article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
with entanglement between two quantum bits , protocols have been demonstrated for teleporting an unknown quantum state @xcite , super dense coding of information @xcite and secure communication @xcite .
an arbitrary qubit can be teleported from one particle to another with the use of an entangled pair of particles , which had been experimentally verified in different quantum systems @xcite .
however , distinguishing all the four bell states of the photonic qubits has remained a fundamental difficulty in achieving 100% teleportation . in the first demonstration of the teleportation with photons @xcite ,
only one of the four bell states was able to distinguish from the others .
thus , the efficiency of teleportation was limited to 25% .
later on , a complete bell state measurement was demonstrated with non - linear interaction of photons@xcite .
even though they could separate all the four bell states , the efficiency was reduced because of the non - linear process involved . in recent years
, complete bell state analysis has been proposed with the use of hyper - entanglement @xcite , where the two photons are entangled in an additional degree of freedom ( dof ) along with polarization .
this method was utilized to increase the channel capacity of super dense coding @xcite .
this was done by projecting each hyper - entangled photon to four single particle two - qubit bell states .
nevertheless , hyper - entanglement assisted bell state analysis is not of much use in teleportation since it requires projecting the unknown state and one of the epr particle state to any of the four bell states . on the other hand
, a hyper - entangled pair of particles can teleport a higher dimensional quantum state using hyper - entangled - bell - state analysis which was described using kerr non - linearity @xcite .
there had been a number of studies which use spin - orbit states of light for quantum information processing @xcite .
khoury and milman @xcite proposed a spin to orbit teleportation scheme with 100% efficiency that uses the oam entanglement between the two photons and a spin - orbit bell state analysis ( soba ) . in this article , we describe a three particle entangled state which finds applications in teleporting two qubits simultaneously and implementing an efficient key distribution protocol . in section [ sc.2 ]
we give a description of the proposed state .
along with the mathematical form of the state we give a schematic for the state preparation . the experimental procedure for the generation of the proposed state
is given in section [ sc.3 ] .
the state can be utilized to teleport two qubits using two sobas and 16 unitary transformations as given in section [ sc.4 ] .
experimental schemes for realizing the @xmath0 gates and soba have also been given in section [ exp ] . in section [ sc.5 ]
we describe a new qkd protocol using the new state , which is more efficient than the traditional ekert protocol .
finally , we conclude in section [ sc.6 ] .
we describe a system of particles in such a way that one particle is entangled to all other particles in different degrees of freedom .
let us consider a system consisting of three photons where photon * 2 * is entangled with photons * 1 * and * 3 * in different degrees of freedom namely oam and polarization respectively .
the polarization state of the photon * 1 * and the oam state of the photon * 3 * are arbitrary or unknown . since the oam of a photon
is expressed in infinite dimensional hilbert space , one can have higher dimensional entangled states .
we take an arbitrary two dimensional subspace of the infinite dimensional oam basis as @xmath1 .
the described state can be prepared using a pair of hadamard and @xmath0 gates in different dofs which correspond to the polarization and the oam . ) and a @xmath0 gate acting on oam basis and polarization " box contains a hadamard ( @xmath2 ) and @xmath0 gates acting on polarization basis . [ fg.2],width=288 ]
the initial states of three particles can be written as @xmath3 where @xmath4 is the unknown state of polarization of the photon * 1 * , @xmath5 is the unknown oam state of the photon * 3 * , @xmath6 represents horizontal and @xmath7 represents vertical polarization states of a photon . the schematic diagram for preparation of the initial state
is given in fig .
[ fg.2 ] . to entangle photons *
1 * and * 2 * in oam , a hadamard gate ( @xmath8 ) on photon * 1 * and subsequent @xmath0 gate on photon * 1 * and * 2 * are applied .
both gates act in the oam degree of freedom .
similarly , another hadamard ( @xmath2 ) acting on photon * 2 * and a @xmath0 between photons * 2 * and * 3 * entangle them in polarization . here , both the gates are acting on the polarization states of photons . after performing the gate operations
mentioned above , the final state becomes , @xmath9
we are proposing a method in which initially two photons are entangled in the oam and third photon is in a pure state of oam and polarization . by polarization gate operations on one of the entangled photons and the independent photon , one can arrive at the described state . experimental scheme for the generation of the state
is given in fig .
[ fg.3 ] . to generate the described state , we start with a type i spontaneous parametric down conversion ( spdc ) of light in a second order nonlinear crystal that gives a pair of photons entangled in the oam @xcite .
a vertically polarized optical vortex beam of azimuthal index 1 has been considered as the pump .
the state corresponding to the pair of photons produced by the spdc of this beam is given by @xmath10 note that the experimental realization of the quantum gates in the oam basis \
{ @xmath11 } are not straightforward . moreover , in the teleportation , it is easier to project the state of a photon to one of its four spin - orbit bell state if we use even / odd basis .
thus for the ease of experimental realization , we reduce the infinite dimensional entangled state to a simple two - qubit entangled state by grouping all the even and the odd oam states and rewrite the expression for the oam state in eq . as @xmath12
we define a transformation from the general oam space to the even - odd oam space as @xmath13 @xmath14 applying @xmath15 on the left - hand - side of eq .
yields @xmath16 . from the conservation of oam
, we have @xmath17 .
thus , we can rewrite eq .
as @xmath18 note that the photons are entangled in the even / odd oam states .
experimentally , this transformation can be implemented by post selection in the even / odd oam basis .
all the oam operations or measurements must be performed in even / odd basis using oam sorter @xcite .
this maps the general oam state to even / odd basis which is mathematically represented by operator `` @xmath15 '' .
let the photon * 1 * pass through a simon - mukunda polarization gadget which can convert its polarization to any arbitrary state @xcite and the photon * 2 * pass through a half wave plate at @xmath19 .
thus polarization state of the photon * 1 * is encoded as the unknown state @xmath20 and the state of the photon * 2 * is encoded as @xmath21 .
action of hwp on the photon * 2 * is a hadamard operation .
thus , the state becomes @xmath22 now , consider the photon * 3 * with unknown superposition state of oam in the even / odd basis and with definite state polarization .
its state can be expressed as , @xmath23 a polarization @xmath0 gate is applied on photon * 2 * ( control ) and * 3 * ( target ) .
this operation leads to a polarization entanglement between photons * 2 * and * 3*. thus , the three particle state becomes @xmath24 this is in the same form as the proposed three particle entangled state described in section * 2 * , eq . .
teleportation of @xmath25 qubits , in general , needs @xmath25 pairs of entangled photons . on the other hand , teleportation of @xmath26-dimensional quantum state
needs @xmath26-dimensional entanglement and one has to do a joint measurement in @xmath27 dimensional bell basis @xcite .
however , teleportation schemes were proposed for a @xmath26 dimensional oam state using bell filter and quantum scissors @xcite . with the entangled state presented above
, we propose an efficient scheme for teleporting a two - qubit state distributed in different dofs . the schematic diagram is given in fig . [ fg.4 ] .
the polarization state of photon * 1 * and the oam state of photon * 3 * form a four dimensional unknown two - qubit state @xmath28 which needs to be teleported .
we can teleport the combined state to a single particle , photon * 2*. the unknown polarization ( spin angular momentum ) and the entangled oam state of photon * 1 * and entangled polarization and the unknown oam state of photon * 3 * in eq .
, can be projected to individual spin - orbit bell states by soba .
alice performs the soba on the photons * 1 * and * 3 * and projects the state of both the particles to the corresponding spin - orbit bell states .
the soba , for which the oam is in the even / odd basis , can be achieved by mach - zehnder interferometers involving an oam sorter and polarizing beam splitters @xcite .
the states can be defined as @xmath29 the soba operation for photon * 1 * can be represented by a polarization controlled oam @xmath0 ( @xmath30 ) gate and a polarization hadamard ( @xmath2 ) gate operation with subsequent detection .
similarly , the soba for photon * 3 * is an oam controlled polarization @xmath0 ( @xmath31 ) gate and oam hadamard ( @xmath8 ) gate operation with subsequent detection .
experimental schemes for these single particle two - qubit @xmath0 gates and soba are given in section [ exp ] . by substituting the single particle spin - orbit bell states into a three particle wave function in eq . , we get @xmath32 where @xmath33 and @xmath34 are single particle spin - orbit bell states corresponding to the photons * 1 * & * 3 * respectively .
@xmath35 is a transformation matrix for the spin - orbit state of photon * 2*. it has been introduced for the compact representation of the 16 states of photon * 2 * given in table [ tb.1 ] . after the soba measurements on the photons * 1 * and * 3 * , state of the photon * 2 * becomes @xmath36 the state corresponding to a detection of @xmath37 in sobas .
the states corresponding to all possible soba outcomes are given in the first and the second columns of the table [ tb.1 ] .
note that the information encoded in the polarization state of the photon * 1 * is transferred to the oam state of the photon * 2 * and the information encoded in the oam state of the photon * 3 * is transferred to the polarization state of the photon * 2*. so , to get the initial state , bob has to do a swapping ( @xmath38 ) between the oam and the polarization along with other unitary transformations which use only polarization operations .
these operations can be implemented with the standard wave plates which makes the experimental realization of this method more feasible . - unitary transformation given in table [ tb.1],width=288 ] there are 16 possible outcomes for alice measurements which demand 16 different unitary operations by bob to complete the teleportation protocol .
alice communicates her measurement outcome ( @xmath39 ) classically to bob .
bob does the corresponding unitary transformation @xmath40 ( given in the third column of table [ tb.1 ] ) on state of photon * 2 * to get @xmath41 which completes the teleportation of unknown state given in eq . .
a swap operation is equivalent to consecutive @xmath42 , @xmath43 and @xmath42 gate operations .
since the operation done in polarization will be transferred to the oam after the swap operation , one can perform all the operations in polarization which are well known .
two half wave plates before and after the swap gate can perform 16 unitary operations which is required for the teleportation .
a circuit diagram of the proposed scheme is given in fig .
[ fg.5 ] . [ cols="^,<,<",options="header " , ] photons * 1 * and * 2 * have oam correlation , so it will violate the following bell s inequality for 3 dimensional case @xcite @xmath44 where @xmath45 the shifts of holograms are chosen in such a way that they maximally violate the bell - type inequality .
@xmath46=0,1,2 corresponds to the detection of oam states 0 , 1 and -1 respectively .
the coincidence measurements with the combinations of @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 , @xmath49 , @xmath52 , @xmath53 and @xmath54 are required for the key distribution and to check the security of the key using eq . and .
our protocol has three advantages compared to traditional ekert protocol @xcite . *
if ekert protocol is used , then around @xmath55 pairs of photons , i.e. , @xmath56 photons ( in practice , little more than @xmath56 ) are required to establish a secret key of length @xmath57 between alice and bob . in our approach , around @xmath58 photons
would be required for the same purpose .
hence , in terms of number of photons required , ekert s protocol is 33% less efficient than ours .
* on the other hand , if the same number of photons are used in ekert and our protocol and if the target key length is also the same , then because of more entanglement resource , our protocol would have more redundancy and hence can tolerate more noise .
it is easy to see that the security in each degree of freedom is equivalent to that of the ekert protocol .
* further , this state can be used for multi - party qkd .
photons 1 , 2 and 3 can be distributed amongst alice , bob and charlie respectively .
now , two sets of independent keys can be generated , one for alice - bob and another for alice - charlie .
possibility of a new entangled state and its application in teleporting a two - qubit oam - polarization quantum state and in the qkd have been discussed .
the new teleportation method overcomes the difficulty of measuring in two particle polarization bell basis , by implementing independent single particle two - qubit bell measurements .
the method critically depends on the experimental realizations of polarization @xmath0 gate as well as single particle two - qubit @xmath0 gates .
all the 16 unitary transformations which are required for this teleportation scheme can be realized with linear optical components along with a swap gate . once the state given in eq .
is achieved , one can have a 100% efficient teleportation of two simultaneous qubits .
extending this to higher dimensions is mathematically trivial though creating entanglement in different dofs has experimental limitations . with the new qkd protocol ,
the sender and the receiver need to use less resource than traditional ekert protocol to share the secret key of the same length .
multi - party schemes also can be developed for the teleportation and the qkd using the described protocols .
authors wish to acknowledge prof .
anirban pathak , jiit noida for fruitful discussions and suggestions .
c. h. bennett , g. brassard , c. crpeau , r. jozsa , a. peres , and w. k. wootters , teleporting an unknown quantum state via dual classical and einstein - podolsky - rosen channels , phys .
70 , 18951899 ( 1993 ) .
m. riebe , h. hffner , c. roos , w. hnsel , j. benhelm , g. lancaster , t. krber , c. becher , f. schmidt - kaler , d. james _ et al .
_ , deterministic quantum teleportation with atoms , nature 429 , 734737 ( 2004 ) .
s. g. reddy , s. prabhakar , a. aadhi , a. kumar , m. shah , r. singh , and r. simon , measuring the mueller matrix of an arbitrary optical element with a universal su ( 2 ) polarization gadget , j. opt .
am . a 31 , 610615 ( 2014 ) .
r. ursin , f. tiefenbacher , t. schmitt - manderbach , h. weier , t. scheidl , m. lindenthal , b. blauensteiner , t. jennewein , j. perdigues , p. trojek _ et al .
_ , entanglement - based quantum communication over 144 km , nature phys .
3 , 481486 ( 2007 ) .
m. mafu , a. dudley , s. goyal , d. giovannini , m. mclaren , m. j. padgett , t. konrad , f. petruccione , n. ltkenhaus , and a. forbes , higher - dimensional orbital - angular - momentum - based quantum key distribution with mutually unbiased bases , phys .
a 88 , 032305 ( 2013 ) . | we present a scheme to generate three particle hyper - entanglement utilizing polarization and orbital angular momentum ( oam ) of a photon .
we show that the generated state can be used to teleport a two - qubit state described by the polarization and the oam .
the proposed quantum system has also been used to describe a new efficient quantum key distribution ( qkd ) protocol .
we give a sketch of the experimental arrangement to realize the proposed teleportation and the qkd . |
in this paper , we examine the one - dimensional zakharov system ( 1d zs ) @xmath7 local well - posedness in the inhomogeneous sobolev spaces @xmath8 has been obtained by means of the contraction method in the bourgain space @xmath9 by bourgain - colliander @xcite and ginibre - tsutsumi - velo @xcite . and
@xmath10 , respectively . ] in the latter paper , the following result is obtained : [ t : gtvprop1.2 ] 1d zs is locally well - posed for initial data @xmath11 provided @xmath12 specifically : 1 . _
existence_. @xmath13 , if @xmath14 , then @xmath15 and a solution @xmath16 to 1d zs on @xmath17 $ ] such that @xmath18 ; h_x^k ) } \leq c\| u_0\|_{h^k } \\
& \|n\|_{c([0,t ] ; h_x^s ) } + \|\partial_t n\|_{c([0,t ] ; h_x^{s-1 } ) } \leq c{\langle}\|u_0\|_{h^k}{\rangle}^2 ( \|n_0\|_{h^s } + \|n_1\|_{h^{s-1 } } ) \end{aligned}\ ] ] and @xmath19 , where @xmath20 is given by table [ tab : gtv ] .
uniqueness_. this solution is unique among solutions @xmath21 belonging to @xmath22 ; h_x^k)\cap x^s_{k , b_1}$ ] . alone with @xmath23 solving appearing as a coefficient .
then , @xmath24 can be expressed in terms of @xmath21 and @xmath23 , and therefore @xmath24 need not enter into the uniqueness claim . ]
_ uniform continuity of the data - to - solution map_. for a fixed @xmath25 , taking @xmath26 as above , the map @xmath27 as a map from the @xmath28-ball in @xmath29 to @xmath22 ; h_x^k ) \times c([0,t ] ; h_x^s ) \times c([0,t ] ; h_x^{s-1})$ ] is uniformly continuous .
the region of local well - posedness in this theorem is depicted in fig .
[ f : gtv ] .
we shall outline the @xcite proof of theorem [ t : gtvprop1.2 ] in [ s : gtv ] since the estimates are needed in the proof of theorem [ t : northip ] in [ s : northip ] .
[ f : gtv ] our goal in this paper is to establish local ill - posedness outside of the well - posedness strip , in particular near the optimal corner @xmath30 , @xmath31 .
that is , we consider the region ( 1 ) @xmath32 ( above the strip ) , and ( 2 ) @xmath33 ( below the strip ) . in the first region , the wave data @xmath34 is somewhat smoother than the schrdinger data @xmath35 . as a result ,
the forcing term @xmath36 of the wave equation , as time evolves , introduces disturbances that are rougher than the wave data , and the wave solution @xmath24 does not retain its higher initial regularity .
this is quantified in theorem [ t : northip ] below . in the second region ,
the schrdinger data @xmath35 is somewhat smoother than the wave data @xmath34 . as a result ,
the forcing term @xmath37 of the schrdinger equation introduces disturbances that are rougher than the schrdinger data , and the schrdinger solution @xmath21 does not retain its higher initial regularity .
this is quantified in theorem [ t : deepsouthip ] and [ t : nearsouthip ] below .
these simplistic explanations are , at least , accurate for @xmath38 . for @xmath39 , there are possibly multiple simultaneous causes for breakdown , although we find that our methods still yield information in this setting
. we will draw upon and suitably modify techniques developed by birnir - kenig - ponce - svanstedt - vega @xcite , christ - colliander - tao @xcite , and bourgain @xcite , who addressed ill - posedness issues for the nonlinear schrdinger equation . for a survey of ill - posedness results for nonlinear dispersive equations ,
see tzvetkov @xcite .
our first result demonstrates that the boundary line @xmath40 in theorem [ t : gtvprop1.2 ] is sharp .
[ t : northip ] let @xmath41 and @xmath32 or @xmath42 and @xmath43 .
there exists a sequence @xmath44 such that @xmath45 for all @xmath46 and the corresponding solution @xmath47 to 1d zs on @xmath17 $ ] with initial data @xmath48 satisfies @xmath49 where @xmath50 .
the time interval @xmath17 $ ] here is independent of @xmath46 .
the form of ill - posedness appearing in theorem [ t : northip ] is referred to as `` norm inflation '' .
the result is first reduced to the case where @xmath38 and @xmath4 is just above the line @xmath51 . in this case ,
theorem [ t : gtvprop1.2 ] applied with @xmath51 ( the wave initial data is @xmath52 ) provides the existence of a solution @xmath47 on a time interval @xmath53 , independent of @xmath46 , with uniform - in-@xmath46 control on @xmath54 .
the estimates of @xcite will enable us to show that @xmath55 is comparable to @xmath56 in a slightly stronger norm than @xmath57 ( on this fixed in @xmath46 time interval ) and then theorem [ t : northip ] follows from the fact that holds with @xmath58 replaced by @xmath59 , which can be directly verified .
is the solution to @xmath60 , @xmath61 , @xmath62 . ]
the proof is given in [ s : northip ] .
our second theorem demonstrates lack of uniform continuity of the data - to - solution map , for any @xmath63 , as a map from the unit ball in @xmath29 to @xmath22 ; h^k)\times c([0,t ] ; h^s ) \times c([0,t ] ; h^{s-1})$ ] for @xmath30 and any @xmath64 .
we first show that if one issue is ignored , we can , in a manner similar to @xcite , make use of an explicit soliton class to demonstrate that for any @xmath63 there are two waves , close in amplitude on all of @xmath17 $ ] , initially of the same phase but that slide completely out of phase by time @xmath53 .
this form of ill - posedness is termed `` phase decoherence '' . the soliton class for 1d zs that we use appears in @xcite @xcite .
the `` ignored issue '' pertains to low frequencies of @xmath65 , and can be resolved by invoking the method of @xcite to construct a `` near soliton '' class offering more flexibility than the exact explicit soliton class in the selection of @xmath65 .
this is , however , not straightforward since 1d zs lacks scaling and galilean invariance , which was used to manufacture the solution class in @xcite .
[ t : deepsouthip ] suppose @xmath30 , @xmath66 .
fix any @xmath63 and @xmath67 .
then there is a pair of schwartz class initial data tuples @xmath68 and @xmath69 giving rise to solutions @xmath16 and @xmath70 on @xmath17 $ ] such that the data is of unit size @xmath71 and initially close @xmath72 but the solutions become well - separated by time @xmath53 in the schrdinger variable @xmath73}^\infty h_x^k } \sim 1 \,.\ ] ] we expect that this result can be extended to all @xmath74 and @xmath66 , although preliminary efforts were abandoned since the computations became very lengthy and technical .
the proof of theorem [ t : deepsouthip ] appears in [ s : cct ] .
our final theorem employs a method of bourgain @xcite .
[ t : nearsouthip ] for any @xmath63 , the data - to - solution map , as a map from the unit ball in @xmath75 to @xmath22 ; h^k)\times c([0,t ] ; h^s ) \times c([0,t ] ; h^{s-1})$ ] fails to be @xmath76 for @xmath74 and @xmath33 .
this is a weaker form of ill - posedness than the phase decoherence of theorem [ t : deepsouthip ] , although it covers the full region below the well - posedness boundary @xmath31 of @xcite .
the proof is given in
[ s : bourgain ] .
+ * acknowledgments*. i would like to thank jim colliander for his clear explanation of how to construct counterexamples to bilinear estimates and for other helpful discussion on this topic .
also , i would like to thank guixiang xu for carefully reading
13 of the paper and pointing out numerous misprints and an error .
finally , i would like to thank the anonymous referee for providing several helpful suggestions for improvement .
we outline and review the local well - posedness argument in @xcite since the estimates will be needed in the proofs of theorem [ t : northip ] , [ t : nearsouthip ] .
let @xmath77\sphat(\xi ) = e^{-it\xi^2}\hat{u}_0(\xi)$ ] and @xmath78 denote the schrdinger group and duhamel operators , respectively .
define the schrdinger bourgain spaces @xmath79 , @xmath80 by the norms @xmath81 consider an initial wave data pair @xmath82 . split @xmath83 into low and high frequencies would have to be modified to have @xmath84 in place of @xmath85 on the right - hand side ] , and set @xmath86 , so that @xmath87 .
let @xmath88 so that @xmath89 by setting @xmath90 , we obtain a solution to the linear homogeneous problem @xmath91 let @xmath92 so that @xmath93 it follows that if we set @xmath94 , then we obtain a solution to the linear inhomogeneous problem @xmath95 define the one - dimensional reduced wave bourgain spaces @xmath96 , @xmath97 as @xmath98 let @xmath99 on @xmath100 $ ] and @xmath101 outside of @xmath102 $ ] .
let @xmath103 , which will serve as a time cutoff for the bourgain space estimates .
for clarity , we write @xmath104 .
we can now recast 1d zs as @xmath105 where @xmath106 , which has the integral equation formulation @xmath107(t ) \\
n_\pm(t ) & = w_\pm(t)(n_0,n_1 ) \mp w_\pm\ast_r(\partial_x|u|^2)(t)\ , .
\end{aligned}\ ] ] [ l : group ] 1 .
[ i : groupsch ] _
schrdinger_. @xmath108 .
[ i : groupwave]_1-d wave_. @xmath109 [ l : duhamel ] suppose @xmath110 . 1 .
[ i : duhamelsch]_schrdinger_. if @xmath111 , @xmath112 , @xmath113 , then @xmath114 . + if @xmath115 , then @xmath116 .
+ @xmath117 .
[ i : duhamelwave]_1-d wave_. if @xmath118 , @xmath119 , @xmath120 , then @xmath121 .
+ if @xmath122 , then @xmath123 .
+ @xmath124 .
[ l : gtv4.3 ] let @xmath125 satisfy @xmath126 then @xmath127 [ l : gtv4.4 ] let @xmath128 satisfy @xmath129 then @xmath130 to obtain theorem [ t : gtvprop1.2 ] , fix @xmath131 , and consider the maps @xmath132 , @xmath133 @xmath134 \\ \label{e:103 } \lambda_{w\pm}(u ) & = \psi_1w_\pm(n_0,n_1 ) \pm \psi_t w_\pm \ast_r(\partial_x |u|^2 ) \,.\end{aligned}\ ] ] for @xmath135 , a fixed point @xmath136 is obtained in @xmath137 satisfying @xmath138 by applying lemmas [ l : group ] , [ l : duhamel ] , [ l : gtv4.3 ] , [ l : gtv4.4 ] with values for @xmath20 , @xmath139 , @xmath140 , @xmath141 given by table [ tab : gtv ] .
.values of @xmath20 , @xmath139 , @xmath140 , @xmath141 meeting the criteria of lemmas [ l : gtv4.3 ] , [ l : gtv4.4 ] for various intervals of @xmath142 . note that @xmath143 and @xmath144 in order to capture a factor @xmath145 from lemma [ l : duhamel ] . also note that @xmath146 and @xmath147 for all cases except @xmath148 .
[ cols="^ , < , < " , ] consider first the case @xmath149 .
we note from table [ tab : gtv ] that @xmath150 , and thus we have the sobolev imbeddings @xmath151 also , @xmath152 and thus @xmath153 similar estimates apply to differences of solutions .
consider now the case @xmath148 , where it is necessary to take @xmath154 .
we return to and estimate directly using lemma [ l : duhamel ] to obtain @xmath155 and by lemma [ l : gtv4.3 ] , @xmath156 where @xmath20 , @xmath140 are as specified in the table [ tab : gtv ] , and the right - hand side is appropriately bounded by , .
the bounds in , apply in this case since @xmath157 .
we further note that we can re - estimate @xmath21 in @xmath158 in to obtain @xmath159
here we prove theorem [ t : northip ] . in steps 13
, the result will be established for @xmath161 and @xmath32 but with @xmath4 _ near _ @xmath162 . in steps 45 , the general case of the theorem is reduced to the case considered in steps 13 .
let @xmath41 .
let @xmath163}(\xi ) \\ \hat{\phi}_{n , b}(\xi ) & = n^{\frac{1}{2}-k}\chi_{[n+1,n+1+\frac{1}{n}]}(\xi)\end{aligned}\ ] ] let @xmath164
. then @xmath165 .
a solution to the integral equation @xmath166\cdot u_n\}(t)\ ] ] provides a solution to 1d zs with initial data @xmath167 when @xmath168 is defined in terms of @xmath55 as @xmath169 by working with the estimates in lemmas [ l : gtv4.3 ] , [ l : gtv4.4 ] ( taking @xmath170 in the discussion of [ s : gtv ] ) , we obtain a solution @xmath55 to in @xmath171 for @xmath172 , on @xmath17 $ ] , where @xmath173 ( thus independent of @xmath46 ) satisfying @xmath174 ; h_x^{k-\sigma } ) } \leq \| u_n\|_{x^s_{k-\sigma , \frac{3}{4}-2\epsilon } } \leq \|\phi_n\|_{h^{k-\sigma } } \sim n^{-\sigma}\ ] ] * step 1*. we show that @xmath175(t ) \|_{h^s } \sim t n^{s-(2k-\frac{1}{2 } ) } \qquad \text{for } n\gtrsim t^{-1}\ ] ] that says that holds provided @xmath176 is replaced by the linear flow @xmath177 in . to show this , note that in the pairing @xmath178 , there are 4 combinations @xmath179 , where @xmath180 .
we claim that @xmath181\sphat(\xi , t ) \ ;
\sim \;i\xi t n^{1 - 2k}e^{-it\xi } h_1(\xi)\ ] ] where @xmath182 is the `` triangular step function '' with peak at @xmath183 , of width @xmath184 , and of height @xmath185 , i.e. @xmath186 \\ ( -2n-1)-\xi & \text{if } \xi \in [ -2n-1-\frac{1}{n } , -2n-1 ] \end{cases}\ ] ] here , the symbol @xmath187 means that the difference between the two quantities has @xmath188 norm of lower order in @xmath46 .
it then follows by taking complex conjugates in that @xmath189\sphat(t,\xi ) \sim i\xi t e^{-it\xi } n^{1 - 2k}h_2(\xi)\ ] ] where @xmath190 is the `` triangular step function '' centered at @xmath191 , of width @xmath184 , and of height @xmath185 , i.e. @xmath192 \\ ( 2n+1+\frac{2}{n } ) - \xi & \text{if } \xi \in [ 2n+1+\frac{1}{n } , 2n+1+\frac{2}{n } ] \end{cases}\ ] ] hence @xmath193(t)\|_{h_x^s } \sim t n^{s-(2k-\frac{1}{2})}\ ] ] we further claim that the @xmath194 and @xmath195 interactions for the @xmath196 term are of lower order in @xmath46 , i.e. specifically , @xmath197 finally , we claim that all of the interactions @xmath194 , @xmath198 , @xmath199 , and @xmath195 for the @xmath200 term are of lower order in @xmath46 , i.e. @xmath201(t ) \|_{h^s } \leq n^{s-(2k-\frac{1}{2})-1 } \quad \text{for } j , k \in \ { \ , a , b \}\ ] ] combining , establishes .
we begin by proving .
note that @xmath202 } e^{ix\xi_1 } e^{-it\xi_1 ^ 2 } \
, d\xi_1 \\ & \overline{u(t)\phi_{n , b}(x ) } = n^{\frac12-k}\int_{\xi_2 \in [ -n-1-\frac{1}{n } , -n-1 ] } e^{ix\xi_2 } e^{it\xi_2 ^ 2 } \ , d\xi_2\end{aligned}\ ] ] after the change of variable @xmath203 in the second equation . for the remainder of the computation ,
@xmath204 is restricted to @xmath205 $ ] and @xmath206 is restricted to @xmath207 $ ] . by ,
@xmath208 where @xmath209 since @xmath210 is confined to a @xmath185-sized interval around @xmath211 and @xmath212 is confined to a @xmath185-sized interval around @xmath52 , we have that @xmath213 is confined to a unit - sized interval around @xmath52 . by the power series expansion for @xmath214
, we have @xmath215 .
@xmath216\sphat(\xi , t ) \\ & = n^{1 - 2k}\int_{\xi_1 } \int_{\xi_2 } i(\xi_1+\xi_2 ) \delta(\xi_1+\xi_2-\xi ) e^{-it(\xi_1 ^ 2-\xi_2 ^ 2 ) } g(t,\xi_1,\xi_2 ) \ , d\xi_1 \ , d\xi_2\end{aligned}\ ] ] using that @xmath217 and that @xmath215 , we obtain . and are proved by a similar computation ; we only present the proof of in the case @xmath218 . for @xmath219
$ ] , @xmath220 where @xmath221\sphat(\xi_1,\tau_1 ) [ \overline{\psi_1 u{\phi}_{n , a}}]\sphat(\xi_2,\tau_2 ) \\ & = \iint_{\substack { \xi=\xi_1+\xi_2 \\ \tau=\tau_1+\tau_2 } } \hat{\psi}_1(\tau_1+\xi_1 ^ 2 ) \hat{\phi}_{n , a}(\xi_1 ) \hat{\psi}_1(\tau_2-\xi_2 ^ 2 ) \overline{\hat{\phi}_{n , a}}(-\xi_2)\end{aligned}\ ] ] in this integral , @xmath204 and @xmath206 are each confined to a @xmath185 sized interval around @xmath222 , forcing @xmath223 to lie in a @xmath185 sized interval around @xmath224 .
the @xmath225 and @xmath226 factors then ( essentially ) restrict @xmath227 to a unit sized interval around @xmath228 and restrict @xmath229 to a unit sized interval around @xmath230 , so that @xmath231 is forced to lie within a unit sized interval around @xmath52 .
consequently , @xmath232\times [ -1,1 ] \\ = 0 & \text{otherwise } \end{cases}\ ] ] on the support of @xmath233 , the factor @xmath234 .
from , @xmath235(t)\|_{h^s } \\
& \leq n^{1 - 2k}\left ( \int_\xi |\xi|^2 { \langle}\xi{\rangle}^{2s } \left [ \int_\tau \frac{|g(\tau,\xi)|}{|\tau+\xi| } d\tau \right]^2 d\xi \right)^{1/2 } \leq n^{s-(2k-\frac{1}{2})-1}\end{aligned}\ ] ] * step 2*. also , on this time interval @xmath17 $ ] independent of @xmath46 , we claim that @xmath236 where @xmath237 note that @xmath238 will be @xmath239 provided @xmath240 is not chosen too large .
this says that @xmath176 is well - approximated by the linear flow @xmath241 in the _ stronger _ norm @xmath242 .
we now prove . from , @xmath243 for @xmath20 as defined above and @xmath244 following with lemma [ l : gtv4.3 ] , @xmath245 where @xmath246 by lemma [ l : gtv4.4 ] , @xmath247 where @xmath248 and @xmath249 , @xmath20 are defined above .
note that @xmath250 .
combining , @xmath251 by , @xmath252 since @xmath253 , provided @xmath46 is taken large enough and @xmath254 , will follow .
* step 3*. here , we establish @xmath255 if @xmath256 and @xmath257 , or if @xmath258 and @xmath259 .
to show this , we note that by and , it suffices to show that @xmath260 writing @xmath261\end{aligned}\ ] ] we see that it suffices to show that @xmath262(t ) \|_{h^s } \leq 1 \notag\\ & \| [ w_\pm \ast_r \partial_x ( u_n-\psi_1u\phi_n )
\overline{\psi_1 u\phi_n}](t ) \|_{h^s } \leq 1 \label{e:213}\\ & \| [ w_\pm \ast_r \partial_x ( \psi_1 u\phi_n \cdot \overline{u_n-\psi_1u\phi_n } ) ] ( t)\|_{h^s } \leq 1 \notag\end{aligned}\ ] ] we focus on the middle estimate ; the other two are handled similarly . as we describe in detail below , by requiring @xmath4 to lie sufficiently close to ( but above ) @xmath162 , we can assign @xmath240 such that @xmath263 and also @xmath264 where @xmath249 is given in . then proceed to estimate the left - hand side of by lemma [ l : duhamel ] as @xmath265 by step 2 and lemma [ l : group ] , @xmath266 by
, it follows that the exponent is @xmath267 .
we now provide the details assigning @xmath268 in terms of @xmath3 and @xmath4 .
the condition is equivalent to the restriction @xmath269 the following assignments meet the criteria and . * if @xmath256 , restrict to @xmath4 such that @xmath257 , and set @xmath270 . *
if @xmath258 , then restrict to @xmath4 such that @xmath259 and set @xmath271 .
* step 4*. suppose @xmath41 and @xmath32 .
let @xmath272 be such that @xmath273 and @xmath272 meets the restrictions outlined in step 3 with @xmath4 replaced by @xmath272 . then by steps 13 ( with @xmath4 replaced by @xmath272 ) @xmath274
so we can take @xmath275 in the statement of the theorem . +
* step 5*. next , suppose @xmath39 and @xmath43 . by the reasoning of step 4
, it suffices to restrict to @xmath276 .
set @xmath277 , and note that @xmath278 .
clearly @xmath279 , so we can just appeal to the conclusion of steps 14 applied with @xmath3 replaced by @xmath280 .
let @xmath282 , which is the unique positive ground state solution to @xmath283 let @xmath284 and set @xmath285 from , it follows that @xmath286 solves 1d zs for all @xmath287 and @xmath288 .
this is the exact soliton class appearing in @xcite and @xcite .
our next goal is to prove theorem [ t : deepsouthip ] demonstrating phase decoherence ill - posedness for @xmath30 , @xmath66 .
we first , however , settle for a partial result ( proposition [ p : deepsouthippartial ] ) using a pair from the above exact explicit soliton class .
we include this result since it is clear and straightforward and exhibits the idea behind the proof of the full result ( theorem [ t : deepsouthip ] ) , which is considerably more technical and appears in the next section .
define the norm @xmath289 as @xmath290 the limitation of the following partial result is the use of @xmath291 and @xmath292 norms as opposed to the full @xmath188 and @xmath293 norms .
[ p : deepsouthippartial ] suppse @xmath281 .
fix any @xmath63 , @xmath67 .
then @xmath294 sufficiently large and @xmath295 sufficiently close to @xmath296 so that if @xmath297 then the solutions are of unit size on @xmath17 $ ] , @xmath298 and are initially close @xmath299 @xmath300 but become fully separated in the @xmath21-variable by time @xmath53 , @xmath301 we will select @xmath302 sufficiently large later . take @xmath303 sufficiently close to @xmath296 so that @xmath304 . then since @xmath305 we have @xmath306 and noting that @xmath307 and @xmath308 gives @xmath309 take @xmath310 sufficiently large so that @xmath311 is sufficiently close to @xmath312 in order to make the above expression @xmath313 .
thus is established .
next , we establish . by the change of variable @xmath314 @xmath315 since @xmath281 we have @xmath316 and
the above difference is made @xmath313 by again taking @xmath310 sufficiently large .
also @xmath317 ( here , the notation @xmath318 indicates the derivative ) .
since @xmath281 we have @xmath316 and the above difference is made @xmath313 by again taking @xmath310 sufficiently large .
the statements are proved by similar change of variable calculations .
the need for the restrictions to @xmath319 in is clear from these calculations .
in fact , one can show that for @xmath320 , we have @xmath321 as @xmath322 due to the @xmath323 frequency contribution .
now we establish .
the key observation here is that while @xmath324 is very small ( as @xmath325 ) , @xmath326 is of fixed size @xmath327 and thus @xmath328 is purely imaginary . now
@xmath329 but the last term on the right - hand side is @xmath330 which , combined with gives .
here , we remove the shortcoming of proposition [ p : deepsouthippartial ] ( high frequency truncated norms @xmath289 , @xmath331 used instead of @xmath188 , @xmath293 ) and prove theorem [ t : deepsouthip ] . the soliton class employed in the proof of proposition [ p : deepsouthippartial ] involved assigning @xmath332 and thus @xmath333 . replace @xmath334 in the definition of @xmath24 by @xmath335 defined by @xmath336 ( i.e. the restriction to frequencies @xmath337 ) and set @xmath338
then @xmath339 unfortunately , @xmath340 is no longer a solution to 1d zs .
we shall thus adapt the method of christ - colliander - tao @xcite to construct a `` near soliton '' class that grants more flexibility in the selection of the wave initial data .
the method procedes by solving a `` small dispersion approximation '' to the equation , and by introducing scaling and phase translation parameters , building the `` near soliton '' class . the main new obstacle , in comparison to the work of @xcite applied to the nonlinear schrdinger equation ,
is that 1d zs does not possess scaling nor the galilean ( phase shift ) identity .
we thus need to carry out the small dispersion approximation for a _ modified _ zakharov system with the property that when scaling and phase shift operations are performed , the modified zakharov system is converted into the true zakharov system .
consider fixed initial data @xmath341 ( to be defined later ) . *
step 1*. the solution to the small dispersion approximation @xmath342 with @xmath343 is @xmath344 * step 2*. for parameters @xmath345 , @xmath346 , @xmath288 , consider the initial - value problem for the modified zakharov system @xmath347 , @xmath348 @xmath349 & \tfrac{1}{2}\left[n_0\left(x+ \frac{\nu(1 + 2n)}{\lambda}t \right)+ n_0\left(x-
\frac{\nu(1 - 2n)}{\lambda}t\right)\right]u \\ & + ( n_++n_-)u \end{aligned}\\ & \frac{\lambda}{(1 - 2n)\nu } \partial_t n_+ + \partial_x n_+ = -\tfrac{1}{2}(1 + 2n ) \partial_x |u|^2 \\
& \frac{\lambda}{(1 + 2n)\nu } \partial_t n_- - \partial_x n_- = \tfrac{1}{2}(1 - 2n ) \partial_x |u|^2 \\ & u(x,0)=u_0(x ) , \qquad n_\pm(x,0)=0 \end{aligned } \right.\ ] ] if @xmath350 and ( implicit constants here depend on @xmath351 and @xmath352 ) @xmath353 then , on @xmath17 $ ] , we have the two estimates @xmath354 to show this , note first that @xmath355 where @xmath356 and @xmath357 , and thus for @xmath350 , @xmath358 by the energy method applied to , we have @xmath359 \ , \overline{\partial_x^ku } \ , dxdt -2{\operatorname{re}}i \int_0^t\int_x \partial_x^k[n_\pm u ] \ , \overline{\partial_x^k u } \ , dxdt \\ & = \text{i } + \text{ii}\end{aligned}\ ] ] term i will be addressed via the gronwall inequality , while in estimating ii we will produce a small coefficient .
@xmath360 term ii is decomposed as @xmath361 from , @xmath362 while for @xmath363 , we integrate by parts ( here @xmath364 means up to terms bounded similarly to ) @xmath365 \ , dxdt \\ & \approx \int_0^t \int_x \partial_x^{k-1}n_\pm \,\partial_x^{k-1}(iu\overline{\partial_x^2 u } - i\bar{u}\partial_x^2u ) \ , dxdt \\ & = -\frac{1}{\nu^2 } \int_0^t \int_x \partial_x^{k-1}n_\pm \ ,
\partial_x^{k-1}\partial_t |u|^2 \ , dxdt \\ & = -\frac{1}{\nu^2 } \int_x \partial_x^{k-1}n_\pm(t ) \ , \partial_x^{k-1}|u|^2(t ) \ , dxdt + \frac{1}{\nu^2 } \int_0^t \int_x \partial_t \partial_x^{k-1}n_\pm \ , \partial_x^{k-1}|u|^2 \ , dxdt \\ & = \text{ii}_{b1 } + \text{ii}_{b2}\end{aligned}\ ] ] from , we have @xmath366 from , @xmath367 , so @xmath368 all together , ( using @xmath369 conservation as well ) , @xmath370 with @xmath371 where the last inequality follows from the assumptions .
by the gronwall inequality , @xmath372 provided we have @xmath373 we obtain , from and a continuity in time argument , the bound @xmath374 now , the assumptions imply ; and implies by the first of the assumptions in .
* step 3*. with @xmath347 as defined in step 2 , @xmath375 as defined in step 1 , and satisfied , we claim that @xmath376 where the implicit constant depends on @xmath377 and @xmath378 . for this
, we appeal to the result of step 2 at the level of @xmath379 derivatives , and then apply the energy method to the difference @xmath380 in @xmath381 : @xmath382 & -2{\operatorname{re}}i\nu^2 \int_0^t \int_x \partial_x^{k+2}u \ , \overline{\partial_x^k(u - v ) } \ , dxdt \\ & -2{\operatorname{re}}i \int_0^t\int_x \partial_x^k\left [ n_0\left ( x \pm \frac{\nu(1\pm 2n)}{\lambda}t\right)u - n_0(x)v\right ] \overline{\partial_x^k(u - v ) } \ , dxdt \\ & -2{\operatorname{re}}i\int_0^t \int_x \partial_x^k[n_\pm u ] \ , \overline{\partial_x^k(u - v ) } \ , dxdt \end{aligned}\\ & = \text{i}+\text{ii}+\text{iii}\end{aligned}\ ] ] direct estimates using , give @xmath383 by and , @xmath384 by rewriting @xmath385 u + n_0(x)(u - v)\end{aligned}\ ] ] term ii can be estimated as @xmath386 & \frac{\nu(1\pm 2n)t^2}{\lambda}\|n_0\|_{h^{k+1}}\|u\|_{l_t^\infty h_x^k } \|u - v\|_{l_t^\infty h_x^k } \\ & + \|n_0\|_{h_x^k } \int_0^t \|(u - v)(t ) \|_{h_x^k}^2 \ , dt \end{aligned}\ ] ] combining , and applying the gronwall inequality , we have @xmath387\ ] ] the result follows from the assumptions .
* step 4*. for @xmath288 , set @xmath388 then @xmath16 solves 1d zs , with @xmath389 and initial data @xmath390 , @xmath391 , @xmath392 .
consider @xmath393 and @xmath345 .
since @xmath394 for all @xmath395 , we have by change of variable @xmath396 also , if @xmath397 for @xmath398 and @xmath399 , then another change of variable gives @xmath400 if @xmath66 and @xmath401 and @xmath402 satisfy @xmath403 then @xmath404 .
* step 5*. fix @xmath405 and @xmath406 , to be chosen momentarily . in terms of @xmath310 and @xmath402 , define the following quantities : let @xmath407 , and set @xmath408 we note that @xmath409 is purely imaginary . note further that @xmath410 take @xmath46 sufficiently close to @xmath296 so that @xmath411 take @xmath412 such that @xmath413 for @xmath414 , and @xmath65 to be the smooth function given on the fourier side as @xmath415 so that in fact @xmath416 now consider the solutions @xmath417 and @xmath418 of the modified zakharov system given at the beginning of step 2 , both in terms of the data @xmath419 and @xmath65 .
define , as in step 4 , the 1d zs solution @xmath420 in terms of @xmath417 and @xmath421 in terms of @xmath418 . by the comments at the end of step 4 ,
@xmath422 where , in order to meet condition , we need @xmath423 . by a change of variable and , @xmath424 by and
the fact that @xmath425 for all @xmath395 and uniformly in all the parameters , we can take @xmath426 sufficiently small so that @xmath427 by the results of step 1 and 3 , again taking @xmath428 sufficiently small , if @xmath429 , then @xmath430 here , the first condition of is met , since the @xmath53 appearing there is our @xmath431 . we see trivally from that @xmath432 but by and and the choice of @xmath419 and @xmath65 , @xmath433}^\infty l_x^2 } = \mathcal{o}(1)\ ] ] we further note that @xmath434 and therefore we can accomodate an arbitrarily small preselected time , @xmath53 as in the statement of the theorem .
in this section , we give the proof of theorem [ t : nearsouthip ] . for fixed @xmath435 data @xmath436 , to be specified later , and a parameter @xmath437 , consider initial data @xmath438 and corresponding 1d zs solutions @xmath439 .
clearly @xmath440 the solution , written in integral equation form , is : @xmath441\ , dt ' \\ &
n(t ) = w(t)(\gamma n_0,\gamma n_1 ) \pm \tfrac{1}{2 } \int_0^t \partial_x |u|^2(x\pm s , t - s ) \ , ds\end{aligned}\ ] ] from which it follows that @xmath442\ , dt ' \\ & \partial_\gamma n(t ) = w(t)(n_0,n_1 ) \pm \tfrac{1}{2}\int_0^t \partial_x ( \partial_\gamma u \ , \bar{u } + u \overline{\partial_\gamma u})(x\pm s , t - s ) \
, ds \end{aligned}\ ] ] by , @xmath443 by applying @xmath444 to and again appealing to , we get @xmath445 by applying @xmath446 to , we obtain @xmath447\ , dt ' \\ & \partial_\gamma^2 n(t ) = \pm \int_0^t \partial_x ( \partial_\gamma^2 u \ , \bar{u } + 2|\partial_\gamma u|^2 + u \overline{\partial_\gamma^2 u})(x\pm s , t - s ) \ , ds\end{aligned}\ ] ] from which we find , together with , that @xmath448\ , dt'\\ & \partial_\gamma^2 n \big|_{\gamma=0}(t ) = \pm \int_0^t \partial_x|u(t')u_0|^2(x\mp s , t - s ) \ , ds\end{aligned}\ ] ] let @xmath449 , @xmath450 .
fix @xmath451 , and let @xmath452 be the solution map @xmath453 .
let @xmath454 be given by @xmath455 .
let @xmath456 so that @xmath457
. then ( here @xmath458 denotes a linear map @xmath459 ) @xmath460 also @xmath461 and since @xmath462 @xmath463 hence @xmath464 we now note how to prescribe an appropriate sequence @xmath465 ( indexed by @xmath46 ) to show that @xmath466 is not a bounded ( continuous ) bilinear map in the two cases ( 1 ) @xmath33 and ( 2 ) @xmath32 . * if @xmath33 , @xmath467}(\xi)\ ] ] @xmath468}(\xi ) + n^{\frac{1}{2}-s}\chi_{[-2n+1-\frac{1}{n},-2n+1]}(\xi)\ ] ] and @xmath469 , then @xmath465 is a sequence such that @xmath470 but @xmath471 we note that the second term in the definition of @xmath472 is included solely to make @xmath65 real . *
if @xmath32 and we set @xmath473}(\xi ) + \chi_{[n+1,n+1+\frac{1}{n}]}(\xi))\ ] ] and @xmath474 , @xmath469 , then @xmath465 is a sequence such that @xmath470 but @xmath475 the second was considered in [ s : northip ] as part of the proof of theorem [ t : northip ] , and thus reproduces a weaker version of that result .
we now carry out a proof of the first case to establish theorem [ t : nearsouthip ] .
since @xmath476\sphat(\xi ) = \int_0^t e^{-i(t - t')\xi^2}[uu_{n,0}(t')w(n_{n,0},0)(t')]\sphat(\xi ) \ , dt'\ ] ] we need to examine @xmath477\sphat(\xi ) = \int_{\xi_1 } e^{-it'\xi_1 ^
2}\hat{u}_{n,0}(\xi_1 ) \cos(t'(\xi-\xi_1 ) ) \hat n_{n,0}(\xi-\xi_1 ) \ , d\xi_1 \\ & = ( e^{-it'n^2}\cos(t'(2n-1))+ \mathcal{o}(t ' ) ) \int_{\xi_1 } \hat{u}_{n,0}(\xi_1 ) \hat n_{n,0}(\xi-\xi_1 ) \ , d\xi_1 \end{aligned}\ ] ] by the support properties of @xmath478 and @xmath479 . directly evaluating
the convolution gives @xmath480 where @xmath481 consists of two triangular step functions , each of height @xmath482 and width @xmath483 , centered at @xmath484 and @xmath485 , respectively .
specifically , @xmath486 @xmath487 we have by the support properties of @xmath488 and @xmath489 , and @xmath490\sphat(\xi)\\ & = \begin{aligned}[t ] & + n^{1-k - s } g_1(\xi ) \int_0^t e^{-i(t - t')(n-1)^2}(e^{-it'(n-1)^2 } + e^{-it'(n^2 + 2n-1 ) } ) \ , dt '
\\ & + n^{1-k - s } g_2(\xi)\int_0^t e^{-i(t - t')(-3n+1)^2}(e^{-it'(n-1)^2 } + e^{-it'(n^2 + 2n-1 ) } ) \ , dt'\\ & + n^{1-k - s}g(\xi)\mathcal{o}(t^2 ) \end{aligned}\end{aligned}\ ] ] evaluating each component separately gives @xmath491\sphat(\xi ) = n^{1-k - s}g_1(\xi)e^{-it(n-1)^2}t + n^{1-k - s}g(\xi)(\mathcal{o}(t^2 ) + \mathcal{o}(n^{-1}))\ ] ] thus , provided @xmath395 is chosen small and @xmath46 sufficiently large , the first term is pointwise dominant , giving @xmath492 completing the proof .
bjrn birnir , carlos e. kenig , gustavo ponce , nils svanstedt , and luis vega , _ on the ill - posedness of the ivp for the generalized korteweg - de vries and nonlinear schrdinger equations _ , j. london math .
( 2 ) * 53 * ( 1996 ) , no . 3 , 551559 . mr mr1396718 ( 97d:35233 ) j. bourgain , _ fourier transform restriction phenomena for certain lattice subsets and applications to nonlinear evolution equations .
i. schrdinger equations _ , geom .
* 3 * ( 1993 ) , no . 2 , 107156 .
( 95d:35160a ) michael christ , james colliander , and terrence tao , _ asymptotics , frequency modulation , and low regularity ill - posedness for canonical defocusing equations _
* 125 * ( 2003 ) , no . 6 , 12351293 . mr mr2018661
( 2005d:35223 ) bo ling guo , _ existence and nonexistence of global solution , soliton solution for a class of the systems of generalized zakharov equations _ , nonlinear evolutions ( balaruc - les - bains , 1987 ) , world sci . publishing , teaneck , nj , 1988 , pp .
mr mr1031546 ( 90m:35169 ) | ginibre - tsutsumi - velo ( 1997 ) proved local well - posedness for the zakharov system @xmath0 for any dimension @xmath1 , in the inhomogeneous sobolev spaces @xmath2 for a range of exponents @xmath3 , @xmath4 depending on @xmath1 . here
we restrict to dimension @xmath5 and present a few results establishing local ill - posedness for exponent pairs @xmath6 outside of the well - posedness regime .
the techniques employed are rooted in the work of bourgain ( 1993 ) , birnir - kenig - ponce - svanstedt - vega ( 1996 ) , and christ - colliander - tao ( 2003 ) applied to the nonlinear schrdinger equation . |
the brightness variations of rv tauri stars are characterized by alternating primary and secondary minima .
the alternation is not strictly regular ; the order of the deep and shallow minima shows occasional reversals ( * ? ? ?
* ( wallerstein 2002 ) ) . according to the most likely explanation the alternating light variation is caused by the 2:1 resonance between the fundamental and the first - overtone mode ( * ? ? ? * ( fokin 1994 ) ) . in this near - resonance state
the system may display low - dimensional chaotic behaviour , as proposed by ( * ? ? ?
* buchler & kovcs ( 1987 ) ) and ( * ? ? ?
* buchler et al . ( 1996 ) ) .
the chaos hypothesis is also supported by the observed interchanges of deep and shallow minima .
similar interchanges were recently discovered in the order of the period - doubled pulsation cycles in rr lyrae stars observed by the _ kepler
_ space telescope ( * ? ? ?
* ( szab et al . 2010 ) ) . in these cases
the period doubling of the fundamental mode is caused by a 9:2 resonance with the ninth overtone . by this time
the four - year long photometric data of _ kepler _ have become available to study the nature of the interchanges of the alternating maxima in these rr lyrae stars and to compare to the similar phenomenon in rv tauri variables .
we analysed three rv tauri stars ( tt oph , uz oph , and u mon ) and three rr lyrae stars ( rr lyr , v808 cyg , v355 lyr ) that clearly show the interchanges .
rv tauri light curves are based on several data sources : visual and photometric measurements of amateur astronomers in the aavso , afoev and vsolj databases , data from the asas ( * ? ? ?
* ( pojmaski 1997 ) ) , nsvs ( * ? ? ?
* ( woniak et al . 2004 ) ) and catalina sky surveys ( * ? ? ?
* ( drake et al . 2009 ) ) , as well as the observations of the _ hipparcos _ satellite ( * ? ? ?
* ( esa 1997 ) ) and the optical monitoring camera aboard the integral satellite ( * ? ? ?
* ( mas - hesse et al .
2003 ) ) . in the case of rr lyrae stars we used
all the available _ kepler _ photometric data between the q1-q15 runs . to determine the extrema of the cycles we applied least - squares parabola and cubic spline fitting .
we followed the method used by ( * ? ? ?
* molnr et al . ( 2012 ) ) by connecting the even- and odd - numbered cycles respectively .
this representation significantly helps to recognize the epoch of the interchanges .
all three rv tauri stars show interchanges on short and long timescales .
the characteristic interval between two interchanges are 14 - 21 cycles for tt oph , and 3 - 9 cycles for u mon .
the longer timescale denote 112 - 119 cycles for tt oph and 39 - 51 for u mon .
interestingly , the light curve of tt oph remains in the same phase during the first 249 cycles .
u mon also shows a section of 136 cycles without interchanges .
the total number of interchanges is low for uz oph , but a short timescale is suspected around 5 - 25 cycles and a long one around 45 - 65 cycles .
the timescales of the interchanges of rr lyrae stars are not divided into two groups ; instead they spread to a broad interval between 3 and 39 , 86 or 60 cycles for v808 cyg , v355 lyr and rr lyr respectively .
most of the intervals are around 6 - 30 cycles for all rr lyrae stars .
we note that determination of the timescales of the interchanges were not unambiguous for v355 lyr and rr lyr , where the alternation seemingly disappears occasionally , making the lengths between the interchanges uncertain up to a few cycles .
no periodicities were recognized in the occurrences of the interchanges for any of the analysed stars .
rv tauri stars can stay for long periods of time in one phase without interchanges .
similar behaviour was not observed in the rr lyrae stars .
the blazhko periods are @xmath0170 cycles long for v808 cyg , @xmath066 cycles for v355 lyr and @xmath070 cycles for rr lyr : the time intervals between the interchanges do not show any correlation with the blazhko periods .
this work has been supported by the hungarian otka grant k83790 , the mb08c 81013 mobility - grant of the mag zrt . , the `` lendlet-2009 '' young researchers programme of the hungarian academy of sciences and the ktia urkut_10 - 1 - 2011 - 0019 grant .
the _ kepler _ team and the _ kepler _ guest observer office are recognized for helping to make the mission and these data possible .
we acknowledge with thanks the variable star observations from the aavso international database , the afoev database , operated at cds , france , and the vsolj database contributed by observers worldwide .
kk is grateful for the support of her marie curie fellowship ( 255267- sas - rrl , fp7 ) . | we characterized the time intervals between the interchanges of the alternating high- and low - amplitude extrema of three rv tauri and three rr lyrae stars . |
a proper phase transition takes place only in the idealised limit of an infinite number of interacting degrees of freedom .
although this limit is never realised in the laboratory ( let alone in numerical simulations ) , everyday experience suggests that macroscopic samples are infinite for all practical purposes .
spin glasses @xcite are an exception .
the problem lies in their sluggish dynamics at the critical temperature and below .
the system remains for very long times , or forever , out of equilibrium .
in fact , letting the system relax for about one hour , the spatial size of the glassy magnetic domains is ( at most ) of the order of one hundred lattice spacings @xcite .
it has become clear lately that , in order to interpret experimental data in spin glasses , the relevant equilibrium properties are those of systems of size similar to that of the experimentally achievable coherence length @xcite . phase transitions on finite systems are actually crossover phenomena describable through the well known theory of finite size scaling ( see e.g. @xcite ) .
however , a conspicuous feature of disordered systems ( and most notably , of spin glasses ) is to undergo strong sample - to - sample fluctuations in many thermodynamic properties .
it is thus natural to ask questions about the probability distribution , induced by the disorder , of the various physical quantities .
typically the size of these fluctuations decreases when enlarging the size of the equilibrated system ; if we wish to have hints about their possible relevance in experimental systems , it is important to know the rate at which fluctuations decrease with system size . this is particularly important if we are to study dynamical heterogeneities @xcite in spin glasses @xcite . in particular ,
a relevant but elusive physical quantity ( potentially relevant to analyse dynamical effects close to the phase transition ) is the finite - system pseudo - critical temperature .
our scope here is to characterise its statistical properties in spin glasses .
this problem has been extensively studied and is well understood for finite - size weakly bond - disordered spin models , below the upper critical dimension @xmath0 . for a system of size @xmath1 and a disorder sample @xmath2 , one can define a pseudo - critical temperature @xmath3 as the location of the maximum of a relevant susceptibility : this definition is clearly non unique , but all sensible definitions lead to the same scaling behaviour as @xmath4 . according to the harris criterion @xcite , a major role @xcite is played here by the value of the thermal critical exponent of the pure system , @xmath5 .
if @xmath6 the disorder is irrelevant , the value of @xmath7 is not modified by the disorder ( i.e. @xmath8 ) , and the width @xmath9 of the probability distribution of the pseudo - critical temperature , defined as @xmath10 , where @xmath11 denotes the disorder average , behaves as @xmath12 as expected naively @xcite . in such a situation in the infinite volume limit the ( disorder induced ) fluctuations of @xmath3 are negligible with respect to the width of the critical region and to the finite - size shift of @xmath13 , that both behave like @xmath14 . in the other case , when @xmath15 .
disorder is relevant , the value of @xmath7 for the disordered model is different from @xmath5 and obeys @xcite the bound @xmath16 . in this case
@xmath9 , the width of the critical region , and the finite - size shift of @xmath13 behave like @xmath17 .
the behaviour @xmath18 that would be naively dominant is destroyed by the disorder .
the case of weakly bond - disordered spin models above the upper critical dimension needs a very careful analysis , as shown in @xcite . to the best of our knowledge
, the distribution of the pseudo - critical temperature in finite size spin glass models has not been studied numerically before : this is the object of the present note .
recent analytical work has predicted @xmath19 ( where @xmath20 is the number of spins , i.e. the system volume ) for the ( mean field ) sherrington - kirkpatrick model ( sk ) for spin glasses @xcite : we establish in this note that the realised scenario is indeed different . very recently , while this work was being completed , ref .
@xcite has also tried ( and failed ) to verify numerically the analytical predictions of @xcite . a former attempt to analyse
numerically the distribution of the pseudo - critical temperature in the sk model was useful to investigate the numerical techniques of choice @xcite . here , we present numerical results both for the three - dimensional ( @xmath21 ) and the mean - field sk spin glass models . in the @xmath21 case ,
we show that the probability distribution of pseudo - critical temperatures verifies finite size scaling . from the scaling of this distribution
we obtain a precise estimate of the critical temperature and of the critical exponent for the correlation length , @xmath7 . on the other hand
, we find that for the mean - field spin glass @xmath22 ( in agreement with analytical findings for the scaling with @xmath20 of disordered - averaged quantities in mean field models @xcite ) .
since this is in plain contradiction with the results of ref .
@xcite , we briefly revisit their analytical argument and show where the error in @xcite stems from .
we also believe that a second analytic conclusion of @xcite , stating that the @xmath23 is distributed according to a tracy widom probability law , is based on very shaky grounds , and we will give hints of the fact that it is not substantiated numerically . our first step is to define a pseudo - critical temperature for a given finite - size sample .
random bond and site diluted models allow @xcite a straightforward definition of @xmath3 as the location of the maximum of the relevant susceptibility @xmath24 . in our case
the situation is more complex ( even if , as we will see , the analysis of the spin glass susceptibility @xmath25 will be very useful and revealing ) . here the relevant diverging quantity is @xcite @xmath25 @xmath26 ( where the @xmath27 are the local spin variables ) that is of order @xmath20 in the whole low temperature phase .
@xmath25 is a continuously decreasing function of the temperature , and has no peak close to @xmath13 : this requires , as we will discuss in the following , a slightly more sophisticated analysis in order to extract a pseudo critical temperature .
an alternative and simpler procedure is very straightforward : let us introduce it first .
we first assume ( as done in @xcite ) that for a given disorder sample @xmath2 the finite - size scaling of an observable @xmath28 of dimension @xmath29 is @xmath30 , where @xmath31 is a @xmath32 and @xmath2 independent finite - size scaling function : the whole disorder sample dependence is encoded inside the pseudo - critical temperature @xmath3 .
this is in fact an approximation since the scaling function has a residual @xmath2 dependence @xcite .
we next build dimensionless combinations of operators : we call them @xmath33 , and they are build in such a way to scale as @xmath34 where @xmath35 is a @xmath36 , @xmath32 and @xmath2 independent finite size scaling function .
a familiar looking combination is the single sample pseudo - binder cumulant @xmath37 ( notice that this is defined for a given disorder realisation ) : the genuine binder cumulant is defined as @xmath38 . for sensible choice of @xmath39
the solution @xmath40 of the equation @xmath41 , with a disorder independent constant @xmath39 , is a proxy of the pseudo - critical temperature , namely @xmath42 with an @xmath32 and @xmath2 independent constant @xmath43 .
for example for a function @xmath44 that in the infinite volume limit is zero in one phase and one in the other phase any constant @xmath39 in the interval @xmath45 $ ] will do : it is wise , in order to minimise the corrections to scaling to choose a legitimate value for @xmath39 such that @xmath40 is typically inside the critical region ( the value of @xmath40 depends on @xmath39 and on the choice made of a dimensionless combination @xmath46 ) .
we are also able to use @xmath47 for determining @xmath3 : in this way we are able to monitor a quantity that diverges in the infinite volume limit , and to use it to extract a pseudo - critical temperature .
the approach used to define @xmath3 in this case is based on the same technique : we compare the single sample spin glass susceptibility to a value close to the average spin glass susceptibility at the critical temperature on a given lattice size .
this measurement is a good proxy for the direct measurement of the position of an emerging divergence .
we have applied these ideas to the edwards - anderson model in @xmath21 and to the sk model .
we used existing data obtained by the janus collaboration on systems with @xmath48 to @xmath49 for the @xmath21 ea model @xcite , and from @xcite for the sk model with @xmath20 ranging from @xmath50 up to @xmath51 . in both cases
the quenched random couplings can take the two values @xmath52 with probability one half .
the layout of the rest of this work is as follows . in sect .
[ sect : ea ] we discuss our numerical methods , and we present our results for the edwards - anderson model . an analogous analysis for the sk model is presented in sect .
[ sect : sk ] .
this study is complemented in sect .
[ sect : tap ] with our analysis of the analytically predicted scaling for the distribution of pseudo - critical temperatures .
we also present in sect .
[ sect : pd ] an analysis of the distribution function of the pseudo critical points . finally , we give our conclusions in sect .
[ sect : conclusions ] .
the hamiltonian of the model is @xmath53 where the sum runs over the couples of first neighbouring sites of a @xmath21 simple cubic lattice with periodic boundary conditions .
the ising @xmath54 spin variables can take the two values @xmath52 and the couplings are quenched binary variables that can take the value @xmath52 with probability one half . in order to analyse the single sample pseudo - critical temperatures of edwards - anderson @xmath21 systems we need to construct several dimensionless quantities .
we define the fourier transform of the replica - field @xmath55 ( @xmath56 and @xmath57 are two real replicas , i.e. two independent copies of the system evolving under the same couplings , but with different thermal noise ) @xmath58 which we use to construct the two - point propagator @xmath59 where @xmath60 denotes the thermal average for the sample @xmath2 .
since the smallest momentum compatible with the periodic boundary conditions is @xmath61 we define @xmath62 and @xmath63 similarly , the second smallest momentum is given by @xmath64 and by the two other possibilities : we use it to define @xmath65 .
we consider the following dimensionless quantities : @xmath66^{1/2}\;,\label{eq : xi } \\
b^j & \equiv \frac{\langle q^4\rangle_j}{\langle q^2\rangle^2_j}\;,\label{bj } \\
b_g^j & \equiv \frac { \sum_i \bigl\langle\bigr[\phi(\boldsymbol k_1^{(i ) } ) \phi(-\boldsymbol k_1^{(i)})\bigr]^2\bigr\rangle_j}{[g^j(\boldsymbol k_1)]^2}\;,\\ r_{12}^j & \equiv \frac{g^j(\boldsymbol k_1)}{g^j(\boldsymbol k_2)}\;.\label{eq : r12}\end{aligned}\ ] ] in eq .
[ bj ] we have used the global spin - overlap , which is defined as @xmath67 where @xmath1 is the total number of spins .
let us start by considering the sample - averaged observables ( which we denote by dropping the super - index @xmath2 ) .
up to scaling corrections , they do not depend @xmath32 at the critical point , @xmath68 .
here @xmath69 , that is a good approximation to @xmath70 , and we plot @xmath71 for three samples with @xmath49 .
the red curve has exactly one solution ( in our range ) for @xmath72 , which defines its @xmath73 : this is the normal behaviour of this observable .
the green sample has three solutions , so we pick the largest one ( green dot ) .
the blue sample has @xmath74 : in this case we take the lowest simulated temperature @xmath75 as an upper bound for the pseudo - critical temperature ( this ignorance will not affect our estimate for the median ) .
error bars are included in the plot and are very small .
[ fig : tcj ] ] for each value of @xmath32 we can search for the temperature @xmath76 such that @xmath77 then , provided we are not very far from the scaling region ( so that @xmath39 is not too different from @xmath78 ) , we expect that @xmath79 where we have included the first corrections to scaling .
we can use this same approach to define a single - sample critical temperature @xmath80 .
let us choose a fixed value of @xmath39 close enough to the value @xmath78 defined from the sample average at the critical point .
for each sample we use cubic splines to determine @xmath81 such that @xmath82 for some samples the @xmath33 turn out not to be monotonic : there can be several solutions to this equation . in those cases
we simply pick the largest solution .
this process is illustrated in figure [ fig : tcj ] .
the motivation for this choice is simple .
the physical meaning of a pseudo - critical temperature is a characteristic temperature that separates the paramagnetic phase from the low - temperature one .
indeed , any temperature @xmath40 , solving the equation @xmath83 , is a temperature where non - paramagnetic behaviour has already arisen .
therefore , only the largest @xmath40 makes sense as a divider among both phases .
in fact , the non - monotonic behaviour of @xmath33 may be due to other reasons ( in particular , temperature chaos ) , unrelated to the paramagnetic / spin - glass phase transition . in any case , our definition will be justified _ a posteriori _
, on the view of the simplicity of the emerging physical picture .
the values of @xmath40 have a very wide probability distribution .
for a few disorder samples the solution of eq .
falls out of our simulated range of temperatures and we only obtain an upper or ( less frequently ) a lower bound ( see the blue curve in figure [ fig : tcj ] ) . in this situation the arithmetic average of the @xmath40 is not well defined : we consider instead the median temperature , that we denote by @xmath84 . since , by definition ,
the median does not change as long as the proportion of samples without a solution is less than 50% , this is a robust estimator in these circumstances ( we are well below this limit for all the cases considered , the typical proportion being about @xmath85 ) . from fig .
[ fig : tcj ] it is also clear that the statistical uncertainty over the determination of @xmath40 in a given sample is very small as compared to the size of sample to sample fluctuations . in analogy with the sample averaged case , we make the ansatz @xmath86 where we have ignored sub - leading corrections . a fit to this equation
would , in principle , yield the values of @xmath87 and @xmath7 .
however , for a fixed value @xmath39 , we do not have enough degrees of freedom to determine simultaneously @xmath7 and @xmath87 .
following the approach of @xcite , we get around this problem by considering @xmath88 values of @xmath39 at the same time : this allows us to fit at the same time all the resulting @xmath89 , with fit parameters @xmath90 : in other words we force the @xmath91 obtained for different @xmath39 values to extrapolate to the same @xmath87 with the same exponent .
this procedure may seem dangerous , since we are extracting several transition temperatures from each of the @xmath33 , that are correlated variables .
however , the effect of these correlations can be controlled by considering the complete covariance matrix of the data .
the set of points @xmath92 are labelled by their @xmath32 and their @xmath39 : we have data for @xmath93 different values of @xmath32 , with @xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 ( @xmath99 samples in all cases but for @xmath49 , where we have @xmath100 ) .
we also select @xmath88 values of @xmath101 in the critical region ( @xmath102 , @xmath103 , @xmath104 ) .
the appropriate chi - square estimator is @xmath105 \sigma^{-1}_{(ia)(jb ) } \bigl [ \tilde t_{y_j}^j(l_b ) - t_\text{c } - a_{y_j } l_b^{-1/\nu}\bigr]\;,\ ] ] where @xmath106 is the covariance matrix of the set of @xmath40 , which we compute by a bootstrap approach @xcite ( each point of the set is identified by @xmath32 and @xmath39 ) : it is a block - diagonal matrix , since the data for different @xmath32 values are uncorrelated .
thus far we have considered just a single observable @xmath44 , for different values of @xmath39 . since
the fitting function is the same for the different @xmath44 we have selected , with common @xmath87 and @xmath7 , and only amplitudes differ , we can include in the same fit data for the four dimensionless quantities ( [ eq : xi][eq : r12 ] ) , considering several values of @xmath39 for each . in order to simplify the notation , from here on
we shall denote our set of points in the fit as @xmath107 , taking @xmath108 as labelling both the observable @xmath44 and the height @xmath39 ( so that it will range from @xmath109 to @xmath110 ) .
we can use the usual disorder averaged spin - glass susceptibility @xmath111 to arrive at yet another determination of the single - sample critical temperature , with the definition @xmath112 with @xmath39 close to one , and we expect @xmath113 to follow the same scaling behaviour of : we include the values of @xmath113 in the global fit to the individual pseudo critical temperatures . in this case
the pathologies that affect the analysis of other single sample observables are far less frequent . computed with different observables .
the parameters in the fit are the common extrapolation point @xmath87 , the common exponent @xmath7 and an individual amplitude for each set of data points .
we include in the fit the data for @xmath114 .
[ fig : fitm ] ] we show the results of this combined fitting procedure in figure [ fig : fitm ] , where we have included data for the four dimensionless ratios @xmath115 , @xmath116 , @xmath117 and @xmath118 , using three values of @xmath39 for each ratio ( we plot the three fits for the same observable with the same colour ) .
we have also used the data for @xmath119 , with @xmath120 ( we select the value of @xmath87 reported in @xcite ) .
we have discarded the @xmath48 data , which showed strong corrections to the leading scaling of .
the best fit gives @xmath121 with @xmath122 for @xmath123 degrees of freedom ( giving a @xmath28 value of @xmath124 ) .
this results nicely agree with the determination of @xcite , @xmath125 ref .
@xcite includes corrections to scaling , as in , with @xmath126 .
@xcite gives a comprehensive list of estimates for @xmath127 and @xmath7 . in order to take these scaling corrections into account and to include the @xmath48 data in the fit ,
we can therefore rewrite as @xmath128 where again we use the same @xmath129 parameter for all the observables and all values of @xmath39 .
unfortunately , our numerical data are not precise enough to allow a reliable determination of @xmath129 , @xmath7 and @xmath87 at the same time ( the resulting error in @xmath129 would be greater than @xmath130 ) .
we have been able to check consistency of our approach by taking the values of @xmath7 and @xmath129 from @xcite and fitting only for @xmath87 and for the amplitudes , including now the data for @xmath48 .
the resulting best fit gives @xmath131 , with @xmath132 for @xmath133 degrees of freedom ( @xmath28 value : @xmath134 ) : this is a satisfactory check of consistency .
the results we have discussed make us confident of the fact that our determination of the single - sample critical temperatures yields reasonable results .
we can now take the analysis one step further and consider the width of the distribution of @xmath80 .
we consider the two temperatures @xmath135 and @xmath136 such that @xmath137 the value @xmath138 is such that the temperature interval @xmath139 $ ] defines the same probability as an interval of two standard deviations around the mean for a gaussian probability distribution .
we define the width @xmath140 as @xmath141 , computed using @xmath142 , @xmath143 and @xmath144 .
[ fig : fitwidth ] ] the simplest ansatz for the scaling behaviour of @xmath140 is @xmath145 in principle we could repeat the global fitting procedure that we have applied to the medians @xmath146 .
unfortunately , not all the observables that we considered in figure [ fig : fitm ] can be used to analyse @xmath140 , since the distribution of some of them is too wide , so that the critical temperature of too many samples falls out of our simulated range of @xmath36 , and the width defined in eq .
[ eq : fitdelta ] is undefined .
because of that we analyse @xmath140 by only using
the @xmath40 derived from @xmath142 , @xmath143 and @xmath144 .
the corrections to scaling are now stronger than for the median @xmath91 , so that we can not obtain a good fit to leading order even if we discard the data for @xmath48 .
using once again as an input the critical exponents from @xcite and fitting for the amplitudes we obtain a very good fit with @xmath147 for @xmath148 degrees of freedom ( @xmath28 value : @xmath149 ) .
the results of the best fit are plotted in figure [ fig : fitwidth ] . according to the ansatz of eq .
[ fss ] , the width of @xmath40 must be equal to the width of @xmath3 ( since @xmath150 is @xmath2 independent ) : the width of @xmath40 should accordingly be @xmath39 independent .
figure [ fig : fitwidth ] shows that this is indeed the case for @xmath151 , but less so for @xmath152 : as mentioned before the disorder independence of the scaling function in eq .
[ fss ] is only approximate @xcite .
the hamiltonian of the sherrington - kirkpatrick mean field model is @xmath153 where the sum runs over all couples of spins of the system , the ising @xmath154 spin variables can take the two values @xmath52 and the couplings are quenched binary variables that can take the value @xmath52 with probability one half .
our analysis of the mean field sk model is very similar to the one we have discussed in the case of the edwards - anderson model .
here we have seven values of the system size @xmath20 ( of the form @xmath155 , with @xmath156 ranging from @xmath157 to @xmath158 : in all cases we have @xmath159 disorder samples but for @xmath160 where we have @xmath161 and for @xmath162 where we have @xmath163 ) , that makes the fitting procedure stable .
it is also of use the fact that in this case the value of the ( infinite volume ) critical temperature , @xmath164 is known exactly .
we consider three definitions of @xmath13 : * the one based on the single sample binder cumulant of eq .
[ bj ] , @xmath165 ; * one based on the low order cumulant @xmath166 and the single sample quantity @xmath167 , i.e. @xmath168 ; * one based on the spin - glass susceptibility @xmath169 . for a given value of @xmath20
, we use for the left hand sides @xmath170 , @xmath171 and @xmath172 the values measured for this same size , at @xmath173 .
we solve eq . by using a simple linear interpolation .
like for the ea model , in some cases eq
. can have more than one solution .
we again choose the largest solution , which in the case of sk turns out to be always the one closer to the infinite volume value @xmath13 .
in a few cases , for small values of @xmath20 , the equation has only solutions outside the range of temperatures that was used in the parallel tempering monte carlo simulation ( @xmath174 ) .
we fix this problem again by basing our statistical analysis on the median of the distribution and on the definition of the width given by eq .
[ width ] .
it turns out that these pathological cases are less numerous for the sk model than for the ea model , and that the width given by eq .
[ width ] is always defined . in terms of the number of sites @xmath20 of the sk fully connected lattice the ansatz of eq .
[ eq : fitm ] becomes @xmath175 we show in fig . [
fig : fitmsk ] the data for the median of the distribution of @xmath176 as a function of @xmath177 , together with the results of two best fits .
we first notice that the data are well compatible with the fact that in the @xmath4 limit @xmath164 .
the first fit is a linear fit to the form @xmath178 ( with @xmath179 and @xmath180 with @xmath181 degrees of freedom ) .
this is a good fit for the large systems , but it fails below @xmath182 .
we also show a ( very good ) best fit including the next to leading corrections , with an exponent @xmath183 ( including all @xmath20 values , @xmath184 with @xmath185 degrees of freedom ) .
the analysis of the data for @xmath186 and @xmath187 leads to the same conclusions : here however the leading term ( @xmath188 ) has a small coefficient and the effect of the next to leading term is stronger . in conclusion
our data are in excellent agreement with an asymptotic @xmath189 behaviour for the median of the distribution .
as function of @xmath177 , together with the results of two best fits : a leading order fit , and a fit including a higher order correction ( see text ) . ]
we show in fig . [
fig : fitwidthsk ] the width of the distribution of @xmath176 as a function of @xmath177 , together with the results of two fits , namely @xmath190 , and @xmath191 respectively .
the data are well compatible with @xmath192 in the limit @xmath4 , as expected .
the leading order fit , including @xmath179 , has a @xmath193 with @xmath181 degrees of freedom .
the two - parameter fit gives an excellent representation of the data ( with a @xmath194 with @xmath185 degrees of freedom ) including now the @xmath195 and @xmath162 points .
very similar results are obtained for @xmath152 and @xmath196 . in conclusion our finite size , numerical analysis of the sk model strongly support an asymptotic @xmath197 scaling behaviour for the width of the distribution of the pseudo - critical temperatures .
as function of @xmath198 , together with the results of two fits : a leading order fit , and a fit including a higher order correction ( see text ) . ]
our results are in contradiction with the claim made in @xcite that the width of the finite - size fluctuations due to quenched disorder of the critical temperature of the sk spin glass scales like @xmath199 . we show here that we can give support to our numerical finding by means of a very simple scaling argument . the sg susceptibility can be computed from the tap free energy @xcite as @xmath200 where @xmath201 is the hessian of the tap free energy at the relevant minimum . if we seat deep in the paramagnetic phase , the only relevant minimum of the tap free energy is @xmath202 for all @xmath203 .
note that in the pseudo - critical region , where @xmath204 , is not at all obvious that the such tap solution is the relevant one ( for instance a sub - extensive set of sites of size @xmath205 , with @xmath206 , could have non - vanishing @xmath207 , or maybe one could have for all sites @xmath208 , with @xmath209 ) : the following discussion is relevant only in the paramagnetic phase and for system sizes so large that @xmath210 .
it was shown some time ago @xcite that , at @xmath211 , the smallest eigenvalue of the hessian at the fully - paramagnetic tap solution is of order @xmath199 . in ref .
@xcite it has been argued that @xmath212 where @xmath213 is a random variable that , in the limit of large @xmath20 , converges in distribution to a tracy - widom random - variable @xcite . in particular , note that at the critical temperature @xmath214 the @xmath199 scaling is recovered .
now , only for the purpose of discussing the crudest features of the scaling laws , let us assume that @xmath215 is dominated by the contribution of the smallest eigenvalue : @xmath216 ^ 2 + \beta \phi^{(n)}_j \right)^2}\,.\label{eq : sk - bad}\end{aligned}\ ] ] the analysis of ref .
@xcite is based entirely on eq .
( [ eq : sk - very - bad ] ) .
now , note that eq .
( [ eq : sk - bad ] ) implies that interesting behaviour appears only when @xmath217 : this makes sensible to replace @xmath218 with @xmath219 . at this point ,
an implication emerges for the scaling with @xmath20 of the average susceptibility .
we have that @xmath220 where the scaling function @xmath221 has the form @xmath222 in the above expression @xmath223 is the tracy - widom probability density function .
( [ eq : g - def ] ) is not acceptable for two main reasons : * the function @xmath224 in eq.([eq : g - def ] ) is ill - defined , as the integrand has a non - integrable singularity at ; * if one devises some regularisation procedure , dimensional analysis would indicate that @xmath225 in the limit of large @xmath226 .
however , in order to recover the correct critical divergence @xmath227 , one obviously needs @xmath228 .
the solution of these two caveats is of course in the fact that the initial assumption , @xmath229 , is incorrect .
the contribution of @xmath230 eigenvalues is crucial in order to recover the correct scaling behaviour @xmath228 .
thus , the only lesson that we may take from this oversimplified analysis is that the @xmath2-dependent sg susceptibility will probably scale as @xmath231+\psi_j^{(n)}\right)\,.\ ] ] this is exactly the scaling ansatz we made at the beginning , where @xmath232 is some random - variable that ( in distribution ) remains of order @xmath109 in the large-@xmath20 limit .
this result is consistent with our numerical findings . merely rewriting eq .
( [ eq : sk - very - bad ] ) as eq .
( [ eq : sk - bad ] ) suffices to make it obvious that the asymptotic statement in ref .
@xcite is incorrect : the width of the distribution of the pseudo - critical temperatures scales with @xmath198 .
indeed , if as done in ref .
@xcite , one simply derives in eq .
( [ eq : sk - very - bad ] ) with respect to @xmath233 in order to get the maximum of the susceptibility , one finds that @xmath234 .
but at such value of @xmath235 we have that @xmath236 , in other words , at the scale of @xmath237 eq .
( [ eq : sk - very - bad ] ) predicts an essentially constant behaviour , hence the supposed maximum of the susceptibility ( recall that at such value of @xmath233 the fully paramagnetic tap minimum is probably no longer the relevant one ) has no physical meaning .
we have discussed in some detail about the features of the pseudo - critical , sample dependent temperatures in the mean field sk theory , and we have determined their scaling properties .
the next step , that we present here , is in the study of their probability distribution .
unfortunately , there is not any clean analytical prediction for the shape that this distribution function should take in the large-@xmath20 limit . the only proposal known to us
was put forward in ref .
@xcite : when properly scaled , the pseudo - critical temperatures should follow a tracy - widom ( tw ) distribution , in our case for the gaussian orthogonal ensemble ( goe ) . unfortunately , as we have explained in sect .
[ sect : tap ] , the reasoning leading to that prediction is flawed ( although , on the long run , the prediction itself could be correct ) .
lacking an analytical guidance , we will simply check whether our numerical data can be described by either a tw distribution , or by the ubiquitous gaussian distribution . within our limited statistics and system sizes ,
the two distributions turn out to be acceptable ( the gaussian hypothesis fits slightly better our data , but a tracy - widom hypothesis is certainly consistent as well ) . given the preliminary nature of this study , we shall restrict ourselves to the simplest determination of pseudo - critical temperatures , the one coming from the spin - glass susceptibility .
let us start by noticing that the difference of a gaussian distribution @xmath238 and a tw distribution for the goe @xmath239 is indeed very small .
we show in figure [ fig : tw1 ] both a gaussian and a tw distribution with zero average and variance equal to one : it is clear that they are very similar .
some numerical values can be of help . in the case of zero average and unitary variance
a gaussian has a fourth moment equal to @xmath240 , as opposed to @xmath241 for a tw .
the gaussian is symmetric and has zero skewness , while the tw distribution has a small asymmetry , with a skewness equal to -0.29 , the gaussian has a kurtosis equal to zero , while a tw has a kurtosis equal to @xmath242 .
we will use these numerical remark at the end of this section for sharpening the outcome of our quantitative analysis .
it is clear that in this situation , where the two target distributions are very similar , one has to keep under very strong control finite size effects , that could completely mask the asymptotic behaviour .
it is important to notice that the effects we are looking at characterise not only the tails but the bulk of the distribution . ,
defined in equation [ eq : wedoitbetter ] , versus @xmath198 , and our best fit including the first scaling corrections .
( middle ) : @xmath243 , defined in equation [ eq : wedoitbetter ] , versus @xmath198 , and our best fit . here
the fitting function only includes the leading term , since this form already gives a good value for @xmath244 .
( bottom ) : collapse of the edf for different @xmath20 values and of the theoretical distribution function , as described in the text . here
we show the case of a gaussian distribution . ] , but for a tw probability distribution . ]
let us give the basic elements of our approach .
we consider the value of the pseudo - critical inverse temperatures computed from the spin glass susceptibility for different @xmath20 values .
we try to verify if a relation of the form @xmath245 ( where @xmath246 is a random variable with null expectation values and unit variance , gaussian or tw like ) can account for our numerical data and , if yes , to determine the scaling behaviour of @xmath247 and of @xmath248 .
we now define the variable @xmath249 .
let us assume that for a system of size @xmath20 we have @xmath250 samples , and therefore @xmath250 pseudo - critical temperatures @xmath251 .
the empirical distribution function ( edf ) is then @xmath252 where @xmath253 is the heaviside step function .
note that the parameters @xmath254 and @xmath243 in eq .
[ eq : wedoitbetter ] are unknown a priori .
they will be determined through a fitting procedure ( see eq .
[ eq : dist1 ] below ) .
in particular , they typically take different numerical values if we adopt the tw hypothesis , or the gaussian hypothesis .
we define the distance among the edf and the theoretical distribution as @xmath255 ^ 2\,.\ ] ] in order to get an error estimate we repeat the procedure for @xmath100 bootstrap samples for each value of @xmath20 . in fig .
[ fig : tw2 ] we consider the theoretical hypothesis of a gaussian distribution .
we show in the top part of the figure @xmath254 versus @xmath198 , and our best fit including the first scaling corrections . in the middle frame
we show @xmath243 versus @xmath198 , and our best fit . here
the fitting function only includes the leading term , since this form already gives a good value for @xmath244 . in the bottom frame
we show the collapse of the edf for different @xmath20 values and of the theoretical distribution function , as described in the text . in fig .
[ fig : tw3 ] we show the same data for the hypothesis of a tw distribution . in short , even if the gaussian hypothesis is slightly favoured over the tw one , this analysis does not allow us to decide clearly in one sense or in the other one .
the values of @xmath256 are always comparable among the two cases .
the estimates of @xmath243 are indeed more consistent for the gaussian case , but the difference of the quality of the two fits does not allow us to say a clear , final word . in order to try to sharpen our analysis we have used the cramer - von mises criterion @xcite
. we will not give here technical details ( see , however , endnote @xcite ) , but will only discuss the most important features and the results .
when we normalise the distribution to zero average we and unitary width we introduce a correlations about the values to be tested : also because of that we find a better fit to our needs the two sample formulation of the criterion , with a non - parametric approach , where we have to start by fitting the test statistics ( since we can not use tabulated values , because we are determining @xmath248 and @xmath247 in eq .
[ eq : wedoitbetter ] from our finite - size statistics ) . again , as in our previous analysis , tests do not allow us to select a gaussian or a tw distribution : they are both characterised by very similar levels of significance . and for data from the pseudo binder parameter , versus @xmath257 .
thick straight line is for a gaussian distribution ( where the value is three ) , while the thinner straight line is for a tw distribution . ]
a very simple analysis is maybe the most revealing .
as we have discussed at the start of this section the fourth moment of a normalised gaussian is equal to three , while the fourth moment of a tw distribution is equal to @xmath241 .
we plot in fig .
[ fig : tw4 ] the measured fourth moment of the probability distribution , both for data from @xmath47 and for data from the pseudo binder parameter , versus @xmath257 .
the thick straight line is for a gaussian distribution ( where the value is three ) , while the thinner straight line is for a tw distribution .
again , these data do not allow for a precise statement , but they seem to favour the possibility of a gaussian behaviour ( the data for @xmath47 give maybe the clearer indication ) .
our conclusions is that , given the quality of our data and the sizes of our thermalised configurations , that do not go beyond @xmath160 , a gaussian distribution is favoured , but we can not give a clear , unambiguous answer .
we have presented a simple method to study the probability distribution of the pseudo - critical temperature for spin glasses .
we have applied this method to the @xmath21 ea ising spin glass and to the fully connected sk models .
our results are in excellent agreement with a median of the distribution that behaves asymptotically like @xmath258 ( or @xmath259 for the sk model ) , and a width of the distribution that behaves like @xmath260 ( or @xmath259 for the sk model ) .
the value of @xmath7 we find for the ea model is compatible with state of the art results .
furthermore , even if our number of samples is modest as compared with ref .
@xcite , our determination of @xmath7 and @xmath13 is competitive .
an analysis of the probability distribution of the pseudo - critical inverse temperatures for the sk mean field model does not lead to firm conclusions , but hints to a gaussian behaviour .
we are indebted to the janus collaboration that has allowed us to use equilibrium spin configurations of the @xmath261 edwards - anderson model @xcite obtained by large scale numerical simulations .
ab thanks ccile monthus and thomas garel for discussions at an early stage of the work and , specially , barbara coluzzi for a sustained collaboration on the study of the sk model .
we acknowledge partial financial support from micinn , spain , ( contract no fis2009 - 12648-c03 ) , from ucm - banco de santander ( gr32/10-a/910383 ) and from the dream seed project of the italian institute of technology ( iit ) .
dy was supported by the fpu program ( spain ) .
r. lvarez baos , a. cruz , l. a. fernandez , j. m. gil - narvion , a. gordillo - guerrero , m. guidetti , a. maiorano , f. mantovani , e. marinari , v. martin - mayor , j. monforte - garcia , a. muoz sudupe , d. navarro , g. parisi , s. perez - gaviro , j. ruiz - lorenzo , s. f. schifano , b. seoane , a. tarancon , r. tripiccione and d. yllanes ( janus collaboration ) , j. stat .
2010 , p06026 .
r. lvarez baos , a. cruz , l. a. fernandez , j. m. gil - narvion , a. gordillo - guerrero , m. guidetti , a. maiorano , f. mantovani , e. marinari , v. martin - mayor , j. monforte - garcia , a. muoz sudupe , d. navarro , g. parisi , s. perez - gaviro , j. ruiz - lorenzo , s. f. schifano , b. seoane , a. tarancon , r. tripiccione and d. yllanes ( janus collaboration ) , phys .
lett . * 105 * , 177202 ( 2010 ) .
c. a. tracy and h. widom , comm .
phys . * 177 * , 727 ( 1996 ) ; m. dieng , _ distribution functions for edge eigenvalues in orthogonal and symplectic ensembles : painlev representation _ phd thesis , uc davis ( 2005 ) , arxiv : math/0506586 ; p. deift and d.gioev , comm .
pure appl . math .
* 60 * , 867 ( 2007 ) , arxiv : math - ph/0507023 . given two independent samples of @xmath20 and @xmath263 values respectively of a random variable @xmath226 ,
whose empirical distribution functions are @xmath264 and @xmath265 , and being @xmath266 the empirical distribution function of the sample obtained by combining the two original samples together , the two - samples cramr - von mises statistics is the distance @xmath267 \int_{-\infty}^{\infty}\left[f_n(x)-g_m(x)\right]^2dh_{n+m}(x)$ ] . | using the results of large scale numerical simulations we study the probability distribution of the pseudo critical temperature for the three dimensional edwards - anderson ising spin glass and for the fully connected sherrington - kirkpatrick model .
we find that the behaviour of our data is nicely described by straightforward finite - size scaling relations . |
there has been recently active interest in tests of quantum mechanics@xcite versus local realism in a high efficiency detection limit .
several authors@xcite including ourselves have considered detection schemes quadrature phase homodyne measurements .
such schemes use strong local oscillators and hence have very high detection efficiency@xcite .
this removes one of the current loopholes@xcite and potentially allows a strong test of quantum mechanics@xcite to be performed .
the original idea of gilchrist _ et .
al._@xcite was to use a _ circle _ or pair coherent state@xcite produced by nondegenerate parametric oscillation with the pump mode adiabatically eliminated .
using highly efficient quadrature phase homodyne measurements , the clauser horne strong bell inequality@xcite could be tested in an all optical regime .
a small ( approximately @xmath0 ) but significant theoretical violation was found for this extremely idea system .
while the mean photon number for the system may be low ( approximately @xmath1 ) , the use of homodyne measurements allow a macroscopic current to be detected . in this article ,
we take an unphysical but interesting approach and answer the following questions : * _ given that your detection scheme is a quadrature phase homodyne measurement , what is the optimal input or correlated photon number state to maximize the potential violation ? _ * _ what is the optimal bell inequality to test ?
_ to begin we will restrict our attention to correlated photon number states of the form @xmath2 .
the most well known is simply the nondegenerate parametric amplifier specified by an ideal hamiltonian of the form@xcite @xmath3 where @xmath4 is field amplitude of a nondepleting classical pump and @xmath5 is proportional to the susceptibility of the medium .
@xmath6 are the boson operators for the orthogonal signal and idler modes . after a time @xmath7
, the state of the system is given by ( [ correlatedpair ] ) with @xmath8 specified by @xmath9}\over{\cosh \left[\chi \epsilon \tau\right ] } } \end{aligned}\ ] ] in the quadrature phase amplitude basis this state has a positive wigner function .
hence it can be described as a local hidden variable theory and thus can not violate a bell inequality .
the other source of highly correlated photon number states exists in nondegenerate parametric oscillation . in the limit of very large parametric nonlinearity and high q cavities , a state of the form@xcite @xmath10
can be generated .
here @xmath11 is the size of the circle of the coherent states and @xmath12 is the zeroth order modified bessel function .
equivalently this state can be written in the form of ( [ correlatedpair ] ) with @xmath8 given by @xmath13 this was the state considered by gilchrist _ et .
al._@xcite given the general form of known correlated number states ( [ correlatedpair ] ) , the next fundamental question that should be initially addressed is what we mean by the bell inequality .
a number of bell inequalities exist , and the particular one depends used heavy on your application and experimental setup .
the bell inequalities to be considered in this article are the clauser horne@xcite , the _ spin_@xcite , and the information - theoretic@xcite bell inequality . a detailed derivation of the various inequalities will not be given , the reader is referred to references @xcite . here
we will consider only strong inequalities , that is inequalities where auxiliary assumptions ( not based on local realism ) are not required .
in fig ( [ fig1 ] ) we depict a very idealized setup for general bell inequality experiment .
probably the most well known inequality is the clauser horne strong bell inequality@xcite given by @xmath14 here @xmath15 is the probability that a `` 1 '' results occurs at each analyzer @xmath16 given @xmath17 .
similarly @xmath18 is the probability that a `` 1 '' occurs at a detector while having no information about the second . for many of the actual experimental considerations
an angle factorization occurs so that @xmath19 depends only on @xmath20 .
also @xmath21 and @xmath22 are independent of @xmath17 . in this case
@xmath23 can be simplified to @xmath24 where @xmath25 and @xmath26 .
the second form of the bell inequality ( sometimes referred to as the _ spin _ or original bell inequality ) is@xcite @xmath27 where the correlation function @xmath28 is given by @xmath29 here as discussed above @xmath15 is probability that a `` 1 '' results occurs at each analyzer @xmath16 given @xmath17 .
@xmath30 is probability that a `` 0 '' results occurs at each analyzer @xmath16 , while @xmath31 ( @xmath32 ) is probability that a `` 1 '' ( `` 0 '' ) results occurs at the analyzer @xmath33 and a `` 0 '' ( `` 1 '' ) at @xmath34 . with the angle factorization
given above , the inequality ( [ spin ] ) can be rewritten as @xmath35 our final form of bell inequality to be considered in this article was developed by braunstein and caves@xcite .
this classical information - theoretic bell inequality has the form @xmath36 where @xmath37 here @xmath38 is given by @xmath39 with @xmath40 being the information gained at @xmath34 given the result at @xmath33 is known .
the conditional information is then given by @xmath38 .
the base of the logarithm determines the units of the information ( base 2 for bits , base e for nats ) . for quantum computing purposes ,
this inequality should prove highly useful as it directly deals with information content .
several other bell inequality do exist such as the chsh inequality@xcite , but these are not considered here due to there weaker nature .
auxiliary assumptions are necessary in there derivation which open up several loopholes@xcite .
from ( [ correlatedpair ] ) we need to find the optimal @xmath8 which gives the largest bell inequality violation . before determining the @xmath8 we need to briefly focus our attention on the quadrature phase homodyne measurement .
a quadrature phase - amplitude homodyne measurement @xmath41 at @xmath33 can achieved by combining a signal field ( say @xmath42 ) with a strong local oscillator field ( say @xmath4 ) to form two new fields given by @xmath43 /\sqrt{2}$ ] . here
@xmath44 is a phase shift which allows the choice of particular observable to be measured , for instance choosing @xmath44 as @xmath45 or @xmath46 allows the measurement of the conjugate phase variables @xmath47 and @xmath48 respectively .
the homodyne measurement gives the photocurrent difference as @xmath49 performing a measurement on the quadrature phase amplitude @xmath41 at @xmath33 yields a result @xmath50 which ranges in size and sign .
similarly a measurement on the quadrature phase amplitude @xmath51 at @xmath34 yields a result @xmath52 . for our state given by ( [ correlatedpair ] ) , the probability of obtaining the result @xmath53 is simply @xmath54 where @xmath55 here @xmath56 is the hermite polynomial and @xmath57 is the phase of the local oscillator .
eqn ( [ prob ] ) can be explicitly written as @xmath58 where @xmath59 , that is our expression depends only on the sum of the individual local angles .
the probability given by ( [ prob1 ] ) is for continuous variables .
the majority of the tests of quantum mechanics versus local realism require a binary result .
hence for a given quadrature measurement @xmath60 we classify the result as `` 1 '' if @xmath61 and the mutually exclusive `` 0 '' if @xmath62 . here
we have set the binning window about @xmath63 .
where this binning window is located is quite arbitrary , but the maximum violation occurs for the value we have selected .
the probability of obtaining both particles in the `` 1 '' bin is @xmath64 while the probability of obtaining both particles in the `` 0 '' bin is @xmath65 the other probabilities such @xmath66 can be calculated in a similar fashion .
the probabilities formulated above are joint probabilities .
various of the strong bell inequalities also require marginal probabilities of the form @xmath67 the above integrals can be easy evaluated using the results@xcite @xmath68 \nonumber\\ & \;&\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;(for\;n\neq m ) \nonumber\\ \int_{-\infty}^{\infty}e^{-x^{2 } } h_{n}(x ) h_{m}(x)&= & 2^{n}n !
\sqrt{\pi } \delta_{n , m } \end{aligned}\ ] ] where @xmath69 is given by @xmath70 with @xmath71 being the gamma function .
performing the integrals for ( [ prob11 ] ) and ( [ prob00 ] ) we find @xmath72^{2}\cos \left[(n - m ) \psi\right]\end{aligned}\ ] ] similarly eqn ( [ marprob ] ) simplifies to @xmath73 which is independent of the sum of the local oscillator angle @xmath74 .
it is also simple to calculate the correlation function @xmath75 @xmath76^{2 } \nonumber \\ & \;&\;\;\;\;\;\;\;\;\;\times \cos \left[(n - m ) \psi\right ] \end{aligned}\ ] ] given the probabilities @xmath15 , @xmath30 , @xmath77 it is also possible to calculate the conditional information @xmath38 @xmath78-p_{00 } \log \left[2 p_{00 } \right]\nonumber \\ & -&p_{10 } \log \left[2 p_{10 } \right]-p_{01 } \log \left[2 p_{01 } \right]\end{aligned}\ ] ] it is now possible to calculate the clauser horne ( [ ch ] ) and spin ( [ spin ] ) and information - theoretic ( [ info ] ) bell inequalities .
some insight into the problem can be achieved by a careful examination of the term @xmath79^{2}\end{aligned}\ ] ] which is present in all the joint probability distributions .
this expression has several interesting features .
first , as the difference between @xmath80 and @xmath81 becomes large , the smaller that the above expression contributes to any of the probability distributions .
the main contribution for the expression comes from the case @xmath82 .
second , when @xmath83 is even , the above expression is zero . finally , as @xmath80 is large the different between the @xmath84 and @xmath85 elements for fixed large @xmath80 vanishes and they reach an asymptotic limit which is smaller than the @xmath86 case .
if these higher order @xmath80 terms dominate due to the choice of the @xmath8 in the probability formula , then the various bell inequalities can not violated .
this also has the implication that the mean photon number can not be high if a violation is to occur and hence it is not a macroscopic test of quantum mechanics .
to begin our investigations of the bell inequalities , consider the case of we have only two photon pair states that is , latexmath:[\ ] ] the informational theoretic bell inequality is given by @xmath98 .
a violation of this inequality is possible if @xmath99 .
unfortunately for all @xmath100 and
@xmath74 we have @xmath101 .
no violation is possible for any of the bell inequalities considered for the ideal state ( [ idealsimple ] ) when the detection scheme is based on homodyne quadrature phase measurements . if more correlated photon pairs are present can a violation be achieved ?
the obvious answer is yes , because of the recent work of gilchrist _ et .
the real question is how large this violation is ?
considering the expression ( [ correlatedpair ] ) for the correlated photon pairs , what are the optimal @xmath8 coefficients to maximize the violation . because of the results indicated by gilchrist _ et .
al . _ and our previous discussion
we anticipate that the mean photon number per mode must be low to obtain a violation .
hence we will truncate the number state basis at @xmath102@xcite photon per mode . performing a numerical optimization over all the @xmath8
, the optimal set is found to maximize the clauser horne and spin bell inequality ( table [ newtable1 ] ) .
a plot of @xmath8 versus @xmath80 is depicted in fig ( [ fig2 ] ) it is interesting to now discuss some properties of these optimal @xmath8 .
first the general shape of the @xmath8 versus @xmath80 curve shown in ( [ fig2 ] ) is similar to that considered in the circle state by gilchrist _ et .
al._@xcite . it is however not exactly the same ( see ( table [ newtable1 ] ) . given this optimal parameter set , what is the maximum violation of the bell inequalities we are considering . in fig ( [ fig3 ] ) we plot both the clauser horne and spin bell inequalities versus @xmath74 .
for the clauser horne bell inequality the maximum violation corresponds to @xmath103 , while the maximum violation for the spin bell inequality corresponds to @xmath104 .
interesting here is that the percentage violation of the spin inequality is approximately @xmath105 compared with the @xmath106 for the clauser horne case .
this significantly increases the potential for an experiment to be performed provided such an experiment were not significantly more difficult .
also the results for the optimal @xmath8 set give a clauser horne bell inequality violation that is approximately @xmath107 greater than the circle state results of gilchrist _ et . al._@xcite .
it is interesting to consider whether a greater violation of the bell inequality can be achieved with the state given by ( [ circlecn ] ) . to this end
we show the effect of the variation of both @xmath11 and @xmath74 ( sum of the local oscillator angles ) for both the clauser horne and spin bell inequalities in fig ( [ fig4 ] ) . as can be seen the spin bell inequality can be violated far more significantly than the similar clauser horne case .
in fact , as occurred previously the percentage maximum violation in the spin inequality is twice that of the clauser horne result .
in any of the analysis considered above we have not discussed errors , there sources and how they effect the potential violation .
we will not present any significant details here in this article but refer the reader to @xcite for such a decision . our final bell inequality to be considered
is the braunstein and caves@xcite information - theoretic case . in fig ( [ fig5 ] ) we plot @xmath108 versus @xmath74 .
no violation of the information - theoretic inequality is possible for any @xmath74 .
a question here to be addressed is why two of the strong inequalities can be violated while this information - theoretic bell inequality is far from being violated . in the binning process to give a binary result for a quadrature measurement , information must be discarded .
the information - theoretic inequality is much more sensitive to this information loss than the clauser horne inequality .
also why would we fundamentally expect all three inequalities to be violated
. a violation of any of the inequalities indicate a discrepancy between quantum mechanics and local realism .
in this article we have place strict bounds on the optimal @xmath8 coefficients for the state ( [ correlatedpair ] ) which maximizes the clauser horne and spin bell inequalities when a homodyne quadrature phase measurements is performed .
the spin bell inequality is violated by approximately @xmath109 while the clauser horne inequality is violated by approximately @xmath110 .
the violation is small however due to the fact that we are discarding information in the binning process .
in fact due to the information loss in the binning process the information theoretic bell inequality is not violated in any regime .
a larger violation can not be obtained using homodyne measurements with the strong inequalities we have considered .
while our optimal @xmath8 coefficient give a slightly better violation than the pair coherent state , it is difficult to see how such a state could be generated . closely examining the spin bell inequality with the pair coherent state still indicates that a greater violation ( approximately twice the size ) is possible than for the other inequalities
this would make the test much more feasible provided the pair coherent state could be generated .
in such a system the mean photon number is small , so this is not strictly a macroscopic test of quantum mechanics .
it does however have a macroscopic nature due the strong local oscillator which means large photodetector currents are obtained . to conclude , quadrature phase homodyne measurement provide a mechanism for performing tests of the bell inequality with highly efficient detection .
this allows one of the loopholes in current experiments to be closed .
however , due to the inherent information loss in the binning process , the violations are small but should be achievable .
a. einstein , b. podolsky , and n. rosen , phys .
rev . * 47 * , 777 ( 1935 ) .
a. gilchrist , p. deuar and m. d. reid , phys .
lett . * 80 * , 3169 ( 1998 ) .
w. j. munro and g. j. milburn , phys .
lett . * 81 * , 4285 ( 1998 ) b. yurke and d. stoler , phys .
* 79 * , 4941 ( 1997 ) .
e. s. polzik , j. carry , and h. j. kimble , phys . rev .
lett * 68 * , 3020 ( 1992 ) j. f. clauser , m. a. horne , a. shimony and r. a. holt , phys .
* 23 * , 880 ( 1969 ) .
g. kwiat , p. h. eberhard , a. m. steinberg and r. y. chiao , phys .
a * 49 * , 3209 ( 1994 ) .
m. freyberger , p. k. aravind , m. a. horne and a. shimony , phys .
a * 53 * , 1232 ( 1995 ) .
e. s. fry , t. walther , and s. li , phys .
a * 52 * , 4381 ( 1996 ) . by a strong test of quantum mechanics versus local realism ,
we mean a test in which no auxiliary assumptions are necessary . g. s. agarwal , phys .
lett . * 57 * , 827 ( 1986 ) .
k. tara and g. s. agarwal , phys .
a * 50 * , 2870 ( 1994 ) .
m. d. reid and l. krippner 93 , phys .
a * 47 * , 552 ( 1993 ) .
j. s. bell , physics ( n.y . ) * 1 * , 195 ( 1965 ) . j. f. clauser and m. a. horne , phys .
d * 10 * , 526 ( 1974 ) .
j. f. clauser and a. shimony , rep .
41 , 1881 ( 1978 ) .
m. d. reid , phys .
a * 40 * , 913 ( 1989 ) s. l. braunstein and c. m. caves , phys .
lett * 61 * , 622 ( 1988 ) .
m. abramowitz and i. a. stegun , _ handbook of mathematical functions _ ,
dover publications inc , eighth dover , new york , 1965 checks were performed to ensure that truncation errors were minimal ( less than @xmath111 ) . | we identify what ideal correlated photon number states are to required to maximize the discrepancy between local realism and quantum mechanics when a quadrature homodyne phase measurement is used .
various bell inequality tests are considered . 2 |
light is the most common probe for investigating complex media at the mesoscopic scale as it offers both an excellent resolution and is non invasive at moderate energies .
nonetheless , due to the inhomogeneous distribution of refractive index , light suffers multiple scattering while propagating in or through the medium .
unveiling the complexity of light scattering then necessary to retrieve the features of an object of interest or of the surrounding environment . in an inhomogeneous medium , it is a classical approach to consider a scattering sample as one realization of a random process , and study statistical physical quantities such as the mean intensity @xcite . under this approach ,
several physical parameters are relevant to characterize wave propagation in scattering media : the scattering mean - free path @xmath0 , the transport mean - free path @xmath1 , the diffusion constant @xmath2 , the absorption length @xmath3 .
classical back scattering imaging techniques , such as optical coherence tomography , fail when multiple scattering predominates @xcite .
however , one can still measure the long - scale spatial variations of the diffusive parameters .
the resulting image is not an image of the refractive index @xmath4 but e.g. , of the diffusion constant @xmath5 with a resolution of the transport mean free path @xmath1 , at best . in the literature ,
diffuse optical tomography is the gold standard technique to reconstruct the spatial distribution of transport parameters at each point of a volume from intensity measurements at the surface @xcite .
unfortunately , this inverse problem is intrinsically nonlinear with respect to the optical properties of the medium .
this method is thus computationally intensive and limited in terms of spatial resolution @xcite ( e.g. 5 mm in human soft tissues ) . in this paper
, we propose a simple and approach that does not require any inversion procedure and provides a direct image of transport parameters with a spatial resolution of the order of a few transport mean free paths . experimentally , it relies on a passive measurement of time - dependent green s functions between every point at the surface of a scattering medium without the use of any coherent source .
this method is based on the following fundamental result : the cross - correlation ( or mutual coherence function ) of an incoherent wave field measured at two points a and b can yield the time dependent green s function @xmath6 between these two points @xcite . previously developed in seismology
@xcite , in acoustics @xcite and in the microwave regime @xcite , the green s function estimation from the cross - correlations of a diffuse wave - field has been recently extended to optics @xcite .
the experimental set up employed for the passive measurement of the point - to - point green s functions at the surface of a scattering sample is displayed in fig .
an incoherent broadband light source ( 650 - 850 nm ) isotropically illuminates a scattering sample .
this incident wave - field exhibits a coherence time @xmath7 fs and a coherence length @xmath8 @xmath9 m @xcite.the backscattered wave - field is collected by a microscope objective and sent to a michelson interferometer , which is here used as a field correlator .
the beams coming from the two interference arms are recombined and focused by a lens .
a ccd camera conjugated with the sample surface records the output intensity .
@xmath10 if the incident light is spatially and temporally incoherent , the time derivative of the correlation function @xmath11 should converge towards the difference of the causal and anti - causal green s function @xcite , such that @xmath12 where @xmath13 is the time - dependent causal green s function between a point source at @xmath14 and a point detector @xmath15 .
this fundamental result means that a passive correlation measurement mimics the following active experiment [ see fig .
[ fig1](d ) ] : ( i ) a virtual light source of size @xmath16 located at point @xmath14 emits a pulse of duration @xmath17 ; ( ii ) a virtual point - like receiver located at point @xmath15 records the time - resolved scattered wave - field .
this property has been recently demonstrated in optics by retrieving the ballistic and multiple scattering components of the green s function between individual scatterers @xcite . following this experimental proof - of - concept
, we here investigate the case of strongly scattering media .
the first sample under study is a 80 @xmath9m - thick layer of @xmath18 nanoparticles .
a set of time - dependent causal and anticausal green s functions , @xmath19 , has been measured following the experimental procedure described above and depicted in fig .
when one tries to image an unknown medium from a set of green s functions , an important issue is the importance of multiple scattering relative to single scattering .
multiple scattering is actually a nightmare for classical imaging techniques , which are based on the first born approximation .
however , one can adopt a probabilistic approach and investigate statistical quantities such as the mean intensity@xcite .
as the causal and anticausal green s functions are , by definition , nil for @xmath20 and @xmath21 , respectively , the mean intensity can be averaged over the positive and negative times , such that : @xmath22 where the symbol @xmath23 denotes an average over the ccd pixels , i.e over the position vector @xmath14 of the virtual source on the sample surface . as we shall see , the evolution of this quantity as a function of the source - receiver relative position @xmath24 and the time of flight @xmath25 allows to describe quantitatively the transport of light through the scattering medium under study .
[ fig2 ] displays the spatio - temporal evolution of the mean intensity deduced from the set of point - to - point green s functions measured in the @xmath18 layer .
each spatial intensity profile is then renormalized by its maximum at each time of flight @xmath25 .
figure [ fig2](b , c , d ) displays the result at different opds ( @xmath2610 , 50 and 100 m , respectively ) .
not surprisingly , we retrieve the feature of a diffusive halo whose spatial extent increases with time at the same speed in all directions .
this isotropic growth is easily accounted for by the fact that the @xmath18 nanoparticles are small compared to the wavelength , randomly packed and display an isotropic differential cross - section .
now that we have observed qualitatively the diffusive spreading of the mean intensity , we aim to precisely characterize the scattering properties of the medium with a quantitative measurement of the diffusion constant @xmath2 .
theoretically , the multiple scattering intensity can be split into two terms .
first , the incoherent contribution , denoted as @xmath27 , corresponds to the interference of the wave with itself . in the deep multiple scattering regime , this term obeys the : @xmath28 with @xmath2 the diffusion constant and @xmath29 a physical quantity that depends on the sample thickness @xmath30 , the transport mean - free path @xmath1 , the absorption length @xmath3 but not on the transverse coordinate @xmath14 [ see link : link[supplement 1 ] ] .
secondly , the coherent contribution , denoted @xmath31 , corresponds to the interference of the wave with its reciprocal counterpart @xcite and is expressed as @xcite : @xmath32 at a given time of flight , the normalized multiple scattering intensity is thus given by : @xmath33}\ ] ] in our configuration , the spatial distribution of the intensity at a given time has the following shape [ see fig .
[ fig2](g ) ] : a narrow , steep peak ( the coherent contribution ) , on top of a wider pedestal that widens with time ( the incoherent contribution ) .
it is worth noting that due to the normalization process , the multiple scattering intensity profile does not depend on the absorption or the thickness of the sample . by fitting this intensity profile at each time of flight with two gaussian curves
, one can separate the coherent and incoherent contributions [ see fig .
[ fig2](g ) ] .
the width @xmath34 of the incoherent background directly accounts for the spatial extent of the diffusive halo .
figure [ fig2](d ) displays the evolution of @xmath35 versus time . from eq.[eq7 ] , we see that the slope of @xmath35 versus time should be equal to @xmath36 .
the linear fit of @xmath35 in fig .
[ fig2](d ) allows to measure the diffusion constant in the @xmath18 scattering sample : we find @xmath37 @xmath38 .
an estimation of the transport mean - free path @xmath1 is also possible using the relation @xmath39 , with @xmath40 the energy velocity .
the size of the nanoparticles being small compared to the wavelength and the medium being diluted , we can consider the energy velocity @xmath40 to be close to the vacuum light velocity @xmath41 @xcite .
thus , we obtain the following qualitative estimation for the transport mean free path : @xmath42 @xmath9 m . the access to transport parameters from the correlation of a diffuse wave - field is a precious information for the optical characterization of a scattering layer .
however , this measurement is averaged over the whole field - of - view ( fov ) and does not provide any local information on disorder .
this issue is tackled in the next subsection . until now
, the disorder was statistically homogeneous in the scattering sample under study .
we now investigate the case of a medium with a space - dependent concentration of scatterers .
for this purpose we have synthesized a heterogeneously disordered layer of zno nanoparticles .
variations in the concentration of scatterers are visible at a scale smaller than 100 @xmath9 m in the microscopic image [ see fig .
[ fig4](a ) ] .
cross - correlations of the diffuse wave - field are recorded using the procedure described in sec . 2 . .
the spatial extent @xmath34 of the diffusive halo is estimated at each time of flight by fitting the normalized multiple scattering intensity profile [ eq . [ eq7 ] ] with two gaussian curves .
first , the mean intensity is averaged over all the active pixels of the ccd .
a linear fit of @xmath35 versus time [ fig .
[ fig4](b ) ] allows an estimation of the diffusion constant average over a fov of 5 @xmath43 : @xmath44 @xmath38 .
the fov being at least one order of magnitude larger than the typical scale at which the concentration of scatterers fluctuates , such a measurement does not provide a satisfying characterization of the disorder in the scattering sample . to cope with this issue , a local approach is needed . instead of considering the mean intensity over the whole sample surface
, one can average the intensity over sub - areas of the ccd camera .
for instance , one can consider sub - areas delineated with red and blue squares in fig .
[ fig4](a ) that display low and high concentration of scatterers , respectively .
the corresponding diffusive halos show a contrasted speed of expansion [ see fig .
[ fig4](b ) ] .
quantitatively , a different diffusion constant is measured in both areas .
we find @xmath45 in the blue area and @xmath46 in the red area .
this is in agreement with the fact that a higher concentration of scatterers implies a smaller diffusion constant .
@xmath2 is actually proportional to the transport mean free path @xmath1 which itself scales as the inverse of the concentration of scatterers @xcite .
hence , the diffusion contrast highlighted by fig .
[ fig4](b ) indicates that the blue area is twice and half more concentrated than the red one .
a further development is to build an image of the medium from local measurements of the diffusion constant .
this is done by considering the mean intensity over sliding windows of @xmath47 @xmath48 .
the spatial extent @xmath35 of the diffusive halo is fitted linearly over time and yields an estimation of the diffusion constant at the center of each window .
the map of the diffusion constant is superimposed to the reflectivity image of the scattering medium in fig .
[ fig4](c ) .
a qualitative agreement is found between the two images since the diffusion constant is larger in areas where the concentration in scatterers is lower .
however , both images bring different information since the microscope image is only related to the concentration of scatterers at the surface of the medium whereas the diffusion constant also depends on the nature of disorder below the surface .
a 3d image could thus be built from a passive green s functions retrieval but it would require inversion schemes like in optical diffuse tomography @xcite . until now , we have only considered the case of isotropic scattering media . however , many complex samples such as biological tissues @xcite , nematic crystals @xcite , porous materials @xcite or fibrous media @xcite give rise to anisotropic light diffusion . in such media , the scalar diffusion constant
should be replaced by a diffusion tensor [ see link : link[supplement 1 ] ] . providing that the principal directions of anisotropy are aligned with the cartesian coordinates , this tensor is diagonal @xcite .
the three non zero coefficient @xmath49 , @xmath50 and @xmath51 correspond to the diffusion constant along the directions @xmath52 , @xmath53 and @xmath54 , respectively . to illustrate the ability of our approach in measuring different components of the diffusion tensor
, we now investigate light diffusion in a piece of thread seal tape ( polytetrafluoroethylene ) which is made of fibers aligned along the x - direction [ see fig .
[ fig3](a ) ] .
strong forward scattering arises along the normal to the fibers direction .
hence , light diffusion is supposed to be slower along the fibers : @xmath55 . to check that hypothesis , the experimental procedure described in sec .
2 is performed .
the measured green s matrix allows to investigate the spatio - temporal evolution of the mean intensity [ eq . [ eq1 ] ] averaged over the whole fov .
[ fig3 ] displays the diffusive halo in the @xmath56 plane at different times of flight .
the intensity distribution clearly presents an elongated shape along the @xmath57direction .
as expected , light diffusion is thus slower along the @xmath58direction .
this confirms the anisotropy of the sample and the ability of our approach to reveal it . in supplement 1
, the growth of the diffusive halo is quantitatively investigated both in the @xmath58 and @xmath57directions .
the components of the diffusion tensor are estimated following the procedure described in sec . 3a .
we find @xmath59 @xmath38 and @xmath60 @xmath38 .
an anisotropy factor @xmath61 can be deduced from the measurement of the diffusion tensor in the @xmath56 plane @xcite , such that @xmath62 .
the anisotropy of diffusion is thus successfully quantified by our approach .
the first point we would like to emphasize is the potential impact of our approach in optical diffuse tomography .
a passive measurement of the green s matrix can also enable a great leap forward in fundamental physics .
in this work , we have shown , for instance , how the anisotropy of light diffusion can be revealed and quantified .
a further step would be to investigate exotic transport phenomena such as super- and sub - diffusion of light @xcite or , more ambitiously , to provide a direct proof for anderson localization @xcite .
past experimental studies of anderson localization in optics have been ambiguous , due either to absorption effects @xcite or fluorescence issues @xcite . on the contrary ,
our approach would remove any possible ambiguity .
space- and time - dependent experiments can enable absorption - independent observations of localization @xcite .
moreover , a white light illumination in a backscattering configuration limits the incident power required for a satisfactory signal - to - noise ratio , contrary to a transmission configuration that implies a much less energetic signal through a thick scattering layer ( @xmath63 ) .
our experimental scheme would thus prevent from fluorescence or non - linear phenomena @xcite .
it also allows to investigate phenomena like coherent backscattering @xcite or recurrent scattering @xcite that constitute relevant observables at the onset of anderson localization
in summary , we have proposed a novel approach to investigate light transport through complex media .
inspired by previous studies in ultrasound and seismology , we have shown how a set of time - dependent green s functions can be passively extracted between every point at the surface of a scattering medium placed under white light illumination .
the experimental access to this green s matrix is essential since it contains all the information about the complex trajectories of light within the medium . adopting a probabilistic and local approach
, one can , for instance , investigate the spatio - temporal expansion of the diffusive halo in different parts of the sample .
on the one hand , this allows a two - dimensional imaging of the diffusion constant at the surface of the scattering medium , with a spatial resolution of the order of few transport mean free paths . on the other hand ,
our approach is able to reveal and quantify the anisotropy of light diffusion , which can be of great interest for optical characterization purposes .
this work opens important perspectives both in optical diffuse tomography with potential applications to biomedical imaging and in fundamental physics with the quest for an experimental demonstration of anderson localization in 3d .
labex wifi ( laboratory of excellence within the french program investments for the future , anr-10-labx-24 and anr-10-idex-0001 - 02 psl * ) ; european research council ( erc synergy helmholtz ) ; `` direction gnrale de larmement''(dga ) ; agence nationale de la recherche ( anr-14-ce26 - 0032 love ) ; high council for scientific and technological cooperation between france and israel ( p2r israel n@xmath64 29704sc )
we thank o. katz for fruitful discussions .
this document provides supplementary information on the convergence of the time derivative of the mutual coherence function towards the green s functions , the measurement of the diffusion constant components in isotropic and anisotropic scattering media and its comparison with a time - of - flight experiment .
the passive measurement of the green s function is based on the fundamental property that the time derivative of the correlation function should converge for infinite integration times towards the difference between the causal and anti - causal green s functions [ see eq.1 of the accompanying paper ] . in practice , this integration time remains fixed to @xmath65 ms in the experiments shown in the accompanying paper .
hence the measured signal may be polluted by noise . to assess the signal - to - noise ratio ( snr ) in our measurements
, the convergence of @xmath66 as a function of the integration time @xmath67 should be investigated .
it can be expressed as the sum of a deterministic term that resists to average , the expected green s functions , and a noise term that should vanish with average @xmath68 where @xmath69 accounts for the incoherent noise term whose coherence time @xmath17 is governed by the bandwidth of the white light source . to estimate the snr , the mean intensity of eq.[conv1 ] should be considered .
it yields @xmath70 the coherence properties of the noise @xmath69 allows to simplify the last equation into @xmath71 not surprisingly , the noise intensity should decrease as the inverse of the integration time @xmath67 .
fig.[convergence ] confirms this behavior by showing the spatially averaged intensity of @xmath72 measured in the tio@xmath73 layer as a function of @xmath67 .
layer*. the spatially averaged intensity of the mutiual coherence functoin , @xmath74 , is plotted as a function of the integration time @xmath67 at different opds .
the continous black line shows for comparison the expected noise intensity decrease as @xmath75 ( see eq.[conv3]).,width=302 ] for short time delays ( @xmath76 @xmath9 m in fig.[convergence ] ) , the signal intensity is sufficiently large to make the convergence of @xmath77 nearly immediate . for intermediate time delays ( @xmath78 @xmath9 m in fig.[convergence ] ) , @xmath74 decreases as @xmath75 for small integration times before saturating at the signal level for large integration times .
this means that the integration time @xmath65 ms is enough to get rid of most of the noise background and have a satisfying estimation of the green s function . on the contrary , for larger time delays ( @xmath79 @xmath9 m in fig.[convergence ] ) , the signal intensity is too weak to emerge from the noise background : @xmath74 decreases as @xmath75 over the whole integration time range , which is characteristic of a predominant noise background . the value of @xmath74 at time @xmath65 ms allows to determine the noise level in our experiment .
a snr can be derived for each opd : snr@xmath80 at @xmath81 @xmath9 m , snr@xmath82 at @xmath83 @xmath9 m and snr@xmath84 at @xmath85 @xmath9 m .
note that this noise background is systematically subtracted from the raw mean intensity prior to investigating the growth of the diffusive halo in each experiment .
in this section we investigate the case of the propagation of light in the multiple scattering regime .
as mentioned in the article , the energy density @xmath34 whose flux corresponds to the incoherent intensity obeys the diffusion equation : @xmath86 with @xmath87 a source function ,
@xmath3 the absorption mean free path and @xmath2 the diffusion constant .
the solution of this equation corresponds to a diffusive halo whose spatial extent broadens with time [ see fig .
its mathematical expression also depends on the boundary conditions . in the accompanying paper , the samples under study consist in strongly scattering layers of thickness @xmath88 80 @xmath9 m .
the characteristic time for the diffusive wave - field to travel through the scattering medium is given by the thouless time @xmath89 ps for @xmath37 @xmath90 . as the temporal range investigated in our study is of the order of 500 fs , the scattering medium can thus be considered as semi - infinite . in these conditions , _ patterson et al . _
@xcite established the expression for the backscattered incoherent intensity : @xmath91 with @xmath92 the extrapolation length @xcite and @xmath93 the source - receiver distance .
the incoherent intensity can thus be written as the product of two terms : @xmath94 with : @xmath95 the first term @xmath29 accounts for the temporal decreasing of the incoherent intensity whereas the second term in @xmath96 accounts for the temporal growth of the diffusive halo . by normalizing the incoherent intensity at each time flight
, one can therefore investigate the the diffusive properties of the scattering medium independently from the absorption losses .
in this section , we investigate the case of the propagation of light in an anisotropic scattering medium . in such a medium , the diffusion equation is now given by @xcite : @xmath97 with @xmath98 the energy density , @xmath87 a source function , @xmath3 the absorption mean free path and @xmath99 the diffusion tensor . according to ref.@xcite
, the corresponding incoherent intensity is given by : @xmath100 with @xmath49 and @xmath50 the in - plane components of the diffusion tensor . @xmath101 and @xmath102 are the projections of the source - receiver relative positions along the @xmath52 and @xmath53-axis . for a semi - infinite medium , @xmath103 is given by : @xmath104 at a given time of flight , the normalized incoherent intensity can be expressed as : @xmath105 as in the isotropic case , the normalized incoherent intensity is independent on the absorption losses .
[ s1 ] displays the spatio - temporal evolution of the mean intensity measured in the stretched teflon tape .
the two components @xmath49 and @xmath50 of the diffusion tensor can be estimated separately .
first , the relative position @xmath101 is set to 0 and the mean intensity is investigated as a function of time and @xmath102 [ see fig .
[ s1](b ) ] . a linear fit of the square width @xmath35 of the diffusive halo versus time allows an estimation of @xmath50=3000 @xmath38 . alternatively , by setting @xmath106 , we can investigate the diffuse intensity as a function of time and @xmath101 [ see fig.[s1](d ) ] .
again , a linear fit of @xmath35 yields an estimation of @xmath49=1100 @xmath38 . .
( b ) measured spatio - temporal evolution of the mean intensity at @xmath107 .
the intensity is renormalized by its maximum at each time of flight .
( c ) sketch of the spatio - temporal evolution of the mean intensity and its section at @xmath106 .
( d ) measured spatio - temporal evolution of the mean intensity for @xmath106 .
the intensity is renormalized by its maximum at each time of flight.,width=302 ]
in this section , we investigate the time - of - flight distribution of the reflected intensity integrated over the surface of the scattering sample , @xmath108 the decay of the reflected intensity with time bears particular signatures of light diffusion . for times of flight smaller than the thouless time @xmath109
, the medium can be considered as semi - infinite . in that case , a power law decay is expected for the reflected intensity at the surface of the sample @xcite : @xmath110 at longer times of flight , the finite sample size should come into play so that the decay becomes similar to that of time - dependent transmission , i.e. , exponential in the diffuse regime @xcite , such that @xmath111 in absence of absorption ( @xmath112 ) , the time - of - flight distribution of the backscattered intensity is thus an alternative way of measuring the diffusion constant @xmath2 .
an experimental set up involving a coherent illumination scheme and an interferometric arm in reception @xcite has been used to measure @xmath113 in the tio@xmath73 layer over an extensive range of time - of - flight .
the result is displayed in fig.[time ] . as predicted by theory [ eq.[power ] ] , the experimental time - of - flight distribution can be qualitatively fitted by the @xmath114 power law for @xmath115 [ see fig.[time](a ) ] .
sample*. ( a ) the time - evolution of @xmath113 ( blue circles ) is plotted in a log - log scale and fitted with the @xmath114 power law over the time range [ 500 fs - 1 ps ] .
( b ) the time - evolution of @xmath113 ( blue circles ) is plotted in a linear - log scale and fitted with an exponential decay over the time range [ 2 ps - 4.5 ps].,width=302 ] in agreement with theory , the time decay of @xmath113 becomes exponential at larger times of flight [ see fig.[time](b ) ] .
if we neglected absorption losses , a fit of the exponential decay would yield an estimation of the diffusion constant @xmath1161250 [email protected]@xmath118 .
this value is quite far from our measurement deduced with the growth of the diffusive halo ( @xmath119 730 @xmath120.s@xmath118 , see sec.3a of the accompanying paper ) .
the mismatch between both values can be easily accounted for by the presence of absorption losses .
considering that @xmath119 730 @xmath120.s@xmath118 , the exponential decay fit performed in fig.[time](b ) leads to the following estimation for the absorption mean free path : @xmath121 @xmath9 m .
this time - of - flight experiment illustrates the benefit of our approach that allows a quantitative measurement of the diffusion constant , independent from the absorption losses . | imaging the propagation of light in time and space is crucial in optics , notably for the study of complex media . we here demonstrate the passive measurement of time - dependent green s functions between every point at the surface of a strongly scattering medium by means of low coherence interferometry .
the experimental access to this green s matrix is essential since it contains all the information about the complex trajectories of light within the medium . in particular
, the spatio - temporal spreading of the diffusive halo can be locally investigated in the vicinity of each point acting as a virtual source . on the one hand
, this approach allows a quantitative imaging of the diffusion constant in the scattering medium with a spatial resolution of the order of a few transport mean free paths . on the other hand ,
our approach is able to reveal and quantify the anisotropy of light diffusion , which can be of great interest for optical characterization purposes .
this study opens important perspectives both in optical diffuse tomography with potential applications to biomedical imaging and in fundamental physics for the experimental investigation of anderson localization . |
the two dimensional electron gas ( 2deg ) in the presence of magnetic field has been of special interest to the condensed matter physicists since the discovery of the quantum hall effect @xcite and the fractional quantum hall effect @xcite .
the 2deg in a periodic potential and magnetic field has been the cradle of several interesting and important theoretical concepts like the identification of a topological invariant with the hall conductivity @xcite and the existence of a fractal structure in the energy gaps , the hofstadter butterfly @xcite .
interest in this phenomenon has been recently revived with the experimental observation of the hofstadter butterfly in graphene superlattices @xcite and the realization of the hofstadter hamiltonian in the optical lattice systems @xcite .
this has motivated us to study the effect of interactions on the hofstadter butterfly .
the hofstadter butterfly emerges from the interplay of two length scales in the lattice systems : the periodicity of the lattice and the magnetic length @xcite .
it is an effect that emerges when we look at the single particle energy gaps and hall conductivities of a large number of physical systems , in principle , an infinite number of systems with flux per plaquette , @xmath0 , where @xmath1 are co - prime integers and @xmath2 . for non - interacting systems with fermion densities corresponding to filled bands , the gaps exhibit a self similar structure .
further , if the hall conductivities , in units of @xmath3 are plotted as a function of @xmath4 and the number of fermions per unit cell , the contours with the same hall conductivities are straight lines with integer intercepts .
the hall conductivity of each contour and the intercept are solutions of a diophantine equation .
this plot , the landau fan diagram , is the experimental evidence of the hofstadter butterfly in graphene superlattices @xcite . interaction can induce charge ordering that can break the translational symmetry of the system .
interaction induced translational symmetry breaking phases have been studied previously in the honeycomb lattice in the absence of magnetic field @xcite .
effect of interactions on the hofstadter butterfly has been studied in the past using mean field approximation in both square lattice @xcite and honeycomb lattice ( for dirac fermions)@xcite . electron electron interaction in square lattice in magnetic field has also been studied using exact diagonalization method @xcite . however
, none of these past works consider translational symmetry breaking and its effect on the fractal structure of the hofstadter butterfly .
moreover , the effect of interactions on the experimental probe of the hofstadter butterfly , the landau fan diagram , has not been mentioned in previous work . in interacting systems ,
the translational symmetry breaking introduces a third length scale in the problem which can affect the self similarity of the hofstadter butterfly . in this paper
, we investigate whether this actually happens . in our recent paper @xcite
, we had studied the effect of interaction in one of the many systems required to realize the hofstadter butterfly in honeycomb lattice , lattice with flux per plaquette @xmath5 ( in units of @xmath6 ) .
we studied the interaction induced translational symmetry broken phases in the hofstadter regime of the honeycomb lattice for this flux value . in this paper , we study the effect of interactions on the spinless fermions of the honeycomb lattice in the presence of magnetic field such that the flux per plaquette is of the form @xmath7 where @xmath1 are co - prime integers .
we consider fermion densities corresponding to filled bands and address following two questions : ( i ) how common are translation symmetry breaking transitions and how often are they accompanied by the change in the hall conductivity ?
( ii ) do they destroy the fractal structure ?
the rest of this paper is organized as follows : in sec .
[ sec : ii ] , we discuss the model and the phase transitions in the system due to the interactions . sec .
[ sec : iii ] , gives a brief review of the non - interacting hofstadter butterfly in the honeycomb lattice and describes the self similarity of the fractal structure of the butterfly .
the effect of interactions on the hofstadter butterfly is described in sec .
[ sec : iv ] . in sec .
[ sec : v ] , the landau fan diagram for both the non - interacting and interacting cases are discussed and compared .
finally , we conclude in sec .
[ sec : vi ] .
the model we consider is spinless fermions on the honeycomb lattice in the hofstadter regime with nearest neighbor hopping and nearest neighbor interaction .
the hamiltonian is @xmath8 where @xmath9 is the annihilation ( creation ) operator for electrons at site @xmath10 on the honeycomb lattice , @xmath11 is the number density operator and @xmath12 is the nearest neighbor hopping parameter and @xmath13 is the nearest neighbor interaction strength .
we consider @xmath14 and @xmath13 is in units of @xmath12 .
@xmath15 are the gauge fields on the nearest neighbor links such that the magnetic flux passing through each plaquette is @xmath16 where @xmath17 are co - prime integers with @xmath18 and @xmath2 .
we refer different values of flux per plaquette as different systems . for the range of @xmath19 considered in this paper ,
there are @xmath20 flux values in total and thus @xmath20 systems . for each of these systems , we solve the interacting problem for the filled band cases only .
hence , for a particular interaction strength , we have @xmath21 cases in total .
the hamiltonian is invariant under magnetic translations @xmath22 and @xmath23 which are along @xmath24 and @xmath25 directions respectively .
@xmath26=0 $ ] .
we choose the magnetic unit cell to be @xmath19 adjoining original unit cells along the @xmath24 direction as shown in fig .
[ fig:0 ] for @xmath27 .
each magnetic unit cell contains @xmath28 sites .
other symmetries of the system are 6-fold rotations about the centers of the hexagons , 3-fold rotations about the sites and 2-fold rotations ( inversion ) about the centers of the links . at half filling , the system also has particle - hole ( chiral ) symmetry , @xmath29 , where @xmath30 for @xmath10 belonging to one of the sublattices and @xmath31 for the other .
the brillouin zone is the set of wave vectors @xmath32 , where @xmath33 are the reciprocal lattice vectors of the underlying triangular lattice with @xmath34 and @xmath35 .
( color online ) honeycomb lattice in magnetic field with flux @xmath5 passing through each plaquette . @xmath36 and @xmath37 are the two sublattices . @xmath38 and @xmath39 represent the basis vectors of the lattice .
the gray portion shows the magnetic unit cell choice considered in this paper .
@xmath40 are sublattices of this magnetic unit cell . ]
to solve this interacting problem , we use mean field approximation discussed in our previous work @xcite , @xmath41 where @xmath42 denotes all the nearest neighbors of @xmath10 .
the self consistency equations , eq . , have to be solved keeping the number density fixed .
we solve them numerically for interaction strength @xmath43 .
for each flux value @xmath7 , there are @xmath44 complex bond order parameters and @xmath28 real charge order parameters .
the choice of the magnetic unit cell is not unique . for the non - interacting case ,
when the translation symmetry is not broken , the choice of the unit cell is irrelevant . for the interacting case , the choice of unit cell matters .
it determines the pattern of the translation symmetry breaking . with increasing @xmath19 ,
the number of distinct magnetic unit cell choices increase and it is not feasible to solve all the possibilities numerically .
we restrict ourselves to the linear magnetic unit cell choice , as shown in fig .
[ fig:0 ] for @xmath45 .
there could be phases with other patterns of translation symmetry breaking with lower energy and they could appear at lower values of the interaction strength .
our analysis thus underestimates the effects of the interactions .
we work with a lattice of @xmath46 magnetic unit cells and a fixed number of particles corresponding to a particular band filling .
the self consistency equations are solved for filled bands till half filling .
the self consistency solutions for the upper half filling cases is same as the lower half due to particle hole symmetry .
we solve the mean field hamiltonian for these cases for the interaction strength @xmath43 .
we observe that for @xmath47 , @xmath48 of the cases show phase transitions and @xmath49 of the cases show topological transitions , for @xmath50 , @xmath51 of the cases show phase transitions and @xmath52 of the cases show topological transitions and for @xmath53 , @xmath54 of the cases show phase transitions and @xmath55 of the cases have topological transitions .
hence , we see that even for interaction strength @xmath47 significant number of cases show phase transitions and topological phase transitions which increases with increase in the interaction strength .
fig.[fig:1 ] shows the probability of phase transitions , @xmath56 , and topological phase transitions , @xmath57 , as a function of the filling fraction , @xmath58 .
@xmath59 is defined as the number of systems with filling between @xmath58 and @xmath60 that show phase transitions ( topological transitions ) divided by the total number of systems with filling between @xmath58 and @xmath60 . in fig.[fig:1
] , we have taken @xmath61 .
+ fig.[fig:1 ] shows that the number of phase transitions peak at half filling and at the dilute limit .
as expected the number increases with increase in the interaction strength .
the peak near half - filling seems intuitively reasonable , since the inter - particle distance decreases with increasing particle density and so the effect of the nearest neighbor interaction increases .
however , by the above reasoning , there should be minimum number of phase transitions in the dilute limit , quite contrary to what we see in fig.[fig:1 ] .
the answer to this puzzle comes from examining the energy bands of the non - interacting system .
we observe that for flux per plaquette @xmath7 , at low filling , @xmath62 bands come close to each other and the energy gap between these bands decreases with the increase in @xmath19 .
since the band gaps tend to become low in these regions for the systems with larger @xmath19 , it is easier for the interaction to mix the bands leading to the transitions .
near half filling @xmath63 bands bunch up and tend to get degenerate at large @xmath19 .
[ fig:01 ] illustrates this for flux value @xmath64 . in the fermi regime , the energy gap between the lowest two bands in fig .
[ fig:01 ] is very small and are bunched together .
these gaps become negligible as @xmath19 increases .
similarly , the four energy bands near the half filling , the dirac regime , bunch together as can be seen from the fig .
[ fig:01 ] .
on increasing @xmath19 , these bands become @xmath63 degenerate .
[ fig:01a](color online ) non - interacting energy band diagram for @xmath65 . in this figure , in the fermi regime , the lowest two bands are very close to each other and bunch together .
similarly in the dirac regime , at half filling , the four energy bands bunch together.,title="fig : " ] this phenomenon of bands bunching as @xmath19 increases can be understood by examining two extreme limits of this problem : ( i ) hofstadter regime ( small @xmath19 ) ( ii ) weak field limit ( @xmath66 ) . at small @xmath19
, we typically have @xmath28 well separated bands , each of them contributing a particle density of @xmath67 per unit cell , when completely filled . at large @xmath19
, we can analyze the system in the continuum limit , separately for the dilute limit and near half - filling . in the dilute limit
, the system behaves like a single species of non - relativistic fermions in a magnetic field . the spectrum in this regime , dubbed as the fermi regime by hatsugai et .
@xcite , consists of landau levels each contributing a particle density of @xmath7 per unit cell , when completely filled .
thus we may expect @xmath62 of the bands to become degenerate in the weak field limit , consistent with the bunching that we observe .
near half - filling is the so called dirac regime @xcite . here
the system behaves like two species of dirac quasiparticles .
the spectrum consists of relativistic landau levels .
since there are two species , each landau level has a particle density of @xmath68 .
thus in this regime we expect a bunching of @xmath63 bands with increasing @xmath19 , which form the degenerate landau level in the @xmath69 limit . from the above argument
, we also expect the chern number of the bunch of @xmath62 bands to sum up to 1 in the dilute limit and that of the bunch of @xmath63 bands to sum up to 2 near half - filling .
we have computed the chern numbers numerically and have found that this is indeed true .
before describing the effect of interactions on the hofstadter butterfly , we briefly review its fractal structure for the non - interacting honeycomb lattice . in this section
, we also show that the range of @xmath70 we are considering is large enough to see the self similarity to first order , namely the first step of the recursion .
( color online ) hofstadter butterfly for the honeycomb lattice . here
the x - axis represents the single particle energy @xmath71 and y - axis is the magnetic flux per plaquette @xmath4 of form @xmath7 . in this plot @xmath72 . ] the hofstadter butterfly for the non - interacting honeycomb lattice @xcite is shown in fig .
[ fig:2 ] .
this plot is self similar , in the sense that there is a mapping of the whole plot into a block in the plot which is described in detail in this section .
when this block is suitably scaled and rotated , it reproduces the original plot .
we describe the recursion below .
first we construct the so called ` skeleton ' of the hofstadter butterfly .
we define @xmath4 of the form @xmath67 and @xmath73 for @xmath74 to be ` pure cases ' .
the skeleton is formed from the set of curves constructed as described below : ( a ) connect the outer edges of @xmath75 and @xmath76 band of neighboring pure cases for @xmath77 .
this forms a huge box which we denote as the @xmath78 block .
we divide this @xmath78 block into sub - blocks .
these sub - blocks are the portions of the @xmath78 block between @xmath79 and @xmath75 flux values .
we label these sub - blocks as @xmath80 as shown in fig .
[ fig:4a ] .
( b ) connect the right outer edges of @xmath81 band and the left outer edges of the lowest band of neighboring pure cases for @xmath77 .
this forms a huge box which we denote as the @xmath82 block .
the @xmath82 block is further divided into @xmath83 and @xmath84 sub - blocks .
connecting the right outer edges of the lowest band of the neighboring pure cases form the @xmath83 sub - block and connecting the left outer edge of the second lowest band and the right outer edge of the @xmath81 bands of the neighboring pure cases form the @xmath84 sub - block .
this construction is shown in fig .
[ fig:4a ] .
the whole plot in a compressed form with some rotation is present inside each of the @xmath78 sub - blocks @xcite .
this statement is quantified using a recursive relation from the original plot to the @xmath78 sub - blocks @xcite . in each of these sub - blocks
, there is a local variable , @xmath85 , for flux per plaquette defined in terms of the variable @xmath4 representing the flux per plaquette of the original plot .
for @xmath86 we define @xmath87 as @xmath88 $ ] . @xmath89 $ ] stands for the greatest integer less than or equal to @xmath90 .
the recursive relation between @xmath4 and @xmath85 is given by @xcite @xmath91 thus the local variable @xmath85 has values in @xmath92 $ ] like the flux in the original plot .
self - similarity requires that the number of energy bands and gaps at a value of @xmath4 in the original plot is the same as the number of energy bands and gaps at @xmath85 in each sub - block .
we now show that @xmath19 in the range @xmath93 $ ] is enough to verify the first step of the recursion . in fig .
[ fig:4a ] , the @xmath78 block has @xmath94 sub - blocks for the considered range of @xmath19 .
each value of @xmath4 in @xmath95 can be mapped to a value of @xmath85 using eq . .
similar to @xmath4 , we can define ` pure cases ' in @xmath85 .
for example : @xmath96 ; in this case @xmath95 sub - block has @xmath97 bands which is the same number of bands for @xmath98 in original plot , @xmath99 ; here @xmath95 sub - block has @xmath100 bands which is the same number of bands for @xmath5 in original plots as seen in fig .
[ fig:4.4 ] . in general , the number of bands for a particular value of @xmath85 in @xmath95 sub - block is same as that of @xmath101 in the original plot . using the same method of constructing blocks in the original unit cell , as described before , we construct the secondary @xmath78 and @xmath82 blocks in the @xmath95 sub - block .
this secondary @xmath78 block is further divided into @xmath102 sub - sub - blocks some of which are shown in fig .
[ fig:4b ] .
we verify that each of these secondary blocks have the same number of energy bands and gaps and are arranged in similar fashion for a particular value of @xmath85 as the original plot for the same value of @xmath4 .
hence , the energy spectrum of the original plot is seen to be repeating in the sub - block and the self - similarity criterion is satisfied upto first order for the @xmath103 $ ] .
the hofstadter butterfly described in the previous section is defined for the non - interacting case in terms of the single particle energies and the single particle gaps .
however , single particle energies are not defined for the interacting system . in presence of interactions ,
the well defined quantities are the many body ground state at each electron densities and the chemical potential at these number densities . in the non - interacting system ,
the chemical potential is the highest occupied single particle energy .
hence , in the interacting case , it is natural to plot the magnetic flux per plaquette with respect to the chemical potential for different fillings .
we solve this interacting problem using mean field theory .
the mean field hamiltonian , @xmath104 . can be defined as @xmath105 where @xmath106 is a @xmath28 component vector with components @xmath107 .
@xmath108 is the fourier transform of the fermion operators @xmath109 and is defined as , @xmath110 where @xmath109 is the fermionic operator that annihilates an electron at unit cell @xmath10 and sublattice index @xmath111 , @xmath87 is the total number of sites and @xmath112 brillioun zone .
@xmath113 is the single particle mean field hamiltonian in the momentum space which is a @xmath114 matrix .
the spectrum of @xmath113 is given by the eigenvalue equation @xmath115 where @xmath116 is the single particle energy for the @xmath117 band and @xmath118 is its eigenfunction .
@xmath119 is the fermionic operator that creates a fermion in the single particle state with quasi momentum vector @xmath120 in the @xmath121 band . for @xmath122 filled bands
, the mean field ground state can be written as @xmath123 where @xmath124 is the vacuum state .
the ground state energy is the sum of the single particle energies @xmath125 .
if the state , @xmath126 is a good approximation of the single quasi - particle state , then the gap is the lowest single particle energy in the @xmath127 band , @xmath128 .
while this is often the case , it is not always so .
for example , in the case of quantum hall skyrmions @xcite , the order parameter deforms locally when one electron is added to the system and the gap can get reduced by a factor of 2 .
when we solve our mean field equations for a dilute density of fermions in the @xmath127 band , we find that even for very small filling the mean field hamiltonian and consequently the ground state changes significantly i.e @xmath129 is quite different from @xmath130 even for small @xmath131 .
we feel that this may imply that the order parameters deform locally when an extra particle is added and hence that the mean field gap is not reliable .
exploring this issue in detail is ongoing work . in this work ,
we consider only the filled band cases where the mean field theory is reliable .
while we do not know the single particle energy gaps , nevertheless , we can take the band edge to be the single particle energy of the highest occupied level ( which is the chemical potential ) .
hence , to investigate the effects of the interactions on the hofstadter butterfly , we first plot the magnetic flux per plaquette versus the maximum energy of the band for the non - interacting case and show that this plot also shows the self - similar structure .
we then compare it with the interacting case .
[ fig:6a ] is the plot for flux per plaquette versus the maximum energy of each band for the non - interacting case .
the plot is restricted to half - filling here . + as seen in fig .
[ fig:6a ] , the plot is divided into blocks @xmath78 and @xmath82 ( including @xmath84 and @xmath83 blocks ) . the @xmath78 block is further divided into sub - blocks where this pattern repeats itself as seen in fig . [ fig:6b ] .
in this case , the self similarity criterion is satisfied by verifying that the number of energy points for a particular value of @xmath85 in @xmath95 sub - block is same as that of @xmath4 = @xmath85 in the original plot .
[ fig:6b ] is the plot of the hofstadter butterfly in the absence of interaction for the flux in the range @xmath132 i.e. the @xmath95 sub - block of the @xmath78 block . for the range of @xmath19 considered in this paper
, we verify that each of these secondary blocks have the same number of energy points and are arranged in similar fashion for a particular value of @xmath85 as the original plot for the same value of @xmath4 .
thus , we show that the recursive pattern is present even for plot of magnetic flux per plaquette versus the maximum energies of the bands upto first step .
now we study the effect of the interaction on fig .
[ fig:6a ] .
as described in our previous work @xcite , there is always a scaling solution for this interacting problem which satisfies the self consistency equations .
the state corresponding to this solution is the same as the noninteracting case .
however , the strength of the hopping parameters get scaled as @xmath133 and the single particle energies get scaled as @xmath134 where @xmath135 is the number of bands filled and @xmath136 .
we call this the symmetric phase as all the symmetries of the system still remain intact in this phase . in this phase ,
the single particle energies just get scaled .
but , these scalings are not uniform and depend on the bands filled .
however , the band gap never closes and the whole fractal structure of the hofstadter butterfly remains intact
. the recursive relations for the non - interacting case still remain valid for the symmetric phase .
the plot can be divided into @xmath78 and @xmath82 blocks .
further , the @xmath78 block can be divided into sub - blocks and in each sub - blocks , the whole energy spectrum is repeated in a similar fashion as discussed in the non - interacting case .
the scaling solution is not always the minimum energy solution . as discussed in section
[ sec : ii ] , there are a large number of flux values and filling fractions where there are phase transitions to translational symmetry broken states . the plot for flux per plaquette versus the maximum energy of each band is given in fig .
[ fig:7 ] for @xmath137 and @xmath53 . as seen from fig .
[ fig:7a ] and fig .
[ fig:7b ] , for @xmath47 , the hofstadter butterfly , like non - interacting case , can be divided into @xmath78 and @xmath82 blocks .
viewing a particular sub - block in @xmath78 block , as shown in fig .
[ fig:7b ] , we see that the form of the energy spectrum is not repeated inside this sub - block .
thus , the plot is not self - similar . for @xmath53 , a larger portion of the fractal structure
gets destroyed compared to @xmath47 as seen from fig .
[ fig:7c ] .
the fact that the fractal structure increasingly disintegrates with the increase in the interaction strength is a consequence of the increase in the number of phase transitions with the increase in the interaction strength .
the @xmath78 block contains energy bands near half filling which , as seen from fig . [ fig:1 ] , have very high probability of phase transitions which increases with @xmath13 .
these phase transitions , except at exact half filling , break the translational symmetries of the system .
thus , the translational symmetry breaking phase transitions seem to play a crucial role in the destruction of the fractal structure .
as we have stated earlier , this can be expected on the basis of general arguments .
for very dilute case , there are high number of phase transitions and they affect the @xmath82 block . however , in this case , for small value of interaction strength , the system is in symmetric phase for @xmath62 bands filled for flux @xmath7 as there is a comparatively high energy gap between the @xmath138 and @xmath139 band .
thus , these high energy gaps in the @xmath82 block still remains for small interaction strength and will slowly vanish with increase in the interaction strength .
experimental evidence for the hofstadter butterfly has come from the landau fan diagram .
each gap in the hofstadter butterfly can be characterized by two integer topological invariants @xmath140 that satisfy the diophantine equation @xcite @xmath141 where @xmath135 labels the gap and the flux passing per plaquette is @xmath142 , number of particles per unit cell is @xmath143 .
@xmath144 where @xmath145 is the hall conductivity at the @xmath146 gap and @xmath147 is the change in the electron density when there is an adiabatic change in the periodic potential @xcite .
the plot of the hall conductivity with respect to the number of particles per unit cell and the magnetic flux passing per plaquette is called the landau fan diagram .
[ fig:8 ] shows the landau fan diagram for the non - interacting case . in fig .
[ fig:8 ] , the points with the same hall conductivities can be joined to give a straight line which when extrapolated meets the x - axis at an integer point .
this intercept gives the value of @xmath147 whereas the slope gives the value of @xmath148 . in this figure ,
the colorbar is restricted to the values of @xmath148 from @xmath149 to @xmath150 for convenience in plotting ; the maximum value of @xmath148 for non - interacting case for @xmath72 is @xmath151 .
hence , @xmath72 is enough to show and analyze the landau fan diagram and realize that these topological invariants indeed satisfy the diophantine equation .
as mentioned earlier , many of the landau transitions are accompanied by topological transitions where the hall conductivity of the system changes . in the presence of interactions , the topological phase transitions get reflected in the landau fan diagram as shown in fig .
[ fig:9 ] for @xmath43 . here
the landau fan diagram is plotted only for bands with non - trivial topology , i.e. removing the points with zero hall conductivity . in fig .
[ fig:9a ] , for @xmath47 , though most of the points with the same hall conductivities can be joined in a straight lines but there are some points in these lines which have different hall conductivities .
but in fig .
[ fig:9c ] , for @xmath53 , the points with the same hall conductivities can not be joined to form a straight line as most of the points are scattered .
this is due to the topological transitions accompanied with the landau phase transitions which increases with the increase in the interaction strength .
moreover , we can see that most of the region near half filling have a topological transition to zero hall conductivity as shown in fig .
[ fig:9 ] .
this region also increases with the increase in the interaction strength .
the maximum value of the hall conductivity , considering all filled bands for all values of flux per plaquette of the form @xmath7 with @xmath72 , decreases with the increase of interaction strength .
for example , in the absence of interactions , the maximum value of the hall conductivity is @xmath152 , while for case of @xmath47 it is @xmath153 , for @xmath50 it is @xmath154 and for @xmath53 the maximum value of the hall conductivity is @xmath155 .
+ hence , from fig .
[ fig:8 ] and fig .
[ fig:9 ] , it is clear that due to the topological transitions accompanying the phase transitions to spatial symmetry breaking phases , the points with same hall conductivity in the landau fan diagram are more scattered and the number of transitions to zero hall conductivity near the half filling increases as the interaction strength is increased .
in addition , on using the same diophantine equation as for the non - interacting case , @xmath147 no longer remains an integer .
for example , for @xmath53 and @xmath156 , @xmath157 , so @xmath158 .
hence , the diophantine equation used for the non - interacting case is no more valid in the presence of interactions .
we have studied spinless fermions on the honeycomb lattice with nearest neighbor hopping and nearest neighbor interaction in the presence of magnetic field .
the magnetic flux per plaquette is of the form @xmath7 with @xmath17 being co - prime integers and @xmath159 , @xmath160 .
we solve this interacting problem in the mean field approximation for the filled band cases .
in particular , we look for translation symmetry broken phases and study the effect of these phase transitions on the hofstadter butterfly and the landau fan diagram . we find that a large number of the systems at different values of flux and filling exhibit these transitions .
many of the transitions are also topological , i.e. the symmetry breaking is accompanied by a change in the hall conductivity . when the number of transitions is plotted with respect to the filling factor , we find that they are peaked near the dilute limit and near half - filling .
we have provided an explanation of this feature based on the bunching of bands in the non - interacting system .
the number of these phase transitions increases with the increase in the interaction strength as expected .
the hofstadter butterfly is generally understood as arising from the interplay of the two length scales in the system , the periodicity of the potential and the magnetic length .
the translation symmetry breaking introduces a third length scale into the system and hence we expect a strong effect of it on the fractal structure .
we have shown that this is indeed so .
the self similarity structure of the energy spectrum disintegrates as a result of these transitions .
the amount of disintegration increases with the strength of the interactions .
this result is with respect to the choice of the unit cell shown in fig.[fig:0 ] , corresponding to a particular pattern of translation symmetry breaking . in this choice
, the translational symmetry is broken only in the @xmath24 direction
. there could be states with different patterns of symmetry breaking with lower energy and hence for many fillings and flux values , the translation symmetry could break at a smaller interaction strength than our case .
thus if all patterns of translation symmetry breaking were taken into account , we may expect the hofstadter butterfly to disintegrate at lower interaction strengths than shown in this work .
landau fan diagram is the experimental manifestation of the hofstadter butterfly .
we show that the change in the hall conductivity in the transitions gets reflected in the landau fan diagram .
the points with same hall conductivity no longer lie in a straight line and are rather scattered . on increasing the interaction strength , there are more number of topological transitions and thus it becomes difficult to join the points with same hall conductivities in a straight line as they get more scattered in the landau fan diagram . in the presence of interaction , the diophantine equation used for non - interacting case
does not hold .
hence we have shown that interactions disintegrate the hofstadter butterfly .
further , the landau fan diagram also drastically changes and the diophantine equation obeyed by the topological invariants in the non - interacting system breaks down in the interacting system . | we study the effect of interactions on the hofstadter butterfly of the honeycomb lattice .
we show that the interactions induce charge ordering that breaks the translational and rotational symmetries of the system .
these phase transitions are prolific and occur at many values of the flux and particle density .
the breaking of the translational symmetry introduces a new length scale in the problem and this affects the energy band diagram resulting in the disintegration of the fractal structure in the energy flux plot , the hofstadter butterfly .
this disintegration increases with increase in the interaction strength .
many of these phase transitions are accompanied with change in the hall conductivity .
consequently , the disintegration of the hofstadter butterfly is manifested in the landau fan diagram also . |
roth s conjecture @xcite in the theory of finite groups asserts that the ad - representation of the group @xmath2 on itself contains every complex irreducible representation of @xmath3 at least once ( where @xmath4 is the centre ) .
this conjecture is false in general , but is known to be true for symmetric groups @xcite and alternating groups @xcite , and , recently , for the sporadic simple groups @xcite using methods from @xcite .
we shall refer to this property , when it holds , as the roth property .
indeed , for simple groups the exceptions according to @xcite appear to be rather few and to amount to some instances of one classical family of lie type over finite fields of particular order . in this paper
we provide a noncommutative - geometrical point of view on this property , relating it to the universal differential calculus on the finite group and nondegeneracy of the corresponding killing form ( these notions will be defined in the preliminaries section of the paper ) .
nondegeneracy of the killing form here appears to be weaker than the roth property and we conjecture that it holds at least for all simple @xmath2 . beyond this , we expect more strongly at least when @xmath2 is simple or @xmath0 that the killing form is nondegenerate quite generally , not only for the universal calculus . a general calculus on a finite group corresponds to an ad - stable subset @xmath5 with equality in the case of the universal calculus .
here @xmath6 is the group identity .
the killing form is defined on @xmath7 and in the case of finite groups reduces to the trace in the representation @xmath8 where @xmath9 is a basis of @xmath10 and @xmath11 is the character or trace in the representation @xmath10 . in our case
the representation @xmath10 is given by the adjoint action hence @xmath12 where @xmath13 is the centraliser of @xmath14 . complementing the case of the universal calculus in section 3 , we focus next on the other extreme where @xmath15 is a single conjugacy class stable under group inversion ( such classes are called ` real ' as the character is real - valued ) .
section 5 proves nondegeneracy for the natural 2-cycles calculus on symmetric groups @xmath0 while section 6 provides computer verification of this conjecture for real conjugacy classes up to @xmath16 some 75,000 .
this includes the smallest sporadic group @xmath17 and the suzuki group @xmath18 .
the geometric intuition is that when @xmath2 is simple _ any _ @xmath15 generates @xmath2 and hence could be viewed as some kind of ` lie algebra ' which we might hope to reflect the simplicity of @xmath2 .
once we have a killing form , nondegenerate or not , we can envisage many applications . in usual lie theory
complex simple lie algebras have up to equivalence a unique compact real form where the killing form is positive definite , leading in the geometry to a riemannian metric with constant positive curvature .
the simplest finite group examples along these lines have been studied ` by hand ' and it was found , for example , that @xmath19 with its 2-cycles conjugacy class is an einstein space with ricci curvature essentially proportional to the metric , while @xmath20 with an order 4 conjugacy class is ricci flat @xcite . with such eventual geometry in mind , we are interested in when the matrix of @xmath21 in our basis @xmath15 is positive definite .
the answer among real conjugacy classes appears to be precisely those which are classes of involutions .
one of these could then be a natural ` compact ' braided lie algebra structure of a finite nonabelian simple group .
we find moreover that the other real conjugacy have the interesting feature of equal numbers of positive and negative eigenvalues when counted with multiplicity .
we will also provide first results and conjectures on the integrality , irreducibility and maximal eigenvalue of the killing form matrix , collected in section 4 .
we also explore some ideas relating to the open problem of associating an irrep to a conjugacy class .
the underlying idea is that for a simple lie algebra the adjoint representation is irreducible , so if a conjugacy class is like a ` lie algebra ' we might expect it to principally consist of an irrep when the group is simple or close to it ( such as the symmetric group ) .
we are thus motivated to speculate that for @xmath0 and all simple groups where the roth property holds one can bijectively associate irreps to conjugacy classes containing them .
we illustrate this in section 4 for @xmath22 and the mathieu group @xmath17 . in section 5
we look in detail at @xmath0 , where we rapidly establish that such a bijection , if it exists , is not any obvious one such as taking the young diagram associated to a conjugacy class and constructing a specht module from the transposed diagram .
however in section [ signrep ] we find that the related question of which conjugacy classes contain the sign representation has a very natural answer .
we note that a first attempt at a geometrical association between conjugacy classes and irreps via the differential calculus was made in @xcite and was equally tentative .
an alternative strategy suggested by our positive - definiteness conjecture is to fix the noncommutative differential geometry as defined by such conjugacy class and to then view different conjugacy classes as leading to irreps in a less direct way as some kind of ` orbit method ' .
finally , section 6 collects the computer data testing the ideas and conjectures in earlier sections on nonabelian finite simple groups up to order 75,000 .
in this section we introduce the basic elements of noncommutative differential geometry of finite groups @xmath2 in the hopf algebraic approach .
it is quite important that these constructions are not invented in an ad - hoc manner just for groups but that they are part of a general quantum groups approach to noncommutative differential geometry .
suffice it to say on that front that differential structures over any unital algebra @xmath23 are expressed as a specification of the space @xmath24 of 1-forms as an @xmath25-bimodule equipped with a differential @xmath26 obeying the leibniz rule .
we also require that the map @xmath27 sending @xmath28 is surjective and , optionally ( one says that the calculus is connected ) that @xmath29 is spanned by @xmath30 . when @xmath23 is a hopf algebra or ` quantum group ' we can require the calculus to be covariant under left or right translation , or both . in the latter case one says that the calculus is bicovariant .
one knows that a ( say ) left - covariant calculus is a free module over its space @xmath31 of left - invariant differentials . in the bicovariant case
there is a canonical extension to a full exterior algebra @xmath32 or ` noncommutative de rahm complex ' .
every unital algebra has a universal differential calculus defined as @xmath33 and @xmath34 . in the hopf algebra case
it is automatically bicovariant .
any other calculus is isomorphic to a quotient of the universal one .
specialising these ideas , differential calculi on finite sets are in correspondence with directed graphs @xcite . in the case of a finite group @xmath2
the left - covariant calculi are in correspondence with cayley graphs defined by non - empty @xmath35 with @xmath36 ( @xmath6 the group identity ) . here
the edges are of the form @xmath37 for @xmath38 and @xmath39 .
such a calculus is also right covariant ( hence ` bicovariant ' ) _ iff _
@xmath15 is ad - stable .
the basis of invariant 1-forms is @xmath40 and @xmath41 where @xmath42 and @xmath43 are right translation and ` left - invariant derivative ' respectively .
explicitly , @xmath44 in terms of kronecker @xmath45-functions at @xmath38 .
the higher @xmath46 are determined in the simplest ` woronowicz ' formulation by skew - symmetrization with respect to a braiding operator @xmath47 the calculus is connected i.e. has @xmath48 iff the graph is connected , which is iff @xmath15 is a generating set .
we refer to @xcite for an introduction to the theory . a nontrivial application of this theory to finite groups appeared in @xcite where the exterior algebra was linked to the fomin - kirillov algebra arising in the study of the cohomology of flag varieties .
the theme there is a kind of extension of schur - weyl duality to a duality between the geometry of the flag variety and that of the associated weyl group . a finite group @xmath2 is simple _ iff _ all its bicovariant calculi are connected .
suppose that @xmath2 is simple and @xmath15 an ad - stable subset ( defining a bicovariant calculus ) .
let @xmath49 the subgroup generated by @xmath15 .
this is clearly normal and contains more than @xmath6 ( as @xmath15 is nonempty ) , hence @xmath50 and the calculus is connected .
conversely , suppose that all nonempty ad - stable subsets @xmath15 generate @xmath2 .
let @xmath51 be normal and @xmath52 .
this is an ad - stable subset and @xmath53 as @xmath54 is a normal subgroup , hence @xmath50 .
next , in the case of finite sets , a morphism @xmath55 between two differential calculi ( a bimodule map forming a commutative triangle with the corresponding @xmath56 ) is fully determined if it exists .
this is because @xmath57 and hence sends an edge of the graph @xmath58 of @xmath24 to the same edge in the graph @xmath59 of @xmath60 if this edge exists , or else to zero .
hence such a map is a surjection iff @xmath61 . because of this
, we see that @xmath15 being a conjugacy class corresponds to @xmath24 having no proper quotients . in order to develop a lie theory , however , we need something slightly different .
namely , we impose connectedness from the start and in this case the ` minimal ' differential calculus with no connected proper quotients is not necessarily given by a conjugacy class , for example , if the group is abelian .
dually , @xmath15 could be required to be minimal but still generating .
a _ minimal lie structure _ on a finite group @xmath2 means a minimal ad - stable generating subset @xmath35 .
a minimal lie structure always exists , namely one can start with @xmath62 ( which defines the universal calculus ) , this is always ad - stable and clearly generates @xmath2 .
we then remove elements of @xmath15 if possible until further removal is no longer possible .
it does not contain @xmath6 as this could be removed and hence @xmath15 would not be minimal . a finite group is simple _ iff _ every nontrivial conjugacy class is a minimal lie structure . in this case the two notions are equivalent .
suppose that @xmath2 is simple and @xmath15 a nontrivial conjugacy class . as @xmath63 is normal and nontrivial we see , as above , that @xmath15 generates @xmath2 .
suppose that @xmath15 is not minimal , so there exists a proper ad - stable generating subset @xmath64 which , in particular , contradicts @xmath15 a conjugacy class .
conversely , suppose that @xmath2 is a finite group such that every nontrivial conjugacy class is a minimal lie structure .
let @xmath65 be a nontrivial normal subgroup and @xmath52 , which is therefore ad - stable .
decompose this into nontrivial conjugacy classes and let @xmath66 be one of these .
then by assumption @xmath66 is a minimal lie structure on @xmath2 , hence @xmath67 . but
@xmath68 as @xmath69 was a subgroup , hence @xmath70 and hence @xmath50 .
hence @xmath2 is simple .
finally , suppose that @xmath2 is simple and @xmath15 a minimal lie structure .
if @xmath71 is a proper ad - stable subset then ( as above ) @xmath72 is a nontrivial normal subgroup and hence @xmath66 generates @xmath2 , which contradicts minimality . hence @xmath15 is a conjugacy class , nontrivial as does not contain @xmath6 . in geometry the dual of the space @xmath31 of left - invariant differential 1-forms on a lie group
has the structure of a lie algebra .
the corresponding notion in the quantum groups approach to noncommutative differential geometry is that of a ` braided lie algebra ' @xcite .
this is defined abstractly as an object @xmath10 in a braided monoidal category equipped with morphisms @xmath73:{{\mathcal{l}}}{\otimes}{{\mathcal{l}}}\to{{\mathcal{l}}}$ ] ( the ` lie bracket ' ) and @xmath74 , @xmath75 forming a coalgebra in the category ( where @xmath76 is the trivial object ) , subject to several axioms . when the category is abelian , there is a certain quadratic ` enveloping algebra ' @xmath77 associated to the braided lie algebra and forming a bialgebra in the category . of further relevance to us is a ` braided killing form ' defined as the braided or categorical trace of two application of @xmath73 $ ] .
the theory was invented particularly to solve the problem of what is the ` lie object ' underlying the standard drinfeld - jimbo quantum enveloping algebras @xmath78 .
it is shown in @xcite that at least for generic @xmath79 there is a braided lie algebra @xmath80 and an algebra surjection @xmath81 . in the classical limit @xmath82 where @xmath83 recovers the classical lie algebra in an extended form where @xmath84={\rm id}$ ] and @xmath85=0 $ ] .
an introduction to the theory is in @xcite .
this theory was connected to noncommutative differential geometry in @xcite , where the general theorem is that for any ` inner ' bicovariant differential calculus @xmath24 on a coquasitriangular hopf algebra , there is a braided lie algebra @xmath10 isomorphic as an object in a braided category to @xmath86 .
this is the case for all standard quantum group coordinate algebras @xmath87 $ ] and completes the picture above .
the term ` inner ' refers to a bi - invariant 1-form @xmath88 that generates the exterior derivative in the form @xmath89 $ ] , a concept that has no analogue in classical differential geometry .
one could view the classical limit @xmath83 as corresponding to a nonstandard higher order differential structure possible within noncommutative geometry .
now let @xmath90 be a bicovariant differential calculus defined by an ad - stable subset @xmath15 and with space of left - invariant 1-forms @xmath31 .
let @xmath91 be dual basis to the classes of @xmath92 providing the basis of @xmath31 .
the bi - invariant element @xmath93 makes the calculus inner and the associated ` braided lie bracket ' on @xmath94 is @xmath95=x_{aba^{-1}}$ ] .
the coproduct and counit take the form @xmath96 , @xmath97 and the category is that of @xmath2-modules with trivial braiding , the usual ` flip ' .
the associated ` enveloping algebra ' @xmath77 is a quadratic algebra with generators @xmath91 and relations @xmath98 .
this comes equipped with a canonical homomorphism @xmath99 to the group algebra , defined by @xmath100 . in our case
because the underlying braiding is trivial the braided trace becomes the usual trace .
then the killing form is @xmath101)=| z(ab)\cap { { \mathcal{c}}}|=\chi_{{\mathcal{l}}}(ab)\ ] ] where @xmath13 is the centraliser of @xmath14 and @xmath11 is the character of the conjugation representation on @xmath10 .
clearly @xmath21 is ad - invariant since @xmath15 is and is @xmath102-symmetric , hence @xmath103 .
it is also actually symmetric since @xmath104 since @xmath11 is a class function .
we are interested in when @xmath21 is nondegenerate .
if @xmath105 for some nontrivial @xmath106 then @xmath21 is degenerate . in particular , if @xmath107 then @xmath21 is degenerate .
looking at @xmath21 as a matrix with rows and columns labelled by @xmath15 . if @xmath108 where @xmath109 and @xmath110 then @xmath111 for all @xmath39 , hence @xmath21 has a repeated column .
if @xmath112 are distinct then @xmath113 fits the first part . if @xmath114 and @xmath115 then the killing form for the universal calculus is degenerate . for the universal calculus @xmath62 and in the preceding lemma we can take any nontrivial @xmath106 , any @xmath116 and @xmath108 . then @xmath117 .
if @xmath2 has order 2 then @xmath21 _ is _ a @xmath118 matrix and _ is _ nondegenerate .
we therefore only need to investigate the universal calculus in the case where @xmath114 and @xmath119 , for example when @xmath2 is simple and nonabelian .
let @xmath2 be a finite group .
we are going to work from the expression ( [ killing ] ) derived above , and we note first that this formulation actually makes sense for a ` killing form ' similarly defined for any representation @xmath120 and on all of @xmath121 , @xmath122 where @xmath123 is a basis of @xmath121 .
it is well - known that such a symmetric bilinear form is nondegenerate if and only if @xmath120 contains every irrep of @xmath2 with positive multiplicity .
this follows from semisimplicity of the group algebra @xmath121 and general facts about semisimple algebras .
( namely , if an algebra is semisimple then it is a direct sum of matrix blocks . if @xmath124 for some multiplicities @xmath125 then @xmath126 has a block form with @xmath125 times the euclidean inner product on each matrix block .
this is because an element in a matrix block corresponding to an irrep acts as zero by left multiplication on any other block and hence in any other irrep .
hence @xmath126 is nondegenerate iff all the @xmath127 ) .
the case @xmath128 with the conjugation representation is therefore nondegenerate
the conjugation representation contains every irrep .
if @xmath4 is trivial then this is the roth property .
for the case of the universal calculus we are interested in the case @xmath129 where we remove the group identity .
we are also interested in restricting the killing form to @xmath130 but we defer this for the moment .
let @xmath2 be a finite group .
suppose @xmath121 under conjugation contains every irreducible of @xmath2 with strictly positive multiplicity .
then so does @xmath129 .
@xmath120 spanned by set @xmath131 still contains a copy of the trivial representation since this set is permuted by conjugation and hence the element @xmath132 , where we sum over @xmath133 , is invariant .
as @xmath134 as a @xmath2-module , if any nontrivial representation is contained in @xmath121 then it must also be in @xmath120 .
hence if the roth property holds for @xmath2 then it also holds for our particular @xmath120 of interest .
let @xmath1 be the number of distinct irreducible representations of @xmath2 , and call these representations @xmath135 .
[ prop ] let @xmath2 be a finite group and @xmath120 be a representation of @xmath2 for which every irreducible representation @xmath136 appears in @xmath120 with strictly positive multiplicity .
then the restriction of @xmath126 to @xmath137 is nondegenerate . since the form @xmath126 is nondegenerate on @xmath121 we know that @xmath138 is one - dimensional
. we will know that @xmath126 is nondegenerate in @xmath130 if there is no element in the perpendicular which also lies @xmath130 .
we prove this by determining explicitly a vector spanning the line @xmath138 , and observing that it does nt lie in @xmath130 .
define @xmath139 this is well - defined since @xmath140 for all @xmath141 by our assumption .
we claim that @xmath142 for all @xmath143 .
note that the coefficient of @xmath144 in @xmath145 is given by the formula @xmath146 so is always strictly positive .
therefore @xmath145 does not lie in @xmath130 and the claim will imply the proposition .
we use the following standard orthogonality formulas : @xmath147 and @xmath148 note also that @xmath149 now we can compute , extending linearly , @xmath150 the last expression vanishes whenever @xmath143 by . let @xmath2 be a finite group for which the roth property holds . then the killing form for the universal differential calculus on @xmath151 is nondegenerate .
as explained in the introduction , the roth property holds for most simple nonabelian groups including all the sporadic ones , see @xcite .
meanwhile , the simplest example where the roth property holds is @xmath152 , the group of permutations on three elements . this is elementary enough that we can , instructively , compute everything in our characters approach by hand . here
@xmath4 is trivial .
let @xmath152 .
the killing form on @xmath121 is @xmath153 in a basis @xmath154 and @xmath155 .
we obtained by writing out the group product table and evaluated @xmath156 for each entry @xmath157 . here
@xmath158 one can then see by computation that the lower right @xmath159 block in @xmath126 is invertible as required by the proposition . to see how this arises from the character theory in the proposition , we look at the trivial representation @xmath160 , the sign ( parity function ) representation @xmath161 and the standard 2-dimensional representation , denoted @xmath162 .
their character values tabulated on conjugacy classes as for @xmath163 are @xmath164 @xmath165 @xmath166 from which we see that @xmath167 .
thus the ` roth property ' holds for @xmath120 as in the lemma above , hence @xmath126 on @xmath121 is nondegenerate .
we also view this last table as the change of basis with the characteristic class functions @xmath168 with support @xmath169 , @xmath170 with support @xmath171 and @xmath172 with support @xmath173 .
now , by direct computation it is clear that @xmath174 viewed as the coefficients of an element of @xmath121 lies in @xmath175 and hence spans this 1-dimensional space . using the table to convert back to the basis of characters ,
we have obtained ` by hand ' an element @xmath176 in the perpendicular complement .
this agrees up to normalisation with the canonical element constructed in the proof of the proposition , namely @xmath177 where the coefficient of @xmath178 is @xmath179 .
finally , @xmath180 has by construction a non - zero coefficient of @xmath6 , so @xmath180 does not lie in @xmath130 .
in fact we conjecture that the conclusion of proposition [ prop ] holds even when the roth property fails .
[ conjuni ] let @xmath2 be a finite simple nonabelian group and @xmath21 the killing form for the universal differential calculus .
then @xmath21 is nondegenerate .
experimental evidence that this holds for the finite simple groups beyond those implied by proposition [ prop ] is limited as even the smallest of the simple groups such as @xmath181 where the roth property is known to fail in @xcite require too much computer memory to check directly .
however , we will have evidence for a more general conjecture where the roth property for the relevant representation and hence proposition [ prop ] do not apply but where the killing form is still nondegenerate .
in particular , we will see in section 6 that nondegeneracy holds for all conjugacy class differential calculi on @xmath181 . also , just as the roth property holds more generally than for certain simple groups , eg for @xmath19 above , we similarly further conjecture that the nondegeneracy may hold more generally for finite groups with trivial centre .
finally , we also conjecture nondegeneracy at the other extreme of ` minimal ' differential calculi .
note that for a simple group any conjugacy class will be generating and define a connected differential calculus .
[ conjcon ] let @xmath2 be a finite simple nonabelian group and @xmath5 a conjugacy class closed under inversion .
then the killing form @xmath21 on @xmath182 is nondegenerate .
this is verified by computer in section 5 for groups up to order 75,000 .
this includes the first sporadic group @xmath17 and the exceptional group of lie type @xmath18 . the requirement on having inverses is included since the experimental evidence shows examples of conjugacy classes without this assumption and where @xmath21 is degenerate ,
although those examples are scarce and generally one has nondegeneracy even without requiring closure under inversion .
conjecture [ conjcon ] and our previous conjecture [ conjuni ] suggest further that nondegeneracy holds for simple @xmath2 and any ad - stable subset @xmath5 closed under inverses , but we have no direct evidence for this .
we also further conjecture that nondegeneracy can hold more generally for groups which may have centre but where @xmath183 is empty and @xmath15 generates the group . for example , @xmath0 with its 2-cycles conjugacy class ( which generates the group ) has nondegenerate killing form yet the analogue of the ` roth property ' clearly does not hold for @xmath184 as most irreps are not contained in it .
we will prove these assertions in section 5 , but for now we directly look at the simplest case of @xmath19 to illustrate why we require @xmath15 to generate the group . for @xmath152 and @xmath15 the order 3 conjugacy class of 2-cycles ( which generates the group )
, the killing form is nondegenerate . here
@xmath2 acting by conjugation permutes the elements of @xmath185 and hence @xmath186 .
this does not have the roth - type property used in proposition [ prop ] and indeed on @xmath121 we have @xmath187 on the three classes and hence @xmath188 in basis order @xmath189 as before , which is degenerate .
however , the @xmath190 submatrix arising from its restriction to @xmath15 is a multiple of the euclidean inner product @xcite and is invertible .
incidentally , its restriction to the other class and to the union of the two nontrivial classes is also nondegenerate . for @xmath152 and @xmath15 the order @xmath191 conjugacy class of 3-cycles ( which does not generate the group ) , the killing form is degenerate .
here each element @xmath192 acts to flip the elements of @xmath15 while @xmath193 of course act trivially .
hence @xmath194 again does not have the roth - type property . in this case
@xmath195 on the three classes .
hence @xmath196 which is degenerate .
its @xmath197 submatrix corresponding to the restriction to @xmath15 is also degenerate ( as is its restriction to the other nontrivial conjugacy class and to the sum of the two classes ) .
we now look at some general results about the killing form of particular relevance to the case of a conjugacy class differential calculus .
let @xmath2 be a a finite group and @xmath15 an ad - stable subset of @xmath131 .
we note that the ` euclidean metric ' @xmath198 given by the kronecker delta - function in basis @xmath15 is also ad - invariant .
we regard both @xmath199 as maps @xmath200 and hence the composite @xmath201 , which we also denote @xmath21 , is a map @xmath202 . here
explicitly @xmath203 by construction this is also ad - invariant .
hence its eigenspaces provide a natural decomposition of @xmath182 into subrepresentations .
note that as @xmath21 is real and symmetric in our basis it has a full diagonalisation with real eigenvalues over @xmath204 .
however , it can also be viewed as a hermitian matrix or self - adjoint operator over @xmath205 .
also the entries of @xmath21 are non - negative integers .
suppose @xmath206 is an irreducible representation of @xmath121 which is defined over @xmath207 .
so @xmath208 for an irreducible representation @xmath209 of @xmath210 .
furthermore , suppose @xmath211 is a conjugacy class in @xmath2 such that @xmath206 occurs in the conjugation representation @xmath212 .
if the isotypical component of @xmath206 in @xmath212 is contained in a single eigenspace of the killing matrix @xmath21 of @xmath211 , then the corresponding eigenvalue lies in @xmath213 .
choose an element @xmath214 and consider the map @xmath215 which is a @xmath2-equivariant surjection from the left - regular representation to the conjugation representation of @xmath2 .
let @xmath216 denote the block of the irreducible representation @xmath206
. by block decomposition of @xmath121 and schur s lemma it follows that @xmath217 restricts to a surjection from @xmath218 to the isotypical component of @xmath206 in @xmath212 .
however , since @xmath206 was defined over @xmath207 it follows that @xmath219 has a basis that lies inside @xmath210 .
moreover by surjectivity of @xmath220 onto the isotypical component there exists such a basis element @xmath221 whose image @xmath222 is nonzero . by the assumptions
, @xmath221 is an eigenvector of the killing matrix .
moreover @xmath221 has rational coefficients as a vector in @xmath212 .
since the entries of @xmath21 are integral and @xmath221 is rational it follows that the eigenvalue of @xmath221 lies in @xmath207 . on the other hand
the integrality of @xmath21 implies that the eigenvalues are all algebraic integers .
so the eigenvalue of @xmath221 is a rational number and an algebraic integer .
therefore it must lie in @xmath213 .
note that this proposition implies in particular , that if @xmath206 is a complex representation of @xmath2 defined over @xmath207 which occurs in @xmath212 with multiplicity @xmath30 , then it lies in an eigenspace of @xmath21 with eigenvalue in @xmath213 . since all representations of the symmetric group are defined over @xmath207 ( over @xmath213 even )
, we have the following corollary .
let @xmath15 be a nontrivial conjugacy class of @xmath0 .
if an irreducible representation of @xmath0 occurs in the conjugation representation @xmath182 with multiplicity one , then it embeds into an eigenspace for the corresponding killing form with eigenvalue in @xmath213 . note
that an irreducible representation is rational if all its character values are rational .
this is because the matrix entries can be obtained by projection via central idempotents in the group algebra with coefficients defined by the characters .
similarly the character determines if an irreducible representation is complex in the sense of not real .
let @xmath5 be an ad - stable subset . 1 .
if a complex irreducible representation @xmath206 occurs in @xmath182 in an eigenspace of the associated killing form matrix , then so does its dual representation ( with complex conjugate character ) .
if we consider the inverse conjugacy class @xmath223 then the eigenvalues of the killing form matrix for @xmath223 are the same as the ones obtained for @xmath15 , and the decompositions of the respective eigenspaces into irreps are isomorphic .
let @xmath15 be an ad - stable subset .
the conjugation representation is clearly defined over @xmath204 by working in the basis @xmath15 ( indeed , over @xmath213 ) . since @xmath21 is real and symmetric in the basis @xmath15 its eigenspaces are also defined over @xmath204 , and hence real as subrepresentations of the conjugation representations .
this implies the first part . for the second part we consider inversion as a bijection between the two ad - stable subsets .
let @xmath224 . clearly @xmath225 commutes with @xmath226 _ iff _
@xmath227 commutes with @xmath228 .
but as the killing forms are symmetric , we see that the killing forms have the same matrices in their respective bases .
if @xmath229 is expanded in the basis @xmath15 we define @xmath230 to be the corresponding vector in @xmath231 with the same coefficients in the corresponding basis , i.e. @xmath232 are represented by the same column vector in their respective bases .
one may readily see that the matrices for the action of an element of @xmath233 in the two cases are also identical .
this implies the second part .
one can extend the first part to a general galois theory argument that if an irrational representation occurs then so do its various ` conjugates ' , with the corresponding change of galois root in the eigenvalue ( we will see this for @xmath22 below ) .
next , we look at the trivial representation in @xmath182 canonically present as spanned by the sum of the elements of @xmath15 . by a slight abuse of notation we denote this element by @xmath88 ( its analogue as a left - invariant 1-form makes the calculus inner ) . we recall that a matrix with non - negative entries is called _ irreducible _ if for all indices @xmath234 there exists @xmath235 such that the matrix entry @xmath236 .
this is equivalent to connectedness of the graph on the set of indices defined by an edge whenever the entry @xmath237 .
[ theta ] let @xmath2 be a finite group and @xmath5 a conjugacy class .
then @xmath21 has a ( positive ) integral maximal eigenvalue @xmath225 , given by the sum of any row of @xmath21 .
moreover , @xmath21 splits onto @xmath238 irreducible components _ iff _ the eigenspace associated to @xmath225 has dimension @xmath238 and in this case all other eigenspace dimensions are divisible by @xmath238 . in particular ,
if @xmath21 is irreducible then the eigenspace associated to @xmath225 is 1-dimensional , generated by the eigenvector @xmath239 .
@xmath240 where @xmath241 is the sum of the @xmath221th column of the matrix of @xmath21 .
however , @xmath242 after a change of variables .
hence @xmath241 is independent of @xmath243 in the case of a conjugacy class .
hence @xmath88 is an eigenvector of @xmath21 with eigenvalue as stated .
moreover , if @xmath21 is irreducible then by perron - frobenius theory there is a 1-dimensional maximal eigenspace with eigenvalue the column sum of @xmath21 , i.e. with eigenvector @xmath88 .
if @xmath21 is not irreducible then after a reordering of the basis it can be presented as a direct sum .
iterating this , we reduce @xmath21 to a direct sum of some number @xmath244 of irreducible blocks .
in fact each block will be , after reordering , a copy of the same irreducible matrix .
this follows from ad - invariance of @xmath21 as follows .
consider an element in @xmath2 that conjugates a corner of the first block to the corresponding corner of another .
all the indices relating to the first block belong to the same connected component of the graph and , by assumption , they are not connected to any of the indices for the other blocks , and this notion is ad - invariant , as @xmath21 is .
hence the indices relating to the conjugated first block must be connected to themselves and not to the first block .
hence the first block maps over to the conjugated block , and all its entries are the same when suitably ordered , again by ad - invariance of @xmath21 . once @xmath21 has been presented as @xmath238 blocks @xmath245 , its eigenvectors will consist of @xmath238 parts forming eigenvectors for each block with the same eigenvalue . however , since these blocks are all irreducible and have the same row sum as @xmath21 , they will each have the same maximal eigenvalue as @xmath21 , and any other eigenvalues will be strictly lower .
this implies the facts stated .
note that the diagonal of @xmath21 is always non - zero as @xmath246 commutes with @xmath247 for all @xmath248 .
hence @xmath249 can only have the same or more positive entries as @xmath250 , so in our case irreducible is equivalent to the existence of @xmath235 such that all entries of @xmath250 are positive , i.e. the _ primitivity _ of the matrix @xmath21 .
let @xmath2 be a finite nonabelian simple group and @xmath5 a conjugacy class not consisting of involutions .
then the associated @xmath21 is irreducible .
this is surmised by looking at finite simple groups up to order 75,000 .
the only observed reducible cases are the classes of involutions for @xmath251 , @xmath252 or @xmath253 for @xmath254 up to the order that we could check .
these are all groups of lie type over finite fields of characteristic 2 and we further conjecture that this is always the case .
results for @xmath0 will be discussed in section [ sn ] .
next , we consider the question : for which conjugacy classes is @xmath21 not only nondegenerate but positive definite ? we have explained in the introduction that if @xmath182 is some kind of braided lie algebra we might expect a simple group to have a unique ` compact real form ' , i.e. a choice of @xmath182 with @xmath21 positive definite . whilst this is not generally true
, we can conjecture a close enough result : [ conpos ] let @xmath2 be a finite nonabelian simple group and @xmath5 a conjugacy class closed under inverses .
if @xmath15 consists of involutions , then the killing form is positive definite .
otherwise the killing form has zero signature ( i.e. the same number of positive and negative eigenvalues counted with multiplicities ) .
this is supported by computer verification to order 75,000 .
note that by the feit - thompson theorem every ( nonabelian ) finite simple group must contain involutions , so a conjugacy class in the terms of the conjecture will always exist .
uniqueness is not true in general , since various simple groups ( eg . the altenating groups @xmath255 for @xmath256 ) contain several different conjugacy classes of involutions .
there are also some examples , see the conjugacy classes 3a and 3b in @xmath257 in section 6 , that yield positive definite killing forms without being closed under inverses .
we also expect a behaviour similar to the conjectured one to hold for groups such as @xmath0 with @xmath15 generating , see the next section .
the second part of conjecture [ conpos ] is not supported by the analogy but it perhaps reminiscent of the split real form in lie theory . finally , bearing in mind that a braided - lie algebra in the case of a classical lie group has the form @xmath258 where @xmath259 is in the adjoint representation and is irreducible when the lie group is simple
, we might hope that @xmath182 will decompose in a similar way and that this might produce an irreducible representation analogous to @xmath259 after removal of the trivial representation .
this is not exactly what happens but we propose to use the killing form to pick out an irreducible component in @xmath182 , and we shall explore this for @xmath0 with its 2-cycles class in the next section . by thinking of any conjugacy class as a braided - lie algebra and using the killing form we might hope to associate irreps to conjugacy classes in a natural way .
this is not so much a matter of precise conjectures at the moment but a programme in the form of a couple of ` principles ' which we propose .
( nondegeneracy principle . ) for large finite nonabelian simple groups @xmath2 the decomposition into eigenspaces of @xmath21 is typically into irreps or into complex conjugate pairs of irreps .
this is certainly not always true but we shall see how it works for the @xmath17 group . note that the strong form stated implies that where an irrep occurs with multiplicity then these copies are generally distinct as well , i.e. @xmath21 typically resolves or ` separates ' the isotypical components in the process .
( correspondence principle . )
let @xmath2 be a finite nonabelian simple group for which roth s property holds .
there is often a ` reasonable ' way up to choice of galois and complex conjugates to bijectively assign nontrivial irreps to non - trivial conjugacy classes containing them in the conjugation representation , in such a way that if a complex irrep is assigned to a conjugacy class not stable under inversion then its complex conjugate is assigned to the inverse class , and where possible the assigned irrep occurs without multiplicity .
we certainly find examples where avoiding multiplicities is not possible , and in this case a weaker requirement is to have a distinct eigenvalue of @xmath21 for the chosen irrep or complex conjugate pair of irreps .
@xmath260 @xmath261 table 1 provides a first impression by looking at @xmath22 . we used sage and some deductive reasoning to determine the decomposition of conjugacy classes into irreps and their eigenvalues .
there are some accidental degenerations , one of them because @xmath21 is reducible , but otherwise the eigenspaces of @xmath21 are irreps or sums of ` conjugate ' irreps ( though not in this case complex conjugate ) . and in the last column we have also selected an irrep to associate to each conjugacy class in such a way that there is a 1 - 1 correspondence .
this is a more or less unique assignment if we want the 5 and the 4 to enter in their respective conjugacy classes without multiplicity and _ if _ we intuitively require the ` conjugate ' pair of 5-cycle classes to map to the ` galois conjugate ' pair of irreps ( which of the pair of classes is assigned to which of @xmath262 is a residual ambiguity ) .
the eigenvalues have been deduced from the knowledge ( see above ) that the trivial representation has maximal eigenvalue followed by comparison of dimensions between eigenspaces and irreps .
this method does not fully identify the eigenvalue when distinct irreps with the same dimension occur , but in the case of @xmath262 this is a matter of definition as to which is called which . however , having defined one as @xmath263 one may verify by explicit calculation that it appears in the last line of the table with the other eigenvalue .
the eigenvalues in these classes are not integer and indeed the eigenspaces are examples of real irrational representations ( however , when they both appear with the same eigenvalue then this is an integer ) . also note that there is a unique conjugacy class where @xmath21 is positive definite and the equal dimensions of the combined positive and negative eigenspaces in each of the other cases .
we contrast both features with @xmath20 , also in the table , where aside from the trivial representations all eigenvalues are zero .
note that @xmath20 is _ not _ simple and indeed there is no way to associate nontrivial irreps and nontrivial conjugacy classes containing them .
table 2 @xmath264 illustrates that the same ideas hold in the simplest sporadic group , @xmath17 , again using sage and deductive reasoning to construct the table . here
again we find for the most part that the eigenspace decomposition is generally a decomposition into irreps or conjugate pairs of irreps ( the sole exception is the 10,11 irreps in the class @xmath265 ) .
we use atlas labelling of conjugacy classes as in section 6 .
we also find a more or less unique 1 - 1 correspondence between conjugacy classes and irreps as shown , as follows .
we see as explained above that inverse pairs of conjugacy classes have the same spectrum and for any complex conjugate pair of irreps if one occurs in a conjugacy then so must the other and with the same eigenvalue . in view of this , inverse pairs of classes
are naturally assigned to a pair of complex conjugate irreps as shown ( which of each pair goes to which is then largely a matter of definition ) .
this forces us to assign @xmath266 as shown as it does not occur in the other inverse pair , and hence @xmath267 to that .
both then occur without multiplicity . of the remaining nontrivial irreps
the only one which occurs without multiplicity in the 2nd row of the table is 11 , so we are led to assign that . similarly in the class @xmath268 , only the 10 occurs without multiplicity so we assign that . of the remaining classes
only the 55 appears in the class @xmath265 without multiplicity , so we assign that .
however , the assignment of the 44 and 45 is not determined by these arguments and this represents the residual ambiguity .
they both occur with multiplicity in the remaining two conjugacy classes .
one of these , the @xmath269 , has no nontrivial irreps occurring with integer eigenvalues at all , so this provides no guide either .
note also the unique conjugacy class @xmath265 that has positive definite @xmath21 and the equal dimensions of the combined positive and negative eisgenspaces in each of the other classes , other than the mutually inverse pair @xmath270 and the mutually inverse pair @xmath271 .
table 2 also shows the need for a more compact account of the killing form eigenvalues and we do this in figure 1 in the form of a spectrogram .
this shows all the eigenvalues of the irreps occurring in the various conjugacy classes together as an overall picture or bar - code of the group , as well as an example of the contribution from just one class .
complex conjugates have the same spectrum so we do not list them and the thickness of lines indicates a preponderance of nearby eigenvalues .
the combined spectrum is that of the direct sum @xmath272 of all the different @xmath21 acting on a hilbert space @xmath273 which is the direct sum of all the spans of conjugacy classes .
the latter if we include the span of @xmath169 carries a faithful representation of the algebra of functions @xmath151 and the triple @xmath274 can be viewed as a discrete ` spectral triple ' cf . @xcite
although obeying a more general set of axioms . this is a direction to be explored more fully in a sequel .
one may compute @xmath275=\sum_{g\in g\setminus\{e\}}e_g { { \partial}}^g f,\quad e_g(x_a):={|{{\mathcal{c}}}|\over|g|}k(x_a , x_{gag^{-1}})x_{gag^{-1}},\quad fx_a:=f(a)x_a\ ] ] for all @xmath276 and @xmath248 . here
@xmath277 defines ` vector fields ' for the conjugation action .
we also have @xmath278 for the noncommutativity of 1-forms and functions . in this sense
@xmath279 $ ] then has the form of an exterior differential with vielbein lengths given by @xmath21 .
the @xmath280 weighting is to adjust for overcounting by the order of the isotropy group and could be omitted .
@xmath281{m11all.pdf}\ ] ] @xmath281{m115a.pdf}\ ] ] finally , for completeness we mention that a more usual application of the killing form in lie theory is to the construction of the quadratic casimir as a central element in the enveloping algebra .
we can indeed do something similar whenever @xmath21 is nondegenerate . when the killing form @xmath21 is nondegenerate we let @xmath282 be its inverse matrix in basis @xmath15
. then the quadratic casimir is defined as @xmath283 .
its image in the group algebra @xmath284 is also called the casimir element . unlike @xmath21 itself
, @xmath285 defines an operator not only on @xmath182 but on every representation .
we can of course use its values to induce a labelling of irreducibles . in the case of @xmath0
this is hardly necessary but could be useful for other finite groups equipped with nondegenerate killing form , much as in lie theory . on the other hand we can also decompose @xmath15 in terms of other ` @xmath88 elements ' consisting of sums over other conjugacy classes .
so this coordinatisation is a change of basis from using the various @xmath88 for different conjugacy classes , i.e. from using the so - called ` central characters ' for the pairing of an irrep with a conjugacy class . here
the value of @xmath286 in an irrep @xmath206 is given by @xmath287 however , more geometrically , the action of the quadratic casimir is the laplace - beltrami operator and this is analogous to the ability to view a classical quadratic differential as a ` linear ' differential with respect to a different ( noncommutative ) differential calculus .
for @xmath22 with its 2 - 2-cycles calculus we have @xmath21 reducible into 5 identical @xmath190 blocks @xmath245 in a suitable basis order , with inverse of each block @xmath288 from this one finds @xmath289 involving the same conjugacy class as defining the calculus .
by contrast , for @xmath290 with its 2-cycles differential calculus we follow the above steps starting with the killing form .
this is also reducible and consists of 3 identical @xmath197 blocks in a suitable basis order with @xmath291 and resulting quadratic casimir @xmath292 this is a combination of @xmath6 and the @xmath88 for the 2 - 2-cycles conjugacy class and is more manifestly ` quadratic ' for our 2-cycles calculus as expected for the natural laplace - beltrami operator on the finite group .
although we are primarily interested in simple groups , the symmetric groups are so sufficiently close and more readily computable that they are useful as a first indication . also in noncommutative geometry @xmath19 in particularly with its 2-cycles class has the same noncommutative de rham cohomology as @xmath293 $ ] and has the flavour of some kind of discrete model of that . in this section
we develop our theory for general @xmath0 .
first we note that in the case of @xmath0 for @xmath294 , with the 2-cycles conjugacy class , one can see from the formulae for the killing form in @xcite that @xmath21 itself has all entries strictly positive .
hence the proposition [ theta ] applies in this case and there is a unique maximal eigenvalue , with eigenspace spanned by @xmath88 . for @xmath19 and @xmath290 , @xmath21 is reducible , and @xmath88 is a maximal eigenvector but each eigenvalue has multiplicity 3 ( as one can see from table 3 below ) .
our main result in this section is , for @xmath0 with its 2-cycles class , an explicit decomposition of @xmath182 into irreducible representations in a manner compatible with the eigenspace decomposition under @xmath21 .
we also find explicit formulae for the eigenvalues . as an application
we obtain that the killing form matrix @xmath21 for this conjugacy class is positive definite . additionally at the end of the section we give a complete analysis for which conjugacy classes contain the sign representation .
first recall that irreducible representations of @xmath0 , called the specht modules , are indexed by partitions @xmath295 , and a partition is represented by its _ young diagram _ or _
shape_. for example the partition @xmath296 of @xmath297 is represented by @xmath298 the specht module associated to such a partition , or shape , is constructed abstractly for example in @xcite .
we will use another construction @xcite of the specht module @xmath299 , namely as a subrepresentation of the regular representation .
we recall this construction now .
note that , since @xmath299 occurs in the regular representation with multiplicity equal to @xmath300 , the construction for given @xmath301 naturally depends on an additional choice . namely we choose a _ tableau _ of shape @xmath301 , a one - to - one labelling of the boxes by the integers @xmath302 . as usual ,
the symmetric group @xmath0 acts on the set of tableaux by permuting the entries . to the chosen tableau
@xmath303 we associate its ` row subgroup ' @xmath304 , consisting of permutations in @xmath0 preserving the row sets , and its ` column subgroup ' @xmath305 of permutations preserving the column sets .
for example , for @xmath306 , for the tableau @xmath303 of shape @xmath307 given by @xmath308 the row sets are @xmath309 giving the product of permutation groups , @xmath310 and the column sets are @xmath311 , so that @xmath312 to the tableau @xmath303 we can then associate an element of the group algebra @xmath313 called the ` young symmetrizer ' @xmath314 where @xmath315 and @xmath316 is the length of a permutation .
to finish the construction we consider the left ideal in the group algebra defined by the young symmetrizer , @xmath317 this defines a subrepresentation of the left regular representation .
clearly the right action of @xmath0 on @xmath313 provides @xmath0-equivariant isomorphisms between the modules @xmath318 for varying @xmath303 .
the isomorphism of any @xmath318 with the abstract specht module @xmath299 constructed in @xcite is straightforward to write down .
moreover , the sum of the subspaces @xmath318 for varying tableaux @xmath303 is the block of @xmath299 inside @xmath313 .
note that there are many more tableaux than the multiplicity of @xmath299 , so this is not a direct sum .
let @xmath319 denote the set of _ standard young tableaux _
, that is tableaux whose entries are strictly increasing in rows and in columns .
then the block of @xmath299 inside @xmath313 is precisely the subspace @xmath320 we now use this theory to give a concrete construction of irreducible subrepresentations inside a conjugation representation . for any partition @xmath321 of @xmath1
we have a corresponding conjugacy class @xmath322 in @xmath0 , namely the one with cycle type @xmath323 .
explicitly @xmath322 is the conjugacy class containing the element @xmath324 if we let @xmath325 denote the centraliser of @xmath326 and identify @xmath327 via @xmath328 , then we obtain a @xmath0-equivariant homomorphism from the left regular representation to the conjugation representation , @xmath329 coming from the linear extension of the quotient map @xmath330 .
[ l : spechtembeddings ] suppose @xmath301 and @xmath323 are partitions of @xmath1 and all notations as above .
the specht module @xmath299 occurs as a subrepresentation of the conjugation representation @xmath331 if and only if there exists a standard young tableau @xmath303 of shape @xmath301 for which @xmath332 in @xmath331 . in that case
, the subrepresentation is explicitly realized as the subspace @xmath333 , where @xmath217 is the projection map from .
if there is a tableau @xmath303 for which @xmath332 , then the restriction of the map @xmath217 from to the subrepresentation @xmath318 of @xmath313 defines a non - zero @xmath0-equivariant map @xmath334 .
since @xmath318 is irreducible and isomorphic to @xmath299 it follows that this map must be an isomorphism onto its image , and therefore @xmath333 is a submodule of the conjugation representation isomorphic to @xmath299 . on the other hand ,
if @xmath335 for all @xmath336 , then the entire block of @xmath299 in @xmath313 lies in the kernel of @xmath217 , and therefore the irreducible representation @xmath299 does not occur in the image of @xmath217 .
since @xmath217 is surjective , this means that @xmath299 is not a subrepresentation of @xmath331 .
we remark that this lemma also holds , of course , with @xmath337 replaced by any cyclic @xmath313-module , with the same proof .
our first application of this construction is to the the @xmath191-cycles class . in the example of @xmath19 , the @xmath191-cycles
class @xmath338 has three elements and it is straightforward to see that the conjugation representation , @xmath339 , is the ( defining ) three - dimensional permutation representation of @xmath19 . in terms of specht modules
this representation decomposes as @xmath340 that is , the trivial representation plus the standard @xmath191-dimensional representation .
the general case is not much different , as we observe below .
we use the notation @xmath341 for the partition @xmath342 which represents the @xmath191-cycles class in @xmath0 .
consider @xmath0 for @xmath343 with the 2-cycles class @xmath344 .
for @xmath345 the decomposition of @xmath182 into irreducibles is given in equation . 1 . for @xmath346 the decomposition of the conjugation representation @xmath182 into irreducible representations
is given by @xmath347 here the first two specht modules @xmath348 are the trivial representation and the standard @xmath349-dimensional representation , respectively .
each irreducible submodule of @xmath182 lies in an eigenspace for the killing form matrix @xmath21 with eigenvalues as follows .
the eigenvalue of @xmath21 for the eigenspace containing @xmath350 ( spanned by the element @xmath88 ) is @xmath351 the eigenvalue of @xmath21 in the eigenspace containing @xmath352 is @xmath353 suppose @xmath346 .
then the eigenvalue of @xmath21 on the eigenspace containing @xmath354 is @xmath355 .
\(1 ) could be checked using character theory .
but we will rather define explicit embeddings of the specht modules , by the method of lemma [ l : spechtembeddings ] , in order to be able to compute the eigenvalues of @xmath21 in the later parts of the proof .
of course the trivial representation embeds into @xmath356 as the subspace spanned by the element @xmath357 , and has multiplicity @xmath30 . for the standard representation @xmath352 we consider the subspace @xmath358 of @xmath359 for @xmath217 from corresponding to the tableau @xmath360 1&2&3&@xmath361 & @xmath362 @xmath1 this is the submodule of @xmath182 obtained by applying @xmath363 to the vector @xmath364 . up to an overall multiple , which we drop
, this vector works out to be @xmath365 since @xmath366 we have found a copy of @xmath352 in @xmath182 . for the next representation @xmath354 we consider the subspace @xmath367 of @xmath182 for @xmath217 from and the tableau @xmath368 1 & 2&3&@xmath361 & @xmath369 & @xmath1 this is the submodule of @xmath182 obtained by applying @xmath363 to the vector @xmath370 .
up to an overall multiple this vector works out to be @xmath371 and since @xmath372 we have found a copy of @xmath354 in @xmath182 .
that we have thereby completely decomposed @xmath182 follows by dimension count : @xmath373 where @xmath374 is computed for example by the hook formula .
( a prettier proof of the above sum decomposition for @xmath375 into dimensions of specht modules goes via symmetric functions ) .
this concludes the proof of ( 1 ) .
that the irreducible subrepresentations lie in eigenspaces of @xmath21 follows immediately from the fact that in the decomposition of @xmath182 each irreducible representation occurs with multiplicity at most one .
we can now compute the eigenvalues . for the trivial representation we compute the column sum @xmath376 over all 2-cycles .
in the basis of the ` triangular ' listing @xmath377 @xmath378 @xmath379 @xmath380 @xmath381 @xmath382 we have for @xmath246 the choice ( 12 ) , or @xmath246 lies in the size @xmath383 region on the left where @xmath246 has one entry in common with ( 12 ) , or @xmath246 lies in the triangle to the right of size @xmath384 where @xmath246 is disjoint from ( 12 ) . using the values of @xmath21 for these three cases in @xcite , we find @xmath385 which computes as stated . for the standard representation we use the vector we constructed in the proof of ( 1 ) , @xmath386 which involves the left and bottom slopes of the triangle leaving out the common vertex .
then the eigenvalue computed as the coefficient of @xmath387 in @xmath388 is @xmath389 @xmath390 which comes out as stated .
both formulae , although computed for @xmath294 in the above counting , also give the right answer for @xmath391 , as computed by hand . for the representation @xmath354 we use the vector @xmath392 from the proof of ( 1 ) and compute the eigenvalue as the @xmath387 coefficient of @xmath393 ,
i.e. as @xmath394 @xmath395 the killing form for @xmath0 , @xmath343 with the 2-cycles conjugacy class @xmath15 is non - degenerate and in fact positive definite .
moreover the decomposition of @xmath182 into irreps consisting of the trivial and the standard representation , and the representation @xmath354 , coincides for @xmath396 with the decomposition of @xmath21 into eigenspaces of respectively the maximal , next to maximal and smallest eigenvalues .
looking at the three expressions for the eigenvalues in the lemmas above it is evident that they have different leading powers of @xmath1 and hence are distinct for all @xmath1 bigger than some value . by inspection ,
the only degeneracies are @xmath345 when the trivial and the standard representation have the same eigenvalue of @xmath21 , @xmath397 when the eigenvalues of the trivial and the @xmath354 coincide , being smaller than the eigenvalue of the standard representation , and @xmath398 when the eigenvalues of the standard representation and of @xmath354 coincide .
after that , the eigenvalue of the trivial exceeds that of the standard representation which exceeds that of @xmath354 as stated .
as all the eigenvalues are positive we conclude that @xmath21 is non - degenerate ( and positive definite when extended as a hermitian inner product ) .
one useful application of the k eigenvalue decomposition is that when it succeeds it directly provides subspaces of the conjugacy class representation which decompose it into irreps , just by linear algebra and entirely bypassing combinatorics .
we illustrate this for @xmath306 and this also serves as a direct check of the above .
[ example ] for @xmath306 with its 2-cycles conjugacy class , we use the explicit formula for the killing form in @xcite and mathematica to compute the eigenvalues and eigenspaces .
the eigenvalues are @xmath399 the unique ` perron - frobenius ' eigenvector for the @xmath400 eigenvalue is @xmath88 as it must be .
the eigenvectors for the 19 eigenvalue are , with @xmath401 , @xmath402 @xmath403 @xmath404 @xmath405 @xmath406 @xmath407 in a basis ordered as above . on inspection
we can see that conjugation by @xmath408 interchanges the @xmath409 as in the standard representation namely @xmath410 and fixing the others , etc .
the action of @xmath411 is more complicated and comes out as @xmath412 for @xmath413 and @xmath414 where @xmath415 this obeys @xmath416 .
we now change to a new basis @xmath417 which does not change the form of the action of @xmath418 by transposition but now @xmath419 for @xmath413 .
we define @xmath420 to identify the representation as the standard one embedded in @xmath421 .
the eigenvectors for the 14 eigenvalue can similarly be seen to form an irrep , the @xmath354 one .
apart from accidental degeneracies and reversals for small @xmath1 , we see that the next - to - maximal eigenvalue associates to the 2-cycles conjugacy class the standard representation , which is the specht module associated to the transposition of the young diagram describing the conjugacy class .
such an association also assigns the trivial representation to the trivial conjugacy class as expected .
the methods we have developed are general and can be applied in principle to other conjugacy classes allowing us to study which if any irreps are picked out in a stable range with a view towards extending such a 1 - 1 correspondence .
such an extended correspondence , however , can not be the obvious one because we find that not every specht module is actually embeddable in the adjoint representation of the ` expected ' conjugacy class under the association given by transposition of the young diagram .
specifically , we show in section [ signrep ] that the sign representation does not occur in the @xmath1-cycle class for all even @xmath1 . at least for small @xmath1 we can see , however , that an extended correspondence is possible .
a complete analysis for @xmath422 using methods as above is summarised in table 1 .
one could obtain these results from sage but here we obtain them as an illustration of the general theory above to conjugacy classes of different shape .
we focus on @xmath290 and we first realise the relevant specht modules @xmath299 as submodules of the left regular representation @xmath423 explicitly as @xmath424 @xmath425 these give rise to explicit embeddings of these specht modules into the conjugation representation as follows . 1 . for the @xmath426-cycle conjugacy class in @xmath290 we have an explicit embedding of the specht module @xmath427 given by applying @xmath313 to the vector @xmath428 as this vector is non - zero , this 2-dimensional representation is contained and by adding up the dimensions @xmath429 next , the killing form matrix has all entries 3 because the product of any 2 - 2-cycles here is another 2 - 2-cycle and all elements of the commute .
hence @xmath430 so this module has zero eigenvalue .
2 . for the @xmath263-cycle conjugacy class in @xmath290
we similarly find @xmath431 with explicit embeddings of the specht modules @xmath432 and @xmath433 given by applying @xmath313 to the non - zero vectors @xmath434 @xmath435 @xmath436 next , all products of 2-cycles are either @xmath6 , a 3-cycle or a 2 - 2-cycle and we compute the number of fixed points ( the character of the conjugation representation ) as @xmath437 respectively .
hence the killing form in basis order @xmath438 comes out as @xmath439 denoting by @xmath440 the stated eigenvector taken without any overall factor , we compute the eigenvalue from the coefficient of ( 123 ) in @xmath441 .
for example , in the first case @xmath442 .
3 . for the @xmath443-cycle conjugacy class in @xmath290
we similarly find @xmath444 with the explicit embeddings of the specht modules @xmath445 and @xmath446 given by applying @xmath313 to the non - zero vectors @xmath447 @xmath448 this time products of 4-cycles are either @xmath6 , 2 - 2-cycles or 3-cycles with number of fixed points or character values 6,2,0 respectively .
hence the killing form in basis order @xmath449 is @xmath450 we use the eigenvectors @xmath440 above without the overall factors and compute the eigenvalue as the coefficient of @xmath451 in @xmath452 .
for example , in the first case @xmath453 .
looking as the tables we see that for @xmath19 the ` expected ' correspondence holds but the eigenvalues of @xmath21 are degenerate in the case of 2-cycles ( as above ) . for @xmath290 there are also some eigenvalue degeneracies and the 4-cycle representation does not contain the ` expected ' specht module @xmath454 . on the other hand ,
we see that a different assignment of irreps to conjugacy classes is possible in such a way as to give a 1 - 1 correspondence between irreps and conjugacy classes .
@xmath455 @xmath456 we end with a couple of more concrete conjectures for @xmath0 based on limited computer verification to @xmath457 .
they remain a topic for further work . for @xmath0 , @xmath294
the conjugacy classes with reducible @xmath21 are precisely the @xmath458-fold @xmath191-cycles for @xmath1 odd . in this case
the maximal eigenvalue has eigenspace decomposition @xmath459 , where @xmath460 is the standard representation . for all @xmath0 , @xmath294
all conjugacy classes have non - degenerate @xmath21 and of these precisely @xmath461 $ ] are positive definite , namely the classes of @xmath145-fold 2-cycles class for @xmath462 $ ] .
here @xmath463 $ ] denotes integer part .
all the other classes have evenly split signature .
note that of the positive - definite classes , the 2-cycles remain the more natural one for the ` compact real form ' of the braided lie algebra of @xmath0 , with the others viewed as defining higher order differential calculi .
not much is known about decomposition of the conjugation representation of @xmath0 into irreducible representations other than roth s property , that every irreducible representation occurs @xcite if @xmath343 , and some asymptotic estimates for the multiplicities @xcite .
the more detailed question of which conjugacy classes a given irreducible representation occurs in looks even more complicated in general .
however for the sign representation we can give a complete answer , which we do here .
first we explain briefly the overall multiplicity of the sign representation in the conjugation representation @xmath313 . by character theory
this multiplicity is precisely the number of conjugacy classes consisting of even permutations minus the number of conjugacy classes of odd permutations ( the row sum in the character table , for the sign representation ) . in analogy with euler s product formula for the number of all partitions ,
one can write a generating function for these multiplicities . namely , if @xmath464 is the multiplicity of the sign representation in @xmath313 , then the above description of @xmath464 implies @xmath465 clearly , multiplying out the right - hand side of the formula above , we get a contribution @xmath466 to @xmath467 for every partition of @xmath1 .
namely the contribution is @xmath468 for every partition of @xmath1 with an even number of even parts and @xmath469 for every partition of @xmath1 with an odd number of even parts .
the former are in bijection with conjugacy classes of even elements , and the latter with conjugacy classes of odd elements .
this proves the identity . by the classical euler identity which reads ( after replacing the usual variable by @xmath470 and inverting ) ,
@xmath471 it follows that the multiplicity of the sign representation in the conjugation representation @xmath313 is equal to the number of partitions of @xmath1 into distinct odd parts .
we refine this observation in the following proposition , which could alternatively be viewed as giving a representation theoretic proof of the euler identity .
[ p : signrep ] the sign representation of @xmath0 appears as a subrepresentation of the conjugation representation @xmath337 if and only if @xmath323 is a partition of @xmath1 into distinct odd parts . if it appears in @xmath337 , then it has multiplicity one .
since the sign representation has multiplicity one in in the left - regular representation @xmath313 and the conjugation representation @xmath337 is a cyclic @xmath313-module , it is clear that the sign representation can have multiplicity at most @xmath30 in @xmath337 .
let us now write @xmath472 for the conjugation action .
fix an element @xmath326 in the conjugacy class @xmath473 . by lemma [ l : spechtembeddings
] , the sign representation appears in @xmath474 if and only if the element @xmath475 in @xmath474 is nonzero .
moreover if it is nonzero then it spans the sign representation .
now suppose @xmath476 is nonzero and let @xmath477 be an element of the centraliser @xmath325 .
then we see that @xmath478 this implies that @xmath477 is even , since @xmath476 spans the sign representation .
therefore if the sign representation occurs in @xmath337 then @xmath325 contains only even permutations .
the converse is true as well .
if all elements in @xmath479 are even , then the coefficient of @xmath326 in @xmath476 comes out to be @xmath480 , implying that @xmath476 is nonzero , and the sign representation occurs in @xmath481 .
it remains to prove that @xmath325 contains only even permutations , precisely if @xmath323 is a permutation of @xmath1 into distinct odd parts . clearly , if @xmath323 has an even part then there is a cycle of even length in @xmath326 , which gives an element of the centralizer that has odd parity .
also if @xmath323 has two parts of size @xmath482 ( we may assume @xmath482 odd , by above ) , then there is an element of the centralizer which exchanges the corresponding two @xmath482-cycles of @xmath326 , which is a product of @xmath482 many @xmath191-cycles .
so again there is an element of odd parity in @xmath325 .
this shows that for the sign representation to occur inside @xmath337 , we must have that @xmath323 is a partition of @xmath1 into distinct , odd parts .
conversely , if @xmath323 is a partition of @xmath1 into distinct odd parts , then the centraliser is generated by the individual cycles in @xmath326 , and these are all even permutations .
we note that since , as in the proof of the previous proposition , one can always give a precise vector spanning a copy of the sign representation , one can in principle use the methods from above to also find its eigenvalue under @xmath21 . as a very special case ,
note that proposition [ p : signrep ] implies in particular that if @xmath1 is even , then the sign representation does not occur in the class of @xmath1-cycles .
this means that , as already remarked , the naive guess at a bijection of conjugacy classes with specht modules contained within them ( given by transposing the diagram ) does not work in general .
nevertheless , for the cases we have looked there _ is _ at least one alternative assignment .
thus for @xmath0 with @xmath343 it is quite possible that one should be able to find each irrep in a distinct conjugacy class , but clearly some further ideas are needed .
in order to provide empirical supporting evidence for our conjectures as well as get a grip on the behaviour of the killing forms associated to minimal calculi for finite simple groups we have performed an extensive amount of computational verifications .
most of our calculations have been performed using the open source computer algebra system sage in a linux workstation , relying heavily on gap for some internal procedures .
the code of the actual implementation is available from the authors upon request . 1 .
pick a generator @xmath233 of @xmath15 , 2 .
compute a section @xmath485 , such that @xmath486 for all @xmath487 , 3 .
compute the function @xmath488 and cache its values , 4 .
compute @xmath489 as @xmath490 . in practical terms
, the suggested method reduces the computation of the killing form to the computation of the first row and the permutations that create all the other rows from the first one .
the resulting algorithm provides a reasonably quick computation of the killing form , allowing to compute them for all conjugacy classes of all finite simple groups of order up to 75000 .
the limiting factors of the implementation are related to exhaustion of the computer memory rather than computing times , with the first conjugacy class out of our reach being the class 6b of elements of order 6 with centralizer of size 6 in the next larger group , the mathieu group @xmath491 of order 95040 .
the naming of the conjugacy classes is according to the convention in the atlas @xcite .
nondegeneracy of @xmath21 is checked by using sage internal algorithms for computing the rank of @xmath21 .
computationally , finding the rank is quick enough and does nt require substantially more resources than computing the matrix itself . up to order 75,000 the killing form is nondegenerate in all cases but the following exceptions : * the ( mutually inverse ) conjugacy classes 7a and 7b of elements of order 7 in the alternating group @xmath492 , * the ( pairwise inverse ) conjugacy classes 4a , 4b , 8a , 8b , 12a , 12b of elements of orders 4 , 8 or 12 in the unitary group @xmath493 . * the two ( mutually inverse ) conjugacy classes 7a and 7b of elements of order 7 in @xmath494 . as mentioned before , all the degenerate cases occur in conjugacy classes that are not closed under inversion , so reality of the conjugacy class appears to be a sufficient condition ( though by no means necessary ) for the nondegeneracy of @xmath21 .
the irreducibility of @xmath21 is tested by using standard algorithms for checking connectedness of the graph @xmath495 with vertices indexed by elements of @xmath15 and containing an edge @xmath496 if and only if @xmath497 . the size of the graph @xmath495 is substantially smaller than the size of @xmath21 and the connectedness checking is also fast , so again this test does not produce much of an additional overhead . generically , the tested killing forms yield irreducible matrices , so that perron - frobenius theorem applies and the eigenspace associated to the maximal eigenvalue is 1-dimensional ; the only observed exceptions to irreducibility are given by the conjugacy classes of involutions in the linear groups @xmath498 , @xmath499 , @xmath500 , @xmath501 , the exceptional suzuki group @xmath502 and the unitary group @xmath181 .
the observed reducible cases suggest a particular behaviour for the classes of involutions only when the group naturally appears as a group of matrices with coefficients on a field of characteristic 2
. however , not every such group and class of involutions is reducible , as shown by the data for @xmath494 .
for this problem we find some additional limiting factors .
the computation of the characteristic polynomial and the determination of the eigenvalues gets slow and as the size of the conjugacy classes increase .
the eigenvalues with their corresponding multiplicity have been computed for all the listed groups .
that computation has revealed that the killing form appears to be positive definite whenever it comes from a conjugacy class consisting of involutions , plus the mutually inverse classes 3a and 3b of elements of order 3 and centralizer of size 648 in the unitary group @xmath257 .
thus to the level visible in the data and for real conjugacy classes the classes of involutions are precisely those where the killing form is positive definite .
moreover , we see also that for real conjugacy classes those which are not classes of involutions have precisely an equal number of positive and negative eigenvalues taken with multiplicity , i.e. zero signature . for the same groups , we have also computed the decomposition into irreps of the adjoint representation on @xmath182 by means of character theory , looking for some correlation between both decompositions . as the group get larger , the observed behaviour is that the dimensions of the eigenspaces coincide with the dimensions of irreducible representations , so as the group size increases we expect that each eigenspace contains exactly one irrep .
the obvious exceptions to this rule are the conjugacy classes yielding reducible killing forms mentioned in the previous paragraph .
we have tried to make the correlation between irreps and eigenspaces more precise , and implemented an algorithm that finds the irrep decomposition of each eigenspace . however , the exact determination of the eigenspaces turns out to be too slow and memory intensive to be of any practical use except for the few smallest conjugacy classes .
we hope we will be able to get around this issue in the near future . here
we summarize some of the obtained data for all finite simple groups up to order 75,000 .
we list whether the conjugacy class is real ( closed for inverses ) , whether the corresponding killing form irreducible , and its signature .
the naming of the conjugacy classes follows the convention at the atlas , and conjugacy classes of elements order with the same centralizer sizes have been amalgamated whenever they show identical behaviour .
listing the actual eigenspace decomposition of the adjoint representation on @xmath182 would be too lengthy and not particularly enlightening , so we shall omit that data here .
whenever the killing form is reducible we have included between brackets the number of irreducible components in the corresponding column .
the signature of the killing form is expressed as the triple @xmath503 where @xmath504 , @xmath1 and @xmath505 are respectively the number ( counted with multiplicities ) of positive , negative and zero eigenvalues ; in particular , nondegeneracy is given by zero as the last number of this triple . in supplementary information
we list the maximal eigenvalue @xmath506 of the killing form , equal to the row sum . for a real conjugacy class @xmath507
is a measure of the typical size of the other entries of the killing form matrix after the principal entry @xmath508 in each row .
we also list the value @xmath509 of the character of the adjoint representation on a typical element of @xmath15 as a measure of the degree to which the braided lie algebra is nonabelian .
it counts the number of elements in @xmath15 that commute with any given element of @xmath15 .
g. heide and a. zalesski , passman s problem on adjoint representations . in : groups , rings and algebras ( proc .
conf . in honour of d.s .
passman ) , contemporary math .
420 , amer .
( 2006 ) 163 176 | we apply a recent approach to noncommutative differential geometry on finite groups to the case of finite nonabelian simple groups and similar groups with trivial centre .
the ` lie algebra ' or bicovariant differential calculus here is provided by an ad - stable generating set and by analogy with lie theory we consider when the associated killing form is nondegenerate .
we prove nondegeneracy for the case of the universal calculus whenever roth s property holds , including for all symmetric groups @xmath0 , all sporadic and most other finite simple nonabelian groups .
we conjecture that nondegeneracy holds more generally and prove it for the 2-cycles calculus on any @xmath0 , and by computer for all real conjugacy classes on finite simple nonabelian groups up to order 75,000 . in all cases we find that the killing form is in fact either positive definite , if the conjugacy class consists of involutions , or otherwise has zero ( evenly split ) signature . as an application of the killing form we find that its eigenspaces typically decompose the conjugacy class representation into irreducibles and we explore the possibility of bijectively assigning an irreducible representation to a conjugacy class containing it .
we prove that the conjugacy classes of @xmath0 containing the sign representation are those corresponding to partitions of @xmath1 into distinct odd parts and relate this observation to a classical identity of euler . |
when a narrow linewidth optical pump wave reaches a certain power threshold in a dielectric material , it can generate a counter - propagating optical stokes wave together with a strong co - propagating acoustic wave through the process of backward stimulated brillouin scattering ( bsbs ) .
the generated stokes wave is doppler red - shifted with respect to the pump wave by the acoustic resonance frequency @xmath0 .
the acoustic resonance frequency of bsbs depends mainly on material properties and on the optical wavelength and is on the order of 10 ghz for optical wavelengths of 1550 nm in glasses @xcite .
bsbs can be cascaded , in the sense that a stokes wave acts as pump waves for higher order stokes waves .
this can be exploited to achieve comb - like optical spectra that contain several spectral components spaced by the acoustic resonance frequency @xmath0 .
cascaded bsbs has been observed in several resonator configurations that include hybrid erbium - brillouin fiber ring lasers @xcite , fabry - perot ( fp ) resonators @xcite and whispering gallery mode resonators @xcite . in recent demonstrations
of cascaded bsbs in fp resonators , stable phase - locking between the optical waves has been observed @xcite , i.e. the generation of brillouin frequency combs ( bfcs ) .
early theoretical studies of cascaded bsbs in resonators did not consider the kerr - nonlinearity and the finite linewidth of the acoustic resonance @xcite .
these studies showed that , in certain configurations , optical waves can be phase - locked by phase - sensitive fwm interactions that are mediated by the acoustic field .
such interactions have been referred to as brillouin enhanced fwm ( befwm ) @xcite .
several more recent theoretical studies on cascaded bsbs @xcite did not include phase - sensitive fwm interactions or used coupled power equations and are therefore not suitable to explain the phase - locking observed in experiments .
recently , we numerically showed that fwm interactions arising from the kerr - nonlinearity ( kfwm ) between co - propagating waves can lead to phase - locking of the optical wave generated by cascaded bsbs in a 38 cm long chalcogenide fiber resonator @xcite . in this numerical study ,
the acoustic field was represented as a superposition of several acoustic waves following the approach of reference @xcite and befwm interactions were not included since the fiber under investigation was much longer than the coherence length of befwm ( @xmath1 mm @xcite ) . here
we provide for the first time , a detailed derivation of the dynamic coupled - mode equations for cascaded bsbs in fibers , which includes all phase - sensitive kfwm and befwm interactions .
such equations are necessary to correctly describe phase - locked cascaded bsbs for a wide range of resonator lengths .
the full equations are derived by describing the acoustic field with a single acoustic wave as well as with a superposition of several acoustic waves .
the validity of both approaches is demonstrated , thereby resolving questions that have recently arisen over the appropriate formalism @xcite .
we use our equations to examine the relative importance of the different fwm interactions with regard to phase - locking and numerically simulate cascaded bsbs in low - finesse fp resonators of different lengths .
we find that depending on the length scale on which the field amplitudes vary , different fwm interactions are relevant for phase - locking of the bfcs .
in this section , the coupled - mode equations for cascaded bsbs are derived for a weakly guiding , single - optical - mode dielectric waveguide consisting of an isotropic material .
the waveguide is oriented along the @xmath2-direction and has transverse dimensions that are large enough so that boundary effects can be neglected .
furthermore , we restrict our analysis to a narrow optical bandwidth and assume that only one polarization mode is excited and that the state of polarization is maintained . in optical fibers , the acoustic pressure waves participating in bsbs are well represented by the density variation @xmath3 from the mean density of the medium @xmath4 .
the wave equation for @xmath3 that governs the propagation of the acoustic wave can be written in the form @xcite @xmath5 where @xmath6 is the velocity of an acoustic pressure wave , @xmath7 is a damping parameter , @xmath8 is the vacuum permittivity and @xmath9 is the electrostrictive constant .
@xmath10 represents a langevin noise source associated with the acoustic dissipation that is necessary to include thermal excitation of acoustic waves @xcite .
the optical waves are represented in a scalar picture by the electric field @xmath11 .
the propagation of the electric field is governed by the wave equation @xcite @xmath12 where @xmath13 is the vacuum speed of light .
the parameter @xmath14 is the linear part of the relative permittivity that can be written in the form @xcite @xmath15 where @xmath16 is the refractive index , @xmath17 is the attenuation coefficient of the optical waves and @xmath18 is the angular frequency of the optical wave . here
, the parameters @xmath16 and @xmath17 are assumed to be wavelength independent due to the narrow optical bandwidth that is considered .
@xmath19 in equation [ eq : theory_el_wave ] represents the nonlinear part of the polarization density . in our analysis
we include the contribution to @xmath19 arising from the optical kerr - effect ( described by the third order susceptibility @xmath20 ) and from the variation of the material density @xmath3 resulting from the presence of acoustic waves : @xmath21 where @xmath22 is the nonlinear part of the relative permittivity .
assuming that the nonlinear polarization can be treated as a small perturbation to the total polarization and that the interaction takes place over a narrow optical bandwidth , we use the following approximation @xcite @xmath23 we describe the electric field as superposition of forward and backward propagating waves with propagation constants @xmath24 , angular frequencies @xmath25 and amplitudes @xmath26 : @xmath27 the index @xmath28 denotes the pump wave , indexes @xmath29 stokes waves and indexes @xmath30 anti - stokes waves
. the carrier frequencies of the pump wave and the order first stokes wave are defined such that their frequency spacing @xmath31 corresponds exactly to the acoustic resonance frequency @xmath32 of a pressure wave that travels with velocity @xmath33 along the @xmath2-axis @xmath34 where @xmath35 is the corresponding propagation constant of the acoustic wave .
the angular frequencies of all higher order stokes and anti - stokes waves are defined as @xmath36 i.e. exactly equally spaced by the acoustic frequency @xmath37 for the interaction between the pump and the first order stokes wave ( see fig .
[ fig : concept ] ) . .
] the parameter @xmath38 in equation [ eq : electric_theory_paper ] scales the optical amplitudes such that @xmath39 corresponds to the optical power carried by each mode .
@xmath40 represents the transverse mode profile for guided optical modes with frequency close to @xmath18 and fulfills the modal equation @xmath41f(x , y ) = 0.\end{aligned}\ ] ] note that in we neglect the frequency dependence of the field profile @xmath40 because of the small frequency range involved in brillouin interactions .
the angular brackets @xmath42 denote averaging over the transverse plane , using the integral @xmath43 . in this section , we derive the coupled - mode equations for cascaded bsbs by expressing the density variation @xmath3 as superposition of only one forward and one backward propagating acoustic wave with propagation constants @xmath44 and angular frequency @xmath32 : @xmath45e^{-i\omega_b t } + c.c.,\end{aligned}\ ] ] where @xmath46 and @xmath47 are the amplitudes of forward and backward propagating acoustic waves , respectively . in general many acoustic modes exist .
however , here we only consider the interaction with the fundamental longitudinal acoustic mode that dominates in the sbs interaction and has mode profile @xmath48 .
@xmath48 fulfills the modal equation @xmath49\xi ( x , y ) = 0.\end{aligned}\ ] ] in order to derive the coupled - mode equations for the optical waves , we start by inserting equations [ eq : electric_theory_paper ] and [ eq : acoustic_theory_paper ] into equation [ eq : theory_el_wave ] and substituting equations [ eq : epsilon_l ] and [ eq : p_nl ] .
we then use the approximation shown in equation [ eq : p_nl_approx ] and apply the slowly varying envelope approximation by omitting second derivatives of the amplitudes @xcite .
we separate the parts with exponential factors @xmath50 and @xmath51 into the following equations @xmath52 with @xmath53 to derive equation [ eq : theory_cme_optical_1acoustic ] , we substituted the modal equation equation [ eq : bg_modal_optical ] . in order to separate the transverse dependency , both sides of the equation were multiplied by @xmath40 , integrated over the transverse plane and divided by @xmath54 .
furthermore , we used the approximations @xmath55 and @xmath56 in prefactors since we only consider a small frequency range and a weakly guiding waveguide . the source term on the right hand side of equation [ eq : theory_cme_optical_1acoustic ] , containing the parameter @xmath57 , describes the coupling of the optical waves @xmath58 to the counter propagating waves @xmath59 ( stokes processes ) and @xmath60 ( anti - stokes processes ) via the acoustic waves @xmath61 and @xmath62 , respectively . here
, the @xmath2 and @xmath63 dependent phase rotations result from the propagation constant mismatches @xmath64 where @xmath65 represents the propagation constant difference between two adjacent spectral components @xmath66 in equation [ eq : theory_cme_optical_1acoustic ] , the term containing @xmath67 represents the interactions via the kerr - nonlinearity .
the sum includes all combinations of the indexes @xmath68 and @xmath69 and the kronecker delta @xmath70 ensures conservation of energy .
the first term of the sum includes kfwm of co - propagating waves , which has no propagation constant mismatch ( under zero dispersion assumption ) .
the second term of the sum contains kfwm of counter - propagating waves , which has a propagation constant mismatch of @xmath71 .
we proceed similarly for deriving the equations for the amplitudes of the acoustic waves and insert equations [ eq : electric_theory_paper ] and [ eq : acoustic_theory_paper ] into equation [ eq : theory_acoustic6 ] .
we then use the slowly varying envelope approximation and separate the terms with the exponential factors @xmath72 and @xmath73 to obtain @xmath74 with @xmath75 in equation [ eq : theory_cme_acoustic_1acoustic ] , we have substituted the decay constant of the acoustic wave @xmath76 , which is related to the phonon lifetime @xmath77 with @xmath78 . in deriving equation [ eq : theory_cme_acoustic_1acoustic ]
, we used the approximation @xmath79 for prefactors .
several terms that contain first derivatives of the amplitudes @xmath80 have been neglected based on the approximation @xmath81 , which represents the ratio between the brillouin gain bandwidth and the brillouin frequency shift @xcite .
we also substituted the modal equation for the acoustic wave ( equation [ eq : bg_modal_acoustic ] ) and separated the transverse dependency by multiplying both sides of the equation with @xmath48 , integrating over the transverse plane and dividing by @xmath82 .
the first term on the right hand side of equation [ eq : theory_cme_acoustic_1acoustic ] shows that the acoustic waves @xmath61 are driven by all pairs of optical waves @xmath83 and @xmath59 .
this term also contains the propagation constant mismatch @xmath84 since the optical waves drive the acoustic field with different propagation constants @xmath85 .
the noise terms @xmath86 equation [ eq : theory_cme_acoustic_1acoustic ] are related to the original noise term @xmath10 by @xmath87 the noise sources @xmath88 and @xmath89 are uncorrelated gaussian random variables with zero mean @xmath90 and are @xmath91 correlated in the sense @xmath92 where the parameter @xmath93 describes the strength of the thermal fluctuations .
an expression for @xmath93 is given by equation 18 in reference @xcite which is based on the equipartition theorem . note that in equations [ eq : theory_cme_optical_1acoustic ] and [ eq : theory_cme_acoustic_1acoustic ] , the phase of the terms containing the propagation constant mismatch @xmath84 rotates by @xmath94 over the `` brillouin - length '' @xcite @xmath95 which is about 1 cm for the parameters of standard silica fiber at 1550 nm @xcite .
equations [ eq : theory_cme_optical_1acoustic ] and [ eq : theory_cme_acoustic_1acoustic ] represent the coupled - mode equations for cascaded bsbs derived using only one forward and one backward propagating acoustic wave .
it is common for modeling cascaded backward bsbs to describe the acoustic field as superposition of several acoustic waves @xcite .
this approach can be insightful and useful for numerical modeling of waveguides with lengths @xmath96 . to derive the corresponding coupled - mode equations
, we express the acoustic field @xmath97 as the superposition of several acoustic waves with different propagation constants ( corresponding to the propagation constants of the driving fields ) @xmath98 but with the same temporal frequency @xmath99 ( see fig . [
fig : theory_dispersion ] ) .
the density variation then reads @xmath100 } + c.c.\end{aligned}\ ] ] .
] to derive the coupled - mode equations , we insert equations [ eq : electric_theory_paper ] and [ eq : bfc_fiber_multiple_acoustic_def ] into equations [ eq : theory_el_wave ] and [ eq : theory_acoustic6 ] and proceed as in the previous section to derive @xmath101}e^{-il \delta k } = ig_1 \sum\limits_l { { a_l^ \pm a{{_{l + 1}^ \mp } ^ * } } e^{-il \delta k}+f^{\pm}}. \end{split}\end{aligned}\ ] ] in equation [ eq : bfc_fiber_several_acoustic ] , only forward and backward propagating acoustic waves were separated .
the detuning term containing @xmath102 now appears in the equation because the acoustic waves with propagation constants , as defined in equation [ eq : theory_propagation_constant_acoustic_waves ] , do not obey the acoustic dispersion relation @xmath103 , except for @xmath104 . in order to obtain separate equations ,
we define @xmath105 to obey the following equation @xmath106 where @xmath107 and @xmath108 is the number of acoustic waves . by comparing equations [ eq : bfc_fiber_several_acoustic ] and [ eq : bfc_fiber_ql ]
, it can be seen that when @xmath109 are defined as in equation [ eq : bfc_fiber_ql ] , equation [ eq : bfc_fiber_several_acoustic ] is also fulfilled .
however , it should be noted that the amplitudes @xmath109 , in general , are not uniquely defined by equation [ eq : bfc_fiber_several_acoustic ] .
the different components of the acoustic field @xmath110 are not always orthogonal functions because they have the same frequency @xmath37 and the spatial spectra of @xmath111 can also overlap if the amplitudes @xmath112 vary on a length scale @xmath113 . with equation [ eq : bfc_fiber_ql ] ,
we have defined the waves @xmath105 as the parts of the acoustic field that are driven by the pairs of adjacent spectral components of the optical field @xmath114 and @xmath115 . .
] the sum over @xmath116 , containing the acoustic waves @xmath117 in equation [ eq : bfc_fiber_aj_with_qls ] , represents the coupling of the optical waves @xmath83 to the counter - propagating waves @xmath59 and @xmath60 , which is mediated by all acoustic waves @xmath117 .
this coupling is only phase - matched for @xmath118 , which results from the phase - rotation terms containing @xmath119 .
for @xmath118 , there is no propagation constant mismatch as these terms then describe the coupling of two optical waves , e.g. @xmath83 and @xmath59 , by the acoustic wave @xmath120 that is also driven by the same pair of optical waves ( fig .
[ fig : theory_fwm_illustration ] ( a ) ) , which is the usual three - wave interaction of bsbs .
the other terms of the sum with @xmath121 , on the other hand , represent terms that we refer to as befwm interactions @xcite .
one example of such an interaction is illustrated in fig .
[ fig : theory_fwm_illustration ] ( b ) .
the waves @xmath83 and @xmath59 are coupled by the acoustic wave @xmath117 with wavevector @xmath122 , which is driven by the waves @xmath123 and @xmath124 .
at the same time , @xmath123 and @xmath124 are coupled by the acoustic field @xmath120 with wavevector @xmath125 , driven by the waves @xmath83 and @xmath59 .
this interaction can be efficient , e.g. when the acoustic waves @xmath111 have significant strength at frequency @xmath126 in momentum - space , which is possible when @xmath109 vary on a length scale @xmath127 .
on the other hand , befwm interactions have no significant contribution if the spatial spectra of the acoustic waves @xmath111 and @xmath128 are well separated in momentum - space , which is the case when the amplitudes vary on a lengthscale @xmath129 . in this case , these terms can be omitted as previously been done @xcite ( note that in this case the correct inclusion of the noise terms @xmath130 has to be reconsidered ) . for comparison , figs .
[ fig : theory_fwm_illustration ] ( c ) and ( d ) show the kfwm interactions of co - propagating waves and counter - propagating waves , respectively , as described in the discussion of equation [ eq : theory_cme_optical_1acoustic ] .
the peak brillouin gain coefficient @xmath131 ( units : m@xmath132w@xmath132 ) is often much larger than the nonlinear coefficient @xmath67 , e.g. in optical fibers . the phase - matched bsbs three - wave interaction ( fig .
[ fig : theory_fwm_illustration ] ( a ) ) is then expected to be the dominant gain process as it is phase - matched and has the larger gain constant .
this interaction , however , can not lead to phase - locking as it only couples pairs of optical waves and is independent of the spectral phase .
the fwm interactions , on the other hand , couple more than two optical waves and are sensitive to the spectral phase . in the next section
, we will numerically investigate the relative importance of the different fwm interactions with respect to phase - locking .
in this section , the dynamic coupled - mode equations , derived in the previous section , are used to study cascaded bsbs in low - finesse , fp fiber resonators of two different lengths .
we qualitatively investigate the importance of the different fwm interactions for phase - locking the optical waves and its dependence on the resonator length . for simplicity , we only include forward and backward propagating pump , first and second order stokes waves and the corresponding acoustic waves that couple the optical waves ( @xmath133 ) .
the dynamic coupled - mode equations for cascaded bsbs were solved using the method of characteristics @xcite adapted to include additional acoustic waves @xcite .
we used the coupled - mode equations [ eq : bfc_fiber_aj_with_qls ] and [ eq : bfc_fiber_ql ] , omitting the propagation of the acoustic wave ( the term @xmath134 ) , which is a common approximation because the acoustic waves are highly damped @xcite . for a complete description of the system , the boundary conditions have to be defined . in the case of a fp resonator
, the boundary conditions can be written as @xcite @xmath135 the parameter @xmath136 is the pump power that is coupled into the resonator at @xmath137 .
@xmath138 are the linear phase shifts of the waves due to transit through the cavity , which are given by @xmath139 the parameter @xmath140 can be described as @xmath141 here , we assume that the parameter @xmath140 is an integer , so that all waves effectively experience the same linear phase shift @xmath142 after one roundtrip , where `` mod '' ( modulo ) denotes the remainder of @xmath143 after division by @xmath144 . additionally , it is assumed that initially at @xmath145 all fields are zero inside the resonator . to identify phase - locking between the waves , we consider the phases @xmath146 of the amplitudes @xmath147 where @xmath148 are real and represent the powers of the waves .
the frequencies of the stokes waves in the steady state do not necessarily coincide with the frequencies defined by equation @xmath149 due to `` stokes pulling '' ( the stokes frequencies lie in between the cavity resonances and the brillouin gain peak @xcite ) .
however , when the system reaches a phase - locked steady state , the individual phases can be expressed in the form @xmath150 , where @xmath151 are frequency detunings that are @xmath2 and @xmath152 independent @xcite .
this means that the actual steady state stokes and anti - stokes frequencies are given by @xmath153 . when the three waves reach a steady state with equally spaced frequencies , the phase - dispersion @xmath154 is time independent , i.e. @xmath155 . for simulations for the longer resonator
, we used the parameters of the as@xmath156se@xmath157 chalcogenide fiber resonator that was investigated in reference @xcite with length @xmath158 cm .
the parameters of this resonator are given by : @xmath159 % , @xmath160 db / m , @xmath161 , @xmath162 m@xmath132w@xmath132 , @xmath163 m@xmath132w@xmath132 , @xmath164 ghz , @xmath165 mhz and @xmath166 m / s , which results in @xmath167 . for the shorter resonator we chose @xmath168 cm . in order to be able to compare the simulation results for the long and the short resonators
more easily , we assumed that for the short resonator the interaction area is 10 times smaller compared to the longer resonator , such that @xmath169 and @xmath67 are increased by a factor of 10 .
we also assumed that the attenuation coefficient @xmath170 is 10 times larger in the shorter resonator .
we calculated the strength parameter @xmath93 of the noise terms using the expression given in equation 18 in reference @xcite using the following parameters : waveguide temperature @xmath171 k , density of as@xmath156se@xmath157 @xcite @xmath172 kg / m@xmath173 and interaction area @xcite @xmath174 @xmath175m@xmath176 .
however , it should be noted that the simulation results discussed in this section are relatively insensitive to the exact noise strength . due to the presence of optical feedback ,
the powers of the stokes waves are rather determined by an interplay of gain and loss . for our qualitative investigation of the importance of the different fwm interactions with regards to achieving phase - locking
, we performed a two dimensional parameter scan of the parameters @xmath177 and @xmath178 , and solved the coupled - mode equations for each pair of parameters .
we scanned the coupled input power @xmath178 from 0.45 w to @xmath179 w in steps of 0.025 w and the phase shift @xmath177 from 0 to @xmath180 ( which corresponds to tuning the pump frequency @xmath18 by one fsr from one cavity resonance to the next one ) in steps of @xmath181 .
the coupled - mode equations were integrated over a time interval of @xmath182s , which is much longer than the phonon and the photon lifetimes in the cavity that determine the time scales of the transient dynamics . and the phase - dispersion @xmath183 at the end of the fiber , for the input power @xmath184 w , and two different values of @xmath185 : ( a ) @xmath186 ( b ) @xmath187 .
( c ) and ( d ) show the computed waveforms corresponding to ( a ) and ( b ) , respectively . ]
figures [ fig : theory_temporal ] ( a ) and ( b ) show two typical evolutions of the powers @xmath188 ( @xmath189 ) and the phase - dispersion @xmath190 of the forward propagating waves at the boundary @xmath191 obtained from the simulation .
figures [ fig : theory_temporal ] ( c ) and ( d ) on the right shows the corresponding time - domain output waveform calculated with @xmath192 the insets in the plots on the right show 1 ns - long zoomed - in sections at different times .
both simulations were performed for the longer resonator for @xmath184 w. for the parameter @xmath185 , we chose the values @xmath193 and @xmath194 .
figures [ fig : theory_temporal ] ( a ) and ( c ) illustrate the case where the system does not reach a phase - locked steady state . here , the phase - dispersion @xmath190 drifts with time .
this also results in an oscillation of the powers @xmath188 since the powers of the waves are coupled to the phases via fwm .
the obtained output waveform in fig .
[ fig : theory_temporal ] ( c ) also changes with time .
figures [ fig : theory_temporal ] ( b ) and ( d ) show the case where the system attained a phase - locked steady state . after an initial power transfer from pump to the first and second order stokes waves , the powers @xmath188 and the phase - dispersion @xmath190 reached a constant value , which resulted in a stable waveform shown on the right .
the results of the entire parameter scans for the long and the short cavity are shown in figs .
[ fig : theory_stability ] ( a ) and ( b ) .
the black area corresponds to parameters for which the dynamic system attained a phase - locked steady state with at least 1 % of the coupled power in the second stokes wave @xmath195 .
the simulation results for these parameters are similar to the ones shown in figs .
[ fig : theory_temporal ] ( b ) and ( d ) .
the white area corresponds to parameters for which the phase - dispersion @xmath190 did not reach a stable value similar to the case shown in figs .
[ fig : theory_temporal ] ( a ) and ( c ) .
the white area also includes parameters for which the second stokes wave was not efficiently excited .
the phase - dispersion @xmath190 was judged to have reached a stable value when its standard deviation fulfilled @xmath196 over the final @xmath197s of the simulation time interval . and @xmath198 to identify domains of phase - locking ( black ) .
parameters that lead to phase - locking were determined by integrating the dynamic coupled - mode equations for cascaded bsbs for a sufficiently long time interval and determining whether @xmath190 reached a stable value when the second stokes wave got sufficiently excited .
the left and right column show results for the long and the short cavity , respectively . in order to identify the importance of the different fwm interactions with regards to phase - locking
, the parameter scan was performed including all fwm interactions ( a ) , ( b ) ; only befwm ( c ) , ( d ) ; only kfwm of co - propagating waves ( e ) , ( f ) ; only kfwm of counter - propagating waves ( g ) , ( h ) .
phase - insensitive bsbs interactions ( illustrated in fig .
[ fig : theory_fwm_illustration ] ( a ) ) , and phase - insensitive terms arising from the kerr - nonlinearity ( self - phase and cross - phase modulation ) were kept in the equations for all simulations . ] to investigate the importance of the different fwm interactions with regard to phase - locking , we performed the same parameter scan for both fiber lengths three additional times . here , we only included fwm terms corresponding to befwm ( figs .
[ fig : theory_stability ] ( c ) , ( d ) ) , only kfwm of co - propagating waves ( figs .
[ fig : theory_stability ] ( e ) , ( f ) ) and only kfwm of counter - propagating waves ( figs . [
fig : theory_stability ] ( g ) , ( h ) ) , respectively .
the terms representing the phase - insensitive bsbs interactions ( illustrated in fig .
[ fig : theory_fwm_illustration ] ( a ) ) , and phase - insensitive terms arising from the kerr - nonlinearity ( self - phase and cross - phase modulation ) were kept in the equations for all simulations .
the analysis shows that in the long resonator ( fig .
[ fig : theory_stability ] , left column ) , mainly kfwm of co - propagating waves leads to phase - locking whereas in the case of the short resonator ( fig .
[ fig : theory_stability ] , right column ) , kfwm of co - propagating waves and befwm can both efficiently phase - lock the waves for a large parameter space .
kfwm of counter - propagating waves does not seem relevant for both resonator lengths in the presence of the other fwm interactions .
for the long cavity , it did not lead to phase - locking for any parameters . in the short cavity ,
phase - locking was observed only for a very small parameter space that is included in the parameter space for phase - locking by befwm ( fig .
[ fig : theory_stability ] ( d ) ) .
the results can be understood considering that the coherence length of befwm and kerr - fwm of counter - propagating waves is @xmath199 .
the long resonator has a length @xmath200 .
therefore , the contributions of the non - phase - matched interactions almost completely cancel along the resonator length . in the case of the shorter resonator , which is only a few times longer that the coherence length @xmath199
, the non - phase - matched befwm interactions can have a strong contribution ( fig .
[ fig : theory_stability ] ( d ) ) .
this is possible , despite the resonator length being longer than @xmath199 , because the optical and acoustic fields are not uniformly distributed along the resonator length due to its low finesse .
the amplitudes can vary on a length scale of @xmath199 since the gain was increased in the case of the shorter resonator .
note that the main reason for the difference of the results shown in figs .
[ fig : theory_stability ] ( e ) and ( f ) are the different ratios between phonon and photon lifetime for the long and the short resonators . in this case
, no interactions were included that are not phase - matched and the waves experience a comparable gain per transit .
however , in the short resonator , the photon lifetime is shorter , which broadens the linewidth of the cavity resonances and leads to a different ratio between cavity linewidth and brillouin gain linewidth .
in this paper , we derived the dynamic coupled - mode equations for cascaded bsbs in fibers including all kfwm and befwm interactions . coupled - mode equations were obtained by expressing the acoustic field with a single acoustic wave and as a superposition of several acoustic waves with different propagation constants .
we used the derived equations to numerically simulate cascaded bsbs in a low - finesse , chalcogenide , fp fiber resonators of two different lengths and qualitatively investigated the relative importance of the different fwm terms with regards to phase - locking of the optical waves . in the longer resonator ,
corresponding to a previous experiment based on a 38 cm long chalcogenide fiber resonator @xcite , kfwm of co - propagating waves is most relevant for phase - locking .
in contrast , in the shorter resonator of 4 cm length , kfwm and the phase - sensitive coupling of several optical waves via befwm contributed to phase - locking . in the model described in this paper , dispersion has been neglected since we assumed that the interaction takes place over a narrow optical bandwidth as in the experiments described in refs .
@xcite . for numerical modeling of broad bfcs that are generated by the interplay of bsbs and the kerr - nonlinearity , as demonstrated in reference @xcite
, dispersion would have to be considered .
equations [ eq : bfc_fiber_aj_with_qls ] and [ eq : bfc_fiber_ql ] were derived by expressing the electrical and the acoustic field as a superposition of several co - propagating waves with different propagation constants .
this approach is advantageous for numerical modeling of waveguides that are long enough that kfwm of counter - propagating waves and befwm can be neglected , e.g. long silica fiber resonators .
this can be understood considering that simulating the interaction of counter - propagating optical waves requires performing an integration at each spatial grid point of the resonator for each time step . describing the acoustic field as a single wave requires the spatial grid to be fine enough to resolve the modulations of the acoustic wave on a length scale below @xmath199 , in order to include bsbs of higher order stokes components ( see equation [ eq : theory_cme_acoustic_1acoustic ] ) . on the other hand , when the acoustic wave is split in components with different propagation constants , as in equation [ eq : bfc_fiber_ql ] , the amplitudes of the acoustic waves @xmath105 vary on the same length scale as the amplitudes of the optical waves , which can be tens of meters , e.g. in the case of a long silica fiber .
this allows a significantly coarser spatial grid for the numerical simulation , which can decrease the required computation time by several orders of magnitude . in principle , cascaded bsbs can also be described by representing both the electrical and acoustic fields with one forward and one backward propagating wave .
the generation of new frequencies then manifests as modulations of the amplitudes of the waves .
this could be advantageous for modeling short resonators as it significantly reduces the number of coupling terms since this number is then independent of the number of generated stokes waves .
a spatial discretisation that is fine enough to resolve the amplitude modulations , representing higher order stokes waves , would be necessary .
for bsbs on very short length scales , the propagation of the acoustic wave may also have to be considered if the amplitudes of the acoustic waves vary on lengths scales that are on the order of the decay length of the acoustic wave ( @xmath201 m ) .
australian research council ( arc ) ( ce110001018 , de130101033 , fl120100029 ) .
the authors thank christian wolff for helpful discussions and bjrn sturmberg for help with parallelizing the numerical routine . | cascaded stimulated brillouin scattering ( sbs ) is a complex nonlinear optical process that results in the generation of several optical waves that are frequency shifted by an acoustic resonance frequency .
four - wave mixing ( fwm ) between these brillouin shifted optical waves can create an equally spaced optical frequency comb with a stable spectral phase , i.e. a brillouin frequency comb ( bfc ) . here , we investigate phase - locking of the spectral components of bfcs , considering fwm interactions arising from the kerr - nonlinearity as well as from coupling by the acoustic field .
deriving for the first time the coupled - mode equations that include all relevant nonlinear interactions , we examine the contribution of the various nonlinear processes to phase - locking , and show that different regimes can be obtained that depend on the length scale on which the field amplitudes vary .
10 url # 1`#1`urlprefix[2]#2 [ 2][]#2 _ _ ( , ) , edn .
_ _ ( , ) , edn . , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . .
_ _ * * , ( ) . & * * , ( ) . , , &
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . .
_ _ * * , ( ) . .
_ _ * * , ( ) . & .
_ _ ( ) . , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . .
_ _ * * , ( ) . |
quantum measurement is one of the simplest , yet most important , physical processes @xcite . without measurement , we see no event and obtain no information .
a quantum measurement is a process that maps the quantum state of a quantum system to the classical state ( probability distribution ) of an external classical system that belongs to the observer side .
it is known that the interface ( border ) between quantum and classical systems can be shifted .
in fact , an observer does not interact with the system itself .
instead , they extract information from another system , a measurement apparatus , that has direct contact with the system . both the system and the measurement apparatus can be treated quantum mechanically .
the main purpose of this paper is to investigate the energy and interaction required for measuring an observable .
more precisely , we investigate the energy of a quantum apparatus and the strength of interaction between a system and the apparatus so that the process is fully described by quantum theory . to put the question from a more pragmatic viewpoint ,
our interest is in the amount of resource " required to perform a measurement ( or information transfer ) task @xcite .
thus , to study this problem , the interface between quantum and classical systems must be located such that the apparatus is treated quantum mechanically .
for instance , although we have an equivalent minimum description ( called an instrument @xcite ) of the dynamics and measurement results that puts the interface between the system and the apparatus , our problem does not allow us to employ it because our interest is in the limitations of the quantum apparatus . in section [ section2 ] , we discuss how large " the quantum side ( or quantum apparatus ) must be to describe a measurement process in a fully quantum manner .
we examine a so - called standard measurement model to find that the model requires an external system switching on the measurement interaction between a system and an apparatus . in this sense ,
the measurement process is not closed - it is not described in a fully quantum manner in this model .
this model does not obey any non - trivial energy - time uncertainty relation .
this conclusion agrees with previous results @xcite .
then , in section [ section3 ] , we rigorously formulate a quantum apparatus as a timing device that autonomously switches on a perturbation .
an argument on the analyticity shows that the total hamiltonian must be two - side unbounded to fulfill the conditions for the timing device .
our main results are presented in section [ section4 ] .
we consider a measurement apparatus that autonomously switches on the interaction at a certain time .
we show that there is an energy - time uncertainty type relation between the fluctuation of the apparatus hamiltonian and time duration of the measurement .
the proof of this trade - off relation is given by combining two kinds of uncertainty relations .
we first observe that a perfect measurement of a given observable implies a perfect distortion of the conjugate states .
this property is called an information - disturbance relation and is related to an uncertainty relation for a joint measurement .
we then employ the mandelstam - tamm uncertainty relation to obtain a restriction on time and energy required for the distortion .
in addition , because the measurement process involves an information transfer from the system to the apparatus , the interaction between them should not be too weak .
in fact , in section [ section5 ] we show a trade - off relation between the strength of interaction and the measurement time .
the robertson uncertainty relation plays an essential role in the proof . in section [ section6 ]
we illustrate an argument of the spacetime uncertainty relation as an application of our result .
let us consider a quantum system described by a hilbert space @xmath3 .
a quantum state is represented as a density operator on @xmath3 .
we first present a description when we put an interface between the quantum and external classical systems just outside this system . by measuring an observable
, we obtain an outcome . in general
, the outcome is not definite and only a probability distribution on the outcome set is determined .
thus the measurement of an observable maps the quantum state of a quantum system to the classical state ( probability distribution ) of an external classical system which belongs to the observer side . for a map to be consistent with an interpretation based on probability , it must be an affine map .
each affine map is completely described by a positive - operator - valued measure ( povm ) in the quantum system ( see e.g. , @xcite ) .
a povm ( with a discrete outcome set @xmath4 ) is defined as a family of positive operators @xmath5 satisfying @xmath6 .
a povm @xmath7 gives a probability distribution @xmath8 $ ] on an outcome set @xmath9 .
a povm @xmath10 is called a projection - valued measure ( pvm ) , or a sharp observable , if each @xmath11 is a projection operator .
next we shift the interface between the quantum and classical systems .
the measurement process is a physical interaction between a system and a measurement apparatus .
we treat the apparatus as a quantum system described by a hilbert space @xmath12 . by taking @xmath12 to be sufficiently large
, we can expect the total system to be ( approximately ) closed .
thus the total dynamics is specified by a hamiltonian @xmath13 .
the total hamiltonian can be decomposed into three parts @xcite : @xmath14 where @xmath15 ( resp .
@xmath16 ) acts only on @xmath3 ( resp .
@xmath12 ) and @xmath17 describes an interaction hamiltonian acting on @xmath18 .
( more precisely , the total hamiltonian should be written as @xmath19 . )
the initial state of the composite system at time @xmath20 is described by @xmath21 with an unknown state @xmath22 of the system and a fixed initial state @xmath23 of the apparatus .
this evolves according to the von neumann equation up to a certain predetermined time @xmath24 , where the time duration @xmath2 is called a measurement time duration .
then an observer outside these quantum systems observes a meter observable @xmath25 , which is a povm on the apparatus .
this whole process is said to describe a measurement process of an observable @xmath26 if @xmath27 = \mbox{tr}[({\mathbf{1}}\otimes { \mathsf{e}}_x)e^{-i\frac{h \tau}{\hbar } } ( \rho \otimes \sigma ) e^{- \frac{h\tau}{\hbar}}]$ ] holds for every @xmath22 .
the above physical measurement model is consistent with the abstract measurement theory by ozawa s extension theorem @xcite and has been shown to be useful in analyzing real measurement processes .
this model , however , is not sufficient for obtaining general results on the minimum energy fluctuation required for a measurement . in order to illustrate this point , we consider the following example , a so - called standard measurement model developed essentially by von neumann @xcite .
suppose that a system is a qubit ( thus @xmath28 ) and an apparatus is a one - particle system on a real line ( see fig .
[ figure_switch11 ] ) .
.,width=340 ] set @xmath29 , @xmath30 and @xmath31 , where @xmath32 represents the momentum operator . as the meter observable
we employ @xmath33 . for an arbitrary time duration @xmath2 , if the initial state at time @xmath20 of the apparatus is prepared in a narrowly - localized state with respect to a position , the accuracy of the measurement of @xmath34 can be made arbitrarily high
. therefore , there is no limitation on the time and energy required for this measurement process . by looking carefully at this model ,
one may notice that the time duration is given by hand . in this model , the interaction between a system and an apparatus is switched on at @xmath20 and switched off at @xmath24 .
the dynamics outside this time interval is not discussed .
an external observer must put these systems together at @xmath20 , which until then must have been somehow independent .
thus the mechanism initiating this switching - on process is supposed to be outside the quantum model . in this sense , this model is not closed and we need to shift the interface further to include the on - and - off switching process . in the discussion below , we develop a general measurement process that also treats this switching - on ( and off ) mechanism quantum mechanically . in some situations , roughly speaking , we shift the quantum - classical interface so that the quantum side includes the experimenter who controls the apparatus . in the next section
we give a mathematically rigorous formulation of the timing device that switches on a perturbation .
as stated above , we investigate an apparatus that works not only as an information extractor but also as a switching device for an interaction ( perturbation ) .
to do so , in this section , we study a formulation of the latter condition and derive some results .
we consider an apparatus that works as a timing device to switch on an interaction at a certain time after @xmath20 .
thus , we assume that at time @xmath20 the state of the total system is a product state .
the apparatus is specified by the hilbert space @xmath12 , hamiltonian @xmath16 and a state @xmath35 at time @xmath20 .
let us first assume that the apparatus is perfectly isolated and has no interaction with the system
. then the time evolution is described by a von neumann equation determined by an apparatus hamiltonian @xmath16 which acts only on @xmath12 . for an arbitrary time @xmath36
, the state at time @xmath37 is written as @xmath38 .
next , we consider an interaction between the system and apparatus .
we denote the hilbert space of the system by @xmath3 .
the individual dynamics of the system is governed by the hamiltonian @xmath15 defined on @xmath3 .
thus the total hamiltonian has three parts , @xmath39 , where @xmath17 is the interaction hamiltonian that acts on the tensor product @xmath18 .
the time evolution of a state @xmath40 ( a density operator ) over the composite system is described by the corresponding von neumann equation , @xmath41 .
\end{aligned}\ ] ] let us consider conditions for @xmath16 , @xmath17 and a state of the apparatus to describe a timing device .
they must satisfy the following general conditions that represent the capability to switch on the interaction at a certain time .
the apparatus evolves freely up to time @xmath20 and then switches on an interaction with the system at some time after @xmath20 .
thus , the state of the apparatus should be decoupled from that of the system until time @xmath20 .
the time evolution of the state is determined entirely once the state at a certain time is given .
we impose the following condition on the dynamics and the state at time @xmath20 : [ cond1 ] ( no interaction up to time @xmath20 ) an apparatus ( specified by @xmath16 , @xmath17 and a state @xmath35 at time @xmath20 ) satisfies _ the no interaction condition up to time @xmath20 _ if for any @xmath42 and for any state @xmath22 of the system , @xmath43 = 0 \label{cond1eq}\end{aligned}\ ] ] holds , where @xmath44 .
we emphasize that in condition [ cond1 ] , the left - hand side of ( [ cond1eq ] ) must be vanishing for any @xmath22 .
this condition implies that only after the apparatus reaches @xmath35 , does @xmath17 induce the apparatus and the system to interact .
in fact the following lemma shows that these conditions are equivalent .
if part " : we differentiate by @xmath37 , @xmath48 to obtain @xmath49 = [ h_s + h_a , \rho(t ) \otimes \sigma(t)].\end{aligned}\ ] ] as the left - hand side is written as @xmath50 $ ] , we obtain @xmath51=0 $ ] . as @xmath22 is arbitrary , we obtain condition [ cond1 ] .
+ only if " part : it holds that for @xmath42 , @xmath52 \\ & = & [ h , \rho(t)\otimes \sigma(t ) ] , \end{aligned}\ ] ] where we used ( [ cond1eq ] ) for @xmath46 .
thus we conclude that @xmath53 satisfies the von neumann equation whose integration is ( [ cond1int ] ) .
[ lemma1 ] if condition [ cond1 ] is satisfied and at a certain time @xmath56 the state has a product form @xmath57 ; then the state @xmath40 at any time @xmath42 becomes the product state ; @xmath58 where @xmath59 .
note that this condition is rather weak .
it still allows the existence of @xmath22 such that @xmath17 acts trivially on @xmath53 for any @xmath63 .
in addition , this condition does not specify exactly the switching time of the perturbation as @xmath20 .
it only requires the existence of the moment @xmath64 at which the perturbation does not vanish .
[ cond3](switching - off time condition ) an apparatus satisfies _ the switching - off time condition _ if there exists a time @xmath65 called a switching off time such that for any @xmath66 and for any state @xmath67 of the system , @xmath68 = 0 \label{cond3eq}\end{aligned}\ ] ] holds .
[ ex1 ] suppose that an apparatus is described by @xmath69 and @xmath70 .
it is coupled with a system described by @xmath71 with orthonormal basis @xmath72 and a trivial hamiltonian @xmath73 .
the interaction term is introduced as @xmath74 , where @xmath75 and @xmath76 is a nonvanishing real function whose support is included in @xmath77 for some @xmath78 . in the position representation @xmath79 is supposed to be strictly localized in the negative real line .
that is , the support of @xmath80 is included in @xmath81 for some @xmath82 .
the freely evolved state of the apparatus can be written as @xmath83 , which has support in @xmath84 .
it is easy to see that this system - plus - apparatus satisfies condition [ cond1 ] .
let us consider the time evolution of the state denoted by @xmath85 . for @xmath86
, while the state of the system remains @xmath87 , the state of the apparatus is affected by the interaction .
we denote the whole state at time @xmath37 by @xmath88 .
it can be shown that at time @xmath37 , the state of the apparatus @xmath89 becomes @xmath90 in the position representation . on the other hand , for @xmath91 , while the state of the system remains @xmath92 , the state @xmath93 of the apparatus at time @xmath37 becomes @xmath94 in the position representation .
thus , for a general initial state @xmath95 , the state of the total system evolves as @xmath96 which coincides with an unperturbed ( freely evolved ) state up to time @xmath97 .
thus it satisfies condition [ cond1 ] .
the state of the system evolves as @xmath98 for @xmath99 .
thus condition [ cond2 ] is also satisfied .
in addition , we put @xmath100 .
taking into account the support of @xmath101 , we obtain for @xmath102 , @xmath103 as their supports do not intersect with the support of @xmath76 , condition [ cond3 ] is satisfied .
suppose that @xmath13 is lower bounded .
( an upper bounded case can be treated similarly . ) by purifying the state @xmath35 of the apparatus , we can assume the state @xmath35 to be pure .
we denote the purified state by @xmath106 . on this enlarged system
, @xmath13 also works as a hamiltonian and is lower bounded .
( more precisely we need @xmath107 , where @xmath108 acts only on an auxiliary hilbert space employed for purification . )
we denote by @xmath12 again the enlarged hilbert space of the apparatus . for an arbitrary state @xmath109 , we have for @xmath42 , @xmath110=0 .
\end{aligned}\ ] ] this implies that for @xmath42 there exists a function @xmath111 such that @xmath112 as @xmath113 is normalized we have @xmath114 we now define latexmath:[\ ] ] thus we can conclude @xmath478 we assume @xmath479 .
combining it with theorem [ ettheorem ] by putting @xmath237 , we obtain @xmath480 this ends the proof .
while a measurement process consists of the information transfer " described by the time evolution of a composite system and the state reduction conditional to each measurement outcome , we treat only the former .
see @xcite for the relevant discussions . for a given total hamiltonian @xmath13
there is an ambiguity in dividing it into three parts as @xmath481 . in this paper , by imposing conditions which are explained later , this arbitrariness is reduced to some extent .
although there remains arbitrariness even with these conditions , our results do not depend on the choice of the division . | quantum measurement is a physical process .
a system and an apparatus interact for a certain time period ( measurement time ) , and during this interaction , information about an observable is transferred from the system to the apparatus . in this study , we quantify the energy fluctuation of the quantum apparatus required for this physical process to occur autonomously .
we first examine the so - called standard model of measurement , which is free from any non - trivial energy - time uncertainty relation , to find that it needs an external system that switches on the interaction between the system and the apparatus .
in such a sense this model is not closed .
therefore to treat a measurement process in a fully quantum manner we need to consider a
larger " quantum apparatus which works also as a timing device switching on the interaction . in this
setting we prove that a trade - off relation ( energy - time uncertainty relation ) , @xmath0 , holds between the energy fluctuation @xmath1 of the quantum apparatus and the measurement time @xmath2 .
we use this trade - off relation to discuss the spacetime uncertainty relation concerning the operational meaning of the microscopic structure of spacetime .
in addition , we derive another trade - off inequality between the measurement time and the strength of interaction between the system and the apparatus . |
let @xmath1 be a complex reductive group with maximal compact subgroup @xmath2 .
it has been shown in @xcite ( see also @xcite ) that smooth compact complex spherical @xmath1-varieties @xmath3 may be characterized by the fact that a moment map @xmath4 separates the @xmath2-orbits in @xmath3 . in
@xcite it has been shown that the so - called gradient maps are the right analogue for moment maps when one is interested in actions of a real reductive group @xmath5 .
spherical gradient manifolds have been introduced in @xcite in order to carry over brion s theorem to the real reductive case . to be more precise
, we call a @xmath6-gradient manifold @xmath7 with gradient map @xmath8 spherical if a minimal parabolic subgroup of @xmath6 has an open orbit in @xmath9 . if @xmath6 is connected complex reductive , then a minimal parabolic subgroup is the same as a borel subgroup of @xmath6 , so that there is no ambiguity in this definition .
the main result of @xcite states that @xmath9 is spherical if and only if @xmath10 almost separates the @xmath11-orbits in @xmath9 .
very recently , real spherical manifolds have attracted attention from the representation theoretical view point ( see @xcite and @xcite ) as well as from a geometric one ( see @xcite and @xcite ) . in @xcite
, the authors have shown that , given a homogeneous real spherical manifold @xmath12 , any minimal parabolic subgroup of @xmath6 has only finitely many orbits in @xmath9 . moreover
, the same paper contains the list of all reductive spherical subalgebras of @xmath13 . in
@xcite the author has found _ all _ decompositions of @xmath14 as the sum of two subalgebras . in this paper
we classify all spherical algebraic subalgebras of @xmath15 where @xmath15 is a simple lie algebra of real rank @xmath0 by methods in the spirit of @xcite .
although a subalgebra @xmath16 of @xmath15 is spherical whenever @xmath17 is a complex spherical subalgebra of @xmath18 , we would like to stress the fact that the converse is not true ( see the example in section 3 ) .
in particular one can not reduce the question to the complex classification .
the tables containing all spherical algebraic subalgebras are given in theorems [ reductiveorthogonal ] and [ nonredorthogonal ] for @xmath13 , in theorems [ reductiveunitary ] and [ nonredunitary ] for @xmath19 , in theorems [ reductivesymplectic ] and [ nonredsympl ] for @xmath20 , and in theorems [ reductiveexceptional ] and [ nonredexc ] for the exceptional lie algebra @xmath21 .
let us outline the main steps of the proof as well as the organization of this paper . in section 2 we show that the homogeneous manifold @xmath12 admits a @xmath6-gradient map if and only if @xmath22 is an algebraic subgroup of @xmath6 .
this is the reason why we classify spherical _ algebraic _ subalgebras . in section 3
we characterize reductive and non - reductive spherical algebraic subgroups @xmath22 of @xmath6 by the fact that a maximal compact subgroup of @xmath22 acts transitively on the spheres in a certain representation related to the inclusion @xmath23 .
more precisely our starting point is the following [ intro ] let @xmath24 be a connected simple lie group of real rank @xmath0 . 1 .
let @xmath25 be a reductive algebraic subgroup of @xmath6
. then @xmath22 is spherical if and only if @xmath26 acts transitively on the connected components of the spheres in @xmath27 .
2 . let @xmath28 be a non - reductive algebraic subgroup with @xmath29
. then @xmath22 is spherical if and only if @xmath30 and @xmath31 acts transitively on the spheres in @xmath32 .
the main technical work consists in reducing considerations to a normal form of the lie algebra of @xmath22 where we can carry out explicit calculations .
then we make essential use of onishchik s classification of transitive actions on spheres in order to single out those normal forms that are spherical . more details on this general scheme are given in section 4 . in the remaining sections 5 to 8 we carry out the necessary calculations case by case for @xmath14 , @xmath33 , @xmath34 and @xmath35 .
after this paper was finished , we learned that kimelfeld has considered the classification problem of algebraic spherical subgroups in real simple lie groups @xmath5 of rank @xmath0 , too . in @xcite he obtains proposition [ intro ] by differential geometric arguments based on the karpelevich compactification of the hyperbolic space @xmath36 and then gives a general description of all spherical algebraic subgroups of @xmath6
however , he does not provide an explicit list of all spherical algebraic subalgebras of @xmath15 .
_ acknowledgments .
_ we would like to thank peter heinzner and valdemar tsanov for helpful discussions on the subject presented here .
we are much obliged to friedrich knop for informing us about kimelfeld s paper @xcite as well as about an inaccuracy in the statement of theorem [ nonredorthogonal ] .
_ notation .
_ we denote lie groups by upper case roman letters while their lie algebras are denoted by the corresponding lower case gothic letters , i.e. , @xmath37 .
in this section , @xmath6 denotes a connected semisimple lie group that embeds as a closed subgroup into its universal complexification @xmath38 .
let @xmath2 be a compact real form of @xmath38 such that @xmath6 is stable under the corresponding cartan involution of @xmath39 .
then we obtain the cartan decomposition @xmath40 , @xmath41 , where @xmath42 is a maximal compact subgroup of @xmath6 and where @xmath43 . by a
_ @xmath6-gradient manifold _
we mean the following .
let @xmath44 be a khler manifold endowed with a holomorphic @xmath38-action such that there exists a @xmath2-equivariant moment map @xmath4 for the @xmath2-action on @xmath44 .
we call such a @xmath3 a _
hamiltonian @xmath38-manifold_. basic examples are given by @xmath38-stable complex submanifolds of some projective space @xmath45 where @xmath46 is a finite - dimensional complex @xmath38-representation space .
in particular , homogeneous algebraic @xmath38-varieties are hamiltonian . identifying @xmath47 with @xmath48 and composing @xmath49 with the orthogonal projection to @xmath50 with respect to a @xmath2-invariant inner product
, we obtain the @xmath11-equivariant _ gradient map _ @xmath51 .
any @xmath6-stable closed real submanifold @xmath9 of @xmath3 is called a @xmath6-gradient manifold with gradient map @xmath52 . for more details and the basic properties of gradient maps
we refer the reader to @xcite and @xcite .
let @xmath6 be a connected semisimple lie group and let @xmath22 be a closed subgroup of @xmath6 .
then @xmath12 is a @xmath6-gradient manifold if and only if @xmath22 is algebraic , i.e. , if @xmath53 holds for the zariski closure @xmath54 of @xmath22 in @xmath38 .
suppose first that @xmath53 holds .
it follows that @xmath12 is a real submanifold of the homogeneous space @xmath55 .
since the latter is quasi - projective , it is in particular khler , and since @xmath38 is semisimple , there exists a unique @xmath2-equivariant moment map on @xmath55 .
then the construction described above yields a @xmath6-gradient map on @xmath12 .
if @xmath12 is a gradient manifold , then by definition there exists a @xmath6-equivariant diffeomorphism @xmath56 where @xmath3 is a hamiltonian @xmath38-manifold . from this
we obtain @xmath57 , and @xmath58 is again khler .
therefore @xmath59 is an algebraic subgroup of @xmath38 , see ( * ? ? ?
* corollary 4.12 ) , hence contains @xmath54 .
this implies @xmath60 , as was to be shown .
as in the previous section let @xmath5 be a connected semisimple lie group .
let @xmath61 be a closed subgroup such that @xmath12 is a @xmath6-gradient manifold with gradient map @xmath8 .
we say that @xmath12 is _ spherical _ if a minimal parabolic subgroup of @xmath6 has an open orbit in @xmath9 .
in this case we call @xmath22 a spherical subgroup of @xmath6 and @xmath16 a spherical subalgebra of @xmath15 .
as shown in @xcite this is equivalent to the fact that @xmath10 almost separates the @xmath11-orbits in @xmath9 , i.e. , that the map @xmath62 induced by @xmath8 has discrete fibers .
for the rest of this paper we assume that _ @xmath6 is simple and has real rank @xmath0_.
then @xmath15 is isomorphic to either @xmath14 ( @xmath63 ) or @xmath33 ( @xmath64 ) or @xmath34 ( @xmath65 ) or the exceptional lie algebra @xmath66 , see ( * ? ? ?
vi.11 ) . as our main result
we will classify the algebraic spherical subalgebras of @xmath15 up to conjugation by an element of @xmath6 , i.e. , those subalgebras @xmath16 for which there exists a closed subgroup @xmath61 having @xmath16 as lie algebra such that @xmath12 is a spherical gradient manifold .
any symmetric subalgebra of @xmath15 is spherical in @xmath15 , see @xcite .
it is not hard to see that , if @xmath17 is spherical in @xmath18 in the usual sense , then @xmath16 is spherical in @xmath15 .
however , as the following examples show , the converse does not hold , i.e. , there are more spherical subalgebras of @xmath15 than just real forms of complex spherical subalgebras of @xmath18 .
the unipotent radical @xmath67 of a minimal parabolic subgroup of @xmath6 is always spherical in @xmath6 .
however , in general @xmath68 is not a spherical subgroup of @xmath38 . as a concrete example one may take @xmath69 .
a semisimple example is given by the spherical subgroup @xmath70 of @xmath71 ( see section [ symplectic ] ) since @xmath72 is not spherical in @xmath73 , see @xcite .
for the classification we distinguish the cases that @xmath22 is reductive or not .
if @xmath22 is reductive , then by definition we may assume that , after conjugation by an element of @xmath6 , we have a cartan decomposition @xmath74 where @xmath75 is a maximal compact subgroup of @xmath22 and where @xmath76 is a @xmath26-invariant subspace of @xmath77 with @xmath78\subset{{\mathfrak{k}}}_h$ ] .
we write @xmath79 with respect to the @xmath11-invariant inner product on @xmath77 that comes from the @xmath2-invariant inner product on @xmath48 . in this situation ,
sphericity of @xmath12 has been characterized in ( * ? ? ?
* proposition 6.1 ) . under the additional assumption
@xmath80 we can make the following more precise statement .
[ lonqperp ] let @xmath5 be a connected simple lie group of real rank @xmath0 and let @xmath81 be a closed reductive subgroup
. then @xmath12 is spherical if and only if @xmath26 has an open orbit in every sphere in @xmath82 . except for @xmath83 proposition [ lonqperp ]
says that @xmath12 is spherical if and only if @xmath26 acts transitively on the spheres in @xmath84 .
in particular , @xmath9 can only be spherical if the @xmath26-representation on @xmath84 is irreducible . by ( * ? ? ? *
theorem 1.1 ) the homogeneous gradient manifold @xmath12 is spherical if and only if any gradient map on it almost separates the @xmath11-orbits .
the mostow decomposition ( see @xcite for a proof using gradient maps ) exhibits @xmath9 as @xmath11-equivariantly isomorphic to the twisted bundle @xmath85 . is a subgroup of @xmath6 and @xmath86 is a set on which @xmath22 acts , then the twisted bundle @xmath87 is defined as the quotient of @xmath88 with respect to the diagonal @xmath22-action @xmath89 . ]
a particular gradient map is given by @xmath90=\operatorname{ad}(k)\xi$ ] for @xmath91 and @xmath92 .
since the @xmath11-orbits in @xmath85 correspond to the @xmath26-orbits in @xmath84 , this map @xmath10 separates the @xmath11-orbits if and only if the map @xmath93 that is induced by the inclusion @xmath94 , has discrete fibers . since these fibers are precisely the @xmath26-orbits in the compact sets @xmath95 with @xmath92 , we note in particular that the fibers have to be finite . in the case
@xmath80 , the @xmath11-orbits in @xmath77 are spheres , so their intersections with any subspace of @xmath77 are again spheres and in particular connected ( unless the subspace is a line ) .
thus , @xmath9 is spherical if and only if @xmath26 has an open orbit in every sphere @xmath95 , @xmath92 , hence in any sphere in @xmath84 .
in the rest of this section we give a similar criterion for non - reductive @xmath22 .
for this we fix a minimal parabolic subalgebra @xmath96 of @xmath15 such that @xmath97 is a maximal abelian subspace ( i.e. , a line ) of @xmath77 .
the corresponding group is @xmath98 where @xmath99 .
let @xmath16 be a non - reductive algebraic subalgebra of @xmath15 and let @xmath22 be a corresponding subgroup of @xmath6 . as is shown in ( * ? ? ?
* lemma 3.1 ) , after conjugation by an element of @xmath6 we may assume that @xmath100 where @xmath101 is reductive in @xmath15 and @xmath102 is a nilpotent ideal of @xmath16 .
on the group level we have @xmath103 with a reductive group @xmath104 and @xmath105 . note that @xmath106 acts by conjugation on @xmath67 and stabilizes @xmath107 , hence acts on @xmath108 . on the lie algebra level we have the decomposition @xmath109 of @xmath110 as an @xmath31-module .
[ nonreductive ] let @xmath12 be a @xmath6-gradient manifold such that @xmath28 is non - reductive .
then @xmath9 is spherical if and only if either 1 .
@xmath111 and @xmath112 is arbitrary , or 2 .
@xmath113 and @xmath30 and @xmath114 is arbitrary , or 3 .
@xmath115 and @xmath30 and @xmath114 acts transitively on the spheres in @xmath32 .
every algebraic subgroup @xmath61 that contains @xmath67 is spherical .
moreover , the proof of proposition [ nonreductive ] will show that every @xmath22 containing @xmath116 with @xmath113 is spherical .
therefore , we will concentrate on the case @xmath117 .
let @xmath22 be a non - reductive algebraic subgroup of @xmath6 of the form @xmath118 .
since we have the @xmath6-equivariant fiber bundle @xmath119 , we see that @xmath9 is @xmath6-equivariantly diffeomorphic to the twisted product @xmath120 . by
ix.1.8 ) the unique open orbit of the opposite minimal parabolic subgroup @xmath121 in @xmath122 is @xmath123 .
this implies that @xmath12 is spherical if and only if there is some @xmath124 such that @xmath125 $ ] is open in @xmath120 .
the latter is the case if and only if @xmath126 has an open orbit in @xmath127 .
using the fact that @xmath107 is normal in @xmath22 , we see that @xmath127 is @xmath128-equivariantly diffeomorphic to @xmath129 where @xmath106 acts by conjugation on @xmath108 .
this proves that @xmath12 is spherical if and only if @xmath106 has an open orbit in @xmath108 . since @xmath130 is an @xmath106-equivariant diffeomorphism ,
we conclude that sphericity of @xmath9 is equivalent to the fact that @xmath106 has an open orbit in @xmath32 .
suppose now that @xmath131 .
then @xmath106 can have an open orbit in @xmath32 only if it contains @xmath132 , i.e. , if @xmath30 . if @xmath133 , this condition is also sufficient . since every @xmath132-orbit in @xmath32 intersects any sphere in @xmath32 precisely once
, the claim follows .
as we have seen , in order to classify non - reductive spherical algebraic subalgebras @xmath134 we may assume without loss of generality @xmath117 .
in particular , @xmath31 must act irreducibly on @xmath32 if @xmath12 is spherical . in closing this section
, we will exploit this observation a little further .
we have @xmath135 according to the restricted root space decomposition of @xmath15 with respect to @xmath97 where @xmath136 is a simple restricted root . if @xmath137 is not a restricted root , we set @xmath138 .
( note that this is only the case for @xmath139 . )
the group @xmath140 acts irreducibly on every root space , see e.g. ( * ? ?
* ch . 8.13 ) .
suppose that @xmath12 is spherical with @xmath141 .
since @xmath142 acts with two different weights on @xmath143 and @xmath144 , we obtain @xmath145 consequently , @xmath31 acts irreducibly on @xmath32 only if @xmath32 contains @xmath143 or @xmath144 .
this proves the following [ n_henthltwurzelraum ] let @xmath146 be a spherical non - reductive algebraic subgroup of @xmath6 .
then @xmath147 contains @xmath143 or @xmath144 .
[ rootspacecontinnh ] arguing case by case we will see that @xmath148= { { \mathfrak{g}}}_{2\alpha}$ ] holds for every simple lie algebra @xmath15 of real rank @xmath0 .
this implies that @xmath147 always contains @xmath144 .
let us outline the principal steps that will lead to the classification result . recall that @xmath5 is a connected simple lie group of real rank @xmath0 that embeds into @xmath38 .
let @xmath134 be an algebraic subalgebra which we assume first to be reductive .
after conjugation by an element of @xmath6 we have @xmath149 .
the first step in the classification consists in analyzing the @xmath11-action on the grassmannians @xmath150 of real @xmath151-planes in @xmath77 for classical @xmath6 so that we can conjugate @xmath152 by an element of @xmath11 into a suitable normal form .
one verifies directly that for classical @xmath6 the @xmath11-representation on @xmath77 is given as follows .
if @xmath153 , then we have the defining representation of @xmath154 on @xmath155 .
for @xmath156 we have @xmath157 and may identify @xmath77 with @xmath158 due to the hermitian structure on @xmath36 . under these identifications
we obtain the defining representation of @xmath159 on @xmath158 . if @xmath160 , then @xmath161 and @xmath162 .
on @xmath163 we consider the quaternionic inner product @xmath164 where @xmath165 is the quaternionic conjugate of @xmath166 . by definition , @xmath167 is the isometry group of @xmath168 .
the action of @xmath169 on @xmath163 is given by @xmath170 .
the map @xmath171 is a complex - linear isomorphism and induces the map @xmath172 , @xmath173 since @xmath168 corresponds under @xmath174 to the pair of the standard hermitian form and the standard complex symplectic form on @xmath175 , we obtain @xmath176 . in the following , we will usually identify @xmath177 with the matrix @xmath178 . under this identification ,
the representation of @xmath179 on @xmath77 is given by @xmath180 [ grassmann ] let @xmath5 be a connected classical simple lie group of real rank @xmath0 .
the orbit structure of @xmath6 on the grassmanian @xmath181 of real @xmath151-planes in @xmath77 is as follows .
a. if @xmath182 , then @xmath154 acts transitively on @xmath150 for any @xmath151 . b. if @xmath183 , then two real @xmath151-planes in @xmath77 lie in the same orbit of @xmath157 if and only if their maximal complex subspaces have the same dimension . c. if @xmath160 , then any real subspace of @xmath77 is after conjugation with an element of @xmath184 equal to @xmath185 for unique @xmath186 and @xmath187 with @xmath188 for @xmath189 .
let us first prove the claim for the case @xmath190 .
let @xmath46 be a real subspace of @xmath158 and @xmath191 be a maximal complex subspace of @xmath46 with orthonormal basis @xmath192 .
let @xmath193 be an orthonormal basis of @xmath194 .
then the matrix @xmath195 is in @xmath196 and maps @xmath197 to @xmath46 .
analogously one proves the claim for @xmath198 .
let us now consider the @xmath199-action on real @xmath151-planes in @xmath163 .
suppose first that @xmath200 .
since @xmath201 acts transitively on @xmath202 , any one- or three - dimensional real subspace of @xmath203 can be mapped onto @xmath204 or @xmath205 , respectively , by a suitable element of @xmath201 .
any real plane can be mapped onto @xmath206 with @xmath207 .
note that there is no element in @xmath201 that maps @xmath208 onto @xmath209 if @xmath210 . in the general case
, we will show that any real @xmath151-dimensional subspace @xmath211 contains a basis of its quaternionic span @xmath212 over @xmath203 that is orthonormal with respect to @xmath168 .
consequently , there exists an element in @xmath167 that maps this basis to a subset of the standard basis of @xmath163 .
re - ordering if necessary , and acting with elements of @xmath213 we obtain therefore the following normal form given by @xmath214 where @xmath187 with @xmath188 for @xmath189 . without loss of generality
we may assume @xmath215 .
let @xmath216 be the maximal quaternionic subspace of @xmath46 .
since @xmath217 , we may assume @xmath218 .
we decompose @xmath46 into the sum @xmath219 , where @xmath220 choosing a @xmath168-orthonormal basis of @xmath46 compatible with this decomposition we proved the existence of the adapted orthonormal basis . in the next step
we conjugate @xmath22 by an element of @xmath11 such that @xmath152 is of the normal form defined by the help of lemma [ grassmann ] . in order to understand all possibilities for @xmath26 we determine @xmath221 and @xmath78 $ ] since we know that @xmath222\subset{{\mathfrak{k}}}_h\subset\mathcal{n}_{{\mathfrak{k}}}({{\mathfrak{p}}}_h)\ ] ] must hold . due to proposition [ lonqperp ]
the subalgebra @xmath16 is spherical if and only if @xmath26 acts transitively on the ( connected components of the ) spheres in @xmath84 .
therefore , in the third and last step we single out those groups @xmath26 fulfilling which do indeed act transitively on the spheres in @xmath152 . in this step
we will make essential use of onishchik s classification of transitive actions on spheres .
we recall his result in the following theorem where we consider the defining representations of @xmath223 on @xmath224 , of @xmath225 on @xmath158 and of @xmath199 on @xmath163 .
[ thm2on ] let @xmath11 be either @xmath226 or @xmath159 or @xmath167 and let @xmath46 denote the respective defining representation of @xmath11 .
the following table lists all connected proper subgroups @xmath227 of @xmath11 that act transitively on the spheres in @xmath46 , up to conjugation in @xmath11 .
note that @xmath228 does not contain such a subgroup . since a connected subgroup of @xmath223 lies in @xmath229 and since @xmath229 ( @xmath63 , @xmath230 ) and @xmath199 are simple , the result follows directly from ( * ? ? ?
* table 8) for these groups .
let us discuss the case @xmath231 .
identifying @xmath232 with @xmath203 one sees that @xmath233 .
the isotropy algebra @xmath234 of @xmath235 corresponds to the diagonal @xmath236 in @xmath237 .
therefore we have to find all subalgebras @xmath238 of @xmath237 ( up to conjugation ) that verify @xmath239 consider the projections @xmath240^{\pi_2}\ar[d]_{\pi_1 } & { { \mathfrak{sp}}}(1)\\ { { \mathfrak{sp}}}(1 ) . & } \ ] ] let @xmath238 be a subalgebra of @xmath237 verifying .
this implies @xmath241 where @xmath242 is an ideal in @xmath238 .
if @xmath243 , then @xmath244 . if @xmath245 , then @xmath246 is a maximal torus in @xmath247 and @xmath248 , hence @xmath249 . finally , suppose that @xmath250 and note that @xmath251 .
if @xmath252 , then @xmath253 . if @xmath254 , then @xmath255 and thus @xmath238 is simple . therefore @xmath256 is either @xmath257 ( and then @xmath258 ) or @xmath259 . in the latter case @xmath260 and @xmath261 are isomorphisms onto @xmath247 and @xmath238 coincides with the graph of @xmath262 .
since we can identify @xmath263 with the space of fixed points @xmath264 in this case and since @xmath265 , we obtain @xmath266 , hence @xmath238 can not verify . finally consider the case @xmath267 .
then @xmath268 and @xmath269 is a one - dimensional ideal in @xmath238 which implies @xmath270 .
thus @xmath271 . in the unitary case we use ( * ? ? ?
* theorem 1.5.1 ) in order to reduce the classification to @xmath272 .
we denote the isotropy algebra @xmath273 of @xmath274 by @xmath275 . as above
we look for subalgebras @xmath238 of @xmath276 that verify @xmath277 . according to (
* theorem 1.5.1 ) this holds if and only if @xmath278 ( where @xmath279 is the derived algebra of @xmath238 ) and @xmath280 where @xmath281 is the projection of @xmath276 onto its center with kernel @xmath282 .
note that the second condition is automatically verified since already @xmath283 .
consequently , the result follows from the classification of subgroups of @xmath196 that act transitively on @xmath284 given in ( * ? ? ?
* table 8) .
the non - reductive case will be treated along the same lines as follows .
as we have seen above , after conjugating by an element of @xmath6 we may suppose that any non - reductive algebraic subalgebra @xmath134 is of the form @xmath285 .
we always assume that @xmath131 so that we have @xmath286 . then @xmath22 is spherical if and only if @xmath31 acts transitively on the connected components of the spheres in @xmath32 . as above
we start by analyzing the @xmath140-representation on @xmath110 in order to find a normal form for @xmath147 .
afterwards we determine @xmath287 for this normal form since this normalizer must contain @xmath31 .
finally , we apply again onishchik s result theorem [ thm2on ] in order to single out those @xmath31 that do act transitively on the spheres in @xmath32 .
let @xmath16 be a reductive algebraic subalgebra of @xmath13 . to simplify the notation we write @xmath288 for its cartan decomposition . on the group level we have @xmath289 .
it follows from lemma [ grassmann ] that @xmath154 acts transitively on @xmath290 .
hence , we may assume @xmath291 for some @xmath292 .
[ [ q , q],n ] for @xmath292 we have @xmath293&=\left\{\left(\begin{smallmatrix}0&0&0\\0&b&\vdots \\ 0&\ldots&0\end{smallmatrix}\right)\in\mathfrak{k } : b\in { \mathfrak{so}}(n - k)\right\},\\ \mathcal{n}_\mathfrak{k}({\mathfrak{q}}_k)&=\left\{\left(\begin{smallmatrix } a&0&0\\0&b&0\\0&0&0 \end{smallmatrix}\right)\in\mathfrak{k } : a\in { \mathfrak{so}}(k ) , b\in { \mathfrak{so}}(n - k)\right\}.\end{aligned}\ ] ] since @xmath294 $ ] is the vector space generated over the reals by elements @xmath295 $ ] , where @xmath296,@xmath297 are in @xmath298 , it suffices to calculate the bracket for basis elements of @xmath298 .
let @xmath299 be the standard basis of @xmath224
. then @xmath300=\left ( \begin{smallmatrix}e_ie_j^t&0\\0&e_i^te_j\end{smallmatrix}\right)-\left ( \begin{smallmatrix}e_je_i^t&0\\0&e_j^te_i\end{smallmatrix}\right)=\left ( \begin{smallmatrix}e_ie_j^t - e_je_i^t&0\\0&0\end{smallmatrix}\right)=\left ( \begin{smallmatrix}e_{ij}-e_{ji } & 0\\0&0\end{smallmatrix}\right).\ ] ] hence @xmath294=\left\{\left(\begin{smallmatrix}0&0&0\\0&a&\vdots \\ 0&\ldots&0\end{smallmatrix}\right)\in\mathfrak{k } : a\in { \mathfrak{so}}(n - k)\right\}$ ] for @xmath301 .
the second claim follows from direct calculation . since @xmath302\oplus{{\mathfrak{q}}}_k\cong{{\mathfrak{so}}}(n - k,1)$ ] ( with @xmath303 )
, we have shown the following any reductive subalgebra @xmath304 of @xmath305 is conjugate under @xmath6 to one of the form @xmath306 , where @xmath307 is any subalgebra of @xmath308 and @xmath292 .
note that the action of @xmath309 on @xmath310 is given by @xmath311 hence , combining this result with proposition [ lonqperp ] and theorem [ thm2on ] we obtain the following list of all spherical reductive algebraic subalgebras of @xmath14 up to conjugacy in @xmath6 .
this list has been found in ( * ? ? ?
* remark 5.1 ) .
[ reductiveorthogonal ] all spherical reductive algebraic subalgebras @xmath304 of @xmath312 are ( up to conjugation in @xmath6 ) one of the following , where @xmath313 and @xmath314 is arbitrary .
according to ( * ? ? ?
* table 2 ) , the symmetric subalgebras are @xmath315 and @xmath316 for @xmath317 . the only subalgebras @xmath16 of @xmath312 in the above list that might be conjugate to each other by an element of @xmath6 are @xmath318 and @xmath319 .
note that these subalgebras are indeed conjugate by an element of @xmath6 if and only if @xmath320 .
in this subsection we classify the spherical non - reductive algebraic subalgebras @xmath16 of @xmath13 . due to proposition [ nonreductive ]
we can assume @xmath321 , where @xmath322 and @xmath323 .
after understanding the @xmath140-action on @xmath324 we are going to find a normal form of @xmath325 and of @xmath326 .
we then only have to classify the subgroups @xmath327 that act transitively on the connected components of the spheres in @xmath328 .
first we determine the structure of a minimal parabolic subalgebra of @xmath13 .
choosing @xmath329 as maximal abelian subspace of @xmath77 , we obtain @xmath330 with @xmath331 moreover , we have @xmath332 identifying @xmath110 with @xmath333 via @xmath334 we see that the @xmath140-representation on @xmath110 is isomorphic to the defining representation of @xmath335 on @xmath333 .
thus , lemma [ grassmann ] applies to show that any subspace @xmath325 of @xmath324 is @xmath140-conjugate to @xmath336 where @xmath151 is the dimension of @xmath328 in @xmath324 .
the normalizer of @xmath337 in @xmath140 is given by @xmath338 and its action on @xmath339 is @xmath340-equivariantly isomorphic to the standard @xmath340-action on @xmath341 . combining these observations with proposition [ nonreductive ] and theorem [ thm2on ] , we obtain the following [ nonredorthogonal ] every spherical non - reductive algebraic subalgebra of @xmath312 is @xmath6-conjugate to one in the following list where @xmath342 and @xmath314 are arbitrary ( under the condition that the maximal compact subalgebra is a lie algebra , see remark [ non - reductive , embedding ] below ) . [ non - reductive , embedding ]
let us explain the notation in theorem [ nonredorthogonal ] .
let @xmath343 denote the projection onto the first factor .
an arbitrary subalgebra @xmath344 is of the form @xmath345 where @xmath346 is a homomorphism of lie algebras . in order to simplify the notation
, we write here and in the rest of this paper @xmath347 where @xmath348 . the condition that @xmath349 is
a subalgebra means that @xmath350 is contained in the centralizer of @xmath351 in @xmath352 .
note that , if @xmath353 is simple , then @xmath174 is either identically zero or injective .
let @xmath354 be a reductive algebraic subalgebra of @xmath305 . due to lemma [ grassmann ] two real subspaces of @xmath77
are conjugate to each other by an element of @xmath11 if and only if they have the same real dimension and their maximal complex subspaces have the same dimension .
hence , we may assume @xmath355 where @xmath356 is the real dimension of @xmath357 and @xmath151 the complex dimension of the maximal complex subspace in @xmath357 .
[ [ qkl , qkl ] ] for @xmath358 we have @xmath359&=\left\{\left(\begin{smallmatrix}0&0&0&0\\0&a&b&0\\ 0&-\overline{b}^t&c&0\\0&0&0&ix\end{smallmatrix}\right)\in\mathfrak{k } : a\in{\mathfrak{so}}(l ) , c\in\mathfrak{u}(n - k - l ) , ix=-\operatorname{tr}(a)-\operatorname{tr}(c)\right\}\\ & \subset\mathfrak{s}(\mathfrak{u}(n - k)\times\mathfrak{u}(1)),\\ \mathcal{n}_\mathfrak{k}({\mathfrak{q}}_{k , l})&=\left\{\left(\begin{smallmatrix } a&0&0&0\\0&b&0&0\\ 0&0&c&0\\0&0&0&ix\end{smallmatrix}\right)\in\mathfrak{k } : a\in \mathfrak{u}(k ) ,
c\in{\mathfrak{u}}(n - k - l ) , b - ix\operatorname{id}\in { \mathfrak{so}}(l),x\in{\mathbb{r}}\right\}.\end{aligned}\ ] ] let @xmath360 be the standard basis of @xmath158 and @xmath361 elements of @xmath362 . then @xmath363 = \left(\begin{smallmatrix}z\overline{w}e_{ij}&0\\0&\overline{z}w\delta_{ij } \end{smallmatrix}\right)-\left(\begin{smallmatrix}w\overline{z}e_{ji}&0\\ 0&\overline{w}z\delta_{ij}\end{smallmatrix}\right)=\left(\begin{smallmatrix } z\overline{w}e_{ij}-w\overline{z}e_{ji}&0\\0&(\overline{z}w-\overline{w } z)\delta_{ij } \end{smallmatrix}\right).\ ] ]
note that @xmath364 the second claim follows from direct calculation .
since @xmath365\subset\mathfrak{l}\subset \mathcal{n}_\mathfrak{k}({\mathfrak{q}}_{k , l})$ ] has to hold , either @xmath366 or @xmath367 has to be zero .
therefore , using @xmath368\oplus { \mathfrak{q}}_{k,0}\cong{\mathfrak{su}}(n - k,1)$ ] and @xmath369\oplus { \mathfrak{q}}_{k , n - k}\cong{\mathfrak{so}}(n - k,1)$ ] for @xmath370 we have the following result .
any reductive subalgebra @xmath304 of @xmath305 is either compact or of the form @xmath371 or @xmath372 , where @xmath307 is a subalgebra of @xmath373 and @xmath370 .
the following lemma implies that subalgebras of the form @xmath372 are almost never spherical .
let @xmath134 be of the form @xmath372 , where @xmath307 is a subalgebra of @xmath373 and @xmath370
. then @xmath16 is spherical in @xmath305 if and only if @xmath374 .
the action of @xmath375 on @xmath376 is given by @xmath377 and is therefore not irreducible if @xmath378 . if @xmath374 then @xmath379 acts transitively on the spheres in @xmath380 .
therefore the claim follows from proposition [ lonqperp ] .
hence , combining these results with proposition [ lonqperp ] and theorem [ thm2on ] we can prove the following list of all spherical reductive algebraic subalgebras of @xmath33 up to conjugacy in @xmath6 .
[ reductiveunitary ] all spherical , reductive subalgebras @xmath304 of @xmath381 are ( up to conjugation in @xmath6 ) one of the following according to ( * ? ? ? * table 2 ) the symmetric subalgebras are @xmath382 and @xmath383 and @xmath384 and @xmath312 .
let @xmath22 now be a non - reductive algebraic subgroup of @xmath6 . due to proposition [ nonreductive ]
we may assume that it is of the form @xmath321 , where @xmath385 and @xmath327 .
after understanding the @xmath140-action on @xmath324 we are going to find a normal form of @xmath325 and of @xmath326 .
we then only have to classify the subgroups @xmath386 that act transitively on the connected components of the spheres in @xmath328 .
choosing @xmath387 as a maximal abelian subspace of @xmath388 , we obtain @xmath389 with @xmath390 moreover , we have @xmath391 the map @xmath392 with @xmath393 is a @xmath394-equivariant isomorphism with respect to the adjoint action of @xmath140 on @xmath324 and the @xmath394-action on @xmath395 given by @xmath396 .
since @xmath397={\mathfrak{g}}_{2\alpha}$ ] holds , lemma [ grassmann ] implies that any subspace @xmath147 of @xmath110 is @xmath140-conjugate to @xmath398 where @xmath356 is the real dimension of @xmath328 in @xmath324 and @xmath151 the complex dimension of the maximal complex subspace of @xmath328 in @xmath324 .
the normalizer of @xmath399 in @xmath140 is given by @xmath400 and its action on @xmath401 is given by @xmath402 .
note that this action is reducible if @xmath151 and @xmath366 are both nonzero . combining these observations with proposition [ nonreductive ] and theorem [ thm2on ]
we obtain the following [ nonredunitary ] every spherical non - reductive algebraic subalgebra of @xmath381 is @xmath6-conjugate to one in the following list where @xmath403 , @xmath404 and @xmath314 are arbitrary ( under the condition that the maximal compact subalgebra is a lie algebra , see remark [ non - reductive , embedding ] ) .
since @xmath406 ( * ? ? ? * ch . x 6.4 )
we assume @xmath407 .
the lie algebras @xmath305 , @xmath408 and @xmath388 have the following form @xmath409 let @xmath354 be a reductive algebraic subalgebra of @xmath410 .
it follows from lemma [ grassmann ] that after conjugation with an element in @xmath11 , we may assume that @xmath411 equals @xmath412 where @xmath186 , @xmath413 and @xmath414 is the standard real basis of the quaternions .
[ [ qklm , qklm ] ] for @xmath415 we have @xmath416\supset\\ & \left\{\left(\left(\begin{smallmatrix}a&-{\overline{b}}\\b & \overline{a}\end{smallmatrix}\right),\left(\begin{smallmatrix}0&0\\0 & 0\end{smallmatrix}\right)\right)\in\mathfrak{k } : a=\kbordermatrix{&k&l&m&p&\\k&0&0&0&0&0\\l&0&\star&\star & * & * \\m&0&\star&\star & * & * \\ p&0 & * & * & * & * \\&0 & * & * & * & * } , b=\kbordermatrix{&k&l&m&p&\\k&0&0&0&0&0\\l&0&0&c_1&\star & * \\m&0&c_1^t&0&\star & * \\ p&0&\star&\star & * & * \\ & 0 & * & * & * & * } , \right.\\ & \left.\quad\quad \star\text { only has real entries } , 0\neq c_1=c_1(\xi_1,\ldots\xi_m)\right\},\\\end{aligned}\ ] ] @xmath417 the claim follows from direct calculation .
since @xmath418\subset\mathfrak{l}\subset \mathcal{n}_\mathfrak{k}({\mathfrak{q}}_{k , l , m , p,\xi})$ ] has to hold , we have shown the following [ threecases ] if @xmath304 is a reductive algebraic subalgebra of @xmath305 with @xmath419 , then either @xmath420 ( i.e. , @xmath421 ) , @xmath422 ( i.e. , @xmath423 ) or @xmath424 ( i.e. , @xmath425 ) has to hold . considering these cases separately we prove the following result .
[ reductivesymplectic ] all spherical reductive algebraic subalgebras @xmath304 of @xmath410 are ( up to conjugation in @xmath6 ) one of the following according to ( * ? ? ?
* table 2 ) the symmetric subalgebras are @xmath426 and @xmath427 and @xmath428 and @xmath429 . if @xmath420 and @xmath430 then @xmath431 is compact
. therefore proposition [ lonqperp ] and theorem [ thm2on ] yield the first three cases in the list .
if @xmath420 and @xmath432 we have @xmath433\oplus{\mathfrak{q}}_{k,0,0,0}\subset\mathfrak{h } \subset \mathcal{n}_\mathfrak{k}({\mathfrak{q}}_{k,0,0,0})\oplus{\mathfrak{q}}_{k,0,0,0 } \cong{\mathfrak{sp}}(k)\oplus{\mathfrak{sp}}(n - k,1)\ ] ] and the action of @xmath434 on @xmath435 is given by @xmath436 therefore proposition [ lonqperp ] and theorem [ thm2on ] yield the next three lines in the list .
now let us consider the case @xmath422 .
if @xmath437 we are in the case we just considered .
we therefore additionally assume @xmath438 .
the action of @xmath439 on @xmath440 is reducible if @xmath441 .
if @xmath374 we can decompose @xmath442 into @xmath443 calculation shows that the action of @xmath442 on @xmath444 is not transitive on the spheres in @xmath444 .
let us now assume @xmath424 .
if @xmath445 we are again in the first case .
we therefore assume @xmath438 and obtain @xmath446 the action of @xmath447 on @xmath448 is reducible if @xmath441 .
we therefore assume @xmath374 . for dimensional reasons
@xmath449 can only act transitively on the spheres in @xmath450 if @xmath451 for all @xmath452 .
we denote @xmath453 , where @xmath454 .
recall that @xmath455 .
if @xmath456 we obtain @xmath457\oplus{\mathfrak{q}}_{0,0,n,0,\xi}\\ & \subset\mathfrak{h}\subset \mathcal{n}_\mathfrak{k}({\mathfrak{q}}_{0,0,n,0,\xi})\oplus{\mathfrak{q}}_{0,0,n,0,\xi}={\mathfrak{u}}(n,1).\end{aligned}\ ] ] if @xmath458 we obtain @xmath459 conjugation with @xmath460 , where @xmath461 maps this lie algebra to @xmath462 conjugation with @xmath463 maps it then to @xmath464 while the algebra @xmath465\oplus{\mathfrak{q}}_{0,0,n,0,\xi}$ ] is mapped to @xmath381 .
if @xmath466 and @xmath467 are non zero we obtain @xmath468 conjugation with @xmath460 , where @xmath469 maps this lie algebra to @xmath470 conjugation with @xmath471 , where @xmath472 , @xmath473 maps it then to @xmath462 hence we obtain again @xmath474\oplus{\mathfrak{q}}_{0,0,n,0,\xi}\cong{\mathfrak{su}}(n,1).\ ] ] the action of @xmath475 on @xmath476 is given by @xmath477 since the compact part of @xmath478 already acts transitively on the spheres in @xmath479 , proposition [ lonqperp ] and theorem [ thm2on ] yield the last two cases in the list . choosing @xmath480 we obtain the following subalgebras of @xmath305 .
@xmath481 moreover , we have @xmath482 the identification of @xmath110 with @xmath483 indicated above is a @xmath140-equivariant isomorphism with respect to the adjoint @xmath140-action on @xmath110 and the @xmath484-action defined by @xmath485 for all @xmath486 , @xmath487 . since @xmath397={\mathfrak{g}}_{2\alpha}$ ] holds , lemma [ grassmann ] shows that we may assume that @xmath147 equals @xmath488 where @xmath186 and @xmath489 . as
before we have @xmath490 since the exponential mapping does not destroy the block structure , we conclude that the action of @xmath491 on @xmath492 is reducible if more than one of the indices @xmath493 is non - zero .
considering each of these cases separately and using proposition [ nonreductive ] and theorem [ thm2on ] we prove the following result .
[ nonredsympl ] every spherical non - reductive algebraic subalgebra of @xmath410 is @xmath6-conjugate to one in the following list , where @xmath494 , @xmath495 , @xmath496 and @xmath314 are arbitrary ( under the condition that the maximal compact subalgebra is a lie algebra , see remark [ non - reductive , embedding ] ) .
if @xmath420 the action of @xmath497 on @xmath498 is given by @xmath499 for all @xmath500 , @xmath501 , @xmath502 and @xmath503 .
the action of @xmath504 on @xmath505 is not transitive on the spheres if @xmath506 .
if @xmath507 the action is given by @xmath508 in the case @xmath509 we have ( analogously to what we obtained in the reductive case ) that the group @xmath510 is @xmath140-conjugate to @xmath511 the action of the above group on @xmath512 is given by @xmath513 respectively @xmath514 .
the action of @xmath515 on @xmath516 is given by @xmath513 respectively @xmath517 .
in this section let @xmath24 be the connected simple group @xmath519 of real rank @xmath0 .
the maximal compact subgroup @xmath11 is isomorphic to @xmath520 while @xmath521 , see @xcite .
we want to classify the spherical reductive algebraic subalgebras @xmath522 of @xmath523 . using the classification of maximal reductive subalgebras of @xmath15 , krtz and schlichtkrull
have shown that a spherical reductive subalgebra of @xmath15 must be contained in a symmetric one , see ( * ? ? ?
* lemma 6.2 ) .
according to @xcite , the symmetric subalgebras of @xmath524 are @xmath525 , @xmath526 and @xmath527 .
the next lemma determines @xmath411 .
[ q = pperp ] let @xmath528 be a symmetric subgroup of @xmath6 ( defined as the fixed point set of an involution @xmath529 that commutes with the cartan involution ) and @xmath530 a subgroup of @xmath531 .
the adjoint @xmath227-action on @xmath388 can only be transitive on the spheres in @xmath532 if @xmath533 . since @xmath304 is a subalgebra of @xmath534
, @xmath411 lies in @xmath535 which implies that @xmath536 is the direct sum of @xmath537 and @xmath538 . since the adjoint @xmath539 action on @xmath388 stabilizes @xmath535 and @xmath537 , @xmath227
can only act irreducibly on @xmath532 if @xmath533 .
if @xmath534 is simple and non - compact , then we have @xmath540 $ ] .
these observations are sufficient to handle the case that @xmath304 is a subalgebra of @xmath525 or @xmath526 . for the case
that @xmath304 is a subalgebra of @xmath527 we also need the following technical lemma .
[ tensor ] let @xmath541 be compact and @xmath46 be an irreducible representation of @xmath6 .
then @xmath46 is @xmath6-equivariantly isomorphic to @xmath542 , where @xmath543 and @xmath544 are @xmath545-vector spaces , @xmath2 is an irreducible @xmath546-representation and @xmath191 is an irreducible @xmath547-representation .
we decompose @xmath46 into isotypical components with respect to @xmath547 , i.e. @xmath548 . the group @xmath546 leaves each @xmath549 invariant since the @xmath546-action on @xmath46 commutes with the @xmath547-action on @xmath46 .
therefore each @xmath549 is @xmath6-invariant .
since @xmath46 is assumed to be irreducible , @xmath46 is just one isotypical component .
therefore we can write @xmath550 , where @xmath191 is an irreducible representation of @xmath547 .
we define the bilinear map @xmath552 by @xmath553 .
note that @xmath191 is a @xmath545-vector space with respect to multiplication from the right .
therefore @xmath550 is also a @xmath545-vector space .
this induces a @xmath545-vector space structure on @xmath554 by @xmath555 for all @xmath556 , @xmath557 and @xmath558 .
the map @xmath559 is @xmath545-bilinear .
hence it induces a map @xmath560 that we also denote by @xmath559 .
we define a @xmath561-action on @xmath562 by @xmath563 , where the action of @xmath547 on @xmath191 is the given one and the action of @xmath546 on @xmath554 is induced by the @xmath546-action on @xmath46 via @xmath564 .
the map @xmath559 is @xmath6-equivariant with respect to the given actions .
since the @xmath6-action on @xmath46 is irreducible ( and @xmath559 is not the zero map ) @xmath559 has to be surjective .
since the dimension of @xmath562 equals the dimension of @xmath46 , it is bijective and therefore a @xmath6-equivariant isomorphism .
note that the @xmath546-action on @xmath554 has to be irreducible in order for the @xmath6-action on @xmath46 to be irreducible .
[ reductiveexceptional ] up to conjugation by an element of @xmath6 , the following tables exhausts the spherical , reductive subalgebras @xmath565 of @xmath524 . according to (
* table 2 ) , the symmetric ones are @xmath566 and @xmath567 and @xmath527 .
let us first consider the case that @xmath568 .
since @xmath533 ( lemma [ q = pperp ] ) and @xmath534 is simple non - compact , @xmath569 has to contain @xmath526 .
hence @xmath570 is the only spherical subalgebra of @xmath305 that lies in @xmath526 .
let us now consider the case that @xmath571 . in this case @xmath572 and @xmath573 .
we are therefore looking for connected subgroups of @xmath520 that act transitively on the spheres in @xmath574 with respect to the adjoint action . due to theorem [ thm2on ] there are none except for @xmath520 itself .
this gives @xmath575 . if @xmath576 then @xmath577 and @xmath578 .
since @xmath533 ( lemma [ q = pperp ] ) and @xmath579 is simple non - compact , @xmath569 has to contain @xmath579 .
hence @xmath580 , where @xmath307 is a subalgebra of @xmath581 .
note that we have @xmath582 the condition for @xmath22 to be spherical in @xmath6 is that @xmath227 acts transitively on the spheres in @xmath583 .
let @xmath584 denote a cartan involution on @xmath305 that commutes with @xmath529 .
then @xmath585 is another involution on @xmath305 defining the lie algebra @xmath586 as fixed point set ( see also the proof of lemma [ q = pperp ] ) . note that @xmath587 has real rank one since @xmath305 has real rank one
. therefore @xmath588 acts transitively on the spheres in @xmath537 .
in particular the adjoint @xmath589-representation on @xmath537 is irreducible .
lemma [ tensor ] implies that @xmath537 decomposes into a tensor product of three @xmath545-vector spaces that are irreducible representations for @xmath590 , @xmath591 and @xmath591 respectively , where @xmath543 . since the adjoint @xmath589-representation on @xmath537 is not only irreducible but transitive on the spheres this tensor product has two factors that are equal to @xmath545 ( since simple tensors can not be mapped to non - simple tensors by this action ) .
using again the fact that @xmath589 acts transitively on the sphere @xmath592 , and the fact that @xmath593 we obtain the following decomposition @xmath594 note that in each case @xmath590 acts by the standard action on @xmath595 .
hence , already @xmath596 acts transitively on the spheres in @xmath357 .
this implies that @xmath597 is arbitrary .
we start by analyzing the structure of a minimal parabolic subgroup @xmath98 of @xmath518 . according to (
* lemma 7.3 ) we have that @xmath598 and that @xmath67 is not abelian .
since the @xmath140-representations on @xmath599 and @xmath600 are irreducible , see ( * ? ? ?
8.13 ) , we obtain @xmath148={{\mathfrak{g}}}_{2\alpha}$ ] where @xmath135 is the restricted root space decomposition .
it is shown in @xcite that @xmath601 and @xmath602 .
[ so(7)subsetso(8 ) ] the @xmath140-representation on @xmath143 is induced by the embedding @xmath603 given by @xmath604 while the @xmath140-representation on @xmath144 is the defining representation of @xmath605 up to the covering @xmath606 .
the embedding from @xmath609 into @xmath610 is described in ( * ? ? ?
* ch.1 5.3 ) as follows .
let @xmath611 then @xmath612 forms a basis of @xmath613 , where @xmath614 is the matrix with a one in line @xmath615 and column @xmath616 and zero everywhere else .
we take the standard orthogonal basis @xmath617 of octonions over @xmath204 with the following multiplication table .
note that the space of pure imaginary octonions @xmath46 is spanned by @xmath618 .
the lie algebra @xmath608 is spanned by elements @xmath619 with @xmath620 .
the map @xmath621 for all @xmath611 is an isomorphism from @xmath613 to @xmath608 .
the map @xmath622 is an embedding of @xmath608 into @xmath623 , where we choose the basis @xmath624 of @xmath625 .
it is explicitly given by the following map as calculation with the multiplication table shows .
@xmath626 note that the induced action of @xmath609 on @xmath625 is transitive on the spheres .
this follows since @xmath627 coincides ( by the multiplication table ) with the purely imaginary octonions , which equals @xmath628 .
therefore the orbit of @xmath609 is open in @xmath629 . since @xmath609 and
hence also its orbits are compact , it coincides with @xmath629 .
in particular , this representation is irreducible . to see this ,
let @xmath631 .
then @xmath632 is a complex simple lie algebra .
we apply the weyl dimension formula ( see e.g. ( * ? ? ?
* ch.v 6 ) ) in order to determine the possible dimensions of irreducible @xmath632-representations .
to do this we use the information on @xmath633 given in ( * ? ? ?
* appendix c 1 ) .
then elementary calculations show that there are unique irreducible representations in complex dimensions @xmath630 and @xmath607 and none in complex dimensions @xmath634 and @xmath635 .
this proves the statement about real representations of @xmath636 .
proposition [ so(7)subsetso(8 ) ] allows us to give a normal form for the subspace @xmath637 . since we already know from remark [ rootspacecontinnh ] that @xmath638 , it is sufficient to give a normal form for subspaces of @xmath143 under the action of @xmath140 .
[ oesubspaceofr8 ] let @xmath609 act irreducibly on @xmath625 and @xmath46 be a @xmath639-dimensional subspace .
then @xmath46 is @xmath609-conjugate to the span of @xmath640 or to the span of @xmath641 , where @xmath642 .
the proof of the last proposition shows that @xmath609 acts transitively on the spheres in @xmath625 .
the stabilizer of @xmath609 in @xmath643 is equal to @xmath547 ( ( * ? ? ?
* ch . 1 5 ) ) . due to onishchik s classification ( ( * ? ? ?
* table 8) ) the group @xmath547 acts transitively on the spheres in @xmath644 and has stabilizer equal to @xmath645 .
the group @xmath645 acts transitively on the sphere @xmath646 but maps complex subspaces to complex subspaces .
+ now let @xmath647 be an orthonormal basis of @xmath46 . if @xmath648 we can map @xmath649 to @xmath650 with @xmath609 , @xmath651 to @xmath652 with @xmath547 and @xmath653 to @xmath654 with @xmath645 .
+ if @xmath320 the orthogonal complement of @xmath46 has dimension less or equal to three . by the previous argument we can map the orthogonal complement of @xmath46 to @xmath655 , thus mapping @xmath46 to the span of @xmath656 .
+ now let @xmath657 . as before we use @xmath609 , @xmath547 and @xmath645 to map @xmath649 to @xmath650 , @xmath651 to @xmath652 and @xmath653 to @xmath654 .
the isotropy @xmath658 of @xmath645 acts on the remaining @xmath659 by @xmath660 , where the @xmath658-action on @xmath661 is the standard one .
if @xmath662 is an element of @xmath663 then @xmath46 is @xmath609-conjugate to the the span of @xmath664 .
if @xmath662 is in @xmath665 we can use the described @xmath658 action to map it to @xmath666 , where @xmath667 and @xmath668 .
the assumption follows since @xmath669 holds .
[ nonredexc ] every non - reductive spherical algebraic subalgebra of @xmath524 is @xmath6-conjugate to one in the following table where @xmath670 and @xmath314 are arbitrary ( under the condition that the maximal compact subalgebra is a lie algebra , see remark [ non - reductive , embedding ] ) .
we assume @xmath671 .
lemma [ oesubspaceofr8 ] shows that @xmath46 may ( after conjugation in @xmath140 ) be chosen as @xmath672 or as the real span of @xmath673 where @xmath151 is the real dimension of @xmath46 and @xmath642 .
the case that @xmath46 is the real span of @xmath641 will be considered at the end of this proof . if @xmath674 we have @xmath675 where @xmath676 is the dimension of @xmath677 .
direct calculation yields @xmath678 moreover , the action of @xmath679 on @xmath680 is given by @xmath681 onishchik s classification theorem [ thm2on ] shows that the subgroup @xmath31 of @xmath679 can only act transitively on the spheres in @xmath682 if the projection of @xmath683 onto @xmath684 is one of the following lie algebras , where @xmath314 is arbitrary .
since @xmath685 holds , our next step is to take each of the lie algebras in the table above and check if it is contained in the projection onto @xmath684 of @xmath686 we start with @xmath374 .
note that @xmath687 in this case .
it is clear that @xmath623 is not contained in @xmath688 , since the dimension of @xmath608 is smaller than the dimension of @xmath623 .
due to ( * ? ? ?
* ch . 1 2.8 and 5.3 ) the representation @xmath689 is injective .
therefore @xmath609 is a subgroup of @xmath690 that acts transitively on the spheres in @xmath324 ( with respect to the action defined by @xmath174 ) .
+ note that @xmath691 is naturally embedded in @xmath692 . and
@xmath692 is embedded in @xmath623 via : @xmath693 calculation shows that the intersection of @xmath694 with @xmath695 does not contain @xmath694 .
hence @xmath590 is not contained in @xmath696 .
therefore the same holds for @xmath692 . + now let @xmath697 . since @xmath698 ( with respect to the action defined by @xmath174 ) it is clear that @xmath547 is contained in @xmath699 .
calculation shows that the projection of @xmath700 onto @xmath613 does not contain @xmath613 .
therefore the projection of @xmath695 onto @xmath613 can not contain @xmath613 .
+ if @xmath701 calculation shows that the projection of @xmath702 onto @xmath703 does not contain @xmath704 , where @xmath705 is the natural embedding of @xmath706 into @xmath703 . since @xmath706 is a subalgebra of @xmath703 the same holds for @xmath703 .
+ if @xmath707 calculation shows that the projection of @xmath708 onto @xmath709 does not contain @xmath709 . + if @xmath710 then the projection of @xmath711 onto @xmath712 is surjective .
furthermore @xmath713 in this case ( since @xmath714 has to hold ) . therefore @xmath715 in this case , where @xmath716 furthermore this shows that the projection of @xmath717 onto @xmath712 also contains @xmath718 .
+ if @xmath719 then the projection of @xmath720 onto @xmath721 is surjective .
+ if @xmath722 then the projection of @xmath723 onto @xmath703 is surjective .
now let us consider the case that @xmath46 is the real span of @xmath726 and @xmath727 .
in this case we denote @xmath728 by @xmath729 .
then @xmath730 equals @xmath731 @xmath732 which is as lie algebra isomorphic to @xmath712 and will be denoted by @xmath733 .
we will denote the induced representation of @xmath733 on @xmath734 by @xmath735 .
we have to show that @xmath736 acts transitively on the spheres in @xmath737 which can be identified with the real span @xmath191 of @xmath738 .
note that @xmath739 lies in the sphere @xmath740 and that @xmath741 hence the orbit of @xmath736 through @xmath742 is open in the sphere in @xmath737 . since @xmath326 is compact its orbit through @xmath742 is equal to the sphere .
note that our calculations also show that the only proper subalgebras of @xmath743 that act transitively on the spheres are @xmath744 and @xmath745 , where @xmath746 is arbitrary .
sigurdur helgason , _ differential geometry , lie groups , and symmetric spaces _ ,
graduate studies in mathematics , vol .
34 , american mathematical society , providence , ri , 2001 , corrected reprint of the 1978 original . | we determine all spherical algebraic subalgebras in any simple lie algebra of real rank @xmath0 . |
we consider a system of @xmath0 distinguishable particles in a volume @xmath1 interacting _ via _ potentials that are translationally invariant , _
e.g. _ , additive pair potentials . in the amorphous solid phase a finite fraction of the particles
are localised at random positions with thermal fluctuations of finite range .
the system will be found in one of many possible equilibrium states .
one such state is singled out arbitrarily , denoted by @xmath2 and specified by nonzero local static density fluctuations @xmath3 . here , @xmath4 is the position of particle @xmath5 and @xmath6 denotes a thermal average restricted to equilibrium state @xmath2 .
the amorphous state has macroscopic translational invariance , so that @xmath7 = 0,\ ] ] where @xmath8 $ ] denotes a disorder average .
localisation in the equilibrium state @xmath2 is detected by higher - order moments , the simplest one being the second : @xmath9 = \delta_{{\bf k}_1+{\bf k}_2,{\bf 0}}\;\omega({\bf k}_1).\ ] ] for this to be nonzero , macroscopic translational invariance requires @xmath10 , and @xmath11 is a real - valued function .
the second moment vanishes in the isotropic fluid phase and diagnoses the localisation of particles .
hence the second moment ( and in fact all higher moments , as will be discussed below ) serves as an order parameter for the transition from the fluid to the amorphous solid state . in a different equilibrium state , denoted by @xmath12 ,
all particles are translated with respect to the equilibrium state @xmath2 by a displacement @xmath13 , so that @xmath14 due to the translational invariance of the hamiltonian there is in fact a manifold of symmetry - related states , generated by spatially uniform translations .
how do we have to generalise the representation of the order parameter ( [ naive_orderpar ] ) , in order to take into account the underlying translational symmetry of the hamiltonian ? for two equilibrium states , @xmath12 and @xmath15 , the second moment is given by @xmath16 = { \rm e}^{i{\bf k}_1\cdot({\bf a}^{\mu}-{\bf a}^{\rho}})\ , \delta_{{\bf k}_1+{\bf k}_2,{\bf 0}}\;\omega({\bf k}_1)\,.\ ] ] we introduce a longitudinal wavevector @xmath17 and a transverse wavevector @xmath18 and rewrite the above expression @xmath19 in terms of a transverse displacement @xmath20 .
thereby , the second moment is represented by its magnitude @xmath21 and the generator of transverse translations @xmath22 .
the interpretation of the order parameter is more intuitive for the second moment as a function of real - space variables : @xmath23 .
it does not depend on @xmath24 , and hence can be visualised as a rectilinear _ hill _ , whose position is specified by a pair of pure states . in a simple model of particle localisation in @xmath25 dimensions , a fraction @xmath26 of the particles are harmonically bound to random positions with random mean square displacements @xmath27 , giving rise to @xmath28 here @xmath29 $ ] is the disorder - averaged distribution of squared localisation lengths of the @xmath30 localised particles .
the width of the hill is thus given by the typical value @xmath31 of the localisation length .
the structure in one space dimension is sketched in fig .
[ fig : hill ] .
that varies with position @xmath32 along the ridge .
note that the scale of this figure is much larger than that used for fig .
[ fig : hill ] . ] that varies with position @xmath32 along the ridge .
note that the scale of this figure is much larger than that used for fig .
[ fig : hill ] . ]
this pattern of symmetry breaking suggests that the goldstone excitations of an equilibrium state are constructed from it _ via _
@xmath33- dependent translations of the hill in the @xmath34 direction , _
i.e. _ , _ ripples _ of the hill and its ridge ( see fig . [
fig : ripples ] ) . in fourier space , the generalisation of eq .
( [ generalop ] ) to inhomogeneous displacements reads @xmath35 the corresponding representation in real space is completely equivalent : @xmath36
in the following , we apply the general concepts of goldstone excitations to a particular amorphous solid state gels and vulcanised matter . for these systems
a semi - microscopic model provides a framework for analysing the transition from the viscous fluid to the gel , as well as the structural and thermal properties of the gel , quantitatively .
here we use it to compute the free energy of goldstone fluctuations , identify it with the elastic free energy , and thereby extract the shear modulus . the transition from the viscous fluid to the gel
is controlled by the number of crosslinks , connecting randomly chosen pairs of monomers .
these monomers can be part of a larger prefabricated structure , such as chains consisting of @xmath37 monomers or stars or dimers _ etc_. for the following , the internal structure of the molecules is almost irrelevant ; it will only determine certain parameter values .
the average over the quenched disorder ( _ i.e. _ the number and location of the crosslinks ) is performed with help of the replica trick .
the resulting , effectively uniform , theory is formulated in terms of a field @xmath38 defined over @xmath39-fold replicated space @xmath40 , so that the argument @xmath40 means the collection of @xmath39 position @xmath25-vectors @xmath41 . according to the replica scheme of deam and edwards @xcite , the disorder - averaged free energy @xmath42 is given by @xmath43 .
where @xmath44 is the temperature .
the replica partition function @xmath45 is given as a functional integral @xmath46 over the field @xmath47 , where @xmath48 denotes wavevectors conjugate to @xmath40 .
the expectation value of the field is given by @xmath49 for this to be nonzero , macroscopic translational invariance requires @xmath50 .
the case with only one of the @xmath51 wavevectors nonzero corresponds to the ( particle number ) density ( [ density ] ) .
density fluctuations are suppressed by excluded volume interactions , which tend to maintain homogeneity .
therefore the theory , in its simplest version , contains only fluctuating fields for which at least two of the @xmath25-component entries in the argument @xmath52 are nonzero .
we refer to such @xmath53 dimensional vectors as the higher - order replica sector ( hrs ) .
fields with @xmath54 non - vanishing @xmath25-dimensional wavevectors correspond to the @xmath54th moment of the local density , so in particular @xmath55 yields the second moment discussed above .
we consider a finite - dimensional system with a small number of nearest neighbours , implying that fluctuations of the local density are in general non - gaussian and all moments need to be taken into account .
the _ translational _ invariance of the system is reflected in the invariance of the effective hamiltonian @xmath56 under _ independent translations of the replicas_. in the amorphous solid state this symmetry is broken down to invariance under the subgroup of common translations of all the replicas .
the manifold of symmetry - related classical solutions ( _ i.e. _ saddle points of @xmath56 ) is given by @xmath57 in analogy to eq .
( [ generalop ] ) .
we have decomposed wavevectors into longitudinal and transverse components with help of a complete orthonormal basis set in replica space @xmath58 .
the `` longitudinal '' direction in @xmath51 dimensional replica space is the unit vector @xmath59 , which is constant in replica space .
all @xmath53 dimensional vectors can be expanded in this basis : _
e.g. _ , @xmath60 and decomposed into longitudinal ( @xmath61 ) and transverse ( @xmath62 ) components : @xmath63 analogous expressions hold for @xmath40 and @xmath64 .
invariance under common translations of all replicas is guaranteed in the classical solution ( [ eq : classicalstate ] ) by the kronecker delta , which enforces the longitudinal wavevector @xmath65 to be zero .
the manifold of symmetry - related states is generated by purely transverse displacements .
the order parameter ( [ eq : classicalstate ] ) exhibits the structure in replicated @xmath40-space discussed above for the second moment of the local density in different equilibrium states .
the classical state @xmath66 can be visualised as a rectilinear hill in @xmath40-space with contours of constant height oriented parallel to @xmath67 .
symmetry - related classical states are obtained by translating the hill rigidly , perpendicular to the ridge - line .
goldstone excitations from the classical state are constructed _ via _ @xmath68-dependent translations of the hill in the @xmath69-direction @xcite , _ i.e. _ , ripples of the hill and its ridge : @xmath70 we are interested here in the low energy excitations only and , hence , consider goldstone fluctuations , whose energy is expected to vanish with a power of the inverse wavelength of the ripples .
we neglect all other fluctuations , which are expected to be massive . for consistency
we thus should restrict ourselves to goldstone fluctuations with wavelengths that are larger than a short distance cutoff @xmath71 .
in particular we require that the displacement @xmath72 varies on length scales larger than the width of the interface and take @xmath73 .
the physical content of this restriction is that , to be meaningful , elastic deformations should occur on length - scales longer than the typical localisation length of the particles in the elastic medium . as discussed after eq .
( [ eq : pdinterpret ] ) , the critical fields of the theory reside in the hrs .
hence we want to make sure that the distortions of the order parameter remain in this sector .
this is guaranteed if the distortions fulfill the incompressibility constraint : @xmath74 , which , for small amplitude distortions , reduces to @xmath75 .
for the present purposes the essential term in the effective hamiltonian @xmath56 @xcite is the quadratic gradient term , @xmath76 where @xmath77 and @xmath78 are , respectively , the number - density and linear size of the underlying entities being constrained and @xmath79 indicates that only wave vectors in the higher - order replica sector are to be included in the summation .
although eq .
( [ eq : landaueffective ] ) was originally obtained _ via _ an expansion valid close to the critical point , such a quadratic gradient term is expected to feature in a description valid deep inside the solid state . if one is seeking as we are the dominant contribution to the elastic free energy governing the displacement field @xmath80 ( _ i.e. _ the leading - order term in @xmath80 and @xmath65 ) then it is adequate to retain solely this quadratic gradient term . inserting the fluctuating field into @xmath56 and computing the _ increase _ , @xmath81 , due to the distortion
@xmath80 ( _ i.e. _ the elastic free energy ) gives the following contribution , which arises solely from the quadratic gradient term in @xmath56 : [ eq : elasticcouple ] @xmath82 here , summation over repeated cartesian indices @xmath83 is implied and the displacement vectors @xmath84 have to fulfill the constraint @xmath85 .
it was shown in refs .
@xcite that in the classical state ( [ eq : classicalstate ] ) @xmath86 with the fraction of localised particles @xmath87 and the distribution of localisation lengths @xmath88 , expressed in terms of a universal classical scaling function @xmath89 . here , @xmath90 and @xmath91 are , respectively , the control parameter for the density of constraints and the nonlinear coupling constant . a semi - microscopic model of vulcanised macromolecular matter yields @xmath92 , @xmath93 , @xmath94 and @xmath95 , where @xmath12 controls the mean number of constraints ( and has critical value @xmath96 ) , @xmath97 is the persistence length of the macromolecules , and @xmath37 is the number of segments per macromolecule . if the network is built up from single monomers , corresponding to @xmath98 , then @xmath97 is interpreted as the length of the tether connecting two point - like particles .
we substitute this classical value of the order parameter into eq .
( [ eq : replicastiffness ] ) and take the replica limit @xmath99 to obtain the shear modulus @xmath100 the shear modulus increases continuously from zero with the third power of the distance from the gel point , in agreement with the previous microscopic calculation of ref . @xcite .
a simple scaling argument is also consistent : the free energy density scales with the correlation length @xmath101 as @xmath102 .
this is to be compared with the elastic free energy density of eq .
( [ eq : elasticenergy ] ) .
since @xmath103 is dimensionless this implies @xmath104 . at @xmath105
we find agreement , which is to be expected since our calculation is based on a gaussian expansion around the classical solution .
we discuss the effects of goldstone fluctuations at the level of gaussian fluctuations ( [ eq : elasticenergy ] ) only .
the elastic green function is given by @xmath106 where @xmath107 . here
, the integration over wave - numbers is restricted to @xmath108 , where @xmath109 is the linear size of the system .
the mean value of the distorted classical state , @xmath110 is determined by the local green function @xmath111 . in space dimension
@xmath112 , @xmath113 is finite , so that the order parameter has a finite expectation value .
the latter is reduced , due to goldstone fluctuations , as one would expect : in addition to independent fluctuations of the localised particles there are also collective fluctuations transverse phonons which effectively increase the extent of spatial fluctuations of the localised particles .
this is best seen by recalling the classical structure for @xmath114 , eq .
( [ eq : mfstate ] ) : @xmath115 order parameter correlations are given within the gaussian approximation by @xmath116 in @xmath117 , the asymptotics of the elastic green function for large separation @xmath118 is well known , @xmath119 , implying for the correlator of the order parameter @xmath120 where @xmath121 . in two space dimensions , @xmath113 increases logarithmically with the system size , so that the expectation value of the order parameter ( [ eq : orderparameter ] ) vanishes in the macroscopic limit .
goldstone fluctuations destroy localisation in two space dimensions , in agreement with the hohenberg - mermin - wagner theorem .
nevertheless , a two - dimensional amorphous solid persists , which is characterised by algebraic decay of order parameter correlations , _
i.e. _ , quasi - long - range ( random ) order .
the asymptotics of the elastic green function for large spatial separation is given by @xmath122 , implying an algebraic decay of the order - parameter correlations with a non - universal exponent @xmath123 : @xmath124 this scenario is similar to the theory of two - dimensional regular solids , taking into account harmonic phonons only , as discussed in ref .
the main difference is the absence of bragg peaks at reciprocal lattice vectors .
instead , there is a divergence in the scattering function at @xmath125 , as can be seen by integrating the correlator , above , over all @xmath126@xcite : @xmath127 .
certain ideas concerning the nature of the vulcanization transition in two spatial dimensions , developed in detail in the present letter , originated in enlightening discussions involving horacio e. castillo and weiqun peng ; see refs . @xcite and @xcite .
it is a pleasure to acknowledge these discussions , as well as enlightening discussions with matthew p. a. fisher and dominique toublan .
we thank for their hospitality the department of physics at the university of colorado boulder ( pmg ) and the kavli institute for theoretical physics at the university of california
santa barbara ( pmg and az ) , where some of the work reported here was undertaken .
this work was supported in part by the national science foundation under grants nsf dmr02 - 05858 ( sm , pmg ) , nsf ph99 - 07949 ( pmg , az ) , and by the dfg through sfb 602 and grant no .
zi 209/6 - 1 ( az ) . | goldstone modes in the amorphous solid state , resulting from the spontaneous breaking of translational symmetry due to random localisation of particles , are discussed . starting from a microscopic model with quenched disorder ,
the broken symmetry is identified to be that of relative translations of the replicas .
goldstone excitations , corresponding to pure shear deformations , are constructed from long wavelength distortions of the order parameter .
the elastic free energy is computed , and it is shown that goldstone fluctuations destroy localisation in two spatial dimensions , yielding a two - dimensional amorphous solid state characterised by power - law correlations .
the _ elastic _ properties of gels and vulcanised matter have received considerable attention in the literature .
phenomenological arguments@xcite as well as microscopic models@xcite have been used to compute , _
e.g. _ , the shear modulus .
the _ structural _ properties of gels have been discussed more recently@xcite , building on the work of edwards and collaborators @xcite .
it has been shown that in the gel phase a finite fraction of particles is localised around random positions with finite thermal excursions .
what has not been analysed so far , and what constitutes the main focus of this letter , is the connection between the spontaneously broken translational symmetry due to particle localisation and shear deformations , which are identified with goldstone excitations as a consequence of the spontaneous breaking of translational symmetry .
we first discuss symmetry - related equilibrium states in the general context of amorphous solids .
we construct low - energy long - wavelength goldstone fluctuations , which correspond to almost - uniform translations of the localised particles .
particular care is taken of the residual symmetry of the amorphous solid state , which remains statistically homogeneous .
we then apply these general ideas to a particular type of amorphous solid gels and vulcanised matter for which a replica field theory is available@xcite .
this allows us to identify the lost symmetry to be that of relative translations of the replicas , whilst common translations of all replicas remain as intact symmetries in the amorphous solid state .
goldstone fluctuations are described as long - wavelength distortions of the order parameter , which are reminiscent of the ripples of an interface between coexisting liquid and gas phases .
the free energy of these fluctuations is computed within a gaussian approximation , and identified with the elastic free energy of a macroscopically homogeneous isotropic amorphous solid . |
the formation , stability and dynamics of vortex - like structures has been a long - standing theme of interest in many areas of physics , including classical fluid mechanics @xcite , superfluidity and superconductivity @xcite , and cosmology @xcite .
moreover , in the past decade , there has been a tremendous growth of excitement in this topic in the branches of atomic and optical physics .
this has been propelled by considerable experimental and theoretical advances in the fields of nonlinear optics @xcite and bose - einstein condensates ( becs ) in dilute alkali vapors @xcite ( see also ref .
@xcite ) .
focusing more specifically on the rapidly growing area of becs @xcite , one can recognize that the study of vortices has been central to the relevant literature .
in particular , as concerns the experimental efforts , the original observations of single @xcite and multiple vortices @xcite was soon followed by the realization of robust lattices of large numbers of vortices @xcite .
subsequent studies turned to higher - charged structures such as vortices of topological charge @xmath0 and even @xmath1 @xcite and illustrating dynamical instability of these topological objects @xcite . on the other hand ,
the abundance of experimental results has stirred an intense theoretical interest in the conditions under which such vortices and vortex lattices would be robust and experimentally observable .
most often , vortex existence and stability issues were examined in the framework of the standard parabolic confining potential ( typically produced by magnetic traps ) . in that framework , and in the two - dimensional ( 2d ) case , vortices of charge @xmath2 were found to be stable , while vortices of higher charge ( @xmath3 ) were shown to be potentially unstable @xcite ( depending on the atom numbers ) .
later , similar results were found for vortices of @xmath1 @xcite , while the availability of more substantial computational resources has more recently led to similar conclusions in the fully three - dimensional ( 3d ) case @xcite .
the studies of refs .
@xcite and @xcite examined the various scenarios of break - up of higher - charge vortices during dynamical evolution simulations for repulsive and attractive interactions respectively .
furthermore , ref .
@xcite considered such vortices riding on the background of not just the ground state , but also of higher , ring - like , excited states of the system .
it should also be mentioned that these advancements have motivated the development of mathematically rigorous tools in order to study the spectrum of such vortex modes .
such methods involve the use of the evans function @xcite , or the use of the index theorem evaluating the number of potentially unstable eigendirections @xcite .
although the existence and stability of fundamental and higher charge vortices has been examined extensively as indicated above , the _ formation _ of such vortex structures is far less studied . in particular , while seminal interference experiments ( demonstrating that becs are coherent matter waves ) were reported as early as a decade ago @xcite , the role of interference between becs in vortex generation was experimentally studied only recently @xcite .
this work proposed and examined the formation of vortices resulting from the interference and controlled merging of three condensed fragments , where the fragments were essentially independent becs separated by an optical potential barrier .
such a process has close ties to elements of topological defect formation in phase transitions , as proposed by kibble @xcite and zurek @xcite .
interestingly , the experimental work was almost concurrent with a theoretical study investigating a simpler elongated barrier separating two independent becs @xcite ; in the latter setting , the interference forms a dark soliton whose bending and subsequent breakup due to the manifestation of the transverse modulational instability also result in vortices . in the present work ,
we expand on these considerations and study in detail , by means of systematic numerical simulations , the formation of vortices in a setting closely matching the one of the experiment in ref .
our purpose is to get a deeper insight into this interference - induced vortex formation mechanism , investigating fundamental features , such as the number and lifetime of ensuing vortices .
this is done upon studying in detail the parametric dependences influencing the relevant experimental observations .
specifically , we quantify the above mentioned features as functions of the elimination / ramping - down time of the optical barrier between the fragments , or the initial relative phases between the original independent fragments .
our investigation chiefly refers to a 2d setting , but we also illustrate how the results are generalized in the pertinent 3d case . notice that our considerations are motivated not only by their direct bearing on the experiments of ref .
@xcite , but also by their relevance to studies of spontaneous symmetry breaking during phase transitions @xcite .
our presentation is structured as follows . in sec .
[ sec : setup ] , we briefly summarize the setup of our computational experiments . in sec .
[ sec : num ] , we study the interference of three bec fragments in the 2d and 3d setup ; special attention is payed at the role of the relative phases between the different condensates , and the ramp - down time of the laser sheet barrier responsible for separating the three fragments . in sec .
[ sec : expt ] , we briefly comment on the relation between numerical and experimental results . finally , in sec .
[ sec : conclusions ] , we summarize our findings , present our conclusions , and discuss some possible extensions of this work .
we consider a bec at a temperature close to zero , where quantum or thermal fluctuations are negligible ( note that finite temperature effects are briefly discussed at the end of sec .
[ sec : num ] ) .
this system can accurately be described by a mean - field theoretical model , namely the gross - pitaevskii equation ( gpe ) @xcite : @xmath4 \psi , \label{gpe}\end{aligned}\ ] ] where @xmath5 is the condensate wavefunction ( with @xmath6 being the atomic density of the condensate ) , @xmath7 is the atomic mass , the coupling constant @xmath8 measures the strength of inter - atomic interactions and @xmath9 is the @xmath10-wave scattering length .
the potential @xmath11 in the gpe is taken to be of the form , @xmath12 where the two components in the right - hand side of eq .
( [ poteq ] ) are a harmonic magnetic trap , @xmath13 , with trapping frequencies @xmath14 hz and @xmath15 hz , and the three - armed time - dependent optical barrier , @xmath16 , used in the experiments of ref . @xcite .
this three - armed potential induces a separation of the ground state of the condensate into three different fragments ( see top - left and top - center panels in fig . [ pot ] and fig .
[ fig_expt](a ) ) .
note that the function @xmath17 in eq .
( [ poteq ] ) describes the ramping down of the optical barrier .
the maximum initial barrier energy for the potential is taken to be @xmath18 nk @xcite , where @xmath19 is boltzmann s constant . in the numerical simulations , the ground state of the system
is obtained by relaxation ( imaginary time integration ) initiated with the thomas - fermi ( tf ) approximation ( see top - center panel in fig . [ pot ] ) @xmath20 where @xmath21 nk @xcite is the chemical potential , and @xmath22 contains the chosen phase of the different wells separated into the three regions a , b and c depicted in the top - right panel of fig .
let us denote by @xmath23 , @xmath24 , and @xmath25 the initial phases in regions a , b and c , respectively .
it is important to note that while in the experiments of ref .
@xcite the different fragments have uncorrelated ( random ) phases , in our numerical simulations we are able to control their initial phases and , more importantly , their relative phases . our numerical experiments , emulating the experimental sequence of ref .
@xcite , are performed with the output of imaginary time relaxation used as initial condition for the full dynamics of eq .
( [ gpe ] ) .
the bottom row of fig .
[ pot ] depicts the evolution of the phase for the case @xmath26 during relaxation .
as can be observed from the figure , the initial `` seed '' ( @xmath27 ) starts with sharp phase boundaries around the localized fragments .
as the relaxation procedure evolves , the phase boundaries become smoother and adjust to the boundaries for the top - right panel of the figure .
the phase profile seems to settle after 10 ms of imaginary time relaxation . in our simulations , 50 ms of imaginary time relaxation was used to ensure proper convergence to the steady - state solution . , @xmath24 , and @xmath25 .
the bottom row of panels depicts the evolution of the phase during our relaxation ( imaginary time relaxation ) towards the initial steady state with different phases for the fragments .
this example shows the phase for the case @xmath26 at the times indicated . in all panels ,
the field of view is approximately 70 @xmath28 m per side .
the axis numbers indicate @xmath29 and @xmath30 coordinates relative to the center of the unsegmented harmonic trap.,title="fig:",width=321 ] + , @xmath24 , and @xmath25 .
the bottom row of panels depicts the evolution of the phase during our relaxation ( imaginary time relaxation ) towards the initial steady state with different phases for the fragments .
this example shows the phase for the case @xmath26 at the times indicated . in all panels ,
the field of view is approximately 70 @xmath28 m per side .
the axis numbers indicate @xmath29 and @xmath30 coordinates relative to the center of the unsegmented harmonic trap.,title="fig:",width=321 ] -0.4 cm ms ( first and second row ) , @xmath31 ms ( third and fourth row ) , and @xmath32 ms ( fifth and sixth row ) .
the times are indicated in the panels in ms and the field of view is approximately 70 @xmath28 m per side . for the initial conditions in all cases , the different condensed fragments have relative phases of @xmath33 , namely , @xmath34 ( @xmath35 ) .
, title="fig:",width=321 ] ms ( first and second row ) , @xmath31 ms ( third and fourth row ) , and @xmath32 ms ( fifth and sixth row ) .
the times are indicated in the panels in ms and the field of view is approximately 70 @xmath28 m per side . for the initial conditions in all cases , the different condensed fragments have relative phases of @xmath33 , namely , @xmath34 ( @xmath35 ) .
, title="fig:",width=321 ] ms ( first and second row ) , @xmath31 ms ( third and fourth row ) , and @xmath32 ms ( fifth and sixth row ) .
the times are indicated in the panels in ms and the field of view is approximately 70 @xmath28 m per side . for the initial conditions in all cases , the different condensed fragments have relative phases of @xmath33 , namely , @xmath34 ( @xmath35 ) .
, title="fig:",width=321 ] -0.4 cm
for the 2d rendering of the experiment of ref .
@xcite , we restrict our system to the @xmath36 coordinates and we use the _ same _ chemical potential as in ref . @xcite .
first we explore the effect of the ramp - down time of the potential barrier on the formation of vortices through the merging of the different fragments of the condensate . for this purpose
we use a linear ramp : @xmath37 where @xmath38 is the maximum barrier energy as defined above and @xmath39 is the ramping time ( in ms ) of the barrier ; note that a similar ramp was used in the experiments of ref .
@xcite .
we monitor the formation of vortices , as well as the overall vorticity of the system , using various diagnostics .
these are based on the corresponding fluid velocity of the superfluid given by @xcite , @xmath40 where @xmath41 stands for complex conjugation .
the fluid vorticity is then defined as @xmath42 .
the results for the merger of the three bec fragments with relative phases @xmath26 , @xmath35 , for different merging times are depicted in fig .
[ bpa2d_allu ] .
as can be seen from the figure , the number of vortex pairs nucleated by the merger is extremely sensitive to the ramping time @xmath39 .
shorter ramping times give rise to an extremely rich vorticity pattern as the fragments merge [ see , for example , the top two rows in fig .
[ bpa2d_allu ] corresponding to the instantaneous ( @xmath43 ) removal of the barrier ] , including the appearance of structures resembling vortex streets .
however , for longer ramp times , i.e. slower ramping , just a handful of vortices are nucleated .
in fact , for @xmath44 ms ( results not shown here ) , the only vortex that is nucleated is the central one .
it is evident that , independently of the ramping time , the central vortex is always formed for this pair of relative phases between the fragments . this vortex is the consequence of the intrinsic vorticity present in the initial condition where the three fragments have been phase imprinted with a total of a @xmath45 phase gain about the condensate center .
ms and @xmath26 ( cf .
middle rows in fig .
[ bpa2d_allu ] ) .
, width=321 ] -0.4 cm in fig .
[ filaments_fig ] we present a spatio - temporal rendering of the vortex formation for the middle row example of fig . [ bpa2d_allu ] ( i.e. , phases given by @xmath46 and a ramp down time of @xmath31 ms ) . in the figure we depict a space - time contour plot of the vorticity where blue / red contours correspond to negatively / positively charged vortices , respectively .
the figure clearly shows the formation of pairs of vortices with opposite charge , some of which self - annihilate at later times , while others oscillate together with the cloud : upon formation , they expand to the rims of the cloud and are subsequently reflected from its outskirts and contract anew , following the density profile oscillations resulting from the merging process .
( namely , @xmath43 ms , @xmath31 ms , and @xmath32 ms , from top to bottom ) with @xmath26 .
the left panels correspond to the total angular momentum [ eq .
( [ lz_eq ] ) ] normalized by @xmath47 , while the middle and right panels correspond to the total fluid velocity [ eq .
( [ s_eq ] ) ] for the whole cloud ( middle ) and the central portion ( see text ) of the cloud ( right ) . , title="fig:",width=321,height=113 ] ( namely , @xmath43 ms , @xmath31 ms , and @xmath32 ms , from top to bottom ) with @xmath26 .
the left panels correspond to the total angular momentum [ eq .
( [ lz_eq ] ) ] normalized by @xmath47 , while the middle and right panels correspond to the total fluid velocity [ eq .
( [ s_eq ] ) ] for the whole cloud ( middle ) and the central portion ( see text ) of the cloud ( right ) .
, title="fig:",width=321,height=113 ] ( namely , @xmath43 ms , @xmath31 ms , and @xmath32 ms , from top to bottom ) with @xmath26 .
the left panels correspond to the total angular momentum [ eq .
( [ lz_eq ] ) ] normalized by @xmath47 , while the middle and right panels correspond to the total fluid velocity [ eq .
( [ s_eq ] ) ] for the whole cloud ( middle ) and the central portion ( see text ) of the cloud ( right ) .
, title="fig:",width=321,height=113 ] -0.4 cm for three different relative phases and a ramp down of @xmath31 ms . the top two , middle two and bottom two series correspond , respectively , to : a ) @xmath48 , b ) @xmath49 , @xmath50 , and @xmath51 , and c ) @xmath49 , @xmath52 , and @xmath53 . , title="fig:",width=321 ] for three different relative phases and a ramp down of @xmath31 ms . the top two , middle two and bottom two series correspond , respectively , to : a ) @xmath48 , b ) @xmath49 , @xmath50 , and @xmath51 , and c ) @xmath49 , @xmath52 , and @xmath53 . , title="fig:",width=321 ] for three different relative phases and a ramp down of @xmath31 ms . the top two , middle two and bottom two series correspond , respectively , to : a )
@xmath48 , b ) @xmath49 , @xmath50 , and @xmath51 , and c ) @xmath49 , @xmath52 , and @xmath53 . , title="fig:",width=321 ] -0.4 cm in order to measure the vorticity generated during the merger of the condensate fragments at any given time , we compute the expectation value of the @xmath54-component of the angular momentum of the bec @xmath55 where @xmath56 and @xmath57 denote the corresponding position and momentum operators , the subscript @xmath54 indicates the component of the corresponding cross - product and @xmath58 indicates the complex inner product defined from @xmath59 . using eq .
( [ lz_eq ] ) and eq .
( [ gpe ] ) , we can evaluate the time derivative of this expectation value : @xmath60 in obtaining this result , we have assumed the isotropy of @xmath61 in the @xmath36 plane .
this result also has some important consequences including the fact that during the ramp down of the optical barrier , we should not expect the angular momentum of the bec to be conserved , while we should expect such a conservation to appear once @xmath62 or , more generally , when the full potential is azimuthally isotropic in the @xmath36 plane ( cf .
[ bpa2d_lz ] and [ bpa2d_lz_phases ] ) . in the left column of fig .
[ bpa2d_lz ] we depict the @xmath54-component of the angular momentum .
the middle column of the same figure shows a quantity that we refer to as total fluid velocity , defined as : @xmath63 where @xmath64 is the total volume of integration . finally , the right column of fig .
[ bpa2d_lz ] shows the same quantity , as defined in eq .
( [ s_eq ] ) , but only for the central portion of the cloud .
the central portion of the cloud was defined as a square , centered at the center of the magnetic trap , with a side equal to 10% of the integration domain .
this area corresponds approximately to the void area between the three initial fragments ( see top - center panel in fig . [ pot ] ) .
the three diagnostics defined above are shown in fig .
[ bpa2d_lz ] for the same cases presented in fig .
[ bpa2d_allu ] .
it is interesting to note how the presence of a time - dependent component in the potential generates angular momentum in the system .
as can be observed in the figure , the total angular momentum increases ( in absolute value ) through the duration of the barrier ramp - down and then settles to a constant value that is larger ( in absolute value ) for slower ramps . also , it is worth noting that the initial value of the angular momentum is different from zero [ @xmath65 due to the intrinsic vorticity carried by the out - of - phase fragments ( see also discussion below ) . for our initial condition ,
most of the total angular momentum is seeded in the weak overlapping region between the fragments where the phase gradient is large .
what we can observe about the second and third diagnostics by comparing ( the second row of ) fig .
[ bpa2d_allu ] and fig .
[ filaments_fig ] with ( the second row of ) fig .
[ bpa2d_lz ] is roughly the following : the integrated ( throughout the cloud ) velocity appears to peak when the filamentation in the pattern of fig .
[ filaments_fig ] is maximal , e.g. , around times of 50 and 90 ms . on the other hand ,
the same diagnostic integrated within the central core of the cloud peaks substantially earlier when the vortices are formed through the collision of the fragments around the end time of the ramp ( i.e. , around 25 ms ) .
subsequently , and as the vortices are advected away from the core , the latter quantity decreases .
( the different panels are presented in the same manner as in fig .
[ bpa2d_lz ] ) . ,
title="fig:",width=321,height=113 ] + -0.2 cm ( the different panels are presented in the same manner as in fig .
[ bpa2d_lz ] ) .
, title="fig:",width=326,height=113 ] + ( the different panels are presented in the same manner as in fig . [ bpa2d_lz ] ) . , title="fig:",width=321,height=113 ] + -0.4 cm is assigned the value of 0 ) .
top : ramp down time of @xmath43 ms ( i.e. , instantaneous removal ) of the barrier .
the different shades of gray indicate the amount of vorticity contained in the central region after 100 ms .
white , light gray , gray , black correspond , respectively , to 0 , 1 , 2 , and 3 vortices .
middle : same as above for a ramp down time of @xmath32 ms .
this case only produces no vortices ( white ) or one vortex ( gray ) .
bottom : difference between top and middle diagrams .
this corresponds to the vortices formed by the collision of the fragments and not by the intrinsic vorticity of the initial configuration ( which depends on the relative phases of the different fragments ) .
, title="fig:",width=226 ] + is assigned the value of 0 ) .
top : ramp down time of @xmath43 ms ( i.e. , instantaneous removal ) of the barrier .
the different shades of gray indicate the amount of vorticity contained in the central region after 100 ms .
white , light gray , gray , black correspond , respectively , to 0 , 1 , 2 , and 3 vortices .
middle : same as above for a ramp down time of @xmath32 ms .
this case only produces no vortices ( white ) or one vortex ( gray ) .
bottom : difference between top and middle diagrams .
this corresponds to the vortices formed by the collision of the fragments and not by the intrinsic vorticity of the initial configuration ( which depends on the relative phases of the different fragments ) .
, title="fig:",width=226 ] + is assigned the value of 0 ) .
top : ramp down time of @xmath43 ms ( i.e. , instantaneous removal ) of the barrier .
the different shades of gray indicate the amount of vorticity contained in the central region after 100 ms .
white , light gray , gray , black correspond , respectively , to 0 , 1 , 2 , and 3 vortices .
middle : same as above for a ramp down time of @xmath32 ms .
this case only produces no vortices ( white ) or one vortex ( gray ) .
bottom : difference between top and middle diagrams .
this corresponds to the vortices formed by the collision of the fragments and not by the intrinsic vorticity of the initial configuration ( which depends on the relative phases of the different fragments ) .
, title="fig:",width=226 ] -0.4 cm + + -0.4 cm we now turn to the examination of the effect of the relative initial phases between the different fragments . as an example
, we show in fig .
[ bpa2d_allu_phases ] the behavior of the cloud density and its vorticity for three different phase combinations for a fixed ramp - down time of @xmath31 ms .
as can be noticed , the complexity of the vorticity field is similar for the three cases shown in the figure .
however , the only phase combination of the ones shown here that produces a vortex at the center of the cloud corresponds to @xmath26 , which can be understood by the intrinsic vorticity already present in the initial configuration ( see explanation below ) . in fig .
[ bpa2d_lz_phases ] we depict the vorticity indicators for the three phase combinations of fig . [ bpa2d_allu_phases ] .
as can be seen in the figure , the angular momentum @xmath66 for @xmath48 ( top - left panel ) and @xmath67 ( bottom - left panel ) suffers almost no change since the initial configuration does not carry any intrinsic vorticity and thus , its interaction with the ramping down barrier does not produce angular momentum .
however , for @xmath26 ( middle - left panel ) , as explained before , the system does gain angular momentum during the barrier removal .
the total fluid velocity indicators ( middle and right columns in fig . [ bpa2d_lz_phases ] behave similarly for the three phase combination with the notable difference that for the @xmath26 ( middle - left panel ) case , its final value is different from zero since the fragments produce a vortex at the center of the trap ( due to the intrinsic vorticity carried by the initial condition ) . in order to follow in more detail the formation of vortices in the central portion of the cloud as a function of the relative initial phases of the different fragments , we perform systematic simulations for a large set of relative initial phases .
the results are presented in fig .
[ diagram ] , where we show the existence of vortices as a function of @xmath68 for @xmath49 .
darker shades correspond to the presence of more vortices .
the top panel of fig .
[ diagram ] corresponds to an immediate release of the barrier ( ramp down time of @xmath43 ms ) where the presence of 0 , 1 , 2 or 3 vortices ( white , light gray , gray , and black respectively ) can be observed for different phase combinations .
the middle panel depicts the same diagram , but for a ramping - down time of @xmath32 ms .
it is clear that , for this relatively slow ramp down , the formation of a vortex in the central region is exclusively determined by the vorticity of the initial configuration .
namely , if any of the relative phases is larger than @xmath69 , the initial configuration resembles more that of a _ discrete vortex _ ( see refs . @xcite and @xcite for reviews ) .
in particular , each fragment can be thought of a `` unit '' and the whole configuration corresponds to a discrete vortex with three units with a net vorticity different from zero .
the middle panel of the figure clearly reveals that _ only _ within an arc of the second ( and essentially symmetrically of the fourth ) quadrant of the plane of the phases @xmath70 , a single vortex will form at the center of the configuration .
it is interesting that discrete vortex - like configurations consisting of three fragments have been considered in the context of nonlinear optics in ref .
@xcite and even in genuinely discrete systems as , e.g. , in ref . @xcite . in the bottom panel of fig . [ diagram ]
we depict the difference between the top and middle panels to show the amount of vortices that are created exclusively by the fragment collision and not by the intrinsic vorticity of the initial configuration .
it is important to mention that some of the vortices counted in the top panel come from vortices that are created _ outside _ the central region but that migrate towards the center as time progresses .
+ + + -0.4 cm + + -0.4 cm as shown in the previous section , the 2d setting lends itself to a more detailed examination of important features such as the parametric dependence on the ramping time and the relative phases of the fragments . nevertheless , it is important to also consider some of the delicate points particular to the 3d nature of the experiments and the observable quantities available within the experimental images . for this reason ,
we now focus on the 3d setting , presenting results of the simulations relevant to the experiment of ref .
the setup is the same as in the previous section but we now use the full 3d space with the _ same _ chemical potential as before .
typical results are shown in figs .
[ 3da][3dc ] for a ramping down time of the potential barrier of 25 ms .
figure [ 3da ] corresponds to the case of equal initial phases @xmath48 , while fig .
[ 3db ] pertains to initial phases @xmath26 , and fig .
[ 3dc ] to phases @xmath67 .
the figures depict contour plots of the density ( top rows ) and vorticity ( middle rows ) , as well as a @xmath54-projection of the density ( bottom rows ) as it would be observed in the laboratory .
the vortex structure is considerably more complex in the 3d scenario because the vorticity does not show up as straight vortex lines but rather as a complex web of vortex filaments in various directions . as in the 2d case
, there is the formation of a vertical vortex line at the center of the cloud ( cf .
second row in fig .
[ 3db ] ) for the appropriate relative phases of the different fragments of the condensate with the same conditions as before . nonetheless , in the 3d case , the central vortex line is prone to bending as can be clearly seen in the later stages of the dynamical evolution presented in the second row of fig .
in fact , the vortex bending is even clearly visible in the @xmath54-projection ( see third row of fig .
[ 3db ] ) .
these 3d numerical experiments are quite revealing in that the laboratory experiments can only show projections of the density and thus missing to a great extent are the intricate vortex line dynamics .
importantly also , between the bending effect and the integrated view used in the experimental images , it is possible for the presence of vortex - like structures or filaments to be blurred ( as in the later stages of fig . [ 3db ] ) or entirely lost ( as in the later stages of fig . [ 3da ] and fig . [ 3dc ] ) . as can be seen from figs .
[ 3da][3dc ] , the vorticity emerges at the early stages of the merger ( @xmath71 ms ) , through vorticity sheets that nucleate some of the vortex line structures . nonetheless , it is interesting to note that most of the vorticity is carried by vortex lines and vortex rings that are _ horizontal _ ( except the notable case of the vertical vortex line depicted in the second row of fig .
[ 3db ] ) .
this fact is also quite visible in the density contour plots for @xmath72 ms where the horizontal vortex lines `` pinch '' the cloud and create peripheral horizontal ridges around the cloud .
it is also possible to observe some vorticity in the bulk of the cloud that does not directly come from the phase differences between the initial fragments , but from the actual turbulence that is created by the fragment collision . as an example ,
two small vortex rings are clearly visible in fig .
[ 3da ] for @xmath72 ms ( one close to the top of the cloud and the other one 1/3 from the bottom ) .
we would like to stress the difficulty of capturing the vorticity at the edge of the cloud ( where most vorticity is actually observed ) in our numerical experiments .
this is due to the fact that the vorticity is defined as the curl of the fluid velocity of eq .
( [ fluid_vel ] ) that is normalized by the density .
the numerical effect is that close to the periphery ( where the density is small ) the fluid velocity corresponds to the ratio of small numbers which imposes great numerical difficulties .
nonetheless , by using a fine grid of @xmath73 we are able to capture most of the delicate vorticity dynamics at the periphery of the cloud consisting , mostly , of horizontal vortex lines that are parallel to the periphery of the cloud .
another interesting phenomenon is the oscillation of the atomic cloud .
the cloud starts with a larger horizontal extent compared to the vertical one and after merger creates an almost spherical cloud , which in turn elongates again in the horizontal direction after the fragments go `` through '' each other .
this behavior repeats a few times until the cloud takes an approximate spherical shape ( results not shown here ) .
we have also monitored the effects of damping due to the coupling of the condensed atoms to the thermal cloud .
equation ( [ gpe ] ) is obtained by supposing a dilute bose gas at a temperature close to _ absolute zero_. however , at finite temperatures , but still smaller than the critical temperature @xmath74 for condensation , a fraction of the atoms are not condensed and form the so - called thermal cloud . in turn
, this thermal cloud induces a damping on the dynamics of the condensed cloud .
we used the approach of phenomenological damping @xcite described in refs .
@xcite that relies on replacing the @xmath75 in front of the time derivative in eq .
( [ gpe ] ) by @xmath76 , where @xmath77 is the damping rate , and by renormalizing the solution at each iteration to keep the initial mass ( number of atoms ) constant during integration .
we tested values of @xmath77 in the interval @xmath78 $ ] that contains the value of 0.03 estimated in ref .
@xcite for a temperature @xmath79 .
the results of the phenomenological damping are , qualitatively , very similar ( results not shown here ) to the effects of ramping down the barrier over longer time scales : larger damping resulting in a stronger suppression of the vorticity generated by the merger of the different cloud fragments .
the simulations described in the present work were directly aimed at developing a more thorough understanding of both the experimental results of ref .
@xcite and the dynamics of vortex formation during bec merging and collisions . in this section ,
we thus briefly discuss the observed similarities and differences between the experimental and theoretical results . in the laboratory experiment ,
an optical potential was used to segment a harmonic trap before condensation was achieved ; with additional evaporative cooling , three initially isolated and mutually independent ( i.e. , uncorrelated phases between the different fragments ) condensates were created . a phase - contrast image of three such becs is given in fig .
[ fig_expt](a ) , which can be directly compared with the simulated data of fig .
[ 3da ] ( bottom row , left image ) .
the initial condition of incoherent condensate fragments serves as the conceptual basis behind the motivation to impose and examine various relative phases between the fragments in this work s simulations . in the experiment ,
the optical barrier was ramped off approximately linearly over time scales between 50 ms and 3 s , significantly longer than the time scales considered in the simulations . at the end of the merging process ,
immediately after the barrier was completely removed , the fully merged bec was released from the harmonic trap and allowed to ballistically expand for 56 ms to enable imaging of vortex cores .
example images of merged and expanded becs are shown in fig .
[ fig_expt](b ) through ( d ) .
these images can be compared with the simulated data of figs .
[ 3da ] through [ 3dc ] ( bottom rows ) ; note that the simulations do not involve an expansion stage . -0.4 cm in both experimental results and in the simulations shown here , we note the following important similarities .
first , it is clear that vortex cores may be formed during the merging process .
second , as noted in the experimental work , the vortex formation process should depend upon relative phases between the condensates .
this conclusion is borne out by the present work .
specifically , the simple analysis of a slow merging process leading to a 25% probability of vortex formation , as presented in the experimental work , matches the results of the simulations summarized in fig .
[ diagram ] .
finally , experimental and simulated results show that faster merging leads to more vortices initially created , but these vortices may self - annihilate with time by holding the fully merged bec in the trap .
however , there is one notable quantitative difference between the experimental and numerical results . in the experimental work , for barrier
ramp - down times longer than 1 s , single vortices were experimentally observed in approximately 25% of the images obtained directly at the conclusion of the merging process .
multiple vortex cores were not observed under these conditions . for faster ramps , experimental images containing either single or multiple cores
were more often obtained , with significantly more than 25% of the images containing at least one vortex core .
the results of the numerical data show that the slow merging limit ( where multiple vortices cease to be created during merging ) is reached for merging times that are much shorter than in the experiment . in other words ,
images with multiple vortices are seen in the experiment under conditions where the numerical results would suggest that a given bec should have at most one vortex .
there are a few possible sources of this discrepancy . at first glance
, it might appear that the spontaneous formation of vortices in becs during evaporative cooling in an axi - symmetric _ harmonic _ trap , as noted in ref .
@xcite , could play a role in higher percentage of vortices seen in the experiment .
such vortex formation processes can not be described by the gpe and are thus not observable in the simulations of this work . however , due to angular momentum damping and self - annihilation of vortices in the asymmetric local potentials in which the three bec fragments grow , we believe that vortices that might be spontaneously created in one ( or more ) of the three becs are unlikely to survive at rates that would affect the experimental observations of ref . @xcite .
this possible source for the quantitative discrepancy could be tested , for example , with simulation methods based on the stochastic gpe @xcite .
perhaps a more likely source of the quantitative discrepancy may lie in the optical potential energy or shape ; differences between the experiment and simulations regarding barrier heights , widths , and ramp - down trajectory might induce more vortices to form during merging .
for example , if center - of - mass oscillations of the cloud were induced in the experiment , atomic fluid flow around the central portion of the optical barrier could induce formation of vortices and lead to increased vortex observation rates in the experiment .
such processes could be studied in future gpe simulations in order to further characterize dynamical processes that may be involved in vortex formation
. it might also be possible that imperfections in the true optical barrier used in the experiment could pin vortices for a portion of the barrier ramp process , and significantly alter the vortex formation and annihilation process . finally , much of the present work focused on the central portion of the trap , a region that encompasses the center of the merged bec . in the experimental work , however , vortices were most often seen further from the bec center .
we have studied the formation and subsequent evolution of vortex structures and filaments in a system directly simulating the experimental setup of ref .
in particular , we have considered the case of three independent fragments ( of variable initial relative phases ) and how these merge upon the ramping down and eventual removal of the optical barrier that separates them . while there are many similarities between the numerical results and the experimental results of ref .
@xcite , the numerical simulations importantly show features and new dynamics not discussed or observed in the experimental work .
the first part of our study concerns the simpler two - dimensional setting , where it is straightforward to observe the interference of the independent matter waves , the ensuing formation of vortices , as well as their motion within the cloud , as a function of different parameters such as the ramping - down time or the initial relative phases between the fragments .
different diagnostics for the vorticity were developed in the process ( such as the @xmath54-component of the angular momentum , or the integrated velocity of the flow throughout the cloud or near its center ) and their dynamics was explained based on the evolution simulations .
principal findings of this part of the work included the formation of smaller numbers of vortices as the ramping - down time was increased and the formation of a single vortex in the core of the condensate for appropriate , discrete - vortex - like relations between the phases of the different fragments . the second part of our work explored how the features found in the two - dimensional setup are generalized in a fully three - dimensional setting , and how these affect the measurement process through , e.g. , the projection of the bec density on the plane .
key features of the latter dynamical evolutions involved the blurring of the vortex dynamics by the projection process coupled with the spontaneous vortex bending even when the different fragments have the appropriate phase relation to generate a vortex through their merging . in the 3d setting ,
the vorticity emerged in the form of vorticity sheets inducing vortex filaments ( most often in a horizontal form ) which led to pinching effects at the vortex cloud periphery and the formation of corresponding ridges in the atomic density profile .
our work is related to a physical mechanism that has previously been discussed in the context of topological defect formation and trapping during phase transitions , often referred to as the kibble - zurek ( kz ) mechanism @xcite . for the case of cooling of an atomic gas through the bec phase transition @xcite , the kz scenario involves the growth and subsequent merging of phase - incoherent regions of the atomic cloud , with vortices being trapped in the merging process as the bec grows in size and atom number .
the full kz mechanism involves physics beyond the scope of the simulations of this work ; however , our simulations show dynamical processes that may relate to the portion of the kz mechanism involving the merging of phase - incoherent regions of condensed atoms .
our numerical results showing the phase relationships involved in vortex formation during merging are consistent with basic notions of this portion of the kz mechanism .
there are many interesting questions to consider for future work in the present framework .
firstly , it is clear that further analysis of experimental data and further variations of simulation parameters will be needed in order to resolve the quantitative differences between experimental and numerical results , as discussed in sec .
[ sec : expt ] .
perhaps additional light on this question ( and an interesting diagnostic in its own right ) would be the examination of the integrated density along the @xmath80 plane which should perhaps detect some of the horizontal vorticity filaments illustrated herein .
a particularly challenging ( and more general ) question along the same vein concerns the extent to which it may be possible to reconstruct the fully three - dimensional cloud density from such projections . both from an experimental and from a theoretical point of view it would be interesting to extend the present considerations also to multi - component condensates in order to examine the potential formation of vortex - like filaments and structures in settings similar to the ones presented herein ( e.g. , containing fragments from different components ) . in the latter setting , there would exist an exciting interplay between the interference mechanisms and the formation of the coherent structures , and the phase separation dynamics between the components ; see e.g. , the recent experimental results of ref .
@xcite and references therein .
we are thankful to matthew davis for providing a careful reading of the manuscript .
pgk and rcg acknowledge the support of nsf - dms-0505663 .
pgk also acknowledges support from nsf - dms-0619492 and nsf - career .
bpa acknowledges support from the army research office and nsf - mps-0354977 .
pitaevskii and s. stringari , _ bose - einstein condensation _ , oxford university press ( oxford , 2003 ) ; c.j .
pethick and h. smith , _ bose - einstein condensation in dilute gases _ , cambridge university press ( cambridge , 2002 ) . | we study the formation of vortices in a bose - einstein condensate ( bec ) that has been prepared by allowing isolated and independent condensed fragments to merge together .
we focus on the experimental setup of scherer _ et al . _
[ phys .
rev .
lett . * 98 * , 110402 ( 2007 ) ] , where three becs are created in a magnetic trap that is segmented into three regions by a repulsive optical potential ; the becs merge together as the optical potential is removed .
first , we study the two - dimensional case , in particular we examine the effects of the relative phases of the different fragments and the removal rate of the optical potential on the vortex formation .
we find that many vortices are created by instant removal of the optical potential regardless of relative phases , and that fewer vortices are created if the intensity of the optical potential is gradually ramped down and the condensed fragments gradually merge . in all cases , self - annihilation of vortices of opposite charge
is observed .
we also find that for sufficiently long barrier ramp times , the initial relative phases between the fragments leave a clear imprint on the resulting topological configuration .
finally , we study the three - dimensional system and the formation of vortex lines and vortex rings due to the merger of the bec fragments ; our results illustrate how the relevant vorticity is manifested for appropriate phase differences , as well as how it may be masked by the planar projections observed experimentally . |
decentralized optimization has recently been receiving significant attention due to the emergence of large - scale distributed algorithms in machine learning , signal processing , and control applications for wireless communication networks , power networks , and sensor networks ; see , for example , @xcite . a central generic problem in such applications
is decentralized resource allocation for a multiagent system , where the agents collectively solve an optimization problem in the absence of full knowledge about the overall problem structure . in such settings ,
the agents are allowed to communicate to each other some relevant estimates so as to learn the information needed for an efficient global resource allocation .
the decentralized structure of the problem is reflected in the agents local view of the underlying communication network , where each agent exchanges messages only with its neighbors . in recent literature on control and optimization ,
an extensively studied decentralized resource allocation problem is one where the system objective function @xmath3 is given as a sum of local objective functions , i.e. , @xmath4 where @xmath5 is known only to agent @xmath6 ; see , for example @xcite . in this case
, the objective function is separable across the agents , but the agents are coupled through the resource allocation vector @xmath7 . each agent maintains and updates its own copy of the allocation / decision vector @xmath7 , while trying to estimate an optimal decision for the system problem .
another decentralized resource allocation problem is the one where the system objective function @xmath3 may not admit a natural decomposition of the form @xmath8 , and the resource allocation vector @xmath9 is distributed among the agents , where each agent @xmath6 is responsible for maintaining and updating only a coordinate ( or a part ) @xmath10 of the whole vector @xmath7 .
such decentralized problems have been considered in @xcite ( see also the textbook @xcite ) . in the preceding work ,
decentralized approaches converge when the agents are using weighted averaging , or when certain contraction conditions are satisfied .
recently , li and marden @xcite have proposed a different algorithm with local updates , where each agent @xmath6 keeps estimates for the variables @xmath11 , @xmath12 , that are controlled by all the other agents in the network .
the convergence of this algorithm relies on some contraction properties of the iterates .
note that all the aforementioned algorithms were developed for offline optimization problems .
our proposed algorithm * oda - ps * is closest to recent papers @xcite .
the papers proposed a decentralized algorithm for online convex optimization which is very similar to * oda - ps * in a sense that they also introduce online subgradient estimations in primal @xcite or dual @xcite space into information aggregation using push - sum . in these papers ,
the agents share a common decision set in @xmath13 , the objective functions are separable across the agents at each time ( i.e. , @xmath14 for all @xmath15 ) , and the regret is analyzed in terms of each agent s own copy of the whole decision vector @xmath16 .
moreover , an additional assumption is made in @xcite that the objective functions are strongly convex .
the paper is organized as follows . in section [
sec : problem ] , we formalize the problem and describe how the agents interact . in section [ sec : basicregret ] , we provide an online decentralized dual - averaging algorithm in a generic form and establish a basic regret bound which can be used later for particular instantiations , namely , for the two algorithms * oda - c * and * oda - ps*. these algorithms are analyzed in sections [ sec : algo - regret ] , where we establish @xmath0 regret bounds under mild assumptions . in section
[ sec : sim ] , we demonstrate our analysis by simulations on a sensor network .
we conclude the paper with some comments in section [ sec : conclusion ] .
* notation : *
consider a multiagent system ( network ) consisting of @xmath17 agents , indexed by elements of the set @xmath18 $ ] .
each agent @xmath19 takes actions in an action space @xmath20 , which is a closed and bounded interval of the real line .
being a compact convex subset of a multidimensional space @xmath13 ; we mainly stick to the scalar case for simplicity . ]
the communication among agents in the network is governed by either one of the two following models : 1 .
an undirected connected graph @xmath21 : if agents @xmath6 and @xmath22 are connected by an edge ( which we denote by @xmath23 ) , then they may exchange information with one another .
thus , each agent @xmath19 may directly communicate only with the agents in its neighborhood @xmath24 .
note that agent @xmath6 is always contained in its own neighborhood .
time - varying digraphs @xmath25 , for @xmath26 : if there exists a directed link from agent @xmath22 to @xmath6 at time @xmath15 ( which we denote by @xmath27 ) , agent @xmath22 may send its information to agent @xmath6 .
we use the notation @xmath28 and @xmath29 to denote the in and out neighbors of agent @xmath6 at time @xmath15 , respectively .
that is , @xmath30 @xmath31 in this case , we assume that there always exists a self - loop @xmath32 for all agent @xmath19 . therefore , agent @xmath6 is always contained in its own neighborhood . also , we use @xmath33 to denote the out degree of node @xmath6 at time @xmath15 .
i.e. , @xmath34 we assume @xmath35-strong connectivity of the graphs @xmath36 with some scalar @xmath37 , i.e. , a graph with the following edge set @xmath38 is strongly connected for every @xmath26 .
the network interacts with an environment according to the protocol shown in figure [ fig : ogo ] .
we leave the details of the signal generation process vague for the moment , except to note that the signals received by all agents at time @xmath15 may depend on all the information available up to time @xmath15 ( including @xmath39 , as well as all of the local information exchanged in the network ) .
moreover , the environment may be adaptive , i.e. , the choice of the function @xmath40 may depend on all of the data generated by the network up to time @xmath15 .
let us denote the network action at time @xmath15 by @xmath41 we consider the _ network regret _ : @xmath42 is the difference between the total cost incurred by the network at time @xmath1 and the smallest total cost that could have been achieved with a single action in @xmath43 in hindsight ( i.e. , with perfect advance knowledge of the sequence @xmath44 ) and without any restriction on the communication between the agents .
the problem is to design the rule ( or policy ) each agent @xmath19 should use to determine its action @xmath45 based on the local information available to it at time @xmath15 , such that the regret in is ( a ) sublinear as a function of the time horizon @xmath1 and ( b ) exhibits `` reasonable '' dependence on the number of agents @xmath17 and on the topology of the communication graphs .
we now introduce a generic algorithm for solving the decentralized online optimization problem defined in section [ sec : problem ] .
the algorithm uses the dual - averaging subgradient method of nesterov @xcite as an optimization subroutine .
we now present a basic regret bound that can be used for any generic algorithm of the form ( [ eqn : gdyn1])-([eqn : gdyn2 ] ) under the following assumption : [ assume : lipschitz ] all functions @xmath46 are lipschitz continuous with a constant @xmath47 : @xmath48 [ thm : main ] let @xmath49 , @xmath19 , be the sequences of the agents primal iterates , let @xmath50 with @xmath51 be the sequence of the agents local updates , and let @xmath52 be generated as @xmath53 then , under assumption [ assume : lipschitz ] , the _ network regret _
@xmath54 in ( [ eq : regret ] ) can be upper - bounded in terms of @xmath55 and @xmath56 as @xmath57 where @xmath58 is the diameter of the set @xmath20 , and @xmath59 . since @xmath60 is a continuous function on the compact set @xmath43 , @xmath61 by the weierstrass theorem .
we can write @xmath62 where the second step follows from convexity of @xmath40 , while the last step uses the fact that all @xmath46 are @xmath47-lipschitz .
we have the following for the first term in : @xmath63 where the equality follows from the definition of @xmath64 in ( [ eq : net - ac ] ) and @xmath65 .
the second term in can be further expanded as @xmath66 now , from relation ( [ eq : xu ] ) we obtain @xmath67 therefore , by ( * ? ? ?
* lemma 3 ) , we can write @xmath68 for the second term on the right - hand side of , we have @xmath69 combining the estimates in eqs . - and taking the supremum over all @xmath70 , we get the desired result . @xmath71
we now introduce a decentralized online optimization algorithm which uses a circulation - based framework for its dual update rule .
we refer to this algorithm as * oda - c * ( online dual averaing with circulation - based communication ) .
* oda - c * uses the network model ( g1 ) for its communication .
fix a vector @xmath72 of positive weights and a nonnegative @xmath73 matrix @xmath74 , such that @xmath75 only if @xmath76 , satisfying the following symmetry condition : @xmath77 then , * oda - c * uses the following instantiation of the update rules in ( [ eqn : gdyn1])-([eqn : gdyn2 ] ) : @xmath78\label{eqn : dyn1}\\ \xb_i(t+1 ) & = \pi_{\sx^n}^{\psi } \left(\zb_i(t+1),\alpha(t)\right),\label{eqn : dyn2}\ ] ] where @xmath79 represents a vector of messages transmitted by agent @xmath22 to agent @xmath6 , provided that @xmath76 .
the dual update rule is inspired by the state dynamics proposed by li and marden @xcite , whereas the primal update rule is exactly what one has in nesterov s scheme @xcite . to complete the description of the algorithm
, we must specify the update policies @xmath80 and the messages @xmath81 .
we assume that all agents receive a complete description of @xmath40 .
agent @xmath6 then computes @xmath82,~t\ge 0.\end{aligned}\ ] ] and feeds this signal back into the dynamics ( [ eqn : dyn1 ] ) .
note , however , that the execution of the algorithm will not change if the agents never directly learn the full function @xmath40 , nor even the full gradient @xmath83 , but instead receive the local gradient signal @xmath84 .
the messages @xmath85 take the form @xmath86 for all @xmath15 and all agents @xmath87 .
let @xmath88 our regret analysis rests on the following simple but important fact : [ lm : mean_field ] the weighted sum @xmath89 evolves according to the linear dynamics @xmath90 where @xmath91 .
we observe that the relation in holds regardless of the choices of decisions @xmath92 and @xmath93 .
moreover , we point out that if @xmath94 , then the combination of and will reduce to a centralized online variant of nesterov s scheme @xcite .
let @xmath95 denote the @xmath73 matrix with entries @xmath96_{ij } = v^k_{j\to i}(t ) - v^k_{i \to j}(t)$ ] . then @xmath97_{ij}\right\ } \\ & = \bz^k(t ) + u_k(t ) + \tr [ \tilde{m}v^k(t)],\end{aligned}\ ] ] where @xmath98 is an @xmath99 matrix with entries @xmath100 .
since @xmath98 is a symmetric matrix , by , and @xmath95 is skew - symmetric , @xmath101 = 0 $ ] , so we obtain .
@xmath71 we now particularize the bound in theorem [ thm : main ] to this scenario under the following additional assumption : [ assume : lipgrad ] all functions @xmath46 are differentiable and have lipschitz continuous gradients with constant @xmath102 : @xmath103 [ thm : local_grad_signals ] under assumptions [ assume : lipschitz][assume : lipgrad ] , the regret of any algorithm of the form ( [ eqn : dyn1])-([eqn : dyn2 ] ) , and with @xmath55 computed according to , can be upper - bounded as follows : @xmath104 the terms on the right - hand side of the bound in theorem [ thm : main ] can be further estimated as follows . since each @xmath105 is @xmath47-lipschitz , @xmath106 it remains to estimate term ( e3 ) in theorem [ thm : main ] .
to that end , we write @xmath107 where we have exploited the fact that the gradients of all @xmath46 are @xmath102-lipschitz . now , by construction , @xmath108 where the last step follows from the fact that the map @xmath109 is @xmath110-lipschitz ( see , e.g. , ( * ? ? ? * lemma 1 ) ) . substituting these estimates into the bound in theorem [ thm : main ]
, we get the result .
@xmath71 [ cor : regret_bound ]
suppose that the policies for computing @xmath80 and @xmath111 are such that , for all @xmath15 and for any sequence @xmath112 , @xmath113 for some finite constant @xmath114 ( which may depend on @xmath17 and on other problem parameters ) .
then , the regret of the algorithm ( [ eqn : dyn1])-([eqn : dyn2 ] ) is bounded by @xmath115\sum^t_{t=1}\alpha(t-1 ) + \frac{c}{\alpha(t)}. \end{aligned}\ ] ] in particular , if we choose @xmath116 for @xmath117 , then the regret is @xmath0 : @xmath118 \sqrt{t } + c\sqrt{t+1 } .\end{aligned}\ ] ] the conditions we have imposed on the pair @xmath119 are equivalent to saying that @xmath74 is the transition probability matrix of a reversible random walk on @xmath120 with invariant distribution @xmath72 @xcite .
[ lem : disagree ] for the policy in - we have @xmath121 for every @xmath26 , where @xmath122 is of the vector @xmath123 , and where @xmath124 denotes the spectral gap of @xmath74 @xcite , i.e. , @xmath125 from @xmath126 thus , we upper - bound the quantity on the right - hand side . from , we can rewrite the dynamics as @xmath127 by unrolling the dynamics and from time 0 to @xmath15 and recalling that @xmath128 for all @xmath6 , we obtain : @xmath129 we have @xmath130 note that @xmath131 . from ( [ eqn : zvec2 ] ) and ( [ eqn : zslr ] ) , we have @xmath132 by the properties of markov matrices @xcite , for any @xmath123 , @xmath133 therefore , @xmath134 from relations ( [ eqn : dyn ] ) and ( [ eqn : right ] ) , we obtain @xmath135 where assumption [ assume : lipschitz ] is used in the last inequality . from this and relation ,
we obtain @xmath136 .
@xmath71 by combining theorem [ thm : local_grad_signals ] and lemma [ lem : disagree ] , we can now provide a regret bound for * oda - c * : [ thm : linear_ave ] @xmath137 with @xmath138 result follows .
we now introduce another decentralized online optimization algorithm which uses the push - sum communication protocol for its dual update rule .
we refer to this algorithm as * oda - ps * ( online dual averaing with push - sum based communication ) .
* oda - ps * uses the network model ( g2 ) for its communication . for * oda - ps
* , each agent @xmath6 maintains an additional scalar sequence @xmath139 .
then , this algorithm particularizes the update rule in ( [ eqn : gdyn1])-([eqn : gdyn2 ] ) as @xmath140_{ij } w_j(t)\label{eqn : algo1}\\ z_i^k(t+1 ) = ~ & n\delta_i^ku_i(t ) + \sum_{j=1}^n [ a(t)]_{ij } z_j^k(t ) , ~k\in [ n]\label{eqn : algo2}\\ \xb_i(t+1 ) = ~ & \pi^\psi_{\sx^n}\paren{\frac{\zb_i(t+1)}{w_i(t+1)},\a(t)}\label{eqn : algo3}\end{aligned}\ ] ] where the weight matrix @xmath141 is defined by the out - degrees of the in - neighbors , i.e. , @xmath142_{ij } = \left\ { \begin{array}{ll } 1/d_j(t ) & \text{whenever } j \in \nin(t)\\ 0 & \text{otherwise . } \end{array } \right.\ ] ] the matrix @xmath141 is column stochastic by construction .
note that the above update rules are based on a simple broadcast communication .
each agent @xmath6 broadcasts ( or _ pushes _ ) the quantities @xmath143 and @xmath144 to all of the nodes in its out - neighborhood @xmath145 .
then , in - each agent simply _ sums _ all the received messages to obtain @xmath146 and @xmath147 . the update rule can be executed locally . to complete the description of the algorithm
, we must specify the update policies @xmath80 . as in * oda - c
* , we assume that the signal agent @xmath6 gets from the environment at time @xmath15 is simply the @xmath6-th coordinate of the gradient of @xmath40 at the agent s primal variable @xmath148 .
thus , we define : @xmath149,~t\ge 0,\ ] ] i.e. , the update performed by agent @xmath6 at time @xmath15 is the simply the @xmath6-th coordinate of the gradient of @xmath40 at the agent s primal variable @xmath148 .
we assume that each agent @xmath6 initializes its updates with @xmath150 and @xmath151 , while @xmath152 can be any arbitrary value in @xmath153 .
we also recall that the local action of agent @xmath6 at time @xmath15 is given by the @xmath6th coordinate of @xmath148 , i.e. , @xmath154 for notational convenience , let us denote the products of the weight matrices @xmath155 by @xmath156 , i.e. , @xmath157 also , we denote @xmath158 for the regret analysis , we first study the dynamics of the dual iterates @xmath159 and its `` mean field '' @xmath160 in the following lemma . we remind that @xmath161 and @xmath162.\end{aligned}\ ] ] [ lem : zbar ] let @xmath128 for all @xmath19 . *
the weighted sum @xmath163 evolves according to the linear dynamics @xmath164 where @xmath165 . * for any @xmath166 $ ] , the iterates in ( [ eqn : algo2 ] ) evolve according to the following dynamics @xmath167_{ik } u_k(s).\ ] ] * from relation ( [ eqn : algo2 ] ) , we have for all @xmath168 $ ] @xmath169_{ij}z_j^k(t)\right]\\ = & ~u_k(t ) + \frac{1}{n}\sum_{j=1}^n z_j^k(t)\sum_{i=1}^n[a(t)]_{ij}\\ = & ~u_k(t ) + \bar{z}^k(t),\end{aligned}\ ] ] where the last equality follows from the column - stochasticity of the matrix @xmath141 .
the desired result follows by stacking up the scalar relation above over @xmath170 .
* by stacking up the equation ( [ eqn : algo2 ] ) over @xmath6 , we have for all @xmath26 and @xmath168 $ ] @xmath171 by unrolling this equation from time 0 to @xmath15 , we obtain @xmath172 where the equalities follows from @xmath173 and the initial condition @xmath128 for all @xmath19 .
we get the desired result by taking the @xmath6-th component of this vector .
@xmath71 we now particularize the bound in theorem [ thm : main ] in this scenario under the additional assumption on the lipschitz continuous gradients ( assumption [ assume : lipgrad ] in section [ sec : algo - regret ] ) .
[ thm:4 ] under assumptions [ assume : lipschitz]-[assume : lipgrad ] , the regret of the algorithm ( [ eqn : algo1])-([eqn : algo3 ] ) with the local update @xmath174 of agent @xmath6 computed according to ( [ eqn : policy1 ] ) can be upper - bounded as follows : for all @xmath175 , @xmath176 since the definition of @xmath55 in * oda - ps * ( cf .
eq . ) coincides with that in * oda - c * ( cf .
, we can reuse all the derivations in the proof of theorem [ thm : local_grad_signals ] except for the network - wide disagreement term : @xmath177 where the last inequality follows from the @xmath110-lipschitzian property of the map @xmath109 ( * ? ? ?
* lemma 1 ) .
@xmath71 we now show that the network - wide disagreement term in theorem [ thm:4 ] is indeed upper - bounded by some constant . for doing this , we first restate a lemma from @xcite .
[ lem : no ] let the graph sequence @xmath178 be @xmath35-strongly connected
. then the following statements are valid .
* there is a sequence @xmath179 of stochastic vectors such that the matrix difference @xmath180 for @xmath181 decays geometrically , i.e. , for all @xmath182 $ ] .
@xmath183_{ij}-\phi_i(t ) } \le
\beta\t^{t - s } \quad \text{for all } t \ge s \ge 0,\ ] ] where we can always choose @xmath184 if in addition each @xmath36 is regular , we may choose @xmath185 or @xmath186 whenever @xmath187 .
* the quantity @xmath188_i}\ ] ] satisfies @xmath189 moreover , if the graphs @xmath36 are regular , we have @xmath190 .
the next lemma provides an upper - bound for @xmath191 .
[ lem : network ] let the sequences @xmath192 and @xmath193 be generated according to the algorithm ( [ eqn : algo1])-([eqn : algo2 ] ) . recall that @xmath194 .
then , we have for all @xmath26 , @xmath195 where the constants @xmath196 , @xmath197 and @xmath198 are as defined in lemma [ lem : no ] . from the definitions of @xmath159 , @xmath160 and @xmath199
, we have @xmath200 thus , we can upper - bound the quantity on the right - hand side . by inspecting equation ( [ eqn : algo1 ] ) , it is easy to see that for any @xmath19 and @xmath26 , we have @xmath201_{i\ell } w_i(0 ) = \sum_{\ell=1}^n[a(t-1:0)]_{i\ell}.\ ] ] from this and lemma [ lem : zbar ] , we have the following chain of relations : @xmath202_{ik}u_k(s)}{\sum_{\ell=1}^n[a(t-1:0)]_{i\ell } } - \sum_{s=0}^{t-1}u_k(s)\nonumber\\ ~= & \sum_{s=0}^{t-1}u_k(s)\frac{\sum_{\ell=1}^n[a(t-1:s+1)]_{ik}-\sum_{\ell=1}^n[a(t-1:0)]_{i\ell}}{\sum_{\ell=1}^n[a(t-1:0)]_{i\ell}}\nonumber\\ ~\le & \sum_{s=0}^{t-1}u_k(s ) \bigg(\frac{\sum_{\ell=1}^n\left([a(t-1:s+1)]_{ik}-\phi_i(t-1)\right)}{\sum_{\ell=1}^n[a(t-1:0)]_{i\ell } } \nonumber\\ ~ & + \frac{\sum_{\ell=1}^n\left(\phi_i(t-1)-[a(t-1:0)]_{i\ell}\right)}{\sum_{\ell=1}^n[a(t-1:0)]_{i\ell}}\bigg ) \nonumber\\ ~\le & \sum_{s=0}^{t-1}u_k(s ) \frac{\beta\t^{t - s-2}+\beta\theta^{t-1}}{\gamma},\end{aligned}\ ] ] where the inequalities follow from adding and subtracting @xmath203 and from lemma [ lem : no ] . from relation ( [ eqn : policy1 ] )
, we have @xmath204 combining this and the fact that @xmath205 for all @xmath206 , we further have @xmath207 substituting this estimate in relation , we get the desired result .
@xmath71 by combining theorem [ thm:4 ] and lemma [ lem : network ] , we can now provide the regret bound of * oda - ps * : [ thm : linear_ave2 ] @xmath208 where the constants @xmath196 , @xmath197 and @xmath198 are as defined in lemma [ lem : no ] . by jensen s inequality ,
we have @xmath209 hence , using lemma [ lem : network ] , we can estimate the network - wide disagreement term as follows : @xmath210 thus , the conditions of corollary [ cor : regret_bound ] with this modified network - wide agreement hold with @xmath211 and the stated result follows .
consider the problem of estimating some target vector @xmath212 using measurements from a network of @xmath17 sensors .
each sensor @xmath6 is in charge of estimating a subvector @xmath213 of @xmath7 , where @xmath214 and @xmath215 is some very large number .
an example includes the localization of multiple targets , where in this case @xmath212 becomes a stacked vector of all target locations .
when there are a number of spatially dispersed targets , we can certainly benefit from distributed sensing .
the sensors are assumed to have a linear model of @xmath216 , where @xmath217 and @xmath218 . at each time @xmath15 , each sensor @xmath19 estimates its portion @xmath219 of the target vector @xmath212 , and then takes a measurement @xmath220 , which is corrupted by observation error and possibly by modeling error .
we assume all sources of errors can be represented as an additive noise , i.e. , @xmath221 where @xmath222 with @xmath223 is a stacked vector of all @xmath224 s and @xmath225 , where @xmath226 is the noise covariance matrix .
the regret is computed with respect to the least - squares estimate of the target locations at time @xmath1 , i.e. , @xmath227 where @xmath228 and we set @xmath229 $ ] . for * oda - c * , we experiment with a @xmath230 node cycle graph whose communication topology is given as : @xmath231 we set @xmath232 , @xmath233 for all @xmath6 , and @xmath234 if @xmath23 . for * oda - ps * , we experiment with a time - varying sequence of digraphs with @xmath230 nodes whose communication topology is changing periodically with period @xmath235 .
the graph sequence is , therefore , @xmath235-strongly connected . in figure
[ fig : graph ] , we depict the repetition of the 3 corresponding graphs . the averaging matrices @xmath141 ( cf .
can be determined accordingly .
we ran our algorithms once for each @xmath236 $ ] .
that is , for a given @xmath1 , the iterates in the algorithms are updated from @xmath237 to @xmath238 .
we used step size @xmath239 for both algorithms . ] vs. iterations for online distributed active sensing using * oda - c * ( left ) and * oda - ps * ( right ) [ fig : sensing],title="fig : " ] vs. iterations for online distributed active sensing using * oda - c * ( left ) and * oda - ps * ( right ) [ fig : sensing],title="fig : " ] in figure [ fig : sensing ] , we depict the average regret @xmath240 over time @xmath1 of the distributed sensing problem when * oda - c * and * oda - ps * are used , respectively .
it shows that the regret is sublinear for both algorithms and the average @xmath240 goes to zero as the time increases .
we have studied an online optimization problem in a multiagent network .
we proposed two decentralized variants of nesterov s primal - dual algorithm , namely , * oda - c * using circulation - based dynamics for time - invariant networks and * oda - ps * using broadcast - based push - sum protocol for time - varying networks .
we have established a generic regret bound and provided its refinements for certain information exchange policies .
the regret is shown to grow as @xmath0 when the step size is @xmath241 . for * oda - c * , the bound is valid for a static connectivity graph and a row - stochastic matrix of weights @xmath242 $ ] which is reversible with respect to a strictly positive probability vector @xmath243 . for * oda - ps * , the bound is valid for a uniformly strongly connected sequence of digraphs and column - stochastic matrices of weights @xmath141 whose components are based on the out - degrees of neighbors .
simulation results on a sensor network exhibit the desired theoretical properties of the two algorithms .
a. martinoli , f. mondada , g. mermoud , n. correll , m. egerstedt , a. hsieh , l. parker , and k. stoy , _ distributed autonomous robotic systems_. 1em plus 0.5em minus 0.4emspringer tracts in advanced robotics , springer - verlag , 2013 .
chang , a. nedi , and a. scaglione , `` distributed constrained optimization by consensus - based primal - dual perturbation method , '' _ ieee transactions on automatic control _
59 , pp . 15241538 , 2014 .
s. s. ram , a. nedi , and v. v. veeravalli , `` distributed stochastic subgradient projection algorithms for convex optimization , '' _ journal of optimization theory and applications _ ,
147 , pp . 516545 , 2010 .
k. tsianos , s. lawlor , and m. rabbat , `` consensus - based distributed optimization : practical issues and applications in large - scale machine learning , '' in _
50th allerton conference on communication , control , and computing _ , 2012 , pp .
15431550 .
d. jakovetic , j. xavier , and j. moura , `` cooperative convex optimization in networked systems : augmented lagrangian algorithms with directed gossip communication , '' _ ieee transactions on signal processing _ , vol .
59 , pp . 38893902 , 2011 .
j. tsitsiklis , d. bertsekas , and m. athans , `` distributed asynchronous deterministic and stochastic gradient optimization algorithms , '' _ ieee transactions on automatic control _ ,
31 , pp . 803812 , 1986 .
f. benezit , v. blondel , p. thiran , j. tsitsiklis , and m. vetterli , `` weighted gossip : distributed averaging using non - doubly stochastic matrices , '' in _ ieee international symposium on information theory proceedings ( isit ) _ , 2010 , pp .
1753 1757 .
m. akbari , b. gharesifard , and t. linder , `` distributed subgradient - push online convex optimization on time - varying directed graphs , '' in _
52nd allerton conference on communication , control , and computing _ , 2014 , pp .
264269 .
j. c. duchi , a. agarwal , and m. j. wainwright , `` dual averaging for distributed optimization : convergence analysis and network scaling , '' _ ieee transactions on automatic control _ , vol .
592606 , 2012 . | we consider a decentralized online convex optimization problem in a network of agents , where each agent controls only a coordinate ( or a part ) of the global decision vector .
for such a problem , we propose two decentralized variants ( * oda - c * and * oda - ps * ) of nesterov s primal - dual algorithm with dual averaging . in * oda - c * , to mitigate the disagreements on the primal - vector updates , the agents implement a generalization of the local information - exchange dynamics recently proposed by li and marden @xcite over a static undirected graph . in * oda - ps * , the agents implement the broadcast - based push - sum dynamics @xcite over a time - varying sequence of uniformly connected digraphs .
we show that the regret bounds in both cases have sublinear growth of @xmath0 , with the time horizon @xmath1 , when the stepsize is of the form @xmath2 and the objective functions are lipschitz - continuous convex functions with lipschitz gradients .
we also implement the proposed algorithms on a sensor network to complement our theoretical analysis . |
recently , many exotic hadrons observed in experiments are attracting our interest .
they are expected to have unexpected structures , because their properties , such as quantum numbers , masses , and decay patterns , can not be explained by the ordinary picture of hadrons , baryons as @xmath11 and mesons as @xmath12 .
as one of new aspects of these structures , the possibility of multiquarks or hadronic molecules have been extensively discussed . it is considered that , particularly near thresholds , hadronic molecules such as loosely bound or resonant states of meson - meson or meson - baryon emerge as one of hadronic configurations realized in nature . as typical candidates , the meson - meson molecules for @xmath13 and @xmath14 have been studied in the charm and bottom quark regions @xcite . the hadronic molecules as bound or resonant states formed by @xmath15 or @xmath16 meson ( @xmath17 or @xmath18 meson ) and a nucleon ( @xmath19 ) are also interesting . since the @xmath0 and @xmath1 ( @xmath2 and @xmath3 ) molecules contain ordinary flavor structures of three quarks , they are intimately related to the ordinary heavy baryons , @xmath20 , @xmath21 , @xmath22 and their excited states for the charm sector , and @xmath23 , @xmath24 , @xmath25 and their excited states for the bottom sector @xcite .
these excited states have been studied in the quark model extensively @xcite . however , near the @xmath0 and @xmath1 ( @xmath2 and @xmath3 ) thresholds , it is expected that the wave functions of the quark model states have a large overlap with the component of @xmath0 and @xmath1 ( @xmath2 and @xmath3 ) , and the properties of such baryon states are strongly affected by @xmath0 and @xmath1 ( @xmath2 and @xmath3 ) states
. there would be even such states that are dominated by the molecular components . in the @xmath0 and @xmath1 ( @xmath2 and @xmath3 ) systems
, there are two important symmetries ; the heavy quark symmetry and chiral symmetry .
the heavy quark symmetry manifests in mass degeneracy of heavy pseudoscalar mesons @xmath26 , @xmath17 and vector mesons @xmath27 , @xmath28 in the heavy quark mass limit @xcite .
indeed , the mass splitting between a @xmath15 ( @xmath17 ) meson and a @xmath29 ( @xmath28 ) meson is small ; 140 mev for @xmath15 and @xmath29 mesons and 46 mev for @xmath17 and @xmath28 mesons .
the small mass splittings in the heavy flavor sectors should be compared with the large mass splittings , @xmath30 mev for @xmath31 and @xmath32 mesons , and @xmath33 mev for @xmath34 and @xmath35 mesons , in the light flavor sectors .
because of this , the heavy quark symmetry introduces the mixing of @xmath36 and @xmath37 states , where both @xmath38 and @xmath39 mesons are considered on the same footing as fundamental degrees of freedom in the dynamics @xcite .
then , physical states are given as a superposition of @xmath36 and @xmath37 states .
for convenience , in the following , we denote @xmath40 to stand for @xmath38 or @xmath39 .
thanks to the mixing of states in @xmath41 systems , a long range force between a @xmath40 meson and a nucleon @xmath19 is supplied by one pion exchange with the same coupling strengths for @xmath42 and @xmath43 vertices @xcite .
it is interesting that the pion exchange can exist in @xmath41 systems . as a matter of fact
, it is known that the pion exchange is important for the binding of atomic nuclei @xcite . there
the tensor force of the pion exchange that mixes channels with different angular momenta , i.e. @xmath44 and @xmath45 , yields a strong attraction to generate a rich structure of bound as well as resonant states .
because the pion exchange is a long range force , it becomes more significant for states with larger orbital angular momenta .
in fact , it was shown that , in the nucleon - nucleon scattering , the phase shifts in the channels with large orbital angular momenta are reproduced almost only by the pion exchange potential @xcite .
the importance of the pion exchange was also discussed in @xmath46 systems .
it was investigated that the properties of bound and resonant states in @xmath46 systems are dramatically changed , if the tensor force is switched off @xcite . in our previous studies
, it was also shown that the pion exchange potential played a significant role for hadronic molecules such as @xmath47 and @xmath48 baryons @xcite , @xmath49 ( @xmath50 ) mesons ( e.g. @xmath14 meson ) @xcite and @xmath51 ( @xmath52 ) mesons @xcite . in reality , however , the hadronic molecules of @xmath41 do not necessarily correspond to the observed states , because there should be couplings not only to three quark states but also to other meson - baryon states such as @xmath53 , @xmath54 and @xmath55 ( @xmath56 , @xmath57 and @xmath58 ) , and so on . however , we may expect that such couplings are small for @xmath0 and @xmath2 baryons near the thresholds .
the reasons are that the wave functions of hadronic molecules are spatially large as compared to the conventional three quark states , and that the transitions , e.g. from @xmath0 ( @xmath2 ) to @xmath54 ( @xmath57 ) , are suppressed by a heavy quark exchange . from those points of view , in the present discussion , we focus on the role of the @xmath59 and @xmath60 sectors and study the bound or resonant states generated by @xmath59 and @xmath60 . recently , there have been several studies of baryon states as @xmath59 ( @xmath60 ) @xcite and @xmath61 ( @xmath62 ) @xcite molecules .
the present study is different from them in that we focus on the states near the thresholds and with a large angular momentum .
this paper is organized as follows . in sec .
, we briefly describe the interaction between a @xmath63 or @xmath64 meson and a nucleon . in sec .
[ result ] , we solve the coupled channel schrdinger equations numerically for bound and resonant states .
we discuss the results of the @xmath59 and @xmath60 molecule in sec .
[ discussion ] , and summarize the present work in the final section .
following our previous studies @xcite , we consider @xmath34 , @xmath35 and @xmath65 exchange potentials between @xmath40 and @xmath19 .
the interaction lagrangians for heavy meson vertices are given by the heavy quark symmetry and chiral symmetry @xcite , @xmath66 \ , , \label{lpihh}\\ { \cal l}_{v hh } & = -i\beta\mbox{tr } \left [ h_b v^\mu(\rho_\mu)_{ba}\bar{h}_a \right ] + i\lambda\mbox{tr } \left [ h_b\sigma^{\mu\nu}f_{\mu\nu}(\rho)_{ba}\bar{h}_a \right ] \ , , \label{lvhh}\end{aligned}\ ] ] where the subscripts @xmath34 and @xmath67 are for the pion and vector ( @xmath35 and @xmath65 ) meson interactions .
the heavy meson field @xmath68 is defined by @xmath69 $ ] with the four - velocity @xmath70 of a heavy meson , where the subscript @xmath71 is for light flavors , @xmath72 , @xmath73 .
the conjugate field is given by @xmath74 .
the axial current of light flavors is written by @xmath75 with @xmath76 and the pion decay constant @xmath77 mev .
the pion field is defined by @xmath78 and the vector meson field by @xmath79 where @xmath80 corresponds to the gauge coupling constant from hidden local symmetry @xcite .
the vector meson field tensor is written by @xmath81 $ ] .
the coupling constants @xmath82 , @xmath83 and @xmath84 are summarized in table [ constants ] .
these coupling constants are essentially the same as those given in refs .
@xcite , except for the signs of @xmath85 , @xmath83 and @xmath84 for vertices of @xmath34 and @xmath65 , which are reversed due to g - parity transformation between @xmath63 ( @xmath64 ) and @xmath86 ( @xmath87 ) . in eqs . and
, we consider the static approximation @xmath88 .
.masses and coupling constants of mesons @xmath89 , @xmath35 , @xmath65 in refs .
@xcite . [ cols="<,^,^,^,^,^,^",options="header " , ] let us move to @xmath90 state . for @xmath59 , we find one resonance near the @xmath0 threshold when the @xmath34 potential is used .
the resonance energy is @xmath91 mev and the half decay width is @xmath92 mev as shown in table [ tabledn ] . in the scattering state , we define the resonance energy @xmath93 by an inflection point of the phase shift @xcite , and the decay width @xmath94 is given by @xmath95
. when the @xmath96 potential is used , we find a bound state with a binding energy @xmath97 mev and with a relative radius @xmath98 fm . the mixing ratios of the bound state are 38.8 % for @xmath99 , 6.0 % for @xmath100 and 55.2 % for @xmath101 , as shown in table . [ mixing ] .
interestingly , @xmath101 is the most dominant channel , although the mass of @xmath4 is heavier than the mass of @xmath0 .
this is because the attraction of tensor force is the strongest for @xmath101 channel . for @xmath60 of @xmath90
, we find bound states for both cases when the @xmath34 and @xmath96 potentials are used .
the binding energies are @xmath102 mev for the @xmath34 potential and @xmath103 mev for the @xmath96 potential , and the relative radii are @xmath104 fm and @xmath105 fm , respectively . for the @xmath96 potential ,
the mixing ratios of the bound state are 28.4 % for @xmath106 , 7.7 % for @xmath107 and 63.9 % for @xmath108 . in this case again , @xmath108 is the most dominant component regardless of its heavy mass . for @xmath109 , we find a resonant @xmath0 state for the @xmath34 potential with the resonance energy @xmath110 mev and the half decay width @xmath111 mev as shown in table [ tabledn ] .
in contrast , for the @xmath96 potential , we find a bound @xmath0 state with the binding energy @xmath112 mev and the relative radius @xmath113 fm .
for the bound state , the mixing ratios are 19.8 % for @xmath114 , 62.8 % for @xmath115 , 14.2 % for @xmath116 and 3.2 % for @xmath117 .
thus , @xmath115 is the dominant channel although its mass is heavier than the mass of @xmath114 ; once again a large attraction due to the tensor force is provided . for @xmath60 state of @xmath109
, we obtain bound states for both cases the @xmath34 and @xmath96 potentials are used .
the binding energies are @xmath118 mev for the @xmath34 potential and @xmath119 mev for the @xmath96 potential with relative radii @xmath120 fm and @xmath121 fm , respectively . for @xmath125 , we find resonant states above the threshold .
for @xmath59 , the resonance energies are @xmath126 mev for the @xmath34 potential and @xmath127 mev for the @xmath96 potential .
the half decay widths are @xmath128 mev and @xmath129 mev , respectively . as compared to the cases of @xmath130 and @xmath131 , the results of the @xmath34 and @xmath96 potentials are not very much different , as indicated in fig .
[ figenergyleveli=0 ] .
since the wave functions are extended due to larger orbital angular momenta of @xmath38-wave ( @xmath132 ) and @xmath133-wave ( @xmath134 ) for @xmath122 , the long range potential of the @xmath34 exchange dominates , while the short range potentials from @xmath35 and @xmath65 exchanges are suppressed . for @xmath60 ,
when the @xmath34 potential is used , we find a resonant state whose resonance energy is @xmath135 mev and half decay width is @xmath136 mev , while when the @xmath96 potential is used , we find a loosely bound state of a binding energy @xmath137 mev and a relative radius @xmath138 fm .
the mixing ratios are 71.2 % for @xmath139 , 6.7 % for @xmath140 , 7.5 % for @xmath141 and 14.6 % for @xmath142 .
the most dominant component is @xmath139 , and the second dominant one is @xmath142 . for @xmath143
, we obtain resonant states for both cases when the @xmath34 and @xmath96 potentials are used . for @xmath59 ,
the resonance energies are 153.6 mev for the @xmath34 potential and 160.0 mev for the @xmath96 potential , which are above the @xmath4 threshold .
the corresponding half decay widths are 671.9 mev and 375.4 mev , respectively .
the difference between the results of the @xmath34 and @xmath96 potentials is once again small , due to the same reason as before with large angular momenta .
for @xmath60 state , we also find resonant states above the @xmath5 threshold .
the resonance positions are @xmath144 mev for the @xmath34 potential and @xmath145 mev for the @xmath96 potential .
the corresponding half decay widths are @xmath146 mev and @xmath147 mev , respectively . for @xmath148 , we find resonant states with narrow widths for both cases when the @xmath34 and @xmath96 potentials are used . for @xmath59 , when the @xmath34 potential is used , the resonance energy is @xmath149 mev and the half decay width @xmath150 mev .
when the @xmath96 potential is used , the resonance energy is @xmath151 mev and the half decay width @xmath152 mev . for @xmath60
, we also find resonances whose energies are @xmath153 mev for the @xmath34 potential and @xmath154 mev for the @xmath96 potential , with the corresponding half decay widths @xmath155 mev and @xmath92 mev , respectively .
again , the results for the @xmath34 and @xmath96 potentials are similar . for @xmath156 , we obtain resonances above the @xmath4 and @xmath5 thresholds . for @xmath157 , there exist resonances at 217.7 mev for the @xmath34 potential and at 220.8 mev for the @xmath96 potential , with half decay widths 182.4 mev and 109.1 mev , respectively . for @xmath60 , there also exist resonances whose energies are 85.6 mev for the @xmath34 potential and 87.5 mev for the @xmath96 potential , with half decay widths @xmath158 mev and @xmath159 mev , respectively . finally ,
for @xmath160 , we find no bound nor resonant state . and
@xmath60 for @xmath161 when the @xmath96 potential is used .
the same convention is used as fig.[figenergyleveli=0 ] .
[ figenergyleveli=1 ] , width=566 ] let us move to the states of isospin triplet ( @xmath161 ) .
we summarize the results for @xmath59 and for @xmath60 in table [ tabledn ] and show the energy levels in fig .
[ figenergyleveli=1 ] . as a result
, we find resonant states only for @xmath10 when the @xmath96 potential is employed .
for @xmath59 , the resonance energy is @xmath162 mev and the half decay width is @xmath163 mev . for @xmath60 state
, we obtain the resonance whose energy is @xmath164 mev and half width @xmath165 mev .
the reason that there are not many states in isospin triplet channel can be understood as follows ; as compared to the isosinglet channel , the attractive force in the isotriplet channel is weak due to small isospin factor ; @xmath166 for isosinglet and @xmath167 for isotriplet .
by solving coupled channel schrdinger equations for @xmath36 and @xmath168 channels , we find many bound and resonant states for @xmath169 and few resonances for @xmath161 . for these states ,
the tensor force of the @xmath34 exchange potential plays a significant role to produce them .
when we ignore the @xmath168 channels and solve the schrdinger equation only with @xmath36 channels , we find no bound state nor resonance .
thus , the @xmath36-@xmath168 mixing and the @xmath34 exchange potential play an important role to generate a rich structure of molecular states . the importance of the @xmath34 exchange potential stands out more in the bottom sector .
the small mass difference between @xmath17 and @xmath28 mesons helps to yield the strong attraction because it induces the strong @xmath2-@xmath5 mixing with the tensor force .
furthermore , the @xmath60 has heavier reduced mass and hence the kinetic term is suppressed . for these reasons ,
the binding energies of @xmath60 states are larger than these of @xmath59 states . when we compare the results of the @xmath34 potential and of the @xmath96 potential , they are quite different for @xmath6 , @xmath131 with @xmath169 , where the @xmath35 and @xmath65 exchange potentials become important to produce attraction . for @xmath59 and @xmath60 states ,
both @xmath35 and @xmath65 exchange potentials are attractive , and hence they increase the binding energy significantly . on the other hand , for @xmath122 , @xmath8 and @xmath9 , the results for the @xmath96 potential are similar to those for the @xmath34 potential . for large @xmath170 states ,
the @xmath34 exchange potential plays a dominant role to generate bound and resonant states , while the @xmath35 and @xmath65 exchanges play only a minor one .
this is attributed to the fact that these states have large orbital angular momenta .
if relevant channels include large orbital angular momenta , the wave functions tend to extend spatially , and the long range force , namely the @xmath34 exchange potential , becomes important while the short range force is suppressed . in the present study
, we discuss the molecular structure formed by the @xmath41 bound and resonant states .
however , the hadronic molecular picture is not applicable to deeply bound states with small radii . for such states ,
the constituent hadrons , namely @xmath63 ( @xmath64 ) meson and nucleon , overlap each other , and therefore we have to consider short range effects such as an internal structure of hadrons , channel couplings to conventional three quark states and so on . as a naive criterion for the hadronic molecule , we have shown the relative radii of the bound states as discussed in ref @xcite .
if the size of the bound state is larger than twice of typical hadron size ( namely 1 fm ) , the state could be well described by a molecular structure . for resonant states ,
we identify the states as the hadronic molecule .
according to the criterion , for the @xmath96 potential , only the bound state for @xmath122 of @xmath60 constructs a hadronic molecule , where the relative radius is 1.81 fm .
contrary to this , the bound states for @xmath6 and @xmath131 with @xmath169 which have a small radius and a large binding energy are not described as simple molecules . we need to consider the short range effects including various channel couplings to do more realistic discussions .
we have investigated heavy baryons as hadronic molecules formed by a heavy meson and a nucleon , @xmath59 and @xmath60 . the interaction is given by the meson exchange potential with respecting the heavy quark and chiral symmetries .
many bound states and resonances are found in isosinglet channel both for @xmath59 and @xmath60 systems . in contrast , there are few resonances in isotriplet channel due to small isospin factor . for @xmath6 and @xmath131 ,
we have found deeply bound states far below @xmath0 or @xmath2 threshold with a small radius . in these states ,
the vector meson exchange potential yields a strong attraction . in order to perform more realistic discussions for such compact states , we need to consider further effects from internal structures of constituent hadrons and channel couplings to quark intrinsic states . on the other hand , for @xmath122 , @xmath8 and @xmath9 with large orbital angular momenta ,
the one pion exchange potential is favored and we have obtained loosely bound states and resonances near the thresholds . there
the short range interactions are rather inactive and the long range force of the one pion exchange dominates , where the tensor force plays an important role to generate these states , and they compose the hadronic molecular structure .
it is expected that the hadronic molecular states from a heavy meson and a nucleon are studied in many accelerator facilities ; the @xmath15 meson production in the antiproton - nucleon annihilation process in fair @xcite and the heavy ion collision in rhic and lhc @xcite .
furthermore , the exotic baryons are investigated in j - parc in the coming future .
this work is supported in part by grant - in - aid for scientific research on priority areas `` elucidation of new hadrons with a variety of flavors(e01 : 21105006)''(s .
y. and a. h. ) from the ministry of education , culture , sports , science and technology of japan and grant - in - aid for `` jsps fellows(24 - 3518)''(y .
y. ) from japan society for the promotion of science .
interaction potentials are derived by using the lagrangians in eqs . , , and as shown in ref .
the potentials for the coupled channel systems are given in the matrix form of @xmath171 for @xmath6 and of @xmath172 for the other states . where @xmath182 the @xmath183 and @xmath184 are the isospin matrices for @xmath40 and @xmath19 .
the functions @xmath185 and @xmath186 are given by @xmath187 with @xmath188 , and @xmath189 denotes the form factor given in eq . .
finally , the kinetic terms are given by @xmath203 @xmath204 @xmath205 @xmath206 @xmath207 @xmath208 @xmath209 @xmath210 where @xmath211 , @xmath212 and @xmath213 .
the total hamiltonian is then given by @xmath214 .
9 s. k. choi _ et al . _ [ belle collaboration ] , phys . rev .
* 91 * ( 2003 ) 262001 [ hep - ex/0309032 ] .
m. b. voloshin , prog . part .
phys . * 61 * ( 2008 ) 455 [ arxiv:0711.4556 [ hep - ph ] ] .
e. s. swanson , phys .
* 429 * ( 2006 ) 243 [ hep - ph/0601110 ] .
n. brambilla , s. eidelman , b. k. heltsley , r. vogt , g. t. bodwin , e. eichten , a. d. frawley and a. b. meyer _ et al .
_ , eur .
j. c * 71 * ( 2011 ) 1534 [ arxiv:1010.5827 [ hep - ph ] ] .
a. bondar _ et al .
_ [ belle collaboration ] , phys .
lett . * 108 * ( 2012 ) 122001 [ arxiv:1110.2251 [ hep - ex ] ] .
i. adachi _ et al . _
[ belle collaboration ] , arxiv:1207.4345 [ hep - ex ] .
m. b. voloshin , phys . rev .
d * 84 * ( 2011 ) 031502 [ arxiv:1105.5829 [ hep - ph ] ] .
a. e. bondar , a. garmash , a. i. milstein , r. mizuk and m. b. voloshin , phys .
d * 84 * ( 2011 ) 054010 [ arxiv:1105.4473 [ hep - ph ] ] .
see for example ref .
@xcite for recent review for charmed and bottom baryons as well as up , down , strangeness baryons .
e. klempt and j. -m .
richard , rev .
* 82 * , 1095 ( 2010 ) [ arxiv:0901.2055 [ hep - ph ] ] .
l. a. copley , n. isgur and g. karl , phys .
d * 20 * , 768 ( 1979 ) [ erratum - ibid .
d * 23 * , 817 ( 1981 ) ] .
w. roberts and m. pervin , int .
j. mod .
a * 23 * ( 2008 ) 2817 [ arxiv:0711.2492 [ nucl - th ] ] .
n. isgur and m. b. wise , phys .
b * 232 * ( 1989 ) 113 .
n. isgur and m. b. wise , phys .
* 66 * ( 1991 ) 1130 .
the notation @xmath38 in the present article is opposite in charge conjugation from those in refs .
@xcite .
s. yasui and k. sudoh , phys .
d * 80 * ( 2009 ) 034008 [ arxiv:0906.1452 [ hep - ph ] ] .
y. yamaguchi , s. ohkoda , s. yasui and a. hosaka , phys .
d * 84 * ( 2011 ) 014032 [ arxiv:1105.0734 [ hep - ph ] ] .
y. yamaguchi , s. ohkoda , s. yasui and a. hosaka , phys .
d * 85 * ( 2012 ) 054003 [ arxiv:1111.2691 [ hep - ph ] ] .
k. ikeda , t. myo , k. kato and h. toki , lect .
notes phys .
* 818 * , 165 ( 2010 ) [ arxiv:1007.2474 [ nucl - th ] ] .
n. kaiser , r. brockmann and w. weise , nucl .
a * 625 * ( 1997 ) 758 [ nucl - th/9706045 ] .
n. kaiser , s. gerstendorfer and w. weise , nucl .
phys . a * 637 * ( 1998 ) 395 [ nucl - th/9802071 ] .
r. j. n. phillips , rev .
* 39 * ( 1967 ) 681 . c. b. dover and j. m. richard , phys .
d * 17 * ( 1978 ) 1770 .
s. ohkoda , y. yamaguchi , s. yasui , k. sudoh and a. hosaka , phys .
d * 86 * ( 2012 ) 014004 [ arxiv:1111.2921 [ hep - ph ] ] .
s. ohkoda , y. yamaguchi , s. yasui , k. sudoh and a. hosaka , arxiv:1202.0760 [ hep - ph ] .
t. mizutani and a. ramos , phys .
c * 74 * ( 2006 ) 065201 [ hep - ph/0607257 ] . c. garcia - recio , v.
k. magas , t. mizutani , j. nieves , a. ramos , l. l. salcedo and l. tolos , phys .
d * 79 * ( 2009 ) 054004 [ arxiv:0807.2969 [ hep - ph ] ] .
o. romanets , l. tolos , c. garcia - recio , j. nieves , l. l. salcedo and r. g. e. timmermans , phys .
d * 85 * ( 2012 ) 114032 [ arxiv:1202.2239 [ hep - ph ] ] . c. garcia - recio , j. nieves , o. romanets , l. l. salcedo and l. tolos , arxiv:1210.4755 [ hep - ph ] .
j. hofmann and m. f. m. lutz , nucl .
a * 763 * , 90 ( 2005 ) [ hep - ph/0507071 ] .
j. hofmann and m. f. m. lutz , nucl .
a * 776 * , 17 ( 2006 ) [ hep - ph/0601249 ] .
j. haidenbauer , g. krein , u. -g .
meissner and l. tolos , eur . phys . j. a * 47 * ( 2011 ) 18 [ arxiv:1008.3794 [ nucl - th ] ] .
j. he , and x. liu , phys .
d * 82 * , 114029 ( 2010 ) [ arxiv:1008.1500 [ hep - ph ] ] .
j. -r .
zhang , arxiv:1211.2277 [ hep - ph ] .
j. -r .
zhang , arxiv:1212.5325 [ hep - ph ] .
j. -j .
wu , r. molina , e. oset and b. s. zou , phys .
* 105 * ( 2010 ) 232001 [ arxiv:1007.0573 [ nucl - th ] ] .
j. -j .
wu , r. molina , e. oset and b. s. zou , phys .
c * 84 * ( 2011 ) 015202 [ arxiv:1011.2399 [ nucl - th ] ] .
j. -j .
wu and b. s. zou , phys .
b * 709 * ( 2012 ) 70 [ arxiv:1011.5743 [ hep - ph ] ] .
a. v. manohar and m. b. wise , camb . monogr .
cosmol .
* 10 * , 1 - 191 ( 2000 ) .
r. casalbuoni , a. deandrea , n. di bartolomeo , r. gatto , f. feruglio and g. nardulli , phys .
* 281 * ( 1997 ) 145 [ arxiv : hep - ph/9605342 ] .
m. bando , t. kugo and k. yamawaki , phys .
rept . * 164 * ( 1988 ) 217 .
r. machleidt , k. holinde and c. elster , phys .
* 149 * ( 1987 ) 1 ; r. machleidt , phys .
c * 63 * ( 2001 ) 024001 [ arxiv : nucl - th/0006014 ] .
k. arai and a. t. kruppa , phys .
c * 60 * ( 1999 ) 064315 .
u. wiedner , prog .
* 66 * ( 2011 )
477 [ arxiv:1104.3961 [ hep - ex ] ] . s. cho _ et al . _
[ exhic collaboration ] , phys .
* 106 * ( 2011 )
212001 [ arxiv:1011.0852 [ nucl - th ] ] . s. cho _ et al . _
[ exhic collaboration ] , phys .
c * 84 * ( 2011 ) 064910 [ arxiv:1107.1302 [ nucl - th ] ] . | we study hadronic molecules formed by a heavy meson and a nucleon , @xmath0 and @xmath1 ( @xmath2 and @xmath3 ) systems .
respecting the heavy quark symmetry and chiral symmetry , we consider the @xmath0-@xmath4 ( @xmath2-@xmath5 ) mixing induced by the one boson exchange potential including the tensor force .
we find many bound and resonant states with @xmath6 , @xmath7 , @xmath8 and @xmath9 in isospin singlet channels , while only a few resonant states with @xmath10 in isospin triplet channels .
the analysis of @xmath0 and @xmath1 ( @xmath2 and @xmath3 ) molecules will be useful to study mass spectra of excited charmed ( bottom ) baryons with large angular momenta , when their masses are close to the @xmath0 and @xmath4 ( @xmath2 and @xmath3 ) thresholds . |
edholm s law states that data rates offered by wire - less communication systems will converge with wired ones , where forward rate extrapolation indicates convergence around the year 2030 @xcite . as a result , the only cables that would require removal are the cables supporting power . in turn ,
wireless power ( transfer ) ( wpt ) is rapidly gaining momentum , see fig .
[ fig : wireless_power_popularity ] , and more companies are trying to capitalize on the wireless energy promise , refer for example to witricity @xcite , ubeam @xcite , ossia @xcite , artemis @xcite , energous @xcite , or proxi @xcite . a natural next step is the deployment of networks of wpt nodes ( denoted throughout this work as wptns ) , i.e. deployed and dedicated wpt devices providing power to nearby energy receivers @xcite ( see also an example of a fully energy autonomous wptn in ( * ? ? ?
1 ) ) . wptns are expected to find numerous applications , e.g. in sensing systems ( vide rechargeable sensor network @xcite ) , biology research ( vide insect fly monitoring @xcite or animal inspection @xcite ) , or implantable networks ( vide brain - machine interface @xcite ) . in all the above applications ,
the use of batteries is prohibitive ( in biology - related applications due to induced weight or prohibitive cabling , in implantable applications due to necessity of surgical battery replacement ) , thus wpt is the only long - term viable option . finally , we speculate that due to continuous decrease of energy consumption of embedded platforms ( * ? ? ? * fig .
1 ) , @xcite , within a decade energy provision through wptns will exceed the required energy cost of the above applications .
the necessity of wpt becomes immanent while observing that powering battery - based platforms using energy harvesting alone is not enough ( * ? ? ?
* section i ) .
simply put , ambient energy harvesting does not guarantee sufficient quality of energy provision ( * ? ? ?
3.3 ) ( * ? ? ?
* section i ) .
for example , large - scale london , uk - based rf far field energy harvesting measurements at 270 london underground stations demonstrate that in the best case only 45% of such locations can sustain the required minimum rectifying operation ( * ? ? ?
* table vi ) ( for a single input source , considering digital tv , gsm 900/1800 and 3 g transmitters ) .
however , wpt(n ) has its own inherent deficiency . while there is a huge focus on making wpt(n ) more efficient considering its hardware and physics , its energy conversion coefficient is still low ( * ? ? ?
* section i ) and absolutely not considered to be `` green '' .
this leads to a large carbon footprint of wptn , which will be amplified as the technology spreads to the mass market .
.energy consumption of various wptn devices , see section [ sec : problem_statement ] [ cols=">,>,>,>,>",options="header " , ] atmega328 dc characteristics follow from ( * ? ? ?
* table 29 - 7 ) digi xbee dc characteristics follow from @xcite although following fig .
[ fig : wptn_hardware_model ] we assume that the passive wakeup radio is used to wake up erx from the off state to communication with etx state , in the calculation we nevertheless include the cost of the idle state of the microcontroller and the radio of the erx .
therefore , we calculate the erx power consumption as @xmath1 , where @xmath2 denoting total energy consumption of digi xbee board for @xmath3s total experiment time , composed of transmission energy ( @xmath4 ) , receive energy ( @xmath5 ) , and idle state energy ( @xmath6 ) , respectively , for and @xmath7 and @xmath8 are the number of packets that the erx transmits and receives , and @xmath9 is the energy consumed by the arduino uno board composed of active state energy ( @xmath10 ) , and idle state energy ( @xmath11 ) , respectively ) and ( [ eq : communication_energy_a ] ) are worst case approximations , we assume for simplicity during transmission and reception arduino was simultaneously in sleep state this is due to a small overhead of energy consumption by transmission and reception compared to the total time when the node was idle . ] .
finally , we measure and analytically evaluate protocol - specific parameter , i.e. time to charge a time between transmission of charge request by the erx to the beginning of charge provision by the first - responding etx .
we consider one erx and @xmath12 etxs , as in the experiment .
erx is in charging range of @xmath13 etxs ( @xmath14 ) and in communication range of all @xmath12 etxs . at a given moment of time ( given erx position )
@xmath13 and @xmath12 is fixed .
our goal is to derive formulas for expected time to charge an erx in wptn . [
[ beaconing ] ] beaconing + + + + + + + + + in beaconing implementation we assumed only one round of charging , after which all etxs within the communication range of erx will be turned on .
the duration of this round is @xmath15 , where @xmath16 denotes an uniform distribution from @xmath17 to @xmath18 .
if erx randomly starts to send charge requests in wptn , then time to charge is @xmath19 . a cumulative distribution function ( cdf ) of @xmath20 under the assumption that @xmath21 is not random but constant and equal to its mean value , @xmath22 , is @xmath23 where @xmath24 $ ] .
[ [ probing ] ] probing + + + + + + + probing works in rounds .
successful round starts with charging request , continues to power probing phase and ends in charging phase , in which the protocol stays , successfully charging erx .
however , if the charging is not successful , the protocol goes back to the charging request phase . in the new charging request phase previous etx , that unsuccessfully attempted to charge erx , is excluded from wptn .
therefore , for the first round , there are @xmath12 etx and @xmath13 etx that could charge erx .
if we choose one of @xmath25 etxs that could not charge erx , we exclude it in the next round , which starts with @xmath26 etx and @xmath13 etx that could charge erx . considering random variable @xmath27erx
was charged in round @xmath28 , then @xmath29 for @xmath30 and @xmath31 for @xmath32 $ ] .
length of a successful round , @xmath33 , is different from the unsuccessful round , @xmath34 . considering average values of those variables , @xmath35 and @xmath36 , respectively , as a consequence the cdf of @xmath37 assumes no randomness of @xmath38 and @xmath39 , giving @xmath40 where @xmath41 . [
[ reference - measurement ] ] reference measurement + + + + + + + + + + + + + + + + + + + + + to calculate accuracy for both protocols we need to measure the reference case first .
the reference will denote whether etx should switch on during a particular time to charge erx .
we measure the reference scenario as follows . 1 .
we mark the appearance time , @xmath42 , and disappearance time , @xmath43 , of the erx at each position depicted in fig .
[ fig : experiment_setup ] .
the erx stays at one location for 20s and is allowed to move to a new position within 15s from switch off , respectively . ] ; 2 . during each round of movement of the erx ,
one etx is charging at a time .
after position 10 was reached by the erx as given in fig .
[ fig : experiment_setup ] , a new etx is turned on and a currently charging etx is switched off ; 3 .
we consider the following situation to be correct : if @xmath44@xmath45at a resistor r4 of erx , then etx @xmath46 should switch on to charge the erx at this position , otherwise it should switch off .
each event of voltage crossing threshold is added to a vector @xmath47 @xmath48 $ ] , where @xmath49 and @xmath50 , @xmath51 denote the start ( @xmath52 ) and stop ( @xmath53 ) of the reference charge and @xmath54 denote its successive number .
we note that the voltage sampling period at resistor r4 of erx is 0.1s , just like in experiments in section [ sec : experiment_scenario ] .
[ [ charge - accuracy - metric ] ] charge accuracy metric + + + + + + + + + + + + + + + + + + + + + + having the reference case we can compare the actual working time sequence of each etx @xmath46 ( for each protocol beaconing and probing ) with the reference vector @xmath55 and calculate charge accuracy as @xmath56 where @xmath57 is the corresponding vector of for protocol @xmath58 and @xmath59 denotes xnor operation . to obtain the metrics of interest from the measurements for the benchmark ( freerun protocol ) , see section [ sec : charger_energy_saving_protocol ] , we use the measured values in @xmath60 , described in section [ sec : charge_accuracy_metric ] , to calculate the harvested power at each position depicted in fig . [
fig : experiment_setup ] .
we then sum up the harvested power of four etx , as the theoretical harvested power in the testing scenario where four etxs are switched on all the time .
then we measure the same performance parameters as for the other two protocols .
results are presented in fig .
[ fig : experiment_results_1 ] and fig .
[ fig : experiment_results_2 ] and discussed in the subsequent sections .
note that all experimental results were plotted using matlab s function .
we have performed the experiment for five different communication threshold values , @xmath61 , to measure wptn performance simulating various etx / erx link qualities .
the result is presented in fig .
[ fig : experiment_results_1 ] . refer to fig .
[ fig : exp1_1 ] . for every value of @xmath61 ,
the energy harvested by beaconing is higher than for probing .
this is due to restriction of probing , where at most one etx can charge the erx during a beacon period .
the beaconing protocol allows multiple etx to be turned on at the same time . as the @xmath61increases , the harvested energy decreases for both protocols beaconing and probing .
naturally , the higher the threshold is , the less probability that the etx would be triggered on by neighboring erx .
as expected , the freerun mode has the highest harvested energy in almost every testing point , because all etxs are switched on all the time .
we are now ready to present the fundamental result of this paper , _ proving the `` green '' aspect of the designed wptn_. in addition to harvested energy we show total energy used ( in kj ) by all etx during the whole experiment , refer to fig .
[ fig : exp1_0 ] .
we clearly see the power saved by the beaconing and probing protocol , compared with the freerun mode ( for probing by almost five times ) . since freerun mode switches etx all the time , the energy consumption by etx is highest and constant over @xmath61 .
we discuss the reason behind this gain in detail in subsequent sections . the power consumption of the probing protocol is higher than for beacon in our measurements .
the main reason is that probing needs erx to receive the probing command from the etx , measure the signal strength and send feedback packets to the charger .
the etxs request erx to measure harvested power in every beacon round .
note that probing protocol uses three message types to trigger the charging phase , see protocol [ alg : erx_states ] , while beaconing protocol uses only one message to trigger charging , see protocol [ alg : beaconing_implementation ] . as the @xmath61increases , the power consumption by the erx with probing decreases .
the reason is that the larger the @xmath61is , the less probability that etx will accept charge request messages from the erx .
then larger @xmath61causes less number of etx to associate with the erx on probing , which further causes less number of communication messages at erx .
compared with the beaconing protocol , probing stays on a stable level for each @xmath61 . at 70dbm ,
the efficiency of probing is around three times larger than beaconing . at 50dbm , the efficiency of beaconing increases .
the main reason is that @xmath61 represents the range that etx evaluates whether the erx can be successfully charged or not .
the larger the @xmath61is , the smaller the threshold range is , and the higher energy erx can harvest in the range .
therefore , smaller @xmath61value results in higher efficiency .
the benefit of using smaller threshold @xmath61is that the power transmission efficiency increases .
the drawback is that decreasing range causes decreasing harvested energy , see fig [ fig : exp1_1 ] .
the freerun mode always takes the lowest charging efficiency because it can not estimate whether erx is inside or outside the wptn and what is the possible harvesting power .
if the receiver is outside the wptn or in the area with very low power radio , switching on etxs will waste a lot of power . in the experiment ,
the disappearing time of erx is 15s .
we conjecture that if the disappearing time increases the efficiency of freerun mode will be even lower .
probing protocol keeps relatively high and stable accuracy from 70dbm to 50dbm .
high accuracy causes the high efficiency in probing based protocol as shown in fig [ fig : exp1_2 ] . in probing ,
only one etx is allowed to charge the erx which potentially decreases the accuracy .
we hypothesize that if multiple etxs could exploit probing - like protocol at the same time the accuracy and efficiency could further increase
. the accuracy of beaconing increases as the threshold increases from 70dbm to 50dbm the higher @xmath61value is , the closer the erx must be to an etx for triggering the charging . and
the closer the erx is to the charger , the higher the probability that the etx can charge the erx .
freerun mode naturally has the worst charging accuracy the erx can hardly harvest sufficient energy at certain positions while etxs are continuously switched on . to verify theoretical analysis of time to charge in both protocols
, we have conducted an experiment , where we have placed erx less than 50 cm to each of etx devices ( to ensure erx is within charging range of all etxs ) . to emulate erx being within charging range of a given erx
, we would connect or disconnect powercast device from arduino microcontroller .
for each value of @xmath13 from @xmath62 ( one etx connected ) to @xmath63 ( four etxs connected ) we have performed an experiment where erx appears randomly in the network 50 times .
afterwards , we measured the time it takes from appearing in the network to being charged .
a cdf values of those experiments are presented in fig .
[ fig : time_to_charge_results ] . in this figure experimental results ( solid lines ) are compared against theoretical results ( dashed lines ) . for fig .
[ fig : beaconing_time_n4k1]fig .
[ fig : beaconing_time_n4k4 ] additionally a cdf of @xmath64 is added .
we see that beaconing is faster in reaching etx than probing , however with increasing @xmath13 time to charge for probing becomes very low as well ( almost instant connection after approximately two seconds ) . for beaconing ,
irrespective of number of etx , the time to charge stays constant .
the discrepancy between experimental and numerical results is due to approximation of not taking into account propagation and processing time .
nevertheless the analysis follow the trends of the experimental results in all cases reasonably well .
in this experiment , we have tested the performance of wptn , in which the erx is inside the communication range while outside the charging range of etx .
we change the experiment setup in fig .
[ fig : wptn_setup ] by turning etx 1 and 3 by 180 degrees around its axis .
results in this testing scenario and previous section are depicted as _
back _ , and _
normal_. etx @xmath61is set to 70dbm . in both normal and back condition ,
the harvested energy of beaconing protocol is larger than the probing protocol . by beaconing protocol
the harvested energy in back condition decreases by 45% from the normal condition , which fits with the experiment setup by turning two chargers 180 degrees back .
the decreasing percent in probing from normal to back condition is 32% , which is smaller than in beaconing protocol .
this is because probing selects the etx that can charge energy over the threshold @xmath61 .
as expected , the power consumption of etxs in beacon protocol and freerun mode maintains at the same level in both normal and back conditions .
beacon protocol does not give etxs the function to know whether the charging power is efficiently harvested not .
therefore , as in the back condition , even if the erx is outside the charging range , the charger still switches on as it hears the request message from the erx .
lots of energy is wasted in the back condition by beaconing protocol . in probing protocol
, the ` back ' etxs ( etx 1 and etx 3 ) will switch off , after they evaluate the harvested energy at the erx is low . for both protocols , beaconing and probing ,
the scheduling of the communication in erx are not influenced by the topology of wptn .
so the power consumption in communication in both normal and back conditions are almost the same . by beaconing protocol
, the efficiency in the back condition decreases by 45% from the normal condition .
the decrease in probing protocol from normal to back condition is 25% , which is much smaller than for beaconing .
the smaller decreasing percent is because probing protocol makes etx work only when the charged energy is over @xmath61 .
the 25% decrease mainly comes from the probing period when the etx is turned on and asks the erx to measure the harvested power . in our hardware implementation
the probing period is very short ( 4s ) and we speculate that increasing the probing period can further increase the efficiency of probing based protocol in the back condition .
the accuracy trends in the non - line of sight scenario are the same as in line - of - sight case , sec .
[ sec : accuracy_los ] . also , as in the previous case probing protocol obtains the highest accuracy and confirms its supremacy over the other two considered approaches .
with these results we do believe we open up a new research direction within wptn and we are aware of many points of improvement
. we list the most important ones here . 1
. considering beaconing protocol , the charge request rate should be optimized with respect to power consumption and harvested energy of the erx .
for example , the beacon period can adapt to the number of erxs in the wptn .
since as long as one erx calls the charger to switch on there is no need to trigger etxs for every erx .
considering probing protocol , the probing frequency should be optimized with respect to power consumption of probing as well .
erxs should optimize the probing scheduling considering static and dynamic conditions .
for example , when the erx is static , there is no need to make the receiver measure harvested power with high frequency , since the harvested power is not expected to fluctuate much in such case .
3 . wptn should optimize the combinations of the subset of switched on etxs to take advantage of the constructive signal combined at the erx . measuring all possible combination of the subset of the neighboring etxs to switch on consumes too much time and power at the erx .
thus a novel charge control algorithms are required to enable green wptn with multiple etxs operating at a time .
in this paper we have introduced a new class of charging control protocol for wireless power transfer networks ( wptns)denoted as ` green'that conserve energy of the chargers .
the purpose of such protocol is to maximize three metrics of interest to wptns ( that we introduced here for the first time ) : ( i ) etx charge accuracy , ( ii ) etx charge efficiency and ( iii ) erx harvested power , which in - turn minimize unnecessary uptime of wptn energy transmitters .
we prove that this problem is np - hard .
to solve it we propose two heuristics , denoted as ` beaconing ' ( where energy receivers simply request power from transmitters ) and ` probing ' ( based on the principle of charge feedback from the energy receivers to the energy transmitters ) .
the strength of our protocols lies in making few assumptions about the wtpn environment .
we conclude that each protocol performs its task best in two special cases .
experimentally we show that for large distances between chargers and receivers , probing is more efficient and accurate but harvests less energy than beaconing and has higher communication cost . as the charger to receiver distance increases ,
the efficiency of the beaconing - based protocol increases ( since communication range is well correlated with charging range ) .
we would like to thank dr .
roberto riggio from create - net for providing the energino platform , prof .
joshua smith s group at the university of washington for providing impinj r1000 rfid reader , and to mark stoopman from tu delft for help at the initial stage of the project .
h. dai , g. chen , c. wang , s. wang , x. wu , and f. wu , `` quality of energy provisioning for wireless power transfer , '' _ ieee trans .
parallel distrib . syst . _ , 2014 , accepted for publications .
[ online ] .
available : http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6762897 r. p. wicaksono , g. k. tran , k. sakaguchi , and k. araki , `` wireless grid : enabling ubiquitous sensor networks with wireless energy supply , '' in _ proc .
ieee vtc - spring _ ,
yokohama , japan , may 1518 , 2011 .
s. j. thomas , r. r. harrison , a. leonardo , and m. s. reynolds , `` a battery - free multichannel digital neural / emg telemetry system for flying insects , '' _ ieee trans . biomed .
circuits syst .
_ , vol . 6 , no . 5 , pp . 424435 , oct . 2012 .
j. holleman , d. yeager , r. prasad , j. r. smith , and b. otis , `` neuralwisp : an energy - harvesting wireless neural interface with 1-m range , '' in _ proc .
ieee biocas _ ,
baltimore , md , usa , nov . 2022 , 2008 . j. r. long , w. wu , y. dong , y. zhao , m. a. t. sanduleanu , j. gerrits , and g. van veenendaal , `` energy - efficient wireless front - end concepts for ultra lower power radio , '' in _ proc .
ieee cicc _ ,
san jose , ca , usa , sep . 2124 , 2008 .
michel , y. k. shen , a. p. aiden , a. veres , m. k. gray , w. brockman , the google books team , j. p. pickett , d. hoiberg , d. clancyand peter norvig , j. orwant , s. pinker , m. a. nowak , and e. l. aiden , `` quantitative analysis of culture using millions of digitized books , '' _ science _ , vol .
6014 , pp .
176182 , jan . 2011 .
[ online ] .
available : http://www.sciencemag.org/content/331/6014/176.full.pdf a. massa , g. oliveri , f. viani , and p. rocca , `` array designs for long - distance wireless power transmission : state - of - the - art and innovative solutions , '' _ proc .
vol . 101 , no .
6 , pp . 14641481 , jun . 2013 .
s. timotheou , i. krikidis , g. zheng , and b. ottersten , `` beamforming with miso interference channels with qos and rf energy transfer , '' _ ieee trans .
wireless commun .
_ , vol . 13 , no . 5 , pp . 26462658 , may 2014
k. gomez , r. riggio , t. rasheed , d. miorandi , and f. granelli , `` energino : a hardware and software solution for energy consumption monitoring , '' in _ proc .
international workshop on wireless network measurements _ , paderborn , germany , may 18 , 2012 .
[ online ] .
available : http://www.energino-project.org k. huang and v. k. n. lau , `` enabling wireless power transfer in cellular networks : architecture , modeling and deployment , '' jul .
24 , 2012 , submitted to ieee trans .
wireless commun .
[ online ] .
available : http://arxiv.org/abs/1207.5640 g. yang , c. k. ho , and y. l. guan , `` dynamic resource allocation for multiple - antenna wireless power transfer , '' nov .
17 , 2013 , submitted to ieee trans . signal processing .
[ online ] .
available : http://arxiv.org/abs/1311.4111 d. w. k. ng and r. schober , `` resource allocation for coordinated multipoint networks with wireless information and power transfer , '' in _ proc .
ieee globecom _ ,
austin , tx , usa , dec . 812 , 2014 .
[ online ] .
available : http://arxiv.org/abs/1403.5730 x. liu , p. wang , d. niyato , and z. han , `` resource allocation in wireless networks with rf energy harvesting and transfer , '' _ ieee network _ , may 22 , 2014 , accepted for publication .
[ online ] .
available : http://arxiv.org/abs/1405.5630 m. y. naderi , p. nintanavongsa , and k. r. chowdhury , `` rf - mac : a medium access control protocol for re - chargeable sensor networks powered by wireless energy harvesting , '' _ ieee trans . wireless commun .
_ , vol . 13 , no . 7 , pp . 39263937 , jul
. 2014 .
h. j. visser and r. j. m. vullers , `` rf energy harvesting and transport for wireless sensor network applications : principles and requirements , '' _ proc .
ieee _ , vol
. 101 , no . 6 , pp . 14101423 , jun .
2013 .
x. lu , d. niyato , p. wang , d. i. kim , and z. han , `` wireless charger networking for mobile devices : fundamentals , standards , and applications , '' oct .
31 , 2014 .
[ online ] .
available : http://arxiv.org/abs/1410.6635 m. zhao , j. li , and y. yang , `` a framework of joint mobile energy replenishment and data gathering in wireless rechargeable sensor networks , '' _ ieee trans .
mobile comput .
_ , vol . 3 , no . 12 ,
26892705 , dec . 2014 .
e. casilari , j. m. cano - garcia , and g. campos - garrido , `` modeling of current consumption in 802.15.4/zigbee sensor motes , '' _ sensors _ , vol .
10 , no .
54435468 , jun . 2010 .
[ online ] .
available : http://www.mdpi.com/1424-8220/10/6/5443 \(2015 ) atmel 8-bit microcontroller with 4/8/16/32kbytes in - system programmable flash .
[ online ] .
available : http://www.atmel.com/images/atmel-8271-8-bit-avr-microcontroller-atmega48a-48pa-88a-88pa-168a-168pa-328-328p_datasheet_complete.pdf , `` xbee 802.15.4 module specification , '' 2015 .
[ online ] .
available : http://www.digi.com/products/wireless-wired-embedded-solutions/zigbee-rf-modules/point-multipoint-rfmodules/xbee-series1-module#specs | a wireless power transfer network ( wptn ) aims to support devices with cable - less energy on - demand .
unfortunately , wireless power transfer itself especially through radio frequency radiation rectification is fairly inefficient due to decaying power with distance , antenna polarization , etc .. consequently , idle charging needs to be minimized to reduce already large costs of providing energy to the receivers and at the same time reduce the carbon footprint of wptns . in turn ,
energy saving in a wptn can be boosted by simply switching off the energy transmitter when the received energy is too weak for rectification . therefore in this paper
we propose , and experimentally evaluate , two `` green '' protocols for the control plane of static charger / mobile receiver wptn aimed at optimizing the charger workflow to make wptn green .
those protocols are : ` beaconing ' , where receivers advertise their presence to wptn , and ` probing ' exploiting the receiver feedback from wtpn on the level of received energy .
we demonstrate that both protocols reduce the unnecessary wtpn uptime , however trading it for the reduced energy provision , compared to the base case of ` wptn charger always on ' .
for example , our system ( in our experiments ) saves at most @xmath080% of energy and increases 5.5 times the efficiency with only @xmath017% less energy possibly harvested . |
. , bias voltage along @xmath58 and tmr along @xmath59 directions . part ( a ) corresponds to magnetic field along the * * ha * * and part ( b ) along * ia . *
( c ) schematic presentation of the model with a single spot and ( d ) the corresponding energy profile .
( e ) calculated conductance ( @xmath46 ) for dmtj with a single spot in p state for different spot dimensions .
r vs. v for dmtj with multiple spots in the p state for different average spot dimensions .
, width=377 ] | we investigate electronic transport in epitaxial fe(100)/mgo / fe / mgo / fe double magnetic tunnel junctions with soft barrier breakdown ( hot spots ) .
specificity of these junctions are _ continious middle layer _ and nitrogen doping of the mgo barriers * * which provides soft breakdown at biases about 0.5v . in the junctions with hot spots
we observe quasi - periodic changes in the resistance as a function of bias voltage which point out formation of quantum well states in the middle fe continuous free layer .
the room - temperature oscillations have been observed in both parallel and antiparallel magnetic configurations and for both bias polarizations . * * a simple model of tunneling through hot spots in the double barrier magnetic junction is proposed to explain qualitatively this effect .
recent theoretical predictions @xcite followed by experimental observations of coherent tunneling in magnetic tunnel junctions ( mtjs ) with mgo barriers @xcite * * have busted research of @xmath0 mtjs @xcite ( here @xmath1 and @xmath2 are ferromagnetic layers and @xmath3 - insulating barrier ) , and opened new perspectives of their applications in spintronic devices . for thin enough electrodes , electron tunneling may reveal resonant features due to quantum well states ( qws ) .
early studies explored the simplest way to realize resonant tunneling by growing a thin nonmagnetic layer ( nm ) between ferromagnetic electrode and barrier in standard mtj structures @xmath4 @xcite .
however , resonant tunneling in double mtjs ( dmtjs ) @xmath5 ( @xmath6 is central layer ) may have advantages in comparison with the standard mtjs , mainly due to their enhanced tunnelling magnetoresistance ( tmr ) @xcite and resonant spin - torque effects @xcite .
tmr in dmtjs only weakly varies at low bias voltages @xcite , which is crucial for applications .
last but not least , the current driven magnetization reversal in dmtjs occurs at relatively low current densities @xcite . in dmtjs
the qws can strongly influence electron transport only if the _
_ f__@xmath7 layer has submonolayer roughness and its thickness exceeds 1 nm minimizing coulomb blockade ( cb ) effects @xcite .
on the other hand , the _
_ f__@xmath7 layer should also be thin enough so that energy separation of qws substantially exceeds the thermal energy .
these conditions are hardly fulfilled in the macroscopic dmtjs @xcite , where resonant tunneling was not observed mainly due to the absence of the atomically flat surfaces over entire junction lateral dimensions .
evidence for a local tunneling through qws in the central fe layer in @xmath8 junctions was provided by iovan et al .
@xcite using the point contact technique .
* * recently , nozaki et al .
@xcite reported on resonant tunneling effects in macroscopic * * dmtjs with fe nano islands incorporated into the thick mgo barrier .
their results have been interpreted as due to combined qws @xcite and cb effects .
this letter reports on the detailed study of electron transport in epitaxial macroscopic fe@xmath9/mgo / fe / mgo / fe junctions with _ continuous middle fe layer _ and current flowing through the hot spots. oscillatory conductance and tmr with applied voltage present clear signatures of the local coherent tunneling through qws .
the oscillations have been observed in _ both parallel and antiparallel magnetic configurations and for both bias polarizations _ , contrary to ref.@xcite where they were seen only in the parallel configuration and for one current orientation .
moreover , we observe oscillations in the room temperature ( rt ) regime , while in ref.@xcite they were seen mainly at low temperatures . finally , the technique we used for their observation is also different and makes use of some features of breakdown junctions , which allow to observe _ quantum oscillations in a continuous layer_. our results have been explained within a simple model which assumes formation of single or multiple hot spots .
the junctions under study have the following structure : mgo//mgo@xmath10 /v@xmath11 /cr@xmath12 /co@xmath13 /fe@xmath14 /mgo@xmath15 /fe@xmath13 /mgo@xmath15 /fe@xmath10 /co@xmath16
/pd@xmath10 /au@xmath10 .
the mtj stacks were grown by molecular - beam epitaxy with the base pressure of 5@xmath1710@xmath18 torr in the presence of atomic nitrogen . the mgo barrier and the fe layers
were grown at rt .
the fe was annealed to 450@xmath19 for flattening .
the high resolution cross - sectional tem images ( fig .
1@xmath20 ) generally corroborate the good structural quality and homogeneity of dmtjs .
since the tem images are deduced from inverse fourier transform of a diffraction pattern integrated across the sample thickness , the amorphous zones , if present , add a diffuse backgound to the diffraction pattern of the single crystal zones .
more details of standard double barrier samples growth may be found in @xcite .
the specificity of the samples studied here is a nitrogen doping of all the layers achieved during the growth , with a nitrogen concentration roughly estimated to less than 2% .
although further studies are needed to determine the concentration of nitrogen inside mgo , recent report @xcite * * revealed * * that mgo barrier in mtjs may be doped up to 2.5% of nitrogen without changes in the crystalline structure . in our samples ,
the structural analysis by rheed , auger spectroscopy and magnetometry demonstrate that , despite the nitrogen doping ( as evidenced from auger , see fig.1b ) , the structural and magnetic properties are not affected .
the rheed patterns of fe and mgo ( not presented here ) are identical to those of nitrogen - free samples .
what is important for the studies presented in this paper is that , from electrical transport point of view , the barrier doping by nitrogen is responsible for local soft dielectric breakdown @xcite with reduced breakdown voltage .
this may be expected to keep barrier and central fe electrode compositions nearly unchanged .
after the mbe growth of the multilayer stack , the mtj structures were patterned to 10@xmath21 by uv lithography and ar - ion etching , step - by - step controlled in situ by auger .
figure 2@xmath20 shows the rt magnetization measured with magnetic field along the fe easy ( 100 ) axis ( * ea)*. one observes well defined transitions of the three distinct fe layers .
this is indirect indication of epitaxy and conservation of magnetic properties ( fourfold anisotropy ) for nitrogen doped fe .
the zero bias tmr is close to 30% ( fig.2b ) .
application of the bias exceeding roughly 500 mv leads to breakdown of the junctions , which in turn reduces zero - bias tmr down to about 4% ( see figs .
2@xmath22 .
two other observations indicate indirectly that this breakdown decreases the effective mgo barrier height most probably by transforming locally nearly crystalline mgo regions near a hot - spot into amorphous .
firstly , the coercive field of the central free fe layer remains unchanged after breakdown ( h@xmath7=45 oe ) , while coercive fields of the upper and bottom electrodes increase substantially about 50% .
these changes are most probably due to hardening of the fe / co interfaces by high current density close to the hot - spot. the second indication for the possible amorphization of the mgo barrier during the breakdown with intact middle electrode is our experimental observation of the signature of qws in the electron transport in some of the broken dmtjs which we discuss below . we have carried out detailed study of the rt resistance as a function of magnetic field with three different field orientations : * ea * , the hard ( 110 ) axis -*ha * , and in intermediate axis ( * ia * ) situated approximately at 10 degrees from the * ea*. here we present results obtained at bias voltages up to 1.5v and with the steps of 25mv for those broken dmtjs which showed reproducible signatures of the changes in the resistance with bias ( fig.3 ) .
the zero bias tmr varies substantially along the three mentioned directions .
as expected , tmr is reduced along * ia * in comparison with * ea * and * ha * directions , where transitions in all three magnetic layers as indicated by arrows are reflected in the tmr . _
_ we shall concentrate further on the electron transport data obtained with field along * ha * and * ia * directions , where the largest relative changes in tmr vs. bias were observed .
figure 3@xmath20 shows typical bias dependence of the resistance @xmath23 for parallel ( p ) and antiparallel ( ap ) states with magnetic field along * ha*. one observes oscillatory behavior of r and tmr with a period close to 150mv in both p and ap states .
it is important to note that these oscillations , more clearly resolved for negative bias when current flows from the upper to bottom electrodes ( figs .
3@xmath24 ) , have period that is in reasonable agreement with the predictions by wang _
et al .
_ @xcite .
figures 3a , c mark with arrows the majority spin qws energies calculated within about 1v above ( red ) and below ( black ) the fermi energy @xcite .
although the absolute values of tmr measured with the field along * ia * are reduced in comparison with those for the fields along * ea * and * ha * , the relative changes of the tmr with bias are substantially enhanced ( see fig .
3__b,__@xmath25 ) ) . in order to understand this effect
we remind that the measurements of tunnel resistance with the field along * ia * are usually observed @xcite * * to be most sensitive to small variations in the angle between magnetizations of the fixed and free layers in comparison with * ea * and * ha * configurations .
we suggest here that the strongest relative changes in tmr(v ) for the * ia * configuration could be a consequence of local spin - torque effects which are predicted to be enhanced with intermediate alignment of the ferromagnetic layers @xcite . figures 4 @xmath26 represent 3d plots of tmr vs magnetic field and bias with magnetic fields applied along * ha * and * ia * respectively .
dependence of tmr on bias is observed to be more asymmetric with field along * ia * ( figs .
4 @xmath26 ) , which is in agreement with possible influence of local spin torque effects in the breakdown regions .
let us now discuss physical mechanisms which could be behind the main experimental findings .
before breakdown , the current is roughly uniform across the junction area and weak interface disorder might introduce decoherence suppressing the effects due to qws . the soft breakdown of the doped mgo may create defects and local amorphization which locally reduce the mgo barriers leading to hot spots which connect fe and fe / co leads with the central fe layer . with hot spots of sufficiently small lateral dimensions , electrons tunneling to the central layer
can sample well defined structure of qws due to lack of decoherence .
the remaining part of the macroscopic dmtj provides then some averaged featureless background signal .
the feasibility of the above scenario is qualitatively supported by the good correspondence of the observed periodic variations in r vs.v and the theory taking into account qws formed within 4.6 nm thick central fe electrode @xcite ( see figure 3a , c ) . in order to describe the observed features
, we consider a dmtj ( fig .
4c , d ) with barriers including a number of hot spots. the average conductance of the structure is @xcite @xmath27where @xmath28 is the conductance due to a single @xmath29-th spot , @xmath30 is the probability of realization of a certain configuration of the @xmath29-th spot , and @xmath31 is a set of parameters characterizing this configuration .
let us assume that a particular hot spot is characterized by its lateral dimension @xmath20 .
the current through the spot can be then calculated as @xmath32 where @xmath33 are the normal and lateral wavevector components of the incoming wave ( layer 1 ) , @xmath34 and @xmath35 are the normal components of the wavevectors in layers 2 and 3 , respectively .
here we assumed that the in - plane component of @xmath36 is conserved for tunneling from layer 2 to 3 , whereas for tunneling through the spot there is no conservation of the in - plane component due to broken translational symmetry , and the scattering can be described by an angle distribution function @xmath37 . in our calculations
we use the approximation @xmath38 , which means that scattering to the state with large in - plane component @xmath39 is effectively suppressed .
equation ( 1 ) includes transmission probabilities @xmath40 and @xmath41 for tunneling from the layer 1 to 2 , and from 2 to 3 , respectively .
the sum over @xmath34 runs over discrete values satisfying the quantization condition @xmath42 .
it should be emphasized that this condition is related only to thickness @xmath43 of the layer 2 and is the same for any other hot spot. calculating the integral over @xmath44 we find @xmath45 the conductance @xmath46 as a function of bias @xmath47 for a single spot is presented in fig .
4e for different values of @xmath20 .
we assumed @xmath48 nm , the barrier width @xmath49 nm , @xmath50 ev , and the barrier height @xmath51 ev . as we see , the oscillation peaks related to the level quantization in the layer 2 are more pronounced for wide spots , and they are effectively damped for small @xmath20 .
this is because the small spot enables tunneling with nonconserved in - plane component of the wave vector .
taking into account tunneling from many different spots , we obtain qualitatively the same picture corresponding to a mean value of @xmath20 , and proportional to the number of spots .
the total conductance of the structure includes a constant non - oscillating part , @xmath52 , related to the tunneling without spots . in fig .
4f we present resistance calculated as @xmath53 , where @xmath54 is the number of spots and @xmath55 is the mean value of @xmath20 .
we note that the variation of barrier heights at the spot does not affect the position of peaks and does not change the shape of peaks , changing only the amplitude .
thus , averaging over randomly distributed @xmath20 and barrier heights @xmath56 gives a picture like for a single spot with some mean values of @xmath20 and @xmath56 .
the applicability of the above model requires a number of conditions .
although location of the oscillation peaks does not depend on the spot dimension ( @xmath20 ) , the hot spot should not exceed average dimensions over which central electrode is atomically flat .
also , as seen from fig.4e , f , decreasing the parameter @xmath20 makes oscillations less visible ( damped ) .
therefore we expect the hot spots to be roughly of _ nm _ lateral size .
secondly , if magnetic moment of one or two layers is reversed ( ap alignment , ) the resistance becomes very large because the minority @xmath57 band will be displaced well above the fermi energy .
however realistically speaking the hot spot region of the mgo barrier may be not fully epitaxial either due to structural defects and/or nitrogen doping .
this can strongly reduce spin filtering by the @xmath57 band in the ap alignment , substantially suppressing tmr in real dmtjs with hot spots. in conclusion , we have presented evidence for local resonant tunneling through quantum well states in the middle continuous free layer of a double magnetic tunnel junction .
the oscillations have been observed at room temperature in both parallel and antiparallel magnetic configurations and for both bias polarizations .
owing to specific features of the breakdown junctions , we were able to observe quantum well states in continuous magnetic layers . to observe similar effects in the nitrogen free dmtj
, the junction area should be smaller than the size of terraces of the fe central layer .
understanding of electron transport in magnetic tunnel junctions with defects and hot spots is of great importance both from fundamental and applied points of view .
recent reports link spin torque oscillations with record low bandwidth to the presence of defects and hot spots inside the mgo barrier of mtjs @xcite .
we thank s. parkin and j. martinek for stimulating discussions , f.greullet for help with experiment and e.snoeck for taking tem images .
the work was supported by spanish micinn ( mat2009 - 10139 , mat2006 - 28183-e , csd2007 - 0010 , integrated action project fr2009 - 0010 ) and cam ( p2009/mat-1726 ) , by the fct grant ptdc / fis/70843/2006 in portugal and by polish ministry of science and higher education as a research project in years 2007 2010 .
99 w. butler , et .
al , _ phys .
rev .
b _ * 63 * , 054416 ( 2001 ) . j. mathon , et .
al , _ phys .
rev .
b _ * 63 * , 220403(r ) ( 2001 ) . m. bowen , et .
al , _ appl.phys.lett_. * 79 * , 1665 ( 2001 ) .
j. faure - vincent , et .
al , _ appl.phys.lett_. * 82 * , 4507 ( 2004 ) . s. parkin , et .
al , _ nat.mater _ * 3 * , 862 ( 2004 ) .
yuasa , et .
al , _ nat.mater _ * 3 * , 868 ( 2004 ) . r. guerrero , et .
al , _ appl.phys.lett_. * 91 * , 132504 ( 2007 ) .
j. moodera , et .
al , _ phys .
rev .
lett _ * 74 * , 3273 ( 1995 ) .
p. leclair , et .
al , _ phys . rev .
lett _ * 84 * , 2933 ( 2000 ) .
yuasa , et .
al , _ science _ * 297 * , 234 ( 2002 ) .
f. greullet , et .
al , _ phys .
rev .
lett _ * 99 * , 187202 ( 2007 ) .
j. barnas , et .
al , _ phys .
rev .
lett _ * 80 * , 1058 ( 1998 ) .
a. petukhov , et .
al , _ phys .
rev .
lett _
* 89 * , 107205 ( 2002 ) .
a. kalitsov , et .
al , _ phys .
rev .
lett _ * 93 * , 046603 ( 2004 ) .
a. vedyaev , et .
al . , _ phys .
lett . a _ * 355 * , 243 ( 2006 ) .
i. theodonis , et .
al , _ phys.rev.b _ * 76 * , 224406 ( 2007 ) . m. watanabe , et .
al , _ appl.phys.lett . _ * 92 * , 082506 ( 2008 ) .
j. peralta - ramos , et .
al , _ phys .
rev .
b _ * 78 * , 024430 ( 2008 ) .
g. feng , et .
al , _ j. appl .
phys .
_ * 105 * , 07c926 ( 2009 ) . c. tiusan , et .
al , _ j.phys . :
cond . mat .
_ * 19 * , 165201 ( 2007 ) .
g. feng , et .
al , _ appl.phys.lett . _
* 89 * , 162501 ( 2006 ) .
a. iovan , et .
al , _ nano letters _ * 8 * , 805 ( 2008 ) . t. nozaki , et .
al , _ phys.rev.lett . _
* 96 * , 027208 ( 2006 ) .
y. wang , et .
al , _ phys .
rev .
lett _ * 97 * , 087210 ( 2006 ) .
s. parkin , _
aps2009 march meeting _ ,
* z32*.00001 ( 2009 ) .
d. dimitrov , et .
al , _ appl.phys.lett_. * 94 * , 123110 ( 2009 ) . x. liu , et .
al , _
j. appl . phys_.
* 92 , * 8 ( 2002 ) .
i. lifshitz , et .
al , _ sov .
. jetp _ * 50 * , 499 ( 1979 ) .
d. houssameddine , et.al , _ prl _ * 102 * , 257202 ( 2009 ) . |
non - decoupling effects of heavy particles in the low energy observables have been for a long time an indirect but crucial test for the discovery of new particles .
the most recent example is the top quark , which contributes significantly to the observables that measure the electroweak ( ew ) radiative corrections , as for instance @xmath0 @xcite .
in particular , the latest global fit of the electroweak parameters from lep data gives a quite strong constraint for the allowed top mass value @xcite .
this indirect search supports the recently announced evidence of @xmath1 production at cdf @xcite and will certainly contribute to get a final confirmation of the existence of the quark top .
the non - decoupling effects of the higgs boson are , however , weaker than in the case of the top quark due to the screening theorem @xcite . according to this theorem ,
the sensitivity in the low energy observables to the higgs boson mass is at most logarithmic at one loop .
this fact has made in the past a difficult task to disentangle the higgs from the dominant fermionic effects and , therefore , no significant bound on the higgs mass has been obtained so far from the analysis of lep data .
this situation , however , will likely improve in the near future due to the increasing precision of electroweak measurements and the forthcoming cdf data .
a confirmation of the top quark is still needed , as well as a precise measurement of the top quark mass , but the important point is that it is now beginning to be plausible the search for evidence of the higgs boson in electroweak radiative corrections .
on the other hand , regarding the theoretical aspects , the leading logarithmic higgs mass effects in the low energy observables are known up to one - loop level since the pioneering works by appelquist and bernard @xcite and by longhitano @xcite .
their strategy was to use the symmetry properties of the @xmath2 gauged non - linear @xmath3-model ( gnl ) @xcite along with a systematic power - counting analysis to provide a list of these logarithmic higgs mass dependent terms .
however , at present , it is known that these leading logarithmic terms are not sufficient to discriminate a heavy higgs possibility from an alternative symmetry breaking scenario to which one requires to respect the same sm symmetries .
these logarithmic contributions are a consequence of the general gauge and custodial symmetry requirements of the low energy structure of ew interactions and therefore they will be the same irrespective of the particular choice for the breaking dynamics , with the higgs mass being replaced by some alternative physical mass .
thus , if one wants to reveal the nature of the symmetry breaking from low energy observables , one has to go beyond the leading logarithmic effects . in this paper , we present the complete calculation of the leading and next - to - leading non - decoupling effects of a heavy higgs boson in the sm to one loop level . these genuine higgs boson effects can not be obtained using the gnl as in @xcite , but must be calculated directly from the evaluation of the one loop diagrams in the sm .
futhermore , in order to classify in a systematic way those effects , we will use here the electroweak chiral lagrangian ( echl ) @xcite .
our approach is based on effective field theory methods , in which the non - decoupling effects of a heavy higgs boson are represented , at energies below the higgs mass , by certain set of gauge invariant effective operators of the echl .
the echl is basically a non - linear sigma model coupled to the @xmath2 gauge fields , where the higgs field has been removed from the physical spectrum of the theory .
the model respects the fundamental symmetries of the sm , namely @xmath2 gauge invariance spontaneously broken to @xmath4 and the custodial symmetry @xmath5 of the pure scalar sector , but does not include explicitely a particular dynamics for the symmetry breaking @xcite-@xcite .
although this higgs - less parametrization of ew interactions is non - renormalizable , the theory can be rendered finite to one loop by adding gauge invariant operators up to dimension four .
these effective operators parametrize the low energy effects of the underlying fundamental dynamics of the symmetry breaking .
in particular , the echl can be regarded as an effective theory of the sm in which the higgs field has been integrated out , and its effects at energies well below the higgs mass are parametrized by the chiral effective operators @xcite .
we believe that this kind of approach may be interesting for several reasons .
first of all , it provides a gauge invariant way of separating the non - decoupling higgs boson effects from the rest of the ew radiative corrections . on the other hand ,
the echl is a general framework in which one can analyze the low energy effects not only of a heavy higgs in the sm , but of more general breaking dynamics characterized by the absence of ligth modes @xcite .
it is then desirable to have the echl that parametrizes a sm higgs as a fundamental reference model . in our previous work @xcite ,
we discussed a general procedure to obtain the echl operators by matching the sm predictions in the limit of large higgs mass with the predictions from the echl to one loop order .
the subset of echl operators involved in the two and three - point green s functions of the gauge fields were also obtained there . in this work ,
we complete the calculation of the echl operators by analyzing the four - point green s functions for gauge fields .
we will also discuss here in more detail the relation between the renormalization of both the sm and the effective theory , which is crucial for understanding the true meaning of the effective operators .
we will show with some explicit examples how to calculate observables in the echl parametrization of ew interactions .
the paper is organized as follows .
section two is a survey of the electroweak chiral lagrangian approach , in which we also include a brief discussion of the formal renormalization procedure of the effective theory and the matching conditions .
section three is devoted to describe the computation of the sm four - point green s functions in the large @xmath6 limit .
the renormalization prescription chosen for the sm is fixed in this section to be the on - shell scheme .
the matching equations will be solved in section four where we present the results for the complete set of echl bare parameters .
section five is devoted to discuss the dependence of the bare echl parameters on the renormalization prescription fixed for the underlying sm .
we explain in section six the relation between the bare echl parameters and the renormalized echl parameters for different renormalizations of the effective theory . in section seven
we explain how to compute observables to one loop with the echl and demonstrate by choosing some simple observables that the result agrees with previous calculations in the literature .
finally , section eight is devoted to the conclusions .
the echl is the most simple effective theory of ew interactions that parametrizes the physics of the @xmath2 breaking dynamics at low energies .
the assumption made in this approach is that , whatever the @xmath2 breaking interactions may be , the particles involved in the symmetry breaking are heavier than the w and z bosons .
the echl is then a low energy formulation of ew interactions which contains just the `` light '' gauge and would - be goldstone fields , satisfying the basic requirement of @xmath2 gauge invariance spontaneously broken to @xmath7 : @xmath8 its basic structure is a gauged non - linear sigma model @xmath9 , where a non - linear parametrization of the would - be goldstone bosons is coupled to the @xmath2 gauge fields @xmath10 + \frac{1}{2}\ ; tr\left [ \wtd \wtu + \btd \btu \right ] + \cl_{\rm r_\xi } + \fpnl , \label{nll}\ ] ] where the bosonic fields have been parametrized as @xmath11 and the covariant derivative and the field strength tensors are defined as @xmath12 \wtd & \equiv & \partial_\mu \wh_\nu - \partial_\nu \wh_\mu - g [ \wh_\mu , \wh_\nu],\nonumber\\[2 mm ] \btd & \equiv & \partial_\mu \bh_\nu - \partial_\nu \bh_\mu . \label{ften}\end{aligned}\ ] ] the physical fields are given by @xmath13 z_\mu & = & c\ ; w^3_\mu - s\ ; b_\mu \nonumber , \\[2 mm ] a_\mu & = & s\ ; w^3_\mu + c\ ; b_\mu,\end{aligned}\ ] ] where @xmath14 and the weak angle is defined by @xmath15 .
the second term in eq.([ecl ] ) includes the complete set of @xmath2 and cp invariant operators up to dimension four proportional to @xmath16 that is @xmath17 conserving but @xmath18 and @xmath19 violating .
it is not relevant in case of absence of fermion contributions and will not be considered here . ]
that were classified by longhitano in @xcite : .
notice that the definition of @xmath20 is different here than in @xcite .
] @xmath21 ^ 2 \nonumber\\[2 mm ] \cl_{1 } & = & a_1 \frac{i g g'}{2 } b_{\mu\nu } tr\left ( t \wtu \right ) \nonumber\\[2 mm ] \cl_{2 } & = & a_2 \frac{i g'}{2 } b_{\mu\nu } tr\left ( t [ v^\mu , v^\nu ] \right ) \nonumber\\[2 mm ] \cl_{3 } & = & a_3 g tr\left ( \wtd [ v^\mu , v^\nu ] \right ) \nonumber\\[2 mm ] \cl_{4 } & = & a_4 \left [ tr\left ( v_\mu v_\nu \right ) \right]^2 \nonumber\\[2 mm ] \cl_{5 } & = & a_5 \left [ tr\left ( v_\mu v^\mu \right ) \right]^2 \nonumber\\[2 mm ] \cl_{6 } & = & a_6 tr\left ( v_\mu v_\nu \right ) tr\left ( t v^\mu \right ) tr\left ( t v^\nu \right)\nonumber\\[2 mm ] \cl_{7 } & = & a_7 tr\left ( v_\mu v^\mu \right ) \left [ tr\left ( t v^\nu \right ) \right]^2\nonumber\\[2 mm ] \cl_{8 } & = & a_8 \frac{g^2}{4 } \left [ tr\left ( t \wtd \right ) \right]^2 \nonumber\\[2 mm ] \cl_{9 } & = & a_9 \frac{g}{2 } tr\left ( t \wtd \right ) tr\left ( t [ v^\mu , v^\nu ] \right ) \nonumber\\[2 mm ] \cl_{10 } & = & a_{10 } \left [ tr\left ( t v_\mu \right ) tr\left ( t v_\nu \right ) \right]^2 \nonumber\\[2 mm ] \cl_{11 } & = & a_{11 } tr\left ( ( d_\mu v^\mu ) ^2 \right ) \nonumber\\[2 mm ] \cl_{12 } & = & a_{12 } tr\left ( t d_\mu d_\nu v^\nu \right ) tr \left ( t v^\mu \right)\nonumber\\[2 mm ] \cl_{13 } & = & a_{13 } \frac{1}{2 } \left [ tr \left ( t d_\mu v_\nu \right ) \right]^2 \label{li}\end{aligned}\ ] ] where the basic building blocks are defined as @xmath22 the above particular base of invariants can be transformed into a new one with only 11 independent structures by making use of the classical equations of motion @xcite .
the new effective lagrangian is then given in terms of a new set of chiral parameters @xmath23 given by : @xmath24 , @xmath25 , @xmath26 , @xmath27 , @xmath28 , @xmath29 , @xmath30 ; @xmath31 . both effective lagrangians , however , will give rise to the same physical on - shell amplitudes . in this work , since we will not restrict ourselves to calculate on - shell matrix elements , we keep the complete basis given in eq.([li ] ) .
we will work in a generic r@xmath32 gauge , the gauge fixing term @xmath33 and the faddeev - popov lagrangian @xmath34 in eq.([ecl ] ) were given in our previous work @xcite .
we refer the reader to this work for the detailed formulas and a discussion on these terms .
it is worth just recalling here that @xmath34 does not coincide with the usual faddeev - popov lagrangian of the sm due to the non - linearity of the would - be goldstone bosons under infinitesimal @xmath2 transformations .
furthermore , because of the non - linear realization of the gauge symmetry , some of the couplings in @xmath9 have different feynman rules than in the sm .
we collect in fig.(1 ) the subset of them that are relevant for the present calculation .
it is also important to mention that in a r@xmath32 gauge , the complete electroweak chiral lagrangian must be brs invariant and include also effective operators involving the ghost fields .
longhitano@xcite showed , however , that the subset of operators given above is sufficient to absorb the divergences of the gauged non - linear sigma model if the landau gauge is chosen .
as it will be shown later , we have demonstrated in this work that the set of operators given in eq.([li ] ) is also enough , in a general r@xmath32 gauge , to render finite the two , three and four - point green s functions with external gauge fields . the non - linear sigma model in eq.([nll ] ) is not a renormalizable theory , as increasing the number of loops in a calculation implies the appearence of new divergent structures of higher and higher dimension .
however , the echl is an effective theory that can be renormalized order by order in the loop expansion . in particular , at one loop order , the new divergences generated by a one loop calculation with @xmath9 can be absorved into redefinitions of the effective operators given in eq.([li ] ) @xcite .
therefore , one can obtain finite renormalized green s functions if one makes a suitable redefinition of the fields and parameters of the echl , among which the chiral parameters @xmath35 must be included @xcite- @xcite .
formally , one defines the renormalized quantities in the effective theory by the following relations @xmath36 \vec{w}_\mu^b \ ; = \ ; \widehat{z}_w^{1/2}\ ; \vec{w}_\mu , & g^b \ ; = \ ; \widehat{z}_w^{-1/2}\ ; ( g - \widehat{\delta g } ) , \\[2 mm ] \vec{\pi}^b \ ; = \ ; \widehat{z}_\phi^{1/2}\ ; \vec{\pi } , & v^b \ ; = \ ; \widehat{z}_\phi^{1/2}\ ; ( v -\widehat { \delta v}),\\[2 mm ] \xi_b^b \ ; = \ ; \xi_b\ ; ( 1 + \widehat{\delta \xi}_b ) , \hspace{1 cm } & \xi_w^b \ ; = \ ; \xi_w \ ; ( 1 + \widehat{\delta \xi}_w ) , \end{array}$ ] @xmath37 where the renormalization constants of the effective theory are @xmath38 and the superscript b denotes bare quantities .
we use the hatted notation to distinguish counterterms and green s functions in the effective theory from the corresponding quantities in the sm .
the 1pi renormalized green s functions of the effective theory to one loop will be generically denoted by @xmath39 where the superscript r denote renormalized function and the superscripts t , c and l denote the tree level , counterterm and loop contributions respectively .
we will discuss in section 6 the on - shell renormalization of the effective theory , giving explicit expressions for the counterterms introduced in eq.([ret ] ) . for the moment , in order to discuss the matching procedure , we will treat the counterterm contributions to the renormalized functions of eq.([gfe ] ) just at a formal level .
we would like to focus now our attention on the chiral parameters .
once a particular renormalization scheme has been chosen to fix the counterterms of the effective theory , the renormalized @xmath40 parameters remain as free parameters that can not be determined within the framework of the low energy effective theory .
the values of the renormalized chiral parameters can be constrained from the experiment , as they are directly related to different observables in scattering processes @xcite,@xcite-@xcite and in precision electroweak measurements ( @xcite-@xcite,@xcite , see also section 7 ) ; but to have any theoretical insight on their values , one has to relate the effective theory with a particular underlying fundamental dynamics of the symmetry breaking .
if the underlying fundamental theory is the standard model with a heavy higgs , the chiral parameters can be determined by matching the predictions of the sm in the large higgs mass limit and those of the echl , at one loop level . by heavy higgs we mean a higgs mass much larger than any external momenta and light particle masses ( @xmath41 ) so that one can make a low energy expansion , but smaller than say 1 tev , so that a perturbative loop calculation is reliable .
we will impose here the strongest form of matching @xcite by requiring that all renormalized one - light - particle irreducible ( 1lpi ) green s functions are the same in both theories at scales @xmath42 .
the 1lpi functions are , by definition , the green s functions with only light particles in the external legs and whose contributing graphs can not be disconnected by cutting a single light particle line .
this matching condition is equivalent to the equality of the light particle effective action in the two descriptions .
some other forms of matching have been discussed in the literature , by requiring the equality of the two theories at the level of the physical scattering amplitudes @xcite or connected green s functions @xcite .
these requirements , however , complicate the calculation unnecesarily while give at the end the same results for the physical observables .
there is also some discussion in the literature @xcite on the dependence of the green s functions on the parametrization chosen for the would - be goldstone bosons .
we will fix here the particular parametrizations of eq.([fpar ] ) in the effective theory and eq.([smpar ] ) in the sm .
of course , the physical observables will not depend on this particular choice . in order to completely determine the chiral parameters in terms of the parameters of the sm
, it is enough to impose matching conditions in the two , three , and four - point 1lpi renormalized green s functions with external gauge fields .
we have worked in a general r@xmath32-gauge to show that the chiral parameters @xmath35 are @xmath43-independent .
we use dimensional regularization and the on - shell substraction scheme .
the sm green s functions are non - local ; in particular , they depend on @xmath44 through the virtual higgs propagators .
one has to make a large @xmath45 expansion to represent the virtual higgs boson effects by the local effective operators @xmath46 . in this step ,
care must be taken since clearly the operations of making loop integrals and taking the large @xmath6 limit do not commute .
thus , one must first regulate the loop integrals by dimensional regularization , then perform the renormalization with some fixed prescription ( on - shell in our case ) and at the end take the large @xmath6 limit , with @xmath6 being the renormalized higgs mass . from the computational point of view , in the large @xmath6 limit we have neglected contributions that depend on @xmath47 and/or @xmath48 and vanish when the formal @xmath49 limit is taken . we show in appendix a one illustrative example of how to take the large @xmath6 expansion of the loop integrals .
the matching procedure can be summarized by the following relation among renormalized 1lpi green s functions @xmath50 where the large higgs mass expansion of the sm green s functions has to be made as explained above .
this matching condition imposes a relation between the renormalization of the sm and the renormalization of the effective theory .
we have chosen to renormalize both theories in the on - shell scheme , so that the renormalized parameters are the physical masses and coupling constants .
therefore , the renormalized parameters are taken to be the same in both theories and the matching conditions will provide relations between the sm and the echl counterterms . ) in both theories has also been discussed . in that case
, the matching procedure relates the running @xmath51-renormalized parameters , that are different in the fundamental and the effective theories . ]
the matching condition ( [ match ] ) represents symbolically a system of tensorial coupled equations ( as many as 1lpi functions for external gauge fields ) with several unknowns , namely the complete set of parameters @xmath52 that we are interested in determining . in our previous work @xcite
, we solved the subset of coupled equations involving the two - point and three - point functions . from this subset
, we were able to determine the chiral parameters @xmath53 and @xmath54 . in section 4
, we will solve the matching equations for the four - point green s functions , thus the set of echl parameters for a heavy higgs will be completed .
but before that , we have to set a renormalization prescription for the standard model .
we start by writing down the sm lagrangian @xmath55\ ] ] where @xmath56 d_\mu \phi & = & ( \partial_\mu + \frac{1}{2 } i g \vec{w}_\mu\cdot\vec{\tau } + \frac{1}{2 } i g ' b_\mu ) \phi . \label{smpar}\end{aligned}\ ] ] @xmath57 are defined in eqs.([fpar],[ften ] ) , @xmath33 and @xmath58 are the usual r@xmath59 gauge fixing and faddeev popov terms of the standard model .
we rescale the fields and parameters as follows @xmath60 \phi^b = z_\phi^{1/2 } \phi , & v^b = z_\phi^{1/2 } ( v - \delta v ) , \\[2 mm ] g^b = z_w^{-1/2 } ( g - \delta g ) , & g'^b = z_b^{-1/2 } ( g ' - \delta g ' ) , \\[2 mm ] \mu^b = z_\phi^{-1/2 } ( \mu - \delta \mu ) , & \lambda^b = \lambda ( 1 - \delta \lambda / \lambda ) , \\[2 mm ] \xi_b^b = \xi_b ( 1 + \delta \xi_b ) , & \xi_w^b = \xi_w ( 1 + \delta \xi_w ) . \end{array } \label{reps}\ ] ] where the renormalization constants of the sm are @xmath61 and the superscript b denotes bare quantities . we have chosen to renormalize the sm in the on - shell scheme .
we choose the physical masses , @xmath6 , @xmath62 , @xmath63 and @xmath64 as our renormalized parameters .
the weak mixing angle is defined in terms of physical quantities , as it is usual in the on - shell scheme @xmath65 and from @xmath64 and @xmath66 one derives the coupling constant @xmath67 .
the 1lpi renormalized green s functions in the standard model to one loop will be generically denoted by @xmath68 where one has to consider the tree , counterterm and loop contributions of all the one light particle irreducible diagrams in the sm ; that is , all the diagrams that can not be disconnected by cutting a light ( non - higgs ) particle line . in principle
, we should give now the whole set of renormalization conditions defining the sm on - shell scheme .
however , to extract from the matching conditions the values of the chiral parameters @xmath35 , we only need for the moment to evaluate explicitely the sm counterterms that enter in the renormalization of the diagrams @xmath69 in figs.(3 ) , that is , the tree level diagrams with an intermediate higgs boson . furthermore
, since we are doing a large @xmath6 expansion , it will not be necessary to give the complete expressions for these sm counterterms , but just the leading terms that give non - negligible contributions ( i.e. non vanishing in the large @xmath6 limit ) once they are plugged into the matching equations of the four - point functions . in summary , these considerations imply that we will need explicit expressions for the tadpole and higgs mass counterterms to order @xmath70 and for the w and z mass counterterms to order @xmath71 .
the other sm counterterms , @xmath72 and @xmath73 , have at most a logarithmic dependence on higgs mass and give subleading contributions to the renormalization of the @xmath69 diagrams .
the @xmath72 and @xmath73 counterterms , however , do contribute to the renormalization of the four - point green s functions through the renormalization of the tree level irreducible diagrams .
we do not need to give explicit expressions for them because , as we will see , they appear in the matching through the differences @xmath74 and @xmath75 .
the values of these differences will be extracted from the matching .
the renormalization of the scalar sector has been done following the work of marciano and willenbrock @xcite . in order to fix the notation and renormalization for the tadpole we first write down the sm lagrangian for the scalar sector in terms of the would - be goldstone boson fields @xmath76 and @xmath77 and the physical higgs boson field @xmath78 . in terms of the bare fields and parameters
, it reads as follows @xmath79 & & - \lambda^b \left [ \phi^{+ b } \phi^{- b } ( \phi^{+ b } \phi^{- b } + ( \chi^b)^2 + ( h^b)^2 ) + \frac{1}{4 } ( ( \chi^b)^2 + ( h^b)^2 ) ^2 \right ] \nonumber\\ & & - \lambda^b v^b \left [ 2 \phi^{+ b } \phi^{- b } h^b + ( \chi^b)^2 h^b + ( h^b)^3 \right ] \nonumber \\[2 mm ] & & - \lambda^b ( v^b)^2 ( h^b)^2 + \frac{\delta t}{v^b } ( \phi^{+ b } \phi^{- b } + \frac{1}{2 } ( \chi^b)^2 + \frac{1}{2 } ( h^b)^2 ) + \delta t h^b \nonumber\\[2 mm ] & & + ( \xi { \rm -dependent \;\ ; terms } ) \label{lesc}\end{aligned}\ ] ] the tadpole counterterm is defined in terms of bare parameters as is opposite to the one in @xcite ] @xmath80 the renormalization condition for the tadpole is fixed such that the tadpole loop corrections @xmath81 are cancelled by the tadpole counterterm @xmath82 , or equivalently , such that the renormalized tadpole vanishes . from the computational point of view , one ignores all tadpole diagrams and tadpole counterterms , and includes the extra contributions to the renormalized scalar green s functions coming from the counterterm @xmath83 in eq.([lesc ] ) that is quadratic in the scalar fields .
the bare masses of the higgs and gauge bosons are taken as @xmath84 so that the following relation among the basic bare parameters holds @xmath85 the renormalized masses are defined by @xmath86 so that the on - shell renormalization conditions @xmath87 = 0,\hspace{1 cm } & t^{\rm r } \equiv t + \delta t = 0,\nonumber\\[2 mm ] & { \rm re } \left [ \sigma_{w}^{\rm r } ( q^2 = m_w^2 ) \right ] = 0 , \hspace{1 cm } & { \rm re } \left [ \sigma_{z}^{\rm r } ( q^2 = m_z^2 ) \right ] = 0 , \label{rcon}\end{aligned}\ ] ] imply that eqs.([mwh],[lamb ] ) are also fulfilled by the renormalized quantities @xmath88 m_w^2 & = & g^2 v^2 / 4 , \nonumber\\[2 mm ] m_z^2 & = & ( g^2 + g'^2 ) v^2 / 4 , \nonumber\\[2 mm ] \lambda & = & \frac { g^2 m_h^2}{8 m_w^2}. \label{core}\end{aligned}\ ] ] the renormalization conditions ( [ rcon ] ) fix the values of the sm counterterms to be , ( v = w , z ) and @xmath89 the direct result from the feynman diagrams .
the tadpole loops are however denoted by @xmath90 . ]
@xmath91 + { \displaystyle \frac{\delta t}{v } } , \hspace{1 cm } & \delta t = - t , \\[4 mm ] \delta m_w^2 = { \rm re } \left [ \sigma_{w}^{\rm l } ( q^2 = m_w^2 ) \right ] , \hspace{1 cm } & \delta m_z^2 = { \rm re } \left [ \sigma_{z}^{\rm l } ( q^2 = m_z^2 ) \right ] .
\vspace{2 mm } \end{array } \label{rc2}\ ] ] if one wishes to keep just the non - vanishing contributions in the large @xmath6 limit to the renormalization of the tree level diagrams @xmath69 , the computation of the unrenormalized self - energies of eq.([rc2 ] ) involve just the leading diagrams collected in fig.(2 ) .
these loop diagrams give the following values of the mass and tadpole counterterms in the on - shell scheme @xmath92 + { \cal o}(1),\nonumber\\[2 mm ] \frac{\delta m_w^2}{m_w^2 } & = & \frac{g^2 m_h^2}{m_w^2 } \frac{- 1}{16 \pi^2 } \frac{1}{8 } + { \cal o}(1),\nonumber\\[2 mm ] \frac{\delta m_z^2}{m_z^2 } & = & \frac{g^2 m_h^2}{m_w^2 } \frac{- 1}{16 \pi^2 } \frac{1}{8 } + { \cal o}(1 ) , \nonumber\\[2 mm ] \frac{\delta t / v}{\mh^2 } & = & - \frac{g^2 m_h^2}{m_w^2 } \frac { 1}{16 \pi^2 } \frac{3}{8 } \left [ \hat{\delta}_\epsilon + 1 \right ] + { \cal o}(1 ) , \label{smcl}\end{aligned}\ ] ] where @xmath93 and @xmath94 is the usual mass scale of dimensional regularization .
we have checked the agreement of our expressions with the results of @xcite .
we have explicitely indicated in the formulas of the sm counterterms ( [ smcl ] ) that these expressions are truncated to a certain order in the 1/@xmath6 expansion .
this truncation is enough to keep the non - vanishing effects of a heavy higgs in the evaluation of the following combination of sm counterterms @xmath95 that comes from the renormalization of the tree level diagrams @xmath69 and appears explicitely in the matching equations for the four - point functions given in appendix b expansion of these terms @xcite . ] .
in this section we present the results of our calculation of the 4-point green s functions , giving the set of matching equations that we have imposed and their solution .
the master equation that summarizes the complete set of matching conditions for the renormalized four - point functions is the following : @xmath96 where @xmath97 .
the calculation of the one loop contributions @xmath98 and @xmath99 is the most involved part .
one must include all the 1pi diagrams in the echl and all the 1lpi diagrams in the sm .
1lpi diagrams are those that can not be disconnected by cutting a single light particle line , that is , a non - higgs particle line .
one must , in principle , account for all kind of diagrams with gauge , scalar and ghost fields flowing in the loops .
however , some simplifications occur .
firstly , a subset of the diagrams that have only light particles in it is exactly the same in both models , and their contribution can be simply dropped out from both sides of the matching equation ( [ mama ] ) .
this is the case , for instance , of the subset of diagrams whith only gauge particles in them .
secondly , calculating explicitely every diagram in the four - point 1lpi sm functions and using the techniques given in app.a , we have checked that the diagrams involving both gauge and higgs particles in the loops give vanishing contributions in the large @xmath6 limit to the four - point functions . only those with just scalars ( goldstone bosons or higgs ) particles in the loops do contribute with non - vanishing corrections in the large @xmath6 limit to the matching equation ( [ mama ] ) limit . ] .
finally , among the diagrams with pure scalar loops , there are some with only goldstone boson particles .
one would expect that these diagrams give the same contributions in the sm and the echl , however they do not .
the reason is the already mentioned differences in the feynman rules of the vertices in fig.(1 ) .
these diagrams ( denoted generically by @xmath100 in figs.(3 ) ) must therefore be included in both sides of the matching equation . in fig.(3 ) , we give the complete list of the tree level ( @xmath69 ) and one loop ( @xmath101 ) diagrams that give a contribution to the matching equations ( [ mama ] ) of the @xmath102 and @xmath103 green s functions .
we give in appendix b the final result of the calculation of the tree , counterterms and loop contributions to the matching equations for the different green s functions .
each matching condition of the form given in eq.(b1 ) is a tensorial equation and therefore it provides several equations corresponding to the various independent tensorial structures .
furthermore , each of these equations can be written in the form of a polynomial in powers of @xmath104 . in summary ,
one gets one equation `` per '' coefficient of the polinomial in each independent tensorial structure and in each green s function .
the result including the equations from the two , three and four - point functions is a linear system with more equations than unknowns .
the system turns out to be compatible , giving a strong consistency check of the calculation . by keeping just the independent set of equations from the four ponit functions ,
one gets : @xmath105\\[2 mm ] \left ( \delta z_w - \frac{\delta g^2}{g^2 } \right ) \frac{1}{g^2 } + a^b_{11 } & = & \frac{1}{16 \pi^2 } \frac{- 1}{12 } \left [ \hat{\delta}_\epsilon + \frac{4}{3 } \right ] \nonumber \\[2 mm ] 2 a^b_3 & = & \frac{1}{16 \pi^2 } \frac{- 1}{12 } \left [ \hat{\delta}_\epsilon + \frac{17}{6 } \right ] \nonumber \\[2 mm ] a^b_3 - a^b_{11 } + a^b_{12 } & = & \frac{1}{16 \pi^2 } \frac{- 1}{24 } \left [ \hat{\delta}_\epsilon + \frac{11}{6 } \right ] \nonumber \\[2 mm ] 2 \left ( a^b_5 + a^b_7 \right ) & = & \frac{1}{16 \pi^2 } \frac{1}{12 } \left [ \frac{43}{2 } \hat{\delta}_\epsilon + \frac{47}{3 } \right ] + \frac{m_w^2}{g^2 m_h^2 } + \delta s \nonumber \\[2 mm ] a_4^b + a^b_6 - a^b_{11 } + 2 a^b_{12 } & = & \frac{1}{16 \pi^2 } \frac{- 1}{12 } \left [ \hat{\delta}_\epsilon + \frac{7}{3 } \right ] \nonumber \\[2 mm ] \left ( \delta z_w - \frac{\delta g^2}{g^2 } \right ) \frac{1}{g^2 } + 2 a^b_3 - a_4^b - a^b_8 + 2 a^b_9 - 2 a^b_{13 } & = & \frac{1}{16 \pi^2 } \frac{- 1}{12 } \left [ \hat{\delta}_\epsilon + \frac{5}{6 } \right ] \nonumber \\[2 mm ] \left ( \delta z_w - \frac{\delta g^2}{g^2 } \right ) \frac{1}{g^2 } + 2 a^b_3 + a_4^b + 2 a^b_5 - a^b_8 + 2 a^b_9 - 2 a^b_{13 } & = & \frac{1}{16 \pi^2 } \frac{1}{24 } \left [ 37 \hat{\delta}_\epsilon + \frac{55}{3 } \right ] + \frac{m_w^2}{g^2 m_h^2 } + \delta s \nonumber \\[2 mm ] 2 \left ( a^b_4 + a^b_5 + 2 ( a^b_6 + a^b_7 + 2 a^b_{10 } ) \right ) & = & \frac{1}{16 \pi^2 } \frac { 1}{8 } \left [ 13 \hat{\delta}_\epsilon + \frac{20}{3 } \right ] + \frac{m_w^2}{g^2 m_h^2 } + \delta s \nonumber\end{aligned}\ ] ] where @xmath106 and @xmath107 have been defined in eqs .
( [ eps ] ) and ( [ dels ] ) respectively and we use the following notation for the differences of counterterms @xmath108 the first four equations provide a check for the values of @xmath109 and @xmath110 that we already obtained in our previous work from the calculation of the two and three - point functions .
finally , using these results and the values of @xmath111 and @xmath112 from @xcite we can extract the genuine parameters of the four - point functions : @xmath113 and @xmath114 . by solving the complete linear system of equations ,
one gets a unique solution for the bare electroweak chiral parameters given by : @xmath115 a_1^b & = & \frac{1}{16 \pi^2 } \frac{1}{12 } \left ( \delta_\epsilon - \log \frac{\mh^2}{\mu^2 } + \frac{5}{6 } \right ) , \nonumber \\[2 mm ] a_2^b & = & \frac{1}{16 \pi^2 } \frac{1}{24 } \left ( \delta_\epsilon - \log \frac{\mh^2}{\mu^2 } + \frac{17}{6 } \right ) , \nonumber \\[2 mm ] a_3^b & = & \frac{-1}{16 \pi^2 } \frac{1}{24 } \left ( \delta_\epsilon - \log \frac{\mh^2}{\mu^2 } + \frac{17}{6 } \right ) , \nonumber \\[2 mm ] a_4^b & = & \frac{-1}{16 \pi^2 } \frac{1}{12 } \left ( \delta_\epsilon - \log \frac{\mh^2}{\mu^2 } + \frac{17}{6}\right ) , \nonumber \\[2 mm ] a_5^b & = & \frac{m_w^2}{2 g^2 \mh^2 } - \frac{1}{16 \pi^2 } \frac{1}{24 } \left ( \delta_\epsilon - \log \frac{\mh^2}{\mu^2 } + \frac{79}{3 } - \frac { 27 \pi}{2 \sqrt { 3 } } \right ) , \nonumber\\[2 mm ] a_{11}^b & = & \frac{-1}{16 \pi^2}\frac{1}{24 } , \nonumber \\[2 mm ] a_6^b & = & a_7^b \ ; = \ ; a_8^b \ ; = \ ; a_9^b \ ; = \ ; a_{10}^b \ ; = \ ; a_{12}^b \ ; = \ ; a_{13}^b \ ; = \ ; 0 . \label{amh}\end{aligned}\ ] ] we would like to make some remarks on this result for the chiral parameters : 1 .
first of all , we agree with the @xmath116 dependence of the @xmath52 parameters that was first calculated by longhitano @xcite looking at the divercences of the non - linear sigma model .
we see therefore that the divergences generated with the @xmath9 to one loop are exactly canceled by the @xmath116 terms in the @xmath52 s .
the values of @xmath117 and @xmath118 agree with the results given in @xcite , where the equivalence theorem was used in comparing the scattering amplitudes for goldstone bosons in the sm @xcite and the echl .
these values , on the other hand , do not coincide with the values in @xcite where just contributions from pure higgs loops were considered .
it is important to realize that the matching procedure fixes completely the values of the bare parameters @xmath52 in terms of the renormalized parameters of the sm .
4 . eqs.([amh ] ) give the complete non - decoupling effects of a heavy higgs , that is , the leading logarithmic dependence on @xmath6 and the next to leading constant contribution to the electroweak chiral parameters .
the @xmath35 s are accurate up to corrections of the order @xmath47 where @xmath119 and higher order corrections in the perturbative expansion .
we demonstrate that the @xmath35 s are gauge independent , as expected .
there is only one effective operator , the one corresponding to @xmath120 , that gets a tree level contribution .
its expression in terms of renormalized sm parameters depends on the renormalization prescription that one has chosen in the standard model , on - shell in our case .
we believe it is important at this point to clarify the relation among the @xmath121 s that correspond to an on - shell renormalization of the sm and their corresponding values if a different renormalization prescription for the sm is chosen .
we will discuss this point in the following section . by putting together the results of the two , three and four - point functions , one also obtains some relations among the counterterms of the two theories @xmath122 \delta z_b & = & \frac{- g'^2}{16 \pi^2 } \frac{1}{12 } \left ( \delta_\epsilon - \log \frac{\mh^2}{\mu^2 } + \frac{5}{6 } \right ) , \nonumber\\[2 mm ] \delta \xi_w & = & \delta z_w , \hspace{1 cm } \delta \xi_b = \delta z_b , \nonumber \\[2 mm ] \frac{\delta g^2}{g^2 } & = & \frac{\delta g'^2}{g'^2 } = 0 , \nonumber \\[2 mm ] \delta z_\phi - 2 \frac{\delta v}{v } & = & \frac{g^2}{16 \pi^2 } \left [ - \frac{\mh^2}{8 m_w^2 } + \frac{3}{4 } \left ( \delta_\epsilon - \log \frac{\mh^2}{\mu^2 } + \frac{5}{6 } \right ) \right .
\nonumber \\[2 mm ] & & \left .
+ \frac{1}{4 } \frac{\xi_z}{c^2 } \left ( \delta_\epsilon - \log \frac{\xi_z m_z^2}{\mu^2 } + 1 \right ) + \frac{1}{2 } \xi_w \left ( \delta_\epsilon - \log \frac{\xi_w m_w^2}{\mu^2 } + 1 \right ) \right]\end{aligned}\ ] ] these equations give the differences among the renormalization constants of the sm in the large @xmath6 limit and those in the echl , when the on - shell renormalization scheme is chosen in both theories .
they are obtained here as a constraint imposed by the matching ; one can also calculate them from the explicit expressions of the on - shell counterterms of the two theories and verify that these relations are indeed satisfied .
in the previous section , we have given the values of the bare electroweak chiral parameters ( [ amh ] ) when the sm is renormalized in the on - shell scheme .
we would like now to discuss how these bare parameters are changed when a different renormalization scheme is chosen for the sm .
we have seen that the chiral parameter @xmath118 is the only one that gets a tree level contribution when a heavy higgs is integrated out , and therefore it is the only one whose expression in terms of the sm renormalized parameters will depend on the renormalization prescription of the sm . by using eqs.([core ] ) and
( [ amh ] ) , one can rewrite @xmath118 in terms of the on - shell renormalized scalar self - coupling @xmath123 : @xmath124 the easiest way to connect with a new renormalization scheme for the sm is to write down @xmath118 in terms of the bare scalar self - coupling @xmath125 . to this end , we first write down the relation is of higher order in the large @xmath45 expansion than the contributions from the renormalization of @xmath45 and @xmath62 and will be ignored here . ] between the renormalized self - coupling @xmath123 and the bare @xmath125 @xmath126 by substituing the values of the mass counterterms given in ( [ smcl ] ) in this equation , we get the relation among the bare and the renormalized self - coupling in the on - shell scheme @xmath127\ ] ] now , by plugging the last equation into eq.([a5lr ] ) one finds the value of @xmath118 in terms of @xmath125 @xmath128 in order to connect with a different renormalization prescription we simply substitute @xmath125 in eq.([a5 lb ] ) by its corresponding definition in terms of the new renormalized self - coupling .
for instance , if one chooses the @xmath51 scheme , where the scalar self - coupling is defined by @xmath129\ ] ] then @xmath118 in terms of the renormalized self - couplig @xmath130 is given by @xmath131 and for the rest of chiral parameters one gets exactly the same values as in eq.([amh ] ) .
another example is the renormalization prescription chosen by gasser and leutwyler in @xcite .
their prescription for the renormalized coupling is given by is what we call here @xmath132 .
the mass apearing in @xmath106 is not the physical mass of the higgs , but what they call @xmath133 .
the renormalized mass @xmath133 was fixed in @xcite such that @xmath134 . ]
@xmath135\ ] ] and the expression of @xmath118 in terms of the renormalized self - coupling @xmath132 is therefore @xmath136 and the rest of parameters remain again the same . for comparison ,
it is interesting to translate the results of eqs.([amh],[a5gl ] ) to the usual notation in chiral perturbation theory @xmath137 l^b_2 & = & \frac{l^b_2}{4 } = a^b_4 = \frac{- 1}{16 \pi^2 } \frac{1}{12 } \left ( \hat{\delta}_\epsilon + \frac{17}{6}\right ) , \nonumber \\[2 mm ] l^b_9 & = & - \frac { l^b_6}{2 } = a^b_3 - a^b_2 = \frac { - 1}{16 \pi^2 } \frac{1}{12 } \left ( \hat{\delta}_\epsilon + \frac{17}{6 } \right ) , \nonumber \\[2 mm ] l^b_{10 } & = & l^b_5 = a^b_1 = \frac{1}{16 \pi^2 } \frac{1}{12 } \left ( \hat{\delta}_\epsilon + \frac{5}{6 } \right ) .
\label{lib}\end{aligned}\ ] ] these chiral parameters agree with the values found by gasser and leutwyler in @xcite ( see appendix b of this reference ) and in @xcite .
we find it a quite remarkable agreement since they used functional methods to integrate out the higgs particle in a linear sigma model where the gauge fields were considered as external sources and they used as well different techniques to study the large @xmath6 limit .
this confirms the fact already mentioned that the contributions to the effective operators of dimension four that come from mixed gauge - scalar loops are subleading , in the large @xmath6 limit , compared to the pure scalar loops contributions .
however , this is not the case for the dimension two custodial breaking operator @xmath20 that gets contributions from mixed gauge - scalar loops @xcite . to conclude this section , we emphasize once more that the bare chiral parameters for a given underlying theory , the sm in our case , must be computed with a choice for the renormalization prescription of this theory .
the explicit expression of the chiral parameters will vary from one prescription to another , but the numerical value remains the same , and the connection between different prescriptions can be clearly and easily established .
in this section we briefly describe the renormalization procedure in the effective theory . given the effective lagrangian of eq.([ecl ] ) , the first step is to redefine the fields and parameters of the lagrangian according to eq.([ret ] ) .
it introduces , at a formal level , the set of counterterms of the effective theory @xmath138 , etc , that need to be computed once a particular renormalization prescription scheme is chosen .
we fix here the on - shell renormalization scheme as we did in the case of the sm . for practical reasons we prefer to choose the renormalization conditions as in reference @xcite , which are the most commonly used for lep physics . in terms of the renormalized selfenergies these renormalization conditions read as follows @xmath139
the renormalized self energies are computed in the effective theory as usual , namely , by adding all the contributions from the one loop diagrams and from the counterterms .
we get the following expressions have different sign than in reference @xcite ] : @xmath140 \widehat{\sigma}^{\rm r}_{w } ( q^2 ) & = & \widehat{\sigma}^{\rm l}_{w } ( q^2 ) + \widehat{\delta z_w } \left ( q^2 - m_w^2 \right ) - \widehat{\delta m_w^2}. \nonumber\\[2 mm ] \widehat{\sigma}^{\rm r}_{z } ( q^2 ) & = & \widehat{\sigma}^{\rm l}_{z } ( q^2 ) + \left ( c^2 \widehat{\delta z_w } + s^2 \widehat{\delta z_b } \right ) ( q^2 - m_z^2 ) - \widehat{\delta m_z^2 } \nonumber\\ & & + 2 g'^2 a_0^b m_z^2 + \left ( 2 s^2 g^2 a^b_1 + c^2 g^2 a_8^b + ( g^2 + g^{\prime 2 } ) a^b_{13 } \right ) q^2 .
\nonumber\\[2 mm ] \widehat{\sigma}^{\rm r}_{\gamma z } ( q^2 ) & = & \widehat{\sigma}^{\rm l}_{\gamma z } ( q^2 ) + s c \left ( \widehat{\delta z_w } - \widehat{\delta z_b } \right ) q^2 - s c\ ; m_z^2 \left ( \frac{\widehat{\delta g'}}{g ' } - \frac{\widehat{\delta g}}{g } \right ) \nonumber\\ & & + \left ( s c g^2 a^b_8 - ( c^2 - s^2 ) g g ' a^b_1 \right ) q^2 .
\label{jql}\end{aligned}\ ] ] where @xmath141 \widehat{\delta m_z^2 } & = & m_z^2 \left ( \widehat{\delta z_\phi } - 2 c^2 \frac{\widehat{\delta g}}{g } - 2 s^2 \frac{\widehat{\delta g'}}{g ' } - 2 \frac{\widehat{\delta v}}{v } - c^2 \widehat{\delta z_w } - s^2 \widehat{\delta z_b } \right),\nonumber\\[2 mm ] m_w^2 & = & g^2 v^2 / 4 , \nonumber\\[2 mm ] m_z^2 & = & ( g^2 + g'^2 ) v^2 / 4 , \label{masas}\end{aligned}\ ] ] and the superscripts r and l denote renormalized and echl loops respectively . from eq.([masas ] ) the following relation among the @xmath142 and @xmath143 mass counterterms is obtained @xmath144 finally , by requiring these renormalized self energies to fulfill eqs.([rce ] ) and taking into account that the @xmath145 ward identity implies @xmath146 one gets the following results for the values of the counterterms in terms of the unrenormalized selfenergies of the effective theory and the bare @xmath35 s : @xmath147 \widehat{\delta m_z^2 } & = & \widehat{\sigma}^{\rm l}_{z } ( m_z^2 ) + m_z^2 \left ( 2 g'^2 a_0^b + 2 s^2 g^2 a^b_1 + c^2 g^2 a_8^b + ( g^2 + g^{\prime 2 } ) a^b_{13 } \right ) , \nonumber\\[2 mm ] \frac{\widehat{\delta g}}{g } & = & \frac{- 1}{s c } \frac{\widehat{\sigma}^{\rm l } _ { \gamma z}(0 ) } { m_z^2 } , \nonumber\\[2 mm ] \frac{\widehat{\delta g'}}{g ' } & = & 0 , \nonumber\\[2 mm ] \widehat{\delta z_w } & = & \frac{c^2}{s^2 } \left(\frac{\widehat{\sigma}^{\rm l}_{z}(m_z^2 ) } { m_z^2 } - \frac{\widehat{\sigma}^{\rm l}_{w}(m_w^2 ) } { m_w^2 } \right ) + 2 \frac{c}{s } \frac { \widehat{\sigma}^{\rm l } _ { \gamma z}(0)}{m_z^2 } - \widehat{\sigma}^{\prime \;\rm l}_\gamma ( 0 ) \nonumber\\ & & + 2 g^2 a^b_0
+ 2 g^2 a^b_1 + \frac{c^2 - s^2}{s^2 } g^2 a^b_8 + \frac{c^2}{s^2 } ( g^2 + g'^2 ) a^b_{13 } , \nonumber\\[2 mm ] \widehat{\delta z_b } & = & \frac{\widehat{\sigma}^{\rm l}_{w}(m_w^2)}{m_w^2 } - \frac { \widehat{\sigma}^{\rm l}_{z}(m_z^2)}{m_z^2 } - 2 \frac{s}{c } \frac{\widehat{\sigma}^{\rm l } _ { \gamma z}(0)}{m_z^2 } - \widehat{\sigma}^{\prime \rm l}_\gamma ( 0 ) \nonumber\\ & & - \left ( 2 g'^2 a^b_0 + g^2 a^b_8 + ( g^2 + g'^2 ) a^b_{13 } \right ) .
\label{ect}\end{aligned}\ ] ] now that we have at hand eqs.([ect ] ) the only parameters of the theory that still need to be renormalized are the electroweak chiral parameters @xmath35 .
the following formal redefinition of the chiral parameters has already been introduced @xmath148 the divergent part of the @xmath52 parameters , or equivalently the divergent part of the counterterms @xmath149 , are fixed by the symmetries of the effective theory and since the work of longhitano @xcite they are known to be @xmath150 & & \delta a_2 |_{div } = \frac{1}{16 \pi^2 } \frac{1}{24 } \delta_\epsilon , \hspace{2cm}\delta a_3 |_{div } = \frac{- 1}{16 \pi^2 } \frac{1}{24 } \delta_\epsilon , \nonumber \\[2 mm ] & & \delta a_4 |_{div } = \frac{- 1}{16 \pi^2 } \frac{1}{12 } \delta_\epsilon , \hspace{2cm}\delta a_5 |_{div } = \frac{- 1}{16 \pi^2 } \frac{1}{24 } \delta_\epsilon , \nonumber \\[2 mm ] & & \delta a_i |_{div } = 0 , \hspace{5 mm } i = 6, .... 13 . \label{pp}\end{aligned}\ ] ] these universal divergent contributions to the chiral bare parameters imply in turn universal renormalization group equations for the renormalized parameters @xmath151 & & a_2(\mu ) = a_2(\mu ' ) + \frac{1}{16 \pi^2 } \frac{1}{24 } \log\frac{\mu^2}{\mu ' ^2 } , \hspace{1 cm } a_3(\mu ) = a_3(\mu ' ) - \frac{1}{16 \pi^2 } \frac{1}{24 } \log\frac{\mu^2}{\mu ' ^2 } , \nonumber\\[2 mm ] & & a_4(\mu ) = a_4(\mu ' ) - \frac{1}{16 \pi^2 } \frac{1}{12 } \log\frac{\mu^2}{\mu ' ^2 } , \hspace{1 cm } a_5(\mu ) = a_5(\mu ' ) - \frac{1}{16 \pi^2 } \frac{1}{24 } \log\frac{\mu^2}{\mu ' ^2 } , \nonumber\\[2 mm ] & & a_i(\mu ) = a_i(\mu ' ) ; \hspace{3 mm } i = 6, ... 13.\end{aligned}\ ] ] the value of the bare chiral parameters @xmath121 , on the other hand , is completely determined by the matching procedure in terms of the renormalized parameters of the underlying physics that has been integrated out , as we have seen for the particular case of a heavy higgs .
however , for a given @xmath121 , we still have to choose how to separate the finite part into the renormalized @xmath40 and the counterterm @xmath149 in eq.([nsq ] ) such that their sum gives @xmath121 .
this second renormalization scheme concerns only to the effective theory .
therefore , in using a set of renormalized parameters @xmath40 for a particular underlying theory , one must specify , in addition , how the finite parts of the counterterms in eq.([nsq ] ) have been fixed . in the case of the sm , where a heavy higgs has been integrated out to one loop ,
the bare chiral parameters are given in eq.([amh ] ) .
they correspond to the on - shell renormalization of the underlying sm .
now , in order to present the corresponding renormalized parameters we have to fix the finite parts of the counterterms .
for instance , if we fix the counterterms to include just the @xmath152 terms as in eq.([pp ] ) , the renormalized chiral parameters for the sm with a heavy higgs are . ] : @xmath153 a_1 ( \mu ) = { \displaystyle \frac{1}{16 \pi^2 } \frac{1}{12 } \left ( \frac{5}{6 } - \log\frac{m_h^2}{\mu^2 } \right ) } , & \hspace{-1cm}a_4 ( \mu ) = { \displaystyle \frac{- 1}{16 \pi^2 } \frac{1}{12 } \left ( \frac{17}{6 } - \log\frac{m_h^2}{\mu^2 } \right)},\\[6 mm ] a_2 ( \mu ) = { \displaystyle \frac{1}{16 \pi^2 } \frac{1}{24 } \left ( \frac{17}{6 } - \log\frac{m_h^2}{\mu^2 } \right ) } , & \hspace{-1cm}a_{11 } ( \mu ) = { \displaystyle \frac{-1}{16 \pi^2}\frac{1}{24}},\\[6 mm ] a_5 ( \mu ) = { \displaystyle \frac{v^2}{8 \mh^2 } - \frac{1}{16 \pi^2 } \frac{1}{24 } \left ( \frac{79}{3 } - \frac { 27 \pi}{2 \sqrt{3 } } - \log\frac{m_h^2}{\mu^2 } \right ) } , \\[6 mm ] a_i ( \mu ) = 0 , \hspace{3 mm } i = 6,7,8,9,10,12,13 .
\end{array } \label{ar}\ ] ] another example is the renormalization scheme chosen by gasser and leutwyler for the linear @xmath154 sigma model in ref.@xcite .
the values of the bare parameters for their choice of the renormalization scheme of the underlying sigma model were given in eqs.([lib ] ) .
now , they fix instead the chiral counterterms to the following values : @xmath155 \delta l_2 & = & \frac{\delta l_2}{4 } = \delta a_4^{\rm gl } = \frac{1}{16 \pi^2 } \frac{- 1}{12 } ( \delta_\epsilon + 1 ) , \nonumber \\[2 mm ] \delta l_9 & = & - \frac{\delta l_6}{2 } = \delta a_3^{\rm gl } - \delta a_2^{\rm gl } = \frac{1}{16 \pi^2 } \frac{- 1}{12 } ( \delta_\epsilon + 1 ) , \nonumber \\[2 mm ] \delta l_{10 } & = & \delta l_5 = \delta a_1^{\rm gl } = \frac{1}{16 \pi^2 } \frac{1}{12 } ( \delta_\epsilon + 1).\end{aligned}\ ] ] consequently , they get the following renormalized chiral parameters for the sigma model the renormalized mass @xmath133 of @xcite . ]
@xmath156 l_2(\mu ) & = & \frac{l_2(\mu)}{4 } = a_4^{\rm gl}(\mu ) = \frac{- 1}{16 \pi^2 } \frac{1}{12 } \left ( \frac{11}{6 } - \log\frac{m^2_{\rm gl}}{\mu^2 } \right ) , \nonumber \\[2 mm ] l_9(\mu ) & = & - \frac { l_6(\mu)}{2 } = a_3^{\rm gl}(\mu ) - a_2^{\rm gl}(\mu ) = \frac { - 1}{16 \pi^2 } \frac{1}{12 } \left ( \frac{11}{6 } - \log\frac{m^2_{\rm gl}}{\mu^2 } \right ) , \nonumber \\[2 mm ] l_{10}(\mu ) & = & l_5(\mu ) = a_1^{\rm gl}(\mu ) = \frac{1}{16 \pi^2 } \frac{1}{12 } \left ( - \frac{1}{6 } - \log\frac{m^2_{\rm gl}}{\mu^2 } \right).\end{aligned}\ ] ]
in this section we will show , as an example , the explicit calculation of the radiative corrections to @xmath0 and @xmath157 within the electroweak chiral lagrangian approach .
these observables are defined in the effective theory in terms of the renormalized self - energies in the same way as in the fundamental sm , namely : @xmath158 \delta r & \equiv & \frac{\widehat{\sigma}^{\rm r}_w ( 0)}{m_w^2 } + { \rm ( vertex + box ) } , \label{ror}\end{aligned}\ ] ] where @xmath159 and the renormalized self - energies can be computed as we have explained in section 6 .
once a renormalization scheme has been chosen , one can always express @xmath0 and @xmath157 in terms of unrenormalized self - energies and the @xmath52 s .
for instance , in the on - shell scheme given by the conditions of eq.([rce ] ) , one gets the particular values of the counterterms given in eqs.([ect ] ) .
next , by plugging these counterterms into eqs.([jql ] ) , one obtains the renormalized self - energies in terms of the unrenormalized ones and the @xmath52 s .
finally , by substituing these formulas into eqs.([ror ] ) the following expressions for @xmath0 and @xmath160 in the on - shell scheme are found @xmath161 \delta r & = & \frac{\widehat{\sigma}^{\rm l}_w ( 0 ) - \widehat{\sigma}^{\rm l}_w ( m_w^2)}{m_w^2 } + \widehat{\sigma}'^{\rm l}_\gamma ( 0 ) + \frac{c^2}{s^2 } \left [ \frac{\widehat{\sigma}^{\rm l}_w ( m_w^2)}{m_w^2 } - \frac{\widehat{\sigma}^{\rm l}_z ( m_z^2)}{m_z^2 } - \frac{2 s}{c } \frac{\widehat{\sigma}^{\rm l}_{\gamma z } ( 0 ) } { m_z^2 } \right ] \nonumber\\[2 mm ] & & - 2 g^2 a^b_0 + \frac{s^2 - c^2}{s^2 } g^2 ( a^b_8 + a^b_{13 } ) - 2 g^2 ( a^b_1 + a^b_{13 } ) + { \rm ( vertex + box)}.\end{aligned}\ ] ] the explicit computation of the bosonic loop contributions in the effective theory , as well as the contributions from just @xmath162 and @xmath163 were found in @xcite .
we present here the complete result @xmath164 + 2 g'^2 a^b_0 , \nonumber \\[2 mm ] \delta r & = & \frac{g^2}{16 \pi^2 } \left [ \frac{11}{12 } \left ( \delta_\epsilon - \log \frac{m_w^2}{\mu^2 } \right ) + f(m_w^2,m_z^2 ) \right ] \nonumber\\[2 mm ] & & - 2 g^2 a^b_0 + \frac{s^2 - c^2}{s^2 } g^2 ( a^b_8 + a^b_{13 } ) - 2 g^2 ( a^b_1 + a^b_{13 } ) .
\label{rorab}\end{aligned}\ ] ] where @xmath165 f(m_w^2,m_z^2 ) & = & \log c^2 \left ( \frac{5}{c^2 } -1 + \frac{3 c^2}{s^2 } - \frac{17}{4 s^2 c^2 } \right ) - s^2 ( 3 + 4 c^2 ) f(m_z^2,m_w , m_w)\nonumber \\ & & + i_2(c^2)(1 - \frac{c^2}{s^2 } ) + \frac{c^2}{s^2 } i_1(c^2 ) + \frac{1}{8 c^2 } ( 43 s^2 - 38 ) \nonumber \\ & & + \frac{1}{18 } ( 154 s^2 - 166 c^2 ) + \frac{1}{4 c^2 } + \frac{1}{6 } + \delta \alpha + \left ( 6 + \frac{7 - 4 s^2 } { 2 s^2 } \log c^2 \right).\end{aligned}\ ] ] and @xmath166 , @xmath167 , @xmath168 and @xmath169 can be found in @xcite . in eqs.([rorab ] ) there are apparently a divergent term and a @xmath94-scale dependent term .
however , when one redefines the bare effective chiral parameters as usual , @xmath170 , it can be easily seen that the divergent terms are cancelled by the divergent parts of the @xmath149 and the @xmath94-scale dependence is cancelled by the scale dependence of the @xmath40 .
the observables @xmath0 and @xmath157 turn out to be finite and scale and renormalization prescription independent , as it must be .
in particular , if we set the substraction scheme for the chiral counterterms to include just the @xmath152 terms as in eq.([pp ] ) , the following expressions for @xmath0 and @xmath157 in terms of renormalized chiral parameters are obtained @xmath171 + 2 g'^2 a_0(\mu ) , \nonumber \\[2 mm ] \delta r & = & \frac{g^2}{16 \pi^2 } \left [ - \frac{11}{12 } \log \frac{m_w^2}{\mu^2 } + f(m_w^2,m_z^2 ) \right ] \nonumber\\[2 mm ] & & - 2 g^2 a_0(\mu ) + \frac{s^2 - c^2}{s^2 } g^2 ( a_8 + a_{13 } ) - 2 g^2 ( a_1(\mu ) + a_{13 } ) .
\label{rrf}\end{aligned}\ ] ] equations ( [ rrf ] ) are general and can be applied to any underlying physics for the symmetry breaking sector .
if we want to recover the values of @xmath0 and @xmath157 in the particular case of the sm with a heavy higgs , one just has to substitute the values of the chiral parameters in eqs.([ar ] ) into eqs.([rrf ] ) to obtain @xmath172 , \nonumber \\[2 mm ] \delta r & = & \frac{g^2}{16 \pi^2 } \left [ \frac{11}{12 } \left(\log \frac{m_h^2}{m_w^2 } - \frac{5}{6 } \right ) + f(m_w^2,m_z^2 ) \right].\end{aligned}\ ] ] which agrees with the result given in @xcite .
one can similarly obtain the heavy higgs contributions to other relevant observables in electroweak phenomenology .
given the present situation of remarkable improvement in electroweak precision measurements and the encouraging prospects for the future , we believe it is now imperative a good undertanding of the sm higgs boson radiative corrections .
a heavy higgs boson do not decouple from the low energy electroweak observables and therefore will leave some trace on them that could be observed with the future precision measurements . in this paper , we have calculated the complete non - decoupling effects of the sm higgs boson to one loop .
they consist of the already known leading logarithmic higgs mass dependent effects and the next to leading constant terms .
both effects are relevant and for not too large @xmath6 values can be of comparable magnitude .
we have classified these non - decoupling effects using the echl approach , an effective field theory that respects the sm symmetries at low energies . within this approach ,
the non - decoupling effects of a heavy higgs boson are represented , at energies below the higgs mass , by a subset of gauge invariant effective operators of the echl .
we have calculated in this work the values of the parameters of these chiral operators that represent the sm higgs .
it is our main result and is given in eq.([amh ] ) .
we get just seven non - vanishing relevant parameters summarizing the whole set of non - decoupling heavy higgs effects to one loop .
we have also discussed in detail in this work the relation between the renormalization of both the sm and the effective theory , with special emphasis in the on - shell scheme , that has been our particular choice here . in conclusion ,
we believe that the approach followed in this work is interesting and useful because it provides a gauge invariant way of separating the non - decoupling higgs boson effects from the rest of the ew radiative corrections and , on the other hand , it is a general framework in which one can analyze the low energy effects not only of an sm heavy higgs , but of alternative strongly interacting syummetry breaking scenarios .
the echl that parametrizes the sm higgs will then serve as a fundamental reference model .
we are indebted to c.p .
martn for his valuable help with some technical aspects of this work and his interesting comments .
we thank a. dobado for many useful conversations and for reading the manuscript .
we appreciate the critical reading of the manuscript done by s. peris .
we would like to thank also the slac theory group and the cern th - division for their hospitality .
m.j.h . acknowledges finantial support from consejera de educacin de la comunidad de madrid during her stay at slac . this work has been partially supported by the spanish ministerio de educacin y ciencia under project cicyt aen93 - 0673
in this appendix , we give an explicit example of the large-@xmath6 techinques used in the calculation of the loop integrals .
we start by giving the m - theorem of g. giavarini , c. p. martin and f. ruiz ruiz @xcite ( reduced to the case of 1-loop integrals ) , which gives sufficient conditions for a loop integral to vanish in the large - m limit .
consider an integral of the form @xmath173 where @xmath174 m_i & = & 0 \;\ ; { \rm or } \;\ ; m\end{aligned}\ ] ] and @xmath175 is a monomial in the components of @xmath176 .
@xmath177 denote an arbitrary real number .
the external momenta @xmath178 lay in a bounded subdomain of @xmath179 .
let @xmath180 be the mass dimension of @xmath181 and @xmath182 the minimum of zero and the infrared degree of @xmath181 at zero external momenta .
then + * m - theorem * : if the integral @xmath181 is both uv and ir convergent by power counting at non - exceptional external momenta and @xmath183 , then @xmath181 goes to zero when @xmath184 goes to infinity . as an example , consider the @xmath185 loop correction to the @xmath142 self - energy given in fig.(2 ) : @xmath186 [ ( k+q)^2 - m_h^2]}\ ] ] where @xmath187 is the @xmath142 external momentum and @xmath188 denotes the space - time dimension in dimensional regularization .
let s work out explicitly the large @xmath6 expansion of the most divergent part of this correction which comes from the integral @xmath189 [ ( k+q)^2 - m_h^2]}\ ] ] the superficial degree of uv divergence of @xmath190 is 2 at @xmath191 .
the first step is to rearrange _ algebraically _ the denominator that includes a light mass @xmath192 so that one gets a term in the integral with the same degree of uv divergence but where the light mass is no more in the denominator , and a second term with still a light mass dependence but with the degree of uv divergence lowered by two . applying this algebraic rearrangement as many times as necessary until the last term gives a convergent integral ( twice in the case of @xmath190 ) , one gets @xmath193 } + \frac { \xi m_w^2 k_\mu k_\nu}{k^4 [ ( k+q)^2 - m_h^2 ] } + \frac{\xi^2 m_w^4 k_\mu k_\nu } { k^4 [ k^2 - \xi m_w^2 ] [ ( k+q)^2 - m_h^2 ] } \right ] \nonumber \\[2 mm ] & = & a_{\mu \nu } + b_{\mu \nu } + c_{\mu \nu}\end{aligned}\ ] ] now the last term @xmath194 is already uv and ir convergent .
one can apply the lebesgue dominated convergence theorem to see that the heavy higgs mass limit can be safely taken inside the integral in this term , and therefore , @xmath195 let s develope now the term @xmath196 , ( the other term @xmath197 can be evaluated using the same procedure ) .
we rewrite the denominators in @xmath196 , using again an algebraic identity @xmath198 [ ( k+q)^2 - m_h^2]}\ ] ] one has to apply this identitity as many times as neccessary until one gets an integral that fulfills the conditions of the m - theorem . using this identity three times in @xmath196 one get s @xmath199 } - \frac{k_\mu k_\nu ( 2 k q + q^2)}{k^2 [ k^2 - m_h^2]^2 } + \frac{k_\mu k_\nu ( 2 k q + q^2)^2}{k^2 [ k^2 - m_h^2]^3 } - \right .
\nonumber \\[2 mm ] & & \left .
\frac{k_\mu k_\nu ( 2 k q + q^2)^3 } { k^2 [ k^2 - m_h^2]^3 [ ( k+q)^2 - m_h^2 ] } \right]\end{aligned}\ ] ] now the last integral is finite at @xmath191 and satisfies the requirements of the m - theorem .
it then goes to zero as @xmath200 and @xmath49 .
the other three terms in @xmath196 can be evaluated using standard techniques , and one gets @xmath201\ ] ] using the same techniques to evaluate the rest of terms one finally gets the large @xmath6 contribution of the @xmath185 correction to the @xmath142 self energy @xmath202\ ] ] as a final comment , we would like to point a difference between these large - m techniques and the commonly used expansion in powers of the external momenta @xmath187 .
this large - m calculation gives us all the existing contributions up to an arbitrary power in the external momenta @xmath187 , as far as they do not vanish in the large-@xmath6 limit . on the other hand , these techniques
provide an extremely easy way to extract the non - vanishing large mass effects of any loop integral .
in this appendix , we present the results of the various terms contributing to the matching equations ( [ mama ] ) . since we are interested mainly in the differences between the sm terms and the corresponding ones of the echl , it is convenient to rewrite eq.([mama ] ) in the following form the direct result from feynman diagrams ] : @xmath203 the tree diagrams contributing to @xmath204 are displayed in fig.(3 ) .
the contributions from the counterterms are generated using eq.([ret ] ) for the 1pi green s functions of the effective theory , @xmath205 , and eq.([reps ] ) for the 1lpi green s functions of the sm , @xmath206 .
for the difference of counterterms , we use the following notation : @xmath207 the loop contributions to @xmath208 come from the explicit evaluation of all the one loop diagrams in fig.(3 ) in the large @xmath6 limit . to perform this calculation
we have used the techniques described in appendix a. there are no differences in the tree level contributions @xmath210 the contributions from the counterterms are @xmath211 g^{\mu \nu } g^{\lambda \rho } \nonumber\\ & & + i g^4 s^2 \left [ \left(\delta z_w - \frac{\delta g^2}{g^2 } \right ) \frac{1}{g^2 } + a^b_{11 } \right ] \left ( g^{\mu \rho } g^{\lambda \nu } + g^{\lambda \mu } g^{\nu \rho } \right).\end{aligned}\ ] ] the evaluation of diagrams in fig.(3.a ) gives @xmath212 g^{\mu \nu } g^{\lambda \rho } + \nonumber \\[2 mm ] & & i \frac{g^4 s^2}{16 \pi^2 } \left [ \frac{1}{12 } \left ( \hat{\delta}_\epsilon + \frac{4}{3}\right ) \right ] \left ( g^{\mu \rho } g^{\lambda \nu } + g^{\lambda \mu } g^{\nu \rho } \right).\end{aligned}\ ] ] there are no differences at tree level @xmath214 the contributions from the counterterms are @xmath215 g^{\mu \nu } g^{\lambda \rho } \nonumber\\ & & + i g^4 \frac{s}{c } \left [ \left ( ( \delta z_w - \frac{\delta g^2}{g^2})\frac{1}{g^2 } + a^b_{11}\right ) c^2 \right .
\nonumber \\ & & \left .
\rule[0mm]{0mm}{6 mm } + a^b_3 - a^b_{11 } + a^b_{12 } \right ] \left ( g^{\mu \rho } g^{\lambda \nu } + g^{\lambda \mu } g^{\nu \rho } \right).\end{aligned}\ ] ] the evaluation of diagrams in fig.(3.b ) gives @xmath216 g^{\mu \nu } g^{\lambda \rho } \nonumber \\ & & + i \frac{g^4 s}{c}\frac{1}{16 \pi^2 } \left [ \frac{1}{12 } \hat{\delta}_\epsilon ( c^2 + \frac{1}{2 } ) + \frac{1}{36 } ( 4 c^2 + \frac{11}{4 } ) \right ] \left ( g^{\mu \rho } g^{\lambda \nu } + g^{\lambda \mu } g^{\nu \rho } \right ) .
\nonumber\end{aligned}\ ] ] in this case , there are already differences at tree level since there is one diagram , @xmath217 in fig.(3c ) , with a tree level exchange of a higgs boson .
the large @xmath6 limit of this diagram gives @xmath218 the contributions from the counterterms are @xmath219 g^{\mu \nu } g^{\lambda \rho } \nonumber\\ & & + i \frac{g^4}{c^2 } \left [ \left ( ( \delta z_w - \frac{\delta g^2}{g^2 } ) \frac{1}{g^2 } + a^b_{11}\right ) c^4 - 2 c^2(-a^b_3+a^b_{11}- a^b_{12 } ) \right .
\nonumber\\ & & \left
. \rule[0mm]{0mm}{6 mm } - ( a^b_4 + a^b_6 - a^b_{11}+ 2 a^b_{12 } ) \right ] \left ( g^{\mu \rho } g^{\lambda \nu } + g^{\lambda \mu } g^{\nu \rho } \right).\end{aligned}\ ] ] where @xmath107 is given by the following combination of sm counterterms @xmath220 and the sm counterterms @xmath221 and @xmath222 are given in eq.([smcl ] ) .
the evaluation of the diagrams in fig.(3.c ) gives @xmath223 g^{\mu \nu } g^{\lambda \rho } \nonumber\\ & & + i\frac{g^4}{c^2}\frac{1}{16 \pi^2 } \left [ \frac{1}{12 } \hat{\delta}_\epsilon ( c^4 + c^2 - 1)+\frac{1}{72 } ( 8 c^4 + 11 c^2 -14)\right ] \nonumber \\ & & \left(g^{\mu \rho } g^{\lambda \nu}+g^{\lambda \mu } g^{\nu \rho}\right).\end{aligned}\ ] ] the differences at tree level are given by diagrams @xmath217 and @xmath224 of fig.(3.d ) . in the large @xmath6 limit
we get @xmath225 the contributions from the counterterms are @xmath226 2 g^{\mu \nu } g^{\lambda \rho } \nonumber\\ & & + i g^4 \left [ - \left ( \delta z_w - \frac{\delta g^2}{g^2}\right ) \frac{1}{g^2 } - 2 a^b_3 - a^b_4 - 2 a^b_5 + a^b_8 - 2 a^b_9 \right . \nonumber\\ & & \left .
\rule[0mm]{0mm}{6 mm } + 2 a^b_{13 } + \delta s \right ] \left ( g^{\mu \rho } g^{\lambda \nu } + g^{\lambda \mu } g^{\nu \rho } \right).\end{aligned}\ ] ] the one loop diagrams are displayed in fig.(3.d ) and their evaluation in the large @xmath6 limit gives @xmath227 2 g^{\mu \nu } g^{\lambda \rho } \nonumber \\ & & + i \frac{g^4}{16 \pi^2 } \left [ \frac{37}{24 } \hat{\delta}_\epsilon + \frac{55}{72 } \right ] \left ( g^{\mu \rho } g^{\lambda \nu } + g^{\lambda \mu } g^{\nu \rho } \right).\end{aligned}\ ] ] the differences at tree level are given by diagrams @xmath217 , @xmath224 and @xmath228 of fig.(3.e ) . in the large @xmath6 limit
we get @xmath229 the contributions from the counterterms are @xmath230 \left ( g^{\mu \nu } g^{\rho \lambda } + g^{\mu \rho } g^{\lambda \nu } + g^{\lambda \mu } g^{\nu \rho } \right).\ ] ] the one loop diagrams are displayed in fig.(3.e ) and their evaluation in the large @xmath6 limit gives @xmath231 \left ( g^{\mu \nu } g^{\lambda \rho } + g^{\mu \rho } g^{\lambda \nu } + g^{\lambda \mu } g^{\nu \rho } \right).\end{aligned}\ ] ]
fig.1 : : feynman rules for the echl couplings that differ from the sm .
we have shown only those that are relevant for the present calculation .
fig.2 : : one - loop diagrams that give a leading contribution to the sm combination of counterterms @xmath107 , as explained in the text .
fig.3 : : 1lpi standard model diagrams relevant for the matching of the four - point green s functions of gauge fields up to one loop .
we have calculated all the existing diagrams including gauge , goldstone boson and higgs fields in the loops .
we plot here only the diagrams that do not cancel at both sides of the matching equation , either because they do not exist in the echl ( @xmath232 , @xmath69 ) or because they are different in the echl and the sm ( @xmath100 ) .
we have also restricted our plot to those diagrams that give a non - vanishing contribution to the matching in the large @xmath6 limit .
diagrams @xmath100 have to be calculated also in the effective theory , @xmath233 , using the feynman rules given in fig.1 . + all the momenta are taken incoming and arrows indicate negative charge flux . + * 3.a * diagrams for the @xmath234 green function .
( p1 ) indicates that the diagram obtained exchanging the @xmath235 and @xmath236 external legs has also to be included .
there are a total of 12 @xmath232 diagrams and 3 @xmath100 diagrams .
+ * 3.b * @xmath237 green function .
( p1 ) represents the diagram obtained exchanging the @xmath235 and @xmath236 external legs .
there are 18 @xmath232 and 3 @xmath100 diagrams .
+ * 3.c * @xmath238 green function .
( p1 ) represents the diagram obtained exchanging the @xmath235 and @xmath236 external legs .
there are 47 @xmath232 , 6 @xmath100 and 1 @xmath69 diagrams .
+ * 3.d * @xmath239 green function .
( p3 ) indicates that the three diagrams obtained by exchanging the ( @xmath240 ) , ( @xmath241 ) , and ( @xmath242 ) external legs have to be also included .
( p3) indicates the substitutions ( @xmath241 ) , ( @xmath243 ) and ( @xmath244 ) .
( p1 ) indicates the exchange ( @xmath241 ) .
there are 58 @xmath232 , 12 @xmath100 and 2 @xmath69 diagrams .
+ * 3.e * @xmath245 green function .
( p5 ) indicates the five diagrams obtained by the following permutations of the @xmath246 external z s : @xmath247 and @xmath248 .
( p2 ) indicates the exchange of @xmath246 by @xmath249 and @xmath250 .
there are 75 @xmath232 , 18 @xmath100 and 3 @xmath69 diagrams .
99 m. veltman , nucl . phys .
* b123 * ( 1977 ) , 89 .
see , for instance : d. schaile , `` precision tests of the electroweak interaction '' , talk presented at the 27-ichep conference , glasgow , 20 - 27 july 1994 .
cdf collaboration : f. abe et al .
* 73 * ( 1994 ) , 225 ; phys . rev .
* d50 * ( 1994 ) , 2966 .
m.veltman , acta phys . pol . * b8 * ( 1977 ) , 475 .
t.appelquist and c.bernard , phys . rev .
* d22 * ( 1980 ) , 200 ; + t.appelquist _ in _ gauge theories and experiments at high energies , ed .
k.c.browner and d.g.sutherland , scottish u. summer school , 1980 .
a.c.longhitano , nucl . phys .
* b188 * ( 1981 ) , 118 ; phys . rev .
* d22 * ( 1980 ) , 1166 .
o. cheyette and m.k .
b197 * ( 1987 ) , 205 ; + o. cheyette , nucl . phys .
* b297 * ( 1988 ) , 183 ; + r. akoury and y.p .
yao , phys .
rev . * d25 * ( 1982 ) , 1605 . m.j .
herrero and e. ruiz morales ; nucl .
b418 * ( 1994 ) , 431 ; + m.j .
herrero and e. ruiz morales , ftuam94/12 , hep - ph 9408252 . for reviews on the use of effective lagrangians in electroweak physics see .
+ f. feruglio , int .
. phys . * a8 * ( 1993 ) , 4937 .
einhorn , `` beyond the standard model with effective lagrangians '' , um - th-93 - 12 .
a. dobado and m.j .
herrero , phys .
lett * b228 * ( 1989),495 ; * b233 * ( 1989),505 .
+ j. donoghue and c. ramirez , phys .
b234 * ( 1990 ) , 361 .
b. holdom and j. terning , phys . lett .
* b247 * ( 1990 ) , 88 .
a. dobado , d. espriu and m.j .
herrero , phys . lett .
* b255 * ( 1991 ) , 405 .
d. espriu and m.j .
herrero , nucl . phys .
* b373 * ( 1992 ) , 117 .
s. dawson and g. valencia , nucl .
b352 * ( 1991 ) , 27 .
+ j. barger , s. dawson and g. valencia , nucl .
* 399 * ( 1993 ) , 364 .
t. appelquist and g .- h wu , phys
. rev . * d48 * ( 1993 ) , 3235 . s. weinberg , physica * 96a * ( 1979 ) , 327 . j. gasser and h. leutwyler , ann
( n.y . ) * 158 * ( 1984 ) , 142 ; j. gasser and h. leutwyler , nucl .
b250 * ( 1985 ) , 465 .
h. georgi , ann .
rev . nucl . part .
* 43 * ( 1993 ) , 209 and references therein .
m. bilenky and a. santamara , nucl . phys .
* b420 * ( 1994 ) , 47 .
a. dobado , m.j .
herrero and j. terron , z. phys .
* c50 * ( 1991 ) , 205 ; z. phys . * c50 * ( 1991 ) , 465 . m.j . herrero and e.
ruiz morales , phys .
* b296 * ( 1992 ) , 397 .
a. dobado and m.t .
urdiales , ftuam94/29 , ftucm94/19 .
lynn , m.e . peskin and r.g .
stuart , cern-86 - 02 ( 1986 ) .
+ m.e . peskin and t. takeuchi ,
* 65 * ( 1990 ) , 964 ; phys . rev . * d46 * ( 1992 ) , 381
. + m. golden and l. randall , nucl .
b361 * ( 1991 ) , 3 .
d. espriu and j. matias , ub - ecm - pr-94/12 .
w. marciano and s. willenbrock , phys . rev . * d40 * ( 1988 ) , 2509
. s. dawson and s. willenbrock , phys . rev .
* d40 * ( 1989 ) , 2880 .
the equivalence theorem in effective lagrangians has been discussed in : + a. dobado and j.r .
pelez , nucl .
b425 * ( 1994 ) , 110 .
he , y .- p .
kuang and x. li , phys .
b329 * ( 1994 ) , 278 .
a. nyffeler and a. schenk , butp-94/12 , hutp-94/a012 .
m. bohm , h. spiesberger and w. hollik , fortschr .
phys . * 34 * , 11 , ( 1986 ) , 687 .
w. marciano and a. sirling , phys . rev . * d22 * ( 1980 ) , 2697 . + g. burgers and w. hollik , in : polarization at lep , cern yellow report , + ed .
g. alexander et al . ,
cern , geneva , 1989 .
+ m. consoli and w. hollik ; g. burgers and f. jegerlehner in : z physics at lep , + cern yellow report . ed .
g. altarelli et al . , cern , geneva , 1989 .
g.giavarini , c.p.martin and f. ruiz ruiz , nucl .
b381 * ( 1992 ) , 222 . | we study the complete non - decoupling effects of the standard model higgs boson to one loop . using effective field theory methods , we integrate out the higgs boson and represent its non - decoupling effects by a set of gauge invariant effective operators of the electroweak chiral lagrangian . in a previous work
, we analyzed the non - decoupling effects in the two and three - point green s functions of gauge fields .
we complete here the calculation of the chiral effective operators by analyzing the four - point functions .
we discuss in detail the relation between the renormalization of both the standard model and the effective theory , which is crucial for a correct understanding and use of the electroweak chiral lagrangian .
some examples have been chosen to show the applicability of this effective lagrangian approach in the calculation of low energy observables in electroweak theory .
= 6.6truein = 8.7truein = 0.1truein = -1truein = -.7truein plus 0.2pt minus 0.1pt = 44by .5 cm .5 cm plus 1pt -18.5 cm |
galaxies in the early universe have clumpy , highly turbulent , gas - dominated discs @xcite and are vigorously forming stars .
they are believed to sustain their high rates of star formation by the accretion of cold gas along filaments into their central regions @xcite .
the most intense phase of star formation takes place at redshifts 1 - 5 @xcite , with steep decline thereafter @xcite .
it is currently unknown how our milky way ( hereafter mw ) galaxy fits into this general scheme . a fundamental piece of the puzzle is missing in our understanding of the star formation history ( hereafter sfh ) of the mw , and its association with galaxies observed in the early universe .
the aim of this study is to present the first results of a new method to measure the sfh in our galaxy using chemical abundances .
the chemical elements formed and ejected by the early generations of stars during the formation of the mw are now found in the atmospheres of old , long - lived stars .
the absolute and relative abundances of these elements are a gauge of the past star formation as the galaxy evolved . observed variations in the elemental abundance patterns as a function of the age of the stars
can be reproduced with a chemical evolution model to infer the sfh of our galaxy .
the method has the huge advantage that , if stars have different origins in the disc , and follow a uniform evolution ( as is the case for the thick disc , see below ) , the measured sfh is non - local , and valid for the entire population . although the approach is model - dependent , and its validity relies on assumptions which are still largely debated - such as the invariance of the imf with time - it is much more economical and straightforward than standard methods , which aim to reconstruct the galactic sfh by determining individual stellar ages for volume complete samples .
our method does not rely on star density , it uses only abundance variations as a function of age , thus is not limited by number densities - related problems such as the design of volume complete samples , scale height corrections , etc .
moreover , it gives access to detailed variations of the sfh at ancient epochs , a domain that is likely to remain inaccessible with standard methods at least until the advent of the esa astrometric mission , gaia .
counting stars as a function of age for ancient populations is indeed impractical for the time being , simply because there are not enough old stars with sufficiently accurate ages .
we base our study on a spectroscopic dataset of 1111 f , g and k dwarfs stars from the solar vicinity @xcite .
ages were obtained using a bayesian estimate of the probability distribution function for each star @xcite and a severe pruning to only include stars with robust age estimates has been applied . moreover , in the present study we exclude outer thin disc stars ( stars on the thin disc sequence
e.g , low - alpha stars , see @xcite with [ fe / h]@xmath1 -0.2 dex ) , because of their possible separate chemical evolution history @xcite .
this reduces the sample to 281 stars .
1 shows the distribution of [ si / fe ] versus age for the final sample .
it illustrates a change of slope in the declining ratio of [ @xmath2/fe ] at about 8 gyr .
two distinct phases in the abundance enrichment history of the mw are clearly evident @xcite . as in @xcite , we _ define _ the thin and thick discs as corresponding to these two different phases , with a transition between these two formation epochs about 89 gyr ago .
starting from this result , we sought to model the corresponding variation in the sfh needed to reproduce these trends in [ @xmath2/fe ] with time .
distribution of stars at the solar vicinity . both the size and the colour code the metallicity of stars , with
the colour coded according to the vertical scale .
the selected stars show two regimes of [ si / fe ] evolution as a function of time , defining the thick ( ages greater than 8 gyr ) and thin ( ages smaller than 8 gyr ) disc populations @xcite . ]
our modelling approach assumes the following : + _ ( i ) _ all the gas is in the system at the beginning , with no inflow or outflow .
this is similar to a closed - box model @xcite , but is equivalent to saying that most accretion in the inner disc ( r@xmath310 kpc ) occured early . + _
( ii ) _ the ism is initially devoid of metals and remains well mixed during its whole evolution . + _ ( iii ) _ because our aim is to recover the galactic sfh , no coupling between the star formation rate and the gas content is assumed . instead , we determine the sfh by fitting our model to the data .
our approach is otherwise typical of chemical evolution models .
we adopt standard theoretical yields @xcite , an imf @xcite independent of time , and a stellar mass - lifetime relation dependent on the metallicity @xcite .
no instantaneous recycling approximation has been implemented .
we search for the sfh that best fits the [ @xmath2/fe]-age relation ( in this case the alpha element used is the silicon ) , rather than [ @xmath2/fe]-[fe / h ] , as is usually done .
our main reason for doing this is that different star formation histories can produce similar chemical [ @xmath2/fe]-[fe / h ] tracks , thus the solution is degenerate in this plane , without age information .
moreover , the evolutionary [ @xmath2/fe]-[fe / h ] track of the thin disc being very limited @xcite , it is not possible to use this sequence to constrain its detailed evolution . in practice , at t = 0 , the system contains a pristine gas of mass 1 . for each time
step the code looks at the provided sfh and calculates the amount of mass that is converted from the ism into stars and subtracts it from the ism .
the mass of o , mg , fe , si , h , corresponding to the abundance of the ism at that time is also removed from the ism and used to calculate the metallicity of the stellar population .
the code then looks at the stars created in every previous time step and calculates the amount of gas and metals that is returned to the ism .
this is done by interpolating the cumulative yields produced for all stars at the end of their lives and taking into account the metallicity of the star at its formation .
this mass is then added to the ism and the fraction of recycled gas is removed from the star and returned to the gas component .
a chemical track is finally generated resulting from this particular sfh , and fitted to the data through a @xmath4 approach , using the difference between the model value of [ si / fe ] at the age of each star in the data and the observed [ si / fe ] . the data is fitted iteratively . at each iteration
an entire sfh is created and the chemical evolution code is used to calculate the corresponding chemical track .
the algorithm requires an initial guess sfh and a user supplied convergence criterion .
a series of best fits are found by bootstraping the sample , and these best fits are averaged in order to smooth the sfh ( see snaith et al .
( 2014 ) for a detailed description ) .
the result of the fitting procedure to the age - silicon data is given in fig .
2(a ) , with the corresponding sfh in fig .
the star formation obtained suggests that : + ( 1 ) the galaxy underwent an intense phase of star formation between 9 and 13 gyr ago , during the formation of the thick disc . at those epochs ,
the sfr peaked at 34 times the value characterising the subsequent thin disc formation phase ; + ( 2 ) this intense sf phase lasted @xmath0 4 gyr and produced about as much mass in stars as that produced in the following 8 gyr , during the thin disc formation ( fig .
in other words , the thick disc of the galaxy is about as massive as the thin disc ( in agreement with recent mass estimates using structural parameters , see sect .
5.1 ) ; + ( 3 ) the estimated sfh shows a distinct dip at @xmath0 8 - 9 gyr . a statistical analysis ( snaith et al . ,
2014 ) indicates that this dip is robust , and implies that the end of the thick disc formation phase is marked by a decrease in the star formation . a caveat must be added . as mentioned in the introduction ,
although our sample is local ( @xmath3100pc ) , as long as the analysis is based on stars with sufficiently large and different eccentricities , the recovered sfh is non - local , and is valid for the whole population .
while this is true for the thick disc , it becomes less correct for younger stars that have orbits more confined to the solar circle . in this respect
, it is difficult to ascertain if the dip in the sfr at 7 - 9 gyr is a local feature or not .
reciprocally , stars that avoid the solar circle will not be represented in the age - abundance plane , and not taken into account in the reconstructed sfh . from that point of view , our position , at the limit between the inner and outer discs , has an advantage : the solar vicinity is a fair sampling of all known populations of the disc , while we know that the inner disc lacks the metal - poor thin disc stars , and the outer disc lacks both the metal - rich alpha - poor thin disc and the alpha - rich metal - poor thick disc . finally , it is interesting to show the chemical evolution in the [ si / fe]-[fe / h ] plane generated by the recovered sfh .
this is done in fig .
2(d ) , which illustrates the [ si / fe]-[fe / h ] chemical track of the model that results from averaging individual [ si / fe]-[fe / h ] tracks .
each of these tracks has been derived using the best fit sfh solution obtained by fitting the age-[si / fe ] relation through bootstraping the data .
it is striking to see that , even if the model has not been designed to fit this relation , the obtained chemical tracks describe very well the thick disc sequence .
in contrast , the thin disc is less well described , as expected , since the mean [ @xmath2/fe ] and metallicity vary very little during the thin disc phase , while the dispersion in metallicity is very large at any given age ( or [ @xmath2/fe ] ) , due to the position of the sun at the transition between the inner metal - rich and the outer metal - poor discs .
overall , the recovered sfh allows us to match , at the same time , the age-[si / fe ] and the [ si / fe]-[fe / h ] relations , thus guaranteeing that it is a robust description of the chemical evolution of stars observed today at the solar vicinity .
these results lead to a fundamental revision of the role played by the thick disc in galactic evolution , as already suggested by @xcite , and more recently , @xcite . with the recent recognition that our mw s stellar mass is dominated by its disc - the contribution of the central spheroidal component being 10% at most @xcite
- the thick disc is arguably the dominant old stellar population of the mw .
the most recent thick disc scale length measurements ( * ? ? ?
* ; * ? ? ?
* but see jayaraman et al .
( 2013 ) ) show that this population has a much shorter scale length than previously thought , of about 2.0 kpc , while the thin disc scale length is @xmath03.6 kpc . from the stellar surface densities derived from the segue survey data @xcite , and assuming that the thick and thin discs separate at [ @xmath2/fe]=0.25 dex ( on the segue scale , * ? ? ?
* ) , the thin disc contributes to @xmath021 [email protected]@xmath6 to the local stellar surface density , and the thick disc to @xmath08 [email protected]@xmath6 . correcting for the scale length effect assuming the values just quoted , we find that the thick disc represents @xmath047% of the stellar mass within 10 kpc of the galactic centre , in accordance with estimates from the sfh given here . in other words ,
the stellar population responsible for the increase of the metallicity in the galaxy from halo to thin disc values is massive ( see also * ? ? ? * for a similar conclusion ) , and essentially concentrated in the inner galaxy .
this is confirmed by the first apogee results ( * ? ? ?
14 ) , where it is shown that the mdf for stars with median galactocentric radii between 4 and 7 kpc has a mode at [ fe / h]@xmath0 -0.2 dex , with roughly equal amount of stars above and below this value .
interestingly , the distribution of alpha abundance as a function of metallicity given in this same plot shows that below [ fe / h]@xmath0 - 0.2 dex , most stars are alpha - enhanced , meaning that they are thick disc objects . as well as a major contributor to the total mass budget of the galaxy , the thick disc also produced as much metals as the thin disc in the following 8 gyr of galactic evolution . this role has been largely over - looked in chemical evolution models of our galaxy ( all models based on long term infall , e.g. * ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , which were built on the assumption of an apparent lack of intermediate metallicity disc stars in the galaxy - the so - called g - dwarf problem @xcite - simply because these stars are poorly represented in the solar neighbourhood .
a consequence is that galactic chemical evolution models so far have underestimated the importance of this population , leading to a biased view of galactic evolution .
the g - dwarf problem as such vanishes because ( 1 ) the local metallicity distribution function ( mdf ) is strictly local and can not be used to constrain models that aim to give a general description of the chemical evolution of the discs and ( 2 ) stars of intermediate metallicities are in fact largely present in the galaxy and correspond to the thick disc .
this is of crucial importance also for the long - term evolution of the galaxy , since it alleviates the need for prolonged infall onto the inner ( @xmath710 kpc ) disc over the last 8 gyr . over this period
, the mw may have sustained its star formation from the gas not consumed during the thick disc formation epoch and by recycling the mass lost from disc stars @xcite .
such a picture is not in accordance with models of galaxy formation .
we find here that there may have been a period of intense accretion very early in the history of the galaxy but , after that , the mw may have evolved mostly by reprocessing the material lost from stars , and by consuming the gas left at the end of the thick disc formation phase .
there is very little need for significant cosmological gas accretion beyond the early phases of the mw formation .
this may call for a substantial revision of our galaxy evolution models , which suggest that slow , long - term accretion of nearly primordial gas is continuously needed in the disc @xcite .
note that these results are derived from inner disc stars , and are not applicable to the outer disc , which could have followed a different chemical evolution @xcite resulting from a mixture of gas ejected by outflows from the forming thick disc , and pristine accreted gas . in this picture , it is expected that the first ( z@xmath82 ) gas accretion phase mainly fuels the building of the inner regions of the disc , while significant accretion may continue in the outer regions afterwards .
the evolution of our galaxy derived from these results fits well with observational studies of distant galaxies , which indicate that mw progenitors assembled more than half of their stellar mass within 5 gyr after the big bang @xcite .
figure [ results ] ( panels ( _ b _ ) and ( _ c _ ) ) illustrates the evolution of the sfr and cumulative stellar mass as a function of age compared to the one obtain by @xcite for milky way - type progenitors at different redshifts . given possible systematic effects in the age scale , good qualitative agreement is obtained between these estimates and our recovered sfh from solar vicinity data .
these direct observations , together with the properties of local disc galaxies , which do not show any presence of substantial old , classical spheroids @xcite , suggest that the stellar mass formed at these early epochs was likely in thick discs .
moreover , spectroscopic observations of distant galaxies show that their discs have high velocity dispersions in their gas , similar to that measured in the present stellar thick disc of the mw @xcite . for these reasons ,
the formation of the thick stellar disc is a fundamental part of the formation of the mw , and disc galaxies generally .
we have presented the first robust measurement of the mw star formation history using the imprint left on chemical abundances of long - lived stars at the solar vicinity .
our results show that the thick disc formed coincident with the maximum star formation activity in the universe , between 12.5 and 9 gyr ago , and play a fundamental role in the chemical evolution of the galaxy , as well as in its global mass budget .
the implication of these results are far reaching .
firstly , although we have no direct clue of what would be a mdf representative of the entire disc , the fraction of stars of intermediate metallicity is significantly higher than described by the local mdf .
a simple closed - box mdf with a maximum at [ fe / h]@xmath0 - 0.1 dex has a median at [ fe / h]@xmath0 - 0.3 dex , which is about the metallicity transition between the thick and thin discs ( see the age - metallicity relation in @xcite ) , and this is in accord with the equal mass balance measured here between the two discs .
it suggests that the simple closed - box may be a good first order model to describe the chemical evolution of the discs .
secondly , the difference in scale length between the two discs , if confirmed , implies that the solar vicinity mdf is strictly local and _ can not _
be used as it is to infer a general chemical evolution of the galactic disc .
thirdly , these results suggest that large amounts of gas must have been available early in the galaxy to sustain the formation of the thick disc , adding support to the choice of adopting a closed - box model to approximate its chemical evolution .
we find that there may have been a period of intense accretion very early in the history of the galaxy , and that , by z@xmath01 , the mw may have evolved mostly by reprocessing the material lost from stars , and by consuming the gas left at the end of the thick disc formation phase .
this finding is at odds with current galaxy evolution models , which suggest that slow , long - term accretion of nearly primordial gas is continuously needed in the mw disc to sustain its star formation .
we wish thank the referee , klaus fuhrmann , for his prompt report and constructive comments .
the authors acknowledge the support of the french agence nationale de la recherche ( anr ) under contract anr-10-blan-0508 ( galhis project ) . | we report the first robust measurement of the milky way star formation history using the imprint left on chemical abundances of long - lived stars .
the formation of the galactic thick disc occurs during an intense star formation phase between 9.0 ( [email protected] ) and 12.5 gyr ( [email protected] ) ago and is followed by a dip ( at [email protected] ) lasting about 1 gyr .
our results imply that the thick disc is as massive as the milky way s thin disc , suggesting a fundamental role of this component in the genesis of our galaxy , something that had been largely unrecognized .
this new picture implies that huge quantities of gas necessary to feed the building of the thick disc must have been present at these epochs , in contradiction with the long - term infall assumed by chemical evolution models in the last two decades .
these results allow us to fit the milky way within the emerging features of the evolution of disc galaxies in the early universe . |
the free - electron laser flash at desy , based on self - amplified spontaneous emission ( sase ) , is the unique user facility operating in the vuv and the soft x - ray wavelength range . since summer 2005
, it provides coherent femtosecond long radiation to user experiments @xcite .
[ figoverview ] shows the current layout of the facility .
electron bunches are produced in an rf gun and accelerated in six cryomodules ( acc1 - acc6 ) each containing eight nine - cell superconducting cavities of the tesla type and a doublet of superconducting quadrupoles .
to achieve high peak current in the undulator , the bunch is longitudinally compressed in two magnetic chicanes .
downstream the first bunch compressor bc2 ( a four bend chicane ) there is a diagnostic section ( dbc2 ) equipped with several otr screens for the measurement of beam profiles .
a second compression stage takes place after the passage through acc2 and acc3 and is performed using a s - type chicane ( bc3 ) .
the last accelerating section , presently containing three cryomodules ( acc4 , acc5 and acc6 ) , accelerates the beam to the chosen final energy ( presently up to 1 gev ) . then , the electron beam is either guided to a bypass beam line ( not shown in fig .
[ figoverview ] ) or brought through a collimator section to the undulator .
the collimator section has a dogleg shape and contains transverse and energy collimators for undulator protection purposes .
finally , the electron beams from the undulator and bypass are dumped in the same absorber ( dump ) .
we will assume that the transverse particle motion is uncoupled in linear approximation and will use the variables @xmath0 and @xmath1 for the description of the horizontal beam oscillations . here , as usual
, @xmath0 is the horizontal particle coordinate and @xmath1 is the horizontal canonical momentum scaled with the kinetic momentum of the reference particle @xmath2 , and the exact meaning of the subscript @xmath3 will be clarified in the following sections .
if the beam undergoes acceleration , @xmath2 is not constant along the accelerator and in this case there are different methods of including the beam energy variation in the lattice functions which could lead , for example , to lattice functions that , for a given periodic structure , decrease with increasing beam energy .
we will use as definition of the twiss parameters the second order moments of the beam distribution , namely @xmath4 where @xmath5 is the rms emittance . with this definition ,
the twiss matrix @xmath6 satisfies the equation @xmath7 where @xmath8 is path length along the accelerator , matrix @xmath9 is the symplectic unit and the matrix @xmath10 is always symmetric ( see equation ( [ equ_40 ] ) below ) .
the solution of ( [ equ_01 ] ) is given by the formula @xmath11 where the matrix @xmath12 satisfies the equation @xmath13 and is symplectic .
alternatively , the matrix @xmath12 can be expressed using twiss parameters , if they are known , in the familiar form @xmath14 here @xmath15 is a 2 by 2 rotation matrix , @xmath16 is the horizontal phase advance ( defined as usual using the integral of the reciprocal of the betatron function ) and @xmath17 note that , if we assume that the dynamics in the variables @xmath0 and @xmath1 is given by the matrix @xmath18 , then the following identity holds @xmath19
the beam line discussed in this paper starts from the quadrupole doublet of the acc1 cryomodule and ends at the entrance of the sase undulator .
we consider neither the beam dynamics in the gun area nor the choice for the setting of the undulator quadrupoles ( important topic , which deserves separate consideration ) .
optics option i , which was prepared for the commissioning is described in @xcite and is shown in fig .
[ op_op_i ] .
it may look somewhat non - typical for the straight beam line with its two bulges , but it was the only way to satisfy certain lattice constraints , which were considered at that time as important .
these constraints include among others a special choice of twiss parameters in the bunch compressors bc2 and bc3 reduces the emittance growth due to coherent synchrotron radiation ( csr ) .
the selection of the optical functions in the collimator section is determined by the need to suppress the dispersion and to shape the beam envelop suitable for collimation purposes .
beam emittance measurements in sections dbc2 and seed are obtained from the beam sizes measured at four separated locations inside the fodo structure .
this measurement technique provides its best accuracy for periodic twiss functions and a @xmath20 phase advance per cell .
optics option i was used during flash commissioning and during the first months of its operation as user facility .
let the index @xmath21 indicate the design twiss parameters and design setting of acceleration and focusing , and a nonzero @xmath3 stands for the beam motion perturbed by different imperfections .
we will characterize the cumulative effect of these imperfections using the value of the mismatch parameter @xmath22 calculated at the undulator entrance .
if we assume that @xmath23 and that the difference between design and actual twiss parameters at the point @xmath24 has the same order of magnitude , then we will have to lowest order ( which , in fact , is quadratic in @xmath3 ) @xmath25 + \ldots .\end{aligned}\ ] ] here @xmath26 @xmath27 @xmath28 the matrix @xmath29 represents the contribution of initial beam mismatch and the matrix @xmath30 adds the mismatch accumulated due to imperfections in acceleration and focusing . in the interesting case for us , when focusing and acceleration is provided by quadrupoles and rotationally symmetric cavities , the matrix @xmath10 can be expressed using the speed of light approximation as follows @xmath31 here @xmath32 is the quadrupole coefficient and the focusing effect of accelerating field is completely hidden in dynamics of the reference momentum @xmath2 .
let us make more simplifications and , first , assume that the effect of an initial beam mismatch is smaller than the cumulative effect of focusing perturbations ( the procedure of matching the beam parameters at the entrance of the bunch compressor bc2 is well established at flash ) , and , second , neglect in the matrix @xmath33 the difference in rf focusing due to difference in design and actual accelerating field
. then we have @xmath34 @xmath35 + \ldots .\end{aligned}\ ] ] as a final step , we introduce a positive weight function @xmath36 and obtain an estimate @xmath37 where @xmath38 . as criteria for the optics sensitivity to quadrupole errors we choose the value @xmath39 the purpose of the weight function is to make the values of @xmath40 approximately equal in order of magnitude for all quadrupoles and , therefore , it reflects our knowledge or our hypothesis about the error distribution , and is , eventually , a designer choice .
in the design of the optics option ii we used @xmath41 as a weight functions , i.e. we placed the main attention to the relative errors in the quadrupole @xmath42-values .
this was motivated not only by possible uncertainties in the acceleration field distribution along the accelerating modules , but also by uncertainties in knowledge of the transfer coefficients between quadrupole fields and power supply currents .
this optics , which is in operation since spring 2006 and is shown in fig .
[ op_op_ii ] , reduces the sensitivity to quadrupole errors by a factor of two as compared with the previous optics using criteria ( [ eeeq_2 ] ) , but does not show the special behavior of the beta functions in the bunch compressor bc3 and moderately changes the beta functions through the collimator section .
with the operational experience gained and with almost all quadrupoles being remeasured , now we are much more confident in the setting of transverse focusing .
therefore optics option ii was further developed and evolved into a focusing setup which we call optics option ii+ , and which is shown in fig .
[ op_op_iip ] .
this optics , which is in operations since the beginning of 2008 , improves the chromatic beam transfer properties of the optics option ii in the dbc2 and seed sections . in the same time it violates , to a greater or lesser extent , all constraints which were considered as important before the beginning of operations .
we thank the colleagues from the flash team for their interest in our optics studies and for their help during practical optics setup and measurements .
we are specially grateful to h. mais for the interesting discussion and the careful reading of the manuscript . | the aim of the flash linac is to create electron bunches of small transverse emittance and high current for the free - electron laser facility flash at desy .
available operational experience indicates that in order to optimize sase signal at different wavelengths or to fine - tune the fel wavelength , empirical adjustment of the machine parameters is required and , therefore , the sensitivity of the beamline to small changes in the beam energy and in the magnet settings becomes one of the important issues which affects both , the final performance and the reproducibility of the results after breaks in operation . in this article the transverse optics of the flash beamline with low sensitivity to changes in beam energy and quadrupole settings is presented .
this optics is in operation since spring 2006 and has shown a superior performance with respect to the previous setup of the transverse focusing . |
quantum phase transitions ( qpts ) are qualitative changes occuring in the properties of the ground state of a many - body system due to modifications either in the interactions among its constituents or in their interactions with an external probe @xcite , while the system remains at zero temperature .
typically , such changes are induced as a parameter @xmath0 in the system hamiltonian @xmath1 is varied across a point at which the transition is made from one quantum phase to a different one
. often some correlation length diverges at this point , in which case the latter is called a _ quantum critical point_. because thermal fluctuations are inhibited , qpts are purely driven by quantum fluctuations : fluctuations or correlations in the value of some observable or observables that occur in a pure state . thus , these are purely quantum phenomena : a classical system in a pure state can not exhibit correlations .
prominent examples of qpts are the quantum paramagnet to ferromagnet transition occurring in ising spin systems under an external transverse magnetic field @xcite , the superconductor to insulator transition in high-@xmath2 superconducting systems , and the superfluid to mott insulator transition originally predicted for liquid helium and recently observed in ultracold atomic gases @xcite . since entanglement is a property inherent to quantum states and intimately related to quantum correlations @xcite , one would expect that , in some appropriately defined sense , the entanglement present in the ground state undergoes a substantial change across a point where a qpt occurs .
recently , several authors attempted to better understand qpts by studying the behavior of different measures of entanglement in the ground state of exactly solvable models ( see @xcite for representative contributions ) .
such investigations primarily focused on characterizing entanglement using information - theoretic concepts , such as the entropy of entanglement @xcite or the concurrence @xcite , developed for _ bipartite _ systems . in particular , a detailed analysis of the two - spin concurrence has been carried out for the xy model in a transverse magnetic field @xcite , whereas the entanglement between a block of nearby spins and the rest of the chain has been considered in @xcite . while a variety of suggestive results emerge from such studies , in general a full characterization of the quantum correlations near and at a quantum critical point will not be possible solely in terms of bipartite entanglement . identifying the entanglement measure or measures that best capture the relevant properties close to criticality , including the critical exponents and universality class of the transition , remains open problems in quantum information and condensed matter theory . in refs .
@xcite , we introduced _ generalized entanglement _ ( ge ) as a notion extending the essential properties of entanglement beyond the conventional subsystem - based framework .
this notion is general in the sense that it is definable relative to _ any distinguished subset of observables _ , without explicit reference to subsystems , which makes it directly applicable to any algebraic language used to describe the system ( fermions , bosons , spins , etc . )
@xcite . founding the notion on a distinguished set of observables
makes it especially well suited to studying qpts , as our definition makes the existence of ge equivalent to the existence of nonzero correlations or fluctuations in those observables .
the basic idea is that any quantum state gives rise to a reduced state on the distinguished subset of observables @xcite .
these reduced states form a convex set ; as with standard quantum states , there are pure ( extremal ) and mixed ( non - extremal ) ones @xcite .
we define a _ generalized entangled _
pure state , relative to a subspace of observables , to be one whose reduced state on that subspace is _
although we will have little occasion in the present context to apply it to states that are mixed relative to the full set of observables , we extend this notion to include mixed states by defining a generalized entangled mixed state to be one that can not be written as a convex combination of generalized - unentangled pure states . the special case in which the observable set is a lie algebra is often important in physics . in a broad class of such algebras described below , the algebra is not only a subspace of operators , but is such that we can define a natural hermitian projection onto that subspace
. then a simple ( global ) measure of ge for quantum states is provided by what we call the _ purity relative to the algebra_. this is defined as the squared length of the projection of the hermitian operator ( density matrix ) representing the state onto the algebra . as argued in @xcite , if the correct algebra is chosen , the purity contains information about the relevant quantum correlations that uniquely identify and characterize qpts of the system . in this paper , we deepen and expand the analysis initiated in @xcite , by focusing on the detection of qpts due to a broken symmetry as revealed by the behavior of an appropriate relative purity of the ground state . in section [ sectionpurity ]
, we recall the relevant mathematical setting and the definition of the relative purity as a function of the expectation values of the distinguished observables . in section
[ examples ] , we discuss several examples where the relative purity is seen to provide a natural measure of entanglement . in section [ purity - qpt ] , we illustrate some physical criteria that are relevant in choosing the appropriate observable subalgebra and using ge as an indicator of qpts . in sections [ lmg ] and [ xymodel ] we explicitly characterize the qpts present in the so - called lipkin - meshkov - glick ( lmg ) model @xcite and in the one - dimensional spin-1/2 anisotropic xy model in a transverse magnetic field , respectively .
this is done by studying the properties of the purity relative to different algebras of observables in the ground state of both models .
we find the relevant critical exponents for these models , and in the case of the anisotropic xy model in a transverse magnetic field , obtain a new `` global '' disorder parameter , the variance of the number of spinless fermionic excitations in a jordan - wigner - transformed representation of the system .
finally , we provide in separate appendices the details underlying various statements made in the main body of the paper .
these include the relationship between standard separability and ge ( appendix a ) , the ge properties of two special classes of spin states , the cluster and valence bond solid states ( appendix b ) , the proof of the relationship between the local purity and the meyer - wallach entanglement measure ( appendix c ) , and the semiclassical properties of the lmg model in the thermodynamic limit ( appendix d ) .
in the ge approach , entanglement is considered as an _ observer - dependent _ property of a quantum state , which is determined by the physically relevant point of view through the expectation values of a distinguished subset of observables . whenever a preferred decomposition into subsystems is specified in terms of an appropriate ( physical or encoded @xcite ) tensor product structure
, ge becomes identical to standard entanglement provided that distinguished observables corresponding to all _ local _ actions on the individual subsystems are chosen : in particular , for @xmath3 with dim(@xmath4 , standard entanglement of states in @xmath5 is recovered as ge relative to @xmath6 @xcite ( see also appendix [ app1 ] ) .
in fact , the subsystems relative to which standard entanglement is defined ( whether directly identifiable with physical degrees of freedom or related to `` encoded '' or `` virtual '' ones ) are always understandable in terms of appropriate ( associative ) algebras of local observables .
this has been observed before , e.g. in @xcite ( see also @xcite for a recent analysis ) .
however , it is important to realize that the ge notion genuinely extends the standard entanglement definition , and does not coincide with or reduce to it in general . on one hand , this may be appreciated by noticing that even for situations where a subsystem partition is naturally present , states which are manifestly _ separable _ relative to such a partition may possess ge relative to an algebra different from @xmath7 ( see the two spin-1 example of section [ examples ] ) . on the other hand , as also emphasized in @xcite , ge is operationally meaningful in situations where _ no _ physically accessible decomposition into subsystems exists , thus making conventional entanglement not directly definable . as mentioned in the introduction , we will focus on the case where the distinguished observables form a lie algebra @xmath8 of linear operators , acting on a finite - dimensional state space @xmath5 for the system of interest , @xmath9 .
( note that we will not usually distinguish between the abstract lie algebra isomorphic to @xmath8 , and the concrete lie algebra @xmath8 of operators that faithfully represents it on @xmath5 . )
we will assume @xmath8 to be a real lie algebra consisting of hermitian operators , with the bracket of two linear operators @xmath10 and @xmath11 being given by = i ( xy - yx ) .
in this way , operators in @xmath8 can be directly associated with physical observables . for the same reason
, we will also use a slightly nonstandard ( but familiar to physicists ) notion of the lie group generated by @xmath8 , involving the map @xmath12 instead of the mathematicians @xmath13 , for @xmath14 .
no assumption that the lie algebra acts irreducibly on @xmath5 ( i.e. , that it admits no nontrivial invariant subspaces ) will be made , but important consequences of making such an assumption will be discussed
. we will also assume the lie algebra to be closed under hermitian conjugation .
this implies that it is a _ reductive _ algebra ( not to be confused with reducibility of the representation ) . in our context
, a reductive lie algebra is best thought of as the product ( direct sum , as a vector space ) of a finite number of _ simple _ lie algebras , and a finite number of copies of a one - dimensional abelian lie algebra .
simple _ lie algebra is a non - abelian one possessing no nontrivial ideals , where an ideal is a subalgebra invariant under commutation with anything in the algebra ; the relevant property of ideals here is that they can be quotiented out of the algebra , allowing it to be written as a nontrivial product of ideal and quotient ; thus simple lie algebras are non - abelian ones that can not be decomposed into factors , so the factorization used in defining reductive lie algebras above is maximal
. the product ( direct vector - space sum ) of a finite number of simple lie algebras is called _ semisimple _ , and thus a reductive algebra is the product of a semisimple and an abelian part .
the reader is referred to @xcite for relevant background on lie algebras and their representation theory .
as this subsection unfolds , we will summarize much of this representation theory in a way suited to our needs , and the reader should concentrate on understanding the content of the statements , and not vex him or herself unduly about understanding why they are true
. we will consider pure quantum states of @xmath9 , @xmath15 , as well as mixed quantum states of @xmath9 , described by density matrices @xmath16 acting on @xmath5 .
since @xmath8 was assumed closed under hermitian conjugation , the projection of a quantum state @xmath16 onto @xmath8 with respect to the trace inner product is uniquely defined .
let @xmath17 denote the projection map , @xmath18 .
if @xmath16 is a pure state , @xmath19 , the _ purity of @xmath20 relative to @xmath8 _ ( or @xmath8-purity ) is defined as the squared length of the projection according to the trace inner product norm @xcite ; that is p _ ( | ) = tr[(p_(|| ) ) ^2 ] .
[ p0 ] the @xmath8-purity may be explicitly evaluated upon selecting an operator basis @xmath21 for @xmath8 . by assuming the @xmath22 to be hermitian , @xmath23 and orthogonal , @xmath24 eq .
( [ p0 ] ) may be rewritten as @xmath25 = \sum \limits_{\alpha=1}^l \langle a_\alpha \rangle^2 \;,\ ] ] where @xmath26 denotes the expectation value of the observable @xmath22 in the pure state @xmath20 .
an important property following is that the @xmath8-purity is _ invariant _ under group transformations : if a new basis for @xmath8 is introduced by letting @xmath27 , with @xmath28 , @xmath29 , and @xmath30 real numbers , then one finds @xmath31 sometimes it is useful to introduce a common normalization factor @xmath32 in order to set the maximum value of the purity to 1 , in which case eq .
( [ purity0 ] ) becomes @xmath33 as mentioned earlier , a pure quantum state @xmath20 is defined to be _
generalized entangled _ ( _ generalized unentangled _ ) relative to @xmath8 if it induces a mixed ( pure ) state on that set of observables .
when @xmath8 is a complex semisimple lie algebra acting _ irreducibly _ on @xmath5 , it was shown in @xcite ( theorem 14 , part ( 4 ) ) that @xmath20 is generalized unentangled with respect to @xmath8 _ if and only _ if it has maximum @xmath8-purity , and generalized entangled otherwise . under the same assumptions , the abovementioned theorem ( part ( 3 ) )
also leads to the identification of the generalized unentangled pure states as the _ generalized coherent states _ ( gcss ) associated with @xmath8 @xcite . in other words ,
all generalized unentangled states are in the ( unique ) orbit of a minimum weight state of @xmath8 ( taken as a reference state ) under the action of the lie group .
remarkably , gcss are known to possess _ minimum invariant uncertainty _ , @xmath34 $ ] @xcite , so that , similar to the familiar harmonic - oscillator ones , they may be regarded in some sense as closest to classical " states . our characterization theorem for generalized unentangled states on irreducible representations used some standard facts from the theory of semisimple lie algebras and their representations that will also be useful in the discussion of reducible representations in the next subsection .
these are the existence of cartan ( in the semisimple context , maximal abelian ) subalgebras , their conjugacy under the action of the lie group associated with the algebra , and the fact that any finite - dimensional representation , given a choice of cartan subalgebra ( csa ) , decomposes into mutually orthogonal `` weight spaces , '' which are simultaneously eigenspaces of all csa elements .
the map from csa elements to their eigenvalues on a given weight space is a linear functional on the csa called the `` weight '' of that weight space .
the theorem also uses the observation that the projection of the state into the lie algebra is necessarily a hermitian element of that algebra , hence semisimple ( diagonalizable ) , hence belonging to some csa , which we call its _ supporting csa_. frequently , semisimple lie algebras are presented by giving a _ cartan - weyl _ basis , consisting of a set of commuting , jointly diagonalizable operators that generate a csa of the algbera , and a set of so - called `` weyl operators '' that are non - diagonalizable , and act to take a state in one weight space to a state in another ( or else annihilate it ) : in physical examples these are often called `` raising and lowering operators . ''
normalized states correspond to normalized linear functionals on the lie algebra ; when a cartan - weyl basis for the algebra is chosen such that the csa distinguished by the basis is the supporting csa for a given state , the state is zero except on the csa part of the basis . on the csa ,
the state is some convex combination of the weights , that is an element of the _ weight polytope _ ( which is defined as the convex hull of the weights ) .
so it turns out that extremal states on the lie algebra correspond to extremal points of the weight polytope .
this applies regardless of whether the representation is irreducible or not . for irreducible representations ( irreps ) ,
the extremal points of the weight polytope are also highest - weight states of the irrep .
reducible representations are discussed in the next subsection ( along with some comments on reductive algebras ) . in preparation for that
, we introduce another aspect of standard lie theory : the weyl group . besides being able to take any csa to any other csa ,
the lie group also acts on the weight polytope for a given csa , by reflections in a set of hyperplanes through the origin .
the group these generate is called the weyl group .
considered together , the hyperplanes divide the weight space into a set of convex cones , sometimes called _ weyl chambers _ , whose points are at the origin , and whose union with the hyperplanes is the entire space . any such cone can be mapped to any other via the weyl group action , and the weight polytope of the representation is the convex hull of the weyl group orbits of the weights in the closure of any single weyl chamber .
it is important to realize that the relationships just mentioned between maximal purity , generalized coherence , and generalized unentanglement established for a pure state relative to an irreducibly represented algebra @xmath8 do _ not _ automatically extend to the case where @xmath8 acts reducibly on @xmath5 .
we will discuss semisimple algebras first and then , because the algebra we use to analyze the lmg model is abelian , the case of reductive algebras .
if @xmath8 is semisimple , a generic finite - dimensional representation of @xmath8 may be decomposed as a direct sum of irreducible invariant subspaces , @xmath35 , with each of the @xmath36 being in turn the direct sum of its weight spaces .
every irrep appearing in the decomposition has a highest ( or lowest ) weight , and for each of these irreps , there is a manifold of gcss for the irrep constructed as the orbit of a highest weight state for that irrep .
the weight polytope for the _ reducible _ representation will be the convex hull of those for all the irreps contained in it .
because of this , the gcss for these irreps will _ not _ , in general , all satisfy the extremality property that defines generalized unentangled states .
this reflects the fact that even for a state belonging to a specific @xmath8-irrep , ge is a property which depends in general on how the state relates to the whole representation , not solely the irrep . nor is there necessarily a single weight , for one of the constituent irreps , that generates ( as the convex hull of the weyl group orbit ) the weight polytope of the reducible representation .
indeed , the extremal weights in the weight polytope , which correspond to generalized unentangled states , need not all have the same length .
since this squared length is the @xmath8-purity ( as defined in eq .
( [ purity0 ] ) ) of the corresponding state , it is thus no longer the case that all generalized unentangled states have maximal lie - algebraic purity .
however , maximal purity remains a _
sufficient _ , though no longer a necessary , condition for generalized unentanglement .
if the algebra is reductive , the expectations of a maximal commutative subalgebra now include ones for the abelian part of the algebra , i.e. operators that commute with the entire algebra
. these must be proportional to the identity on each irrep , but may have different eigenvalues ( possibly degenerate ) on different irreps .
states on this algebra then involve not just weights for the semisimple part of the algebra , but expectation values for the abelian part of the algebra as well .
these can distinguish different subsets of the irreps , and so irreps whose highest weight for the semisimple part is not extremal for the semisimple part , may become extremal ( generalized unentangled ) in the full reductive algebra .
however , maximal quadratic purity will remain a sufficient , though in general still not necessary , condition for a state being generalized unentangled .
more intuition about ge , purity , and gcss may be gained from simple examples .
consider a physical system which is composed of two spin-@xmath37 ( namely , two qubits ) , and let them be labeled by @xmath38 , with @xmath39 , and corresponding @xmath40 generators @xmath41 , @xmath42 , @xmath43 .
consider ge relative to a _ global _ representation of @xmath40 , whose total - spin generators are @xmath44 .
this representation splits into two irreps , the one - dimensional singlet representation with @xmath45 and the three - dimensional triplet representation with @xmath46 .
the generalized unentangled states relative to this representation of @xmath40 are those for which there exists an @xmath47 such that the state is a @xmath48 eigenstate of @xmath49 . with respect to the csa @xmath50 , those are the states @xmath51 , which are also gcss ( with purity equal to 1 ) . no generalized unentangled state
is contained in the singlet irrep . in particular ,
neither the spin - zero state in the triplet , nor that which spans the singlet , are generalized unentangled ( they both have purity equal to 0 ) , nor are they on highest - weight orbits ( thus gcss ) . as another example consider a single spin-1 system , whose state space @xmath52 carries an irrep of @xmath40 @xcite . in this case , for any choice of spin direction ( say @xmath53 ) only the @xmath54 eigenstates are generalized coherent .
there is also a one - dimensional @xmath55 eigenspace .
the maximal - purity states are also the highest - weight states ; however , the pure @xmath56 eigenstate is not a gcs , has zero purity , and is generalized entangled .
if , for the same system , a distinguished algebra @xmath57 generated by @xmath58 alone is chosen , then the representation reduces as the direct sum of the three invariant one - dimensional subspaces corresponding to @xmath59 . in this case , three different orbits exist in the representation , each of them consisting of only one state up to phases .
however , only the states with @xmath60 are extremal , whereas the state with @xmath56 is not : as one can easily verify from the fact that the reduced state is now just the expectation value of @xmath58 , an equal mixture of a @xmath61 and a @xmath62 state has the same reduced state as a @xmath56 state , so the latter remains , as in the irreducible case , generalized entangled .
a generalization of the latter example , which is relevant to the lmg model we will study in section [ lmg ] , is the case of a spin-@xmath63 system with a distinguished abelian subalgebra generated by @xmath58 .
again , one can see that only the states with maximal magnitude of @xmath58 are generalized unentangled , and only they have maximal purity . by definition ,
note that the relative purity and the invariant uncertainty functionals as defined in the previous section relate to each other via @xmath64 where @xmath65 denotes the quadratic casimir invariant of the lie algebra and @xmath66 is given by eq .
( [ purity0 ] ) ( prior to rescaling ) .
because , by standard representation theory , @xmath67 , with @xmath68 within each irrep , relative purity and invariant uncertainty essentially provide the same information if @xmath8 acts irreducibly .
this , however , is no longer true in general in the reducible case . in the above two - spin-1/2 example , for instance , the two measures agree on the singlet sector ; for triplet states , @xmath69 , thus the invariant uncertainty value is 1 ( same as @xmath66 ) for @xmath60 ( generalized unentangled ) states , whereas it yields 2 for the ( zero - purity ) state with @xmath56 in the triplet sector . for mixed states on @xmath5 , the direct generalization of the squared length of the projection onto @xmath8 as in eq .
( [ p0 ] ) does _ not _ give a ge measure with well - defined monotonicity properties under appropriate generalizations of the locc semigroup of transformations @xcite . a proper extension of the quadratic purity measure defined in the previous section for pure states to mixed states
may be naturally obtained via a standard convex roof construction . if @xmath70 , with @xmath71 and @xmath72 , the latter is obtained by calculating the maximum @xmath8-purity ( minimum entanglement ) over all possible convex decompositions @xmath73 of the density operator @xmath16 as a pure - state ensemble . in general , similarly to what happens for most mixed - state entanglement measures , the required extremization makes the resulting quantity very hard to compute .
while a more expanded discussion of mixed - state ge measures is given in @xcite , we focus here on applying the notion of ge to characterize qpts in different lattice systems .
because the latter take place in the limit of zero temperature , the ground state of the system may be assumed to be pure under ideal conditions .
accordingly , eq . ( [ purity1 ] ) will suffice for our current purposes .
we now apply the concept of relative purity to different physical systems in order to understand its meaning as a measure of entanglement for pure quantum states .
first , we will concentrate on spin systems , showing that for particular subsets of observables , the @xmath8-purity can be reduced to the usual notion of entanglement : the pure quantum states that can be written as a product of states of each party will be generalized unentangled .
however , for other physically natural choices of observable sets , this is no longer the case .
next , we study the @xmath8-purity as a measure of entanglement for fermionic systems , since this is a good starting point for the analysis of the qpt present in the anisotropic xy model in a transverse magnetic field ( section [ xymodel ] ) .
in particular , we show that if a fermionic state can be represented as a single slater determinant , it is generalized unentangled relative to the lie algebra @xmath74 , which is built from bilinear products of fermionic operators .
these examples illustrate how the concept and measure of ge is applicable to systems described by different operator languages , in preparation for the study of qpts .
let us introduce the following representative quantum states for @xmath75 spins of magnitude @xmath76 : @xmath77 where the product state @xmath78 , and @xmath79 denotes the state of the @xmath80th party with @xmath53-component of the spin equal to @xmath81 ( defining the relevant computational basis for the @xmath80th subsystem ) . for simplicity , we begin by studying the ge of a two - qubit system ( two spin-1/2s ) , where the most general pure quantum state can be written as @xmath82 , with the complex numbers @xmath83 , @xmath84 , @xmath85 , and @xmath86 satisfying @xmath87 .
the traditional measures of pure - state entanglement in this case are well understood , indicating that the bell states @xmath88 @xcite ( and its local spin rotations ) are maximally entangled with respect to the local hilbert space decomposition @xmath89 . on the other hand , calculating
the purity relative to the ( irreducible ) lie algebra of all _ local observables _ @xmath90 classifies the pure two - spin-1/2 states in the same way as the traditional measures do ( see fig .
[ twoparties ] ) . here , the operators @xmath91 and @xmath92 are the pauli operators acting on spin 1 and 2 , respectively , and @xmath93 in the basis where @xmath94 and @xmath95 . in this case , eq .
( [ purity1 ] ) simply gives @xmath96 where bell s states are maximally entangled ( @xmath97 ) and product states of the form @xmath98 ( gcss of the local algebra @xmath8 above ) are generalized unentangled , with maximum purity .
therefore , the normalization factor @xmath99 may be obtained by setting @xmath100 in such a product state . as explained in section [ sectionpurity ] , @xmath101 is invariant under group operations , i.e. , in this case , local rotations .
since all gcss of @xmath8 belong to the same orbit generated by the application of group operations to a particular product state ( a reference state like @xmath102 ) , they all consistently have maximum @xmath8-purity ( @xmath100 ) .
another important insight may be gained by calculating the purity relative to the algebra of _ all _ observables for the system , @xmath103 in this case .
one finds that _ any _ two spin-1/2 pure state @xmath20 ( including bell s states ) is then generalized _ un_entangled ( @xmath100 , see also fig .
[ twoparties ] ) .
this property is a manifestation of the relative nature of ge , as considering the set of all observables as being physically accessible is equivalent to not making any preferred subsystem decomposition .
accordingly , in this case any pure quantum state becomes a gcs of @xmath104 .
system . the quantum states
@xmath105 and @xmath106 are defined in eqs .
( [ statedef]).,height=566 ] in fig .
[ twoparties ] we also show the ge for systems of two parties of spin-@xmath76 relative to different algebras .
we observe that the purity reduces again to the traditional concept of entanglement for higher spin if it is calculated relative to the ( irreducible ) lie algebra of _ all _ local observables @xmath107 .
for example , if we are interested in distinguishing product states from entangled states in a two - spin-1s system , we need to calculate the purity relative to the ( irreducible ) algebra @xmath108 , where the @xmath109 hermitian and traceless matrices @xmath110 are the well known gell - mann matrices @xcite : @xmath111 which satisfy @xmath112 = \delta_{\alpha,\beta}$ ] . in this basis , the computational spin-1 states are represented by the 3-dimensional vectors @xmath113 then , the relative purity for a generic pure state @xmath20 becomes @xmath114 where @xmath115 denotes the expectation value of @xmath116 in the state @xmath20 .
in this way , product states like @xmath98 are generalized unentangled ( @xmath100 ) and states like @xmath117 ( and states connected through local spin unitary operations ) , are maximally entangled in this algebra ( @xmath97 ) .
different results are obtained if the purity is calculated relative to a _ subalgebra of local observables_. for example , the two - spin-1 product state @xmath118 where both spins have zero projection along @xmath53 becomes generalized entangled relative to the ( irreducible ) algebra @xmath119 of local spin rotations , which is generated by @xmath120 , the spin-1 angular momentum operators @xmath121 for each spin being given by @xmath122 notice that access to local angular momentum observables suffices to operationally characterize the system as describable in terms of two spin-1 particles ( by imagining , for instance , performing a stern - gerlach - type of experiment on each particle ) .
thus , even when a subsystem decomposition can be naturally identified from the beginning in this case , states which are manifestly separable ( unentangled ) in the standard sense may exhibit ge ( see also appendix [ app1 ] ) . on the other hand
, this is physically quite natural in the example , since there are no su(2 ) @xmath123 su(2 ) group operations ( local rotations ) that are able to transform the state @xmath124 into the unentangled product state @xmath125 .
the examples described in this section together with other examples of states of bipartite quantum systems are shown in fig .
[ twoparties ] .
it is clear that calculating the purity relative to different algebras gives information about different types of quantum correlations present in the system .
the traditional concept of pure multipartite entanglement in an @xmath75 spin-@xmath76 quantum system refers to quantum states that can not be written as a product of states of each party .
the @xmath8-purity distinguishes pure product states from entangled ones if it is calculated relative to the ( irreducible ) algebra of local observables @xmath126 ( see appendix [ app1 ] ) . by eq .
( [ pinvariance ] ) , the measure @xmath101 is invariant under local unitary operations as desired . in particular , the usual concept of entanglement in an @xmath75-qubit quantum state ( @xmath75 spin 1/2s ) can be recovered if the purity is calculated relative to the local algebra @xmath127 , where the pauli operators @xmath128 ( @xmath129 ) are now @xmath130 and the @xmath131 matrices @xmath132 and @xmath133 are given in eq .
( [ pauli1 ] ) . then , the local purity becomes @xmath134 where again the normalization factor @xmath135 is obtained by setting @xmath100 in any product state like @xmath136 ( a gcs in this algebra ) . with this definition ,
states like @xmath137 , @xmath138^{\otimes n}$ ] ( with obvious notations ) , and the cluster states @xmath139 introduced in ref .
@xcite ( see also appendix [ appb ] ) , will be maximally entangled ( @xmath97 ) .
remarkably , as announced in @xcite , after some algebraic manipulations ( see appendix [ app2 ] ) , one can prove that @xmath140 where @xmath141 is the ( pure - state ) measure of _ global entanglement _ for @xmath75 spin 1/2s systems originally introduced by meyer and wallach in @xcite .
a similar relation was independently derived in @xcite .
see also @xcite for additional related considerations . in fig .
[ nparties ] we display some examples of the purity relative to the local algebra @xmath142 for a @xmath75 spin-@xmath76 system .
we also show the purity relative to the algebra of all observables @xmath143^n)$ ] , where any pure quantum state is a gcs , thus generalized unentangled ( @xmath100 ) .
spin-@xmath76 system .
the quantum states @xmath144 , @xmath145 , and @xmath146 are defined in eqs .
( [ statedef]).,width=472,height=566 ] we now apply the concept of ge to a physical system consisting of @xmath75 ( spinless ) fermion modes @xmath148 , each mode being described in terms of canonical creation and annihilation operators @xmath149 , @xmath150 respectively , satisfying the following anti - commutation rules : @xmath151 for instance , different modes could be associated with different sites in a lattice , or to delocalized momentum modes related to the spatial modes through a fourier transform . in general , for any @xmath152 unitary matrix @xmath153 , any transformation @xmath154 maps the original modes into another possible set of fermionic modes . using the above commutation relations
, one also finds that @xmath155 = \delta_{jk } c^{\dagger}_i c^{\;}_l - \delta_{il } c^{\dagger}_k c^{\;}_j \;.\ ] ] thus , the set of bilinear fermionic operators
@xmath156 provides a realization of the unitary lie algebra @xmath147 in the @xmath157-dimensional fock space @xmath158 of the system .
the latter is constructed as the direct sum of subspaces @xmath159 corresponding to a fixed fermion number @xmath160 , with dim(@xmath161 $ ] . for our purposes
, it is convenient to express @xmath147 as the linear span of a hermitian , orthonormal operator basis , which we choose as follows : @xmath162 ( we use henceforth the notational convention that the large left curly bracket means `` is the span of '' ) .
the action of @xmath147 on @xmath163 is reducible , because any operator in @xmath147 conserves the total number of fermions @xmath164 .
it turns out that the irrep decomposition of @xmath147 is identical to the direct sum into fixed - particle - number subspaces @xmath159 , each irrep thus appearing with multiplicity one . using eq .
( [ purity1 ] ) , the @xmath8-purity of a generic pure many - fermion state relative to @xmath147 becomes @xmath165 + \frac{4}{n } \sum\limits_{j=1}^n \langle c^{\dagger}_j c^{\;}_j -1/2 \rangle^2\;.\ ] ] here , we took @xmath166 , for reasons that will become clear shortly . in this case , the fermionic product states ( slater determinants ) of the form @xmath167 , with @xmath168 denoting the reference state with no fermions and @xmath169 labelling a particular set of modes , are the gcss of the @xmath147 algebra @xcite .
because a slater determinant carries a well defined number of particles , each gcs belongs to an irrep space @xmath159 for some @xmath170 , states with different @xmath170 belonging to different orbits under @xmath74 .
a fixed gcs has maximum @xmath8-purity when compared to any other state within the same irrep space .
remarkably , it also turns out that any gcs of @xmath171 gives rise to a reduced state which is extremal ( thus generalized unentangled ) regardless of @xmath170 , the @xmath8-purity assuming the same ( maximum ) value in each irrep . using this property , the normalization factor @xmath172
was calculated by setting @xmath100 in an arbitrary slater determinant .
thus , the purity relative to the @xmath147 algebra is a good measure of entanglement in fermionic systems , in the sense that @xmath100 in any fermionic product state , and @xmath173 for any other state , irrespective of whether the latter has a well defined number of fermions or not .
notice that , thanks to the invariance of @xmath101 under group transformations ( eq .
( [ pinvariance ] ) ) , the property of a state being generalized unentangled is independent of the specific set of modes that is chosen .
this is an important difference between our ge and the mode entanglement approach @xcite .
as already mentioned , although many measures of entanglement have been defined in the literature , assessing their ability to help us better understand qpts in quantum systems largely remains an open problem . in the following two sections we attempt to characterize the qpts present in the lmg model and in the anisotropic xy model in an external magnetic field through the ge notion , relative to a particular subset of observables which will be appropriately chosen in each case .
interestingly , for both these models the ground states can be computed exactly by mapping the set of observable operators involved in the system hamiltonian to a new set of operators which satisfy the same commutation relations , thus preserving the underlying algebraic structure . in the new operator language , the models are seen to contain some symmetries that make them exactly solvable , allowing one to obtain the ground state properties in a number of operations that scales polynomially with the system size ( see also @xcite for related discussions ) .
it is possible then to understand which quantum correlations give rise to the qpts in these cases . several issues should be considered when looking for an algebra @xmath8 of observables that may make the corresponding relative purity a good indicator of a qpt .
a first relevant observation is that in each of these cases a preferred lie algebra exists , where the respective ground state would have maximum @xmath8-purity independently of the interaction strengths in the hamiltonian .
the purity relative to such an algebra remains constant , therefore it does not identify the qpt .
( in these cases , this algebra is in fact the lie algebra generated by the parametrized family of model hamiltonians , as the parameters are varied . )
thus , one needs to extract a subalgebra relative to which the ground state may be generalized entangled , depending on the parameters in the hamiltonian .
a second , closely related observation is that the purity must contain information about quantum correlations which undergo a qualitative change as the transition point is crossed : thus , the corresponding degree of entanglement , as measured by the purity , must depend on the interaction strengths governing the phase transition .
finally , whenever a degeneracy of the ground state exists or emerges in the thermodynamic limit , a physical requirement is that the purity be the same for all ground states .
although these restrictions together turn out to be sufficient for choosing the relevant algebra of observables in the following two models , they do not provide an unambiguous answer when solving a non - integrable model whose exact ground state solution can not be computed efficiently . typically , in the latter cases the ground states are gcss of lie algebras each of whose dimension increases exponentially with the system size . choosing the observable subalgebra that contains the proper information on the qpts ( such as information on critical exponents ) then becomes , in general , a difficult task .
on the other hand , a concept of _ generalized mean - field hamiltonian _ emerges from these considerations . given a hilbert space @xmath5 of dimension @xmath174 ( with @xmath175 an integer @xmath176 )
, we will define a mean - field hamiltonian as an operator @xmath177 that is an element of an irreducibly represented lie algebra of hermitian operators @xmath178 whose dimension scales polynomially in @xmath75 that is , @xmath179 .
when the ground state of such an @xmath180 is non - degenerate , it turns out to be a gcs of @xmath8 @xcite , while the remaining eigenstates ( some of which may also be gcss ) and energies can be efficiently computed .
the connection between lie - algebraic mean - field hamiltonians and their efficient solvability deserves a careful analysis in its own right , which we will present elsewhere @xcite .
originally introduced in the context of nuclear physics @xcite , the lipkin - meshov - glick ( lmg ) model is widely used as a testbed for studying critical phenomena in ( pseudo)spin systems @xcite .
this model was shown to be exactly - solvable in @xcite . in this section ,
we investigate the critical properties of this model by calculating the purity relative to a particular subset of observables , which will be chosen by analyzing the _ classical _ behavior of the ground state of the system . for this purpose ,
we first need to map the model to a _ single _
spin , where it becomes solvable and where the standard notion of entanglement is not immediately applicable .
the model is constructed by considering @xmath75 fermions distributed in two @xmath75-fold degenerate levels ( termed upper and lower shells ) .
the latter are separated by an energy gap @xmath181 , which will be set here equal to 1 .
the quantum number @xmath182 ( @xmath183 or @xmath184 ) labels the level while the quantum number @xmath185 denotes the particular degenerate state in the level ( for both shells , @xmath186 ) .
in addition , we consider a `` monopole - monopole '' interaction that scatters pairs of particles between the two levels without changing @xmath185 .
the model hamiltonian may be written as @xmath187 where @xmath188 , and the fermionic operators @xmath189 ( @xmath190 ) create ( annihilate ) a fermion in the level identified by the quantum numbers @xmath191 and satisfy the fermionic commutation relations given in section [ sectionunpurity ] .
thus , the interaction @xmath192 scatters a pair of particles belonging to one of the levels , and the interaction @xmath193 scatters a pair of particles belonging to different levels .
note that the factor @xmath194 must be present in the interaction terms for stability reasons , as the energy per particle must be finite in the thermodynamic limit .
upon introducing the pseudospin operators @xmath195 which satisfy the @xmath40 commutation relations of the angular momentum algebra , @xmath196 & = & \pm j_{\pm}\ ; , \label{angmom0}\\ \left [ j_+ , j_- \right ] & = & j_z \ ; , \label{angmom}\end{aligned}\ ] ] the hamiltonian of eq .
( [ lmghamilt1 ] ) may be rewritten as @xmath197 as defined by eq .
( [ lmghamilt2 ] ) , @xmath198 is invariant under the @xmath199 inversion symmetry operation @xmath200 that transforms @xmath201 , and it also commutes with the ( casimir ) total angular momentum operator @xmath202 .
therefore , the non - degenerate eigenstates of @xmath198 are simultaneous eigenstates of both @xmath200 and @xmath203 , and they may be obtained by diagonalizing matrices of dimension @xmath204 ( whereby the solubility of the model ) .
notice that , by definition of @xmath58 as in eq .
( [ pseudospin3 ] ) , the maximum eigenvalue of @xmath58 and @xmath205 is @xmath206 . in particular , for a system with @xmath75 fermions
as assumed , both the ground state @xmath207 and first excited state @xmath208 belong to the largest possible angular momentum eigenvalue @xmath209 @xcite ( so - called half - filling configurations ) ; thus , they can be computed by diagonalizing a matrix of dimension @xmath210 .
the hamiltonian ( [ lmghamilt2 ] ) does not exhibit a qpt for finite @xmath75 .
it is important to remark that some critical properties of the lmg model in the thermodynamic limit @xmath211 can be understood by using a semiclassical approach @xcite ( note that the critical behavior is essentially mean - field ) : first , we replace the angular momentum operators in @xmath212 ( with @xmath198 given in eq.([lmghamilt2 ] ) ) by their classical components ( fig .
[ angmomcoord ] ) ; that is @xmath213 where @xmath214 is the resulting classical hamiltonian and @xmath215 , @xmath216 . in this way
, one can show that in the thermodynamic limit ( see appendix [ app3 ] ) @xmath217 so the ground state energy per particle @xmath218 can be easily evaluated by minimizing @xmath219 as mentioned , the ground and first excited states have maximum angular momentum @xmath220 . in fig .
[ classicreplmg ] we show the orientation of the angular momentum in the ground states of the classical hamiltonian @xmath214 , represented by the vectors @xmath221 , @xmath222 , and @xmath223 , for different values of @xmath224 and @xmath225 .
when @xmath226 we have @xmath227 and the classical angular momentum is oriented in the negative @xmath53-direction . however , when @xmath228 we have @xmath229 and the classical ground state becomes two - fold degenerate ( notice that @xmath214 is invariant under the transformation @xmath230 ) . in this region and for @xmath231 the angular momentum is oriented in the @xmath232 plane ( @xmath233 ) while for @xmath234 it is oriented in the @xmath235 plane ( @xmath236 ) .
the model has a gauge symmetry in the line @xmath237 , @xmath238 , where @xmath239 can take any possible value .
going back to the original hamiltonian of eq .
( [ lmghamilt1 ] ) , the quantum system undergoes a second order qpt at the critical boundary @xmath240 , where for @xmath241 the ground and first excited states @xmath207 and @xmath208 become degenerate in the thermodynamic limit and the inversion symmetry @xmath200 breaks .
the order parameter is given by the mean number of fermions in the upper shell @xmath242 , which in the thermodynamic limit converges to its classical value , @xmath243 obviously , for @xmath244 we have @xmath245 , and @xmath246 otherwise ( see fig.[classicreplmg ] ) .
the critical exponents of the order parameter are easily computed by making a taylor expansion near the critical points ( @xmath247 ) . defining the quantities @xmath248 and @xmath249
, we obtain @xmath250 where the critical exponents are @xmath251 and @xmath252 . in fig .
[ lmgenergyplot ] we show the exact ground state energy per particle @xmath218 ( with @xmath253 ) as a function of @xmath224 and @xmath225 in the thermodynamic limit ( eqs .
( [ lmgenergy ] ) ) .
one can see that also in the broken symmetry region ( @xmath254 ) the system undergoes a first order qpt at @xmath237 ; that is , the first derivative of the ground state energy with respect to @xmath224 is not continuous in this line .
the standard notion of entanglement is not directly applicable to the lmg model as described by eq .
( [ lmghamilt2 ] ) , for this is a single spin system and no physically natural partition into subsystems is possible .
therefore , using the @xmath8-purity as a measure of entanglement becomes an advantage from this point of view , since the latter only depends on a particular subset of observables and no partition of the system is necessary .
the first required step is the identification of a relevant lie algebra of observables relative to which the purity has to be calculated .
since both the ground and first excited states of the quantum lmg model may be understood as states of a system carrying total angular momentum @xmath209 , a first possible algebra to consider is the @xmath255 algebra acting on the relevant @xmath256-dimensional eigenspace .
relative to this algebra , @xmath257 is generalized unentangled for arbitrary values of @xmath258 thus the corresponding purity remains constant and does not signal the qpts .
however , the family of hamiltonians ( [ lmghamilt2 ] ) do not generate this lie algebra , but rather an @xmath40 algebra , so perhaps @xmath255 is not a natural choice physically @xcite .
thus a natural choice , suggested by the commutation relationships of eqs .
( [ angmom0 ] ) and ( [ angmom ] ) , is to study the purity relative to the spin-@xmath206 representation of the angular momentum lie algebra @xmath259 : @xmath260 \;,\ ] ] where the normalization factor @xmath261 is chosen to ensure that the maximum of @xmath101 is equal to 1 . with this normalization factor
, @xmath101 can be calculated exactly in the thermodynamic limit by relying on the semi - classical approach described earlier ( see appendix [ app3 ] and eq .
( [ angmomtrans ] ) ) . for @xmath237 and arbitrary @xmath262 ,
@xmath263 which is a gcs of @xmath40 and has @xmath100 . for generic interaction values such that @xmath264 , the classical angular momentum depicted in fig .
[ classicreplmg ] is oriented along the @xmath53-direction and is not degenerate : because @xmath265 , only @xmath266 contributes to @xmath101 ; by recalling that @xmath267 , this gives @xmath100 , so that _ as far as relative purity is concerned the ground state behaves asymptotically like a coherent state in the thermodynamic limit_. physically , this means that with respect to the relevant fluctuations , gcss of @xmath40 are a good approximation of the quantum ground state for large particle numbers , as is well established for this model @xcite .
however , in the region @xmath228 the ground state ( both classical and quantum ) is two - fold degenerate in the @xmath268 limit , and the value of @xmath101 depends in general on the particular linear combination of degenerate states .
this can be understood from fig .
[ classicreplmg ] , where different linear combinations of the two degenerate vectors @xmath222 and @xmath223 imply different values of @xmath269 for @xmath231 and different values of @xmath270 for @xmath234 , while @xmath266 remains constant . with these features
, the purity relative to the @xmath40 algebra will not be a good indicator of the qpt .
an alternative option is then to look at a subalgebra of @xmath40 . in particular , if we only consider the purity relative to the single observable @xmath271 ( i.e. , a particular csa of @xmath40 ) , and retain the same normalization as above , we have @xmath272 this new purity will be a good indicator of the qpt , since @xmath100 only for @xmath264 in the thermodynamic limit , and in addition @xmath101 does not depend on the particular linear combination of the two - fold degenerate states in the region @xmath228 , where @xmath173 .
obviously , in this case @xmath101 is straightforwardly related to the order parameter ( eq .
( [ lmgop ] ) ) ; the critical exponents of @xmath273 are indeed the same ( @xmath251 and @xmath252 ) . note that the purity defined by eq .
( [ jzpuritylmg ] ) does not always take its maximum value for gcss of @xmath274 ( eigenstates of @xmath58 ) . in the region @xmath275 where @xmath276 , the quantum ground state of the lmg model ( eq .
( [ lmghamilt2 ] ) ) does not have a well defined @xmath53-component of angular momentum except at @xmath277 ( @xmath278 \neq 0 $ ] if @xmath279 ) , thus in general it does not lie on a coherent orbit of this algebra for finite @xmath75 . however , as discussed above , it behaves asymptotically ( in the infinite @xmath75 limit ) as a gcs ( in the sense that @xmath280 ) .
moreover , in section [ sectionpurity ] we showed that for @xmath58-eigenstates with eigenvalues @xmath281 , we also obtain @xmath173 . in fig .
[ lmgpurityplot ] we show the behavior of @xmath101 as a function of the parameters @xmath224 and @xmath225 .
interestingly , the purity relative to @xmath58 is a good indicator not only of the second order qpt but also of the first order qpt ( the line @xmath237 , @xmath238 ) . in the ground state of the lmg model.,width=453 ]
in this section , we exploit the purity relative to the @xmath147 algebra ( introduced in section ( [ sectionunpurity ] ) ) as a measure able to identify the paramagnetic to ferromagnetic qpt in the anisotropic one - dimensional spin-1/2 xy model in a transverse magnetic field and classify its universality properties .
the model hamiltonian for a chain of @xmath75 sites is given by ( see fig.[figxymodel ] ) @xmath282 + \sum\limits_{i=1}^n \sigma_z^ i \;,\ ] ] where the operators @xmath283 ( @xmath284 ) are the pauli spin-1/2 operators on site @xmath80 ( defined in eqs .
( [ pauli1 ] ) and ( [ pauli3 ] ) ) , @xmath0 is the parameter one may tune to drive the qpt , and @xmath285 is the amount of anisotropy in the @xmath286 plane . in particular , for @xmath287 eq .
( [ ham1 ] ) reduces to the ising model in a transverse magnetic field , while for @xmath288 the model becomes isotropic .
periodic boundary conditions were considered here , that is @xmath289 , for all @xmath80 and @xmath47 . when @xmath290 and @xmath287 the model is ising - like .
in this limit , the spin - spin interactions are the dominant contribution to the hamiltonian ( [ ham1 ] ) , and the ground state becomes degenerate in the thermodynamic limit , exhibiting ferromagnetic long - range order correlations in the @xmath291 direction : @xmath292 , where @xmath293 is the magnetization in the @xmath291-direction . in the opposite limit where @xmath294 , the external magnetic field becomes important , the spins tend to align in the @xmath53 direction , and the magnetization in the @xmath291 direction vanishes : @xmath295 .
thus , in the thermodynamic limit the model is subject to a paramagnetic - to - ferromagnetic second order qpt at a critical point @xmath296 that will be determined later , with critical behavior belonging to the 2-@xmath297 ising model universality class .
.,width=434 ] this model can be exactly solved using the jordan - wigner transformation @xcite , which maps the pauli ( spin 1/2 ) algebra into the canonical fermion algebra through @xmath298 where the fermionic operators @xmath149 ( @xmath150 ) have been introduced in section [ sectionunpurity ] and @xmath299 is the raising spin operator . in order to find the exact ground state , we first need to write the hamiltonian given in eq .
( [ ham1 ] ) in terms of these fermionic operators , @xmath300 where @xmath301 is an operator that commutes with the hamiltonian , and @xmath302 is the total number operator ( here , we choose @xmath75 to be even ) .
then , the eigenvalue of @xmath200 is a good quantum number , and noticing that @xmath303 we obtain @xmath304 whenever the ( non - degenerate ) eigenstate of @xmath198 is a linear combination of states with an even ( odd ) number of fermions . in particular
, the numerical solution of this model in finite systems ( with @xmath75 even ) indicates that the ground state has eigenvalue @xmath305 , implying anti - periodic boundary conditions in eq .
( [ ham2 ] ) .
the second step is to re - write the hamiltonian in terms of the fermionic operators @xmath306 ( @xmath307 ) , defined by the fourier transform of the operators @xmath149 ( @xmath150 ) : @xmath308 where the set @xmath224 of possible @xmath185 is determined by the anti - periodic boundary conditions in the fermionic operators : @xmath309 $ ] .
therefore , we rewrite the hamiltonian as @xmath310 the third and final step is to diagonalize the hamiltonian of eq .
( [ ham3 ] ) using the bogoliubov canonical transformation @xmath311 where the real coefficients @xmath312 and @xmath313 satisfy the relations @xmath314 where @xmath315 with @xmath316 given by @xmath317 in this way , the quasiparticle creation and annihilation operators @xmath318 and @xmath319 , satisfy the canonical fermionic anti - commutation relations of eq .
( [ anticom ] ) , and the hamiltonian may be finally rewritten as @xmath320 where @xmath321 is the quasiparticle energy . since in general @xmath322
, the ground state is the quantum state with no quasiparticles ( bcs state @xcite ) , such that @xmath323 .
thus , one finds @xmath324 where @xmath168 is the state with no fermions ( @xmath325 ) .
excited states with an even number of fermions ( @xmath305 ) can be obtained applying pairs of quasiparticle creation operators @xmath318 to the @xmath326 state .
however , one should be more rigorous when obtaining excited states with an odd number of particles , since @xmath327 implies periodic boundary conditions in eq .
( [ ham2 ] ) , and the new set of possible @xmath185 s ( wave vectors ) is @xmath328 $ ] ( different of @xmath224 ) . in fig .
[ orderparameter ] we show the order parameter @xmath329 as a function of @xmath0 in the thermodynamic limit and for different anisotropies @xmath330 @xcite . we observe that @xmath331 for @xmath332 and @xmath333 for @xmath334 , so the critical point is located at @xmath335 , regardless of the value of @xmath330 . the value of @xmath296 can also be obtained by setting @xmath336 in eq .
( [ diago ] ) , where the gap vanishes .
notice that the hamiltonian of eq .
( [ ham1 ] ) is invariant under the transformation that maps @xmath337 ( @xmath199 symmetry ) , implying that @xmath338 for all @xmath0 .
however , since in the thermodynamic limit the ground state becomes two - fold degenerate , for @xmath334 , it is possible to build up a ground state where the discrete @xmath199 symmetry is broken , i.e. @xmath339 .
this statement can be easily understood if we consider the case of @xmath287 , where for @xmath340 the ground state has no magnetization in the @xmath291 direction : for @xmath341 , the spins align with the magnetic field , while an infinitesimal spin interaction disorders the system and @xmath342 . on the other hand , for @xmath343 the states @xmath344
$ ] and @xmath345,$ ] with @xmath346 $ ] and @xmath347 $ ] become degenerate in the thermodynamic limit , and a ground state with @xmath348 can be constructed from a linear combination . in the thermodynamic limit as a function of @xmath0 for different anisotropies @xmath330 .
the critical point is at @xmath335 .
, width=434 ] remarkably , this paramagnetic - to - ferromagnetic qpt does not exist in the isotropic limit ( @xmath349 ) . in this case , the hamiltonian of eq .
( [ ham1 ] ) has a continuous @xmath350 symmetry ; that is , it is invariant under any @xmath351 rotation of the form @xmath352 $ ] . since the model is one - dimensional , this symmetry can not be spontaneously broken , regardless of the magnitude of the coupling constants .
nevertheless , a simple calculation of the ground state energy indicates a divergence in its second derivative at the critical point @xmath335 , thus , a second order non - broken symmetry qpt . for @xmath353 all the spins ( in the ground state ) are aligned with the external magnetic field , with total magnetization in the @xmath351 direction @xmath354 , and the quantum phase is gapped . for @xmath355 , the total magnetization in the @xmath351 direction
is @xmath356 , the gap vanishes , and the quantum phase becomes critical ( i.e. , the spin - spin correlation functions decay with a power law ) , with an emergent @xmath350 gauge symmetry @xcite . then , in terms of fermionic operators ( eq .
( [ ham2 ] ) ) , an insulator - metal ( or superfluid ) like second order qpt exists at @xmath296 for the isotropic case , with no symmetry order parameter .
it is a lifshitz transition . the @xmath357 state of eq .
( [ ground ] ) is a gcs of the algebra of observables @xmath358 , spanned by an orthonormal hermitian basis which is constructed by adjoining to the basis of @xmath147 given in eq .
( [ unbasis ] ) the following set @xmath359 of number - non - conserving fermionic operators : @xmath360 then , the @xmath357 state is generalized unentangled with respect to the @xmath361 algebra and its purity @xmath101 ( eq . ( [ purity1 ] ) ) contains no information about the phase transition : @xmath362 .
therefore , in order to characterize the qpt we need to look at the possible subalgebras of @xmath363 . a natural choice is to restrict to operators which preserve the total fermion number that is , to consider the @xmath147 algebra defined in section [ sectionunpurity ] , relative to which the @xmath357 state may become generalized entangled .
( note that as mentioned in section [ sectionunpurity ] , the @xmath147 algebra can also be written in terms of the fermionic operators @xmath364 and @xmath365 , with @xmath185 belonging to the set @xmath224 . ) in the @xmath357 state , @xmath366 only if @xmath367 , thus using eq .
( [ unpurity ] ) the purity relative to @xmath171 is : @xmath368 where the coefficients @xmath313 can be obtained from eqs .
( [ relation1 ] ) and ( [ relation2 ] ) .
in particular , for @xmath341 the spins are aligned with the magnetic field and the fully polarized @xmath369 state is generalized unentangled in this limit ( a gcs of @xmath147 with @xmath100 ) . in the thermodynamic limit , the purity relative to the @xmath147 algebra can be obtained by integrating eq .
( [ unpurity2 ] ) : @xmath370 leading to the following result : @xmath371 & \text { if } g \leq 1/2 \\ \frac{1}{1 + \gamma } & \text { if } g>1/2 \end{array } \right.\ ; .\ ] ] although this function is continuous , its derivative is not and has a drastic change at @xmath372 , where the qpt occurs .
moreover , @xmath101 is minimum for @xmath373 implying maximum entanglement at the transition point and in the ordered ( ferromagnetic ) phase .
remarkably , for @xmath373 and @xmath211 , where the ground state of the anisotropic xy model in a transverse magnetic field is two - fold degenerate , @xmath101 remains invariant for arbitrary linear combinations of the two degenerate states . as defined , for large @xmath0 the purity @xmath101 approaches a constant value which depends on @xmath330 .
it is convenient to remove such dependence in the ordered phase by introducing a new quantity @xmath374 ( shifted purity ) .
we thus obtain @xmath375 & \text { if } g \leq 1/2 \\ 0 & \text { if } g>1/2 \end{array } \right . \;. \label{disorder0}\ ] ] the new function @xmath376 behaves like a _ disorder parameter _ for the system , being zero in the ferromagnetic ( ordered ) phase and different from zero in the paramagnetic ( ordered ) one .
the behavior of @xmath376 as a function of @xmath0 in the thermodynamic limit is depicted in fig .
[ figpur ] for different values of @xmath330 . in the special case of the ising model in a transverse magnetic field ( @xmath287 ) ,
one has the simple behavior @xmath377 for @xmath378 and @xmath379 if @xmath373 . of the @xmath357 as a function of @xmath0 for different anisotropies @xmath330 , eq .
( [ disorder0 ] ) .
@xmath380 behaves like a disorder parameter for this model , sharply identifying the qpt at @xmath335
. , width=529 ] the critical behavior of the system is characterized by a power - law divergence of the _ correlation length _
@xmath181 , which is defined such that for @xmath381 , @xmath382 .
thus , @xmath383 signals the emergence of long - range correlations in the ordered region @xmath373 . near the critical point ( @xmath384 ) the correlation length behaves as @xmath385 ,
where @xmath386 is a critical exponent and the value @xmath387 corresponds to the ising universality class .
let the parameter @xmath388 .
the fact that the purity contains information about the critical properties of the model follows from the possibility of expressing @xmath376 for @xmath389 as a function of the correlation length , @xmath390\ ] ] where a known relation between @xmath0 , @xmath330 , and @xmath391 has been exploited @xcite . performing a taylor expansion of eq .
( [ disorder ] ) in the region @xmath384 , we obtain @xmath392 with @xmath393 and @xmath394 ( fig .
[ scaling ] ) .
thus , the name disorder parameter for @xmath376 is consistent . .
the exponent @xmath393 belongs to the ising universality class.,width=453 ] some physical insight in the meaning of the ground - state purity may be gained by noting that eq .
( [ unpurity2 ] ) can be written in terms of the fluctuations of the total fermion operator @xmath395 @xmath396 where the @xmath357-property @xmath397 has been used .
in general , the purity relative to a given algebra can be written in terms of fluctuations of observables @xcite .
since fluctuations of observables are at the root of qpts it is not surprising that this quantity succeeds at identifying the critical point .
interestingly , by recalling that @xmath398 , the @xmath147-purity can also be formally expressed as @xmath399 where the sum only extends to the non - number - conserving @xmath363-generators belonging to the set @xmath359 specified in eq .
( [ so2nbasis ] ) .
thus , the purity is entirely contributed by expectations of operators connecting different @xmath147-irreps , the net effect of correlating representations with a different particle number resulting in the fluctuation of a _ single _ operator , given by @xmath400 . in fig .
[ fluctuation ] , we show the probability @xmath401 of having @xmath170 fermions in a chain of @xmath402 sites for @xmath287 . we observe that for @xmath373 the fluctuations remain almost constant , and so does the purity .
state for a chain of @xmath402 sites and anisotropy @xmath287.,title="fig:",width=415 ] [ distribution ] again , the isotropic case ( @xmath349 ) is particular in the sense that @xmath100 ( or @xmath379 , see fig . [ figpur ] ) , without identifying the corresponding metal - insulator qpt .
the reason is that in this limit , the hamiltonian of eq .
( [ ham2 ] ) contains only fermionic operators that preserve the number of particles ( i.e. , @xmath403 ) , and the ground state of the system is always a gcs of the @xmath147 algebra .
therefore , in order to obtain information about this qpt , one should look into algebras other than @xmath147 , relative to which the ground state is generalized entangled .
for example , in sec .
[ localalgebra ] we study the purity relative to the local algebra of observables and in fig .
[ figpur2 ] we show that it succesfully identifies the qpt in the isotropic case , being maximum for @xmath404 ( thus implying generalized unentanglement ) . as mentioned , the critical behavior of the xy model in a transverse field has also been investigated by looking at various quantities related to the concurrence , which is intrinsically a measure of bipartite entanglement . for a generic mixed state @xmath16 of two qubits ,
the latter is calculated as @xcite @xmath405 where @xmath406 are the square roots of the eigenvalues of the matrix @xmath407 and @xmath408 .
the concurrence for the reduced density operator @xmath409 of two nearest - neighbor qubits ( @xmath410 ) and next - nearest - neighbor ( @xmath411 ) qubits on a lattice has been investigated in detail in ref .
since , thanks to translational invariance , @xmath409 depends on the qubit indexes only via their distance , we will use the notation @xmath412 , @xmath413 for the resulting quantities as in @xcite .
while the results reported in the above work nicely agree with the scaling behavior expected for this model , the emerging picture based on concurrence can not be regarded as fully satisfactory . as stressed in @xcite , the entanglement as quantified by the nearest - neighbor concurrence
is _ not _ directly an indicator of the qpt in this model , showing maximum entanglement at a point which is not related to the qpt .
however , the derivative @xmath414 of the concurrence with respect to the spin - spin coupling parameter can be seen to diverge logarithmically at the critical point for @xmath394 , and with a power law for the isotropic case [ fig .
( [ isoconc ] ) ] , identifying the critical point in this model .
such a divergence is not found when analyzing , at the isotropic point , other qpts in models of interest , like the one - dimensional anisotropic heisenberg chain ( see , for instance , @xcite ) .
therefore , it suggests that the identification of a critical point using concurrence could be a hard task in general .
state as a function of @xmath0 in the isotropic xy model , @xmath415 .
both curves correspond to the exact solution in the thermodynamic limit .
the value of @xmath416 below @xmath296 is also zero as @xmath412 ( not shown ) .
, width=377 ] finally , we have also investigated the behavior of the purity of the @xmath357 state relative to the algebra of local observables @xmath142 .
( [ purity3 ] ) , this is physically related to the total magnetization @xmath417 along @xmath53 .
the resulting behavior is plotted in fig . [ figpur2 ] as a function of @xmath0 and @xmath330 . as explained in section [ nspinexamples ] , this is a measure of the usual notion of entanglement in the @xmath75-spin-1/2s system . in particular
, the @xmath357 state is unentangled for @xmath418 ( where @xmath419 ) , thus @xmath420 in this limit .
moreover , for @xmath343 we have @xmath421 ( up to local rotations ) , thus @xmath357 becomes maximally entangled , and @xmath422 .
state relative to the local algebra @xmath423 , as a function of @xmath0 for different anisotropies @xmath330 ( @xmath335 ) . the number of sites @xmath402 as in fig .
[ fluctuation ] .
, width=510 ] state relative to the local algebra as a function of @xmath0 for @xmath424 and different lattice sizes.,width=434 ] compared to the purity relative to the @xmath74 algebra , the purity relative to @xmath142 is not as good an indicator of the phase transition when @xmath425 , in the sense that it does not present any drastic change in its behavior .
however , its derivative with respect to the spin - spin coupling parameter diverges at the critical point in this model [ fig .
[ derivative ] ] . only in the isotropic case ( @xmath349 )
the purity relative to the local algebra presents a drastic change at the critical point ( see fig .
[ figpur2 ] ) . in this case , the operator @xmath426 for @xmath427 scales as @xmath428 with the exponent being @xmath429 . on the other hand , this exponent can also be obtained from the purity relative to the local algebra , in the same limit : @xmath430 therefore , this measure of entanglement is also a good indicator of the qpt for the isotropic case .
in this paper , we have explored the usefulness of generalized entanglement ( ge ) for characterizing the broken ( and one example of non - broken ) symmetry quantum phase transitions ( qpts ) present in different lattice systems .
as we focused on situations where the physically relevant observables form a lie algebra , a natural ge measure provided by the relative purity of a state relative to the algebra has been used to identify and characterize these transitions . in sections
[ spinexamples ] and [ nspinexamples ] , we showed using several illustrative examples how the concept of @xmath8-purity can be useful for different spin systems , by encompassing the usual notion of entanglement if the family of all local observables is distinguished .
in addition , the possibility to directly apply the ge notion to arbitrary quantum systems , including indistinguishable particles , was explicitly shown in section [ sectionunpurity ] , using fermionic systems as a relevant case study .
depending on the subset of observables chosen , the @xmath8-purity contains information about different @xmath170-body correlations present in the quantum state , allowing for a more general and complete characterization of entanglement . finally , in sections [ lmg ] and [ xymodel ] we showed that the @xmath8-purity successfully distinguishes between the different phases present in two lattice systems , where the critical points are characterized by a broken symmetry ( or non - broken symmetry in the case of the isotropic xy model in an external magnetic field ) and the usual notion of entanglement can not be straightforwardly applied . as also discussed in section [ purity - qpt ] ,
the most critical step is to determine which subset of observables may be relevant in each case , since the @xmath8-purity must contain information about the quantum correlations that play a dominant role in the qpt . in particular , the ground state of the two models we considered can be exactly calculated and the relevant quantum correlations in the different phases are well understood , thus choosing this subset of observables becomes relatively easy . applying these concepts to a more general case , where the ground state of the system can not be exactly computed , can be done in principle by following the same strategy . however , determining in a systematic way the minimal subset of observables @xmath8 whose purity is able to signal and characterize the qpt , thereby providing the relevant correlations , requires an elaborate analysis .
even more interesting , perhaps , is the open question of finding the minimal number of ge measures , possibly including measures of ge relative to different observable sets , needed to unambiguously characterize the universality class of a transition , obtaining all of its critical exponents .
finally , a fascinating direction for further investigation is to explore the significance of the ge notion within topological quantum - information settings @xcite and to understand what generalizations might be needed to handle topological qpts .
it is a pleasure to thank james gubernatis , leonid gurvits , juan pablo paz , and wojciech zurek for discussions .
we acknowledge support from the us doe through contract no .
h. b. and l. v. gratefully acknowledge financial support from the los alamos office of the director .
given a quantum system @xmath9 whose states @xmath20 belong to a hilbert space @xmath5 of dimension dim@xmath431 , the purity relative to the ( real ) lie algebra of all traceless observables @xmath432 spanned by an orthogonal , commonly normalized hermitian basis @xmath433 , @xmath434 , is , according to eq .
( [ purity1 ] ) , given by : @xmath435 the normalization factor @xmath436 depends on @xmath86 and is determined so that the maximum purity value is 1 . if tr@xmath437 ( as for the standard spin-@xmath438 gell - mann matrices ) , then @xmath439 , whereas in the case tr@xmath440 ( as for ordinary spin-@xmath441 pauli matrices ) , @xmath442 .
recall that any quantum state @xmath15 can be obtained by applying a group operator @xmath153 to a reference state @xmath443 ( a highest or lowest weight state of @xmath444 ) ; that is @xmath445 with @xmath446 , and @xmath447 real numbers . therefore , any quantum state @xmath20 is a gcs of @xmath444 , thus generalized unentangled relative to the algebra of all observables : @xmath448 for all @xmath20 .
let now assume that @xmath9 is composed of @xmath75 distinguishable susbsytems , corresponding to a factorization @xmath449 , with dim@xmath450 , @xmath451 .
then the set of all _ local _ observables on @xmath9 becomes @xmath452 .
an orthonormal basis which is suitable for calculating the local purity @xmath101 may be obtained by considering a collection of orthonormal bases @xmath453 , @xmath454 , each acting on the @xmath148th subsystem that is , @xmath455 where @xmath456 .
then for any pure state @xmath457 one may write @xmath458 \:.\ ] ] by letting @xmath459 be the lie algebra of traceless hermitian operators acting on @xmath460 alone , the above equation also is naturally rewritten as @xmath461 the @xmath462-purity @xmath463 may be simply related to the conventional subsystem purity .
let @xmath464 be the reduced density operator describing the state of the @xmath148th subsystem . because the latter can be represented as @xmath465 one can also equivalently express eq .
( [ puritysub ] ) as @xmath466 \:,\ ] ] that is , @xmath467 . clearly , the maximum value of either eqs .
( [ puritysub1 ] ) or ( [ puritysub2 ] ) will be attained when , and only when , each of the conventional purities @xmath468 @xmath469 @xmath470 for all @xmath148 , which allows determining the @xmath471-normalization factor as @xmath472 accordingly , @xmath473 and the equivalence with the standard notions of separability and entanglement are recovered .
note that for the case of @xmath75 qubits considered in section iiib , the above value simplifies to @xmath474 which in turn gives the purity expression of eq .
( [ purity3 ] ) once the standard unnormalized pauli matrices are used ( @xmath475 , thus removing the overall factor 2 ) .
in ref . @xcite ,
briegel and raussendorf introduced the so - called cluster states for a system of @xmath75 qubits in @xmath297 space dimensions which , in the computational basis , are expressed as @xmath476 where @xmath477 defines the cluster ( @xmath478 ) and @xmath330 denotes some nearest neighbor qubits in the cluster : @xmath479 for @xmath480 , @xmath481 for @xmath482 , @xmath483 for @xmath484 , etc .
we consider @xmath485 when @xmath486 is not in @xmath477 .
the usual notion of entanglement , as applied to a cluster state , is recovered when the @xmath8-purity is calculated relative to the local algebra @xmath487 ( see appendix [ app1 ] ) . for this purpose
, we first calculate the expectation values @xmath488 , with @xmath129 .
one can immediately realize that @xmath489 , @xmath490 , since @xmath491 is an hermitian operator ( i.e. , @xmath492 ) that acting on the @xmath148-th qubit s state ( in the natural basis ) introduces a phase factor @xmath493 , and the coefficients of eq .
( [ cluster ] ) are all real .
moreover , @xmath494 , @xmath490 , since the weight of every state of the natural basis is the same in eq .
( [ cluster ] ) .
in other words , we have a linear combination of basis states where each single qubit has the same probability of pointing up or down .
finally , one can also prove that @xmath495 , @xmath490 .
this can be done by using the eigenvalue equations @xmath496 , for the family of operators @xmath497 , where @xmath498 denotes the set of all nearest - neighbor qubits to the @xmath148-th qubit .
therefore , @xmath499 .
again , since eq .
( [ cluster ] ) is a combination of all the states of the computational basis with the same probability , we obtain @xmath495 . in this way , the @xmath8-purity ( eq . ( [ purity3 ] ) )
is @xmath97 , and the cluster states are maximally entangled relative to the local set @xmath500 .
another important class of spin states is the one defined by the so - called _ valence bond solid _ ( vbs ) states .
these states have been introduced in the context of heisenberg - like magnets , and have been recently revisited in the context of quantum computation @xcite .
their general form is @xmath501 where @xmath502 represent nearest - neighbor bonds of a @xmath297-dimensional lattice of coordination @xmath503 , @xmath504 and @xmath505 are schwinger - wigner boson ( creation ) operators on site @xmath148 whose relation to @xmath40 spin-@xmath76 generators is @xmath506 with the constraint @xmath507 , and @xmath508 .
@xmath509 being an integer makes the possible values of @xmath76 to depend upon the connectivity of the lattice , which is defined by @xmath503 .
we start by showing that the bond operators @xmath510 are invariant under global spin rotations .
the schwinger - wigner boson operators transform as vectors for @xmath40 rotations @xmath511 under an arbitrary spin rotation on lattice site @xmath148 , defined by @xmath512 then , we can use this result to prove that @xmath513 implying , for @xmath514 , @xmath515 therefore , @xmath516 belongs to the singlet irrep of the total spin @xmath517 ( i.e. , @xmath518 .
we want to show now that @xmath519 .
we first observe that @xmath520 , because the transformation that maps @xmath521 and @xmath522 ( i.e. , a global spin rotation about the @xmath523-axis , setting @xmath524 and @xmath525 in eq . ( [ swtransform ] ) ) implies @xmath526 .
then , from the invariance under global spin rotations and the singlet nature of @xmath527 , we obtain @xmath528 .
in other words , the purity relative to the algebra @xmath529 vanishes , meaning that @xmath516 is maximally generalized entangled relative to this algebra . however , in order to make contact with the standard notion of entanglement ( appendix [ app1 ] ) we need to address the ge relative to the algebra @xmath530 , that is , relative to the set of all local observables . for simplicity , we only discuss the 1@xmath297 case for @xmath531 ( i.e. , @xmath532 in eq .
( [ vbs ] ) ) but the reader could use the same techniques to obtain results in higher @xmath297 dimensions and spin magnitude @xmath76 .
the algebra @xmath533 , @xmath534=\delta_{j j ' } \ ( \delta_{\mu'\nu } \ { \cal s}^{j}_{\mu \nu'}-\delta_{\mu \nu ' } \ { \cal s}^{j}_{\nu \mu ' } ) , \label{conm3}\ ] ] can be written in terms of the @xmath40 generators as @xcite @xmath535 \ , \nonumber \\
{ \cal s}^j_{0 1}&=&\frac{1}{2\sqrt{2 } } [ s_-^j + \left \ { s_-^j , s_z^j \right \ } ] \ ,
\nonumber \\ { \cal s}^j_{2 0}&=&\frac{1}{2\sqrt{2 } } [ s_-^j - \left \
{ s_-^j , s_z^j \right \ } ] \ , \nonumber \\ { \cal s}^j_{0 2}&=&\frac{1}{2\sqrt{2 } } [ s_+^j - \left \ { s_+^j , s_z^j \right \ } ] \ , \nonumber \\ { \cal s}^j_{1 2}&= & \frac{i}{2 } \left \ { s_x^j , s_y^j \right \}+(s_x^j)^2 + \frac{1}{2}(s_z^j)^2 - 1 \ , \nonumber \\ { \cal s}^j_{2 1}&= & \frac{1}{2i } \left \ { s_x^j , s_y^j \right \}+(s_x^j)^2 + \frac{1}{2}(s_z^j)^2 - 1 \ , \label{igene}\end{aligned}\ ] ] with @xmath536 . from the spin - rotational invariance of the state @xmath516
we get @xmath537 and , since @xmath531 , we obtain @xmath538 . moreover , the spin - rotational invariance also implies that @xmath539 remains the same constant @xmath540 .
then , for example , applying a global @xmath541-rotation about the @xmath523-axis to @xmath516 ( i.e. , the operation that maps @xmath542 and @xmath543 ) we obtain @xmath544 , hence , @xmath545 .
therefore , the state @xmath516 ( @xmath546 ) is maximally entangled when using the standard notion of entanglement ( @xmath97 , for the algebra of all local observables @xmath547 ) .
in ref . @xcite , meyer and wallach define a measure of entanglement @xmath141 for pure states of qubit systems , that is invariant under local unitary operations ( local rotations ) . for this purpose
, they first define the mapping @xmath548 acting on product states as @xmath549 where @xmath84 and @xmath550 are either the states @xmath551 or @xmath552 , and @xmath553 denotes the absence of the @xmath148-th qubit . on the other hand ,
any @xmath75-qubits pure quantum state can be written in the natural basis ( @xmath53-component of the spin equal to @xmath554 ) as @xmath555 { |{\phi_i}\rangle},\ ] ] where @xmath556 and @xmath557 are complex coefficients , and the orthonormal states @xmath558 of @xmath559 qubits
( absence of the @xmath148-th qubit ) are also written in the natural basis .
therefore , the action of @xmath548 on @xmath20 is @xmath560 then , they define the entanglement @xmath561 as @xmath562 and @xmath563 is @xmath564 therefore , @xmath565 denotes complex conjugate . after some simple calculations we obtain the following relations @xmath566 $ ] . since @xmath567 contains a sum over all qubits ( see eq .
( [ meyer ] ) ) , we finally obtain @xmath568 = 1 - p_{\fh } ( { |{\psi}\rangle})\;,\ ] ] where @xmath101 is the purity relative to the local algebra @xmath569 defined in section [ nspinexamples ] .
as we mentioned in section [ lmg ] , some critical properties of the lmg , such as the order parameter or the ground state energy per particle in the thermodynamic limit , may be obtained using a semi - classical approach . in this section
we sketch a rough analysis of why such approximation is valid ( for a more extensive analysis , see ref .
@xcite ) .
we first define the collective operators @xmath570 where @xmath571 and the fermionic operators @xmath189 ( @xmath190 ) have been defined in section [ lmg ] .
the collective operators satisfy the @xmath572 commutation relations ( section [ sectionunpurity ] ) ; that is @xmath573 = \delta_{\sigma ' \sigma '' } e_{(\sigma,\sigma ' '' ) } - \delta_{\sigma \sigma ' '' } e_{(\sigma'',\sigma ' ) } .\ ] ] if the number of degenerate levels @xmath75 is very large , it is useful to define the intensive collective operators @xmath574 , with commutation relations @xmath575 = \frac{1}{n } \left ( \delta_{\sigma ' \sigma '' } \hat{e}_{(\sigma,\sigma ' '' ) } - \delta_{\sigma \sigma ' '' } \hat{e}_{(\sigma'',\sigma ' ) } \right).\ ] ] therefore , the intensive collective operators commute in the limit @xmath576 , they are effectively classical and can be simultaneously diagonalized .
similarly , the intensive angular momentum operators @xmath577 , @xmath578 , and @xmath579 ( with @xmath49 defined in eqs .
( [ pseudospin1 ] ) , ( [ pseudospin2 ] ) , and ( [ pseudospin3 ] ) ) commute with each other in the thermodynamic limit , so they can be thought of as the angular momentum operators of a classical system .
since the intensive lmg hamiltonian @xmath212 , with @xmath198 given in eq .
( [ lmghamilt2 ] ) , can be written in terms of the intensive angular momentum operators , it can be regarded as the hamiltonian describing a classical system .
the ground state of the lmg model @xmath207 is then an eigenstate of such intensive operators when @xmath268 : @xmath580 , @xmath581 being the corresponding eigenvalue .
in other words , when obtaining some expectation values of intensive operators such as @xmath582 or @xmath212 the ground state @xmath207 can be pictured as a classical angular momentum with fixed coordinates in the three - dimensional space ( see fig .
[ angmomcoord ] ) .
this point of view makes it clear why such operators ought to be intensive .
otherwise , such a classical limit is not valid and terms of order 1 would be important for the calculations of the properties of the lmg model .
obviously , all these concepts can be extended to more complicated hamiltonians such as the extended lmg model , or even hamiltonians including interactions of higher orders as in @xcite .
the definition of entanglement - like notions in systems composed of indistinguishable particles has attracted considerable attention recently .
see for instance k. eckert , j. schliemann , d. bruss , and m. lewenstein , ann .
( n. y. ) * 299 * , 88 ( 2002 ) . similarly , there are ways to rewrite eq .
( [ lmghamilt1 ] ) in terms of some algebras that are products of @xmath75 `` local '' @xmath40 s .
this is related to the approach taken in j. vidal , g. palacios , and r. mosseri , phys .
a * 69 * , 022107 ( 2004 ) which uses a standard entanglement measure , the concurrence , to study phase transitions in the lmg model .
such an approach implies algebras which are much larger than the @xmath40 subalgebra generated by the lmg hamiltonians , and we would argue that the additional detail embodied in the partitions into `` local '' spins is distracting rather than helpful if one wants to study phase transitions in this particular model . indeed , the critical behavior of the lmg model is governed by the expectations of _ collective _ operators . | characterizing and quantifying quantum correlations in states of many - particle systems is at the core of a full understanding of phase transitions in matter . in this work
, we continue our investigation of the notion of generalized entanglement [ barnum _ et al .
_ , phys .
rev .
a * 68 * , 032308 ( 2003 ) ] by focusing on a simple lie - algebraic measure of purity of a quantum state relative to an observable set . for the algebra of local observables on multi - qubit systems ,
the resulting local purity measure is equivalent to a recently introduced global entanglement measure [ meyer and wallach , j. math .
phys .
* 43 * , 4273 ( 2002 ) ] . in the condensed - matter setting , the notion of lie - algebraic purity is exploited to identify and characterize the quantum phase transitions present in two exactly solvable models : the lipkin - meshkov - glick model , and the spin-1/2 anisotropic xy model in a transverse magnetic field . for the latter ,
we argue that a natural fermionic observable - set arising after the jordan - wigner transformation , better characterizes the transition than alternative measures based on qubits .
this illustrates the usefulness of going beyond the standard subsystem - based framework while providing a global disorder parameter for this model .
our results show how generalized entanglement leads to useful tools for distinguishing between the ordered and disordered phases in the case of broken symmetry quantum phase transitions .
additional implications and possible extensions of concepts to other systems of interest in condensed matter physics are also discussed . |
the bethe ansatz method @xcite has been applied to various problems in condensed matter physics .
the heisenberg @xmath0 spin chain , the hubbard model , the kondo model , and many other models are solved by the bethe ansatz or its variants @xcite .
the heisenberg @xmath0 spin chain describes various physical systems including interacting fermion systems . also , it is related to the transfer matrix of the six - vertex model @xcite , the asymmetric exclusion process , and a kardar - parisi - zhang - type growth model @xcite . for the @xmath0 hamiltonian ,
the bethe ansatz expresses energy eigenvalues and eigenstates as functions of quasi - particle momenta which should satisfy a set of algebraic equations , the bethe ansatz equations ( bae ) .
various numerical and analytic methods have been developed to solve the bae , from which the ground state energy and low - lying excitation spectrum are calculated @xcite . combined with a finite - size - scaling theory
, they provide useful tools to study equilibrium and nonequilibrium critical phenomena .
when applying the bethe ansatz method to the heisenberg @xmath0 spin chain , one may doubt whether the bethe ansatz produces the complete set of eigenstates .
the completeness of the bethe ansatz is proved for an inhomogeneous generalization of the @xmath0 chain threaded by the aharonov - bohm flux @xcite .
it implies that the bethe ansatz for the @xmath0 hamiltonian is also complete in a general sense .
one may obtain the eigenstate of the @xmath0 hamiltonian from that of the generalized hamiltonian by taking a limit where the inhomogeneity and the flux vanish .
on the other hand , it has also been investigated for a long time whether the bethe ansatz produces all eigenstates of the @xmath0 hamiltonian without the help of the limiting procedure .
there have been the arguments for the completeness assuming a string conjecture . for the @xmath7 chain , which has isotropic spin - spin interaction ,
takahashi @xcite showed that the number of the states constructed from the bethe ansatz using the string conjecture is equal to the total number of the spin states .
the string conjecture was also adopted to prove the completeness of the bethe ansatz for a generalized @xmath0 chain @xcite and the hubbard model @xcite .
later , the string conjecture was shown to be invalid by essler _
et al . _
@xcite , who analyzed the algebraic bethe ansatz equations in the two down - spin sector .
they found that some solutions expected from the string conjecture are missing and that there exist solutions violating the string conjecture .
nevertheless , they concluded that the bethe ansatz is complete at least in the two down - spin sector because the numbers of the former and the latter are the same .
however one of the solutions found in ref .
@xcite is _ singular _ ; the corresponding bethe ansatz wave function is not well - defined .
in fact the state @xmath8 , which is an eigenstate of the @xmath0 hamiltonian , is not produced from the _ regular _ ( non - singular ) solutions of the bethe ansatz ( see below for the definition of the state @xmath9 ) .
quasi - particle momenta corresponding to the state are divergent @xcite and the bethe ansatz wave function is ill - defined .
an eigenstate that is not produced from the regular solutions of the bethe ansatz will be referred to as a singular state .
siddharthan @xcite reported that there also exist the singular states in the three down - spin sector .
it was claimed that a symmetry property might be important since the singular states have a definite symmetry property .
the singular state does not imply the incompleteness of the bethe ansatz for the @xmath0 hamiltonian .
one may obtain the wave functions for the singular states of the @xmath0 hamiltonian from those of the generalized hamiltonian by taking the limit where the inhomogeneity and the flux vanish ( see , for example , ref .
however it should be examined whether the singular states are relevant to the involved physical quantities of the heisenberg @xmath0 spin chain since , if so , one must first consider the generalized model and take the suitable limit @xcite .
therefore , it is helpful to answer the question why there appear the singular states and how many there are in general down - spin number sectors . in this paper , we investigate the symmetry properties of the @xmath0 hamiltonian and the bethe ansatz wave functions to show that the regular solution of the bethe ansatz with well - defined wave functions does not contain a certain class of eigenstates .
those states are the singular states .
we also estimate the number of the singular states approximately at each sector with fixed number of down spins and find that the number of them diverges exponentially with the chain length .
the heisenberg @xmath0 spin chain of length @xmath10 has the hamiltonian @xmath11 where @xmath12 are the pauli matrices at site @xmath13 , @xmath14 .
the periodic boundary condition is imposed , @xmath15 .
we assume @xmath10 is even unless stated otherwise .
it reduces to the ferromagnetic ( antiferromagnetic ) @xmath7 chain in the case of @xmath16 ( @xmath17 ) . since the hamiltonian commutes with the magnetization @xmath18
, one can work in the @xmath19 down - spin sector in the diagonal basis of @xmath20 .
a state in the @xmath19 down - spin sector is spanned by @xmath21 state vectors @xmath22 with @xmath23 ( @xmath24 for @xmath25 ) denoting the position of the @xmath26th down spin .
the eigenstate of @xmath27 can be written as @xmath28 by @xmath29 where the sum is over all permutations @xmath30 of integers @xmath31 .
the quasi - particle momenta are complex numbers with @xmath32 .
the amplitudes @xmath33 s for different permutations @xmath30 and @xmath34 , which are identical except for two neighboring integers such that @xmath35 , and @xmath36 for @xmath37 , are related as @xmath38 the periodic boundary condition implies the bethe ansatz equations ( bae ) for the quasi - particle momenta : @xmath39 the state constructed from the bethe ansatz is called the bethe state . it has the energy @xmath40 . in the special case of @xmath41 , where the bae becomes @xmath42 ,
the bethe ansatz produces all eigenstates of @xmath27 .
a solution is obtained by selecting @xmath19 different values among the @xmath10 values of @xmath43 integers ( half - integers ) for odd ( even ) @xmath19 .
there are @xmath44 different ways for each @xmath19 , which is equal to the possible spin states .
naively one may think that the solutions at @xmath41 evolve continuously as @xmath45 is turned on .
however it will turn out to be false .
the @xmath0 hamiltonian under the periodic boundary condition commutes with the shift operator @xmath1 , which shifts the position of each down spin to the left by one unit . by construction
, the bethe state is the eigenstate of @xmath1 with the eigenvalue @xmath46 $ ] .
since @xmath47 is the identity operator , @xmath48 should be unity .
the hamiltonian also commutes with the lattice inversion operator @xmath2 defined by @xmath49 with @xmath50 .
since @xmath51 is the identity operator , the eigenvalue of @xmath2 , which will be denoted by @xmath6 , takes the value of @xmath52 .
the inversion operator does not commute with the shift operator in general .
but it is easy to show that they commute with each other in the subspace of @xmath3 ; @xmath53 | \psi \rangle = 0 $ ] if @xmath54
. then the eigenstate of @xmath27 can be made as the simultaneous eigenstate of @xmath1 and @xmath2 in the subspace .
the bethe state with @xmath3 is not the eigenstate of @xmath2 necessarily .
if a bethe state @xmath55 with @xmath3 with well - defined wave function is a simultaneous eigenstate of @xmath2 ( @xmath56 ) , then one can show that @xmath6 should be equal to @xmath4 : @xmath57 it is proved as follows .
suppose that @xmath55 is a bethe state in the @xmath19 down - spin sector with the quasi - particle momenta @xmath58 satisfying eq .
( [ bae ] ) .
the wave function is given as in eq .
( [ psi ] ) .
when one applies the lattice inversion operator @xmath2 to the state @xmath59 , the wave function of the state @xmath60 is given by @xmath61 \ , \label{vwf}\end{aligned}\ ] ] where @xmath62 with a parity permutation @xmath63 which maps @xmath64 to @xmath65 , and we used that @xmath66 = \omega$ ] and @xmath67 .
one can easily verify that @xmath68 satisfies the bae in eq .
( [ bae ] ) and that the amplitude @xmath69 satisfies eq .
( [ perm ] ) with @xmath70 replaced by @xmath68 .
this shows that @xmath71 is a bethe state with the quasi - particle momenta @xmath72 .
therefore a bethe state could be an eigenstate of @xmath2 only when the two sets @xmath58 and @xmath72 are identical .
we rewrite the wave function in eq .
( [ vwf ] ) , introducing the permutation @xmath73 defined by the relation @xmath74 , as @xmath75 \nonumber \\ & = & \omega\sum_p a(\tilde{p}p\bar{p})\exp\left[i\sum_m k_{p_m}x_m\right ] \ .\end{aligned}\ ] ] comparing it with eq .
( [ psi ] ) , one can find @xmath6 from the ratio of @xmath76 and @xmath77 for a certain permutation @xmath30 , e.g. , the identity permutation @xmath78 .
if the amplitude @xmath33 is nonzero and finite , @xmath6 is given by @xmath79 note that the set @xmath70 is equal to @xmath68 .
on one hand , unless both @xmath80 and @xmath81 are present in the set @xmath70 , one can rearrange the momenta in such a way that the permutation @xmath73 and the parity permutation @xmath63 are the same
. then one has @xmath82 and hence @xmath83 .
on the other hand , if both @xmath80 and @xmath81 are present in @xmath70 , @xmath19 should be even and one can rearrange the momenta in the following way : @xmath84 with @xmath85 and @xmath86 . with this arrangement , the permutation @xmath87 maps @xmath88 to @xmath89 , @xmath89 to @xmath88 , and @xmath90 to @xmath91 .
then , using eq .
( [ perm ] ) , the ratio between @xmath92 and @xmath93 is given by @xmath94 therefore we find that @xmath95 if @xmath55 is an eigenstate of @xmath2 .
this proves eq .
( [ lemma ] ) .
equation ( [ lemma ] ) of previous section shows that the bethe state with well - defined wave function can not be the simultaneous eigenstate of @xmath1 and @xmath2 with @xmath5 .
but there exist many eigenstates of @xmath27 with such symmetry .
suppose an eigenstate of @xmath27 possesses the quantum numbers @xmath96 .
the state may be degenerate ( having the same energy ) with another state which has @xmath97 in the same @xmath4 sector . in the degenerate case
, the bethe state may be a linear superposition of the degenerate pair , which does not lead to the contradiction to eq .
( [ lemma ] ) . on the other hand ,
if the eigenstate does not have the degenerate partner with @xmath97 in the same @xmath4 sector , such a state can not be a bethe state , hence is absent in the regular bethe ansatz solutions . in the remaining part ,
we show that there exist nondegenerate states with @xmath5 for nonzero @xmath45 in the general @xmath19 down - spin sector .
it is convenient to work with the second quantized form of the hamiltonian . under the jordan - wigner and fourier transformations , defined as
@xmath98 $ ] and @xmath99 , respectively , the @xmath0 hamiltonian in the @xmath19 down - spin sector reads as @xmath100 where @xmath101 and @xmath102 here @xmath103 and @xmath104 are anticommuting fermion operators with the bare momentum @xmath105 and @xmath106 is the number operator with @xmath107 .
the momentum @xmath108 takes the real value of @xmath109 integer ( half - integer ) for odd ( even ) @xmath19 .
consider a symmetric set @xmath110 which consists of @xmath111 bare momenta satisfying @xmath112 ( mod @xmath113 ) .
a set @xmath114 is symmetric if it contains @xmath115 for all @xmath116 . for a given @xmath110
, @xmath117 is defined as the set of all momenta in the interval @xmath118 except @xmath119 or @xmath120 ( mod @xmath113 ) for @xmath121 .
note that @xmath117 is also a symmetric set .
then , at @xmath41 , the following states @xmath122 with @xmath123 are degenerate eigenstates of @xmath27 in the @xmath19 down - spin sector having @xmath124 . here
@xmath125 denotes the vacuum , i.e. , @xmath126 for all @xmath108 .
they are not the eigenstates of @xmath2 .
one can easily show that @xmath127 figure [ circle ] illustrates an example of the set @xmath110 and corresponding @xmath117 with the states in eq .
( [ sq2 ] ) for @xmath128 and @xmath129 . as @xmath45 is turned on
, @xmath130 generates the overlap between the degenerate states , which may cause the degeneracy to split . to see the degeneracy splitting , we treat @xmath130 as a perturbation and apply the degenerate perturbation theory . in this scheme ,
the leading order correction to the energy is given by the eigenvalue of the perturbation matrix @xmath131 with @xmath132 .
the eigenstates are given by the eigenvectors of the perturbation matrix .
a straightforward calculation shows that the matrix element is given by @xmath133 where we neglect a constant diagonal element .
it has the nondegenerate eigenvector @xmath134 with the eigenvalue @xmath135 .
the other states remain to be degenerate with their common eigenvalue @xmath80 .
if one applies @xmath2 to the state @xmath136 and uses eq .
( [ vps ] ) , one can see that the nondegenerate state is the eigenstate of @xmath2 with @xmath97 while @xmath124 .
therefore , it can not be produced from the regular solution of the bethe ansatz by eq .
( [ lemma ] ) .
the remaining degenerate states may split further in the higher order perturbations .
so the number of the nondegenerate states with @xmath137 is equal to or greater than the number of different realizations of the set @xmath110 . for even @xmath19 ,
@xmath110 consists of @xmath138 values of @xmath108 selected randomly in the interval @xmath139 and their negative partners . for odd @xmath19 , @xmath110 consists of @xmath140 , @xmath141 @xmath108 s randomly - selected in the interval @xmath139 , and their negative partners .
so , there are at least @xmath142 ( @xmath143 ) states which have @xmath137 and they are those states which can not be produced from the bethe ansatz in the even ( odd ) @xmath19 down - spin sector .
if we sum over all @xmath19 , the number of such states is of the order of @xmath144 . in the two down - spin sector ,
the perturbation calculation becomes exact since the states in eq .
( [ sq2 ] ) with the empty set @xmath110 span the whole two down - spin sector with @xmath124 .
so the state @xmath145 is the eigenstate of @xmath27 at all @xmath45 , which can not be produced from the bethe ansatz .
if one performs the inverse jordan - wigner and fourier transformations , the state can be written as @xmath146 .
this is the example of the singular states mentioned in @xcite .
consider also the states @xmath147 with @xmath148 , constructed from a symmetric set @xmath149 consisting of @xmath150 momenta satisfying @xmath151 ( mod @xmath113 ) and the set @xmath117 of all momenta in the interval @xmath152 except @xmath119 or @xmath120 ( mod @xmath113 ) for @xmath153 .
they are degenerate eigenstates of @xmath27 at @xmath41 with @xmath154 .
the same perturbation analysis will show that the degeneracy splits for finite @xmath45 and there appear nondegenerate states with @xmath155 .
those states can not be produced by the bethe ansatz from eq .
( [ lemma ] ) either .
there are @xmath156 ( @xmath157 ) different ways in choosing the set @xmath149 for even ( odd ) @xmath19 . for each set
, there exists at least one state with @xmath158 .
so the total number of such states is again of order of @xmath144 .
the same analysis can be applied to the states from the symmetric sets @xmath159 , with @xmath160 ( mod @xmath113 ) and symmetric sets @xmath161 containing @xmath162 and satisfying @xmath163 ( mod @xmath113 ) . for each case , there are at least a number of order of @xmath144 singular states .
the state counting discussed in the previous section is not exact .
states from different @xmath114 sets may be degenerate at @xmath41 . in that case perturbation calculations
should be carried out in an enlarged space and it is not guaranteed that each @xmath114 set generates at least one singular state .
and the singular states with @xmath5 may become degenerate accidentally with a state with @xmath164 for a particular value of @xmath45 .
then the bethe ansatz may produce the linear superpositions of them .
we believe that those effects do not affect the leading order of magnitude of the total number of the singular states .
we checked this by diagonalizing the @xmath0 hamiltonian exactly with @xmath165 .
the symmetry property of each eigenstate is checked and the nondegenerate states with @xmath158 and ( @xmath166 ) are identified as the singular states .
figure [ level ] shows the energy level , as a function of @xmath45 , of each state with @xmath3 for @xmath167 in the four down - spin sector .
the singular states are denoted by dotted lines .
note that each singular state evolves from the degenerate states at @xmath168 as the degeneracy is lifted for nonzero @xmath45 . at @xmath169 ,
many singular states become degenerate with non - singular states and they are not counted as the singular states . in table 1 , we show the total number of the singular states at @xmath170 , and @xmath171 . due to the accidental degeneracy
the number varies with @xmath45 .
it is confirmed that the number increases approximately as @xmath144 at the three values of @xmath45 .
the singular behaviors of some bethe ansatz solutions have been noticed for the heisenberg @xmath0 model ( see , e.g. , @xcite ) . so the generalizations to the inhomogeneous @xmath0 model with aharonov - bohm flux were considered and their bethe ansatz solutions were shown to be complete @xcite . in those cases the hamiltonian does not possess the lattice inversion symmetry and the present analysis is not applicable .
the singular states include low - lying states in the half - filling case , @xmath172 .
figures 1 ( b ) and ( c ) illustrate degenerate first excited states at @xmath41 in the @xmath154 sector .
they are not the eigenstates of the inversion operator @xmath2 .
for nonzero @xmath45 the degeneracy is removed , and the states evolve into the eigenstates of @xmath2 with @xmath173
. therefore the bethe ansatz fails to produce one of them with @xmath174 .
those states are important in identifying the operator content of the @xmath0 model in the context of the conformal field theory ( see @xcite and references therein ) .
they have the scaling dimension @xmath175 with @xmath176 and @xmath177 . to recover the whole operator content from the bethe ansatz
, one should introduce the aharonov - bohm flux and take the limit where the flux vanishes as done in @xcite .
so far we have assumed @xmath10 is even . for odd @xmath10
, one can not find a pair of bare momenta @xmath108 and @xmath119 satisfying @xmath178 and the previous analysis can not be applied .
so the above arguments are not valid for odd @xmath10 , though eq .
( [ lemma ] ) holds good .
we have performed the same numerical analysis for odd @xmath179 and found that the nondegenerate states with @xmath5 do not exist .
in summary , we have shown that the symmetry properties of the hamiltonian and the bethe wave functions prevent the regular solutions of the bethe ansatz from containing some eigenstates .
for the fixed number of down spins , the @xmath0 hamiltonian can be simultaneously diagonalized with the shift operator @xmath1 and the lattice inversion operator @xmath2 in the subspace where the eigenvalue of @xmath1 , @xmath4 , is @xmath180 , and each simultaneous eigenstate has one of the following four pairs of eigenvalues , @xmath181 .
but there is no regular bethe ansatz wave function with @xmath182 , which leads to the appearance of the singular bethe states .
we estimated the number of the singular states as of @xmath183 for even @xmath10 .
the number of the singular states is exponentially large though the fraction to the total number of the states , @xmath184 , is vanishingly small .
moreover , the singular states include low - lying states in the half - filling case where the number of down spins is @xmath185 .
h. bethe , z. phys .
* 71 * ( 1931 ) 205 . v. e. korepin , n. m. bogoliubov , and a. g. izergin , _ quantum inverse scattering method and correlation functions _ ( cambridge university press , 1993 ) .
r. j. baxter , _ exactly solved models in statistical mechanics _
( academic press , new york , 1982 ) .
b. derrida , phys .
* 301 * ( 1998 ) 65 .
f. c. alcaraz , m. c. barber , and m. t. bachelor , ann . phys .
( n.y . ) * 182 * ( 1988 ) 280 .
d. kim , phys .
e * 52 * ( 1995 ) 3512
. n. m. bogoliubov , a. g. izergin , and v. e. korepin , nucl .
b * 275 * ( 1986 ) 687 .
n. m. bogoliubov , a. g. izergin , and n. yu reshetikhin , j. phys .
a * 20 * ( 1987 ) 5361 .
f. woynarovich and h .-
eckle , j. phys .
a * 20 * ( 1987 ) l97 .
j. d. noh and d. kim , phys .
e * 53 * ( 1996 ) 3225 .
r. p. langlands and y. saint - aubin , preprint v. tarasov and a. varchenko , internat .
notices * 13 * ( 1995 ) 637 .
m. takahashi , prog .
* 46 * ( 1971 ) 401 . a. n. kirillov and n. a. liskova , j. phys .
a * 30 * ( 1997 ) 1209 .
f. h. l. essler , v. e. korepin , and k. schoutens , phys .
* 67 * ( 1991 ) 3848 .
f. h. l. essler , v. e. korepin , and k. schoutens , j. phys .
a * 25 * ( 1992 ) 4115 .
r. siddharthan , cond - mat/9804210 .
a. wal , t. lulek , b. lulek , and e. kozak , int .
b * 13 * ( 1999 ) 3307 . | we investigate symmetry properties of the bethe ansatz wave functions for the heisenberg @xmath0 spin chain .
the @xmath0 hamiltonian commutes simultaneously with the shift operator @xmath1 and the lattice inversion operator @xmath2 in the space of @xmath3 with @xmath4 the eigenvalue of @xmath1 .
we show that the bethe ansatz solutions with normalizable wave functions can not be the eigenstates of @xmath1 and @xmath2 with quantum number @xmath5 where @xmath6 is the eigenvalue of @xmath2 . therefore the bethe ansatz wave functions should be singular for nondegenerate eigenstates of the hamiltonian with quantum number @xmath5 .
it is also shown that such states exist in any nontrivial down - spin number sector and that the number of them diverges exponentially with the chain length .
_ pacs _ : 75.10.jm ; 05.50+q + _ keywords _ : heisenberg @xmath0 spin chain ; singular bethe ansatz solution ; symmetry 2 |
since the seminal work of murakami @xcite , a bilayer ( bl ) of bismuth in the ( 111 ) orientation is theoretically considered @xcite as one of the few two - dimensional ( 2d ) electron systems , besides hgte/(hg , cd)te quantum wells @xcite , which should show the quantum spin hall effect , and enter the quantum anomalous spin hall phase by magnetic doping @xcite . the experimental realization of the bi bl as a 2d topological insulator ( ti ) in a solid state environment remains challenging because of the need to find a suitable substrate which ( i ) enables epitaxial growth of the bi bl and ( ii ) sufficiently decouples the electronic states of the bl from the substrate electrons in order to protect the predicted 1d topological state ( ts ) at the edges of the bl .
recently , it was experimentally shown that an epitaxial bl can be grown on the 3d tis bi@xmath0te@xmath1 @xcite and bi@xmath0se@xmath1 @xcite .
scanning tunneling spectroscopy ( sts ) showed indications for the survival of the 1d ts at the edges of bl islands grown on te terminated surfaces of binary @xcite and ternary @xcite bi - chalcogenides .
however , the edge state s energetic position is still controversial @xcite , partly because the band structure is complicated by the complex coexistence of the 1d bi bl edge state and the 2d ts of the hybrid bi bl - substrate system @xcite . here
, we experimentally investigate the band structure and scattering of the electron system of a bi bl grown on bi@xmath0se@xmath1 by observation of the quasi - particle interference ( qpi ) @xcite via fourier - transform sts ( ft - sts ) @xcite .
the ft - sts images are interpreted with the help of the joint - density of states ( jdos ) approach , which has been used successfully to help the interpretation of ft - sts images for high temperature superconductors @xcite , quantum - hall electron phases in semiconductors @xcite , and , more recently , in the field of tis @xcite .
the jdos method is usually applied with a spectral function @xmath2 determined from angle resolved photoemission spectroscopy ( arpes ) .
deducing @xmath3 from arpes presents three major inconveniences : ( i ) a limitation to the occupied states .
( ii ) very high quality arpes data , which is needed in order to disentangle the spectral features , is often hindered by the challenge of obtaining a high quality crystal surface over micro meter distances .
( iii ) in systems with multiple bands in a small energy and momentum range , the interpretation of arpes data can be problematic due to matrix element effects . instead , here
, a semi - theoretical approach is developed , determining @xmath3 from a density functional theory ( dft ) calculation of the band structure as input for the jdos simulation , where the structural parameters have been optimized in order to fit the arpes - measured band structure .
the jdos approach , in comparison to the more rigorous stationary phase approach @xcite or transfer matrix method @xcite , bears sufficient simplicity in order to easily investigate the effects of different spin - dependent scattering matrices @xmath4 on the simulated qpi patterns . on the other hand ,
the `` on - shell '' assumption used in the jdos approach has been shown to be a sufficiently accurate simplification by comparison between jdos and transfer matrix results for high - temperature superconductors @xcite .
the first objective of this paper is to demonstrate that the spin - dependent jdos method using the dft - calculated surface - localized spectral function applied here , can serve to interpret the complex ft - sts measured qpi patterns of bi bl on bi@xmath0se@xmath1 , involving up to three scattering vectors at a given energy .
the second objective is the experimental investigation of the band structure and electron scattering in this ti .
it is shown that the 2d ts of this system coexists over a large energy range with rashba spin - split quantum well states in the conduction band ( cb ) and valence band ( vb ) of the substrate .
the band gap of the bi bl which hosts the 1d edge state is relocated several hundred mev above the fermi level by charge transfer , and thus may be accessible to transport experiments only after further surface doping of the bl .
scanning tunneling microcopy ( stm ) and spectroscopy ( sts ) experiments were performed in a multi - chamber ultrahigh - vacuum ( uhv ) system with a base pressure below @xmath5 using a home - built variable temperature stm similar to the one described in ref . [ ] . both the tip , electrochemically etched from polycrystalline w wire , and sample were cooled by a continuous flow he cryostat to @xmath6 .
constant current stm images were taken in constant current mode at a tunneling current @xmath7 with the bias voltage @xmath8 applied to the sample .
maps of the differential conductance @xmath9 , called sts images in the following , were measured with closed feedback in constant current mode via lock - in technique with a small modulation voltage @xmath10 added to @xmath8 .
such maps are closely related to the local electron density of states ( ldos ) of the surface at an energy @xmath11 with respect to the fermi energy @xmath12 .
the calculation of the ft - sts images is described below .
arpes measurements of the band dispersion of the sample were performed on the sgm-3 beamline of the astrid synchrotron radiation facility @xcite .
the arpes spectra have been acquired using a photon energy of @xmath13 and the temperature of the sample was kept at @xmath14 .
the energy and angular resolution for arpes measurements were better than @xmath15 and @xmath16 , respectively .
the stoichiometric bi@xmath0se@xmath1 single crystal used as substrate was grown and characterized as described in ref . [ ] , and is highly @xmath17-doped resulting from bulk defects .
the crystalk was cleaved by the scotch tape method inside uhv at room temperature resulting in a termination by an intact quintuple layer @xcite . within 15 min after cleavage ,
the substrate was cooled down to @xmath18 , and bi was deposited at a rate of @xmath19 from a crucible with high purity material using an electron beam evaporator .
the growth temperature was optimized for a smooth growth of a closed ( 111 ) bl .
the coverage of bi was calibrated using stm images of well - known ( 111 ) bl islands on bi@xmath0te@xmath1 @xcite and is given in fractions of a ( 111 ) bl in the following . for the arpes measurements
, the samples were annealed after the growth at @xmath20 for 10 min in order to increase the size of the bl islands in these samples .
the dft calculations were performed using the full - potential linearized augmented plane wave method as implemented in the fleur code @xcite .
dft is employed in the generalized gradient approximation as given by ref .
[ ] including spin - orbit coupling self - consistently .
using the experimental bi@xmath0se@xmath1 lattice parameters , the bi bl was assumed to sit epitaxially on one side of six quintuple - layers of the substrate and the atomic positions of the bl and the first substrate layers were relaxed . in order to match the experimental dispersions , the distance between substrate and bl had to be further increased , as described in sec .
[ sec : dft_spinpol ] .
se@xmath1 ( @xmath21 , @xmath22 ) .
inset : atomically resolved stm image taken on a flat area of the bl ( @xmath23 , @xmath24 ) .
( b ) bi bl islands grown by deposition of 60% bl ( @xmath21 , @xmath22 ) .
( c ) height profile taken along the line in ( b ) . ]
figure [ fig : figure1 ] ( a ) shows an stm image of an almost perfectly closed bi bl on bi@xmath0se@xmath1 .
there are only a few remaining vacancy islands ( dark spots ) where the bi@xmath0se@xmath1 substrate is exposed , and a few islands of excess bi grown on top of the first bl ( bright spots ) .
atomically resolved stm images ( inset ) reveal an in - plane lattice constant of the bl of @xmath25 which is almost pseudomorphic to the bi@xmath0se@xmath1 lattice , resulting in a lateral compression of the bl of about 10% with respect to the bulk bi lattice constant .
deposition of less than 1 bl results in the growth of irregularly shaped bi bl islands on the bare substrate [ fig .
[ fig : figure1 ] ( b ) ] .
note , that , unlike for the case of bi@xmath0te@xmath1 @xcite , it is not possible to grow triangular shaped bl islands , even if the growth temperature is reduced , indicating a weak bonding of the bl to the bi@xmath0se@xmath1 substrate .
stm height profiles [ fig .
[ fig : figure1 ] ( c ) ] reveal a surprisingly large height of the bl of at least @xmath26 , slightly dependent on the bias voltage .
compared to the height of the bi bl grown on si(111 ) of only @xmath27 @xcite which fits with the bulk bi lattice constant along the trigonal c axis , this indicates a large van der waals gap between the bl and the bi@xmath0se@xmath1 substrate of roughly @xmath28 .
there are only few defects at the surface of the bl which appear as atomic size depressions [ see inset of fig . [
fig : figure1 ] ( a ) ] . instead
, the visible extended triangular shaped protrusions are due to subsurface defects , most probably located at the interface between the bl and bi@xmath0se@xmath1 @xcite .
in order to study the electronic structure of the bl on bi@xmath0se@xmath1 system , sts images have been taken on a @xmath29 area of the closed bl without vacancy and second bl islands ( fig .
[ fig : figure2 ] ) . as visible already in the stm image ( a )
, there are qpi patterns centered mostly around the surface defects , which result from the scattering of the electrons at these defects . the dispersion of the qpi patterns visible in the sts images taken as a function of bias voltage @xmath8 ( b - i ) shows a complex behaviour indicating the contribution of more than one electron band .
in addition to the rather weak @xmath9 corrugation due to qpi , there is a rather strong long range contrast which inverts between @xmath30 and @xmath31 , and again between @xmath32 and @xmath33 .
this contrast reveals potential disorder on a length scale of @xmath34 which is probably caused by donor- or acceptor like subsurface defects @xcite . in order to deduce the origin of the electrons contributing to the qpi , the 2d fast fourier - transformations of the sts images in fig .
[ fig : figure2 ] ( b - i ) have been calculated , with the convention of @xmath35 , resulting in so called ft - sts images . in such images taken at a bias voltage @xmath8 , the intensity at a given @xmath36
is related to the probability of an electron scattering from an initial state @xmath37 into a final state @xmath38 with energy @xmath39 and wave vectors @xmath40 and @xmath41 , respectively , where @xmath42 is the scattering vector @xcite .
the following commonly used @xcite image processing steps have been done , in order to reduce the noise in the ft - sts images : first , the sts images have been smoothened by using a median filter with 5 pixel averaging along the slow scan direction .
second , in order to suppress effects due to the infinitely sharp edges of the sts image on the ft , the images have been processed using a window filter with a smooth transition to zero intensity at the rim on a length of about @xmath43 .
third , the calculated ft - sts image is symmetrized accoring to the pseudo sixfold symmetry of the underlying bandstructure .
finally , the ft - sts images have been smoothened using a gaussian filter .
the ft - sts images resulting from the sts images in fig . [ fig : figure2 ] ( b - i ) are shown in ( j - q ) . upon the first inspection
, there are two dominant features : below the fermi energy @xmath44 the ft - sts signal has a hexagonal rim with the flat side along @xmath45-@xmath46 , which disperses to smaller @xmath47 values for increasing energy . above @xmath44 , the hexagon gets increasingly warped and finally transforms into a star - like shape with the six arms along @xmath45-@xmath46 . upon further inspection , there are additional features of smaller size inside the outer hexagon and star , which also disperse as a function of energy , indicating the contribution of two to three bands to the observed qpi throughout the energy window .
we took particular care , that these features we analyzed in the following , were not artificially introduced by the image processing steps described above , but are already visible in the raw 2d fast fourier - transformations of the raw sts images .
moreover , we observed the same energetic evolution of ft - sts features on five sample areas from different cleaves .
the comparison of these features from ft - sts data taken on different sample locations showed some energetic shift of at most @xmath48 , which we assign to a slight lateral variation of the doping by surface defects .
however , the data set we analyzed in the remaining part of this manuscript was taken from one sample location and did nt show such a variation . due to the contribution of many bands to the observed qpi , it is a highly nontrivial task to deduce the bandstructure @xmath49 from the distribution of @xmath36-vectors in the ft - sts - images . on the other hand , by comparison of the @xmath36-vectors from ft - sts images to @xmath49 determined from complementary arpes measurements or dft calculations ,
it is possible to relate the presence or absence of @xmath36-vectors in ft - sts images to the scattering - electron spin dependent matrix elements . for this purpose
, the ft - sts images are simulated from a dft calculated bandstructure and compared to the measured ft - sts images in the following .
the ft - sts images are simulated using the jdos approximation .
@xcite this approach entails computing the self - correlation function of the 2d constant energy contour ( cec ) of the surface - localized spectral function @xmath50 in @xmath40-space at a given energy @xmath51 : @xmath52 thereby , the number of all possible @xmath36 within the same cec is counted and weighted by the spin dependent scattering matrix element @xmath53 . in order to have access to @xmath3 we develop a semi - theoretical approach , as illustrated in fig .
[ fig : figure3 ] .
first , the band structure is calculated from an lda slab calculation of the bi bl / bi@xmath0se@xmath1 system [ fig .
[ fig : figure3 ] ( a ) ] , where the slab has been adjusted to the geometry observed by stm and to the bi bl bands below @xmath44 as measured by arpes [ fig .
[ fig : figure3 ] ( c ) , see sec . [ sec : dft_spinpol ] ] .
the band structure is computed around @xmath45 along three different directions in a 0.7 @xmath54 range in the surface brillouin zone ( sbz ) ( @xmath45-@xmath55 , @xmath45-@xmath46 and @xmath45-@xmath56 , where @xmath56 is defined as the point along the perimeter of the sbz that sits at equal distance between @xmath55 and @xmath46 ) .
second , the relevant eigenvalues @xmath57 for each band have been isolated and symmetrized in * k*-space according to the pseudo sixfold symmetry of the system , which resulted in the description of the band structure along 12 different directions .
a 3d interpolation of the data was used to obtain a full space profile of the single electron bare energy @xmath57 .
note that this method does not require fitting the dispersion profile with any function but uses a voronoi natural neighbors interpolation of close scattered data points . a full 3d spectral function in the energy range between @xmath58 and @xmath59
is then produced according to @xmath60 ^ 2+\sigma''(e,\textbf{k})^2}*w(e,\textbf{k } ) \label{eq:3}\ ] ] where @xmath61 and @xmath62 are real and imaginary parts of the electron self - energy , here set to @xmath63 and @xmath64 , respectively .
@xmath65 is a weighting factor accounting for different localization of the electron wave function of each state @xmath66 in the bi bl . in this way it is possible to take care for
the surface sensitivity of the sts technique by neglecting all the bands which have low contribution in the bl accoording to dft , and will therefore not contribute to the qpi pattern .
@xmath67 is defined in all 3d grid points through the same symmetrization - interpolation procedure used for @xmath57 starting from values associated to the individual points from the dft calculation [ see size of points in fig . [
fig : figure3 ] ( a ) ] .
figure [ fig : figure3 ]
( b ) shows a 3d perspective of the resulting surface - localized spectral function .
third , we cut @xmath50 at different @xmath51 to obtain the cecs as shown in fig .
[ fig : figure3 ] ( c ) . at this point ,
a good agreement between the simulated cecs and the cecs as measured by arpes ( same figure ) demonstrates the soundness of the method .
finally , the jdos simulation is performed as described in equ .
[ eq:2 ] assuming a spin - dependent @xmath4 ( see sec .
[ sec : dft_spinpol ] ) .
the jdos can be directly compared to the ft - sts data [ fig .
[ fig : figure3 ] ( d ) ] . before
this comparison is analysed in detail , the results of the dft calculations are described in the following .
the slab used for the dft calculations has been adjusted to the geometry as measured by stm ( sec .
[ sec : stm ] ) .
the in - plane lattice constant has been set to that of bi@xmath0se@xmath1 consistent with the pseudomorphic growth . while the distance between the bi layers in the bl was kept constant at the bulk bi value of @xmath68 , the distance between the topmost se layer and the topmost bi layer in the bl ( surface layer )
was varied between @xmath27 and @xmath69 by increasing the gap between the bl and the substrate , in order to find the best match between the calculated band structure and the arpes data .
this is achieved for @xmath69 , consistent with the large distance between bl and substrate observed by stm ( sec .
[ sec : stm ] ) .
the resulting band structure shown in fig .
[ fig : figure3 ] ( a , c ) reveals four principal bands , which dominate the surface layers by their relatively strong surface localization : the green band can be identified with the bi bl valence band ( vb ) , which is shifted by about @xmath70 up in energy with respect to the free bi bl @xcite due to charge donation to the substrate .
a similar charge transfer effect was already proposed for the bi bl grown on bi@xmath0te@xmath1 and was attributed to the difference in work function between the film and the substrate @xcite . consequently , there is a strong electric field in the first layers of the substrate leading to a downwards band - bending induced quantum well hosting a two - dimensional electron system ( 2des ) of the bi@xmath0se@xmath1 cb in light blue and vb in dark blue , which reveal a similar shape as the respective bands observed for bi@xmath0se@xmath1 covered with dilute amounts of adsorbates @xcite .
moreover , the electric field causes a rashba - like spin - splitting as revealed by the two branches of each of the three bands .
the band crossing at @xmath45 of the two spin - split branches of the bi@xmath0se@xmath1 cb 2des was previously named substrate dirac point ( dp ) @xcite . however , its origin is of rashba - type spin - orbit splitting and not of topological nature .
in addition to these bands derived from the bi bl or from quantum well states of the bi@xmath0se@xmath1 substrate , there is a non - trivial 2d ts with a dirac - like dispersion ( red ) , whose dp appears about @xmath71 below @xmath44 .
this ts is strongly localized in the bi bl and partially in the first quintuple layer of the substrate ( see fig . [
fig : figure3 ] ( a ) and ref .
it _ replaces _ the original 2d ts of the substrate bi@xmath0se@xmath1 , now coexisting with the rashba - split bi@xmath0se@xmath1 cb 2des .
hybridization between the bi bl vb and the bi@xmath0se@xmath1 cb 2des causes a gap in these two bands at about @xmath72 above @xmath44 which is crossed by the ts .
note , that this hybridization gap is only along @xmath45@xmath46 , and not along @xmath45@xmath55 [ fig .
[ fig : figure3 ] ( b , c ) ] .
therefore , the necessary condition of an energy gap in the bi bl bands for the formation of the bl edge state is not fulfilled in this hybridization gap region , as opposed to the argumentation in ref .
in contrast , the edge state is expected to be positioned in the original band gap between vb and cb of the bl at about @xmath73 above @xmath44 . in order to take into account the spin texture of the rashba - split bands and of the ts , the spin - dependence of the matrix elements @xmath4 is considered to lowest approximation ,
i.e. neglecting possible out - of plane components of the spin and spin - orbit scattering involving changes in the orbital moments @xcite out - of plane spin components derive from third order rashba spin - orbit splitting . considering the very low anisotropic nature of the split these components
are thus expected to be small . representing the spin part of the electron wave functions by normalized spinors @xmath74 , the matrix elements are calculated as overlap probabilities @xmath75 where @xmath76 is the angle between * k * and * k * and @xmath77 takes into account the identity of the two band branches @xcite . thereby , the matrix element depends on the projection of the spin in @xmath37 onto the spin in @xmath38 .
the spin directions in a cec are exemplarily represented in fig .
[ fig : figure3 ] ( c ) .
figure [ fig : figure4 ] shows a comparison of the simulated cecs , the resulting jdos images and the measured ft - sts images . first focusing on ( a - e ) [ as well as fig . [
fig : figure3 ] ( c ) ] , the cecs show the following structures : below @xmath44 up to about @xmath78 , the most dominating features are a central concentric ring stemming from the largely isotropic ts , and two outer hexagonal contours stemming from the spin - split bi - bl , which has a hexagonal anisotropy reflecting crystal symmetry . in between , there are faint features stemming from the bi@xmath0se@xmath1 vb .
note , that these cecs are also confirmed by the arpes measurements [ fig .
[ fig : figure3 ] ( c ) ] . around @xmath44 , or ,
more specifically , going from @xmath78 to @xmath79 , the ts circle vanishes at the dp and opens again , while the bi - bl hexagons gradually merge with complex features stemming from the onsetting bi@xmath0se@xmath1 cb 2des . finally , in the unoccupied dos above @xmath80 ,
the two inner cecs stem from the spin - split bi - bl vb , which get more and more isotropic approaching the band maximum . the star - like shape with the arms along @xmath45@xmath46 originates from the bi@xmath0se@xmath1 cb which has a strong trigonal anisotropy . taking into account the spin - polarization within these cecs to first approximation ( sec .
[ sec : dft_spinpol ] ) , the jdos simulation of the qpi results in the images shown in fig .
[ fig : figure4 ] ( f - j ) .
the following principal features can be observed and ascribed to particular scattering transitions inside the cecs ( in fig .
[ fig : figure4 ] the corresponding scattering vectors have been mostly drawn along @xmath45@xmath46 , as an example , while the corresponding vectors along @xmath45@xmath55 are left out , in order to increase visibility ) .
below @xmath44 [ c.p . figs .
[ fig : figure4 ] ( a , b , c , f , g , h ) ] intraband scattering through ( @xmath81 ) and along the side ( @xmath82 ) of the almost degenerate spin - split hexagonal bi bl cecs produces the inner star ( green line ) and outer hexagonal rim ( green arrow ) in the jdos , respectively .
interband scattering between the ts and the bi bl ( @xmath83 and @xmath84 ) results in two additional inner hexagonal features ( orange and pink arrows ) in the jdos . below the onset of the bi@xmath0se@xmath1 vb
, there is an additional interband scattering channel ( @xmath85 ) from the bi@xmath0se@xmath1 vb into the bi - bl vb [ grey arrow in ( a ) ] , which is not clearly visible in the jdos ( f ) . above @xmath44 [ c.p .
[ fig : figure4 ] ( d , e , i , j , ) ] the intraband bi bl scattering between the two branches of opposite spin chirality ( @xmath81 ) and the intraband scattering @xmath86 along the flat side of the bi@xmath0se@xmath1 cb cec have almost equal scattering vectors and lead to the inner star shaped feature in the jdos ( green arrow and dark blue line ) .
the different intraband scattering vectors involving the six arms of the star - shaped bi@xmath0se@xmath1 cb cec ( @xmath87 , @xmath88 , @xmath89 ) lead to the outer star shape of the jdos with 12 arms .
the additional interband scattering vectors ( not shown ) lead to very complex features in the @xmath36-space in between these two intraband scattering features .
comparison of these simulated jdos images and the measured ft - sts images [ c.p .
[ fig : figure4 ] ( f - j ) to ( k - o ) ] leads to the conclusion that the overall trend in the experimentally observed qpi patterns can be understood as follows .
while below @xmath44 the dominant hexagonal outer shape in the ft - sts images results from intraband scattering @xmath81 in the bi bl , the dominant star - like shape above @xmath44 comes from one of the intraband scattering vectors , i.e. , @xmath87 .
moreover , by thorough comparison almost all of the possible interband and intraband scattering vectors found in the simulated jdos images can be identified in the measured ft - sts images [ see coloured arrows in ( f - j ) and ( k - o ) ] .
most importantly , the interband scattering between the ts and the bi - bl , @xmath83 and @xmath84 , are observed over the entire energy range below @xmath44 down to almost @xmath90 .
another important feature visible in the jdos and ft - sts images is the flower - like pattern marked in figs .
[ fig : figure4 ] ( j , o ) in yellow .
this pattern stems from scattering vectors @xmath91 between the star - shaped cec of the bi@xmath0se@xmath1 cb and the circular cec of the bi bl [ fig .
[ fig : figure4 ] ( e ) ] and proves the coexistance of these two bands in the unoccupied states region .
some of the scattering vectors identified in the jdos are missing in the ft - sts images .
the @xmath81 scattering channel is absent in the ft - sts images at very large negative bias voltage [ green cross in ( k ) ] .
also , from the three intraband bi@xmath0se@xmath1 cb scattering vectors dominating the jdos , only @xmath87 is visible in the ft - sts [ light blue arrow and crosses in ( o ) ] .
these discrepancies are probably explained by deficiencies in the matrix elements @xmath4 , as will be discussed in sec .
[ sec : discussion ] .
as will be shown in the following section , a comparison of the dispersions of the @xmath36 vectors , which are independent of scattering matrix element effects , reveals an astonishingly good agreement between measurement and simulation .
the length of the scattering vectors ( @xmath47 ) along @xmath45@xmath46 of all experimentally detected qpis has been extracted from the ft - sts images and is plotted in fig .
[ fig : figure5 ] ( a ) as a function of energy ( dots ) together with the dispersion extracted from the spectral function and jdos images ( lines ) . for comparison , the dft - calculated band structure which is the basis for the simulation ,
is plotted in the same figure together with the according scattering vectors ( b ) .
there is an excellent overall agreement between the measured and simulated dispersions over most of the energy range .
note , that @xmath92 is plotted only for reference in fig .
[ fig : figure5 ] ( a ) , but detected neither in the jdos nor in the ft - sts images , satisfying the avoided backscattering as a well known property of time - reversed band partners in general and a ts in particular . also , the bl intraband scattering appears exclusively between the two branches of equal spin helicity [ fig .
[ fig : figure5 ] ( b ) ] . only in a small energy window from @xmath44
down to roughly @xmath93 , there is a considerable deviation for @xmath94 and @xmath95 .
this deviation occurs in an energy range , where there are principally a number of additional possible bi@xmath0se@xmath1 cb intraband scattering vectors which might overlap with @xmath83 ( grey dotted lines ) . above @xmath44 , @xmath81 and @xmath86
are almost degenerate , such that the experimentally detected dispersion is probably a mixture of these two scattering channels .
taking into account the relative simplicity of the used jdos method , and the dft based approach for the spectral function , which typically comprises some error in the energetic positions of the bands around @xmath44 , the correspondence between both qpi interference patterns ( fig .
[ fig : figure4 ] ) and dispersion ( fig .
[ fig : figure5 ] ) in simulation and experiment is astonishingly good .
the average value of the cross correlation between the experimental ft - sts and the simulated jdos images defined by @xmath96}{\sqrt{\sum\limits_{i , j}(i_{\mathrm{exp}}(i , j)-<i_{\mathrm{exp}}>)^2}\cdot\sqrt{\sum\limits_{i , j}(i_{\mathrm{sim}}(i , j)-<i_{\mathrm{sim}}>)^2}}\ ] ] where @xmath7 are the intensities at point @xmath97 of the images and @xmath98 denote their mean intensity values , amounts to 80% , with a variation of @xmath99% across the whole investigated energy range .
figure [ fig : figure5 ] nicely illustrates , that the intraband scattering inside the ts is strongly suppressed by time reversal symmetry due to its peculiar spin texture over a large energy range .
nevertheless , _ interband _ scattering between the ts and the bi bl band is allowed .
the approximate linearity of @xmath94 , @xmath100 , and @xmath95 , taking into account the inversion symmetry of the band structure with respect to @xmath45 , experimentally proves the presence and linear dispersion of the ts .
this holds even in the energy range where the ts is almost degenerate with the substrate vb .
this extraordinary property of the particular ts in bi bl / bi@xmath0se@xmath1 is caused by its localization in the bi bl which suppresses hybridization with bulk vb states in contrast to other 3d tis .
however , the observed interband scattering from the ts into the bi bl vb excludes the possibility of using this ts for carrying dissipationless spin currents on the surface .
on the other hand , there are obvious discrepancies between the ft - sts and jdos images shown in fig .
[ fig : figure4 ] , indicating a too simplistic form of the assumed scattering matrix elements @xmath4 .
the absence of @xmath81 at low energy [ c.p .
green arrow and cross in fig .
[ fig : figure4 ] ( f , k ) ] and of the two largest bi@xmath0se@xmath1 cb intraband scattering vectors [ c.p . light blue dotted and dashed arrows and crosses in fig .
[ fig : figure4 ] ( d , e , i , j , n , o ) ] , in the ft - sts images might be explained by the sensitivity of sts to the vacuum density of states whereas , the jdos approach considers the density of states _ in _ the surface . the vacuum density of states is dominated by states with small @xmath40 , damping qpi patterns involving large @xmath36 . by switching off the @xmath81 scattering channel in the jdos simulations a very good agreement with the ft - sts image is found .
another explanation for the missing bi@xmath0se@xmath1 cb intraband scattering vectors is spin selection , which may be oversimplified in the jdos approach , in particular for the strongly anisotropic cb where the constraint to in - plane components of the spin breaks down .
the inclusion of a more rigorous treatment of the electron spin in the matrix elements is beyond the scope of the present paper .
the experimentally determined band structure substantiates that the original band gap of the bi bl is moved up to @xmath101 above @xmath44 .
this doping effect comes about by the charge transfer from the bl to the substrate , which is also responsible for the rashba spin - split substrate cb 2des states visible in the band structure of fig .
[ fig : figure5 ] ( b ) .
the strong rashba splitting has been recently observed for bi bl grown on a single quintuple layer of bi@xmath0se@xmath1(or bi@xmath0te@xmath1 ) @xcite , although , there , it does not coexist with a ts .
it remains to be experimentally studied , whether the energetic position of the topological edge state of the bl is in the bl band gap , as predicted theoretically @xcite , and thus not accessible to transport experiments .
in this case , it might be possible to shift the edge state to @xmath44 by surface doping using decoration of the bl with alkali adatoms @xcite .
finally , our paper illustrates a methodology for the analysis of ft - sts images by dft - based spin - dependent jdos simulations enabling to pin point the origin of very complex qpi interference patterns which stem from the electrons in several energetically degenerate bands .
we thank tien - ming chuang for his help concerning the fast fourier transformation procedures , and lucas barreto and wendell silva for the help with the arpes data .
computing time on the supercomputer juropa at the jlich supercomputing centre ( jsc ) is gratefully acknowledged .
we acknowledge financial support from the dfg via spp 1666 `` topological insulators : materials - fundamental properties - devices '' .
ae , jw , and rw acknowledge financial support from the dfg via sfb 668 .
aak acknowledges project no .
kh324/1 - 1 from the emmy - noether - program of the dfg .
mm , xgz , and phh acknowledge financial support from the villum foundation and from the aarhus university research foundation .
jlm and bbi acknowledge the danish national research foundation ( dnrf93 ) .
45ifxundefined [ 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/physrevlett.97.236805 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1073/pnas.1218104110 [ * * , ( ) ] http://dx.doi.org/10.1038/ncomms2387 [ * * , ( ) ] link:\doibase 10.1021/nl402450b [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.016801 [ * * , ( ) ] link:\doibase 10.1103/physrevb.89.155436 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) http://stacks.iop.org/0022-3727/44/i=46/a=464010 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevlett.96.067005 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.108.127001 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.116805 [ * * , ( ) ] http://dx.doi.org/10.1038/nature08308 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.88.125141 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) \doibase http://dx.doi.org/10.1063/1.2972971 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.88.041404 [ * * , ( ) ] @noop @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.108.206402 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1126/science.1181083 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.4747715 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.89.085401 [ * * , ( ) ] http://stacks.iop.org/1367-2630/16/i=4/a=045017 [ * * , ( ) ] @noop ( ) , @noop * * , ( ) | the band structure , intra- and interband scattering processes of the electrons at the surface of a bismuth - bilayer on bi@xmath0se@xmath1 have been experimentally investigated by low - temperature fourier - transform scanning tunneling spectroscopy .
the observed complex quasiparticle interference patterns are compared to a simulation based on the spin - dependent joint density of states approach using the surface - localized spectral function calculated from first principles as the only input .
thereby , the origin of the quasiparticle interferences can be traced back to intraband scattering in the bismuth bilayer valence band and bi@xmath0se@xmath1 conduction band , and to interband scattering between the two - dimensional topological state and the bismuth - bilayer valence band .
the investigation reveals that the bilayer band gap , which is predicted to host one - dimensional topological states at the edges of the bilayer , is pushed several hundred milli - electronvolts above the fermi level .
this result is rationalized by an electron transfer from the bilayer to bi@xmath0se@xmath1 which also leads to a two - dimensional electron state in the bi@xmath0se@xmath1 conduction band with a strong rashba spin - splitting , coexisting with the topological state and bilayer valence band . |
despite of the fact that x - ray observations of star - forming regions have been a powerful tool for discovering young stellar objects ( yso ) and t tauri stars ( tts ) since _ einstein _ ( references in feigelson & montmerle 1999 ) , long wavelength ( ir and submillimeter ) emission is used to define evolutionary classes for young stars .
classes i - iii are solely based on the excess seen in the ir spectral energy distribution ( sed ) with respect to stellar blackbody photospheric emission , as measured by the spectral index @xmath4 = d log ( @xmath5 f@xmath6 ) / d log @xmath5 between @xmath5 = 2.2 and 10 - 25@xmath7 m : class i , ii and iii correspond to @xmath4@xmath8 0 , -2 @xmath9@xmath4 @xmath9 0 , and @xmath4 @xmath9 -2 ( lada 1987 ) .
class 0 stars represent the earliest phase of star formation and are identified by the ratio of submillimeter to bolometric luminosity ( andr et al .
1993 ) and are seen as condensations in submillimeter far - infrared dust continuum maps ; these condensations often show collimated co outflows or internal heating sources .
x - ray bright t tauri stars usually belong to either class ii or iii .
high resolution images obtained by the _ chandra _
x - ray observatory open a new era in the study of star formation because the 1@xmath10 resolution of _ chandra _ is necessary to resolve individual sources in nearby star forming regions , and because confusion due to foreground and background stars is much less important at x - ray energies than in the optical and near - infrared regime .
recent _ chandra _ observations of orion @xcite detected hundreds of x - ray sources , e.g. pre - main sequence stars ( pms ) with masses in the range 0.05 m@xmath11 m@xmath12 and a combined infrared and x - ray study suggests that the x - ray luminosity of pms depends on stellar mass , rotational history , and magnetic field @xcite . a high percentage of class i pms were also found to be x - ray emitters ; in @xmath13 oph , 70% of identified class i stars are x - ray bright .
strong x - ray flares from pms in @xmath13 oph @xcite , monoceros r2 @xcite , and orion @xcite were detected from class i , ii and iii objects , possibly because of magnetic activity .
moreover , x - ray emission from class 0 candidates was detected in omc-3 @xcite .
the trifid nebula ( m20 ) is one of the best - known astrophysical objects .
it is a classical nebula ionized by an o7.5 star , hd164492a , and the ionized nebula glows in red light .
the nebula is trisected by obscuring dust lanes giving the trifid its name .
a blue reflection nebula appears to the north of the red nebula . at an age of @xmath14 years
, the trifid is a young region .
recent studies using the infrared space observatory ( iso ) and the hubble space telescope ( hst ) ( cernicharo et al .
1998 ; lefloch & cernicharo 2000 ; hester et al . 1999 ) show the trifid to be a dynamic , pre - orion " star forming region containing young stars undergoing episodes of violent mass ejection , with protostars like hh399 @xcite losing mass and energy to the nebula in optically bright jets .
four massive ( 1760 m@xmath15 ) protostellar cores were discovered in the trifid from millimeter - wave observations .
these cores are associated with molecular gas condensations at the edges of clouds ( lefloch & cernicharo 2000 ) .
t tauri stars and young stellar object candidates were also identified using near - infrared color - color diagrams from 2mass data ( rho et al .
2001 ) . unlike better - studied nearby star - forming regions such as orion and @xmath13 oph , star forming activities in the trifid
have only recently been recognized @xcite , and as a result of this ( and also due to contamination by foreground and background stars in the optical and ir ) , the population of pms in the trifid has not been fully investigated .
the distance to the trifid nebula is between 1.68 and 2.84 kpc ( lynds et al .
1985 ; kohoutek et al . 1999 and references therein ) and a distance of 1.68 kpc
is adopted in this paper . in this paper
, we report results of _ chandra _ observations of the trifid nebula . _
chandra _ resolved the hd 164492 multiple system into a number of discrete x - ray sources and we present their x - ray properties which include components a ( an o star ) , c ( a b6v star ) and b ( a2ia ) . our _ chandra _ observations revealed 304 sources and we found that 30% of the sources have hard emission similar to that from pms . among these candidate pms , we report properties of about a dozen flare sources , which include unusual variability from the o star and unusual emission from an a - type supergiant .
we also discuss the x - ray properties of the hd 164492 complex , the properties of x - ray sources which are apparently associated with two protostellar cores , and also the properties of some apparently strongly variable objects .
the trifid nebula was observed with the advanced ccd imaging spectrometer ( acis ) detector on board the _ chandra x - ray observatory _
@xcite on 2002 june 13 .
the results presented here arise from the imaging array ( acis - i ) , which consists of four 1024@xmath161024 front - side illuminated ccds .
the array was centered at r.a .
@xmath17 and dec.@xmath1856@xmath19 ( j2000 ) and covered an area in the sky of about @xmath20 .
the total exposure time of the acis observations was 58 ksec .
this observation is sensitive to x - ray luminosities of 5@xmath16 10@xmath21 erg s@xmath3 , assuming an appropriate pms x - ray spectrum ( temperature of 1 kev and an absorption 1.6@xmath16 10@xmath22 cm@xmath3 ) at the distance ( 1.68 kpc ) of the trifid .
we started data analysis with the level 1 processed event list provided by the pipeline processing at the _ chandra _ x - ray center .
the energy and grade of each data event were corrected for charge transfer inefficiency ( cti ) , applying the algorithm described by @xcite .
the event file was filtered to include event grades of 0 , 2 , 3 , 4 , and 6 , and filtered by time intervals to exclude background flaring intervals or other bad times .
the filtering process was done using the _ chandra _ interactive analysis of observations ( ciao ) package provided by the _ chandra _ x - ray center .
figure [ chandrazoom ] shows the _ chandra _ acis - i three color image of the trifid .
hundreds of point sources are detected with little diffuse emission .
x - ray sources were located using the _ wavdetect _ tool within the ciao package . this tool performs a mexican hat wavelet decomposition and reconstruction of the image after accounting for the spatially varying point spread function as described by @xcite .
we used wavelet scales ranging from 1 to 16 pixels in steps of @xmath24 , and a default source threshold probability of @xmath25 .
the _ wavdetect _ tool was run using an exposure map and it produced a catalog of 353 sources from the entire acis - i fov .
then we identified false sources produced by cosmic rays or cases in which the source counts are below the background counts .
this observation finally resulted in 304 x - ray sources detected from the total acis - i fov ( @xmath20 ) .
the full source list is given in table [ catalog ] in order of r.a .
thirty percent of these sources are shown to be hard ( shown in blue in figure [ chandrazoom ] ) ; the hard sources have a spectral hardness ratio ( shr ) @xmath26 , where shr is the ratio of the net counts in the hard @xmath27 kev band to those in the soft @xmath28 kev band .
diffuse emission was not obvious in the _ chandra _ images of the trifid nebula with our current data processing .
none - detection of diffuse emission in m20 is consistent with a claim that high mass star - forming regions without stars earlier than o6 may be unlikely to exhibit diffuse soft x - rays ( townsley et al .
2003 ; abbott 1982 ) the brightest source in the field corresponds to the o star hd164492a ( source 102 ) and has 884 counts .
this is equivalent to 0.047 photons per frame , which is small enough so as not to be affected by pileup effects ( the pileup fraction is @xmath29 ) . in order to verify the hard emission and fe k line detected in the _ asca _ spectra of hd164492a ( rho et al .
2001 ) , we extracted a spectrum using the same extraction region as was used in the _ asca _ analysis , and confirmed that the _ chandra _ and _ asca _ spectra are the same within the errors .
however , we found that hd164492a is actually a soft source ( see section 6 for details ) while the hard emission and fe k line ( which were attributed to the o star in the _ asca _ analysis ) is actually produced by the hard sources resolved by _
chandra _ as shown figure [ ostarcomplex ] .
in addition , the emission previously attributed entirely to hd164492a in an analysis of _ rosat _ pspc data ( rho et al . 2001 ) is now resolved into a dozen x - ray sources as shown in figure [ chandrazoom ] and [ ostarcomplex ] . to confirm the x - ray astrometry
, we used the 2mass final release point source catalog to search for near - ir counterparts to the x - ray sources in the full source list .
we cross - correlated the positions of the 2mass sources with the positions of our x - ray sources . within @xmath30 of
the _ chandra _ aimpoint , we found 72 2mass sources that coincide with x - ray source positions .
we computed the total r. a. and dec offset of these sources and obtained offsets of @xmath31 in r. a. , and @xmath32 in dec , which are less than one third of a pixel .
therefore , the astrometry of trifid acis image was confirmed with the minor systematic errors in the _ chandra _ aspect solution .
the hd164492 complex is a multiple stellar system composed of 7 physically related components , a - g @xcite .
component hd164492a is the o7.5 iii((f ) ) star responsible for ionizing the nebula .
this complex was observed as an unresolved bright x - ray source in pspc images ( rho et .
al . 2001 ) .
our _ chandra _ images ( figures [ chandrazoom ] and [ ostarcomplex ] ) resolve this complex into a number of discrete x - ray sources .
x - ray emission is clearly detected from components a , b , and c as shown in table 2 ( fig . [ ostarcomplex ] ) .
hd164492a is the brightest x - ray source in this complex .
component hd164492c ( a b6v star , gahm et al .
1983 ) is a bright x - ray source ( source 94 ) and is barely resolved from a nearby x - ray source to the nw of c ( source 91 ) which has no obvious optical counterpart ( fig . [ ostarcomplex ] ) .
source d ( a be star ) is a strong h-@xmath33 emitter ( herbig 1957 ) and has a disk ( resolved by a radio image ) , similar to proplyds in the orion nebula ( odell 2001 ) .
the component d is much fainter in x - rays than component c. the x - ray light curve of the hd164492a shows small but significant variability ( @xmath34 in the 10 hour observation , figure [ lcpms ] ) .
this is unsual since most ob stars are not known x - ray variables .
x - ray variability is not widely known from early type stars based on previous _ einstein _ and _ rosat _ surveys @xcite .
however , recent _ chandra _ observations show significant variability from some early type stars @xcite .
for example ,
@xmath35 oria , an o9.5 star , shows a 50% drop in 10 hr with multiple 10 - 20% flares @xcite .
some level of time variability is also observed from other early type stars in orion @xcite .
the spectrum of the o star shows thermal emission as shown in fig .
[ ostarspec]a .
we fit the x - ray spectra of hd164492a using two collisional ionization equilibrium ( cie ) thermal models ( a mewe - kaastra plasma model ( kaastra 1992 ) and an updated raymond - smith model ( raymond & smith 1977 ) ; mekal and apec in xspec ) . both models gave similar results .
we assumed sub - solar metal abundance ( 0.3 solar ) ; assuming solar abundances produced less than a 20% change in the derived spectral parameters .
the best fit of the spectrum of o star yields @xmath36 @xmath37 and a temperature of @xmath38 kev , which implies an x - ray luminosity of 2.8@xmath39 erg s@xmath3 .
the derived temperature is similar to the x - ray temperatures of other single massive stars ( corcoran et al .
1994 ; moffat et al .
although some of the luminosity may be due to an unresolved low - mass companion such as a t - tauri star which could be responsible for the variability , the low temperature obtained from the spectral fit is largely consistent with a single massive star . assuming a bolometric luminosity of l@xmath40 @xmath410.51.6@xmath42 ergs s@xmath3 , the ratio of x - ray to bolometric luminosity is @xmath43 , which is smaller than the canonical value @xmath44 but it is within the scatter @xcite .
the lack of high temperature x - ray emission in hd164492a suggests that a magnetically confined wind shock has not developed in the o star of hd164492a as has been suggested in other variable o stars showing high temperature x - ray emission ( like @xmath45 ori a , c , and e @xcite ) .
similarly the lack of hard emission in hd164492a also suggests that colliding wind emission ( as considered in @xcite ) is not important .
x - rays from hd164492a are likely produced by shocks distributed through the star s massive stellar wind .
the variability may be from instabilities within a radiatively driven wind ( lucy & white 1980 ; feldmeier et al .
1997 ) , but if so the rapid timescale of the changes ( @xmath46 hours ) suggests that the x - ray emission is dominated by a small number of strong shocks rather than a large distribution of weak shocks .
we also modelled the extracted spectrum from source c , which is blended with the unidentified x - ray source to the nw ( source 91 : component c2 ) which is barely resolved from source c. the component d , a be star , is not identified as a source by wavelet analysis ; however , there is some faint x - ray emission coinciding with the be star as shown in figure 2 .
component c2 ( source 91 ) shows variability , and the light curve is shown in fig .
[ lcpms ] . the best two - temperature absorbed thermal fit in fig .
[ ostarspec]b has @xmath47 @xmath37 and a temperature of @xmath48 kev , and @xmath49 @xmath37 and a temperature of @xmath50 kev .
the luminosity of soft and hard components are 1.4@xmath51 and 6@xmath52 erg s@xmath3 , respectively .
the lower absorption @xmath53 @xmath37 is comparable to the optical extinction @xmath54 mag @xcite , suggesting that the wind(s ) and the strong ionizing radiation from the o and b stars strip the surrounding materials .
the 6 kev component is partially responsible for the spectra seen by _ asca _ ( rho et al .
the x - ray emission from b and be stars is not well characterized at present since they are often faint .
the @xmath55 ratio is below 10@xmath56@xmath41 10@xmath57 for stars of spectral type b1.5(iii - v ) and later @xcite , and the late type stars have a luminosity range of 10@xmath21 - 10@xmath58 erg s@xmath3 .
thus the observed x - ray emission of 2@xmath59erg s@xmath3 from the source c blend is difficult to reconcile with emission from a typical b6v type star or a late type main - sequence companion star .
we extracted separate spectra from both components c and c2 , but the spectra could not be distinguished from each other and additionally we lose the hard component emission due to lack of photon statistics in the hard energy band , although separate light curves show stronger time variability in the component c2 .
the b6v star is an x - ray source as shown in figure [ ostarcomplex ] , and the component c2 is a companion candidate which is also strong x - ray emitter .
the companions of later b - type stars with strong x - ray emission , are suggested to be often pms ( stelzer et al . 2003 ) , which show characteristics of high x - ray luminosities and hard x - rays .
the presence of an unidentified x - ray source near the b6v star and the presence of hard x - ray emission in this system suggest that c2 may be a pms .
other possibilities such as emission from corona of the herbig be star , or a mechanism related to the proplyds remain open at this point .
component b ( source 106 ) is classified as a2ia ( gahm et al .
1983 ) or possibly as an a2 iii ( lindroos 1985 ) .
a - type stars are not often detected as x - ray sources @xcite since they lack subsurface convection zones to power strong magnetic fields , and they lack a strong uv flux to power massive stellar winds .
this may indicate that this star has a low - mass , x - ray bright companion or the spectral type may be incorrect .
its x - ray luminosity is listed in table 2 ; we assumed a temperature of 0.15 kev from other a stars ( simon & drake , 1993 ) and the absorption of 1.6@xmath1610@xmath22 @xmath37 .
we created lightcurves for each of the detected x - ray sources , using a bin length of 2500 sec , which gives a total number of 24 bins .
we calculated the ratio between max count in the highest bin and minimum count in the lowest bin and found @xmath4140 sources have the ratio greater than 1 after accounting for the errors .
if a source was only detected during a flare , we used the mean count rate instead of minimum count rate . in table 2
we list sources in which this ratio greater than 3 ( sources 8 , 23 , 97 , 166 , 170 , 194 , 211 , 237 , 246 , 256 , 283 and 285 ) .
these sources should represent the most strongly variable sources . to confirm this variability we also performed a simple @xmath60 test against an assumed the constant light curve .
the probability of constancy for each source from this test is given in table 2 , for which we used additional binning as needed for each source .
nine sources ( sources 23 , 91 , 97 , 166 , 194 , 237 , 256 , 283 and 285 ) in table 2 are definitely variable ( null probability @xmath9 10% ) based on the @xmath60 test and our ratio test .
we show sample light curves of the variable sources in fig .
[ lcpms ] .
we searched for near - infrared and optical counterparts for these variable sources .
unfortunately no previous optical identification of t tauri stars exists in the trifid
. the only identified t tauri stars or young stellar objects are 85 pms candidates identified from infrared excesses using 2mass second release data @xcite . here
we re - evaluate the identification of pms candidates tts or yso from near - infrared color - color diagrams using the 2mass final release data .
the final release data have improved photometry by allowing multiple - components of point spread function for the blended sources with the improved zero point , and a complete source catalog by thoroughly identifying artifacts . among variable sources in table 2 ,
only source 283 was previously identified as a yso @xcite and now sources 23 , 97 , 166 , 194 and 237 are identified as tts or yso based on the 2mass data .
we also searched for optical counterparts to these x - ray sources using the guide star catalog ( gsc 2.2 ) and usno - b catalog , but since these catalogs do not include any sources near bright optical diffuse emission , we also directly compared the x - ray sources in table 1 to digital all sky images ( dposs ) . among the variable sources listed in table 2 , five ( source 97 , 166 ,
237 , 256 and 285 ) have shr @xmath26 ( and they appear blue in figure [ chandrazoom ] ) .
sources 8 , 23 , 194 , 211 , 246 and 283 have both optical and 2mass counterparts .
these sources are not embedded and thus are likely in later stage of pms evolution , possibly class ii or iii .
sources 8 , 211 and 246 have 2mass counterparts , but not identified as tts or yso from near - infrared colors .
they are still likely pms , because the near - ir excess is shown to be time - dependent ( carpenter et al . 2000 ) .
sources 97 , 166 , 237 , 256 and 285 have no optical counterparts and have shr @xmath61 and the light curve of source 166 is shown in fig .
[ lcpms ] . in particular , source 256 and 285 have neither near - ir nor optical counterparts but exhibit flares in their x - ray light curves as shown in fig .
[ lcpms ] .
no radio counterparts are known for these sources .
we suggest these x - ray flare stars are in an early pre - main sequence stage ( class i or earlier ) .
source 211 has sufficient counts for spectral fitting , and the best - fit using a one - temperature thermal model yields @xmath62 @xmath37 and a temperature of @xmath63 ( @xmath81 ) kev .
the inferred x - ray luminosity of 1.1@xmath39 erg s@xmath3 , is comparable to those from tts and yso @xcite , and brighter than those of typical low - mass main - sequence stars ( @xmath4110@xmath64 erg s@xmath3 ) .
the sources with flares seem to be found preferentially concentrated along the dust lane and at the edge of the hii region , i.e. along the ionization fronts .
this reinforces the conclusion that most of flaring sources are pms .
we estimated the x - ray luminosities of sources 8 , 23 , 97 , 166 , 170 , 194 , 246 , 256 and 283 assuming the spectral parameters are the same as those of source 211 . for sources 237 and 285
, we assumed the spectral parameters are the same as those of tc1 source , because their hardness ratios are comparable to that of tc1 .
we compared the x - ray sources with the known massive star forming cores detected in the 1300@xmath7 m dust continuum map @xcite .
there are five such cores , four of which are known to have bipolar wings .
we detected x - ray emission from tc1 ( source 117 ) and possibly tc4 ( source 38 ) , two cores with bipolar wings and associated class 0 candidates @xcite .
the sources associated with both tc1 and tc4 are hard x - ray sources ( which appear as blue sources in figure [ chandrazoom ] ) .
figure [ tc1tc4 ] shows the positions of the x - ray sources superposed on the 1300@xmath7 m dust continuum map .
the x - ray source associated with tc1 has no 2mass counterpart .
but the x - ray identification of the source associated with the tc4 core is less clear , because this source is located at the edge of the acis field of view where the width of psf is @xmath65 .
the tc4 source coincides with the 1300@xmath7 m dust continuum source within @xmath66 , and also has two 2mass counterparts within @xmath67 .
the closest 2mass source is 2mass180212.5 - 230549 for which j band emission is not detected and h(=14.8mag ) and k@xmath68 ( = 12.9mag ) magnitudes are contaminated by a nearby star .
it is not clear if this 2mass source is related to the tc4 and the x - ray source .
tc4 satisfies the class 0 definition of @xcite in terms of its dust temperature ( @xmath69 k ) , the ratio of its submillimeter to bolometric luminosity , and the presence of a bipolar outflow .
an outflow from tc4 shows a kinematic age of 6.8@xmath1610@xmath70 yr and suggests it to be an intermediate- or high - mass object @xcite . for tc1 ,
the sed is not available so it s not clear if tc1 satisfies the class 0 criteria or not ; however , the object is deeply embedded in a dust core so it is at least a class i object , and possibly younger than a class i source .
tc1 and tc4 have hardness ratios of @xmath71 and @xmath72 , respectively ( see table 1 and 2 ) .
the source associated with tc4 is one of the hardest sources in table 1 .
these sources are either highly absorbed or have high temperatures .
our analysis of the x - ray spectra of tc1 and tc4 shows that the best fit thermal model parameters are n@xmath73=6@xmath74 @xmath37 , @xmath75 kev and @xmath76 erg s@xmath3 for the tc 4 source , and @xmath77 @xmath37 , kt = 1.1 kev , and @xmath78 erg s@xmath3 for the tc1 source .
the large column densities we derive show that these sources are highly embedded .
the only detection of x - ray emission from a class 0 object until now is a source located in omc-3 ( cso6 ) @xcite . in comparison ,
the class 0 source in omc-3 ( cso6 ) has @xmath79 @xmath37 and a luminosity of @xmath80 erg s@xmath3 .
x - ray emission from class 0 objects which are in the dynamical infall phase is poorly understood since so far only a few class 0 candidates have been identified and their properties are not well constrained . it is unclear if the x - ray emission is due to solar - type magnetic activities as in class i objects .
the fact that these two counterparts have high hardness ratios does not support the x - ray emission is from a low - mass companion to the protostellar core .
if these objects are accreting , then the x - ray emission may be related to the accumulation and release of angular momentum toward the growing central star by outflow processes .
young hii regions like the trifid are rich sources of x - ray emitters .
our _ chandra _ images reveal a few hundred x - ray sources including variable and hard sources , along with pre - main sequence stars and more evolved ob stars .
we summarize our findings here .
the multiple star system hd164492 is resolved for the first time in x - rays into individual components .
x - ray emission is detected from components a , b , c ( a b6v star ) , which is blended with an unidentified source in the _ chandra _ images .
this blend has comparable x - ray brightness to the o star .
the o star hd 164492a shows small but significant variability and has a soft spectrum with a temperature of 0.5 kev .
the temperature is comparable to those of other single massive stars and the ratio of x - ray and bolometric luminosities is smaller than the canonical value @xmath44 but it is within the scatter of distribution .
the lack of any hard component suggests that neither a magnetically confined wind shock nor colliding wind shock is needed to describe the x - ray emission from the o star .
the variability of the x - ray emission implies that the emission is produced by a small number of strong shocks in the wind of hd 164492a .
the x - ray spectrum from the component c blend requires a two - temperature thermal model with @xmath81 kev and @xmath82 kev .
the inferred x - ray luminosity is 2@xmath83 erg s@xmath3 .
this blend is highly variable in x - rays which suggests that one of the stars dominates the emission .
we found a dozen stars which show evidence of flaring activity and there could be many as @xmath41 40 variable stars in the full source list .
nine sources ( sources 8 , 23 , 97 , 166 , 194 , 237 , 256 , 283 and 285 in table 2 ) have significant variability based on the @xmath60 statistics .
we searched for their near - infrared and optical counterparts , and found that six stars have both optical and 2mass counterparts .
these sources are likely in later stages of pms evolution .
four sources which have no optical counterparts and have shr greather than @xmath84 are likely in early stages of pms , possibly class i or earlier .
there are a few stars with neither near - ir nor optical counterpart whose light curves show strong flares , suggesting that they are very early stage pre - main sequence stars .
we detected x - ray emission from tc1 and possibly tc4 , two massive star forming cores with bipolar wings and associated class 0 candidates .
both tc1 and tc4 show extremely hard x - rays and their spectra imply very high absorption ( n@xmath73=3.5 - 6 @xmath74 @xmath37 ) and high luminosity ( @xmath2 erg s@xmath3 ) . only one other detection of x - ray emission from a class 0 object has been previously reported .
thus our two detections imply the second and third x - ray detection of a class 0 object .
it is unclear if the x - ray emission of these objects is due to solar - type magnetic activities as in class i objects .
if the x - ray emission is from the accreting stage , the x - ray emission may be related to the competing processes of accumulation of angular momentum toward the growing central star and release of angular momentum by outflow processes .
thirty percent of the x - ray sources in the full field are shown to be hard ( shr @xmath26 ) sources ( shown in blue in figure [ chandrazoom ] ) , and 16 percent of sources are extremely hard sources ( shr @xmath85 ) .
a high proportion of these sources are probably pms because they are either highly embedded or extemely high temperature .
these hard sources along with the sources near hd 164492c are responsible for the hard spectra seen by _
asca_. the accurate positions from the complete list of the x - ray sources are provided in table 1 .
these sources provide an opportunity to identify interesting pms using the multi - wavelength follow - up observations , which will help us to further understand the population and evolution of protostellar objects . partial support for this work
was provided by nasa through _ chandra _ grant g02 - 3095a .
j. r. and s. v. r. acknowledge the support of california institute of technology , the jet propulsion laboratory , which is operated under contract with nasa .
andre , p. , ward - thompson , d. , & barsony , m. , 1993 , apj , 406 , 122 - 141 berghfer , t. w. , schmitt , j. h. m. m. , & cassinelli , j. p. , 1996,a&as , 118 , 481 carpenter , j. m. , hillenbrand , l.a .
& skrutskie , m. f. , 2001 , aj , 121 , 3160 cassinelli , j. p. & cohen , d. h. , 1994 , iaus , _ pulsation , rotation and mass loss in early - type stars _
( kluwer academic publishers ) , 162 , 189 cernicharo et al .
1998 , science , 282 , 462 ( clc98 ) chlebowski , t. , harnden , f. r. , & sciortino , s. 1989 , , 341 , 427 rccrrr [ catalog ] 1 & 177 & 18:01:50.6122:54:39.6 & 12 @xmath86 4 & 3 & -0.23 @xmath86 0.20 + 2 & 208 & 18:01:50.7922:52:18.5 & 22 @xmath86 7 & 3 & 0.09 @xmath86 0.13 + 3 & 206 & 18:01:52.5422:57:33.6 & 22 @xmath86 6 & 4 & -0.26 @xmath86 0.16 + 4 & 207 & 18:01:53.5722:51:35.6 & 24 @xmath86 7 & 4 & 0.56 @xmath86 0.16 + 5 & 333 & 18:01:57.1823:01:22.1 & 58 @xmath86 9 & 11 & -0.49 @xmath86 0.14 + 6 & 344 & 18:01:57.6423:00:01.7 & 13 @xmath86 4 & 3 & -0.19 @xmath86 0.20 + 7 & 197 & 18:01:58.3322:52:26.5 & 21 @xmath86 6 & 5 & -0.36 @xmath86 0.14 + 8 & 196 & 18:01:59.0722:52:27.5 & 73 @xmath86 10 & 12 & -0.49 @xmath86 0.11 + 9 & 352 & 18:01:59.1223:00:55.3 & 20 @xmath86 5 & 5 & 0.76 @xmath86 0.22 + 10 & 195 & 18:01:59.8522:53:14.8 & 41 @xmath86 7 & 9 & -0.62 @xmath86 0.16 + 11 & 203 & 18:02:02.0422:54:19.1 & 14 @xmath86 4 & 4 & -0.36 @xmath86 0.23 + 12 & 348 & 18:02:02.0723:00:56.2 & 13 @xmath86 4 & 4 & 0.69 @xmath86 0.10 + 13 & 332 & 18:02:02.5423:02:32.5 & 103 @xmath86 11 & 20 & 0.81 @xmath86 0.12 + 14 & 343 & 18:02:03.1322:59:09.6 & 13 @xmath86 4 & 4 & -0.33 @xmath86 0.19 + 15 & 201 & 18:02:03.4522:56:07.2 & 13 @xmath86 4 & 4 & -0.69 @xmath86 0.34 + 16 & 194 & 18:02:03.8622:52:58.0 & 45 @xmath86 7 & 10 & 0.84 @xmath86 0.18 + 17 & 313 & 18:02:04.7523:04:03.0 & 11 @xmath86 4 & 3 & 0.29 @xmath86 0.19 + 18 & 189 & 18:02:05.5022:55:01.2 & 19 @xmath86 5 & 6 & -0.65 @xmath86 0.25 + 19 & 341 & 18:02:06.0523:00:49.9 & 14 @xmath86 4 & 4 & -0.60 @xmath86 0.30 + 20 & 175 & 18:02:07.0222:57:25.8 & 13 @xmath86 4 & 6 & -0.53 @xmath86 0.27 + 21 & 174 & 18:02:07.6222:57:42.7 & 102 @xmath86 10 & 37 & -0.51 @xmath86 0.11 + 22 & 188 & 18:02:07.7522:57:23.4 & 22 @xmath86 5 & 8 & 0.92 @xmath86 0.28 + 23 & 173 & 18:02:07.7722:55:33.9 & 58 @xmath86 8 & 19 & -0.58 @xmath86 0.15 + 24 & 172 & 18:02:08.0922:57:18.6 & 39 @xmath86 6 & 14 & -0.85 @xmath86 0.21 + 25 & 289 & 18:02:08.5222:59:00.0 & 62 @xmath86 8 & 24 & -0.76 @xmath86 0.15 + 26 & 354 & 18:02:09.1823:02:36.9 & 5 @xmath86 2 & 2 & 0.00 @xmath86 0.22 + 27 & 205 & 18:02:09.4222:50:56.1 & 10 @xmath86 4 & 3 & -0.12 @xmath86 0.20 + 28 & 339 & 18:02:09.9223:04:27.5 & 43 @xmath86 8 & 8 & -0.47 @xmath86 0.15 + 29 & 186 & 18:02:10.4622:55:13.8 & 28 @xmath86 5 & 11 & -0.66 @xmath86 0.22 + 30 & 185 & 18:02:10.8022:55:32.6 & 14 @xmath86 4 & 6 & -0.25 @xmath86 0.26 + 31 & 353 & 18:02:11.2723:04:11.9 & 18 @xmath86 5 & 4 & 0.14 @xmath86 0.17 + 32 & 160 & 18:02:11.3622:56:50.8 & 74 @xmath86 9 & 29 & -0.53 @xmath86 0.13 + 33 & 246 & 18:02:11.6322:59:33.9 & 189 @xmath86 14 & 70 & -0.54 @xmath86 0.08 + 34 & 204 & 18:02:11.6622:50:10.8 & 21 @xmath86 6 & 4 & 0.31 @xmath86 0.19 + 35 & 170 & 18:02:11.7822:56:45.8 & 34 @xmath86 6 & 14 & 0.24 @xmath86 0.18 + 36 & 159 & 18:02:11.9522:55:43.6 & 294 @xmath86 17 & 92 & -0.70 @xmath86 0.07 + 37 & 351 & 18:02:12.2623:03:10.4 & 14 @xmath86 4 & 3 & 0.33 @xmath86 0.22 + 38 & 347 & 18:02:12.4723:05:48.1 & 62 @xmath86 9 & 9 & 0.67 @xmath86 0.13 + 39 & 311 & 18:02:12.6822:58:52.2 & 6 @xmath86 2 & 2 & 1.00 @xmath86 0.45 + 40 & 199 & 18:02:12.7922:57:00.0 & 6 @xmath86 3 & 3 & 0.80 @xmath86 0.40 + 41 & 288 & 18:02:13.1523:01:59.3 & 92 @xmath86 10 & 27 & -0.66 @xmath86 0.12 + 42 & 338 & 18:02:13.3823:04:13.5 & 9 @xmath86 4 & 3 & 0.38 @xmath86 0.27 + 43 & 192 & 18:02:13.5022:57:31.4 & 6 @xmath86 2 & 2 & 0.00 @xmath86 0.35 + 44 & 287 & 18:02:13.8322:58:36.1 & 30 @xmath86 6 & 13 & -0.69 @xmath86 0.21 + 45 & 286 & 18:02:13.9523:02:12.7 & 115 @xmath86 11 & 29 & -0.23 @xmath86 0.09 + 46 & 168 & 18:02:14.3722:54:21.5 & 55 @xmath86 7 & 22 & -0.75 @xmath86 0.17 + 47 & 285 & 18:02:15.5623:02:22.5 & 47 @xmath86 7 & 14 & -0.49 @xmath86 0.15 + 48 & 350 & 18:02:15.7923:05:56.1 & 20 @xmath86 6 & 4 & -0.43 @xmath86 0.14 + 49 & 284 & 18:02:15.8522:58:27.4 & 14 @xmath86 4 & 6 & -0.60 @xmath86 0.30 + 50 & 167 & 18:02:16.1722:57:23.4 & 9 @xmath86 3 & 4 & -1.00 @xmath86 0.47 + 51 & 244 & 18:02:16.4422:59:39.5 & 36 @xmath86 6 & 15 & -0.78 @xmath86 0.21 + 52 & 283 & 18:02:16.6122:58:36.8 & 11 @xmath86 3 & 5 & -0.67 @xmath86 0.35 + 53 & 282 & 18:02:16.8123:03:47.0 & 167 @xmath86 14 & 31 & -0.63 @xmath86 0.09 + 54 & 280 & 18:02:17.0122:59:17.8 & 13 @xmath86 4 & 6 & -1.00 @xmath86 0.41 + 55 & 279 & 18:02:17.2822:57:50.9 & 22 @xmath86 5 & 10 & 0.91 @xmath86 0.28 + 56 & 202 & 18:02:17.3622:53:44.6 & 7 @xmath86 3 & 3 & -0.56 @xmath86 0.38 + 57 & 156 & 18:02:17.6522:56:18.0 & 26 @xmath86 5 & 12 & -0.85 @xmath86 0.26 + 58 & 310 & 18:02:17.9123:04:07.3 & 10 @xmath86 4 & 3 & -0.67 @xmath86 0.35 + 59 & 309 & 18:02:17.9723:00:14.0 & 9 @xmath86 3 & 4 & -0.82 @xmath86 0.39 + 60 & 331 & 18:02:18.1823:03:43.4 & 30 @xmath86 6 & 8 & -0.89 @xmath86 0.22 + 61 & 337 & 18:02:18.7823:03:42.4 & 37 @xmath86 7 & 10 & -0.57 @xmath86 0.17 + 62 & 198 & 18:02:18.9022:51:45.1 & 26 @xmath86 6 & 7 & -0.66 @xmath86 0.20 + 63 & 336 & 18:02:19.0923:03:32.7 & 11 @xmath86 4 & 3 & 0.00 @xmath86 0.24 + 64 & 155 & 18:02:19.1722:53:46.3 & 97 @xmath86 10 & 34 & -0.77 @xmath86 0.13 + 65 & 243 & 18:02:19.1922:58:36.7 & 15 @xmath86 4 & 6 & -0.60 @xmath86 0.30 + 66 & 242 & 18:02:19.7022:59:40.6 & 11 @xmath86 3 & 5 & -1.00 @xmath86 0.45 + 67 & 278 & 18:02:19.9823:01:21.3 & 17 @xmath86 4 & 7 & -0.14 @xmath86 0.22 + 68 & 330 & 18:02:20.4323:03:15.7 & 18 @xmath86 5 & 5 & 0.68 @xmath86 0.24 + 69 & 165 & 18:02:20.5722:53:56.2 & 18 @xmath86 4 & 7 & -0.37 @xmath86 0.24 + 70 & 346 & 18:02:20.7623:02:41.1 & 15 @xmath86 4 & 4 & -0.08 @xmath86 0.20 + 71 & 329 & 18:02:21.0522:59:02.5 & 6 @xmath86 2 & 2 & -1.00 @xmath86 0.53 + 72 & 308 & 18:02:21.0623:01:43.6 & 40 @xmath86 7 & 12 & -0.45 @xmath86 0.17 + 73 & 241 & 18:02:21.0723:03:13.9 & 507 @xmath86 23 & 97 & -0.51 @xmath86 0.05 + 74 & 328 & 18:02:21.2023:00:17.3 & 3 @xmath86 2 & 1 & 0.33 @xmath86 0.61 + 75 & 240 & 18:02:21.2622:57:50.1 & 54 @xmath86 7 & 21 & -0.53 @xmath86 0.16 + 76 & 335 & 18:02:21.2723:01:28.8 & 9 @xmath86 3 & 3 & -0.57 @xmath86 0.31 + 77 & 239 & 18:02:21.4623:01:02.5 & 66 @xmath86 8 & 23 & -0.55 @xmath86 0.14 + 78 & 238 & 18:02:21.9323:00:57.7 & 114 @xmath86 11 & 39 & -0.72 @xmath86 0.12 + 79 & 307 & 18:02:21.9623:02:07.2 & 31 @xmath86 6 & 8 & -0.38 @xmath86 0.17 + 80 & 190 & 18:02:22.0422:54:15.7 & 9 @xmath86 3 & 4 & -0.64 @xmath86 0.36 + 81 & 277 & 18:02:22.2423:01:50.5 & 38 @xmath86 7 & 8 & -0.44 @xmath86 0.15 + 82 & 306 & 18:02:22.2723:00:46.4 & 14 @xmath86 4 & 6 & -0.76 @xmath86 0.31 + 83 & 049 & 18:02:22.3622:49:09.3 & 41 @xmath86 8 & 7 & 0.68 @xmath86 0.16 + 84 & 179 & 18:02:22.3622:56:35.2 & 9 @xmath86 3 & 4 & 0.20 @xmath86 0.32 + 85 & 276 & 18:02:22.4223:02:10.7 & 80 @xmath86 9 & 18 & -0.67 @xmath86 0.13 + 86 & 275 & 18:02:22.6123:01:58.9 & 116 @xmath86 12 & 20 & -0.41 @xmath86 0.09 + 87 & 305 & 18:02:22.7823:00:18.5 & 7 @xmath86 3 & 3 & 1.00 @xmath86 0.45 + 88 & 327 & 18:02:22.8123:01:52.3 & 5 @xmath86 3 & 1 & -0.67 @xmath86 0.25 + 89 & 237 & 18:02:22.8622:59:35.6 & 29 @xmath86 5 & 12 & -0.03 @xmath86 0.19 + 90 & 274 & 18:02:22.9023:01:28.5 & 35 @xmath86 6 & 10 & -0.12 @xmath86 0.15 + 91 & 235 & 18:02:23.0523:01:58.1 & 857 @xmath86 30 & 129 & -0.47 @xmath86 0.03 + 92 & 273 & 18:02:23.0623:01:25.3 & 8 @xmath86 3 & 3 & -0.64 @xmath86 0.36 + 93 & 236 & 18:02:23.0923:01:44.8 & 317 @xmath86 18 & 52 & -0.63 @xmath86 0.06 + 94 & 234 & 18:02:23.1623:02:00.3 & 831 @xmath86 29 & 119 & -0.50 @xmath86 0.03 + 95 & 304 & 18:02:23.1923:00:23.6 & 9 @xmath86 3 & 4 & -0.69 @xmath86 0.34 + 96 & 233 & 18:02:23.2123:00:12.3 & 17 @xmath86 4 & 7 & -0.58 @xmath86 0.27 + 97 & 272 & 18:02:23.2523:01:35.0 & 48 @xmath86 7 & 12 & -0.18 @xmath86 0.13 + 98 & 303 & 18:02:23.3223:03:11.3 & 20 @xmath86 5 & 6 & 0.00 @xmath86 0.20 + 99 & 232 & 18:02:23.4123:01:41.5 & 90 @xmath86 10 & 20 & -0.75 @xmath86 0.11 + 100 & 326 & 18:02:23.4123:02:22.4 & 19 @xmath86 5 & 5 & -0.71 @xmath86 0.23 + 101 & 231 & 18:02:23.5123:00:23.3 & 65 @xmath86 8 & 25 & -0.13 @xmath86 0.12 + 102 & 230 & 18:02:23.5423:01:50.9 & 884 @xmath86 30 & 114 & -0.89 @xmath86 0.04 + 103 & 270 & 18:02:23.6323:03:27.7 & 155 @xmath86 13 & 33 & -0.18 @xmath86 0.08 + 104 & 268 & 18:02:23.6523:00:55.6 & 40 @xmath86 6 & 15 & -0.29 @xmath86 0.16 + 105 & 269 & 18:02:23.6623:01:57.3 & 35 @xmath86 7 & 8 & -0.60 @xmath86 0.16 + 106 & 229 & 18:02:23.7123:01:45.4 & 299 @xmath86 18 & 47 & -0.64 @xmath86 0.07 + 107 & 267 & 18:02:23.8322:59:39.0 & 21 @xmath86 5 & 9 & -0.45 @xmath86 0.23 + 108 & 164 & 18:02:23.9522:54:30.7 & 23 @xmath86 5 & 10 & -0.67 @xmath86 0.22 + 109 & 302 & 18:02:23.9823:04:19.2 & 204 @xmath86 15 & 35 & -0.51 @xmath86 0.08 + 110 & 301 & 18:02:24.1723:00:36.6 & 7 @xmath86 3 & 3 & -0.80 @xmath86 0.40 + 111 & 020 & 18:02:24.1922:51:13.2 & 12 @xmath86 4 & 4 & -0.87 @xmath86 0.34 + 112 & 228 & 18:02:24.2922:59:27.1 & 72 @xmath86 9 & 30 & -0.75 @xmath86 0.15 + 113 & 266 & 18:02:24.5723:00:20.7 & 20 @xmath86 5 & 8 & -0.24 @xmath86 0.22 + 114 & 227 & 18:02:24.6323:01:03.7 & 76 @xmath86 9 & 26 & 0.26 @xmath86 0.12 + 115 & 154 & 18:02:24.9022:55:26.4 & 18 @xmath86 4 & 8 & -0.89 @xmath86 0.31 + 116 & 265 & 18:02:25.1223:02:17.4 & 79 @xmath86 9 & 23 & -0.76 @xmath86 0.14 + 117 & 264 & 18:02:25.1623:01:26.7 & 41 @xmath86 7 & 15 & 0.24 @xmath86 0.16 + 118 & 010 & 18:02:25.1722:53:41.0 & 14 @xmath86 4 & 6 & -0.47 @xmath86 0.28 + 119 & 323 & 18:02:25.3423:03:55.5 & 83 @xmath86 10 & 17 & -0.56 @xmath86 0.13 + 120 & 324 & 18:02:25.3523:02:56.7 & 32 @xmath86 6 & 9 & -0.57 @xmath86 0.19 + 121 & 226 & 18:02:25.3723:00:36.3 & 25 @xmath86 5 & 10 & -0.77 @xmath86 0.25 + 122 & 300 & 18:02:25.4323:01:39.9 & 27 @xmath86 5 & 9 & -0.47 @xmath86 0.20 + 123 & 262 & 18:02:25.6323:00:56.9 & 50 @xmath86 7 & 19 & -0.37 @xmath86 0.15 + 124 & 261 & 18:02:25.8423:01:42.6 & 25 @xmath86 5 & 9 & -0.61 @xmath86 0.21 + 125 & 225 & 18:02:25.9323:00:11.6 & 69 @xmath86 8 & 28 & -0.67 @xmath86 0.10 + 126 & 224 & 18:02:25.9423:00:34.6 & 87 @xmath86 9 & 33 & -0.57 @xmath86 0.12 + 127 & 260 & 18:02:25.9522:58:03.2 & 16 @xmath86 4 & 7 & 0.05 @xmath86 0.23 + 128 & 223 & 18:02:25.9923:00:13.4 & 170 @xmath86 13 & 62 & -0.77 @xmath86 0.09 + 129 & 178 & 18:02:25.9922:55:27.8 & 4 @xmath86 2 & 2 & -0.20 @xmath86 0.46 + 130 & 259 & 18:02:26.0123:02:27.7 & 61 @xmath86 8 & 19 & -0.60 @xmath86 0.14 + 131 & 258 & 18:02:26.0923:01:28.6 & 5 @xmath86 2 & 2 & -1.00 @xmath86 0.63 + 132 & 321 & 18:02:26.4723:01:55.9 & 16 @xmath86 4 & 5 & -0.40 @xmath86 0.24 + 133 & 222 & 18:02:26.8523:00:14.3 & 61 @xmath86 8 & 24 & -0.48 @xmath86 0.14 + 134 & 019 & 18:02:26.9522:51:07.9 & 108 @xmath86 11 & 25 & -0.68 @xmath86 0.11 + 135 & 256 & 18:02:27.1323:00:50.6 & 9 @xmath86 3 & 4 & -0.09 @xmath86 0.30 + 136 & 255 & 18:02:27.1723:03:33.4 & 215 @xmath86 15 & 45 & -0.75 @xmath86 0.08 + 137 & 220 & 18:02:27.3222:59:37.7 & 34 @xmath86 6 & 14 & -0.71 @xmath86 0.21 + 138 & 349 & 18:02:27.4523:04:13.7 & 17 @xmath86 5 & 4 & -0.56 @xmath86 0.20 + 139 & 299 & 18:02:27.5623:00:20.3 & 8 @xmath86 3 & 3 & -0.56 @xmath86 0.38 + 140 & 254 & 18:02:27.9722:58:56.9 & 11 @xmath86 3 & 5 & -0.82 @xmath86 0.39 + 141 & 298 & 18:02:28.0323:01:43.0 & 19 @xmath86 5 & 6 & -0.39 @xmath86 0.22 + 142 & 320 & 18:02:28.1323:02:11.1 & 11 @xmath86 4 & 4 & -0.75 @xmath86 0.31 + 143 & 253 & 18:02:28.4122:59:47.3 & 25 @xmath86 5 & 10 & -0.54 @xmath86 0.22 + 144 & 004 & 18:02:28.4422:55:44.5 & 21 @xmath86 5 & 9 & -1.00 @xmath86 0.32 + 145 & 319 & 18:02:28.4423:01:14.1 & 11 @xmath86 3 & 4 & -0.69 @xmath86 0.34 + 146 & 219 & 18:02:28.5123:00:37.7 & 297 @xmath86 17 & 95 & -0.36 @xmath86 0.06 + 147 & 297 & 18:02:28.5223:00:14.6 & 7 @xmath86 3 & 3 & 1.00 @xmath86 0.45 + 148 & 048 & 18:02:28.5422:51:45.8 & 13 @xmath86 4 & 4 & -0.39 @xmath86 0.22 + 149 & 218 & 18:02:28.6523:03:03.4 & 432 @xmath86 21 & 86 & -0.53 @xmath86 0.05 + 150 & 318 & 18:02:28.6623:01:36.8 & 6 @xmath86 3 & 2 & -0.25 @xmath86 0.26 + 151 & 217 & 18:02:28.7422:59:47.4 & 269 @xmath86 16 & 91 & -0.62 @xmath86 0.07 + 152 & 216 & 18:02:28.8022:59:00.1 & 54 @xmath86 7 & 22 & -0.41 @xmath86 0.15 + 153 & 317 & 18:02:28.8123:02:25.5 & 21 @xmath86 5 & 7 & -0.89 @xmath86 0.22 + 154 & 252 & 18:02:29.1023:00:40.9 & 19 @xmath86 4 & 8 & 0.20 @xmath86 0.23 + 155 & 009 & 18:02:29.1422:52:52.6 & 108 @xmath86 11 & 32 & -0.75 @xmath86 0.12 + 156 & 215 & 18:02:29.3523:01:01.6 & 29 @xmath86 5 & 11 & -0.47 @xmath86 0.20 + 157 & 214 & 18:02:29.4722:59:14.7 & 25 @xmath86 5 & 11 & -0.56 @xmath86 0.22 + 158 & 018 & 18:02:29.5322:55:36.5 & 13 @xmath86 4 & 5 & -0.73 @xmath86 0.32 + 159 & 117 & 18:02:29.5922:56:54.4 & 4 @xmath86 2 & 2 & -1.00 @xmath86 0.71 + 160 & 213 & 18:02:29.6022:58:59.3 & 16 @xmath86 4 & 7 & -0.50 @xmath86 0.28 + 161 & 040 & 18:02:29.6822:50:12.5 & 14 @xmath86 4 & 4 & 0.90 @xmath86 0.29 + 162 & 003 & 18:02:29.7922:56:03.5 & 24 @xmath86 5 & 11 & -0.67 @xmath86 0.25 + 163 & 345 & 18:02:30.0323:02:02.5 & 4 @xmath86 2 & 1 & 1.00 @xmath86 0.39 + 164 & 250 & 18:02:30.1923:00:39.5 & 30 @xmath86 6 & 12 & 0.44 @xmath86 0.19 + 165 & 251 & 18:02:30.2022:59:50.2 & 10 @xmath86 3 & 4 & -0.45 @xmath86 0.33 + 166 & 212 & 18:02:30.3722:59:29.1 & 135 @xmath86 12 & 57 & -0.19 @xmath86 0.09 + 167 & 249 & 18:02:30.3823:01:34.5 & 96 @xmath86 10 & 30 & -0.74 @xmath86 0.12 + 168 & 053 & 18:02:30.4722:48:58.6 & 40 @xmath86 8 & 7 & 0.47 @xmath86 0.14 + 169 & 032 & 18:02:30.8922:53:03.8 & 16 @xmath86 4 & 5 & -0.79 @xmath86 0.29 + 170 & 211 & 18:02:30.9323:02:35.6 & 73 @xmath86 9 & 21 & -0.79 @xmath86 0.15 + 171 & 210 & 18:02:30.9823:00:43.6 & 128 @xmath86 11 & 43 & 0.48 @xmath86 0.10 + 172 & 316 & 18:02:31.1023:00:31.5 & 9 @xmath86 3 & 4 & 0.33 @xmath86 0.30 + 173 & 296 & 18:02:31.2023:01:24.4 & 36 @xmath86 6 & 13 & -0.45 @xmath86 0.17 + 174 & 295 & 18:02:31.2223:01:20.8 & 18 @xmath86 4 & 7 & -0.50 @xmath86 0.21 + 175 & 294 & 18:02:31.5523:01:49.4 & 28 @xmath86 6 & 9 & -0.71 @xmath86 0.23 + 176 & 017 & 18:02:31.6722:55:12.1 & 10 @xmath86 3 & 4 & 0.45 @xmath86 0.33 + 177 & 209 & 18:02:31.7123:02:16.8 & 97 @xmath86 10 & 30 & -0.59 @xmath86 0.11 + 178 & 315 & 18:02:31.8223:03:33.9 & 13 @xmath86 4 & 4 & -0.65 @xmath86 0.29 + 179 & 293 & 18:02:31.8423:00:24.9 & 8 @xmath86 3 & 3 & -0.85 @xmath86 0.36 + 180 & 096 & 18:02:31.8922:58:02.0 & 7 @xmath86 3 & 3 & 0.43 @xmath86 0.41 + 181 & 140 & 18:02:31.9423:00:24.7 & 5 @xmath86 2 & 2 & -0.83 @xmath86 0.38 + 182 & 248 & 18:02:31.9923:02:40.4 & 164 @xmath86 13 & 38 & -0.79 @xmath86 0.10 + 183 & 016 & 18:02:32.0022:53:59.8 & 15 @xmath86 4 & 6 & -0.18 @xmath86 0.25 + 184 & 095 & 18:02:32.3022:59:27.9 & 8 @xmath86 3 & 3 & -0.75 @xmath86 0.44 + 185 & 314 & 18:02:32.3823:04:54.8 & 40 @xmath86 7 & 8 & -0.42 @xmath86 0.15 + 186 & 008 & 18:02:32.4022:56:16.0 & 14 @xmath86 4 & 6 & 0.86 @xmath86 0.35 + 187 & 015 & 18:02:32.6022:53:25.6 & 8 @xmath86 3 & 4 & -0.67 @xmath86 0.35 + 188 & 292 & 18:02:32.6123:01:16.4 & 21 @xmath86 5 & 9 & -0.82 @xmath86 0.28 + 189 & 094 & 18:02:32.6122:59:24.5 & 15 @xmath86 4 & 7 & -0.75 @xmath86 0.31 + 190 & 132 & 18:02:32.6322:59:37.6 & 10 @xmath86 3 & 4 & -0.27 @xmath86 0.31 + 191 & 093 & 18:02:32.6722:59:49.5 & 27 @xmath86 5 & 11 & -0.50 @xmath86 0.21 + 192 & 116 & 18:02:32.6822:59:46.3 & 14 @xmath86 4 & 6 & 0.12 @xmath86 0.25 + 193 & 115 & 18:02:32.7423:00:46.7 & 16 @xmath86 4 & 7 & -0.18 @xmath86 0.25 + 194 & 291 & 18:02:32.7523:04:21.0 & 74 @xmath86 9 & 14 & -0.58 @xmath86 0.13 + 195 & 114 & 18:02:32.7922:59:20.5 & 8 @xmath86 3 & 3 & 1.00 @xmath86 0.45 + 196 & 290 & 18:02:33.0223:03:10.8 & 73 @xmath86 9 & 18 & 0.83 @xmath86 0.14 + 197 & 002 & 18:02:33.0322:55:47.7 & 22 @xmath86 5 & 9 & 0.13 @xmath86 0.21 + 198 & 334 & 18:02:33.1623:02:37.4 & 9 @xmath86 3 & 3 & -0.62 @xmath86 0.29 + 199 & 092 & 18:02:33.3423:01:11.3 & 9 @xmath86 3 & 4 & -0.54 @xmath86 0.32 + 200 & 091 & 18:02:33.5323:01:17.4 & 34 @xmath86 6 & 13 & -0.63 @xmath86 0.19 + 201 & 113 & 18:02:33.6722:59:54.3 & 9 @xmath86 3 & 4 & -1.00 @xmath86 0.43 + 202 & 039 & 18:02:33.6722:52:57.4 & 10 @xmath86 3 & 3 & 1.00 @xmath86 0.41 + 203 & 001 & 18:02:33.7822:55:37.5 & 99 @xmath86 10 & 40 & -0.72 @xmath86 0.12 + 204 & 074 & 18:02:33.7822:58:40.9 & 24 @xmath86 5 & 11 & -0.92 @xmath86 0.27 + 205 & 031 & 18:02:33.9022:53:22.8 & 18 @xmath86 4 & 7 & -0.57 @xmath86 0.22 + 206 & 090 & 18:02:33.9122:59:52.6 & 25 @xmath86 5 & 10 & -0.04 @xmath86 0.20 + 207 & 030 & 18:02:34.0222:50:13.4 & 77 @xmath86 10 & 15 & -0.71 @xmath86 0.13 + 208 & 247 & 18:02:34.0323:03:17.9 & 182 @xmath86 14 & 36 & -0.50 @xmath86 0.08 + 209 & 073 & 18:02:34.3323:01:15.4 & 154 @xmath86 13 & 47 & -0.66 @xmath86 0.10 + 210 & 112 & 18:02:34.4522:58:10.8 & 11 @xmath86 3 & 5 & -0.23 @xmath86 0.28 + 211 & 072 & 18:02:34.5122:59:38.0 & 312 @xmath86 18 & 107 & -0.47 @xmath86 0.06 + 212 & 071 & 18:02:34.5723:00:30.9 & 45 @xmath86 7 & 16 & -0.32 @xmath86 0.16 + 213 & 047 & 18:02:34.7022:51:25.3 & 10 @xmath86 4 & 3 & 0.14 @xmath86 0.27 + 214 & 111 & 18:02:34.8523:00:46.1 & 9 @xmath86 3 & 4 & 1.00 @xmath86 0.35 + 215 & 131 & 18:02:34.9123:01:48.7 & 16 @xmath86 4 & 5 & 0.14 @xmath86 0.22 + 216 & 089 & 18:02:35.0523:00:52.6 & 79 @xmath86 9 & 28 & -0.68 @xmath86 0.13 + 217 & 110 & 18:02:35.2423:01:02.3 & 10 @xmath86 3 & 4 & -0.20 @xmath86 0.26 + 218 & 130 & 18:02:35.3323:00:09.7 & 8 @xmath86 3 & 3 & -0.33 @xmath86 0.35 + 219 & 070 & 18:02:35.5522:59:55.5 & 92 @xmath86 10 & 36 & -0.74 @xmath86 0.13 + 220 & 109 & 18:02:35.7423:02:58.6 & 145 @xmath86 12 & 35 & -0.41 @xmath86 0.09 + 221 & 088 & 18:02:35.9522:59:47.0 & 21 @xmath86 5 & 8 & -0.70 @xmath86 0.24 + 222 & 068 & 18:02:35.9823:01:40.8 & 182 @xmath86 14 & 52 & -0.55 @xmath86 0.08 + 223 & 057 & 18:02:36.0722:49:39.5 & 14 @xmath86 5 & 3 & 0.11 @xmath86 0.17 + 224 & 108 & 18:02:36.0922:59:19.9 & 8 @xmath86 3 & 3 & -0.11 @xmath86 0.34 + 225 & 087 & 18:02:36.1123:01:31.9 & 60 @xmath86 8 & 19 & -0.78 @xmath86 0.16 + 226 & 107 & 18:02:36.2922:57:33.3 & 9 @xmath86 3 & 4 & -0.64 @xmath86 0.36 + 227 & 067 & 18:02:36.4522:59:59.6 & 24 @xmath86 5 & 9 & -0.76 @xmath86 0.25 + 228 & 086 & 18:02:36.7022:58:04.2 & 32 @xmath86 6 & 13 & -0.88 @xmath86 0.23 + 229 & 066 & 18:02:36.8123:00:24.0 & 48 @xmath86 7 & 18 & -0.65 @xmath86 0.17 + 230 & 139 & 18:02:36.8323:04:00.6 & 19 @xmath86 5 & 6 & -0.31 @xmath86 0.18 + 231 & 151 & 18:02:36.9223:04:14.8 & 16 @xmath86 5 & 4 & -0.48 @xmath86 0.23 + 232 & 046 & 18:02:37.2722:47:42.9 & 30 @xmath86 7 & 5 & -0.54 @xmath86 0.13 + 233 & 129 & 18:02:37.3323:01:04.1 & 11 @xmath86 4 & 4 & -0.69 @xmath86 0.34 + 234 & 106 & 18:02:37.4423:00:48.7 & 9 @xmath86 3 & 4 & -0.71 @xmath86 0.33 + 235 & 128 & 18:02:37.7623:02:11.9 & 10 @xmath86 3 & 3 & -0.85 @xmath86 0.36 + 236 & 065 & 18:02:38.2622:58:54.8 & 322 @xmath86 18 & 121 & -0.75 @xmath86 0.07 + 237 & 029 & 18:02:38.8522:50:06.1 & 81 @xmath86 10 & 13 & 0.24 @xmath86 0.09 + 238 & 045 & 18:02:38.9422:50:21.0 & 22 @xmath86 6 & 5 & 0.72 @xmath86 0.21 + 239 & 127 & 18:02:39.0322:59:50.5 & 13 @xmath86 4 & 5 & -0.20 @xmath86 0.26 + 240 & 055 & 18:02:39.0722:48:22.8 & 43 @xmath86 9 & 6 & -0.03 @xmath86 0.11 + 241 & 104 & 18:02:39.5923:03:44.0 & 87 @xmath86 10 & 17 & -0.53 @xmath86 0.11 + 242 & 064 & 18:02:39.9423:00:12.9 & 119 @xmath86 11 & 43 & -0.58 @xmath86 0.11 + 243 & 103 & 18:02:40.1123:01:45.8 & 23 @xmath86 5 & 7 & -0.70 @xmath86 0.24 + 244 & 084 & 18:02:40.1823:04:04.5 & 48 @xmath86 8 & 11 & 0.84 @xmath86 0.17 + 245 & 054 & 18:02:40.2122:48:34.8 & 26 @xmath86 7 & 4 & -0.05 @xmath86 0.11 + 246 & 063 & 18:02:40.4222:57:53.3 & 58 @xmath86 8 & 23 & -0.86 @xmath86 0.17 + 247 & 083 & 18:02:40.6123:01:28.4 & 85 @xmath86 9 & 26 & -0.52 @xmath86 0.12 + 248 & 014 & 18:02:40.6622:55:17.5 & 14 @xmath86 4 & 6 & -0.71 @xmath86 0.33 + 249 & 138 & 18:02:40.8323:00:54.6 & 6 @xmath86 3 & 2 & 0.45 @xmath86 0.33 + 250 & 052 &
18:02:41.2522:49:05.4 & 46 @xmath86 8 & 7 & -0.17 @xmath86 0.09 + 251 & 082 & 18:02:41.2523:01:14.0 & 110 @xmath86 11 & 31 & -0.61 @xmath86 0.11 + 252 & 062 & 18:02:41.3722:57:42.4 & 72 @xmath86 9 & 24 & 0.51 @xmath86 0.13 + 253 & 126 & 18:02:41.5022:58:20.8 & 5 @xmath86 2 & 2 & 0.33 @xmath86 0.43 + 254 & 137 & 18:02:41.9123:00:57.9 & 9 @xmath86 3 & 3 & -0.67 @xmath86 0.28 + 255 & 061 & 18:02:42.1922:57:52.1 & 30 @xmath86 6 & 12 & -0.87 @xmath86 0.24 + 256 & 125 & 18:02:42.3623:04:35.8 & 72 @xmath86 9 & 13 & -0.17 @xmath86 0.11 + 257 & 060 & 18:02:42.8222:58:35.9 & 109 @xmath86 11 & 40 & -0.54 @xmath86 0.11 + 258 & 100 & 18:02:42.9523:03:12.4 & 123 @xmath86 12 & 24 & -0.59 @xmath86 0.10 + 259 & 028 & 18:02:43.0322:55:38.5 & 17 @xmath86 4 & 6 & 0.24 @xmath86 0.22 + 260 & 027 & 18:02:43.2022:54:35.4 & 22 @xmath86 5 & 7 & -0.83 @xmath86 0.27 + 261 & 099 & 18:02:43.5123:01:35.4 & 58 @xmath86 8 & 15 & -0.06 @xmath86 0.13 + 262 & 080 & 18:02:43.5422:57:36.0 & 12 @xmath86 4 & 5 & -0.65 @xmath86 0.29 + 263 & 124 & 18:02:44.0422:57:15.6 & 5 @xmath86 2 & 2 & -0.50 @xmath86 0.40 + 264 & 123 & 18:02:44.1823:02:45.1 & 50 @xmath86 8 & 11 & -0.51 @xmath86 0.14 + 265 & 145 & 18:02:44.7622:58:15.4 & 8 @xmath86 3 & 3 & 0.69 @xmath86 0.34 + 266 & 079 & 18:02:44.7822:59:51.9 & 47 @xmath86 7 & 15 & -0.37 @xmath86 0.15 + 267 & 078 & 18:02:45.4522:59:18.4 & 48 @xmath86 7 & 14 & -0.67 @xmath86 0.17 + 268 & 136 & 18:02:45.4622:58:37.4 & 9 @xmath86 3 & 3 & 0.69 @xmath86 0.34 + 269 & 144 & 18:02:45.7422:59:44.2 & 9 @xmath86 3 & 3 & -0.38 @xmath86 0.30 + 270 & 135 & 18:02:45.9922:57:19.6 & 9 @xmath86 3 & 3 & 0.86 @xmath86 0.35 + 271 & 026 & 18:02:46.6322:50:32.2 & 86 @xmath86 11 & 13 & -0.80 @xmath86 0.12 + 272 & 051 & 18:02:46.9522:53:22.5 & 10 @xmath86 4 & 3 & -0.09 @xmath86 0.21 + 273 & 037 & 18:02:47.2022:55:02.4 & 11 @xmath86 4 & 4 & -0.44 @xmath86 0.26 + 274 & 044 & 18:02:47.4922:48:20.4 & 42 @xmath86 9 & 6 & -0.20 @xmath86 0.10 + 275 & 121 & 18:02:47.7222:58:39.3 & 23 @xmath86 5 & 8 & -0.85 @xmath86 0.25 + 276 & 120 & 18:02:47.8222:59:42.2 & 22 @xmath86 5 & 6 & 0.33 @xmath86 0.22 + 277 & 036 & 18:02:47.9022:49:11.5 & 57 @xmath86 9 & 8 & -0.61 @xmath86 0.11 + 278 & 119 & 18:02:48.3823:00:30.6 & 40 @xmath86 7 & 12 & -0.27 @xmath86 0.14 + 279 & 077 & 18:02:48.6322:59:32.4 & 73 @xmath86 9 & 18 & -0.82 @xmath86 0.15 + 280 & 035 & 18:02:48.7022:54:17.6 & 20 @xmath86 5 & 6 & -0.67 @xmath86 0.25 + 281 & 043 & 18:02:48.7522:51:57.1 & 13 @xmath86 4 & 3 & -0.58 @xmath86 0.24 + 282 & 143 & 18:02:49.0222:56:57.4 & 8 @xmath86 3 & 3 & -0.68 @xmath86 0.28 + 283 & 076 & 18:02:49.1622:59:45.5 & 120 @xmath86 11 & 27 & -0.32 @xmath86 0.10 + 284 & 042 & 18:02:49.2422:50:18.7 & 71 @xmath86 10 & 11 & -0.41 @xmath86 0.11 + 285 & 025 & 18:02:49.3622:48:32.6 & 93 @xmath86 12 & 10 & 0.11 @xmath86 0.07 + 286 & 041 & 18:02:49.6622:48:55.6 & 100 @xmath86 12 & 12 & -0.59 @xmath86 0.07 + 287 & 098 & 18:02:49.8422:58:49.6 & 26 @xmath86 5 & 9 & -0.77 @xmath86 0.25 + 288 & 142 & 18:02:50.1223:00:13.4 & 12 @xmath86 4 & 3 & -0.75 @xmath86 0.31 + 289 & 056 & 18:02:50.3622:47:41.0 & 37 @xmath86 10 & 4 & 0.20 @xmath86 0.09 + 290 & 034 & 18:02:50.5222:48:50.3 & 93 @xmath86 12 & 11 & -0.55 @xmath86 0.07 + 291 & 134 & 18:02:51.3723:02:07.4 & 16 @xmath86 4 & 4 & 0.71 @xmath86 0.23 + 292 & 118 & 18:02:51.7123:02:34.3 & 184 @xmath86 15 & 26 & -0.60 @xmath86 0.08 + 293 & 050 & 18:02:52.2722:49:48.8 & 18 @xmath86 6 & 3 & 0.03 @xmath86 0.13 + 294 & 150 & 18:02:53.0923:03:39.0 & 22 @xmath86 6 & 5 & 0.38 @xmath86 0.15 + 295 & 149 & 18:02:54.9022:58:22.1 & 14 @xmath86 4 & 4 & 0.83 @xmath86 0.27 + 296 & 023 & 18:02:57.0822:54:20.1 & 168 @xmath86 14 & 22 & -0.67 @xmath86 0.09 + 297 & 153 & 18:02:57.5523:02:54.4 & 21 @xmath86 6 & 4 & -0.46 @xmath86 0.18 + 298 & 148 & 18:02:57.8323:00:46.4 & 13 @xmath86 5 & 3 & 0.24 @xmath86 0.22 + 299 & 141 & 18:02:58.0622:56:02.4 & 13 @xmath86 4 & 3 & 0.17 @xmath86 0.21 + 300 & 033 & 18:03:00.1922:52:55.7 & 89 @xmath86 12 & 11 & -0.47 @xmath86 0.10 + 301 & 147 & 18:03:01.6123:00:53.6 & 25 @xmath86 6 & 5 & -0.29 @xmath86 0.16 + 302 & 152 & 18:03:02.1923:02:48.7 & 45 @xmath86 9 & 6 & 0.58 @xmath86 0.13 + 303 & 146 & 18:03:03.3223:00:19.2 & 25 @xmath86 7 & 4 & -0.19 @xmath86 0.15 + 304 & 133 & 18:03:03.9323:01:04.2 & 79 @xmath86 11 & 10 & -0.58 @xmath86 0.11 + llllccccccccc [ tinter ] 8 & 196 & 18:01:59.07 - 22:52:27.4 & 73 & 12 & [email protected] & [email protected] & [email protected] & 55% & [email protected] & y & [ y ] & 0.32 + 23 & 173 & 18:02:07.77 - 22:55:33.8 & 58 & 19 & [email protected] & [email protected] & [email protected] & 4% & [email protected] & tts & [ y ] & 0.27 + 38 ( tc4 ) & 347 & 18:02:12.47 - 23:05:48.0 & 62 & 9 & [email protected] & [email protected] & [email protected] & 100% & [email protected] & y ? & n & 1.9 + 91 ( c2)&235 & 18:02:23.05 - 23:01:58.1 & 857 & 129 & [email protected] & [email protected] & [email protected] & 2% & [email protected] & n & n & 6@xmath87 + 94 ( cd ) & 234 & 18:02:23.16 - 23:02:00.2 & 831 & 119 & [email protected] & [email protected] & [email protected] & 79% & [email protected] & y & [ y ] & 14@xmath87 + 97 & 272 & 18:02:23.25 - 23:01:34.9 & 48 & 12 & [email protected] & [email protected] & [email protected] & 8% & [email protected] & yso & n & 0.22 + 102 ( a ) & 230 & 18:02:23.54 - 23:01:50.9 & 884 & 114 & [email protected] & [email protected] & [email protected] & 46% & [email protected] & y & [ y ] & 2 + 106 ( b ) & 229 & 18:02:23.71 - 23:01:45.4 & 299 & 47 & [email protected] & [email protected] & [email protected] & 100% & [email protected] & y & y & 5 + 117 ( tc1 ) & 264 & 18:02:25.16 - 23:01:26.7 & 41 & 15 & [email protected] & [email protected] & [email protected] & 100% & [email protected] & n & n & 4.7 + 166 & 212 & 18:02:30.37 - 22:59:29.0 & 135 & 57 & [email protected] & [email protected] & [email protected] & 0% & [email protected] & yso & u & 0.62 + 170 & 211 & 18:02:30.93 - 23:02:35.6 & 73 & 21 & [email protected] & [email protected] & [email protected] & 68% & [email protected] & n & y & 0.33 + 194 & 291 & 18:02:32.75 - 23:04:21.0 & 74 & 14 & [email protected] & [email protected] & [email protected] & 2% & [email protected] & ( yso ) & [ y ] & 0.34 + 211 & 72 & 18:02:34.51 - 22:59:37.9 & 312 & 107 & [email protected] & [email protected] & [email protected] & 43% & [email protected] & y & [ y ] & 1.1 + 237 & 29 & 18:02:38.85 - 22:50:06.0 & 81 & 13 & [email protected] & [email protected] & [email protected] & 2% & [email protected] & ( yso ) & n & 6.32 + 246 & 63 & 18:02:40.42 - 22:57:53.3 & 58 & 23 & [email protected] & [email protected] & [email protected] & 57% & [email protected] & y & [ y ] & 0.25 + 256 & 125 & 18:02:42.36 - 23:04:35.7 & 72 & 13 & [email protected] & [email protected] & [email protected] & 0% & [email protected] & n & n & 0.33 + 283 & 76 & 18:02:49.16 - 22:59:45.4 & 120 & 27 & [email protected] & [email protected] & [email protected] & 3% & [email protected] & yso & [ y ] & 0.55 + 285 & 25 & 18:02:49.36 - 22:48:32.5 & 93 & 10 & [email protected] & [email protected] & [email protected] & 2% & [email protected] & n & n & 6.2 + | the trifid nebula , a young star - forming hii region , was observed for 16 hours by the acis - i detector on board of the _ chandra_x - ray observatory .
we detected 304 x - ray sources , thirty percent of which are hard sources and seventy percent of of which have near - infrared counterparts .
_ chandra _ resolved the hd164492 multiple system into a number of discrete x - ray sources .
x - ray emission is detected from components hd164492a ( an o7.5iii star which ionizes the nebula ) , b and c ( a b6v star ) , and possibly d ( a be star ) .
component c is blended with an unidentified source to the nw .
hd164492a has a soft spectrum ( @xmath0 kev ) while the component c blend shows much harder emission ( @xmath1 6 kev ) .
this blend and other hard sources are responsible for the hard emission and fe k line seen by _ asca _ , which was previously attributed entirely to hd 164492a .
the soft spectrum of the o star is similar to emission seen from other single o stars and is probably produced by shocks within its massive stellar wind .
lack of hard emission suggests that neither a magnetically confined wind shock nor colliding wind emission is important in hd164492a .
a dozen stars are found to have flares in the field and most of them are pre - main sequence stars ( pms ) .
six sources with flares have both optical and 2mass counterparts .
these counterparts are not embedded and thus it is likely that these sources are in a later stage of pms evolution , possibly class ii or iii .
two flare sources did not have any near - ir , optical , or radio counterparts .
we suggest these x - ray flare stars are in an early pre - main sequence stage ( class i or earlier ) .
we also detected x - ray sources apparently associated with two massive star forming cores , tc1 and tc4 .
the spectra of these sources show high extinction and x - ray luminosities of @xmath2 erg s@xmath3 .
if these sources are class 0 objects , it is unclear if their x - ray emission is due to solar - type magnetic activities as in class i objects , or some other mechanism . |
the optical afterglow of grb 000301c , at a redshift of @xmath6 , exhibited variability prior to the break in the power law decline in its lightcurve .
this variability has been interpreted as gravitational microlensing ( garnavich , loeb & stanek 2000 ) , as first proposed by loeb & perna ( 1998 ) .
while there is no obvious intervening galaxy which might provide lenses in the form of stars or machos , we will take this possibility seriously .
koopmans and wambsganss ( 2001 ) considered lensing by machos ( see also wyithe & turner ( 2002b ) ) , but as we will explain in [ discussion ] we find a rate of macho lensing five times larger than theirs
. the rate of microlensing by stars is much lower , and is naively of order one in three hundred grbs at redshift two .
we note that microlensing is not the only possible explanation of this lightcurve ( panaitescu 2001 ) .
several authors have considered using microlensing to study the properties of the underlying afterglow ( granot & loeb 2001 ; gaudi & loeb 2001 ; gaudi , granot & loeb 2001 ; ioka & nakamura 2001 ) . in this work
we will only make statements about afterglows in the aggregate .
assuming that there is a fraction @xmath7 of dark matter in machos , it is not surprising that a lensing event was seen .
however , if the only cosmological microlenses are stars , the observation of an event is somewhat problematic .
roughly thirty afterglows have been well characterized ( starting with metzger et al .
( 1997 ) and kulkarni et al .
( 1998 ) and summarized in frail et al .
( 2001 ) and bloom , frail & kulkarni ( 2003 ) ) .
we investigate whether magnification bias can make a significant difference , considering the two extreme possibilities that the optical and gamma ray fluxes are perfectly correlated or perfectly uncorrelated . in appendix
[ appdx : binary ] we consider binary lenses , as discussed by mao & loeb ( 2001 ) .
neither of these two effects seems to be able to increase the probability by a factor of ten .
we thus conclude that the observation of a second microlensing event by a star in a sample twice as large is quite unlikely , though with a large macho fraction the probability can be significant . in [ physics ]
we summarize the basic physics involved in gamma ray burst fireballs , and give an overview of the multiwavelength observations . in [ microlensing ]
we derive the cosmological microlensing rate ( relegating the details for binary lenses to appendix [ appdx : binary ] ) . in the remaining sections ( [ bias ] [ sample ] ) we discuss the enhancement of the lensing probability due to magnification bias ( turner , gott & ostriker 1984 ) . magnification bias is the effect that a given _ detected _ source is more likely to be lensed than a source chosen randomly , simply because magnified sources are easier to detect .
we will consider both the usual single magnification bias , as well as double magnification bias which might be relevant here because of the way gamma - ray burst afterglows are selected .
the selection is complicated , involving gamma ray , x - ray and optical wavelengths .
we can only hope to capture the basic features . in [ bias ] & [ doublebias ] we consider how the magnification bias affects bursts at a single redshift , while in [ sample ] we consider the magnification bias on the sample of well measured afterglows in bloom et al .
( 2003 ) .
we primarily consider the gravitational lensing caused by point mass lenses .
we define the distance to the source @xmath8 , lens @xmath9 , and deflector source distance @xmath10 .
these distances are all angular diameter distances in physical units , and we take a flat cosmology where @xmath11 . we define the einstein radius and the einstein angle for the system as @xmath12 where @xmath13 is the mass of the point lens .
the basic picture of a gamma - ray burst ( grb ) afterglow is a fireball with a relativistic jet with some small opening angle ( e.g. waxman 1997a ) .
the observed afterglow light is emitted in a ring of opening angle @xmath14 , where @xmath15 is the lorentz factor of the ejecta moving at velocity @xmath16 .
this ring expands at an apparent rate @xmath17 , the so called superluminal expansion .
this superluminal expansion is what allows for rapid microlens variability , even at cosmological distances ( loeb & perna 1998 ) .
we take the radius of the afterglow ring in units of the einstein radius from waxman ( 1997b ) and garnavich et al .
( 2000 ) , @xmath18 where @xmath19 is the isotropic burst energy in units of @xmath20 ergs , and @xmath21 is the ambient gas density in units of 1 @xmath22 .
we will assume for this work that we want a microlensing event to occur at some minimum time after the grb , to allow for optical observations . to enforce this assumption
we take @xmath23 , typically achieved roughly one tenth day after the burst at a redshift of @xmath2 .
we define the fractional apparent width of the emitting region @xmath24 , where the emission comes in the annulus between @xmath25 and @xmath26 .
in fact , there is some emission throughout the interior of the ring , but the inner part is less bright ( granot , piran & sari 1999 ) . for our later calculation , we need to know the intrinsic ( not apparent ) luminosity function of grbs , at different wavebands . this is a subject of much uncertainty at the moment .
schaefer , deng & band ( 2001 ) found that the luminosity function @xmath27 at gamma - ray energies is a power law of slope @xmath28 at the faint end .
this result is obtained by assuming a lag variability relation .
note that bursts of all durations are included in their analysis , whereas bursts for which afterglows have been observed fall into the long duration category ( as localization of the bursts is typically done with bepposax observations which are insensitive to bursts with durations less than about a second ) .
frail et al .
( 2001 ) demonstrated using about 20 afterglows that the total energy output is remarkably constant from burst to burst , about @xmath29 ergs , most of which comes out in gamma - ray .
bloom et al .
( 2003 ) confirmed this result with a larger sample of about 30 afterglows .
however , what is important to us is not the true total energy , but rather the isotropic equivalent energy .
it is the latter that determines the effective luminosity of the burst ( i.e. the observed fluence is proportional to the isotropic equivalent energy divided by the luminosity distance squared ) .
the two are related via a geometrical factor determined by the opening angle of the burst .
the luminosity function is therefore determined by the distribution of opening angles .
frail et al .
( 2001 ) derived this distribution using the observed jet break times .
however , it is important to keep in mind that the distribution of opening angles is obtained from a flux - limited sample . to relate that to the intrinsic distribution of opening angles , and therefore luminosities , would require detailed modeling of the selection , which is beyond the scope of this paper .
krumholz , thorsett & harrison ( 1998 ) derived constraints on the grb luminosity function by matching the number - peak - flux relation in the batse catalog ( see also rutledge , hui & lewin 1995 ) , as well as making use of three bursts with known redshifts at the time . assuming that the redshift distribution follows the global star formation history ( see madau et al .
1996 ; madau , pozzetti & dickinson 1998 ; wijers et al .
1998 ) , they find that a power - law slope of @xmath30 to @xmath31 is consistent with observations ( see fig . 1 in krumholz et al . 1998 ) .
the luminosity function at optical wavebands is even more uncertain , because of the limited number of afterglows . on theoretical grounds ,
one expects the gamma - ray emission to be well correlated with x - ray , but not with the optical .
this is because the x - ray frequencies are generally above the various characteristic frequencies of the relativistic fireball , such as e.g. the synchrotron self absorption frequency , whereas the ( unknown ) relative ordering of the characteristic frequencies affects the optical flux significantly
. some observational evidence of this can be found in freedman & waxman ( 2001 ) ( see also useful discussions in kumar ( 2000 ) and sari , piran & narayan ( 1998 ) ) . for our later investigation , we will examine both the case where the gamma - ray / x - ray and optical emission are well - correlated , and the case where they are not .
as far as the slope of the optical luminosity function is concerned , we will assume , for simplicity and the lack of evidence to the contrary , it is the same as that for the gamma - ray emission .
observation of an optical afterglow has a complicated selection process , which we will outline here .
first , the burst is detected in gamma - rays ( @xmath32 kev or so ) by batse , but it is at lower x - ray energies that the bursts are localized by e.g. bepposax . after a localization ,
an attempt to measure the optical afterglow can be made .
thus , to detect an optical afterglow , there are effectively three thresholds that must be exceeded . the gamma - ray and x - ray fluxes
are believed to be well correlated , but the correlation with the optical flux is uncertain ( granot & sari 2002 ) .
the microlensing optical depth is defined as the number of lenses within some minimum impact parameter of the line of sight to an object .
the basic optical depth , where the impact parameter is taken to be an einstein radius , is given by the following for a flat cosmology ( press & gunn 1973 ) : @xmath33 where @xmath34 is the dimensionless coordinate distance , i.e. the luminosity distance is @xmath35 . in fig .
[ fig : tau ] we plot the optical depth as a function of redshift for several values of @xmath36 in a flat cosmology , dividing out the @xmath37 dependence .
the quantity @xmath37 is the fraction of critical density today that is in objects capable of causing microlensing .
we have assumed that the comoving number density of such objects is constant if it decreases towards higher redshifts , the optical depth will be lower . for @xmath38 and @xmath2 ,
the optical depth is @xmath39 .
fukugita , hogan & peebles ( 1998 ) estimated @xmath37 in stars to be @xmath40 . with this value , constant out to redshift @xmath2 , the optical depth is 1/440 for @xmath38 , and is as large as 1/260 for @xmath41 , though the latter value for the matter density is strongly disfavored . for machos
, we will simply assume that @xmath42 , namely that the halo macho fraction is universal . according to fukugita et al .
( 1998 ) , roughly half of the baryons are accounted for at redshift zero , so the density of baryonic machos should not exceed @xmath43 , and thus @xmath44 for @xmath38 and @xmath45 .
of course non - baryonic machos could have @xmath46 .
we will thus take @xmath47 as the largest plausible value , though even this value has problems ( graff et al .
1999 ; fields , freese & graff 2000 ) .
there are limits on the cosmological abundance of machos of any kind ( dalcanton et al .
1994 ) , but @xmath48 is allowed for stellar mass lenses . in the baryonic case ,
the optical depth to redshift @xmath2 is as large as 1/75 , which is not excessively smaller than the observed rate . though we discussed excising the central region @xmath49 of the einstein ring in the previous section , to account for afterglow lensing events that happen too quickly to be detected
, doing so has only a small effect on the rate , in fact this region is only 4% of the einstein ring .
this cut will have a larger effect in the next section , where we will be discussing magnification bias .
the actual optical depth of relevance depends on the magnification one is interested in .
assuming @xmath50 is the magnification threshold of interest , the optical depth is : @xmath51=\tau_0\beta^2.\ ] ] where @xmath52 is the peak magnification for the given lens system at a distance @xmath53 in einstein units from the burst . here , @xmath54 is the distance from the lens in einstein units where the peak magnification is @xmath50 .
the symbol @xmath55 denotes a step function ( @xmath56 if @xmath57 , and zero otherwise ) . ignoring magnification bias , the optical depth @xmath58 is equal to the rate of microlensed grbs . for grb 000301c ,
the peak magnification was approximately @xmath59 . in table
[ tab : tau ] we give the optical depth to lensing for several different magnification thresholds and we also consider three cases for the fractional width of the emitting ring in the gamma ray burst afterglow .
we find that the consideration of binary lenses typically increases the optical depth , but not by a large amount ( see appendix [ appdx : binary ] ) .
the two main effects that we see are that binaries give a larger increase for narrower source rings , and a larger increase for larger thresholds .
note that for this calculation , we only require that the peak magnification exceed the threshold , and we do not consider the detailed structure of the lightcurve . in the remainder of this paper
, we will assume that @xmath38 , @xmath59 and @xmath60 , which gives @xmath61 .
we will consider both stellar lenses with @xmath62 and machos with @xmath42 .
for microlensing of point sources , the probability of an impact parameter less than @xmath54 in einstein units , and thus a magnification greater than @xmath63 is simply given by @xmath64 we will use the approximate expression , and thus we need to truncate to ensure that @xmath65 , @xmath66 the probability density corresponding to this expression is now @xmath67 i.e. @xmath68 gives the probability that the magnification lies in the range @xmath69 . we will briefly discuss the effects of an inhomogeneous background mass distribution in [ discussion ] .
we now consider the magnification bias ( ignoring the complication of binary lenses , which as we have seen does not significantly affect the lensing rate ) .
we compute the probability of a detectable microlensing event on the afterglow of a grb detected at high energy .
the detectability of the grb at high energy will be affected by magnification bias as a point source .
we then simply model the microlensing event as a detected grb with a lens between 0.2 and 1.2 einstein radii distant .
the inner boundary allows a minimum time before peak magnification ( so that a bump in the lightcurve can be observed ) , and the outer boundary simply enforces that the magnification is in excess of roughly 2 in the peak ( for @xmath60 ) .
the precise values of these parameters do not significantly affect our results , however , for luminosity functions with slopes @xmath70 .
implicit in this prescription is the assumption that the magnification of the the gamma ray flux is the same as the magnification of the optical flux , when the optical flux is first detected .
this is a good assumption as typically the afterglow would be detected in the regime where the emission region is small compared to the einstein radius , meaning the afterglow is effectively a point source when initially detected .
we now derive expressions for the number of detected sources , given lensing effects , following the discussion in turner , ostriker & gott ( 1984 ) .
we assume a simple selection function @xmath71 which gives the efficiency of detecting sources with flux @xmath72 , and an absolute luminosity function @xmath73 , where @xmath74 , and @xmath75 is the number of sources per luminosity range . in this section and
the next , we will consider bursts at a fixed redshift , thus the apparent luminosity function is trivially related to the true luminosity function . without lensing ,
the number of detected sources is then @xmath76 ) n(l ) .
\label{n0}\ ] ] taking into account lensing , the number of detected sources is @xmath77)\int \frac{da}{a}\,\,\frac{dp}{da}\,n\left(\frac{l}{a}\right ) .
\label{nlens}\ ] ] we will consider a simple power law luminosity function , with a low and high flux cutoffs , @xmath78 and a simple choice of selection function @xmath79 where @xmath80 .
we define @xmath81 and @xmath82 ( we expect @xmath83 ) . here
we note that if the slope is in the range @xmath84 , the cutoffs can be neglected .
limit comes from the convergence of the integrated luminosity function ( @xmath85 ) at the low luminosity end , while the @xmath86 limit arises from the the convergence of @xmath87 in the large @xmath88 limit , where @xmath89 . ] with these simple forms , we find the following expression for the unlensed source count , which of course is not measurable , @xmath90 not explicitly stated in eq .
[ nlens ] is the fact that one is generally interested in the number of sources with magnification larger than some value , or in some particular range .
we define the quantity @xmath91 which is equal to the number of sources with magnification @xmath92 , normalized by @xmath93 , the total number of detected sources if lensing were absent ( eq . [ n0 ] ) . with the power - law luminosity function , this takes the following form @xmath94.\ ] ] we can compute @xmath95 explicitly for the analytic @xmath96 ( eq . [ ptau0 ] ) .
first , we will need the integral over the analytic probability density of an arbitrary power of amplification , given by @xmath97 and we have used the definition @xmath98 with @xmath99 being the standard hypergeometric function . we notice that this function has simple poles when @xmath100 is a non - positive integer , and their residues are independent of @xmath101 , thus when two such functions are subtracted , all poles vanish .
we define the quantity @xmath102 , which gives us the ratio of the number of all detected sources taking into account lensing ( with all possible magnification ) to the number of all detected sources ignoring lensing .
note that by virtue of the approximation made in eq .
[ patau0 ] , we approximate @xmath103 by @xmath104/\sqrt{1 + 4 \tau_0})$ ] . instead of @xmath105 ( see eq . [ ptau0beta ] )
results in negligible changes to our results . ] expanding @xmath106 in the limit of @xmath107 , we find where we have used the following expansion of @xmath108 in the limit of small @xmath109 : @xmath110 we are interested in the number of microlensed sources , which we defined previously as sources falling in an annulus about a lens , with inner and outer radii of 0.2 and 1.2 in einstein units , respectively .
the number of detected microlensed sources is then just the difference of the bias factors @xmath111 for minimum magnifications corresponding to the inner and outer edges of the annulus .
thus we find that the ratio of the number of microlensed sources to the number of detected sources , or equivalently the probability of a given burst being microlensed , is given by @xmath112 where the most naive expectation ( ignoring magnification bias but taking into account lensing ) is that @xmath113 . in fig .
[ fig : single ] we plot @xmath114 as a function of @xmath15 for several values of @xmath115 and @xmath116 , with the constraint @xmath117 , in other words the luminosity function ranges over four orders of magnitude . and
@xmath118 , defined after eq . [ sf ] , are both small . their exact size , as can be seen from fig .
[ fig : single ] , does not matter a great deal .
we choose the range @xmath119 to span four orders of magnitude as schaefer et al .
( 2001 ) give the luminosity function over this range .
if @xmath119 were larger , our results are not much affected .
] we have used the magnification distribution derived from the distribution in @xmath120 , though the results do not differ much from those of the analytic @xmath121 .
we see that the magnification bias is only significant when the slope @xmath15 becomes significantly larger than one .
note that we have fixed @xmath122 in fig .
[ fig : single ] , but the curves depend on the value of @xmath109 only weakly . in other words ,
the lensing probability is nearly linear in @xmath109 . using the value of @xmath123 for the faint - end luminosity function from schaefer et al .
( 2001 ) , we find the magnification bias as a function of the ratios @xmath115 and @xmath116 , @xmath124 one can see that if @xmath115 and @xmath116 can be ignored , @xmath125 . even with this slope significantly less than three
, the magnification bias causes an enhancement of the probability of a factor of roughly fifty percent . we should emphasize that magnification bias offers a smaller enhancement if @xmath15 were smaller
in the previous section we effectively assumed that the gamma - ray / x - ray and optical luminosities were perfectly correlated . we now take the other extreme position , assuming that the gamma - ray / x - ray and optical luminosities are completely uncorrelated , though , for simplicity , both drawn from luminosity functions with the same shape but with different thresholds ( see discussion in [ physics ] ) .
it has been shown that this situation can give rise to significantly larger magnification bias ( borgeest , van linde & refsdal 1991 ; wyithe , winn & rusin 2002 ) .
as before , the unlensed number counts are given by @xmath126 ) n(l_v)\int dl_x\,s_x ( f [ l_x , z ] ) n(l_x).\ ] ] where the subscripts @xmath127 and @xmath128 stand for the visible and x - ray wavebands relevant for selection of sources .
note that here @xmath129 gives the number of sources with the luminosities in the two wavebands falling into the respective ranges .
taking into account lensing , the number of detected sources is @xmath130)\int dl_x\,s_x ( f[l_x , z ] ) \int \frac{da}{a^2}\,\,\frac{dp}{da}\ , n\left(\frac{l_v}{a}\right)n\left(\frac{l_x}{a}\right).\ ] ] we will now define some similar variables to the previous section .
we take the luminosity functions in the two bands to be identical , with the same slope and range between lower and upper cutoffs , though the detection thresholds differ . and
@xmath131 , and the cutoffs for another : @xmath132 and @xmath133 , can be completely different .
what we assume here is that the range @xmath134 equals @xmath135 .
our expressions here implicitly assume a rescaling has been done so that @xmath136 and @xmath137 .
the rescaling factors are omitted from our expressions , because they get canceled out in the end as far as the probability of lensing is concerned .
] we define @xmath115 and @xmath116 as before , based on the _ smaller _ detection threshold ( relative to the lower luminosity function cutoff ) , and we define @xmath138 as the ratio of the smaller to larger threshold .
the unlensed source counts are given by @xmath139 where @xmath140 is now the lower of the two thresholds .
similarly to the previous section , we compute the excess source counts above a given magnification @xmath50 due to the double magnification bias , @xmath141 ( eq . [ ba0 ] ) : @xmath142 \left[\max\left(\eta^{-1}a^{-1},{f_{\rm l}}\right)^{1-\gamma}-{f_{\rm h}}^{\gamma-1}\right ] .
\nonumber\end{aligned}\ ] ] we now repeat the analysis of the previous section for the double magnification bias , with the same range of allowed magnifications , again using eq . [
eq : plens ] .
we plot the ratio as before in fig .
[ fig : double ] , where we again fix @xmath122 .
we will take @xmath143 , and the results are insensitive to the exact value .
we see that the magnification bias is insignificant for the case where @xmath144 , but it can be as large as three or four in the case where @xmath145 ( motivated by the results of schaefer et al .
2001 ; see [ physics ] for more discussions ) .
note that , however , if @xmath15 gets close to @xmath146 or larger , the enhancement in lensing probability due to double magnification bias can become very large , limited only by the cut - offs in the luminosity functions
i.e. the result will be quite cut - off dependent .
in fact in some cases a downturn is expected , as the very high magnification regime yields microlensing events too short to be detected .
lastly , if the two luminosity function slopes are different , the important quantity is their average , so if @xmath147 , then the double magnification bias becomes large .
the results of the previous two sections are valid for bursts at a single redshift only . to be more precise , we will consider a population of afterglows from a range of redshifts .
we choose 27 detected afterglows as our sample , taken from bloom et al .
( 2003 ) and listed in table [ tab : grb ] .
while we caution that the sample is a heterogeneous one , with a probably quite complex selection process , we will continue to model only the part of the selection that has to do with the observed flux ( in gamma - ray as well as in optical ) that our results are not sensitive to the precise thresholds chosen ( unless @xmath15 is very close to @xmath148 or larger for single magnification bias , or @xmath146 or larger for double magnification bias ) will be regarded as a partial justification for our model .
we calculate the microlensing optical depth as a function of redshift , as before assuming the flat @xmath38 cosmology , and assuming that the shape of the luminosity function does not evolve . for each redshift
, we compute the probability of a detectable microlensed grb . for the full sample ,
we then simply compute the probability that at least one observed afterglow was microlensed . in fig .
[ fig : sample ] we show results for various values of the thresholds , and for both @xmath117 and @xmath149 . in principle
we could use the flux information for each burst to construct a probability , but as we show in appendix [ appdx : flux ] , this would make little difference .
we have thus far assumed that lenses are uniformly distributed ( i.e. not clustered ) , an approximation that can fail badly . for lenses distributed in isothermal galactic halos , as
machos would be wyithe & turner ( 2002a ) found that the typical cosmological microlensing event occurs at small optical depth , and thus the isolated lens approximation is justified . in the top panel of fig .
[ fig : sample ] we illustrate the probability to see at least one event in the sample assuming @xmath150 ( machos making up half of @xmath151 ) .
this probability is quite high : about 25% with no magnification bias and approaching 35% and 50% for single and double magnification bias with @xmath145 , respectively .
these probabilities make it likely that another microlensed grbs could be observed if the sample were to be doubled .
in contrast , wyithe & turner ( 2002a ) found that microlensing by stars typically occurs at optical depths of order unity , and the isolated lens approximation is not appropriate .
koopmans & wambsganss ( 2001 ) have addressed this problem by simulating lens systems at optical depths of order unity .
they find a microlensing probability proportional to optical depth for small optical depths , but turning over at optical depths around 1/10 . choosing a typical value of @xmath152 according to wyithe & turner ( 2002a ) ,
the effectiveness of lenses is of order only one third the naive value . in discussing stellar lenses
we will thus take an effective lens density @xmath153 . in the bottom panel of fig .
[ fig : sample ] we illustrate the probability of observing at least one grb lensing event due to a star in the sample . without magnification bias , this is about 1.5% , rising to 2% and 3.5% for single and double magnification bias respectively .
it thus seems unlikely that a stellar lens could be responsible .
our estimate of the microlensing rate for gamma ray burst afterglows is in significant disagreement with previous literature , which we outline here . if @xmath46 , then the optical depth to @xmath2 is @xmath154 , in significant disagreement with koopmans and wambsganss ( 2001 ) . computing the mean value of @xmath120 from their eq . 4 ( this is identified with our @xmath109 ) , we find @xmath155 , a factor of five smaller .
they have cut their isothermal halos at radii that are too small , and have thus neglected 80% of their mass .
they have done this to enforce that the halos do not overlap along the line of sight , but the fact is that they do , and significantly .
they have thus underestimated the lensing probability by a factor of five .
let us explain this in more detail .
in [ microlensing ] we assumed an optical depth for microlensing consistent with a uniform distribution of lenses .
this is clearly not the case , as mass in clustered in halos .
we assume that all of the matter is in singular isothermal sphere halos with cutoffs ( @xmath156 ) .
these will obey a luminosity velocity dispersion relation , have luminosities taken from a press - schechter distribution , and have relation between cutoff radii and velocity dispersion , @xmath157 following koopmans & wambsganss ( 2001 ) , we will take @xmath158 , @xmath159 , as the fiducial case , with @xmath160 mpc@xmath161 , @xmath162 km s@xmath163 . to prescribe the cutoff radii ,
we take @xmath164 , but the results are insensitive to the precise value of beta as long as it is not close to zero .
we will compute @xmath165 by normalizing the total mass density to account for @xmath38 , namely all mass lies in halos .
we find @xmath166 we now can easily compute the geometric cross section of halos , and the mean number of halos intersected on a line of sight to a particular redshift , @xmath167 even to moderate redshift , a significant number of halos are intersected . for grb 000301c at @xmath6 , @xmath168 , and
the approximation that halos do not overlap is therefore not accurate .
this is the origin of our disagreement .
we have shown that a large number of halos are typically intersected .
however , there will be variations in the optical depth ( or equivalently , convergence ) between lines of sight .
we have done a simple test , simulating the distribution of convergences and modifying the magnification distribution @xmath121 accordingly .
the magnification bias we find does not vary by more than 15% in this case .
we have shown that the consideration of magnification bias ( both single and double ) is important , in that the expected probability that a given observed gamma ray burst is microlensed can be roughly three times the naive optical depth estimate .
machos making up half of the baryonic matter have a significant probability to cause such an event , but we find that stellar lenses alone have a very small probability ( less than 3% for any event in the afterglow sample ) to do so .
we find that binaries can affect the gamma ray burst microlensing rate at the 25% level , which though surprisingly large , is not very significant .
a single observed event with a low probability is not in itself problematic .
however , should another microlensed gamma ray burst be observed in another thirty afterglows , we believe it would be a significant indication that there are significantly more lenses in the universe than can be inferred from stars . in that case
, the effects we have calculated could be quite important in determining the exact density of lenses .
we thank jules halpern for numerous enlightening discussions , and andrei gruzinov , leon koopmans , pawan kumar , dani maoz , reem sari and eli waxman for useful conversations .
eab acknowledges support from the columbia university academic quality fund .
lh acknowledges support by an outstanding junior investigator award from the doe , grant ast-0098437 from the nsf and grant nag5 - 10842 from nasa .
this research was supported in part by the nsf under grant phy99 - 07949 at the kavli institute for theoretical physics at the university of california , santa barbara .
we discuss the two point mass gravitational lens , a model that has been studied extensively ( schneider & wei 1986 ; erdl & schneider 1993 ) .
we quote some relevant results .
we take the masses of the lenses to be @xmath169 and @xmath170 , with @xmath171 , and we define dimensionless masses @xmath172 such that @xmath173 . furthermore , we define @xmath174 , with @xmath175 without loss of generality . note that taking @xmath176 yields the single point mass lens .
the excess optical depth due to binaries is given by @xmath177 for the binary systems , we take a distribution flat in @xmath178 , and also flat in @xmath179 , where @xmath180 is the projected separation of the lenses on the sky , for the ensemble average ( baltz & gondolo 2001 ) .
we adopt a complex parameterization ( witt 1990 ) of the lens system .
we introduce complex angular coordinates on the plane of the sky , @xmath181 . given two lenses at angular positions @xmath182 and @xmath183 , a source at @xmath184 will have images at the solutions @xmath185 of the lens equation @xmath186 in this formalism , the jacobian of @xmath187 in eq .
[ eq : zeta ] is @xmath188 the magnification of an image at @xmath185 of a point source at @xmath184 is given by @xmath189 , with the sign of @xmath190 giving the parity @xmath191 of the image .
the total magnification is the sum of the individual magnifications of the images . for a source of uniform surface brightness
, the magnification is just the image area divided by the source area ( gould & gaucherel 1997 ) , @xmath192\bigg{/ } \int_0^{2\pi}d\phi\;{\rm im}\left[\overline{\zeta}\frac{d\zeta}{d\phi } \right].\ ] ] we never need to use numerical differencing as long as we can analytically differentiate the expression for the source boundary @xmath193 . as we are concerned with annular sources
, the source boundaries will always have the form @xmath194 , which can be easily differentiated .
differentiating the lens equation , we find @xmath195 solving for @xmath196 , we find the following result , depending only on the numerical solution of the lens equation , and not on any numerical derivatives @xmath197 we can take care of the required parities trivially . since we effectively need @xmath198 , we simply take the absolute value of the jacobian factor @xmath190 in eq .
[ eq : dzdphi ] , since @xmath199 .
given that we have the gamma ray fluxes for this sample of bursts , we could consider using that information in the determination of the probability .
in particular , we could write the probability for an observed burst with apparent flux @xmath200 and redshift @xmath201 , @xmath202 this compares with eq .
[ eq : plens ] for the single magnification bias , which can be written in a similar form @xmath203 for a power law luminosity function with no cutoffs , as is approximately the case when @xmath84 , these two probabilities are identical burst by burst .
summing over the sample reduces the small discrepancy between them when the luminosity function is not a pure power law .
ccccccccccccc 5.0 & & 0.135 & 0.142 & 4.93% & & 0.145 & 0.179 & 23.6% & & 0.185 & 0.259 & 40.0% + 3.0 & & 0.341 & 0.359 & 5.22% & & 0.459 & 0.525 & 14.4% & & 0.599 & 0.763 & 27.3% + 2.0 & & 1.007 & 1.134 & 1.77% & & 1.399 & 1.503 & 7.47% & & 1.849 & 2.121 & 14.7% + 1.5 & & 2.785 & 2.793 & 0.30% & & 4.022 & 4.111 & 2.21% & & 5.534 & 5.901 & 6.63% lccccc 970228 & 0.695 & 0.5817 & 0.972 & 11.0 & 10.69 + 970508 & 0.835 & 0.6730 & 1.525 & 3.17 & 4.834 + 970828 & 0.958 & 0.7473 & 2.141 & 96.0 & 205.5 + 971214 & 3.418 & 1.5712 & 48.19 & 9.44 & 454.9 + 980613 & 1.096 & 0.8245 & 2.987 & 1.71 & 5.108 + 980703 & 0.966 & 0.7519 & 2.185 & 22.6 & 49.38 + 990123 & 1.600 & 1.0609 & 7.608 & 268 . & 2040 .
+ 990506 & 1.300 & 0.9280 & 4.556 & 194 . &
883.9 + 990510 & 1.619 & 1.0687 & 7.834 & 22.6 & 177.0 + 990705 & 0.840 & 0.6762 & 1.548 & 93.0 & 144.0 + 990712 & 0.433 & 0.3888 & 0.310 & 6.50 & 2.015 + 991208 & 0.706 & 0.5891 & 1.010 & 100 . & 101.0 + 991216 & 1.020 & 0.7827 & 2.500 & 194 . &
485.0 + 000131 & 4.500 & 1.7496 & 92.60 & 41.8 & 3871 .
+ 000210 & 0.846 & 0.6801 & 1.577 & 61.0 & 96.20 + 000301c & 2.034 & 1.2208 & 13.72 & 4.10 & 56.25 + 000418 & 1.119 & 0.8367 & 3.143 & 20.0 & 62.86 + 000911 & 1.059 & 0.8041 & 2.740 & 230 . & 630.2 + 000926 & 2.037 & 1.2218 & 13.77 & 6.20 & 85.37 + 010222 & 1.477 & 1.0089 & 6.244 & 120 . & 749.3 + 010921 & 0.451 & 0.4029 & 0.342 & 15.4 & 5.267 + 011121 & 0.362 & 0.3312 & 0.203 & 24.0 & 4.872 + 011211 & 2.140 & 1.2552 & 15.53 & 5.00 & 77.65 + 020405 & 0.690 & 0.5782 & 0.955 & 38.0 & 36.29 + 020813 & 1.254 & 0.9057 & 4.167 & 38.0 & 158.3 + 021004 & 2.332 & 1.3135 & 19.15 & 3.20 & 61.28 + 021211 & 1.006 & 0.7748 & 2.416 & 1.00 & 2.416 | the microlensing interpretation of the optical afterglow of grb 000301c seems naively surprising , since a simple estimate of the stellar microlensing rate gives less than one in four hundred for a flat @xmath0 cosmology , whereas one event was seen in about thirty afterglows .
considering baryonic machos making up half of the baryons in the universe , the microlensing probability per burst can be roughly @xmath1 for a grb at redshift @xmath2 .
we explore two effects that may enhance the probability of observing microlensed gamma - ray burst afterglows : binary lenses and _ double _ magnification bias .
we find that the consideration of binary lenses can increase the rate only at the @xmath3 level . on the other hand , because gamma - ray bursts for which afterglow observations exist are typically selected based on fluxes at widely separated wavebands which are not necessarily well correlated ( e.g. localization in x - ray , afterglow in optical / infrared ) , magnification bias can operate at an enhanced level compared to the usual single bias case . using a simple model for the selection process in two bands ,
we compute the enhancement to microlensing rate due to magnification bias in two cases : perfect correlation and complete independence of the flux in the two bands .
we find that existing estimates of the slope of the luminosity function of gamma - ray bursts , while as yet quite uncertain , point to enhancement factors of more than three above the simple estimates of the microlensing rate .
we find that the probability to observe at least one microlensing event in the sample of 27 measured afterglows can be @xmath4 for stellar lenses , or as much as @xmath5 for baryonic machos .
we note that the probability to observe at least one event over the available sample of afterglows is significant only if a large fraction of the baryons in the universe are condensed in stellar mass objects . |
the recent discovery of high redshift quasars with @xmath1 ( fan et al .
2001 ) implies that the formation of supermassive black holes proceeded in less than @xmath2 yr . also , the recent compilation of the kinematical data of galactic centers in both active and inactive galaxies has shown that a central massive dark object ( mdo ) , which is the nomenclature for a black hole ( bh ) candidate , correlates with the properties of galactic bulges .
the demography of mdos have revealed the following relations : \1 ) the bh mass exhibits a linear relation to the bulge mass with the ratio of @xmath3 as a median value ( kormendy & richstone 1995 ; richstone et al .
1998 ; magorrian et al . 1998 ; gebhardt et al .
2000a ; ferrarese & merritt 2000 ; merritt & ferrarese 2001a ; mclure & dunlop 2002 ) .
\2 ) the bh mass correlates with the velocity dispersion of bulge stars with a power - law relation as @xmath4 , @xmath5 ( gebhardt et al .
2000b ) , 4.72 ( ferrarese & merritt 2000 ; merritt & ferrarese 2001a , b ) , or @xmath6 ( tremaine et al . 2002 ) .
\6 ) the @xmath7 in seyfert 1 galaxies is not well converged , which may be considerably smaller than 0.001 ( wandel 1999 ; gebhardt et al .
2000a ) or similar to that for ellipticals ( mclure & dunlop 2002 ) , while the bh mass - to - velocity dispersion relation in seyfert 1 galaxies seems to hold in a similar way to elliptical galaxies ( gebhardt et al .
2000a ; nelson 2000 ) .
on the other hand , the x - ray emission ( brandt et al . 1997 ) or pa@xmath8 lines ( veilleux , sanders , & kim 1999 ) intrinsic for active nuclei have been detected in roughly one forth of ultraluminous infrared galaxies ( ulirgs ) .
furhtermore , it has been revealed that qso host galaxies are mostly luminous and well evolved early - type galaxies ( mcleod & rieke 1995 ; bahcall et al .
1997 ; hooper , impey , & foltz 1997 ; mcleod , rieke , & storrie - lombardi 1999 ; brotherton et al .
1999 ; kirhakos et al . 1999 ; mclure et al . 1999 ; mclure , dunlop , & kukula 2000 ) .
all these findings on qso hosts and supermassive bhs imply that the formation of a qso , a bulge , and a supermassive bh is mutually related .
some theoretical models have been hitherto considered to explain the bh - to - bulge correlations ( silk & rees 1998 ; adams , graff , & richstone 2001 ; ostriker 2000 ) .
but , little has been elucidated regarding the physics on the angular momentum transfer which is requisite for bh formation , since the rotation barrier by the tidal spin up in a growing density fluctuation is given by @xmath9 in units of the schwarzshild radius @xmath10 , where @xmath11 is the baryonic mass , @xmath12 is the cosmological redshift , and @xmath13 is the spin parameter which provides the ratio of circular velocity to velocity dispersion of dark matter ( umemura 2001 ) . furthermore , required mechanisms for bh formation must work effectively in a spheroidal system like a bulge .
the @xmath8-viscosity or non - axisymmetric gravitational instabilities would effectively transfer angular momentum once a disk - like system forms , but they are not likely to work in a spheroidal system . in this paper , as a potential mechanism that works in a spheroidal system , the relativistic drag force by the radiation from bulge stars is considered . as a result , the bh - to - bulge ratio is basically determined by the energy conversion efficiency for nuclear fusion of hydrogen to helium , @xmath0 .
also , in relation to bh growth , a new scenario for qso formation is addressed .
a radiation hydrodynamic model which could account for the putative correlations between smbhs and bulges is recently proposed by umemura ( 2001 ) , where the relativistic drag force by the radiation from bulge stars is considered .
the radiation drag can extract angular momentum from gas and allow the gas to accrete onto the center ( umemura , fukue , & mineshige 1997 , 1998 ; fukue , umemura , & mineshige 1997 ) .
for the total luminosity @xmath14 of a uniform bulge , the radiation energy density is given by @xmath15 where @xmath16 is the light speed and @xmath17 is the radius of the bulge .
then , the angular momentum loss rate by the radiation drag is given by @xmath18 where @xmath19 is the total angular momentum of gaseous component and @xmath20 is the mass extinction coefficient which is given by @xmath21 with dust absorption coefficient @xmath22 and gas density @xmath23 .
therefore , in an optically - thin regime , @xmath24 where @xmath25 is the total optical depth of the system and @xmath26 is the total mass of gas . in an optically - thick regime , the radiation drag efficiency
is saturated due to the conservation of the photon number ( tsuribe & umemura 1997 ) .
thus , an expression of the angular momentum loss rate which includes both regimes is given by @xmath27 then , the mass accretion rate is estimated to be @xmath28 in an optically - thick regime , this gives simply @xmath29 which is numerically @xmath30 this rate is comparable to the eddington mass accretion rate for a black hole with @xmath31 , that is , @xmath32 where @xmath33 is the energy conversion efficiency .
unless otherwise stated , @xmath34 for an extreme kerr black hole is assumed .
the timescale of radiation drag - induced mass accretion is @xmath35 where @xmath36 and @xmath37 is the metallicity of gas .
it is noted that the gas which is more abundant in metals accretes in a shorter timescale , because the extinction is predominantly given by the dust opacity . for the moment , an optically - thick stage is considered . due to the mass accretion induced by the radiation drag , a massive dark object ( mdo ) forms at the center of bulge .
then , the mass of mdo is estimated in terms of @xmath38 in practice , it is likely that optically - thin surface layers are stripped from optically - thick clumpy clouds by the radiation drag , and the stripped gas losing angular momentum accretes onto the center ( kawakatu & umemura 2002 ) .
next , we employ a simplest analytic model for bulge evolution .
the star formation rate is assumed to be a schmidt law , @xmath39 .
if we invoke the instantaneous recycling approximation , the star formation rate is given by @xmath40 where @xmath8 is the net efficiency of the conversion into stars after subtracting the mass loss .
the radiation energy emitted by a main sequence star is @xmath41 to the rest mass energy of the star , @xmath42 , where @xmath43 is the energy conversion efficiency of nuclear fusion from hydrogen to helium , which is 0.007 .
thus , the luminosity of the bulge is estimated to be @xmath44 by substituting this in ( [ m - mdo ] ) , @xmath45 the term @xmath46 represents just the stellar mass in the system which is @xmath47 observationally . as a consequence ,
the mdo mass to bulge mass ratio is given by @xmath48 where @xmath49 .
it should be noted that the final mass is basically determined by @xmath43 .
also , the proportionality of smbh mass to bulge mass is a natural consequence in the present mechanism . in the above , the prediction for the mdo - to - bulge mass relation is made in the context of radiation drag - induced mass accretion .
however , the mdo itself is not a supermassive black hole , because the radiation drag is not likely to remove the angular momentum thoroughly , but some residual angular momentum will terminate the radial contraction .
hence , we should consider the further collapse of the mdo through other physical mechanisms . in the mdo , the viscosity is expected to work effectively because the timescale for viscous accretion , @xmath50 , is shortened by the radiation drag , where @xmath51 is the specific angular momentum , @xmath52 is the viscous parameter , and @xmath53 is the sound velocity .
thus , the mdo is likely to be a massive self - gravitating viscous disk . for a self - gravitating viscous disk ,
some self - similar solutions are known to give an inside - out disk collapse ( mineshige & umemura 1996 ; mineshige & umemura 1997 ; tsuribe 1999 ) .
in particular , tsuribe ( 1999 ) provided a convenient formula for the inside - out mass accretion rate , @xmath54 where @xmath53 is the sound speed and @xmath55 is the toomre s q which is @xmath56 for the epicycle frequency @xmath57 and the surface density @xmath58 .
tsuribe ( 1999 ) has found a solution of stable accretion with @xmath59 .
the critical accretion rate is given by @xmath60 this rate is again comparable to the eddington mass accretion rate for a black hole with @xmath31 if @xmath61k [ see ( [ m - dot - edd ] ) ] . through this inside - out collapse
, a central core grows .
the core is expected to be a rigidly rotating supermassive star because the viscous transfer of angular momentum works to smear out any differential rotation in a self - gravitating system . the equilibrium configuration and the stability of a rigidly rotating supermassive star has been scrutinized by baumgarte & shapiro ( 1999 ) .
they found that a rotating supermassive star becomes unstable for @xmath62 .
as for the dynamical collapse of a rotating supermassive star , saijo et al .
( 2002 ) performed post - newtonian calculations and found that if a rotating supermassive star is in rigid rotation , it can collapse towards the horizon scale without undergoing bar - mode instability .
the final stage of the collapse was investigated by shibata & shapiro ( 2002 ) with a full general relativistic approach and the emergence of an apparent horizon of a kerr black hole was shown .
the resultant spin parameter of the black hole is around 0.75 .
hence , an mdo formed by radiation drag - induced mass accretion could evolve into a supermassive kerr black hole through the inside - out viscous collapse and the general relativistic instability .
the present picture for the formation of a supermassive black hole is summarized in figure [ fig1 ] .
if the bh accretion causes the nuclear activity , the further radiative mass accretion can be induced by the nuclear luminosity @xmath63 .
this feedback works positively to grow the smbh .
consequently , it is predicted that the bh mass to bulge mass ratio becomes @xmath64 this is just comparable to the observed ratio . if the bh mass is determined by the present mechanism , the bh mass to velocity dispersion relation is naturally understood in the context of a cold dark matter ( cdm ) cosmology . supposing the bulge is a virialized system , then @xmath66 and @xmath67 where @xmath68 is the radius at the maximum expansion epoch @xmath69 .
if a cdm cosmology is assumed , then @xmath70 where @xmath71 around @xmath72 , almost regardless of the cosmological parameters ( bunn & white 1997 ) .
combining all these relations , we find @xmath73 with @xmath74 , which is 4 for @xmath75 .
this result is just corresponding to the inferred relation between the bh mass and the stellar velocity dispersion .
the radiation drag efficiency could be strongly subject to the effect of geometrical dilution ( umemura , fukue , & mineshige 1998 ) .
if the system is spherical , the emitted photons are effectively consumed within the system , whereas a large fraction of photons can escape from a disk - like system and also the drag efficiency is considerably reduced across the optically - thick disk .
this may be the reason why @xmath7 is observed to be significantly smaller than 0.001 in disk systems . in the present model ,
the theoretical upper limit of @xmath76 is @xmath0 .
the low mass end of the bh hierarchy is suggested by the recent observations on massive bhs in globular clusters , e.g. , @xmath77 in g1 ( gebhardt , rich , & ho 2002 ) and @xmath78 in m15 ( van der marel et al . 2002 ; gessen et al . 2002 ) .
it is found that @xmath65 relation still holds in such small spheroidal systems .
the present radiation hydrodynamic mechanism is expected to work in forming a bh even in a small spheroidal system , if the system retains interstellar medium ( ism ) sufficient to become optically thick .
although in globular clusters a few supernovae can easily expel the ism in an early evolutionary phase , the mass loss from low mass stars in later phases may accumulate a sufficient ism . on the other hand
, m33 appears not to have a massive bh with an upper limit of @xmath79 ( gebhardt et al .
an intriguing difference is that m33 contains a significant population of stars younger than a few gyrs ( e.g. , oconnell 1983 ) , in contrast to other globular clusters which possess massive bhs .
thus , m33 might not have been able to retain a sufficient ism owing to successive supernova explosions . in other words ,
the bh mass in globular clusters may sensitively reflect its star formation history .
as for the high mass end of the bh hierarchy , whether hypermassive bhs in clusters of galaxies exist is a crucial test . in galaxy clusters , it is not likely that optically - thick intracluster medium is supplied by galactic winds or stripping of ism .
hence , in the present scenario , hypermassive bhs will not form in the centers of galaxy clusters . therefore ,
the bh hierarchy is predicted to be cut off at the maximum size in galaxies , @xmath80 . and
@xmath63 are the bulge luminosity and the black hole accretion luminosity , respectively .
@xmath81 is the mass of smbh .
the ordinary qso phase is preceded by a host luminosity - dominant `` proto - qso phase''.,width=453 ] here , we construct a picture of qso formation based on the present scenario for smbh formation .
if the mass accretion driven by the viscosity on to the bh horizon is determined by an order of eddington rate , the bh mass grows according to @xmath82 where @xmath83 is the ratio of bh accretion rate to the eddington rate and @xmath84 is the eddington timescale , @xmath85yr .
@xmath86 is the mass of a seed bh with @xmath87 ( shibata & shapiro 2002 ) . in order to incorporate the chemical evolution of host galaxy
, we use an evolutionary spectral synthesis code pegase ( fioc & rocca - volmerange 1997 ) , and also employ a galactic wind model with the wind epoch of @xmath88yr to match the present - day color - magnitude relation ( arimoto & yoshii 1987 ) .
the system is assumed to change from optically - thick to optically - thin phase at @xmath89 .
based on the present coevolution model , the evolution of bulge luminosity ( @xmath90 ) , the agn luminosity ( @xmath63 ) , and the mass of smbh ( @xmath81 ) are shown in figure [ fig2 ] , assuming the constant eddington ratio ( @xmath91 ) .
the @xmath81 reaches @xmath92 at a time @xmath93 , so that the bh fraction becomes @xmath94 , which is just comparable to the observed ratio . in the optically - thin phase after the galactic wind ( @xmath95 )
, @xmath81 still continues to grow until @xmath93 and therefore the agn brightens with time .
after @xmath63 exhibits a peak at @xmath93 , it fades out abruptly due to exhausting the fuel of the mdo .
this fading nucleus could be a low luminosity agn ( llagn ) .
it is found that the era of @xmath96 can be divided into two phases with a transition time @xmath97 when @xmath98 ; the earlier phase is the host luminosity - dominant phase and the later phase is the agn luminosity - dominant phase .
the lifetimes of both phases are comparable to each other , which is about @xmath99yr .
the agn - dominant phase is likely to correspond to ordinary qsos , while the host - dominant phase is obviously different from observed qsos so far .
we define this phase as `` a proto - qso '' ( kawakatu & umemura 2003 ) .
the observable properties of proto - qsos are predicted as follows : ( 1 ) the width of broad emission line is narrower , which is less than @xmath100km / s . ( 2 ) @xmath7 rapidly increases from @xmath101 to @xmath102 in @xmath103 years .
( 3 ) the colors of @xmath104 at rest bands and @xmath105 at observed bands are about 0.5 magnitude bluer than those of qsos .
( 4 ) in both proto - qso and qso phases , the metallicity of gas in galactic nuclei is @xmath106 , and that of stars weighted by the host luminosity is @xmath107 , which are consistent with the observations for qsos and the elliptical galaxies . ( 5 ) a massive dusty disk ( @xmath108 ) surrounds a massive bh , and it may obscure the nucleus in the edge - on view to form a type 2 nucleus .
the predicted properties of proto - qsos are quite similar to those of radio galaxies at high redshifts .
thus , high-@xmath12 radio galaxies are a key candidate for proto - qsos .
the proto - qso phase is preceded by a bright and optically thick phase , which may correspond to a ultraluminous infrared galaxy ( ulirg ) phase .
also , the precursor of ulirgs is an optically - thin and very luminous phase with the lifetime of @xmath109 years .
this may correspond to the assembly phase of lbgs or ly@xmath8 emitters . in this phase ,
the metallicity is subsolar ( @xmath110 ) , and the hard x - ray luminosity is @xmath111 if @xmath112 .
such a coevolution scenario of a smbh and the host is summarized in figure [ fig3 ] .
the smbh could form through the relativistic radiation - hydrodynamic processes from galactic scale to horizon scale .
the resultant mass of a smbh is in proportion to the bulge mass , and the smbh mass fraction is determined by the nuclear energy conversion efficiency from hydrogen to helium , @xmath0 , which is the theoretical upper limit in the present mechanism .
massive bhs can form also in globular clusters , but hypermassive bhs are unlikely to form in the centers of galaxy clusters . as for qso formation , it is predicted that a host luminosity - dominant `` proto - qso phase '' exists before the qso phase .
the proto - qso is a phase of growing bh with broad emission line width of less than @xmath100km / s .
the proto - qso phase is preceded by an optically - thick ultraluminous infrared galaxy ( ulirg ) phase . before the ulirg phase
, an optically thin starburst phase should exist .
adams , f. c. , graff , d. s. , & richstone , d. o. 2001 , apj , 551 , l31 arimoto , n. , & yoshii , y. 1986 , a&a , 164 , 260 bahcall , j. n. , et al .
1997 , apj , 479 , 642 baumgarte , t. w. , & shapiro , s. l. 1999 , apj , 526 , 941 brandt , w. n. , et al .
1997 , mnras , 290 , 617 brotherton , m. s. , et al . 1999 ,
apj , 520 , l87 bunn , e. f. , & white , m. 1997 , apj , 480 , 6 fan , x. et al .
2001 , aj , 122 , 2833 ferrarese , l. , & merritt , d. 2000 , apj , 539 , l9 fioc , m. , & rocca - volmerrange , b. 1997 , a&a , 326 , 950 fukue , j. , umemura , m. , & mineshige , s. 1997 , pasj , 49 , 673 gebhardt , k. , et al .
2000a , apj , 539 , l13 gebhardt , k. , et al .
2000b , apj , 543 , l5 gebhardt , k. , et al .
2001 , aj , 122 , 2469 gebhardt , k. , rich , r. m. , & ho , l. c. 2002 , apj , 578 , l41 gessen et al .
2002 , aj , 124 , 3270 hooper , e. j. , impey , c. d. , & foltz , c. b. 1997 , apj , 480 , l95 kawakatu , n. & umemura , m. 2002 , mnras , 329 , 572 kawakatu , n. & umemura , m. 2003 , apj , in press kirhakos , s. , et al . 1999 ,
apj , 520 , 67 kormendy , j. , & richstone , d. 1995 , araa , 33 , 581 laor , a. 1998 , apj , 505 , l83 magorrian , j. , et al .
1998 , aj , 115 , 2285 mcleod , k. k. , & rieke , g. h. 1995 , apj , 454 , l77 mclure , r. j. , & dunlop , j. s. 2002 , mnras , 331 , 795 mclure , r. j. , dunlop , j. s. , & kukula , m. j. 2000 , mnras , 318 , 693 mclure , r. j. , et al .
1999 , mnras , 308 , 377 merrifield , m. r. , forbes , duncan a. , & terlevich , a. i. 2000 , mnras , 313 , l29 merritt , d. , & ferrarese , l. 2001a , mnras , 320 , l30 merritt , d. , & ferrarese , l. 2001b , apj , 547 , 140 mineshige , s. & umemura , m. 1996 , apj , 469 , l49 mineshige , s. & umemura , m. 1997 , apj , 480 , 167 nelson , c. h. 2000 , apj , 544 , l91 oconnell , r. w. 1983 , apj , 267 , 80 ostriker , j. p. 2000 , phys .
lett . , 84 , 5258 richstone , d. , et al .
1998 , nature , 395a , 14 saijyo , m. , baumgarte , t. w. , shapiro , s. l. , & shibata , m. 2002 , apj , in press salucci , p. , et al .
2000 , mnras , 317 , 488 sarzi , m. , et al .
2001 , apj , 550 , 65 shibata , m. & s. shapiro , 2002 , apj , 572 , l39 shields , g. a. et al .
apj , in press ( astro - ph/0210050 ) silk , j. , & rees , m. 1998 , a&a , 331 , l1 tremaine , t. et al .
2002 , apj , 574 , 740 tsuribe , t. 1999 , apj , 527 , 102 tsuribe , t. , & umemura , m. 1997 , apj , 486 , 48 umemura , m. 2001 , apj , 560 , l29 umemura , m. , fukue , j. , & mineshige , s. 1997 , apj , 479 , l97 umemura , m. , fukue , j. , & mineshige , s. 1998 , mnras , 299 , 1123 van der marel , r. p. et al .
2002 , aj , 124 , 3255 veilleux , s. , sanders , d. b. , & kim , d .-
1999 , apj , 522 , 139 wandel , a. 1999 , apj , 519 , l39 | a novel mechanism to build up a supermassive black hole ( smbh ) is proposed . here , relativistic radiation - hydrodynamic processes are considered from a galactic scale to a horizon scale . in the present scenario ,
the mass of a smbh is predicted to be in proportion to the bulge mass , and it turns out that the smbh - to - bulge mass ratio is basically determined by the nuclear energy conversion efficiency from hydrogen to helium , @xmath0 .
the bh hierarchy is also addressed in relation to the bh formation in globular clusters .
futhermore , a new picture for qso formation is proposed based on the present smbh formation scenario .
it is predicted that a host luminosity - dominant `` proto - qso phase '' exists before the qso phase , and the proto - qso phase is preceded by an optically - thick ultraluminous infrared galaxy ( ulirg ) phase .
[ 1996/06/01 ] a&a |
understanding , modeling and predicting the volatility of financial time series has been extensively researched for more than 30 years and the interest in the subject is far from decreasing .
volatility prediction has a very wide range of applications in finance , for example , in portfolio optimization , risk management , asset allocation , asset pricing , etc .
the two most popular approaches to model volatility are based on the autoregressive conditional heteroscedasticity ( arch ) type and stochastic volatility ( sv ) type models .
the seminal paper of @xcite proposed the primary arch model while @xcite generalized the purely autoregressive arch into an arma - type model , called the generalized autoregressive conditional heteroscedasticity ( garch ) model .
since then , there has been a very large amount of research on the topic , stretching to various model extensions and generalizations .
meanwhile , the researchers have been addressing two important topics : looking for the best specification for the errors and selecting the most efficient approach for inference and prediction .
besides selecting the best model for the data , distributional assumptions for the returns are equally important .
it is well known , that every prediction , in order to be useful , has to come with a certain precision measurement . in this way
the agent can know the risk she is facing , i.e. uncertainty .
distributional assumptions permit to quantify this uncertainty about the future .
traditionally , the errors have been assumed to be gaussian , however , it has been widely acknowledged that financial returns display fat tails and are not conditionally gaussian .
therefore , it is common to assume a student - t distribution , see @xcite , @xcite and @xcite , among others .
however , the assumption of gaussian or student - t distributions is rather restrictive .
an alternative approach is to use a mixture of distributions , which can approximate arbitrarily any distribution given a sufficient number of mixture components .
a mixture of two normals was used by @xcite , @xcite and @xcite , among others .
these authors have shown that the models with the mixture distribution for the errors outperformed the gaussian one and do not require additional restrictions on the degrees of freedom parameter as the student - t one . as for the inference and prediction ,
the bayesian approach is especially well - suited for garch models and provides some advantages compared to classical estimation techniques , as outlined by @xcite .
firstly , the positivity constraints on the parameters to ensure positive variance , may encumber some optimization procedures . in the bayesian setting
, constraints on the model parameters can be incorporated via priors . secondly , in most of the cases we are more interested not in the model parameters directly , but in some non - linear functions of them . in the maximum likelihood ( ml )
setting , it is quite troublesome to perform inference on such quantities , while in the bayesian setting it is usually straightforward to obtain the posterior distribution of any non - linear function of the model parameters .
furthermore , in the classical approach , models are usually compared by any other means than the likelihood . in the bayesian setting , marginal likelihoods and bayes factors allow for consistent comparison of non - nested models while incorporating occam s razor for parsimony .
also , bayesian estimation provides reliable results even for finite samples . finally , @xcite add that the ml approach presents some limitations when the errors are heavy tailed , also the convergence rate is slow and the estimators may not be asymptotically gaussian .
this survey reviews the existing bayesian inference methods for univariate and multivariate garch models while having in mind their error specifications .
the main emphasis of the paper is on the recent development of an alternative inference approach for these models using bayesian non - parametrics .
the classical parametric modeling , relying on a finite number of parameters , although so widely used , has some certain drawbacks .
since the number of parameters for any model is fixed , one can encounter underfitting or overfitting , which arises from the misfit between the data available and the parameters needed to estimate . then , in order to avoid assuming wrong parametric distributions , which may lead to inconsistent estimators , it is better to consider a semi- or non - parametric approach . bayesian non - parametrics may lead to less constrained models than classical parametric bayesian statistics and provide an adequate description of the data , especially when the conditional return distribution is far away from gaussian .
up to our knowledge , there has been a very few papers using bayesian non - parametrics for garch models .
these are @xcite for univariate garch , @xcite and @xcite for mgarch .
all of them have considered infinite mixtures of gaussian distributions with a dirichlet process ( dp ) prior over the mixing distribution , which results into dp mixture ( dpm ) models .
this approach so far proves to be the most popular bayesian non - parametric modeling procedure .
the results over the papers have been consistent : the bayesian non - parametric approach leads to more flexible models and is better in explaining heavy - tailed return distributions , which parametric models can not fully capture .
the outline of this survey is as follows .
section [ section : univ_garch ] shortly introduces univariate garch models and different inference and prediction methods .
section [ section : mult_garch ] overviews the existing models for multivariate garch and different inference and prediction approaches .
section [ section : non_param ] introduces the bayesian non - parametric modeling approach and reviews the limited literature of this area in time - varying volatility models .
section [ section : illustration ] presents a real data application .
finally , section [ section : conclusions ] concludes .
as mentioned before , the two most popular approaches to model volatility are garch type and sv type models . in this survey
we focus on garch models , therefore , sv models will not be included thereafter .
also , we are not going to enter into the technical details of the bayesian algorithms and refer to @xcite for a more detailed description of bayesian techniques .
the general structure of an asset return series modeled by a garch - type models can be written as : @xmath0 where @xmath1 $ ] is the conditional mean given @xmath2 , the information up to time @xmath3 , @xmath4 is the mean corrected returns of the asset at time @xmath5 , @xmath6 $ ] is the conditional variance given @xmath2 and @xmath7 is the standard white noise shock .
there are several ways to model the conditional mean , @xmath8 .
the usual assumptions are to consider that the mean is either zero , equal to a constant ( @xmath9 ) , or follows an arma(@xmath10,@xmath11 ) process .
however , sometimes the mean is also modeled as a function of the variance , say @xmath12 , which leads to the garch - in - mean models . on the other hand , the conditional variance , @xmath13 ,
is usually modeled using the garch - family models . in the basic garch model the conditional variance of the returns depends on a sum of three parts : a constant variance as the long - run average , a linear combination of the past conditional variances and a linear combination of the past mean squared returns . for instance , in the garch(1,1 ) model , the conditional variance at time @xmath14 is given by @xmath15 , for @xmath16 .
there are some restrictions which have to be imposed such as @xmath17 , @xmath18 for positive variance , and @xmath19 for the covariance stationarity .
@xcite proposed the exponential garch ( egarch ) model that acknowledges the existence of asymmetry in the volatility response to the changes in the returns , sometimes also called the `` leverage effect '' , introduced by @xcite .
negative shocks to the returns have a stronger effect on volatility than positive .
other arch extensions that try to incorporate the leverage effect are the gjr model by @xcite and the tgarch of @xcite , among many others . as @xcite puts it
, `` there is now an alphabet soup '' of arch family models , such as aarch , aparch , figarch , starch etc , which try to incorporate such return features as fat tails , volatility clustering and volatility asymmetry .
papers by @xcite , @xcite , @xcite , @xcite provide extensive reviews of the existing arch - type models .
@xcite review arch type models , discuss their extensions , estimation and testing , also numerous applications .
also , one can find an explicit review with examples and applications concerning garch - family models in @xcite and chapter 1 in @xcite .
the main estimation approach for garch - family models is the classical maximum likelihood method .
however , recently there has been a rapid development of bayesian estimation techniques , which offer some advantages compared to the frequentist approach as already discussed in the introduction .
in addition , in the empirical finance setting , the frequentist approach presents an uncertainty problem .
for instance , optimal allocation is greatly affected by the parameter uncertainty , which has been recognized in a number of papers , see @xcite and @xcite , among others .
these authors conclude that in the frequentist setting the estimated parameter values are considered to be the true ones , therefore , the optimal portfolio weights tend to inherit this estimation error
. however , instead of solving the optimization problem on the basis of the choice of unique parameter values , the investor can choose the bayesian approach , because it accounts for parameter uncertainty , as seen in @xcite and @xcite , for example .
a number of papers in this field have explored different bayesian procedures for inference and prediction and different approaches to modeling the fat - tailed errors and/or asymmetric volatility .
the recent development of modern bayesian computational methods , based on monte carlo approximations and mcmc methods have facilitated the usage of bayesian techniques , see e.g. @xcite .
the standard gibbs sampling procedure does not make the list because it can not be used due to the recursive nature of the conditional variance : the conditional posterior distributions of the model parameters are not of a simple form .
one of the alternatives is the _ griddy - gibbs _ sampler as in @xcite .
they discuss that previously used importance sampling and metropolis algorithms have certain drawbacks , such as that they require a careful choice of a good approximation of the posterior density .
the authors propose a griddy - gibbs sampler which explores analytical properties of the posterior density as much as possible . in this paper
the garch model has student - t errors , which allows for fat tails .
the authors choose to use flat ( uniform ) priors on parameters @xmath20 with whatever region is needed to ensure the positivity of variance , however , the flat prior for the degrees of freedom can not be used , because then the posterior density is not integrable .
instead , they choose a half - right side of cauchy .
the posteriors of the parameters were found to be skewed , which is a disadvantage for the commonly used gaussian approximation . on the other hand , @xcite modeled the errors of a garch model with a mixture of two gaussian distributions .
the advantage of this approach compared to that of student - t errors , is that if the number of the degrees of freedom is very small ( less than 5 ) , some moments may not exist .
the authors have chosen flat priors for all the parameters , and discovered that there is little sensitivity to the change in the prior distributions ( from uniform to beta ) , unlike in @xcite , where the sensitivity for the prior choice for the degrees of freedom is high .
more articles using a griddy - gibbs sampling approach are by @xcite , who have modeled asymmetric volatility with gaussian innovations and have used uniform priors for all the parameters , and by @xcite , who explored an asymmetric garch model with student - t errors .
another mcmc algorithm used in estimating garch model parameters , is the _ metropolis - hastings _ ( mh ) method , which samples from a candidate density and then accepts or rejects the draws depending on a certain acceptance probability .
@xcite modeled the errors as gaussian distributed with zero mean and unit variance while the priors are chosen as gaussian and a mh algorithm is used to draw samples from the joint posterior distribution .
the author has carried out a comparative analysis between ml and bayesian approaches , finding , as in other papers , that some posterior distributions of the parameters were skewed , thus warning against the abusive use of the gaussian approximation .
also , @xcite has performed a sensitivity analysis of the prior means and scale parameters and concluded that the initial priors in this case are vague enough .
this approach has been also used by @xcite , @xcite and @xcite , among others .
a special case of the mh method is the random walk metropolis - hastings ( rwmh ) where the proposal draws are generated by randomly perturbing the current value using a spherically symmetric distribution .
a usual choice is to generate candidate values from a gaussian distribution where the mean is the previous value of the parameter and the variance can be calibrated to achieve the desired acceptance probability .
this procedure is repeated at each mcmc iteration .
@xcite have also carried out a comparison of estimation approaches , griddy - gibbs , rwmh and ml . apparently , rwmh has difficulties in exploring the tails of the posterior distributions and ml estimates may be rather different for those parameters where posterior distributions are skewed . in order to select one of the algorithms
, one might consider some criteria , such as fast convergence for example .
@xcite numerically compares some of these approaches in the context of garch .
the griddy - gibbs method is capable in handling the shape of the posterior by using smaller mcmc outputs comparing with other methods , also , it is flexible regarding parametric specification of a model .
however , it can require a lot of computational time .
this author also investigates mh , adaptive rejection metropolis sampling ( arms ) , proposed by @xcite , and acceptance - rejection mh algorithms ( armh ) , proposed by @xcite . for more in detail about each method in garch models see @xcite and @xcite , among others . using simulated data , @xcite calculated geometric averages of inefficiency factors for each method .
inefficiency factor is just an inverse of @xcite efficiency factor . according to this ,
the armh algorithm performed the best .
also , computational time was taken into consideration , where armh clearly outperformed mh and arms , while griddy - gibbs stayed just a bit behind .
the author observes that even though the armh method showed the best results , the posterior densities for each parameter did not quite explore the tails of the distributions , as desired . in this case
griddy - gibbs performs better ; also , it requires less draws than armh .
@xcite investigate one more convergence criteria , proposed by @xcite , which is based on cumulative sum ( cumsum ) statistics .
it basically shows that if mcmc is converging , the graph of a certain cumsum statistic against time should approach zero .
their employed griddy - gibbs algorithm converged in all four parameters quite fast .
then , the authors explored the advantages and disadvantages of alternative approaches : the importance sampling and the mh algorithm . considering importance sampling , one of the main disadvantages , as mentioned before ,
is to find a good approximation of the posterior density ( importance function ) .
also , comparing with griddy - gibbs algorithm , the importance sampling requires much more draws to get smooth graphs of the marginal densities . for the mh algorithm , same as in importance sampling ,
a good approximation needs to be found .
also , compared to griddy - gibbs , the mh algorithm did not fully explore the tails of the distribution , unless for a very big number of draws . another important aspect of the bayesian approach , as commented before , are the advantages in model selection compared to the classical methods .
@xcite reviews some bayesian model selection methods using mcmc for garch - type models , which allow for the estimation of either marginal model likelihoods , bayes factors or posterior model probabilities .
these are compared to the classical model selection criteria showing that the bayesian approach clearly considers model complexity in a more unbiased way .
also , @xcite includes a revision of bayesian selection methods for asymmetric garch models , such as the gjr - garch and threshold garch .
they show how using the bayesian approach it is possible to compare complex and non - nested models to choose for example between garch and stochastic volatility models , between symmetric or asymmetric garch models or to determine the number of regimes in threshold processes , among others .
an alternative approach to the previous parametric specifications is the use of bayesian non - parametric methods , that allow to model the errors as an infinite mixture of normals , as seen in the paper by @xcite .
the bayesian non - parametric approach for time - varying volatility models will be discussed in detail in section [ section : non_param ] .
to sum up , considering the amount of articles published quite recently regarding the topic of estimating univariate garch models using mcmc methods indicates still growing interest in the area .
although numerous garch - family models have been investigated using different mcmc algorithms , there are still a lot of areas that need further research and development .
returns and volatilities depend on each other , so multivariate analysis is a more natural and useful approach . the starting point of multivariate volatility models is a univariate garch , thus the most simple mgarch models can be viewed as direct generalizations of their univariate counterparts . consider a multivariate return series @xmath21 of size @xmath22 .
then @xmath23 where @xmath24 $ ] , @xmath25 are mean - corrected returns , @xmath7 is a random vector , such that @xmath26=0}$ ] and @xmath27=i_k}$ ] and @xmath28 is a positive definite matrix of dimensions @xmath29 , such that @xmath30 is the conditional covariance matrix of @xmath31 , i.e. , @xmath32=h_{t}^{1/2 } \tx{cov}[\epsilon_t ] ( h_{t}^{1/2})'=h_t$ ] .
there is a wide range of mgarch models , where most of them differ in specifying @xmath30 . in the rest of this section we will review the most popular and widely used and the different bayesian approaches to make inference and prediction . for general reviews on mgarch models ,
see @xcite , @xcite and @xcite ( chapter 10 ) , among others .
regarding inference , one can also consider the same arguments provided in the univariate garch case above .
maximum likelihood estimation for mgarch models can be obtained by using numerical optimization algorithms , such as fisher scoring and newton - raphson . @xcite
have estimated several bivariate arch and garch models and found that some classical estimates of the parameters were quite different from their bayesian counterparts .
this was due to the non - normality of the parameters .
thus , the authors suggest careful interpretation of the classical estimation approach . also , @xcite found it difficult to evaluate the classical estimates under the stationarity conditions , and consequently the resulting parameters , evaluated ignoring the stationarity constraints , produced non - stationary estimates .
these difficulties can be overcome using the bayesian approach .
the vec model was proposed by @xcite , where every conditional variance and covariance ( elements of the @xmath30 matrix ) is a function of all lagged conditional variances and covariances , as well as lagged squared mean - corrected returns and cross - products of returns .
using this unrestricted vec formulation , the number of parameters increases dramatically .
for example , if @xmath33 , the number of parameters to estimate will be 78 , and if @xmath34 , the number of parameters increases to 210 , see @xcite for the explicit formula for the number of parameters in vec models . to overcome this difficulty , @xcite simplified the vec model by proposing a diagonal vec model , or dvec , as follows : @xmath35 where @xmath36 indicates the hadamard product , @xmath37 , @xmath38 and @xmath39 are symmetric @xmath29 matrices . as noted in @xcite , @xmath30 is positive definite provided that @xmath37 , @xmath38 , @xmath39 and the initial matrix @xmath40 are positive definite .
however , these are quite strong restrictions on the parameters .
also , dvec model does not allow for dynamic dependence between volatility series . in order to avoid such strong restrictions on the parameter matrices , @xcite propose the bekk model , which is just a special case of a vec and , consequently , less general .
it has the attractive property that the conditional covariance matrices are positive definite by construction .
the model looks as follows : @xmath41 where @xmath42 is a lower triangular matrix and @xmath43 and @xmath44 are @xmath29 matrices . in the bekk model
it is easy to impose the definite positiveness of the @xmath30 matrix .
however , the parameter matrices @xmath43 and @xmath44 do not have direct interpretations since they do not represent directly the size of the impact of the lagged values of volatilities and squared returns .
@xcite present a paper that compares the performance of various bivariate arch and garch models , such as vec , bekk , etc , estimated using bayesian techniques .
as the authors observe , they are the first to perform model comparison using bayes factors and posterior odds in the mgarch setting .
the algorithm used for parameter estimation and inference is metropolis - hastings , and to check for convergence they rely on the cumsum statistics , introduced by @xcite , and used by @xcite in the univariate garch setting .
using the real data the authors found that the t - bekk models performed the best , leaving t - vec not so far behind ; t - vec model , sometimes also called t - vech , is a more general form of a dvec , seen above , where the mean - corrected returns follow a student - t distribution .
the name comes from a function called @xmath45 , which reshapes the lower triangular portion of a symmetric variance - covariance matrix into a column vector . to sum up
, the authors choose t - bekk model as clearly better than the t - vec , because it is relatively simple and has less parameters to estimate . on the other hand
, @xcite developed a prior distribution for a vech specification that directly satisfy both necessary and sufficient conditions for positive definiteness and covariance stationarity , while remaining diffuse and non - informative over the allowable parameter space .
these authors employed mcmc methods , including metropolis - hastings , to help enforce the conditions in this prior .
more recently , @xcite use the bekk - garch model to show the usefulness of a new posterior sampler called the adaptive hamiltonian monte carlo ( ahmc ) .
hamiltonian monte carlo ( hmc ) is a procedure to sample from complex distributions .
the ahmc is an alternative inferential method based on hmc that is both fast and locally adaptive .
the ahmc appears to work very well when the dimension of the parameter space is very high .
model selection based on marginal likelihood is used to show that full bekk models are preferred to restricted diagonal specifications .
additionally , @xcite suggests an approach called constrained hamiltonian monte carlo ( chmc ) in order to deal with high dimensional bekk models with targeting , which allow for a parameter dimension reduction without compromising the model fit , unlike the diagonal bekk .
model comparison of the full bekk and the bekk with targeting is performed indicating that the latter dominates the former in terms of marginal likelihood .
factor - garch was first proposed by @xcite to reduce the dimension of the multivariate model of interest using an accurate approximation of the multivariate volatility .
the definition of the factor - garch model , proposed by @xcite , says that bekk model in is a factor - garch , if @xmath43 and @xmath44 have rank one and the same left and right eigenvalues : @xmath46 , @xmath47 , where @xmath48 and @xmath49 are scalars and @xmath50 and @xmath51 are eigenvectors .
several variants of the factor model have been proposed .
one of them is the full - factor multivariate garch by @xcite : @xmath52 where @xmath53 is a @xmath54 vector of constants , which is time invariant , @xmath55 is a @xmath29 parameter matrix , @xmath56 is a @xmath54 vector of factors and @xmath57 is a @xmath29 diagonal variance matrix such that @xmath58 , where @xmath59 is the conditional variance of the @xmath60th factor at time @xmath5 such that @xmath61 , @xmath62 , @xmath63 .
then , the factors in the @xmath56 vector are garch(1,1 ) processes and the vector @xmath25 is a linear combination of such factors . it can be easily shown that @xmath30 is always positive definite by construction . however , the structure of @xmath30 depends on the order of the time series in @xmath31 .
@xcite have considered this problem to find the best ordering under the proposed model .
furthermore , @xcite investigate a full - factor mgarch model using the ml and bayesian approaches .
the authors compute maximum likelihood estimates using fisher scoring algorithm . as for the bayesian analysis ,
the authors have adopted a metropolis - hastings algorithm , and found that the algorithm is very time consuming , especially in high - dimensional data .
to speed - up the convergence , @xcite have proposed reparametrization of positive parameters and also a blocking sampling scheme , where the parameter vector is divided into three blocks : mean , variance and the matrix of constants @xmath55 . as mentioned before , the ordering of the univariate time series in full - factor models is important , thus to select `` the best '' model one has to consider @xmath64 possibilities for a multivariate dataset of dimension @xmath65 . instead of choosing one model and making inference ( as if the selected model was the true one ) , the authors employ a bayesian approach by calculating the posterior probabilities for all competing models and model averaging to provide `` combined '' predictions .
the main contribution of this paper is that the authors were able to carry out an extensive bayesian analysis of a full - factor mgarch model considering not only parameter uncertainty , but model uncertainty as well . as already discussed above
, a very common stylized feature of financial time series is the asymmetric volatility . @xcite
have proposed a new class of tree structured mgarch models that explore the asymmetric volatility effect .
same as the paper by @xcite , the authors consider not only parameter - related uncertainty , but also uncertainty corresponding to model selection .
thus in this case bayesian approach becomes particularly useful because an alternative method based on maximizing the pseudo - likelihood is only able to work after selecting a single model .
the authors develop an mcmc stochastic search algorithm that generates candidate tree structures and their posterior probabilities .
the proposed algorithm converged fast .
such modeling and inference approach leads to more reliable and more informative results concerning model - selection and individual parameter inference .
there are more models that are nested in bekk , such as the orthogonal garch for example , see @xcite and @xcite , among others .
all of them fall into the class of direct generalizations of univariate garch or linear combinations of univariate garch models .
another class of models are the nonlinear combinations of univariate garch models , such as conditional correlation ( ccc ) , dynamic condition correlation ( dcc ) , general dynamic covariance ( gdc ) and copula - garch models .
a very recent alternative approach that also considers bayesian estimation can be found in @xcite who proposes a new dynamic component models of returns and realized covariance ( rcov ) matrices based on time - varying wishart distributions . in particular , bayesian estimation and model comparison is conducted with a existing range of multivariate garch models and rcov models .
the ccc model , proposed by @xcite and the simplest in its class , is based on the decomposition of the conditional covariance matrix into conditional standard deviations and correlations .
then , the conditional covariance matrix @xmath30 looks as follows : @xmath66 where @xmath67 is diagonal matrix with the @xmath68 conditional standard deviations and @xmath69 is a time - invariant conditional correlation matrix such that @xmath70 and @xmath71 . the ccc approach can be applied to a wide range of univariate garch family models , such as exponential garch or gjr - garch , for example .
@xcite have estimated some real data using a variety of bivariate arch and garch models in order to select the best model specification and to compare the bayesian parameter estimates to those of the ml .
these authors have considered three arch and three garch models , all of them with constant conditional correlations ( ccc ) .
they have used a metropolis - hastings algorithm , which allows to simulate from the joint posterior distribution of the parameters . for model comparison and selection , @xcite
have obtained predictive distributions and assessed comparative validity of the analyzed models , according to which the ccc model with diagonal covariance matrix performed the best considering one - step - ahead predictions . a natural extension of the simple ccc model are the dynamic conditional correlation ( dcc ) models , firstly proposed by @xcite and @xcite .
the dcc approach is more realistic , because the dependence between returns is likely to be time - varying .
the models proposed by @xcite and @xcite consider that the conditional covariance matrix @xmath30 looks as @xmath72 , where @xmath73 is now a time varying correlation matrix at time @xmath14 .
the models differ in the specification of @xmath73 . in the paper by @xcite ,
the conditional correlation matrix is @xmath74 , where @xmath75 and @xmath76 are non - negative scalar parameters , such that @xmath77 , @xmath69 is a positive definite matrix such that @xmath78 and @xmath79 is a @xmath80 sample correlation matrix of the past @xmath81 standardized mean - corrected returns @xmath82 . on the other hand , in the paper by @xcite ,
the specification of @xmath83 is @xmath84 , where @xmath85 , @xmath86 is a mean - corrected standardized returns , @xmath48 and @xmath49 are non - negative scalar parameters , such that @xmath19 and @xmath87 is unconditional covariance matrix of @xmath88 . as noted in @xcite , the model by @xcite
does not formulate the conditional correlation as a weighted sum of past correlations , unlike in the dcc model by @xcite , seen above .
the drawback of both these models is that @xmath75 , @xmath76 , @xmath48 and @xmath49 are scalar parameters , so all conditional correlations have the same dynamics .
however , as @xcite notes it , the models are parsimonious .
moreover , as financial returns display not only asymmetric volatility , but also excess kurtosis , previous research , as in univariate case , has mostly considered using a multivariate student - t distribution for the errors . however , as already discussed above , this approach has several limitations .
@xcite propose a mgarch - dcc model , where the standardized innovations follow a mixture of gaussian distributions .
this allows to capture long tails without being limited by the degrees of freedom constraint , which is necessary to impose in the student - t distribution so that higher moments could exist .
the authors estimate the proposed model using the classical ml and bayesian approaches . in order to estimate model parameters , dynamics of single assets and dynamic correlations , and the parameters of the gaussian mixture ,
@xcite have relied on rwmh algorithm .
bic criteria was used for selecting the number of mixture components , which performed well in simulated data . using real data ,
the authors provide an application to calculating the value at risk ( var ) and solving a portfolio selection problem .
mle and bayesian approaches have performed similarly in point estimation , however , the bayesian approach , besides giving just point estimates , allows the derivation of predictive distributions for the portfolio var .
an extension of the dcc model of @xcite is the asymmetric dcc also proposed by @xcite , which incorporates an asymmetric correlation effect .
it means that correlations between asset returns decrease more in the bear market than they increase when the market performs well .
@xcite generalizes the adcc model into the agdcc model , where the parameters of the correlation equation are vectors , and not scalars .
this allows for asset - specific correlation dynamics . in the agdcc model ,
the @xmath89 matrix in the dcc model is replaced with : @xmath90 where @xmath91 are mean corrected standardized returns , @xmath92 selects just negative returns , `` diag '' stands for either taking just the diagonal elements from the matrix , or making a diagonal matrix from a vector , @xmath93 is a sample correlation matrix of @xmath94 , @xmath95 and @xmath96 are @xmath97 vectors , @xmath98 , @xmath99 and @xmath100 . to ensure positivity and stationarity of @xmath89 , it is necessary to impose @xmath101 and @xmath102 , @xmath103 .
the agdcc by @xcite is just a special case where @xmath104 , @xmath105 and @xmath106 .
up to our knowledge , the only paper that considers the agdcc model in the bayesian setting is @xcite that propose to model the distribution of the standardized returns as an infinite scale mixture of gaussian distributions by relying on bayesian non - parametrics .
this approach is presented in more detail in section [ section : non_param ] .
the use of copulas is an alternative approach to study return time series and their volatilities .
the main convenience of using copulas is that individual marginal densities of the returns can be defined separately from their dependence structure .
then , each marginal time series can be modeled using univariate specification and the dependence between the returns can be modeled by selecting an appropriate copula function .
a @xmath65-dimensional copula @xmath107 , is a multivariate distribution function in the unit hypercube @xmath108^k}$ ] , with uniform @xmath108}$ ] marginal distributions . under certain conditions
, the sklar theorem affirms that ( see , @xcite ) , every joint distribution @xmath109 , whose marginals are given by @xmath110 , can be written as @xmath111 , where @xmath112 is a copula function of @xmath113 , which is unique if the marginal distributions are continuous . the most popular approach to volatility modeling through copulas
is called the copula - garch model , where univariate garch models are specified for each marginal series and the dependence structure between them is described using a copula function .
a very useful feature of copulas , as noted by @xcite , is that the marginal distributions of each random variable do not need to be similar to each other .
this is very important in modeling time series , because each of them might be following different distributions .
the choice of copulas can vary from a simple gaussian copula to more flexible ones , such as clayton , gumbel , mixed gaussian , etc . in the existing literature different parametric and non - parametric specifications
can be used for the marginals and copula function @xmath114 .
also , the copula function can be assumed to be constant or time varying , as seen in @xcite , among others .
the estimation for copula - garch models can be performed in a variety of ways .
maximum likelihood is the obvious choice for fully parametric models .
estimation is generally based on a multi - stage method , where firstly the parameters of the marginal univariate distributions are estimated and then used to condition in estimating the parameters of the copula .
another approach is non- or semi - parametric estimation of the univariate marginal distributions followed by a parametric estimation of the copula parameters .
as @xcite has showed , the two - stage maximum likelihood approach lead to consistent , but not efficient estimators .
an alternative is to employ a bayesian approach , as done by @xcite .
the authors have developed a one - step bayesian procedure where all parameters are estimated at the same time using the entire likelihood function and , provided the methodology , for obtaining optimal portfolio , calculating var and cvar .
@xcite have used a gibbs sampler to sample from a joint posterior , where each parameter is updated using a rwmh . in order to reduce computational cost ,
the model and copula parameters are updated not one - by - one , but rather by blocks , that consist of highly correlated vectors of model parameters .
@xcite have also used bayesian inference for copula - garch models .
these authors have proposed a methodology for modelling dynamic dependence structure by allowing copula functions or copula parameters to change across time .
the idea is to use a threshold approach so these changes , that are assumed to be unknown , do not evolve in time but occur in distinct points .
these authors have also employed a rwmh for parameter estimation together with a laplace approximation .
the adoption of an mcmc algorithm allows the choice of different copula functions and/or different parameter values between two time thresholds .
bayesian model averaging is considered for predicting dependence measures such as the kendall s correlation .
they conclude that the new model performs well and offers a good insight into the time - varying dependencies between the financial returns .
@xcite developed bayesian inference of a multivariate garch model where the dependence is introduced by a d - vine copula on the innovations .
a d - vine copula is a special case of vine copulas which are very flexible to construct multivariate copulas because it allows to model dependency between pairs of margins individually .
inference is carried out using a two - step mcmc method closely related with the usual two - step maximum likehood procedure for estimating copula - garch models .
the authors then focus on estimating the var of a portfolio that shows asymmetric dependencies between some pairs of assets and symmetric dependency between others .
all the previously introduced methods rely on parametric assumptions for the distribution of the errors .
however , imposing a certain distribution can be rather restrictive and lead to underestimated uncertainty about future volatilities , as seen in @xcite
. therefore , bayesian non - parametric methods become especially useful , since they do not impose any specific distribution on the standardized returns .
bayesian non - parametrics is an alternative approach to the classical parametric bayesian statistics , where one usually gives some prior for the parameters of interest , whose distribution is unknown , and then observes the data and calculates the posterior .
the priors come from the family of parametric distributions .
bayesian non - parametrics uses a prior over distributions with the support being the space of all distributions .
then , it can be viewed as a distribution over distributions .
one of the most popular bayesian non - parametric modeling approach is based on dirichlet processes ( dp ) and mixtures of dirichlet processes ( dpm ) .
dp was firstly introduced by @xcite .
suppose that we have a sequence of exchangeably distributed random variables @xmath115 from an unknown distribution @xmath116 , where the support for @xmath117 is @xmath37 . in order to perform bayesian inference , we need to define the prior for @xmath116 .
this can be done by considering partitions of @xmath37 , such as @xmath118 , and defining priors over all possible partitions .
we say that @xmath116 has a dirichlet process prior , denoted as @xmath119 , if the set of associated probabilities given @xmath116 for any partition follows a dirichlet distribution , @xmath120 , where @xmath121 is a precision parameter that represents our prior certainty of how concentrated the distribution is around @xmath122 , which is a known base distribution on @xmath37 .
the dirichlet process is a conjugate prior .
thus , given @xmath123 independent and identically distributed samples from @xmath116 , the posterior distribution of @xmath116 is also a dirichlet process such that @xmath124 where @xmath125 is the empirical distribution function .
there are two main ways for generating a sample from the marginal distribution of @xmath126 , where @xmath127 and @xmath128 : the polya urn and stick breaking procedures . on the one hand ,
the polya urn scheme can be illustrated in terms of a urn with @xmath48 black balls ; when a non - black ball is drawn , it is placed back in the urn together with another ball of the same color . if the drawn ball is black , a new color is generated from @xmath129 and a ball of this new color is added to the urn together with the black ball we drew .
this process gives a discrete marginal distribution for @xmath126 since there is always a probability that a previously seen value is repeated .
on the other hand , the stick - breaking procedure is based on the representation of the random distribution @xmath116 as a countably infinite mixture : @xmath130 where @xmath131 is a dirac measure , @xmath132 and the weights are such that @xmath133 where @xmath134 .
this implies that the weights @xmath135 as @xmath136 .
the discreteness of the dirichlet process is clearly a disadvantage in practice .
a solution was proposed by @xcite by using dpm models where a dp prior is imposed over over the distribution of the model parameters , @xmath137 , as follows : @xmath138 observe that @xmath139 is a random distribution drawn from the dp and because it is discrete , multiple @xmath140 s can take the same value simultaneously , making it a mixture model .
in fact , using the stick - breaking representation , the hierarchical model above can be seen as an infinite mixture of distributions : @xmath141 where the weights are obtained as before : @xmath142 , @xmath143 , for @xmath144 , and where @xmath145 and @xmath146 .
regarding inference algorithms , there are two main types of approaches . on the one hand , the marginal methods , such as those proposed by @xcite , @xcite and @xcite , which rely on the polya urn representation .
all these algorithms are based on integrating out the infinite dimensional part of the model .
recently , another class of algorithms , called conditional methods , have been proposed . these approaches , based on the stick - breaking scheme ,
leave the infinite part in the model and sample a finite number of variables .
these include the procedure by @xcite , who introduces slice sampling schemes to deal with the infiniteness in dpm , and the retrospective mcmc method of @xcite , that is later combined by @xcite with slice sampling method by @xcite to obtain a new composite algorithm , which is better , faster and easier to implement .
generally , the stick - breaking compared to the polya urn procedures produce better mixing and simpler algorithms .
as mentioned above , so far there has been little research in modeling volatilities with mgarch using the dpm models . up to our knowledge , these only include : @xcite for univariate garch , and @xcite and @xcite , for mgarch .
@xcite have applied semi - parametric bayesian techniques to estimate univariate garch - type models .
these authors have used the class of scale mixtures of gaussian distributions , that allow for the variances to change over components , with a dirichlet process prior on the mixing distribution to model innovations of the garch process . the resulting class of models is called dpm - garch type models . in order to perform bayesian inference on the new model , the authors employ a stick - breaking sampling scheme and make use of the ideas proposed in @xcite , @xcite and @xcite .
the new scale mixture model was compared to a simpler mixture of two gaussians , student - t and the usual gaussian models .
the estimation results in all three cases were quite similar , however , the scale mixture model is able to capture skewness as well as kurtosis and , based on the approximated log marginal likelihood ( lml ) and @xmath147 , provided the best performance in simulated and real data .
finally , @xcite have applied the resulting model to perform one - step - ahead predictions for volatilities and var . in general
, the non - parametric approach leads to wider bayesian credible intervals and can better describe long tails .
@xcite propose a bayesian non - parametric modeling approach for the innovations in mgarch models .
they use a mgarch specification , proposed by @xcite , which is a different representation of a well known dvec model , introduced above .
the innovations are modeled as an infinite mixture of multivariate normals with a dp prior .
the authors have employed polya urn and stick - breaking schemes and , using two data sets , compared the three model specifications : parametric mgarch with student - t innovations ( mgarch - t ) , garch - dpm-@xmath148 that allows for different covariances ( scale mixture ) and mgarch - dpm , allowing for different means and covariances of each component ( location - scale mixture ) . in general , both semi - parametric models produced wider density intervals . however , in mgarch - t model a single degree of freedom parameter determines the tail thickness in all directions of the density , meanwhile the non - parametric models are able to capture various deviations from normality by using a certain number of components .
these results are consistent with the ones in @xcite . as for predictions , both semi - parametric models performed equally good and outperformed the parametric mgarch - t specification .
finally , the paper by @xcite can be seen as a direct generalization of the paper by @xcite to the multivariate framework . here , same as in @xcite ,
the authors have proposed using an infinite scale mixture of normals for the standardized returns . for the mgarch model
a gjr - adcc was chosen , allowing for asymmetric volatilities and asymmetric time - varying correlations .
moreover , the authors have carried - out a simulation study that illustrated the adaptability of the dpm model .
finally , the authors provided one real data application to portfolio decision problem concluding that dpm models are less restrictive and more adaptive to whatever distribution the data comes from , therefore , can better capture the uncertainty about financial decisions .
to sum up , the findings in the above papers are consistent : the bayesian semi - parametric approach leads to more flexible models and is better in explaining heavy - tailed return distributions , which parametric models can not fully capture .
the parameters are less precise , i.e. wider bayesian credible intervals are observed because the semi - parametric models are less restricted .
this provides a more adequate measure of uncertainty .
if in the gaussian setting the credible intervals are very narrow and the real data is not gaussian , this makes the agent overconfident about her decisions , and she takes more risk than she would like to assume .
@xcite observes that the combination of bayesian methods and mcmc computational algorithms provide new modeling possibilities and calls for more research regarding non - parametric bayesian time series modeling .
this illustration study using real data has basically two goals : firstly , to show the advantages of the bayesian approach , such as the ability to obtain posterior densities of quantities of interest and the facility to incorporate various constraints on the parameters . secondly , to illustrate the flexibility of the bayesian non - parametric approach for garch modeling .
the data used for estimation are the log - returns ( in percentages ) , obtained from close prices adjusted for dividends and splits , of two market indices : ftse100 and s&p500 from november 10th , 2004 till december 10th , 2012 , resulting into a sample size of 2000 observations .
ftse100 is a share index of the 100 companies listed on the london stock exchange with the highest market capitalization .
s&p500 is a stock market index based on the common stock prices of 500 top publicly traded american companies .
the data was obtained from yahoo finance .
figure [ f : returns ] and table [ t : descriptive_table ] present the basic plots and descriptive statistics of the two log - return series .
[ f : returns ] ccc ' '' '' & ftse100 & s&p500 + mean & 0.0112 & 0.0099 + median & 0.0344 & 0.0779 + variance & 1.7164 & 1.9617 + skewness & -0.0974 & -0.3001 + kurtosis & 10.5464 & 12.5674 + correlation & 0.6060 + as seen from the plot and descriptive statistics , the data is slightly skewed and with high kurtosis , therefore , assuming a gaussian distribution for the standardized returns would be inappropriate
. therefore , we estimate this bivariate time series using an adcc model by @xcite , presented in section 3.4 . , which incorporates an asymmetric correlation effect .
the univariate series are assumed to follow gjr - garch@xmath149 models in order to incorporate the leverage effect in volatilities . as for the errors , we use an infinite scale mixture of gaussian distributions .
therefore , we call the final model gjr - adcc - dpm . inference and prediction is carried out using bayesian non - parametric techniques , as seen in @xcite .
the selection of the mgarch specification is arbitrary and other models might work equally well .
for the sake of comparison , we estimate a restricted gjr - adcc - gaussian model using maximum likelihood and bayesian approaches .
the estimation results are presented in table [ table : estimation ] .
[ table : estimation ] c c c c c c c & & & + & estimate & st .
& mean & @xmath150 ci&mean & @xmath150 ci + + @xmath151 & 0.0166 & 0.0020 & 0.0192 & ( 0.0130 , 0.0258 ) & 0.0181 & ( 0.0104 , 0.0264 ) + @xmath152 & 0.0190 & 0.0016 & 0.0249 & ( 0.0174 , 0.0316 ) & 0.0219 & ( 0.0153 , 0.0293 ) + @xmath153 & @xmath154 & @xmath155 & 0.0058 & ( 0.0002 , 0.0177 ) & 0.0046 & ( 0.0003 , 0.0112 ) + @xmath156 & @xmath157 & @xmath158 & 0.0053 & ( 0.0002 , 0.0173 ) & 0.0059 & ( 0.0002 , 0.0151 ) + @xmath159 & 0.9087 & 0.0045 & 0.9010 & ( 0.8841 , 0.9152 ) & 0.8956 & ( 0.8762 , 0.9139 ) + @xmath160 & 0.9079 & 0.0050 & 0.8888 & ( 0.8705 , 0.9088 ) & 0.8851 & ( 0.8675 , 0.9041 ) + @xmath161 & 0.1535 & 0.0085 & 0.1587 & ( 0.1351 , 0.1871 ) & 0.1586 & ( 0.1057 , 0.2089 ) + @xmath162 & 0.1483 & 0.0092 & 0.1737 & ( 0.1398 , 0.2020 ) & 0.1758 & ( 0.1134 , 0.2142 ) + @xmath163 & 0.0075 & 0.0020 & 0.0071 & ( 0.0014 , 0.0145 ) & 0.0095 & ( 0.0040 , 0.0156 ) + @xmath164 & 0.9898 & 0.0029 & 0.9818 & ( 0.9665 , 0.9898 ) & 0.9806 & ( 0.9693 , 0.9901 ) + @xmath96 & @xmath165 & @xmath166 & 0.0076 & ( 0.0002 , 0.0153 ) & 0.0039 & ( 0.0001 , 0.0114 ) + the estimated parameters are very similar for all three approaches , except for @xmath167 , the asymmetric correlation coefficient @xmath96 .
since @xmath167 and @xmath96 are so close to zero , the ml has some trouble in estimating those parameters .
overall , the @xmath96 is small , indicating little evidence of asymmetrical behavior in correlations .
figure [ fig : tails ] shows the estimated marginal predictive densities of the one - step - ahead returns in log scale using the bayesian approach .
we can observe the differences in tails arising from different specification of the errors .
the dpm model allows for a more flexible distribution , therefore , for more extreme returns , i.e. fatter tails .
the estimated densities were obtained using the procedure described in @xcite . for bayesian gaussian and dpm models , scaledwidth=100.0% ]
table [ table : volatilities ] presents the estimated mean , median and @xmath150 credible intervals of one - step - ahead volatility matrices in bayesian context .
the matrix element ( 1,1 ) represents the volatility for the ftse100 series , ( 2,2 ) for the s&p500 , and the elements in the diagonal ( 1,2 ) and ( 2,1 ) represent the covariance of both financial returns .
figure [ fig : vols_returns ] draws the posterior distributions for volatilities and correlation .
the estimated mean volatilities for both , dpm and gaussian approaches , are very similar , however , the main differences arise from the shape of the posterior distribution .
@xmath150 credible intervals for dpm model correlation are wider providing a more realistic measure of uncertainty about future correlations between two assets .
this is a very important implication in financial setting , because if an investor chooses to be gaussian , she would be overconfident about her decision and unable to adequately measure the risk she is facing .
see @xcite for a more detailed comparison of dpm and alternative parametric approaches in portfolio decision problems .
[ table : volatilities ] c c c c c c c & & & + & constant & ml gaussian & mean & @xmath150 ci&mean & @xmath150 ci + & & & median & ci length&median & ci length + + @xmath168 & 1.7164 & 0.4007 & 0.4098 & ( 0.3681 , 0.4538 ) & 0.3996 & ( 0.3550 , 0.4512 ) + & & & 0.4099 & 0.0857 & 0.3983 & 0.0962 + @xmath169 & 1.1120 & 0.2911 & 0.2800 & ( 0.2571 , 0.3077 ) & 0.2751 &
( 0.2421 , 0.3123 ) + & & & 0.2790 & 0.0506 & 0.2742 & 0.0702 + @xmath170 & 1.9617 & 0.4939 & 0.4635 & ( 0.4159 , 0.5193 ) & 0.4431 & ( 0.3912 , 0.5059 ) + & & & 0.4606 & 0.1034 & 0.4408 & 0.1146 + to sum up , this illustration has shown the main differences between the standard estimation procedures and the new non - parametric approach .
even though the point estimates for the parameters and the one - step - ahead volatilities are very similar , the main differences arise from the thickness of tails of predictive distributions of one - step - ahead returns and the shape of the posterior distribution for the one - step - ahead volatilities .
in this paper we reviewed univariate and multivariate garch models and inference methods , putting emphasis on the bayesian approach .
we have surveyed the existing literature that concerns various bayesian inference methods for mgarch models , outlining the advantages of the bayesian approach versus the classical procedures .
we have also discussed in more detail the recent bayesian non - parametric method for garch models , which avoid imposing arbitrary parametric distributional assumptions .
this new approach is more flexible and can describe better the uncertainty about future volatilities and returns , as has been illustrated using real data .
we are grateful to an anonymous referee for helpful comments .
the first and second authors are grateful for the financial support from mec grant eco2011 - 25706 .
the third author acknowledges financial support from mec grant eco2012 - 38442 .
81 natexlab#1#1[1]`#1 ` [ 2]#2 [ 1]#1 [ 1]http://dx.doi.org/#1 [ ] [ 1]pmid:#1 [ ] [ 2]#2 , & ( ) . . , .
( ) . . , _
_ , . . , & ( )
( ) . . , _
_ , . , & ( ) . . in ( ed
. ) , _ _ chapter .
( pp . ) . : .
( ) . . , _
_ , . . , & ( )
, , & ( ) . . , _
_ , . . , & ( )
, , & ( ) . . , _
_ , . , , & ( ) . . ,
_ , . , & ( ) .
, _ _ , . , & ( ) . . ,
_ , . , & ( ) . . ,
, ( pp . ) .
( ) . . , _
( ) . . , _
( ) . . , _
_ , . , , & ( ) .
, , & ( ) . . in , & ( eds . ) , _ _ ( pp . ) . : .
, , & ( ) . . , _
( ) . . , . , & ( )
, , & ( ) . . , _
, , & ( ) . . , _
_ , . . , & ( )
_ , . , & ( ) .
( ) . . , _
( ) . . , _
( ) . . , _
( ) . . , _
_ , . , & ( ) .
, , & ( ) . . , _
_ , . , & ( ) .
( ) . . , _
_ , . , & ( ) .
( ) . . , _
, , & ( ) . . , _
, , & ( ) . . , _
, , & ( ) . . , _
, , & ( ) . . ,
_ , . , & ( ) .
_ , . , & ( ) .
_ , . , & ( ) .
, & ( ) . . , _
_ , . , & ( ) .
. , ( pp . ) . , & ( ) .
. in , , , & ( eds . ) , _ _ chapter .
( pp . ) . : .
( ed . ) . , &
_ _ , . , & ( ) .
( ) . . , _
( ) . . , _
, , & ( ) . . , _
( ) . . , _
( ) . . , _
_ , . , & ( ) .
, _ _ , . , & ( ) .
( ) . . , _
( ) . . , _
( ) . . , _ _ , . , & ( ) . . , _
, ( pp . ) . , & ( ) . . , _
( ) . . , _
( ) . . in , , , & ( eds . ) , _ _ chapter .
( pp . ) . : . , & ( ) . .
( ed . ) . .
, & ( ) . . in , , , & ( eds . ) , _ _ ( pp . ) . : .
( ) . . , _
( ) . . in , & ( eds . ) , _ _ . : .
( ) . . in , , , & ( eds . ) , _ _ ( pp . ) . : . ( ) . . , _
( ed . ) . : , & ( ) . . , _ _ , . .
( ) . . , _
, , & ( ) . . , . , , & ( ) . . , _
, , & ( ) . . , _
, , & ( ) . . , _
( ) . . , _ _ , . .
( ) . . , _
_ , . . , & ( )
( ) . . , _ | this survey reviews the existing literature on the most relevant bayesian inference methods for univariate and multivariate garch models .
the advantages and drawbacks of each procedure are outlined as well as the advantages of the bayesian approach versus classical procedures .
the paper makes emphasis on recent bayesian non - parametric approaches for garch models that avoid imposing arbitrary parametric distributional assumptions .
these novel approaches implicitly assume infinite mixture of gaussian distributions on the standardized returns which have been shown to be more flexible and describe better the uncertainty about future volatilities .
finally , the survey presents an illustration using real data to show the flexibility and usefulness of the non - parametric approach .
* keywords : * bayesian inference ; dirichlet process mixture ; financial returns ; garch models ; multivariate garch models ; volatility . |
in the past few years there has been significant progress in the study of the chiral - continuum extrapolation problem for lattice qcd with different types of fermions ( see , for example , refs .
extrapolation of lattice qcd simulation results to the physical ( light ) quark masses is a nontrivial task due to the nonanalytic variation of hadronic properties with quark masses which is a consequence of spontaneous chiral symmetry breaking . to perform an extrapolation to the continuum limit we need to account for direct chiral , taste ( in the staggered case ) and rotational symmetry breaking by finite lattice spacing . at present ,
lattice qcd simulations with staggered fermions defined in ref
. @xcite reached smaller quark masses @xcite compared with other types of lattice fermions . for example , compared with unquenched simulations using ginsparg - wilson fermions ( defined in ref .
@xcite ) simulations with staggered fermions are computationally less demanding .
the computational cost depends on many factors like lattice spacing and the practical implementation of the fermions in the simulations ( see ref .
@xcite and references therein ) . from ref .
@xcite it follows that the simulations with ginsparg - wilson fermions may be about ten to one hundred times more expensive than with improved staggered fermions at comparable masses . however , the price of these advantages of staggered fermions is that they contain an internal flaw . in those simulations , to reduce the number of taste degrees of freedom , the fourth - root trick is employed . in the continuum limit
this trick is consistent because there are no taste - changing interactions , however , when one includes these interactions the finite lattice spacing makes the root of the determinant non - local and the trick remains controversial ( see , for example , ref.@xcite ) . in order to include the relatively large discretization effects in staggered calculations , staggered chiral perturbation theory has been developed ( see refs.@xcite ) .
application of the staggered lagrangian allowed one to determine the pseudoscalar masses and decay constant @xcite .
for example , recently in ref .
@xcite there has been considerable success in determining the @xmath1 meson decay constant using the milc collaboration unquenched gauge configurations with three flavors of light sea quarks . in these calculations both valence and sea light quarks were represented by the highly improved ( asqtad ) staggered quark action which allows for a much smoother chiral extrapolation to physical up and down quarks than has been possible in the past .
being computationally very costly , ginsparg - wilson fermions , on the other hand , do not suffer from the inconsistency of staggered fermions .
moreover , these fermions have an exact chiral symmetry .
it has been shown in refs .
@xcite that it is possible to combine the good chiral properties of ginsparg - wilson fermions , taking them as valence quarks where they are computationally less demanding , with numerically fast staggered fermions as sea quarks .
extrapolation for the hybrid ( mixed fermion ) program has recently been addressed for baryons by tiburzi in ref .
@xcite . here
we consider the application of this program for the @xmath2 meson , which is a classic testing ground for qcd with light quarks .
+ in this report we extend the study of qcd for staggered sea and valence ginsparg - wilson quarks in ref.@xcite to calculate the mass of the vector meson at one - loop level up to order @xmath3 .
qcd with staggered sea and valence ginsparg - wilson quarks , which as in @xcite we will refer to as mixed qcd , consists of @xmath4 fully quenched ( valence ) quarks ( with @xmath4 ghost - quarks ) and @xmath5 staggered ( sea ) quarks .
the masses of the sea and valence quarks , in principle , could be chosen to be completely arbitrary , except that the quenched quark masses are fixed to be equal to the masses of their ghost partners .
the full graded chiral symmetry of mixed qcd in the chiral and continuum limits is described via the semi - direct product @xmath6 , where @xmath7 and @xmath4 are the numbers of unquenched ( staggered ) and quenched quarks , respectively . as in full qcd
, we expect that the axial @xmath8 symmetry is also broken by the anomaly .
we also phenomenologically include in our calculations the one - loop contribution from the @xmath9 decay which can not be consistently incorporated in the heavy meson formalism because of the high momentum involved in this process .
however , the inclusion of this loop is justified by the physical results obtained , for example , in refs .
@xcite and references therein .
in order to consistently include lattice discretization effects the symanzik action @xcite needs to be constructed , based on the symmetry constraints of the underlying lattice theory , with further projection to the low energy effective theory .
this task was carried out in refs.@xcite and here , we will only briefly mention the main results . in general , the symanzik action up to order @xmath10 has the form : @xmath11 where @xmath12 is the action corresponding to continuum partially quenched theory described in refs .
@xcite , @xmath13 ( @xmath14 ) is the action consisting of dimension @xmath15 operators . here
, we will require that the heavy meson chiral lagrangian , corresponding to @xmath12 , matches the corresponding partially quenched chiral lagrangian , similar to that in ref .
@xcite or in ref .
@xcite when the lattice spacing is set to zero . from discussions in ref .
@xcite ( see also refs .
@xcite ) the actions @xmath16 and @xmath17 are equal to zero because of the specific character of the fermions we use .
the action @xmath18 , according to ref .
@xcite , contains only six allowed operators which break taste or rotational symmetries ( see also for details refs .
the chiral lagrangian corresponding to the action @xmath19 will give only a tree level contribution of order @xmath20 .
although allowed in our counting scheme , this contribution will be disregarded , as currently we are interested in the @xmath0 analytic behavior of vector meson masses . using the results from the operator construction in refs .
@xcite we will study the analytic behavior of the vector meson mass in both the chiral and continuum limits , without constructing the explicit expression for the heavy meson pqchpt .
mixed chpt in the goldstone meson sector with staggered and ginsparg - wilson quarks was first studied in ref.@xcite .
we first give a brief description of that work .
+ assuming that the @xmath21 symmetry is spontaneously broken down to its vector part , the particle spectrum will contain light pseudoscalar bosons .
these goldstone meson fields can be written in terms of a @xmath22 , unitary matrix field , @xmath23 , defined as : @xmath24 where @xmath25 , @xmath26 is the bosonic matrix and @xmath27 is a low energy constant . in particular , we are interested in the case with @xmath28 and @xmath29 , for which the bosonic matrix has the following form : @xmath30 \pi^- & d & k^0 & q_{dx } & q_{dy } & ... & ... \\[5pt ] k^- & \bar{k}^0 & s & q_{sx } & q_{sy } & ... & ... \\[5pt ] q^\dag_{ux } & q^\dag_{dx } & q^\dag_{sx } & x & p^+ & r^\dag_{\tilde{x}x } & r^\dag_{\tilde{y}x } \\[5pt ] q^\dag_{uy } & q^\dag_{dy } & q^\dag_{sy } & p^- & y & r^\dag_{\tilde{x}y } & r^\dag_{\tilde{y}y } \\[5pt ] ... & ... & ... & r_{\tilde{x}x } & r_{\tilde{x}y } & \tilde{x } & \tilde{p}^+ \\[5pt ] ... & ... & ... & r^\dag_{\tilde{y}x } & r^\dag_{\tilde{y}y } & \tilde{p}^- & \tilde{y } \\[5pt ] \end{array } \right)\ ] ] following the notation of ref .
@xcite , the following correspondence is implied : @xmath31 , @xmath32 , @xmath33 , ( @xmath34 , @xmath35 and @xmath36 are the analogous combinations of valence ghost quarks ) , @xmath37 , similarly for @xmath38 , @xmath39 , and @xmath40 .
@xmath41 , where @xmath42 and @xmath43 ( note that @xmath44 is a @xmath45 matrix in taste ) .
the pseudoscalar states @xmath46 , @xmath47 , @xmath48 , ... , etc . in eq .
( [ phi ] ) , which consist of staggered quarks only , form the `` staggered '' sector of the mixed pseudoscalar boson chpt .
this sector is well studied in refs .
each state in the `` staggered '' sector can be represented as : @xmath49 ( similarly for @xmath47 , @xmath50 , etc ) where @xmath51 are the sixteen euclidean gamma matrices associated with taste ( see , for example , ref.@xcite ) . under chiral symmetry transformations
: @xmath52 where @xmath53 . as in refs .
@xcite , we adopt the following counting scheme : @xmath54 where @xmath55 denotes the typical qcd scale . this counting scheme is relevant for simulations with improved staggered quarks @xcite .
the leading order chiral lagrangian ( in this counting scheme ) that contains the terms of order @xmath56 is of the form : @xmath57 here @xmath58 denotes a supertrace in flavor space and @xmath59 is a low - energy constant . in our case
the diagonal mass matrix is : @xmath60 for convenience , to derive the flavor neutral propagators , the singlet mass parameter @xmath61 in eq .
( [ lagrangian ] ) is allowed to be finite , however , at the end of the calculations we always could take the limit @xmath62 .
the potential @xmath63 in the leading order lagrangian comprises all terms proportional to @xmath64 . at this stage
we will not go deep into describing the form of the @xmath63 ( we refer to refs .
@xcite for more information ) . here
, we present only those results required for our further study . at tree level , because of the specific properties of potential @xmath65 , the valence - valence mesons with only ginsparg - wilson quarks obey the continuum - like mass relations : @xmath66 in our calculations the only relevant valence - valence goldstone bosons will be flavor neutral because the valence quarks in the sea loops will be canceled by their ghost partners and , in addition , g - parity suppresses the process and @xmath67 have even g - parity , while the pseudoscalar meson has odd g - parity ] @xmath68 .
we also will be interested only in flavor - neutral and taste singlet staggered quark combinations , so in the sea - sea sector , for a meson in a singlet taste channel @xmath69 made of sea quarks of flavor @xmath70 , we have : @xmath71 where the constant @xmath72 is given , for example , in ref .
@xcite ( eq . ( 27 ) ) .
finally , the masses of the valence - sea mesons ( @xmath73 or @xmath74 ) are given as : @xmath75 where @xmath76 is an undetermined parameter in this mixed theory which is taste independent .
the flavor - charged ( non - diagonal ) fields in eq .
( [ phi ] ) have only connected propagators ( @xmath77 ) , in the quark - flow sense , which are given by the expression : @xmath78 where @xmath79 , @xmath80 @xmath81 \{ghost , quark } , @xmath82 if @xmath83 quark and @xmath84 if @xmath85 ghost .
disconnected propagators for flavor neutral mesons ( like , for example , @xmath46 or @xmath86 ) contain hairpin - like interactions ( through the @xmath61 term ) .
the sum of all these hairpin - like diagrams ( in the way described in ref.@xcite ) for the disconnected @xmath87 propagators ( @xmath88 $ ] ) will give ( as in ref .
@xcite ) the following result : @xmath89 = -\frac{m^2_0}{3}\frac{(q^2 + m^2_{u_i})(q^2 + m^2_{d_i})(q^2 + m^2_{s_i})}{(q^2 + m^2_{x})(q^2 + m^2_{y})(q^2 + m^2_{\pi^0_i})(q^2 + m^2_{\eta_i})(q^2 + m^2_{\eta'_i})}\end{aligned}\ ] ] where @xmath90 , @xmath91 and @xmath92 are the diagonalized flavor - neutral , singlet taste - channel mass eigenstates .
although there should be vector and axial - vector hairpin contributions , as in ref .
@xcite , we do not include them here because they are irrelevant in our later calculations .
chpt with staggered sea and chiral ( ginsparg - wilson ) valence quarks was successfully applied to calculate the corrections to pseudoscalar masses and decay constants to one loop in ref .
@xcite and recently , in ref .
@xcite , this approach was applied to calculate masses and magnetic moments of baryons . in this work
we will calculate vector meson mass corrections at one - loop level , without explicitly constructing the corresponding heavy meson lagrangian but rather using general principles .
+ from eq .
( [ symanzik ] ) it follows that continuum heavy meson chiral lagrangian will consist of the parts : @xmath93 and @xmath94 . as shown in ref .
@xcite , from the comparison with the simulations in ref .
@xcite , the contribution from the terms of order @xmath95 may be neglected .
the chiral lagrangian , @xmath96 , corresponding to @xmath97 in eq .
( [ symanzik ] ) , will consist of terms of order @xmath98 . and finally , we will not have terms of order @xmath99 ( @xmath14 ) in the heavy meson chiral lagrangian because they can be eliminated by using the equations of motion @xcite . now , it is easy to see that @xmath100 gives one - loop chiral corrections to the vector meson mass see the diagrams @xmath101 shown in fig . 1 .
the terms in @xmath102 and @xmath96 will give tadpole as well as tree level contributions .
from these considerations for the total tree level correction @xmath103 to the vector meson mass ( up to order @xmath10 ) it follows that : @xmath104 where , in general , @xmath105 ( @xmath106 ) with @xmath107 as the renormalization scale . for concreteness we will calculate the mass correction of a charged @xmath2 meson at rest .
( the calculation is somewhat more complicated for a neutral @xmath2 meson , but because of isospin symmetry , the mass correction of all @xmath2 mesons must be the same . ) in the case @xmath28 and @xmath29 we have staggered @xmath108 , @xmath109 and @xmath110 sea quarks with @xmath111 and @xmath112 valence quenched quarks ( corresponding to different flavors ) .
the quark flow diagrams for this case are presented in fig . 1 , with the propagators for the connected diagrams ( fig .
1 @xmath113 ) given by : @xmath114 \frac{1}{((p - q)^2 + m^2 ) } \ , \\ g^{con}_{vpp } & = \sum_{f = u , d , s } \frac{1}{(q^2 + m^2_{q_{fx}})((p - q)^2 + m^2_{q_{fy } } ) } , \end{aligned}\ ] ] where @xmath115 is the initial momentum of the vector meson with mass @xmath116 ( @xmath117 in euclidean space ) , @xmath118 is the momentum of intermediate pseudoscalar meson state and instead of using a heavy meson propagator we just used the usual one ( which is practically the same ) with the mass of intermediate vector meson equal to the mass of the initial one .
the subindices on the propagators remind us that in the self - energy integral we need to take into account the vector - vector - pseudoscalar ( @xmath119 ) and vector - pseudoscalar - pseudoscalar ( @xmath120 ) nature of the meson interactions . for the diagrams in fig .
1@xmath121 and @xmath122 in the limit @xmath123 , we have : @xmath124 ^ 2 .
\end{aligned}\ ] ] notice , that for case shown in fig .
1@xmath125 there are no additional double hairpin diagrams because of g - parity .
now , let us take @xmath126 ( @xmath127 picture ) as in milc simulations @xcite . in this case :
@xmath128 let us also require that @xmath129 and @xmath130 or equivalently @xmath131 . the first choice will help us to find the correction to the @xmath2 meson mass because it consists of two light quarks .
the second choice is natural because in the continuum limit it gives @xmath132 .
taking these two conditions into account we can write the above propagator in the form : @xmath133 \frac{1}{((p - q)^2 + m^2)}. \end{aligned}\ ] ] using the method in ref .
@xcite for the total self - energy we have : @xmath134 % \sigma^{nlna}_{total } & = -\frac{f^2_{\rho \pi \pi}}{4\pi^2 m^2 } % \left[2m^4_{q_{ux}}ln(m_{q_{ux } } ) + m^4_{q_{sx}}ln(m_{q_{sx } } ) \right ] \end{aligned}\ ] ] where @xmath135 \sigma_{vpp}(m^2 , \lambda ) & = -\frac{f^2_{\rho \pi \pi}}{6\pi^2}\int^\infty_0 dk \frac{k^4 u^2_{vpp}(k^2 , \lambda)}{\omega(\omega^2 - \frac{m^2}{4 } ) } , \end{aligned}\ ] ] we used the notation @xmath136 and the functions @xmath137 are finite range regulators ( @xcite ) .
for example , the leading non - analytic ( lna ) contribution to the self energy is : @xmath138 in the continuum limit ( where we have @xmath139 ) the lna contribution is : @xmath140 this result agrees with that in ref .
@xcite for the vector @xmath2 meson - where in the limit @xmath141 they found : @xmath142 .
\end{aligned}\ ] ] note that while the coefficients in these two expressions are written differently , since @xmath143 for values of @xmath144 @xmath145 ( @xcite ) , @xmath146 ( @xcite ) and @xmath147 @xmath148 , the expressions are equal
. it would be also interesting to consider the case , when @xmath149 , @xmath150 and @xmath151 ( @xmath152 in the continuum limit ) . using these conditions and recalling that the @xmath153 vector meson consists of one light and one strange quark we find the self energy contribution : @xmath154 in the continuum limit ( @xmath155 and @xmath152 ) the lna contribution to the total self - energy is : @xmath156 again , comparing this result with that in ref .
@xcite for the @xmath157 meson mass correction ( in the limit @xmath158 ) : @xmath159 \end{aligned}\ ] ] we see good agreement with our result for the lna contribution in the continuum limit .
now , returning to @xmath2 meson , let us take into account the tadpole contribution from @xmath160 .
this case is similar to @xmath100 but without intermediate vector meson propagators and we do nt need to include the diagram with two intermediate pseudoscalars .
the total self - energy correction from @xmath161 is : @xmath162 with the definition : @xmath163 where , in general , @xmath164 , @xmath165 and @xmath166 are undetermined parameters and @xmath167 is a finite range regulator @xcite .
the term proportional to @xmath165 is absent in the continuum limit ( see , for example , ref .
@xcite when lattice spacing is set to zero ) .
this means that the tadpole contribution comes only from the operators in @xmath96 .
then , for example , if @xmath168 ( applying dimensional regularization ) we have : @xmath169 the calculations above were performed in the case where @xmath111 and @xmath170 correspond to different flavors . to calculate the mass of the neutral vector meson we need to change our quark flow diagrams and add one more with dhp insertions on each intermediate pseudoscalar line .
however , as noted above , isospin symmetry for the vector meson , means that we need not perform separate calculations for the neutral vector meson mass .
imposing the conditions @xmath171 with @xmath172 we can simplify the expression for the tree level contribution and the vector meson mass in both chiral and continuum expansions can be written in the form : @xmath173 where @xmath174 , @xmath175 and @xmath176 , @xmath177 are new dimensionful parameters . the dependence on @xmath178
may be absorbed in the coefficients @xmath176 and @xmath179 if one fixes the mass of the strange quark in the simulations . from the definition of @xmath180 and @xmath181 using the appendix in ref .
@xcite we can separate the leading non - analytic contributions from @xmath182 and @xmath183 @xmath184 , \\[20pt ] \nonumber \widetilde{\sigma}^{lna}_{total } & = \frac{\beta a^2}{8\pi^2}\left[2m^2_{q_{ux}}ln \frac{m_{q_{ux}}}{\mu } + m^2_{q_{sx}}ln \frac{m_{q_{sx}}}{\mu } - m^2_{x}ln \frac{m_{x}}{\mu } + \frac{1}{3}m^2_{\eta_i}ln \frac{m_{\eta_i}}{\mu } \right ] \end{aligned}\ ] ] where @xmath185 is the physical mass of the @xmath2 meson and using eqs .
( [ val - val][mixed ] ) we can express all meson masses as functions of @xmath186 and @xmath178 : @xmath187 m^2_{\eta_i } & = \frac{1}{3}m^2_{\pi } + \frac{2}{3}m^2_{s_i}.\end{aligned}\ ] ] it is also of practical interest to consider the general case with only two conditions : @xmath172 and @xmath129 . in this case
the total self - energies are : @xmath188 @xmath189 where @xmath190 \nonumber \tilde{\alpha } & = ( m^2_{u_i } - m^2_{x})\left[1 - \frac{m^2_{\eta_i } - m^2_{s_i}}{m^2_{\eta_i } - m^2_{x}}\right ] \end{aligned}\ ] ] from which one can see if @xmath191 then we have the same result as in eqs .
( [ sunrise],[tadpole ] ) . to perform chiral - continuum extrapolation of lattice data for the @xmath192 meson with ginsparg - wilson quarks in a staggered sea the consistent numerical evaluation of self - energy integrals @xmath193 and @xmath194 needs to be done using the particular type of finite range regulator .
the extrapolation formula , in a numerical simulation with fixed strange quark mass , will have five coefficients , some of which are functions of the frr cutoff .
based on the recently developed mixed pqchpt for pseudoscalar meson sector we have extracted a chiral - continuum extrapolation formula for the vector meson masses at one - loop up to order @xmath0 .
instead of writing explicit expression for the vector meson lagrangian we have used general principles in order to calculate the mass corrections at one - loop order .
up to order @xmath10 , vector meson masses , in general , depend on six , presently unknown , parameters : @xmath176 , @xmath177 , @xmath166 and @xmath76 . the low energy coefficient
@xmath195 is model independent and may also be estimated from the extrapolation of pseudoscalar masses .
if one , however , is interested in vector meson mass corrections up to order @xmath196 then the number of parameters reduces by two : @xmath197 this can be used to test the theory predictions up to order @xmath196 .
it is also possible to reduce the number of extrapolation parameters by one if we fix the strange quark mass in the simulations . during the derivations
the following approximations were used : @xmath198 , @xmath172 and @xmath199 for the @xmath2 meson .
this approximation may be adjusted , for example , to the @xmath200 vector meson if we assume @xmath172 , @xmath199 and @xmath201 .
the idea beyond this is that in the continuum limit the masses of the valence quarks , @xmath202 and @xmath203 , are required to correspond to the masses of valence quarks from which the vector meson is constructed . in our calculations
we include ( phenomenologically ) the relativistic @xmath9 decay process , which although can not be consistently included in the framework of low energy effective field theory , is required to give consistent physical predictions .
finally , we did not take into account the finite volume effects from the lattice , which are already well studied in refs .
h.r.g . would like to thank jerry p. draayer for his support , the southeastern universities research association ( sura ) for a graduate fellowship and louisiana state university for a scholarship partially supporting his research .
he would also like to acknowledge s. vinitsky and g. pogosyan for support at the joint institute of nuclear research .
finally , this work was supported in part by doe contract de - ac05 - 84er40150 , under which sura operates jefferson laboratory .
o. br _ et al .
_ , phys .
d * 67 * , 114505 ( 2003 ) [ arxiv : hep - lat/0210050 ]
. o. br _ et al .
_ , phys .
d * 70 * , 034508 ( 2004 ) [ arxiv : hep - lat/0306021 ]
. o. br _ et al .
_ , phys .
d * 72 * , 054502 ( 2005 ) [ arxiv : hep - lat/0503009 ] . b. c. tiburzi , arxiv : hep - lat/0508019 . l. susskind , phys . rev .
d * 16 * , 3031 ( 1977 ) . c. w. bernard _ et al .
_ , phys .
d * 64 * , 054506 ( 2001 ) [ arxiv : hep - lat/0104002 ] .
p. h. ginsparg and k. g. wilson , phys .
d * 25 * , 2649 ( 1982 ) .
a. d. kennedy , nucl .
phys . proc .
* 140 * , 190 ( 2005 ) [ arxiv : hep - lat/0409167 ] .
b. bunk _ et al .
_ , nucl .
b * 697 * , 343 ( 2004 ) [ arxiv : hep - lat/0403022 ] .
w. j. lee and s. r. sharpe , phys .
d * 60 * , 114503 ( 1999 ) [ arxiv : hep - lat/9905023 ] .
r. s. van de water and s. r. sharpe , arxiv : hep - lat/0507012 . c. r. allton _ et al .
_ , phys .
b * 628 * , 125 ( 2005 ) [ arxiv : hep - lat/0504022 ] . c. aubin and c. bernard , phys .
d * 68 * , 034014 ( 2003 ) [ arxiv : hep - lat/0304014 ] . c. aubin and c. bernard , nucl .
suppl . * 129 * , 182 ( 2004 ) [ arxiv : hep - lat/0308036 ] .
s. r. sharpe and r. s. van de water , phys .
d * 71 * , 114505 ( 2005 ) [ arxiv : hep - lat/0409018 ] .
a. gray _ et al . _ [ hpqcd collaboration ] , arxiv : hep - lat/0507015 .
h. r. grigoryan and a. w. thomas , arxiv : hep - lat/0507028 , to appear in phys .
b. d. b. leinweber _ et al .
_ , phys .
d * 64 * , 094502 ( 2001 ) [ arxiv : hep - lat/0104013 ] .
k. symanzik , nucl .
b * 226 * , 187 ( 1983 ) .
k. symanzik , nucl .
b * 226 * , 205 ( 1983 ) .
a. morel , j. phys .
( france ) 48 , 1111 ( 1987 ) . c. w. bernard and m. f. l. golterman , phys
. rev .
d * 49 * , 486 ( 1994 ) [ arxiv : hep - lat/9306005 ] . c. k. chow and s. j. rey , nucl .
b * 528 * , 303 ( 1998 ) [ arxiv : hep - ph/9708432 ] .
m. luscher , phys .
b * 428 * , 342 ( 1998 ) [ arxiv : hep - lat/9802011 ] .
b. sheikholeslami and r. wohlert , nucl .
b * 259 * , 572 ( 1985 ) .
s. r. sharpe , nucl .
suppl . * 34 * , 403 ( 1994 ) [ arxiv : hep - lat/9312009 ] .
c. aubin _ et al . _
( milc ) , phys .
d * 70 * , 114501 ( 2004 ) [ arxiv : hep - lat/0407028 ] .
y. b. luo , phys .
d * 55 * , 353 ( 1997 ) [ arxiv : hep - lat/9604025 ] . c. arzt , phys .
b * 342 * , 189 ( 1995 ) [ arxiv : hep - ph/9304230 ] . c. aubin
_ et al . _ ,
d * 70 * , 094505 ( 2004 ) [ arxiv : hep - lat/0402030 ] . m. a. pichowsky
_ et al . _ ,
phys . rev .
d * 60 * , 054030 ( 1999 ) [ arxiv : nucl - th/9904079 ] .
j. f. donoghue _ et al .
_ , phys .
d * 59 * , 036002 ( 1999 ) [ arxiv : hep - ph/9804281 ] .
w. detmold _ et al .
_ pramana * 57 * , 251 ( 2001 ) [ arxiv : nucl - th/0104043 ] .
e. jenkins _ et al .
_ , phys .
* 75 * , 2272 ( 1995 ) [ arxiv : hep - ph/9506356 ] .
d. arndt and c. j. d. lin , phys .
d * 70 * , 014503 ( 2004 ) [ arxiv : hep - lat/0403012 ] .
f. bedaque _ et al .
d * 71 * , 054015 ( 2005 ) [ arxiv : hep - lat/0407009 ] .
b. orth _ et al .
d * 72 * , 014503 ( 2005 ) [ arxiv : hep - lat/0503016 ] .
g. colangelo , nucl .
* 140 * , 120 ( 2005 ) [ arxiv : hep - lat/0409111 ] . | 0.2 cm we consider partially quenched , mixed chiral perturbation theory with staggered sea and ginsparg - wilson valence quarks in order to extract a chiral - continuum extrapolation expression for the vector meson mass up to order @xmath0 , at one - loop level . based on general principles
, we accomplish the task without explicitly constructing a sophisticated , heavy vector meson chiral lagrangian . 0.2 cm |
we consider collective decision making processes such as a market that acts as a central mechanism for coordinating the actions of autonomous participants .
we address the questions : how does one measure the quality of the collective decision making process , and how weak can the central market mechanism be ?
in many applications , there is significant interest in decentralizing computation while still being able to arrive at results that can not be computed entirely locally .
we use a simple model to capture the informational complexity of computing global functions by aggregating results from participants who are endowed with arbitrary allotments of local information .
this allows us to draw conclusions about the requirements on allotments and protocols , for efficient collective information processing .
a key aspect of our model is the specification of meta - information based on distinguishing perfect information , single - blind arrangements , and double - blind arrangements .
our technical framework is built on the notions of communication complexity .
we assume that participants possess information which is not available to other participants ; we call this the _ private _ information .
this work is motivated by several applications .
a rather timely application is found in the domain of participants in electronic markets . often , such as in financial markets , participating agents would benefit from an understanding of the global system dynamics @xcite . for instance
, agents might like to have signals that indicate the presence of herding , bubbles and other aggregate phenomena .
typically , the local view of a single agent does not provide sufficient information to reliably detect this .
moreover , in such a domain , one is tightly constrained by what information can be revealed , incentives to reveal this information , and other aspects related to privacy in computation .
if we seek efficient decentralized information processing mechanisms under these constraints , then we would like to be able to determine what is or is not possible , employing only a coarse characterization of resources and endowment of information .
recent studies such as in anonymized financial chat rooms @xcite provide interesting insights into the behaviour of such collectives , such as the characterization of equilibria in which a subset of traders profit from the information of others .
this is but one example of a larger body of economic literature related to phenomena in networked markets @xcite .
however , in that literature , it is not typical to investigate our question of how the allotment structure and communication protocols relate to the efficiency with which specific types of computation are achieved .
for instance , change detection @xcite is of fundamental importance in financial markets how weak a protocol is sufficient to decide a change has occurred ?
recent work on the topic of complexity of financial computations by indicates that this is a fertile direction to pursue .
similar issues arise in many other application domains , such as mobile sensor networks and distributed robotics .
describe a mobile sensor network for optimal data gathering , using a combination of underwater and surface level sensing robots to optimally gather information such as chemical and thermal dynamics in a large volume of water ( typically measured in square miles ) .
similar systems have been utilized for tracking oil spills and algal blooms .
a key computational method utilized by such distributed robotic networks involves distributed optimization @xcite . the deployment of modules in such a network needs to satisfy a spatial coverage requirement , described as a cost function , so that each module plans trajectories to optimize this criterion .
the sensor fusion problem , to determine a combined estimate of an uncertain parameter , may also be posed an as optimization problem in the sense of maximizing information gain . despite this rigorous approach , relatively little is known about how to compare different formulations of these optimization problems given that we are interested in a certain type of global function ( say , number of peaks in a chemical concentration profile or some distributional aspect of the overall field ) using weak local sensing and the ability to move sensing nodes , how does one compare or otherwise characterize protocols and other aspects of the problem formulation ?
a line of work that begins to touch upon some of these questions is that of ghrist and collaborators .
use tools from algebraic topology to solve the problem of counting targets using very weak local information ( such as unlabelled counts in a local neighbourhood ) .
present an approach to detection of holes in coverage through decentralized computation of homology . here again , the focus being on aspects of the specific function being computed , the authors do not address the relationship between the protocols and problem formulation , and the efficiency of computation .
extending the idea of decentralized computation in social systems , consider the problem faced by a program committee , such as one that might review this paper .
we seek a decentralized computation of a ranking problem .
similar ranking problems also occur in executive decision making such as the hiring decision in academic departments .
the key issue here is that of parsimonious information sharing , coupled implicitly or explicitly with the meta - information problem .
these challenges arise due to limitations on the capacities of the decision makers to exchange information with each other .
the common theme underlying all of these applications is the computation of a function based on allotment of portions of the information to parties who have reasonable amounts of computational resource but would like to keep exchange of information limited .
we wish to understand how weak the corresponding protocols can be , for various types of functions .
major categories of functions of interest include change detection and ranking .
we model change detection by an abstract version of the key underlying problem , of determining whether the data forms a constant or a step function .
the main statistical property we consider here is the computation of the mean .
we are interested in understanding just how much communication must occur to answer various questions of interest .
therefore , instead of working with detailed models for the questions about market behaviour or sensor networks discussed previously , we have deliberately kept the models we study as _ simple _ as possible .
_ this makes our lower bound results stronger .
_ determining an answer to any more realistic question will require even more information to be exchanged than in these simple models , as long as the more realistic model includes the simpler problem at its core .
it therefore makes sense in our setting to study the simplest possible embodiment of each of the core problems . for the upper bounds , our results are a first step and will need to be extended to more realistic models .
our model is based on the notion of _ communication complexity _
@xcite , which has been highly influential in computer science .
a _ boolean _
function models yes / no decisions , by requiring that the function take either the value 0 or the value 1 . a quantity with value either 0 or 1 is known as a _ bit _ , and quantities that are drawn from a larger range of values can be expressed by using multiple bits ; a function that is defined over a domain containing @xmath0 different values is said to have an @xmath1-bit _ input_. say two players alice and bob wish to compute a boolean function @xmath2 on a @xmath3-bit input , but alice only has access to the first @xmath1 bits and bob to the other @xmath1 bits .
alice and bob are not _ computationally _ constrained , but they are _ informationally _ constrained .
the question now is : how many bits of information do alice and bob need to exchange to compute @xmath2 on a given @xmath3-bit input ? a _ protocol _ for this problem specifies , given the inputs to the players and the communication so far , which player is the next to send information , as well as what information is actually sent .
there is a trivial protocol where alice merely sends her part of the input to bob .
bob now has all the information he needs to compute @xmath2 , and he sends back the 1-bit answer to alice .
the _ cost _ of a protocol is the total number of bits that are exchanged ; this simple protocol has a cost of @xmath4 for _ any _ function .
the field of two - party communication complexity studies , for various functions @xmath2 of interest , whether more efficient protocols exist . as an example , for the equality function which tests whether alice and bob s inputs are exactly the same , it is known that the communication upper bound of @xmath4 is tight for _ deterministic _ protocols , but there is an improved protocol with cost @xmath5 when the players messages are allowed to be randomized and it is sufficient for the final answer to be correct with high probability .
the notion of communication complexity can be generalized from the two - party setting to the _ multi - party _ setting @xcite . here
the number of players is not limited to two , each player has some information about the global input , and they wish to compute some boolean function of the global input .
there are two standard models for how the input is distributed among the players : the number - in - hand model ( nih ) and the number - on - forehead ( nof ) model .
suppose there are @xmath6 players and the global input is @xmath7 bits long .
in the nih model , there is some fixed partition of the global input into @xmath6 parts , and each player gets one of these parts . in the nof model , again there is a fixed partition into @xmath6 parts , but the @xmath8-th player gets all the parts _ except _ the @xmath8-th part .
the main motivation for our model is that in many situations , such as financial markets or sensor networks , information is distributed among players in a more complex fashion than in the noh or nif models .
moreover , the players might not have control over which pieces of information they have access to the _ allotment _ of inputs to players might be arbitrary , perhaps even adversarial . as an example , creators of financial instruments may decide which assets to bundle into pools that are then offered for sale .
purchasers of such instruments might wish to check fairness of allocation , without revealing to each other their precise holdings ; or regulators might wish to check that sellers behaved impartially but without relying on full disclosure . yet , the players might still wish to compute some function of their global input in this less structured setting .
now different kinds of question arise than in the standard communication complexity setting . for a given function ,
which kinds of allotment structures allow for protocols ? does the meta - knowledge of what the allotment structure actually is make a difference to whether there is an efficient protocol or not ?
these questions are interesting even for simple functions which have been thoroughly investigated in the standard setting . to be more formal , let @xmath2 be a function which @xmath6 players wish to compute on a global input @xmath9 of size @xmath7 bits .
an @xmath10 allotment structure is a sequence of @xmath6 subsets @xmath11 of @xmath12 = \{1,2,\dots , n\}$ ] .
an _ allotment structure _ corresponds to an allotment of input bits among players in the following way : player @xmath8 receives all bits @xmath13 for @xmath14 .
note that unlike in the noh and nif models , this allotment of input bits is completely general it might be the case that two players receive the same set of bits , for example .
this is the main novelty of our approach .
our intention here is to model two kinds of situations . in the first , the players have little control over which pieces of information they can access
they have to do the best they can , with the available information . in the second ,
the allotment is made by a centralized authority , and it is of interest to study _ which _ allotment would most facilitate the computation in question . a @xmath6-player _ macroscope _ on @xmath7 bits is simply a function @xmath2 on @xmath7 input bits together with an @xmath10 allotment structure .
we will abuse notation and sometimes use a macroscope to refer to a _ sequence _ of functions @xmath15 , where each function @xmath16 depends on @xmath7 bits .
this will enable us to pose and study the question of the asymptotic efficiency of protocols for macroscopes .
( x1 ) at ( 0,0 ) @xmath17 ; ( x2 ) at ( 0,1 ) @xmath18 ; ( x3 ) at ( 0,2 ) @xmath19 ; ( x4 ) at ( 0,3 ) @xmath20 ; ( xdots ) at ( 0,4 ) @xmath21 ; ( xn ) at ( 0,5 ) @xmath22 ; ( s1 ) at ( 2,1 ) @xmath23 ; ( s2 ) at ( 2,2 ) @xmath24 ; ( sdots ) at ( 2,3 ) @xmath21 ; ( sk ) at ( 2,4 ) @xmath25 ; ( x1 ) ( s1.west ) ; ( x2 ) ( s1.west ) ; ( xn ) ( s1.west ) ; ( x1 ) ( s2.west ) ; ( x3 ) ( s2.west ) ; ( x4 ) ( s2.west ) ; ( xn ) ( s2.west ) ; ( x2 ) ( sk.west ) ; ( x3 ) ( sk.west ) ; ( x4 ) ( sk.west ) ; ( x ) at ( -0.5,2.5 ) @xmath9 ; ( 0,-0.1 ) ( 0,5.1 ) ; the generalized modelling of the allotment of inputs raises the issue of _ meta - information _ how much do players know about the allotment of inputs , and how can they take advantage of this ? in the case of the nih and nof models , the allotment is implicitly known to all players because it is fixed in advance .
however , in our setting , there are two different kinds of situations the _ single - blind _ situation and the _ double - blind _ situation . in a single - blind macroscope , all players know the allotment structure , however player @xmath8 does not know the values of any input bits apart from the ones whose indices are in @xmath26 . in a double - blind macroscope ,
the players are more hampered in that they do not even know the allotment structure , however they do know the indices of the bits they receive .
it remains to formally define what a _ protocol _
is in our model . to keep things simple
, we focus on _ simultaneous - message _ protocols , where each player broadcasts a sequence of bits to all players ; this is often presented figuratively as each player writing their bit string on a universally viewable blackboard .
a protocol _ solves _ a macroscope if each player can determine the value of the function on the global input simply by looking at its own input bits as well as the information written on the blackboard .
the cost of a protocol for a macroscope is then the total length of strings written by the players . in protocols for single - blind macroscopes ,
the message of player @xmath8 is a function of the values of bits whose indices are in @xmath26 as well as of the allotment structure . for double - blind macroscopes , the message of player
@xmath8 is a function only of the values of bits whose indices are in @xmath26 .
we make a deliberate choice in our modelling to be highly general in terms of the allotment structure , and to be specific in terms of the structure of the actual communication .
this is because our main aim is to understand the impact of the allotment structure on efficiency of communication .
our model can be extended to allow more degrees of freedom with regard to the communication structure .
one way in which this can be done is to allow multiple - round protocols , where players communicate in turns , with the protocol specifying whose turn it is to communicate .
another is by allowing randomness here each player
is assumed to have access to a private source of randomness , on which its message can depend .
a third way is to restrict communication to take place between specified pairs of players , i.e. , there is an implicit topology of communication .
this third approach is taken in the field of distributed algorithms @xcite , where however input allotment is not modelled in a flexible way .
we are interested in protocols which have communication as low as possible .
this is desirable not just in terms of efficiency , such as meeting bandwidth constraints , but also in terms of _ privacy_. in applications such as financial markets , the players would like to obtain some global knowledge without revealing their own inputs .
thus , player @xmath8 has more than one reason for not following the trivial protocol of publishing the values of all bits in @xmath26 .
the _ lower _ the communication , the _ less _ the information revealed about the values of bits held by individual players ; we will rely on this link between parsimony of communication and the weakness of the coordination mechanism .
privacy requirements are modelled more explicitly in sub - areas of cryptography such as secure multi - party computation @xcite .
we prefer not to model these requirements explicitly so as not to complicate our model too much .
we make no assumption about the relationship between the number of players and the number of bits in the global input . in an application such as sensor networks
, there might be few players ( sensors ) , each having a large amount of information , whereas in the financial markets application , there are typically many players each having few pieces of information .
our model deals equally well with both extremes .
a first observation is that to compute a non - trivial function over the global input , i.e. a function that depends on all the input bits , the allotment structure must satisfy the _ covering _ property each index @xmath27 $ ] lies in at least one set @xmath26 of the allotment structure . if the covering property did not hold , consider an index @xmath28 which does not belong to the allotment , and an input @xmath29 such that @xmath2 is _ sensitive _ to @xmath29 at index @xmath28 , meaning that @xmath30 is different from @xmath31 , where @xmath32 is @xmath29 with the value of the @xmath28th bit flipped . by the non - triviality of @xmath2 ,
such an input @xmath29 must exist .
clearly any protocol outputs the same answer for @xmath29 as for @xmath32 since @xmath28 does not belong to the allotment , and hence the protocol can not be correct . henceforth , we automatically assume that a macroscope has the covering property .
there are no general necessary conditions on the allotment structure beyond the covering property for computation of non - trivial functions .
but intuitively , the more `` even '' the allotment is , in the sense of each bit being allotted to the same number of players , the easier it is to compute a symmetric function of the inputs .
we define an _ even _
@xmath10 allotment structure as an allotment structure for which there is a number @xmath33 such that each index @xmath34 $ ] belongs to exactly @xmath33 distinct subsets @xmath26 , and each subset @xmath26 is of the same size .
clearly , for such an allotment structure , each set @xmath26 is of size @xmath35 .
our first results address the question of what we can say in general about the cost of single - blind and double - blind macroscopes .
[ gen - single - blind ] every single - blind macroscope on @xmath7 bits has a protocol with cost @xmath7 . moreover ,
this bound is optimal .
note that the upper bound _
does not _ depend on the number of players .
the proof of the upper bound in theorem [ gen - single - blind ] takes advantage of the global knowledge the players have about the allotment structure .
consider a protocol in which each input bit @xmath36 has a player `` responsible '' for it
player @xmath28 is responsible for input bit @xmath36 iff @xmath37 and @xmath38 for @xmath39 .
it follows from the covering property that each input bit has a player responsible for it .
it should also be clear that at most one player is responsible for any given input bit .
the protocol consists of players sending the values of all the bits they are responsible for .
since the macroscope is single - blind , each player knows who is responsible for which input bits .
hence each player can reconstruct the input from the information sent in the protocol , and therefore also compute the function on the input .
the cost of the protocol is @xmath7 , since each bit has exactly one player responsible for it . to see that this bound is optimal ,
consider the macroscope consisting of the parity function together with an @xmath40 allotment structure which allots each input bit to a distinct player .
suppose there is a protocol for this macroscope where one of the players does not send a message .
assume , without loss of generality , that this player holds the @xmath8-th bit .
then the parity function can not depend on the @xmath8-th bit , which is a contradiction .
[ gen - double - blind ] every @xmath6-player double - blind macroscope on @xmath7 bits has a protocol with cost at most @xmath41 .
the protocol giving the upper bound of theorem [ gen - double - blind ] is a simple one .
the @xmath28-th player sends an @xmath7-bit string specifying its allotment @xmath42 , as well as at most @xmath7 bits which specify the values of the bits whose indices are in the allotment @xmath42 .
once this information is made public by each player , the players can each reconstruct the input and hence compute the function on that input .
we do not know whether the bound in theorem [ gen - double - blind ] is optimal in general , but we believe this to be the case for protocols that use only one round of communication .
our primary focus is on studying the complexity of macroscopes for problems which arise in contexts such as electronic markets and distributed sensor networks .
for each of these problems , we are interested in issues such as the optimal cost of solving a macroscope , the differential cost of meta - information ( the reduction in cost when using a single - blind macroscope rather than a double - blind macroscope ) , and for single - blind macroscopes , the dependency of the cost on the allotment structure .
a fundamental problem in the context of electronic markets is the constancy problem of detecting whether a given function is constant or not .
we model this problem as a boolean function on @xmath43^n$ ] , which is 1 if all the inputs are equal , and 0 otherwise .
this requires a slight adaptation of our model to inputs which are @xmath44-ary rather than binary , but this adaptation can be done in a natural way .
we are able to characterize the cost of protocols for single - blind constancy macroscopes _ optimally _ in terms of the allotment structure . given an @xmath10 allotment structure , we define the intersection graph of the structure as follows .
the graph has @xmath7 vertices , and there is an edge between vertex @xmath8 and vertex @xmath28 for @xmath45 $ ] iff @xmath46 .
[ single - blind - constancy ] every @xmath6-player single - blind constancy macroscope on @xmath7 @xmath44-ary inputs can be solved with cost @xmath47 , where @xmath48 is the number of connected components of the intersection graph of the allotment structure associated with the macroscope . moreover ,
this bound is optimal to a constant factor .
note that the bound does not actually depend on @xmath7 , the number of inputs !
since the macroscope is single - blind , each player knows the identities of all the other players in the same connected component in the intersection graph of the allotment structure .
this is because these identities depend only on the allotment structure and not on the input .
the protocol is as follows : for each connected component , it is only the player with the smallest index in that connected component who sends a `` long '' message of @xmath49 bits long , specifying a value in @xmath43 $ ] that occurs in its portion of the input .
in addition , each player sends a 1-bit message saying whether its portion of the input is constant or not .
the players know that constancy holds if each 1-bit message encodes `` yes '' , and in addition , the values in @xmath43 $ ] sent in the long messages are all the same . indeed , if the function is constant , it is clear that all the 1-bit messages are `` yes '' , and that the values in the long message are the same . to argue the converse , just notice that if a fixed player in a connected component of the intersection graph sees a constant value @xmath50 $ ] , and if all other players in the connected component see a constant value , it follows that all players in the connected component see the same constant value @xmath51 .
we argue that this bound is optimal up to a factor of 2 .
we will give separate arguments that @xmath6 bits of communication are required and that @xmath52 bits of communication are required . from these separate bounds , it follows that @xmath53 bits of communication are required . to see that @xmath6 bits of communication are necessary , consider any allotment structure such that @xmath54 is non - empty for each player @xmath8 .
define a function @xmath55 \rightarrow [ n]$ ] such that @xmath56 for each player @xmath8 .
suppose there is a player @xmath8 who does not send a message .
then the protocol gives the same answer on both the all 1 s input and the input that is all 1 except at @xmath57 , since the communication pattern is the same for both of these inputs .
however the constancy function differs on these inputs .
next we show the @xmath52 lower bound .
consider any allotment structure whose intersection graph contains @xmath48 connected components .
suppose that fewer than @xmath52 bits of communication are sufficient for solving the macroscope on this allotment structure . then there is some connected component @xmath33 of the intersection graph such that the corresponding players send less than @xmath58 bits of communication in all .
this implies that there are two values @xmath59 $ ] such that the communication pattern of players in @xmath33 is exactly the same when the players in @xmath33 all receive the input @xmath60 as when they all receive the input @xmath61 .
now consider two inputs
input @xmath9 in which all co - ordinates are the constant @xmath60 and input @xmath62 in which all co - ordinates outside @xmath33 have value @xmath60 and co - ordinates in @xmath33 have value @xmath61 .
the communication pattern of the protocol is the same for @xmath9 and @xmath62 , however the constancy function is true for @xmath9 and false for @xmath62 .
this is a contradiction .
thus , for single - blind constancy macroscopes , the critical property of the allotment structure is the number of connected components of the intersection graph .
the fewer the number of connected components , the more efficiently the macroscope can be solved .
we next study the situation for double - blind macroscopes .
[ double - blind - constancy ] every @xmath6-player double - blind constancy macroscope on @xmath7 @xmath44-ary inputs can be solved with cost @xmath63 . moreover , there are @xmath6-player double - blind constancy macroscopes which require cost @xmath64 .
the protocol giving the upper bound is simple .
each player sends a message encoding one of @xmath65 possibilities : either the players portion of the input is non - constant , or if it is constant , which of the @xmath44 possible values it is .
the protocol accepts if each message encodes the same value @xmath66 $ ] . for the lower bound ,
since each player is unaware of players allotments other than its own , the lower bound of @xmath67 in the proof of theorem [ single - blind - constancy ] holds with the maximum possible value of @xmath48 , namely @xmath68 .
thus , in the case of the constancy function , the differential cost of meta - information can be quite significant , especially when the intersection graph of the allotment structure is connected .
next we consider a formalization of the change detection problem . the boolean step function ( bsf ) problem
is defined as follows : a string @xmath69 evaluates to 1 if there is an index @xmath8 such that @xmath70 for @xmath71 and @xmath72 for @xmath73 , or to 0 otherwise .
= [ shape = circle , fill , inner sep=0pt , minimum size=4pt ] ( 0.7,1 ) ( 8.3,1 ) ; at ( 1,0.75 ) @xmath17 ; at ( 2,0.75 ) @xmath18 ; at ( 3,0.75 ) @xmath19 ; at ( 4,0.75 ) @xmath20 ; at ( 5,0.75 ) @xmath74 ; at ( 6,0.75 ) @xmath75 ; at ( 7,0.75 ) @xmath76 ; at ( 8,0.75 ) @xmath22 ; at ( 1,1 ) ; at ( 2,1 ) ; at ( 3,1 ) ; at ( 4,2 ) ; at ( 5,2 ) ; at ( 6,2 ) @xmath75 ; at ( 7,2 ) ; at ( 8,2 ) ; possibilities for different structures of allotment lead to a twofold challenge in a bsf macroscope .
first , no party may see the step , so parties need to share some information about the values they see .
second , if several parties detect a step , they then need to determine whether they are observing the same step .
= [ shape = circle , fill , inner sep=0pt , minimum size=4pt ] ( 0.7,1 ) ( 9.3,1 ) ; at ( 1,0.75 ) @xmath17 ; at ( 2,0.75 ) @xmath18 ; at ( 3,0.75 ) @xmath19 ; at ( 4,0.75 ) @xmath20 ; at ( 5,0.75 ) @xmath74 ; at ( 6,0.75 ) @xmath75 ; at ( 7,0.75 ) @xmath77 ; at ( 8,0.75 ) @xmath76 ; at ( 9,0.75 ) @xmath22 ; at ( 1,1 ) ; at ( 2,1 ) ; at ( 3,1 ) ; at ( 4,2 ) ; at ( 5,2 ) ; at ( 6,2 ) @xmath75 ; at ( 7,2 ) ; at ( 8,1 ) ; at ( 9,2 ) ; [ double - blind - bsf ] every @xmath6-player double - blind bsf macroscope on @xmath7 bits can solved with cost @xmath78 .
the protocol is as follows : for each @xmath79 $ ] , player @xmath8 sends two indices @xmath80 and
@xmath81 , where @xmath51 is the largest index in @xmath26 for which @xmath82 and @xmath83 is the smallest index in @xmath26 for which @xmath84 .
given all these messages , each player can calculate the value of the smallest index @xmath83 for which @xmath84 simply by taking the minimum of @xmath81 over all players @xmath8 , as well as the largest index @xmath51 for which @xmath82 , simply by taking the maximum of @xmath80 over all players @xmath8 .
note that @xmath85 iff @xmath86 .
we conjecture that the bound of theorem [ double - blind - bsf ] is tight for double - blind bsf macroscopes for protocols that use only one round of communication . for single - blind macroscopes , however , we can do better .
[ single - blind - bsf ] every @xmath6-player single - blind bsf macroscope on @xmath7 bits can be solved with cost @xmath87 . the protocol witnessing the upper bound is as follows .
each player sends a message consisting of two parts .
the first part is 2 bits long , and specifies which of the following is the case : ( 1 ) the player s portion of the input is constant , ( 2 ) there is a single transition from 0 to 1 in the player s input , ( 3 ) neither ( 1 ) nor ( 2 ) holds . the second part is @xmath88 bits long .
the interpretation of the second part of the message is as follows : if case ( 1 ) holds for the first part , then the second part encodes which constant ( either 0 or 1 ) the player is given .
if case ( 2 ) holds , then the second part encodes the index at which a transition occurs , i.e. , a number @xmath27 $ ] such that @xmath14 ( assuming that the player in question is player @xmath8 ) and such that for all @xmath89 , @xmath90 and @xmath91 .
if case ( 3 ) holds , the contents of the second part of the message are irrelevant . from the messages , the players can either reconstruct the input @xmath9 , if case ( 1 ) or ( 2 ) holds for each player , or conclude directly that @xmath85 , if case ( 3 ) holds for any player . from the input @xmath9 ,
each player can compute @xmath92 on its own . in this case , too , the advantage of using single - blind macroscopes can be seen , though a matching lower bound in theorem [ double - blind - bsf ] is needed to prove this .
next , we attempt to model the _ averaging _ function
. distributional statistics of different kinds need to be computed in a decentralized way in various contexts such as sensor networks . to model this
, we again depart from the framework of bit strings as inputs .
the input is now a sequence of real numbers @xmath93 , @xmath94 $ ] for each @xmath95 . given a parameter @xmath96 , we study the cost of protocols for @xmath97-averaging macroscopes , where each player needs to arrive at an @xmath98-additive approximation to the average of the numbers @xmath99 by using the protocol to communicate . [ single - blind - averaging ] let @xmath96 be fixed .
every @xmath6-player single - blind @xmath97-averaging macroscope on @xmath7 inputs can be solved with cost @xmath100 .
notice again that there is no dependence of the cost on @xmath7 , merely on the number of players and the approximation error .
we again use critically the meta - information of players about the allotment structure .
each player @xmath8 knows , for each index @xmath14 , the number @xmath101 of distinct players receiving input @xmath13 .
the message sent by player @xmath8 is an @xmath102-additive approximation to the quantity @xmath103 .
since the quantity is between @xmath104 and @xmath105 , the approximation can be specified using @xmath106 bits . given the messages of all players ,
each player can compute an @xmath98-approximation to the average simply by summing all the individual approximation .
since the individual approximations are @xmath102-additive approximations , the sum will be an @xmath98-approximation to the average .
in the case of averaging macroscopes , we _ can _ show that the double - blind restriction is a significant one , in that it leads to a dependence of the cost on the number of players .
the proof uses a dimensionality argument .
[ double - blind - averaging ] let @xmath96 be a parameter .
there are @xmath107-player double - blind @xmath98-averaging macroscopes on @xmath7 inputs which require cost @xmath108 .
consider a @xmath107-player double - blind macroscope for @xmath97-averaging , with player 1 receiving an allotment @xmath109 $ ] and player 2 receiving an allotment @xmath110 $ ] .
our assumption that the allotment structure is covering implies @xmath111 $ ] , but neither player knows anything about the other player s allotment beyond this fact .
we show that a player s message must essentially reveal its input .
consider player 1 , and let @xmath112 . now for each @xmath113 , consider @xmath114 \setminus \{i\}$ ] .
if player 1 has allotment @xmath115 and player 2 has allotment @xmath116 , then in a correct protocol , player 2 eventually knows an @xmath98-approximation to @xmath117 } x_j / n$ ] .
since it also knows all values except @xmath99 , this implies that it knows an @xmath118-approximation to @xmath99 .
thus an @xmath118-approximation to @xmath99 should be extractable from player 1 s message .
this should hold for every @xmath8 , which means player 1 must send at least @xmath119 bits . by a symmetric argument ,
player 2 must send at least @xmath120 bits , which means that at least @xmath108 bits are communicated in all .
theorem [ double - blind - averaging ] has the disadvantage that it does not say much about @xmath98 , which may be large in comparison to @xmath7 .
however , we believe that a refinement of the proof which argues about information revealed about subsets of inputs rather than individual inputs can be used to establish an improved lower bound .
the previous two sections have focused on the details of a model for collective information processing and properties of this model .
we now take a step back to discuss why these results are of relevance to the motivating examples identified in the introduction .
consider the problem of agents in financial markets .
it is increasingly the case that , with the emergence of diverse communication methods and agents deciding at time scales ranging from microseconds to days , common knowledge is nt so commonly available in practice .
this has significant implications for dynamics @xcite and much of the modern discussion regarding markets is related to such issues @xcite . in this
setting , there is a need for diagnostic tools that could provide useful signals by computing global properties , i.e. functions , based on local information that can be used subject to limitations on protocols .
for instance , has there been a change from a ` constant ' level in a global sense ? or , is there a significant difference in the form of a step change between segments of a networked market ?
we illustrate the use of concepts from complexity theory to address abstract versions of such questions .
indeed , our techniques could be used to answer further such questions about statistical distributions , ranking queries , etc .
a key novelty , even in comparison to the state of the art in communication complexity , is that we consider an arbitrary endowment of inputs and meta - information such as single and double blind protocols .
an important general direction for future work in this area would be to extend our analysis to more directly address the subtleties of the above mentioned dynamics as they occur in applications of interest .
also , we would like to better understand the relationships between our model of decentralized computation under informational constraints and previous established models , such as @xcite , which employ different methods and focus more on temporally extended sequential protocols such as auctions . in terms of more specific questions , our model could be extended in various ways .
we have considered the case of a _ simple _ structure of communication , with simultaneous messages sent in one round of communication , and a possibly complex allotment structure of the inputs .
we could allow more sophisticated communication structure .
for example , double - blind protocols with two rounds of communication can emulate single - round single - blind protocols with some loss in efficiency , simply by using the first round to share publicly information about the input allotment , and then running the single - blind protocol in the second round .
more generally , communication might be restricted to occur only between specific _ pairs _ of players . in the context of sensor networks , for instance
, it is natural to model both the location of information and the structure of communication as governed by the topology of the ambient space .
even studying very simple functions such as constancy in such general models appears to be interesting .
we emphasize that we are interested in understanding the communication requirements of even very simple functions in modelling frameworks that render them non - trivial .
this distinguishes our work from the existing research on communication complexity , where functions such as constancy and bsf are trivial because the model of communication is so simple .
we are especially interested in modelling some aspects of collective information processing , such as information overlap and meta - information , which have been neglected so far and which we believe can have a significant impact on the efficiency of communication .
the way we capture these notions is quite flexible and can be used both to model computation of continuous quantities such as average and computation of discrete quantities such as boolean functions , as we have illustrated with our results .
another direction that we find compelling is modelling meta - information in a more sophisticated way . as of now
, we have the single - blind model and the double - blind model .
but there are various intermediate notions that are reasonable to study . for example
, each player might know the number of players and the number of inputs but nothing about which inputs are given to which other players . or a player might know which other players also receive the inputs it receives , but nothing about inputs it does not receive . or some global property of the allotment structure , such as that the allotment structure is even , might be known . notice that in the case of averaging macroscopes , there _ is _ an efficient protocol whose cost does nt depend on the number of players if the allotment structure is even and the size of each allotment is known .
the protocol simply involves the players summing all their inputs and dividing by a universal constant . in general
, one can ask : assuming that there are efficient single - blind protocols known , what is the minimal information about the allotment structure required to give efficient protocols ?
ranking is an important subject that we have not addressed in this paper .
some interesting problems can be captured as ranking macroscopes , and we leave their study for further work . we have seen that in some cases more information hinders making a global decision , rather than helping . more generally , why is allotment structure important ? we are striving to fully understand how the allotment structure of information affects our ability to efficiently answer questions that require global information .
s.r . would like to acknowledge the support of the uk engineering and physical sciences research council ( grant number ep / h012338/1 ) and the european commission ( tomsy grant agreement 270436 , under fp7-ict-2009.2.1 call 6 ) .
a.s . acknowledges the support of epsrc via platform grant ep / f028288/1 .
r.s . acknowledges the support of epsrc via grant ep / h05068x/1 .
arora , s. , barak , b. , brunnermeier , m. ge , r. 2011 , ` computational complexity and information asymmetry in financial products ' , _ communications of the acm _ * 54*(5 ) , 101107 .
http://dx.doi.org/10.1145/1941487.1941511[doi:10.1145/1941487.1941511 ] chandra , a. k. , furst , m. l. lipton , r. j. 1983 , multi - party protocols , _ in _ ` stoc 1983 : proceedings of the fifteenth annual acm symposium on theory of computing ' , acm , pp .
http://dx.doi.org/10.1145/800061.808737[doi:10.1145/800061.808737 ] darley , v. outkin , a. v. 2007 , _ a nasdaq market simulation : insights on a major market from the science of complex adaptive systems _ , vol . 1 of _ complex systems and interdisciplinary science _ , world scientific
. leonard , n. , paley , d. , lekien , f. , sepulchre , r. , fratantoni , d. davis , r. 2007 , ` collective motion , sensor networks , and ocean sampling ' , _ proceedings of the ieee _ * 95*(1 ) , 4874 .
http://dx.doi.org/10.1109/jproc.2006.887295[doi:10.1109/jproc.2006.887295 ] yao , a. 1982 , protocols for secure computations , _ in _ ` focs 1982 : proceedings of 23rd annual symposium on foundations of computer science ' , ieee computer society , pp . | we introduce a new model of collective decision making , when a global decision needs to be made but the parties only possess partial information , and are unwilling ( or unable ) to first create a global composite of their local views .
our macroscope model captures two key features of many real - world problems : allotment structure ( how access to local information is apportioned between parties , including overlaps between the parties ) and the possible presence of meta - information ( what each party knows about the allotment structure of the overall problem ) . using the framework of communication complexity , we formalize the efficient solution of a macroscope .
we present general results about the macroscope model , and also results that abstract the essential computational operations underpinning practical applications , including in financial markets and decentralized sensor networks .
we illustrate the computational problem inherent in real - world collective decision making processes using results for specific functions , involving detecting a change in state ( constant and step functions ) , and computing statistical properties ( the mean ) . |
abundance determinations of stars in galactic globular clusters can provide valuable information about important astrophysical processes such as stellar evolution , stellar structure , galactic chemical evolution and the formation of the milky way . with the advent of efficient high resolution spectrographs on 8 - 10 m telescopes , it is now possible to reach at least the luminous rgb stars in even the most distant galactic globular clusters ( gcs ) . in our previous work in this area
we have explored the abundances for large samples of stars in the canonical relatively nearby gcs m71 @xcite , m5 @xcite m3 , and m13 @xcite , as well as in pal 12 @xcite , a cluster associated with the sgr dwarf galaxy @xcite . for the case of pal 12
, in addition to the difference in age of several gyr , it being younger than the bulk of the galactic gcs @xcite , we found evidence for substantive differences between the chemical history of pal 12 and that of the `` normal '' halo gcs of similar [ fe / h ] is given by @xmath4 on a scale where @xmath5 h atoms . then [ x / h ] = log[n(x)/n(h ) ] @xmath6 log[n(x)/n(h ) ] , and similarly for [ x / fe ] . ] . in the present work ,
we study the outer halo gc ngc 7492 , a cluster which is not suspected ( yet ) of being part of any known stream or otherwise abnormal .
we compare the abundance ratios deduced from high resolution , high signal - to - noise ratio for four giants in this distant outer halo gc with those from our recent analysis of a large sample of stars in the relatively nearby and well studied gcs m3 and m13 , which have metallicities close to that of ngc 7492 .
we look for evidence in the deduced abundance ratios of some difference in the formation mechanisms or chemical history of gcs in the `` normal '' outer halo .
there has been no previous high dispersion study of ngc 7492 .
@xcite obtained [ fe / h ] = @xmath7 dex using their narrow band q39 photometric system , @xcite derived @xmath8 dex from the @xmath9 method applied to two rr lyrae variables in the cluster , while moderate dispersion spectroscopy in the region of the infrared ca triplet by @xcite gave [ fe / h ] = @xmath10 dex .
given the large distance of ngc 7492 , we select the brightest possible stars on the upper rgb for observation , making no attempt to reach fainter luminosities . these stars were picked from the photometric study of this cluster by @xcite . since this is a rather sparse cluster , these stars , which are the four brightest suspected members ,
lie distributed along the upper rgb , not concentrated at the rgb tip .
the positions of these stars in a @xmath12 cmd are illustrated in figure [ figure_cmd ] superposed on the predicted cluster isochrone from the @xcite evolutionary tracks ; see figure 10 of @xcite for a @xmath13 cmd of ngc 7492 . although the cluster is distant , there are so few luminous giants that they must be observed as single stars ; two of them can not be fit into a single 7 arcsec long slit .
spectra were obtained with hires @xcite at the keck observatory 20 - 22 august 2003 .
the instrument configuration covered the range 4650 to 7010 , with small gaps between the orders at the red end .
this is the `` yellow '' configuration described in @xcite .
these data were reduced using a combination of figaro @xcite scripts and the software package makee .
the desired minimum snr was 90 over a 4 pixel resolution element for a wavelength near the center of the hires detector .
this is calculated strictly from the counts in the object spectrum , and excludes noise from cosmic ray hits , sky subtraction , flattening problems , etc . since the nights were relatively dark , sky subtraction is not an issue except at the specific wavelengths corresponding to strong night sky emission lines , such as the na d doublet .
the seeing was extremely good during these nights , making the exposures shorter than normal , and enabling us to reach this goal for the stars reported here .
table [ table_sample ] gives details of the hires exposures for each star , with the total exposure time for each object .
all long integrations were broken up into separate exposures , each 1200 sec long , to optimize cosmic ray removal .
the last column of the table gives the heliocentric radial velocity for each star , measured from the hires spectra ; see @xcite for the details of the procedure used to determine @xmath14 . the radial velocity of ngc 7492is large and negative , and the cluster abundance is low .
it was easy to tell after one integration whether or not a star is a member of the cluster .
approximate measurements of @xmath14 were made on line ; all stars attempted turned out to be radial velocity members of ngc 7492 .
the four rgb stars have a mean @xmath14 of @xmath15 km s@xmath16 , agreeing well within the errors with the early measurement of @xcite of a single bright cluster member , but somewhat larger than the value of @xmath17 km s@xmath16 found from moderate resolution spectra by @xcite .
our more accurate @xmath14 for ngc 7492 should be used in future computations of the mass of the galaxy which rely on the orbits of its outlying satellites .
the velocity dispersion from the four members , with no correction for an instrumental contribution , is @xmath18 = 1.2@xmath191 km s@xmath16 , reflecting the low mass of this sparse cluster ; the observed @xmath18 is only slightly larger than the expected instrumental uncertainties .
we follow the procedures developed in our earlier work on globular cluster stars and described in @xcite to determine the stellar parameters for the four rgb stars in ngc 7492 .
@xmath11 is derived by comparing reddening - corrected broad band colors with the predictions of grids of model atmospheres .
we utilize here the grid of predicted broad band colors and bolometric corrections of @xcite based on the marcs stellar atmosphere code of @xcite .
we normally utilize @xmath20 and @xmath21 colors to determine @xmath11 .
the infrared colors were taken from 2mass @xcite .
@xcite provided @xmath22 photographic photometry over a large field including the entire cluster of ngc 7492 .
ccd photometry in @xmath13 was presented by @xcite , but their field is smaller than the cluster and the calibration of their photometry is not secure .
@xcite , in the course of a study of blue stragglers in this cluster , obtained ccd @xmath13 photometry for a large sample of stars , but the data tables were never published and have subsequently been lost ( ct , private communication , 2004 ) .
the two published photometric studies of ngc 7492 give @xmath23 mags which are inconsistent by several tenths of a mag ; the differences are not just a simple offset .
we therefore measured @xmath12 for our sample stars from andicam images taken for this purpose on aug . 1 , 2004 with the 1.3 m telescope at ctio operated by the smarts consortium ;
these values are given in table [ table_sample ] .
andicam is a dual channel cammera constructed by the ohio state university instrument group , but only the optical channel was used .
our andicam program requires photometric conditions , and additional standard star fields , charged to our andicam allocation through noao , are always taken for us .
we adopt a distance for ngc 7492 of 26.2 kpc @xcite ( as compared to 7.5 kpc for m13 and 10.4 kpc for m3 ) , with a reddening of e(b v ) = 0.036 mag from @xcite . the relative extinction in various passbands
is taken from @xcite ( see also * ? ? ?
the adopted stellar parameters are given in table [ table_teff ] .
to the maximum extent possible , the atomic data and the analysis procedures used here are identical to those we used in our recently completed analysis of a large sample of stars in m3 and m13 @xcite .
in particular , see 3 of that paper for a description of the measurement of the equivalent widths ( listed for the four stars in ngc 7492 in table [ table_eqw ] ) , 4 for a discussion of the atomic parameters , 4.2 for our adopted solar abundances ( tabulated in table 2 of our earlier paper , and repeated as table [ table_sun ] here ) , and 6 for a description of our abundance analysis procedures . as in our earlier work , the abundance analysis is carried out using a current version of the lte spectral synthesis program moog @xcite .
we employ the grid of stellar atmospheres from @xcite without convective overshoot , when available .
the template file of suitable unblended lines with their adopted atomic parameters is identical to that we used in our recently completed analysis of m3 and of m13 @xcite .
three of the four stars gave @xmath24 km s@xmath16 based on deriving a uniform fe abundance as a function of @xmath25 for the large set of fe i lines .
the fourth star ( star r ) gave 1.8 km s@xmath16 ; this value was also set to 2.0 km s@xmath16 .
lines with @xmath25@xmath26 175 mwere ignored , except for the ba ii lines in the coolest star in our sample in ngc 7492 . the resulting abundance ratios for the four rgb stars in ngc 7492 are given in tables [ table_abund_a ] to [ table_abund_e ] .
species with only one detected line are assigned an uncertainty of 0.10 dex .
table 7 of @xcite indicates the changes in derived abundance ratios for small changes in the adopted stellar parameters , the [ fe / h ] for the adopted model atmosphere , or the set of @xmath25for the lines of each species , and is appropriate for use here as well .
the mean abundance and 1@xmath18 variance for the species observed in ngc 7492 are listed in table [ table_abundsig ] .
the oxygen abundance is derived from the forbidden lines at 6300 and 6363 .
the subtraction of the night sky emission lines for the forbidden lines was reasonably straightforward given that the radial velocity of ngc 7492 is sufficiently different from 0 km s@xmath16 that their @xmath25 can be reliably measured .
the c / o ratio was assumed to be solar .
cn and ni i contamination is negligable ( see * ? ? ?
[ o / fe ] is given with respect to fe ii ; abundance ratios for all other elements are given with respect to fe i. the deduced mean [ fe / h ] of ngc 7492 value is in good agreement with that inferred by @xcite from moderate dispersion spectra in the region of the infrared ca triplet .
the na abundance was obtained from the 5680 doublet for all four stars .
we have , as in our previous papers , not used any non - lte corrections for na .
calculations by @xcite suggest values of between 0.1 and 0.2 dex are appropriate for our sample in ngc 7492 , with the coolest star having the largest value .
calculations by @xcite suggest somewhat smaller values .
the ba abundance of star 950 in ngc 7492 appears to be @xmath270.2 dex larger than that of the other cluster members studied here .
however , the detected lines of ba ii are all within the range where substantial hfs corrections occur .
table 7 of @xcite shows the very high sensitivity of the deduced ba abundances to small uncertainties in the equivalent width and microturbulence of the ba ii lines we use . on the other hand , the deduced [ y / fe ]
is also somewhat high for this star , while the deduced fe is the lowest of all the stars in our sample .
perhaps a slight adjustment of @xmath11 for this star is required . at this point
, we assume this is the result of observational and modelling uncertainties and does not indicate a real spread in [ ba / fe ] within ngc 7492 , but further verification of this is desirable .
the abundances of the elements with respect to fe , [ x / fe ] , as a function of @xmath11 are shown in figure [ figure_o_si ] , covering o , na , mg and si , figure [ figure_ca_v ] , which includes ca , sc , ti and v , figure [ figure_cr_ni ] , which includes cr , mn , co and ni , figure [ figure_cu_zr ] , which includes cu , zn , y and zr , and figure [ figure_ba_dy ] , for ba , la , nd , eu and dy . note the apparent star - to - star variation in [ o / fe ] and in [ na / fe ] , which becomes undetectably small , if it exists at all , for the elements heavier than na .
the scatter for [ ca / fe ] and for [ ni / fe ] , with 8 to 13 detected absorption lines in each star , is remarkably small , @xmath28 dex over the four star sample in ngc 7492 .
to check for the presence of star - to - star variations in abundance ratios within the small sample of rgb stars in ngc 7492 , we use a parameter we call the `` spread ratio '' ( @xmath29 ) .
the numerator of @xmath29 is the 1@xmath18 rms variance for the sample of four stars in ngc 7492 about the mean abundance for each atomic species ( @xmath30 ) with detected absorption lines , denoted @xmath18 ; the relevant values are given in the first three columns of table [ table_abundsig ] .
the denominator of @xmath29 is the total expected uncertainty , @xmath31 , which is the sum in quadrature of the known contributing terms .
included are a term corresponding to an uncertainty of 50 k in @xmath11 , the same for an uncertainty of 0.2 dex in log(@xmath32 ) , and for an uncertainty of 0.2 km s@xmath16 in @xmath33 , and the observed uncertainty [ @xmath34 , the parameter @xmath35 , which is calculated from data given in tables [ table_abund_a ] to [ table_abund_e ] , is taken as the variance about the mean abundance for a given species in a given star , i.e. the 1@xmath18 rms value about the mean abundance of species @xmath30 in a given star/@xmath36 , where @xmath37 is the number of observed lines of species @xmath30 .
it includes contributions from errors in the measured @xmath25 , random errors ( i.e. between lines of a given species ) in the adopted @xmath38 values , etc . some species , an example being fe i with its very large value of @xmath37 , have unrealistically small values of @xmath35 ; we adopt a minimum of 0.05 dex for this parameter .
the ratio @xmath39 is an indication of whether there is any intrinsic star - to - star variation in [ x / fe ] .
a high value of this `` spread ratio '' , tabulated in the fifth column of this table , suggests a high probability of intrinsic scatter for the abundance of the species @xmath30 .
ideally the mean @xmath29 for those elements with no star - to - star variation should be unity ; for many species the measured @xmath29 is close to that value , certainly closer here than for the sample in m3 and in m13 we studied earlier @xcite .
inspection of table [ table_abundsig ] shows that for all but two species @xmath40 for the sample of four stars in ngc 7492 , indicating little sign of an intrinsic star - to - star range in abundance .
o i and na i , however , have the two largest values of @xmath29 , 1.3 and 3.0 respectively . note that @xmath29 for mg i is 0.5 , suggesting no real star - to - star abundance variations for this element .
we therefore assume that the range of abundances seen in our sample of rgb stars in ngc 7492for na i and o i represent real star - to - star abundance variations ; while no other element shows definite evidence for such variations from this simple analysis .
c , n , o , na , mg , and al are known to show correlated abundance variations from star - to - star among the most luminous stars in globular clusters ; see , e.g. the review of @xcite .
our simple spread ratio analysis ( see [ section_spread ] ) shows definite star - to - star variations in abundance of both o and na in our small sample of rgb stars in ngc 7492 .
variations in mg , if present are smaller and subtle .
it is well established that o and na are anti - correlated among luminous giants in globular clusters , see , e.g. @xcite .
furthermore , @xcite compiled the data from the literature , combined it with their own , and showed that the same linear relation can be used to fit the o and na data for all globular clusters studied in detail thus far .
the latest addition to the clusters studied in detail , ngc 2808 , by @xcite , does so as well .
figure [ figure_ona ] shows the relationship between na and o abundances ( both with respect to fe ) for our sample in ngc 7492 .
also superposed is the line representing the fit for this anti - correlation determined by @xcite for the luminous giants in m3 , shifted by @xmath60.07 dex and + 0.1 dex in the vertical and horizontal axis as compared to the relation we found for m13 @xcite .
the first and last quartiles of the o na anti - correlation seen by @xcite in their sample of luminous giants in m3 are indicated .
there is a reasonably clear anti - correlation which corresponds well with that seen for luminous giants in other well studied galactic gcs ( see , e.g. the compilation of * ? ? ?
a similar correlation is detected in the outer halo gc ngc 7006 by @xcite .
there is marginally statistically significant evidence for a correlation between [ na / fe ] and [ mg / fe ] for our small sample in ngc 7492 , similar to that shown in other gcs ( see , for example , figures 12 and 13 of * ? ? ?
the larger uncertainty in our deduced [ mg / fe ] ratios makes this result quite uncertain .
we now turn to what we can learn about the chemical history of the galaxy by comparing the abundance ratios in ngc 7492 , at a galactocentric distance of 25 kpc , with those from our recent analysis of the inner halo gcs , m3 and m13 @xcite in m3 and m13 , with @xmath0 of 12 and 9 kpc respectively .
we note that the atomic parameters , the analysis procedures , and the software packages used are identical in both of these studies
. hence we should be able to detect small differences in the relative values of the abundance ratios of these three gcs .
table [ table_abundcomp ] gives the parameter @xmath41[x / fe ] , which is , for each species with detected lines , the mean abundance ratio [ x / fe ] for ngc 7492 with the average of the same parameter for m3 and for m13 subtracted . because of the large star - to - star differences in o / fe seen in m3 and especially in m13 , we subtract the mean [ o / fe ] for stars in m13 of luminosities comparable to those we observe in ngc 7492 .
similarly , @xcite found a luminosity dependence of [ mg / fe ] in m13 , most probably due to the luminosity dependence of non - lte corrections , which were ignored . again in this case we subtract the mean [ mg / fe ] of stars of comparable luminosity in m13 to those observed in ngc 7492 .
if we allow @xmath42 dex as a tolerable range given the potential internal and systematic errors in the analyses of these three gcs , we find that 80% of the 20 elements in common have a difference of 0.0@xmath42 dex , with only si , cr and co outside that range .
figure [ figure_abundcomp_m3m13 ] shows the resulting differences in [ x / fe ] as a function of atomic number
. the abundance ratios of [ x / fe ] for si and co are each larger in ngc 7492 than they are in m3 and m13 , while that of cr is smaller .
the largest magnitude of the set of @xmath41[x / fe ] occurs for si , and is + 0.23 dex .
galactic chemical evolution produces trends in abundance ratios as a function of metallicity .
the case of [ si / fe ] is illustrated in figure [ figure_sife ] , where metallicity is parameterized by [ fe / h ] ; examples for the elements ca , ti and ba are given in figures 21 , 22 and 23 of @xcite .
we use the same set of of high precision analyses of gcs as in @xcite , specifically ngc 6528 @xcite , ngc 6553 @xcite , 47 tuc @xcite , m71 @xcite , m5 @xcite , ngc 288 @xcite , ngc 362 @xcite , ngc 6752 @xcite , m3 @xcite , m13 @xcite , ngc 6397 @xcite and m15 @xcite , adding in ngc 7492 as well . to characterize the behavior of the metal - poor halo field stars , we adopt abundance ratios from recent large surveys of @xcite , @xcite , @xcite , and by @xcite .
no effort has been made to homogenize these analyses , but since the field star surveys were carried out over the course of more than a decade , we have corrected for the difference in the solar fe abundance adopted by each .
given that ngc 7492 has [ fe / h ] 0.35 dex smaller than the mean value for m3 and m13 , we have attempted to evaluate the correction to the difference caused by these global trends .
we can only do this for about half of the elements studied here .
this correction is given in the last column of table [ table_abundcomp ] , and is to be subtracted from the value of @xmath41[x / fe ] to form @xmath43[x / fe ] .
these corrections , which do not exceed 0.1 dex in magnitude , bring @xmath43[si / fe ] , @xmath43[cr / fe ] and @xmath43[co / fe ] within the range consistent with no difference ( 0.0@xmath42 dex ) , while not causing any additional elements to exceed the allowed range for equality .
the corrections for si , cr and co are shown on figure [ figure_abundcomp_m3m13 ] as well .
thus after implementing the corrections for global chemical evolution , all the elements in common show identical abundance ratios [ x / fe ] for ngc 7492 as for m3 and m13 , allowing for a @xmath42 dex tolerance ; 75% of the 20 values of @xmath43[x / fe ] lie within the range @xmath44 to @xmath45 dex .
this suggests that the galactic chemical evolution of the outer halo at @xmath0 25 kpc has been identical to that of the well studied inner halo gcs , at least up to the time of the formation of the old globular clusters ngc 7492 , m3 and m13
. in particular , our limited evidence , based on ba , la and eu abundance ratios , suggests the neutron capture processes , both @xmath46 and @xmath47 , appear to have had similar histories throughout the spatial extent of the halo for old gc stars .
we have carried a detailed abundance analysis for 21 elements in a sample of four rgb stars in the metal poor distant outer halo globular cluster ngc 7492([fe / h ] @xmath48 dex ) .
the analyzed spectra , obtained with hires at the keck observatory , are of high dispersion ( r=@xmath3=35,000 ) .
most elements show no sign of star - to - star variation within our limited sample .
we have , however , detected an anti - correlation between o and na abundances similar to that seen in our previous analyses of inner halo gcs as well as in studies of relatively nearby gcs by other .
a correlation between mg and na abundance may also be present .
we have compared the abundance ratios in ngc 7492 with those we previously determined for the much closer old halo gcs m3 and m13 @xcite , hoping that since all these analyses were carried out by the same two people within a timespan of only a few months in a completely consistent manner , with the same line lists , the same atomic parameters , the same analysis codes and procedures , etc . that small differences in the abundance ratios might be detectable .
we evaluate the trends of abundance ratio with metallicity for old halo stars from our data combined with published large surveys of halo field star abundances .
we then apply corrections to the abundances we derived for m3 and m13 for each species , when feasible , to extrapolate them to the 0.35 dex smaller [ fe / h ] of ngc 7492 . after making such corrections , all the elements in common show identical abundance ratios for ngc 7492 and for m3 and m13 , allowing for a @xmath42 dex tolerance , and 75% of them are then within the tolerance @xmath190.10 dex .
this suggests that the galactic chemical evolution of the outer halo at @xmath0 25 kpc has been identical to that of the well studied much closer inner halo gcs , at least up to the time of the formation of the old globular clusters ngc 7492 , m3 and m13
. in particular , our limited evidence , based on ba , la and eu abundance ratios , suggests the neutron capture processes , both @xmath46 and @xmath47 , appear to have had similar histories throughout the spatial extent of the halo for old gc stars as well .
the presence of the o / na anti - correlation in ngc 7492 , with @xmath0 of 25 kpc , and the similarity of its chemical history to that of the well studied nearby gcs , provide new constraints on any model of gc formation in the galactic halo .
the entire keck / hires user communities owes a huge debt to jerry nelson , gerry smith , steve vogt , and many other people who have worked to make the keck telescope and hires a reality and to operate and maintain the keck observatory .
we are grateful to the w. m. keck foundation for the vision to fund the construction of the w. m. keck observatory .
the authors wish to extend special thanks to those of hawaiian ancestry on whose sacred mountain we are privileged to be guests . without their generous hospitality , none of the observations presented herein would have been possible .
this publication makes use of data from the two micron all - sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center , funded by the national aeronautics and space administration and the national science foundation .
we are grateful to the national science foundation for partial support under grant ast-0205951 to jgc .
barnes , s. a. , 1968 , , 73 , 579 carretta , e. , cohen , j. g. , gratton , r. g. & behr , b. b. , 2001 , , 122 , 1469 carretta , e. , gratton , r. , bragaglia , a. , bonifacio , p. & pasquini , l. , 2004 , , 416 , 925 carretta , e. , bragaglia , a. & cacciari , c. , 2004 , , in press cohen , j. g. , frogel , j. a. , persson , s. e. & elias , j. h. , 1981 , , 249 , 481 , 1981 fulbright , j. p. , 2000 , , 120 , 1841 gratton , r. g. , carretta , e. , eriksson , k. & gustafsson , b. , 1999 , , 350 , 955 gratton , r. g. & sneden , c. , 1991 , , 241 , 501 gratton .
r. g. , _ et al .
_ , 2001 , , 369 , 87 gustafsson , b. , bell , r.a . ,
eriksson , k. & nordlund , , 1975 , a&a , 42 , 407 skrutskie , m. f. , schneider , s.e . ,
stiening , r. , strom , s.e . , weinberg , m.d .
, beichman , c. , chester , t. _ et al .
_ , 1997 , in _ the impact of large scale near - ir sky surveys _ , ed .
et al . _
( dordrecht : kluwer ) , p. 187
sneden , c. , kraft , r. p. , shetrone , m. d. , smith , g. h. , langer , g. e. & prosser , c. f. , 1997 , , 114 , 1964 sneden , c. , kraft , r. p. , guhathakurta , p. , peterson , r. c. & fulbright , j. p. , 2004
, , 127 , 2162 takeda , y. , zhao , g. , takada - hidai , m. , chen , y. , saito , y. & zhang , h. w. , 2003 , chin.j.astron.astrophys . , 3 , 316 thevenin , f. , carbonnel , c. , de freitas pacheco , j. a. , idiart , t. p. , jasniewsicz , g. , de lavery , p. & plez , b. , 2001 , , 373 , 905 vogt , s. e. _ et al .
_ 1994 , spie , 2198 , 362 lcrrrrrr h , 231 & 23 08 22.32 @xmath49 37 43 & 14.71 & 13.36 & 20/08/2003 & 800 & 88 & @xmath50 + t , 458 & 23 08 25.75 @xmath49 37 10 & 15.50 & 14.41 & 21/08/2003 & 3600 & 92 & @xmath51 + r & 23 08 29.46 @xmath49 36 32 & 15.51 & 14.40 & 20,21/08/2003 & 3000 & 92 & @xmath52 + k , 950 & 23 08 20.83 @xmath49 36 20 & 15.77 & 14.72 & 22/08/2003 & 3600 & 105 & @xmath15 + llrrrrrrrrrrrrrrr oi & 6300.30 & 0.00 & @xmath69.78 & 30.0 & 23.5 & 13.5 & 27.3 + oi & 6363.78 & 0.02 & @xmath610.30 & 9.0 & & 8.0 & 9.5 + nai & 5682.63 & 2.10 & @xmath60.70 & 46.0 & 32.1 & 8.0 & 16.0 + nai & 5688.19 & 2.10 & @xmath60.42 & 61.0 & 49.6 & 15.0 & 28.5 + nai & 6160.75 & 2.00 & @xmath61.23 & 18.5 & & & + mgi & 4703.00 & 4.34 & @xmath60.67 & 157.2 & 126.7 & 115.8 & 136.0 + mgi & 5528.40 & 4.34 & @xmath60.48 & 170.2 & 148.0 & 133.8 & 145.8 + mgi & 5711.09 & 4.34 & @xmath61.67 & 82.7 & 64.0 & 30.0 & 46.0 + sii & 5690.43 & 4.93 & @xmath61.87 & 27.0 & 20.0 & & 23.5 + sii & 5948.54 & 5.08 & @xmath61.23 & 42.0 & 36.0 & 33.0 & 27.0 + sii & 6155.13 & 5.62 & @xmath60.76 & 24.0 & 20.0 & 17.0 & 31.0 + sii & 6237.32 & 5.62 & @xmath61.01 & 15.0 & & & + cai & 5512.99 & 2.93 & @xmath60.27 & 56.5 & 40.0 & 23.0 & 33.7 + cai & 5581.96 & 2.52 & @xmath60.47 & 82.7 & 58.3 & 44.8 & 48.1 + cai & 5588.75 & 2.52 & 0.44 & 133.4 & 116.7 & 95.3 & 103.0 + cai & 5590.11 & 2.52 & @xmath60.71 & 80.7 & 61.1 & 46.7 & 46.9 + cai & 5601.28 & 2.52 & @xmath60.44 & 77.3 & 68.3 & 49.0 & 41.5 + cai & 6161.30 & 2.52 & @xmath61.03 & 49.2 & 24.0 & & 22.0 + cai & 6162.17 & 1.90 & @xmath60.09 & & 145.4 & 132.0 & 148.0 + cai & 6166.44 & 2.52 & @xmath61.05 & 51.3 & 31.0 & 11.0 & 25.5 + cai & 6169.04 & 2.52 & @xmath60.54 & 72.5 & 54.5 & 35.9 & 43.0 + cai & 6169.56 & 2.52 & @xmath60.27 & 93.8 & 65.6 & 46.7 & 61.2 + cai & 6471.66 & 2.52 & @xmath60.59 & 83.6 & 61.9 & 38.2 & 50.8 + cai & 6493.78 & 2.52 & 0.14 & 123.8 & 90.9 & 77.8 & 89.6 + scii & 5526.79 & 1.77 & 0.13 & 94.7 & 88.6 & 78.2 & 82.6 + scii & 5657.90 & 1.51 & @xmath60.50 & 93.4 & 76.1 & 66.6 & 72.7 + scii & 5667.15 & 1.50 & @xmath61.24 & 54.0 & 41.3 & 31.0 & 32.0 + scii & 5669.04 & 1.50 & @xmath61.12 & 60.8 & 45.0 & 34.8 & 44.8 + scii & 5684.20 & 1.51 & @xmath61.08 & 77.0 & 41.0 & 36.0 & 51.0 + scii & 6245.64 & 1.51 & @xmath61.13 & 56.9 & 35.0 & 32.2 & 34.4 + scii & 6604.60 & 1.36 & @xmath61.48 & 57.2 & 34.0 & 19.0 & 45.0 + tii & 4981.74 & 0.85 & 0.50 & 167.7 & 111.3 & 97.5 & 120.7 + tii & 4999.51 & 0.83 & 0.25 & 157.4 & 113.0 & 86.3 & 106.4 + tii & 5022.87 & 0.83 & @xmath60.43 & 107.2 & 71.0 & 41.8 & 62.3 + tii & 5039.96 & 0.02 & @xmath61.13 & 148.0 & 94.4 & 73.6 & 75.7 + tii & 5426.26 & 0.02 & @xmath63.01 & 33.7 & & & + tii & 5471.20 & 1.44 & @xmath61.39 & 15.0 & & & + tii & 5490.15 & 1.46 & @xmath60.93 & 34.7 & & & + tii & 5644.14 & 2.27 & 0.05 & 41.0 & & & + tii & 5662.16 & 2.32 & @xmath60.11 & 24.5 & & & + tii & 5937.81 & 1.07 & @xmath61.89 & 16.4 & & & + tii & 5941.75 & 1.05 & @xmath61.52 & 43.6 & 17.3 & & 14.0 + tii & 5953.16 & 1.89 & @xmath60.33 & 41.1 & & & + tii & 5978.54 & 1.87 & @xmath60.50 & 29.0 & & & + tii & 6258.10 & 1.44 & @xmath60.35 & 75.0 & 37.3 & 25.4 & 35.7 + tii & 6261.10 & 1.43 & @xmath60.48 & 72.2 & 36.5 & 18.0 & 28.7 + tii & 6303.76 & 1.44 & @xmath61.57 & 9.0 & & & + tii & 6312.22 & 1.46 & @xmath61.55 & 9.5 & & & + tii & 6743.12 & 0.90 & @xmath61.63 & 45.8 & & & + tiii & 4657.20 & 1.24 & @xmath62.32 & 84.0 & 97.0 & 59.3 & 83.9 + tiii & 4708.67 & 1.24 & @xmath62.37 & 96.8 & 81.8 & 71.1 & 87.7 + tiii & 4865.62 & 1.12 & @xmath62.81 & 75.0 & 84.0 & 52.7 & 64.1 + tiii & 5185.91 & 1.89 & @xmath61.46 & 98.0 & 85.9 & 74.8 & 81.9 + tiii & 5336.79 & 1.58 & @xmath61.63 & 121.0 & 105.0 & 93.2 & 105.4 + vi & 5670.85 & 1.08 & @xmath60.43 & 35.0 & & & + vi & 5703.57 & 1.05 & @xmath60.21 & 34.3 & & & + vi & 6081.44 & 1.05 & @xmath60.58 & 29.8 & & & + vi & 6090.22 & 1.08 & @xmath60.06 & 41.7 & & & + vi & 6199.20 & 0.29 & @xmath61.28 & 32.0 & & & + vi & 6251.82 & 0.29 & @xmath61.34 & 35.4 & & & + vi & 6274.64 & 0.27 & @xmath61.67 & 21.8 & & & + vi & 6285.14 & 0.28 & @xmath61.51 & 46.2 & & & + cri & 5345.81 & 1.00 & @xmath60.97 & 134.1 & 104.3 & 75.4 & 85.6 + cri & 5348.33 & 1.00 & @xmath61.29 & 115.8 & 74.5 & 54.7 & 63.9 + cri & 5409.80 & 1.03 & @xmath60.71 & 169.5 & 126.0 & 88.8 & 106.9 + cri & 5787.96 & 3.32 & @xmath60.08 & 12.0 & & & + mni & 4754.04 & 2.28 & @xmath60.09 & 107.2 & 74.2 & 49.8 & 62.2 + mni & 4783.42 & 2.30 & 0.04 & 121.2 & 95.0 & 69.2 & 72.9 + mni & 4823.51 & 2.32 & 0.14 & 116.5 & 75.0 & 67.5 & 78.0 + mni & 5537.74 & 2.19 & @xmath62.02 & 28.0 & & & + fei & 4788.77 & 3.24 & @xmath61.81 & 49.5 & & & + fei & 4891.50 & 2.85 & @xmath60.11 & 170.7 & 144.8 & 137.7 & 142.3 + fei & 4919.00 & 2.86 & @xmath60.34 & 168.5 & 141.0 & 117.4 & 141.4 + fei & 5083.34 & 0.96 & @xmath62.96 & 165.2 & 131.2 & 112.3 & 126.2 + fei & 5166.28 & 0.00 & @xmath64.20 & & 141.4 & 118.9 & 139.7 + fei & 5194.95 & 1.56 & @xmath62.09 & 174.6 & 130.3 & 121.0 & 129.1 + fei & 5232.95 & 2.94 & @xmath60.10 & 172.0 & 139.0 & 129.6 & 139.5 + fei & 5324.19 & 3.21 & @xmath60.10 & 160.4 & 135.8 & 117.7 & 125.8 + fei & 5393.18 & 3.24 & @xmath60.72 & 114.4 & 93.0 & 81.8 & 86.1 + fei & 5410.92 & 4.47 & 0.40 & 87.1 & 73.4 & 56.0 & 68.3 + fei & 5415.21 & 4.39 & 0.64 & 105.0 & 91.0 & 66.4 & 86.9 + fei & 5424.08 & 4.32 & 0.51 & 114.7 & 87.8 & 83.1 & 99.2 + fei & 5445.05 & 4.39 & @xmath60.03 & 79.3 & 70.0 & 47.0 & 67.2 + fei & 5473.90 & 4.15 & @xmath60.69 & 42.2 & 43.9 & 20.0 & 24.5 + fei & 5493.50 & 4.10 & @xmath61.68 & 15.9 & & & + fei & 5497.52 & 1.01 & @xmath62.83 & & 147.7 & 134.3 & 143.9 + fei & 5501.46 & 0.96 & @xmath63.05 & & 138.6 & 118.5 & 136.5 + fei & 5506.79 & 0.99 & @xmath62.79 & & 146.2 & 130.0 & 142.4 + fei & 5525.55 & 4.23 & @xmath61.08 & 21.0 & & & 13.5 + fei & 5554.88 & 4.55 & @xmath60.35 & 33.9 & 35.0 & 22.0 & 25.6 + fei & 5567.39 & 2.61 & @xmath62.67 & 64.8 & 39.0 & & 31.2 + fei & 5569.62 & 3.42 & @xmath60.49 & 123.0 & 92.0 & 81.1 & 91.2 + fei & 5572.84 & 3.40 & @xmath60.28 & 146.1 & 108.6 & 100.1 & 104.3 + fei & 5576.09 & 3.43 & @xmath60.92 & 109.1 & 83.5 & 63.7 & 72.3 + fei & 5586.76 & 3.37 & @xmath60.14 & 144.8 & 117.7 & 105.1 & 114.1 + fei & 5641.44 & 4.26 & @xmath61.08 & 30.0 & & & + fei & 5662.52 & 4.18 & @xmath60.57 & 65.0 & 48.5 & 37.2 & 39.8 + fei & 5679.02 & 4.65 & @xmath60.82 & 25.0 & & & + fei & 5701.54 & 2.56 & @xmath62.14 & 106.0 & 76.3 & 46.9 & 66.7 + fei & 5705.98 & 4.61 & @xmath60.49 & 25.5 & & & 29.0 + fei & 5752.04 & 4.55 & @xmath60.94 & 19.2 & & & + fei & 5753.12 & 4.26 & @xmath60.69 & 52.5 & 38.9 & 30.0 & 30.4 + fei & 5762.99 & 4.21 & @xmath60.41 & 70.2 & 44.9 & 36.5 & 48.0 + fei & 5775.06 & 4.22 & @xmath61.30 & 22.6 & & & + fei & 5778.46 & 2.59 & @xmath63.43 & 19.6 & & & + fei & 5806.72 & 4.61 & @xmath60.95 & 11.0 & & & + fei & 5859.60 & 4.55 & @xmath60.55 & 34.0 & 22.0 & 15.6 & 17.7 + fei & 5862.35 & 4.55 & @xmath60.33 & 52.2 & 30.0 & 21.0 & 24.2 + fei & 5883.81 & 3.96 & @xmath61.26 & 27.0 & 17.5 & & 19.0 + fei & 5930.17 & 4.65 & @xmath60.14 & 47.4 & 36.7 & 21.3 & 18.0 + fei & 5934.65 & 3.93 & @xmath61.07 & 42.0 & 42.0 & 29.5 & 21.8 + fei & 5952.72 & 3.98 & @xmath61.34 & 53.3 & 35.5 & 20.0 & 24.3 + fei & 5956.69 & 0.86 & @xmath64.50 & 110.7 & 60.7 & 34.5 & 51.2 + fei & 5976.79 & 3.94 & @xmath61.33 & 41.7 & 18.0 & 17.5 & 19.3 + fei & 5983.69 & 4.55 & @xmath60.66 & 37.5 & & 16.8 & 20.0 + fei & 6024.05 & 4.55 & 0.03 & 67.4 & 40.5 & 38.9 & 48.9 + fei & 6027.05 & 4.07 & @xmath61.09 & 38.3 & & & + fei & 6055.99 & 4.73 & @xmath60.37 & 29.2 & 28.0 & & 15.0 + fei & 6065.48 & 2.61 & @xmath61.41 & 143.9 & 111.4 & 90.6 & 97.2 + fei & 6078.50 & 4.79 & @xmath60.33 & 29.2 & 25.5 & 10.5 & 12.5 + fei & 6137.69 & 2.59 & @xmath61.35 & 144.3 & 110.6 & 99.2 & 109.8 + fei & 6151.62 & 2.18 & @xmath63.37 & 66.0 & 37.8 & 17.6 & 30.1 + fei & 6157.73 & 4.07 & @xmath61.16 & 42.5 & 26.2 & 8.0 & 20.9 + fei & 6165.36 & 4.14 & @xmath61.47 & 24.9 & & & + fei & 6173.34 & 2.22 & @xmath62.88 & 91.0 & 62.6 & 36.0 & 51.6 + fei & 6180.20 & 2.73 & @xmath62.65 & 55.0 & 29.8 & 14.0 & 27.7 + fei & 6187.99 & 3.94 & @xmath61.62 & 14.0 & & & + fei & 6191.56 & 2.43 & @xmath61.42 & 174.9 & 129.4 & 99.3 & 128.8 + fei & 6200.31 & 2.61 & @xmath62.37 & 88.5 & 59.7 & 31.5 & 47.1 + fei & 6240.65 & 2.22 & @xmath63.17 & 59.3 & 35.2 & 16.0 & 23.0 + fei & 6246.32 & 3.60 & @xmath60.88 & 91.9 & 74.3 & 48.9 & 57.7 + fei & 6252.55 & 2.40 & @xmath61.77 & 146.8 & 106.8 & 100.0 & 112.0 + fei & 6254.26 & 2.28 & @xmath62.43 & 122.1 & 95.9 & 73.0 & 79.3 + fei & 6265.13 & 2.18 & @xmath62.54 & 118.9 & 89.8 & 59.1 & 77.7 + fei & 6297.79 & 2.22 & @xmath62.64 & 108.1 & 79.9 & 44.1 & 59.4 + fei & 6301.51 & 3.65 & @xmath60.72 & 93.7 & 71.0 & 50.0 & 65.1 + fei & 6315.31 & 4.14 & @xmath61.23 & 25.0 & & & + fei & 6355.03 & 2.84 & @xmath62.29 & 69.3 & 45.1 & 19.0 & 33.0 + fei & 6380.75 & 4.19 & @xmath61.38 & 22.3 & & & + fei & 6393.60 & 2.43 & @xmath61.58 & 156.6 & 118.3 & 100.8 & 115.0 + fei & 6408.03 & 3.69 & @xmath61.02 & 86.7 & 59.3 & 40.2 & 49.6 + fei & 6411.65 & 3.65 & @xmath60.72 & 109.2 & 75.1 & 58.2 & 75.7 + fei & 6421.35 & 2.28 & @xmath62.01 & 145.9 & 101.0 & 85.7 & 101.2 + fei & 6475.63 & 2.56 & @xmath62.94 & 58.8 & 33.2 & 14.6 & 19.5 + fei & 6481.87 & 2.28 & @xmath63.01 & 86.3 & 46.5 & 30.6 & 43.2 + fei & 6498.94 & 0.96 & @xmath64.69 & 98.1 & 51.2 & 31.9 & 43.5 + fei & 6546.24 & 2.76 & @xmath61.54 & 137.2 & 99.1 & 74.1 & 92.7 + fei & 6581.21 & 1.48 & @xmath64.68 & 42.7 & & & + fei & 6592.91 & 2.73 & @xmath61.47 & 130.9 & 97.4 & 81.7 & 90.8 + fei & 6593.87 & 2.43 & @xmath62.37 & 99.2 & 67.4 & 48.0 & 64.2 + fei & 6608.02 & 2.28 & @xmath63.93 & 16.0 & & & + fei & 6609.11 & 2.56 & @xmath62.66 & 79.1 & 45.5 & 26.5 & 37.5 + fei & 6633.75 & 4.79 & @xmath60.80 & 27.0 & & & 12.0 + fei & 6648.12 & 1.01 & @xmath65.92 & 13.0 & & & + fei & 6739.52 & 1.56 & @xmath64.79 & 27.0 & & & + fei & 6750.15 & 2.42 & @xmath62.58 & 101.8 & 68.0 & 46.6 & 60.2 + fei & 6839.83 & 2.56 & @xmath63.35 & 27.9 & & & + fei & 6843.65 & 4.55 & @xmath60.83 & 25.6 & & & + fei & 6855.18 & 4.56 & @xmath60.74 & 34.6 & & & + fei & 6861.95 & 2.42 & @xmath63.85 & 20.0 & & & + fei & 6978.85 & 2.48 & @xmath62.45 & 104.8 & 72.6 & 44.6 & 66.5 + fei & 6988.52 & 2.40 & @xmath63.56 & 39.7 & & & + fei & 7022.95 & 4.19 & @xmath61.15 & 38.0 & & & + fei & 7038.22 & 4.22 & @xmath61.20 & 30.1 & & & + feii & 4923.93 & 2.88 & @xmath61.32 & 138.4 & & & + feii & 5197.58 & 3.23 & @xmath62.23 & 74.0 & 64.0 & 70.9 & 76.8 + feii & 5234.63 & 3.22 & @xmath62.22 & 75.1 & 67.6 & 74.3 & 83.6 + feii & 5414.08 & 3.22 & @xmath63.62 & 22.0 & 23.5 & 10.0 & 16.5 + feii & 5425.26 & 3.00 & @xmath63.24 & 29.0 & 27.0 & 27.0 & 37.5 + feii & 5534.85 & 3.25 & @xmath62.64 & 64.3 & 56.8 & 53.2 & 52.9 + feii & 5991.38 & 3.15 & @xmath63.57 & 30.0 & & 12.0 & 27.4 + feii & 6149.26 & 3.89 & @xmath62.69 & 18.0 & 25.2 & 16.4 & 26.9 + feii & 6247.56 & 3.89 & @xmath62.36 & 40.0 & 31.9 & 31.0 & 31.6 + feii & 6369.46 & 2.89 & @xmath64.20 & 10.0 & & & + feii & 6416.92 & 3.89 & @xmath62.69 & 19.0 & 15.9 & & 15.8 + feii & 6516.08 & 2.89 & @xmath63.45 & 64.0 & 58.3 & 47.5 & 51.8 + coi & 5530.79 & 1.71 & @xmath62.06 & 30.0 & & 9.0 & 10.0 + coi & 5647.23 & 2.28 & @xmath61.56 & 21.5 & & & + coi & 6189.00 & 1.71 & @xmath62.45 & 18.0 & & & + nii & 5578.72 & 1.68 & @xmath62.64 & 86.1 & 44.1 & 32.0 & 33.0 + nii & 5587.86 & 1.93 & @xmath62.14 & 81.4 & 36.9 & 20.5 & 39.0 + nii & 5682.20 & 4.10 & @xmath60.47 & 18.0 & & & + nii & 5748.35 & 1.68 & @xmath63.26 & 41.0 & & 7.0 & 14.0 + nii & 5846.99 & 1.68 & @xmath63.21 & 34.1 & & & + nii & 6128.97 & 1.68 & @xmath63.33 & 30.0 & 23.5 & 10.4 & 10.5 + nii & 6175.37 & 4.09 & @xmath60.54 & 22.0 & & & + nii & 6176.81 & 4.09 & @xmath60.53 & 21.8 & & & + nii & 6177.24 & 1.83 & @xmath63.51 & 15.0 & & & + nii & 6482.80 & 1.93 & @xmath62.63 & 47.8 & 21.0 & 13.3 & 19.6 + nii & 6586.31 & 1.95 & @xmath62.81 & 45.5 & 22.3 & 8.0 & 13.0 + nii & 6643.63 & 1.68 & @xmath62.30 & 126.3 & 83.0 & 56.5 & 73.9 + nii & 6767.77 & 1.83 & @xmath62.17 & 106.4 & 74.8 & 51.7 & 64.1 + cui & 5105.54 & 1.39 & @xmath61.50 & 82.0 & 46.0 & 24.0 & 28.0 + cui & 5782.12 & 1.64 & @xmath61.78 & 41.5 & & & + zni & 4722.16 & 4.03 & @xmath60.39 & 45.0 & 36.6 & 41.0 & 44.0 + zni & 4810.54 & 4.08 & @xmath60.17 & 46.0 & 54.6 & 52.4 & 48.5 + yii & 4883.69 & 1.08 & 0.07 & 85.0 & 71.3 & 74.5 & 71.8 + yii & 5087.43 & 1.08 & @xmath60.17 & 71.0 & 55.0 & 56.2 & 60.1 + yii & 5200.42 & 0.99 & @xmath60.57 & 56.0 & 42.1 & 32.7 & 44.0 + zri & 6127.44 & 0.15 & @xmath61.06 & 17.0 & & & + zri & 6134.55 & 0.00 & @xmath61.28 & 13.0 & & & + zri & 6143.20 & 0.07 & @xmath61.10 & 10.0 & & & + baii & 5853.70 & 0.60 & @xmath61.01 & 117.3 & 98.1 & 99.9 & 108.2 + baii & 6141.70 & 0.70 & @xmath60.07 & 172.8 & 134.7 & 152.3 & 155.9 + baii & 6496.90 & 0.60 & @xmath60.38 & 186.6 & 146.8 & 154.9 & 163.0 + laii & 5122.99 & 0.32 & @xmath60.85 & 27.8 & & & + laii & 6390.48 & 0.32 & @xmath61.41 & 21.5 & & & + ndii & 4947.02 & 0.56 & @xmath61.13 & 18.0 & & & + ndii & 4959.12 & 0.06 & @xmath60.80 & 66.6 & 34.0 & 23.5 & 32.0 + ndii & 5092.79 & 0.38 & @xmath60.61 & 44.0 & 28.5 & 17.4 & 16.9 + ndii & 5212.35 & 0.20 & @xmath60.96 & 48.0 & & & + ndii & 5249.58 & 0.98 & 0.20 & 49.9 & 30.5 & 33.5 & 31.5 + ndii & 5319.81 & 0.55 & @xmath60.14 & 61.9 & 33.2 & 33.8 & 27.8 + euii & 6645.11 & 1.38 & 0.12 & 25.5 & & & + lclc o & 8.85 & fe & 7.45 + na & 6.33 & ni & 6.25 + mg & 7.54 & cu & 4.21 + al & 6.47 & zn & 4.60 + si & 7.55 & ba & 2.13 + ca & 6.36 & y & 2.24 + sc & 3.10 & zr & 2.60 + ti & 4.99 & la & 1.14 + v & 4.00 & nd & 1.45 + cr & 5.67 & eu & 0.51 + mn & 5.39 & dy & 1.10 + lcrlrrcrcrcccc 231 & @xmath61.76 @xmath19 0.05 * & 90 & @xmath61.72 @xmath19 0.06 & 12 & 0.26 @xmath19 0.05 * & 2 & 0.10 @xmath19 0.05 & 3 & 0.67 @xmath19 0.16 & 3 + r & @xmath61.77@xmath19 0.05 * & 66 & @xmath61.80 @xmath19 0.08 & 9 & 0.49 @xmath19 0.10 & 1 & 0.19 @xmath19 0.05 * & 2 & 0.52 @xmath19 0.06 & 3 + 458 & @xmath61.79 @xmath19 0.05 * & 70 & @xmath61.77 @xmath19 0.05 & 10 & 0.55 @xmath19 0.05 * & 2 & @xmath60.10 @xmath19 0.05 * & 2 & 0.55 @xmath19 0.20 & 3 + 950 & @xmath61.94 @xmath19 0.05 * & 64 & @xmath61.88 @xmath19 0.05 & 9 & 0.52 @xmath19 0.15 & 2 & @xmath60.28 @xmath19 0.05 * & 2 & 0.45 @xmath19 0.20 & 3 + llrrrrrlrrrlr 231 & 0.33 @xmath19 0.09 & 4 & 0.09 @xmath19 0.05 * & 11 & 0.20 @xmath19 0.07 & 7 & 0.16 @xmath19 0.05 * & 18 & @xmath60.07 @xmath19 0.06 & 8 + r & 0.36 @xmath19 0.10 & 3 & 0.11 @xmath19 0.05 * & 12 & 0.09 @xmath19 0.05 * & 7 & 0.12 @xmath19 0.08 & 7 & & 0 + 458 & 0.45 @xmath19 0.12 & 3 & 0.05 @xmath19 0.05 * & 12 & 0.10 @xmath19 0.06 & 7 & 0.18 @xmath19 0.07 & 7 & & 0 + 950 & 0.42 @xmath19 0.11 & 2 & 0.07 @xmath19 0.05 & 11 & 0.15 @xmath19 0.05 * & 7 & 0.07 @xmath19 0.05 * & 6 & & 0 + llrlllllrrrr 231 & @xmath60.24 @xmath19 0.09 & 4 & @xmath60.24 @xmath19 0.09 & 4 & 0.11 @xmath19 0.05 * & 3 & @xmath60.06 @xmath19 0.05 * & 13 & @xmath60.63 @xmath19 0.06 & 2 + r & @xmath60.16 @xmath19 0.11 & 3 & @xmath60.48 @xmath19 0.13 & 3 & & 0 & @xmath60.07 @xmath19 0.08 & 7 & @xmath60.57 @xmath19 0.10 & 1 + 458 & @xmath60.29 @xmath19 0.06 & 3 & @xmath60.43 @xmath19 0.05 * & 3 & 0.18 @xmath19 0.10 & 1 & @xmath60.09 @xmath19 0.05 * & 8 & @xmath60.72 @xmath19 0.10 & 1 + 950 & @xmath60.29 @xmath19 0.05 * & 3 & @xmath60.39 @xmath19 0.08 & 3 & 0.32 @xmath19 0.10 & 1 & @xmath60.09 @xmath19 0.06 & 8 & @xmath60.62 @xmath19 0.10 & 1 + lrlllrrrrrrrrrr 231 & @xmath60.09 @xmath19 0.08 & 2 & @xmath60.29 @xmath19 0.05 * & 4 & 0.36 @xmath19 0.10 & 3 & 0.24 @xmath19 0.10 & 3 & 0.12 @xmath19 0.16 & 2 + r & @xmath60.09 @xmath19 0.08 & 2 & @xmath60.35 @xmath19 0.05 * & 4 & & 0 & 0.05 @xmath19 0.09 & 3 & & 0 + 458 & @xmath60.16 @xmath19 0.05 & 2 & @xmath60.31 @xmath19 0.06 & 4 & & 0 & 0.34 @xmath19 0.07 & 3 & & 0 + 950 & 0.13 @xmath19 0.05 * & 2 & @xmath60.11 @xmath19 0.08 & 4 & & 0 & 0.50 @xmath19 0.08 & 3 & & 0 + lrrcrrrrrr oi & 0.46 & 0.13 & 0.10 & 1.30 & 4 + nai & @xmath60.02 & 0.21 & 0.07 & 3.00 & 4 + mgi & 0.55 & 0.09 & 0.20 & 0.45 & 4 + sii & 0.39 & 0.06 & 0.10 & 0.60 & 4 + cai & 0.08 & 0.03 & 0.12 & 0.25 & 4 + scii & 0.14 & 0.05 & 0.21 & 0.24 & 4 + tii & 0.13 & 0.05 & 0.12 & 0.42 & 4 + tiii & 0.34 & 0.06 & 0.15 & 0.40 & 4 + vi & @xmath60.07 & & 0.23 & & 1 + cri & @xmath60.25 & 0.06 & 0.17 & 0.35 & 4 + mni & @xmath60.39 & 0.10 & 0.18 & 0.55 & 4 + fei & @xmath61.82 & 0.08 & 0.10 & 0.80 & 4 + feii & @xmath61.79 & 0.07 & 0.13 & 0.54 & 4 + coi & 0.20 & 0.11 & 0.12 & 0.92 & 3 + nii & @xmath60.08 & 0.02 & 0.10 & 0.20 & 4 + cui & @xmath60.64 & 0.06 & 0.16 & 0.38 & 4 + zni & @xmath60.03 & 0.11 & 0.13 & 0.85 & 4 + yii & @xmath60.27 & 0.11 & 0.15 & 0.73 & 4 + zri & 0.36 & & 0.16 & & 1 + baii & 0.28 & 0.19 & 0.22 & 0.86 & 4 + laii & 0.12 & & 0.18 & & 1 + ndii & 0.32 & 0.11 & 0.11 & 1.00 & 4 + euii & 0.61 & & & & 1 + lrrrr o & 0.46 & 0.15 & 0.02 & + na & @xmath60.02 & 0.21 & @xmath60.10 & 0.00 + mg & 0.55 & 0.09 & 0.12 & 0.05 + si & 0.39 & 0.06 & 0.23 & 0.08 + ca & 0.08 & 0.03 & @xmath60.09 & 0.07 + sc & 0.14 & 0.05 & 0.14 & 0.07 + ti & 0.24 & 0.06 & 0.08 & 0.07 + v & @xmath60.07 & & 0.01 & + cr & @xmath60.25 & 0.06 & @xmath60.22 & @xmath60.10 + mn & @xmath60.39 & 0.10 & @xmath60.04 & @xmath60.05 + co & 0.20 & 0.11 & 0.21 & 0.07 + ni & @xmath60.08 & 0.01 & @xmath60.01 & 0.00 + cu & @xmath60.64 & 0.06 & @xmath60.02 & + zn & @xmath53 & 0.11 & 0.02 & + y & @xmath60.27 & 0.11 & @xmath60.04 & + zr & 0.36 & & 0.40 & + ba & 0.28 & 0.019 & 0.05 & @xmath60.04 + la & 0.12 & & 0.03 & + nd & 0.32 & 0.11 & 0.10 & + eu & 0.61 & & 0.07 & 0.00 + | we have carried out a detailed abundance analysis for 21 elements in a sample of four rgb stars in the outer halo globular cluster ngc 7492 ( @xmath0 25 kpc ) ; we find [ fe / h ] = @xmath1 dex inferred from fe i lines ( @xmath2 from fe ii ) using high dispersion ( r=@xmath3=35,000 ) spectra obtained with hires at the keck observatory .
most elements show no sign of star - to - star variation within our limited sample .
we have , however , detected an anti - correlation between o and na abundances similar to that seen in our previous analyses of inner halo gcs as well as in studies of relatively nearby gcs by others .
we compare the abundance ratios in ngc 7492 with those we previously determined for the much closer old halo gcs m3 and m13 . after making corrections for trends of abundance ratio with metallicity characteristic of halo stars
, we find that for for these three gcs for each of the elements in common we deduce identical abundance ratios with respect to fe to within the probable measurement uncertainties .
thus , the chemical history of the outer halo as exemplified by the metal - poor outer halo globular cluster ngc 7492is indistinguishable from that of the inner halo , exemplified by m3 and m13 , at least through the epoch of formation of these old globular clusters .
this applies to the neutron capture processes as well . |
by means of the code named cuba ( described in hmi ) , we have calculated the ratio @xmath4 for a set of qso spectral models as discussed in hmii .
the hi and heii opacities computed with various qso background models have been then tested against the observed values .
together with the already mentioned heii data , we have taken a collection of hi data .
the calculated redshift dependent heii @xmath5 ( the fractional flux decrement shortward of ly@xmath6 ) is shown in figure for different input qso spectral models , together with the available data points . in the left upper panel we plotted also hi @xmath5 against the data . | the detection of a heii absorption trough in the spectra of three high - redshift quasars provides new constraints on the spectral shape of the uv extragalactic background .
the relative strengths of the observed flux decrements at the rest - frame wavelengths of 1216 and 304 require a relatively soft radiation field at 4 ryd compared to 1 ryd .
if the ionizing metagalactic flux is dominated by the integrated light from qsos , and the heii reionization of the universe was completed well before @xmath0 , then the uv / soft x - ray spectrum of individual quasars at these epochs must , once the cosmological `` filtering '' through material along the line of sight is taken into account , satisfy similar constraints on the average .
we model the propagation of agn - like ionizing radiation through the intergalactic medium using cuba , a numerical code developed in our prior work .
we show that , in order to explain the reported heii absorption , any thermal component responsible for the `` soft x - ray excess '' observed in the spectra of agns at @xmath1 must , at @xmath0 , have a typical temperature @xmath2 ev , and a luminosity not exceeding @xmath3% of that of the `` uv bump '' . |
the thermoelectric ( te ) effect allows for generation of electricity from temperature gradients .
this opens the path for self - sustaining sensors , power generation from waste heat or a complementary to photovoltaics@xcite . in recent years
, several approaches for different material systems and device layouts have been developed to achieve a high efficiency at low cost.@xcite while the highest efficiencies have been reported for te elements based on inorganic semiconductors , recently , doped polymers gained attention because despite having lower efficiency they have the potential of a strong cost reduction.@xcite additionally , these organic compounds allow for fabrication of te devices on flexible foils which are easier to handle and can e.g. be installed on large and curved surfaces like in cooling towers . in this work ,
another class of organic semiconducting compounds is studied : small molecules , which so far have hardly been examined with respect to their te properties , but are predicted to be a promising material class@xcite .
small molecules have the advantage to polymers that due to their low molecular weight they can be deposited by thermal evaporation in vacuum , allowing for high material purity and arbitrary layer stacks , while mass production on large substrates via roll - to - roll fabrication is feasible as well .
doping of small molecules has led to novel devices like organic light - emitting diodes ( oleds ) and organic photovoltaic cells.@xcite the electrical conductivity of such doped layers of small molecules typically increases with temperature@xcite , while their thermal conductivity is rather low@xcite .
the scope of this publication is the comparison of the te potential of several material combinations of small molecular hosts and dopants .
it is found that @xmath0-doped c60samples show several orders of magnitude higher power factors , compared to @xmath1-doped hole transporting compounds .
the highest values are obtained for low concentrations of cr2(hpp)4)or high concentrations of in c60 .
the resulting of these first tests for this material class are as high as 16% of the current record reported for optimized devices employing the doped polymer .
samples are prepared and measured in - situin a vacuum setup ( base pressure @xmath2pa ) , which has been described in detail earlier@xcite . a doped organic layer of mostly thickness is deposited by thermal co - deposition of host and dopant onto a glass substrate with two parallel , thick gold contacts of distance .
this sample layout allows for measurement of both , conductivity and seebeck coefficient at different temperatures , because the substrate is mounted onto two separately heated copper blocks below the gold contacts .
all samples are thermally annealed at in vacuum prior to investigation to ensure reproducibility of temperature - dependent measurements .
electrical measurements are performed using a keithley 236 source measure unit . for probing the conductivity ,
a voltage of is applied to the gold contacts and the current flow is measured and averaged over 2 minutes to compensate for statistical fluctuations . as for all samples a linear and symmetric current response is found for @xmath3 to , we assume ohmic injection . for measuring the seebeck coefficient , a temperature difference of between the gold contacts is used and the thermovoltage is measured and averaged for several minutes , followed by a measurement at to compensate for measurement artifacts . both , and ,
are measured at different mean temperature and for various samples of different doping concentration . in the following ,
is expressed in terms of molar ratio ( mr ) , being the ratio of the numbers of dopant to host molecules .
hole transporting materials investigated here are the hosts _ n_,_n_,_n_,_n_-tetrakis 4-methoxyphenyl - benzidine ( ) and _ n_,_n_-bis(9,9-dimethyl - fluoren-2-yl)-_n_,_n_-diphenyl - benzidine ( ) @xmath1-doped by 1,3,4,5,7,8-hexafluorotetracyanonaphthoquinodimethane(f6-tcnnq ) and 36-fold fluorinated c60(c60f36).@xcite as electron transporting material the host c60is studied , comparing 5 @xmath0-dopants : tetrakis(1,3,4,6,7,8-hexahydro-2h - pyrimido[1,2-a]pyrimidinato)dichromium(ii)and -ditungsten ( ii ) ( cr2(hpp)4)@xcite and w2(hpp)4)@xcite ) , 3,6-bis(dimethylamino)acridine(aob@xcite ) , 2-(1,3-dimethyl-1__h__-benzoimidazol-3-ium-2-yl)phenolatehydrate(@xcite ) and 2-(2-methoxyphenyl)-1,3-dimethyl-1__h__-benzoimidazol-3-ium iodide(@xcite ) .
some of the conductivity and seebeck data presented in this work have been published earlier in the given references , but so far no conclusion for the te properties have been deduced .
the thermoelectric potential of a material is determined by the material s figure of merit @xmath4 with seebeck coefficient , electrical conductivity and thermal conductivity , measured at the mean temperature .
hence , for te applications , materials of low but high and are desired .
the numerator @xmath5 in equation ( [ eq : ztm ] ) is called power factor . doping a semiconducting te material allows for manipulating these parameters .
the presence of dopants leads to an increasing density of free charge carriers and thus along with a reduction of , due to the shift of the fermi level towards the transport level@xcite . at low doping concentrations ,
the thermal conductivity is attributed to phonons while at high contributions from the electrons dominate .
consequently , the optimum in doping concentration needs to be determined . for the most extensively studied inorganic te material system bi@xmath6te@xmath7 , record values in the order of at were achieved@xcite , while for the doped polymer recently a promising record of at room temperature has been reported@xcite .
the figure of merit largely affects the efficiency of power generation , which is given by the ratio of electrical energy @xmath8 generated to the net heat flow rate @xmath9@xcite : @xmath10 were and are the temperatures at the hot and cold side of the material , respectively . for @xmath11 , equation ( [ eq : eff ] )
approaches the carnot limit .
layers of doped small organic molecules have been found to show an increasing with temperature in the range of 30 to , while stays almost constant @xcite .
these properties make this class of materials a promising candidates for te application .
unfortunately , of such doped layers is in most cases unknown and difficult to measure , due to degradation effects by air - exposure@xcite .
hence , instead of , the power factor @xmath5 is used .
this evaluation is performed for a variety of material combinations and doping concentrations .
data measured at are chosen , because that temperature was most reliable to control for all samples .
all four @xmath1-doped material combinations displayed in fig .
[ fig : fig1-zt - vs - c ] ( a ) show an increasing power factor with doping concentration and a saturation at high .
the largest values are obtained for doped by f6-tcnnqwith a maximum of .
several orders of magnitude higher values are obtained for @xmath0-doped c60samples , as plotted in fig .
[ fig : fig1-zt - vs - c ] ( b ) .
this is attributed to the extraordinary high electron mobility of c60@xcite , determining the conductivity .
hence , c60is the most promising candidate for an organic small molecular te device of all studied host materials . applying the dopant results in the highest power factors with a maximum of at ( @xmath12 ) . at lower concentrations the dopant cr2(hpp)4)results in the best power factors .
interestingly , the dopants cr2(hpp)4 ) , w2(hpp)4)and aobyield almost identical values at @xmath13 .
the highest value obtained for c60@xmath0-doped by is comparable to the record reported for layers of c60@xmath0-doped by the inorganic dopant cs@xmath6co@xmath7 in an optimized sample geometry@xcite . for the purely organic doped layers of pentacene @xmath1-doped by f4-tcnq ,
the current record reported in literature@xcite is one order of magnitude lower , despite employing an optimized geometry .
-doped by of . ] for intrinsic c60 , the thermal conductivity has been reported to be almost temperature - independent between 30 and at a value around @xcite , which is 4 orders of magnitude lower than values observed for @xcite .
assuming that is mainly attributed to phonons and hence unaltered by introduction of dopant molecules , the value for intrinsic c60can be used to derive , which is shown by the right hand axes of fig .
[ fig : fig1-zt - vs - c](b ) .
a maximum of at is found for the sample of c60@xmath0-doped by of .
this value is 16% of the current record reported for of at room temperature@xcite .
as was found to increase with temperature for all investigated samples , it is interesting to study the temperature dependencies of ( equation ( [ eq : ztm ] ) ) and ( equation ( [ eq : eff ] ) ) for the sample of highest . here , the temperatures and , are , respectively , used during the seebeck measurements .
[ fig : eta - vs - t-106 ] shows that both and show an approximately linear increase with . raising from to , almost doubles while rises by 50% . at a typical te device operating temperature of , the values of and
are derived .
the efficiency reported here is rather low , compared to optimized polymer devices .
based on equation ( [ eq : eff ] ) , it is expected that increasing should result in an efficiency gain .
for example using @xmath14 and @xmath15 , hence instead of while keeping and hence , should result in a 16-fold efficiency gain , reaching .
however , since in our setup it was impossible to measure at such a large , this work requires a different experimental approach and will be done in future .
this simple estimation of the thermoelectric properties of doped organic small molecules yields that the studied @xmath1-doped material systems generates rather low power factors , while @xmath0-doped c60is a promising candidate for organic te devices .
the best results are obtained for c60doped by low cr2(hpp)4)or high doping concentrations . assuming the thermal conductivity of intrinsic c60is valid for doped layers as well , a maximum of at is derived , being 16% of the current record reported for optimized layers . and hence is found to increase with for this material class .
it is expected that applying a larger temperature difference of at an efficiency of 0.606% can be reached . in a future work ,
the thermal conductivity of @xmath0-doped c60layers needs to be studied to verify these results .
the author thanks karl leo and hans kleemann for fruitful discussions and novaled ag , germany for supplying the dopants f6-tcnnq , cr2(hpp)4)and w2(hpp)4 ) .
23ifxundefined [ 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 10.1038/nmat3013 [ * * , ( ) ] link:\doibase 10.1557/mrs2006.44 [ * * , ( ) ] link:\doibase 10.1038/nmat3635 [ * * , ( ) ] link:\doibase 10.1038/nmat3824 [ * * , ( ) ] link:\doibase 10.1021/ct3004436 [ * * , ( ) ] link:\doibase 10.1039/c2cp42710a [ * * , ( ) ] link:\doibase 10.1002/pssa.201228310 [ * * , ( ) ] link:\doibase 10.1016/0038 - 1098(94)90208 - 9 [ * * , ( ) ] link:\doibase 10.1063/1.122718 [ * * , ( ) ] link:\doibase 10.1126/science.259.5098.1145 [ * * , ( ) ] link:\doibase 10.1016/j.orgel.2012.09.024 [ * * , ( ) ] _ _ , http://www.dr.hut-verlag.de/9783843911771.html [ ] , ( ) link:\doibase 10.1016/j.orgel.2013.11.033 [ * * , ( ) ] link:\doibase 10.1063/1.3689778 [ * * , ( ) ] link:\doibase 10.1021/ja211382x [ * * , ( ) ] link:\doibase 10.1016/0038 - 1098(71)90096 - 2 [ * * , ( ) ] link:\doibase 10.1063/1.1735380 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.91.148301 [ * * , ( ) ] link:\doibase 10.1002/pssa.201330049 [ * * , ( ) ] link:\doibase 10.1002/adma.200502752 [ * * , ( ) ] link:\doibase 10.1063/1.3631633 [ * * , ( ) ] link:\doibase 10.1063/1.3456394 [ * * , ( ) ] link:\doibase 10.1088/0256 - 307x/25/6/076 [ * * , ( ) ] | doped films of organic small molecules are investigated with respect to their thermoelectric properties . a variety of hosts and dopants , for both @xmath0 and @xmath1-doping , are compared .
c60@xmath0-doped by cr2(hpp)4)or are found to be the most promising material systems with a maximum of at , assuming a doping - independent thermal conductivity due to phonon - based heat transport .
this value is 16% of the current record reported for optimized devices employing the doped polymer .
[ 1 ] [ ] |
the rapidly variable , aperiodic x - ray continuum emission from seyfert agn has long supported the notion that the x - ray emission originates from within a few tens of schwarzschild radii of the central supermassive black hole . in the late 1980 s , _ exosat _ probed x - ray variability on time scales of a few days or less , yielding power spectral density functions ( psds ) covering temporal frequencies down to @xmath6 hz .
these psds established the scale - invariant `` red - noise '' nature , i.e. , larger variability amplitudes toward larger time scales , characterized by a psd which increases toward lower temporal frequencies .
however , no characteristic time scale , such as a `` break '' in the psd power - law slope , was detected ( e.g. , lawrence et al .
1987 ; mchardy & czerny 1987 ; green , mchardy & lehto 1993 ) .
the _ rossi x - ray timing explorer _ ( _ rxte _ ) was launched in 1995 ; its unique attributes ( large effective area , high throughput , fast slewing , and flexible scheduling ) permitted multi - time scale monitoring campaigns that probed x - ray variability on time scales from hours to years ( temporal frequencies from @xmath7 hz ) .
the resulting seyfert broadband psds yielded evidence for breaks at temporal frequencies @xmath8 with psd power - law slopes breaking from @xmath92 to @xmath91 above and below @xmath8 , respectively ; the break frequencies corresponded to break time scales @xmath1 of a few days or less ( e.g. , edelson & nandra 1999 ; pounds et al .
2001 ; uttley , mchardy & papadakis 2002 ( hereafter u02 ) ; markowitz et al .
2003b ( hereafter m03 ) ; papadakis , reig & nandra 2003 ; mchardy et al . 2004 , 2005 ;
uttley & mchardy 2005 ) .
it was noted by m03 that a small sample of seyfert psds was consistent with a relation between @xmath1 and black hole mass @xmath3 estimated via reverberation mapping , in the sense that relatively larger - mass black holes are associated with psds with larger values of @xmath1 , albeit with some scatter . using a larger sample , mchardy et al .
( 2006 ) demonstrated that the scatter could be explained by an additional dependence of @xmath1 on @xmath10 , the accretion rate relative to eddington , in that for a given @xmath3 , sources accreting at relatively higher rates have psds shifted toward higher @xmath8 ( smaller values of @xmath1 ) . in the absence of an accurate estimate of @xmath3 by other means , x - ray timing can thus be used to estimate @xmath3 , if @xmath1 and bolometric luminosity @xmath11 are both known . in this paper , we present the broadband 310 kev psd of ic 4329a , an x - ray bright and variable seyfert located at a redshift of @xmath12 = 0.01605 ( willmer et al .
ic 4329a had been monitored in the optical with the intent of using reverberation mapping to estimate @xmath3 in this object , but the constraints were quite poor ( winge et al . 1996 ; peterson et al .
2004 ) . here , we demonstrate that there is significant evidence for a break in the x - ray psd of ic 4329a .
we also present psds for the 35 , 510 , 1020 , and 2040 kev bands .
section 2 reviews the multi - time scale sampling strategy and describes the observations , data reduction , and light curve sampling .
section 3 describes the methods of measurement and modeling of the psds .
the results of model fits to the psds are presented in @xmath134 . in @xmath135 , the psd break frequency is used to yield an estimate of @xmath3 based on x - ray timing , which we compare to estimates obtained by various other methods .
we followed the observing strategy of edelson and nandra ( 1999 ) , u02 , m03 , etc . , in which a source is monitored with
even , regular sampling on multiple time scales such that the resulting set of light curves yields individual psds covering complementary temporal frequency ranges . with the reasonable assumption
that the underlying psd shape has remained constant throughout all the monitoring observations and/or x - ray luminosity , we expect significant changes in the shapes of agn psds only on timescales of centuries to millennia . ] , the individual psds are combined to produce the final broadband psd .
ic 4329a was monitored with _
rxte _ once every 4.26 days ( 64 orbits ) for a duration of 4.3 years , from 2003 apr 8 to 2007 aug 7 ( modified julian day ( mjd ) 5273754319 ; observation identifiers 80152 - 03- * , 90154 - 01- * , 91138 - 01- * , and 92108 - 01- * ) .
this sampling , which probes variability on time scales from @xmath9 a week to a few years , is henceforth called `` long - term '' sampling .
each observation lasted approximately 1 ks . there were four gaps due to sun - angle constraints in october
november of each year ; each gap was @xmath14 days long .
more intensive monitoring with _ rxte _ was done to probe variability on time scales from several hours to a month ( `` medium - term '' sampling ) .
_ rxte _ observed ic 4329a once every @xmath15 ks ( three orbits ) for a duration of 34.1 days , from 2003 jul 10 at 18:05 ut to 2003 aug 13 at 20:35 ut ( mjd 52830.7552864.86 ; observation identifiers 80152 - 04- * ) .
again , each visit lasted @xmath9 1 ks .
finally , a continuous , 133 ks observation of ic 4329a with _ xmm - newton _ ( @xmath132.3 ) allowed us to probe variability on time scales from @xmath9an hour to 1 day .
the proportional counter array ( pca ; swank 1998 ) consists of five large - area , collimated proportional counter units ( pcus ) .
reduction of the pca data followed standard extraction and screening procedures , using heasoft version 6.5.1 software .
pca standard-2 data were collected from pcu 2 only ; pcus 1 , 3 and 4 have been known to suffer from repeated breakdown during on - source time , and pcu 0 lost its propane veto layer in 2000 .
data were rejected if they were gathered less than 10@xmath16 from the earth s limb , within 30 minutes of satellite passage through the south atlantic anomaly ( saa ) , if the satellite s pointing offset was greater than 0@xmath172 , or if electron2 was @xmath18 0.1 ) . as the pca is a non - imaging instrument ,
the background was estimated using the `` l7 - 240 '' background models , appropriate for faint sources ; see e.g. , markowitz , edelson & vaughan ( 2003a ) and edelson & nandra ( 1999 ) for details on pca background subtraction , the dominant source of systematic uncertainty ( e.g. , in total broadband count rate ) in these data .
spectral fitting for each observation was done using xspec version 11.3.2ag , assuming a galactic column of @xmath19 @xmath20 ( kalberla et al .
as the response of the pca slowly hardens slightly over time due to the gradual leak of xenon gas into the propane layer in each pcu , response files were generated for each separate observation .
fluxed light curves were extracted in the 310 kev band by fitting a power - law to that band only ( 210 kev is a `` standard '' hard x - ray measurement band , but the 23 kev effective area of the pca is extremely small , so we use 310 kev here ) .
light curves for the 35 and 510 kev sub - bands and the 1020 kev band were also extracted .
errors on each light curve point were derived from the standard error of 16-s count rate light curve bins within each observation .
the total number of data points after screening was 372 points for each long - term light curve , with 14.5@xmath21 of the data missing due to , e.g. , sun - angle constraints or screening .
each medium - term light curve contained 173 points , with only 4.1@xmath21 of the data missing .
the 310 kev light curves are plotted in figure 1 .
we also extracted 2040 kev pca light curves , as it would have been desirable to have pca data simultaneous with the 2040 kev high - energy x - ray timing experiment ( hexte ) light curves ( see below ) , but in this band , the source was only @xmath97@xmath21 of the total pca background ; @xmath222@xmath21 systematic uncertainties in the pca background on time scales of @xmath912 ks ( jahoda et al .
2006 , figure 29 ) yield uncertainties of @xmath23 uncertainties in the net source count rate , and so we do not consider the pca above 20 kev . the high - energy x - ray timing experiment ( hexte ) aboard _ rxte _ consists of two independent clusters ( a and b ) , each containing four nai(tl)/csi(na ) phoswich scintillation counters ( see rothschild et al . 1998 ) which share a common 1@xmath16 fwhm field of view .
each of the eight detectors has a net open area of about 200 @xmath24 .
source and background spectra were extracted from each individual _ rxte _
visit using science event data and standard extraction procedures .
the same good time intervals used for the pca data ( e.g. , earth elevation and saa passage screening ) were applied to the hexte data . to measure real - time background measurements ,
the two hexte clusters each undergo two - sided rocking to offset positions , in this case , to 1.5@xmath25 off - source , switching every 32 s. there is a galaxy cluster ( abell 3571 , located at a redshift of @xmath12 = 0.039 ) located about 2@xmath25 south of ic 4329a , at r.a .
= 13h 47.5 m , decl .
= 32@xmath16 52 m. this source is detected in the _ rxte _
all - sky slew survey ( xss ; revnivtsev et al .
2004 ) , which shows the 820 kev flux of this source to be about half that of ic 4329a ( at 2@xmath16 off - axis , the count rate in the pca is negligible , and thus the effect of contamination from abell 3571 on pca light curves and psds is negligible ; jahoda et al .
however , _
bepposax_pds observations have shown no detection of abell 3571 above 15 kev ( nevalainen et al .
the 820 kev emission seen by the xss must therefore be emission only between 8 and @xmath26 kev , and in this paper we use the 2040 kev band , so the presence of a3571 can be safely ignored as far as contaminating hexte background data obtained within @xmath912@xmath16 of the center of a3571 is concerned .
cluster a data taken during the following times were excluded , as the cluster did not rock on / off - source : 2004 dec 13 2005 jan 14 , 2005 dec 12 2006 jan 4 , during 2006 jan 25 , and after 2006 mar 14 .
detector 2 aboard cluster b lost spectral capabilities in 1996 ; these data were excluded .
light curves were extracted in 16 s bins in the 2040 kev band .
deadtime corrections were applied to account for cluster rocking , pulse analyzer electronics , and the recovery time following scintillation pulses caused by high - energy charged particles ; typically , the hexte deadtime is @xmath27 .
the light curves were then binned to every 12.79 days ( long - term ) or 50.2 ks ( medium - term ) .
this action minimized variability associated with background systematics .
the hexte background is relatively stable over long time scales but rapidly variable within a single satellite orbit as the spacecraft moves in the geomagnetic environment . for a typical background rate ,
the poisson error in a @xmath9600 s good time exposure is @xmath281@xmath21 ( gruber et al .
in the case of ic 4329a , the net source flux is roughly 9@xmath21 of the total background in the 2040 kev band , and each observation was only 1 ks in duration , yielding a @xmath2810@xmath21 systematic uncertainty in the net source count rate .
in addition , there are likely systematic uncertainties associated with the deadtime correction , @xmath290.5@xmath21 in a 1 ks duration observation .
errors on each point were determined by the standard error of the 16 s points . for observations using both clusters , light curves from clusters a and b were added
; all light curves are in units of ct s@xmath30 detector@xmath30 ( that is , ct s@xmath30 per 7 detectors when both clusters were in operation , or per 3 detectors when only cluster b was operating normally ; there were no observations where cluster a was the only cluster operating normally ) .
the long - term light curve had 125 points , with 13 points ( 10.4@xmath21 ) missing ; the medium - term light curve had 60 points , with no points missing .
the 2040 kev light curves are plotted in figure 1 .
the hexte gain and response are both very stable over time scales of years , permitting us to work in units of count rates .
however , the hexte light curves , plotted in figure 1 , have had count rates converted to fluxes for consistency with the way the pca light curves are plotted ; we use a conversion rate of 1 ct s@xmath30 per hexte detector in 2040 kev corresponding to 2.6 @xmath31 erg @xmath20 s@xmath30 for a source with a photon index of 1.7 .
visually , there appears to be some mild variability in the 2040 kev light curves not present in the 310 kev light curve , likely from the aforementioned systematic uncertainties associated with background subtraction and deadtime correction .
however , we note that the overall shape of the resulting long- and medium - term combined psd ( see @xmath134 ) is roughly consistent with that obtained for the 1020 kev band at the temporal frequencies probed , and we are confident that the systematic effects associated with hexte background subtraction and deadtime correction do not significantly affect our conclusions for the 2040 kev psd .
_ xmm - newton _ observed the nucleus of ic 4329a during 2003 aug 67 ( observation identifier 0147440101 ) , for a duration of 136 ks . in this paper ,
we only consider data collected using the european photon imaging camera ( epic ) pn camera .
light curves were extracted using xmm science analysis software version 7.1.0 , xselect version 2.4a , and the latest calibration files .
source photons were extracted from a circular region of radius 40@xmath32 ; backgrounds were extracted from circles of identical size , centered @xmath93@xmath33 away from the core .
hot , flickering , or bad pixels were excluded .
we inspected the 1013 kev pn background light curve for flares , and excluded periods when the 1013 kev background rate exceeded 0.1 ct s@xmath30 .
we extracted a light curve in the 310 kev band binned to 2000 s ; variability at shorter time scales was dominated by poisson noise .
the light curve had 67 points , with 3 points , or 4.5@xmath21 , missing due to the above background screening .
it is plotted in figure 1 .
light curves for the 35 and 510 kev sub - bands were also extracted , binned to 5000 s ( 27 points ) .
light curve sampling parameters , including mean net source and background count rates , for all light curves are listed in table 1 .
also listed in table 1 are variability amplitudes @xmath34 ( see vaughan et al .
2003 for definition of @xmath34 and its uncertainty ) for all light curves .
previous works have noted that , from 2 to @xmath920 kev , @xmath34 is generally observed to decrease with increasing photon energy ( e.g. , markowitz , edelson & vaughan 2003a ; miniutti et al .
2007 ) ; possible explanations include the dilution of the observed variability of the coronal power - law component by the compton reflection hump or `` pivoting '' of the power - law component .
the values of @xmath34 for the 35 , 510 , and 1020 kev bands are consistent with this trend ( in the case of the 2040 kev light curve , it is not immediately obvious how much variability is intrinsic and how much is due to the systematic effects discussed in @xmath132 ) .
table 2 lists psd measurement parameters . the 310 kev band has the best temporal frequency coverage ; this band and the 35 and 510 kev sub - bands each contain long- , medium- and short - term sampling . as _ xmm - newton _ lacks coverage above 12 kev , there is no short - term sampling to complement the 1020 kev and 2040 kev psds .
nonetheless , the 2040 kev psd presented here is , to our knowledge , the first x - ray psd of a seyfert above 20 kev .
psds were measured for each light curve separately , as described in @xmath133.1 .
however , as described in @xmath133.2 , these `` observed '' psds suffer from measurement distortion effects ( aliasing and red - noise leak ) and the errors are not well - determined .
a model - dependent monte carlo method was used to assign proper errors and to recover the intrinsic , underlying psd shape .
these steps are reviewed only briefly in the next two subsections ; for further details , the reader is referred to u02 and m03 .
initial psd construction closely followed @xmath133.1 of m03 .
light curves were linearly interpolated across gaps , though such gaps were rare .
each light curve s mean was subtracted .
periodograms were constructed using a discrete fourier transform ( e.g. , oppenheim & shafer 1975 ; see also @xmath133.1 of m03 for definitions of fourier frequencies @xmath35 and the periodogram ) .
following papadakis & lawrence ( 1993 ) and vaughan ( 2005 ) , the periodogram was logarithmically binned every factor of 1.6 in @xmath35 ( 0.20 in the logarithm ) to produce the observed psd @xmath36 ; the two lowest temporal frequency bins were widened to accommodate three periodogram points .
the @xmath95 lowest - temporal frequency bins in each individual psd typically contained @xmath3715 periodogram points , precluding us from assigning normal errors . the individual long- , medium- , and short - term psds were combined to yield the final , broadband observed psds @xmath36 , which are plotted in figure 2 , as well as in the top panels of figures 3 , 4 , and 5 in @xmath38 space .
no renormalization of the individual psds was done .
the constant level of power due to poisson noise was not subtracted from these psds , but instead modeled in the monte carlo analysis below .
visual inspection of figure 2 supports , at first glance , the notion that all five psds are at least roughly consistent in broadband shape .
however , to account for psd measurement effects ( the reader is referred to u02 and m03 for descriptions of aliasing and red - noise leak , and how the method accounts for them ) and to assign proper errors , we use a version of the monte carlo technique `` psresp '' introduced by u02 .
we summarize this technique as follows : one first assumes an underlying broadband model psd shape , and , for each individual observed light curve sampling pattern , one uses the algorithm of timmer & knig ( 1995 ) to generate @xmath39 light curves .
the simulated light curves are each resampled to match the observed light curve , then psds are calculated .
the model average psd @xmath40 , which accounts for the distortion effects , is calculated from the @xmath39 psds ; errors for each psd bin are assigned based on the rms spread of the individual simulated psds at a given temporal frequency bin .
the constant level of power due to poisson noise @xmath41 ( see table 2 ) is added to the model as @xmath42 , where @xmath43 and @xmath44 are the source and background count rates , respectively ; for non - continuously sampled light curves , this must be multiplied by the ratio of @xmath45 , the average sampling time , to the average exposure time per observation . for the hexte light curves , the estimates of @xmath41
were calculated taking into consideration the total number of good detectors ( e.g. , times when cluster a data were not used ) .
the goodness of fit is determined .
a statistic @xmath46 which compares the observed psd @xmath36 to the model average psd @xmath40 , using the errors from the model ( as opposed to using poorly - determined errors from the observed psd ) , is calculated .
the best - fitting normalization of the @xmath47 is found by renormalizing @xmath40 until the value of observed @xmath46 is minimized .
10,000 combinations of the long- , medium- , and short - term psds @xmath47 are randomly selected to model the @xmath46 distribution , comparing @xmath47 to @xmath40 .
the probability that the model psd can be rejected , @xmath48 , is defined as the percentile of the 10,000 simulated @xmath46 values exceeded by the value of the observed @xmath46 . a range of underlying psd model shapes is thus tested in this way , and the model with the lowest rejection probability ( highest likelihood of acceptance @xmath49 ) can be identified .
we tested psd model shapes consisting of simple unbroken power - law models and singly - broken power - law models ; the quality of each psd precludes more complex shapes such as single or multiple lorentzian components or quasi - periodic oscillations routinely modeled in the psds of x - ray binary systems .
the unbroken power - law model was of the form @xmath50 , where @xmath51 is the power - law slope and the normalization @xmath52 is the psd amplitude at @xmath53 , arbitrary chosen to be @xmath54 hz .
@xmath41 is added to each simulated psd but is not explicitly listed here since @xmath41 is different for each psd segment .
the model was tested by stepping through @xmath51 from 0.0 to 3.2 in increments of 0.1 , each time with @xmath39=300 simulations done to calculate @xmath40 .
the best - fitting models are plotted in the second panels of figures 3 , 4 and 5 .
the best - fitting values of @xmath51 , @xmath52 , and likelihood of acceptance @xmath55 are listed in table 3 .
the errors on @xmath51 correspond to values 1@xmath56 above the rejection probability @xmath57 for the best - fit value on a gaussian probability distribution ; for example , if the best - fit model had @xmath57=95.45@xmath21 ( 2.0@xmath56 on a gaussian probability distribution ) , errors correspond to @xmath57=99.73@xmath21 ( 3.0@xmath56 ) .
the errors on @xmath52 were determined from the rms spread of the @xmath58 randomly selected sets of simulated psds .
the high rejection probabilities for the 310 , 35 , and 510 kev psds ( each @xmath18 90@xmath21 ) and the residuals plotted in figures 3b and 4b suggest that a more complex psd model shape may be appropriate for these three psds .
the 1020 kev and 2040 kev psds cover less dynamic range in temporal frequency than their @xmath3710 kev counterparts due to the lack of short - term sampling ; the rejection probabilities are consequently much lower , @xmath3790@xmath21 , along with smaller residuals ( figure 5b ) .
to test for the presence of breaks in the psd , we then tested a singly - broken psd model shape of the form @xmath59 where the normalization @xmath60 is the psd amplitude at the break frequency @xmath8 , and @xmath61 and @xmath62 are the low- and high - frequency power law slopes , respectively , with the constraint @xmath63 .
( we also tested a more slowly - bending psd model of the form @xmath64 , but given the psd quality , there were degeneracies between @xmath8 , @xmath61 , and @xmath62 such that reasonable constraints on @xmath8 could not be attained . )
the range of slopes tested was 0.03.2 in increments of 0.1 .
break frequencies were tested in the log from 7.4 to 4.9 in increments of 0.1 , corresponding to @xmath65 in the linear scale .
100 simulated psds were used to determine @xmath40 .
the best - fit model parameters , along with likelihoods of acceptance @xmath66 , are listed in table 4 ; residuals are plotted in the bottom panels of figures 3 and 4 for the 310 , 35 , and 510 kev psds .
figure 6 shows contour plots of @xmath62 versus @xmath8 for these three psds at the respective best - fit values of @xmath61 .
we use the ratio of the likelihoods of acceptance @xmath67 between the broken and unbroken power law model fits to establish if a break is significant ( we use @xmath67 at least @xmath9 10 ) . for the 310 kev psd ,
we conclude that a break is significantly detected at @xmath68 hz = @xmath69 hz , which corresponds to a time scale of @xmath2 days .
breaks are also significantly detected in each 35 and 510 kev sub - band psd , with best values of @xmath8 also near @xmath54 hz . because of the limited dynamic range in temporal frequency for the 1020 and 2040 kev psds , the improvement in fit when adding a break to the model is not high , with @xmath70 .
parameters for the broken power - law model are thus listed in parentheses in table 4 for these two psds and data / model residuals are not plotted .
we next explored the possibility that all five psds could be consistent with the same psd shape , namely that of the 310 kev psd , with a low - frequency power law slope equal to 1.0 , a high frequency slope near 2 , and a break frequency @xmath71 hz . we re - tested the 35 and 510 kev psds with a broken power - law model , keeping @xmath61 fixed at 1.0 .
acceptable fits were obtained , with best - fit values for @xmath8 and @xmath62 consistent with those measured for the 310 kev psd ; results are listed in table 4 .
the 1020 and 2040 kev psds only cover up to roughly @xmath72 hz , and there is not much `` leverage '' above @xmath73 hz to either confirm or refute the presence of a break at those temporal frequencies .
however , the best - fit power law slopes in the unbroken power law fits are flat , 1.11.2 ( compared to 1.51.7 for the @xmath37 10 kev psds , which extend up to roughly @xmath74 hz , well above the best - fit 310 kev psd break frequency ) .
we conclude that the 1020 and 2040 kev psds are at least roughly consistent with the presence of breaks at or near that found in the @xmath3710 kev psds , and with the same @xmath61 , and that the @xmath1810 kev psds probe temporal frequencies primarily below the break frequency .
we have used complementary _ rxte _ and _ xmm - newton _ monitoring of the x - ray bright seyfert agn ic 4329a to measure the 310 kev psd , and we find significant evidence for a break in the modeled power - law slope at a temporal frequency of @xmath0 hz , which corresponds to a break time scale @xmath1 of @xmath2 days .
best - fit power law slopes above and below the break are @xmath75 and @xmath76 , respectively . for a discussion of candidate physical mechanisms to explain the turnover , the reader is referred to , e.g. , edelson & nandra ( 1999 ) and arvalo & uttley ( 2006 ) .
we can derive a new estimate for @xmath77 from the psd break , using the empirical relation between @xmath1 , bolometric luminosity @xmath11 , and @xmath77 quantified by mchardy et al .
( 2006 ) , log(@xmath1(days ) ) = 2.1 log(@xmath78 )
0.98log(@xmath79 erg s@xmath30 ) @xmath80 .
the average absorbed 310 kev flux from the long - term pca monitoring is @xmath81 erg @xmath20 s@xmath30 . to find the unabsorbed flux
, we used a spectral model in xspec consisting of a power law with photon index 1.74 ( markowitz , reeves & braito 2006 ) , absorbed by the galactic column , and five layers of absorption modeled by steenbrugge et al .
( 2005 ) : a column of cold gas intrinsic to the host galaxy with column density @xmath82 @xmath20 , and four ionized absorbers , with column densities 1.3 , 0.32 , 6 , and 2 @xmath83 @xmath20 and ionization parameters log@xmath84 , where @xmath85 is the isotropic 11000 ryd ionizing continuum luminosity , @xmath86 is the electron number density , and @xmath87 is the distance from the central continuum source to the absorbing gas . ] = 1.37 , 0.56 , 1.92 , and 2.70 erg cm s@xmath30 , respectively .
( a fifth ionized absorber quantified by markowitz , reeves & braito 2006 , with best - fit modeled values of log@xmath84 = 3.73 erg cm s@xmath30 and @xmath88 @xmath20 , is expected to produce only a narrow fe k absorption feature and is ignored . ) to model each ionized absorber , an xstar component was used ( bautista & kallman 2001 ) . using this model , a luminosity distance of 78.6 mpc ( mould et al .
2000 ) , and assuming a cosmology with @xmath89 km s@xmath30 mpc@xmath30 and @xmath90 , the unabsorbed , rest - frame 210 kev luminosity @xmath91 is thus estimated to be @xmath92 erg s@xmath30 . using marconi et al .
( 2004 , their figure 3b ) , @xmath93 , and @xmath94 erg s@xmath30 . using the relation of mchardy et al .
( 2006 ) , we derive @xmath95 ( a 7@xmath21 uncertainty in the distance to ic 4329a ( mould et al . 2000 ) translates into an additional @xmath96 uncertainty on @xmath3 ) .
the accretion rate relative to eddington , @xmath10 , is thus estimated to be @xmath5 .
this estimate of @xmath3 is in agreement with several estimates obtained by other methods .
reverberation mapping has been known to yield highly accurate estimates of @xmath3 , but only if the data are of sufficiently high quality . in the case of ic 4329a , the estimate of @xmath3
is poorly constrained due to relatively low - quality optical spectra , and formally only an upper limit , @xmath97 ( see peterson et al .
2004 for details ) . however , many other methods of estimating @xmath3 in ic 4329a yield estimates closer to @xmath98 .
this includes other methods based on estimating the distance @xmath99 from the central continuum source to the broad line region , such as the empirical relation noted by kaspi et al .
( 2000 ) between @xmath99 and the optical continuum luminosity , `` recalibrated '' by vestergaard & peterson ( 2006 ; their equation 5 ) . using the value of h@xmath100 fwhm ( rms ) from wandel , peterson & malkan ( 1999 ) , @xmath101 km s@xmath30 , and a value for @xmath102@xmath103(5100 ) of @xmath104 erg s@xmath30 from kaspi et al .
( 2000 ) , the vestergaard & peterson ( 2006 ) equation yields @xmath105 .
vestergaard & peterson ( 2006 ; their equation 6 ) also provide a recalibrated formula for estimates of @xmath3 based on h@xmath100 luminosity and optical continuum luminosity .
winge et al .
( 1996 ) report a mean h@xmath100 flux of @xmath106 erg @xmath20 s@xmath30 ; assuming a luminosity distance of @xmath107 mpc ( mould et al .
2000 ) yields an h@xmath100 luminosity of @xmath108 erg s@xmath30 .
the resulting estimate of @xmath3 is thus @xmath109 .
a virial estimate based on the photoionization method ( wandel , peterson & malkan 1999 ) is @xmath110 .
the black hole mass can also be estimated via the well - known relation between stellar velocity dispersion @xmath111 and @xmath3 .
we use equation 19 of tremaine et al .
( 2002 ) , @xmath112 ( @xmath113(@xmath114 km s@xmath30))@xmath115 .
oliva et al . ( 1999 ) report @xmath116 km s@xmath30 or @xmath117 km s@xmath30 using 1.59@xmath43 m si or 1.62 @xmath43 m co(6,3 ) features , respectively .
using the average of these two values yields @xmath118 . finally , nikolajuk , papadakis & czerny ( 2004 ) suggested a prescription to estimate @xmath3 based on short - term ( @xmath119 day duration ) x - ray variability amplitude measurements , provided that the variability amplitude measured probes temporal frequencies above the psd break and that the psd power law slope at high temporal frequencies is 2 ; this assumption is consistent with the current psd results .
nikolajuk , czerny & papadakis ( 2004 ) estimated @xmath120 based on short - term _ rxte _ light curves , and markowitz , reeves & braito ( 2006 ) estimated @xmath121 using a 2.512 kev light curve from the 2003 _ xmm - newton _ pn observation .
several of the x - ray timing studies discussed in @xmath131 have noted the remarkable similarity between seyferts and black hole x - ray binaries in terms of their broadband psd shapes ; furthermore , the @xmath1@xmath77@xmath11 relation extrapolates down to stellar - mass black holes over 67 orders of magnitude in both @xmath3 and x - ray luminosity .
however , it is not immediately obvious from the shape of ic 4329a s psd and from the derived value of @xmath10 if ic 4329a is an analog to a high / soft or low / hard state black hole x - ray binary system .
there is not enough temporal frequency coverage below the break in the psd of ic 4329a to determine if the power - law slope of @xmath91 persists for several decades of temporal frequency below the break , which would identify it as an analog of cyg x-1 in the high / soft state ( e.g. , axelsson , borgonovo & larsson 2005 ) ; monitoring covering temporal frequencies down to and below @xmath122 hz would be needed to determine the ultra - low temporal frequency psd shape of ic 4329a .
the psds presented here are roughly consistent with same underlying broadband psd shape at all energies probed : a break near @xmath123 hz , and power - law slopes of @xmath124 and @xmath125 below and above the break frequency , respectively .
the 1020 and 2040 kev psds are probing temporal frequencies primarily below the break frequency but are consistent with the presence of a break in the psd up to 2040 kev .
arvalo & uttley ( 2006 ) noted that an increase in @xmath8 with photon energy is expected in a model incorporating inwardly propagating variations in the local mass accretion rate modifying the central x - ray emission ( lyubarskii 1997 , kotov et al .
2001 ) , with relatively harder x - ray bands associated with more centrally concentrated emissivity profiles .
however , no significant breaks were detected in the 1020 and 2040 kev psds .
the current 35 and 510 kev psds lack the necessary high temporal frequency resolution to discern any dependence of @xmath8 on photon energy . a much more dense temporal frequency sampling covering at least the range @xmath126 hz would be required to adequately test any energy dependence of @xmath8 in this object .
a.m. thanks the _ rxte _ science operations staff , particularly the _ rxte _ schedulers for ensuring that the long - term monitoring observations were scheduled so evenly all these years .
this work has made use of heasarc online services , supported by nasa / gsfc , the nasa / ipac extragalactic database , operated by jpl / california institute of technology under contract with nasa , and the nist atomic spectra database .
a.m. acknowledges financial support from nasa grant nas5 - 30720 .
arvalo , p. & uttley ,
mnras , 367 , 801 axelsson , m. , borgonovo , l. & larsson , s. 2005 , a&a , 438 , 999 bautista , m.a . & kallman , t. 2001 , apjs , 134 , 139 edelson , r. & nandra , k. 1999 , apj , 514 , 682 green , a. , mchardy , i. & lehto , h. 1993 , mnras , 265 , 664 gruber , d.e . ,
blanco , p.r .
, heindl , w.a . , pelling , m.r .
, rothschild , r.e . & hink , p.l .
1996 , a&as , 120 , 641 jahoda , k. , markwardt , c.b . ,
radeva , y. , rots , a. , stark , m.j . , swank , j. , strohmayer , t. & zhang , w. 2006 , apjs , 163 , 401 kalberla , p.m.w .
2005 , a&a , 440 , 775 kaspi , s. , smith , p.s .
, netzer , h. , maoz , d. , jannuzi , b.t .
& giveon , u. 2000 , apj , 533 , 631 kotov , o. , churazov , e. & gilfanov , m. 2001 , mnras , 327 , 799 lawrence , a. , watson , m. , pounds , k. & elvis , m. 1987 , nature , 325 , 694 lyubarskii , y.e . 1997 , mnras , 292 , 679 marconi , a. , risaliti , g. , gilli , r. , hunt , l.k . , maiolino , r. & salvati , m. 2004 , mnras , 351 , 169 markowitz , a. , edelson , r. & vaughan , s. 2003a , apj , 598 , 935 markowitz , a. , edelson , r. , vaughan , s. , et al .
2003b , apj , 593 , 96 ( m03 ) markowitz , a. , reeves , j.n . ,
george , i.m .
, et al . , 2009 ,
apj , 691 , 922 markowitz , a. , reeves , j.n . & braito , v. 2006 , apj , 646 , 783 mchardy , i.m . & czerny , b. 1987 , nature , 325 , 696 mchardy , i.m . ,
gunn , k.f .
, uttley , p. & goad , m.r .
2005 , mnras , 359 , 1469 mchardy , i.m .
, koerding , e. , knigge , c. , uttley , p. & fender , r.p .
2006 , nature , 444 , 730 mchardy , i.m . ,
papadakis , i.e. , uttley , p. , page , m.j .
& mason , k.o .
2004 , mnras , 348 , 783 miniutti , g. , fabian , a.c . , anabuki , n. , et al .
2007 , pasj , 59s , 315 mould , j.r . ,
huchra , j.p . ,
freedman , w.l .
, et al . 2000 , apj , 529 , 786 nevalainen , j. , oosterbroek , t. , bonamente , m. & colafrancesco , s. 2004 , apj , 608 , 166 nikolajuk , m. , papadakis , i. & czerny , b. 2004 , mnras , 350 , l26 oliva , e. , origlia , l. , maiolino , r. & moorwood , a.f.m . 1999 ,
a&a , 350 , 9 oppenheim , a. & shafer , r. 1975 , `` digital signal processing '' ( prentice - hall publishing ) papadakis , i.e. & lawrence , a. 1993 , mnras , 261 , 612 papadakis , i.e. , reig , p. & nandra , k. 2003 ,
mnras , 344 , 993 peterson , b.m . ,
ferrarese , l. , gilbert , k.m . , et al .
2004 , apj , 613 , 682 pounds , k. , edelson , r. , markowitz , a. , & vaughan , s. 2001 , apj , 550 , l15 revnivtsev , m. , sazonov , s. , jahoda , k. & gilfanov , m. 2004 , a&a , 418 , 927 rothschild , r. , blanco , p.r . , gruber , d.e .
, heindl , w.a . , macdonald , d.r . ,
marsden , d.c .
, pelling , m.r . & wayne , l.r .
1998 , apj , 496 , 538 steenbrugge , k.c . ,
kaastra , j.s . ,
sako , m. , branduardi - raymont , g. , behar , e. , paerels , f.b.s . ,
blustin , a.j . & kahn , s.m .
2005 , a&a , 432 , 453 swank , j. 1998 , in nuclear phys .
b ( proc . suppl . ) : the active x - ray sky : results from bepposax and rossi - xte , rome , italy , 1997 october 21 - 24 , eds .
l. scarsi , h. bradt , p. giommi , & f. fiore , nucl .
b suppl .
( the netherlands : elsevier science b.v . ) , 69 , 12 timmer , j. & knig , m. 1995 , a&a , 300 , 707 tremaine , s. , gebhardt , k. , bender , r. , et al .
2002 , apj , 574 , 740 uttley , p. & mchardy , i.m .
2005 , mnras , 363 , 586 uttley , p. , mchardy , i.m .
& papadakis , i.e. 2002 , mnras , 332 , 231 ( u02 ) vaughan , s. 2005 , a&a , 431 , 391 vaughan , s. , edelson , r. , warwick , r. & uttley , p. 2003 , mnras , 345 , 1271 vestergaard , m. & peterson , b.m .
2006 , apj , 641 , 689 wandel , a. , peterson , b.m . & malkan , m. 1999 , apj , 526 , 579 willmer , c.n.a .
, focardi , p. , chan , r. , pilligrini , p.s . & nicolaci da costa , l. 1991 , aj , 101 , 57 winge , c. , peterson , b.m . ,
pastoriza , m.g . & storchi - bergmann , t. 1996 , apj , 469 , 648 llccccccc 310 & long & pca & 4.26 d & @xmath81 & 11.57 & 2.99 & @xmath127 + & medium & pca & 17.1 ks & @xmath128 & 10.12 & 2.98 & @xmath129 + & short & pn & 3000 s & @xmath130 & 4.86 & 0.03 & @xmath131 + 35 & long & pca & 4.26 d & @xmath132 & 4.94 & 1.05 & @xmath133 + & medium & pca & 17.1 ks & @xmath134 & 4.28 & 1.05 & @xmath135 + & short & pn & 5000 s & @xmath136 & 2.77 & 0.02 & @xmath137 + 510 & long & pca & 4.26 d & @xmath138 & 6.63 & 1.94 & @xmath139 + & medium & pca & 17.1 ks & @xmath140 & 5.84 & 1.94 & @xmath141 + & short & pn & 5000 s & @xmath142 & 2.06 & 0.01 & @xmath143 + 1020 & long & pca & 4.26 d & @xmath144 & 2.33 & 2.15 & @xmath145 + & medium & pca & 17.1 ks & @xmath146 & 2.07 & 2.14 & @xmath147 + 2040 & long & hexte & 12.79 d & @xmath148 & 0.32 & 3.54 & @xmath149 + & medium & hexte & 50.2 ks & @xmath150 & 0.31 & 3.44 & @xmath151 llcc 310 & long & @xmath152 & 80.1 + & medium & @xmath153 & 4.27 + & short & @xmath154 & 0.41 + 35 & long & @xmath152 & 181 + & medium & @xmath153 & 9.72 + & short & @xmath155 & 0.72 + 510 & long & @xmath152 & 144 + & medium & @xmath153 & 7.62 + & short & @xmath155 & 0.98 + 1020 & long & @xmath152 & 608 + & medium & @xmath153 & 32.8 + 2040 & long & @xmath156 & 5940 + & medium & @xmath157 & 180 lcccccc 310 & @xmath168 & @xmath69 & @xmath169 & @xmath170 & 87.8 & 39.9 + 35 & @xmath171 & @xmath172 & @xmath173 & @xmath174 & 97.1 & 10.3 + & 1.0(fixed ) & @xmath175 & @xmath176 & @xmath177 & 90.4 & 9.6 + 510 & @xmath178 & @xmath179 & @xmath180 & @xmath181 & 82.5 & 20.6 + &
1.0 ( fixed ) & @xmath182 & @xmath183 & @xmath184 & 79.2 & 19.8 + 1020 & ( @xmath185 ) & ( @xmath186 ) & ( @xmath187 ) & ( @xmath188 ) & ( 29.4 ) & ( 1.7 ) + 2040 & ( @xmath189 ) & ( @xmath190 ) & ( @xmath191 ) & ( @xmath192 ) & ( 75.9 ) & ( 3.4 ) | we present the x - ray broadband power spectral density function ( psd ) of the x - ray - luminous seyfert ic 4329a , constructed from light curves obtained via _ rossi x - ray timing explorer _ monitoring and an _ xmm - newton _ observation . modeling the 310 kev psd using a broken power - law psd shape , a break in power - law slope
is significantly detected at a temporal frequency of @xmath0 hz , which corresponds to a psd break time scale @xmath1 of @xmath2 days . using the relation between @xmath1 , black hole mass @xmath3 , and bolometric luminosity as quantified by mchardy and coworkers
, we infer a black hole mass estimate of @xmath4 and an accretion rate relative to eddington of @xmath5 for this source .
our estimate of @xmath3 is consistent with other estimates , including that derived by the relation between @xmath3 and stellar velocity dispersion .
we also present psds for the 1020 and 2040 kev bands ; they lack sufficient temporal frequency coverage to reveal a significant break , but are consistent with the same psd shape and break frequency as in the 310 kev band . |
statistical aspects of the evolution of languages have attracted , in the last few years , a great deal of attention among physicists and mathematicians.@xcite one of the better established quantitative empirical facts about extant languages is their size distribution , namely , the frequency of languages with a given number of speakers . naturally , explaining the origin of this distribution is one the aimed goals of mathematical modeling in this field .
recent work has built up on variations of two basic models of language evolution , schulze s model@xcite and viviane s model,@xcite both of them mostly focused on the effects of mutation of linguistic features which give rise to new languages , and including the possibility of language extinction .
these models , however , disregard the fact that , over periods which are short as compared with the typical time scales of language evolution , the speakers of a given language can substantially vary in number just by the effect of population dynamics . for instance , in the last five centuries a period which includes the culturally devastating european invasion of the rest of the globe perhaps 50 % of the world s languages went extinct ( among them , two thirds of the 2,000 preexisting native american languages@xcite ) or changed drastically . in the same period ,
however , the world s population grew by a factor of twelve or more.@xcite demographic effects have been very recently incorporated to a model of language evolution by tuncay,@xcite in the form of a stochastic multiplicative model for population growth ( see also ref . ) . with suitable tuning of its several parameters , tuncay s model is able to reasonably reproduce the observed distribution of language sizes , as a result of numerical simulations.@xcite the complexity of the model which , in its full form,@xcite includes population growth along with language inheritance , branching , assimilation , and extinction makes it however difficult to identify the specifc mechanism that shapes the distribution of sizes . in this paper ,
i show that the empirical distribution of language sizes can be accurately explained taking into account just the effect of demographic processes .
this possibility was already pointed out , for the specific case of the languages of new guinea , by novotny and drozd.@xcite i propose a two - parameter stochastic model where the populations speaking different languages evolve independently of each other . during the evolution which , in the realization of the present model , is assumed to span 1,000 years language creation , assimilation , and extinction are disregarded .
this assumption does not discard mutations inside a given language , which may lead to its internal evolution , but each language preserves its identity as a cultural and demographic unit along the whole period .
the model is analytically tractable , and its two parameters can be fitted _ a priori _ from empirical data .
numerical simulations confirm the prediction that the distribution is essentially independent of details in the initial condition the distribution of sizes ten centuries ago so that , in a sense , the present distribution is the unavoidable consequence of just demographic growth .
the model is further validated by showing that the distribution of sizes of languages belonging to a given family has the same shape as the overall distribution .
these results strongly suggest that population dynamics is a necessary ingredient in models of linguistic evolution .
it is a well established empirical fact that the frequency of languages with a given number of speakers @xmath0 is satisfactorily approximated by a log - normal distribution.@xcite accordingly , the distribution of language sizes as a function of the log - size , @xmath1 , is approximately given by a gaussian , @xmath2,\ ] ] where @xmath3 and @xmath4 are , respectively , the mean value and the mean square dispersion of @xmath5 .
the quantity @xmath6 gives the fraction of languages with log - sizes in @xmath7 .
significant departures from the log - normal distribution are limited to small language sizes , up to the order of a few tenths speakers.@xcite ethnologue statistical summaries,@xcite whose data correspond to collections done mostly during the 1990s , list the number of languages with sizes within decade bins ( 1 to 9 speakers , 10 to 99 speakers , 100 to 999 speakers , and so on ) , and give the number of speakers within each bin .
the total number of languages in the list is @xmath8 , accounting for an overall population a little above @xmath9 speakers .
the sizes of @xmath10 languages of the database are unknown . in terms of the distribution @xmath11 ,
the number of languages in the bin between @xmath12 and @xmath13 speakers is @xmath14 while the total population speaking the languages in the same bin is @xmath15 the values of @xmath16 and @xmath17 provided by the ethnologue statistical summaries can thus be used to estimate the parameters @xmath3 and @xmath4 in the distribution @xmath11 of eq .
( [ gauss ] ) .
least - square fitting yields @xmath18 figure [ f1 ] shows , as a histogram over the variable @xmath5 , ethnologue s data for @xmath16 .
the curve is a plot of the function @xmath19 with the parameters of eq .
( [ fit ] ) , whose integral over the histogram bins approximates the empirical values of @xmath16 . for easier comparison with the histogram , @xmath19
is further multiplied by the bin width @xmath20 .
the aim in the following is to provide a model which explains the fitted distribution @xmath11 .
the log - normal shape of the distribution of language sizes suggests that a multiplicative stochastic process is at work in the evolution of the number of speakers of each language .
this is in turn consistent with the hypothesis that , over sufficiently long time scales , the population speaking a given language evolves autonomously , driven just by demographic processes.@xcite let @xmath21 be the number of speakers of language @xmath22 at time @xmath23 , and assume that at time @xmath24 one year later , say the population has changed to@xcite @xmath25 where the growth rate @xmath26 is a positive stochastic variable drawn from some specified distribution .
as shown below , the mean value and the mean square dispersion over this distribution can be estimated from empirical data . in terms of the initial population @xmath27 , the number of speakers at time
@xmath23 is @xmath28 i suppose now that , during the whole @xmath23-step process , the distribution of the growth rate @xmath26 is ( i ) the same for all languages , and ( ii ) does not vary with time .
moreover , ( iii ) no language is created or becomes extinct .
admittedly , these are rather bold assumptions for the world s history during the last 1,000 years . however , in view of the lack of reliable data over such period , they are at least justified by the sake of simplicity .
i identify the evolution of the world s languages as @xmath29 realizations of the multiplicative stochastic process ( [ pt0 ] ) . by virtue of assumption ( i ) ,
all the realizations are statistically equivalent . in this interpretation , the present distribution of log - sizes , given by eqs .
( [ gauss ] ) and ( [ fit ] ) , is the probability distribution for the variables @xmath30 obtained from those realizations .
thus , my aim is to quantitatively relate the distribution @xmath11 to the outcome of the stochastic process . the total population @xmath31 at time @xmath23 is @xmath32 averaging this expression over realizations of the stochastic variable @xmath26 , and assuming that the growth rate is not self - correlated in time , we find @xmath33 , where @xmath34 is the mean growth rate and @xmath35 is the initial total population .
in order to apply this analysis to the world s population in the last ten centuries , let us take @xmath36 , which is the estimated population by the year 1000.@xcite ascribing the total population accounted for by ethnologue s data to the year 2000 , and associating this number with the population averaged over realizations of the growth rate , we have @xmath37 and @xmath38 .
this makes it possible to evaluate the mean growth rate per year as @xmath39 to evaluate the dispersion of the growth rate , it is useful to introduce its relative deviation with respect to the average , @xmath40 , as @xmath41 .\ ] ] the average value and the mean square dispersion of the deviation @xmath40 are , respectively , @xmath42 where @xmath43 is the mean square dispersion of the growth rate . assuming that @xmath40 is always sufficiently small to approximate @xmath44 , eq .
( [ pt ] ) can be rewritten for the log - size @xmath45 as @xmath46 .\ ] ] this equation shows explicitly that , besides the deterministic growth given by the term @xmath47 , the evolution of the logarithm of the population speaking a given language is driven by an additive stochastic process .
thus , by virtue of the central limit theorem,@xcite the distribution @xmath11 must converge to a gaussian like in eq .
( [ gauss ] ) , starting from any distribution of initial log - sizes @xmath48 . for the time
being , however , the question remains whether the times relevant to the process are enough to allow for the development of the gaussian shape and , in particular , to suppress any effect ascribable to a specific initial distribution .
unfortunately , the initial sizes @xmath27 the number of speakers of each language 1,000 year ago are not known for most languages .
however , their effect on the present size distribution can be readily assessed .
averaging eq .
( [ qt ] ) over realizations of the stochastic process and over the distribution of initial log - sizes and always assuming that the deviations @xmath40 are small yields , for the mean square dispersion of @xmath49 , @xmath50 where @xmath51 is the mean square dispersion of the initial log - sizes .
the empirical estimation for @xmath52 is the value of @xmath4 given in eq .
( [ fit ] ) . in turn
, an upper bound can be given for @xmath51 , as the maximal mean square dispersion in the log - size distribution of @xmath29 languages with a total population of @xmath36 speakers , and at least one speaker per language .
this maximal dispersion is obtained with @xmath53 languages with exactly one speaker , and just one language with the remaining @xmath54 speakers .
clearly , this is an unlikely distribution for the languages 1,000 years ago , but represents the worst - case instance , with the largest contribution of the initial condition to the present dispersion of log - sizes . in this extreme situation ,
the estimation for the initial mean square dispersion is @xmath55 .
meanwhile , according to eq .
( [ fit ] ) , @xmath56 . in the right - hand side of eq .
( [ sq2 ] ) , therefore , the largest contribution by far is given by the second term , and that of the initial distribution is essentially negligible .
this makes it possible to calculate the mean square dispersion of the deviations @xmath40 as @xmath57 for @xmath38 , thus completing the statistical characterization of the growth rate per year , @xmath26 .
note that this value of @xmath58 is in agreement with the assumption of small relative deviations in the growth rate . summarizing the results of this section ,
i have argued that the present log - normal distribution of language sizes can be seen as the natural consequence of population dynamics driven by a stochastic multiplicative process , eq .
( [ pt0 ] ) , where the evolution of each language is interpreted as a realization of the process . using data on the total population growth during the last 1,000 years a period over which i neglect language birth and death and fitting only one parameter ( @xmath58 ) from the distribution itself , i was able to statistically characterize the growth rate per year which explains the present distribution , giving its mean value and mean square dispersion .
also , i have advanced that the dispersion of language sizes ten centuries ago has essentially no effect on its present value .
it is now useful to validate these conclusions with numerical realizations of the model , and with applications within language families .
in this section , i present results for series of @xmath29 numerical realizations of the multiplicative stochastic process ( [ pt0 ] ) .
the mean value and the mean square dispersion of the growth rate @xmath59 are fixed according to the values estimated in section [ sect3 ] , eqs .
( [ alfaav ] ) and ( [ sigmad ] ) . in order to speed up the computation , individual values of @xmath59
are drawn from a square distribution centered at @xmath34 , with a width which insures the correct mean square dispersion . in agreement with my main assumptions ,
i avoid the possibility that languages die out by replacing the absorbing boundary at @xmath60 , below which a language should become extinct , by a reflecting boundary . in view of the discussion in the previous section ,
the convergence of the distribution of log - sizes to a gaussian is guaranteed by the central limit theorem .
the emphasis in the simulations is thus put on the possible effects of the distribution of initial sizes @xmath61 .
figure [ f2 ] shows , as normalized histograms , numerical results for single series of @xmath62 realizations of the stochastic process after @xmath38 steps , from four different initial conditions . in ( a ) , all the languages have exactly the same initial size @xmath63 . in ( b ) , the initial sizes are uniformly distributed between @xmath64 and @xmath65 . in ( c ) , the distribution of initial sizes is also uniform , but spans the interval @xmath66 .
finally , in ( d ) the initial distribution is more heterogeneous , with half the languages having size @xmath67 and the other half having size @xmath68 .
the parameters @xmath63 , @xmath65 , @xmath69 , and @xmath67 , which characterize these initial distributions , are fixed by the condition that the total population is @xmath70 .
the curve in all plots is the gaussian of eq .
( [ gauss ] ) with the parameters of eq .
( [ fit ] ) .
the agreement in cases ( a ) to ( c ) is excellent . only in case
( d ) , where the initial distribution is specially heterogeneous and , certainly , not a likely representation of the distribution of language sizes ten centuries ago the deviations are larger , though the agreement is still very reasonable .
the difference between the numerical results of case ( d ) and the expected gaussian function resides not only in the width of the distribution but also in its mean value . to a much lesser extent , this discrepancy is also visible in case ( b ) .
this shift between the distribution peaks can be understood in terms of the average of eq .
( [ qt ] ) over both the realizations of the growth rate @xmath59 and the initial log - sizes @xmath71 , @xmath72 besides the contribution of the multiplicative stochastic process , given by the term proportional to the time @xmath23 , the mean value @xmath73 in eq .
( [ mq ] ) depends on the average initial log - size @xmath74 . in spite of the fact that the total initial population and the number of languages are the same for all simulations
which always gives the same average size per language the average log - size depends on the specific initial distribution .
thus , the final mean values for different initial conditions are generally shifted with respect to each other . as a consistency test for the assumption that languages do not die out along the evolution period considered here , i have also run simulations taking into account the absorbing boundary at @xmath60 .
namely , when the size of a language decreases below one speaker during its evolution , it is considered to become extinct and that particular realization of the stochastic process is interrupted . among the four initial conditions considered above , those who undergo larger extinctions are , not unexpectedly ,
( b ) and ( d ) as they produce the largest shifts to the left in the log - size distributions . in both cases , however , the fraction of extinct languages is around 1 % , which validates the above assumption quite satisfactorily . a crucial consequence of the hypotheses on which the present model is based in particular
, the mutual independence of the size evolution of different languages is that its predictions should hold not only for the ensemble of all the world s languages , but also for any sub - ensemble to which the homogeneity assumptions ( i ) and ( ii ) reasonably apply . in other words ,
the final log - normal shape of the size distribution should also result from the evolution of , say , the languages of a given geographical region , or belonging to a given language family .
this can be readily assessed from empirical data on the number of speakers of individual languages and , in fact , has already been pointed out for a set of some 1,000 new guinean languages.@xcite here , i analyze the size distribution of languages belonging to each one of the four largest families , according to ethnologue s classification.@xcite population data for individual languages were obtained from ethnologue s online databases .
figure [ f3 ] shows histograms of log - sizes for those families . to ease the comparison , the column width and the horizontal scale are the same as in fig .
curves stand for least - square fittings with gaussian functions as in eq .
( [ gauss ] ) .
the resulting parameters are @xmath75 and @xmath76 for niger - congo ; @xmath77 and @xmath78 for austronesian ; @xmath79 and @xmath80 for trans - new guinea ; and @xmath81 and @xmath82 for indo - european . the quality of the agreement between the data and the gaussian fitting is clearly comparable to that of the whole language ensemble in fig . [ f1 ] . note the interesting fact that the mean square dispersion @xmath4 which , according to the present model , results from the dispersion in the population growth rate is sensibly larger for the indo - european family than for the other three
. surely , this is a direct consequence of the highly diverse fate of european languages in the last few centuries . in any case , the four mean square dispersions are not far from the overall value given in eq .
( [ fit ] ) .
in this paper , i have argued that the present log - normal distribution of language sizes is essentially a consequence of demographic dynamics in the population of speakers of each language .
in fact , an isolated population can largely vary in number within time scales which are short as compared with those involved in substantial language evolution . to support this suggestion ,
i have proposed a stochastic multiplicative process for the population dynamics of individual languages , where language birth and death are disregarded . within some bold assumptions on the geographical and temporal homogeneity of the process
, the model is completely specified by two parameters , which give the average growth rate of the population and its mean square dispersion .
the average growth rate is completely defined by the initial and the final world population .
i have chosen to apply the model on the period spanning the last 1,000 years , for which reliable data on the world s population growth are available .
the mean square dispersion of the growth rate is the only parameter which i fitted in an admittedly _ ad - hoc _ manner , using the present distribution of language sizes .
it seems unlikely to find estimations for this parameter from independent historical sources , which would require to have reliable records on the population change year by year .
note that the dispersion in the growth rate is determined by a variety of factors , including fluctuations in birth and mortality frequencies and migration events.@xcite once the two parameters are fitted , the model is able to produce , as the result of evolution along ten centuries , excellent predictions of the present distribution of language sizes .
numerical simulations show that the final distribution is largely independent on the initial condition .
this emphasizes the point that , irrespectively of the long - range historical processes that may have determined the distribution of language sizes of 1,000 years ago including language birth and death , branching , mutation , competition , assimilation , and/or replacement population dynamics is by itself able to explain the present distribution .
this conclusion had already been advanced for the case of new guinean languages in ref . .
in fact , realizing that the same log - normal profile is found in the size distribution inside language families , is a further validation of the present model . in view of the present arguments
, one can moreover safely assert that the distribution of language sizes was already a log - normal function , with different parameters , in year 1000 .
it is clear that in the last few years with the advent of a host of new mechanisms of globalization which endanger cultural diversity many , or most , of the world s languages are threatened by the risk of extinction.@xcite this risk is peculiarly acute for those languages whose number of speakers is below a few hundreds including the range where the distribution of sizes differs from the log - normal profile ( cf . fig .
it should be a program of obviously urgent interest to study in detail what are the relevant processes at work in that range , even if they escape the domain of the statistical physicists approaches .
i am grateful to susanna c. manrubia , who brought to my attention the significance of the distribution of language sizes to the problem of language evolution in 1996 .
also , i acknowledge critical reading of the manuscript by a. a. budini .
l. campbell and m. mithun , eds . , _ the languages of native america : a historical and comparative assessment _
( university of texas press , austin , 1979 ) .
see also http://www.nativeamericans.com/natives.htm . | it is argued that the present log - normal distribution of language sizes is , to a large extent , a consequence of demographic dynamics within the population of speakers of each language .
a two - parameter stochastic multiplicative process is proposed as a model for the population dynamics of individual languages , and applied over a period spanning the last ten centuries .
the model disregards language birth and death .
a straightforward fitting of the two parameters , which statistically characterize the population growth rate , predicts a distribution of language sizes in excellent agreement with empirical data .
numerical simulations , and the study of the size distribution within language families , validate the assumptions at the basis of the model . |
the hot intergalactic medium in low redshift clusters of galaxies has been enriched in iron to @xmath3 solar ( @xcite , @xcite ) by processes that , if understood , are a powerful diagnostic of galaxy formation . since the mass in the x - ray emitting gas exceeds that in the stars in galaxies by factors of 2 - 10 ( assuming @xmath4 km s@xmath5 mpc@xmath5 as we do throughout this paper ; we also assume @xmath6 ) , the igm is the main reservoir of metals in clusters .
moreover , the strong correlation of iron mass with optical light ( @xcite ) implies a coupling of the constraints on the evolution of metals from the x - ray data on high redshift clusters and the optical data on the evolution of elliptical galaxies . if , as many theorists believe , clusters are fair samples of the universe , as well as the only systems likely to have retained all the metals produced by stars , then studies of the evolution of the igm metallicity strongly constrain the star formation history in the universe .
there were few measurements of fe in clusters at @xmath7 before the launch of the _ asca _ satellite ( @xcite ) .
neither early results from _ asca _ ( @xcite , @xcite ) , nor subsequent detailed individual measurements ( @xcite , @xcite , @xcite ) , showed evolution in the fe abundance out to @xmath1 . here
we report on a large sample of high signal - to - noise _ asca _ spectra of clusters at @xmath8 , and find no evidence for evolution of the cluster metal abundance .
since recent optical data reveals only passive evolution in cluster ellipticals out to @xmath9 , most of the cluster metals were produced at @xmath2 .
we thus infer much higher star formation rates in the @xmath2 universe than derived from optical / uv surveys .
we extracted and analyzed @xmath10 spectra of clusters at @xmath8 from data in the _ asca _ public archives before 12/96 .
we concentrate on the 21 high signal - to - noise observations with 90% confidence error in fe abundance @xmath11 a moderately large sample upon which simple and unambiguous statistical tests can be made . at @xmath12 kev ,
the product of the flux , @xmath13 , and exposure time must exceed @xmath14 erg @xmath15 to attain this accuracy .
for a typical _ asca _ exposure time of 40 ks ( net exposures for two clusters in our sample , ms1358 and ac118 , were @xmath16 ks ) , a minimum flux @xmath17 erg @xmath15 s@xmath5 is thus required . before the release of the _ rosat _ all - sky survey , such objects were rare : e.g. , in the ipc database only 19 clusters have @xmath8 and @xmath18 .
cross - correlation of the _ rass _ and abell catalogs yields 50 clusters with @xmath19 at @xmath8 ( @xcite ) , but further cross - correlation with the _ asca _ observation log identifies only 19 clusters with @xmath18 .
thus we include most of the high signal - to - noise , high-@xmath20 clusters .
( the flux from one cluster that apparently met our criterion , abell 1774 , turned out to be dominated by an agn . )
this subsample ( table 1 ) spans the bolometric x - ray luminosity ( about twice the 2 - 10 kev luminosity ) range from @xmath21 to @xmath22 erg s@xmath5 , overlapping the low-@xmath20 sample measured by previous satellites .
we applied standard processing to the data ( e.g. , @xcite ) ; and , for both the sis and gis , used a @xmath23 extraction radius , corresponding to 3 mpc at @xmath1 sufficiently extended to include the total flux . at @xmath24 ,
10 - 30% of the flux may fall outside the extraction radius .
we used all - sky backgrounds accumulated during the _ asca _ deep survey pointings for clusters with @xmath25 , and background from the same field for the higher @xmath20 systems .
spectra were fit to an isothermal , collisional equilibrium plasma model at the optical redshift of the source .
the redshifts obtained from _ asca _
data are within 0.013 of the optical values ( e.g. , for the relatively faint ms1358 the fitted redshift is @xmath26 0.323(+.011,-0.022 ) compared to @xmath27 0.327 , while for the brighter zw3146 @xmath26 0.292 ( + 0.004,-0.0061 ) and @xmath27 0.291 ) .
cooling - flow - plus - isothermal - plasma models were used in cases ( zw3146 , a2390 ) where a cooling flow is clearly present in the _ rosat _ image ( cf .
as found by @xcite , the extra component has little effect on the derived abundances of the isothermal plasma .
our fits are in excellent agreement with results for those clusters in table 1 that were previously analyzed : rxj 1347 ( @xcite ) , zw3146 , a1835 and e1455 ( @xcite ) , a1204 ( @xcite ) , and ms0451 ( @xcite ) .
( the results are new for all others . ) the parameters derived from _ asca _ spectra ( figure 1 ) are robust for these high flux objects .
cccccc a1204 & 45.08 & 0.587 & 0.350 & 0.07 & 0.170 a1722 & 45.23 & 0.768 & 0.250 & 0.11 & 0.270 a1246 & 45.18 & 0.798 & 0.223 & 0.08 & 0.187 ms2137 - 24 & 45.25 & 0.643 & 0.410 & 0.12 & 0.313 a963 & 45.30 & 0.829 & 0.291 & 0.08 & 0.206 ms1358 + 62 & 45.30 & 0.813 & 0.272 & 0.10 & 0.327 a1413 & 45.36 & 0.797 & 0.289 & 0.05 & 0.143 a1763 & 45.38 & 0.953 & 0.263 & 0.09 & 0.187 a2218 & 45.41 & 0.857 & 0.180 & 0.065 & 0.171 a773 & 45.45 & 0.952 & 0.240 & 0.08 & 0.197 ms1455 & 45.47 & 0.660 & 0.330 & 0.08 & 0.258 a520 & 45.48 & 0.934 & 0.250 & 0.20 & 0.201 ms0451 - 03 & 45.60 & 1.01 & 0.157 & 0.12 & 0.539 ac118 & 45.65 & 0.970 & 0.228 & 0.09 & 0.308 a2390 & 45.73 & 0.949 & 0.220 & 0.06 & 0.230 zw3146 & 45.74 & 0.802 & 0.240 & 0.05 & 0.290 a1689 & 45.77 & 0.947 & 0.260 & 0.06 & 0.180 a2204 & 45.78 & 0.903 & 0.450 & 0.05 & 0.153 a2219 & 45.92 & 1.07 & 0.250 & 0.07 & 0.228 a1835 & 45.92 & 0.910 & 0.320 & 0.05 & 0.252 rxj1347 - 11 & 46.28 & 0.968 & 0.330 & 0.10 & 0.451
our sample of 21 clusters has nearly identical fe abundance properties to the low-@xmath20 _ ginga _ sample ( @xcite , @xcite ) , _ i.e. _ a mean abundance @xmath28fe@xmath29 and variance 0.076 ( figure 2 ) compared to @xmath28fe@xmath30 and variance 0.062 ( all errors are 90% confidence for one parameter of interest ) .
a k - s test shows no significant difference between the samples .
we find no evidence of systematic abundance variations with luminosity .
the weak abundance - temperature correlation ( figure 3 ) is not significant at 90% confidence .
we conclude that this sample , with @xmath28z@xmath31 , shows no statistical differences in fe abundance from a similar sample of low-@xmath20 objects .
results for the systems with larger abundance errors agree with the above conclusions .
although our sample is not flux- or redshift - limited , it has no obvious selection biases and should be representative of @xmath32 clusters .
its mean luminosity , @xmath28log @xmath33 , and mean temperature , @xmath34 kev , are slightly higher than in the low-@xmath20 sample drawn from a flux limited survey ( @xmath28log @xmath35 , @xmath36 kev ) .
the temperature - luminosity relationships of the @xmath32 and low-@xmath20 samples show no significant differences ( @xcite ) .
in high-@xmath37 hierarchical clustering scenarios ( @xcite ) , the mass of rich clusters grows by @xmath38 from @xmath39 to the present .
moreover , systems at @xmath9 evolve differently from a cluster of the same richness observed today , with high-@xmath20 clusters assembled over a shorter time interval and undergoing more merging in the few gyr prior to the epoch when they are observed .
thus the present and low-@xmath20 samples should differ if the accreted material has distinct properties , and the surprising constancy of both the mean and variance of the fe distribution over this redshift range ( corresponding to a lookback time of @xmath401/4 - 1/3 the age of the universe ) argues for a very homogeneous accretion process or a low-@xmath37 universe where cluster growth is essentially completed at @xmath41 ( @xcite ) .
the substantial spread in cluster metallicity at @xmath42 and @xmath43 is unexpected within purely gravity - driven hierarchical clustering scenarios .
many of these clusters include an excess of blue galaxies compared to low-@xmath20 systems ( the `` butcher - oemler effect '' ) .
none of these have anomalous metallicity : whatever processes converted high-@xmath20 blue spiral galaxies into low-@xmath20 spheroidal systems had no major impact on the igm .
the absence of evolution over the past @xmath44 gyr implies early enrichment of the igm .
thus , ram pressure stripping of infalling galaxies at low-@xmath20 can not dominate the enrichment , but much of the cluster fe could be produced by snia ( @xcite ) : reconciling large fe - mass - to - light ratios with modest present - day snia rates requires the rate to have a short rise time followed by steep negative evolution such that 90% of the fe is produced at @xmath45 .
most of the cluster metals , perhaps including most of the fe , were produced at high @xmath20 by snii ( @xcite ) .
the stars in elliptical galaxies that we see today come from the same population that produced the metals in the igm ( e.g. , @xcite ) ; moreover , the elliptical galaxies in several rich @xmath1 - 0.4 clusters of galaxies show only passive evolution ( e.g. , @xcite ) .
therefore , the star formation in elliptical galaxies that produced the cluster metals must have stopped @xmath46 gyr before @xmath9 , placing the period of cluster metal formation at @xmath2 in excellent agreement with the lack of evolution inferred from _ asca _ data .
this is in contrast to the history of metal production by disk galaxies , where star formation peaked at @xmath47 and continues through the present epoch ( @xcite , @xcite ) .
the containment by the deep cluster potential of the metals synthesized and ejected by supernovae during the epoch when the stars that make up present - day elliptical galaxies formed , enables us to calculate the @xmath2 metal production by ( proto-)ellipticals and estimate their contribution to the total metals produced at @xmath2 and over the hubble time . assuming that relative abundances in clusters are constant
, the mass in metals produced per unit ( present - day ) blue - band elliptical galaxy luminosity is @xmath48 in solar units . if the star formation ( though not necessarily the _ dynamical _ evolution ) in elliptical galaxy precursors proceeded in a similar way in the field and in clusters , then the product of the metal mass - to - light ratio and the early - type galaxy luminosity density ( @xcite ) yields the total mass density of metals from early - type galaxies , @xmath49 this is consistent with the value , derived by @xcite , required to account for the increase in stellar mass - to - light ratio with luminosity as due to stellar remnants from primordial star formation with an imf biased towards high mass stars .
the metals ejected by proto - ellipticals , as calculated above , exceed what is estimated to be locked inside the stars in normal massive galaxies today by 2 - 5 ( @xcite ) .
since the latter is consistent with the star formation rate derived from galactic uv light , star formation in proto - ellipticals must have proceeded in an obscuring environment or in sub - luminous fragments ( @xcite ) .
since half the star formation in disks has occurred since @xmath50 but , star formation in ellipticals was completed by @xmath50 ( cf .
@xcite ) , proto - spheroids increasingly dominate star formation at early epochs .
if elliptical galaxy star formation predominantly occurred at @xmath51 , the average metal production rate was @xmath52 compared to averages from high-@xmath20 , uv surveys of galaxies of 0.89 , 0.46 , and 0.17 @xmath53 yr@xmath5 mpc@xmath54 for @xmath2 , @xmath55 , and @xmath56 , respectively .
thus , global star formation rates derived from uv surveys are severe underestimates . at @xmath57 , where @xmath58 redshifts out of the ground - based optical band , [ oii]3727 or the uv continuum
are used as star formation indicators .
recent astro-2 uit and hut observations ( @xcite ) show that there is often a shift of a factor 2 in specific star formation rates using the uv continuum instead of @xmath58 ; and furthermore , comparison of iras @xmath59 luminosities with @xmath58 values indicates another factor of 2 offset ( see also , @xcite , @xcite ) .
since even small amounts of reddening are sufficient to remove objects from the high-@xmath20 galaxy samples and to change their inferred star formation rates , we believe that the cluster data are much more robust indicators of the true metal formation history of the universe .
the fe abundance of the intergalactic medium in rich clusters of galaxies undergoes little if any evolution from @xmath1 to @xmath60 .
this strongly constrains high-@xmath37 hierarchical theories in which clusters rapidly grow during this period .
a real and constant spread in the metallicity at all observed redshifts implies stochasticity , not only in the construction of these massive systems , but also in the efficiency and/or timing of star formation within their constituent galaxies .
a self - consistent history of star formation in disk galaxies has emerged from a synthesis of observations of damped @xmath61 systems , deep ground based and _ hubble space telescope _ surveys of star forming galaxies , and spectral synthesis and chemical evolution modeling . however , this census of star formation in the universe is incomplete , since most of the metals produced at high @xmath20 were expelled from galaxies .
a complementary picture of star formation in proto - spheroids is emerging , driven by the lack of evolution in igm metallicity reported here and optical evidence that together indicate that cluster metals originate from the precursors of elliptical galaxies at @xmath2 .
the @xmath62-dependence of @xmath63 ( @xcite ) and gross and relative igm elemental abundances ( @xcite ) are explained if this star formation produces an imf skewed toward high masses , precipitating a massive outflow of metal - enriched gas .
this star formation must occur in sub - luminous fragments or be dust - enshrouded to avoid detection by uv surveys , but the ir and/or uv background produced is within reach of near - future measurements ( @xcite ) .
the igm of rich clusters are reservoirs for primordial metal - rich galactic outflows driven by this star formation , and their investigation with _
asca _ has enabled us to discover that @xmath64 times more metals were created at @xmath2 than are inferred from high redshift galaxy data .
anders , e. , & grevesse , n. 1989 , geochimica et cosmochimica acta , 53 , 197 allen , s. w. , fabian , a. c. , edge , a. c. , bautz , m. w. , furuzawa , a. , & tawara , y. 1996 , mnras , 283 , 263 arnaud , m. , rothenflug , r. , boulade , o. , vigroux , l. , & vangioni - flan , e. 1992 , a & a , 254 , 49 bautz , m. w. , mushotzky , r. , fabian , a. c. , yamashita , k. , gendreau , k. c. , arnaud , k. a. , crew , g. b. , & tawara , y. 1994 , pasj , 46 , l131 bender , r. , ziegler , b. , & bruzual , g. 1996 , apj , 463 , l51 buat , v. 1992 , a&a , 264 , 444 butcher , j. 1995 , ph.d .
thesis , leicester university cowie , l. , songaila , a. , hu , e. m. , & cohen , j. g. 1996 , aj , 112 , 839 deharveng , j .- m . , sasseen , t. p. , buat , v. , bowyer , s. , lampton , m , & wu , x. 1994 , a&a , 289 , 715 donahue , m. 1996 , apj , 468 , 79 ebeling , h. , voges , w. , bohringer , h. , edge , a. c. , huchra , j. p. , briel , u. g. 1996 mnras , 281 , 799 edge , a. , & stewart , g. 1991 , mnras , 252 , 428 elbaz , d. , arnaud , m. , & vangioni - flam , e. 1995 , a&a , 303 , 345 fanelli , m. et al .
1996 , preprint ( astro - ph/9612086 ) kauffmann , g. 1995 , mnras , 274 , 153 loewenstein , m. , & mushotzky , r. f. 1996 , apj .
466 , 695 loveday , j. , peterson , b. a. , efstathiou , g. , & maddox , s. j. 1992 , apj , 390 , 338 madau , p. , ferguson , h. c. , dickinson , m. , giavalisco , m. , steidel , c. c. , & fruchter , a. 1996 , preprint ( astro - ph/9607172 ) matsuura , m. , miyoshi , s. j. , yamashita , k. , tawara , y. , furuzawa , a. , lasenby , a. n. , sanders , r. , jones , m. , & hatsukade , i. 1996 , apj , 466 , l75 mushotzky , r. f. 1996 , in roentgenstrahlung from the universe , ed .
h. u. zimmerman , j. trumper , and h. yorke ( mpe report 263 ) , 545 mushotzky , r. f. , loewenstein , m. , arnaud , k. a. , tamura , t. , fukazawa , y. , matsushita , k. , & kikuchi , k. 1996 , apj , 466 , 686 mushotzky , r. f. , & scharf , c. a. 1997 , apj , submitted pei , y. c. , & fall , s. m. 1995 , apj , 454 , 69 renzini , a. , ciotti , l. , dercole , & pelligrini , s. 1993 , apj , 419 , 52 schindler , s. , hattori , m. , neumann , d. m. , bohringer , h. 1996 , a&a , submitted tanaka , y. , inoue , h. , & holt , s. s. 1994 , pasj , 46 , l37 tsuru , t. , koyama , k. , hughes , j. p. , arimoto , n. , kii , t. , & hattori , m. 1996 , in uv and x - ray spectroscopy of astrophysical and laboratory plasmas , ed . k. yamashita and t. watanabe ( tokyo : universal academy press ) , 375 white , s. d. 1996 , preprint ( astro - ph/9608044 ) yamashita , k. 1992 , in frontiers of x - ray astronomy , ed . y. tanaka and k. koyama ( tokyo : universal academy press ) , 475 zepf , s. e. , & silk , j. 1996 , apj . | we present the first large sample of accurate iron abundances and temperatures for clusters at redshifts @xmath0 .
we find that the fe abundance shows little or no evolution out to @xmath1 .
this and the early formation epoch of elliptical galaxies in clusters indicate that most of the enrichment of the intracluster medium occurred at @xmath2 .
if clusters represent fair samples of the universe , then global metal production is 2 - 5 times greater than is inferred from recent studies of galaxies at low and high redshifts . |
a technological break - through in the fabrication of nanodimensional clusters and other metallic nanoparticles gave rise to a development of such nano - technological and nanooptical branch as nanoplasmonics , which is of great interest to physicists , chemists , material engineers , it specialists , and biologists .
the nanoplasmonics deals with conduction electron liquid oscillations in metallic nanostructures and nanoparticles , and an interaction of those oscillations with light ( plasmon - polariton ) ref1,ref2 .
such studies are also aimed at the oscillations of a crystal lattice ( sic , for example ) in the nanoparticles , whose interaction with light has much in common with the plasmon oscillations ( phonon - polariton ) @xcite .
the important feature of nanoplasmonic phenomena is the combination of a strong spatial localization and high - frequency ( from ultraviolet to infrared ) of electron oscillations . in its turn
, strong localization leads to a giant enhancement of the local optical and electrical fields .
these important features of plasmonic particles made it possible to discover quite a number of new effects .
one of the most developed is the use of large local fields near plasmonic nanoparticles for enhancement of the raman scattering cross - section .
recent experiments have shown that such an increase may achieve 10 - 14 orders of magnitude , which may help to resolve single molecules @xcite-@xcite .
the local enhancement of the fields can also be used to increase the fluorescence intensity and to determine the structure of a single dna strand without using the fluorescent labels ref8-@xcite . by using the nanoparticles of more complex
shape one can provide enhancement of both the absorption and the emission of light by natural and artificial fluorophores @xcite . on the other hand ,
the plasmon nanoparticles are proposed to be used in nanolasers @xcite and to stimulate plasmonic oscillations in nanoparticles by means of the optical emission ( spaser ) @xcite . beside these new applications of the plasmonic nanoparticles
, one can essentially increase the efficiency - cost ratio , for example , in solar batteries or light emitting diodes ref13,ref14 by using the achievements in nanoplasmonics . and finally , it is awaited that the nanoplasmonics will make it possible to create a new element base ( genuine integrated optics and ultracompact optical components ) for the computers and data processing equipment by taking an advantage of small dimensions of metallic nanoparticles and fast speed of optical processes @xcite . an intricate spatial structure of the physical phenomena , which form the basis of nanoplasmonics , impedes a development of the latter . very often , the numerical studies do not allow one to explain the physics of the observed phenomena , while the analytical studies are mostly devoted to the case of spherical and spheroidal nanoparticles , which are very far from the synthesized nanostructures from the viewpoint of geometry and physics .
separate nanoparticles are often approximated by point dipoles .
this permits one to explain several phenomena ( for example , a propagation of plasmons in a chain of nanoparticles @xcite ) .
but as a whole , the physics of plasmonic oscillations in complex nanostructures and nanoparticles remains insufficiently studied , and to make progress in this field one needs new approaches and new ideas . as one of such approaches we suggest that the plasmonic effects observed in clusters of nanoparticles should be described by the paradigm of plasmonic atoms and molecules , which has much in common with the paradigm of normal atoms and molecules
this would allow one to explain the complex phenomena of nanoplasmonics in terms of atomic and molecular physics , and to propose quite a number of new applications of the nanoplasmonics . within the framework of our approach we propose to treat the plasmonic oscillations of the connected nanoparticles as the plasmonic atoms .
the plasmonic atoms are analogous , in a certain way , to the electron oscillations at the nanocrystal quantum dots , which are also often called the artificial atoms @xcite .
the bound states of plasmonic atoms in the clusters of nanoparticles are proposed to be called the plasmonic molecules , and finally , the plasmonic oscillations in the nanoparticle gratings are proposed to be called the plasmonic crystals .
such an approach seems to be efficient and useful as it follows from the strongly localized plasmonic oscillations recently discovered in the clusters of spherical nanoparticles @xcite .
these oscillations , being localized and of short - range , represent the plasmonic molecules , and on their basis it is possible to create different kinds of plasmonic nanostructures and to develop new nanodevices ( optical nanosensors of single molecules , nanoelectromechanical systems ( nems @xcite ) , and even plasmonic computers ) .
let us consider in more detail the analogy between the plasmonic and ordinary atoms . in the simplest case of a hydrogen atom ,
the energy levels are defined by the bohr equation @xmath0 where @xmath1=13.55 ev is the ionization energy of hydrogen atom .
the wave function of an electron has the form , in this case : @xmath2 where @xmath3 is the radial part of the wave function , and @xmath4 , the spherical harmonic .
for example , in the case of the first excited ( 2p ) atomic state , the expression for the radial function takes the form : @xmath5 where @xmath6 is the normalized constant , and @xmath7 , the radius of the first bohr orbit .
function ( [ eq3 ] ) is represented in fig .
[ fig:1 ] . in the case of a spherically symmetric plasmonic atom ,
i.e. the plasmonic oscillations in a metallic spherical nanoparticle of the radius @xmath8 , under the drude dispersion law @xmath9 ( where @xmath10 is bulk plasmon frequency ) , the spectrum of the plasmonic oscillations will take the form : @xmath11 whereas the electric field potential , in respect to ( [ eq4 ] ) , will have the form corresponding to ( [ eq2 ] ) with the radial wave function @xmath12 where @xmath13 and @xmath14 is some constant .
function ( eq5 ) for the index @xmath15 is illustrated in fig .
[ fig:1 ] by a dotted line .
it is seen from this figure that wave functions of the ordinary and plasmonic atoms are similar .
the spectra ( [ eq1 ] ) and ( [ eq4 ] ) also have much in common .
the plasmonic oscillations have similar properties in the other spherically symmetrical layered systems ( nanomatreshka ) @xcite and spheroidal nanoparticles ( nanorice ) @xcite .
the resonant plasmonic frequencies and potentials of plasmonic atoms in more complex and less symmetrical nanoparticles have a more complex form .
for example , in the case of a metallic nanoparticle having a form of a three - axial ellipsoid with the semi - axes @xmath16 , the plasmonic frequencies will be defined by the ratio @xcite @xmath17 where @xmath18 and @xmath19 are the internal and external lame functions @xcite , and a stroke denotes the derivative of function by its argument .
the dependences of the resonant dielectric permeabilities for @xmath20 are shown in fig .
[ fig:2 ] .
the explicit expressions for eigen frequencies and eigen functions , e.g. , for the case of @xmath20 and @xmath21 , have the form ( guzatov , klimov to be published ) @xmath22 where @xmath23 is a constant ; and @xmath24 .
the distribution of a surface charge for the plasmonic oscillations with @xmath25 and @xmath21 derived from eq .
( [ eq7 ] ) has the form depicted in fig .
[ fig:3 ] ( guzatov , klimov to be published ) .
even more complex properties characterize the cubic and polyhedral plasmonic atoms @xcite-@xcite .
the spectrum of the cubic plasmonic atoms was studied in @xcite , and some of the first values take the form @xmath26 figure [ fig:4 ] illustrates schematically the surface charge distribution for several states of a cubic plasmonic atom as found in ref27 .
the experiments studying the cubic plasmonic atoms in @xcite proved to be in an agreement with the theoretical calculations .
it is very important that both the plasmonic atom wavefunction ( electric potential ) and the field intensity are changing within the whole volume of a nanoparticle and the adjacent space , despite the fact that the structure of plasmonic atoms may be rather complex . by analogy with that ,
the surface charge in any plasmonic atom changes noticeably over the whole nanoparticle surface .
such a smeared spatial structure of plasmonic atoms is their characteristic feature , and can be used in quite a number of applications such as nanosensors of single molecules .
however , as the distance between nanoparticles decreases they start to interact more intensively . and instead of the plasmonic atoms it is expedient to use the conception of plasmonic molecules to describe the plasmonic oscillations in the closely situated and strongly interacting nanoparticles .
as in the case of ordinary atoms and molecules , the question about the existence of bound states of the plasmonic atoms emerges here first of all .
the evidence for the existence of bound states of the plasmonic atoms was first provided in @xcite by the example of a cluster of two closely sitiated spherical nanoparticles ( fig .
[ fig:5 ] ) . generally speaking
, the plasmonic oscillations in the cluster of two nanoparticles were studied earlier too @xcite-@xcite .
however , it was assumed , both explicitly and implicitly , that the bound states are not formed in such a cluster , and the interaction assumes a scattering of plasmonic atoms on each other with a respective splitting of energy levels , and hybridization of wave functions ( potentials ) of single plasmonic atoms @xcite .
that was the reason why the plasmonic molecules had not been predicted earlier .
the discovered by us spectrum of the plasmonic oscillations in a cluster of two spherical nanoparticles is shown in fig .
[ fig:6 ] for the case of identical spheres ( @xmath27 . there are only symmetrical ( t - modes ) and antisymmetrical ( l - modes ) states of the scattering within the range of @xmath28 ( @xmath29 ) .
they exist at any distance between the particles , and continuously transit into the respective states of the weakly interacting plasmonic atoms with the characteristics ( [ eq4 ] ) and ( [ eq5 ] ) , within large distances between the nanospheres .
just those scattering states had become the subject of investigations in @xcite-@xcite . in the region of @xmath30 ( @xmath31 )
the plasmonic oscillations may exist only at small distances between the nanoparticles , and correspond to the bound states of the plasmonic atoms and molecules ( m - modes ) .
note that the existence of the plasmonic molecules is by no means connected with a quasistatical approximation , and the calculations made within the framework of the total system of maxwell s equations , with account for retardation effects , provide evidence for the existence of plasmonic molecules in the case of a small gap as compared with the radii of spheres ( guzatov , klimov unpublished ) .
the analogous spectra appear for higher azimuthal numbers @xmath32 , although there are some peculiarities in the spectra of the unbound plasmonic atoms . in the region of small distances between the nanospheres , the properties of the plasmonic molecules may be studied analytically within quasistatic approximation .
for the drude dispersion law , the spectrum of the plasmonic molecules takes the form : @xmath33 where @xmath34 is the distance between the centers of spheres and @xmath35 the potential of the axis - symmetric ( @xmath32=0 ) state of the plasmonic molecule in a space between the spheres ( @xmath36 has the form : @xmath37 where @xmath38 is the associated legendre function and @xmath39 , the bispherical coordinates @xmath40 and where @xmath41 . for the antisymmetric unbound plasmonic atom we have
respectively @xmath42 and @xmath43 in the case of large azimuthal numbers , @xmath44 , one can also find simple asymptotic equations for the spectra of plasmonic oscillations in a cluster of two spheres .
this case is of importance for calculation of van der waals forces between nanoparticles .
consider in more detail a structure of the wave functions ( electric potentials ) of the plasmonic molecules .
figure [ fig:7 ] illustrates a wave function spatial distribution of a plasmonic molecule and unbound states of plasmonic atoms in the @xmath45-@xmath46 plane .
spatial structure of antisymmetrical ( l - modes ) and symmetrical ( t - modes ) wave functions of the unbound plasmonic atoms , in the axis - symmetrical case ( @xmath47 ) corresponds to the structure of wave functions of the isolated atoms .
namely , a positive charge lies in one part of a sphere , while a negative charge , which is equal to the positive one due to the electrical neutrality of spheres , lies in the opposite part of the latter .
an interaction between the plasmonic atoms is reduced , in this case , to a charge re - distribution on opposite semi - spheres . in the case of plasmonic molecules ( symmetrical m - modes ) ,
the situation is the opposite , and both positive and negative charges are concentrated in a small region near a gap between the nanospheres .
far from the gap , the wave functions of plasmonic molecules practically vanish . in the case of @xmath48 , where the wave functions have an angular dependence @xmath49 or @xmath50
, the situation is analogous .
the only difference is that the dipole moments of the unbound plasmonic atoms and molecules are directed along the @xmath45- or @xmath51-axis . as a whole ,
the wave function of the plasmonic atoms remains more or less distributed over the volume of both nanospheres , whereas the plasmonic molecules are strongly localized near the gap between the nanospheres in a region with a characteristic dimension of the order of the gap size with the maxima at spherical surfaces .
a strong localization of the plasmonic modes is also well seen in fig .
fig:8 , which shows a distribution of the ( surface ) charge over a surface of spheres . to plot these distributions
we have used a polar system of coordinates ( @xmath52 , @xmath53 ) for each sphere in plane @xmath54 .
the surface charge distribution was depicted with @xmath55 polar curve where @xmath56 is the dimensionless surface charge density as a function of polar angle of corresponding sphere .
as the distance between the spheres increases , the localization of the plasmonic molecules decreases . at critical distance between the spheres the plasmonic molecules disappear , whereas the unbound atoms do not change .
the localization of plasmonic molecules and unbound plasmonic atoms is different , which defines their different dependence on the exciting radiation .
the polarizability of the unbound plasmonic atoms is of the order of a nanosphere volume @xmath57 , and they effectively interact with the uniform external fields of the respective orientation and symmetry . and on the contrary , the plasmonic molecules have a comparatively small polarizability @xmath58 , where @xmath59 is the value of the gap between the spheres . as a result ,
the plasmonic molecules are weakly excited by the uniform optical fields ( as compared to the unbound plasmonic states ) .
on the other hand , the plasmonic molecules interact effectively with strongly inhomogeneous fields that are localized near the gap between the spheres .
such fields are produced by the atomic and molecular radiation near the gap .
this makes the plasmonic molecules extremely perspective from the viewpoint of a development of nanosensors and elements of the nanodevices , which are sensitive to single molecules . quite probably , that the sers from single molecules @xcite-@xcite is due to their interaction with the plasmonic molecules .
it seems that hot points in fractal structures ( shalaev s hot spots @xcite ) are also due to the plasmonic molecules .
above we considered the plasmonic molecules and unbound states of the plasmonic atoms formed in a cluster of two identical nanospheres . however , the plasmonic molecules may be formed within a much wider set of clusters of the nanoparticles .
the only necessary condition for the formation of plasmonic molecules is smallness of the gap between the nanoparticles of the finite volume as compared to their characteristic dimensions .
the plasmonic molecules may also be formed in connected particles with two or more nanobubbles or dielectric nanospheres inside them ( fig .
[ fig:9 ] ) .
first of all , the plasmonic molecules can exist in a cluster of two different nanospheres . here
the spectra and wave functions of unbound plasmonic atoms and plasmonic molecules are completely analogous to the case of identical nanospheres .
if the distance between the spheres is small , then the plasmonic molecule spectrum may be written in the form : @xmath60 , \notag \\ n & = & 1,\;2,\;3,\;\ldots \quad m\gg 1 , \label{eq15}\end{aligned}\ ] ] where the parameter @xmath61 can be found from the expression @xmath62 and where @xmath63 are the dielectric constants of the first and second spheres and the space between them , correspondingly .
for the dispersion law of drude for both spheres it can be found from ( eq15 ) @xmath64 \right\ } ^{1/2 } , \label{eq17}\end{aligned}\ ] ] where @xmath65 and @xmath66 are the bulk plasmon frequencies of spheres .
strong localization of plasmonic molecules allows one to suggest that their characteristics depend mainly on the radius of curvature of almost contacting surfaces and the distance between the centers of the inscribed ( or circumscribed ) spheres .
so , for any smooth nonspherical particle the properties of plasmonic molecules formed in the gap may be estimated by considering the plasmonic molecules in two spheres , which approximate the nonspherical particles in the contact point ( see fig . [ fig:10 ] ) .
for the closely set semi - infinite bodies the situation turns to be more intricate , since in this case part of the charges may go arbitrary far from the contact area .
nevertheless , here exist a localized plasmonic oscillations , which are analogous to the antisymmetric unbound plasmonic atoms ( l - modes ) and plasmonic molecules ( m - modes ) . in case of closely set semi - infinite hyperboloids ( fig .
[ fig:11]a ) the variables in laplace equation can be separated within prolate spheroid coordinates @xcite , and the localized plasmonic oscillation spectra have the form @xmath67 where @xmath68 and the indices @xmath69 , @xmath70 , which characterize the modes , continuously vary from 0 to the infinity ; @xmath71 ( @xmath72 is half of the smallest distance between the hyperboloids and @xmath73 is half of the distance between the hyperboloids focuses ) .
the corresponding wave functions ( electric potentials ) in the gap ( @xmath74 ) are @xmath75 where the prolate spheroid coordinates @xmath76 are related with the cartesian coordinates by the equations @xmath77 inside the hyperboloids the expressions for wave function of plasmonic atoms and plasmonic molecules are analogous to ( [ eq19 ] ) .
the axis - symmetrical ( @xmath47 ) spectra for several values of @xmath69 , @xmath70 = 1 , 3 , 5 , 10 are shown in fig .
[ fig:11]b as functions of distance between hyperbolic tips . the spatial distribution of corresponding wavefunctions ( electric potentials ) for @xmath69 , @xmath70 = 1 are given in fig .
[ fig:12 ] . as seen from the figures , for both the symmetrical and antisymmetrical plasmonic oscillations there
is observed a strong localization of the modes , and this ensures an effective interaction with the ordinary atoms , molecules or quantum dots with plasmonic atoms and molecules near the gap between tips .
one more interesting fact is that the spectra of these oscillations are quite close to the oscillations in a two - sphere cluster .
this allows one to consider the localized plasmonic oscillations in the gap between semi - infinite bodies as the virtual plasmonic atoms and molecules .
obviously , strong localization of plasmonic molecules may be suitable for a number of applications .
this is particularly important when one needs to ensure effective interaction between the nanolocalized light sources ( molecules and nanocrystal quantum dots ) and the nanoparticles and nanostructures , as well as the nano - electromechanical devices @xcite , in which the van der waals forces play an important role @xcite .
as is well known , the van der waals forces are associated with spatial dependence of the vacuum fluctuation energy density .
if the distance between the plasmonic nanoparticles is small , then the main contribution to the van der waals energy comes from zero ( vacuum ) oscillations of the plasmonic modes , i.e. the plasmonic atoms and molecules @xcite . in case of two identical plasmonic nanospheres the contribution to the van der
waals energy comes from zero oscillations of the antisymmetric and symmetric unbound plasmonic atoms ( l- and t - modes ) and plasmonic molecules ( m - modes ) @xmath78 in spite of the fact that the expressions for the van der waals energy of different modes are formally similar , the noted contributions result in totally different physical consequences .
figure [ fig:6 ] shows that the energy of antisymmetric unbound states of the plasmonic atoms grows with distance , and this results in the particle attraction . in a similar manner ,
the energy of the plasmonic molecules diminishes with distance , and thus results in the repulsion of the nanospheres .
symmetric states of unbound plasmonic atoms also lead to a very weak repulsion .
the attraction or repulsion conditioned by different plasmonic oscillations is of a simple physical nature .
first consider the axis - symmetrical oscillations ( @xmath47 ) . here
the longitudinal antisymmetric modes are conditioned by the oscillations of the parallel dipoles which are attracting in accordance with the law of electrostatics ( see fig .
[ fig:14 ] and fig .
[ fig:7 ] ) . in case of symmetrical oscillations
the dipoles have opposite directions , and this results in the dipole repulsion ( see fig .
[ fig:14 ] and fig .
[ fig:7 ] ) . for @xmath48 ,
i.e. when the wave functions are described by the law of @xmath79 or @xmath50 ( @xmath80 is the angle describing the rotation around symmetry axis ) , the directions of the dipoles corresponding to different plasmonic states are changed . for antisymmetric modes
the dipole moments in different spheres should be crosswise to the symmetry axis and should have opposite orientations ( fig . [ fig:14 ] and fig .
[ fig:7 ] ) .
it is easily seen that this results again in the effect of attraction of the unbound plasmonic atoms . for the plasmonic molecules the dipole momenta should also be crosswise to the symmetry axis , but
, due to the symmetry with respect to z coordinate , should have the same direction ( see fig .
this again leads to the repulsion of nanospheres with plasmonic molecules .
[ fig:15 ] illustrates the dependence of the contributions from different plasmonic states into the van der waals energy on the distance between the spheres .
this was obtained by direct summation of all the plasmonic modes in ( [ eq21 ] ) . as it was clear beforehand , the symmetric modes lead to the repulsion , while the longitudinal antisymmetric modes lead to the attraction of the nanoparticles .
unexpectedly , the repulsion contribution from plasmonic molecules is almost equal to the attraction contribution of the unbound plasmonic atoms . as
a result , the total van der waals energy grows with distance , i.e. it is the attraction energy , but the amount of this energy is by an order smaller than the energy to be obtained without taking into account the contribution from the plasmonic molecules .
direct measurement of the van der waals force between the plasmonic nanoparticles may form the basis of the experimental proof of the plasmonic molecule existence .
since the plasmonic molecules particularly effectively interact with highly inhomogeneous electric fields , then one can develop different detectors of the ordinary single molecules or quantum dots .
[ fig:16 ] illustrates the dependence of the radiation decay rate of spontaneous emission of different molecules , which fall within the gap between the two spheres , on the radiation wavelength .
it is assumed that the nanospheres are made of na with the plasmonic resonances in the visible range ( fig .
[ fig:16]a ) or sic with the phonon - polariton resonances in the ir range ( fig .
[ fig:16]b ) .
the peaks in the right - hand part of the figures correspond to the interaction of an ordinary molecule and the unbound plasmonic atoms , while the peaks in the left - hand part correspond to the interaction with the plasmonic molecules .
the analysis of the figures shows that , similar to the case of the van der waals energy , the interaction with plasmonic molecules is of greater importance than that with the symmetric unbound plasmonic atoms .
of fundamental importance is the fact that in a homogeneous external field the plasmonic molecules are not effectively excited .
this results in the following : the clusters of two and more nanoparticles ( where the geometry allows one to excite the plasmonic molecules of a certain frequency ) , can be used in the development of effective nanodetectors of single molecules of high signal / noise ratio , since the external fields at the plasmonic molecule frequency fail to excite the plasmonic molecule effectively . on the other hand ,
the effective interaction between the plasmonic and ordinary atoms and molecules can be applied both to the development of spaser and single atom nanolaser , and the sets of such devices .
probably , the most interesting consequence of the plasmonic molecule strong localization near the gap between the nanoparticles of arbitrary shapes lies in possible existence of plasmonic molecules in the clusters with more than two particles .
figure [ fig:13 ] illustrates a cluster of nine metallic nanoparticles , in which different configurations of plasmonic molecules may exist . in each of the twelve gaps
a plasmonic molecule having the properties defined by the curvature radii of the closest surfaces and given by expressions ( [ eq9])-([eq11 ] ) may be excited . in this system
one can also excite the pairs , triples , quadruples , etc .
, of plasmonic molecules . under simultaneous excitation of several plasmonic molecules a weak interaction between the molecules may be easily taken into account using the perturbation theory .
the account for such an interaction results in the hybridization of the wave functions of single plasmonic molecules and a corresponding splitting of the spectrum .
similar reasoning is also fully applicable to the large - size arrays of nanoparticles , and this allows one to speak about plasmonic molecular crystals . in other words ,
a strong spatial localization of plasmonic molecules and their weak interaction with each other allows an almost linear scaling of any simpler structure of plasmonic molecules .
to put it in simpler words , the plasmonic molecules allow one , similar to the lego bricks , to produce the optoelectronic nanodevices of any complexity . in principle , the excitation of plasmonic molecules in the clusters and arrays may be performed with the help of homogeneous optical fields of a corresponding resonant frequency .
however , such a method of excitation is not highly effective due to weak polarizability of plasmonic molecules .
a positive feature of such inefficiency is a weak sensitivity of plasmonic molecules to the external noise fields .
a more effective method to excite the plasmonic molecules lies in the excitation by means of spontaneous or stimulated radiation ( spaser @xcite)of usual molecules at a frequency of plasmonic molecule oscillation . in this case , depending on the orientation and location of the ordinary molecules ( or nanocrystal quantum dots ) with respect to the exciting field , the plasmonic molecules may be produced in the given part of a cluster or array , and , thus , to produce different initial conditions for the data processing and transfer .
in this work the analysis has been made of the plasmonic oscillations in single nanoparticles and the clusters of nanoparticles .
it has been shown that all the plasmonic oscillations may be classed by the degree of their localization .
such oscillations where the localization is mostly defined by the size of single nanoparticles we propose to name `` the plasmonic atoms '' , and those strongly localized in the nanoparticle contact zones ( without electrical contact ) we propose to name " the plasmonic molecules , since they are the bound states of plasmonic atoms .
it has been demonstrated that the plasmonic molecules almost do not interact between each other , and , so , can form the structures of arbitrary spatial complexity .
the methods of plasmonic molecule excitation and its applications in the description of nems and the nanodetectors of single molecules are briefly discussed as well .
the authors are grateful to the russian foundation for basic research ( grants 05 - 02 - 19647 and 07 - 02 - 01328 ) and the russian academy of sciences and centre de recherche scientifique ( france ) for financial support of this work .
one of the authors ( d.v.g . ) is thankful to the young scientists support program of the educational - scientific center of p.n .
lebedev physical institute and the ras presidium program `` support of young scientists '' for a partial financial support of the work .
* figure [ fig:1 ] * radial wave functions ( a.u . ) of usual and plasmonic atoms .
the parameter @xmath81 or @xmath82 .
the solid line corresponds to an ordinary atom and the dotted line , to the plasmonic atom .
* figure [ fig:3 ] * surface charge distribution ( a. u. ) for a plasmonic atom in triaxial metallic nanoellipsoid for @xmath20 and @xmath21
. the ratios of ellipsoid semi - axes @xmath84 = 0.7 and @xmath85 = 0.4 .
* figure [ fig:7 ] * spatial distribution of electric potential of the modes with l , m , t = 1 in @xmath45-@xmath46 plane .
a , case @xmath47 .
b , case @xmath48 .
radii of the spheres , @xmath8=50 nm ; distance between their centers , @xmath34=105 nm . surface of the nanospheres is denoted by a dotted line .
* figure [ fig:8 ] * surface charge ( a.u . ) of different modes with l , m , t= 1 , 2 , 3 at @xmath48 .
radii of the spheres @xmath8=50 nm ; distance between their centers , @xmath34=105 nm . surface of the nanospheres is denoted by the dotted line .
* figure [ fig:12 ] * spatial distribution of wave functions for localized plasmonic oscillations in the gap between two hyperboloids for @xmath47 and @xmath86 @xmath70 = 1 .
distance between hyperboloids , 2@xmath72=5 nm ; the least curvature radius is 50 nm ( cf .
[ fig:7]a ) .
solid line depicts surfaces of hyperboloids ; dotted line corresponds to the cluster of two nanospheres with @xmath8 = 50 nm and @xmath34 = 105 nm .
* figure [ fig:16 ] * relative radiative decay rate of a molecule located between two nanospheres versus the radiation wavelength .
a , nanospheres made of sodium ( na ) .
b , nanospheres made of silica - carbone ( sic ) . in both cases , @xmath8 = 50 nm and @xmath34 = 105 nm .
direction of the molecular dipole moment is shown by arrow . the dielectric constant for na
is borrowed from @xcite , for sic , from @xcite . * figure [ fig:13 ] * excitation of plasmonic molecules in clusters of nine nanospheres with usual molecules .
a , usual and plasmonic molecules are not excited .
b , external field with horizontal polarization excite the usual molecules .
c , radiation of usual molecules excite the plasmonic molecules with horizontal polarization . | the proposed paradigm of plasmonic atoms and plasmonic molecules allows one to describe and predict the strongly localized plasmonic oscillations in the clusters of nanoparticles and some other nanostructures in uniform way .
strongly localized plasmonic molecules near the contacting surfaces might become the fundamental elements ( by analogy with lego bricks ) for a construction of fully integrated opto - electronic nanodevices of any complexity and scale of integration . |
water is a key ingredient in many astrophysical environments , including circumstellar disks .
it is the main constituent of the icy mantles coating interstellar grains , and can be one of the most abundant molecules in the gas phase . for these reasons
, water may dominate both the chemistry ( being a major oxygen reservoir ) and the thermal balance ( i.e. cooling ) of the gas component of protostellar disks .
in addition , water vapor and water ice in protoplanetary disks are the major reservoirs of water later to be found in planets , asteroids and comets and , therefore , are of great relevance for understanding the origin of the solar system and the distribution of volatiles within it .
unfortunately , the observation of water molecules in interstellar space is nearly impossible from the ground because of the absorption by the terrestrial atmosphere .
so far , just a handful of space - based instruments ( iso , swas and odin ) have carried out observations of water vapor lines in interstellar molecular clouds , but these instruments did not have the sensitivity to observe water protoplanetary disks . however , the up - coming esa / nasa mission herschel will provide a sensitive instrument , hifi , for these studies
@xcite . also , sofia has potential for water observations .
meanwhile , hdo , as the most important isotope of water , is currently the best ground - based probe of the presence of water in astrophysical environments .
recently , @xcite reported the first discovery of hdo ( and therefore also the first discovery of water ) in a protoplanetary disk , namely dm tau .
the observations show an absorption line of gas phase hdo with a line - to - continuum ratio of 0.9 which translates into an average column density through the disk of @xmath7 @xmath5 . in the region where hdo is located ,
its fractional abundance is estimated to be about @xmath8 .
if the abundance of gaseous h@xmath3o is similar to the value observed in molecular clouds and protostars , @xmath9 , hdo / h@xmath3o ratio is 0.01 .
the fact that the line is a deep absorption line immediately indicates that hdo must be present above the midplane ( from where the sub - millimeter continuum is emitted ) .
it must also be present especially in the outer radial regions ( @xmath10 au ) of the disk since the continuum originates from an extended region .
the presence of a large column of hdo in this location was a surprise , since the grain temperatures in this region are below 25 k , and h@xmath3o and hdo should be almost entirely frozen out onto grains .
chemical models including _ thermal _ desorption of water ice and gas - phase formation of h@xmath3o predict a h@xmath3o column of 10@xmath11 @xmath5 at about 400au , only a factor of 5 larger than the observed hdo column .
@xcite included photo - desorption by a strong stellar uv field and found h@xmath3o column densities of about @xmath12@xmath5 at 700au .
an extended component of hdo ( and consequently water ) therefore indicates a desorption agent from grain surfaces .
important possibilities are x - rays originating from the star , and penetrating throughout the entire outer disk , and uv photons , either from the star or simply due to the interstellar radiation field .
high energy photons can in principle act to remove h@xmath3o molecules from the ice layers on dust grains .
x - rays are energetic enough to remove a number of molecules , if sufficient energy can be concentrated close to the surface of the ice mantle on a grain .
this idea has been explored by @xcite who considered x - ray heating of small grains , or spot heating of larger grains .
small grains , or small thermally insulated spots on large grain can be heated strongly enough by a single x - ray absorbed photon to lead to thermal evaporation of part of the ice mantle .
an alternative mechanism is desorption by fuv photons @xcite .
an fuv photon absorbed in the surface layer of an ice mantle can lead to the release of a molecule into the gas . in this letter
we explore the effect of just the interstellar fuv radiation field irradiating the disk surface of dm tau .
we show that such a scenario leads to a layer of water vapor in the disk surface to an fuv optical depth around unity . the column densities reached in steady state appear to be a viable explanation for the hdo absorption line seen in dm tau . to show this we developed a simple model of the chemistry leading to the presence of water vapor in irradiated gas containing grains covered by layers of water ice .
a much more detailed discussion in the context of a full photo - dissociation region ( pdr ) model , as well as analytical formulas to estimate h@xmath3o column densities under such conditions will be given by hollenbach et al ( 2005 , in preparation ) .
here we report the application of this model to the case of dm tau .
a forthcoming paper will present the results of a larger parameter space study , which will explore the dependence of the model predictions especially on the characteristics of the grains ( ceccarelli et al . in preparation ) .
the model computes the h@xmath3o abundance across a protoplanetary disk , as a function of the radius and height . for the physical structure we used the dust density and temperature profile which fits the spectral energy distribution ( sed ) of dm tau @xcite , and shown in fig .
[ fig : structure ] .
the gas is assumed to be fully coupled with the dust in terms of density distribution and temperature . due to the low temperature , the dust grains are covered with an ice layer . in our model ,
uv photons entering the disk surface are absorbed by dust grains , and cause photo - desorption of water molecules from the ice .
only uv photons absorbed directly at the surface of the ice mantle can desorb a molecule , so that only a small fraction of absorbed photons will cause the ejection of h@xmath3o .
laboratory experiments at lyman @xmath13 wavelength have shown that the involved yields are typically between 10@xmath14 and 10@xmath15 molecules per photon
. the yield increases to some extent depending on the uv dose in the experiment , apparently because radical formation on the ice surface aids the desorption process @xcite .
it is unclear how important this effect is in astrophysical environments , and we adopt a conservative value for the yield of 10@xmath14 molecules per incident fuv ( [email protected] ev ) photon .
the yield probably introduce a factor of a few uncertainty into the computations ( see below ) .
the desorption process at a given location will also depend on the attenuated fuv flux , and the rate per unit volume can be written as : @xmath17 where @xmath18 is the fuv field in @xcite units .
photons/@xmath20/s is the fuv flux for the standard habing interstellar field ( @xmath18=1 ) .
@xmath21 is the photo - desorption yield .
@xmath22 is h@xmath3 column density that gives @xmath23 between 6 and 13.6 ev ; we adopted a value equal to @xmath24 @xmath5 @xcite .
@xmath25 is the grain geometrical cross section , equal to @xmath26 , where we assumed an average grain size of 0.1 @xmath27 m .
@xmath28 is the grain number density .
the product @xmath29 is approximatively @xmath30 @xmath5 times the gas density @xmath31 , obtained assuming a gas to dust ratio in mass equal to 1:100 . after being released into the gas phase , water either freezes out back onto a grain , or it is photo - dissociated by the fuv photons at the rates : @xmath32 and @xmath33 where @xmath34 is the sticking coefficient , for which we use 1 @xcite .
@xmath35 is h@xmath3o molecule thermal velocity .
@xmath36s@xmath37 is the rate of fuv dissociation in unshielded @xmath38 field .
gaseous atomic oxygen can freeze out onto grains , where it is assumed to form water ice by reactions with hydrogen atoms on the grain surfaces .
o atoms may also form water in the gas phase via the standard sequence of reactions started by the reaction between o and h@xmath39 .
the gas phase formation of water thus proceeds at a rate : @xmath40 where @xmath41 is equal to @xmath42 s@xmath37 cm@xmath14 ( the rate coefficient for the reaction o + h@xmath39 @xmath44 oh@xmath45 + h@xmath3 ) times 0.33 ( the last factor accounts for the fraction of h@xmath46o@xmath45 recombinations with electrons forming h@xmath3o ) .
the atomic oxygen abundance is also computed by the steady state equilibrium between formation and destruction processes : photo - dissociation of gas phase water ( eq . ( [ eq:3 ] ) ) for the formation , and formation of gaseous water via reaction with h@xmath39 ( eq . ( [ eq:4 ] ) ) and
freezing onto the grains for the destruction of o. we assume that these reactions are the dominating processes , and that all the oxygen not contained in co or silicates is contained in o , h@xmath3o , and h@xmath3o ice . finally , the h@xmath39 abundance is computed as follows .
h@xmath39 is formed by cosmic ray ionization of h@xmath3 to form h@xmath47 , followed by the reaction with h@xmath3 to form h@xmath39 . in the regions where
co molecules are not frozen onto the grain mantles , h@xmath39 is destroyed by the reaction with co. elsewhere , the h@xmath39 abundance is computed following the model described in @xcite , which takes into account all three deuterated forms of h@xmath39 and solves the chemical composition by considering the reactions between h@xmath45 , e@xmath48 , grains , and h@xmath39 isotopologues .
figure [ fig : abundance ] shows the h@xmath3o abundance ( with respect to h@xmath3 ) across the disk , for a standard interstellar uv field ( @xmath38 ) .
figure [ fig : pdr ] shows a vertical cross - section of the chemical species involved in water formation at a radius of 700 au . despite the relatively large densities in the disk , water vapor has an abundance of @xmath2 in a large fraction of the outer disk , in the layers just above the midplane , where the a@xmath49 to the disk surface is lower than @xmath50 mag . by midplane here we mean the region where more than 2/3 of co molecules are frozen onto the grain mantles .
this occurs at an height of about 230 au at a radius of 700 au , and 100 au at a radius of 400 au . above this height
, co is desorbed thermallyo abundance in our model . ] .
the gas phase water abundance peaks near the surface where the fuv desorbing flux is high and where the h@xmath39 abundance is high . in practice , the freeze - out rate of the h@xmath3o molecules is larger than the fuv photo - desorption rate of the grain mantles only at densities larger than about @xmath51 cm@xmath14 . at lower densities , assuming water is in the grain surfaces , water vapor is formed mostly through direct photo - desorption from mantles , and , to a lesser extent , by gas phase reactions occurring among atomic oxygen ( a product of h@xmath3o photo - dissociation ) and h@xmath39 . in the upper layers ,
atomic oxygen is predicted to be very abundant .
the resulting average gas phase h@xmath3o column density of a face - on disk is @xmath52 @xmath5 .
figure [ fig : coldens ] shows how the column density is distributed as function of the radius .
the column density is remarkably constant over most of the disk , although the column does rise slightly with radius .
the first conclusion of our study is that _ vapor water can indeed be found with relatively high abundances ( @xmath53 ) and column densities ( @xmath4 @xmath5 ) in the protoplanetary disks which surround low luminosity protostars illuminated by the interstellar radiation field ( isrf ) , like in the case of dm tau_. this is caused by the photo - desorption of the icy grain mantles by the average isrf .
increasing the fuv field by a factor of ten and hundred leads to h@xmath3o column densities larger by only factor 1.5 and 2 respectively . decreasing the fuv field by a factor ten results in decreasing the h@xmath3o column density by a factor a bit more than a factor 2 .
therefore , the h@xmath3o column density is not sensitive to the addition of a possible fuv field from the central source or a nearby hot star ( see hollenbach et al .
2005 for details and an analytical proof of the insensitivity to g@xmath54 ) .
the fundamental reason for this insensitivity is that in higher fields , the increased photo - desorption of the ice is balanced by the increased photo - dissociation of gas phase h@xmath3o .
as discussed above , the value of the photo - desorption yield , @xmath21 , is only constrained within a factor roughly three to ten by laboratory experiments .
this uncertainty causes an uncertainty in the predicted h@xmath3o column density by the same factor .
another important , and poorly known parameter that enters in these computations is the dust - to - gas ratio . in the standard case we assumed the canonical mass dust - to - gas ratio equal to 1% . if this value is increased by a factor ten ( which means decreasing by a factor 10 the gas density in the plot of fig .
[ fig : structure ] ) , the h@xmath3o column density increases by a factor @xmath55 . on the contrary ,
a decrease by a factor ten of the dust - to - gas ratio leads to a decrease by a factor 1.2 of the h@xmath3o column density .
this is due to the fact that the uv optical depth is determined by the dust distribution only and independent of the gas density .
as long as photo - desorption , photo - dissociation and freeze - out are the dominant processes , the derived densities and column densities ( as opposed to the abundances ) of h@xmath3o are independent of the gas density .
the small changes are due to regions in which the gas formation rout of h@xmath3o is important .
therefore , also a drastic change in the dust - to - gas ratio does not cause much of a change in the h@xmath3o vapor column density .
finally , the results will also depend on the grain size distribution .
if much smaller grains than 0.1@xmath27 dominate the grain surface area , the balance between the different formation and destruction rates will be shifted .
this question will be discussed in a followup paper . in conclusion , we predict a h@xmath3o column density in protostellar disks similar to that surrounding dm tau of a few times @xmath56 @xmath5 , rather insensitive on the external fuv field and/or the dust - to - gas ratio in the disk .
@xcite reported the detection of hdo in dm tau , with an observed column density of @xmath57 @xmath5 .
based on the present theoretical model , the h@xmath3o column density is predicted to be @xmath58 @xmath5 , within about a factor of a few , given the uncertainties in the yield , fuv field and the dust - to - gas ratio . this implies a hdo / h@xmath3o ratio equal to about 1% .
this is consistent with estimates of the hdo / h@xmath3o ratio in embedded low mass protostars .
measured a hdo / h@xmath3o ratio equal to 3% in the sublimated ices surrounding iras16293 - 2422 , and a much lower ( @xmath590.2% ) value in the outer envelope where the gas phase chemistry dominates the hdo and h@xmath3o formation .
note that the large hdo / h@xmath3o ratio of 3% refers only to the sublimated ices , which are responsible for the water abundance @xmath60 in this object .
very likely , the sublimated ices are only the `` last coating '' of the mantles where the molecular deuteration is the largest , due to the history of ice formation . in agreement with this interpretation
, nir observations of hdo ice have , so far , failed to detect it , at a level of hdo / h@xmath3o@xmath61% @xcite therefore , only a small fraction of elemental deuterium is iced on the mantles .
the hdo / h@xmath3o ratio in dm tau , equal to @xmath6% , is just slightly lower than the value measured in iras16293 - 2422 and attributed to sublimated ices .
if the values were indeed different , this might point to reprocessing of the ices in the disk - however , for this an actual measurement of the vapor water in dm tau is required .
the measurement of the h@xmath3o column density would also allow to measure the photo - desorption yield .
in addition , measures of the hdo / h@xmath3o ratio in protoplanetary disks will probe the history of the ice formation , with important implications for the chemical and isotopic composition of solar system objects . based on our model
, we predict the ground state transition of h@xmath3o at 557 ghz to be in absorption similar to the hdo ground transition at 464 ghz .
the line is predicted to be highly optically thick , with @xmath62 .
unfortunately , neither the currently available odin nor the planned hso satellites will be able to detect the continuum of dm tau , because of the large beam dilution of the signal from the disk .
the results reported in this _ letter _
lead to the following important conclusions : + 1 .
an external uv radiation field as weak as the average habing field ( g@xmath63=1 ) can produce a layer of h@xmath3o on top of a protoplanetary disk ( like that surrounding dm tau ) with a column of a few times 10@xmath64@xmath5 .
this column density is insensitive to external parameters like the strength of the radiation field : variations in g@xmath63 by two orders of magnitude only change n(h@xmath3o ) by a factor of 2 .
+ 3 . similarly , changing the dust - to - gas ratio ( i.e. the gas density at a given dust density ) has only weak influence on the derived column densities .
because of the weak dependence on parameters , the measurement of the h@xmath3o column density in protoplanetary disks is an excellent tool to measure the photo - desorption yield @xmath21 .
+ 5 . combined with previous observations of hdo @xcite , we predict an hdo / h@xmath3o ratio in the outer disk of dm tau of 0.01 , with an uncertainty of about a factor of a few .
we predict that the h@xmath3o ground state line at 557 ghz is undetectable in dm tau with odin and/or hso - hifi observations .
+ finally , we remark that other molecules trapped in the water ices could also be released into the gas phase , keeping `` alive '' the gas chemistry in the layers above the midplane of the outer disk .
cd and cc acknowledge travel support through the dutch / french van gogh program , project vgp 78 - 387 and the hospitality at nasa ames during the development of this work , during which cd received travel funds from the university of california , berkeley .
dh acknowledges support from the nasa astrophysical theory , dynamics and origins of solar systems . this work has benefited from research funding from the european community s sixth framework programme .
we thank the referee juri aikawa for a detailed report .
n(h@xmath3o)/n(h@xmath3 ) as a function of position in the disk .
the solid line shows where the uv optical depth measured vertically to the disk plane is unity : @xmath23 ( see text ) .
the dashed lines show iso density contours : from the top to the bottom they are @xmath65 , @xmath66 , @xmath67 and @xmath51 cm@xmath14 .
the white region was excluded from the calculation because of the low density there.,width=340 ] the abundances and the visual extinction a@xmath68 ( from the disk surface ) as a function of the height above midplane at 700 au from the star .
the solid and dashed lines show the abundance of h@xmath3o and o , respectively .
the dashed - dotted and the dotted - dashed lines show the abundance of co and h@xmath39 .
the dotted line indicates the @xmath69 .
the most important contribution to the h@xmath3o column density is produced between @xmath70 and @xmath71 mag , and a height between 100 and 250 au .
co freezes out at a height of @xmath72 au .
this causes a small increase of the h@xmath39 abundance , since co ( along with o ) is the main destroyer of h@xmath39 . the high abundance of co at very low a@xmath73 may be artificial because we have ignored co photo - dissociation
the water ice abundance is not shown because it is almost constant @xmath74 throughout the plot .
, width=340 ] gas - phase h@xmath3o column density measured perpendicular to the disk midplane as function of the radius . in the very inner regions the h@xmath3o column density decreases because the gas densities at a@xmath75 are higher , leading to a greater proportion of h@xmath3o that is frozen out onto grains as opposed to photo - desorbed .
once the grains are hotter than 100 k ( @xmath76au ) , water will not freeze - out and gas phase abundances will rise again ( not shown).,width=340 ] | recent observations of the ground state transition of hdo at 464 ghz towards the protoplanetary disk of dm tau have detected the presence of water vapor in the regions just above the outer disk midplane @xcite . in the absence of non - thermal desorption processes
, water should be almost entirely frozen onto the grain mantles and hdo undetectable . in this letter
we present a chemical model that explores the possibility that the icy mantles are photo - desorbed by fuv ( 6ev @xmath0 13.6ev ) photons .
we show that the average interstellar fuv field is enough to create a layer of water vapor above the disk midplane over the entire disk . assuming a photo - desorption yield of @xmath1 , the water abundance in this layer
is predicted to be @xmath2 and the average h@xmath3o column density is @xmath4 @xmath5 .
the predictions are very weakly dependent on the details of the model , like the incident fuv radiation field , and the gas density in the disk . based on this model
, we predict a gaseous hdo / h@xmath3o ratio in dm tau of @xmath6% .
in addition , we predict the ground state transition of water at 557 ghz to be undetectable with odin and/or hso - hifi . |
the discovery of the accelerated expansion of the universe @xcite has generated high interest in consistent long - distance modification of gravity .
although gravity might be modified so as to account for cosmic acceleration , modification must be suppressed at short distances because general relativity is confirmed to very high accuracy in the solar system .
thus , for a modified theory of gravity to be consistent , it is important to equip a mechanism to screen the effect of modification that could otherwise persist down to small scales . a wide class of modified gravity theories can be ( at least effectively ) described in terms of a scalar - tensor theory , and the screening mechanisms can be understood by inspecting the structure of the scalar - tensor lagrangian .
mainly two mechanisms have been proposed so far .
one is the chameleon mechanism @xcite , in which a density - dependent effective potential is introduced to make the scalar field sufficiently massive in a high - density environment .
this mechanism is utilized for example in @xmath0 gravity @xcite .
another mechanism is known as the vainshtein effect@xcite , which operates in models with nonlinear derivative interactions . in such models
, nonlinearity becomes large in the vicinity of a matter source to suppress the scalar - mediated force .
this mechanism is incorporated for example into the dgp model @xcite , the galileon model @xcite , and massive gravity @xcite . in this paper
, we discuss how the vainshtein effect works in horndeski s scalar - tensor theory @xcite .
horndeski s theory is the most general theory composed of a single scalar field and a metric that gives rise to second - order field equations , and therefore includes a large number of concrete models of modified gravity as specific cases .
we present an algebraic master equation that determines a gravitational field around any static and spherically symmetric distribution of nonrelativistic matter .
the similar problem of deriving the vainshtein solution in the cosmological background was addressed in ref .
there , however , complete analysis was hindered by the complexity of the governing equations . in the case
where the effect of the cosmic expansion may be ignored , the relevant equations are simplified so that the problem actually reduces to what has been done in the context of the galileon model @xcite and massive gravity @xcite .
it is therefore possible to determine completely the conditions under which the vainshtein solutions are available .
our solution thus obtained is useful in testing modified gravity . among various cosmological observations , we are particularly interested in cluster lensing , which can provide the convergence over a wide range of radii by combining strong- and weak - lensing measurements ( see refs .
@xcite , and references therein ) .
the precise measurement of the structure of galaxy clusters offers us an important clue to test modified theories of gravity @xcite .
the first paper on gravitational lensing in the context of massive gravity is ref .
the gravitational sector considered there is given by a subclass of horndeski s theory
. however , ref .
@xcite considered nonminimal and disformal coupling to matter , while we assume that matter is minimally coupled to gravity .
working in horndeski s theory , one can examine modified gravity endowed with the vainshtein screening mechanism in a generic manner . in this paper
, we compute the lensing convergence of galaxy clusters in horndeski s theory based on the general spherically symmetric solution . as a specific example
, we demonstrate how nonlinear massive gravity proposed recently in ref .
@xcite is included in our analysis , and explore the possibility of detecting the signatures of modified gravity in the context of massive gravity .
this paper is organized as follows : in the next section , we define horndeski s lagrangian and explain our procedure to obtain static and spherically symmetric solutions .
then , in sec . [ sec : eq ]
, we derive the vainshtein solution and present the conditions for the existence of the solution .
we show the relation between horndeski s theory and the decoupling limit of massive gravity in sec .
[ sec : proxy ] . in sec .
[ sec : lens ] we study the effect of modification of gravity on gravitational lensing and explore its implications for cluster lensing observations . section [ sec : concl ] is devoted to conclusions .
we consider the most general scalar - tensor theory constructed by horndeski .
the action we study is given by @xmath1 + s_{\rm m}\left[g_{\mu\nu},\psi_{\rm m}\right],\end{aligned}\ ] ] where @xmath2 is the action for matter @xmath3 , which is assumed to be minimally coupled to gravity .
the first term is composed of the four lagrangians , @xmath4 where @xmath5 , \\ { \cal l}_5&= & g_5(\phi , x)g^{\mu\nu}\nabla_\mu\nabla_\nu\phi -\frac{1}{6}g_{5x}\bigl[(\box\phi)^3 \nonumber\\&&\quad -3\box\phi(\nabla_\mu\nabla_\nu\phi)^2 + 2(\nabla_\mu\nabla_\nu\phi)^3\bigr],\end{aligned}\ ] ] with @xmath6 . the second term is the action for matter fields . here
, @xmath7 is the ricci scalar , @xmath8 is the einstein tensor , and @xmath9 and @xmath10 are arbitrary functions of @xmath11 and @xmath12 . here
we have written the lagrangians in the equivalent form called the generalized galileon rather than the original one . in this paper , we use notation such as @xmath13 and @xmath14 .
we are interested in spherical overdensities on subhorizon scales , so that we will neglect the effect of cosmic expansion . in ref .
@xcite , the vainshtein mechanism in horndeski s theory has been investigated taking into account the background evolution of the scalar field , @xmath15 , and it was shown that newton s `` constant '' evolves in time through its dependence on @xmath16 and @xmath17 ( see also ref .
@xcite for a similar argument )
. however , the time variation of newton s `` constant '' is strongly constrained from experiments @xcite .
this leads us to assume that the scalar field has a negligible time dependence at least during the relevant period , @xmath18 const and @xmath19 . our background solution is thus taken to be @xmath20 in order for the theory to admit this solution , we require that @xmath21 and @xmath22 .
spherically symmetric perturbations produced by a nonrelativistic matter lump on top of the background ( [ bg ] ) can be written as @xmath23{{\rm d}}t^2+[1 - 2\psi(r)]\delta_{ij}{{\rm d}}x^i{{\rm d}}x^j , \\
\phi & = & \phi_0+\varphi(r),\end{aligned}\ ] ] where @xmath24 is the usual radial coordinate , @xmath25 .
the time - time component of the gravitational field equations is @xmath26'}{2r^2 } \nonumber\\&&\quad + g_{5x}\frac{[(\varphi')^3]'}{6r^2 } = - \frac{1}{4}t_{t}^{\;t},\label{f1}\end{aligned}\ ] ] while the space - space component reduces to @xmath27 where a prime stands for differentiation with respect to @xmath24 .
finally , from the scalar - field equation of motion we obtain @xmath28'}{r^2 } \nonumber\\ & & + 2g_{4\phi } \frac{[r^2(2\psi -\phi)']'}{r^2 } + 4(g_{4x}-g_{5\phi})\frac{[r\varphi'(\psi'-\phi')]'}{r^2 } \nonumber\\ & & + 2\left(g_{4xx}-\frac{2}{3}g_{5\phi x}\right ) \frac{[(\varphi')^3]'}{r^2}+2g_{5x}\frac{[(\varphi')^2\phi']'}{r^2 } \nonumber\\ & & = -k_{\phi\phi}\varphi.\label{f3}\end{aligned}\ ] ] here , all the functions in the coefficients are evaluated at @xmath29 and @xmath30 . from now on , we will ignore the mass term @xmath31 , because we focus only on modified gravity endowed with the vainshtein mechanism .
one sees that eqs .
( [ f1 ] ) and ( [ f3 ] ) can be integrated once to give algebraic equations for @xmath32 . in doing
so it is convenient to use the enclosed mass defined as @xmath33 the resultant equations coincide with those derived from taking the limit @xmath34 const and @xmath30 in the result of ref .
@xcite .
let us introduce six dimensionless parameters , @xmath35 , and @xmath36 , as well as the planck mass @xmath37 and a new mass scale @xmath38 , to rearrange and simplify the expressions .
those dimensionless quantities are related to the coefficients in the above equations as @xmath39 and @xmath40 we also define @xmath41 both of which are dimensionless .
now the gravitational field equations ( [ f1 ] ) and ( [ f2 ] ) reduce to @xmath42 substituting eqs .
( [ psi - eq ] ) and ( [ phi - eq ] ) to the scalar - field equation of motion ( [ f3 ] ) , we arrive at @xmath43x^2 + \left(\nu+2\alpha^2 + 4\beta\xi\right)x^3 - 3\beta^2x^5=0.\label{basic - equation}\end{aligned}\ ] ] solving the algebraic equation @xmath44 for @xmath45 , one obtains the radial profile of @xmath45 in terms of @xmath46 .
it is then straightforward to determine the two metric potentials by using eqs .
( [ phi - eq ] ) and ( [ psi - eq ] ) .
note in passing that if @xmath47 then we have a trivial solution @xmath48 which is not interesting . in the rest of the paper
we therefore assume that @xmath49 .
since there is a sufficient number of parameters , at this stage the coefficients of the polynomial @xmath50 are free in principle .
however , it is important to note that the structure of @xmath50 in the most general case is still essentially the same as the corresponding equation in massive gravity : it is quintic and the matter source term @xmath51 appears only in the zeroth - order and quadratic terms .
this structure allows us to proceed following closely the previous analysis in massive gravity @xcite .
in this section , we solve the equation @xmath52 for @xmath53 and @xmath54 , and single out a solution appropriate for our current purpose in each domain .
we then derive the conditions under which the two solutions are matched smoothly in an intermediate region .
the procedure here is basically the same as that of ref .
the region far from the source corresponds to @xmath54 , while it is assumed that @xmath53 in the vicinity of the source , where the vainshtein mechanism is expected to operate .
it is therefore appropriate to define the vainshtein radius @xmath55 by @xmath56 in the outer region ( @xmath54 ) , there is always a decaying solution , @xmath57 which is obtained by neglecting the nonlinear terms in @xmath50 .
we are interested only in this solution , because the other solutions , if they exist , do not correspond to an asymptotically flat spacetime . the stability of the solution in the linear regime can be studied by using the truncated action @xmath58.\end{aligned}\ ] ] working in this action , one can derive the stability condition easily in the same way as in the brans - dicke theory : _ in the einstein frame _ , the kinetic term for small fluctuations has the right sign provided that @xmath59 we require the condition ( [ stability - linear ] ) for the stability of the solution @xmath60 .
the same condition can also be derived from @xmath61 .
( see appendix a for further details . )
let us turn to identifying the desired inner solution .
the inner solution is different depending on whether @xmath62 or @xmath63 , because the structure of @xmath50 for @xmath53 is crucially different . in the inner region ( @xmath53 ) ,
( [ basic - equation ] ) reduces to @xmath64 the behavior of the solution to this equation depends on the sign of @xmath65 .
+ ( i ) @xmath66 . in this case , the second and the third terms in eq .
( [ large - a ] ) balance , so that the solution is @xmath67 for this inner solution we find @xmath68 and @xmath69 , which is not at all standard gravity .
we therefore discard this possibility .
\(ii ) @xmath70 . in this case
, we have the same solution as above , @xmath71 , which is to be discarded for the same reason .
we have another solution , @xmath72 for which the first and the second terms in eq .
( [ large - a ] ) balance . for this latter solution
we have the correct newtonian behavior : @xmath73 .
this is therefore the vainshtein solution we are looking for .
+ having thus identified the desired solutions in the inner and outer regions , we consider the matching of the two . for simplicity
we focus on the case of @xmath74 .
the solution matching in the case of @xmath75 can also be done essentially in the same way as follows . for @xmath74 ( and hence @xmath76 )
, the outer solution @xmath77 can be matched only to the inner solution @xmath78 smoothly .
, the solution @xmath79 is stable , but the other one , @xmath80 , is not . ]
the smooth matching of the two solution is possible if and only if _ @xmath44 has a single root in @xmath81 for any @xmath82 .
_ since @xmath83 , if @xmath84 the intermediate value theorem guarantees the existence of at least one root in @xmath81 .
note that @xmath85 does not in fact depend on @xmath51 . using the intermediate value theorem again in @xmath86 and @xmath87
, it can be shown that one or three roots exist in general in @xmath88 .
let @xmath89 and @xmath90 be the solution to @xmath91 if such @xmath92 and @xmath93 do not exist , it is obvious that there is only a single root in @xmath81 . in this case , the smooth matching is possible .
if such @xmath89 and @xmath90 exist , @xmath44 would in general have three roots in @xmath81 for some interval of @xmath51 .
since @xmath44 for @xmath53 [ eq . ( [ large - a ] ) ] has a single root in @xmath81 , two of the three roots in @xmath94 disappear as @xmath51 increases . for parameters satisfying @xmath95 , the solution corresponding to @xmath96 disappears as @xmath51 increases , implying that the smooth matching is impossible in this case . however , if @xmath97 the solution corresponding to @xmath96 remains and hence the smooth matching is still possible .
indeed , there is a single root for any @xmath82 in this case because a simple manipulation shows that @xmath98 increases with increasing @xmath51 , where @xmath99 is the locus of extrema in @xmath81 . summarizing , the outer and inner solutions can be matched smoothly provided that eq .
( [ cond1 ] ) and either of the following two conditions are satisfied : ( i ) @xmath92 and @xmath93 do not exist satisfying eq .
( [ exist2 ] ) ; ( ii ) such @xmath92 and @xmath93 exist , but they satisfy eq .
( [ cond2 ] ) . here
again we focus on the case of @xmath74 for simplicity .
the problem reduces to solving the cubic equation , @xmath44 , where now @xmath100 the inner solution is @xmath101 ( we assume that @xmath102 . )
this solution shows the correct newtonian behavior : @xmath73 .
the inner and outer solutions can be matched smoothly only for @xmath103 , which is also required from stability of the inner solution .
smooth matching also requires that there is a single root in @xmath104 for any @xmath82 , or , equivalently , that @xmath50 has no local extrema in @xmath104 .
the two local extrema are in @xmath105 provided that @xmath106 where we used the condition ( [ stability - linear ] ) .
otherwise , one must require that the discriminant of @xmath107 is negative , _ i.e. , _ @xmath108 summarizing , smooth matching is possible if eq .
( [ c3 - 1 ] ) or eq .
( [ c3 - 2 ] ) is satisfied .
let us confirm that the conditions for smooth matching indeed reproduce the previous result obtained in the context of massive gravity @xcite .
to do so , we start with finding out the concrete form of @xmath109 corresponding to the decoupling limit of massive gravity . the correspondence can be seen more clearly if we move to the covariantized version of the decoupling limit lagrangian , _
i.e. , _ the `` proxy theory '' proposed in ref .
@xcite . since the proxy theory contains the riemann dual tensor while the lagrangian of the generalized galileon not
, one may wonder how the former is included in the latter .
actually , @xmath113 corresponds to the term containing the riemann dual tensor in the proxy theory .
the easiest way to verify this is to compare the field equations of the two theories . from eq .
( [ proxy - g ] ) one finds @xmath114 so that the parameter space collapses to a two - dimensional space .
the inner solution @xmath115 exists only for @xmath76 and is given by @xmath116 .
let us define @xmath117 .
then , the condition ( [ cond1 ] ) reads @xmath118 solving the equation @xmath119 , which does not in fact depend on @xmath90 , one finds @xmath120^{1/2}.\end{aligned}\ ] ] this exists if latexmath:[\ ] ] we can specify the einasto profile with the virial cluster mass @xmath136 , the special radius @xmath197 , and the slope index @xmath167 . based on the millennium simulation @xcite , the authors of ref .
@xcite claimed that cold dark matter halos can be more properly described by the einasto profile than by the nfw profile .
they also argued that the best - fit value of @xmath167 increases gradually with the increase of the virial mass , from @xmath218 for galaxy halos to @xmath219 for the most massive clusters .
a. g. riess _ et al .
_ [ supernova search team collaboration ] , astron . j. * 116 * , 1009 ( 1998 ) [ astro - ph/9805201 ]
. s. perlmutter _ et al .
_ [ supernova cosmology project collaboration ] , astrophys .
j. * 517 * , 565 ( 1999 ) [ astro - ph/9812133 ] .
j. khoury and a. weltman , phys .
lett . *
93 * , 171104 ( 2004 ) [ astro - ph/0309300 ] .
a. de felice and s. tsujikawa , living rev .
* 13 * , 3 ( 2010 ) [ arxiv:1002.4928 [ gr - qc ] ] .
a. i. vainshtein , phys .
b * 39 * , 393 ( 1972 ) .
g. r. dvali , g. gabadadze and m. porrati , phys .
b * 485 * , 208 ( 2000 ) [ hep - th/0005016 ] .
a. lue , phys .
* 423 * , 1 ( 2006 ) [ astro - ph/0510068 ] .
a. nicolis , r. rattazzi and e. trincherini , phys .
d * 79 * , 064036 ( 2009 ) [ arxiv:0811.2197 [ hep - th ] ]
. m. fierz and w. pauli , proc .
a * 173 * , 211 ( 1939 ) .
v. a. rubakov and p. g. tinyakov , phys .
usp . * 51 * , 759 ( 2008 ) [ arxiv:0802.4379 [ hep - th ] ] .
g. w. horndeski , int .
j. theor .
* 10 * , 363 ( 1974 ) .
c. deffayet , x. gao , d. a. steer and g. zahariade , phys .
d * 84 * , 064039 ( 2011 ) [ arxiv:1103.3260 [ hep - th ] ] .
t. kobayashi , m. yamaguchi and j. i .
yokoyama , prog . theor
. phys . * 126 * , 511 ( 2011 ) [ arxiv:1105.5723 [ hep - th ] ] .
r. kimura , t. kobayashi and k. yamamoto , phys .
d * 85 * , 024023 ( 2012 ) [ arxiv:1111.6749 [ astro-ph.co ] ] .
m. wyman , phys .
lett . * 106 * , 201102 ( 2011 ) [ arxiv:1101.1295 [ astro-ph.co ] ]
. g. chkareuli and d. pirtskhalava , phys .
b * 713 * , 99 ( 2012 ) [ arxiv:1105.1783 [ hep - th ] ] .
s. sjors and e. mortsell , arxiv:1111.5961 [ gr - qc ] .
f. sbisa , g. niz , k. koyama and g. tasinato , phys .
d * 86 * , 024033 ( 2012 ) [ arxiv:1204.1193 [ hep - th ] ] .
m. bartelmann and p. schneider , phys .
rept . * 340 * , 291 ( 2001 ) [ astro - ph/9912508 ] .
k. umetsu , arxiv:1002.3952 [ astro-ph.co ] . m. postman , d. coe , n. benitez , l. bradley , t. broadhurst , m. donahue , h. ford and o. graur _ et al .
_ , astrophys .
j. suppl .
* 199 * , 25 ( 2012 ) [ arxiv:1106.3328 [ astro-ph.co ] ] .
l. lombriser , f. schmidt , t. baldauf , r. mandelbaum , u. seljak and r. e. smith , phys . rev .
d * 85 * , 102001 ( 2012 ) [ arxiv:1111.2020 [ astro-ph.co ] ] .
t. narikawa and k. yamamoto , jcap * 1205 * , 016 ( 2012 ) [ arxiv:1201.4037 [ astro-ph.co ] ] .
c. de rham , g. gabadadze and a. j. tolley , phys .
lett . *
106 * , 231101 ( 2011 ) [ arxiv:1011.1232 [ hep - th ] ] .
e. babichev , c. deffayet and g. esposito - farese , phys .
lett . *
107 * , 251102 ( 2011 ) [ arxiv:1107.1569 [ gr - qc ] ] .
j. g. williams , s. g. turyshev and d. h. boggs , phys .
lett . * 93 * , 261101 ( 2004 ) [ gr - qc/0411113 ] . c. de rham and l. heisenberg , phys .
d * 84 * , 043503 ( 2011 ) [ arxiv:1106.3312 [ hep - th ] ] .
j. f. navarro , c. s. frenk and s. d. m. white , astrophys .
j. * 462 * , 563 ( 1996 ) [ astro - ph/9508025 ] .
j. f. navarro , c. s. frenk and s. d. m. white , astrophys .
j. * 490 * , 493 ( 1997 ) [ astro - ph/9611107 ] . t. j. broadhurst _ et al .
_ [ acs collaboration ] , astrophys . j. * 621 * , 53 ( 2005 ) [ astro - ph/0409132 ] . k. umetsu and t. broadhurst , astrophys .
j. * 684 * , 177 ( 2008 ) [ arxiv:0712.3441 [ astro - ph ] ] . k. umetsu , t. broadhurst , a. zitrin , e. medezinski and l. -y .
hsu , astrophys . j. * 729 * , 127 ( 2011 ) [ arxiv:1011.3044 [ astro-ph.co ] ] .
k. umetsu , t. broadhurst , a. zitrin , e. medezinski , d. coe and m. postman , astrophys . j. * 738 * , 41 ( 2011 ) [ arxiv:1105.0444 [ astro-ph.co ] ] .
h. zhao , mon . not .
278 * , 488 ( 1996 ) [ astro - ph/9509122 ] .
jing and y. suto , astrophys .
j. * 529 * , l69 ( 2000 ) [ astro - ph/9909478 ] .
r. mandelbaum , u. seljak , r. j. cool , m. blanton , c. m. hirata and j. brinkmann , mon . not .
soc . * 372 * , 758 ( 2006 ) [ astro - ph/0605476 ] .
j. einasto , trudy inst .
astrofiz .
alma - ata , * 5 * , 87 ( 1965 ) .
j. f. navarro , e. hayashi , c. power , a. jenkins , c. s. frenk , s. d. m. white , v. springel and j. stadel _ et al .
_ , mon . not .
soc . * 349 * , 1039 ( 2004 ) [ astro - ph/0311231 ] .
l. gao , j. f. navarro , s. cole , c. frenk , s. d. m. white , v. springel , a. jenkins and a. f. neto , mon . not .
soc . * 387 * , 536 ( 2008 ) [ arxiv:0711.0746 [ astro - ph ] ] .
e. hayashi and s. d. m. white , mon . not .
* 388 * , 2 ( 2008 ) [ arxiv:0709.3933 [ astro - ph ] ] . c. de rham , g. gabadadze , l. heisenberg and d. pirtskhalava , phys .
d * 83 * , 103516 ( 2011 ) [ arxiv:1010.1780 [ hep - th ] ] .
k. koyama , g. niz and g. tasinato , jhep * 1112 * , 065 ( 2011 ) [ arxiv:1110.2618 [ hep - th ] ] . c. de rham , comptes rendus physique * 13 * , 666 ( 2012 ) [ arxiv:1204.5492 [ astro-ph.co ] ] .
g. damico , c. de rham , s. dubovsky , g. gabadadze , d. pirtskhalava and a. j. tolley , phys .
d * 84 * , 124046 ( 2011 ) [ arxiv:1108.5231 [ hep - th ] ] .
a. de felice , a. e. gumrukcuoglu and s. mukohyama , phys .
lett . * 109 * , 171101 ( 2012 ) [ arxiv:1206.2080 [ hep - th ] ] .
a. e. gumrukcuoglu , c. lin and s. mukohyama , phys .
b * 717 * , 295 ( 2012 ) [ arxiv:1206.2723 [ hep - th ] ] .
g. damico , g. gabadadze , l. hui and d. pirtskhalava , arxiv:1206.4253 [ hep - th ] .
v. springel , s. d. m. white , a. jenkins , c. s. frenk , n. yoshida , l. gao , j. navarro and r. thacker _ et al .
_ , nature * 435 * , 629 ( 2005 ) [ astro - ph/0504097 ] . | we explore the possibility of testing modified gravity exhibiting the vainshtein mechanism against observations of cluster lensing .
we work in the most general scalar - tensor theory with second - order field equations ( horndeski s theory ) , and derive static and spherically symmetric solutions , for which the scalar field is screened below a certain radius .
it is found that the essential structure of the problem in the most general case can be captured by the program of classifying vainshtein solutions out of different solutions to a quintic equation , as has been performed in the context of massive gravity .
the key effect on gravitational lensing is that the second derivative of the scalar field can substantially be large at the transition from screened to unscreened regions , leaving a dip in the convergence .
this allows us to put observational constraints on parameters characterizing the general scalar - tensor modification of gravity .
we demonstrate how this occurs in massive gravity as an example , and discuss its observational signatures in cluster lensing . |
in recent years , quantum statistical mechanics and quantum information theory have played an increasingly central role in quantum gravity .
such interplay has proved particularly insightful both in the context of the holographic duality in ads / cft @xcite , as well as for the current background independent approaches to quantum gravity , including loop quantum gravity ( lqg ) @xcite , the related spin - foam formulation @xcite , and group field theory @xcite .
interestingly , the different background independent approaches today share a microscopic description of space - time geometry given in terms of discrete , pre - geometric degrees of freedom of combinatorial and algebraic nature , based on spin - network hilbert spaces @xcite . in this context , entanglement has provided a new tool to characterise the quantum texture of space - time in terms of the structure of correlations of the spin networks states . along this line
, several recent works have considered the possibility to use specific features of the short - range entanglement in quantum spin networks ( area law , thermal behaviour ) to select quantum geometry states which may eventually lead to smooth spacetime geometry classically @xcite .
this analysis usually focuses on states with few degrees of freedom , leaving open the question of whether a _ statistical
_ characterisation may reveal new structural properties , independent from the interpretation of the spin network states . in this work we proposes the use of the information theoretical notion of quantum _ canonical typicality _ , as a tool to investigate and characterise universal local features of quantum geometry , going beyond the physics of states with few degrees of freedom . in quantum statistical mechanics ,
_ canonical typicality _ states that almost every pure state of a large quantum mechanical system , subject to the fixed energy constraint , is such that its reduced density matrix over a sufficiently small subsystem is approximately in the _ canonical _ state described by a thermal distribution la gibbs @xcite .
such a statement goes beyond the thermal behaviour . for a generic closed system in a quantum pure state ,
subject to some _ global constraint _
, the resulting canonical description will not be thermal , but generally defined in relation to the constraint considered @xcite .
again , in this case , some specific properties of the system emerge at the local level , regardless of the nature of the global state .
these properties depend on the physics encoded in the choice of the global constraints . within this generalised framework ,
we exploit the notion of _ typicality _ to study whether and how `` universal '' statistical features of the local correlation structure of a spin - network state emerge in connection with the choice of the global constraint .
we focus our analysis on the space of the @xmath1-valent @xmath0 invariant intertwiners , which are the building blocks of the spin network states . in lqg
, such intertwiners can be thought of dually as a region of @xmath2 space with an @xmath3 boundary @xcite .
we reproduce the typicality statement in the full space of @xmath1-valent intertwiners with fixed total area and we investigate the statistical behaviour of the canonical reduced state , dual to a small patch of the @xmath3 boundary , in the large @xmath1 limit . eventually , we study the entropy of such a reduced state and its area scaling behaviour in different thermodynamic regimes .
the content of the manuscript is organised as follows .
section [ tysn ] introduces the statement of canonical typicality in a formulation particularly suitable for the spin network hilbert space description .
section [ snst ] shortly reviews the notion of state of quantum geometry in terms of spin network basis . in section [ cano ]
we reformulate the statement of quantum typicality in this context .
we derive the notion of canonical reduced state of the @xmath1-valent intertwiner spin network system in section [ redu ] and we prove the existence of a regime of typicality for such system in section [ typ ] .
the entropy of the typical state and its thermodynamical interpretation are investigated in section [ thermo ] .
we conclude in section [ fine ] with a short discussion of our results .
technical details of all the computations are given in the supplementary material .
we start with a brief summary of the result achieved in @xcite .
suppose we have a generic _ closed _ system , which we call `` universe '' , and a bipartition into `` small system '' @xmath4 and `` large environment '' @xmath5 .
the universe is assumed to be in a pure state .
we also assume that it is subject to a completely arbitrary _ global constraint _
for example , in the standard context of statistical mechanics it can be the fixed energy constraint .
such constraint is concretely imposed by restricting the allowed states to the subspace @xmath7 of the states of the total hilbert space @xmath8 which satisfy the constraint @xmath6 : @xmath9 @xmath10 and @xmath11 are the hilbert spaces of the system and environment , with dimensions @xmath12 and @xmath13 , respectively .
we also need the definition of the canonical state of the system @xmath14 , obtained by tracing out the environment from the microcanonical ( maximally mixed ) state @xmath15 @xmath16 & & \mathcal{i}_{\mathcal{r } } \equiv \frac{\mathbb{1}_{\mathcal{r } } } { d_{\mathcal{r}}}\end{aligned}\ ] ] where @xmath17 is the projector on @xmath7 , and @xmath18 .
this corresponds to assigning _ a priori _ equal probabilities to all states of the universe consistent with the constraints @xmath19 . in this
setting , given an arbitrary pure state of the universe satisfying the constraint @xmath6 , i.e. @xmath20 , the reduced state @xmath21 $ ] will almost always be very close to the canonical state @xmath14 .
concretely , such a behaviour can be stated as a theorem @xcite , showing that for an arbitrary @xmath22 , the distance between the reduced density matrix of the system @xmath23 and the canonical state @xmath14 is given probabilistically by @xmath24 } { \mathrm{vol } \left [ \phi \in \mathcal{h}_{\mathcal{r } } \right ] } \leq \eta ' \label{result}\end{aligned}\ ] ] where the _ trace - distance _
@xmath25 is a metric . ] on the space of the density matrices @xcite , while @xmath26 & & \eta = \varepsilon + \frac{1}{2 } \sqrt{\frac{d_s}{d_e^{\mathrm{eff } } } } , \end{aligned}\ ] ] with the effective dimension of the environment defined as @xmath27 } \geq \frac{d_{\mathcal{r}}}{d_s } \label{deff}.\end{aligned}\ ] ] the bound in eq .
( [ result ] ) states that the fraction of the volume of the states which are far away from the canonical state @xmath14 more than @xmath28 decreases exponentially with the dimension of the `` allowed hilbert space '' @xmath29 and with @xmath30 .
this means that , as the dimension of the hilbert space @xmath31 grows , a huge fraction of states gets concentrated around the canonical state .
the proof of the result relies on the _ concentration of measure phenomenon_. the key tool to prove this result is the _ levy _ lemma , which we briefly report for completeness in appendix [ levy ] .
+ in the following we will reconsider this argument for a peculiar class of high dimensional quantum systems , the quantum spin network states .
the interest in testing the notion of typicality in quantum gravity resides in the kinematic nature of the statement , a fundamental feature to study the possibility of a thermal characterisation of reduced states of quantum geometry , regardless of any hamiltonian evolution in time .
in several background - independent approaches to quantum gravity , the spin network states provide a _
kinematical _ description of quantum geometry , in terms of superpositions of graphs @xmath32 labelled by group or lie algebra elements representing holonomies of the gravitational connection and their conjugate triad @xcite .
these states are constructed as follows ( for a thorough introduction to spin - networks we refer to @xcite ) . to each edge
@xmath33 one associates an @xmath0 irreducible representation ( irrep ) labelled by a half - integer @xmath34 called spin . the representation ( hilbert )
space is denoted @xmath35 and has dimension @xmath36 .
to each vertex @xmath37 of the graph one attaches an intertwiner @xmath38 , which is @xmath0-invariant map between the representation spaces @xmath35 associated to all the edges @xmath39 meeting at the vertex @xmath37 , @xmath40 one can alternatively consider @xmath38 as a map from @xmath41 and call the intertwiner an invariant tensor or a singlet state between the representations attached to all the edges linked to the considered vertex .
once the @xmath42 s are fixed , the intertwiners at the vertex @xmath37 actually form a hilbert space , which we will call @xmath43\end{aligned}\ ] ] a spin network state @xmath44 is defined as the assignment of representation labels @xmath42 to each edge and the choice of a vector @xmath45 for the vertices .
the spin network state defines a wave function on the space of discrete connections @xmath46 , @xmath47 = \langle h_e | \mathcal{i}_v \rangle \equiv \text{tr } \bigotimes_e d^{j_e}(h_e ) \otimes \bigotimes_v \mathcal{i}_v\end{aligned}\ ] ] where we contract the intertwiners @xmath38 with the ( wigner ) representation matrices of the group elements @xmath48 in the chosen representations @xmath42 .
therefore , upon choosing a basis of intertwiners for every assignment of representations @xmath49 , the spin networks provide a basis of the space of wave functions associated to the graph @xmath32 , @xmath50= \bigoplus_{\{j_e\ } } \bigotimes_v \mathcal{h}_v .\end{aligned}\ ] ] such discrete and algebraic objects provide a description of the fundamental excitations of quantum spacetime . from a geometrical point of view , classically , given a cellular decomposition of a three - dimensional manifold , a spin - network graph with a node in each cell and edges connecting nodes in neighbouring cells is said to be dual to this cellular decomposition .
therefore , each edge of the graph is dual to a surface patch intersecting the edge and the area of such patch is proportional to the representation @xmath42 .
analogously , vertices of a spin network can be dually thought of as chunks of volume ( see fig.[duality ] for an example ) .
see @xcite for the geometric interpretation of spin - networks states as collection of polyhedra .
, while @xmath38 labels the intertwiner tensor at the node.,width=2 ] in the following , we will focus on a fundamental building block of a spin network graph , the hilbert space of a single intertwiner with @xmath1 legs .
now we consider a large quantum system given by a collection of @xmath1 edges , represented by @xmath1 independent edges states .
the hilbert space of the system is the direct sum over @xmath51 s of the tensor product of @xmath1 irreducible representations @xmath52 , @xmath53 this set of independent edges plays the role of the `` universe '' .
notice that , despite its extreme simplicity , this system has a huge hilbert space .
the single representation space @xmath54 has finite dimension @xmath55 .
however , @xmath56 is summed over all @xmath57 .
therefore each wilson line state ( edge ) lives in an infinite dimensional hilbert space . in the following
, we will always consider a cut - off in the value of the @xmath0 representation labelling the edge .
this is usually done in lqg to include a cosmological constant .
] , @xmath58 this will allow us to deal with a very large but finite dimensional space .
now , we want to split the universe into system and environment .
we do so simply by defining two subsets of edges @xmath5 and @xmath4 , with @xmath59 and @xmath60 , such that @xmath61 .
consequently , we can write the hilbert space of the universe as the tensor product @xmath62 . we would like to stress here that , for our result to hold the @xmath63 links of the system do not need to be adjacent .
despite that , we are interested in the local properties of an intertwiner therefore we will always think about these links as adjacent and forming a simply connected 2d patch .
+ the next step toward typicality consists in defining the constraint which restricts the allowed states of the system and environment to a subspace of the total hilbert space .
two main ingredients are necessary to the definition of the constraint .
the first one is the @xmath0 gauge invariance .
this choice reduces the universe hilbert space to the collection of the @xmath0-invariant linear spaces @xmath64,\end{aligned}\ ] ] spanned by @xmath1-valent intertwiner states .
invariance under @xmath0 is the the first ingredient defining our subsystem constrained space . -legged intertwiner system describes a convex polyhedron with @xmath1 faces , with the topology of a _
2-sphere_. the @xmath1 ( oriented ) edges are dual to the elementary @xmath1 surfaces comprising the surface of the polyhedron @xcite .
the intertwiner contains information on how the elementary surfaces , dual to the links , are combined together to form a surface boundary of the space region dual to the node @xcite.,width=2 ] it has been proven in @xcite that the hilbert space of the @xmath1-valent intertwiners naturally decomposes into subspaces of constant total area is favoured by the forthcoming approach involving the @xmath65 structure of the intertwiner space . ]
which , following the notation in @xcite we call @xmath66 . therefore @xmath67 .
we further constrain our system by considering only the invariant _ tensor product _
hilbert space , with total spin fixed to @xmath68 ( see fig . [ sphere ] ) .
this is the last ingredient .
eventually , the constrained hilbert space is given by @xmath69 .
+ it was also proven in @xcite that each subspace @xmath70 of @xmath1-valent intertwiners with fixed total area @xmath71 carries an irreducible representation of @xmath65 . in this context , one can interpret @xmath72 as the total area dual to the set of @xmath1 legs of the intertwiner .
the main reason behind the choice of @xmath73 as constrained space is that in the semiclassical limit one can think of this system , dually , as a closed surface with area @xmath74 , where @xmath75 is the planck length .
+ once the constrained space has been defined , in order to compute the canonical reduced state , we need the expression of the maximally mixed state @xmath15 over @xmath7 .
this is formally given by @xmath76 where @xmath77 projects the states of @xmath78 onto the @xmath0 gauge invariant subspace with fixed total spin number @xmath73 .
when dealing with @xmath0 quantum numbers there are two common choices for the basis of the hilbert space : the coupled and the decoupled basis .
the coefficients which connect the two basis are the well - known clebsh - gordan coefficient . considering that the main task is to perform the partial trace of @xmath15 over the environment , a suitable basis to write the projector
@xmath77 is a _ semi - decoupled _ basis in which all the quantum numbers within the system and within the environment , respectively , are coupled , but the environment and the system are not . + onto the constrained hilbert space @xmath7 . for a recent work on the splitting of a gauge - invariant system we suggest @xcite.,width=3 ] using such semi - decoupled basis
we can write the projector as @xmath79 where with @xmath80 we mean @xmath81 , @xmath82 and @xmath83 means that we are summing only over the configurations of the spins @xmath84 such that @xmath85 .
the quantum numbers @xmath86 and @xmath87 stand for the recoupling quantum numbers necessary to write the state in the coupled basis , respectively within the system and the environment .
eventually , @xmath88 and @xmath89 are , respectively , the norm of the total angular momentum of the system and its projection over the @xmath90 axis ; @xmath91 and @xmath92 have the same meaning but they refer to the environment .
the details of the generic element of the semi - decoupled basis and of the way in which we obtain the projector can be found in the supplementary material .
+ the dimension of the constrained hilbert space @xmath93 counts the degeneracy of the @xmath1-valent intertwiners with fixed total spin @xmath72 . given the equivalence between the space @xmath73 of @xmath1-valent intertwiners with fixed total area @xmath94 ( including the possibility of trivial su(2 ) irreps ) and the irreducible representation of @xmath65 formalism for @xmath0 intertwiners @xcite , @xmath31 can be calculated as the dimension of the equivalent maximum weight @xmath65 irrep with young tableaux given by two horizontal lines with equal number of cases @xmath72 , @xmath95 thanks to the tensor product structure of the semi - decoupled basis , with respect to the bipartition of the universe into system and environment , we can easily perform the partial trace operation over the environment .
the details of the computation can be found in the supplementary material .
the final expression of the canonical state of the system is @xmath96 defining the `` system '' .
the `` environment '' is identified with the complementary 2d - surface associated to the set of links @xmath97 , with @xmath98.,width=2 ] where @xmath99 means that we are summing over the configurations of the spins of the system @xmath100 such that @xmath101 .
moreover the definition of the @xmath102functions is @xmath103 we also define the following short - hand notation @xmath104
. we will also call @xmath105 the dimension of the system s degeneracy space with fixed area @xmath106 and closure defect @xmath107 , derived from the equivalent @xmath65 representation as for the case of the environment in eq .
( [ canoni ] ) .
the canonical weight @xmath108 encodes all the information about the local structure of correlations of the reduced intertwiner state .
the specific form of this factor tells us about the physics of the system , defined by the specific choice of constraints : the su(2 ) gauge symmetry and the fixed total area constraint .
given the global constraint , the split in system and environment breaks the gauge symmetry . due to the presence of the constraint , onto @xmath7 the quantum numbers of system @xmath109
are intertwined with those of the environment @xmath110 .
this is why , beside the expected dependence on the total area of the system @xmath106 , the canonical weight carries some interesting extra information on the local closure defect @xmath107 .
in this section we study the region of the space of the parameters @xmath111 where the canonical reduced state is typical . in other words ,
we investigate the distance of the canonical state from a randomly chosen pure state in @xmath7 . concretely , following the approach described in section [ tysn ] , we want to show that for the overwhelming majority of intertwiner states @xmath112 , the trace distance @xmath113 between the reduced density matrix of the system @xmath114 and the canonical state @xmath14 is extremely small is the probability two tell apart @xmath115 and @xmath116 , by means of the most effective quantum measurement@xcite ] .
this amounts to prove two things : first , that the hilbert space average of such trace distance is itself quite small in the regime in which we are interested in @xmath117 \ll 1 \,\,,\end{aligned}\ ] ] where @xmath118 indicates the hilbert space average performed using the unique unitarily invariant haar measure @xcite .
second , that the fraction of states for which such distance is higher than a certain @xmath119 is exponentially vanishing in the dimension of the hilbert space
. + now , following @xcite , one can recast the condition in eq .
( [ result ] ) with the following bound on the average distance , @xmath120 \leq \sqrt{\frac{d_s}{d_e^{\mathrm{eff } } } } \leq \frac{d_s}{\sqrt{d_{\mathcal{r}}}}\end{aligned}\ ] ] concretely , the first step toward the statement of typicality in our context amounts to study in which region of the parameters space @xmath121 we have @xmath122 .
the hilbert space of the system is the tensor product hilbert space of the set of irreps @xmath52 with a given cutoff @xmath123 .
we assume @xmath124 , in order to be sure that @xmath125 will always carry an irreducible representation of @xmath65 .
each @xmath126 has dimension @xmath127 .
therefore , considering the set of @xmath63 edges comprising the system , we have @xmath128 analogously , for the environment we get @xmath129 . since @xmath31 is given in eq .
( [ dime ] ) , we can focus on the last inequality in eq .
( [ bbound ] ) and define the regime where @xmath130 \ll 1 $ ] . studying the ratio @xmath131 we can see that @xmath1 and @xmath72 play a rather symmetric role in making this quantity small .
the region of interest is certainly @xmath132 or @xmath133 , or both .
as we will argue in the next section , @xmath134 is precisely the regime of interest for the thermodynamical limit .
therefore we focus on this region , where there are two different regimes : @xmath135 or @xmath136 . in both cases there
are wide regions of the parameters space where the inequality @xmath130 \ll 1 $ ] holds .
we were able to extract the following two conditions which guarantee an exponential decay of eq .
( [ ratio ] ) , either on @xmath1 or on @xmath72 : @xmath137 the details can be found in the supplementary material but we would like to present a physically motivated argument to provide a meaningful value for the cut - off @xmath123 and check the plausibility of the given bounds . as argued in @xcite , if we look at a sphere with small radius @xmath138 , placed at a large distance @xmath139 , we will see it within a small angle @xmath140 . therefore using the scale of the radius of the observed universe @xmath141 and assuming that there is nothing with size smaller than the planck length @xmath75 , we will never see something with angular extension smaller than @xmath142 .
+ a spherical harmonics of representation @xmath143 is able to discriminate angular distances of the order @xmath144
. therefore the existence of @xmath145 means that there is an upper bound to the representation which we need to consider which is @xmath146 . using this argument
we obtain the following cut - off @xmath147 putting the numbers in eq .
( [ cut1 ] ) and eq .
( [ cut2 ] ) we obtain @xmath148 following @xcite , we can use levy s lemma ( see appendix [ levy ] ) to bound the fraction of the volume of states which are @xmath149 more distant than @xmath150 from @xmath14 as @xmath151 } { \mathrm{vol } \left[|\mathcal{i}\rangle \in \mathcal{h}_{\mathcal{r } } \right ] } \leq b_\epsilon(d_{\mathcal{r}})\\ \nonumber & b_{\epsilon } ( d_{\mathcal{r}})\equiv 4 \ , \mathrm{exp } \left[-\frac{2}{9\pi^3 } d_{\mathcal{r } } \varepsilon^2 \right].\end{aligned}\ ] ] the dimension @xmath31 can be evaluated numerically because we have an exact expression .
we give a numeric example to show that it is not necessary to have huge areas or number of links for the typicality to emerge .
suppose we have a huge sensitivity on the trace distance : @xmath152 .
moreover , @xmath153 . with these numbers we have @xmath154.\end{aligned}\ ] ] suppose we look at the most elementary patch , just a few links ( @xmath155 ) .
the set of numbers @xmath156 gives the following bounds , using a cut - off given by the cosmological horizon @xmath157 \ll 1\end{aligned}\ ] ] as we can see , the typicality emerges quite easily , due to the exponential - like growth of the constrained hilbert space on the number of links @xmath1 and on the total area @xmath72 .
+ the existence of a typical behaviour indicates the emergence of a regime where the properties of the reduced state of the @xmath1-valent intertwiner state are _
universal_. the structure of _ local _ correlations carried by the reduced state is independent from the specific shape of the pure intertwiner state and it is locally the same everywhere . due to the global symmetry constraint though , the canonical weight presents a very involved analytic form , despite the extreme simplicity of the system under study . in order to extract some physical information from this coefficient we are going to study its behaviour in the thermodynamic limit
in this section we investigate the behaviour of the entropy of the reduced state , in the thermodynamic limit . in the standard context of statistical mechanics ,
when performing the thermodynamic limit the density of particles must be finite otherwise the energy density would diverge : @xmath158 with @xmath159 . as we will see in the forthcoming argument ,
the area is playing here the role of the energy , therefore we think that the correct way of performing the thermodynamic limit consists in taking @xmath160 with @xmath161 , where @xmath162 is the average spin of the intertwiner . +
the entropy of the system is given by the von neumann entropy , @xmath163\end{aligned}\ ] ] given the diagonal form of the canonical reduced density matrix @xmath14 in eq .
( [ canonico ] ) , this can be written as @xmath164 within the typicality regime ( @xmath165 ) we can use the stirling approximation for the factorials , to simplify the form of the binomial coefficients in @xmath166 and @xmath31 . we will study separately the three regimes @xmath167 , @xmath168 and @xmath169 .
the details of the computation can be found in the supplementary material , here we only summarise the results . in the case of small average spin , the leading term in the thermodynamic limit
is @xmath170 where @xmath171 is the quantum mechanical average , on the canonical state @xmath14 , while @xmath172 is formally identified as the `` temperature '' of the environment .
it turns out to be a function of the averaged spin of the environment . despite being quite far from the standard setting ,
a hint toward a thermodynamical interpretation of this result comes from the @xmath65 description of the @xmath0 intertwiner space . using the schwinger representation of the @xmath173
lie algebra @xcite , one can describe the @xmath1-valent intertwiner state as a set of @xmath174 oscillators , @xmath175 .
the quadratic operators @xmath176 acts on couples of punctures @xmath177 and form a closed @xmath178 lie algebra .
the @xmath179 casimir operator is given by the oscillators energy operator @xmath180 , with @xmath181 , and its value on a state gives twice the sum of the spins on all legs , @xmath182 .
therefore , one can interpret @xmath5 as measuring ( twice ) the total area of the boundary surface around the intertwiner .
+ in statistical mechanics , the thermal behaviour of the canonical state relies on the constraint of energy conservation .
the emergence of the canonical state from the micro - canonical occurs as the degeneracy of the environment grows exponentially with the energy , hence decreasing exponentially with the system energy . in these terms
, constraining the total area is equivalent to fix a shell of eigenvalues ( in fact a single eigenvalue ) of the energy operator acting on the full system . in the limit @xmath183
, the degeneracy of the single energy level grows exponentially .
for such a reason the area scaling described by eq .
( [ area ] ) is consistent with a thermal interpretation for our reduced surface state .
it is also worth to mention that the departure from the exact thermal behaviour , la gibbs , is a signature of the breaking of the global su(2 ) symmetry ( closure defect ) , witnessed by the explicit dependence of the reduced state on @xmath88 .
here we study the behaviour of the entropy in the regime @xmath184 . up to @xmath185 the logarithm of the normalised canonical weight is given by @xmath186 interestingly , the leading term does not depend on the quantum numbers of the system
. therefore the entropy is counting the number of orthogonal states on which the canonical state has non - zero support @xmath187 this makes the entropy extensive in the number of edges comprising the dual surface of the system . in this sense
, the term @xmath188^k$ ] defines some kind of _ effective _ dimension of the system , suggesting that the following two things happen in such regime : first , the canonical state has approximately a tensor product structure ; second that the total spin is equally distributed among all spins in the universe therefore the accessible hilbert space of each spins is roughly limited by a representation of the order of @xmath162 .
the validity of this interpretation can be checked assuming a tensor product structure of @xmath63 links with single - link hilbert space limited to the representation @xmath189 and computing the entropy @xmath190 as the logarithm of the dimension of this space .
if we can find an @xmath191 such that the difference @xmath192 is proportional only to small corrections @xmath193 , we can say that our argument is not too far from what is happening in such a regime . with these assumptions
the effective dimension of the hilbert space of the system is @xmath194 in the @xmath167 regime we can write it as @xmath195 which gives @xmath196 the difference between the two entropies @xmath197 is given only by small corrections of order @xmath198 when @xmath199 .
this simple computation provides evidence that the result in eq.([eq : ent2 ] ) follows from the two aforementioned assumptions . eventually , we compute the behaviour of the entropy in the intermediate regime @xmath201 . with respect to the previous cases ,
this regime does not add anything new to the analysis .
the observed behaviour is extensive in the number of links of the system , with a coefficient which is slightly different from the previous one : @xmath202)\end{aligned}\ ] ] the relevant computation can be found in the supplementary material .
in this manuscript we extend the so - called typicality approach , originally formulated in statistical mechanics contexts , to a specific class of tensor network states given by @xmath0 invariant spin networks . in particular , following the approach given in @xcite ,
we investigate the notion of canonical typicality for a simple class of spin network states given by @xmath1-valent intertwiner graphs with fixed total area .
our results do not depend on the physical interpretation of the spin - network , however they are mainly motivated by the fact that spin networks provide a gauge - invariant basis for the kinematical hilbert space of several background independent quantum gravity approaches , including loop quantum gravity , spin - foam gravity and group field theories .
the first result is the very existence of a regime in which we show the emergence of a canonical typical state , of which we give the explicit form .
geometrically , such a reduced state describes a patch of the surface comprising the volume dual to the intertwiner .
the structure of correlations described by the state should tell us how local patches glue together to form a closed connected surface in the quantum regime .
we find that , within the typicality regime , the canonical state tends to an exponential of the total spin of the subsystem with an interesting departure from the gibbs state .
the exponential decay la gibbs of the reduced state is perturbed by a parametric dependence on the norm of the total angular momentum vector of the subsystem ( closure defect ) .
such a feature provides a signature of the non local correlations enforced by the global gauge symmetry constraint .
this is our second result .
we study some interesting properties of the typical state within two complementary regimes , @xmath203 and @xmath204 . in both cases , we find that the area - law for the entropy of a surface patch must be satisfied at the local level , up to sub - leading logarithmic corrections due to the unavoidable dependence of the state from the closure defect . however , the area scaling interpretation of the entropy in the two regimes is quite different . in the @xmath203 regime ,
the result is related to the definition of a generalised gibbs equilibrium state .
the area is playing the role of the energy , as imposed by the specific choice of the global constraint , requiring total area conservation . on the other hand , in the @xmath205 regime
, the area scaling is given by the extensivity of the entropy in the number of links comprising the reduced state , as for the case of the generalised ( non @xmath0-gauge invariant ) spin networks @xcite .
in this regimen , each link contributes independently to the result , indicating that the global constraints are very little affecting the local structure of correlations of the spin network state .
still , interestingly , the remainder of the presence of the constraints can be read in the definition of what looks like an effective dimension for the single link hilbert space .
we interpret these results as the proof that , within the typicality regime , there are certain ( local ) properties of quantum geometry which are `` universal '' , namely independent of the specific form of the global pure spin network state and descending directly from the physical definition of the system encoded in the choice of the global constraints .
we would like to stress that our result is purely kinematic , being a statistical analysis on the hilbert space of spin - network states .
for the case of a simple intertwiner state , such study necessarily requires to consider a system with a _ large number _ of edges , beyond the very large dimensionality of the hilbert space of the single constituents . in this sense ,
the presented statistical analysis and thermal interpretation is very different from what recently done in @xcite , considering quantum geometry states characterised by few constituents with a high dimensional hilbert space .
in fact , we expect a large number statistical analysis to play a prominent role in facing the problem of the continuum in quantum gravity .
therefore we think it is important to propose and develop new technical tools which are able to deal with a large numbers of elementary constituents and extract physically interesting behaviours .
+ the kinematic nature of the statement of typicality , together with its general formulation in terms of constrained hilbert spaces given in @xcite , provide an important tool to study the possibility of a thermal characterisation of reduced states of quantum geometry , regardless of any hamiltonian evolution in time . beyond the simple case considered in the paper and in a more general perspective
, we expect typicality to be useful to understand how large the effective hilbert space of the theory can be , given the complete set of constraints defining it .
it will also help in understanding which typical features we should expect to characterise a state in such space .
if we think of dynamics as a flow on the constrained hilbert space , we generally expect that , even if the initial state is highly un - typical after a certain `` time''-scale we will find the system in a state which is extremely close to the typical state .
this happens because , as it has been shown in the original paper on typicality , the number of states close to the typical state are the overwhelming majority . + finally , it is interesting to look at the proposed `` generalised '' thermal characterisation of a local surface patch , within the standard lqg description of the horizon , as a closed surface made of patches of quantized area . differently from the _ isolated horizon _
analysis ( see e.g. @xcite ) , in our description the thermal character of the local patch is not ( semi)classically induced by the thermal properties of a black hole horizon geometry , but emerges from a purely quantum description . in this sense , our picture goes along with the informational theoretic characterisation of the horizon proposed in @xcite .
in fact , we think that typicality could be used to define an information theoretic notion of quantum horizon , as the boundary of a generic region of the quantum space with an emergent thermal behaviour .
we leave this for future work .
the authors are grateful to daniele oriti , aldo riello and thibaut josset for interesting discussions and careful readings of the draft of the paper .
in order to better understand the result it is useful to look at its most important step , which is the so - called levy - lemma . take an hypersphere in @xmath206 dimensions @xmath207 , with surface area @xmath208 .
any function @xmath209 of the point which does not vary too much will have the property that its value on a randomly chosen point @xmath211 will approximately be close to the mean value .
@xmath212}{\mathrm{vol}\left [ \phi \in s^d \right ] } \leq 4 \ ,
\mathrm{exp } \left [ - \frac{d+1}{9 \pi^3 } \epsilon^2 \right]\end{aligned}\ ] ] where @xmath213 $ ] stands for `` the volume of states @xmath211 such that @xmath214 '' .
@xmath215 is the average of the function @xmath209 over the whole hilbert space and @xmath216 $ ] is the total volume of the hilbert space .
integrals over the hilbert space are performed using the unique unitarily invariant haar measure . + the levy lemma is essentially needed to conclude that all but an exponentially small fraction of all states are quite close to the canonical state .
this is a very specific manifestation of a general phenomenon called `` concentration of measure '' , which occurs in high - dimensional statistical spaces @xcite .
the effect of such result is that we can re - think about the `` a priori equal probability '' principle as an `` apparently equal probability '' stating that : as far as a small system is concerned almost every state of the universe seems similar to its average state , which is the maximally mixed state @xmath217 .
+ j. c. baez , _ spin networks in gauge theory _ , advances in mathematics 117 253 - 272 ( 1996 ) ; j. c. baez , _ spin networks in nonperturbative quantum gravity _ , in the interface of knots and physics , louis kauffman , ed .
( american mathematical society , providence , rhode island , 1996 ) . | in this work we extend the so - called typicality approach , originally formulated in statistical mechanics contexts , to @xmath0 invariant spin network states .
our results do not depend on the physical interpretation of the spin - network , however they are mainly motivated by the fact that spin - network states can describe states of quantum geometry , providing a gauge - invariant basis for the kinematical hilbert space of several background independent approaches to quantum gravity .
the first result is , by itself , the existence of a regime in which we show the emergence of a typical state .
we interpret this as the prove that , in that regime there are certain ( local ) properties of quantum geometry which are `` universal ''
. such set of properties is heralded by the typical state , of which we give the explicit form .
this is our second result . in the end
, we study some interesting properties of the typical state , proving that the area - law for the entropy of a surface must be satisfied at the local level , up to logarithmic corrections which we are able to bound . |
superconducting devices based on the josephson effect are an important testbed for investigating deep aspects of quantum mechanics such as macroscopic quantum phenomena @xcite-@xcite and circuit quantum electrodynamics ( cqed ) @xcite-@xcite .
such devices are moreover promising candidates for the practical implementation of solid state quantum computing @xcite-@xcite .
this thanks to the possibility to arrange superconducting circuits in a desired way with great flexibility @xcite-@xcite and also because their behavior can be analyzed by means of equivalent mechanical models @xcite , describing the motion of fictitious particles moving in an effective potential , with a supposed quantum behavior at low temperature .
for example , it is possible to realize josephson anharmonic oscillators that can be used as artificial atoms , which can be manipulated by microwaves with nmr - like techniques @xcite , and which can also be coupled with superconducting resonators for single photon experiments in cavities @xcite , @xcite .
moreover , it is often possible to control and modify the effective potential shape in time with a fast and accurate timing .
this allows , for example , the observation of very fast coherent oscillations ( up to 20 ghz ) of the magnetic flux states in a squid ( superconducting quantum interference device ) qubit @xcite , @xcite , obtained just by quickly and strongly modifying the effective potential shape ( from a symmetric double - well to a single - well , and back to the double - well ) .
this is done by simply applying flux pulses , in the absence of microwaves . for this kind of manipulation
it is of great importance the rapidity of the modification of the effective potential shape . for example , if we consider quantum computing applications , the manipulation must be fast enough in order to induce non adiabatic landau - zener transitions between the first two energy levels ( used as computational space ) , but also slow enough in order to avoid transitions to upper levels ( non computational space ) .
fortunately generally speaking this is possible thanks to an appropriate energy gap existing between the first couple of levels and the upper ones , but the transition rate is an aspect that must be accurately considered and calibrated @xcite . in this work we investigate experimentally and theoretically the effect of the speed of modification of the potential shape in a double squid flux qubit , presenting the experimental observation and the theoretical analysis of an interesting quantum effect due to the interplay of landau - zener transitions and resonant tunnelling .
a similar phenomenon has been studied in ref.@xcite , @xcite , even if in a different system and contest .
the device we consider is the so called double squid @xcite , consisting of a superconducting loop of inductance @xmath0 interrupted by a dc squid , a second smaller superconducting loop of inductance @xmath1 interrupted by two identical josephson junctions , each of ( nominally ) identical critical current @xmath2 and capacitance @xmath3 ( fig . [
figschema]a ) .
the dc squid behaves approximately like a single junction of capacitance @xmath4 and tunable critical current @xmath5 ( where @xmath6 being @xmath7 the flux quantum ) , which is controlled by a magnetic flux @xmath8 applied to the small loop ( this approximation holds if the loop is small enough , i.e for @xmath9 ) .
note that @xmath10 can also be negative , and in this case the dc squid behaves as a pi - junction .
the double squid behavior can be controlled by two distinct magnetic fluxes , one applied to the large loop ( @xmath11 ) and the second to the small one ( @xmath8 , previous mentioned ) .
the squid dynamics can be described by an equivalent mechanical model , with effective mass @xmath12 , effective position corresponding to the total magnetic flux threading the large loop ( @xmath13 ) , and potential @xmath14 this effective potential can have one or two distinct wells , according to the adimensional parameter @xmath15 : in the particular case @xmath16 there will be a single - well for @xmath17 ( approximately an harmonic potential with a characteristic frequency controlled by @xmath8 , fig .
[ figschema]b ) , and two distinct wells separated by a barrier for @xmath18 ( with barrier height controlled by @xmath8 , fig .
[ figschema]c ) . the flux @xmath11 controls the potential symmetry : for @xmath16 the potential is symmetric , otherwise it is tilted ( fig .
[ figschema]c ) .
in order to study the effect of speed versus adiabaticity on the system , we concentrate our attention on a fast and large modification of the potential shape , from the single - well to the double - well case .
this is just an half of the complete manipulation of the qubit state presented in ref .
initially the system is maintained in the singlewell case for a rest time @xmath19 , waiting for the complete relaxation to the ground state .
then it is moved rapidly to the double - well case , where an high barrier separates the two minima , and this is obtained by changing the control flux @xmath8 with a characteristic sweeping rate @xmath20 .
finally it is done a read out of the squid flux state , corresponding to observe which of the two minima is occupied at the end .
this is performed by an inductively coupled readout squid used as a magnetometer by means of measurements of the switching current @xcite .
the sequence is repeated many times in order to estimate the occupation probability @xmath21 of the final flux ( for example , the probability to obtain a final right flux state ) .
the complete operation is repeated for different unbalancing fluxes @xmath11 . for slow ( adiabatic ) modifications
we expect that the system remains always in its ground state : the left flux state when the left minima is the lower one ( for @xmath22 ) , and the right flux state in the opposite case ( for @xmath23 ) , with a sweet transition between these opposite cases around the symmetry point ( @xmath24 ) . in this case the probability @xmath21 as a function of the unbalancing flux @xmath11 presents a sigmoidal shape . by increasing the sweeping rate @xmath25 we expect an excitation of upper levels due to non adiabatic transitions , with a possible emerging of effects related to this population .
we performed the measurements on devices realized by standard trilayer nb / alox / nb technology , with nominal parameters @xmath26 , @xmath27 , @xmath28 and @xmath29 , in a dilution refrigerator with base temperature @xmath30 arranged for ultra low noise qubit measurements ( mu - metal , superconducting and normal metal shields , thermocoax and l - c - l filters on dc lines , different attenuator stages on the signal line ) . a preliminary study of the switching current in the readout dc squid gives an escape temperature of about 250 mk , compatible with the crossover temperature within the experimental errors .
this indicates the absence of an excess temperature due to noise @xcite .
the probability @xmath21 is evaluated by repeating the preparation - modification - readout cycles for 1000 times at a rate of @xmath31 .
the initial preparation is obtained by waiting for a time @xmath32 in the single - well condition ( fig .
[ figlivelli]a ) , for @xmath33 . in this condition
the system is well approximated by an harmonic oscillator with characteristic frequency @xmath34 , corresponding to a level spacing of about @xmath35 , very high with respect to the thermal bath temperature , so that we expect a negligible thermal excitation .
m@xmath36 ) .
( b ) energy potential of the system and relative eigenstates in the double - well case ( for @xmath37 m@xmath36 ) , with a slight asymmetry ensuring the degeneracy ( for @xmath38 m@xmath39 ) .
( c ) variation of the energy levels positions for different fluxes @xmath8 ( for @xmath40 ) .
note that for convenience all energies are expressed as frequencies in ghz , and are shifted by subtracting the ground state energy.,width=302 ] the potential shape modification is driven by a fast pulse generator , presenting signals with a typical rise time @xmath41 that can be changed by using an home made tunable l - c - l filter .
the modified pulse is fully characterized thanks to a fast oscilloscope , in particular it is possible to check the pulse shape and the actual rise time .
the fast signal is transmitted to the device thanks to a @xmath42 matched coaxial cable interrupted by three 20db attenuators placed at @xmath43 , @xmath44 and @xmath45 stages .
more details on the setup can be found in ref .
we tested the entire line at room temperature and in the absence of the chip , while the present setup does not allow to test the entire system ( line plus chip ) at low temperature .
for this reason we can expect a large error in the determination of the real signal shape and rise time at the device level .
the applied signal modifies the potential from the single - well condition , at @xmath46 , to the double - well case , at @xmath47 , passing through the critical condition @xmath48 where there is the transition between the single - well and double - well conditions ( fig .
[ figlivelli ] ) . at the end of each cycle
the final flux state is measured by the coupled readout dc squid .
this is done by applying a current ramp to the squid and recording the switching current , which is directly related to the qubit flux .
the sequence is repeated for different unbalancing fluxes @xmath11 , ranging from @xmath49 to @xmath50 , obtaining the probability curves plotted in figure [ figmisure2d ] .
these curves are obtained for three different rise times , @xmath51 , @xmath52 and @xmath53 . to measure a right flux state for different unbalancings @xmath11 at three different rise times @xmath54 .
resonance peaks are visible for faster transitions ( arrows in second and third plots),width=264 ] in the top plot we observe the sigmoidal function expected for a slow rate . in the middle and lower plots two distinct order of peaks appear , respectively at about @xmath55 and @xmath56 .
the measurement can be repeated for different rise times obtaining the 3-dimensional curve showed in fig .
[ figmisure3d ] . to measure a right flux state for different unbalancings @xmath11 at different rise times
@xmath54 , plotted as a 3d surface.,width=302 ] from fig .
[ figmisure3d ] we can note some enlightening characteristics .
first of all the position of peaks ( in @xmath11 ) corresponds to the conditions for which different levels in the two wells are aligned ( degenerate ) ( fig . [
figlivelli]b ) .
this strongly suggest that the presence of peaks is a manifestation of resonant tunneling between wells .
secondly we note that the appearing of peaks requires rise times below a particular critical value , namely it is necessary to have an high enough sweep - rate in order to observe peaks .
thirdly , there is an alternation of peaks orders : when the second order of peaks appears the first order disappears . in fig .
[ figlivelli]c it is plotted the modification of the first nine energy levels in the passage from the single - well to the double - well condition ( in the degenerate case for @xmath57 ) .
this figure can help us in a qualitative explanation of the observed peaks . in the single - well condition ( on the left )
it appears reasonable to suppose that only the ground state is populated .
close to the critical point @xmath58 , where the barrier appears to separate two distinct wells , landau - zener transitions populate the upper levels , with an efficiency depending on the sweep - rate @xmath59 .
these excited states can cross the barrier thanks to a resonant tunneling when the alignment condition is meet .
the experimental results thus suggest that these two effects combine together and produce the observed peaks , due to an excess of population in the upper well when the resonant and the non - adiabatic conditions are both fulfilled .
the region where this effect is active is small , of the order of @xmath60 of the entire span of the flux @xmath8 ( @xmath61 ) , and this region is crossed in a similar fraction of the entire rise time duration , about @xmath62 .
we stress again that the combination of landau - zener and resonant tunneling can explain the first two observations ( position of peaks and appearing of them only below a critical rise time ) , but not the third one , that is the alternation of peaks orders .
in order to gain information about possible physical mechanisms and/or properties of the system which may be responsible for the appearance of the alternation of peaks order as in fig . [ figmisure3d ] , in what follows we develop a more quantitative analysis exploiting a simple quantum model useful to describe the system under scrutiny . to do this we start considering the hamiltonian model relative to the potential ( [ h ] ) , rewritten in a more convenient way : where @xmath64 , @xmath65 and @xmath66 .
moreover @xmath67 .
the hamiltonian ( [ mainhamilt ] ) is well suited to describe the time evolution of a particle in one dimensional time - dependent potential .
in particular , appropriately choosing the function @xmath68 , it is possible to vary the potential shape going , in an interval of time @xmath69 from a single to a double - well , thus reproducing the initial and final conditions of the experiment before discussed .
the problem therefore consists in finding , as function of both the unbalancing parameter @xmath70 as well as of the rise time @xmath54 , the probability @xmath21 that , at the end of the process , the particle is found in the right well .
let s observe that knowing the state of the system @xmath71 at the time instant @xmath69 , this probability can be simply evaluated as let s indicate by @xmath73 a set of instantaneous eigenfunctions of the hamiltonian ( [ mainhamilt ] ) : @xmath74 exploiting these states , we can write @xmath75 \cdot s_n(t ) \cdot \ket{\psi_{n}(t)}.\ ] ] where the function @xmath76 are solutions of the following set of integro - differential equations : @xmath77s_k(t).\ ] ] with @xmath78 .
starting from eqs .
( [ pl ] ) and ( [ psi_t ] ) , the probability @xmath21 can be then written as an exact analytical resolution of the coupled integro - differential equations ( [ main diff eq ] ) is not easy .
thus we proceed further by performing numerical simulations of the dynamical behavior of the system carefully taking into account both the non adiabadicity in the system dynamics and the possible emergence of resonant tunneling processes . as first step , considering @xmath68 as a parameter , we numerically diagonalize the hamiltonian ( [ mainhamilt ] ) at a generic time instant @xmath81 , finding its instantaneous eigenvectors @xmath73 and the correspondent eigenvalues @xmath82 . considering values of @xmath54 of interest in the context of this paper , we have evaluated the quantity @xmath83 , defined in eq .
( [ ln ] ) , in correspondence to different values of @xmath84 verifyng , as expected , that , at least for not too large @xmath84 , @xmath83 is almost equal to one or negligible witnessing that the first eigenstates are practically localized for @xmath85 . in our simulation
however we use the numerical value of @xmath83 instead of 0 or 1 . to evaluate the probability @xmath21 we thus need to calculate the populations @xmath87 by numerically solving the system ( [ main diff eq ] ) explicitly giving the way in which the potential shape modifies itself going from the initial condition to the final one during the time @xmath69 . in other words
we now have to choose the function @xmath68 appearing in eq .
( [ mainhamilt ] ) .
we wish to underline that this is a very delicate point .
it is undoubted indeed that the dynamics of the system will be deeply affected from the way of varying the potential shape .
thus we expect to find different results in correspondence to different choice of @xmath68 .
a sigmoidal function allows a simple and reasonable description , at least from a qualitative point of view because of experimental uncertainties on the exact @xmath68 shape as discussed in section iii .
thus we fix it as @xmath88 with @xmath89 , @xmath90 , choosing in particular @xmath91 and @xmath92 .
it is the case to stress that varying these two parameters implies , as a consequence , a changing in the rise time @xmath54 . in order to investigate on the appearance of peaks in fig .
[ figmisure3d ] , we have calculated the probability of finding the particle in the right well for different values of @xmath93 supposing , as it appears physically reasonable , that at @xmath86 the system is in its ground state .
in particular we evaluated such a probability considering a range of @xmath11 in correspondence of which peaks appear , as suggested by the experimental data .
considering values of @xmath54 like in figure [ figmisure3d ] , our simulation does not evidence the existence of significant peaks .
however , taking into account the fact that the peaks in the probability @xmath21 arise reducing the rise time and in view of the experimental uncertainties before discussed , we have simulated the behavior of the system exploring smaller @xmath54 .
the results obtained are reported in figure ( [ ground ] ) where the probability @xmath21 is plotted as function of both @xmath11 and @xmath54 . to find the system in the right well against the unbalancing @xmath11 and rise time @xmath54 , supposing that at @xmath86 it is in the linear superposition @xmath94 of its ground and first excited states in correspondence to @xmath95 and @xmath96.,width=321 ] as expected , peaks of resonance appear in correspondence to different values of @xmath11 .
however the position of such peaks with respect to the rise time @xmath54 does not reflect , not only quantitatively but also qualitatively , the experimental observations .
in other words even if a dependence of @xmath21 on @xmath54 is evident , the function @xmath97 is very different from the experimental one .
in particular the probability @xmath21 obtained by simulation ( see fig . [ ground ] ) is not characterized by the alternation of peaks order as in figure [ figmisure3d ] .
such result moreover does not seem to be imputable to the particular choice of the @xmath68 we made .
we have indeed verified that this is the case by choosing a linear function and obtaining the same qualitative behavior of that shown in fig .
[ ground ] . if it is true that the manner in which we modify the potential shape deeply affects the dynamics of the system and thus the probability @xmath21 that the system is found in the right well at the end of the potential modification , another key ingredients to be considered is surely the state of the system at @xmath86 . taking into account the fact that the temperature at which the experiment is performed is @xmath98 , it is reasonable to suppose that at @xmath86 there is a small , but not zero ( of the order of few percent ) , probability that the system is in its first excited state .
thus could be reasonable to assume that the preparation step leaves the system in a mixture @xmath99 of the ground and the first excited state .
performing simulation starting from this mixture instead of the ground state , we have verified that , considering values of @xmath100 compatible with @xmath101 , we do not get significant differences with respect to the results displayed in fig .
[ ground ] .
we have also checked that increasing the relative weight _ x _ of the first excited state in the initial mixture worsens the accordance between theoretical predictions and experimental results .
the theoretical prediction instead drastically changes if we suppose that quantum coherences are present in the initial state of the system .
such an assumption can be justified by considering the fact that the waiting interval of time @xmath19 was not long enough to allow the complete destruction of the coherences between the ground and the first excited state of the double squid .
if this is the case it is reasonable to assume that at @xmath86 the system is in a quantum superposition @xmath94 of the first two low - lying states , instead of a mixture of the two states as before supposed . starting from this initial state the probability @xmath21 shows a dependence on both @xmath11 and @xmath54 as displayed in fig .
( [ combinazione1 ] ) where we have considered a smaller range of @xmath54 to better appreciate the behavior of @xmath21 . as expected , also in view of experimental uncertainties on the @xmath68 function as well as on the parameters defining the system , figure ( [ combinazione1 ] ) does not exactly match the experimental results presented before , even if the qualitative behavior of @xmath21 seems to be well reproduced .
more in detail the most important aspect of the results shown in fig .
( [ combinazione1 ] ) consists in the fact that , as experimentally observed , there is an alternation of the peaks order determined by both the asymmetry in the potential governed by the value of the unbalancing parameter @xmath11 , and on the rise time @xmath54 required to go from a single to a double - well .
the theoretical analysis developed in this paper has the merit to disclose the role played by the persistence of quantum coherences in the initial state of the double squid .
we wish to stress indeed that starting from an initial state as the ground state of the qubit or a mixture of the same ground state and the first excited one , even if leading to the appearance of peaks , is completely unable to predict the alternance of minima and maxima as requested by the experimental results .
thus our assumption , that is the persistence of quantum coherences , leads to predictions in good qualitative agreement with the experimental results .
the intriguing point is that all the alternative seemingly more reasonable assumptions concerning the initial state of the squid predict a behavior not compatible with some aspect of the observed one . before concluding we wish to spend some words about possible decoherence effects in the dynamical behavior of the system .
it is important to underline that , as we have previously discussed , the temporal interval where the physical mechanisms at the basis of the observed effects , are active is a small fraction ( of the order of 0.1 ns ) of the total duration of the experiment ( about 1 ns ) . as first approximation
it is thus reasonable to neglect decoherence effects in the system dynamics , which present time scales of the order of nanoseconds .
anyway it is the case to underline that in this context the source of noise is the @xmath102 noise @xcite that generally speaking acts modifying the effective control parameter as for example @xmath11 @xcite-@xcite .
we thus expect that the effects of such a noise on the results reported in figure [ combinazione1 ] would consist at most in a broadening of the observed peaks . summarizing
, we have experimentally verified that a fast modification of the potential shape of a double squid gives rise to the activation of quantum resonance phenomena manifesting themselves as peaks in the probability of measuring a right flux state of the squid at the end of the non adiabatic transition .
the theoretical analysis we have performed seems moreover to lead to the conclusion that the characteristic behavior of such a probability is also a direct consequence of the presence of quantum coherences in the initial state of the system .
this fact suggests to perform other experiments on the system under scrutiny aimed at revealing quantum interference effects in the behavior of some no - diagonal physical observables . | by quickly modifying the shape of the effective potential of a double squid flux qubit from a single - well to a double - well condition , we experimentally observe an anomalous behavior , namely an alternance of resonance peaks , in the probability to find the qubit in a given flux state .
the occurrence of landau - zener transitions as well as resonant tunneling between degenerate levels in the two wells may be invoked to partially justify the experimental results .
a quantum simulation of the time evolution of the system indeed suggests that the observed anomalous behavior can be imputable to quantum coherence effects .
the interplay among all these mechanisms has a practical implication for quantum computing purposes , giving a direct measurement of the limits on the sweeping rates possible for a correct manipulation of the qubit state by means of fast flux pulses , avoiding transitions to non - computational states . |
a distinguishing characteristic of type ia supernovae ( sne ) is the large amount of iron that they are thought to produce mainly through the decay of radioactive ni@xmath3 .
the nuclear decay of this material powers the sn light curve , thereby providing a constraint on the quantity of material ejected in individual sn events .
although variation in the amount and distribution of radioactive ejecta likely accounts for the modest range of peak absolute magnitudes seen in type ia sne , by and large sne ia are strikingly homogeneous in their properties . various observational constraints ( e.g. , the absence of hydrogen in optical spectra near optical maximum , occurrence in older stellar populations , and the relatively large amount of ni@xmath3 produced ) have led to the current framework involving carbon deflagration / detonation in a white dwarf driven to the chandrasekhar limit by accretion .
the most likely progenitor system ( branch et al .
1995 ) is a c - o white dwarf accreting h / he - rich gas from a companion , either from its wind or through roche lobe overflow .
remnants of sn ia should contain about one chandrasekhar mass of ejecta comprised of 0.60.8 @xmath2 of iron with intermediate - mass elements ( o , mg , si , s , ca ) distributed in the outer layers ( e.g. , iwamoto et al .
current models predict that sne ia should leave no compact remnant .
one of the goals of supernova remnant ( snr ) research has been to use remnant properties to infer the nature of the progenitor .
such properties as the partially neutral nature of the surrounding ambient medium ( tuohy et al . 1982 ) or the fe - rich composition of the ejecta ( hughes et al .
1995 ) have been used to argue for a sn ia origin for some snrs . in this article
we investigate the large magellanic cloud ( lmc ) snr 71 ( [email protected] ) which exhibits both of these features .
71 ( davies , elliott , & meaburn 1976 ) was first proposed as a supernova remnant based on the x - ray survey of the lmc ( long , helfand , & grabelsky 1981 ) .
follow - up optical spectroscopy confirmed this identification and revealed the remnant to be balmer - dominated ( tuohy et al .
1982 ; smith et al . 1991 ) , that is to say , one whose optical spectrum is dominated by hydrogen emission with little or no emission from forbidden lines of [ ] or [ ] . the remnants of tycho s sn and sn1006 are galactic examples of such snrs ( kirshner & chevalier 1978 ; schweizer & lasker 1978 ) .
x - ray studies of 71 ( hughes , hayashi , & koyama 1998 ) yielded an age of @xmath15000 yr and evidence for the presence of sn ejecta in the form of an enhanced global abundance of fe . here
we present results from new observations of the remnant that bear closely on the nature of the originating sn .
we observed 71 using the back - side - illuminated chip ( s3 ) of the advanced ccd imaging spectrometer ( acis - s ) instrument in full - frame timed exposure mode starting on 2000 january 04 for 45.4 ks ( obsid 775 ) .
the reduction of these data proceeded as described in rakowski et al .
our image was constructed from a subset of a 3-d position - velocity data cube obtained on 71 using the rutgers fabry - perot ( rfp ) at the focus of the ctio 1.5-m on 1998 february 18 .
the rfp data reduction is described in ghavamian et al .
( 2002 ) . the soft x - ray image of 71 from ( fig
. 1 , middle panel ) shows emission from an egg - shaped outer rim that matches nearly perfectly the optical image ( fig . 1 , left panel ) .
features in common include matched variations in surface brightness around the rim as well as the presence of multiple nested filaments , which are particularly noticeable along the southern edge .
faint diffuse emission in the interior of the snr agrees well too .
the remarkable detailed similarity of the x - ray and optical morphology is uncommon among supernova remnants in general .
nearly all of the optical emission in 71 arises in high velocity ( 5001000 km s@xmath5 ) nonradiative shocks , with little contribution from the slow ( @xmath1100 km s@xmath5 ) radiative shocks typically seen in middle - aged remnants .
furthermore the x - ray emission shows broadly similar intrinsic emissivities around the rim ( there are no large variations in abundance or thermodynamic state , see rakowski , ghavamian , & hughes 2002 ) so that surface brightness variations largely trace variations in the ambient gas density around the remnant .
north - south the outer rim spans 83@xmath6 , while in the east - west direction it is only 60@xmath6 across .
any differences between the x - ray and optical sizes are small , no more than a 0.5@xmath6 difference in diameter , with the optical dimensions being marginally larger than the x - ray ones . the hard x - ray image of 71 ( fig
. 1 , right panel ) shows emission from the outer rim that resembles the soft x - ray and images .
however , this band also reveals a new spatial component of x - ray emission that has no counterpart in the other bands .
this emission almost fills the interior and appears as a modestly limb - brightened , elliptical shell of emission slightly offset from the geometric center of the outer rim .
the emission extends 40@xmath6 by 32@xmath6 at a position angle ( for the major axis ) of @xmath112@xmath7 east of north and is centered at 05@[email protected]@xmath10 @xmath467@[email protected]@xmath6 ( j2000 ) .
hereafter we refer to this as the `` core '' emission .
spectra were extracted from four regions in the remnant s southeastern quadrant ( see figure 1 ) to investigate the soft and hard x - ray emission components .
two spectra of the soft rim ( `` outer '' and `` inner '' ) , corresponding to the two nested filaments that appear here , were extracted .
likewise two spectra of the core x - ray emission ( again `` outer '' and `` inner '' ) were extracted .
background - subtracted count rates are given in table 1 .
background represented less than 0.5% of the source count rate in the 0.25 kev band as taken from a 1@xmath132@xmath13 annulus outside the remnant . before fitting ,
source spectra were rebinned to a minimum of 10 counts per channel .
the rim data sets ( 2 left panels of fig . 2 ) are quite similar as are the core data sets ( 2 right panels ) , but the rim and core spectra differ significantly from each other .
= 18truecm the abundances and emission properties of the gas were constrained through fits of nonequilibrium ionization ( nei ) planar - shock thermal plasma models ( see hughes , rakowski , & decourchelle 2000 ) .
the abundances of o , ne , mg , si , s , and fe were allowed to vary , while c , n , ar and ca were fixed at 20% of solar and the ni abundance was tied to fe .
the fits were generally acceptabl weith similar best - fit values for similar regions ( see table 1 ) .
absorption of low energy x - rays by intervening gas and dust was included with the hydrogen column density , @xmath14 , as a free parameter .
the values obtained , @xmath14 = 4@xmath15 atoms @xmath16 , are consistent with previous x - ray spectral fits ( hughes et al . 1998 ) and the galactic column density in this direction ( dickey & lockman 1999 ) .
table 1 + nei thermal model fits for regions in 71 + @lcccc@ + parameter & outer rim & inner rim & outer core & inner core + + rate ( s@xmath5 ) & 0.29 & 0.20 & 0.098 & 0.093 + @xmath14 ( @xmath17 @xmath16 ) & @xmath18 & @xmath19 & @xmath20 & @xmath21 + @xmath22 ( kev ) & @xmath23 & @xmath24 & @xmath25 & @xmath26 + @xmath27 & @xmath28 & @xmath29 & @xmath30 & @xmath31 + o & @xmath32 & @xmath33 & @xmath34 & @xmath35 + ne & @xmath36 & @xmath37 & @xmath38 & @xmath39 + mg & @xmath40 & @xmath41 & @xmath42 & @xmath43 + si & @xmath44 & @xmath45 & @xmath46 & @xmath47 + s & @xmath48 & @xmath490.30 & @xmath492.6 & @xmath492.1 + fe & @xmath50 & @xmath51 & @xmath52 & @xmath53 + @xmath54/d.o.f & 103.1/100 & 79.4/91 & 85.6/73 & 102.6/69
+ + the ionization timescale of the rim region indicates that the gas is out of equilibrium , as expected .
the fitted abundances are sub - solar and generally consistent with the gas phase lmc values ( russell & dopita 1992 ) .
the fe abundance is derived from the l - shell lines that produce , at ccd spectral resolution , a quasi - continuum blend of emission over the 0.71.5 kev band .
note that our nei spectral model is derived from the original raymond & smith ( 1977 ) code ( and subsequent revisions ) which accurately models the overall flux of the fe l - shell emission , but is less accurate at modeling its detailed spectral variation with photon energy , due to limitations in the availability of atomic data .
this is the case to a greater or lesser extent for all nei spectral models currently available .
typically with such models , the fitted fe abundance is low , while the abundances of ne and mg , whose k - shell lines fall in this energy band , are enhanced to compensate for deficiencies in the modeled fe l - shell emission . nevertheless these spectral fits ( as well as the more comprehensive study of 71 s outer rim in rakowski et al .
2002 ) confidently confirm that the rim region is swept - up interstellar medium ( ism ) heated to x - ray emitting temperatures by a high velocity blast wave .
we note that there is no evidence for a hard power - law tail beyond the thermal emission in the spectrum of the rim .
in marked contrast to the rim regions , the core spectra are dominated by the fe l - shell blend , which peaks near 0.8 kev , and have no significant o line emission at @xmath10.65 kev .
the fitted parameters ( table 1 ) reveal that the abundances of all species , with the exception of o , are considerably greater than the lmc mean , while the abundances of some species , ne and fe in particular , are even greater than solar .
the abundance ratio of iron to oxygen is @xmath05 times the solar ratio .
these large absolute abundances , in addition to the non - solar abundance ratios , signal the presence of supernova ejecta in the core of the remnant .
here we quantitatively examine the properties of the core region to identify the evolutionary state of the remnant and the nature of the originating sn .
the large iron - to - oxygen ratio of the core x - ray emission , as well as its highly - symmetric , well - ordered spatial distribution , immediately suggests a sn ia origin for 71 .
a critical diagnostic of the sn type is the total mass of ejecta .
we estimate this in two ways , first using dynamical considerations and then from the properties of the core x - ray emission .
the position of the reverse shock ( or contact discontinuity ) relative to that of the blast wave can constrain the evolutionary state of a remnant in the context of self - similar models ( e.g. , truelove & mckee 1999 ) . once the profiles of the ism and ejecta are chosen , this constraint is unique .
although subject to hydrodynamic instabilities , the position of the contact discontinuity between the sn ejecta and the ambient medium is reasonably well identified in our image it is marked by the outer extent of the core hard emission .
this has a geometric mean radius of 4.3 pc assuming an lmc distance of 50 kpc .
71 s outer blast wave has a geometric mean radius of 8.6 pc , yielding a ratio for the radii of @xmath12 . in their study of instabilities and clumping in sne ia , wang & chevalier ( 2001 )
present the 1-d radial evolution of the contact discontinuity and blast wave for an exponential ejecta density profile interacting with a constant density ambient medium . for this model ,
the radius of the contact discontinuity is approximately one - half that of the blast wave when the normalized radius of the forward shock is @xmath55 .
the physical radius of the forward shock , @xmath56 , depends on the normalized radius , @xmath57 ; the ejected mass , @xmath58 ; and the ambient hydrogen number density , @xmath59 , through a scaling relation .
inverting , we obtain a relation for the ejected mass in terms of the chandrasekhar mass , @xmath60 @xmath61 which evaluates to @xmath62 for @xmath63 , @xmath64 pc and a mean ambient density of 0.5 @xmath65 ( ghavamian et al .
2002 ) . at this time
the entire ejecta have been thermalized , the reverse shock has just reached the center , and the ejecta are moving outward relatively slowly at @xmath66 100 km s@xmath5 . for our second ejecta mass estimate ,
we extracted spatially resolved spectra of the core from five concentric elliptical rings spaced by 5@xmath6 ( in semi - major axis length ) .
the annuli had constant ellipticity ( an axial ratio of 1.25 ) and position angle ( 12@xmath7 east of north ) .
each of these spectra contain projected emission from the blast wave , which we removed by subtracting off a spectrum of the bright rim emission scaled independently for each annular ring to zero out the resultant spectrum over the 0.20.6 kev band .
this band was chosen because it shows no evidence for enhanced central emission and thus plausibly corresponds to emission from the blast wave alone .
the count rates of the various spectra range from 0.068 s@xmath5 to 0.38 s@xmath5 .
these spectra all show the broad fe l - shell blend and clearly resolved si and s k@xmath67 lines , but no obvious k@xmath67 lines of o , ne , or mg .
we fit these spectra with the nei spectral model including a minimal set of elemental species : si , s , ar , ca , fe , ni , and a h continuum . to further simplify the model we restricted the number of free parameters in the fits .
the global nei parameters , temperature and ionization timescale , and the ni to fe ratio were constrained to be the same for all five spectra .
we allowed the emission measures ( ems ) of si and fe to vary freely in all the spectra ; however , the s , ar , and ca ems varied along with si at their relative solar abundance ratios .
ems are defined as @xmath68 , where @xmath69 and @xmath70 are the electron and ion number densities and @xmath71 is the emitting volume .
we fit both the low ( @xmath72 kev ) and high ( @xmath73 kev ) energy portions of the spectra separately in order to ensure that the fits were not being dominated by the `` shape '' of the fe l - shell blend .
in fact both portions of the spectra are broadly consistent with @xmath74 kev and @xmath75 .
finally the fits were performed for three values of the column density , @xmath76 atoms @xmath65 , fixed separately for three different sets of fits , in order to estimate uncertainties on the parameters .
the fits are not particularly good from a @xmath54 point - of - view , but most of the residuals are associated with the fe - l shell blend . from the fits we obtain numerical values for the ems of the fe and si - group elements as a function of radius in the core region .
for simplicity we assume spherical geometry for the deprojection .
if the true 3-d shape of the core region were ellipsoidal , then our mass estimates would be some 13% higher ( prolate ) or lower ( oblate ) .
we assume that the fe and si - group components are co - spatial , although we allow for the components to contribute in differing proportions in each radial shell .
the mean charge state of each species in the fit is calculated using the values quoted above for @xmath22 and @xmath77 .
for example , the mean number of electrons per ion is 11.8 for si and 18.2 for fe .
= 0.98 the main uncertainty in converting ems to fe and si density and mass estimates is not knowing which species contribute electrons . in a solar abundance plasma
, there are @xmath1@xmath78 electrons ( mostly from hydrogen ) per fe ion , compared to the roughly 50 electrons per fe ion in a pure metal plasma .
thus if even just a small quantity of hydrogen were mixed into the ejecta , it would drastically affect the derived densities .
our spectral fits are inconclusive regarding the amount of hydrogen present in the core region .
the spectral fits discussed here require a relatively high level of broadband continuum emission , yet in other fits we performed a wide range of continuum intensities were inferred , while the ems of the metals remained fairly stable . furthermore , although we have modeled a hydrogen continuum , any fully stripped ion could produce the requisite featureless spectrum .
thus in lieu of a definitive result we present a couple of astrophysically plausible estimates . for one
we assume that the metals are the sole source of electrons , which yields maximal density estimates .
for the other we make the plausible assumption that a comparable mass of hydrogen has been well mixed into the metal rich ejecta during its evolution .
the deprojected fe and si densities versus radius are plotted in figure 3 .
each clearly shows enhanced density around 34 pc indicating a thick shell structure with the si slightly ahead of the fe .
fe is present even in the center of the remnant , while si is absent there . integrating the density profiles we obtain total masses of 1.5 @xmath2 of fe and 0.24 @xmath2 of si assuming a pure metal composition for the emitting material .
the curves for the case with a modest admixture of hydrogen are quite similar in shape to those of the pure metal case ( see figure 3 ) , although the ion densities we infer are about a factor of two lower .
the total amount of fe is 0.8 @xmath2 and that of si is 0.12 @xmath2 .
the spatial and spectral analysis of the acis data reveal 71 to be a textbook example of a supernova remnant : an outer blast wave shock propagating through the ism that surrounds a core of sn ejecta heated to x - ray emitting temperatures by the inward moving reverse shock .
our analysis strongly argues for a type ia sn origin for 71 .
specifically we find ( 1 ) a large fe to o ratio in the ejecta x - ray emission ; ( 2 ) @xmath11.5 @xmath2 of ejecta from dynamical considerations ; ( 3 ) a relatively low mass ratio of si to fe ( @xmath10.15 ) ; and ( 4 ) a large mass of ejected fe ( @xmath10.8 @xmath2 ) .
these results are all consistent with sn ia ejecta and further may have some power to discriminate between models for how the flame front propagates in these explosions ( see , e.g. , iwamoto et al .
the ejecta are mostly contained within a thick shell and are partially stratified . according to our deprojection analysis
, fe extends throughout the entire ejecta even into the center , while si appears to be contained only within the outer half of the ejecta . a considerably deeper observation of 71 has been approved for observation in cycle 4 .
this will allow us to search for and study spectral variations within the fe - rich ejecta and to determine its morphology in greater detail .
measuring the kinematics of the fe - rich ejecta is an important next step .
unfortunately this is likely to be beyond the capabilities of , , and even 2 , although the constellation - x facility , if it attains angular resolution of 10@xmath6 , should be able to detect the expected motions of @xmath1100 km s@xmath5 .
we are also pursuing ground - based optical searches for coronal [ ] @xmath795303 line emission from the ejecta , which could provide complementary insights into the nature and dynamics of the ejecta .
we thank ted williams for assistance with the fabry - perot observations , leisa townsley for allowing us to use her cti corrector software , and john nousek and dave burrows for help with the original proposal .
partial financial support was provided by nasa contract nas8 - 39073 ( sao ) , grants go0 - 1035x and go1 - 2052x ( rutgers ) , and a nasa graduate student researchers program fellowship to cer .
davies , r. d. , elliott , k. h. , & meaburn , j. 1976 , mmras , 81 , 89 dickey , j. m. , & lockman , f. j. , 1990 , araa , 28 , 215 dwarkadas , v. v. , & chevalier , r. a. 1998 , apj , 497 , 807 ghavamian , p. , rakowski , c. e. , hughes , j. p. , & williams , t. b. 2002 , apj , submitted .
hughes , j. p. , hayashi , i. , & koyama , k. 1998 , apj , 505 , 732 hughes , j. p. , rakowski , c. e. , & decourchelle , a. 2000 , apj , 543 , l61 iwamoto , k. , brachwitz , f. , nomoto , k. , kishimoto , n. , umeda , h. , hix , w. r. , & thielemann , f. 1999 , , 125 , 439 kirshner , r. p. , & chevalier , r. a. 1978 , a&a , 67 , 267 long , k. s. , helfand , d. j. , & grabelsky , d. a. 1981 , apj , 248 , 925 rakowski , c. e. , ghavamian , p. , & hughes , j. p. 2002 , apj , submitted .
raymond j. c. , & smith , b. w. 1977 , apjs , 35 , 419 russell , s. c. , & dopita , m. a. 1992 , apj , 384 , 508 schweizer , f. , & lasker , b. m. 1978 , apj , 226 , 167 smith , r. c. , kirshner , r. p. , blair , w. p. , & winkler , p. f. 1991 , 375 , 652 .
truelove , j. k. , & mckee , c. f. 1999 , , 120 , 299 ( erratum : , 128 , 403 ) wang , c .- y . , & chevalier , r. a. 2001 , apj , 549 , 1119 | x - ray observations of 71 , a supernova remnant ( snr ) in the large magellanic cloud ( lmc ) , reveal a clear double shock morphology consisting of an outer blast wave shock surrounding a central bright region of reverse - shock heated ejecta .
the abundances of the outer shock are consistent with lmc values , while the ejecta region shows enhanced abundances of si , fe , and other species .
however , oxygen is not enhanced in the ejecta ; the fe / o abundance ratio there is @xmath05 times the solar ratio .
based on the relative positions of the blast wave shock and the contact discontinuity in the context of snr evolutionary models , we determine a total ejecta mass of @xmath11.5 @xmath2 .
ejecta mass estimates based on emission measures derived from spectral fits are subject to considerable uncertainty due to lack of knowledge of the true contribution of hydrogen continuum emission .
maximal mass estimates , i.e. , assuming no hydrogen , result in 1.5 @xmath2 of fe and 0.24 @xmath2 of si . under the assumption that an equal quantity of hydrogen has been mixed into the ejecta
, we estimate 0.8 @xmath2 of fe and 0.12 @xmath2 of si .
these characteristics support the view that in 71 we see fe - rich ejecta from a type ia sn several thousand years after explosion .
@footnotetext = footnotetext |
under the suggestion of the pioneering griffith s approach to fracture in brittle - elastic solids , a variational model in fracture mechanics has been proposed in @xcite .
it is based on a requirement of minimality of the overall energy @xmath2 which accounts for the macroscopic deformation and the possible presence of cracks , and is defined by @xmath3where @xmath0 is the regular region occupied in the three - dimensional ambient space by the body , @xmath4 the spatial derivative evaluated at @xmath5 of the differentiable transplacement map @xmath6 the map defining the actual ( deformed ) place @xmath7
@xmath8 the representation in @xmath0 of a surface - like crack occurring in @xmath7 , @xmath9 the elastic energy , @xmath10 a constant surface energy , @xmath11 the two - dimensional hausdorff measure .
the map @xmath12 is also assumed , as usual , to be one - to - one outside @xmath13 , orientation preserving ( i.e. @xmath14 ) , and such that the global invertibility condition@xmath15holds for all @xmath16 , a condition allowing frictionless self - contact of the boundary while still preventing self - penetration ( for details on this last condition see @xcite ) .
in other words , @xmath12 is an orientation preserving homeomorphism outside the subset of @xmath13 containing its jump set .
the requirement is then that _ at each instant _ @xmath17 $ ] _ of a cracking process the pair _ @xmath18 _
realize a minimum of the global energy _
@xmath2with @xmath13 an admissible crack .
admissibility is intended in the sense that @xmath13 is a rectifiable set ( specifically the image of a countable number of lipschitz maps ) with zero volume measure .
the interval of time is then discretized and minimality is required at time steps .
minimizers are pairs @xmath19 : the minimum problem has two variables .
notwithstanding simplicity and elegance of this model , the evaluation of minima of the energy at each time step involves a number of analytical problems .
the essential difficulty arises in controlling in three dimensions minimizing sequences of surfaces leading to the possible actual crack , or better to its picture @xmath13 in the reference place .
however , when the entire crack is open , @xmath13 coincides with the jump set of the transplacement field @xmath20 . the convenient simplification of identifying cracks with the jump sets of displacement fields has been then adopted .
bounded variation ( bv ) or special bounded variation ( sbv ) functions have been then involved as candidates to be minimizers of the energy considered as a functional of the sole @xmath21 .
the energy is then considered as the one of a simple cauchy s body and minimizers are sought in a space of maps including candidates to be reasonable descriptors of the elastic - brittle behavior .
remind that the choice of function spaces where one researches minimizers has constitutive nature .
a difficulty has been however encountered : theorems allowing the selection of fields with discontinuity sets describing reasonable ( physically significant ) crack patterns seem to be not available at least in the current literature ( see @xcite for a pertinent review of the results along this line ; also @xcite , @xcite , @xcite ) .
moreover , this kind of approach seems to be not able to account for partially opened cracks .
in fact , in this case the transplacement field is continuous across the closed part of the crack although the material bonds are broken in the actual place .
the description of partially closed cracks can have even stringent interest in the time - discretized procedure sketched above .
in fact , when such a procedure is applied by taking into account a loading program described by time - dependent boundary conditions , a program implying even the growth of pre - existing cracks besides the nucleation of a new fracture at the @xmath22th instant , it can happen that the cracks may close even partially , and then they re - open at subsequent time steps .
moreover , the discovery of physically appropriate crack patterns is another key point .
such questions have been tackled from a different point of view in gmmm ( see also additional comments in @xcite , @xcite ) by using tools from geometric analysis .
the skeletal features of the theory are listed in the ensuing items .
( _ i _ ) : : distinction is made between cracks and jump set of the transplacement field , as in @xcite .
the latter set is however constrained to be contained within the crack pattern .
differently from all previous proposals , the crack pattern is described through measures over a fiber bundle with typical fiber the grassmanian of ` planes ' through @xmath0 , the so - called curvature varifolds .
( _ ii _ ) : : a generalized notion of curvature can be associated with curvature varifolds .
it enters the constitutive structure of the surface energy along the crack margins . the resulting energy differs from griffith s proposal for the presence of the generalized curvature of the varifold and the energy along the tip in three dimensions . in this sense
the model is an evolution of griffith s scheme .
the curvature - dependence of the surface energy has analytical advantages and permits the control of minimizing sequences .
a theorem showing the existence of pairs of crack patterns and transplacement fields in appropriate measure and function spaces is then available ( see @xcite ) .
it has some implications : ( _ a _ ) : : the crack pattern results a rectifiable set .
although it can be very irregular , it has the features that our intuition assigns to a fracture .
( _ b _ ) : : balance equations can be derived in weak form from the first variation even for a _ generic _ rectifiable set .
notice that if a crack is assumed to be coincident with the discontinuity set of the transplacement field only , to obtain balance equations , stronger regularity assumptions on the geometry of the crack pattern are necessary ( see detailed analyses in @xcite , @xcite , @xcite , @xcite , @xcite ) .
these consequences imply that , beyond the analytical advantages , the physical meaning of items ( _ i _ ) and ( _ ii _ ) deserves additional analyses .
they can be developed by going along the essential steps of the theory . in continuum
mechanics the choice of function spaces as ambient for solutions to equilibrium or evolution problems has constitutive nature .
the properties of the members of a given space carry a physical meaning about .
the characteristic physical features of the problem under analysis have then to address the functional choice . in the case under scrutiny ,
the idea is that the material is elastic - brittle .
it means that the material is elastic up to a certain threshold after which a crack is created while outside the crack the material is still in elastic phase .
the threshold can be expressed in terms of deformation , stress , energy , depending on circumstances . moreover , as it will be clear later , the threshold can be also not expressed directly .
it is , in a sense , included in the choice of the function space that one selects .
let us focus the attention on pure elasticity first .
if the material is purely elastic , it can deform at will , without any threshold and , in principle , without end .
the deformation can be also perfectly recovered , after unloading . in this sense ,
phenomena like cavitation in solids are ascribed to elastic - brittle behavior rather than perfect elastic setting .
if this view is accepted , the consequence would be that a perfectly ( hyper)elastic body should be such that any compatible transplacement ( or displacement , depending on the choice ) field does not describe the nucleation of fractures or holes .
it is well known that under conditions of polyconvexity of the elastic energy density , minimizers of the elastic energy of a simple body can be found in the sobolev space @xmath128 a space hosting maps with first distributional derivative having integrable @xmath129power , i.e. the first derivative is in @xmath130 . however , when @xmath131 non negligible is the presence of transplacement maps with graphs admitting boundaries with projections into the interior of @xmath0 .
such boundaries describe the formation of ` holes ' and/or open ` fractures ' of various nature , so they are undesirable when a purely elastic material is under analysis , at least if the view on elasticity sketched above is accepted .
the difficulty can be overcame .
in fact , there is a global way to check eventually to control the presence of boundaries in the graph of a map with projection into the domain of the map itself through linear functionals .
indicate them by @xmath132 , with the indices @xmath21 suggesting that the functional @xmath133 is associated with the transplacement @xmath21 . in the case
treated here , such functionals are linear over smooth @xmath134forms with compact support in @xmath135 . the general technique , which is valid in any finite dimension @xmath136 of the ambient space , is described in the monograph @xcite .
here , i sketch only minimal ideas , furnishing the essential picture and physical interpretations in the case under scrutiny . preliminarily ,
remind that a @xmath137vector over a linear space @xmath138 is a rank@xmath139 skew - symmetric tensor , that is an element of the skew - symmetrization of @xmath140 .
the space of @xmath137vectors is indicated here by @xmath141 .
it has a natural dual @xmath142 .
any map of the type @xmath143 is called a @xmath137form as a real vector bundle over @xmath1 of fiber dimension @xmath136 , that is a family of @xmath87dimensional vector spaces parametrized by points of @xmath0 . ] .
the space of all @xmath144forms of the type just defined is indicated by @xmath145 for the mechanics treated here interesting is the case of @xmath134vectors over @xmath146 , the space hosting the graph of the deformation . consider a deformation ( transplacement ) @xmath147 , @xmath5 . amid all possible @xmath134vectors over @xmath148 ,
it is possible to define at every @xmath5 the @xmath134vector @xmath149 associated with the gradient of deformation @xmath150 .
detailed definition and various properties can be found in @xcite , here , for the expository purposes declared at the beginning , it is only necessary to know that at each @xmath25 its components are the entries of @xmath4 , @xmath151 , @xmath152 . in @xmath149 ,
then , all elements characterizing the deformation of lines , areas , and volume of the body in @xmath1 are included . in this sense
, @xmath149 characterizes completely the deformation .
its dual counterpart
the value at the same @xmath25 of some form in @xmath153 is then a sort of generalized stress .
given a transplacement @xmath6 , @xmath154 , the @xmath134**current integration * * @xmath132 ( * current * for short ) over the graph of @xmath21 is defined to be a linear functional over smooth @xmath134forms @xmath155 with compact support in @xmath156 , namely@xmath157where the angle brackets indicate the natural action over @xmath158 of its dual counterpart , that is the part of the graph that can be seen as the graph of lipshitz maps . ] .
essentially , @xmath159 plays the role of generalized internal power .
the number @xmath160 indicates here the so - called * mass of the current * and is defined by@xmath161where @xmath162 is the modulus of @xmath163 , evaluated in the standard way for tensors .
the symbol @xmath164 indicates the * total variation * of the current and is defined as usual for functionals .
a * boundary current * can be associated with @xmath132 : it is indicated by @xmath165 and defined by duality , that is and acts as @xmath166 . ]
@xmath167with @xmath168 the space of @xmath169forms over @xmath148 with compact support in @xmath170 .
the notion of boundary current has not only formal nature .
it has an immediate physical interpretation : when the graph of @xmath21 is free of boundaries inside the interior of @xmath0 , @xmath171 for any @xmath172 .
essentially , this zero boundary condition prevents the formation of cracks or holes inside the actual place @xmath173 of the body .
such a condition has been used ( see @xcite and the other references of its authors mentioned therein ) to define a class of transplacements the so - called * * weak diffeomorphisms * * _ _ _ _ which is ` constitutively ' an appropriate choice for describing what one imagines to be a pure elastic deformation , as sketched above .
here the situation is a bit more complicated .
the deformation has to be a weak diffeomorphism outside a subset of the support of the varifolds describing the crack pattern , if the minimizing procedure provides a non - empty family of minimizing varifolds .
inside that subset , the transplacement admits jumps , so it is not purely a weak diffeomorphism .
the definition has natural extension in @xmath186 : the summation in the item ( _ v _ ) should be extended up to @xmath121 .
of course , the definition of currents and related boundaries holds in dimension @xmath136 : in that case @xmath149 is tested over @xmath87forms . here and in the whole paper , the restriction @xmath187 is essentially motivated by the physics under scrutiny ( see @xcite for the abstract theory ) .
the first item indicates the possibility of evaluating the gradient of deformation .
the second item is another regularity condition .
it implies that one can in principle measure the average of the gradient of deformation , the volume change , the overall deformation of surfaces .
the third item is the standard condition that a transplacement be an orientation preserving map
. item ( _ iv _ ) is the condition mentioned at the beginning of section 1 .
it permits to move @xmath0 along @xmath21 into a region @xmath188 in such a way that self - contact between parts of the boundary @xmath189 be allowed while self - penetration excluded .
notice that in item ( _ v _ ) the action of the projector @xmath190 on the total variation of the boundary current is motivated by the fact that the latter behaves substantially as a measure .
essential properties for the space of extended weak diffeomorphisms are shown in @xcite ( see there the relevant theorems and proofs ) .
standard weak diffeomorphisms are @xmath191 maps that satisfy the items ( _ ii _ ) , ( _ iii _ ) , ( _ iv _ ) in previous definition while item ( _ v _ ) which is substituted by the zero boundary condition @xmath192 for any @xmath193 . from a kinematic point of view , in going from @xmath194 to @xmath195 , one transits from pure elastic setting to elastic - brittle behavior . in this sense ,
the requirement of minimality of the energy includes the possibility of finding minimizers if any in terms of weak diffeomorphisms and null varifolds , and in terms of extended weak diffeomorphisms and non - null varifolds .
the transition from a situation to another is morally the threshold from the elastic to the elastic - brittle behavior . in this sense ,
also , there is no need in principle of adding another condition defining the threshold itself .
there could be also minimizers for which the transplacement field is simply a weak diffeomorphism but the varifolds are not null .
this situation describes presence of closed cracks only : in other words , material bonds are broken but the crack remains closed and the transplacement field does not jump across the crack facies .
such a situation can occur in a step - by - step minimization program obtained by updating in time steps the boundary conditions and requiring minimality of the energy at each step . at the step @xmath196 a crack pattern can occur
, at the step @xmath197 the deformation closes the cracks , at the step @xmath198 there is a purely elastic continuation , then , at further steps , new crack patterns accrue . of course , proving existence of minimizers is a crucial step for attributing sense to the previous reasonings .
existence depends on the characteristic properties of the energy and the boundary conditions .
once the existence of minimizers is established , the characterization of them along the physical suggestions collected above is matter of regularity theorems .
the question is open .
actually , any regularity theorem is available .
different is the case of the existence problem .
the technique discussed previously is a general tool for the description of phenomena in which energy can in principle be concentrated over submanifolds of a certain manifold , and this energy depends on the geometry of the submanifold itself .
it can be used to analyze either specific situations of physical interest or to formulate and analyze abstract mathematical problems . *
phenomena of physical interest that can be described by using the tools mentioned hitherto deal for example with the mechanics of linear defects like dislocations and discontinuity surfaces . for dislocations the choice of the varifolds play a crucial role . for interfaces ,
the functional setting has to be changed and a new special class of extended weak diffeomorphisms arises .
the new choice requires proof of completeness of the new functional class and evaluation of the applicability of lower semicontinuity results . *
another point of discussion is also the evaluation of the crack nucleation and growth in complex bodies .
the adjective ` complex ' distinguishes bodies characterized by a prominent influence of changes in material texture ( the * microstructure * ) on the macroscopic behavior , an influence exerted through inner actions requiring a representation going beyond the common picture in terms of standard stresses .
quasicrystals , ferroelectrics , magnetoelastic materials , polymeric bodies of various nature , including elastomers , fullerene - based composites , porous bodies , bodies with continuous distributions of dislocations , multiphase materials are paradigmatic examples . notwithstanding
the variety of special models , a unitary picture of the mechanics of complex bodies exists . within it ,
the representation of bodies goes beyond standard cauchy s approach in the sense that every material element is viewed as a system rather than a black box individuated by a single point in the ambient space , which is cauchy s view .
such a description is multifield and intrinsically multiscale .
a * morphological descriptor * field @xmath247 , @xmath248 , of the essential geometrical features of the material microstructure is then introduced . to construct the essential structures of the relevant mechanics ,
it is just necessary to presume that @xmath249 is an element of a set @xmath250 which has just the structure of a differentiable manifold .
it is assumed to be finite - dimensional for the sake of simplicity .
@xmath250 is called the * manifold of substructural shapes*. the interest of mentioning here complex bodies is motivated by data showing that the microstructural changes may influence in non negligible way the force driving crack tips along evolution processes . theoretical analysis of this phenomenon within the setting of the general model building framework of the mechanics of complex bodies ( that is without specifying the type of microstructure ) has been developed in @xcite from a point of view different from the one adopted here .
the simplest extension of the theory discussed here to complex bodies is given by an energy of the type@xmath251with the natural modifications in generic dimension @xmath136 .
in analyzing the existence of minimizers , an essential point is the choice of the space hosting the morphological descriptor maps .
such a choice could require the embedding of @xmath250 into a linear space .
such embedding always exists because @xmath250 is finite dimensional , also it can be isometric when @xmath252 is riemannian . in all cases , however , it is not unique so that it becomes a ingredient of the model , a sort of constitutive choice .
another point is the link of the jump set of the morphological descriptor field with the varifolds . here
the underlying physics is subtle . in principle
one can accept that @xmath247 may have jumps even outside the support of the varifolds and there it may be even continuous .
jumps outside the varifolds can be justified by the formation of domains of microstructures , like polarization or magnetization domains .
the meaning of the possible continuity on the support of the varifolds is associated with the question whether in cracking a body one alters along the margins of the crack the microstructure , in a sense determining a new type of microstructure , or , else , the microstructure remains the same across the margins of the crack . in the philosophy of continuum mechanics , the remark above coincides with asking whether a crack just divides neighboring material elements or breaks the material elements met in front of the tip .
the answer can not be definitive and is matter of modelling .
the situation becomes also more complicated when the structure of the surface energy involving the generalized curvature of the varifold is enriched by making more articulated assumptions .
relevant investigations are actually open . * a point which may deserve to be noted is that the scheme discussed in previous sections has intrinsic similarity with the general framework of the mechanics of complex bodies which has been sketched rapidly in the last item .
in fact , when curvature @xmath34varifolds with boundary are chosen to represent cracks , in principle the region where they localize is not known in other words one does not know where the support of the varifolds is placed in @xmath0 , that is where the crack is . in this setting , every point can be crossed in principle by a crack .
the minimization procedure tells us that the crack is here or there , before nothing is known about its position . instead of assigning to each material element a morphological descriptor
@xmath253 selected in a finite dimensional differentiable manifold @xmath252 , _ one is then assigning to each material element a measure_. in this sense , the scheme discussed here is driven by the ideas of the mechanics of complex bodies , and , in some sense , it goes beyond them a bit .
* acknowledgements*. discussions with mariano giaquinta , giuseppe modica , gianfranco capriz , gilles francfort , luigi ambrosio have pushed me in various manners to write this paper .
i thank them for the impulse .
this work has been developed within the activities of the research group in ` theoretical mechanics ' of the ` centro di ricerca matematica ennio de giorgi ' of the scuola normale superiore at pisa .
the support of the gnfm - indam is acknowledged .
giaquinta , m. , mariano , p. m. , modica , g. , mucci , d. ( 2009 ) , stratified energies : ground states with cracks , _ proceedings of the iutam symposium progress in the theory and numerics of configurational mechanics _ , p. steinmann ed . ,
springer verlag , p. 159 - 168 .
giaquinta , m. , mariano , p. m. , modica , g. , mucci , d. ( 2010 ) , currents and curvature varifolds in continuum mechanics , _ in `` nonlinear partial differential equations and related topics '' , a volume dedicated to n. n. uralceva _ , a. a. arkhipova and a. nazarov edts . , american mathematical society , _ in print_. giaquinta , m. , martinazzi , l. ( 2005 ) , _ an introduction to the regularity theory for elliptic systems , harmonic maps and minimal graphs _ , lecture notes of the scuola normale superiore ( new series ) ,
edizioni della normale , pisa .
consider the place @xmath0 of a body , selected as a reference , as an open , connected set in the three - dimensional ambient space , with surface - like boundary oriented by the normal at each point , to within a finite number of corners and edges . if a @xmath23 crack is formed in the deformed configuration @xmath7 @xmath21 the transplacement and crosses a generic point @xmath24 with @xmath25 in @xmath0 , its ` direction ' is locally described by tangent plane to the crack at @xmath24 , when the crack is smooth .
when the crack margins have a corner at @xmath24 , a cone of planes has to be considered .
crack patterns can be however very irregular .
one can accept a set as a representative of a crack pattern when it is just rectifiable as mentioned hitherto .
so , an approximate notion of tangent plane is available in geometric measure theory ( see @xcite ) .
crack patterns can have a fictitious representation in the reference place @xmath1 ( let say through sets with zero volume measure ) the reference place is now mutant because the macroscopic structure of the body is changing with the nucleation and growth of a crack pattern .
let @xmath26 indicate a two - dimensional plane or a straight line in three dimensional ambient space where @xmath0 is contained .
the pair @xmath27 gives in @xmath0 local information on the geometry of the crack crossing possibly @xmath24 .
of course , up to when a real crack is not realized , any @xmath26 in the star at @xmath25 can be a candidate to describe locally the direction of a possible crack pattern .
the pair @xmath28 can be viewed as a typical point of a fiber bundle @xmath29 , @xmath30 , with natural projector @xmath31 and typical fiber @xmath32 , the grassmanian of @xmath33planes or straight lines associated with @xmath0 .
a @xmath34**varifold * * over @xmath35 is a non - negative radon measure @xmath36 over the bundle @xmath37 ( see almgren , @xcite , @xcite , @xcite , mantegazza ) .
the measure @xmath36 has a projection over @xmath35 obtained by using the _ projector of measures _ indicated here by @xmath38 associated with the natural projection @xmath39 defining the fiber bundle @xmath40 .
such a projection , namely @xmath41 , is radon measure over @xmath35 and is also indicated for short by @xmath42 .
it is called the weighed measure of the varifold and defines the so - called _ mass _ @xmath43 of the varifold itself through the relation @xmath44 . for the purpose of describing crack patterns through measures ,
the essential point is the possibility of constructing varifolds over a subset @xmath45 of @xmath35 .
the subset @xmath45 can have variegate nature . here , for the physical purpose at hand , it is assumed that it is an admissible crack in the sense sketched above .
let @xmath46 be the @xmath34dimensional hausdorff measure in @xmath47 , @xmath30 .
it is assumed that @xmath45 is a @xmath48measurable , @xmath34rectifiable subset of @xmath35 . for @xmath49
a function in @xmath50 , the approximate tangent @xmath51space allows one to treat in a unitary way surface and linear cracks . ]
( here @xmath52 or @xmath23 ) @xmath53 to @xmath45 at almost every @xmath25 in @xmath45 is defined for @xmath54 a.e .
the symbol @xmath56 indicates the orthogonal projection onto @xmath53 .
a varifold @xmath57 , restricted to @xmath45 , can be then defined .
it is called the * rectifiable varifold * associated with @xmath45 , with density @xmath49 , and is such that@xmath58for any @xmath59 .
rectifiable sets can be considered a sort of generalized surfaces ( see almgren ) .
a subclass of them admits a notion of generalized mean curvature vector ( see @xcite , @xcite ) . for members of such subclass ( not all )
a notion of second fundamental form can be defined ( see hutchinson ) . here
the attention is on varifolds admitting density @xmath49 with integer values , the so - called * integer rectifiable varifolds*. they allow the definition of a special class of varifolds ( see mantegazza ) which is essential for the ensuing developments .
a new ingredient has to be inserted .
it is a third - rank tensor field @xmath60 over @xmath61 , with components @xmath62 .
it plays the role of a * generalized curvature*. a varifold @xmath36 is called a * curvature * @xmath34**varifold with boundary * * if ( _ i _ ) @xmath36 is an integer , rectifiable @xmath34varifold @xmath63 associated with the triple @xmath64 , ( _ ii _ ) there exists a function @xmath65 , _ _ _ _ and a vector radon measure @xmath66 _ _ _ _ such that , for every _ _ _ _ @xmath67 , one gets@xmath68the vector measure @xmath66 _ _ _ _ is called the * varifold boundary measure * @xcite . the subclass of varifolds with generalized curvature @xmath69 in @xmath70 is indicated here by @xmath71 .
it is possible to show ( see @xcite ) that if @xmath72 , with @xmath73 , @xmath74 is locally the graph of a multivalued function of class @xmath75 , @xmath76 , far from @xmath66 .
* varifolds with boundary can be used as descriptors of crack patterns : ( _ i _ ) the set @xmath45 has the minimal geometrical properties of an admissible crack , at least in the sense mentioned above .
( _ ii _ ) the density @xmath49 furnishes information on its possible faceted shape there is a smooth surface , @xmath77 , when there is a net fold , @xmath78 , and so on . ] .
( _ iii _ ) the local orientation of the crack pattern is accounted for through @xmath26 .
( _ iv _ ) curvature is considered , although in the generalized ( weak ) form specified above .
( _ v _ ) the boundary of the crack it includes the tip is described by the boundary of the varifold . *
consider a smooth crack crossing a body in @xmath0 and intersecting somewhere its boundary but maintaining the tip in the interior of @xmath0 .
a two - dimensional ( @xmath79 ) varifold @xmath80 describes the crack , its boundary measure @xmath81 is supported by the entire boundary of the crack itself . to represent separately the crack tip , that is the part of the boundary of the crack in the interior of @xmath0 , a specific one - dimensional varifold @xmath82 has to be inserted .
it is supported by the tip alone .
its boundary is supported by possible corners along the tip and the points determining the intersection of the tip with the external boundary of the body @xmath83 .
the insertion of @xmath84 allows one to assign later energy to the tip of the crack .
different properties can be also assigned to the corners of the tip by using @xmath85 . of course , to capture the intuitive structure of the geometry under scrutiny , the varifolds @xmath80 and @xmath82 have to satisfy a certain link .
a definition presented in @xcite specifies the link : a family @xmath86 of @xmath34varifolds with boundary in @xmath87dimensional ambient space is said to be * stratified * when @xmath88 for all @xmath89 s . in the special case treated here the condition of stratification reduces to @xmath90 . * the choice @xmath30 for constructing @xmath37 and the associated varifolds allows one to consider not only two - dimensional cracks with the relative tips but also additional linear defects ( @xmath91 ) which can be cracks included in very thin tubes
material bonds are broken along a line for some reason or even dislocations . in the latter case
, through a one - dimensional varifold one can describe dislocations emanating from a crack tip in a three - dimensional body . in the former case one
manages crack patterns stratified over various dimensions .
of course , the analogous description can be adopted in space dimension @xmath92 and stratification of defects of various nature can be accounted for . a new form of the energy for a body undergoing fractures , based on the description of cracks in terms of varifolds , has been proposed in @xcite ( see also comments in @xcite , @xcite ) . such an energy is indicated below by @xmath93 .
it differs from the expression @xmath94coming from the traditional griffith s proposal and is an extension of it . for a three - dimensional body it reads:@xmath95where @xmath96 , @xmath97 , @xmath98 and @xmath99 are constitutive coefficients .
in particular , @xmath96 , @xmath97 , @xmath98 are positive numbers , so the contribution of the generalized curvature of the varifolds is always present , even if it can be extremely small .
the density @xmath100 is defined as the difference @xmath101 between the * bulk elastic energy * @xmath102 and the * potential * @xmath103 * of external body forces*. a number of comments on its physical meaning are necessary . *
the addendum @xmath104 has the role of the last integral in @xmath94 , that is the role of griffith s surface energy : @xmath105 has the same meaning of @xmath106 in @xmath94 . of course , the first addendum is the bulk elastic energy with density @xmath107 the body is then simple but anisotropic and coincides with the first integral in @xmath94 .
the other terms are not standard . *
the addendum @xmath108 counts energy along the tip .
such an energy is proportional to the length of the tip itself , namely to the mass @xmath109 of the one - dimensional varifold supported by the tip itself .
* the term @xmath110 adds possible energy concentrated at the tip corners where material bonds can be entangled in principle in a way different from the other parts of the tip .
* the two addenda@xmath111mark in a more pronounced manner the difference with respect to @xmath112 . they have pure configurational nature : they do not involve directly the gradient of deformation @xmath4 .
the first one accounts for the ( generalized ) curvature of the varifold describing the surface of the crack , the second one includes the curvature of the varifold describing the tip .
influence of the curvature of the crack , above all in the proximity of the tip , has been recognized in @xcite , a work devoted however to other aspects of fracture processes , namely grinfeld s instability .
the curvature here aims to account at macroscopic level of local microstructural effects at low scale , occurring in the cracking process .
curvature energy can be associated with bending effects in breaking material bonds when a crack is determined consider for example a material in which the inner bonds are modeled through beam - like interactions .
a point has to be stressed : bending occurs in the current configuration @xmath7 while , as a function of @xmath25 , @xmath113 is defined over the reference configuration @xmath0 .
however , when bending of material bonds in the actual configuration breaks the bonds themselves , such a bending has a configurational effect because it contributes to the mutation of @xmath0 due to the nucleation and possible propagation of a crack .
all configurational effects are measured in @xmath0 , as it is commonly accepted .
additionally , analogies with different approaches to crack analyses in complex bodies ( see @xcite ) and bodies with strain - gradient effects ( see @xcite , @xcite ) can be called upon to enforce the interpretation of the presence of @xmath114 as a configurational indicator of the effects due to latent microstructures .
the terms including @xmath114 in the energy tell us essentially that , once boundary conditions are prescribed , one has to pay in energy for curving the crack .
consider a rectangular planar sheet of a homogeneous material half of it including a straight crack in the middle as in figure 1 .
apply boundary conditions in terms of transplacement ( dirichlet boundary conditions ) in mode 1 .
if the boundary conditions are such that the crack can growth , it remains straight , unless some additional agency occurs paying energy in curving it .
* consider a body in an initial configuration in which no crack is present .
for example , @xmath0 is a three - dimensional ball .
after some loading program , a certain configuration in which a crack occurs is reached .
clearly one could consider such a configuration as the initial one .
the possibility stresses the point that ` being cracked ' is only a relative concept
. configurations or better states have to be compared to affirm that a body is cracked .
moreover , in sequences of configurations ( or states ) in which one excludes the possibility of restoring cracks by gluing the matter across crack facies , an order relation has to be considered .
such an order is given by monotonicity in crack patterns : if in a certain configuration there is a crack with respect to an uncracked configuration in the sequence , a subsequent configuration can have a crack pattern that coincides with or includes the previous crack . in terms of varifolds
such a point of view is expressed by affirming the existence a family of _ comparison varifolds _ @xmath115 such that the family of varifolds @xmath116 describing the actual crack pattern is constrained by @xmath117 for any @xmath89 and @xmath118 .
the assignment of @xmath115 does not mean that one is considering in a given configuration a pre - existing crack pattern always , because the comparison varifold family can be also empty . *
the assumption used so far that the ambient space is three - dimensional has been accepted only to be close to the standard physical intuition .
there is no obstruction to consider higher dimensional spaces .
consider the dimension of the ambient space to be @xmath92 .
the previous treatment can be extended to this case straight away .
the only modification , including the expression of the energy is that the range of @xmath89 must be considered to be @xmath119 , at least in principle .
the rest remains unchanged .
this choice allows one to consider stratified families of defects at various scales with various physical meanings , depending on the circumstance . of course , there could be cases in which @xmath89 could not take all the natural values from @xmath120 to @xmath121 . even in three - dimensional ambient space
one could have only @xmath91 , for example , being in the condition to consider only linear defects like nets of dislocations .
such a case would deserve perhaps a treatment a part .
the case of four manifolds , @xmath122 , could be called upon when the description of crack in relativistic elastic bodies is considered , with all the necessary changes in the representation of the energy , adapted to the relativistic setting . in this case , however , care must be taken because in the relativistic setting a representation of continuous bodies based on the back - to - label representation seems to be preferable .
the circumstance then would change the stage and would include a sort of ` mixed ' representation of the energy .
solution to this problem , if any , is a pair @xmath126 .
minimization over a class of varifolds has the meaning of minimization over a class of bodies : every possible crack pattern represented over @xmath0 ( remind that nucleation andor growth of cracks occur in the current place ) defines a body .
different crack patterns indicate different bodies : @xmath0 , in fact , changes .
the family of varifolds @xmath116 represents the crack pattern so it selects a body and the field @xmath127 describes the deformation of such a body .
the two ` objects ' are correlated .
in fact , in the actual configuration possible nucleation , growth and/or opening of a crack are consequences of the deformation , the varifolds over @xmath0 are representatives of what happens in the actual configuration .
the choice of the function space hosting the generic @xmath21 is then another key point of the treatment .
it has to be linked to the varifolds supported on @xmath0 .
the existence result for the minimum problem stated above for the energy @xmath199 of an elastic brittle solids has been proven in @xcite .
boundary conditions of dirichlet type can be presumed .
they are given by prescribing the transplacement field along the boundary @xmath83 of @xmath200 .
the discussion of the existence is developed by taking first a subspace of @xmath201 , precisely the space @xmath202 defined by@xmath203for some @xmath204 . essentially , the choice of @xmath205 with @xmath204 is a request of additional regularity which is sometimes necessary for physical needs .
combination with the space of varifolds allows one to recognize a natural ambient in which the existence of minimizers of the energy @xmath93 can be investigated .
such a space is indicated by @xmath206 and defined by@xmath207where @xmath208 and @xmath209 are comparison varifolds describing possible initial cracks . in particular , the subspace@xmath210with @xmath211 the part of the boundary of the body where the transplacement field is prescribed , takes into account the boundary conditions of dirichlet type mentioned above . in all these definitions ,
another regularity requirement is prescribed .
in fact , the condition @xmath212 imposes that the essential supremum of @xmath21 is almost everywhere with respect to the lebesgue measure bounded .
in fact , a priori it is not possible to exclude that , if one is able to prove under some conditions the minimality of the energy @xmath213 over some space of extended weak diffeomorphisms and varifolds , the minimizing varifold does not describe a fragmentation of the body let say a crack cutting a piece of matter from the rest . in this case
, a transplacement field could be such that the cut piece , now free from boundary conditions , can be translated rigidly to infinity . by imposing that @xmath214 , with @xmath215 a real number
, then , one wants to avoid the situation just sketched .
so , in this sense the assignment of @xmath215 has not properly constitutive nature .
it is not related to some property of the material , rather it is a parameter selecting admissible deformation processes , admissibility considered with reference to the possible unconstrained extraction of pieces of matter from the body . for @xmath136 the dimension of the ambient space and @xmath89 ranging from @xmath120 to @xmath121 , analogous definitions of @xmath216 hold and the theory can be generalized ( see relevant results in @xcite ) .
another crucial point in the path leading to the proof of existence theorem of minimizers of the energy is the discussion of the structural properties of the energy .
they have constitutive nature , of course . in non - linear
elasticity of simple cauchy s bodies , common assumptions about the structure of the energy @xmath217 are well known . by indicating by @xmath218 the space of @xmath219 matrices with positive determinant , the energy density @xmath9 is considered as a map@xmath220\]]with values @xmath221 .
remind that positiveness of the determinant of @xmath4 is the condition assuring that the transplacement be orientation preserving .
the properties h1-h4 below are then assumed to hold .
h1 : : @xmath222 $ ] is * continuous * in @xmath223 .
h2 : : the map @xmath224 is * polyconvex * : that is there exists a borel function @xmath225 acting as@xmath226with values @xmath227 , which is continuous in @xmath228 for every @xmath229 , convex and lower semicontinuous in @xmath230 for every @xmath228 , and such that and @xmath150 on @xmath25 is now suppressed for the sake of brevity . ] @xmath231 for any list of entries @xmath232 with @xmath233 .
h3 : : the energy density @xmath9 satisfies the growth condition@xmath234 h4 : : for every @xmath5 and @xmath235 , if for some @xmath236 the inequality @xmath237 is satisfied , then @xmath233 .
the physical nature of these assumptions is discussed in various treatises ( see @xcite , @xcite ) . the standard presence in the polyconvex energy of the determinant of the gradient of deformation and the relevant adjugate
is summarized here in the functional dependence on @xmath149 .
the choice is not only formal .
it furnishes a rapid path toward the extension of the treatment to @xmath87dimensional cases ( see @xcite ) .
however , the essential point is to underline that , in the setting explored here , h1-h4 do not need to be supplemented by additional structural assumptions on the energy to assure the existence of minimizers of @xmath238 .
the relevant theorem reads as follows : * since no additional structural hypotheses besides h1-h4 of standard non - linear elasticity need to be added , the information about the possible nucleation of a crack or growth of an existing one is furnished by the presence of the terms ruled by varifolds in the energy and the _ constitutive _ choice of @xmath216 as functional setting .
it is just the latter choice that avoids the introduction of an external criterion for the nucleation of a crack or the growth of an existing one .
in fact , the energy is minimized over a class of possible bodies . * more in general than other descriptions , it is possible to determine the weak form of balance equations for crack patterns which are just rectifiable sets .
the result is not discussed here for the sake of conciseness .
it is presented in @xcite and opens the way to computational opportunities not explored yet .
the balance equations derived naturally in @xcite for very general crack geometries as mentioned above the crack pattern has to be just a rectifiable set are the ones obtained by horizontal variations , that are variations of the reference place .
thus they have configurational nature : they involve in fact the hamilton - eshelby tensor and non - standard terms deriving from the variations of the terms including the varifolds , the ones directly related with the geometry of the crack pattern . *
the existence result holds also at dimensions greater than 3 , provided that the obvious variations in previous definitions ( see @xcite , 2 for the abstract theory ) . *
an analogous result holds also for the generalized energy@xmath244with @xmath245 a convex real - valued function satisfying the condition @xmath246 .
the interest of this remark is not only technical .
it gives a grater degree of freedom in selecting further constitutive structures under the suggestions of possible experimental evidences and numerical tests .
* the existence result does not exclude the possibility that the minimization procedure foresees stratified varifolds supported on the boundary of @xmath0 . in this case
one can say that a boundary crack appears .
the meaning of such a boundary cracks is not exotic .
in fact , on a part of the boundary where the transplacement is imposed a crack can occur so that the boundary condition is ` broken ' .
consider for example a beam jointed at one of its ends .
boundary conditions and properties of the material can be such that a crack occurs just at the interface between the joint and the beam .
however , in principle a boundary crack can appear on a free part of the boundary .
such a situation can describe the fragmentation of a thin film at the boundary , which is , essentially , the abrasion of the boundary itself . | the nucleation and/or growth of cracks in elastic - brittle solids has been recently described in @xcite in terms of a special class of measures and with a variational technique requiring the minimization of a certain energy over classes of bodies . here ,
the physical foundations of the theory and the basic ideas leading to it are described and commented further on .
a view on certain possible developments and shifts toward different settings is also given .
this article has expository character .
fracture mechanics , elastic - brittle materials , currents , curvature varifolds , classes of bodies , extended weak diffeomorphisms when the adjective ` variational ' is attributed to models of some physical phenomena , impulsively one may think that the physical situation under scrutiny is conservative .
variational means , in fact , that one is managing some functional , essentially an energy , and is asking that it attains its minimum on some function space . the situation is well known and is typical , for example , of the determination of ground states of elastic bodies under prescribed boundary conditions . the energy is defined on the reference place @xmath0 occupied by the body in the ambient space , a place fixed once and for all .
its minimum is required to be attained over a class of one - to - one orientation preserving maps .
variational descriptions of the nucleation and the evolution of defects like cracks however exist . at least in principle
, the question of their physical appropriateness can be posed , due to the dissipative character of the phenomena they are referred to .
of course , one can say that the quest of something to be minimized has rather instrumental character because in nature some economic principle always appears somewhere and the calculus of variations is at a stage of development that it can be desirable to be under conditions of using it .
this point of view can be an answer .
however , the answer can be even deeper .
when one think of the nucleation of defects , in fact , one manages a _
mutant body_. mutations occur , in fact , in the gross shape of the body at the continuum level , and they can be naturally pictured as mutations of the reference place @xmath0 which is now _ not _ fixed once and for all as for example in the standard formulation of elasticity . in principle ,
one can imagine of having at disposal a class of possible bodies occupying @xmath1 , every member of the class differing from the others by the defect pattern .
once boundary conditions are prescribed , a variational principle can select the resulting body within a class . this way , a variational approach to the defect nucleation
the latter imposed by the interplay between boundary condition and nature of the material is physically significant when it involves a minimization process of the free energy ( for example ) over an entire class of possible bodies .
this point of view is the one adopted here and is also the answer to the question of physical appropriateness of some variational approaches .
the energy dissipated in the nucleation ( or growth ) process is individuated in the gap arising from the body in the original shape the one existing before the assignment of boundary conditions and the actual shape ( i.e. the body plus the defect pattern ) obtained by the minimization procedure . of course the explicit evaluation of this gap can be arduous .
this point of view interprets the physical aspects of two existing variational models of fracture mechanics .
both models are recalled below .
the attention is focused on the second one in which cracks are described through appropriate measures , the so - called curvature varifolds .
the physical significance of the approach is discussed here .
it is shown also that the point of view can be adopted in other circumstances dealing with the nucleation of defects stratified over manifolds with different dimensions . |
inferring the nature of dust and dust heating requires knowledge about the dust emissivity , and its power law index .
there is some evidence supporting variation of dust emissivity with environmental conditions ( e.g. lisenfeld et al .
2000 , paradis et al .
the dust spectral index could change depending on grain properties , e.g. geometry , size distribution , and chemical composition .
dust grains might be affected by different physical processes like shattering , sputtering , grain - grain collision , condensation and coagulation .
hence , depending on recent star formation activities , dust / gas content and metallicity changes across galaxies , variations in the dust spectral index in inner / outer disks or arm / inter - arm regions may occur .
herschel observations of the nearest scd galaxy m33 , ( herm33es , kramer et al .
2010 , boquien et al .
2010 ) ideally enable us to derive distribution of @xmath0 across m33 .
we use the herschel pacs & spire data at 160 , 250 , 350 , and 500@xmath2 m together with the spitzer mips data at 70@xmath2 m to investigate the physical properties of a standard dust model ( big grains+vsg+pah , krgel 2003 and references therein ) .
these wavelengths spread over a range where the dust emission mainly emerges from the big grains , which are believed to be in thermal equilibrium with the interstellar radiation field .
we use the newton - raphson iteration method to derive @xmath3 , @xmath0 and column density , pixel by pixel , in two different approaches : + 1 ) we assume that dust emitting at the spire bands is mainly heated by the interstellar radiation field ( isrf ) and obeys single - component gray body radiation . in this case
we use the data at longer wavelengths ( 250 , 350 and 500@xmath2 m ) .
+ 2 ) we consider another dust component heated by young massive stars with higher temperature ( warm dust ) besides the dust heated by the isrf ( cold dust ) , hence , dust emits as a two - component gray body . here
, we use the data from 70@xmath2 m to 500@xmath2 m wavelengths .
+ using a monte - carlo simulation , we further investigate the reliability of each approach and estimate the errors and uncertainties .
we derive the following results : + + as shown in fig . 1 , in the central disk ( @xmath4 kpc ) , the dust spectral index is higher in the central disk ( @xmath51.4 ) than in the outer disk .
similarly , both t and column density are higher in the central disk .
the flocculent spiral structure of m33 is evident in the column density map .
the average t and @xmath0 for the entire galaxy are 22@xmath69 k and [email protected] , respectively .
+ the cold and warm dust components are separated assuming a variable @xmath0 for the dominant colder component .
although we have to use a fixed value of @xmath7 for the warm dust , we repeat the analysis for different possibilities of @xmath7=1 , 1.5 , and 2 .
interestingly , we find that @xmath0 is again higher in the central disk and even shows the arm / inter - arm contrast , irrespective of @xmath7 assumed . also in all three cases , the cold dust follows spiral arms and dust lanes and is denser where the temperature is lower .
the warm dust ( both temperature and column density ) follows the distribution of h@xmath1 and star forming regions . for the entire galaxy ,
we derive t@xmath8=16 @xmath63k , t@xmath9=40@xmath613k , and @[email protected] assuming @xmath7=2 .
our simulations show that the two - component approach , generally , constrains the parameters better than the single - component approach , particularly the temperature .
the m33 s outer / inner disk and even the spiral arms are visible in the @xmath0-maps . having similar distribution as of the molecular gas and radiation field traced by h@xmath1 , @xmath0 seems to have a meaningful change across the galaxy .
figure 2 shows correlations between @xmath0 and h@xmath1 , co(2 - 1 ) and hi integrated intensities ( gratier et al . 2010 ) as tracers of the ionized , molecular , and atomic phases of the ism , respectively . in both approaches , @xmath0 is found to be better correlated with both h@xmath1 and co than with hi .
this could indicate that @xmath0 is not the same in star forming and non - starforming regions in a galaxy .
star formation could in principle influence the grain composition ( via a change in metallicity ) and size distribution ( via shattering of dust grains due to strong shocks ) , which could modify the dust emissivity index i.e. @xmath0 . | using the herschel pacs and spire fir / submm data , we investigate variations in the dust spectral index @xmath0 in the nearby spiral galaxy m33 at a linear resolution of 160pc .
we use an iteration method in two different approaches , single and two - component modified black body models . in both approaches
, @xmath0 is higher in the central disk than in the outer disk similar to the dust temperature .
there is a positive correlation between @xmath0 and h@xmath1 as well as with the molecular gas traced by co(2 - 1 ) .
a monte - carlo simulation shows that the physical parameters are better constrained when using the two - component model . |
let @xmath14 denote a sample of the discrete gaussian free field ( gff ) on @xmath3 pinned to 0 at the origin .
explicitly , @xmath15 is a centered gaussian process such that @xmath16 where @xmath17 is the green function in @xmath18 ; i.e. , the expected number of visits to @xmath19 for the simple random walk on @xmath3 started at @xmath20 and killed upon reaching the origin . for @xmath21 and conditional on the sample @xmath6 of the gff , let @xmath22 be a discrete - time markov chain with transition probabilities given by @xmath23 where @xmath24 denotes the @xmath25-norm on @xmath3 .
we will write @xmath26 for the law of the above random walk such that @xmath27 and use @xmath28 to denote the corresponding expectation .
we also write @xmath29 for the law of the gff and use @xmath30 ( as above ) to denote the expectation with respect to @xmath29 .
the transition kernel @xmath31 depends only on the differences @xmath32 whose law is , as it turns out , invariant and ergodic with respect to the translates of @xmath3 .
the markov chain @xmath33 is thus an example of a random walk in a stationary random environment .
the main conclusion we prove about this random walk is then : [ thm_main_spectral ] for each @xmath0 and each @xmath34 , @xmath35 furthermore , @xmath36 is recurrent for @xmath29-almost every @xmath6 .
the transition probabilities @xmath31 are such that the walk prefers to move along the edges where @xmath6 increases ; the walk is thus driven towards larger values of the field .
this has been predicted ( e.g. , in @xcite ) to result in a subdiffusive behavior .
we prove a version of subdiffusivity for the expected exit time from large balls : [ thm_main_exit_time ] let @xmath37 denote the first exit time of @xmath38 from @xmath39 ^ 2\cap{\mathbb{z}}^2 $ ] . for each @xmath34
, we then have @xmath40 where @xmath41 4\gamma/\gamma_{{\text{\rm c}}},\qquad&\text{otherwise}. \end{cases}\ ] ] the bounds on the expected hitting time indicate that @xmath42 should scale as @xmath43 for large @xmath44 .
although we expect this to be true , we have so far only been able to prove a corresponding lower bound : [ thm_main_diffusive_exponent ] for @xmath29-almost every @xmath6 and each @xmath34 , @xmath45 where @xmath46 is as in .
we note that theorems [ thm_main_exit_time ] and [ thm_main_diffusive_exponent ] are consistent with the predictions in @xcite for general log - correlated fields .
in particular , confirms the prediction for the diffusive exponent of the walk from @xcite as a lower bound .
the reason why the bounds in are not sufficient is that we do not know whether @xmath37 scales with @xmath10 proportionally to its expectation .
a full proof of subdiffusive behavior thus remains elusive .
the technical approach that makes our analysis possible stems from the following simple rewrite of the transition kernel , @xmath47 this represents @xmath33 as a random walk among random conductances , or a random conductance model to which a large body of literature has been dedicated in recent years ( see @xcite for reviews ) .
an immediate benefit of the rewrite is that the process is now reversible with respect to the measure @xmath48 on @xmath3 defined by @xmath49 a price to pay is that the conductance @xmath50 of edge @xmath51 now depends on @xmath6 and not just its gradients , and the law of the conductances is thus not translation invariant .
as it turns out , the change of the behavior of the expected exit time at the critical point @xmath52 ( see theorem [ thm_main_exit_time ] ) arises , in its entirety , from the asymptotic @xmath53 this is , roughly speaking , because point - to - point effective resistances in the associated random conductance network @xmath54 behave , for points at distance @xmath10 , as @xmath13 for every @xmath0 .
the precise statement is the subject of : [ thm - effective - resistance ] let us regard @xmath39 ^ 2\cap{\mathbb{z}}^2 $ ] as a conductance network where edge @xmath51 has conductance @xmath50 .
let @xmath55 denote the effective resistance between @xmath20 and @xmath19 in network @xmath56 .
for each @xmath0 there are @xmath57 such that @xmath58 holds for each @xmath59 and each @xmath60 . moreover , for the corresponding network @xmath54 on all of @xmath3 , there is a constant @xmath61 such that @xmath62 and , for each @xmath0 and each @xmath34 , also @xmath63 the effective resistance and further background on the theory of resistor networks are discussed in detail in section [ sec : sec_gen_par_ser ] .
we note that , in light of monotonicity of @xmath64 , the bounds in theorem [ thm - effective - resistance ] readily imply recurrence of the random walk as well . runs of @xmath65 steps of the random walk with transition probabilities confined ( through reflecting boundary conditions ) to a box of side - length @xmath66 .
labelled left to right , the plots correspond to @xmath67 equal to @xmath68 , @xmath69 and @xmath70 ; time runs upward the vertical axis .
trapping effects are quite apparent.,title="fig:",scaledwidth=25.0% ] runs of @xmath65 steps of the random walk with transition probabilities confined ( through reflecting boundary conditions ) to a box of side - length @xmath66 . labelled left to right ,
the plots correspond to @xmath67 equal to @xmath68 , @xmath69 and @xmath70 ; time runs upward the vertical axis .
trapping effects are quite apparent.,title="fig:",scaledwidth=25.0% ] runs of @xmath65 steps of the random walk with transition probabilities confined ( through reflecting boundary conditions ) to a box of side - length @xmath66 .
labelled left to right , the plots correspond to @xmath67 equal to @xmath68 , @xmath69 and @xmath70 ; time runs upward the vertical axis .
trapping effects are quite apparent.,title="fig:",scaledwidth=25.0% ] 0.0 cm runs of the liouville random walk ( see ) for time @xmath65 in the same environments , and for same values of @xmath71 , as in fig .
the difference between these walks is quite obvious , particularly so for larger @xmath71.,title="fig:",scaledwidth=25.0% ] runs of the liouville random walk ( see ) for time @xmath65 in the same environments , and for same values of @xmath71 , as in fig .
the difference between these walks is quite obvious , particularly so for larger @xmath71.,title="fig:",scaledwidth=25.0% ] runs of the liouville random walk ( see ) for time @xmath65 in the same environments , and for same values of @xmath71 , as in fig .
the difference between these walks is quite obvious , particularly so for larger @xmath71.,title="fig:",scaledwidth=25.0% ] 0.0 cm closely related to our problem is the recently - defined _ liouville brownian motion _ ( lbm ) , which is basically just a time change of the standard brownian motion by an exponential of the continuum gaussian free field .
the construction of the process was carried out in @xcite , with the associated heat kernel constructed in @xcite . in @xcite ,
the spectral dimension ( defined as 2 times the exponent for the return probability computed in almost sure sense with respect to the underlying random environment ) for lbm was computed , and in @xcite some nontrivial bounds for off - diagonal lbm heat kernel were established .
a random walk naturally associated with lbm is the continuous time simple symmetric random walk with exponential holding time at @xmath72 having parameter @xmath73 where , in our notation , @xmath74 .
a more natural ( albeit qualitatively similar , as far as long - time behavior is concerned ) modification is to use @xmath75 ( see ) in instead of @xmath76 ; we will refer to the associated process as the liouville random walk ( lrw ) below .
formally , this process is a continuous - time markov chain on @xmath3 with generator @xmath77.\ ] ] the nature of the transition rates of the lrw precludes formulation using conductances and , no surprise , our analysis is thus quite different from those mentioned above .
for instance , unlike for the lrw , our random walk moves preferably towards neighbors with a higher potential , emphasizing the trapping effects of the random environment ; see fig .
the off - diagonal heat kernel computation in @xcite is also of a different flavor : our control of the return probability relies crucially on the electric - resistance metric while the off - diagonal lbm heat kernel is expected to be related to the liouville first passage ( liouville fpp ) percolation metric ( see @xcite ) .
notwithstanding the above differences , both the lrw and our random walk share the following fact : @xmath78 defined above is a stationary measure ( conditional on @xmath6 ) for both processes . the same thus applies to any interpolation between the lrw and our random walk ; namely , the continuous - time markov chain with generator @xmath79\ ] ] for any @xmath80 $ ] . assuming that the scaling limits of these random walks can eventually be extracted
, one may thus obtain a one - parameter family of natural diffusions evolving on the background of a continuum gaussian free field .
another series of related works is on random walks on random planar maps .
this is thanks to the conjectural relation between lqg and random planar maps ( note that part of the conjecture has been established in @xcite ) .
building on ideas from the theory of circle packings @xcite , the authors of @xcite proved that the uniform infinite planar triangulation and quadrangulation are both almost surely recurrent . in @xcite , it was shown that the random walk on the uniform infinite planar quadrangulation is sub - diffusive , where an upper bound of @xmath81 on the exponent was given while the conjectured exponent is @xmath82 .
as mentioned above , our work relies on estimates of effective resistances , which is a fundamental metric for a graph . recently ,
some other metric properties of gff have been studied , including the pseudo - metric defined via the zero - set of the gff on the metric graph @xcite , the liouville fpp metric @xcite ( which is roughly the graph distance on the network @xmath83 if we regard edge conductances as passage times ) and the chemical distance for the level - set percolation @xcite .
these studies reveal different facets of the metric properties of the gff . in particular , by @xcite and the present paper , we see that putting random weights / conductances as exponential of the gff substantially distorts the graph distance of @xmath2 but has much less of an effect on the resistance metric of @xmath2 . in light of the connection between random walks and effective resistances ( see , e.g. , @xcite for some background ) , the principal step ( and the bulk of the paper ) is the proof of theorem [ thm - effective - resistance ] .
this theorem is proved by a novel combination of planar and electrostatic duality , gaussian concentration inequality and the russo - seymour - welsh theory , as we outline below .
duality considerations for planar electric networks are quire classical .
they invariably boil down to the simple fact that , in a planar network , every harmonic function comes hand - in - hand with its harmonic conjugate .
an example of a duality statement , and a source of inspiration for us , is ( * ? ? ?
* proposition 9.4 ) , where it is shown that , for locally - finite planar networks with sufficient connectivity , the wired effective resistance across an edge ( with the edge removed ) is equal to the free effective conductance across the dual edge in the dual network ( with the dual edge removed ) . however , the need to deal with more complex geometric settings steered us to develop a version of duality that is phrased in purely geometric terms .
in particular , we use that , in planar networks with a bounded degree , cutsets can naturally be associated with paths and _ vice versa_. the starting point of our proofs is thus a representation of the effective resistance , resp . ,
conductance as a variational minimum of the dirichlet energy for _ families _ of paths , resp .
, cutsets .
although these generalize well - known upper bounds on these quantities ( e.g. , the nash - williams estimate ) , we prefer to think of them merely as extensions of the parallel and series law . indeed , the variational characterizations are obtained by replacing individual edges by equivalent collections of new edges , connected either in series or parallel depending on the context , and noting that the said upper bounds become sharp once we allow for optimization over all such replacements .
we refer to propositions [ prop : fundamental_prop_resistance ] and [ prop : fundamental_prop_conductance ] in section [ sec : sec_gen_par_ser ] for more details .
another useful fact that we rely on heavily is the symmetry @xmath84 which implies that the joint laws of the conductances are those of the resistances . using this we can _
almost _ argue that the law of the effective resistance between the left and right boundaries of a square centered at the origin is the same as the law of the effective conductance between the top and bottom boundaries .
the rotation symmetry of @xmath6 and the ( electrostatic ) duality between the effective conductance and resistance would then imply that the law of the effective resistance through a square is the same as that of its reciprocal value .
combined with a gaussian concentration inequality ( see @xcite ) , this would readily show that , for the square of side @xmath10 , this effective resistance is typically @xmath13 .
however , some care is needed to make the `` almost duality '' argument work .
in fact , we do not expect an exact duality of the kind valid for critical bond percolation on @xmath3 to hold in our case .
indeed , such a duality might for instance entail that the law of the conductances on a minimal cutset ( separating , say , two points ) in the primal network is the same as the law of the resistances on the dual path `` cutting through '' this cutset .
although the gffs on a graph and its dual are quite closely related ( see , e.g. , @xcite ) , we do not see how this property can possibly be true .
notwithstanding , we are more than happy to work with just an approximate duality which , as it turns out , requires only a uniform bound on the _ ratio _ of resistances of neighboring edges .
this ratio would be unmanageably too large if applied the duality argument to the network based on the gff itself .
for this reason , we invoke a decomposition of gff ( see lemma [ lem : field_decomp_smooth ] ) into a sum of two independent fields , one of which has small variance and the other is a highly smooth field .
we then apply the approximate duality to the network derived from the smooth field , and we argue that the influence from the other field is small since it has small variance .
we have so far explained only how to estimate the effective resistances between the boundaries of a _
square_. however , in order to prove our theorems , we need to estimate effective resistances between vertices , for which a crucial ingredient is an estimate of the effective resistances between the two short boundaries of a _ rectangle_. questions of this type fall into the framework of the russo - seymour - welsh ( rsw ) theory .
this is an important technique in planar statistical physics , initiated in @xcite with the aim to prove uniform positivity of the probability of a crossing of a rectangle in critical bernoulli percolation .
recently , the theory has been adapted to include fk percolation , see e.g. @xcite , and , in @xcite , also voronoi percolation .
in fact , the beautiful method in @xcite is widely applicable to percolation problems satisfying the fkg inequality , mild symmetry assumptions , and weak correlation between well - separated regions .
for example , in @xcite , this method was used to give a simpler proof of the result of @xcite , and in @xcite , a rsw theorem was proved for the crossing probability of level sets of the planar gff . the current flow realizing ( through thomson s principle ) the effective resistance between two points in the network associated with a sample of the gff in a square of side @xmath85 and @xmath86 ( left ) and @xmath87 ( right ) .
the source / sink of the current lie on the diagonal of the square ( see the dark spots in the left figure ) .
the intensity of the shade increases with the value of the current .
, title="fig:",scaledwidth=42.0% ] the current flow realizing ( through thomson s principle ) the effective resistance between two points in the network associated with a sample of the gff in a square of side @xmath85 and @xmath86 ( left ) and @xmath87 ( right ) .
the source / sink of the current lie on the diagonal of the square ( see the dark spots in the left figure ) .
the intensity of the shade increases with the value of the current . ,
title="fig:",scaledwidth=42.0% ] 0.0 cm our rsw proof is _ hugely _ inspired by @xcite , with the novelty of incorporating the ( resistance ) metric rather than merely considering connectivity .
we remark that in a recent work @xcite , a rsw result was established for the liouville fpp metric , again inspired by @xcite .
it is fair to say that the rsw result in the present paper is less complicated than that in @xcite , for the reason that we have the approximate duality in our context which was not available in @xcite .
however , our rsw proof has its own subtlety since , for instance , we need to consider crossings by whole collections of paths simultaneously . the rsw proof is carried out in section [ sec : rsw ] .
once the effective resistances are under control , we move on to the proof of the results on random walks .
the upper bound on the return probability is proved in section [ subsec : spectral_dim_upper ] using the methods drawn from @xcite .
the lower bound on the return probability is more subtle as it requires showing that the effective resistivity from @xmath88 to @xmath19 in @xmath56 is bounded by the sum of the resistances from @xmath88 to @xmath89 and from @xmath19 to @xmath90 ) .
this amounts to bounding a _ difference _ of effective resistances , which is not immediate from the estimates obtained thus far .
we approach this by invoking a concentric decomposition of the gff along a sequence of annuli , which permits representing of the typical value of the resistance as an exponential of a random walk .
the law of the iterated logarithm then shows that the natural fluctuations of the effective resistance ( which are of order @xmath91 ) can be beaten in at least one of the annuli .
these key steps are the content of proposition [ prop-5.8 ] and lemma [ lemma-5.11 ] . as an immediate consequence ,
we then get recurrence and , in fact , also the bounds in theorem [ thm - effective - resistance ] .
voltage profiles for the same geometric setting as in fig . 1 but for two different samples of gff at @xmath92 ( left ) and @xmath87 ( right ) .
notice that the profile represents the probability that a random walk hits the highest point before hitting the lowest point in this graph.,title="fig:",scaledwidth=45.0% ] -8 mm 0.0 cm we feel that our method of estimating effective resistances provides a novel framework which may have applications in other planar random media .
in fact , from our proofs we should be able to see that our method can be adapted to some other log - correlated gaussian fields such as those considered in @xcite .
we refrain ourselves from doing so , for the reason that we do not yet know how to characterize the class of log - correlated gaussian fields with subpolynomial ( i.e. , @xmath13-like ) growth of the effective resistances .
one important , and perhaps less conspicuous , ingredient of our proofs is the estimate of the effective resistance by means of the gaussian concentration inequality .
when the underlying random media is not a function of a gaussian process , a derivation of such a concentration inequality seems to be a challenge .
a natural class of non - gaussian models where one should try to prove an analogue of theorem [ thm - effective - resistance ] is that of gradient fields with uniformly convex interactions .
indeed , there the required concentration is implied by the brascamp - lieb inequality .
concerning our future goals for the problem at hand , our first attempt will aim at the computation of the spectral dimension ( which amounts to an almost sure version of ) and an upper bound on the diffusive exponent matching the lower bound in theorem [ thm_main_diffusive_exponent ] .
our ultimate goal is to prove existence of an appropriate scaling limit of the whole problem .
this applies not only to the walk itself , but also to the resistance metric as well as the associated current and voltage configurations ; see fig .
[ fig3 ] and [ fig4 ] for illustrations .
we wish to thank takashi kumagai , hubert lacoin , rmi rhodes , steve lalley , vincent vargas and ofer zeitouni for helpful discussions .
as noted above , our asymptotic statements on the random walk hinge on estimates of effective resistance between various sets in @xmath3 .
these will in turn rely crucially on a certain duality between the effective resistance and the effective conductance which will itself be based on the distributional equality of @xmath6 with @xmath93 .
the exposition of our proofs thus starts with general versions of these duality statements .
these can be viewed as refinements of ( * ? ? ?
* proposition 9.4 ) and are therefore of general interest as well .
let @xmath94 be a finite , unoriented , connected graph where each edge @xmath95 is equipped with a resistance @xmath96 , where @xmath97 denotes the set of positive reals .
we will use @xmath94 to denote both the corresponding network as well as the underlying graph .
let @xmath98 and @xmath99 respectively denote the set of vertices and edges of @xmath94 .
we assume for simplicity that @xmath94 has no self - loops although we allow distinct vertices to be connected by multiple edges .
for the purpose of counting we identify the two orientations of each edge ; @xmath99 thus includes both orientations as one edge .
two edges @xmath95 and @xmath100 of @xmath94 are said to be _ adjacent _ to each other , denoted as @xmath101 , if they share at least one endpoint .
similarly a vertex @xmath19 and an edge @xmath95 are adjacent , denoted as @xmath102 , if @xmath19 is an endpoint of the edge @xmath95 .
@xmath103 is a sequence of vertices of @xmath94 such that any two successive vertices are adjacent .
we also use @xmath103 to denote the subgraph of @xmath94 induced by the edge set of @xmath103 . for @xmath104 , a _ flow _
@xmath105 from @xmath20 to @xmath19 is an assignment of a number @xmath106 to each _ oriented _ edge @xmath107 such that @xmath108 and @xmath109 whenever @xmath110 .
the _ value of the flow _
@xmath105 is then the number @xmath111 ; a unit flow then has this value equal to one . with these notions in place ,
the effective resistance @xmath112 between @xmath20 and @xmath19 is defined by @xmath113 where the infimum ( which is achieved because @xmath94 is finite ) is over all unit flows from @xmath20 to @xmath19 .
note that we sum over each edge @xmath114 only once , taking advantage of the fact that @xmath115 appears in a square in this , and later expressions .
recall that a multiset of elements of @xmath116 is a set of pairs @xmath117 for some @xmath118 for each @xmath119 .
we have the following alternative characterization of @xmath112 : [ prop : fundamental_prop_resistance ] let @xmath120 denote the set of all multisets of simple paths from @xmath20 to @xmath19 .
then @xmath121 where @xmath122 is the set of all assignments @xmath123 such that @xmath124 the infima in are ( jointly ) achieved .
let @xmath125 denote the right hand side of .
we will first prove @xmath126 .
let thus @xmath127 and @xmath128 subject to be given .
we will view each edge @xmath95 in @xmath94 as a _
parallel _ of a collection of edges @xmath129 where the resistance on @xmath130 is @xmath131 and , if the inequality in for edge @xmath95 is strict , a dummy edge @xmath132 with resistance @xmath133 such that @xmath134 . in this new network
, @xmath135 can be identified with a collection of _ disjoint _ paths where ( by the series law ) each path @xmath136 has total resistance @xmath137 .
the parallel law now guarantees @xmath138 which proves @xmath139 as desired .
next , we turn to proving that @xmath140 and that the infima in are achieved . to this end , let @xmath141 be the flow that achieves the minimum in . in light of the inequality @xmath142 it suffices to construct a collection of paths @xmath143 and an assignment of resistances @xmath144 such that @xmath145 the argument proceeds by constructing inductively a sequence of flows @xmath146 from @xmath20 to @xmath19 ( whose value decreases to zero ) and a sequence of collections of paths @xmath147 as follows .
we initiate the induction by setting @xmath148 and @xmath149 and employ the following iteration for @xmath150 : * if @xmath151 for all @xmath114 , then set @xmath152 and stop . * otherwise , there exists a path @xmath153 from @xmath20 to @xmath19 such that @xmath154 for all @xmath155 .
denote @xmath156 . *
set @xmath157 and let @xmath158 for all @xmath155 .
* set @xmath159 for all @xmath155 and @xmath160 for all @xmath161 , and repeat .
since the set @xmath162 is strictly increasing ( and our graph is finite ) , the procedure will stop after a finite number of iterations ; the quantity @xmath163 then gives the number of iterations used .
note that the same also shows that the paths @xmath153 are distinct .
we will now show the desired inequality with @xmath164 and @xmath165 for @xmath166 .
first , abbreviating @xmath167\colon = \{1,\dots , j\}$ ] , we have @xmath168\colon e\in p_j } \alpha_j = \theta^\star_e\ ] ] for each @xmath114 . employing the definition of @xmath169 we get @xmath168\colon e\in p_j } \alpha_j^2 r_{e
, p_j } = \sum_{j\in [ j]\colon e\in p_j } \alpha_j \theta^\star_e r_e = r_e\,(\theta^\star_e)^2\ ] ] and so @xmath170\colon e\in p_j } \alpha_j^2 r_{e , p_j } = \sum_{e\in e(\mathfrak g)}r_e\,(\theta^\star_e)^2\ , . \ ] ] rearranging the sums yields @xmath170\colon e\in p_j } \alpha_j^2 r_{e , p_j } = \sum_{j\in [ j ] } \alpha_j^2 \bigl(\,\sum_{e\in p_j } r_{e , p_j}\bigr ) , \ ] ] where @xmath171 } \alpha_j = 1 $ ] .
abbreviating @xmath172 , the right hand side of the preceding equality is minimized ( subject to the stated constraint ) at @xmath173 } 1/r_j}$ ] , and therefore @xmath168 } \alpha_j^2 \bigl(\,\sum_{e\in p_j } r_{e , p_j}\bigr ) \geq \bigl(\,\sum_{j\in [ j ] } \frac{1 } { r_j}\bigr)^{-1}\,.\ ] ] this completes the desired inequality including the construction of a minimizer in . a slightly augmented version of the above proof in fact yields : [ fundamental_cor_resistance ] let @xmath174 be the set of all multisets of edges of @xmath94 that , if considered as a graph on @xmath98 , contains a path between @xmath20 and @xmath19 .
then @xmath175 where @xmath176 is the set of all assignments @xmath177 such that @xmath178 the infima are jointly achieved for @xmath179 being a subset of @xmath180 .
let @xmath125 denote the right - hand side of .
obviously , @xmath181 so restricting the first infimum to @xmath182 , proposition [ prop : fundamental_prop_resistance ] shows @xmath183 .
( this will also ultimately give that the minimum is achieved over collections of paths . ) to get @xmath184 , let us consider an assignment @xmath185 satisfying .
for each @xmath186 , let @xmath187 denote an arbitrarily chosen simple path between @xmath20 and @xmath19 formed by edges in @xmath44 . then , defining @xmath188 for each @xmath186 , we find that the assignment @xmath189 satisfies .
now the claim follows from the simple observation that @xmath190 .
an alternative way to approach an electric network is using conductances .
we write @xmath191 for the edge conductance on @xmath95 , and define the effective conductance between @xmath20 and @xmath19 by @xmath192 ^ 2\,,\ ] ] where @xmath193 are the two endpoints of the edge @xmath95 ( in some _ a priori _ orientation ) and the infimum is over all functions @xmath194 satisfying @xmath195 and @xmath196 .
the infimum is again achieved by the fact that @xmath94 is finite .
the fundamental electrostatic duality is then expressed as @xmath197 and our aim is to capitalize on this relation further by exploiting the geometric duality between paths and cutsets . here
we say that a set of edges @xmath198 is a cutset between @xmath20 and @xmath19 ( or that @xmath198 separates @xmath20 from @xmath19 ) if each path from @xmath20 to @xmath19 uses an edge in @xmath198 .
[ prop : fundamental_prop_conductance ] let @xmath199 denote the set of all finite collections of cutsets between @xmath20 and @xmath19 .
then @xmath200 where @xmath201 is the set of all assignments @xmath202 such that @xmath203 the infima in are ( jointly ) achieved .
the proof is structurally similar to that of proposition [ prop : fundamental_prop_resistance ] .
denote by @xmath204 the quantity on the right hand side of .
we will first prove @xmath205 .
pick @xmath206 and @xmath207 subject to .
now view each edge @xmath95 as a _ series _ of a collection of edges @xmath208 where the conductance on @xmath209 is @xmath210 and , if the inequality in is strict , a dummy edge @xmath132 with conductance @xmath211 such that @xmath212 . in this new network
, @xmath213 can be identified with a collection of _ disjoint _ cutsets , where the cutset @xmath214 has total conductance @xmath215 .
the nash - williams criterion then shows @xmath216 thus proving @xmath217 as desired .
next , we turn to proving @xmath218 and that the infima in are attained .
let @xmath219 be a function that achieves the infimum in .
this function is discrete harmonic in the sense that @xmath220 for @xmath110 , where @xmath221.\ ] ] in light of the inequality @xmath222 , it suffices to construct a collection of cutsets @xmath223 and conductances @xmath224 such that @xmath225 ^ 2\,.\ ] ] we will now define a sequence of functions @xmath226 satisfying @xmath227 and a sequence of collections of cutsets @xmath228 as follows .
initially , we set @xmath229 and @xmath230 .
abbreviating @xmath231 , we employ the following iteration for @xmath232 : * if @xmath233 is constant on @xmath98 , then set @xmath152 and stop .
* otherwise , by ( and positivity of all @xmath234 s ) we have @xmath235 and hence there exists a cutset @xmath236 separating @xmath20 from @xmath19 such that @xmath237 for all @xmath155 .
we take @xmath236 to be the closest cutset to @xmath20 that is , one that is not separated from @xmath20 by another such cutset and define @xmath238 . *
set @xmath239 and let @xmath240 for all @xmath155 . * set @xmath241 for all @xmath242 , where @xmath243 denotes the endpoint of @xmath95 with a larger value of @xmath233 . for all other vertices @xmath72 , set @xmath244 , and repeat .
we see that the above procedure will stop after a finite number of iterations , since all the cutsets @xmath236 are different by our construction .
the number @xmath163 is then the total number of iterations used .
the validity of for all @xmath245 follows directly from the construction . in order to prove ,
we now proceed as follows .
first , we have @xmath168\colon e\in \pi_j } \alpha_j = df^\star(e)\ ] ] and so , by the definition of @xmath246 , @xmath168\colon e\in \pi_j } \alpha_j^2 c_{e , \pi_j } = \sum_{j\in [ j]\colon e\in p_j } \alpha_j \,df^\star(e ) c_e = ( df^\star(e))^2 c_e\,.\ ] ] it follows that @xmath170\colon e\in \pi_j } \alpha_j^2 c_{e , \pi_j } = \sum_{e\in e(\mathfrak g ) } c_e\,\bigl[f^\star(e_+ ) - f^\star(e_-)\bigr]^2 \,.\ ] ] rearranging the sums yields @xmath170\colon e\in \pi_j } \alpha_j^2 c_{e , \pi_j } = \sum_{j\in [ j ] } \alpha_j^2 \bigl(\,\sum_{e\in \pi_j } c_{e , \pi_j}\bigr),\ ] ] where @xmath171 } \alpha_j = 1 $ ] .
abbreviating @xmath247 , the right hand side of the preceding equality is minimized ( subject to the stated constraint ) at @xmath248 } 1/c_j}$ ] .
therefore , @xmath168 } \alpha_j^2 \bigl(\,\sum_{e\in \pi_j } c_{e , \pi_j}\bigr ) \geq \bigl(\,\sum_{j\in [ j ] } \frac{1}{c_j}\bigr)^{-1}\ ] ] which completes the proof of including the existence of minimizers in .
propositions [ prop : fundamental_prop_resistance ] and [ prop : fundamental_prop_conductance ] naturally lead to restricted notions of resistance and conductance obtained by limiting the optimization to only _ subsets _ of paths and cutsets , respectively . for the purpose of current paper
we will only be concerned with effective resistance . to this end , for each collection @xmath249 of finite sets of elements from @xmath250 ) , we define @xmath251 where @xmath252 and where @xmath253 is the set of all @xmath254 such that @xmath255 we refer to @xmath256 as the effective resistance _ restricted to @xmath257_. by taking suitable @xmath131 , the map @xmath258 is shown to be non - decreasing with respect to the set inclusion
. we will mostly be interested in @xmath256 when @xmath249 is a set of simple paths from @xmath20 to @xmath19 .
the following result is analogous to metric property of effective resistance .
[ lem : series_law ] let @xmath259 be collections of paths such that for any choice of @xmath260 from @xmath261 for each @xmath262 , the graph union @xmath263 contains a path between @xmath20 and @xmath19
. then @xmath264 define the edge sets @xmath265 recursively by setting @xmath266 and letting @xmath267 for @xmath268 .
let @xmath269 be a vector in @xmath270 satisfying for all @xmath271 .
for each @xmath272 and each @xmath273 , define @xmath274 by @xmath275 also for @xmath276 and @xmath277 in @xmath278 respectively , define @xmath279 notice that for any @xmath276 , @xmath280 where the first equality follows from the fact that @xmath281 for all @xmath282 and the last inequality is a consequence of .
the above definitions also immediately give @xmath283 as holds , proposition [ fundamental_cor_resistance ] with @xmath44 being the set of edges in @xmath284 yields @xmath285^{-1 } \sum_{\substack{p_j \in \mathcal p_j , \\
1 \leq j \leq k } } \prod_{1 \leq j \leq k}\rho_{j , p_j } \biggr)^{-1 } \\ & = \sum\limits_{1 \leq i \leq k}\bigl(\,\sum _ { p \in \mathcal p_i}\frac{1}{\sum_{e \in e(p)}r_{e , p}}\bigr)^{-1}\ , , \end{aligned}\ ] ] where we again used that @xmath281 in the last step .
since holds for all choices of @xmath269 satisfying , the claim follows from . a similar upper bound holds also for the effective conductance .
[ lem : parallel_law ] let @xmath286 be such that every path from @xmath20 to @xmath19 lies in @xmath287 .
then @xmath288 this is a straightforward consequence of proposition [ prop : fundamental_prop_resistance ] .
indeed , write @xmath289 as suprema of @xmath290 over @xmath135 and @xmath131 satisfying .
next bound the sum over @xmath103 by the sum over @xmath272 and the sum over @xmath291 and observe , since @xmath292 , we have @xmath293 as this holds for all @xmath135 and all admissible @xmath131 , the claim follows from .
we note ( and this will be useful later ) that , in standard treatments of electrostatic theory on graphs , the notions of effective resistance / conductance are naturally defined between subsets ( as opposed to just single vertices ) of the underlying network .
a simplest way to reduce this to our earlier definitions is by `` gluing '' vertices in these sets together .
explicitly , given two non - empty disjoint sets @xmath294 consider a network @xmath295 where all edges in @xmath296 have been removed and the vertices in @xmath116 identified as one vertex @xmath297 with all edges in @xmath94 with exactly one endpoint in @xmath116 now `` pointing '' to @xmath297 in @xmath295 and the vertices in @xmath298 similarly identified as one vertex @xmath299 . then we define @xmath300 note that , for one - point sets , @xmath301 coincides with @xmath302 , and similarly for the effective conductance .
the electrostatic duality also holds , @xmath303 .
the similarity of the two formulas and naturally leads to the consideration of self - dual situations i.e. , those in which the resistances @xmath304 can somehow be exchanged for the conductances @xmath234 .
an example of this is the network @xmath305 where the distributional identity @xmath306 makes the associated resistances @xmath307 equidistributed to the conductances @xmath308 . to formalize this situation , given a network @xmath94 we define its _ reciprocal _ @xmath309 as the network with the same underlying graph but with the resistances swapped for the conductances .
an edge @xmath95 in network @xmath309 thus has resistance @xmath310 , where @xmath304 is the resistance of @xmath95 in network @xmath94 .
[ lem : dual_couple ] let @xmath311 denote the maximum vertex degree in @xmath94 and let @xmath312 denote the maximum ratio of the resistances of any pair of adjacent edges in @xmath94 . given two pairs @xmath313 and @xmath314 of disjoint , nonempty subsets of @xmath98 , suppose that every path between @xmath116 and @xmath298 shares a vertex with every path between @xmath315 and @xmath316 . then @xmath317 the proof is based on the fact that every path @xmath103 between @xmath315 and @xmath316 defines a cutset @xmath318 between @xmath116 and @xmath298 by taking @xmath318 to be the set of all edges adjacent to any edge in @xmath103 , but not including the edges in @xmath296 . by the electrostatic duality
we just need to show @xmath319 to this end , given any @xmath320 let us pick positive numbers @xmath321 such that @xmath322 for any edge @xmath95 and any path @xmath323 , let @xmath324 , @xmath325 and @xmath326 denote the sets of all edges in @xmath327 , @xmath328 and @xmath99 that are adjacent to @xmath95 , respectively . for any @xmath329 and any @xmath330 , let @xmath331 and note that @xmath332 s are positive numbers satisfying @xmath333 for all @xmath334 . as a consequence , if we define @xmath335 then @xmath336 satisfies .
now fix a path @xmath103 in @xmath135 and compute , invoking the definitions of @xmath311 , @xmath312 and also jensen s inequality in the second step : @xmath337 hence we get @xmath338 as this holds for any choice of @xmath135 and positive numbers @xmath339 satisfying , we get as desired . a crucial fact underlying the proof of the previous lemma was that one could obtain a cut set for @xmath135 from a path @xmath103 in @xmath135 by taking union of all edges adjacent to vertices in @xmath103 . in the same setup
, we get a corresponding result also for effective conductances .
indeed , we have : [ remark_dual_couple ] for the same setting and notation as in lemma [ lem : dual_couple ] , assume that for every cutset @xmath198 between @xmath315 and @xmath316 , the subgraph induced by the set of all edges that are adjacent to some edge in @xmath198 contains a path in @xmath340 .
then @xmath341 for any cutset @xmath198 between @xmath315 and @xmath316 , let @xmath342 denote the set of all edges that are adjacent to some edge in @xmath198 . thus @xmath342 contains a path in @xmath343 by the hypothesis of the lemma . now given any @xmath344 , we pick positive numbers @xmath345 such that @xmath346 following the exact same sequence of steps as in the proof of lemma [ lem : dual_couple ] , we now find @xmath347 satisfying such that @xmath348 proposition [ fundamental_cor_resistance ] then implies @xmath349 as this holds for all choices of @xmath213 and @xmath350 satisfying , we get the desired inequality .
before we move on to the main line of the proof , we need to develop some preliminary control on the underlying gaussian fields .
the goal of this section is to amass the relevant technical claims concerning gaussian processes and , in particular , the gff .
an impatient , or otherwise uninterested , reader may consider only skimming through this section and returning to it when the relevant claims are used in later proofs .
we start by recalling , without proof , a few standard facts about general gaussian processes : [ lem : borell_ineq ] given a finite set @xmath116 , consider a centered gaussian process @xmath351 .
then , for @xmath352 , @xmath353 where @xmath354 .
[ lem : basic_chaining ] let @xmath355 be a finite metric space with @xmath356 .
suppose that there are @xmath357 such that for every @xmath358 $ ] , the @xmath359-covering number @xmath360 of @xmath361 obeys @xmath362 .
then for any @xmath363 and any centered gaussian process @xmath364
satisfying @xmath365 we have @xmath366 where @xmath367 with @xmath368 . as a consequence of lemma [
lem : basic_chaining ] we get the following result which we will use in the next subsection . [ lem : max_expectation ] let @xmath369 be squares in @xmath3 of side lengths @xmath370 respectively and let @xmath371}b_j$ ] .
there exists an absolute constant @xmath372 such that , if @xmath373 is a centered gaussian process satisfying @xmath374 then @xmath375 the following lemma , taken from @xcite , is the fkg inequality for gaussian random variables
. we will refer to this as the fkg in the rest of the paper .
[ lem : fkg_gaussian ] consider a gaussian process @xmath376 on a finite set @xmath116 , and suppose that @xmath377 then @xmath378 holds for any bounded , borel measurable functions @xmath379 on @xmath380 that are increasing separately in each coordinate . as a corollary to fkg , we get : [ cor : square_root_trick ] consider a gaussian process @xmath376 on a finite set @xmath116 such that holds .
if @xmath381 are all increasing ( or all decreasing ) , then @xmath382}{\mathbb p}(\mathcal e_i ) \geq 1 - \biggl(1 - { \mathbb p}\bigl(\bigcup_{i \in [ k]}\mathcal e_i\bigr)\biggr)^{1/k}\,.\ ] ] this is known as the `` square root trick '' in percolation literature ( see , e.g. , @xcite ) . moving to the specific example of the gff we note that one of the most important properties that makes the gff amenable to analysis is its behavior under restrictions to a subdomain .
this goes by the name gibbs - markov , or domain - markov , property . in order to give a precise statement ( which will happen in lemma [ lem : markov ] below ) we need some notations .
given a set @xmath383 , let @xmath384 denote the set of vertices in @xmath385 that have a neighbor in @xmath116 .
recall that a gff in @xmath386 with dirichlet boundary condition is a centered gaussian process @xmath387 such that @xmath388 where @xmath389 is the green function in @xmath116 ; i.e. the expected number of visits to @xmath19 for the simple random walk on @xmath3 started at @xmath20 and killed upon entering @xmath385 .
we then have : [ lem : markov ] consider the gff @xmath387 on a set @xmath386 with dirichlet boundary condition and let @xmath390 be finite . define the random fields @xmath391 and @xmath392 by @xmath393 then @xmath394 and @xmath395 are independent with @xmath396 .
moreover , @xmath395 equals @xmath397 on @xmath398 and its sample paths are discrete harmonic on @xmath298 .
this is verified directly by writing out the probability density of @xmath397 or , alternatively , by noting that the covariance of @xmath399 is @xmath400 , which is harmonic in both variables throughout @xmath298 .
we leave further details to the reader . by way of reference to the spatial scales that these fields will typically be defined over
, we refer to @xmath394 as the _ fine _ field and @xmath395 as the _ coarse _ field .
however , this should not be confused with the way their actual sample paths look like .
indeed , the samples of @xmath394 will typically be quite rough ( being those of a gff ) , while the samples of @xmath395 will be rather smooth ( being discrete harmonic on @xmath298 ) .
our next goal is to develop a good control of the smoothness of @xmath395 precisely .
a starting point is the following estimate : [ lemma-3.7 ] there is an absolute constant @xmath401 such that , given any @xmath402 with @xmath403 connected and denoting @xmath404 ^ 2\cap { \mathbb{z}}^2 \subseteq b\bigr\},\ ] ] the coarse field @xmath395 on @xmath298 obeys @xmath405 where @xmath406 denotes the length of the shortest path in @xmath403 connecting @xmath72 to @xmath407 .
let @xmath408 first be nearest neighbors and let @xmath409 .
using @xmath410 to denote the canonical inner product in @xmath411 with respect to the counting measure , the gibbs - markov property gives @xmath412 since @xmath413 is increasing ( as an operator @xmath414 ) with respect to the set inclusion , the worst case that accommodates the current setting is when @xmath116 is the complement of a single point and @xmath298 is the square @xmath415 ^ 2\cap{\mathbb{z}}^2 $ ] .
focusing on such @xmath116 and @xmath298 from now on and shifting the domains suitably , we may assume @xmath416 .
then @xmath417 where @xmath418 is the potential kernel defined , e.g. , as the limit value of @xmath419 as @xmath420 .
the relevant fact for us is that @xmath421 admits the asymptotic form @xmath422 where @xmath423 and @xmath424 is a ( known ) constant .
there is another representation of @xmath425 in terms of harmonic measures which follows from the discrete harmonicity of the coarse field .
let @xmath426 , for @xmath427 and @xmath428 , denote the harmonic measure ; i.e. , the probability that the simple random walk started from @xmath72 first enters @xmath429 at @xmath407 .
then @xmath430 where @xmath431\delta_z(\cdot).\ ] ] in order to make use of this expression
, we will need are suitable estimates for the harmonic measure : there are constants @xmath432 such that for all @xmath433 , any neighbor @xmath19 of @xmath20 and @xmath434 , from , e.g. , ( * ? ? ?
* proposition 8.1.4 ) , we have @xmath435 and @xmath436 for our special choice of @xmath116 , using we now write @xmath437\bigl[h^b(u,\tilde z)-h^b(v,\tilde z)\bigr]\bigl({\mathfrak a}(z)+{\mathfrak a}(\tilde
z)-{\mathfrak a}(z-\tilde z)\bigr ) .
\quad\end{gathered}\ ] ] since @xmath438 is a probability measure for each @xmath20 , the contribution of the terms @xmath439 and @xmath440 vanishes . for the same reason
, we may replace @xmath441 with @xmath442 in .
now we apply with the result @xmath443 invoking and , the two sums are now readily bounded by a constant independent of @xmath444 .
this gives for neighboring pairs of vertices .
for the general case we just apply the triangle inequality for the intrinsic ( pseudo)metric @xmath445^{1/2}$ ] along the shortest path in @xmath403 between @xmath20 and @xmath19 in the graph - theoretical metric . using the above variance bound , we now get : [ cor : field_smoothness1 ] for each set
@xmath446 , let us write @xmath447 for the diameter @xmath116 in the graph - theoretical metric on @xmath116 .
for each @xmath34 there are constants @xmath448 such that for all sets @xmath449 with @xmath403 connected and obeying @xmath450 ^ 2\cap { \mathbb{z}}^2 \subseteq b\bigr\ } \ge\delta\,\text{\rm diam}_{\tilde b}(\tilde b)\ ] ] and for @xmath395 denoting the coarse field on @xmath298 for the gff @xmath397 on @xmath116 , we have @xmath451 for each @xmath60 . the condition ensures , via lemma [ lemma-3.7 ] , that the variance of @xmath452 is bounded by a constant times @xmath453 with @xmath10 as in .
the assumption then ensures that this is at most a @xmath454-dependent constant .
writing this constant as @xmath455 and denoting @xmath456 lemma [ lem : borell_ineq ] gives @xmath457 it remains to show that @xmath458 is bounded uniformly in @xmath116 and @xmath298 satisfying . for this
we note that , again by lemma [ lemma-3.7 ] , an @xmath359-ball in the intrinsic metric @xmath459^{1/2}$ ] on @xmath403 contains an order-@xmath460 ball in the graph - theoretical metric on @xmath403 which itself contains an order-@xmath461 ball in the @xmath25-metric on @xmath298 .
lemmas [ lem : max_expectation ] thus applies with @xmath462 and @xmath463 .
the proof of the rsw estimates will require controlling the expectation of the gff on concentric annuli , conditional on the values of the gff on the boundaries thereof .
we will conveniently represent the sequence of these expectations by a random walk .
annulus averages and the associated random walk have been central to the study of the local properties of nearly - maximal values of the gff in @xcite .
however , there the emphasis was on estimating the probability that the random walk stays above a polylogarithmic curve for a majority of time , while here we are interested in a different aspect ; namely , the law of iterated logarithm .
the conclusions derived here will be applied in the proof of proposition [ lem : tassion_main ] .
we begin with a quantitative version of the law of the iterated logarithm for a specific class of gaussian random walks .
[ lem : finite_lil ] set @xmath464 for @xmath465 and let @xmath466 be independent random variables with @xmath467 for some @xmath468 .
let @xmath469 and suppose that there are @xmath470 and @xmath471 such that @xmath472 then there are @xmath473 , @xmath474 and @xmath475 , depending only on @xmath476 and @xmath477 , such that for all @xmath478 , the random walk @xmath479 obeys @xmath480 since @xmath481 is regularly varying at infinity with exponent @xmath482 and @xmath483 is within distance @xmath476 of a linear function , one can find @xmath484 and @xmath485 sufficiently large ( and depending only on @xmath477 and @xmath476 ) such that @xmath486 and @xmath487 hold true .
now define a sequence of random variables as @xmath488 then @xmath489 are independent with @xmath490 . then , for each @xmath491 with @xmath492 , the inequality and a straightforward gaussian tail estimate show @xmath493 for some constant @xmath494 depending only on @xmath477 and @xmath476 .
thus , whenever @xmath495 is such that @xmath496 holds true , we have @xmath497 for some @xmath498 . by independence of @xmath499 @xmath500 ,
the chebyshev inequality gives @xmath501 for some constant @xmath502 .
a computation using a gaussian tail estimate gives @xmath503 for all @xmath504 .
therefore @xmath505 for some constant @xmath506 .
on @xmath507 , gives @xmath508 and so the bounds and imply .
we will apply lemma [ lem : finite_lil ] to a special sequence of random variables which arise from averaging the gff along concentric squares . for integers @xmath59 ,
@xmath509 and @xmath510 , denote @xmath511 and , for each @xmath512 , define @xmath513 notice that we can also write @xmath514 due to the gibbs - markov property of the gff .
we then have : [ lem : ellipicity_upper ] for each integer @xmath515 as above , there are constants @xmath516 and @xmath517 such that for all @xmath59 and all @xmath518 the sequence @xmath519 ( with @xmath520 ) satisfies the conditions of lemma [ lem : finite_lil ] with these @xmath521 . since the @xmath522 s are differences of a gaussian martingale sequence , they are independent normals .
so we only need to verify the constraints on the variances . denoting @xmath523
, the gibbs - markov property of the gff implies @xmath524 recalling our notation @xmath426 for the harmonic measure , the representation @xmath525 gives @xmath526 now substitute the asymptotic form and notice that the terms arising from @xmath424 exactly cancel , while those from the error @xmath527 are uniformly bounded .
concerning the terms arising from the term @xmath528 , here we note that @xmath529 which follows by using @xmath530 to approximate the sum by an integral .
hence we get @xmath531 with @xmath532 bounded uniformly in @xmath59 , @xmath509 and @xmath533 .
using the above setup , pick two ( possibly real ) numbers @xmath534 and define @xmath535 the point of working with the conditional expectations of @xmath536 evaluated at the origin is that these expectations represent very well the typical value of the same conditional expectation anywhere on @xmath537 .
namely , we have : [ lem : anchoring ] denote @xmath538 for each @xmath510 ( and each @xmath539 as above ) there are @xmath540 and @xmath541 such that for all @xmath542 and all @xmath509 , @xmath543 denote @xmath544 and for @xmath545 abbreviate @xmath546 from the gibbs - markov property we also have @xmath547 as soon as @xmath10 is sufficiently large , the domains @xmath548 , @xmath549 and @xmath550 obey condition with some @xmath551 for all @xmath509 and all @xmath552 .
corollary [ cor : field_smoothness1 ] then gives @xmath553 for some constants @xmath554 independent of @xmath10 , @xmath495 and @xmath491 .
this shows that the oscillation of @xmath555 on @xmath537 has a uniform gaussian tail , so in order to bound @xmath556 uniformly for @xmath557 , it suffices to show that , for just one @xmath557 , also @xmath558 has such a tail .
since this random variable is a centered gaussian , it suffices to estimate its variance . here gives @xmath559
corollary [ cor : field_smoothness1 ] can now be applied with @xmath548 , @xmath560 and @xmath561 to bound the right - hand side by a constant uniformly in @xmath10 , @xmath495 and @xmath533 . combined with
, the union bound shows @xmath562 with @xmath563 independent of @xmath10 , @xmath495 and @xmath491 .
another use of the union bound now yields , thus proving the claim . in this subsection
, we estimate the cardinality of the sets of points where the gff equals ( roughly ) a prescribed multiple of its absolute maximum .
recall that from @xcite we know that the family of random variables @xmath564 is tight as @xmath420 .
the level sets we are interested in are of the form @xmath565 where @xmath566 and @xmath567 .
our conclusion about these is as follows : [ thm - levelset ] for any @xmath568 there are @xmath569 and @xmath570 such that for all @xmath571 and all @xmath572 the bound @xmath573 holds for all @xmath10 sufficiently large .
the same statement holds also for the gff on @xmath574 .
the exponent linking the cardinality of the level set to the linear size of the underlying domain has been computed in @xcite building on @xcite where the leading - order growth - rate of the absolute maximum was determined .
while much progress on the maxima of the gff has been made recently , notably with the help of modified branching random walk ( mbrw ) introduced in @xcite , the methods used in these studies do not seem to be of much use here .
indeed , in order to make use of the modified branching random walk one needs to invoke a comparison between the gff and mbrw , which is conveniently available for the maximum ( using slepian s lemma @xcite ) , but does not seem to extend to the cardinality of the level sets . another possible approach to consider is the intrinsic dimension of the level sets ( see @xcite ) , but this would not give a sharp estimate as we desire .
our approach to theorem [ thm - levelset ] is much simpler , being a combination of the second moment method ( which directly applies to gff ) and the `` sprinkling method '' which was employed in @xcite in the context of the gff .
we remark that the second moment method has recently been used to prove that a suitably - scaled size of the whole level set admits a non - trivial distributional limit @xcite .
the proof is actually quite easy when @xmath575 , but becomes more complicated in the complementary regime of @xmath576 .
this is due to well known failure of the second - moment method in these problems and the need for a suitable truncation to make it work again . the first half of the proof thus consists of the set - up , and control , of the truncation .
pick @xmath59 large and let @xmath577 . for @xmath578 , write @xmath579 and , for @xmath580 , set , abusing of our earlier notation , @xmath581 .
note that @xmath582 for all @xmath583 .
then for all @xmath584 and with @xmath423 , @xmath585 and @xmath586 hold with @xmath532 uniformly bounded in @xmath10 and @xmath587 as above .
next denote @xmath588 a straightforward calculation then shows that @xmath589 and @xmath590 again , with @xmath532 uniform in @xmath10 .
it follows that there are numbers @xmath591 with @xmath592 and a gaussian process @xmath593 which is independent of @xmath594 and obeys @xmath595 such that @xmath596 further , we have that @xmath597 again with @xmath532 uniform in @xmath10 . for @xmath598 , @xmath599 and @xmath600 ,
define the event @xmath601 + r\bigr\}\ , .
\qquad\end{gathered}\ ] ] we claim that for @xmath602 and @xmath603 sufficiently large , we have @xmath604 in order to prove , note that by @xmath605 and @xmath606 abbreviating @xmath607 + r$ ] , from these observations we have @xmath608 where the last inequality holds for all @xmath609 where @xmath610 . this yields .
now we are ready to apply the second moment method .
we will work with @xmath611 from and a calculation for the gaussian distribution we get @xmath612 for some constant @xmath494 .
our next task is a derivation of a suitable upper bound on @xmath613 . from
and we get that , for any @xmath578 and with @xmath614 a constant depending on @xmath615 but not on @xmath19 or @xmath10 , @xmath616}\ , .
\end{aligned}\ ] ] here the last inequality follows from the fact that , once we write the integral using the explicit form of the law of @xmath617 , the integrand is maximized at @xmath618 and decays exponentially when @xmath619 is away from @xmath620 .
combined with , the preceding inequality implies that @xmath621 } = o(1)\,.\ ] ] this implies @xmath622 for some @xmath569 sufficiently small uniformly in @xmath623 for some @xmath624 large .
it remains to enhance the lower bound in to a number sufficiently close to one . to this end , pick an integer @xmath444 with @xmath625 , let @xmath626 and consider a collection of boxes @xmath627 of the form @xmath628 contained in @xmath629 . for @xmath630 , @xmath631 , define the coarse fields @xmath632 by lemma [ lem : max_expectation ] and ( * ? ? ?
* lemma 3.10 ) , we get that @xmath633 in addition , as is easy to check , @xmath634 . introducing the event
@xmath635 we obtain that @xmath636 conditioning on @xmath637 and on the values @xmath638 , the gff in each square of @xmath639 are independent of each other .
further , the gaussian field on @xmath639 dominates the field obtained from subtracting @xmath640 from the gff on @xmath639 with dirichlet boundary condition on @xmath641 .
write @xmath642 by a straightforward first moment computation , we see that @xmath643 therefore , applying to @xmath639 we get that @xmath644 by conditional independence , we then get that @xmath645 combined with , it gives @xmath646 choosing @xmath444 so large that @xmath647 ( assuming that @xmath454 is sufficiently small ) , this readily gives the claim for the gff on @xmath56 with dirichlet boundary condition . in the case that the gff on @xmath648 , the same calculation goes through by considering instead
the level set restricted to the square @xmath649 and replacing @xmath650 in by @xmath6 .
we leave further details to the reader . as a final item of concern in this section
we note that , apart from the gibbs - markov property , our proofs will also make use of another decomposition of the gff which is based on a suitable decomposition of the green function
. this decomposition will be of crucial importance for the development of the rsw theory in section [ sec : rsw ] .
[ lem : field_decomp_smooth ] let @xmath651 be the gff on @xmath56 with dirichlet boundary condition .
then there are two independent , centered gaussian fields @xmath652 and @xmath653 such that the following hold : 1 .
@xmath654 a.s .
@xmath655 uniformly for all @xmath656 .
3 . @xmath657 uniformly for all @xmath658 such that @xmath659 .
the distribution of @xmath660 is invariant under reflections and rotations that preserve @xmath56 . throughout the proof of the current lemma , we let @xmath661 be the _ lazy _ discrete - time simple symmetric random walk on @xmath2 that , at each time , stays put at its current position with probability @xmath482 , or transitions to a uniformly chosen neighbor with the complementary probability . we denote by @xmath662 the law of the walk with @xmath663 and
write @xmath664 to denote the expectation with respect to @xmath662 .
let @xmath665 be the first hitting time to the boundary @xmath89 .
it is clear that @xmath666 in addition , thanks to laziness of @xmath667 , the matrix @xmath668 is non - negative definite for each @xmath669 .
therefore , there are independent centered gaussian fields @xmath670 and @xmath671 such that @xmath672 and @xmath673 at this point , it is clear that we can couple the processes together so that property ( a ) holds .
property ( b ) holds by crude computation which shows @xmath674 it remains to verify property ( c ) . for any @xmath675 and @xmath676
, we have that @xmath677 since @xmath678 ( see , e.g. , ( * ? ? ?
* exercise 2.2 ) ) , the first term on the right hand side is bounded by @xmath679 .
the second term is @xmath680 by ( * ? ? ?
* theorem 4.4.6 ) and the fact that @xmath681 .
this completes the verification of property ( c ) .
having dispensed with preliminary considerations , we now ready to develop a rsw theory for effective resistances across rectangles .
throughout we write , for @xmath682 , @xmath683 \times [ -m , m ] \bigr)\cap{\mathbb{z}}^2\ ] ] for the rectangle of @xmath684 vertices centered at the origin .
recall that @xmath685 .
the principal outcome of this section are corollary [ cor-4.3 ] and proposition [ prop-4.7 ] . in corollary
[ cor-4.18 ] , these yield the proof of one half of theorem [ thm - effective - resistance ] . the proof of the other half comes only at the very end of the paper ( in section [ sec - proofs ] ) . in bernoulli percolation ,
the rsw theory is a loose term for a collection of methods for extracting uniform lower bounds on the probability that any rectangle of a given aspect ratio is crossed by an occupied path along its longer dimension .
the starting point is a duality - based lower bound on the probability of a left - right crossing of a square . in the present context
, the crossing probability is replaced by resistance across a square and duality by consideration of a reciprocal network .
an additional complication is that our problem is intrinsically spatially - inhomogeneous .
this means that all symmetry arguments , such as rotations and reflections , require special attention to where the underlying domain is located .
in particular , it will be advantageous to work with the gff on finite squares instead of the pinned field in all of @xmath3 . if @xmath686 is a rectangular domain in @xmath3 , we will write @xmath687 , @xmath688 , @xmath689 and @xmath690 to denote the sets of vertices in @xmath691 that have a neighbor in @xmath692 to the left , down , right and up of them , respectively .
( notice that , unlike @xmath693 , these `` boundaries '' are subsets of @xmath686 . ) given any field @xmath694 recall that @xmath695 denotes the network on @xmath686 associated with @xmath696 .
we then abbreviate @xmath697 and @xmath698 our first estimate concerning these quantities is : [ prop : zero_mean ] let @xmath699 denote the gff on @xmath700 with dirichlet boundary conditions .
there is @xmath701 and for each @xmath598 there is @xmath702 such that for all @xmath542 and all @xmath703 , @xmath704 the same result holds also for @xmath705 , which is equidistributed to @xmath706 .
the proof requires some elementary observations that will be useful later as well : [ lem : resistance_decomp ] let @xmath116 be a finite subset of @xmath3 and @xmath707 , @xmath708 be two random fields on @xmath116 .
then for any @xmath709 we have , @xmath710 furthermore , @xmath711 and @xmath712 let @xmath105 be a unit flow from @xmath20 to @xmath19
. then implies @xmath713 ^ 2 { \mathrm{e}}^{-\gamma(\chi_{1 , u ' } + \chi_{1 , v ' } ) } { \mathrm{e}}^{-\gamma(\chi_{2 , u ' } + \chi_{2 , v'})}\,.\ ] ] hereby follows by bounding the second exponential by its maximum over all pairs of nearest neighbors in @xmath116 and optimizing over @xmath105 .
the estimate is obtained similarly ; just take the conditional expectation before optimizing over @xmath105 .
the proof of exploits the similarity between and and is thus completely analogous .
our aim is to use the fact that , in any gaussian network , the resistances are equidistributed to the conductances .
we will apply this in conjunction with the estimate in lemma [ remark_dual_couple ] .
unfortunately , this estimate requires a hard bound on the maximal ratio of resistances at neighboring edges .
these ratios would be undesirably too large if we work with the gff network directly ; instead we will invoke the decomposition of @xmath714 into the sum of gaussian fields @xmath715 and @xmath716 as stated in lemma [ lem : field_decomp_smooth ] and apply lemma [ remark_dual_couple ] to the network associated with @xmath717 only .
we begin by estimating the oscillation of @xmath717 across neighboring vertices . from property ( c ) in the statement of lemma [ lem : field_decomp_smooth ] and a standard bound on the expected maximum of centered gaussians
, we first get @xmath718 using this bound and property ( c ) , lemma [ lem : borell_ineq ] shows that for each @xmath598 there is @xmath719 such that for all @xmath59 , @xmath720 now observe that the pairs @xmath721 and @xmath722 satisfy the conditions of lemma [ remark_dual_couple ] .
using @xmath723 to denote the top - to - bottom resistance in the reciprocal network , combining with the last display yields @xmath724 a key point of the proof is that , since the law of @xmath725 is symmetric with respect to rotations of @xmath700 , the fact that @xmath726 implies @xmath727 the union bound then shows @xmath728 lemma [ lem : resistance_decomp ] and the independence of @xmath729 and @xmath717 now give @xmath730 lemma [ lem : field_decomp_smooth ] shows @xmath731 for some constant @xmath732 and so the maximum on the right of is at most @xmath733 .
taking @xmath734 , the desired bound now follows ( for @xmath10 sufficiently large ) from ( [ e:4.10][e:4.11 ] ) and markov s inequality . with only a minor amount of additional effort , we are able to conclude a uniform _ lower _ bound for the resistance across rectangles . [ cor-4.3 ]
let @xmath735 be as in proposition [ prop : zero_mean ] .
for each @xmath598 there is @xmath736 such that for all @xmath737 , all @xmath738 and all translates @xmath686 of @xmath739 contained in @xmath740 , we have @xmath741 the same applies to @xmath742 for any translate @xmath686 of @xmath743 contained in @xmath740 . replacing effective resistances by effective conductances in the proof of proposition [ prop : zero_mean ] ( and relying on lemma [ lem : dual_couple ] instead of lemma [ remark_dual_couple ] ) yields @xmath744 for all @xmath542 .
since @xmath745 this bound extends to the rectangle @xmath739 . now consider a translate @xmath686 of this rectangle that is contained in @xmath740 .
taking @xmath746 and let @xmath747 be the translate of @xmath748 that is centered at the same point as @xmath686 . considering the gibbs - markov decomposition into a fine field @xmath749 and a coarse field @xmath750 on @xmath747 , we then get @xmath751 since @xmath686 and @xmath747 are centered at the same point , the first probability is at least @xmath752 by our extension of to rectangles .
the second probability can be made arbitrarily small uniformly in @xmath10 by taking @xmath753 large .
the claim follows . despite our convention that constants such as @xmath754 , etc may change meaning line to line
, the constant @xmath735 will denote the quantity from proposition [ prop : zero_mean ] throughout the rest of this paper .
as noted already in the introduction , our approach to the rsw theory is strongly inspired by @xcite which is itself based on inductively controlling the crossing probability ( in bernoulli percolation ) between @xmath755 and a _ portion _ of @xmath756 .
we will now setup the relevant objects and notations and prove estimates that will later serve in an argument by contradiction . for the square @xmath56 and @xmath757\cap{\mathbb{z}}$ ] with @xmath758
, consider the subset of @xmath756 defined by @xmath759}b(n):= ( \{n\}\times[\alpha,\beta])\cap{\mathbb{z}}^2\ ] ] let @xmath760}$ ] denote the set of paths in @xmath56 that use only the vertices in @xmath761)\cap{\mathbb{z}}^2 $ ] except for the initial vertex , which lies in @xmath755 , and the terminal vertex , which lies in @xmath762}b(n)$ ] . with these notions in place
, we now introduce the shorthand @xmath763,\chi}:=r_{b(n)_\chi}\bigl(\mathcal p_{n;[\alpha,\beta]}\bigr ) = r_{b(n)_\chi}\bigl({\partial_\mathrm{left}}b(n),{\partial_\mathrm{right}}^{[\alpha,\beta]}b(n)\bigr).\ ] ] our first goal is to define a quantity @xmath764 which will mark , in rough terms , the point of transition of @xmath765,\chi_{2n}}$ ] from large to small values .
an illustration of the geometric setting underlying the definition of the restricted effective resistance @xmath766,\chi}$ ] in .,scaledwidth=45.0% ] 0.0 cm we first need a couple of simple observations . note that @xmath767}\cup \mathcal p_{n;[-n,0]}$ ] includes all paths starting on @xmath755 and terminating on @xmath756 .
lemma [ lem : parallel_law ] then shows @xmath768,\chi_{2n}}}+\frac1{r_{n,[-n,0],\chi_{2n}}}\ ] ] while the symmetry of both the law of @xmath769 and the square @xmath56 with respect to the reflection through the @xmath72 axis implies @xmath770,\chi_{2n } } { \overset{\text{\rm law}}=}r_{n,[-n,0],\chi_{2n}}$ ] . by proposition [ prop : zero_mean ] , there is @xmath771 such that @xmath772 as soon as @xmath542 .
the square - root trick in corollary [ cor : square_root_trick ] then shows @xmath773,\chi_{2n } } > 2{\mathrm{e}}^{\hat c\log\log(2n)}\bigr ) \leq \sqrt{2/3 } < 0.82\ ] ] as soon as @xmath542 .
next we note that , by lemma [ lemma-3.7 ] , @xmath774 hence , there is @xmath775 such that @xmath776 obeys @xmath777 for all @xmath59 .
now set @xmath778 , define @xmath779 $ ] by @xmath780,\chi_{2n}}>(4+c_1)\,{\mathrm{e}}^{\hat c\log\log(2n)}\bigr)\ ] ] and , noting that @xmath781 is non - decreasing with @xmath782 ( cf ) , let @xmath783 this definition implies the following inequalities : [ lem : tassion1 ] for @xmath784 as in , define @xmath785 and let @xmath786 , @xmath771 and @xmath787 be as above . then the following two properties hold for all @xmath542 : 1
. for all @xmath788 , @xmath789,\chi_{2n } } \leq 5c_2{\mathrm{e}}^{\hat c\log\log(2n)}\bigr ) \geq 0.005.\ ] ] 2 .
if @xmath790 , then for all @xmath791 , @xmath792,\chi_{2n } } \geq ( 4 + c_1)\,{\mathrm{e}}^{\hat c\log \log ( 2n)}\bigr ) > 0.99\ ] ] and @xmath793,\chi_{2n } } \leq 4{\mathrm{e}}^{\hat c\log\log(2n)}\bigr ) \geq 0.17\,.\ ] ] we begin with ( p1 ) .
since @xmath794 for @xmath795 , for all such @xmath576 we have @xmath796,\chi_{2n}}\leq ( 4 + c_1)\,{\mathrm{e}}^{\hat c\log\log(2n ) } ) \geq 0.01\,.\ ] ] in order to deal with @xmath797 , we will will need : [ lemma - reroute ] for @xmath776 and @xmath19 being the point with coordinates @xmath798 , we have @xmath799,\chi_{2n } } > & \,c_1 r_{n,[\alpha_n-1,n],\chi_{2n}}\bigr\ } \\ & \subseteq \bigl\{\max\bigl\{\chi_{v - e_2 } - \chi_{v + e_1 } , \chi_{v - e_2 + e_1 } + \chi_{v - e_2 } - \chi_{v } - \chi_{v + e_1}\bigr\}\ge c'\bigr\}. \end{aligned}\ ] ] deferring the proof of this lemma until after this proof , we now combine for @xmath800 with to get @xmath801,\chi_{2n}}\le \,(4+c_1)c_1{\mathrm{e}}^{\hat c\log\log(2n)}\bigr ) \\ & \qquad\ge { \mathbb p}\bigl(r_{n,[\alpha_n-1,n],\chi_{2n}}\le ( 4+c_1)\,{\mathrm{e}}^{\hat c\log\log(2n)},\,\,r_{n,[\alpha_n , n],\chi_{2n } } \le c_1r_{n,[\alpha_n-1,n]}\bigr ) \\ & \qquad\geq 0.01 - 0.005 = 0.005\ , .
\end{aligned}\ ] ] since @xmath802 , the bound holds for @xmath803 as well .
thanks to the upward monotonicity of @xmath804,\chi_{2n}}$ ] , the inequality then extends to all @xmath805 .
the first inequality in @xmath806 evidently holds by our choice of @xmath764 . as for the second inequality ,
lemma [ lem : parallel_law ] shows @xmath807,\chi_{2n}}}\le\frac1{r_{n,[0,\alpha],\chi_{2n}}}+\frac1{r_{n,[\alpha , n],\chi_{2n}}}\ ] ] and this then implies @xmath808,\chi_{2n } } \leq 2{\mathrm{e}}^{\hat c\log\log(2n ) } , r_{n,[\alpha , n],\chi_{2n } } > ( 4 + c_1)\,{\mathrm{e}}^{\hat c\log\log(2n)}\bigr\ } \\
\subseteq \bigl\{r_{n,[0,\alpha],\chi_{2n } } \leq 4{\mathrm{e}}^{\hat c\log\log(2n)}\bigr\}. \qquad\end{gathered}\ ] ] invoking and the definition of @xmath764 , the probability of the event on the right is than at most @xmath809 .
we still owe to the reader : suppose @xmath696 is such that the complementary event to that on the right of occurs .
we will show that then the complement of the event on the left occurs as well .
for this , let @xmath105 be the optimal flow realizing the effective resistivity in and let @xmath106 denote its value on edge @xmath107 . to reduce clutter of indices , write @xmath810 for the resistance of edge @xmath107 .
abbreviate @xmath811 , @xmath812 and @xmath813 .
our aim is to reroute @xmath814 through @xmath20 to @xmath815 .
define a flow @xmath816 by setting @xmath817 , @xmath818 and @xmath819 and letting @xmath820 for all other edges @xmath95 .
the only edges where @xmath821 might expend more energy than @xmath105 are the edges @xmath822 and @xmath823 . to bound the change in energy , we note @xmath824 ^ 2 \\ & \leq 2r(v , u)\theta(v , u)^2 + 2r(v , t)\,{\mathrm{e}}^{c'\gamma}\theta(v , t)^2 \end{aligned}\ ] ] with the second inequality due to the containment in the complement of the event on the right of .
similarly we have @xmath825 hence we get @xmath826,\chi_{2n } } \leq ( 2 + 4{\mathrm{e}}^{c'\gamma})r_{n , [ \alpha_n , n],\chi_{2n } } = c_1 r_{n , [ \alpha_n , n],\chi_{2n}}$ ] , thus proving .
we now move to bounds on resistance across rectangular domains . as in bernoulli percolation ,
a fundamental tool in this endeavor is the fkg inequality which , in our case , will be used in the following form : [ lem : resistance_chaining1 ] consider a finite @xmath827 and a gaussian process @xmath828 with @xmath829 for all @xmath830 .
suppose that @xmath831 are collections of paths in @xmath686 that satisfy the conditions of lemma [ lem : series_law ] for a pair of disjoint subsets @xmath313 of @xmath686 .
then for any @xmath832 , we have @xmath833 this is an immediate consequence of lemma [ lem : series_law ] , the monotonicity of @xmath834 in individual edge resistances , and the fkg inequality in lemma [ lem : fkg_gaussian ] .
the principal outcome of this subsection is : [ lem : tassion2 ] there are @xmath835 such that for all @xmath542 for which @xmath836 holds , all @xmath837 and any shift @xmath686 of @xmath739 satisfying @xmath838 , @xmath839 the same applies to @xmath742 for any shift @xmath686 of @xmath743 that obeys @xmath838 . by proposition [ prop : zero_mean ]
the bound holds for left - to - right resistance of centered squares .
we will employ a geometric argument combined with the fkg inequality to extend the bound from squares to rectangular domains .
the main technical tool is lemma [ lem : series_law ] which , in a sense , permits us to bound resistance by path - connectivity considerations only .
we will actually use a different argument depending on whether @xmath764 equals , or is less than @xmath840 .
here we will need the bound , but for the underlying domain not necessarily centered at the box which defines the underlying field .
thus , for @xmath686 a translate of the square @xmath56 such that @xmath838 , let @xmath841,\chi_m}$ ] denote the quantity corresponding to @xmath766,\chi_m}$ ] for the square @xmath686 and the underlying field given by @xmath714 . in light of , corollary [ cor : field_smoothness1 ] and lemma [
lem : resistance_chaining1 ] show that , for some constant @xmath842 depending only on @xmath787 and @xmath843 , @xmath844,\chi_m } \leq c_3'{\mathrm{e}}^{\hat c\log\log m}\bigr)\ge0.001\ ] ] holds for all @xmath542 , all @xmath837 and all squares @xmath686 as above that are contained in @xmath845 .
thanks to invariance of the law of @xmath714 under rotations of @xmath700 , the same bound holds also for the `` rotated '' quantities ; namely , those dealing with up - down resistivities . now let @xmath686 be a translate by @xmath846 of the rectangle @xmath739 such that @xmath838 and let us regard @xmath686 as the union of the squares @xmath847 for each @xmath848 , consider the following collections of paths : first , let @xmath261 be the set of all paths in @xmath849 that cross @xmath849 left to right ( with only the initial and terminal point visiting the left and right boundaries of @xmath849 )
. then ( referring to parts of the boundary as if @xmath849 were the square @xmath56 ) , let @xmath850 be the collection of paths that connects the bottom of the square to the @xmath851 $ ] portion of the top boundary , and let @xmath852 be the path between the bottom of the square to the @xmath853 $ ] portion of the top boundary .
the key point ( implied by the fact that @xmath854 ) is now that , for any choice of paths @xmath855 , @xmath856 and @xmath857 and any @xmath858 , the graph union of the triplet of paths @xmath859 is connected and , for each @xmath860 , the graph union of @xmath859 is connected to the graph union of @xmath861 ; see fig .
[ fig6 ] .
the setting of the proof of proposition [ lem : tassion2 ] , case @xmath854 .
the collection of paths shown suffices to ensure a left - to right crossing through the four shown translates of @xmath56 .
the key points to observe are that @xmath260 intersects both @xmath862 and @xmath863 while @xmath863 intersects @xmath864 , for each @xmath271.,scaledwidth=98.0% ] 0.0 cm it follows that the graph union of the seven triplets of paths contains a left - to - right crossing of the rectangle @xmath686 and , by lemma [ lem : series_law ] , we thus get @xmath865 in light of the definition ( and , for simplicity of computation , restricting @xmath261 to paths that terminate only at the top @xmath853 $ ] portion of the right boundary ) , and the fkg inequality now give with @xmath866 and @xmath867 . here , in addition to which ,
as before , we bring to the form , we will also need this is why we need @xmath868 which we extend using corollary [ cor : field_smoothness1 ] and lemma [ lem : resistance_chaining1 ] to the form @xmath869,\chi_m } \leq c_3''{\mathrm{e}}^{\hat c\log\log m}\bigr)\ge0.01\ ] ] for some @xmath870 , all @xmath542 and all translates @xmath686 of @xmath56 such that @xmath838 .
the same bound holds also for all rotations and reflections of these quantities .
an illustration of the geometric setting underlying the key argument in the proof of proposition [ lem : tassion2 ] , case @xmath790 . here
@xmath871 and @xmath872 .
examples of paths @xmath873 , @xmath874 , @xmath875 and @xmath876 are shown in black .
together with any choice of paths @xmath877 , @xmath878 and @xmath879 ( shown in gray ) , these enforce a left - to - right crossing of the rectangle.,scaledwidth=65.0% ] 0.0 cm abbreviate @xmath871 and note that @xmath880 for @xmath10 large enough .
let us first deal with @xmath686 being a translate of the rectangle @xmath881\times[-n , n])\cap{\mathbb{z}}^2 $ ] by some @xmath846 subject to the restriction @xmath882 .
consider the squares @xmath883 and @xmath884 ^ 2\cap{\mathbb{z}}^2\ ] ] and note that @xmath885 and @xmath886 ; see fig .
define the following collections of paths : first , let @xmath887 be all paths in @xmath888 from the left side to the @xmath889 $ ] portion of the right side .
similarly , let @xmath890 be all paths in @xmath891 from the @xmath889 $ ] portion of the left side to the right side of @xmath891 .
next we define the following collections of paths in @xmath892 : 1 .
the set @xmath893 of all paths from the top to the bottom sides of @xmath892 , 2 .
the set @xmath894 of all paths from the left side of @xmath892 to the @xmath895 $ ] portion of the right side , 3 .
the set @xmath896 of all paths from the left side of @xmath892 to the @xmath897 $ ] portion of the right side , 4 .
the set @xmath898 of all paths from the @xmath895 $ ] portion of the left side of @xmath892 to the right side , and 5 .
the set @xmath899 of all paths from the @xmath897 $ ] portion of the left side of @xmath892 to the right side .
the key point is that , thanks to the assumption @xmath872 , for any choice of paths @xmath855 , the graph union of these paths will contain a left - to - right path crossing @xmath686 ; see fig .
[ fig6 ] . by lemma [
lem : series_law ] , @xmath900 where @xmath901 . from here
we get for all @xmath902 rectangles @xmath903 with @xmath904 and @xmath905 . in order to prove the desired claim ,
consider a translate @xmath686 of @xmath739 by @xmath846 entirely contained in @xmath906 and note that , letting @xmath907 , and we can cover @xmath686 by the family of rectangles @xmath908 and @xmath909 defined as follows : @xmath910\times[-n , n]\bigr)\cap{\mathbb{z}}^2,\quad j=0,\dots , k,\ ] ] where @xmath911 for all @xmath912 and @xmath913e_1 $ ] , which ensures that all @xmath914 lie inside @xmath686 ( and thus inside @xmath845 ) , and @xmath915\times[0,2(2n - k)]\bigr)\cap{\mathbb{z}}^2,\quad j=1,\dots , k-1,\ ] ] where @xmath916 are such that all @xmath917 lie in @xmath906 ( this is possible because @xmath918 ) and such that @xmath919 for each @xmath920 .
assuming each @xmath921 and @xmath917 contains a path connecting the shorter sides of the rectangle , the graph union of these paths then contains a left - to - right crossing of @xmath686 .
lemma [ lem : series_law ] then gives @xmath922 in light of our earlier proof of for rectangles of dimensions @xmath923 , we get for @xmath924 rectangles as well with @xmath925 and @xmath926 .
it appears that proposition [ lem : tassion2 ] could be more than sufficient for proving uniform upper bound on resistance across rectangles , provided we can somehow guarantee that @xmath927 does not grow faster than exponentially with @xmath10 .
this is the content of : [ lem : tassion_main ] for each @xmath928 and each @xmath929 , there exists an integer @xmath930 such that if , for some @xmath59 , @xmath931 holds all translates or rotates @xmath686 of @xmath739 contained in @xmath932 , then we have @xmath933 for at least one @xmath934 .
the proof will be based on the following lemma : [ lem : tassion3 ] suppose that , for some @xmath835 and some @xmath59 , holds for all translates and rotates of @xmath739 contained in @xmath932 .
there are @xmath935 and @xmath936 , depending only on @xmath424 and @xmath937 , respectively , such that whenever @xmath938 is such that @xmath939 and @xmath940 , @xmath941 holds for all translates and rotates of @xmath942 contained in @xmath943 . the geometric setup for the proof of lemma [ lem : tassion3 ] .
the graph union of paths @xmath944 contains a left - to - right crossing of the @xmath945-rectangle.,scaledwidth=85.0% ] 0.0 cm we will first prove this for rectangles @xmath686 of the form @xmath946 .
consider the squares @xmath947 ^ 2\cap{\mathbb{z}}^2 $ ] and @xmath948 and let @xmath949 be the four maximal rectangles of dimensions @xmath950 , labeled counterclockwise starting from the one at the bottom , contained in the annulus @xmath951 .
let @xmath952 be a path in @xmath888 connecting the left - hand side to the @xmath953 $ ] portion of the right - hand side and , similarly , @xmath954 is the path in @xmath891 connecting the @xmath953$]-portion of the left - hand side to the right hand side .
let @xmath955 be paths ( in @xmath949 , respectively ) between the shorter sides of @xmath949 , respectively .
then the assumption @xmath956 implies that the graph union of @xmath944 contains a path in @xmath686 connecting the left side to the right side ; see fig .
[ fig8 ] . combining
with ( in which @xmath10 is replaced by @xmath957 ) , we get the claim for @xmath686 with @xmath958 and @xmath959 . to extend this to rectangles @xmath686 of the form @xmath942 , we note that these can be covered by four translates and two rotates of @xmath946 such that the existence of a crossing between the shorter sides in each of these rectangles forces a crossing of @xmath686 .
thanks to lemma [ lem : resistance_chaining1 ] , the desired bound then holds for @xmath686 as well ; we just need to multiply the above @xmath936 by @xmath960 and raise the above @xmath935 to the sixth power .
we are now ready to give : the proof is by way of contradiction ; indeed , we will prove that if such @xmath961 does not exist , then we will ultimately violate the first inequality in ( p2 ) in lemma [ lem : tassion1 ] for a sufficiently large square .
this will be done by showing that a path from the left side of the square @xmath962 to the @xmath963 $ ] part of the right side can be re - routed to instead terminate in the @xmath964$]-part of the right side .
the re - routing will be achieved by showing existence of a path winding around an annulus of inner `` radius '' at least @xmath965 centered at the point @xmath966 .
we will focus on @xmath961 of the form @xmath967 , where @xmath968 and @xmath509 .
fix such an @xmath495 ( and thus @xmath961 ) and , for @xmath583 , let @xmath969 .
consider also the annulus @xmath970 and define the conditional field @xmath971 by the gibbs - markov property of the gff , @xmath972 has the law of the values on @xmath537 of the gff in @xmath973 with dirichlet boundary condition .
let @xmath974 denote the sum of the resistances between the shorter sides of the four maximal rectangles contained in @xmath537 , in the field @xmath975 .
assuming @xmath976 , lemma [ lem : tassion3 ] in conjunction with corollary [ cor : field_smoothness1 ] and lemma [ lem : resistance_chaining1 ] show that , for some @xmath977 and @xmath978 : @xmath979 let @xmath980 be the smallest integer such that @xmath981 , let @xmath787 be as in the first inequality in ( p2 ) in lemma [ lem : tassion1 ] and let @xmath982 be the constant from lemma [ lem : ellipicity_upper ] .
define @xmath983 lemma [ lem : ellipicity_upper ] ( dealing with the lil for the sequence @xmath984 ) and lemma [ lem : anchoring ] ( dealing with the deviations @xmath985 ) tell us that there is a positive integer @xmath986 satisfying @xmath987 putting together , , the choices of @xmath980 and @xmath988 along with lemmas [ lem : anchoring ] and [ lem : resistance_decomp ] we get for all @xmath10 such that @xmath989 , @xmath990 where @xmath991 . the geometric setting for a key argument in the proof of of proposition [ lem : tassion_main ] : once @xmath764 is less than the inner radius of the depicted annulus , @xmath992})$ ] is bounded by @xmath993})$ ] plus the sum of the resistances between the shorter sides of the four maximal rectangles contained in the annulus . ,
scaledwidth=65.0% ] 0.0 cm we are now ready to derive the desired contradiction .
lemma [ lem : series_law ] gives us that if @xmath994 for all @xmath995 , then @xmath996 } ) \leq r_{b(n')_{\chi_{n'}}}(\mathcal p_{n ' ; [ 0 , \alpha_{n ' } ] } ) + c_1{\mathrm{e}}^{\hat c\log\log n}\bigr ) \\
\geq 1 - 0.03 = 0.97\,.\ ] ] from the second inequality in ( p2 ) in lemma [ lem : tassion1 ] we have @xmath997 } ) \leq 4{\mathrm{e}}^{\hat c\log\log n'}\bigr ) \geq 0.17\,.\ ] ] the last two displays and the fkg imply @xmath996 } ) \leq ( 4 + c_1)\,{\mathrm{e}}^{\hat c\log\log n'}\bigr ) > 0.17 \times 0.97 > 0.16\,.\ ] ] in contradiction with the first inequality in ( p2 ) in lemma [ lem : tassion1 ] .
the claim follows . as a consequence of the above arguments , we are now ready to state our first _
unrestricted _ general upper bound on the effective resistance across rectangles : [ prop-4.7 ] there are constants @xmath998 and @xmath999 such that for all @xmath1000 , all @xmath1001 and for every translate @xmath686 of @xmath739 contained in @xmath845 , we have @xmath1002
the same applies to @xmath742 for translates @xmath686 of @xmath743 with @xmath838 .
we begin by showing that holds ( with the same constants ) along an exponentially growing sequence of @xmath10 .
this is where proposition [ lem : tassion2 ] and proposition [ lem : tassion_main ] come together .
[ lemma-4.7 ] let @xmath424 and @xmath937 be as in proposition [ lem : tassion2 ] .
there is @xmath401 and an increasing sequence @xmath1003 of positive integers such that , for each @xmath1004 , we have @xmath1005 and the bound @xmath1006 holds for all translates @xmath686 of @xmath1007 contained in @xmath1008
. we will construct @xmath1003 by induction .
suppose that @xmath1009 have already been defined .
since holds for @xmath1010 , proposition [ lem : tassion_main ] shows the existence of an @xmath1011 $ ] with @xmath1012 .
define a sequence @xmath1013 by @xmath1014 and @xmath1015 and note that @xmath1016 for some numerical constant @xmath732 .
now if @xmath1017 is true for @xmath1018 , then @xmath1019 the fact that @xmath1020 implies that this must fail once @xmath282 is sufficiently large ; i.e. , for some @xmath1021 , where @xmath1022 depends only on @xmath1023 .
we thus let @xmath1024 be the smallest such that @xmath1025 and set @xmath1026 . then holds by the inequality on the right of and the fact that @xmath1027 .
the bound is implied by proposition [ lem : tassion2 ] . to start the induction
, we just take the above sequence @xmath1028 with @xmath1029 and find the first index @xmath282 for which @xmath1025 .
then we set @xmath1030 and argue as above .
from here we now conclude : let @xmath1031 be the sequence from lemma [ lemma-4.7 ] . invoking corollary [ cor : field_smoothness1 ] and lemma [ lem : resistance_chaining1 ]
, the bound shows that , for each @xmath1032 and any translate @xmath686 of @xmath1007 contained in @xmath845 , @xmath1033 holds with some constants @xmath1034 independent of @xmath491 and @xmath444 . by invariance of the law of @xmath699 with respect to rotations of @xmath700 ,
the same holds for the resistance @xmath742 for all rotations of @xmath1007 contained in @xmath845 .
now pick @xmath623 and let @xmath491 be such that @xmath1035 . for @xmath1036 , consider a translate @xmath686 of @xmath739 contained in @xmath845 .
let @xmath1037 ; by this @xmath980 is bounded uniformly in @xmath491 .
we then find rectangles @xmath849 , @xmath1038 that are translates of @xmath1007 such that @xmath1039 for each @xmath1040 and are centered along the same horizontal line as @xmath686 and positioned in such a way that they all lie inside @xmath845 .
next we find translates @xmath1041 of @xmath1042 such that @xmath1043 , which is a translate of @xmath56 , is contained in @xmath914 for each @xmath1040 .
we can again position these so that @xmath1044 for each @xmath271 .
it is clear from the construction that if , for each @xmath1045 , we are given a path in @xmath849 and , for each @xmath1040 , a path in @xmath914 and these paths connect the shorter sides of the rectangle they lie in , then the graph union of all these paths contains a path in @xmath686 between the left side and right side thereof .
lemma [ lem : series_law ] then gives @xmath1046 all of the rectangles lie in @xmath845 and so applies to the resistivities on the right of .
lemma [ lem : resistance_chaining1 ] then readily gives with @xmath1047 and @xmath1048 .
in addition to resistance across rectangles , the proofs in section [ sec-5 ] will also require an lower bound for resistances across annuli . for @xmath1049 , let @xmath1050 and denote @xmath1051 note that @xmath1052 as well as @xmath1053 .
we have : [ lemma - annuli ] there @xmath1054 such that for all @xmath10 sufficiently large and @xmath1055 , @xmath1056 let @xmath1057 denote the four maximal rectangles contained in @xmath116 .
we assume that the rectangles are labeled clockwise starting from the one on the right .
now observe that every path in @xmath116 from @xmath1058 to @xmath1059 contains a path that is contained in , and connects the longer sides of , one of the rectangles @xmath1057 .
it follows that @xmath1060 the claim will follow from the fkg inequality if we can show that , for some @xmath1061 and @xmath1062 , @xmath1063 holds for all translates @xmath686 of @xmath1064\times[0,4n])\cap{\mathbb{z}}^2 $ ] contained in @xmath1065 and all @xmath10 sufficiently large .
( indeed , then @xmath1066 and @xmath1067 . )
we will show this using the duality in lemma [ lem : dual_couple ] , but for the we will first need to invoke the decomposition @xmath1068 from lemma [ lem : field_decomp_smooth ] .
first , for any @xmath1069 , @xmath1070 passing over to conductances , from lemma [ lem : resistance_decomp ] we then get , as before , @xmath1071 while the duality in lemma [ lem : dual_couple ] gives , as in the proof of proposition [ prop : zero_mean ] , @xmath1072 finally , we use lemma [ lem : resistance_decomp ] one more time to get @xmath1073 if we set @xmath1074 , proposition [ prop-4.7 ] bounds the first probability below by @xmath1075 . now take @xmath1076 for @xmath315 large and work your way back to get . in order to get the tail estimate on the effective resistance in theorem [ thm - effective - resistance ] , we need to invoke a concentration - of - measure argument for the quantity at hand .
recall the notation @xmath1077 for the effective resistance in network @xmath1078 restricted to the collection of paths in @xmath135 .
[ lem : gaussian_conc ] suppose @xmath696 is a gaussian field on @xmath56 with @xmath1079 for all @xmath1080 and @xmath935 independent of @xmath10 .
let @xmath1081 be a subnetwork of @xmath1082 and let @xmath135 be a finite collection of paths within @xmath116 between some given source and destination .
there is a constant @xmath1083 such that for all @xmath59 , all @xmath60 and all @xmath0 , @xmath1084 for the proof , we will need : [ lem : lipschitz ] let @xmath116 be a subnetwork of @xmath56 and @xmath135 be a finite collection of paths within @xmath116 between some given source and destination .
let @xmath1085 be defined by @xmath1086 where @xmath1087 is the set of all @xmath1088 such that @xmath1089 then @xmath1090 is a lipschitz function relative to the @xmath1091 norm on @xmath1092 with lipschitz constant @xmath1093 .
define a new real - valued function , also denoted by @xmath1090 , on @xmath1094 via @xmath1095 then for any @xmath1096 and @xmath1097 it is clear that @xmath1098 hence @xmath1099 is @xmath1093- lipschitz relative to the @xmath1091 norm as well .
this follows directly from the gaussian concentration inequality ( see @xcite ) and lemma [ lem : lipschitz ] .
we are now ready to give a version of the upper bound in theorem [ thm - effective - resistance ] , albeit for a network arising form a gff on a finite subset of @xmath3 : [ prop : thm1.4 ] there is @xmath1100 depending only on @xmath71 and a constant @xmath1101 such that @xmath1102 holds for all @xmath1103 , all @xmath1104 and all @xmath1105 .
combining proposition [ prop-4.7 ] with corollary [ cor-4.3 ] , for each @xmath598 there is @xmath1106 such that if @xmath1107 , @xmath1108 and @xmath686 is a translate of @xmath739 contained in @xmath845 , then we have @xmath1109 decomposing @xmath1110 on @xmath700 into a fine field @xmath1111 and a coarse field @xmath1112 , the fact that @xmath1113 along with @xmath1114 shows @xmath1115 the last probability tends to zero as @xmath1116 uniformly in @xmath433 and so , by choosing @xmath753 large , there is a constant @xmath1117 such that , for all @xmath1107 , @xmath1118 holds for all @xmath1119 and all translates of @xmath739 contained anywhere in @xmath700 . since gives us an interval of width of order @xmath1120 where @xmath1121 keeps a uniformly positive mass , the gaussian concentration in proposition [ lem : gaussian_conc ] shows that , for some constants @xmath1122 , @xmath1123 and also @xmath1124 hold for every @xmath60 .
the proof has so far assumed @xmath1107 ; to eliminate this assumption we note that @xmath1125 uniformly in @xmath1126 and so the union bound gives @xmath1127 since @xmath1128 while @xmath1129 is at most @xmath1130 times an @xmath1131-dependent constant , by adjusting @xmath1132 we make hold for all @xmath59 . due to rotation symmetry ,
the same bound holds also for @xmath1133 and any translate @xmath686 of @xmath743 contained in @xmath700 .
now fix @xmath1134 and let @xmath1135
. then one can find a collection of rectangles of the form @xmath743 or @xmath739 with @xmath1136 that are contained in @xmath700 and satisfy : 1 .
there are at most @xmath1137 of such rectangles with @xmath1138 independent of @xmath444 .
2 . if a path is chosen connecting the shorter sides in each of these rectangles , then the graph union of these paths contains a path from @xmath20 to @xmath19 . by lemma [ lem : series_law ]
, this construction dominates @xmath1139 by the sum of the resistances between the shorter sides of these rectangles . the fkg inequality , and a union bound
then imply @xmath1102 this is the desired claim . in order to extend this to the network with the underlying field @xmath6 , we first note : let @xmath6 denote the gff on @xmath3 pinned at the origin .
there are @xmath1140 and @xmath999 such that for all @xmath1000 , all @xmath1001 and for every translate @xmath686 of @xmath739 contained in @xmath845 , we have @xmath1141 the same applies to @xmath742 for translates @xmath686 of @xmath743 with @xmath1142 .
we will assume that @xmath444 is the minimal integer such that @xmath1142 .
note that this means that @xmath1143 is bounded .
we proceed in two steps , first reducing @xmath6 to the gff in @xmath1144 and then relating this field to @xmath699 . using the gibbs - markov property , the field @xmath6
can be written as @xmath1145 , where @xmath1146 , the fine field , has the law of the gff on @xmath1147 while the coarse field @xmath1148 is @xmath6 conditional on its values outside of @xmath700 .
now pick an @xmath1149 such that @xmath72 is at least @xmath1150 lattice steps from both @xmath845 and @xmath1151 . for any @xmath1069
we then have @xmath1152 noting that both events are increasing functions of @xmath6 , for the first probability on the right we get @xmath1153 using the fkg inequality . for the second probability we set @xmath1154 and note , since @xmath1155
, we have @xmath1156 but the above definition ensures that @xmath1157 is independent of @xmath1158 and a calculation using the explicit form of the law of @xmath1159 gives that @xmath1160 is bounded by a constant independent of @xmath444 .
markov s inequality and then bound the last probability in by @xmath1161 for some constant @xmath732 independent of @xmath116 or @xmath444 .
next let @xmath1162 $ ] be discrete harmonic on @xmath1147 with @xmath1163 and @xmath1164 whenever @xmath1165 .
let @xmath1166 have the law of @xmath1167 but assume that @xmath1166 is independent of @xmath1146 .
the gibbs - markov property shows @xmath1168 a direct use of lemma [ lem : resistance_decomp ] is hampered by the fact that @xmath1169 is of order @xmath1170 .
however , this is not a problem when @xmath686 is at least distance @xmath1171 from the origin because then @xmath1172 . letting @xmath1173
, we now note that each translate @xmath686 of @xmath739 contains a translate @xmath747 of @xmath1174 which is at least distance @xmath10 from the origin and is aligned with one of the longer side of @xmath686 .
lemma [ lem : resistance_decomp ] then gives , for any @xmath1175 , @xmath1176 for some @xmath1177 .
hence @xmath1178 now set @xmath1179 , @xmath1180 and pick any @xmath1181 .
then the last probability in both and tends to zero as @xmath420 , while , as soon as @xmath787 is large enough , the first probability on the right of is uniformly positive by proposition [ prop-4.7 ] and a routine use of the fkg inequality ( to get us from rectangles of the form @xmath1174 to those with aspect ratio @xmath1182 ) .
the claim follows . using exactly the same argument as in the proof of lemma [ prop : thm1.4 ] , we then get : [ cor-4.18 ] let @xmath6 be the gff in @xmath1183 .
there are @xmath57 such that @xmath1184 holds for all @xmath1103 and all @xmath1105 .
this is one half of theorem [ thm - effective - resistance ] ; the other half will be shown in section [ sec - proofs ] .
here we use the techniques developed earlier in this paper to finally prove our main results .
we begin with some preparatory claims ; the actual proofs start to appear in section [ sec-5.3 ]
. our proofs will require restricting to subsets of @xmath3 of points with only a moderate value of the effective resistance to the origin and/or the boundary of a box centered there in .
here we give the needed bounds on cardinalities of such sets .
abbreviate , as in , @xmath423 .
we will proceed by a straightforward first - moment estimate , but first we have to localize the problem to a finite box . write @xmath1187 where @xmath1188 is the fine field on the box @xmath1189 . since @xmath1190 ,
the variance of @xmath1159 is bounded by a constant times @xmath1191 uniformly on @xmath56 and so , combining corollary [ cor : field_smoothness1 ] with a bound at one vertex , @xmath1192 on the event when @xmath1193 we have @xmath1194 and so comparing this with the restriction on the effective resistivity in we may as well estimate the probability in for @xmath6 replaced by @xmath1195 . here
we will still need to employ a truncation to keep the field @xmath1195 below its typical maximum scale .
the following crude estimate based on a union bound is sufficient , @xmath1196 for some constants @xmath448 .
writing @xmath1197 for the complementary event and inserting @xmath1197 in the probability in with @xmath6 replaced by @xmath1195 , markov s inequality bounds the result by @xmath1198 now @xmath1199 is increasing while @xmath1200 is a decreasing event . since the conditioning on @xmath1197 preserves the fkg inequality , the quantity in is no larger than @xmath1201 corollary [ cor-4.18 ] bounds the last probability by @xmath1202 so we just have to compute the sum of the expectations of @xmath1203 s .
pick a pair of nearest neighbors @xmath20 and @xmath19 , with @xmath1204 , and let @xmath1205 . disregarding the event @xmath1197 , a straightforward moment computation using @xmath1206 for @xmath1204 shows @xmath1207 on the other hand , a change of measure argument gives @xmath1208 for @xmath1209 , the probability itself decays as @xmath1210 . invoking the definition of @xmath46 in
, the inequalities ( [ e:5.13][e:5.14 ] ) thus give @xmath1211 summing over @xmath1204 , the claim follows . in light of lemma [ lemma-5.3 ]
, it suffices to show that @xmath1214 thanks to the gibbs - markov property , it actually suffices to show this ( with @xmath454 replaced by @xmath1215 ) for @xmath6 replaced by @xmath650 and @xmath1216 replaced by a box @xmath56 .
( indeed , we just need to take a translate @xmath298 of @xmath56 with @xmath1217 and then use the gibbs - markov property on a translate of @xmath1218 centered at the same point as @xmath298 .
the contribution of the coarse field is estimated using corollary [ cor : field_smoothness1 ] . )
the argument for is different depending on the relation between @xmath71 and @xmath52 .
for @xmath1219 we use that the maximum of the gff has doubly - exponential lower tails ( see @xcite ) . invoking the gibbs - markov property
we then conclude that , with probability at least @xmath1220 , for some @xmath494 , there is at least one point @xmath20 where @xmath1221 for some large enough @xmath1222 . as @xmath1223 , for @xmath20 and @xmath19 neighbors ,
have bounded ( in fact , stationary ) variances , a union bound shows that will hold also for the neighbors of @xmath20 . on this event , and denoting by @xmath19 a neighbor of @xmath20 , @xmath1224 since @xmath1225 equals @xmath46 for @xmath1219 , we are done here .
concerning @xmath1226 , here will will apply theorem [ thm - levelset ] for @xmath1227 .
recall the notation @xmath1228 for the level set in . a straightforward computation
using the explicit form of the gaussian probability density shows @xmath1229 and so @xmath1230 .
theorem [ thm - levelset ] now guarantees that @xmath1231 occurs with probability @xmath1232 .
this statements permits even setting @xmath1233 , whereby the claim readily follows .
the starting point of our proofs is an upper bound on the return probability for the random walk .
we remark that numerous methods exist in the literature to derive such bounds .
some of these are based on geometric properties of the underlying markov graph such as isoperimetry and volume growth , others are based on resistance estimates .
the most natural approach to use would be that of @xcite ( see also @xcite ) ; unfortunately , this does not seem possible due to our lack of required uniform control of the resistance growth . instead , we base our presentation on the general strategy outlined in ( * ? ? ?
* chapter 21.5 ) .
we begin by restating , and proving , one half of theorem [ thm_main_spectral ] : pick @xmath34 and a large integer @xmath44 , and recall the notation @xmath1236 for the set in .
consider the random walk @xmath1237 on the network @xmath1238 ; this walk starts at @xmath88 and moves around @xmath1239 indefinitely using the transition probabilities that are modified on the boundary of @xmath1239 so that jumps outside @xmath1239 are suppressed .
let @xmath1240 record the successive visits of @xmath1241 to @xmath1236 .
then @xmath1242 is a markov chain on @xmath1236 with stationary distribution @xmath1243 let @xmath1244 , @xmath1245 , @xmath1246 , etc be the times of the successive visits of @xmath1242 to @xmath88 .
define @xmath1247 then we have @xmath1248 where the first inequality comes from the monotonicity of @xmath1249 and the second inequality reflects the fact that @xmath1250 .
since @xmath1251 , by , e.g. , ( * ? ? ?
* lemma 10.5 ) we have @xmath1252 ( this is proved by noting that the object on the left is a stationary measure for the walk @xmath1242 of total mass @xmath1253 . ) by conditioning on @xmath1254 we further estimate @xmath1255 where @xmath1256 and note that @xmath1257 by the commute - time identity of @xcite ( cf ( * ? ? ? * corollary 2.21 ) ) . combining this with ( [ e:5.21ua][e:5.22ua ] ) and
we then get @xmath1258 which proves because , due to the jumps being only to nearest neighbors , the walk @xmath1241 coincides with the walk @xmath1259 up to time at least @xmath1260 . a standard calculation based on reversibility and
the cauchy - schwarz inequality yields @xmath1261 invoking the upper bound on the heat - kernel and lemma [ lem - volume - upper ] , we get that with probability tending rapidly to one as @xmath10 and @xmath44 tend to infinity , we have @xmath1262 ^ 2\,.\ ] ] setting @xmath1263 gives the desired claim .
we now move to the proofs of the requisite lower bounds . here
the focus will be trained on the expected exit time which we write using the hitting time identity as @xmath1266 where , using our convention that @xmath89 is the external boundary of @xmath56 , @xmath1267 is the electrostatic potential , a.k.a .
voltage , in @xmath56 with @xmath1268 and @xmath481 vanishing on @xmath89 . estimating from below naturally requires finding a sufficiently good lower bound on @xmath481 .
the idea is to recast the problem using a simple electric network and invoke suitable effective resistance estimates .
the following computation will be quite useful : consider a resistor network with three nodes , @xmath1269 , and for each @xmath1270 let @xmath1271 denote the conductance of the edge @xmath1272 .
let @xmath1273 denote the effective resistance between node @xmath271 and node @xmath282 .
then , @xmath1274 let us represent the network by an equivalent network , now with nodes @xmath1275 whose only edges are from @xmath88 to each of @xmath1276 . denoting the conductances of these edges by @xmath1277 respectively
, the @xmath1242-@xmath1278 transform shows @xmath1279 next let us introduce the associate resistances @xmath1280 .
the series law then gives @xmath1281 for all @xmath1282 .
a computation shows that , for all cyclic permutations @xmath1283 of @xmath1284 , @xmath1285 some algebra then shows that the ratio on the left of equals @xmath1286 .
this is then checked to agree with the right - hand side . as @xmath1289
, we may apply the network reduction principle to represent the problem on an effective network of three nodes , with node @xmath1290 labeling @xmath19 , node @xmath1291 marking the origin and node @xmath1292 standing for @xmath89 .
since @xmath481 is harmonic on @xmath648 , it is also harmonic on the effective network .
but there @xmath1293 is just the probability that the random walk at @xmath19 jumps right to @xmath88 in the first step .
using conductances , this probability is exactly the expression on the left of . plugging in the effective resistances , the claim follows . for the proof we recall the annulus decomposition of the gff from section [ subsec : smoothness ] .
let @xmath968 and for a given @xmath59 and @xmath1297 , set @xmath967 .
define the annuli @xmath1298 and @xmath1299 note that @xmath1300 .
write @xmath1301 , where @xmath1159 is the coarse field on @xmath1302 and @xmath1303 is the corresponding fine field .
denote @xmath1304 define @xmath984 as in and for @xmath1305 let @xmath1306 ( both objects are measurable with respect to @xmath6 . ) similarly to lemma [ lem : anchoring ] we get @xmath1307 as soon as @xmath1308 is sufficiently large .
let @xmath1309 denote the fine field on @xmath1310 , @xmath1311 ( we think of @xmath1312 as set to zero outside @xmath1313 ) and @xmath1314 be the corresponding coarse field .
the definitions ensure @xmath1315 note also that @xmath984 and @xmath1316 are independent as long as @xmath1317 .
next recall that @xmath1318 , for @xmath116 an annulus in @xmath3 , denotes the sum of the effective resistances in network @xmath1319 between the shorter sides of the four maximal rectangles contained in @xmath116 .
recall also that @xmath1320 denotes the effective resistance in @xmath1319 between the inner and outer boundaries of annulus @xmath116 .
we define the events : @xmath1321 and @xmath1322 here @xmath786 is the constant proposition [ prop : zero_mean ] and @xmath204 is fixed via : [ lemma-5.11 ] for each @xmath34 there are @xmath1323 , @xmath541 , @xmath1138 such that one can choose @xmath1324 in the definitions of @xmath1325 and @xmath1326 so that , for all @xmath542 and all @xmath1327 , @xmath1328 abbreviate by @xmath1329 the first event on the right of . this event is measurable with respect to @xmath1312 and so @xmath1330 are independent . by lemma [ lemma - annuli ]
, @xmath1331 holds for some @xmath1061 and all @xmath491 as soon as @xmath542 .
we are first interested in a simultaneous occurrence of @xmath1329 and @xmath1332 . recalling that @xmath1333 is a random walk ,
define the stopping time @xmath1334 then , for @xmath204 sufficiently small , lemma [ lem : finite_lil ] shows @xmath1335 for some constant @xmath1138 . since the increments of @xmath984 are independent centered gaussians with a uniform bound on their tail , for the event @xmath1336 the fact that @xmath1337 yields @xmath1338 as soon as @xmath495 is larger than a positive constant . under a similar restriction on @xmath495
, we then also have @xmath1339 therefore , on the event on the left , @xmath1340 will not occur for some @xmath1341 only if the sequence @xmath1342 contains a run of 1 s of length at least @xmath1343 .
this has probability @xmath1344 . as @xmath1061
, we get @xmath1345 as soon as @xmath495 is larger than some positive constant . for event @xmath1326 , the fact that the coarse field @xmath1159 on @xmath1346 has uniformly bounded variances implies , via corollary [ cor : field_smoothness1 ] , @xmath1347 for some @xmath498 .
proposition [ prop-4.7 ] in turn shows that the first event on the right of has a uniformly positive probability .
the claim then follows as before .
fix @xmath1348 large and , given @xmath1295 , let @xmath495 be the largest integer such that @xmath1349 .
( we are assuming the setting of lemma [ lemma-5.11 ] . ) abbreviate @xmath1350 and suppose that the event @xmath1351 occurs for some @xmath1352 with @xmath1353 .
then @xmath1354}\,{\mathrm{e}}^{-3\hat c\log\log(n ' ) } \\ & \ge { \mathrm{e}}^{\tilde c\sqrt{k^\star\log\log k^\star } } \end{aligned}\ ] ] holds for some constant @xmath1355 , where we used that @xmath1356 implies @xmath1357 as soon as @xmath961 is sufficiently large .
similarly , abbreviating @xmath1358 , we get @xmath1359 where we again used that @xmath1357 .
now observe that if @xmath1360 , then the nash - williams estimate implies @xmath1361 while the series law gives @xmath1362 since @xmath1363 implies that @xmath1364 lies outside @xmath1365 , we also have @xmath1366 combining ( [ e:5.62][e:5.65 ] ) we thus get that @xmath1367 for all @xmath1360 as soon as the event in occurs .
the claim now follows ( for @xmath10 replaced by @xmath961 ) from and lemma [ lemma-5.11 ] .
using the same proof , lemma [ lemma-5.3 ] applies also for resistivity @xmath1372 . in light of @xmath1373 corollary [ cor-4.18 ]
applies to @xmath1374 just as well . combining this with
, we now proceed as in the proof of lemma [ lem - good - volume ] to get the result .
the upper bound has already been shown in lemma [ lemma-5.7 ] , so we just need to derive the corresponding lower bound .
for this we write as a bound and apply with proposition [ prop-5.8 ] to get that , with probability tending to one as @xmath420 , @xmath1375 the claim then follows from lemma [ lem - good - volume2 ] .
the upper bound on the return probability has already been proved in lemma [ lemma-5.6 ] , so we will focus on the lower bound and recurrence .
consider again the random walk @xmath1241 on @xmath1376 and let @xmath1242 be its trace on @xmath1377 .
let @xmath1378 .
then @xmath1379 the hitting time estimate in conjunction with the definition of @xmath1377 gives @xmath1380 whereby we get @xmath1381 since applies also for the expectation of @xmath1382 , the choice @xmath1383 implies @xmath1384 and thus , using one more time , @xmath1385 but @xmath1386 and so we get @xmath1387 as well . using this in , the desired lower
bound then follows from , e.g. , .
it remains to show recurrence .
here we note that and along with lemma [ lemma-5.11 ] imply that @xmath1388 in probability along a sufficiently rapidly growing deterministic sequence of @xmath10 s .
since the sequence of resistances is increasing in @xmath10 , the convergence holds almost surely . by a well known criterion ,
this implies recurrence . here
the bound has already been shown in corollary [ cor-4.18 ] , so we just have to focus on ( [ e:1.11ua][e:1.12ua ] ) .
we will use a decomposition of @xmath6 from ( * ? ? ?
* proposition 3.12 ) .
let @xmath968 and consider the annuli @xmath1389 and @xmath1390 for all @xmath1391
. then @xmath1392,\ ] ] where @xmath1393 is a function such that @xmath1394 while @xmath1395 are random variables and @xmath1396 and @xmath1397 are random fields ( all measurable with respect to @xmath6 ) that are independent of one another and distributed as centered gaussian with the specifics of the law determined as follows : 1 .
@xmath1398 , 2 .
writing @xmath1399 for the coarse field obtained as the conditional expectation of the gff on @xmath1400 given its values on @xmath1401 , we have @xmath1402 3 .
@xmath1403 is the fine field on @xmath1404 .
clearly , only one of the fine fields @xmath1312 can contribute in for each given @xmath19 and @xmath1407 unless @xmath1408 . setting ( with some abuse of our earlier notation ) ,
@xmath1409 ( * ? ? ? *
lemma 3.8 ) shows that , for some constants @xmath1410 , @xmath1411 the first half of then lets us write @xmath1412 we now set @xmath1413 and note that the nash - williams estimate and lemma [ lem : resistance_decomp ] imply @xmath1414\ ] ] where @xmath1415 .
our aim is to study the maximum in and show that it grows at least as exponential of @xmath1416 . to this end
, we define the sequence of record values of the sequence @xmath1417 as follows : set @xmath1244 and for @xmath1418 let @xmath1419 then we have : [ lemma - excursions ] @xmath1420 are independent with a uniform bound on their tail , @xmath1421 for some constant @xmath494 .
in particular , for each @xmath34 there is @xmath732 such that @xmath1422 holds for all @xmath1418 . postponing the proof temporarily , we note that shows @xmath1423
a borel - cantelli argument then gives @xmath1424 ( this is first proved for @xmath980 running along powers of @xmath1291 and then extended by monotonicity of both numerator and denominator . ) in particular , for @xmath495 large enough , the sequence @xmath1425 will see at least @xmath1416 record values as defined above .
if it were not for the terms @xmath1426 and @xmath1427 , this observation would bound the maximum in by what we want , so we have to ensure that these terms do not spoil this . consider the events @xmath1428 and @xmath1429 by and markov s inequality , there is an a.s .
finite @xmath1430 such that @xmath1431 ( again , we prove this for @xmath495 running along powers of @xmath1291 and then fill the gaps by monotonicity . ) this means that at least half of the record values by time @xmath495 occur at indices where @xmath1432 occurs , i.e. , @xmath1433 as soon as @xmath495 is large enough .
but the events @xmath1434 are independent of each other and of _ all _ of @xmath1435 s and @xmath1436 s and , since lemma [ lemma - annuli ] tells us @xmath1437 for some @xmath1222 , the longest run of 1 s in the sequence @xmath1438 has length at most @xmath1439 .
it follows that , for @xmath495 large , the event @xmath1440 occurs for some @xmath491 of the form @xmath1441 for some @xmath1442 subject to @xmath1443 and @xmath1444 for @xmath1445 .
this shows @xmath1446 and so @xmath1447 once @xmath495 is large enough . from
we now conclude @xmath1448 for some constants @xmath1410 as soon as @xmath495 is large enough .
since also @xmath1440 occur for @xmath1449 , using this in yields @xmath1450 the bound follows .
we will follow the proof of ( * ? ? ? * lemma 4.16 ) .
since the sequence @xmath1451 has independent ( centered ) gaussian increments , we can embed it into a path of standard brownian motion by putting @xmath1452 where @xmath1453 . by property ( 1 ) above
, we have @xmath1454 as @xmath1455 .
consider the process @xmath1456 which is zero outside the interval @xmath1457 $ ] and on this interval , @xmath1458 the independence of increments of brownian motion now gives @xmath1459\bigr ) \\ \ge
p\bigl(b_{t_j}-b_{t_k}\le 1\colon j = t_k,\dots , t_n\bigr)\prod_{j = k}^{n-1}p\bigl(\max_{s\in[t_j , t_{j+1}]}w^{(j)}(s)\le1+\log(1+t_j - t_k)\bigr).\end{gathered}\ ] ] since @xmath1456 are brownian bridges on intervals of bounded length , and maxima thereof thus have a uniformly gaussian tail , the product on the right - hand side is positive uniformly in @xmath495 .
it follows that , for @xmath1460 being a uniform lower bound on the product , @xmath1461 where @xmath1462 .
the probability on the right is at most @xmath1463 by , e.g. , ( * ? ? ? * proposition 4.9 ) .
this proves .
the bound now follows from standard estimates of sums of independent heavy - tailed random variables .
t. kumagai . ,
volume 2101 of _ lecture notes in mathematics_. springer , cham , 2014 .
lecture notes from the 40th probability summer school held in saint - flour , 2010 , cole dt de probabilits de saint - flour .
[ saint - flour probability summer school ] . | given any @xmath0 and for @xmath1 denoting a sample of the two - dimensional discrete gaussian free field on @xmath2 pinned at the origin , we consider the random walk on @xmath3 among random conductances where the conductance of edge @xmath4 is given by @xmath5 . we show that , for almost every @xmath6 , this random walk is recurrent and that , with probability tending to 1 as @xmath7 , the return probability at time @xmath8 decays as @xmath9 .
in addition , we prove a version of subdiffusive behavior by showing that the expected exit time from a ball of radius @xmath10 scales as @xmath11 with @xmath12 for all @xmath0 .
our results rely on delicate control of the effective resistance for this random network .
in particular , we show that the effective resistance between two vertices at euclidean distance @xmath10 behaves as @xmath13 . |
this paper provides the mathematical foundation for stochastically continuous affine processes on the cone of positive semidefinite symmetric @xmath0-matrices @xmath1 .
these matrix - valued affine processes have arisen from a large and growing range of useful applications in finance , including multi - asset option pricing with stochastic volatility and correlation structures , and fixed - income models with stochastically correlated risk factors and default intensities . for illustration ,
let us consider a multi - variate stochastic volatility model consisting of a @xmath2-dimensional logarithmic price process with risk - neutral dynamics @xmath3 and stochastic covariation process @xmath4 , which is a proxy for the instantaneous covariance of the price returns . here
@xmath5 denotes a standard @xmath2-dimensional brownian motion , @xmath6 the constant interest rate , @xmath7 the vector whose entries are all equal to one and @xmath8 the vector containing the diagonal entries of @xmath9 . the necessity to specify @xmath9 as a process in @xmath1 such that it qualifies as covariation process is one of the mathematically interesting and demanding aspects of such models .
beyond that , the modeling of @xmath9 must allow for enough flexibility in order to reflect the stylized facts of financial data and to adequately capture the dependence structure of the different assets .
if these requirements are met , the model can be used as a basis for financial decision - making in the area of portfolio optimization , pricing of multi - asset options and hedging of correlation risk .
the tractability of such a model crucially depends on the dynamics of @xmath9 .
a large part of the literature in the area of multivariate stochastic volatility modeling has proposed the following affine dynamics for @xmath9 : @xmath10\\[-8pt ] x_0&=&x\in s_d^+ , \nonumber\end{aligned}\ ] ] where @xmath11 is some suitably chosen matrix in @xmath1 , @xmath12 some invertible matrices , @xmath13 a standard @xmath0-matrix of brownian motions possibly correlated with @xmath5 , and @xmath14 a pure jump process whose compensator is an affine function of @xmath9 .
the main reason for the analytic tractability of this model is that , under some technical conditions , the following affine transform formula holds : @xmath15 = e^{\phi(t , z , v)+{\operatorname{tr}}(\psi(t , z , v ) x)+ v^\top y}\ ] ] for appropriate arguments @xmath16 and @xmath17 . the functions @xmath18 and @xmath19 solve a system of nonlinear ordinary differential equations ( odes ) , which are determined by the model parameters . setting @xmath20 , @xmath21 and @xmath22 and taking @xmath23
, we arrive at @xmath24 = e^{-\phi(t , u)-{\operatorname{tr}}(\psi(t , u ) x)},\qquad u\in s_d^+.\ ] ] in this paper , we characterize the class of all stochastically continuous time - homogeneous markov processes with the key property ( [ introaffine])henceforth called affine processes on @xmath1 .
our main result shows that an affine process is necessarily a feller process whose generator has affine coefficients in the state variables .
the parameters of the generator satisfy some well - determined admissibility conditions , and are in a one - to - one relation with those of the corresponding odes for @xmath25 and @xmath26 .
conversely , and more importantly for applications , we show that for any admissible parameter set there exists a unique well - behaved affine process on @xmath1 .
furthermore , we prove that any stochastically continuous infinitely decomposable markov process on @xmath1 is affine with zero diffusion , and vice versa . on the one hand ,
our findings extend the model class ( [ eq : cov process ] ) , since a more general drift and jumps are possible .
indeed , we allow for full generality in @xmath11 , as long as @xmath27 , for a general linear drift part @xmath28 and for an inclusion of ( infinite activity ) jumps
. this of course enables more flexibility in financial modeling . for example , due to the general linear drift part , the volatility of one asset can generally depend on the other ones , which is not possible for @xmath29 . on the other hand , we now know the exact assumptions under which affine processes on @xmath1 actually exist .
our characterization of affine processes on @xmath1 is thus exhaustive . beyond that ,
the equivalence of infinitely decomposable markov processes with state space @xmath1 and affine processes without diffusion is interesting in its own right .
this paper complements duffie , filipovi and schachermayer @xcite , who analyzed time - homogeneous affine processes on the state space @xmath30 .
have been explored in @xcite . ]
matrix - valued affine processes seem to have been studied systematically for the first time in the literature by bru , who introduced the so called wishart processes .
these are generalizations of squares of matrix ornstein uhlenbeck processes , that is , of the form ( [ eq : cov process ] ) for @xmath31 and @xmath32 , for some real parameter @xmath33 .
note that @xmath33 is a stronger assumption than what we require on @xmath11 and @xmath34 .
bru @xcite then establishes existence and uniqueness of a local @xmath1-valued solution to ( [ eq : cov process ] ) under the additional assumptions that @xmath35 has distinct eigenvalues , @xmath36 , and that @xmath37 and @xmath38 commute ( see @xcite , theorem 2@xmath39 ) . in the more special case
where @xmath40 and @xmath41 , bru @xcite shows global existence and uniqueness for ( [ eq : cov process ] ) for any @xmath35 with distinct eigenvalues ( see @xcite , theorem 2 and last part of section 3 ) .. but these are degenerate solutions , as they are only defined on lower - dimensional subsets of the boundary of @xmath1 ( see @xcite , corollary 1 ) . ]
bru s results concerning strong solutions have recently been extended to the case of matrix valued jump - diffusions ; see @xcite .
wishart processes have subsequently been introduced in the financial literature by gourieroux and sufana @xcite and gourieroux et al .
financial applications thereof have then been taken up and carried further by various authors , including da fonseca et al .
@xcite and buraschi , cieslak and trojani @xcite .
grasselli and tebaldi @xcite give some general results on the solvability of the corresponding riccati odes .
barndorff - nielsen and stelzer @xcite provide a theory for a certain class of matrix - valued lvy driven ornstein
uhlenbeck processes of finite variation .
leippold and trojani @xcite introduce @xmath1-valued affine jump diffusions and provide financial examples , including multi - variate option pricing , fixed - income models and dynamic portfolio choice .
all of these models are contained in our framework .
we want to point out that the full characterization of positive semidefinite matrix - valued affine processes needs a multitude of methods . in order to prove the fundamental property of regularity of affine processes
another adaption of the famous analysis of montgommery and zippin is necessary , which has been worked out in @xcite and @xcite for the state space @xmath42 . for the necessary conditions on drift , diffusion and jump parameters we need the theory of infinitely divisible distributions on @xmath1 .
most interestingly , the constant drift part @xmath11 must satisfy a condition depending on the magnitude of the diffusion component ( see proposition [ th : constant drift ] ) , which is in accordance with the choice of the drift in bru s work @xcite on wishart processes , as explained above .
this enigmatic additional condition on the drift @xmath11 is derived by studying the process with respect to well chosen test functions , including in our case the determinant of the process .
it is worth noting , as already visible in dimension one , that a naive application of classical geometric invariance conditions does not bring the correct necessary result on the drift but a stronger one .
indeed , take a one - dimensional affine diffusion process @xmath9 solving @xmath43 then a back - of - the - envelope calculation would yield the stratonovich drift at the boundary point @xmath44 of value @xmath45 , leading to the necessary parameter restriction @xmath46 , which is indeed too strong .
it is well known that the correct parameter restriction is @xmath47 .
we see two reasons why geometric conditions on the drift can not be applied : first , precisely at the boundary of our state spaces the diffusion coefficients are not lipschitz continuous anymore , and , second , the boundary of the cone of positive semi - definite matrices is not a smooth submanifold but a more complicated object . for the sufficient direction
refined methods from stochastic invariance theory are applied .
having established viability of a particular class of jump - diffusions on @xmath1 , existence of affine processes on @xmath1under the necessary parameter conditions is shown through the solution of a martingale problem .
uniqueness follows by semigroup methods which need the theory of multi - dimensional riccati equations . summing up
, we face two major problems in the analysis of positive matrix valued affine processes .
first , the candidate stochastic differential equations necessarily lead to volatility terms which are not lipschitz continuous at the boundary of the state space .
this makes every existence , uniqueness and invariance question delicate .
second , the jump behavior transversal to the boundary is of finite total variation . for affine processes on @xmath1 ,
results and proofs deviate in essential points from the theory on state spaces of the form @xmath30 given in @xcite , which is a consequence of the more involved geometry of this nonpolyhedral cone .
the program of the paper as outlined below therefore includes a comparison with the approach in @xcite .
section [ section : definition of affine property ] contains the main definition and a summary of the results of this article . in section [ subsec : regular and feller ] , we then derive two main properties , namely the regularity of the process and the feller property of the associated semigroup .
the feller property , in turn , is a simple consequence of an important positivity result of the characteristic exponents @xmath48 , which is proved in lemma [ prop : feller prop ] .
this lemma is further employed as a tool for the treatment of the generalized riccati differential equations in section [ section : riccati ] ( see proof of proposition [ prop_ricc_sol ] ) . the global existence and uniqueness of these equations
is then used to show uniqueness of the martingale problem for affine processes ( see proof of proposition [ th : existence markov ] ) . in section
[ section : nec paramter restrictions ] , we define a set of admissible parameters specifying the infinitesimal generator of affine semigroups and prove the necessity of the parameter restrictions ( see proposition [ th : necessary admissibility ] ) . the sufficient direction is then treated in section [ sec : existence ] .
it is known that , for @xmath49 , there exist continuous affine processes on @xmath1 which are in contrast to those on the state space @xmath30_not _ infinitely divisible ( see example [ counterex : bru ] ) . the analysis of this paper reveals the failure of infinite divisibility as a consequence of the drift condition ( see proof of theorem [ th : char infdec infdiv ] ) .
this has substantial influence on the approach chosen here to prove existence of affine processes associated with a given parameter set : being in general hindered to recognize the solutions of the generalized riccati differential equations as cumulant generating functions of sub - stochastic measures , as done in @xcite , section 7 , we solve the martingale problem for the associated lvy type generator on @xmath1 , as exposed in section [ sec : existence ] and appendix [ app : viability ] . in section [ subsec : alternative existence ] , however , we deliver a variant of the existence proof of @xcite for pure jump processes , which is possible in this case due to the absence of a diffusion component . finally , section [ secproofs ] contains the proofs of the main results which build on the propositions of the previous sections . for the stochastic background and notation , we refer to standard text books such as @xcite and @xcite .
we write @xmath50 and @xmath51 . moreover : * @xmath52 denotes the space of symmetric @xmath0-matrices equipped with the scalar product @xmath53 .
note that @xmath52 is isomorphic , but not isometric , to the standard euclidean space @xmath54 .
we denote by @xmath55 the standard basis of @xmath52 , that is , the @xmath56th component of @xmath57 is given by @xmath58 , where @xmath59 denotes the kronecker delta .
additionally , we sometimes consider the following basis elements @xmath60 which are positive semidefinite and form a basis of @xmath52 : @xmath61 * @xmath1 stands for the cone of symmetric @xmath0-positive semidefinite matrices , @xmath62 for its interior in @xmath52 , the cone of strictly positive definite matrices .
the boundary is denoted by @xmath63 , the complement is denoted by @xmath64 , and @xmath65 denotes the one - point compactification .
recall that @xmath1 is self - dual [ w.r.t .
the scalar product @xmath66 , that is , @xmath67 both cones , @xmath1 and @xmath62 , induce a partial and strict order relation on @xmath52 , respectively : we write @xmath68 if @xmath69 , and @xmath70 if @xmath71 .
* @xmath72 is the space of @xmath0-matrices and @xmath73 the orthogonal group of dimension @xmath2 over @xmath74 .
* @xmath75 denotes the @xmath0-identity matrix . throughout this paper , a function @xmath76
is understood as the restriction @xmath77 of a function @xmath78 which satisfies @xmath79 for all @xmath80 . without loss of generality @xmath81 .
we avoid using the @xmath82 operator , that is , to identify @xmath83 with a vector in @xmath54 by stringing the columns of @xmath84 together , while only taking the entries @xmath85 with @xmath86 . throughout this article
, we shall consider the following function spaces for measurable @xmath87 .
we write @xmath88 for the borel @xmath89-algebra on @xmath90 .
@xmath91 corresponds to the banach space of bounded real - valued borel measurable functions @xmath92 on @xmath90 with norm @xmath93 .
we write @xmath94 for the space of real - valued continuous functions @xmath92 on @xmath90 , @xmath95 for @xmath96 , @xmath97 for the space of functions @xmath98 with compact support and @xmath99 for the banach space of functions @xmath98 with @xmath100 and norm @xmath101 .
furthermore , @xmath102 is the space of @xmath103 times differentiable functions @xmath92 on @xmath104 , the interior of @xmath90 , such that all partial derivatives of @xmath92 up to order @xmath103 belong to @xmath94 . as usual
, we set @xmath105 , and we write @xmath106 and @xmath107 , for @xmath108 .
we consider a time - homogeneous markov process @xmath9 with state space @xmath1 and semigroup @xmath109 acting on functions @xmath110 , @xmath111 we note that @xmath9 may not be conservative .
then there is a standard extension of the transition probabilities to the one - point compactification @xmath112 of @xmath1 by defining @xmath113 for all @xmath114 and @xmath115 , with the convention that @xmath116 for any function @xmath92 on @xmath1 .
thus @xmath9 becomes conservative on @xmath112 .
[ def : affine process sd+ ] the markov process @xmath9 is called _ affine _ if : it is stochastically continuous , that is , @xmath117 weakly on @xmath1 for every @xmath114 and @xmath118 , and its laplace transform has exponential - affine dependence on the initial state @xmath119 for all @xmath114 and @xmath120 , for some functions @xmath121 and @xmath122 .
note that stochastic continuity of @xmath9 implies that @xmath123 and @xmath124 are jointly continuous in @xmath125 ; see lemma [ lem : order preserving affine](iii ) below .
moreover , due to the markov property , this also means that @xmath126 for all @xmath127 and @xmath128 .
in contrast to @xcite , we take stochastic continuity as part of the definition of affine processes , and consider the laplace transform instead of the characteristic function .
the latter is justified by the nonnegativity of @xmath9 , the former is by convenience since , as we will see in proposition [ th : regularity ] below , it automatically implies regularity in the following sense . the affine process @xmath9 is called _ regular _ if the derivatives @xmath129 exist and are continuous at @xmath130
. we remark that there are simple examples of markov processes which satisfy definition [ def : affine process sd+](ii ) but are not stochastically continuous ; see @xcite , remark 2.11 .
however , such processes are of limited interest for applications and will not be considered . in the following
, we shall provide an equivalent characterization of the affine property in terms of the generator of @xmath9 . as we shall see in ( [ eq : generator ] ) , the diffusion , drift , jump and killing characteristics of @xmath9 depend in an affine way on the underlying state .
we denote by @xmath131 some bounded continuous truncation function with @xmath132 in a neighborhood of @xmath133
. then the involved parameters are admissible in the following sense .
[ def : necessary admissibility ] an _ admissible parameter set _ @xmath134 associated with @xmath135 consists of : * a linear diffusion coefficient @xmath136 * a constant drift term @xmath137 * a constant killing rate term @xmath138 * a linear killing rate coefficient @xmath139 * a constant jump term : a borel measure @xmath140 on @xmath141 satisfying @xmath142 * a linear jump coefficient : a @xmath0-matrix @xmath143 of finite signed measures on @xmath141 such that @xmath144 for all @xmath145 and the kernel @xmath146 satisfies @xmath147 * a linear drift coefficient : a family @xmath148 such that the linear map @xmath149 of the form @xmath150 satisfies @xmath151\\[-8pt ] & & \eqntext{\mbox{for all $ x , u\in s_d^+$ with $ \langle x , u\rangle=0$.}}\end{aligned}\ ] ] we shall comment more on the admissibility conditions in section [ subsecdis ] below .
the following three theorems contain the main results of this article .
their proofs are given in section [ secproofs ] .
first , we provide a characterization of affine processes on @xmath1 in terms of the admissible parameter set introduced in definition [ def : necessary admissibility ] . as for the domain of the generator ,
we consider the space @xmath152 of rapidly decreasing @xmath153-functions on @xmath1 , defined in ( [ defs+ ] ) below .
it is shown in appendix [ secstoneweier ] that @xmath154 , for @xmath155 , as well as @xmath156 . [
th : main theorem ] suppose @xmath9 is an affine process on @xmath1 . then @xmath9 is regular and has the feller property
let @xmath157 be its infinitesimal generator on @xmath158 .
then @xmath159 and there exists an admissible parameter set @xmath160 such that , for @xmath161 , @xmath162\\[-8pt ] & & { } + \int_{s_d^+\setminus\{0\ } } \bigl(f(x+\xi)-f(x)\bigr)m(d\xi)\nonumber\\ & & { } + \int_{s_d^+\setminus\{0\}}\bigl(f(x+\xi)-f(x)-\langle \chi(\xi ) , \nabla f(x)\rangle\bigr)m(x , d\xi),\nonumber\end{aligned}\ ] ] where @xmath163 is defined by ( [ eq : betaijdef ] ) , @xmath164 by ( [ eq : mudef ] ) and @xmath165 moreover , @xmath123 and @xmath124 in ( [ def : affine process ] ) solve the generalized riccati differential equations , for @xmath166 , @xmath167 with @xmath168\\[-8pt ] & & { } -\int_{s_d^+\setminus\{0\ } } \biggl(\frac{e^{-\langle u , \xi\rangle}-1+\langle \chi(\xi ) , u \rangle}{\|\xi\|^2
\wedge 1}\biggr)\mu(d \xi),\nonumber\end{aligned}\ ] ] where @xmath169 .
conversely , let @xmath170 be an admissible parameter set
. then there exists a unique affine process on @xmath1 with infinitesimal generator ( [ eq : generator ] ) and ( [ def : affine process ] ) holds for all @xmath171 , where @xmath123 and @xmath124 are given by ( [ eq : f - riccati ] ) and ( [ eq : r - riccati ] ) .
[ remcons ] it can be proved as in @xcite that @xmath9 is conservative if and only if @xmath172 and @xmath173 is the only @xmath1-valued local solution of ( [ eq : r - riccati ] ) for @xmath130 .
the latter condition clearly requires that @xmath174 .
hence , a sufficient condition for @xmath9 to be conservative is @xmath172 and @xmath174 and @xmath175 where @xmath176 denotes the jordan decomposition of @xmath177 .
indeed , it can be shown similarly as in @xcite , section 9 , that the latter property implies lipschitz continuity of @xmath178 on @xmath52 . due to the feller property ,
as established in theorem [ th : main theorem ] , any affine process @xmath9 on @xmath1 admits a cdlg modification , still denoted by @xmath9 ( see , e.g. , @xcite , chapter iii.2 ) .
it can and will thus be realized on the space @xmath179 of cdlg paths @xmath180 with @xmath181 for @xmath182 whenever @xmath183 or @xmath184 . for every @xmath115 ,
we denote by @xmath185 the law of @xmath9 given @xmath186 and by @xmath187 the natural filtration generated by @xmath188 .
we also consider the usual augmentation @xmath189 of @xmath187 , where @xmath190 is the augmentation of @xmath187 with respect to @xmath191 .
then @xmath192 is right continuous and @xmath9 is still a markov process under @xmath192
. we shall now relate conservative affine processes to semimartingales , where semimartingales are understood with respect to the stochastic basis @xmath193 for every @xmath84 .
[ thmsemim ] let @xmath9 be a conservative affine process on @xmath1 and let @xmath194 be the related admissible parameter set associated with the truncation function @xmath135 .
then @xmath9 is a semimartingale whose characteristics @xmath195 with respect to @xmath135 are given by @xmath196 where @xmath163 is given by ( [ eq : betaijdef ] ) , @xmath197 by ( [ eq : cijkl ] ) and @xmath164 by ( [ eq : mudef ] ) .
furthermore , there exists , possibly on an enlargement of the probability space , a @xmath198-matrix of standard brownian motions @xmath13 such that @xmath9 admits the following representation : @xmath199 where @xmath200 satisfies @xmath201 and @xmath202 denotes the random measure associated with the jumps of @xmath9 .
hence , @xmath9 is continuous if and only if @xmath140 and @xmath203 vanish .
let @xmath204 be the set of all families of probability measures @xmath205 on the canonical probability space @xmath206 such that @xmath207 is a stochastically continuous markov processes on @xmath1 with @xmath208=1 $ ] , for all @xmath115 .
note that in contrast to @xcite , there is no need to impose regularity of @xmath9 . for two probability measures
@xmath209 on @xmath206 , the convolution @xmath210 is defined as the push - forward of @xmath211 under the map @xmath212 .
[ def : infinite divisibility ] an element @xmath213 is called : infinitely decomposable , if for each @xmath214 , there exists @xmath215 such that @xmath216 infinitely divisible , if the one - dimensional marginal distributions @xmath217 are infinitely divisible , for all @xmath218 .
in @xcite it was shown that regular affine processes on @xmath30 are infinitely decomposable markov processes , and vice versa .
in fact , this property was the core for the existence proof of affine processes in @xcite .
on @xmath1 the situation is different .
the following counterexample reveals that not all affine processes on @xmath1 are infinitely divisible .
[ counterex : bru ] the affine process @xmath9 on @xmath1 corresponding to the parameter set @xmath219 , where @xmath220 , is the diffusion process initially studied by bru @xcite . by @xcite ,
theorem 3 , the laplace - transforms @xmath221=\bigl(\det(i+2tu)\bigr)^{-\delta/2}e^{-\langle ( i+2tu)^{-1}u , x\rangle}\ ] ] are those of noncentral wishart distributions @xmath222 . by a well - known result due to paul lvy ,
these wishart distributions are not infinitely divisible if @xmath223 ( see @xcite , section 2.c ) . here , is our main result on infinite divisibility of affine processes on @xmath1 .
[ th : char infdec infdiv ] let @xmath224 and @xmath225 .
the following assertions are equivalent : @xmath205 is infinitely decomposable .
@xmath207 is affine with vanishing diffusion parameter @xmath226 .
@xmath207 is affine and infinitely divisible .
we discuss and highlight some properties of the admissible parameter set @xmath170 of an affine process .
let us therefore define the normal cone @xmath227 containing the inward pointing normal vectors , to @xmath1 at @xmath118 . ) from the general definition in ( [ defnorcone ] ) below by choosing @xmath228 and @xmath229 , and using the self - duality of @xmath230 for all @xmath231 . ]
it will be shown in lemma [ lem : zero divisor ] below that @xmath232 only for boundary elements @xmath233 .
the diffusion term does not admit a constant part , and its linear part is of the very specific form @xmath234 this property of @xmath235 has also been stated in the setting of symmetric cones in @xcite .
we could thus write the second order differential operator in ( [ eq : generator ] ) as @xmath236 the reason why we introduce and use the symmetrization ( [ eq : cijkl ] ) of @xmath235 is that it corresponds to the quadratic characteristic ( [ eq : char c ] ) of the semimartingale @xmath9 .
the remarkable drift condition ( [ eq : b ] ) has been assumed in many previous papers . here is the first time where necessity and sufficiency of ( [ eq : b ] ) are proved in the full generality in the presence of jumps .
note that in dimension @xmath237 , the drift condition simply reduces to nonnegativity @xmath238 .
but for dimension @xmath49 , the boundary of the state space @xmath1 becomes curved and kinked , implying a nontrivial trade - off between diffusion @xmath239 and @xmath11 .
concerning the form of @xmath5 , let us note the following : condition ( [ eq : betaij ] ) implies in particular @xmath240 where for any matrix @xmath241 , @xmath242 denotes the matrix where the @xmath243th row and column are deleted .
indeed , inserting @xmath244 in condition ( [ eq : betaij ] ) yields @xmath245 for all @xmath246 with @xmath247 .
since the @xmath243th column and row of such an element @xmath246 is zero , it follows that @xmath248\\[-8pt ] & & \qquad=\bigl\langle\beta^{ii}_{\setminus\{i\}},u_{\setminus\{i\ } } \bigr\rangle-\int_{s_d^+\setminus\{0\ } } \frac{\langle\chi(\xi)_{\setminus\{i\}},u_{\setminus\{i\}}\rangle } { \|\xi\|^2 \wedge1 } \mu_{ii}(d\xi)\geq0.\nonumber\end{aligned}\ ] ] by choosing appropriate elements @xmath249 , we can further derive the integrability of @xmath250 for all @xmath251 , which implies @xmath252 as ( [ eq : bchi ] ) and ( [ eq : intchi ] ) must hold true for all @xmath249 , assertion ( [ eq : betaii ] ) is proved . note that the @xmath253th component of the adjoint operator @xmath254 is given by @xmath255 since @xmath256 . in most previous papers ,
@xmath163 is of the form @xmath257 in this case , @xmath258 and hence ( [ eq : betaij ] ) is equivalent to @xmath259 for all @xmath260 with @xmath261 . if @xmath163 is of the form @xmath262 where @xmath263 and @xmath264 linear satisfying @xmath265 , then , in view of ( [ eq : wishart drift ] ) , condition ( [ eq : betaij ] ) holds true as long as @xmath266 for all @xmath260 with @xmath261 . as a bold conjecture
, we claim that any @xmath163 satisfying ( [ eq : betaij ] ) is of form ( [ eq : form b ] )
. here is a simple example where @xmath163 is of the form ( [ eq : form b ] ) but not of the usual form ( [ eq : wishart b ] ) : let @xmath267 and @xmath268 it can be easily checked that ( [ eq : betaij ] ) is satisfied , while @xmath163 can not be brought into the form ( [ eq : wishart b ] ) . if @xmath269 models the ( squared ) volatility of the @xmath243th stock price , as in ( [ stockex ] ) , then this drift specification admits level impacts of the volatility of stock 1 on the volatility of stock 2 , and vice versa . see remark [ remcons ] .
condition ( [ eq : m ] ) means that jumps described by @xmath140 , which can for instance appear at @xmath270 , should be of finite variation entering the cone @xmath1 , since infinite variation transversal to the boundary would let the process leave the state space .
similarly , condition ( [ eq : mu ] ) asserts finite variation for the inward pointing directions , while we could a priori have a general jump behavior ( supported by @xmath1 due to the affine structure ) parallel to the boundary .
note that in the case @xmath237 , which corresponds to @xmath271 , the linear jump part can have infinite total variation ( see @xcite , equation ( 2.11 ) ) .
however , due to the geometry of the cone @xmath1 , we conjecture that in higher dimensions @xmath224 such a behavior is no longer possible and that all jumps are in fact of finite total variation . in any case , for @xmath224 , affine positive matrix valued diffusion processes can not be approximated ( in law ) by pure jump processes , since this would yield a contradiction to condition ( [ eq : b ] ) .
see also remark [ remapproxinfdiv ] below .
suppose @xmath9 is an affine process on @xmath1 .
the main result of this section is that @xmath9 is regular in the sense of definition [ def : affine process sd+ ] .
in addition , we shall prove that @xmath272 is a feller semigroup on @xmath158 . in order to show both properties , we shall mainly rely on lemma [ prop : feller prop ] below .
the feller property is then a simple consequence of this statement and regularity is obtained by arguing as in keller - ressel , schachermayer and teichmann @xcite , who obtained the corresponding statements for affine processes on the state space @xmath30 ; see @xcite , theorem 4.3 , and also the ph.d .
thesis of keller - ressel @xcite .
we observe that most arguments of @xcite translate to our setting without major changes .
it is only required to tailor some technicalities to the cone @xmath1 .
we start with the following elementary observations .
[ lem : boundary ] if @xmath273 and @xmath274 , then @xmath275 .
let @xmath276 such that @xmath261 .
then , @xmath274 implies @xmath277 . hence , @xmath278 .
we now derive some first properties of the functions @xmath25 and @xmath26 in ( [ def : affine process ] ) .
[ lem : order preserving affine ] let @xmath9 be an affine process on @xmath1 .
then , we have : the functions @xmath25 and @xmath26 satisfy @xmath279 for all @xmath280 . for all @xmath281 with @xmath282 and for all @xmath128 , the order relations @xmath283 hold true .
the functions @xmath25 and @xmath26 are jointly continuous in @xmath284
. furthermore , @xmath285 and @xmath286 are analytic on @xmath62 .
assertion ( i ) follows directly from the chapman
kolmogorov equation , @xmath287 for the proof of ( ii ) , note that @xmath288 is equivalent to @xmath289 for all @xmath127 . by the monotonicity of the exponential function
, we have for all @xmath115 and for all @xmath290 , @xmath291 and the assertion follows by taking logarithms . concerning statement ( iii ) ,
note that stochastic continuity of @xmath9 implies joint continuity of @xmath292 in @xmath293 ( this follows , e.g. , from @xcite , lemma 23.7 ) , for all @xmath115 .
this in turn yields continuity of the functions @xmath294 and @xmath295 .
the second assertion follows from analyticity properties of the laplace transform .
the following property of @xmath26 is crucial .
[ prop : feller prop ] let @xmath296 be any map satisfying @xmath297 and the properties of lemma [ lem : order preserving affine ] ( regarding the function @xmath298 ) .
then @xmath299 for all @xmath300 .
we adapt the proof of @xcite , proposition 1.10 , to our setting .
assume by contradiction that there exists some @xmath301 such that @xmath302 .
let us consider the interval @xmath303\neq\varnothing$ ] , where @xmath304 denotes the smallest eigenvalue of @xmath305 .
then for all @xmath306 $ ] we have @xmath307 . since @xmath124 admits property ( ii ) of lemma [ lem : order preserving affine ] ,
we obtain @xmath308 for all @xmath309 $ ] .
consequently , lemma [ lem : boundary ] yields that @xmath310 .
hence , @xmath311 for all @xmath309 $ ] .
the analyticity of @xmath312 on @xmath62 carries over to @xmath313 and implies that @xmath314 for all @xmath315 .
indeed , the set of zeros of @xmath316 has an accumulation point in @xmath317 , which implies that @xmath316 vanishes entirely on @xmath317 .
the same statement holds true for @xmath114 replaced by @xmath318 . indeed ,
if @xmath319 , then the assertion is shown by the same arguments as above . otherwise ,
if @xmath320 , we have for all @xmath315 with @xmath321 , that is , for all @xmath322 $ ] @xmath323 which yields again @xmath324 and @xmath325 for all @xmath326 , @xmath327 $ ] . the same reasoning as before then leads to @xmath325 for all @xmath328
by reapplying this argument , we finally get for every @xmath329 and for all @xmath315 @xmath330 from the continuity of the function @xmath331 and of the determinant , we deduce that for any @xmath315 , @xmath332 a contradiction , and the assertion is proved
. we may now formulate the main result of this section .
[ th : regularity ] let @xmath9 be an affine process with state space @xmath1 . then , we have : @xmath9 is a feller process .
@xmath9 is regular . in order to prove ( i ) , it suffices to show that for all @xmath333 @xmath334 ( see , e.g. , @xcite , propostion iii.2.4 ) .
property ( [ eq : fellerprop1 ] ) is a consequence of stochastic continuity , which implies for all @xmath333 and @xmath115 @xmath335 concerning ( [ eq : fellerprop2 ] ) , it suffices to verify this property for a dense subset of @xmath158 . by a locally compact version of stone
weierstrass theorem ( see , e.g. , @xcite ) , the linear span of the set @xmath336 is dense in @xmath158 .
indeed , it is a subalgebra of @xmath158 , separates points and vanishes nowhere , as all elements are strictly positive functions on @xmath1 . from lemma [ prop : feller prop ]
, we can deduce that @xmath337 if @xmath338 , since @xmath339 and @xmath340 for @xmath341 implying that @xmath342 goes to @xmath133 as @xmath343 .
hence , statement ( i ) is proved .
the proof of ( ii ) follows precisely the lines of @xcite , proof of theorem 4.3 . using lemma [ prop : feller prop ]
, one may mimic the proof of @xcite , theorem 4.3 , to obtain that differentiability of @xmath124 in @xmath155 , which follows from lemma [ lem : order preserving affine](iii ) , implies differentiability of @xmath124 in @xmath114 for @xmath344 and for all @xmath345 . by the regularity of @xmath9
, we are now allowed to differentiate the equations ( [ eq : flow phi ] ) and ( [ eq : flow psi ] ) with respect to @xmath114 and evaluate them at @xmath344 . as a consequence , @xmath25 and
@xmath26 satisfy the system of differential equations @xmath346 where @xmath347 and @xmath348 are defined as in ( [ eq : f , r ] ) .
the analysis of these ( generalized riccati ) differential equations is subject of section [ section : riccati ] , whereas the specific form of @xmath347 and @xmath348 is elaborated in the following .
in this section , we derive _ necessary _ parametric restrictions , that is , given an affine process on @xmath1 , we determine necessary implications on a set of parameters which only ensue from definition [ def : affine process sd+ ] .
these conditions are precisely the conditions on the admissible parameter set as of definition [ def : necessary admissibility ] .
the form of the functions @xmath347 and @xmath348 as defined by ( [ eq : f , r ] ) is then characterized by means of this parameter set , which is stated in proposition [ th : necessary admissibility ] below . for its proof , we first provide a number of technical prerequisites .
[ lem : zero divisor ] let @xmath260 and @xmath349 be the diagonalization of @xmath84 with @xmath350 and @xmath351 .
then the following assertions are equivalent : @xmath352 , @xmath353 , @xmath305 is of form @xmath354 with @xmath355 .
the direction ( i)@xmath356(ii ) is obvious . in order to prove the implication ( ii)@xmath356(iii ) ,
define @xmath357 as @xmath358
. then we have @xmath359 which implies @xmath360 for all @xmath361 and by the positive definiteness of @xmath357 it must then be of form @xmath362 with @xmath355 .
thus , @xmath305 is given by ( [ eq : form zerodivisor ] ) .
this then implies that @xmath352 , which proves the direction ( iii)@xmath356(i ) .
[ lemprojector ] let @xmath363 be an orthogonal projector , that is , @xmath364 and @xmath365 ( see , e.g. , kato @xcite , section ) , and define @xmath366 .
then @xmath367 is an orthogonal projector and the orthogonal complement of @xmath363 in @xmath1 equals @xmath368 that @xmath367 is an orthogonal projector follows by inspection .
the diagonalization of @xmath363 is of the form @xmath369 with @xmath370 , and thus @xmath371 . in view of lemma [ lem : zero divisor ] , we conclude that @xmath372 is orthogonal to @xmath363 if and only if @xmath373 .
this proves the assertion .
[ lem : linear map ] let @xmath305 be in @xmath52 and @xmath374 such that @xmath352 .
then , the linear map @xmath375 defined by @xmath376 has the following properties : @xmath375 is self - adjoint and @xmath377 .
there exists an element @xmath378 such that @xmath379 .
the assertion ( i ) is obvious , since for every @xmath380 , @xmath381 and @xmath382 if @xmath383 . for proving part ( ii )
, we use the fact that @xmath84 is of form ( [ eq : form x ] ) and that all zero divisors @xmath305 in @xmath52 of @xmath84 can be represented by ( [ eq : form zerodivisor ] ) with @xmath384 . thus , setting @xmath385 where @xmath386 satisfies @xmath387 , yields @xmath379 .
[ lem : linear map cone ] let @xmath388 denote a vector space below , @xmath388 corresponds to @xmath52 , the vector space of linear maps @xmath389 , or the vector space of finite signed measures on @xmath52 . ] over @xmath74 .
let @xmath390 be an additive ( resp .
, homogeneous additive ) map , that is , for all @xmath391 and @xmath392 ( resp .
, for all @xmath393 ) we have @xmath394 then @xmath395 is the restriction of an additive ( resp . , @xmath396-linear ) map on @xmath52 .
we define the map @xmath397 as @xmath398 @xmath399 is well defined , as for @xmath400 such that @xmath401 we have @xmath402 since @xmath403 , the domain of @xmath399 is all of @xmath52
. also , @xmath404 by the additivity of @xmath405 .
hence , @xmath406 is the restriction of @xmath399 to @xmath1 .
homogeneity of @xmath399 holds , as for @xmath407 we have by definition @xmath408 finally , we show additivity of @xmath399 . choose @xmath409 such that @xmath410 , hence @xmath411 . by the definition of @xmath412 , we have @xmath413 and by the additivity of @xmath405 we obtain @xmath414 we now provide a convergence result for laplace transforms ( in fact laplace fourier transforms ) , which is most relevant for the analysis of affine processes .
[ lem : fourierlaplace ] let @xmath415 be a sequence of measures on @xmath52 with @xmath416 pointwise , for some finite function @xmath405 on @xmath1 continuous at @xmath417 .
then @xmath415 converges weakly to some finite measure @xmath418 on @xmath52 and the fourier
laplace transforms converge for @xmath419 and @xmath420 to the fourier
laplace transforms of @xmath421 , that is , @xmath422 in particular , @xmath423 and @xmath424 for all @xmath425 . instead of @xmath417
we could take any set @xmath426 of points at the boundary @xmath427 : if we assume continuity of @xmath405 at points in @xmath426 , then we obtain the equality of @xmath428 with the laplace transform of @xmath421 for all points in @xmath426 . additionally , continuity is too strong an assumption , since we only need right continuity of @xmath405 along the segment @xmath429 for @xmath430 at the points from the boundary under consideration .
proof of lemma [ lem : fourierlaplace ] since @xmath431 is bounded , we know by general theory that @xmath432 has a vague accumulation point @xmath418 , which is a finite measure on @xmath52 . since @xmath433 on @xmath1
, it follows by well - known regularity properties of laplace transforms ( see , e.g. , @xcite , lemma 10.8 ) that the functions @xmath434 admit an analytic extension on the strip @xmath435 , still denoted by @xmath434 : @xmath436 moreover , pointwise convergence of the finite convex functions @xmath434 to @xmath405 on @xmath1 implies that this convergence is in fact uniform on compact subsets of @xmath437 ( see , e.g. , rockafellar @xcite , theorem 10.8 ) .
hence , the functions @xmath434 are uniformly bounded on compact subsets of @xmath62 and since @xmath438 , also on compact subsets of @xmath439 .
therefore , and since @xmath437 is a set of uniqueness in @xmath440 , it follows by vitali s theorem ( @xcite , chapter 1 , proposition 7 ) that the analytic functions @xmath434 converge uniformly on compact subsets of @xmath435 to an analytic limit thereon . by lvy s continuity theorem
, we therefore know that for any @xmath441 the finite measures @xmath442 converge weakly to a limit , which by uniqueness of the weak limit has to equal @xmath443 .
whence the only vague accumulation point of @xmath432 is @xmath421 .
vague convergence implies weak convergence if mass is conserved .
continuity of @xmath444 at @xmath417 implies this mass conservation : indeed , by weak convergence of @xmath445 we arrive at @xmath446 and therefore by dominated convergence we obtain that the limit @xmath447 yields @xmath448 which is the desired mass conservation , hence weak convergence , which means in turn convergence of the fourier laplace transform at @xmath417 .
finally , let us state a general comparison result for odes and hereto introduce the notion of _ quasi - monotonicity _ , which we shall need several times throughout this article , in particular in the proofs of propositions [ th : necessary admissibility ] and [ prop_ricc_sol ] below .
[ def : quasimono ] let @xmath449 be an open set .
a function @xmath450 is called _ quasi - monotone increasing _ if for all elements @xmath451 , @xmath452 which satisfy @xmath68 and @xmath453
, @xmath454 holds true .
accordingly , we call @xmath92 _ quasi - constant _ if both @xmath92 and @xmath455 are quasi - monotone increasing
. the following comparison result can be deduced from a more general theorem proved by volkmann @xcite .
[ th : volkmann ] let @xmath449 be an open set .
let @xmath456 be a continuous locally lipschitz map such that @xmath457 is quasi - monotone increasing on @xmath90 for all @xmath458 .
let @xmath459 and @xmath460 be differentiable maps such that @xmath461 and @xmath462 then we have @xmath463 for all @xmath464 .
the main result of this section characterizes the form of the functions @xmath347 and @xmath348 as defined by ( [ eq : f , r ] ) .
[ th : necessary admissibility ] let @xmath9 be an affine process with state space @xmath1 .
then there exist parameters @xmath465 , where @xmath466 satisfy the admissibility conditions of definition [ def : necessary admissibility ] and @xmath467 , such that the functions @xmath347 and @xmath348 are of the form ( [ eq : f ] ) and ( [ eq : r ] ) . note that for the moment we only obtain @xmath468 , and not ( [ eq : b ] ) . proof of proposition [ th : necessary admissibility ] as the proof of proposition [ th : necessary admissibility ] is rather long , we divide it into several steps : _ step _ 1 . _ necessary admissibility conditions for @xmath469 . _ in order to derive the particular form of @xmath347 and @xmath348 with the above parameter restrictions , we follow the approach of keller - ressel @xcite , theorem 2.6 .
note that the @xmath114-derivative of @xmath470 at @xmath344 exists for all @xmath471 , since @xmath472\\[-8pt ] & = & \bigl(-f(u)-\langle r(u),x\rangle\bigr)e^{-\langle u , x\rangle}\nonumber \ ] ] is well defined by proposition [ th : regularity ] .
moreover , we can also write @xmath473 by the above equalities and the fact that @xmath474 , we then obtain for @xmath130 @xmath475 setting @xmath476 and @xmath477 yields @xmath478 as in ( [ eq : c ] ) and @xmath479 as in ( [ eq : gamma ] ) .
we thus obtain @xmath480\\[-8pt ] & & \qquad=\lim_{t\rightarrow0^+ } \frac{1}{t}\int_{s_d^+-x}\bigl(e^{-\langle u,\xi\rangle}-1\bigr)p_t(x , d\xi+x).\nonumber\end{aligned}\ ] ] for every fixed @xmath481 , the right - hand side of ( [ eq : frinfdiv ] ) is the logarithm of the laplace transform of a compound poisson distribution supported on @xmath482 with @xmath483 and compounding distribution @xmath484 . concerning the support , note that the compounding distribution is concentrated on @xmath485 , which implies that the compound poisson distribution has support on the convex cone @xmath486 . by lemma [ lem : fourierlaplace ] ,
the pointwise convergence of ( [ eq : frinfdiv ] ) for @xmath487 to some function being continuous at @xmath133 , implies weak convergence of the compound poisson distributions to some infinitely divisible probability distribution @xmath488 supported on @xmath482 . indeed
, this follows from the fact that any compound poisson distribution is infinitely divisible and the class of infinitely divisible distributions is closed under weak convergence ( @xcite , lemma 7.8 ) .
again , by lemma [ lem : fourierlaplace ] the laplace transform of @xmath488 is then given as exponential of the left - hand side of ( [ eq : frinfdiv ] ) .
in particular , for @xmath270 , @xmath489 is an infinitely divisible distribution with support on the cone @xmath1 . by the lvy
khintchine formula on proper cones ( see @xcite , theorem 3.21 ) , its laplace transform is therefore of the form @xmath490 where @xmath468 and @xmath140 is a borel measure supported on @xmath1 such that @xmath491 yielding ( [ eq : m ] ) .
therefore , @xmath492 _ step _ 2 . _ necessary admissibility conditions for @xmath493 .
_ we next obtain the particular form of @xmath348 .
observe that for each @xmath127 and @xmath494 , @xmath495 is the laplace transform of the infinitely divisible distribution @xmath496 , where @xmath497 denotes the @xmath498 convolution power . for @xmath499
, these laplace transforms obviously converge to @xmath500 pointwise in @xmath305 . using again the same arguments as before [ an application of lemma [ lem : fourierlaplace ] as below ( [ eq : frinfdiv ] ) ]
, we can deduce that @xmath496 converges weakly to some infinitely divisible distribution @xmath501 on @xmath482 with laplace transform @xmath502 for @xmath452 . by the lvy khintchine formula on @xmath52 ( @xcite , theorem 8.1 , indeed on @xmath503 by modifying the scalar product appropriately )
, the characteristic function of @xmath504 has the form @xmath505 for @xmath506 , where @xmath235 is a symmetric positive semidefinite linear operator on @xmath52 , @xmath507 , @xmath508 a measure on @xmath509 satisfying @xmath510 and @xmath135 some appropriate truncation function .
furthermore , by @xcite , theorem 8.7 , @xmath511\\[-9pt ] & & \qquad\stackrel{t \rightarrow0}{\longrightarrow } \int_{s_d\setminus\{0\}}f(\xi)m(d\xi)+\int_{s_d\setminus\{0\ } } f(\xi)m(x , d\xi)\nonumber\end{aligned}\ ] ] holds true for all @xmath512 which are bounded , continuous and vanishing on a neighborhood of @xmath133 .
we conclude that @xmath164 has support in @xmath485 .
therefore , the characteristic function @xmath513 admits an analytic extension to @xmath514 , which then has to coincide with the laplace transform for @xmath452 .
we conclude that , for all @xmath115 , @xmath515 & & \qquad=\frac{1}{2}\langle u , a(x ) u \rangle-\langle b(x),u\rangle\\[-1pt ] & & \qquad\quad{}+\int_{s_d\setminus\{0\ } } \bigl(e^{-\langle u , \xi\rangle}-1+\langle \chi(\xi ) , u \rangle\bigr)m(x , d \xi),\qquad u\in s_d^+.\nonumber\end{aligned}\ ] ] as the left - hand side of ( [ eq : levy khintchine r ] ) is linear in the components of @xmath84 , it follows that @xmath516 , @xmath517 as well as @xmath518 for every @xmath519 are homogeneous additive maps on @xmath1 in the sense of lemma [ lem : linear map cone ] .
this then implies that they are restrictions of linear maps on @xmath52 , such that we can write @xmath520 \int_e(\|\xi\|^2 \wedge1 ) m(x , d\xi)&=&\langle x , \mu(e)\rangle=\sum_{i , j}\mu_{ij}(e)x_{ij},\end{aligned}\ ] ] where ( recall that @xmath57 denotes the standard basis of @xmath52 defined in section [ subsec : notation ] ) : @xmath521 \beta^{ij}&=&\beta^{ji}=(1+\delta_{ij})\frac{b(c^{ij})}{2}\in s_d\end{aligned}\ ] ] and @xmath522 are finite signed measures on @xmath509 .
the fact that @xmath508 is a nonnegative measure for each @xmath115 implies immediately that @xmath523 is a positive semidefinite matrix . in ( [ eq : levymeasure limit ] ) ,
take now @xmath524 and nonnegative functions @xmath525 with @xmath526 on @xmath527 .
then for each @xmath528 the left - hand side of ( [ eq : levymeasure limit ] ) is zero since the @xmath529 is concentrated on @xmath527 . as @xmath530 , the first integral on the right vanishes as well .
hence , @xmath531 for any nonnegative function @xmath532 with @xmath526 on @xmath527 implies that @xmath533 for each @xmath528 .
thus , we can conclude that @xmath534 for all @xmath535 .
now let @xmath536 be any linear map with the property @xmath537
. then @xmath538 .
this implies that the pushforward @xmath539 of @xmath504 under @xmath540 is an infinitely divisible distribution supported on @xmath1 . by the lvy khintchine formula on proper cones ( see @xcite , theorem 3.21 , and by @xcite , proposition 11.10 ) this implies that for all @xmath127 @xmath541 where @xmath542 denotes some truncation function associated with @xmath539 and @xmath543 the pushforward of @xmath544 under @xmath540 .
due to ( [ eq : projlevy ] ) , we can set @xmath545 .
thus , ( [ eq : tborig ] ) becomes @xmath546 moreover , equations ( [ eq : tat ] ) , ( [ eq : tb ] ) and ( [ eq : projlevy ] ) are equivalent to @xmath547 in particular , we claim that @xmath548\\[-8pt ] & & \eqntext{\mbox{for all } u \in s_d \mbox { s.t . } ux = xu=0,}\\ \label{eq : zero divisor zerob } \hspace*{-52pt}\langle b(x ) , u\rangle- \int_{s_d^+\setminus\{0\ } } \langle\chi(\xi),u\rangle m(x , d \xi ) & \geq & 0\nonumber\\[-8pt]\\[-8pt ] \eqntext{\mbox{for all } u\in s_d^+ \mbox { s.t . }
ux = xu=0,}\\ \label{eq : zero divisor levy } \int_{s_d^+\setminus\{0\}}\langle\chi(\xi),u \rangle m(x , d\xi ) & < & \infty \nonumber\\[-8pt]\\[-8pt ] & & \eqntext{\mbox{for all } u \in s_d^+ \mbox { s.t . }
ux = xu=0.}\end{aligned}\ ] ] indeed , if @xmath84 is invertible then @xmath549 is equivalent to @xmath130 and the assertions are obvious . otherwise , if @xmath84 is in @xmath550 , the linear map @xmath375 defined in lemma [ lem : linear map ] is self - adjoint and satisfies @xmath551 .
furthermore , by lemma [ lem : linear map](ii ) , there exists an element @xmath378 such that @xmath379 .
hence , @xmath552 it follows from the proof of lemma [ lem : linear map ] that for @xmath246 , @xmath357 is an element of @xmath1 as well and we have @xmath553 and @xmath554 . equation ( [ eq : zero divisor levy ] ) is obtained by choosing @xmath555 in ( [ eq : projlevy 1 ] ) .
indeed , @xmath556 from these arguments and lemma [ lem : zero divisor ] , properties ( [ eq : betaij ] ) and ( [ eq : mu ] ) can be derived so far .
thus , only ( [ eq : alpha ] ) remains to be shown .
_ step _ 3 . _
necessary admissibility condition for @xmath239 .
_ due to the linearity of @xmath235 , @xmath557 can be written as @xmath558 , where the @xmath253th component of @xmath559 is defined by @xmath560 .
note that @xmath561 is defined on all of @xmath52 .
given that for all @xmath562 , @xmath235 is a positive semidefinite operator on @xmath52 , @xmath563 for all @xmath241 and therefore , by the self duality of @xmath1 , @xmath564 . by ( [ eq : zero divisor zero ] ) , we have for all @xmath305 such that @xmath352 @xmath565 next , we show that @xmath561 is quasi - constant , that is , @xmath566 for all @xmath567 with @xmath568 ( see definition [ def : quasimono ] )
. indeed , pick @xmath569 with @xmath568 . according to our assumptions , @xmath570 , due to ( [ eq : vartheta ] ) and the positivity of @xmath571 . hence ,
@xmath572 where the second last equality holds in view of the symmetry of @xmath235 .
we now claim that there exists some @xmath573 such that @xmath574 , for each .
it is sufficient to show that this statements holds for all orthogonal projectors @xmath575 , that is , there exists some @xmath573 such that @xmath576 for all orthogonal projectors @xmath363 . indeed ,
if this is the case , we can derive the general statement in the following way : take @xmath577 , then by spectral decomposition
there are numbers @xmath578 and orthogonal projectors @xmath579 , which are mutually orthogonal , such that @xmath580 ( see , e.g. , kato @xcite , section i.6.9 ) .
since the assertion holds for all orthogonal projectors , we have that @xmath581 by the property that @xmath582 is quadratic . since @xmath583 is again an orthogonal projector ,
we obtain the result .
we prove the assertion on orthogonal projectors by quasi - constancy .
take an arbitrary orthogonal projector @xmath584 and define @xmath585 .
additionally , we define @xmath586 . by quasi - constancy , we obtain @xmath587 and @xmath588 for all @xmath589 with @xmath590 and @xmath591 . therefore , @xmath592 and @xmath593 are orthogonal to the orthogonal complement of @xmath584 in @xmath594 ( i.e. , the positive symmetric matrices of the form @xmath595 by lemma [ lemprojector ] ) , and @xmath596 and @xmath597 are orthogonal to the orthogonal complement of @xmath598 in @xmath594 ( the positive symmetric matrices of the form @xmath599 by lemma [ lemprojector ] ) .
this means that we can write @xmath600 where the symmetric matrix @xmath601 is orthogonal to all elements which are orthogonal to @xmath584 and @xmath602 ( in @xmath603 ) , that is , @xmath604 is orthogonal to the linear span of matrices of the form @xmath605 and @xmath606 .
however , such a decomposition is unique , since all vectors in the sum are mutually orthogonal , and the decomposition is given by @xmath607 therefore , we can conclude the assertion @xmath608 .
since @xmath584 was arbitrary the assertion is proved .
finally , all the derived restrictions on the parameters together with ( [ eq : levy khintchine r ] ) then yield ( [ eq : r ] ) . an alternative proof for the special form of the diffusion matrix @xmath235 can also be established by stokes theorem @xcite on riccati odes .
the aim of this section is to prove the form of the infinitesimal generator as stated in ( [ eq : generator ] ) .
[ th : generator ] the infinitesimal generator @xmath609 of an affine process on @xmath1 satisfies @xmath610 and is of the form ( [ eq : generator ] ) for all @xmath161 and @xmath127 . as already mentioned in the proof of proposition [ th : necessary admissibility ] , the @xmath114-derivative of @xmath470 at @xmath344 exists pointwise for all @xmath611 and is given by ( [ asharpeux ] ) .
furthermore , @xmath612 , for @xmath613 . as @xmath614 is a feller semigroup on @xmath158 , it follows from @xcite , lemma 31.7 , that @xmath615 and @xmath616 combined with proposition [ th : necessary admissibility ] , we thus obtain @xmath617 indeed , in order to obtain the form of the diffusion part , observe that we have by symmetrization @xmath618 see ( [ eq : cijkl ] ) . according to theorem [ th : density ] , the linear hull @xmath619 of @xmath620 is dense in @xmath621 with respect to the family of seminorms @xmath622 defined in ( [ eq : seminorm ] ) .
denoting the right - hand side of ( [ eq : generator ] ) by @xmath623 , we now claim that for every @xmath624 @xmath625 where @xmath626 such that @xmath627 for every @xmath103 .
indeed , this is obvious for the differential operator part of @xmath623 . by choosing @xmath628 and by denoting @xmath629
, we obtain the following estimate for the integral part : @xmath630 where @xmath631 and @xmath632 denote some constants and @xmath633 correspond to the jordan decomposition @xmath176 . in the second last inequality ,
we use the estimate @xmath634 . the same as above
can be shown for the measure @xmath635 , whence ( [ eq : asharp ] ) holds true . as by the first part of the proof , we have @xmath636 for all elements of @xmath637 , ( [ eq : asharp ] ) implies @xmath638 since the infinitesimal generator of every feller process is a closed operator , it follows that @xmath639 and @xmath640 on @xmath621 . in this subsection
, we shall deal with linear transformations of affine processes .
the proposition below states how the parameters of an affine process on @xmath1 change under such linear maps , which allows us to establish a canonical representation of an affine process .
[ prop : lin trafo ] suppose @xmath9 is an affine process on @xmath1 with parameters @xmath641 , @xmath642 as specified in definition [ def : necessary admissibility ] and @xmath468 .
furthermore , let @xmath643 be an automorphism , where @xmath644 is invertible .
then , @xmath645 is an affine process on @xmath1 , whose parameters , denoted by @xmath646 , are given as follows with respect to the truncation function @xmath647 : @xmath648 where @xmath649 ( @xmath650 ) is the pushforward of the measure @xmath140 ( @xmath203 , resp . ) .
let us consider the process @xmath651 for which we have @xmath652\\ & & \qquad= { { \mathbb e}}\bigl[\exp\bigl(-\bigl\langle u , gx_t^{g^{-1}y(g^\top)^{-1}}g^\top\bigr\rangle \bigr)\bigr]\\ & & \qquad={{\mathbb e}}\bigl[\exp\bigl(-\bigl\langle g^\top u g , x_t^{g^{-1}y(g^\top)^{-1 } } \bigr\rangle\bigr)\bigr]\\ & & \qquad=\exp\bigl(-\phi(t , g^\top u g)-\langle\psi(t , g^\top u g),g^{-1}y(g^\top)^{-1}\rangle\bigr)\\ & & \qquad=\exp\bigl(-\phi(t , g^\top u g)-\langle(g^\top)^{-1}\psi(t , g^\top u g)g^{-1},y\rangle\bigr).\end{aligned}\ ] ] define now @xmath653 and @xmath654 by @xmath655 to see that @xmath656 is an affine process on @xmath1 . using ( [ eq : f - riccati ] ) and ( [ eq : f ] ) , we consequently obtain @xmath657 due to the uniqueness of the lvy khintchine decomposition
, this implies that @xmath11 transforms to @xmath658 , @xmath659 remains constant and @xmath140 becomes @xmath660 .
for @xmath654 we proceed similarly , that is , we have @xmath661 from which it can be seen that @xmath239 transforms to @xmath662 , @xmath663 becomes @xmath664 , and @xmath203 changes to @xmath665 for every @xmath666 .
moreover , since @xmath667 @xmath668 by means of proposition [ prop : lin trafo ] , we can derive a canonical representation for affine processes .
[ prop : canonic ] let @xmath9 be an affine process on @xmath1 with parameters @xmath669 , @xmath670 as specified in definition [ def : necessary admissibility ] and @xmath468 .
then there exists an automorphism @xmath671 such that the parameters of the affine process @xmath672 , denoted by @xmath646 , are as in proposition [ prop : lin trafo ] with @xmath673 where we define @xmath674 by proposition [ prop : lin trafo ] , the parameters of @xmath672 transform as @xmath675 since @xmath239 and @xmath468 , they are jointly diagonalizable through an automorphism on @xmath1 .
more precisely , there exists an invertible matrix @xmath676 such that @xmath677 and @xmath678 where @xmath679 denotes the rank of a matrix . for the proof of this fact
, we refer to @xcite , theorem 8.7.1 .
this subsection is devoted to show that condition ( [ eq : b ] ) holds true for any affine process @xmath9 on @xmath1 . since the automorphism @xmath680 in proposition [ prop : canonic ] is order preserving , it suffices to consider affine processes of the canonical form as specified in proposition [ prop : canonic ] .
the following result is a consequence of the lvy khintchine formula on @xmath681 .
[ lemprop : characteristicsdet ] let @xmath656 be an affine process of canonical form as specified in proposition [ prop : canonic ] with parameters denoted by @xmath646 . then , for any @xmath682 , we have @xmath683 and @xmath684\\[-8pt ] & & \qquad{}+\frac{1}{2}\sum_{i , j , k , l}\widetilde{a}_{ijkl}(y)\,\partial _ { ij}\,\partial_{kl}\det(y ) \geq0.\nonumber\end{aligned}\ ] ]
let @xmath682 and let @xmath685 be a function with @xmath686 and @xmath687 for all @xmath84 in a neighborhood of @xmath688 .
then , for any @xmath689 , the function @xmath690 lies in @xmath691 and thus in @xmath692 , where @xmath693 denotes the infinitesimal generator of @xmath656 .
note that @xmath694 . hence , the limit @xmath695 exists for any @xmath696 , where @xmath697 denotes the transition function of @xmath656 , and @xmath698 is the pushforward of @xmath699 under @xmath92 , which is a probability measure supported on @xmath681 . using the same arguments as in proposition [ th : necessary admissibility ] [ i.e. , applying lemma [ lem : fourierlaplace ] as done below ( [ eq : frinfdiv ] ) ] , and noting that @xmath694 , we conclude that @xmath700 is the logarithm of the laplace transform of an infinitely divisible distribution on @xmath681 . note that @xmath701 hence , @xmath702 and the admissibility condition ( [ eq : mu ] ) implies ( [ mtild0inf ] ) . by the lvy khintchine formula on @xmath681 ( see @xcite , theorem 3.21 ) , the linear coefficient in @xmath357 in ( [ atildelk ] ) has to be nonpositive .
but this is now just ( [ eq : inv det explicit ] ) , whence the lemma is proved .
it now remains to show that ( [ eq : b ] ) follows from ( [ eq : inv det explicit ] ) .
for this purpose , it suffices to evaluate ( [ eq : inv det explicit ] ) at diagonal elements @xmath703 .
thus , we state the following lemma .
[ lem : quadraticvar ] let @xmath704 be diagonal , and let @xmath705 .
then we have @xmath706 obvious .
next , we calculate the partial derivatives of the determinant .
[ lem : detder ] let @xmath707 be diagonal , @xmath708
. then we have @xmath709 and @xmath710 where the empty product is defined to be @xmath711 . in dimension @xmath267 , the assertion is easily checked , as @xmath712 .
therefore , we have @xmath713 as well as @xmath714 for dimension strictly larger than @xmath715 , we employ a combinatorial argument .
recall leibniz s definition of the determinant , @xmath716 where @xmath89 is an element of the permutation group @xmath38 on the set @xmath717 and @xmath718 denotes the signum function on @xmath38 , that is , @xmath719 if @xmath89 is an even permutation and @xmath720 if it is odd .
differentiation of ( [ eq : leibnizdet ] ) with respect to @xmath85 yields @xmath721 thus , for the second derivative we have @xmath722 where the last equality holds since @xmath688 is diagonal . for @xmath723 ,
the statement is obvious .
we are prepared to prove the admissibility condition on the constant drift .
[ th : constant drift ] let @xmath9 be an affine process on @xmath1 , then ( [ eq : b ] ) holds , that is , @xmath724 since the automorphism @xmath680 in proposition [ prop : canonic ] is order preserving , it suffices to show that ( [ eq : inv det explicit ] ) in lemma [ lemprop : characteristicsdet ] implies @xmath725 we show that @xmath726 , if @xmath727 . to this end , take again some diagonal @xmath682 of form @xmath728 . by lemmas [ lem : quadraticvar ] and [ lem : detder ] , we obtain @xmath729 as @xmath730 and by ( [ eq : betaij ] ) also @xmath731 for @xmath732 , letting @xmath733 yields @xmath726 for @xmath727 .
relabeling of indices then proves ( [ eq : canonic drift ] ) .
in this section , we prove that for a given admissible parameter set @xmath239 , @xmath11 , @xmath734 , @xmath659 , @xmath663 , @xmath140 , @xmath203 satisfying the conditions of definition [ def : necessary admissibility ] , there exists a unique affine process on @xmath1 , whose infinitesimal generator @xmath609 is of form ( [ eq : generator ] ) . our approach to derive this result is to consider the martingale problem for the operator @xmath157 . in order to prove uniqueness for this martingale problem
, we shall need the following existence and uniqueness result for the generalized riccati differential equations ( [ eq : f - riccati ] ) and ( [ eq : r - riccati ] ) .
we first derive some properties of the function @xmath348 given in ( [ eq : r ] ) .
[ lem : r quasi monotone ] @xmath348 is analytic on @xmath62 and quasi - monotone increasing on @xmath1 . that @xmath348 is analytic on @xmath62 follows by dominated convergence ( see , e.g. , @xcite , lemma a.2 ) .
now let @xmath735 , and define @xmath736 now , the map @xmath737 is quasi - monotone increasing , as it is shown in step 3 of the proof of proposition [ th : necessary admissibility ] .
furthermore , it follows from the admissibility condition ( [ eq : betaij ] ) that @xmath738 is a quasi - monotone increasing linear map on @xmath1 .
finally , the quasi - monotonicity of @xmath739 is a consequence of the monotonicity of the exponential and that @xmath740 . by dominated convergence
, we have @xmath741 pointwise for each @xmath452 .
hence , the quasi - monotonicity carries over to @xmath348 .
indeed , choose @xmath742 such that @xmath743 and @xmath744 .
then we have for all @xmath745 , @xmath746 , @xmath747 .
thus , @xmath748 as @xmath749 , which proves that @xmath348 is quasi - monotone increasing .
[ lemkestr ] there exists a constant @xmath426 such that @xmath750 we may assume , without loss of generality , that the truncation function in definition [ def : necessary admissibility ] takes the form @xmath751 [ otherwise adjust @xmath752 accordingly ]
. then , for all @xmath452 we have @xmath753\\[-8pt ] & \preceq&-2u\alpha u+b^\top(u)+\gamma+\mu(s_d^+\cap\{\|\xi\| > 1\ } ) \nonumber\\ & \preceq & b^\top(u)+\gamma+\mu(s_d^+\cap\{\|\xi\| > 1\}),\nonumber\end{aligned}\ ] ] where we use that @xmath754 set now @xmath755 by ( [ est : crucial for upper bounds ] ) , we obtain , for @xmath756 , that @xmath757 from which we derive the existence of a positive constant @xmath426 such equation ( [ eq : above estimate r(u ) on cone ] ) holds . here is our main existence and uniqueness result for the generalized riccati differential equations ( [ eq : f - riccati ] ) and ( [ eq : r - riccati ] ) .
[ prop_ricc_sol ] for every @xmath155 , there exists a unique global @xmath758-valued solution @xmath759 of ( [ eq : f - riccati ] ) and ( [ eq : r - riccati ] ) .
moreover , @xmath123 and @xmath124 are analytic in @xmath760 .
we only have to show that , for every @xmath155 , there exists a unique global @xmath62-valued solution @xmath26 of ( [ eq : r - riccati ] ) , as then @xmath25 is uniquely determined by integrating ( [ eq : f - riccati ] ) and has the desired properties by admissibility of the parameter set .
let @xmath155 . since @xmath348 is analytic on @xmath62 , standard ode results ( e.g. , @xcite , theorem 10.4.5 ) yield there exists a unique local @xmath62-valued solution @xmath124 of ( [ eq : r - riccati ] ) for @xmath761 , where @xmath762 it thus remains to show that @xmath763 . that @xmath124 , and hence @xmath123 , is analytic in @xmath760 then follows from @xcite , theorem 10.8.2 .
since @xmath348 may not be lipschitz continuous at @xmath764 ( see remark [ reminwpointing ] below ) , we first have to regularize it .
we thus define @xmath765 it then follows as in lemmas [ lem : r quasi monotone ] and [ lemkestr ] that @xmath766 is quasi - monotone increasing on @xmath1 and that ( [ eq : above estimate r(u ) on cone ] ) holds for some constant @xmath767 .
moreover , @xmath766 is analytic on @xmath52 .
hence , for all @xmath768 , there exists a unique local @xmath52-valued solution @xmath654 of @xmath769 for all @xmath770 with maximal lifetime @xmath771 from ( [ eq : above estimate r(u ) on cone ] ) , we infer that for all @xmath452 and @xmath772 , @xmath773 gronwall s inequality ( e.g. , @xcite , ( 10.5.1.3 ) ) implies @xmath774 hence , @xmath775 for @xmath776 .
as @xmath766 is quasi - monotone increasing on @xmath1 , volkmann s comparison theorem [ th : volkmann ] now yields @xmath777 therefore and since @xmath778 is also analytic in @xmath305 , lemma [ prop : feller prop ] implies that @xmath779 for all @xmath780 .
we now carry this over to @xmath124 and assume without loss of generality , as in the proof of lemma [ lemkestr ] , that the truncation function in definition [ def : necessary admissibility ] takes the form @xmath751 .
then @xmath781 hence , for @xmath155 and @xmath782 , we have @xmath783 theorem [ th : volkmann ] thus implies @xmath784 hence , @xmath785 . using ( [ eq : above estimate r(u ) on cone ] ) again , we now can show as for @xmath654 that @xmath786 hence @xmath787 , as desired .
[ reminwpointing ] lemma [ lem : r quasi monotone ] states that the admissibility of the parameters @xmath788 implies quasi - monotonicity of @xmath348 on @xmath1 .
is quasi - monotone on @xmath1 and @xmath740 if and only if the parameters @xmath788 are admissible .
] moreover , quasi - monotonicity just means that @xmath348 is `` inward pointing '' close to the boundary @xmath1 .
indeed , let @xmath120 with @xmath789
. then @xmath790 .
hence , if @xmath348 were lipschitz continuous on @xmath1 , a deterministic variant of theorem [ th : martingale problem convex ] would imply the invariance of @xmath1 with respect to ( [ eq : r - riccati ] ) right away .
however , the map @xmath348 might fail to be lipschitz at @xmath764 ( see the one - dimensional counterexample @xcite , example 9.3 ) , even though it is analytic on the interior @xmath62 . here , quasi - monotonicity plays the decisive role .
it leads to the phenomenon that @xmath124 stays away from the boundary @xmath550 for @xmath155 , which is of crucial importance in our analysis .
we are now prepared to study the martingale problem for the operator @xmath157 given by ( [ eq : generator ] ) . for the notion of martingale problems ,
we refer to @xcite , chapter 4 .
we shall proceed in four steps .
first , we approximate @xmath609 by regular operators @xmath791 on the space @xmath152 of rapidly decreasing @xmath153-functions on @xmath1 , defined in ( [ defs+ ] ) .
second , using theorem [ th : martingale problem convex ] below , we show that there exists an @xmath1-valued cdlg solution of the martingale problem for @xmath791 .
third , a subsequence of these solutions is shown to converge to an @xmath112-valued cdlg solution of the martingale problem for @xmath609 .
finally , we show that this solution is unique , markov and affine , as desired .
note that we can not employ stroock s @xcite seminal existence and uniqueness results for martingale problems , since those are solved on @xmath792 and require uniform elliptic diffusion parts .
neither of these is satisfied in our case .
now let @xmath793 be some admissible parameter set .
fix some @xmath794 and @xmath795 . in order to bound the coefficients and cut off the small jumps , we let
@xmath796 we then define the bounded and smooth parameters @xmath797 concerning the diffusion function @xmath197 given by ( [ eq : cijkl ] ) , we first find an appropriate factorization which will allow us to write the continuous martingale part of @xmath9 as a stochastic integral .
thereto observe that any @xmath1-valued solution , presumed that it exists , of the following symmetric matrix - valued diffusion sde : @xmath798 where @xmath13 is a standard @xmath198-matrix brownian motion and @xmath200 with @xmath201 , has quadratic variation @xmath799 .
define now @xmath800 by @xmath801 where @xmath802 .
then ( [ eq : wishart diff ] ) can be written as @xmath803 and @xmath804 . since @xmath805 involves the matrix square root , which is neither lipschitz continuous nor bounded nor
globally defined , we need to introduce some approximating regularization in order to meet the assumptions of theorem [ th : martingale problem convex ] below .
thereto fix some truncation function @xmath806 and define @xmath807 note that @xmath808 satisfies : * @xmath809 , * @xmath810 on @xmath1 , * @xmath811 . with this
, we can now define the regularization of @xmath812 by @xmath813 which then satisfies the smoothness condition of theorem [ th : martingale problem convex ] .
finally , we set @xmath814 & = & ( s^2_{\varepsilon , n}(x))_{ik}\alpha_{jl}+(s^2_{\varepsilon , n}(x))_{il}\alpha_{jk}\\[-1.5pt ] & & { } + ( s^2_{\varepsilon , n}(x))_{jk}\alpha _ { il}+(s^2_{\varepsilon , n}(x))_{jl}\alpha_{ik},\nonumber\end{aligned}\ ] ] and define the corresponding regularized operator on @xmath815 @xmath816\\[-8pt ] & & { } + \int_{s_d^+\setminus\{0\ } } \bigl(f(x+\xi)-f(x)\bigr)m^{\delta}(d\xi)\nonumber\\ & & { } + \int_{s_d^+\setminus\{0\}}\bigl(f(x+\xi)-f(x)-\langle \chi(\xi ) , \nabla f(x)\rangle\bigr ) m^{\delta , n}(x , d\xi).\nonumber\end{aligned}\ ] ] we now show that @xmath817 approximates @xmath609 .
we let @xmath818 and @xmath152 denote the locally convex spaces of rapidly decreasing @xmath153-functions on @xmath52 and @xmath1 defined in ( [ defs+ ] ) below , respectively .
[ lemaconva ] @xmath819 and , for every @xmath161 , @xmath820 since @xmath821 as defined in ( [ eq : varphi ] ) converges uniformly on compact sets to 1 , this is clear for the differential operator part .
concerning the integral part , we have @xmath822 by dominated convergence the second term goes uniformly in @xmath84 to @xmath133 , thus we only have to consider the first one . by splitting the first integral into @xmath823
, we note that @xmath824 converges uniformly in @xmath84 to @xmath133 .
hence , it remains to analyze @xmath825 which can be estimated by @xmath826 where @xmath827 correspond to the jordan decomposition of @xmath176 .
as @xmath92 lies in @xmath152 , the second term converges uniformly to @xmath133 . for the first one ,
observe that for every @xmath528 @xmath828 such that we can apply dominated convergence .
again , since @xmath92 lies in @xmath152 , the first integral converges uniformly in @xmath84 to @xmath133 as well .
hence ( [ eq : gen convergence ] ) holds true , and @xmath819 follows similarly .
we now establish existence for the martingale problem for @xmath817 .
[ lemprop : martingale problem approx ] for every @xmath115 there exists an @xmath1-valued cdlg solution @xmath9 to the martingale problem for @xmath817 with @xmath186 .
that is , @xmath829 is a martingale , for all @xmath830 .
consider the following sde of type ( [ eq : jump diffusion ] ) : @xmath831\\[-8pt ] & & { } + \sum_{k , l}^d \int_0^t\sigma^{kl}_{\varepsilon , n}(x^{\varepsilon , \delta , n}_s)\,dw_{s , kl}+j_t,\nonumber\end{aligned}\ ] ] where @xmath13 is a @xmath198-matrix of standard brownian motions and @xmath14 a finite activity jump process with compensator @xmath832 .
note that the quadratic variation of the continuous martingale part of @xmath833 is given by @xmath834 as defined in ( [ eq : approx c ] ) .
it thus follows by inspection that any cdlg solution @xmath835 of ( [ eq : approx sde ] ) solves the martingale problem for @xmath817 .
hence , it remains to show that there exists an @xmath1-valued cdlg solution of ( [ eq : approx sde ] ) .
let us recall the normal cone ( [ eq : normal cone sd+ ] ) to @xmath1 .
as @xmath836 , @xmath837 and @xmath838 are designed to satisfy the assumptions of theorem [ th : martingale problem convex ] and since @xmath839 for all @xmath127 , we only have to show that for all @xmath840 and @xmath841 @xmath842\\[-8pt ] & & \hspace*{45pt}{}-\frac{1}{2}\sum_{k , l=1}^d d\sigma_{\varepsilon , n}^{kl}(x)\sigma_{\varepsilon , n}^{kl}(x),u\biggr\rangle \geq0.\nonumber\end{aligned}\ ] ] due to the definition of @xmath843 , respectively , the definition of @xmath844 given in ( [ eq : sigmavareps ] ) , condition ( [ eq : volapara ] ) is satisfied .
concerning ( [ eq : stratodrift cond ] ) , we have by ( [ eq : betaij ] ) @xmath845 moreover , it is shown in lemma [ prop : strato ] below that @xmath846 the lemma now follows from theorem [ th : martingale problem convex ] .
[ prop : strato ] let @xmath847 where @xmath848 contains the eigenvalues in decreasing order and let @xmath849 be defined by ( [ eq : reg sigmakl ] ) .
then , for all , @xmath850 where @xmath851 and @xmath852 . furthermore , if @xmath853 then @xmath854 for all @xmath374 and for all @xmath841 .
let us denote @xmath855 and notice that @xmath856 we now use the following formula from @xcite , theorem 6.6.30 : @xmath857m^{jj}\biggr ) o(t)^\top,\ ] ] where @xmath858 is a family of symmetric matrices and @xmath859 for @xmath860 and @xmath861 .
this holds true if @xmath862 is continuously differentiable for @xmath863 and @xmath864 is continuously differentiable on an open real interval which contains all eigenvalues of @xmath865 for all @xmath866 .
we now apply this formula to our case , where @xmath867 and @xmath868 since we take the derivative at @xmath344 , we only have to consider @xmath869 where @xmath870 is the orthogonal matrix diagonalizing @xmath84 and @xmath871 note that we do not have an explicit contribution of @xmath872 which is part of the definition of @xmath808 , since @xmath873 and @xmath874 .
some lines of calculations then yield ( [ eq : strato corr ] ) .
let us now verify ( [ eq : inward cond ] ) .
take an arbitrary @xmath875 and assume first that it has rank @xmath876 , that is , @xmath877 and all other eigenvalues are strictly positive . by lemma [ lem : zero divisor ] and
( [ eq : normal cone sd+ ] ) , the elements of @xmath878 can then be written as @xmath879 , where @xmath880 with @xmath881 .
thus , ( [ eq : inward cond ] ) now reads @xmath882_{dd}.\ ] ] as @xmath883_{dd}=2 \delta_{id}(o^{\top}\sigma^{\top}\sigma o)_{id}$ ] and @xmath884 , we have @xmath885_{dd}=\sum_{j\neq d}\frac{\varphi_n(x)(\sqrt{\lambda_j\varphi_n(x ) + \varepsilon}-\sqrt{\varepsilon})}{\sqrt{\lambda_j\varphi _ n(x)+\varepsilon}+\sqrt{\varepsilon}}[o^{\top}\sigma^{\top } \sigma o]_{dd}.\ ] ] since @xmath886 , we obtain by condition ( [ cond : drift eps ] ) @xmath887_{dd}\geq \bigl[o^{\top}\bigl(b-(d-1)\sigma^{\top}\sigma\bigr)o\bigr]_{dd } \geq0,\ ] ] which proves ( [ eq : inward cond ] ) for @xmath374 with @xmath888 . in the general case
, we can proceed similarly . for @xmath840 with @xmath889 ,
the elements of @xmath878 are given by @xmath879 , where @xmath890 with @xmath891 .
this follows again from lemma [ lem : zero divisor ] and ( [ eq : normal cone sd+ ] ) .
now , ( [ eq : inward cond ] ) can be written as @xmath892 which proves the assertion . combining lemmas [ lemaconva ] and [ lemprop : martingale problem approx ] , we obtain the announced existence result for the martingale problem for @xmath609 .
[ lemth : stoch exist ] for every @xmath115 , there exists an @xmath112-valued cdlg solution @xmath9 to the martingale problem for @xmath157 with @xmath186 .
that is , @xmath893 is a martingale , for all @xmath894 .
by lemma [ lemprop : martingale problem approx ] , there exists a solution @xmath835 to the martingale problem for @xmath895 with sample paths in @xmath896 ( the space of @xmath1-valued cdlg paths ) , and hence also in @xmath897 .
we now claim that @xmath898 is relatively compact considered as a sequence of processes with sample paths in @xmath897 .
is relatively compact , that is , the closure of @xmath899 in @xmath900 is compact . here
, @xmath900 denotes the family of probability distributions on @xmath897 and @xmath901 the distribution of @xmath835 . ] for the proof of this assertion , we shall make use of theorems 9.1 and 9.4 in chapter 3 of @xcite . in order to meet the assumption of @xcite , chapter 3 , theorem 9.4
, we take @xmath691 as subalgebra of @xmath902 . then , for every @xmath903 and @xmath685 , we have @xmath904}}|\mathcal{a}^{\varepsilon , \delta , n}f(x_{t}^{\varepsilon , \delta , n})|\bigr ] < \infty,\ ] ] since there exists a constant @xmath905 such that @xmath906 for all @xmath907 , where @xmath622 are the semi - norms as defined in ( [ eq : seminorm ] ) ( see also the proof of proposition [ th : generator ] ) .
thus , the requirements of @xcite , chapter 3 , theorem 9.4 , are satisfied .
note that @xmath656 in the notation of @xcite , chapter 3 , theorem 9.4 , corresponds in our case to @xmath908 such that @xcite , chapter 3 , condition ( 9.17 ) , is automatically fulfilled .
it then follows by the conclusion of @xcite , chapter 3 , theorem 9.4 , that @xmath909 is relatively compact [ as family of processes with sample paths in @xmath910 for each @xmath911 .
furthermore , since we consider @xmath912 , the compact containment condition is always satisfied , that is , for every @xmath913 and @xmath903 , there exists a compact set @xmath914 for which @xmath915$}\bigr ] \geq1-\eta\ ] ] holds true .
by @xcite , chapter 3 , theorem 9.1 , and the fact that @xmath916 is dense in @xmath917 , we therefore obtain that @xmath918 is relatively compact in @xmath897 .
thus , there exists a subsequence @xmath919 of the probability distributions associated to @xmath920 which converges in the prohorov metric to some limit probability distribution .
by @xcite , chapter 3 , theorem 3.1 , this implies weak convergence of @xmath921 and hence the subsequence @xmath922 converges in distribution to some limit process @xmath9 in @xmath923 . combining this with lemma [ lemaconva ] and @xcite , chapter 4 , lemma 5.1 ,
we conclude that @xmath9 is a solution to the martingale problem for @xmath157 .
hence , the lemma is proved .
we can now prove the existence and uniqueness of an affine process for any admissible parameter set .
[ th : existence markov ] let @xmath170 be an admissible parameter set .
then there exists a unique affine process on @xmath1 with infinitesimal generator ( [ eq : generator ] ) , and ( [ def : affine process ] ) holds for all @xmath924 , where @xmath123 and @xmath124 are given by ( [ eq : f - riccati ] ) and ( [ eq : r - riccati ] ) .
suppose first that @xmath172 and @xmath174 .
let @xmath127 .
then lemma [ lemth : stoch exist ] implies the existence of an @xmath112-valued cdlg solution @xmath9 of the martingale problem for @xmath609 with @xmath186 .
we now show that @xmath9 is unique in distribution .
thereto , note that by @xcite , chapter 4 , theorem 7.1 , @xmath925 is a martingale for all rapidly decreasing functions @xmath926 , similarly defined as @xmath152 in ( [ defs+ ] ) .
now let @xmath25 and @xmath26 be the unique solutions of the generalized riccati differential equations ( [ eq : f - riccati ] ) and ( [ eq : r - riccati ] ) , given by proposition [ prop_ricc_sol ] .
fix @xmath481 , @xmath155 , and some @xmath927 such that @xmath928 then @xmath929 in view of ( [ eqmartftxt ] ) , the laplace transform of @xmath188 at @xmath305 is thus given by @xmath930={{\mathbb e}}_x[f(t , x_t)]=f(0,x)-0=e^{-\phi(t , u)-\langle\psi ( t , u),x\rangle}.\ ] ] since @xmath155 was arbitrary , we conclude that the distribution of @xmath188 is uniquely determined for all @xmath481 . from @xcite ,
chapter 4 , theorem 4.2 , we infer that @xmath9 is a markov process with generator @xmath609 on @xmath152 and thus unique in law as solution of the martingale problem for @xmath609 . moreover , by ( [ eqexiproof ] ) , @xmath9 is stochastically continuous and affine .
thus , the proposition is proved under the premise that @xmath172 and @xmath174 . for general parameters @xmath659 and @xmath663
, we employ a feynman kac argument .
denote by @xmath931 and @xmath932 the affine generator and corresponding feller semigroup associated with @xmath793 from the first part of the proof , respectively .
since @xmath933 is nonnegative on @xmath1 , it follows along the lines of @xcite , proposition 11.1 , that @xmath934\ ] ] defines a feller semigroup @xmath614 on @xmath158 with infinitesimal generator @xmath935 for @xmath936 , which is the desired solution . for affine processes without diffusion component (
i.e. , the admissible parameter @xmath239 vanishes ) , the existence question can be handled entirely as in the case of affine processes on @xmath937 @xcite , section 7 . in this section ,
we elaborate an alternative existence proof in this specific case , by following the lines of @xcite .
note that the ou - type processes driven by matrix lvy subordinators @xcite are contained in the class of pure jump processes of this section .
we call a function @xmath938 of lvy khintchine form on @xmath1 , if @xmath939 where @xmath940 and @xmath941 is a borel measure supported on @xmath1 such that @xmath942 once again , we recall that a distribution on @xmath1 is infinitely divisible if and only if its laplace transform takes the form @xmath943 , where @xmath92 is of the above form ( see also step 1 in the proof of proposition [ th : necessary admissibility ] ) . similarly to @xcite , we introduce the sets @xmath944 the following technical statement can be obtained easily by mimicking the proofs of the corresponding statements in @xcite , proposition 7.2 and lemma 7.5 : [ lemma : three dfs statements ] we have : @xmath945 , @xmath946 are convex cones in @xmath947 .
@xmath948 , @xmath949 imply @xmath950 .
@xmath951 imply @xmath952 .
if @xmath953 converges to a continuous function @xmath25 on @xmath1 , then @xmath948 .
a similar statement holds for sequences in @xmath946 .
let @xmath954 be an admissible parameter set .
then @xmath955 locally uniformly as @xmath749 , where @xmath956 corresponds to the admissible parameter set @xmath957 .
( note that there is one fixed truncation function . )
[ prop : c cs semiflow ] let @xmath958 be an admissible parameter set .
then for all @xmath290 , the solutions @xmath959 of ( [ eq : f - riccati ] ) and ( [ eq : r - riccati ] ) lie in @xmath960 .
suppose first that , this would already cover all possible jump measures if @xmath224 . ]
@xmath961 for all @xmath962
. then equation ( [ eq : r - riccati ] ) is equivalent to the integral equation @xmath963 where @xmath964 and @xmath965 is given by @xmath966 here , @xmath967 is the notation for the semi - group induced by @xmath968 , @xmath969 , @xmath970 .
hence , the variation of constants formula yields ( [ eq : var of const ] ) . due to admissibility condition ( [ eq : betaij ] ) , we have that @xmath971 is a linear drift which is `` inward pointing '' at the boundary of @xmath1 , which is equivalent to @xmath972 being a positive semi - group , that is , @xmath973 maps @xmath1 into @xmath1
. therefore , @xmath974 and since @xmath975 is given by @xmath976 with @xmath203 satisfying ( [ finite var ] ) , we also have @xmath977 using picard s iteration and lemma [ lemma : three dfs statements ] , it follows that the sequence @xmath978 defined as @xmath979 lies in @xmath980 , for each @xmath290 , hence so does its limit @xmath981 . since @xmath982
, we have again by lemma [ lemma : three dfs statements ] @xmath983 . by an application of lemma [ lemma : three dfs statements](v ) , the general case is then reduced to the former , since @xmath956 clearly satisfies ( [ finite var ] ) .
we are prepared to provide an alternative proof of proposition [ th : existence markov ] under the additional assumption @xmath226 : by proposition [ prop : c cs semiflow ] , @xmath984 lie in @xmath985 .
hence for all @xmath290 , @xmath115 , there exists an infinitely divisible sub - stochastic kernel @xmath986 with laplace - transform @xmath987 . the chapman
kolmogorov equations hold in view of properties ( [ eq : flow phi ] ) and ( [ eq : flow psi ] ) .
whence , proposition [ th : existence markov ] follows .
[ remapproxinfdiv ] we note that the proof of statement ( v ) in lemma [ lemma : three dfs statements ] is much easier than the one of @xcite , lemma 7.5 , because @xmath226
. however , for @xmath988 and @xmath224 , @xmath348 can not be locally uniformly approximated by functions @xmath989 of a pure jump type such as in lemma [ lemma : three dfs statements ] .
indeed , otherwise one could infer as above the existence of an affine process which is infinitely decomposable and has nonvanishing diffusion component .
this is in contradiction with proposition [ th : char infdec infdiv ] and in the case of pure diffusions it contradicts example [ counterex : bru ] .
the first part is a summary of propositions [ th : regularity ] , [ th : necessary admissibility ] , [ th : generator ] and [ th : constant drift ] .
the second part follows from proposition [ th : existence markov ] .
let @xmath9 be a conservative affine process .
it is shown in proposition [ th : generator ] that @xmath990 .
hence , @xmath991 is a @xmath992-martingale with @xmath993 defined in ( [ eq : filt ] ) . from @xcite , theorem ii.2.42 , combined with ( [ aexpux ] ) and remark [ remcons ]
, it then follows that @xmath9 is a semimartingale with characteristics ( [ eq : char c])([eq : char nu ] ) . the canonical semimartingale representation ( @xcite , theorem ii.2.34 ) of @xmath9 is thus given by @xmath994 where @xmath995 denotes the continuous martingale part , and @xmath202 the random measure associated with the jumps of @xmath9 .
in order to establish representation ( [ eq : bm rep ] ) , we find it convenient to consider the vectorization , @xmath996 , of @xmath995 .
the aim is now to find a @xmath997-dimensional brownian motion @xmath998 on a possibly enlarged probability space and a @xmath999-matrix - valued function @xmath89 such that @xmath1000 thus , @xmath89 has to fulfill @xmath1001\\[-8pt ] & = & ( \sigma(x_t)\sigma^{\top}(x_t))_{ijkl}.\nonumber\end{aligned}\ ] ] as suggested by ( [ sigmakldef ] ) , we define the entries of the @xmath1002-matrix @xmath1003 in terms of @xmath805 given in ( [ sigmakldef ] ) by @xmath1004 note that the @xmath56th column of @xmath1003 is just the vectorization of the matrix @xmath805 .
we thus obtain @xmath1005 .
hence , @xmath1003 satisfies ( [ eq : quad var brow ] ) .
analogous to the proof of @xcite , theorem 20.1 , we can now build a @xmath997-dimensional brownian motion @xmath998 on an enlargement of the probability space such that ( [ eq : cont mart rep ] ) holds true .
as the @xmath253th entry of @xmath995 is given by @xmath1006 where @xmath13 is the @xmath198-matrix brownian motion satisfying @xmath1007 , we obtain the desired representation .
we first prove some technical lemmas .
[ lem : cauchy functional equation ] let @xmath1008 be an additive function , that is , @xmath1009 satisfies cauchy s functional equation @xmath1010 then @xmath1009 can be extended to an additive function @xmath1011 .
moreover , if @xmath1009 is measurable on @xmath1 then @xmath92 is measurable on @xmath52 . in that case
, @xmath92 is a continuous linear functional , that is , @xmath1012 for some @xmath1013 .
the first part follows from lemma [ lem : linear map cone ] .
concerning measurability , let @xmath1014 , a borel measurable set .
then we have by the additivity of @xmath92 , @xmath1015 which is again a measurable set , in view of the measurability of @xmath1009 on @xmath1 . for @xmath83
we write @xmath1016 , where @xmath1017 .
we introduce the additive functions @xmath1018 via @xmath1019 . by the just proved measurability of @xmath92
, we infer that all @xmath1020 are measurable functions on @xmath1021 .
by @xcite , chapter 2 , theorem 8 , any additive measurable function on the real line is a continuous linear functional .
hence for each @xmath243 , we infer the existence of @xmath1022 such that @xmath1023 holds .
since @xmath1024 , it follows that @xmath1025 for some @xmath1026 . also , we consider cauchy s exponential equation for @xmath1027 , that is , @xmath1028 [ lem : cauchy exp eq . ]
suppose @xmath1029 is measurable , strictly positive , and satisfies ( [ eq : multiplicative functionals ] ) .
then @xmath1030 , for some @xmath1026 .
if @xmath1031 , then @xmath1032 . since @xmath1033 is strictly positive , its logarithm yields the well defined function @xmath1008 , @xmath1034
clearly @xmath1009 is additive , hence by the first part of lemma [ lem : cauchy functional equation ] , there exists a unique additive extension @xmath76 .
also , @xmath92 is measurable on @xmath1 , hence by the second assertion of lemma [ lem : cauchy functional equation ] we have @xmath1035 , for some @xmath1026 .
the last statement follows from the monotonicity of the exponential and the self duality of @xmath1 .
[ rem : missinglink ] the assumption of strict positivity of @xmath1033 in the preceding lemma is essential . otherwise , there exist solutions @xmath1033 which are not of the asserted form .
lemma [ lem : cauchy exp eq . ] is the main ingredient of the proof of the following characterization concerning @xmath103-fold convolutions of markov processes .
[ lema : dfslemma10.3 ] let @xmath1036 ( @xmath1037 ) .
then @xmath1038 if and only if for all @xmath1039 and @xmath1040 , , there exists @xmath1041 and @xmath1042 such that @xmath1043 , @xmath1044 and @xmath1045=\rho^{(j)}(\mathbf{t},\mathbf{u } ) e^{-\langle\psi(\mathbf{t},\mathbf{u}),x\rangle}\nonumber\\[-8pt]\\[-8pt ] & & \eqntext{\forall x\in s_d^+ , j=0,1,\ldots , k.}\end{aligned}\ ] ] we proceed similarly as in the proof of @xcite , lemma 10.3 .
fix @xmath1046 , @xmath1047 , @xmath1048 and set @xmath1049.\ ] ] by the definition of the convolution , ( [ eq : kfoldconv ] ) is equivalent to the following : @xmath1050\\[-8pt ] & & \eqntext{\forall x^{(i)}\in s_d^+,\qquad x = x^{(1)}+\cdots+x^{(k)}.}\end{aligned}\ ] ] hence , the implication ( [ eq : char kfoldconv])@xmath356([eq : equivalent to kfoldconv ] ) is obvious . for the converse direction
, we observe that @xmath1051 are strictly positive on all of @xmath1 .
thus , by ( [ eq : equivalent to kfoldconv ] ) we have @xmath1052 and @xmath1009 is a measurable , strictly positive function on @xmath1 satisfying ( [ eq : multiplicative functionals ] ) . hence , an application of lemma [ lem : cauchy exp eq . ]
yields the validity of ( [ eq : char kfoldconv ] ) , where @xmath1053 . by the definition of @xmath1051
, it follows that @xmath1054 and .
we are prepared to prove theorem [ th : char infdec infdiv ] : ( i)@xmath356(ii ) : due to lemma [ lema : dfslemma10.3 ] , infinite decomposability implies that @xmath9 is affine .
also , by the definition of infinite decomposability and by lemma [ lema : dfslemma10.3 ] we have that the @xmath103th root @xmath1055 for each @xmath214 is an affine process with state space @xmath1 with exponents @xmath124 and @xmath1056 .
this implies that @xmath1057 has admissible parameters @xmath1058 .
hence , the admissibility condition proved in proposition [ th : constant drift ] implies @xmath1059 , for each @xmath103 , which is impossible , unless @xmath226 or @xmath237 .
( ii)@xmath356(iii ) : follows from proposition [ prop : c cs semiflow ] , in view of the lvy khintchine form of @xmath1060 , for each @xmath481 .
( iii)@xmath356(i ) : by definition , every transition kernel @xmath1061 of @xmath9 is infinitely divisible with laplace transform @xmath1062 . for each @xmath214 , the maps @xmath1063 , @xmath1064 satisfy the properties ( [ eq : flow phi ] ) and ( [ eq : flow psi ] ) . also , infinite
divisibility implies that for each @xmath1065 , @xmath1066 is the laplace transform of a sub - stochastic measure on @xmath1 . in conjunction with properties ( [ eq :
flow phi ] ) and ( [ eq : flow psi ] ) we may conclude that @xmath1067 gives rise to a feller semigroup on @xmath158 , which is affine in @xmath1068 .
hence , we have constructed for each @xmath214 a @xmath103th root of @xmath9 which is stochastic continuous by the definition of its characteristic exponents @xmath1069 .
thus theorem [ th : char infdec infdiv ] is proved .
in this section , we study existence and viability in a nonempty closed convex set @xmath1070 of solutions to the equation @xmath1071 where @xmath1072 , @xmath1073 are lipschitz continuous maps , @xmath13 a standard @xmath140-dimensional brownian motion and @xmath14 a finite activity jump process with state - dependent , absolutely continuous compensator @xmath1074 .
we further assume that @xmath1075 is bounded .
we tackle this problem in three steps .
first , we derive some regularity and existence results for diffusion sdes .
these results are not in the standard literature , we thus provide full proofs .
second , we prove existence of a cdlg solution @xmath9 for ( [ eq : jump diffusion ] ) . finally , we provide sufficient conditions for @xmath9 to be @xmath1076-valued .
let @xmath1077 be a filtered probability space satisfying the usual conditions and carrying an @xmath140-dimensional standard brownian motion @xmath13 .
we consider the following diffusion sde : @xmath1078 where @xmath1079\times{{\mathbb r}}^n$ ] and @xmath11 and @xmath89 are as above . recall that @xmath9 is a solution of ( [ sde ] ) if @xmath9 is continuous and ( [ sde ] ) holds for all @xmath128 a.s .
in particular , note that this null set depends on @xmath1080 . fix @xmath1081 and
let @xmath1082 .
furthermore , let @xmath1083 be stopping times and for @xmath1084 , @xmath1085 , @xmath1086-measurable random variables . consider the following equations : @xmath1087 then there exists a constant @xmath905 depending only on @xmath363 , @xmath540 , @xmath528 , the lipschitz constants of @xmath11 and @xmath89 and @xmath1088 such that for @xmath1089 , @xmath1090\nonumber\\ & & \qquad\leq c\mathbb{e}\biggl[\|u_1-u_2\|^p+|\theta_1 \wedge t-\theta_2 \wedge t|^{{p/2}}\\ & & \qquad\quad\hspace*{70.5pt}{}+{\int_0^t\sup_{u\leq s}}\|x_u - y_u\|^p \,ds\biggr].\nonumber\end{aligned}\ ] ] by the same arguments as in the proof of @xcite , lemma 11.5 , we first obtain the following estimate : @xmath1091 moreover , @xmath1092 for the stochastic integral part
, we apply the burkholder davis
gundy inequality @xmath1093\\ & & \qquad\leq k\mathbb{e}\biggl[\biggl(\int_0^t\bigl\|\sigma(x_u)1_{\{\theta_1\le u\ } } -\sigma(y_u)1_{\{\theta_2\le u\}}\bigr\|^2 \,du\biggr)^{{p/2}}\biggr]\\ & & \qquad\leq k\mathbb{e}\biggl[\biggl(\int_{\theta_1 \wedge t}^{(\theta_1 \vee\theta _
2)\wedge t}\|\sigma(x_u)\|^2 \,du\biggr)^{p/2}+\biggl(\int_{\theta_2 \wedge t}^{(\theta_1 \vee\theta_2)\wedge t}\|\sigma(y_u)\|^2 \,du\biggr)^{p/2}\\ & & \qquad\quad\hspace*{125.4pt } { } + \biggl(\int_{(\theta_1 \vee\theta_2)\wedge t}^t\|\sigma(x_u)-\sigma ( y_u)\|^2 \,du\biggr)^{{p/2}}\biggr]\\ & & \qquad \leq k\mathbb{e}\biggl[|\theta_1\wedge t-\theta_2 \wedge t|^ { { p/2}}+\int_{0}^t\|\sigma(x_s)-\sigma(y_s)\|^p \,ds\biggr]\\ & & \qquad \leq k\mathbb{e}\biggl[|\theta_1 \wedge t-\theta_2 \wedge t|^{{p/2}}+\int_{0}^t\sup_{u\leq s}\|x_u - y_u\|^p \,ds\biggr],\end{aligned}\ ] ] where @xmath426 always denotes a constant which varies from line to line .
the last estimate in both inequalities follows from the the lipschitz continuity of @xmath11 and @xmath89 . by assembling these pieces ,
the proof is complete . here
is a fundamental existence result , which is not stated in this general form in the standard literature .
therefore , we provide a full proof .
[ th : regulardiff ] there exists a function @xmath1094\times{{\mathbb r}}^n \times \omega\times{{\mathbb r}}_+ \rightarrow{{\mathbb r}}^n$ ] with the following properties : @xmath1095 is continuous in @xmath1096 for all @xmath1097 .
@xmath1098 is @xmath1099\times{{\mathbb r}}^n)\otimes\mathcal{p}$]-measurable .
denotes the predictable @xmath89-field . ]
@xmath1095 solves ( [ sde ] ) for all @xmath1080 .
let @xmath1100 be a stopping time and @xmath90 an @xmath1101 measurable random variable , then @xmath1102 solves @xmath1103 for every @xmath1079\times{{\mathbb r}}^n$ ] , there exists a unique solution @xmath1104 of ( [ sde ] ) , which is continuous in @xmath114 .
this is a consequence of the lipschitz continuity of @xmath1105 and @xmath1106 .
uniqueness is meant modulo indistinguishability . from estimate ( [ eq :
estimate ] ) , we can deduce for @xmath1107 , @xmath1108^n$ ] , @xmath1109 and @xmath1110 , @xmath1111\\ & & \qquad\leq k\biggl(\|x - y\|^{p/2}+|\theta_1-\theta_2|^{p/2}\\ & & \qquad\quad\hspace*{17.3pt}+\int_0^t \mathbb{e}\bigl[\sup_{u\leq s}\|\widetilde{z}(\theta_1,x , u)-\widetilde{z}(\theta_2,y , u)\|^p\bigr ] \,ds\biggr)\end{aligned}\ ] ] for some constant @xmath426 . hence , by gronwall s lemma , @xmath1112&\leq & ke^{kt}(\|x - y\|^{p/2}+|\theta _ 1-\theta_2|^{p/2})\\ & \leq & c\|(\theta_1,x)-(\theta_2,y)\|^{p/2}.\end{aligned}\ ] ] let now @xmath1113 be the set of dyadic rational numbers in @xmath74 and @xmath1114 the set of dyadic rational numbers in @xmath792 .
furthermore , we define @xmath544 by @xmath1115 \times[-t , t]^n)$ ] . by setting @xmath1116
, we can apply kolmogorov s lemma .
indeed , analogous to the proof of @xcite , theorem 2.8 , we derive for all @xmath1117 with @xmath1118 , where @xmath1033 is a positive valued random variable , and for all @xmath1119 , where @xmath1120 is some set depending on @xmath540 with @xmath1121 , the following estimate : @xmath1122 here , @xmath1123 and @xmath745 is some positive constant .
let us now define @xmath1098 : if @xmath1124 , then @xmath1125 for @xmath1126 . for @xmath1119 and @xmath1127 , @xmath1128 for @xmath1126 .
if @xmath1129 , we choose a sequence @xmath1130 such that @xmath1131 . by estimate ( [ eq : kolmogorov estimate ] ) , @xmath1132 is a cauchy sequence converging with respect to .
we can therefore set @xmath1133 .
as we have uniform convergence in @xmath114 and as @xmath1134 is continuous in @xmath114 , the resulting process @xmath1098 is jointly continuous in @xmath1135 .
furthermore , for every @xmath1080 , @xmath1098 is indistinguishable from @xmath1134 , that is , @xmath1136=1.\ ] ] indeed , for @xmath1127 , this is clear and for @xmath1137 , we have for @xmath1138 with @xmath1131 @xmath1139\leq c\varepsilon^{-p}\|(\theta_n , x_n)-(\theta , x)\|^{p/2},\ ] ] which implies that @xmath1140 in probability , uniformly in @xmath114 . as @xmath1141 a.s . , and thus in particular in probability , it follows that @xmath1142 a.s . for all @xmath1143 . letting @xmath1144 proves assertion ( i ) .
statement ( ii ) is then a consequence of ( i ) and the @xmath1145-measurability of @xmath1146 , which is satisfied since @xmath1147 contains all null sets of @xmath1148 .
furthermore , property ( [ eq : cont version ] ) implies that @xmath1149 is a solution of ( [ sde ] ) for all @xmath1080 , which yields assertion ( iii ) . in order to prove ( iv ) , we proceed in two steps : _ step _ 1 .
we first assume that @xmath1100 and @xmath90 take finitely many values @xmath1150 $ ] and @xmath1151 , respectively .
denote @xmath1152 then @xmath1153 does the job .
indeed , as @xmath1154 are disjoint and @xmath1155 for all @xmath1156 , we have ( see , e.g. , @xcite , page 39 ) @xmath1157 for all @xmath1158 a.s .
_ step _ 2 . for general @xmath1100 ,
@xmath90 , approximate @xmath1159 by the simple stopping times @xmath1160 let @xmath1161 be a sequence of @xmath1162-measurable random variables , each @xmath1161 taking finitely many values , and @xmath1163 in @xmath1164 ( such @xmath1161 obviously exists ) .
moreover , @xmath1165 for all @xmath1156 ( see @xcite , chapter 1 , problem 2.24 ) . by step 1 , each @xmath1166 satisfies the respective sde .
moreover , from estimate ( [ eq : estimate ] ) and grownwall s lemma we deduce that for any @xmath903 , there exists a constant @xmath905 such that @xmath1167 \\ & & \qquad\le c e^{ct}\mathbb{e}\bigl[\bigl\|u^{(k)}-u^{(k')}\bigr\|^2+\bigl|\theta^{(k)}\wedge t-\theta^{(k')}\wedge t\bigr|\bigr].\end{aligned}\ ] ] hence , @xmath1166 is a cauchy sequence and thus converging with respect to @xmath1168 $ ] , for all @xmath903 , to some continuous process @xmath9 satisfying ( [ sde2 ] ) . on the other hand , by the continuity of @xmath1169 , we know that @xmath1170 for all @xmath1097 and @xmath1158 .
again , by continuity of @xmath1171 , we conclude that @xmath1172 up to indistinguishability , which proves the claim .
we now provide a constructive proof for the existence of a solution of ( [ eq : jump diffusion ] ) on a specific stochastic basis which is defined as follows : * @xmath1173 is a filtered space , where @xmath1174 , @xmath1175 and @xmath1176 are precisely defined below .
note that we do not have a measure on @xmath1177 for the moment .
the generic sample element will be denoted by @xmath1178 .
* @xmath1179 is some filtered probability space satisfying the usual conditions and carrying an @xmath140-dimensional standard brownian motion @xmath13 .
we shall consider the above diffusion sde ( [ sde ] ) on @xmath1180 and thus obtain the respective solution @xmath1181 satisfying the regularity properties of theorem [ th : regulardiff ] . * @xmath1182 is the canonical space for @xmath792-valued marked point processes ( see , e.g. , @xcite ) : @xmath1183 consists of all cdlg , piecewise constant functions @xmath1184 , @xmath1185 with @xmath1186 and @xmath1187 , where @xmath1188 , defined by @xmath1189 and @xmath1190 are the successive jump times of @xmath1191 .
we denote by @xmath1192 the canonical jump process , and let @xmath1193 be its natural filtration with @xmath1194 .
note that @xmath1195 are @xmath1196 and @xmath1197-stopping times if interpreted as @xmath1198 .
the following statement is meant to be pointwise , referring to the filtered measure space @xmath1199 without reference to a probability measure .
[ lemma : 1 ] let @xmath1200 be as of theorem [ th : regulardiff ] .
then for an @xmath1201-measurable random variable @xmath1202 the process @xmath1203 is : continuous in @xmath114 for all @xmath1204 , @xmath1145-adapted on @xmath1205 .
the first assertion is a consequence of theorem [ th : regulardiff](i ) .
the second one follows from the @xmath1099\times{{\mathbb r}}^n)\otimes \mathcal{p}$]-measurability of @xmath1181 , as stated in theorem [ th : regulardiff](ii ) , and the fact that @xmath1195 and @xmath90 are @xmath1145-measurable on @xmath1205 .
here is our existence result for ( [ eq : jump diffusion ] ) .
there exists a cdlg @xmath1206-adapted process @xmath9 and a probability measure @xmath1207 on @xmath1208 with @xmath1209 , such that @xmath9 is a solution of ( [ eq : jump diffusion ] ) on @xmath1210 .
we follow the arguments in the proof of @xcite , theorem 5.1 , which is based on @xcite , theorem 3.6 , and proceed in three steps .
_ step _ 1 .
we start by solving ( [ eq : jump diffusion ] ) along every path @xmath1191 . to this end , let us define recursively : @xmath1211 , @xmath1212 , and for @xmath1213 : @xmath1214 where @xmath1215 is any fixed point in @xmath1216 and @xmath1098 satisfies the properties of theorem [ th : regulardiff ] . by lemma [ lemma : 1 ] ,
every @xmath1217 is continuous in @xmath114 for all @xmath1204 and @xmath1145-adapted on @xmath1205 since @xmath1218 is @xmath1201-measurable .
thus , the process @xmath1219 is cdlg @xmath1145-adapted and solves ( [ eq : jump diffusion ] ) on @xmath1220 for @xmath1221 and any fixed path @xmath1191 .
_ step _ 2 .
it remains to show that there exists a probability measure @xmath1222 such that @xmath1223 is the compensator of @xmath14 and @xmath1224 holds true .
for this purpose , we shall make use of @xcite , theorem 3.6 .
let us define the following random measure @xmath418 by @xmath1225 observe that @xmath418 is predictable , since @xmath188 is cdlg and @xmath1145-adapted .
theorem 3.6 in @xcite now implies that there exists a unique probability kernel @xmath1226 from @xmath1180 to @xmath1227 , such that @xmath418 is the compensator of the random measure @xmath203 associated to the jumps of @xmath14 .
on @xmath1177 we then define the probability measure @xmath1222 by @xmath1228 whose restriction to @xmath1229 is equal to @xmath1230 . _ step _ 3 .
we finally show that @xmath9 defined by ( [ xdef ] ) solves ( [ eq : jump diffusion ] ) on @xmath1231 for all @xmath1158 .
note that @xmath1232 is an @xmath1077-brownian motion .
this implies that @xmath1233 is a solution of ( [ sde ] ) on @xmath1077 , satisfying the properties of theorem [ th : regulardiff ] .
it thus remains to show that @xmath1234 @xmath1207-a.s .
let @xmath203 be the random measure associated with the jumps .
as @xmath1235 is bounded , we have for all @xmath1236 , @xmath1237\times { { \mathbb r}}^n)\bigr]=\mathbb{e}_{\mathbb{p}}\bigl[\nu([0,t]\times { { \mathbb r}}^n)\bigr]=\mathbb{e}_{\mathbb{p}}\biggl[\int_0^t k(x_t,{{\mathbb r}}^n)\,dt\biggr]\leq c t\ ] ] for some constant @xmath905 .
this implies that @xmath1238\times{{\mathbb r}}^n )
< \infty$ ] a.s .
for all @xmath1236 and hence @xmath1239=0 $ ] or equivalently @xmath1240 a.s . consider a nonempty closed convex set @xmath1070 .
we now provide sufficient conditions for the solution @xmath9 in ( [ xdef ] ) to be @xmath1076-valued .
this result is based on @xcite , theorem 4.1 .
we recall the notion of the _ normal cone _ @xmath1241 of @xmath1076 at @xmath1242 , consisting of inward pointing vectors .
see , for example , @xcite , definition iii.5.2.3 , except for a change of the sign .
[ th : martingale problem convex ] assume that @xmath89 also has a lipschitz continuous derivative .
suppose furthermore that @xmath1243 for all @xmath1244 and @xmath1245 , where @xmath1246 denotes the @xmath243th column of @xmath89 .
then , for every initial point @xmath1247 , the process @xmath9 defined in ( [ xdef ] ) is a @xmath1076-valued solution of ( [ eq : jump diffusion ] ) .
we have to show that @xmath1248 a.s . for all @xmath1158 .
we proceed by induction on @xmath528 . for @xmath1249 , @xmath1250 ,
is simply given by @xmath1251 due to @xcite , theorem 4.1 , conditions ( [ eq : invariance cond1 ] ) and ( [ eq : invariance cond2 ] ) imply that for all , @xmath1252 a.s . let us now assume that for all @xmath1253 , @xmath1254 a.s .
, thus in particular @xmath1255 a.s . if @xmath1256 , then we immediately obtain @xmath1257 otherwise , let @xmath1258 satisfy @xmath1259
. then , @xmath1260&=&\mathbb { e}[f(x_{t_{n-1}-}+\delta j_{t_{n-1}})]\\ & = & \mathbb{e}\biggl[\int_{{{\mathbb r}}^n\setminus\{0\}}f(x_{t_{n-1}-}+\xi ) k(x_{t_{n-1}-},d\xi)\biggr]=0,\end{aligned}\ ] ] since by ( [ eq : cond jumps ] ) , @xmath1261 a.s . and @xmath1262 .
hence , @xmath1263 a.s . , implying that @xmath1264 a.s .
as this holds true for all @xmath1265 with @xmath1266 , it follows that @xmath1267 a.s .
thus , again by @xcite , theorem 4.1 , and conditions ( [ eq : invariance cond1 ] ) and ( [ eq : invariance cond2 ] ) @xmath1268 a.s . takes values in @xmath1076 , which proves the induction hypothesis .
the definition of @xmath9 then yields the assertion .
in this section , we deliver a differentiable variant of the stone weierstrass theorem for @xmath153-functions on @xmath1 .
this approximation statement is essential for the description of the generator of an affine semigroup , as is elaborated in section [ section : inf generator ] .
we employ multi - index notation in the sequel . for @xmath1269 ,
a multi - index is an element @xmath1270 having length @xmath1271 .
the factorial is defined by @xmath1272 .
the partial order @xmath1273 is understood componentwise , and so are the elementary operations @xmath1274 .
that is , @xmath1275 if and only if @xmath1276 for @xmath1277 .
moreover , for @xmath1278 , the multinomial coefficient is defined by @xmath1279 we define the monomial @xmath1280 , and the differential operator @xmath1281 . corresponding to a polynomial @xmath1282 , we introduce the differential operator @xmath1283 .
let @xmath818 denote the locally convex space of rapidly decreasing @xmath153-functions on @xmath52 ( see @xcite , chapter 7 ) , and define the space of rapidly decreasing @xmath153-functions on @xmath1 via the restriction @xmath1284 equipped with the increasing family of semi - norms @xmath1285 @xmath152 becomes a locally convex vector space ( see @xcite , theorem 1.37 ) . for technical reasons , we also introduce for @xmath1286 the semi - norms @xmath1287 on @xmath1288 , where @xmath1289 denotes the closed ball with radius @xmath6 and center @xmath688 .
note that @xmath1290 .
we first give an alternative description of @xmath152 .
[ lemc1 ] we have @xmath1291 the inclusion @xmath1292 is trivial .
hence , we prove @xmath1293 .
so let @xmath1294 for some @xmath1295 with @xmath1296 for all @xmath1297 and some @xmath1298 .
we choose a standard mollifier @xmath1299 supported in @xmath1300 and satisfying @xmath1301 , @xmath1302 . for @xmath735
we introduce the neighborhoods @xmath1303 of @xmath1 .
the convolution @xmath1304 of the indicator function for @xmath1305 with @xmath1306 satisfies @xmath1307 on @xmath1 and it vanishes outside @xmath1308 .
furthermore , all derivatives of @xmath1309 are bounded , since @xmath1310 where the last estimate holds because @xmath1311 . now we set @xmath1312 . by construction @xmath1313 , @xmath1314 and @xmath347 vanishes outside @xmath1315 , because @xmath1309 does .
what is left to show is that @xmath1316 . since @xmath347 vanishes outside @xmath1315 , it is sufficient to deliver all estimates of its derivatives on @xmath1315 .
let @xmath1317 , then we have by the leibniz rule @xmath1318 by assumption @xmath1319 is bounded on @xmath1315 , and @xmath1320 is bounded on all of @xmath52 .
hence , by the last equation , we have @xmath1321 for all , which by definition means @xmath1322 . [ lemc2 ]
let @xmath155 . then for each @xmath1286 , and for all @xmath1297 we have @xmath1323 .
in particular , we have @xmath1324 that is , @xmath1325 for some @xmath1326 . since @xmath155 , there exists a positive constant @xmath659 such that @xmath1327 , for all @xmath115 .
hence , we obtain by a straightforward calculation , @xmath1328 , for all @xmath1297 .
next , let @xmath1298 , and write @xmath1329 , where @xmath707 and @xmath1330 and pick multi - indices @xmath1317 .
then we have by the binomial formula @xmath1331 now since @xmath1332 ranges in a compact set , and since @xmath1333 we see that @xmath1334 must be bounded uniformly in @xmath1335 .
hence , @xmath1336 , for all @xmath1297 . together with lemma [ lemc1 ]
, this implies @xmath1337 .
we are now prepared to deliver the following density result for the @xmath1021-linear hull @xmath619 of @xmath1338 in @xmath152 .
[ th : density ] @xmath619 is dense in @xmath621 .
denote by @xmath1339 and @xmath1340 the topological dual of @xmath1341 and @xmath152 , respectively .
the former , @xmath1342 , is known as the space of tempered distributions .
the distributional action is denoted by @xmath1343 and @xmath1344 for @xmath1345 and @xmath1346 , respectively .
now suppose by contradiction , that @xmath619 is not dense in @xmath621 .
then by @xcite , theorem 3.5 , there exists some @xmath1347 such that @xmath1348 on @xmath1349 .
hence , @xmath1350 , for all @xmath155 .
the restriction @xmath1351 yields a continuous linear embedding @xmath1352 .
hence , the restriction @xmath540 of @xmath1353 to @xmath1354 , given by @xmath1355 yields an element of @xmath1345 with @xmath1356 . pick an @xmath1326 according to lemma [ lemc2 ] . by the definition of @xmath540 , we have @xmath1357 , for all @xmath155 . by the bros
glaser theorem ( see @xcite , theorem ix.15 ) , there exists a function @xmath1358 with @xmath1359 , polynomially bounded [ i.e. , for suitable constants @xmath1360 we have @xmath1361 , for all @xmath115 ] and a real polynomial @xmath1363 such that @xmath1364 in @xmath1365 . hence , we obtain for any @xmath155 @xmath1366 but the last factor is just the laplace transform of @xmath1367 .
this implies @xmath1368 , hence @xmath1369 , which in turn implies that @xmath1353 vanishes on all of @xmath621 , a contradiction .
we thank martin keller - ressel and alexander smirnov for discussions and helpful comments . | this article provides the mathematical foundation for stochastically continuous affine processes on the cone of positive semidefinite symmetric matrices .
this analysis has been motivated by a large and growing use of matrix - valued affine processes in finance , including multi - asset option pricing with stochastic volatility and correlation structures , and fixed - income models with stochastically correlated risk factors and default intensities . , , + and . . |
rotating black holes are described by the kerr solution of general relativity @xcite .
matter particles can perform stable circular orbits in the equatorial plane around these compact objects as long as the orbital radius is greater than the marginally stable orbit @xcite .
the introduction of a locally non rotating frame ( lnrf ) offers a fairly easy way for studying particle motions in the kerr geometry @xcite . in a sense , the observer s frame co rotates with spacetime thereby cancelling frame dragging effects as much as possible .
recently , an unexpected behaviour of the radial dependence of the orbital lnrf velocity of test masses with a keplerian angular velocity distribution around a rapidly rotating , non charged black hole has been discovered @xcite : in vacuum spacetime of very fast spinning kerr black holes with the rotational parameter @xmath0 , there occurs an unexpected dip of the orbital lnrf velocity as function of orbital radius .
this minimum maximum structure of the orbital velocity emerges close to the black hole , at radii @xmath1 .
orbital radii are given in units of the gravitational radius that is defined as @xmath2 with newton s constant g , vacuum speed of light c and black hole mass @xmath3 .
we set g = c=@xmath3=1 throughout the paper for convenience .
the gradient of the orbital velocity is positive in a small radial range .
the radii associated with the local extrema are greater than the innermost stable circular orbit ( isco ) and smaller than the last stable orbit at @xmath4 for any value of @xmath5 .
hence , this effect occurs always within the ergosphere of a kerr black hole .
the velocity difference between the extrema , i.e. the slow down amounts approximately to 1% of c at thorne s spin limit @xmath6 @xcite but is higher for larger black hole spin .
the non monotonic behaviour of the orbital lnrf velocity is a pure effect of general relativity and has been overlooked up to 2004 .
a follow up investigation has shown that the non monotonic behaviour remains even for non keplerian distributions of angular momentum ( @xmath7 ) in the kerr vacuum @xcite .
the critical value of the black hole spin that guarantees the emergence of the effect is higher in this case , i.e. @xmath8 .
the motivation for this work is the question whether the non monotonicity occurs also in more general kerr spacetimes or might actually been removed . in the present work ,
the lnrf orbital velocity is generalized for kerr de sitter and kerr anti de sitter spacetimes , i.e. including the cosmological constant @xmath9 . as for the kerr spacetime
proper , we use the boyer
lindquist form @xcite to study the motion of a particle .
then , the line element of the ordinary kerr geometry holds @xmath10 with the functions ( @xmath11 ) @xmath12 where @xmath3 and @xmath5 denote black hole mass and spin , respectively .
the line element for the kerr
de sitter ( kds , @xmath13 ) or kerr anti de sitter ( kads , @xmath14 ) metric , respectively , is significantly more complicated @xcite @xmath15 ^ 2 + \rho^2\,\left(\frac{dr^2}{\delta_r}+\frac{d\theta^{2}}{\delta_\theta}\right),\end{aligned}\ ] ] where we have the generalized functions @xmath16 @xmath9 denotes the cosmological constant .
these functions reduce to the ordinary kerr geometry by setting @xmath17 : @xmath18 now , we aim to cast the kds / kads line element into a form analogous to equation ( [ eq : kerr ] ) @xmath19 where @xmath20 are the generalizations of the functions @xmath21 , respectively .
indeed , this is possible and we find @xmath22}{\delta_\theta\,(r^2+a^2)^2-\delta_r\,a^2\sin^2\theta } , \\
\alpha^2_{\lambda } & = & \frac{\delta_r\,\delta_\theta\,\rho^2}{\chi^2\,\left[\delta_\theta\,(r^2+a^2)^2-\delta_r\ , a^2\sin^2\theta\right ] } , \\
\tilde{\omega}^2_{\lambda } & = & \frac{\sin^2\theta}{\chi^2\,\rho^2}\left[\delta_\theta\,(r^2+a^2)^2-\delta_r\,a^2\sin^2\theta\right].\end{aligned}\ ] ] note that the first expression represents the generalization of the frame dragging frequency , @xmath23 . considering that the shift vector in the kerr geometry satisfies @xmath24 in boyer
lindquist form @xcite , it is interesting to investigate its generalization , @xmath25 .
it is easy to show that @xmath26 is larger for increasing values of the cosmological constant , i.e. the frame dragging effect becomes stronger with increasing @xmath9 .
further , it is even possible to reverse frame dragging with a sufficiently low and negative @xmath9 . in this case , the radial position where the reversion takes place is determined by the root of @xmath26 . with these generalizations
it is straightforward to compute the orbital velocity component relative to the lnrf for kerr spacetimes with non
zero @xmath9 just analogous to the ordinary kerr geometry , see e.g. @xcite : @xmath27}{\rho^2\,\sqrt{\delta_r\,\delta_\theta}}\,\left\{\omega-\frac{a\,[(r^2+a^2)\delta_\theta-\delta_r]}{\delta_\theta\,(r^2+a^2)^2-\delta_r\,a^2\sin^2\theta}\right\}.\end{aligned}\ ] ] we investigate this completely general radial profile of the orbital lnrf velocity by specifying the angular velocity @xmath28 for two cases : keplerian or constant specific angular momentum rotation .
both distributions are motivated by astrophysical situations .
keplerian angular momentum distributions can be found in accretion flows described by the standard disc model @xcite .
torus solutions exhibit constant angular momentum distributions @xcite .
the non monotonic orbital velocity profile has originally been discovered for prograde orbits .
a similar effect has not been found by us for retrograde orbits .
hence we assume a prograde ( @xmath29 ) keplerian angular velocity distribution of the orbiting test particles that satisfies @xmath30 we can compute the velocity profile from equation ( [ eq : velkds ] ) .
the result is shown in figure 1 which demonstrates that the minimum
maximum structure depends on @xmath9 . as visible in the plot , a cosmological constant close to zero self
consistently approaches the ordinary case for the kerr geometry .
further , the strength of the slow
down effect increases as @xmath9 decreases , i.e. kerr anti de sitter ( @xmath31 ) exhibits a stronger effect than kerr de sitter ( @xmath32 ) .
orbital lnrf velocity profiles around a black hole with @xmath33 for keplerian orbits in the equatorial plane ( @xmath34 ) and different values of the cosmological constant , @xmath35 . the orbital velocity is given in units of c ( @xmath36).,width=302 ] particles around black holes may not follow keplerian orbits but show some distribution of the specific angular momentum @xmath37 , e.g. @xmath7 . in this case , the general expression for the angular velocity in the kerr geometry can be specified for the kerr(anti)de sitter geometry @xmath38 the specific angular momentum @xmath37 has to be chosen in a range between marginally stable and marginally bound orbits , i.e. @xmath39 , see @xcite .
fixing @xmath37 in this interval again reveals distinct minimum
maximum structures in the velocity profile .
figure 2 displays the velocity profile computed from equations ( [ eq : velkds ] ) and ( [ eq : genomeg ] ) for a kerr black hole with @xmath40 and @xmath41 . in figure
2 the values for @xmath5 and @xmath37 are chosen as an example and to make our work comparable to the work of @xcite . for @xmath42 their result is confirmed ( solid curve ) .
the sign of @xmath9 can be easily determined by comparing the profile s position relative to the case @xmath42 .
similar to the keplerian distribution , the trend remains : a negative @xmath9 amplifies the slow down effect .
orbital lnrf velocity profiles around a black hole with @xmath40 for @xmath7 orbits in the equatorial plane ( @xmath34 ) and different values of the cosmological constant , @xmath43.,width=302 ]
the examples discussed so far demonstrate an interesting interplay of black hole spin @xmath5 and @xmath9 .
we have investigated this more deeply since the interaction seems to control the occurrence and modulation depth of the minimum maximum structure .
we consider only the equatorial plane ( @xmath34 ) and compute the gradient of the orbital lnrf velocity for prograde keplerian angular momentum distribution , @xmath44 , and find @xmath45\slash \{2\,\sqrt{3r}\,(a+r^{3/2})^2\,\sqrt{a^2\,(3-\lambda r^2)-r(6 - 3r+\lambda r^3 ) } \nonumber\\ \times[\,a^2\,(\lambda r^2 - 3)+r(6 - 3r+\lambda r^3)]\}.\end{aligned}\ ] ] analogously , the gradient of the orbital lnrf velocity can be analytically computed by plugging equation ( [ eq : genomeg ] ) into equation ( [ eq : velkds ] ) , i.e. for constant angular momentum distributions .
this gradient satisfies @xmath46\}\slash \nonumber\\ \{\sqrt{a^2 - 2r+r^2-a^2\lambda r^2/3-\lambda r^4/3 } \nonumber\\ \times[\,a^4\lambda r -a^3\lambda l r+3
r^3-a l ( 6+\lambda\,r^3)+a^2(6 + 3r+\lambda r^3)]^2\}.\end{aligned}\ ] ] we study both gradient equations in the following and start with the prograde keplerian angular velocity . the roots of the gradient determine the radial positions of the local extrema .
in general , the existence and the positions of extrema are controlled by @xmath5 and @xmath9 . with decreasing @xmath9 the slow
down effect is enhanced for fixed and sufficiently high black hole spin .
a gradient with only one root has a point of inflexion in the velocity profile , whereas more roots yield minima and maxima of the slow
down structure . for each given @xmath9 , there exists a critical value of the black hole spin @xmath47 that is associated with a gradient vanishing at one specific radius . for spin values
greater than @xmath47 , there exists a radial range where the gradient becomes positive . in the kerr vacuum ( @xmath42 ) , it is @xmath48 for keplerian orbiters .
however , generalizing to non zero @xmath9 it is even possible that a high value of @xmath9 is associated with a lower value of critical spin . a graphical analysis of the gradient for fixed @xmath9 delivers the critical spins @xmath47 beyond which the minimum maximum structure appears . the critical spin @xmath47 scales linearly with @xmath9 for @xmath49 ( figure 3 ) .
the study for ordinary kerr black holes by aschenbach @xcite is confirmed and generalized to cases with non zero cosmological constant .
unexpectedly , a similar linear correlation between @xmath47 and @xmath9 can be found for @xmath7 angular momentum distributions ( figure 4 ) .
both the slope and the offset of @xmath5 for @xmath42 differ slightly .
the critical spin grows more steeply and the offset is smaller for keplerian angular momentum distributions .
further , the linear correlation is constrained to a narrow interval of the cosmological constant .
we find in our analysis that the minimum
maximum structure appears only for @xmath50 .
critical spin values as function of @xmath9 for keplerian angular momentum distribution .
the limiting values for the black hole spin @xmath47 are plotted as function of @xmath9 for keplerian distributions of the specific angular momentum .
there appears to be an almost positive linear correlation between @xmath47 and @xmath9 . for spins larger than @xmath47 the slow
down effect occurs .
however , the linear relation breaks down for @xmath51 , because of @xmath52.,width=302 ] critical spin values as function of @xmath9 for @xmath7 angular momentum distribution .
the limiting values for the black hole spin @xmath47 are plotted as function of @xmath9 for @xmath7 specific angular momentum distributions .
similar to the keplerian case an positive linear correlation between @xmath47 and @xmath9 remains .
the analysis of the orbital velocity profiles shows that the the minimum
maximum structure occurs only for @xmath50.,width=302 ]
the slow down effect occurs for rapidly spinning black holes , @xmath53 .
both observations and theory suggest the existence of rapidly spinning black holes . rapid rotation of supermassive black holes is suggested by the observation of broad iron k x ray lines @xcite , and by flares observed from the galactic centre black hole @xcite .
black hole theory suggests that supermassive black holes are endowed with high angular momentum due to the black hole growth history @xcite .
jet launching mechanisms such as the blandford
znajek process @xcite also involve rapidly spinning black holes .
general relativistic magnetohydrodynamics simulations support this idea because strong outflows are driven by poynting fluxes near black holes only for high spins @xcite .
if the minimum maximum structure around a fast spinning black hole could be detected , its radial profile constrains both black hole spin and @xmath9 . if black hole mass and spin were known from observations , e.g. from quasi
periodic oscillations , it might be possible to constrain the cosmological constant .
but only extraordinarily high values of @xmath9 would produce a significant change in the feature as demonstrated in the numerical examples .
the cosmic concordance model @xcite suggests that @xmath9 is by many orders of magnitudes too small to enforce a significant change in the minimum maximum structure .
currently , there is no hope to detect this for the kerr black hole candidates . on the other hand
, we would like to stress that theory of dark energy physics is limited by a number of uncertainties allowing a plethora of models .
the extremes include proposals advocating for a cosmological constant that was significantly higher in earlier cosmological epochs , e.g. involving false vacua @xcite , or even that there is no need for a cosmological constant because density inhomogeneities drive a modification of cosmic expansion @xcite .
another model involves a modification of 4d gravity @xcite .
the new gr effect outlined here might be exploited to probe general relativity for the strong field case .
it is currently hard to say whether this new effect has any practical implication for cosmological studies .
we conclude that a very fast spinning black hole ( @xmath54 for @xmath14 or @xmath0 for @xmath13 ) causes a slow down of the orbital velocity at distances within two gravitational radii .
closer to the rotating black hole the orbital velocity steeply increases again just as expected from newtonian physics .
it has been demonstrated in this paper that the gr slow
down effect survives as the kerr black hole is immersed into a @xmath9 fluid which is described by the kerr
de sitter or kerr anti de sitter solution .
this is true for keplerian distributions of angular momenta as well as for distributions with constant angular momenta .
the parameter study reveals a trend that the slow
down effect is more pronounced for negative values of @xmath9 in both cases , the keplerian and @xmath7 orbiters .
the minimum
maximum structure occurs for black hole spins close to 1 and fairly high @xmath9 .
this is a new gr effect in kerr(anti)de sitter spacetimes .
we thank an anonymous referee who helped us to improve the paper .
kerr r p 1963 _ phys .
lett . _ * 11 * 237 bardeen j m , press w h and teukolsky s a 1972 _ astrophys . j. _ * 178 * 347 bardeen j m 1970 _ astrophys . j. _ * 162 * 71 aschenbach b 2004 _ astron .
_ * 425 * 1075 thorne k s 1974 _ astrophys . j. _ * 191 * 507 stuchlik z , slany p , trk g and abramowicz m a 2005 _ phys .
d _ * 71 * 024037 boyer r h and lindquist r w 1967 _ j. math .
* 8 * 265 carter b 1968 _ phys .
lett . _ * a26 * 399 gibbons g w and hawking s w 1977 _ phys .
d _ * 15 * 2738 stuchlik z and calvani m 1991 _ gen .
* 23 * 507 thorne k s , price r h and macdonald d a 1986 _ black holes : the membrane paradigm _
( new haven and london : yale university press ) mller a and camenzind m 2004 _ astron .
astrophys . _ * 413 * 861 shakura
n i and sunyaev r a 1973 _ astron .
astrophys . _
* 24 * 337 abramowicz m , jaroszynski m and sikora m 1978 _ astron .
astrophys . _ * 63 * 221 tanaka y et al .
1995 _ nature _ * 375 * 659 genzel r , schdel r , ott t , eckart a , alexander t , lacombe f , rouan d and aschenbach b 2003 _ nature _ * 425 * 934 porquet d , predehl p , aschenbach b , grosso n , goldwurm a , goldoni p , warwick r s and decourchelle a 2003 _ astron .
astrophys . _
* 407 * l17 aschenbach b , grosso n , porquet d and predehl p 2004 _ astron . astrophys . _ * 417 * 71 shapiro s l 2005 _ astrophys .
j. _ * 620 * 59 volonteri m , madau p , quataert e and rees m j 2005 _ astrophys
. j. _ * 620 * 69 blandford r d and znajek r l 1977 _ mon . not .
r. astron .
soc . _ * 179 * 433 krolik j h , hawley j f and hirose s 2005 _ astrophys . j. _ * 622 * 1008 mckinney j c 2006 _ mon . not
. r. astron .
_ * 368 * 1561 spergel d n et al .
2006 _ preprint _ astro - ph/0603449 volovik g e 2005 _ ann .
, lpz _ * 14 * 165 kolb e w , matarrese s , notari a and riotto a 2005 _ phys . rev .
d _ * 71 * 023524 dvali g , gabadadze g and porrati m 2000 _ phys . lett .
b _ * 485 * 208 | it has been recently demonstrated that the orbital velocity profile around kerr black holes in the equatorial plane as observed in the locally non rotating frame exhibits a non monotonic radial behaviour .
we show here that this unexpected minimum maximum feature of the orbital velocity remains if the kerr vacuum is generalized to the kerr - de sitter or kerr anti de sitter metric .
this is a new general relativity effect in kerr spacetimes with non vanishing cosmological constant . assuming that the profile of the orbital velocity is known , this effect constrains the spacetime parameters .
received 15 january 2007 , in final form 5 april 2007 accepted for publication in _ classical and quantum gravity _ |
the study of finite - type invariants of immersions of the circle into the plane was initiated by v. arnold . in his paper ( @xcite )
, he gave a description of the three basic invariants , corresponding three different components of the discriminant subspace in the space of the immersions : the closure of the immersions with a direct self - tangency ( @xmath0 ) , the closure of the immersions with an inverse self - tangency ( @xmath1 ) , and the closure of the immersions with a triple point ( @xmath2 ) ( for the details see @xcite ) .
he also noticed , that circle immersions into a plane are closely related to legendrian knots in a solid torus @xmath3 .
indeed , a projection of a legendrian knot along the @xmath4 is a plane curve .
we call the union of @xmath0 and @xmath1 components of the discriminant subspace the @xmath5-discriminant . a finite - type invariant associated with the @xmath5-discriminant
is referred to as @xmath5-invariant . in the paper @xcite v.
goryunov introduces the notion of a marked chord diagrams , describing the combinatorics of a non - generic legendrian knot in the solid torus , and presents a universal finite - type invariant for such immersions , which gives rise to the universal finite - type @xmath0-invariant of circle immersions . in his 1997 article , j.
w. hill shows that any relation on symbols of finite - type invariants of legendrian knots without so - called dangerous self - tangencies in a solid torus is a consequence of certain relations , which are called @xmath6-marked 4t - relations . for the details
see @xcite . the notion of framed chord diagram ( see the definition below ) appeared in an attempt to understand the combinatorics @xmath5-invariants in the paper of s. lando ( see @xcite ) . by an analogy with the knots
finite - type invariant theory , a framed chord diagram represents the combinatorial type of a non - generic immersion , admitting simple self - tangencies ( both direct and inverse ) as singularities .
similarly to the knots case , the symbol of a finite - type invariant of immersions gives rise to a functional on the space of framed chord diagrams .
of course , not every functional comes from a finite - type invariant , and it turns out that those that actually come from a finite - type invariant must satisfy the framed 4t - relations ( see the definitions below ) .
hill s results imply , that the framed 4t - relations are not only necessary , but also sufficient for a functional to come from a @xmath5-invariant .
the filtered space of knot finite - type invariants is dual to the bialgebra of knots with singular knot filtration ( see , for example , @xcite ) .
the induced graded hopf algebra structure on the space of ordinary chord diagrams implies that the latter is just a polynomial algebra .
@xmath5-invariants admit a natural product , which gives rise to a well - defined coproduct on the vector space of framed chord diagrams . however , it is not clear yet whether the space of framed chord diagrams admits a nice product , which makes the study of the space more difficult . in this note
we partly fix the lack of algebra structure on the space of framed chord diagrams by providing it with a hopf module structure over the hopf algebra of ( non - framed ) chord diagrams . weight system ( or , simply speaking , linear map from the space of chord diagrams satisfying 4t - relations ) is a very important notion in the knots finite - type invariants theory , as , due to the vassiliev - kontsevich theorem , it allows to construct knots invariants .
we prove , that any weight system admits a lifting to a framed weight system via the discoloration operator ( see the definition in the section 3 ) .
the comodule structure on the space of framed chord diagrams also gives a way to produce framed weight systems .
the space of framed chord diagrams admits a well - defined map to the space of framed graphs ( introduced in @xcite ) , induced by the operation of taking the intersection graph of a diagram .
it turns out , that a framed version of the chromatic polynomial can be defined , and also used to construct a framed weight system .
this note is the natural development of the ideas of @xcite .
it consists of the introduction and three sections . the hopf module structure on the space of framed chord diagrams
is introduced in section 1 .
section 2 is devoted to similar properties of the space of framed graphs .
the description of a framed chromatic polynomial , along with two methods of construction of framed weight systems are presented in the section 3 .
the author would like to express his sincere gratitude to s. duzhin for his support , numerous discussions , and the permission to use pictures from @xcite .
i also thank s. lando for the interest and valuable comments that helped to make the text easier to understand .
the work was inspired by the article of v. kleptsyn and e. smirnov `` plane curves and bialgebra of lagrangian subspaces . ''
( @xcite ) . in this note
, @xmath7 always denotes a chosen field of characteristics 0 .
let us first recall the definition of the algebra of chord diagrams @xmath8 .
let @xmath9 .
an _ order @xmath10 chord diagram _ is a collection of @xmath10 chords pairs of pairwise distinct points on an oriented circle , considered up to orientation preserving diffeomorphisms of the circle . in the figures below ,
the points in a pair forming a chord are drawn connected by an arc .
the space @xmath8 is the space spanned by the set of the chord diagrams modulo the _ 4t - relations _ , shown in figure [ 4 t ] ( a ) . the pictures in figure [ 4 t ] should be understood as follows .
it is assumed that the dashed part of the circle may contain other chords endpoints , but their combinatorial structure is the same for all the four diagrams entering the relation .
only the chords whose endpoints are inside solid intervals are shown explicitly .
one of the explicitly shown chords is again the same for all the diagrams involved .
we refer to the chord that is not shared by all the diagrams in the relation as the _ jumping chord_. the space @xmath8 is endowed with a natural product @xmath11 , a natural coproduct @xmath12 , the unit @xmath13 , the counit @xmath14 , and the antipode @xmath15 , making @xmath8 into a commutative cocommutative hopf algebra .
the multiplication is given by glueing two diagrams , and the coproduct is given by summing up the tensor products of pairs of chord diagrams formed by a decomposition of the set of chords into two complimentary subsets .
for the precise definitions of the maps , we refer the reader to @xcite , section 4.4 .
we define the space @xmath16 of framed chord diagrams following lando @xcite .
let @xmath9 .
order @xmath10 framed chord diagram _ is a chord diagram along with a fixed _ framing_. a framing is a map from the set of chords to @xmath17 .
the preimages of @xmath18 under this map are called _ oriented chords _ , the preimages of @xmath19 are called _ disorienting chords_. an example of a framed chord diagram is shown in figure [ fcd ] .
a framed chord diagram is said to be _ white _ if all its chords are disorienting .
a framed chord diagram is said to be _ black _ if all its chords are oriented .
we consider framed chord diagrams as generating elements of a @xmath7-vector space @xmath20 over @xmath7 .
the space @xmath16 of framed chord diagrams is the quotient space of @xmath20 modulo the relations shown in figure [ 4 t ] ( a ) and ( b ) . and
@xmath16.,title="fig:",width=56 ] and @xmath16.,title="fig:",width=56 ] and @xmath16.,title="fig:",width=56 ] and @xmath16.,title="fig:",width=56 ] ( a ) and @xmath16.,title="fig:",width=56 ] and @xmath16.,title="fig:",width=56 ] and @xmath16.,title="fig:",width=56 ] and @xmath16.,title="fig:",width=56 ] and @xmath16.,title="fig:",width=56 ] and @xmath16.,title="fig:",width=56 ] ( b ) we also refer to these relations as to _ 4t - relations_. clearly , the space @xmath16 is naturally graded by the total number of chords .
unfortunately , a natural way to define a multiplication on @xmath16 is not known yet . however , the multiplication of a framed diagram by a black diagram is well - defined .
let @xmath21 be a black diagram , and @xmath22 an arbitrary framed chord diagram .
then the product of @xmath21 and @xmath22 , defined by cutting and glueing the two circles ( see figure [ cd_prod ] ) , is well - defined modulo 4t - relation , i.e. it does not depend on the choice of the points where the diagrams are cut . like in the non - framed diagrams case
, this easily follows from the framed 4t - relations .
it is enough to use only the relations that do not change the number of oriented chords .
the proof of the fact that the product does not depend on the choice of the cut - point of @xmath21 repeats word - for - word the proof that the natural map from the algebra of circular chord diagrams to the algebra of linear chord diagrams is well - defined ( see @xcite , section 4.7 ) .
the cut - point of @xmath22 choice independence follows by a direct generalization of the corresponding proof in the non - framed case ( see @xcite , lemma 4.13 ) , since the framing of the jumping chord remains unchanged under a jump along an oriented chord .
the bilinear extension of the above described product @xmath23 makes @xmath16 into a @xmath8-module .
now we endow @xmath16 with an additional algebraic structure .
let @xmath24 be a hopf algebra over the field @xmath7 with a coproduct @xmath25 , and a counit @xmath26 .
then a @xmath7-vector space @xmath27 is called a ( left ) @xmath24-comodule if there is a map @xmath28 satisfying the identities @xmath29 @xmath30 where @xmath31 is the canonical isomorphism @xmath32 . recall ( see @xcite ) that the space @xmath16 admits a well - defined coalgebra structure given by the map
@xmath33 : @xmath34 where the summation is carried out over all possible ways to decompose the set of chords of @xmath22 into two non - intersecting subsets @xmath35 and @xmath5 , and @xmath36 denotes the subdiagram of @xmath22 formed by the chords belonging to @xmath35 .
there is an inclusion @xmath37 .
define the linear map @xmath38 as @xmath39 the well - definedness of @xmath40 is clear . put @xmath41 as it is just a composition of linear maps , it is also a well - defined linear map .
the map @xmath42 endows @xmath16 with an @xmath8-comodule structure . the identities @xmath43 and @xmath44 are straightforward .
the first one follows from the presentation of the both sides of the equality as the sums over the decompositions of the set of oriented chords of a diagram into three non - intersecting sets .
the second one follows from the fact that the only diagram in @xmath8 having a non - vanishing image under @xmath45 is the empty diagram .
the module and the comodule structures on @xmath16 are tightly connected .
let @xmath24 be a hopf algebra over @xmath7 , with a product @xmath46 and a coproduct @xmath25 , and let @xmath27 be a @xmath7-vector space .
then @xmath27 is called a ( left ) _ @xmath24-hopf module _
if * there is a map @xmath47 endowing @xmath27 with a ( left ) @xmath24-module structure ; * there is a map @xmath48 endowing @xmath27 with a ( left ) @xmath24-comodule structure ; * the map @xmath49 is a morphism of @xmath24-modules , where the @xmath24-module structure on @xmath50 is given by the map @xmath51 where @xmath52 is the map permuting the tensor factors .
the maps @xmath53 and @xmath42 endow @xmath16 with a structure of an @xmath8-hopf module .
we have to check that the coproduct map @xmath42 is a morphism of @xmath8-modules @xmath16 and @xmath54 . indeed , consider a product @xmath55 , where @xmath21 is a black diagram , and @xmath22 is an arbitrary framed chord diagram . by definition , @xmath56 where @xmath57 denotes the set of oriented chords of the framed chord diagram @xmath58 , @xmath59 denotes the subdiagram consisting of the chords of the set @xmath35 only ( with the chords combinatorial structure preserved ) , and @xmath60 is the diagram with the chords from the set @xmath35 _ removed_. note that every subset @xmath61 can be presented as a disjoint union @xmath62 .
hence , @xmath63 which clearly equals to @xmath64 .
the main structure result of the theory of chord diagrams is a direct corollary of the milnor - moore theorem .
it states that every commutative cocommutative connected graded hopf algebra over a field of characteristic 0 is a polynomial algebra generated by a basis of the subspace of its primitive elements .
there is a similar structure theorem for hopf modules . to formulate it ,
we need one more definition .
let @xmath24 be a hopf algebra over @xmath7 , and let @xmath27 be a ( left ) @xmath24-hopf module with the comodule structure map @xmath49 .
the _ covariant submodule _
@xmath65 is the subspace of @xmath27 spanned by the covariant elements , i.e. , elements @xmath66 satisfying @xmath67 .
let @xmath24 be a commutative cocommutative hopf algebra over @xmath7 with the coproduct @xmath25 , and let @xmath27 be an @xmath24-hopf module with the module structure given by the product @xmath68 .
then @xmath27 is trivial , i.e. the restriction of the product map @xmath69 is an @xmath24-hopf modules isomorphism .
the @xmath24-hopf module structure on @xmath70 is given by the maps @xmath71 @xmath72 an immediate corollary of the larson - sweedler theorem is the @xmath8-hopf module @xmath16 is free .
it is generated by a basis of its covariant submodule @xmath73 .
this statement means that every element of @xmath16 is representable as a linear combination of products of the type @xmath74 , where @xmath21 is a black diagram , and @xmath22 is a covariant element .
the space @xmath73 admits a nice description .
[ structure ] the space @xmath73 is spanned by white diagrams .
it is clear that the subspace spanned by white diagrams is a subspace of @xmath73 .
now we are going to show that every chord diagram modulo 4t - relations is representable as a linear combination of products of black diagrams by white diagrams .
take an arbitrary framed chord diagram @xmath22 .
if all its chords have the same framing , then there is nothing to prove .
now suppose that the diagram contains chords of both possible framings .
choose a point @xmath75 different from the chords endpoints on the supporting circle .
let @xmath76 be an endpoint of an oriented chord . by the _ remoteness _ @xmath77 of @xmath76 with respect to @xmath75 we mean the number of endpoints of disoriented chords one meets while traveling along the circle with respect to its orientation from @xmath76 to @xmath75 .
the number @xmath78 where @xmath79 and @xmath80 denote the number of oriented and disorienting chords , respectively , is called the _ complexity _ of @xmath22 with respect to @xmath75 .
note that if the complexity equals 0 , then @xmath22 obviously is representable as a product of a black and a white diagram .
the minimal possible non - zero complexity of a diagram with @xmath79 oriented chords and @xmath80 disorienting chords is @xmath81 .
travelling along the circle starting from @xmath75 in a clockwise direction , consider the first encountered oriented chord endpoint with a non - zero remoteness .
apply the 4t - relation , as shown in figure [ 4 t ] ( b ) , trying to reduce the remoteness of the oriented chord endpoint under consideration .
in such a way we present @xmath22 as a linear combination of three diagrams , with the basis point @xmath75 inherited to all three .
one of these diagrams differs from @xmath22 by the position of one of the oriented chord endpoints only , and its complexity is clearly smaller than that of @xmath22 .
the number of oriented chords in each of the two other diagrams is one less than in @xmath22 .
if they do not contain oriented chords at all , then we are done . otherwise , we take into account that the largest possible complexity of a diagram with @xmath82 oriented chords and @xmath83 disorienting chords is @xmath84 which is evidently smaller than @xmath81 .
another corollary of the larson
sweedler theorem is the space @xmath16 admits a bigrading .
indeed , by the larson - sweedler theorem , @xmath16 is a tensor product of the graded algebra @xmath8 and the vector space spanned by white diagrams .
the latter is also graded since the framed 4t - relations are homogeneous with respect to the number of chords .
the assertion follows .
_ remark : _
proposition [ structure ] means that the symbol of any finite - type j - invariant is a linear combination of products of symbols of j@xmath85-invariants by symbols of j@xmath86-invariants .
the space @xmath73 has non - trivial relations among the generating white diagrams .
up to degree 4 , there are no relations , and the first example of a non - trivial relation in degree 4 was found by s. duzhin .
examples of relations in @xmath73 are shown in figure [ white ] .
unfortunately , a complete description of the set of relations in @xmath73 is not known .
.,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ] .,title="fig:",width=37 ]
the proof that the listed identities are satisfied is the same for each of the series : it is possible to split the diagrams taking part in a relation into 4 pairs .
the difference of the diagrams forming a pair equals to the difference of a pair of diagrams with one oriented chord .
the resulting linear combination of diagrams with one oriented chords vanishes identically .
the dimensions of the graded components of the spaces @xmath16 and @xmath73 for small number of chords obtained by computer calculations made in sage ( @xcite ) are [ cols="^,^,^,^,^,^",options="header " , ]
recall that a linear map from @xmath8 to an abelian group @xmath87 is traditionally called an _ @xmath87-valued weight system_. weight systems are important objects ; they allow one to construct finite - type knot invariants , due to the vassiliev kontsevich theorem ( see , for example @xcite ) .
gven an arbitrary weight system , it is possible to construct a framed weight system .
recall , that for the description of the @xmath8-hopf module structure on @xmath16 , we exploited the projection operator @xmath88 .
let us introduce one more projection operator @xmath89 .
define the action of the _ discoloration operator _
@xmath90 on a framed chord diagram @xmath22 by the rule : @xmath91 where @xmath80 is the number of disorienting chords of @xmath22 , and @xmath92 is obtained from @xmath22 by setting the framings of all the chords to be 0 .
extend @xmath93 to @xmath16 by linearity .
_ remark : _ the discoloration operator can also be used to define a comultiplication operator @xmath94 as @xmath95 the operator @xmath96 , along with @xmath53 , provides @xmath16 with one more @xmath8-hopf module structure , different from the one we have actually used .
the intersection graph of a ( non - framed ) chord diagram @xmath22 is defined as the graph @xmath98 whose vertices are in one - to - one correspondence with the chords of @xmath22 , and two vertices are connected by an edge if the corresponding chords intersect ( see @xcite ) .
this construction was first proposed by s. chmutov , s. duzhin , and s. lando in 1994 .
let @xmath22 be a framed chord diagram .
the _ intersection graph of @xmath22 _ is the graph @xmath98 whose vertices are in one - to - one correspondence with the chords of @xmath22 .
two vertices of @xmath99 are connected by an edge if and only if the corresponding chords of @xmath22 intersect ( this means that their endpoints @xmath100 and @xmath101 appear in the interlacing order @xmath102 while travelling along the circle ) .
the framing of @xmath98 is inherited from that of @xmath22 .
the map @xmath35 is a well - defined linear map .
the squares @xmath106^{i\otimes i } \ar[d]^{m_\mathcal m } & \mathcal g\otimes \mathcal h \ar[d]^{m_\mathcal h}\\ \mathcal m \ar[r]^i & \mathcal h } \ ] ] and @xmath107^{i } \ar[d]^{\delta_\mathcal m } & \mathcal h \ar[d]^{\delta_\mathcal h}\\ \mathcal a\otimes \mathcal m \ar[r]^{i\otimes i } & \mathcal g\otimes \mathcal h } \hspace{1 cm } \xymatrix { \mathcal m \ar[r]^{i } \ar[d]^{\delta_\mathcal m } & \mathcal h \ar[d]^{\delta_\mathcal h}\\ \mathcal m\otimes \mathcal m \ar[r]^{i\otimes i } & \mathcal h\otimes \mathcal h } \ ] ] are commutative .
we have to check that a linear combination of framed chord diagrams vanishing due to the 4t - relation maps to a linear combination of framed graphs vanishing due to the 4t - relations as well .
it is a straightforward generalization of the corresponding statement in the non - framed case , and the proof is the same .
it is clear that the multiplication of chord diagrams corresponds to the disjoint union of graphs .
this implies the commutativity of the first square .
the commutativity of the second and the third square is also straightforward .
a natural extension of the chromatic polynomial to the framed graphs takes values in the polynomial algebra @xmath113 $ ] .
let @xmath99 is a framed graph , @xmath114 is an edge of @xmath99 , and @xmath115 and @xmath116 are the vertices adjacent to @xmath114 .
we define the framed chromatic polynomial , which we also denote by @xmath110 , via the skein - relation @xmath117 and the additional relations @xmath118 @xmath119 @xmath120 where the graph @xmath121 is obtained from @xmath99 by deleting the edge @xmath114 , and the graph @xmath122 is obtained from @xmath99 by contracting @xmath114 .
the framings of the vertices of @xmath121 , and all the vertices of @xmath122 , except for the vertex @xmath123 appearing as the result of edge contraction , are inherited from that of the corresponding vertices of @xmath99 . in the case of contraction
, we set @xmath124 if @xmath125 , and 1 otherwise . ] .
the symbols @xmath126 and @xmath127 denote the graph with one black vertex and no edges , the graph with one white vertex and no edges , and the empty graph , respectively .
the value of the chromatic polynomial on any graph containing loops is 0 .
consider a framed graph @xmath99 .
first note that using the first skein relation one is able to present @xmath99 as a linear combination of edgeless graphs .
we need to check that the resulting combination does not depend on the order we get rid of the edges when applying the skein relation .
this check is straightforward .
now it is clear that the framed chromatic polynomial is uniquely defined by the skein relation .
finally , we have to check that the appropriate linear combination of the values of the framed chromatic polynomial on a 4-tuple of graphs taking part in a 4t - relation , vanishes . as in the non - framed case this follows from the fact that both sides of a 4 t relation are the same modulo the skein relation defining @xmath110 .
_ remark : _ the polynomial algebra @xmath113 $ ] also admits a hopf algebra structure ( with @xmath130 and @xmath115 being the primitive generators of degree 1 ) . the chromatic polynomial @xmath131 $ ] is a morphism in the category of hopf algebras .
recall , that given a @xmath87-valued weight system @xmath80 , and @xmath132-valued weight system @xmath133 , the composition @xmath134 is an @xmath135-valued weight system ( the _ convolution product _ of weight systems ) . using the comodule structure @xmath42 on @xmath16 we can extend this operation to produce framed weight system let @xmath136 are abelian groups , @xmath80 is an @xmath87-valued weight system , and @xmath133 is an @xmath132-valued framed weight system . then the composition map @xmath137 is an @xmath135-valued framed weight system .
there are weight systems that are not determined by the intersection graph only .
for example , there exists an order 11 weight system , that distinguishes two mutant knots @xcite , and thus , by a theorem of chmutov and lando ( see @xcite ) , can not be obtained from a linear map from @xmath111 .
note , that even putting in the previous theorem @xmath138 , by a suitable choice of @xmath80 , it is possible to construct framed weight system not determined by the intersection graph only .
arnold , _ plane curves , their invariants , perestroikas and classifications . _ in _ singularities and bifurcations _ ,
volume 21 of adv .
soviet math .
, 33 - 91 .
soc . , providence , ri , 1994 . with an appendix by f. aicardi .
v.v.goryunov , _ finite order invariants of framed knots in a solid torus and in arnold s ( j+)-theory of plane curves , _ lecture notes in pure and applied mathematics , vol.184 ( 1997 ) ` geometry and physics ' ( j.e.andersen , j.dupont , h.pedersen and a.swann , eds . ) , marcel dekker , inc . , new york basel
hong kong , 549 - 556 .
j. murakami , _ finite type invariants detecting the mutant knots . _
knot theory , a volume dedicated to professor kunio murasugi for his 70th birthday . edited by m. sakuma et al . ,
osaka university , march 2000 . | this note is dedicated to the study of a hopf module structures on the space of framed chord diagrams and framed graphs .
we also introduce a framed version of the chromatic polynomial and propose two methods to construct framed weight systems . |
removal of the smallest eigenvalues of a matrix can significantly improve the performance of iterative methods .
this is important for lattice qcd where stochastic methods for disconnected diagrams and new matter formulations such as overlap fermions are computationally intensive .
in addition , there is often a need for these systems to be shifted to take account of multiple quark masses or other parameters .
we think that it is important for any iterative method we develop to be applicable to general non - hermitian systems , for this is often the case for lattice systems .
gmres @xcite is an iterative method that can be used in lattice studies with stochastic noise methods .
convergence can be improved by deflating eigenvalues .
two years ago it was shown that deflating eigenvalues @xcite can be especially helpful for lattice problems with multiple right - hand sides @xcite . here
we report on generalizations of these techniques .
first we show how deflated restarted gmres can accomodate multiple shifts .
we give numerical results for a wilson matrix with very small eigenvalues .
we then consider the case of combining multiple shifts with multiple right - hand sides . finally we also discuss and give numerical results on the possibility of combining deflated gmres for the first system with a bicgstab algorithm which uses those deflated eigenvalues for the subsequent solutions .
the shifted systems are expressed by @xmath1 where the s(i ) are the numerical shifts . the method proposed in @xcite
is based upon keeping the residual vectors for the shifted systems parallel to one another .
that is , one uses only a single krylov subspace to solve all the systems .
this is possible for gmres only by forcing the residuals to be parallel to one another after a restart and thus the minimum residual condition is applied to only the unshifted ( most difficult ) system .
we find that this also results in an almost optimal solution for the shifted systems .
= 2.9 in -.1 cm -.5 cm our method for deflating gmres ( called gmres - dr @xcite ) simulataneously solves linear equations and computes eigenvalues .
this is possible because it has a krylov subspace for solving the linear equations as well as krylov subspaces designed for solving the eigenvalue equation .
these approximate eigenvectors are refined and improved as the gmres - dr cycles proceed . we have determined that gmres - dr can be efficiently applied to multiply shifted systems because of its krylov properties .
fig.1 shows the result of a shifted run of gmres - dr(50,30 ) , i.e. , the subspace is 50-dimensional and includes 30 approximate eigenvectors .
it is compared to the residual norm for regular gmres(50 ) .
one can see the deflation start to take place around 200 matrix - vector products ( the approximate eigenvectors are accurate enough by then ) .
typically , many right - hand sides are used to do stochastic estimates of matrix elements .
it would be beneficial if the deflated eigenvalues and eigenvectors calculated for the first right - hand side could be reused .
then deflation could begin from the start instead of having to wait for the required eigenvectors to be recalculated , with more dramatic results than in fig.1 .
a method to accomplish this called gmres - proj was given in @xcite .
gmres - proj alternates cycles of gmres with projections over the approximate eigenvectors .
next we consider using this for the case of multiple shifts for each of the multiple right - hand sides . gmres - dr can deflate eigenvalues and handle multiple shifts at the same time for the first right - hand side ( the right - hand sides for each shifted system can be kept parallel ) .
however , for the second and subsequent right - hand sides there is no way to deflate eigenvalues by projecting over the previously computed approximate eigenvectors and still keep the right - hand sides for the different shifted systems parallel .
if the eigenvectors were exact , this would be possible .
so one approach is to use only fairly accurate approximate eigenvectors and do a correcting iteration at the end for each shifted system separately .
we now propose a better fix for this problem which does not require accurate eigenvectors . gmres - dr produces approximate eigenvectors that are represented in the form of an arnoldi - like recurrence relation : @xmath2 where the columns of @xmath3 span the space of approximate eigenvectors , @xmath4 has one extra column added to @xmath3 , and @xmath5 is a @xmath6 by @xmath7 matrix . using this
, we can perform the projection over the approximate eigenvectors so that the right - hand sides for the different shifted systems are all multiples of each other , except for error in the direction of @xmath8 , the last column of @xmath4 .
this error can be easily corrected at the end , if we are willing to solve one additional right - hand side , namely @xmath8 .
there is also potential to develop a deflated version of bicgstab for multiply shifted systems using this idea .
this will be reported later , but for now we will look at deflating bicgstab for multiple right - hand sides in the non - shifted case .
the deflation on the additional right - hand sides can employ other methods after the initial eigenvalue generation by gmres - dr .
we have looked at bicgstab , a non restarted method popular in the lattice community , in this context .
we again combine a projection over approximate eigenvectors with the iterative method .
however , the projection can only be performed once before the bicgstab iteration begins .
therefore this projection must reduce the critical eigencomponents to a low enough level that no further reduction is needed during the bicgstab iteration .
the minimal residual projection is the best possible one in the sense that it minimizes the norm of the residual vector .
however , it can be shown that even if one knows the exact eigenvectors , the corresponding component in the residual vector will not generally be zeroed out ( unless the matrix is hermitian ) . to deal with this problem , we will give a projection that is better for this task .
it uses both left and right approximate eigenvectors .
left - right projection for deflating eigenvectors : \1 .
let the current approximate solution be @xmath9 and the current system of equations be @xmath10 .
let @xmath11 be orthonormal with columns spanning the subspace of approximate right eigenvectors .
let @xmath12 similarly be orthonormal with columns from approximate left eigenvectors .
solve @xmath13 , where @xmath14 .
the new approximate solution is @xmath15 .
this projection can reduce the small eigenvector components better than minimum residual and in the case of exact left and right eigenvectors , the components can be zeroed out . of course , it is necessary to compute left eigenvectors .
however , for many lattice problems , including the wilson - dirac matrix , there is a simple relationship between the left and right eigenvectors .
= 2.9 in -.1 cm -.5 cm fig.2 above shows a run on a smaller @xmath16 lattice ( using matlab ) of the deflated bicgstab algorithm using eigenvectors calculated from an initial gmres - dr solution of the first right - hand side .
the deflated solutions take a smaller number of iterations than the nondeflated implementation .
notice that deflation does not care about the nature of the right - hand sides used , so these may be noise vectors or units vectors , as one desires .
this applies to all techniques used .
this material is based upon work supported by the national science foundation under grants no .
0070836 ( theoretical physics ) and 0310573 ( computational mathematics ) and ncsa and utilized the sgi origin 2000 system at the university of illinois .
rm acknowledges the baylor university summer sabbatical program .
y. saad and m. h. schultz , gmres : a generalized minimum residual algorithm for solving nonsymmetric linear systems `` , siam j. sci .
comput . 7 , 856 ( 1986 ) .
r. b. morgan , gmres with deflated restarting '' , siam j. sci . comput .
24 , 20 ( 2002 ) .
r. b. morgan and w. wilcox , nuclear physics b ( proc .
suppl . ) 106 , 1067 ( 2002 ) .
a. frommer , restarted gmres for shifted linear systems " , siam j. sci . comput .
53 , 15 ( 1998 ) . | work on generalizing the deflated , restarted gmres algorithm , useful in lattice studies using stochastic noise methods , is reported .
we first show how the multi - mass extension of deflated gmres can be implemented .
we then give a deflated gmres method that can be used on multiple right - hand sides of @xmath0 in an efficient manner .
we also discuss and give numerical results on the possibilty of combining deflated gmres for the first right hand side with a deflated bicgstab algorithm for the subsequent right hand sides . |
there is a great deal of current interest in quasi one - dimensional quantum spin models , such as spin ladders , spin chains with dimerization and ( or ) frustration , and various generalizations of the above .
the studies of these models have been mostly triggered by the experimental discoveries of several spin - peierls and spin - ladder compounds @xcite .
the theoretical efforts have been focused mainly on the nature of the ground state , the excitation spectrum , as well as thermodynamic properties of these models .
the system of coupled quantum @xmath0 chains ( spin - ladder ) was found to behave differently depending on whether the number of chains is even or odd @xcite .
for an even number , the excitation spectrum is generically gaped and short - range correlations dominate in the ground state . the single @xmath0 quantum spin chain with dimerization exhibits
quite similar properties , since it is known that dimerization produces an energy gap and destroys the quasi long - range order of the integrable uniform chain . in a recent paper , ref.@xcite , m.a .
martin - delgado , r. shankar , and g. sierra have proposed a model of coupled spin chains with staggered dimerization which possesses a rich phase diagram , quite different from that of the simple spin ladder ( no dimerization ) .
consider the simplest version of the model , containing two coupled @xmath0 heisenberg chains ( fig . 1 . ) : @xmath1{\bf s}_{\alpha , i}\cdot { \bf s}_{\alpha , i+1}\ ] ] here @xmath2 is labeling the two chains ( legs of the ladder ) .
the coupling @xmath3 is the inter - chain ( rung ) interaction and @xmath4 are the nearest neighbor interactions along the chains ( all couplings are antiferromagnetic ) .
notice that the dimerization @xmath5 is staggered in both directions ( along and perpendicular to the chains ) , i.e. the dimerization in each chain is in antiphase with its neighbor chain .
the model eq.(1 ) was analyzed in ref.@xcite by mapping onto a non - linear sigma model ( nlsm ) . in the nlsm approach the value of the topological angle
@xmath6 determines the nature of the possible phases of the model @xcite . for @xmath7
the model is massive while it is massless for @xmath8 . for the model at hand two
massive ( gaped ) phases were found in the parameter plane @xmath9 with a massless ( critical ) line separating them , determined by the equation : @xmath10 a symmetric branch with @xmath11 also exists but we will assume @xmath12 from now on .
the two massive phases separated by this line are basically phases where the dominant correlations are along the rungs or the stronger bonds along the legs .
the existence of a critical line is quite unusual since individually both interactions , @xmath13 and @xmath5 are gap producing .
it is clear from fig .
1 . that there are two points @xmath14 and @xmath15 where the system is a single uniform chain and hence integrable and massless .
the nlsm approach eq.(2 ) does not correctly reproduce the point @xmath16 which is quite natural since it is expected to work in the vicinity of the point @xmath15 only .
it has been conjectured in ref.@xcite essentially from continuity , that a critical line connects the two integrable points .
let us mention that upon introducing an additional spin alternation @xcite and ( or ) considering ladders of spin @xmath17 or higher @xcite the phase diagram of the system will be even more complex .
the basic origin for this complexity however is still the staggered dimerization in the rung direction .
thus , form now on we will consider only the minimal " model eq.(1 ) .
the purpose of the present work is to analyze the excitation spectrum and the phase diagram of the model eq.(1 ) at @xmath18 by a combination of numerical and analytical techniques . in sec .
ii we present exact diagonalization results for the energy gap and the spin - spin correlations , confirming the existence of two gaped phases separated by a gapless line . in sec .
iii we develop two kinds of strong - coupling approaches in the phase dominated by strong interchain ( rung ) correlations : ( a ) dimer series expansions , and ( b ) diagrammatic treatment of the effective hamiltonian , written in terms of bond triplet operators .
the spectrum near the critical line is calculated and the relevant interactions , responsible for the closing of the gap are identified .
iv contains our conclusions and discussion of future prospects .
first we present results from finite lattice diagonalizations of the hamiltonian eq.(1 ) , for systems of up to @xmath19 spins .
periodic boundary conditions are used throughout . for reasons of symmetry the number of spins must be a multiple of @xmath20 , and most results are obtained from extrapolation of @xmath21 spin systems .
we find empirically that the ground state is a singlet @xmath22 whereas the first excited state is a triplet @xmath23 .
the computed singlet - triplet gaps are extrapolated to the thermodynamic limit @xmath24 via the ansatz : @xmath25 where @xmath26 and @xmath27 are constants .
the last term allows for the logarithmic corrections arising through conformal invariance in spin chains and effectively accounts for the residual curvature observed in direct ratio plots . in fig .
we show the variation of energy gaps versus the interchain coupling @xmath28 for various @xmath29 , for a fixed value of the dimerization @xmath5 ( in this case @xmath30 ) .
it is clear qualitatively that the variation in @xmath31 is small for both small and large values of @xmath28 whereas the variation is large near @xmath32 where the individual curves have a minimum .
the dashed lines , based on the extrapolated formula ( 3 ) clearly show the gap vanishing at a single point on the @xmath28 axis .
similar results are obtained for other values of @xmath5 although for small @xmath5 the convergence is poorer .
this is consistent with the increasing quantum fluctuations and larger finite size corrections expected near the uniform limit .
this analysis leads to fig .
3 . which shows the estimated critical line in the @xmath33 plane . comparing with eq.(2 ) one can see that the nlsm result can not be trusted numerically , but the existence of a gapless line is correctly predicted . in fig .
we show various spin correlations versus @xmath28 for @xmath30 .
as intuitively expected the phase with @xmath34 is characterized by strong inter - chain ( rung ) and weak intra - chain correlations .
when @xmath28 decreases and ultimately crosses the critical point , the behavior of the correlation functions is reversed , signaling a transition into a phase , dominated by strong intra - chain correlations on the stronger bonds .
we find that the spin correlations as well as the excitation gap change continuously through the quantum transition point .
in this section we present analysis based on strong - coupling expansions around the limit @xmath35 .
two approaches are used : the dimer series expansion and the bond operator mapping onto an effective interacting bose gas .
the two approaches are similar in spirit and produce quite similar result even though the technical details are different .
we start with results obtained by the linked - cluster dimer series expansion method .
the linked - cluster expansion method has been previously reviewed in several articles@xcite , and will not be repeated here .
the basic idea is to divide the hamiltonian into two parts : @xmath36 where @xmath37{\bf s}_{\alpha , i}\cdot { \bf s}_{\alpha , i+1 } \nonumber\end{aligned}\ ] ] by fixing values of @xmath5 , we can construct an expansion in @xmath38 by taking @xmath39 as unperturbed hamiltonian and @xmath40 as a perturbation .
the zeroth order @xmath41 approximation corresponds to isolated dimers ( singlets ) on the rungs .
each singlet can be excited into a triplet state and thus the excitation gap is @xmath13 .
finite @xmath42 introduces interactions between the rungs which modifies substantially the spectrum . in order to consider
the physically interesting coupling region @xmath43 the perturbation series has to be developed to high order and then extrapolated to the relevant point .
we have carried out an expansion up to order @xmath44 for the ground state energy , to order @xmath45 for antiferromagnetic susceptibility , and to order @xmath46 for the triplet excitation spectra for several values of @xmath5 .
there are only 12 graphs that contribute to the ground state energy and dispersion , and 14 graphs that contribute to the antiferromagnetic susceptibility .
the series coefficients are not presented here , but are available upon request .
then we have used integrated differential approximants and pad approximants@xcite to extrapolate the series .
it is sometimes useful to compare the present results with the spectrum of a spin ladder without dimerization @xmath47 , which can be found in ref .
@xcite ( see fig .
4 . ) . in fig . 5 .
we present the triplet excitation spectrum @xmath48 for @xmath30 , obtained by the dimer series expansions .
the triplet gap @xmath49 decreases with the decrease of @xmath28 , vanishing ( within error bar ) at the critical value @xmath50 . by comparing the spectra of fig .
5 . with the corresponding spectra for @xmath51 ( see fig .
4 . of ref.@xcite for the value @xmath52 )
one can see that the renormalization due to finite @xmath5 is the strongest at @xmath53 while in the vicinity of the point @xmath54 the energy is almost unrenormalized .
this is due to the fact that the first three leading orders in @xmath38 do not change @xmath55 ( see also sec .
iii.b ) .
a more accurate determination of the critical line can be achieved by the dlog pad approximants to the antiferromagnetic susceptibility series , and the results are presented in fig .
the phase boundary obtained within the dimer series approach is in excellent agreement with the exact diagonalization results in the region @xmath56 . for small values of @xmath13
the convergence of the dimer series becomes poorer ( larger error bars ) , as the critical point occurs at much larger @xmath38 . in fig .
several critical spectra , for parameters on the massless line , are plotted .
notice that for @xmath57 when eq.(1 ) reduces to a snake " , i.e. one - dimensional heisenberg chain , the dimer series reproduces correctly the spinon dispersion , which is known from the exact bethe ansatz solution of the problem @xcite .
the latter is given by the formula ( for a chain with exchange @xmath42 ) : @xmath58 .
indeed , the lower curve in fig . 6 . practically coincides with this formula . using the long series for the ground state energy , we were able to obtain the most accurate estimates of the ground state energy , as it was done before for the normal two - chain ladder@xcite .
for example for @xmath59 and @xmath60 which is near the critical line , the ground - state energy per site is estimated to be @xmath61 beside the above inter - chain dimer expansions about the limit @xmath35 , we can also construct another dimer expansion , the intra - chain dimer expansion , by taking the stronger intra - chain bonds ( the bonds couping @xmath62 ) as unperturbed hamiltonian , and the rest of the bonds as perturbation , and carry out dimer series expansions in @xmath63 for fixed values of @xmath64 .
the series have been computed up to order @xmath65^{11}$ ] for the ground state energy , to order @xmath65^{6}$ ] for the antiferromagnetic susceptibility , and to order @xmath65^{7}$ ] for the triplet excitation spectra for several values of @xmath64 . with this expansion , one can study the excitation spectrum and the ground state properties for the parameters located within the intra - chain dimer phase .
the gapless critical line derived from this expansion agrees well with that by inter - chain dimer expansions , but with much less accuracy .
here we will not present any detailed results from this dimer expansion , however the series are available upon request . in this section
we describe a mapping of the model onto an interacting bose gas , which can be achieved by using the bond operator representation for spins @xcite .
denote by @xmath66 the singlet state formed by two spins on rung @xmath67 .
then , in the strong - coupling limit @xmath68 the excitations are well described by triplets , created by the ( bond ) operators @xmath69 , @xmath70 , where @xmath71 are the three components of the triplet and the usual bosonic commutation relations hold @xcite .
the hamiltonian eq.(1 ) expressed in terms of the triplet operators is @xcite : @xmath72 where we have separated the part due to dimerization : @xmath73 + [ i \leftrightarrow i+1 ] \right \}\ ] ] and @xmath74 will be defined below . for @xmath51 the hamiltonian ( 7 )
coincides with that of the simple two - leg ladder @xcite . in order for the triplet operators to represent only the physical spin triplet states
, they have to satisfy the on - site hard - core constraint : @xmath75 .
this restriction on the hilbert space can be taken into account by introducing an infinite on - site repulsion between the triplets , as discussed in our previous work @xcite : @xmath76 our goal is to develop a diagrammatic treatment of the interactions in the hamiltonian ( 7 ) in order to understand which terms are most relevant to the renormalization of the spectrum .
the quadratic part ( first line in eq.(7 ) ) is diagonalized in momentum space by a bogoliubov transformation @xmath77 , which leads to the spectrum @xmath78 , with @xmath79 .
the bogoliubov coefficients are defined as : @xmath80 .
the spectrum of the hamiltonian eq.(7 ) without the dimerization term @xmath81 was investigated diagrammatically in refs.@xcite .
here we only summarize the results .
the on - site repulsion @xmath74 gives the dominant contribution to the spectrum renormalization , while the two - particle inter - site interaction with strength @xmath82 ( first term in the second line of eq.(7 ) ) is a relatively minor effect @xcite . even though the on - site interaction is infinite , the scattering amplitude @xmath83 is finite ( as physically expected ) , which can be seen by resumming the ladder series , shown in fig .
7(a ) and setting @xmath84 .
the scaterring amplitude and the corresponding self - energy ( fig .
7(b ) ) are @xcite : @xmath85^{-1 } = \sum_{q}\frac{1}{-\omega + \omega_{q}+ \omega_{k - q } } \ , \
\sigma_{u}(k,\omega ) = 4 \sum_{q } v_{q}^{2 } \gamma(k + q,\omega - \omega_{k}).\ ] ] here , and in all future equations we set , for convenience , the number of rungs to unity . as emphasized in ref.@xcite , eq.(10 ) represents the dominant contribution from @xmath74 , provided the density of triplets @xmath86 is small .
thus eq.(10 ) should be viewed as the first term in an expansion in powers of @xmath87 . for the simple ladder ( @xmath51 )
we find ( after solving the corresponding dyson equation , see below ) @xmath88 for @xmath89 , and @xmath90 for @xmath91 .
therefore the dilute gas approximation is expected to work quite well even for @xmath43 .
naturally when @xmath92 the above picture , based on strong rungs , breaks down . instead , the excitations in this regime should be viewed as weakly bound spinons , thus making the local triplets an inadequate staring point .
we have estimated that the bond operator formalism describes well the excitation spectrum for @xmath56 @xcite .
next we turn to the three - particle term @xmath81 which represents the dimerization part of the hamiltonian .
to lowest ( one - loop ) order there are contributions to the normal and anomalous self - energies , drawn in fig .
7(c ) and fig .
7(d ) , respectively .
the corresponding formulas are : @xmath93^{2}}{\omega - \omega_{q+\pi } - \omega_{k - q } } + \left\ { \begin{array}{c } u \leftrightarrow v \\
\omega \rightarrow -\omega \end{array } \right\},\ ] ] @xmath94 where the following definitions are used
@xmath95 @xmath96 the function @xmath97 originates from the fourier transform of the three - particle vertex eq.(8 ) .
equations ( 11 ) and ( 12 ) can be derived by evaluating all possible internal loops in the diagrams in figs .
7(c),(d ) ,
where , as usual , lines with a single arrow stand for the normal green s function @xmath98 and lines with double arrows represent the anomalous green s functions @xmath99 .
finally , the coupled dyson s equations for the normal and anomalous green s functions have to be solved leading to the result ( more details can be found in @xcite ) : @xmath100 [ \omega - a_{k } - \sigma^{n}(k,\omega ) ] + [ b_{k } + \sigma^{a}(k,\omega)]^{2}},\ ] ] where the total normal and anomalous self - energies are defined as @xmath101 in order to get a feeling for the effect of @xmath81 on the spectrum , it is instructive to examine the second order perturbation theory result .
it can be obtained by noticing that to leading order @xmath102 .
thus in evaluating the leading order correction arising from the dimerization term one can set @xmath103 which means that only the normal self - energy contributes : @xmath104 one can see that to lowest order the renormalization of the spectrum for @xmath53 is the strongest while @xmath54 is not renormalized . for a fixed value of @xmath5 ,
the gap @xmath105 decreases with decreasing @xmath28 and vanishes at a critical ratio , signaling an instability of the rung dimer phase .
we have found the spectrum numerically by solving for the poles of the exact green s function eq.(15 ) , thus resumming the perturbation series to all orders .
the results for @xmath30 are presented in fig .
5 . the diagrammatic results are in good agreement with the dimer series expansions , however with decreasing @xmath28 the disagreement increases .
we would like to point out that even though the two methods have the same starting point ( strong rungs ) and typically agree quite well in the region of strong coupling , for smaller coupling the dimer series expansion performs better .
the reason is that the series expansion takes into account all terms to a particular ( finite ) order , while the diagrammatic approach sums up only the most dominant subclass of diagrams .
the small parameter @xmath87 controlling the diagrammatic expansion grows with decreasing @xmath28 and at @xmath106 ( the critical point , as estimated by the dimer series and exact diagonalization ) has the value @xmath107 .
this relatively large value of the triplet density is related to the considerably decreased magnitude of the spin - spin correlation function on the rungs@xcite ( see the ( 1,2 ) correlations in fig .
therefore the agreement between the diagrammatic analysis and the dimer series is , in fact , even better than expected .
in conclusion , we have investigated the excitation spectrum and the ground state phase diagram of the quantum spin ladder with staggered dimerization and have found that a transition takes place form a phase dominated by strong rung correlations into a phase , characterized by intrachain dimers .
the two phases are separated by a gapless line .
the location of the boundary , the triplet excitations , as well as the spin - spin correlations in the rung dimer phase were calculated by dimer series expansions , diagrammatic analysis of the effective hard - core bose gas of rung triplets , and exact diagonalizations .
we have identified the three - particle scattering @xmath81 , in terms of the rung triplets , to be responsible for the instability of the rung dimer phase ( closing of the gap ) .
it is interesting to note that this is indeed the term which , in the continuous field theory formulation , contributes to the topological angle @xmath6 and leads to the transition .
indeed , from the representation of the spin operator on a given rung @xmath67 in terms of the triplets @xcite one can easily see that @xmath108 , @xmath109 .
these are precisely the two vector fields used in the derivation of the non - linear sigma model @xcite .
therefore the terms of the form @xmath110 which appear in our lattice hamiltonian @xmath81 would be precisely the ones which give a non - trivial @xmath6 after the continuum limit is taken .
there are several additional aspects of the problem that can be addressed within the strong - coupling formalism described in this paper .
as suggested in ref.@xcite the spinons become unconfined on the gapless line separating the two phases of the model ( see fig .
3 . ) . since in the two massive phases confinement definitely takes place , we would expect that the triplet spectral weight decreases as the critical line is approached and ultimately goes to zero .
in addition , at the point of deconfinement the same energy is required to create triplet and singlet excitations .
therefore , close to the critical line , a low - energy singlet excitation must appear in the spectrum . within the strong - coupling approach
a singlet can be viewed as a collective bound state of two triplets @xcite .
while in the spin ladder without dimerization the singlet bound state is high in energy ( small binding energy ) , in the present model we expect it to become a truly low - energy state ( strong triplet binding ) .
we leave these issues for a future study @xcite .
while this manuscript was being prepared for publication we became aware of two recent preprints devoted to the same model @xcite .
both of these works have studied the phase diagram numerically via exact diagonalization and their results for the location of the critical line are in agreement with ours .
we would like to thank o.p .
sushkov for numerous stimulating discussions and critical reading of the manuscript .
this work was supported by a grant from the australian research council .
the computations have been performed on silicon graphics power challenge and convex machines , and we thank the new south wales centre for parallel computing for facilities and assistance with the calculations . for a review of the nlsm approach , see : i. affleck , in _ fields , strings , and critical phenomena _ , edited by e. brezin and j. zinn - justin ( north - holland , amsterdam , 1989 ) , p.565 ; g. sierra , cond - mat/9610057 ( unpublished ) . | we study the excitation spectrum and ground state properties of the two - leg @xmath0 quantum spin ladder with staggered dimerization .
two massive phases , separated by a critical line are found , as predicted by previous analysis , based on the non - linear sigma model ( nlsm ) .
we have used dimer series expansions , exact diagonalization of small clusters and diagrammatic analysis of an effective interacting bose gas hamiltonian , obtained by using the bond operator representation for spins .
we compute the excitation spectrum in the phase , dominated by strong rungs in the parameter regimes far away and close to the point of instability .
the exact location of the phase boundary is determined . 0.6 cm |
+ this contribution deals with the implementation of procedures and methods , worked out in our group during the last few years , in order to provide algorithmic solutions to the problem of determining the first passage time ( fpt ) probability densities ( pdf ) and its relevant statistics for continuous state - space and continuous parameter gaussian processes describing the stochastic modeling of a single neuron s activity . in most modeling approaches , it is customary to assume that a neuron is subject to input pulses occurring randomly in time , ( see , for instance , @xcite and references therein ) . as a consequence of the received stimulations , it reacts by producing a response that consists of a spike train .
the reproduction of the statistical features of such spike trains has been the goal of many researches who have focused the attention on the analysis of the interspike intervals .
indeed , the importance of the interspike intervals is due to the generally accepted hypothesis that the information transferred within the nervous system is usually encoded by the timing of occurrence of neuronal spikes . to describe the dynamics of the neuronal firing we consider a stochastic process @xmath0 representing the change in the neuron membrane potential between two consecutive spikes ( cf . , for instance , @xcite ) . in this context
, the threshold voltage is viewed as a deterministic function @xmath1 and the instant when the membrane potential crosses @xmath1 as a fpt random variable .
the modeling of a single neuron s activity by means of a stochastic process has been the object of numerous investigations during the last four decades . a milestone contribution in this direction
is the much celebrated paper by gerstein and mandelbrot @xcite in which a random walk and its continuous diffusion limit ( the wiener process ) was proposed with the aim of describing a possible , highly schematized , spike generation mechanism .
however , despite the excellent fitting of a variety of data , this model has been the target of severe criticism on the base of its extreme idealization in contrast with some electrophysiological evidence : for example , this model does not take into account the spontaneous exponential decay of the neuron membrane potential .
an improved model is the so called ornstein - uhlenbeck ( ou ) model , that embodies the presence of such exponential decay .
however , the ou model does not allow to obtain any closed form expression for the firing pdf , except for some very particular cases of no interest within the neuronal modeling context .
rather cumbersome computations are thus required to obtain evaluations of the statistics of the firing time .
successively , alternative neuronal models have been proposed , that include more physiologically features .
the literature on this subject is too vast to be recalled here .
we limit ourselves to mentioning that a review of most significant neuronal models can be found in @xcite , @xcite and in the references therein .
in particular , in @xcite it is presented an outline of appropriate mathematical techniques by which to approach the fpt problem in the neuronal context .
we shall now formally define the firing pdf for a model based on a stochastic process @xmath0 with continuous sample paths .
first , assume @xmath2=1 $ ] , with @xmath3 i.e. we view the sample paths of @xmath0 as originating at a preassigned state @xmath4 at initial time @xmath5 .
then , @xmath6 is the fpt of @xmath0 through @xmath1 , and @xmath7 = \frac{\partial}{\partial t } p(t_{x_0 } < t)\ ] ] is its pdf .
henceforth , the fpt pdf @xmath8 $ ] will be identified with the firing pdf of a neuron whose membrane potential is modeled by @xmath0 and whose firing threshold is @xmath1 . throughout this paper
, we shall focus our attention on neuronal models rooted on diffusion and gaussian processes , partially motivated by the generally accepted hypothesis that in numerous instances the neuronal firing is caused by the superposition of a very large number of synaptic input pulses which is suggestive of the generation of gaussian distributions by virtue of some sort of central limit theorems .
it must be explicitly pointed out that models based on diffusion processes are characterized by the lack of memoryas a consequence of the underlying markov property . however , in the realistic presence of correlated input stimulations , the markov assumption breaks down and one faces the problem of considering more general stochastic models , for which the literature on fpt problem appears scarce and fragmentary .
simulation procedures then provide possible alternative investigation tools especially if they can be implemented on parallel computers , ( see @xcite ) .
the goal of a typical simulation procedure is to sample @xmath9 values of the fpt by a suitable construction of @xmath9 time - discrete sample paths of the process and then to record the instants when such sample paths first cross the boundary .
in such a way , one is led to obtain estimates of the firing pdf and of its statistics , that may be implemented for data fitting purposes .
the aim of this paper is to outline numerical and theoretical methods to characterize the fpt pdf for gaussian processes .
attention will be focused on markov models in section [ section2 ] , and on non - markov models in section [ section3 ] .
finally , section [ section4 ] will be devoted to the description of some computational results .
+ we start briefly reviewing some essential properties of gauss - markov processes .
let @xmath10 , where @xmath11 is a continuous parameter set , be a real continuous gauss - markov process with the following properties ( cf .
@xcite ) : _ ( i ) _ : : @xmath12 $ ] is continuous in @xmath11 ; _ ( ii ) _ : : the covariance @xmath13\,[x(t)-m(t)]\bigr\}$ ] is continuous in @xmath14 ; _ ( iii ) _ : : @xmath0 is non - singular , except possibly at the end points of @xmath11 where it could be equal to @xmath15 with probability one .
a gaussian process is markov if and only if its covariance satisfies @xmath16 it is well known @xcite , that well - behaved solutions of ( [ eq:(2.1 ) ] ) are of the form @xmath17 where @xmath18 is a monotonically increasing function by virtue of the cauchy - schwarz inequality , and @xmath19 because of the assumed nonsingularity of the process on @xmath11 .
the conditional pdf @xmath20 of @xmath0 is a normal density characterized respectively by conditional mean and variance @xmath21\\ v(t|t_0 ) & = & h_2(t)\,\biggl[h_1(t)-{h_2(t)\over h_2(t_0)}\;h_1(t_0)\biggr],\end{aligned}\ ] ] with @xmath22 .
it satisfies the fokker - planck equation and the associated initial condition @xmath23+{1\over 2}\;{\partial^2 \over \partial x^2}\ ; [ a_2(t)\,f(x , t|y,\tau)],\\ \\ & & \lim_{\tau \uparrow t}\,f(x , t|y,\tau)=\delta(x - y),\end{aligned}\ ] ] with @xmath24 and @xmath25 given by @xmath26\;{h_2^{\prime}(t)\over h_2(t)}\,,\qquad a_2(t ) = h_2 ^ 2(t)\;r^{\prime}(t),\ ] ] the prime denoting derivative with respect to the argument . the class of the gauss - markov processes @xmath27 , such that @xmath28 is characterized by means and covariances of the following two forms : @xmath29 @xmath30 + @xmath31 or @xmath29 @xmath32 $ ] + @xmath33 .
the first type includes the wiener process , used in @xcite to describe the neuronal firing , while the second type includes the ornstein uhlenbeck process that has often been invoked as a model for neuronal activity ( see , for instance , @xcite ) . any gaussian process with covariance as in ( [ eq:(2.2 ) ] )
can be represented in terms of the standard wiener process @xmath34 as @xmath35 , \label{eq:(2.4)}\ ] ] and is therefore markov .
this last equation suggests the way to construct the fpt pdf of a gauss - markov process @xmath0 in terms of the fpt pdf of the standard wiener process @xmath36 as an example , from ( [ eq:(2.4 ) ] ) for the conditioned fpt pdf one has : @xmath7 = { dr(t)\over dt}\;g_w\bigl\{s^*[r(t)],r(t)|x_0^*,r(t_0)\bigr\ } , \label{eq:(2.12)}\ ] ] where @xmath37 is defined in ( [ eq:(2.3 ) ] ) and @xmath38 $ ] is the fpt pdf of @xmath39 from @xmath40 at time @xmath41 to the continuous boundary @xmath42 , with @xmath43 \over h_2[r^{-1}(\vartheta_0)]},\qquad s^*(\vartheta)={s[r^{-1}(\vartheta)]-m[r^{-1}(\vartheta ) ] \over h_2[r^{-1}(\vartheta)]}\,.\ ] ] results on the fpt pdf for the standard wiener process can thus in principle be used via ( [ eq:(2.12 ) ] ) to obtain the fpt pdf of any continuous gauss - markov process .
for instance , if @xmath42 is linear in @xmath44 , @xmath38 $ ] is known and @xmath8 $ ] can be obtained via ( [ eq:(2.12 ) ] ) . instead
, if @xmath38 $ ] is not known , a numerical algorithm , or a simulation procedure , should be used for the standard wiener process and , after that , @xmath8 $ ] can be obtained via the indicated transformation .
the above procedure often exhibits the serious drawback of ensuing unacceptable time dilations ( see @xcite ) .
for instance , exponentially large times are involved when transforming the ornstein - uhlenbeck process to the wiener process , which makes such a method hardly viable .
hence , it is desirable to dispose of a direct and efficient computational method to obtain evaluation of the firing pdf .
along such a direction , in @xcite it has been proved that the conditioned fpt density of a gauss - markov process can be obtained by solving the non - singular volterra second kind integral equation @xmath45 \!=\ ! - 2 \psi[s(t),t|x_0,t_0 ] + 2 \int_{t_0}^t\ !
g[s(\tau),\tau|x_0,t_0]\ , \psi[s(t),t|s(\tau),\tau]\;d \tau\nonumber\\ & & \hspace*{10cm}\bigl(x_0<s(t_0)\bigr ) \label{eq:(3.1)}\end{aligned}\ ] ] with @xmath46 and @xmath47 = \biggl\{{s^{\prime}(t ) - m^{\prime}(t)\over 2 } \ ; -{s(t)-m(t)\over 2}\;{h^{\prime}_1(t)h_2(\tau ) -h^{\prime}_2(t)h_1(\tau)\over
h_1(t)h_2(\tau)-h_2(t)h_1(\tau ) } \nonumber \\ & & \hspace*{3.3 cm } - \ ; { y - m(\tau)\over 2}\;{h^{\prime}_2(t)h_1(t)-h_2(t ) h^{\prime}_1(t)\over h_1(t)h_2(\tau)-h_2(t)h_1(\tau)}\biggr\ } \ ; f[s(t),t\,|\,y,\tau ] \label{psi}\end{aligned}\ ] ] where @xmath48 $ ] is the transition pdf of @xmath49 closed form solutions of ( [ eq:(3.1 ) ] ) are available in @xcite for different families of boundaries . by making use of this result , in @xcite
an efficient numerical procedure based on a repeated simpson s rule has been proposed to evaluate fpt densities of gauss - markov processes , that can be implemented to obtain reliable evaluations of firing densities for neuronal models based on gauss - markov processes .
+ the methods proposed in the previous section rest on the strong markov assumption on the stochastic process modeling the neuron s membrane potential , which grants the possibility of making use of the mentioned analytic and computational methods for fpt pdf evaluations .
this is not the case when the stochastic process used to model the neuron s firing mechanism is non - markov . here
we shall focus our attention on gauss non - markov processes .
however , thus doing we face the lack of effective analytical methods for obtaining manageable closed - form expressions for the fpt pdf , although some preliminary analytical results have been obtained by ricciardi and sato in @xcite for a class of stationary gaussian processes .
indeed , if @xmath0 is one - dimensional non - singular stationary gaussian process mean square differentiable , a series expansion for the fpt pdf was derived ( see , @xcite ) . in the following , for convenience
, we shall take @xmath50 as initial time and , without loss of generality , assume that @xmath51=0 $ ] and @xmath2=1 $ ] , with @xmath4 an arbitrarily specified initial state .
furthermore , the covariance function @xmath52 : = \gamma ( t-\tau ) $ ] will be assumed to be such that @xmath53 and @xmath54 ( this last assumptions being equivalent to the mean square differentiable property ) .
the fpt pdf of @xmath0 through @xmath1 is then given by the following expression @xmath55=w_1(t|x_0)+\sum_{i=1}^\infty ( -1)^i\int_0^t \!\ ! dt_1 \!\ !
\int_{t_1}^t \!\ !
dt_2 \cdots \!\ !
\int_{t_{i-1}}^t \!\ ! \!\ !
dt_i w_{i+1}(t_1,\ldots , t_i , t|x_0 ) , \label{(0)}\ ] ] with @xmath56\ , p_{2n } [ s(t_1),\ldots , s(t_n ) ; z_1 , \ldots , z_n|x_0],\nonumber\end{aligned}\ ] ] where @xmath57 is the joint pdf of @xmath58 @xmath59 @xmath60 conditional upon @xmath61 . due to the great complexity of the involved multiple integrals , expression ( [ ( 0 )
] ) does not appear to be manageable for practical uses , even though it has recently been shown that it allows to obtain some interesting asymptotic results @xcite .
since ( [ ( 0 ) ] ) is a leibnitz series for each fixed @xmath62 estimates of the fpt pdf can in principle be obtained since its partial sum of order @xmath63 provides a lower or an upper bound to @xmath64 depending on whether @xmath63 is even or odd .
however , also the evaluation of such partial sums is extremely cumbersome . in conclusion , at the present time for this class of gaussian processes , no effective analytical methods , nor viable numerical algorithms
are available to evaluate the fpt pdf .
a simulation procedure seems to be the only residual way of approach . to this aim
, we have restored and updated an algorithm due to franklin @xcite in order to construct sample paths of a stationary gaussian process with spectral density of a rational type and deterministic starting point .
the idea is the following .
let us consider the linear filter @xmath65 where @xmath66 is the impulse response function and @xmath67 is the input signal . by fourier transformation , ( [ ( 3 ) ] ) yields @xmath68 where @xmath69 and @xmath70 are the spectral densities of input @xmath67 and output @xmath0 , respectively , and where @xmath71 denotes the fourier transform of @xmath72 equation ( [ ( 4 ) ] ) is suggestive of a method to construct a gaussian process @xmath0 having a preassigned spectral density @xmath73 .
it is indeed sufficient to consider a white noise @xmath74 having spectral density @xmath75 as the input signal and then select @xmath66 in such a way that @xmath76 . if the spectral density of @xmath0 is of rational type , namely if @xmath77 where @xmath78 and @xmath79 are polynomials with @xmath80 setting @xmath81 , from ( [ ( 4 ) ] ) it follows @xmath82 where @xmath83 to calculate the output signal @xmath0 it is thus necessary to solve first the differential equation @xmath84 to obtain the stationary solution @xmath85 , and then evaluate @xmath86 the simulation procedure is designed to construct sample paths of the process @xmath0 at the instants @xmath87 where @xmath88 is a positive constant time increment .
the underlying idea can be applied to any gaussian process having spectral densities of a rational type and , since the sample paths of the simulated process are generated independently of one another , this simulation procedure is particularly suited for implementation on supercomputers .
extensive computations have been performed on parallel computers to explore the different shapes of the fpt pdf as induced by the oscillatory behaviors of covariances and thresholds ( cf .
, for instance , @xcite and @xcite ) .
grafici / fig1.eps 8 cm plot of the boundary @xmath1 given in ( [ soglia ] ) for @xmath89 and @xmath90 ( bottom to top ) fig : fig1 grafici / fig2.eps 10 cm plots refer to fpt pdf @xmath91 through the boundary ( [ soglia ] ) with @xmath89 and @xmath92 for a zero - mean gaussian process characterized by correlation function ( [ ( funcorr ) ] ) . in figure 2(a ) @xmath91 given by ( [ ( fptclosed ) ] ) has been plotted .
the estimated fpt pdf @xmath93 with @xmath94 is shown in figure 2(b ) , with @xmath95 in figure 2(c ) and with @xmath96 in figure 2(d ) .
fig : fig2 grafici / fig3.eps 10 cm same as in figura [ fig : fig2 ] with @xmath97 .
+ the aim of this section is to compare the behavior of the fpt pdf s among gauss - markov processes and gaussian non - markov processes , in order to analyze how the lack of memory affects the shape of the density , also with reference to the specified type of correlation function . for simplicity ,
set @xmath98 and @xmath99=1,$ ] so that in the following we shall consider the fpt pdf @xmath100 $ ] . to be specific , we consider a stationary gaussian process @xmath0 with zero mean and correlation function @xmath101 which is the simplest type of correlation having a concrete engineering significance @xcite .
when the correlation function is of type ( [ ( funcorr ) ] ) , @xmath0 is not mean square differentiable , since @xmath102 .
thus the series expansion ( [ ( 0 ) ] ) does not hold . however , specific assumptions on the parameter @xmath103 help us characterize the shape of the fpt pdf .
we start assuming @xmath104 so that the correlation function ( [ ( funcorr ) ] ) can be factorized as @xmath105 hence , choosing @xmath106 and @xmath107 in ( [ eq:(2.2 ) ] ) , @xmath0 becomes gauss - markov .
therefore , for any boundary @xmath108 the fpt pdf @xmath91 can be numerically evaluated by solving the integral equation ( [ eq:(3.1 ) ] ) . in the following , we consider boundaries of the form @xmath109\biggr\ } , \label{soglia}\ ] ] with @xmath110 .
it is evident that @xmath111 and that @xmath1 tends to 0 as @xmath112 increases .
furthermore , as @xmath113 decreases , the boundary becomes flatter . in figure
[ fig : fig1 ] , the boundary @xmath1 given in ( [ soglia ] ) is plotted for @xmath89 and for two choices of the parameter @xmath113 , i.e. @xmath92 and @xmath97 . as proved in @xcite for boundaries of the form ( [ soglia ] ) , the fpt pdf @xmath91 of a gauss - markov process admits the following closed form : @xmath114 , \label{(fptclosed)}\ ] ] where @xmath115 $ ] is the transition pdf of the gauss - markov process @xmath49 for a zero - mean gauss - markov process characterized by the correlation function ( [ ( funcorr ) ] ) with @xmath89 and @xmath116 , the fpt pdf @xmath91 given by ( [ ( fptclosed ) ] ) through the boundary ( [ soglia ] ) is plotted in figure 2(a ) for @xmath92 and in figure 3(a ) for @xmath97 . note that as @xmath113 increases the mode increase , whereas the corresponding ordinate decrease . setting @xmath117 in ( [ ( funcorr ) ] ) , the gaussian process @xmath0 is no longer markov and its spectral density is given by @xmath118 thus being of a rational type .
since in ( [ ( spectral ) ] ) the degree of the numerator is less than the degree of the denominator , it is possible to apply the simulation algorithm described in section 3 in order to estimate the fpt pdf @xmath93 of the process .
the simulation procedure has been implemented by a parallel fortran 90 code on a 128-processor ibm sp4 supercomputer , based on mpi language for parallel processing .
the number of simulated sample paths has been set equal to @xmath119 .
the estimated fpt pdf @xmath93 through the boundary ( [ soglia ] ) with @xmath89 and @xmath92 are plotted in figures 2(b)@xmath1202(d ) for gaussian processes with correlation function ( [ ( funcorr ) ] ) having @xmath121 , respectively . for the same processes , figures 3(b)@xmath1203(d )
show the estimated fpt pdf @xmath93 through the boundary ( [ soglia ] ) with @xmath89 and @xmath97 .
note that as @xmath103 increases , the shape of the fpt pdf @xmath93 becomes flatter and the related mode increases .
furthermore , as figures 2(a)-2(b ) and figures 3(a)-3(b ) show , @xmath93 is very similar to @xmath91 for small values of @xmath103 .
work performed within a joint cooperation agreement between japan science and technology corporation ( jst ) and universit di napoli federico ii , under partial support by indam ( g.n.c.s ) .
99 first passage time densities evaluation for simulated gaussian process . _ cybernetics and system _ * 1 * , ( 2000 ) , 301306 .
simulation of gaussian processes and first passage time densities evaluation . _ lecture notes in computer science _ * 1798 * , ( 2000 ) , 319333 .
parallel simulations in fpt problems for gaussian processes . science and supercomputing at cineca .
report 2001 , ( 2001 ) , 405412 . a computational approach to first - passage - time problem for gauss - markov processes .
prob . _ * 33 * , ( 2001 ) , 453482 . on the asymptotic behavior of first passage time densities for stationary gaussian processes and varying boundaries . _ methodology and computing in applied probability _
( to appear ) .
numerical simulation of stationary and non stationary gaussian random processes .
_ siam review _ * 7 * , ( 1965 ) , 6880 . random walk models for the spike activity of a single neuron .
_ biophysical journal _ * 4 * , ( 1964 ) , 4168 .
certain properties of gaussian processes and their first - passage times . _ j. r. statist .
soc . _ _ ( b ) _ * 27 * , ( 1965 ) , 505522 .
diffusion processes and related topics in biology .
springer - verlag , new york , ( 1977 ) .
an outline of theoretical and algorithmic approaches to first passage time problems with applications to biological modeling . _ math .
japonica _ * 50 * , ( 1999 ) , 247322 . a note on first passage time for gaussian processes and varying boundaries .
_ ieee trans .
inf . theory _ * 29 * , ( 1983 ) , 454457 . on the evaluation of first passage time densities for gaussian processes .
_ signal processing _ * 11 * , ( 1986 ) , 339357 .
diffusion models of neuron activity . in _ the handbook of brain theory and neural networks _
arbib , ed.),the mit press , cambridge , ( 2002 ) , 343 - 348 . _
e. di nardo : _ dipartimento di matematica , universit della basilicata , contrada macchia romana , potenza , italy + _ a.g .
nobile : _ dipartimento di matematica e informatica , universit di salerno , via s. allende , i-84081 baronissi ( sa ) , italy + _ e. pirozzi and l.m .
ricciardi : _ dipartimento di matematica e applicazioni , universit di napoli federico ii , via cintia , napoli i-80126 , italy | this paper focuses on the outline of some computational methods for the approximate solution of the integral equations for the neuronal firing probability density and an algorithm for the generation of sample - paths in order to construct histograms estimating the firing densities .
our results originate from the study of non - markov stationary gaussian neuronal models with the aim to determine the neuron s firing probability density function . a parallel algorithm has been implemented in order to simulate large numbers of sample paths of gaussian processes characterized by damped oscillatory covariances in the presence of time dependent boundaries .
the analysis based on the simulation procedure provides an alternative research tool when closed - form results or analytic evaluation of the neuronal firing densities are not available . |
recently modern information - communication - technology ( ict ) has opened us access to large amounts of stored digital data on human communication , which in turn has enabled us to have unprecedented insights into the patterns of human behavior and social interaction .
for example we can now study the structure and dynamics of large - scale human communication networks @xcite and the laws of mobility @xcite , as well as the motifs of individual behavior @xcite .
one of the robust findings of these studies is that human activity over a variety of communication channels is inhomogeneous , such that high activity bursts of rapidly occurring events are separated by long periods of inactivity @xcite .
this feature is usually characterized by the distribution of inter - event times @xmath0 , defined as time intervals between , e.g. , consecutive e - mails sent by a single user .
this distribution has been found to have a heavy tail and show a power - law decay as @xmath1 @xcite . in human behavior
obvious causes of inhomogeneity are the circadian and other longer cycles of our lives as results of natural and societal factors .
malmgren _ et al . _
@xcite suggested that an approximate power - law scaling found in the inter - event time distribution of human correspondence activity is a consequence of circadian and weekly cycles affecting us all , such that the large inter - event times are attributed to nighttime and weekend inactivity .
as an explanation they proposed a cascading inhomogeneous poisson process , which is a combination of two poisson processes with different time scales .
one of them is characterized by the time - dependent event rate representing the circadian and weekly activity patterns , while the other corresponds to the cascading bursty behavior with a shorter time scale .
their model was able to reproduce an apparent power - law behavior in the inter - event time distribution of email and postal mail correspondence .
in addition they calculated the fano and allan factors to indicate the existence of some correlations for the email data as well as for their model of inhomogeneous poisson process , with quite good comparison @xcite .
however , the question remains whether in addition to the circadian and weekly cycle driven inhomogeneities there are also other correlations due to human task execution that contribute to the inhomogeneities observed in communication patterns , as suggested , e.g. , by the queuing models @xcite .
there is evidence for this by goh and barabsi @xcite , who introduced a measure that indicates the communication patterns to have correlations .
recently , wu _ et al .
_ have studied the modified version of the queuing process proposed in @xcite by introducing a poisson process as the initiator of localized bursty activity @xcite .
this was aimed at explaining the observation that the inter - event time distributions in short message ( sm ) correspondence follow a bimodal combination of power - law and poisson distributions .
the power - law ( poisson ) behavior was found dominant for @xmath2 ( @xmath3 ) .
since the event rates extracted from the empirical data have the time scales larger than @xmath4 ( also measured empirically ) , a bimodal distribution was successfully obtained . however , in their work the effects of circadian and weekly activity patterns were not considered , thus needing to be investigated in detail . as the circadian and weekly cycles affect human communication patterns in quite obvious ways , taking place mostly during the daytime and differently during the weekends , our aim in this paper is to remove or de - season from the data the temporal inhomogeneities driven by these cycles .
then the study of the remaining de - seasoned data would enable us to get insight to the existence of other human activity driven correlations .
this is important for two reasons .
first , communication patterns tell about the nature of human behavior .
second , in devising models of human communication behavior the different origins of inhomogeneities should be properly taken into account ; is it enough to describe the communication pattern by an inhomogeneous poissonian process or do we need a model to reflect correlations in other human activities , such as those due to task execution ? in this paper , we provide a systematic method to de - season the circadian and weekly patterns from the mobile phone communication data .
firstly , we extract the circadian and weekly patterns from the time - stamped communication records and secondly , these patterns are removed by rescaling the timings of the communication events , i.e. phone calls and sms .
the rescaling is performed such that the time is dilated or contracted at times of high or low event activity , respectively .
finally , we obtain the inter - event time distributions by using the rescaled timings and comparing them with the original distributions to check how the heavy tail and burstiness behavior are affected . as the main results we find that the de - seasoned data still shows heavy tail inter - event time distributions with power - law scalings thus indicating that human task execution is a possible cause of remaining burstiness in mobile phone communication .
this paper is organized as follows . in section [
sect : analysis ] , we introduce the methods for de - seasoning the circadian and weekly patterns systematically in various ways . by applying these methods the values of burstiness of inter - event time distributions are obtained and subsequently discussed .
finally , we summarize the results in section [ sect : summary ] .
we investigate the effect of circadian and weekly cycles on the heavy - tailed inter - event time distribution and burstiness in human activity by using the mobile phone call ( mpc ) dataset from a european operator ( national market share @xmath5 ) with time - stamped records over a period of @xmath6 days starting from january 2 , 2007 .
the data of january 1 , 2007 are not considered due to its rather unusual pattern of human communication .
we have only retained links with bidirectional interaction , yielding @xmath7 users , @xmath8 links , and @xmath9 events ( calls ) .
for the analysis of short message ( sm ) dataset , see appendix .
we perform the de - seasoning analysis by defining first the observable . for an individual service user @xmath10
, @xmath11 denotes the number of events at time @xmath12 , where @xmath12 ranges from @xmath13 seconds , i.e. the start of january 2 , 2007 at midnight , to @xmath14 seconds ( 119 days ) .
the total number of events @xmath15 is called the strength of user @xmath10 . in general , for a set of users @xmath16 , the number of events at time @xmath12 is denoted by @xmath17 .
@xmath16 can represent one user , a set of users , or the whole population .
when the period of cycle @xmath18 is given , the event rate @xmath19 with @xmath20 is defined as @xmath21 for convenience , we redefine the periodic event rate with period @xmath18 as @xmath22 with any non - negative integer @xmath23 for @xmath24 . by means of the event rate
, we define the rescaled time @xmath25 as following @xcite @xmath26 this rescaling corresponds to the transformation of the time variable by @xmath27 with @xmath28 . here
@xmath28 means that there exists no cyclic pattern in the frame of rescaled time .
the time is dilated ( contracted ) at the moment of high ( low ) activity . in order to check whether the rescaling affects the inter - event time distributions and
whether still some burstiness exists we reformulate the inter - event time distributions by using rescaled event times and compare them with the original distributions .
the definition of the rescaled inter - event time from the rescaled time is straightforward . considering two consecutive events of a user @xmath29 occurring at times @xmath30 and @xmath31 ,
the original inter - event time is @xmath32 , then the corresponding rescaled inter - event time is defined as follows @xmath33 to find out how much the de - seasoning affects burstiness , we measure the burstiness of events , as proposed in @xcite , where the burstiness parameter @xmath34 is defined as @xmath35 here @xmath36 and @xmath37 are the standard deviation and the mean of the inter - event time distribution @xmath38 , respectively . the value of @xmath34 is bounded within the range of @xmath39 $ ] such that @xmath40 for the most bursty behavior , @xmath41 for neutral or homogeneous poisson behavior , and @xmath42 for completely regular behavior .
the burstiness of the original inter - event time distribution , denoted by @xmath43 , is to be compared with that of @xmath44 of the rescaled inter - event time distribution for given period @xmath18 . with the de - seasoning the burstiness
is expected to decrease and here we are most interested in by what amount the burstiness decreases when using @xmath45 day or @xmath46 days , i.e. removing the circadian or weekly patterns . [ cols="^,^ " , ] we apply the same de - seasoning analysis described in the main text to the sm dataset .
figure [ fig : sms_individual ] shows the various circadian activity patterns of individual users .
the main difference from the mpc dataset is that the activity peak is around 11 pm .
this feature becomes evident if the averaged event rates are obtained from the same strength groups or from the groups with broad ranges of strength , shown in the left columns of figs [ fig : sms_strength_merge ] and [ fig : sms_group ] .
for the details of the strength dependent grouping , see table [ table : sms_group ] .
the inter - event time distributions and their values of burstiness are also compared . at first
, the distributions can not be described by the simple power - law form but by the bimodal combination of power - law and poisson distributions as suggested by @xcite . as shown in fig .
[ fig : sms_burstiness ] , the values of burstiness slowly decrease as the period @xmath18 increases up to 7 days , which implies that de - seasoning the circadian and weekly patterns does not considerably affect the bursty behavior . finally , we perform the power spectrum analysis for sm dataset in fig . [
fig : sms_powerspec ] and find again that the cyclic patterns longer than @xmath18 can not be removed by the de - seasoning with period @xmath18 .
all these results confirm our conclusion that the heavy tail and burstiness are not only the consequence of circadian and other longer cycle patterns but also due to other correlations , such as human task execution .
financial support by aalto university postdoctoral program ( hj ) , from eu s 7th framework program s fet - open to ictecollective project no .
238597 and by the academy of finland , the finnish center of excellence program 2006 - 2011 , project no .
129670 ( mk , kk , jk ) , as well as by tekes ( fidipro ) ( jk ) are gratefully acknowledged .
the authors thank a .-
barabsi for the mobile phone communication dataset used in this research .
10 j .- p .
onnela , j. saramki , j. hyvnen , g. szab , m. argollo de menezes , k. kaski , a .-
barabsi , and j. kertsz .
analysis of a large - scale weighted network of one - to - one human communication .
, 9:179 , 2007 .
r. d. malmgren , j. m. hofman , l. a. n. amaral , and d. j. watts . characterizing individual communication patterns . in _
kdd 09 : proceedings of the 15th acm sigkdd international conference on knowledge discovery and data mining _ , pages 607616 , new york , ny , usa , 2009 .
acm press . | the temporal communication patterns of human individuals are known to be inhomogeneous or bursty , which is reflected as the heavy tail behavior in the inter - event time distribution . as the cause of such bursty behavior two main mechanisms
have been suggested : a ) inhomogeneities due to the circadian and weekly activity patterns and b ) inhomogeneities rooted in human task execution behavior . here
we investigate the roles of these mechanisms by developing and then applying systematic de - seasoning methods to remove the circadian and weekly patterns from the time - series of mobile phone communication events of individuals .
we find that the heavy tails in the inter - event time distributions remain robustly with respect to this procedure , which clearly indicates that the human task execution based mechanism is a possible cause for the remaining burstiness in temporal mobile phone communication patterns . |
it has been known well that , in the limit of vanishing binding energy , the root - mean - square radius diverges for a wave function in a short range potential well with orbital angular momenta of @xmath0 = 0 and 1 @xcite .
halo nuclei , which are characterized by an extended density distribution @xcite , have been interpreted to be due to such divergence with a single - particle wave function for @xmath0 = 0 and 1 @xcite .
@xcite for recent review articles on halo nuclei . for even - even nuclei
, the pairing correlation among valence neutrons plays a decisive role in the structure of weakly bound nuclei @xcite .
bennaceur , dobaczewski , and ploszajczak have demonstrated that the root - mean - square radius does not diverge for even - even nuclei as the pairing correlation largely suppresses the halo structure in odd - mass nuclei , which has been referred to as the pairing anti - halo effect @xcite . in refs .
@xcite , we have argued that the odd - even staggerings observed in reaction cross sections @xcite can be interpreted in terms of the pairing anti - halo effect ( see also refs .
@xcite ) .
the pairing anti - halo effect has been studied using mainly the hartree - fock - bogoliubov ( hfb ) method @xcite . in this approach ,
the pairing anti - halo effect occurs because the quasi - particle energy remains a finite value even when a single - particle energy vanishes@xcite .
a key issue for this argument is that the pairing gap needs to be finite in the zero binding limit .
many hfb calculations have actually shown that it is indeed the case @xcite , leading to a suppression of the halo structure in even - even systems .
although the hfb method provides a clear mathematical interpretation of the pairing anti - halo effect , its physical mechanism is less transparent .
the aim of this paper is to propose a more intuitive idea on the pairing anti - halo effect , using a three - body model with a core nucleus and two valence neutrons .
this model is formulated with a simpler single - particle basis , providing a complementary interpretation to the one based on the hfb method .
the paper is organized as follows . in sec .
ii , we first show how the pairing anti - halo effect is realized in the hfb method .
we then introduce the three - body model and discuss the pairing anti - halo effect in this model . in sec .
iii , we introduce a quasi - particle wave function within the three - body model and investigate its structure .
we show that a coherent superposition of a weakly bound state and continuum states is a key ingredient of the pairing anti - halo effect .
we then summarize the paper in sec .
before we discuss the pairing anti - halo effect with a three - body model , we first show how it is understood in the hfb method .
this is also to clarify the notation used in this paper .
for a two - body system which consists of a valence neutron and a core nucleus , we assume that the wave function for the relative motion obeys the schrdinger equation given by , @xmath1\psi_{nljm}({\mbox{\boldmath $ r$ } } ) = \epsilon_{nlj}\,\psi_{nljm}({\mbox{\boldmath $ r$ } } ) , \label{two - body}\ ] ] where @xmath2 is the reduced mass and @xmath3 is the potential between the valence neutron and the core nucleus .
we have assumed that @xmath3 is local and has spherical symmetry so that the wave function is characterized by the orbital angular momentum @xmath0 , the total angular momentum @xmath4 and its @xmath5-component , @xmath6 , as well as the radial quantum number , @xmath7 . here , @xmath8 is the energy eigenvalue . for simplicity
, we consider only an @xmath9 wave solution of this schrdinger equation .
the radial wave function @xmath10 , defined with the spin - angular function , @xmath11 , by @xmath12 behaves asymptotically as , @xmath13 where @xmath14 is defined as @xmath15 .
the expectation value of @xmath16 then reads , @xmath17 which apparently diverges in the limit of @xmath18 . in many - body systems with the pairing correlation
, one may consider the hfb equations given by @xcite , @xmath19 where @xmath20 is the mean - field hamiltonian given in eq .
( [ two - body ] ) , @xmath21 is the pairing potential , and @xmath22 is the chemical potential .
we have again assumed that both the mean - field potential in @xmath20 and the pairing potential , @xmath21 , are spherical and local functions of @xmath23 . in the hfb equations ,
( [ hfb ] ) , @xmath24 is a quasi - particle energy and @xmath25 and @xmath26 are the upper and the lower components of a quasi - particle wave function , respectively .
these are ortho - normalized according to , @xmath27 = \delta_{\alpha,\beta } , \label{normalizationhfb}\ ] ] where @xmath28 and @xmath29 are shorthanded notations for @xmath30 .
the one - particle density , @xmath31 , is given in terms of @xmath26 by , @xmath32 in the bcs approximation , @xmath26 is expressed by a product of the occupation factor @xmath33 and the single - particle wave function , @xmath34 @xcite .
in contrast , in the hfb , the asymptotic form of the radial part for the lower component , defined similarly to eq .
( [ radialwf2 ] ) , reads @xcite , @xmath35 for @xmath36 , where @xmath37 is given as @xmath38 .
the expectation value of @xmath16 with this wave function reads , @xmath39 notice that the quasi - particle energy , @xmath24 , is given in the bcs approximation as , @xmath40 , where @xmath41 is the pairing gap .
this implies that @xmath24 behaves as @xmath42 in the zero binding limit with @xmath43 .
the expectation value of @xmath16 then reads , @xmath44 which remains finite as long as the pairing gap , @xmath41 , is finite .
this is nothing but the pairing anti - halo effect proposed in ref .
an essential point for the pairing anti - halo effect is that the single - particle energy , @xmath8 , is replaced by the quasi - particle energy , @xmath45 , reflecting the pairing correlation , which then induces a shrinkage of wave function according to eq .
( [ low - component ] ) . in order to achieve a simple but still physical concept for the pairing anti - halo effect ,
let us now introduce a three - body model which consists of the core nucleus and two valence neutrons .
the hamiltonian for the three - body model reads @xcite , @xmath46 where the single - particle hamiltonian , @xmath20 , is the same as the one in eq .
( [ two - body ] ) and @xmath47 is a pairing interaction between the two valence neutrons .
the last term in this equation is the two - body part of the recoil kinetic energy of the core nucleus , whose mass is denoted by @xmath48 . using the eigen - functions of the single - particle hamiltonian @xmath20 , that is , the wave functions @xmath34 in eq .
( [ two - body ] ) , the two - particle wave function for the ground state of the three - body system with spin - parity of @xmath49 is given as , @xmath50 with @xmath51^{j=0 } , \\ & = & \sum_m\frac{(-1)^{j - m}}{\sqrt{2j+1}}\ , \psi_{nljm}({\mbox{\boldmath $ r$}}_1)\psi_{n'lj - m}({\mbox{\boldmath $ r$}}_2 ) .
\nonumber \\\end{aligned}\ ] ] ( for simplicity of the notation , we do not use here the anti - symmetrized basis @xcite .
the anti - symmetrization is realized by setting @xmath52 in eq .
( [ twoparticle ] ) ) . the one - particle density constructed with this two - particle wave function
is then given by @xcite , @xmath53 using the spherical reduction for the wave functions ( see eq .
( [ radialwf ] ) ) , one can show that the one - particle density is expressed as @xcite , @xmath54 where @xmath55 is defined as @xmath56 .
o. only the @xmath9-wave single - particle states are included in the calculations . in the uncorrelated case ,
the two valence neutrons occupy the 2@xmath57 state at @xmath58 mev , while they are scattered into the continuum states in the correlated case .
a zero - range pairing interaction is employed , which yields the ground state energy of @xmath59 mev .
( lower panel ) the decomposition of the correlated density into three components .
the ( bb ) component ( the dashed line ) corresponds to the one in which both of the two valence neutrons occupy the bound 2@xmath57 state , while in the ( bc ) component shown by the dot - dashed line , one of them is scattered to a continuum state .
the ( cc ) component shown by the dotted line corresponds to the one in which both of the valence neutrons are scattered into continuum states .
the total correlated density is also shown by the solid line . ]
the upper panel of fig .
1 shows a one - particle density in the three - body model . to draw this figure
, we consider the @xmath60o nucleus ( @xmath60o = @xmath61o + @xmath7 + @xmath7 ) , and employ a contact pairing interaction , @xmath62 , with @xmath63 mev @xmath64 , together with the cutoff energy of @xmath65 mev .
the continuum states are discretized with the box boundary condition with the box size of @xmath66 = 30 fm .
we use a woods - saxon potential for the mean - field potential , @xmath3 , with the radius parameter of @xmath67 = 3.5 fm and the diffuseness parameter of @xmath68 = 0.67 fm @xcite .
the depth of the woods - saxon potential is somewhat arbitrarily chosen to be @xmath69 = @xmath70 mev , which has the 2@xmath57 state at @xmath71 mev .
for simplicity of the discussion , we include only @xmath0 = 0 in eq .
( [ twoparticle ] ) , for which the 1@xmath57 state is assumed to be occupied by the core nucleus and is explicitly excluded in the summation . in this model space
, there are one bound state , 2@xmath57 , at @xmath72 mev , and five discretized @xmath9-wave continuum states up to 10 mev .
this calculation yields the ground state energy of @xmath59 mev .
the dashed line in the figure shows the one - particle density in the absence of the pairing interaction , which is proportional to the square of the 2@xmath57 wave function , that is , @xmath73 .
since the 2@xmath57 state is a weakly bound @xmath9-wave state , the resultant density has an extended long tail .
in contrast , in the correlated density distribution shown by the solid line , the density distribution is considerably shrunk compared to the uncorrelated density .
the root - mean - square radii are @xmath74 = 5.18 and 8.83 fm for the correlated and the uncorrelated cases , respectively .
this is a clear manifestation of the pairing anti - halo effect discussed in the previous sub - section .
the lower panel of fig .
1 shows a decomposition of the correlated density .
here , we decompose it into three components , that is , ( i ) ( bb ) : both @xmath7 and @xmath75 in eq .
( [ rho1 ] ) belong to the weakly bound state , 2@xmath57 , ( ii ) ( bc ) : one of them belongs to the bound state while the other belongs to a continuum state , and ( iii ) ( cc ) : both of them belong to continuum states .
the ( bb ) component in fact has the same radial profile as the uncorrelated density shown in the upper panel , having an extended tail .
the ( bc ) component behaves similarly to the ( bb ) component inside the potential , while it has the opposite sign to the ( bb ) component in the tail region . because of this
, the density in the inner part is enhanced while the ( bb ) and ( bc ) components are largely canceled out in the outer part .
one can thus find that the scattering of a particle to the continuum spectrum due to the pairing interaction plays an essential role in the pairing anti - halo effect .
the ( cc ) component , on the other hand , provides only a small portion of the correlated density , even though it is not negligible .
this component is positive in a wide range of radial coordinate , as one can see in the figure .
in order to get a deeper insight into the pairing anti - halo effect in the three - body model , we next re - express the one - particle density in a different form .
to this end , we first notice that the two - particle wave function , eq . ( [ twoparticle ] ) ,
is expressed as , @xmath76^{j=0},\ ] ] with @xmath77 the one - particle density , eqs .
( [ rho1 - 0 ] ) and ( [ rho1 ] ) , is then given as , @xmath78 where @xmath79 is defined as , @xmath80 notice that this is in a similar form to the one - particle wave function in the hartree - fock - bogoliubov approximation , especially if one expands the quasi - particle wave function , @xmath81 , on the hartree - fock basis , @xmath82 @xcite .
for this reason , we shall call @xmath83 a `` quasi - particle '' wave function hereafter .
notice that the quasi - particle wave functions @xmath84 are not orthonormalized , just the same as the hfb wave functions @xmath85 ( see eq .
( [ normalizationhfb ] ) ) . , defined by eq . ( [ qpwf - radial ] ) , for the weakly bound 2@xmath57 state .
the solid line shows the total wave function , while the dashed and the dot - dashed lines denote its bound state and continuum state contributions as defined by eq .
( [ decomposition ] ) , respectively . ] the solid line in fig .
2 shows the radial dependence of the quasi - particle wave function for the weakly - bound 2@xmath57 state , that is , @xmath79 with @xmath86 , for the three - body hamiltonian introduced in the previous subsection .
the dashed and the dot - dashed lines show its decomposition into the bound state and the continuum state contributions , respectively . they are defined as @xmath87 one can see that the main feature of this quasi - particle wave function is similar to the one - particle density shown in fig . 1 ( b ) .
that is , the bound state and the continuum state contributions are largely canceled with each other outside the potential while the two components contribute coherently in the inner region .
we notice that the localization due to a coherent superposition of continuum states is the same mechanism as a formation of a localized wave packet .
this is an essential ingredient of the pairing anti - halo effect , that is , a formation of localized wave packet induced by a pairing interaction .
state ( the solid line ) and the lowest discretized @xmath9-wave state at @xmath88 = 0.51 mev ( the dashed line ) .
3(b ) shows the radial wave function @xmath89 while fig .
3(a ) shows @xmath90 .
3(c ) : the integrand of the matrix element for the zero - range pairing interaction between the components @xmath91 $ ] and @xmath92 $ ] , where @xmath93 is the anti - symmetrizer and @xmath94 and @xmath95 are the wave functions for the weakly bound state and the lowest continuum state , respectively . ]
a question still remains concerning why the superposition is in such a way that the tail part of the bound wave function is suppressed . in order to clarify this point ,
let us strict ourselves only to two single - particle states , one is the weakly bound 2@xmath57 state at @xmath96 mev and the other is the lowest discretized @xmath9-wave continuum state .
for the potential given in the previous section , the latter state is at @xmath97 = 0.51 mev for @xmath66 = 30 fm .
3(a ) and 3(b ) show the radial part of the wave functions for these states .
the solid and the dashed lines correspond to the bound state and the scattering states , respectively .
3(b ) shows the radial wave functions , @xmath89 , while fig .
3(a ) shows @xmath98 .
in the inner part , the two wave functions behave similarly to each other , because the absolute value of the single - particle energies , @xmath99 , is small for both the states , so that @xmath100 . in the outer region where the potential @xmath3 disappears , the two wave functions should behave differently .
since they behave similarly in the inner region , the two wave functions have to have opposite sign in the outer region in order to fulfill the orthogonal condition . with these two single - particle states , we assume , for simplicity , that the two - particle wave function is given by , @xmath101 , \label{twoparticle-2}\ ] ] where @xmath94 and @xmath95 are the bound and the scattering wave functions , respectively , and @xmath93 is the anti - symmetrizer .
the coefficients @xmath102 and @xmath103 are obtained by solving the eigen - value equation , @xmath104 where @xmath105 and @xmath106 are the diagonal components of the three - body hamiltonian including the pairing matrix elements ( in the present case , @xmath105 and @xmath106 are @xmath107 and @xmath108 mev , respectively ) .
@xmath109 is the matrix element of the pairing interaction between the two configurations , that is , @xmath110^*[\phi_b(r)\phi_c(r)].\ ] ] the integrand is shown in fig .
as one can see , the integrand is positive , except for large values of @xmath23 , for which the contribution is negligibly small , and thus @xmath109 is negative for an attractive pairing interaction with @xmath111 0 .
the eigen - values @xmath112 and the corresponding eigen - vectors of eq .
( [ 2x2 ] ) read , @xmath113 where @xmath114 is the normalization factor . for the lower eigen - value , @xmath115 , the quantity @xmath116 reads @xmath117 which is apparently negative .
since @xmath109 and @xmath116 are both negative , @xmath102 and @xmath103 thus have the same sign to each other ( see eq .
( [ eigenvector ] ) ) , leading to the quasi - particle wave function which has a suppressed tail as shown in fig .
2 . this feature remains the case even when higher continuum states and/or the configuration with @xmath118 are included in the two - particle wave function .
there is a freedom for the phase of single - particle wave functions to take a positive value or a negative value at the origin . in fig .
3(b ) , the two @xmath57 wave functions are taken to be positive at the origin .
we notice that the shrinkage of the halo wave function is independent of the choice of the sign of the wave function .
that is , if one takes the negative sign for the continuum wave function at the origin , the sign of the pairing matrix @xmath109 turns to be positive so that @xmath102 and @xmath103 have a different sign from one another .
however , the one particle density as well as the quasi - particle wave function remain the same , since the sign of the amplitude @xmath102 in eq .
( [ eigenvector ] ) and that of the single - particle wave function , @xmath119 , in eq .
( [ twoparticle-2 ] ) are simultaneously altered , whereas the sign of @xmath103 remains the same .
we have discussed the pairing anti - halo effect from a three - body model perspective .
in contrast to the conventional understanding based on a hartree - fock - bogoliubov ( hfb ) wave function , the present study provides a simple and intuitive concept for the pairing anti - halo effect .
namely , we have found that an essential ingredient of the pairing anti - halo effect is a coherent superposition of a loosely - bound and continuum states due to a pairing interaction , which leads to a localized wave function as a wave packet .
the coherence of the wave functions results in an enhancement of one - particle density in the inner region while the long tail of a weakly bound wave function is largely canceled out with continuum wave functions .
the present study offers a complementary understanding for the pairing anti - halo effect to the one with the hfb approximation .
in fact , we have shown that the one - particle density with a three - body model can be cast into a similar form of the density in the hartree - fock - bogoliubov approximation .
we have pointed out that such `` quasi - particle '' wave functions show the shrinkage effect as a consequence of a coherent superposition of a weakly bound and continuum states .
this work was partly supported by by jsps kakenhi grant numbers jp16k05367 . | the pairing anti - halo effect is a phenomenon that a pairing correlation suppresses a divergence of nuclear radius , which happens for single - particle states with orbital angular momenta of @xmath0 = 0 and 1 in the limit of vanishing binding energy .
while this effect has mainly been discussed in terms of hartree - fock - bogoliubov ( hfb ) theory , we here use a three - body model and provide its new intuitive concept as a localized wave packet for a quasi - particle , that is , a coherent superposition of a weakly bound and continuum wave functions due to a pairing interaction .
we show that the one - particle density in the three - body model can be directly expressed with such quasi - particle wave functions , which have a close analog to wave functions in the hfb approximation . |
radiative - transfer modeling of supernova ( sn ) ia is a challenging enterprise ( @xcite , hereafter d13 ) . for a start ,
the initial ejecta conditions for such simulations are uncertain .
the possibility of both single- and double - degenerate progenitor systems suggests that the ejecta mass likely varies amongst sn ia ejecta ( e.g. , @xcite ) .
the ejecta composition that results from the combustion of a c / o white dwarf is not known with confidence because the explosion scenario , besides being non unique , involves hard - to - model combustion physics @xcite . as a result , for each scenario , simulations of the explosion , whether 1-d or multi - d , are parametrized rather than modeled consistently from first principles @xcite .
the progenitor is likely to depart from a simple hydrostatic configuration for many possible reasons including fast rotation @xcite , the conditions produced by the smoldering phase and ignition ( see , e.g. , @xcite ) , or from the complex dynamical evolution in a merger event @xcite .
even if the ejecta properties were accurately known , the modeling of the radiation would remain challenging because of the prevalence of line opacity , the strong influence of scattering resulting from the very low gas densities in the fast expanding low mass ejecta , the importance of non - lte and time - dependent effects , and non - thermal processes .
perhaps even more important are the numerous processes that take place between the various constituents of the gas ( electrons and ions ) , involving thousands of atomic levels , and which control its thermodynamic state . historically , this complexity has generally been interpreted in terms of an opacity " problem ( see , e.g. , @xcite ) . in @xcite , we demonstrate that the temperature and ionization distribution , which are difficult to determine accurately , are also important ( not surprisingly ) because the thermodynamic state of the gas determines which ions provide the opacity .
we found , for example , that forbidden lines are crucial , as early as bolometric maximum , in controling the cooling of the ejecta .
paradoxically , these lines have low oscillator strengths , and thus tend to be ignored in simulations prior to @xmath0100d after explosion .
finally , inaccuracies in the atomic data , or the lack of atomic data altogether , introduce a major source of uncertainty in any radiative transfer modeling of sne ia .
a central problem with sne in general is that lines are strongly doppler broadened by the fast expansion of the ejecta .
the large metal mass fraction in sne ia leads to the presence of forests of lines which overlap , typically preventing the identification of a
clean " line anywhere in their spectra .
the emerging radiation is in addition strongly influenced by a few strong lines , such as si6355 or the ca triplet , giving the wrong impression that the spectrum is analogous to a blackbody influenced by a few spectral features of large optical depth .
this situation is particularly problematic when sne ia evolve past maximum because at such times , the optical depth clearly drops , the ejecta turns nebular , but strong lines are still present .
there are no regions with negligible flux , even though the continuum optical depth is well below unity even at bolometric maximum @xcite .
after bolometric maximum , the peaks and valleys of sn ia spectra are a complex blend of numerous lines , some thick , others thin , each interacting with hundreds of other lines either locally ( within a sobolev length ) or non - locally ( because redshifted into resonance with a redder line ) .
this conspires to produce confusion about spectrum formation in sne ia . the concept of a photosphere " is routinely used but
the notion of a sharp boundary from where radiation would escape does not hold for sne ia . @xcite
propose that most / all lines at nebular times are permitted ( generally resonance ) transitions , while numerous papers emphasize the near exclusive presence of forbidden - line transitions @xcite . to give additional evidence for the importance of forbidden lines in sne ia @xcite
, we focus here on the 5900 feature observed in post - maximum sn ia spectra . in recent years
, this feature has been associated with nad , although this association remains suspicious and puzzling .
models sometimes provide a very good fit ( * ? ? ?
3 ) , a poor fit @xcite , or predict no feature at that location @xcite . in the next section , we give an historical perspective of earlier work that modeled or discussed the spectral feature at 5900 .
we then present results from our grid of delayed - detonation and pulsational - delayed - detonation models covering a range of mass and show that this feature can be explained as [ co]5888 emission , across the range from sub - luminous to standard sne ia ( section [ sect_res ] ) .
in other words , we find that such ejecta models naturally produce an emission feature at 5900 , and that this emission in our models is systematically associated with [ co]5888 .
our conclusions and a discussion of future work are presented in section [ sect_conc ] .
far from being a banal characteristic of sn ia spectra , the observation of the [ co]5888 line indicates that forbidden line transitions have to be incorporated in any sn ia model at and beyond bolometric maximum , not just to reproduce observations , but also to compute correctly the cooling of sn ia ejecta .
the spectral feature at 5900 in sne ia has been discussed repeatedly in the last two decades .
we can separate the studies focusing on the photospheric " phase ( epochs until soon after bolometric maximum ) and those devoted to advanced " nebular phase ( beyond 100d after explosion when the sn exhibits an apparently pure - emission spectrum ) .
@xcite is probably the first to study nebular - phase spectra of sne ia , and he associates unambiguously the 5900 feature with the forbidden transition of co at 5888 .
later , @xcite present numerical developments incorporated into the code eddington , and apply their technique to spectrum formation in a sn ia ejecta at 250d after explosion .
they propose as the origin of the 5900feature ( see their fig . 5 ) .
@xcite use the simultaneous presence of [ co]5888 and [ fe]4658 ( strictly speaking , the 4500 - 5000 region contains lines from both feand fe ) in sn ia spectra to confirm the radioactive decay of at the origin of the sn ia luminosity .
indeed , they find the flux ratio of these two lines ( ignoring line overlap and some complications of the radiative transfer ) can be explained from the decay of to .
they discard the possible association of the 5900 feature with na .
@xcite study sn1991bg at both early times and late times . at nebular times , they propose that the 5900 feature is primarily associated with [ co ] , and show how this co forbidden transition may be used to set constraints on the original mass .
their nebular model reproduces the 5900 feature , in both strength and width ( see their fig .
14 ) , supporting the same assessment made by @xcite .
.summary of nucleosynthetic yields for the chandrasekhar - mass delayed - detonation models used in this work .
numbers in parenthesis correspond to powers of ten . [ cols="<,^,^,^,^,^ " , ] a further confirmation of this association is that the observed strength of the 5900 feature typically increases after bolometric maximum . in the context of [ co ] emission , this also makes sense since the strength of forbidden lines relative to other lines should increase as the ejecta density drops .
we show in fig .
[ fig_seq ] a montage of spectra for the evolution of sn2005cf and model pddel3 @xcite from bolometric maximum ( i.e. , + 0d ) until + 80d ( we note that model ddc10 studied in detail in d13 does a somewhat better job in the @xmath1-band region , but model pddel3 is more compatible with the narrow line profiles of sn2005cf ; both models are equally suitable for the present discussion ) . as in fig .
[ fig_grid ] , we show the flux associated with the [ co]5888 line as a filled red area .
it is clearly evident that early after bolometric maximum , flux from that forbidden line contributes to the emergent radiation , and that this flux contribution increases with time , as observed .
we note that at bolometric maximum , [ co]5888 line emission is already a strong coolant of the co - rich layers ; [ co]5888 line photons are not seen earlier on because these regions are located below the last scattering / absorbing layer .
there are other [ co ] lines in the 6000 region ( table [ tab_co3 ] ) .
of all these , the transition at 5888 is the strongest .
the transition at 5906 is expected to be weaker and it also overlaps with the 5888 transition .
the transition at 6195 should be about 1/3 of the strength of 5888 but the model predicts essentially no flux in this line .
we find that the 6195 suffers absorption by overlapping lines , in particular from the si doublet at 6355 , which is strong at those epochs .
numerous other lines are present in this spectral region , while fewer reside in the 5900 region , allowing the [ co]5888 to escape .
we also find that these low - lying states are in lte with the ground state ( they have the same departure coefficients ) , although we obtain strong depopulation ( and strong departure from lte ) of the ground state of [ co ] through non - thermal ionization and excitation ( this departure is also epoch dependent ) .
scattering is thus expected to influence little the formation of these forbidden - line transitions .
we also see that in our simulations , the delayed detonation models systematically show a range of optical colors at + 40d , while observations look more similar at this time ( fig .
[ fig_grid ] ) .
this is particularly visible for the low - luminosity sn ia 1999by , which shows a stronger [ co]5888 line and relatively less flux in the red than in our model ddc25 .
this range in colors in our models reflects the trend in ionization level at the corresponding epoch , as evidenced by the ionization state of co in the in the inner ejecta where the spectrum forms ( fig .
[ fig_ion ] ) .
this could arise from problems with the atomic physics , or from issues with the structure computed for the assumed progenitor model .
another possible explanation may be that , while the mass is accurate in each model / observation pair ( tightly constrained by the peak luminosity ) , the ejecta mass of 1.38 , which is kept fixed here , may also vary and in particular involve sub - chandrasekhar mass wds ( see , e.g. , @xcite ) . for a fixed mass , reducing the ejecta mass leads naturally to a significant increase in ejecta ionization ( see , e.g. , @xcite ) .
from this work , we unambiguously discard nad as the origin of the 5900feature in sne ia after bolometric maximum based on 1 ) the prediction of [ co]5888 line emission from all our delayed - detonation models after bolometric maximum ; 2 ) the strengthening of that [ co ] line with time in the first 2 months after bolometric maximum and 3 ) the satisfactory match of our synthetic spectra to the observations .
this finding is not trivial taxonomy because it further confirms our conclusions from the more theoretical discussion in d13 .
although the [ co ] line was identified three decades ago in nebular phase spectra of sne ia @xcite we are discovering the role of [ co ] in sne ia _ as early as bolometric maximum _ , and confirming its identification at later times . lately , a controversial association has been made with nad but this association is not supported by models .
one explanation for the mis - identification is that [ co ] lines are not included in the associated radiative transfer calculations @xcite , as was also done in our earlier modeling of sn ia .
when they are included , the line shows up @xcite .
another explanation is that sn radiative transfer is often done by separate codes for photospheric " phase and nebular phase studies , and photospheric codes typically neglect forbidden lines .
unfortunately , this creates a boundary between the two regimes that is artificial .
nebular lines are indeed seen at photospheric epochs , e.g. , the [ ca7300 doublet at the end of the plateau in sne ii - p @xcite or [ co]5888 early after bolometric maximum ; this work ) .
similarly , strong p - cygni profiles , typical of photospheric - phase spectra persist well into the nebular phase . in the future
, we will investigate the radiative properties of sne ia as they turn nebular .
as shown here , our chandrasekhar - mass delayed - detonation models exhibit a large range of ionization whereas observations appear somewhat degenerate in that respect .
our simulations differ in mass but have the same ejecta mass of 1.4 .
while the mass determines the peak luminosity , the ratio of to ejecta mass is a key ingredient controlling the ionization state of the gas .
hence , we will investigate whether a range of ejecta masses , tied to a narrow range of to ejecta mass ratio , can reduce the ionization disparity of our models after maximum .
ld and sb acknowledge financial support from the european community through an international re - integration grant , under grant number pirg04-ga-2008 - 239184 , and from agence nationale de la recherche " grant anr-2011-blanc - simi-5 - 6 - 007 - 01 . djh acknowledges support from stsci theory grant hst - ar-12640.01 , and nasa theory grant nnx10ac80 g . this work was also supported in part by the national science foundation under grant no .
phys-1066293 and benefited from the hospitality of the aspen center for physics .
ak acknowledges the nsf support through the nsf grants ast-0709181 and tg - ast090074 .
this work was granted access to the hpc resources of cines under the allocation c2013046608 made by genci ( grand equipement national de calcul intensif ) . | the high metal content and fast expansion of supernova ( sn ) ia ejecta leads to considerable line overlap in their optical spectra .
uncertainties in composition and ionization further complicate the process of line identification . in this paper
, we focus on the 5900 emission feature seen in sn ia spectra after bolometric maximum , a line which in the last two decades has been associated with [ co]5888 or nad . using non - lte time - dependent radiative - transfer calculations based on chandrasekhar - mass delayed - detonation models , we find that nad line emission is extremely weak at all post - maximum epochs .
instead , we predict the presence of [ co]5888 after maximum in all our sn ia models , which cover a range from 0.12 to 0.87 of .
we also find that the [ co]5888 forbidden line is present within days of bolometric maximum , and strengthens steadily for weeks thereafter .
both predictions are confirmed by observations . rather than trivial taxonomy
, these findings confirm that it is necessary to include forbidden - line transitions in radiative - transfer simulations of sne ia , both to obtain the correct ejecta cooling rate and to match observed optical spectra .
[ firstpage ] radiative transfer supernovae : general supernovae : individual : 2005cf stars : white dwarfs |
core - collapse supernovae ( ccsne ) are inherently asymmetric events , due to the asymmetric nature of the explosion mechanism . probing the shapes of these events can provide , therefore , important clues as to the nature of the explosion mechanism and its influence on the standard photometric and spectroscopic observables used to measure their properties . + thomson scattering is a polarizing process , with the polarization of the photon aligned in the direction orthogonal to the plane of scattering ( i.e. the plane defined by the ingoing and outgoing photon directions and the point of scattering ; @xcite ) . in sne
the principal site for this process is at the photosphere , where the opacity is dominated by free electrons . for a resolved spherical
photosphere the most significant polarization would be seen to arise from the limb , where only a narrow range of scattering angles ( for photons exiting the photosphere and being scattered into the line of sight ) are permissible . for an unresolved spherical photosphere ,
the polarized components arising from regions separated by @xmath5 further around the photosphere are equal but opposite , cancelling out and leading to a zero net observed polarization . in the presence of asymmetries
, there is incomplete cancellation of the polarization components arising from different points in the photosphere , leading to a net observed polarization with the degree of polarization related to the magnitude of the asymmetry and the polarization angle corresponding to the orientation of the asymmetry on the sky ( rotated by @xmath5 ) .
this produces a wavelength-_independent _ continuum polarization .
+ the interaction of continuum photons with the line forming region also produces a wavelength-_dependent _ polarization .
although resonant scattering is inherently depolarizing , absorption along the line of sight removes blue - shifted photons ( leading to the standard absorption component of p cygni profiles ) .
if the line forming region does not equally cover the photosphere , due to some large scale structure or clumping , then unequal contributions of the polarization components arising from the photosphere will be removed , leading to an increase in polarization across the absorption and the so - called _ inverted p - cygni profiles _ in the polarization spectrum @xcite .
the properties of the polarization at the absorption line are related to the distribution of the line forming element about the photosphere .
because photons scattered into the line of sight by resonant scattering are depolarized , the red - shifted emission components are inherently depolarized .
polarimetry provides , therefore , a unique and powerful way of directly measuring the shapes of sne at early , optically thick times .
+ previous polarimetric studies of ccsne have shown that the continua of type ibc sn are generally more polarized at early times than type ii sne .
in addition , the increase in polarization across the absorption features of p cygni profiles , coupled with the rotation of the polarization , have demonstrated that the geometries of ccsne are not described by just a single global axial symmetry @xcite .
+ here we report two epochs of vlt spectropolarimetry of the type ib sn , associated with the x - ray flash ( xrf ) 080109 .
xrf 080109 was discovered on 2008 jan 10 @xcite , in the galaxy ngc 2770 , a normal sa galaxy @xcite .
the location of xrf 080109/sn 2008d , relative to its host galaxy , is shown as fig .
[ fig : intro : posisp ] .
the xrf was followed by an ultra - violet transient .
an associated sn coincident with the xrf was identified ( 2008d ; @xcite ) and was spectroscopically confirmed to be of type ib @xcite .
@xcite determined that the xrf was non - relativistic , unlike xrfs associated with the gamma - ray burst phenomenon , rather arising from the shock breakout arising from a sn explosion .
the properties of the xrf suggested it arose from breakout at a radius consistent with a wolf - rayet ( wr ) progenitor star , into a dense wind ( @xmath6 ) . @xcite and @xcite modelled the explosion as spherically symmetric , but both found that sn 2008d was only slightly more energetic than normal type ibc sne ( @xmath7 ) and less energetic than so - called hypernovae ( e.g. sn 2002ap ) .
@xcite found , however , that poor fits of synthetic spectra to observed line profiles indicated the role of significant asymmetries .
here we directly establish those asymmetries by means of spectropolarimetry .
+ the paper is organised as follows : in section [ sec : obs ] we describe our spectropolarimetric observations of sn 2008d .
the results of these observations are presented in [
sec : results ] , and these results are analysed in
[ sec : analysis ] .
the results and analysis are discussed in [ sec : discussion ] and we present our conclusions in [ sec : conclusions ] .
spectropolarimetry of sn 2008d was conducted at two epochs , 2008 jan 31 and 2008 feb 15 ( corresponding to approximately v - band light - curve maximum and 15 days after maximum , respectively ; @xcite ) , using the eso vlt fors1 instrument in the pmos mode @xcite . a journal of these observations is shown as table [ tab : obs : log ] .
all observations were conducted with the 300v grism , providing a wavelength range of @xmath8 , and with no order separation filter used .
observations of sn 2008d , at both epochs , consisted of two sets of four exposures at each of the four half - wavelength retarder plate angles . in addition , at each epoch , a flux standard was observed with the full polarimetry optics in place at a single retarder plate angle ( @xmath9 ) .
the observations were reduced using iraf and our own specially written software , following the scheme presented by @xcite .
+ the spectropolarimetric observations at each epoch were combined using a weighted averaging scheme , with the stokes parameters weighted by the flux . due to its relatively high declination ,
sn 2008d was only observed at low altitude ( @xmath10 ) by the vlt .
the fors1 instrument uses a mosaic of two thinned , backside - illuminated 2k@xmath114k e2v ccd44 - 82 detectors , with increased sensitivity in the blue , but severe fringing and second - order contamination at @xmath12 ( c.f . spectropolarimetry with the previous tektronix ccd detector ; e.g. @xcite ) .
the nature of the observations , of a highly reddened object at high - airmass with the slit oriented at @xmath13 rather than the parallactic angle , was beneficial in that it significantly reduced the flux in the blue ( see figs .
[ fig : obs : epoch1 ] and [ fig : obs : epoch2 ] ) and , hence , minimised order contamination of blue flux in the red ( and hence contamination of the resulting measured polarization ) .
since polarimetry uses a relative measurement of flux , the severe reddening of the spectrum has no effect on the final measured degree of polarization . + in order to assess the role of any instrumental changes , including evolution of the fringing pattern ,
the instrumental signature corrections ( @xmath14 and @xmath15 ; ) were compared for each pair of datasets and found to be consistent to within the measurement error of @xmath16 .
+ the spectra were flux calibrated , with respect to the observed flux standard , but were _ not _ corrected for atmospheric extinction ( such that the flux spectra presented here are significantly redder than presented elsewhere ) .
the calibration of the pmos mode of fors1 was checked using observations of the polarized standard star vela 1 95 .
the data were corrected for the heliocentric recessional velocity of the host galaxy of @xmath17 @xmath18 .
+ synthetic broad v - band polarimetry was measured from the data , weighting the observed spectropolarimetry as a function of wavelength with the johnson v passband response function .
the observed spectropolarimetry of sn 2008d at 31 jan 2008 and 15 feb 2008 is presented as figs . [
fig : obs : epoch1 ] and [ fig : obs : epoch2 ] .
we summarise the spectroscopic properties of sn 2008d at the two epochs of our observations , to serve as an orientation for the discussion of the polarimetric properties of this object .
previously , spectra of sn 2008d have been presented by @xcite , @xcite , @xcite , @xcite and @xcite .
here we adopt the line identifications made by @xcite and @xcite .
+ at the two epochs of our observations the spectrum of sn 2008d is composed of broad p cygni profiles of , , and .
+ h&k and the ir triplet ( ir3 ) are present at the blue and red extremes of the spectra . at the first epoch ,
the full p cygni profile of h&k is observed , with the absorption minimum at -14100@xmath18 .
in addition , weak narrow absorptions are also observed superposed on the emission feature , at the rest wavelength in the host galaxy , arising in the interstellar medium . at the second epoch ,
the absorption component is no longer detected , due to the spectrum becoming redder and steeper , such that only the emission component and the decline towards the minimum are observed .
the ir3 feature is detected at both epochs . at the first epoch ,
the emission component is truncated in the red by an line @xcite .
the absorption component , corresponding to a velocity of @xmath19 @xmath18 , is a broad single - minimum feature . at the second epoch
there is little evidence of the redward absorption due to , but the absorption minimum of ir3 is separated into seemingly high - velocity ( hv ) and low - velocity minima at @xmath20 @xmath18and @xmath21 @xmath18 , respectively ( but note that the hv component is moving at lower velocities than the single component observed at the first epoch ) .
+ similar to the spectrum of sn 2005bf , ( 42 ) is observed as separate resolved absorption minima at the first epoch ( @xmath22 @xmath18 ) ; at the second epoch , due to a general decrease in observed velocities the features are blended into a single profile , with a likely significant contribution from @xmath23 , commensurate with the observed increase in the strength of other lines elsewhere in the spectrum . a combined @xmath24 and ( 37,38 ) feature
is observed at @xmath25 .
+ at the first epoch a set of lines ( @xmath26 ) are observed in the spectra , with the two redder lines being observed as slight notches ( corresponding to @xmath27 @xmath18 ) .
the stronger @xmath28 line may also be blended with , which may explain the apparent disparity between line strengths in the series . at the second epoch the lines have significantly increased in strength , although with a slight decrease in velocity to @xmath29 @xmath18 .
a narrow absorption is observed at the rest wavelength of d at the host galaxy .
+ at the first epoch , @xmath30 is observed as a shallow p cygni profile , that is potentially blended with a high - velocity component of @xmath31 @xcite . assuming this feature is purely due to the absorption minimum corresponds to @xmath32 @xmath18 .
this feature is absent from the data of the second epoch .
there is a hint of @xmath4 in emission , with the corresponding absorption trough coincident with the nearby telluric feature ( suggesting an approximate velocity @xmath33 @xmath18at the first epoch , and @xmath34 @xmath18at the second ) . @xcite
observe the full p cygni profile of the line in their spectroscopic data , at 2 days prior to our first observation . + the galactic foreground reddening to sn 2008d is given as @xmath35 @xcite .
the equivalent width of d measured from our low resolution data is a poor indicator of reddening arising in the host , as previous high - resolution studies of sn 2008d have shown it to be a superposition of a number of different , saturated absorption systems @xcite .
previous studies have determined the reddening to fall in the range @xmath36 @xcite . here
, we adopt the value @xmath37 , the most commonly used value , and assume the foreground reddening is negligible .
the spectropolarimetric data are plotted on the stokes @xmath38 plane for both epochs on figs .
[ fig : obs : epoch1qu ] and [ fig : obs : epoch2qu ] .
sn 2008d is observed to be significantly polarized at both epochs , with a baseline polarization of @xmath39 .
as shown on figs .
[ fig : obs : epoch1 ] and [ fig : obs : epoch2 ] this level of polarization is common at both epochs .
most of this polarization is in the @xmath40 stokes parameter , as evident in figs .
[ fig : obs : epoch1qu ] and [ fig : obs : epoch2qu ] , as the data points are clustered in the same location of the stokes @xmath38 plane and the polarization angle is @xmath41 across most of the spectrum .
the apparent rise in the polarization at the blue and red extremes of the data is consistent with an increase in the uncertainties on the measured polarization , due to declining signal - to - noise in these portions of the spectrum and the possible wavelength dependence of the the interstellar polarization ( isp ) .
+ the most significant deviations from this baseline level of polarization are associated with spectral features . at the first epoch a peak polarization of @xmath42 is observed at the h&k absorption component .
this is contrasted for the ir3 absorption , for which the absorption minimum is associated with a minimum ( 0.6% ) in the polarization spectrum . on the @xmath38 plane ,
the data corresponding to the ir3 absorption is observed to lie distinctly separate from the rest of the data .
significant polarization is associated with the @xmath28 line , and the redder he lines are also polarized albeit at a lower degree . with the increase in the strength of the lines at the second epoch , there is a correlated increase in the polarization .
the resolved absorption components of the ( 42 ) multiplet and the @xmath24 line , at the first epoch , are associated with strong polarization at the level of @xmath43 .
+ given the substantial degree of reddening associated with this sn ( see [ sec : results : spec ] ) , there is the expectation that the isp component is non - zero .
as has been demonstrated in a number of cases , the determination and subtraction of the isp component is vital to determining the intrinsic polarization ( @xmath44 ) and , hence , geometry of sne .
importantly , the isp can cause intrinsically unpolarized features to be observed to be polarized and make polarized features appear depolarized .
the standard expectation for p cygni profiles , in the flux spectrum , is that significant polarization is associated with the absorption component while the emission component is unpolarized .
we see that , at the first epoch , the absorption component of ir3 is unpolarized .
furthermore , we note that the general behaviour of the data on the stokes @xmath38 plane is constant with time , occupying the same location ( despite variability in the polarization associated with spectral lines ) . + synthetic v - band polarimetry of this data yielded @xmath45 ( @xmath46 ) and @xmath47 ( @xmath48 ) , at 31 jan 2008 and 15 feb 2008 respectively .
@xcite present broad v - band polarimetric measurements which , not only agreeing with our synthetic v - band polarimetry , also show that sn 2008d occupies the same location of the @xmath38 plane prior to and following our two observations ( see [ sec : discussion ] ) .
the determination of the isp is crucial to determining the intrinsic polarization of the sn .
the reddening towards sn 2008d can be used to provide a constraint on the maximum degree of isp .
the total reddening to sn 2008d ( see [ sec : results : spec ] ) constrains the isp to be @xmath49 ( assuming a standard serkowski - galactic type isp ) , arising predominantly in the host galaxy . in order to make a precise estimate of the isp component
particular assumptions must be used , principally based on there being portions of the spectrum that are intrinsically unpolarized , such that the observed polarization of these regions is due to the isp alone .
we attempt to remove only a single total isp , rather than account for individual isps arising in the galaxy and the host galaxy , as the total isp is a vector sum of all individual components .
a number of techniques , based on this primary assumption , have been used and we discuss in turn the alternative values of the isp indicated .
+ under the assumption that particular wavelength regions of the spectrum are intrinsically depolarized due to the blending of numerous overlapping fe lines in the range 4800 - 5600 , @xcite used the observed polarization properties at particular wavelengths as measures of the isp .
the multiplet 42 lines are clearly resolved at both epochs @xcite , however immediately redward in the range 5200 - 5500 there are a number of overlapping features . under the assumption that the line blanketing opacity dominates over electron scattering in this wavelength range , such that this region is intrinsically depolarized , we measure the median average values of the stokes parameters as @xmath50 and @xmath51 at the first epoch and @xmath52 and @xmath53 at the second epoch .
these values of the isp stokes parameters are identical , within the uncertainties , and the average value corresponds to @xmath54 and @xmath55 .
+ an additional measurement can be derived using the technique of @xcite , under the assumption that the emission component of a p cygni profile is intrinsically unpolarized .
the polarization of the emission component , corrected for the polarization of the continuum ( actually a mixture of the intrinsic continuum polarization and the isp ) , reflects the isp component alone and can be measured by a least - squares fit to the equations @xmath56 and @xmath57 to determine the stokes parameters of the emission line @xmath58 and @xmath59 ( in the presence of no isp @xmath60 ) .
this technique was applied to the emission components of the @xmath614470 / and ir3 lines at the second epoch , where a good approximation for the continuum level could be made . at the first epoch ,
the emission components of these lines are truncated by blends with other features , whereas at the second epoch these features appear unblended .
the isp component was measured for both lines to be @xmath0 and @xmath62 ( where the quoted error is the scatter in fits using different wavelength ranges and different continuum levels ) .
+ the fact that the isp is time - invariant supports these values , given the apparent constant level of polarization and the constant location of the data on the @xmath38 plane at the two observational epochs .
@xcite observed similar stokes parameters with their broad - band polarimetry , between 3.6 and 78.5 days post - explosion , suggestive that the bulk of the observed polarization ( excluding polarization associated with strong lines ) was due to a unvarying isp .
+ in general , the isp is expected to be wavelength - dependent , but characterising this dependence is difficult for objects such as sn 2008d which show strong intrinsic wavelength - dependent polarization associated with spectral lines . for a serkowski - type isp , for milky way - like dust ,
the difference in isp between the peak polarization , at a characteristic wavelength of @xmath63 , and the isp polarization measured for the emission components of the lines at the blue and red ends of the spectrum is smaller than the error of the measured stokes parameters of the isp .
the small degree of wavelength dependence justifies the use of a single - valued , wavelength independent isp as inferred from this data . for a number of sne , the wavelength - dependence of the isp , arising from their host galaxies ,
has been observed to deviate significantly from a standard @xcite - type isp law @xcite .
in the case of sn 2008d , there is no evidence for the magnitude of the polarization rising significantly at either the blue or red extremes of the observed spectrum ( suggesting the peak in the isp occurs within the wavelength range of these observations ) .
in fact , the baseline polarization of the data is approximately constant across the observed wavelength range , with a possible peak around @xmath63 ( see above ) , which lends further credence to the to the assumption that the isp can be considered approximately wavelength independent for our observations ( with any wavelength dependence of the isp over the wavelength range being smaller than the uncertainties of our wavelength independent estimate ) .
+ the weighted average of the measured stokes parameters for the isp , from the different methods , are @xmath64 and @xmath65 , which correspond to @xmath66 and @xmath67 .
the direction of the isp vector is shown on fig .
[ fig : intro : posisp ] .
contrary to the standard prediction that the isp vector be aligned with the spiral arms of the host galaxy , due to the alignment of dust grains by magnetic field lines parallel to the spiral arm @xcite , the isp is not aligned with the nearest spiral arm .
this may indicate that the isp determined may either be incorrect , that the magnetic field in the host is not aligned with the spiral arms at large radii or that the isp is the sum of a number of separate components , perhaps arising in the galaxy , the host galaxy , the intervening medium and , importantly , in the circumstellar environment of the sn .
+ after correcting the observed data for the isp , we note that away from obvious spectral features the intrinsic polarization is generally @xmath2 .
furthermore , at both epochs the wavelength region @xmath68 appears to have a relatively constant low polarization , with only weak features appearing in the flux spectrum . using this region as representative of the continuum
, we measure the intrinsic polarization to be @xmath69 and @xmath70 at the first and second epochs respectively ( where these values are weighted means over the data in this wavelength range ) . assuming a spheroidal photosphere
, this polarization corresponds to an apparent axial ratio @xmath71 .
significant intrinsic polarization is associated with the lines at both epochs . at the second epoch , higher levels of polarization
are measured for the redder lines , in line with the absorption becoming stronger ( i.e. the fraction of continuum photons being removed by the line is larger at the second epoch ) . if the line forming region , for a general line,1 blocks all the unpolarized light from a spheroidal photosphere , then the maximum degree of polarization expected at absorption minimum ( @xmath73 ) is : @xmath74 following the prescription of @xcite , where @xmath75 is the degree of polarization measured in adjacent `` continuum '' regions and @xmath76 and @xmath77 are the fluxes measured at the nearby continuum and at the absorption component minimum , respectively .
there is some difficulty with the measurement of the maximum polarization of @xmath28 in the intrinsic polarization spectrum , as there is variability at and around the position corresponding to the absorption line minimum in the flux spectrum .
we note , however , that the redder lines show the expected inverted p - cygni profile in the polarization spectrum . at the second epoch , for @xmath72
we estimate theoretical polarization limits of @xmath78 , @xmath79 and @xmath80 , respectively , as opposed to the observed values of @xmath81 , @xmath82 , and @xmath83 .
@xcite , @xcite and @xcite present a number of reasons why this theoretical limit may not be actually observed : depolarization of resonantly scattered continuum photons in optically thick lines , the contribution of strong emission line components or clumpy ejecta .
+ a significant clue as to the true reason behind the significant polarization is observed in the rotation of the polarization angle across the @xmath28 line ( and to a lesser extent for @xmath84 ) .
this rotation corresponds to a velocity - dependent loop structure , which is shown as fig .
[ fig : obs : heline ] .
such a loop feature is generally interpreted as a departure from a single global axial symmetry , such that the axis of symmetry of the line forming region is not aligned with that of the photosphere and the line forming region also only covers a fraction of the photosphere @xcite .
the similarity of the polarization angle of three lines , @xmath85 , @xmath86 and @xmath87 , at the first epoch suggest a common geometry for the line forming regions of these lines .
the two red lines have similar polarization angles at the second epoch , @xmath88 and @xmath89 , suggesting the line forming region is the same as in the first epoch . the polarization angle associated with @xmath28 changes to @xmath90 , suggesting possible blending due to d. + at the second epoch , @xmath23 is also expected to be a prominent feature , although coincident with a number of lines .
the presence of this line is suggested by the similar polarization angle ( @xmath91 ) associated with the absorption at 4890 .
+ we conclude that the violation of eq .
[ eq : res : leonard ] and the observed rotation of the polarization angle for the lines is due to the lines arising in an excited clump that is off - axis relative to the axis of symmetry of the photosphere and does not cover the full area of the photosphere .
[ [ sec : analysis : intrinsic : caii ] ] at the first epoch , h&k is observed to be polarized ( with an inverted p - cygni profile ) at @xmath92 , a similar level to that observed for the ir3 at @xmath93 .
the ir3 absorption feature is observed to be polarized at both epochs , with the subtraction of the isp , producing the expected `` inverted p - cygni profiles '' in the polarization spectra .
as it is unclear if the flux at the absorption minimum of the h&k feature is significantly detected at the second epoch ( [ sec : results : spec ] ) , we can not ascertain the real degree or significance of the observed polarization around the wavelength of that feature at that epoch . at the second epoch , in which a probable hv component is observed , a polarization of ir3
is observed to be @xmath94 .
this suggests the apparent hv component at the second epoch also produces the absorption minimum at the first epoch , and the two components were not resolved at the first epoch . similarly to [ sec : analysis : intrinsic : hei ] , the theoretical limiting polarization for is @xmath95 , such that the observed data suggests a similar scenario for ( as suggested for ) being formed in a region with a different axial symmetry to that of the photosphere and incomplete coverage . at the first epoch
the polarization angle at the ir3 absorption minimum is @xmath96 , whereas at the second epoch the angle is @xmath97 .
importantly , these angles are significantly different than those observed for , and , suggesting the ir3 line forms in a significantly different portion of the ejecta to the lines of the other species .
+ at the first epoch , the absorption component of ir3 is obviously polarized , although the data is not of sufficient quality ( due to the severity of fringes ) to discern a loop ( see fig .
[ fig : obs : caloop1 ] ) . by the time of the second epoch , a loop
is easily identified ( fig .
[ fig : obs : caloop2 ] ) corresponding to the hv component of the absorption profile .
[ [ sec : analysis : intrinsic : feii ] ] the observed level of polarization associated with the ( 42 ) lines is significantly reduced upon subtraction of the isp , as indicated in figs .
[ fig : obs : epoch1 ] and [ fig : obs : epoch2 ] .
some polarization is still associated with these lines at the first epoch ( @xmath98 ) , but a larger polarization ( @xmath99 ) is observed at this wavelength at the second epoch , due to the increase in strength of the @xmath23 line ( see [ sec : analysis : intrinsic : hei ] ) .
+ at first epoch , the peak polarization for the bluest absorption of ( 37,38 ) , at @xmath100@xmath18 , is 0.9% .
another polarization peak at 1% is observed at @xmath101 , without an obvious counterpart in the flux spectrum .
it may correspond to the @xmath102 line , in the same multiplet , at the velocity determined above .
[ [ sec : analysis : intrinsic : oi ] ] a weak polarization signature is associated with @xmath103 measured to be @xmath104 at @xmath105 and @xmath106 at @xmath107 , at the first and second epochs respectively .
the large errors are due to the coincident telluric feature , but still imply the polarization of is not as large as that observed for and .
[ [ sec : analysis : intrinsic : siii ] ] at the first epoch the feature is polarized , although with a peculiar polarization signature tracing the shallow line profile in the flux spectrum . at the absorption minimum ( @xmath32@xmath18 )
the corresponding polarization is @xmath108 .
the maximum in polarization ( @xmath109 ) actually occurs significantly redward of the minimum at 6260(@xmath110@xmath18 ) .
following @xcite and @xcite ( [ sec : results : spec ] ) , this feature may be a blend of and a hv component of @xmath31 .
if the polarization maximum is due to @xmath31 , rather than , it would imply that the absorption minimum of the @xmath31 hv component has a velocity of @xmath111@xmath18 .
a similar shallow absorption profile was observed for sn 2005bf , although significant polarization was detected at the absorption minimum corresponding to @xmath112 rather than @xcite .
+ as discussed in
[ sec : results : spec ] , the feature is no longer apparent in the flux spectrum of the second epoch .
in addition , we note that there is no significant signature in the polarization spectrum in the wavelength range where is expected to lie , such that the feature has not been blended with another feature and instead it has disappeared at that time . + following the scheme of @xcite , the dominant axis of the data on the stokes @xmath38 plane
was determined by the weighted least - squares fit of a straight line to the data at both epochs .
the angle @xmath113 , between the polarization angle of the dominant axis and the stokes @xmath114 axis were determined to be @xmath115 and @xmath116 at the first and second epochs , respectively .
the dominant ( @xmath117 ) and orthogonal ( @xmath118 ) stokes parameters were determined by rotating the data about the origin of the stokes @xmath38 plane by an angle @xmath119 , where : @xmath120 the determination of the polarization along and perpendicular to the dominant axis provides information on the principal asymmetry of the sn , and deviations from it ( e.g. by the line forming regions of particular species in the spectrum ) and removes the effects of the position angle at which the sn was observed ( although certain inclination effects remain ) .
the spectropolarimetry projected into the dominant and orthogonal stokes parameters is shown as figs .
[ fig : analysis : epoch1 ] and [ fig : analysis : epoch2 ] . in the case of sn
2008d , like sn 2002ap @xcite , at both epochs the dominant axes were found to not pass through the origin of the stokes @xmath38 plane , such that the presence of continuum polarization is observed as an offset in the orthogonal polarization component .
+ at the first epoch , /@xmath31 and lines are predominantly polarized along the dominant axis , whereas h&k and ir3 and the have components in both the dominant and orthogonal directions .
a strong orthogonal polarization is observed close to the expected position of the @xmath4 absorption .
+ at the second epoch the situation is reversed , with the lines being stronger in the orthogonal component . despite the apparent decrease in the strength of the absorptions , these features are still present in the dominant component , although the strong feature observed in @xmath118 is due to @xmath23 .
+ caution is encouraged in interpreting the data on the dominant and orthogonal axes . the apparent shift of from the dominant axis ( first epoch ) to the orthogonal axis ( second epoch ) is primarily due to the apparent change in angle of the dominant axes ( @xmath121 ) between the two epochs , due to the evolving flux and polarization of lines in the spectrum between the two epoch ; as opposed to a bulk change in the polarization properties of the sn between the two epochs ( such as observed for sn 2001ig ; @xcite ) .
examination of the data on the stokes @xmath38 plane in figs .
[ fig : obs : epoch1qu ] and [ fig : obs : epoch2qu ] shows the data to be clustered in the same location at the two epochs . + as discussed in
[ sec : analysis : geometry ] , the relative polarization of different species in the dominant and orthogonal directions indicates that , on the sky , the line forming elements have very different distributions relative to eachother within the ejecta .
the true power of spectropolarimetry , to study the shapes of sne , is its ability to provide information of the geometry in the plane of the sky .
comparison of the degree of polarization and polarization angle against the velocities observed in the flux spectrum permits a three dimensional tomographic analysis indicating the distribution of various elements within the ejecta in addition to the shape of the photosphere .
+ the polarization angles and velocities at absorption minimum for important lines in the spectrum are presented for both epochs in fig .
[ fig : analysis : polar ] .
the polarization angle measured at the absorption minimum corresponds to the polarized component of continuum light that is _ not _ absorbed by the line forming region , hence the approximate distributions of the various line forming regions on the sky are rotated @xmath5 to the positions shown in fig .
[ fig : analysis : polar ] .
+ the observed polarization angle will be the result of mixing of polarized components from the continuum ( if the photosphere is asymmetric ) and the line s effective polarization ( due to the shadow it casts on the photosphere ) . in the case of sn
2008d we have established , however , that the observed line polarization is significantly larger than the continuum polarization , and will dominate the measured polarization angle . + as shown in fig .
[ fig : analysis : polar ] , at the first epoch and are found in similar parts of the ejecta , with similar velocities ( @xmath122@xmath18 ) , while the @xmath30 line is formed in a similar location , but at lower velocities .
both the h&k and ir3 lines have the same polarization angle , but are offset from the lines by @xmath123 ( or @xmath124 ) .
this implies that there are , at least , two distinct ejecta components and that the lines , at much higher velocities , are disjoint from the rest of the ejecta . + at the second epoch , all velocities have decreased and and most of the lines are observed at approximately the same position angles as at the first epoch .
the feature at @xmath28 is separate from the other lines , with a polarization angle more consistent with . at this epoch
, the @xmath23 line is visible , and is grouped with the higher wavelength lines at @xmath125 .
this suggests that the feature associated with @xmath28 is actually a blend with d @xcite .
( 42 ) is observed at an intermediate polarization angle .
+ it is unclear at the first epoch , given the large degree of uncertainty , whether the line forming region occupies a similar space to the line forming region or if it is consistent with the region ( due to the fact that a full rotation of the stokes parameters occurs over only @xmath126 ) .
there is significant uncertainty in both the degree of polarization and polarization angle associated with due to coincidence with a strong telluric feature . as polarimetry measures differences in flux ,
the telluric absorption ( assumed to be unpolarized ) should not have any effect on the measured polarization , except for a reduction in the signal - to - noise ratio .
following the classification scheme of , sn 2008d / xrf 080109 is classified as being of type d1+l()+l ( ) .
this classification reflects the fact that a dominant axis can be fit to the data , although the polarization is better described as an ellipse on the stokes @xmath38 plane . furthermore , distinct loop features have been identified for and .
sn 2008d shows a number of similarities and differences with the spectropolarimetry of other ccsne , which can be used to compare and contrast likely geometries of these events .
the intrinsic level of polarization of the continuum , dependent on the shape of the photosphere , is significantly detected but low , suggesting asymmetries of @xmath127 .
similar low levels of continuum polarization have been observed over the range of type ibc sne , including 2002ap , 2005bf , 2006aj and 2007gr .
the starkest contrast between these events is , however , the polarization associated with specific spectral lines . + for sn 2002ap , significant polarization associated with the @xmath128 ( 0.9% ; @xcite ) and ir3 @xcite absorption components was observed . in that case , however , the progenitor star was a non - degenerate c / o star without he , whereas for sn 2008d the presence of a strong feature specifically implies the explosion of he - rich star @xcite .
@xcite interpreted the lack of polarization associated with in spectropolarimetry of the type ib sn 2005bf as indicating `` shielding '' of the o - rich core by the he envelope . on the other hand , 2007gr was a type ic sn , for which no he layer expected , which also showed little polarization associated with @xcite .
+ @xcite argue against using sn 2005bf as the comparison for sn 2008d on photometric and spectroscopic grounds , due to the double peaked light curve and the appearance of in the spectrum at 6 days prior to the second light curve maximum ( @xmath129 days post - explosion ) .
we note that the polarization properties of sn 2008d and sn 2005bf are actually quite similar .
the loop feature for the @xmath28 line for sn 2008d at the second epoch is similar to that observed for sn 2005bf @xcite ; except in this instance the signal - to - noise ratio is lower , making the loop less clear in fig .
[ fig : obs : heline ] .
further similarity is observed for the resolved lines of in the flux spectrum , the low level of line polarization and the presence of and hv @xmath31 absorption features .
importantly , lines of both sn 2008d and 2005bf are observed to have minima at substantially higher velocities than , and ( see fig .
[ fig : discussion:05bf ] ) and are significantly polarized .
this suggests that , at least for the second light curve maximum , sn 2005bf is similar to sn 2008d at light curve maximum for the purposes of comparing their polarimetry and , ultimately , the natures of their explosions .
+ hv ca components have proven quite common in type ia sne ( sn 2001el - @xcite ; sn 2006x - @xcite ; @xcite ) ; here we note that they also seem rather common in snib / c as well .
spectropolarimetry has shown , however , that these features are highly polarized and that the polarization angles of these features are substantially different from the rest of ejecta .
our spectropolarimetry of sn 2008d shows clear evidence that the ejecta are _ chemically segregated_. + fig . [
fig : analysis : polar ] shows that the h&k and ir3 lines are formed at higher velocities than all the other spectral features studied in the spectrum .
the lines all have approximately similar polarization angles at both epochs , except at the second epoch where one feature ( @xmath28 ) is observed to have a polarization close to that of .
+ the observed maximum polarizations for and , which exceed the theoretical limit for simple blocking of the unpolarized light , implies that the line forming regions for these species are not distributed evenly across the photosphere .
the hv components of the lines at both epochs , along with the significantly different polarization angle to the rest of the constituents of the ejecta , suggest the may be formed in a clump with a significant projected velocity along the line - of - sight . for sn 2007gr
, @xcite inferred that the hv calcium was constrained in a bipolar distribution , with the other elements such as forming a torus in the orthogonal plane ( due to the different polarization angles ) .
such a bipolar distribution of , for these type ibc sne , may be related to the action of a jet - like flow .
the low continuum polarization may indicate that the ejecta are viewed along , or close to , a principal axis of symmetry for the ejecta .
+ the presence of loop structures has been observed for nearly all ccsne @xcite and some type ia sne .
these features have been interpreted by @xcite and @xcite as being due to the line - forming region , of the specific species , having an axial symmetry inclined with respect to the axial symmetry of the photosphere .
this implies that there is no single global axial symmetry , and rather that the geometries of the ejecta are different for clumps of different species .
if a single axial symmetry were present , then the data would be well described by a single dominant axis , and for sn 2008d it has been shown ( see [ sec : analysis : decomp ] and figs . [
fig : analysis : epoch1 ] and [ fig : analysis : epoch2 ] ) to not be the case .
+ the low level of continuum polarization determined here suggests the deviation of the photosphere from a spherical symmetry is low .
this implies that the photosphere is receding through the ejecta at an approximately isotropic rate , which further suggests that the photosphere did not cross a significant change in the excitation structure between the two epochs of spectropolarimetry . in the cases of sne 2004dj @xcite , sn 2001ig @xcite and 2001dh @xcite significant changes in the degree of polarization and the polarization angle with time implied depth dependent geometry discontinuities . in the case of type iip and type iib sne
, such a significant change in polarization is expected once the photosphere crosses the chemical boundary between the h and he layers , where the photosphere can recede more quickly since is less ionized and , hence , presents less opacity than h. the geometry of is expected , therefore , to more closely match the underlying ni distribution due to the asymmetric apparent luminosity of the photosphere @xcite .
another such discontinuity is expected when the core products of nucleosynthesis and explosive burning are revealed , for which hydrodynamic simulations predict complex three - dimensional structures @xcite .
+ @xcite observed small changes in the broad - band polarization properties of sn 2008d over a larger time frame , from 2008 jan 13 to 2008 mar 28 , which are consistent with our synthetic v - band polarimetry ( see [ sec : results : pol ] ) .
while the true stability of the polarimetric standard adopted by @xcite , itself a sn , is debatable , the uncorrected polarimetry of sn 2008d shows only small variability .
this implies that a significant geometry discontinuity was not just unobserved due to the paucity of our own observations .
our spectropolarimetry suggests , however , that while some of the variability may truly arise in the continuum , an additional factor that may contribute to the measured broad - band polarization is evolving line polarization .
the most important feature possibly affecting the v - band polarization is the / feature , which is stronger , narrower and more polarized at the second epoch compared to the first .
the striking absence of a discrete geometry discontinuity suggests that any axisymmetric structure directly associated with the explosion mechanism is not observed at the times of our observations or those of @xcite , and rather that the continuum polarization arises from only an elongation of the core , similar to that observed at early times for sn 2002ap @xcite .
+ the apparent geometry of sn 2008d , at the two epochs , suggests the possibility of a jet induced explosion . in the jet - torus explosion model , a jet - like flow propagates through the progenitor
such a flow may originate from an asymmetry of neutrino deposition in the ejecta , the standing accretion shock instability @xcite or a magneto - rotational instability @xcite .
hv material is expected to arise from material impacted by the jet , while lateral shocks compress the main portion of the ejecta into a toroidal configuration @xcite .
+ the lack of a detected geometry discontinuity is suggestive that either a jet - like flow had already been revealed at the time of our first observation or that it still remained shielded inside the core at the time of our second observation . the properties of jets which can either punch through the ejecta or stall within the core have been studied previously , and shown to be dependent on the relative mix of thermal and kinetic energy in the jet @xcite .
we propose two models ( scenarios a and b ; schematic diagrams of these configurations are presented as fig .
[ fig : discussion : scenario ] ) , in the jet - torus paradigm , which may have given rise to the spectropolarimetric properties : the jet - like flow ( thermal energy dominated ) stalled within the core .
high - velocity is produced approximately along the direction of the jet - like flow , while the main portion of the ejecta exterior to the photosphere at both epochs ( , , and ) has a squashed spherical / toroidal geometry orthogonal to that of .
@xcite constructed a tilted jet model to explain the apparent lack of a single global axial symmetry for sn 2005bf .
similar to sn 2002ap @xcite , the low degree of asymmetry of the photosphere is taken as an indication that such jets had not reached the height defined by the photosphere as measured at the epochs of the two observations .
on the other hand the significant polarization associated with particular spectral features , with differing polarization angles for different species , is indicative of differing distributions of these elements within the ejecta , such that a single spherical symmetry is not appropriate .
that loops are also observed for some spectral features is perhaps supportive of sn 2008d having a similar configuration as sn 2005bf , and a tilted axis of excitation with respect to the axis of symmetry of the photosphere .
this may be evidence that there has not been full scale transport of material that would have led to homogeneity , and that the elements reside in different portions of the ejecta .
this scenario can be directly tested with observations at later epochs , when a large increase in the continuum polarization is expected as the asymmetric core - layers , containing jet material , are revealed .
+ _ scenario b : _ the jet - like flow ( kinetic energy dominated ) completely broke out of the progenitor prior to our first spectropolarimetric observation ( and those of @xcite , at 18 days prior to our first observation ) , such that no discontinuity would be apparent as in the case of the jet stalling within the core .
such a scenario was suggested for sn 2002ap , where a polarization peak in the vicinity of @xmath4/ ir3 was interpreted as being due to light reflected by electrons in a high - velocity ni - rich jet - like clump , exterior to the main ejecta , moving at @xmath130 @xcite .
@xcite rejected this picture because : 1 ) the polarization was associated with moving at @xmath131 and 2 ) no features due to ni or co ( in the range 1.6 - 1.8 @xmath132 ) were present in the flux spectrum at early times , as would be expected if these radioactive elements had been deposited outside the main ejecta . for sn
2008d we note similarly that @xcite did not detect ni or co features in their ir spectra nor do we detect the asymmetric excitation of the ejecta illuminated by exterior ni - rich clumps ( with an obviously bipolar distribution ) .
similarly , the low degree of polarization , suggests there was little / no transport of core - material to the outer layers of the ejecta by the action of a jet - like flow .
+ we conclude that a jet - like flow , if present , stalled inside the co core and , therefore , support scenario a. an important consequence of scenario a is , however , that the hv is _ not _ formed from products of nucleosynthesis from the progenitor , that was transported to the high - velocity ( large radius ) layer by a jet - like flow , rather it arose from primordial ca abundance of the progenitor .
+ the question remains , however , as to the nature of the control mechanism which dictates why some type ibc sne are associated with the grb phenomenon and , hence , with highly - energetic jets which penetrate to the exterior of the progenitor and others , such as sne 2008d and 2005bf show evidence for jets which may have only remained in the core @xcite .
the suggestion from spectropolarimetric observations that any jet - like flows did not completely breakout has important implications for the interpretation of the x - ray emission of the associated xrf 080109 .
+ in addition to our spectropolarimetry , @xcite report peculiar emission line profiles in late - time ( 91d after max . ) nebular spectra of sn 2008d , at an epoch when the sn is optically thin , which is claimed to trace the asymmetric ejecta ( see also * ? ? ?
we note that our early time spectropolarimetric observations of asymmetries in the distributions of a number of species within the ejecta are consistent with this picture ( for example ) . at such late times , however , the ejecta can be shaped by a number of competing factors : the interaction between the circumstellar medium ( csm ) and the ejecta and the evolution and interaction of physical processes ( such as shocks and instabilities ) over the lifetime of the sn .
importantly , at early times the measured geometry reflects the asymmetric nature of the explosion mechanism with little evolution from the moment of explosion . in this case
, the observed geometry is more closely related to the geometry from which the early x - ray emission originated than the late - time observations of @xcite may be .
+ there are a number of open issues regarding the early x - ray emission from sn 2008d : whether it is due to thermal emission of shock break out @xcite or not @xcite ; whether the emission is non - thermal ( @xcite ) , a single black body ( @xcite ) , or multiple black bodies ( @xcite ) ; whether there is a dense wind surrounding the progenitor ( @xcite ) or not ( @xcite ) ; whether non - thermal emission arises from a shock - mediated fermi acceleration ( @xcite ) or a variation of the forward / reverse shock paradigm of grbs ( @xcite ) .
the common feature of these analyses is that they assume , or conclude , that the associated processes are spherically symmetric , or nearly so .
our spectropolarimetry data show that this is not the case , certainly by nearly maximum light when the observations are made , and presumably not as the shock is erupting from the surface of the star and propagating into any circumstellar medium .
+ our observations do not directly constrain the geometry at breakout , but they give a strong caution that asymmetric shock breakout should be considered .
asymmetric shock breakout is unlikely to be accompanied by a single black body spectrum , so arguments that a single black body does not fit the x - ray spectrum are questionable .
we also note that the x - ray spectrum softens considerably over the course of the x - ray outburst ( @xcite , supplemental material ) , so using the mean x - ray spectrum to characterize the physical nature of the outburst requires caution .
arguments that the x - ray burst can not be a shock break out ( @xcite ) are subject to uncertainties in the color temperature compared to the effective temperature ( @xcite ) , so must also be considered with caution . given the concrete evidence for asymmetry and the likelihood that the shock erupted from restricted areas of the surface of the progenitor , all of these issues need to be reconsidered in the context of jet - like shock breakout .
sn 2008d has been observed to be significantly polarized at maximum and two - weeks after maximum light .
a significant amount of polarization is associated with the interstellar polarization , @xmath133 and @xmath134 , arising in the host galaxy .
+ sn 2008d is intrinsically polarized , with a continuum polarization implying an asymmetry of the photosphere of @xmath135 .
strong polarization associated with spectral lines shows evidence for significant asymmetries in the line forming region of the ejecta above the photosphere .
loops on the stokes @xmath38 plane are observed for lines of , in particular @xmath28 , and . a low level of polarization is potentially associated with @xmath4 , although this feature is complicated by blending with a strong telluric feature .
significant polarization is also observed for and ; in the case of the latter , polarization of the highest velocity component of ir3 was measured to be @xmath93 and @xmath136 at the first and second epochs respectively .
the polarization of the ir3 feature shows a significantly different polarization angle , at both epochs , to other species in the spectrum , suggesting it arises in a different spatially distinct , higher - velocity portion of the ejecta than the lines of other species ( i.e. the line forming region for does not just lie at higher velocities than the line forming regions of other species , but also with a different geometry ) .
+ the strong polarization associated with these lines suggests incomplete coverage of the photosphere by the line forming regions for these species , and the differing polarization angles imply that the line forming regions are different for different species .
the absence of a strong polarization signature associated with @xmath103 , the absence of ni or co in early time ir spectra , and a relatively low level of continuum polarization suggests no core material was transported to the outer layers of the ejecta by , for example , the action of a jet - like flow .
+ the observed asphericity of the photosphere ( albeit at a relatively low level ) and the chemically dependent structure of the line forming regions above the photopshere demonstrates that sn 2008d was not spherically symmetric .
+ if sn 2008d was the result of a jet - induced explosion , then we conclude that the jet stalled within the core , causing the core and , hence , photosphere to become slightly elongated .
the degree of elongation is observed to change as a function of depth over the period of our spectropolarimetry and over the period for which broad - band polarimetry has also been reported . in terms of spectropolarimetry ,
sn 2008d is found to be closely related to sn 2005bf ( albeit at 6 days before the second light curve maximum of the latter ) .
the spectropolarimetric study of the geometries of sn 2002ap , 2005bf , and 2008d are shown to be consistent within the jet - torus paradigm , but in which the jet remains within the cores of the progenitor .
we hypothesise that relatively late - time polarimetric observations of ccsne , before the optical depth to electron scattering drops below unity , will reveal a geometry discontinuity consistent with the photosphere having reached the position of a stalled jet . +
the observation of asymmetries of sn 2008d at early times has important implications for the study of the x - ray and uv shock breakout and the assumption of spherical symmetry .
the research of jrm is funded through the sophie & tycho brahe fellowship .
the dark cosmology centre is supported by the dnrf .
the research of jcw is supported in part by nsf grant ast-0707769 .
the authors are grateful to the european organisation for astronomical research in the southern hemisphere for the generous allocation of observing time .
they especially thank the staff of the paranal observatory for their competent and never - tiring support of this project in service mode .
cccccc object & date & exposure & median & median & type + & ut & ( s ) & airmass & altitude & + ltt 3218 & 2008 jan 31.16 & 30 & 1.027 & 76.74 & flux std . + sn 2008d & 2008 jan 31.22 & @xmath137 & 1.894 & 31.80 & object + + sn 2008d & 2008 feb 15.18 & @xmath138 & 1.899 & 31.69 & object + gd 108 & 2008 feb 15.23 & 50 & 1.056 & 71.33 & flux std .
+ vela 1 95&2008 feb 15.24 & @xmath139 & 1.156 & 59.93 & pol .
std . + , ( 42 ) , @xmath30 , @xmath4 and ir3 for the first and second observations of sn 2008d .
the angle of each arc corresponds to the observed polarization angle on the sky ( the arc length is @xmath140 ) and the radius of the arc corresponds to the velocity . in this figure , north is up and east is left , with the corresponding orientation of the stokes parameters indicated in the bottom left of the left panel.,width=340 ] features arise in a shell at the edge of the ejecta , with the high velocity material aligned with the jet axis . , and occur in the main portion of the ejecta .
the ni - rich jets are contained within core , below the depth of the photosphere at the first and second epochs ( dotted line ) .
b ) kinetic energy dominated jet explosion .
ni - rich material and other core - material , such as and , are transported to the surface .
the outer edges of the ejecta are excited in the vicinity of the ni - rich clumps ( heavy black dotted lines ) .
the position of the photosphere , at the two observational epochs , are indicated by light dotted lines.,width=359 ] | spectropolarimetry of the type ib sn 2008d , associated with the xrf 080109 , at two separate epochs , are presented .
the epochs of these observations correspond to v - band light curve maximum and 15 days after light curve maximum ( or 21 and 36 days after the xrf ) .
we find sn 2008d to be significantly polarized , although the largest contribution is due to the interstellar polarization component of @xmath0 and @xmath1 . at the two epochs ,
the spectropolarimetry of sn 2008d is classified as being d1+l()+l ( ) . the intrinsic polarization of continuum wavelength regions is @xmath2 , at both epochs , implying an asymmetry of the photosphere of @xmath3 .
similar to other type ibc sne , such as 2005bf , 2006aj and 2007gr , we observed significant polarization corresponding to the spectral features of , , , and , possibly , @xmath4 , about a close - to - spherically - symmetric photosphere .
we introduce a new plot showing the chemically distinct line forming regions in the ejecta and comment on the apparent ubiquity of highly polarized high - velocity features in type ibc sne .
the polarization angle of ir triplet was significantly different , at both epochs , to those of the other species , suggesting high - velocity forms in a separate part of the ejecta .
the apparent structure in the outer layers of sn 2008d has implications for the interpretation of the early - time x - ray emission associated with shock break - out .
we present two scenarios , within the jet - torus paradigm , which explain the lack of an apparent geometry discontinuity between the two observations : 1 ) a jet which punched a hole straight through the progenitor and deposited ni outside the ejecta and 2 ) a jet which stalled inside the radius of the photosphere as observed at the second epoch .
the lack of a peculiar polarization signature , suggesting strongly asymmetric excitation of the ejecta , and the reported properties of the shock - breakout favour the second scenario . |
the authors acknowledge fruitful discussions with l. gerland and a. dumitru .
s. hossenfelder wants to thank the land hessen for financial support .
l. randall and r. sundrum , phys .
* 83 * ( 1999 ) 4690 [ arxiv : hep - th/9906064 ] .
l. randall and r. sundrum , phys .
* 83 * ( 1999 ) 3370 [ arxiv : hep - ph/9905221 ] .
y. uehara , mod .
phys . lett .
a * 17 * ( 2002 ) 1551 [ arxiv : hep - ph/0203244 ] .
k. cheung , [ arxiv : hep - ph/0003306 ] .
presented at 7th international symposium on particles , strings and cosmology ( pascos 99 ) , granlibakken , tahoe city , california , 10 - 16 dec 1999 . published in * lake tahoe 1999 , particles , strings and cosmology * 145 - 155 t. g. rizzo , [ arxiv : hep - ph/9910255 ] .
proceedings of 2nd international conference physics beyond the standard model : beyond the desert 99 : accelerator , nonaccelerator and space approaches , ringberg castle , tegernsee , germany , 6 - 12 jun 1999 . published in * tegernsee 1999 , beyond the desert 1999 * 23 - 44 s. hofmann , m. bleicher , l. gerland , s. hossenfelder , k. paech and h. stcker , j. phys .
g * 28 * ( 2002 ) 1657 .
m. bleicher , s. hofmann , s. hossenfelder and h. stcker , phys . lett . *
b548 * , 73 - 76 ( 2002 ) , [ arxiv : hep - ph/0112186 ] .
a. ringwald and h. tu , phys .
b * 525 * , 135 ( 2002 ) [ arxiv : hep - ph/0111042 ] .
y. uehara , prog .
phys . *
107 * , 621 ( 2002 ) [ arxiv : hep - ph/0110382 ] .
g. landsberg , phys .
lett . * 88 * , 181801 ( 2002 ) [ arxiv : hep - ph/0112061 ] .
a. v. kotwal and c. hays , phys . rev .
d * 66 * ( 2002 ) 116005 [ arxiv : hep - ph/0206055 ] .
r. c. myers and m. j. perry ann .
phys . * 172 * , 304 - 347 ( 1986 ) j. preskill , [ arxiv : hep - th/9209058 ] presented at international symposium on black holes , membranes , wormholes and superstrings , woodlands , tx , 16 - 18 jan 1992 . in * houston 1992 , proceedings , black holes , membranes , wormholes and superstrings
* 22 - 39 .
a. r. liddle and a. m. green , phys .
* 307 * ( 1998 ) 125 [ arxiv : gr - qc/9804034 ] .
d. blais , c. kiefer and d. polarski , phys .
b * 535 * ( 2002 ) 11 [ arxiv : astro - ph/0203520 ] .
j. d. barrow , e. j. copeland and a. r. liddle , phys .
d * 46 * ( 1992 ) 645 . | recent calculations applying statistical mechanics indicate that in a setting with compactified large extra dimensions a black hole might evolve into a ( quasi-)stable state with mass close to the new fundamental scale @xmath0 .
black holes and therefore their relics might be produced at the lhc in the case of extra - dimensional topologies . in this energy regime , hawking s evaporation scenario is modified due to energy conservation and quantum effects .
we reanalyse the evaporation of small black holes including the quantisation of the emitted radiation due to the finite surface of the black hole .
it is found that observable stable black hole relics with masses @xmath1 would form which could be identified by a delayed single jet with a corresponding hard momentum kick to the relic and by ionisation , e.g. in a tpc .
# 1@xmath2#1 the idea of large extra dimensions ( lxds ) which was recently proposed in @xcite might allow to study interactions at trans - planckian energies in the next generation collider experiments . here , the hierarchy - problem is solved or at least reformulated in a geometric language by the existence of @xmath3 compactified lxds in which only the gravitons can propagate .
the standard - model particles are bound to our 4-dimensional sub - manifold , often called our 3-brane .
the strength of a force at a distance @xmath4 generated by a charge depends on the number of space - like dimensions . for distances smaller than the compactification length @xmath5
, the gravitational interaction drops faster compared to the other interactions . for distances much bigger than @xmath5 gravity
is described by the well known potential law @xmath6 . however , starting from @xmath7 the force lines are diluted into the extra dimensions resulting in a smaller effective coupling constant for gravity .
this scenario would lead to the following relation between the four - dimensional planck mass @xmath8 and the higher dimensional planck mass @xmath0 , which is the new fundamental scale of the theory @xmath9 the lowered fundamental scale would induce a vast number of observable phenomena for quantum gravity at energies in the range @xmath0 .
in fact , the non - observation of these predicted features gives first constraints on the parameters of the model , the number of extra dimensions @xmath3 and the fundamental scale @xmath0 @xcite .
on the one hand , this scenario would have major consequences for cosmology and astrophysics such as the modification of inflation in the early universe and enhanced supernova - cooling due to graviton emission @xcite . on the other hand , additional processes had to be expected in high - energy collisions @xcite : production of real and virtual gravitons @xcite and the creation of black holes at energies that can be achieved at colliders in the near future .
especially the possibility of black hole production in lxds at the lhc and from cosmic rays has received great attention @xcite .
black holes produced in such interactions would be tiny and may decay on fm / c time scales @xcite .
thus , the decay of these objects ( black holes , p - branes , string balls ) could be studied in detail in the laboratory .
unfortunately , very little is known about the final stages of black hole evaporation .
extensive speculations about the final fate of black holes have been brought forward in the literature and will be discussed in detail later .
the general notion is that a small black hole stops evaporating particles when its mass approaches the planck scale , resulting in the exciting possibility of forming a ( quasi-)stable relic . in this letter , we study a model for the radiation from small black holes assuming a geometrical quantisation of the emitted radiation .
the quantisation of the radiation can lead to black hole relics of masses around 1 - 3 tev with small electric charge .
those relics might be observable at the lhc .
let us start with the properties of black holes which may be accessible in the next generation colliders .
a black hole which might be produced with @xmath10 tev would have a radius much smaller than the size @xmath5 of the lxds @xcite . in this case
one can neglect the periodic boundary conditions due to compactification and approximate the space - time to be spherically symmetric . for these black holes ,
the metric is given by the @xmath11 - dimensional schwarzschild metric @xcite .
following ref.@xcite and implying the extra dimensions via eq .
( [ master ] ) , the schwarzschild radius is given by @xmath12 the metric takes the familiar form @xmath13 with @xmath14 being the surface element of the @xmath15-dimensional sphere , containing @xmath16 angles and @xmath17 given by @xmath18 from this one gets the surface gravity : @xmath19 which is the newtonian force at the horizon in the schwarzschild case .
the surface of the black hole is @xmath20 with @xmath21 denoting the surface of the unit @xmath22-sphere @xmath23 the production cross section for black holes in parton - parton or @xmath24 collisions can be estimated on geometrical grounds and is of order @xmath25 @xcite .
detailed studies to support this estimate can be found in the recent works by jevecki and thaler@xcite and by eardley and giddings@xcite . assuming the validity of the classical approximation and setting @xmath26 tev and @xmath27 one finds @xmath28 tev@xmath29 pb . the luminosity of pp interactions at the lhc would then allow the production of approximately @xmath30 black holes per year @xcite .
the fate of these small black holes is difficult to estimate .
there final evaporation is closely connected to the information loss puzzle .
the black hole emits thermal radiation , whose sole property is the temperature , regardless of the initial state of the collapsing matter .
so , if the black hole completely decays into statistically distributed particles , unitarity can be violated .
this happens when the initial state is a pure quantum state and then evolves into a mixed state @xcite . when one tries to avoid the information loss problem two possibilities are left .
the information is regained by some unknown mechanism or a stable black hole remnant is formed which keeps the information . besides the fact that it is unclear in which way the information should escape the horizon @xcite there are several other arguments for black hole relics @xcite : * the uncertainty relation : the schwarzschild radius of a black hole with planck mass is of the order of the planck length . since the planck length is the wavelength corresponding to a particle of planck mass , a problem arises when the mass of the black hole drops below planck mass .
then one has trapped a higher mass , @xmath31 , inside a volume which is smaller than allowed by the uncertainty principle @xcite . to avoid this problem ,
zeldovich has proposed that black holes with masses below planck mass should be associated with stable elementary particles @xcite * corrections to the lagrangian : the introduction of additional terms , which are quadratic in the curvature , yields a dropping of the evaporation temperature towards zero @xcite .
this holds also for extra dimensional scenarios @xcite and is supported by calculations in the low energy limit of string theory @xcite . * further reasons for the existence of relics
have been suggested to be black holes with axionic charge @xcite , the modification of the hawking temperature due to quantum hair @xcite or magnetic monopoles @xcite .
coupling of a dilaton field to gravity also yields relics , with detailed features depending on the dimension of space - time @xcite .
let us now compare the classical micro - canonical emission scenario to an approach that takes into account the effects of the geometrical quantisation of the emitted radiation .
note that this approach is different from a quantisation of the event horizon . in the present model
, all horizon configurations can still be realized .
generally , black holes emit particles via the hawking mechanism @xcite .
the temperature of the radiation is : @xmath32 with @xmath33 given by ( [ kappa ] ) . from dimensional aspects we expect the entropy to be @xmath34 . from thermodynamics we know that , @xmath35 where @xmath36 is interpreted as the conserved energy of the system . inserting eqs .
( [ ssr ] ) , ( [ kappa ] ) and ( [ tempkappa ] ) one obtains a mass dependence of @xmath37 with the exponent @xmath38 . integration yields @xmath39 in the limit where the energies of the emitted particles are small compared to the mass of the black hole the grand - canonical ensemble can be used .
the number density of particles with energy @xmath40 is then @xmath41 with which one derives the higher dimensional analogue of the stefan - boltzmann law @xcite @xmath42 the spectrum of the emitted radiation has a maximum at frequencies of the order of the temperature .
thus , when the mass of the black hole approaches the planck scale , the energy of the emitted quanta can no longer be neglected .
since we are interested in the late evaporation stage , when the black hole is small and hot , an appropriate statistical description is given by the micro - canonical ensemble @xcite . here , the single - particle number density is given by @xmath43}}{{\rm exp}[s(m)]}\quad,\ ] ] where @xmath44 denotes the entropy of the black hole .
let us shortly examine the limit of huge black hole masses in the micro - canonical approach .
the limit of the grand - canonical number density is wien s limit , i.e. @xmath45 . in the micro - canonical case @xmath46 in leading order in @xmath47 .
thus , wien s limit is recovered .
next we investigate the multi - particle spectrum in the micro canonical description : @xmath48}{{\rm exp}[s(m ) ] } \quad,\ ] ] with @xmath49 being the smallest integer next to @xmath50 .
this cut - off assures that the total energy of the emitted quantum does not exceed the mass of the black hole . after substituting @xmath51
one obtains for the total energy density which is radiated off by the black hole @xmath52 the evaporation rate per degree of freedom is given by @xmath53 fig .
[ dmdtq ] ( thin dotted lines ) shows the evaporation rate ( [ mdoteq ] ) as a function of the initial mass @xmath36 of the black hole . in the limit @xmath54
, the micro canonical evaporation rate reproduces the hawking evaporation in @xmath15 space - like dimensions : @xmath55 which is , using ( [ tempkappa ] ) , @xmath56 now we address the model with geometrical quantisation of the emitted radiation .
the radiation of a black hole derived by semi - classical quantum field theory in curved space @xcite yields a black body spectrum in @xmath22 dimensions .
spherical symmetry is taken into account by making the usual separation ansatz for the wave equation : @xmath57 which factorizes the full wave function into an amplitude @xmath58 , a radial wave function @xmath59 and a set of spherical harmonics @xmath60 . here
@xmath61 is an abbreviation for the occurring @xmath62 angles .
the coefficients in the multipole expansion suppress the energy emission in higher order harmonics , therefore we consider only the @xmath63 mode . in this case
the dispersion relation is the usual one for massless particles @xmath64 .
the boundary conditions of the black body lead to a restriction of the possible momenta . that was discussed in ref . @xcite . ]
this results in a geometrical quantisation of the momentum spectrum @xmath65 as a consequence , the multi - particle spectrum ( [ mehrteil ] ) is modified , since the emitted particles have energies in integer multiples of a minimal energy quantum @xmath66 : @xmath67}{{\rm exp}[s(m ) ] } \theta\left(m - l \delta \omega \right ) \quad.\ ] ] here the @xmath68-function cuts off the spectrum when the energy of one particle exceeds the mass of the black hole .
the energy density of the radiation is derived by summation over momentum space @xmath69 in the limit of large black hole masses @xmath70 one has @xmath71 and regains the continuous emission spectrum from eq .
( [ epsq ] ) . the evaporation rate with respect to the geometrically quantised spectrum
is shown in fig .
[ dmdtq ] ( thick lines ) and compared to the continuous spectrum case ( thin dotted lines ) .
the spacing of energy levels gets smaller with increasing @xmath36 , and whenever it is possible to occupy an additional level the evaporation rate exhibits a step .
these steps naturally appear in spacings @xmath72 tev , because in the mass range of interest it is @xmath73 the exact value thereby depending on the number of extra dimensions .
the evaporation process occurs in quantised steps .
it can not proceed further when the lowest lying quantum state allowed exceeds already the mass of the black hole .
evaporation is halted at a finite mass value for a certain fraction of initial masses above the fundamental scale .
it should be noted that the transverse momentum spectra of the radiation is modified as compared with a simple extrapolation of the hawking formula to small masses : here the final quanta possess energies of the order @xmath74 , and not a continuous spectrum with @xmath75 .
if the mass were slightly above the planck scale , it would still be possible for the black hole to emit a particle carrying away most of its energy .
this might leave a stable relic with mass below planck mass .
however , since physics below the planck scale is unknown , we can not be sure about the existence and properties of relics with masses below the fundamental scale .
thus , in the following we will focus on those relics with masses above the fundamental scale .
it is interesting to ask for the spectrum of final relic masses @xmath76 depending on the initial mass @xmath77 of the black hole .
this relation is shown in fig .
[ miniend ] for the case @xmath78 .
the most probable case is the exclusive emission of minimal energy quanta during the evaporation process , depicted by solid lines in fig .
[ miniend ] .
black holes with @xmath79 are stable with @xmath80 .
the inclusion of higher modes in the evaporation process , which becomes more important with increasing initial mass , is shown by the dashed and dotted lines . when going even further , to masses @xmath81 , the whole range of end masses would become accessible . because the black holes accessible at the lhc would have mainly masses slightly above @xmath0 , most of these black holes were stable with masses around @xmath0 .
indirect constraints on the existence of black hole relics can be obtained from the decay of primordial black holes .
their modified and dimension dependent energy spectrum influences observables , e.g. the cosmic microwave background and the baryogenesis @xcite .
furthermore , the black hole relics from primordial density fluctuations may be a candidate for dark matter @xcite .
upper limits on the relative contribution of those relics to the critical energy density in the universe are on the order of @xmath82 @xcite .
thus , the observation of relics in a collider experiment is of highest interest .
the final and most interesting question is : how could one observe these relics ? * a black hole relic with a mass of @xmath83 tev would have a spectrum that just fails to allow for a last emission of a quantum ( cf . fig .
[ miniend ] ) . therefore
, if its mass were increased only slightly by the energy @xmath84 , it would enable the black hole to evaporate again , emitting a high energetic quantum and leaving a tiny mass relic .
this might result in a delayed flash of hard photons , leptons or qcd jets compared to the collision dynamics encountered at the lhc .
the fraction of black holes evaporating in this manner can be estimated from the mass spectrum of black holes produced at the lhc and is @xmath85 for @xmath86 gev .
note that relics from primordial black holes might also lead to observable air showers , if a black hole relic evaporates in the atmosphere .
* a certain fraction of the black holes produced in parton - parton collisions would carry a small charge of order @xmath87 .
this might allow to identify the charged black hole relics , e.g. by ionisation in a time projection chamber .
* the thermal evaporation spectrum would be much softer as expected in the literature @xcite which assumes total decay of tev black holes .
however , the final stages would be governed by non - thermal particle emission .
to summarise , we have analysed the late stages of black hole evaporation , including the geometrical quantisation of the emitted radiation . in this model setting , the production of stable black hole relics would be possible at the lhc .
these relics may be observable in a late burst of jets if they capture additional energy .
charged black hole relics may eventually be directly detected , e.g. in a tpc by ionisation . |
when a short and intense pulse of light is launched into a microstructured fiber ( mf ) , nonlinear effects can significantly broaden the spectrum to much more than an octave @xcite .
these supercontinua are used in a variety of applications such as ultrafast optical switching , spectroscopy , optical coherent tomography , optical clocks , etc .
routinely , around 100-fs pulses are used at a wavelength close to a zero - group velocity dispersion wavelength ( zdw ) of the fiber . using nearly octave spanning input pulses @xcite , the phenomenon of ` pulse trapping ' @xcite can be observed . in this effect
a non - dispersing pulse can form at the short wavelength end of the broad spectrum .
this is surprising because the pulse exists in a region of normal dispersion in the fiber , where the dispersion - induced chirp can not be cancelled by self - phase modulation .
the non - dispersing pulse is trapped behind a fundamental soliton in the anomalous dispersion regime that was generated by soliton fission .
soliton fission can create multiple fundamental solitons , which shift to longer wavelengths via the soliton self - frequency shift ( ssfs ) @xcite .
nishizawa and goto @xcite were the first to demonstrate that a soliton undergoing the ssfs can trap light behind it .
the trapped light adjusts to a wavelength that is group velocity matched to the soliton .
because it is forced to travel with the soliton and lies in the normal dispersion regime , it has to shift to shorter wavelengths in order to keep the same group - velocity as the soliton .
gorbach and skryabin provided an alternative view of pulse trapping @xcite : the light is trapped because the soliton accelerates . in simulations they turned off the ( negative ) acceleration induced by the ssfs and the trapping ceased .
the acceleration of the soliton provides a ` gravity - like ' potential to the trapped pulse . in this paper
we study experimentally the phenomenon of pulse trapping in a fiber with two zdws .
we use intense few - cycle pulses to generate a soliton and a trapped pulse and focus in particular on the trap dynamics as the soliton reaches the longer zero - dispersion wavelength and decays .
we observe how the light escapes the trap , expanding to a few times the trap length . for higher input pulse energies ,
a further component in the spectrum at even shorter wavelengths is formed , which dominates the blue end of the evolving supercontinuum .
as previously shown , pulse trapping can be explained by an effective potential that is produced by an accelerating pulse the fundamental soliton undergoing ssfs @xcite .
the discussion below follows this analysis .
the soliton induces a nonlinear modification of the refractive index , @xmath2 , via the optical kerr effect @xcite : @xmath3 where @xmath4 is the intensity and @xmath5 the nonlinear index coefficient , assuming an instantaneous response , and @xmath6 is the linear index . any light field @xmath7 in the fiber that interacts with
the soliton sees the nonlinear contribution to the refractive index , @xmath8 , and experiences cross - phase modulation ( xpm ) @xcite .
the evolution of the slowly varying envelope @xmath9 of this light is governed by @xcite : @xmath10 @xmath11 and @xmath12 are space and time in the laboratory frame . @xmath13 and @xmath14 are the dispersion parameters , where @xmath15 evaluated at the frequency of @xmath7 .
since this frequency lies in the normal dispersion regime , @xmath14 is positive .
@xmath16 is the xpm - term and @xmath17 , where @xmath18 is the envelope of the soliton and @xmath19 is the nonlinear parameter . in the co - moving frame given by @xmath20 and @xmath21 , eqn .
( [ agpulseprop ] ) is transformed into the nonlinear schrdinger equation @xmath22 in analogy to the ordinary schrdinger equation in quantum mechanics , the xpm contribution @xmath23 induced by the soliton plays the part of a potential barrier
. equation ( [ agpscheqn ] ) is transformed into an accelerating frame @xcite : @xmath24 here @xmath25 describes the rate of the ssfs , which determines the soliton acceleration . for a soliton , @xmath26 @xcite .
@xmath27 is the raman time ; the soliton length is @xmath28 .
@xmath29 is the group velocity dispersion at the soliton wavelength , and the minus sign indicates that the accelerating soliton is redshifting and slowing down .
the transformation fixes the peak of the accelerating soliton at @xmath30 . in this frame , the propagating light @xmath7 acquires a phase @xmath31 , so @xmath32 is replaced with @xmath33 @xcite : @xmath34 the propagation equation ( [ agpscheqn ] ) becomes @xmath35 where @xmath36 therefore , in a frame that accelerates with the soliton , other waves propagate in the presence of an effective potential @xmath37 .
the first term in equation [ eqpot ] is the barrier due to xpm and the second term is a linear potential due to the ( negative ) acceleration @xmath25 .
@xmath38 is sketched in figure [ potentialpic ] .
the barrier is located at @xmath30 and at @xmath39 the linear increase reaches the height of the barrier . .
the soliton acceleration causes the slope in the potential .
[ potentialpic ] ] gorbach and skryabin pointed out that the potential is ` gravity - like ' , i.e. similar to the gravitational potential on earth . standing in a lift ,
the floor acts as a strong potential barrier against the gravitational force . without gravity ,
the same potential would be created if the lift had a constant upward acceleration .
so a uniformly accelerating barrier can trap objects just as a barrier in a gravitational field does . as seen in figure
[ potentialpic ] , the potential can trap light between @xmath30 and @xmath40 .
we can estimate the temporal length of the trap by calculating @xmath41 using @xmath42 .
approximating @xmath43 and inserting @xmath25 we obtain for an accelerating soliton @xmath44 @xmath41 depends on the ssfs as well as the ratio of group velocity dispersions for the trapped light and the soliton . in our experiments
the trap length is relatively constant , so that the trapped light is expected to have a well defined temporal length .
the trap is positioned immediately behind the soliton center at @xmath45 . in consequence ,
the trapped light travels at the group velocity of the soliton and slows down with it .
the spectrum of the trapped light shifts to a wavelength associated with the same group velocity as the soliton .
accordingly , the potential determines both the spectral and temporal properties of the trapped light .
, of the fiber nl - pm-760 @xcite and the initial spectrum of the few - cycle pump pulses ( inset ) .
the dispersion zeroes at 760 nm and 1160 nm . [ dispersion and input spec pic ] ] .
the pulse energy is 39pj .
a gdd of 7.5fs@xmath46 broadens an unchirped 7-fs pulse by 10% in time .
[ varyinginputchirppic ] ] to create the trapping soliton as well as the trapped light we couple a few - cycle pulse into a microstructured fiber .
we use pulses from a @xmath477fs ti : sapphire laser with a repetition rate of 78mhz ( rainbow , femtolasers gmbh ) .
a reflective - optics telescope expands the beam and a parabolic mirror of high numerical aperture focuses the light for fiber coupling .
the use of mirrors minimizes dispersion - induced pulse broadening .
the 1.8m - long fiber ( nl - pm-760 , crystal fiber a / s ) has zero dispersion at 760 nm and 1160 nm .
the dispersion of the fiber and the initial pulse spectrum are shown in figure [ dispersion and input spec pic ] ; the nonlinear coefficient is @xmath48 at @xmath49 nm @xcite .
we rotate the fiber to align the fiber axes to the polarization of the incoming light . as a test of the pulse length at the fiber input face we replaced the fiber with a mirror and backreflected the pulse through the reflective focusing optics . compensating for group delay dispersion ,
no significant pulse broadening was observed after the double pass .
after the fiber we use a microscope objective to collimate the light and analyze it temporally with an autocorrelator ( pulsecheck , ape gmbh ) . by tilting the nonlinear crystal of the autocorrelator we perform frequency - resolved optical gating ( frog ) measurements @xcite .
we also pick off part of our laser output beam , reduce the spectral bandwidth with a 60-nm interference filter , and use it as a reference pulse to record x frog traces .
the reference pulse was characterized using a frog measurement to be 35-fs in length ( fwhm ) .
simultaneous to the frog and x - frog traces , we measure the spectrum of the trapped light and trapping solitons on a compact ccd - based spectrometer and an optical spectrum analyzer ( osa ) for longer wavelengths . figure [ varyinginputchirppic ] demonstrates that the fiber output spectra are very sensitive to the dispersion - induced chirp on the input pulse . in all experiments we adjusted the input chirp to obtain maximum output spectral width ( black curve in figure [ varyinginputchirppic ] ) .
we create solitons in the near - infrared and trapped light in the visible . because the input spectrum lies partially in the anomalous dispersion regime beyond 760 nm wavelength , a soliton forms and raman shifts via the ssfs . the part of the input light that is in the normal dispersion regime and that is trailing the soliton is trapped .
in addition there is interaction between light in the two dispersion regimes via self - phase modulation and phasematched amplification of dispersive waves @xcite .
therefore , both the trap and trapped light are created by launching a single pulse into the fiber . since the raman - shifting soliton traps light
, the trapped light has to maintain the soliton group velocity @xmath50 even though the soliton slows down .
the trapped light impinges onto the soliton which leads to a blueshift and eventually the maintenance of group velocity matching @xcite . increasing the input pulse energy in the fiber speeds up the ssfs and secondary fundamental solitons are created . for high input pulse energies the primary soliton shifts up to the long zdw , which it can not pass .
the ssfs abruptly ends and the soliton decays . as the acceleration of the soliton is terminated , the ` gravity - like ' part of the potential disappears , while the barrier decays slowly , i.e. the trapped light can escape behind the soliton . at the same time the wavelength shift of the trapped light ceases . when sufficient pulse energy is provided , a soliton is generated and redshifted due to the ssfs .
figure [ lowpowerspecevolution ] shows that the rate of shifting increases with pulse energy as longer wavelengths are reached within our finite fiber length . below a wavelength of 700 nm a trapped pulse is created that frequency shifts with the soliton , but in opposite direction .
the relative rate of shifting is determined by group velocity matching .
some of the initial pulse energy remains between 700 nm and 900 nm and propagates ahead of the slowing soliton .
note the feature near 1300 nm , which indicates the onset of ` cherenkov ' radiation , as discussed in the next section .
similar spectra have been observed in other experiments , see for example @xcite .
spectra for higher pulse energies are shown in figure [ highpowerspecevolution ] .
the ssfs has ceased and the soliton has reached a final wavelength of 1150 nm , just below the zdw of 1160 nm . approaching this wavelength , the soliton produces dispersive cherenkov waves around 1300 nm . the production of cherenkov radiation balances the stimulated raman effect in the soliton and the soliton redshift terminates .
since the soliton energy is gradually being transferred to the cherenkov radiation , the soliton decays @xcite .
figure[highpowerspecevolution ] also shows that nearly all of the trapped light shifts to wavelengths shorter than the group velocity matched wavelength .
these are propagating slower than the soliton - , that is the light escapes behind the soliton .
light at wavelengths travelling faster than the soliton collides with the soliton one last time and shifts to the blue beyond the group velocity matched wavelength .
this is seen particularly clearly by examining the soliton and the trapped light in the time domain , which we describe next . ,
title="fig : " ] , title="fig : " ] ] for low input pulse energies the soliton undergoes ssfs throughout the fiber and hence it is always slowing down . the ` gravity - like ' part of the trap potential still exists at the fiber end .
the trapped light is confined behind the barrier created by the soliton . in figure [ low power shg frog ] second harmonic - frog ( shg - frog ) measurements @xcite of the visible part of the spectrum
show the trapped pulse .
all of the frog and x - frog traces in this paper use the same linear color scale shown .
the trapped pulse has a duration of @xmath51ps .
the time - bandwidth product ( fwhm ) of this pulse is more than ten the spectral width could support a sub 50-fs pulse .
the frog retrieval requires a frog grid as large as 2048x2048 pixels . before retrieval , the wavelength is scaled by a factor to fit the data into a square grid @xcite .
after retrieval the wavelength is unscaled to recover the physical scale . the frog error in figure [ low power shg frog ] is 0.00468 ( left column ) and 0.00557 ( right column ) .
shg - frog has an ambiguity regarding the direction of time and the frog measurement is not suitable to resolve the relative positions in time of the trapped pulse and soliton .
therefore we performed x - frog measurements on the solitons and the trapped pulse by mixing the light from the fiber with the reference pulse ( fwhm 35fs ) derived before the fiber and spectrally filtered .
the x - frog traces can be interpreted directly .
results are shown in figure [ low power xfrog pic ] , where the propagation direction is to the left . again the ps - wide trapped pulse is measured .
it is confined to the region behind the soliton . as the soliton redshifts and delays , the trapped light blueshifts to maintain group velocity matching with the soliton .
note that the trapped pulse trails the soliton by about 0.5ps .
this delay is due to the normal group velocity dispersion of the microscope objective used to collimate the light from the fiber .
the measurements clearly show how the trapped pulse is strongly blocked by the soliton induced barrier .
the retrieved pulse shapes in figure [ low power shg frog ] ( left ) indicate that the trapped pulse has a small , low intensity shoulder on one side .
the x - frog measurements in figure [ low power xfrog pic ] show that this shoulder extends behind the trapping soliton and reveal some ps temporal oscillations not resolved in the retrieved frog traces .
the shoulder seems induced by the asymmetry of the confining potential .
the linearly increasing part of the trap potential caused by the ssfs is much more shallow than the sharp edge of the barrier .
, title="fig : " ] , title="fig : " ] the size of the trap can be estimated by equation [ eqtauprime ] .
the soliton , when shifted to a wavelength around 1000 nm , maintains a near constant width of @xmath52fs for different input pulse energies as measured separately by autocorrelation . with @xmath53fs @xcite and a ratio @xmath54
( see figure [ dispersion and input spec pic ] ) this gives a trap length of 4ps .
the actual length of a pulse in this trap will be somewhat shorter as the pulse energy will not reach the height of the potential barrier of the trap due to tunneling .
the observed pulse length of about 1ps is in good agreement with this estimate .
] when the input pulse energy in the fiber is sufficient for the soliton to reach the longer zero - dispersion wavelength , the behavior of the trapped light changes .
shg - frog measurements of the trapped light are shown in figure [ high power shg frog ] .
( frog retrieval errors : 0.00449 ( left ) , 0.00484 ( right ) ) .
the pulse peak has widened considerably beyond 1ps and the small shoulder of figure [ low power shg frog ] has developed into a long tail . with more energetic input pulses the tail extends further .
this behavior again is a direct consequence of the soliton dynamics .
when the soliton wavelength is stabilized by cherenkov radiation , the acceleration terminates rapidly , but the soliton decays slowly @xcite .
the barrier part of the potential remains , but the gravity - like part disappears .
hence one side of the potential well vanishes and the light escapes the trap to one side creating the pronounced tail in figure [ high power shg frog ] .
a higher launched pulse energy leads to a faster ssfs and the soliton acceleration turns off earlier on in the fiber .
thus the trapping ceases earlier and the trapped light can spread out more under dispersion during the remaining length of fiber .
this is seen in figure [ high power shg frog ] as the tail extends further for higher pulse energies .
corresponding x - frog traces of the trapped pulse and soliton after the soliton redshift has ceased are shown in figure [ high power xfrog pic ] .
the soliton remains a barrier and prevents the trapped pulse from overtaking as before . on the other side
the developing tail is clearly visible , extending more than 10ps behind the soliton barrier ( see also figure [ high power shg frog ] ) .
again the tail bears oscillations not resolved in the retrieved frog traces .
the shorter wavelengths in the tail travel slower due to the normal dispersion .
the spectrum of the trapped pulse in figure [ high power shg frog ] ( right ) is centered at 580 nm with a spectral width of 40 nm . at this wavelength
the fiber dispersion is @xmath55 ps@xmath46/km ( figure [ dispersion and input spec pic ] ) .
this leads to an expected length of the tail of about 10ps per meter , in good agreement with the measured length .
thus as soon as the trapping terminates , the trapped light spreads quickly .
the trace is similar to previous frog measurements of the trapped light after the trapping has ended ( see figure [ high power shg frog ] ) .
[ veryhighpowershg ] ] by sacrificing the reference beam used in the x frog measurements we are able to further increase the pulse energy into the fiber .
the soliton now reaches the wavelength stabilized by cherenkov radiation even earlier in the fiber .
unfortunately , the large temporal width of the pulse and a small signal appearing at longer wavelengths made it impossible for our code to retrieve the frog trace .
the frog trace is similar to traces of the trapped light at lower powers ( compare to figure [ high power shg frog ] ) .
the length of the tail is determined by the dispersion over the full fiber length .
the wavelength of the pulse peak is arrested at @xmath0590 nm .
this stabilizes the trapped pulse and makes it independent of coupled input energy in this energy range .
the spectrum continues to reach shorter wavelengths , considerably below the group velocity - matched point ( figure [ veryhighpowerspecevolution ] ) .
we attribute the additional peak of light appearing below 550 nm to nonlinearly phase matched resonant wave mixing at the very input of the fiber @xcite .
this is consistent with the observation that the peak shifts to the blue with increasing energy .
similar behaviour has been seen in @xcite . in our case
there is a particularly pronounced spectral gap at 550 nm , independent of input energy and of constant shape . in the infrared
we see that multiple solitons , created from the input pulse by soliton fission , have shifted towards the long zdw .
the soliton that is arrested just short of the long zdw is at a minimum of group velocity .
the initially trapped light still trails the soliton and thus is of a wavelength shorter than the group velocity - matched wavelength .
light in the trap that had been propagating @xmath56 faster than the soliton before release has undergone a spectral blueshift at the soliton to a group velocity @xmath56 slower than the soliton @xcite .
by measurements in both the temporal and spectral domains , we show how the dynamics of a soliton determine both the wavelength and pulse - like nature of the short wavelength end of the supercontinuum .
light in the normal dispersion regime forms a trapped pulse governed by a potential consisting of a nonlinear barrier and a ` gravity - like ' linear potential . the linear part is due to the ssfs which causes a negative acceleration of the soliton .
as long as the soliton is accelerating , the trapping confines the light in time .
when the ssfs terminates , the trapped light falls behind the soliton , moving to a wavelength associated with a group velocity slower than the soliton .
this effect dominates the shape of the supercontinuum at short wavelengths until resonant wave mixing with the input pulse further broadens the spectrum .
this setup allows for the creation and release of wavelength tunable picosecond pulses in the normal dispersion regime at any location along a microstructured fiber . the wavelength for the trapped pulse
is obtained by choosing an appropriate dispersion profile of the fiber .
the most important parameter is the longer zero dispersion wavelength .
if the fiber has a long zdw that is longer than 1160 nm , then the ssfs would take the soliton further into the infrared before the acceleration terminates .
the trapped pulse would shift to even shorter wavelengths and still be confined in time .
once the soliton is arrested , the trapped pulse has reached the final wavelength which is largely independent of pulse energy .
hence , the trapped pulse wavelength could be increased from this by use of lower input pulse energies or chirped input pulses .
we are indebted to dmitry skryabin , andrey gorbach , scott robertson , franz krtner , peter staudt , klaus metzger and wilson sibbett for discussions and technical support .
this work is supported by the epsrc , the royal society , and the leonhardt group aue . | we observe the dynamics of pulse trapping in a microstructured fiber .
few - cycle pulses create a system of two pulses : a raman shifting soliton traps a pulse in the normal dispersion regime .
when the soliton approaches a wavelength of zero group velocity dispersion the raman shifting abruptly terminates and the trapped pulse is released . in particular , the trap is less than 4ps long and contains a 1ps pulse . after being released
, this pulse asymmetrically expands to more than 10ps . additionally , there is no disturbance of the trapping dynamics at high input pulse energies as the supercontinuum develops further .
99 j. m. dudley , g. genty , s. coen , `` supercontinuum generation in photonic crystal fiber , '' 11351184 ( 2006 ) .
n. ishii , c. y. teisset , s. kohler , e. e. serebryannikov , t. fuji , t. metzger , f. krausz , a. baltuska , a. m. zheltikov , `` widely tunable soliton frequency shifting of few - cycle laser pulses , '' 036617 ( 2006 ) .
e. e. serebryannikov , a. m. zheltikov , n. ishii , c. y. teisset , s. kohler , t. fuji , t. metzger , f. krausz , a. baltuska , `` soliton self - frequency shift of 6-fs pulses in photonic - crystal fibers , '' 585588 ( 2005 ) .
p. falk , m. h. frosz , o. bang , l. thrane , p. e. andersen , a. o. bjarklev , k. p. hansen , and j. broeng , `` broadband light generation at @xmath0 1300 nm through spectrally recoiled solitons and dispersive wave , '' 621623 ( 2008 ) .
n. nishizawa and t. goto , `` characteristics of pulse trapping by use of ultrashort soliton pulses in optical fibers across the zero - dispersion wavelength , '' 11511159 ( 2002 ) .
n. nishizawa , t. goto `` pulse trapping by ultrashort soliton pulses in optical fibers across zero dispersion wavelength , '' 152154 ( 2002 ) .
n. nishizawa , t. goto `` ultrafast all optical switching by use of pulse trapping across zero dispersion wavelength , '' 359365 ( 2003 ) .
a. v. gorbach , d. v. skryabin `` light trapping in gravity like potentials and expansion of supercontinuum spectra in photonic
crystal fibres , '' nature photonics * 1 , * 653656 ( 2007 ) .
j. c. travers , a. b. rulkov , b. a. cumberland , s. v. popov , and j. r. taylor `` visible supercontinuum generation in photonic crystal fibers with a 400w continuous wave fiber laser , '' 1443514447 ( 2008 ) .
b. a. cumberland , j. c. travers , s. v. popov , and j. r. taylor `` toward visible cw - pumped supercontinua , '' 21222124 ( 2008 ) .
a. v. husakou and j. herrmann , `` supercontinuum generation of higher - order solitons by fission in photonic crystal fibers , '' 203901 ( 2001 ) .
g. p. agrawal , _ nonlinear fiber optics _ ( academic press , 2006 ) . t. g. philbin , c. kuklewicz , s. robertson , s. hill , f. knig , u. leonhardt , `` fiber optical analog of the event horizon , '' science * 319 , * 13671370 ( 2008 ) .
l. gagnon , p. a. blanger , `` soliton self
frequency shift versus galilean like symmetry , '' 466468 ( 1990 ) . measured data provided by crystal fibre a / s , denmark .
f. x. krtner , _ few - cycle laser pulse generation and its applications : vol 95 _ ( springer , 2004 ) .
j. m. dudley , l. provino , n. grossard , h. mailotte , r. s. windeler , b. j. eggleton , s. coen `` supercontinuum generation in air - silica microstructured fibers with nanosecond and femtosecond pulse pumping , '' 765771 ( 2002 ) .
j. herrmann , u. griebner , n. zhavoronkov , a. husakou , d. nickel , j. c. knight , w. j. wadsworth , p. st .
j. russell and g. korn , `` experimental evidence for supercontinuum generation by fission of higher - order solitons in photonic fibers , '' 173901 ( 2002 ) .
k. m. hilligse , t. v. andersen , h. n. paulsen , c. k. nielsen , k. mlmer , s. keiding , r. kristiansen , k. p. hansen , and j. j. larson , `` supercontinuum generation in a photonic crystal fiber with two zero dispersion wavelengths , '' 10451054 ( 2004 ) .
g. genty , m. lehtonen , h. ludvigsen , and m. kaivola `` enhanced bandwidth of supercontinuum generated in microstructured fibers , '' 34713480 ( 2004 ) .
m. h. frosz , p. falk , and o. bang `` the role of the second zero - dispersion wavelength in generation of supercontinua and bright - bright soliton - pairs across the zero - dispersion wavelength , '' 61816192 ( 2005 ) . c. cheng ,
x. wang , z. fang , b. shen , `` nonlinear copropagation of two optical pulses of different frequencies in photonic crystal fiber , '' 291294 ( 2005 ) .
d. v. skryabin , f. luan , j. c. knight , p. st . j. russell , `` soliton self
frequency shift cancellation in photonic crystal fibers , '' science * 301 , * 17051708 ( 2003 ) .
the applied scaling factor was as follows : fig .
[ low power shg frog ] left:3 , right:2 ; fig .
[ high power shg frog ] left:3 , right:2 .
a. v. yulin , d. v. skryabin , and p. st .
russell `` four - wave mixing of linear waves and solitons in fibers with higher - order dispersion , '' 24112413 ( 2004 ) . m. l. v. tse , p. horak , f. poletti , n. g. r. broderick , j. h. v. price , j. r. hayes , and d. j. richardson `` supercontinuum generation at 1.06@xmath1 m in holey fibers with dispersion flattened profiles , '' 44454451 ( 2006 ) |
experiments on rapidly rotating bose - einstein condensates ( bec ) in harmonic magnetic traps have observed triangular lattices of vortices @xcite . the lattice is expected to melt at sufficiently rapid rotation rates , @xmath0 , close to the radial trap frequency , @xmath1 @xcite , at sufficiently high energy temperature @xcite . in the rapidly rotating regime ,
the condensate wave function is a linear superposition of single particle states of the lowest landau level ( lll ) in the coriolis force ; quantum fluctuations are predicted @xcite to melt the lattice when the number of vortices , @xmath2 , is so large that the number of particles per vortex ( or filling factor ) , @xmath3 , falls to @xmath4 6 - 10 . while quantum fluctuations increase with decreasing @xmath3 , thermal fluctuations at finite temperature increase more rapidly .
the ratio of the mean square vortex displacements due to thermal fluctuations to those due to quantum fluctuations is @xmath5 @xcite . under typical experimental conditions ,
melting of the vortex lattice induced by thermal fluctuations is more likely to be observed before melting due to quantum fluctuations . in this paper
, we generalize the halperin and nelson ( hn ) theory of dislocation - mediated melting @xcite to apply to the thermal melting of vortex lattices .
the hn theory describes thermal melting of lattices as a kosterlitz - thouless transition @xcite due to unbinding of pairs of elastic dislocations in the lattice .
the complication in generalizing the theory is that the forces between vortices in neutral condensates are long ranged , unlike the forces in a crystal lattice .
the hn theory has been applied to vortex lattices in type - ii superconductors @xcite , where the forces between vortices are short ranged .
it has also been discussed in superfluid helium films @xcite under the very limiting assumption that the lattice is incompressible .
we consider a dilute gas of atoms trapped in an external potential and rotating about the z - axis , taking a zero range s - wave interacation between particles with coupling constant , @xmath6 , where @xmath7 is the s - wave scattering length for two bosons ; we work in the usual mean field approximation for the interaction energy .
furthermore , we describe the elastic behavior of a vortex lattice in a rapidly rotating bec using the coarse - grained , long wavelength elastohydrodynamics description @xcite . in lieu of treating the vortices as discrete entities ,
we express the deviation of the vortices from their equilibrium positions ( the triangular lattice sites ) as a continuous displacement field , @xmath8 .
although the condensates are three dimensional , the dominant motion is in the rotational plane , to which we confine our attention .
the equations of motion couple the local ( coarse - grained ) velocity field , @xmath9 , the local ( coarse - grained ) particle density , and the vortex displacement field . in refs .
@xcite , we used the elastohydrodynamic equations of motion to derive the modes and various correlation functions of system . here , we use the time - independent form of the same equations to understand the elastic behavior and thermal melting of a superfluid vortex lattice .
dislocations are expected to form through nonlinear instabilities of vortex lattice excitations , i.e. , tkachenko modes .
however , to determine the melting , we need consider only the thermodynamics of dislocations . in doing so
, we separate the elastic displacement field into two parts : a fluctuating part from the tkachenko modes and a singular part from the dislocations . in the absence of nonlinearities
these two terms are independent . prior to deriving the thermal melting of a superfluid vortex lattice , we first review the elastic equations for a superfluid vortex lattice in sec . [
sec : equ ] . then in sec .
[ sec : dis ] , we find the elastic displacement field and energy of a single dislocation , and the interaction energy of a dislocation pair . in sec .
[ sec : melt ] , we generalize the hn theory of two dimensional melting to a superfluid vortex lattice . finally , in sec .
[ sec : inhomo ] , we consider melting in an inhomogeneous system .
in this section we discuss the elastic properties of the vortex lattice , working in the frame in corotating with the lattice . considering only motion in the transverse plane , we write the energy as a function of the two - dimensional density @xmath10 , @xmath9 and the vortex displacement field , @xmath8 as @xcite @xmath11d^2r.\end{aligned}\ ] ] the first term in eq .
( [ eden ] ) is the kinetic energy of the superfluid surrounding the vortex lattice , in the second @xmath12 is the trapping potential , and third term is the centrifugal potential ; the fourth term is the interaction energy of the bec , with @xmath13 the usual two - dimensional coupling constant ( @xmath14 , where @xmath15 is the axial thickness of the condensate ) ; the final term is the elastic energy density . to second order in @xmath16 , @xmath17 +
\gamma n_{\rm v } { \mbox{\boldmath $ \epsilon$ } } \cdot{\mbox{\boldmath $ \nabla$ } } n.\end{aligned}\ ] ] the first two terms in eq .
( [ eel ] ) are the compression and shear energies , respectively , of the vortex lattice ; @xmath18 and @xmath19 are the ( two - dimensional ) density - dependent compressional and shear moduli ( essentially the usual three dimensional constants @xcite integrated over the thickness of the system in the axial direction ) @xcite . in the incompressible limit , e.g. , in liquid helium , @xmath20 @xcite , while in the lll limit ( @xmath21 ) , @xmath22 @xcite , and @xmath18 is expected to be small @xcite .
the final term , in which @xmath23 is the equilibrium vortex density , is the coupling of the vortex displacements to density inhomogeneities , with @xmath24 an @xmath0-dependent constant @xcite . in the incompressible limit , @xmath25 , where @xmath26 is the vortex ( wigner - seitz ) cell radius ,
@xmath27 , and @xmath28 is the vortex core radius @xcite ; in the lll limit , @xmath29 @xcite .
we first calculate the response of the vortex lattice to a stationary stress to linear order , assuming that the system is effectively two dimensional with motions only in the plane perpendicular to the rotational axis . to linear order , thermal and quantum fluctuations of the vortex positions do not influence the elastic response .
the velocity field entering eq .
( [ eden ] ) is governed , in a stationary configuration , by the displacement field of the vortices .
conservation of circulation implies that a shift of the positions of the vortices from their equilibrium positions alters the local velocity of the fluid according to @xcite : @xmath30 where @xmath31 is the non - singular superfluid phase .
the curl of this equation , @xmath32 , relates the transverse component of the velocity field to the longitudinal component of the displacement field .
the steady state velocity field created by a distortion of the vortex lattice obeys @xmath33 . for a homogeneous system , @xmath34 ,
i.e. , the longitudinal component of the velocity vanishes .
variation of eq .
( [ eden ] ) with respect the particle density and displacement field gives two first order equations of motion , @xmath35 where @xmath36 is the chemical potential ; and @xmath37 where @xmath38 is the elastic stress @xcite , and @xmath39 is the elastic stress due to density inhomogeneity .
( unlike in @xcite we separate out the @xmath40 term explicitly in defining @xmath41 . ) in general the elastic constants , dependent on the density , appear within the derivatives .
we first consider a homogeneous system , and return to the problem of an inhomogeneous system in sec .
[ sec : inhomo ] . the fluid velocity induced by a displacement of the vortices results in a coriolis force which balances the elastic and external stresses , @xmath42 ( including @xmath43 ) : @xmath44 the curl and divergence of eq.([main1 ] )
yield the transverse and longitudinal elastic responses , @xmath45 and @xmath46 , independently .
since @xmath33 , @xmath47 and @xmath48\epsilon_{\rm l } = \zeta_{\rm l}.\ ] ] fourier transforming eqs .
( [ tterm ] ) and ( [ lterm ] ) , we find the elastic green s functions , @xmath49 @xcite , @xmath50 @xmath51 and @xmath52 transverse stresses give rise only to transverse displacements , and longitudinal stresses only to longitudinal displacements . in terms of the elastic green s functions , the dynamic equations for displacement field are @xcite , @xmath53 unlike in conventional elastic systems , the longitudinal elastic green s function contains an additional length scale , an elastic penetration depth " : @xmath54 note that @xmath55 while @xmath18 is generally negative @xcite , and thus @xmath56 can be positive or negative . in the incompressible regime @xcite , @xmath57 , while in the lll regime , @xmath58 , where @xmath59 is the healing length in the condensate . in the lll regime , @xmath60 , and hence in both regimes @xmath61 . thus within the regime of validity of the coarse - grained continuum approximation , @xmath62 , and @xmath63 . the elastic displacement due to the ( longitudinal ) driving force @xmath39 is thus @xmath64 consistent with the results of refs .
corrections to this result using the elastic green s functions for an inhomogeneous system are of order @xmath65 .
to describe dislocation - mediated melting of a superfluid vortex lattice , we first investigate the structure of elastic dislocations , which are topological defects in the vortex lattice .
such dislocations obey the condition , @xmath66 for any closed path around the dislocation , where @xmath67 is the burger s vector @xcite .
the minimum length of a burger s vector is the lattice spacing .
we first solve eqs .
( [ veldisrel ] ) and ( [ main1 ] ) for an elastic dislocation in a superfluid vortex lattice , leaving mathematical details for the appendix . for @xmath68 and @xmath69 ,
the displacement field is , @xmath70 \\
\nonumber \label{eybxd0 } \epsilon_y & = & \frac{b}{2\pi}\left[\frac12\ln\left(x^2+y^2\right ) + \frac{x^2}{r^2 } + \frac{c_2}{\omega^2mn}\frac{y^2-x^2}{(x^2+y^2)^2}\right ] .
\nonumber \\\end{aligned}\ ] ] this displacement field is transverse , with vanishing divergence away from the origin , as expected since the longitudinal green s function vanishes away from the dislocation .
in addition we find that the energy of a single dislocation is @xmath71\right],\ ] ] where @xmath72 are modified bessel functions of the second kind , and @xmath73 is the radius of the dislocation core ; @xmath74 , the lattice spacing .
we include the dislocation core energy in the chemical potential of the system . in the limit @xmath75 , the logarithmic term in eq .
( [ sdenergy ] ) is the dominant term .
to second order in the displacement field the interactions between dislocations are pairwise .
the energy of two separated dislocations at positions @xmath76 and @xmath77 , with burger s vectors @xmath78 , is , as derived in the appendix , @xmath79 , \nonumber\end{aligned}\ ] ] where @xmath80 is the two dimensional anti - symmetric tensor , the first term in eq.([interdise ] ) grows with increasing separation of the dislocation pairs and is therefore long ranged ; the final term decreases with the separation at a rate depending on @xmath81 . in the limit @xmath82 , where one recovers classical elasticity
, the final term extend to long distances ; in this limit , eq.([interdise ] ) reduces to the dislocation interaction energy of ref .
@xcite .
two dislocations with opposite burger s vectors are attracted with an interaction which grows logarithmically .
dislocations thus bind in pairs at low temperature .
dislocation - mediated melting is driven , as we discuss in the following section , by unbinding of such pairs .
the core radius , @xmath73 , the length scale at which the dislocation can no longer be described by a _ continuous _ displacement field , as well as the dislocation core energy , can only be computed from the local microscopic condensate wave function . in the incompressible limit ,
the energy of the vortex lattice depends entirely on the kinetic energy of the surrounding superfluid , and thus the core energy is linearly proportional to the density .
similarly , in the rotating frame in the lll limit , only the interaction energy changes with vortex position , and thus the core energy is proportional to the square of the density .
we approximate the dislocation core energy by , @xmath83 where @xmath84 and @xmath85 are positive functions of the rotation rate , with the approximate dependence , @xmath86 and @xmath87 .
since @xmath88 depends on the position of the vortices within the dislocation core , where the continuum limit is not valid , exact values of @xmath84 and @xmath85 must be calculated in terms of the local condensate wave function .
the dislocation - mediated melting of the vortex lattice is a kosterlitz - thouless transition @xcite arising from unbinding of pairs of elastic dislocations .
thermal fluctuations cause dislocations to enter large systems in pairs with opposite burger s vectors . below the melting temperature ,
pairs are bound ; pairs of burger s vectors of equal magnitude but opposite direction produce no long range elastic deformation , and are thus most stable than isolated dislocations .
if two opposite dislocations overlap , they mutually annihilate . above the melting temperature
, dislocations are unbound , and to a first approximation can be treated independently . in the thermodynamic limit ( @xmath75 ) ,
the energy ( [ sdenergy ] ) is minimized for a dislocation with the minimum magnitude of the burger s vector , the lattice spacing , @xmath89 , and thus @xmath90 plus a small constant .
[ for a triangular lattice , @xmath91 . ]
this logarithmic form allows us to apply the simple physical argument of kosterlitz and thouless to determine when dislocations are thermodynamically stable in the vortex lattice .
since there are @xmath92 possible locations for a single dislocation in the vortex lattice , the entropy of a single dislocation is @xmath93 .
hence , the helmholtz free energy per dislocation is @xmath94 thus for @xmath95 , dislocations ( of size @xmath96 ) are thermodynamically stable .
isolated dislocations remain in the system only above this temperature . in the absence of interactions between dislocation pairs
, @xmath97 would be the melting temperature . as the temperature increases to @xmath98
the mean separation between two dislocations within a pair diverges ; in general the mean separation of a pair @xmath99 is @xcite , @xmath100 for @xmath101 the mean separation is finite , while for @xmath102 , the logarithmic term in eq .
( [ interdise ] ) leads to a divergent separation length within a dislocation pair , indicating that the pair has become unbound . to obtain a more accurate description of melting requires considering interactions of dislocation pairs , which can be taken into account by the renormalization - by - decimation technique in refs .
this technique requires @xmath103 to be small .
we do not give the details here , but note that the net effect is to renormalize the shear modulus @xmath19 to a temperature dependent function , @xmath104 .
the shear modulus has an intrinsic dependence on the temperature , since it is proportional to the superfluid density .
the melting temperature is then @xcite @xmath105 previous estimates for @xmath106 for a lattice of particles interacting logarithmically ( e.g. , vortices ) have shown @xmath107 where @xmath108 @xcite .
the melting temperature depends only on the shear modulus .
it is instructive to compare the origin of this dependence in a system of vortices with the corresponding result in a traditional elastic system ( e.g. , an isotropic crystal ) @xcite .
the first two terms of eq .
( [ interdise ] ) have the same form for the interaction energy of dislocations in an elastic system .
the third term of eq .
( [ interdise ] ) , containing an additional length scale , @xmath81 , is particular to the vortex system ; this term is a short range interaction , @xmath109 . in smectic liquid crystals near the nematic - to - smectic - a transition
, the interaction energy of two dislocations also contains an additional length scale , a penetration depth ; when this length is small enough to screen the dislocations in the liquid crystal beyond their core , the dislocation pairs unbind at @xmath110 @xcite .
in such a liquid crystal , _ all _ the energy channels ( splay , twist , and bend ) are screened by the compression or stretching of the smectic layers . in the superfluid vortex lattice , however , only compressional effects are screened ( by the rigid body motion ) while the shearing is unaffected . in the vortex system the unbinding of the dislocations
is promoted by shearing alone . for temperatures approaching @xmath111 from below ,
the displacement correlation function is independent of the system parameters @xcite : @xmath112 in the solid phase , the vortex lattice exhibits algebraic long - range order .
however , above @xmath111 , the displacement correlations function decay roughly exponentially with distance , as one sees from renormalization group calculations @xcite .
extended two dimensional bose systems , e.g. , helium films , or atoms in optical lattices @xcite , undergo a ( conventional ) kosterlitz - thouless transition @xcite to a superfluid state at @xmath113 , where @xmath114 is the two - dimensional superfluid mass density .
when rotating at low temperature , such a system should contain a vortex lattice , which melts via a dislocation mediated kosterlitz - thouless transition at a temperature @xmath115 . using the incompressible shear modulus in @xmath111
one finds @xmath116 @xcite , where @xmath117 ; @xmath118 in @xmath119he films is measured to be @xmath120 @xcite .
thus @xmath121 ; renormalization of the shear modulus should reduce this result at most by a factor of @xmath122 @xcite .
cozzini et al .
find from numerical computation of the @xmath110 shear modulus vs. @xmath123 from the incompressible regime ( @xmath124 ) to the lll regime ( @xmath125 ) that with the exception of a small increase just above the incompressible regime , @xmath126 @xcite .
this result suggests that @xmath127 decreases with increasing @xmath128 .
preliminary calculations for the parameters of the @xmath129rb experiment of ref .
@xcite give @xmath130 @xcite , indicating @xmath131 prior to renormalization effects .
we expect thermal melting in a rapidly rotating bec in a finite inhomogeneous geometry also to be driven by formation of dislocations arising from thermal fluctuations .
dislocations have in fact been observed in the lattices in these systems ; see fig .
5a of ref @xcite , however , describing the structure and unbinding of dislocation pairs is much more difficult than in a homogeneous system .
on the one hand , the elastic constants , @xmath18 and @xmath19 , depend on the local density , and dislocation energies depend on position , while the elastic theory contains second order terms explicitly dependent on @xmath8 in addition to the usual terms dependent on derivatives of @xmath8 @xcite .
furthermore , in a finite system dislocations can be created not only in pairs in the bulk of the system , but an isolated dislocation can also be formed at the edge , which separates from its image dislocation and moves into the bulk .
such additional features must be taken into account numerically to determine the details of melting , a problem we leave for the future @xcite .
here we simply determine an upper bound on the melting temperature .
we assume that the maximum density , @xmath132 , of the bec occurs at the center of the trap , where a dislocation has maximum energy .
furthermore , this maximum energy is less than the energy of a dislocation in a condensate of the same size but with uniform density , @xmath132 .
as a dislocation approaches the edge of the trap , its energy goes to zero .
thus the dislocation energy in an inhomogeneous system is bounded above by @xmath133 where @xmath134 is the ( unrenormalized ) shear modulus at the center of the trap , and @xmath135 is the thomas - fermi radius of the rotating cloud . using an upper bound to the dislocation energy should yield an upper bound on the melting temperature , and thus eq .
( [ disthermo ] ) indicates that the temperature at which dislocations are thermodynamically stable is bounded above by @xmath136 .
in fact , dislocations should appear near the edge of the bec for temperatures below @xmath137 and creep towards the center as the @xmath138 approaches @xmath137 , while for @xmath139 , we expect dislocations to occur throughout the system .
melting generally occurs below the bose - einstein condensation temperature , @xmath140 , where @xmath141 and @xmath1 are the azimuthal and radial angular frequencies of the trap respectively .
we first estimate @xmath142 in the limit of a nearly incompressible fluid ( @xmath143 ) in trap geometry , and then consider the lll limit . in the former regime , @xmath144 , where @xmath145 the ( two dimensional ) superfluid mass density at the center of the trap .
the shear modulus depends on the superfluid mass density a function of temperature , which for a weakly interacting bose gas is approximately the product of the mass , the mean thickness , and the three dimensional condensate density , @xmath146 . at the center of the trap
the condensate density , coarse - grained over the vortices , is approximately homogeneous and does not feel the effect of the condensate edge ; at @xmath110 , @xmath147^\frac25.\ ] ] in writing eq .
( [ tf_den ] ) , we approximate the number of particles in the condensate by the total particle number @xmath148 ; @xmath149 is the radial trap length , and @xmath150 is the abrikosov parameter @xcite . at finite temperature , @xmath151 .
the thickness , @xmath15 , of the condensate at the center of the trap is approximately @xmath152 , where @xmath153 is the azimuthal trap length .
thus we find @xmath154 as the solution of @xmath155 for the experimental parameters of refs .
@xcite in @xmath129rb ( @xmath156 nm , @xmath157 @xmath36 m , @xmath158 @xmath36 m ) with @xmath159 particles and @xmath160 , we have @xmath161 . for @xmath162 this
ratio increases to @xmath163 .
if @xmath154 is only slightly less than unity , then the thermal melting of the vortex lattice will be hard to observe experimentally .
higher rotation rates help to lower the melting temperature , but in most experiments , the number particles in the system decreases too , making thermal melting harder to resolve . note that eq .
( [ inhomo - melt - cond ] ) predicts that @xmath154 decreases with increasing scattering length , allowing for the possibility of inducing thermal melting via a feshbach resonance . in the lll limit ,
the density at the center of the trap is @xcite , @xmath164 along the axis , the condensate particles are in the lowest harmonic oscillator state , so the thickness of the condensate at the center of the trap is approximately @xmath165 .
also at the center of the trap , the condensate is in a harmonic oscillator state along the axis and approximately homogeneous in the plane of rotation , so the density as a function of temperature is proportional to @xmath166 , which we take as the temperature dependence of @xmath19 in the lll regime .
the melting temperature is given by the solution of @xmath167 figure [ fig : rrphase ] shows schematically the phase diagram of a rapidly rotating bec .
the y - axis is the temperature divided by the condensation temperature , and the x - axis is the rotational rapidity , " @xmath168 @xcite . in this figure
we use the parameters of refs .
@xcite as above .
the curve separating the vortex lattice and vortex liquid phases is given by eq .
( [ inhomo - melt - cond ] ) at low rotation , ( [ tmlll ] ) at high rotation rates , and an interpolation between the two limits ; the two kinks in the curve are artifacts of using these two limits with the interpolation .
we assume that the melting leads directly to a vortex liquid , neglecting a possible hexatic phase intermediate between the vortex lattice and liquid which exhibits no long - range order of the vortex positions but maintains algebraic long - range order of the triangular orientation of the lattice @xcite . computing the phase boundary between such a hexatic phase and a vortex liquid requires microscopic details of the dislocations ( e.g. , core radius and energy ) beyond the scope of this paper .
the vortex lattice is predicted to melt via quantum fluctuations into a new strongly correlated regime .
details of this regime are discussed in refs .
@xcite . with the density eq .
( [ lllden ] ) and @xmath169 , we estimate the phase boundary for the strongly correlated system to be at @xmath170^{-\frac12}.\ ] ] in fig .
[ fig : rrphase ] , where we take a critical filling fraction , @xmath171 @xcite , the phase boundary is indicated by the vertical dotted line , at @xmath172 ( corresponding to @xmath173 ) , the phase diagram remains to be studied in detail , e.g , near the region where the vortex liquid phase boundary intercepts the strongly correlated regime , and the behavior at higher temperatures . at low rotation rates ( @xmath174 ) , the melting temperature may be too close to the condensation temperature to discern the melting of the vortex lattice experimentally .
higher rotation rates provide a better opportunity to observe melting since @xmath111 decreases with @xmath175 . however , at sufficiently high rotation rates ( e.g. , at @xmath176 in fig .
[ fig : rrphase ] ) , it is not clear how to experimentally distinguish thermal melting from quantum melting , since we do not have an adequate understanding of the melted states at finite temperature . ; the rotational rapidity " is defined as @xmath177 @xcite . to the right of the dotted line ,
the system at lower temperatures enters a strongly correlated regime .
, title="fig:",width=288 ]
the elasticity of a superfluid vortex lattice is determined by the energy of the vortex cells and the global rotational fluid flow ; the kinetic energy of the rigid - body " rotational velocity stabilizes the system against a negative compression modulus , @xmath18 . because of the rotational motion , the effect of elastic compression is significantly weakened , and shearing is the dominant degree of freedom for the vortex lattice .
correspondingly , we find that the melting temperature of the vortex lattice depends only on the ( renormalized ) shear modulus
. among the unanswered questions for future examination are , first , to delineate the properties of the dislocation core , which will require extensive computation to unveil .
dislocation cores play a significant role in finite sized systems ; in experiments to date , core radii may be only an order of magnitude smaller than the system size .
also , one needs to take into account a possible transition to a hexatic phase , and then the transition of the hexatic to a vortex liquid .
the latter process involves the unbinding of the dislocations into disclination pairs , whose description requires the energy and size of the dislocation core @xcite .
dislocations in non - uniform condensates present further unsolved problems , analogous to those for vortices in inhomogeneous systems , for example , the dependence of the energy of a dislocation on position .
this dependence is needed to determine whether it is energetically preferable for dislocations to form at the edge , or by unbinding in the bulk of the condensate .
we thank markus holzmann for helpful discussions and joshua rubin .
this work was supported in part by nsf grants phy03 - 55014 , phy05 - 00914 , and phy07 - 01611 .
in this appendix we derive the velocity and displacement fields around an elastic dislocation in a homogeneous two - dimensional vortex lattice , in particular , for superfluid vortex lattices , a two - dimensional edge dislocation @xcite .
the challenge is to find a solution of eqs .
( [ veldisrel ] ) and ( [ main1 ] ) that satisfies eq.([burgerscond ] ) for the burger s vector .
we first recast eq.([burgerscond ] ) as @xmath178 where @xmath80 is the anti - symmetric tensor , and @xmath179 .
note that @xmath8 is not continuous within the entire plane ( i.e. , @xmath180 ) ; otherwise , eq.([burgerscond2 ] ) is not valid . to satisfy the conditions , @xmath34 and @xmath181 , we define two dual ( gauge ) fields , the stream function , @xmath182 , and a modified airy stress function , @xmath183 . using the elastic stress tensor , @xmath184 , defined from eq.([elforce ] )
such that @xmath185 , we write , @xmath186 substituting these expressions into the curl of eq.([veldisrel ] ) and eq.([burgerscond2 ] ) , we have , @xmath187 where @xmath188 is the position of the dislocation . taking @xmath189
, we find @xmath190\\ \label{psi } \psi({\mbox{\boldmath $ r$ } } ) & = & \varepsilon_{ij}b_j\frac{\partial}{\partial x_i}\left[\frac{2c_2\omega\delta^2}{(2c_1+c_2)\pi}\left(\ln{\vert { { \mbox{\boldmath $ r$}}}\vert } + k_0\!\left(\frac{{\vert { { \mbox{\boldmath $ r$}}}\vert}}{\delta}\right)\right)\right].\end{aligned}\ ] ] the airy stress function contains two basic terms . for finite @xmath81 , the first term in eq .
( [ chi ] ) is long ranged while the second term dies off .
the stream function is proportional to this second term . for imaginary @xmath81
we replace @xmath191 by @xmath192 ( where @xmath193 is a modified bessel function of the first kind ) since @xmath194 and @xmath195 are real .
[ for @xmath82 , we find the typical results for an edge dislocation in two dimensions ( see ref @xcite ) , however , this limit is not physically relevant here . ]
according to eq .
( [ burgerscond ] ) , the displacement field , @xmath8 , along any closed path around a dislocation does not return to its initial value .
instead , the displacement field at the close of the loop is equal to its initial value plus the burger s vector .
thus , the system includes a
branch cut " from the dislocation core to the edge of the system , as in fig .
[ fig : branchcut ] ; the displacement field for a dislocation is single valued .
( this structure is similar to that of the phase , @xmath196 , around a vortex , which changes from @xmath197 to @xmath198 ; jumping by @xmath198 across the branch cut . )
the energy of an elastic displacement in the superfluid vortex lattice is , @xmath199\right ] d^2r , \nonumber\end{aligned}\ ] ] where @xmath200 is the transverse flow velocity .
integrating by parts and using @xmath201 , from eq .
( [ veldisrel ] ) , we have @xmath202d\sigma_j \nonumber \\ & + & \int { \mbox{\boldmath $ \epsilon$}}\cdot\left[-2c_1{\mbox{\boldmath $ \nabla$}}({\mbox{\boldmath $ \nabla$}}\cdot{\mbox{\boldmath $ \epsilon$ } } ) - c_2\nabla^2{\mbox{\boldmath $ \epsilon$ } } + nm{{{\mbox{\boldmath $ \omega$}}}\!\times\!{{\mbox{\boldmath $ v$}}}}\right]d^2r\nonumber,\end{aligned}\ ] ] where @xmath203 is the normal component of the surface element of the section of the boundary ( including the branch cut ) . in the absence of external forces , eq.([main1 ] ) implies that the integrand in the second integral in eq.([energystuff ] ) is zero .
the integrals at the boundary of the system and around the core are negligible ; the branch cut contribution is dominant . since @xmath204 on the lower part of the branch cut , and @xmath205 on the upper part ( fig .
[ fig : branchcut ] ) , we find finally @xmath206d\sigma_j^-,\end{aligned}\ ] ] where the integral is along the lower branch .
computing this expression using eqs .
( [ streamf ] ) and ( [ airyf ] ) , relating the derivatives of @xmath8 to @xmath184 @xcite , and ( [ chi ] ) and ( [ psi ] ) , we obtain eq .
( [ sdenergy ] ) .
+ to obtain a more realistic description of the system , requires taking into account the interactions of dislocations . to compute the energy of a two - dimensional gas of interacting dislocations we first substitute eq.([streamf ] ) and eq.([airyf ] ) into eq.([vlenergy ] ) by relating the derivatives of @xmath8 to @xmath184 @xcite , and find , @xmath207 d^2r.\end{aligned}\ ] ] we consider an aggregate of dislocations with burger s vectors , @xmath208 , such that the sum over all burger s vectors is zero and @xmath194 and its derivatives are zero at infinity unlike a single dislocation , a pair of dislocations with equal and opposite burger s vectors is not long ranged .
integrating eq.([elasticenergy2 ] ) by parts we obtain , @xmath209 and substituting in eq.([dualdis ] ) and eq.([chi ] ) , and integrating , we derive eq .
( [ interdise ] ) .
tkachenko , zh . esp .
. fiz . * 49 * , 1875 ( 1965 ) [ sov .
jetp * 22 * , 1282 ( 1966 ) ] ; zh . esp .
. fiz . * 50 * , 1573 ( 1966 ) [ sov .
jetp * 23 * , 1049 ( 1966 ) ] ; zh . esp .
. fiz . * 56 * , 1763 ( 1969 ) [ sov .
jetp * 29 * , 245 ( 1969 ) ] . since @xmath216 in ( and near ) the incompressible limit
, @xmath217 changes sign for @xmath218 , which according to eq .
( [ gfe ] ) should be accompanied by a resonant @xmath219 becoming negative .
this change of sign does not imply an instability of the lattice , since it occurs for wavelengths @xmath220 the vortex lattice spacing , a length scale at which the elastohydrodynamic description is no longer valid . in general ,
a conventional elastic system is unstable if @xmath221 @xcite ; however , in the vortex case the kinetic energy term in eq .
( [ eden ] ) guarantees the stability of the lattice . for @xmath222
, the elastic green s function of the vortex lattice decays exponentially , while , if @xmath223 , then the penetration depth is imaginary , suggesting ripples in the displacement field . in the regime where @xmath56 is negative ,
the quasi - wavelengths of any undulations in the displacement are @xmath224 .
the dependence of @xmath111 on @xmath81 is negligible in the region @xmath225 . if one assumes incompressibility , @xmath226 , as in the derivation of @xmath111 in ref .
@xcite , for which @xmath81 becomes effectively infinite , the melting temperature is given by the limit of @xmath227 , where the @xmath228 s are the renormalized elastic moduli .
however , this limit is unphysical , since the tkachenko mode frequencies are finite only if @xmath229 is finite . in sec .
[ sec : equ ] , we separated out the elastic force due to an inhomogeneous density , @xmath39 , and treated it as an external stress .
this force only affects the displacement field of the vortex lattice , according to eq .
( [ rad - dis ] ) , and does not influence the displacement field of the dislocation .
however , the coupling between the particle density and the vortex density contributes a term to the dislocation energy , @xmath230 using eqs .
( [ veldisrel ] ) , ( [ streamf ] ) , and ( [ psi ] ) , we find for a vortex at @xmath231 , @xmath232 where @xmath233 . but because @xmath234 , this term is small and does not contribute significantly to the overall energy . | we describe thermal melting of the two - dimensional vortex lattice in a rotating superfluid by generalizing the halperin and nelson theory of dislocation - mediated melting . and derive a melting temperature proportional to the renormalized shear modulus of the vortex lattice .
the rigid - body rotation of the superfluid attenuates the effects of lattice compression on the energy of dislocations and hence the melting temperature , while not affecting the shearing . finally , we discuss dislocations and thermal melting in inhomogeneous rapidly rotating bose - einstein condensates ; we delineate a phase diagram in the temperature rotation rate plane , and infer that the thermal melting temperature should lie below the bose - einstein transition temperature . |
the study of stellar winds is crucial for our understanding of the evolution of massive stars and their interactions with their environment .
in fact , mass - loss plays a fundamental role in the evolution of massive stars ( e.g. @xcite ) as well as in the mechanical and chemical feedback of these stars into the interstellar medium ( e.g. @xcite ) . over the last decade ,
our view on the mass - loss of massive stars has changed as the importance of wind fragmentation became clear .
direct observational evidence for the existence of clumps in the winds of o - type stars was obtained from the low - level , rapid variability of the @xmath04686 emission line in the spectrum of @xmath2puppis ( @xcite ) .
indirect evidence stems from the analysis of the spectra of o - type stars .
indeed , with observations from the @xmath3 satellite , it became clear that the modelling of lines of ions such as , and with model atmosphere codes such as cmfgen , is impossible if clumping is not included ( e.g. @xcite ) .
the existence of such small - scale wind structures has important consequences on many diagnostics of mass - loss rates .
indeed , strong clumping often implies a substantial reduction , by factors between a few and about ten , of the overall mass - loss rate compared to a homogeneous stellar wind . in this context
, our ignorance of the properties of these wind structures translates into uncomfortably large uncertainties on the actual mass - loss rates .
for instance , whilst the space between clumps is generally supposed to be totally empty , some recent studies @xcite showed that it is necessary to include the effects of a tenuous inter - clump medium to correctly predict the strength of various lines in the observed uv spectra of o - type stars . by doing so
, these authors even found that the observed profiles might again be compatible with the conventional mass - loss rates .
mass - loss rate diagnostics in different wavelength domains are strongly sensitive to different aspects of the wind fragmentation . for instance , h@xmath4 and radio emission are collisional processes which scale with density squared and are thus sensitive to the clumping factor regardless of the size of the clumps . in clumped winds with a void inter - clump medium , the velocity law is intermittent and some velocities are not present along a given sight line .
this porosity in the velocity field is called vorosity @xcite .
the latter strongly modifies the sobolev length and its impact is mainly visible on the uv resonance lines . the present paper deals with the impact of wind structure on the morphology of x - ray spectra of massive stars .
high - resolution @xmath5 and @xmath6 x - ray spectra of presumably single o - type stars revealed less absorption of the x - rays by the cool wind than expected for a homogeneous wind with conventional mass - loss rates .
this property can be explained in the case of fragmented stellar winds ( e.g. @xcite ; @xcite ; @xcite ; see also @xcite for a review ) , although @xcite argue in favour of an overall reduction of the mass - loss rate of the wind without the need to include porosity effects . in a fragmented wind with optically thick clumps ,
x - ray lines are sensitive to porosity , which depends on clump continuum optical depth , and thus clump size as well as the clumping factor .
that is why high - resolution x - ray spectra are important tools to study the structure of the wind and help us determine mass loss rates which must be consistent with the other wavelength domain analyses .
previous works in the x - ray domain ( e.g. @xcite ) , were performed on a line - by - line basis ; the spectra were never analysed as a whole at once . using multi - temperature plasma models in future work
, the global fit of observations will give information concerning the different shock temperatures as well as the size and the position of the different x - ray emitting shells . with this in mind ,
we have designed a new code for the purpose of simulating the full high - resolution x - ray spectrum of a single massive star .
our method properly accounts for the effect of line blending on the line profiles , and allows testing the consistency of current models for the x - ray emission of massive stars over the full range of the x - ray domain .
therefore , our code is meant as a further step towards extracting the maximum amount of information from the high - resolution x - ray spectra . in this paper
, we introduce the different free parameters and investigate their impact on single temperature synthetic x - ray spectra .
we further discuss the main assumptions and physics in our code and compare different formulations of the wind fragmentation .
in particular , we investigate for the first time the impact of a tenuous , but non - void , inter - clump medium on the synthetic x - ray spectra .
finally , we present the first application of our code to fit the region around the @xmath0 18.97@xmath1 line in the @xmath7 spectrum of @xmath2puppis and illustrate the degeneracies on the stellar wind parameters that exist when a single line is fitted . in future work , we will apply this code to the full x - ray band spectra of @xmath2puppis .
before using a multi temperature plasma model to analyse the x - ray spectra of o - type stars in detail , we need to calculate single temperature synthetic spectra with the most recent stellar wind theories ( emission and absorption ) and with as few as possible free parameters .
we decide to follow the wind embedded shocks scenario .
consequently , we need to know where the shocks arise in the wind and what is the temperature of the resulting post - shock hot plasma . using up - to - date emissivities
, we can then determine the x - ray emission of a hot plasma as a function of temperature , abundances and quantity of matter present in the shocks .
next , we determine how the cool wind material absorbs the x - ray photons and how the wind structure ( i.e homogeneous or porous ) impacts on the emergent x - ray flux . in the following subsections ,
we describe the tools and assumptions considered to build our code and to calculate synthetic spectra .
previous works on x - ray line profile fitting have always normalized the line profiles , i.e the overall line flux was never physically modelled .
we attempt a different approach here , by using the true x - ray emissivity of an optically thin plasma . for this purpose , we use atomdb which models the emission from a collisionally ionized , optically thin isothermal plasma in thermal equilibrium . many processes are included in the model : collisional excitation , ionization , and dielectronic recombination drive the ionization balance and emission spectra from the plasma .
continuum processes such as two - photon decay and bremsstrahlung are also included .
photon - driven processes such as photo - excitation and photo - ionization are not included in the model .
the assumption of an optically thin plasma requires a low density ( @xmath8@xmath9 ) , therefore three body effects are not included .
the standard atomdb model data has been recalculated on a finer 0.025 kev / step electron temperature grid for this analysis and we use an energy resolution of 0.2 ev .
in addition to the strongest lines that dominate the spectrum , the high - resolution theoretical spectra reveal the presence of numerous weak lines and also discontinuities in the continuum ( fig . [ figpre2 ] ) . their contribution to the emergent flux
is non - null and it is necessary to take them into account for the computation of emergent spectra as they certainly modify the emissivities and profiles of the strongest observed lines .
when fitting actual observations , one further needs to account for the presence of plasma at different temperatures , located at different places in the wind . therefore the normalization of the model will result from the combination of several model spectra weighted according to their relative emission measure .
this will be dealt with in our forthcoming paper on model fits of @xmath2puppis spectrum.in addition , in the x - ray emitting plasma , some ions are present in a helium - like form .
the treatment of the transition between the @xmath10 and @xmath11 levels must be done separately .
indeed , the presence of a strong uv radiation modifies the populations of the metastable upper level of the forbidden line via a pumping to the upper level of the inter - combination line ( @xcite ; @xcite ) . we will discuss this fundamental effect in our more complete analysis of @xmath2puppis presented in a forthcoming paper ( herv et al .
, in preparation ) .
we consider the hot plasma to be embedded in the cool wind and to move outwards along with the latter .
we decompose the emitting volume into small cells where we calculate the x - ray emission of each cell and its shift in wavelength in the observer s frame of reference due to the overall wind motion .
the x - ray emitting plasma is assumed to be optically thin to x - rays .
an x - ray photon emitted by the hot plasma , either escapes from the wind or is absorbed by the cool wind material .
thus the radiative transfer is reduced to the calculation of @xmath12 , the optical depth along the line - of - sight due to the cool wind material . following the approach of @xcite , the absorption by the stellar wind
is essentially characterized by the @xmath13 parameter and the mass - absorption coefficient @xmath14 , which are linked by the relation @xmath15 .
these absorption coefficients are dependent on the wavelength and the abundances of the different elements .
to evaluate the mass - absorption coefficients , we use the radiative transfer code cmfgen ( hillier @xmath16 miller 1998 ) which computes the ionization structure of the cool wind as a function of temperature , abundances of many elements and mass - loss rate . in the ionization calculations , cmfgen accounts for different physical processes such as collisional excitation , photo - ionisation by photospheric light and x - rays , and radiative and dielectronic recombination . from the known ionization structure
, we can then compute @xmath14 . in the x - ray domain ,
the opacity is dominated by the photo - ionization of the metals .
this absorption is by far the dominant source of opacity and we can safely neglect other physical processes such as free electron scattering . the wavelength dependence of @xmath14 ( fig .
[ figpre1 ] ) and therefore @xmath17 , is quite sensitive to the model elemental abundaces and to the ionization fraction of singly ionized helium . in this respect ,
we stress that the large values of the mass absorption coefficient , used in the present paper , reflect the fact that the sum of the cno abundances in the best - fit model of the uv and optical spectrum of @xmath2puppis is supersolar ( bouret et al . , in preparation ) .
the detailed behaviour of @xmath14 as a function of radius and wavelength ( especially at longer wavelengths ) also depends upon the ionization structure of the wind and more specifically on the details of the recombination of he@xmath18 into he@xmath19 @xcite .
however , for simplicity , we assume here that @xmath14 is independent of the position in the wind out of the optical depth integral ( eq.[eqn3 ] ) and thus to achieve an analytical evaluation of this integral . ] and we evaluate its value at @xmath20r@xmath21 . in our detailed analysis of the rgs spectrum of @xmath2puppis ( herv et al .
, in preparation ) , we will drop this assumption and rather introduce a radial dependence of @xmath14 .
finally , the values of the mass - absorption coefficient or the wind optical depth as a function of the wavelength are used in eq.[eqn3 ] and [ eqn4 ] ( see sect.[fragment ] below ) to determine the absorption of x - rays by the stellar wind .
previous studies of stellar winds revealed that they are not homogeneous but fragmented .
the absorption of x - rays by a clumpy wind has been addressed by several authors who adopted slightly different formalisms for the same phenomenon .
the _ porosity length prescription _ is based on the concept of the mean free path of photons between two successive interactions with clumps , and the opacity of the latter @xcite . in the _ fragmentation frequency prescription _ , the opacity of the wind in the x - ray band is estimated by the frequency of clumps passing through a reference radius @xcite . in this section
, we briefly summarize the main aspects of these models . at the microscopic level
, the absorption of x - rays by a stellar wind is mainly due to the bound - free transitions from the k and l energy levels .
to start , we assume that the clumps are spherical and very small .
we consider clumps over a wide range of optical depth , from optically thin to optically thick . in a situation where the individual clumps are optically thick ,
the effective opacity is essentially determined by the geometrical properties of the clumps @xcite and can be written @xmath22 where @xmath14 is the mass - absorption coefficient , @xmath23 is the geometrical scale of the clump , @xmath24 its mass and @xmath25 its optical depth .
this latter can be written as @xcite : @xmath26 where @xmath27 is the mean density of the stellar wind , @xmath28 the volume filling factor , @xmath29 is the mass - loss rate of the star and @xmath30 the velocity of the wind at the radius @xmath31 which is determined by a @xmath32-law ( i.e @xmath33 with @xmath34 throughout this work ) .
+ under this assumption , it appears that the effective reduction depends on the ratio between the clump scale and filling factor , also called the porosity length @xcite .
the concept of effective opacity can be extended to the optically thin limit using a simple bridging law @xcite : @xmath35 using this result in a steady state ( i.e. non variable ) wind , one obtains the wind optical depth in conventional @xmath36 coordinates @xcite : @xmath37 where @xmath38 , @xmath39 is the optical depth parameter of the wind , @xmath40 is the terminal velocity of the wind , @xmath41 is the radius of the star and @xmath42 the porosity length . in the following we adopt @xmath43 .
the fragmentation of stellar wind shells is a consequence of hydrodynamical instabilities . rather than being spherically symmetric , the resulting clumps
could actually have a flattened , elongated shape . using the number @xmath44 of fragments passing through some reference radius per unit time ,
also called the _ fragmentation frequency _ , and the opacity of each fragment at this reference radius , @xcite showed that the wind opacity can be written in the general case as : @xmath45 where @xmath46 is the optical depth of a flattened fragment @xmath47 along the line - of - sight and @xmath48 represents its orientation .
the optical depth along the line - of - sight : @xmath49 with @xmath14 the mass - absorption coefficient and @xmath50 the average radial optical depth of a fragment located at distance @xmath31 @xcite . in the case of an isotropic opacity ( i.e. small - scale spherically symmetric clumps )
, @xmath51 is taken equal to unity ( i.e. the line - of - sight crosses each clump along a radial direction ) . in section 3.1
, we compare the profiles obtained with @xmath12 given by eq.[eqn4 ] to those computed with eq.[eqn3 ] , first in the case of an isotropic opacity where @xmath51 is taken equal to one , then in a case of an anisotropic opacity where the clumps are assumed to be flattened ( i.e @xmath52 ) .
as we will show below , the porosity length prescription and the fragmentation frequency prescription are essentially equivalent to first order , although in their original work , @xcite concentrated on spherical clumps , whilst @xcite accounted for flattened structures by introducing the line - of - sight orientation via the @xmath51 parameter . ] .
actually , the geometry assumed by @xcite is even more idealized than a pure spherical geometry since it assumes that the line - of - sight always crosses the clump along the radial direction . strictly speaking
, this is valid only if the clumps are very small .
the main differences between the two approaches hence concern the geometry of the wind fragments and the way the optical depth of a clump is included in the model , i.e. eq.[eqn2 ] for the porosity prescription , and the @xmath53 term in eq.[eqn4 ] for the fragmentation frequency . for small optical depths and isotropic clumps ( @xmath54 ) , both models should be essentially equivalent . in a first step ,
we reproduce the profile of various lines ( table[tab1 ] ) for each prescription and compare them with each other and with the results for a homogeneous wind model .
we consider a distribution of spherically symmetric clumps and we thus set @xmath54 in eq.[eqn4 ] and [ eqn5 ] .
our general results agree well with those of previous studies .
first , the larger the porosity of the wind , the larger the mean free path of a photon in the stellar wind and the lower the absorption @xcite .
second , the larger the fragmentation frequency , the lower the opacity in the fragmentation prescription , see @xcite .
we have simulated four specific lines with @xmath55 @xmath56s@xmath57 adopted from @xcite .
this parameter is found by the authors to best fit the @xmath0 18.97 @xmath1 line in the hetg and rgs spectra of @xmath2puppis .
then we derive the porosity length , @xmath58 , needed to reproduce the line profiles computed with this fragmentation frequency ( fig .
[ fig1 ] ) . for different values of @xmath13 , hence different values of wavelength ( see table 1 ) , the two line profiles are , in each case , very close .
note however that we obtain different values for the porosity length when matching different lines computed with the same fragmentation frequency : we have to decrease the porosity parameter when @xmath59 increases .
in other words , when we compare simulated global spectra using a single porosity parameter to those obtained for a specific value of the fragmentation frequency , we find that the spectra are exactly the same in the wavelength band near the line used to find the parameter @xmath58 that best matches the value of @xmath60 ( in our case the @xmath0 18.9 @xmath1 line ) , but differ over the other wavelength domains ( fig.[fig2 ] ) . more specifically , the porosity model predicts a slightly lower flux in the short wavelength domain and a higher flux in the long wavelength band
. there are some underlying assumptions in both models , and with current data it seems difficult to tell which prescription provides the better description of reality .
the porosity prescription has the advantage to provide an analytical form of the opacity that speeds up the computation .
therefore , in the following , we concentrate mainly on the porosity prescription .
the overall results and conclusions obtained in the study for the different parameters , in the next section , are the same for the fragmentation model in an isotropic ( @xmath61 ) case .
for each simulation in this section , we use a terminal velocity @xmath40 of 2250kms@xmath57 .
we further assume a @xmath62 of 0.217kev , solar abundances for the chemical composition of the plasma , a stellar radius @xmath41 of 18.6@xmath63 and a mass - loss rate of 4.2 @xmath64yr@xmath57 .
these stellar and wind parameters correspond to those used by oskinova et al ( 2006 ) in their work to fit lines of @xmath2puppis , except for the plasma temperature which corresponds to the closest value in atomdb to the result of @xcite . in the remainder of this section
, we simulate a series of spectra for such a single o star in order to illustrate the sensitivity of the synthetic spectra on the different model parameters .
the first explored parameter is the porosity length .
the impact of this parameter for broad band spectra was never presented , in the literature .
fig.[fig5 ] illustrates our results .
we have chosen to explore a wide range of values of the parameter @xmath58 .
as expected , we find a strong impact on the emergent flux . at the shortest wavelengths , where @xmath14 is lowest
, the porosity length has little impact on the spectrum .
however , the situation is radically different at longer wavelengths , where the wind opacity @xmath14 becomes huge ( fig.[tau0 ] , tab.[tab1 ] ) . here
, the impact of the porosity length is much more important .
we find differences of several orders of magnitude between the spectrum for a homogeneous wind and the simulations with the largest porosity .
for a highly porous wind , the variations of @xmath58 have less impact as the wind is already so fragmented that the photons have few remaining interactions with the clumps . as a second step ,
we consider the effect of the shape of fragmented shells .
for this purpose , the factor @xmath51 in eq.[eqn4 ] and [ eqn5 ] , which represents the orientation of the fragmented shells with respect to the line - of - sight , is re - introduced into our simulations .
compared to the isotropic case , the opacity of the wind decreases ( see fig.[fig3 ] ) .
the radiation from the back side of the star is less attenuated than in the case of isotropic clumps .
indeed , the fact that fragments are flattened and that photons coming from the rear side of the wind cross these shells mainly perpendicularly implies that they have to cross less matter , and are thus less absorbed , than in the isotropic case . as a result ,
the total flux and intensity are higher in the anisotropic case . also , the profiles are less asymmetrical and the full width at half maximum increases . the consequences are more intense spectra ( by a factor 1.2 in the short wavelengths and 3.0 in the longer wavelengths domain ) and an increased overlap of lines in the anisotropic model ( see fig.[fig4 ] ) .
another parameter of our model is the size of the x - ray emitting plasma ( @xmath65 ) .
as the emissivity is proportional to @xmath66 , the outer parts of the wind contribute less to the total intrinsic ( i.e. before wind absorption is accounted for ) x - ray emission , but can not be neglected for the wind - absorbed emission as can be seen for specific line profiles in fig.[fig6 ] .
the effect is by far strongest in the long wavelength domain where @xmath67 is largest .
the intensity of the lines in fig.[fig6 ] varies by a factor between 1.1 and @xmath68 80 ( from shorter to longer wavelengths ) when the outer radius of the emitting volume is multiplied by a factor 2.0 .
moreover , extending the emitting region to larger distances implies that the velocity of the x - ray emitting plasma ( assumed to move along with the cool wind ) becomes larger if the outer radius is still located inside the wind acceleration zone .
this in turn increases the full width at half maximum of the lines as well as the line width at the continuum level .
the effect is best seen in the red wing of the line which is most strongly depressed by the wind absorption . with a larger emitting volume , a larger fraction of the redshifted photons from the rear outer parts of the wind
escape the wind since they cross layers of lower density .
the impact of the increase of the size of the emitting region on the overall spectrum can be seen in fig.[fig7 ] .
the location of the inner boundary of the emission region is one of the open questions in the understanding of the x - ray emission of massive stars .
however , a variation of this parameter can have important consequences . as we pointed out in the previous section , the density of matter increases as one moves inwards in the stellar wind , and therefore the contribution to the intrinsic x - ray emission increases as well .
still , as far as the observable emission is concerned , part of this increase in emission measure is compensated by the increase of the amount of absorbing material that the photons have to cross before they escape .
if we adopt an inner radius of 2.5@xmath69 instead of 1.5@xmath69 , with an outer radius of 10.@xmath69 , for the emitting plasma , we note that the line intensity in the short wavelength range , where @xmath17 is low , decreases roughly by a factor two . in the long wavelength domain , where @xmath17 is high ,
the reduction of the stellar wind absorption is more important than that of the emission . in consequence ,
the inner part of an x - ray emission shell is fully absorbed by the cool material .
the result is that the lines at longer wavelength are rather insensitive to the location of the inner boundary of the emission region ( fig.[fig8 ] ) .
the same qualitative considerations hold for the overall spectrum which is shown in fig.[fig9 ] .
we note that the fir triplet of helium - like ions provides in principle a powerful diagnostic of the inner emission boundary @xcite .
however , in the case where the observed x - ray emission stems from a multi - temperature plasma with each component possibly having its own value of the inner boundary , it must be stressed that the resultant triplets are the sum ( weighted by the emission measures ) of the triplets in the various components .
this situation renders the interpretation of the fir diagnostics more complex than expected from a single temperature plasma . in their work , @xcite have shown the need to introduce a homogeneous inter - clump medium to better fit the profile of the @xmath701032,1038 doublet with the model atmosphere code cmfgen .
these authors conclude that the inclusion of a small percentage @xmath71 of homogeneous material in between the clumps is needed to achieve a good fit of the line .
similar conclusions were reached by @xcite .
what is the consequence of such a tenuous , but homogeneous , inter - clump medium on the x - ray line profiles ? to answer this question , we incorporate this situation in our model . in practice
, we work with a constant mass - loss rate and we consider that the size of the clumps is small .
we distribute an important percentage of matter ( 90@xmath72 , 95@xmath72 or 97@xmath72 ) in clumps and we compute the optical depth due to a totally clumped wind . as a first assumption , we consider that the ionization is the same in the cool clump as in the inter - clump medium .
next , we compute the optical depth for an homogeneous wind with the remaining matter . finally we take the sum of the two contributions and determine the absorption for a mixed wind model .
as a homogeneous wind absorbs more photons , the intensity of the line profiles of the mixed model should actually be lower than for a fully clumped model ( see fig.[fig10 ] ) and the impact is stronger at longer wavelengths , and hence at larger opacities @xmath14 .
these considerations are confirmed by our simulations of the overall x - ray spectrum ( see fig.[fig11 ] ) . as a second step ,
we attempt to fit the spectrum of the mixed model with 97% of matter clumped to a model spectrum for a fully clumped wind .
fig.[fig12 ] illustrates the results : the individual line profiles of the mixed model are almost perfectly fitted by a fully clumped model with a slightly lower porosity parameter .
this has important consequences in practice : whilst the x - ray line profiles might allow us to infer the presence of clumps and to constrain their geometry , it seems almost impossible to distinguish between a fully clumped wind and a mixed model without constraints from observations at other wavelengths ( e.g. uv , optical).we thus conclude that x - ray line profiles alone can not reveal the presence of an inter - clump medium .
in the previous sections , we have discussed the impact of different parameters and of different fragmentation prescriptions on the line profiles and on the emergent x - ray spectra . in this section
, we use our code to fit the lyman @xmath73 doublet @xmath018.97 @xmath1 and the neighbouring @xmath0 19.36 @xmath1 , @xmath0 19.82 @xmath1 lines of the massive star @xmath2puppis and determine some stellar wind parameters .
we decided to work on an extended region around the strong line since fig.[figpre2 ] reveals the presence of weak lines that affect the wings of the line and produce a pseudo - continuum that contributes of order 5% of the flux of the line .
for this purpose , we have extracted , reduced and combined 18 rgs spectra of @xmath2puppis from the @xmath6 archive ( @xcite ; the different aspects of the data reduction are treated in detail in that paper ) .
we also apply the absorption by the interstellar medium ( @xmath74 , @xcite ) to the synthetic spectra .
then , we convolve the spectra with a lorentzian profile with a half - width at half - maximum of 0.07 @xmath1 to account for the instrumental resolution of the rgs spectrograph onboard @xmath6 . finally , we compare the results to the observation . as shown in fig.[fig13 ] and
[ oviiizoom ] , we obtain a degeneracy of the stellar wind parameters . indeed , for the line profile , we demonstrate that we can use a homogeneous or a porous wind model to fit the data equally well ( see fig.[fig13 ] and table 2 ) though with different values for the hot gas filling factors .
note that these values are in agreement with the results of @xcite .
we only need to adapt the mass - loss rate and the abundance of oxygen to obtain similar results . in a denser wind ( i.e. with a larger mass loss rate ) , the increase of absorption dominates the increase of x - ray photon production at longer wavelengths where the opacity is largest . in consequence
we have two ways to increase the number of x - ray photons which escape from a denser wind .
the first solution is to increase the number of x - rays resulting from shocks .
to first order , we could imagine that the number of shocks , and hence the hot gas filling factor , would be larger in a denser wind ( tab.[tab2 ] , models a and c ) .
alternatively , we can increase the number of photon which escape .
consequently , we have to increase the free mean path of photons in the stellar wind .
this can be achieved with a more porous wind ( tab.[tab2 ] , models a and d ) .
normally , the line profiles of the two previous solutions should be different .
unfortunately , the instrument resolution is not sufficient to distinguish the subtle differences .
furthermore , we can also fit this line equally well with a model where the inter - clump medium is not void ( fig.[fig14 ] , on the left ) .
it is thus obvious that a single line does not permit to discriminate between models .
these preliminary results therefore emphasize the necessity to work on the full x - ray spectrum .
the analysis of the full rgs spectrum is beyond the scope of the present paper , but will be presented in a future work ( herv et al .
, in preparation ) . as we demonstrated in sect . 3
, each parameter has a different impact over different parts of the spectrum and we are therefore confident that some degeneracies on e.g. porosity or mass - loss rate , will then be lifted . nevertheless , already on this small region of the spectrum , three results are found .
first , the possibility of an anisotropy ( i.e. , @xmath75 , in the fragmentation prescription ) of the wind absorption appears unlikely .
indeed , the flux from the back side of the star would be too large ( fig.[fig3 ] , on the top right ) and hence the width of the synthetic line emission would be too large in comparison to the observations ( fig.[fig14 ] , on the right ) .
second , a good fit to the line and its close neighbours is only found if one uses an overabundance of n and a depletion of o ( tab.[tab2 ] and fig.[fig13 ] ) . third , a quick look at a wider wavelength band ( fig.[fig15 ] ) shows that a single temperature plasma can not reproduce the whole spectrum of @xmath2puppis .
a multi - temperature plasma model is required for a full analysis of the x - ray spectrum .
thus the observed line profiles and intensities are a combination of the contribution of the different temperature plasmas present in the wind .
finally , it should be noted that part of the degeneracy of the results is due to the limited resolution of present - day high - resolution x - ray spectrographs . in sect.3
, we show the impact on a line profile of a change in porosity length or inner radius for example , but the convolution by the response matrix of current instruments washes out part of the difference induced by the different parameters .
in this paper , we have studied the impact of the two different prescriptions of stellar wind fragmentation on the emergent broad - band x - ray spectrum . for spherical clumps with an isotropic opacity ,
the differences between the two prescriptions are the most obvious when comparing the overall x - ray spectrum . indeed ,
in this case , we find that a given simulation with a single value of the fragmentation frequency can not be represented by a spectrum calculated with a single value of the porosity parameter .
however , in practice , with the quality of current high - resolution x - ray spectra of o - type stars , it seems almost impossible to distinguish the two models in a real case when working on a line - by - line basis .
this is because the differences are often small and are washed out by the coarse resolution of present - day instruments .
moreover , the differences are most easily seen when comparing the shape of the full spectrum which is seldom done .
in addition , one must keep in mind that the shape of spectrum is also affected by the amount of interstellar absorption and the fact that the x - ray emission from a real stellar wind most likely spans a range of plasma temperatures rather than a single temperature as assumed here ( see e.g. @xcite ; @xcite ) . in both prescriptions
we find that the x - ray emission is more intense in a fragmented stellar wind than in a homogeneous wind.the introduction of an anisotropic opacity in the fragmentation frequency prescription yields spectra that are more intense than in the isotropic case and of course than in an homogeneous wind .
sophisticated , 3-dimensional hydrodynamical calculations might eventually be needed to clarify the issue of the geometry of the clumps . as pointed out previously ,
both models rely on a series of assumptions and , currently , we do not have enough knowledge on the stellar wind fragmentation to privilege one model over the other.our investigation of the impact of the various model parameters on the emerging spectrum showed that both the position and the size of the x - ray emitting plasma are very important .
an increase in the size leads to a more luminous spectrum , although we have shown that the amplitude of the intensity increase depends upon wavelength . for a given mass - loss rate ,
the differences between fully clumped and homogeneous wind models are quite large . however , in light of the simulations presented here , it will be very difficult to estimate correctly the value of the porosity if we include the effect of an inter - clump medium .
in fact , the incorporation of a few percent of the wind material in the inter - clump medium yields the same emergent spectrum as obtained for a purely clumped model with a slightly less fragmented wind . finally , we present preliminary fits of the region near the @xmath0 18.97@xmath1 emission line of @xmath2puppis with our code .
our first results reveal that anisotropic clumps are unlikely , and that non - solar abundances ( overabundance of n , depletion of o ) are needed to reproduce the x - ray spectrum .
however , there is also a degeneracy of stellar wind parameters . even though a porous wind is closest to the observational data
, we can not totally exclude an homogeneous wind .
the mass - loss rate , the porosity parameter and the hot gas filling factor are strongly tied .
however , simultaneous fits of the total spectrum will likely allow to reduce the degeneracy on the porosity and other parameters .
but , it appears that a single temperature x - ray plasma is insufficient to fit the global spectrum of @xmath2puppis .
this result implies the necessity to use several temperatures and we will have to include the contribution of each temperature plasma to the line intensity and profile of @xmath0 18.97@xmath1 ( and all the other lines ) but also to the continuum . 18.97 @xmath1 line profile of the non - void inter - clump medium model e ( in dashed red line ) to the observational data ( in solid black line ) .
_ right _ : comparison of the anisotropic model f ( in dashed red line ) to the observational data ( in solid black line ) .
model parameters are given in table 2 .
[ fig14],width=340 ] 18.97 @xmath1 line is well fitted with an homogeneous model ( model a in dashed red line ) , the triplet and lines in the model are too weak .
that indicates the need to use a second temperature in order to fit the whole spectrum of @xmath2puppis .
consequently , the hot gas filling factor of the first plasma temperature will be lower in order to take account of the contribution , in the intensity and line profile , of the second plasma temperature emission .
[ fig15],width=340 ] this research is supported by the fnrs ( belgium ) and by the communaut franaise de belgique - action de recherche concerte ( arc ) - acadmie wallonie
europe and an xmm+integral prodex contract .
( belspo ) blumenthal , g. r. , drake , g. w. f. , & tucker , w. h. 1972 , , 172 , 205 bouret , j .- c . , lanz , t. , & hillier , d.j .
2005 , a&a 438 , 301 cohen , d.h . ,
leutenegger , m.a . ,
wollman , e.e . ,
zsarg , j. , hillier , d.j . ,
townsend , r.h.d .
, & owocki , s.p .
2010 , mnras , 405 , 2391 diplas , a. , & savage , b.d .
1994 , , 93 , 211 dopita , m.a .
2008 , in _ massive stars as cosmic engines _ , proc .
iau symp . , 250 , 367 eversberg , t. , lpine , s. , & moffat , a.f.j .
1998 , , 494 , 799 feldmeier , a. , oskinova , l. & hamann , w .- r .
2003 , , 403 , 217 gdel , m. , & naz , y. 2009 , astron .
astrophys .
, 17 , 309 hillier , d.j . , kudritzki , r.p . ,
pauldrach , a.w . ,
baade , d. , cassinelli , j.p . ,
puls , j. , & schmitt , j.h.m.m .
1993 , , 276 , 117 hillier , d.j . , & miller , d.l .
1998 , , 496 , 407 hirschi , r. 2008 , in _ clumping in hot - star winds _ , eds .
hamann , a. feldmeier , & l. oskinova , 9 , arxiv0709.0392 kramer , r.h . ,
cohen , d.h .
, & owocki , s.p .
2003 , , 592 , 532 leutenegger , m.a . ,
paerels , f.b.s . , kahn , s.m .
, & cohen , d.h . 2006 , , 650 , 1096 leutenegger , m.a . ,
cohen , d.h . ,
zsarg , j. , martell , e.m . ,
macarthur , j.p . ,
owocki , s.p . ,
gagn , m. , & hillier , d.j .
2010 , , 719 , 1767 miller , n.a . ,
cassinelli , j.p . ,
waldron , w.l . ,
macfarlane , j.j .
, & cohen , d.h .
2002 , , 577 , 951 naz , y. 2009 , a&a 506 , 1055 naz , y. , flores , c.a .
, & rauw , g. 2012 , a&a , in press , arxiv1112.0862 oskinova , l.m . , feldmeier , a. , & hamann , w .- r .
2004 , , 422 , 675 oskinova , l.m . ,
feldmeier , a. , & hamann , w .- r .
2006 , mnras , 372 , 313 owocki , s. 2011 , bulletin de la societe royale des sciences de liege , 80 , 16 owocki , s.p . , & cohen , d.h .
2001 , , 559 , 1108 owocki , s.p . , & cohen , d.h .
2006 , , 648 , 565 porquet , d. , mewe , r. , dubau , j. , raassen , a.j.j . , & kaastra , j.s .
2001 , , 376 , 1113 smith , r.k . ,
brickhouse , n.s . ,
liedahl , d.a . , & raymond , j.c .
2001 , , 556 , l91 sundqvist , j.o . ,
puls , j. , & feldmeier , a. 2010 , , 510 , a11 zhekov , s.a . ; palla , f. 2007 , mnras , 382 , 1124 zsarg , j. , hillier , d.j . , bouret , j .- c . ,
lanz , t. , leutenegger , m.a . , & cohen , d.h .
2008 , , 685 , l149 | high - resolution x - ray spectra of o - type stars revealed less wind absorption than expected from smooth winds with conventional mass - loss rates .
various solutions have been proposed , including porous winds , optically thick clumps or an overall reduction of the mass - loss rates .
the latter has a strong impact on the evolution of the star .
our final goal is to analyse high resolution x - ray spectra of o - type stars with a multi temperature plasma model in order to determine crucial stellar and wind parameters such as the mass loss rate , the cno abundances and the x - ray temperature plasma distribution in the wind . in this context
we are developing a modelling tool to calculate synthetic x - ray spectra .
we present , here , the main ingredients and physics necessary for a such work .
our code uses the most recent version of the atomdb emissivities to compute the intrinsic emissivity of the hot plasma as well as the cmfgen model atmosphere code to evaluate the opacity of the cool wind .
following the comparison between two formalisms of stellar wind fragmentation , we introduce , for the first time in x - rays , the effects of a tenuous inter - clump medium .
we then explore the quantitative impact of different model parameters on the x - ray spectra such as the position in the wind of the x - ray emitting plasma .
for the first time , we also show that the two formalisms of stellar wind fragmentation yield different results , although the differences for individual lines are small and can probably not be tested with the current generation of x - ray telescopes . as an illustration of our method
, we compare various synthetic line profiles to the observed @xmath0 18.97 @xmath1 line in the spectrum of @xmath2puppis .
we illustrate how different combinations of parameters can actually lead to the same morphology of a single line , underlining the need to analyse the whole spectrum in a consistent way when attempting to constrain the parameters of the wind . |
accurate astrometry is essential for measuring parallaxes for stars in the solar neighborhood ( @xmath330 pc ) and proper motions for members of nearby open clusters and star - forming regions ( 100300 pc ) .
the coldest brown dwarfs near the sun and the more obscured members of young clusters can be difficult to detect with the optical and near - infrared ( ir ) instruments that are normally used for such measurements , but they are much brighter at mid - ir wavelengths .
the best available mid - ir instrument for astrometry of this kind is the infrared array camera ( irac ; * ? ? ?
* ) on the _ spitzer space telescope _
irac has already contributed to the measurement of parallaxes for most of the coldest known brown dwarfs @xcite . meanwhile ,
irac observations at multiple epochs spanning a decade are now available for several nearby star - forming regions , making it possible to search for new members via their proper motions ( e.g. , program 90071 ; a. kraus ) . in the two most sensitive bands of irac for stars and brown dwarfs ( 3.6 and 4.5 @xmath0 m ) , the largest sources of uncertainty in astrometry are ( 1 ) systematic offsets in the measured pixel coordinates of stars due to varying intra - pixel sensitivity combined with an under - sampled point spread function ( psf ) , ( 2 ) @xmath2100 mas systematic errors in the pipeline s 3@xmath4 order distortion corrections , and ( 3 ) random errors in the measured pixel coordinates of stars , which can depend significantly on the algorithm adopted for measuring those positions , especially at lower signal to noise ( s / n ) . to date , the first source has been addressed primarily in the context of high - precision photometry of single objects rather than astrometry of multiple sources across the irac arrays ( see @xcite ) . meanwhile , the pipeline distortion corrections have been greatly improved upon with newly developed 5@xmath1 order corrections @xcite , although these new corrections are not yet publicly available . finally , comparisons of algorithms that estimate stellar positions in irac images have been performed only with synthetic data @xcite or at high s / n ratios @xcite . in this paper , we seek to more fully realize irac s capabilities for high - precision astrometry .
we begin by describing the sets of archival irac data that we have selected for our analysis ( section 2 ) . using these data ,
we characterize the systematic offsets due to varying intra - pixel sensitivity and compare the astrometric errors produced by several algorithms for measuring stellar positions ( section 3 ) .
we then measure accurate , high - order distortion corrections for the 3.6 and 4.5 @xmath0 m bands and quantify the astrometric errors produced by combining the new corrections and our preferred algorithm ( section 4 ) .
the irac camera is described in detail by @xcite . in summary , it contains four 256 @xmath5 256 arrays that have plate scales of @xmath6 pixel@xmath7 , resulting in fields of view of @xmath8 .
the arrays collect images at 3.6 , 4.5 , 5.8 , and 8.0 @xmath0 m , which are denoted as [ 3.6 ] , [ 4.5 ] , [ 5.8 ] , and [ 8.0 ] .
the images produced by the camera exhibit a fwhm of @xmath9 . _
spitzer _ was cooled with liquid helium from its launch in august 2003 until the depletion of the helium in may 2009 , which is known as the cryogenic mission .
it has continued to operate in a post - cryogenic phase with the [ 3.6 ] and [ 4.5 ] bands of irac .
the ideal dataset for simultaneously measuring both the distortion and the potential biases in algorithms for estimating stellar positions would have the following characteristic : it would contain a sufficiently large number of images of moderately dense fields of stars such that each pixel in the array experiences the detection of a star on many occasions .
given the data available in the archive , we have selected post - cryogenic observations of dense fields within the galactic plane from program 70157 ( l. allen ) .
these images contain 0.3 million stars with 3.8 million detections at s /
n @xmath10 15 , resulting in an average of 29 detections per pixel in each band . to test the validity of applying the distortion corrections derived from those data to images obtained in other years during the mission , we also have made use of data from programs 104 ( t. soifer ) , 146 ( e. churchwell ) , 20201 ( e. churchwell ) , 40184 ( j. hora ) , 60101 ( r. arendt ) , 70072 ( b. whitney ) , 80074 ( b. whitney ) , and 90071 ( a. kraus ) , all of which are comparable in density of detections to the dataset of the galactic plane .
in addition , to compare the accuracies of different algorithms for measuring stellar positions , we need observations designed to maintain point sources at the same positions on the array among many images . for this purpose , we have selected the time - series photometric monitoring data from program 80179 ( s. metchev ) .
the sensitivity of an individual irac pixel varies from the center to the edge .
because the psfs in the [ 3.6 ] and [ 4.5 ] images are under sampled , the varying sensitivity leads to correlated noise in irac photometry and systematic offsets in the measured positions @xcite .
in addition , the astrometric uncertainties can vary significantly among different algorithms for measuring positions . to correct for the effect of intra - pixel sensitivity and to identify the optimum choice of algorithm ,
we have compared five commonly used routines : the idl functions cntrd , gcntrd , and box_centroider , which use first derivatives , gaussian fitting , and flux - weighted means ( i.e. , the first moment ) , respectively ; the iraf routine ofilter , which employs optimum filtering and a triangular weighting function ; and the point response function ( prf ) fitting routine implemented by the astronomical point source extractor ( apex ; * ? ? ?
* ) , which is the extractor for the mosaicking and point source extraction package developed by the science center ( ssc ) .
although location - dependent prfs are available from the ssc , we use only the central prf for extraction since the use of multiple prfs produces discontinuities in the distortion .
we performed prf - fitting throughout this study using a 5 @xmath5 5 pixel box around each source , which is three times larger than the fwhm .
due to systematic offsets in the calculated positions between some of the algorithms ( see section 4 ) , we began by measuring separate preliminary distortion corrections with post - cryogenic data from program 70157 for the five algorithms we considered .
we then computed the deviations between the true " positions ( see section [ sec : meth ] ) and the measured positions of stars . in figure
[ fig : phase ] , we show the x deviations in [ 3.6 ] as a function of x pixel phase , which is defined as the distance of the measured position from the center of the pixel in the x direction . the residuals are qualitatively similar for the y pixel phase in [ 3.6 ] and the two components in [ 4.5 ] .
we find that the positions from cntrd , gcntrd , box_centroider , and ofilter exhibit large systematic errors ( @xmath10 100 mas ) , which is a reflection of the varying intra - pixel sensitivity .
those x deviations are also correlated with the y pixel phase . for gcntrd
, we note that the offsets near the edge of the pixel form two distinct sequences separated by @xmath270 mas . the large scatter in the residuals of box_centroider is a reflection of that algorithm s low precision ( see figure [ fig : prec ] ) .
for the same post - cryogenic data analyzed with the other four algorithms , prf fitting with apex produces much smaller systematic offsets in the measured positions , as shown in figure [ fig : phase ] .
we also tested prf fitting on the various sets of cryogenic data from section 2 , arriving at the same result , as illustrated with the data from program 40184 in figure [ fig : phase ] .
the [ 3.6 ] cryogenic data do show a small systematic offset ( @xmath210 mas ) , which can be fit with a third - order function in each pixel direction ( see section [ sec : con ] ) .
an offset of this kind was not found in [ 4.5 ] .
it is unclear why a bias should be present only in the [ 3.6 ] cryogenic data . to quantify the precision of the five methods for measuring positions
, we make use of the data from program 80179 because they consist of many images taken at the same position on the sky without dithering . for the four algorithms with large systematic errors due to the varying intra - pixel sensitivity ( figure [ fig : phase ] )
, we applied a two - dimensional correction to the measured positions .
we computed the median absolute deviation ( mad ) of the corrected positions for each star .
we plot the resulting mads for the x direction in [ 3.6 ] as a function of s / n in figure [ fig : prec ] . although all of the methods exhibit similar precisions at high s / n ( @xmath220 mas at s / n = 100 ) , prf fitting produces the smallest errors at lower s / n ( @xmath2100 mas at s / n = 3 ) . because prf fitting with apex produces the smallest systematic offsets due to intra - pixel sensitivity variations and the highest precision , we used it for deriving our distortion correction , and recommend its use when precision irac astrometry is required .
however , ofilter is a reasonable alternative since its systematic errors can be corrected and its precision is only slightly worse than that of apex .
we define our distortion correction in a way that is compatible with the simple imaging polynomial ( sip ) convention ( @xcite ) , which was developed for early in its mission . following the syntax of @xcite ,
we define @xmath11 and @xmath12 to be the positions measured by apex after subtracting 128 .
@xmath11 and @xmath12 are converted to world coordinates ( right ascension and declination ) in the following manner : 1 . if the positions are measured from [ 3.6 ] cryogenic images , @xmath11 and @xmath12 are corrected for varying intra - pixel sensitivity as follows : @xmath13 where @xmath14 and @xmath15 are defined as @xmath16 the values of @xmath17 and @xmath18 are provided in an r package described at the end of section 4.3 . as discussed in section 3
, positions from [ 3.6 ] post - cryogenic images and [ 4.5 ] images do not require such corrections ( i.e. , @xmath19 and @xmath20 ) .
2 . @xmath21 and @xmath22 are converted to intermediate coordinates @xmath23 and @xmath24 by @xmath25 where @xmath26 is the position angle of the y axis , @xmath27 is a time - dependent scale factor applied to the distortion correction ( see section [ sec : time ] ) , and @xmath28 and @xmath29 contain the distortion correction defined in the following manner : @xmath30 as with @xmath17 and @xmath18 , the values of @xmath31 and @xmath32 are found in the r package in section 4.3 .
finally , world coordinates are computed by following the gnomonic projection as described by @xcite , which takes @xmath23 , @xmath24 , and a frame s central world coordinates as input .
we also use the inverse of these steps ( i.e. , the conversion from world coordinates into any of the intermediate steps ) in our measurement of @xmath17 , @xmath18 , @xmath31 , and @xmath32 .
the second and third steps have analytic inversions , while the distortion corrections themselves do not , which introduces an insignificant error ( @xmath30.01 pixel ) to the conversion . to measure distortion corrections with the data from program 70157 ( section 2 )
, we could either adopt an independent catalog of high - precision astrometry or apply self calibration ( i.e. , simultaneously measure an astrometric catalog , distortion corrections of [ 3.6 ] and [ 4.5 ] , and relative offsets and orientations among images ) .
we choose to employ the latter because any independent catalog would have different resolution and sensitivity than the irac images .
because self - calibration involves millions of parameters that depend on each other nonlinearly , we derived initial distortion corrections and measured relative offsets and orientations using astrometry from the two micron point source catalog ( 2mass , * ? ? ? * ) and then we refined the distortion corrections iteratively in the following manner .
( 1 ) we constructed an astrometric catalog by calculating the mean position for each star from individual detections with s / n @xmath10 15 using our current distortion correction and relative offsets and orientations .
( we refer to the mean position as a star s true " position , and the detections as the star s measured " positions . ) to quantify the goodness of fit for the current corrections and offsets , we calculated the rms of all measured positions relative to the true positions .
( 2 ) we alternated solving for new distortion corrections and the relative offsets using least - squares algorithms defined in r ( an open source statistical software package ; * ? ? ?
( 3 ) we continued iterating these two steps until the rms declined to a roughly constant value . in the second step , after applying the appropriate partial inverse transformations to the true positions , we used the linear modeling function lm in r to evaluate the distortion - corrections coefficients .
meanwhile , we measured the relative orientations and offsets between frames using the nonlinear r algorithm optim . after completing an initial series of iterations and converging on a solution , we identified and removed detections on bad pixels and on the edges of the arrays where apex produces erroneous positions and then repeated the iterative process .
to determine the optimum orders for equations ( 6 ) and ( 7 ) , we considered a wide range of orders ( 3 to @xmath1020 ) when analyzing the data from program 70157 .
we then applied the solution derived for each order to the subset of observations from program 90071 that encompassed the taurus star - forming region . for comparison
, we also applied the pipeline s 3@xmath4 order corrections to those data as well .
we measured the relative orientations and offsets between images using the same self - calibration method as before , except with the distortion corrections fixed . to quantify the errors for each set of distortion corrections , we calculated the median deviation of measured positions from true positions in taurus data for 4@xmath54 pixel bins across the [ 3.6 ] and [ 4.5 ] arrays . in figure
[ fig : comp ] , we show the median deviations in the y direction of the [ 4.5 ] array produced by ( 1 ) the pipeline correction and ofilter positions , ( 2 ) the pipeline correction and apex positions , ( 3 ) our 5@xmath1 order correction and apex positions , and ( 4 ) our 8@xmath1 order correction and apex positions .
a time - dependent scale factor described in section [ sec : time ] was applied to our distortion corrections .
the pipeline correction produces residuals ranging from @xmath3380 to 120 mas and @xmath331000 to 450 mas when applied to ofilter and apex positions , respectively .
the difference between the residuals produced by the two algorithms is a reflection of offsets in the measured pixel positions between apex and ofilter , which are likely due to variations of the psf as a function of array location .
this comparison demonstrates that the systematic astrometric errors can depend significantly on the adopted algorithm used to measure positions for a given distortion correction . _ indeed , our distortion corrections are only valid when used in conjunction with apex prf fitting . _
the bayesian information criterion @xcite indicated an optimum order of @xmath220 for our distortion corrections when applied to the dataset used to measure the distortion .
however , when we examined residual maps like those in the top panel in figure [ fig : comp ] for different orders on an independent dataset , we found that the improvements beyond orders of @xmath27 for [ 3.6 ] and @xmath28 for [ 4.5 ] were much smaller than the minimum errors in the measurements of pixel coordinates ( @xmath220 mas , fig .
[ fig : prec ] ) .
therefore , we adopted 7@xmath1 order corrections for [ 3.6 ] and 8@xmath1 order corrections for [ 4.5 ] . to illustrate the improvement of 8@xmath1 order relative to lower orders , we show in figure [ fig : comp ] the residual maps using 5@xmath1 and 8@xmath1 order corrections . in figure
[ fig : resids ] , we plot the median x and y deviations for both bands in the taurus data produced by our adopted corrections .
the large residuals near the edges in figure [ fig : resids ] are due to the fact that they were omitted from the fitting process because of the erroneous positions from apex at those locations , as noted earlier .
we investigated whether our adopted distortion corrections are valid throughout the mission by applying them to the datasets selected in section 2 .
for each dataset , we found a systematic low - order offset in the residual maps .
as an illustration , in the top panel of figure [ fig : scalecorec ] we show the residual maps for the y component of [ 4.5 ] for programs 20201 , 80074 , and 90071 , which have epochs of 2008.33 , 2012.50 , and 2013.91 , respectively . at each epoch , there are two or three sections of the array with distinctly different median residuals , which range from about @xmath3325 to 25 mas . similar trends in the x direction are seen in the residual maps of the x component of [ 4.5 ] . the [ 3.6 ] residual maps also show offsets of this kind , except with a smaller range .
this pattern is likely a reflection of the tiling of images in a given dataset .
we find that these offsets can be can be eliminated by multiplying the distortion correction by a scale factor , as demonstrated in the bottom panel of figure [ fig : scalecorec ] .
we plot these scale factors for all of the datasets that we have analyzed as a function of epoch in figure [ fig : scale ] .
the value of the factor appears to vary on a time scale of several months with the largest change occurring at the transition between the cryogenic and post - cryogenic phases of the mission .
the sizes of the scale factors are equivalent to changes of a few millimeters in the focal length of the telescope .
the scale factor for an epoch not represented in our selected datasets can be interpolated from the scale factors in figure [ fig : scale ] or measured independently in the way that we have done if the dataset is large enough . in figure
[ fig : final ] , we show maps of the magnitude and direction of the distortion implied by our adopted corrections for [ 3.6 ] and [ 4.5 ] at epoch 2013.91 . because the number of coefficients within a single distortion correction is high ( @xmath270 ) and our corrections were measured with r , we have created an r package iracpm " that contains functions and instructions for applying the corrections .
although the formalism of section 4.1 is compatible with sip , we do not recommend simply updating the headers of images with our distortion correction because [ 3.6 ] cryogenic images benefit from a correction for varying intra - pixel sensitivity prior to the application of the distortion and our corrections are only applicable to positions measured by apex ( see figure [ fig : comp ] ) .
to illustrate the astrometric accuracy that can be achieved by using our new distortion corrections , we applied them to archival multi - epoch images of the chamaeleon i star - forming region ( t. esplin , in preparation ) .
we present the mads as a function of s / n for the final epoch of observations ( program 90071 ) in figure [ fig : cham ] .
the median positional errors are @xmath270 mas at s / n = 10 ( 90% of the errors are between 25 and 230 mas ) and @xmath220 mas at s / n = 100 ( 90% between 7 and 42 mas ) .
most of the error at low s / n is likely introduced in the measurement of the pixel coordinates of a star ( see fig . [
fig : prec ] ) with only minor contributions from the relative offsets and orientations between images and the distortion corrections .
we have sought to characterize and correct for the largest sources of uncertainty in astrometry in the [ 3.6 ] and [ 4.5 ] bands of irac .
we began by comparing five commonly used routines for measuring the pixel coordinates of stars that implement flux - weighted means , gaussian fitting , triangular weighting , first derivatives , and prf fitting .
we have found that prf fitting with apex produces the smallest systematic errors from varying intra - pixel sensitivity . after correcting for those systematic errors for each algorithm
, apex provides the greatest precision in the pixel coordinates of sources at all values of s / n . for these reasons , we recommend the use of prf fitting with apex for irac analysis that requires precise astrometry ( e.g. , photometric monitoring for transits or variability and proper motion surveys ) . by applying self - calibration
to irac data in the galactic plane , we have derived 7@xmath34 and 8@xmath34 order distortion corrections for the [ 3.6 ] and [ 4.5 ] arrays , respectively . to test these distortion corrections ,
we have applied them to datasets spanning the mission of _ spitzer_. the resulting residual maps contain low - order systematic offsets that are well - fit by a time - dependent scale factor that is multiplied by the distortion correction . after correcting for this effect , these datasets exhibited residuals that have a range of roughly @xmath3510 mas , which reflects the accuracy of our distortion corrections .
when analyzing data in the chamaeleon i star - forming region by combining apex prf fitting with our new distortion corrections , we have achieved total astrometric errors of @xmath220 and 70 mas at s / n = 10 and 100 , respectively . to promote the use of irac for precision astrometry , we have made our distortion corrections publicly available in a new r package called iracpm " .
we note that our corrections are only valid when used in conjunction with positions measured with apex prf fitting .
we acknowledge support from grant ast-1208239 from the national science foundation .
we thank patrick lowrance and john stauffer for helpful discussions regarding the distortion correction for irac .
2mass is a joint project of the university of massachusetts and the infrared processing and analysis center at caltech , funded by nasa and the nsf .
the center for exoplanets and habitable worlds is supported by the pennsylvania state university , the eberly college of science , and the pennsylvania space grant consortium . | the infrared array camera ( irac ) on the _ spitzer space telescope _ currently offers the greatest potential for high - precision astrometry of faint mid - ir sources across arcminute - scale fields , which would be especially valuable for measuring parallaxes of cold brown dwarfs in the solar neighborhood and proper motions of obscured members of nearby star - forming regions . to more fully realize irac s astrometric capabilities , we have sought to minimize the largest sources of uncertainty in astrometry with its 3.6 and 4.5 @xmath0 m bands . by comparing different routines that estimate stellar positions , we have found that point response function ( prf ) fitting with the _ spitzer _ science center s astronomical point source extractor produces both the smallest systematic errors from varying intra - pixel sensitivity and the greatest precision in measurements of positions . in addition
, self - calibration has been used to derive new 7@xmath1 and 8@xmath1 order distortion corrections for the 3.6 and 4.5 @xmath0 m arrays of irac , respectively .
these corrections are suitable for data throughout the mission of _ spitzer _ when a time - dependent scale factor is applied to the corrections . to illustrate the astrometric accuracy that can be achieved by combining prf fitting with our new distortion corrections , we have applied them to archival data for a nearby star - forming region , arriving at total astrometric errors of @xmath220 and 70 mas at signal to noise ratios of 100 and 10 , respectively . |
to study the complexity of continuous , or discrete , dynamical systems , a large number of concepts have been introduced @xcite .
a non exhaustive list includes the kolmogorov - sinai metric entropy @xcite , the adler - konheim - mcandrew topological entropy @xcite , the arnold complexity @xcite , the lyapounov characteristic exponents , the various fractal dimensions , @xcite the feigenbaum s numbers of period - doubling cascades @xcite , @xmath0 .
many authors have tried to study and discuss the relations between these various notions in an abstract framework @xcite .
inequalities have been shown , for instance the metric entropy is bounded by the topological entropy , let us also mention the kaplan - yorke relation @xcite .
furthermore , many specific dynamical systems have been introduced enabling to see these notions at work .
some of the most popular are the lorentz system @xcite , the baker map @xcite , the logistic map @xcite , the henon map @xcite .
each of these systems has been useful to understand and exemplify the previous complexity measures . here
, we introduce another two - parameter family of mapping of two variables , originating from lattice statistical mechanics for which much can be said .
in particular , we will conjecture an exact algebraic value for the exponential of the topological entropy and arnold complexity .
furthermore , these two measures of complexity are found to be equal for all the values of the two parameters , generic or not ( the notion of genericity is explained below ) .
a fundamental distinction must be made between the previously mentioned complexity measures according to their invariance under certain classes of transformations .
one should distinguish , at least , two different sets of complexity measures , the ones which are invariant under the larger classes of variables transformations , like the _ topological entropy _ or the _ arnold complexity _
@xcite , and the other measures of complexity which also have invariance properties , but under a `` less larger '' set of transformations , and are therefore more sensitive to the details of the mapping ( for instance they will depend on the metric ) .
we now introduce the following two parameters family of birational transformations @xmath1 : @xmath2 which can also be written projectively : @xmath3 as far as complexity calculations are concerned , the @xmath4 case is singled out . in that case , it is convenient to use a change of variables ( see appendix [ appendixa ] ) to get the very simple form @xmath5 : @xmath6 or on its homogeneous counterpart : @xmath7 these transformations derive from a transformation acting on a @xmath8 matrices @xmath9 @xcite : @xmath10 where @xmath11 permutes the entries @xmath12 with @xmath13 , and @xmath14 is the homogeneous inverse : @xmath15 .
transformations of this type , generated by the composition of permutations of the entries and matrix inverse , naturally emerge in the analysis of lattice statistical mechanics symmetries @xcite .
the correspondence @xcite between transformations @xmath16 and @xmath1 , more specifically between @xmath17 and @xmath1 , is given in appendix [ appendixa ] .
it will be shown below that , beyond this correspondence , @xmath17 and @xmath1 share properties concerning the complexity .
transformation @xmath16 is homogeneous and of degree @xmath18 in the @xmath19 homogeneous entries .
when performing the @xmath20 iterate one expects a growth of the degree of each entries as @xmath21 .
it turns out that , at each step of the iteration , some factorization of all the entries occurs .
the common factor can be factorized out in each entry leading to a reduced matrix @xmath22 , which is taken as the representent of the @xmath20 iterate in the projective space .
due to these factorizations the growth of the calculation is not @xmath21 but rather @xmath23 where generically @xmath24 is the largest root of @xmath25 ( i.e. 1.46557123 @xmath26 ) @xcite .
we call @xmath24 the complexity growth or simply the complexity .
this result is a consequence of a _
stable factorization scheme _ given in appendix [ appendixb ] , from which two generating functions @xmath27 and @xmath28 can be constructed . generating function
@xmath27 keeps track respectively of the degrees of the determinants of the successive reduced matrices and @xmath28 of the degrees of the successive common factors .
the function @xmath27 should not be confused with the parameter @xmath29 .
the actual value of @xmath24 is the inverse of the pole of @xmath28 ( or @xmath27 ) of smallest modulus .
the algebraicity of the complexity is , in fact , a straight consequence of the rationality of functions @xmath27 and @xmath28 with integer coefficients @xcite .
the same calculations have also been performed on transformations eq .
( [ uv ] ) and eq .
( [ uvt ] ) . in that case
, factorizations also occur at each step , and generating functions can be calculated .
these generating functions are , of course , different from the generating functions for @xmath30 ( see @xcite ) but they have the _ same _ poles , and consequently the same complexity growth .
one sees that , remarkably , the complexity @xmath24 does not depend on the birational representation considered : @xmath17 for any value of @xmath31 , @xmath1 or the homogeneous transformation eq .
( [ uvt ] )
. it will be usefull to define some degree generating functions @xmath32 : @xmath33 where @xmath34 is the degree of some quantities we look at , at each iteration step ( numerators or denominators of the two components of @xmath35 , degree of the entries of the `` reduced '' matrices @xmath22 s , degree of the extracted polynomials @xmath36 s in appendix [ appendixb ] ... ) .
the complexity growth @xmath24 is the inverse of the pole of smallest modulus of any of these degree generating functions @xmath37 : @xmath38 in the @xmath40 case , which corresponds to a codimension one variety of the parameter space ( see appendix [ appendixa ] and [ appendixc ] ) , additional factorizations occur reducing further the growth of the complexity .
the generating functions are modified and the new complexity is given , for @xmath16 , by the equation : @xmath41 i.e. @xmath42 . for @xmath43 , which corresponds to @xmath17 , the equation reads : @xmath44 leading to the complexity @xmath45 .
not surprisingly , the complexity of the mappings @xmath46 for @xmath4 ( see ( [ uv ] ) ) and mapping @xmath43 ( see ( [ yz ] ) ) , are the same : complexity @xmath24 corresponds to the asymptotic behavior of the degree of the successive quantities encountered in the iteration ( see ( [ lambda ] ) ) .
clearly , this behavior remains unchanged under simple changes of variables . note that this complexity growth analysis can be performed directly on transformation @xmath47 , or on its homogeneous counterpart eq .
( [ yzt ] ) .
the number of generating functions in the two cases is not the same , but all these functions lead to the same complexity . in fact , complexity @xmath24 is nothing but the arnold complexity @xcite , known to be invariant under transformations corresponding to a change of variables ( like the change of variables from eq .
( [ uv ] ) ( for @xmath4 ) to eq .
( [ yz ] ) or to eq .
( [ yzt ] ) ) .
let us also recall that the arnold complexity counts the number of intersection between a fixed line - th iterate of any fixed algebraic curve together with any other possibly different but fixed algebraic curve .
] and its @xmath20 iterate , which clearly goes as @xmath23 .
conversely , all these calculations can be seen as a handy way of calculating the arnold complexity .
all these considerations allow us to design a semi - numerical method to get the value of the complexity growth @xmath24 for any value of the parameter @xmath48 .
the idea is to iterate , with ( [ yz ] ) ( or ( [ uv ] ) ) , a generic _ rational _ initial point @xmath49 and to follow the magnitude of the successive numerators and denominators . during the first few steps , some accidental simplifications may occur , but , after this transient regime , the integer denominators ( for instance ) grow like @xmath23 where @xmath50 is the number of iterations .
typically , a best fit of the logarithm of the numerator as a linear function of @xmath50 , between @xmath51 and @xmath52 , gives the value of @xmath24 within an accuracy of @xmath53 .
an integrable mapping corresponds to a polynomial growth of the calculations : the value of the complexity @xmath24 has to be numerically very close to @xmath54 .
[ f : fig1 ] shows the values of the complexity as a function of the parameter @xmath48 .
the calculations have been performed using an infinite - precision c - library @xcite . for most of the values of @xmath48
we have found @xmath55 , in excellent agreement with the value predicted in eq .
( [ complex ] ) . in @xcite
, it has been shown that the simple rational values @xmath56 yield integrable mappings .
for these special values one gets @xmath57 corresponding to a _ polynomial _ growth @xcite .
in addition , fig .
[ f : fig1 ] singles out two sets of values @xmath58 and @xmath59 , suggesting two infinite sequences @xmath60 and @xmath61 is an involution . ] for @xmath50 and @xmath62 integers such that @xmath63 and @xmath64 and @xmath62 odd .
we call `` specific '' the values of @xmath48 of one of the two forms above ( together with the integrable values ) , and `` generic '' the others . to confirm this conjecture , we go back to ( the matrix ) transformation @xmath65 , for @xmath66 , to get a generating function of the degrees of some factors ( the @xmath67 s in appendix [ appendixb ] ) extracted at each step of iteration , namely , with the notations of @xcite and of appendix [ appendixb ] , function @xmath28 . from now on
, we will give below , instead of @xmath28 , the expression of the following complexity generating function defined , for @xmath68 matrices , as : @xmath69 in the following the calculations are often displayed for @xmath70 matrices and @xmath71 will simply be denoted @xmath72 .
let us recall that the value of the complexity @xmath73 is the inverse of the root of smallest modulus of the denominator of this rational function .
examples of these calculations in order to get the corresponding factorization scheme and deduce the generating function @xmath28 or @xmath74 , are given in appendix [ appendixb ] . in appendix
[ appendixc ] , we show how to choose an initial matrix to iterate , the matrix satisfying @xmath4 and @xmath75 for any integers @xmath76 and @xmath31 . first , we have obtained ( see appendix [ appendixb ] ) the generating function @xmath77 in the generic case for @xmath4 : @xmath78 we also got the generating function @xmath77 for the different `` specific '' cases : @xmath79 @xmath80 and : @xmath81 for the two integrable values @xmath82 and @xmath83 . for @xmath84 ( @xmath85 ) and @xmath86 ( @xmath87 and @xmath62 odd ) , the corresponding complexities are the inverse of the roots of smallest modulus of polynomial : @xmath88 in agreement with the values of fig . [
f : fig1 ] . in this figure
the @xmath48-axis has been discretized as @xmath89 ( @xmath90 integer ) and the extra values 1/7 , 1/11 , 1/13 and 5/7 have been added .
this semi - numerical method acts as an ` integrability detector ' and , further , provides a simple and efficient way to determine the complexity of an algebraic mapping . applied to mappings eq .
( [ uv ] ) , eq .
( [ k ] ) , or eq . ( [ yz ] ) , it shows that the complexity is , generically , _
independent _ of the value of the parameter @xmath48 , except for the four integrable points , and for two denombrable sets of points .
it is worth noticing that these results are not specific to @xmath91 matrices , for example relation eq .
( [ first ] ) is actually valid simply replacing @xmath72 by @xmath71 .
these complexity growth calculations can straightforwardly be generalized to @xmath92 . as explained in appendix [ appendixb ] ,
the `` generic '' generating function is : @xmath93 the pole of smallest modulus of eq .
( [ alpeps ] ) gives @xmath94 for the value of the complexity for the matrix transformation @xmath95 .
the complexity for the transformation @xmath1 is the square of this value : @xmath96 .
[ f : fig2 ] shows , for @xmath97 , complexity @xmath24 as a function of the parameter @xmath98 , obtained with the semi - numerical method previously explained .
even with such a `` small value '' of @xmath99 the expected drastic change of value of the complexity ( namely @xmath100 ) is non - ambiguously seen .
moreover , fig .
[ f : fig2 ] clearly shows that , besides the value @xmath101 known to be integrable whatever @xmath29 @xcite , at least the following values @xmath102 , @xmath103 and @xmath104 are associated with a significantly smaller complexity , at least for the discretization in @xmath48 we have investigated . from these numerical results and by analogy with @xmath105 , one could figure out that all the @xmath106 are also non - generic values of @xmath48 .
in fact a factorization scheme analysis like the one depicted in appendix [ appendixb ] ) shows that @xmath107 or @xmath108 actually correspond to the generic generating function eq .
( [ alpeps ] ) .
we got similar results for other values of @xmath92 .
however , when varying @xmath29 and keeping @xmath48 fixed , new values of the complexity @xmath24 occur , @xmath24 being some `` stair - case '' function of @xmath29 .
we will not exhaustively describe the rather involved `` stratified '' space in the @xmath109 plane , corresponding to the various `` non generic '' complexities .
let us just keep in mind that , besides @xmath101 and @xmath110 , at least @xmath111 , @xmath112 and @xmath113 are singled out for @xmath92 in our semi - numerical analyzis .
the generic expression ( for @xmath91 matrices ) for the generating function @xmath114 eq .
( [ alpeps ] ) is replaced , for the `` non - generic '' @xmath115 ( with @xmath92 ) , by : @xmath116 for the other `` non - generic '' value of @xmath48 , @xmath117 , the complexity generating function reads : @xmath118 for the `` non - generic '' value @xmath119 , the complexity generating function reads : @xmath120
it is well known that the fixed points of the successive powers of a mapping are extremely important in order to understand the complexity of the phase space .
a lot of work has been devoted to study these fixed points ( elliptic or saddle fixed points , attractors , basin of attraction , etc ) , and to analyse related concepts ( stable and unstable manifolds , homoclinic points , etc ) .
we will here follow another point of view and study the generating function of the _ number _ of fixed points . by analogy with the riemann @xmath121 function , artin and
mazur @xcite introduced a powerful object the so - called _ dynamical zeta function _ : @xmath122 where @xmath123 denotes the number of fixed points of @xmath124 .
the generating functions @xmath125 can be deduced from the @xmath121 function : @xmath126 the topological entropy @xmath127 is related to the singularity of the dynamical @xmath121 function : @xmath128 if the dynamical zeta function can be interpreted as the ratio of two characteristic polynomials of two linear operators @xmath129 and @xmath130 , namely @xmath131 , then the number of fixed points @xmath123 can be expressed from @xmath132 . in this linear operators framework
, the _ rationality _ of the @xmath121 function , and therefore the algebraicity of the exponential of the topological entropy , amounts to having a _ finite dimensional representation _ of the linear operators @xmath133 and @xmath130 . in the case of a rational @xmath121 function ,
the exponential of the topological entropy is the inverse of the pole of smallest modulus .
since the number of invariant points remains unchanged under topological conjugaison ( see smale @xcite for this notion ) , the @xmath121 function is also a topologically invariant function , invariant under a large set of transformations , and does not depend on a specific choice of variables .
such invariances were also noticed for the complexity growth @xmath24 .
it is then tempting to make a connection between the _ rationality _ of the complexity generating function previously given , and a possible _ rationality _ of the dynamical @xmath121 function .
we will also compare the arnold complexity @xmath24 and the exponential of the topological entropy @xmath127 .
we try here to get the expansion of the dynamical zeta function of the mapping @xmath43 ( see eq .
( [ yz ] ) ) , for generic values of @xmath48 which are neither of the form @xmath134 , nor of the form @xmath135 .
we concentrate on the value @xmath136 .
this value is close to the value 1/2 where the mapping is integrable @xcite .
on can gain an idea of the number , and localization , of the ( real ) fixed points looking at the phase portrait of fig .
[ f : fig3 ] . the elliptic fixed point @xmath137 is well seen , as well as the five elliptic points and the five saddle points of @xmath138 .
many points of higher degree are also seen .
transformation @xmath43 has a single fixed point for any @xmath48 .
this fixed point is elliptic for @xmath139 and localized at @xmath140 .
transformation @xmath141 has only the fixed point inherited from @xmath47 .
the new fixed points of @xmath142 are @xmath143 , @xmath144 and @xmath145 .
transformation @xmath146 has four new fixed points . at this point
the calculation are a bit too large to be carried out with a literal @xmath48 , and we particularize @xmath147 . for @xmath148
we have five new elliptic points and five new saddles points . the coordinates @xmath149 and @xmath150 of these points are roots of the two polynomials : @xmath151 the five pairings of the seven roots of eq .
( [ fixedptsk5_1 ] ) and eq .
( [ fixedptsk5_2 ] ) , giving the five elliptic points , are ( 0.530283 , -0.107335 ) , @xmath152 ( -0.050283 , -0.24 ) , ( 0.372665 , -0.372665 ) , ( 0.107335,-0.530283 ) , ( 0.24 , 0.050283 ) and the five pairings giving the five hyperbolic - saddle points are ( 0.372665 , -0.075431 ) , ( 0.107335 , -0.107335 ) , ( 0.404568 , -0.24 ) , ( 0.075431 , -0.372665 , ) , ( 0.24 , -0.404568 ) .
this is clearly seen on fig .
[ f : fig3 ] where the occurrence of five `` petals '' corresponding to five elliptic points are obvious , the five hyperbolic points being located between the petals . for transformation @xmath153 , beyond the fixed points of @xmath154 and @xmath155 , one gets two complex saddle fixed points , i.e. transformation @xmath154 has two 6-cycles . for transformation @xmath156 , one obtains one elliptic real fixed point , one saddle real fixed point and and two complex saddle fixed points . for transformation @xmath157 ,
one obtains one saddle real fixed point and four complex saddle fixed points . for transformation @xmath158 ,
one obtains one elliptic real fixed point , three saddle real fixed points and and four complex saddle fixed points . for transformation @xmath159 ,
one obtains one elliptic real fixed point , one saddle real fixed point and and three complex elliptic fixed points and six saddle complex fixed points .
the two elliptic fixed points of @xmath160 ( 0.24 , -0.874 ) and ( 0.874 , -0.24 ) are seen as `` ellipse '' on fig .
( [ f : fig3 ] ) . for transformation @xmath161 ,
one obtains one elliptic real fixed point , five saddle real fixed point and and twelve complex saddle fixed points . on fig .
( [ f : fig3 ] ) a fixed point of @xmath162 lying on @xmath163 is seen near @xmath164 .
the polynomials , similar to eq .
( [ fixedptsk5_1 ] ) and eq .
( [ fixedptsk5_2 ] ) ( or to eq .
( [ p ] ) given in appendix [ appendixd ] ) , as well as the specific pairing of roots , for the successive iterates @xmath165 , are available in @xcite .
it is worth noticing , that among the 53 cycles of @xmath43 of length smaller , or equal , to 11 , as much as 44 have a representent on the line @xmath163 , six have one on the line @xmath166 .
two of the three remaining cycles are of length 11 , while the last is of length eight .
the particular role played by the @xmath167 line can be simply understood .
let us calculate the inverse of the birational transformation ( [ yz ] ) .
it has a very simple form : @xmath168 which is nothing but transformation ( [ yz ] ) where @xmath169 and @xmath170 have been permuted .
the @xmath171 symmetry just corresponds to the _ time - reversal symmetry _
@xmath172 transformation .
the @xmath167 line is the _ time - reversal invariant line_. also note that only one of the 31 complex cycles is of the form @xmath173 where @xmath174 and @xmath175 is the complex conjugate .
the 30 remaining complex cycles are actually 15 cycles and their complex conjugates .
eventually , we observe an _ area preserving _
@xcite property _ in the neighborhood of all the fixed points of _ @xmath176 : the product of the modulus of the two eigenvalues of the jacobian ( i.e. the determinant ) of @xmath176 , at all fixed points for @xmath177 , is equal to @xmath54 .
this _ local _ property is rather non trivial : the determinant of the product of the jacobian over an _ incomplete _ cycle is very complicated and only when one multiply by the last jacobian does the product of the determinants shrink to @xmath178 .
the total number of fixed points of @xmath179 for @xmath180 running from @xmath181 to @xmath182 , yields the following expansion , up to order eleven , for the generating function @xmath183 of the number of fixed points : @xmath184 this expansion coincides with the one of the _ rational _ function : @xmath185 which corresponds to a very _ simple rational _ expression for the dynamical zeta function : @xmath186 expansion ( [ g ] ) remains unchanged for all the other generic values of @xmath187 we have also studied .
we conjecture that : _ the simple rational expression _ eq .
( [ conjec ] ) _ is the actual expression of the dynamical zeta function for any generic value of @xmath48_. comparing the expression eq .
( [ complex ] ) with eq .
( [ conjec ] ) , one sees that the singularities of the dynamical zeta function happen to coincide with the singularities of the generating functions of the arnold complexity . in particular , the complexity growth @xmath24 and the exponential of the topological entropy @xmath127 are _ equal_. when mentioning zeta functions , it is tempting to seek for simple _ functional relations _ relating @xmath188 and @xmath189
. let us introduce the following `` avatar '' of the dynamical zeta function : @xmath190 the transformation @xmath191 is an involution .
one immediately verifies that @xmath192 corresponding to ( [ conjec ] ) verifies two extremely simple and remarkable functional relations : @xmath193 or on the zeta function @xmath194 : @xmath195 the generating function ( [ gg ] ) verifies : @xmath196 an alternative way of writing the dynamical zeta functions relies on the decomposition of the fixed points into _ cycles _ which corresponds to the weyl conjectures @xcite .
let us introduce @xmath197 the number of irreducible cycles of @xmath198 : for instance for @xmath199 we count the number of fixed points of @xmath200 , that are not fixed points of @xmath5 , @xmath201 , @xmath202 or @xmath203 , and divide by twelve .
one can write the dynamical zeta function as : @xmath204 the combination of the @xmath205 s , inherited from the product ( [ weyl ] ) , automatically takes into account the fact that the total number of fixed points of @xmath206 can be obtained from fixed points of @xmath207 , where @xmath208 divides @xmath209 , and from irreducible fixed points of @xmath210 itself ( see @xcite for more details ) .
a detailed analysis of this cycle decomposition ( [ weyl ] ) for generic values of @xmath98 will be detailed elsewhere @xcite .
the previous exhaustive list of fixed points ( up to order twelve ) can be revisited in this irreducible cycle decomposition point of view .
the results of @xcite yield : @xmath211@xmath212 .
one actually verifies easily that ( [ conjec ] ) and ( [ weyl ] ) have the same expansion up to order twelve with these values of the @xmath205 s .
the next @xmath205 s should be @xmath213 @xmath214 it should be noticed that if one introduces some generating function for the _ real _ fixed points of @xmath165 , this generating function has the following expansion , up to order eleven , for @xmath215 : @xmath216 this series is a quite `` checkered '' one . furthermore , its coefficients depend very much on parameter @xmath98 .
in contrast with generating function ( [ genptfixe ] ) , the generating function @xmath217 has _ no universality property _ in @xmath48 .
this series does not take into account the topological invariance in complex projective space : it just tries to describe the dynamical system in the real space .
this series @xmath218 corresponds to the `` complexity '' as seen on the phase portrait of fig .
( [ f : fig3 ] ) .
one sees here the quite drastic opposition between the notions well - suited to describe transformations in complex projective spaces and the ones aiming at describing transformations in real variables . to further investigate the identification of these two notions ( arnold complexity - topological entropy )
, we now perform similar calculations ( of fixed points and associated zeta dynamical functions ) for @xmath219 with @xmath220 and @xmath221 with @xmath222 odd .
the calculations have been performed for @xmath223 for @xmath224 and @xmath225 , giving the expansion of @xmath226 up to order eleven . for @xmath227
this gives : @xmath228 for @xmath229 : @xmath230 for @xmath231 : @xmath232 and for @xmath233 : @xmath234 all these expressions are compatible with this single expression of the @xmath121 function : @xmath235 we conjecture that this expression is exact at every order and for every value of @xmath236 .
again all the singularities of this expression coincide with those of generating function corresponding to the arnold complexity ( see eq .
( [ complexspec ] ) ) .
as far as functional relations relating @xmath188 and @xmath237 are concerned , recalling ( [ avatar ] ) , one immediately verifies that @xmath238 corresponding to ( [ zetam ] ) verifies the simple functional relation : @xmath239 actually @xmath240 has a very simple @xmath50-th root of unity form : @xmath241 also note that when @xmath62 is odd , and only in that case , @xmath240 _ also satisfies _ the functional relation : @xmath242 no simple functional relation , similar to ( [ gg ] ) , can be deduced on @xmath183 .
similar calculations can also be performed for the second set of non - generic values of @xmath243 , namely @xmath244 with @xmath64 , @xmath62 odd . for @xmath245 , that is @xmath246 , one gets , up to order eleven , the _ same expansion _ as eq .
( [ g7 ] ) : @xmath247 suggesting , again , the dynamical zeta function : @xmath248 for @xmath249 , that is @xmath250 , one gets @xmath251 a compatible zeta function could be . ] : @xmath252 this form is not the same as eq .
( [ zetam ] ) , however it has the same poles . comparing these rational expressions for the dynamical zeta function ( ( [ conjec ] ) , ( [ zetam ] ) ,
... ) , and the rational expressions for the generating functions of the arnold complexity ( ( [ betabeta1 ] ) , ( [ betabeta2 ] ) , ( [ betabeta3 ] ) , ...
) for the generic , and non - generic , values of @xmath98 , one sees that one actually has the same singularities in these two sets of generating functions ( note that @xmath11 has to be replaced by @xmath253 since @xmath254 is associate to transformation @xmath255 and not @xmath256 ) .
the identification between the arnold complexity and the ( exponential of the ) topological entropy is thus valid _ for generic values of @xmath48 , _ and _ even for non - generic ones_. it is worth noticing that , due to the topological character of the dynamical zeta function , these results are of course not specific of the @xmath150 and @xmath149 representation of the mapping ( [ yz ] ) but are also valid for the @xmath257 representation ( [ uv ] ) : in particular the exact expressions of the dynamical zeta functions ( namely ( [ conjec ] ) , ( [ zetam ] ) ) , remain unchanged and , of course , the denominators of the complexity generating functions are also the same for generic , or non - generic , values of @xmath258 . the _ local area preserving _ property _ in the neighborhood of all the fixed points of _
@xmath176 previously noticed for @xmath4 , @xmath48 generic , is also verified for these non generic values of @xmath48 .
this ( generic ) identification is not restricted to @xmath40 .
one can also consider mapping ( [ uv ] ) for arbitrary values of @xmath259 and @xmath243 and calculate the successive fixed points .
of course , as a consequence of the higher complexity of the @xmath260 situation ( as shown in section [ secb ] , the complexity immediately jumps from @xmath261 to @xmath262 ) the number of successive fixed points is drastically increased and the calculations can not be performed up to order eleven anymore . in the generic case , the expansion of the generating function @xmath183 of the number of fixed points can be obtained up to order seven : @xmath263 one has two fixed points for @xmath154 , no new fixed points for @xmath264 , three sets of three new fixed points for @xmath155 ( giving @xmath265 fixed points ) , four sets of four new fixed points for @xmath266 ( giving @xmath267 fixed points ) , nine sets of five new fixed points for @xmath268 ( giving @xmath269 fixed points ) , fourteen sets of six new fixed points for @xmath270 ( giving @xmath271 fixed points ) .
this expansion corresponds to the following order seven expansion for the dynamical zeta function : @xmath272 thus yielding to the following rational expression for the dynamical zeta function : @xmath273 this expression can also be written : @xmath274 let us recall the `` alternative '' zeta function ( [ avatar ] ) . it verifies the simple functional relation : @xmath275 this new rational conjecture ( [ zetaalpha ] ) corresponds to the following expression for @xmath183 : @xmath276 comparing the denominators of eq .
( [ zetaalpha ] ) and eq .
( [ alpeps ] ) , one sees that , like for the case @xmath4 , there is an identification between the arnold complexity and the ( exponential of the ) topological entropy @xmath277 heuristically , this identification can be understood as follows .
the components of @xmath165 , namely @xmath278 and @xmath279 , are of the form @xmath280 and @xmath281 , where @xmath282 , @xmath283 , @xmath284 and @xmath285 are polynomials of degree asymptotically growing like @xmath286 .
the arnold complexity amounts to taking the intersection of the @xmath287-th iterate of a line ( for instance a simple line like @xmath288 where @xmath289 is a constant ) with another simple ( fixed ) line ( for instance @xmath288 itself or any other simple line or any _ fixed _ algebraic curve ) . for instance , let us consider the @xmath287-th iterate of the @xmath288 line , which can be parameterized as : @xmath290 with line @xmath288 itself .
the number of intersections , which are the solutions of @xmath291 , grows like the degree of @xmath292 : asymptotically it grows like @xmath293 . on the other hand the calculation of the topological entropy corresponds to the number of fixed points of @xmath165 , that is to the number of intersection of the two curves : @xmath294 which are two curves of degree growing asymptotically like @xmath293 . the number of fixed points is obviously bounded by @xmath295 but one can figure out that it should ( generically ) grow like @xmath293 .
this is fully confirmed by our exact calculations
. the eulerian product weyl - decomposition ( [ weyl ] ) of the dynamical zeta function ( [ zetaalpha ] ) corresponds to the following numbers of @xmath209-cycles : @xmath296 @xmath297@xmath298@xmath299@xmath300@xmath301 . for a `` non - generic '' value of @xmath48 when @xmath92 , namely @xmath82 , the expansion of the generating function @xmath183 and of the dynamical zeta function read respectively : @xmath302 a possible rational expression for the dynamical zeta function is for instance : @xmath303 this last result has to be compared with ( [ alpeps1sur2 ] ) .
for another `` non - generic '' value of @xmath48 when @xmath92 , namely @xmath83 the expansion of the generating function @xmath183 and of the dynamical zeta function read respectively : @xmath304 a possible rational expression for the dynamical zeta function is for instance : @xmath305 this last result with has to be compared with ( [ alpeps1sur3 ] ) .
these results , @xmath306 , we do not have enough coefficients in the expansion of the dynamical zeta function to actually compare it with ( [ g35 ] ) . ]
are again in agreement with an arnold - complexity - topological - entropy identification .
the _ local area preserving _ property in the neighborhood of all the fixed points of @xmath307 previously noticed for @xmath4 , is also verified for @xmath92 for ( [ uv ] ) _ for generic values of _
@xmath48 _ generic as well as these non generic values of _ @xmath48 .
based on analytical and semi - numerical calculations we have conjectured rational expressions with integer coefficients for the generating functions of the complexity and for the dynamical zeta functions for various values of the parameters of a family of birational transformations .
according to these conjectures , the arnold complexity and the exponential of the topological entropy are algerbraic numbers . moreover , these two numbers are _ equal _ for all the values of the parameters . from a general point of view
, rational zeta dynamical functions ( see for instance @xcite ) are known in the literature through theorems where the dynamical systems are asked to be _ hyperbolic _ , or through combinatorial proofs using symbolic dynamics arising from markov partition @xcite and even , far beyond these frameworks @xcite , for the so - called `` isolated expansive sets''(see @xcite for a definition of the isolated expansive sets ) .
there also exists an explicit example of a rational zeta dynamical function but only in the case of an _ explicit linear _
dynamics on the torus @xmath308 , deduced from an @xmath309 matrix , namely the cat map @xcite ( diffeomorphisms of the torus ) : @xmath310 \ , , \ , \,\qquad b \ , = \,\left [ \begin { array}{cc } 1&0\\ \noalign{\medskip}0&1 \end { array } \right ] \ , , \qquad \zeta \ , = \ , \ , { { { \rm det}(1-z \cdot b ) } \over { { \rm det}(1-z \cdot a ) } } \ , = \ , \ , { { ( 1-z)^2 } \over { 1 - 3 \cdot z + z^2}}\end{aligned}\ ] ] note that golden number singularities for complexity growth generating functions have already been encountered ( see equation ( 7.28 ) in @xcite or equation ( 5 ) in @xcite ) .
in our examples , we are not in the context where the known general theorems can apply straightforwardly .
the question of the demonstration of the rationality of zeta functions we have conjectured remains open . in the framework of a `` diffeomorphisms of the torus '' interpretation , the degree of the denominator of a rational dynamical zeta function gives a lower bound of the dimension @xmath311 of this `` hidden '' torus @xmath312 . on expression ( [ zetam ] ) valid for @xmath313 and @xmath106 , one notes that dimension @xmath314 grows linearly with @xmath62 .
the iteration of some birational transformations which densify abelian surfaces ( resp .
varieties ) has been seen to correspond to polynomial growth of the calculations @xcite . introducing well - suited variables @xmath315 ( @xmath316 ) to uniformize the abelian varieties the iteration of these birational transformations just corresponds to a shift @xmath317 . for such polynomial growth situations ,
matrix @xmath129 can be thought as the jordan matrix associated with this translation , its characteristic polynomial yielding eigenvalues equal to @xmath318 .
many denominators of rational zeta functions encountered here are of the form : @xmath319 where @xmath320 is product of cyclotomic polynomials @xcite .
we have encountered : @xmath321 more generally the rational dynamical zeta functions , or the rational functions @xmath322 encountered here , are of the form : @xmath323 ( for @xmath324 ) or @xmath325 ( for @xmath121 ) where @xmath326 and @xmath327 have some kind of decomposition on _ cyclotomic polynomials _ : @xmath328 this is particularly obvious on expressions ( [ betabeta3 ] ) but also on expressions ( [ betabeta2 ] ) , or ( [ zetaalpha1sur2 ] ) , or even ( [ zetam ] ) .
we do not have yet any @xmath329-adic cohomology interpretation ( see for instance @xcite page 453 ) of this cyclotomic polynomials `` encoding '' of the zeta functions or the complexity functions @xmath330 .
most of these rational expressions for zeta functions satisfy very simple functional relations but one also expects , for ( [ zetaalpha1sur2 ] ) or ( [ zetaalpha1sur3 ] ) for instance , more involved but , still simple , functional relations similar to the ones obtained by voros in @xcite .
many of the generating functions @xmath331 can also be seen to satisfy simple functional relations relating @xmath332 and @xmath333 .
this will be detailed elsewhere given by equation ( 5 ) in @xcite verifies @xmath334 . ] .
in practice it is numerically easier to get the generating functions of arnold complexity than getting the dynamical zeta functions .
if one assumes the rationality of the dynamical zeta function and the identification between arnold complexity and ( exponential of the ) topological entropy , getting the generating functions of arnold complexity is a simpler way to `` guess '' the denominator of the dynamical zeta functions .
the analysis developed here can be applied to a very large set of birational transformations of an _ arbitrary _ number of variables , always leading _ rational _ generating functions @xcite . moreover , these generating functions are always simple rational expressions with integer coefficients ( thus yielding _ algebraic numbers _ for the arnold complexity ) .
they even have the previously mentioned `` cyclotomic encoding '' . at this point
, the question can be raised to see if the iteration of _ any _ birational transformation of an arbitrary number of variables always yields rational generating functions for the arnold complexity .
we have even found rational generating functions of arnold complexity for rational transformations _ which are not birational _ ( see ( 7.7 ) and ( 7.28 ) in
@xcite ) : any proof of these rationalities should not depend too heavily on a naive _ reversibility _ of the mapping @xcite .
we have also calculated lyapunov exponents @xcite in order to study the _
metric entropy_. these numerical calculations will be detailed elsewhere @xcite for transformation ( [ yz ] ) for @xmath335 .
these results share quite small values of the lyapunov exponents , the largest of which being much smaller than the topological entropy .
we thus infer that , in this very example , the metric entropy is much smaller than the topological entropy .
we have here an opposition between topological concepts originating from complex projective spaces and the metric concepts of real analysis .
the `` non - topological '' complexity measures do not seem to be able to identify with the previous topological and algebraic quantities . on the birational examples studied here
, the metric entropy does not seem to share the same algebraic values as the topological complexity measures .
in previous papers @xcite , it has been noticed that the successive _ even _ iterates of @xmath339 , acting on an initial matrix @xmath9 , actually belong to a plane which contains matrices @xmath9 , @xmath340 and @xmath341 , or equivalently a fixed matrix @xmath342 ( see @xcite ) : @xmath343 for any integer @xmath50 ( even relative integer ) .
in fact one even has the following property : any point of the plane containing @xmath9 , @xmath344 and @xmath342 is transformed , by the even iterates @xmath345 , into another point of this plane .
this can be used to define the two dimensional mapping @xmath346 compatible with mapping @xmath256 .
let s @xmath9 be an arbitrary @xmath8 matrix , and let us define should be replaced by @xmath347 yielding ( see ( [ aam ] ) and ( [ am ] ) ) : @xmath348 @xmath349 . ] : @xmath350 where @xmath351 with @xmath352 , and @xmath353 and s and @xmath354 s are not exactly the same as the ones given in @xcite , in equations ( 6.13 ) and ( 6.14 ) : the @xmath355 s and @xmath354 s in @xcite are homogeneous expressions and the @xmath355 s and @xmath354 s we introduce here are inhomogeneneous true invariants which can be deduced from the ones in @xcite dividing them be @xmath356 or @xmath357 . ] : @xmath358 then one has the @xmath359 invariance of @xmath29 and @xmath48 : @xmath360 and @xmath361 can be seen as a representation of @xmath255 : @xmath362 where @xmath363 and @xmath243 are precisely the values given by eq .
( [ am ] ) .
transformation @xmath364 reads : @xmath365 in the @xmath366 case , this transformation simplifies and one can introduce new variables @xmath150 and @xmath149 given by : @xmath367 with these new variables , @xmath1 reads : @xmath368 for @xmath369 , transformation @xmath370 is integrable @xcite the invariant being ( see ( 6.38 ) in @xcite ) : @xmath371 this algebraic expression is of course only well - suited for @xmath372 . the variable @xmath149 amounts to considering @xmath373 for arbitrary @xmath48 s .
for @xmath8 matrices ( @xmath374 ) the factorizations corresponding to the iterations of @xmath95 read : @xmath375 and for arbitrary @xmath50 : @xmath376 where @xmath377 for @xmath378 ( mod 4 ) , @xmath379 for @xmath380 ( mod 4 ) , @xmath381 for @xmath382 ( mod 4 ) and @xmath383 for @xmath384 ( mod 4 ) and @xmath385 also depends on the truncation .
factorization relations _ independent of @xmath31 _ , _ occur _ : @xmath386 let us introduce @xcite the generating functions @xmath27 and @xmath28 of the degree of the @xmath387 s and @xmath67 s .
their exact expressions read : @xmath388 it is clear that one has an exponential growth of exponents @xmath389 s , @xmath390 s , @xmath391 s and @xmath392 s : these coefficients grow like @xmath393 where @xmath394 this displays the `` generic '' factorization scheme . however , on various subvarieties like codimension one subvariety @xmath40 , the factorization scheme can be modified as a consequence of additional factorizations occurring at each iteration step , thus yielding a smaller value for the complexity @xmath395 . for @xmath366 the previous factorization scheme becomes for @xmath396 matrices matrices , they are just a bit more involved . ] : @xmath397 and for arbitrary @xmath50 : @xmath398 for @xmath50 even and : @xmath399 for @xmath50 odd . the exact expressions of the generating functions @xmath27 and @xmath28 read ) corresponds to a very simple expression for another generating function introduced in @xcite , namely the function @xmath400 ( see for instance equation ( 8.12 ) in @xcite ) . ] : @xmath401 it is important to note that factorization scheme ( [ newnew ] ) is _ actually stable , but of a slightly more general form , _ as compared to ( [ rappel ] ) , or the ones described in @xcite : recalling the generating functions @xmath402 and @xmath403 of the exponents that occur in the factorization scheme ( see equation ( 8.6 ) and ( 8.10 ) in @xcite ) , one must now introduce _ two sets _ of such exponents generating functions , @xmath404 , @xmath405 , @xmath406 , @xmath407 , in order to keep track of the parity of @xmath50 , and even split these four functions into their odd and even parts : @xmath408 we must also decompose @xmath27 and @xmath28 in odd and even parts : @xmath409 namely : @xmath410 instead of functional relations ( 8.6 ) and ( 8.10 ) in @xcite , one now has the following relations : @xmath411 where the odd and even part of the exponents generating functions @xmath412 , @xmath413 , @xmath414 , @xmath415 , read : @xmath416 period four in the factorization scheme ( [ newnew ] ) corresponds to the occurrence of a @xmath417 singularity for these exponents generating functions .
the `` stability '' of factorization scheme ( [ rappel ] ) corresponds to the following @xmath418-property : the exponents of the @xmath67 s occurring at the @xmath62-th step of iteration are also the one s at @xmath419-th step of iteration , the @xmath67 s being changed into @xmath420 : at each new iteration step one only needs to find the exponent of @xmath421 ( if any ) .
the `` stability '' of factorization scheme ( [ newnew ] ) is a straight generalization mod.2 . of the previous property : the exponents of the @xmath67 s occurring at the @xmath62-th step of iteration are also the one s at @xmath422-th step of iteration the @xmath67 s being changed into @xmath423 .
let us come back to @xmath425 with the non - generic value @xmath426 .
we consider here @xmath427 .
the factorization scheme reads : @xmath428 up to the thirteenth iteration one has the previously described @xmath418-property , but this property is broken with @xmath429 in favor of the @xmath430-property encountered with ( [ newnew ] ) .
the previously introduced odd - even - parity dependent exponents generating functions @xmath431 and @xmath432 now read : @xmath433 from which one deduces , from relations ( [ relationsalphabeta ] ) , the rational expressions of the @xmath434 s and @xmath435 s : @xmath436 yielding the rational expression for @xmath437 : @xmath438 the complexity growth corresponds to the ( smallest ) root of : @xmath439 these results have also been checked using the previously depicted semi - numerical complexity growth evaluation method for @xmath440 and @xmath441 .
the following value for the complexity has been obtained : @xmath442 in good agreement with the exact algebraic value deduced from ( [ b16 ] ) , namely : @xmath443 ( to be compared with the generic algebraic value of @xmath444 , @xmath445 associated with @xmath446 ) .
the singularities of ( [ betaunsurdeux ] ) are in agreement with the dynamical zeta function calculated for these values of @xmath29 and @xmath48 : @xmath447 these calculations can also be performed , for @xmath92 , for the other non - generic value of @xmath448 : @xmath449 .
as far as the factorization scheme is concerned one gets exactly the same scenario as the one described in ( [ newnew1/2 ] ) , the breaking of the @xmath450-property and the occurrence of a @xmath451-property taking place with @xmath452 instead of @xmath453 previously . for @xmath449 and , for instance , for @xmath454 , one gets the following expression for @xmath455 : @xmath456 again these results have been compared with the complexity growth deduced from the semi - numerical method , for @xmath440 and @xmath457 .
we have obtained the following value for the complexity : @xmath458 in good agreement with the exact algebraic value deduced from ( [ betaunsurtrois ] ) , namely : @xmath459 .
the singularities of ( [ betaunsurtrois ] ) are in agreement with the dynamical zeta function calculated for these values of @xmath29 and @xmath48 : @xmath460
we present , in this section , a possible choice of an initial @xmath91 matrix corresponding to a prescripted value of @xmath29 and @xmath48 . from the results of appendix [ appendixa ]
, one has : @xmath461 and : @xmath462 in order to perform our complexity growth calculations to get the factorization scheme of the transformation , one needs to iterate a non - trivial , initial matrix as simple as possible , in the @xmath40 case and for specific values of @xmath448 ( @xmath463 , @xmath464 , ... ) .
actually , let us consider a matrix of the form : @xmath465\end{aligned}\ ] ] the @xmath40 condition factorizes as follows : @xmath466 on the other hand , expression of @xmath467 is also very simple since it also factorizes : @xmath468
the fixed points of @xmath470 can be found as suitable pair of roots of two polynomials @xmath471 and @xmath472 .
the number of pairs of roots being relavitely small ( degre(@xmath342 ) @xmath473 degre(@xmath474 ) ) , it is straightforward to check which are the admissible pairs . for @xmath475 and @xmath476 , the two polynomials happen to verify @xmath477 .
we give below the expression of @xmath471 : @xmath478 the actual value of @xmath149 for the fixed point on @xmath479 is : @xmath480 @xmath481 . polynomials @xmath471 and their partners @xmath472 corresponding to the fixed points of @xmath482 , @xmath483 , and @xmath484 , are available in @xcite as well as their respective pairings of roots . | we propose a conjecture for the exact expression of the dynamical zeta function for a family of birational transformations of two variables , depending on two parameters .
this conjectured function is a simple rational expression with integer coefficients .
this yields an algebraic value for the topological entropy .
furthermore the generating function for the arnold complexity is also conjectured to be a rational expression with integer coefficients with the same singularities as for the dynamical zeta function .
this leads , at least in this example , to an equality between the arnold complexity and the exponential of the topological entropy .
we also give a semi - numerical method to effectively compute the arnold complexity .
.2 cm pacs numbers : 05.45.+b , 47.52.+j .2 cm * key words :* rational dynamical zeta functions , discrete dynamical systems , birational mappings , cremona transformations , arnold complexity , topological entropy . |
charmonium history began in @xmath10 with the discovery of the so - called @xmath1 particle . soon after that
, it was realized that @xmath1 could be described as a bound state of a charm quark and a charm antiquark , namely as a two - body problem . moreover , owing to the large mass of the charm quark , the system could be considered in a first place by its nonrelativistic nature and thus solving the schrdinger equation with a well motivated qcd potential was a successful approach to the problem at hand .
the discovery in @xmath11 of the @xmath2 resonance made evident that the naive quark model picture is not enough to describe its properties .
for instance , the large mass of the @xmath2 and its zero isospin value point out that it is made by charm quarks ; on the other hand , the isospin violating decay @xmath12 makes the @xmath2 difficult to accommodate as charm - anticharm state .
however , the decay mentioned above can be naturally explained in a picture in which the @xmath2 is understood as a @xmath13 bound state , namely a four - quark bound - state problem . during the last decade ,
experimental observations have revealed the existence of a large number of unexpected states such as the @xmath2 , culminating recently with the observation by the lhcb collaboration in @xmath14 of two charmonium pentaquark states @xmath3 and @xmath4 @xcite .
therefore , we have gone from a two - body problem to a five - body problem passing trough a four - body one in less than 15 years . in this proceedings contribution
we present a unified description of all these states within the framework of a constituent model of quarks .
the model has already been applied to two- and four - quark structures and is now being applied to the study of five - quark states .
in addition , we discuss herein a set of four resonances recently measured in the lhcb @xcite : @xmath5 , @xmath6 , @xmath7 and @xmath8 , which have been considered as new multiquark states @xcite but in our scheme most of them appear as simple quark - antiquark structures .
our constituent quark model ( cqm ) is based on the assumption that the light massless light quark acquires a dynamical , momentum dependent mass namely the constituent quark mass as a consequence of the dynamical chiral symmetry breaking of the qcd lagrangian at some momentum scale .
this constituent quark mass is frozen at low momenta at a value around @xmath15 and explains the @xmath16 of the mass of the proton .
the simplest lagrangian which mimics this situation must contain chiral fields to compensate the constituent mass term .
these chiral fields induce boson exchanges between quarks . in the heavy quark sector ,
chiral symmetry is explicitly broken and we do not need additional fields . however , the chiral fields introduced above provide a natural way to incorporate one - pion ( ope ) interactions in the molecular ( four - quark ) dynamics .
higher pion exchanges than ope are effectively included by scalar boson exchanges . besides the dynamically broken chiral symmetry ,
the other two main properties of qcd that are incorporated in our quark model are confinement and asymptotic freedom . to mimic such features we use a linear screened confining potential and a perturbative one - gluon exchange interaction .
the detailed explanation and parametrization of all these interactions can be found in ref .
@xcite , updated in ref .
@xcite . to obtain the solutions in the two - body case , we solve the schrdinger equation using the gaussian expansion method @xcite whereas the four and five - body problem are solved using the resonating group method @xcite .
two - quark and multiquark configurations are coupled using the @xmath17 method @xcite .
details of the calculations are given in refs .
@xcite [ t1 ] cccccc molecule & @xmath18 & @xmath19 & @xmath20 ( mev ) & @xmath21 ( mev ) & @xmath22 ( mev ) + @xmath23 & @xmath24 & @xmath25 & @xmath26 & @xmath27 & @xmath28 + @xmath29 & @xmath30 & @xmath25 & @xmath31 & @xmath32 & @xmath33 + @xmath29 & @xmath24 & @xmath25 & @xmath34 & @xmath35 & @xmath36 + @xmath29 & @xmath37 & @xmath25 & @xmath38 & @xmath39 & @xmath40 +
we consider the @xmath41 and @xmath42 thresholds which are the only ones close to the mass of the @xmath3 and @xmath4 resonances and also where a sizable residual interaction mainly due to pion exchanges can be expected .
one can see in table [ t1 ] that in the mass region of the @xmath3 we obtain a @xmath41 bound state with @xmath43 .
its mass is very close to the experimental one and should , in principle , be identified with the observed @xmath3 structure . referring to the channel @xmath42 , we find three almost degenerated states with a mass around @xmath44 that makes them natural candidates for the @xmath4 resonance .
the total spin and parity of these states are @xmath45 , @xmath46 and @xmath47
. their degeneration may be the origin of the uncertainty in the experimental determination of the total spin and parity for the @xmath4 resonance .
it is remarkable that the three bound states predicted by our model decay into @xmath48 final state with a partial decay width which is generally equal to or larger than the width corresponding to the @xmath49 decay channel .
this suggests that the @xmath50 final state is a suitable decay channel for studying the properties of these resonances .
in particular , the width of the predicted @xmath29 resonance with @xmath45 is twelve times larger through the @xmath51 channel than through the @xmath49 channel , making this decay an important check of our prediction .
concerning the parity of the states , a molecular scenario is not the most convenient to obtain positive parity states because , being the @xmath52 mesons and the @xmath53 baryons of opposite parity , the relative angular momentum should be at least @xmath54 ( p - wave ) which will be above s - waves .
this is reflected in the fact that the states with positive parity in table [ t1 ] are those with smaller binding energies .
[ tab : qqbar ] ccccc state & @xmath55 & @xmath56 & theory ( mev ) & experiment ( mev ) + @xmath57 & @xmath58 & @xmath59 & @xmath60 & @xmath61 + & & @xmath62 & @xmath63 & @xmath64 + & & @xmath65 & @xmath66 & @xmath67 + @xmath68 & @xmath69 & @xmath59 & @xmath70 & @xmath71 + & & @xmath62 & @xmath72 & @xmath61 + & & @xmath65 & @xmath73 & @xmath61 + [ tab : r3 ] cccccc mass & width & @xmath74 & @xmath75 & @xmath76 & @xmath77 + @xmath78 & @xmath79 & @xmath80 & @xmath81 & @xmath82 & @xmath83 + [ tab : r4 ] ccccccc mass ( mev ) & @xmath74 & @xmath84 & @xmath85 & @xmath86 & @xmath87 & @xmath88 + @xmath78 & @xmath80 & @xmath89 & @xmath90 & @xmath91 & @xmath92 & @xmath93 +
table [ tab : qqbar ] shows the calculated naive quark - antiquark spectrum in the region of interest of the lhcb for the @xmath94 and @xmath69 channels . a tentative assignment of the theoretical states with the experimentally observed mesons at the lhcb experiment is also given .
it can be seen that the naive quark model is able to reproduce all the new lhcb resonances except the @xmath5 .
the @xmath6 , @xmath7 and @xmath8 appear as conventional charmonium states with quantum numbers @xmath95 , @xmath96 and @xmath97 , respectively .
a complete study of the decay widths of these states has been performed in ref .
@xcite showing that the total decay width of the @xmath6 , @xmath7 and @xmath8 are , withing errors , in reasonable agreement with the data . to gain some insight into the nature of the @xmath5 ,
that does not appear as a quark - antiquark state , and to see how the coupling with the open - flavour thresholds can modify the properties of the naive quark - antiquark states predicted above , we have performed a coupled - channel calculation including the @xmath98 , @xmath99 , @xmath100 and @xmath101 channels for the @xmath94 sector ; and the @xmath102 , @xmath100 and @xmath101 ones for the @xmath103 sector .
these are the allowed channels whose thresholds are in the region studied by the lhcb .
our results for the @xmath94 channel can be found in ref .
therein , we conclude first that the net effect of coupling the thresholds to both naive quark - antiquark states is to modify the mass of the bare @xmath0 states in a modest amount .
the second observation is that the total decay widths are significantly reduced . for the coupled - channel calculation in the @xmath103 channel
, we found only one state with mass @xmath104 and total decay width @xmath105 .
this state is made by @xmath106 of the @xmath59 charmonium state and by @xmath107 of the @xmath102 component .
our theoretical results are given in tables [ tab : r3 ] and [ tab : r4 ] .
as we do not find any signal for the @xmath5 , neither bound nor virtual , we analyze the line shape of the @xmath101 channel as an attempt to explain the @xmath5 as a simple threshold cusp ( see again ref .
@xcite for details ) .
figure [ fig : f2 ] compares our result with that reported by the lhcb collaboration in the @xmath108 decays .
the rapid increase observed in the data near the @xmath101 threshold corresponds to a bump in the theoretical result just above such threshold .
this cusp is too wide to be produced by a bound or virtual state below the @xmath101 threshold .
as a summary , our results confirm the fact that there are several states with a @xmath109 structure in the vicinity of the masses of the @xmath3 and @xmath4 pentaquarks reported by the lhcb . concerning the other resonances , three of them , namely @xmath6 , @xmath7 and @xmath8 , are consistent with bare quark - antiquark states with quantum numbers @xmath110 , @xmath111 and @xmath112 , respectively . in the @xmath69 sector , we do not find any pole in the mass region of the @xmath5 .
however , the scattering amplitude shows a bump just above the @xmath101 threshold which reproduces the fast increase of the experimental data .
therefore , the structure showed by this data around @xmath113 should be interpreted as a cusp due to the presence of the @xmath102 threshold .
this work has been partially funded by ministerio de ciencia y tecnologa under contract no .
fpa2013 - 47443-c2 - 2-p , by the spanish excellence network on hadronic physics fis2014 - 57026-redt , and by the junta de castilla y len under contract no .
p. g. ortega acknowledges the financial support of the spanish ministerio de economa y competitividad and european feder funds under the contracts fis2014 - 51948-c2 - 1-p .
j. segovia acknowledges the financial support from alexander von humboldt foundation . | the original charmonium two - body problem , like the @xmath0 structure of the @xmath1 meson , has become more involved in the last few years with the discovery of new resonances such as the tentative molecular state @xmath2 or the possible pentaquark one @xmath3 .
we discuss herein how these exotic states ( and others ) can be described in a unified way adding higher fock state components to the naive quark model picture .
in particular , we present our theoretical results on the pentaquark states @xmath3 and @xmath4 , and on the new charmonium - like resonances @xmath5 , @xmath6 , @xmath7 and @xmath8 that have been reported in @xmath9 by the lhcb collaboration . |
an @xmath15 linear collider ( lc ) is probably the most compelling case as the next - generation particle collider for high - precision physics .
one of the lc options , the international linear collider , has recently completed its technical design , and there is hope that funding to begin its construction could be gathered up in a near future , especially if japan s proposal to host it goes forward .
the physics motivation for a linear collider has been strengthened even more after the discovery of a new particle compatible with a higgs boson at the lhc whose interactions could be studied at the lc with sufficient precision to test the electroweak symmetry breaking ( ewsb ) mechanism of the standard model .
the lc would also probe the dynamics behind the symmetry breaking mechanism through high - precision measurements of the properties of the top quark , the heaviest of the fundamental fermions and thus the most strongly coupled to the ewsb sector .
the flexibility in energy of a lc would allow the top
antitop threshold behaviour to be mapped out in detail .
particularly , a theoretically well - defined top mass with a total uncertainty below 100 mev could be extracted by means of such a threshold scan @xcite , substantially beyond the precision achieved at tevatron , @xmath16 gev @xcite , and at the lhc at 7 tev , @xmath17 gev @xcite .
while the statistical uncertainty of the top mass measurement at the lhc is expected to improve in future runs , the systematic ( theoretical ) uncertainties related to the connection between the mass parameter used in the theory and the one measured in the experiment are a limiting factor for further improving the accuracy of the top - quark mass measurement at hadron colliders .
the perturbative nature of the @xmath18 system which is produced near threshold in @xmath15 annihilation was recognized long ago @xcite , and the leading - order coulomb force was treated to all orders in @xmath19 using a non - relativistic approach @xcite .
the matching between qcd and non - relativistic qcd ( nrqcd ) @xcite provided the necessary pieces to compute the next - to - next - to - leading order ( nnlo ) qcd corrections to the @xmath18 production cross section @xcite in the region defined by relative velocities of the top and antitop @xmath20 . in this fixed - order approach which achieves a systematic summation of terms @xmath21 with @xmath22 at order n@xmath23lo , up to n@xmath24lo corrections to resonant @xmath18 production
are known @xcite . in parallel ,
the advances in the formulation of the non - relativistic effective theory allowed for renormalization - group improved calculations for the @xmath18 system produced at threshold .
within the potential nrqcd ( pnrqcd ) @xcite and velocity nrqcd ( vnrqcd ) @xcite formalisms the systematic summation of potentially large logarithmic terms @xmath25 originating from ratios of the top - mass scale @xmath26 , the non - relativistic three - momentum @xmath27 and the kinetic energy @xmath28 , was carried out to next - to - next - to - leading logarithmic ( nnll ) order @xcite for the total cross section , which accounts for all terms proportional to @xmath29 with @xmath30 .
the effective field theory ( eft ) computations above account for the qcd interactions among nearly on - shell top and antitop quarks .
however , the predictions can only be evaluated for all threshold energies after the top decay width is included in the eft quark propagator , @xmath31 , thus providing an infrared cutoff for the top kinetic energy .
the counting @xmath32 is naturally enforced in this way , which is also justified numerically in the standard model , where @xmath33 gev due to the electroweak interaction . once the top width is included ,
the physical final state is @xmath34 at least if we assume that @xmath35 , and consider @xmath36 bosons as stable . beyond leading order
, the production of the final state @xmath34 can also occur through non - resonant processes that do not involve a nearly on - shell @xmath18 pair , and which are thus not described by the standard nrqcd formalism . in the counting scheme where the electroweak coupling scales as @xmath37 , the leading non - resonant effects are nlo for the total cross section and reproduce the full - theory contributions where one of the @xmath8 pairs is produced from a nearly on - shell top , while the other is produced either from a highly virtual top or directly without an intermediate top . the unstable - particle eft @xcite provides the framework for a systematic computation of resonant and non - resonant contributions while maintaining an expansion in the small parameters of the problem .
the nlo non - resonant corrections , calculated within this formalism in @xcite , represent the leading electroweak correction to the @xmath18 cross section below the threshold , where the lo ( resonant ) result rapidly vanishes , reaching up to 20% .
this had been noticed before in refs .
@xcite , obtaining the dominant nlo non - resonant corrections when moderate invariant - mass cuts on the @xmath8 pairs are applied within the so - called _ phase - space matching _ approach , based on vnrqcd . aside from the sizeable corrections induced by the non - resonant production
, there is a further conceptual reason to term the pure qcd resonant result alone that is usually shown in the literature as incomplete . the resonant cross section at nnlo shows _ finite - width divergences _ , _ i.e. _ uncanceled divergences proportional to the top width , which in dimensional regularization have the form @xmath38 and arise from the logarithmic divergences in the imaginary part of the two - loop non - relativistic correlation function .
these are also known as _ phase - space divergences _ because they can be traced back to uv - divergences in the nrqcd @xmath18 phase space integrations @xcite that originate because the unstable - particle propagators describing the top quark in the eft allow for contributions to the forward - scattering amplitude from intermediate top and antitop states which have arbitrarily large invariant masses ( see @xcite for a detailed explanation ) .
the occurrence of finite - width divergences is an evidence that the pure resonant result must be supplemented with additional short - distance information from the full @xmath39 process . in the unstable - particle eft
the additional input is given by diagrams corresponding to off - shell top quark decay that contribute to the non - resonant part at nnlo @xcite .
in this paper we show that the nnlo non - resonant contributions generate infrared divergences when the momentum of the virtual top - quark lines approaches the endpoint at @xmath40 , which precisely cancel the finite - width divergences ( [ eq : sigmadiv ] ) from the resonant part .
this is proved by explicit computation of the nnlo non - resonant contributions , given by the @xmath41 corrections to the nlo non - resonant diagrams . for the extraction of the endpoint - singular terms
we use the method of regions to asymptotically expand the loop and phase - space integrals around the endpoint . as a byproduct
, we obtain the first terms in the expansion in @xmath42 of the nnlo non - resonant contributions , where @xmath7 is a cut on the invariant masses of the top and antitop decay products satisfying @xmath9 .
this approximation provides an accurate estimate of the nnlo non - resonant contributions to the @xmath18 inclusive cross section with moderate invariant - mass cuts in the @xmath8 systems , and , as we prove , confirms the result obtained for the same observable within the phase - space matching approach in @xcite .
the structure of the paper is the following .
in section [ sec : eftdivs ] we recall the issue of the uncanceled divergences proportional to the top width in resonant @xmath18 production at nnlo and collect the total divergent result . the framework to account for non - resonant production at nlo and nnlo
is described in section [ sec : nonresonantdivs ] .
the origin of the endpoint singularities that arise in the non - resonant amplitudes and the method used for their analytic extraction is explained in the latter section , which includes the introduction of the scale @xmath7 .
the results for the endpoint - singular contributions to the nnlo non - resonant corrections are summarized in section [ sec : results ] for the various sets of diagrams .
readers not interested in the individual results might jump directly to section [ sec : nnlores ] , where the formula containing all endpoint - singular non - resonant nnlo contributions to the @xmath43 cross section with an invariant - mass cut @xmath7 is given and the cancellation of the finite - width divergences against the endpoint divergences is made explicit . in section [ sec : comparison ] we compare our findings with those of the phase - space matching approach @xcite and also comment on the approximation to the nnlo non - resonant contributions obtained in another work @xcite by expanding in @xmath44 .
finally , in section [ sec : finalresults ] we compare the non - resonant contributions computed in this work with the nlo ones as a function of the cut @xmath7 , and then show their numerical impact as a function of the energy relative to the leading - order qcd calculation of resonant @xmath18 production .
our conclusions are given in section [ sec : conclusion ] .
close to the top antitop production threshold , the @xmath45 final state is produced from @xmath15 collisions predominantly by intermediate top and antitop quarks with small virtuality ( resonant ) , _ i.e. _ @xmath46 , where @xmath47 and @xmath48 is the non - relativistic kinetic energy .
the qcd dynamics of the nearly on - shell top and antitop quarks can be described within the nrqcd approach , an effective field theory that is built upon integrating out the hard modes with scale @xmath49 .
since the inclusive cross section for the @xmath39 process can be obtained from the @xmath50 cuts of the @xmath15 forward - scattering amplitude , the resonant contribution to this observable is given in the eft formalism by the imaginary part of the matrix element @xmath51 where @xmath52 ( @xmath53 ) are operators describing the production ( decay ) of the resonant @xmath54 pair from @xmath15 , and @xmath55 are short - distance coefficients . at leading order in the non - relativistic expansion
, the @xmath54 pair is produced in an @xmath56-wave , and the first term in ( [ eq : ares ] ) is of order @xmath57 .
the corresponding lo production operators can be found in @xcite . at nnlo ,
@xmath58-wave production operators , as well as new ( @xmath59-suppressed ) @xmath56-wave operators contribute ( for explicit expressions see @xcite ) .
the perturbative contributions to the resonant amplitude ( [ eq : ares ] ) are characterized by top and antitop quark lines with time and spatial components of the momenta obeying the potential scaling , @xmath60 and @xmath61 ( in the centre - of - mass system or in a reference frame differing from it only by a small non - relativistic velocity ) .
as mentioned in the introduction , at nnlo the resonant amplitude ( [ eq : ares ] ) shows an uncanceled finite - width divergence in the imaginary part .
the finite - width divergences arise from two - loop diagrams with a coulomb gluon and a @xmath59-suppressed insertion .
the insertion may correspond to a @xmath59-suppressed potential , an nnlo correction to the kinetic lagrangian , a @xmath58-wave or a @xmath59-suppressed @xmath56-wave current producing the top
antitop pair , or an insertion of the absorptive part of matching coefficients of production operators describing finite lifetime corrections .
the different contributions to the nnlo finite - width divergence are collected in the following formula : @xmath62 2\,n_c\ , \big ( \ , -4\pi a \ , { \rm div}\big [ { \rm im}\ , g_r \big ] + { \rm div}\big[{\rm im}\ , g_{\rm kin } \big ] \nonumber \\ & \hspace{4.9 cm } + { \rm div}\big[{\rm im}\ , g_{\rm dil}\big ] + { \rm div}\big[{\rm im}\ , g_{v^2}\big ] \big ) \nonumber \\ & + \left[\big(c_{p , p\rm \text{-wave}}^{(v ) } \big)^2 + \big ( c_{p , p\rm \text{-wave}}^{(a ) } \big)^2\right ] \frac{4n_c}{3 m_t^2 } \,{\rm div}\big[{\rm im}\,g_{p\text{-wave } } \big ] \nonumber \\ & + \left [ c_p^{(v)}c_p^{(v),\rm abs}+ c_p^{(a)}c_p^{(a),\rm abs}\right ] 4 n_c \,{\rm div}\big[{\rm re}\,g^{(0)}_{\rm c } \big ] \ , , \label{divsigma}\end{aligned}\ ] ] where @xmath63 \quad , \quad c_p^{(a ) } = -4\pi\alpha \,\frac{a_e v_t}{s - m_z^2 } \ , , \\ c_{p ,
p\rm \text{-wave}}^{(v ) } = 4\pi\alpha\ , \frac{v_e a_t}{s - m_z^2 } \quad , \quad c_{p , p\rm \text{-wave}}^{(a ) } = -4\pi\alpha\,\frac{a_e a_t}{s - m_z^2}\end{aligned}\ ] ] are the tree - level coefficient functions of the leading - order @xmath56- and @xmath58-wave production operators , and @xmath64 , with @xmath65 the casimir operator of the fundamental su(3 ) representation
. the vector ( @xmath66 ) and axial - vector ( @xmath67 ) couplings of the fermions to the gauge bosons are given in eq .
( [ eq : vfaf ] ) below , and the remaining symbols are defined as in @xcite .
the @xmath68 are the absorptive parts of the matching coefficients of the leading - order @xmath56-wave production operators .
the latter contributions are part of the corrections induced by the top - quark instability , as they can be interpreted as top - pair production where one of the tops arises from a @xmath8 system with invariant mass very close to @xmath69 .
the coefficients @xmath70 and @xmath71 were calculated in ref .
@xcite ( there denoted @xmath72 and @xmath73 , respectively ) , and their explicit expressions can be found therein .
they are suppressed by @xmath74 with respect to the leading - order @xmath75 , and therefore amount to nnlo corrections .
the absorptive parts in the matching coefficients of the @xmath18 production currents lead to a dependence of the cross section on the real part of the zero - distance coulomb green function @xmath76 , which has a divergence from the two - loop graph with a coulomb potential shown in fig .
[ figcoulomb ] , @xmath77 = \frac{m^2_t a}{16\pi}\,\frac{1}{\epsilon } \
, , \label{divregc}\end{aligned}\ ] ] here regulated dimensionally . the remaining contributions to @xmath78 in ( [ divsigma ] ) arise from divergent contributions to the imaginary part of the green function : @xmath79 , @xmath80 and @xmath81 are first - order corrections to the zero - distance green function from the potential @xmath82 ( in momentum space , leaving out the corresponding wilson coefficient ) , the insertion of the kinetic energy correction @xmath83 , and the insertion of the lifetime - dilatation operator @xmath84 , respectively .
the explicit expressions for @xmath79 , @xmath80 and @xmath81 in @xmath85 dimensions can be found in ref .
their divergent parts satisfy @xmath86 = -{\rm div}\big[{\rm im}\,g_{\rm dil } \big ] = -(4\pi a ) \,{\rm div}\big[{\rm im}\,g_{r } \big ] = \frac{m_t a}{16\pi}\,\frac{\gamma_t}{\epsilon } \ , .
\label{divregrgkingdil}\end{aligned}\ ] ] the term @xmath87 corresponds to the zero - distance coulomb green function obtained with the top
antitop pair produced by the @xmath59-suppressed @xmath56-wave current , which can easily be related to the leading - order coulomb green function by the non - relativistic equation of motion of the top quark , @xmath88
. therefore , @xmath89 = -\frac{\gamma_t}{3m_t } { \rm div}\big[{\rm re}\,g^{(0)}_{\rm c } \big ] = -\frac{m_t a}{48\pi}\,\frac{\gamma_t}{\epsilon } \ , .
\label{divimgv2}\end{aligned}\ ] ] finally , @xmath90 is the @xmath91 component of the coulomb green function at zero distance and describes the production of the top
antitop pair in a @xmath58-wave , which first contributes at nnlo . from the explicit expression of @xmath90
( see _ e.g. _ ref .
@xcite , denoted @xmath92 therein ) we have @xmath93 = \frac{m_t^3 a}{16\pi}\,\frac{\gamma_t}{\epsilon } \ , .
\label{divimgpwave}\end{aligned}\ ] ] using the results ( [ divregc])([divimgpwave ] ) in ( [ divsigma ] ) , we get the total finite - width divergence at nnlo : @xmath94\ , .
\label{divsigmatotal}\end{aligned}\ ] ] the result in ( [ divsigmatotal ] ) agrees with the total uv divergence generated in the nnlo effective - theory matrix elements contributing to @xmath95 as obtained in @xcite .
we note that in the approach followed in refs .
@xcite , the finite - width divergences ( named phase - space divergences therein ) are absorbed by the counterterms @xmath96 associated to @xmath97 forward - scattering operators , and the corresponding phase - space logarithms are resummed using renormalization - group techniques known from effective theories for the coefficients @xmath98 .
the matching conditions for these coefficients at the hard scale , @xmath99 , are related to the non - resonant contributions in the unstable - particle effective theory approach which we discuss in the following section .
the non - relativistic eft formalism has to be extended in order to account for non - resonant production of the physical final state @xmath45 , which involves processes where the @xmath8 pairs are produced by highly virtual top or antitop quarks , @xmath100 , or without intermediate tops .
the leading non - resonant contributions are caused by the top - quark instability , and are thus of electroweak origin . adopting the counting scheme where @xmath37 , a systematic separation of resonant and non - resonant effects can be achieved within the unstable - particle effective theory for pair production near threshold @xcite .
non - resonant effects take place at short distances , as compared to the length scales governing the qcd interaction of the non - relativistic top
antitop pair .
the non - resonant contributions to the @xmath15 forward - scattering amplitude are thus reproduced in this formalism by the matrix element of four - electron production
decay operators @xmath101 @xcite : @xmath102 with short - distance coefficients @xmath103 that are determined by the hard contributions of the @xmath15 forward - scattering amplitude .
explicit expressions for the operators @xmath104 can be found in @xcite . for the @xmath45 inclusive cross section ,
only the imaginary parts of @xmath103 are needed .
the first non - vanishing contribution arises from the cut two - loop diagrams of order @xmath105 shown in fig .
[ fig1 ] , and is thus suppressed by @xmath106 ( nlo ) with respect to the leading term from the resonant part .
contributions where the @xmath107 or the @xmath108 pair in the final state is reached through the resonant decay of a higgs or @xmath109-boson , since the latter constitute a reducible background which can be eliminated in the @xmath18 resonance region by applying cuts on the invariant masses of the final - state particle pairs @xcite . ]
cuts contributing to the nlo non - resonant cross section .
@xmath110 cuts and symmetric diagrams are not shown .
this figure is reproduced from ref .
@xcite.,scaledwidth=90.0% ] the diagrams in fig .
[ fig1 ] are obtained by asymptotically expanding the full electroweak theory diagrams of the @xmath15 forward - scattering amplitude assuming that the momenta in the top and antitop lines are hard , so that the top quarks are far off - shell , @xmath111 .
this expansion implies that the top - quark self - energy insertions ( resummed in the full - theory diagrams in the top - quark propagator ) , @xmath112 , must be treated perturbatively .
accordingly , top - quark propagators in the non - resonant diagrams have no width , contrary to the case of resonant top quarks .
the leading - order imaginary contribution in this non - resonant expansion of the top propagator is proportional to @xmath113 , yielding the 3-particle final states @xmath114 ( fig .
[ fig1 ] ) and @xmath110 instead of the physical 4-particle final state @xmath45 .
in addition , the amplitudes corresponding to the diagrams in fig . [ fig1 ] have to be expanded in @xmath115 in the threshold region .
altogether , the non - resonant contribution at nlo amounts to the calculation of the squared and phase - space integrated amplitudes for the on - shell processes @xmath116 and @xmath117 at the centre - of - mass energy @xmath118 in ordinary perturbation theory ( see @xcite ) .
the nlo non - resonant corrections to the @xmath45 cross section , also including cuts on the invariant masses of the @xmath8 systems , were determined in @xcite and later confirmed by @xcite .
divergences in the non - resonant part can arise when the top ( or antitop ) propagators go on - shell @xcite .
this is a consequence of the hard - momentum region expansion , which forces us to drop the top width from the top
antitop propagators .
given that dimensional regularization is used to deal with divergences in the resonant amplitude , it must also be used here consistently in order to regulate phase - space singularities from top - quark propagators going on - shell . for the 3-particle @xmath114 phase space
, the outer integration variable can be chosen as the squared invariant mass of the @xmath119 subsystem , @xmath120 , where @xmath121 .
the antitop momentum is on - shell here , @xmath122 . setting @xmath118 , as dictated by the asymptotic expansion ,
the kinematics of the process provides the restriction @xmath123 , where @xmath124 is introduced to allow for loose cuts ( @xmath125 or @xmath126 ) in the @xmath127 invariant masses of the form @xcite @xmath128 for the @xmath110 phase space , the roles of @xmath120 and @xmath129 are reversed . to recover the total cross section we have to set @xmath130 . integrating over all other kinematic variables but @xmath120 ,
each non - resonant contribution involves an integral of the form @xmath131 where the endpoint singularity at @xmath132 for @xmath133 has been regularized in @xmath85 dimensions , which drops the scaleless singular contribution from the upper boundary @xmath132 .
the integrals ( [ eq : endpointsing ] ) arise from expanding the non - singular parts of the numerator of the contributions about the endpoint @xmath132 . at nlo , only the diagram @xmath134 has an endpoint divergence , with @xmath135 , and the dimensionally regularized result is therefore finite in the limit @xmath136 . at nnlo ,
however , integrands with @xmath137 are found which generate @xmath1 terms of the form ( [ eq : sigmadiv ] ) .
these shall cancel the finite - width divergences in the resonant part of the full - theory diagrams . from the unstable - particle eft power - counting @xmath138
, nnlo non - resonant corrections can only arise from qcd corrections to the nlo ones .
more precisely , the non - resonant nnlo corrections are obtained from the nlo contributions by adding to the diagrams @xmath134@xmath139 in fig .
[ fig1 ] a virtual gluon or qcd counterterm ( on one side of the cut ) or a real gluon ( going through the cut ) in all possible ways .
the number of diagrams contributing in this way is well above 100 .
fortunately , only a few of them are endpoint - singular , as we show next , first for the virtual contributions and afterwards for the real - gluon radiation diagrams .
nnlo amplitudes with virtual gluons ( see fig .
[ fig2 ] ) involve the same 3-particle cuts ( @xmath114 or @xmath110 ) as the nlo diagrams .
virtual corrections to the two - loop forward - scattering diagrams with @xmath114 cuts which are endpoint - singular .
@xmath110 cuts and symmetric diagrams are not shown .
the @xmath15 external legs have not been drawn , except for the @xmath140 diagram .
, title="fig:",scaledwidth=24.0% ] virtual corrections to the two - loop forward - scattering diagrams with @xmath114 cuts which are endpoint - singular .
@xmath110 cuts and symmetric diagrams are not shown .
the @xmath15 external legs have not been drawn , except for the @xmath140 diagram .
, title="fig:",scaledwidth=24.0% ] virtual corrections to the two - loop forward - scattering diagrams with @xmath114 cuts which are endpoint - singular .
@xmath110 cuts and symmetric diagrams are not shown .
the @xmath15 external legs have not been drawn , except for the @xmath140 diagram .
, title="fig:",scaledwidth=24.0% ] virtual corrections to the two - loop forward - scattering diagrams with @xmath114 cuts which are endpoint - singular .
@xmath110 cuts and symmetric diagrams are not shown .
the @xmath15 external legs have not been drawn , except for the @xmath140 diagram .
, title="fig:",scaledwidth=24.0% ] virtual corrections to the two - loop forward - scattering diagrams with @xmath114 cuts which are endpoint - singular .
@xmath110 cuts and symmetric diagrams are not shown .
the @xmath15 external legs have not been drawn , except for the @xmath140 diagram .
, title="fig:",scaledwidth=24.0% ] virtual corrections to the two - loop forward - scattering diagrams with @xmath114 cuts which are endpoint - singular .
@xmath110 cuts and symmetric diagrams are not shown .
the @xmath15 external legs have not been drawn , except for the @xmath140 diagram .
, title="fig:",scaledwidth=24.0% ] virtual corrections to the two - loop forward - scattering diagrams with @xmath114 cuts which are endpoint - singular .
@xmath110 cuts and symmetric diagrams are not shown .
the @xmath15 external legs have not been drawn , except for the @xmath140 diagram .
, title="fig:",scaledwidth=24.0% ] virtual corrections to the two - loop forward - scattering diagrams with @xmath114 cuts which are endpoint - singular .
@xmath110 cuts and symmetric diagrams are not shown .
the @xmath15 external legs have not been drawn , except for the @xmath140 diagram .
, title="fig:",scaledwidth=26.0% ] virtual corrections to the two - loop forward - scattering diagrams with @xmath114 cuts which are endpoint - singular .
@xmath110 cuts and symmetric diagrams are not shown .
the @xmath15 external legs have not been drawn , except for the @xmath140 diagram .
, title="fig:",scaledwidth=26.0% ] virtual corrections to the two - loop forward - scattering diagrams with @xmath114 cuts which are endpoint - singular .
@xmath110 cuts and symmetric diagrams are not shown .
the @xmath15 external legs have not been drawn , except for the @xmath140 diagram .
, title="fig:",scaledwidth=28.5% ] in order to isolate the endpoint - singular contributions , the integrand of the @xmath120-integration in the 3-particle phase space is asymptotically expanded in powers of @xmath141 , where @xmath142 .
a factor @xmath143 is provided by the kinematics of the phase space and originates from splitting the total momentum @xmath144 into @xmath145 and the on - shell antitop momentum @xmath146 .
negative integer powers of @xmath141 are introduced by top propagators .
this simple counting provides the leading term @xmath147 in the limit @xmath148 for the integrand of the nlo diagram @xmath134 , while diagrams @xmath149@xmath150 , with one intermediate top propagator less , get an additional power of @xmath141 and are endpoint - regular .
likewise , the integrands for diagrams @xmath151@xmath152 , with no internal top line , exhibit the overall scaling @xmath143 and are well - behaved at the endpoint .
gluon insertions can give rise to an additional power @xmath153 , where only @xmath154 is relevant for an endpoint - singular behaviour , and we note that half - integer values of @xmath155 are needed in order to produce @xmath1 endpoint divergences
. the overall power in @xmath141 of the loop integrals involved in the virtual corrections can easily be inferred by using the expansion by regions @xcite , as it is explained in sec .
[ sec : calcmethod ] below .
negative powers of @xmath141 can arise from 1-loop integrals which ( after integration over all phase - space variables but @xmath156 ) are singular when @xmath156 approaches its maximum .
this is only achieved by the gluon correction to the @xmath18-vertex and by the virtual gluon connecting the antitop with the on - shell bottom quark radiated from a top line .
the former yields negative powers with @xmath157 and thus produces endpoint - singular contributions when it corrects diagrams @xmath158@xmath150 ( nnlo diagrams @xmath159 of fig .
[ fig2 ] ) .
in particular , the negative half - integer power introduced by the @xmath18-vertex correction in the nnlo diagrams @xmath160@xmath161 gives integrands scaling as @xmath162 ( for @xmath163 ) and @xmath164 ( for @xmath159 ) , where the latter produce @xmath165 endpoint divergences according to ( [ eq : endpointsing ] ) . on the other hand ,
the virtual gluon connecting the antitop with the on - shell bottom quark radiated from a top line yields factors @xmath164 and @xmath166 .
however , this loop integral replaces one top propagator of the corresponding nlo diagram , so the additional powers are @xmath167 and @xmath168 .
thus this loop integral is only of relevance when inserted into diagram @xmath134 ( nnlo diagram @xmath169 of fig .
[ fig2 ] ) , yielding singular integrand terms scaling as @xmath170 and @xmath164 , respectively .
the remaining virtual @xmath19-corrections to diagram @xmath134 , namely the top - bottom vertex corrections ( in diagram @xmath171 ) , the top self - energy plus the mass counterterm ( @xmath172 plus @xmath173 ) , as well as the @xmath174- and @xmath18-vertex counterterms ( @xmath175 and @xmath176 ) , yield factors @xmath177 at leading order and @xmath178 at next - to - leading order , such that the corresponding diagrams have the same degree of divergence at the endpoint as the nlo diagram @xmath134 , and subleading terms are endpoint - regular .
the same applies to the remaining virtual corrections to diagrams @xmath179 , which are not shown in fig .
[ fig2 ] . in summary ,
the potentially endpoint - singular cases that can be found from a virtual gluon at nnlo are ( @xmath180 ) : @xmath181 only in the last case ( [ eq : intep1 ] ) we obtain a @xmath165 singularity , which is labeled @xmath182 in order to mark it as an endpoint divergence . the cases ( [ eq : intep2 ] ) and ( [ eq : intep32 ] ) are also endpoint - divergent , but finite ( in the limit @xmath136 ) through dimensional regularization . the asymptotic expansion of the phase - space integral near the endpoint translates into an expansion in powers of @xmath42 in the results .
virtual corrections to diagrams @xmath183@xmath152 are all endpoint - regular and contribute first at @xmath184 in this expansion .
let us turn to the nnlo non - resonant contributions with real - gluon emission which involve a 4-particle cut , @xmath185 ( see fig .
[ fig3 ] ) or @xmath186 .
real - gluon corrections to the two - loop forward - scattering diagrams with @xmath187 cuts which are endpoint - singular .
@xmath186 cuts and symmetric diagrams are not shown .
the @xmath15 external states have been omitted .
, title="fig:",scaledwidth=24.0% ] real - gluon corrections to the two - loop forward - scattering diagrams with @xmath187 cuts which are endpoint - singular .
@xmath186 cuts and symmetric diagrams are not shown .
the @xmath15 external states have been omitted .
, title="fig:",scaledwidth=24.0% ] real - gluon corrections to the two - loop forward - scattering diagrams with @xmath187 cuts which are endpoint - singular .
@xmath186 cuts and symmetric diagrams are not shown .
the @xmath15 external states have been omitted .
, title="fig:",scaledwidth=24.0% ] real - gluon corrections to the two - loop forward - scattering diagrams with @xmath187 cuts which are endpoint - singular .
@xmath186 cuts and symmetric diagrams are not shown .
the @xmath15 external states have been omitted .
, title="fig:",scaledwidth=24.0% ] real - gluon corrections to the two - loop forward - scattering diagrams with @xmath187 cuts which are endpoint - singular .
@xmath186 cuts and symmetric diagrams are not shown .
the @xmath15 external states have been omitted .
, title="fig:",scaledwidth=24.0% ] real - gluon corrections to the two - loop forward - scattering diagrams with @xmath187 cuts which are endpoint - singular .
@xmath186 cuts and symmetric diagrams are not shown .
the @xmath15 external states have been omitted .
, title="fig:",scaledwidth=24.0% ] the corresponding phase space can be written with an additional integration over the variable @xmath188 with @xmath189 , where @xmath190 is the gluon momentum .
the kinematic restriction for the total cross section reads @xmath191 , where @xmath192 .
now we want to impose an invariant - mass cut in accordance with the one for the 3-particle phase space described above .
an infrared - safe definition requires that the case where no gluon is emitted and the case where the bottom quark emits a collinear gluon should be indistinguishable . note that when the bottom quark emits a collinear gluon , splitting its momentum into @xmath193 , we may have @xmath194 significantly smaller than @xmath195 . without real - gluon emission ,
the invariant - mass cut defined above imposes a lower bound on the variable @xmath156 , _ i.e. _ @xmath196 with @xmath197 . with gluon emission from the bottom quark
, the new variable @xmath198 represents the quantity which before was described by the variable @xmath156 . then , for consistency , the lower cut has to be imposed on @xmath198 , _ i.e. _ @xmath199 .
if , instead , we imposed a lower bound on @xmath156 in the real - gluon case , the situation @xmath200 could arise , thus excluding events where the gluon is detected separately ( @xmath201 ) and including events where the gluon is part of the bottom jet ( @xmath199 ) .
therefore , a collinear - safe observable is only obtained by imposing the lower limit on the variable @xmath198 which corresponds to the 3-particle variable @xmath156 when the collinear gluon is not detected separately in the detector . with this choice ,
we obtain integrals for the 4-particle phase space parametrized as @xmath202 to extract the endpoint behaviour , these integrals are asymptotically expanded for small @xmath203 , using the strategy of regions as explained in sec .
[ sec : calcmethod ] below . in this parameterization
, the size of the variable @xmath204 , which is proportional to the components of @xmath190 , characterizes the different regions for expanding the inner integral in the limit @xmath205 .
we therefore need to keep track of factors of @xmath203 and @xmath204 ( or of @xmath206 ) appearing in the amplitudes .
the phase - space splitting of the total momentum into @xmath207 and the antitop momentum @xmath146 provides a factor @xmath208 , and the gluon emission , @xmath209 , yields @xmath210 .
top propagators with momenta @xmath145 and @xmath207 contribute negative integer powers of @xmath141 and @xmath203 , respectively , whereas bottom propagators with momentum @xmath211 and antitop propagators with momentum @xmath212 have a negative power of @xmath204 each .
other possible propagators emerging from diagrams @xmath158@xmath152 after the addition of a gluon going through the cut are @xmath213 in the limit @xmath214 , @xmath215 .
the overall scaling at leading order in @xmath203 results as @xmath216 where @xmath217 , @xmath218 and @xmath219 are the numbers of top propagators with momenta @xmath220 , @xmath221 and @xmath145 , respectively , @xmath222 stands for the number of bottom propagators with momentum @xmath223 , and @xmath224 is a function of @xmath225 for @xmath226 .
two different regions are identified for the inner @xmath156-integration : the hard region where @xmath227 , and the ultrasoft region where @xmath228 . after performing the integral over @xmath156 in the hard region and in the ultrasoft region
, we obtain contributions with scalings @xmath229 and @xmath230 , respectively , under the @xmath198-integral .
since the @xmath231 s are integer quantities , the endpoint - singular cases require either @xmath232 or @xmath233 .
it is straightforward to check that these conditions are only accomplished by the real - gluon corrections to the nlo diagram @xmath134 , which consist of three squared @xmath234 diagrams and three interference diagrams , as shown in fig .
[ fig3 ] . moreover , the amplitudes from fig . [ fig3 ]
involve only endpoint - singular cases of the form @xmath235 , producing contributions proportional to @xmath236 according to ( [ eq : intep32 ] ) .
subleading terms in the expansion in @xmath203 do not yield half - integer powers , so no @xmath165 endpoint singularities arise from the real - gluon emission diagrams that contribute to the nnlo non - resonant cross section .
the endpoint - singular contributions contained in the amplitudes of figs . [ fig2 ] and [ fig3 ] arise from terms in the integrand of the outer phase - space integration of the form @xmath237 , with @xmath238 for the virtual - gluon corrections , and of the form @xmath239 for the case of real - gluon emission .
the extraction of such terms is greatly simplified if the 1-loop and phase - space integrals over other variables are asymptotically expanded in @xmath141 and @xmath203 prior to their evaluation using the strategy of expansion by regions @xcite .
the relevant regions are determined and their completeness is confirmed by asymptotically expanding mellin
barnes representations of the full integrals .
such an interplay between expansion by regions and mellin
barnes representations has been described in detail in @xcite . for the 3-particle ( virtual - gluon ) diagrams ,
three regions are found in the loop integration over the ( gluon ) momentum @xmath240 : the _ hard region _ where @xmath241 , the _ ultrasoft region _ where @xmath242 , and the _ potential region _ where @xmath243 and @xmath244 ( defined in the rest frame of @xmath145 ) .
note that these regions are analogous to the ones that are found in the threshold expansion @xcite if we identify @xmath59 there with @xmath141 .
for the 4-particle ( real - gluon ) diagrams , only the hard region with @xmath245 and the ultrasoft region with @xmath246 contribute to the phase - space integral over the variable @xmath156 .
it is illustrative to show how the method of regions applies to our case by means of two non - trivial examples .
consider first the virtual gluon connecting the antitop with the on - shell bottom quark , diagram @xmath169 of fig .
the relevant 1-loop scalar integral reads @xmath247 where we have used the on - shell conditions for the antitop ( @xmath248 ) and the bottom quark ( @xmath249 ) . in order to expand the denominators for each region we need to know the scaling in @xmath141 of the final - state particle momenta . working in the top rest frame , where @xmath250 , simple kinematics gives , at leading order in @xmath141 , @xmath251 note that we regard neither @xmath252 nor @xmath253 as small quantities for the purpose of the @xmath141-expansion . in the hard region where @xmath254 , all the propagators scale as order @xmath255 , and @xmath256 .
taking into account the factor @xmath143 from the @xmath257 phase - space measure and @xmath258 from the top propagator on the r.h.s .
of the cut , the hard - region contribution scales as @xmath166 , and is thus endpoint - regular . in the ultrasoft region
, we have to consider @xmath259 , which allows to drop @xmath260 from the fermion propagators as well as @xmath261 , obtaining @xmath262 where @xmath263 denotes the first term in the @xmath148 expansion of the corresponding momentum @xmath264
. therefore we get that @xmath265 , and an overall endpoint - singular contribution to diagram @xmath169 from the ultrasoft contribution scaling as @xmath170 .
an additional factor @xmath266 in the integrand of @xmath267 ( coming , for instance , from the next - to - leading term in the expansion of the antitop propagator in the ultrasoft region ) could potentially yield a @xmath268 endpoint divergence .
however , the resulting angular integral can be shown to vanish , so there are no @xmath268 endpoint divergences from the ultrasoft region . finally , the expansion of @xmath269 in the potential region where @xmath270 yields at leading order @xmath271 from ( [ eq : i1bpot ] ) we find @xmath272 , and thus a leading - order scaling for the potential - region contribution to diagram @xmath169 . subleading terms of order arise from the next - to - leading term in the expansion of the bottom propagator ( suppressed by @xmath273 ) as well as from an additional factor of either @xmath274 or @xmath275 ( both @xmath276 ) in the numerator of the full @xmath169 amplitude .
these subleading terms generate the @xmath268 endpoint divergences from diagram @xmath169 .
let us now consider an example involving a real - gluon correction , namely diagram @xmath277 from fig .
[ fig3 ] . in the rest frames of either @xmath145 or @xmath207 ,
the components of the on - shell gluon momentum are proportional to the variable @xmath204 , defined in sec .
[ sec : nonresonantdivs ] , such that both the bottom propagator with momentum @xmath211 and the antitop propagator with momentum @xmath212 provide a negative power of @xmath204 each .
so , according to ( [ eq : realgluon ] ) with @xmath278 , the integral over the variable @xmath156 has the form @xmath279 where @xmath280 for @xmath226 , and we have introduced @xmath281 . applying the expansion by regions to integrals with finite boundaries or theta functions is explained in @xcite .
we are interested in the expansion of this integral in @xmath203 . as can easily be confirmed by writing a mellin
barnes representation of the integral , for @xmath205 it receives contributions from the hard region , @xmath282 , and from the ultrasoft region , @xmath283 . in the former case we can expand out factors of @xmath203 in the denominator of ( [ eq : g5 ] ) and set @xmath284 in the numerator for the leading - order contribution : @xmath285 so the overall scaling of the hard contribution is given by the remaining factor in ( [ eq : realgluon ] ) , @xmath286 , which is endpoint - regular .
however , in the ultrasoft region we have to set @xmath287 and @xmath288 and keep the first propagator unexpanded .
thus @xmath289 which provides the scaling @xmath290 for the leading ultrasoft term .
by explicit computation it can be shown that no @xmath291 contribution is generated by subleading terms in the expansion of the ultrasoft region because they involve angular integrations of the form @xmath292 which vanish .
the following results contain all endpoint - singular contributions to the individual diagrams in figs .
[ fig2 ] and [ fig3 ] , _ i.e. _ , within the asymptotic expansion described above , the terms of ( [ eq : endpointsing ] ) with @xmath293 . endpoint - regular contributions with @xmath294 are omitted . from the viewpoint of the results , all terms of the expansion in powers of @xmath42
are provided up to and including the order @xmath295 , as shown in eqs .
( [ eq : intep2])([eq : intep1 ] ) .
the omitted terms are of order @xmath296 without endpoint - singular @xmath165 terms or logarithms of @xmath7 , or of higher order in @xmath6 ; they are regular in the limit @xmath297 .
the results presented here get dominant in the limit @xmath297 , where they are _ a priori _ not valid because our effective - field - theory treatment requires @xmath125 ( see @xcite ) .
however , our result taking @xmath7 as a tiny ( @xmath298 ) unphysical scale provides a precise approximation of the amplitude at the endpoint , which could be complemented by a numerical evaluation of the contributions outside this tight @xmath7-cut ( where the top propagators never go on shell ) in order get a precise evaluation of the full nnlo non - resonant corrections to the total cross section with or without invariant - mass cuts .
all the diagrams relevant to this nnlo analysis with the exception of @xmath140 have the structure of a hadronic tensor @xmath299 connected to a leptonic tensor via two photon or @xmath109 propagators .
while the leptonic tensor contains the @xmath10 pairs on the left - hand and right - hand sides of the forward - scattering amplitude , the hadronic tensor involves everything between the two photon/@xmath109 propagators , including the cut through @xmath300 , with the exception of two powers of the elementary charge @xmath301 removed from the two external vertices . in the case of photon/@xmath109 couplings to quarks , these two vertices on the left - hand and right - hand side of the hadronic tensor read @xmath302 respectively , where @xmath303 and @xmath304 are the vector and axial - vector couplings of the corresponding quark ( @xmath305 ) . in the case of a @xmath109-boson , they are given by @xmath306 where @xmath307 ( @xmath308 ) is the sine ( cosine ) of the weak mixing angle , @xmath309 the electric charge of the fermion ( @xmath310 , @xmath311 ; similarly @xmath312 ) , and @xmath313 the third component of the weak isospin of the fermion .
for a photon , @xmath314 and @xmath315 . in the case of diagram
@xmath316 the relevant 3-gauge - boson coupling is @xmath317 , multiplying the usual momentum - dependent terms as defined in ( [ eq : iwwl ] ) below .
we have @xmath318 for a photon attached to the left of the hadronic tensor and @xmath319 for a @xmath109-boson .
we note that for all diagrams except for @xmath320 , ( which are already symmetric ) there is a symmetric contribution where the right and left parts of the diagram have been exchanged . for the diagrams with the hadronic - tensor structure ( all except @xmath140 ) ,
the corresponding amplitude for the symmetric amplitude is obtained by replacing the coupling factors on the sides of the hadronic tensor as @xmath321 where @xmath322 , and then taking the complex conjugate of the whole amplitude . the results in this section correspond to the amplitudes with @xmath323 cuts , as shown in figs .
[ fig2 ] and [ fig3 ] .
the additional diagrams with @xmath324 cuts also have to be taken into account when computing the non - resonant contribution to @xmath325 , but their contribution is equal to the corresponding one with @xmath300 cuts by virtue of @xmath326-invariance .
the hadronic tensor has the lorentz structure @xmath327 with the total momentum @xmath328 and @xmath329 .
only the first term contributes to the cross section , it is projected out by @xmath330 for the propagators and polarization sums of gluons , feynman gauge is used , while unitarity gauge is employed for @xmath36-bosons .
all nnlo contributions are proportional to the factor @xmath331 where @xmath332 is the scale introduced in dimensional regularization and @xmath333 is the number of qcd colours .
@xmath334 is the tree - level top decay width , @xmath335 obtained from the amplitude @xmath336 with the bottom - quark mass set to zero . in the presentation of the results that follows ,
we shall mark if the @xmath165 poles are of ultraviolet , infrared or endpoint - singular origin by writing @xmath337 , @xmath338 or @xmath182 , respectively .
we first give results for the hadronic tensor coefficient @xmath339 from the virtual @xmath41 corrections to diagram @xmath134 , _
i.e. _ diagrams @xmath340 , depicted in fig .
[ fig2 ] , that generate the three types of endpoint - divergent contributions listed in ( [ eq : intep2])([eq : intep1 ] ) .
it is a general feature of these corrections that @xmath341 and @xmath295 terms arise only from the potential region in the 1-loop integration over the gluon momenta , whereas the @xmath4 terms come from either hard or ultrasoft gluons ( with the only exception of diagram @xmath169 , as explained below ) .
the amplitude for @xmath163 arises from inserting the virtual - gluon correction in the @xmath18-vertex and has the highest degree of endpoint singularity , as explained in sec .
[ sec : nonresonantdivs ] .
the result reads @xmath342 \nonumber \\ & \ ; + \left ( \frac{1}{{\epsilon}_{\text{ep } } } + 2 \ln\frac{m_t^2}{\lambda^2 } \right ) \left [ - v_t^{\text{l}}v_t^{\text{r}}\ , \frac{2 \ , ( 2 + 2x+5x^2)}{3 \ , ( 1-x ) \ , ( 1 + 2x ) } - \frac{1}{4 } v_t^{\text{l}}a_t^{\text{r}}+ \frac{1}{6 } a_t^{\text{l}}a_t^{\text{r}}\right ] \
, \biggr\ } \,.\end{aligned}\ ] ] terms of order @xmath296 without @xmath182 terms or logarithms of @xmath7 and endpoint - regular contributions of @xmath343 are omitted in our results , as explained above .
also terms of @xmath344 are dropped everywhere . for the virtual gluon connecting the bottom and the antitop quark , diagram @xmath169
, the hadronic tensor coefficient reads @xmath345 \nonumber \\ & \ ; + \left ( \frac{1}{{\epsilon}_{\text{ep } } } + 2 \ln\frac{m_t^2}{\lambda^2 } \right ) \left [ v_t^{\text{l}}v_t^{\text{r}}\ , \frac{3 \ , ( 1+x+2x^2)}{4 \ , ( 1-x ) \ , ( 1 + 2x ) } - ( v_t^{\text{l}}a_t^{\text{r}}- a_t^{\text{l}}v_t^{\text{r}})\frac{1 - 2x}{6(1 + 2x ) } \right ] \ , \biggr\ } \,.\end{aligned}\ ] ] the single and double infrared singularities in ( [ eq : h1b ] ) are related to the emission of the virtual gluon from the massless bottom quark .
the contribution proportional to @xmath346 of order @xmath236 in the second - to - last line of ( [ eq : h1b ] ) arises from potential - gluon momentum , but , as it is purely imaginary , it cancels with the symmetric contribution where the gluon is exchanged on the r.h.s . of the cut .
the virtual gluon correcting the @xmath174-vertex , diagram @xmath171 , gives @xmath347 \,.\end{aligned}\ ] ] the remaining virtual corrections to diagram @xmath134 correspond to renormalization and self - energy contributions . at @xmath41 ,
the top- and bottom - quark fields and the top mass need to be renormalized , which is done in the on - shell scheme .
since the non - resonant nlo diagrams are purely of electroweak origin , they do not involve qcd couplings , so the renormalization of @xmath19 is irrelevant for the current nnlo analysis .
the insertion of the renormalized self - energy into a top / antitop line with momentum @xmath348 next to a cut vanishes , since this contribution is proportional to @xmath349 in the on - shell limit , and the cutkosky rules prescribe a factor @xmath350 which sets @xmath349 to zero as a scaleless term in dimensional regularization .
the same is true for a cut bottom line , because the bottom self - energy is scaleless and vanishes for @xmath249 .
therefore we do not need to consider self - energy insertions in lines which are cut . on the other hand , when the top self - energy is inserted into an internal top line ,
the top - field renormalization parts of the vertex counterterms from the two adjacent vertices cancel exactly the contribution of the field - renormalization to the 2-point top counterterm in the renormalized self - energy insertion , so the complete correction to an internal top line is equal to the insertion of the bare self - energy plus the mass - renormalization part of the 2-point top counterterm .
this correction only generates an endpoint - singular contribution when inserted into the internal top lines of the nlo diagram @xmath134 .
the corresponding nnlo diagrams are @xmath172 and @xmath173 in fig .
[ fig2 ] , and the sum of both gives the coefficient @xmath351\,.\end{aligned}\ ] ] the bare self - energy entering diagram @xmath172 receives contributions from hard and ultrasoft loop momenta , whereas the mass- as well as the field - renormalization constants are entirely determined by hard contributions . for the computation of @xmath173 ( and also for @xmath175 and @xmath176 )
we need the result for the nlo amplitude @xmath134 retaining @xmath352 terms , which was not necessary for the calculation performed in @xcite . for completeness
we write here the hadronic tensor coefficient for the endpoint - singular term of @xmath134 with the full @xmath353-dependence : @xmath354 although the renormalized self - energy insertion into a cut bottom line vanishes , there is a contribution from the bottom - field renormalization part @xmath355 of the counterterm of the @xmath356 vertex , diagram @xmath175 .
this correction is obtained by multiplying the nlo result @xmath357 ( [ eq : h1 ] ) with @xmath355 , where @xmath358 is defined from the relation between the bare and renormalized bottom - quark field , @xmath359 , @xmath360 obtaining @xmath361 which is effectively zero , but exhibits separate ultraviolet and infrared singularities . finally , there is a contribution from top - field renormalization , diagram @xmath176 , attributed to the antitop part @xmath362 of the counterterm of the @xmath18-vertex .
the top - field counterterm @xmath363 , defined analogously to @xmath358 , is determined from the derivative of the bare self - energy with respect to @xmath364{/}\hspace*{-0.55\dslashwidth } p}_t$ ] , giving @xmath365 multiplying @xmath362 with ( [ eq : h1 ] ) we get for the hadronic tensor coefficient from diagram @xmath176 : @xmath366 \,.\end{aligned}\ ] ] let us note that in the sum of all virtual - gluon corrections and renormalization contributions to the diagram @xmath134 listed above , the ultraviolet @xmath337 singularities cancel out , such that the overall @xmath41 correction is ultraviolet - finite .
( recall that the @xmath165 divergences in the real - gluon corrections discussed next can only be of infrared origin . ) as explained in sec .
[ sec : nonresonantdivs ] , for endpoint - singular contributions involving real gluons we need to consider only the forward - scattering amplitudes with a 4-particle cut corresponding to real - gluon corrections to the nlo diagram @xmath134 .
these have been shown explicitly in fig .
[ fig3 ] for the case of @xmath367 cuts .
the real - gluon corrections involve three squared @xmath368 amplitudes ( diagrams @xmath369 ) and three interference amplitudes ( diagrams @xmath370 ) that have a symmetric counterpart which is obtained by mirroring the diagram across the cut ( without reversing the fermion flow ) .
the contributions to the hadronic tensor from the symmetric diagrams read @xmath371 \,,\end{aligned}\ ] ] @xmath372\,,\end{aligned}\ ] ] @xmath373\,.\end{aligned}\ ] ] while diagram @xmath374 receives contributions from both hard and ultrasoft gluons , diagram @xmath375 only gets contributions from the hard region and diagram @xmath376 only from the ultrasoft region . in the case of the interference diagrams ,
the contributions of both @xmath377 and @xmath277 are entirely produced by ultrasoft gluons , whereas @xmath378 receives as well contributions from hard gluons .
the results for the interference diagrams read @xmath379 @xmath380\,,\end{aligned}\ ] ] @xmath381\,.\end{aligned}\ ] ] it can be checked that in the sum of the real - gluon and virtual - gluon corrections to diagram @xmath134 , the infrared - singular terms cancel out completely .
the cancellation of infrared divergences holds independently for the combinations @xmath382 , @xmath383 , @xmath384 and @xmath385 , where the factor @xmath386 in front of @xmath175 and @xmath176 accounts for the symmetric contribution .
it is also interesting to note that the sum of the ultrasoft - gluon ( real and virtual ) contributions in the @xmath41 corrections to diagram @xmath134 above vanishes . in particular ,
the ultrasoft pieces in the combinations @xmath387 , @xmath382 , @xmath383 and @xmath388 , cancel out separately .
since there are no potential - region contributions of order @xmath236 and the remaining endpoint - singular diagrams @xmath389 , give only @xmath390 terms , the order @xmath236 in the non - resonant cross section originates purely from the hard region .
this also implies that logarithms @xmath391 will be absent in the @xmath236 term : in individual contributions , they arise from expanding hard - region terms @xmath392 and ultrasoft - region terms @xmath393 in powers of @xmath353 . due to the different scaling of these two regions with @xmath394 , logarithms @xmath391
may remain in a contribution even if it is finite for @xmath395 , cf . the amplitude ( [ eq : hg1 ] ) for diagram @xmath374 .
but where the contributions from both hard and ultrasoft regions are separately finite or , as here , the ultrasoft contribution vanishes completely , leaving only an ultraviolet- and infrared - finite hard contribution , there the logarithms @xmath391 must disappear . the cancellation of ultrasoft terms in the nnlo non - resonant calculation
is closely connected with the absence of contributions from ultrasoft gluons in the nnlo matrix element ( [ eq : ares ] ) containing the interactions among the resonant top and antitop quarks .
we postpone the discussion of this issue to sec .
[ sec : psm ] .
the nnlo diagrams @xmath397 , @xmath316 and @xmath140 of fig .
[ fig2 ] produce endpoint - singular contributions of the form ( [ eq : intep1 ] ) .
the additional power @xmath398 needed to produce the @xmath182 from the nlo diagrams @xmath399 , @xmath400 and @xmath401 arises from the potential region of the @xmath18-vertex correction .
the hadronic tensor coefficients from @xmath397 and @xmath316 read @xmath402 where the coupling factor @xmath317 in ( [ eq : h3a ] ) is defined from the 3-gauge - boson vertex feynman rule @xmath403 \,,\ ] ] omitting the elementary charge @xmath301 .
the total momentum @xmath404 is incoming from the left - hand side ( index @xmath332 ) , the lower @xmath405-boson has outgoing momentum @xmath406 ( index @xmath407 ) , and the upper @xmath408-boson has outgoing momentum @xmath409 ( index @xmath410 ) . for diagram
@xmath140 we have to evaluate directly its contribution to the unpolarized @xmath411 cross section , since it does not have the structure of a hadronic tensor contracted with a leptonic tensor .
the cross - section contribution of diagram @xmath140 , already summed over the two gauge bosons , photon and @xmath109 , in the @xmath412-channel propagator reads @xmath413
\left ( \frac{1}{{\epsilon}_{\text{ep } } } + 2 \ln\frac{m_t^2}{\lambda^2 } \right ) \,,\end{aligned}\ ] ] where the squared centre - of - mass energy @xmath412 has been set to @xmath414 only in the non - trivial parts of the loop and phase - space integrations , but kept general in the photon and @xmath109 propagators and in the kinematic factors of the cross section .
this corresponds to the same prescription used for the nlo non - resonant contributions computed in @xcite .
with all the results from the individual diagrams of figs .
[ fig2 ] and [ fig3 ] at hand , we can compute the complete endpoint - singular non - resonant nnlo contribution to the @xmath43 cross section .
first , we need to sum up the contributions which come in the form of a hadronic tensor coefficient ( all diagrams except @xmath140 ) .
the real - gluon contributions @xmath415 ( [ eq : hg1 ] ) , @xmath416 ( [ eq : hg2 ] ) and @xmath417 ( [ eq : hg3 ] ) from symmetric diagrams have to be counted once .
for all other diagrams , there are corresponding symmetric diagrams , so the virtual - gluon contributions @xmath418 ( [ eq : h1a ] ) , @xmath419 ( [ eq : h1b ] ) , @xmath420 ( [ eq : h1c ] ) , @xmath421 ( [ eq : h2a ] ) and @xmath422 ( [ eq : h3a ] ) , the renormalization and self - energy contributions @xmath423 ( [ eq : h1de ] ) , @xmath424 ( [ eq : h1f ] ) and @xmath425 ( [ eq : h1 g ] ) , and the real - gluon interference contributions @xmath426 ( [ eq : hg4 ] ) , @xmath427 ( [ eq : hg5 ] ) and @xmath428 ( [ eq : hg6 ] ) have to be symmetrized before adding them together .
this is done by replacing pairs of coupling factors as specified in ( [ eq : xlyr ] ) .
the contribution of each diagram ( with hadronic tensor coefficient @xmath339 ) to the unpolarized @xmath429 cross section is then given by @xmath430 where @xmath431 is the electromagnetic coupling and @xmath432 and @xmath433 are the masses of the gauge bosons to the left - hand and right - hand sides of the hadronic tensor ; the latter are attached to the @xmath15 initial state via couplings @xmath434 and @xmath435 , defined analogously to ( [ eq : hcouplings ] ) .
the contribution ( [ eq : deltasigmah ] ) has to be summed over the four combinations of gauge bosons ( @xmath436 ) and , in addition , it must be counted twice , because for every contribution with @xmath300 cut presented here , an equivalent one with @xmath437 cut exists .
the result from the correction @xmath438 ( [ eq : sigmah4a ] ) must be counted four times ( symmetrization and counting both cuts @xmath439 and @xmath440 ) . after adding this piece , the following complete endpoint - singular non - resonant nnlo contribution to the cross section results : @xmath441 \ , \biggl\ { 2 { c_{\text{f}}}\ , \frac{\alpha_s}{\pi } \ , \ , \frac{m_t^2}{\lambda^2 } \ , \nonumber \\
& \qquad{}\qquad{}\qquad{}\quad { } + \frac{2\sqrt{2}}{\pi^2 } \ , \frac{m_t}{\lambda } \ , \big ( \delta \gamma_t^{(1 ) } - 4{c_{\text{f}}}\ , \frac{\alpha_s}{\pi } \big ) \biggl\ } \nonumber \\ & \quad\quad\;\ ; + \biggl\ { - \bigl [ q_t^2 \ , c_{\gamma\gamma}(s ) - 2 q_t v_t \ , c_{\gamma z}(s ) + v_t^2 \ , c_{zz}(s ) \bigr]\ , \frac{7 + 7x+22x^2}{6(1-x ) \ , ( 1 + 2x ) } \nonumber \\ & \quad\qquad\quad\ ; + \frac{1}{3 } a_t^2 \ , c_{zz}(s ) + \frac{1}{2 } q_t a_t \ , c_{\gamma z}(s ) - \frac{1}{2}v_t a_t \ , c_{zz}(s ) \nonumber \\ & \quad\qquad\quad\ ; + \bigl [ q_t q_b \ , c_{\gamma\gamma}(s ) - \bigl ( q_t \ , ( v_b+a_b ) + q_b v_t \bigr ) \ , c_{\gamma z}(s ) + v_t \ , ( v_b+a_b ) \ , c_{zz}(s ) \bigr ] \nonumber \\ * & \qquad{}\qquad{}\qquad{}\quad{}\times \frac{1 - 5x-2x^2}{6(1+x ) \ , ( 1 + 2x ) } \nonumber \\ & \quad\qquad\quad\ ; + \left [ q_t \ , c_{\gamma\gamma}(s ) - \left ( v_t + q_t \ , \frac{{c_{w}}}{{s_{w } } } \right ) c_{\gamma z}(s ) + v_t \ , \frac{{c_{w}}}{{s_{w } } } \ , c_{zz}(s ) \right ] \frac{2 + 5x-2x^2}{6x \ , ( 1 + 2x ) } \nonumber \\ & \quad\qquad\quad\ ; - \bigl [ q_t \ , c_\gamma(s ) + v_t \ , c_z(s ) \bigr ] \ , \left [ \ln\!\left(\frac{2}{x}-1\right ) + \frac{(1-x ) \ , ( 1 - 2x-23x^2)}{12x^2 } \right ] \nonumber \\ * & \qquad{}\qquad{}\qquad{}\quad { } \times \frac{x}{4(1-x)^3 \ , ( 1 + 2x ) } \,\biggr\}\ , { c_{\text{f}}}\ , \frac{\alpha_s}{2\pi } \,\left ( \frac{1}{{\epsilon}_{\text{ep } } } + 2 \ln\frac{\mu_{\rm soft}^2}{\lambda^2 } \right ) \biggr\ } \,,\end{aligned}\ ] ] where we have used the functions @xmath442 which arise from the photon and @xmath109-boson propagators , and dropped @xmath443 terms . in order to improve the quality of the eft expansion , we have kept @xmath329 general in all photon and @xmath109 propagators as well as in the kinematic factors of the cross section and set @xmath118 only in the non - trivial parts of the loop and phase - space integrations , as done in @xcite .
we comment below on the introduction of the scale @xmath444 in ( [ eq : sigmatot ] ) .
the coefficient @xmath445 in the @xmath236 term of ( [ eq : sigmatot ] ) is equal to @xmath446 \,,\end{aligned}\ ] ] which agrees with the first - order qcd correction to the top decay width neglecting the bottom mass , first obtained in @xcite .
@xmath445 is given by the sum of the hard - region contributions from diagrams @xmath171 , @xmath175 , @xmath374 , @xmath375 , @xmath378 and from one half of @xmath447 .
the hard - region contribution from diagrams @xmath172 and @xmath173 is actually equal to the top - field renormalization contribution from the counterterms of both vertices adjacent to the top line ( @xmath448 multiplied with the nlo diagram @xmath134 ) .
the above - mentioned set of diagrams provides virtual and real corrections which only affect the upper top line . in the on - shell top case
which is effectively taken by the leading hard - region contribution , these corrections precisely correspond to the @xmath41 corrections to the on - shell decay process @xmath336 . in the result ( [ eq : sigmatot ] ) ,
@xmath445 multiplies exactly the endpoint - divergent @xmath449 term in the nlo non - resonant result of @xcite , such that @xmath450 may be dropped from our nnlo result if the top width in the nlo result is replaced by its @xmath19-corrected version @xmath451 .
( [ eq : sigmatot ] ) is the main result of this work . as mentioned before ,
all ultraviolet ( @xmath337 ) and infrared ( @xmath338 ) singularities are canceled ( independently of each other ) in this total contribution .
the only remaining poles are found at order @xmath296 ; they represent @xmath182 endpoint singularities and originate purely from the potential region .
when the @xmath165 finite - width divergences ( [ divsigmatotal ] ) from the nnlo resonant contributions in nrqcd are added to the nnlo non - resonant result ( [ eq : sigmatot ] ) , we observe a total cancellation of the @xmath165 divergences . and
@xmath452 , as done in the calculation of the absorptive matching coefficients @xmath68 @xcite , in order to find an exact cancellation . ] for the cancellation of the dependence on the renormalization scale @xmath453 associated with the finite - width divergences we have replaced consistently @xmath454 in the nnlo non - resonant amplitudes and taken @xmath455 .
this is justified by the fact that our @xmath182 singularities originate from one hard integration ( associated with the decay @xmath456 ) and two integrations with contributions from a smaller scale @xmath457 , stemming from the on - shell limit of the top quarks and from potential - region gluon momenta .
also in the resonant nnlo amplitude , one of the subgraphs always corresponds to a hard contribution , _
i.e. _ to a matching coefficient between qcd and nrqcd which is evaluated with @xmath455 ( and in @xmath458 dimensions ) . by setting consistently @xmath455 in our contributions , too , the explicit dependence on @xmath444 associated with the top - quark instability in the nnlo resonant contributions ( _ i.e. _ the @xmath459 terms proportional to @xmath460 or @xmath68 ) cancels against the @xmath444-dependence of the nnlo non - resonant endpoint - singular contributions ( [ eq : sigmatot ] ) , together with the cancellation of the @xmath165 poles .
the remaining logarithms of kinematic scales in this combination of resonant and non - resonant pieces for the inclusive @xmath18 cross section with @xmath8 invariant - mass cuts are of the form @xmath461 , in addition to logarithms of the velocity in the resonant eft matrix elements .
our result ( [ eq : sigmatot ] ) for the non - resonant contributions to the total cross section with an invariant - mass cut @xmath462 on the @xmath8 subsystems was obtained for the case of loose cuts , @xmath463 .
the condition of loose cuts implies that the invariant - mass constraint only enters in the computation of the non - resonant contributions , while matrix elements on the resonant side are unaffected , as explained in refs .
the endpoint - divergent terms presented here are equivalent to the first terms in the expansion in @xmath42 of the full nnlo non - resonant result . interestingly enough
, an alternative approach @xcite ( named _ phase - space matching _ ) has determined the matching coefficients of the four - electron operators providing the non - resonant contributions to the cross section with invariant - mass cuts on the @xmath8 pairs in the same range , @xmath464 , but through calculations involving only the matrix elements in the non - relativistic effective theory . in the phase - space matching ( psm )
approach , symmetric cuts on the invariant masses of the top and antitop decay products restrict the integration over the top and antitop momenta in the resonant diagrams contributing to ( [ eq : ares ] ) .
then the resulting cut integrals are expanded assuming that @xmath125 , but still smaller than @xmath69 so that the non - relativistic expansion carried out by nrqcd is still valid , yielding @xmath465\bigg(\alpha_s \frac{m_t}{\lambda}\bigg)^k + \ , \sigma_{\rm nrqcd}(\infty ) \,,\ ] ] where @xmath466 is the nrqcd resonant cross section without invariant - mass restrictions ( equivalently , for @xmath467 ) . the first term in ( [ eq : psm ] ) with powers of @xmath7
can be understood as minus the contributions to the nrqcd matrix elements from the regions where the invariant - mass constraints are not fulfilled , which are then subtracted from the unrestricted cross section by means of ( [ eq : psm ] ) .
the powers of @xmath468 in ( [ eq : psm ] ) arise from cut diagrams with relativistic corrections , while factors of @xmath469 are introduced by cut diagrams with coulomb - like gluons .
@xmath470 is a hard coefficient coming , for instance , from the matching of the vector current that produces the non - relativistic @xmath18 .
assuming for the power - counting that @xmath471 , the nlo terms in the phase - space matching approach correspond to the terms @xmath472 in ( [ eq : psm ] ) with @xmath473 , while at nnlo we have to retain those with @xmath474 and @xmath475 plus the @xmath476 ) correction to @xmath477 times the nlo term .
the whole computation is equivalent to the non - relativistic expansion of the full - theory squared matrix elements containing the double - resonant diagrams for @xmath478 and their interference with the diagrams for @xmath325 having only either the top or the antitop in intermediate stages .
however , the full - theory contributions coming from the square of single - top and pure background diagrams ( the so - called remainder contributions in @xcite ) can not be reproduced by eft resonant diagrams , and thus have to be computed with external tools in the phase - space matching approach . for moderate invariant - mass cuts
it was shown in @xcite through a numerical comparison with the full - theory tree - level @xmath325 cross section that the remainder contributions are small at nlo .
indeed , in the computation of non - resonant corrections at this order within the unstable - particle eft ( _ i.e. _ the @xmath472 terms in ( [ eq : psm ] ) ) , the remainder contributions are reproduced by diagrams @xmath183@xmath152 ( see fig . [ fig1 ] ) that contribute first with @xmath479 terms in the @xmath42-expansion , corresponding to @xmath480 in ( [ eq : psm ] ) .
for the first two nlo terms in the series of ( [ eq : psm ] ) , @xmath481 , it was checked in @xcite that they agree with the corresponding terms in the expansion of the full nlo non - resonant result . under the assumption that the qcd corrections to the remainder contributions are also small in the above - mentioned range of invariant - mass cuts , the phase - space matching contributions at @xmath41
were also computed in @xcite .
we can now compare the nnlo terms of the phase - space matching series with those computed in this work , eq .
( [ eq : sigmatot ] ) : the same coefficients in the @xmath394 , @xmath236 and @xmath482 contributions are found from that comparison . in ( [ eq : sigmatot ] )
is generated by including in the nrqcd top - quark propagators the @xmath41 corrections to the top width , which was not done explicitly in the analysis of @xcite because @xmath460 was considered an input parameter there .
the rest of the @xmath236 correction in ( [ eq : sigmatot ] ) arises in the phase - space matching approach from the product of the @xmath41 correction to the vector - current matching coefficient ( @xmath483 ) at both @xmath18-vertices with the nlo phase - space matching contributions . ]
a simple argument can be used to explain why the two methods yield the same series expansion in @xmath6 of the non - resonant contributions , despite the different starting points in the respective calculations . in our method , the full - theory amplitude is expanded first assuming that top - quark lines are off - shell , _
i.e. _ @xmath484 , which drops the top self - energy corrections from the full - theory top - quark propagator , and then the @xmath6 series arises from further expanding the resulting amplitude around the on - shell limit , _
i.e. _ for @xmath485 .
the last expansion is equivalent to considering that the momentum @xmath145 is potential , since the antitop on - shell condition sets @xmath486 . on the other side ,
the psm computation starts from the full - theory amplitude expanded for nearly on - shell ( potential ) top quarks , with @xmath487 , producing @xmath488 and @xmath489 corrections .
then the latter are transformed into @xmath490 terms from hard-@xmath145 momentum regions ( @xmath484 ) where we can expand the integrand taking the off - shellness of @xmath145 much larger than the non - relativistic scales @xmath460 and @xmath491 .
the first important observation is that in both methods a double expansion of the integrand in @xmath145 ( according to hard and potential region scalings ) is performed , but in reversed order . in the method of regions , such double expansions are known as overlap contributions @xcite . adopting the notation of the latter reference
, we can denote as @xmath492 the integrand resulting from expanding the full - theory integrand @xmath493 ( including the dirac delta functions from the cutkosky rules ) according to first hard and then potential momentum , and @xmath494 the same expansion taken in the reversed order , as effectively done in the psm approach . within the expansion by regions , especially in simple cases with only two relevant regions ( here hard and potential ) , double expansions usually yield the same doubly - expanded integrand , whether the two expansions are executed in one order or the other .
so we can expect that the identity @xmath495 holds .
let us consider in the following at first the case without invariant - mass restrictions , so that @xmath493 does not depend on any cut @xmath7 .
then our second important remark is that the integration of the integrand @xmath496 over @xmath145 vanishes in dimensional regularization , @xmath497 because after doubly expanding the integrand according to hard and potential momenta , there is no scale left in the integrand ( recall that we have not imposed any invariant - mass constraints in @xmath493 , so the integration limits are unbounded ) .
if we now rewrite ( [ eq : fhp ] ) using @xmath498 as @xmath499 we readily identify the first term as the taylor series in @xmath6 of the non - resonant contributions computed in our approach for an invariant - mass cut of the form ( [ eq : deltam ] ) .
the second term corresponds to the resonant nrqcd amplitude further expanded assuming hard @xmath145 and integrated outside the region of @xmath120 allowed by the invariant - mass cuts .
that is precisely the quantity that yields minus the series in the first term of ( [ eq : psm ] ) with powers of @xmath7 , obtained in the phase - space matching approach .
therefore , from ( [ eq : cancel ] ) we conclude that the psm contributions must be equal to the series expansion obtained with the unstable - particle eft , as we have shown by explicit computation up to terms of order @xmath482 .
the equivalence between both series is lost at the order where terms coming from full - theory diagrams with just one top or antitop , or with no tops at all , which are not described by nrqcd , first contribute . as mentioned before ,
these terms arise at nlo from diagrams @xmath151@xmath152 , and are of order @xmath500 .
since the @xmath41 corrections to diagrams @xmath151@xmath152 do not introduce negative powers of @xmath141 , we also expect that the nnlo non - resonant contributions in the phase - space matching series start to differ from ours at order @xmath501 .
the reasoning above also provides further insight on the regions ( of gluon momenta ) which contribute to the endpoint - singular nnlo corrections . in
scaleless overlap contributions like ( [ eq : fhp ] ) where the integrand is doubly expanded according to both regions , singularities from domains of hard and potential top momenta @xmath145 cancel each other to yield zero .
exactly the same cancellation of singularities happens between the contributions of the individual ( hard or potential ) regions , originating from the respective integration domain where the top momentum approaches the scaling of the opposite region : the ultraviolet singularities ( from the hard-@xmath145 limit ) present in the integrals of the resonant contribution ( which have been expanded for potential @xmath145 ) are canceled by the endpoint singularities ( from the potential-@xmath145 limit ) of the non - resonant contribution ( whose integrals have been expanded for hard @xmath145 ) .
this cancellation occurs individually for the different scalings of gluon momenta .
it is known that the @xmath165 finite - width divergences on the resonant side originate purely from potential ( coulomb ) gluons .
so their cancellation must be provided purely by potential gluon momenta in the endpoint - divergent non - resonant diagrams .
this is exactly what we have found in our calculation for the origin of the @xmath182 terms from potential - region contributions .
similarly , it is known that ultrasoft gluons do not contribute to the total resonant nrqcd amplitudes at nnlo .
so it comes without surprise that we observe a cancellation of the ultrasoft - region contributions to the endpoint - divergent non - resonant corrections ( see the remarks at the end of sec .
[ subsec : realgluon ] ) .
a different path has been taken by the authors of ref .
@xcite to provide an estimate of the nnlo non - resonant corrections .
they have expanded the nnlo non - resonant contributions to the @xmath18 total cross section in powers of @xmath502 , where @xmath503 , and calculated the first two terms in this expansion . for the nlo non - resonant corrections ,
the leading - order term in @xmath407 has been shown to deviate from our exact result @xcite by less than 5% , despite the fact that the approximation for the individual diagrams is much less accurate .
such a numerical comparison is not possible at nnlo where the exact result is not known , and our approximation , eq .
( [ eq : sigmatot ] ) , does not correspond to the total cross section , but to the cross section with invariant - mass cuts applied in the @xmath8 pairs .
we notice though important differences between the approach followed in @xcite and ours for the nnlo result . according to the authors of @xcite the leading - order term in @xmath407 of the nnlo non - resonant contributions arises from diagram @xmath163 and contains an energy - dependent term of the form @xmath504 .
the latter is generated in their approach because the scale @xmath505 in the top - quark propagators is kept as an infrared regulator for the ( infrared - divergent ) loop integration of the non - resonant amplitude , despite the fact that the integral is saturated by the region @xmath506 .
let us remark that in our approach logarithms of the velocity can only appear in the resonant matrix elements , because the hard - momentum expansion performed in the non - resonant diagrams drops terms of order @xmath507 from the top - quark propagators and , as a consequence , an energy - dependence there only arises beyond nnlo as powers of @xmath491 .
the infrared divergence in @xcite is analogous to the @xmath182 endpoint divergences we find within the unstable - particle eft formalism , and must be compensated by an ultraviolet divergence arising from a resonant contribution in order to render the leading - order result in the @xmath407-expansion independent of the regularization scheme .. while the nnlo relativistic corrections to the resonant diagram with a coulomb potential , fig .
[ figcoulomb ] , only give terms scaling as @xmath508 , the corrections arising from the insertion of the ( electroweak ) absorptive matching coefficients @xmath68 into fig .
[ figcoulomb ] yield terms scaling also as @xmath509 .
part of the latter provides the counterpart for the infrared divergence in the leading - order result from @xcite . ]
the need for an ultraviolet counterpart for the result of @xcite becomes even clearer if the infrared divergence on the non - resonant side is regulated dimensionally , which substitutes the @xmath510 above by a @xmath511 plus ( potentially ) additional finite terms , thus introducing an explicit scheme - dependence through the renormalization scale @xmath332 .
we can easily do this exercise by taking the integral form of the leading - order term provided in eq .
( 4.1 ) of @xcite and performing it in @xmath512 dimensions . with the appropriate normalization factor for the cross section
, this contribution reads @xmath513 \ , \frac{{n_{\text{c}}}\gamma_t^{\rm born}}{m_t } \ , \delta_{1a}^{(1 ) } \big|_{\rho^{-1 } } \,,\end{aligned}\ ] ] with @xmath514 \frac{{c_{\text{f}}}\alpha_s}{\rho } \,.\end{aligned}\ ] ] as expected , the coefficient of the infrared - divergent @xmath165 term in @xmath515 above agrees with the @xmath182 term which arises from our result for diagram @xmath163 , eq .
( [ eq : h1a ] ) plus symmetric contributions , expanded at the leading order for small @xmath407 ( note that @xmath516 and that the coefficient of @xmath517 in ( [ eq : h1a ] ) is proportional to @xmath518 ) .
a quick inspection of the rest of the @xmath182 endpoint - divergent results computed in sec . [ sec : results ] reveals that diagram @xmath169 also contains an infrared @xmath182 divergence proportional to @xmath519 .
it can also be checked that the sum of the @xmath520 divergences from diagrams @xmath163 and @xmath169 ( including symmetric contributions ) cancels against the first term in the @xmath407-expansion of the finite - width divergence on the resonant side , eq .
( [ divsigmatotal ] ) .
the contribution from diagram @xmath169 has not been considered , however , in the analysis of the @xmath521 term of @xcite .
we therefore conclude that in its present form the leading - order nnlo non - resonant result of @xcite is incomplete , and can not be consistently added as a correction to the known nnlo contributions from the resonant side .
finally , for implementing a loose cut on the @xmath8 invariant masses in the integrals for the nlo non - resonant contributions expanded in the parameter @xmath407 , the authors of @xcite have provided the replacement rule @xmath522 applied to the arguments of the theta - functions of their integrals . adopting the same rule for the @xmath41 contribution that yields the leading non - resonant term at nnlo in the approach of @xcite , the same @xmath394 term as in our result ( [ eq : sigmatot ] ) is obtained .
our subleading terms of order @xmath236 can not directly be compared to @xcite .
in this final section we compare numerically the first terms in the @xmath6-series of the nnlo non - resonant corrections with the nlo ones as well as with the leading - order eft approximation .
this is done for the @xmath523 cross section with invariant - mass cuts in the @xmath8 subsystems of the form ( [ eq : deltam ] ) , where @xmath124 .
the nnlo non - resonant terms are given by the endpoint - singular contributions computed in sec .
[ sec : results ] , and have been collected in eq .
( [ eq : sigmatot ] ) .
the nlo non - resonant corrections , @xmath524 , are obtained from the results given in @xcite , which are too lengthy to be reproduced here .
we present nlo and nnlo contributions separately in this section , _
i.e. _ `` nnlo '' always refers to the pure second - order corrections without including the nlo result .
a dependence on the invariant - mass cut @xmath525 enters first at nlo through the non - resonant contributions for the case of loose cuts ( _ i.e. _ @xmath126 ) . the leading - order cross section @xmath526
is given entirely by the leading - order resonant contribution ( [ eq : ares ] ) , which sums coulomb corrections proportional to @xmath527 to all orders in the strong coupling .
its analytic expression , following the same conventions as in this paper , can be found in @xcite .
@xmath526 depends on the renormalization scale @xmath444 only through @xmath528 .
cross section computed at @xmath118 as a function of the invariant - mass cut @xmath529 .
the solid ( blue ) line corresponds to @xmath530 in ( [ eq : sigmatot ] ) dropping the @xmath182 term , and using @xmath531 with @xmath532 .
the shaded ( blue ) band shows the same result varying @xmath444 in the interval 1560 gev . for comparison ,
the nlo non - resonant contribution ( dashed red line ) , @xmath524 from @xcite , is also shown.,scaledwidth=80.0% ] figure [ fig4 ] shows the contribution to the cross section from the nnlo endpoint - singular terms as a function of the invariant - mass cutoff exactly at the threshold ( @xmath118 ) .
the nnlo contribution corresponds to @xmath530 in ( [ eq : sigmatot ] ) with the endpoint divergence @xmath182 removed .
it uses @xmath533 plus the standard - model input parameters @xmath534 @xmath535 whereas the on - shell weinberg angle @xmath536 and the fine - structure constant in the @xmath537-scheme , @xmath538 , are derived quantities .
the solid ( blue ) line is obtained for @xmath539 gev , where @xmath540 , whereas the shaded band displays the effect of varying the scale @xmath444 from 15 to 60 gev ( lower values of the scale corresponding to more negative contributions ) .
the dashed ( red ) line is the nlo non - resonant contribution @xmath524 from @xcite which is also shown for comparison .
both the nlo and nnlo non - resonant corrections give a negative shift .
the values shown in fig . [ fig4 ] range from @xmath541 gev up to the maximum value allowed by the kinematics , @xmath542 gev , which corresponds to the total cross section .
recall that the nnlo non - resonant terms computed in this work are a valid description for moderate invariant - mass cuts , satisfying @xmath543 .
for tight cuts ( @xmath544 ) , which are not studied here , the expansion by regions dictates that the dependence on @xmath525 is taken into account in the resonant part of the amplitude , while the non - resonant contributions are absent in this case ( see @xcite ) .
the further requirement @xmath545 is a consequence of the expansion around the endpoint that produces the result ( [ eq : sigmatot ] ) for the non - resonant contributions . from fig .
[ fig4 ] we observe that in a moderate @xmath525-range ( @xmath546 ) the nnlo non - resonant corrections are always smaller ( in absolute value ) than the nlo ones , and ( because of the higher singularity in @xmath7 or @xmath525 ) they become more negative when the available phase space for the @xmath8 pairs gets restricted by tightening the invariant - mass cut .
the ratio between the nnlo and nlo non - resonant contributions ranges approximately from 0.9 to 0.5 for @xmath525 in the interval @xmath547 gev .
the relative size of the endpoint - singular nnlo corrections with respect to the lo cross section as a function of the centre - of - mass energy in the threshold region is displayed in figure [ fig5 ] for two different values of the invariant - mass cut , @xmath548 gev ( upper blue solid line ) and @xmath549 gev ( upper blue dashed line ) .
the corresponding curves for the nlo result are also shown ( lower black lines ) . at threshold energies , @xmath550 , the non - resonant corrections are almost energy - independent , with a mild linear energy - dependence introduced from the @xmath109 and photon propagators .
the nnlo non - resonant corrections for the chosen values of @xmath525 give a constant negative shift of about @xmath551 above the threshold where the lo cross section is also constant . below the peak region , where the lo result vanishes rapidly , the relative size of the non - resonant corrections is very large , up to 10% for @xmath548 gev in the energy range shown in fig .
[ fig5 ] . in absolute value , the endpoint - singular nnlo non - resonant corrections amount to for @xmath549 gev and for @xmath548 gev when the centre - of - mass energy is varied within the interval @xmath552 gev .
lo cross section in percent : @xmath553 ( upper blue lines ) and @xmath554 ( lower black lines , from @xcite ) .
solid ( dashed ) lines correspond to an invariant - mass cut @xmath548 gev ( @xmath549 gev ) .
the renormalization scale in the nnlo non - resonant contribution has been set to @xmath539 gev , and we have chosen @xmath555 for the value of the qcd coupling that enters also in the lo result.,scaledwidth=80.0% ] for an analysis of the impact of higher - order non - resonant corrections for moderate invariant - mass cuts not considered in the present work , namely the non - logarithmic terms of order @xmath296 and a large part of the n@xmath24lo contributions , we refer the reader to @xcite .
the corrections induced by off - shell top - quark decay and by other non - resonant production processes of the physical @xmath45 final state are a missing piece towards the prediction of the threshold top - quark pair production cross section at the third order ( nnlo ) .
phenomenologically , they are needed to match the accuracy of the well - known qcd corrections to the resonant production , as required for the precision attainable in the top - mass determination at a future @xmath15 collider .
the computation of the full set of nnlo non - resonant corrections represents a non - trivial task , involving @xmath556 diagrams that include 1-loop virtual - qcd as well as tree - level gluon - radiation corrections to the @xmath557 final state . at the theory level ,
the nnlo non - resonant corrections are also mandatory , because the purely resonant cross section contains at the same order an uncanceled ultraviolet divergence , @xmath558 ( [ divsigmatotal ] ) , which must be compensated by a divergence with opposite sign on the non - resonant side in order to yield a regularization - independent result for this observable . in this work
we have identified the divergences in the nnlo non - resonant amplitudes which provide such a cancellation .
they originate at the endpoint of the phase - space integration over the @xmath8 invariant mass in virtual diagrams with @xmath557 final states .
we have extracted these endpoint divergences through an expansion of the relevant nnlo non - resonant diagrams around the endpoint .
the expanded integrals involve the scale @xmath462 of the invariant - mass cut in the @xmath8 system . in this way , apart from the @xmath165 divergences , we obtain the endpoint - singular terms @xmath3 , @xmath4 and @xmath5 which correspond to the first terms in the expansion in powers of @xmath6 . this series provides a rigorous approximation of the nnlo non - resonant contributions to the @xmath325 cross section with symmetric invariant - mass cuts of size @xmath559 applied to the @xmath8 pairs , as long as @xmath525 is much smaller than the top mass but significantly larger than the top width ( equivalently , if @xmath560 ) .
our analytic result agrees with the one obtained for the same observable within the phase - space matching approach @xcite . on the other hand , by comparing the infrared structure of our result with the one obtained in @xcite at leading order in the expansion for small @xmath561 , we conclude that the latter misses one contribution and that it can not be combined with the nnlo resonant corrections in a regularization - scheme independent way .
numerically we find that within the above range of invariant - mass cuts the nnlo non - resonant contributions produce a negative shift of about @xmath551 with respect to the leading - order @xmath18 cross section above threshold . for energies below the peak
, where non - resonant production is known to dominate over the ( subleading ) resonant terms , the corrections reach up to @xmath562 .
the presence of endpoint singularities in the nnlo non - resonant contributions represents an additional complication for their calculation , since these singularities have to be subtracted from the amplitude , together with the standard soft - collinear divergences due to gluon radiation .
the analysis performed in this work , which identifies and evaluates the endpoint - singular terms , thus provides a necessary step towards the computation of the full set of nnlo non - resonant contributions to @xmath39 and , consequently , towards having a complete nnlo theoretical prediction for @xmath18 production near threshold that accounts for the effects related to top - quark decay in a consistent manner .
we thank m. beneke for helpful discussions and for his comments on the manuscript .
we also thank a. a. penin for discussions on the comparison with their result .
this work is supported by the dfg sonderforschungsbereich / transregio 9 `` computergesttzte theoretische teilchenphysik '' .
the work of p. r. is partially supported by mec ( spain ) under grants fpa2007 - 60323 and fpa2011 - 23778 and by the spanish consolider - ingenio 2010 programme cpan ( csd2007 - 00042 ) .
feynman diagrams have been drawn with the packages axodraw @xcite and jaxodraw @xcite .
we acknowledge the use of the computer programs form @xcite and feyncalc @xcite for some parts of the calculation .
k. seidel , f. simon , m. tesar and s. poss , eur .
j. c * 73 * ( 2013 ) 2530 , arxiv:1303.3758 [ hep - ex ] .
t. aaltonen _ et al . _
[ cdf and d0 collaborations ] , phys .
d * 86 * ( 2012 ) 092003 , arxiv:1207.1069 [ hep - ex ] .
atlas collaboration , atlas - conf-2012 - 095 .
i. i. y. bigi , y. l. dokshitzer , v. a. khoze , j. h. khn and p. m. zerwas , phys .
b * 181 * ( 1986 ) 157 .
m. beneke , y. kiyo and k. schuller , nucl .
b * 714 * ( 2005 ) 67 , hep - ph/0501289 .
m. beneke , y. kiyo and k. schuller , pos ( radcor 2007 ) 051 , arxiv:0801.3464 [ hep - ph ] . m. beneke and y. kiyo , phys .
b * 668 * ( 2008 ) 143 , arxiv:0804.4004 [ hep - ph ] .
a. pineda and j. soto , nucl .
* 64 * ( 1998 ) 428 , hep - ph/9707481 .
m. beneke , hep - ph/9806429 .
m. beneke , a. signer and v. a. smirnov , phys . lett .
b * 454 * ( 1999 ) 137 , hep - ph/9903260 .
n. brambilla , a. pineda , j. soto and a. vairo , nucl . phys .
b * 566 * ( 2000 ) 275 , hep - ph/9907240 .
m. e. luke , a. v. manohar and i. z. rothstein , phys .
d * 61 * ( 2000 ) 074025 , hep - ph/9910209 .
a. v. manohar and i. w. stewart , phys .
d * 62 * ( 2000 ) 014033 , hep - ph/9912226 .
a. h. hoang and i. w. stewart , phys .
d * 67 * ( 2003 ) 114020 , hep - ph/0209340 .
a. h. hoang , a. v. manohar , i. w. stewart and t. teubner , phys .
* 86 * ( 2001 ) 1951 , hep - ph/0011254 .
a. h. hoang , a. v. manohar , i. w. stewart and t. teubner , phys .
d * 65 * ( 2001 ) 014014 , hep - ph/0107144 .
a. pineda and a. signer , nucl .
b * 762 * ( 2007 ) 67 , hep - ph/0607239 .
m. stahlhofen and a. hoang , pos ( radcor 2011 ) 025 , arxiv:1111.4486 [ hep - ph ] .
m. beneke , a. p. chapovsky , a. signer and g. zanderighi , phys .
lett . * 93 * ( 2004 ) 011602 , hep - ph/0312331 .
m. beneke , a. p. chapovsky , a. signer and g. zanderighi , nucl .
b * 686 * ( 2004 ) 205 , hep - ph/0401002 .
m. beneke , p. falgari , c. schwinn , a. signer and g. zanderighi , nucl
. phys .
b * 792 * ( 2008 ) 89 , arxiv:0707.0773 [ hep - ph ] . a. h. hoang , c. j. reier and p. ruiz - femena , nucl .
* 186 * ( 2009 ) 403 , arxiv:0810.2934 [ hep - ph ] .
a. h. hoang , c. j. reier and p. ruiz - femena , phys .
d * 82 * ( 2010 ) 014005 , arxiv:1002.3223 [ hep - ph ] .
a. a. penin and j. h. piclum , jhep * 1201 * ( 2012 ) 034 , arxiv:1110.1970 [ hep - ph ] . m. beneke and v. a. smirnov , nucl .
b * 522 * ( 1998 ) 321 , hep - ph/9711391 .
v. a. smirnov , _ applied asymptotic expansions in momenta and masses _ , vol .
177 of _ springer tracts in modern physics _ , springer , germany ( 2002 ) .
b. jantzen , jhep * 1112 * ( 2011 ) 076 , arxiv:1111.2589 [ hep - ph ] . m. jezabek and j. h. khn , nucl .
b * 314 * ( 1989 ) 1 .
s. actis , m. beneke , p. falgari and c. schwinn , nucl .
b * 807 * ( 2009 ) 1 , arxiv:0807.0102 [ hep - ph ] .
j. a. m. vermaseren , comput .
commun .
* 83 * ( 1994 ) 45 .
d. binosi , j. collins , c. kaufhold and l. theussl , comput .
commun .
* 180 * ( 2009 ) 1709 , arxiv:0811.4113 [ hep - ph ] .
j. a. m. vermaseren , math - ph/0010025 .
r. mertig , m. bhm and a. denner , comput .
commun .
* 64 * ( 1991 ) 345 . | we analyse the subleading non - resonant contributions to the @xmath0 cross section at energies near the top antitop threshold .
these correspond to next - to - next - to - leading - order ( nnlo ) corrections with respect to the leading - order resonant result .
we show that these corrections produce @xmath1 endpoint singularities which precisely cancel the finite - width divergences arising in the resonant production of the @xmath2 final state from on - shell decays of the top and antitop quarks at the same order .
we also provide analytic results for the @xmath3 , @xmath4 and @xmath5 terms that dominate the expansion in powers of @xmath6 of the complete set of nnlo non - resonant corrections , where @xmath7 is a cut imposed on the invariant masses of the @xmath8 pairs that is neither too tight nor too loose ( @xmath9 ) .
ttk-13 - 17 + sfb / cpp-13 - 43 + ific/13 - 29 + 1.5 cm * nnlo non - resonant corrections to threshold top - pair production from @xmath10 collisions : endpoint - singular terms * b. jantzen@xmath11 and p. ruiz - femena@xmath12 + _ @xmath13institut fr theoretische teilchenphysik und kosmologie , + rwth aachen university , + d-52056 aachen , germany _ + _ @xmath14instituto de fsica corpuscular ( ific ) , csic - universitat de valncia + apartado de correos 22085 , e-46071 valencia , spain _ + |
the cosmic microwave background ( cmb ) is an almost perfect , nearly isotropic blackbody radiation now at a temperature of 2.725 k emanating from the big bang itself . as a practical matter , it is the oldest fossil remnant of the primordial universe available to us for precision characterization .
the precise mapping of the cmb anisotropies in both temperature and polarization has allowed us to reconstruct the initial conditions of the universe .
as we have learned , these initial conditions are remarkably simpler than one might have imagined .
twenty years ago people wanting to model how the recent universe evolved from initial conditions were forced to insert in their codes initial conditions that were at best tentative guesses , but today we believe that we know what the correct initial conditions were , with uncertainties in the parameters presently at about the few percent level .
today the greatest uncertainty in such studies of how the late universe evolved out of these initial conditions arises from the modeling of the late time physics itself .
while the physics of the cmb is linear and simple , the late time physics of the universe is nonlinear and complex .
processes such as gravitational clustering , hydrodynamics , magnetogenesis , mhd , star formation , supernova explosions and their aftermath , to give an incomplete list must be modeled , either by starting from first principles or in many cases by using more _ ad hoc _ models and relying on late time observational data to fix the parameters of these phenomenological models .
the cmb by contrast offers a relatively clean probe of the state of the universe before nonlinear physics kicked in .
even the quantum field theory that enters into computing the generation of the primordial density during a prior epoch of inflation is conceptually relatively simple : it is , for the most part , free field theory , which some quantum field theorists would regard as classical , because it is tree level , or order @xmath0 in the loop expansion . in other words , one takes the modes of the classical field theory , normalizing them according to classical canonical commutation relations , and one replaces the @xmath1-number coefficients with annihilation and creation operators , just as for the simple harmonic oscillator .
the study in the 1970s and 1980s of semi - classical quantum field theory in a curved spacetime ( see for example the book by birrell and davies@xcite and references therein ) raised some interesting new issues and became one of the principal theoretical tools for the subsequent calculation in the mid-1980s of the perturbations predicted from the leading order quantum corrections to the classical inflationary cosmology .
this program requires some courage .
we do not know the correct theory of quantum gravity , but we linearize the classical theory and proceed to quantize this linearized theory making predictions as if we had a theory of quantum gravity .
perhaps one of the virtues of inflationary cosmology as developed in the 1980s is that it protects us from our ignorance .
inflation provides an ingenious mechanism for hiding any clues as to what happened during a possible earlier planck epoch , presumably governed by strongly coupled gravity at the quantum level , for which we today have hardly any idea as to the character of the underlying theory . to be sure , we have superstring theory , which is very promising at the formal level .
but we do not know how to calculate predictions relevant for the very early universe .
cosmic inflation is an incomplete theory .
some call it the inflationary paradigm , " but i do not like this term , which alludes to thomas kuhn s paradigm shifts , " which are momentous and revolutionary discoveries after which science emerges not the same as before .
however the older meaning of paradigm , " referring to latin grammar , is less glorious . according to webster s third international dictionary,@xcite
a paradigm is an example of a conjugation or declension showing a word in all its inflectional forms . "
unfortunately , the number of inflationary potentials that have been proposed in the literature is immense .
the problem is that the theory does not make a definite prediction for the form of the inflationary potential , so there are , at least technically speaking , almost an infinite number of free parameters , corresponding to a free function .
we can , however , make predictions by assuming the absence of ` features , ' or bumps and dips in the potential .
assuming that the inflationary potential is relatively smooth and well behaved , we can , for example , expand the inflationary potential as a power series within the observable region where the cosmological perturbations visible to us today were imprinted , and retaining only the first few terms , we obtain approximate predictions .
but there is something unsatisfying about the lack of an unambiguous base theory .
we can speak of the emergence of a standard model of cosmology , but the analogy to the standard electroweak model of weinberg and salam is imperfect . in that case , there are precisely 26 undetermined parameters , @xmath2 and @xmath3 ) , 2 higgs potential coupling constants , 3 charged lepton masses , 6 quark masses , 3 neutrino masses , 4 parameters for the ckm quark mixing matrix and 4 more for the neutrino mixing matrix , and finally @xmath4 and when we talk about physics beyond the standard model , " we know exactly what we are talking about . for my talk
i was asked to provide an overview of the observations of the cosmic microwave background ( cmb ) including its history , the present state of our knowledge of the cmb , and also some indications about the future of this field with a description of the experiments now in progress or being planned , as well as describing some of the challenges that must be overcome in order to make the observations that we would like to make .
this is a broad remit , and what is included here constitutes an arbitrary selection , reflecting my own personal biases . for a more detailed and complete overview ,
the reader is invited to consult my review article,@xcite and in particular the many references cited therein .
here i will try to emphasize some lesser known aspects of the story of the cmb , which has become part of the canon of modern cosmology , referring the reader elsewhere for a more complete account.@xcite the history of the cmb reaches back much earlier than most people realize , with key observations pre - dating the the prophetic set of papers around 1948 of which the most cited is alpher , bethe , and gamow,@xcite in which a cmb temperature of approximately 5 k was predicted . the observational evidence for the cmb with a measurement of its temperature already existed in studies of the relative intensities of absorption lines in spectra of nearby hot stars taken at optical frequencies .
although the primary optical transitions are electronic , the electronic levels are split into both vibrational and rotational sublevels .
the relative populations of these sublevels can under the right conditions serve as molecular thermometers that can be used to measure the thermodynamic brightness temperature of the radiation field at the frequency corresponding to rotational transitions between the sublevels . in the general case ,
this procedure does not work , because several competing effects enter into determining the relative level populations ( e.g. , collisions , excitation by uv photons , etc . ) , each process being characterized by its own effective temperature .
however , to the extent that the coupling to the radiation field dominates over other effects , the relative level populations provide an accurate measure of the radiation field temperature .
the three molecules originally studied were the diatomic molecules @xmath5 @xmath6 and @xmath7 the expectation based on laboratory experiments carried out at room temperature was that a whole series of ` bands ' resulting from rotational - vibrational splittings into sublevels should be visible for all these molecules .
but instead only two sublevels were observed for the cn molecule , corresponding to the rotational @xmath8@xmath9 splitting of the ground state , and for @xmath6 and @xmath10 only a single line was observed , suggesting that all these molecule are in their rotational ground state .
the @xmath8@xmath9 rotational transitions for the @xmath5 @xmath6 and @xmath10 molecules correspond to the frequencies 228 ghz , 535 ghz , and 834 ghz ( or the wavelengths 1.32 mm 0.56 mm , and 0.36 mm ) , respectively . for comparison ,
the modern value of the cmb temperature converted to a frequency is @xmath11 ghz . for the cn relative populations
, mckellar@xcite deduced an effective temperature of around 2.1 k ; however , he did not go further to analyze the possible consequences of his finding .
it was only after the discovery by penzias and wilson that the coincidence of their direct microwave temperature measurement with the temperatures recorded using these molecular thermometers was noted .
it was then recognized that the cmb could have been discovered earlier if these measurements had been correctly interpreted .
having described the first observation of the cmb , we now go on to discuss the first theoretical work by alpher , bethe , and gamow ( although bethe s name was added only to complete the pun ) .
it should be noted that the work by alpher , bethe , gamow , and hermann received little attention at the time , partially for reasons that in retrospect are not unreasonable . today
` primordial nucleosynthesis ' is regarded as one of the ` observational pillars ' of the hot big bang model , but the observational data needed to arrive deductively at this conclusion did not exist at the time . in 1948 the challenge was to explain the abundances of all the elements seen in the present universe , ranging from hydrogen to iron and beyond , and primordial nucleosynthesis predicted that the baryons in the universe would end up almost exclusively in the form of @xmath12h and @xmath13he , with literally trace amounts of a handful of other light elements : @xmath14d , @xmath15he , @xmath16li , and @xmath16be . around 1948
most researchers looked toward processes within stars for producing the heavier elements .
they were not aware that stars alone could not account for the entire observed helium abundance and that something more than nuclear burning in stars was required to explain the observed relative isotope abundances .
it is ironic that in the 1990s before the hot big bang cosmology had become thoroughly incorporated into the received wisdom , some critics cited as one of their arguments against the big bang the lack of agreement of the 1948 alpher , gamow , and hermann prediction of a 5 k cmb temperature@xcite with the modern observed value of 2.725 k. yet if we examine this prediction in light of the knowledge at the time and redo the theoretical analysis more carefully , we arrive at the conclusion that alpher et al .
were lucky not to have been off by an even greater factor .
it turns out that unless one does extremely sensitive measurements , which have been carried out in the meantime , the basic prediction that about a quarter of the baryonic mass is in the form of @xmath13he results over an extremely broad range of values for the input parameter @xmath17 ( the baryon - to - photon number density ratio ) , which is the only adjustable free parameter of primordial nucleosynthesis . at the time , attention was not so much focused on the possibility of primordial nucleosynthesis but more on the nuclear processes in stars , especially for generating the heavier elements .
it was not then evident that both primordial and stellar nucleosynthesis were required to explain the current observed relative element abundances .
in many accounts in the secondary literature of the discovery of the cmb by penzias and wison in 1965 the word ` serendipitous ' ( not exactly among the most commonly used english words ) pops up over and over .
while there is some truth in this description , in reality the discovery of the cmb was less accidental than sometimes depicted .
the frontiers of electronics and communication have greatly shifted since the mid-1960s .
low noise and high frequency had quite a different meaning then , and the problem penzias and wilson had set out to explore was the ultimate limits of noise for communication in the frequency range around 4 ghz .
they embarked on the systematic study of this question in an effort spanning many years and exploring many possible sources .
the account depicting of dicke s princeton experiment on the verge of discovering the vestiges of the big bang but being scooped by some folks at bell labs who did not really know what they were doing does not do justice to the contribution of penzias and wilson . at nearly the same time , robert dicke and collaborators at princeton , unaware of the prior work by gamow and collaborators , had speculated that there should be a cmbr and had constructed an apparatus to measure its signal .
however before they could complete their measurement , the news arrived that penzias and wilson had detected the signal that they had set out to seek , and the outcome was that two papers were published back to back in the astrophysical journal , one by penzias and wilson , preceded by a paper by dicke , peebles , roll and wilkinson explaining how this signal could be explained in a hot big bang cosmological model .
there is a long gap between the discovery of penzias and wilson of the cmb at zeroth order and the discovery of the cmb anisotropy by the cobe dmr in 1992 , which was accompanied by what remains to date the best measurement of the cmb frequency spectrum by the cobe firas instrument .
this long gap of almost thirty years can largely be understood by the fact that the technology to make the necessary measurements was lacking and not that the importance of looking for anisotropies and deviations from a blackbody spectrum had not been recognized .
a detailed history of the pre - cobe experiments searching for a possible anisotropy in the cmb temperature and for deviations from the blackbody form for the frequency spectrum can be found in the reviews by weiss@xcite and by readhead and lawrence.@xcite the technical challenges to be overcome to carry out measurements at the required accuracy were formidable , and it took quite some time for the upper limits to reach the level of the actual cmb anisotropy as finally discovered by cobe . in 1992
the cobe team announced the first discovery of an anisotropy of the cosmic microwave background using the dmr ( differential microwave radiometer ) instrument.@xcite the dmr consisted of pairs of horns pointing in directions separated by approximately @xmath18 in the sky .
the detectors were rapidly switched between the horns and only the difference in signal was recorded . through the spinning and precession of the satellite , a data stream consisting of differences in the cmb temperature was recorded at three frequencies and sky maps were constructed using this data . with the data smoothed to @xmath19 the cobe team found an rms anisotropy of @xmath20 ( or @xmath21 for the primordial blackbody signal .
the shape of the spectrum was found consistent with scale invariance ( although with large uncertainties , i.e. , @xmath22 ) as predicted by inflationary models .
the cobe firas instrument probed the absolute frequency spectrum of the cmb by rapidly switching a fourier transform spectrometer between the sky and an artificially constructed blackbody source .
cobe firas constrained the rms deviations of the cmb from a perfect blackbody to be less than 5 parts in @xmath23 over the frequency range 60 630 ghz.@xcite competing theoretical cosmological models had an undetermined or floating normalization , but when normalized to cobe , a prediction for the normalization of the primordial power spectrum of density fluctuations followed , usually expressed in terms of @xmath24 ( i.e. , the rms fractional mass fluctuation @xmath25 within a sphere of radius @xmath26 ) , could be obtained , and this predicted @xmath24 could be compared to data from galaxy surveys .
models with adiabatic fluctuations , as produced in the simplest inflationary models , produced higher values of @xmath24 than models with isocurvature perturbations or models with structure formation through topological defects .
the next step was to explore how the cmb angular power spectrum extended to smaller angles .
a host of ground and balloon based experiments followed on the cobe detection , and it took some time for the observational situation to become clear . for a number of years , a plot summarizing the observational situation showing all the experiments extending the cobe dmr experiment to smaller angular scales somewhat resembled a target where the marksman often altogether missed the target .
theorists could hand pick the points they liked , claiming that based on inside information certain points should be believed more than others .
( see for example the figure of max tegmark in ref . . )
the observations improved gradually .
it is usually the boomerang @xcite and maxima @xcite experiments that are credited with having provided the first clear and convincing evidence of the acoustic oscillations .
the next major step was the wmap space mission , often described as the second - generation " cmb space mission , following on cobe .
the nasa wmap satellite was launched in 2001 and delivered its first results in 2003 concerning the cmb temperature anisotropy over the whole sky .
wmap s first cmb polarization results were reported in 2006 , and wmap continued taking data until 2010 .
periodic updates of the wmap results were subsequently published taking into account more data and refining the analysis .
like cobe , wmap surveyed the entire sky , but because wmap used a pair of telescopes rather than horns pointing directly at the sky , wmap had a superior angular resolution .
wmap observed in five frequencies : 23 , 33 , 41 , 61 , and 94 ghz .
the lowest frequency bands were the most contaminated by galactic synchrotron radiation , and the highest ( or highest two ) frequency band was the cleanest .
because of the hemt ( high electron mobility transistor ) ( i.e. , coherent amplification ) detection technology employed , wmap could not reach to higher frequencies were galactic thermal dust contamination starts to become dominant .
i vividly recall the first press conference when the wmap collaboration announced their first scientific results , and given that this was the first post - cobe cmb space mission , there were great expectations that some bombshell was going to be dropped . boomerang@xcite and maxima@xcite
had presented fairly convincing evidence for the first doppler peak at precisely the position predicted by @xmath27cdm combined with simple inflationary models .
even though these observations killed off a lot of models or at least made most of these models very difficult to defend , the error bars were still quite large , and there remained the possibility that wmap with its high precision and full sky coverage would conclude `` none of the above '' for the many still contending theoretical models .
as a theorist , this was certainly my hope , because simple models of single scalar field inflation are rather boring .
at the press conference , john bahcall summed up that the greatest surprise [ from wmap ] was that there was no surprise . "
indeed wmap found that their measured temperature power spectrum was consistent with the simplest six - parameter model .
the six parameters were @xmath28 @xmath29 @xmath30 @xmath31 @xmath32 and @xmath33 or equivalently , one could use any other six independent parameters derived from these parameters .
one of the interesting novelties from among the wmap results was the measurement of the spectral index of the power spectrum , which prior to wmap was consistent with @xmath34 which would be the prediction if there were an exact unbroken scale symmetry . while inflationary potentials can be constructed or some might say ` designed'through fine tuning to give @xmath35 arbitrarily close to one , generically inflation predicts some deviation from exact scale invariance , most likely but not inexorably on the side @xmath36 the three - year wmap results in the framework of the previously described six - parameter model measured @xmath37 thus excluding a scale - invariant primordial power spectrum at approximately @xmath38 for many this result strengthened the case for inflation , as opposed to some unspecified more symmetric theory for imprinting the primordial cosmological perturbations .
wmap also was able to place interesting constraints on primordial non - gaussianity and discovered a few ` anomalies ' at modest statistical significance , a topic to which we shall return below in our discussion of planck .
until the planck 2013 cosmology results were announced , the wmap characterization of the temperature power spectrum on intermediate scales remained unrivaled , although ground - based experiments , most notably act@xcite and spt,@xcite succeeded in extending the measurement of @xmath39 to smaller scales extending to around @xmath40 in these proceedings i shall not dwell on the esa planck space mission because planck has over the last few years released two sets of cosmology results , one in 2013 and another in 2015 .
these results have been the subject of many plenary conference presentations .
it is therefore doubtful that i can in the small number of pages available tell you something that you have not already heard or read elsewhere .
for this reason i invite you the consult the relevant 2013 and 2015 papers for a complete account .
let me however give you my own take on the main scientific impact of the planck .
planck had the capability to improve on wmap in several ways owing to its ten times better sensitivity and two times better angular resolution.@xcite the dimensions of the planck and wmap telescopes were roughly comparable , with a usable diameter of slightly over a meter , the main difference between the two missions being the types of detectors deployed .
wmap deployed coherent detectors passively cooled to about 95 k and less sensitive , while planck used , at least for the six highest hfi ( high frequency instrument ) frequency channels , more sensitive bolometers that were cryogenically cooled to @xmath41 mk . on a space mission , bolometers are more risky because failure of the cooling system to perform adequately would prevent any useful data from being taken .
bolometers however present two significant advantages : ( 1 ) bolometers function nearly at the limit of the inherent quantum noise of the incident photon flux , whereas the sensitivity of coherent detectors is worse than this fundamental limit by a substantial factor , and ( 2 ) coherent detectors for measuring cmb anisotropies are able to operate only up to about 100 ghz , whereas bolometers function successfully at much higher frequencies extending into the wien tail of the cmb blackbody frequency spectrum .
wmap mapped the microwave sky in five bands centered at 23 , 33 , 41 , 61 , and 94 ghz , whereas planck mapped the sky in nine bands centered at 30 , 44 , 70 , 100 , 143 , 217 , 353 , 545 , and 857 ghz .
the extended reach of planck thanks to these higher frequency bands allowed measurements where the dust contributes substantially to the sky signal and even dominates over the cmb in the highest bands , thus allowing better cleaning of nonprimordial contaminants and substantially reducing uncertainties in the possible role of dust contamination in the final cleaned cmb maps . the highest planck frequency ( at 857 ghz )
is situated far into the wien tail of the cmb , so that except very near the galactic poles , one sees hardly any cmb .
almost all the signal comes from thermal dust emission from our galaxy as well as from infrared galaxies .
the presence of these higher frequency channels constitutes an additional advantage . since cmb measurements are almost always diffraction limited , given the roughly comparable dimensions of the telescopes , planck benefits from a better angular resolution
. power spectrum ( where @xmath42 ) compared to the best fit six - parameter theoretical model ( red ) , and the lower plot shows a zoom of the residuals .
the fit is quite good except for a dip around @xmath43 whose statistical significance is around @xmath44 _
( credit : esa/ planck collaboration ) _ , width=453 ] -30pt the temperature power spectrum reported in the planck 2013 cosmological results release was consistent with the same six - parameter model used for the wmap analysis .
figure [ planckttfig ] shows the temperature power spectrum from the 2015 cosmology release .
( for more details , see refs . and .
) one could say that the john bahcall s 2001 wmap summary that the `` greatest surprise was that there was no surprise '' could equally well be applied to the planck results , despite the considerable scope to detect anomalies as explained above .
the planck @xmath39 power spectrum shows several improvements of a qualitative nature relative to the wmap @xmath39 power spectrum . while only three acoustic oscillations are clearly visible in the wmap power spectrum before the error bars start to blow up at higher @xmath45 the planck power spectrum shows clearly five acoustic peaks , and moreover the decay of the damping tail is precisely mapped out to about @xmath46 where things start to fall apart . at this point beam
smearing combined with the limited sensitivity causes the error bars to almost literally blow up toward higher @xmath45 so that beyond @xmath47 there is little useful data .
the other result from 2013 that was perhaps surprising to a lot of theorists was the stringent upper limits established by planck on the possible primordial non - gaussianity . in the wmap data
, there were some hints at very modest statistical significance of an @xmath48 yadav and wandelt@xcite reported that according to their independent analysis of the wmap data , @xmath49 $ ] at 99.5% confidence level with a central value of @xmath50 other analyses found somewhat lower values , but despite disagreement on the details , most analyses found a hint of a signal with @xmath51 at low statistical significance . in the wake of these developments , theorists worked feverishly to devise models predicting a large primordial bispectral non - gaussianity .
many of these theorists anticipated a planck confirmation of the hints found by wmap , but in its 2013 analysis based on the temperature maps , the planck collaboration reported @xmath52 thus laying to rest the hint of a detection that some had seen in the wmap data .
the 2015 planck results , which additionally included polarization , gave consistent constraints .
( left ) and @xmath53 ( right ) power spectra reported in the planck 2015 release show that the observed polarized anisotropies are broadly consistent with the six - parameter concordance model .
the error bars shown should be understood as unreliable and as a lower bound on the actual errors , because as the planck 2015 papers reported , there remains some evidence of unaccounted or uncorrected systematics in the polarization maps , most notably @xmath54 leakage . _
( credit : esa/ planck collaboration ) _ , title="fig:",width=226 ] ( left ) and @xmath53 ( right ) power spectra reported in the planck 2015 release show that the observed polarized anisotropies are broadly consistent with the six - parameter concordance model .
the error bars shown should be understood as unreliable and as a lower bound on the actual errors , because as the planck 2015 papers reported , there remains some evidence of unaccounted or uncorrected systematics in the polarization maps , most notably @xmath54 leakage . _
( credit : esa/ planck collaboration ) _ , title="fig:",width=226 ] -20pt the planck 2015 results also included the cmb @xmath55 auto - correlation power spectrum as well as the @xmath53 cross - correlation power spectrum .
( see fig .
[ planckpolarizationfigure ] . ) the polarization of the cmb had first been discovered by dasi in 2002@xcite and was later mapped over the whole sky by wmap at lower resolution and sensitivity than in the 2015 planck results .
the first wmap polarization results were released in 2005 based on the wmap 3-year data , and updated using the wmap 5-year , 7-year , and 9-year data were also released . when analyzed independently without using the temperature data , the planck polarization power spectrum gives cosmological parameters consistent with those obtained using the tt data alone and with those using all the data .
this fact indicates that the cmb polarization tells a consistent story , corroborating the story told by the temperature anisotropies alone .
the signal - to - noise ratio of the planck @xmath56 s hovers near or below unity until about @xmath57 beyond where the error bars start to blow up .
the planck team also studied the various anomalies found by the wmap team at modest statistical significance .
these include bipolar disorder ( or dipole modulation of the power spectrum ) , alignments of certain low @xmath58 multipole moments , and the ` cold spot , ' all of which are results at low statistical significance .
this means that one is free either to reject them as statistical flukes , or to take them as interesting hints of new physics , following one s own personal tastes .
some believed that planck with its higher frequency channels might make these anomalies go away by showing that they resulted from underestimating or improperly correcting for galactic dust contamination .
this however did not happen .
the anomalies seen by wmap were broadly confirmed by planck at comparable statistical significance .
this result is not at all surprising because the limiting factor for both experiments is cosmic variance .
unfortunately , future measurements of the cmb temperature can not hope to shed additional light on this question .
they too will be seeing that same sky and will be limited by exactly the same cosmic variance .
there is however room for modest improvement resulting from using polarization , perhaps from the late-2016 planck low @xmath58 polarization results when these become available .
in early 2015 the planck collaboration released its first cosmological results including the polarization data .
the measurements of the ee polarization power spectrum at @xmath59 is consistent with the 2013 results using tt alone , and perhaps more importantly , the cosmological parameters found using ee alone tell a story consistent with the tt only results , showing that the measured polarization power spectrum corroborates the story told by tt . in the 2015 cosmology release
, however , it was noted that the polarization data contained some evidence of residual systematic errors and that the polarization error bars given are not fully trustworthy .
moreover , the low @xmath58 polarization power spectrum ( for @xmath60 ) had not been released because the analysis was still ongoing . in late 2016
the planck collaboration intends to release a new set of results that will , most importantly , provide planck measurements of the polarization extending all the way down to @xmath61 this data will help pin down the reionization optical depth as well as giving planck constraints on @xmath62 from the reionization bump .
it is however doubtful that these constraints will be better than the most recent constraints established by bicep2/keck array team .
we now turn to discussing the b modes of the cmb polarization , which have received a lot of attention , especially since the march 2015 announcement by the bicep2 team of a purported discovery of b modes using measurements in just one frequency channel at 150 ghz.@xcite based on cmb polarization measurements in a patch covering only about 2% of the sky chosen for its low level of galactic foregrounds as measured using temperature maps , the bicep2 team found a b mode amplitude that if entirely primordial would correspond to @xmath63 with @xmath64 disfavored at @xmath65 according to the interpretation put forth by the bicep2 team , an explanation based on galactic foregrounds could be excluded because of the extremely large magnitude of the observed signal . this claim made the front pages of newspapers around the world and sparked great enthusiasm in the popular science press .
bicep2 claimed to have investigated the predictions of six dust models and to have found that none of these models could account for dust contamination having such a large magnitude .
there was only a slight problem : there was no publicly available data to fix the parameters of these models , a difficulty not sufficiently appreciated by the four bicep2 pis .
the only data that could settle this question was in the planck high frequency polarized dust maps ( particularly the polarized maps at @xmath66 and @xmath67 ghz whose signal is almost exclusively thermal dust emission from our galaxy ) , but at that time the planck collaboration had not publicly released these maps , nor had they yet presented their analysis . in the early summer a paper by flauger , spergel , and hill@xcite appeared arguing that all the data available at that time concerning polarized dust emission implied that all the signal could be attributed to galactic dust contamination , although a primordial signal could not be ruled out . in september 2015
the planck collaboration finally published the results of their polarized dust analysis showing that in fact the polarized dust magnitude measured by planck ( primarily using the 353 ghz channel ) in the bicep2 field was sufficiently high to account for all the signal.@xcite subsequently planck and bicep2 agreed to carry out a joint cross - correlation study in order to determine the actual level of dust contamination .
the result of this analysis was that the detection , initially claimed at @xmath68 went away , and a new upper bound on @xmath69 with @xmath70 at 95 % confidence was established.@xcite subsequent work by the bicep2/keck array team incorporating more recent data improved this bound to @xmath71 at a 95 % confidence level.@xcite there are at present many ground and balloon based experiments underway , for example act , bicep2/keck array , class , ebex , piper , polar bear / simons array , quijote , qubic , spider , and spt , aiming to improve the upper bounds on @xmath62 or to make a first detection .
we discuss some of the most ambitious of the initiatives in the next two sections .
with some risk of oversimplification , the status of the cmb today could be summed up as follows . the planck space mission
together with ground based experiments such as act and spt extending to smaller scales have mapped the cmb temperature anisotropies to reach almost the cosmic variance limit , and at present the frontier of cmb research is the search for primordial b modes that were presumably generated during cosmic inflation , as well as an ultra - precise characterization of the absolute frequency spectrum by improving on cobe firas . before going on to discuss the search for b modes , let us nuance this claim somewhat . at low multipole number @xmath45 the planck determination of the temperature power spectrum is limited by cosmic variance that is , the fact that we can observe only one sky .
if we accept the hypothesis that the primordial scalar perturbations were imprinted by an isotropic gaussian stochastic process , all useful information in the cmb temperature anisotropies can be compressed into the measured power spectrum @xmath72 and even with the most perfect measurement , there always remains a residual rms fractional uncertainty error of @xmath73 in the measurement of each @xmath74 for the underlying isotropic statistical process that generated our random sky map .
even with the best possible measurements , there is a maximum @xmath58 beyond which very little precise information can be obtained regarding the primordial perturbations given that beyond @xmath75 other effects rapidly kick in and dominate the sky anisotropy signal .
only measurements of the primordial signal averaged over broad bands and with large fractional error bars can be obtained beyond this approximate threshold .
this transition from a regime at low @xmath58 where the primordial cmb signal dominates to a regime at high @xmath58 where the primordial signal becomes a sideshow is quite abrupt for a number of reasons .
most of the foregrounds that dominate on small angular scales have the power spectrum of white noise or of point sources that is , @xmath76 the primordial cmb temperature power spectrum , by contrast , at low @xmath58 scales very approximately as @xmath77 ( if one ignores the acoustic oscillations ) but at higher @xmath58 falls exponentially due to silk damping or ` viscosity . '
point sources can be masked , but there are unsurmountable limitations to how much contamination can be mitigated by fancy ` component separation ' techniques because each point source has its own unique frequency spectrum .
the end result , as estimated by fisher forecasts for future cmb space missions , is that the error bars on the cosmological parameters may shrink each by a factor of approximately 23 .
it has been argued that one can do better with measuring the ee polarization power spectrum because despite the fact that the cmb polarization is harder to measure experimentally on account of its smaller absolute magnitude , the non - primordial contaminants have a smaller fractional polarization than the primordial cmb , and thus for polarization the wall at around @xmath75 moves to somewhat higher @xmath78 the preceding comments should not be misinterpreted as implying that exploring the cmb beyond @xmath79 is not of great interest . as the papers from the act and spt collaborations have shown
, there is a lot of interesting science to be done by observing the microwave temperature and polarization anisotropy on very small angular scales , but the primary science goals are not the primordial scalar mode power spectrum but rather the science of what are often termed as the `` secondary anisotropies . ''
gravitational lensing of the cmb is a very interesting ( and clean ) probe of the inhomogeneities of the distribution of matter lying between ourselves and the last scattering surface at @xmath80 in a certain sense , the idea behind these observations is the same as for studies of weak lensing in the optical and infrared bands , where one observes the ellipticities of galaxies after smoothing .
the simplest model would be that except for the lensing shear , the ellipticities would be random and uncorrelated , so that after averaging over many galaxies one recovers the lensing field .
however intrinsic alignments add extra noise and biases to the reconstructed lensing field maps
. a few differences between cmb and other lensing surveys are worth highlighting : ( 1 ) the cmb constitutes the most distant exploitable source plane , whereas weak lensing of galaxies does not extend as far back in redshift .
this means that linear theory is more applicable to interpreting cmb lensing results than to the weak lensing of galaxies .
( 2 ) for constructing a lensing map based on the alignments of the ellipticities of galaxies , it is assumed that there are no intrinsic alignments , which would act to bias measurements .
cmb lensing is free of this complication and can be used by studying cross correlations to test for and to characterize such intrinsic alignments . moreover , combining with cmb lensing also provides more powerful ` tomography , ' in order to try to recover depth information .
another frontier of cmb research is the measurement of the absolute frequency spectrum of the cmb .
it is remarkable that the cobe firas experiment of 25 years ago still remains the best measurement of the absolute frequency spectrum , notwithstanding some improvement at low frequencies by the arcade balloon borne experiment .
by contrast , great progress has been made improving on the cobe dmr first detection of the primordial cmb anisotropy , and much of this improvement derives from ground and balloon based experiments and not just the two post - cobe cmb space missions wmap and planck . while differential measurements can be carried out through the atmosphere , absolute measurements require subtractions that are not feasible .
technology has greatly improved since firas , and an experiment called pixie@xcite has been proposed very roughly speaking to redo the firas experiment with polarization and with a sensitivity approximately two orders of magnitude better .
we also parenthetically note that pixie also proposes to search for b modes with the same instrument ( using a substantially different technique ) . for an interesting summary of the new science made possible through measuring the absolute spectrum , see ref . and references therein .
we have seen that observations of the cmb both of its anisotropies in temperature and polarization and of its absolute frequency spectrum have played a key role in putting modern cosmology on a firm observational footing , so that what is now known as the standard model of cosmology " or the early universe is not just theoretical speculation .
we should be careful not to exaggerate the successes of this model , because there remain a lot of open questions , and in certain respects ( e.g. , the functional form of the primordial power spectrum ) , the model provides what may be fairly described more as a fitting formula than a complete theory .
for example , the standard model as proposed by weinberg and salam , even though theorists may not like it so much for various well justified reasons , has a well defined number of parameters .
there are no free functional forms . even including massive neutrinos , which are now known to exist and often described as ` physics beyond the standard model , ' one must admit that the few extra terms and parameters from the theoretical point of view hardly constitute a revolutionary update of the theoretical physics of the late 1960s and 1970s .
indeed there is room for a compelling theoretical model realizing inflation that would provide a prediction for the functional form of the inflationary potential , or for an alternative to inflation .
but this is a subject on which we have seen many papers , but unfortunately there have been no qualitative breakthroughs . to my mind
, the most interesting future challenge of cmb research is the search for primordial b modes . as explained above
, the scalar power spectrum has already been characterized close to the ultimate limit imposed by cosmic variance . on the other hand , regarding b modes , all that we have for the moment is ever improving upper bounds .
the last five years have seen a remarkable improvement in the quality of these bounds , to the point now that the best bounds on @xmath62 derive from b modes and no longer from the shape of the @xmath81 power spectrum at @xmath82 the bounds derived from the shape of the temperature spectrum are highly model dependent , whereas the bounds from b modes are extremely robust .
there remains substantial scope for improving on measurements of the b modes .
the most ambitious projects currently in the planning stage aim to be able to make a detection of @xmath83 at high statistical significance , and if one is very ( perhaps unrealistically ) optimistic about the unknowns ( e.g. , complexity of foregrounds , ability to overcome systematic uncertainties ) , one might hope to be able to push beyond the goal . at this point ,
apart from the experiments already underway , several ambitious initiatives are being planned . in the suborbital class
, there is the ground - based american s4 ( stage 4 " ) program , whose details remain to be defined .
s4 aims to deploy of order @xmath84 detectors from the ground in order to detect primordial b modes and also map the lensing potential from the ground . what can actually be achieved from the ground
is subject to considerable uncertainty on account of atmospheric interference and the paucity of bands through which the atmospheric contamination is not overwhelming . from space three satellite missions
are presently under consideration : the japanese - led litebird mission ( described in more detail in the contribution by hirokazu ishino in these proceedings ) , an esa m5 proposal whose details are not yet defined , and pixie . the esa m5 proposal could be similar to the previous c0re and core+ proposals , or it could take the form of a joint esa - jaxa - nasa mission building on the success of the litebird jaxa / nasa joint phase a study now underway .
another possibility is a new version of the pixie proposal . in any case , it is clear that there still remains much exciting new science to be done in the area of cmb observation .
it is difficult to predict exactly what path this field will take .
mb would like to thank the organizers of the cospa 2015 12th international symposium on cosmology and particle astrophyscis held from 12 - 16 october in daejeon , korea for their hospitality and for an impeccably well organized and interesting conference .
spergel , l. verde , h.v .
peiris et al . ,
first year wilkinson microwave anisotropy probe ( wmap ) observations : determination of cosmological parameters , " ap . j. suppl . 148
( 2003 ) 175 ( arxiv : astro - ph/0302209 ) planck collaboration ( r. adam et al . ) , planck intermediate results .
the angular power spectrum of polarized dust emission at intermediate and high galactic latitudes , " astron .
astrophys . 586
( 2016 ) a133 ( arxiv:1409.5738 ) the bicep2/keck and planck collaborations , a joint analysis of bicep2/keck array and planck data , " phys .
( 2015 ) 101301 ( arxiv:1502.00612 ) keck array , bicep2 collaborations ( p.a.r .
ade et al . ) , bicep2/keck array vi : improved constraints on cosmology and foregrounds when adding 95 ghz data from keck array , " phys .
( 2016 ) 031302 ( arxiv:1510.09217 ) s. das , t. louis , m.r .
nolta et al . , the atacama cosmology telescope : temperature and gravitational lensing power spectrum measurements from three seasons of data , " jcap 04 ( 2014 ) 014 ( arxiv:1301.1037 ) | the 2.725 k cosmic microwave background has played a key role in the development of modern cosmology by providing a solid observational foundation for constraining possible theories of what happened at very large redshifts and theoretical speculation reaching back almost to the would - be big bang initial singularity . after recounting
some of the lesser known history of this area , i summarize the current observational situation and also discuss some exciting challenges that lie ahead : the search for b modes , the precision mapping of the cmb gravitational lensing potential , and the ultra - precise characterization of the cmb frequency spectrum , which would allow the exploitation of spectral distortions to probe new physics . |
attempts to categorify the cluster algebras of fomin and zelevinsky @xcite have led to the investigation of categories with the 2-calabi - yau property ( 2-cy for short ) and their cluster - tilting objects .
main early classes of examples were the cluster categories associated with finite dimensional path algebras @xcite and the preprojective algebras of dynkin type @xcite .
this paper is centered around the more general class of stably 2-cy and triangulated 2-cy categories associated with elements in coxeter groups @xcite ( the adaptable case was done independently in @xcite ) , and their relationship to the generalized cluster categories from @xcite ( see section 4 for the definition ) .
let @xmath0 be a finite connected quiver with vertices @xmath9 , and @xmath1 the complete preprojective algebra of the quiver @xmath0 over a field @xmath2 .
denote by @xmath10 the distinguished generators in the corresponding coxeter group @xmath11 . to an element @xmath3 in @xmath11 ,
there is associated a stably 2-cy category @xmath12 and a triangulated 2-cy category @xmath13 .
the definitions are based on first associating an ideal @xmath14 in @xmath1 to each @xmath15 , hence to any reduced word by taking products .
this way we also get a finite dimensional algebra @xmath16 .
objects of the category @xmath12 are submodules of finite dimensional free @xmath5-modules .
the cluster category is then equivalent to @xmath13 with @xmath17 , where @xmath7 is a coxeter element such that @xmath18 is a reduced expression @xcite .
when @xmath1 is a preprojective algebra of dynkin type , then the category @xmath19 as investigated in @xcite is also obtained as @xmath12 where @xmath3 is the longest element ( * ? ? ?
* iii 3.5 ) . using the construction of ideals we get for each reduced expression
@xmath20 a chain of ideals @xmath21 which gives rise to an interesting set of @xmath1-modules : @xmath22 which all turn out to be indecomposable and to lie in @xmath12 .
the investigation of this set of modules , which we call _ layers _ , from different points of view , including connections with tilting theory , is one of the main themes of this paper , especially for a class of words called @xmath7-sortable .
the modules @xmath23 provide a natural filtration for the cluster - tilting object @xmath24 associated with the reduced expression @xmath25 ( see section 1 ) .
these modules can be used to show that the endomorphism algebras @xmath26 are quasi - hereditary @xcite . here
we show that these modules are rigid ( theorem [ 2spherical ] ) , that is @xmath27 and that their dimension vectors are real roots ( theorem [ realroot ] ) , so that there are unique associated indecomposable @xmath8-modules @xmath28 ( which are not necessarily rigid ) .
the situation is especially nice when all layers are indecomposable @xmath8-modules , so that @xmath29 .
this is the case for @xmath7-sortable words .
an element @xmath3 of @xmath11 is @xmath7-sortable when there exists a reduced expression of @xmath3 of the form @xmath30 with @xmath31 where @xmath7 is a coxeter element , that is , a word containing each generator @xmath15 exactly once , and in an order admissible with respect to the orientation of @xmath0 .
starting with the tilting @xmath8-module @xmath8 ( when @xmath32 ) , there is a natural way of performing exchanges of complements of almost complete tilting modules , determined by the given reduced expression .
we denote the final tilting module by @xmath33 , and the indecomposable @xmath8-modules used in the sequence of constructions by @xmath34 for @xmath35 .
we show that @xmath36 is a @xmath8-module in this case and that @xmath37 for all @xmath38 ( theorem [ lequalt ] ) and we also show that the indecomposable modules in the torsionfree class @xmath39 are exactly the @xmath34 ( theorem [ twtilting ] ) . in particular this gives a one - one correspondence between @xmath7-sortable words and torsionfree classes , as first shown in @xcite using different methods ( see also @xcite ) .
there is another sequence @xmath40 of indecomposable @xmath8-modules , defined using restricted reflection functors , which coincide with the above sequences .
this is both interesting in itself , and provides a method for proving @xmath37 for @xmath35 . in another paper @xcite ,
we give a description of the layers from a functorial point of view .
when the @xmath7-sortable word is @xmath41 , and @xmath42 , then the successive layers are given by @xmath43 for the indecomposable projective @xmath8-modules @xmath44 , where @xmath45 denotes the ar - translation .
in the general case we will give a description of the layers using specific factor modules of the above modules .
the generalized cluster categories @xmath46 for algebras @xmath47 of global dimension at most two were introduced in @xcite .
it was shown that for a special class of words @xmath3 , properly contained in the dual of the @xmath7-sortable words , the 2-cy category @xmath13 is triangle equivalent to some @xmath46 .
we point out that the procedure for choosing @xmath47 works more generally for any dual of a @xmath7-sortable word ( theorem [ equivalence ] ) .
the paper is organized as follows .
we start with some background material on 2-cy categories associated with reduced words , on complements of almost complete tilting modules and on reflection functors . in section 2
we show that for any reduced word @xmath3 , the associated layers are indecomposable rigid modules , which also are positive real roots . hence there are unique associated indecomposable @xmath8-modules . in section 3
we show that our three series of indecomposable modules @xmath48 , @xmath49 and @xmath50 coincide in the @xmath7-sortable case .
section 4 is devoted to examples and questions beyond the @xmath7-sortable case .
some of this work was presented at the conferences ` homological and geometric methods in representation theory ' in trondheim in august 2009 , ` interplay between representation theory and geometry ' in beijing in may 2010 , and in seminars in bonn and torun in 2010 . throughout @xmath2
is an algebraically closed field .
the tensor product @xmath51 , when not specified , will be over the field @xmath2 . for a @xmath2-algebra @xmath47 ,
we denote by @xmath52 the category of finitely presented right @xmath47-modules , and by @xmath53 the category of finite length right @xmath47-modules . for a quiver @xmath0
we denote by @xmath54 the set of vertices and by @xmath55 the set of arrows , and for @xmath56 we denote by @xmath57 its source and by @xmath58 its target .
this work was done when the first author was a postdoc at ntnu , trondheim .
she would like to thank the research council of norway for financial support .
part of this work was done while the second author visited ntnu during march and august 2009 .
he would like to thank the people in trondheim for their hospitality .
the authors would also like to thank the referee for reading the paper thoroughly , and for several useful suggestions .
let @xmath0 be a finite connected quiver without oriented cycles and with vertices @xmath59 . for @xmath60
we denote by @xmath61 the positive integer @xmath62 the _ coxeter group _ associated to @xmath0 is defined by the generators @xmath63 and relations * @xmath64 , * @xmath65 if @xmath66 , * @xmath67 if @xmath68
. in this paper @xmath6 will denote a word ( _ i.e. _ an expression in the free monoid generated by @xmath69 ) , and @xmath3 will be its equivalence class in the coxeter group @xmath11 .
an expression @xmath25 is _ reduced _ if @xmath70 is smallest possible .
an element @xmath71 is called a _ coxeter element _ if @xmath72 and @xmath73 .
we say that a coxeter element @xmath74 is _ admissible _ with respect to the orientation of @xmath0 if @xmath75 when there is an arrow @xmath76 .
the preprojective algebra associated to @xmath0 is the algebra @xmath77 where @xmath78 is the double quiver of @xmath0 , which is obtained from @xmath0 by adding for each arrow @xmath79 in @xmath55 an arrow @xmath80 pointing in the opposite direction .
we denote by @xmath1 the completion of the preprojective algebra associated to @xmath0 and by @xmath81 the category of right @xmath1-modules of finite length .
the algebra @xmath1 is finite - dimensional selfinjective if @xmath0 is a dynkin quiver .
then the stable category @xmath82 satisfies the _ 2-calabi - yau property _ ( 2-cy for short ) , that is , there is a functorial isomorphism @xmath83),\ ] ] where @xmath84 and @xmath85:=\omega^{-1}$ ] is the suspension functor @xcite ( see @xcite for a complete proof ) . when @xmath0 is not dynkin , then @xmath1 is infinite dimensional and of global dimension @xmath86 . in this case
the triangulated category @xmath87 is @xmath86-cy @xcite ( see @xcite for a complete proof ) .
we now recall some work from @xcite . for each @xmath88
we have an ideal @xmath89 in @xmath1 , where @xmath90 is the idempotent of @xmath1 associated with the vertex @xmath91 .
we write @xmath92 when @xmath93 is a expression of @xmath94 .
we denote by @xmath95 the simple bimodule corresponding to the vertex @xmath91 .
we collect the following information which is useful for section 2 : @xcite[basicprop ] let @xmath1 be a complete preprojective algebra . * if @xmath25 and @xmath96 are two reduced expressions of the same element in the coxeter group , then @xmath97 . *
if @xmath98 with @xmath99 reduced , then @xmath100 .
moreover @xmath6 is reduced if and only if @xmath101 . and for @xmath102 we have @xmath103 .
if @xmath1 is not of dynkin type we have moreover : * any finite product @xmath104 of the ideals @xmath105 is a tilting module of projective dimension at most one , and @xmath106 . *
if @xmath107 is a simple @xmath1-module and @xmath104 is a tilting module of projective dimension at most one , then @xmath108 or @xmath109 .
* if @xmath110 , then @xmath111 for a tilting module @xmath104 of projective dimension at most one . by @xmath112 the ideal @xmath113 does not depend on the choice of the reduced expression @xmath6 of @xmath3 .
therefore we write @xmath114 for the ideal @xmath113 and @xmath16 when @xmath6 is an expression of @xmath3 .
this is a finite dimensional algebra .
we denote by @xmath12 the category of submodules of finite dimensional free @xmath5-modules .
this is a _ frobenius category _ , that is , an exact category with enough projectives and injectives , and the projectives and injectives coincide .
its stable category @xmath13 is a triangulated category which satisfies the 2-calabi - yau property @xcite .
the category @xmath12 is then said to be _
stably 2-calabi - yau_. recall that a _ cluster - tilting object _ in a frobenius stably 2-cy category @xmath115 with finite dimensional morphisms spaces is an object @xmath116 such that * @xmath117 * @xmath118 implies that @xmath119 . for any reduced word @xmath25 ,
we write @xmath120 ( * ? ? ?
* thm iii.2.8)[birsc ] for any reduced expression @xmath25 of @xmath94 , the object @xmath121 is a cluster - tilting object in the stably 2-cy category @xmath12 . for any reduced word @xmath25
, we have the chain of ideals @xmath21 which is strict by proposition [ basicprop ] ( b ) . for @xmath122
we define the _ layer _ @xmath123 using proposition [ basicprop ] ( b ) it is immediate to see the following we have isomorphisms in @xmath81 : @xmath124 & m^i_{\mathbf{w}}}),\ ] ] where @xmath91 is the greatest integer satisfying @xmath125 and @xmath75 .
( if such @xmath91 does not exist , then we define @xmath126 to be @xmath127 . )
therefore the layers @xmath128 give a filtration of the cluster - tilting object @xmath24 .
let @xmath0 be a finite connected quiver with vertices @xmath129 and without oriented cycles .
a basic @xmath8-module @xmath130 is called a _ tilting module _
if @xmath131 and it has @xmath132 non - isomorphic indecomposable summands . for each indecomposable summand @xmath133 of @xmath130
, it is known that there is at most one indecomposable @xmath134 such that @xmath135 is a tilting module @xcite , and that there is exactly one if and only if @xmath136 is a sincere @xmath8-module @xcite .
we then say that @xmath133 ( and possible @xmath137 ) is a _ complement _ for the almost complete tilting module @xmath136 .
the ( possible ) other complement of @xmath136 can be obtained using the following result : @xcite[tilting mutation ] * if the minimal left @xmath138-approximation @xmath139^f & b}$ ] is a monomorphism , then @xmath140 is a complement for @xmath136 . *
if the minimal right @xmath138-approximation @xmath141^g & t_i } $ ] is an epimorphism , then @xmath142 is a complement for @xmath136 .
there is a one - one correspondence between tilting modules @xmath130 and contravariantly finite torsionfree classes @xmath143 containing the projective modules .
let @xmath0 be finite quiver with vertices @xmath129 and without oriented cycles .
let @xmath144 be a source .
then the quiver @xmath145 is obtained by replacing all arrows starting at the vertex @xmath91 by arrows in the opposite direction .
write @xmath146 where @xmath147 is the indecomposable projective @xmath8-module associated with the vertex @xmath38 . then using results of @xcite and @xcite we have functors : @xmath148^{r_i } & & \ar@<1mm>[ll]^{r_i^-}{{\sf mod \hspace{.02 in } } } kq'}\ ] ] where @xmath149 , @xmath150 and @xmath151 which induce inverse equivalences ( recall that in this paper we work with right modules ) @xmath152\ar@<1mm>[rr]^{r_i } & & \ar@<1mm>[ll]^{r_i^-}({{\sf mod \hspace{.02 in } } } kq')/[e_i dkq']},\ ] ] where @xmath153 $ ] ( resp .
$ ] ) is obtained from the module category @xmath155 ( resp .
@xmath156 ) by annihilating morphisms factoring through @xmath157 ( resp .
@xmath158 ) . since @xmath91 is a source ( resp .
a sink ) of @xmath0 ( resp .
@xmath159 ) we can regard the category @xmath153 $ ] ( resp .
$ ] ) as a full subcategory of @xmath155 ( resp .
@xmath156 ) .
when the vertex @xmath91 is not a sink or source , a reflection is still defined on the level of the grothendieck group @xmath161 .
the grothendieck group is constructed as the group with generators @xmath162 $ ] for @xmath163 and relations @xmath162+[z]=[y]$ ] if there is a short exact sequence @xmath164 & y\ar@{->>}[r ] & z}.$ ] this is a free abelian group with basis @xmath165,\ldots[s_n]\}$ ] , where @xmath166 are the simple @xmath8-modules . with respect to this basis
we define @xmath167)= [ s_j]+(m_{ij}-2\delta_{ij})[s_i],\ ] ] where @xmath61 is the number of edges of the underlying graph of @xmath0 as before .
this definition is coherent with the previous one .
indeed if @xmath91 is a source and @xmath168 is an indecomposable @xmath8-module which is not isomorphic to @xmath44 , then we have @xmath169)=[r_i(m)].\ ] ]
throughout this section let @xmath3 be an element in the coxeter group of an acyclic quiver @xmath0 , and fix @xmath25 a reduced expression of @xmath3 .
for @xmath170 we have defined in section 1 the layer @xmath36 as the quotient @xmath123 in this section , we investigate some main properties of these layers .
we show that each layer can be seen as the image of a simple @xmath1-module under an autoequivalence of @xmath87 . hence they are rigid indecomposable @xmath1-modules of finite length , and we compute explicitly their dimension vectors and show that they are real roots . hence to each layer
we can associate a unique indecomposable @xmath8-module with the same dimension vector @xcite , but which is not necessarily rigid .
note that some of the results of this section have been proven independently in @xcite but with different proofs .
the following easy observation is useful .
[ extend support ] let @xmath171 be an acyclic quiver and @xmath0 be a full subquiver of @xmath171 . for any reduced expression @xmath6 of @xmath94 ,
the module @xmath172 ( respectively , @xmath36 ) for @xmath171 is the same as @xmath172 ( respectively , @xmath36 ) for @xmath0 .
let @xmath173 and @xmath174 be the corresponding complete preprojective algebras .
let @xmath175 then we have @xmath176 and @xmath177 for any @xmath144 .
thus the assertions follow .
[ layerimagesimple ] let @xmath0 be an acyclic quiver and @xmath1 the complete preprojective algebra .
let @xmath25 be a reduced expression . 1
. for @xmath178 we have isomorphisms of @xmath1-modules : @xmath179 2 .
if @xmath0 is non - dynkin , then for @xmath178 we have isomorphisms in @xmath180 : @xmath181 we divide the proof into two cases , according to whether @xmath1 is of non - dynkin type or of dynkin type . _ non - dynkin case _ : we set @xmath182 and @xmath183 .
since @xmath184 is reduced , by proposition [ basicprop](e ) we have @xmath185 and hence we get the second isomorphism . since @xmath186 is reduced , we have @xmath187 , and therefore @xmath188 by proposition [ basicprop ] ( d ) .
thus we have @xmath189 _ dynkin case _ : we take a non - dynkin quiver @xmath171 containing @xmath0 as a full subquiver .
let @xmath190 be the complete preprojective algebra of @xmath171 and @xmath191 for @xmath192 . using the non - dynkin case and lemma [ extend support ]
, we have @xmath193 for the idempotent @xmath194 , the twosided ideal @xmath195 annihilates @xmath196 .
since @xmath197 holds , we have @xmath198 and @xmath199 thus the assertion follows .
immediately we have the following result , which implies that @xmath36 is an indecomposable rigid @xmath1-module of finite length . [
2spherical ] for @xmath178 we have * if @xmath1 is of non - dynkin type : @xmath200 * if @xmath1 is of dynkin type : @xmath201 note that one can write down explicitly the dimension for the other @xmath91 in the dynkin case by using @xmath202 @xcite . in the non - dynkin case
, @xmath36 is then _
2-spherical _ in the sense of seidel - thomas @xcite .
we divide the proof into two cases , according to whether @xmath1 is of non - dynkin type or of dynkin type .
_ non - dynkin case _ : by proposition [ basicprop ] ( c ) , @xmath203 is a tilting @xmath1-module with @xmath204 . hence the functor @xmath205 is an autoequivalence of @xmath180 .
we have @xmath206 and hence @xmath207 since @xmath87 is 2-cy .
moreover since @xmath0 has no loops , @xmath208 vanishes and since @xmath1 is known to have global dimension @xmath86 , @xmath209 vanishes for @xmath210 .
hence @xmath211 is 2-spherical .
since by proposition [ layerimagesimple ] the layer @xmath36 is the image of the simple @xmath211 by an autoequivalence of @xmath87 , it follows that @xmath36 is also 2-spherical .
_ dynkin case _ : we take a non - dynkin quiver @xmath171 containing @xmath0 as a full subquiver .
let @xmath190 be the complete preprojective algebra of @xmath171 .
then @xmath19 can be seen as a full and extension closed subcategory of @xmath212 . using the non - dynkin case and lemma [ extend support ] , we get @xmath213 using the fact that @xmath19 is stably 2-cy we get @xmath214 . here
we state a property about two consecutive layers associated with the same vertex , which gives rise to special non - split short exact sequences in @xmath81 .
[ dimext1 ] let @xmath215 be integers such that @xmath216 and such that @xmath38 is the only integer satisfying @xmath217 and @xmath216 .
then we have @xmath218 in order to prove this proposition , we first need a lemma . for @xmath219 , we denote as before by @xmath220 the @xmath1-module @xmath221 . [ petitlemme ]
let @xmath217 be as in proposition [ dimext1 ] . 1 .
the map @xmath222 induced by the irreducible map @xmath223 is an epimorphism .
the image of the map @xmath224 induced by the irreducible map @xmath225 is @xmath226 .
\(1 ) since @xmath217 , then by lemma iii.1.14 of @xcite , we have isomorphisms @xmath227 where @xmath228 is the idempotent @xmath229 .
then the map @xmath230 is the epimorphism @xmath231 induced by the inclusion @xmath232 .
\(2 ) it is clear that the image is contained in the radical . by lemma iii.1.14 of @xcite
, we have isomorphisms @xmath233 the map @xmath234 is induced by the inclusion of ideals @xmath235 .
but since @xmath38 is the only integer satisfying @xmath217 and @xmath236 , we have @xmath237 and hence the map @xmath234 is an isomorphism . by the definition of the layers , we have the following short exact sequences @xmath238 & m_{\mathbf{w}}^j\ar@{->>}[r ] & m_{\mathbf{w}}^i } \quad\textrm{and } ( k)\quad \xymatrix { l^k_{\mathbf{w}}\ar@{>->}[r ] & m_{\mathbf{w}}^k\ar@{->>}[r ] & m_{\mathbf{w}}^j}\ ] ] let @xmath239 be the kernel of the composition map @xmath240 .
then we have a short exact sequence @xmath241 & m_{\mathbf{w}}^k\ar@{->>}[r ] & m_{\mathbf{w}}^i}\ ] ] which gives rise to the following long exact sequence in @xmath242 , where @xmath243 : @xmath244 & d{{\sf hom } } _
\lambda(k , m_{\mathbf{w}})\ar[r ] & d{{\sf hom } } _ { \lambda}(m_{\mathbf{w}}^k , m_{\mathbf{w}})\ar[r ] & d{{\sf hom } } _ { \lambda}(m_{\mathbf{w}}^i , m_{\mathbf{w}})\ar[r ] & 0}\ ] ] the space @xmath245 is zero by lemma iii.2.1 of @xcite , and the @xmath246-module @xmath247 is indecomposable injective .
therefore the module @xmath248 has simple socle , and hence @xmath239 is indecomposable .
moreover from the sequences @xmath249 , @xmath250 and @xmath251 , we deduce that we have a short exact sequence @xmath252 & k\ar@{->>}[r ] & l^j_{\mathbf{w}}}$ ] which is non - split since @xmath239 is indecomposable .
hence we get @xmath253 from @xmath249 we deduce the long exact sequence @xmath254 & { { \sf hom } } _
\lambda(m_{\mathbf{w}}^k , m_{\mathbf{w}}^j)\ar[r ] & { { \sf hom } } _ \lambda(m_{\mathbf{w}}^k , m_{\mathbf{w}}^i)\ar[r ] & { { \sf ext }
} ^1_\lambda(m_{\mathbf{w}}^k , l_{\mathbf{w}}^j)\ar[r ] & { { \sf ext } } ^1_\lambda(m_{\mathbf{w}}^k , m_{\mathbf{w}}^j)=0}.\ ] ] hence by lemma [ petitlemme ] ( 1 ) we get @xmath255 . from @xmath249
we also deduce the long exact sequence @xmath256 & { { \sf hom } } _
\lambda(m_{\mathbf{w}}^i , m_{\mathbf{w}}^j)\ar[r ] & { { \sf hom } } _
\lambda(m_{\mathbf{w}}^j , m_{\mathbf{w}}^j)\ar[r ] & { { \sf hom } } _ \lambda(l^j_{\mathbf{w}},m_{\mathbf{w}}^j)\ar[r ] & { { \sf ext } } ^1_\lambda(m_{\mathbf{w}}^i , m_{\mathbf{w}}^j)=0}.\ ] ] hence by lemma [ petitlemme ] ( 2 ) we get @xmath257 which is one dimensional since @xmath258 is indecomposable . finally using @xmath250
we get the long exact sequence @xmath254 & { { \sf ext } } ^1_\lambda(m_{\mathbf{w}}^k , l^j_{\mathbf{w}})\ar[r ] & { { \sf ext } } ^1_\lambda(l^k_{\mathbf{w}},l^j_{\mathbf{w}})\ar[r ] & { { \sf ext } } ^2_\lambda(m_{\mathbf{w}}^j , l^j_{\mathbf{w}})\ar[r ] & \cdots}\ ] ] by the 2-cy property and the previous remarks we have @xmath259 and therefore @xmath260 in this section we investigate the action of the functor @xmath261 at the level of the grothendieck group of @xmath87 when @xmath1 is not of dynkin type .
we show that this action has interesting connections with known actions of coxeter groups .
we denote by @xmath262 $ ] the induced automorphism of @xmath263 .
[ dimvectlemma ] let @xmath0 be a non - dynkin quiver . for all @xmath264 in @xmath54
we have @xmath265=[s_j ] + ( m_{ij } -2 \delta_{ij})[s_i]\ ] ] in @xmath263 , where @xmath61 is the number of arrows between @xmath91 and @xmath38 in @xmath0 .
since @xmath266 , we have @xmath267 as @xmath1-bimodules .
hence we have the following isomorphisms in @xmath268 : @xmath269 therefore we have @xmath270= ( \sum_{t}(-1)^t\dim{{\sf ext } } ^t_\lambda(s_j , s_i))[s_i]=(2\delta_{ij}-m_{ij})[s_i].\ ] ] from the triangle @xmath271\ar[r ] & i_i\ar[r ] & \lambda\ar[r ] & s_i}$ ] we get a triangle @xmath272\ar[r ] & s_j{\overset{\mathbf{l}}{{\otimes}}}_\lambda i_i\ar[r ] & s_j \ar[r ] & s_j{\overset{\mathbf{l}}{{\otimes}}}_\lambda s_i}.\ ] ] hence we have @xmath273=[s_j]-[s_j{\overset{\mathbf{l}}{{\otimes}}}_\lambda s_i]=[s_j]-(2\delta_{ij}-m_{ij})[s_i].$ ] from lemma [ dimvectlemma ] , we deduce the following results .
[ realroot ] let @xmath1 be the complete preprojective algebra of _ any _ type . 1
. for @xmath178 we have @xmath274=r_{u_1}\ldots r_{u_{j-1}}([s_{u_j}])$ ] , where the @xmath275 are the reflections defined in section 1 .
in particular all @xmath274 $ ] are positive real roots
@xmath276,\ldots,[l^l_{\mathbf{w}}]$ ] are pairwise different in @xmath263 .
3 . for @xmath122
, there exists a unique indecomposable @xmath8-module @xmath277 such that @xmath274=[(l^j_{\mathbf{w}})_q]$ ] .
\(1 ) as in the previous subsection we treat separately the dynkin and the non - dynkin case . the non - dynkin case is a direct consequence of lemma [ dimvectlemma ] and proposition [ layerimagesimple ] . for the dynkin case , we can follow the strategy of the proof of theorem [ 2spherical ] .
we introduce an extended dynkin quiver containing @xmath0 as subquiver . then applying reflection functors associated to the vertices of @xmath0 to modules
whose support do not contain the additional vertex is the same as applying the reflection functors of @xmath0 . thus using lemma
[ extend support ] , the equality coming from the non - dynkin quiver gives us the equality for @xmath0 .
hence the @xmath278 $ ] are real roots , which are clearly positive since @xmath279 is a module .
\(2 ) by ( * ? ? ?
4.4.4 ) , @xmath280 $ ] , @xmath281),\ldots , r_{u_1}\ldots r_{u_{l-1}}([s_{u_l}])$ ] are pairwise different .
thus the assertion follows from ( 1 ) .
\(3 ) from ( 1 ) we know that the dimension vector of the layer @xmath36 is a positive real root , and we get the result applying kac s theorem @xcite .
the layer @xmath36 is always rigid as a @xmath1-module , but the associated indecomposable @xmath8-module @xmath28 is not always rigid as shown in the following .
[ examplelqnonrigid ] let @xmath0 be the quiver @xmath282 & \\ 1\ar[ur]\ar[rr ] & & 3}$ ] , and @xmath283 .
then we have @xmath284 thus the associated indecomposable @xmath8-modules are the following : @xmath285 the module @xmath286 lies in the tube of rank 2 , with indecomposable objects @xmath287 and @xmath288 on the border of the tube . since @xmath286 is not on the border of the tube , it is not rigid .
@xcite let @xmath0 be an acyclic quiver with @xmath132 vertices , and @xmath11 be the coxeter group of @xmath0 .
let @xmath289 be the vector space with basis @xmath290 .
the _ geometric representation _
@xmath291 of @xmath11 is defined by reflections @xmath292 the _ contragradient of the geometric representation _
@xmath291 is then @xmath293 the grothendieck group @xmath263 has a basis consisting of the simple @xmath1-modules , and @xmath294 has a basis consisting of the indecomposable projective @xmath1-modules .
let @xmath1 be the complete preprojective algebra of non - dynkin type . 1 .
the coxeter group @xmath11 acts on @xmath263 by @xmath295 $ ] as the geometric representation .
2 . the coxeter group @xmath11 acts on @xmath296 by @xmath295 $ ] as the contragradient of the geometric representation .
\(1 ) this follows directly from lemma [ dimvectlemma ] .
\(2 ) this is shown in ( * ? ? ?
* theorem 6.6 ) .
it is assumed in @xcite that @xmath0 is extended dynkin , but this assumption is not used in the proof for this statement . in this subsection , we state some basic properties of the first layers . in particular
we show that the equivalence @xmath297 , when @xmath0 is not dynkin , can be interpreted as a reflection functor of the category @xmath87 .
[ lemmelayer1 ] let @xmath0 be an acyclic quiver , and @xmath298 .
let @xmath299 be a coxeter element admissible with respect to the orientation of @xmath0 .
let @xmath144 be a source of @xmath0 and @xmath300 be the reflection functor for @xmath145 .
then we have the following : 1 .
@xmath301 in @xmath19 .
so we view @xmath8-modules as @xmath1-modules annihilated by @xmath302 .
if @xmath99 is a subsequence of @xmath6 , and if @xmath6 is a subsequence of @xmath303 ( where @xmath99 , @xmath303 and @xmath6 are reduced expressions in @xmath11 ) , then @xmath304 is a @xmath8-module .
3 . @xmath305 is a @xmath306-module and isomorphic to @xmath307 as a @xmath306-module , where @xmath157 is the indecomposable projective @xmath8-module associated to @xmath91 and @xmath45 is the ar - translation of @xmath155 .
\(1 ) this is propositions ii.3.2 and ii.3.3 of @xcite .
\(2 ) since @xmath308 , we have that @xmath304 is annihilated by @xmath302 .
\(3 ) note that by proposition [ basicprop ] ( b ) we have @xmath309 and @xmath310 if @xmath102 .
therefore by ( 1 ) it is enough to prove that @xmath311 .
the projective resolution of @xmath312 in @xmath19 has the form : @xmath313 & \bigoplus_{a\in \bar{q}_1 , s(a)=i}e_{t(a)}\lambda\ar[r ] & e_i i_i\ar[r ] & 0.}\ ] ] applying the functor @xmath314 to the exact sequence @xmath315 , we get an exact sequence @xmath316 & \bigoplus_{a\in \bar{q}_1 , s(a)=i}e_{t(a)}\dfrac{\lambda}{i_c}\ar[r ] & e_i\dfrac{i_i}{i_ii_{c}}=e_i\dfrac{i_i}{i_{cs_i}}\ar[r ] & 0}.\ ] ] since @xmath91 is a source in @xmath0 , we have the set equality @xmath317 therefore by ( 1 ) the exact sequence @xmath318 is @xmath319 & \bigoplus_{a\in q_1 , s(a)=i}e_{t(a)}kq\ar[r ] & e_i\dfrac{i_i}{i_{cs_i}}\ar[r ] & 0}.\ ] ] hence we have @xmath320 . from lemma [ lemmelayer1 ]
we deduce the following result which gives another interpretation of the reflection functor .
[ commutativediagram ] let @xmath0 be an acyclic quiver and @xmath298 . 1 .
let @xmath144 be a sink of @xmath0 .
let @xmath145 .
then the following diagram commutes @xmath321 \ar[r]^{r_i^{- } } \ar@{^(->}[d ] & { { \sf mod \hspace{.02 in } } } kq'/ [ e_i kq']\ar@{^(->}[d]\\ { { \sf f.l .
\hspace{.02 in } } } \lambda\ar[r]_{-{\otimes}_{\lambda}i_i } & { { \sf f.l .
\hspace{.02 in } } } \lambda},\ ] ] where the vertical functors are the natural inclusions .
if @xmath0 is not dynkin , then the following diagram commutes @xmath321 \ar[r]^{r_i^{- } } \ar@{^(->}[d ] & { { \sf mod \hspace{.02 in } } } kq'/ [ e_i kq']\ar@{^(->}[d]\\ { \mathcal{d}}^b({{\sf f.l .
\hspace{.02 in } } } \lambda)\ar[r]_{-{\overset{\mathbf{l}}{{\otimes}}}_{\lambda}i_i } & { \mathcal{d}}^b({{\sf f.l .
\hspace{.02 in } } } \lambda)},\ ] ] where the vertical functors are the natural inclusions .
2 . let @xmath322 be a coxeter element and @xmath323 the coxeter functor .
then the following diagram commutes @xmath324 \ar[r]^{c^{- } } \ar@{^(->}[d ] & { { \sf mod \hspace{.02 in } } } kq/ [ kq]\ar@{^(->}[d]\\ { { \sf f.l .
\hspace{.02 in } } } \lambda\ar[r]_{-{\otimes}_{\lambda}i_c } & { { \sf f.l .
\hspace{.02 in } } } \lambda},\ ] ] where the vertical functors are the natural inclusions
. if @xmath0 is not dynkin , then the following diagram commutes @xmath324 \ar[r]^{c^{- } } \ar@{^(->}[d ] & { { \sf mod \hspace{.02 in } } } kq/ [ kq]\ar@{^(->}[d]\\ { \mathcal{d}}^b({{\sf f.l .
\hspace{.02 in } } } \lambda)\ar[r]_{-{\overset{\mathbf{l}}{{\otimes}}}_{\lambda}i_c } & { \mathcal{d}}^b({{\sf f.l .
\hspace{.02 in } } } \lambda),}\ ] ] where the vertical functors are the natural inclusions .
in particular we have @xmath325 \(1 ) denote by @xmath7 the coxeter element admissible with respect to the orientation of @xmath0 , and by @xmath326 the coxeter element admissible with respect to the orientation of @xmath159 .
we have the following isomorphisms in @xmath81 . @xmath327 thus on @xmath328 $ ] , we have @xmath329
. the latter assertion can be shown quite similarly since we have @xmath330 by proposition [ layerimagesimple ] .
\(2 ) this is a direct consequence of ( 1 ) .
in this section @xmath0 is a finite acyclic quiver , @xmath1 is the complete preprojective algebra associated with @xmath0 and @xmath7 a coxeter element admissible with respect to the orientation of @xmath0 .
the purpose of this section is to investigate the layers for words @xmath6 satisfying a certain property called @xmath7-sortable .
@xcite let @xmath7 be a coxeter element of the coxeter group @xmath11 .
usually we fix a reduced expression of @xmath7 and regard @xmath7 as a reduced word . an element @xmath3 of @xmath11
is called _ c - sortable _ if there exists a reduced expression @xmath6 of @xmath3 of the form @xmath30 where all @xmath331 are subwords of @xmath7 whose supports satisfy @xmath332 for @xmath144 , if @xmath15 is in the support of @xmath331 , by abuse of notation , we will write @xmath333 . then @xmath7-sortability does not depend on the choice of reduced expression of @xmath7 .
it is immediate that the expression @xmath30 is unique for any @xmath7-sortable element of @xmath11 @xcite .
let @xmath3 be an element of @xmath11 , and @xmath25 a reduced expression .
recall from section 1 that for @xmath334 the layer @xmath36 is defined to be the @xmath1-module : @xmath335 where @xmath336 satisfies @xmath337 and is maximal with this property .
[ calculation of l and m ] let @xmath0 be the quiver @xmath338 & \\1\ar[rr]\ar[ur ] & & 3}$ ] and @xmath339
. the standard cluster - tilting object @xmath24 in @xmath12 has the following indecomposable direct summands @xmath340 then we can easily compute the layers @xmath341 .
they are the indecomposable summands of the @xmath126 as @xmath8-modules : @xmath342 here is a nice characterization of @xmath7-sortable words .
let @xmath3 be an element of @xmath11 and @xmath343 be a reduced expression of @xmath3 .
then we have the following : 1 .
if there exists a coxeter element @xmath7 such that @xmath3 is @xmath7-sortable and @xmath6 is the @xmath7-sortable expression of @xmath3 , then @xmath36 is in @xmath155 for all @xmath122 , where @xmath0 is admissible for the coxeter element @xmath7 ; 2 .
if for all @xmath122 the layer @xmath36 is in @xmath155 for a certain orientation of @xmath0 , then @xmath3 is @xmath7-sortable , where @xmath7 is the coxeter element admissible for the orientation of @xmath0 .
\(1 ) assume that @xmath25 is a @xmath7-sortable word .
let @xmath344 , and @xmath2 be the ( possibly ) last index such that @xmath345 and @xmath336 . since @xmath6 is @xmath7-sortable , the word @xmath346 is a subsequence of @xmath347 .
therefore we have that @xmath348 is a @xmath8-module by lemma [ lemmelayer1](2 ) .
( 2 ) we prove this assertion by induction on the length of the word @xmath3 .
for @xmath349 the result is immediate . by lemma
[ extend support ] we can assume that the support of @xmath6 is @xmath54 .
assume that ( 2 ) is true for any word @xmath6 of length @xmath350 and let @xmath351 be a reduced expression such that @xmath36 is a @xmath8-module for all @xmath178 .
we first show that @xmath352 is a source of @xmath0 .
assume it is not , then there exists @xmath353 such that there is an arrow @xmath354 in @xmath0 .
take the smallest such number .
it is then not hard to check that the top of @xmath355 is the simple @xmath356 and that the kernel of the map @xmath357 contains @xmath358 in its top .
thus @xmath355 is not a @xmath8-module , which is a contradiction .
therefore @xmath352 is a source of @xmath0 .
we have @xmath359 for @xmath360 by proposition [ layerimagesimple ] , where @xmath361 . by theorem [ realroot ]
( 1 ) we have @xmath274=r_{u_1}\circ \ldots \circ r_{u_{j-1}}([s_{u_j}])$ ] in the grothendieck group @xmath263 . by theorem [ realroot ]
( 2 ) we then have @xmath274\neq [ s_{u_1}]$ ] for @xmath362 .
thus @xmath36 is not isomorphic to the simple projective @xmath363 if @xmath364 .
then by corollary [ commutativediagram ] , we get @xmath365\ ] ] where @xmath366 . by the induction hypothesis we get that @xmath99 is @xmath367-sortable where @xmath367 is the coxeter element admissible for the orientation of @xmath159 , _
i.e. _ @xmath368 .
we get the conclusion using the following criterion which detects @xmath7-sortability : ( * ? ? ?
* lemma 2.1 ) let @xmath369 be a coxeter element . if @xmath370 , then @xmath3 is @xmath7-sortable if and only if @xmath371 is @xmath372-sortable . to the @xmath7-sortable word @xmath373 ,
we associate two series of @xmath8-modules @xmath34 and @xmath374 , and show that they coincide with @xmath36 . for @xmath178
, we define @xmath375-modules @xmath376 . for @xmath377 , @xmath34 is the projective @xmath375-module @xmath378 . for @xmath379 ,
let @xmath2 be the maximal integer such that @xmath336 and @xmath337 .
we define @xmath34 as the cokernel of the map @xmath380 where @xmath381 is a minimal left @xmath382-approximation .
[ calculation of t ] let @xmath0 be the quiver @xmath338 & \\1\ar[rr]\ar[ur ] & & 3}$ ] and @xmath339 in example [ calculation of l and m ] .
let us compute the @xmath34 .
for @xmath383 the @xmath34 are the projective @xmath8-modules , thus we have @xmath384 then we have to compute approximations .
we have a short exact sequence @xmath256 & { { \begin{smallmatrix}}1{\end{smallmatrix}}}\ar[r ] & { { \begin{smallmatrix}}2\\1{\end{smallmatrix}}}\oplus{{\begin{smallmatrix}}&3&&\\1&&2&\\&&&1{\end{smallmatrix}}}\ar[r ] & { { \begin{smallmatrix}}2&&3&&\\&1&&2&\\&&&&1{\end{smallmatrix } } } \ar[r ] & 0},\ ] ] where the map @xmath385 & { { \begin{smallmatrix}}2\\1{\end{smallmatrix}}}\oplus{{\begin{smallmatrix}}&3&&\\1&&2&\\&&&1{\end{smallmatrix}}}}$ ] is the minimal left @xmath386-approximation of @xmath387 .
hence we have @xmath388 .
we have an exact sequence @xmath256 & { { \begin{smallmatrix}}2\\1{\end{smallmatrix}}}\ar[r ] & { { \begin{smallmatrix}}&3&&\\1&&2&\\&&&1{\end{smallmatrix}}}\oplus{{\begin{smallmatrix}}2&&3&&\\&1&&2&\\&&&&1{\end{smallmatrix}}}\ar[r ] & { { \begin{smallmatrix}}\\&3&&&&&\\1&&2&&3&&\\&&&1&&2&\\&&&&&&1{\end{smallmatrix}}}\ar[r ] & 0},\ ] ] where @xmath389 & { { \begin{smallmatrix}}&3&&\\1&&2&\\&&&1{\end{smallmatrix}}}\oplus{{\begin{smallmatrix}}2&&3&&\\&1&&2&\\&&&&1{\end{smallmatrix}}}}$ ] is the minimal left @xmath390-approximation of @xmath391 .
hence we have @xmath392 .
there is an exact sequence @xmath256 & { { \begin{smallmatrix}}2&&3&&\\&1&&2&\\&&&&1{\end{smallmatrix}}}\ar[r ] & { { \begin{smallmatrix}}\\&3&&&&&\\1&&2&&3&&\\&&&1&&2&\\&&&&&&1{\end{smallmatrix}}}\ar[r ] & { { \begin{smallmatrix}}3\\1{\end{smallmatrix}}}\ar[r ] & 0},\ ] ] hence @xmath393 .
so we have @xmath394 for any @xmath38 .
to define the @xmath8-modules @xmath374 , the following notion is convenient .
an _ admissible triple _ is a triple @xmath395 consisting of an acyclic quiver @xmath0 , a coxeter element @xmath7 admissible with respect to the orientation of @xmath0 , and a @xmath7-sortable word @xmath30 .
we denote by @xmath396 the quiver @xmath0 restricted to the support of @xmath397 .
let @xmath395 be an admissible triple with @xmath398 . for @xmath178
, we define @xmath8-modules @xmath399 by induction on @xmath70 .
if @xmath400 then we define @xmath401 , the simple projective @xmath375-module associated to the vertex @xmath352 . assume @xmath402 .
then we write @xmath403 .
it is not hard to check that the triple @xmath404 is an admissible triple with @xmath405 .
therefore by the induction hypothesis we have @xmath406-modules @xmath407 .
for @xmath360 we define @xmath408 where @xmath409 is the reflection functor @xmath410^(.6){r_{u_1}^{- } } & & { { \sf mod \hspace{.02 in } } } kq^{(0)}}\ ] ] at the source @xmath352 of @xmath411 .
[ calculation of u ] let @xmath0 be the quiver @xmath338 & \\1\ar[rr]\ar[ur ] & & 3}$ ] and @xmath339 in examples [ calculation of l and m ] and [ calculation of t ] .
let us now compute the @xmath374 s . by definition @xmath412 .
then we have @xmath413 and finally we have @xmath414 and @xmath415 where @xmath416 is the reflection functor associated to the quiver @xmath417 & 2}$ ]
. therefore we have @xmath418 so we have @xmath419 for any @xmath38 .
[ lequalt ] let @xmath25 be a @xmath7-sortable word where @xmath7 is admissible for the orientation of @xmath0 . 1
we have @xmath420 for @xmath122 .
we have @xmath421 for @xmath122 and @xmath381 is a monomorphism for @xmath422 .
\(1 ) by definition @xmath423 and @xmath424 .
hence we get @xmath425 .
let @xmath99 be the word @xmath426 .
we will prove that @xmath427 for @xmath362 .
by lemma [ lemmelayer1 ] ( 3 ) we have @xmath428
. we can write @xmath429 we have the following exact sequence : @xmath430 & \frac{y}{x}{\otimes}_{\lambda}i_{u_1 } \ar^{b}[r ] & \frac{y}{x}{\otimes}_{\lambda}\frac{i_{u_1}}{i_{c'}i_{u_1 } } \ar[r ] & 0}\ ] ] by proposition [ basicprop ] ( b ) we have the inclusion @xmath431 since @xmath432 is a subword of @xmath433 .
thus we have @xmath434 for any @xmath435 , @xmath436 and @xmath437 .
thus @xmath438 and @xmath439 is an isomorphism .
we have isomorphisms @xmath440 \(2 ) we will now prove that @xmath441 . for @xmath442
this is clear because of a basic property of reflection functors .
assume @xmath443 .
let @xmath2 be the maximal integer such that @xmath337 and @xmath336 .
it exists because @xmath379 and @xmath6 is @xmath7-sortable .
we define the subwords @xmath444 and @xmath445 of @xmath6 .
let @xmath367 be @xmath446 , and @xmath159 be the quiver @xmath447 .
then @xmath448 is an admissible triple .
we have @xmath449 and @xmath450 , thus we have an almost split sequence : @xmath451 applying the reflection functor @xmath452 to this short exact sequence we still get a short exact sequence : @xmath453 which is @xmath454 and the left map is a left @xmath455-approximation , thus a left @xmath456-approximation .
[ remark to lequalt ] the statements @xmath457 for @xmath122 in theorem [ lequalt ] is also true for non - reduced words @xmath458 such that @xmath99 is reduced and that all @xmath331 are subwords of @xmath7 whose supports satisfy @xmath332 the proof above works without any change . note that in this situation , the morphism @xmath381 is a monomorphism for @xmath459 , but the morphism @xmath460 may not be a monomorphism and @xmath461 and @xmath462 may be zero .
[ propertiesofl ] let @xmath6 be a @xmath7-sortable word , where @xmath7 is admissible with respect to the orientation of @xmath0 .
then the @xmath8-modules @xmath36 satisfy the following properties : 1 .
they are non - zero , indecomposable and pairwise non - isomorphic .
2 . the space @xmath463 vanishes if @xmath464 .
\(1 ) since @xmath6 is reduced , @xmath36 is non - zero by proposition [ basicprop ] ( b ) .
since reflection functors preserve isoclasses , the @xmath374 are indecomposable and pairwise non - isomorphic .
\(2 ) using reflection functors , we can assume that @xmath465 is simple projective , and then this is clear .
[ twtilting ] let @xmath466 be an admissible triple .
for @xmath467 , denote by @xmath468 the maximal integer such that @xmath469 .
let @xmath470 1 .
@xmath471 is a tilting @xmath375-module .
we have @xmath472 \(1 ) we prove that @xmath33 is a tilting @xmath375-module by induction on @xmath473 .
if @xmath474 , then the assertion is clear since @xmath475 by definition .
we consider the case @xmath476 .
let @xmath2 be the maximal integer such that @xmath477 and @xmath478 . by the induction hypothesis
we know that @xmath479 is a tilting @xmath375-module where @xmath99 is the word defined by @xmath480 .
by definition we have an exact sequence @xmath481 & e\ar[r ] & t^l_{\mathbf{w}}\ar[r ] & 0}\ ] ] with a minimal left @xmath482-approximation @xmath483 .
then @xmath483 is a minimal left @xmath484-approximation , using the fact that @xmath6 is @xmath7-sortable and corollary [ propertiesofl ] ( 2 ) .
moreover @xmath483 is a monomorphism by theorem [ lequalt ] . by proposition [ tilting mutation ] , we have that @xmath33 is a tilting @xmath375-module .
\(2 ) we prove that @xmath485 by induction on @xmath486 . if @xmath349 , then the assertion is clear .
assume that @xmath402 and write @xmath403 .
_ case 1 : @xmath352 is in the support of @xmath99 _ : this means that @xmath487 .
thus we have @xmath488 then using the induction hypothesis we get @xmath489 moreover we have @xmath490 by definition of the @xmath34 there exists a short exact sequence : @xmath491 & e\ar[r ] & t^j_{\mathbf{w}}\ar[r ] & 0}\ ] ] where @xmath492 is in @xmath493 and where @xmath38 is the minimal integer such that @xmath494 and @xmath495 .
it exists since @xmath352 is in the support of @xmath99 .
the approximation map is a monomorphism by theorem [ lequalt ] ( 2 ) , thus @xmath496 is in @xmath497 .
thus we have @xmath498 .
_ case 2 : @xmath352 is not in the support of @xmath99 .
_ then it is easy to see that @xmath499 and we get @xmath500 * the short exact sequence @xmath252^f & e\ar@{->>}[r ] & l^j_{\mathbf{w}}}$ ] in @xmath155 is an almost split sequence in the category @xmath39 .
* this almost split sequence is an element of @xmath501 , which is the ` 2-calabi - yau complement ' of the short exact sequence @xmath502 & k\ar@{->>}[r ] & l^k_{\mathbf{w}}}$ ] of proposition [ dimext1 ] .
[ calculation of sub t ] let @xmath0 be the quiver @xmath338 & \\1\ar[rr]\ar[ur ] & & 3}$ ] and @xmath339 in example [ calculation of t ] .
the module @xmath33 is by definition @xmath503 .
it is easy to check theorem [ twtilting ] .
the module @xmath33 is a tilting module over @xmath8 , and we have @xmath504 in this section we establish the converse of theorem [ twtilting ] .
hence we get a natural bijection between tilting @xmath8-modules with finite torsionfree class and @xmath7-sortable elements in @xmath11 .
let us start with some preparation . to any ( not necessarily reduced ) word @xmath373 such that all @xmath331 are subwords of @xmath7
whose supports satisfy @xmath505 we can associate @xmath375-modules @xmath34 for @xmath122 and @xmath33 in the same way as in the @xmath7-sortable case . [ csortableifftilting ]
let @xmath506 be as above .
assume that @xmath6 is non - reduced and that @xmath99 is reduced .
then the number of indecomposable summands of @xmath33 is strictly less than @xmath507 . by remark [ remark to lequalt ]
, we have @xmath508 .
since @xmath6 is not reduced , this is zero by proposition [ basicprop ] ( b ) .
since @xmath99 is reduced , all @xmath509 ( @xmath510 ) are indecomposable by theorem [ 2spherical ] .
therefore we have the assertion . [ ar things ]
let @xmath0 be an acyclic quiver and @xmath130 be a tilting @xmath8-module . 1 .
the category @xmath511 has almost split sequences .
2 . if @xmath511 has finitely many indecomposable modules , then the ar - quiver of @xmath511 is a full subquiver of the translation quiver @xmath512 .
\(1 ) this is well - known @xcite .
\(2 ) we can clearly assume that @xmath0 is connected .
since @xmath130 is a tilting module , then all indecomposable projectives are in @xmath513 .
the irreducible maps between projectives in @xmath513 coincide with the irreducible maps between projectives in @xmath155 , so that @xmath0 is a full subquiver of the ar - quiver of @xmath513 .
moreover , for any indecomposable module in @xmath513 , there is a nonzero map from an indecomposable projective module . since @xmath513 is of finite type and @xmath0
is connected , it follows that the ar - quiver of @xmath513 is connected .
we now claim that each indecomposable module in @xmath513 is of the form @xmath514 , where @xmath45 is the ar - translate in @xmath513 and @xmath515 is indecomposable projective . if not
, then since @xmath513 is of finite type , there is some @xmath45-periodic indecomposable @xmath516 .
then , since the quiver of @xmath513 is connected , there must be an irreducible map between some periodic indecomposable @xmath516 and some @xmath514 with @xmath515 indecomposable projective . applying @xmath517
we can assume that the second module is @xmath515 .
if @xmath518 is irreducible , then @xmath516 is projective , a contradiction . if @xmath519 is irreducible , then @xmath520 is irreducible ,
so @xmath521 is projective , a contradiction .
thus each indecomposable of @xmath513 is of the form @xmath514 , where @xmath515 is indecomposable projective .
then using the fact that @xmath0 is a full subquiver of the ar - quiver of @xmath513 , we deduce that the ar - quiver of @xmath513 is a full subquiver of @xmath522 . from lemmas
[ csortableifftilting ] and [ ar things ] we deduce a nice consequence .
[ subtfinite ] let @xmath0 be an acyclic quiver .
let @xmath7 be a coxeter element admissible with respect to the orientation of @xmath0 .
let @xmath130 be a tilting @xmath8-module .
assume that @xmath511 has finitely many indecomposable modules .
then there exists a unique @xmath7-sortable word @xmath6 such that @xmath523 . without loss of generality , we assume @xmath524 .
we denote by @xmath45 the ar - translation of @xmath513 . for any @xmath144 ,
we denote by @xmath525 the minimal number satisfying @xmath526 , which exists by lemma [ ar things ] .
then for @xmath527 we look at the set @xmath528 and define @xmath529 then the word @xmath530 where @xmath531 satisfies @xmath532 for each expression @xmath533 , we define @xmath534 as the number of @xmath15 ( @xmath144 ) appearing in @xmath99 . by using induction on @xmath535
, we have @xmath536 by using the almost split sequences in @xmath513 and the shape of the ar - quiver of @xmath513 given in lemma [ ar things ] ( 2 ) .
in particular the number of indecomposable direct summands of @xmath537 is exactly @xmath132 since @xmath538 for any @xmath144 .
moreover we have @xmath539 since @xmath540 for any @xmath144 .
we only have to check that @xmath6 is reduced .
otherwise we take an expression @xmath533 such that @xmath99 is non - reduced and @xmath535 is minimal with this property . by lemma [ csortableifftilting ] , the number of indecomposable direct summands of @xmath537 is less than @xmath132 , a contradiction .
thus @xmath6 is reduced . as a consequence
we get the following : if @xmath130 is a tilting @xmath8-module such that @xmath513 is of finite type , then all indecomposables in @xmath513 are rigid as @xmath8-modules . combining theorem [ subtfinite ] with theorem [ twtilting ]
we get the following result which was first proved using other methods in @xcite .
there is 1 - 1 correspondences * @xmath541 & \ { c \textrm{-sortable words with } c^{(0)}=c\}}.$ ] * @xmath542 & \ { c \textrm{-sortable words}\}}.$ ] dually , we can state the defintion .
let @xmath7 be a coxeter element of the coxeter group @xmath11 admissible with respect to the orientation of @xmath0 .
an element @xmath3 of @xmath11 is called _ co - c - sortable _ if there exists a reduced expression @xmath543 of @xmath3 of the form @xmath544 where all @xmath331 are subwords of @xmath7 whose supports satisfy @xmath332 note that this definition is equivalent to the fact that @xmath545 is @xmath546-sortable .
we denote by @xmath396 the quiver @xmath0 restricted to the support of @xmath397 . from a @xmath7-sortable word @xmath6
we define in this subsection , @xmath375-modules @xmath547 , @xmath548 and @xmath549 in a dual manner to the modules @xmath34 , @xmath33 and @xmath374 defined in the previous subsections . for @xmath178
, we define @xmath375-modules @xmath547 . for @xmath377
, @xmath550 is the injective @xmath375-module @xmath551 . for @xmath379 ,
let @xmath2 be the maximal integer such that @xmath336 and @xmath337 .
we define @xmath550 as the kernel of the map @xmath552 where @xmath553 is a minimal right @xmath554-approximation .
then we define a @xmath8-module @xmath548 as the direct sum @xmath555 , where @xmath468 is the maximal integer such that @xmath469 .
a _ co - admissible triple _ is a triple @xmath395 consisting of an acyclic quiver @xmath0 , a coxeter element @xmath7 admissible with the orientation of @xmath0 , and a co-@xmath7-sortable word @xmath544 .
let @xmath395 be a co - admissible triple with @xmath556 . for @xmath178 , we define @xmath8-modules @xmath549 by induction on @xmath70 .
if @xmath400 then we define @xmath557 , the simple injective @xmath375-module associated to the vertex @xmath352 .
assume @xmath402 .
then we write @xmath558 .
it is not hard to check that the triple @xmath404 is a co - admissible triple with @xmath405 .
therefore by the induction hypothesis we have @xmath406-modules @xmath559 .
for @xmath360 we define @xmath560 where @xmath561 is the reflection functor @xmath410^(.6){r_{u_1 } } & & { { \sf mod \hspace{.02 in } } } kq^{(0)}}\ ] ] at the sink @xmath352 of @xmath411 .
then a dual version of theorems [ lequalt ] ( 2 ) and [ twtilting ] hold .
more precisely we have the following .
let @xmath0 be an acyclic quiver and @xmath7 be the associated coxeter element . then for a co-@xmath7-sortable element @xmath6 the following hold .
* for all @xmath334 , we have @xmath562 , and @xmath563 is an epimorphism . * the subcategory @xmath564 is finite , and @xmath565 .
dually to theorem [ subtfinite ] , we also have the following . for any tilting module @xmath566 such that @xmath567 is finite , then the following hold . *
the ar - quiver of @xmath568 is a full subquiver of @xmath522 .
* there exists a ( unique ) co-@xmath7-sortable word @xmath6 ( which can be constructed from the ar - quiver of @xmath568 ) such that @xmath569 .
there is 1 - 1 correspondences * @xmath570 & \ { \textrm{co-}c \textrm{-sortable words with } c^{(0)}=c\}}.$ ] * @xmath571 & \ { \textrm{co-}c \textrm{-sortable words}\}}.$ ] let @xmath6 be a co-@xmath7-sortable word with @xmath32 .
we denote by @xmath572 ( resp .
@xmath573 ) the auslander algebra of the category @xmath564 ( resp .
@xmath574 ) , that is @xmath575 following ( * ? ? ?
* theorem 5.21 ) we get the following result : [ equivalence ] there exists a commutative diagram of triangle functors : @xmath576 ^ -{\textrm{res . } } \ar[d ] & { \mathcal{d}}^b(a_{\mathbf{w}})\ar[r]^{-{\overset{\mathbf{l}}{{\otimes}}}_{a_{\mathbf{w}}}m_{\mathbf{w } } } & { \mathcal{d}}^b(\lambda_w)\ar[d ] \\
{ \mathcal{c}}_2(\overline{a}_{\mathbf{w}})\ar[rr]^f & & \underline{{{\sf sub \hspace{.02 in } } } } \lambda_w , } \ ] ] where @xmath577 is the generalized @xmath86-cluster category defined in @xcite associated with the algebra @xmath573 of global dimension at most 2 , and where @xmath578 is an equivalence of categories .
the proof in @xcite deals with @xmath548 in the preinjective component of the ar - quiver of @xmath155 ( that is @xmath579 concealed ) , but the proof only uses the fact that @xmath564 is finite and that the ar - quiver of @xmath564 is a full subquiver of @xmath522 . note that for any element @xmath3 , the @xmath86-cy category @xmath13 is equivalent to a generalized @xmath86-cluster category @xmath580 @xcite , but the algebra @xmath47 of global dimension 2 is constructed in a very different way . a link between the construction given here and the construction of @xcite is given in @xcite .
the proof in @xcite does not carry over for @xmath7-sortable words .
indeed , for a general co-@xmath7-sortable word , the ar - quiver of @xmath564 is a subquiver of the quiver of @xmath26 , fact which is used in the proof . for a @xmath7-sortable word @xmath6 ,
the ar - quiver of @xmath39 is not a subquiver of the quiver of @xmath26 .
in this section we discuss some possible generalizations of the description of the layers in terms of tilting modules , beyond the @xmath7-sortable case . we pose some problems and give some examples to illustrate limitations for what might be true .
recall from section 2 that to a reduced expression @xmath6 of an element @xmath3 in @xmath11 we have associated a set @xmath48 of @xmath581 indecomposable rigid @xmath1-modules which we call layers , and which are indecomposable rigid @xmath8-modules when @xmath6 is @xmath7-sortable , where @xmath7 is admissible with respect to the orientation of @xmath0 . under the same assumption ( _ i.e. _ @xmath6 is @xmath7-sortable ) , we constructed a set @xmath49 of @xmath581 indecomposable @xmath8-modules via minimal left approximations , starting with the tilting module @xmath8 , and ending up with a tilting module @xmath33 .
all minimal left approximations were monomorphisms .
we showed that the two sets of indecomposable modules coincide .
in particular , the module @xmath582 , where for @xmath583 the integer @xmath468 is the position of the last reflection @xmath15 in the word @xmath6 , is a tilting module over @xmath8 .
we now consider the case of words @xmath6 with the assumption that @xmath584 , where @xmath7 is a coxeter element admissible with respect to the orientation of @xmath0 .
when @xmath585 is a word , we define @xmath33 to be a tilting module associated with @xmath6 if it is possible to carry out the following .
start with @xmath586 , where @xmath44 is the indecomposable projective @xmath8-module associated with the vertex @xmath91 .
if possible , exchange @xmath587 with a non - isomorphic indecomposable @xmath8-module to get a tilting module @xmath588 , then replace summand number @xmath589 in @xmath590 by a non - isomorphic indecomposable @xmath8-module to get a new tilting module @xmath591 , etc .
if an exchange is possible at each step , we obtain a tilting module @xmath33 .
we say that a word @xmath584 starting with a coxeter element is _ tilting _
if @xmath33 exists , and @xmath6 is _ monotilting _ if morever @xmath33 is obtained by only using left approximations .
hence @xmath7-sortable words are examples of monotilting words .
* characterize the tilting words @xmath6 .
in particular is every reduced word @xmath584 starting with a coxeter element tilting ? * characterize the monotilting words .
* when do two tilting words @xmath592 and @xmath593 give rise to the same tilting module ? or formulated differently , for which tilting words @xmath6
do we have @xmath594 ?
let @xmath0 be the quiver @xmath282 & \\ 1\ar[dr]\ar[ur ] & & 4\\ & 3\ar[ur ] & } $ ] and @xmath595
. then @xmath6 is not reduced .
one can easily check that we have @xmath596 then the minimal left @xmath597-approximation of @xmath598 is @xmath599 .
it is a monomorphism whose cokernel is @xmath600 the minimal left @xmath601-approximation of @xmath602 is @xmath603 .
it is a monomorphism whose cokernel is @xmath604 .
hence @xmath6 is monotilting .
let @xmath0 be the quiver @xmath282 & \\ 1\ar[dr]\ar[ur ] & & 4\\ & 3\ar[ur ] & } $ ] and @xmath606 .
then one can show that @xmath6 is monotilting and that @xmath607 then one can check easily that all the modules of the form @xmath608 are in @xmath609 .
however , it may happen that @xmath605 is of finite type for a tilting word @xmath6 which is not @xmath7-sortable .
it follows from theorem [ subtfinite ] that there exists a unique @xmath7-sortable word @xmath610 such that @xmath611 .
we then pose the following .
[ examplecompletion ] let @xmath0 be the quiver @xmath282 & \\ 1\ar[dr]\ar[ur ] & & 4\\ & 3\ar[ur ] & } $ ] and @xmath614 .
then @xmath6 is monotilting and we have @xmath615 then @xmath6 is not @xmath7-sortable , @xmath605 is finite and one can check that @xmath616 . when @xmath6 is @xmath7-sortable , @xmath6 is a monotilting word and @xmath33 coincides with @xmath617 given by the layers . in general @xmath617
is not a @xmath8-module , but as we have seen there is an indecomposable @xmath8-module associated with each indecomposable summand of @xmath617 , and hence a @xmath8-module @xmath618 associated with @xmath617 . in this connection we have the following questions : * for which @xmath6 does the following hold ? * * each indecomposable summand of @xmath618 is rigid , * * @xmath618 is a tilting module , * * @xmath6 is tilting and @xmath619 , * * @xmath6 is monotilting and @xmath619 . * if @xmath6 is monotilting and @xmath618 is rigid , do we have @xmath619 ?
let @xmath0 be the quiver @xmath282 & \\ 1\ar[ur]\ar[rr ] & & 3}$ ] , and @xmath620 .
the word @xmath621 is monotilting and we have @xmath622 to exchange @xmath287 we have to use the minimal right approximation @xmath623 & { { \begin{smallmatrix}}3\\1{\end{smallmatrix}}}}.$ ] hence @xmath6 is a tilting word which is not monotilting and we get @xmath624 the cluster - tilting object @xmath24 of @xmath12 associated with @xmath6 has the indecomposable summands : @xmath625 we then see that @xmath619 , even though @xmath6 is not a monotilting word .
let @xmath0 and @xmath6 be as in example [ examplecompletion ] .
then we have @xmath626 therefore we obtain @xmath627 each indecomposable summand is rigid , but @xmath618 is not a tilting module
. therefore we can have ( a ) without ( b ) .
k. erdmann , n. snashall , _ preprojective algebras of dynkin type , periodicity and the second hochschild cohomology _ ,
algebras and modules , ii ( geiranger , 1996 ) , 183193 , cms conf .
, 24 , amer .
soc . , providence , ri , 1998 . | let @xmath0 be an acyclic quiver and @xmath1 be the complete preprojective algebra of @xmath0 over an algebraically closed field @xmath2 . to any element @xmath3 in the coxeter group of @xmath0 , buan , iyama , reiten and scott
have introduced and studied in @xcite a finite dimensional algebra @xmath4 . in this paper
we look at filtrations of @xmath5 associated to any reduced expression @xmath6 of @xmath3 .
we are especially interested in the case where the word @xmath6 is @xmath7-sortable , where @xmath7 is a coxeter element .
in this situation , the consecutive quotients of this filtration can be related to tilting @xmath8-modules with finite torsionfree class . |
pulsating white dwarfs show brightness variations at a range of frequencies , with amplitudes of up to a few per cent .
they appear to be normal white dwarfs , except for the fact that they lie in a special temperature range , or instability strip , in which they are unstable to pulsations .
there are instability strips for all three of the main white dwarf types , da , db and do ( atmosphere dominated by , and , respectively ) , around 12000 , 23000 and 100000k , respectively .
pulsating white dwarfs offer the prospect of physical insight via a number of different methods .
one is asteroseismology , the determination of a white dwarf s internal structure through the comparison of observed periodicities with model predictions of stellar eigen - modes .
another method analyses which modes are present , what their amplitudes are and how the amplitudes vary . the aim is to learn about the physics of the upper layers of the star , by determining the way in which modes interact with different parts of the star and with each other .
robinson , kepler & nather ( @xcite ) showed that white dwarfs pulsate in non - radial gravity modes , and that the observed brightness variations result from perturbations of their temperature , not of their geometry or gravity .
they also estimated the expected variations in the line - of - sight velocity and the line profile , and concluded that these would be too small to measure with then available instrumentation .
as we will show below , this statement no longer holds .
the pulsations were originally thought to be excited in an ionisation zone , as in classical pulsators such as the cepheids .
the dynamics in da and db pulsating white dwarfs are rather unlike those in classical pulsators , however , in that for the white dwarfs the surface convection zone can respond almost instantaneously to gravity - mode pulsations , because the turnover times are much shorter than the mode periods . in a series of papers , brickhill
( @xcite , @xcite , @xcite,@xcite ) has described the response in detail , and concludes it leads to significant `` convective driving '' , which will be sufficient to overcome the damping in the radiative interior for modes with periods of order of or shorter than the thermal response time of the convection zone .
brickhill s partly numerical results have recently been confirmed analytically ( wu @xcite ; goldreich & wu @xcite ) .
the da and db pulsating white dwarfs often show power at frequencies that are the sums or differences of frequencies with stronger power . based on the density of the theoretical gravity - mode spectra in the relevant frequency range
, it has been argued that these `` combination frequencies '' do not represent real eigen - modes , but are instead produced by nonlinear mixing of pulsation signals of real eigen - modes ( `` parent modes '' ) in the outer layer of the star .
some support for this idea comes from observed correlations in power .
brickhill ( @xcite , b ) has suggested that the mixing is a by - product of the interaction between the pulsations and the convection zone .
because of the relatively slow thermal response of the convection zone , flux variations take some time to appear at the surface , and they are reduced in amplitude . but the thickness of the convection zone is modulated by the pulsation , so that the signal is not only diminished and delayed , but also distorted .
it is the latter distortion which translates into the power seen at combination frequencies . both to be able to make an asteroseismological analysis and to verify the pulsation models
, it is essential to identify periodicities with eigen - modes , i.e. , obtain their radial order , spherical degree and azimuthal order . for some db and do variables , it has been possible to obtain reasonably sound identifications from a direct comparison of observed frequencies with model predictions .
however , for many other stars , including all da variables , there are either too few modes , or the mode structure is too variable .
robinson et al .
( ) implemented a method to identify spherical degree , which combined broad - band measurements of mode amplitudes at optical and uv wavelengths .
it makes use of the wavelength dependence of limb darkening , which causes the pulsation amplitudes to vary with wavelength in a manner that depends on .
based on its success , we were motivated to implement a variation of this method which uses high signal - to - noise time - resolved spectrophotometry at optical wavelengths only .
as will become clear , our initial data on the star showed variations not only in flux but also in line - of - sight velocity .
the broadened scope , as well as the increased complexity in modeling the flux variation as a function of wavelength , made us decide to present the amplitudes and phases of the modes as a function of wavelength in a separate paper ( clemens et al.@xcite , hereafter paper ii ) . in this paper , we focus on the implications of the line - of - sight velocity variations .
we first briefly review the basic properties of ( ) , and then describe the observation and data reduction in detail ( ) . in , we analyse the periodicities in the light curve , and in we search for corresponding variations in the line - of - sight velocity curve .
we discuss the implications of our results in , both in a model - independent way and in the context of the convective driving models described above .
[ [ sec : zzpsc ] ] ( zz psc ; wd 2326 + 029 ) was discovered as a variable by shulov & kopatskaya ( @xcite ) and studied in some detail by mcgraw and robinson ( @xcite ) .
it is bright ( @xmath7 ) , at the cool end of the instability strip ( @xmath8k , bergeron et al .
paper ii ) and shows large - amplitude modulations ( up to a few per cent ) .
the source remained relatively obscure until zuckerman & becklin ( @xcite ) reported an infrared excess , which is still unexplained ( for recent developments , see kleinman et al . ; koester , provencal & shipman @xcite ) . in 1987
, it was first subjected to scrutiny with the whole earth telescope ( wet ; winget et al . ) . unlike for the variable dbs and dos , however , the observations while of excellent quality yielded more questions than answers .
partly , this is because of the complex and variable frequency spectrum of : typically , dozens of periodicities are seen , with many occurring at frequencies that are differences or sums of the frequencies of other periodicities .
kleinman et al .
( ) discussed an extensive collection of data sets , obtained over many seasons .
they find that many periodicities are variable and unstable from year to year , but some do recur . by considering the whole ensemble
, they could identify with some certainty which periodicities were associated with real modes and which with combination frequencies .
they found nineteen real modes , the majority of which very likely have @xmath9 .
however , due to the vagaries of the mode characteristics especially the enigmatic , confusing behavior of multiplets , showing different and changing splitting their results are suggestive , but not definitive .
spectra of were taken on 1996 november 19 using the low resolution imaging spectrometer ( oke et al . ) on the 10-m keck ii telescope . from 4:42 till 9:25
ut , a continuous series of 700 12s exposures was made .
only 100 pixels in the spatial direction ( corresponding to @xmath10arcsec ) , binned by a factor 2 , were read out through two amplifiers , leading to a read - out and re - set time in between exposures of about 12s .
the @xmath11 grating was used , covering the range 34505950 at @xmath12 .
the weather started photometric , and we used a @xmath13arcsec slit to make use of that .
hence , the seeing of @xmath14arcsec determined the wavelength resolution of @xmath15 . in principle
, it would have been best to keep the slit at the parallactic angle in order for refraction not to influence the wavelength scale , but this would have required non - standard guiding and we decided to keep the slit at position angle 0 instead . at the end of the series ,
a bank of cloudlets came in , causing the last 55 spectra , as well as one spectrum some ten minutes earlier , to have a reduced source and increased sky rate ( the moon was up , so these are easy to identify ) .
we have not used any of these spectra in our analysis .
the series of exposures was followed by fifteen 4s integrations on the spectrophotometric flux standard 191-b2b , using the same setup .
each of the series was followed by hg / kr wavelength calibration and halogen flat - field frames .
the reduction of all spectra was done as follows .
the frames were bias - corrected separately for the two amplifiers using the overscan regions , and then corrected for the gain difference between the two amplifiers as determined from halogen spectra .
the sky background was estimated by fitting parabolae for each position in the dispersion direction , excluding 21 binned pixels ( @xmath16 ) centred on the stellar spectrum .
the latter pixels were used to estimate the stellar flux using an optimal weighting scheme similar to that of horne ( @xcite ) . note that we did not flat - field the data .
the reason is that we had been unable , due to an instrument failure shortly after taking the exposures on g191-b2b , to obtain a large enough number of halogen spectra to construct a flat - field that matched the signal - to - noise ratio of our observations , especially in the blue . since the flat fields that we did have appeared very smooth , we decided to forego flat - fielding altogether
the resulting spectra look smooth ( ) , except for a few pixels that are easy to identify .
note the presence of metals
@xmath17 and @xmath18 uncovered only recently by koester et al .
( @xcite ) .
accurate flux calibration was not possible , since the 191-b2b spectra were taken while the bank of cloudlets was still passing .
we selected the three most exposed spectra of the series of fifteen acquired , as these have comparable count rates and hence may not be influenced too much , making the relative flux calibration more reliable magnitude of @xmath19 ( @xmath20mjy ; mccook & sion @xcite ) . ] . from these three , we derived the instrumental response using the model fluxes provided by bohlin , colina & finley ( @xcite ) , and the atmospheric extinction curve of beland , boulade & davidge ( @xcite ) .
typical examples of relatively high and low - flux spectra , as well as the average spectrum , are shown in .
special care was taken to correct the wavelength scale for instrumental effects due to flexure and atmospheric refraction .
both effects can be seen in the pixel positions on the detector as a function of time of the @xmath21 sky line and the balmer lines ( a ) . here , the positions were derived by cross - correlation of the spectra with the average of those spectra taken with the slit position angle less than 20 away from the parallactic angle .
the break at @xmath22s in the balmer - line positions is due to guiding being stopped and re - started ( to take a guider image ) .
the flexure is described well by a cubic fit to the sky - line positions .
after correction for the flexure , as well as for the jump due to switching the guider off and on , the positions of the balmer lines vary linearly with what is expected from refraction ( @xmath23 , where @xmath24 is the difference between parallactic and position angle , and @xmath25 the zenith distance ) .
the different slopes indicate that our guiding wavelength was approximately 7000 , as expected for the ccd camera used .
we used this value to calculate the wavelength corrections for the whole spectrum .
since our spectra were taken through a wide slit , small movements of the star will result in random apparent wavelength shifts for individual spectra .
the effect can be estimated by measuring the shifts along the slit from the spatial profiles ( averaged over some number of pixels along the dispersion direction ) .
we find a root - mean - square variation in these of 0.084 binned pixels , corresponding to 0036 .
assuming the variations in the dispersion direction are similar , one thus expects variations of about 0.2 , or 14 at h@xmath26 .
apparent brightness variations can result from varying slit losses due to seeing - related changes in the width of the stellar images . from the spatial profiles , we find 10% root - mean - square variations around the average full width at half maximum of 3 binned pixels ( 12 ) .
using images taken with lris for other projects , we estimate that the slit losses with an @xmath27 aperture are about 1% on average . assuming that the slit losses scale approximately quadratically with the seeing , one thus expects apparent brightness variations of the order of 0.2% . this is smaller than the poisson noise in individual wavelength elements , but dominates in the average light curve discussed below .
it is also much larger than variations due to scintillation ( @xmath28 ; young @xcite ) .
in b , we show the continuum light curve , constructed by averaging the 52005500 region .
the light curve shows the presence of a number of periodicities , and is similar to light curves observed before ( see ) .
the fourier transform is shown in a. most of the power is at frequencies below 6mhz .
an idea of the background level can be obtained from the 1020mhz region ( see inset in figure ) , where little evidence for any periodicity is seen . in this region
, we find a root - mean - square amplitude of about 0.025% .
the corresponding root - mean - square variation is half that , and one infers root - mean - square variations in the fluxes of about @xmath29 ( where @xmath30 is the number of data points ) , more or less consistent with the estimate made above based on varying slit losses .
we believe essentially all the signal seen in the main panel of a , which is well above 0.025% , reflects stellar variability .
we identified the periodicities sequentially .
we started by measuring the approximate frequency of the highest peak and fitting the data to a function of the form @xmath31 , where @xmath32:52:08ut is the time relative to the mid - observing time , solving for pulsation frequency @xmath33 , phase @xmath34 and amplitude @xmath35 , plus a constant offset @xmath36 .
next , we identify the highest remaining peak in the fourier transform of the residuals and include another cosine wave in the fit , etc . for combination frequencies , we held the frequencies fixed at the corresponding combinations of parent - mode frequencies .
the decision whether a peak corresponded to a combination frequency or to a real mode , was generally made consulting the work of kleinman ( @xcite ) ; in all cases , the amplitude of the combination frequency was the smallest of the peaks involved .
continuing as long as the next largest peak had not been influenced greatly by removing the previous peaks , we found eleven periodicities with amplitudes larger than 0.55% .
@lrrrrrrrr@ real mode & & & & & & & & + & & & & & & & & + @xmath37 & 614 & & & & & & & @xmath38 & 818 & & & & & & & @xmath39 & 653 & & & & & & & @xmath40 & 776 & & & & & & & @xmath41 & 283 & & & & & & & @xmath42 & 430 & & & & & & & combination & & & & & & & & + frequency & & & & & & & & + @xmath43 & 351 & & & & & & & @xmath44 & 2469 & & & & & & & @xmath45 & 409 & & & & & & & @xmath46 & 317 & & & & & & & @xmath47&10323 & & & & & & & of the eleven periodicities found , there appear to be six real and five combination frequencies .
these are listed in and the corresponding peaks are identified in a. comparing our periodicities with those found by kleinman ( @xcite ) , we find that most periodicities have been seen before .
the exceptions are : ( i ) f6 : power at 430s has only been seen as an apparent harmonic of a strong 860s mode ; ( ii ) f3+f1 : power at f1 ( 614s ) and f3 ( 653s ) has not been seen simultaneously before , and hence neither has this combination ( although power at 317s due to a different combination has been observed ) ; and ( iii ) f2@xmath48f1 and f3@xmath48f1 : at 2469 and 10322s , these periods are longer than any generally considered ( the identification of f3@xmath48f1 should be considered tentative here too ) .
absent is power at 400s , which has been present in almost all observations so far . from the residuals
, it is clear that the eleven modes do not account well for the observed light curve ( with our estimate of 0.2% root - mean - square uncertainties , we find a formal @xmath49 ) .
we have tried continuing the identification process ( see appendix ) , but fail to account for all the variability .
therefore , we decided to focus on the eleven modes for which we have full confidence .
we note , however , that because of the additional power and the resulting non - normal deviations , the formal errors on the parameters are not very meaningful .
to get an idea of relative uncertainties , we have scaled up the errors on the data points such that the reduced @xmath50 equals unity .
a puzzling aspect of our results is that the three largest periodicities appear to have frequencies related to each other by simple integer ratios : @xmath51 , @xmath52 .
both are correct to within 0.2% . after identifying more modes and including these in the fit ( see appendix ) ,
the ratios become more accurate ( to within 0.1% ) . furthermore
, another mode , f8 , is identified for which @xmath53 , again to within 0.1% .
similar close coincidences have not been found in the observations of kleinman ( @xcite ) and it may be that they are in some way due to our relatively short time span , perhaps in combination with the badness of the fit , especially near the light maxima ( see b ) . on the other hand , these ratios may represent unexplored physics underlying the pulsation spectrum . for the present time , we will ignore possible complications arising from this .
the positions of the balmer lines on the detector , as derived by the cross - correlations described in , appear to show modulation on time - scales similar to those shown by the light curve ( see a ) .
this modulation might reflect apparent variations in the line - of - sight velocity of . in order to study this in more detail
, we determined doppler shifts for all the spectra by fitting the h@xmath54 , h@xmath26 and h@xmath55 lines ( details in below ) .
the resulting line - of - sight velocity curve is shown in c ( upper curve ) .
the fourier spectrum , shown in b , shows one strong peak , at a frequency of @xmath56hz , i.e. , consistent with that of the strongest periodicity in the light curve .
inspired by the close correspondence between the frequencies of these signals , we looked for power at frequencies corresponding to the other ten periodicities in the light curve and found small peaks at some but not all of them ( ) .
we determined their amplitudes and phases by fitting the velocity curve with a combination of eleven cosine waves with frequencies fixed at those determined from the light curve ( plus a third - order polynomial to account for slow variations ) .
the results are listed in and indicated in b ; the residuals are shown in c ( bottom curve ) .
note that unlike for the light curve , the fit to the line - of - sight velocities is formally acceptable ( @xmath57 ) and we expect our error estimates to be reliable .
could these line - of - sight velocity variations reflect pulsational motion ?
the pulsations are g - modes and the associated motion is largely horizontal ( e.g. , dziembowski @xcite ) . the observed line - of - sight velocity , i.e.
, the physical velocities projected on the light of sight and integrated over the stellar disk , will thus be dominated by motion near the limb .
the observed amplitude will be smaller than the physical velocity amplitude by an amount depending on the spherical degree and the direction of the pulsation axis relative to the observer .
robinson et al .
( @xcite ) estimated the expected line - of - sight velocity amplitudes for typical pulsation amplitudes and found that they should be a few .
this is comparable to the amplitudes we find .
if we suppose that the line - of - sight velocity variations reflect horizontal motion associated with the white dwarf pulsation , it is very interesting that all five apparently significant detections ( @xmath58 ) are for periodicities we identified as real modes above ; no significant detections are found for the combination frequencies .
before we turn to the implications , however , we need to verify that our error estimates are reasonable and the velocity detections significant ; and if so , whether the inferred line - of - sight velocities reflect intrinsic velocity variations rather than some systematic effect in our measurement procedure . we will address these two points in turn .
looking at b , all peaks but that associated with f1 look insignificant .
indeed , they would be formally insignificant if one did not know the frequency at which to look in advance .
however , the pulsation frequencies are determined accurately from the light curve . in the absence of systematic effects ( considered separately in )
, a rough estimate for the significance of a peak with a certain amplitude at a frequency specified in advance can be made by counting the number of other peaks in the velocity spectrum at or above that amplitude .
for instance , there are 13 peaks with amplitudes above 2 at frequencies below 7mhz ( b ) . with a frequency resolution of @xmath59hz ,
the total number of independent frequency elements in this frequency range is 109 , and thus the probability of finding a @xmath60 peak by chance at a given frequency is roughly 1:8 .
one would estimate 1:14 if one counts peaks up to the nyquist frequency ( 21mhz ; 23 peaks over 2 in 322 [ @xmath61 frequency elements ) .
thus , f2 , f3 and f4 , all of which have amplitudes in excess of @xmath62 , are likely real .
there is no other peak as large as f1 , hence the probability that it is a chance coincidence is @xmath63 . in order to obtain more quantitative estimates , we ran monte - carlo simulations , in which we fitted scrambled velocity curves in the same way as we fitted the observations .
the scrambled curves were created by first taking out the slow third - order variation in the measured doppler shifts , and then re - distributing the resulting velocities randomly over the measurement times . by counting the number of artificial data sets for which the fit produces an amplitude larger than that observed
, one can infer the likelihood of a false detection .
the likelihood distributions for all periodicities but f3@xmath48f1 are consistent with each other ( because of its long period , f3@xmath48f1 is covariant with the third - order polynomial , and its likelihood distribution is wider ) . in , we show the cumulative distribution of the amplitudes for all modes but f3@xmath48f1 derived from one thousand simulations .
we also indicate the observed amplitudes .
we infer that f1 , f2 , f3 and f4 are detected at the @xmath64 , 98 , 98 and 95% confidence levels , respectively ; f5 is marginally detected ( @xmath65 confidence ) ; and the signals at all other periodicities are not significant . as mentioned above , the fit to the line - of - sight velocities is formally acceptable ( @xmath57 ) .
the errors , dominated by the star wandering around in the slit ( ) , should be more or less normally distributed , and thus our error estimates likely reasonable .
indeed , from the simulations one infers similar numbers : @xmath66^{1/2}\simeq0.8\,\kms$ ] ( where the factor @xmath67 accounts for the random phase ) , and @xmath68deg .
the exception is f3@xmath48f1 , for which @xmath69 , indicating that the uncertainty for this periodicity is underestimated in the least - squares fit ( ) . this reflects the large covariance with the third - order polynomial .
while the error estimates should be reliable , we caution that the uncertainty distributions are not expected to be normal .
this is a consequence of our choice of parameters @xmath70 and @xmath71 .
for instance , from , the uncertainty on @xmath71 is 28 for f5 . the @xmath72 uncertainty , however
, is far larger than 56 : indeed , at 95% confidence @xmath71 can have any value , as the detection of velocity variations at f5 is significant at @xmath73 confidence .
the uncertainties on @xmath74 and @xmath75 are normally distributed , and one can use these to determine reliable uncertainty estimates for @xmath70 and @xmath71 at different confidence levels . note that the same holds for @xmath76 and @xmath77 , but the more significant detections make this less of an issue .
above , we have shown that at frequencies at which the flux of the star varies strongly , significant variations are also present in the line - of - sight velocity curve .
the question that remains is whether these variations reflect the intrinsic motion in the white dwarf s outer layers expected from the estimates of robinson et al.(@xcite ) , or whether they might be biased or perhaps even result from inadequacies in the way we determined the velocities . for instance , through changes in continuum slope , the measured doppler shift might be influenced by the variations in temperature with pulsation phase . for any systematic effect related to the flux variations ,
the line - of - sight velocity variations have to be either in phase or in anti - phase with the light variations , i.e. , the phase difference between velocity and light @xmath78 or @xmath79deg .
this is not what we observe : for f1 , @xmath80deg .
thus , no more than @xmath81 could be related to systematic effects of this kind .
the component out of phase with the light variations has an amplitude of @xmath82 , which we can be confident is real given the results of the monte - carlo simulations ( ) .
while the above shows that at least part of the line - of - sight velocity signals are intrinsic to the star , the component of the variation that is in phase or anti - phase with the light variations may still be influenced .
in order to get a handle on the importance of possible systematics , we tried different methods to determine the line - of - sight velocities , looked at differences between the velocities inferred for different lines , and tried fitting artificial data generated with the models described in paper ii .
we found that cross - correlation was very susceptible to systematics .
for instance , the line - of - sight velocity amplitudes derived from the cross - correlation discussed in ( a ) are systematically different for different balmer lines , both in phase and in amplitude .
most likely , this reflects the influence of the varying temperature and hence continuum slope , which is not accounted for in the cross - correlation procedure . to get unbiased velocities
, one needs to normalize all spectra , but it is not clear how to do this properly .
we therefore focused on line - profile fitting , of h@xmath54 , h@xmath26 and h@xmath55 ( in the wavelength intervals 46225102 , 42204540 and 40404200 , respectively ) .
to each line , we fitted a sum of a lorentzian and a gaussian profile , with their central wavelengths forced to be the same , and a continuum represented by either a linear or a parabolic function .
we tried fitting to @xmath83 in addition to fitting to @xmath84 directly , the rationale being that since absorption is a multiplicative rather than an additive process , a sum of components can represent @xmath83 better than @xmath84 . for all types of fits ,
the resulting line - of - sight velocity curves were fitted with the combination of eleven cosine waves .
the results were systematically different from those inferred from the cross - correlation velocities , with amplitudes generally being somewhat smaller , and phase differences between light and velocity being further away from zero .
the systematic differences between lines were smaller , but not absent .
no apparent effects are present in the derived phases for the different periodicities , but the amplitudes increase systematically from h@xmath54 to h@xmath26 to h@xmath55 . also , the zero points ( i.e. , systemic velocities ) derived for the different lines are very different .
this is probably unrelated , however , as we find similar differences in systemic velocity for our flux standard 191b2b , which has much narrower lines .
instead , it likely reflects problems with the wavelength calibration , possibly related to the way the light of the arc lamp is projected on the slit ( t. bida , 1997 , private communication ) .
such an effect should not affect the results for the periodic variations in velocity .
this leaves the systematic differences in the amplitudes found for the individual balmer lines .
we can discard the possibility that these are intrinsic to the star based on a simple analytic derivation : while the intrinsic velocity and light variations are not independent , their dependency is only a second - order effect and hence very small ( paper ii ) .
most likely , therefore , the observed systematic effects in amplitude are still related to our way of fitting , perhaps simply because our choice of lorentzian plus gaussian is an inadequate model for the observed line profiles . to get a better handle on the problem , we fitted a number of artificial data sets , produced using the procedures described in paper ii , for different settings of temperature amplitude , horizontal surface velocity amplitude , phase difference between light and velocity , and spherical degree . for all artificial data sets and fitting methods ,
the results for the different lines showed systematic differences of comparable magnitude to those seen in the real data .
a set with zero input temperature variation gave the same velocity amplitude and phase for all three lines , as should be the case , but different systemic velocity , with h@xmath26 being close to zero , and h@xmath54 and h@xmath55 being negative and positive , respectively , by an amount that depended on the fitting method .
the amount was smaller for fitting @xmath83 than for fitting @xmath84 directly , as expected ( @xmath85 , @xmath86 , @xmath87 for h@xmath54 , h@xmath26 , h@xmath55 , respectively ) . for data sets with non - zero temperature variations ,
h@xmath26 again reproduced the input values best , in the sense that the systemic value was still closest to zero , and the velocity phase delay closest to the input value .
h@xmath54 and h@xmath55 tended to lower and higher amplitude and to larger and smaller phase delay , respectively , with the values for both amplitude and phase being more or less symmetric around those for h@xmath26 . from a set with no velocity variations put in ,
we find that for a temperature variation corresponding to 2.8% modulation at 5500 , the apparent velocity shifts in h@xmath54 and h@xmath55 are about @xmath88 in amplitude .
the systematic differences with h@xmath26 for sets including velocity variations are consistent with this . the amplitude should be compared with the observed radial velocity amplitude of @xmath89 and the uncertainty of @xmath90 .
given the results on the artificial data and taking into account that the observed velocity signals have low signal - to - noise , we decided to use the average of the h@xmath54 , h@xmath26 and h@xmath55 velocities , determined by fitting to @xmath83 ( with a linear slope for the continuum ) .
the resulting line - of - sight velocity curve is the one discussed above and shown in c. it has uncertainties dominated by the effects of the star wandering around in the slit . given the simulations with the artificial data , we feel reasonably confident that systematic effects in the fitted amplitudes and phases will be smaller than the quoted uncertainties derived from the least - squares fit ( as listed in ) . for completeness , we note that the result would be similar if one were to use the velocities derived from h@xmath26 only .
we conclude that we have detected line - of - sight velocity variations which are due to intrinsic processes in the white dwarf , and which have amplitudes comparable with those expected from the integration over the visible disk of the surface motion due to the pulsations ( robinson et al .
we have shown that our estimates of the uncertainties are robust and that systematic effects , while problematic , most likely will be within these uncertainties . out of the eleven strongest modes present in the light curve ,
five out of six of the real modes show significant modulation , while none of the combination frequencies does , even though f2+f1 and 2f2 have light amplitudes similar to those of the weaker real modes f4 and f5 .
in this section , we discuss the observational results for the real modes and the combination frequencies in some detail . for each group , we describe the general properties we regard as important , mention what can be understood from geometrical arguments alone , and discuss the results in the context of convective - driving models .
we mentioned already that only the real modes show significant modulation in the line - of - sight velocities .
as measures of the relative amplitudes and phases of velocity and light , we use @xmath91 and @xmath92 ( see ) . here , @xmath70 and @xmath76 are the observed line - of - sight velocity and fractional light amplitudes , @xmath71 and @xmath77 the corresponding phases , and @xmath33 the frequency of the mode . the ratio does not depend on azimuthal order and aspect ( see dziembowski @xcite ; paper ii ) .
the rationale for including @xmath33 is that with this definition depends only on spherical degree for adiabatic pulsations . for adiabatic pulsations ,
one also has @xmath93 , i.e. , velocity maximum arrives a quarter cycle after light maximum . from , we see that the values of both and show no significant variation among the different real modes .
all values of are between 0 and 90 , i.e. , velocity lags light , but by less than a quarter cycle .
the largest value of is found for mode f4 . while the difference with the values for the other modes is not formally significant , we note that a larger value is consistent with this mode having @xmath94 and the others @xmath9 , as inferred in paper ii from the wavelength dependences of the pulsation amplitudes .
compared to an @xmath9 mode , the light variations for an @xmath94 mode are subject to stronger cancellation , while the surface velocities are actually easier to observe ( see dziembowski @xcite ) . in the interior of a white dwarf , pulsation associated with a real mode should be largely adiabatic ; maximum flux coincides with maximum compression and is followed a quarter cycle later by matter moving away at maximum velocity .
this will change as the mode propagates upward and enters the outer regions , where non - adiabatic effects become important . using the observed relations between light and line - of - sight velocity variations
, we can therefore hope to infer properties of the surface layers . in the convective - driving picture ( brickhill @xcite ; goldreich & wu @xcite ) , the main source of change is the surface convection zone .
as this region bottles up heat when it is heated , the flux perturbations at the photosphere of the star are diminished in magnitude and delayed in phase compared to those entering the bottom of the convection zone .
in contrast , the horizontal velocities associated with the pulsation are nearly independent of depth inside the convection zone , because of the strong turbulent viscosity ( brickhill @xcite ; goldreich & wu @xcite ) .
thus , values of should be larger than those expected for adiabatic pulsations , and values of smaller than 90 .
the effects are expected to become more prominent with increasing values of @xmath95 , where @xmath96 is the angular frequency of the mode and @xmath97 is the so - called thermal adjustment time ( goldreich & wu @xcite ; quantity @xmath98 in brickhill @xcite ) , which is a few times the thermal time at the bottom of the convection zone . with increasing mode frequency , therefore , should increase and should deviate more and more from 90 , tending towards 0 . indeed , the observed values of are smaller than 90 and those for larger by a factor 23 than expected if one were to neglect non - adiabatic effects ( e.g. , @xmath99 for mode f1 ) .
unfortunately , the observations do not allow one to verify the predicted trends , but it may be worth noting that , if anything , shows a trend opposite to that predicted , i.e. , less of a deviation from 90 for shorter periods .
we remind the reader that the errors in are not distributed normally ( ) .
it is difficult to make a more quantitative comparison , as this requires evaluating the thermal adjustment time @xmath97 , which depends on details of the convection .
ideally , one would like to use and to estimate @xmath97 and thus constrain the convection properties .
our data only allow a rough determination : a 800-s mode in a star with @xmath100s has @xmath101 and @xmath102 ( wu & goldreich @xcite ) , similar to what is observed for mode f2 .
an independent constraint to @xmath97 can be inferred from the longest - period mode that is present .
this is because to excite a gravity - mode , the convection zone needs to extend deep enough for its thermal response time to be of order of or longer than the mode period , i.e. , @xmath103 ( brickhill @xcite ; goldreich & wu @xcite ) .
the longest - period mode is f2 at 818s ( or f7 at 1106s ; see appendix ) , and thus @xmath104s ( @xmath105s ) , consistent with the above value .
consistent estimates of @xmath97 are also obtained using combination frequencies ( see below ) .
the lack of line - of - sight velocity signals at the frequencies which we associate with combinations of real modes , confirms the notion that combination frequencies do not reflect physical pulsation , but rather mixing of the parent - mode signals by a non - linear transformation occurring in the outer layers of the white dwarf .
we quantify the combination frequencies using an amplitude ratio @xmath106 and a phase difference @xmath107 . here ,
@xmath108 and @xmath109 are the observed amplitude and phase of a combination at frequency @xmath110 of real modes @xmath111 and @xmath112 ( with frequencies @xmath113 and @xmath114 , amplitudes @xmath115 and @xmath116 , and phases @xmath117 and @xmath118 ) ; @xmath119 is the number of possible permutations : @xmath120 for @xmath121 and @xmath122 for @xmath123 ;
@xmath111 and @xmath112 are chosen such that difference frequencies are positive .
we include @xmath119 to ensure that the harmonic of a single mode and a combination of two modes with almost identical properties would lead to the same ( cf .
brickhill @xcite ) .
similar definitions are used for the 3-mode combinations ( see ) ; below , we focus on combinations of two modes .
the observed values of and ( tables [ tab : freqs ] and [ tab : freqs - app ] ) vary significantly from mode to mode , although part of the variation is due to only a few outliers ( especially 2f4 , which we will discuss in more detail below ) .
there appears to be some correlation between the value of and the amplitudes of the parent modes , with increasing for decreasing @xmath124 .
the phase differences cluster around zero ( within @xmath125 ) ; this reflects the sharp maxima and shallow minima in the light curve .
if the combination frequencies indeed result from nonlinear mixing in the upper layers of the star , then independent of a specific theory one expects a combination frequency to have at any point on the surface a flux amplitude proportional to the product of its parent - mode amplitudes at that point .
the surface distributions of the parent modes are described by spherical harmonics .
therefore , the surface distribution of the combination frequency will be proportional to the product of spherical harmonics , which can in turn be described by a linear superposition of spherical harmonics with different (see , e.g. , abramowitz & stegun @xcite ; we adopt their notation ) . in the following discussion
, we will assume all parent modes have only @xmath126 components .
we consider the more general case later .
first , we study the first harmonic of a single parent mode .
the first harmonic of a parent mode with @xmath127 has a surface distribution described by @xmath128 , which can be decomposed into a sum of @xmath129 and @xmath130 distributions .
the first harmonic of a parent mode with @xmath131 has a surface distribution that is the sum of three components , @xmath129 , @xmath130 and @xmath132 .
the numerical factors in front of the @xmath129 terms are the same for both decompositions .
when integrated over the visible hemisphere , the ( 2,0 ) parent mode will suffer stronger cancellation than the ( 1,0 ) parent mode . on the other hand , their harmonics will typically be dominated by @xmath129 components that do not suffer from cancellation .
therefore , will be higher for the harmonic of the ( 2,0 ) mode .
this conclusion is independent of the the angle between the line - of - sight and the pulsation axis . in , the value of for 2f4 is much larger than , e.g. , that for 2f1 and 2f2 .
while we should caution that 2f4 could be blended with other combinations , this is exactly what one would expect from the example given above , given our identifications of f1 and f2 with @xmath9 and f4 with @xmath94 ( paper ii ) .
note that one would not expect , e.g. , f4+f1 to have a value of larger than those for combinations of two @xmath9 modes , such as f2+f1 , f3+f1 and f5+f1 ( as observed ; ) .
this is because the product @xmath133 has no @xmath129 component and therefore suffers cancellation .
looking in detail at for all combinations involving f1 and another @xmath9 mode , e.g. , 2f1 , f2+f1 , we find that the values are not consistent with being the same .
there appears to be a trend of increasing with decreasing parent - mode amplitude .
a similar trend appears to be present for the 3-mode combination frequencies .
this is not expected , as generally the efficiency of mixing should be relatively independent of the amplitudes of the input signals when these signals are weak .
we should caution that above we have been neglecting the possible effect of different values of . for instance , the harmonic of a @xmath9 , @xmath134 parent mode is described by a @xmath135 distribution only ; it would have a low value of . in our short observations
, we can not resolve the rotation - induced frequency splitting among different components that has been seen previously for some of the modes ( kleinman et al . ) .
thus , the power at both the harmonic and its parent mode may result from a number of combinations , which makes our conclusions uncertain .
they can likely be verified , however , using existing , much longer time series from wet .
we now turn from these general considerations , valid for any local nonlinear mixing scheme , to a comparison of our data with numerical results and analytic calculations based on the convective - driving mechanism . as mentioned , in this picture flux perturbations appearing at the photosphere
are diminished and delayed relative to those that enter the convection zone , by an amount that depends on the depth of the convection zone ( and thus @xmath97 ) .
the depth of the convection zone , however , depends sensitively on the photospheric temperature .
the latter varies due to the pulsations , and , therefore , the depth of the convection will vary as well .
the associated variation in reduction and delay leads to a distortion of the pulsation signal at the photosphere , which translates into the power seen at the combination frequencies ( brickhill @xcite ; wu @xcite ) .
brickhill ( @xcite ) found that will be @xmath136 for the first harmonic of an @xmath127 mode and @xmath137 for that of a ( 2,0 ) mode , the exact values depending on the frequencies , @xmath97 and the angle between the line of sight and the pulsation axis .
he found sum frequencies share similar values of with the harmonics .
] , while difference frequencies from the same pair of parent modes have lower values .
these results were confirmed analytically by wu ( @xcite ) , who found that for combinations with relatively high frequency ( @xmath138 ) , should be approximately constant and approximately zero , while towards lower frequencies ( @xmath139 ) , should be lower and should shift towards @xmath140 .
the observed values of and compare , on average , favourably to the expectations .
the typical values of are reproduced with @xmath141s ( wu @xcite ) , which is consistent with the values of @xmath97 inferred above from the longest - period overstable mode and the velocity - light comparison ( ) .
little evidence is found for the expected trends of either or with frequency of the combinations . in part
, this might be due to our inability to resolve components of different .
again , it would be good to verify these conclusions with longer time series , such as those already available from wet .
as mentioned , a trend of with parent - mode amplitude is not expected in any simple non - linear mixing theory ( e.g. , wu @xcite ) .
nevertheless , the numerical calculations presented by brickhill ( @xcite ) show an amplitude dependence of for 3-mode combinations that is similar to that observed ( and with similar numerical values ) . at present
, we do not understand why this dependence arises in the simulations ; unfortunately , brickhill does not expand on it .
we have applied the tool of time - resolved spectrophotometry to the study of pulsating white dwarfs . in this paper
, we showed that at our very high signal - to - noise ratios we can detect the small line - of - sight velocity variations associated with the g - mode pulsations , at amplitudes similar to those expected ( robinson et al.@xcite ) .
significant velocity variation is seen for five out of the six modes we believed were real .
for all five modes , the phase difference between velocity and light is between 0 and 90 , i.e. , velocity maximum occurs later than flux maximum , but by less than the quarter cycle expected for adiabatic pulsations .
the amplitude ratios are similar for the five modes .
the largest value occurs for mode f4 ( at 776s ) , which is consistent with mode f4 having @xmath94 and the others @xmath9 , as was inferred from the wavelength dependence of the pulsation amplitudes in paper ii .
no significant line - of - sight velocity variation is seen for any of the five combination frequencies , even though two of these have light amplitudes comparable to that of the fourth strongest real mode .
this provides independent confirmation that the combination frequencies do not reflect physical pulsation .
rather , they likely result from mixing of mode power by a non - linear transformation in the outer layers of the star .
the phase differences between combination frequencies and parent modes are all close to zero , reflecting the sharp maxima and shallow minima in the light curve .
the amplitude ratios show significant differences , but are generally within a factor two of each other .
the exception is the four times larger value for the harmonic of mode f4 .
this difference is again consistent with the differences in between mode f4 and the others .
models for the interaction between pulsation and convection can broadly reproduce the observed values of and for the real modes , as well as those of and for the combination frequencies ( brickhill @xcite ; wu @xcite ) .
the models are internally consistent , in that the inferred thermal properties of the convection zone are such that modes at the observed periods are expected to be overstable . in detail , there are some problems , with showing a trend with frequency opposite of that expected , and the mixing strength , reflected in , not showing the expected correlation with frequency , but rather a possible dependency on parent - mode amplitude . both these need confirmation ; for the former , further time - resolved spectroscopy is needed , but for the latter , excellent data already exist .
we also hope that our results will lead to further theoretical efforts . in summary ,
time - resolved spectroscopy offers insight in the white dwarf interiors at three levels . by measuring the effects of temperature variation at different wavelengths , the temperature structure of the atmosphere can be calibrated . using the velocity changes and the combination frequencies
, one can constrain the convection zone below the photosphere . and
finally , with the determination of the spherical degree and hence the more secure identifications of periodicities with eigen - modes , one can use asteroseismology to probe the interior with greater confidence .
we gratefully acknowledge help of wayne wack with the observations , useful discussions with tom bida about intricacies of lris and with peter goldreich , scot kleinman and rob robinson about white - dwarf pulsations , as well as sceptical and therefore very useful remarks by an anonymous referee .
m.h.v.k . acknowledges a nasa hubble fellowship while at caltech , and a fellowship of the royal netherlands academy of arts and sciences at utrecht .
this research has made use of the simbad database , operated at cds , strasbourg , france .
the observations were obtained at the w. m. keck observatory , which is operated by the california association for research in astronomy , a scientific partnership among the california institute of technology , the university of california , and the national aeronautics and space administration .
it was made possible by the generous financial support of the w. m. keck foundation .
the reduction was done within the environment of the munich image data analysis system , which is developed and maintained by the european southern observatory .
abramowitz m. & stegun i. a. ( eds ) 1972 , handbook of mathematical functions ( new york : dover ) beland s. , boulade o. , davidge t. , 1988 , cfht info .
, 19 , 16 bergeron p. , wesemael f. , lamontagne r. , fontaine g. , saffer r. a. , allard n. f. , 1995 , apj , 449 , 258 bohlin r. c. , colina l. , finley d. s. , 1995 , aj , 110 , 1316 brickhill a. j. , 1983 , mnras , 204 , 537 brickhill a. j. , 1990 , mnras , 246 , 510 brickhill a. j. , 1991a , mnras , 251 , 673 brickhill a. j. , 1991b , mnras , 252 , 334 brickhill a. j. , 1992a , mnras , 259 , 519 brickhill a. j. , 1992b , mnras , 259 , 529 clemens j. c. , 1993 , baltic astr .
, 2 , 407 clemens j. c. , van kerkwijk m. h. , wu y. , 1998 , mnras , submitted ( paper ii ) dziembowski w. , 1977 , acta astr .
, 27 , 203 goldreich p. , wu y. , 1999a , apj , 511 , 904 goldreich p. , wu y. , 1999b , apj , submitted ( astro - ph/9810038 ) horne k. , 1986 , pasp , 98 , 609 kawaler s. d. , bradley p. a. , 1994 , apj , 427 , 415 kleinman s. j. , 1995 , phd thesis , university of texas at austin kleinman s. j. , et al . , 1994 , apj , 436 , 875 kleinman s. j. , et al .
, 1998 , apj , 495 , 424 koester d. , provencal j. , shipman h. l. , 1997 , a&a , 230 , l57 mccook g. p. , sion e. m. , 1987 , apjs , 65 , 603 mcgraw j. t. , robinson e. l. , 1975 , apj , 200 , l89 nather r. e. , winget d. e. , clemens j. c. , hansen c. j. , hine b. p. , 1990 ,
apj , 361 , 309 oke j. b. , et al .
, 1995 , pasp , 107 , 375 robinson e. l. , kepler s. o. , nather r. e. , 1982 , apj , 259 , 219 robinson e. l. , et al . , 1995 , apj , 438 , 908 shulov o. s. , kopatskaya e. n. , 1974 , astrofizika , 10 , 117 winget d. e. , et al . , 1990 , apj , 357 , 630 winget d. e. , et al . , 1991 , apj , 378 , 326 winget d. e. , et al . , 1994 , apj , 430 , 839 wu y. , 1997 , phd thesis , california institute of technology wu y. , goldreich p. , 1999 , apj , accepted ( astro - ph/9812085 ) young a. t. , 1967 , aj , 72 , 747 zuckerman b. , becklin e. e. , 1987 , nature , 330 , 138
the light curve we obtained is not described well by variations at the eleven frequencies we identified in .
in particular , there are a quite a few more peaks present in the fourier transform at combination frequencies .
we have continued the identification process until no peaks with amplitudes larger than 0.3% remained .
this gave us 29 frequencies .
at this point , the fourier spectrum of the residuals became too noisy to identify with any confidence more peaks in the main region of power ( 14mhz ) .
in the cleaner region at higher frequencies , however , there were still four peaks coincident with frequencies expected for 3-mode combinations .
since these might be of particular interest for modeling the pulsations , we included these in a final fit .
the frequencies , amplitudes and phases for all 33 periodicities are listed in , and indicated in a. we also fitted the velocities to this set of 33 periodicities .
the results are indicated in b ; calculated values of and for the real modes are listed in .
we conclude with a few remarks .
first , we note that possible significant periodicity is also present for 2f3 ( @xmath142 , @xmath143 , @xmath144 , @xmath145 ) and f5+f2 ( @xmath146 , @xmath147 , @xmath148 , @xmath149 ) , while no significant signal is seen at 2f5 ( @xmath150 , @xmath151 ) .
one might have expected power at a number of combination frequencies not present in , such as f3@xmath48f2 and f7@xmath152f1 .
these may indeed be present , but their frequencies are so close to other combinations that it is impossible to include them in the fit . which of the possible combinations one uses is a matter of choice .
we have been biased by theory in preferring those for which the parent modes have larger amplitudes , or , if similar , those for which the fitted value of was closest to 0 .
second , the periodicity f9 , which is very close to f1 in period , is rather uncertain .
the modulation could be fitted equally well with a variable amplitude for f1 .
if f9 is real , it should have @xmath153 given its proximity to f1 , which has @xmath9 .
if it has @xmath94 and a substantial @xmath126 component , one might expect a large harmonic , like for f4 .
this harmonic might in part account for the peak now described by f9+f1 ; if we fit 2f9 instead ( we can not fit both jointly ) , we find @xmath154 , @xmath155 , which corresponds to @xmath156 , similar to that of 2f4 , but @xmath157 , much further removed from zero than for any other combination frequency . finally , while for most combination frequencies we do not find significant velocity variation , there are two exceptions : f4@xmath48f1 at 2900s and f6@xmath48f1 at 1447s .
these two periodicities appear harmonically related to each other , and we believe the associated velocity signals reflect either instrumental or atmospheric effects rather than physical variations .
we briefly wondered about possible orbital modulation , but observations at higher resolution did not confirm these periodicities .
@lrrrrrrrr@ real mode & & & & & & & & + & & & & & & & & + @xmath37 & 613 & & & & & & & @xmath38 & 818 & & & & & & & @xmath39 & 655 & & & & & & & @xmath40 & 777 & & & & & & & @xmath41 & 284 & & & & & & & @xmath42 & 431 & & & & & & & @xmath158 & 1106 & & & & & & & @xmath159 & 920 & & & & & & & @xmath160 & 601 & & & & & & & @xmath161 & 500 & & & & & & & 2-mode & & & & & & & & + combination & & & & & & & & + @xmath162 & 306 & & & & & & & @xmath43 & 350 & & & & & & & @xmath44 & 2444 & & & & & & & @xmath45 & 409 & & & & & & & @xmath46 & 317 & & & & & & & @xmath47 & 9612 & & & & & & & @xmath163 & 364 & & & & & & & @xmath164 & 343 & & & & & & & @xmath165 & 2900 & & & & & & & @xmath166 & 389 & & & & & & & @xmath167 & 194 & & & & & & & @xmath168 & 529 & & & & & & & @xmath169 & 253 & & & & & & & @xmath170 & 1447 & & & & & & & @xmath171 & 368 & & & & & & & @xmath172 & 303 & & & & & & & @xmath173 & 275 & & & & & & & 3-mode & & & & & & & & + combination & & & & & & & & + @xmath174 & 245 & & & & & & & @xmath175 & 228 & & & & & & & @xmath176 & 223 & & & & & & & @xmath177 & 209 & & & & & & & @xmath178 & 220 & & & & & & & @xmath179 & 147 & & & & & & & | we present time - resolved spectrophotometry of the pulsating da white dwarf . as in previous broad - band photometry , the light curve shows the presence of a large number of periodicities .
many of these are combination frequencies , i.e. , periodicities occurring at frequencies that are sums or differences of frequencies of stronger , real modes .
we identify at least six real modes , and at least five combination frequencies .
we measure line - of - sight velocities for our spectra and detect periodic variations at the frequencies of five of the six real modes , with amplitudes of up to @xmath0 .
we argue that these variations reflect the horizontal surface motion associated with the g - mode pulsations .
no velocity signals are detected at any of the combination frequencies , confirming that the flux variations at these frequencies do not reflect physical pulsation , but rather mixing of frequencies due to a non - linear transformation in the outer layers of the star .
we discuss the amplitude ratios and phase differences found for the velocity and light variations , as well as those found for the real modes and their combination frequencies , both in a model - independent way and in the context of models based on the convective - driving mechanism . in a companion paper , we use the wavelength dependence of the amplitudes of the modes to infer their spherical degree .
# 1section [ sec:#1 ] # 1fig .
[ fig:#1 ] # 1table [ tab:#1 ] @xmath1 # 1@xmath2 # 1#2 + @xmath2#2 # 1 # 1 # 1 + # 1 notes .
# 1 # 1#2#2>30 # 1#2@xmath3 # 1#2@xmath4 # 1#2@xmath5 # 1#2#1.#2 # 1#2 # 12#1 # 1#1 # 10to0#1 # 1#2#1@xmath6roman#2 cite#1#2#1@tempswa , # 2 internalcite stars : individual ( ) stars : oscillations white dwarfs |
hadronic one- and two - particle spectra provide for each particle species information about the phase - space distribution @xmath4 of the hadronic emission region @xcite .
once reconstructed from the measured spectra , @xmath4 allows to distinguish between different dynamical scenarios of heavy ion collisions .
it provides an experiment - based starting point for a dynamical back extrapolation into the hot and dense early stage of the collision , where quarks and gluons are expected to be the relevant physical degrees of freedom .
our reconstruction of @xmath4 is based on the hadronic one- and two - particle spectra @xmath5\ , .
\label{2 } \end{aligned}\ ] ] the triple - differential hadronic one - particle spectrum ( [ 1 ] ) tests the momentum - dependence of @xmath4 only .
its azimuthal @xmath6-dependence with respect to the reaction plane @xmath7 is parametrized by the harmonic coefficients @xmath8 @xcite .
information about the space - time structure of @xmath4 can be obtained from the relative momentum dependence of the two - particle correlator @xmath9 , @xmath10 .
this @xmath11-dependence is usually parametrized via the hanbury - brown twiss ( hbt ) radius parameters @xmath12 which depend on the average pair momentum @xmath13 .
depending on the gaussian parametrization adopted , the indices @xmath14 , @xmath15 in ( [ 2 ] ) run either over the cartesian directions _ long _ ( parallel to the beam ) , _ out _ ( parallel to the transverse component @xmath16 ) and _ side _ , or over the corresponding yano - koonin coordinates @xmath17 , @xmath18 and @xmath19 @xcite . [ fig1 ]
a typical data analysis starts from a simple ansatz for the phase space distribution @xmath4 in terms of very few , physically intuitive fit parameters , @xcite : @xmath20 this model e.g. assumes local thermalization at freeze - out with temperature @xmath21 within a space - time region of transverse gaussian width @xmath22 , emission duration @xmath23 , longitudinal extension @xmath24 , where @xmath25}$ ] , and average emission time @xmath26 .
the model allows for dynamical source correlations via the hydrodynamic flow @xmath27-velocity @xmath28 .
we assume a linear transverse flow profile @xmath29 with variable strength @xmath30 , and bjorken scaling of the flow component in the longitudinal direction , @xmath31 , @xmath32 = \eta$ ] .
resonances are produced in thermal abundances with proper spin degeneracy @xmath33 for each particle species @xmath14 .
their contribution to the pion yield is obtained by propagating them along their classical path @xmath34 according to an exponential decay law @xcite , @xmath35 } s_r^{\rm dir}(x , p)\ , , \label{5}\ ] ] where @xmath36 is the integral over the available resonance phase space for isotropic decays .
we include all pion decay channels of @xmath37 , @xmath38 , @xmath39 , @xmath40 , @xmath41 , @xmath42 , @xmath43 , @xmath44 , @xmath45 and @xmath46 with branching ratios larger than 5 percent .
the model parameters @xmath21 , @xmath30 , @xmath22 , @xmath47 , @xmath23 , @xmath26 can then be determined via the following strategy : 1 . _
transverse one - particle spectrum @xmath48 determines blue - shifted temperature @xmath49 : _ + the slope of @xmath48 is essentially given by @xmath50 .
hence , different combinations of @xmath21 and @xmath30 can account for the same data , see fig .
@xmath51 disentangle temperature and transverse flow .
@xmath52 fixes transverse extension @xmath22 : _ + the @xmath53-slope of @xmath52 is proportional to @xmath54 , @xmath55 @xcite .
this slope allows in combination with @xmath48 to specify @xmath21 and @xmath30 , see fig .
the overall size of @xmath52 determines the gaussian width @xmath22 .
we find @xmath3 , @xmath56 mev , @xmath57 fm .
@xmath58 determines @xmath26 , width of rapidity distribution @xmath59 determines @xmath47 : _ + in principle , @xmath58 depends on @xmath26 , @xmath47 and @xmath23 @xcite .
we have fixed @xmath60 by matching the width of the pion rapidity distribution .
the data then favour clearly a lifetime of @xmath61 fm / c ( see fig . [ fig2 ] ) but are not very sensitive to the emission duration @xmath23 @xcite .
the present plots are obtained with @xmath62 fm / c .
@xmath63 discards opaque sources _ : + for the model ( [ 3]-[5 ] ) , the ykp - parameter @xmath63 is mainly sensitive to the temporal aspects of the source .
the large statistical uncertainties for @xmath63 do not allow to constrain the model parameter space further , see fig .
models of opaque sources including an opacity factor in ( [ 4 ] ) are excluded already by the present data @xcite .
the radius of a cold pb nucleus corresponds to @xmath64 fm in the gaussian parametrization ( [ 4 ] ) , i.e. , the experimental data indicate a very large source @xmath65 .
this is dynamically consistent with a collision scenario in which the initially produced pressure gradients result in a significant transverse flow @xmath66 , driving the expansion of the system over a time of 9 fm / c to twice its initial size .
the temperature may have decreased substantially during this expansion ; the data indicate 130 mev at freeze - out .
these conclusions are further supported by the analysis presented by g. roland @xcite , which is based on approximate analytical formulas .
we have extracted the model parameters by comparison to numerical model calculations , following the above strategy .
they are not obtained from a simultaneous fit to all observables , and hence we do not quote errors .
the reaction plane analysis of triple - differential one - particle spectra ( [ 1 ] ) has been discussed extensively in this conference e.g. by a. poskanzer , j .- y .
ollitrault , and s. voloshin . the general strategy for linking this analysis to an azimuthally sensitive particle interferometry
is based on the @xmath67-dependence of the hbt - radii @xcite , where @xmath67 measures the azimuthal angle of @xmath68 relative to the reaction plane , @xmath69 here , the many harmonic coefficients @xmath70 make a direct comparison to experimental data impossible
. however , various relations hold amongst these coefficients , since the leading anisotropy in realistic source models @xmath4 can be quantified by very few parameters only . using the symmetries of the system and assuming that elliptic deformations dominate we find @xcite @xmath71 a violation of eqs .
( [ 8])-([9 ] ) by experiment would indicate strong higher order deformations and rule out many model scenarios . on the basis of eqs .
( [ 7])-([9 ] ) , an azimuthally sensitive parametrization of the two - particle correlator involves only two additional fit parameters , @xmath72 the anisotropy parameter @xmath73 vanishes at mid - rapidity or if the source contains no dynamical correlations .
it characterizes anisotropic dynamics .
the parameter @xmath74 characterizes the elliptic geometry .
the parameters @xmath73 and @xmath74 can be determined from event samples in spite of the uncertainty in the eventwise reconstruction of the reaction plane . for details , see ref .
@xcite .
9 u. heinz , nucl - th/9609029 .
s.a . voloshin and y. zhang , z. phys . c*70 * ( 1996 ) 665 .
p. jones for the na49 coll . ,
quark matter 96 , nucl . phys .
* a610 * ( 1996 ) 188c .
u.a . wiedemann and u. heinz , phys .
rev . c*56 * ( 1997 ) 3265 ; ibidem r610 .
h. appelshuser , na49 phd - thesis .
s. chapman , r. nix and u. heinz , phys .
rev . c*52 * ( 1995 ) 2694 .
wiedemann , p. scotto and u. heinz , phys .
rev . c*53 * ( 1996 ) 918 .
b. tomik and u. heinz , nucl - th/9707001 , and poster at this conference .
g. roland for the na49 coll . ,
these proceedings .
u.a . wiedemann , phys .
c*57 * ( 1998 ) 266 . | the preliminary cern sps na49 pb+pb 158 gev / a one- and two - particle @xmath0-spectra at mid - rapidity are consistent with a source of temperature @xmath1 mev , lifetime @xmath2 fm / c , transverse flow @xmath3 , and a transverse geometric size which is twice as large as the cold pb nucleus . |
the fractional quantum hall ( fqhe ) was discovered@xcite when experimenters were searching for the wigner crystal@xcite ( wc ) .
the expectation was that once the kinetic energy is quenched by forcing all electrons into the lowest landau level ( ll ) , the interaction energy would govern the physics and induce a wc.@xcite however , it turns out that , in a large range of filling factors of the lowest ll , the interaction energy favors the formation of composite fermions ( cfs ) instead , which form a liquid and produce the fqhe.@xcite however , when the interparticle separation is large compared to the size of a localized particle , namely the magnetic length @xmath1 , the wc should be stabilized .
this condition can be achieved in two ways : ( i ) by reducing the magnetic length , which can be attained by going to very high fields ( or low filling factors ) ; and ( ii ) by increasing the interparticle separation , which can be conveniently accomplished by considering systems close to integer fillings , where , to the zeroth order approximation , only the electrons or holes in the topmost partially filled ll are relevant .
it is useful to differentiate between these two kinds of wcs , both because they originate from different interparticle interactions , and because the density of the relevant particles in the latter actually depends on the magnetic field , vanishing at the magnetic field where the filling is an integer .
we will therefore label these crystals type - i " and type - ii " wigner crystals , respectively .
much experimental work has been performed to investigate such crystalline states.@xcite recently , microwave resonance experiments@xcite have extensively probed regions at low fillings and also near integer fillings and revealed resonances in the conductivity that are interpreted as the pinning mode of a wc . interestingly , even though the formation of composite fermions was responsible for inhibiting the _ electron _ wc in a broad range of filling factors , composite fermions themselves can form a wc , which is called a composite fermion wc ( cfwc ) .
composite fermions@xcite are bound states of electrons and quantized vortices .
they experience an effective magnetic field @xmath2 and form landau - like levels called @xmath3 levels ( @xmath3ls ) .
their filling factor @xmath4 is related to the electron filling factor by @xmath5 , where @xmath6 is the number of vortices attached to composite fermions . as for electrons , there exist two kinds of cfwcs : _ type - i cfwc _ : it has been shown theoretically that at very low fillings , the cfwc has lower energy than the electron wc , i.e. the actual ground state is at type - i cfwc.@xcite essentially , electrons capture fewer than the maximum number of vortices available to them , and use the remaining degrees of freedom to form a wc .
the cfwc has been shown to be an excellent approximation of the exact state in numerical diagonalization studies.@xcite these studies predict that composite fermions with a large number of attached vortices can occur in type - i cfwc , and that there should be transitions , as a function of the filling factor , from wc of one flavor of composite fermions to another . _
type - ii cfwc _ :
the cfwc can also occur at cf fillings close to @xmath7 , which correspond to electron fillings close to @xmath8 .
the state at @xmath7 contains an integer number of fully occupied @xmath3ls , and at nearby fillings the system has either a small density of composite fermions in an otherwise empty @xmath3l or a small density of cf holes in an otherwise full @xmath3l .
these will form a type - ii cfwc.@xcite we compute in this paper the static and dynamic properties of the type - ii cfwc .
the primary motivation of our work comes from the remarkable recent experiments by zhu _
et al._@xcite where a resonance was observed in a narrow region around @xmath0 and interpreted as the pinning mode of the type - ii cfwc . a comparison of the results of our calculation with the experimental observations shows a good agreement , thus supporting the formation of a cfwc .
we study the cfwc by making an exact mapping of the problem of composite fermions in a partially filled @xmath3l to that of fermions in the lowest ll interacting via an effective interaction .
various quantities can then be computed using the familiar methods developed in refs .
( the type - ii cfwc was also studied by scarola _
et al._@xcite in a variational study that compared the energies of the wc , bubble crystal , or stripe phases of composite fermions .
that work , however , did not compute the shear modulus of the cfwc or the magnetophonon and magnetoplasmon dispersions , and also did not focus on the phase diagram of the cfwc . )
our main conclusions are as follows .
we estimate the ranges of stability for the cfwc around the filling factors @xmath9 .
the theoretical range of stability around @xmath10 is consistent with the interpretation of microwave resonance experimental results around @xmath11 as a signature of the cfwc.@xcite we predict cfwc also around @xmath12 and @xmath13 , except in much narrower ranges of filling factor .
we present results for the dispersions of the phonons as well as magnetoplasmon of the cfwc . using a model of chitra et .
al.@xcite we argue that the frequency of the pinning resonance of the cfwc is expected to be of similar magnitude as that of the corresponding electron wc , and also determine the magnetic field dependence ; these are also consistent with experiments .
a curious observation is that the hartree fock treatment is much more successful for the cfwc than for the electron wc .
for example , maki and zotos@xcite found , with similar methods as those employed below , that the electron wc in the lowest ll is stable in the range @xmath14 ( and @xmath15 by particle hole symmetry in the lowest ll ) , which is grossly inconsistent with experiments that demonstrate wc only below approximately @xmath16 .
on the other hand , the stability range we find for the cfwc is quite consistent with experiment . while this difference may seem surprising , it is easily understood .
the hartree - fock approach fails for electrons in the lowest ll because they form strongly correlated states of composite fermions , not captured by the hartree fock approach .
in contrast , the interaction between composite fermions is such that the cfwc state is destabilized before composite fermions can form their own strongly correlated fqhe states .
for example , the @xmath17 fqhe state , which is a _
fractional _ qhe of composite fermions@xcite at the composite fermion filling factor of @xmath18 , is outside the cfwc region , which persists , according to our calculation , only up to @xmath19 .
the remainder of the paper is organized as follows .
we describe the mapping from composite fermions in a partially filled @xmath3l to fermions in the lll .
next , we solve the equations of motion for the cfwc , and calculate the shear modulus and the dispersion relations for the magnetophonon and magnetoplasmon collective modes of the this crystal .
finally , we conclude with a discussion of how our results compare with experimental observations .
electrons in the lowest landau level exhibit exotic properties , which result from the formation of composite fermions , bound states of electrons and an even number of topological vortices . at the rational filling factors
@xmath20 the ground state contains precisely @xmath21 filled @xmath3ls of composite fermions . of interest in this paper
are filling factors close to these rational values , when the cf filling factor is given by @xmath22 in what follows below , we assume that the magnetic field is sufficiently strong that the system is fully spin polarized and ll mixing is negligible .
our program proceeds along the following steps : \(i ) we treat the @xmath21 filled @xmath3ls as inert , and formulate the problem in terms of either solely the composite fermions in the @xmath23 @xmath3l ( for @xmath24 ) or the missing composite fermions in the @xmath25 @xmath3l ( for @xmath26 ) . the physics that we wish to investigate lies fully in the dynamics of these composite fermions . these are sometimes called cf quasiparticles and cf quasiholes , but we will refer to them simply as composite fermions below , while remembering that their filling factor is @xmath27 , and that their density vanishes at @xmath28 ( where @xmath29 ) .
we do not include the physics of @xmath3l mixing in our work ; it will modify the form of the inter - cf interaction , but we do not believe that would significantly alter the results .
\(ii ) the two - body coulomb interaction between electrons generates a complex interaction between composite fermions that contains two , three and higher body terms .
we assume that the physics is dominated by the two - body interaction ( determined in the next section ) .
this should be a good approximation at sufficiently small densities of the composite fermions ; because the cf filling is found to be small in the entire range of stability of the cfwc determined below , we believe that the approximation of the neglect of three and higher body interaction terms in the cf hamiltonian does not cause significant corrections .
there is also numerical evidence that this is a good approximation@xcite , but a systematic study of the importance of higher order interactions between particles as a function of system size has not been undertaken.@xcite \(iii ) we define the equivalent problem of fermions at filling factor @xmath27 in the lowest ll interacting via an appropriate effective interaction . the effective interaction used here has been shown to reproduce the desired interaction pseudopotentials with extremely good accuracy .
\(iv ) we calculate the properties of the wc of fermions at @xmath27 in a standard hartree fock approach.@xcite in particular , the instability of the cfwc is signaled by the shear modulus becoming negative .
\(v ) disorder is neglected throughout .
this makes our results unreliable at very small values of @xmath27 , where the distance between the ( relevant ) composite fermions is so large that disorder will likely dominate over interaction , producing a localized or a glassy phase .
however , we expect the neglect of disorder to be a good first approximation for weak disorder and not too large a lattice spacings ; here the wc order ought to persist meaningfully over several lattice spacings . , both for composite fermions in the second @xmath3 level ( labeled quasiparticle " ) and cf quasiholes in the lowest @xmath3 level ( quasihole " ) .
at long distances , both interactions are coulombic , but have more complicated forms at short distance . , scaledwidth=45.0% ] the cf wave functions can be obtained from the electron wave functions by the standard mapping : @xmath30 here @xmath31 is a quantum number labeling the state , @xmath4 is the cf filling factor , and @xmath32 is the lll projection operator .
the interaction between two composite fermions in the @xmath33 @xmath3l is determined by computing the cf pseudopotentials,@xcite i.e. the energies of the pair in definite relative angular momentum states , by analogy to the electron pseudopotentials in the @xmath34 ll .
the pseudopotential for two electrons in relative angular momentum @xmath35 is defined as@xcite : @xmath36 where @xmath37 is the wave function of two electrons in the @xmath34 ll in relative angular momentum @xmath35 .
the cf pseudopotentials are analogously defined as @xmath38 where @xmath39 while for the calculation of of the electron pseudopotentials @xmath40 , it is sufficient to consider only two electrons in the @xmath34 ll , the evaluation of the cf pseudopotentials requires a full many - particle calculation , because the jastrow factor correlates all composite fermions with one another ( including those in lower @xmath3 levels ) .
nonetheless , using the standard methods@xcite , the relevant integrals can be evaluated by the metropolis monte carlo method , and the thermodynamic limits for the cf pseudopotentials can be obtained .
we refer the reader to the literature for further details.@xcite we next define a problem of fermions confined to the lowest ll that have an effective interaction @xmath41 that produces the above pseudopotentials : @xmath42 where the superscript @xmath43 refers to the lowest ll .
there are many possible choices for @xmath44 which produce the same effective interaction.@xcite we find it convenient to use the effective interaction given by lee , scarola , and jain@xcite : @xmath45\ ] ] the asterisk on @xmath46 indicates that the distance is being measured in units of the effective magnetic length @xmath47 .
the last term captures the long distance limit of the interaction between composite fermions : @xmath48 , which is the coulomb interaction between particles of fractional charge @xmath49 at a distance @xmath50.@xcite the number of terms kept in the sum , and the coefficients @xmath51 , are determined to produce a good approximation for the pseudopotentials obtained from the microscopic theory .
we use below the @xmath51 given in ref .
[ ] . in what follows
, we will use @xmath52 as the unit of length and @xmath53 as the unit of energy .
we will omit the asterisk from the lengths for notational convenience .
we will further set @xmath54 and @xmath55 .
in certain places of relevance , the units will be explicitly restored .
now , we consider the dynamics of fermions in the lll interacting with an effective interaction . since we are dealing with fermions in the lll , they all have the same kinetic energy .
therefore , the ground state of the system is decided entirely by interactions between the fermions .
since we will be working at low filling factors , we make the ansatz that our system is described by a crystal state@xcite , for which we choose the hexagonal ( or triangular ) lattice symmetry , which it is known to give the lowest energy for the 2d classical wc.@xcite the localized single particle states are wave packets : @xmath56 centered at hexagonal lattice sites @xmath57 where @xmath58 are integers and @xmath59 the lattice constant .
the maki - zotos@xcite wave function is obtained by placing a gaussian wave packet at each lattice site and then antisymmetrizing : @xmath60 the problem can be recast as a semiclassical crystal dynamics problem by calculating the expectation value of the effective interaction : @xmath61 where the sums on the right are infinite sums over hexagonal lattice sites , @xmath62 , and @xmath63 is given by : @xmath64 the derivation of these expressions closely follows the work of maki and zotos@xcite , and is given in appendix a. as stated above , we neglect @xmath65 body terms in eq .
( [ vmz ] ) .
the individual two body terms are interpreted as semiclassical energy of particles located at the lattice sites of a hexagonal lattice .
we have calculated the energy for both composite fermions around each of the filling factors @xmath9 using the effective real space interactions discussed above . with the particular form of the effective interaction used (
( [ effint]))@xcite , the integrals in eq .
( [ v2 ] ) can be evaluated analytically using mathematica , but the resulting expression is too long to reproduce here . in fig .
[ effpot ] we plot the semiclassical interaction given by eq .
( [ v2 ] ) using the values of @xmath51 for composite fermions around @xmath0 given by [ ] . at short range ,
these interactions have a complicated form as a function of the distance , and can even be attractive , but at long distance the interaction is always repulsive .
the attractive portion of the interaction between the composite fermions might seem to suggest that the crystal is unstable to pairing , but the issue is somewhat subtle because the gain in energy due to pairing is accompanied by a competing enhancement in the interaction energy between the pairs .
it has been shown by an explicit calculation reported in ref .
[ ] that for sufficiently small values of @xmath27 a wc state is energetically preferred over the paired liquid state , although the formation of bubble crystals , where each lattice site supports more than one composite fermion , can result from an attractive interaction in some filling factor ranges .
we will not consider the possibility of bubble crystals below , and assume that there is a single cf at each lattice site .
next we calculate the lattice dynamics for a crystal of composite fermions in the harmonic approximation.@xcite we interpret the sum @xmath66 as the zero - point energy of the cfwc with composite fermions located at the lattice sites of a hexagonal lattice .
we perturb the particle s equilibrium location @xmath67 by a small , time dependent quantity @xmath68 : @xmath69 we expand @xmath70 to quadratic order in @xmath71 and write the equation of motion for the @xmath72 particle : @xmath73\ ] ] @xmath74\ ] ] in writing these equations , we have assumed that the dynamics of composite fermions@xcite corresponds to particles of mass @xmath75 in an effective magnetic field @xmath76 . the cf mass @xmath75 is unrelated to the electron mass.@xcite the factor of @xmath77 results from expressing the lengths in units of @xmath52 , while recognizing that the argument of @xmath78 is already in units of @xmath52 .
this is consistent with the expectation that the magnetic field renormalization from @xmath79 to @xmath2 upon the formation of composite fermions is associated with an analogous electric field renormalization.@xcite assuming a solution of the form @xmath80 the equations of motion in the @xmath81 and @xmath82 directions take the form : @xmath83u_{{\mbox{\boldmath$\scriptstyle{k}$}}}^y+\omega^*_c\phi_{xx}\!({\mbox{\boldmath$k$}})u_{{\mbox{\boldmath$\scriptstyle{k}$}}}^x\ ] ] @xmath84u_{{\mbox{\boldmath$\scriptstyle{k}$}}}^x+\omega^*_c\phi_{yy}\!({\mbox{\boldmath$k$}})u_{{\mbox{\boldmath$\scriptstyle{k}$}}}^y\ ] ] where @xmath85 and @xmath86 is given by @xmath87 here , @xmath31 and @xmath88 denote @xmath81 and @xmath82 components .
equations ( [ xeq ] ) and ( [ yeq ] ) can be diagonalized to obtain the dispersion @xmath89 : @xmath90\right)\pm\sqrt{\frac{1}{4}(\omega^{*2}_c+\omega^*_c[\phi_{xx}({\mbox{\boldmath$k$}})+\phi_{yy}({\mbox{\boldmath$k$}})])^2+\omega^{*2}_c[\phi_{xy}^2({\mbox{\boldmath$k$}})-\phi_{xx}({\mbox{\boldmath$k$}})\phi_{yy}({\mbox{\boldmath$k$}})]}\ ] ] to ascertain the regions of filling factor where the cf crystal is stable , we need to examine the phonon dispersion in the long wavelength limit , @xmath91 . in the limit of small @xmath92
we can write @xmath86 as@xcite : @xmath93 where @xmath94 is the shear modulus and @xmath95 the coefficient in the longitudinal mode .
both have units of energy . in two dimensions
, it can be shown that the dynamical matrix of a two dimensional hexagonal crystal with central forces characterized by two parameters , here conveniently chosen to be @xmath94 and @xmath95.@xcite as shown in ref .
, the classical value of @xmath94 for particles with charge @xmath96 in a magnetic field @xmath2 and filling factor @xmath27 is given by @xmath97 , which , when converted into our energy units of @xmath53 , reduces to @xmath98 .
the parameter @xmath99 is given by @xmath100.@xcite non - classical properties of the cfwc can be captured by measuring the variation of @xmath94 and @xmath95 with respect to their classical counterparts in the region where the crystal is stable . substituting into eq .
( [ disp ] ) produces : @xmath101 where we have kept terms up to quadratic order in @xmath92 . expanding in powers of @xmath92 , we get : @xmath102 @xmath103 the @xmath104 mode is called the magnetoplasmon mode and the @xmath105 mode is called the magnetophonon mode . for negative values of @xmath94 the magnetophonon frequency becomes imaginary in the long wave length limit , indicating an instability of the cf crystal .
( top ) and the cf filling factor @xmath4 ( bottom ) .
, scaledwidth=45.0% ]
we give in this section the results of our calculation , and , where possible , compare them to experiments .
to calculate the shear modulus , we evaluate the dynamical matrix @xmath86 of eq .
( [ phiab ] ) numerically for several different values of @xmath92 and fit the result to the small @xmath92 limit form of eq .
( [ lphiab ] ) .
further details are given in appendix b. in fig .
[ c13qp ] we show our results for the cfwc moduli around @xmath106 . in this figure
, we plot @xmath94 and @xmath95 normalized with respect to the classical shear modulus of particles with fractional charge as a function of @xmath27 .
notice that as @xmath27 approaches zero , all of the crystal moduli become equal to the classical crystal moduli .
we find that at sufficiently low filling factors ( approximately @xmath107 ) , the cf moduli are indistinguishable from their classical crystal counterparts .
the non - classical behavior of the cfwc becomes manifest only at somewhat larger values of @xmath27 and , indeed , makes a significant correction to the phase boundary marked by the position where the shear modulus changes its sign .
the cfwc becomes unstable when @xmath94 becomes negative .
we depict the stability regions around the filling factors @xmath108 in fig .
[ stabreg ] , which is one of the principal results of our work .
the predicted range near @xmath0 is nicely consistent with the experiment of zhu _
et al._@xcite , who find a range @xmath109 @xmath110 , determined by where a microwave resonance is resolved .
this differs by an order magnitude from the range @xmath111 for the electron crystal.@xcite in fact , as also noted by zhu _
et al . _ , when measured in terms of the cf filling , the range near 1/3 is given by @xmath112 , which is more similar to the observed stability region of the @xmath113 wc .
overall , our calculations provide strong support to the identification of the microwave resonance experiments in terms of the cfwc .
we also predict regions of stability for the cfwc around @xmath114 , which , however , are much narrower and may be harder to observe , especially considering that at very small @xmath27 the disorder may dominate producing a localized glassy phase .
we note that we mark the region of stability as the filling @xmath27 where the shear modulus _ first _ becomes negative as the @xmath27 is increased from zero . according to our calculations the shear modulus becomes positive again at some larger fillings , to which we do not assign any physical significance
we believe that those regions are outside the regime where our approximations are valid .
it was noted in previous calculations@xcite that composite fermions can possibly form bubble crystal and stripe phases at larger values of @xmath27 . in fig .
[ 13qp ] , we plot the dispersions of the magnetoplasmon and magnetophonon collective modes for the cfwc around the irreducible element of the first brillouin zone ( fbz ) around @xmath0 . for small wave vectors , the dispersion has the form @xmath115 , which is identical to the small @xmath92 behavior of the classical wc dispersion .
the dispersion relations for the magnetophonons and magnetoplasmons for the cfwc around @xmath12 and @xmath13 , shown in figs .
[ 25qp ] and [ 37qp ] , have nearly the same shape and appear to be scaled versions of the dispersion relations shown for the crystals around @xmath0 .
the magnetophonon and magnetoplasmon dispersions of the wc obtained above are valid for an ideal system with no disorder . at small wave vectors ,
a gap opens due to disorder , resulting in a pinning mode that has been studied extensively and is the topic of the subsequent section . as noted in ref .
, it is in principle possible to obtain information about the small @xmath116 dispersion by comparing the pinning mode resonances measured from nearly identical samples that have different metal film coplanar waveguide ( cpw ) slot widths @xmath117 ( see any of [ ] for experimental details . ) measurement of the dispersion at large @xmath116 is complicated by the fact that light ideally couples only to excitations with very small wave vectors .
a possible method for accessing larger momenta through light scattering would be to use gratings or to exploit piezoelectric coupling to surface phonons to define a wave length , which has proven useful in the study of collective mode dispersion of the liquid states of composite fermions.@xcite the extremal points of the dispersion may also provide a signature in inelastic light scattering , in the presence of disorder , because of the singularity in the density of states at the corresponding energies.@xcite a number of theoretical studies have considered the electromagnetic response of a pinned winger solid in a magnetic field,@xcite and find that the cyclotron resonance is shifted by an amount related to the pinning frequency , and at the same time also broadened .
the actual frequencies of the pinning mode depend on the details of the potential and are difficult to predict in a quantitative manner .
we use below the model of chitra _
et al._@xcite , where they use a gaussian variational method to study the elastic hamiltonian of a wigner crystal to deduce its properties in the presence of disorder .
we first address the relation between the pinning mode frequency of the cfwc at an effective filling @xmath4 to that of the electron wc at the same filling ; we believe that such a relationship might be more stable against the nature of the disorder than the actual pinning frequency . according to chitra et .
model@xcite for the pinning frequencies of the wc , different disorder length scales produce significantly different @xmath79 dependence for the the pinning frequencies .
their approximate expression for pinning mode frequencies is given by : @xmath118 where @xmath119 , @xmath120 is the classical shear modulus for the electron wc in units of energy density , @xmath121 , @xmath59 is the lattice spacing , @xmath122 is a measure of the electrostatic coupling of the particles in the lattice to disorder ( and is proportional to the square of the charge of the particle in the lattice ) , @xmath123max[@xmath124 , @xmath125 is the mass density , @xmath126 is the particle density and @xmath127 is the disorder correlation length .
the pinning frequencies have a different b - field dependence depending on which of @xmath127 or @xmath1 is larger . putting everything together , we find @xmath128 noting that @xmath59 , @xmath129 , @xmath130 , and @xmath131 are identical for electrons near @xmath132 and composite fermions near @xmath7 ( the quantity @xmath130 is the same because it is independent of the mass , and @xmath131 is the same because the magnetic length and the disorder correlation lengths are unchanged . ) , the ratio of the pinning mode frequency of composite fermions and electrons , @xmath133 and @xmath134 , respectively , is given by @xmath135 furthermore , @xmath136 and @xmath137 are approximately equal in the classical limit , because both @xmath138 and @xmath122 are proportional to @xmath139 whereas @xmath140 and @xmath141 are proportional to @xmath142 . therefore , if we take the coupling to the disorder potential to be the same , the pinning frequencies of the electron wc at @xmath143 and the cfwc at @xmath4 are equal in the classical limit , and likely of the same order in the region where the system is not too far from the classical region . while clearly an approximation , this result is consistent with the experimental finding of zhu _ et al._,@xcite and
, we believe , explains why the pinning frequencies of the electron and cfwcs have similar magnitude in a range of parameters .
we next come to the @xmath79-field dependence of the pinning frequency .
there is an interesting difference in @xmath79-field dependence of a type - i and type - ii wc . for a type - i crystal ,
the @xmath79-field dependence of @xmath144 depends on the relative magnitudes of @xmath1 and @xmath127 .
the choice @xmath145 gives @xmath146 , whereas @xmath147 produces @xmath148 .
this difference in behavior is easy to identify experimentally and the @xmath79-field dependence of @xmath144 thus provides evidence for the relevant length scale for disorder .
the @xmath79-field dependence of the type - ii wc is more complicated , because the @xmath79 dependence of the cf quasiparticle / quasihole particle density @xmath149 of the type - ii wc must be taken into account when examining the @xmath79-field dependence of @xmath144 .
this leads to a simplification for the @xmath79 dependence of the pinning frequency of the cfwc , because the density variation with @xmath79 dominates .
let us consider , for specificity , filling factors in the vicinity of @xmath106 , where @xmath150 and @xmath151 .
as we move slightly away from the rational fillings , we get : @xmath152 where @xmath153 ( which is equivalent to the relation @xmath154 ) . for the type - i
wc we have ( using the classical shear modulus ) @xmath155 , where @xmath129 is independent of @xmath79 and fixed once the sample is created .
for the type - ii crystal , when @xmath156 , we have @xmath157 .
the dependence of @xmath144 on @xmath158 dominates the @xmath79-field dependence of the pinning mode , regardless of the choice for the disorder length .
this leads to the prediction that pinning frequency of the cfwc always decreases as one moves away from the rational fillings , except possibly in the vicinity of the melting transition where the shear modulus becomes small .
this is consistent with the experimental findings@xcite .
similarly , the @xmath79-field dependence of the pinning modes of the type - ii crystals observed around integer fillings will also be dominated by @xmath158 and the frequency of the pinning mode will decrease as one moves away from integer fillings , again consistent with the experimental observations@xcite .
the @xmath79-field dependence of the pinning mode thus does not shed qualitative light on the nature of disorder for type - ii wcs .
[ edcfddimtable ] .the peak melting temperatures for cfwc below and above @xmath159 , 2 , and 3 , which correspond to electrons filling factors in the vicinity of @xmath0 , 2/5 , and 3/7 .
we assume a density of @xmath160 @xmath161 and parameters appropriate for gaas . [ cols="^,^",options="header " , ] another question of interest is the melting temperature of the cfwc .
as for two dimensional crystals , it is natural to expect that the melting is described by the kosertlitz - thouless mechanism@xcite , with melting temperature given by : @xmath162 in fig .
[ mt ] we plot the melting temperature of the cfwcs near @xmath163 and @xmath164 in units of @xmath165}\;\text{k}\approx 3.32\times 10^{-4}\sqrt{\rho/\nu}$ ] k for gaas - algaas heterostructures , with magnetic field quoted in tesla , @xmath129 in @xmath161 , and the dielectric function taken to be @xmath166 . for a typical density of @xmath167 @xmath161 at the fillings @xmath106 , @xmath168 corresponds to the temperatures @xmath169 , respectively .
we also include as reference a plot of the melting temperature of the wc of point particles with charge @xmath96 . for illustration ,
we give in table i the peak melting temperatures along with the filling factors where they occur . as a function of the partial filling factor @xmath170 where @xmath4 is the cf filling factor .
the melting temperature is quoted in units of @xmath168 ( with @xmath171 ) .
the melting temperature of point particles of charge @xmath96 is also included for reference.,title="fig:",scaledwidth=45.0% ] as a function of the partial filling factor @xmath170 where @xmath4 is the cf filling factor .
the melting temperature is quoted in units of @xmath168 ( with @xmath171 ) .
the melting temperature of point particles of charge @xmath96 is also included for reference.,title="fig:",scaledwidth=45.0% ] as a function of the partial filling factor @xmath170 where @xmath4 is the cf filling factor .
the melting temperature is quoted in units of @xmath168 ( with @xmath171 ) .
the melting temperature of point particles of charge @xmath96 is also included for reference.,title="fig:",scaledwidth=45.0% ] several interesting features are worth noting .
the melting temperature at 1/3 is in the same ball park as the experimental temperature of 50 mk where data in zhu _ _ e__t .
al . was collected .
however , taken at face value , our results would imply a cfwc at the low filling factor side of 1/3 but not at filling factors higher than 1/3 .
the reason for the discrepancy is unclear at this stage , and may have to do either with some simplifying assumptions of the model ( such as neglect or disorder , finite width corrections , or @xmath3 level mixing ) , or with slight intrinsic inaccuracies in the estimation of the effective cf interaction .
we also note that the temperature estimated here is when infinitely long range crystalline order is lost ; short range crystalline order will possibly persist to higher temperatures .
the cfwc melting temperature decreases as we go to 2/5 and 3/7 , as expected , making their observation even more challenging ; the reason for anomalously high @xmath172 for the cf - hole wc at 3/7 is not understood , although it must have to do with the form of the effective interaction . an interesting feature is a significant asymmetry between the behavior of cfwcs above and below @xmath173 integer , as expected from rather general considerations , given that the interaction between cf particles is different from that between cf holes .
( similar asummetry is to be expected for electron wcs above and below integer fillings . )
the experimental results appear to suggest a smaller asymmetry than theory ; we believe that disorder can possibly wash out the asymmetry .
also , note that at low fillings , the melting temperature for the type - ii wc near @xmath132 is @xmath174 larger than the corresponding cfwc near @xmath7 ( since @xmath175 . )
the significantly larger melting temperature of the type - ii wc near @xmath132 explains the relative ease with which its pinning mode resonance was detected as compared to the resonance near @xmath0 .
we are grateful to the national science foundation by partial support under grant no .
dmr-1005536 , and to the penn state high - performance computing cluster for computer time .
as indicated in the text , our calculations are based on the maki - zotos crystal wave function constructed by placing a single particle wave packet at each lattice site @xmath176 of a hexagonal ( or triangular ) lattice and antisymmetrizing the product . in this section
we outline the derivation of the effective interaction for the cfwc , given in eq .
( [ v2 ] ) .
it is not possible to perform this calculation exactly , and one must resort to approximations .
we restrict ourselves here to including only the two - body direct and exchange terms since these make the greatest contribution to the energy.@xcite the expression for the denominator of eq .
( [ vmz ] ) is already given to us by [ ] : @xmath177 where @xmath178 and @xmath62 .
following appendix a of [ ] : @xmath179 where @xmath180}- \nonumber\\ & & e^{-\frac{1}{4 } \left[({\mbox{\boldmath$\scriptstyle{r}$}}_1-{\mbox{\boldmath$\scriptstyle{r}$}}_i)^2+({\mbox{\boldmath$\scriptstyle{r}$}}_2-{\mbox{\boldmath$\scriptstyle{r}$}}_j)^2+({\mbox{\boldmath$\scriptstyle{r}$}}_1-{\mbox{\boldmath$\scriptstyle{r}$}}_j)^2+({\mbox{\boldmath$\scriptstyle{r}$}}_2-{\mbox{\boldmath$\scriptstyle{r}$}}_i)^2 + 2i\left[(x_1-x_2)(y_i - y_j)-(y_1-y_2)(x_i - x_j)\right]\right]}\big{)}\end{aligned}\ ] ] transforming to the center of mass and relative coordinates , the first ( direct ) term is given by : @xmath181 and the second ( exchange ) term is given by : @xmath182 here , @xmath183 and @xmath184 are the bessel function and the modified bessel function of the first kind .
the nearest neighbor term @xmath185 is of the order @xmath186 for @xmath187 . because we are not concerned with properties of the system when the filling factor is much larger than this , the sum @xmath188 is negligible for our purposes . substituting our results for @xmath189 and @xmath190 into eq .
( [ vmz ] ) gives : @xmath191}{1-\frac{1}{2}\sum_{i\neq j}s_{ij } } \cong \sum_{i\neq j}\left[\frac{i_{1,ij}-i_{2,ij}}{1-s_{ij}}\right]\ ] ] @xmath192 the substitution @xmath193 for @xmath41 reproduces the maki - zotos result . for the effective interaction between the composite fermions ,
the integrals can be performed analytically , but the expressions are quite cumbersome and are most conveniently dealt with in a computer program such as mathematica .
in this section we describe our method for evaluating @xmath86 , eq .
( [ phiab ] ) .
we divide @xmath86 into two parts , the @xmath92 independent part and the @xmath92 dependent part : @xmath194 it is necessary to find a way to compute these sums in a manner so that the error can be estimated in a reliable manner , which is especially important for small @xmath92 , where @xmath195 can be quite small . since both sums are absolutely convergent , we are free to rearrange the terms to produce best convergence . for specificity , we describe our method for evaluating @xmath196 ; extending this discussion to @xmath197 and @xmath198 is straightforward .
we begin with the @xmath92 dependent term in eq .
( [ phiab2 ] ) .
our goal is to utilize the periodicity of the @xmath199 terms to produce a rapidly convergent alternating series .
because of inversion symmetry , we have @xmath200 where the sum is over the lattice sites @xmath201 . using @xmath202 ( with @xmath52 as the unit of length ) , we rewrite this as : @xmath203 where @xmath58 are integers , @xmath204 and @xmath205 .
we wish to rearrange the sum on the right hand side of eq .
[ kdepsum ] into an alternating series so that it converges more quickly . to this end
, we divide the integration region into successive light and dark blocks as shown in fig .
[ kxky ] , where each point represents a lattice vector @xmath206 .
the regions are chosen in a manner that @xmath207 at any point in the light ( dark ) regions gives a positive ( negative ) contribution .
individual positive or negative terms in the alternating series are constructed by summing over all light or dark blocks along the diagonal " lines shown in fig .
we then apply the euler transformation@xcite to the resulting alternating series , which produces a rapidly converging sum . ) .
the points on the graph mark the locations @xmath208 , where @xmath58 are positive integers that occur in the expression eq .
( [ r ] ) .
the quantities @xmath209 are defined in appendix [ dm ] .
, scaledwidth=45.0% ] we organize the non @xmath92-dependent sum in a different manner .
we divide the entire hexagonal lattice into successive annuli ( labeled by the index @xmath21 ) as in fig .
[ nkxky ] , where , beyond certain distance , the inner radius of each successive annulus is twice that of the previous annulus ( e.g. the radii of 5 , 10 , 20 in fig .
[ nkxky ] ) .
the partial sum over the points in the @xmath21th annulus is labeled @xmath210 . the behavior @xmath211 at long distances ensures that the sum @xmath210 decreases by approximately a factor of two for each successive annulus for large @xmath21 .
defining @xmath212 , with @xmath213 , the sum can be expressed exactly as : @xmath214 because we do not know the proper @xmath215 _ a priori _ , we resort to approximating the remaining sum as a power series by setting @xmath216 .
the sum is then given by : @xmath217 to estimate the error in this approximation , we define @xmath218 so that @xmath219
. with some algebra , and assuming @xmath220 , this inequality becomes : @xmath221 which establishes the error to be bounded by @xmath222 in our numerical calculations , @xmath215 becomes rapidly smaller as we increase @xmath21 , allowing us to calculate the sum to a very high degree of accuracy .
we implement the sum using the epsilon algorithm given in ref .
[ ] .
han zhu , yong p. chen , p. jiang , l. w. engel , d. c. tsui , l. n. pfeiffer , and k. w. west , phys .
. lett . * 105 * , 126803 ( 2010 ) .
y. p. chen , r.m .
lewis , l.w .
engel , d.c .
tsui , p.d . ye , l.n .
pfeiffer , k.w .
west , phys .
* 91 * , 016801 ( 2003 ) .
a. pinczuk , b. s. dennis , l. n. pfeiffer , k. west , phys .
* 70 * , 3983 ( 1993 ) ; m. kang , a. pinczuk , b. s. dennis , l. n. pfeiffer , k. w. west , _ ibid .
_ * 86 * , 2637 ( 2001 ) ; t. d. rhone _ et al .
_ , _ ibid .
_ * 106 * , 196805 ( 2011 ) ; s.m .
girvin , a.h .
macdonald , and p.m. platzman , phys .
* 54 * , 581 ( 1985 ) ; v.w .
scarola , k. park and j.k .
jain , phys .
b * 61 * , 13064 ( 2000 ) . | the wigner crystal of composite fermions is a strongly correlated state of complex emergent particles , and therefore its unambiguous detection would be of significant importance .
recent observation of optical resonances in the vicinity of filling factor @xmath0 has been interpreted as evidence for a pinned wigner crystal of composite fermions [ zhu _ et al .
_ , phys .
rev .
lett . * 105 * , 126803 ( 2010 ) ] .
we evaluate in a microscopic theory the shear modulus and the magnetophonon and magnetoplasmon dispersions of the composite fermion wigner crystal in the vicinity of filling factors 1/3 , 2/5 , and 3/7 .
we determine the region of stability of the crystal phase , and also relate the frequency of its pinning mode to that of the corresponding electron crystal near integer fillings .
these results are in good semiquantitative agreement with experiment , and therefore support the identification of the optical resonance as the pinning mode of the composite fermions wigner crystal .
our calculations also bring out certain puzzling features , such as a relatively small melting temperature for the composite fermion wigner crystal , and also suggest a higher asymmetry between wigner crystals of composite fermion particles and holes than that observed experimentally . |
the wrapped fukaya category @xmath4 of a symplectic manifold @xmath0 with convex contact boundary , introduced in @xcite , is of increasing importance in both symplectic topology ( topology of stein manifolds @xcite , the nearby lagrangian conjecture @xcite ) and homological mirror symmetry @xcite .
the wrapped category is a version of the fukaya category which incorporates both closed lagrangian submanifolds and non - compact lagrangians modelled on a legendrian cone at infinity , and which therefore takes into account the dynamics of the reeb flow at the contact boundary . incorporating non - compact lagrangians
typically leads to a substantial increase in complexity ; the wrapped floer cohomology of a cotangent fibre is isomorphic to the homology of the based loop space @xcite , in particular it is of infinite rank . despite its increasing prominence , there are surprisingly few cases in which the wrapped category has been computed .
essentially complete descriptions are known for cotangent bundles @xcite and for punctured spheres @xcite , whilst we have partial information for certain stein manifolds ( given as lefschetz fibrations @xcite or plumbings @xcite ) , and some general structural information concerning the behaviour of the category under weinstein surgery @xcite .
following abouzaid , we study the open - closed string map @xmath5 from hochschild homology to symplectic cohomology .
this is compatible with the decomposition of the wrapped category into subcategories indexed by the spectrum @xmath6 of quantum multiplication by the first chern class , @xmath7 ( we work throughout with coefficients in the novikov field @xmath8 defined in [ subsection novikov ring ] ) .
we prove two main results .
[ theorem introduction 2 ] let @xmath9 be a full subcategory . if @xmath10 hits an invertible element in @xmath11 , then @xmath12 split - generates @xmath13 .
the theorem subsumes two statements .
first , we extend abouzaid s generation criterion @xcite from the exact to the monotone case : as discussed further below in section [ subsection linear vs quadratic hams ] , this is a fairly substantial undertaking .
second , we prove that @xmath1 is a symplectic cohomology module map , so hitting any invertible in a given eigensummand is as good as hitting the identity .
the module structure was observed independently , in the exact case , by ganatra @xcite , and the underlying @xmath14-module structure is used in forthcoming work @xcite on generation in the closed case .
our second main result studies a basic compatibility of @xmath1-maps with the maps induced by functors of fukaya categories arising from lagrangian correspondences .
consider a monotone symplectic manifold @xmath15 , a convex monotone symplectic manifold @xmath0 , and a closed monotone lagrangian correspondence @xmath16 .
composing monotone orientable lagrangians @xmath17 with the correspondence @xmath18 gives rise to a set @xmath19 . to ensure that @xmath20 is a monotone orientable lagrangian
, one needs to ensure that the composition @xmath21 is transverse and embedded , and that @xmath20 is monotone .
call @xmath18 _ very good _ if this is the case for all @xmath22 .
we will discuss the condition that @xmath18 is `` very good '' in section [ subsection phi lands in fuk e ] . to achieve this a priori condition for all possible @xmath22 in such a general context
, one would need to impose a strong condition on @xmath18 .
for instance , it holds if @xmath18 is the total space of a fibre bundle over @xmath15 , the projection @xmath23 is injective , and @xmath24 has torsion kernel .
however , in practice , for the purposes of proving a generation result ( such as via corollary [ corollary introduction 1 ] below ) it suffices that @xmath18 is `` very good '' just with respect to a selected collection of lagrangians @xmath17 whose compositions with @xmath18 one hopes will split - generate @xmath4 ( or a summand thereof ) .
indeed , one applies our machinery to the subcategory generated by the selected @xmath22 and the subcategory generated by their compositions @xmath25 .
if one can deduce that @xmath1 restricted to the second subcategory hits an invertible element , then by theorem [ theorem introduction 2 ] those compositions @xmath25 actually split - generate the whole category .
let @xmath26 be the open - closed map from the ordinary fukaya category of @xmath15 to small quantum cohomology .
[ theorem introduction 1 ] for monotone @xmath15 , monotone convex @xmath0 and very good @xmath16 as above , and for @xmath27 , there is a commutative diagram @xmath28 \ar@{->}_{\mathrm{oc}_b}[d ] & \mathrm{hh}_*({\mathcal{w}(e ) } ) \ar@{->}^{\mathrm{oc}_e}[d ] \\
qh^*(b ) \ar@{->}[r]^-{q_{\gamma,\gamma } } & sh^*(e ) . } \ ] ] the vertical maps are @xmath14 respectively @xmath29-module maps .
the horizontal arrows in the diagram are constructed using the quilt theory of wehrheim - woodward @xcite .
there are corresponding diagrams on the subcategories for particular @xmath30-eigenvalues .
[ corollary introduction 1 ] if for some full subcategory @xmath31 the map @xmath32 hits an @xmath33-invertible element , then the lifts of lagrangians @xmath34 to @xmath0 via @xmath18 split - generate @xmath4 , and for any monotone @xmath35 legendrian at infinity , the wrapped floer cohomology @xmath36 has finite total rank .
the possibility of incorporating the bulk deformation class @xmath27 is essential in applications .
for instance , if @xmath37 is the total space of a negative line bundle over @xmath15 , the undeformed composite map @xmath38 typically vanishes identically for degree reasons , whilst the deformed map can still be non - trivial .
the diagram allows further deformations ( for instance bulk deformations of @xmath39 and the corresponding @xmath40-map ) , although for simplicity we do not state the results in the more general case . en route to proving theorem [ theorem introduction 1 ]
, we also construct for general monotone convex @xmath0 the commutative diagram @xmath41 \ar@{->}_{\mathrm{oc}_e}[d ] & \mathrm{hh}_*({\mathcal{w}(e ) } ) \ar@{->}^{\mathrm{oc}_e}[d ] \\
qh^*(e ) \ar@{->}[r]^-{c^ * } & sh^*(e ) , } \ ] ] which we call _ acceleration diagram _ , which relates the compact fukaya category @xmath42 with the wrapped category .
the analogue of theorem [ theorem introduction 2 ] also holds for the compact category ( namely : if @xmath43 is a full subcategory for which @xmath10 hits an invertible element in @xmath44 , then @xmath12 split - generates @xmath45 ) .
the following is an illustrative application of the acceleration diagram and the generation theorem for the compact category .
let @xmath0 be a convex toric fano variety .
let @xmath46 be a critical point of the superpotential with non - zero critical value @xmath47 .
if the generalized eigensummand @xmath48 is @xmath49-dimensional then the lagrangian torus @xmath50 corresponding to @xmath51 split - generates @xmath52 and @xmath53 , in particular @xmath53 is cohomologically finite .
there are two known constructions of the wrapped category of a stein manifold , which take advantage of hamiltonian functions which have either linear slope at infinity or which grow quadratically on the contact cone . in either set - up one must deal with the fact that products in floer theory are most naturally defined as maps @xmath54 which involve ( at least ) multiples of a given hamiltonian function ( used to displace @xmath22 slightly to achieve self - transversality ) , and the fact that the floer complex on the right of may not be isomorphic , at chain level , to the factors on the left , even if all the @xmath55 co - incide . in the setting of linear slope hamiltonians , as in @xcite ,
one defines the morphism groups via a telescope " ( homotopy direct limit ) construction , @xmath56 $ ] , involving arbitrarily large multiples of a fixed hamiltonian of linear slope ; whilst in the setting of quadratic hamiltonians , one uses a rescaling trick ( conjugating by a suitable multiple of the liouville flow ) to identify the output chain complex in with the input complexes appearing on the left .
unfortunately , the rescaling trick is only available in the exact case , with no analogue for a general monotone convex symplectic manifold .
for instance , for negative line bundles @xmath57 , the liouville flow does not extend across the zero - section .
on the other hand , abouzaid proved the generation criterion using quadratic hamiltonians , accordingly restricted to the exact case .
similar remarks apply to ganatra s discussion of the module structure of the open - closed string map .
the proof of each of the module structure of the @xmath1-map and the generation criterion in the telescope model is surprisingly involved : the necessary modifications are numerous and non - trivial , and constitute one of the main contributions of this paper .
at the cost of appealing to kuranishi spaces or polyfolds , the arguments we give for theorem [ theorem introduction 2 ] would adapt to hold for general convex symplectic manifolds , without any monotonicity assumptions . whilst this paper is largely conceived as foundational ,
we include two illustrative applications of theorems [ theorem introduction 2 ] and [ theorem introduction 1 ] . 1 . the total space of the line bundle @xmath2 is monotone whenever @xmath58 .
when @xmath59 , @xmath60 is known explicitly through work of the first author @xcite . making heavy use of the module structure and compatibility with eigenvalue splittings , we compute the open - closed string map explicitly , and prove that the wrapped category is cohomologically finite , split - generated by closed lagrangian submanifolds . by contrast , in all previous situations in which the wrapped fukaya category has been computed ( punctured spheres , certain plumbings , cotangent bundles ) , it either vanishes or is cohomologically infinite .
the special case of @xmath61 had been studied previously , compare to ( * ? ? ? * sec.4.4 ) .
the quilt diagram in theorem [ theorem introduction 1 ] can be interpreted as a transfer - of - generation " criterion , enabling one to show that lifts of lagrangians under a very good correspondence define split - generators in terms of geometry on the domain of the correspondence . as a simple example ,
we show that if a finite group @xmath62 acts freely by symplectomorphisms on a closed monotone symplectic manifold @xmath15 with quotient @xmath63 , and if @xmath64 hits an invertible element when restricted to the subcategory generated by a collection of lagrangians @xmath65 , then @xmath66 are split - generators for @xmath67 and their preimages @xmath68 split - generate @xmath39 .
it is not hard to envisage other applications . in a series of papers ,
abouzaid @xcite proved that the cotangent fibre @xmath69 generates the wrapped category @xmath70 of the cotangent bundle of a closed manifold equipped with its canonical exact symplectic structure @xmath71 . from a dynamical point of view
, there is a great deal of interest in the convex non - exact manifolds defined by cotangent bundles equipped with magnetic forms , i.e. by @xmath72 where @xmath73 is some closed but not necessarily exact form pulled back from the base .
it seems likely that our results can be used to show a cotangent fibre still generates the wrapped category after turning on the magnetic potential . in another direction
, morse - bott degenerations provide one class of examples to which theorem [ theorem introduction 1 ] can be applied , cf .
subsection [ subsection morse bott degenerations ] , either to give information on the @xmath74 map , or conversely , when the @xmath74 map is known , to constrain the behaviour of quilt maps on quantum cohomology , which are notoriously hard to compute in general .
theorem [ theorem introduction 1 ] can also be considered a prototype for the more difficult case in which the correspondence @xmath18 is itself allowed to be non - compact .
whilst quilt theory is not manifestly sufficiently developed at present to allow such a generalisation , ganatra s recent work @xcite takes many steps in the necessary direction .
a commutative diagram of the shape presented in theorem [ theorem introduction 1 ] , but entwining open - closed maps on wrapped categories , would be of obvious interest in mirror symmetry . local calabi - yau s ( total spaces of canonical bundles over fano varieties ) are well - known testing grounds for many aspects of mirror symmetry , with rich symplectic topology @xcite .
monotone negative line bundles form another class of `` local symplectic manifolds '' , which arise naturally in symplectic birational geometry , cf . @xcite . as remarked above , whilst in known computations in the stein case the wrapped category @xmath4 either vanishes or is extremely complicated , the line bundles @xmath75 show qualitatively different behaviour .
[ conjecture introduction 1 ] let @xmath76 be a monotone negative line bundle .
if @xmath40 hits an invertible element , then @xmath4 is cohomologically finite .
conjecture [ conjecture introduction 1 ] asserts an open - string analogue of a recent result of the first author @xcite , who gave an explicit description of the symplectic cohomology of a general negative line bundle @xmath0 in terms of gromov - witten theory on @xmath15 .
the symplectic cohomology is a quotient of quantum cohomology , so of finite rank .
the hypothesis that @xmath40 hits an invertible is a version of a `` resolution of the diagonal '' statement for @xmath15 . using the cyclicity of the fukaya category @xcite , and recent work of abouzaid - fukaya - oh - ohta - ono @xcite ,
it is equivalent to the injectivity of the closed - open string map , a natural ring map @xmath77 the condition that @xmath40 hits @xmath78 is known to hold in several monotone examples : it holds for toric fano varieties @xmath15 by @xcite , and it holds for products @xmath79 if it holds for the @xmath80 by @xcite .
there are also non - toric examples , for instance it holds for the fano varieties given by a quadric and a complete intersection of two even - dimensional quadrics @xcite .
monotonicity of the line bundle @xmath0 is a fairly strong requirement : for simply connected @xmath15 , any such @xmath0 arises as the line bundle with first chern class @xmath81 , where the first chern class of @xmath15 is a non - trivial multiple @xmath82 of a primitive class @xmath83 , and where the multiple @xmath84 satisfies @xmath85 .
typical examples are the bundles @xmath86 for @xmath87 , the square root of the canonical bundle for any spin fano variety , or appropriate bundles over the twistor space of a hyperbolic @xmath88-manifold with @xmath89 ( * ? ? ?
* prop.33 ) .
there is a canonical lagrangian correspondence @xmath18 in @xmath90 , given by the coisotropic submanifold which is the circle bundle of @xmath0 of suitable radius , and the sequel @xcite studies the diagram of theorem [ theorem introduction 1 ] in this special case .
for instance , if @xmath15 and @xmath0 are ( monotone ) toric , direct computations with the landau - ginzburg superpotential of @xmath0 imply that the unique toric fibre with non - trivial floer cohomology is indeed the image , under @xmath18 , of the corresponding unique monotone toric fibre in @xmath15 , cf .
@xcite .
any precise categorical statement of the relationship between floer theory in @xmath15 and in @xmath0 , however , is necessarily rather involved . because of monotonicity , the fukaya categories @xmath39 and @xmath4 both split into collections of mutually orthogonal subcategories indexed by eigenvalues of the quantum product action of the first chern class .
consideration of the behaviour of these eigenvalues under composition with a lagrangian correspondence , cf .
section [ subsection additivity of the m0 values ] , shows that the image of @xmath39 in @xmath4 under a functor defined by @xmath18 will account for only particular idempotent summands , usually at most one . in the toric examples , whilst the monotone lagrangian tori of interest in @xmath15 and @xmath0 are superficially matched up by the correspondence @xmath18 , if one looks more closely at the local systems with which these tori must be equipped for floer cohomology to be non - vanishing , discrepancies start to appear . to hope to obtain the whole of @xmath4 functorially from @xmath18 ,
one should apparently incorporate bulk deformations of @xmath15 , and deformations of the correspondence @xmath18 itself by non - trivial self - maurer - cartan elements .
the resulting moduli spaces contributing to @xmath1 are typically obstructed , and explicit computations require virtual perturbations . we intend to address some of these issues in a sequel to this paper . + * acknowledgements . *
we are indebted to mohammed abouzaid for suggesting this project , and in particular suggesting the commutative diagram of theorem [ theorem introduction 1 ] , and for his continuing interest in and feedback on the project .
we are grateful to denis auroux and sheel ganatra for helpful conversations .
in this section we summarize the general algebraic constructions that we need to describe the geometric constructions of section [ section geometric constructions ] .
in particular , the reader might like to view the somewhat complicated looking @xmath91-relations of this section simply as a transcription of the natural bubbling phenomena shown in the figures of section [ section geometric constructions ] .
we restrict ourselves to non - curved @xmath91-categories .
later , we will view fukaya categories as direct sums of uncurved categories indexed by the eigenvalues of quantum product by the first chern class .
recall that an @xmath91-category @xmath92 is : a collection of objects @xmath93 ; graded vector spaces @xmath94 for all @xmath95 , working over a chosen ground field ; composition maps @xmath96 where we abbreviate @xmath97 for @xmath98 ( for @xmath99 define it to equal the ground field ; and we will typically denote generators by @xmath100 or @xmath101 ) ; and finally the @xmath102 must satisfy the @xmath91-relations : @xmath103 summing over all obvious choices of @xmath104 , and we now explain the sign and the gradings .
let @xmath105 be the grading of a morphism in @xmath92 , and define @xmath106 the reduced grading .
then grading signs are determined by the convention that the compositions @xmath102 are acting from the right with degree @xmath49 .
abbreviate @xmath107 which is the reduced grading of @xmath108 .
the second expression emphasizes the letter @xmath109 which is used to denote the variables indexed by @xmath110 .
a @xmath92-bimodule @xmath111 is a collection of graded vector spaces @xmath112 for any objects @xmath113 of @xmath92 , together with multi - linear maps @xmath114 for @xmath115 , and any objects @xmath116 of @xmath92 , satisfying the @xmath91-relations ( figure [ figure ainfinity eqns for bimodule b ] ) : @xmath117 summing over all obvious choices of @xmath104 , and where the signs are @xmath118 note : generators of the domain of @xmath119 are written @xmath120 , where we underline the bimodule element .
@xmath121 is a @xmath92-bimodule , with @xmath122 and composition maps @xmath123 .
a morphism of @xmath92-bimodules @xmath124 is a collection of multi - linear maps @xmath125 for @xmath115 , satisfying the @xmath91-relations ( figure [ figure ainfinity f ] ) : @xmath126 summing over the obvious choices of @xmath104 , and where the signs are as before .
let @xmath111 be a @xmath127-bimodule .
its cyclic bar complex in dimension @xmath128 is the vector space @xmath129 a typical generator will be denoted @xmath130 .
we underline the element belonging to the bimodule .
the grading for @xmath131 is : @xmath132 the bar differential @xmath133 on a generator @xmath134 is @xmath135 summing over all obvious choices of @xmath136 , where @xmath137 .
the hochschild homology is the resulting homology : @xmath138 @xmath139 is the hochschild homology of @xmath92 ( see above example ) .
let @xmath140 be an @xmath91-functor , meaning a map @xmath141 on objects together with a grading - preserving multi - linear map @xmath142 for each @xmath143 ( using reduced gradings ) , satisfying the @xmath91-relations ( figure [ figure ainfinity for phi ] ) : @xmath144 summing over the obvious @xmath136 , and over all partitions @xmath145 and all @xmath146 .
[ lemma change of rings ] an @xmath91-functor @xmath140 induces a _ change of rings _ which turns an @xmath147-bimodule @xmath148 into a @xmath92-bimodule @xmath149 .
moreover , there is a tautological chain map @xmath150 for any @xmath151 , define @xmath152 we extend the natural map @xmath153 to define new @xmath154 : @xmath155 summing over all partitions @xmath156 and @xmath157 , and all @xmath158 . the tautological chain map @xmath159 is given by @xmath160 summing over all partitions @xmath161 and all @xmath162 . the fact that this is a chain map is a consequence of the @xmath91-relations satisfied by @xmath163 . [ lemma functoriality of hh ] a morphism of @xmath92-bimodules @xmath164 induces a chain map @xmath165 define @xmath166 by sending @xmath130 to @xmath167 where @xmath168 .
this is a chain map .
let @xmath111 be a @xmath92-bimodule , and @xmath148 an @xmath147-bimodule .
lemma [ lemma change of rings ] defines @xmath149 .
[ corollary induced map on hh ] a map @xmath169 of @xmath92-bimodules induces natural homomorphisms @xmath170 at the chain level , the first map is @xmath171 , determined by lemma [ lemma functoriality of hh ] , and the second map is the tautological map of lemma [ lemma change of rings ] .
this section is needed only in the proof of the generation criterion [ subsection outline of the proof of generation ] .
let @xmath172 be an @xmath91-category .
suppose @xmath173 are respectively a left and a right @xmath172-module .
this means to each object @xmath174 in @xmath172 we have associated objects @xmath175 of @xmath172 , and maps @xmath176 we now define : a chain complex @xmath177 , and an @xmath172-bimodule @xmath178 , built so that @xmath179 the chain complex is given by @xmath180 graded by the sum of the degrees , using reduced degrees for @xmath172 , and the differential ( of degree @xmath181 ) on a generator @xmath182 is defined by @xmath183 summing over the obvious @xmath104 .
the @xmath172-bimodule is defined by @xmath184 with composition maps @xmath185 defined on generators by @xmath186 and @xmath187 is zero whenever @xmath136 are both non - zero .
there is an obvious map from the cyclic bar complex for @xmath178 to the chain complex @xmath177 which simply reorders the positions of @xmath188 and @xmath189 ( with the obvious sign change due to reordering dictated by the gradings ) .
after the reordering , the bar differential becomes precisely the differential defined for @xmath177 .
thus @xmath190 . for ease of reference ,
we summarize here all grading conventions and results arising in the paper , and we do not discuss these further ( we refer to @xcite for a detailed discussion of gradings ) . on first reading this section should be skipped . 1 .
the novikov variable @xmath191 for @xmath0 has cohomological degree @xmath192 ; 2 .
unreduced gradings for lagrangian floer cohomology are as in seidel @xcite ; 3 . by convention , @xmath193 has degree @xmath194 , 4 .
and so @xmath195 has degree @xmath194 ; 5 .
products preserve unreduced gradings , units lie in unreduced grading @xmath194 ; 6 .
we never reduce gradings of module inputs / outputs or of @xmath29 ; 7 .
we reduce gradings of morphisms of @xmath91-categories , so all @xmath196 have degree @xmath181 ; 8 .
@xmath197 has degree @xmath194 ; 9 .
@xmath198 has degree @xmath194 ( unreduced gradings ) ; 10 .
@xmath199 has degree @xmath200 ; 11 .
@xmath201 has degree @xmath202 ; 12 .
@xmath203 has degree @xmath200 ( the underlined terms are viewed as modules , so their degrees are not reduced ) ; 13 .
@xmath204 has degree @xmath205 if @xmath206 ; 14 .
@xmath207 has degree @xmath84 if @xmath206 .
we now define the wrapped fukaya category @xmath208 of a monotone symplectic manifold @xmath0 which is conical at infinity .
we will mimick the construction of abouzaid - seidel @xcite which dealt with exact @xmath0 .
the construction of abouzaid @xcite for exact @xmath0 using quadratic hamiltonians is somewhat simpler than @xcite , but relies heavily on having a globally defined liouville flow ( which is used to offset the problem that the @xmath91-composition maps increase the slope of the hamiltonian ) .
this construction can not be applied to monotone @xmath0 , because the liouville flow is not globally defined : it is only defined outside of a compact subset of @xmath0 .
@xmath209 will always denote a ( non - compact ) symplectic manifold which is _ conical at infinity _ , which means outside of a bounded domain @xmath210 there is a symplectomorphism @xmath211 where @xmath212 is a contact manifold , and @xmath213 is the coordinate on @xmath214 .
the conical condition implies that outside of @xmath215 the symplectic form becomes exact : @xmath216 where @xmath217 .
it also implies that the liouville vector field @xmath218 ( defined by @xmath219 ) will point strictly outwards along @xmath220 .
it also implies that @xmath221 is induced by the flow of @xmath222 for time @xmath223 , so we can write @xmath224 , @xmath225 ( pull - back ) .
we will call @xmath226 the _ conical end _ of @xmath0 . by _
conical structure _ @xmath227
we mean an @xmath228-compatible almost complex structure on @xmath0 ( so @xmath229 is a @xmath227-invariant metric ) satisfying the _ contact type condition _ @xmath230 for large @xmath213 . on @xmath231
this implies @xmath232 where @xmath233 is the reeb vector field for @xmath212 defined by @xmath234 , @xmath235 . by choosing @xmath236 or @xmath231 generically ,
one ensures that @xmath236 is sufficiently generic so that the periods of the closed orbits of @xmath233 form a countable closed subset of @xmath237 .
we always assume @xmath209 is _ monotone _ : @xmath238 = \lambda_e \omega[u ] \textrm { for all } [ u]\in \pi_2(e)}\ ] ] for some monotonicity constant @xmath239 in @xmath240 , where the notation means we are integrating the cohomology class over the sphere @xmath241 . a submanifold @xmath242 is lagrangian if @xmath243 and @xmath244 .
we will always assume that the lagrangians are _ monotone _ , meaning : @xmath245)\,\lambda_{l } \quad \textrm{for } [ u]\in \pi_2(e , l)}\ ] ] for some constant @xmath246 , where @xmath247 is the maslov index .
recall the maslov index @xmath248 is associated to any continuous disc @xmath249 , and @xmath250 is even for orientable @xmath22 .
the maslov index only depends on the @xmath251 class of @xmath241 .
moreover `` gluing in a sphere ''
@xmath252 causes @xmath253 ) + \mu(u)$ ] .
this gluing relation forces @xmath0 to be monotone , and the monotonicity constants are related by : @xmath254 we will always work over the field @xmath255 called the _ novikov ring _
, where @xmath256 is some chosen ground field , and @xmath191 is a formal variable . the choice of @xmath256 will not matter , except in the toric examples at the end of the paper , where we will specialize to @xmath257 .
the novikov ring is graded by placing @xmath191 in ( real ) degree @xmath258 in all future floer constructions , solutions will be counted with novikov weights .
for example for the floer differential , the floer solutions @xmath241 are counted with weight @xmath259}$ ] , which lies in degree @xmath260 = 2c_1(te)[u]$ ] .
we will not discuss the orientation signs @xmath261 in this paper ( see @xcite ) .
the objects of @xmath208 are the closed monotone orientable lagrangian submanifolds of @xmath0 , together with all non - compact monotone orientable lagrangian submanifolds @xmath263 which are _ conical at infinity _ , by which we mean : 1 .
[ item transverse to bdry ] @xmath22 intersects @xmath231 transversally ; 2 .
[ item conical on the end ] the pull - back @xmath264 near @xmath231 and on the conical end .
the above conditions ensure that @xmath265 is a legendrian submanifold of the contact manifold @xmath266 , and that @xmath22 has the form @xmath267 near the conical end .
conversely , a monotone lagrangian @xmath268 with @xmath264 near @xmath231 can be extended to an @xmath22 of the above form .
in fact , any monotone lagrangian @xmath268 which intersects @xmath231 in a legendrian submanifold , can be hamiltonianly isotoped in @xmath215 relative to @xmath220 so that @xmath269 will vanish near @xmath231 . this can be proved as in ( * ? ? ?
* lemma 3.1 ) ( it is a local argument , and the liouville flow is defined near @xmath231 ) . finally , as will be explained in detail in [ subsection the role of monotonicity ] , one must in fact restrict @xmath270 to lagrangians having the same @xmath271-value .
for @xmath4 to be graded one would require @xmath272 . for @xmath4 to be defined over a field of characteristic @xmath273 one would require @xmath274 and
then an @xmath22 should come with a choice of pin structure .
for these details we refer to @xcite : we will not discuss gradings or orientation signs for sake of brevity .
fix a hamiltonian @xmath275 which on the conical end has the form @xmath276 for @xmath277 .
later , in [ subsection symplectic cohomology ] ( see the convention ) , we will make a global rescaling of @xmath278 , but the reader can ignore this for now .
let @xmath174 be the hamiltonian vector field : @xmath279 .
the reeb vector field on @xmath231 is @xmath280 .
fix a countable collection @xmath281 , so they give rise to @xmath282 .
a _ reeb chord of period _
@xmath283 is a trajectory @xmath284\to \sigma$ ] of @xmath285 with ends on @xmath286 ( for some , possibly non - distinct , indices @xmath287 ) .
it is an _ integer _
reeb chord if @xmath283 is an integer . by rescaling @xmath228 ( or by choosing @xmath231 generically )
, there are no integer reeb chords .
an _ integer @xmath174-chord of weight @xmath283 _ is an orbit @xmath284\to e$ ] of @xmath288 with ends on @xmath289 for integer @xmath290 .
these lie in @xmath215 ( otherwise , there would be one in the conical end projecting to an integer reeb chord in @xmath231 ) . for a generic choice of @xmath278 ( see @xcite ) , integer @xmath174-chords are non - degenerate and any point of @xmath291 can not be both an endpoint and a starting point of two ( possibly non - distinct ) integer @xmath174-chords .
so only finitely many integer @xmath174-chords have ends on given @xmath289 , and they lie in @xmath215 . given @xmath293 , let @xmath294 denote the free @xmath8-module generated by the ( finitely many ) integer @xmath174-chords of weight @xmath283 with ends on @xmath289 .
fix a conical structure @xmath227 on @xmath0 ( we remark that the contact type condition in [ subsection symplectic manifolds conical at infinity ] on the conical end is equivalent to @xmath295 ) .
the chain differential @xmath296 is defined as follows on a generator @xmath297 .
let @xmath298 \to e$ ] be an isolated ( modulo @xmath240-translation ) solution of floer s equation @xmath299 , satisfying : lagrangian boundary conditions @xmath300 and @xmath301 ; and asymptotic conditions @xmath302 as @xmath303 respectively , where @xmath109 is an @xmath174-chord
. then @xmath241 contributes @xmath259}\ , x$ ] to @xmath304 , where @xmath305=\int u^*\omega\in { \mathbb{r}}$ ] .
we will always omit discussing orientation signs @xmath261 , which is done carefully in @xcite by replacing each @xmath306 by an orientation line . if @xmath307 , then the floer cohomology of @xmath289 using @xmath308 is defined as @xmath309 since @xmath0 is monotone , @xmath307 precisely when @xmath289 have the same @xmath271-value ( which we define and explain in [ subsection the role of monotonicity ] ) .
we keep the discussion quite general by supposing that we have a @xmath49-dimensional family of floer - type solutions ( such as those arising in the maps we just mentioned ) and we study the possible bubblings .
first observe that sphere bubbling is ruled out for dimension reasons , since by monotonicity a non - constant holomorphic sphere will increase the index of the family by at least @xmath310 .
so only disc bubbling can arise . by disc bubble ,
we mean a non - constant holomorphic disc @xmath311 , bounding a single lagrangian @xmath312 say , with one boundary marked point @xmath313 landing at a prescribed generic point @xmath314 ( so @xmath315 plays the role of the node when our family breaks ) .
the index of families is additive , so only isolated discs will bubble off in our @xmath49-dimensional family and so , by monotonicity and a dimension count , the disc must have maslov index @xmath316 .
now consider the moduli space @xmath317 of all such maslov @xmath310 discs in the class @xmath318 but do not impose a condition on the value at the marked point .
let @xmath319 be the evaluation at the marked point .
then @xmath320 $ ] determines a locally finite cycle in @xmath22 ( it is a cycle and not just a chain , because @xmath317 can not have boundary degenerations , since by monotonicity @xmath310 is the minimal maslov index of non - constant disc bubbles ) . for dimension reasons
this lf - cycle must be a scalar multiple of @xmath321 $ ] ( viewed as a locally finite cycle since @xmath22 may be non - compact - we explain this in [ subsection construction of m0 ] ) .
then define @xmath322}\ , \mathrm{ev}_*[\mathcal{m}_1(\beta)]= m_0(l)\ , [ l ] \in c_{\mathrm{dim}(l)}^{\mathrm{lf}}(l;\lambda ) , \textrm { where } m_0(l)\in \lambda.\ ] ] when there are no such @xmath323 discs , then @xmath324 , and if this holds for all objects @xmath22 then the @xmath91-category is called _ unobstructed _ , and no disc bubbling discussion is necessary . in the applications we have in mind , we typically expect @xmath323 discs .
so we briefly explain why in the monotone case , thanks to the machinery of fukaya - oh - ohta - ono @xcite , all bubbling contributions will in fact cancel out , so one can ignore all contributions coming from @xmath325 . in other words , monotonicity allows us to pretend that our @xmath91-categories @xmath326 are all unobstructed .
using monotonicity to deduce that the only non - constant holomorphic discs bounding @xmath22 have maslov index at least @xmath310 , the machinery of fukaya - oh - ohta - ono @xcite then yields a chain level model for @xmath327 using locally finite chains @xmath328 for which @xmath327 is a well - defined ring in which @xmath321 $ ] represents the unit .
we briefly explain this . at the chain level
the first @xmath91-equation for @xmath22 is @xmath329 for @xmath330 .
the fact that the right - hand side vanishes relies on the observation that these @xmath331 terms come from counting discs with three boundary marked points , but for which the incidence condition @xmath321 $ ] is automatic .
so forgetting that unnecessary marked point shows that we are sweeping a chain contained inside @xmath332 , which has dimension one less than @xmath109 ( unless the disc is constant ) .
so only constant discs can contribute , and so it turns out that @xmath333)=x$ ] and @xmath334,x)=(-1)^{|x|}x$ ] , so the above right - hand side becomes a multiple of @xmath335 .
so @xmath336 really is a differential .
similarly , feeding a copy of @xmath321 $ ] ( or a copy of @xmath325 ) into an entry of any of the higher @xmath196 composition maps ( using only copies of @xmath337 ) , then again we get zero , and for dimension reasons , @xmath338)=0 $ ] .
so @xmath321 $ ] is a strict unit for the subcategory generated by @xmath22 ( * ? ? ?
* sec.2a ) .
these arguments generalize to the case when one considers different lagrangians : for @xmath339 one gets @xmath340 provided that @xmath341 .
so @xmath342 is only defined if @xmath341 , and similarly the higher order compositions @xmath343 are only defined when all the lagrangians have the same @xmath271 value .
so the ( wrapped ) fukaya category is actually a family of @xmath91-categories indexed by the possible @xmath271 values .
since this is central to our treatment , we state it as a lemma ( the observation relating @xmath271 to the eigenvalues of @xmath344 is due to kontsevich , seidel and auroux ( * ? ? ?
* sec.6 ) ) : [ lemma fukaya cats split according to evalues ] the category @xmath4 is a family of @xmath91-categories @xmath326 indexed by @xmath345 , where @xmath326 is generated by the lagrangians @xmath22 with @xmath346 . since @xmath0 is monotone , the non - triviality of @xmath326 implies that @xmath347 is an eigenvalue of quantum cup product by @xmath344 : @xmath348 the analogous statement holds for the compact category @xmath42 or the fukaya category @xmath39 of a closed ( rather than convex ) monotone manifold @xmath15
. * convention . * from now on , we always treat @xmath4 as a non - curved @xmath91-category which splits into non - interacting pieces , not as a curved category .
so it is always assumed that when we consider maps on @xmath4 , we in fact mean some @xmath326 so that all the lagrangians involved have the same @xmath271-value .
the same comment pertains to @xmath39 and @xmath42 .
+ our discussion shows that the @xmath91-equations for the wrapped category naturally involve also terms containing @xmath349 s , but that these terms in fact cancel out ( since we assume that all lagrangians involved have the same @xmath271-value ) .
this implies that also , for instance , the bar differential for @xmath350 will have terms involving @xmath349 s , and that these cancel .
finally we make some technical remarks about why @xmath349 is defined in the non - compact setup ( the machinery of fukaya - oh - ohta - ono @xcite constructs @xmath349 in the closed monotone setup ) .
we begin with an observation : let @xmath0 be conical at infinity .
let @xmath263 be any lagrangian submanifold conical at infinity ( [ subsection the objects of the wrapped fukaya category ] )
. then all pseudo - holomorphic discs @xmath351 have @xmath352 lying either in @xmath215 or in a contact hypersurface @xmath353 of the conical end . for general reasons of convexity ( * ? ? ?
* sec.7c ) , which essentially reduce to green s formula , any pseudo - holomorphic map @xmath354 defined on a riemann surface @xmath355 ( which can have corners ) with @xmath356 landing in the conical end of @xmath0 , with @xmath357 , must in fact lie entirely in @xmath353 .
this maximum principle uses the fact that @xmath353 is a contact hypersurface in @xmath0 on which the liouville flow is outward - pointing .
the claim follows immediately by taking @xmath358 to be the restriction of @xmath241 to @xmath359 ( for generic @xmath213 so that @xmath353 intersects the image of @xmath241 transversely ) . by the lemma ,
given a point @xmath360 of @xmath22 , all holomorphic discs bounding @xmath22 satisfying @xmath361 must lie entirely in a compact region of @xmath0 determined by @xmath360 .
thus all the constructions relating to the moduli space of holomorphic discs and in particular the construction of @xmath325 reduce to the constructions in the closed case , provided we replace cycles / chains by locally finite cycles and locally finite chains .
for example , suppose @xmath362 are monotone .
then , using the analysis @xcite known for the closed setup , @xmath349 is , up to a scalar multiple , a sum of chains which on each compact @xmath363 is building the fundamental cycle for the pair @xmath364 . exhausting @xmath0 by a sequence of larger and larger compacts
@xmath365 proves that @xmath325 is , up to a scalar multiple , an infinite sum of chains which is precisely the locally finite cycle @xmath321 $ ]
. denote @xmath366 the floer continuation map @xmath367 determined by a monotone homotopy from @xmath368 to @xmath308 .
this involves counting isolated solutions similar to those in the definition of @xmath292 , except now @xmath369 depend on the domain coordinate @xmath370 and there is no @xmath240-reparametrization .
section [ subsection the auxiliary data for wrapped category ] describes the auxiliary data carefully ( for @xmath371 one uses : @xmath372 , @xmath373 , @xmath374 , @xmath375 ) .
the maps @xmath371 are the chain maps which at the level of cohomology are used to define the _ cohomology direct limit _ as @xmath376 : @xmath377 because the @xmath91-operations act on the chain level and require changing the slope @xmath283 of @xmath308 , the wrapped floer complex has to be built as a homotopy direct limit , in other words by a telescope construction ( rather than taking the direct limit at the level of cohomology ) . following ( * ? ? ?
* sec.3 g , sec.5a ) , [ definition wrapped floer complex ] the wrapped complex is @xmath378}\ ] ] where @xmath379 is a formal variable of degree @xmath380 satisfying @xmath381 .
this is a module over the exterior algebra @xmath382 $ ] , where the differentiation operator @xmath383 has degree @xmath181 .
the chain differential is chosen to respect this module structure : @xmath384 observe that for a @xmath292-cocycle @xmath297 , the role of @xmath385 is to identify @xmath297 and @xmath386 at the cohomology level , as expected in the cohomology direct limit ; whereas the role of the subcomplex @xmath387 is to yield representatives of @xmath388 . indeed , recall ( * ? ? ?
* sec.3 g ) the following about @xmath389\oplus cf^*(l_i , l_j;bh)$ ] : 1 . [ item 0 ] @xmath390 ; 2 .
[ item 1 ] @xmath391 is identifiable with @xmath392 ( change signs in the differential ) ; 3 .
[ item 2 ] @xmath393 is quasi - isomorphic to @xmath394 ; 4 .
[ item 3 ] @xmath395 is a quasi - isomorphism for @xmath396 , hence so is @xmath397 ; 5 .
[ item 4 ] @xmath397 is a quasi - isomorphism ; 6 .
[ item 5 ] @xmath398 and @xmath399 commute via ( [ item 4 ] ) , up to chain homotopy ; 7 .
[ item 6 ] @xmath400 canonically ; ( [ item 0 ] ) is because @xmath401 is an exhausting increasing filtration of @xmath394 .
the @xmath402 , @xmath403 are decreasing filtrations for @xmath394 , @xmath401 respectively ; so ( [ item 2])-([item 4 ] ) follow : successive quotients are cones on the identity , so acyclic . combining yields ( [ item 6 ] ) . by @xcite , there is a universal family @xmath404 over the moduli space of stable discs , with elements @xmath405 where ( figure [ figure popsicle ] ) : 1 . @xmath406 is a finite set of indices @xmath407 ( with @xmath408 , @xmath409 ) ; 2 .
a map @xmath410 defining labels @xmath411 ( possibly not distinct ) ; 3 . _
( disc with @xmath412 ends ) .
_ @xmath355 is a riemann surface isomorphic to @xmath413 with @xmath412 distinct boundary punctures @xmath414 ( of which @xmath315 is distinguished , and the others are ordered according to the orientation of @xmath415 ) ; 4 .
[ item preferred points ] _
( strip - like parametrizations ) . _
@xmath416 are holomorphic @xmath417 extending to smooth isomorphisms on the compactifications @xmath418 , with @xmath419 , @xmath420 ( where @xmath222 is the strip @xmath421 $ ] viewed as a disc with @xmath310 ends ) .
this is the same as a choice of a _ preferred point _ on the hyperbolic geodesic joining @xmath315 , @xmath422 ( corresponding via @xmath423 to the point @xmath424 and to the geodesic @xmath425 ) ; we will omit the discussion of finite symmetries ( * ? ? ?
* sec.2c , 2f , lemma 3.7 ) , but we mention that all auxiliary data constructions are done so as to preserve the symmetries which exchange preferred points lying on the same geodesic .
the above family is compactified by boundary strata described by the natural bubbling configurations one expects when a subset of the preferred points of ( [ item preferred points ] ) converges towards a boundary puncture ( * ? ? ?
* fig.2 , fig.3 ) .
we omit discussing gluing ( * ? ? ?
* sec.2d ) , but we mention one important technical aspect : _ remark 1 .
one actually allows the geodesics in ( [ item preferred points ] ) to be deformed ( except near the ends ) .
we explain why in an example .
suppose we want to glue the first two discs in figure [ figure popsicle ] at one puncture .
call the discs @xmath426 and we want to glue @xmath427 with @xmath428 .
gluing gives a @xmath429-punctured disc with a geodesic and 2 curves which are not geodesics ( they overlap along a common curve carrying the preferred point @xmath430 ) but they can be deformed into two geodesics .
the `` glued '' @xmath406 is @xmath431 .
there are four choices for the `` glued '' @xmath432 : @xmath433 can be any of the @xmath434 coming from @xmath435 , and in the cases @xmath436 and @xmath437 we would need to draw a new geodesic before gluing , so that we obtain glued data as in ( [ item preferred points ] ) ( up to deforming ) . now observe the third disc in figure [ figure popsicle ] .
the dotted line is a `` cut '' : we consider a @xmath49-parameter family of such discs in which the auxiliary data contained on one side of the cut converges towards a common boundary puncture . in the limit ,
the disc breaks into two , namely @xmath438 glued with @xmath435 with `` glued '' @xmath432 determined by @xmath439 , and the new geodesic we draw on @xmath435 is remembering that @xmath430 came from the geodesic joining @xmath315 to @xmath437 before breaking . _ now assign further auxiliary data : 1 . _ ( weight ) .
_ @xmath440 are positive integers with @xmath441 ; 2 . _
( strip - like ends ) .
_ holomorphic embeddings @xmath442 , for @xmath443 , of @xmath444\subset z$ ] mapping the boundaries @xmath445 to @xmath415 and converging to @xmath446 as @xmath447 .
similarly @xmath448 for the negative strip @xmath449\times [ 0,1]$ ] ; 3 . _
( closed @xmath49-forms ) .
_ closed @xmath49-forms @xmath450 on @xmath355 for @xmath443 pulling back to : @xmath194 on @xmath415 ; @xmath451 via @xmath452 on @xmath453 ; @xmath194 via the other @xmath454 on @xmath455 .
( subclosed @xmath49-forms ) .
_ @xmath49-forms @xmath456 on @xmath355 with @xmath457 ( for the complex orientation of @xmath355 ) , @xmath458 near @xmath415 , and such that @xmath456 pulls back to : @xmath194 on @xmath415 ; @xmath451 via @xmath459 on @xmath460 ; @xmath194 via the other @xmath461 on @xmath462 . in practice
, one always chooses @xmath456 so that @xmath458 except on a small neighbourhood of the preferred point @xmath423 .
the above data determines @xmath463 , a @xmath49-form on @xmath355 with @xmath464 , @xmath465 near @xmath415 .
also @xmath466 pulls - back to : @xmath194 on @xmath415 ; @xmath467 via @xmath454 on @xmath453 .
observe that this is consistent with stokes s theorem , @xmath468 .
the choices of @xmath469 can be made to vary smoothly over the moduli space ( * ? ? ?
* sec.2e,2f ) consistently with the boundary strata , by an inductive argument .
we make one important technical remark about gluing : _ remark 2 . when gluing a disc @xmath438 at @xmath470 with a disc @xmath435 at @xmath471 , we require the weights to agree : @xmath472 .
the @xmath236 forms glue naturally . as for the @xmath473 forms : they are indexed by @xmath406 , not @xmath474 , so define @xmath475 on @xmath438 and @xmath476 on @xmath435 , then the @xmath473 s glue in a natural way to yield @xmath456 s on the gluing indexed by @xmath477 .
_ the last piece of auxiliary data is the almost complex structure : 1 . _
( almost complex structures ) . _
a family @xmath478 of conical structures which pull - back via @xmath454 on @xmath479 to almost complex structures @xmath480 depending only on @xmath191 ( which are chosen before - hand for each integer @xmath283 )
. this choice can be made smoothly over the moduli space , and compatibly with gluing ; 2 . _
( infinitesimal deformations ) .
_ certain deformation data @xmath481 for @xmath482 ( * ? ?
* sec.3b ) ; 3 . _
( regular almost complex structures ) _ the @xmath483 give rise to a family of perturbed almost complex structures @xmath484 , agreeing with @xmath482 asymptotically , and which vary smoothly over the moduli space .
let @xmath485 be @xmath174-chords of weight @xmath486 with ends on @xmath487 , @xmath488 , @xmath489 , @xmath490 , @xmath491 , respectively for @xmath492 , and where @xmath281 .
denote by @xmath493 the moduli space of solutions @xmath494 of @xmath495 = 0\ ] ] with lagrangian boundary conditions @xmath496 where @xmath497 is the component of @xmath415 between @xmath498 and @xmath499 ; and asymptotic conditions @xmath500 as @xmath501 . for a generic choice of @xmath481
the @xmath493 are smooth manifolds of the expected dimension ( * ? ? ?
* thm 3.5 ) . in the exact setup of @xcite
, there is a natural compactification of @xmath493 modeled on the compactification of @xmath502 ( * ? ? ?
* sec.3e . ) .
the compactness relied on two ingredients : ( 1 ) a maximum principle forcing solutions to lie in @xmath215 ( * ? ? ?
* sec.7d ) ; ( 2 ) an a priori energy estimate : @xmath503 using @xmath504 , @xmath464 , and by stokes s theorem the topological energy @xmath505 is bounded ( in the exact setup ) in terms of the asymptotics @xmath506 using action functionals ( * ? ? ?
. in our non - exact setup , ( 1 ) still holds since the argument only relied on exactness of @xmath228 on the conical end .
we no longer have an a priori estimate for @xmath505 since the action functionals are multi - valued .
however , since we will count solutions with weight @xmath507 ( notice this is a homotopy invariant of @xmath241 relative to the ends ) , we only need to ensure compactness of @xmath508 which is the union of the components which have @xmath509 . but
this now follows as in the exact case ( 2 ) since we artificially imposed an a priori bound on @xmath510 via @xmath511 .
observe that , because we work over the novikov ring , we are not using the monotonicity assumptions here to obtain an energy - index relation which in turn would give a priori bounds on @xmath510 in terms of the index ( unlike @xcite , who use this approach since they are working over @xmath512 ) .
we only use the monotonicity assumptions in [ subsection the role of monotonicity ] and in [ subsection phi lands in fuk e ] to control @xmath349 . _
we now define , for @xmath513 , the maps @xmath514 : @xmath515\otimes \cdots \otimes cf^*(l_0,l_1;w_1h)[\mathbf{q } ] \to cf^*(l_0,l_d;w_0h)[\mathbf{q}]\ ] ] first we determine the @xmath516 coefficient of the image .
an isolated @xmath517 will contribute @xmath518 to @xmath519 where @xmath520 for @xmath521 , @xmath522 for @xmath523 , and it will contribute @xmath194 otherwise ( for orientation signs , see ( * ? ? ? * sec.3h ) ) .
the reason for the mysterious - looking conditions on @xmath474 is that when @xmath520 the formal variable @xmath524 plays the algebraic role of remembering that there is a preferred point on the geodesic in @xmath355 connecting @xmath525 . for symmetry reasons (
* lemma 3.7 ) , there is a cancellation of the counts of pseudo - holomorphic maps involving auxiliary data where more than one preferred point lies on a geodesic , so we only need to keep track of whether there is or there is nt a preferred point on each geodesic ( encoded by @xmath524 , @xmath516 ) .
encoding this information in the algebra is necessary , so that the breaking analysis for the family @xmath404 will indeed yield the @xmath91-equations for the @xmath196 that we define below ( see remarks 3 and 4 below , or see ( * ? ? ?
* ( 3.29 ) ) ) . the @xmath524 coefficient of the image is determined by the requirement that the map @xmath514 respects the @xmath383 operator as follows , for @xmath513 : @xmath526 where @xmath527 is defined in [ subsection grading conventions ] .
observe that this equation determines all possible @xmath519 inductively on the value @xmath528 ( for an example , see ( * ? ? ?
* ex.3.14 ) ) .
the geometrical meaning of @xmath383-equivariance is explained in remark 3 below . finally , recall that @xmath381 , so we have determined the image .
summing up all @xmath514 as @xmath529 vary , we obtain the @xmath91-operations for @xmath513 : @xmath530 we conclude with two technical remarks relating to gluing : _ remark 3 . in remark 1 of [ subsection the auxiliary data for wrapped category ] we described the gluings of the first two discs @xmath426 in figure [ figure popsicle ] for four choices of
`` glued '' data @xmath432 .
let us pretend @xmath531 is not present ( otherwise @xmath435 will give cancelling contributions due to the symmetry which swaps @xmath532 ) .
the discs @xmath426 define two operators @xmath533 , whose @xmath516-output is non - zero only if we input terms of the form @xmath534 , @xmath535 respectively , and the @xmath516-outputs say @xmath536 are then a count of isolated solutions for the relevant data .
the @xmath533 are in fact contributions to @xmath537 .
gluing means composing operators , but @xmath438 requests a @xmath524-input at the gluing puncture ( namely @xmath538 ) .
the @xmath524-output of @xmath539 is determined by @xmath383-equivariance : for example @xmath540 ( ignore signs for simplicity ) .
so in the composition , we take @xmath541 : those correspond to two of the four gluing choices ( the other two involve @xmath381 ) . for example , @xmath542 corresponds to the choice @xmath439 , and it arises from the breaking of figure [ figure popsicle ] ( right ) . in particular , at the level of composing operators , each term of the composition corresponds to precisely one type of breaking : so the ambiguity of the gluing choice in the auxiliary data disappears .
this is crucial since distinct operator composites should correspond to distinct types of breakings . _
+ _ remark 4 .
we now explain the @xmath91-equations for the @xmath196 in an example .
we ignore signs , so we work modulo @xmath310 .
recall @xmath543 and @xmath544 . consider the @xmath91-equation @xmath545 .
evaluate this equation on @xmath546 .
first , the @xmath524-coefficient is : @xmath547 , which is the usual @xmath91-equation without @xmath379 s coming from the 3 possible cuts of a @xmath548-punctured discs with no geodesics ( in fact there are also @xmath349 terms , corresponding to cuts which do nt have any puncture on one side of the cut , but these cancel - see [ subsection the role of monotonicity ] ) .
secondly , the @xmath516-coefficient : this has 7 terms of which two cancel , leaving @xmath549 .
these are the breakings coming from the cuts shown in figure [ figure cutting ] . in general
, @xmath550 corresponds to the disc with 2 punctures , 1 geodesic , 1 preferred point ; @xmath551 arises as an incoming disc with @xmath310 punctures , 1 geodesic , no preferred point ( i.e. the @xmath292 disc with a new geodesic drawn ) . _
we refer to salamon @xcite for a precise definition of the floer cohomology @xmath552 for a hamiltonian @xmath553 defined on a closed monotone symplectic manifold @xmath554 .
in particular ( * ? ? ?
* lecture 3 ) discusses the role of monotonicity .
we recall that at the chain level , @xmath555 is generated , over the novikov ring @xmath8 , by all @xmath49-periodic hamiltonian orbits of @xmath556 .
the differential @xmath557 is defined as follows on generators : there is a contribution @xmath558 } x_-$ ] to @xmath559 for each isolated ( up to @xmath240-reparametrization ) map @xmath560 solving floer s equation @xmath561 with asymptotic conditions @xmath562 as @xmath563 , where @xmath564 is a fixed almost complex structure on @xmath15 compatible with @xmath565 . up to isomorphism , the @xmath552 do not depend on the choice of @xmath566 .
indeed , a homotopy @xmath567 which is constant for @xmath479 , induces a _ continuation map _
@xmath568 by counting solutions @xmath241 of floer s equation above where now @xmath569 depend on the coordinate @xmath570 ( and we no longer have an @xmath240-reparametrization action ) .
this continuation map is always an isomorphism ( the inverse is defined using the `` reversed homotopy '' @xmath571 ) .
when @xmath556 is @xmath572-small and morse , and @xmath566 are time - independent , then @xmath555 reduces to the morse complex for @xmath556 , in particular all @xmath49-periodic hamiltonian orbits are constant , and so by invariance we have an isomorphism @xmath573 which by pss @xcite ( see [ subsection pss ] ) intertwines the quantum product and the pair - of - pants product . for symplectic manifolds @xmath0 conical at infinity ( see [ subsection symplectic manifolds conical at infinity ] )
@xmath574 heavily depends on the choice of @xmath575
because the larger @xmath576 is on the conical end the likelier it is that there are many @xmath49-periodic hamiltonian orbits there .
the continuation maps are no longer isomorphisms , because they can no longer be reversed : only certain homotopies @xmath577 with @xmath578 are guaranteed to be well - defined .
so to construct an invariant one needs to choose a class of hamiltonians which grow in a controlled way at infinity .
we consider @xmath576 which at infinity are linear of positive slope in the radial coordinate @xmath213 ( see [ subsection symplectic manifolds conical at infinity ] ) .
one then takes a direct limit over the continuation maps which increase the slope of @xmath576 at infinity : @xmath579 where @xmath580 .
this defines the _ symplectic cohomology _ @xmath581 the symplectic cohomology in the non - exact setup was first constructed by ritter @xcite , and was discussed further in @xcite .
the reader might like to consider first the original construction in the exact setup , due to viterbo @xcite , which is surveyed by seidel @xcite and whose algebraic structures such as the pair - of - pants product are constructed in ritter @xcite . when @xmath576 is @xmath572-small on @xmath215 , morse , time - independent , and of small enough slope on the conical end , the chain complex @xmath582 reduces to the morse complex for @xmath0 , and since this is part of the direct limit construction there is always a canonical map ( which can be non - injective and can be non - surjective ) : @xmath583 @xmath584 splits into a sum @xmath585 indexed by the eigenvalues of the quantum product action of @xmath586 on @xmath587 ( see theorem [ theorem oc is a module map ] ) . in this paper
we follow the conventions described in @xcite , except we use the simpler novikov ring defined in [ subsection novikov ring ] since we will always work in the monotone setting .
+ * warning ( grading ) .
* @xmath33 is typically not @xmath512-graded .
indeed , floer cohomology is graded by the conley - zehnder index , which is a @xmath588-grading where @xmath589 .
+ for the purpose of defining the open - closed string map in section [ section open - closed string map ] , it is actually convenient to view also @xmath33 as a homotopy direct limit ( recall [ subsection the wrapped floer complex ] ) .
namely , define @xmath590\ ] ] with @xmath336 defined by the same formula as in definition [ definition wrapped floer complex ] , where @xmath292 is now the differential defining hamiltonian floer cohomology and @xmath371 are the floer continuation maps @xmath591 analogous to [ subsection kappa map ] . + * convention . * in [ subsection reeb chords and hamiltonian chords ] , we chose @xmath278 to have slope @xmath49 at infinity , but we could have rescaled @xmath278 by a constant @xmath592 ( once and for all ) . from now on , we assume @xmath278 is morse and @xmath572-small on @xmath215 , and linear on the conical end with slope @xmath592 smaller than the minimal reeb period of the closed reeb orbits in @xmath231 .
this ensures that the only @xmath49-periodic orbits of @xmath278 lie in @xmath215 and are constant ( on the other hand @xmath308 typically will have non - constant @xmath49-periodic orbits on the conical end for large @xmath283 ) .
[ lemma sh is h of sc and c is inclusion ] @xmath593 and the canonical map @xmath594 arises as the inclusion of the subcomplex @xmath595 .
the first claim follows by the same arguments as in [ subsection the wrapped floer complex ] , the second claim follows by definition ( and required the rescaling in the convention ) .
the floer differential preserves the free homotopy class of the orbits , so one can restrict the chain complexes @xmath555 and @xmath596 to only contractible orbits .
denote their cohomologies by @xmath597 and @xmath598 . by invariance , and considering small @xmath556 as mentioned in [ subsection hamiltonian floer cohomology ] , we deduce @xmath599 for any @xmath556 .
similarly when @xmath576 has small slope , such as the @xmath278 in [ subsection symplectic cohomology ] , we have @xmath600 , however this may not hold for general @xmath576 .
moreover @xmath601 is a subalgebra containing @xmath602 and the unit @xmath603 , and there is a projection map @xmath604 .
when @xmath605 , @xmath606 .
there are mutually inverse pss - isomorphisms @xcite @xmath607 defined by counting spiked planes as follows .
parametrize @xmath608 by @xmath609 for @xmath610 .
consider solutions @xmath611 of floer s equation @xmath612 where @xmath613 $ ] is a decreasing cut - off function which equals @xmath49 for @xmath614 and equals @xmath194 for @xmath615 .
notice @xmath241 is @xmath564-holomorphic near the puncture @xmath616 .
the _ energy _ of @xmath241 is defined as @xmath617 .
if the energy is finite , then @xmath241 converges to a hamiltonian orbit of @xmath556 as @xmath618 , and @xmath241 extends @xmath564-holomorphically over @xmath616 . for a generic cycle @xmath619 , define @xmath620 by counting @xmath194-dimensional moduli spaces of such finite energy solutions @xmath241 with @xmath621 .
the definition of @xmath622 is similar , except parametrize @xmath623 by @xmath624 . in this case
, solutions @xmath241 converge to a hamiltonian orbit as @xmath447 , and @xmath241 is @xmath564-holomorphic for @xmath460 and extends @xmath564-holomorphically over the puncture @xmath616 . in @xcite
, it is proved that there are mutually inverse pss - isomorphisms also for @xmath0 , @xmath625 where @xmath575 is radial at infinity and converges fast to a constant at infinity ( so @xmath576 is globally bounded ) .
homotopying @xmath278 to @xmath576 , where @xmath278 is as in [ subsection symplectic cohomology ] , defines a continuation map @xmath626 which is an isomorphism and the composite @xmath627 equals the @xmath628 of [ subsection symplectic cohomology ] ( for details , see @xcite ) .
[ subsection fukaya category of e ] let @xmath0 be a monotone symplectic manifold conical at infinity , and let @xmath15 be a closed monotone symplectic manifold . denote their fukaya categories by @xmath629 , @xmath630 . recall that their objects are constructed using only _ closed _
monotone lagrangians as objects ( compare : @xmath270 includes also non - compact lagrangians conical at infinity ) .
we briefly discuss @xmath147 , but much of what we say applies also to @xmath92 .
suppose @xmath631 are lagrangians in @xmath0 ( we continue with the convention in [ subsection the role of monotonicity ] ) .
if the @xmath291 are pairwise transverse then no hamiltonians arise in the definition of @xmath632 : one just counts suitable holomorphic polygons in @xmath0 .
indeed , the analogues of all constructions for the compact category would correspond to allowing only weights @xmath633 and only @xmath516 terms , in particular @xmath634 is generated by the intersections @xmath635 . when the lagrangians do nt intersect transversely , one only uses @xmath572-small compactly supported hamiltonians to flow the lagrangians so that their intersections become transverse . for example , the ordinary lagrangian floer complexes are related to the wrapped versions via @xmath636 where @xmath637 is the time @xmath49 flow of @xmath638 , and where one must choose suitable almost complex structures ( for example @xcite explains this ) . when @xmath289 are not transverse , then @xmath634 is in fact _ defined _ as above using a @xmath572-small compactly supported @xmath278 so that @xmath639 are transverse ( how one sets up the @xmath91-category to keep track of the hamiltonian perturbation data is described in detail in seidel @xcite ) .
so the construction of @xmath147 is dramatically simpler than @xmath4 : one no longer needs most of the auxiliary data in [ subsection the auxiliary data for wrapped category ] .
the composition maps for the @xmath92-bimodule @xmath92 , @xmath641 [ figure ainfinity eqns for bimodule b ] for @xmath115 and any monotone lagrangians @xmath642 , are defined by counting the @xmath194-dimensional moduli spaces of maps described in figure [ figure bimodule compositions for b ] .
the domain is a disc @xmath643 with punctures on the boundary : two distinguished fixed boundary punctures denoted by dashes at @xmath644 ( @xmath181 receives the module input from the @xmath645 map , and @xmath380 receives the output ) , the small circular dots which lie on the two open arcs of @xmath646 denote boundary punctures which are free to move ( subject to being distinct and ordered on the respective arcs ) . in particular , @xmath647 .
the @xmath91-equations are a standard consequence of a bubbling analysis argument , by considering what happens when a subset of the free points converge together .
the types of bubbles that appear are sketched in figure [ figure ainfinity eqns for bimodule b ] .
define @xmath649 by counting maps as on the left in figure [ figure oc map ] .
more precisely : the domain is the disc @xmath650 , with one interior puncture and @xmath651 boundary punctures of which one is distinguished .
we fix the parametrization by fixing the distinguished boundary puncture at @xmath181 ( which will receive the @xmath652 input of @xmath653 ) , and the interior puncture at @xmath194 ( which will define the output in @xmath555 of @xmath653 ) .
the other @xmath654 boundary punctures can move freely subject to being ordered ( they receive the @xmath655 inputs of @xmath653 ) . then @xmath656 counts isolated maps @xmath657 which are holomorphic away from @xmath194 and which satisfy floer s equation near @xmath194 .
more precisely , near the puncture @xmath616 , we fix a cylindrical parametrization @xmath658\times s^1 $ ] where the @xmath659 is parametrized by the argument of @xmath51 .
then we require that @xmath241 satisfies the floer continuation equation @xmath660 on this cylindrical parametrization , where @xmath613 $ ] is a decreasing cut - off function which equals @xmath49 for @xmath614 and equals @xmath194 for @xmath615 . on the right in figure
[ figure oc map ] is an equivalent way of viewing the map @xmath40 : the domain is a half - infinite cylinder @xmath658\times s^1 $ ] with @xmath651 punctures on the circle @xmath661 , one of which is a distinguished puncture at @xmath662 , and we require the maps to satisfy the above floer equation .
the map @xmath40 is a chain map .
this can be seen by considering the possible degenerations of the @xmath49-dimensional moduli spaces : ( 1 ) if marked points on the boundary move together , we obtain @xmath663 or @xmath664 contributions ; and ( 2 ) if the floer trajectory breaks ( on the region where @xmath665 ) , we obtain contributions described by the floer differential @xmath666 .
thus we obtain a map on homology : @xmath667 the constructions for @xmath147 are analogous to those for @xmath92 .
in particular , since all lagrangians considered are closed , the maximum principle mentioned in [ subsection moduli space of pseudoholo maps in wrapped case ] ensures that all solutions lie in a compact subset determined by the lagrangians .
so the non - compactness of @xmath0 plays no role ( unlike the case of @xmath4 where lagrangians can be non - compact ) .
composing @xmath74 with the pss isomorphism @xmath670 of section [ subsection pss ] corresponds to gluing a spiked disk onto the centre of the disk . by homotopying the hamiltonian to zero for the glued configuration ,
one deduces that the composite map @xmath671 is defined by the counts of holomorphic polygons which intersect a given test cycle in @xmath0 ( we replaced the spike of the pss - plane by a cycle intersection condition ) .
we will now build an @xmath91-functor @xmath673 from the compact category to the wrapped category . on objects , it is just the inclusion of the closed lagrangians @xmath263 into the larger class of lagrangians that define @xmath270 . to simplify the following discussion , we will in fact assume that we only consider a full subcategory of @xmath147 generated by a subcollection of lagrangians which are pairwise transverse ( something which can always be achieved for a countable collection of lagrangians by flowing each one by a small generic compactly supported hamiltonian flow ) .
this is in order to avoid the tedious discussion of the auxiliary hamiltonian perturbation data which is needed in the construction of the compact category , mentioned in section [ subsection fukaya categories b and e ] . if one identifies @xmath674 , so the generators are the intersection points @xmath635 and the differential counts holomorphic strips , then one would like to define @xmath172 on morphisms as the inclusion @xmath675 \subset \bigoplus_{w=0}^{\infty } cf^*(l_i , l_j;wh)[\mathbf{q}],\ ] ] where one allows @xmath633 in the definition of @xmath394 in section [ definition wrapped floer complex ] .
unfortunately , in general one can not allow weight @xmath633 in the definition of @xmath394 when @xmath289 are both non - compact lagrangians , since then they may intersect non - transversely at infinity , and it would no longer be transparent how to consistently define non - compactly supported hamiltonian perturbation data .
however , allowing weight @xmath633 is not problematic when @xmath289 are compact lagrangians , which is all we need .
define the @xmath91-category @xmath676 to have the same objects @xmath270 and the same morphism spaces @xmath394 as @xmath4 except when @xmath289 are both compact lagrangians , in which case @xmath677 \oplus cw^*(l_i , l_j)$ ] .
the @xmath91-operations are defined by the same disc counts as in the construction of @xmath4 , with the novelty that if one of the boundary punctures carries an input from @xmath678 $ ] then no closed @xmath49-form @xmath236 is associated to that puncture ( see [ subsection the auxiliary data for wrapped category ] ) , so the equation @xmath679 turns into the @xmath227-holomorphic equation @xmath680 near that puncture .
[ lemma the inclusion of w0 ] the inclusion @xmath681 is a quasi - isomorphism . the @xmath91-functor @xmath407 is the identity on objects , the obvious inclusion on morphisms , and zero on higher order tensors . at the cohomology level
@xmath407 is an isomorphism since , as explained in section [ subsection the wrapped floer complex ] , the wrapped cohomology is a direct limit over the weights @xmath376 , so it does not matter if the weights start at @xmath633 or @xmath682 .
the _ acceleration functor _ is the @xmath91-functor given by the composite @xmath683 where the second map is a quasi - isomorphism which inverts the above @xmath407 up to homotopy , and the first map is the inclusion on objects and morphisms , and is zero on higher order tensors . as mentioned above , to simplify the discussion we assumed that in @xmath147 we only considered a countable collection of compact lagrangians which are pairwise transverse .
so in the above discussion this restriction would also apply to @xmath4 .
however , the inclusion of this somewhat smaller @xmath4 into the full @xmath4 is a quasi - equivalence , since it is essentially surjective : a compact lagrangian is isomorphic to a compactly supported hamiltonian isotopy of itself .
the analogous statement holds for @xmath42 .
so the acceleration functor @xmath684 can be defined by avoiding the discussion of hamiltonian perturbations altogether .
the acceleration functor determines an induced map @xmath685 in section [ section open - closed string map ] we will extend @xmath686 to a map @xmath687 , taking @xmath688 .
[ theorem acceleration commutative diagram ] there is a commutative diagram @xmath689 \ar@{->}_{\mathrm{oc}_e}[d ] & \mathrm{hh}_*({\mathcal{w}(e ) } ) \ar@{->}^{\mathrm{oc}_e}[d ] \\
qh^*(e ) \ar@{->}[r]^-{c^ * } & sh^*(e ) } \ ] ] we need to prove that the following diagram commutes , @xmath690 \ar@{->}_{\mathrm{oc}_e}[d ] & \mathrm{hh}_*(\mathcal{w}_0(e ) ) \ar@{->}_{\mathrm{oc}_e}[d ] \ar@{<-}^{f}[r ] & \mathrm{hh}_*({\mathcal{w}(e ) } ) \ar@{->}_{\mathrm{oc}_e}[d ] \\
qh^*(e ) \ar@{->}[r]^-{c^ * } & sh^*(e ) \ar@{<-}[r]^-{\mathrm{id } } & sh^*(e ) } \ ] ] the square on the right tautologically commutes since @xmath407 is the inclusion ( lemma [ lemma the inclusion of w0 ] ) .
we remark that the lower horizontal map in that square at the chain level is the inclusion @xmath691\oplus sc^*(e)$ ] , where @xmath692 $ ] .
so @xmath693 plays the role of the `` weight zero '' summand , where we extend @xmath292 and @xmath366 to the new summand to be respectively the usual singular boundary operator on lf - cycles and @xmath694 to be the pss map .
the map @xmath628 arises as an inclusion at the level of complexes , see lemma [ lemma sh is h of sc and c is inclusion ] .
so to prove that the left square commutes , we need to show that the extension @xmath687 coincides with @xmath686 on the image of the inclusion @xmath695 . to show that the two @xmath1 maps coincide it is enough to prove that no new solutions arise in the definition of @xmath1 restricted to compact lagrangians , when one increases the slope of @xmath278 on the conical end .
since we are now working at the cohomology level , we just need to check that the @xmath1 maps agree on a combination of generators and we may assume that we increase the slope of @xmath278 only outside of a large compact set of @xmath0 which contains the compact lagrangians involved in that combination of generators .
without loss of generality , we may assume that compact set is @xmath215 .
so we need to check that there are no solutions @xmath696 of @xmath679 bounding compact lagrangians @xmath697 which converge to a closed hamiltonian orbit outside @xmath215 . here
@xmath355 is a disc with negative punctures on the boundary and one positive puncture at the centre of the disc , and @xmath466 is a @xmath49-form on @xmath355 satisfying @xmath464 ( for a detailed description see section [ section open - closed string map ] ) .
it turns out that such solutions @xmath241 can not enter @xmath226 : the proof of this is essentially ( * ? ? ?
* lemma 7.2 ) , except that in our case @xmath355 also has a negative puncture at the centre where @xmath241 converges to a closed hamiltonian orbit in @xmath226 .
but the lemma of abouzaid - seidel can be adapted to this case ( see for instance the remarks in the `` no escape lemma '' section at the end of @xcite ) , provided that the action of closed hamiltonian orbits in @xmath226 is negative ( which can always be ensured , by choosing @xmath278 appropriately , see ( * ? ? ?
* section 8) ) .
we emphasize that the non - exactness of the symplectic form on @xmath215 is not an issue , since only the exact conical end @xmath226 is used in the argument .
recall the construction of @xmath596 in [ subsection symplectic cohomology ] .
we now define the analogue of [ subsection open closed map oc_b ] for @xmath4 at the chain level : @xmath698 .
we first need to define a map @xmath699 \otimes cf^*(l_{n-1},l_n;w_n h)[\mathbf{q}]\otimes \cdots \\ \qquad\qquad\qquad\qquad\qquad \cdots \otimes cf^*(l_0,l_1;w_1h)[\mathbf{q } ] \to cf^*(w_0h)[\mathbf{q } ] . \end{array}\ ] ] observe figure [ figure popsicle 2 ] .
consider discs @xmath355 as in [ subsection the auxiliary data for wrapped category ] for @xmath700 , except now @xmath315 is an interior puncture ( this will receive the hamiltonian orbit output coming from @xmath596 ) .
in addition , the last boundary puncture @xmath701 is distinguished and we denote by a dash in figure [ figure popsicle 2 ] ( this will receive the module input from @xmath350 ) .
we can fix the domain s parametrization by taking @xmath355 to be a punctured disc with @xmath702 located at @xmath703 . in [ subsection the auxiliary data for wrapped category ] , we modify @xmath459 to @xmath704\times s^1\to s$ ] ( asymptotic to @xmath315 ) , and modify @xmath423 to @xmath705 , @xmath706 , @xmath707 .
see the left of figure [ figure popsicle 2 ] .
analogously to the @xmath493 defined in [ subsection moduli space of pseudoholo maps in wrapped case ] , except for the aforementioned modifications , we obtain a moduli space of maps solving @xmath679 which we denote @xmath708 .
in particular , note that @xmath679 turns into @xmath709 via @xmath459 near @xmath315 .
to define @xmath710 we mimick the definition of the @xmath711 in [ subsection wrapped a infinity structure ] .
first we determine the @xmath712 coefficient of the image .
an isolated @xmath713 will contribute @xmath518 to @xmath714 where @xmath520 for @xmath521 , @xmath522 for @xmath523 , and it will contribute @xmath194 otherwise ( for signs mimick ( * ? ? ?
* sec.3h ) ) . the @xmath715 coefficient of the image is determined by the requirement that the map @xmath514 respects the @xmath383 operator .
+ * example . *
an isolated solution for the left configuration in figure [ figure popsicle 2 ] contributes @xmath716 to @xmath717 , using asymptotic data @xmath718 at @xmath719 .
+ finally , sum up the @xmath710 as @xmath529 vary : @xmath720 @xmath74 is a chain map , so it induces @xmath199 .
the proof that @xmath74 is a chain map is similar to the proof of the @xmath91-equations for @xmath721 ( see remarks 3 and 4 in [ subsection wrapped a infinity structure ] ) .
we illustrate the argument in a concrete example . in figure
[ figure popsicle 2 ] consider the left configuration : if the preferred point @xmath430 moves towards @xmath437 , the limit contributes to @xmath722 which is a part of @xmath723 , and this corresponds to the right - most picture in figure [ figure popsicle 2 ] .
if @xmath430 moves towards @xmath315 , in the limit a @xmath310-punctured sphere will bubble off with the preferred point @xmath430 lying on the geodesic joining the two punctures ( see the broken configuration in the middle of figure [ figure popsicle 2 ] ) . by the construction of the parametrizations near the ends , this @xmath310-punctured sphere can be identified with a cylinder @xmath724 with the preferred point @xmath430 lying at @xmath662 on the geodesic line @xmath725 .
so this breaking contributes to @xmath726 .
consider the @xmath49-dimensional components of the compactified moduli space @xmath708 .
the oriented sum of the possible boundary degenerations sums up to zero , and corresponds to the equation @xmath727 where @xmath728 is the bar differential .
we warn the reader that @xmath199 is not a ring homomorphism ( and can not be since @xmath729 does not admit a product , unlike @xmath730 ) .
however , we now prove that @xmath199 respects a certain @xmath33-module structure on @xmath729 , and for the compact category @xmath731 respects a certain @xmath587-module structure , and that the two actions are compatible with the acceleration diagram of theorem [ theorem acceleration commutative diagram ] .
the module structure was observed independently , in the exact case , by ganatra @xcite , and the underlying @xmath14-module structure is used in forthcoming work by
abouzaid - fukaya - oh - ohta - ono @xcite in the closed case .
[ theorem oc is a module map ] 1 .
@xmath732 is a @xmath587-module ; + @xmath587 is a @xmath587-module via the quantum product ; + + 2 .
@xmath729 is an @xmath33-module ; + @xmath33 is an @xmath33-module via the pair - of - pants product ; + + 3 .
@xmath729 , @xmath33 are @xmath587-modules via @xmath733 ; + + and the acceleration diagram of theorem [ theorem acceleration commutative diagram ] respects the @xmath587-actions .
@xmath74 splits up according to generalized eigensummands of the @xmath586-action , @xmath734 where @xmath347 are the eigenvalues of the quantum product by @xmath586 on @xmath587 , and @xmath735 is the fukaya category generated by lagrangians with @xmath346 ( lemma [ lemma fukaya cats split according to evalues ] ) . 5 . similarly for the wrapped category , @xmath736 6 .
the acceleration diagram therefore splits up according to the eigenvalues @xmath347 of @xmath586 : @xmath737 \ar@{->}_{\mathrm{oc}_e=\mathrm{oc}_{\lambda}}[d ] & \mathrm{hh}_*(\mathcal{w}_{\lambda}(e ) ) \ar@{->}^{\mathrm{oc}_e=\mathrm{oc}_{\lambda}}[d ] \\
qh^*(e)_{\lambda } \ar@{->}[r]^-{c^*_{\lambda } } & sh^*(e)_{\lambda } } \ ] ] remark : ( 1 ) and ( 4 ) also hold if we replace @xmath0 by a _ closed _ monotone symplectic manifold @xmath15 . *
* the @xmath587-action on @xmath342 . * + the action @xmath738 is described in more detail in [ subsection deformed quilt morphism ] ( and in great detail in auroux ( * ? ? ?
* sec.6.1 ) ) .
observe the first picture in figure [ figure oc_qhaction0 ] .
consider the disc @xmath413 with two fixed boundary punctures at @xmath739 ( respectively receiving the output and the input chain from @xmath339 ) , with an interior marked point @xmath740 ( which receives as input the locally finite cycle @xmath741 from @xmath693 ) .
we fixed the @xmath742-parametrization of the disc by fixing the punctures at @xmath743 and by fixing at @xmath744 the marked point @xmath745 which is constrained to lie on the hyperbolic geodesic connecting the two punctures @xmath743 .
_ then , for a generator @xmath746 of @xmath339 , one defines @xmath747 , summing over generators @xmath748 of @xmath339 , where the discs @xmath749 are isolated among solutions which intersect @xmath741 at @xmath750 and converge to @xmath751 at @xmath752 .
it should be clear from the context what `` solutions '' we mean : here it means @xmath241 is @xmath227-holomorphic , it satisfies the obvious lagrangian boundary conditions @xmath753 , and the asymptotic conditions @xmath751 at @xmath743 , and @xmath754 .
we now clarify more precisely the condition at @xmath745 .
consider the moduli space @xmath755 of all such solutions lying in a fixed homotopy class @xmath756 but without imposing the condition @xmath757 .
now fix a cohomology class @xmath758 \in h^*(e)$ ] , and within it we pick a representative locally finite cycle @xmath741 which is transverse to the evaluation map @xmath759 at @xmath760 . then @xmath761 is the space of all solutions @xmath241 described above satisfying the condition @xmath757 . call @xmath762 the count of the isolated points ( i.e. the zero - dimensional components ) of @xmath761 with appropriate orientation signs . define @xmath763 , and extend @xmath221 so that it is @xmath8-linear in @xmath339 . _
technical remark .
two possible degenerations of a @xmath49-family of such solutions involve a disc bubbling off at @xmath380 or at @xmath181 , which respectively correspond to the floer differential on the output and the input , and the count of these degenerations would show that the action of @xmath741 is a chain map ( see the second and third picture in figure [ figure oc_qhaction0 ] ) .
there is however also another degeneration , when a disc bubbles off at a boundary point @xmath764 different from @xmath743 .
this corresponds to a higher - order map which counts solutions as above but with an extra boundary marked point at @xmath764 , whose input chain at @xmath764 is @xmath349 .
but our standing assumptions in [ subsection the role of monotonicity ] ensure that @xmath349 is a multiple of @xmath321 $ ] or @xmath765 $ ] ( depending on which arc of @xmath766 the point @xmath764 lies ) .
so these solutions are never isolated since the condition @xmath767 $ ] or @xmath765 $ ] at @xmath764 is automatic since it is the lagrangian boundary condition . for a more detailed discussion see auroux ( * ? ? ?
* lemma 6.5 ) .
analogous remarks apply to the actions that we will define below , which we therefore omit
. we will also omit the discussion that the module actions we define on cohomology only depend on the class @xmath758\in qh^*(e)$ ] rather than on the chosen representative lf - cycle @xmath741 , since this is a fairly standard argument .
_ we now prove that , on cohomology , @xmath221 is a module action .
instead of having just one interior marked point @xmath744 , consider now having two distinct interior marked points @xmath740 and another @xmath768 which is allowed to vary on the geodesic joining @xmath745 to @xmath181 .
consider a @xmath49-parameter family of solutions @xmath241 which intersect lf - cycles @xmath769 at the respective marked points ( the asymptotic data @xmath751 being fixed ) .
apart from the usual disc bubbling on the boundary , there are two new degenerations that occur when @xmath770 converges to @xmath744 or to @xmath181 . in the first case , as sketched in the fourth picture in figure [ figure oc_qhaction0 ] , a sphere bubble forms at @xmath194 which carries the two marked points , corresponding to the quantum product @xmath771 ( of which the lowest order @xmath191 term , namely when the sphere bubble is constant , corresponds to the classical interesection product of the lf - cycles @xmath769 ) . in the second case
the bubbling corresponds to composing the actions of @xmath741 and @xmath772 .
thus the count of these boundary degenerations yields the equation @xmath773 on cohomology , where @xmath774 denotes the action . + * step 2 .
* * the @xmath587-action on @xmath775 . * + if we replace @xmath339 by @xmath776 in the above discussion ( where @xmath753 are now objects in the wrapped fukaya category ) , then as in the construction of the wrapped category in [ subsection the auxiliary data for wrapped category ] and [ subsection moduli space of pseudoholo maps in wrapped case ] , there will now be additional auxiliary data , in particular a hamiltonian and preferred points , which determines a @xmath49-form @xmath466 on @xmath777 ; and in this context a `` solution '' will mean that @xmath241 satisfies the floer equation @xmath679 ; and @xmath778 is now defined as in [ subsection moduli space of pseudoholo maps in wrapped case ] .
there is however one crucial complication : in the proof that @xmath221 is a module action , when @xmath770 converges to @xmath745 we want the degeneration to give rise to a _
holomorphic _ sphere bubble , so that standard gluing theorems for holomorphic maps apply .
it is therefore necessary to ensure that the floer equation @xmath679 turns into the @xmath227-holomorphic equation @xmath680 near @xmath745 by constructing @xmath466 to be zero near @xmath760 , and similarly we need to ensure that @xmath779 near both @xmath780 when we prove that @xmath221 is a module action .
so we need to define more precisely the auxiliary data for the domain @xmath781 of the solutions counted by @xmath782 where , recall , @xmath783 $ ] .
recall the auxiliary data described in [ subsection the auxiliary data for wrapped category ] . in this case
, we have 1 . a negative strip - like end @xmath459 for the puncture @xmath784 ; 2 . a positive strip - like end @xmath785 for the puncture @xmath786 ; 3 . only one closed @xmath49-form @xmath787 on @xmath355 associated with the end @xmath786 , which pulls back to @xmath451 near @xmath743 ; 4 .
the labels @xmath788 for all @xmath789 , so the preferred points @xmath423 ( if any are present ) all lie on the geodesic joining @xmath784 and @xmath786 , and one can ensure that the sub - closed forms @xmath456 satisfy @xmath458 except in a small neighbourhood @xmath790 of @xmath423 . to ensure @xmath779 at @xmath744 , we would need to impose @xmath791 and @xmath792 near @xmath745 , since @xmath793 where the weight @xmath794 is specified by the input at @xmath746 coming from @xmath795 .
if the @xmath796 do not contain @xmath745 , then one can always ensure @xmath797 near @xmath745 by adding suitable exact forms supported near @xmath745 to @xmath798 , because @xmath798 are closed near @xmath745 .
the issue is that one can not do this consistently in families , since the preferred point @xmath423 may converge to @xmath745 . to fix this issue ,
observe the first picture in figure [ figure oc_qhsupport0 ] : we simply deform the geodesic joining @xmath799 in a fixed small neighbourhood of @xmath744 , say a hyperbolic disc @xmath800 of radius @xmath801 centred at @xmath745 , so that the deformed geodesic avoids @xmath802 by moving along the clockwise arc of @xmath803 from @xmath804 to @xmath801 .
moreover , we impose that the support of @xmath805 must be contained within a hyperbolic disc @xmath806 of radius @xmath807 centred at the preferred point @xmath423 .
this ensures that @xmath458 in @xmath808 , so we can always construct @xmath798 to be zero on @xmath808 . as already mentioned in remark 1 of [ subsection the auxiliary data for wrapped category ] ,
deforming the geodesics away from the ends is not an issue and is anyway a necessity in the gluing arguments .
we could equivalently have said that we keep the geodesic undeformed , but we stipulate a canonical way of determining where @xmath790 must lie with respect to the preferred point @xmath423 : this point of view is illustrated in the second picture of figure [ figure oc_qhsupport0 ] .
namely , if @xmath423 lies outside @xmath809 then the centre @xmath810 of the hyperbolic disc of radius @xmath807 which contains @xmath790 equals the preferred point : @xmath811 , otherwise if @xmath812 then define @xmath813 where : @xmath814 is the upper half - disc which bounds the geodesic @xmath815 and the arc of @xmath415 carrying the boundary condition @xmath22 , and @xmath816 is the hyperbolic geodesic through @xmath423 orthogonal to the geodesic @xmath815 . in this point of view , the preferred point does not play the role of the centre of the support disc @xmath817 , but instead plays the role of fixing the strip - like end parametrization as in [ subsection the auxiliary data for wrapped category ] which identifies @xmath818 with the strip @xmath819 $ ] so that @xmath820 , @xmath821 , @xmath822 and @xmath823 . in particular , this identification determines the image of @xmath802 in @xmath222 , and the image of @xmath816 is a vertical segment @xmath824 in @xmath222 .
the @xmath49-parameter family of automorphisms of @xmath777 which moves the preferred point along the geodesic @xmath825 corresponds to @xmath240-translation on @xmath222 , so the images of @xmath826 in @xmath222 will simply translate in @xmath222 if we vary the preferred point in @xmath355 .
therefore , by this point of view , it becomes clear that one can build the forms @xmath456 consistently in families , since the problem reduces to constructing families of model forms @xmath456 on @xmath222 .
the definition of @xmath827 \to cw^{*-\textrm{codim(c)}}(l , l')$ ] now follows the usual routine : the @xmath516-coefficient of the image in @xmath828 counts the isolated floer solutions which do not involve preferred points ; the @xmath516-coefficient in @xmath829 counts the isolated solutions involving exactly one preferred point ( @xmath456 bumps up the output weight by @xmath49 ) ; as usual , by symmetry arguments ( * ? ? ?
* lemma 3.7 ) , the counts of floer solutions which involve more than one preferred point on the geodesic cancel out ; and finally we extend @xmath830 by requiring it to be @xmath383-equivariant .
the symmetry argument relies on showing that the transposition of two preferred points lying on the same geodesic switches the orientation sign with which the isolated solution is counted ( * ? ? ?
* lemma 9.1 ) .
this result works generally for moduli spaces of punctured discs provided that one shows that the geodesic / preferred point decorations can be constructed consistently over the moduli space , which is what we have discussed above .
_ as in the previous technical remark , the degenerations of @xmath49-families of such solutions which prove that @xmath830 is a chain map involve disc bubbling at @xmath743 .
the novelty in the wrapped setup is that , if a preferred point is present , the preferred point may also bubble off by converging to @xmath743 . but
this is precisely what we expect since the floer differential @xmath336 for @xmath776 in definition [ definition wrapped floer complex ] counts discs with possibly a preferred point , and we refer the interested reader to the remarks in [ subsection wrapped a infinity structure ] which explain how the condition of @xmath383-equivariance accurately reflects these bubbling phenomena .
before we can mimick the proof from step 1 that @xmath830 is a module map on cohomology , we need to explain how to make @xmath779 near both @xmath744 and @xmath768 .
we make @xmath779 on @xmath808 as before , by ensuring that the preferred point ( if present ) lies outside @xmath802 .
recall that this means we deformed the geodesic @xmath815 so that it avoids @xmath802 by going around @xmath803 clockwise . similarly , as shown in figure [ figure oc_qhsupport00 ] , we now make a further deformation of the ( deformed ) geodesic so that it avoids @xmath831 by going around @xmath832 clockwise ( we emphasize that these are hyperbolic discs , so from the euclidean point of view @xmath833 shrinks to a point if @xmath770 converges to @xmath181 ) . then , as before , we require that the preferred points @xmath423 lie on this new deformed geodesic , and that the supports @xmath790 of @xmath805 lie inside @xmath806 .
thus , @xmath798 are closed on @xmath834 so we can always construct @xmath798 to be zero there ( by adding exact forms supported there ) , so @xmath779 there .
the proof that @xmath835 is a module map now follows by the same argument as in step 1 except we are using the auxiliary data described above .
+ * step 3 . * * the @xmath587-action on @xmath732 . *
+ consider the compact category @xmath836 ( so in the absence of hamiltonians ) , and consider a generator of the @xmath127-bimodule @xmath127 , @xmath837 in the notation from [ subsection grading conventions]-[subsection bimodules ] , with @xmath838 for closed lagrangians @xmath839 .
for a locally finite cycle @xmath840 , define @xmath841 by : @xmath842 summing over all obvious choices of @xmath104 , where we use the sign notation @xmath843 defined in [ subsection grading conventions ] , where @xmath844 is the cohomological degree , and where @xmath845 is the count of discs @xmath846 illustrated in the first picture of figure [ figure oc_qhaction1 ] .
the domain @xmath355 in that picture is a disc @xmath413 with @xmath847 positive boundary punctures , of which the dash at @xmath181 marks the input for the module element ; one negative boundary puncture at @xmath380 indicated by a dash ; and one interior marked point at @xmath740 ( as in step 1 , the choices @xmath848 fix the parametrization of the disc ) .
the definition of @xmath849 is analogous to that of @xmath850 except now there is additionally an interior marked point @xmath740 and we impose that floer solutions @xmath241 satisfy @xmath851 in the same sense as explained in step 1 .
we emphasize that the boundary punctures which carry the inputs @xmath852 and @xmath853 are free to move in an order - preserving way along the two open arcs joining @xmath739 .
the analogue here of the technical remark from step 1 is illustrated in the second and third picture in figure [ figure oc_qhaction1 ] .
these show two typical degenerations that can occur in a @xmath49-family of such solutions @xmath241 , when a subset of the boundary punctures converge towards @xmath181 and @xmath380 respectively .
observe that the bubbled discs which do not carry the marked point @xmath854 are precisely the discs counted by the bimodule homomorphisms @xmath855 .
if a subset of the boundary punctures converge to a boundary point different from @xmath743 , then a bubble appears which is counted by @xmath856 .
these degenerations prove that the action of @xmath741 is a morphism of @xmath127-bimodules @xmath857 ( see the definition in [ subsection bimodules ] ) .
therefore , @xmath858 induces a chain map on the hochschild complex by lemma [ lemma functoriality of hh ] .
thus , on homology , we obtain : @xmath859 we omit the standard argument that this map only depends on the cohomology class @xmath758\in qh^*(e)$ ] .
to prove that this is a @xmath587-module action , we mimick step 1 using two ordered interior marked points @xmath860 .
if @xmath770 converges to @xmath745 in a @xmath49-family , then as usual a bubble counted by the quantum product appears as in the last picture in figure [ figure oc_qhaction1 ] .
if @xmath770 instead converges to @xmath181 , then a subset of the boundary punctures may converge together with @xmath770 to @xmath181 , so the bubble that appears at @xmath181 is of the same type as the first picture in figure [ figure oc_qhaction1 ] with @xmath770 in place of @xmath745 .
so these degenerations account for all the contributions of the composite of the actions of @xmath741 and @xmath772 . +
* step 4 . *
* the @xmath587-action on @xmath729 . * + for the wrapped category @xmath861 ( so when hamiltonians are present ) , we have the same complication as in step 2 : we need to ensure that the auxiliary data can be constructed so that @xmath779 near @xmath745 ( and near @xmath770 in the proof of the module action ) , so that the floer equation @xmath679 turns into the @xmath227-holomorphic equation near @xmath854
. observe figure [ figure oc_qhaction2 ] : this is the analogue of figure [ figure oc_qhaction1 ] , now decorated with a typical configuration of geodesics and preferred points @xmath423 .
the auxiliary data is now be constructed as in [ subsection the auxiliary data for wrapped category ] , but we need to tweak the construction just like in step 2 by deforming the geodesics so that they avoid the disc @xmath800 .
the only novelty , is that there are two different ways to deform the geodesics : for the geodesics which lie in the upper half - disc @xmath814 ( which bounds the geodesic @xmath815 and the arc of @xmath415 carrying the boundary conditions @xmath862 ) we will deform the geodesics to avoid @xmath802 by moving clockwise along @xmath803 ; whereas for the geodesics in the lower half - disc @xmath863 we will instead move anticlockwise along @xmath803 . for the geodesic connecting @xmath380 to @xmath181 one needs to make a choice amongst these two possibilities : we will do the same as in step 2 , so we go around @xmath803 clockwise in this case .
although it seems that we are breaking the symmetry of the problem , this is only apparent because the module input at @xmath181 has already divided the set of positive boundary punctures into two distinct subsets , depending on whether they lie in @xmath814 or in @xmath864 , so we are only stipulating that @xmath181 will belong to the former subset .
the upshot , just as in step 2 , is that the auxiliary forms @xmath865 can all be consistently constructed so that they vanish on @xmath808
. now consider the wrapped category @xmath866 and a generator of the @xmath127-bimodule @xmath127 , @xmath867 in the notation from [ subsection grading conventions]-[subsection bimodules ] , with @xmath868 for ( possibly non - compact ) lagrangians @xmath839 , where @xmath869 , @xmath870 , @xmath871 for some weights @xmath872 .
for an lf - cycle @xmath840 , define the action @xmath873 by @xmath874 where @xmath875 takes into account the @xmath380 grading of @xmath379 , and the @xmath516-coefficient of @xmath876\ ] ] counts the isolated solutions @xmath846 of floer s equation @xmath679 such as the first picture in figure [ figure oc_qhaction2 ] ( except the geodesics have been deformed as explained above , which ensured that the @xmath865 are zero on @xmath808 and so the same is true for @xmath877 ) . in particular , the weight @xmath283 above is @xmath878 , where @xmath879 is the number of preferred points . as usual , we extend @xmath880 so that it is @xmath383-equivariant and @xmath8-multi - linear .
the proof that this action is a bimodule morphism is analogous to step 3 , and examples of the relevant degenerations of @xmath49-families are illustrated in the second and third picture in figure [ figure oc_qhaction2 ] : so observe that the bubbled discs which do not carry the marked point @xmath854 are precisely the discs counted by the bimodule homomorphisms @xmath881 .
as in step 3 , by lemma [ lemma functoriality of hh ] , @xmath858 induces a chain map on the hochschild complex .
the proof that this is a @xmath587-module action on homology is analogous to step 3 ( and an example of a relevant degeneration arising in the proof is illustrated in the last picture of figure [ figure oc_qhaction2 ] ) .
+ * step 5 . * * the @xmath587-action on @xmath33 . * + we will now construct the @xmath587-action @xmath882 .
recall @xmath883 $ ] from section [ subsection symplectic cohomology ] , which we already used in [ subsection open - closed with q coefficients ] . in analogy with the wrapped category , by `` solutions '' @xmath884 we now mean solutions to floer s equation @xmath679 where @xmath466 depends on the auxiliary data .
let us recall this data in the case of the differential @xmath336 .
the domain @xmath724 can be viewed as the sphere @xmath885 together with a negative puncture , a positive puncture , and a geodesic connecting the punctures corresponding to the line @xmath725 .
the auxiliary data is analogous to [ subsection the auxiliary data for wrapped category ] , except the ends are cylindrical rather than strip - like since we replace the time - interval @xmath284 $ ] by @xmath659 . in particular , we have a closed form @xmath236 which pulls back to @xmath451 in the cylindrical ends near the punctures ; and a preferred point @xmath423 ( if present ) on the geodesic @xmath725 corresponds to the translation @xmath886 , @xmath887 .
these translations allow one to consistently build @xmath49-forms @xmath456 with @xmath457 , and @xmath888 outside of a small neighbourhood of @xmath423 , analogously to the construction of the wrapped category .
to define the action @xmath889 of an lf - cycle @xmath890 we mimick step 1 : on the domain we just described above , insert a new marked point @xmath891 on the geodesic @xmath725 ( as usual , we chose the specific point @xmath892 in order to fix the domain s parametrization ) .
now require solutions @xmath241 to satisfy @xmath757 ( in the sense explained in step 1 ) .
just as in step 2 , we need to ensure that @xmath779 near @xmath745 : this can be achieved by deforming the geodesics as in step 2 so that they avoid a hyperbolic disc @xmath893 by moving clockwise along @xmath803 , and one ensures @xmath894 and @xmath792 on @xmath808 .
the proof that this is a chain map follows as usual : in a @xmath49-family of solutions , @xmath745 may converge to one of the two punctures which gives rise to a new twice - punctured sphere bubble carrying @xmath745 ; whereas the twice - punctured sphere which does not carry @xmath745 corresponds to the differential @xmath336 ( which involves either the floer differential @xmath292 if there are no preferred points , or the continuation map @xmath366 if there is a preferred point , as defined in section [ subsection symplectic cohomology ] ) . the proof that this is a module action on cohomology is analogous to the proof in step 2 , by placing two interior marked points @xmath895 on the geodesic @xmath725 .
one deforms geodesics so that they avoid @xmath802 , @xmath831 by going around their boundary clockwise , so that all auxiliary forms vanish on @xmath808 and @xmath896 . then in a @xmath49-family , if @xmath770 converges to @xmath745 it will give rise to a holomorphic bubble ( since @xmath779 near @xmath745 ) counted by the quantum product , whereas if @xmath770 converges to @xmath745 the twice - punctured sphere breaks up into two such spheres corresponding to the composite of the actions of @xmath741 and @xmath772 . + * step 6 . * * the @xmath33-action on @xmath33 . * + the pair - of - pants product @xmath897 is constructed in analogy with @xmath898 , except the domain is now a @xmath548-punctured sphere ( that is , a pair - of - pants ) , rather than a @xmath548-punctured disc .
two of the punctures are positive : they receive inputs from @xmath899 , cf^*(w_2h)[\mathbf{q}]$ ] , whereas the negative puncture receives the output from @xmath900 $ ] .
the count of solutions on these domains ( with auxiliary data , such as preferred points ) defines the contribution @xmath901 \otimes cf^*(w_1h)[\mathbf{q } ] \to cf^*(w_0h)$ ] to @xmath897 . as a comparison with the previous constructions
, we could equivalently have said that we insert a new positive interior puncture @xmath745 on the twice - punctured sphere .
so rather than an interior marked point @xmath745 as in steps 1 - 5 , we now have an interior _ puncture _ together with a choice of positive cylindrical - end near @xmath745 ; and rather than requiring that @xmath779 near @xmath745 , we now require that @xmath902 in the cylindrical - end near @xmath745 so that @xmath679 turns into floer s equation for the hamiltonian @xmath903 ( rather than the @xmath227-holomorphic equation ) . on cohomology ,
the pair - of - pants defines an action of @xmath33 on itself by multiplication .
we need to check that the action of @xmath587 on @xmath33 defined in step 5 agrees with this @xmath33-action via @xmath195 . by lemma [ lemma sh is h of sc and c is inclusion ]
, @xmath628 arises from a composition : the pss map @xmath904 from [ subsection pss ] composed with the inclusion of the subcomplex @xmath595 . at the level of cohomology , this composition corresponds to the gluing of a spiked plane ( see [ subsection pss ] ) onto the pair of pants .
this gluing gives rise to a cylinder with a spike ending at an interior marked point of the cylinder .
observe that the spike is just an intersection condition with an lf - cycle at the marked point , except it is phrased using morse chains rather than singular chains .
this concludes the proof that the two actions agree via @xmath628 . for more details on this gluing argument and on the pair - of - pants product , see @xcite . + * step 7 . * * the @xmath33-action on @xmath729 . *
+ to upgrade the @xmath587-action on @xmath729 from step 4 to an @xmath33-action , we make the following changes : * the interior marked point @xmath905 in step 4 gets replaced by a positive interior puncture @xmath906 , in analogy with the discussion in step 6 : so it comes with a choice of a cylindrical end , namely a holomorphic embedding @xmath907 converging to @xmath745 as @xmath908 ; * associated to @xmath745 there is a weight @xmath909 and a closed form @xmath910 which pulls back to @xmath451 in the strip - like end near @xmath380 and in the cylindrical end near @xmath745 .
the asymptotic condition for solutions @xmath696 at @xmath745 is now a generator @xmath911 ; * a new type of geodesic may decorate the domain , as part of the auxiliary data : namely the geodesic @xmath912 connecting the module output at @xmath380 with the puncture @xmath744 . a preferred point on
this geodesic specifies a cylindrical parametrization , more precisely : it specifies a unique holomorphic map @xmath913 which extends to a smooth isomorphism @xmath914 on the compactifications such that @xmath915 , @xmath916 ( where @xmath917 are the circles compactifying the cylinder @xmath918 ) , and such that @xmath919 and @xmath920 .
although , as usual , we will often just abusively call @xmath423 the preferred point . recall that these isomorphisms are used to consistently build the auxiliary sub - closed forms @xmath456 , with @xmath805 supported near the preferred point . * just as discussed in step 4 , we need to deform all the geodesics so that the auxiliary forms ( apart from those involving the puncture @xmath745 ) vanish on the hyperbolic disc @xmath921 . more precisely :
the geodesics connecting @xmath380 to a boundary puncture in the upper half - disc @xmath814 are deformed to avoid @xmath800 by going around @xmath803 clockwise ( this includes , by convention , the geodesic @xmath815 connecting the module input / output ) ; those connecting @xmath380 to a boundary puncture in the lower half - disc @xmath864 go around @xmath803 anticlockwise .
the geodesic described in the previous paragraph ( if present ) stays undeformed , since in this case by construction @xmath792 near @xmath745 and so @xmath922 near @xmath745 ( since all other auxiliary forms vanish on @xmath808 ) .
the remainder of the construction of the @xmath4-bimodule action of a cycle @xmath923 on @xmath4 is completely analogous to step 4 , so we obtain a chain map on the hochschild complex , and thus an @xmath33-action on @xmath729 . to prove that it is a module action on cohomology , we now consider two positive interior punctures @xmath744 and @xmath768 ( rather than interior marked points , as in step 4 ) .
just as in step 4 , all geodesics involving pairs of boundary punctures need to be deformed so that they avoid @xmath809 and @xmath924 so that we can ensure that the relevant auxiliary forms vanish on @xmath808 and @xmath896 ( and as usual , the clockwise / anticlockwise convention for the deformation is as above ) .
the novelty here is the possibile presence of the geodesic @xmath925 connecting @xmath380 to @xmath926 . this geodesic @xmath927 does not need to be deformed in @xmath924 since the auxiliary forms for @xmath772 are constructed so that they pull - back correctly in the cylindrical - end near @xmath770 ( and we just ensured that all other auxiliary forms vanish near @xmath770 ) .
however , we need to deform @xmath927 near @xmath745 so that the preferred point @xmath423 ( if present ) on @xmath927 will not collide with @xmath745 in a @xmath49-family ( unless also @xmath770 converges to @xmath745 ) . we need to consistently deform @xmath927 without changing @xmath927 near the ends , so that it avoids a neighbourhood of @xmath745 : but to do this ,
we need to allow the diameter of the neighbourhood to depend on the relative distance between @xmath780 .
observe figure [ figure oc_qhsupport000 ] .
define @xmath928 where @xmath929 refers to the hyperbolic distance .
then we deform @xmath927 so that it avoids @xmath930 by going clockwise around @xmath931 , and we require that @xmath790 lies inside the disc @xmath932 so that @xmath458 on @xmath933 , and thus we can always construct @xmath456 so that @xmath792 on @xmath933 . in other words , for the geodesic @xmath927 instead of using @xmath809 we now use @xmath930 for a radius @xmath934 which shrinks to zero as @xmath770 approaches @xmath745 , in particular we do not deform @xmath927 near the ends .
figure [ figure oc_qhsupport000 ] compares how we deform @xmath927 ( and we indicate a typical preferred point @xmath423 and the support @xmath790 ) versus how we deform the geodesics which connect @xmath380 to a boundary puncture .
note that if @xmath770 lies outside @xmath802 then @xmath935 : so if @xmath770 converges to @xmath181 then the deformed geodesic @xmath927 converges to the usual deformation of the geodesic connecting @xmath380 to @xmath181 .
the remainder of the proof of the module action follows as in step 4 : in a @xmath49-family of solutions , @xmath770 may converge to the puncture @xmath745 or to @xmath49 . in the first case , a three - punctured sphere carrying @xmath780 appears , corresponding to the pair - of - pants product @xmath936 ( notice that preferred geodesics , if present , connect @xmath780 to the output puncture , as it should be ) . in the second case , a disc bubble appears carrying the puncture @xmath770 , and the broken configuration corresponds to composing the actions of @xmath741 and of @xmath772 .
the fact that , on cohomology , the @xmath587-action on @xmath729 from step 4 agrees with the @xmath33-action via @xmath733 is proved as in step 6 .
namely , we glue a pss - plane onto the puncture @xmath745 , thereby `` filling it '' and replacing it with an interior marked point @xmath745 as in step 4 , the spike of the pss - plane being the lf - cycle intersection condition . + * step 8 . * * the @xmath1-maps are compatible with the actions .
* + we will prove that @xmath937 preserves the @xmath33-action .
the proof that the @xmath1-map for the compact category preserves the @xmath587-action is analogous : just ignore all the decorations on the domains .
consider the domain of the solutions which define @xmath1 : recall this is a disc with positive boundary punctures , of which one is distinguished ( receiving the module input ) , and with one negative interior puncture ( receiving the @xmath596-output ) .
we fix the parametrization of the domain by placing the distinguished boundary puncture at @xmath49 and the interior puncture at @xmath194 .
now introduce a new positive interior puncture @xmath854 which is free to move on the geodesic @xmath938 connecting the punctures @xmath194 and @xmath49 .
a typical domain with decorations is shown in the first picture of figure [ figure oc_qhaction3 ] .
the solutions @xmath241 of @xmath679 will be required to satisfy floer s equation near @xmath745 just as in step 6 .
recall that , as in the construction of the @xmath1-map , the preferred geodesics connect the output interior puncture at @xmath194 with a boundary puncture .
the preferred point specifies a unique holomorphic map @xmath939 extending to an isomorphism on the compactifications such that @xmath940 , @xmath941 , @xmath942 , @xmath920 . as in step 4
, we need to ensure that the auxiliary forms ( except those related to @xmath745 ) vanish on @xmath808 , so as usual we need to deform the geodesics so as to avoid the hyperbolic disc @xmath800 by going around it ( with the usual clockwise / anticlockwise convention depending on whether the geodesic lies in the upper or lower half - disc of @xmath943 we comment further on this below ) .
the novelty here is : * we emphasize that the interior puncture @xmath745 is now allowed to move freely on @xmath938 ( since we used the punctures @xmath944 to fix the parametrization of the disc ) ; * a new type of geodesic can arise , see the first picture in figure [ figure oc_qhsupport0000 ] : the geodesic @xmath945 connecting the output interior puncture @xmath194 to the interior puncture @xmath745 .
this geodesic does not need to be deformed , since the relevant auxiliary forms for this geodesic are constructed so that they pull - back correctly in the cylindrical - ends near @xmath194 and @xmath745 ; * if @xmath745 is sufficiently close to @xmath194 , then @xmath800 contains @xmath194 .
so in this case we can not deform geodesics to avoid @xmath802 since then they would never reach @xmath194 .
so instead we apply a similar trick to step 7 ( figure [ figure oc_qhsupport000 ] ) : instead of @xmath800 we always use the disc @xmath930 with radius @xmath946 when deforming geodesics ( we will explain later why this time we square the hyperbolic distance between the two punctures ) .
notice @xmath930 shrinks to zero if @xmath745 approaches the puncture @xmath194 , whereas it coincides with @xmath802 whenever @xmath745 lies outside of @xmath947 ( and , since it is a hyperbolic disc , if @xmath745 approaches @xmath181 it will shrink to a point when viewed as a euclidean disc ) .
observe the second and third pictures in figure [ figure oc_qhsupport0000 ] : we deform ( if necessary ) any geodesic which connects @xmath194 to a boundary puncture so that it avoids the disc @xmath930 by moving along @xmath931 ( with the usual clockwise / anticlockwise convention , in particular for the geodesic @xmath938 from @xmath194 to @xmath181 we stipulate that the deformed geodesic goes around @xmath931 clockwise : this does not break any symmetry in the boundary punctures since @xmath181 already broke that symmetry by being the distinguished boundary puncture which receives the module input ) .
the reason for making @xmath934 decrease quadratically in the hyperbolic distance between @xmath194 and @xmath745 , is to ensure that if we fix the slope of a straight line geodesic @xmath948 connecting @xmath194 to a boundary point @xmath109 of the disc , then @xmath948 will not intersect @xmath930 when @xmath745 is sufficiently close to zero , and thus @xmath948 does not need to be deformed .
therefore , in a @xmath49-family , if @xmath745 converges to @xmath194 , and @xmath109 converges to @xmath949 , then the deformations of @xmath948 will converge to the undeformed geodesic @xmath950 .
this also holds in the borderline case @xmath951 since @xmath930 shrinks to a point as @xmath952 .
this convergence result is necessary to ensure that when counting floer solutions on these domains , and @xmath745 bubbles off to @xmath194 , then the resulting domain ( without the bubble ) is the domain used for the @xmath1-map , namely where geodesics are undeformed straight line segments from @xmath194 to a boundary puncture .
let us call @xmath953 the map which counts isolated solutions of @xmath679 for the above domains with the indicated auxiliary data . now consider a @xmath49-family of such solutions .
for example , the first picture in figure [ figure oc_qhaction3 ] ( for simplicity we draw the undeformed geodesics ) .
the possible degenerations are : 1 .
@xmath745 converges to the output @xmath194 ( possibly together with a subset of preferred points ) .
these are counted by the action of @xmath741 on the @xmath596-output of @xmath1 ( see the second picture in figure [ figure oc_qhaction3 ] ) ; 2 .
@xmath745 converges to the input @xmath49 : in this case a subset of the boundary punctures and preferred points can converge together with @xmath745 to @xmath49 .
these degenerations are counted by the action of @xmath741 on the input of @xmath1 ( see the third picture in figure [ figure oc_qhaction3 ] ) ; 3 .
@xmath745 does not converge to one of the punctures , but a subset of the boundary punctures converges to a boundary point @xmath764 giving rise to a disc bubble .
these degenerations are counted by the map @xmath953 composed with @xmath954 when @xmath949 , or @xmath953 composed with @xmath955 when @xmath951 .
in other words , @xmath953 composed with the hochschild differential @xmath956 defined in [ subsection cyclic bar complex ] ; 4 .
@xmath745 does not converge to one of the punctures , but there is a twice - punctured sphere that bubbles off at @xmath194 .
this degeneration is counted by @xmath957 composed with @xmath953 . in conclusion ,
the counts of these degenerations correspond to the equation @xmath958 which shows that @xmath953 is the chain homotopy required to prove that on cohomology @xmath1 preserves the @xmath596-action .
the theorem now follows from steps 1 - 8 .
in particular , observe that if @xmath959 is an @xmath213-module homomorphism , then @xmath960 is compatible with splittings into generalized eigenspacs of any @xmath961 : @xmath962 implies @xmath963 .
let @xmath965 be the subring of series having only non - negative powers of @xmath191 ( see [ subsection novikov ring ] ) .
let @xmath966 be the multiplicative group of units : the series having non - zero @xmath967-coefficient .
so @xmath968 is a local ring with maximal ideal @xmath969 the kernel of the evaluation @xmath970 of the @xmath967-coefficient , so the series with only strictly positive powers of @xmath191 .
the original motivation for local systems of coefficients came from the syz - viewpoint on mirror symmetry @xcite , and is very natural in the context of toric varieties more generally .
accordingly , we modify the @xmath91-category @xmath4 to @xmath971 where now the objects @xmath22 come equipped with a local system of coefficients @xmath972 with fibre @xmath8 and structure group @xmath973 .
so : 1 .
@xmath974 over each @xmath975 ; 2 . for any path @xmath466 in @xmath22 from @xmath109 to @xmath297
, there is an isomorphism @xmath976 given by multiplication by an element in @xmath973 depending only on @xmath977\in \pi_1(l , x , y)$ ] ; 3 .
concatenating paths yields composed @xmath978 s . for such @xmath979 and hamiltonian @xmath308
the floer complex is now @xmath980 summing over all integer @xmath174-chords @xmath109 of weight @xmath283 , and a floer solution @xmath981 asymptotic to hamiltonian chords @xmath982 at @xmath739 , which contributed @xmath259 } x$ ] to the old differential @xmath983 , now contributes @xmath984 } p_{\gamma_1}^{l_1 } \circ \phi_y \circ p_{\gamma_0}^{l_0 } \in \mathrm{hom}_{\lambda}(\underline{\lambda}_{x(0)}^{l_0},\underline{\lambda}_{x(1)}^{l_1}),\ ] ] where @xmath985 are the paths in @xmath979 swept by @xmath986 along the oriented arcs in @xmath987 connecting @xmath380 to @xmath181 , and @xmath181 to @xmath380 .
this defines the new differential @xmath988 .
equivalently , one considers flat @xmath973-connections on the trivial line bundle @xmath989 over @xmath22 .
so the holonomy is a homomorphism @xmath990 , which determines a class @xmath991\in h^1(l;\lambda_0^{\times})$ ] , and the parallel transport @xmath992 for a path @xmath466 from @xmath109 to @xmath297 is given by multiplication by @xmath993\in \lambda_0^{\times}$ ] ( evaluation of cocycles on chains ) .
independence of the homotopy class @xmath466 is recorded by the fact that @xmath994 is a cocycle .
so we get the usual @xmath8-module @xmath995 generated by @xmath174-chords of weight @xmath283 , however the above @xmath241 contributes @xmath259}b_{l_0}[\gamma_0]b_{l_1}[\gamma_1]\ , x$ ] to @xmath996 . for zero @xmath997 , which is a trivial local system ,
this recovers the usual @xmath998 .
the key observation is that if we consider a @xmath49-family of discs @xmath241 as above with fixed asymptotics @xmath982 , then the homotopy class of the arcs @xmath985 is constant in the family , so @xmath999}b_{l_0}[\gamma_0]b_{l_1}[\gamma_1]$ ] is constant .
thus @xmath1000 if and only if the old @xmath1001 .
similarly , the analogue @xmath1002 of [ subsection kappa map ] is defined as follows : @xmath259}b_{l_0}[\gamma_0]b_{l_1}[\gamma_1]\ , x$ ] is a contribution to @xmath1003 precisely when @xmath259 } x$ ] is a contribution to @xmath386 .
together with @xmath988 this defines @xmath1004 .
similarly , @xmath1005 now counts solutions @xmath241 with weight @xmath259}b_{l_0}[\gamma_0]\cdot b_{l_1}[\gamma_1]\cdots b_{l_n}[\gamma_n]$ ] where @xmath1006 are the images via @xmath241 of the oriented arcs in @xmath1007 which land in @xmath291 .
analogously , one defines the map @xmath1008 . for fukaya categories since
@xmath278 is not present and assuming @xmath1009 , the @xmath109 are just intersections points in @xmath1010 . in particular , in [ subsection the role of monotonicity ] , for @xmath1011 we model @xmath337 by locally finite chains @xmath1012 with local coefficients .
mimicking [ subsection the role of monotonicity ] , the new @xmath1013 s now depend on @xmath991\in h^1(l;\lambda_0^{\times})$ ] : @xmath1014 } b_l[\partial\beta]\ , \mathrm{ev}_*[\mathcal{m}_1(\beta ) ] \in c_{\mathrm{dim}\ , l}^{\mathrm{lf}}(l,\underline{\lambda}^l)\ ] ] summing over the homotopy classes @xmath318 of maslov index @xmath310 , where @xmath1015 corresponds to the disc bubbles @xmath241 in class @xmath1016=\beta$ ] ( see [ subsection the role of monotonicity ] ) . as argued in [ subsection the role of monotonicity ] , @xmath317 is a @xmath1017-cycle so @xmath320 $ ] is a multiple of @xmath321 $ ] , and finally we define @xmath1018 by @xmath1019.\ ] ] just like in the absence of local systems , @xmath1020 splits up according to @xmath271-values ( see [ subsection the role of monotonicity ] ) .
even though typically the @xmath271-values have changed by introducing local systems ( i.e. @xmath1021 ) , the possible @xmath271-values are still constrained to lie in the spectrum of eigenvalues of @xmath1022 because lemma [ lemma fukaya cats split according to evalues ] holds also in the presence of local systems .
the advantage of local systems is that a wider range of eigenvalues can now arise geometrically as @xmath271-values
. we will argue in [ subsection review of the results ] that all our results for @xmath208 hold also for @xmath971 .
a full subcategory @xmath1023 _ split - generates _ an @xmath91-category @xmath172 if for each @xmath1024 , there is a twisted complex built from @xmath1025 admitting @xmath1026 as a summand . concretely : @xmath1027 is obtained by starting from @xmath1025 , forming mapping cones and shifts , repeating that process arbitrarily often , and then taking images under all idempotent endomorphisms of the resulting objects . if it is not necessary to take images under idempotents , then we say @xmath1023 _ generates _ @xmath172 .
we sometimes say @xmath1025 _ split - generates _ or _ generates _
@xmath172 , respectively .
[ theorem abouzaid generating criterion ] let @xmath1028 be an exact symplectic manifold @xmath1029 conical at infinity , and @xmath1023 a full subcategory of @xmath1030 .
if @xmath49 lies in the image of @xmath1031 & \mathrm{hh}_*(\mathcal{w}(m ) ) \ar@{->}[r]^-{\mathrm{oc}_m } & sh^*(m ) , } \ ] ] then @xmath1023 split - generates @xmath1030 . by theorem
[ theorem oc is a module map ] , it in fact suffices that @xmath1032 contains an @xmath1033-invertible : [ lemma qh 1 is enough ] if @xmath1034 is invertible in @xmath1033 then @xmath1035 .
@xmath1036 for some @xmath1037 , so since @xmath1038 is an @xmath1033-module map by theorem [ theorem oc is a module map ] ( which of course holds also in the exact setup ) , @xmath1039 . in this section , we will prove that the generation criterion holds in the monotone setting .
[ theorem generation in monotone setting ] for any monotone symplectic manifold conical at infinity , the generation criterion holds for the wrapped fukaya category defined as in [ subsection the objects of the wrapped fukaya category ] .
[ remark about idempotent summands ] analogously , if an idempotent summand @xmath1040 of @xmath49 is in the image of @xmath74 on a full subcategory @xmath1023 of @xmath4 , then @xmath1023 split - generates the summand of @xmath4 corresponding to @xmath1040 .
we explain this further in remark [ remark idempotent summands ] .
the reason the proof of abouzaid does not immediately carry over to the monotone setting is that the @xmath91-categories have been constructed in significantly different ways .
abouzaid s construction relies on a subtle use of the liouville flow , which is globally defined in the exact case but not in the monotone case .
fix an object @xmath481 in @xmath1041 , where @xmath1029 is a monotone symplectic manifold which is conical at infinity .
recall @xmath1023 is a full subcategory of @xmath1030 .
let @xmath20 , @xmath1042 denote the yoneda @xmath1023-modules @xmath1043 where @xmath1044 .
recall @xmath1042 represents the object @xmath481 ( see ( * ? ? ?
* sec.(2 g ) ) ) . using these as the @xmath20 , @xmath1042 in [ subsection tensor products ] for the @xmath91-category @xmath1023
, we deduce @xmath1045 in [ subsection the coproduct ] , we will construct the coproduct @xmath1046 , which is an @xmath1023-bimodule map @xmath1047 in particular @xmath1048 is the usual coproduct . in section [ subsection the oc map ]
we will construct the closed - open string map @xmath1049 , constructed in analogy with the @xmath1-maps .
this yields two interesting composite maps : @xmath1050 & sh^*(m ) \ar@{->}^-{\mathrm{co}}[r ] & hw^*(k , k)\\ \mathrm{c}_2:\mathrm{hh}_*(\mathcal{s},\mathcal{s } ) \ar@{->}^-{\mathrm{hh}_*(\delta)}[r ] & \mathrm{hh}_*(\mathcal{s},\mathcal{l}\otimes \mathcal{r } ) \cong h^*(\mathcal{r}\otimes_{\mathcal{s}}\mathcal{l } ) \ar@{->}^-{\mu_{\mathcal{s}}}[r ] & hw^*(k , k ) } \ ] ] using the @xmath1051 to compose all morphisms of the chain complex @xmath1052 .
the key ingredient of the proof is to show that these composites are equal , which we will prove in [ subsection the key ingredient ] .
the hypothesis that @xmath1 hits @xmath1053 implies @xmath1054 hits the identity @xmath1055 since @xmath1056 is a ring homomorphism .
therefore @xmath1057 hits @xmath1058 . for purely algebraic reasons ( * ? ? ?
* appendix a ) this implies that @xmath1042 ( which represents @xmath481 ) is a summand of a twisted complex built from objects in @xmath1023 .
therefore any object @xmath481 in @xmath1030 is split - generated by objects in @xmath1023 , as required .
we illustrate that purely algebraic reasoning in a very simple example : when @xmath1023 is generated by a single object @xmath174 , and we assume that @xmath1059 for some @xmath1060 , @xmath1061 .
let s abbreviate that by @xmath1062 .
then @xmath1063 is an idempotent : @xmath1064 .
thus @xmath1065 \ar@<-2pt>@{<-}_-{\alpha}[r ] & hw^*(x , x ) & \alpha\cdot \beta = e_k , \ ; \beta\cdot \alpha = p}\ ] ] displays @xmath1066 as a summand of @xmath1067 : the image of the projection @xmath764 .
[ remark idempotent summands ] suppose @xmath1023 is a full subcategory of @xmath4 such that @xmath1068 hits the identity @xmath1058 , and @xmath1069 lies in the summand @xmath13 of @xmath4 corresponding to the eigenvalue @xmath347 of @xmath1070
. then in fact @xmath1071 hits @xmath1058 , since the maps are @xmath587-module maps ( by theorem [ theorem oc is a module map ] ) .
so it suffices to consider lagrangians with a single @xmath271-value in the subsequent proof of the generation criterion . in particular
, we never encounter holomorphic annuli with boundaries on lagrangians with different @xmath271-values .
this outline is precisely the argument of abouzaid @xcite , so the only difference for our setup is that we need to construct @xmath1056 , @xmath1046 and prove the equality of @xmath1072 in the monotone setting .
@xmath1073 arises at the chain level as the sum of certain maps @xmath1074 \to cw^*(k , k;w_0h)[\mathbf{q}]$ ] .
we now define @xmath1075 by mimicking [ subsection open - closed with q coefficients ] .
consider @xmath1076 ( recall we used @xmath1077 for @xmath1078 ) .
denote the interior puncture by @xmath1079 and the boundary puncture by @xmath784 . mimicking [ subsection the auxiliary data for wrapped category ] ,
the auxiliary data is : an index set @xmath406 with a ( constant ) map @xmath1080 , weights @xmath1081 with @xmath1082 , a closed form @xmath1083 , and sub - closed forms @xmath456 indexed by @xmath406 . the total @xmath49-form on @xmath355 is @xmath1084 , which pulls - back to @xmath1085 near @xmath1086 and @xmath1087 near @xmath315 .
we can assume @xmath406 is either @xmath1088 or empty , depending on whether we have a geodesic with one preferred point joining @xmath1089 or not ( the count of other configurations will cancel for symmetry reasons as in ( * ? ? ?
* lemma 3.7 ) ) .
define the @xmath1090 coefficient of @xmath1091 as the count with weight @xmath1092 of isolated solutions @xmath696 of @xmath679 with lagrangian boundary condition @xmath481 and asymptotic conditions @xmath1093 at @xmath1089 , using auxiliary data @xmath1094 if @xmath1095 and
@xmath1096 if @xmath1097 . the @xmath524 coefficient of @xmath1091 is determined by requiring @xmath1075 is @xmath383-equivariant . an easy check shows @xmath1056 is indeed a chain map . to prove that @xmath1056 is a ring homomorphism on cohomology
, we argue as in theorem [ theorem oc is a module map ] : consider the domains as above except @xmath1086 is replaced by two interior puntures @xmath1098 lying on the geodesic joining @xmath1099 symmetrically about @xmath194 . in a @xmath49-family with fixed auxiliary data , if @xmath1098 converge together to @xmath194 , a @xmath548-punctured bubble breaks off giving rise to the product on @xmath29 ; if @xmath1098 converge to the boundary then two disc bubbles form , contributing to the composite @xmath1100 .
[ remark co full map ] one can extend the above closed - open string map to @xmath1101 by having several outputs as opposed to just one output , and then introducing auxiliary data as we did for the open - closed string map @xmath74 of section [ section open - closed string map ] . in particular , this map is an acceleration of the analogous homomorphism @xmath1102 constructed using a small hamiltonian @xmath1103 ( compare [ subsection the acceleration diagram ] ) .
the map @xmath1104 is a module homomorphism over @xmath587 and also over @xmath33 ( which is proved as for @xmath74 in theorem [ theorem oc is a module map ] ) , and it is a ring homomorphism ( unlike @xmath74 ) . for weights @xmath1105 , abbreviate @xmath1106\otimes \cdots \otimes cf^*(l_0,l_1;w_1h)[\mathbf{q } ] .
\end{array}\ ] ] we need to define for @xmath115 , a map @xmath1107 @xmath1108 \otimes \mathcal{w}(l_0',\ldots , l_s';\mathbf{w } ' ) \\
\longrightarrow cf^*(k , l_r;w_0h)[\mathbf{q}]\otimes cf^*(l_s',k;w_0'h)[\mathbf{q } ] , \end{array}\ ] ] observe figure [ figure coproduct ] .
we are now considering a universal family @xmath1109 in which there are two distinguished boundary punctures @xmath1110 on the riemann surface @xmath355 which serve as the outputs for @xmath1046 , and one distinguished puncture @xmath1111 which serves as the input @xmath1046 receives from the module @xmath1112 .
these distinguished @xmath1113 are fixed to lie at @xmath1114 , after identifying @xmath1115 with the disc @xmath413 .
this kills the reparametrization group of @xmath1116 and the local coordinates for @xmath1116 are given by the angular coordinates of the punctures @xmath1117 lying on the arc @xmath1118 and of the punctures @xmath1119 lying on the arc @xmath1120 .
the punctures are ordered as shown in figure [ figure coproduct ] .
we construct inductively two sets of auxiliary data , mimicking [ subsection the auxiliary data for wrapped category ] .
so @xmath1105 and @xmath1121 are integer weights ; @xmath1122 are finite sets of indices @xmath1123 , coming with maps @xmath1124 , @xmath1125 defining labels @xmath1126 ( possibly not distinct ) , where we use @xmath1127 to denote @xmath1128 ; we have sub - closed forms @xmath456 , @xmath1129 ; we have closed forms @xmath1130 and @xmath1131 .
finally this defines the total @xmath49-form on @xmath355 : @xmath1132 with @xmath1133 , @xmath465 near @xmath415 .
the pull - back of @xmath466 to @xmath415 is @xmath194 .
define @xmath1134 and @xmath1135 .
near the ends , in the relevant parametrization , @xmath466 has the form : @xmath1136 , @xmath1137 , @xmath1138 , @xmath1087 , @xmath1139 respectively near @xmath446 , @xmath1111 , @xmath1140 , @xmath315 , @xmath1141 .
this is consistent with stokes : @xmath1142 . by mimicking [ subsection moduli space of pseudoholo maps in wrapped case ]
, we obtain a moduli space @xmath1143 of solutions @xmath696 of @xmath679 satisfying the relevant asymptotic and boundary conditions . arguing as in ( * ? ? ?
* lemma 3.7 ) , we can always assume that there is never more than one preferred point on a geodesic , since for symmetry reasons the contributions of isolated solutions will cancel when @xmath432 or @xmath1144 is not injective .
let us abbreviate by @xmath221 the map we need to define above .
we will define it inductively on the total degree of @xmath379 arising in the input ( compare the definition of the @xmath1 maps in [ subsection open - closed with q coefficients ] ) .
we will later define the @xmath1145-coefficient of the output of @xmath221 . once this is done , by imposing @xmath383-equivariance of @xmath221 , the @xmath1146- and the @xmath1147-coefficients are determined .
finally the @xmath1148-coefficient is determined by imposing @xmath383-equivariance of @xmath221 .
an issue might arise at this stage : @xmath1149 ( recall @xmath379 has degree @xmath380 ) , so for consistency one needs in general that for a tensor @xmath283 the output @xmath1150 has @xmath1151 and @xmath1152 arising with the same coefficient up to @xmath1153 .
but this is indeed the case by considering @xmath1154 ( using @xmath1155 ) . since @xmath381
, we have determined @xmath221 .
we now define the @xmath1145-coefficient of the output of @xmath221 . just as in [ subsection open - closed with q coefficients ] ,
the only case when the output is non - zero is when the input @xmath1156 comes with the correct powers of @xmath379 , that is @xmath520 iff @xmath521 ( respectively @xmath1157 iff @xmath1158 ) , in other words there is a @xmath379 iff there is a geodesic ( with one preferred point ) joining the relevant puncture with @xmath315 ( respectively @xmath1141 ) .
if this is the case , then the @xmath1159 coefficient of the output is a count with weight @xmath1092 of the isolated solutions @xmath1160 . that @xmath1046 satisfies the @xmath91-equations required from an @xmath1023-bimodule map @xmath1161 is now an exercise in bookkeeping ,
compare remark 4 in [ subsection wrapped a infinity structure ] .
the key observation is that in codimension @xmath49 , either a subset of the punctures ( and possibly some preferred points ) converges to a common point on the boundary that is not @xmath1162 , or a subset of the punctures and/or preferred points converges to one of @xmath1162 . in the first case it gives rise to a @xmath1163-term ( which is part of the bar differential for the @xmath1023-bimodule @xmath1023 ) . in the second case : if the common point is @xmath315 it gives rise to a @xmath1164-term ; if @xmath1141 , it gives rise to a @xmath1165-term ; if @xmath1111 , it gives rise to a @xmath1166-term .
in particular if no punctures but exactly one preferred point converges to a puncture then we get the @xmath366-contributions in @xmath1167 , @xmath1168 , @xmath1169 .
we now prove that the composites @xmath1072 from [ subsection outline of the proof of generation ] agree .
this involves studying a moduli space of annuli , which in abouzaid s case is described in ( * ? ?
* appendix c4,c5 ) , and in our case will be endowed with additional decorations coming from geodesics and preferred points .
keeping track of this additional data may seem cumbersome at first , but it has the advantage that all gluings are natural : so in fact we will not need the first homotopy of abouzaid ( * ? ? ?
* sec.6.2 ) which changes the weights so that gluing becomes possible , and we will only need to construct the analogue of his second homotopy ( * ? ? ? * sec.6.3 ) .
consider the moduli space of all holomorphic annuli with two boundary punctures , one on each component of the boundary .
for example the family of annuli @xmath1170 by doubling up an annulus , one obtains an elliptic curve @xmath1171 with two marked points together with a real structure ( an antiholomorphic involution @xmath843 , whose fixed locus are the two boundary circles of the annulus , and it fixes the two marked points ) .
the pairs ( @xmath1171 , @xmath843 ) form , via an tale forgetful map , a subset of the deligne - mumford space @xmath1172 ( recall @xmath1173 ) .
this is described in detail in fukaya - oh - ohta - ono ( * ? ? ?
* chp.2 ) .
the family @xmath1174 determines a real @xmath49-dimensional family in @xmath1172 parametrized by the modular parameter @xmath1175 $ ] .
a typical @xmath1176 degeneration of @xmath1172 involves the shrinking of a hyperbolic geodesic in the punctured elliptic curve , which in the limit gives rise to a nodal curve with one node .
however , since our family @xmath1174 comes with real structures , the boundary strata will in fact carry two nodes ( and @xmath843 either fixes or interchanges the two nodes ) , and so the elliptic curves in the family @xmath1174 will degenerate into two spheres joined at two nodes .
when @xmath1177 , one of the two nodes is forming in the interior of our annulus , thus in the limit the annulus breaks into two discs joined at an interior point as in picture 1 of figure [ figure coproduct2 ] ignoring all decorations ( this is the case when @xmath843 swaps the nodes ) . when @xmath1178 , the two nodes arise as a limit of two pairs of boundary points of the annulus connected by two shrinking geodesic arcs . in the limit
this gives rise to two discs joined at two nodes as in picture 5 of figure [ figure coproduct2 ] ignoring all decorations ( this is the case when @xmath843 fixes the nodes ) .
observe that in both cases the two original marked points will lie on opposite components of the nodal curve ( in the second case this is because one disc would otherwise be unstable ) .
an analogous argument can be carried out when there are several marked points on the outer circle @xmath1179 , which we view as a family lying over our family @xmath1174 via a forgetful map .
thus the punctures at @xmath1180 are distinguished , which is why in figure [ figure coproduct2 ] we mark them with a dash , and we mark the other punctures with a cross .
the dotted lines in pictures 2 and 3 of figure [ figure coproduct2 ] are examples of hyperbolic geodesics which shrink in the limit to give respectively pictures 1 and 5 . in picture 2 , we work in the @xmath1181 plane , where @xmath1182 parametrizes the annulus , so we ve passed to the cover @xmath1183 of the angular variable .
the limit @xmath1177 corresponds to stretching the strip 2 near the dotted line .
picture 4 is biholomorphic to 3 , and shrinking the dotted lines in 3 corresponds to stretching the two handles in 4 .
the dotted lines play the same role as the `` cuts '' in figures [ figure popsicle ] and [ figure cutting ] .
observe our moduli space of annuli in picture 2 or 3 of figure [ figure coproduct2 ] .
the distinguished boundary punctures are @xmath1184 ( which will be the output ) and @xmath1185 ( which will be the bimodule input ) , and the other boundary punctures @xmath1186 ordered counterclockwise on the boundary @xmath1187 . mimicking [ subsection the auxiliary data for wrapped category ] ,
the auxiliary data is : a finite indexing set @xmath406 with a map @xmath1188 which keeps track of the presence of preferred points on geodesics joining @xmath1111 with the other punctures ( with the important convention that the geodesic joining @xmath1189 , if present , is always the one which increases @xmath191 ) ; weights @xmath1190 ; closed forms @xmath1191 ; sub - closed forms @xmath456 .
these define a total @xmath49-form @xmath1192 , which pulls back to @xmath1136 , @xmath1193 , @xmath1194 near @xmath1195 , where @xmath1196 .
consider a @xmath49-dimensional moduli space of solutions @xmath696 of @xmath679 defined on a fixed annulus @xmath355 , and satisfying the relevant boundary and asymptotic conditions .
pictures 1 and 5 of figure [ figure coproduct2 ] are two examples of boundary points of this moduli space .
notice there are geodesics in pictures 1 and 5 which do not carry preferred points : this is a natural phenomenon that arises also in the breaking analysis for the @xmath1197-maps , and remarks 1 and 2 in [ subsection the auxiliary data for wrapped category ] and remarks 3 and 4 in [ subsection wrapped a infinity structure ] explain this : these geodesics encode how the breaking has occurred , and each breaking yields distinct composites of the relevant operators .
if the cut in picture 2 had included more than one preferred point on the left , then picture 1 would have two or more preferred points on the same geodesic , and those broken configurations would cancel out at the level of operators ( algebraically : @xmath381 ) .
we conclude that ( ignoring all @xmath379 s , to keep the discussion comprehensible ) that picture 1 is a contribution to the coefficient of @xmath1198 in @xmath1199 whereas picture 5 is a contribution ( ignoring all @xmath379 s ) to the coefficient of @xmath1198 in @xmath1200 where @xmath1201 is the reordering isomorphism from [ subsection tensor products ] which induces @xmath1202 ( notice above that we are inputting into @xmath1201 a typical contribution to @xmath1203 as prescribed by lemma [ lemma functoriality of hh ] ) .
this example now easily generalizes to show that the boundary points of such @xmath49-dimensional moduli spaces include the counts involved in the maps @xmath1204 and @xmath1205 .
there are however also two other contributions which we now describe .
let @xmath1206 denote the weighted count of isolated solutions @xmath241 defined on an annulus as above .
now consider again a @xmath49-dimensional family of solutions on the annulus .
firstly , when a subset of the outer boundary punctures converges to a common point , we obtain bubbling configurations counted by @xmath1207 where @xmath728 is the bar differential for @xmath1208 .
secondly , when bubbling occurs on the inner boundary circle of the annulus , the broken configurations are counted by @xmath1209 .
we remark on a particular case of the first type of bubbling : when the bubble carries none of the boundary punctures .
these are the terms involving the @xmath349 s that are supposed to arise in the bar differential and in the @xmath1197-maps ( see [ subsection the role of monotonicity ] ) .
these terms will all cancel out since they cancel out for the @xmath91-composition maps . since the oriented count of boundary points of a compact @xmath49-dimensional manifold is zero ,
we conclude that @xmath1054 and @xmath1210 differ at the chain level by @xmath1211 , therefore they agree on cohomology .
this concludes the proof of theorem [ theorem generation in monotone setting ] .
from now on : 1 .
@xmath8 is the novikov ring ( see [ subsection novikov ring ] ) ; 2 .
@xmath554 is a closed monotone symplectic manifold ; 3 .
@xmath92 is the fukaya category @xmath39 of @xmath15 ( see [ subsection fukaya categories b and e ] ) ; 4 .
@xmath553 is a hamiltonian on @xmath15 ( see [ subsection hamiltonian floer cohomology ] ) ; 5 .
@xmath209 is a monotone symplectic manifold conical at infinity , see [ subsection symplectic manifolds conical at infinity ] ( in section [ section generation and functoriality for the compact category ] we will discuss the case when @xmath0 is compact rather than non - compact ) ; 6 .
@xmath147 is the generalized fukaya category of @xmath0 ( discussed in [ subsection generalized fukaya category ] ) ; 7 .
@xmath4 is the wrapped fukaya category of @xmath0 ( see [ subsection the objects of the wrapped fukaya category ] ) ; 8 .
@xmath575 is a hamiltonian on @xmath0 linear at infinity of small slope ( see [ subsection symplectic cohomology ] ) ; 9 .
@xmath1213 denotes the product @xmath1214 with symplectic form @xmath1215 ( the rescaling makes @xmath1213 monotone with monotonicity constant @xmath1216 ) ; 10 .
@xmath18 is a closed lagrangian correspondence from @xmath15 to @xmath0 , that is : @xmath1217 is a closed monotone oriented lagrangian submanifold . in this section ,
we use the quilt machinery of wehrheim - woodward @xcite .
this allows us to define maps by pictures of quilts and allows us to do gluing arguments pictorially .
define @xmath1219 by counting isolated quilt maps as in figure [ figure quilt morph on qh ] ( we use cohomological conventions ) , where @xmath688 is small ( see [ subsection symplectic cohomology ] , [ subsection pss ] ) and where @xmath1220 is the degree of the map which we will determine later . the quilt domain is an infinite cylinder @xmath724 , with the seam condition @xmath18 on the circle @xmath661 .
after folding , the quilt maps are of the form @xmath1221 \times s^1 \to \overline{b}\times e\ ] ] satisfying the floer equation @xmath1222[\partial_s u - c(s ) x_{h^b}\oplus x_{h^e } ] = 0,\ ] ] with boundary condition @xmath1223 , where @xmath613 $ ] is a decreasing cut - off function which equals @xmath49 for @xmath614 and equals @xmath194 for @xmath615 . in general , such maps
@xmath1224 have been investigated before in ( * ? ? ?
* sec.6.1 ) .
abbreviate dimensions by @xmath1226 [ theorem qgamma morphism ] the composite ( using the pss maps from [ subsection pss ] ) @xmath1227 is identifiable with the homomorphism @xmath1228\ ] ] where @xmath774 is the intersection product @xmath1229 , and @xmath1230 . by poincar duality
, we identify @xmath1231 for details about locally finite homology , we refer to @xcite .
consider the intersection product @xmath1232 in general , suppose we are given a locally finite pseudocycle @xmath1233 defined as an evaluation map at some marked point for some moduli space @xmath1234 .
say @xmath1235 = \sum \lambda_i \gamma_i$ ] , where @xmath1236 and @xmath1006 is a basis of lf - cycles for @xmath0 .
then @xmath1237 is the ( novikov - weighted ) count of solutions in @xmath1234 which intersect the cycle @xmath1238 at that marked point , where @xmath1238 is the dual to @xmath1006 with respect to @xmath774 .
using poincar duality , @xmath1239 , then using @xmath774-duality : @xmath1240 by a standard gluing argument , this counts certain quilt maps whose domain is a sphere @xmath1241 , with seam condition @xmath18 along the equator . by a standard floer homotopy argument
we can assume the hamiltonians have been homotoped to zero .
after folding the sphere to obtain a disc with boundary condition @xmath18 , @xmath1242 counts isolated @xmath1243-holomorphic discs @xmath1244 intersecting @xmath1245 at @xmath616 ( here @xmath1246 ) .
but no such non - constant @xmath241 will be isolated : there is a rotational @xmath659-symmetry of the domain ( more precisely : the geometric image of the evaluation map from the fiber product of the moduli space of discs crossed with the domain , modulo reparametrization , evaluating into the manifold @xmath1213 will sweep a cycle that is too small to intersect the cycle condition imposed at the centre of the disc ) .
so the only solutions being counted are the constants @xmath1247 which yield the ordinary intersection product .
thus @xmath1248 .
finally apply knneth s theorem .
we remark that the map in the theorem , which is part of a commutative diagram involving the @xmath1 map and the map @xmath1249 , can be constructed entirely in the holomorphic world ( that is without appealing to @xmath1250 groups ) .
the fact that it agrees with the above glued map boils down to the fact that the two pss maps are indeed inverse to each other at the level of cohomolgy . by just working in the holomorphic world
, one would bypass the `` homotope @xmath278 to zero '' part of the argument .
the way the argument is written above , one would actually need to explain why , after homotopying @xmath278 to zero , there are no gromov limits involving sphere / disc bubbles .
the reason is that the principal component of the gromov limit has to have full index otherwise the moduli space will not intersect the cycle condition , therefore the bubbles have to have index zero . but
chern number zero spheres impose a real codimension @xmath1251 condition so this bubbling does not occur for dimension reasons ; and maslov index zero discs are constant by monotonicity so this is not actually a bubbling .
this argument relies on the fact that the almost complex structure that we use when counting solutions must also be regular for the moduli spaces of spheres and of discs bounding the relevant lagrangians , which of course is assumed and not problematic in the monotone setting .
if @xmath18 is null - homologous , then @xmath1253 .
so to have a chance at making @xmath1254 non - zero , we would need to kill the @xmath659-freedom which arose in the previous proof by introducing an extra boundary marked point .
we now show how this gives rise to a deformed map depending on a class in @xmath27 .
fix a puncture labeled @xmath1255 on the seam @xmath18 in figure [ figure quilt morph on qh ] .
define @xmath1256 by counting quilt maps as before , with the additional condition that @xmath1255 is an input for the floer cycle @xmath1257 .
observe that locally near @xmath1255 one can fold the solutions @xmath241 to give a local map @xmath1258 defined near @xmath1259 , with lagrangian boundary condition @xmath1260 , so the input @xmath466 at @xmath616 is well - defined .
moreover , observe that the following degenerations of @xmath49-families of solutions can occur : breakings at the ends which define the floer differentials for @xmath556 and @xmath576 , and discs bubbling on @xmath18 .
the discs bubbling on @xmath18 must carry the boundary puncture @xmath1255 , otherwise they would have too high index ( recall @xmath18 is monotone and oriented ) .
the disc bubbles carrying the boundary puncture @xmath1255 correspond to the floer differential @xmath1261 , but this vanishes since @xmath466 is a floer cocyle . therefore @xmath1256 is a chain map .
capping off the ends , using the pss - maps , defines the map @xmath1262 .
it is not a priori known that @xmath1263 .
the undeformed map @xmath1254 corresponds to the case @xmath1264
\in hf^*(\gamma,\gamma)$ ] .
[ theorem qgamma morphism deformed ] the deformed quilt homomorphisms @xmath1265 are determined by the homomorphism @xmath1266 the first map is quantum intersection product , the second is the augmentation which projects to the coefficient of the generator @xmath1267 $ ] in @xmath1268 . arguing as for theorem [ theorem qgamma morphism ] , we may identify @xmath1262 with the map @xmath1269 \ } \subset qh_{k}(\overline{b}\times e ) \otimes h_{*}(\gamma;\lambda ) \to \lambda\ ] ] which maps a generic cycle @xmath1270 to the ( novikov - weighted ) count of isolated @xmath1243-holomorphic @xmath1271 , with @xmath1272 and @xmath1273 ( where @xmath1274 is the fixed marked point ) .
for dimension reasons dictated by the quantum product , this forces @xmath1275 ( using the homological grading for @xmath466 ) .
next , we recall the quantum intersection product action in our setup , and we refer to auroux ( * ? ? ? *
sec.6.1 ) for details .
fix the three marked points @xmath1276 on the disc @xmath1277 ( alternatively , if one later quotients by domain reparametrizations , take two distinct freely moving boundary marked points and an interior marked point free to move along the hyperbolic geodesic joining the two boundary marked points ) .
consider the moduli space @xmath1278 of @xmath1243-holomorphic discs @xmath1279 , lying in class @xmath1016=b\in \pi_2(\overline{b}\times e,\gamma)$ ] .
let @xmath1280 denote the evaluations at @xmath743 and @xmath194 respectively .
given chains @xmath1281 , @xmath1282 such that @xmath1283 is transverse to @xmath1284 , define @xmath1285 define the weight @xmath1286 = \int_{{\mathbb{d } } } u^*\omega$ ] for any @xmath1287 .
the grading convention [ subsection novikov ring ] is @xmath1288}| = 2\lambda_e \omega[b]= 2\lambda_e\lambda_{\gamma } \mu(b)=\mu(b).\ ] ] the quantum intersection product is by definition @xmath1289}\;\left((\alpha,\beta ) * _ { b } \gamma\right ) \in c_*(\gamma;\lambda ) \end{array}\ ] ] and it is degree - preserving if one uses cohomological gradings .
as explained more generally in the proof of theorem [ theorem qgamma morphism ] , the requirement that @xmath1290 sweeps a multiple of the cycle @xmath1267 $ ] in @xmath18 can be replaced by the condition that the discs @xmath241 intersect the @xmath774-dual cycle @xmath1291 $ ] inside @xmath18 at the marked point @xmath1292 .
but the latter condition is automatically satisfied , so composing the quantum intersection product with the augmentation gives precisely the count of holomorphic discs which define @xmath1262 .
observe that , since @xmath0 is non - compact , one must take care to distinguish between @xmath1293 which is represented by cycles , and @xmath1294 which is represented by locally finite cycles .
the theorem involves the former .
recall that the moduli space @xmath1278 defined above has @xmath1295 where @xmath1296 is the maslov index .
therefore , in the definition of @xmath1297 the only @xmath728 which contribute have maslov index satisfying @xmath1298 so @xmath1299 the grading of @xmath1297 .
thus @xmath1300 where the sum over @xmath473 ranges over a basis of @xmath1301 , with @xmath774-dual basis @xmath1302 in @xmath1303 , and @xmath1304 is the homological grading of @xmath466 ( so @xmath1305 if @xmath206 ) .
consider the open - closed string map @xmath1306 for @xmath1217 , and view @xmath466 as a floer cocycle in @xmath1268
. then @xmath1307\ ] ] where @xmath1308 are respectively the cycle and the lf - cycle dual to the cycles @xmath469 via the ordinary intersection product , and we sum over a basis @xmath1309 $ ] of @xmath1310 .
this follows immediately upon comparing with the discs counted by @xmath1262 , described in the proof of theorem [ theorem qgamma morphism deformed ] , and upon observing that the condition of intersecting @xmath1245 at the centre of the disc is equivalent to requiring that the moduli space of such discs sweep @xmath1311 under evaluation at the centre . for @xmath1312 , the map @xmath1262 factors
as : @xmath1313 where @xmath1314 is the floer triangle product by @xmath466 , and @xmath1315 is slant product by the point class in @xmath15 . let @xmath960 denote the map defined by the factorization in the claim .
observe figure [ figure qfactorization ] .
on the left are the configurations which count the coefficient of the locally finite cycle @xmath1171 in @xmath0 in the output of @xmath1316 , where @xmath473 is a cycle in @xmath15 , in particular the target space of the maps defined on those discs is the space @xmath1213 .
observe in the figure that we replaced the condition of sweeping @xmath1317 by the condition of intersecting @xmath1318 at the left interior marked point , where @xmath1319 is the cycle in @xmath0 dual to @xmath1171 via the ordinary intersection product .
now consider the figure on the right : discs with one boundary puncture at @xmath1320 and two interior marked points at @xmath1321 where @xmath1322 . in a @xmath49-family , as @xmath1178 those configurations break to give the figure on the left .
on the other hand , as @xmath1323 , a bubble appears at @xmath616 carrying the two interior punctures : this is the quantum product of @xmath1324 and @xmath1325 . since @xmath1326,[e]$ ] are the units for quantum product in @xmath1327 respectively , we have @xmath1328\cdot [ \beta , e ] = [ \beta,\varepsilon^{\vee}]$ ] .
so , up to chain homotopy , @xmath1329 .
even though @xmath1330 are @xmath14-module maps , the map @xmath1331 is not in general a @xmath14-module map ( in any reasonable sense ) due to the slant - product in the above factorization . indeed , suppose we introduce a new interior marked point @xmath1332 on the ray @xmath1333 inside the quilt cylinder @xmath724 from figure [ figure quilt morph on qh ] , and we impose the intersection condition at @xmath1332 with a cycle @xmath1334 in @xmath15 . in a @xmath49-family , @xmath1332 may diverge to infinity giving rise to the floer product by @xmath1334 .
the other possible breaking is if @xmath1332 converges towards the marked point @xmath1255 on the seam , giving rise to a disc bubble in @xmath1213 bounding @xmath18 , carrying @xmath1332 as an interior marked point and @xmath1255 as a boundary marked point .
we can run a similar argument for a marked point @xmath1335 on @xmath658\times \{0\}$ ] and an lf - cycle @xmath1336 in @xmath0 .
thus , abbreviating @xmath1337 , this proves that : @xmath1338 where @xmath1339 denotes the quantum intersection product of @xmath1340 acting on @xmath1341 .
so even if @xmath1336 were obtained by a push - pull construction from @xmath1334 , the difference @xmath1342 may be non - zero , depending on the quantum cohomology action on @xmath1268 .
for now we denote by @xmath1343 the generalized fukaya category of @xmath0 as defined by wehrheim - woodward @xcite using generalized lagrangian correspondences as objects . _ generalized lagrangian correspondence _
@xmath1344 will mean : @xmath1345 is a monotone closed oriented lagrangian submanifold , @xmath1346 are monotone symplectic manifolds , @xmath1347 , @xmath1348 ( with @xmath1346 conical at infinity if @xmath1346 is non - compact ) .
the morphism spaces of @xmath1349 are the floer cohomology groups @xmath1350 defined in @xcite ( we later mention briefly what they are in our context ) . for sufficiently nice @xmath18 one can in fact later switch back to the fukaya category @xmath629 ( see [ subsection phi lands in fuk e ] ) , but we will keep things general for now .
we emphasize : there is no wrapping for now , and all lagrangians are closed .
we aim to study the natural functor @xmath1353 which on objects is @xmath1354 where a monotone lagrangian @xmath1355 is viewed as a correspondence in @xmath1356 . on morphisms ,
the map is given by `` gluing in the identity @xmath1357 '' , more precisely for monotone lagrangians @xmath1358 , the morphism @xmath1359 is defined by counting quilt maps schematically drawn in figure [ figure phi quilt ] .
we extend @xmath1360 to an @xmath91-functor @xmath140 : @xmath1361 counts quilt maps as in figure [ figure phi quilt ainfinity ] . in the second picture ,
the domain is a disc @xmath1362 , there is a distinguished fixed puncture at @xmath1292 drawn as a dash ( corresponding to the output of @xmath163 ) , and the remaining boundary punctures , drawn as small circular dots , are free to move on the arc @xmath1363 ( subject to being distinct and ordered anticlockwise ) .
the @xmath91-equations are a standard consequence of a bubbling analysis argument , by considering what happens when a subset of the free points converge together .
the types of bubbles that appear are sketched in figure [ figure ainfinity for phi ] .
[ definition composition of lagrangians ] for a lagrangian @xmath17 , denote by the curly letter @xmath20 the geometric composition with @xmath18 , which is defined by the push - pull construction : @xmath1366\\ & = & \left\ { e\in e : \exists b\in b \;\mathrm{with}\ ; b \in
l \;\mathrm{and}\ ; ( b , e ) \in \gamma \right\ } \end{array}\ ] ] where @xmath1367 are the projections .
the composition is _ transverse _ if the intersection in the second line is transverse , and is _ embedded _ if @xmath1368 on the resulting intersection is injective .
[ definition good gamma ] [ definition very good gamma ] call @xmath18 _ good _ if for all monotone orientable lagrangians @xmath1355 , the composite @xmath1369 is transverse and embedded ( so @xmath20 is an orientable lagrangian ) . call a good @xmath18 _ very good _ if for all such @xmath17 , the composed lagrangian @xmath1369 is monotone .
[ lemma properties of good gamma ] if the projection @xmath1370 is a smooth fibre bundle and @xmath1371 is injective , then @xmath18 is good . if @xmath1372 has torsion kernel , then @xmath18 is monotone .
@xmath1373 being a smooth fiber bundle is equivalent to @xmath1373 being a surjective submersion .
surjectivity of @xmath1374 implies transverseness .
injectivity of @xmath1368 on @xmath18 implies embeddedness . for the final claim ,
consider the long exact sequence for the pair @xmath1375 : @xmath1376 by assumption @xmath1377 is torsion , so also @xmath1378 is .
so @xmath466 is surjective modulo torsion .
torsion does not affect maslov indices as the maslov index is additive on relative @xmath1379 .
so it suffices to verify monotonicity on spheres in @xmath1213 ( discs mapping the boundary to a point in @xmath18 ) .
so monotonicity of @xmath18 follows from the monotonicity of @xmath1213 ( recall [ subsection notation ] ) .
[ example coisotropic ] let @xmath363 be a closed coisotropic submanifold .
then @xmath1380 is an integrable distribution by ( * ? ? ?
* lemma 5.33 ) , so @xmath365 is foliated by isotropic leaves .
assume @xmath365 is _ regular _ : the leaves are connected compact submanifolds of @xmath365 .
denote by @xmath1381 the quotient @xmath1382 , identifying points lying on the same leaf .
then @xmath1383}q = t_c c / t_c c^{\omega}$ ] and @xmath1381 inherits a symplectic form @xmath1384 which is natural in the sense that @xmath1385 ( * ? ? ?
* lemma 5.35 ) .
also @xmath365 naturally gives rise to a lagrangian submanifold @xmath1386 where @xmath1387 is @xmath1381 with the symplectic form @xmath1388 . by construction @xmath1389
is good , and @xmath1390 is the map which takes the union of leaves corresponding to points in @xmath22 .
[ lemma coisotropic ] let @xmath363 be a regular coisotropic with reduction @xmath1381 as above .
suppose @xmath1391 is monotone . by rescaling ,
assume @xmath1392 , so also @xmath1393 is monotone with @xmath1394 .
assume further that @xmath1395 has torsion kernel .
then @xmath18 is monotone and very good . since @xmath1396 , @xmath1397 , the map @xmath1398
can be identified with @xmath1372 .
so monotonicity of @xmath18 follows by lemma [ lemma properties of good gamma ] .
suppose @xmath1399 is a monotone lagrangian .
we now verify monotonicity for a disc @xmath1400 .
consider a neighbourhood @xmath1401 of @xmath363 .
we can assume @xmath1401 retracts onto @xmath365 ( so @xmath1402 is fixed ) and @xmath1403 intersects @xmath241 transversely along some circles .
the concatenation of these circles is trivial in @xmath1404 since it is homotopic to @xmath352 , which has a null - homotopy @xmath1405 in @xmath0 .
since @xmath1395 has torsion kernel , we can assume the concatenation is null - homotopic in @xmath1401 ( it suffices to verify monotonicity on a multiple of the class @xmath1016\in \pi_2(e,\mathcal{l})$ ] ) . by adding and subtracting the null - homotopy
, we can assume @xmath241 is a gluing of a disc @xmath1406 and some spheres in @xmath0 .
for spheres monotonicity follows from monotonicity of @xmath0 , so we can now assume @xmath1407 . by retracting @xmath1401
we can assume @xmath1408 .
so @xmath241 induces @xmath1409 where @xmath1410 is @xmath241 composed with an orientation reversing isomorphism of @xmath413 . by additivity for maslov indices , @xmath1411 , and for areas : @xmath1412 .
but @xmath1413 are monotone with @xmath1414 ( see [ subsection monotonicity assumptions ] ) so monotonicity of @xmath20 with @xmath1415 follows . if @xmath1370 is a smooth fibre bundle and @xmath1371 is injective , then @xmath18 arises by reduction from the coisotropic submanifold @xmath1416 .
define @xmath1417 so that @xmath1418 .
so @xmath1419 .
via the diffeomorphism @xmath1420 any @xmath1421 gives rise to a unique @xmath1422 with @xmath1423 .
since @xmath18 is lagrangian , @xmath1424 . to prove that @xmath365 is coisotropic , it remains to show that @xmath1425 , and since the inclusion @xmath1426 is clear it suffices to compare their real dimensions . via the diffeomorphism , @xmath1427 . since @xmath1428 is a fiber bundle , @xmath1429 .
finally : @xmath1430 . thus the dimensions agree since @xmath1431 .
we discuss in detail in [ subsection additivity of the m0 values ] that the next lemma makes sense when @xmath1432 because then also @xmath1433 .
in particular , we ensure that @xmath1327 have the same monotonicity constant @xmath1216 by rescaling @xmath565 to @xmath1434 .
[ lemma from generalized fukaya to fukaya ] for good @xmath18 , @xmath1435 this is a standard consequence of the `` strip - shrinking '' theorem of wehrheim - woodward @xcite , but we include a sketch of the key ideas of the proof for the convenience of the reader .
consider the first quilt in figure [ figure phi quilt 2 ] . recall that the generators of the floer chain group for @xmath1436 are constant such quilts , and the differential is defined by counting the @xmath49-dimensional moduli spaces ( modulo @xmath240-translation ) of such quilts .
consider shrinking the width of the strips mapping into @xmath15 ( so we are deforming the domain of the quilt maps ) .
a standard parametrized moduli space argument shows that the floer chain groups change by a quasi - isomorphism if we change the width . wehrheim - woodward show that when the width is very small , the quilt maps become approximately constant on the thin strips .
an implicit function theorem argument then shows that there is a bijection between the quilt maps for small width and the second type of quilt maps in figure [ figure phi quilt 2 ] where the pair of seams @xmath291 and @xmath18 are replaced by the composite seam @xmath1437 .
therefore , the generators can be canonically identified , and under this identification the floer differentials agree .
so the floer cohomologies are isomorphic .
the lemma proves @xmath1438 , and for generalized correspondences of type @xmath1439 the @xmath1440 are defined exactly like the @xmath1441 maps .
thus [ lemma phi lands in f(e ) ] @xmath163 lands in a full subcategory of @xmath1349 which is quasi - equivalent to @xmath42 . by shrinking one strip at a time in the proof of the previous lemma , defines @xmath1442 & \cong & e_{(l,\gamma ) } , \end{array}\ ] ] where @xmath1443 are identity elements .
similarly , one obtains @xmath1444\in hf^*(\mathcal{l},(l,\gamma))$ ] . a gluing argument for quilts and
a strip - shrinking argument shows that @xmath1445 on cohomology , and similarly @xmath1446 .
so @xmath20 and @xmath1447 are quasi - isomorphic in @xmath1349 .
consider the two full subcategories @xmath1448 of @xmath1349 generated respectively by all objects @xmath1369 and @xmath1447 .
since each @xmath20 is quasi - isomorphic to @xmath1447 , the inclusions @xmath1449 and @xmath1450 are quasi - equivalences of @xmath91-categories ( seidel ( * ? ? ?
* sec.10a ) ) . therefore @xmath1448 are quasi - equivalent , and the claim follows .
we will assume @xmath1327 have the same monotonicity constant @xmath1216 by rescaling @xmath565 to @xmath1434 .
for good @xmath18 , we always abbreviate @xmath1369 ( see [ subsection phi lands in fuk e ] ) .
[ lemma additivity of m0 values ] for very good @xmath18 , if @xmath1451 are monotone oriented with equal @xmath271-values then the @xmath1452 are monotone oriented with equal @xmath271-values , indeed : @xmath1453 the lemma is a consequence of the strip - shrinking theorem of wehrheim - woodward ( * ? ? ?
5.4.2(b ) ) .
we sketch the argument .
consider the floer chain complex @xmath1454 of the cyclic correspondence @xmath1455 .
note the lagrangians are monotone and orientable , and the symplectic manifolds @xmath1327 are monotone with the same monotonicity constant @xmath1216 .
consider the floer differential @xmath1456 : the obstruction to @xmath1457 are disc bubbles of maslov index @xmath310 ( see [ subsection the role of monotonicity ] ) , giving rise to : @xmath1458
if @xmath1459 have the same @xmath271-values then @xmath1457 so @xmath1460 is well - defined . by strip - shrinking the pair @xmath1461 ( see the proof of lemma [ lemma from generalized fukaya to fukaya ] ) , one deduces that @xmath1457 also for the cyclic correspondence @xmath1462 , and strip - shrinking @xmath1463 we obtain @xmath1457 for @xmath1464 .
thus : @xmath1465 _ remark .
our @xmath1466 is defined over the novikov ring , whereas in ( * ? ? ?
5.4.1 ) the notation @xmath1467 refers to the same maslov 2 disc count over the integers .
the two are related by : @xmath1468 since this lies in grading @xmath310 ( see [ subsection monotonicity assumptions ] ) _ [ corollary phi functor respects m0 values ] for very good @xmath18 , the functor @xmath1469 respects @xmath271-values : @xmath1470 so @xmath1471 can only be non - zero on summands with @xmath1472 and @xmath1473 ( compare lemma [ lemma fukaya cats split according to evalues ] ) .
recall from section [ section fukaya with local systems ] the construction of the ( wrapped ) fukaya category with local systems .
we now briefly explain why our results for @xmath208 will hold also for @xmath971 .
inserting local systems of coefficients does not affect the moduli spaces of solutions that we want to count , it only changes the weight with which we count solutions .
so it suffices to show that the counts are consistent with breaking . just as for @xmath988 ,
the key observation is that once the asymptotic data is chosen and we consider any family of solutions with that asymptotic data , then the total weight associated with solutions in the family is constant , even for broken solutions arising in the boundary strata of the family .
this is because by ( 2 ) in [ subsection lagrangians with local systems ] the weight is constant along a @xmath49-family of paths with fixed ends , and by ( 3 ) this holds even if the path breaks ( the weight of a concatenation of paths equals the product of the weights for the components of the broken path ) . finally , consider the new functor @xmath1474 , where for now @xmath1475 is the generalized fukaya category for @xmath0 with local systems .
for simplicity , we will not put a local system on @xmath18 ( one can easily do this as well ) .
the lagrangian boundary conditions along the outermost seams of the quilt domains involved in @xmath163 are always objects @xmath1476 , so it suffices to keep track of the local systems on the @xmath291 .
so we only need to comment on [ subsection phi lands in fuk e ] , when we compose @xmath1369 .
we now assume @xmath18 is good .
then a local system on @xmath22 induces one on @xmath18 by pulling back via the projection @xmath1477 .
since @xmath1478 is injective , we can uniquely lift to @xmath18 any paths and homotopies in @xmath20 , so the local system on @xmath18 induces one on @xmath20 ( equivalently , we are taking the direct image of sheaves ) .
we need to show that this canonical choice is compatible with strip - shrinking .
suppose a quilt map @xmath241 has a seam condition @xmath1479 , where @xmath466 is a path on the quilt , then in the limit of strip - shrinking one obtains @xmath1480 for some lagrangian @xmath291 ( after subdividing the arc @xmath466 according to the lagrangian targets ) . by construction , @xmath1481 = b_{\gamma}[(u_{\overline{b}}(\gamma),u_e(\gamma ) ) ] = b_{l_i}[u_{\overline{b}}(\gamma)]$ ] .
so the parallel transports equal in @xmath973 .
thus , after strip - shrinking , we obtain @xmath1482 , which on objects is given by @xmath1483 and lemma [ lemma phi lands in f(e ) ] becomes : @xmath1484 .
we now continue with @xmath1485 , without local systems , and we do not assume that @xmath18 is good .
the bimodule maps @xmath1486 for @xmath147 are analogous to those for @xmath92 : in figure [ figure bimodule compositions for b ] replace @xmath15 by @xmath0 , and replace @xmath1487 by lagrangian correspondences @xmath1488 involving any monotone symplectic manifolds @xmath1489 .
the @xmath91-relations for @xmath1486 are proved just as for @xmath92 ( in figure [ figure ainfinity eqns for bimodule b ] make the analogous replacements ) . by lemma [ lemma change of rings ] , the functor @xmath140 induces a change of rings which turns @xmath147 into a @xmath92-bimodule @xmath1491 .
recall that the new composition maps @xmath1492 involve composing the @xmath1493 maps with several @xmath1494 maps , see figures [ figure bimodule e before gluing ] and [ figure bimodule e after gluing ] .
+ [ figure ainfinity for new e ] define @xmath1496 by counting quilt maps as shown in figure [ figure bimodule morphism ] .
+ [ figure ainfinity f ] by corollary [ corollary induced map on hh ] , the morphism @xmath1498 of @xmath92-bimodules together with the functor @xmath140 induce a homomorphism @xmath1499 by mimicking [ subsection open closed map oc_b ] , figure [ figure oc map ] , define @xmath1500 as in [ subsection bimodule structure for e over b ] , one replaces the @xmath291 in figure [ figure oc map ] by lagrangian correspondences @xmath1501 . the fact that @xmath1502 is decreasing ensures a maximum principle ( [ subsection moduli space of pseudoholo maps in wrapped case ] ) prohibiting solutions from escaping to infinity ( the radial coordinate is bounded by the maximum that the radial coordinate attains on the ( compact ! ) lagrangian correspondences or on the hamiltonian orbit at @xmath1503 ) .
so far , we have defined an @xmath91-functor @xmath1505 , and a morphism @xmath1506 where we emphasize the dependence on @xmath18 .
we now deform @xmath1507 to @xmath1508 depending on a class @xmath466 in @xmath1268 in analogy with subsection [ subsection deformed quilt morphism ] .
fix a cycle @xmath466 representing a poincar dual class in @xmath1509 .
introduce a new fixed marked point @xmath1255 on the seam @xmath18 in figure [ figure bimodule morphism ] .
we impose the new condition that the quilt maps @xmath241 must intersect @xmath466 at the new marked point ( after folding @xmath241 near @xmath1255 to obtain a local map into @xmath1213 ) .
more precisely , we have an evaluation map @xmath1510 defined on each moduli space of such quilt maps ( having imposed boundary conditions , and having fixed a homology class of such maps ) . we pick @xmath466 generically so that @xmath1510 is transverse to it , and finally we consider @xmath1511 .
the map @xmath1512 counts the resulting zero dimensional components .
we emphasize that the marked point @xmath1255 is fixed , so the @xmath91-relations and the compatibility with the ( undeformed ) bimodule structure continue to hold and are proven in exactly the same manner . the original map @xmath407 is given by @xmath1513}$ ] .
there is a commutative diagram @xmath1514 \ar@{->}_{\mathrm{oc}_b}[d ] & \mathrm{hh}_*({\mathcal{b}},\widetilde{{\mathcal{e } } } ) \ar@{->}^{\widetilde{\mathrm{oc}}_e}[d ] \ar@{->}^{\mathrm{tautological}}[r ] & \mathrm{hh}_*({\mathcal{e } } ) \ar@{->}^{\mathrm{oc}_e}[d ] \\
hf^*(h^b ) \ar@{->}[r]^-{h^*(q_{\gamma,\gamma } ) } & hf^*(h^e ) \ar@{->}[r]^-{\mathrm{identity } } & hf^*(h^e ) } \ ] ] using the tautological map from lemma [ lemma change of rings ] , and the map @xmath1515 defined by counting quilts shown on the left in figure [ figure diagram 1 ] . consider the square on the left in the diagram .
the composite @xmath1516 is shown on the left in figure [ figure commutativity 5 ] ( after gluing ) .
observe this is equivalent to the quilt on the right .
now consider the disc on the right in figure [ figure commutativity 5 ] , except we make a hyperbolic translation of the seam - circle : we assume it is a hyperbolic circle , whose hyperbolic centre is the interior puncture @xmath1517 which we allow to freely move on the geodesic joining @xmath1518 , and we fix the boundary point @xmath1255 to lie at @xmath1519 .
if @xmath1517 is at @xmath194 we get precisely the quilt on the right in figure [ figure commutativity 5 ] . in a @xmath49-family with fixed auxiliary data ,
if @xmath1517 converges to @xmath380 then we obtain the broken configuration on the right in figure [ figure diagram 1 ] .
there are other possible degenerations in the above two cases .
let @xmath960 be the count of isolated configurations shown in figure [ figure commutativity 5 ] . in a @xmath49-family ,
when a subset of the boundary punctures converge together , we get contributions to @xmath1520 where @xmath728 is the bar differential for @xmath1521 , and when a floer cylinder breaks off at the puncture @xmath1517 we get contributions to @xmath1522 .
so the difference at the chain level between the two composites in the left square is counted by @xmath1523 , so the composites agree on homology , and the square commutes .
now consider the right square in the diagram .
the composite of the tautological map and the @xmath74 map is precisely the count of quilt maps shown in figure [ figure bimodule e before gluing ] except we replace the output dash at @xmath380 by an interior puncture @xmath1517 at @xmath616 .
so the proof that the square on the right commutes is analogous to the proof of the @xmath91-equations for the functor @xmath163 shown in figure [ figure ainfinity for phi ] ( except that figure is rotated by @xmath1524 so that the output dash at @xmath380 becomes our input dash at @xmath181 and we have additionally an interior puncture @xmath1517 at @xmath1519 or @xmath194 ) .
[ theorem second commutative diagram ] for very good @xmath1525 ( definition [ definition very good gamma ] ) , there is a commutative diagram @xmath1514 \ar@{->}_{\mathrm{oc}_b}[d ] & \mathrm{hh}_*({\mathcal{e } } ) \ar@{->}^{\mathrm{hh}_*(\mathcal{a})}[r ] \ar@{->}_{\mathrm{oc}_e}[d ] & \mathrm{hh}_*({\mathcal{w}(e ) } ) \ar@{->}^{\mathrm{oc}_e}[d ] \\
hf^*(h^b ) \ar@{->}[r]^-{h^*(q_{\gamma,\gamma } ) } & hf^*(h^e ) \ar@{->}[r]^-{c^ * } & sh^*(e ) } \ ] ] in particular , @xmath628 is a ring homomorphism ; @xmath1526 , @xmath40 , @xmath74 are @xmath14-module maps ; and the second @xmath74 is an @xmath29-module map . consider the diagram from section [ subsection the first commutative diagram ] .
we can replace @xmath1343 by @xmath1527 by lemma [ lemma phi lands in f(e ) ] since @xmath18 is very good .
in particular , we use that the @xmath1528 of the proof of lemma [ lemma phi lands in f(e ) ] have isomorphic @xmath1529 since they are quasi - equivalent @xmath91-categories ( the @xmath91-analogue of morita equivalence ) .
moreover , a strip - shrinking argument as in [ lemma from generalized fukaya to fukaya ] shows that the @xmath74 maps on cohomology commute with the isomorphism @xmath1530 .
this yields the left square and proves that it commutes .
the square on the right commutes by theorem [ theorem acceleration commutative diagram ] .
our previous discussion always assumed that @xmath0 was a ( non - compact ) monotone symplectic manifold conical at infinity .
however , several of the results hold also if @xmath0 is a closed monotone symplectic manifold in fact they become easier to prove since they do not require accelerating the hamiltonian , so much of the auxiliary data which decorates the disc counts can be omitted .
we state the two main results , and in the following sections we will show applications thereof . [ theorem generation for compact categories ] let @xmath15 be a closed monotone symplectic manifold , and let @xmath1023 a full subcategory of @xmath39 . if the composite @xmath1031 & \mathrm{hh}_*({\mathcal{f}}(b ) ) \ar@{->}[r]^-{\mathrm{oc}_b } & qh^*(b ) , } \ ] ] hits an invertible element , then @xmath1023 split - generates @xmath39 .
the analogous statement holds if one restricts to generalized eigensummands by the remark in theorem [ theorem oc is a module map ] and remark [ remark about idempotent summands ] .
[ theorem second commutative diagram in compact setup ] let @xmath1531 be closed monotone symplectic manifolds . for very good @xmath1532 ( definition [ definition very good gamma ] ) , and any @xmath27 , there is a commutative diagram @xmath1533 \ar@{->}_{\mathrm{oc}_{b_1}}[d ] & \mathrm{hh}_*(\mathcal{f}(b_2 ) ) \ar@{->}_{\mathrm{oc}_{b_2}}[d ] \\
qh^*(b_1 ) \ar@{->}[r]^-{q_{\gamma,\gamma } } & qh^*(b_2 ) } \ ] ] in particular , @xmath1534 , @xmath1535 are @xmath14-module maps .
we briefly recall some basic facts from algebraic topology .
firstly , the pull - back @xmath1541 is injective , since @xmath1542 where @xmath1543 sends a chain @xmath843 to the sum of its lifts @xmath1544 .
secondly , by naturality of the first chern class , @xmath1545 in particular @xmath1546 have the same monotonicity constant ( note that holomorphic spheres lift , since @xmath1547 ) .
one can verify directly that the pull - back @xmath1548 preserves the quantum product , since spheres always lift since @xmath1547 .
the corollary can easily be reformulated at the floer level ( recall that @xmath1550 for closed monotone symplectic manifolds ) .
the pull - back map at the floer level is given by summing the possible lifts of a generator : @xmath1551 at the floer level , one can see that the pair - of - pants product is indeed preserved since there is a bijection between the pair - of - pants moduli spaces : @xmath1552 ( once you fix the lift @xmath1553 of an asymptote of a pair - of - pants map @xmath1554 , there is a unique lift of @xmath241 to @xmath1540 ) .
the fact that @xmath1555 preserves the pop - product then follows since in @xmath1540 the pop - moduli spaces have a free @xmath62-action given by applying the deck transformations to the pop - maps @xmath1556 .
let @xmath18 be the transpose of the graph of @xmath1557 , @xmath1558 where @xmath1559 means @xmath1029 with the opposite symplectic form and opposite almost complex structure .
then @xmath18 is very good ( definition [ definition very good gamma ] ) .
it suffices to verify that lemma [ lemma properties of good gamma ] applies .
the projection @xmath1560 is a smooth fibre bundle since it is a covering .
the projection @xmath1561 is injective .
finally , @xmath1562 has torsion kernel since @xmath1563 includes into @xmath1564 .
the diagram from theorem [ theorem second commutative diagram in compact setup ] , with no marked point on the correspondence @xmath18 ( or equivalently , taking @xmath1264 $ ] ) , becomes @xmath1565 \ar@{->}_{\mathrm{oc}_m}[d ] & \mathrm{hh}_*({\mathcal{f}}(\widetilde{m } ) ) \ar@{->}_{\mathrm{oc}_{\widetilde{m}}}[d ] \\
qh^*(m ) \ar@{->}[r]^{q_{\gamma } } & qh^*(\widetilde{m})}\ ] ] by theorem [ theorem qgamma morphism ] , the lower horizontal map @xmath1567 is determined by the intersection product with @xmath1291 $ ] : @xmath1568)\ , \beta^{\vee}\ ] ] where @xmath1569 , @xmath1570 , @xmath1571 is the intersection dual to @xmath473 , and the sum is over a basis of cycles @xmath473 representing @xmath1572 . for dimension reasons we need @xmath1573 .
thus @xmath1574 ) = ( ( [ \overline{m}],[\mathrm{pt}])\bullet [ \gamma])\ , [ \mathrm{pt}]^{\vee } = [ \widetilde{m } ] = 1 \in qh^*(\widetilde{m})\ ] ] since they intersect once transversely at @xmath1575 .
let @xmath1585 be a closed real @xmath88-dimensional symplectic manifold together with an effective hamiltonian action of the @xmath654-torus @xmath1586 .
this action determines a _ moment map _
@xmath1587 , which is determined up to an additive constant ( we tacitly identify @xmath1588 with the dual of the lie algebra of @xmath1586 ) .
the image @xmath1589 is a convex polytope , called the _
moment polytope_. by delzant s theorem @xmath1046 determines , up to isomorphism , @xmath1585 together with the action .
the moment polytope has the form @xmath1590 where @xmath1591 are parameters and @xmath1592 are the primitive inward - pointing normal vectors to the codimension 1 faces of @xmath1046 . for delzant s theorem to hold , we always assume that at each vertex @xmath764 of @xmath1046 there are exactly @xmath654 edges of @xmath1046 meeting at @xmath764 ; that the edges are rational , that is they are of the form @xmath1593 for @xmath1594 ; and that these @xmath1595 are a @xmath512-basis for @xmath1596 . _
* example : @xmath1597 . *
the natural rotation action of @xmath1598 on the entries of @xmath1599 extends to @xmath1600 .
choosing @xmath1601 and @xmath1602 , the associated moment map is @xmath1603 and the moment polytope is @xmath1604 _ to the moment polytope one associates the _ landau - ginzburg superpotential _
@xmath1605 , where @xmath8 is the novikov ring defined over @xmath257 , given by @xmath1606 _
* example : @xmath1597 .
* using the above choices , @xmath1607 .
_ let @xmath1609 denote the valuation for the @xmath191-filtration , whose value on a laurent series is the lowest exponent of @xmath191 arising in the series .
the condition @xmath1610 is equivalent to the equations of the moment polytope .
so restricting to those @xmath1611 which satisfy that positivity condition , each @xmath51 determines a toric lagrangian @xmath1612 by taking @xmath1613 , together with a choice of holonomy around each generating circle of the torus @xmath1612 given by @xmath1614 ( recall @xmath973 from section [ subsection novikov rings for local systems ] ) .
abbreviate by @xmath50 the lagrangian @xmath1612 with the holonomy determined by @xmath51 . from now on we work with local systems in the sense of section [ subsection lagrangians with local systems ] .
we will only consider cases where @xmath174 is monotone , so in particular @xmath50 does not bound any non - constant holomorphic discs of maslov index strictly less than @xmath310 . as a consequence of the machinery of fukaya - oh - ohta - ono @xcite ( see the sketch proof of auroux ( * ? ? ?
* lemma 6.4 ) ) , @xmath1615\ ] ] is a multiple of the fundamental class ( see [ subsection the role of monotonicity ] and [ subsection lagrangians with local systems ] ) and therefore @xmath1616 is well - defined .
moreover , @xmath1616 is a ring via the @xmath331-product with unit @xmath1617 $ ] , and @xmath1616 as a module over @xmath1618 satisfies @xmath1619 = m_0(l_z)\ , [ l_z].\ ] ] the superpotential @xmath1620 in fact expresses the same count of maslov @xmath310 discs as @xmath1621 ( see auroux ( * ? ? ?
* definition 3.3 ) ) , so @xmath1622 the key relationship between critical points of @xmath1620 and the floer cohomology of the @xmath50 is the following result of cho - oh ( see also the discussion thereof in ( * ? ? ?
* proposition 6.9 ) ) .
[ theorem pt class is cycle for crit lag ] @xmath1623 if and only if @xmath1624 , indeed this holds if and only if the point class @xmath1267\in c_*(l_z)$ ] is a floer cycle in @xmath1616 . in particular , if this holds , then @xmath50 is not displaceable by a hamiltonian isotopy . the original application which we had in mind , which spurred the general results of this foundational paper , was to apply our machinery in the toric setting to negative line bundles in order to prove generation statements for the wrapped category of the total space in terms of generation statements for the fukaya category of the base .
a general treatment is given in the sequel @xcite , but here we indicate why the situation is more delicate than one might at first suspect .
recall that a complex line bundle @xmath1625 over a closed symplectic manifold @xmath554 is called a _ negative line bundle _
if @xmath1626 \ ; \textrm { for some real } k>0.\ ] ] by convention , @xmath1627 will always mean the first chern class of the line bundle @xmath0 , whereas @xmath586 denotes the first chern class of the total space @xmath0 .
we remark that a symplectic manifold @xmath554 admits a hermitian line bundle @xmath1628 with a connection with curvature form @xmath1629 if and only if @xmath1630 $ ] is in the image of @xmath1631 ( the map induced by the exact sequence @xmath1632 ) .
as explained for example in @xcite ( taking @xmath1633 in ( * ? ? ? * sec.7.2 ) ) , the negativity condition ensures that there is a symplectic form @xmath228 on @xmath0 such that @xmath1634 so @xmath1635 \equiv k[\omega_b]$ ] via @xmath1636 .
it is shown carefully in @xcite that @xmath0 is a symplectic manifold conical at infinity ( see [ subsection symplectic manifolds conical at infinity ] ) , and the contact hypersurface @xmath231 can be chosen to be any sphere subbundle of @xmath0 .
the reeb flow is given by the @xmath659-rotation in the fibres .
we always assume that we are in the monotone setting : @xmath1637(a),\;\ ; \forall a\in \pi_2(e ) \qquad \textrm { some } \lambda_e>0.\ ] ] this is equivalent , via @xmath1638 , to the condition that @xmath15 is monotone with @xmath1639(a),\;\ ; \forall a\in \pi_2(b ) , \textrm { where } 0 < k < \lambda_b.\ ] ] our motivation for studying this particluar class of non - compact manifolds , is because it is the only family of examples outside of exact cotangent bundles where the symplectic cohomology has been computed : typically one can not compute @xmath29 unless the relevant moduli spaces of solutions to the floer equation are empty , because one can rarely solve the floer equation explicitly for a generic almost complex structure . in the case of negative line bundles
the relevant counts of moduli spaces can be rephrased in terms of gromov - witten invariants involving explicit obstruction classes , which can be computed by algebro - geometric techniques .
[ lemma description of gamma for line bdles ] let @xmath1217 be the @xmath659-bundle over the diagonal @xmath1648 which fibrewise is the circle of radius @xmath1649 .
then @xmath18 is monotone and very good ( definition [ definition very good gamma ] ) .
apply example [ example coisotropic ] to the coisotropic submanifold @xmath1650 given by the sphere subbundle of radius @xmath1651 .
then @xmath1217 is the @xmath659-bundle over @xmath1652 which fibrewise is the circle of radius @xmath1651 .
also @xmath1653 , and via the projection @xmath1654 we may identify @xmath1655 to make @xmath1393 monotone , we want @xmath1656 , so choose @xmath1651 so that @xmath1657 .
finally , we check @xmath1658 has torsion kernel , so the claim will follow by lemma [ lemma coisotropic ] .
the @xmath659-fibration @xmath1659 yields @xmath1660 the map @xmath1661 is in fact evaluation by the euler class of @xmath0 after identifying @xmath1662 ( this can be checked by comparing the gysin sequence for @xmath1663 and the long exact sequence for the pair @xmath1664 , for any sphere @xmath1665 ) . in our case , @xmath1666\neq 0 \in h^2(b)$ ] , so @xmath1667 is torsion .
finally , @xmath1668 has torsion kernel since the composite @xmath1669 has torsion kernel @xmath1670 . for our @xmath18 , in theorem
[ theorem qgamma morphism ] we obtain @xmath1253 since @xmath1291=0 $ ] ( a homology is given by the disc bundle over @xmath1652 ) .
so instead we wish to use the @xmath1262 map of [ subsection deformed quilt morphism ] for the class @xmath1671 using the poincar duality @xmath1672 .
it turns out that the lowest - order @xmath191 term in @xmath1673 counts the standard maslov @xmath310 disc in a reference fibre @xmath1674 of @xmath0 bounding the circle @xmath659 in the sphere bundle @xmath1675 of radius @xmath1649 .
thus : @xmath1676 where @xmath1677 is a normalization constant : @xmath1678 = c\ , [ \omega_b^{\dim_{{\mathbb{c}}}b}]$ ] .
suppose that the base @xmath15 has the property that @xmath1679 hits an invertible for some full subcategory @xmath1680 .
since @xmath40 is a @xmath1681-module map , it also hits @xmath49 . thus if we could verify that @xmath1673 is invertible , then the diagram of theorem [ theorem second commutative diagram ] would imply that @xmath1682 hits an invertible
this in turn implies by theorem [ theorem generation in monotone setting ] that @xmath1683 split - generates @xmath4 .
unfortunately , without performing further deformations of the correspondence @xmath18 ( i.e. allowing several marked points with cycle conditions in @xmath1268 , rather than just one marked point ) it appears that @xmath1673 vanishes in @xmath33 , at least in the case @xmath1684 . for the sake of space
, we will not discuss this phenomenon in detail here , nor will we pursue the aspect of these further deformations we will relegate such discussions to later work @xcite .
we will however state the following general result about negative line bundles , which we prove in @xcite by a careful analysis of the superpotential , which shows that the wrapped category for a large class of negative line bundles is non - trivial .
then @xmath1686 is non - nilpotent and @xmath1687 .
moreover , there is a non - displaceable monotone lagrangian torus @xmath1688 with @xmath1689 and @xmath1690 .
if @xmath17 is the lagrangian torus corresponding to the barycentre of the moment polytope of @xmath15 , then geometrically @xmath1691 is the torus in the sphere bundle @xmath1675 lying over @xmath1355 .
we will not discuss this result since it relies heavily on computations with the superpotential , which are beyond the purpose of this paper .
rather , for the purposes of this paper , we wish to give a quick and elegant concrete application of the @xmath33-module structure of the @xmath1-map and of the generation criterion in the monotone setting .
[ theorem qh and sh of o(-k ) ] let @xmath1696 with @xmath1697 .
then @xmath587 and @xmath33 are generated over @xmath8 by @xmath1698\in qh^*(e)$ ] subject to missing in version 1 of @xcite , which will be corrected in revision 2 .
the sign arises in theorem 63 : @xmath1699 .
this is because , within the proof , @xmath1700 should be @xmath1701 because the duality sign on the @xmath1702 reverses the sign of the weights.]@xmath1703 used in this paper corresponds to the @xmath191 used in @xcite . in @xcite , @xmath1704\in h_2({\mathbb{p}}^m,{\mathbb{z}})$ ] .
but in the conventions of this paper , this variable comes with a power of the area @xmath1705}\omega_e
@xmath1706 for @xmath1707 , @xmath1708 _ ( lower order _
_ terms)__@xmath1709 in @xmath587 .
recall that @xmath1710/(w^{1+m } - t)$ ] , @xmath1711 , where @xmath1712 is the novikov variable for @xmath1713 .
so the eigenvalues of @xmath1714 on @xmath1681 are @xmath1715-th roots of unity ( up to a fixed constant factor ) . on the other hand , for @xmath1716
the above calculation shows that the eigenvalues of @xmath1022 on @xmath587 are @xmath1717-th roots of unity ( up to a constant factor ) .
the eigenvalues in this example highlight the difficulty mentioned in section [ subsection generation results for toric negative line bundles ] , regarding the necessity of further deformations of @xmath18 .
indeed , by the additivity formula in lemma [ lemma additivity of m0 values ] and corollary [ corollary phi functor respects m0 values ] , the functor @xmath1471 translates @xmath271 values by @xmath1718 . in general , therefore , one would expect @xmath1471 to be non - trivial on at most one summand @xmath1719 ; in many cases one sees that @xmath18 induces the zero functor . as a toric variety
, @xmath1684 can be described by the fan whose edges in @xmath1720 are given by @xmath1721 where @xmath1722 are the edges for the fan for @xmath1600 : @xmath1723 , @xmath490 , @xmath1724 , @xmath1725 .
the moment polytope for @xmath1600 is given by @xmath1726 , and the polytope for @xmath0 is @xmath1727 the landau - ginzburg superpotential associated to this polytope is @xmath1728 the critical points of @xmath1729 are the following : for each solution @xmath283 of @xmath1730 , there is one critical point : @xmath1731 the jacobian ring @xmath1732/(\partial_{z_1}w_e , \ldots , \partial_{z_{m+1}}w_e)$ ] has rank @xmath1733 : @xmath1734/(w^{1+m - k}-(-k)^k t^k).\ ] ] now @xmath39 is generated by the lagrangian toric fibre @xmath1737 over the barycentre of the polytope for @xmath1713 , taken with finitely many appropriate local systems , with one local system for each eigenvalue of @xmath1738 ( corresponding to the @xmath271 value ) .
these local systems on the monotone torus @xmath1737 arise precisely as the critical points of the superpotential for @xmath15 , which is @xmath1739 .
a calculation shows that the lift @xmath1740 of @xmath1737 to the sphere bundle @xmath1675 in @xmath1716 is a monotone lagrangian toric fibre of @xmath0 and that the critical points of @xmath1729 are precisely @xmath1740 with finitely many local systems .
the @xmath271-values for @xmath1740 with these local systems are precisely the critical values @xmath1741 above , so they exhaust all non - zero eigenvalues of @xmath1022 . again , this highlights the necessity of further deformations of @xmath18 , since the the naive lift of @xmath1737 with its local systems to the sphere bundle would be @xmath1740 with local systems which necessarily are trivial in the last factor .
so the holonomies are not being lifted correctly by the functor @xmath1471 as written , and this is consistent with the issue of the additivity of the @xmath271-values mentioned in the previous remark . 1 . the wrapped category @xmath208 is compactly generated , indeed split - generated by one lagrangian torus @xmath1743 with @xmath1733 choices of holonomy ; 2 .
@xmath208 is cohomologically finite ( so @xmath1744 for any objects @xmath1459 ) ; 3 .
@xmath1745 , so @xmath20 is not hamiltonianly displaceable . by theorem [ theorem qh and sh of o(-k ) ] , for non - zero eigenvalues @xmath347 ,
the eigensummands of @xmath1627 acting on @xmath587 are @xmath49-dimensional over the novikov field @xmath1746 of formal laurent series in @xmath191 .
therefore @xmath587 splits as a direct sum of the generalized @xmath194-eigenspace and @xmath49-dimensional field summands corresponding to the non - zero eigenvalues .
the map @xmath195 is the quotient by the generalized @xmath194-eigenspace , by theorem [ theorem about sh of neg l b ]
. then @xmath1748 satisfies @xmath1749 ) = -kw\cdot \pi^*\omega_{{\mathbb{p}}^m}^m + ( \textrm{higher order } t\textrm { terms } ) \neq 0 \in qh^*(e),\ ] ] where the first term arises from the regular maslov 2 disc in the fibre ( the standard disc filling @xmath1750 ) , and @xmath1751 is the holonomy for the @xmath659-fibre direction of @xmath20 . observe that there is no order @xmath967 term because the constant disc would sweep the lf - cycle @xmath1267\in h_0^{lf}(e)\cong h^{2m+2}(e)=0 $ ] .
the @xmath1752 factor arises because the fibre disc intersects once transversely the cycle intersection condition @xmath1753\in c_*(e)$ ] , and the intersection dual of this is the lf - cycle @xmath1754\in h_2^{lf}(e)$ ] which corresponds to @xmath1755 ( the fibre is the preimage of a point in the base , and taking preimages is poincar dual to pull - back on cohomology ) .
it is a general feature of the toric setting that the maslov index of a disc bounding a toric lagrangian is twice the number of intersections of the disc with the anticanonical divisor ( see auroux ( * ? ? ?
* lemma 3.1 ) ) . in our case , the anticanonical divisor viewed as an lf - cycle , consists of the base @xmath1600 together with the preimage under projection of the canonical divisor in the base @xmath1600 .
so the fibre disc is the only maslov two disc that can intersect the cycle @xmath1753 $ ] . by monotonicity ,
all other discs with maslov index greater than @xmath310 will contribute higher order @xmath191 terms .
this proves that @xmath1756)\neq 0 $ ] ( in particular , this uses that @xmath1267 $ ] is a cycle by theorem [ theorem pt class is cycle for crit lag ] ) .
the key observation is to consider the quantum module structure of the @xmath1 map and the eigensummands @xmath44 of @xmath587 : since the critical points of the superpotential show that there is such a lagrangian @xmath20 for each eigensummand of @xmath208 , this proves the claim .
the finite - dimensionality claim follows because the split - generators are closed lagrangians .
in particular , @xmath1757 is generated at the chain level by the finitely many intersections of @xmath20 with a hamiltonianly deformed copy of @xmath20 .
we emphasise that although the wrapped category is cohomologically finite , it does contain objects which are non - compact lagrangians such as real line bundles over suitable lagrangians in the base or lefschetz thimbles for appropriate lefschetz fibrations whose wrapped floer cochain complexes are infinitely generated .
the module structure of the @xmath1-map played a crucial role : without this module structure one would need to compute all the terms in the expansion of @xmath1756)$ ] .
it turns out @xcite that in addition to the discs which can be detected in the usual trivializations of @xmath1758 ( which are easily shown to be regular ) there are also `` hidden discs '' that do not entirely lie in one given trivialization and they are not regular for the standard complex structure ( in fact , they are broken configurations consisting of a disc bounding @xmath20 with a holomorphic sphere bubble in @xmath1600 attached to the disc ) .
these hidden discs arise for maslov indices @xmath1759 and above , and they perfectly cancel the terms involving @xmath1760 contributed by the `` visible discs '' .
this is forced to happen by the module structure , since @xmath1756)$ ] must be divisible by @xmath1761 since it must vanish in the @xmath194-eigensummand of @xmath587 .
it would be difficult to justify this computation without the module structure , as it would involve discussing gluing results to turn the irregular hidden discs into genuine pseudo - holomorphic discs for a perturbed almost complex structure .
another short and elegant application of the quantum module structure of the @xmath1 map and of the generation theorem in the monotone setting applies to closed toric varieties which are fano ( that is , monotone ) .
the following result is a special case of a theorem of abouzaid , fukaya , oh , ohta and ono @xcite .
we include a proof since it gives a nice illustration of the usefulness of the module structure of the @xmath1 map .
let @xmath15 be a closed toric fano variety , such that the eigenvalues of @xmath1738 acting on @xmath1681 are all distinct and have multiplicity one .
then @xmath1762 is split - generated by the lagrangian tori with suitable holonomies arising as the critical points of the superpotential for @xmath15 .
the assumption on the eigenvalues of @xmath1738 ensures that @xmath1681 splits into @xmath49-dimensional eigensummands .
a classical fact about closed fano varieties is that quantum cohomology is isomorphic to the jacobian ring of the superpotential .
therefore , to each @xmath49-dimensional eigensummand there corresponds a critical point @xmath51 of the superpotential , and therefore a toric lagrangian @xmath50 with suitable holonomy ( as discussed in section [ section brief survey on landau ginzburg ] ) . but
now observe that @xmath1763 satisfies @xmath1749 ) = \textrm{vol}_b + ( \textrm{higher order } t\textrm { terms } ) \neq 0 \in qh^*(b),\ ] ] where the first term corresponds to the constant disc , and we used the fact that the poincar dual of @xmath1267\in h_0(b)$ ] is the cohomology class of the volume form .
but by the @xmath1681-module structure of @xmath1 , the element @xmath1756)$ ] can only be non - zero in the eigensummand @xmath1764 corresponding to the relevant eigenvalue @xmath1765 of @xmath1738 .
being non - zero in a field implies invertibility , so @xmath1 hits an invertible element in the relevant eigensummand .
so by the generation theorem [ theorem generation for compact categories ] , the lagrangians @xmath50 split - generate @xmath1762 .
abouzaid , fukaya , oh , ohta and ono work in the more general setting of not - necessarily - fano closed toric varieties , with no assumption on the simplicity of the spectrum of quantum product by the first chern class , and prove that the fukaya category and its bulk - deformations are still split - generated by toric fibres .
let @xmath209 be a symplectic manifold conical at infinity , satisfying the monotonicity assumption @xmath1766 \qquad ( \textrm{some } \lambda_e>0)\ ] ] ( this is equivalent to the monotonicity definition in section [ subsection monotonicity assumptions ] when @xmath0 is simply connected ) . 1
. @xmath1756 ) \neq 0 \in qh^*(e)$ ] ; 2 .
if the generalized eigensummand @xmath44 is @xmath49-dimensional , then @xmath22 split - generates @xmath45 and @xmath1770 , in particular @xmath1770 is cohomologically finite . by kontsevich , seidel and auroux ( * ? ? ?
* sec.6 ) , the image under @xmath1771 of @xmath586 is @xmath1772 $ ] ( this relies on @xmath362 being monotone ) . the proof involves showing that the count of holomorphic discs @xmath1773 bounding @xmath22 hitting an lf - cycle representative of @xmath1774 at @xmath1775 and passing through a generic @xmath1267\in c_*(l)$ ] at @xmath1776 is precisely @xmath1466 .
now @xmath586 is a multiple of @xmath1635 $ ] , and @xmath228 is exact at infinity .
therefore @xmath586 can be represented by a closed de rham form which is compactly supported , which in turn is poincar dual to a cycle @xmath1777 .
thus we can represent @xmath586 by a _ compact _
lf - cycle @xmath365 .
since the disc count above does not depend on the choice of representative , we can use @xmath365 .
but then that disc count is precisely the coefficient of @xmath1778 in @xmath1756)$ ] , where @xmath1779 is the intersection dual of @xmath1777 .
the other terms in the expansion of @xmath1756)$ ] are obtained by extending @xmath1778 to a basis of @xmath1780 and counting discs which hit the poincar - dual basis representatives in @xmath1781 at the centre . in particular
, the other terms can not cancel the term contributed by @xmath1778 because of linear independence over @xmath1780 , thus : @xmath1749 ) = m_0(l)\ , \mathrm{pd}(c^{\vee } ) + \textrm{linearly independent terms } \in qh^*(e).\ ] ] this proves the first claim since @xmath1782 by assumption .
_ * example : * to clarify the argument , consider the example @xmath1783 .
since @xmath1753 $ ] as an lf - cycle represents @xmath1784 $ ] , and @xmath1785 , we can take @xmath1786 $ ] . then @xmath1787 $ ] .
finally @xmath1788 ( pull - back is poincar dual to taking preimages , and @xmath1267 $ ] is pd to @xmath1789 in @xmath1600 ) . by the calculations in section [ subsection generation for o of -k ] , the non - zero eigenvalues of @xmath586 are @xmath1790 for solutions of @xmath1730 .
thus the term we are getting in @xmath1756)$ ] is @xmath1791 , which agrees with the contribution of the fibre disc computed in the proof of theorem [ theorem wrapped cat of o of -k ] . _
let @xmath0 be any ( non - compact ) toric fano variety convex at infinity .
let @xmath46 be a critical point of the superpotential @xmath1620 with non - zero critical value @xmath47 .
if the generalized eigensummand @xmath48 is @xmath49-dimensional then the lagrangian torus @xmath50 split - generates @xmath52 and @xmath53 , in particular @xmath53 is cohomologically finite .
the fact that @xmath1267 $ ] is a floer cycle in @xmath1616 follows by theorem [ theorem pt class is cycle for crit lag ] . the lagrangian @xmath50 ( with holonomy ) associated to @xmath51 has eigenvalue @xmath1792 .
so the claim follows by the previous theorem .
finally , we point out another interesting class of examples where the diagram of theorem [ theorem second commutative diagram ] applies .
these examples arise when a quasi - projective variety @xmath1793 is the total space of a morse - bott degeneration to @xmath1794 such that @xmath1795 is closed and monotone .
recall this means that there is a lefschetz fibration @xmath1796 with singular fibre @xmath1794 such that the hessian of @xmath1654 is non - degenerate on the normal bundle of @xmath1797 .
this ensures that there is a lagrangian correspondence in @xmath1798 where @xmath0 is a generic smooth fibre of @xmath1796 , since the vanishing cycle of the degeneration is an @xmath1799-bundle over @xmath15 , which embeds as a coisotropic submanifold in @xmath0 .
these spaces were studied by perutz ( * ? ? ? * def.2.1 ) .
a. abbondandolo , m. schwarz , _
floer homology of cotangent bundles and the loop product _ ,
3 , 15691722 , 2010 . m. abouzaid , _ a geometric criterion for generating the fukaya category _
hautes tudes sci .
112 , 191240 , 2010 .
k. fukaya , p. seidel , i. smith , _ the symplectic geometry of cotangent bundles from a categorical viewpoint _ , in homological mirror symmetry : new developments and perspectives , lecture notes in phys .
757 , springer , 126 , 2009 . | let @xmath0 be a monotone convex symplectic manifold .
we define and study the open - closed string map @xmath1 from hochschild homology of the wrapped fukaya category of @xmath0 to symplectic cohomology .
we prove @xmath1 is a symplectic cohomology module map ( proved independently by ganatra in the exact setup ) , and we extend abouzaid s generation criterion from the exact to the monotone setting .
we furthermore prove @xmath1 commutes with the quilt maps defined by lagrangian correspondences from closed manifolds . as sample applications ,
we show the wrapped category of @xmath2 is cohomologically finite for @xmath3 , and we prove that lifts of split - generators for the compact fukaya category under a finite free quotient are again split - generators . |
since the mit group succeeded in trapping a @xmath4na condensate in an optical potential @xcite , the spin degrees of freedom are liberated , that give rise to a rich variety phenomena @xcite such as spin domains @xcite , textures @xcite , spin mixing dynamics law98,pu99,wenxianzhang , and fragmentation of condensate @xcite .
the properties of such a three - component spinor condensate were first studied using mean - field theories ( mft ) @xcite and polar and ferromagnetic spinor condensates had been implemented experimentally @xcite .
it was initially predicted that the ground state of @xmath4na bec ( @xmath5 ) is either polar ( @xmath6 ) or anti - ferromagnetic ( @xmath7 in the mean - field theory .
however , the results from many body theory @xcite pointed out that the ground state of @xmath4na atoms is a spin singlet with properties drastically different from those of mean field theories ( @xmath8 ) .
it was shown @xcite that the singlet ground state in zero field is a fragmented condensate with anomalously large number fluctuations ( @xmath9 ) and thus has fragile stability .
the exact quantum eigenstates in the spin-2 case @xcite are also found and their magnetic response to a weak magnetic field is compared with their mean field counterpart @xcite .
mixtures of scalar condensates with more than one atomic species or state are actively studied theoretically @xcite .
experimentally , the feshbach resonance has exploited to create a double species condensate with tunable interactions and the dynamics of the superfluid and controllable phase separation are observed myatt , modugno , thalhammer , papp . by adjusting the two s - wave scattering lengths @xmath10 and @xmath11 through the so - called optical feshbach resonances @xcite , the spin exchange interaction between individual atoms can be precisely tuned .
the theoretical studies on mixtures of spinor condensates attracted much attention recently and both mean field and quantum many body theories have been applied to this novel system @xcite .
a temporal modulation of spin exchange interaction , which is tunable with optical feshbach resonance , was recently proposed to localize the spin mixing dynamics in a @xmath12rb condensate @xcite .
the interspecies scattering parameters between @xmath12rb and @xmath4na are calculated resorting to the simple approach of the degenerate internal - state approximation ( dia ) @xcite as the low - energy atomic interactions can be mostly attributed to the ground - state configurations of the two valence electrons .
the interspecies scattering lengths for singlet and triplet electronic states are approximately determined already dia1,dia2 , given by @xmath13 and @xmath14 , where @xmath10 is the bohr radius .
the interspecies interactions between @xmath12rb and @xmath4na atoms are then parametrized by three scattering lengths ( see eq .
( scatter ) bellow ) , each being a linear combination of @xmath15 and @xmath16 weighted by the appropriate @xmath17 coefficients for the total combined spins of @xmath18 , and @xmath19 . within this approximation
, it is found coincidently that , the parameter for the interspecies singlet - pairing interaction @xmath20 , is equal to zero @xcite .
however , for the more general case of an arbitrary mixture of spin-1 condensate , the hyperfine interaction between nuclear spin and electron spin gives in general non - neglegible interspecies pairing and dia approximation is not applicable @xcite . in this paper , we first study the various quantum phases of the binary mixture of spin-1 condensates in the ground state ignoring the interspecies singlet pairing .
then the situation with competition between intra- and inter - species singlet pairings is considered . using the full quantum approach of exact diagonalization
, we present the detailed phases diagram for the @xmath3 case .
we take the intra - condensate atomic interaction in the form @xmath21 with @xmath22 for the two species and the inter - species interaction is described as @xmath23 in contrast to intra - condensate interactions between identical atoms luo , xu09 , the collision between atoms belonging to different species can occur in the total spin @xmath24 channel , which makes the mixture more interesting .
here@xmath25 with @xmath26 is the scattering lengths in the channels of total spin @xmath27 , and @xmath19 , respectively , and @xmath28 denotes the reduced mass for the pair of atoms , one each from the two different species with masses @xmath29 and @xmath30 respectively .
@xmath31 is the corresponding projection operator with the relationship @xmath32 and @xmath33 , from which we get @xmath34with the parameters @xmath35 @xmath36 @xmath37 and @xmath20 @xmath38 @xmath37 .
@xmath39 projects an inter - species pair into spin singlet state @xcite .
furthermore , these parameters can be related to the singlet and triplet scattering lengths by means of a method based on @xmath17 coefficient @xcite @xmath40 denote the atomic field operators for the spin state @xmath41 as @xmath42 for species 1 and @xmath43 for species 2 , the hamiltonian for the mixture system in the second quantization is represented by @xmath44@xmath45 is the same as @xmath46 with the substitution of subscript @xmath47 by @xmath19 and @xmath42 by @xmath43 and @xmath48 . through the control of the trapping frequency
, we can make the two species sufficiently overlapped and adopt the single spatial - mode approximation ( sma ) @xcite for each of the two spinor condensates with modes @xmath49 and @xmath50 , i.e. @xmath51with @xmath52 ( @xmath53 ) the annihilation operator for the ferromagnetic ( polar ) atoms satisfying @xmath54 = 0 $ ] and @xmath55 = \delta _ { ij}$ ] ( and the same form of commutations for @xmath53 ) .
the density - density interaction part is a constant .
hence we only focus on the spin - dependent hamiltonian @xmath56with @xmath57 ( @xmath58 ) defined in terms of the @xmath59 spin-1 matrices @xmath60 @xmath61 .
the operator @xmath62creates a singlet pair with one atom each from the two species , in much the same way that the operators @xmath63create pairs with two atoms from the same species @xcite .
the interaction coefficients are @xmath64 @xmath65 and @xmath66 . in the following
we perform all analysis in the single - mode regime in the absence of external fields .
as learned from previous studies @xcite , sma is shown to be exact for atomic interaction of the ferromagnetic type . for polar interaction , if the magnetization @xmath67 , the sma wave function is still exact and becomes invalid only if @xmath68 is large . in our mixture , we can safely apply the sma to wave function of both polar and ferromagnetic atoms in the absence of an external magnetic field .
the ground state for polar atoms is a fragile fragmented state with @xmath67 .
small external fields or spatial dependence drive the system to symmetry broken states which are better captured by mean field theory .
very recently the symmetry broken phase has been found in this binary mixture in the presence of a weak magnetic field @xcite . in our previous study
@xcite we report the anomalous fluctuations for the numbers of atoms in the mixture of @xmath4na ( polar ) and @xmath12rb ( ferromagnetic ) condensates in their @xmath24 manifold .
dia has been adopted to ignore the @xmath69 term .
the ferromagnetic @xmath70rb condensate provides a smooth background where the quantum many body states are hardly affected by the fluctuation , while the fragile polar atoms @xmath4na are easier influenced . especially in the ground state of the aa phase , the interspecies anti - ferromagnetic spin - exchange is large enough to polarize both species and a maximally entangled state is realized between two species with total spin @xmath71 .
in this section , we consider the simple situation of @xmath2 with hamiltonian @xmath73or in an alternative form @xmath74where @xmath75 and @xmath76 is the total angular momentum operator .
a constant energy shift of @xmath77 has been trivially eliminated .
the results may serve as reference states for the complete ground - state phases .
the eigenstates of the hamiltonian ( [ ham3 ] ) are then simply the common eigenstates of the commutative operators @xmath78 and @xmath79 , denoted by @xmath80where the states in uncoupled representation @xmath81span a hilbert space of dimension @xmath82 , respectively @xcite . here @xmath83 s are the clebsch - gordon coefficients , @xmath84 are the normalization constants and @xmath85 is the lowering operator for @xmath86(@xmath87 ) . minimizing the corresponding eigenenergies of the hamiltonian @xmath88we can get the ground state energy determined by different parameters @xmath89 , @xmath90 and @xmath91 . in fig .
1 , using the full quantum approach of exact diagonalization we calculate ground state order parameters @xmath92 @xmath93 and @xmath94 to illustrate the different phases as @xmath91 changes .
the results are presented for three typical cases ( a ) @xmath95 , ( b ) @xmath96 and ( c ) @xmath97 with equal atomic numbers in two species , i.e. @xmath98 .
we see that the results agree fairly well with the simulated annealing approach in the mean field theory @xcite except for some small deviations originated from pure quantum effect . for instance , the maximum and minimum values are @xmath99 and @xmath100 for @xmath101 and @xmath102 , but they are @xmath103 and @xmath104 for @xmath105 . at fixed values of ( a ) @xmath106 @xmath107 ; ( b ) @xmath108 @xmath109 ; ( c ) @xmath106 @xmath110 ; and @xmath111 ( all in units of @xmath112 ) .
black solid lines , red dashed lines and blue dot - dashed lines denote respectively the order parameters @xmath113 @xmath114 and @xmath115 .
, width=312 ] for the mixture of two ferromagnetic condensates , there are generally two phases ff and aa separated by the critical point @xmath100 as shown in fig
the ff phase is described by a set of degenerate states generated by repeatedly applying the lowering operators @xmath118 on the extreme states for @xmath119 times @xmath120with @xmath121 .
the extreme states @xmath122can be simply described as @xmath123 the aa phase is a singlet @xmath124 with all states obey the condition @xmath125 .
all channels of total spin zero have to be taken into account and we have@xmath126 the mixture of two polar condensates allows for five distinct phases ff , mm@xmath129 pp , mm@xmath130 and aa separated by four critical points @xmath131 and @xmath132 corresponding to @xmath133 , @xmath134 , @xmath135 and @xmath136 in fig . [ fig1]b . in the region @xmath137
, the mm@xmath138 phase takes the same form as eq .
( ff ) , with extreme states represented by @xmath139or @xmath140 . in mm@xmath138 phase
the atoms in species 1 are totally polarized in one direction and form a steady magnetic field ( black solid line ) , and those in species 2 are partially polarized in the same direction . increasing the strength of coupling interaction ( @xmath141 ) breaks singlet pairs in species 2 one by one , and results in the increase of the total spin . in the region
@xmath91 @xmath142 , the two species are essentially independent for a weak interspecies spin - exchange interaction .
the pp phase is a total spin singlet described by the direct product of the well known polar ground state @xmath143 @xcite , giving rise to @xmath144 @xmath145 and @xmath146 . in the region of @xmath147
, however , the mm@xmath148 phase favors that the atoms in species 2 are polarizing to the opposite direction of those in species 1 and the total spin gradually decreases .
this situation is much more complicated because all states that satisfy the condition @xmath149 are involved and the summation index runs over all possible clebsch - gordon coefficients , giving rise to the following extreme state @xmath150 in the case of a mixture of a ferromagnetic and a polar condensate , four possible phases ff , mm@xmath151 mm@xmath130 and aa are separated by three critical points @xmath152 and @xmath153 corresponding to @xmath154 in fig . 1 ( c ) .
it has been shown that the ground state of polar atoms ( @xmath155 ) only is a fragmented condensate with anomalously large number fluctuations and fragile stability @xcite , which can be described as @xmath156 . the @xmath157 and @xmath158 are invariant under any rotation of the system , and commute with @xmath159 and @xmath160 . however , for a ferromagnetic condensate ( @xmath161 ) the ground state favors that all atoms are aligned in the same direction ( i.e. , @xmath162 ) and much more stable .
so when these two kinds of atoms are mixed together , the polar atoms are more easier to be influenced , but their back action on to the stable ferromagnetic atoms is negligible .
this can be seen from the constant black solid line in fig.[fig1 ] ( c ) .
an interesting observation is that for large and negative ( positive ) value of parameter @xmath163 the system enters the same phase ff ( aa ) , no matter how the atoms interact inside each species .
detailed calculation show that in ff phase all atoms are polarized in the same direction and the total spin reaches its maximum value .
the ground state is highly degenerate with degeneracy @xmath164 .
take the state @xmath165 as an example , the atomic populations are @xmath166 which reduces to @xmath167 for each species for large @xmath168 .
the number fluctuations @xmath169 are in order of @xmath170 for large @xmath168 . on the other hand , in aa phase ,
atoms in species 1 and 2 are fully polarized , however , in opposite directions and the total spin vanishes zj .
the singlet ground state is a fragmented condensate @xcite and the single particle density matrix is diagonal with atoms equally populated @xmath171 or @xmath172 for each species .
the fluctuations @xmath173 are anomalously large ( in order of @xmath168 ) thus the fragmented condensate is fragile .
the last term @xmath175 in hamiltonian ( [ ham ] ) is related to the inter - species singlet paring .
although [ @xmath176=0 $ ] , we notice that [ @xmath177\neq 0 $ ] , and [ @xmath178\neq 0 $ ] .
thus in general they do not belong a set of commutative operators and the system is not solvable . in the special case of @xmath179=@xmath90=@xmath180 , we found that the spin - dependent hamiltonian reduces to a sum of commutative operators .
the eigenstates can be constructed by several building blocks @xcite which can be found via generating function method . to see more clearly the role played by the @xmath20 term
, we focus on a special case in which @xmath181 is much larger than the other parts of the hamiltonian and @xmath161 , @xmath155 . we know that the @xmath20 term encourages pairing two different types of atoms into singlets when @xmath182 . in this case
the interspecies singlet - pairing interaction dominates the system and the total spin vanishes , namely @xmath183 there are two typical cases obeying the above condition . in one case
we have @xmath184 @xmath185 and @xmath186 , which means the singlet formation occurs inside each species , and the atoms in the same species are all paired with no net spin left . in the other case
we have @xmath187 , which means some intra - species pairs are broken , meanwhile singlet formation occurs between different species . under the condition @xmath188 ,
numerical results show that the above two cases indeed exist and well determined by the parameter @xmath189 . at fixed values of @xmath106 @xmath190 and @xmath191 ( in the unit of @xmath192 ) .
black solid lines , red dashed lines , blue dot - dashed lines and green dotted lines denote respectively the order parameters @xmath193 @xmath194 , @xmath195 , and @xmath196,width=312 ] we consider the direct product of the fock states of the two species @xmath197 , which may be equivalently defined as @xmath198here @xmath199 are the corresponding magnetization specified as @xmath200 and @xmath201 is the total magnetization . for simplification , we restrict ourselves into the subspace that the total magnetization is conserved @xmath202 , in which case all states are non - degenerate .
the hamiltonian ( [ ham ] ) is then represented in a sparse matrix and with the approach of exact diagonalization we numerically get the ground state of the system , on which the order parameters are calculated . for different inter - species paring interaction parameter @xmath203 .
, width=336 ] fig .
2 shows the dependence of the four order parameters on @xmath163 at fixed values of @xmath204 @xmath205 and @xmath206 .
we find that in the region of @xmath207 , @xmath101 , @xmath208 , and @xmath209 are all approximatively equal to zero , corresponding to the first case of vanishing total spin .
however , for positive @xmath189 , atoms in each species begin to polarize and are fully polarized for @xmath210 , with the negative value of @xmath211 showing that they are polarized to the opposite directions . during the whole process
the three order parameters always obey the condition @xmath212 .
the distinction in the envelope of @xmath213 and @xmath114 in the region @xmath214 lies in the choice of @xmath204 @xmath215 .
the fourth order parameter @xmath216 reflects the feature of the total spin @xmath217 of the system .
it takes a constant value , which is actually the maximum value , in entire region of @xmath91 in figure 2 ( see the green line ) , while the total spin reaches its minimum value @xmath100 . in order to see more clearly the competition between the interspecies coupling and singlet pairing interactions , in figure 3
we illustrate four order parameters for different values of @xmath175 .
an obvious variation can be easily seen from , for example , figure 3d , where the black solid line ( @xmath218 ) gradually changes to the purple dashed line ( @xmath219 ) .
we find that when @xmath20 = 0 the order parameters @xmath220 becomes zero in the ff phase implying that the total spin @xmath217 gets to its maximum ( fig .
3a ) , while in the aa phase it equals to a constant value and the total spin @xmath217 vanishes . in between the two limits 0 and `` constant ''
, the intermediate value of the order parameters @xmath221 indicates that the system is a mixture with both singlet pairs and nonzero net magnetization .
this feature can be well understood for the special case when @xmath2 and @xmath222 , marked as a small dip in the black solid line in fig .
we find that @xmath223 , which can be obtained analytically .
in fact , the state at this point can be simply expressed as a direct product of two well - known states ( ferromagnetic and polar ) , @xmath224 , and we find that on this state @xmath225which agrees with the numerical result .
we notice that the system is a mixture with @xmath226 polar pairs and net magnetization @xmath227
. in our system there exist three states , on which the average of total spin amounts to zero .
although we have @xmath228 @xmath229 and @xmath230 , the difference between these states can be easily seen from an example @xmath231 .
we find that @xmath232from the relation between these three states ( [ 3states ] ) , we see that the aa phase of our system includes at lest two pairing mechanism , i.e. @xmath233 and @xmath234 . as total spin @xmath217 vanishes , the number distributions of these three states are all @xmath235 , but the number fluctuation on these states are quite different . for the state @xmath236 hoyip
, the results are @xmath237on the state @xmath238 @xcite , we have obtained in eq .
( [ flu ] ) , in the case of large @xmath168 , satisfying @xmath239while for the state @xmath240 we find that the number fluctuations are equally distributed @xcite , i.e. @xmath241 for different inter - species paring interaction parameter @xmath242 .
( b ) the dependence of ground - state order parameters on @xmath243 at fixed values of @xmath106 @xmath190 and @xmath244 , width=288 ] finally we discuss briefly the situation when @xmath245 , in which case the ground state favors minimizing the @xmath20 term .
we illustrate the competition between the interspecies coupling and singlet pairing interactions numerically in fig .
4a . if @xmath175 is far more larger than any other parameters , the order parameter @xmath246 will vanish ( see the dashed line in figure 4a ) .
all order parameters take the value @xmath100 or @xmath247 when @xmath248 , with the boundary determined by the corresponding amplitudes of spin - coupling strengths as shown in fig .
4b for @xmath249 .
in summary we studied the interspecies singlet pairing in the ground state of a binary mixture of spin-1 condensates in the absence of a magnetic field . in the case of @xmath250
, the exact quantum states can be constructed from angular momentum theory for the mixture of two ferromagnetic , two polar , and ferromagnetic - polar condensates .
the ground state is classified into five types according to the inter - species coupling parameter @xmath91 . by means of the full quantum approach of exact diagonalization , more general case of @xmath3
is considered .
we illustrate the competition between the two interspecies interaction @xmath251 and @xmath252 , and find that if @xmath253 the ground state is a singlet of the total spin .
there , however , exist different types of singlet formations determined by @xmath189 .
this work is supported by the nsf of china under grant nos .
10774095 and 11074153 , the national basic research program of china ( 973 program ) under grant nos . 2010cb923103 , 2011cb921601 , the nsf of shanxi province , shanxi scholarship council of china , and the program for new century excellent talents in university ( ncet ) .
p. o. fedichev , yu .
kagan , g. v. shlyapnikov , and j. t. m. walraven , phys .
lett . * 77 * , 2913 ( 1996 ) ; f. k. fatemi , k. m. jones , and p. d. lett , _ ibid . _ * 85 * , 4462 ( 2000 ) ; m. theis , g. thalhammer , k. winkler , m. hellwig , g. ruff , r. grimm , and j. h. denschlag , _ ibid .
_ * 93 , * 123001 ( 2004 ) ; c. d. hamley , e. m. bookjans , g. behin - aein , p. ahmadi , and m. s. chapman , phys . rev .
a * 79 * , 023401 ( 2009 ) . | we study the ground state properties of a mixture formed by two spin-1 condensates in the absence of an external magnetic field . as the collisional symmetry between interspecies bosonic atoms
is broken , the interspecies coupling interaction ( @xmath0 ) and interspecies singlet pairing interaction ( @xmath1 ) arise . the ground state can be calculated using the angular momentum theory analytically for @xmath2 .
the full quantum approach of exact diagonalization is adopted numerically to consider the more general case as @xmath3 .
we illustrate the competition between the two interspecies interactions and find that as singlet pairing interaction dominates ( or the total spin vanishes ) , there are still different types of singlet formations which are well determined by @xmath0 . |
carbon nitrides with n : c ratios @xmath71 form a class of solid state compounds with properties that are being investigated for applications ranging from gas sorption and catalysis to energy conversion and storage @xcite .
experimentally known structures are based on @xmath3-bonded carbon atoms and include polymeric to graphitic layered compounds . within the c - n - h system ,
materials related to liebig s melon ( ( c@xmath8n@xmath9h@xmath0)@xmath10 ) contain tri-@xmath11-triazine ( heptazine : c@xmath8n@xmath12 ) structural units linked @xmath13 -nh- groups @xcite .
additional graphitic materials containing intercalated li@xmath14 , cl@xmath15 and br@xmath15 ions are formed by polytriazine imide ( pti ) layers consisting of triazine ( c@xmath0n@xmath0 ) rings linked by -n= or -nh- species @xcite .
synthesis of macroscopic flakes of triazine - based graphitic carbon nitride ( tgcn ) composed of nitrogen - linked triazine ( c@xmath0n@xmath0 ) units was reported recently @xcite .
this structure was first proposed in the mid-1990 s as `` graphitic carbon nitride '' ( `` g - c@xmath0n@xmath1 '' ) by analogy with the structurally related graphite @xcite .
a nanocrystalline material with c@xmath0n@xmath1 stoichiometry had been reported from chemical vapor deposition @xcite and related models were studied theoretically @xcite .
triazine - based carbon nitrides can also be exfoliated to produce layered materials analogous to few - layered graphene @xcite .
theory shows that the optical bandgap ( 1.62.7 ev ) important in photocatalysis can be tuned by controlling the layer buckling at high pressure @xcite .
interest in dense carbon nitride polymorphs based on @xmath4 bonded c atoms began with predictions from density functional theory ( dft ) calculations of superhard properties @xcite .
first investigations began by considering known structures such as those of si@xmath0n@xmath1 ceramic phases , and more recently using structure searching methods @xcite . however
none of the predicted high density c@xmath0n@xmath1 polymorphs has been demonstrated experimentally to date @xcite .
one carbon nitride solid containing @xmath4 bonded c atoms was produced by laser heating dicyandiamide ( dcda : c@xmath6n@xmath1h@xmath1 ) at @xmath16 gpa in a diamond anvil cell @xcite .
this material with a defective wurtzite structure could be recovered to ambient conditions @xcite .
a recent study has identified formation of a cn phase that also contains @xmath4 bonded carbon @xcite .
experimental compression studies and dft calculations have also revealed the existence of carbon nitride structures containing both @xmath3 and @xmath4 bonded c atoms obtained from the graphitic layered pti compound c@xmath8n@xmath9h@xmath0@xmath17hcl .
layer buckling causes c - n distances in adjacent layers to approach a covalently bonded value and an interlayer bonded ( ilb ) form becomes stabilized above 47 gpa @xcite .
this result shows that carbon nitride framework structures containing mixed @xmath3 and @xmath4 bonding can exist over a range of densities .
microporous carbon nitride solids have also been produced at ambient pressure by linking c@xmath0n@xmath0 triazine units into three - dimensional covalent networks @xcite .
the structural polymorphism of @xmath4-bonded carbon nitride is usefully compared with that of elemental carbon . here
high - pressure synthesis is well known to produce dense structures @xcite .
low pressure allotropes dominated by @xmath3 bonding consist of planar graphite and graphene structures formed from fused six - membered rings . introducing curvature into the layers leads to fullerenes and c - nanotubes , while hierarchical low density mesoporous solids assembled at different length scales are important for nanotechnology and planetary science studies @xcite .
theoretical modelling now also points to the potential existence of metastable zeolitic @xmath4-bonded carbon framework structures in the 020 gpa range @xcite . in equilibrium at low temperatures and pressures above 0.76 gpa @xcite
the dense @xmath4-bonded structure of diamond becomes the most stable form and no further structural transformations are expected until a body - centered ( bc8 ) polymorph is predicted above 1000 gpa ( 1 tpa ) @xcite .
other new dense allotropes relevant to understanding c - rich exoplanets are predicted to exist at extreme pressures extending into the tpa range @xcite .
here we used _
ab initio _ random structure searching ( airss ) @xcite to explore carbon nitride polymorphism as a function of density throughout a wide pressure range extending between 01000 gpa ( 01 tpa ) , focusing on behavior at the c@xmath0n@xmath1 composition . at low density
the results reveal new crystalline architectures including previously unsuspected framework structures based on mixed @xmath3-@xmath4 bonding
. one of these structures ( @xmath2 or its enantiomorph @xmath18 ) has been verified experimentally in our study .
the airss results allow a ranking of stability among predicted c@xmath0n@xmath1 polymorphs up to 650 gpa , above which c@xmath0n@xmath1 becomes unstable relative to diamond + pyrite - structured cn@xmath6 .
airss is a powerful theoretical prediction technique that has led to the discovery of new structures subsequently verified by experiment , in materials ranging from hydrogen @xcite , ammonia @xcite , ammonia hydrates @xcite , aluminium hydride @xcite , silane @xcite to xenon oxides @xcite . within the airss method
a cell volume and shape are selected at random , atoms are added at random positions to provide the desired composition , and the system is relaxed until the forces on the atoms are negligible and the pressure takes the required value .
this procedure is repeated many times , giving a relatively sparse sampling of the `` structure space '' .
we typically constrain the minimum separations between atom pairs ( c c , c
n ) in the initial structures , which helps to space out the atoms appropriately , while retaining a high degree of randomness .
these minimum distances are obtained from preliminary short airss runs . in almost all of the subsequent searches we impose symmetry constraints on the structures , for example
, we might demand that all structures belong to randomly selected space groups with 8 or more symmetries . in the next stage of searching
we might choose to allow structures with 4 or more symmetries , etc . with this approach
we obtain a very good sampling of the high - symmetry structures , which can be extended towards lower symmetries as the available computing resources allow .
the use of symmetry constraints greatly reduces the size of the `` structure space '' , although they also break it up into disconnected regions which can prevent some structures from relaxing fully .
however , there is a strong tendency for low energy structures to possess symmetry @xcite , and we find that the application of symmetry constraints throughout the searching is a very useful tool .
searches for low pressure structures are performed using heptazine- and triazine - based units and we consider structures containing up to sixteen formula units ( f.u . ) .
the structure searches were performed using airss and first - principles dft methods , version 8.0 of the castep plane - wave basis set pseudopotential code @xcite , default castep ultrasoft pseudopotentials@xcite , a plane - wave basis set energy cutoff of 280 ev , and an initial brillouin zone sampling grid of spacing [email protected] @xmath20 .
the pbe generalized gradient approximation density functional @xcite was used for the structure searches .
the final numerical results reported in this paper were obtained using the pbesol functional with a basis set energy cutoff of 700 ev , castep `` on - the - fly '' ultrasoft pseudopotentials , as described in the supplemental material@xcite , and a @xmath21-point spacing of [email protected]@xmath20 , except for the data in table i , which were calculated with a finer @xmath21-point spacing of [email protected] @xmath20 .
estimating the pressure of polymorphic transitions among @xmath3 and @xmath4 bonded phases for carbon nitride materials is difficult because no reliable data for the structures and energetics of @xmath4 bonded c@xmath0n@xmath1 polymorphs are available . in order to choose a density functional that would provide reasonable transition pressures for comparison with experiment we considered the well - known transition from @xmath3-bonded graphite to @xmath4-bonded diamond in elemental carbon , which has distinct similarities to c@xmath0n@xmath1 .
the structures and lattice constants of diamond and graphite are known , which allows accurate comparisons to be made with the results of dft calculations @xcite . in fig .
[ fig : enthalpy_dia_graph ] we plot the differences in enthalpy between the graphite and reference diamond structures calculated using five density functionals : the local density approximation ( lda ) @xcite , the semi - local pbe @xcite , wc @xcite and pbesol generalized gradient approximations @xcite , and the pbe+g06 generalized gradient approximation with empirical van der waals corrections @xcite .
the calculated graphite / diamond transition pressures ranged from -1 gpa ( lda ) to + 6 gpa ( pbe ) , which provides an estimate of the range of errors in the pressure arising from the approximate density functionals of about @xmath224 gpa , see fig . [ fig : enthalpy_dia_graph ] .
the most important factor for obtaining an accurate graphite / diamond transition pressure is to use a functional that provides a good account of the volume of diamond and the in - plane lattice constant of graphite . here
the effects of zero - point vibrational motion are estimated to be small and have been neglected .
the lda normally tends to underestimate volumes , while pbe tends to overestimate them .
pbesol , wc , and pbe+g06 give volumes for diamond intermediate between those of lda and pbe and are in reasonable agreement with experiment .
the description of graphite is more variable .
the graphite / diamond transition pressure is calculated to be 0.7 gpa using pbesol , in close agreement with experiment .
wc and pbe - g06 both give 2 gpa , while pbe predicts the transition at 6 gpa .
we therefore elected to use pbesol for our studies of c@xmath0n@xmath1 polymorphs .
the pbe and pbesol functionals gave the same sequence of polymorphic phase transitions although the pbesol transition pressures are somewhat larger .
enthalpy - pressure relations for c@xmath0n@xmath1 polymorphs found by airss using the pbesol functional are shown in fig.[fig : enthalpy_c3n4_pbe_0 - 400_gpa ] . below approximately 5 gpa
we find that carbon nitride compounds are metastable with respect to elemental carbon ( diamond or graphitic polymorphs ) and n@xmath6 phases . at low pressure a wide variety of corrugated layers or open framework structures based on linked heptazine or triazine motifs
are predicted to occur and the number of potential polymorphs and structure types increases dramatically with system size @xcite .
these low - density materials typically have very large unit cell volumes and they rapidly become destabilized with increasing pressure @xcite .
the importance of such mesoporous phases and their relationship to layered and framework - structured carbon nitrides is discussed below .
heptazine - based structures of @xmath23 , @xmath24 and @xmath25 symmetries are the most stable below 2 gpa in pbesol , and the @xmath2 structure with mixed @xmath3-@xmath4 bonding is stable in the range 211 gpa . at higher pressures we find structures with @xmath4 bonding around the carbon atoms to be the most stable .
we found the following sequence of transitions among the most stable phases as a function of pressure : @xmath26 the numbers above the arrows indicate the pbesol transition pressures in gpa .
( the @xmath27 structure found in dft searches by dong _
et al . _
@xcite is not predicted to be stable at any pressure because our study has discovered new structures that are more stable . )
structures of the high - density phases predicted to become stable at high pressure are illustrated in fig.[fig : structures_high_pressure ] .
the low - enthalpy structures contain only c
n bonds , no c c or n n bonding was found to occur in energetically competitive structures over the pressure range studied of 01 tpa . between 1147 gpa ( using pbesol ) the @xmath28-c@xmath0n@xmath1 structure with @xmath29 symmetry analogous to the ceramic phase @xmath28-si@xmath0n@xmath1
is predicted to be stable .
this is one of the polymorphs considered by cohen , sung and others to constitute a likely ultra - low compressibility , superhard material @xcite .
the @xmath28-c@xmath0n@xmath1 structure contains @xmath4 bonded 4-coordinated c atoms and @xmath3 bonded three - fold coordinated n atoms characteristic of other low - enthalpy structures at high pressures .
in addition to its low compressibility and potentially high hardness it has been predicted to have the interesting materials property of a negative poisson s ratio @xcite .
the @xmath30-c@xmath0n@xmath1 phase considered in early discussions is not stable at any pressure within the range studied . above 47 gpa
we find that @xmath28-c@xmath0n@xmath1 is overtaken in stability by a @xmath31 structured phase that remains the most stable polymorph until 100 gpa ( fig.[fig : structures_high_pressure ] ) . at this pressure
a structure with @xmath32 symmetry makes a brief appearance , before the introduction of a cubic ( @xmath33 ) structure @xcite containing nearly tetrahedrally bonded n atoms along with the @xmath4-bonded c atom centers .
neither the @xmath31 nor @xmath32 polymorphs were identified as stable phases in structure searches reported by dong _
et al . _
@xcite . at 195 gpa
another cubic ( @xmath34 ) phase makes a brief appearance , followed by a @xmath35 structured polymorph that is stable until 238 gpa .
this phase was described by dong _
_ as occurring between 224300 gpa .
it appears to resolve the competing structural situation between c- and n - based bonding with the emergence of clearly separated c- and n - atomic layers within a structure that could be described as `` defective '' diamond or c - bn . above 238 gpa
the most stable predicted polymorph has orthorhombic @xmath5 symmetry .
this remains the most stable phase until 650 gpa where it decomposes into diamond and a pyrite - structured cn@xmath6 polymorph according to @xmath36
to begin to test our theoretical predictions we carried out an exploratory synthesis experiment in a laser - heated diamond anvil cell ( dac ) using a sample of crystalline tgcn g - c@xmath0n@xmath1 as a starting material @xcite .
this compound has been shown to have @xmath37 or @xmath38 symmetry corresponding to abc or ab stacking of the buckled graphitic layers .
a thin sample flake of approximately 60 @xmath39 60 micrometers in lateral dimensions was loaded into the dac chamber and isolated from the diamond window using three small ruby spheres .
the pressure was determined by ruby fluorescence from a different ruby sphere placed away from the sample .
the sample was compressed using ne as the pressure - transmitting medium ( ptm ) . _ in situ _ x - ray diffraction experiments were carried out at beamline id27 of the european synchrotron radiation facility using monochromatic radiation ( @xmath40 = 0.3738 ) . the starting material exhibits an intense peak at 2@xmath41 = 6.6@xmath42 due to the 002 reflection of the g - c@xmath0n@xmath1 tgcn phase ( fig .
[ fig : experimental_data](a ) ) . during initial compression to 53 gpa
this peak first shifts to larger 2@xmath41 values then broadens and finally disappears above approximately 40 gpa ( fig .
[ fig : experimental_data](a ) ) .
this most likely occurs due to structural disordering associated with layer buckling and possible interlayer c - n bonding at high pressure as previously observed for structurally related c@xmath8n@xmath9h@xmath0@xmath17hcl @xcite . by 53 gpa the only peaks remaining in the pattern correspond to the face centered cubic lattice of solid ne used as the ptm ( fig .
[ fig : experimental_data](a ) ) . following initial pressurization the material was heated at 1545@xmath2280 k using a co@xmath6 laser beam focused on the sample while monitoring the x - ray diffraction pattern @xcite . after approximately 5 minutes new single crystal spots appeared in the 2d diffraction pattern that had been previously dominated by broad diffraction rings from the compressed tgcn phase and ne ptm material ( fig.[fig : experimental_data](a ) and ( b ) ) .
interestingly , the appearance of the crystalline peaks was accompanied by a pressure drop inside the dac chamber to 44 gpa indicated by ruby fluorescence , as well as by an 18 % lattice expansion of the solid ne used as ptm ( fig .
[ fig : experimental_data](a ) ) .
this possibly indicates a corresponding densification within the c@xmath0n@xmath1 sample .
analysis of our x - ray data showed no evidence for diamond or any elemental n@xmath6 phases formed during the experiment .
there was likewise no evidence for any of the other predicted carbon nitride phases with different n / c ratios such as cn or cn@xmath6 . we conclude that the sample retained its initial composition close to c@xmath0n@xmath1 throughout the dac laser heating experiment .
the crystalline material produced in the experiment corresponded to only a few crystallites distributed throughout the sample . while the sample remained at high pressure in the dac we carried out a mapping study using a focused ( @xmath432 micrometers ) x - ray beam that revealed crystals present in only 3 out of 36 patterns obtained .
a typical 2d diffraction image is shown in fig .
[ fig : experimental_data](b ) showing the crystalline spots against the sharper rings from the ne ptm and diffuse scattering from the densified disordered c@xmath0n@xmath1 material . during recovery to ambient conditions the diffraction pattern emerged more clearly .
once removed from the dac the sample which remained mounted inside its re gasket was examined by x - ray diffraction .
although there were more crystalline features than at high pressure the powder averaging remained poor and it was not possible to carry out _ ab initio _ or rietveld structure refinements ( ( fig . [ fig : experimental_data](c ) ) .
however we tested the range of diffraction peaks against calculated patterns for the various polymorphs produced in our airss study as well as other theoretically predicted structures .
only one phase contained peaks within the range of @xmath44 spacings observed experimentally .
this corresponded to a new c@xmath0n@xmath1 polymorph with @xmath2 symmetry as predicted by the airss calculations .
this phase is predicted to have a stability range between 2 to 11 gpa using pbesol . at 50 gpa
it is highly metastable relative to the predicted @xmath28-c@xmath0n@xmath1 ( @xmath29 ) or @xmath31 structures , but it lies lower in energy than the metastably compressed graphitic forms ( fig.[fig : enthalpy_c3n4_pbe_0 - 400_gpa ] ) .
we found an excellent match between all the observed @xmath44 spacings for the new crystalline phase and predicted reflections for the @xmath2 polymorph .
we then used le bail analysis to obtain refined unit cell parameters @xmath45 = @xmath46 = 4.021 , @xmath47 = 13.922 , @xmath48 = 3.46 , and @xmath49 = 56.26 @xmath50 per f.u .
, with structural residual or reliability factors r@xmath51 = 6.49 and r@xmath52 = 4.23 ( fig .
[ fig : experimental_data](c ) ) .
these cell parameters are in reasonable agreement with the predicted values for the new structure type at 0 gpa ( table [ table : exp_theory ] ) .
the observed pattern of x - ray intensities differs substantially from those expected for a fully averaged powder diffraction pattern because only a few crystallites were present within the laser heated sample .
in addition the crystallites might exhibit preferred orientation further complicating any more detailed analysis of the experimental data .
we were able to generate patterns that approximately modelled the observed intensity distribution by constructing composites containing a few oriented crystallites with fixed atomic positions from the @xmath2 theory predictions , applying a march - dollase function to correct for preferred orientation . using this approach we obtained solutions to a constrained rietveld refinement of the structure with large but not unreasonable residual values , e.g. , with r@xmath51 and r@xmath52 on the order of @xmath4320 ( fig .
[ fig : experimental_data](d ) ) . to test the possibility that other atomic arrangements could give rise to the same unit cell we then used airss to examine alternative structure solutions within the experimentally determined cell shape and volume under ambient conditions . using the pbesol functional for these runs gave a very small residual pressure of about 0.4 gpa in the experimental cell while pbe indicated a pressure of 3 gpa .
our searches consistently returned @xmath2 as the most stable phase , with the next possibility lying at much higher energy ( 0.75 ev / f.u . ) .
we conclude from this analysis that the newly predicted @xmath2 polymorph was produced in our high pressure - high temperature experiment .
this is the first recorded experimental synthesis of a crystalline c@xmath0n@xmath1 framework structure predicted by _ ab initio _ theory .
along with the defective - wurtzite structure c@xmath6n@xmath0h @xcite and the recent report of a new cn phase @xcite it constitutes a new example of a crystalline carbon nitride containing @xmath4-bonded carbon atoms .
various schematic drawings of the new @xmath2 polymorph are shown in fig.[fig : structure_p4_32_12_ambientstack ] to highlight different important features of the structure .
it corresponds to a new framework type containing fused triazine rings linked via @xmath4-bonded c atoms .
this is important because mixed @xmath53 bonding around the c atoms does not occur for pure carbon polymorphs in equilibrium at any pressure .
in addition the geometrical arrangement containing @xmath4-bonded carbon linked to four nitrogen atoms is highly unusual and perhaps absent in all organic chemistry except perhaps among transition states .
our experimental findings indicate that the probability of forming the new carbon nitride phase is low .
this can be ascribed to high kinetic barriers as well as the difficulty of forming the unusual c - n linked tetrahedral geometry associated with the formation reaction . [ cols="^,^,^,^,^",options="header " , ]
in the low pressure range where c@xmath54n@xmath55 compounds are predicted to be metastable with respect to elemental carbon and nitrogen @xcite , it has typically been assumed by analogy with elemental carbon that layered `` graphitic '' motifs should constitute the ground state for carbon nitride materials at low pressures , taking into account the existence of `` voids '' within the layers due to the different valencies of c and n atoms @xcite .
one basic unit of connectivity is the planar triazine ( c@xmath0n@xmath0 ) ring that may be interconnected via linking -n= bonds to form polytriazine imide ( pti ) layers with c@xmath0n@xmath1 stoichiometry @xcite .
the polymeric tgcn material synthesized by algara - siller _ _ et al.__@xcite and used as the precursor for our high pressure experiments was based on such pti layers with either ab ( space group @xmath38 ) or abc ( @xmath37 ) stacking .
the most stable tgcn polymorph found in our airss study had @xmath37 symmetry ( fig .
[ fig : low_pressure_structures ] ) .
it is predicted to lie at 0.329 ev / f.u . above the ground state at ambient pressure
( see supplemental table i ) .
it has been shown that layers based on linked tri - s - triazine ( polyheptazine : ph ) units are more stable than the pti motifs @xcite .
various levels of complexity can appear among both pti and ph connectivity patterns within a single layer , and the layers are predicted to be far from planar .
substantial buckling around the bridging -n= units that act as mechanical hinges lead to structural models with `` corrugated '' layers @xcite .
dong _ et al .
_ @xcite found a @xmath23 structure to be the ground state at 0gpa , and we have found a very similar structure of @xmath24 symmetry that is only a fraction of an mev per f.u .
higher in energy than @xmath23 ( see fig . [
fig : low_pressure_structures ] and supplemental material @xcite ) . we have also found a new corrugated ph structure of @xmath25 symmetry that is predicted to be about 7 mev per f.u .
higher in energy than @xmath23 andon analysis is the same as a structure found by gracia and kroll but in a larger unit cell with @xmath56 symmetry .
we predict that @xmath25 is the most stable phase in the range 0.052 gpa .
the @xmath23 , @xmath24 , and @xmath25 structures are based on the thsi@xmath6 motif @xcite .
the @xmath25 phase , which has a starting volume near 73 @xmath50/f.u . is expected to cross the @xmath37 enthalpy - pressure curve at approximately 25 gpa and so would not be accessible to the experimental synthesis procedure ( fig .
[ fig : enthalpy_c3n4_pbe_0 - 400_gpa ] ) .
gracia and kroll @xcite noted tubular ph structures that could lead to macroscopic forms analogous to carbon nanotubes .
many of these structure types have very large unit cell volumes ( @xmath7100 @xmath50/f.u . ) that are rapidly destabilized at high pressure , see fig . [
fig : enthalpy_c3n4_pbe_0 - 400_gpa ] .
others with energies up to 1 ev / f.u . above the ground state have volumes in the 6590 @xmath50/f.u .
range and they appear as metastable structure solutions in fig .
[ fig : structures_high_pressure ] .
one structure stands out .
this is the new @xmath2 phase described above that we have synthesized experimentally ( fig.[fig : structure_p4_32_12_ambientstack ] ) .
the experimental volume of @xmath2 of 56.26 @xmath50/f.u . is a little smaller than the pbesol volume of 59.97 @xmath50/f.u .
a further new phase of @xmath24 symmetry consisting of triazine rings linked into a three - dimensional framework via @xmath3-bonded atoms is found to occur at 0.152 ev / f.u . above the ground state at ambient pressure ( fig .
[ fig : low_pressure_structures ] ) .
the identification of such a new family of framework structures built from thermally stable c - n bonds opens up new directions for carbon nitride research .
the challenge for carbon nitride synthesis is now to produce some of these predicted phases and their unusual bonded architectures in a controlled way , and verify their structures and determine their properties experimentally .
of particular interest is the behavior of the @xmath37 phase that forms the main constituent of tgcn used as a starting material for our laser heated diamond anvil cell experiments . at ambient pressure
it lies at 0.317 ev / f.u . above the ground state and at approximately 0.2 ev / f.u . above @xmath2 , see fig.[fig : enthalpy_c3n4_pbe_0 - 400_gpa ] . during initial compression
the tgcn phase loses its initial crystallinity but it can be expected to continue along a similar enthalpy - pressure path to reach approximately 0.5 ev / f.u . above the @xmath2 energy by 50 gpa . by this pressure
@xmath2 has become substantially metastable ( by @xmath433 ev / f.u . ) relative to @xmath28-c@xmath0n@xmath1 and other dense phases but it is the next lowest in enthalpy that we have found following destabilization of the disordered and metastably compressed @xmath37 material . these predicted results can allow us to better understand the nucleation of only a few crystals of the @xmath2 phase following laser heating within the metastably compressed @xmath37 tgcn compound .
interestingly , several other c@xmath0n@xmath1 polymorphs are also predicted to occur within the 2080 gpa pressure range , with enthalpies extending to @xmath430.5 ev / f.u . above the ground state .
we might expect that further high-@xmath57,@xmath58 experiments carried out on metastably compressed starting materials could lead to identification of these new c@xmath0n@xmath1 polymorphs . structure .
( a ) ball and stick model : carbon atoms - grey , nitrogen atoms - blue .
( b ) structure drawing emphasizing the @xmath4-bonded carbon atoms ( red ) .
( c ) abstract `` lego '' representation featuring parallel , 1d struts of quasi - aromatic imidamides ( shown in blue ) linked together by tetrahedral @xmath4-bonded carbon joints ( shown in red ) .
( d ) a fragment of the hybrid @xmath3-@xmath4 bonded carbon nitride.,scaledwidth=40.0% ]
extended @xmath3-bonded framework and layered carbon nitride structures are found to be thermodynamically stable at pressures below about 2 gpa .
the lowest energy structures are built from linked polyheptazine ( ph ) units : at higher energies layered and 3d microporous frameworks based on triazine rings linked by imide groups ( pti structures ) appear . the enthalpies and volumes per f.u .
are reported in the supplemental material @xcite .
we find that the microporous ph @xmath23 structure is the most stable at zero pressure , although the similar @xmath24 structure is only a fraction of an mev per f.u .
higher in energy .
a layered @xmath25 ph structure is predicted to be stable in the range 0.052 gpa .
the @xmath2 structure is a new structure prototype not found in databases that we have discovered using airss .
it has mixed @xmath3 and @xmath4 bonding and is predicted to be stable in the range 211 gpa .
c@xmath0n@xmath1 therefore differs from pure carbon in that there is a thermodynamically stable phase between the @xmath3 and @xmath4 bonded forms .
this structure could be accessible through suitable precursors and synthetic routes at low pressure .
the relative stabilities of the most favorable c@xmath0n@xmath1 have uncertainties due to the use of approximate density functionals which amount to a few gpa at low pressures .
we have , however , sought to assess the sizes of these errors by comparing with results for the related graphite / carbon diamond system .
airss has predicted new @xmath25 , @xmath2 , @xmath31 , @xmath32 , and @xmath5 c@xmath0n@xmath1 structures to have regions of thermodynamic stability and , in addition , the new @xmath24 structure is almost stable .
our calculations predict decomposition of c@xmath0n@xmath1 at pressures above 650 gpa into diamond and pyrite - structured cn@xmath6 .
our experimental study has provided strong evidence for the synthesis in laser - heated diamond anvil cell experiments of the newly - predicted @xmath2 structure .
the identification of a new family of structures with mixed @xmath3 and @xmath4 bonding opens up a new area of exploration for these important photo- and catalytically active solid materials .
r. j. n. acknowledges financial support from the engineering and physical sciences research council ( epsrc ) of the u.k .
[ ep / j017639/1 ] .
c. j. p. acknowledges financial support from epsrc [ ep / g007489/2 ] and [ ep / k013688/1 ] , and a royal society wolfson research merit award. r. j. n. and c. j. p. acknowledge use of the archer facilities of the u.k.s national high - performance computing service ( for which access was obtained via the ukcp consortium [ ep / k013688/1 ] and ep / k014560/1 ) .
p. f. m. was supported by epsrc grant [ ep / l01709/1 ] .
e. kroke and m. schwarz , _ novel group 14 nitrides _ , coordination chemistry reviews * 248 * , 493 ( 2004 ) .
a. schwarzer , t. saplinova , and e. kroke , _ tri - s - triazines ( s - heptazines)from a `` mystery molecule '' to industrially relevant carbon nitride materials _
, coordination chemistry reviews * 257 * , 2032 ( 2013 ) .
a. thomas , a. fischer , f. goettmann , m. antonietti , j .- o .
mller , r. schlgle , and j. m. carlsson , _ graphitic carbon nitride materials : variation of structure and morphology and their use as metal - free catalysts _ , j. mater
. chem . * 18 * , 4893 ( 2008 ) .
g. goglio , d. foy , and g. demazeau , _ state of art and recent trends in bulk carbon nitrides synthesis _ , materials science and engineering : r : reports * 58 * , 195 ( 2008 ) .
t. malkow , _
critical observations in the research of carbon nitride _
, materials science and engineering : a * 292 * , 112 ( 2000 ) .
x. wang , k. maeda , a. thomas , k. takanabe , g. xin , j. m. carlsson , k. domen , and m. antonietti , _ a metal - free polymeric photocatalyst for hydrogen production from water under visible light _ , nature materials * 8 * , 76 ( 2009 ) .
b. v. lotsch , m. dblinger , j. sehnert , l. seyfarth , j. senker , o. oeckler , and w. schnick , _ unmasking melon by a complementary approach employing electron diffraction , solid - state nmr spectroscopy , and theoretical calculations structural characterization of a carbon nitride polymer _ , chem .
j. * 13 * , 4969 ( 2007 ) .
b. jurgens , e. irran , j. senker , p. kroll , h. muller , and w. schnick , _ melem ( 2,5,8-triamino - tri - s - triazine ) , an important intermediate during condensation of melamine rings to graphitic carbon nitride : synthesis , structure determination by x - ray powder diffractometry , solid - state nmr , and theoretical studies _ , j. am .
125 * , 10288 ( 2003 ) .
b. v. lotsch and w. schnick , _ new light on an old story : formation of melam during thermal condensation of melamine _ , chem .
j. * 13 * , 4956 ( 2007 ) .
m. j. bojdys , j .- o .
mueller , m. antonietti , and a. thomas , _ ionothermal synthesis of crystalline , condensed , graphitic carbon nitride _ ,
chem .- eur .
j. * 14 * , 8177 ( 2008 ) .
e. wirnhier , m. dblinger , d. gunzelmann , j. senker , b. v. lotsch , and w. schnick , _ poly(triazine imide ) with intercalation of lithium and chloride ions [ ( c@xmath0n@xmath0)@xmath6(nh@xmath54li@xmath59)@xmath0@xmath17licl ] : a crystalline 2d carbon nitride network _ ,
chem .- eur .
j. * 17 * , 3213 ( 2011 ) .
s. y. chong , j. t. a. jones , y. z. khimyak , a. i. cooper , a. thomas , m. antonietti , and m. j. bojdys , _ tuning of gallery heights in a crystalline 2d carbon nitride network _ , j. mater .
chem . a * 1 * , 1102 ( 2013 ) .
z. zhang , k. leinenweber , matt bauer , laurence a. j. garvie , p. f. mcmillan , and g. h. wolf , _ high - pressure bulk synthesis of crystalline c@xmath8n@xmath9h@xmath0@xmath17hcl : a novel c@xmath0n@xmath1 graphitic derivative _ , j. am .
soc . * 123 * , 7788 ( 2001 ) .
g. algara - siller , n. severin , s. y. chong , t. bjrkman , r. g. palgrave , a. laybourn , m. antonietti , y. z. khimyak , a. v. krasheninnikov , j. p. rabe , u. kaiser , a. i. cooper , a. thomas , and m. j. bojdys , _ triazine - based graphitic carbon nitride : a two - dimensional semiconductor _ , angew .
* 53 * , 7450 ( 2014 ) .
a. y. liu and r. m. wentzcovitch , _ stability of carbon nitride solids _ ,
b * 50 * , 10362 ( 1994 ) . d. m. teter and r. j. hemley , _ low - compressibility carbon nitrides _
, science * 271 * , 53 ( 1996 ) .
m. todd , j. kouvetakis , t. l. groy , d. chandrasekhar , d. j. smith , and p. w. deal , _ novel synthetic routes to carbon nitride _ ,
* 7 * , 1442 ( 1995 ) .
j. kouvetakis , m. todd , b. wilkens , a. bandari , and n. cave , _ novel synthetic routes to carbon - nitrogen thin films _ ,
* 6 * , 811 ( 1994 ) .
j. ortega and o. f. sankey , _ relative stability of hexagonal and planar structures of hypothetical c@xmath0n@xmath1 solids _ , phys .
b * 51 * , 2624 ( 1995 ) .
j. e. lowther , _ relative stability of some possible phases of graphitic carbon nitride _ , phys .
b * 59 * , 11683 ( 1999 ) .
m. j. bojdys , n. severin , j. p. rabe , a. i. cooper , a. thomas , and m. antonietti , _ exfoliation of crystalline 2d carbon nitride : thin sheets , scrolls and bundles via mechanical and chemical routes _ , macromol .
rapid commun . * 34 * , 850 ( 2013 )
. t. y. ma , y. tang , s. dai , and s. z. qiao , _ proton - functionalized two - dimensional graphitic carbon nitride nanosheet : an excellent metal-/label - free biosensing platform _
, small * 10 * , 2382 ( 2014 ) . m. deifallah , p. f. mcmillan , and f. cora , _ electronic and structural properties of two - dimensional carbon nitride graphenes _ , j. phys .
c * 112 * , 5447 ( 2008 ) .
s. zuluaga , l. h. liu , n. shafiq , s. m. rupich , j .- f .
veyan , y. j. chabal , and t. thonhauser , _ structural band - gap tuning in g - c@xmath0n@xmath1 _ , phys .
* 17 * , 957 ( 2015 ) .
m. l. cohen , _ calculation of bulk moduli of diamond and zinc - blende solids _ ,
phys . rev .
b * 32 * , 7988 ( 1985 ) . a. y. liu and m. l cohen , _ prediction of new low compressibility solids _ , science * 245 * , 4920 ( 1989 ) . a. y. liu and m. l. cohen , _ structural properties and electronic structure of low - compressibility materials : @xmath30-si@xmath0n@xmath1 and hypothetical @xmath30-c@xmath0n@xmath1 _ , phys . rev .
b * 41 * , 10727 ( 1990 ) . c .- m .
sung and m. sung , _ carbon nitride and other speculative superhard materials _ , materials chemistry and physics * 43 * , 1 ( 1996 ) .
h. dong , a. r. oganov , q. zhu , and g .-
qian , _ the phase diagram and hardness of carbon nitrides _ , scientific reports * 5 * , 9870 ( 2015 ) .
e. kroke , _ gt - c@xmath0n@xmath1the first stable binary carbon(iv ) nitride _ , angewandte chemie international edition * 42 * , 11134 ( 2014 ) .
e. horvath - bordon , r. riedel , p. f. mcmillan , p. kroll , g. miehe , p. a. van aken , a. zerr , p. hoppe , o. shebanova , i. mclaren , s. lauterbach , e. kroke , and r. boehler , _ high - pressure synthesis of crystalline carbon nitride imide , c@xmath6n@xmath6(nh ) _ , angew .
int . ed . *
46 * , 1476 ( 2007 ) .
a. salamat , k. woodhead , p. f. mcmillan , r. q. cabrera , a. rahman , d. adriaens , f. cora , and j .-
perrillat , _ tetrahedrally bonded dense c@xmath6n@xmath0h with a defective wurtzite structure : x - ray diffraction and raman scattering results at high pressure and ambient conditions _ , phys .
b * 80 * , 104106 ( 2009 ) .
e. stavrou , s. lobanov , h. dong , a. r. oganov , v. b. prakapenka , z. konopkova , and a. f. goncharov , _ synthesis of ultra - incompressible sp@xmath50-hybridized carbon nitride _
, arxiv e - prints * 1412.3755 * , ( 2014 ) .
a. salamat , m. deifallah , r. q. cabrera , f. cora , and p. f. mcmillan , _ identification of new pillared - layered carbon nitride materials at high pressure _
, scientific reports * 3 * , 2122 ( 2013 ) .
d. t. vodak , k. kim , l. iordanidis , p. g. rasmussen , a. j. matzger , and o. m. yaghi , _ computation of aromatic c@xmath0n@xmath1 networks and synthesis of the molecular precursor n(c@xmath0n@xmath0)@xmath0cl@xmath8 _ , chem .
j. * 9 * , 4197 ( 2003 ) .
s. ren , m. j. bojdys , r. dawson , a. laybourn , y. z. khimyak , d. j. adams , and a. i. cooper , _ porous , fluorescent , covalent triazine - based frameworks via room - temperature and microwave - assisted synthesis _ ,
* 24 * , 2357 ( 2012 ) .
m. j. bojdys , j. jeromenok , a. thomas , and m. antonietti , _ rational extension of the family of layered , covalent , triazine - based frameworks with regular porosity _ , adv . mater . * 22 * , 2202 ( 2010 ) .
li , b .- w .
sun , l. sui , d .-
qian , and m. chen , _ preparation of water - dispersible porous g - c@xmath0n@xmath1 with improved photocatalytic activity by chemical oxidation _ ,
* 17 * , 3309 ( 2015 ) . c. reece , d. j. willock , and a. trewin ,
_ modelling analysis of the structure and porosity of covalent triazine - based frameworks _ , phys .
* 17 * , 817 ( 2015 ) .
a. r. oganov , r. j. hemley , r. m. hazen , and a. p. jones , _ structure , bonding , and mineralogy of carbon at extreme conditions _ , reviews in mineralogy and geochemistry * 75 * , 47 ( 2013 ) .
m. lazzeri and a. barreiro , _ carbon - based nanoscience _ , elements * 10 * , 447 ( 2014 ) .
y. gong , z. wei , j. wang , p. zhang , h. li , and y. wang , _ design and fabrication of hierarchically porous carbon with a template - free method _ , scientific reports * 4 * , 6349 ( 2014 ) .
i. a. baburin , d. m. proserpio , v. a. saleev , and a. v. shipilova , _ from zeolite nets to @xmath4 carbon allotropes : a topology - based multiscale theoretical study _ , phys .
* 17 * , 1332 ( 2015 ) .
f. p. bundy , h. p. bovenkerk , h. m. strong , and r. h. wentorf jr .
, _ diamond - graphite equilibrium line from growth and graphitization of diamond _ , j. chem . phys . * 35 * , 383 ( 1961 ) .
r. z. khaliullin , h. eshet , t. d. khne , j. behler , and m. parrinello , _ nucleation mechanism for the direct graphite - to - diamond phase transition _ , nature materials * 10 * , 693 ( 2011 )
. m. martinez - canales , c. j. pickard , and r. j. needs , _ thermodynamically stable phases of carbon at multiterapascal pressures _
108 * , 045704 ( 2012 ) .
r. f. smith , j. h. eggert , r. jeanloz , t. s. duffy , d. g. braun , j. r. patterson , r. e. rudd , j. biener , a. e. lazicki , a. v. hamza , j. wang , t. braun , l. x. benedict , p. m. celliers , and g. w. collins , _ ramp compression of diamond to five terapascals _ , nature * 511 * , 330 ( 2014 ) .
c. j. pickard and r. j. needs , _ piling up the pressure _ , nature news and views * 511 * , 294 ( 2014 ) .
c. j. pickard and r. j. needs , _ high - pressure phases of silane _ , phys .
lett . * 97 * , 045504 ( 2006 ) .
c. j. pickard and r. j. needs , _ ab initio random structure searching _ , j. phys . : condens .
matter * 23 * , 053201 ( 2011 ) . c. j.
pickard and r. j. needs , _ structure of phase iii of solid hydrogen _ , nature physics * 3 * , 473 ( 2007 ) . c. j. pickard ,
m. martinez - canales , and r. j. needs , _ density functional theory study of phase iv of solid hydrogen _ , phys .
b * 85 * , 214114 ( 2012 ) : phys . rev .
b * 86 * , 059902(e ) ( 2012 ) . s. ninet , f. datchi , p. dumas , m. mezouar , g. garbarino , a. mafety , c. j. pickard , r. j. needs , and a. m. saitta , _ experimental and theoretical evidence for an ionic crystal of ammonia at high pressure _ , phys .
b * 89 * , 174103 ( 2014 ) .
a. d. fortes , e. suard , m .- h .
lemee - cailleau , c. j. pickard , and r. j. needs , _ crystal structure of ammonia monohydrate ii _ , j. am .
soc . * 131 * , 13508 ( 2009 ) .
g. i. g. griffiths , a. d. fortes , c. j. pickard , and r. j. needs , _ crystal structure of ammonia dihydrate ii _ , j. chem .
phys . * 136 * , 174512 ( 2012 ) . c. j.
pickard and r. j. needs , _ metallization of aluminum hydride at high pressures : a first - principles study _ , phys .
b * 76 * , 144114 ( 2007 ) .
a. dewaele , n. g. worth , c. j. pickard , r. j. needs , s. pascarelli , o. mathon , m. mezouar , and t. irifune , _ xenon oxides under pressure _ , unpublished .
l. pauling , _ the principles determining the structure of complex ionic crystals _ , j. am . chem . soc . * 51 * , 1010 ( 1929 ) . s. j. clark , m. d. segall , c. j. pickard , p. j. hasnip , m. i. j. probert , k. refson , and m. c. payne , _ first principles methods using castep _
, z. kristallogr . *
220 * , 567 ( 2005 ) .
d. vanderbilt , _ soft self - consistent pseudopotentials in a generalized eigenvalue formalism _ , phys .
b * 41 * , 7892 ( 1990 ) .
j. p. perdew , k. burke , and m. ernzerhof , _ generalized gradient approximation made simple _ , phys .
lett . * 77 * , 3865 ( 1996 ) ; * 78 * , 1396(e ) ( 1997 ) .
see supplemental material at http://link.aps.org/supplemental/ ..... for additional information about the c@xmath0n@xmath1 structures .
w. kohn and l. j. sham , _ self - consistent equations including exchange and correlation effects _ , phys .
* 140 * , a1133 ( 1965 ) .
z. wu and r. e. cohen , _ more accurate generalized gradient approximation for solids _ , phys .
b * 73 * , 235116 ( 2006 ) .
j. p. perdew , a. ruzsinszky , g. i. csonka , o. a. vydrov , g. e. scuseria , l. a. constantin , x. zhou , and k. burke , _ restoring the density - gradient expansion for exchange in solids and surfaces _ , phys
lett . * 100 * , 136406 ( 2008 ) ; * 102 * , 039902 ( 2009 ) .
s. grimme , _ semiempirical gga - type density functional constructed with a long - range dispersion correction _ , j. comput .
* 27 * , 1787 ( 2006 ) .
j. gracia and p. kroll , _ corrugated layered heptazine - based carbon nitride : the lowest energy modifications of c@xmath0n@xmath1 ground state _ , j. mater
. chem . * 19 * , 3013 ( 2009 ) .
y. guo , w. a. goddard iii , _ is carbon nitride harder than diamond ?
no , but its girth increases when stretched ( negative poisson ratio ) _ , chem .
. lett . * 237 * , 72 ( 1995 ) .
s. petitgirard , a. salamat , p. beck , g. weck , and p. bouvier , _
strategies for in situ laser heating in the diamond anvil cell at an x - ray diffraction beamline _ , j. synchrotron rad . * 21 * , 89 ( 2014 ) . | we used _ ab initio _ random structure searching ( airss ) to investigate polymorphism in c@xmath0n@xmath1 carbon nitride as a function of pressure .
our calculations reveal new framework structures , including a particularly stable chiral polymorph of space group @xmath2 containing mixed @xmath3 and @xmath4-bonding , that we have produced experimentally and recovered to ambient conditions . as pressure
is increased a sequence of structures with fully @xmath4-bonded c atoms and three - fold coordinated n atoms is predicted , culminating in a dense @xmath5 phase above 250 gpa . beyond 650 gpa
we find that c@xmath0n@xmath1 becomes unstable to decomposition into diamond and pyrite - structured cn@xmath6 . |
the author thanks to anonymous referee for helpful and constructive comments .
99 j. e. hirsch , proc .
nat . acad .
sci . * 46 * , 16569 ( 2005 ) [ arxiv : physics/0508025 ] .
p. ball , nature * 436 * , 900 ( 2005 ) .
d. f. taber , science * 309 * , 4 ( 2005 ) .
l. bornmann and h .- d .
daniel , embo reports * 10(1 ) * , 2 ( 2009 ) .
w. glnzel , scientometrics * 67 * , 315 ( 2006 ) .
l. bornmann and h .- d .
daniel , scientometrics * 65 * , 391 ( 2005 ) .
s. lehmann , a. d. jackson and b. e. lautrup , arxiv : physics/0512238 ; nature * 444 * , 1003 ( 2006 ) .
j. e. hirsch , proc .
sci . * 104 * , 19193 ( 2007 ) . [ arxiv:0708.0646 ] .
l. egghe , issi newsletter * 2(1 ) * , 8 ( 2006 ) .
s. b. popov , arxiv : physics/0508113 .
d. katsaros , l. akritidis and p. bozanis , j. am . soc . inf .
* 60n5 * , 1051 ( 2009 ) .
m. schreiber , europhys .
lett . * 78 * , 30002 ( 2007 ) [ arxiv : physics/0701231 ] . c. e. shannon , bell syst
. tech .
j. * 27 * , 379 ( 1948 ) ; * 27 * , 623 ( 1948 ) .
r. p. feynman , _
feynman lectures on computation , _ edited by anthony j.g .
hey and robin w. allen ( addison - weslay , reading , 1996 ) , p. 123
. s. kullback and r. a. leibler , annals math .
statist .
* 22 * , 79 ( 1951 ) .
t. m. cover and j. a. thomas , _ elements of information theory _ ( wiley - interscience , new york , 1991 ) .
r. gray , _ entropy and information theory _ ( springer - verlag , new york , 1990 ) .
s. guiasu , _ information theory with applications _ ( mcgraw - hill , new york , 1977 ) .
p. g. hall , statist .
* 24 * , 25 ( 2009 ) [ arxiv:0910.3546 ] .
r. adler , j. ewing , p. taylor , statist .
* 24 * , 1 ( 2009 ) [ arxiv:0910.3529 ] .
b. w. silverman , statist .
* 24 * , 15 ( 2009 ) [ arxiv:0910.3532 ] .
m. schreiber , arxiv:1005.5227v1 [ physics.soc-ph ] .
s. redner , j. stat .
l03005 ( 2010 ) [ arxiv:1002.0878 ] . m. mitzenmacher , internet mathematics * 1n2 * , 226 ( 2004 )
. g. k. zipf , _ human behavior and the principle of least effort , _
( addison - wesley , cambridge , 1949 ) .
w. li , glottometrics * 5 * , 14 ( 2003 ) . m. v. simkin and v. p. roychowdhury , arxiv : physics/0601192v2 [ physics.soc-ph ] .
m. e. j. newman , contemp .
phys . * 46 * , 323 ( 2005 ) [ arxiv : cond - mat/0412004 ] .
d. j. de s. price , science * 149 * , 510 ( 1965 ) .
s. redner , eur .
j. b * 4 * , 131 ( 1998 ) [ arxiv : cond - mat/9804163 ] .
z. k. silagadze , complex syst .
* 11 * , 487 ( 1997 ) [ arxiv : physics/9901035 ]
. this paper demonstrates a violation of causality : written in 1999 , it was published in 1997 !
l. egghe and r. rousseau , scientometrics * 69 * , 121 ( 2006 ) .
k. barcza and a. telcs , scientometrics * 81 * , 513 ( 2009 ) .
j. e. iglesias and c. pecharromn , scientometrics * 73 * , 303 ( 2007 ) . m. v. simkin and v. p. roychowdhury , complex syst .
* 14 * , 269 ( 2003 ) [ arxiv : cond - mat/0212043 ] .
m. v. simkin and v. p. roychowdhury , annals improb .
11 * , 24 ( 2005 ) [ arxiv : cond - mat/0305150 ] .
albert einstein , quoted in @xcite .
f. radicchi , s. fortunato , b. markines and a. vespignani , phys .
e * 80 * , 056103 ( 2009 ) [ arxiv:0907.1050v2 ] . | a new indicator , a real valued @xmath0-index , is suggested to characterize a quality and impact of the scientific research output .
it is expected to be at least as useful as the notorious @xmath1-index , at the same time avoiding some its obvious drawbacks .
however , surprisingly , the @xmath1-index is found to be quite a good indicator for majority of real - life citation data with their alleged zipfian behaviour for which these drawbacks do not show up .
the style of the paper was chosen deliberately somewhat frivolous to indicate that any attempt to characterize the scientific output of a researcher by just one number always has an element of a grotesque game in it and should not be taken too seriously .
i hope this frivolous style will be perceived as a funny decoration only .
[ [ section ] ] sound , sound your trumpets and beat your drums ! here
it is , an impossible thing performed : a single integer number characterizes both productivity and quality of a scientific research output . suggested by jorge hirsch @xcite , this simple and intuitively appealing @xmath1-index has shaken academia like a storm , generating a huge public interest and a number of discussions and generalizations @xcite . a russian physicist with whom
i was acquainted long ago used to say that the academia is not a christian environment .
it is a pagan one , with its hero - worship tradition . but hero - worshiping requires ranking . and
a simple indicator , as simple as to be understandable even by dummies , is an ideal instrument for such a ranking .
@xmath1-index is defined as given by the highest number of papers , @xmath1 , which has received @xmath1 or more citations .
empirically , @xmath2 with @xmath3 ranging between three and five @xcite . here @xmath4 stands for the total number of citations . and
now , with this simple and adorable instrument of ranking on the pedestal , i m going into a risky business to suggest an alternative to it .
am i reckless ? not quite .
i know a magic word which should impress pagans with an irresistible witchery .
claude shannon introduced the quantity @xmath5 which is a measure of information uncertainty and plays a central role in information theory @xcite . on the advice of john von neumann ,
shannon called it entropy . according to feynman @xcite
, von neumann declared to shannon that this magic word would give him `` a great edge in debates because nobody really knows what entropy is anyway '' .
armed with this magic word , entropy , we have some chance to overthrow the present idol .
so , let us try it !
citation entropy is naturally defined by ( [ sentropy ] ) , with @xmath6 where @xmath7 is the number of citations on the @xmath8-th paper of the citation record .
now , in analogy with ( [ hind ] ) , we can define the citation record strength index , or @xmath0-index , as follows @xmath9 where @xmath10 is the maximum possible entropy for a citation record with @xmath11 papers in total , corresponding to the uniform citation record with @xmath12 .
note that ( [ sind ] ) can be rewritten as follows @xmath13 where @xmath14 is the so called kullback - leibler relative entropy @xcite , widely used concept in information theory @xcite .
for our case , it measures the difference between the probability distribution @xmath15 and the uniform distribution @xmath16 .
the kullback - leibler relative entropy is always a non - negative number and vanishes only if @xmath15 and @xmath17 probability distributions coincide . that s all .
here it is , a new index @xmath0 afore of you .
concept is clear and the definition simple .
but can it compete with the @xmath1-index which already has gained impetus ?
i do not know .
in fact , it does not matter much whether the new index will be embraced with delight or will be coldly rejected with eyes wide shut .
i sound my lonely trumpet in the dark trying to relax at the edge of precipice which once again faces me .
nevertheless , i feel @xmath0-index gives more fair ranking than @xmath1-index , at least in the situations considered below
. some obvious drawbacks of the @xmath1-index which are absent in the suggested index are the following : * @xmath1-index does not depend on the extra citation numbers of papers which already have @xmath1 or more citations .
increasing the citation numbers of most cited @xmath1 papers by an order of magnitude does not change @xmath1-index .
compare , for example , the citation records @xmath18 @xmath19 and @xmath20 which have @xmath21 and @xmath22 respectively .
* @xmath1-index will not change if the scientist losses impact ( ceases to be a member of highly cited collaboration ) .
for example , citation records @xmath23 and @xmath24 both have @xmath25 , while @xmath0-index drops from 4.8 to 3.0 . * @xmath1-index will not change if the citation numbers of not most cited papers increase considerably .
for example , @xmath26 @xmath27 and @xmath28 + @xmath29 both have @xmath25 , while @xmath0-index increases from 3.0 to 6.9 .
of course , @xmath0-index itself also has its obvious drawbacks .
for example , it is a common case that an author publishes a new article which gains at the beginning no citations . in this case
the entropy will typically decrease and so will the @xmath0-index .
i admit such a feature is somewhat counter - intuitive for a quantity assumed to measure the impact of scientific research output .
however , the effect is only sizable for very short citation records and in this case we can say that there really exists some amount of objective uncertainty in the estimation of the impact .
anyway , you can hardly expect that a simple number can substitute for complex judgments implied by traditional peer review .
of course , the latter is subjective .
nothing is perfect under the moon .
it is tempting , therefore , to try to overcome this possible subjectivity of peer review by using `` simple and objective '' numerical measures . especially because some believe academic managers `` do nt have many skills , but they can count '' @xcite . in reality , however , it is overwhelmingly evident @xcite that simple numerical indicators , like @xmath0-index proposed here , neither can eliminate subjectivity in management science nor prevent a dull academic management .
but they can add some fun to the process , if carefully used : `` citation statistics , impact factors , the whole paraphernalia of bibliometrics may in some circumstances be a useful servant to us in our research .
but they are a very poor master indeed '' @xcite
. it will be useful to compare two `` servants '' on some representative set of real - life citation records and fig.[sh ] gives a possibility .
one hundred citation records were selected more or less randomly from the _ citebase _
@xcite citation search engine .
fig.[sh ] shows @xmath1-index plotted against @xmath0-index for these records .
what a surprise ! @xmath1 and @xmath0 indexes are strongly correlated and almost equal for a wide range of their values .
of course , the coefficient , @xmath30 , in ( [ sind ] ) was chosen to make these two indexes relatively equal for _ some _ citation records , but i have not expected them to remain close for _ all _ citation records .
there is some mystery here .
let us try to dig up what it is .
recent studies indicate that from the one hand the normalized entropy does not change much from record to record with @xmath31 @xcite , and on the other hand , the scaling law ( [ hind ] ) for the @xmath1-index is well satisfied with @xmath32 @xcite .
these two facts and a simple calculation imply that @xmath0 and @xmath1 indexes are expected to be approximately equal .
therefore , the real question is why these regularities are observed in the citation records ?
common sense and some experience on the citation habits tell us that these habits are the subject of preferential attachment the papers that already are popular tend to attract more new citations than less popular papers .
it is well known that the preferential attachment can lead to power laws @xcite .
namely , with regard to citations , if the citations are ranked in the decreasing order @xmath33 then the zipf s law @xcite says that @xmath34 empirical studies reveal @xcite that the zipf s law is a ubiquitous and embarrassingly general phenomenon and in many cases @xmath35 .
the citation statistics also reveals it @xcite .
therefore let us assume the simplest case , distribution ( [ zipf ] ) with @xmath36 .
then @xmath37 condition determines the hirsch index @xmath38 and we see that , if the individual citation records really follow the zipf distribution @xmath39 , the hirsch index is a really good indicator as it determines the only relevant parameter @xmath40 of the distribution .
in fact , the number of papers are finite and we have a second parameter , the total number of papers @xmath11 . for sufficiently large @xmath11 , @xmath41 therefore , from ( [ hindc ] ) we get the following scaling law @xmath42 instead of ( [ hind ] ) .
however , @xmath43 varies from @xmath44 to @xmath45 then @xmath11 varies from @xmath46 to @xmath47 and this explains the observations of @xcite .
note that the relation ( [ hindc ] ) was already suggested by egghe and rousseau , on the basis of zipf s law , in @xcite . for other references where
the connections between zipf s law and @xmath1-index are discussed see , for example , @xcite .
as for @xmath0-index , zipfian distribution @xmath39 implies the probabilities ( assuming @xmath11 is large ) @xmath48 and , hence , the following entropy @xmath49 but @xmath50 therefore , @xmath51 this expression gives @xmath52 for @xmath53 and @xmath54 for @xmath55 which are quite close to what was found in @xcite ( although for a small data set ) . because @xmath56 is small
, we have the following scaling behaviour for the @xmath0-index from ( [ sind1 ] ) : @xmath57 fig.[ssn ] and fig.[hhn ] demonstrate an empirical evidence for these scaling rules from the _ citebase _ citation records mentioned above . as we see , the scalings ( [ sindn ] ) and ( [ hindn ] ) are quite pronounced in the data . however , there are small number of exceptions
. therefore , citation patterns are not always zipfian .
inspection shows that in such cases the zipfian behavoir is spoiled by the presence of several renowned papers with very high number of citations .
if they are removed from the citation records , the zipfian scalings for @xmath0 and @xmath1 indexes are restored for these records too .
to conclude , we wanted to overthrow the king but the king turned out to be quite healthy .
the secret magic that makes @xmath1-index healthy is zipfian behaviour of citation records .
under such behaviour , a citation record really has only one relevant parameter and the hirsch index just gives it .
however , not all citation records exhibit zipfian behaviour and for such exceptions the new index related to the entropy of the citation record probably makes better justice .
but , i m afraid , this is not sufficient to sound our trumpets and cry `` le roi est mort .
vive le roi ! ''
the magic has an another side however .
the zipfian character of citation records probably indicate the prevalence of preferential attachment and some randomness in the citation process .
we can use a propagation of misprints in scientific citations to estimate how many citers really read the original papers and the striking answer is that probably only 20% do , the others simply copying the citations @xcite . of course , it will go too far to assume that `` copied citations create renowned papers '' @xcite , but these observations clearly give a caveat against taking all the indexes based on the number of citations too seriously .
`` not everything that can be counted counts , and not everything that counts can be counted '' @xcite .
such local measures of the research impact estimation should be taken with a grain of salt and at least supplemented by different instruments for analyzing the whole citation network , like the one given in @xcite . |
as the increase of connectivity in and between different complex systems , spread of many diseases or viruses is becoming more and more prevalent in our society@xcite .
for instance , outbreaks of many infectious diseases , including severe acute respiratory syndromes ( sars ) , swine flu ( h1n1 ) , and the recent ebola virus , caused great damage and loss of life .
the spread of computer and mobile phone viruses brought about a great deal trouble to human life and serious damage to economy .
understanding the intrinsic mechanisms of those spreading processes and designing efficient control strategies become very important and urgent tasks , which bring together a lot of researchers from areas of biology , sociology , mathematics , physics , engineering , etc@xcite .
mathematical modeling of epidemic spreading has a long history of more than two hundred years@xcite .
generally , the population is divided into several classes : susceptible , infected and recovered individuals .
susceptible individuals represent those who can contract the infection .
infected individuals were previously susceptible individuals and got infected by the disease .
recovered individuals are those who have recovered from the infection . in the susceptible - infected - susceptible ( sis ) model@xcite , infected individuals can recover from the disease and become susceptible individuals again . while in the susceptible - infected - recovered ( sir ) model@xcite , infected individuals no longer get infected after recovery from the disease , which are assumed to get the permanent immunity . in classical epidemiology
, a common assumption is that individuals in a class is treated similarly , and have equal probability to contact with everyone else@xcite .
however , the recent abundance of data demonstrates that both the connectivity pattern and the contact rate are heterogeneous among real - world complex networks@xcite , which means the traditional deterministic differential equations and many other related results of epidemic processes are inadequate in real - world situations .
this great stimulates the research of epidemics on real - world complex networks@xcite . due to the complexity of real - world networks ,
the mean - field approach@xcite and the generating function approach@xcite are used to drive the analytic results of epidemics spreading .
one of the remarkable results obtained by pastor - satorras and vespignani@xcite shows that in the limit of a network of infinite size , the epidemic threshold of the sis model tends to zero asymptotically in scale - free networks with power - law parameter in ( 2 , 3 ] . for sir model , it was found that in the thermodynamic limit , not only the threshold tends to vanish , but also the time for the stabilization of the infection becomes very small@xcite . by using the message - passing approach ,
karrer and newman@xcite calculated the probabilities for any node and any time to be in state s , i , and r on tree structure .
many other explicit results of sir model are obtained by mapping the sir model to the percolation process@xcite . also , effects of degree correlations@xcite , clustering@xcite , weights and directions of edges@xcite on epidemic spreading are broadly discussed . on the other hand ,
various efficient immunization protocols@xcite have been designed for controlling the spread of epidemics on networks .
recently much attention has been transferred to epidemic spreading in temporal and multiplex networks@xcite .
addition to diseases or viruses , there are usually many other substances spreading in networks like information packets , goods , ideas , etc . , which depend on the specific types of the networks .
epidemic spreading is often coupled with the delivery of these substances .
for example , hiv spreads through the exchange of body fluids among individuals in contact networks .
computer viruses spread with the delivery of information packets in computer networks .
flu often spreads by air traffics among different spatial areas .
therefore , understanding the mechanisms of these coupled spreading processes and how these processes affect each other is significant for designing efficient epidemic immunization strategies .
meloni et al@xcite first studied the effects of traffic flow on epidemic spreading .
they found that the epidemic threshold in the sis model decreases as flow increases , and emergence of traffic congestion slows down the spread of epidemics .
then , yang et al@xcite further studied the relation between traffic dynamics and the sis epidemic model , and found that the epidemic can be controlled by fine tuning the local or global routing schemes .
furthermore , they obtained that the epidemic threshold can be enhanced by cutting some specific edges in the network@xcite .
the impacts of traffic dynamics on sir epidemic model havent been reported in literature . in this paper , we study the traffic - driven sir spreading dynamics in complex networks .
we focus on the instantaneous size of infected population , and the final size of ever infected population . based on these two properties ,
we study how the packets transmission process governed by given routing protocols affects the epidemic spreading .
our model includes two coupled processes : packet delivery process and the epidemic spreading process .
we will introduce our model in the context of computer networks .
we assume that nodes in the network are identical which can generate , receive and deliver information packets .
each node has a queue obeying the first - in - first - out ( fifo ) rule for storing packets .
the length of the queue is set infinite .
load of a node is the number of packets in its queue .
every node generates packets at a rate @xmath1 .
for example if @xmath2 , each time an arbitrary node generates one packet definitely and another one with probability 0.5 .
the destination nodes of the packets are chosen randomly , and the packets will be removed from the network after arriving at the destination nodes .
the transmission of packets is governed by the efficient routing protocol proposed by yan et al@xcite . for an arbitrary path @xmath3 of length @xmath4 between node @xmath5 and @xmath6 , denoted as @xmath7 , its routing cost is defined as follows : @xmath8 where @xmath9 is the degree of node @xmath10 , and @xmath0 is a control parameter .
the sum runs over all the intermediate nodes of path @xmath3 .
the efficient paths for delivering packets are defined to be those which have the minimum routing costs .
if there are many efficient paths between two nodes , we randomly chose one for delivering packets . according to eq . 1
, @xmath0 determines the routing cost of a path .
when @xmath11 , paths with large - degree nodes usually have large routing costs .
thus , efficient paths tend to be those paths composed of small - degree nodes when @xmath11 , and vice versa . each time a node can deliver at most @xmath12 packets .
when @xmath13 , all the packets can be delivered without delay , there is no traffic congestion .
the overall load of the network is constant after a short transient time .
when @xmath12 is a constant value , there is a critical packet generation rate @xmath14 .
when @xmath15 , there is no traffic congestion .
when @xmath16 , the network generates more packets than it can deliver .
as a result , the overall load of the network increases with time , which is the traffic congestion phenomenon .
we use the order parameter to characterize the traffic congestion , which is as follows@xcite : @xmath17 where @xmath18 is the total number of packets in the network at time @xmath19 .
when @xmath20 , the generated and delivered packets are balanced , and the network is under the free flow state . when @xmath21 , packets accumulate continuously in the network , which indicates that there exists traffic congestion . as in the traditional sir model ,
nodes in the network are divided into three classes : susceptible nodes , infected nodes and recovered nodes .
initially , all the nodes in the network are susceptible nodes , which perform the normal function of generating , delivering , and receiving packets , and the network flow is stable
. then we randomly select a node to be an infected one which is the original source of the infection .
infected nodes generate infection packets instead of normal packets at each time step . with the delivery of these infection packets ,
more and more susceptible nodes get infected after receiving the infection packets . infected nodes get recovered and become recovered nodes with probability @xmath22 at each time step . recovered nodes generate normal packets , and they can also make the infection packets into normal packets . thus , recovered nodes block the epidemic spreading .
the epidemic spreading ends when all the infected nodes become recovered nodes .
the transitions between the susceptible , infected , and recovered nodes for our model are shown in figure 1 .
the underlying networks are random networks generated by the erds - rnyi ( er ) model@xcite and scale - free networks generated by the static model@xcite .
also , some real - world networks@xcite are used in the simulations .
we assume that , at time @xmath19 , the numbers of susceptible nodes , infected nodes and recovered nodes are @xmath23 , @xmath24 and @xmath25 respectively .
first , we study the time evolution of our model on the er model , the static model , and the email network@xcite .
we add the infection to the network at @xmath26 when the network is under free flow state , by randomly selecting a node to become infected . in figure 2 ( a ) , ( b ) and ( c )
, we see that @xmath23 decreases with @xmath19 greatly , and then tends to be stable . on the contrary
, @xmath25 increases with @xmath19 abruptly , and then converges at @xmath27 , which is the maximum number of recovered ( or ever infected ) node .
@xmath28 reflects the range of the infection . differently from @xmath23 and @xmath25
, @xmath24 increases with @xmath19 first , then decreases with @xmath19 .
the peak of @xmath24 , denoted by @xmath29 , represents the maximum instantaneous number of infected nodes during the epidemic spreading process .
@xmath29 reflects how intense the epidemic spreading is . at any time @xmath19 ,
the sum of @xmath23 , @xmath24 , and @xmath25 equals the size of the network .
routing parameter @xmath0 is one of the important factors in our model .
@xmath0 determines the efficient paths for the delivery of normal and infection packets . in figure 2(d ) , ( e ) , and ( f ) , we see that @xmath23 , @xmath24 and @xmath25 all vary with @xmath0 , and for any of the three , the trends of the curves of different @xmath0 are similar .
we focus on the maximum instantaneous size of infected population @xmath29 , and the final size of ever infected population @xmath28 . in figure 3 ( a ) , ( b ) and ( c )
, we see that both @xmath29 and @xmath28 increase with @xmath0 first , then decrease with @xmath0 .
there are optimal @xmath0 which correspond to the maximum @xmath29 and @xmath28 respectively .
note that the optimal @xmath0 for @xmath29 and @xmath28 are close , but not necessarily the same .
there are jumps in both @xmath29 and @xmath28 when @xmath30 is near zero .
to explain these results , we calculate the load variance @xmath31 of nodes , which is defined as follows : @xmath32 where @xmath33 is the load of node @xmath10 at time @xmath19 .
@xmath34 is a constant value and is large enough to ensure accurate calculation of the average load @xmath35 of node @xmath10 .
when @xmath36 is small , the load distribution is relatively even in the network , and vice versa . in figure 3 ( d ) , ( e ) and ( f )
, @xmath36 decreases first and then increases with @xmath0 .
there is optimal @xmath0 which leads to the minimum @xmath36 .
interestingly , the values of the optimal @xmath0 for @xmath36 , @xmath29 and @xmath28 are close , which indicates that homogeneous load distribution facilitates the epidemic spreading .
there is also abrupt decrease in @xmath36 when @xmath0 is near zero .
this is because the efficient paths are very different for @xmath37 and @xmath38 , and the load is abruptly redistributed from large - degree nodes to small - degree nodes when @xmath0 increases from below zero to above zero .
this also accounts for the jumps in @xmath29 and @xmath28 .
the results are consistent for both the models networks and the email network as shown in figure 3 . ,
@xmath28 and @xmath36 vs. @xmath0 .
the network models , the email network , as well as all the parameters are the same as in figure 2 .
the results are the average over 100 independent runs .
, width=384,height=288 ] packet generation rate @xmath1 also has great impacts on the epidemic spreading . in figure 4
, we see that the peak of @xmath39 for @xmath40 is almost 4 times of the peak of @xmath28 for @xmath41 .
the peak of @xmath29 for @xmath40 is almost 6 times of the peak of @xmath29 for @xmath41 .
also the positions of the peaks for @xmath42 and 0.5 are different . in figure 5 ( a )
, we clearly see both the maximum @xmath29 and the maximum @xmath28 increase with @xmath1 .
when @xmath1 becomes large , infected nodes will generate more infected packets , which facilitates the epidemic spreading . in figure 5 ( b ) , we see that @xmath43 decreases with @xmath1 , which indicates an increase of dependency on large - degree nodes . to illustrate this , we focus on the top 1% largest degree nodes , and
calculate the average ratio @xmath44 of the number of infection packets a large degree node delivered when it was infected to the number of infection packets it cured after it recovered . in figure 6
, @xmath44 increases with @xmath1 , which means that the role large - degree nodes play in facilitating the epidemic spreading becomes more and more remarkable compared to the role they play in inhibiting the epidemic spreading .
this explains why the epidemic spreading becomes more dependent on large - degree nodes to spread widely when @xmath1 increases as shown in figure 5 ( b ) . and @xmath28 vs. @xmath0 for different @xmath1 .
the results are the average over 100 independent runs . , width=480,height=192 ] , the maximum @xmath28 and the corresponding @xmath43 vs. @xmath1 . the network model and
the parameters are the same as in figure 4 .
the results are the average over 100 independent runs .
, width=384,height=192 ] vs. @xmath1 .
the results are the average over 10 independent runs . , width=288,height=288 ]
first , we show the impacts of network density on the epidemic spreading . in figure 7
( a ) and ( c ) , we see that the maximum @xmath29 and the maximum @xmath28 increase with average degree @xmath45 for both random networks and scale - free networks .
this indicates that more edges facilitate the epidemic spreading . in figure 7 ( b ) and
( d ) , @xmath43 decreases with @xmath45 , but is large than zero .
this means that , to realize an intense and wide epidemic spreading , generally the paths for the delivery of infection packets should be biased towards including small - degree nodes , but when the network becomes dense , the degree of the dependence of large - degree nodes in the transmission of infection packets should be increased accordingly . , the maximum @xmath28 and the corresponding @xmath43 vs. @xmath45 .
the results are the average over 100 independent runs .
, width=384,height=288 ] then we show the impacts of power - law parameter @xmath46 on the epidemic spreading in figure 8 .
both the maximum @xmath29 and the maximum @xmath28 increase with @xmath46 first , and then tend to be stable .
the @xmath43 increases slightly with @xmath46 .
we infer from figure 8 that homogeneous network structure facilitates the epidemic spreading . , the maximum @xmath28 and @xmath43 vs. @xmath46
. the results are the average over 100 independent runs .
, width=384,height=192 ] when delivery capacity @xmath47 is a constant value , and the packet generation rate @xmath1 is large enough , the packets ca nt be delivered in time , and then the number of packets accumulated in the network increases with time , which is the traffic congestion phenomenon . in the simulation , @xmath12 is set to 10 , @xmath48 is calculated according to eq . 2 to quantify the degree of traffic congestion .
in figure 9 ( a ) , we see that only when @xmath0 is in [ 0.6 , 1 ] , there is no traffic congestion in the network , which is indicated by @xmath20 .
otherwise , there exists traffic congestion where @xmath49 . in figure 9 ( b ) , we see that both @xmath29 and @xmath28 increase with @xmath0 first , and then decrease with @xmath0 .
the optimal @xmath0 for @xmath29 and @xmath28 are 0.7 and 0.9 respectively , where there is no traffic congestion .
also , we obtain that when the traffic congestion is not serious like @xmath0 in [ 1.5 , 2 ] ( figure 9 ( a ) ) , the epidemic spreading can still spread intensely and widely , which is inferred by @xmath29 and @xmath28 in figure 9 ( b ) . , @xmath29 and @xmath28 vs. @xmath0 .
the results are the average over 100 independent runs .
, width=384,height=192 ] then we fix @xmath0 to be zero , and study how the maximum @xmath29 and the maximum @xmath28 vary with @xmath1 . in figure 10 , we obtain that both the maximum @xmath29 and the maximum @xmath28 increase first and then decrease with @xmath1 , which is consistent for both random networks and scale - free networks .
the reason for these results is that when @xmath1 is small , there is no traffic congestion , and the infection becomes intense and wide spread with increase of @xmath1 .
however , when @xmath1 is large enough , traffic congestion appears , which inhibits the epidemic spreading . and the maximum @xmath28 vs. @xmath1 .
the results are the average over 100 independent runs .
, width=384,height=192 ] in addition to the efficient routing protocol , we also study the impacts of the other static routing protocols on the epidemic spreading . if @xmath50 is replaced with @xmath51 in eq .
1 , then we get the cost function of the optimal routing protocol@xcite . for the optimal routing protocol , we only present the results of @xmath52 and @xmath28 vs. @xmath0 on three real - world networks@xcite in figure 11 .
there are also optimal @xmath0 which correspond to the maximum @xmath29 and the maximum @xmath28 respectively . when @xmath0 is near zero , there are also jumps in @xmath29 and @xmath28 due to the significant change of paths for delivering packets . for the optimal routing and the efficient routing ,
either the maximum @xmath29 or the maximum @xmath28 is very close .
the difference lies in that @xmath29 and @xmath28 of the optimal routing vary much slower than that of the efficient routing . and @xmath28 vs. @xmath0 for the optimal routing and the efficient routing on three real - world networks .
the results are the average over 100 independent runs .
, width=480,height=240 ]
in summary , we propose a traffic - driven sir epidemic model and study the impacts of several factors on our model . we find that the epidemic spreading is greatly affected by the load distribution , and homogeneous load distribution
facilitates the epidemic spreading . increasing the network density or network homogeneity
will enhance the epidemic spreading .
large - degree nodes have dual effects on the epidemic spreading , since large - degree infected nodes facilitate the epidemic spreading , while large - degree recovered nodes greatly inhibit the epidemic spreading . to realize an intense and wide epidemic spreading , the paths for the delivery of packets
are generally biased towards including small - degree nodes .
increasing packet generation rate generally favors the epidemic spreading .
however , when the amount of generated packets is larger than the delivery capacity of the network , there will exit traffic congestion , which blocks the epidemic spreading .
also , we find similar impacts of different static routing protocols on the traffic - driven sir epidemic spreading .
our work helps understanding the interplay between traffic dynamics and epidemic spreading , and provides some clues for network immunization .
this work was supported by the natural science foundation of china ( grant no .
61304154 ) , the specialized research fund for the doctoral program of higher education of china ( grant no .
20133219120032 ) , and the postdoctoral science foundation of china ( grant no .
2013m541673 ) .
10 s. n. dorogovtsev , a. v. goltsev , j. f. f. mendes , _ rev . mod .
phys . _ * 80 * ( 2008 ) 1275 .
a. barrat , m. barthelemy , a. vespignani , dynamical processes on complex networks , cambridge university press , cambridge , 2008 .
m. e. j. newman , networks : an introduction , oxford university press , 2009 .
r. pastor - satorras , c. castellano , p. van mieghem , a. vespignani , arxiv:1408.2701 , 2014 .
a. l. barabsi , _ science _ * 325 * ( 2009 ) 412 .
r. pastor - satorras , a. vespignani , _ phys .
_ * 86 * ( 2001 ) 3200 .
z. yang , t. zhou , _ phys .
* 85 * ( 2012 ) 056106 .
f. d. sahneh , c. scoglio , p. van mieghem , _ ieee / acm transactions on networking ( ton ) _ * 21 * ( 2013 ) 1609
. m. e. j. newman , _ phys .
* 66 * ( 2002 ) 016128 .
m. barthlemy , a. barrat , r. pastor - satorras , a. vespignani , _ phys .
* 92 * ( 2004 ) 178701 .
m. barthlemy , a. barrat , r. pastor - satorras , a. vespignani , _ journal of theoretical biology _ * 235 * ( 2005 ) 275 .
b. karrer , m. e. j. newman , _ phys .
* 82 * ( 2010 ) 016101 .
e. kenah , j. m. robins , _ phys .
* 76 * ( 2007 ) 036113 .
j. miller , _ phys .
* 76 * ( 2007 ) 010101 .
a. v. goltsev , s. n. dorogovtsev , j. f. f. mendes , _ phys . rev . e _ * 78 * ( 2008 ) 051105
. a. v. goltsev , s. n. dorogovtsev , j. g. oliveira , j. f. f. mendes , _ phys .
* 109 * ( 2012 ) 128702 .
m. a. serrano , m. bogu , _ phys .
* 97 * ( 2006 ) 088701 .
j. miller , _ phys .
* 80 * ( 2009 ) 020901 .
y. gang , z. tao , w. jie , f. zhong - qian , w. bing - hong , _ chinese physics letters _ * 22 * ( 2005 ) 510 .
x. chu , z. zhang , j. guan , s. zhou , _
physica a _ * 390 * ( 2011 ) 471 .
r. pastor - satorras , a. vespignani , _ phys .
* 65 * ( 2002 ) 036104 .
p. van mieghem , _
computer communications _ * 35 * ( 2012 ) 1494 .
m. starnini , a. machens , c. cattuto , a. barrat , r. pastor - satorras , _ journal of theoretical biology _ * 337 * ( 2013 ) 89 .
c. granell , s. gmez , a. arenas , _ phys .
* 97 * ( 2013 ) 128701 .
d. w. zhao , l. h. wang , s. d. li , z. wang , l. wang , b. gao , _ plos one _ * 9 * ( 2014 ) e112018 .
s. meloni , a. arenas , y. moreno , _ pnas _ * 106 * ( 2009 ) 16897 .
h. x. yang , w. x. wang , y. c. lai , y. b. xie , b. h. wang , _ phys . rev .
* 84 * ( 2011 )
045101 . h. x. yang , z. x. wu , _ j. stat . mech . _ * 3 * ( 2014 ) p03018 .
h. x. yang , z. x. wu , b. h. wang , _ phys .
* 87 * ( 2013 ) 064801 .
g. yan , t. zhou , b. hu , z. q. fu , b. h. wang , _ phys .
* 73 * ( 2006 ) 046108 .
a. arenas , a. daz - guilera , r. guimer , _ phys .
* 86 * ( 2001 ) 3196 .
p. erds , a. rnyi , _ publ .
inst . hung .
_ * 5 * ( 1960 ) 17 .
goh , b. kahng , d. kim , _ phys .
* 87 * ( 2001 ) 278701 .
r. guimer , l. danon , a. diaz - guilera , f. giralt , a. arenas , _ phys .
* 68 * ( 2003 ) 065103(r ) .
s. w. sun , l. j. ling , n. zhang , g. j. li , r. s. chen , _ nucleic acids research _ * 31 * ( 2003 ) 2443 . l. a. adamic , n. glance , in proceedings of the www-2005 workshop on the weblogging ecosystem , 2005 .
k. wang , y. zhang , s. zhou , w. pei , s. wang , t. li , _ physica a _ * 390 * ( 2011 ) 2593 . | we propose a novel sir epidemic model which is driven by the transmission of infection packets in networks .
specifically , infected nodes generate and deliver infection packets causing the spread of the epidemic , while recovered nodes block the delivery of infection packets , and this inhibits the epidemic spreading .
the efficient routing protocol governed by a control parameter @xmath0 is used in the packet transmission .
we obtain the maximum instantaneous population of infected nodes , the maximum population of ever infected nodes , as well as the corresponding optimal @xmath0 through simulation . we find that generally more balanced load distribution leads to more intense and wide spread of an epidemic in networks .
increasing either average node degree or homogeneity of degree distribution will facilitate epidemic spreading .
when packet generation rate @xmath1 is small , increasing @xmath1 favors epidemic spreading .
however , when @xmath1 is large enough , traffic congestion appears which inhibits epidemic spreading . |
the recent discovery of a new boson at the large hadron collider ( lhc)@xcite will lead to a detailed study of its properties in the next years , and will require a careful comparison of the experimental data with the results of precise calculations for the higgs sector of the standard model ( sm ) . vector boson fusion ( vbf ) , @xmath0 , is the second largest channel for detecting the higgs boson .
it is a pure electroweak process and its study is very important for determining the couplings and for a deeper knowledge of the higgs sector@xcite .
however , at the 10 tev energy scale of lhc , even the electroweak radiative corrections become important @xcite and should be included in computer simulations together with qcd corrections @xcite . with a partonic center - of - mass energy @xmath1 of several tev - more than an order of magnitude larger than the masses @xmath2 of the gauge bosons -
the radiative corrections contain large sudakov logarithmic terms @xmath3 where @xmath4 are the weak coupling constants and the logarithms @xmath5 emerge from the two different energy scales .
these terms dominate the perturbative expansion and may even require resummation when the fixed order perturbation expansion breaks down
. however , for vbf the scattering amplitude is proportional to the vacuum expectation value ( vev ) , and standard resummation methods do not apply because the effective operator is not a gauge singlet .
that makes vbf a special interesting process to deal with .
the sudakov logarithms can be regarded as infrared logarithms since they diverge as @xmath6 , and by using an effective theory they can be converted to ultraviolet logarithms and summed by standard renormalization group techniques .
quite recently the soft - collinear effective theory ( scet ) @xcite has been shown to provide a simple way to obtain the sum of the series of leading - logs ( ll ) @xmath7 , next - to - leading - logs ( nll ) @xmath8 , next - to - next - to - leading - logs ( nnll ) @xmath9 , etc .
@xcite . in the effective theory the single terms contributing to the scattering amplitude
are multiplied by the general factor @xcite @xmath10 { \nonumber}\\ & \quad \times \exp\left\{-\int_{\mu_l}^{\mu_h } \frac { { \rm d}\mu}{\mu } \left [ a(\alpha(\mu))\log\frac{\mu^2}{\mu^2_h}+b ( \alpha(\mu ) ) \right ] \right\ } { \nonumber}\\ & \quad \times \exp f ( \alpha(\mu_h))\end{aligned}\ ] ] where @xmath11 is the low energy scale and @xmath12 is the high energy scale .
the coefficient @xmath13 is the high scale matching coefficient , @xmath14 and @xmath15 are the low scale matching coefficients , @xmath16 is the non - cusp anomalous dimension and @xmath17 is the coefficient of the cusp anomalous dimension .
the ll series is summed by the one - loop cusp anomalous dimension ; the sum of the nll series requires the two - loop cusp anomalous dimension , the one - loop non - cusp anomalous dimension , and the one - loop low scale matching coefficient @xmath15 ; the nnll series is given by the the three - loop cusp , two - loop @xmath16 and @xmath15 , one - loop @xmath14 and @xmath13 , etc .
in fact the exponentiated form of eq.([amp ] ) only requires the inclusion of electroweak corrections at low orders , while the unexponentiated form of fixed - order calculations would require electroweak corrections of any order for achieving the same accuracy . in some recent papers @xcite the method as been used for calculating the electroweak corrections to higgs production via vbf .
numerical results at nll order were obtained for the cross section , including the effect of parton distribution functions ( pdfs ) @xcite .
most of the corrections were obtained in analytical form by scet and might be easily included in the other software packages that have been developed , without the need of tedious loop calculations .
while pdfs imply a 3% error , many of the retained terms at nll order are very small , less than 1% , and could be neglected in order to speed up the numerical integration of the cross section . in this paper
the simpler ll order resummation is compared with the full nll calculation for higgs production via vbf , and the accuracy of the two approximations is discussed .
the ll calculation of the cross section is found to deviate from the nll result by less than @xmath18 up to 10 tev center - of - mass energy .
while the present study gives a useful evaluation of the accuracy in the scet resummation , it provides a very simple and fast analytical way for including the electroweak radiative corrections in software packages .
in fact at nll order the one - loop anomalous dimension turns out to be a @xmath19 matrix in the operator basis @xcite , and the running integral in eq.([amp ] ) requires a numerical computation .
moreover two - loop beta functions are required for the running of the couplings at the same order , yielding coupled equations that again must be solved by a numerical routine . on the other hand , at ll order the only one - loop term that is required is the cusp anomalous dimension , and by use of the simple uncoupled one - loop beta functions the running integral in eq.([amp ] ) is analytical , yielding a diagonal correction factor for the differential cross section . moreover
, an exact cancellation of terms yields the same correction factor that would be found for the quark scattering @xmath20 , plus higgs rescattering and wave function renormalization terms .
nll terms are still shown to be relevant for a full description of the dependence on scattering angles of the differential cross section , but such small dependence is averaged in the integrated cross section . the paper is organized as follows : in sec .
[ sec : kinematics ] the kinematics of the vbf process is described and some details on the integration of the cross section are reported ; the general calculation framework is described in sec .
[ sec : operators ] where the operator basis set is defined for the effective theory and matched onto the low scale and high scale gauge theories ; in sec .
[ sec : running ] the running of the wilson coefficients is discussed and the anomalous dimension is shown to take a simple diagonal form at ll order ; explicit analytical expressions are derived for the running integral in sec .
[ sec : llorder ] and the numerical results at ll order and nll order ar compared with other fixed order perturbative calculations in sec .
[ sec : numerical ] .
[ fig : tree ] , @xmath21 and @xmath22-channel diagrams are reported.,title="fig:",scaledwidth=48.0% ] at tree - level , the feynman diagrams for the process @xmath23 are shown in fig.1 .
the two outgoing jets have a large rapidity gap that characterizes the vbf channel , and the background is usually suppressed by cuts ( for vbf cuts see e.g. refs .
@xcite ) .
we denote by @xmath24 , @xmath25 the momenta of the incoming quarks , and by @xmath26 , @xmath27 the momenta of the outgoing quarks ( jets ) .
the momentum of the higgs boson is denoted by @xmath28 , and it is assumed to be on - shell , @xmath29 .
quark masses are neglected , @xmath30 , and all momenta are taken to be incoming , according to the conventions of ref .
the kinematic can be expressed in terms of generalized mandelstam variables : @xmath31 in scet , a set of light - cone vectors is associated to each collinear direction .
they are defined as @xmath32 , @xmath33 , with the plus ( minus ) sign for incoming ( outgoing ) particles . for the quarks
@xmath34 while for the higgs @xmath35 .
the colliding quarks ( @xmath36 ) carry a fraction @xmath37 of the total hadron momenta , @xmath38 where @xmath39 is the center - of - mass energy , and the fractions @xmath37 are integrated over in the cross section by use of pdfs that describe their distribution in the proton .
the momenta of the outgoing jets can be written as @xmath40 where @xmath41 are the energies and @xmath42 the angles with the axis of the beam , while @xmath43 is the azimuthal angle between the outgoing quarks .
the momentum of the higgs is fixed by momentum conservation . in the effective theory , in order to evaluate the cross section
, we square the matrix elements of the effective operators , sum over helicities , flavors , channels and integrate over phase space : @xmath44 where @xmath45 are the effective operators and @xmath46 are the pdf for the flavor @xmath47 at momentum fraction @xmath48 , and a factor of 1/2 must be included for identical particles in symmetric phase - space integrations .
the wilson coefficients @xmath49 are obtained in terms of tree - level high scale coefficients @xmath50 by matching at the high scale , running down to the low scale and matching at the low scale according to eq.([amp ] ) , @xmath51 which must be regarded as a matrix product in the operator basis set .
the three - body phase space @xmath52 reads@xcite @xmath53 \ , { \mathrm{d}}\varphi\ , \theta\bigg(\sum_{j=1}^4 p_j^0\bigg ) { \nonumber}\\ & \quad \times { \delta}\bigg[\bigg(\sum_{j=1}^4 p_j\bigg)^2 - m_h^2\bigg ] { \nonumber}\\ & = \frac{1}{2 ^ 6 \pi^4 } \int\ ! { \mathrm{d}}(\cos \theta_3)\ , { \mathrm{d}}(\cos \theta_4)\ , { \mathrm{d}}\varphi\ , { \mathrm{d}}e_3\ , \frac{e_3 f_1}{f_2 ^ 2 } \,.\end{aligned}\ ] ] where @xmath54 { e_\mathrm{cm}}e_3 { \nonumber}\\ f_2 & = [ x_1+x_2 + ( x_2 - x_1)\cos \theta_4 ] { e_\mathrm{cm}}-2[1-\cos \theta_3 \cos \theta_4 { \nonumber}\\ & \quad - \cos \varphi \sin \theta_3 \sin \theta_4 ] e_3\,.\end{aligned}\ ] ] the remaining integrals in eq .
must be carried out numerically with the boundary conditions @xmath55{e_\mathrm{cm } } } \,,\end{aligned}\ ] ] while we already included the bounds of the pdfs , the boundaries can be restricted by further cuts that we might impose on the phase space in the integration .
electroweak corrections can be obtained by scet in the framework of ref .
the extension to the vbf process was derived in ref .
@xcite , and the explicit expressions are reported in ref .
here we only consider the sm gauge group .
extensions like the minimal left - right symmetric gauge group@xcite will be the subject of an other paper .
the first step consists of matching onto scet at a high scale @xmath56 . here
the effects of symmetry breaking are suppressed , and the matching can be done in the unbroken gauge theory .
next we run the effective operators down to a low scale @xmath57 by renormalization group equations . at the low - scale , the @xmath58 and @xmath59 boson are integrated out : we match onto a @xmath60 effective theory , only containing gluons and photons . in this low - scale matching ,
the effects of @xmath61 symmetry breaking must be considered .
they only enter in this low - energy matching . at the high scale , the operators for vbf are given by @xcite @xmath62 in the higgs sector we introduce the operators @xmath63 @xmath64 where @xmath65 is the collinear scalar doublet @xmath66 that gives rise to a higgs boson , with its collinear wilson line @xmath67 .
the field @xmath68 is a soft scalar that gives rise to a vev when the symmetry is broken .
for the quarks we introduce the operators @xmath69 with the subscript @xmath70 that labels the momentum @xmath71 of the particle .
collinear wilson lines @xmath72 are included in all these fermion fields , according to collinear gauge invariance . in order to keep the notation as general as possible , the projectors @xmath73 have been suppressed , and both left- and right - handed quarks have been allowed .
of course the field @xmath74 is supposed to be a fermion doublet ( singlet ) if it is left - handed ( right - handed ) .
each helicity is considered separately and the contributions are combined at the end .
it is quite obvious that the operators @xmath75 and @xmath76 can only make sense if the quarks are left - handed , while the operators @xmath77 and @xmath78 are allowed if at least one of the incoming particles is left - handed . as discussed in ref.@xcite
this operator basis is not complete , because it only suffices for quarks and for the @xmath79-channel .
however for incoming anti - quarks and for the @xmath80-channel the corresponding operators are obtained by interchanging the particle labels .
thus we will only discuss the case of incoming quarks in the @xmath79-channel . at
ll and nll order the tree - level high - scale matching suffices , and can be done in the unbroken phase of the electroweak gauge theory .
the coupling of the scalar doublet to the gauge fields is described by the lagrangian terms @xmath81 in fig .
1 we only need to consider the @xmath79-channel , because the @xmath22-channel contribution is suppressed in vbf . matching onto the operators in eq . , yields @xmath82 while the other coefficients vanish at tree level . in order to keep the notation as general as possible ,
a new variable @xmath83 has been defined : @xmath84 if the particle with label @xmath47 is left - handed and @xmath85 if the particle is right - handed .
the hypercharge is @xmath86 for left - doublets , while for right - handed particles @xmath87 for up - type quarks and @xmath88 for down - type quarks .
the running of the wilson coefficients follows by the rg equation @xmath89 in terms of the anomalous dimension @xmath90 of the operators . that is a matrix equation , corresponding to the 10 operators in eq . .
using the notation of eq.([amp ] ) the anomalous dimension can be written as the sum of cusp and non - cusp terms @xmath91 we only need the one - loop cusp anomalous dimension at ll order , while two - loop cusp and one - loop non - cusp terms were required at nll order and were reported in detail in ref .
@xcite .
finally , let us consider the low energy matching .
the tree - level result suffices at ll order .
the one - loop calculation was required at nll order and was also reported in ref .
@xcite . at low energies we must take into account the effects of electroweak symmetry breaking .
we match onto a basis of operators in the broken phase of the gauge group . for the higgs part of the operator , given in eq .
, the soft scalar field simply attains a vev and the collinear scalar field produces a higgs .
thus at tree - level the low energy matching yields @xmath92 where @xmath93 is the vev and @xmath94 is the higgs field .
the quark part in eq .
gets matched onto the set @xmath95 here @xmath80 and @xmath96 denote up and down - type quarks . for operators @xmath97 , the pairs of fields @xmath98 can be left - handed or right - handed , whereas in @xmath99 and @xmath100 all fields must be left handed .
we match \{@xmath101 , @xmath102 } onto the operators @xmath103 in eq . , while we ignore @xmath104 and @xmath105 which vanish at tree level .
the matching is described by the @xmath106 matrix @xmath107 which is defined according to @xmath108 so that @xmath109 the matrix @xmath107 follows @xmath110
at ll order , one - loop corrections enter through the cusp anomalous dimension . for vbf
the one - loop terms contributing to the anomalous dimension were derived in ref .
@xcite and reported in detail in ref .
they can be written as the sum of three terms @xmath111 the first term @xmath112 is the collinear anomalous dimension , it is diagonal and contains the large logarithms that contribute to the cusp anomalous dimension , plus wave function renormalization terms .
resummation at ll order requires the inclusion of these one - loop terms , and they are easily obtained by the sum of single - particle collinear terms @xcite for all the external particles @xmath113 as reported in ref .
@xcite , with the shorthand @xmath114 the quark one - loop collinear terms @xmath115 read @xmath116 l_i - \frac{9 { \alpha}_2}{16\pi } \eta_i - \frac{3 { \alpha}_1}{4\pi } y_i^2\,,\ ] ] while the higgs one - loop collinear term @xmath117 is @xmath118 l_h - \frac{3 { \alpha}_2}{4\pi } - \frac{{\alpha}_1}{4\pi } + \frac{3y_t^2}{16\pi^2}\,,\ ] ] where we included the contribution from the top yukawa @xmath119 to the higgs wave function renormalization .
the second term @xmath120 in eq.([eq : ga_total ] ) is the soft anomalous dimension that is obtained by summing over the soft functions @xcite , and is reported in ref .
@xcite in some detail .
this soft term does not contain any large logarithm and does not contribute to the cusp anomalous dimension .
it only depends on the scattering angles of the external particles , and its effect largely cancels in the integration of the cross section . at ll order this term can be neglected .
the third term @xmath121 in eq.([eq : ga_total ] ) is a specific new contribution occurring in the vbf process @xcite , and can be written as the sum of @xmath122 and @xmath123 parts plus a term @xmath124 arising from the rescattering of the higgs boson .
the first two parts contain cusp diagonal terms proportional to @xmath125 , and smaller non - cusp off - diagonal terms that do not depend on the scale @xmath126 . at ll order
we only need to retain the diagonal cusp terms and the rescattering term that is non - cusp but is large because of the large self - coupling @xmath127 of the higgs boson .
we also include the diagonal wave function renormalization terms . according to ref .
@xcite , the cusp parts of the @xmath122 contribution read @xmath128\ ] ] for the coefficients @xmath129 , @xmath130 , @xmath131\ ] ] for the coefficient @xmath132 and @xmath133\ ] ] for the coefficient @xmath134 . the cusp part of the @xmath123 contribution is @xmath135,\ ] ] and the higgs rescattering contribution is given by @xmath136 where @xmath137 for the coefficient @xmath138 .
summing all the terms , we find that the dependence on @xmath139 cancels ( up to non - cusp terms like the difference @xmath140 that can be neglected at ll order ) , yielding the following simple result for the total anomalous dimension @xmath141 where the sum is over the four external quarks . here @xmath142 is the sum of the retained diagonal non - cusp terms , including constant wave function renormalization and higgs rescattering terms @xmath143 -\frac{3{\alpha}_1}{4\pi}\left [ \sum_i y_i^2 + \frac{1}{2}\right ] + \frac{3y_t^2}{16\pi^2}+\hat { \gamma}^\lambda\ ] ] where @xmath144 for the coefficient @xmath138 .
some of these terms are small , but they are constant , do not depend on scattering angles , and their weight might sum up in the integration of the cross section .
all other non - cusp terms have been neglected .
the simple result of eq.([eq : ga_tot ] ) says that the one - loop cusp anomalous dimension is the same that we would obtain by the sum of the collinear terms for the external quarks , neglecting the higgs particle .
however the higgs momentum would affect the kinematic of the quarks anyway .
moreover the higgs rescattering and wave function renormalization terms are not small and have been included in the non - cusp part @xmath145 of the anomalous dimension .
at ll order the running of the coupling constant can be evaluated by uncoupled one - loop beta functions .
that , together with the simple diagonal form of the one - loop cusp anomalous dimension in eq.([eq : ga_tot ] ) , allows for a fully analytical evaluation of the electroweak radiative corrections . by insertion of eq.([eq : d0 ] ) in eq.([amp ] ) , the low scale wilson coefficients follow from eq.([uc ] ) that now reads @xmath146 where the product runs over the four couplings @xmath147 , having defined @xmath148 and @xmath149 , and the exponentiated running factors follow by integration of the corresponding anomalous dimension @xmath150 here @xmath151 is the term proportional to the coupling @xmath152 in the anomalous dimension for the wilson coefficient @xmath153 . by inspection of eq.([eq : ga_tot ] ) , we see that such terms can be written as functions of @xmath154 @xmath155\ ] ] where for each coupling @xmath152 there is a different coefficient @xmath156 , while @xmath157 is a function of the external momenta @xmath158 that in general also depends on the chosen coefficient @xmath153 .
the explicit expressions of the functions are @xmath159 \ , , { \nonumber}\\ { \cal f}^{(2)}_x(p)=&\frac{3}{4\pi}\sum_i \eta_i \log({\bar{n}}_i \cdot p_i ) - \frac{9}{16\pi}\left [ \sum_i \eta_i + d^{{\mathcal{o}}}\right ] \ , , { \nonumber}\\ { \cal f}^{(y)}_x(p)=&\frac{3}{4\pi } \ , , { \nonumber}\\ { \cal f}^{(\lambda)}_x(p)=&\frac{c^{{\mathcal{o}}}}{\pi } \ , , \end{aligned}\ ] ] while the coefficients are @xmath160 in all these definitions the sums are over the external quarks .
the couplings also have an implicit dependence on @xmath79 , dictated by the one - loop beta - functions @xmath161 these can be easily integrated , yielding @xmath162 where the one - loop coefficients are well known@xcite @xmath163 by a simple integration , the running factors in eq.([eq : a ] ) take the explicit general form @xmath164 } } \,\end{aligned}\ ] ] and their product in eq.([eq : wc ] ) gives an analytical expression for the electroweak radiative correction at ll order .
the cross section follows by insertion of the low scale wilson coefficients from eq.([eq : wc ] ) in the phase space integral of eq.([eq : si_ew ] ) and integrating by pdfs .
the closed analytical form of eq.([eq : aint ] ) greatly speeds up the numerical integration of the cross section at ll order compared with nll order .
we discuss the accuracy of the two orders in the next section by a direct comparison .
in this section the accuracy of the approximation is tested by a comparison between ll and nll orders .
all the details of the numerical integration are kept exactly the same as reported in ref .
@xcite where the cross section was evaluated at nll order . we summarize them briefly .
the low energy matching scale is chosen to be @xmath165 .
the couplings and parameters of the standard model have been set at the electroweak scale @xmath165 according to the data of ref.@xcite .
the higgs mass is assumed to be @xmath166 gev .
the high scale @xmath167 is set at the larger value between @xmath168 and the geometric average @xmath169 . at this scale the sum of logarithmic terms @xmath170 reaches its minimum in the one - loop matching . as discussed in ref .
@xcite this choice has the merit of stopping the running whenever one of the mandelstam variables is too small , while keeping the neglected one - loop matching terms as small as possible .
on the other hand the sensitivity to the choice of the high scale was shown to be small and comparable to the sensitivity of the standard tree - level cross section .
should make the omitted terms small , as for the method of minimal variance@xcite .
] we use cteq6 pdfs@xcite and neglect the very small contribution of t and b quarks .
moreover we neglect the s - channel contribution and interference terms that are known to be small with vbf cuts .
the masses of the vector gauge bosons are restored in the denominators of the coefficients in eq .
, as their effects become important when the total cross section is evaluated . hereafter , in order to compare with the nll calculation of ref .
@xcite , we adopt the same cuts on angles and transverse moment : @xmath171 , @xmath172 , @xmath173 gev .
no qcd corrections have been included in both calculations , and only the virtual electroweak corrections are considered by the method .
the terms contributing to the cross section at tree - level are reported in fig.2 .
the @xmath174 process dominates , followed by the other left - left terms .
the left - right contributions are two order of magnitude smaller , while the right - right terms are very small and have been neglected . in fig.2 a sum over the generations of quarks is included through the pdfs . as a first comparison we study the single terms in the phase space , and for each of them we calculate the k - factor which is defined as the ratio between the cross section with electroweak corrections on , and the tree - level cross section without any radiative correction @xmath175 where the differential cross sections are evaluated for fixed values of @xmath176,@xmath177 , @xmath178 and @xmath39 , and for a given set of angles . here
the simple ll calculation is compared with the nll result of ref .
@xcite . in figs . 3 - 10 the k - factor
is reported as a function of @xmath179 with @xmath180 and @xmath181 at a typical set of parameters : @xmath182 tev , @xmath183 and @xmath184 .
the agreement of ll and nll results is very good at small angles . as expected , at ll order the dependence on angles
is reduced in comparison with the nll result , because the neglected terms have a larger dependence on the scattering angles .
however for @xmath185 , in the physically relevant phase - space region , the difference is small and the ll result ( solid line ) interpolate the nll data ( squares ) quite well .
the dependence on the azimuthal angle @xmath186 is very small and negligible for @xmath185 , in perfect agreement with the nll order .
we find a better agreement for the left - right processes ( fig .
7 - 10 ) where the radiative corrections are larger , than for left - left processes ( fig . 3 - 6 ) where the corrections are smaller .
thus the very large suppression of the left - right processes is mainly due to the role played by the collinear anomalous dimension .
it is remarkable that such large suppression has no relevant effects on the total cross section which is dominated by the left - left processes .
the single terms are integrated by pdfs over the phase space and summed up , yielding the cross section @xmath187 . by comparison with the tree level cross section
, the relative electroweak correction is defined as the ratio @xmath188 and is displayed in fig.11 as a function of the center - of - mass energy . for comparison the nll result of ref .
@xcite is also reported in fig.11 together with the output of the code hawk@xcite that is based on a fixed order perturbative calculation . while differences are negligible below 2 tev , at large energies ll and nll corrections grow faster than predicted by fixed order one - loop calculations , reaching 9% at the full lhc energy @xmath182 tev , to be compared with 5% predicted by hawk . at ll order
the correction is a bit smaller with respect to nll order , but the difference is less than 1% up to @xmath189 tev .
thus the simple ll calculation provides an analytical electroweak correction that seems to be more reliable than fixed order calculations at high energies .
while fixed - order perturbative calculations might miss part of the correction at the lhc energy scale , the simple ll resummation contains the main terms , and gives integrated corrections that are more accurate than pdfs .
insertion of the simple analytical result of eqs.([eq : wc]),([eq : aint ] ) in software packages would be straightforward , and would increase the accuracy of simulations . | using soft - collinear effective theory , the leading - log radiative electroweak corrections are written in a closed and analytical form for the hadronic cross section of higgs production through vector boson fusion , @xmath0 , one of the most promising channels for studying the higgs boson at the lhc . the simple leading - log resummation is compared with a full next - to - leading - log calculation , and its accuracy is found to be of order 1% up to 10 tev , i.e. better than the accuracy of pdfs .
corrections are found to be larger than predicted by one - loop fixed order approximations at lhc energies .
the method provides a simple way of incorporating the electroweak corrections in software packages , improving the accuracy of simulations . |
a major goal in the study of galaxy evolution is understanding the interplay between active galactic nuclei ( agn ) and star formation in their host galaxies .
a correlation exists between the mass of central black holes , which power agn , and stellar bulge mass , which suggests that the growth of black holes by accretion is linked to the growth of stellar mass by star formation ( e.g. , * ? ? ?
gas accretion onto a central black hole can spur agn activity , including winds and high energy radiation , which may disrupt the interstellar medium ( ism ) of the host galaxy , quenching star formation ( e.g. , * ? ? ?
* ) . on the other hand , agn activity , through jets and winds for example
, can contribute an additional pressure to the ism of the host galaxy , leading to cloud collapse , and thereby bolstering star formation ( both theoretically and observationally , e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
here we investigate the relationship between star formation and agn activity in the bright radio galaxy 3c368 at redshift @xmath3 @xmath4 1.131 ( as determined by the [ oii ] 372.8 nm line , * ? ? ? * ) .
we observe 3c368 as it was in the epoch of both peak star formation and agn activity in the universe .
3c368 was discovered as part of the third cambridge radio catalogue @xcite . since then
, it has become one of the best studied , high - redshift , radio - loud , fanaroff - riley class ii ( fr - ii ) type galaxies , with observations spanning the electromagnetic spectrum from the radio ( e.g. , * ? ? ?
* ) to the x - ray ( e.g. , * ? ? ?
* ) . not only does it exhibit considerable agn activity , with radio emission extended over 73kpc @xcite , 3c368 also has a substantial stellar mass ( @xmath2 3.6@xmath510@xmath6@xmath7 , * ? ? ?
* ) , and is undergoing a period of vigorous ( 350@xmath8yr@xmath9 , calculated by spectral energy distribution ( sed ) modeling in the far - infrared , * ? ? ?
* ) and extended ( over @xmath2 2@xmath104kpc , * ? ? ?
* ) star formation .
@xcite suggest a link between agn activity and star formation in this source , in that the jets emanating from the agn seem to be spurring star formation . to date , however , no co emission , which traces the molecular gas necessary for star formation , has been detected in this source ( e.g. , * ? ? ?
* ) . to further explore the properties of the stellar population in 3c368 , including its age and spatial distribution
, we observed the [ oi ] 63@xmath0 line , the [ oiii ] 52 and 88@xmath0 lines , and the [ oiv ] 26@xmath0 line using the _ pacs _ spectrometer on the _ herschel space observatory _ , the [ nii ] 205@xmath0 line with the atacama large millimeter / submillimeter array ( alma ) , and the co(21 ) line using both alma and the plateau de bure interferometer ( pdbi ) .
the [ oi ] line arises from dense photodissociated gas on the surfaces of molecular clouds , and combined with our prior detection of the [ cii ] 158@xmath0 line with zeus on the caltech submillimeter observatory ( cso ) @xcite , traces the strength of the uv radiation field .
the [ oiii ] 52 and 88@xmath0 lines taken together allow us to estimate the density within h ii regions , and the [ oiv ] 26@xmath0 line , with an ionization potential of 54.93ev , traces the narrow line region ( nlr ) of the agn .
taken together with spectroscopic observations conducted with the _ infrared spectrograph ( irs ) _ on board the _ spitzer space telescope _ , including [ neii ] and [ neiii ] lines in the mid - infrared , these new data allow us to estimate the age of the starburst in 3c368 , furthering the theory of agn - driven star formation in this powerful , yet enigmatic , galaxy .
we assume a flat @xmath11cdm cosmology , with @xmath12 = 0.27 , @xmath13 = 0.73 , and h@xmath14 = 71kms@xmath9mpc@xmath9 , throughout this paper @xcite .
the oxygen fine - structure lines were observed in 3c368 using the _ photodetector array camera and spectrometer _ ( _ pacs _ ) @xcite on board the _ herschel space observatory _ @xcite .
the [ oi ] 63@xmath0 and [ oiii ] 88@xmath0 lines were observed in line scan chop - nod mode ( obs .
i d : 1342243546 ) , with a total observing time of @xmath2124minutes , before the [ oiii ] 52@xmath0 and [ oiv ] 26@xmath0 lines were observed simultaneously in range scan mode ( obs .
i d : 1342243547 ) , with an observing time of @xmath238minutes .
all observations took place on march 25 , 2012 .
these observations were reduced using the herschel interactive processing environment ( hipe , version 13 ) @xcite . for each of the oxygen lines , with the exception of the [ oi ] 63@xmath0 line ,
the central 9@xmath154@xmath59@xmath154 spatial pixel ( spaxel ) was used to generate the spectrum , and the results were point source corrected .
for the [ oi ] 63@xmath0 line , the inner 3@xmath53 spaxels of the integral field spectrometer were combined , and point source corrected , to create one spectrum .
this step was taken in order to correct what appears to be pointing jitter with this observation , as indicated by the presence of line flux on multiple spaxels .
significant line flux was not seen to extend over multiple spaxels for any of the other lines observed by _
pacs_. these line fluxes are reported in table 1 , and the line spectra are shown in figure 1 . as a follow - up to our detection of the [ cii ] line
, we attempted to detect the co(21 ) rotational line with the plateau de bure interferometer ( pdbi ) in 3c368 .
the observations were carried out on may 31 and june 1 , 2010 , with five antennas in the d configuration , producing a synthesized beam of size 4@xmath1569@xmath54@xmath1526 ( fwhm ) .
total on source integration was 5.25hours , with the band-1 receivers tuned to the redshifted frequency of the line , 108.183ghz .
we used the widex correlator , which provides an instantaneous 3.6 ghz coverage in both polarizations .
for these observations , we used mwc349 for absolute flux calibration .
the data were calibrated and imaged using gildas , resulting in a 1@xmath16 sensitivity of 279@xmath17jybeam@xmath9 for these observations , calculated over an assumed line width of 500kms@xmath9 ; to match the width of the [ cii ] line .
the line was not detected , with a 3@xmath16 upper limit , assuming the source is small with respect to the ( 39@xmath535kpc ) beam , of 0.42jykms@xmath9 , or equivalently 1.5@xmath510@xmath18wm@xmath19 .
we detect the 1.3 mm rest - frame continuum in both the northern and southern lobes of 3c368 .
we report these values using our subsequent , more sensitive , alma observations .
having made no detection with the pdbi , we next attempted to detect the co(21 ) line in 3c368 with the atacama large millimeter / submillimeter array ( alma ) .
these band 3 observations were performed on january 31 , 2015 , with a synthesized beam size of 3@xmath1512@xmath51@xmath1581 ( fwhm ) .
our on - source integration time was @xmath237minutes . for these observations , which were conducted with a precipitable water vapor ( pwv ) measurement of [email protected] mm , we used titan as a flux calibrator , at a distance of 41.7@xmath21 from the source , and j1751 + 0939 as both a phase and bandpass calibrator , located 3.6@xmath21 from the source .
characteristic absolute flux calibrations with alma are accurate to @xmath2 10% ( alma technical handbook ) .
the data were reduced , imaged , and cleaned with the common astronomy software application ( casa ) , version 4.3.1 .
the 1@xmath16 sensitivity for these observations is 134@xmath17jybeam@xmath9 , over an assumed line width of 500kms@xmath9 .
again the line is not detected .
a 3@xmath16 detection limit for a source enclosed by the [email protected] beam yields an upper limit on the line intensity of 0.201jykms@xmath9 , or equivalently 7.3@xmath510@xmath22wm@xmath19 .
this time , we detect the continuum in three components of 3c368 : the northern lobe at 673@xmath2039@xmath17jy , the southern lobe at 712@xmath2045@xmath17jy , and the core at 123@xmath2034@xmath17jy ( see figure 2 ) .
in addition to the co(21 ) line , we attempted to detect the [ nii ] 205@xmath0 line with alma .
these band 9 observations were carried out on february 21 and april 13 , 2014 , with a synthesized beam of size 0@xmath1541@xmath50@xmath1529 ( fwhm ) , and an on - source integration time of @xmath254minutes . for these observations , which were conducted with a pwv of [email protected] mm , we used titan as a flux calibrator , at a distance of 47.9@xmath21 from the source , j1751 + 0939 as a phase calibrator , located 3.6@xmath21 from the source , and j1924 - 2914 as a bandpass calibrator , 44.77@xmath21 from the source .
characteristic absolute flux calibrations for observations performed in cycle 1 with the alma band 9 receivers are accurate to @xmath2 15% ( alma technical handbook ) .
the data were reduced , imaged , and cleaned with casa , version 4.2.1 .
we obtain a 1@xmath16 sensitivity of 891@xmath17jybeam@xmath9 , over an assumed line width of 500kms@xmath9 .
presuming that the line source is extended over @xmath2 2 beams ( as the continuum contours in figure 2 suggest ) , the 3@xmath16 upper limit is 3.78jykms@xmath9 or 8.65@xmath5@xmath23wm@xmath19 .
we detect the rest - frame 205@xmath0 continuum , which traces thermal dust emission , from only the core component of 3c368 , with an integrated intensity of [email protected] ( see figure 2 ) .
the [ neii ] 12.81@xmath0 , [ neiii ] 15.56@xmath0 , and [ nev ] 14.32@xmath0 lines were obtained from the cornell atlas of _ spitzer / irs _ sources ( cassis , version 6 ) @xcite , and were observed as part of a program to study multiple radio galaxies and quasars near the peak of cosmic star formation using the _ spitzer space observatory _ ( pi : martin haas , aorkey=22912000 ) .
the observations of 3c368 were carried out on may 1 , 2008 , with the _ infrared spectrograph ( irs ) _ @xcite in low - resolution , long - slit mode , ( ll1 ) , covering a wavelength range of 19.5@xmath1038@xmath0 .
the total on - source integration time for these observations was @xmath248minutes .
the data were reduced and extracted using the cassis software , which employs an optimal " extraction technique in order to produce a spectrum with the highest possible signal to noise ratio @xcite .
the line fluxes from the three neon lines are reported in table 1 , and the spectrum is plotted in figure 3 .
in order to determine the line fluxes for the fine - structure lines observed with _
herschel / pacs _ , we sum the flux from the line channels ( shaded in figure 1 ) , and propagate the statistical uncertainties per channel to calculate the signal to noise ratios .
the channels associated with the spectral lines are determined by a gaussian fit to the line profiles ( see table 1 ) , with two exceptions . first , the central line velocity and full - width - half - maximum ( fwhm ) obtained by a gaussian fit to the [ oi ] 63@xmath0 line , our most significant _ herschel / pacs _ detection , are used to determine the line channels for the [ oiii ] 88@xmath0 line .
this is necessary due to the low signal to noise in the [ oiii ] 88@xmath0 line spectrum .
second , a gaussian fit to the [ oiii ] 52@xmath0 line shows that it is considerably broader ( @xmath24v @xmath2 760kms@xmath9 ) than any of the other lines which we observed in 3c368 ( green dashed line , figure 1 ) .
the reported flux for this line ( see table 1 ) is calculated by summing the flux from all channels under this best - fit gaussian ( shaded both yellow and cyan in figure 1 ) .
given that the signal to noise per channel in the [ oiii ] 52@xmath0 line spectrum is @xmath252 , and that the fitted velocity width is so much larger than that of the [ oi ] 63@xmath0 line ( our most significant oxygen line detection ) , we additionally calculate the flux by summing over only the channels coincident with our [ oi ] 63@xmath0 line ( red dashed line and yellow shaded channels in figure 1 ) .
it is this flux value , 6.8@xmath510@xmath26wm@xmath19 , which we use in all calculations involving line ratios , since the flux excluded from this value originates within a different velocity component of 3c368 . with the exception of the [ oiii ] 52@xmath0 line , the central line velocities , @xmath2 + 200@xmath10300kms@xmath9 ( calculated using @xmath3 = 1.131 ) , and velocity widths , @xmath2 200@xmath10500kms@xmath9 , calculated for the fine structure lines observed with _
herschel / pacs _ are very similar .
the flux values , along with central line velocities and fwhms , for the higher signal to noise [ neii ] 12.81@xmath0 and [ neiii ] 15.56@xmath0 lines observed with _
spitzer / irs _ were obtained from the infrared database of extragalactic observables from spitzer ( ideos ) ( * ? ? ?
* spoon et al . in prep . ) .
the lower signal to noise [ nev ] 14.32@xmath0 line flux was obtained by summing the line channels , exactly as done for the _ herschel / pacs _ spectra .
the central line velocities , determined by the gaussian fits to the [ neii ] 12.81@xmath0 and [ neiii ] 15.56@xmath0 lines , overlap with those calculated for the oxygen lines observed with _
herschel / pacs _ at the < 2@xmath16 uncertainty level , indicating that these neon lines likely originate within the same regions of 3c368 as do the oxygen lines .
( contours ) , rest - frame continua in 3c368 .
the maps are primary - beam corrected .
the contour levels are @xmath203 , @xmath204 , @xmath205 , and @xmath206@xmath16 ( negative contours are dashed , 1@xmath16 @xmath4 0.4mjybeam@xmath9 ) .
the beams are shown at the bottom - left corner of the image , with the large ( red ) and small ( black ) ellipses for the 1.3 mm and 205@xmath0 maps , respectively .
the inset shows the 205@xmath0 rest - frame continuum from the core component of 3c368.,scaledwidth=50.0% ] , and density , n , allowed by the [ cii]+[oi]/fir ( [ oi]/[cii ] ) ratio , and is bounded by the uncertainty in the ratio ( dashed lines ) .
the three filled contours represent @xmath27 fitting values of 0.25 , 0.5 , and 1.0 .
the fit favors pdrs with @xmath28 @xmath2 280 and n @xmath2 7,500@xmath29 .
the low @xmath28 solution is disregarded because it would result in an emitting region with an unphysically large size.,scaledwidth=50.0% ] lccccccccccc critical density ( @xmath29 ) & 7@xmath510@xmath30 & 3@xmath510@xmath31 & 3@xmath510@xmath30 & 1@xmath510@xmath32 & 1@xmath510@xmath32 & 3.6@xmath510@xmath33 & 4.7@xmath510@xmath30 & 510 & 2.8@xmath510@xmath33 & 48 & 1.1@xmath510@xmath32 + flux ( @xmath34wm@xmath19 ) & [email protected] & 1.5 & [email protected] & 8.5 & 5.2 & 13.4 & 6.4 & 3.0 & 5.1 & < 0.0865 & < 0.00073 + significance ( @xmath16 ) & - & 2.9 & - & 3.8 & 2.9 & 5.5 & 5.0 & 2.4 & 6.4 & - & - + line center ( km s@xmath9 ) & -40@xmath20210 & - & -40@xmath20100 & 290@xmath20130 & 220@xmath2060 & 420@xmath20120 & 240@xmath2040 & 240 & -140 & - & - + line width ( km s@xmath9 ) & 3100@xmath20700 & - & 3000@xmath20200 & 510@xmath20330 & 190@xmath20120 & 760@xmath20290 & 320@xmath2090 & 320 & - & - & - the mid - ir fine - structure lines in the _
spitzer / irs _ spectral regime that arise from the three ionization states of neon are excellent probes of the hardness of the ambient uv radiation fields in galaxies .
it takes 21.56ev photons to form ne@xmath35 , 40.96ev photons to form ne@xmath36 , and 97.11ev photons to form ne@xmath37 .
therefore , the 12.81@xmath0 [ neii ] line arises from h ii regions formed by o / b stars , the 15.56@xmath0 [ neiii ] line arises from h ii regions formed by o stars , and the 14.32@xmath0 [ nev ] line requires the very hard uv radiation fields found near agn or planetary nebula created by hot white dwarfs .
this series of lines is quite useful since extinction corrections between these lines are modest , and the lines have similar critical densities ( table 1 ) .
furthermore , since neon is a noble gas , its gas phase abundance is not dependent on local conditions , since it is neither depleted onto grains nor incorporated into molecules .
[ nev ] line emission is known to be associated with planetary nebulae exposed to the hard uv radiation of very young white dwarfs ( cf . * ? ? ?
* ) , but on galactic scales , detectable [ nev ] is predominantly associated with the nlr of agn . however , detectable [ neiii ] emission can arise from both the nlr of agn and from h ii regions formed by hot o stars . @xcite
studied the neon fine - structure line emission from a variety of galaxies , including 77 3c radio sources , and found that the amount of [ neiii ] line emission from the nlr of agn is directly proportional to the [ nev ] line emission .
therefore , a simple scaling law , based on the observed [ nev ] line emission from a galaxy , can be applied to estimate the fraction of the observed [ neiii ] line emission that arises from any central agn .
we use this scaling law tocorrect " the [ neiii ] line emission for the agn contribution before applying any h ii region models to the star forming regions in 3c368 . applying the relation of @xcite , @xmath38 to the observed luminosity of the [ nev ] 14.32@xmath0 line
, we estimate that 20% of the observed [ neiii ] emission originates from the agn , leaving 80% from star forming regions .
agn - corrected " 80% is the flux which we use in our star - formation - driven h ii region models .
we note that 3c368 is comparable to the higher luminosity sources used in the determination of this [ neiii]/[nev ] relation ( * ? ? ?
* figure 1 , right panel ) , where there are fewer sources and larger scatter in the trend , and so we also consider several limiting cases in the following analysis
. given the large number of fine - structure lines observed in 3c368 , we can determine several properties of the h ii regions within this source .
the ground - state - term level populations within the o@xmath36 ion are density sensitive , so that the [ oiii ] 52@xmath0/[oiii ] 88@xmath0 line ratio yields the ionized gas density .
we find a line ratio of 2.3 , which indicates h ii region gas densities of n@xmath39 @xmath2 1000@xmath29 ( we here use the collision strengths from * ? ? ?
* ) . then , using the models of @xcite , we can combine the [ oiii ] , [ neii ] , and
agn - corrected " [ neiii ] fine - structure lines to arrive at a consistent model for the h ii regions in 3c368 .
we find that the fit favors regions of star formation heated by stars with effective temperature @xmath2 37,000 k and with gas densities @xmath2 1000@xmath29 .
the density is consistent with our estimates from the [ oiii ] 52/88@xmath0 line ratio , and the ionization state is consistent with the neon line ratios , provided that neon is overabundant to oxygen in 3c368 by a factor of 3 compared with either the n " models , which have orion nebular abundances ( o / h = 4.0@xmath510@xmath40 and n / h = 4.5@xmath510@xmath41 ) , or the d " models , with all metals depleted by a factor of @xmath42 from the n models .
this factor of three overabundance is consistent with the recent results of @xcite , who , using _
spectroscopy together with their ionization models , find higher neon to sulfur ratios ( @xmath212 ) , for a wide variety of galaxies , than previously expected .
this suggests that the cannonical neon to sulfur ratio ( @xmath23.7 ) , which was used in the h ii region models employed here , is too low by a factor of three or so . since neon , sulfur , and oxygen
are all primary elements , their abundances should scale together , and so our neon overabundance when compared to oxygen is consistent with this newly measured ne / s ratio . given that the agn contribution to the [ neiii ] line is only @xmath2 20% , and considering the uncertainty in both the line fluxes and the scatter in the [ neiii]/[nev ] correlation ( 0.2 dex in log space ) , we consider both the limiting case of no agn contribution to the [ neiii ] line , and the case of the maximum allowed agn contribution to the [ neiii ] line ( @xmath2 40% ) . in all cases ,
we find that the models favor regions of star formation heated by stars with effective temperature @xmath2 37,000 k and with gas densities @xmath2 1000@xmath29 . to retain the necessary effective stellar temperatures indicated by our modeling , at the observed far - ir luminosity in 3c368 ( 2.0@xmath5@xmath43 @xmath44 , * ?
* ) , we require a starburst with the equivalent of @xmath2 1.2@xmath5@xmath45 o8 v stars @xcite .
such a starburst would have an age of @xmath2 6.5 myr @xcite , as constrained by the stellar lifetimes of the most massive stars .
using the 73kpc radio source size of 3c368 from @xcite , and a range of jet propagation speeds from 0.03 @xmath10 0.3c @xcite , we estimate the duration of the latest episode of agn activity to be between 0.4 @xmath10 4myr , comparable in time to the age of the starburst .
this concurrence lends further support to the possibility that we may be witnessing agn - driven star formation in 3c368 , as has been suggested previously from the alignment of the radio and optical axes , which trace agn jets and stars , respectively , seen in many of the 3cr galaxies , including 3c368 @xcite . in addition to the optical continuum
, @xcite find the 2.2 @xmath0 continuum , also associated with stellar emission , to be aligned with both of the other axes .
it has also been previously suggested that 3c368 may be undergoing a major merger @xcite , providing another possible mechanism for triggering both black hole accretion and star formation simultaneously .
while we can not rule out this possibility , the [ cii]/@xmath46 ratio of 0.5% in 3c368 ( see sections 3.5 & 3.6 ) is more than an order of magnitude larger than the typical values found in the local ultraluminous infrared galaxies ( ulirgs ) , which are predominantly powered by major - merger - driven starbursts ( e.g. , * ? ? ?
the physical parameters derived from our h ii region modeling also explain our non - detection of the [ nii ] 205@xmath0 line . at such large effective temperatures , @xcite
calculate that @xmath2 86@xmath47 of the nitrogen contained within the h ii regions of 3c368 would be in the n@xmath36 state .
unfortunately , the [ niii ] 57@xmath0 line is redshifted into an absorption band in the atmosphere for 3c368 with @xmath3 = 1.131 ( @xmath2 122@xmath0 ) , making observations impossible from the ground , and even beyond the reach of airborne facilities .
these observations will have to wait for the next generation of far - ir space telescopes , such as the space infrared telescope for cosmology and astrophysics ( spica ) and the origins space telescope ( ost ) .
following @xcite , we can estimate the minimum mass of oxygen in the o@xmath36 state in 3c368 , from the luminosity in the oxygen fine - structure lines : @xmath48 where @xmath49 is the flux in the fine - structure line between the upper ( @xmath50 ) state and the lower ( @xmath51 ) state , @xmath52 is the luminosity distance ( 7.735gpc ) , @xmath53 is the mass of an oxygen atom , @xmath54 and @xmath55 are the statistical weights of the upper and lower states , respectively , @xmath56 is the partition function ( the sum of the statistical weights of all relevant states available to the o@xmath36 ion at t @xmath4 8,000 k ) , @xmath57 is the einstein coefficient for the relevant transition ( 2.6@xmath510@xmath41s@xmath9 for the [ oiii ] 88@xmath0 line , * ? ? ? * ) , and @xmath58 the frequency of that transition . using the [ oiii ] 88@xmath0 line ,
we obtain a minimum @xmath59 mass of 4.4@xmath510@xmath60 @xmath7 . similarly
, we can use our 3@xmath16 upper limit for the [ nii ] 205@xmath0 line flux ( @xmath57 @xmath4 2.1@xmath510@xmath61s@xmath9 , * ? ? ?
* ) to put a bound on the mass of n@xmath35 in 3c368 : @xmath62 we obtain a limit of < 3.2@xmath510@xmath60 @xmath7 .
we use the models of @xcite to calculate the total mass in oxygen and nitrogen by scaling from the fraction in the ionization state that we observe back to the total abundance of the element in question . with @xmath2 84@xmath47 of the oxygen in the o@xmath36 state , and @xmath2 14@xmath47 of the nitrogen in the n@xmath63 state , within the h
ii regions of 3c368 , we calculate masses of 5.3@xmath510@xmath60 @xmath7 and < 2.3@xmath510@xmath64 @xmath7 , for oxygen and nitrogen respectively . using these values , we find an upper limit for the n / o ratio in 3c368 of 5.0 .
@xcite report a solar n / o abundance ratio of 0.126 .
while the result which we obtain is consistent with the solar n / o ratio , it also allows for enhanced nitrogen relative to oxygen .
we are continuing our campaign to detect the [ nii ] 205@xmath0 line in 3c368 , and were recently granted alma time to push our sensitivity deeper in this source .
these new observations will effectively double our on - source integration time , improving our sensitivity .
in addition to the lines associated with h ii regions , we have also detected the [ oi ] 63@xmath0 line associated with photodissociation regions ( pdrs ) .
[ oi ] and [ cii ] are the two dominant coolants of pdrs , so that the sum of the line fluxes divided by the far - ir flux ( which tracks the impinging fuv ( 6 @xmath10 13.6ev ) flux ) tracks the efficiency of photoelectric heating . parameterizing the fuv in terms of the local ( habing ) interstellar radiation field ,
the photoelectric heating is sensitive to the ratio of fuv field strength to gas density , g@xmath65/n . with different critical densities
the line ratio constrains the pdr gas density , so that the line intensities and their ratio constrain g@xmath65 and n independently .
we use our prior detection of [ cii ] @xcite , together with the pdr toolbox @xcite , for our analysis .
this software fits input line flux ratios to a bank of line ratios calculated using a radiative transfer code that assumes a plane parallel geometry for the gas . before using this software ,
we correct for the different optical depths of the lines used in modeling the pdrs .
the [ oi ] 63@xmath0 line is likely to be optically thick for pdrs with a@xmath66 of the order of a few ( e.g. , * ? ? ?
* ; * ? ? ?
we , therefore , multiply the [ oi ] 63@xmath0 line flux by a factor of two , in order to account for its opacity ( as in * ? ? ?
* ) , taking the uncertainty in this correction to be equal to the correction which we applied . due to the lesser abundance of c@xmath35 , and the greater populations in its excited levels
, the [ cii ] line is expected to have significantly smaller optical depth than the [ oi ] line .
observations of the hyperfine structure lines of @xmath67c@xmath35 indicate optical depths less than or of the order of 0.5 to 1.5 for galactic pdrs ( cf . * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
pdr models indicate optical depths of the same order @xcite .
therefore , we make no correction to the [ cii ] 158@xmath0 line for opacity effects .
we perform an additional correction to remove any contributions to the [ cii ] 158@xmath0 line emission originating from h ii regions , before performing our pdr analysis . with an ionizing energy of only 11.26ev ,
ionized carbon can be found both in the neutral hydrogen phases of the ism ( 13.6ev ionizing energy ) , and also in the ionized gas phase where species like ionized nitrogen ( 14.5ev ) exclusively reside @xcite .
since the [ nii ] 205@xmath0 and [ cii ] 158@xmath0 lines have very similar critical densities in ionized gas regions , 48@xmath29 and 50@xmath29 , respectively , their ratio can be used to calculate the percentage of c@xmath35 coming from h ii regions @xcite .
we calculate a flux ratio of [ cii ] 158@xmath0 / [ nii ] 205@xmath0 > 60 .
@xcite find that if all of the c@xmath35 observed were to come from h ii regions , this ratio would vary from 3.1 for a low density gas to 4.3 for a high density gas , so that < 7% of the [ cii ] 158@xmath0 line flux in 3c368 comes from within h ii regions .
since this contribution is very small and within the uncertainties , we make no correction to the [ cii ] 158@xmath0 line flux for our pdr analysis . finally , we correct the observed fir luminosity to include only the component produced by the uv flux from young stars which has been reprocessed by the dust into the fir .
we adopt the fit from @xcite , who model the sed of 3c368 with three components : one for the agn - fueled warm dust component ( using a library of torus models from * ? ? ?
* ) , one black - body for the optical / nir stellar emission , and one gray - body in the fir / sub - mm with the dust emissivity index ( @xmath68 ) as a free parameter .
we use the luminosity in this last component for our pdr analysis . @xcite
also use this fir luminosity to calculate a star formation rate in 3c368 , obtaining a value of 350@xmath8yr@xmath9 . inserting these three observations ( see table 2 ) ,
the corrected [ cii ] 158@xmath0 and [ oi ] 63@xmath0 line intensities , and the fitted fir luminosity , into the pdr toolbox software @xcite , we obtain best fits for pdrs with @xmath28 @xmath2 280 and n @xmath2 7,500@xmath29 ( see figure 4 ) . using the scaling relations found in @xcite , and adopting an fir luminosity of 2.0@xmath5@xmath43 @xmath44 @xcite , we calculate that these pdrs are extended over 1.9 @xmath10 4.8kpc , consistent with our casa 2-d gaussian fit to the source from the rest - frame 205@xmath0 continuum map ( @[email protected] , see figure 2 ) .
one potential caveat for our pdr modeling results is that the observed [ cii ] and [ oi ] lines may originate within the x - ray dominated region ( xdr ) surrounding the agn of 3c368 ( e.g. , * ? ? ?
* ) , with the [ oiii ] lines coming from the closer - in nlr ( e.g. , * ? ? ?
given that we can form a consistent picture of star formation in 3c368 using our observations , and that optical fine - structure lines have also been observed to be extended over several arc seconds around the central source ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , the theory of large - scale star formation in 3c368 seems more plausible .
this theory can be tested by spatially resolving the [ cii ] line emission in 3c368 .
extended [ cii ] emission may confirm the existence of large - scale star forming regions , while confined emission may suggest an agn origin .
we have been awarded alma time to carry out these very observations ( see future observations section ) .
lccccccc uncorrected flux ( @xmath34wm@xmath19 ) & 6.4 & 5.1 & 1075.0 + corrected flux ( @xmath34wm@xmath19 ) & 12.8 & 5.1 & 1075.0 + uncertainty ( @xmath34wm@xmath19 ) & 6.4 & 1.5 & 107.5 while the fine - structure lines from pdrs allow us to model these regions in 3c368 , the complete picture of star formation can only come from including molecular gas observations as well .
we have attempted to observe the co(21 ) transition with both alma and pdbi , and like others in the past ( e.g. , * ? ? ?
* ) , we have not detected any emission .
our alma observations probe down to a sensitivity of 134@xmath17jybeam@xmath9 , over an assumed line width of 500kms@xmath9 . assuming a ratio of [ cii ] to co(10 ) of @xmath2 4,100 , which is typical of normal metallicity star forming regions in the milky way and starburst galaxies @xcite , and a co(21 ) to co(10 ) flux ratio of 7.2 ( 90@xmath47 of the high - temperature thermalized value )
, we would expect a [ cii ] to co(21 ) flux ratio of @xmath2 570 in 3c368 .
the 3@xmath16 co upper limit from our alma observations , however , yields a limit of [ cii]/co(21 ) > 7,000 , which is more than 12@xmath5 the expected value .
following @xcite , we can estimate the total molecular gas mass in 3c368 using the 850@xmath0 scuba observations from @xcite . assuming a dust temperature of 35k , consistent with the sed fit from @xcite
, we obtain a value of 1.57@xmath510@xmath6 m@xmath69 .
then , using our upper limit for the co(21 ) line , and assuming that the gas is fully thermalized , we obtain a 3@xmath16 limit for l@xmath70<3.45@xmath510@xmath71kkms@xmath9pc@xmath72 . using these values
, we estimate an @xmath73 > 45.5m@xmath69(kkms@xmath9pc@xmath72)@xmath9 in 3c368 . this @xmath73 value is an order of magnitude larger than the typical range of values quoted in the literature , @[email protected] ( e.g. , * ? ? ?
* ; * ? ? ?
previous attempts to detect molecular gas using the co(10 ) line in high redshift radio galaxies have been largely unsuccessful ( e.g. , * ? ? ?
our limit of l@xmath74@xmath75<3.45@xmath510@xmath71kkms@xmath9pc@xmath72 is a factor of @xmath26.4 deeper than the lowest limit reported in @xcite ( l@xmath74@xmath75<2.2@xmath510@xmath76kkms@xmath9pc@xmath72 ) .
comparing the observed [ cii]/@xmath46 and upper - limit co(10)/@xmath46 ratios for 3c368 to those of other sources ( * ? ? ?
* figure 3 ) , we find that 3c368 has a higher [ cii]/@xmath46 ratio ( @xmath2 0.5% ) than either the starburst nuclei or galactic star forming regions , and deficient co for a source with this [ cii ] flux .
there are several factors which could explain this anomalously high [ cii]/co(21 ) line ratio , including the metallicity , age , and degree of fractionation of the molecular clouds in 3c368 .
moving toward the core of a molecular cloud from its surface , a transition is made from c@xmath35 to co as carbon - ionizing and co - dissociating photons are extincted by dust .
this transition occurs at a@xmath66 @xmath2 3 @xcite .
if the dust - to - gas ratio follows metallicity , then it follows that the penetration depth of these carbon - ionizing photons can be very large in a low - metallicity ism , making the co core relatively small compared to the c@xmath35 envelope , and leading to large [ cii]/co ratios ( e.g. , * ? ? ?
* ; * ? ? ?
while the co emitting core of a molecular cloud retreats as the metallicity decreases , the size of the molecular region , defined by the transition from h to h@xmath77 , does not .
this transition from h to h@xmath77 is set by the self shielding of h@xmath77 molecules , and is therefore metallicity independent .
co would therefore be a poor tracer of the total molecular gas in low - metallicity molecular clouds , which could still be present in the near absence of co ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , the low metallicity case is unlikely for 3c368 , which has a stellar mass of @xmath2 3.6@xmath510@xmath6@xmath7 @xcite . at the current star formation rate
, it would take @xmath2 10@xmath78years to accumulate such a stellar population . assuming that each star formation episode lasts @xmath2 10@xmath64 @xmath10 10@xmath79years , 3c368 would have been through @xmath2 10 @xmath10 100cycles of star formation , making a low metallicity ism seem improbable .
additionally , 3c368 lies on the galaxy main sequence " for a source at redshift 1.131 ( using the definition of * ? ? ?
* ) , furthering the idea that a low metallicity ism is not responsible for the lack of co in this case . a different scenario which could be used to explain the lack of co is that the ism in 3c368 may be highly fractionated into small cloudlets .
such a fractioned ism would have modest extinction to the core of individual cloudlets , allowing for considerable c@xmath35 emission with relatively little co emission .
since the upper limit from our co observations show that the line is suppressed from the expected value by more than 12@xmath5 , the co emitting cores of the molecular clouds in 3c368 must correspondingly be more than @xmath80@xmath5 smaller in radius than the entire clouds . taking a@xmath66 @xmath2 3 , or a corresponding column density of 6@xmath510@xmath81cm@xmath19 , as the co depth , and our fitted pdr density of @xmath2 7,500@xmath29
, we calculate that the clouds in 3c368 must have size @xmath20.3pc , smaller than the sizes derived for clouds in the highly disturbed ism of m82 by @xcite ( 0.4 @xmath10 1.0pc ) , while the co emitting cores of these clouds must have size < 0.1pc .
yet another possible explanation for the lack of co in 3c368 is that the molecular clouds in the galaxy may be chemically young . @xcite
have shown that c@xmath35 recombines with free electrons to form c@xmath83 in less than @xmath2 0.1myrs , while the subsequent formation of co from the neutral atoms can take significantly longer , @xmath2 1@xmath103myrs
. it could be that we are observing these molecular clouds before they have had the chance to form an appreciable amount of co. the work of @xcite suggests that a chemically young ism may be possible here .
cloudlets within a clumpy ism , moving relative to one other , can periodically shield each other from the intense uv radiation field present in the ism . in this scenario ,
the [ cii ] which we observe would originate from the surfaces of dense cloudlets exposed to the intense uv field .
the co emitting cores of the cloudlets would retreat toward the center each time they were exposed to the full uv field , and then take an appreciable time to reform after the next shielding event .
this effect would be particularly pronounced if the cloudlets were small , and hence much of their volume was penetrated by uv photons between shielding events , as seems to be the case here .
a testable prediction of this model would be the existence of considerable [ ci ] emission , since c@xmath35 recombines quickly after the shielding event is initiated .
for this reason , we plan to do follow up observations of 3c368 in [ ci ] , in order to try and resolve the mystery of the missing co.
in an effort to answer some of the remaining questions pertaining to 3c368 , we have been awarded alma time to observe the [ cii ] 158@xmath0 line , at 0@xmath152 spatial resolution , to determine the extent of star formation in this source .
the models which we have employed here predict star formation extended over kilo - parsec scales , which , if correct , will appear in the alma observations as an extended source over @xmath210beams .
in addition to the ionized carbon line , we have also been awarded time to observe the [ ci ] 609@xmath0 line in 3c368 ( the 370@xmath0 line is blocked by the atmosphere ) .
we expect that if the molecular clouds in the galaxy are indeed chemically young , the neutral carbon line will be very bright when compared to our co line upper limit , since the co would not have had sufficient time to form .
we have observed the fine - structure lines constituting the oxygen toolkit " in 3c368 , including the [ oi ] 63@xmath0 , [ oiii ] 52@xmath0 and 88@xmath0 , and [ oiv ] 26@xmath0 lines . using these oxygen lines together with archival fine - structure neon lines in the mid - infrared ,
we have modeled the h ii regions in 3c368 , using the models of @xcite , and found them to be ionized by starbursts headed by o8 stars , with an age of @xmath2 6.5myr .
this age is comparable to the estimated age of the latest episode of agn activity in 3c368 , determined by the propagation of the radio lobes .
given the rate of star formation that we are witnessing in 3c368 , with the alignment of the optical and radio axes , and estimated age of both the starburst and most recent epoch of agn activity , this source is a strong candidate for further study in the field of galaxy - bh coevolution .
we have also modeled the pdrs in 3c368 , and found them to be consistent with star formation spread over kilo - parsec distance scales .
however , we have not detected any co in 3c368 .
our 3@xmath16 limit for the co(21 ) line is < 0.201jykms@xmath9 , or equivalently < 7.3@xmath510@xmath22wm@xmath19 , a level 12@xmath5 smaller than the value expected from standard pdr models .
the lack of co emission may be due to a low - metallicity , highly fractionated , or possibly chemically young ism .
we find that the most likely explanation is that the ism is highly fractionated into small clouds , perhaps as a result of the interactions between agn winds and the star forming molecular clouds , making it appear chemically young .
future alma observations will enable us to test this theory of large - scale star formation in the highly fractionated ism of 3c368 .
we thank the anonymous referee for the insightful comments and suggestions which helped to improve this manuscript .
we additionally thank t. k. daisy leung and a. gowardhan for help with preparing this manuscript in latex .
zeus observations were supported by nsf grant ast-01109476 .
c.l . acknowledges support from an nrao student observing support award , sospa3 - 011 , and a new york space grant award . d.b .
acknowledges partial support from alma - conicyt fund no .
31140010 . d.r . acknowledges support from the national science foundation under grant number ast-1614213 to cornell university .
this paper makes use of the following alma data : ads / [email protected] and ads / [email protected] .
alma is a partnership of eso ( representing its member states ) , nsf ( usa ) and nins ( japan ) , together with nrc ( canada ) and nsc and asiaa ( taiwan ) and kasi ( republic of korea ) , in cooperation with the republic of chile .
the joint alma observatory is operated by eso , aui / nrao and naoj . | we present the detection of four far - infrared fine - structure oxygen lines , as well as strong upper limits for the co(21 ) and [ nii ] 205@xmath0 lines , in 3c368 , a well - studied radio - loud galaxy at @xmath1 .
these new oxygen lines , taken in conjunction with previously observed neon and carbon fine - structure lines , suggest a powerful active galactic nucleus ( agn ) , accompanied by vigorous and extended star formation .
a starburst dominated by o8 stars , with an age of @xmath2 6.5 myr , provides a good fit to the fine - structure line data .
this estimated age of the starburst makes it nearly concurrent with the latest episode of agn activity , suggesting a link between the growth of the supermassive black hole and stellar population in this source .
we do not detect the co(21 ) line , down to a level twelve times lower than the expected value for star forming galaxies .
this lack of co line emission is consistent with recent star formation activity if the star - forming molecular gas has low metallicity , is highly fractionated ( such that co is photodissociated through much of the clouds ) , or is chemically very young ( such that co has not yet had time to form ) .
it is also possible , though we argue unlikely , that the ensemble of fine structure lines are emitted from the region heated by the agn . |
in many glass - forming system , like amorphous polymers or supercooled liquids near the glass transition temperature , the relaxation spectrum exhibits strongly non - exponential behavior .
dielectric spectroscopy shows evidence of two relaxation processes : the so - called @xmath14 and @xmath15-relaxation .
the asymmetric @xmath16-relaxation peak flattens into an @xmath15-relaxation at high frequency domain @xcite . in general
, the @xmath16-relaxation corresponds to the atomic motion of the clusters of atoms themselves or atomic transport between clusters , while the @xmath15-relaxation corresponds to the motions within the clusters . in a phenomenological approach
these two types of relaxation are usually described by the sum of empirical , non - debye laws , namely the sum of havriliak - negami ( hn ) @xcite relaxation functions @xcite or their sum with the stretched exponentials @xcite .
the stretched exponentials ( named also the kohlrausch - williams - watts ( kww ) functions ) are extensively studied in many theoretical papers , see e.g. @xcite , and experimental papers , see e.g.@xcite . this behavior of the dielectric function can be also described by the excess wing @xcite .
this paper is devoted to the hn function which was introduced in @xcite : @xmath17^{\beta } } , \quad 0 < \alpha < 1 , \quad \beta > 0.\ ] ] here , the parameters @xmath16 and @xmath15 do not denote the @xmath14 and @xmath15-relaxation .
they are called the width and symmetry parameters , respectively .
the symbol @xmath18 denotes the frequency and @xmath1 is an effective time constant .
the symbols @xmath19 and @xmath20 denote the relative permittivity and the dielectric constant respectively .
note that eq .
generalizes the cole - cole ( cc ) @xcite relaxation ( eq . for @xmath9 ) and the cole - davidson ( cd ) @xcite relaxation ( eq . for @xmath21 ) . among experimental papers devoted to the hn functions we mention ref .
@xcite where the hn function has been measured during the monitoring of contamination in sandstone .
the hn relaxation is also observed in a complex system representing plant tissues of fresh fruits and vegetables in the frequency range @xmath22 hz , see @xcite .
it can be also applied to model a thermal flux in the field of machining by turning , in the time domain @xcite several authors have investigated the hn relaxation from different points of view .
for example , the relation between cc relaxation and the stretched exponential is established in @xcite .
a transparent subordination approach to anomalous diffusion processes underlying the hn relaxation has been proposed in @xcite .
the analytic expressions in the time domain for the hn relaxation in terms of the fox @xmath23 functions are presented in @xcite for the real values of parameters @xmath16 and @xmath15 .
( remark that the fox @xmath23 functions are defined via the mellin transform @xcite . ) the non - rational values of @xmath16 and @xmath15 are obtained by making the numerical fitting of the hn function with the experimentally obtained data .
for this reason the solutions expressed via the fox @xmath23 functions are correct but impractical because of the fox @xmath23 functions are not yet implemented in the computer algebra systems .
therefore the expression involving the fox @xmath23 functions for general parameter values are purely symbolic and , for the moment , can not be used in actual calculations .
the purpose of this paper is to express the relevant functions , like the response functions , the probability densities and the relaxation functions , related to the hn relaxation in terms of the special functions implemented in the computer algebra systems . that can significantly simplify the numerical calculations
. to realize this purpose we approach the non - rational value of parameter @xmath16 by the rationals such that @xmath8 . hence
, everywhere in the paper we take @xmath24 with @xmath6 and @xmath7 relative prime integers .
another aim of the paper is to propose the probability density related to the response function .
that will hopefully allow one to find localization of the relaxed center in the given sample .
the present paper is organized as follows .
ii contains the basic facts about the relaxation theory , namely the relation between the response functions , the probability densities and the relaxation functions . in sec .
iii , the response functions @xmath2 , @xmath8 and @xmath25 , are represented via the prabhakar function which contains the ( three - parameter ) generalized mittag - leffler functions . for @xmath4 , where @xmath6 and @xmath7 are integers
, we represent the ( three - parameter ) generalized mittag - leffler functions as the finite sum of the generalized hypergeometric functions .
next , we find the asymptotic properties of the response functions and the ( three - parameter ) generalized mittag - leffler functions . by expressing the response function in terms of the meijer g function in sec .
iv we calculate the probability density @xmath26 connected with the hn relaxation .
we show that for @xmath5 , @xmath26 is positively defined and normalized .
thereafter , we find all the moments and asymptotics of @xmath3 . in sec .
v the relaxation functions appropriate to the hn relaxation denoted as @xmath27_{\alpha , \beta}$ ] are derived .
there the integral form of evolution equation is also found . the new relation between @xmath2 and the one - sided lvy stable densities
is found in sec .
the paper is concluded in sec .
in the theory of relaxation the complex frequency - dependent absolute permittivity ( dielectric constant ) of the material is given by @xcite @xmath28.\ ] ] the symbol @xmath29 = \int_{0}^{\infty } e^{-p t } f(t/\tau_0 ) dt$ ] , @xmath30 , defines the laplace transform of @xmath31 @xcite .
the response function @xmath32 ( the number of polarized center per unit time ) is related to the ratio of polarization at time @xmath33 to all possible polarization via @xmath34 . inserting this relation into the laplace transform in eq . , comparing with eq . , and using eq .
( 3 - 4 - 1 ) of @xcite we have @xmath35^{-\beta } = 1 -i\omega\ , \mathcal{l}\left[\frac{n(t)}{n_{0 } } ; i\omega\right].\ ] ] from eq . , hilfer calculates all possible relaxations ( called also the relaxation functions ) @xmath36 in @xcite using the fox @xmath23 function and the series .
his relaxation function expressed via the fox @xmath23 function can be obtained by calculating the laplace transform in eq . and , thereafter , inverting it .
on the other hand @xmath36 can be considered as the relaxation of the sample which contains @xmath37 centers of oriented polarization .
each of these centers relaxes with a different relaxation time @xmath38 .
the relaxation of all sample should be the weighted sum of the debye s relaxations .
that is @xmath39 the probability distribution @xmath40 is a positive function which satisfies @xmath41 .
taking the infinitesimally small @xmath42 and going with @xmath37 to infinity we get @xmath43 where @xmath44 . taking the derivative over time of the first equality in eq .
the response function reads @xmath45 we point out that the equality of @xmath36 derived by hilfer ( see eq . where the information about the debye s relaxation is absent ) to @xmath36 given by eq .
would mean that the hn relaxation is somehow built from the debye s relaxation .
let us denote the response function in time domain related to the hn relaxation as @xmath46 .
the subscripts @xmath16 and @xmath15 are the width and symmetry parameters in eq . .
we calculate the function @xmath2 by comparing eqs . with , and , next , we invert the laplace transform . thus , by using eq .
( 2.5 ) of @xcite for @xmath47 , @xmath48 and @xmath49 the response function can be expressed via the so - called prabakhar function @xcite : @xmath50 ( the subscript @xmath51 , as the various subscripts below , is only added to emphasize the reference from which the formula was taken . )
the ( three - parameter ) generalized mittag - leffler function @xmath52 is commonly given through the series @xcite @xmath53 where @xmath54 denotes the pochhammer symbol .
remark that for @xmath55 eq . gives @xmath56 which is the ( two - parameter ) mittag - leffler function @xcite .
for @xmath57 eq . is equal to the ( classical ) mittag - leffler function @xmath58 @xcite , and for @xmath55 and @xmath59 we have @xmath60 .
the extensive numerical calculation of @xmath61 is considered in @xcite , where the authors test the stability and the validity range for the parameters @xmath16 and @xmath62 .
their algorithm is based on integral representations and exponential asymptotics . using this algorithm
they simulate the behavior of @xmath61 also at zero and at infinity , and estimate the error of their method .
we attempt to express the generalized mittag - leffler function of eq . through a finite sum of the generalized hypergeometric function @xmath63 @xcite .
for this reason , we take the rational @xmath16 , @xmath64 , and change the summation index as follows : @xmath65 , where @xmath66 is the index which appears in eq . .
here , we introduce the index @xmath67 which indicates how many generalized hypergeometric functions should be in the sum .
thus , we get @xmath68 with @xmath69 .
the first list of `` upper '' parameters is equal to @xmath70 followed by @xmath71 , whereas the second list of `` lower '' parameters is the union of @xmath72 and @xmath73 .
let us now check if eq
. reconstructs the exponential behavior of @xmath61 used in @xcite . starting from eq . and the last unnumbered formula on p. 155 of @xcite we can estimate @xmath74 for large values of @xmath75 . according to it @xmath76 function is equal to @xmath77 , where @xmath78 as @xmath79 , where we employed the gauss - legendre multiplication formula for gamma functions . substituting it to eq .
we have @xmath80 which for @xmath24 and @xmath55 reproduces the dominant term of ( * ? ? ?
* eq . ( 2.4 ) ) which presents the exponential behavior of @xmath81 .
moreover , @xmath52 for @xmath82 and @xmath83 can be alternatively represented via the one - sided lvy stable distribution @xmath84 @xcite : @xmath85 eq . can be proved by writing the exponential function in the series form and using the explicit values of moments of @xmath86 , see @xcite . for closely related algebraic and transformation properties of lvy stable functions
consult @xcite .
note that eq .
generalizes the known relation between the classical mittag - leffler function and the one - sided lvy stable distribution , see , e.g. , ( * ? ? ?
* eq . ( 7 ) ) or ( * ? ? ?
* eq . ( 11 ) ) .
inserting the asymptotics of @xmath87 for large @xmath88 given by ( * ? ? ?
* eq . ( 5 ) ) into eq .
we get @xmath89 . after substituting this result into eq .
we reconstruct the asymptotics at infinity given in the last unnumbered formula on p. 76 of @xcite , namely @xmath90 .
the behavior of @xmath2 for @xmath91 looks like the asymptotics of @xmath86 for @xmath92 . that suggests the link between the prabhakar function and the one - sided lvy stable distribution .
we investigate this link in some detail in sec .
vi . from eqs .
and with @xmath82 and @xmath93 we write @xmath94 as the finite sum of @xmath76 s functions .
moreover , @xmath94 can be expressed through the standard special functions for @xmath95 only : @xmath96 where @xmath97 is the parabolic cylinder function @xcite . in the derivation of eq .
we employ eqs .
( 7.11.1.9 ) and ( 7.11.1.10 ) on p. 579 of @xcite . to simplify the calculation in sec .
iv , we insert eq . with @xmath84
given by ( * ? ? ?
* eq . ( 2 ) ) into eq . and
employ formula ( 2.24.3.1 ) on p. 295 of @xcite .
thus , the prabakhar function can be expressed in terms of the meijer g function @xmath98 @xcite : @xmath99 the upper and lower lists of parameters are equal to the union of @xmath100 and @xmath101 , respectively . from eq .
and eq . ( 2.24.2.1 ) on p. 293 of @xcite we calculate its @xmath102-th stieltjes moment @xmath103 which has the form @xmath104 where @xmath105 $ ] denotes the stieltjes moment of the the one - sided lvy stable distribution @xcite .
the stieltjes moment @xmath106 is finite for @xmath107 and is infinite otherwise .
note that @xmath108 is finite in the larger range of @xmath102 , namely for @xmath109 .
@xmath108 does not exist for @xmath110 .
( color online ) plot of @xmath111 given by eq . for @xmath112 , @xmath113 and @xmath114 ( i ; red ) , @xmath115 ( ii ; blue ) , and @xmath116 ( iii ; green ) . ] eq .
can be readily used to study the prabakhar functions graphically .
the prabakhar functions @xmath117 given by eq . for @xmath118 are plotted in fig .
[ fig0 ] . in the limit of @xmath119
they go to infinity for @xmath120 , @xmath121 for @xmath115 , and @xmath122 vanishes for @xmath123 . that exemplifies the asymptotic behavior of @xmath124 for @xmath119 @xcite which goes to @xmath70 for @xmath125 .
it also shows that @xmath117 vanishes when @xmath33 tends to infinity .
this confirms the asymptotics found below eq . .
the inverse laplace transform allows one to pass from the time domain @xmath126 to the space domain @xmath127 and to address the question of emergence of probability distribution functions ( pdf ) in @xmath88 .
the function @xmath3 is denoted here by @xmath128 . for @xmath129 , where @xmath7 and @xmath6 are integers such that @xmath130 , @xmath26 can be presented as the inverse laplace transform of eq .
: @xmath131.\ ] ] substituting the prabhakar function given by eq . into eq . and employing eq .
( 3.38.1.2 ) on p. 393 of @xcite we represent @xmath26 in the form of the meijer g function @xcite : @xmath132 the symbol @xmath133 is defined after eq . .
the numerical tests show that @xmath26 is a positive function for @xmath134 , whereas it has a negative part for @xmath135 , see fig .
[ fig0a ] .
( the analytical confirmation of that fact is below eq . .
) ( color online ) plot of @xmath26 given by eq . for @xmath113 and @xmath114 ( i ; red ) , @xmath115 ( ii ; blue ) , and @xmath116 ( iii ; green ) .
observe that line iii ( green ) is negative for @xmath136 . ] employing eq .
( 2.24.2.1 ) on p. 293 of @xcite we show that all their fractional moments @xmath137 are equal to @xmath138 for real @xmath139 , @xmath140 and @xmath141 , they are finite , and they are infinite otherwise .
it means that @xmath26 are normalized and their higher moments like the mean value or variance do not exist .
similar but not identical behavior is observed for the one - sided lvy stable distributions @xmath86 for which only fractional moments exist and all the higher moments are infinite @xcite .
the existence of normalization and positivity of @xmath26 for @xmath134 permits to conclude that @xmath26 for @xmath5 , @xmath6 and @xmath7 integers , are the pdf .
furthermore , it turns out that using eq .
( 8.2.2.4 ) on p. 617 of @xcite and the gauss - legendre multiplication formula for gamma functions in eq .
, we can rewrite @xmath26 as a finite sum of @xmath6 generalized hypergeometric functions : @xmath142 \\ & \times { _ { 1+k}f_{k}}\left({1 , \delta(k , \beta+j ) \atop \delta(k , 1+j ) } ; \frac{(-1)^{l - k}}{u^{l}}\right ) .
\end{split}\end{aligned}\ ] ] eq .
gives a closed form of @xmath143 .
for example , for the cc relaxation ( @xmath114 ) it can be observed that the appropriate cancelation in @xmath144 s gives @xmath145 which , after applying eq .
( 7.3.1.1 ) on p. 453 of @xcite , yields @xmath146 $ ] . employing eq .
( 1.353.1 ) on p. 38 of @xcite to the remaining sum over @xmath147 we get @xmath148 where @xmath149 .
we point out that eq . was obtained in @xcite by using a different method ,
( 3.24 ) on p. 245 there .
all the distributions @xmath150 share the following features : ( i ) @xmath150 for @xmath8 and @xmath151 is positive ; ( ii ) @xmath150 goes to infinity at @xmath152 , and it vanishes for @xmath153 ; ( iii ) @xmath150 is an decreasing function for @xmath154 , it contains the flat sector near @xmath155 , and it has two extrema for @xmath156 . for @xmath155 the derivative @xmath157 is zero at the point @xmath158 .
the numerical calculation shows that for the cc relaxation we have @xmath159 and @xmath160 ; ( iv ) @xmath150 has a maximum at @xmath161 and a minimum at @xmath162 for @xmath156 , where @xmath163/(1+\alpha)$ ] .
the upper sign is for maximum value of @xmath88 and the lower sign is for minimum of @xmath88
offers an unlimited number of solutions for @xmath26 . for @xmath164
only it can be written down in terms of standard special functions .
for example , for @xmath165 it is equal to @xmath166 which is plotted for @xmath167 in fig .
[ fig1 ] , see line i in red , whereas @xmath26 for @xmath168 is equal to @xmath169 in the derivation of eq .
we have used eqs .
( 7.3.3.1 ) and ( 7.3.3.2 ) on p. 486 of @xcite .
. has been obtained by employing eqs .
( 7.4.1.30 ) - ( 7.4.1.32 ) on p. 499 of @xcite . from eq . , after analyzing the behavior of sine , we conclude that @xmath170 is positive for @xmath171 , whereas it is negative for @xmath172 and @xmath173^{2}$ ] . in fig .
[ fig1 ] we exhibit the probability densities of @xmath26 given by eq . for @xmath174 and @xmath175 .
( color online ) plot of @xmath3 given by eq . for @xmath174 and @xmath176 ( i
; red ) , @xmath113 ( ii ; blue ) , and @xmath177 ( iii ; green ) . ]
[ fig1 ] shows that for fixed value of @xmath15 there exists @xmath178 for which @xmath179 .
for example , for @xmath174 it is @xmath180 and @xmath181 . for @xmath174 and @xmath182
the maximum is higher and it moves in the direction of larger values of @xmath88 .
we now find the series representation of @xmath3 for rational @xmath24 .
to realize that objective we use the series representation of the generalized hypergeometric function , see eq .
( 7.2.3.1 ) on p. 437 of @xcite .
in this way we obtain the double sum : one over @xmath183 , @xmath184 , which is the index appearing in the series representation of @xmath63 , and another one over @xmath147 , @xmath67 , is from eq . .
next , we change the summation index as follows @xmath185 .
thus , @xmath3 has the form @xmath186}{u^{1+\alpha(\beta+r)}}.\ ] ] it turns out that writing sine as the imaginary part of @xmath187 , taking the integral representation of the gamma function ( * ? ? ?
* eq . ( 8.312.2 ) ) , and applying eq . , we get the integral form of eq . :
@xmath188 the auxiliary function @xmath189 written in terms of the ( three - parameter ) generalized mittag - leffler function is equal to @xmath190 using eq . in eq .
it is easy to demonstrate the following features of @xmath189 .
it can be shown that @xmath189 vanishes at @xmath191 .
@xmath189 goes to @xmath70 in the limit of @xmath192 .
the asymptotics at @xmath193 is presented in unnumbered formula for @xmath194 and @xmath195 on p. 76 of @xcite . for the cc relaxation ( @xmath9 )
, @xmath189 can be expressed via the classical mittag - leffler function , namely @xmath196 which is quoted in ( * ? ? ?
* eq . ( 9 ) on p. 185 ) . for the cd relaxation ( @xmath197 ) we have @xmath198 @xcite , where @xmath199 is the incomplete gamma function
. moreover , @xmath200 which results from eq .
( 1.9.6 ) on p. 46 of @xcite . by applying ( * ? ? ?
( 2.5 ) ) to eq . with eq . and
by employing de moivre s formula to calculate the imaginary part of so obtained equation , we derive the modification of @xmath3 which is given by eq .
( 8) for @xmath201 , @xmath202 , and @xmath195 of @xcite .
namely , if @xmath8 and @xmath25 we get @xmath203}{[u^{2\alpha } + 2u^{\alpha}\cos(\pi\alpha ) + 1]^{\beta/2}}\ ] ] with @xmath204 ( defining the range of @xmath88 , see below ) , and @xmath205.\ ] ] for @xmath206 eq .
works for @xmath207 and @xmath208 . for @xmath209 eq . with @xmath207 defines @xmath3 for @xmath210^{1/\alpha}$ ] . eq . with @xmath211 gives @xmath3 for @xmath212^{1/\alpha}$ ] . in this way we can define @xmath3 for all @xmath127 .
note that the presence of @xmath213 in eq . allows us to build the positively defined and normalized pdf @xmath3 for arbitrary real @xmath16 , @xmath8 and @xmath214 .
this property is absent in the solution of ( * ? ? ?
* eq . ( 8) ) for @xmath201 , @xmath215 , and @xmath216 , which is defined for @xmath8 and @xmath211 .
for example ,
@xmath217 , where @xmath218 is given by ( * ? ? ?
* eq . ( 8) ) , is negative for @xmath219 . introducing the term @xmath213 we improve ( * ? ? ?
we remark also that eqs .
, , and gives the same results .
using the various but equivalent forms of @xmath3 make it easier to show many of its properties .
for example , from eqs . and
it can be shown that @xmath3 is positive for @xmath220 , whereas it contains the negative parts for @xmath221^{1/\alpha}$ ] for @xmath222 . from eq .
, in a simple way , we get that @xmath3 goes to infinity in the limit of @xmath223 and @xmath3 vanishes for @xmath224 .
the benefit of use eqs . or rather than eq .
is that it is the only formula which defines @xmath3 for @xmath209 and @xmath208 .
we define now the relaxation function @xmath225_{\alpha , \beta}$ ] whose explicit and exact form is obtained by calculating the laplace transform with @xmath3 given in eq . .
here , we use the formulas from @xcite , i.e. eq .
( 2.24.3.1 ) on p. 350 , eq .
( 8.2.2.14 ) on p. 619 , and eq .
( 8.2.2.3 ) on p. 618 , and , then , we compare the results with eqs . and .
thus , we get ( * ? ? ?
( 30 ) ) : @xmath226_{\alpha , \beta } = 1 - g_{\alpha , \beta}(\ulamek{t}{\tau_{0}}),\end{aligned}\ ] ] where @xmath227 is given in eq . with eq . .
( color online ) plot of @xmath27_{\alpha , \beta}(x)$ ] given in eqs .
and for @xmath167 and @xmath176 ( i ; red ) , @xmath113 ( ii ; blue ) , and @xmath177 ( iii ; green ) . ]
remark that eq . for @xmath228 is equal to one and it vanishes for @xmath33 going to infinity , see fig .
eq . resembles the series representation of the relaxation function obtained in @xcite .
it suggests that the hn relaxation could be explained by using the debye s relaxation .
for instance , for @xmath9 the cc relaxation can be build from the debye s relaxations @xcite . in @xcite
. is also represented in terms of the fox @xmath23 functions which however are not accessible in the computer algebra systems . the advantage of our solution , eq . with eqs . and
, is clearly seen in practice . since in recent versions of the computer algebra systems
the generalized hypergeometric functions @xmath63 as well as the meijer @xmath229 functions are fully implemented , their use permits high - precision calculations . the explicit form of relaxation function @xmath27_{\alpha , \beta}$ ] given by eq . and
allow us to derive the self - similar properties of @xmath3 .
rewriting the first equality in eq .
, where instead of @xmath230 we take @xmath231 , @xmath232 , and taking into account eq .
, we get @xmath233 where @xmath234 . an one - variable function @xmath3 is uniquely extended to a two - variable @xmath235 one : @xmath236 we stress that eq .
is the self - similar property which is also obeyed by the ( classical ) mittag - leffler function @xcite and the lvy stable distribution @xcite . from the second equality in eq . with @xmath237 , @xmath238 , and for @xmath239
we get the laplace - like convolution properties : @xmath240 the similar property is fulfilled also by the one - sided lvy stable distribution , see ( * ? ? ?
* eq . ( 12 ) ) and ( * ? ? ?
* eq . ( 13 ) ) .
. differs from the standard laplace convolution of _ one _ variable function . here
, we have the integral form of evolution equation of _ two _ variables density distribution @xmath241 where the _ both _ variables are changed .
. can be proved by employing eqs . for @xmath242 , @xmath238 , and .
observe also that eq .
is the evolution equation written in the integral form .
following @xcite the relation between hn relaxation and the one - sided lvy density @xmath86 is usually understood in the form of ( * ? ? ?
* eq . ( 7 ) ) or ( * ? ? ?
* eq . ( 11 ) ) . here
, we propose a new kind of link which will show the correlation of @xmath2 , for arbitrary @xmath8 and @xmath244 , with @xmath243 , @xmath8 .
to achieve this task , we reparametrize in eq .
@xmath10 , @xmath11 and @xmath245 throughout this section .
the range of parameter @xmath13 , @xmath12 , assures that @xmath244 .
the derivative of eq . over @xmath246 taken with the opposite sign is the probability density function @xmath247^{\frac{1}{1-q}}\ ] ] called the @xmath13-weibull distribution @xcite .
we point out that eq .
goes to the weibull distribution @xcite for @xmath248 and the weibull distribution is the derivative of the stretched exponential .
thereafter , we take the prabhakar function given in eq . with the ( three - parameter )
generalized mittag - leffler function given via the finite sum of the generalized hypergeometric functions , see eq . .
next , in fig .
[ fig4 ] we compare @xmath249 for @xmath176 and @xmath250 with the @xmath251 .
( color online ) plot of @xmath252 given by eq .
, where the prabhakar function eq .
is defined for @xmath11 , @xmath253 , and @xmath254 ( i ; red ) , @xmath255 ( ii ; blue ) , and @xmath256 ( iii ; green ) .
the black , dashed line shows the one - sided lvy stable distribution for @xmath176 , i.e. @xmath257/(2\sqrt{\pi } t^{3/2})$ ] . ] in fig .
[ fig4 ] we observe that @xmath252 goes over to the one - sided lvy stable distribution @xmath251 when @xmath13 is approaching one . the behavior of @xmath258 in the limit of @xmath248 can be also shown by using in eq . and the series representation of the ( three - parameter ) generalized mittag - leffler function given in eq . .
after applying the gauss - legendre multiplication formula to gamma functions in the denominator of eq .
we get @xmath259^{\alpha(n + \frac{2-q}{q-1 } ) - 1 } \\ & \times \gamma[1-\alpha(n+\ulamek{2-q}{q-1})]\ , \sin[\alpha(n+\ulamek{2-q}{q-1 } ) ] .
\end{split}\end{aligned}\ ] ] writing the sine function as the imaginary part of @xmath260 $ ] and employing eq .
( 8.312.2 ) on p. 892 of @xcite , the integral representation of @xmath258 is found in the form @xmath261^{-\frac{2-q}{q-1 } } \frac{dy}{\pi } \right\}.\ ] ] in the limit of @xmath248 eq . goes to the one - sided lvy stable distribution @xmath243 given in the first unnumbered equation in the last page of @xcite .
the stieltjes moments of @xmath2 given in eq .
go to the stieltjes moments of @xmath243 for @xmath248 . to show that we express @xmath1 and @xmath15 through @xmath13 , where @xmath12 , namely by taking @xmath11 and @xmath10 .
thereafter , we apply the stirling formula for @xmath262 and @xmath263 where @xmath15 is appropriately modified .
starting form the basic facts from relaxation theory we have introduced the response functions , the probability densities and the relaxation functions of the hn relaxation .
the response functions are given via the prabhakar functions related to the ( three - parameter ) generalized mittag - leffler functions whose representation in terms of the finite sum of the generalized hypergeometric functions for rational @xmath16 is found .
we also express the ( three - parameter ) generalized mittag - leffler functions via the one - sided lvy stable distributions .
that relation for @xmath114 ( the cc relaxation ) goes over to the known relation between the ( classical ) mittag - leffler function and the one - sided lvy stable distribution . in this way it generalizes the latter mentioned relation .
we have also identified the values of parameters @xmath15 , i.e. @xmath220 , for which the normalized function @xmath3 connected with the response function can be called the probability density .
the moments of @xmath3 were also calculated .
for @xmath220 the mean values , the variance and the higher moments of @xmath3 do not exist .
similar properties characterize the one - sided lvy stable distribution .
we also derived the laplace - like convolution properties which are the integral form of evolution equations .
moreover , we present the evidence that @xmath2 and the appropriate @xmath243 present the identical asymptotic behavior at infinity .
this suggests the existence of the link between @xmath2 and @xmath243 . this link has been also considered
the main benefit of the paper is that , in addition to finding the explicit and exact forms of the relaxation functions , we represent them as the generalized hypergeometric functions which are implemented in the computer algebra systems .
the use of the generalized hypergeometric functions have significantly simplified the calculation and in the fast way allowed us to obtain many results .
another advantage is that we found restrictions on @xmath15 for which @xmath3 is the probability density . for this value of @xmath15
, @xmath3 gives localization of relaxing centers in the sample . that could give the intuition in which way we should prepare the experimentally measured sample .
we have also elucidated the way in which the various versions of @xmath2 , @xmath3 , and @xmath27_{\alpha , \beta}$ ] appearing in the literature can be connected with each other .
we have also shown the existence of new relations between the response function related to the hn relaxation , and the one - sided lvy stable distribution , which can allow us to find the differential equation related to the integral laplace - like convolution .
k. g. , a. h. and k. a. p. were supported by the pan - cnrs program for french - polish collaboration .
moreover , k. g. thanks for support from mnisw ( warsaw , poland ) , `` iuventus plus 2015 - 2016 '' , program no ip2014 013073 . | we study the functions related to the havriliak - negami frequency relaxation @xmath0^{-\beta}$ ] with @xmath1 characteristic time , measured in many experiments .
we furnish exact and explicit expression for the response function @xmath2 in time domain and a probability density @xmath3 in space domain for @xmath4 and @xmath5 , with @xmath6 and @xmath7 positive integers . for @xmath8 and @xmath9
we reproduce the functions related to the cole - cole relaxation .
we use the method of integral transforms .
we show that @xmath2 with @xmath10 and @xmath11 , @xmath12 , goes over to the one - sided lvy stable distribution when @xmath13 tends to one .
moreover , applying the self - similar property of @xmath3 we introduce two - variable density which satisfies the integral form of evolution equation . |
0 the aim of this paper is to investigate the properties of the finite - temperature ( ft ) effective potential of the gauge models in the de sitter space .
the temperature @xmath7 is introduced as the temperature of quantum fields which are in thermal equilibrium in the static de sitter coordinates .
thus , the corresponding quantum state is determined by the green function which is periodical along the time coordinate of the static frame with period @xmath8 .
this enables one to use the euclidean formulation of the theory .
in particular , one can define the partition function @xmath9 for arbitrary @xmath10 as a functional integral , where the fields configurations belong to the spherical domain @xmath11 with conical singularities on the euclidean horizon .
there are at least two motivations for this work .
the first one is to obtain the one - loop effective potential on such non - trivial spaces like @xmath11 in an analytical form .
this is a continuation of the pioneering studies by shore @xcite and allen @xcite which investigated the effective actions on hyperspheres @xmath12 .
furthermore , the paper extends the results of ref.@xcite concerning the scalar fields , to more interesting gauge models with spontaneous symmetry breaking .
a second motivation is related to the possible application of these results to cosmological problems since the exponentially expanding phase of the early universe @xcite is described by the de sitter geometry .
our analysis has become possible after the spectrum of vector laplacians on singular @xmath13-spheres has been found explicitly in ref.@xcite . in this paper a special attention is paid to calculation and comparison of the effective potentials for two physically relevant quantum states : the de sitter invariant ( dsi ) and the zero - temperature ( zt ) states .
they both can be considered as possible vacua in quantum field theory in the de sitter space .
the dsi state resembles the poincar invariant vacuum of the minkowski space - time since it preserves the symmetry of the de sitter space .
the corresponding temperature is equal to @xmath14 , where @xmath1 denotes the de sitter radius @xcite and the euclidean section of the de sitter space is the hypersphere @xmath12 . in analogy with black hole physics this temperature
can be called the hawking temperature . in the gibbons
hawking path integral @xcite the euclidean gravitational action with a positive cosmological constant has the extremum on a 4sphere @xmath12 .
thus , the choice of dsi state has a natural explanation in the semi - classical treatment of the quantum cosmology .
for this reason the one - loop effective action for quantum fields on @xmath12 has been studied in a number of works @xcite-@xcite , @xcite . in particular , some authors @xcite investigated the phase structure of the gut models with respect to the value of the de sitter radius @xmath1 . the dsi vacuum is analogous to the hartle hawking vacuum @xcite introduced for quantum fields around an eternal black hole . from this point of view
the zero temperature state ( @xmath15 ) in the de sitter space is analogous to the boulware vacuum @xcite .
the properties of this state are much less investigated , and we use here the explicit expressions of the effective potential to see how the choice of the vacuum affects the phase structure of the gauge theories with spontaneous symmetry breaking . the paper is organized as follows .
we briefly discuss in section 2 why the euclidean functional integral on the spherical domains @xmath11 can be interpreted as a partition function in the static de sitter space .
the ft effective potential for gauge models is computed with the @xmath0function method in section 3 . in section 4
we compare the symmetry breaking mechanisms at different values of @xmath1 in the two vacua for models of scalar electrodynamics .
summary of the results is given in section 5 .
technical details concerning the derivation of the @xmath0function are left for appendix .
0 in the static coordinates the line element of the de sitter space with radius @xmath1 can be written in the form @xmath16 where @xmath17 , @xmath18 , and @xmath19 .
the above coordinates ( [ ii.1 ] ) cover only a part of the space .
one can consider a killing vector field generating a one parameter group of isometries , subgroup of so(1,4 )
. then coordinates ( [ ii.1 ] ) can be related to the time like part of the killing field associated with the translations along the time @xmath20 .
this region is restricted by the killing horizon where the killing field is null .
it also coincides with the event horizon for the observers with coordinates @xmath21 .
let @xmath22 be a scalar green function in the de sitter space - time for dsi state ( its some explicit expressions can be found in @xcite@xcite ) .
this function is periodic when the time coordinates of points @xmath23 or @xmath24 are independently increased by @xmath25 @xcite where @xmath26 .
thus the euclidean green function @xmath27 which is obtained from @xmath28 with the help of a wick rotation , is defined on a 4-sphere @xmath12 with the line element @xmath29 where @xmath30 .
note that because of the periodicity property in imaginary time the function @xmath22 can be interpreted as a green function for a canonical ensemble at temperature @xmath31 .
the physical meaning behind this interpretation is that any freely moving observer experiences the dsi state as a thermal bath at temperature @xmath31 , see ref.@xcite
. a natural generalization of @xmath22 is a green function @xmath32 in the de sitter space which is periodic in imaginary time with an arbitrary period @xmath10 .
functions of this kind can be constructed from @xmath22 with the help of a _ reperiodization _ formula suggested by dowker @xcite .
one can interpret @xmath32 as a green function of a canonical ensemble of particles in the static part ( [ ii.1 ] ) of de sitter space at temperature @xmath7 .
it is also possible to introduce the euclidean function @xmath33 by using a definition analogous to eq.([ii.4 ] ) .
the space where @xmath33 is set on is a spherical domain @xmath34 , which is described by the metric ( [ ii.5 ] ) of the 4-sphere , but now with period @xmath10 along @xmath35 .
the main property of @xmath34 is that it has conical singularities at the points @xmath36 , near which the space looks as @xmath37 , where @xmath38 is a cone with deficit angle @xmath39 .
their relation was investigated in the case of black hole geometries with the aim to establish the statistical explanation of the black hole entropy , see for instance ref.s@xcite@xcite .
we will not dwell on this issue further since it is not relevant for our purposes .
it has to be also mentioned that in the presence of the killing horizon the function @xmath49 does not have the hadamard form @xcite when @xmath50 . in this case
the stress energy tensor of a quantum field has a non - integrable divergence on the horizon of the chosen static coordinate system @xcite .
as was discussed in ref.@xcite , this property alone may be not sufficient to exclude ft states as unphysical , since the computation of the stress tensor neglects the backreaction effects .
such effects are very strong near the horizon and a reliable computation requires a nonperturbative approach which is not still developed . however , a non - hadamard form of @xmath49 is not an obstacle for the definition of the partition function ( [ ii.6 ] ) .
the conical singularities of @xmath11 result only to a number of additional ultraviolet divergent terms in the effective action @xmath51 and one can use the _ standard _ renormalization procedure to give a meaning to the integral ( [ ii.6 ] ) .
we will discuss this later .
0 we focus now on the definition and computation of the ft effective potential for the scalar electrodynamics in the de sitter space .
a similar computation has been done before by shore @xcite and allen @xcite for the dsi state .
we restrict our analysis to the abelian theories because the generalization to non - abelian case , as was shown in ref.@xcite , is almost straightforward . as was shown in ref.@xcite , a point @xmath76 where @xmath69 has a minimum , coincides with the one - loop value of the average @xmath67 .
the real part of @xmath69 is a sum of the classical potential energy @xmath77 and a quantum correction .
if the field configuration @xmath76 is unstable , then @xmath78 has a nonvanishing imaginary part which determines the decay probability of this configuration .
hereafter , we will be interested in the situation where the one - loop quantum effects can change significantly the form of the classical potential . as was pointed out by coleman and weinberg @xcite , who investigated the model ( [ iii.1 ] ) in the flat space
, one has to assume to this aim that the gauge coupling @xmath89 is of the order of @xmath56 . in this case
the quantum corrections due to the gauge fields will be comparable with the classical potential . on the other hand , to provide the convergence of the perturbation expansions one has to assume that @xmath90 . in this case
one can safely neglect the contributions to the effective potential of the scalar loops which will be proportional to @xmath91 .
let us note that the @xmath0function method automatically gives a finite expression for the quantum determinants .
the last term in the right hand side of eq.([iii.13 ] ) corresponds to the finite counterterms which are always present after renormalization .
the conical singularities are known to introduce additional ultraviolet divergences , see for instance ref s @xcite .
the renormalization on manifolds with conical singularities has been discussed recently in a number of papers @xcite-@xcite,@xcite . in the case of the hodge - derham operator
all one - loop divergences which are linear in the conical deficit angle , as shown in ref.@xcite , are removed under standard renormalization of the gravitational couplings @xcite in the bare gravitational action .
the renormalization of the remaining divergences of order @xmath101 in the deficit angle or higher is more involved , and requires additional counterterms in the effective action .
these terms have the form of integral invariants defined on the singular surface @xcite .
the values of the additional couplings can not be predicted in the theory , and so to simplify the analysis we will assume that these couplings in @xmath79 are absent .
what is interesting is that @xmath120 can be found in an analytical form in the two most interesting limits : in the dsi state and in the zt state .
we make use of this fact to consider the phase transitions in the gauge model at different values of the de sitter radius and compare the phase structures of the theory in these two cases .
it is also interesting to compare the form of potentials ( [ iii.17 ] ) and ( [ iii.19 ] ) in the opposite limit , when the curvature of the space - time is very large , namely @xmath150 . since we are interested in the phase transitions , terms in the potential which do not depend on @xmath70 are irrelevant .
hence , from now on it is more convenient to deal with the difference @xmath151 which vanishes for @xmath152 .
0 in order to compare phase transitions in the scalar electrodynamics ( [ iii.1 ] ) in dsi and zt states we rewrite expressions ( [ iii.17 ] ) and ( [ iii.19 ] ) in the following dimensionless form ) was suggested in @xcite . ]
@xmath159 = 4 \frac xy\left(p+{13 \over 6}-\log y\right)\ ] ] @xmath160 @xmath161\ ] ] @xmath162 @xmath163 here @xmath164 and the renormalization parameter @xmath100 has been expressed , according to eq.([iii.22 ] ) , in terms of the vector boson mass as @xmath165 .
interestingly , all the information about the constants @xmath166 , @xmath56 and @xmath57 of the considered model is contained in the parameters @xmath167 and @xmath168 defined in eqs.([iii.26 ] ) and ( [ iii.29 ] ) , respectively . for arbitrary values of constants @xmath166 , @xmath56 and @xmath57 the parameters @xmath167 and @xmath168 are independent .
they can be related to each other only in the case of the minimal coupling @xmath169 .
we consider now the phase structure of the theory in the limit of large and small curvatures of the de sitter space , where it can be investigated analytically . in this limit @xmath170 and one
has @xmath171 @xmath172 eq.([iv.4 ] ) coincides with the result of ref.@xcite .
as was shown in ref.@xcite the symmetry breaking in this limit in dsi state is characterized by first order phase transitions . under such transitions ,
when the curvature radius @xmath1 reduces to a critical value @xmath173 , the mass of the vector boson changes discontinuously from a value @xmath174 to zero . by comparing eqs .
( [ iv.4 ] ) and ( [ iv.5 ] ) one can see that the effective potential at zero temperature has a similar structure , so in this case one may expect an analogous behavior
. we can use in this regime asymptotics ( [ iii.25 ] ) and ( [ iii.28 ] ) . as one can see from these equations , the terms proportional to @xmath195 change the sign from positive to negative when @xmath1 becomes larger than a critical value @xmath196 . in this case to have the symmetry breaking in the theory at @xmath197 the terms @xmath198 in ( [ iii.25 ] ) and ( [ iii.28 ] ) have to be positive .
when @xmath1 becomes smaller than @xmath196 the mass of the vector boson gradually vanishes .
this situation corresponds to second order phase transitions .
the critical radius @xmath173 in dsi state is @xcite @xmath199 in order to have a second order phase transition one must ensure that the coefficient of the @xmath198 term in ( [ iii.25 ] ) is positive . by substituting ( [ iii.27 ] ) in eq.([iii.25 ] ) we find that the above condition is satisfied @xcite for @xmath200 the value @xmath201 represents in fact a critical boundary which separates the models of the scalar electrodynamics with different kinds of phase transition . for models with @xmath167 above @xmath202 the system undergoes first order phase transitions , whereas for models with @xmath167 below @xmath202 one has second order phase transitions .
the value of @xmath202 is also confirmed by the numerical analysis .
a similar expression for the critical radius @xmath186 at zero temperature can be found from eq.([iii.28 ] ) @xmath203 the difference between the values of the critical radii can be estimated by their ratio @xmath204 it depends only on @xmath57 and @xmath205 , and for @xmath206 one can conclude that @xmath207 . for the minimal coupling ( @xmath169 ) @xmath208 .
in the case of zt state the critical value @xmath209 can be determined as well . by substituting eq.([iii.31 ] ) in the coefficient of the quartic term of eq.([iii.28 ] ) one gets @xmath210 thus @xmath211 and it is confirmed by the numerical computations .
the typical behavior of the effective potentials in case of second order phase transitions is illustrated by figures 3 and 4 , which are obtained by using formulas ( [ iv.1 ] ) and ( [ iv.2 ] ) .
figure 3 depicts the function @xmath191 for @xmath212 and figure 4 gives @xmath192 for @xmath213 . in both cases
the critical radii @xmath173 and @xmath214 are in agreement with eqs.([iii.27 ] ) and ( [ iii.31 ] ) .
as we have shown , the first order phase transitions always take place in the models of scalar electrodynamics ( [ iii.1 ] ) when constants @xmath167 and @xmath168 are large and positive . on the other hand ,
if these constants are negative with large absolute values , the symmetry breaking corresponds to second order phase transitions .
an analytical study of the region of parameters @xmath57 , @xmath56 and @xmath166 ( or @xmath202 and @xmath209 ) , where the kind of the phase transition changes from the first to the second order is difficult because the effective potentials have quite involved forms ( [ iv.1 ] ) , ( [ iv.2 ] ) .
thus , in principle , one should use here numerical methods .
it is interesting , however , that the analytical estimates @xmath215 and @xmath216 of section 4.2 are in very good agreement with the numerical results . for dsi state
this was first pointed out by allen @xcite .
as we already mentioned , the symmetry breaking mechanism in the considered models is completely determined by the parameters @xmath167 and @xmath168 . by using their definitions ( [ iii.26 ] ) , ( [ iii.29 ] ) one finds that @xmath217 note that the parameter @xmath56 must be chosen positive in order to have a scalar potential bounded from below .
we also assumed that @xmath218 in order to have a considerable quantum correction to the classical potential coming from the gauge fields .
for values @xmath219 our perturbative approach is not reliable , so we have to restrict ourselves to the interval of parameters @xmath220 .
thus , by taking into account that @xmath221 , we find with the help of eq.([iii.ab ] ) the allowed region in the @xmath167@xmath168 plane @xmath222 .
there we can make definite predictions about the properties of the considered models .
this region is shown on figure 5 .
it consists of three subregions ( i ) , ( ii ) and ( iii ) , bounded by dashed lines .
each subregion indicates a class of models with the same properties .
+ region ( i ) is determined by inequalities @xmath223 and @xmath224 .
if the coupling constants @xmath225 satisfy these restrictions , the corresponding models undergo second order phase transitions in both dsi and zt states .
+ in region ( ii ) one has @xmath226 and @xmath227 and first order phase transitions in both dsi and zt vacua . + region ( iii ) is the most interesting , since in this case @xmath228 but @xmath227 , thus the system undergoes a first order phase transition in zt case , whereas it has a second order phase transition in dsi case . in case of minimal coupling ( @xmath169 ) , according to eqs.([iii.26 ] ) and ( [ iii.29 ] ) , @xmath229 and @xmath230 , and so only second order phase transitions are possible in both states .
we investigated the effective potential for the gauge models on the singular spherical domains @xmath11 .
the spectrum of the corresponding wave operators on these backgrounds can be found exactly and it enables one to calculate the potential with the help of the @xmath0function .
we have shown that for arbitrary values of @xmath10 there is a simple relation ( [ iii.c ] ) between the one - loop corrections to the potential from gauge and scalar fields .
thus in studying the gauge models in the de sitter space one can make use of the computations of the scalar potential of ref.@xcite .
the effective action on @xmath11 can be interpreted as the free energy of fields in static de sitter space , where the parameter @xmath10 is the inverse temperature of the system .
thus , by changing @xmath10 one can learn how the presence of temperature affects the properties of such a quantum theory .
the aim of the paper was to compare the phase structures of the gauge models in the de sitter invariant vacuum , which is usually used in the quantum cosmology , with the structure of the normal vacuum which is the state at zero temperature .
according to our analysis both vacua have very similar properties in the extreme regimes of very small and very large curvatures .
if the restoration of the gauge symmetry happens at large radius @xmath1 ( small curvature ) of the de sitter space , then the corresponding phase transition is of first order regardless what vacuum is considered .
analogously , if the symmetry is restored at small values of @xmath1 ( large curvature ) , one has second order phase transitions .
the critical masses and radii are completely determined by the constants @xmath167 and @xmath168 , which are the functions of the parameters @xmath57 , @xmath166 and @xmath56 .
expressions ( [ iv.11 ] ) , ( [ iv.12 ] ) and ( [ iii.31 ] ) for the critical parameters in zt state represent the new result .
we have also found the critical value @xmath211 for the parameter @xmath168 , which separates models with second ( @xmath231 ) and first ( @xmath232 ) order phase transitions in zt state .
finally we proved the existence of a class of the models ( @xmath228 and @xmath227 ) where the kinds of phase transition in the two vacua are different .
the detailed analysis of the effective potential for all @xmath10 s is not given here , but it can be carried out in principle by making use of the properties of the @xmath0function . in the most interesting cases , however , it is sufficient to restrict the study to the expansions near the zero and hawking temperatures , as it was done in ref.@xcite .
* acknowledgements * : this work was supported in part by the natural sciences and engineering research council of canada .
0 here we consider @xmath0function @xmath233 for the transverse operator in eq.([iii.12 ] ) which is defined for convenience in the dimensionless form as @xmath234 . according to @xcite the degeneracies in four dimensions read @xmath235 while the eigen
values @xmath236 are @xmath237 where @xmath238 . as in section 3 , it is suitable to parametrize the function ( [ a.1 ] ) by the parameters @xmath10 and @xmath239 . thus one can write @xmath240 where @xmath241 function ( [ a.1a ] ) has an interesting relation to the @xmath0function of the scalar laplace operator @xmath242 on the same spherical domain @xmath11 . these operators were studied in @xcite , where they were shown to possess the eigenvalues coinciding with ( [ a.2a ] ) .
so the scalar @xmath0function is represented in the form @xmath243 where the degeneracies @xmath244 differ from the degeneracies ( [ a.2 ] ) of the vector operator . by comparing eqs .
( [ a.3a ] ) and ( [ a.4a ] ) with ( [ a.1a ] ) and ( [ a.2 ] ) one obtains @xmath245^{-z}=\ ] ] @xmath246 where @xmath247 . to get the second equality of this equation one has to decompose the last term in @xmath248 and make use of the definition of the riemann _
zeta_function @xmath249 .
the function @xmath250 has been studied in detail in the previous works @xcite , and this makes the analysis of @xmath251 much more easy .
in particular , by taking into account the result of @xcite and eq.([a.11 ] ) one obtains @xmath252 @xmath253+{11 \over 12}-\sigma~~~. \label{a.13}\ ] ] however , the analysis of the derivative of the generalized @xmath0function @xmath254 @xmath255 \label{a.14}\ ] ] is more involved and we consider only two of the most interesting cases where the result can be obtained in the simple analytical form .
this case was investigated by b. allen in ref.@xcite and our aim here is to rederive his result for @xmath256 in a different way with the help of eq.([a.11 ] ) . for the scalar @xmath0function
one can find the following representation @xcite @xmath257 @xmath258~~~,\ ] ] where @xmath259 . by using relation
@xmath260 it is possible to show that @xmath261 thus by substituting eqs.([a.15 ] ) and ( [ a.16 ] ) into ( [ a.11 ] ) and using the properties of the @xmath262function one gets @xmath263 @xmath264 @xmath265+o\left((\beta-\beta_h)^2\right)~~~. \label{a.17}\ ] ] as one can see , this expression coincides at @xmath48 with the result reported in ref.@xcite . to derive @xmath266 in the zero temperature limit one can use the method suggested in @xcite . unfortunately ,
the final formula given in @xcite has a wrong form because of a misprint , so here we use the opportunity to represent the correct answer .
according to relation ( [ a.11 ] ) and eq.(b2 ) of ref.@xcite , one has @xmath267
@xmath268 \equiv 3f(z,\sigma)~~~.\ ] ] the function @xmath269 was introduced in @xcite , where it was shown to be related to @xmath270 , see ( [ a.15 ] ) , by the differential equation @xmath271=3\sigma^{z-1 } \zeta^{(0)}\left(0,\beta_h,\sigma \right)~~~.\ ] ] thus , one can write @xmath272\ ] ] @xmath273 by making use of ( [ a.15 ] ) in ( [ a.21 ] ) one finally finds @xmath274 @xmath275 @xmath276 + g.m .
shore , ann .
* 128 * ( 1980 ) 376 .
b. allen , nucl .
* b226 * ( 1983 ) 228 .
b. allen , ann .
* 161 * ( 1985 ) 152 .
fursaev and g. miele , phys . rev .
* 49 * ( 1994 ) 987 .
a.d.linde , _ particle physics and inflationary cosmology _
( harwood academic publishers , new york 1990 ) . l. de nardo , d.v . fursaev and g. miele , _ heat - kernel coefficients and spectra of the vector laplacians on spherical domains with conical singularities _ , hep - th/9610011 , to appear in class . quantum gravity .
gibbons and s.w . hawking ,
phys . rev .
* d15 * ( 1977 ) 2738 . g.w . gibbons and s.w . hawking ,
phys . rev . * d15 * ( 1977 ) 2752
. f. buccella , g. esposito and g .miele , class .
quantum grav .
* 9 * ( 1992 ) 1499
. g. esposito , g. miele and l. rosa , class .
quantum grav .
* 11 * ( 1994 ) 2031 .
hartle and s.w . hawking , phys .
rev . * d13 * ( 1976 ) 2188 .
boulware , phys . rev . * 11 * ( 1975 ) 1401 ; ibid * 12 * ( 1975 ) 350 .
tagirov , ann .
, ny * 76 * ( 1973 ) 561 .
p. candelas and d. raine , phys . rev . * d12 * ( 1975 ) 965 .
dowker and r. critchley , phys . rev .
* d13 * ( 1976 ) 224 .
dowker , phys . rev . * 18 * ( 1978 ) 1856 .
dowker , j. phys . * a10 * ( 1977 ) 115 .
dealvis and n. ohta , phys . rev .
* d52 * ( 1995 ) 3529 .
demers , r. lafrance and r.c .
myers , phys .
* d52 * ( 1995 ) 2245 .
solodukhin , phys . rev .
* d53 * ( 1996 ) 824 .
frolov , d.v .
fursaev and a.i .
zelnikov , nucl . phys . *
b486 * ( 1997 ) 339 .
b.s.kay and r.m.wald , phys . rep .
* 207*(2 ) ( 1991 ) 49 .
fursaev and g. miele , class .
quantum grav .
* 12 * ( 1995 ) 393 .
s. coleman and e. weinberg , phys . rev . *
d7 * ( 1973 ) 1888 . s.w . hawking , commun .
phys * 55 * ( 1977 ) 133 .
g. cognola , k. kirsten and l. vanzo , phys . rev .
* d49 * ( 1994 ) 1029 .
fursaev , phys . lett .
* b334 * ( 1994 ) 53 .
l. susskind and j. uglum , phys .
* d50 * ( 1994 ) 2700 .
solodukhin , phys . rev . * d51 * ( 1995 ) 609
fursaev , mod .
. lett . * a10 * ( 1995 ) 649 .
fursaev and s.n .
solodukhin , phys .
* b365 * ( 1996 ) 51 .
d. kabat , nucl . phys .
* b453 * ( 1995 ) 281 .
fursaev and g. miele , nucl . phys . * b484 * ( 1997 ) 697 .
birrell and p.c.w .
davies , _ quantum fields in curved space _ , cambridge university press , cambridge 1982 .
m. abramowitz and i.a .
stegun , _ handbook of mathematical functions _ ( dover publications , inc . , new york ) . | the one - loop effective potential for gauge models in static de sitter space at finite temperatures is computed by means of the @xmath0function method .
we found a simple relation which links the effective potentials of gauge and scalar fields at all temperatures . in the de sitter invariant and zero - temperature states the potential for the scalar electrodynamics
is explicitly obtained , and its properties in these two vacua are compared . in this theory
the two states are shown to behave similarly in the regimes of very large and very small radii @xmath1 of the background space .
for the gauge symmetry broken in the flat limit ( @xmath2 ) there is a critical value of @xmath1 for which the symmetry is restored in both quantum states .
+ moreover , the phase transitions which occur at large or at small @xmath1 are of the first or of the second order , respectively , regardless the vacuum considered . the analytical and numerical analysis of the critical parameters of the above theory is performed .
we also established a class of models for which the kind of phase transition occurring depends on the choice of the vacuum . -48pt
+ * finite temperature effective potential for gauge models in de sitter space * * lara de nardo@xmath3 , dmitri v. fursaev*@xmath4 and * gennaro miele*@xmath3 _ @xmath3 dipartimento di scienze fisiche , universit di napoli - federico ii - , and infn + sezione di napoli , mostra doltremare pad . 20 , 80125 , napoli , italy _ _ @xmath5 theoretical physics institute , department of physics , university of alberta , + edmonton , canada t6 g 2j1 _ _ @xmath6 joint institute for nuclear research , bogoliubov laboratory of theoretical physics , + dubna moscow region , russia _ _
pacs number(s ) : 04.62.+v ; 11.10.wx ; 05.70.fh;_ e - mail : [email protected] ; [email protected] ; [email protected] = .6 cm |
running current through a sample will create electromagnetic fields in the vicinity of the sample and these can deflect the photo - electrons on the way to the analyzer
. a full analysis of the electrons trajectories in the vacuum chamber can be quite complicated and so we have estimated the contributions of the magnetic and electric fields separately . for the magnetic field we calculated the defalcation of an electron leaving the sample in normal emission , at the analyzer entrance slit .
we calculated the field of a slab 50@xmath4 m wide , 200 nm thick and 2 mm long carrying 10ma , which is the highest current we used in the experiments .
we ignored the return current reduces the magnetic field .
we found that the deflection under such conditions is 0.8deg for an electron with 2.1ev kinetic energy .
this is less than 15% of the deflection we found in the experiment for the sample with the smallest deflection ( largest superfluid stiffness ) . estimating the effect of the electric field created by the current is more difficult .
there are no electric fields within the sc ; fields can only develop across the vortex channels where there are sharp changes in the chemical potential .
we used simion to calculate the electron trajectories in the case of one such channel 1@xmath4 wide .
we assumed the voltage across the channel is 300mev , larger than any voltage drop in our experiments .
simion solves the electron s equations of motion using specific boundary conditions .
we used the real dimensions of the sample , chamber and electron analyzer .
the maximal deflection for 2.1ev electrons was 0.6deg , much smaller than the deflections we observed in the experiment . to further test the direct effect of the current on the photo - electrons we conducted the following experiment : we prepared an optimally doped film and evaporated the 4 au contacts on it . we did not , however , patterned it into a narrow bridge .
the cross section of the film was1000 larger than the cross section of the narrow bridge so we do not expect any momentum shift due to the low current density .
we ran current through the film up to 60ma .
the contact resistance was such that at 60ma the voltage that developed on the current contact was about 250mev at 60ma .
the results are shown in fig .
[ film ] . in the upper part of the figure we show the raw detector images for zero current and for @xmath3560ma .
there is a shift in the dispersion when we plot the emission angle vs. kinetic energy .
when the angle is translated to momentum , however , no change is seen in k@xmath9 of the node as a function of the current as shown in lower part of fig .
the reason for the `` disappearing '' of the angular shift is the large shift in the kinetic energy .
all our measurements were done in the flux - flow regime where the samples are dissipative .
the maximal power supplied during the arpes measurements was 10ma @xmath5 500mv = 5mw , this is relatively low power compared with the cooling power of the lhe cryostat , which is well above 1w at these temperatures .
nevertheless , heat is transferred to the cold - finger through the insulating substrate , and temperature gradients can develop . since we can run a real four - terminal measurement of the iv curve of the sample during the arpes measurements , we know the exact temperature of the bridge during the experiments . unlike a thermometer , which measures the temperature of the sample holder
, the iv curve gives an accurate measurement of the temperature of the bi2212 film itself .
we did not find any significant change in the temperature during our measurements .
[ t_scan ] presents data for a underdoped ( t@xmath0=73k ) bridge measured at several temperatures .
these measurements were done at different currents chosen such that the voltage across the bridge was 100mev . the split could be observed only below t@xmath0 , where even at t=73k@xmath7t@xmath0 a small secondary dispersion can be observed .
several interesting things can be learned from this , the energy difference between the dispersions is much smaller although the voltage across the bridge is the same , 100mev .
the reason for this is that most of the resistance at t@xmath0 is not related to flux - flow but rather to normal parts of the sample . above t@xmath0 ,
the sample is a plain resistor .
the broadening due to the variation of the electro - chemical potential across the bridge is not significant .
the 100mev develops across a 300@xmath4 m bridge , so an emission spot of about 100@xmath4 m will result in a broadening of no more than 30mev , which is consistent with our data . the arpes intensity is proportional to the spectral function that is given by the imaginary part of green s function .
we use the usual bcs green s function to calculate the spectral function : @xmath36 where @xmath37 is the tight - binding dispersion , @xmath38 is the d - wave gap and @xmath39 is a single - particle scattering - rate .
using this approximation , the arpes data may be fit below t@xmath0 @xcite as was shown before .
as shown by franz and millis @xcite a uniform super - current will modifies the green s function as follows : @xmath40 where @xmath26 is current re - normalization parameter .
the doppler shift in d - wave superconductors creates particle and hole pockets . in fig.[ak ] we show the spectral function at the fermi level for the case where @xmath12 is in the ( @xmath41 ) direction .
panel ( a ) presents the spectral function , ignoring the momentum boost ( without taking @xmath42 to @xmath43 ) for @xmath44 . in this case
we get pockets of size q centred around the original nodes . the spectral weight is higher on one side of the pockets and , along the nodal direction in particular , the spectral weight vanishes on the `` dark '' side of the pockets . in panel ( b ) we add the momentum boost , but maintain @xmath44 .
as can be seen , the center of the pockets is shifted by q/2 leaving k@xmath9 of the `` bright '' side of the pockets exactly at the nodes .
panel ( c ) shows the spectral function for @xmath45 ; the result is smaller pockets of size @xmath46 .
since the shift due to the momentum boost remains @xmath47 we do observe a shift in k@xmath9 . | there are several ways to turn a superconductor into a normal conductor : increase the temperature , apply a high magnetic field , or run a large current .
high - t@xmath0 cuprate superconductors are unusual in the sense that experiments suggest that destroying superconductivity by heating the sample to temperatures above t@xmath0 or by applying a high magnetic field result in different normal states .
spectroscopic probes show that above t@xmath0 , in the pseudogap regime , the fermi surface is partly gapped and there are no well - defined quasiparticles .
transport measurements , on the contrary , reveal quantum oscillations in high magnetic fields and at low temperatures , suggesting a more usual fermi liquid state . studying the electronic structure while suppressing superconductivity by using current , will hopefully shed new light on this problem . in type ii superconductors , such as the cuprates , the resistive state created by
the current is a result of vortex motion .
the large dissipation in the flux - flow regime makes spectroscopic measurements in the presence of current challenging .
we performed angle - resolved photoemission experiments in thin films of bi@xmath1sr@xmath1cacu@xmath1o@xmath2 while running high - density current through the samples .
clear evidence was found for non - uniform flux flow , leaving most of the sample volume free of mobile vortices and dissipation .
the super - current changes the electronic spectrum , creating quasiparticle and quasihole pockets .
the size of these pockets as a function of the current is found to be doping dependent ; it depends both on the superfluid stiffness and on the strength of interactions .
the normal state of high - temperature cuprate superconductors ( htsc ) is very unusual . in a simple metallic superconductor ( sc ) ,
superconductivity can be destroyed by either increasing the temperature or applying a high magnetic field ; in both cases the result will be the same metallic normal state .
for htscs , however , warming above t@xmath0 results in a pseudo gap ( pg ) state with no well defined - quasiparticles ( qp ) and with a partially gapped fermi - surface @xcite while applying a large magnetic field seems to result in a metallic state that can be described using the fermi liquid theory @xcite .
there is a third way to destroy a sc ; and that is by driving a high enough current through the sc .
the ability to carry electrical current without dissipation is the hallmark of superconductivity , but at current densities above some critical value , the system becomes dissipative . in type ii superconductors , such as the cuprates ,
there are at least two different critical current densities : the lower critical current density is associated with dissipation due to magnetic flux motion , while at a much higher current density one expects the cooper pairs to break due to the current and the sample to become normal . in the case of the cuprates , due to their low superfluid density , there may be a third kind of critical current that is set by the current density at which the superfluid stiffness vanishes @xcite .
the result is a state with finite pairing but with no phase stiffness .
a similar state was suggested to describe the behaviour in the pg state above t@xmath0 , where thermal fluctuation may eliminate the superfluid stiffness without destroying the pairing @xcite .
it will be very interesting to compare the electronic spectrum in these two states .
a study describing an attempt to destroy phase coherence using current was recently published @xcite . in this letter
we present a quantitive estimate of the effect of a super - current on the spectrum of bi@xmath1sr@xmath1cacu@xmath1o@xmath2 ( bi2212 ) by performing arpes measurements in the presence of electrical current in the flux - flow regime .
to have good control over the dissipation in the sample during the experiment while running high current densities , we needed to design a suitable sample .
we chose to work with bi2212 thin films grown on laalo@xmath3 substrates , and so for the purpose of this experiment , we prepared 200nm - thick bi2212 films using dc sputtering . using deep uv lithography , we patterned the films into narrow bridges , 50 @xmath4 m in width and 300@xmath4 m in length ( shown in the inset of fig .
[ fig1 ] ) .
we evaporated four au contacts onto the film ; the outer contacts served as the current contacts while the inner contacts enabled accurate measurement of the voltage drop across the bridge as the current was increased ( shown also in the inset of fig .
[ fig1 ] ) .
the four - contact configuration is essential for several reasons : ( _ i _ ) it enables exact measurement of the critical current , ( _ ii _ ) it enables exact measurement of the power that is dissipated in the bridge and , maybe most important , ( _ iii _ ) it provides a simple way to measure the sample temperature and assure it remains superconducting .
fig.[fig1 ] presents a typical current - voltage ( iv ) curve for one of our bridges . at very low currents , no voltage develops across the bridge .
once the current is increased beyond the de - pining current , magnetic flux flows and there is voltage across the bridge .
all the arpes data were collected in the flux - flow regime . further increasing the current eventually leads to instability and to a thermal run - away .
a systematic measurement of the doping and temperature dependence of the critical current for the same bi2212 bridges was published recently @xcite .
the samples were cleaved in the arpes uhv - chamber at a pressure lower than 1 @xmath5 10@xmath6 torr . to make sure that electrons are emitted only from the bridge area , where the current density is high
, we used a very fine needle that was glued onto the bridge . using these fine needles
, we were able to cleave only the bridge without exposing any parts of the electrodes .
we used a uv - laser as our photon source .
photon energy was set to 6.52ev and the laser was focused to a @xmath7100@xmath4 m diameter spot .
the advantage of using low - energy photons is the improved momentum resolution obtained ; the drawback is the limitation in momentum range that can be mapped .
our setup enabled the measurement of the nodal region only .
each sample was measured at a temperature at which the de - pinning current was no more than 2ma .
fig.[fig2 ] presents the arpes spectra of a bridge for three cases : no current , positive current of 4.8ma and negative current of 4.8ma .
very unusual behaviour is displayed : instead of the expected energy broadening , the spectra show an energy split , with no notable energy broadening .
although the definition of the fermi level for each dispersion is unclear , a sharp edge is clearly discernable .
the dispersion that is shifted less with respect to the no - current dispersion is labeled `` main '' , and the other dispersion is labeled `` secondary '' . when the current is reversed , the main and secondary dispersions exchange position .
the energy difference between the two dispersions is about 100mev .
in addition , the main dispersion is shifted slightly , by a few mev , relative to the fermi level in the absence of current .
4.8ma of current . in the presence of current ,
the spectrum is split into two distinct dispersions , which are shifted in energy .
the dispersion that is shifted less with respect to the no - current dispersion is labeled `` main '' , and the other dispersion is labeled `` secondary''.,width=340 ] next , we show data from a different sample as a function of current .
the upper panel of fig .
[ fig3 ] shows the arpes spectra for several current densities . in this case
the spectra are split into three parts .
the energy difference between the different dispersions grows with the current density .
the black symbols in the lower panel of fig .
[ fig3 ] represent the energy difference between the main dispersion and the secondary dispersion for which the energy difference is maximal , as a function of the current .
the red line is the iv curve measured during the arpes measurement .
a perfect agreement is observed between the energy shift and the voltage drop across the bridge .
this agreement suggests the following : for current densities larger than the de - pinning current density , vortices move through the bridge creating a voltage drop , but the flux flow is not uniform . instead
, the movement of vortices is restricted to narrow regions or channels .
all of the voltage drop occurs in these channels , while most of the sample remains free of dynamic vortices and , as a result , free of phase fluctuations . in our experiment ,
the large majority of electrons are emitted from these parts of the sample , as the flux channels occupy a negligible volume . in this flux - flow configuration ,
the chemical potential does not change linearly along the current direction but instead forms sharp steps and regions of constant chemical potential .
the chemical potential was measured at both ends of the bridge by measuring the fermi edge of the au contacts .
the shift of the fermi edge of the first contact with respect to the ground increases linearly with the current and is a result of contact resistance .
this energy shift is identical to the energy shift of the main dispersion .
the shift of the fermi edge of the second contact equals the shift due the contact resistance plus the shift due to the voltage drop across the bridge due to the flux - flow .
we can fully reproduce the iv measurements using the photoemission from the au contacts .
all of our sample exhibited the same behavior : some of the samples displayed two dispersions , other had three .
in some samples there was no energy split , but even then there was no energy broadening ; in those cases there is no flux flow in the cleaved area of the bridge but rather in other parts .
we do not believe that the splitting of the spectra is unique to our samples .
indeed , a large body of literature exists that studied the morphology of the flux flow using many different techniques in many different samples .
it seems that non - uniform flux flow is the rule and not the exception @xcite .
the split in the spectra means that all phase fluctuations induced by the current are concentrated in a small part of the sample that was not probe in our arpes experiment .
these conditions create an opportunity to study scs in the presence of very high current densities without any noticeable flux motion .
next , we describe the changes in the spectra as a result of the current . direction .
( b ) shift of the mdc peak for the same current values .
the red point represents the node at zero current and the black line is the femi - surface from a tight - binding calculation .
, width=340 ] fig .
[ fig4](a ) presents the momentum distribution curves ( mdcs ) for several current values obtained at an energy that corresponds to the position of the intensity edge of the main dispersion .
these data were obtained along the nodal direction .
it is clear to see that the peak in mdcs shifts away from the nodal point as the current increases .
we also observe some broadening of the mdc peaks .
the momentum shift in the main and secondary dispersions is identical , within our experimental error .
these momentum shifts are quite significant , in fig.[fig4](b ) we show the position of the mdc peak for different current values . for a current of about 10ma the momentum shift is about 10% of k@xmath8 .
since current is being run through the sample , thus inducing a potential difference across the sample , we expect to have stray magnetic and electric fields in the vicinity of the sample . in the supplementary material section
, we estimate these stray fields and show that their effect on the electron trajectories can account for no more than 15% of the momentum shifts we observe .
moreover , a simple calculation shows that the magnetic field created by the current will deflect the electrons in the opposite direction to the direction found in the experiment .
we therefore argue that most of the momentum shift is a result of the changes in the spectrum induced by the super - current .
next we estimated the current density .
since the samples are cleaved , we would not use the cross - section of the bridge to calculate the current density , and so instead we used the critical current as a gauge .
we measured carefully the doping and temperature dependence of the critical - current of a set of bridges for which we measured the cross section @xcite .
since the de - pining current density is a fundamental property of the material , we assume that the cleaved bridges have the same critical current - density , which is measured in the arpes chamber after cleaving . for currents larger than the de - pining current , the current flows uniformly through the narrow bridge @xcite .
fig .
[ fig5 ] shows the shift in k@xmath9 as a function of the current density for 4 different samples .
the shift is in the direction of the current .
what is the origin of the observed momentum shift ? in the presence of a super - current the quasiparticle spectrum is doppler shifted in a unique way @xcite .
the bogoliubov spectrum is modified : @xmath10 where @xmath11 is the single particle dispersion .
the doppler shift is momentum dependent and is proportional to the super - current wave vector , @xmath12 , which represents the center - of - mass momentum of the pairs .
@xmath13 is proportional to the current density . for a layered material such as bi2212 it makes sense to use the 2d expression : @xmath14 where @xmath15 is the current density and @xmath16 is the superfluid stiffness of the sc . for a d - wave sc , such as bi2212
, the doppler shift will create particle and hole pockets around the nodal points depending on the current direction @xcite .
these pockets can be measured indirectly by measuring the magnetic - field dependence of the specific - heat of a d - wave sc .
the magnetic field penetrates the sc in the form of a vortex lattice .
the super - current circulating the vortex cores shifts the quasiparticles , creating pockets that contribute to the specific heat , a term that is proportional to @xmath17 @xcite .
this dependence , known as the volovik effect , was measured in the cuprates @xcite . for small values of @xmath13 along the nodal direction
, we can rewrite eq .
[ bdg ] as : @xmath18 where @xmath19 is the fermi velocity along the nodal direction .
the size of the pocket created by the current along the nodal direction is q. eqs .
[ q_def ] and [ nodal_shift ] show that for a given current density , the size of the pocket is inversely proportional to the superfluid density , which depends on both the doping and the temperature .
it is tempting to think of the momentum shift we find as a direct measurement of the pocket size .
the slope of a linear fit to the data in fig .
[ fig5 ] is given by @xmath20 where @xmath21 is the c - axis lattice parameter of bi2212 .
the penetration depth , @xmath22 , can now be calculated using the expression for the 2d superfluid stiffness : @xmath23 the values we obtained are shown in the legend of fig.[fig5 ] .
these values are of the same order of magnitude as values found in measurements of the absolute value of the penetration depth of bi2212 using different techniques @xcite .
since the arpes intensity is proportional to the spectral function , a better understating of the data requires that we consider the effect of the super - current on the spectral function .
it was shown @xcite that for small current densities and keeping only terms that are linear in q , the spectral function in the presence of current can be written as @xmath24 , where @xmath25 is the spectral function without current .
we see that in addition to the doppler shift there is also a momentum boost , which completely cancels out the doppler shift along the nodal direction , and so we should not expect a momentum shift at all along that direction @xcite .
a possible explanation for the observed shift are the interactions , which should not be neglected .
interactions can re - normalize the current carried by the quasiparticles by a factor @xmath26 @xcite , and so in eq .
[ nodal_shift ] we should have , in fact , used the re - normalized term @xmath27 to describe the doppler shift
. if @xmath28 , the doppler shift will no longer be cancelled by the momentum boost and k@xmath9 will be shifted .
see the supplementary material for more details .
the london penetration depth as a function of the temperature and the doping level of the same bi2212 films , was measured using a mutual inductance technique @xcite .
we found that the penetration depth extracted from the momentum shift was shorter than the values presented in ref .
@xcite .
the current re - normalization parameter can be estimated from the ratio of these values , @xmath29 .
for the t@xmath0=78k sample we found @xmath30 , for the moderately underdoped samples ( t@xmath0=58k and 62k ) @xmath31 , and for the extremely underdoped sample we find a very small value of @xmath32 was obtained .
the results are in reasonable agreement with theory @xcite and with experimental estimates @xcite .
fig .
[ fig5 ] also reveals that for all of the samples shown , the dependence of the momentum shift dependence on the current density is not exactly linear ; for high current densities it seems that @xmath33 increases , indicating that the superfluid density decreases .
this is in agreement with theory @xcite , which predicts that @xmath34 . in the present experiment
we were not able to reach the critical current density at which @xmath16 is expected to vanish , as this would require increasing the current density by an order of magnitude .
we hope to accomplish this in the future , using smaller bridges .
in addition , all of the data presented is limited to the nodal direction ; we hope in the future to be able to measure entire pockets by extending the momentum range we can cover .
we have shown that arpes in the presence of current can be a useful way to understand the dynamics of flux in htscs and can be used to measure basic properties such as the super - fluid density and the current re - normalizaion parameter .
we hope that our work will encourage others to measure different systems using similar methods .
this work was supported by the bsf , grant no .
2010313 .
we acknowledge fruitful discussions with lilach goren , ehud altman , erez berg , netanel lindner and boris shapira .
99 a. kanigel et .
al .
prl * 101 * 137002 ( 2008 ) .
n doiron - leyraud et al .
nature * 447 * 565 ( 2007 ) .
lilach goren and ehud altman , prb * 79 * , 174509 ( 2009 ) .
l.b .
ioffe , a.j .
millis , j. phys .
chem . of sol . * 63 * , 2259 ( 2002 ) .
v. j. emery and s. a. kivelson , nature * 374 * , 434 ( 1995 ) .
m. franz and a. j. millis , phys .
rev .
b*58 * , 14572 ( 1998 ) .
erez berg and ehud altman , prl * 99 * , 247001 ( 2007 ) .
m. khodas and a. m. tsvelik , prb * 81 * , 094514 ( 2010 ) .
a. kaminski , et al .
, arxiv:1509.07937 .
muntaser naamneh , juan carlos campuzano , and amit kanigel , prb * 90 * , 224501 ( 2014 ) .
g. e. volovik , jetp lett .
* 58 * , 469 ( 1993 ) .
k. a. moler et al .
prb * 55 * , 3954 ( 1997 ) .
huebener r. p. , _ magnetic flux structures in superconductors _ , springer - verlag berlin ( 2010 ) .
zeldov e. et al . , prb * 49 * , 9802 ( 1994 ) .
m. tinkham , _ introduction to superconductivity _ second edition ( dover books on physics ) ( 2004 ) .
a.j .
millis , s.m .
girvin , l.b .
ioffe , and a.i .
larkin , j. phys . and
chem .
solids * 59 * , 1742 ( 1998 ) .
x.g .
wen and p.a .
lee , phys .
rev .
lett . * 80 * , 2193 ( 1998 ) .
p. zimmermann _
et al._phys .
rev . *
b52 * , 541 ( 1995 ) . o. waldmann _
et al._phys . rev . * b53 * , 11825 ( 1996 ) .
w. anukool _
et al._phys . rev . *
b80 * , 024516 ( 2009 ) .
jie yong , mj hinton , andy mccray , m randeria , m naamneh , a kanigel , tr lemberger , phys .rev .
* b85 * , 180507 ( 2012 ) . c. p. nave , d.a . ivanov and p.a .
lee , phys . rev . *
b73 * , 104502 ( 2006 ) .
j. mesot , m. r. norman , h. ding , m. randeria , j. c. campuzano , a. paramekanti , h. m. fretwell , a. kaminski , t. takeuchi , t. yokoya , t. sato , t. takahashi , t. mochiku , and k. kadowaki , phys .
rev .
lett * 83 * , 840 ( 1999 )
. may chiao , r. w. hill , christian lupien , louis taillefer , p. lambert , r. gagnon , and p. fournier , phys .
rev . * 62 * , 3554 ( 2000 ) .
m.r .
norman , h. ding , m. randeria and j.c .
campuzano , phys . rev . *
b57 * , r11093 ( 1998 ) . |
the doorway states for one - nucleon transfer reactions are eigenstates of nucleon in the static field of nucleus which is the convolution of the free - space nucleon - nucleon forces with the nucleon density distributions in nucleus @xcite .
they are model - independent quantities because ( a ) the free - space @xmath0 forces are independent of nuclear medium effects and ( b ) the nucleon density distributions are deduced from the electron
nucleus @xcite and proton
nucleus @xcite elastic scattering data .
the corresponding eigenvalue problem is that of nucleon in central field which can be solved with any desired accuracy . for these reasons
the doorway states can be used as a very trustworthy test for current nuclear models .
let us discuss some results of ref.@xcite from this point of view .
the nuclear relativity within the walecka @xcite model is confirmed to be actually existing phenomenon .
the dominant contribution to the isovector part of the static field arises from the many - particle @xmath0 forces because the @xmath1 meson ( vector isovector ) and @xmath2 meson ( scalar isovector ) fields arising from the two - particle forces nearly cancel each other . at the same time
the isovector nuclear potential is exclusively of the @xmath1 meson origin within the quantum hadrodynamics @xcite .
the reason for this wrong qhd result is the neglect of the @xmath2 meson field in spite of the fact that both the @xmath1 and @xmath2 meson exchanges are taken into account in the two - particle @xmath0 forces @xcite . \3 .
the contributions from the two - particle , three - particle and four - particle forces to the isoscalar part of the static field are found to be @xmath3mev , @xmath4mev , @xmath5mev .
this is in conflict with such leading principles of the effective field theory as the naive dimensional analysis and the naturalness @xcite .
indeed , the values of the forces which are estimated according to the above principles are @xcite @xmath6 ( @xmath7 is the mass of nucleon ) and therefore the expected relation between @xmath8 , @xmath9 and @xmath10 is @xmath11 .
in fact it is @xmath12 thus suggesting that there is something wrong with the naturalness .
discussion of this point is continued in the next section .
as discussed in ref.@xcite the isoscalar part of the static field may contain contributions from higher ( five - particle , six - particle , etc . ) many - particle forces .
they could be taken into account by increasing the number of terms in the power series expansion @xmath13 for the static field @xcite
( @xmath14 is the nucleon density distribution ) thus introducing an indeterminate number of additional adjustable parameters .
instead we use the fact that ultimately the underlying reason for many - particle forces is the nonlinearity of strong interaction .
we introduce an auxiliary scalar - isoscalar field @xmath15 with the lagrangian density @xmath16 thus obeying the following equation : @xmath17 @xmath18 is the nuclear scalar density . as discussed in ref.@xcite the many - particle forces of all ranks
are taken into account in this way .
the field @xmath15 contains the `` two - particle '' component @xmath19 obeying the equation @xmath20 this component must be eliminated because the `` two - particle '' contribution to the static field is determined by the free - space two - particle forces @xcite .
so the many - particle contribution to the scalar
isoscalar field is @xmath21 let us analyse this expression disregarding for a moment the laplace terms of eqs .
( 5 ) and ( 6 ) which are responsible for the finite range of the forces , although these terms are included in the actual calculations . but
as demonstrated in ref.@xcite they are of little importance thus not affecting the results of the below analysis .
as demonstrated in ref.@xcite the radial dependence of @xmath22 has the form which is schematically shown in fig.1 . as seen from the figure it is negative at @xmath23 , positive at @xmath24 with a maximum @xmath25 in this region , and vanishing at @xmath26 . without the laplace terms
@xmath27 so @xmath28 but as follows from ( 6 ) and ( 7 ) @xmath29 and therefore @xmath30 where @xmath31 .
let us introduce the dimensionless quantities @xmath32 and @xmath33 , @xmath34 in these units @xmath35 the maximum @xmath25 occurs at @xmath36 , so @xmath37 the parameter @xmath38 is negative since @xmath39 , see the figure .
the parameter @xmath40 is negative too provided the coupling constant @xmath41 is positive .
actually the sign of @xmath41 is insignificant since the physical field is @xmath42 thus being expressed through @xmath43 .
the potential energy of the scalar field is also of the form ( 4 ) within the relativistic mean - field approach @xcite , the parameters @xmath40 and @xmath38 being negative too . in this way the sign of the rmf parameters is confirmed .
it should be mentioned that the values of the rmf parameters are determined from the experimental data which include the important correlation effects ( binding energies , density distributions , low - energy spectra etc . ) , and therefore they are model - dependent ( the model - independent treatment of the correlations does not exist ) . in contrast , our parameters are determined from the doorway state energies thus being model - independent . in terms of the @xmath44 and @xmath45 quantities the contribution to the scalar
isoscalar field from the many - particle forces is @xmath46 the second term in the rhs of ( 15 ) arises from the symmetry energy .
the quantities @xmath32 and @xmath33 obey the equations @xmath47 the nda prescription @xcite for the scalar field potential energy is @xcite @xmath49 where @xmath50 mev . according to the concept of naturalness all the coefficients @xmath51
must be of order of unity .
comparison between ( 20 ) and ( 4 ) together with ( 14 ) and ( 16 ) gives @xmath52 as follows from the values ( 19 ) of the parameters @xmath53 , @xmath54 , the concept of naturalness thus being not confirmed . as demonstrated by the calculations for the few - nucleon systems
the effect of many - particle forces is relatively small @xcite .
this result is confirmed but the underlying physical reason is different from that provided by the effective field theory . according to the latter it is decrease of the strength with increasing rank of the force , see eq.(1 ) .
as follows from above this scenario does not hold : the actual reason is the cancellation of the contributions from many - particle forces of different ranks ( the physics is believed to be the same for complex nuclei and few - nucleon systems ) . | the naturalness concept of the effective field theory is not confirmed . |
quantum computers are believed to be more powerful than their classical counterpart , resulting in tremendous efforts to implement quantum computation with different physical systems .
the model of one - way quantum computer @xcite has opened a novel approach towards the possible experimental realization of quantum computation . in the one - way model ,
quantum computation starts from preparing certain universal resource states , namely the cluster states @xcite , and is achieved by merely performing single - qubit measurements on these states .
the principal task turns out to be the preparation of these highly entangled resource states .
one straightforward way is to apply entangling gates to couple a lattice of qubits @xcite .
it is however clearly appealing if there exists a universal resource as the unique ground state of a naturally occurring gapped two - body local hamiltonian , for the advantage of flexible state preparation as well as the stability against the local perturbations .
recall that the cluster states can not occur as a non - degenerate ground state of any two - local spin-@xmath1 hamiltonian @xcite , and have singular entanglement features , e.g. vanishing two - point correlation functions @xcite .
perturbative hamiltonians have been proposed with ( encoded ) graph states as the approximate ground states @xcite , which however require a highly precise control over system parameters and the spectral gap would get significantly smaller according to the order of the perturbation .
recently , a gapped two - body hamiltonian of six - level particles @xcite has been constructed with the so called tri - cluster state as its unique ground state , which is universal for measurement - based quantum computation . here , we construct a new type of gapped local hamiltonian , with the constitutional two - body affleck - kennedy - lieb - tasaki ( aklt ) @xcite and exchange - type interactions acting on four - level particles .
the simplicity of the hamiltonian might allow one to identify possible realizations in models of condensed matter systems or in quantum optical set - ups .
the present approach differs from the previously used methods to construct a parent hamiltonian for projected entangled pair states by finding the local support subspace @xcite , but can be utilized to construct a family of local hamiltonians with the same properties . in this paper , motivated by the usage of the gapped ground state of the 1d aklt spin-1 chain as a universal quantum wire @xcite ( cf.@xcite )
, we start from 1d _ quasi - chain _ of spin @xmath0 particles as a building block .
from there we construct a full 2d resource that allows a deterministic decoupling of the 1d quantum wire structure ( in the similar way as the cluster state ) . after proving that the 1d quasi - chain is gapped with an unique ground state
, we construct a gapped two - body hamiltonian on an octagonal 2d lattice , which is transitionally invariant , and consists of only nearest neighbor aklt- together with exchange - type interactions .
we demonstrate how to implement universal quantum computation by simply making single - spin measurements on the individual four - level particles of its unique ground state .
this hamiltonian provides the example that can be utilized as a complete measurement - based ground - code quantum computer without the demand of dynamical coupling @xcite .
extensions of our approach to the other geometric configurations , e.g. 2d lattice , or other hamiltonians are also possible .
we first consider the 1d aklt model defined on the quasi - chain as in fig .
[ qcha ] and show that the model is gapped and has a unique ground state .
the quasi - chain consists of spin-@xmath0 ( a ) and spin-@xmath2 particles ( b ) coupled with nearest - neighbor two - body interactions @xmath3 \label{chainhamil}\ ] ] where @xmath4 represents the projector onto the spin - s irreducible representation of the total spin for particles @xmath5 and @xmath6 ( cf .
@xcite ) . as the spin per particle equals to half of the local coordination number ( i.e. the number of the bonds from a particle ) , the 1d aklt quasi - chain in eq .
( [ chainhamil ] ) has an unique ground state @xmath7 @xcite , which can be obtained via a projector which maps the symmetric part of three spin-@xmath1 ( virtual qubits ) of maximally entangled pairs into one spin-@xmath0 physical particle . to prove that the 1d aklt quasi - chain is gapped in the thermodynamic limit , we first regroup the hamiltonian in eq .
( [ chainhamil ] ) into blocks as @xmath8 , see fig [ qcha ] , where @xmath9 is the sum of the aklt interactions among the particles inside the block .
we consider the ground state @xmath10 and denote the support subspace of the reduced density matrix corresponding to each block as @xmath11 , and the projector onto the orthogonal subspace by @xmath12 . from @xmath13
it follows that @xmath7 is also the unique ground state of the projective hamiltonian @xmath14 .
one can directly calculate the energy gap @xmath15 of @xmath16 and finds @xmath17 , which leads to @xmath18 we further write the block hamiltonian from @xmath19 on @xmath20 units as @xmath21 . if @xmath22 , the results in @xcite imply that @xmath23 our calculations show that @xmath24 for @xmath25 .
we thus conclude that @xmath19 is gapped as @xmath26 , where better bounds can in principle be obtained for higher @xmath6 .
therefore , the energy gap of the 1d aklt quasi - chain is lower bounded by @xmath27 .
with the above established results for the 1d aklt quasi - chain , we will show how to obtain a model on an octagonal lattice with two - body interactions that is gapped and has a unique ground state .
we start from a number of independent 1d aklt quasi - chains , and introduce the unitary transformation @xmath28 which maps two spin-@xmath1 particles into one spin-@xmath0 particle by @xmath29 where the labels denote @xmath30 respectively .
notice that in principle _ any _ unitary operation can be used at this stage , each leading to a gapped model with a unique ground state that can be used for universal measurement - based computation . using such a transformation , we can merge a number of 1d quasi - chains and get a 2d hamiltonian as @xmath31 with @xmath32 , where @xmath33 denote two neighboring spin-@xmath1 particles in the same merging circle , see fig .
[ conf2d ] .
thus , @xmath34where @xmath35 , @xmath36 , @xmath37 and @xmath38 , with @xmath39 , @xmath40 , @xmath41 , @xmath42 represent different types of couplings , see fig .
[ conf2d ] .
the coupling between @xmath43 and @xmath44-type particles , @xmath45 are effectively exchange - type interaction with @xmath46 and @xmath47 where @xmath48 is the effective spin-@xmath1 operator defined on two levels @xmath49 .
it is easy to verify that @xmath50 and @xmath51 satisfy the commutation relations analogous to the spin angular momentum .
as @xmath52 is equivalent up to a unitary transformation to @xmath53 independent aklt quasi - chains , the spectrum and the corresponding eigenvalues can be trivially obtained from the spectrum and eigenvalues of @xmath54 . in particular
, it follows that @xmath52 is gapped ( with the same constant energy gap @xmath55 as @xmath54 ) , and has the unique ground state @xmath56 measurement - based quantum computation -------------------------------------
we will demonstrate how to use the resource state @xmath57 for measurement - based quantum computation , in following the notation and scheme of refs .
the state @xmath57 can be represented as a projected entangled pair state @xcite , in which a number of maximally entangled pairs @xmath58 of virtual qubits are mapped into physical particles , see fig .
[ uqc ] for one computational block .
the corresponding tensor matrices for the physical site @xmath59 are @xmath60&=&|1\rangle _ { r}\langle 0|_{l}\otimes \langle 1|_{d}\\ a_{u}[-\frac{3}{2 } ] & = & |0\rangle _ { r}\langle 1|_{l}\otimes \langle 0|_{d}\\ a_{u}[+\frac{1}{2}]&=&- 1/\sqrt{3}(z\otimes \langle 1|_{d}+|1\rangle _ { r}\langle 0|_{l}\otimes \langle 0|_{d})\\ a_{u}[-\frac{1}{2}]&=&1/\sqrt{3 } ( z\otimes \langle 0|_{d}-|0\rangle _ { r}\langle 1|_{l}\otimes \langle 1|_{d})\end{aligned}\ ] ] the local tensor @xmath61 can be written in a similar way .
the tensor matrices for site @xmath44 are @xmath62&=&|0\rangle _ { u}\langle 1|_{d}\\ b[- \frac{3}{2}]&=&-|1\rangle _ { u}\langle 0|_{d}\\ b[+\frac{1}{2}]&=&|1\rangle _ { u}\langle 1|_{d}\\ b[-\frac{1}{2}]&=&-|0\rangle _ { u}\langle 0|_{d}\end{aligned}\ ] ] measurement - based quantum computation on such a resource state can be understood as follows ( we refer the reader to @xcite for details ) : the logical information in carried by the virtual qubit , and the measurement in a certain basis on the physical particle will induce either unitary or readout operators on the virtual qubits according to the above tensor matrices .
each logical virtual qubit can be initialized to @xmath63 or @xmath64 by measuring the left end spin-@xmath1 particle in the @xmath65 basis @xmath66 .
quantum computation then proceeds gradually from left to right by measuring each computational block ( @xmath59 , @xmath61 and @xmath44 ) , see figs .
( [ conf2d],[uqc ] ) . before performing computational measurements on each block
, we need to introduce a _ pre - normalization step _ as follows to equalize the different coefficients ( 1 _ vs. _ @xmath67 ) in the tensor matrices @xmath59 ( @xmath61 ) .
we first apply the local filter operation @xmath68 on sites @xmath59 and @xmath61 with @xmath69 an outcome @xmath70 corresponds to an unsuccessful filter attempt and leads after suitable measurements on sites @xmath71 to the transport of quantum information to the next block up to a local pauli operation @xmath72 or @xmath73 .
the procedure can hence be repeated until it succeeds ( i.e. the outcome @xmath74 occurs ) .
one finds that @xmath75 trials lead to an overall success probability of @xmath76 , i.e. the process is efficient .
the required measurements are given by @xmath77 for site @xmath44 , where @xmath78 for sites @xmath59 and @xmath61 .
in the following , we will assume that the filter operation was successful ( i.e. the result @xmath74 was obtained ) .
we show that a suitable choice of the measurement in @xmath44 allows one to either decouple 1d chains and obtain single - qubit operations or read - out , or alternatively to couple two chains directly and obtain a two - qubit gate .
two chains are decoupled by a measurement of site @xmath44 in the @xmath65 basis @xmath79 , leading to vertical virtual qubits @xmath80 and @xmath81 corresponding to sites @xmath82 respectively ( see fig . [ uqc ] ) to be prepared into either @xmath83 or latexmath:[$
@xmath83 , the effective tensor matrices for @xmath59 is given by @xmath85&=&0 , a[+\frac{1}{2}]=-\vert 1\rangle _ { r}\langle 0\vert _ { l}\\a[-\frac{1}{2}]&=&z , a[-\frac{3}{2}]=\vert 0\rangle _ { r}\langle 1\vert _ { l}\end{aligned}\ ] ] for other outcomes and the tensor @xmath61 , the reduced tensor matrices are equivalent up to a local basis change . without loss of generality
, we use the effective 1d tensor matrices @xmath43 to show how to implement arbitrary single - qubit rotations following a similar protocol as in @xcite ( note that the effective tensor matrices are not equivalent to the 1d aklt spin-1 chain ) , and read out logical quantum information .
the readout is realized by simply measuring site @xmath43 in the @xmath65 basis @xmath86 .
once we get the outcome @xmath87 , one can infer that the logical qubit is @xmath88 .
otherwise , the logical qubit gets a @xmath89 by - product operator and we repeat the above procedure .
an arbitrary single - qubit operation can be decomposed into three rotations around the @xmath89 and @xmath72 axes with three euler angles . in order to implement a @xmath89 rotation @xmath90 , we measure site @xmath43 in the basis @xmath91
the first outcome is not possible , while the fourth induce a by - product operator @xmath89 .
if however the outcome is the second or third one , we implement the rotation @xmath92 with a pauli by - product operator @xmath72 or @xmath93 .
an @xmath72-rotation @xmath94 is implemented in a similar way .
we apply the local filter operation @xmath68 at the initial pre - normalization step in the @xmath95 basis @xmath96 .
@xmath72 rotations can then be realized by the same protocol as the @xmath89 rotations , only with the exchange of @xmath97 and @xmath95 basis , where the measurement basis on site @xmath44 at the pre - normalization and decoupling step correspondingly changes as @xmath98 .
the basic idea for the implementation of an entangling gate is to prepare the vertical virtual qubits @xmath99 corresponding to sites @xmath59 and @xmath61 , into an entangled state by a suitable measurement of @xmath44 . in order to determine the kind of two - qubit gate that is implemented for different measurement outcomes , we first rewrite the tensor matrices @xmath100 in the aforementioned basis of @xmath101 .
we further define another basis for site @xmath44 as @xmath102
the contracted tensor from @xmath103 $ ] , @xmath104 $ ] and @xmath105 $ ] , results in one of the two - qubit entangling gates @xmath106 where @xmath107 or @xmath108 .
moreover , any pauli by - product operator can propagate through the above entangling gates as @xmath109 , where both @xmath110 and @xmath111 are pauli operators .
the explicit procedure to implement a specific entangling gate is as follows : we first measure site @xmath59 and @xmath61 in the basis @xmath112 ; if the outcomes correspond to the desired entangling gate out of @xmath113 , we proceed to measure site @xmath44 in the basis @xmath114 . otherwise if we do not obtain the desired outcomes from the measurements on site @xmath59 and @xmath61 , we measure site @xmath44 in the @xmath65 basis @xmath115 , which decouples the chains and leads to transport of quantum information to the next block up to a by - product pauli operator
. in this case , we have to repeat the above procedure in order to obtain the target two - qubit gate , but an arbitrary high success probability is achievable efficiently as well .
the present protocol offers the flexibility to choose a two - qubit gate on demand from a set of entangling gates .
if the bulk hamiltonian is maintained during the measurement - based computation as proposed in the ground - code scheme @xcite , the spectral gap appears to provide certain protection again local noises , so that it may make the computer more robust and easier to meet a stringent fault - tolerant error threshold for quantum error correction .
however , potential advantages of the gap in protecting quantum information is currently intensively being studied in the context of the topological memory @xcite . the general question as to
what extent the passive hamiltonian protection is helpful in measurement - based computation , which is far from the equilibrium , is more involved , and addressed elsewhere .
below , we describe the complete scheme with the hamiltonian present .
we first look at the residual hamiltonian of 1d aklt quasi - chain after measuring the first @xmath116 particles , @xmath117\ ] ] it is gapped and twofold degenerate , which can encode one logical qubit .
one can show that the operators @xmath118^{(a_{i})}\otimes\sigma^{(b_{i})}\}\otimes \sigma^{(b_{n+1})}\ ] ] with @xmath119 form the representation of @xmath120 , and the degenerate ground states are connected only by these non - local operators .
the computation on the 2d resource equips similar robustness , as the hamiltonian of the 2d model is locally unitary equivalent to @xmath53 independent chains . to utilize the gap protection
, one needs to turn off the interactions that couple the computational block ( see fig . [ uqc ] ) to the bulk together with those inside the block , prior to any measurement for this block . as the hamiltonian @xmath52 is frustration - free ,
this can be done in a constant time .
also particles already measured need to stay decoupled from the remaining bulk . in a potential implementation of @xmath52 with trapped polar molecules in an optical lattice @xcite
, nearest - neighbor interactions can be turned off by changing the potential depth of local wells which in turn suppresses the tunneling rate between two neighboring wells . an alternative method without turning off interactions
is to apply fast measurements and remove the particle from the system after the measurement , or drive it to a dark state ( which does not interact with neighboring particles anymore ) .
our approach can be extended from the octagonal lattice to the 2d square lattice , for instance .
the corresponding 1d aklt quasi - chain consists of spin-@xmath121 particles , each of which is connected with two spin-@xmath1 particles . in a similar way , we can show that the energy gap is lower bounded by @xmath122 with @xmath123 and @xmath124 .
we can merge a number of such 1d aklt quasi - chains into a 2d resource state as well , and the computational protocol is similar .
we have proposed a novel translational - invariant gapped hamiltonian of spin-@xmath0 particles with nearest - neighbor two - body aklt- and exchange - type interactions .
its unique ground state is proved to be universal for measurement - based quantum computation .
the hamiltonian inherits important properties from the original aklt model , while at the same time , it has distinct features , e.g. a strictly proved energy gap . further study on such a hamiltonian and
its order parameter might reveal new aspects of many - body physics regarding computational capability .
j.m.c . , w.d . , h.j.b .
thank m. van den nest for helpful discussions and acknowledge the support from the fwf ( sfb - foqus , j.m.c . through the lise meitner program ) and the european union ( qics , scala , namequam ) .
a.m. acknowledges the support by the government of canada through industry canada and by ontario - mri .
r. alicki , m. fannes , and m. horodecki , _ j. phys .
a _ * 42 * , 065303 ( 2009 ) ; s. chesi , d. loss , s. bravyi , and b.m .
terhal , arxiv:0907.2807 ; f. pastawski , a. kay , n. schuch , and j.i .
cirac , arxiv:0911.3843 ; s. bravyi , m. hastings , and s. michalakis , arxiv:1001.0344 . | we show that a local hamiltonian of spin-@xmath0 particles with only two - body nearest - neighbor affleck - kennedy - lieb - tasaki and exchange - type interactions has an unique ground state , which can be used to implement universal quantum computation merely with single - spin measurements .
we prove that the hamiltonian is gapped , independent of the system size .
our result provides a further step towards utilizing systems with condensed matter - type interactions for measurement - based quantum computation . |
we say that a graph @xmath4 embedded in a @xmath5-manifold @xmath0 is _ achiral _ , if there is an orientation reversing homeomorphism @xmath6 of @xmath0 leaving @xmath4 setwise invariant .
if such an @xmath6 exists , we say that it is a homeomorphism of the pair @xmath7 .
if no such homeomorphism exists , we say that the embedded graph @xmath4 is _
chiral_. we can think of a knot as an embedding of a circular graph with some number of vertices in @xmath8 .
independent of the number of vertices on the graph , some embeddings of it are chiral while others are achiral .
by contrast , there exist graphs which have the property that all of their embeddings in @xmath8 are chiral .
such a graph is said to be _ intrinsically chiral _ in @xmath8 because its chirality depends only on the intrinsic structure of the graph and not on the extrinsic topology of an embedding of the graph .
the following theorem provides a method of constructing graphs that are intrinsically chiral in @xmath8 .
@xcite [ ic ] every non - planar graph @xmath2 with no order @xmath9 automorphism is intrinsically chiral in @xmath8 .
since chirality is defined for graphs embedded in any 3-manifold , it is natural to ask whether a graph which is intrinsically chiral in @xmath8 would necessarily be intrinsically chiral in other 3-manifolds .
our first result shows that no graph can be intrinsically chiral in every 3-manifold , or even in every `` nice '' @xmath5-manifold . for every graph @xmath2
, there are infinitely many closed , connected , orientable , irreducible 3-manifolds @xmath0 such that some embedding of @xmath2 in @xmath0 is pointwise fixed by an orientation reversing involution of @xmath0 .
theorem [ theorem1 ] can be thought of as a generalization of the fact that every planar graph has an embedding in @xmath8 which is pointwise fixed by a reflection of @xmath8 .
on the other hand , our main result is a generalization of theorem [ ic ] , which shows that for any given `` nice '' 3-manifold @xmath0 , there are infinitely many graphs with no automorphism of order @xmath9 are intrinsically chiral in @xmath0 . in particular , a graph @xmath2 is said to be _
@xmath5-connected _ , if at least three vertices together with the edges containing them must be removed to disconnect @xmath2 or reduce @xmath2 to a single vertex .
a graph @xmath10 is said to be a _
minor _ of a graph @xmath11 , if @xmath10 can be obtained from @xmath11 by deleting and/or contracting some number of edges .
our main result is as follows .
[ icmanifolds ] for every closed , connected , orientable , irreducible 3-manifold @xmath0 , there is an integer @xmath1 such that any graph with no order @xmath9 automorphism which has a 3-connected minor @xmath12 with @xmath13 is intrinsically chiral in @xmath0 .
note that by contrast with theorem [ icmanifolds ] which is for @xmath5-manifolds without boundary , ikeda @xcite has shown in the theorem below that for `` nice '' 3-manifolds with aspherical boundary , any graph with large enough genus which has a certain type of automorphism of order @xmath9 has an achiral hyperbolic embedding in the double of the manifold .
@xcite let @xmath0 be a compact , connected , orientable , 3-manifold with non - empty aspherical boundary .
then there is an integer @xmath3 such that for any abstract graph @xmath12 with @xmath13 and no vertices of valence 1 , any automorphism of order @xmath9 of @xmath12 that does not restrict to an orientation preserving automorphism of a cycle in @xmath12 can be induced by an orientation reversing involution of some hyperbolic embedding of @xmath12 in the double of @xmath0 .
since the word `` graph '' is used in different ways by different authors , we note here that in the current paper as well as in @xcite a _ graph _ is defined to be a connected @xmath14-complex consisting of a finite number of vertices and edges such that every edge has two distinct vertices and there is at most one edge with a given pair of vertices . in section 2 ,
we prove theorem [ theorem1 ] . in section 3 , we prove theorem [ icmanifolds ] making use of a proposition , which we then prove in section 4 . throughout the paper we work in the smooth category .
the goal of this section is to prove theorem [ theorem1 ] . to that end , we prove the following proposition . note that we use @xmath15 to denote the dimension of the first @xmath16-homology group of @xmath0 and we use @xmath17 to denote the dimension of the first @xmath18-homology group of @xmath0 .
let @xmath19 be a closed , orientable surface . then for infinitely many closed , connected , orientable , irreducible 3-manifolds @xmath20 such that @xmath21 , there is an embedding of @xmath19 in @xmath20 which is pointwise fixed by an orientation reversing involution of @xmath20 .
[ prop : inf ] the proof of proposition [ prop : inf ] will make use of the idea of a _ disk - busting _ curve , which is a simple closed curve in a handlebody that intersects every essential , properly embedded disk in the handlebody .
for example , a core of a solid torus is disk busting in the solid torus . for a genus 2 handlebody with fundamental group generated by @xmath22 and @xmath23 an example of a disk - busting curve is one that includes into the fundamental group of the handlebody as @xmath24 ( see figure [ diskbust ] ) .
all handlebodies have disk - busting curves and richard stong @xcite gives an algorithm to recognize them . for more on disk - busting curves see @xcite or @xcite .
let @xmath25 be the genus of a closed orientable surface @xmath19 .
let @xmath0 be the manifold obtained by gluing genus @xmath25 handlebodies @xmath26 and @xmath27 together along @xmath19 in such a way that there is an orientation reversing involution @xmath6 interchanging @xmath26 and @xmath27 which pointwise fixes the surface @xmath19 .
now @xmath0 is a closed , connected , orientable 3-manifold @xmath0 such that @xmath28 .
however , @xmath0 is reducible . in order to create an irreducible 3-manifold
, we first remove neighborhoods @xmath29 and @xmath30 of identical disk busting curves @xmath31 and @xmath32 in the interiors of the handlebodies @xmath26 and @xmath27 such that @xmath29 and @xmath30 are interchanged by the involution @xmath6 .
note that since @xmath33 consists of two identical handlebodies from which neighborhoods of disk busting curves have been removed , the inclusion of @xmath19 in each component of @xmath33 is incompressible .
now we sew in identical non - trivial knot complements @xmath34 and @xmath35 along @xmath36 and @xmath37 respectively so that the seifert surfaces of the @xmath38 are glued in where the meridians of the @xmath39 were .
let @xmath20 denote the 3-manifold obtained in this way .
then the restriction @xmath40 can be extended to an orientation reversing involution of @xmath20 that pointwise fixes @xmath19 .
note that the components of @xmath41 that contain @xmath19 are homeomorphic to the corresponding components of @xmath42 that contain @xmath19 .
* claim 1 : * the surfaces @xmath19 , @xmath43 , and @xmath44 are each incompressible in @xmath20 * proof of claim 1 : * assume one of @xmath19 , @xmath43 , or @xmath44 is compressible in @xmath20 . then there is a compressing disk @xmath45 for one of these surfaces that meets @xmath46 transversally in a minimal number of components .
if the interior of @xmath45 intersects one of the surfaces @xmath19 , @xmath43 , or @xmath44 , then an innermost loop on @xmath45 bounds a compressing disk @xmath47 for that surface whose interior is disjoint from the other surfaces .
now since each @xmath38 is a knot complement , @xmath47 can not be contained in @xmath38 .
thus @xmath47 must be contained in some component of @xmath48 .
also , since the @xmath49 are disk busting for the @xmath50 , @xmath51 can not be in @xmath19 .
so @xmath47 is a compressing disk for one of the @xmath52 in @xmath53 . by compressing along @xmath47
we obtain a sphere in @xmath53 .
hence @xmath39 is contained in a ball in @xmath50 , contradicting the fact that @xmath49 was disk busting .
* claim 2 : * @xmath20 is irreducible .
* proof of claim 2 : * let @xmath54 be a sphere in @xmath20 and assume without loss of generality that @xmath54 intersects @xmath46 transversally in a minimal number of components .
if @xmath54 intersects any of @xmath43 , @xmath44 , or @xmath19 , then there is an innermost loop on @xmath54 bounding a disk @xmath45 that is a compressing disk for @xmath43 , @xmath44 , or @xmath19 in @xmath55 . but this violates claim 1 .
thus @xmath54 is contained in some @xmath38 or @xmath56
. however , the @xmath38 and the @xmath56 are irreducible .
therefore @xmath20 is irreducible .
now , recall that @xmath28 .
also , it can be seen using a mayer - vietoris sequence that replacing the handlebodies @xmath29 and @xmath30 by the knot complements @xmath34 and @xmath35 does not change the first homology , since a meridional disk of @xmath39 is replaced by a seifert surface in @xmath38 .
thus @xmath21 , and hence @xmath20 has the properties required by the proposition . in order to prove that we can find infinitely many such 3-manifolds @xmath20 , we first show as follows that if @xmath38 is the complement of a connected sum of distinct knots @xmath57 , then @xmath20 contains at least @xmath58 disjoint non - parallel incompressible tori .
the first such torus is a boundary parallel torus in @xmath38 .
the second such torus is a follow - swallow torus that swallows @xmath59 and follows the other @xmath60 tori .
the third swallows @xmath61 and follows the other @xmath62 and so on , each time swallowing one more knot and following one less until the @xmath63 torus swallows all of the @xmath64 with @xmath65 .
since we have @xmath66 such disjoint tori in each @xmath38 , this gives us @xmath58 such disjoint tori in @xmath20 .
these tori can not be parallel to each other since they bound distinct knot complements in @xmath38 . to see that these @xmath58 tori are incompressible in @xmath20 , suppose there is a compressing disk for one of the tori that intersects @xmath67 transversally in a minimal number of components .
an innermost loop on the disk would be a compressing disk for @xmath43 , @xmath44 or @xmath19 , again contradicting claim 1 .
thus the torus would have to be compressible in one of the @xmath38 .
but it is well known that follow - swallow tori are incompressible in knot complements , so the @xmath58 tori must be incompressible in @xmath20 as well .
thus @xmath20 contains @xmath58 disjoint , non - parallel , incompressible tori . on the other hand , it follows from kneser - haken finiteness , that for a given compact , orientable manifold , such as @xmath20 , there is some finite constant @xmath68 such that @xmath20 can not contain more than @xmath68 disjoint closed , non - parallel , incompressible surfaces ( for example , see proposition 1.7 in @xcite or see @xcite ) .
thus @xmath69 . to get a manifold @xmath70 which is not homeomorphic to @xmath20 ,
we replace each @xmath38 by the complement of a knot that is the connected sum of more than @xmath71 knots .
now @xmath70 will contain @xmath72 disjoint , non - parallel , incompressible tori with @xmath73 , and thus @xmath70 is distinct from @xmath20 . by repeating this process
, we can create an infinite sequence of such manifolds each containing more disjoint , non - parallel , incompressible tori than the previous manifold did .
since every graph embeds in a closed orientable surface , the following theorem is an immediate consequence of proposition [ prop : inf ] .
[ theorem1 ] for every graph @xmath2 , there are infinitely many closed , connected , orientable , irreducible 3-manifolds @xmath0 such that some embedding of @xmath2 in @xmath0 is pointwise fixed by an orientation reversing involution of @xmath0 .
the goal of this section is to prove our main result .
we begin with the following definition .
we define the _ genus of a closed surface _ @xmath19 by @xmath74 then the _ genus of a graph _ is defined as the minimum genus of any closed surface in which the graph embeds .
it is worth pointing out that the genus of a non - orientable surface is not consistently defined in the literature .
some papers use our definition , while others define the projective plane as having genus 1 instead of @xmath75 . let @xmath0 be a closed , connected , orientable , irreducible 3-manifold
. then by kneser - haken finiteness @xcite , @xmath0 contains at most finitely many non - parallel disjoint incompressible tori .
let @xmath76 denote either the maximal possible number of non - parallel disjoint incompressible tori in @xmath0 or 2 , whichever is larger .
now we define @xmath77 .
we will refer to the constants @xmath3 and @xmath76 in the statements and proofs of theorem [ icmanifolds ] and proposition [ mainprop ] .
let @xmath2 be a 3-connected graph with genus at least 2 , and let @xmath4 be an embedding of @xmath2 in a closed , connected , orientable , irreducible 3-manifold @xmath0 such that @xmath7 has an orientation reversing homeomorphism fixing every vertex of @xmath4 .
then there is an embedding @xmath78 of @xmath2 in a closed , connected , orientable 3-manifold @xmath79 such that @xmath80 has an orientation reversing involution pointwise fixing @xmath78 and @xmath81 the point of this proposition is that if we have an embedding @xmath4 of a graph @xmath2 in a manifold @xmath0 such that @xmath82 has an orientation reversing _ homeomorphism _ , then we can find a ( possibly different ) manifold @xmath79 and an embedding @xmath78 of @xmath2 in @xmath79 such that @xmath80 has an orientation reversing _
involution_. furthermore , even though @xmath79 might be homologically more complicated than @xmath0 , there is a bound on @xmath83 which depends only on @xmath0 and not on the graph @xmath2 or the particular embedding of @xmath2 in @xmath0 .
note that the orientation reversing homeomorphism in the hypothesis of proposition [ mainprop ] , leaves each edge of @xmath4 setwise invariant since , according to our definition of a graph , there is at most one edge between any pair of vertices .
while this homeomorphism does not necessarily pointwise fix @xmath4 , it is isotopic to a homeomorphism which does fix @xmath4 pointwise .
proposition [ mainprop ] will be proved in the next section .
we now prove theorem [ icmanifolds ] ( restated below ) by making use of proposition [ mainprop ] together with the following result of kobayashi .
@xcite let @xmath84 be a closed , orientable , 3-manifold admitting an orientation reversing involution @xmath6 .
then @xmath85 for every closed , connected , orientable , irreducible 3-manifold @xmath0 , there is an integer @xmath1 such that any abstract graph with no order @xmath9 automorphism which has a 3-connected minor @xmath12 with @xmath13 is intrinsically chiral in @xmath0 .
let @xmath2 be a graph with no order @xmath9 automorphism .
suppose for the sake of contradiction that @xmath2 has an achiral embedding @xmath4 in the manifold @xmath0 .
let @xmath3 and @xmath76 be the constants associated with @xmath0 that were defined at the beginning of the section .
let @xmath12 be a 3-connected minor of @xmath2 .
we will now show that @xmath12 satisfies the inequality @xmath86 first observe that if @xmath87 , then the above inequality is immediate since @xmath88 .
thus we assume that @xmath89 .
since @xmath4 is an achiral embedding of @xmath2 in @xmath0 , there is an orientation reversing homeomorphism @xmath90 of the pair @xmath7 .
let @xmath91 denote the automorphism that @xmath90 induces on the graph @xmath4 .
now even though @xmath90 does not necessarily have finite order , @xmath91 has finite order because @xmath4 has a finite number of vertices .
hence we can express the order of @xmath91 as @xmath92 , where @xmath93 and @xmath94 is odd .
now since @xmath90 is orientation reversing and @xmath94 is odd , @xmath95 is also an orientation reversing homeomorphism of @xmath7 .
also , @xmath25 induces the automorphism @xmath96 on @xmath4 and @xmath97 .
in particular , @xmath98 fixes every vertex of @xmath4 . if @xmath99 , then @xmath100 would induce an order two automorphism on @xmath4 . as we assumed
that no such automorphism exists , we must have @xmath101 .
thus @xmath102 is an orientation reversing homeomorphism of @xmath7 which fixes every vertex of @xmath4 .
since @xmath12 is a minor of the abstract graph @xmath2 , by deleting and/or contracting some edges of the embedding @xmath4 of @xmath2 in @xmath0 , we obtain an embedding @xmath103 of @xmath12 in @xmath0 .
furthermore , by composing the homeomorphism @xmath25 with an isotopy in a neighborhood of each edge that was contracted , we obtain an orientation reversing homeomorphism of @xmath104 which fixes every vertex of @xmath103 .
since @xmath12 is a 3-connected graph with @xmath89 , we can now apply proposition [ mainprop ] to get an embedding @xmath105 of @xmath12 in a closed , connected , orientable , 3-manifold @xmath79 such that @xmath106 has an orientation reversing involution @xmath6 pointwise fixing @xmath105 and @xmath107
let @xmath54 be the component of the fixed point set @xmath108 containing @xmath105 , and let @xmath109 .
we put a metric @xmath110 on @xmath79 , and define a new metric @xmath111 by @xmath112 .
then any neighborhood of @xmath113 with respect to the metric @xmath111 will be setwise invariant under @xmath6 .
now we can pick a neighborhood @xmath114 with respect to @xmath111 such that @xmath114 is homeomorphic to a ball . then by smith theory @xcite , since @xmath115 is an orientation reversing involution of the ball @xmath114 , the fixed point set of @xmath115 is either a single point or a properly embedded disk . since @xmath116 contains more than one point
, @xmath117 is a properly embedded disk , and hence @xmath54 is a closed surface .
thus @xmath118 @xmath119 hence we have the inequality @xmath120 also , since @xmath6 is an orientation reversing involution and @xmath79 is a closed orientable manifold , we can apply kobayashi s theorem @xcite to obtain the inequality @xmath121 it follows that @xmath122 combining the above inequalities , we now have @xmath123 but @xmath79 was given by proposition [ mainprop ] such that @xmath124 hence we obtain the required inequality @xmath86 it now follows that if @xmath2 has a 3-connected minor whose genus is greater than @xmath3 , then @xmath2 must be intrinsically chiral in @xmath0 .
in the course of the proof of proposition [ mainprop ] , we will use the well known
half lives , half dies " theorem , which we state below .
see @xcite or @xcite for a proof of this theorem .
let @xmath0 be a manifold which has a torus boundary component @xmath126 .
then for any pair of generators @xmath22 and @xmath23 of @xmath127 , at least one of @xmath22 and @xmath23 is non - trivial in @xmath128 .
[ cor : hlhd ] suppose for the sake of contradiction that the generators @xmath22 and @xmath23 are both trivial in @xmath128 . attach handlebodies to all boundary components of @xmath0 except @xmath126 to form a new manifold @xmath129 with a single boundary component .
then @xmath22 and @xmath23 are both trivial in @xmath130 .
since @xmath22 and @xmath23 generate the homology of the only boundary component of @xmath129 , we see that @xmath131 . but
this contradicts the half lives , half dies theorem .
thus at least one of the generators of @xmath127 must have been non - trivial in @xmath0 .
note that it is tempting to assume that the half lives , half dies theorem implies that one of @xmath22 or @xmath23 must be trivial in @xmath129 .
but this is not always true . in particular , let @xmath0 denote the product of a torus and an interval .
then no non - trivial curve in the boundary of @xmath0 is in the kernel .
[ mainprop ] let @xmath2 be a 3-connected graph with genus at least 2 , and let @xmath4 be an embedding of @xmath2 in a closed , connected , orientable , irreducible 3-manifold @xmath0 such that @xmath7 has an orientation reversing homeomorphism @xmath25 fixing every vertex of @xmath4
. let @xmath103 denote either @xmath4 or @xmath4 with one edge deleted , and suppose that @xmath103 is contained in a ball @xmath133 in @xmath0 . since @xmath134 is isotopic to @xmath133 in @xmath0 , we can assume that @xmath25 leaves @xmath133 setwise invariant .
also , since @xmath25 fixes every vertex of @xmath103 and @xmath103 has at most one edge between any pair of vertices , @xmath25 takes each edge to itself .
thus @xmath135 is isotopic to a homeomorphism which pointwise fixes @xmath103 .
so we assume that @xmath25 leaves @xmath133 setwise invariant and pointwise fixes @xmath103 .
let @xmath90 be an embedding of @xmath136 in @xmath8 .
then @xmath137 is an orientation reversing homeomorphism of @xmath138 pointwise fixing @xmath139 .
now , @xmath137 can be extended to an orientation reversing homeomorphism of @xmath8 pointwise fixing @xmath139 .
however , jiang and wang @xcite showed that no graph containing @xmath140 or @xmath141 has an embedding in @xmath8 which is pointwise fixed by an orientation reversing homeomorphism of @xmath8 .
thus @xmath103 can not contain @xmath140 or @xmath141 , and hence is abstractly planar .
but this implies that @xmath142 , which is contrary to hypothesis .
thus neither @xmath4 nor @xmath4 with an edge deleted can be contained in a ball in @xmath0 .
we will use this result later in the proof .
let @xmath144 and @xmath145 be the sets of vertices and edges of @xmath4 respectively . for each vertex @xmath146 , define @xmath147 to be a ball around @xmath148 in @xmath0 ( i.e. , a 0-handle containing @xmath148 ) , and let @xmath149 denote the union of the balls around the vertices . also , for each edge @xmath150 , let @xmath151 be a solid tube around @xmath152 in @xmath0 ( i.e. , a 1-handle containing the portion of @xmath153 outside of the 0-handles ) .
let @xmath154 denote the union of the tubes around the edges .
then for each @xmath153 in @xmath154 the intersection @xmath155 follows the standard convention for attaching 1-handles to 0-handles . in other words , in @xmath156
this intersection consists of the disks @xmath157 and @xmath158 and in @xmath159 .
then @xmath160 is a neighborhood of @xmath4 . for convenience
we introduce the following terminology . for each vertex @xmath148 , we let @xmath161 denote the sphere with holes @xmath162 , and for each edge @xmath153 we let @xmath163 denote the annulus @xmath164 .
thus @xmath165 .
now since @xmath166 fixing each vertex of @xmath4 , we know that @xmath167 is isotopic to @xmath143 setwise fixing @xmath4 and fixing each vertex .
thus we can modify @xmath25 by an isotopy ( and by an abuse of notation , still refer to the map as @xmath25 ) so that @xmath166 for each vertex @xmath148 and edge @xmath153 we have @xmath168 and @xmath169 . because this modification was by an isotopy , our new @xmath25 is still orientation reversing .
since @xmath0 is irreducible and we have assumed that @xmath4 is not contained in a ball , @xmath170 is irreducible .
thus we can apply the characteristic decomposition theorem of jaco - shalen @xcite and johannson @xcite to get a minimal family of incompressible tori @xmath171 for @xmath170 such that each closed up component of @xmath172 is either seifert fibered or atoriodal .
since the characteristic family @xmath171 is unique up to isotopy , we can again modify @xmath25 by an isotopy ( and again by an abuse of notation still refer to the map as @xmath25 ) so that @xmath173 and still have @xmath166 , @xmath168 and @xmath169 for each vertex @xmath148 and edge @xmath153 .
let @xmath84 be the closed up component of @xmath172 containing @xmath174 ( see for example figure [ jsj ] )
. then @xmath175 . also ,
since @xmath4 is 3-connected , @xmath176 .
thus the component @xmath84 is not seifert fibered , and hence is atoriodal .
let @xmath177 denote the set of torus boundary components of @xmath84 together with the annuli that make up the components of @xmath178 .
since @xmath4 is 3-connected , @xmath179 is incompressible in @xmath170 .
it follows that @xmath180 is incompressible in @xmath84 .
furthermore , @xmath84 is irreducible since @xmath170 is irreducible and @xmath84 is a component of the jsj decomposition of @xmath170 .
* step 3 : we show that any sphere obtained by capping off an annulus @xmath181 in the jsj decomposition of @xmath182 bounds a ball @xmath183 such that if the components of @xmath184 are in different components of @xmath159 , then @xmath133 meets @xmath185 in a single edge , and otherwise @xmath133 is disjoint from @xmath185 . *
we now apply the characteristic decomposition theorem for pared manifolds @xcite to the pared manifold @xmath182 . since @xmath84 is atoroidal , this gives us a characteristic family @xmath186 of incompressible annuli in x with boundaries in @xmath187 such that if @xmath188 is the closure of any component of @xmath189 , then the pared manifold @xmath190 is either simple , seifert fibered , or i - fibered ( see @xcite for the necessary definitions ) . once again , since the characteristic family @xmath186 is unique up to isotopy , we can modify @xmath25 by an isotopy ( and again by an abuse of notation , still refer to the map as @xmath25 ) so that @xmath191 .
let @xmath181 be an annulus component of @xmath192 , and let @xmath19 denote the sphere obtained by capping off @xmath181 by a pair of disjoint disks @xmath193 and @xmath194 in @xmath195 and @xmath196 , where @xmath197 and @xmath198 may or may not be distinct vertices .
suppose that each component of @xmath199 intersects more than one edge of @xmath185 . then by removing the vertices @xmath197 and @xmath198 and the edges that contain them we would obtain two non - empty subgraphs ( see figure [ annulus ] ) .
but this contradicts our hypothesis that @xmath4 is 3-connected .
thus one of the components of @xmath199 meets @xmath185 in at most one edge of @xmath4 .
now , since @xmath0 is irreducible , one of the closed up components of @xmath199 is a ball @xmath133 .
however , we assumed at the beginning of our proof that neither @xmath4 nor @xmath4 with an edge removed can be contained in a ball in @xmath0
. thus @xmath133 must be the closed up component of @xmath199 intersecting at most one edge of @xmath185 .
now since the annulus @xmath181 is incompressible in @xmath84 , if @xmath200 then there is some edge @xmath153 with vertices @xmath197 and @xmath198 such that @xmath133 contains @xmath201 . on the other hand ,
suppose that @xmath202 .
then by our definition of a graph , every edge in @xmath4 has two distinct vertices .
thus the component of @xmath199 which meets @xmath185 in at most one edge , is actually disjoint from @xmath185 . in this case
, @xmath133 must be disjoint from @xmath185 as illustrated in figure [ v1v2 ] .
note that since a ball can not contain an incompressible torus , no torus boundary component of @xmath84 can occur inside of @xmath133 .
it follows that every torus boundary component of @xmath203 must also be a boundary component of @xmath204 .
let @xmath181 be an annulus in @xmath192 with one boundary in @xmath211 and the other boundary in @xmath196 where @xmath200 . by capping off @xmath181 with disks in @xmath211 and @xmath196
we obtain a sphere which ( as we saw in step 3 ) bounds a ball @xmath133 that meets @xmath4 in @xmath201 for some edge @xmath153 .
note that by our definition of a graph , @xmath153 is the only edge of @xmath4 whose vertices are @xmath197 and @xmath198 .
thus we can let @xmath212 denote the collection of all annuli in @xmath192 with one boundary in @xmath211 and the other boundary in @xmath196 .
now we can cap off the annuli in @xmath212 with disks in @xmath211 and @xmath196 to obtain a collection of spheres which bound nested balls containing @xmath201 .
let @xmath213 denote the annulus in @xmath212 which , when capped off in this way , is outermost with respect to the nested balls .
observe that the boundaries of @xmath214 bound disks @xmath215 and @xmath216 which each meet @xmath4 in a single point of @xmath153 , and the sphere @xmath217 bounds a ball @xmath218 which contains @xmath201 and all of the other annuli in @xmath219 .
furthermore , as we saw at the end of step 3 , no torus boundary component of @xmath84 can be contained in @xmath220 .
we see as follows that @xmath220 does not separate @xmath84 .
if @xmath221 , then @xmath222 and hence @xmath223 .
thus we assume that @xmath224 . in this case , @xmath225 is obtained by removing the interior of the @xmath14-handle @xmath156 from the ball @xmath220 . for each @xmath226 , let @xmath227 . then each @xmath228 is adjacent to both @xmath213 and @xmath163 .
now @xmath229 , and hence @xmath230 consists of a single component .
since @xmath231 , it now follows that @xmath232 has a single component .
we will use this at the end of this step .
we repeat the above construction for each annulus in @xmath192 with boundaries in distinct components of @xmath159 .
this gives us a collection of pairwise disjoint balls @xmath205 ,
, @xmath206 .
observe that for every edge @xmath233 , the annulus @xmath234 is in @xmath235 and its boundaries are in distinct components of @xmath159 .
thus every edge @xmath233 in @xmath4 is contained in some @xmath236 .
it follows that @xmath237 contains @xmath238 and contains every annulus in @xmath192 with boundaries in distinct components of @xmath159 .
next we consider an annulus @xmath54 in @xmath186 both of whose boundaries are in a single @xmath239 .
we saw in step 3 that if we cap off @xmath54 by a pair of disjoint disks in @xmath147 , we obtain a sphere which bounds a ball that is disjoint from @xmath240
. thus the components of @xmath241 are parallel in @xmath161 .
we now cap off every such annulus to obtain a collection of spheres which bound balls that are either nested or disjoint .
thus we can choose an annulus @xmath242 which , when capped off with disks in @xmath147 , bounds an outermost ball @xmath243 with respect to this nesting ( see figure [ fi ] ) .
observe that , as we saw for the @xmath236 , the ball @xmath243 can not contain any torus boundary components of @xmath84 .
however , in this case @xmath244 is obtained by removing the interior of the @xmath9-handle @xmath245 from the ball @xmath243 .
thus @xmath246 where @xmath247 is an annulus in @xmath248 .
this means that @xmath249 has a single component , and hence @xmath250 also has a single component .
now for each annulus @xmath251 with both components of @xmath241 in a single @xmath239 such that @xmath54 is not contained in one of the balls @xmath236 , we use the above argument to define a ball @xmath252 . in this way
, we obtain a collection of pairwise disjoint balls @xmath205 , , @xmath206 , @xmath207 , , @xmath208 .
furthermore , each of the sets @xmath253 and @xmath254 has a single component , and each @xmath255 and @xmath256 is an annulus in @xmath257 .
now it follows that the manifold @xmath258 is the closure of a single component of @xmath210 ( see figure [ annulus ] ) .
recall that @xmath25 fixes each vertex and leaves each edge setwise invariant .
also , @xmath260 , @xmath261 , and @xmath191 .
now , since the sets of balls @xmath262 and @xmath263 were chosen to be outermost , each of these sets is also invariant under @xmath25 . furthermore , we know from step 4 that each @xmath236 intersects @xmath264 only in @xmath233 .
now since @xmath25 leaves each @xmath233 setwise invariant , @xmath25 must also leave each @xmath236 setwise invariant .
it follows that @xmath265 must also be setwise invariant under @xmath25 ; and since each vertex is fixed by @xmath25 , each boundary component of @xmath265 is also setwise invariant under @xmath25 .
now let @xmath148 be a vertex such that some @xmath266 has both its boundary components in @xmath239 .
since @xmath266 is incompressible in @xmath84 , the components of @xmath267 bound disjoint disks in @xmath239 which each nest at least one edge of @xmath4 . since every vertex of @xmath4 is fixed by @xmath25 , it follows that @xmath266 and its boundary components are also setwise invariant under @xmath25 .
thus @xmath268 is setwise invariant under @xmath25 . finally , since all of the @xmath236 and @xmath268 are setwise invariant under @xmath25 , we know that @xmath188 must be setwise invariant under @xmath25 as well . to show that the pared manifold @xmath190 is simple , first recall that @xmath188 is the closure of a single component of @xmath189 . hence by jsj for pared manifolds @xcite , @xmath190 is either @xmath269-fibered , seifert fibered , or simple as a pared manifold .
we see that @xmath190 can not be seifert fibered or @xmath269-fibered as follows .
first observe that for every vertex @xmath148 , there is some edge @xmath153 such that the ball @xmath220 meets @xmath239 .
it follows that @xmath270 meets every component of @xmath159 .
furthermore , since every vertex @xmath148 has valence at least three , @xmath161 is a sphere with at least three holes .
also each @xmath236 contains at most one boundary component of @xmath161 , and each @xmath268 contains no boundary components of @xmath161 . note that if @xmath266 has its boundaries in @xmath239 , then @xmath268 separates @xmath271 into two components .
this means that for each vertex @xmath148 , @xmath272 is homeomorphic to @xmath161 possibly with the interior of a set of pairwise disjoint annuli deleted . since @xmath161 is a sphere with at least three holes , deleting a collection of annuli may create new components , but some component must still be a sphere with at least three holes .
it follows that the component of @xmath270 meeting @xmath174 has genus more than one , and thus the pared manifold @xmath190 can not be seifert fibered .
next , suppose for the sake of contradiction that the pared manifold @xmath190 is @xmath269-fibered . by definition of @xmath269-fibered for pared manifolds
, this means that there is an @xmath269-bundle map of @xmath188 over a base surface @xmath273 such that @xmath274 is in the pre - image of @xmath275 .
it follows that @xmath273 must be homeomorphic to a component of @xmath276 .
this means that the base surface @xmath273 must be a sphere with holes .
in particular , the base surface @xmath273 is orientable .
now since @xmath188 is an orientable @xmath5-manifold which is @xmath269-fibered over the orientable surface @xmath273 , it follows that @xmath188 must actually be a product @xmath277 .
thus @xmath278 , and @xmath279 and @xmath280 are the only components of @xmath281 .
however , since @xmath270 meets every component of @xmath159 , this means that @xmath4 contains at most two vertices . but this contradicts our hypothesis that @xmath4 is 3-connected .
therefore , the pared @xmath190 is not @xmath269-fibered , and since it is also not seifert fibered , it must be simple .
now it follows from thurston s hyperbolization theorem for pared manifolds @xcite applied to the simple pared manifold @xmath284 that @xmath285 admits a finite volume complete hyperbolic metric with totally geodesic boundary .
let @xmath45 denote the double of @xmath285 along its boundary .
then @xmath45 is a finite volume hyperbolic manifold , and @xmath282 can be doubled to obtain an orientation reversing homeomorphism of @xmath45 ( which we still call @xmath25 ) taking each copy of @xmath285 to itself .
now by mostow s rigidity theorem @xcite applied to @xmath45 , the homeomorphism @xmath286 is homotopic to an orientation reversing finite order isometry @xmath287 that restricts to an isometry of @xmath288 . by removing horocyclic neighborhoods of the cusps of @xmath289
, we obtain a copy of the pair @xmath283 which is contained in @xmath290 and is setwise invariant under @xmath6 .
we abuse notation and now consider @xmath6 to be an orientation reversing finite order isometry of @xmath283 instead of this copy .
furthermore , @xmath6 induces isometries on the collection of tori and annuli in @xmath291 with respect to a flat metric .
furthermore , the sets @xmath281 , @xmath292 , and @xmath293 are each setwise invariant under @xmath6 .
finally , it follows from waldhausen s isotopy theorem @xcite that @xmath6 is isotopic to @xmath282 by an isotopy leaving @xmath291 setwise invariant .
now , recall that the boundary components of @xmath188 consist of tori in @xmath171 , and the union of spheres with holes in @xmath281 together with annuli in @xmath192 .
recall from the first paragraph in step 5 that @xmath25 setwise fixes each annulus @xmath294 with boundaries in distinct components of @xmath159 , each annulus @xmath295 with both boundaries in a single component of @xmath159 , each component of @xmath296 , and each component of @xmath267 .
since @xmath6 is isotopic to @xmath282 by an isotopy leaving @xmath291 setwise invariant , @xmath6 leaves invariant the same sets as @xmath25 .
it follows that for each vertex @xmath148 , we have @xmath297 , and @xmath6 takes each component of @xmath272 to itself , leaving each boundary component setwise invariant .
since @xmath6 has finite order , @xmath6 restricts to a finite order homeomorphism of every component of @xmath298 .
we saw in step 5 that for every vertex @xmath148 , at least one component @xmath299 of @xmath272 is a sphere with at least three holes . since @xmath6 restricts to a finite order homeomorphism of @xmath299 taking each boundary component of @xmath300 to itself , @xmath6 must be a reflection of @xmath299 which also reflects each component of @xmath301 .
now @xmath302 is a finite order , orientation preserving isometry of @xmath188 that pointwise fixes the surface @xmath299 .
it follows that @xmath302 is the identity , and hence @xmath6 is an involution of @xmath188 . as we saw in step 6 , for each vertex @xmath148 , @xmath6 reflects some component @xmath299 of @xmath272 which is a sphere with at least three holes , and @xmath6 reflects every component of @xmath301 .
let @xmath304 denote some boundary component of @xmath299 .
then @xmath304 is also a boundary component of either an annulus @xmath265 or an annulus @xmath266 . since @xmath6 reflects @xmath304 , we know that @xmath6 must also reflect the annulus @xmath265 or @xmath266 , whichever contains @xmath304 in its boundary .
since the boundaries of the annulus are not interchanged , @xmath6 must also reflect each boundary component of @xmath265 or @xmath266 .
below we extend @xmath6 to @xmath236 or @xmath268 .
first we consider the case where @xmath304 is in the boundary of an annulus @xmath294 .
let @xmath305 and @xmath306 denote the disks whose union is in @xmath307 .
then @xmath305 and @xmath308 each meet @xmath4 in a single point of @xmath233 .
since @xmath6 reflects the annulus @xmath265 together with each boundary component of @xmath265 , we can extend @xmath6 radially to the disks @xmath305 and @xmath308 to get a reflection of the sphere @xmath309 pointwise fixing a circle containing the points @xmath310 and @xmath311 . recall that the sphere @xmath309 bounds the ball @xmath236 in @xmath0 .
now , we can express @xmath236 as a product @xmath312 whose core @xmath313 has endpoints @xmath310 and @xmath311 ( see figure [ knotted ] ) .
now we extend @xmath6 from a reflection of the sphere @xmath309 to a reflection of the product @xmath312 which pointwise fixes the core @xmath313 .
next we consider the case where @xmath304 is a boundary component of an annulus @xmath295 which has both boundaries in a single @xmath239 .
recall that @xmath314 consists of disks @xmath315 and @xmath316 properly embedded in @xmath147 . since @xmath6 reflects the annulus @xmath266 together with each of its boundary components , we can extend @xmath6 radially to the disks @xmath315 and @xmath316 to get a reflection of the sphere @xmath317 pointwise fixing a circle containing the points of @xmath318 and @xmath319 .
thus we can extend @xmath6 to a reflection of the ball @xmath268 which pointwise fixes a disk containing @xmath320 ( see figure [ bi ] ) .
now the extension of @xmath6 reflects the sphere with holes @xmath299 , and one of the balls @xmath236 or @xmath268 depending on whether @xmath304 is a boundary component of @xmath265 or @xmath266 , respectively .
next we let @xmath322 denote the union of @xmath299 together with the annulus @xmath265 or @xmath266 glued along @xmath304 . now
@xmath6 reflects @xmath322 taking every boundary component of @xmath322 to itself , and hence reflecting every boundary component of @xmath322 .
let @xmath323 be a boundary component of @xmath322 .
if @xmath323 is not the other boundary of the annulus @xmath265 or @xmath266 , then we repeat the above argument with @xmath323 in place of @xmath304 .
if @xmath323 is the other boundary of the annulus @xmath265 or @xmath266 , then @xmath323 is also a boundary of some other component @xmath324 of @xmath298 , as illustrated in figure [ induction ] . in this case , since @xmath323 is reflected by @xmath6 and every boundary component of @xmath324 is setwise invariant under @xmath6 , we know that @xmath6 must reflect @xmath326 .
now let @xmath327 denote a boundary component of @xmath328 , and repeat the above argument with @xmath327 in place of @xmath304 . in general , for a given surface @xmath329 obtained in this way ,
the surface @xmath330 is the union of @xmath329 together with either an annulus of the form @xmath265 or @xmath266 or a sphere with at least two holes contained in @xmath298 .
furthermore , @xmath330 is reflected by @xmath6 .
this process will only stop when the surface that we obtain has no boundary components .
since @xmath270 has only one component which intersects @xmath159 , the closed surface that we obtain in this way must be @xmath270 .
thus we have extended @xmath6 to an orientation reversing involution of each of the balls @xmath205 , , @xmath206 , @xmath207 , , @xmath208 .
now let @xmath331 .
then @xmath332 is a collection of disjoint balls ( for example in figure [ bi ] , @xmath333 is two balls one of which contains @xmath148 ) . also , @xmath6 is a reflection of each component of @xmath334 that fixes each point in @xmath335 .
now we extend @xmath6 radially to a reflection of each ball of @xmath332 in such a way that @xmath6 pointwise fixes each component of @xmath336 .
thus @xmath6 is defined as a reflection of each component of @xmath149 which pointwise fixes @xmath337 .
recall from the end of step 4 that @xmath339 and @xmath270 have the same collection of tori in their boundary components .
furthermore , we have filled in the boundary component of @xmath188 meeting @xmath174 with a collection of balls in @xmath84 and @xmath143 .
thus in fact @xmath340 . finally , we define a new embedding @xmath78 of @xmath2 in @xmath303 as follows .
let @xmath341 .
then for each edge @xmath233 define an embedding of @xmath342 in @xmath78 as the core @xmath313 of @xmath343 , which we know is pointwise fixed by @xmath6 according to the way we extended @xmath6 to @xmath236 ( recall figure [ knotted ] ) .
suppose that an essential curve @xmath347 on some @xmath348 compresses in @xmath0 .
let @xmath315 be a compressing disk for @xmath347 whose intersection with the set of tori @xmath345 is minimal .
let @xmath349 if the interior of @xmath315 is disjoint from @xmath348 .
otherwise , there exists some @xmath45 in the interior of @xmath315 such that @xmath45 is a compressing disk for @xmath348 whose interior is disjoint from @xmath348 . in either case , the intersection of @xmath45 with @xmath345 is minimal .
suppose that @xmath45 contains at least one curve of intersection in its interior .
hence there is an innermost disk @xmath47 on @xmath45 which is a compressing disk for some @xmath350 with @xmath351 . since @xmath350 compresses in @xmath0 but is incompressible in @xmath346 , we know that @xmath47 intersects @xmath4 . since @xmath0 is irreducible ,
any compressible torus is separating in @xmath0 .
thus we can let @xmath352 denote the closed up component of @xmath353 containing @xmath84 and let @xmath354 denote the closed up component of @xmath353 whose interior is disjoint from @xmath84 .
now let @xmath19 denote the region of @xmath45 which is adjacent to the innermost disk @xmath47 .
then @xmath355 , since @xmath356 .
also , @xmath357 implies that @xmath19 is adjacent to another region of @xmath45 which is contained in @xmath352
. in particular , there must be another circle of intersection @xmath358 of @xmath359 which bounds a disk @xmath360 such that @xmath361 contains @xmath362 .
we illustrate the abstract disk @xmath45 and its intersections with @xmath350 in figure [ disk ] .
the white regions in the figure are contained in @xmath354 , and the grey regions are contained in @xmath352 . note we do not illustrate any circles of intersection of @xmath45 with any @xmath363 with @xmath364 . now , since the intersection of @xmath45 with the tori @xmath365 ,
, @xmath366 is minimal , all of the curves of intersection of @xmath359 must be essential on @xmath350 .
in particular , @xmath51 and @xmath358 must both be essential on @xmath350 .
since there can not be two essential , disjoint , non - parallel curves on a torus , this means that @xmath358 is parallel to @xmath51 on @xmath350 .
it follows that @xmath358 must bound a disk @xmath367 which is parallel to @xmath47 in @xmath0 .
in particular , since the interior of @xmath47 is disjoint from @xmath365 , , @xmath366 , the interior of @xmath367 is as well .
but now by replacing the disk @xmath361 with the disk @xmath367 in the compressing disk @xmath45 we obtain a new compressing disk @xmath368 for @xmath348 which has fewer curves of intersection with @xmath365 , ,
@xmath366 than @xmath45 has . from this contradiction
we conclude that the interior of @xmath45 must be disjoint from @xmath369 , and hence @xmath370 . if @xmath371 , then @xmath347 compresses in @xmath303 as required .
otherwise , the compression disk @xmath45 was contained in the interior of the original disk @xmath315 and @xmath372 . in this case , since the intersection of @xmath315 with the set of tori @xmath345 was minimal , @xmath373 is essential in @xmath348 .
but now @xmath373 and @xmath347 are disjoint essential curves on @xmath348 . hence as we saw above , the disks @xmath45 and @xmath315 must be parallel in @xmath0 . now , since @xmath370 , it follows that @xmath347 must compress in @xmath303 as well .
* condition : * _ if @xmath348 is compressible in @xmath0 , then every element of @xmath374 is trivial in @xmath375 , and if @xmath348 is incompressible in @xmath0 then at least one non - trivial element of @xmath374 is trivial in @xmath375 . _
let @xmath348 be a component of @xmath344 . by corollary [ cor : hlhd ] at the beginning of section 4 there is a curve @xmath376 on @xmath348 which is non - trivial in @xmath377 .
also , we know from step 8 that if some essential curve @xmath347 on @xmath348 compresses in @xmath0 , then @xmath347 also compresses in @xmath303 . in particular , @xmath347 is not homologous in @xmath348 to @xmath376 .
now suppose that for some @xmath378 , the involution @xmath6 interchanges @xmath348 and @xmath350 . since @xmath379 is a homeomorphism and @xmath376 is a curve on @xmath348 which is non - trivial in @xmath377 , we know that @xmath380 is a curve on @xmath350 which is also non - trivial in @xmath381 .
now we fill @xmath303 along @xmath348 by adding a solid torus @xmath50 with its meridian attached to the non - trivial curve @xmath376 , and we fill along @xmath350 by adding a solid torus @xmath354 with its meridian attached to @xmath380
. then we extend the involution @xmath6 radially in @xmath382 ( abusing notation and still calling the involution @xmath6 ) .
we repeat this process for every boundary component of @xmath303 which is not setwise invariant under @xmath6 .
thus for every @xmath348 along which we have glued a solid torus @xmath50 , the curve @xmath376 on @xmath348 is now trivial in @xmath383 .
furthermore , if @xmath348 is compressible in @xmath0 , then there is an essential curve @xmath347 on @xmath348 which compresses in @xmath303 .
hence , @xmath347 is not homologous to @xmath376 in @xmath374 , and together they generate @xmath374 .
furthermore , both @xmath347 and @xmath376 are trivial in @xmath383 .
let @xmath384 be the manifold that we have obtained by filling all of the boundary components of @xmath303 which are not setwise fixed by @xmath6 , and let @xmath348 be a component of @xmath385 .
recall from step 7 that @xmath78 is an embedding of @xmath2 in @xmath303 which is pointwise fixed by @xmath6 .
now @xmath386 is an orientation reversing involution pointwise fixing @xmath78 , and @xmath348 is setwise invariant under @xmath6 .
furthermore , by the proof of step 6 , @xmath387 is an isometry with respect to a flat metric . since @xmath6 is an orientation reversing involution of @xmath384 and @xmath348
is a boundary component of @xmath384 , @xmath387 is also an orientation reversing involution .
thus @xmath387 is either a reflection pointwise fixing two parallel circles on @xmath348 , or a composition of a reflection and an order @xmath9 rotation of @xmath348 . in either case , for any non - trivial element @xmath388 , there is a non - trivial curve @xmath389 such that @xmath390 and @xmath391 meet transversely in a single point and @xmath392 is homologous to @xmath393 in @xmath394 .
now suppose that some essential curve @xmath347 on @xmath348 compresses in @xmath0 . then by step 8
, @xmath347 also compresses in @xmath303 , and hence in @xmath384 .
pick a non - trivial curve @xmath391 on @xmath348 that intersects @xmath347 in a single point such that @xmath392 is homologous to @xmath393 in @xmath394 . then @xmath395 .
also , since @xmath347 is null homologous in @xmath396 , by corollary [ cor : hlhd ] , @xmath391 is non - trivial in @xmath396 .
now we fill @xmath384 along @xmath348 by adding a solid torus @xmath50 with its meridian attached to the non - trivial curve @xmath391 . since @xmath392 is homologous to @xmath397 in @xmath394 , we can extend @xmath6 radially to an orientation reversing involution of the solid torus @xmath50
. then @xmath398 is an orientation reversing involution , and both @xmath347 and @xmath391 are trivial in @xmath399 .
on the other hand , suppose that some @xmath348 is incompressible in @xmath0 . by corollary [ cor : hlhd ]
, there is some curve @xmath391 on @xmath348 which is non - trivial in @xmath396 , and @xmath391 is homologous to @xmath400 in @xmath394 .
now we fill @xmath348 by adding a solid torus @xmath50 with its meridian attached to the curve @xmath391 , and then extend @xmath6 to @xmath50 .
then @xmath398 is again an orientation reversing involution , and @xmath391 is trivial in @xmath399 . in this way , we glue a solid torus to each of the @xmath348 in @xmath401 to obtain a closed manifold @xmath79 satisfying the required condition .
since @xmath402 , this gives us an embedding @xmath78 of @xmath2 in @xmath79 .
furthermore , we have extended @xmath6 to an orientation reversing involution of @xmath403 which pointwise fixes @xmath78 .
recall from section 3 that @xmath76 is either the number of non - parallel disjoint incompressible tori in @xmath0 or @xmath9 , whichever is larger .
if no pair of distinct tori of @xmath401 are parallel in @xmath0 , then at most @xmath76 of the components of @xmath344 are incompressible in @xmath0 .
thus we assume that some pair of distinct tori @xmath348 and @xmath350 of @xmath344 are parallel in @xmath0 .
then @xmath348 and @xmath350 co - bound a region @xmath404 in @xmath0 which is homeomorphic to a product of a torus and an interval .
however , since @xmath348 and @xmath350 are in the characteristic family for @xmath346 , they can not be parallel in @xmath346 .
thus @xmath404 intersects @xmath4 .
but since @xmath405 and @xmath4 is disjoint from @xmath406 , this implies that @xmath407 .
it now follows that @xmath408 .
now suppose that some component @xmath363 of @xmath401 is incompressible in @xmath0 and distinct from @xmath348 and @xmath350
. then @xmath409 .
but since @xmath410 , either @xmath363 is parallel in @xmath404 to both @xmath348 and @xmath350 , or @xmath363 is compressible in @xmath404 .
however , since @xmath411 , the latter would imply that @xmath363 is compressible in @xmath0
. thus @xmath363 must be parallel to both @xmath348 and @xmath350 in @xmath411 .
since @xmath363 does not intersect @xmath4 , this means that @xmath363 is parallel to one of @xmath348 or @xmath350 in @xmath346 .
but this is impossible since @xmath348 , @xmath350 , and @xmath363 are all in the characteristic family for @xmath346 .
thus the only components of @xmath344 which could be incompressible in @xmath0 are @xmath348 and @xmath350 .
hence @xmath344 has at most @xmath412 components which are incompressible in @xmath0 .
recall that @xmath79 is obtained from @xmath303 by adding a collection of solid tori @xmath26 ,
, @xmath414 along the components @xmath365 , ,
@xmath366 of @xmath344 . also , by the condition in step 9 , we know that if @xmath348 is compressible in @xmath0 , then every element of @xmath374 is trivial in @xmath375 ; and if @xmath348 is incompressible in @xmath0 then there exists some curve @xmath415 on @xmath348 such that @xmath415 is non - trivial in @xmath374 but trivial in @xmath375 .
it follows that for every @xmath348 which is incompressible in @xmath0 , there is at most one curve @xmath416 in @xmath348 which is non - trivial in @xmath375 .
thus every curve in @xmath374 is either trivial in @xmath375 or homologous to @xmath416 in @xmath417 .
now let @xmath358 be a curve in @xmath79 which is non - trivial in @xmath417 , and for each solid torus @xmath50 , let @xmath49 denote its core . then by general position , we can choose @xmath358 to be disjoint from @xmath418 , and hence disjoint from @xmath419 .
now since @xmath420 , we can also consider @xmath358 in @xmath421 .
suppose that @xmath358 is trivial in @xmath422 .
it follows that @xmath358 is homologous in @xmath423 to a collection of curves on @xmath424 which are trivial in @xmath422 but non - trivial in @xmath417 . since for any @xmath348 which is compressible in @xmath0 every element of @xmath374 is trivial in @xmath375
, @xmath358 is homologous in @xmath417 to a collection of curves on only those @xmath348 which are incompressible in @xmath0 .
it now follows that @xmath358 is homologous in @xmath417 to a sum of @xmath416 s on @xmath348 s that are incompressible in @xmath0 .
but by step 10 , there are at most @xmath76 such tori .
hence there are at most @xmath76 distinct @xmath416 which are non - trivial in @xmath417 .
it follows that these @xmath76 curves generate every non - trivial element of @xmath417 which is trivial in @xmath422 .
this gives us the required inequality : | the main result of this paper is that for every closed , connected , orientable , irreducible 3-manifold @xmath0 , there is an integer @xmath1 such that for any graph @xmath2 which has a 3-connected minor whose genus is greater than @xmath3 , if @xmath2 has no involution then every embedding of @xmath2 in @xmath0 is chiral .
by contrast , the paper also proves that for every graph @xmath2 , there are infinitely many closed , connected , orientable , irreducible 3-manifolds @xmath0 such that some embedding of @xmath2 in @xmath0 is pointwise fixed by an orientation reversing involution of @xmath0 . |
entropy @xmath0 as a function of extensive variables characterizes macroscopic properties of materials in a unified manner @xcite
. its functional form for a given material is completely determined by thermodynamic measurements of the heat capacity , the compressibility and the other relevant susceptibilities .
let @xmath1 be an unconstrained variable of a composite isolated system .
we denote the @xmath1 dependence of @xmath0 by @xmath2 .
then , the equilibrium value of @xmath1 is determined by @xmath3 , \label{ext}\ ] ] which is an example of the variational principle of thermodynamics .
furthermore , according to einstein s fluctuation theory @xcite , the probability density of @xmath1 is expressed as @xmath4 where @xmath5 is the boltzmann constant , and @xmath6 represents the asymptotic equivalence of the exponential factor in the thermodynamic limit .
more precisely , let @xmath7 be the number of the degrees of freedom .
then , the scaled entropy @xmath8 with @xmath9 fixed satisfies @xmath10 in the probability theory , the right - hand side provides a large deviation function ( up to the minus sign ) @xcite .
that is , the scaled entropy @xmath11 is equivalent to the large deviation function .
now , suppose that the scaled entropy @xmath11 is defined by ( [ rate ] ) with the principle of the equal weight for microscopic states .
then , the variational principle ( [ ext ] ) can be derived from the definition .
a surprise in this approach is that the entropy thus defined characterizes thermodynamic properties completely through the fundamental relation of thermodynamics . after establishing the connection to thermodynamics ,
one can obtain the scaled entropy by thermodynamic measurements .
in other words , the large deviation function , which characterizes macroscopic fluctuations , is obtained without measuring fluctuations .
such formulas are not restricted to the case of fluctuations in isolated systems . as demonstrated for a simple model in appendix [ appendix_thermodynamics ] ,
several formulas are derived .
one example is that a scaled cumulant generating function of a thermodynamic variable , which is given by the legendre transform of the large deviation function , is related to a free energy density in thermodynamics . from this relation ,
it is found that the derivative of the scaled cumulant generating function with respect to its argument is equal to the expectation value of the thermodynamic variable in a modified system with an extra external force added .
see ( [ g_equilibrium ] ) for such an example .
the relations between large deviation functions and thermodynamic functions are thought to be well - known , and the relations lead to fluctuation - response relations , one of which , connects the susceptibility and the intensity of fluctuations of magnetization @xcite .
the main difference of non - equilibrium systems from equilibrium systems is the existence of a current . in particular ,
the expectation value of the current in a linear response regime is connected with the intensity of fluctuations of time - averaged current in equilibrium @xcite .
this is one of fluctuation - dissipation relations , which might correspond to fluctuation - response relations in equilibrium statistical mechanics .
since the time - averaged current appears to be important at least for small fluctuations in the linear response regime , it is natural to consider the large deviation function of time - averaged current
. then , from the analogy with thermodynamics , one may expect that there is a formula that connects the large deviation function with a macroscopic quantity and that the fluctuation - dissipation relations are derived from the formula when the linear response regime is focused on .
indeed , onsager and onsager - machlup attempted to express the large deviation function of time - averaged current in terms of dissipation functions @xcite , as pointed out by oono @xcite .
this is successful for small fluctuations in the linear response regime , but the expression can not be applied to general cases with large fluctuations . in the last two decades
, the understanding of the large deviation function of time - averaged current has been substantially developed .
one remarkable result is the discovery of the fluctuation theorem which represents a simple and beautiful symmetry property of the large deviation function of time - averaged entropy production @xcite .
the symmetry property was understood from the microscopic reversibility of the system and it was generalized to identities for any quantities @xcite .
this progress enables us to revisit a linear response theory and to re - derive known formulas including a non - linear response formula @xcite and an expression of steady state distribution @xcite .
@xcite as a review of these re - derivations .
in addition to the re - derivation of the known formulas , the symmetry property plays an essential role in deriving an extension of the clausius relation @xcite on the basis of a neat expression of steady state distribution @xcite . as another direction of the progress
, the large deviation function of time - averaged current was explicitly calculated for lattice models @xcite .
the results evolved into a variational form , which was called an additivity principle @xcite . when the system behavior is assumed to be described by fluctuating hydrodynamics ,
the condition for the validity of the additivity principle can be derived from the analysis of the path - integral expression @xcite .
note that the additivity principle is less universal than the fluctuation theorem , but still characterizes a universal aspect of the large deviation functions of time - averaged current in a wide range class of systems .
@xcite for a recent related study . based on these developments ,
we consider a possibility that the large deviation function of time - averaged current is expressed by a time - averaged quantity as if the framework corresponds to einstein s fluctuation theory .
this formula , if it exists , is regarded as an extension of onsager s theory to that valid even for large fluctuations .
although the motivation seems natural , the problem would not be solved easily . nevertheless , since a new operational variational principle associated with the large deviation function of time - averaged current has been found recently @xcite , it is a good opportunity to study the problem .
in this paper , specifically , we study interacting brownian particles under non - equilibrium conditions .
the motion of the particles are assumed to be described by a set of langevin equations .
we then consider the large deviation function of time - averaged velocity of the particles .
the main results in this paper are formulas ( [ cumulantvariational_expression ] ) and ( [ mainclaim_of_the_paper ] ) .
the formula ( [ cumulantvariational_expression ] ) indicates that the scaled cumulant generating function , which is given by the legendre transform of the large deviation function , is determined from a variational function in a modified system with an inhomogeneous external force added .
essentially the same formula was reported in ref .
the formula ( [ mainclaim_of_the_paper ] ) claims that the derivative of the scaled cumulant generating function is equal to the time - averaged velocity in the modified system , which corresponds to ( [ g_equilibrium ] ) in equilibrium statistical mechanics .
the formula ( [ mainclaim_of_the_paper ] ) also turns out to be an extension of the fluctuation - dissipation relation .
since these new formulas take simple forms , we expect that our results advance the understanding of large deviation function of time - averaged quantity .
this paper is organized as follows . in section [ section_model ] , we introduce a model and define quantities we study in this paper . in section [ section_main ] , we first summarize main results of the paper and after that we derive them . in section
[ section_whatismaximized ] , we address three remarks on the variational function that appears in our main results . concretely , we re -
express the variational function in terms of entropy production rates , mention a similarity with the donsker - varadhan formula , and explain a connection to the principle of the least dissipation of energy . in section [ section_application ]
, we present a few applications of our formula by restricting our study to a single brownian particle on a ring .
first , on the basis of our formula , we re - derive exact formulas of the diffusion constant and the mobility in a compact manner .
second , starting from our formula , we show a variational expression associated with an additivity principle .
third , we propose a method for experimentally obtaining the scaled cumulant generating function .
the final section is devoted to concluding remarks .
some technical details and related subjects are discussed in appendices . in the argument below , the boltzmann constant @xmath5 is set to unity .
we consider @xmath7 colloidal particles immersed in a solvent at temperature @xmath12 .
we describe the system by a set of positions of the particles in a @xmath13-dimensional region @xmath14 . for mathematical simplicity ,
we impose periodic boundary conditions . as a typical example
, one may consider a three - dimensional region , where one direction is confined by walls and no walls in other directions .
even such a case is included in our model by setting up a periodic potential whose height is extremely large at the walls .
another example of the region @xmath14 is a one - dimensional circuit which can be generated by an optical potential .
we denote the positions of all the particles by @xmath15 , where @xmath16 represents the @xmath17-th component of the position of the @xmath18-th particle . in this paper , bold - face fonts ( e.g. @xmath19 ) represent @xmath20-dimensional vectors .
we also express the position of the @xmath18-th particle by the @xmath13-dimensional vector @xmath21 .
the force acting on the particles consists of three types of force , ( i ) an external force , ( ii ) an interaction force exerted from other particles , and ( iii ) a force from solvent molecules . in ( i ) , the external force is further divided into a homogeneous part @xmath22 and a periodic part determined by a potential @xmath23 . since the existence of @xmath24 makes the system out of equilibrium , we call @xmath24 a driving force . in ( ii ) , the interaction force is determined by a two - body interaction potential @xmath25 between the @xmath18-th particle and the @xmath17-th particle under the assumption that three- and higher - body interactions among particles are not taken into account .
the total potential @xmath26 is written as @xmath27 the total force @xmath28 that originates from ( i ) and ( ii ) is expressed as @xmath29 in ( iii ) , the force from solvent molecules is divided into two parts , a friction force and a fluctuating force , where we assume that hydrodynamic interactions are ignored . the friction force is proportional to the particle velocity @xmath30 so that it can be written in the form of @xmath31 with friction constants @xmath32 .
we assume that the fluctuating force is described as gaussian white noise satisfying the fluctuation - dissipation relation of the second kind .
that is , it is expressed by @xmath33 , where @xmath34 satisfies @xmath35 @xmath36 from these arguments , we obtain the equations of motion for colloidal particles . by neglecting inertial effects ,
these are written as @xmath37 where @xmath38 .
the equations are called langevin equations @xcite and have been studied in many situations including brownian motors @xcite , polymer dynamics @xcite and dynamics of mesoscale objects in solids .
see introduction of refs .
@xcite and chap .
11 of ref .
it should be noted that some recent discoveries of non - equilibrium relations have been tested in experiments corresponding to ( [ langevin ] ) with @xmath39 @xcite .
we assume that the particles obey a probability density @xmath40 at @xmath41 .
after that ( @xmath42 ) , the particles move according to ( [ langevin ] ) .
we denote by @xmath43 the expectation value of @xmath44 with respect to the initial distribution @xmath40 and history of @xmath34 .
we define the probability density of @xmath45 by @xmath46 here , the superscript @xmath47 of @xmath48 represents @xmath49 in ( [ langevin ] ) .
we use this superscript throughout the paper , because the force dependence plays an important role in our formulation . the evolution equation for @xmath50
is derived as @xmath51 with a probability current @xmath52 the equation is called the fokker - planck equation @xcite . the stationary distribution and current , which are denoted by @xmath53 and @xmath54 , respectively ,
are determined by @xmath55 @xmath56 the expectation value of a quantity @xmath57 $ ] in the steady state is denoted by @xmath58 .
mathematically , the expectation value is calculated by replacing initial distribution @xmath40 by @xmath59 . in this paper
, we particularly focus on statistical properties of time - averaged velocities @xmath60 the scaled cumulant generating function @xmath61 of the time - averaged velocities is defined by @xmath62 where @xmath63 . in the expansion form
@xmath64 the coefficient @xmath65 is related to the cumulant @xmath66 @xcite .
concretely , it can be shown that @xmath67 here , by substituting ( [ langevin ] ) into ( [ average_velocity ] ) and taking the expectation value , we obtain @xmath68 this leads to @xmath69 furthermore , by substituting ( [ stationarryrho ] ) into ( [ stationary3 ] ) and by eliminating boundary terms due to periodic boundary conditions , we obtain @xmath70
the main purpose of this paper is to find an expression of @xmath71 which shares a common structure in equilibrium statistical mechanics . as reviewed in appendix [ appendix_thermodynamics ] , the cumulant generating function of a thermodynamic variable
is related to a free energy function and this relation leads to the important result that the first derivative of the cumulant generating function is equal to the expectation value of the thermodynamic variable in a modified system .
see ( [ g_equilibrium ] ) for such an example .
we want to have a similar expression for @xmath71 .
since the relation containing ( [ g_equilibrium ] ) leads to the fluctuation - response relation for thermodynamic variables , we expect that there might be a formula similar to ( [ g_equilibrium ] ) , which leads to fluctuation - dissipation relations for currents . in order to investigate this possibility
, we focus on the linear response regime . first , by noting @xmath72 , we have @xmath73 here , the fluctuation - dissipation relation is written as @xmath74 see refs .
@xcite for a derivation .
see also refs .
@xcite for a recent understanding of the derivation on the basis of the fluctuation theorem .
the substitution of ( [ linear_response ] ) into the first term of the right - hand side of ( [ linear_response2 ] ) and the evaluation of the second term at @xmath75 yield @xmath76 by using ( [ linear_response ] ) again in the right - hand side of ( [ linear_response3 ] ) , we obtain @xmath77 this implies that when @xmath24 and @xmath78 are small , the first derivative of @xmath71 is equal to the expectation value of the time - averaged velocity in a modified system with an extra force @xmath79 added .
since ( [ linear_response ] ) is derived from ( [ linear_response2 ] ) and ( [ differential_linear ] ) , the expression ( [ differential_linear ] ) is equivalent to the fluctuation - dissipation relation ( [ linear_response ] ) .
now , we seek for an extension of ( [ differential_linear ] ) to that valid for any @xmath78 and @xmath24 . here
, one may recall the fluctuation theorem as an extension of the fluctuation - dissipation relation , which claims @xmath80 in the system under consideration @xcite . although ( [ linear_response ] ) is derived from ( [ fluctuation_teorem ] ) , a clear extension of ( [ differential_linear ] ) is not obtained as far as we attempted . putting aside the fluctuation theorem , we consider the problem from a different direction .
the basic idea is to consider a modified system with a spatially inhomogeneous extra force @xmath81 added .
precisely writing , the modified system is described by the langevin equation ( [ langevin ] ) with the replacement of @xmath49 by @xmath82 .
we then define a functional @xmath83 by @xmath84 \label{hannkannsuu}.\ ] ] the first result of this paper is a variational expression @xmath85 let @xmath86 be the optimal @xmath87 that maximizes ( [ hannkannsuu ] ) for given @xmath78 and @xmath88 .
that is , @xmath89
. \label{argment}\ ] ] the second result is a formula @xmath90 this is thought to be the generalization of ( [ differential_linear ] ) . indeed , there exists a positive function @xmath91 such that @xmath92 with a condition @xmath93 before deriving the main results , we confirm that the fluctuation - dissipation relation ( [ linear_response ] ) is obtained from our result ( [ mainclaim_of_the_paper ] ) with ( [ wopt_psi ] ) . we start with @xmath94 from a basic property that there is no persistent current in systems without a driving force , we have an identity @xmath95 this leads to @xmath96(f_j+2th_j ) \\ & \quad + \mathcal o ( ( { { \boldsymbol f}}+2t{{\boldsymbol h}})^2 ) .
\label{derivation_linear_response2 } \end{split}\ ] ] by using ( [ derivation_linear_response1 ] ) , we further rewrite the right - hand side of ( [ derivation_linear_response2 ] ) as @xmath97(f_j+2th_j ) + \mathcal o ( { { \boldsymbol f}}^2,{{\boldsymbol h}}^2,{{\boldsymbol f } } { { \boldsymbol h } } ) .
\label{derivation_linear_response3}\ ] ] thus , we arrive at @xmath98(f_j+2th_j ) \\ & + \mathcal o ( { { \boldsymbol f}}^2,{{\boldsymbol h}}^2 , { { \boldsymbol f } } { { \boldsymbol h } } ) .
\label{linear_derivation } \end{split}\ ] ] by differentiating ( [ linear_derivation ] ) with respect to @xmath99 and setting @xmath75 and @xmath100 , we obtain @xmath101.\ ] ] this is equivalent to ( [ linear_response ] ) .
the set of displacements of particles during the time interval @xmath102 $ ] is written as @xmath103 where it should be noted that @xmath104 by differentiating ( [ total_displacement ] ) with respect to @xmath105 , we obtain @xmath106 we denote the joint probability density of @xmath107 and @xmath108 by @xmath109 .
the idea of considering such a joint probability density of @xmath19 and @xmath110 ( or accumulated work , heat ) was employed in refs .
@xcite . from the langevin equations ( [ langevin ] ) and ( [ langevin_x ] )
, we derive the fokker - planck equation @xmath111 where the fokker - planck operator @xmath112 is defined by @xmath113 .
\label{fpoperator } \end{split}\ ] ] here , we consider a quantity @xmath114 which obeys @xmath115 where the operator @xmath116 is calculated as @xmath117 .
\label{modifiedfpoperator } \end{split}\ ] ] then , from ( [ xandv ] ) and ( [ definitionofq ] ) , we have @xmath118 let @xmath119 be the largest eigenvalue of @xmath116 .
( [ timeevolution_q ] ) and ( [ e^thv ] ) lead to the asymptotic form @xmath120 when @xmath105 is sufficiently large .
thus , from the definition of @xmath71 given in ( [ cumulantgeneratingfunction ] ) , we obtain @xmath121 the result that a scaled cumulant generating function is equal to the largest eigenvalue of an operator is standard @xcite . the fluctuation theorem was studied with the aid of this result @xcite .
recently , a relation similar to ( [ derivation_largesteigenvalue ] ) was employed for calculations of the large deviation function of entropy production @xcite , where a kink in the large deviation function has been pointed out .
@xcite for a recent related result .
we denote the adjoint operator of @xmath122 by @xmath123 , which is calculated as @xmath124 . \label{modifiedadjointfpoperator1 } \end{split}\ ] ] let @xmath125 and @xmath126 @xmath127 be all the eigenvalues of @xmath128 and @xmath123 , respectively .
we label the eigenvalues such that @xmath129 and @xmath130 for @xmath131 , where @xmath132 represents the real part of @xmath133 .
we may set in general @xmath134 with @xmath135 , where @xmath136 represents the complex conjugate of @xmath133 . as shown in appendix c ,
the largest eigenvalues @xmath119 and @xmath137 are real , and thus @xmath138 furthermore , the eigenfunctions of the largest eigenvalue are positive . from ( [ derivation_largesteigenvalue ] ) and ( [ largest_equal ] ) ,
we obtain @xmath139 where @xmath140 is the positive eigenfunction corresponding to the largest eigenvalue @xmath137 . here , by noting @xmath141 , we define a force @xmath142 as @xmath143 which corresponds to the cole - hopf transformation @xcite . by using trivial identities @xmath144
@xmath145 we rewrite ( [ eigenad ] ) as @xmath146 .
\label{cumulant } \end{split}\ ] ] now , we consider a modified system with an arbitrary external force @xmath81 added .
the stationary probability density in the modified system , @xmath147 , satisfies ( [ stationarryrho ] ) and ( [ stationarrycontinue ] ) with the replacement of @xmath88 by @xmath148 . by multiplying ( [ cumulant ] ) by @xmath149 , integrating it with respect to @xmath19 , and integrating by parts in the third term ,
we obtain @xmath150 .
\label{derivation_cumulant1 } \end{split}\ ] ] then , we substitute ( [ stationarryrho ] ) into ( [ derivation_cumulant1 ] ) and we replace @xmath151 by @xmath152 . furthermore , by substituting ( [ u_hdef ] ) into the term @xmath153 , we rewrite ( [ derivation_cumulant1 ] ) as @xmath154 \\ & + \int _ { \omega ^n } d{{\boldsymbol x}}\sum_{i=1}^{nd } \frac{p_{\rm st}^{{{{\boldsymbol f}}}+{{{\boldsymbol w}}}}({{{\boldsymbol x } } } ) } { 4t\gamma_i } \left ( u_{{{{\boldsymbol h}}},i}^{{{\boldsymbol f}}}({{{\boldsymbol x } } } ) - w_i({{{\boldsymbol x } } } ) \right ) ^2 , \label{before_g(h)experiment } \end{split}\ ] ] where we have used ( [ stationarrycontinue ] ) .
finally , by recalling ( [ averaged ] ) and ( [ hannkannsuu ] ) , we arrive at @xmath155 it should be noted that the external force @xmath156 in ( [ g(h)experiment ] ) is arbitrary . here ,
the second term of the right - hand side in ( [ g(h)experiment ] ) is not negative and becomes zero only when @xmath157 .
this leads to @xmath158 where the equality holds only when @xmath157 .
this result is the variational expression ( [ cumulantvariational_expression ] ) .
the optimal force @xmath159 defined by ( [ argment ] ) is now derived as @xmath160 next , the derivative of ( [ g(h)experiment ] ) with respect to @xmath161 yields @xmath162 .
\end{split}\ ] ] since @xmath87 is arbitrary , we may set @xmath163 in the expression .
this provides the result ( [ mainclaim_of_the_paper ] ) .
furthermore , from ( [ u_hdef ] ) and ( [ derivation_a ] ) , we obtain ( [ wopt_psi ] ) with ( [ derivation_linear_response1 ] ) , because the eigenfunction @xmath164 satisfies @xmath165 .
in this section , we present three remarks on the variational function @xmath167 . before entering the main part
, we first show a slightly different expression of @xmath167 .
since we have ( [ wopt_psi ] ) , we may restrict @xmath168 to the form @xmath169 throughout this section , the extra force @xmath81 is always connected to the potential function @xmath170 through ( [ deltaw_potential ] ) .
we then rewrite ( [ cumulantvariational_expression ] ) as @xmath171 where the optimal potential @xmath172 satisfies @xmath173 furthermore , by noting @xmath174 and by using ( [ averaged ] ) , we express @xmath175 as @xmath176 } .
\label{variational_expression1_2 } \end{split}\ ] ] the expression of ( [ cumulantvariational_expression2_2 ] ) with ( [ variational_expression1_2 ] ) was reported in ref .
@xcite .
this section is organized as follows . in section [ defference_of_entropy_production ] ,
we rewrite the variational function ( [ variational_expression1_2 ] ) as the difference between two entropy production rates . in section [ relation_to_donsker_varadhan ] , we point out a similarity between the donsker - varadhan formula and our formula
. finally , in section [ relationship_to_the_least ] , we show that the cumulant generating function of time - averaged current in a simple electric circuit is expressed in terms of the least energy dissipation rate associated with a variational principle that determines the voltage distribution . in this subsection
, we express the variational function ( [ variational_expression1_2 ] ) in terms of entropy production rate .
first , the probability current for a given probability density @xmath178 is written as @xmath179 where the slightly heavy notation @xmath180 is used in order to distinguish @xmath88 and @xmath181 dependence in the current . here , we define the functional @xmath182 as @xmath183 by substituting ( [ def_q ] ) into ( [ def_psi ] )
, we express @xmath184 as @xmath185 the integration by parts of the first term leads to @xmath186 we denote the first and second terms of the right - hand side in ( [ psi_expression ] ) by @xmath187 and @xmath188 , respectively .
then , @xmath189 is equal to the increasing rate of the shannon entropy and @xmath188 corresponds to the energy - dissipation rate divided by @xmath12 .
see appendix [ appendix_entropyproduction ] for a more detailed explanation .
since @xmath188 is identical to the increasing rate of the thermodynamic entropy of the heat bath , we identify @xmath190 with entropy production rate of the total system .
now , from ( [ def_q ] ) and ( [ def_psi ] ) , we calculate @xmath191 .
\label{sigma_and_q } \end{split}\ ] ] by noting @xmath192 and comparing ( [ sigma_and_q ] ) with ( [ variational_expression1_2 ] ) , we obtain @xmath193 .
\label{variational_expression3}\ ] ] the first term of the right - hand side in ( [ variational_expression3 ] ) is proportional to the entropy production rate of the modified system in the steady state .
the second term of ( [ variational_expression3 ] ) is a little bit complicated .
@xmath194 is the entropy production rate just after the extra force @xmath87 is turned off suddenly in the steady state of the modified system . in this manner ,
the expression ( [ variational_expression3 ] ) provides us a thermodynamic interpretation of the variational function @xmath166 . for a given path @xmath195
, we define the occupation density as @xmath196 we denote the probability density of @xmath197 by @xmath198 . we write it formally as @xmath199 when @xmath200 is sufficiently large , the asymptotic form of @xmath201 is given by @xmath202 which is the large deviation property of the occupation density with the large deviation function @xmath203 .
let @xmath204 be the adjoint fokker - planck operator defined by @xmath205 .
\label{adjoin_fokker_planck_dv}\ ] ] then , the donsker - varadhan formula in the langevin system is written as @xmath206 such a formula was rigorously derived for general markov stochastic processes @xcite .
its universal form suggests that it might be connected to a framework of non - equilibrium statistical mechanics .
indeed , in ref .
@xcite , it has been pointed out that the minimum entropy production principle is related to the donsker - varadhan formula ( [ dv_variational ] ) .
this relation was obtained by noting the detailed balance condition in equilibrium systems .
similarly , in equilibrium dynamics , the variational principle determining the activity was recognized as a form similar to the donsker - varadhan formula @xcite .
more recently , even without the detailed balance condition , the variational function in the donsker - varadhan formula was conjectured to be related to the activity @xcite . based on these achievements ,
we rewrite ( [ dv_variational ] ) so as to see a similarity with our variational function ( [ variational_expression3 ] ) .
first , since @xmath207 in ( [ dv_variational ] ) is positive , we define the potential @xmath170 by @xmath208 then , using ( [ cole - hopf - identity1 ] ) and ( [ cole - hopf - identity2 ] ) , we rewrite ( [ dv_variational ] ) as @xmath209 .
\label{relation_dv4 } \end{split}\ ] ] the integration by parts in the second term leads to @xmath210 .
\end{split}\ ] ] thus , by using ( [ def_q ] ) with the replacement of @xmath88 by @xmath211 , we obtain @xmath212 .
\label{relation_dv2 } \end{split}\ ] ] furthermore , by comparing ( [ relation_dv2 ] ) with ( [ sigma_and_q ] ) , we find @xmath213 . \label{relation_dv3}\ ] ] these expressions , ( [ relation_dv2 ] ) and ( [ relation_dv3 ] ) , were derived in ref .
@xcite by using another method . for the large deviation function of occupation density in jump processes ,
a similar expression in terms of escape rates instead of the entropy production rates has been reported in ref .
@xcite .
it is seen that there is a strong similarity between ( [ relation_dv3 ] ) and ( [ cumulantvariational_expression2_2 ] ) with ( [ variational_expression3 ] ) .
this suggests that two formulas might be described in a unified manner . we will report a unified treatment of the donsker - varadhan formula and ours in another paper
we consider an electric circuit in which @xmath214 resistances ( @xmath215 ) are connected in series .
we impose an electric potential @xmath216 on the circuit by using a battery as shown in the left side of fig .
[ fig1 ] .
resistances ( @xmath217 ) are connected in series .
we impose an electric potential @xmath216 on the circuit .
right : by using @xmath214 batteries , we impose an electric potential @xmath218 on each resistance @xmath219.,width=302 ] let @xmath220 be the difference of the electric potential at the resistance @xmath219 in the electric circuit .
this @xmath220 is determined by ohm s law @xmath221 where @xmath222 is the electric current in the circuit . from the constraint condition @xmath223 and ohm s law ( [ ohms_law ] ) , we obtain @xmath224 @xmath225 the principle of the least dissipation of energy claims that @xmath220 is characterized by the least dissipation of energy @xcite . in order to see this statement
, we consider an electric circuit displayed in the right side of fig .
[ fig1 ] . by using @xmath214 batteries
, we impose an electric potential @xmath218 on each resistance @xmath219 such that @xmath226 . under the assumption that the energy dissipation is determined by the joule heating , the energy dissipation rate @xmath227
is calculated as @xmath228 then , one can directly confirm that @xmath220 is determined as @xmath229 where @xmath230 means the constraint condition for @xmath231 .
this is an example of the principle of the least dissipation of energy .
we define the least energy dissipation rate @xmath232 as @xmath233 next , we consider fluctuations of the time - averaged electric current @xmath234 during a time interval @xmath235 $ ] . in a manner similar to ( [ cumulantgeneratingfunction ] ) , we define the scaled cumulant generating function of the time - averaged current as @xmath236 for simplicity , we assume that the current fluctuations are described by johnson noise .
see the first chapter of ref .
@xcite for a brief review of johnson noise .
the johnson noise satisfies the fluctuation dissipation relation @xmath237 we also assume that the fluctuations are gaussian . under these assumptions ,
we write @xmath238 as @xmath239 now , we connect the least energy dissipation rate @xmath232 and the cumulant generating function @xmath238 .
first , from ( [ voltage ] ) and ( [ leastenergydissipationrate ] ) , we find @xmath240 then , by replacing @xmath216 by @xmath241 in ( [ leastdissipation ] ) , we obtain @xmath242 where we have used ( [ electric_cumulantgenerating ] ) .
the subtraction of ( [ leastdissipation ] ) from ( [ leastdissipation2 ] ) leads to @xmath243 . \label{leastdissipation_difference}\ ] ] this result that @xmath238 is written by the difference of the energy dissipation rates for two different states reminds us of ( [ cumu_vs_gibbs ] ) in equilibrium statistical mechanics .
our formula ( [ cumulantvariational_expression2_2 ] ) with ( [ variational_expression3 ] ) is a generalization of ( [ leastdissipation_difference ] ) .
in this section , we present three applications of our result by restricting our investigation to a brownian particle on a ring . we write an explicit setting .
let @xmath244 be a position of the particle along the ring of size @xmath245 .
we assume that the motion of the particle is described by @xmath246 with the periodic boundary condition , where @xmath247 is a deterministic force given by @xmath248 such a force can be realized by using a rotating periodic potential @xcite .
note that ( [ 1langevin ] ) with ( [ 1langevin_force ] ) corresponds to the case that @xmath39 in the model studied in section [ section_model ] .
the stationary distribution @xmath249 and the stationary current @xmath250 are determined by @xmath251 with the periodic boundary condition @xmath252 and the normalization condition @xmath253 . for this system , in section [ section_diffusionconstant ] , we derive the diffusion constant @xmath254 and the mobility @xmath255 by utilizing our formula .
then , in section [ appendix_additivity ] , we show that our variational expression ( [ cumulantvariational_expression ] ) with ( [ hannkannsuu ] ) leads to a formula derived from an additivity principle @xcite .
finally , in section [ section_measurement ] , we present an experimental method for determining the cumulant generating function without the information of @xmath247 . in the system under study , the stationary distribution @xmath249 , the diffusion constant @xmath256 and the mobility @xmath257 are known to be expressed in terms of the following functions @xcite : @xmath258 concretely , the expressions are @xmath259 @xmath260 and @xmath261 where @xmath262 is the normalization constant given by @xmath263 here , one may wonder about the origin of such compact expressions .
below , within our formulation , we uncover a simple structure behind the expressions .
first , we consider the diffusion constant .
we start with the formula ( [ g(h)experiment ] ) with ( [ hannkannsuu ] ) . in the case
that @xmath39 , it becomes @xmath264 , \label{remarks_reimann_1 } \end{split}\ ] ] where it should be noted that @xmath265 is arbitrary . by setting @xmath266 in ( [ remarks_reimann_1 ] ) , we obtain @xmath267 since @xmath268 ( see ( [ derivation_linear_response1 ] ) and ( [ u_hdef ] ) ) , @xmath269 is expanded as @xmath270 the substitution of ( [ remarks_reimann_3 ] ) into ( [ remarks_reimann_2 ] ) yields @xmath271 which leads to @xmath272 next , we consider the mobility .
let us fix a function @xmath273 . by setting @xmath274 in ( [ remarks_reimann_1 ] ) ,
we rewrite ( [ remarks_reimann_1 ] ) as @xmath275 we then differentiate ( [ remarks_derivatives_1 ] ) with respect to force @xmath44 .
the result is @xmath276 since @xmath277 is arbitrary in this expression , we may set @xmath278 .
this leads to @xmath279 by substituting ( [ remarks_reimann_3 ] ) , ( [ remark_reimann_8 ] ) into ( [ remarks_derivatives_2 ] ) , we obtain @xmath280 as shown in ( [ remark_reimann_6 ] ) and ( [ remark_derivative_3 ] ) , @xmath256 and @xmath257 are expressed by the second and the first moments of @xmath281 , respectively . here , we determine @xmath281 .
we substitute ( [ remarks_reimann_3 ] ) into ( [ cumulant ] ) and extract the terms proportional to @xmath282 .
we then obtain @xmath283 note the periodic boundary condition @xmath284 and the normalization condition @xmath285 which comes from ( [ u_hdef ] ) . by comparing ( [ determination_of_u1 ] ) with ( [ 1d - statonarrycurrent ] ) , we find that @xmath281 can be solved in the manner similar to @xmath249 .
the result is @xmath286 by substituting ( [ remark_reimann_5 ] ) and ( [ remark_reimann_4 ] ) into ( [ remark_reimann_6 ] ) and ( [ remark_derivative_3 ] ) , we obtain ( [ remark_reimann_9 ] ) and ( [ remark_derivative_3_a ] ) .
furthermore , we can systematically evaluate higher - order cumulant coefficients by formulating an iterative relation .
see appendix [ appendix_iteratibe_expression ] . at the end of this subsection
, we study the violation of the fluctuation - dissipation relation .
first , from ( [ stationary3 ] ) , we have @xmath287 we substitute ( [ disccusion_of_fdt ] ) into ( [ remarks_derivatives_2 ] ) and combine it with our main result ( [ mainclaim_of_the_paper ] ) .
we then obtain @xmath288 which leads to @xmath289 furthermore , by considering an expansion form @xmath290 we rewrite the right - hand side of ( [ discussion_of_fdt3 ] ) as @xmath291 the expression ( [ degree_of_violation ] ) represents a degree of the violation of the fluctuation - dissipation relation .
recently , several attempts have been presented so as to characterize the violation of fluctuation - dissipation relation @xcite .
it might be interesting to find a connection of ( [ degree_of_violation ] ) with them .
when the driving force @xmath44 is absent , @xmath292 becomes zero so that the fluctuation - dissipation relation holds .
we show this fact explicitly .
we substitute ( [ pexpansion ] ) into ( [ 1d - statonarrycurrent ] ) with the replacement of @xmath247 with @xmath293 , and extract the terms proportional to @xmath282 .
we then obtain the following differential equation for @xmath292 : @xmath294 = 0 .
\label{q_1determine}\ ] ] with the periodic boundary condition @xmath295 and the normalization condition @xmath296 which comes from ( [ pexpansion ] ) . here , from ( [ remark_reimann_4 ] ) and ( [ remark_reimann_5 ] )
, we find that @xmath297 is constant in @xmath298 when @xmath299 . in this case , ( [ q_1determine ] ) provides us @xmath300 , where @xmath301 is a constant .
then , ( [ q_1_normalization ] ) leads to @xmath302 .
thus , recalling ( [ degree_of_violation ] ) , we conclude that the right - hand side of ( [ discussion_of_fdt3 ] ) is equal to zero when @xmath44 is absent .
note that the discussion above applies to many - body systems and higher - dimensional cases .
we start with the variational expression ( [ cumulantvariational_expression ] ) with ( [ hannkannsuu ] ) .
since the probability current is independent of @xmath298 in the one - dimensional case , we can rewrite ( [ cumulantvariational_expression ] ) as @xmath303 , \label{relation_additivity1}\ ] ] where @xmath304 and @xmath305 are the stationary probability current and density in the modified system .
note that @xmath304 and @xmath305 can be determined uniquely from ( [ 1d - statonarrycurrent ] ) for a given extra force @xmath265 .
conversely , when a constant current @xmath306 and a probability density function @xmath307 are given , we can define the extra force @xmath265 by @xmath308 thus , there is a one - to - one correspondence between @xmath265 and @xmath309 .
from this fact , we can rewrite ( [ relation_additivity1 ] ) as @xmath310 . \end{split}\ ] ] by taking the maximum of @xmath307 first , we obtain @xmath311 \bigg ] .
\end{split}\ ] ] therefore , we find that the legendre transform of @xmath312 with respect to @xmath313 , which is the large deviation function @xmath314 , becomes @xmath315 .
\end{split}\ ] ] this expression is the same as the variational expression derived from the additivity principle @xcite .
when the system is described by fluctuating hydrodynamics , the variational expression is obtained under the assumption that the most dominant contribution to a path integration is given as a stationary configuration @xcite . here , since the langevin equation ( [ langevin ] )
can be mapped to a stochastic partial differential equation @xcite , which is equivalent to fluctuating hydrodynamics , we can derive the variational expression in our system if the assumption is satisfied .
our result ( [ additivity ] ) indicates that the assumption is valid . in the derivation of ( [ additivity ] )
, the spatial homogeneity of the stationary current plays an important role .
however , since this property is specific to one - dimensional systems , it is not obvious whether the same variational principle does hold in higher - dimensional cases @xcite .
furthermore , by noticing that the probability current and the probability density considered here are equivalent to the particle current and the particle density in non - interacting particle systems , one might find that it is a highly non - trivial problem to derive ( [ additivity ] ) for interacting particle systems .
we consider an experimental method for determining @xmath312 without the information of @xmath247 .
since the direct evaluation of cumulants becomes harder and harder as the order of the cumulants increases , it may be useful if we could present an efficient method for evaluating @xmath312 on the basis of our formula under the assumption that we measure only the trajectories @xmath316 and values of @xmath12 and @xmath317 .
the basic idea is to utilize ( [ g(h)experiment ] ) , which was a cornerstone in the derivation of the main results ( [ cumulantvariational_expression ] ) and ( [ mainclaim_of_the_paper ] ) .
first , we add some external forces @xmath265 to the system and measure trajectories of the particle .
we then notice that the left - hand side of ( [ g(h)experiment ] ) is independent of external forces @xmath265 .
since we can estimate the right - hand side of ( [ g(h)experiment ] ) by using the measured data , we obtain equations which yield @xmath318 .
essentially the same idea was reported in ref .
below , we present a slightly different one from that in ref .
@xcite . concretely , in terms of a complete set of orthogonal functions @xmath319
, we expand @xmath320 as @xmath321 where @xmath322 is a truncation number of the expansion . as @xmath322 is increased , the accuracy of the expansion ( [ expansion1 ] ) increases . for later convenience
, we choose @xmath323 .
since @xmath324 ( see ( [ wopt_psi ] ) ) , the coefficient @xmath325 is derived as @xmath326 we shall determine the other expansion coefficients @xmath327 . first , by differentiating ( [ g(h)experiment ] ) with respect to @xmath282 , we obtain @xmath328 we then expand @xmath265 as @xmath329 by substituting ( [ expansion1 ] ) and ( [ expansion2 ] ) into ( [ 1g(h ) ] ) , we rewrite ( [ 1g(h ) ] ) as @xmath330 here , by defining @xmath331 we rewrite ( [ 1experiment1 ] ) as @xmath332 where we have used ( [ conditionua_0 ] ) .
we prepare @xmath333 external forces @xmath334 .
recalling that ( [ gdiffexperiment ] ) is valid for any external force @xmath265 , we equate the right - hand side of ( [ gdiffexperiment ] ) for the case @xmath335 to the right - hand side of ( [ gdiffexperiment ] ) for the other cases @xmath336 @xmath337 .
the obtained equations are @xmath338 where @xmath339 .
we further define an @xmath340 matrix as @xmath341 with the assumption that @xmath342 is invertible , we derive @xmath343 . \label{rennritu } \end{split}\ ] ] this expression implies that @xmath344 is expressed in terms of @xmath345 , @xmath346 and @xmath347 .
now , from trajectories of the particle in the modified systems with the external forces @xmath336 added , we approximately obtain @xmath346 and @xmath347 in the form @xmath348 @xmath349 with @xmath350 , @xmath351 .
note that the approximation becomes exact in the limit @xmath352 .
therefore , we can express @xmath344 by @xmath345 from experimental data . by integrating @xmath353 with the condition @xmath354 , @xmath355 , we obtain @xmath345 as a function of @xmath282 .
similarly , ( [ gdiffexperiment ] ) leads to the expression of @xmath356 in terms of @xmath357 . the integration yields @xmath312 , where we have used @xmath358 .
we performed a numerical experiment of the driven brownian particle on a ring subject to a periodic potential @xmath359 .
we set @xmath360 @xmath361 , @xmath362 @xmath363 .
the @xmath333 external forces @xmath364 are chosen as @xmath365 , where @xmath366 is a constant . following the recipe we described above , we obtain @xmath312 and @xmath318 from trajectories of the particle .
the results are displayed in fig .
[ figexperiment ] .
we also calculate @xmath312 and @xmath318 by approximately evaluating the largest eigenvalue and the corresponding eigenfunction of the operator @xmath367 .
see appendix [ appendix_largesteigenvalue ] for the calculation method .
these two results are close to each other .
it turns out that our experimental determination method is useful to obtain the accurate value of @xmath312 from measured trajectories .
( up ) and @xmath368 with @xmath369 ( down ) .
quantities are converted to dimensionless forms by setting @xmath370 .
we fix @xmath371 and @xmath372 .
we assumed to know @xmath373 and experimentally determined @xmath312 and @xmath374 from trajectories @xmath375 following the method described in the text .
we set @xmath376 , @xmath377 and @xmath378 . by taking @xmath379 samples
, we estimated @xmath312 and @xmath368 .
the obtained results are displayed with green dashed lines .
error - bars are within the lines .
the red lines were obtained from the evaluation of the largest eigenvalue of @xmath367 .
, title="fig:",width=302 ] ( up ) and @xmath368 with @xmath369 ( down ) .
quantities are converted to dimensionless forms by setting @xmath370 .
we fix @xmath371 and @xmath372 .
we assumed to know @xmath373 and experimentally determined @xmath312 and @xmath374 from trajectories @xmath375 following the method described in the text .
we set @xmath376 , @xmath377 and @xmath378 . by taking @xmath379 samples
, we estimated @xmath312 and @xmath368 .
the obtained results are displayed with green dashed lines .
error - bars are within the lines .
the red lines were obtained from the evaluation of the largest eigenvalue of @xmath367 .
, title="fig:",width=302 ]
in this paper , we have presented the new formula ( [ mainclaim_of_the_paper ] ) for the cumulant generating function of time - averaged velocity .
the formula enables us to determine the cumulant generating function in terms of the expectation value of time - averaged current in a modified system .
this corresponds to ( [ g_equilibrium ] ) in the example of equilibrium statistical mechanics , and it is also regarded as an extension of the fluctuation - dissipation relation .
furthermore , we have pointed out that our formula is related to the additivity principle , the donsker - varadhan formula , and the least dissipation principle . before ending the paper ,
we address problems that should be studied in future .
first , since our formulas are rather formal , we should explore further applications of our formulation to various physical problems . with regard to this aspect , in appendix [ appendix_simplemodel ] ,
we present a simple model in which we can explicitly calculate the cumulant generating function and the optimal force .
we hope that this model might help readers to grasp our formulation and promote the applications . furthermore , although our formulation is valid for interacting particle systems , useful predictions are not presented yet .
a key step in the analysis of many - body problems is to extract slowly varying effective variables from many - body distribution functions . here
, we may employ some approximations for extracting effective variables .
the result of the approximate calculation with our formulation might provide a new insight into properties of current fluctuations because our formulation is qualitatively new .
next , we ask the range of the applicability of our formulation
. the first question may be to consider similar formulas in under - damped langevin systems .
the formulas that correspond to ( [ cumulantvariational_expression ] ) and ( [ mainclaim_of_the_paper ] ) can be derived in a straight - forward manner .
this fact is not surprising at all when the formulation of this paper is properly understood .
however , it may be difficult to find a physical interpretation of the formulas .
for example , it seems that `` the extra force '' appearing in the variational formula can not be realized in laboratory experiments .
more importantly , as far as we attempted , the variational function may not be expressed in terms of the entropy production rates , which is different from the over - damped case .
we do not understand whether or not these are fatal in our formulation . we need further investigations . here , with regard to the range of the applicability of our formulas , we restrict our concerns to a formal aspect by putting aside physical interpretations . concretely , in markov jump processes , we can derive the formulas that correspond to ( [ cumulantvariational_expression ] ) and ( [ mainclaim_of_the_paper ] ) . as seen in ( [ jump2 ] ) and ( [ jump4 ] ) in appendix [ appendix_markov_jump ] , the mathematical expressions are quite simple .
we thus expect that there might be physical examples for which these formulas are useful .
furthermore , since these are expressed in terms of escape rates , it is conjectured that similar variational principles are formulated in many - body hamiltonian systems .
it is stimulating to prove this conjecture . by combining physical interpretations and formal aspects ,
we wish to seek for a universal structure behind our formulation . finally , we go back to our basic motivation .
our naive idea is that the large deviation theory for current fluctuations might be associated with an operational framework .
this was conjectured from the well - established fact that the large deviation theory for fluctuations of thermodynamic variables is tightly connected to equilibrium thermodynamics . the progress developed in this paper
does not provide a final answer to the problem , but the results are encouraging us to continue our efforts . in the best case , we might have an operational framework of thermodynamics for currents . then , we will consider a possibility to unify such a framework with equilibrium thermodynamics .
it may be plausible that a unified scheme is related to the so - called steady state thermodynamics @xcite .
with dreaming such possibilities , we challenge to study operational viewpoints in non - equilibrium statistical mechanics .
the authors thank h. tasaki for continuous discussions on non - equilibrium statistical mechanics .
they also thank f. van wijland for indicating the exactly calculable model in appendix [ appendix_simplemodel ] .
the present study was supported by grants from the ministry of education , culture , sports , science , and technology of japan , nos .
22340109 and 23654130 .
we review the relation between the cumulant generating function of a thermodynamic variable and a thermodynamic function .
specifically , we consider a spin model with finite range interaction in a finite dimensional lattice @xmath380 .
let @xmath381 , @xmath382 , be a spin variable whose value is either @xmath383 or @xmath384 .
we denote @xmath385 by @xmath386 collectively .
we assume that the hamiltonian for the system under a magnetic field @xmath387 is given by @xmath388 where @xmath257 is the magnetic moment . according to equilibrium statistical mechanics , for any observable @xmath389
, the expectation value @xmath390 in the system at temperature @xmath12 is given by @xmath391 where @xmath392 as an observable , we consider the magnetization density @xmath393 where @xmath394 .
now , we define the free energy density @xmath395 by @xmath396 then , the fundamental relation of thermodynamics is written as @xmath397 where @xmath398 is the entropy density and @xmath322 is the expectation value of the magnetization density @xmath399 .
notice that @xmath395 is determined from the measurement of the heat capacity @xmath400 and @xmath401 .
putting aside the thermodynamic function , we consider the scaled cumulant generating function of @xmath399 , which is defined as @xmath402 from ( [ gibbs_free_energy ] ) and ( [ zikacumulant ] ) , we find the following relation between the cumulant generating function and the free energy : @xmath403 \label{cumu_vs_gibbs}.\ ] ] in particular , from the thermodynamic relation ( [ thermodynamic_relation ] ) , we obtain @xmath404 for any @xmath282 . by differentiating ( [ g_equilibrium ] ) with respect to @xmath282 and set @xmath405 , we obtain @xmath406 which is referred to as a fluctuation - response relation @xcite .
we show that the largest eigenvalue of the operators @xmath407 and @xmath408 is real and that the corresponding eigenfunctions are positive .
the argument presented here is not mathematically rigorous , but provides a practical idea for understanding the claim .
although we focus on one - dimensional case for simplicity , the extension to higher - dimensional cases are straight - forward .
we express the operator @xmath409 in the form @xmath410 where @xmath411 .
we study the eigenvalue problem of @xmath409 in the space of periodic functions defined on the interval @xmath412 $ ] .
we define @xmath413 , @xmath414 , and let @xmath415 be the distance between @xmath416 .
we replace the differential operator @xmath417 and @xmath418 by differences .
concretely , the following replacements are considered : @xmath419 @xmath420 for @xmath421 .
a special care is needed for the cases @xmath422 and @xmath423 .
for example , @xmath424 @xmath425 then , we define an @xmath426 matrix @xmath427 such that @xmath428 where the @xmath7 dimensional vector @xmath429 is defined by @xmath430 .
since @xmath427 is an approximation of the operator @xmath409 , we assume that the eigenvalues and eigenfunctions of @xmath409 are determined from the limit @xmath431 for the eigenvalues and eigenvectors of @xmath427 .
thus , below , we study the eigenvalue problem of @xmath432 instead of the eigenvalue problem of @xmath409 .
we show that the largest eigenvalue of @xmath427 is real and that all components of the corresponding left and right eigenvectors are positive , when @xmath7 is sufficiently large .
this claim is understood from the perron - frobenius theory .
a real matrix @xmath433 is called an irreducible ml - matrix @xcite if @xmath433 satisfies the following two conditions : ( i ) all non - diagonal components of @xmath433 are non - negative .
( ii ) there exists a positive number @xmath434 such that the matrix @xmath435 is irreducible , where @xmath234 is the unit matrix .
the perron - frobenius theory leads to the statement that the irreducible ml - matrix @xmath433 has an eigenvalue @xmath436 that satisfies the following conditions : 1 .
@xmath436 is real .
the associated left and right eigenvectors are strictly positive and unique up to constant multiples . 3 . for any eigenvalue @xmath257 except for @xmath436 ,
the real part of @xmath257 is less than @xmath436 .
see theorem 2.6 of ref .
@xcite . since @xmath411 , the matrix @xmath427 for sufficiently large @xmath7 turns out to be an irreducible ml - matrix .
we thus obtain the claim that the largest eigenvalue of @xmath427 is real and that all components of the corresponding left and right eigenvectors are positive .
we explain thermodynamic interpretations of the first and second terms of ( [ psi_expression ] ) , which were denoted by @xmath187 and @xmath188 in section [ defference_of_entropy_production ] , respectively .
we first consider the first term , @xmath187 .
we assume that the system is set under the force @xmath88 and that the probability density is given by @xmath181 .
we define a probability density @xmath438 as the solution of @xmath439 with the initial condition @xmath440 here , we take the limit @xmath441 in ( [ time_derivative_p ] ) . by noting ( [ initial_condition_p ] ) in the right - hand side of ( [ time_derivative_p ] )
, we obtain @xmath442 thus , the first term of ( [ psi_expression ] ) , @xmath187 is expressed as @xmath443 \frac{\partial}{\partial t } p^{{{\boldsymbol f}}}({{\boldsymbol x}},t ) \right \ } .
\label{appendix_s}\ ] ] since @xmath444 , ( [ appendix_s ] ) is further rewritten as @xmath445 this shows that @xmath189 is the increasing rate of the shannon entropy for a given distribution @xmath181 .
next , we consider the second term , @xmath446 . we shall show that @xmath446 is the expectation value of the energy dissipation rate divided by @xmath12 when the distribution is given by @xmath181 . according to the energetic interpretation of langevin equations @xcite , the energy dissipation rate @xmath447 for each realization
is given by @xmath448 where the symbol @xmath449 represents the multiplication in the sense of the stratonovich @xcite .
it should be noted that @xmath450 is a stochastic variable . by substituting ( [ langevin ] ) into ( [ dissipation_rate_sekimoto ] )
, we write @xmath451 here , we change the multiplication rule from the stratonovich type ( @xmath449 ) to the ito type ( @xmath452 ) @xcite .
then , ( [ dissipation_rate_sekimoto2 ] ) becomes @xmath453 by taking the expectation value of ( [ appendix_psihat ] ) and integrating by parts the second term , we obtain @xmath454 where we have used @xmath455 .
the last expression claims that @xmath446 is the expectation value of the energy dissipation rate divided by @xmath12 when the distribution is given by @xmath181 .
for the system studied in section [ section_application ] , we determine the cumulant coefficients in an iterative manner .
we start with the expression @xmath456 which is obtained from ( [ cumulant ] ) with @xmath457 , where @xmath269 satisfies @xmath458 @xmath459 from a trivial relation @xmath460 , \end{split}\ ] ] we rewrite ( [ cumuderivation1 ] ) as @xmath461 \\ & = e^ { \int _ { 0}^{x}dyf(y)/t } \left ( 2\gamma g^f(h ) - \frac{1}{2t}u^f_h(x)^2 \right ) .
\label{iterative1 } \end{split}\ ] ] by integrating ( [ iterative1 ] ) with respect to @xmath298 , we obtain @xmath462 , \label{cumuderivation4 } \end{split}\ ] ] where @xmath366 is an integration constant . here , by direct calculation , we can derive an identity @xmath463 \label{cumuderivation6 } \end{split}\ ] ] for any periodic function @xmath464 .
we then set @xmath465 in ( [ cumuderivation6 ] ) and substitute it into the second term of the right - hand side of ( [ cumuderivation4 ] ) .
the result is @xmath466 , \label{d9 } \end{split}\ ] ] where @xmath467 is a constant .
the condition @xmath468 leads to @xmath469 . through the transformation of the integration variable @xmath470 , we rewrite ( [ d9 ] ) as @xmath471 , \label{cumuderivation7 } \end{split}\ ] ] where we have used ( [ iplusminus ] ) . by integrating the both sides of ( [ cumuderivation7 ] ) with respect to @xmath298 and using the condition ( [ cumuderivation2 ] )
, we derive @xmath472 where @xmath262 was defined in ( [ ipm_alpha ] ) .
it should be noted that @xmath269 satisfies a self - consistent equation .
indeed , by substituting ( [ g(h)iteration ] ) into ( [ cumuderivation7 ] ) , we find @xmath473 . \label{u(x)iteration } \end{split}\ ] ] by substituting ( [ remarks_reimann_3 ] ) into ( [ u(x)iteration ] ) , we can determine the coefficient @xmath474 iteratively . concretely , by noting @xmath475 we derive a recursion equation @xmath476 \end{split}\ ] ] for @xmath477 .
for example , by recalling ( [ remark_reimann_5 ] ) , we calculate @xmath478 .
\end{split}\ ] ] furthermore , we expand @xmath312 as @xmath479 by substituting ( [ u_hsquare ] ) and ( [ g_hexpansion ] ) into ( [ g(h)iteration ] ) , we obtain @xmath480 and @xmath481 for @xmath482 . with this formula ,
the cumulant coefficients @xmath483 can be calculated in an iterative manner .
for example , the third - order cumulant coefficient is derived as @xmath484 reimann _ et al _ found the exact expression of the diffusion constant ( [ remark_reimann_9 ] ) from the first and second cumulants of first passage time @xcite . here , we remember an iterative formula for the moments of the first passage time @xcite . by comparing this iterative formula with our iterative formula
, one might find a general relationship between the cumulant generating function and statistical properties of the first passage time beyond the first two cumulants .
we focus on the case that @xmath486 and @xmath487 , which is studied in section [ section_application ] .
we numerically solve the largest eigenvalue problem ( [ eigenad ] ) .
first , we consider a linear equation @xmath488 since @xmath312 is the largest eigenvalue of @xmath489 , the long time behaviors of solutions are described as @xmath490 now , we assume @xmath491 for any @xmath105 by choosing appropriate initial conditions . then , we define @xmath492 by dividing ( [ phi_zikannhattenn ] ) by @xmath493 and using the same method deriving ( [ cumulant ] ) in section iii.a , we obtain @xmath494 finally , by differentiating ( [ logphi_timeevolution ] ) with respect to @xmath298 , we rewrite ( [ logphi_timeevolution ] ) as @xmath495 with a current @xmath496 let @xmath497 be the stationary solution given as the long - time limit of @xmath498 .
then , from ( [ phi_zennkin ] ) and ( [ u(x , t)def ] ) , we obtain @xmath499 by recalling ( [ wopt_psi ] ) , we have @xmath500 furthermore , by comparing ( [ experiment_current ] ) with ( [ cumulant ] ) , we obtain @xmath501 therefore , by numerically solving ( [ experiment_continue ] ) , we can calculate @xmath312 and @xmath485 .
we illustrate our theoretical idea by explicitly deriving the cumulant generating function and the optimal force for a linear langevin equation .
the model describes the motion of a single brownian particle in two dimensions , which corresponds to the case that @xmath486 and @xmath502 in ( [ langevin ] ) . by setting @xmath503 for notational simplicity , we write the equation as @xmath504 with @xmath505 where @xmath506 . from ( [ stationarryrho ] ) and ( [ stationarrycontinue ] ) , we obtain the stationary distribution as @xmath507 with a normalization constant @xmath366 .
although the stationary distribution is gaussian , the detailed balance condition is not satisfied because of the circulation force @xmath508 .
one can check this fact immediately from ( [ stationarryrho ] ) and ( [ stationarydist ] ) .
it should be noted that the linear langevin equation was studied with a novel idea , irreversible circulation , which characterizes the violation of detailed balance @xcite .
the quantity we consider here is a time - averaged angular momentum ( or irreversible circulation ) defined as @xmath509 for the scaled cumulant generating function of @xmath245 , @xmath510 we can derive the variational principle by using the same method as section [ subsection_derivation ] . first , we define an operator @xmath511 as @xmath512 since @xmath238 is equal to the largest eigenvalue of the operators @xmath511 and @xmath513 , we have @xmath514 where @xmath207 is the positive eigenfunction corresponding to the largest eigenvalue of @xmath513 .
next , we define a force @xmath515 as @xmath516 by dividing ( [ eigenproblem ] ) by @xmath207 and using the definition of @xmath515 , we obtain @xmath517 this takes the same form as ( [ cumulant ] ) so that we can follow the argument from ( [ cumulant ] ) to ( [ g(h)experiment ] ) .
we then arrive at @xmath518 with @xmath519 here , it should be noted that @xmath520 ( [ key ] ) is the key identity of our formulation , which corresponds to ( [ g(h)experiment ] ) .
indeed , from ( [ key ] ) we obtain @xmath521 @xmath522 where the optimal force @xmath523 is given as @xmath524 finally , by solving ( [ aftercolehopf ] ) , we derive the explicit expression of @xmath515 and @xmath238 .
suppose that @xmath525 where @xmath434 is a constant .
this yields @xmath526 by substituting ( [ f19 ] ) into ( [ aftercolehopf ] ) , we obtain @xmath527.\ ] ] since the left - hand side is independent of @xmath19 , we determine the constant @xmath434 as @xmath528 here , we can confirm that ( [ eigenproblem ] ) holds for ( [ f18 ] ) with ( [ optimal_a ] ) , because ( [ eigenproblem ] ) is equivalent to ( [ aftercolehopf ] ) .
we then obtain @xmath529 from ( [ f19 ] ) and ( [ optimal_a ] ) , the force in the optimal system is explicitly written as @xmath530
the model we consider here is a continuous - time markov process of a finite space @xmath531 .
we assume that the transition rate , @xmath532 for @xmath533 , is irreducible .
we set @xmath534 for notational simplicity .
the escape rate @xmath535 is then defined by @xmath536 let @xmath537 be a probability of @xmath538 , where the superscript @xmath539 represents the transition rate dependence of the probability @xmath540 .
then , @xmath537 satisfies the master equation @xmath541 p^{w}({{\boldsymbol a}}^{\prime } , t).\ ] ] for later convenience , we define a matrix by @xmath542 the stationary distribution @xmath543 satisfies @xmath544 we denote by @xmath545 the expectation value of a function @xmath546 with respect to the stationary distribution @xmath547 .
we fix a time interval @xmath235 $ ] . for a given history @xmath548 , where @xmath549 , a sequence of transition times @xmath550 is determined , and @xmath551 is expressed as @xmath552 for @xmath553 with @xmath554 and @xmath555 . here
, we consider a quantity @xmath556 defined for the jump event @xmath557 at @xmath558 ( @xmath559 ) .
its time - averaged value in the history is written as @xmath560 we then study the scaled cumulant generating function of @xmath561 , which is defined by @xmath562 where @xmath563 represents the sample average over realizations of history with an initial distribution .
it is known that @xmath238 is equal to the largest eigenvalue of the matrix @xmath564 given by @xmath565 see ref .
@xcite for the derivation .
let @xmath566 be the left eigenvector for the largest eigenvalue of @xmath564 .
that is , @xmath567 since @xmath568 ( see appendix [ appendix_largest_eigenvalue ] ) , we may define a potential function @xmath569 as @xmath570 dividing the both sides of ( [ eigenvalue_equation_jump ] ) by @xmath566 , we obtain @xmath571}. \label{eigenvalue_equation_jump2 } \end{split}\ ] ] here , for any potential function @xmath572 , we define a modified transition rate @xmath573 as @xmath574 the corresponding escape rate @xmath575 becomes @xmath576 and ( [ eigenvalue_equation_jump2 ] ) is rewritten as @xmath577 this expression might be known , although we do not find its explicit presentation in previous studies .
the derivation is the following .
first , we consider the variation @xmath581 in @xmath582 , which is expressed as @xmath583\\ & = \sum_{{{\boldsymbol a } } } \delta p_{\rm st}^{u_h^v}({{\boldsymbol a } } ) \left [ \lambda^{v}_h({{\boldsymbol a } } ) - \lambda({{\boldsymbol a } } ) \right ] + \sum_{{{\boldsymbol a } } } p_{\rm st}^{u_h^v}({{\boldsymbol a } } ) \delta \lambda^{v}_h({{\boldsymbol a } } ) .
\label{jumpvariation } \end{split}\ ] ] from ( [ jumpmodifiedtransition ] ) and ( [ jumpmodifiedescaperate ] ) , we rewrite the second term of ( [ jumpvariation ] ) as @xmath584.\ ] ] since this becomes zero from the condition of the stationary distribution , the variational equation @xmath585=0 $ ] leads to @xmath586 where @xmath227 is a constant corresponding to a lagrange multiplier . here , ( [ jump3 ] )
is equivalent to @xmath587 by comparing this with ( [ eigenvalue_equation_jump ] ) , we find that @xmath588 as the result of the perron - frobenius theory .
next , we consider the modified system with the transition rate @xmath589 . by multiplying the both sides of ( [ jump1 ] ) by @xmath590 and summing over @xmath591
, we obtain @xmath592 where we have used @xmath588 .
we thus conclude ( [ jump2 ] ) .
furthermore , by differentiating ( [ jump1 ] ) with respect to @xmath282 , multiplying the both sides by @xmath590 , and summing over @xmath591 , we derive @xmath593 the second term of ( [ jump_diff1 ] ) turns out to be zero from the condition of the stationary distribution .
the final expression is ( [ jump4 ] ) . at the end of this appendix , we point out that the expression ( [ jump2 ] ) is similar to the formula of the large deviation function of the empirical measure in markov jump processes @xcite
. this similarity corresponds to the relation between ( [ variational_expression3 ] ) and ( [ relation_dv3 ] ) in langevin equations .
d. n. zubarev , _ nonequlibrium statistical thermodynamics _ ( consultants bureau , new york , 1974 )
. j. a. mclennan , _ introduction to nonequilibrium statistical mechanics _
( prentice - hall , englewood cliffs , nj , 1989 ) .
t. s. komatsu , n. nakagawa , s .-
sasa , and h. tasaki , phys .
lett . * 100 * , 230602 ( 2008 ) .
t. s. komatsu , n. nakagawa , s .-
sasa , and h. tasaki , j. stat . phys . *
142 * , 127 ( 2011 ) .
k. saito and h. tasaki , arxiv:1105.2168 .
t. bodineau and b. derrida , phys .
* 92 * , 180601 ( 2004 ) .
t. bodineau and b. derrida , j. stat .
* 123 * , 277 ( 2006 ) .
t. bodineau and b. derrida , c. r. physique * 8 * , 540 ( 2007 ) .
b. derrida , j. stat .
( 2007 ) p07023 .
l. bertini , a. de sole , d. gabrielli , g. jona - lasinio , and c. landim , phys .
94 * , 030601 ( 2005 ) .
l. bertini , a. de sole , d. gabrielli , g. jona - lasinio , and c. landim , j. stat . phys . * 123 * , 237 ( 2006 ) . t. bodineau and b. derrida , phys .
e * 72 * , 066110 ( 2005 ) .
c. appert - rolland , b. derrida , v. lecomte , and f. van wijland , phys .
e * 78 * , 021122 ( 2008 ) .
a. imparato , v. lecomte , and f. van wijland , phys .
e * 80 * , 011131 ( 2009 ) .
p. reimann , c. van den broeck , h. linke , p. hnggi , j. m. rubi , and a. prez - madrid , phys .
lett . * 87 * , 010602 ( 2001 ) .
p. reimann , c. van den broeck , h. linke , p. hnggi , j. m. rubi , and a. prez - madrid , phys .
e * 65 * , 031104 ( 2002 ) .
v. blickle , t. speck , c. lutz , u. seifert , and c. bechinger , phys .
lett . * 98 * , 210601 ( 2007 ) .
j. r. gomez - solano , a. petrosyan , s. ciliberto , r. chetrite , and k. gawdzki , phys .
lett . * 103 * , 040601 ( 2009 ) .
s. toyabe , t. okamoto , t. watanabe - nakayama , h. taketani , s. kudo , and e. muneyuki , phys .
* 104 * , 198103 ( 2010 ) . | the cumulant generating function of time - averaged current is studied from an operational viewpoint .
specifically , for interacting brownian particles under non - equilibrium conditions , we show that the first derivative of the cumulant generating function is equal to the expectation value of the current in a modified system with an extra force added , where the modified system is characterized by a variational principle .
the formula reminds us of einstein s fluctuation theory in equilibrium statistical mechanics .
furthermore , since the formula leads to the fluctuation - dissipation relation when the linear response regime is focused on , it is regarded as an extension of the linear response theory to that valid beyond the linear response regime .
the formula is also related to previously known theories such as the donsker - varadhan theory , the additivity principle , and the least dissipation principle , but it is not derived from them .
examples of its application are presented for a driven brownian particle on a ring subject to a periodic potential . |
following a series of papers in 1978 @xcite , it has been widely accepted that drift waves in a plasma in a sheared magnetic field , in the absence of parallel current , temperature gradients , and magnetic curvature , are absolutely stable . in other words ,
the `` universal mode '' driven by the density gradient in slab geometry@xcite becomes absolutely stable if any magnetic shear is included .
this conclusion is important as a basic problem of plasma physics , and for understanding the physical mechanisms underlying instabilities in more complicated configurations in laboratory or space plasmas .
for example , the slab limit can provide insight into the edge of magnetically confined plasmas , in which the slab - like drive from radial equilibrium gradients ( @xmath8 for scale length @xmath9 ) exceeds the curvature drive ( @xmath10 for major radius @xmath11 ) , and in which the temperature gradient may tend to be relatively weaker than the density gradient since the former is more strongly equilibrated over ion orbits @xcite .
however , in contrast to the 1978 work ( and later studies @xcite ) , here we demonstrate that the universal mode is in fact unstable for a range of perpendicular wavenumbers @xmath3 satisfying @xmath12 , where @xmath2 is the ion gyroradius .
we reach a different conclusion to the 1978 work because these earlier calculations relied on an eigenmode equation derived under the assumption @xmath0 , where @xmath1 is the radial scale length of the fluctuating mode in untwisted coordinates .
in contrast , our gyrokinetic approach does not share this limitation .
in slab geometry with magnetic shear , the magnetic field in cartesian coordinates @xmath13 is @xmath14b$ ] for some constant @xmath15 and shear length @xmath6 .
a field - aligned coordinate system is introduced : @xmath16 , so @xmath17 and @xmath18 .
we consider electrostatic fluctuations satisfying the standard gyrokinetic orderings @xcite : the gyroradii are of the same formal order as the perpendicular wavelengths of the fluctuations , and these scales are one order smaller than the scale lengths of the equilibrium density and magnetic field .
we therefore may use the linear electrostatic gyrokinetic and quasineutrality equations derived in @xcite , dropping toroidal effects , temperature gradients , and collisions ( except where noted in figure [ fig : tnu ] . ) for perturbations varying in @xmath19 as @xmath20 and independent of @xmath21 , the gyrokinetic equation is @xmath22 and the quasineutrality condition is @xmath23 = 0 .
\label{eq : qn}\ ] ] here @xmath24 denotes species , @xmath25 is the electrostatic potential , @xmath26 is the nonadiabatic distribution function , @xmath27 , @xmath28 , and @xmath29 is the leading - order maxwellian . the scale length of the equilibrium electron density @xmath30 is @xmath31 , @xmath32 is the proton charge , @xmath33 is the species temperature , @xmath34 , @xmath35 , and @xmath36 , @xmath37 is a bessel function .
as @xmath38 was ordered @xmath39 , ( [ eq : gke])-([eq : qn ] ) are valid for both @xmath40 and @xmath41 ; the same is true of @xmath42 where @xmath43 is the electron gyroradius . for eigenmodes @xmath44 , a dispersion equation can be obtained by introducing @xmath45 satisfying @xmath46 with an analogous transform for @xmath47 .
we use @xmath48 as the transform variable because @xmath45 represents the radial mode in the untwisted coordinates . using ( [ eq : transform ] ) , @xmath49 in ( [ eq : gke ] ) becomes @xmath50 , so ( [ eq : gke ] ) may be solved for @xmath51 . using ( [ eq : qn ] ) and approximating @xmath52 , we obtain @xcite @xmath53\hat\phi(x ) \nonumber \\ & & + \left(\frac{t_e}{t_i}+\frac{\omega_*}{\omega}\right ) \left(\frac{k_y}{2\pi
l_s}\right)^2 \int dx ' \int dx '' \int dz \int dz ' \nonumber \\ & & \times \gamma_{0i}(z , z ' ) \hat\phi(x '' ) \exp\left(\frac{i k_y}{l_s}\left [ x'z - x'z'-xz+x''z'\right]\right ) \nonumber \\ & & \times \left|\frac{l_s}{k_y x'}\right|\frac{\omega}{v_i } z\left ( \left|\frac{l_s}{k_y x'}\right|\frac{\omega}{v_i } \right)\end{aligned}\ ] ] where @xmath54\right ) .
\nonumber\end{aligned}\ ] ] here , @xmath55 , @xmath56 , @xmath57 is a modified bessel function , @xmath58 is the plasma dispersion function , and @xmath59 is the drift frequency . the multiple fourier transforms in ( [ eq : integral ] ) arise because the @xmath58 function arguments are naturally written in terms of @xmath60 , whereas the bessel function arguments are naturally written in terms of @xmath61 .
notice the integral equation ( [ eq : integral ] ) from this gyrokinetic approach is manifestly different from the differential eigenmode equations in refs .
we solve ( [ eq : gke])-([eq : qn ] ) using two independent codes , gs2 @xcite and gene @xcite .
a demonstration of linear instability is given in figure [ fig : convergence ] for @xmath62 .
in this figure and hereafter , @xmath63 . due to the literature asserting stability , we have taken great care to verify that this instability is robust and not a numerical artifact . to this end
, we demonstrate in figure [ fig : convergence ] precise agreement between the two codes , which use different velocity - space coordinates and different numerical algorithms .
the figure shows calculations from gs2 run as initial - value simulations , and from gene run as a direct eigenmode solver using the slepc library @xcite . for each code , results are plotted for many different resolutions to demonstrate convergence .
( for gs2 , results are overlaid for 8 sets of numerical parameters : a base resolution , @xmath64 number of parallel grid points , @xmath65 parallel box size , @xmath64 number of energy grid points , @xmath64 maximum energy , @xmath64 number of pitch angle grid points , @xmath66 timestep , and @xmath64 maximum time . for gene ,
results are again overlaid for 8 sets of numerical parameters : a base resolution , @xmath64 number of parallel grid points , @xmath64 parallel box size , @xmath64 number of @xmath67 grid points , @xmath64 maximum perpendicular energy , @xmath64 number of @xmath68 grid points , @xmath64 maximum parallel energy , and @xmath64 numerical hyperviscosity . ) to further minimize the possibility of numerical artifacts , for this figure a reduced mass ratio @xmath69 was employed . as the differences between the 16 series displayed in each plot are nearly invisible , the instability is clearly physical and not numerical . using calculations from two independent codes
gs2 run as an initial - value simulation and gene run as a direct eigenvalue solver with results agreeing to high precision . for each code , results from many different numerical resolutions are overlaid in the figure .
a reduced mass ratio @xmath69 is used here to further minimize the possibility of numerical artifacts .
[ fig : convergence],width=336 ]
figure [ fig : landscape ] displays the dependence of the frequency and growth rate on perpendicular wavenumber and magnetic shear , this time using the true deuterium - electron mass ratio .
as one expects from the zero - shear universal instability , the phase velocity direction is @xmath70 ( appearing in the figures as @xmath71 due to the codes sign convention ) .
the frequency satisfies @xmath72 and @xmath73 . the growth rate is smaller in magnitude than the real frequency .
when @xmath74 , instability is found for @xmath75 , extending beyond @xmath76 for the highest values of @xmath77 .
stability for @xmath78 can be understood from the fact that @xmath79 in ( [ eq : gke])-([eq : qn ] ) becomes small , as the gyro - motion averages out very small - scale fluctuations .
stability is the opposite limit of small @xmath3 is expected since refs .
@xcite are valid there .
a region of stability also exists around @xmath80 when a realistic mass ratio is used .
no such notch in the growth rate is found when the shear is exactly zero . this abrupt change in eigenvalue when a small change is made to the physical system
in this case the addition of small magnetic shear for @xmath80 is a phenomenon seen in many systems with non - orthogonal eigenmodes @xcite such as this one . for such `` non - normal '' systems ,
the behavior over a finite time is typically a much less sensitive function of parameters than the eigenvalues , which represent behavior for @xmath81 when all points along the field line have had time to communicate .
indeed , in initial - value computations we find no notch near @xmath80 in the growth for short times @xmath82 , before ions have have traveled far enough in @xmath61 to `` realize '' there is shear ; these results will be presented in a separate publication .
instability is found for at least some values of @xmath3 whenever @xmath83 exceeds a critical value of @xmath84 .
note that in a tokamak , @xmath85 can exceed this threshold value . here
@xmath86 is the safety factor , @xmath11 is the major radius , @xmath87 , and @xmath88 is the minor radius .
the general trend in figure [ fig : landscape ] of decreasing growth rate with increasing shear can be understood from the observation that in the zero - shear limit , instability is found when @xmath89 is below a @xmath90-dependent threshold .
when sufficient shear is included , all of these unstable parallel wavelengths are sufficiently long for there to be a significant increase in @xmath91 , causing @xmath92 to be averaged out by gyroaveraging as discussed above .
for some @xmath38 such as @xmath93 , the growth rate does not increase monotonically with @xmath77 .
this effect is another example of eigenvalues being sensitive and perhaps misleading , typical in non - normal systems @xcite such as this one , for we find the short - time amplification to increase with @xmath77 monotonically . .
calculations are performed with the gs2 code .
[ fig : landscape],width=336 ] a typical unstable eigenmode is shown in figure [ fig : eigenmodes ] .
this example is obtained using gs2 at @xmath94 and @xmath95 , near the maximum growth rate in figure [ fig : landscape ] . in the field - aligned parallel coordinate @xmath61
, the mode has a parallel extent @xmath96 .
fourier transforming , the extent of the radial eigenmode is @xmath97 .
most of the mode structure occurs where the ion @xmath58 function has an argument of order 1 , @xmath98 , so the electron @xmath58 function has small argument .
the eigenmode amplitude is small in the inner electron region where @xmath99 .
the electrons are highly nonadiabatic : noting the ratio of electron nonadiabatic / adiabatic terms in ( [ eq : integral ] ) is @xmath100 , then the nonadiabatic electron term exceeds the adiabatic term wherever @xmath101 , everywhere the eigenfunction is significant . , @xmath102 , and @xmath103 ( near the maximum growth rate in figure [ fig : landscape ] ) .
in ( b ) , noting @xmath104 , vertical lines indicate the values of @xmath21 at which the arguments of the electron and ion @xmath58 functions in ( [ eq : integral ] ) have unit magnitude . [
fig : eigenmodes],width=336 ] dependence of the instability on temperature gradients and collisionality is shown in figure [ fig : tnu ] , for the case @xmath105 and @xmath102 .
collisions ( using the operator in @xcite ) reduce the growth rate , particularly at the highest @xmath3 values .
this behavior is expected because ( i ) in the free energy moment of ( [ eq : gke ] ) ( eq ( 222 ) of @xcite ) collisions provide an energy sink ; ( ii ) the gyrokinetic collision operator @xcite contains terms @xmath106 ; and ( iii ) the modes at highest @xmath3 have fine velocity - space structure ( as high resolution in energy and pitch - angle is required for numerical convergence ) which collisions destroy . for sufficient collisionality , @xmath107 or more ,
the mode is stabilized at all @xmath3 ( for this @xmath77 ) . and
@xmath102 , showing trends with @xmath108 and collisionality .
[ fig : tnu],width=336 ]
let us now examine in greater detail why we find instability whereas previous authors found stability . first , consider ref .
@xcite , which gave an analytical proof of stability in the limit @xmath109 .
we find no contradiction using gyrokinetic simulations , obtaining instability only when @xmath110 is not small .
( recall @xmath63 for all figures here . ) next , consider refs . @xcite .
these authors considered a differential eigenvalue equation , not equivalent to ( [ eq : integral ] ) , derived assuming @xmath111 ( see e.g. @xcite . )
no such assumption was made in deriving ( [ eq : integral ] ) , which is therefore more general .
considering the typical eigenmode in figure [ fig : eigenmodes].b , there is structure on scales smaller than @xmath2 , violating ( [ eq : differentiallimit ] ) .
therefore the approach used in refs .
@xcite is inapplicable for the modes that are unstable .
@xcite claims to give a proof of stability which does not rely on ( [ eq : differentiallimit ] ) .
however , this reference uses an integral eigenmode equation which differs from ( [ eq : integral ] ) , so it is not surprising that we reach different conclusions .
the eigenmode equation in @xcite is effectively a wkb approximation , the zero - shear dispersion relation but with @xmath112 replaced by @xmath113 .
this wkb approximation is not justified for modes such as the one in figure [ fig : eigenmodes ] , in which the wavelengths in @xmath21 are comparable to or much longer than the scale of variation in the ion and electron @xmath58 functions .
the problems with the integral equation in @xcite are discussed in the introduction of @xcite and references therein .
ref @xcite appears to give a proof of stability using the full gyrokinetic integral eigenmode equation ( [ eq : integral ] ) .
although we retain @xmath114 for all numerical results shown here while @xmath114 is set to 1 in ref .
@xcite , this difference is unimportant : we find little change in the instability for @xmath115 if @xmath114 is set to 1 in gs2 .
the proof in @xcite is accomplished by multiplying the electron nonadiabatic term in ( [ eq : integral ] ) ( @xmath116 ) by a parameter @xmath117 .
the authors first prove stability in the adiabatic electron limit @xmath118 .
we do not disagree , as we find no instability numerically with adiabatic electrons .
however we do disagree with the final step of the proof , where it is argued that solutions for @xmath119 can not cross the real axis as @xmath117 is increased from 0 to 1 to recover ( [ eq : integral ] ) .
including @xmath117 in gs2 , we find that modes do continuously transform from damped to unstable as @xmath117 is increased from 0 to 1 . for the case @xmath105 and @xmath120 shown in figure [ fig : eigenmodes ] , marginal stability occurs when @xmath121 .
therefore , the problematic step in @xcite is item ( iii ) on p750 . indeed , when @xmath122 , a nonadiabatic electron term must be included in ( 38)-(55 ) , and this term can have opposite sign to the other terms in ( 55 ) , so ( 56 ) no longer follows .
in summary , we find the plasma slab with weak or moderate magnetic shear ( @xmath123 ) is generally unstable at low collisionality , even when temperature gradients are weak . the density - gradient - driven electron drift wave , known to be absolutely unstable in the absence of magnetic shear @xcite , is not stabilized at all wavelengths when small magnetic shear is introduced , counter to previous findings .
this instability is seen robustly in multiple gyrokinetic codes , and occurs for @xmath124 .
previous work that apparently showed stability of the universal mode @xcite assumed either ( [ eq : differentiallimit ] ) or @xmath109 , whereas we make neither assumption . ref . @xcite employed a less accurate dispersion relation , and @xcite appears incorrect for nonadiabatic electrons .
though it has sometimes been assumed that an electrostatic instability seen in a gyrokinetic simulation propagating in the electron diamagnetic direction must be a trapped electron mode or electron temperature gradient mode , our results indicate neither trapped particles nor temperature gradients are necessary for instability in this phase - velocity direction .
as the universal mode can indeed be absolutely unstable in the presence of magnetic shear , it should be considered alongside temperature - gradient - driven modes , trapped particle instabilities , ballooning modes , and tearing modes as one of the fundamental plasma microinstabilities .
this material is based upon work supported by the u.s .
department of energy , office of science , office of fusion energy science , under award numbers defg0293er54197 and defc0208er54964 .
computations were performed on the edison system at the national energy research scientific computing center , a doe office of science user facility supported by the office of science of the u.s .
department of energy under contract no .
de - ac02 - 05ch11231 .
we wish to thank tobias grler for providing the gene code .
we also acknowledge conversations about this work with peter catto , greg hammett , adil hassam , edmund highcock , wrick sengupta , jason tenbarge , and george wilkie .
20ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12
& 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | we demonstrate that the universal mode driven by the density gradient in a plasma slab can be absolutely unstable even in the presence of reasonable magnetic shear .
previous studies from the 1970s that reached the opposite conclusion used an eigenmode equation limited to @xmath0 , where @xmath1 is the scale length of the mode in the radial direction , and @xmath2 is the ion larmor radius . here we instead use a gyrokinetic approach which does not have this same limitation .
instability is found for perpendicular wavenumbers @xmath3 in the range @xmath4 , and for sufficiently weak magnetic shear : @xmath5 , where @xmath6 and @xmath7 are the scale lengths of magnetic shear and density .
thus , the plasma drift wave in a sheared magnetic field may be unstable even with no temperature gradients , no trapped particles , and no magnetic curvature . |
anomalous x - ray pulsars ( axps ) and soft - gamma repeaters are neutron stars many of whose attributes at x - ray , gamma - ray , and infrared wavelengths ( see * ? ? ? * for a review ) are best understood in the context of the magnetar model @xcite , according to which their high - energy emission results from the rearrangement and decay of ultra - strong magnetic fields .
much remains to be learned about magnetars , of which only a dozen are known . is an axp with spin period @xmath3s , unusual in being transient .
identified in early 2003 when its x - ray luminosity increased 100-fold @xcite , by 2007 it has returned to the quiescent state it maintained for at least 24 years @xcite .
uniquely for a magnetar it emits radio waves @xcite , that turned on by early 2004 . unlike in ordinary rotation - powered pulsars , the radio pulses have a flat spectrum and vary in luminosity and shape on daily timescales @xcite . radio emission from links magnetars and ordinary pulsars , and provides a new window for learning about the physical characteristics of a magnetar .
for instance , while in principle radio emission could be generated in the corona from closed or open magnetic field lines , the large pulse profile and flux density changes observed on short timescales @xcite appear to point to the latter ( cf . * ? ? ?
* ) . here
we report on observations of the polarized emission from in an attempt to shed some light on the geometry of the radio - emitting regions of this magnetar .
we have observed with the parkes 64-m telescope in new south wales , australia , in full - stokes polarimetry mode for a total of 20hr on - source between 2006 april and november .
table [ tab : obs ] summarizes the relevant observations .
llclr 53852/0427 & 1.369 & 0.9 & wbc & 650 + 53862/0507 & 1.369 & 0.2 & wbc & 350 + 53879/0524 & 1.369 & 0.2 & wbc & 320 + 53913/0627 & 1.369 & 2.0 & dfb & 1500 + 53986/0908 & 1.369 & 1.2 & dfb & 70 + 53989/0911 & 8.356 & 5.0 & wbc & 45 + 53993/0915 & 3.222 & 0.3 & dfb & 20 + 54002/0924 & 1.369 & 3.2 & dfb & 50 + 54021/1013 & 1.369 & 0.4 & dfb & 20 + 54021/1013 & 3.222 & 1.8 & dfb & 10 + 54022/1014 & 3.222 & 3.3 & dfb & 20 + 54060/1121 & 1.369 & 1.6 & dfb & 20 + we have used the three available receiver / feed combinations that have well - characterized polarization properties : h - oh ( 1.4ghz ) , 10/50 cm ( 3.2ghz ) , and mars ( 8.4ghz ) . due to common availability , at 1.4ghz we have also used the central beam of the multibeam receiver , although its polarimetric characteristics are less ideal compared to those of the h - oh receiver @xcite . the h - oh and 10/50 cm systems have orthogonal linear feeds , while the mars package receives dual circular polarizations . in all cases
a pulsed calibrating signal can be injected at an angle of 45 to the feed probes . to record data we used either the digital filterbank ( dfb ) or the wide - band correlator ( wbc ) .
the bandwidth , frequency- and time - resolution varied depending on receiver and spectrometer , but typical values were , respectively , 256mhz , 128 channels , and 2048 bins across the pulse profile ( see table [ tab : obs ] for details ) .
an integer number of pulse periods were folded and recorded to disk in psrfits format for off - line analysis . because the dump time of the spectrometers is @xmath4s , a minimum of two pulse periods were folded in each subscan , with @xmath5 more common .
typically , scans lasting up to @xmath6hr were interspersed with @xmath7min observations of the pulsed calibrator in order to determine the relative gain and phase between the two feed probes . for our purposes ,
the main difference between the 3-level sampling / correlation wbc and the 8-bit precision dfb was the latter s much greater sensitivity to radio frequency interference , which we excised in the frequency- and time - domain during analysis .
we used existing observations of the flux calibrator hydra a , whose flux density is 43.1 , 20.3 and 8.4jy at 1.4 , 3.2 and 8.4ghz respectively , to determine the system equivalent flux density for the receivers and to flux - calibrate the pulse profiles .
all data were analyzed with the psrchive software package @xcite .
as part of the analysis we corrected the stokes parameters ( @xmath8 , @xmath9 , total intensity @xmath10 , and circular polarization @xmath11 ) for the position of the feed probes relative to the telescope meridian and for the parallactic angle of the observation .
we also observed strong pulsars with known polarization characteristics ( such as the vela pulsar ) to provide a check on our polarimetric calibration .
analysis of these pulsars yielded linear polarization @xmath12 , position angle of linear polarization @xmath13 , and @xmath11 matching those in the literature ( e.g. , * ? ? ? * ; * ? ? ? * ) .
to complete polarization calibration for we had to compute the amount of faraday rotation suffered by the radiation in its passage through the galactic magnetic field .
we determined the rotation measure by measuring pa as a function of frequency within the 256-mhz band at 1.4ghz when the pulsar was strong .
the resulting value , @xmath14radm@xmath1 , did not vary within the quoted uncertainty either as a function of pulse phase or time .
the rm was then used to correct the measured pas and frequency - integrated @xmath15 at all frequencies to their values at infinite frequency so that a comparison could be made between frequencies ( e.g. , * ? ? ?
the pas and @xmath15 shown in figure [ fig : pol ] therefore represent those emitted at the pulsar .
we also display in the figure the stokes @xmath10 and @xmath11 .
together with the integrated column density of free electrons to , @xmath16@xmath17pc @xcite , the rm can be used to determine the average magnetic field strength parallel to the line of sight weighted by electron density , @xmath18 g .
this fairly small value appears reasonable given the location of the pulsar , @xmath19 and @xmath20kpc , for which the large - scale galactic field is mostly in the perpendicular direction , and with at least one reversal along the line of sight ( e.g. , * ? ? ?
in spite of the variability of the profiles shown in figure [ fig : pol ] there are three striking and constant aspects to the polarization profiles .
first , the fractional linear polarization is extremely high , close to 100% , and remains high at all frequencies measured here .
secondly , there is a shallow increase in the pa as a function of rotational phase , which remains essentially unchanged regardless of time or frequency of the observations .
the rate of change is reasonably constant over the `` main '' pulse profile components and is around @xmath21deg@xmath22 . finally , there is little or no circular polarization ( @xmath23 ) in the integrated profiles at any frequency ( fig .
[ fig : pol ] ) , or in individual pulses at 1.4ghz except for occasional levels up to @xmath24 of total intensity in the `` precursor '' pulse components ( cf . fig .
[ fig : sp ] ) .
the emission from changed in character in late 2006 july @xcite . while daily variations continue unabated , generally the pulse profiles are broader ( compare figs . [
fig : pol ] [ a ] and [ b ] ) and the fluxes are lower ( the peak flux densities listed in table [ tab : obs ] attest to this ) .
in contrast , the general polarization characteristics do not seem to vary .
this suggests that the gross observed changes in profile morphology are not due to detectable changes in the underlying magnetic field geometry of the emission regions . with the variability of the integrated profiles as a caveat
, we nevertheless attempt to compare the profiles at 1.4 , 3.2 and 8.4ghz . in order to isolate long - term variations ,
we consider for this purpose only data taken in a one - week period in 2006 september .
the double peaked profile gets narrower as the frequency increases and the ratio of the leading to trailing component becomes larger ( fig .
[ fig : pa ] ) .
also , the slow pa sweep ( and absolute value of the pa ) is identical at all frequencies , as expected in the `` rotating vector model '' of @xcite .
this is consistent with the radius - to - frequency mapping paradigm in which lower frequencies are emitted farther from the star than higher frequencies ( e.g. , * ? ? ?
* ) . without detailed geometrical information ,
however , it is difficult to quantify this effect ( for a brief discussion of radius - to - frequency mapping concerning , see * ? ? ?
* ; * ? ? ?
* ) . in the very early days of pulsar astronomy
, it was realized that the observed pa swing could be used to derive the geometry of the star under the assumption that the pa was related to the projection of the dipolar field lines on the plane of the sky @xcite .
unfortunately , it is difficult to determine the geometry in the majority of pulsars mainly because of the small longitude range over which they emit ( e.g. , * ? ? ?
this is true of also .
in addition , it is a priori unclear whether a dipolar field structure holds true in this pulsar , although we proceed on the assumption that it might and see where that leads us . in our post-2006 july data , neither @xmath25 ( the angle between the magnetic and rotation axes ) nor @xmath26 ( the angle of closest approach of the line of sight to the magnetic axis ) can be constrained . the earlier data , with the appearance of pulse components far from the main component ( fig . [ fig : sp ] ) , are more promising in this regard . here , however , the main uncertainty is whether there is 90 of pa rotation between the widely spaced components .
formal fits to the data both with and without an extra 90 of phase are reasonable ( see fig . [
fig : pafit ] ) . in the former case ,
the fits yield values of @xmath25 near 70 and high values of @xmath26 near 2025 . without the added orthogonal jump ,
the fits yield @xmath27 and @xmath28 implying that the magnetic and rotation axes are almost aligned .
the polarization characteristics of are very similar to those seen in young pulsar profiles @xcite .
they too are highly linearly polarized , often have double profiles , and show a slow swing of pa across a wide profile .
@xcite showed that a single cone of emission originating from relatively high in the pulsar magnetosphere could explain the observed characteristics of young pulsars and it is tempting to make the same case here .
however , there is a significant difference in the polar cap radius , @xmath29 ( and light cylinder radius @xmath30 ) , between a young pulsar with a period of 0.1s and with its 5.5s period .
this makes it difficult to see how such a wide ( @xmath31 ) observed pulse profile can be produced unless ( a ) the emission height is very large or ( b ) the magnetic and rotation axes are almost aligned .
we will discuss these two possibilities in turn . in the first case , knowledge of the geometry and the observed pulse width
can be used to compute an emission height . for values of @xmath25 near 70 and @xmath32 , one can use equation ( 2 ) of @xcite to derive the cone opening angle @xmath33 . in turn
the emission height can be computed as @xmath34 , or @xmath35 km .
this is about 10% of the light cylinder radius similar to the value in other young pulsars @xcite .
if this scenario were typical of magnetars in general then the beaming fraction would be high and most bright radio active magnetars would likely be detectable in pulsations . in the second case , for small values of @xmath25 ,
the line of sight could remain wholly within the emission beam leading to the observed wide profile . in this case , if @xmath25 were to vary slightly with time ( for reasons unknown ) , there could be a large effect on both the observed beam shape and torque , both of which have been observed to vary significantly @xcite .
perhaps the emission in this particular magnetar could be in part a direct function of the quasi - alignment between the rotation axis and magnetic axis , or perhaps the alignment might occur as a natural process in magnetars . in either case ,
the small polar cap size would make the beaming fraction of such magnetars rather low .
the main difficulty with this interpretation is that the radio and x - ray beams appear to be nearly aligned @xcite and the observed modulation of thermal x - rays is very large ( @xmath36 ; * ? ? ?
* ) , which would be hard to obtain from a nearly aligned rotator . in summary , the polarized emission from shares many characteristics of those in young pulsars generally .
the emission is highly linearly polarized with little evolution with frequency , the pulse profile is wide and double , and there is only a shallow swing of pa through the main pulse .
this leads to the possibility that the `` standard '' pulsar ideas of emission along open magnetic field lines also hold here . in this case , either the magnetic and rotation axes are almost aligned , or the emission originates high above the surface of the star , which is our preferred interpretation .
obvious remaining differences between and other pulsars are its pulse profile variability ( which does not appear to be accompanied by corresponding gross changes in the magnetic field geometry ) , fluctuating flux density , and flat spectrum .
we thank john sarkissian for help with observations , and aidan hotan and aris karastergiou for discussions .
the parkes observatory is part of the australia telescope , which is funded by the commonwealth of australia for operation as a national facility managed by csiro .
fc acknowledges the nsf for support through grant ast-05 - 07376 .
representative polarimetric profiles of over time and frequency .
the _ red _ and _ blue _ lines represent the amount of linear and circular polarization , respectively , and the _ black _ traces give total intensity .
the position angles ( pas ) are displayed for bins where the linear polarization signal - to - noise ratio @xmath37 and have been corrected for rotation measure . in order to highlight the emission region ,
we show 90 of pulse phase and a pa range of 3090 .
each data set was folded using a different ephemeris , and no attempt was made to align the panels precisely .
panel ( a ) shows a 5min integration with 2048-bin resolution when the pulsar was strong ( emission on this day was also detected at other phases ; see fig .
[ fig : sp ] ) .
panel ( b ) displays a 19min integration when the pulsar had weakened and its profile had broadened considerably .
panel ( c ) shows a 2.7hr integration , and ( d ) a 3.9hr integration .
panels ( b)(d ) are displayed with 512 bins across the full pulse period . ]
selected polarimetric profiles of at 1.4ghz on mjd 53913 ( see also fig . [
fig : pol ] [ a ] for data collected 6hr earlier ) .
each panel contains 16.6s of data ( three pulses ) folded modulo the pulsar period .
the top two panels are contiguous in time , with the one at bottom recorded 6min later . in order to highlight the emission region , we show in each panel 150 of pulse phase . traces and symbols are as in figure [ fig : pol ] . ] position angles ( _ top _ ) and total intensity profiles ( _ bottom _ ) for at three frequencies ( _ blue _ corresponds to 1.4ghz data , _ green _ to 3.2ghz , and _ red _ to 8.4ghz ) .
all data were obtained within a one - week period , and 125 of phase are displayed with 512 bins across the full pulse period .
the 1.4ghz and 8.4ghz data were presented in a somewhat different form in figure [ fig : pol ] , while those at 3.2ghz ( 195 pulses , or 18min ) are from mjd 53993 . starting with the previous best determination of rm ( @xmath38radm@xmath1 ) , we aligned the profiles by adjusting both phase and pa in all three frequencies , obtaining a more precise value @xmath39radm@xmath1 .
longitude 0 on the plot is arbitrary . ]
_ bottom panel _ : polarimetric profile of at 1.4ghz based on 17min of data from mjd 53913 .
the full 360 of pulse phase are shown .
the _ black _ trace shows the total intensity emission , while _ red _ displays linear polarization and _ green _ represents the level of circular polarization . _ top panel _ : position angle data ( _ black _ points ) and the best - fit rotating vector model curves for the almost aligned ( _ blue _ line ) and almost orthogonal ( _ green _ line ) cases discussed in the text . in the latter instance the emission from the main pulse component and the precursor component must arise from different orthogonal modes , thus necessitating the 90 phase jumps in the fitted curve . ] | we have used the parkes radio telescope to study the polarized emission from the anomalous x - ray pulsar at frequencies of 1.4 , 3.2 , and 8.4ghz .
we find that the pulsed emission is nearly 100% linearly polarized .
the position angle of linear polarization varies gently across the observed pulse profiles , varying little with observing frequency or time , even as the pulse profiles have changed dramatically over a period of 7 months . in the context of the standard pulsar `` rotating vector model , ''
there are two possible interpretations of the observed position angle swing coupled with the wide profile . in the first ,
the magnetic and rotation axes are substantially misaligned and the emission originates high in the magnetosphere , as seen for other young radio pulsars , and the beaming fraction is large . in the second interpretation , the magnetic and rotation axes are nearly aligned and the line of sight remains in the emission zone over almost the entire pulse phase .
we deprecate this possibility because of the observed large modulation of thermal x - ray flux . we have also measured the faraday rotation caused by the galactic magnetic field , @xmath0radm@xmath1 , implying an average magnetic field component along the line of sight of @xmath2 g . |
decays of @xmath1 mesons into charmonium are the subject of much theoretical and experimental study .
for example , the cp asymmetry of the decay @xmath2 provides a very high precision determination of the unitarity angle @xmath3 .
this is a clean determination because in the standard model ( sm ) of electroweak interactions the cp asymmetry does not depend on the decay rate .
however , it is an interesting challenge in theoretical physics to determine from first principles the decay rate in the sm .
doubtless this would shed light into the underlying qcd dynamics .
we propose to construct an effective field theory that captures the essential dynamics of decays of a @xmath1 meson into charmonium plus light mesons .
the theory we propose is a systematic expansion in small parameters and therefore the resulting approximations can be systematically improved .
the building blocks of the effective theory are largely known : heavy quark effective theory ( hqet ) to describe the @xmath4 quark in the decaying meson @xcite , soft - collinear effective theory ( scet ) to account for the energetic @xmath5 quark in the final state @xcite , and non - relativistic qcd ( nrqcd ) to describe the quark pair in the charmonium final state @xcite-@xcite .
the main adaptation of these theories to our case involves re - formulating nrqcd covariantly in an arbitrary frame ( normally , nrqcd is formulated in the rest - frame of the charmonium state ) . the resulting effective theory , which we call covariant nrqcd ( cnrqcd ) accounts for the charm and anti - charm quark fields through four component spinors , to preserve covariance .
we aim to show that the amplitudes for @xmath1 decay to charmonium factorize .
what is meant by this is that the amplitude for @xmath6 , where @xmath7 is a charmonium state and @xmath8 the light hadron , is the product of the amplitudes for @xmath9 and @xmath10 , where @xmath11 are quark currents .
that is , the @xmath12 is given by the form factor of the current @xmath13 between @xmath1 and @xmath8 states times the decay constant of the @xmath7 state .
there is a simple physical picture which suggests factorization .
violations to factorization arise only if gluons are exchanged between the @xmath7 state and the @xmath1 or @xmath8 states .
the effect of very energetic ( hard ) gluons affects the process at very short distances only , but will not affect the dynamical picture , so their effect can be absorbed into an overall constant coefficient .
moreover , this constant is calculable since hard gluons are perturbative .
the effect of low energy gluons , on the other hand , is non - perturbative .
however , the @xmath7 is a very compact bound state . in a multi - pole expansion
, a long wavelength gluon interacts with the @xmath7 color - charge distribution through it s color - dipole moment since the state is itself color neutral .
but the dipole is very small because the @xmath7 is small . in the theoretical limit of very heavy charm
, this coupling to the dipole vanishes .
it should be noted that the weak interaction can also produce a charm - anti - charm pair with non - zero total color charge , in the octet configuration . the physical argument above indicates that in this configuration the quark pair does interact with soft gluons even at leading order .
and while a color octet can not form a ( color neutral ) charmonium state , the emission of a soft gluon can in principle turn the pair into a color neutral state .
it would then seem that this electroweak contribution to the amplitude for @xmath7 production in @xmath1 decays does not factorize .
however , a long wavelength gluon interacting with a color octet leave the state in this octet configuration , and hence in a state which can not produce a physical charmonium state . as soon as the gluon wavelength is short enough to discern the quark - anti - quark nature of the octet
, a transition to the color singlet state can occur .
but this is suppressed by the perturbative coupling constant of short wavelength gluons , so factorization holds at leading order and is expected to break down at the first order in the small expansion parameters .
while this physical argument is not a proof , it does capture the basic ingredients needed to construct a more formal argument . as we will see
, it will be important to separate the gluon degrees of freedom according to the wavelengths and frequencies with which different particles interact .
this is accomplished by using the effective theory approach , combining hqet , scet and cnrqcd .
degrees of freedom that are left out in this classification produce what amount to short distance corrections , which can be absorbed into re - definitions of coupling coefficients . on the other hand , active degrees of freedom exhibit some simplifications , as for example spin symmetries , that are of paramount importance in the solution to the problem .
there is a quite extensive literature on the theory of @xmath1 decays into charmonium . the naive factorization hypothesis in exclusive @xmath14 , with @xmath7 and @xmath8 denoting charmonium and strange states , respectively , has been tested against data , using models of the @xmath1 to @xmath15 form factor ; see , e.g. , refs .
several theoretical approaches , distinct from ours , to factorization in qcd have been applied to exclusive @xmath1 decays to charmonium .
these include `` perturbative qcd '' @xcite ( pqcd ) and `` qcd factorization '' @xcite . however , these approaches are not without trouble . ref .
@xcite shows that in qcd factorization infrared divergences arising from nonfactorizable vertex corrections invalidate factorization . and
the pqcd approach does not find factorization but rather computes the non - factorizable part .
one may also attempt to formulate factorization for inclusive charmonium production in @xmath1 decays , as first proposed by bjorken @xcite ( for semi - inclusive @xmath16 see @xcite ) .
this program was formulated more precisely in what is now also commonly known as nrqcd @xcite .
this formulation , however , does not address factorization of amplitudes but rather of decay rates .
moreover , the program has run into difficulties as predictions of polarization of charmonium @xcite have missed the experimental mark @xcite .
namely , the mechanism which resolves the factor of 30 discrepancy between theoretical predictions and experimental measurements of prompt @xmath17 production at the tevatron involves gluon fragmentation to a sub - leading fock component in the @xmath17 wave - function and yields 100% transversely aligned @xmath17 s to lowest order , which is not observed . the effective field theory approach presented in this paper
may be used to study inclusive charmonium production in @xmath1 decays .
the paper is organized as follows . in section [ sec:2 ]
we discuss the kinematics of the decay @xmath18 and outline the effective theory used to derive the decay lagrangian . in section [ sec:3 ]
we list the degrees of freedom of the effective theory relevant in the leading logarithm approximation and give the feynman rules necessary to do one - loop calculations in the effective theory .
also there we discuss the covariant modification of nrqcd .
the effective theory electroweak lagrangian in the leading logarithm approximation and at the leading order in the effective theory power expansion is derived in section [ sec:4 ] . in section [ sec:5 ]
the simple formal proof of the factorization of the decay amplitude in the leading logarithm approximation is presented . in section [ sec:6 ]
we perform the tree - level matching between the effective theory and full theory amplitudes with the soft ( c)nrqcd gluons included and discuss the possibility of factorization beyond the leading logarithms . in appendix [ sec:11 ] the differential decay rate for the semi - inclusive decay @xmath0 is evaluated .
the initial values of the wilson coefficients in the next - to - leading logarithm approximation at the matching scale of the decay are given in appendix [ sec:12 ] .
at the quark level the decay @xmath6 can be described as the @xmath4-quark decaying due to a weak interaction @xmath19-boson exchange into a @xmath20-pair and the @xmath5-quark .
the spectator quark in the @xmath1-meson becomes the second quark in @xmath8 , see fig .
( [ fig : bdecay ] ) .
-quark , the heavy quark / antiquark @xmath20-pair , the light @xmath5-quark , and the spectator quark are shown.[fig : bdecay],height=113 ] the starting point of our analysis is the effective lagrangian @xmath21 of @xmath22 @xmath1-decays of electroweak interactions obtained by decoupling heavy degrees of freedom of the sm such as the @xmath19- , @xmath23-boson and the @xmath24-quark at the scale @xmath25 of the order of the @xmath19-boson mass . throughout
we will refer to @xmath21 as the full theory .
the relevant part which governs at the quark level the decay @xmath26 is given by @xmath27 it has been obtained at the first order in perturbation theory in the electroweak coupling @xmath28 ( fermi coupling ) and the ckm - matrix elements @xmath29 . the flavor changing @xmath22 operators @xmath30 are products of two quark currents with the spinor content @xmath31[\bar{c } \ldots c]$ ] which is the preferable choice when considering the formation of a @xmath20-bound state as a final state .
they read @xmath32[\bar{c } \gamma^\mu p_l c ] , & \op_8 & = [ \bar{s } \gamma_\mu p_l t^a b][\bar{c } \gamma^\mu p_l t^a c ] .
\end{aligned}\ ] ] these operators are related to the electroweak @xmath19-boson exchange operators with the `` natural '' spinor content @xmath33[\bar{c } \ldots b]$ ] by a fierz transformation . in the following
we will frequently refer to @xmath34 and @xmath35 as to the singlet and octet operators , respectively .
the wilson coefficients @xmath36 depend on the scale @xmath37 which is of the order of the @xmath4-quark mass and contain the resummed leading qcd logarithms of the form @xmath38 to all orders in @xmath39 .
currently they are known up to the nnlo in qcd @xcite in a different operator basis though .
the details of the transformation between two operator bases can be found in @xcite and the exact definition of the evanescent operators involved in this transformation is given in appendix [ sec:9 ] .
here we give only the numerical values of the coefficients [ eft:7 ] c_i^nlo ( ) = c_i^(0 ) + c_i^(1 ) ( ) .
we find for the basis ( [ singloct ] ) at the nlo the central values @xmath40 and @xmath41 with the input values @xcite : @xmath42 , @xmath43 , @xmath44 , @xmath45 , @xmath46 and @xmath47 . it should be noted firstly , that the octet wilson coefficient is one order larger than that one for the singlet and secondly , that the nlo correction to the singlet wilson coefficient is roughly @xmath48 since at lo : @xmath49 ( whereas @xmath50 ) .
furthermore , the scale uncertainties amount to intervals @xmath51 $ ] and @xmath52 $ ] when varying the matching scale of electroweak interactions @xmath53 \gev$ ] and the scale @xmath54 \gev$ ] .
this variation is mainly due to @xmath55 dependence which will be cancelled in physical observables once the matrix elements of the operators are calculated ( up to a higher order @xmath39 residual scale dependence ) . finally , at the scale @xmath37 the decay amplitude is proportional to the matrix element @xmath56 in the following we will set up the kinematics of the underlying partonic decay @xmath26 in the full theory which will enable the construction of an effective theory in order to further investigate the structure of this matrix element .
the wavelengths of the quarks participating in the decay are reasonably less than the wavelength associated with the confinement scale @xmath57 .
the @xmath4-quark and the @xmath20-pair are considered to be heavy since their masses are roughly one order of magnitude and @xmath58 times larger than the confinement scale , respectively .
the @xmath5-quark is very light but it is also very energetic in the restframe of the @xmath4-quark , so its wavelength is small .
the numerical values of the quark masses are about @xmath59 the masses of @xmath4 and @xmath60 quarks quoted are obtained from continuum determinations in the @xmath61 scheme @xcite .
the mass of the @xmath5-quark is not relevant for our calculation .
we will work in the @xmath4-quark restframe ( @xmath62 ) and decompose the momenta of the quarks accordingly .
the @xmath4-quark momentum can be decomposed into ( omitting lorentz indices ) @xmath63 with the scaling of the four velocitiy @xmath64 and the residual momentum @xmath65 .
we assume the @xmath20-pair to have a small relative velocity @xmath66 in their center of mass frame ( @xmath67 ) in order to form the @xmath20-bound state @xmath7 . in this kinematical configuration
they are non - relativistic with a momentum scaling @xmath68 .
this scaling persists also to an arbitrary boosted frame @xmath69 where the four veloctiy of the center of mass of the @xmath20-pair scales as @xmath70 , the perpendicular components as @xmath71 and the residual momenta @xmath72 . throughout the symbol @xmath73 will denote the relative velocity of the @xmath20-pair in their @xmath67 whereas four velocities are denoted by @xmath74 with @xmath75 and @xmath76 .
the relative velocity @xmath73 of the quarks of the @xmath20-pair can be found from the self - consistency condition @xcite : [ scale ] v=_s(m_c v),m_c = m_c(m_c)^4/_0 , where the first equation follows from the virial theorem applied to the ground state of the hydrogen - like atom : @xmath77 with @xmath78 is the reduced mass .
the second of eqs .
( [ scale ] ) is the lo expression for the running quark mass in qcd . for 5 flavors @xmath79 and @xmath80 gev .
numerical solution of eqs .
( [ scale ] ) gives : @xmath81 , @xmath82 gev , so that @xmath83 gev and @xmath84 gev .
the relative velocity @xmath73 is not small although @xmath85 gev is close to the charmonium mass and @xmath86 is comparable to @xmath87 .
since the effective theory is an expansion in powers of @xmath73 around @xmath88 , the process @xmath26 at the leading order in @xmath73-expansion is essentially a two - body decay and we can estimate the energy / momentum of the outgoing @xmath5-quark to be @xmath89 in the @xmath62 .
then the energetic @xmath5-quark momentum can be decomposed as follows @xmath90 here the two vectors @xmath91 and @xmath92 with @xmath93 being a unit vector ( the components are given in the @xmath62 ) are light - like vectors , i.e. @xmath94 and @xmath95 . introducing @xmath96 and
@xmath97 the momentum @xmath98 scales as @xmath99 with @xmath100 when @xmath101 is chosen to point approximately in the same direction as @xmath93 .
so , the wavelengths of all quarks involved are reasonably large compared to @xmath87 except for the spectator quark . before the decay the spectator quark belongs to the @xmath1-meson and after the decay it must become the constituent quark of @xmath8 . temporarily neglecting the spectator quark
we will develop an effective theory for the matrix element ( [ eft:1 ] ) in the limit where @xmath102 , @xmath103 , and @xmath104 .
we keep the ratio @xmath105 finite , otherwise the above limits are taken independently .
the latter means that to develop the effective theory for the decay we can simply combine the leading orders of hqet ( see @xcite , @xcite , and @xcite ) , nrqcd @xcite , and scet @xcite .
we want to describe the decay in the @xmath62 , so we have to develop a covariant formulation of nrqcd .
in this section we introduce the quark and gluon degrees of freedom of the effective theory and discuss a covariant form of the nrqcd .
the feynman rules necessary for the one - loop calculation in the effective theory are shown in fig .
[ frules ] .
we do not discuss hqet and scet sectors of the effective theory and refer the reader to the extensive literature .
a ) propagator of collinear quark ; @xmath106 is the lightlike vector of the @xmath5-quark .
b ) interaction vertex of the ultrasoft gluon and collinear quark .
c ) interaction vertex of the collinear gluon and collinear quark .
d ) propagator of the heavy @xmath4-quark .
e ) interaction vertex of the ultrasoft gluon and the heavy @xmath4-quark .
f ) propagator of @xmath60-quark and/or @xmath107-antiquark in covariant nrqcd ; @xmath108 is the 4-velocity of com of @xmath20 pair .
the second expression is the propagator of the charge conjugated field @xmath109 .
g ) & h ) interaction vertex of ultrasoft and/or potential gluon and @xmath60-quark in covariant nrqcd .
negative sign corresponds to the @xmath107-quark .
i ) propagator of the potential gluon in feynman gauge .
momentum @xmath110 is transverse with respect to the 4-velocity of the @xmath20-pair , @xmath111 .
j ) & k ) propagators of the ultrasoft and collinear gluons in feynman gauge.,title="fig:",height=56 ] a ) propagator of collinear quark ; @xmath106 is the lightlike vector of the @xmath5-quark .
b ) interaction vertex of the ultrasoft gluon and collinear quark .
c ) interaction vertex of the collinear gluon and collinear quark .
d ) propagator of the heavy @xmath4-quark .
e ) interaction vertex of the ultrasoft gluon and the heavy @xmath4-quark .
f ) propagator of @xmath60-quark and/or @xmath107-antiquark in covariant nrqcd ; @xmath108 is the 4-velocity of com of @xmath20 pair .
the second expression is the propagator of the charge conjugated field @xmath109 .
g ) & h ) interaction vertex of ultrasoft and/or potential gluon and @xmath60-quark in covariant nrqcd .
negative sign corresponds to the @xmath107-quark .
i ) propagator of the potential gluon in feynman gauge .
momentum @xmath110 is transverse with respect to the 4-velocity of the @xmath20-pair , @xmath111 .
j ) & k ) propagators of the ultrasoft and collinear gluons in feynman gauge.,title="fig:",height=75 ] a ) propagator of collinear quark ; @xmath106 is the lightlike vector of the @xmath5-quark .
b ) interaction vertex of the ultrasoft gluon and collinear quark .
c ) interaction vertex of the collinear gluon and collinear quark .
d ) propagator of the heavy @xmath4-quark .
e ) interaction vertex of the ultrasoft gluon and the heavy @xmath4-quark .
f ) propagator of @xmath60-quark and/or @xmath107-antiquark in covariant nrqcd ; @xmath108 is the 4-velocity of com of @xmath20 pair .
the second expression is the propagator of the charge conjugated field @xmath109 .
g ) & h ) interaction vertex of ultrasoft and/or potential gluon and @xmath60-quark in covariant nrqcd .
negative sign corresponds to the @xmath107-quark .
i ) propagator of the potential gluon in feynman gauge .
momentum @xmath110 is transverse with respect to the 4-velocity of the @xmath20-pair , @xmath111 .
j ) & k ) propagators of the ultrasoft and collinear gluons in feynman gauge.,height=83 ] a ) propagator of collinear quark ; @xmath106 is the lightlike vector of the @xmath5-quark .
b ) interaction vertex of the ultrasoft gluon and collinear quark .
c ) interaction vertex of the collinear gluon and collinear quark .
d ) propagator of the heavy @xmath4-quark .
e ) interaction vertex of the ultrasoft gluon and the heavy @xmath4-quark .
f ) propagator of @xmath60-quark and/or @xmath107-antiquark in covariant nrqcd ; @xmath108 is the 4-velocity of com of @xmath20 pair .
the second expression is the propagator of the charge conjugated field @xmath109 .
g ) & h ) interaction vertex of ultrasoft and/or potential gluon and @xmath60-quark in covariant nrqcd .
negative sign corresponds to the @xmath107-quark .
i ) propagator of the potential gluon in feynman gauge .
momentum @xmath110 is transverse with respect to the 4-velocity of the @xmath20-pair , @xmath111 .
j ) & k ) propagators of the ultrasoft and collinear gluons in feynman gauge.,title="fig:",height=56 ] a ) propagator of collinear quark ; @xmath106 is the lightlike vector of the @xmath5-quark .
b ) interaction vertex of the ultrasoft gluon and collinear quark .
c ) interaction vertex of the collinear gluon and collinear quark .
d ) propagator of the heavy @xmath4-quark .
e ) interaction vertex of the ultrasoft gluon and the heavy @xmath4-quark .
f ) propagator of @xmath60-quark and/or @xmath107-antiquark in covariant nrqcd ; @xmath108 is the 4-velocity of com of @xmath20 pair .
the second expression is the propagator of the charge conjugated field @xmath109 .
g ) & h ) interaction vertex of ultrasoft and/or potential gluon and @xmath60-quark in covariant nrqcd .
negative sign corresponds to the @xmath107-quark .
i ) propagator of the potential gluon in feynman gauge .
momentum @xmath110 is transverse with respect to the 4-velocity of the @xmath20-pair , @xmath111 .
j ) & k ) propagators of the ultrasoft and collinear gluons in feynman gauge.,title="fig:",height=75 ] a ) propagator of collinear quark ; @xmath106 is the lightlike vector of the @xmath5-quark .
b ) interaction vertex of the ultrasoft gluon and collinear quark .
c ) interaction vertex of the collinear gluon and collinear quark .
d ) propagator of the heavy @xmath4-quark .
e ) interaction vertex of the ultrasoft gluon and the heavy @xmath4-quark .
f ) propagator of @xmath60-quark and/or @xmath107-antiquark in covariant nrqcd ; @xmath108 is the 4-velocity of com of @xmath20 pair .
the second expression is the propagator of the charge conjugated field @xmath109 .
g ) & h ) interaction vertex of ultrasoft and/or potential gluon and @xmath60-quark in covariant nrqcd .
negative sign corresponds to the @xmath107-quark .
i ) propagator of the potential gluon in feynman gauge .
momentum @xmath110 is transverse with respect to the 4-velocity of the @xmath20-pair , @xmath111 .
j ) & k ) propagators of the ultrasoft and collinear gluons in feynman gauge.,height=75 ] a ) propagator of collinear quark ; @xmath106 is the lightlike vector of the @xmath5-quark .
b ) interaction vertex of the ultrasoft gluon and collinear quark .
c ) interaction vertex of the collinear gluon and collinear quark .
d ) propagator of the heavy @xmath4-quark .
e ) interaction vertex of the ultrasoft gluon and the heavy @xmath4-quark .
f ) propagator of @xmath60-quark and/or @xmath107-antiquark in covariant nrqcd ; @xmath108 is the 4-velocity of com of @xmath20 pair .
the second expression is the propagator of the charge conjugated field @xmath109 .
g ) & h ) interaction vertex of ultrasoft and/or potential gluon and @xmath60-quark in covariant nrqcd .
negative sign corresponds to the @xmath107-quark .
i ) propagator of the potential gluon in feynman gauge .
momentum @xmath110 is transverse with respect to the 4-velocity of the @xmath20-pair , @xmath111 .
j ) & k ) propagators of the ultrasoft and collinear gluons in feynman gauge.,title="fig:",height=86 ] a ) propagator of collinear quark ; @xmath106 is the lightlike vector of the @xmath5-quark .
b ) interaction vertex of the ultrasoft gluon and collinear quark .
c ) interaction vertex of the collinear gluon and collinear quark .
d ) propagator of the heavy @xmath4-quark .
e ) interaction vertex of the ultrasoft gluon and the heavy @xmath4-quark .
f ) propagator of @xmath60-quark and/or @xmath107-antiquark in covariant nrqcd ; @xmath108 is the 4-velocity of com of @xmath20 pair .
the second expression is the propagator of the charge conjugated field @xmath109 .
g ) & h ) interaction vertex of ultrasoft and/or potential gluon and @xmath60-quark in covariant nrqcd .
negative sign corresponds to the @xmath107-quark .
i ) propagator of the potential gluon in feynman gauge .
momentum @xmath110 is transverse with respect to the 4-velocity of the @xmath20-pair , @xmath111 .
j ) & k ) propagators of the ultrasoft and collinear gluons in feynman gauge.,title="fig:",height=56 ] a ) propagator of collinear quark ; @xmath106 is the lightlike vector of the @xmath5-quark .
b ) interaction vertex of the ultrasoft gluon and collinear quark .
c ) interaction vertex of the collinear gluon and collinear quark .
d ) propagator of the heavy @xmath4-quark .
e ) interaction vertex of the ultrasoft gluon and the heavy @xmath4-quark .
f ) propagator of @xmath60-quark and/or @xmath107-antiquark in covariant nrqcd ; @xmath108 is the 4-velocity of com of @xmath20 pair .
the second expression is the propagator of the charge conjugated field @xmath109 .
g ) & h ) interaction vertex of ultrasoft and/or potential gluon and @xmath60-quark in covariant nrqcd .
negative sign corresponds to the @xmath107-quark .
i ) propagator of the potential gluon in feynman gauge .
momentum @xmath110 is transverse with respect to the 4-velocity of the @xmath20-pair , @xmath111 .
j ) & k ) propagators of the ultrasoft and collinear gluons in feynman gauge.,title="fig:",height=56 ] a ) propagator of collinear quark ; @xmath106 is the lightlike vector of the @xmath5-quark .
b ) interaction vertex of the ultrasoft gluon and collinear quark .
c ) interaction vertex of the collinear gluon and collinear quark .
d ) propagator of the heavy @xmath4-quark .
e ) interaction vertex of the ultrasoft gluon and the heavy @xmath4-quark .
f ) propagator of @xmath60-quark and/or @xmath107-antiquark in covariant nrqcd ; @xmath108 is the 4-velocity of com of @xmath20 pair .
the second expression is the propagator of the charge conjugated field @xmath109 .
g ) & h ) interaction vertex of ultrasoft and/or potential gluon and @xmath60-quark in covariant nrqcd .
negative sign corresponds to the @xmath107-quark .
i ) propagator of the potential gluon in feynman gauge .
momentum @xmath110 is transverse with respect to the 4-velocity of the @xmath20-pair , @xmath111 .
j ) & k ) propagators of the ultrasoft and collinear gluons in feynman gauge.,title="fig:",height=56 ] in short , we need six fields to reproduce the infrared ( ir ) limit of the full theory amplitude : 1 .
the outgoing @xmath5-quark is described by the scet spinor @xmath112 ( see e.g. @xcite , @xcite , and @xcite ) . the large component of the collinear momentum @xmath113 becomes the label @xmath114 on the field and the residual @xmath115-dependence is ultrasoft .
the scet collinear gluon @xmath116 transferring the collinear momentum .
the momentum components scale like those of the collinear quark .
the incoming @xmath4-quark is described by the hqet spinor @xmath117 @xcite .
the outgoing @xmath60- and @xmath107-quarks are described by cnrqcd spinors @xmath118 and @xmath119 , respectively .
see the next subsection for the discussion .
the ultrasoft field @xmath120 . this field transfers momentum @xmath121 .
scet and nrqcd also include the soft gluons which transfer momenta @xmath122 and @xmath123 , respectively . at one loop the soft gluons
do not contribute to the effective theory decay amplitude , so they re not listed above .
the soft gluons are discussed in section [ sec:6 ] where tree - level matching to all orders in @xmath124 is performed . in the center - of - mass frame of the @xmath20 pair ( @xmath67 ) the effective lagrangian is provided by nrqcd .
the desired covariant lagrangian can be obtained from the nrqcd lagrangian @xcite in the @xmath67 by a lorentz boost to the frame where the center - of - mass velocity is @xmath108 . at the leading order in @xmath73-expansion
this gives : [ cclag2 ] _ , p = | _ _ , _ _ , - |^c _ _ , ^c _ _ , . here
the covariant derivative @xmath125 contains only the ultrasoft field @xmath126 .
when calculating one - loop qcd corrections to the effective theory lagrangian it is more convenient to work with the charge conjugated field @xmath127 because it creates the anti - quark @xmath107 in the final state and this is exactly what we need in the effective theory for the @xmath6 decay . at the leading order in @xmath73-expansion the effective theory dirac spinors @xmath128 and @xmath127
are related to the full theory field operator @xmath7 as : [ operators ] ( x)=e^-im_c_x_0e^-i x__,(x)+e^im_c_x_0e^i x^c__,(x)+ , with @xmath129 the covariant potential labels .
we use symbols @xmath130 and @xmath131 for the quark and anti - quark fields following the conventional nrqcd notation .
however unlike in the standard approach we keep all four components of the spinors to ensure relativistic covariance .
the 4-spinors satisfy the constraints @xmath132 here @xmath133 .
the solutions of ( [ const ] ) are such that in the @xmath67 their upper components become the conventional nrqcd two - component pauli spinors @xmath134 and @xmath135 and the lower components vanish : @xmath136 in other words @xmath137 and @xmath138 are the nrqcd spinors @xmath134 and @xmath135 boosted into the frame with velocity @xmath139 with respect to the @xmath67 . in the @xmath67 @xmath140 and the standard form of the nrqcd lagrangian is reproduced .
this formalism is essentially the same as the one discussed in @xcite .
the spinor @xmath127 satisfies [ eq14 ] = , where @xmath141 , and transforms under the same representation of @xmath142 as @xmath137 .
the explicit form of @xmath127 in terms of pauli spinor @xmath135 is given by ( [ vspin3 ] ) .
nrqcd also includes the off - shell quark modes called soft quarks .
the energy and momentum of the soft quark both scale like @xmath123 and therefore the lagrangian for the soft quark at the leading order in @xmath73-expansion is the same as hqet lagrangian .
however we will see in section [ sec:4 ] that at one loop the soft gluons do not contribute to the decay amplitude and so we do not discuss them here .
the nrqcd part of the decay amplitude also includes a @xmath143 ( coulomb ) singularity .
this singularity is reproduced by the local ( on the ultrasoft scale ) four - quark operator ( see @xcite and @xcite ) corresponding to the coulomb interaction between the heavy quarks . in this paper instead we have adopted a simplistic approach : we introduce the potential gluon that transfers the frequency @xmath144 and momentum @xmath123 ( in the @xmath67 ) and does not remove the quarks off - shell @xcite .
this is enough to reproduce the coulomb singularity at one loop .
the lagrangian that couples the potential gluon and the on - shell quarks changes the momentum label @xmath129 on the quark fields @xmath130 and @xmath109 and gives rise to the same interaction vertex as the ultrasoft gluon . matching a full qcd matrix element onto the matrix element of the cnrqcd requires writing down the full qcd spinors @xmath145 and @xmath73 in terms of the cnrqcd spinors @xmath130 and @xmath131 .
this is done as in nrqcd only the frame of reference is not specified .
for example : [ uspin ] & & ( m+_c - m)u = 0 , & & ( -2mp^-_+_c)u = 0 , & & ( -2mp^-_+p^-__c)(p^+_+p^-_)u = 0 , & & ( -2mp^-_+p^-__cp^+_-p_c p^-_)u = 0 , & & p^+_u = _ _ , , p^-_u = u_- , u = _ _ , + u_- , & & -(2m+p_c)u_-+p^-__c _ _ , = 0 , & & u_-=p^-__c _ _ , , & & u = _ _ , ( c _ * p * + _ * p * ) . here
@xmath146 denotes the sum of the perpendicular and residual components in the momentum decomposition given in ( [ pc : scaling ] ) . in the equation
the effective theory spinor @xmath147 is introduced as a projection of the full qcd spinor @xmath145 onto the subspace restricted by the constraint ( [ const ] ) .
the second half of the bispinor subspace @xmath148 accounts for the difference of the order @xmath73 between the full and effective theories .
one can see that in the @xmath67 the full qcd spinor is reproduced .
unlike the @xmath130-spinor which has only two upper components in the @xmath67 , the full qcd spinor has also two non - zero lower components suppressed by the first power of velocity in the @xmath67 . to parameterize the @xmath73-spinor of the full theory by the @xmath109-spinor notice that the charge conjugated spinor @xmath149 satisfies the same equation as the @xmath145-spinor .
the cnrqcd lagrangian for the anti - quark is derived from the full theory using charge conjugation .
therefore the @xmath150-spinor must be written in terms of the effective theory @xmath131-spinor in the same way as the @xmath145-spinor is written in terms of the @xmath130-spinor and then the @xmath73-spinor follows after charge conjugation : [ vcspin ] v= i^2v^c^ & & = i^2 ^ ^ _ _ , , & & = ( i^2^ _ _ , ) , & & = ^c _ _ , . here , componentwise : [ vspin3 ] ^c _ _ , = ( i^2^ _ _ , ) = ( c ( -i^2_*p*^ ) + ( -i^2_*p*^ ) ) ( -i^2^ ) = ( -^_,^_)^t . notice that the spinor @xmath151 satisfies ( [ eq14 ] ) , i.e. for the charge conjugated spinor the 4-velocity is reversed
to obtain the effective theory electroweak ( ewet ) lagrangian we perform the tree - level matching between the full theory amplitude of @xmath152 decay and the decay amplitude in the effective theory to all orders in @xmath124 and at the leading order in @xmath73 and @xmath153
. then the result is rg - improved using the anomalous dimension matrix extracted from the results of one - loop matching which corresponds to summing the leading sudakov logarithms and gives the ewet lagrangian in the leading logarithm approximation .
the soft gluons do not contribute at one loop and therefore do not modify the anomalous dimension matrix in the leading logarithm approximation . from this
we infer that in this approximation the soft gluons can be ignored . at tree level
the matching is done by replacing the full theory fields of ( [ eft:4 ] ) with the effective theory fields of section [ fields ] and using the reduction formula ( see appendix [ sec:10 ] ) [ etstruct1 ] _ p_l^p_lp_r _
-(_n)p_r_5 - _ -_+ to decompose the full theory dirac structure into the sum of the three dirac structures of the effective theory .
as we have inferred the collinear gluons are the only degrees of freedom giving rise to the local operators on the effective theory side in the leading logarithm approximation
. then the result of tree - level matching follows easily from the @xmath154 calculation and the scet invariance of the effective theory lagrangian that requires the collinear gluons and the collinear quark fields to combine in the jet field @xcite : [ ewetmatch2 ] ^()= [ |_n , pw_j * c*h__b ] [ |___j*c*^c__- ] . here
@xmath155 and @xmath156 .
the sum over potential label @xmath129 is understood .
the operator @xmath19 is the wilson line [ wl ] w=_(-g_s|na_n , q ) where @xmath157 is the operator that picks up the net label momentum on the right .
momentum conservation implies that at each order in @xmath124 the labels on the collinear fields satisfy @xmath158 .
the corresponding feynman diagram in the effective theory is shown in fig .
[ fig:4 ] .
-quark ( labeled by 4-velocity @xmath159 ) , the dashed line corresponds to the outgoing light @xmath5-quark ( labeled by collinear momentum @xmath160 ) , the solid lines correspond to the outgoing quarks of @xmath20-pair ( labeled by the potential momenta @xmath161 ) . for the @xmath107 anti - quark the fermion flow is reversed as in the full theory .
only one collinear gluon incoming to the electro - weak vertex is shown.[fig:4],height=102 ] here we present the results of one - loop calculation of the decay amplitude in the effective and full theories .
feynman diagrams are calculated in the @xmath163-scheme .
the momenta of the four external particles are set on - shell and the corresponding ir divergences are regulated by the dimensional regularization .
overall in the effective theory there are eight amputated diagrams .
seven of them are due to loop corrections to the zero order term in the @xmath124 expansion of the tree - level lagrangian ( [ ewetmatch2 ] ) : & & ^(0)_ij= [ |_n , p_j * c*_i h__b ] [ |___j*c*_i^c__- ] , [ gamma:1 ] where @xmath164 stands for the singlet and octet operators and @xmath165 for the dirac structures introduced in ( [ etstruct1 ] ) .
six diagrams where the ultrasoft gluon is exchanged between the quarks are shown in fig .
[ fig : usoft ] . diagrams with the ultrasoft gluon exchange in the effective theory.,height=56 ] diagrams with the ultrasoft gluon exchange in the effective theory.,height=56 ] diagrams with the ultrasoft gluon exchange in the effective theory.,height=56 ] diagrams with the ultrasoft gluon exchange in the effective theory.,height=56 ] diagrams with the ultrasoft gluon exchange in the effective theory.,height=56 ] diagrams with the ultrasoft gluon exchange in the effective theory.,height=56 ] two more diagrams in fig .
[ fig : pot ] correspond to the potential gluon exchange between the quarks of the @xmath166-pair and the emission of collinear gluon from the electro - weak vertex .
-pair and collinear gluon emission from the electro - weak vertex .
[ fig : pot],height=56 ] -pair and collinear gluon emission from the electro - weak vertex .
[ fig : pot],height=75 ] the diagrams involving the soft cnrqcd gluons emitted from the electro - weak vertex are shown in fig .
[ fig : soft ] .
all these diagrams include also the off - shell soft quarks .
the propagators for the soft quarks are hqet propagators because the term @xmath167 and the kinetic energy term in the cnrqcd propagator @xmath168 must be discarded .
this makes feynman integrals corresponding to the diagrams in fig .
[ fig : soft ] scaleless and thereby identically zero in the dimensional regularization .
so the soft gluons do not contribute to the amplitude at one loop .
-pair and soft gluon emission from the electro - weak vertex .
the zigzag line corresponds to soft gluon .
the double lines for @xmath60 and @xmath107-quarks stand for the soft propagators .
these diagrams are identically zero in the dimensional regularization.[fig : soft],height=64 ] -pair and soft gluon emission from the electro - weak vertex .
the zigzag line corresponds to soft gluon .
the double lines for @xmath60 and @xmath107-quarks stand for the soft propagators .
these diagrams are identically zero in the dimensional regularization.[fig : soft],height=64 ] -pair and soft gluon emission from the electro - weak vertex .
the zigzag line corresponds to soft gluon .
the double lines for @xmath60 and @xmath107-quarks stand for the soft propagators .
these diagrams are identically zero in the dimensional regularization.[fig : soft],height=64 ] the non - renormalized amplitude is given by the sum of the diagrams in figs .
[ fig : usoft ] and [ fig : pot ] .
the ultraviolet ( uv ) divergences are then removed by the field renormalization factors for the heavy and light quarks , @xmath169 and @xmath170 , respectively , and by the renormalization matrix of the electroweak operators for @xmath171 , @xmath172 : & & z_h=1 + 2c_f , z_l=1-c_f , c_f= , & & z_00=1-c_f\ { + + } , z_08=0,z_80=0 , & & z_88=1-\ { c_f + + + } , [ tech:1 ] here @xmath173 is the energy of the @xmath5-quark in the @xmath62 .
finally the @xmath163-renormalized amplitude is multiplied by the lehmann - symanzik - zimmermann ( lsz)-factor @xmath174 . the latter is computed from the residues at the poles of heavy and light quark propagators also renormalized in @xmath163-scheme .
this gives the on - shell matrix element : [ tech:2 ] = = 1-c_f .
the resulting expression for the amplitude is given by : _
i=0,8 c_i_j=1,2,3[|_n , p_j * c*_i h__b ] [ |___j*c*_i^c__- ] , [ tech:4 ] & & c_0=1+c_f\ { - - -+ } , & & c_8=1+\ { c_f - - - & & - } .
[ tech:3 ] at one loop there is no mixing between the singlet and the octet operators : @xmath175 .
one can see why by writing down the color structure of the effective theory diagrams .
the diagrams of the first column in fig .
[ fig : usoft ] and the diagrams in fig .
[ fig : pot ] do not alter the color structure of the vertex .
the remaining four diagrams in fig .
[ fig : usoft ] actually mix the singlet and octet operators but the interaction vertices for the @xmath60 and @xmath107 quarks have different signs as it follows from the feynman rules of the effective theory and therefore the color structure of the sum of the second and the third and/or the fifth and the sixth diagrams in fig .
[ fig : usoft ] is either t^a*c*t^a . for @xmath176
the commutator vanishes and for @xmath177 the direct product is reduced to @xmath178 . in the full theory
the nonrenormalized amplitude is given by the sum of six diagrams in fig .
[ fig : ft ] .
diagrams contributing to the amplitude @xmath179 in the full theory.,height=56 ] diagrams contributing to the amplitude @xmath179 in the full theory.,height=56 ] diagrams contributing to the amplitude @xmath179 in the full theory.,height=56 ] diagrams contributing to the amplitude @xmath179 in the full theory.,height=64 ] diagrams contributing to the amplitude @xmath179 in the full theory.,height=56 ] diagrams contributing to the amplitude @xmath179 in the full theory.,height=56 ] we are using the ndr scheme for @xmath180 .
the on - shell value of the full theory amplitude follows when the sum of the six diagrams is multiplied by the field renormalization factors @xmath181 , the operator renormalization matrix @xmath182 , and the lsz factor : & & z_q=1-c_f , z_ij = _ ij + ( cc 0 & 6 + & -2 ) , + & & = 1+(--2 - 4 - 8 ) . the final step is projecting the full theory dirac structures to the dirac structures of the effective theory which corresponds to taking the limit @xmath183 and @xmath184 .
this is done by means of the reduction formulae listed in appendix [ sec:10 ] .
the projected full theory amplitude contains only the dirac structures of eq .
( [ etstruct1 ] ) . at one loop
the matrix element on the full theory side upon expanding the dirac structures in terms of the effective theory structures becomes : [ wils:7 ] [ a^(0)_i+a^(1)_i]o_i . here
the index @xmath185 is cumulative and labels both the dirac and color structure .
on the effective theory side the matrix element has the form : [ wils:8 ] , where @xmath186 is the first order correction to the wilson coefficient and the second bracket is the matrix element in the effective theory at @xmath162 as it comes from the one - loop diagrams .
equating ( [ wils:7 ] ) and ( [ wils:8 ] ) gives the matching condition at one - loop : [ wils:9 ] c^(0)_i = a^(0)_ic^(1)_i = a^(1)_i - c^(0)_jx^(1)_ji = a^(1)_i - a^(0)_jx^(1)_ji .
the coefficients @xmath187 have been introduced in the tree - level matching formula ( [ etstruct1 ] ) : c^(0)_i(r)\{,- , -}r=. [ wils:2 ] the analytic expressions of the coefficients @xmath186 are given in appendix [ sec:12 ] .
their numerical values at the matching scale of the decay are discussed in section [ sec:6b ] .
the coefficients @xmath188 are given by eq .
( [ tech:3 ] ) and contain the ir divergences of the effective theory .
the coefficients @xmath189 contain both the finite part and the ir divergences of the full theory .
we confirm the cancellation of the ir - divergences between the full and effective theory amplitudes . from this fact
we infer that at the leading order in the effective theory power expansion and in the leading logarithm approximation the effective degrees of freedom include only those listed in section [ fields ] .
this is the main result of the paper .
the renormalization matrix @xmath172 in eq .
( [ tech:1 ] ) is diagonal and contains the terms @xmath190 , @xmath191 , and @xmath192 .
the terms proportional to @xmath193 and @xmath191 give rise to the leading sudakov logarithms @xmath194 as we will see shortly .
the terms @xmath192 give rise to the next - to - leading logarithms @xmath195 and should be combined with the next - to - leading logarithms coming from @xmath196 terms in the renormalization matrix @xmath172 calculated at two loops ( see table ii in @xcite for the counting scheme of the divergent terms ) . discarding the terms @xmath192
gives the renormalization factors @xmath197 which turn out to be equal to the renormalization factor for the @xmath198-current ( see @xcite ) : [ rencon ] z_00=z_88=1- c_f .
therefore the anomalous dimension matrix ( adm ) in the leading logarithm approximation is the same for all operators @xmath199 and coincides with the adm of the @xmath198-current in scet whose wilson coefficient is calculated in @xcite : [ eft:56 ] c^ll_j(,r)=c^(0)_j(r ) ( _ n=1^^n _ s^n(2e_s)^n+1 ) .
we have chosen the form ( [ eft:56 ] ) for the wilson coefficient to make it explicit that solving the rg - equation is indeed equivalent to summing up the ir double sudakov logarithms .
scet gauge invariance of the jet field @xmath200 $ ] ensures that loop corrections to this operator and to @xmath201 are the same @xcite .
this observation allows one to combine the tree - level matching formula ( [ ewetmatch2 ] ) with the wilson coefficient ( [ eft:56 ] ) by treating the latter as the operator function of @xmath202 inserted on the right to the jet field @xcite : _ w^ll(,_b)= - v_cb v^_cs _ i=0,8_j=1 ^ 3 c_i(_b ) [ |_n , pwc^ll_j(,r)_j * c*_ih__b ] [ |___j*c*_i^c__- ] .
[ wils:1 ] here @xmath203 picks up the net label momentum of the collinear operators on the left and @xmath204 are wilson coefficients ( [ eft:7 ] ) .
the operator @xmath202 picks up the total momentum of the jet at each order in the @xmath124 expansion . at tree - level
this amounts to replacing @xmath205 but when evaluating @xmath206-corrections the presence of the operator function in ( [ wils:1 ] ) becomes nontrivial .
the dependence of @xmath21 on the scales @xmath207 and @xmath55 is to remind that the lagrangian is obtained in two steps : firstly by running from the electroweak scale to @xmath208 and then to @xmath207 .
applying the ewet lagrangian ( [ wils:1 ] ) to the decay of the @xmath1-meson requires evaluating the matrix elements of the lagrangian between the hadronic states of the effective theory . in this section
we prove that in the leading logarithm approximation the matrix element of the ewet lagrangian between the effective theory states factorizes into the product of the matrix elements of the heavy - to - light current and decay constant of the charmonium state .
as we have seen the ultrasoft gluons are the only degrees of freedom common to all three sectors of the effective theory in the leading logarithm approximation .
the following field redefinition ( see e.g. @xcite ) makes it possible to rewrite the ewet lagrangian ( [ wils:1 ] ) in terms of the ultrasoft - free field operators by incorporating the ultrasoft gluons into wilson lines : [ usdec ] & & _ n , p = y_n_n , p^(0 ) , a_n , q = y_na_n , q^(0)y_n^,w = y_n w^(0)y_n^,&&h__b = y__bh__b^(0),&&__=y____,^(0 ) , ^c__-=y__^c(0)__,- .
here @xmath209 , with @xmath210 , or @xmath108 , stands for the path ordered exponent : y_l(x)=p(i g_s ^x_-ds la_us(ls ) ) .
it is a well - known fact that ultrasoft gluons do not renormalize the four - quark coulomb interaction term in nrqcd ( see e.g. @xcite ) and so , as long as soft gluons are ignored , the ultrasoft - free cnrqcd lagrangian includes only the ultrasoft - free potential quarks and potential gluons . applying these field redefinitions to ( [ wils:1 ] )
gives : [ der:22 ] ^ll_ij=[|_n , p^(0)w^(0 ) c^ll_j(,r)y_n^_j * c*_i y__bh__b^(0 ) ] [ |__^(0 ) y__^_j * c*_i y__^c(0)__- ] .
it is instructive to pause and check that at one loop corrections to ( [ der:22 ] ) exactly reproduce the effective theory amplitude given by ( [ tech:4 ] ) and ( [ tech:3 ] ) .
expanding the ultrasoft and collinear wilson lines in ( [ der:22 ] ) to the first order in @xmath124 reproduces the diagrams shown in fig .
[ fig : usoft ] and the second diagram in fig .
[ fig : pot ] .
terms of the order @xmath211 coming from expanding the @xmath212 are self - energy diagrams and must be discarded since only the amputated diagrams contribute according to lsz prescription .
wilson lines @xmath212 bring the renormalization factors which reproduce the renormalization factors of the quark fields due to the ultrasoft self - energy diagrams .
expanding @xmath213 reproduces only the hqet part of the potential quark propagator in fig .
[ frules ] .
this is sufficient because the corresponding diagrams ( 2nd , 3rd , 5th , and 6th diagrams in fig .
[ fig : usoft ] ) do not have @xmath143-divergences .
physically for the ultrasoft gluon the cnrqcd quark is the same as hqet quark as long as the relative motion of the quarks in the @xmath67 is ignored .
consider the matrix element of ( [ der:22 ] ) between the states @xmath215 and @xmath216 , where @xmath217 is any state which includes a collinear @xmath5-quark .
the states @xmath218 and @xmath219 do not contain ultrasoft gluons , so the matrix element of the singlet operator factorizes : [ eft:33 ] [ |_n , p^(0)w^(0)c^ll_j(,r)_j y_n^ y__bh__b^(0 ) ] [ |__^(0)_j ^c(0)__- ] . now let us show that the matrix element of the octet operator ( @xmath177 ) vanishes and therefore the color state of the @xmath20-pair does not contribute to the decay rate ( in the leading logarithm approximation ) . to this end
it would suffice to show vanishing of the correlator [ eft:31 ] & & \ { _ ^(0)(x ) [ |_n , p^(0)w^(0)c^ll_j(,r)_j y_n^ t^a y__bh__b^(0 ) ] & & ( y ) } , where @xmath220 is the ultrasoft - free interpolating operator carrying quantum numbers of @xmath221 , the bound state of the @xmath20-pair .
the matrix elements ( [ eft:31 ] ) and ( [ eft:33 ] ) with @xmath222 are related by lsz reduction formula and vanishing ( [ eft:31 ] ) implies that ( [ eft:33 ] ) is also zero .
the diagrams contributing to the t - product in ( [ eft:31 ] ) are the ladder diagrams shown in fig .
[ ladder ] , where the potential gluons are exchanged between the quarks of @xmath20-pair .
ladder diagrams that contribute to the t - product ( [ eft:31 ] ) at the leading order in @xmath223 .
wiggly lines represent ultrasoft gluons which couple between themselves and to the external state @xmath1 .
the collinear gluons ( not shown ) couple to the collinear @xmath5-quark line and to the collinear degrees of freedom in @xmath224.,height=188 ] the interpolating operator @xmath220 and the currents in ( [ eft:31 ] ) are color singlets , therefore the color structure of the diagram in fig .
[ ladder ] is : [ eft:32 ] ~a^a , a_1,a_2
\{t^b_1t^b_2 [ [t^a , t^a_1],t^a_2], ] t^b_2t^b_1 } here @xmath225 and @xmath226 are adjoint indices .
the commutators inside the trace arise upon expansion of the ultrasoft operators @xmath213 .
the insertions of @xmath227 on both sides are due to potential gluon exchange .
the quantity @xmath228 represents the color structure arising when contracting the ultrasoft fields from the @xmath20-current with the fields of the heavy - to - light current , and with the ultrasoft degrees of freedom in the @xmath229 state .
expanding the commutator inside the trace gives the structure proportional to @xmath230 .
contracting the indices @xmath226 on both sides of @xmath230 again gives a structure proportional to @xmath230 .
therefore the whole trace is proportional to the trace of @xmath230 and vanishes .
if it were not for the soft nrqcd gluons next - to - leading corrections could be included systematically into @xmath231 of lagrangian ( [ wils:1 ] ) . without the soft gluons we simply have three hqet quarks , two of them propagate along the same wilson line in the opposite directions and as the matching shows decouple from the @xmath198 current . to see explicitly whether soft gluons violate factorization in the next - to - leading logarithm approximation two loop calculation in the effective theory is required .
we have not attempted this calculation , although made the first step in this direction by performing the tree - level matching for the ewet lagrangian that includes both collinear and soft nrqcd gluons .
the matching is performed using the technique explained in the appendix a of @xcite .
the method is to introduce the auxiliary fields corresponding to the off - shell quarks and gluons , to write down the lagrangian for the auxiliary and on - shell fields , and then to integrate out the auxiliary fields by solving the equations of motion ( eom ) .
the integration corresponds to the tree - level matching to all orders in @xmath124 .
the result given by ( [ eweta3c ] ) does not exactly reproduce the corresponding result for the heavy - to - light current in scet ( see e.g. eq . ( a22 ) in @xcite ) because the soft nrqcd gluons are not protected by the scet symmetry under the soft gauge transformations . in this section
we briefly outline the derivation .
the first step is to draw all tree - level diagrams that introduce the off - shell fields .
figures [ aux2 ] and [ aux5 ] show how the off - shell modes arise for the quarks of @xmath20-pair and @xmath5-quark , respectively .
-quark and the off - shell mode @xmath232 .
interaction between collinear gluon and the off - shell mode @xmath232 is given by the same vertex .
b ) collinear gluon interacting with @xmath233-field of the antiquark and the off - shell mode @xmath234 . interaction between collinear gluon and the off - shell mode @xmath234 is given by the same vertex .
the overall sign is opposite relative to a ) due to the reversed fermion flow .
[ aux2],title="fig:",height=75 ] -quark and the off - shell mode @xmath232 .
interaction between collinear gluon and the off - shell mode @xmath232 is given by the same vertex .
b ) collinear gluon interacting with @xmath233-field of the antiquark and the off - shell mode @xmath234 .
interaction between collinear gluon and the off - shell mode @xmath234 is given by the same vertex .
the overall sign is opposite relative to a ) due to the reversed fermion flow .
[ aux2],title="fig:",height=75 ] -quark and the off - shell @xmath235-field .
interaction between collinear gluon and the off - shell mode @xmath235 is given by the same vertex.[aux5],height=75 ] for the @xmath4-quark we have to match in two steps .
both soft and collinear gluons give the @xmath4-quark an off - shell momentum but the corresponding off - shellnesses are of different orders of magnitude .
a soft momentum is @xmath236 and collinear momentum is @xmath237 .
once the collinear momentum is transferred to the @xmath4-quark any off - shell momentum brought by the soft gluons is subleading compared to the collinear momentum propagating along the line .
a diagram where a soft gluon is inserted between collinear gluon(s ) and the electro - weak vertex contributes the factor ~ , where @xmath238 is the collinear momentum propagating along the @xmath4-quark line .
( a gauge field scales like the momentum it transfers . )
therefore only the diagrams with the soft gluons next to the heavy quark field @xmath239 and the collinear gluons next to the electro - weak vertex shown in fig . [ bmatch ] will contribute .
soft gluons ( zigzag lines ) must be next to the heavy quark field.[bmatch],height=68 ] to account for this we introduce two off - shell fields for the @xmath4-quark : @xmath240 which is off - shell by @xmath236 and @xmath241 which is off - shell by @xmath242 , ( see fig . [ aux3 ] ) .
-quark and the off - shell mode @xmath240 .
the interaction between the soft gluon and the off - shell mode @xmath240 is given by the same vertex .
b ) collinear gluon interacting with the off - shell mode @xmath240 and the off - shell mode @xmath241 .
the interaction between the collinear gluon and the off - shell mode @xmath241 is given by the same vertex.[aux3],title="fig:",height=75 ] -quark and the off - shell mode @xmath240 .
the interaction between the soft gluon and the off - shell mode @xmath240 is given by the same vertex .
b ) collinear gluon interacting with the off - shell mode @xmath240 and the off - shell mode @xmath241 .
the interaction between the collinear gluon and the off - shell mode @xmath241 is given by the same vertex.[aux3],title="fig:",height=75 ] in addition to the soft and collinear gluons we should take into account the mode arising when a soft gluon fuses with a collinear one , see fig . [ aux6 ] .
the auxiliary gluon field @xmath243 transfers momentum @xmath244 in the light - cone coordinates ( compare to the collinear momentum @xmath245 ) .
recall that @xmath246 for the decay kinematics . at
the leading order in the effective theory power expansion the interaction vertex of the field @xmath243 with the heavy quarks is the same as for collinear gluons .
the interaction vertex of @xmath243 and the @xmath5-quark is the same as for a soft gluon .
the gluon lagrangian for @xmath243 is discussed in @xcite .
.[aux6],height=86 ] writing down the auxiliary lagrangians which generate the feynman rules listed above is straightforward .
the ewet operator is given by : [ ewetmatch3 ] ^()= [ ( |_n , p+|s)_j * c * ( b+b ) ] [ ( |__+|c)_j*c*(^c+^c__- ) ] . here
the auxiliary fields are understood as perturbative expansions of solutions for the corresponding eom following from the auxiliary lagrangians . for
the @xmath4-quark we have to solve first the eom for the @xmath240 field and then use the solution to solve eom for @xmath241 .
the solutions are [ solstuff ] & & |c=|__,(w_x^-1),&&^c=(w_x-1)^c__,-,&&|s=|_n , p(s_n , x^-1),&&b=(w_x-1)b,&&b=s__bh__b , where @xmath247 , @xmath248 , and @xmath249 are wilson lines [ wxsx ] & & s__b=_(g_s _ ba_s , q),&&w_x=_(- g_s [ |na_x , q+|na_c , q]),&&s_n , x^= _
( g_s [ na_x , q+na_s , q ] ) .
combining eqs .
( [ ewetmatch3 ] ) , ( [ solstuff ] ) , and ( [ wxsx ] ) gives : [ ewetmatch ] ^()= [ |_n , ps_x^w_x_j * c * s__bh__b ] [ |___j*c*^c__- ] .
solving the eom for the auxiliary field @xmath243 makes it possible to write @xmath250 in terms of @xmath251 and @xmath252 only .
this remarkable calculation is performed in appendix a of @xcite , to which the reader is referred .
the only difference is that here we work with the soft nrqcd gluons instead of the soft scet gluons .
the result is : [ ansatz ] s^_xw_x = ws_n^ , where @xmath19 is the collinear wilson line ( [ wl ] ) and @xmath253 is given by : [ ws ] s_n^= _
( g_s na_s , q ) .
using ( [ ansatz ] ) eliminates the auxiliary field @xmath243 from ( [ ewetmatch ] ) which gives : [ eweta3b ] ^()= [ |_n , pws^_n_j * c*s__bh__b ] [ |___j * c * ^c__- ] .
it is easy to verify this formula at @xmath154 by explicit matching . now
the net momentum of the @xmath20-pair is transversal with respect to the 4-velocity @xmath108 .
therefore the net momentum @xmath254 of the soft gluons emitted from the ewet vertex must be transversal as well because they end up inside the @xmath166-current : @xmath255 .
this observation suggests that ( [ eweta3b ] ) must be written as [ eweta3c ] ^()= [ |_n , pw__s^_n_j * c*s__bh__b ] [ |___j*c*^c__- ] , where the kronecker delta enforces transversality of the net soft momentum transferred from @xmath256- to @xmath20-current .
equation ( [ eweta3c ] ) is the final result .
equation ( [ eweta3c ] ) does not hint that the factorization holds in the next - to - leading logarithms approximation and beyond .
however , the transversality constraint results in something resembling the factorization . consider coulomb gauge . in this gauge gluons are transversal , so the soft gluons obey : @xmath257 . now
let us apply the transversality constraint in ( [ eweta3c ] ) _ separately _ to the soft momenta emitted from the @xmath4- and @xmath5-quark lines .
( actually the constraint restricts the _ net _ momentum . )
then the wilson lines @xmath253 and @xmath247 become equal . to see this notice that at the leading order in power expansion the momentum conservation reads : _
b=_+n . applying this equation to the argument of @xmath247 ( [ wxsx ] ) together with the transversality constraints yields : = s_n = s__b .
the last equation allows one to move the soft gluons from the @xmath198 current to @xmath166- current , so the factorization `` holds '' even when the soft nrqcd gluons are included .
however the attempts to make anything rigorous out of the above observation were unsuccessful .
most probably because the factorization is broken beyond the leading logarithm approximation . the one - loop matching ( see section [ sec:4b ] ) gives the initial values of the wilson coefficients @xmath186 at the matching scale @xmath259 . at this scale the ewet lagrangian in the next - to - leading logarithm approximation is : [ wils:4 ] & & _ w^nll(_b,_b)= - v_cb v^_cs _ i , k=0,8_j=1 ^ 3 c_i(_b)\{c^(0)_j(r)_ik+ c^(1)_ik;j(,r)}^(0)_kj .
the numerical values of @xmath260 for @xmath261 and @xmath262 are : & & c^(1)_ik;1(1,0.30)= ( cc -11.27 & -11.27 - 3.41 i + - 2.50 - 0.76 i & 8.94 - 11.55 i ) , & & c^(1)_ik;2(1,0.30)= ( cc 16.11 & 19.64 + 6.58 i + 4.36 + 1.46 i & -13.56 + 20.29 i ) , & & c^(1)_ik;3(1,0.30)= ( cc -10.75 & 10.81 + 1.23 i + 2.40 + 0.27 i & -7.87 + 9.00 i ) . [ wils:5 ] here indices @xmath185 and @xmath263 label the rows and columns , respectively .
comparatively large numerical values of the non - diagonal elements @xmath264 of the coefficients @xmath260 together with the observation that numerically the wilson coefficient @xmath265 is about ten times larger than the wilson coefficient of the singlet ( see section [ sec:2 ] ) suggest that the next to the leading logarithm correction to ( [ wils:1 ] ) is of the same order of magnitude .
in this paper we have proposed the effective theory for the decay @xmath6 , where @xmath7 is a charmonium state and @xmath8 a light hadron neglecting the effects due to the spectator quark .
we have identified the relevant degrees of freedom and verified explicitly by doing the matching calculation at one loop that the effective theory reproduces all the ir divergences of the perturbative qcd .
we have also shown that in the leading logarithm approximation the effective theory decay amplitude factorizes into the decay constant of the charmonium state and the matrix element of heavy - to - light current . the latter has been extensively studied . however including into consideration the soft nrqcd gluons most probably ruins the factorization beyond the leading logarithms .
at least the tree - level matching result ( [ eweta3c ] ) does not suggest the opposite .
the comparatively large value of @xmath266 in charmonium implies that the higher order corrections can not be ignored . to achieve a better quantitative understanding of non - leptonic @xmath1-decays into charmonium
we should definitely go beyond the leading logarithms .
our results can be used e.g. for calculating the rg - improved lagrangian in the next - to - leading logarithm approximation provided the anomalous dimension matrix of the effective theory is known to two loops .
m.s . is grateful to prof .
s. fleming for numerous conversations during the work on the project and to prof .
a. manohar for discussing the results .
the work of c.b . was supported by the bundesministerium fr bildung und forschung , berlin - bonn . the work of b.g . and m.s was supported in part by the us department of energy under contract de - fg03 - 97er40546 .
in this section we apply the ewet lagrangian ( [ wils:1 ] ) to the calculation of the semi - inclusive decay rate of @xmath267 . due to factorization of the charmonium state in the leading logarithm approximation
the decay can be treated along the same lines as the semi - leptonic decay @xmath268 @xcite .
the derivation below is presented in some detail .
the result is given by ( [ ddratefinal2 ] ) .
the differential decay rate is given by [ ddrate2 ] = _ x_s , ( 2)^4 ^ 4(p_b - p_j/-p_x ) . here
the differential decay rate depends only on @xmath270 that varies in the range [ ejrange ] e_j/. there is a region of @xmath270 where our effective theory can be used . in this region
the state @xmath224 must be collinear , i.e. the scet parameter @xmath271 , where @xmath272 is the invariant mass of the jet , is small .
on the other hand @xmath272 must be reasonably larger than @xmath87 , so the jet is inclusive .
it is customary to take @xmath273 as a value of scet parameter for a jet which is both collinear and inclusive .
the kinematic expressions for @xmath274 and @xmath272 in terms of @xmath270 are [ ddrate4 ] e_x = m_b - e_j / m_x^2=m_b^2+m_j/^2 - 2m_be_j/ , so we have to look for a region in ( [ ejrange ] ) where [ condition ] ~1 . for @xmath275
gev such a region exists near the center of the range ( [ ejrange ] ) where @xmath276 and @xmath277 gev .
the scet parameter is reasonably small , on a par with @xmath278 in the charmonium .
the matrix element in ( [ ddrate2 ] ) factorizes according to ( [ eft:33 ] ) .
so , the first step is the evaluation of the matrix element of the @xmath166-current ( see ( [ etstruct1 ] ) ) : [ mejpsi5 ] [ |__^(0)_j ^c(0)__- ] , _ j=\{_,_5,_+}. the matrix element of @xmath180 vanishes by parity ; it is non - zero for the pseudoscalar state @xmath280 .
the other two can be expressed via the @xmath279 decay constant . in the full theory
it is defined as ( see @xcite ) : [ deconstj ] [ |(x)_(x)]=-i f_j / m_j/_^(p ) e^i px . here
@xmath281 is the mass , and @xmath282 is the polarization vector of the state .
the decay constant determined from the leptonic decay mode @xmath283 is @xmath284mev @xcite . to match the left - hand side of ( [ deconstj ] ) to the cnrqcd at the leading order in @xmath73-expansion we replace the full theory operators with their expressions in terms of the ultrasoft - free effective theory fields ( see ( [ operators ] ) and ( [ usdec ] ) ) .
the states of the full theory are replaced with the states of ultrasoft - free cnrqcd . in the sum over @xmath129 and @xmath285 only the terms where @xmath286 survive .
also we should take into account the wilson coefficient @xmath287 , where @xmath288 is the reduced mass of the @xmath20-pair , that comes upon matching between the full and effective theory decay constants ( the coefficient @xmath289 in @xcite ) .
the exponent on the left - hand side is simply @xmath290 since taking into account the binding energy of the @xmath20-pair ( see discussion below ( [ scale ] ) ) would amount to replacing @xmath291 , which brings corrections of @xmath292 . on the right - hand side of ( [ deconstj ] ) we substitute @xmath293 .
equating both sides gives : [ deconstjet ] _
0[|__,^(0)^ ^c(0)__,-]=-i c^-1_j/(m_c/2)f_j / m_j/_^ ( _ ) . the hamiltonian density @xmath294 in the leading logarithm approximation is given by @xmath295 and its matrix element by ( [ eft:33 ] ) .
we write it here explicitly for reference : [ mejpsi ] & & ^ll_w(,_b)= v^_cb v^_cs c_0 ( _ b ) _
p0_p_0_je^i(m_j/_+-m_b_b ) x & & [ |_n , p^(0)w^(0)c^ll_j(,r)_j y_n^(x ) h__b(x ) ] [ |__^(0)_j ^c(0)__- ] . replacing the matrix element of the @xmath20-current with ( [ deconstjet ] )
gives : [ mejpsi2 ] & & ^ll_w(,_b)= v^_cb v^_cs c_0 ( _ b ) [ -i c^-1_j/(m_c/2 ) f_j
/ m_j/]&&_p0\ { j_n,_b(p_r;x ) [ ^(_)n _ ] - j_n,_b(_-(n);x ) [ ^(_)_+(n ) ] } , where [ etcurrent ] j_n,_b(_j;x)=e^i(m_j/_+-m_b_b)x[|_n , p^(0)w^(0)c^ll()_j y_n^h__b](x ) is the standard leading order expression for the scet heavy - to - light current slightly modified here by the presence of the phase factor of the @xmath20-state .
the function @xmath296 represents the exponential factor of ( [ eft:56 ] ) , the coefficients @xmath297 are written explicitly ( see ( [ wils:2 ] ) ) .
the sum over collinear labels in ( [ mejpsi2 ] ) is approximated by the single term corresponding to the stationary point of the exponent .
the label @xmath160 is chosen so that the momentum components in the directions @xmath106 and @xmath298 vanish .
in other words only oscillations on the ultrasoft scale are permitted in the exponential factor in ( [ etcurrent ] ) . in so
doing we have to keep the residual component along @xmath299 which is of the same order @xmath300 as the ultrasoft momentum of the matrix elements in ( [ mejpsi2 ] ) . expanding the exponent in ( [ etcurrent ] ) in light - cone coordinates in the @xmath1-meson frame , where @xmath301 , gives : [ momcon10 ] [ m_j/(_|n)+|n - m_b]+_+[m_j/(_n)-m_b ] . at the stationary point : [ labelcon ] m_j/(_|n)+|n - m_b=0_=0 .
momentum conservation in the @xmath1-meson frame reads : [ kinematika ] m_b_b = m_j/_+(e_x+|_x|)+(e_x-|_x| ) _ j/=_x .
using it we solve the first equation in ( [ labelcon ] ) for @xmath302 : @xmath299 in terms of @xmath270 .
finally , taking the sum over collinear labels in ( [ mejpsi ] ) amounts to replacing the effective theory current ( [ etcurrent ] ) by the expression [ etcurrent2 ] j_n,_b(_j;x)=e^i(e_j/+|*p*_j/|-m_b)(|nx)[|_n , p^(0)w^(0)_j y_n^h__b](x ) and substituting for the operator function @xmath296 its value at the stationary point @xmath303 .
note that although @xmath304 is set to zero , the @xmath305 is the invariant mass of the jet @xmath306 ( see ( [ ddrate4 ] ) ) and must be kept .
taking the square of ( [ mejpsi2 ] ) and summing over polarizations of @xmath279 gives : [ mejpsi3 ] & & |^ll_w(,_b)|^2=2 ^2|v^_cb v^_cs|^2 ^2f_j/^2&&\{m_b^2| j_n,_b(p_r;0 ) |^2+m_j/^2| j_n,_b(_-(n);0 ) summing over the inclusive collinear states in the last line of ( [ mejpsi3 ] ) is performed by means of the standard trick .
the sum is equal to the imaginary part of the t - product of the effective theory currents ( [ etcurrent2 ] ) : [ etjjvsdr ] - 2 t_j(p_j/ ) t_j(q)=-id^4 x e^-i qx \{j^_n,_b(_j;x)j_n,_b(_j;0)}. time - ordering in this expression is applied only to the fields which carry labeled momenta .
the ultrasoft fields belong to the zero bin @xcite of the sum over collinear momenta and are not affected by the time ordering , the corresponding operators are simply averaged over the @xmath229-state . the matrix element in ( [ etjjvsdr ] ) is then reduced to a convolution of the perturbative jet - function and a shape function of @xmath1-meson which is a non - perturbative object .
those functions are defined as : [ jetfunction ] & & [ w^(0)_n , p^(0)](x)[|_n , p^^(0)w^(0)]=ie^-ikxj_p(k^+),&&s(l^+)=[|h__b(ind_us - l^+)h__b ] . here
@xmath307 and @xmath308 .
the calculation is almost identical to @xmath268 @xcite to which the reader is referred \,f(x),\ ] ] where @xmath309 with @xmath310 being a ( matrix ) function of a single variable .
the identity is not difficult to prove by showing that both the path - ordered exponent and the series change identically under small variations of @xmath115 . ] .
the only difference is due to dirac structures . in our case
@xmath311 is either @xmath312 or @xmath313 . to average the dirac structures over @xmath229-meson state we use the fact that the b - meson is a pseudoscalar and in hqet the spin of the heavy quark decouples from the gluon field at the leading order in @xmath314-expansion
the calculation is straightforward and gives : [ pr3eps3 ] ^0p_r^^0 p_r=^0_-(n)^^0 _
-(n)=. finally the differential decay rate in terms of the jet and shape functions becomes : [ ddratefinal2 ] & & = ^2|v^_cb v^_cs|^2 ^2f_j/^2 ( m_b+2 ) & & dk^+ s(k^+ , ) .
the @xmath207-dependence of the jet and shape functions cancels that one of the wilson coefficients , so the differential decay rate is @xmath207-independent .
the argument of the jet function contains the invariant mass of the jet , @xmath315 given by ( [ ddrate4 ] ) .
the scale @xmath55 is the matching scale of the order of the @xmath4-quark mass @xmath316 .
here we give the explicit expressions for the coefficients @xmath317 ( see ( [ wils:4 ] ) ) . to simplify the output
we introduce functions @xmath318 , @xmath319 , and @xmath320 : & & f_1(r,/m_b)=- ^2- -+_2 + ^2,&&f_2(r)=(-i ) , & & f_3(r,/m_b)=- ^2 + ( --)-3 i & & + ( li_2-li_2)-3 li_2-_2&&+- ^2 + 3 -&&+ + + + [ 2]^2 .
the coefficients @xmath260 are enumerated according to the effective theory dirac structures @xmath321 . & & c^(1)_00;1=(f_1(r,/m_b)-+ ) , & & c^(1)_08;1= ( -6 + f_2(r)-23+),&&c^(1)_80;1= c^(1)_08;1,&&c^(1)_88;1= ( f_3(r,/m_b)-.&&.+-3[4r]-+).&&c^(1)_00;2=-(f_1(r,/m_b ) -+),&&c^(1)_08;2=-(-6 + f_2(r)-),&&c^(1)_80;2=c^(1)_08;2,&&c^(1)_88;2= -(f_3(r,/m_b)-+-3 [ 4r].&&.+).&&c^(1)_00;3=-(f_1(r,/m_b)-+(4- ) ) , & & c^(1)_08;3= -(-6+f_2(r)-20 - -),&&c^(1)_80;3=c^(1)_08;3,&&c^(1)_88;3=- ( f_3(r,/m_b)-6 - -+[r].&&.- ) .
in this section we present without derivation the relations used to calculate the wilson coefficients @xmath322 and @xmath323 given in eq .
( [ eft:7 ] ) .
the wilson coefficients for the set of operators [ misyak ] p_1 = [ |s _ p_l t^a c ] [ |c ^ p_l t^a b ] , p_2 = [ |s _ p_l c ] [ |c ^ p_l b ] .
have been calculated in @xcite at the nnlo in @xmath223 .
we need however the wilson coefficients in the different operator basis [ our ] _ 1 , _ 2 . at the lo
the two bases are related by a linear transformation which follows when one applies the fierz transformations in the color and spinor spaces : ( c _ 1 + _ 2 ) = ( cc - & + 2 & ) ( c p_1 + p_2 ) .
[ trafo : d4 ] at the nlo the one - loop corrections have to be taken into account and that brings complications due to the evanescent operators .
the dimensional regularization used when calculating feynman integrals does not have an unambiguous definition for matrix @xmath180 , which is essentially a four - dimensional object .
the systematic way of coping with this difficulty ( see , e.g. @xcite ) consists in expanding the operator basis to include operators vanishing in the limit @xmath324 ( evanescent operators ) . in our case
the ambiguity is brought by the fiertz transformation ( [ trafo : d4 ] ) which is not defined at @xmath325 .
working out the relation between the bases ( [ misyak ] ) and ( [ our ] ) at the nlo requires introduction of two evanescent operators ( c f_1 + f_2 ) = ( c _ 1 + _ 2 ) - ( cc - & + 2 & ) ( c p_1 + p_2 ) [ trafo : d41 ] and calculating one - loop diagrams analogous to those shown in fig .
[ fig : ft ] with the operators ( [ trafo : d41 ] ) inserted .
it suffices to extract only the uv - divergent part of the additional diagrams .
the wilson coefficients of operators ( [ misyak ] ) and ( [ our ] ) are related by [ wc : trafo : mu ] ( ) = ( cc - & 2 + & ) + ( cc 0 & 6 + - & 0 ) ( c c_1^(0 ) ( ) + c_2^(0 ) ( ) ) . here
@xmath326 are the wilson coefficients for the basis ( [ our ] ) and @xmath327 for the basis ( [ misyak ] ) .
the first matrix in eq . ( [ wc : trafo : mu ] ) is the transposed matrix in eq .
( [ trafo : d4 ] ) corresponding to the tree - level transformation of the operator basis . the second matrix in ( [ wc : trafo : mu ] )
is the contribution of operators ( [ trafo : d41 ] ) .
the explicit expressions for the wilson coefficients @xmath327 can be found in eqs .
( 39)-(44 ) in @xcite . substituting them into eq . ( [ wc : trafo : mu ] ) gives the required expressions for @xmath322 and @xmath323 .
in this section we present the reduction formulae necessary to project the matrix elements of the full theory into the matrix elements of the effective theory .
the generic dirac structure on the full qcd side has the form @xmath328 \left [ \bar{c}\ , \gamma_2\ , c \right],\ ] ] where @xmath329 is one of the following products of dirac matrices @xmath330 here the last four entries come from one - loop corrections to the tree - level full theory operator . on the effective theory side
the generic matrix element is @xmath331 [ \bar{\xi}_{\beta_{\cc}\,\pp}\gamma^{et}_{\cc}\,\eta^c_{\beta_{\cc}\,-\pp}],\ ] ] where @xmath332 and @xmath333 are the set of operators which form the complete set on the corresponding subspace of dirac spinors .
the dirac structures @xmath333 follow by sandwiching the full theory structures of eq .
( [ set ] ) between the projectors @xmath334 and @xmath335 p _ _ p_-_^et_. working out the dirac structures @xmath332 requires more effort because now the original dirac structure from the list ( [ set ] ) is sandwiched between the projectors on different subspaces : @xmath336 and @xmath337
. the most straightforward way to work out the necessary relations is using eq .
( 17 ) from @xcite that reads [ complete ] ^et_sb= - _ 5 + ^_. here @xmath338 . the projection formulae are considerably simplified in the restframe of the @xmath4-quark ( @xmath62 ) where @xmath339 .
the four - momentum of the @xmath5-quark in the @xmath62 is specified by its energy : @xmath340 . to simplify the results further we use the relations [ polarization ] ^_- i^ _ _
= -2^_+_- , ^_+ i^ _ _ = -2^_-_+ , where @xmath341 are the polarization vectors of @xmath5-quark with @xmath342 corresponding to the spin direction along / opposite to the quark momentum . here
these relations can be verified explicitly in the @xmath62 where @xmath344 , @xmath345 , @xmath346 , and @xmath347 .
equations below summarize the results . on the @xmath20-side
the following relations hold : [ ccbarr ] ^p_l & & ^_- ^__5 , ^p_r & & ^_+ ^__5 , ^[^]p_l & & ( _ ^^_- _ ^^ _ ) - ^ _ _ _ , ^[^]p_r & & ( _ ^^_- _ ^^ _ ) + ^___. here @xmath348 and @xmath349}= \frac{1}{2}[\g^\mu,\g^\nu]$ ] . on the @xmath256-side we use relations [ sbr ] ^p_l&&n^p_r- ^_+_- , ^p_r&&n^p_l- ^_-_+ , ^[^ ] p_l&&(n^_b^- n^_b^ )
p_l + i^_p_l + ( n^^_- - n^^_-)_+ , ^[^ ] p_r&&(n^_b^- n^_b^ ) p_r -i^_p_r + ( n^^_+ - n^^_+)_- .
using eqs .
( [ ccbarr ] ) and ( [ sbr ] ) it is straightforward to verify ( [ etstruct1 ] ) . finally , there is one more relation holding at the leading order in @xmath73-expansion that allowed us to reduce the number of the effective theory dirac structures when processing the output of the full qcd calculation : [ czero ] _
b=_. it follows from conservation of momentum , @xmath350 , and observation that on the @xmath20-side only the components perpendicular to @xmath108 contribute , which is obvious from ( [ ccbarr ] ) .
n. isgur and m. b. wise , phys .
b * 232 * , 113 ( 1989 ) .
n. isgur and m. b. wise , phys .
b * 237 * , 527 ( 1990 ) .
b. grinstein , nucl .
b * 339 * , 253 ( 1990 ) .
h. georgi , phys .
b * 240 * , 447 ( 1990 ) .
a. f. falk , h. georgi , b. grinstein , and m. b. wise , nucl .
b * 343 * , 1 ( 1990 ) .
a. f. falk , b. grinstein , and m. e. luke , nucl .
b * 357 * , 185 ( 1991 ) . c. w. bauer , s. fleming and m. luke , phys .
d * 63 * , 014006 ( 2000 ) .
[ arxiv : hep - ph/0005275 ] c. w. bauer , s. fleming , d. pirjol , and i. w. stewart , phys .
d * 63 * , 114020 ( 2001 ) .
[ arxiv : hep - ph/0011336 ] c. w. bauer and i. w. stewart , phys .
b * 516 * , 134 ( 2001 ) .
[ arxiv : hep - ph/0107001 ] c. w. bauer , d. pirjol , and i. w. stewart , phys . rev .
d * 65 * , 054022 ( 2002 ) .
[ arxiv : hep - ph/0109045 ] j. chay and c. kim , phys . rev .
d * 65 * , 114016 ( 2002 ) .
[ arxiv : hep - ph/0201197 ] a. v. manohar , t. mehen , d. pirjol , and i. w. stewart , phys .
b * 539 * , 59 ( 2002 ) .
[ arxiv : hep - ph/0204229 ] w. e. caswell and g. p. lepage , phys .
b * 167 * , 437 ( 1986 ) .
g. t. bodwin , e. braaten , and g. p. lepage , phys .
d * 51 * , 1125 ( 1995 ) ; [ erratum - ibid .
d * 55 * , 5853 ( 1997 ) ] .
[ arxiv : hep - ph/9407339 ] m. e. luke , a. v. manohar , and i. z. rothstein , phys .
d * 61 * , 074025 ( 2000 ) .
[ arxive : hep - ph/9910209 ] m. gourdin , a. n. kamal , and x. y. pham , phys . rev .
lett . *
73 * , 3355 ( 1994 ) .
a. n. kamal and a. b. santra , phys .
d * 51 * , 1415 ( 1995 ) .
[ arxiv : hep - ph/9409364 ] p. colangelo , c. a. dominguez , and n. paver , phys .
b * 352 * , 134 ( 1995 ) .
[ arxiv : hep - ph/9502374 ] m. gourdin , y. y. keum , and x. y. pham , phys .
d * 52 * , 1597 ( 1995 ) .
[ arxiv : hep - ph/9501257 ] c. h. chou , h. h. shih , s. c. lee , and h. n. li , phys .
d * 65 * , 074030 ( 2002 ) .
[ arxiv : hep - ph/0112145 ] m. beneke , g. buchalla , m. neubert , and c. t. sachrajda , nucl .
b * 591 * , 313 ( 2000 ) .
[ arxiv : hep - ph/0006124 ] h. y. cheng , y. y. keum , and k. c. yang , phys .
d * 65 * , 094023 ( 2002 ) .
[ arxiv : hep - ph/0111094 ] z. z. song and k. t. chao , phys .
b * 568 * , 127 ( 2003 ) .
[ arxiv : hep - ph/0206253 ] j. d. bjorken , nucl .
b , proc .
suppl . * 11 * , 325 ( 1989 ) .
m. beneke , f. maltoni , and i. .z .
rothstein , phys .
d * 59 * , 054003 ( 1999 ) .
[ arxiv : hep - ph/9808360 ] g. t. bodwin , e. braaten , t. c. yuan , and g. p. lepage , phys .
d * 46 * , r3703 ( 1992 ) .
[ arxiv : hep - ph/9208254 ] p. l. cho and m. b. wise , phys .
b * 346 * , 129 ( 1995 ) .
[ arxiv : hep - ph/9411303 ] a. a. affolder _ et al .
_ [ cdf collaboration ] , phys . rev .
lett . * 85 * , 2886 ( 2000 ) .
[ arxiv : hep - ex/0004027 ] m. gorbahn and u. haisch , nucl .
b * 713 * , 291 ( 2005 ) .
[ arxiv : hep - ph/0411071 ] particle data group , http://pdg.lbl.gov/. | we consider an effective field theory for the nonleptonic decay in which a heavy quark decays into a pair of a heavy quark and antiquark having a small relative velocity and one relativistic ( massless ) quark .
this effective theory is a combination of hqet , scet , and a covariant modification of nrqcd . in the leading logarithm approximation
the effective theory decay amplitude factorizes into the product of matrix elements of heavy - to - heavy and heavy - to - light currents .
we discuss a possibility of factorization beyond the leading logarithm approximation and find it doubtful .
the wilson coefficients of the effective theory electro - weak ( ewet ) lagrangian in the next - to - the leading logarithm approximation are calculated at the matching scale of the decay .
the differential decay rate for the inclusive decay @xmath0 in the effective theory framework is evaluated . |
catastrophic changes in the coronal magnetic field topology are believed to trigger energetic transient events , such as , solar flares and cmes .
the mhd catastrophe leads to the reconnection of magnetic field lines in the corona that results in a wealth of observed post - flare phenomena .
as flares derive their energy from stressed magnetic field , it is expected that the magnetic stress or non - potentiality would relax toward a lower energy state after the release of excess energy available in the active region .
it was first suggested several decades ago by that flare energy release should be associated with observable magnetic field changes .
extensive efforts have been made since then to detect such changes using photospheric magnetic field measured by ground - based instruments @xcite , and more recently , by space - borne instruments @xcite .
a wide variety of results have been reported on the pre- and post - flare changes in magnetic field parameters .
however , these remained mostly unreliable because of poor sensitivity , spatial resolution , cadence and coverage of the available magnetographs .
furthermore , serious questions have also been raised about the nature of observed magnetic field changes as these measurements are expected to be affected by flare - induced line profile changes @xcite .
nevertheless , with the availability of recent high - quality , high - cadence magnetic field observations , there is a mounting evidence of rapid , permanent changes in longitudinal and transverse magnetic fields during the course of large flares @xcite .
there are some other physical processes accompanying the sudden energy release occurring in the solar corona that may influence the magnetic field measurements . a large number of charged particles , _
i.e. _ , electrons and protons , accelerated during the flare energy release in the corona can precipitate along magnetic field - lines and lose their energy in the lower atmosphere . during this process
, microwave radiation is produced by gyrosynchrotron process and hard x - ray ( hxr ) emission through a thick - target bremsstrahlung process @xcite .
white - light flares ( wlfs ) , the most energetic of all flares , were suggested to be associated with direct heating by nonthermal or accelerated particles , specifically quasi - relativistic electrons @xcite .
this is observationally supported by the hxr footpoints , _
i.e. _ , sites of nonthermal particle acceleration , matching well with wlf ribbons @xcite .
however , only very high energy electrons having energy exceeding a few mev are expected to penetrate to deeper photospheric layers due to the increasing density .
only these high energy electrons can contribute to direct heating that may not be adequate to produce wlf emission .
it was inferred from the trace wl and _ yohkoh_/hxt data that wlf ribbons originate in the chromosphere and the temperature minimum region , and that the enhanced wl emission is caused by ionization and subsequent recombination of hydrogen .
energy thus deposited in the chromosphere by the electron beam is then transported to the lower atmosphere by a back - warming process . observed changes in magnetic field parameters during the impulsive phase of a flare would be affected by a variety of these effects .
localized sign reversal of magnetic polarity , termed as `` magnetic anomaly '' , has been observed during the impulsive phase of some flares @xcite .
such a sign reversal has also been attributed to the change in the spectral line profile , from absorption to emission , by numerical models @xcite . however , non - lte calculations for the spectral line ni i 6768 , used for magnetic field measurements in gong and soho / mdi instruments , have shown that this absorption line can turn into emission only by a large increase of electron density and not by heating of the atmosphere by any other means .
consistent with this , discovered sign reversals in locations of strong hxr emission formed near the cooler sunspot umbral / penumbral areas . injection and transport of high - energy electrons was observationally inferred from a m9.8 flare observed in microwave by owens valley solar array , and in hxr by hard x - ray telescope on board _ yohkoh _ @xcite .
_ simulated the gong and mdi observations for solar flares and concluded that as compared to doppler velocity measurements the magnetic field measurements are less sensitive to the line shape changes .
in addition , their numerical simulation also showed that the observed transient sign reversal may be produced when the ni i 6768 line profile turns into emission as a result of non - thermal beam impact on the atmosphere in regions of strong magnetic fields .
the effect of line profile changes on magnetic field estimates has been reported also by .
more recently , flare - associated helioseismic effects have been reported using different local helioseismology techniques , _
e.g. _ , detection of acoustic sources by holography @xcite , flare - induced seismic waves using time - distance analysis @xcite and post - flare amplitude enhancement of acoustic or p - modes by ring - diagram technique @xcite . have found seismic sources to be located near the acoustic sources discovered earlier by during the large flare in noaa 10486 .
several other flare - productive regions have also been extensively studied .
while examining the gong magnetogram movie of the x17/4b flare of 28 october 2003 in noaa 10486 , rapidly `` moving '' transient magnetic features ( mfs ) were detected during the impulsive phase @xcite .
these interesting transient features moved away from the flare site with separation velocities ranging from 30 to 50 km s@xmath0 comparable to the velocity of seismic waves reported by .
these mfs do not appear to be instrumental artifacts as gong and mdi movies both exhibited these features .
the exact nature of the mfs is , however , not well understood and further investigations are required .
if the mfs were indeed associated with the white - light flare by way of the absorption line profile turning into emission , they are expected to be correlated with wlf kernels . in particular
, it would also be interesting to examine whether these features were essentially `` moving '' sign - reversal anomalies associated with hxr sources produced by nonthermal electron beams @xcite .
it is to note that we also detected `` doppler velocity transients '' during the impulsive phase of this flare from gong and mdi dopplergram movies .
these are similar to the localized doppler velocity enhancements , co - spatial with h@xmath1 flare kernels , earlier reported by . in this paper , we discuss spatial and temporal correlations of the `` moving '' magnetic and doppler velocity transient features observed during the energetic white - light flares of 28 and 29 october 2003 .
we examine the properties of these mfs in the light of recent studies related to flare - associated changes in magnetic and doppler velocity fields .
we also attempt to find their correlation with flare kernels as observed in the lower atmosphere , _
i.e. _ , photosphere and chromosphere and the corresponding features in upper layers using the available multi - wavelength observations . in section [ s - data ] , we describe the observational data used in this study . the data analysis technique is given in section [ s - analysis ] .
properties of mfs and associated flare kernels are described in sections [ s - flare28][s - flare29 ] .
finally , discussions and conclusions are presented in section [ s - discussions ] .
we have used magnetograms , dopplergrams and white - light images obtained by gong and soho / mdi to study the wl flares of 28 and 29 october 2003 .
these observations provide information on the temporal and spatial evolution of photospheric magnetic field , doppler velocity field and intensity in noaa 10486 . to examine properties of the magnetic and doppler mfs and their correlation with the flare kernels , we have used white - light intensity images for the photospheric layer . for upper atmospheric layers ,
we have used trace uv 1600 for the temperature minimum region , chromospheric h@xmath1 observations obtained from uso and mlso , and trace uv 284 for the transition region . to compare the locations of flare kernels and mfs with hxr sources , we have used hxr data obtained by rhessi .
details of these observational data are described below : * * gong data : * magnetograms , dopplergrams and white - light images obtained from gong instrument have spatial and temporal resolutions of 2.5 arcsec / pixel and 1 min , respectively @xcite .
the instrument uses spectral line centered at ni i 6768 and works on the principle of phase shift interferometry ( harvey 2008 , private communication ) . *
* soho / mdi data : * in addition to the gong data , we have also used the corresponding soho / mdi data @xcite to inter - compare the observed changes in magnetic , doppler velocity and intensity maps .
the soho / mdi data have spatial resolution of 2.0 arcsec / pixel and temporal cadence of 1 min . *
* uso h@xmath1 : * we have used high spatial and temporal resolution h@xmath1 filtergrams obtained from uso , udaipur ( india ) for studying the chromospheric evolution of the x17/4b flare of 28 october 2003 .
these filtergrams were taken by a 15-cm aperture f/15 spar telescope at a cadence of 30 s during the quiet phase and 3 s in flare mode .
the spatial resolution of uso filtergrams is 0.4 arcsec / pixel at the ccd detector plane . * * mlso h@xmath1 : * mlso h@xmath1 filtergrams have been used for the x10/2b flare of 29 october 2003/20:49 ut as this event occurred during the night - time of the uso , udaipur site .
this data has spatial and temporal resolutions of 2.9 arcsec / pixel and 3 min , respectively .
* * trace uv : * the data corresponding to the temperature minimum and transition regions were obtained from trace @xcite which correspond to 1600 and 284 data sets , respectively .
these images have spatial and temporal resolutions of the order of 0.5 arcsec / pixel and 35 s , respectively .
* * rhessi hxr : * we have used x - ray data obtained from rhessi @xcite which observes solar hxr and @xmath2-rays from 3 kev to 17 mev with spatial resolution as high as @xmath3 2.3 arcsec with a full - sun field of view .
the raw format data ( _ i.e. _ , modulated fourier components ) for both events were obtained from the rhessi data site . hxr maps for the events
were then constructed using pixon method @xcite under solar software ( ssw ) for the dimension of @xmath4 pixels centered at the flaring regions .
the spatial and temporal resolutions of the constructed images are 4 arcsec and 1 min , respectively .
we chose different energy bands for the two flares due to their differing energetics : higher energy ( 100 - 200 kev ) for the 28 october 2003 event and the lower energy ( 50 - 100 kev ) for the 29 october 2003 event .
we have used time - series of full - disk gong and mdi magnetograms and dopplergrams for identifying various transient features during the course of evolution of the energetic flares of 28 and 29 october 2003 in noaa 10486 .
full disk images were aligned by de - rotating with the solar rotation .
we ignored the effect of solar differential rotation while correcting the images , since ( i ) the active region was located close to the solar equator , ( ii ) selected areas - of - interest were much smaller than the disk , and ( iii ) durations of the peak phase of flare - events were relatively short , _
i.e. _ , 10 - 15 min .
we applied a cross - correlation method based on the fourier technique to obtain co - aligned magnetograms registered within a sub - pixel accuracy .
all the images were aligned using this procedure .
we overlaid contours of magnetic flux on various observational images for identifying and selecting corresponding areas - of - interest by visual inspection of various structures , such as , filaments and sunspots .
figures [ fig : immos28 ] and [ fig : immos29 ] show the selected areas of noaa 10486 from trace uv 1600 , uso h@xmath1 , trace uv 284 , mdi magnetogram / dopplergram , and gong white - light images for the flares of 28 and 29 october 2003 , respectively . and
trace uv 284 ( top row , from left to right columns ) , and the mdi magnetogram , dopplergram and gong white - light image ( bottom row , from left to right columns ) obtained around the peak phase of the white - light x17/4b flare of 28 october 2003 .
the arrow and the box indicate locations of observed `` moving '' magnetic transients.,scaledwidth=100.0% ]
the two - ribbon x17/4b flare of 28 october 2003 was one of the most energetic events that occurred in noaa 10486 .
the active region was located close to the solar disk center at the time of this flare ; therefore , projection effects did not pose any serious difficulty in the analysis . according to goes x - ray flux data ,
this flare started at 09:51 ut and ended at 11:24 ut , with the maximum phase at 11:10 ut . however , as seen in h@xmath1 , the flare lasted over a considerably longer period .
significant changes were observed in photospheric magnetic and doppler velocity fields as seen from the movies of gong and mdi observations for 28 october 2003 .
these included gradual , evolutionary changes in the active region during the course of the day .
in addition , there were also abrupt and persistent changes that occurred at some localized sites in the neighborhood of flare kernels observed in white - light @xcite .
figure [ fig : immos28 ] shows multi - wavelength images corresponding to different layers of solar atmosphere obtained during the peak phase of this flare .
the top row , left to right columns , shows the flare kernels observed in the temperature minimum region , the chromosphere and the transition region .
it may be noted that the flare kernels displayed structural similarity in these layers .
images corresponding to the photospheric level are shown in the bottom row , left to right columns : mdi magnetogram , mdi dopplergram and gong white - light .
the white - light image clearly shows emission kernels of this energetic wlf . from the movie of gong magnetograms obtained during this super - flare
, interesting `` magnetic '' transient features were observed moving rapidly away from the flare site .
figure [ fig : immos28 ] shows these features as seen at around 11:06 ut .
one of these features was observed to be moving toward the leading sunspot ( marked by a rectangular box ) , while , the other moved perpendicular to the magnetic neutral line ( marked by an arrow ) .
these features appeared and faded in a few minutes period during the impulsive phase of the flare ( , @xcite ; ) .
both gong and mdi images exhibited these moving features .
similar , albeit fainter , doppler velocity transients associated with the wlf event were also detected in the dopplergrams . .
ab marks the direction of motion of wlf kernels and mfs .
distances are estimated from a reference point e ( left panel).,scaledwidth=90.0% ] .
solid and dotted profiles represent magnetic flux at 10:59:03 ut and 11:07:03 ut , respectively .
polarity reversals at points g and h are evident at 11:07:03 ut , _
i.e. _ , around the peak phase of the flare .
else where along the two lines , magnetic flux remained nearly unchanged .
, scaledwidth=90.0% ] corresponding to the observed moving transients , we noticed a sign reversal of magnetic polarity ( also , doppler velocity ) at some locations during the impulsive phase of the flare . to precisely identify locations of sign reversal , we plotted magnetic flux profiles along a horizontal raster moving from the bottom to the top of selected magnetograms during the pre- and peak - phases of the flare .
we found two locations g and h around which magnetic polarity sign reversals occurred ( figure [ fig : imbd28 ] ) .
we have drawn lines rs and pq passing through these two points over the magnetograms at 10:59:03 ut ( solid lines ) and 11:07:03 ut ( dotted lines ) , respectively .
the corresponding magnetic flux profiles along these lines are shown in figure [ fig : fluxbd28 ] .
it is evident that magnetic polarity reversals occurred at g and h as the moving `` magnetic '' features passed over these locations at 11:07:03 ut .
else where , magnetic flux matched fairly well along these lines at the two time instants .
such sign reversals were already reported several years ago by for some large flares .
this was attributed to the heating of lower atmosphere by the flare such that the core of the absorption line , used for measurement , turned to emission @xcite .
more recently , this effect was reported for the large x - class flare of 6 april 2001 using mdi magnetograms @xcite . however , sign reversals observed by mdi instrument are difficult to be explained by the sudden heating of the lower atmosphere .
this is because ni i 6768 line ( used by mdi and gong instruments ) is formed in the temperature minimum region and is rather stable against temperature changes @xcite .
interestingly , noted that sign - reversals occurred near locations of strong magnetic field , cooler sunspot umbrae that are exactly co - aligned with hxr sources .
this suggested that the sign reversal anomaly was associated with non - thermal electron beam precipitating to the lower atmosphere near cool sunspots .
this inference was further supported by a numerical simulation of mdi measurements incorporating the flare effects .
the transient sign - reversals observed during the impulsive phase of the x17/4b flare in noaa 10486 appear to be similar in nature as reported by .
however , in this case we found two locations that somewhat differed in their basic nature
( i ) the sign reversal at h was located in strong magnetic field region of the leading , negative polarity umbra ; similar to that reported by , and remained nearly stationary during the entire impulsive phase , ( ii ) the sign reversal at g first appeared near the neutral line , _ i.e. _ , a weak field location , and subsequently moved rapidly toward stronger magnetic field region of the following , positive polarity sunspots . in the following section ,
we examine these moving features and their association with the flare kernels observed at various solar atmospheric layers .
+ + + + + + on a visual inspection of magnetograms / dopplergrams and filtergrams of the active region taken at various wavelengths , the magnetic transients appeared to be spatially correlated with flare kernels . for establishing this association
further , we enhanced the transients using maximum entropy method ( mem ) @xcite .
it is a deconvolution algorithm which operates by minimizing a smoothness function ( `` entropy '' ) in an image . before applying this tool for each type of observation
, we subtracted the square root of a reference image from the square root of corresponding images obtained at different instants of time .
figure [ fig : contours28 ] ( a e ) shows contours of suitably enhanced features observed during the impulsive phase of the flare in different sets of images , _
i.e. _ , mdi magnetograms and dopplergrams , gong white - light , trace uv 1600 , uso h@xmath1 and trace uv 284 .
these contours display the spatial and temporal evolution of magnetic / doppler transients and flare kernels observed at various atmospheric layers with increasing height from the photosphere to the transition region .
the solid line in the figure represents an arbitrary reference line , rs , drawn along the neutral line from where distances of various features were measured ( _ cf .
_ , figure [ fig : imbd28 ] ) .
it is evident that magnetic / doppler transients and flare kernels separated away from the neutral line with time during the impulsive phase of the flare ( rows a
e , in each column ) .
this follows the usual behavior of two - ribbon flares .
we also note that the evolution of magnetic transients ( column 1 ) is spatially and temporally well correlated with the flare kernels observed in different layers ( columns 36 ) .
locations of observed transients are marked by arrows and boxes in each frame . along with magnetic transients ,
comparatively fainter doppler transients are also discernible during this period ( column 2 ) .
these are similar to the doppler - ribbons reported by .
magnetic and doppler transients are observed to be particularly well correlated with wlf kernels observed during the impulsive phase of the flare .
, , trace uv 1600 , magnetogram , dopplergram and white - light ( from top to bottom rows ) .
measured distances are marked by `` @xmath5 '' , while the solid profiles represent the boltzmann sigmoid fitting through these points .
right panel : separation velocities derived for various features from the fitted distances corresponding to the left panel .
the dash - dotted line in rhessi velocity panel shows the hxr light curve.,scaledwidth=98.0% ] to quantitatively examine motion of the observed features , we carried out the following procedure .
positions of transients and flare kernels were measured from a reference point e ( figure [ fig : imbd28 ] , left panel ) using an automatic algorithm implemented in idl ( interactive data language ) .
the maximum value was followed along ea , _
i.e. _ , the direction of motion of features in the enhanced images obtained from mem .
distance between the point of maximum value within a given feature ( say e@xmath6 : @xmath7 ) and the reference point e ( @xmath8 ) was calculated in pixels using @xmath9^{1/2}$ ] .
it was then converted into kilometers using the pixel resolution of the corresponding instrument ( table [ t - distvel28 ] ) .
it is evident from figure [ fig : distvel28 ] that magnetic / doppler transients and flare kernels separated away from the neutral line during the impulsive phase of the x17/4b flare , _
i.e. _ , 11:00 - 11:15 ut .
their distances from the reference point e , along ab , increased rapidly during the initial phase , _
i.e. _ , 11:00 - 11:10 ut , as seen in each row ( corresponding to different observations ) in the left column of the figure .
thereafter , the rate of increase of separation slowed down .
table [ t - distvel28 ] shows an interesting trend in the maximum separation attained for the flare kernels .
it was shortest , _
i.e. _ , @xmath10 mm at the transition layer observed in uv 284 ( trace ) , and moving toward the lower layers , it increased to @xmath11 mm at the chromosphere ( uso h@xmath1 ) , @xmath12 mm at the temperature minimum region and the photosphere ( _ cf .
_ , trace uv 1600 , and gong and mdi photospheric data ) .
this trend is consistent with the classical flare model of an expanding loop structure with footpoints anchored in the lower atmosphere .
the hxr feature also showed a separation of @xmath13 mm , similar to that observed in the chromospheric h@xmath1 .
this is consistent with the formation of hxr footpoint sources in the chromosphere or in the lower corona depending on the energy of penetrating particles .
the magnetic and doppler transients , on the other hand , showed a larger separation of @xmath14 mm as obtained for the photospheric wlf kernels .
hxr light curve is shown as the dash - dotted line in hxr velocity panel of figure [ fig : distvel28 ] .
it is evident that the separation velocity of hxr footpoint @xmath15 ( _ cf .
_ , figure [ fig : hessi28 ] ) and the overall hxr flux decreased with time during the post - flare phase , except that there is a small bump in the hxr flux at around 11:12 ut . on examining the hxr images we found that the bump corresponds to an enhancement of hxr flux near the footpoint @xmath16 .
it may be noted that the major contribution to the hxr light curve arises from the integrated flux over both the footpoints @xmath16 and @xmath15 while the separation velocity plotted here pertains only to the footpoint @xmath15 .
we derived the relative velocities of various features moving along the direction ab , away from the reference line rs , by fitting a boltzmann sigmoid ( best fitting ) to the observed positions .
this is given by @xmath17 where , @xmath18 and @xmath19 are fitting parameters ; @xmath20 - top ( _ i.e. _ , maximum of @xmath21 ) , @xmath22 - bottom ( _ i.e. _ , minimum of @xmath21 ) , @xmath23 - the time at which distance is halfway between bottom and top , and @xmath19 - the steepness of the curve , with a larger value denoting a shallower curve .
separation velocity @xmath24 is then obtained by taking time derivative of the fitted distance function @xmath21 .
numerical differentiation was carried out by three point lagrangian interpolation .
figure [ fig : distvel28 ] shows the fitted distances ( solid curves ) passing through the measured distances marked by `` @xmath5 '' symbols ( left panel ) , and the derived velocities ( right panel ) .
velocities of flare kernels and magnetic / doppler transients reached a peak ( or maximum ) at around 11:06 ut .
it is also clear from this figure and table [ t - distvel28 ] that the separation velocity of flare kernels was larger , _
i.e. _ , 45 - 50 km s@xmath0 at lower atmosphere ( temperature minimum region and photosphere ) and it decreased with height , _
i.e. _ , 38 km s@xmath0 at the chromosphere and 29 km s@xmath0 at the transition region
. it is known that a systematic increase in the separation of h@xmath1 flare kernels is the chromospheric signature of magnetic reconnection process occurring progressively at higher levels in the corona .
rather small separation velocities in the range of @xmath3 3 - 10 km s@xmath0 have been reported for two - ribbon flares associated with filament eruptions as compared to the velocities in the range of 30 - 50 km s@xmath0 obtained for the flares in noaa 10486 ( table [ t - distvel28 ] ) .
however , large separation velocities in the range 20 - 100 km s@xmath0 have also been reported during some recent two - ribbon flares occurring in complex active regions @xcite . as the flare - ribbons approach regions of strong magnetic fields , their speeds are known to decrease .
this is also observed for the flare kernels and moving magnetic transients during the flare of 28 october 2003 .
figure [ fig : flux_ba ] shows the separation velocity profile ( dark curve ) of the moving magnetic feature from the neutral line along the direction ba .
it is clear that the separation velocity first increased with magnetic flux ( white curve ) , attained a peak value , then decreased and vanished at the location where magnetic flux attained a maximum value .
this is consistent with the general behavior of flare - ribbons moving toward the region of strong magnetic fields .
800 , @xmath25 1500 gauss levels , while the dotted contours mark the magnetic neutral lines .
rhessi hxr fluxes ( white contours ) in the energy range
100200 kev are overlaid at the levels 20 , 40 , 60 , 80@xmath26 of the maximum value .
red patches mark the enhanced wlf kernels ( top rows 1 - 2 ) and magenta patches represent the enhanced magnetic transients ( bottom rows 3 - 4).,title="fig:",scaledwidth=92.0% ] 800 , @xmath25 1500 gauss levels , while the dotted contours mark the magnetic neutral lines .
rhessi hxr fluxes ( white contours ) in the energy range 100200 kev are overlaid at the levels 20 , 40 , 60 , 80@xmath26 of the maximum value .
red patches mark the enhanced wlf kernels ( top rows 1 - 2 ) and magenta patches represent the enhanced magnetic transients ( bottom rows 3 - 4).,title="fig:",scaledwidth=92.0% ] figure [ fig : hessi28 ] shows temporal and spatial evolution of enhanced wlf kernels ( red patches ) observed in gong intensity maps ( green half - tone images ) obtained during the impulsive phase of the x17/4b flare of 28 october 2003 ( top rows 1 - 2 ) . similarly ,
evolution of moving magnetic transients ( magenta patches ) is shown over the same background images in the bottom rows ( 3 - 4 ) .
white contours represent rhessi hxr flux in the energy range 100 - 200 kev associated with the flare during its impulsive phase , _
i.e. _ , 11:07:03 - 11:12:03 ut , at 20 , 40 , 60 , 80@xmath26 levels of the maximum hxr flux . the central maximum of the hxr source ( marked by `` + '' sign in yellow color ) was found to be moving in the direction of arrow with separation velocity of 43 km s@xmath0 from the reference line , _ i.e. _ , the solid blue line drawn along the neutral line .
separation velocity of the hxr feature was similar to that of the moving magnetic / doppler transient and wlf kernels ( _ cf .
_ , table 1 ) , however , their spatial positions differed some what .
hxr features were earlier reported to form during the impulsive phase of an x5.6 flare of 6 april 2001 in strong magnetic field regions @xcite .
figure [ fig : contours28 ] shows that the moving magnetic transient , where polarity sign reversal occurred , first appeared around 11:03 ut close to the neutral line , _ i.e. _ , at a weak magnetic field location .
thereafter , it moved toward the direction of stronger magnetic field of the following ( positive ) polarity umbra .
unfortunately , we do not have the entire temporal sequence in rhessi hxr due to the non - availability of data before 11:07:16 ut .
therefore , it is not clear whether the hxr source had also formed near the neutral line in the initial phases
. noted that locations of sign - reversal `` anomaly '' coincided well with hxr footpoints .
they suggested the observed anomalous magnetic features to be related to electron precipitation into the lower atmosphere but not with all wlf kernels .
they gave the energetics for the reversal of line - profiles in the cooler , strong magnetic field umbral regions as compared to that in the hotter , weak magnetic field locations .
in agreement with , we observed that hxr footpoint @xmath16 formed in this case also at strong magnetic field location of the leading umbra .
this hxr source was nearly stationary and matched well with both the wlf kernel @xmath27 ( rows 1 - 2 ) and the magnetic transient feature ( rows 3 - 4 ) . on the other hand , hxr source @xmath15 did not match well with the wlf kernel .
we observed that the wlf kernel @xmath28 first appeared at one edge of @xmath15 contours ( 11:07:16 ut ) , then decayed and appeared at the opposite edge ( 11:11:16 ut ) .
around the same time , wlf kernel @xmath29 appeared which was not associated with any hxr feature .
the moving magnetic feature at @xmath15 appear to be better associated with hxr than wlf kernels . thus , association of hxr source with anomalous magnetic polarity feature , as reported by , does appear to hold in this case . and
trace uv 284 ( top row , from left to right columns ) and the mdi magnetogram , dopplergram and gong white - light image ( bottom row , from left to right columns ) obtained during the peak phase of the white - light x10/2b flare of 29 october 2003 .
arrow shows the location of the `` moving '' magnetic transient .
, scaledwidth=100.0% ]
this white - light flare , classified as x10/2b , occurred in noaa 10486 on 29 october 2003 when the active region was located close to the disk center at s15w02 .
the flare began at 20:37 ut , reached peak phase at 20:49 ut and decayed at 21:01 ut .
figure [ fig : immos29 ] shows a mosaic of flare images as observed in the trace uv 1600 , h@xmath1 and trace uv 284 ( top row ) .
the photospheric magnetogram , dopplergram and white - light images are shown in the bottom row . as detected during the white - light flare of 28 october 2003 , we found moving transient features around the impulsive phase of this energetic event also .
however , this flare occurred in another part of the active region away from the event of 28 october 2003 .
locations of magnetic transients and wlf kernels are marked by arrow in each frame of figure [ fig : immos29 ] .
the characteristics of mfs and flare kernels are described in the following . .
sign reversals in flux polarity were detected at `` b '' and `` c '' across the neutral line .
point `` d '' on the vertical line no represents the reference from which distances of flare kernels were measured along line dx.,scaledwidth=100.0% ] ) .
solid and dotted profiles represent magnetic flux obtained at 20:36:03ut and 20:42:03 ut , respectively .
polarity sign reversals are evident at the locations b and c marked in figure [ fig : imbdf29 ] during the peak phase of the white - light flare at 20:42:03 ut , whereas the flux remained nearly unchanged else where along xy and lm.,scaledwidth=90.0% ] + + + + + ( mlso ) , hxr ( rhessi ) , uv 1600 ( trace ) and white - light ( gong ) . measured distances are marked by symbols @xmath30 , @xmath31 , @xmath32 and @xmath33 for h@xmath1 , hxr , uv 1600 , and wl , respectively , while solid lines represent the straight - line fitting through the corresponding data points.,scaledwidth=60.0% ] a movie of mdi magnetograms was used to study the evolution of magnetic flux and detect any magnetic transient features during the impulsive phase of this flare also ( _ cf .
_ , figure [ fig : imbdf29 ] ) .
we identified locations b and c where polarity sign reversals occurred at 20:42:03 ut by examining magnetic flux along a horizontal raster moving from the bottom to the top of magnetograms selected during the pre - peak and peak phases of the flare .
magnetic flux along xy and lm is plotted in figure [ fig : mflux29 ] .
solid and dotted lines in the figure correspond to the times of pre - peak phase ( 20:36:03 ut ) and the peak phase ( 20:42:03 ut ) , respectively .
it is observed that magnetic flux changed sign around b and c , while it remained nearly unchanged else where along the lines xy and lm . a careful examination of mdi magnetogram and dopplergram movies of the 29 october 2003 event revealed moving magnetic and doppler features along with wlf kernels during the impulsive phase . spatial and temporal evolution of mfs and flare kernels observed during this event
are illustrated in figure [ fig : magmos29 ] .
contour levels were derived from the same technique ( mem ) as used for the previous event of 28 october 2003 .
different sets of images , _
i.e. _ , photospheric magnetograms ( mdi ) , dopplergrams ( mdi ) , white - light ( gong ) , chromospheric h@xmath1 ( mlso ) and temperature minimum uv 1600 ( trace ) are shown from the left to the right columns .
the mfs appeared to be correlated with flare kernels observed in white - light and other wavelengths corresponding to various layers of the solar atmosphere .
temporal evolution of mfs and flare kernels is shown from the top ( a ) to the bottom ( e ) rows in each column .
a solid line drawn in each frame marks the reference direction perpendicular to which mfs and flare kernels moved , and the arrow shows the location of the mf in each set of the observations .
distances of flare kernels from the reference line no along the line dx are determined by following the position of the maximum value within a feature of interest .
separations of various flare kernels are shown in figure [ fig : magmos29 ] . in order to find separation speeds of flare kernels , we fitted straight lines ( best fitting ) through the measured distances
. slopes of these fitted lines provided the respective velocities ( figure [ fig : distvel29 ] ) .
maximum distances and velocities are listed in table [ t - distvel29 ] .
velocities of flare kernels showed similar trend as in the earlier case of the flare of 28 october 2003 ; it generally increased from the upper to the lower atmospheric layers .
however , separation velocities obtained for this flare were found to be much smaller as compared to the x17/4b flare .
another difference was that the magnetic and doppler transients remained nearly stationary during this flare .
the profiles for distance and separation velocity were also different from that of the event of 28 october 2003 , where we could do a boltzmann sigmoid fitting as against the straight line fitting for this flare .
it is unlikely that the separation velocity of flare kernels would remain constant throughout the flare duration as appears to be the case in this event .
it is to note that flare kernels became fainter in intensity or got fragmented as they moved away from the reference line , and were generally difficult to identify and follow after 20:51 ut . for h@xmath1 kernel , however , there is an indication that the rate of change of separation decreased after this time .
800 , @xmath25 1500 gauss , and dotted contours mark the magnetic neutral lines .
rhessi hxr fluxes in the energy range 50 - 100 kev are overlaid at 20 , 40 , 60 , 80@xmath26 levels of the maximum value ( blue contours ) .
red patches ( top rows 1 - 2 ) show the positions of wlf kernels , while magenta patches ( bottom rows 3 - 4 ) mark the magnetic transients.,title="fig:",scaledwidth=92.0% ] 800 , @xmath25 1500 gauss , and dotted contours mark the magnetic neutral lines .
rhessi hxr fluxes in the energy range 50 - 100 kev are overlaid at 20 , 40 , 60 , 80@xmath26 levels of the maximum value ( blue contours ) .
red patches ( top rows 1 - 2 ) show the positions of wlf kernels , while magenta patches ( bottom rows 3 - 4 ) mark the magnetic transients.,title="fig:",scaledwidth=92.0% ] figure [ fig : hessi29 ] shows temporal and spatial evolution of the enhanced wlf kernels ( red patches ) observed in gong intensity maps ( green half - tone images ) during the impulsive phase of the x10/2b flare of 29 october 2003 ( top rows 1 - 2 ) .
similarly , evolution of magnetic transients ( magenta patches ) is shown over the same background images in the bottom rows ( 3 - 4 ) .
blue contours represent rhessi hxr flux in the energy range 50 - 100 kev associated with the flare during its impulsive phase , _
i.e. _ , 20:40:16 - 20:50:16 ut , at 20 , 40 , 60 , 80@xmath26 levels of the maximum flux .
the central maximum of the hxr source is marked by `` + '' sign in blue color .
a solid blue line is drawn along the neutral line as a reference to measure the positions of various flare kernels observed to be moving in the direction of the arrow .
corresponding to the white - light flare of 29 october 2003 , xu _ et al .
_ obtained 1.56 @xmath34 m near - infrared ( nir ) images of the region @xmath3 50 km below the photosphere at the opacity minimum layer using the dunn solar telescope at the national solar observatory / sacramento peak .
these images indicated extremely energetic activities associated with this flare as only the most energetic electrons can penetrate this deep in the photosphere .
locations of gong wlf kernels matched very well with the nir continuum flare patches .
however , not all these wlf kernels were associated with rhessi hxr sources , as seen in figure [ fig : hessi29 ] ( top rows 1 - 2 ) .
also , locations of mfs , where sign - reversal anomaly appeared , did not conform well with these hxr sources except at 20:40:16 ut ( first two frames in row 3 ) .
thereafter , the hxr source intensified at the opposite side of the neutral line , where the other part of the two - ribbon wlf kernel was located ( top rows 1 - 2 ) , but no magnetic transient feature ( magenta patches in rows 3 - 4 ) was present .
interestingly , in disagreement with , we note that the strong hxr source ( and the wlf kernel ) which intensified at the west side of the neutral line was located in a weak field region . on the other hand ,
the magnetic transient observed at stronger field positive polarity umbra was generally associated with wlf kernel but no hxr source .
this is in contrast with the white - light flare of 28 october 2003 , and also contradicts the suggested association of sign reversal with hxr sources .
however , it may be noted that found , for a white - light flare of 30 september 2002 , that motions of the continuum and hxr sources show similar trend .
we observed moving transients and sign reversals of magnetic polarity at some locations of noaa 10486 during the impulsive phases of both the white - light flares of 28 and 29 october 2003 .
comparatively fainter doppler transients were discernible along with magnetic transients during these flares .
the magnetic and doppler transients were observed to be well correlated .
sign reversals in magnetic polarity and doppler velocity also occurred at the locations of these moving transients .
the two flares were quiet different in their energetics - the 28 october flare was x17/4b , while the 29 october flare was x10/2b .
they occurred at topologically different locations within the active region .
we have attempted to examine properties and relationship of the observed mfs and flare kernels in various atmospheric layers for the two events .
there are some interesting questions related to these observed transients : what kind of physical process caused the moving magnetic and doppler velocity transients ?
is there any relationship between moving transients , hxr sources and flare kernels ? the gong and mdi instruments use the same spectral line , _ i.e. _ , ni i 6768 , but different principles of observations ( harvey 2008 , private communication ) .
nevertheless , the observed motions in magnetic , doppler velocity and intensity fields obtained from both instruments were similar in nature .
sign reversals in magnetic and doppler velocity fields were detected by both instruments during the impulsive phases of the flares of 28 and 29 october 2003 .
some of these transients occurred near the cooler umbral / penumbral sites in the active region noaa 10486 , but there were also several exceptions . in the case of the x17/4b flare of 28 october 2003
, we found that the two locations of magnetic transients were associated with sign reversals in magnetic polarity ( and doppler velocity ) .
these features differed in their basic characteristics as follows : ( i ) the location of sign reversal that occurred in strong magnetic field of the leading ( negative ) umbra was nearly stationary .
both hxr source and wlf kernel were associated with this feature .
( ii ) the other location of sign reversal observed in the weak field area near the neutral line moved rapidly toward the following ( positive ) polarity umbra , and was better associated with hxr as compared to the wlf kernels . on the other hand , magnetic and doppler transients observed on 29 october 2003 were nearly stationary during the course of the x10/2b flare .
the locations of magnetic transients , where sign - reversal anomaly appeared , did not conform well with the hxr sources .
interestingly , we observed a strong hxr source ( and wlf kernel ) in weak field region without corresponding sign reversal anomaly . on the other hand , there was a wlf kernel but no hxr source associated with the magnetic transient observed in strong field location of the positive polarity umbra .
this is in contrast with the hxr - magnetic anomaly relation observed for the white - light flare of 28 october 2003 , and also in contradiction with the earlier suggestions of about the association of hxr sources with polarity sign reversal phenomenon .
we found that magnetic / doppler transients generally followed the usual behavior of separation of flare kernels observed in two - ribbon flares .
let us first discuss the motion of magnetic transients observed during the flare of 28 october 2003 .
there was an interesting trend in maximum separation attained for the flare kernels observed in successively lower layers of solar atmosphere : the shortest , _
i.e. _ , @xmath10 mm at the transition layer as observed in uv 284 ( trace ) , increasing to @xmath11 mm at the chromosphere ( uso h@xmath1 ) and @xmath12 mm in the temperature minimum region and the photosphere ( uv 1600 trace , and gong & mdi photospheric data ) .
this trend is consistent with the classical flare model of an expanding loop structure with footpoints anchored in the lower atmosphere .
the hxr feature also showed separation of @xmath13 mm , _
i.e. _ , nearly the same as the separation observed in chromospheric h@xmath1 .
this is consistent with the formation of footpoint sources in the chromosphere or in the lower corona depending on the energy of particles .
magnetic and doppler transients , on the other hand , showed a larger separation of @xmath14 mm which matched the separation obtained for the photospheric wlf kernels .
furthermore , we found that flare kernels separated faster in the lower atmosphere ( the temperature minimum region and the photosphere ) with velocity of 45 - 50 km s@xmath0 as compared to that in the upper atmosphere , _
i.e. _ , 38 km s@xmath0 at the chromosphere , and 29 km s@xmath0 at the even higher transition region .
however , separation velocities for the x10/2b flare of 29 october 2003 were found to be much smaller as compared to the x17/4b flare of the previous day .
, using an atmospheric model , showed that spectral line inversion occurs near sunspot penumbral area if electron density rises significantly .
it is to note that a rise in the electron density was found during the impulsive phase of the flare of 28 october 2003 @xcite , also consistent with the rhessi hxr observations .
therefore , occurrence of moving transients appears to be related to line profile changes due to the process associated with electron - beams and not due to real changes in photospheric magnetic and doppler fields .
however , no clear conclusions are discernible in the case of the x10/2b flare of 29 october 2003 . also , away from these features ,
permanent changes in magnetic fluxes from the pre- to post - flare phases have been reported that do not seem to be affected by the wlf or hxr related effects @xcite . from a study of
the flare of 29 september 2002 , found that the continuum enhancement at hxr footpoints is not proportional to the electron beam flux , but it depends on the atmospheric conditions . using h@xmath1 line profiles they explained that the atmosphere was heated considerably during the impulsive phase , which increased the coronal pressure preventing nonthermal electrons from effectively penetrating into the chromosphere .
but during the preflare phase the heating is low that allows an electron beam to easily penetrate into the chromosphere and produce the continuum enhancement via radiative back - warming effect .
it is interesting to note that the observed flare kernels and moving transients associated with these events possessed large velocities .
this is expected due to the large magnetic energy stored in the magnetically complex active region which produced these exceptionally energetic super - flares .
large separation velocities of flare - ribbons and hxr footpoints are the signatures of rapid reconnection of successive magnetic field lines at higher levels of corona .
finally , answer to the question of association characteristics of the moving magnetic transients ( or magnetic anomaly ) still remains ambiguous as a clear and consistent correlation of these features with hxr sources ( or wlf kernels ) did not emerge from the study of the two white - light flares of noaa 10486 .
this work utilizes data obtained by the global oscillation network group ( gong ) project , managed by the national solar observatory which is operated by aura inc .
, under a cooperative agreement with the national science foundation .
the gong data were acquired by instruments operated by the big bear solar observatory , high altitude observatory , learmonth solar observatory , udaipur solar observatory , instituto de astrofisico de canarias and cerro tololo inter - american observatory .
this work utilises data from the solar oscillations investigation/ michelson doppler imager ( soi / mdi ) on the solar and heliospheric observatory ( soho ) .
soho is a project of international cooperation between esa and nasa .
mdi is supported by nasa grants nag5 - 8878 and nag5 - 10483 to stanford university .
we would like to express thanks for the h@xmath1 data to mauna loa solar observatory ( mlso ) .
mlso is operated by the high altitude observatory ( hao ) , a division of the national center for atmospheric research ( ncar ) and funded by the national science foundation ( nsf ) . | rapidly moving transient features have been detected in magnetic and doppler images of super - active region noaa 10486 during the x17/4b flare of 28 october 2003 and the x10/2b flare of 29 october 2003 .
both these flares were extremely energetic white - light events .
the transient features appeared during impulsive phases of the flares and moved with speeds ranging from 30 to 50 km s@xmath0 .
these features were located near the previously reported compact acoustic @xcite and seismic sources @xcite .
we examine the origin of these features and their relationship with various aspects of the flares , _ viz .
_ , hard x - ray emission sources and flare kernels observed at different layers - ( i ) photosphere ( white - light continuum ) , ( ii ) chromosphere ( h@xmath1 6563 ) , ( iii ) temperature minimum region ( uv 1600 ) , and ( iv ) transition region ( uv 284 ) . |
a mesoscopic corbino disk is a unique system to study the dynamics of self - organized vortex matter in small - size superconductors .
the interplay between the vortex - vortex and vortex - boundary interactions in mesoscopic superconductors leads to shape - induced giant vortex states @xcite , concentric _ shells _ of vortices and symmetry - induced vortex - antivortex `` molecules '' in mesoscopic squares and triangles @xcite . in the corbino geometry @xcite
an applied current creates a gradient in the current density and thus the lorentz force , i.e. , introducing a shear driving force between the rings of vortices .
this gives us the unique opportunity to study various dynamical effects related to vortex motion , e.g. , the transition from elastic to plastic motion , channeling @xcite , vortex friction @xcite , etc .
the dynamics of self - organized mortex matter in mesoscopic disks has many common features to , e.g. , atomic matter , charged particles in coulomb crystals , vortices in rotating bose - einstein condensates , magnetic colloids , synthetic nanocrystals , etc .
@xcite , or even large charged balls diffusing in macroscopic wigner rings @xcite and can provide us with a deeper understanding of , e.g. , the microscopic nature of friction , transport , magnetic , optical and mechanical properties of various physical and biological systems . in a corbino disk ,
the applied current is injected at the center and removed at the perimeter ( see fig . 1 ) to induce a radial current density @xmath4 that decays as @xmath5 along the radius @xcite . as a result , vortices near the center of the disk experience a stronger lorentz force @xmath6 than those near the disk s edge . for small @xmath4 ,
the local shear stress is small and the whole vortex pattern moves as a rigid body .
larger @xmath4 result in a strong spatially inhomogeneous stress that breaks up the vortex solid and concentric annular regions move with different angular velocities .
the voltage profiles measured in experiments @xcite reflect different dynamical phases ( elastic motion , shear - induced plastic slip ) of vortex motion .
the onset of plasticity in large corbino disks was theoretically analyzed within a continuous model in @xcite . within molecular dynamics
( md ) simulations of interacting vortices at @xmath7 , the nucleation and motion of dislocations in the vortex lattice was studied in @xcite .
recently , using the bitter decoration technique , the first direct observation of rings of vortices for mesoscopic superconducting nb disks was reported @xcite . for vorticities @xmath8 to 40 , the circular symmetry led to the formation of concentric _ shells _ of vortices , similar to electron shells in atoms or in nano - clusters @xcite .
the analysis of shell filling revealed `` magic - number '' configurations ( mnc ) @xcite corresponding to a commensurability between the shells which occurs when the numbers of vortices of each shell have a common divider @xcite . here , we study the dynamics of _ vortex shells _ in _ mesoscopic _ corbino disks .
our system has the added flexibility that we have several experimentally accessible tuning parameters as , e.g. , the driving shear force , the separation between the vortex rings ( through the external magnetic field ) and the commensurability between the vortex rings ( through the relative number of vortices in each ring ) . in mesoscopic corbino disks , we reveal a non - monotonous dependence of the critical current @xmath9 separating two dynamical regimes , a `` rigid body '' or separate shell rotation , on the magnetic field @xmath10 , and the appearance of dynamical instabilities associated with a jump in @xmath11 .
we show that this unusual behaviour is related to a `` structural transition '' , i.e. , an inter - shell vortex transition . for _ non - zero temperature _
, thermally - activated externally - driven flux motion is investigated , and we explain the observed two - step melting transition in corbino disks @xcite .
we place a corbino disk which has thickness @xmath12 and radius @xmath13 in a perpendicular external magnetic field @xmath14 . the corbino setup is shown in fig .
1 . an external current flows radially from the center to the edge of the disk and results in the inhomogeneous sheath current density @xmath15 , which makes vortices closer to the center feel a stronger force compared to the ones near the edge .
the lorentz force ( per unit length ) acting on vortex @xmath16 , @xmath17 , resulting from the external current is : @xmath18 where @xmath19 is the vortex position , @xmath20 , and @xmath21 is the unit vector along the magnetic field direction which is taken perpendicular to the disk . here
@xmath22 is the unit of force , @xmath23 , and @xmath24 .
( shown by dark blue / dark grey arrows ) that decays as @xmath5 along the radius . as a result , vortices near the center of the disk experience a stronger lorentz force @xmath6 ( shown by yellow / light grey arrows ) than those near the disk s edge .
the vortices are shown by red - to - yellow / grey - to - light grey tubes and by red / grey spots on the surface .
the direction of the external applied magnetic field @xmath25 , which is perpendicular to the surface of the disk , is shown by dark yellow / grey arrow .
, width=302 ] in a thin superconducting disk such that @xmath26 , the vortex - vortex interaction force @xmath27 and the force of the vortex interaction with the shielding currents and with the edge @xmath28 can be modelled respectively by @xcite @xmath29 where @xmath30 , @xmath31 is the position of the @xmath16th vortex , and @xmath32 is the number of vortices , or the vorticity .
our numerical approach is based on the langevin dynamics algorithm , where the time intergartion of the equations of motion is performed in the presence of a random thermal force .
the overdamped equations of motion becomes : @xmath33 here @xmath34 is the driving force ( eq .
( [ eqfj ] ) ) , @xmath35 is the force due to vortex - pin interactions @xcite , and @xmath36 is the thermal stochastic force , obeying the fluctuation dissipation theorem @xmath37 where greek and italic indices refer to vector components and vortex labels , @xmath38 is the viscosity .
the ground state of the system is obtained by simulating field - cooled experiments @xcite .
( orange / light grey solid circles ) and @xmath39 ( blue / dark grey open squares ) , measured in units of @xmath40 , of vortices in the first and second shells , respectively , as a function of applied current @xmath0 , measured in units of @xmath41 , for @xmath42 and different magnetic fields @xmath10 : ( a ) @xmath43 , a rigid - body rotation for @xmath44 ; the shells rotate separately for @xmath45 ; ( b ) @xmath46 , the critical current decreases , @xmath47 ; ( c ) @xmath48 , the @xmath9 further decreases , @xmath49 , and the motion becomes unstable ; ( d ) @xmath50 , _ bistable _ motion : a _ second _ critical current @xmath51 appears ; ( e ) @xmath52 , the motion stabilizes , and first critical current @xmath9 disappears , and new critical current @xmath53 .
, width=302 ]
first , we consider the smallest mesoscopic corbino disk which shows the main physics , and it has @xmath54 vortices which form the mnc ( 1,6,12 ) , as shown in the left - hand inset of fig . 3 . at weak
applied magnetic field @xmath43 all the vortices are packed in an almost perfect triangular abrikosov lattice .
( here we assume a perfect disk with no pinning , and temperature is set to zero after the `` annealing '' process . )
we apply an increasing external current @xmath0 , and we study the average angular velocity @xmath2 of vortices in each shell . for small @xmath0 ,
the shear produced by the gradient of the lorentz force is insufficient to break the vortex lattice .
it produces only _ elastic _ deformations in the lattice which rotates as a rigid body with angular velocity @xmath55 where @xmath56 is the average square radius for the vortices in the disk . with increasing @xmath0 ,
the lorentz force gradient reaches a critical value at which the shells start to slide with respect to each other and rotate with different velocities @xmath57 .
we call this the `` critical current '' @xmath9 , and for the case shown in fig
. 2(a ) we found @xmath58 .
now @xmath59 , and the inner ( outer ) shell rotates faster ( slower ) than the rigid body . for higher applied magnetic field ,
e.g. , @xmath46 , the critical current decreases , @xmath47 ( fig .
this is related to a deformation of the shells , which approach a circular shape , and therefore can more easily slide . at the same time
the shells move closer to each other increasing the dynamical friction between them . at @xmath48 ,
the @xmath9 further decreases down to @xmath49 , and the motion becomes unstable : for @xmath60 , the inner shell rotates with angular velocity @xmath61 , although at some values of @xmath0 in the region @xmath62 , @xmath39 _ drops _ down to @xmath63 ( fig .
the appearance of this irregular motion , or `` stochastization '' , is related to the strong dynamical friction between the shells , which can even `` lock in '' the shells at some values of @xmath0 .
a further increase of the magnetic field , @xmath50 ( fig .
2(d ) ) , results in _
bistable _ motion of the shells : a _ second _ critical current @xmath51 appears , and for currents @xmath64 the system _ either _ rotates as a rigid body _ or _ the shells rotate separately .
finally , at even higher field , e.g. , @xmath52 , the upper branch ( corresponding to higher angular velocity ) of the function @xmath65 for the inner shell ( shown by orange / light grey dotted line in fig .
2(d ) ) disappears .
therefore , the first critical current @xmath9 also disappears , and the motion stabilizes with the only critical current @xmath66 ( fig .
2(e ) ) . the non - monotonous change of the critical current @xmath9 with increasing applied magnetic field @xmath10 is summarized in fig .
, @xmath9 decreases , then instabilities develop in the system which in general are indicative of , and precede , a phase transition .
the transition to a second critical field occurs through a bistable state characterized by _ two _ critical currents in the system .
this unusual behaviour can be understood by analyzing the critical current together with the structure of the vortex shells in the region of this `` phase transition '' .
the vortex patterns for magnetic field before ( low fields ) and after ( high fields ) the transition are presented in the left - hand side and in the right - hand side insets of fig .
the jump in the critical current is correlated to a `` structural transition '' in the system where the distribution of vortices over the shells is altered , i.e. , there is an inter - shell vortex transition from a higher orbit to a lower one , which is similar to the coster - kronig transition in hollow atoms ( see , e.g. , @xcite ) .
this structural transition is accompanied by a local re - distribution of the flux inside the disk , or by the appearance of flux jump instabilities , which might have an additional triggering mechanism caused by the viscosity .
( in units of @xmath41 ) versus @xmath10 ( in units of @xmath67 ) near the structural transition between the states ( 1,6,12 ) ( solid black circles ) and ( 1,7,11 ) ( solid red / grey triangles ) , insets .
empty symbols correspond to bistable vortex motion , with two critical currents , @xmath9 and @xmath68 .
the jump in @xmath9 is related to an inter - shell vortex transition from a higher to a lower orbit , similar to the coster - kronig transition in hollow atoms .
, width=302 ] we found that this jump in @xmath9 is also observed for other vortex configurations and is generic for vortex structures confined in corbino disk .
for instance , in a disk with @xmath69 it is related to the transition between the states ( 1,8,13 ) and ( 2,8,12 ) at @xmath70 . note that a similar behavior ( a jump in the mobility ) is also found in the frenkel - kontorova model for the locked - to - sliding transition for chains moving on commensurate potentials @xcite .
consider now a corbino disk containing a larger number of vortex shells , e.g. , @xmath71 where vortices form three shells .
note that the chosen vorticity allows the triangular - lattice mnc ( 1,6,12,18 ) shown in the inset of fig .
4 . as in the above case of two shells , for small @xmath0
the system displays a rigid - body rotation .
however , the distance between the inner and the outer shells is larger , and , thus , the lorentz force gradient results in a stronger elastic deformation of the vortex configuration as compared to the two - shells case . the shear stress is stronger close to the center of the disk , and as a result , the _ inner shell splits off first _ at @xmath72 , with increasing @xmath0 , while second and third shells still rotate with the same angular velocity up to some critical current value @xmath73 ( see fig . 4 ) .
( in units of @xmath40 ) versus @xmath0 ( in units of @xmath41 ) in first ( orange / light grey open circles ) , second ( black solid line ) and third ( blue / dark grey open squares ) shells , for a three - shell vortex configuration with @xmath74 .
the inset shows a triangular - lattice ground - state configuration ( 1,6,12,18 ) for @xmath75 at @xmath76 .
the critical currents are : @xmath77 , and @xmath78 .
, width=302 ] the three - shell system displays a number of inter - shell vortex transitions when changing the applied magnetic field .
similar to the case of two shells , a structural `` phase transition '' occurs where vortices from a higher shell transit to a lower shell when the magnetic field increases .
this leads to non - monotonic dependences of the critical currents @xmath72 and @xmath79 on @xmath10 and to instabilities in the functions @xmath80 at different @xmath10 .
( in units of @xmath40 ) versus @xmath3 ( in units of @xmath81 ) in first , second and third shells ( using same symbols as in fig . 4 ) , for @xmath74 , @xmath82 , @xmath83 .
the onset of motion occurs in the form of rigid - body rotation . at @xmath84 ,
the inner shell splits off , the second shell unlocks at @xmath85 , and shells rotate independently ( _ angular _ melting ) .
the _ radial _ melting , i.e. , dissolving of the shells , occurs at @xmath86 .
the inset : the experimental @xmath3-dependence of @xmath2 obtained from the resistivity measurements at three different distances from the center of a corbino disk @xcite .
, width=321 ] now we will vary the temperature @xmath3 and study the _ thermally - activated _ externally - driven dynamics of vortices in the shells . for this purpose
, we introduce pinning centers randomly distributed over the disk ( for brevety we assume dense narrow pinning centers with maximum pinning force @xmath87 ) and apply a very low current @xmath0 that results in a lorentz force which is less than the pinning force @xmath87 for a vortex in _ any _ shell . at low @xmath3
all the vortices are pinned and the shells do not move . with increasing @xmath3 ,
the random - noise force is added to the ( weak ) lorentz force .
intuitively one expects that vortices in the inner shell , where the lorentz force is maximum , will depin first and start to rotate while vortices in the other shells are still pinned .
however , the vortex - vortex interaction prevents this scenario , and the inner - shell vortices , if unpinned , become `` trapped '' by the vortex lattice itself that leads to an elastic deformation .
the vortex - vortex interaction locks in the vortex configuration , and when for further increasing @xmath3 , the onset of the motion occurs in the form of a _ rigid - body _ rotation ( fig . 5 ) . at some temperature @xmath88 ,
the inner shell splits off and starts moving with a higher angular velocity , while higher shells stay locked up to a `` second critical '' temperature @xmath89 .
the transition from a rigid body to a separate rotation of shells is called _ angular melting _ ( i.e. , the `` vortex solid '' to `` vortex shells '' transition ) .
this is a multi - step process which starts when first shell splits off ( i.e. , at @xmath88 in our three - shell model ) and it finishes when all the shells rotate independently ( @xmath90 ) , but they keep their identity and contain a constant number of vortices with well - defined average radius @xmath91 .
the _ radial melting _ is associated with a _ dissolving _ of the shells ( i.e. , the `` vortex shells '' to `` vortex liquid '' transition ) and occurs at a higher temperature , @xmath92 ( fig .
this three - shell model explains qualitatively the experimentally observed thermally - induced `` solid - liquid '' melting transition in a corbino disk @xcite with three probes ( see inset to fig .
4 ) . also , the behaviour of @xmath80 ( fig .
4 ) is in agreement with the experimental results . in the experiment ,
the superconducting disk contained a large number of vortices , and the measurements were done for rings rather than for shells .
to model larger systems , we used a disk with , e.g. , 123 vortices forming six shells , and we calculated average velocities in three rings ( each containing two shells ) .
we found that the results for larger disks are similar to the above three - shell system , but the radial melting temperature @xmath92 is lower , in agreement with the experiment .
we predict an unusual non - monotonous behavior of the critical current for unlocking of the vortex rings with magnetic field which originates from a `` structural transition '' where a vortex jumps from a higher shell to a lower shell similarly to the coster - kronig transition in hollow atoms . pushing the experiments of ref . into the mesoscopic regime
will allow one to detect experimentally the predicted behaviour .
the vortex motion in the presence of pinning centers reveals the onset of a rigid body rotation , due to thermally - activated externally - driven flux motion . with increasing @xmath3 ,
the inner shell splits off first and subsequently all the shells start moving separately .
the present numerical study explains the experiments of ref . ,
in which vortex velocities in adjacent layers in a corbino disk were studied as a function of @xmath93 and @xmath3 .
we thank b. janko , f. nori , and u. welp for useful discussions and hospitality .
this work was supported by the flemish science foundation , and the belgian science policy . v.r.m .
acknowledges a partial support through pod .
chibotaru , a. ceulemans , v. bruyndoncx and v. v. moshchalkov , nature ( london ) * 408 * , 833 ( 2000 ) ; phys .
lett . * 86 * , 1323 ( 2001 ) ; v.r .
misko , v. m. fomin , j. t. devreese , v. v. moshchalkov , phys .
lett . * 90 * , 147003 ( 2003 ) .
crabtree , d. lopez , w.k .
kwok , h. safar , and l.m .
paulius , j. low temp .
phys . * 117 * , 1313 ( 1999 ) ; d. lopez , w.k .
kwok , h. safar , r.j .
olsson , a.m. petrean , l.m .
paulius , and g.w .
crabtree , phys .
rev . lett . * 82 * , 1277 ( 1999 ) . | in mesoscopic superconducting disks vortices form shell structures as recently observed in nb disks .
we study the dynamics of such vortices , driven by an external current @xmath0 , in a corbino setup . at very low @xmath0 ,
the system exhibits rigid body rotation while at some critical current @xmath1 vortex shells rotate separately with angular velocities @xmath2 .
this critical current @xmath1 has a remarkable non - monotonous dependence on the applied magnetic field which is due to a dynamically - induced structural transition with a rearrangement of vortices over the shells similar to the coster - kronig transition in hollow atoms .
thermally - activated externally - driven flux motion in a disk with pinning centers explains experimentally observed @xmath2 as a function of @xmath0 and @xmath3 and the dynamically - induced melting transition . |
4u 1820@xmath8303 ( hereafter 4u 1820 ) located in the globular cluster ngc 6624 is an ultracompact x - ray binary ( ucxb ) , which is defined by an ultra - short orbital period ( usually less than 1 hour ) . since its first discovery as a bright x - ray source @xcite ,
4u 1820 has been extensively observed with many x - ray telescopes .
its peak - to - peak modulation period @xmath9 s @xcite is generally believed as the orbital period of a white dwarf orbiting a neutron star @xcite . as the most compact ucxb
, the formation and evolution history of 4u 1820 has been widely studied so far .
considering the high stellar density of globular cluster , @xcite suggested that this system was formed via spiral - in phase of a neutron star into a red giant after their direct collision .
taking common envelope spiral - in into consideration , @xcite calculated an evolutionary sequence from 4u 2127@xmath1012 in m15 to 4u 1820 . in their scenario ,
the binary system was produced via tidal capture between a main - sequence star and a neutron star .
similarly , @xcite proposed that this source originated from an exchange interaction between a neutron star and a primordial binary including two main sequence stars , in which a common envelope evolution phase would subsequently be expected .
although adopting two different formation channels for 4u 1820 @xcite and @xcite agreed that the mass transfer is dominantly driven by gravitational radiation . based on a similar model , @xcite calculated the evolutionary sequence of 4u 1820 , and found that its donor star is a helium white dwarf of mass @xmath11 @xmath12 with a radius within @xmath13 of a completely degenerate configuration .
they also predicted an x - ray luminosity of @xmath14 @xmath15 @xmath16 , and an orbital - period derivation of @xmath17 @xmath18 driven by mass transfer from the white dwarf to the neutron star .
however , @xcite obtained an orbital - period derivative as @xmath19 @xmath18 by ginga observations .
subsequently , @xcite also reported a negative period derivative at a significance of @xmath20 . employing simultaneous rosat / ginga observations , @xcite reported an average orbital - period derivative @xmath21 @xmath18 . based on the further observation by rosat , @xcite obtained @xmath22 @xmath18 .
@xcite also reported a negative derivative of @xmath23 @xmath1 .
recently , @xcite analyzed the 16 yr _ rossi x - ray timing explorer _
( _ rxte _ ) data of 4u 1820 , and refined the negative orbital - period derivative to be @xmath24 @xmath18 at a @xmath25 level . obviously , all of the observations confirmed a negative orbital - period derivative for 4u 1820 , which is contrary to the theoretical prediction given by @xcite .
how to explain the difference in the orbital evolution between the binary evolutionary theories and the observations ? to resolve this problem , many authors proposed various scenarios , which can be divided into two categories . in the first one
it is thought that the orbital - period derivative arises from the binary evolution .
considering the secondary is a non - degenerate helium - star of @xmath26 @xmath27 orbiting a @xmath28 @xmath27 neutron star , @xcite calculated the evolution of a compact binary with initial period 37 minutes .
they found that the orbital period could be as low as 11 minutes and a negative period derivative could be attained also .
their calculation indicates that the donor star is a helium - burning star with mass of @xmath29 @xmath27 , which is not expected to exist in old stellar population like globular clusters @xcite . assuming that stars more massive than the turn - off mass could be formed during the close encounter in globular clusters , @xcite restudied the possibility of helium - burning donor star , and obtain a conclusion similar to @xcite .
@xcite found that a circumbinary disk ( cb disk ) around the binary can drive low - mass x - ray binaries to ultra - short periods as short as 6 min .
recently , @xcite investigated an alternative formation channel toward ucxbs with an orbital period of 11 minutes , which evolved from intermediate - mass x - ray binaries driven by magnetic braking of ap / bp stars .
their simulations also indicate a long - term period - decreasing phase . in the other case , some researchers suggested that the apparent orbital period derivative can be interpreted by an accelerating motion of the binary towards observers .
@xcite argued the acceleration may arise from a third body or the cluster potential .
@xcite explored the possibility of a stellar mass remnant close to 4u 1820 or an intermediate - mass black - hole inside ngc 6624 .
@xcite even suggested that the secular period change was dominated by systematic or random change of position and shape of occulting bulge on the accretion disk rim .
taking the orbital period change as the result of binary evolution , here we suggest that the negative orbital period derivative of 4u 1820 could be produced by a temporary cb disk , which can extract the orbital angular momentum via resonant interaction from the binary system .
our theoretical analysis and numerical simulation indicate that the cb disk mass that can account for the observed orbital - period derivative is approximately @xmath30 @xmath31 . in the following section
, we present a simple theoretical analysis for the current orbital - period derivative of 4u 1820 , and describe the cb disk model in section 3 . the numerical simulation method and simulated results
are described in detail in section 4 . in section 5 , we present a brief summary and discussion .
considering a binary consisting of a neutron star ( of mass @xmath32 ) and a he white dwarf secondary ( of mass @xmath33 ) in a circular orbit , the orbital angular momentum is @xmath34 . here
@xmath35 is the reduced mass , and @xmath36 is the orbital separation . the spin angular momentum of the donor star is neglected because it is much less than the orbital angular momentum . by a simple logarithmic differentiation of the angular momentum , we have @xmath37 taking the eddington accretion rate @xmath38 into account , the accretion rate of the neutron star is assumed to be @xmath39 , where @xmath40 is a constant .
the accretion efficiency of the neutron star is defined as @xmath41 therefore , when @xmath42 , @xmath43 while @xmath44 , @xmath45 .
x - ray observation performed by @xcite reported an x - ray luminosity of @xmath46 @xmath15 @xmath47 , which implies the accretion rate of 4u 1820 @xmath48 @xmath31 @xmath49 .
magnetic braking would turn off for a fully convective star @xcite , hence we consider angular momentum loss due to gravitational radiation and mass loss : @xmath50 where @xmath51 , and @xmath52 are the angular - momentum - loss rate by gravitational radiation , and mass loss , respectively .
the angular - momentum - loss rate by gravitational radiation is given by @xcite : @xmath53 where @xmath54 is the light velocity , @xmath55 is the gravitational constant , and @xmath56 is the total mass of the binary .
taking @xmath57 @xmath31 @xcite , @xmath58 @xmath31 @xcite and @xmath59 @xmath60 @xcite , we can derive @xmath61 @xmath49 .
the angular - momentum - loss rate by mass loss can be written as @xmath62 where @xmath63 is the specific angular momentum of the ejected matter in units of @xmath64 . in this work , the mass loss during the accretion is assumed to form an isotropic wind in the vicinity of the neutron star , and carry away its specific orbital - angular momentum , i.e. @xmath65 .
based on some parameters mentioned above , one can find @xmath66 @xmath49 .
comparing with gravitational radiation , the angular - momentum loss due to mass loss can be ignored .
based on some parameters mentioned above , equation ( 1 ) can yield the current orbital period derivative of 4u 1820 . for a conservative mass transfer ,
the current period derivative should be @xmath67 @xmath1 , while @xmath68 @xmath1 for nonconservative mass transfer with @xmath69 @xcite .
similar to the discussion given by @xcite , both cases predict a positive orbital period derivative , which is contradicted with the observation .
to interpret the negative orbital - period derivative observed in 4u 1820 , here we propose an evolutionary cb disk model .
the influence of cb disk on the evolution of cataclysmic variables @xcite , black - hole x - ray binaries @xcite , algol binaries @xcite , and ucxbs @xcite have been studied extensively .
all these works show that cb disk can efficiently extract the orbital angular momentum from the binary system , and enhance the mass transfer rate and accelerate the evolution process . in this work
, we adopt a different cb disk model .
the resonant interaction between the binary and the cb disk has been well studied @xcite , and its predicted orbital angular momentum loss rate is given by @xcite : @xmath70 here @xmath71 is the mass of cb disk , @xmath72 and @xmath73 ( @xmath74 , where @xmath75 , and @xmath76 are the inner and outer radius of the cb disk , respectively ) are the thickness and the half angular momentum radius of the disk , respectively . @xmath77 and @xmath78 are integers describing the binary potential , @xmath79 is the viscosity parameter of the disk .
@xcite studied the tidal truncation between circumstellar and cb disks in binaries , and found the inner radius of cb disc is @xmath80 .
following their study , assuming the disk extends from @xmath81 to @xmath82 , the half angular momentum radius of the disk is @xmath83 . according to the study of @xcite , the relative thickness of cb disk near the inner edge is @xmath84 . assuming the disk is thin and the non - axisymmetric potential perturbations are small we take @xmath85 and @xmath86 in this work .
according to above equations a cb disk with mass of @xmath87 @xmath31 can reproduce the observed negative orbital - period derivative for both conservative and the nonconservative mass transfer cases .
certainly , the cb disk would slowly induce a small eccentricity @xcite .
however , the tidal interaction between the donor star with a relatively deep convective envelope and the neutron star in such a compact orbit would rapidly circularize the orbit .
we propose the material of the cb disk comes from some _ rare _ superburst events .
superbursts are hour - long thermonuclear runaway burning processes which were first observed by @xcite from the neutron star low mass x - ray binary 4u 1735@xmath8444 .
a superburst of 4u 1820 was discovered by @xcite and @xcite as the second superburst event ever observed .
subsequently , many other superburst events have been discovered from other lmxbs such as ks 1731@xmath8260 , serpens x@xmath81 gx 3@xmath101 and so on . at present ,
about one dozen superburst sources ( candidates ) have been reported , and one third of these sources were observed recurrently @xcite .
interestingly , although the recurrence times estimated from observation of other sources are on the order of a year @xcite , the second superburst of 4u1820 was observed to be 11 years after the first event @xcite , which is consistent to the prediction ( @xmath88 ) by @xcite .
compared with more frequent type i bursts with short duration ( typically @xmath89 s ) , superbursts release x - ray energy @xmath90 @xmath15 during an event , which is three orders of magnitude higher than that of type i burst . considering a substantial fraction of the released energy during the event
is carried away by neutrinos , and is conducted to the inner part of the neutron star , the total energy released during the event is much larger @xmath91 @xmath15 @xcite .
observational and theoretical studies indicate that it is a thermonuclear runaway burning process , and the energy source is most likely carbon and/or oxygen on the surface of the neutron star @xcite .
a detailed analysis indicated the mass of the burning - carbon layer is probably @xmath92 @xmath93 @xcite . in principle
, there exist a range for the released energy of different superburst events @xcite . here
we assume that some peculiar superburst events ( with a probability of @xmath94 , hereafter rare superbursts ) in a layer of @xmath95 @xmath93 ( @xmath96 @xmath31 ) may be responsible for the formation and evolution of cb disk . in our calculation
, 20 percent of the burning - carbon layer was assumed to escape from the neutron - star surface , and 10 percent of the ejected material ( 2 percent of total burning mass , i.e. @xmath97 @xmath31 ) suddenly feed the cb disk during every rare - superburst event . take the recurrence time of superburst events of 4u1820 is 10 yr , as a result , the cb disk would experience a mass feed with @xmath98 every @xmath99
. meanwhile , the mass of cb disk is assumed to decrease at a rate @xmath100 ( the cb disk only remains @xmath101 at the beginning of next rare superburst ) due to the photo - evaporation of the neutron star s spin - down energy @xcite . according to equation ( 5 ) , the angular - momentum loss due to sudden mass loss from the neutron - star surface induced by ( normal or rare ) superburst is @xmath102 @xmath49 , and can be ignorable .
to evaluate the cb disk scenario in detail , we develop a fast binary - evolution program for 4u 1820 . as shown in fig .
1 , our calculation starts from gravitational wave emission dominated detached binary consisting of a neutron star and a low - mass he white dwarf . to fit the current mass of the neutron star ( @xmath103 @xmath31 , @xcite ) and the donor star ( @xmath104 @xmath31 , @xcite ) , the initial input parameters is set to @xmath105 @xmath31 , @xmath106 @xmath27 , and orbital period @xmath107 hr .
such an orbital - period can ensure a he white dwarf to decouple from its roche lobe and the steady mass transfer condition is not introduced manually but satisfied via slowly adjust of gravitational wave emission .
this is different from @xcite where the evolution was studied from roche lobe filled binary with simple relation between the mass of white dwarf donor and the orbital period : @xmath108 . once the donor star overflows its roche lobe due to gravitation radiation , it would transfer material to the neutron star at a rate @xcite : @xmath109 where @xmath110 is a constant factor . in calculation , we take @xmath111 .
@xmath112 is the radius of the secondary , and @xmath113 is the roche - lobe radius of the donor star .
the function @xmath114 only relate to the mass ratio @xmath115 . in this paper
we adopt the approximately description given by @xcite , i.e. @xmath116 utilizing a simple polytropic index @xmath117 , the donor - star radius is given by @xcite @xmath118 where @xmath119 is the mass abundance of hydrogen ( @xmath120 for a pure helium white dwarf in this paper ) , and @xmath121 ( @xmath122 in this paper ) is the radius ratio between the donor star and the white dwarf with same mass , which is completely degenerate and only supported by the fermi pressure of the electrons . to test the influence of input parameters on the evolution and compare the results with other works
, we have run four different models as follows : * model 1:@xmath123 , @xmath124 ; * model 2:@xmath123 , @xmath125 @xmath31 ; * model 3:@xmath69 , @xmath124 ; * model 4:@xmath69 , @xmath125 @xmath31 ; fig .
2 shows the evolution of an ucxb under different models in the @xmath126 plane .
it is clear that the orbital - period evolution can be divided into two stages . in the first stage
, gravitational radiation dominates the angular momentum loss of the binary , and causes the orbit to sharply shrink .
after the donor star overflows its roche lobe , the orbital decay becomes slower because the material is transferred from the less massive donor star to the more massive accretor .
subsequently , the orbital period gradually increases , and the second stage starts where the evolution roughly obeys an orbital period - donor mass relation as @xmath108 .
our simulation also shows that at the orbital period of @xmath127 min the donor star mass is in the range of @xmath128 , in good agreement with the predication by @xcite . the cb disk formed by the first rare superburst
is assumed to experience a mass feed if the mass - transfer rate @xmath129 .
it seems that four models have a similar evolutionary tracks .
however , in the mini panel of fig .
2 , models 2 and 4 predict short - term orbital - decay episodes , which were induced by sudden mass feed of the cb disk . [ cols="^,^ " , ]
mass transfer from the less massive donor star to the more massive accretor always induces an expanding orbit . in the case of stable mass transfer ,
the orbital - period derivative is proportional to the mass - transfer rate and the rate of angular - momentum loss @xcite .
therefore , the negative orbital - period derivative of 4u 1820 remains mysterious for a white - dwarf binary . in this work
, we propose a cb disk scenario with a cycle - mass feed to interpret the anomalous orbital - period derivative of this source . in our model ,
the runaway burning process during a superburst event may carry away the material of burning layer .
if a small fraction of the material feeds a cb disk around the binary rather than leaves it , the related abrupt change in the total angular - momentum - loss rate destroy the condition for a stable mass transfer , and induce the observed negative orbital - period derivative during the recovery of the stable mass transfer .
although the mass of the burning - carbon layer on the neutron - star surface is only @xmath130 @xmath12 during normal superburst events , we assume that some rare superbursts with a probability of 0.1% have a burning - layer mass of @xmath131 @xmath12 . in calculation ,
rare superbursts are assumed to reoccurrence events with a cycle period of 10,000 yr , and 2 percent of the burning - layer mass feed into a cb disk around the binary . in principle
, photo - evaporation process by the neutron star s spin - down energy would decrease the mass of the cb disk @xcite .
however , the mass - loss rate of the cb disk sensitively depends on the photo - evaporation efficiency , the binary separation , and the spin period of the pulsar @xcite .
therefore , we simply assume that the cb disk lost 0.1% of its current mass every year .
our numerical calculation show that such a cb disk model can account for the donor - star mass , orbital period , and orbital - period derivative observed in 4u 1820 .
if the rare superburst events have a recurrence time of 10,000 yr , the cb disk model predicts a 900 yr timescale in the period - decreasing stage , and gives a probability of @xmath132 to observe a negative period derivative . certainly , because of the recovery of stable mass transfer , the absolute value of orbital - period derivative should slowly decrease which can be tested by future further observations . in addition , because the continuum contribution of dust emission from the cb disk could observed in l band ( @xmath133 ) @xcite .
therefore , we expect future detailed multi - waveband observations for 4u 1820 to confirm or refute our scenario . | it is generally thought that an ultracompact x - ray binary is composed with a neutron star and a helium white dwarf donor star . as one of the most compact binaries , 4u 1820 - 303 in globular cluster ngc 6624
was predicted an orbital - period derivative at a rate of @xmath0 @xmath1 if the mass transfer is fully driven by gravitational radiation .
however , the recent analysis of the 16 @xmath2 data from _ rossi x - ray timing explorer _ and other historical records yielded a negative orbital - period derivative in the past @xmath3 yr . in this work , we propose an evolutionary circumbinary ( cb ) disk model to account for this anomalous orbital - period derivative .
4u 1820 - 30 was known to undergo superbust events caused by runaway thermal nuclear burning on the neutron star .
we assume that for a small fraction of the superbursts , part of the ejected material may form a cb disk around the binary . if the recurrence time of such superbursts is @xmath4 yr and @xmath5% of the ejected mass feeds a cb disk , the abrupt angular - momentum loss causes a temporary orbital shrink , and the donor s radius and its roche - lobe radius do not keep in step . driven by mass transfer and angular - momentum loss
, the binary would adjust its orbital parameters to recover a new stable stage .
based on the theoretical analysis and numerical simulation , we find that the required feed mass at the cb disk is approximately @xmath6 @xmath7 . |
mp centauri ( hd 308976 ) is an eclipsing binary , listed in the hd catalog as having a b3 spectral type .
its orbital period of 2.99 days makes it a very difficult target to obtain complete light curves from a single location in one observing season .
it is a member a rare group : massive eclipsing binaries with short orbital periods .
it lies in the galactic plane ( @xmath4 , @xmath5 ) and its eclipsing nature allows for the determination of its absolute parameters and distance , as well as the reddening along the line of sight .
we have obtained @xmath6 photometry over several years and the system was also observed in @xmath7 and @xmath8 by the all sky automated survey ( @xcite ; hereafter asas ) . as part of a program on early - type overcontact and near - contact binaries , we observed mp cen with the eso la silla 2.2 m telescope and the feros spectrograph at high resolution ( r=48,000 ) in february , 2003 . combining radial velocities from these spectra with the @xmath6 photometry and @xmath7 and @xmath8 data from asas , we present the first comprehensive study of this evolved binary .
gw observed mp cen in 1982 and 1989 using the 24-inch ( 0.6 meter ) telescope and single channel photometers equipped with @xmath6 filters at the mt .
john observatory at lake tekapo , new zealand .
the comparison star used was hd 102139 , for which the simbad database gives a b4 iii spectral type .
a 17 arcsec aperture was used and offset slightly to avoid contamination from nearby companions .
the asas project has also obtained @xmath7 and @xmath8 ccd photometry of mp cen . because the goal is to survey large areas of the sky ,
the asas setup is not optimal for photometry of stars in crowded fields like mp cen .
however , mp cen is the brightest object in the measuring aperture and a proper treatment of the third light ( @xmath9 ) contamination makes it possible to use the asas data .
the scatter in all passbands was about 2% .
seven eso 2.2 m + feros spectra have been secured , in two groups of three spectra each on 2003 feb 23 and 24 near quadratures and a single spectrum on 2004 feb 25 in primary eclipse ( table 1 ) .
the wavelength range of the spectra extends from 3900 to 9200 with @xmath10 and a 1200 second exposure time .
the s / n is evaluated around 5850 and the orbital phase is computed with the ephemeris given in section [ analysis ] .
[ speclist ] [ cols="^,^,^,^,^",options="header " , ] quoted errors are formal 1-@xmath11 errors from the solution .
@xmath9 values are in units of total system light at phase 0.25 .
luminosity errors are estimates based on an uncertainty of 2000 k in the effective temperature of the primary .
the error contributions to the luminosities due to the errors in the radii are more than an order of magnitude smaller .
the @xmath12 values are in cgs units . and
@xmath8 photometry and computed light curves.,width=317 ] the asas data show that the system has @xmath13 and with the third light correction from our light curve solution , @xmath14 for the binary at maximum brightness . correcting this value for the reddening ( assuming a ratio of total to selective absorption of @xmath15 and our @xmath16 ) , we find that the apparent v magnitude of the binary is 9.08 . using the @xmath7 luminosity ratio @xmath17=0.36 from our light curve solution , this transforms into johnson v magnitudes of @xmath18=9.41 and @xmath19=10.53 . with our log luminosity of @xmath20 for the primary , the distance to the primary amounts to 3.0 kpc using a bolometric correction of -1.75 from @xcite .
the same calculation for the secondary results in the same distance .
we therefore adopt 3.0 kpc as the distance to mp cen , placing it on the far side of the sagittarius arm .
mp cen lies in the galactic plane and we can compare our estimate of the reddening with the @xcite extinction maps .
mp cen lies at the boundary of four cells ( 294/-1 , 296/-1 , 297/0 and 292/2 ) .
given the low resolution of the cell boundaries , we can not unambiguousily assign mp cen to a particular cell .
our value of @xmath21=0.30 leads to @xmath22 .
looking at the individual cells , we find * 294/-1 distance greater than 1 kpc * 296/-1 the distance could be as high as 4 kpc * 297/0 distances from 1 to 4 kpc supported * 292/2 distances from 0.5 to 4 kpc supported thus , the @xcite maps do not provide a conclusive comparison with our derived distance , but their figure 9 supports the idea that the line of sight to mp cen goes through clouds a and g and just misses clouds l , m and r.
we have presented preliminary values for the parameters of mp cen based on photometry and limited spectroscopic data .
the masses and radii of the stars , along with the semidetached configuration , indicate that the system is evolved .
the radii , @xmath23 and @xmath24 , are consistent with a giant or sub - giant classification for both stars for the derived masses of @xmath25 and @xmath26 .
the @xmath27 we have measured implies a conservative mass transfer rate of @xmath28 , making it a fairly active system .
light curve variations from season to season , most likely arising from this mass transfer activity , are apparent in the photometric data .
we also see emission features in our spectra but the limited phase coverage of the spectra makes any discussion of the circumstellar environment little more than speculation at this point .
more extensive spectroscopy and polarimetry will be required to fully map the circumstellar environment of the system .
this research has made use of the simbad database , operated at cds , strasbourg , france .
andersen , j. & grnbech , b. 1975 , a&a , 45 , 107 bessell , m. s. , castelli , f , b. & plez 1998 , a&a , 333 , 231 bessell , m. s. 2000 , pasp , 112 , 961 cox , a.n .
2000 , allen s astrophysical quantities , ( american institute of physics press ) flower , p. j. 1996 , apj , 469 , 355 kurucz , r. l. 1993 , in light curve modeling of eclipsing binary stars , , ed .
e. f. milone ( new york : springer - verlag ) , 93 leung , k .- c .
& wilson , r. e. 1977 , apj , 211 , 853 munari , u. & tomasella , l. 1999 , a&a , 343 , 806 munari , u. & zwitter , t. 1997 , a&a , 318 , 269 neckel , t. & klare , g. 1980 , a&as , 42 , 251 pojmanski , g. 2002 , acta astron .
, 52,397 terrell , d. , munari , u. , zwitter , t. & nelson , r. h. 2003 , aj , 126 , 2988 van hamme , w. 1993 , aj , 106 , 2096 wilson , r. e. 1979 , apj , 234 , 1054 wilson , r. e. 1990 , apj , 356 , 613 wilson , r. e. 2001 , inf .
variable stars , no .
5076 wilson , r. e & devinney , e. j. 1971 , apj , 166 , 605 ( wd ) wilson , r. e & terrell , d. 1998 , mnras , 296 , 33 | we present photometric and spectroscopic data on the early - type binary mp centauri .
the photometric data are analyzed simultaneously with radial velocities to derive preliminary absolute dimensions for the binary components .
analysis of the spectra shows that the stars rotate synchronously and that the line of sight to the system crosses two kinematically sharp and well - separated interstellar reddening sources .
it is shown that mp cen consists of a b3 primary with @xmath0 , @xmath1 and a lobe - filling b6-b7 secondary with @xmath2 , @xmath3 [ firstpage ] binaries : eclipsing binaries : spectroscopic stars : individual ( mp cen ) stars : evolved . |
we show that a simple , yet previously overlooked , mechanism may give rise to a substantial velocity dispersion in the proton component of gamma - ray burst fireballs .
namely , in grb fireballs in which neutron decoupling occurs , large center of mass energy collisions between decoupled neutrons and protons will induce a velocity dispersion in the proton component . for certain fireball parameters
the timescale for re - thermalization of these protons is longer than the dynamical timescale , and so the protons will retain their acquired dispersion .
this has consequences for the dynamics of the fireball . since the ultra - relativistic proton shell
is directly tied to gamma - ray photon production @xcite , the distribution of proton energy can affect the dynamics of photon emission .
in the internal shock grb model the @xmath0-ray photons are produced through internal multiple - shock collisions involving the protons .
the efficiency for the conversion of shock kinetic energy to photon energy increases with increasing relative shock velocities @xcite .
in addition , the energetic proton shock produces particle cascades leading to multi - tev neutrino emission @xcite .
therefore , both photon and neutrino radiation may be affected by the process discussed here .
more basically , the process we discuss here represents a new aspect of our understanding of the evolution of homogeneous relativistic fireballs .
the standard picture holds that a thermal plasma undergoes an initial acceleration phase , lasting until the energy in relativistic particles is transferred to the kinetic energy of protons , or , in the case of extremely baryon - dilute plasmas , until the plasma becomes optically thin to photons @xcite .
following the acceleration phase a coasting phase ensues . in the coasting phase
the baryons are assumed thermal with typically low temperatures , @xmath1 , in the plasma rest frame .
recently it has been recognized that in fireballs satisfying certain initial conditions the neutrons will dynamically decouple @xcite from the expanding @xmath2/photon / proton plasma .
there is an interesting connection between this decoupling and the net number of electrons per baryon , @xmath3 where @xmath4 is the electron fraction , @xmath5 and @xmath6 are the number densities of positrons and electrons respectively , and @xmath7 and @xmath8 are the number densities of neutrons and protons .
the connection is that when conditions near the fireball source lead to a low @xmath4 in the outflow , substantial differences between the final neutron and proton lorentz factors are possible @xcite .
we will demonstrate a second connection : when the electron fraction in the fireball is low , high energy collisions between decoupled neutrons and protons induce a velocity dispersion in the protons and that these `` hot protons '' are not re - thermalized .
the sensitivity of this effect to @xmath4 in the fireball is interesting because , in analogy to type ii supernovae , the electron fraction may mirror weak physics in the central engine @xcite .
the essential features of the physics of baryon flow in relativistic fireballs are obtained by considering a two component [ ( i ) @xmath9/photon / proton and ( ii ) neutron ] homogeneous fireball with initial radius , temperature , and electron fraction @xmath10 and @xmath11 respectively . in terms of these quantities the entropy per baryon in units of @xmath12 is @xmath13 here @xmath14 is the ratio of total energy to rest mass in the fireball .
a central issue for grbs is the `` baryon loading problem '' , the statement that @xmath15 must be large in order to get the fireball moving with a high lorentz factor . for relativistic fireballs ( @xmath16 a few ) , numerical and analytic work show that to first approximation the fireball evolves as @xcite @xmath17 here @xmath0 is the lorentz factor of the fireball .
these relations follow from entropy and energy conservation and are violated at the beginning and end of the fireball s evolution ( see below ) . in terms of the time @xmath18 as measured by an observer comoving with the plasma , the lorentz factor and temperature evolve as @xmath19 . here
@xmath20 characterizes the fireball source size .
( we adopt units where the speed of light is unity .
) observations of time variability in grbs give the constraint @xmath21 , while the smallest proposed grb sources are solar mass scale compact objects with @xmath22 .
particles in the plasma suffer a 4-acceleration with magnitude gamma @xmath23 .
neutrons have such a small magnetic dipole moment that they are essentially only coupled to the accelerating plasma via strong scatterings with protons ( near the decoupling point @xmath24 ) where @xmath25 is the thomson cross section . to an excellent approximation ,
then , the neutron - proton collision timescale in terms of the comoving proton number density @xmath26 in the plasma rest frame is @xmath27 where @xmath28 is the relative neutron proton velocity .
apart from small corrections stemming from inelastic nucleon - nucleon scatterings , entropy is conserved throughout the acceleration of the fireball .
this implies @xmath29 except for the brief period when the electron / positron pairs annihilate and transfer their entropy to the photons .
decoupling occurs if the dynamic and collision timescales become comparable before the end of the acceleration phase of the fireball s evolution .
this condition is @xmath30_{\rm dec } } } > 1.\ ] ] here @xmath31 is the dynamic timescale in units of @xmath32 , and @xmath33 is the product of the relative velocity and neutron / proton cross section in units of 10 @xmath34 at the decoupling point .
the precise @xmath28 at which this latter quantity should be evaluated is unclear from our simple argument .
this is not crucial because @xmath35 $ ] only varies by a factor of 5 from unity as @xmath28 increases from @xmath36 to @xmath37 .
the final lorentz factor of the neutrons is estimated by evaluating @xmath0 at the decoupling point , @xmath38_{\rm dec } t_{0,{\rm mev}}^3 y_e \tau_{5}}{s_5 } \right)^{1/3}.\ ] ] following decoupling , the proton/@xmath2/photon plasma begins to accelerate away from the neutrons .
energy conservation gives an estimate of the final mean lorentz factor of the protons , @xmath39.\ ] ] the term @xmath40 is present to account for energy lost to neutrinos arising from the decay of pions created in inelastic nucleon - nucleon collissions . here
@xmath41 is the nucleon mass , @xmath42 is the number of pions created per baryon ( at most of order unity ) , @xmath43 is the average center of mass energy lost to neutrinos per pion decay , and @xmath44 is the average lorentz factor of created pions . for pions created in n - p collisions , the branching ratio for charged pion production below the two pion threshold is roughly 1/2 , and @xmath45 .
it is difficult to analytically estimate @xmath44 , but as most inelastic nucleon - nucleon collisions occur near the decoupling point , it is presumably of order @xmath46 . in the numerical examples we present below
, neutrino energy loss is estimated and is found to be small enough so as not to alter the qualitative picture presented here .
subsequent to neutron decoupling , protons will undergo collisions with decoupled neutrons .
the first requirement for the presence of a `` hot '' proton component is that some of these collisions are high energy collisions .
if in the lab frame a proton and neutron have lorentz factors @xmath47 and @xmath48 respectively , then the @xmath0 of the neutron as seen by the proton is @xmath49 a high energy neutron - proton collision implies then : @xmath50 this condition is intendend only as a rough guide for the fireball parameters needed to drive the proton component hot . for reasonable fireball parameters
it is difficult to attain substantial @xmath51 unless @xmath52 .
it has been argued that a low electron fraction will naturally be obtained in many of the proposed grb environments @xcite .
in addition , in calculations of grbs arising from neutron star mergers @xmath53 has been estimated @xcite .
it is not clear if such low electron fractions are also obtained , for example , in the collapsar model for grb central engines @xcite .
protons undergoing collisions with neutrons will lose energy to the background photons and electrons . whether or not these protons re - thermalize with the plasma depends on their thermalization timescale @xmath54 .
the protons will remain `` cool '' ( _ i.e. , _ well coupled and thermal ) if @xmath55 , where the last inequality holds near and after decoupling .
if the opposite case holds , @xmath56 , the protons could keep the velocity dispersion they acquire through collisions with neutrons ( _ i.e. _ become `` hot '' ) .
there are two energy loss mechanisms for protons with a velocity relative to the background plasma .
the first is proton - electron scattering .
bethe s formula @xcite gives this rate as : @xmath57\ ] ] here @xmath58 is the electron number density in units of @xmath59 and @xmath60 and @xmath0 refer to the velocity and lorentz factors respectively , of the proton relative to the plasma .
the relations derived above imply that decoupling occurs at a temperature @xmath61 , with @xmath62 .
this clearly implies that decoupling occurs after @xmath2 annihilation . noting that the ionization electron number density is @xmath63 , and noting that typical kinetic energies of protons relative to the plasma are of the order of a few gev , we see that the thermalization timescale via proton / electron scattering is @xmath64 ref .
@xcite made the observation that a pion - induced electromagnetic cascade may lead to an increase in the number density of electrons and positrons in the fireball .
the greatest this increase could be occurs when all of the pion mass ( minus neutrino losses ) goes into @xmath65 pairs and when , in addition , the processes creating @xmath65 pairs are sufficiently rapid that they come into equilibrium with electron / positron pair annihilation . by considering this limiting case
, one finds that a pion - induced electromagnetic cascade can delay the onset of a dispersion in the proton component ( _ i.e. _ make @xmath66 ) only for very small electron fractions , @xmath67 .
( the number 4 appearing in parentheses here arises from requiring that a large center of mass energy collision is needed to drive a proton dispersion and is explained in more detail below ) .
therefore , a pion induced electromagnetic cascade has little or no leverage in delaying the development of a dispersion in the protons for moderate electron fractions .
the second energy loss mechanism for protons arises because , when the protons acquire a dispersion , the electrons must rearrange themselves in order to preserve charge neutrality .
this rearrangement , on its own , has a negligible effect on the proton dispersion because of the smallness of the ratio of the electron to proton mass .
however , compton scattering following this rearrangement can be a source of energy loss for the protons .
in other words , runaway protons may be viewed as being tightly paired with electrons via the coulomb force , and the proton loses energy not only at the rate given in eq .
[ tautherm1 ] , but in addition at the rate at which the electron to which it is coupled loses energy to the background photons via thomson / compton drag : @xmath68 with @xmath69 the photon energy density , giving , @xmath70 the total thermalization timescale is given by @xmath71 a hot proton component arises , then , if neutron decoupling occurs , and at the end of the fireball s acceleration phase @xmath72 .
applying the simple picture given above for the evolution of the fireball gives the second condition for a hot proton component : @xmath73 in fact , the fireball is still accelerating for fireball radius values greater than @xmath74 ( @xcite , and see below ) .
this implies that the temperature during the acceleration phase becomes lower than @xmath75 and , hence , that the condition for the development of a hot proton component is somewhat weaker than that given above . in fig .
1 we illustrate the decrease in degree of thermalization with increasing initial @xmath76 for a proton moving relative to a background plasma and losing energy via the the processes described in eqs .
[ tautherm1 ] and [ tautherm2 ] .
( i.e. to different plasma temperatures ) at @xmath77 cm .
the different values of @xmath78 are given next to the curves.,width=288 ] in addition to the requirement of charge neutrality leading to proton energy loss via electron compton scattering , there are other collisionless processes that may affect the dynamics of the protons .
indeed , the proton and electron plasma frequencies are comparable to or larger than the collision frequencies after decoupling , @xmath79 with @xmath80 and @xmath81 the proton and electron plasma frequencies .
the post decoupling proton dispersed fireball may provide fertile and novel ground for the study of plasma instabilities .
in particular , the protons may be subject to an analog of the two - stream instability because of collisionally produced bumps in the proton distribution function at the bulk plasma and near the mean neutron component velocities .
the plasma may also be subject to velocity space anisotropy - driven electromagnetic instabilities as the transverse velocity freezes out during the expansion of the fireball .
relevant for the present work , it has been shown that magnetic trapping can saturate these instabilities at substantial ion anisotropies @xcite , and that they are moderately stabilized by relativistic effects @xcite .
a detailed study of the saturation level of the proton anisotropy in this system would be interesting because lower saturation levels imply less dispersion reduction due to adiabatic cooling of the tranverse momenta . in the presence of a magnetic field , either generated by the central engine or created through instabilities , other instabilities , involving ,
for example , ion - cyclotron waves , may operate .
instabilities may play an important role in the setting of the proton distribution function .
they are not expected to alter our result of a dispersion in the proton component , because for the closely related system of counter - streaming ion beams they are well studied and a substantial reduction in an initial dispersion is not seen @xcite .
these instabilities have also been studied in the regime where collisions are important , and in this case counter - streaming ion beams take , as expected , a few collision timescales to slow @xcite .
however , studies of the coupling between plasma instabilities and the hydrodynamic evolution of the protons in the peculiar high photon entropy / small dynamic timescale environment characteristic of grb fireballs have apparently not been done , and may give interesting and unexpected results .
the fact that plasma instabilities play a role during the acceleration phase of some relativistic fireballs is quite interesting , because the previous picture was that collisional processes are sufficient to describe the early evolution in the absence of central engine - generated magnetic fields . a diagram illustrating the various influences on the proton distribution function , including the trend toward a smoother distribution function resulting from plasma instabilities , is shown in fig .
2 . represent the processes in eqs .
[ tautherm1 ] and [ tautherm2 ] , respectively .
, width=288 ] when neutron decoupling occurs , and is pronounced , a rough estimate gives the dispersion in proton lorentz factor as @xmath82 ( a somewhat more precise measure of dispersion will be given below . ) for example , consider a fireball described by @xmath83 and suppose that @xmath84 and @xmath85 ( implying @xmath86 ) .
we find @xmath87 and @xmath88 implying a dispersion in the proton component of @xmath89 .
we have defined the dispersion in terms of the central engine rest frame . for some purposes it is more useful to express the dispersion in terms of the mean proton rest frame .
a simple approximate measure of the dispersion in this frame is @xmath90 which is at most of order a few .
a more precise determination of the dispersion in the protons requires a transport calculation describing the evolution of the fireball .
a simple single angular zone , hydrodynamically consistent , steady state relativistic wind transport calculation has been performed .
neutrino energy loss is estimated by assuming that following an inelastic collision the available kinetic energy is shared equally between two baryons and a pion .
the pion energy is then assumed lost from the system . in the calculation ,
protons suffering large energy ( @xmath91 ) collisions with neutrons are assumed to decouple from the plasma if @xmath92 , where @xmath93 is taken as a free parameter . in the limit where neutrons and protons are coupled to the photon/@xmath2 plasma our calculation reduces to that done in the pioneering study by paczyski @xcite on ultra - relativistic winds from compact sources . in fig .
3 we present the results of a calculation for the parameters listed above : @xmath94 , @xmath84 , @xmath85 and for the cases where @xmath95 and @xmath96 .
results for different values of @xmath4 have been presented to illustrate that the lower the electron fraction , the more pronounced the effects of collisions between neutrons and protons . .
the dot dashed line is for case ( i ) , where baryons are assumed coupled to the plasma ; the dashed line is for case ( ii ) , where neutron velocities are described by a velocity distribution function and where neutrons are allowed to decouple while protons are assumed frozen into the plasma ; and the solid line is for case ( iii ) , where neutrons and protons are both allowed to decouple . note the change in scale between the upper and lower figures.,width=288 ] to illustrate the effects discussed here we present results for the case where ( i ) protons and neutrons are assumed coupled to and thermalized with the plasma ; ( ii ) a distribution function is used to describe the neutrons but the protons are assumed coupled to and thermalized with the @xmath97photon plasma ( the agreement between the final proton lorentz factors in this case and the simple estimate given by eq .
[ gammabarp ] is good ) ; and ( iii ) a velocity distribution function is used to describe the neutron velocities , and protons are assumed to decouple from the plasma once they have undergone a collision with a neutron where @xmath91 . for part ( iii ) of the calculation , the parameter @xmath93 was taken to be 5 .
for the example parameters we have chosen the condition @xmath98 is always satisfied at the stage in the fireball s evolution when @xmath91 collisions occur , and so the first condition is somewhat superfluous here .
perhaps the most striking feature of this calculation is the increase in lorentz factor of the highest energy protons as the protons acquire a dispersion . because energy conservation implies that the mean lorentz factor of the protons is unchanged between cases ( ii ) and ( iii ) ( modulo changes in the net energy carried by the neutrons and pions due to the protons acquiring a velocity dispersion ) the difference in final lorentz factors between cases ( ii ) and ( iii ) provides a measure of the proton dispersion .
note that because our calculation does not allow for a proton to interact with the expanding plasma once it has undergone a high energy collision with a neutron and once @xmath99 our calculation may somewhat overestimate the proton dispersion . in fig .
4 we display the evolution of a measure of dispersion in the proton component of the fireball ( @xmath100 ) . in this figure @xmath101
is defined to be the difference in proton lorentz factors between models ( ii ) and ( iii ) for the case @xmath102 shown in fig . 1 . with radius for the case @xmath103 and the parameters given in the text.,width=288 ]
we have pointed out and made estimates of an overlooked ( and possibly generic ) feature of the baryon ( neutron / proton ) flow in grb fireballs : the acquisition of a large dispersion in lorentz factor of the proton component subsequent to neutron decoupling .
this represents a qualitatively new feature of grb fireballs because the previous picture of the evolution of the proton component held that the protons were thermal , even when neutron decoupling occurs .
a dispersion in the proton component also implies that plasma instabilities may play an important role and suggests new studies of the role of collisionless processes during the acceleration and later shocking phase of the fireball s evolution .
because the effect we have described is principally sensitive to the neutron to proton ratio in the fireball , this work represents a new connection between the central engine weak physics that sets @xmath4 and grb fireball dynamics .
because shocks involving protons are thought to give rise to the photon signal characteristic of grbs , a proton dispersion also has implications for the details and efficiency of photon production .
a direct consequence for the charged particle dynamics of neutron decoupling is interesting because the signal from the decay of pions created in inelastic collisions during neutron decoupling is expected to be weak .
detector event rates for the neutrinos from pion decay are estimated at a few per year @xcite .
direct evidence of neutron decoupling in a grb event would provide information about the progenitor fireball parameters and perhaps give a clue about the properties of the central engine .
this work was partially supported by nsf grant phy98 - 00980 at ucsd , an igpp mini - grant at ucsd , and a nasa gsrp for ka .
we are indebted to neal dalal , c.y .
cardall , tom oneil , pat diamond , and dan dubin for useful insights . | we demonstrate a qualitatively new aspect of the dynamics of gamma - ray burst ( grb ) fireballs : the development of a substantial dispersion in the proton component in fireballs in which neutron decoupling occurs and is sufficiently pronounced .
this effect depends sensitively on the neutron to proton ratio in the fireball , becoming more dramatic with increasing neutron excess .
simple physical arguments and transport calculations indicate that the dispersion in lorentz factor of the protons can be of order the final mean lorentz factor of the fireball .
we show how plasma instabilities could play an important role in the evolution of the fireball and how they might ultimately govern the development of such a velocity dispersion in the proton component .
the role of these instabilities in setting / diminishing a proton lorentz factor dispersion represents a new and potentially important venue for the study of plasma instabilities .
significant dispersion in the proton velocities translates into _ fewer _ protons attaining the highest lorentz factors .
this is tantamount to a reduction in the total energy required to attain a given lorentz factor for the highest energy protons .
as well , a proton component dispersion can have consequences for the electromagnetic and neutrino signature of grbs . |
the modern paradigm of hierarchical galaxy formation is rooted in the notion that large stellar systems are formed through the mergers of smaller ones .
understanding the creation and evolution of a particular galaxy requires a knowledge of the merger history of dark halos and the accompanying gas dynamics .
different environments necessarily lead to differing merger histories , resulting in differences in the observable properties of each galaxy .
in particular , denser environments lead to more frequent mergers , which cause significant structural changes to galaxies .
also , gravitational harassment among galaxies in dense regions can affect the morphological properties of galaxies . within the context of the current paradigm of galaxy formation , it is possible to create computer simulations that begin with some matter distribution and
then follow the creation and evolution of structure ( e.g. see @xcite ) .
observational astronomy , by contrast , is unable to follow the evolution of a particular galaxy .
however , statistical correlations between properties such as galaxy color , luminosity , kinematics , and shape have been found .
attempts to understand how such trends are related to the underlying physics of galaxy evolution are a major focus of extragalactic astronomy .
the purpose of this study is to examine both new and well known trends that relate a galaxy s structure to its environment through the use of observable quantities
. historically , @xcite classified galaxies by their visual appearance on photographic plates .
elliptical galaxies have smooth elliptical isophotes ; spiral galaxies have spiral arms that wind outward from a central bulge or bar . by measuring the surface brightness @xmath7 along the major axis of a galaxy s image
, it was noticed that bright ( @xmath8 ) ellipticals have light profiles that are well fit by log@xmath9 @xmath10 @xmath11r@xmath12 @xcite . on the other hand , the azimuthally
averaged surface brightness profiles of spiral galaxies were discovered to have log@xmath9 @xmath10
@xmath11r @xcite .
such observations reveal a correlation of surface brightness profile with hubble type . furthermore , elliptical galaxies are slowly rotating systems characterized by triaxial shapes ; in contrast , spiral galaxies are rapidly rotating flattened structures .
it has also been found that the average color of galaxies differs for the different galaxy types ; ellipticals tend to be redder , while spirals are bluer .
such correlations with hubble type exist for a host of galaxy properties .
there are well known trends relating morphological type to the local density of galaxies @xcite . in general , bright elliptical galaxies
are preferentially found in regions of higher galaxy number density ; on the other hand , spirals account for the majority of bright galaxies in the field .
it is worth noting , however , that such studies have been focused on galaxies with @xmath13 .
new , large galaxy surveys open up the possibility of exploring such trends in different magnitude bands . images of galaxies are a projection of the true shape .
most studies of the axis ratio of galaxies have concentrated on understanding what the distribution of apparent axis ratios tells us about the intrinsic axis ratios ( e.g. @xcite ) .
such information is a necessary ingredient for , among other things , understanding the kinematic properties of galaxies .
generally , the distribution of apparent axis ratios for thin , nearly circular disks is an approximately uniform function , giving an average axis ratio of @xmath14 . for elliptical galaxies , which tend to be triaxial @xcite ,
the distribution of apparent axis ratios is a steeply peaked function with an average value of @xmath15 .
it is also found that the average axis ratio for ellipticals depends on luminosity , with brighter ellipticals being rounder @xcite .
as previously mentioned , bright ellipticals tend to be found in high density environments .
this suggests that the eventful merger histories of these systems lead to systematic change in average axis ratio .
one under - studied area , related to galaxy shape , is the dependence of axis ratio on environment .
@xcite briefly comment on this dependence for elliptical galaxies .
however , a continued investigation of this field will help complete understanding of the kinematical properties of galaxies . in this paper , we use photometric data from the sloan digital sky survey data release 4 ( sdss dr4 ) @xcite to study the correlations of two structural properties of galaxies with environment ; namely , surface brightness profile and apparent axis ratio .
the sdss provides an unprecedentedly large data set with a uniform data reduction scheme , allowing for a consistent study .
sdss data has already been used to quantify correlations between many galaxy properties @xcite . in [ sec - data ] of this paper
, we describe the sdss , our morphological classification scheme , and the methods by which we determine the apparent axis ratio of galaxies . in
[ sec - prof ] , we present our results on the surface brightness profile type environment relationship .
we make two extensions to previous studies .
first , we use galaxies in the relatively broad absolute magnitude range @xmath16 $ ] .
second , we subdivide the data into narrow magnitude bins to determine the luminosity dependence of the profile type environment relation . in [ sec - axis ] , we describe the results of our study of the apparent axis ratio environment correlation .
motivation for this research was initiated by a prior analysis of galaxy axis ratios in the sdss data @xcite , where it was proposed that axis ratio could be correlated with environment . in [ sec - disc ]
we place our results in the context of the current understanding of galaxy formation and evolution .
the sloan digital sky survey @xcite will , when complete , provide a map of nearly one - fourth of the celestial sphere . a ccd mosaic
camera @xcite images the sky in five photometric bands ( @xmath17 ; @xcite ) .
the follow - up spectroscopic survey provides spectra for galaxies to a limiting petrosian magnitude @xmath18 .
the sloan digital sky survey data release 4 , released to the astronomical community in 2004 october , covers a photometric area of @xmath19 and a spectroscopic area of @xmath20 ( @xcite ; see also @xcite , @xcite , @xcite , and @xcite ) .
the sdss dr4 data processing pipeline provides a morphological star / galaxy separation , with extended objects being classified as ` galaxies ' and unresolved objects being classified as ` stars ' . for each galaxy , in each photometric band , two models are fitted to the two - dimensional galaxy image .
the first model has a de vaucouleurs surface profile @xcite : @xmath21 \right ) \ , \ ] ] which is truncated beyond @xmath22 to go smoothly to zero at @xmath23 .
the second model has an exponential profile : @xmath24 \right ) \ , \ ] ] which is truncated beyond @xmath25 to go smoothly to zero at @xmath26 . for each model , the apparent axis ratio @xmath27 and the phase angle @xmath28
are assumed to be constant with radius .
the parameters @xmath27 , @xmath28 , @xmath29 , and @xmath30 are varied to give the best @xmath31 fit to the galaxy image , after convolution with a double - gaussian fit to the point spread function
. a further fit to each galaxy is made by taking the best de vaucouleurs model and the best exponential model , and finding the linear combination of the two that gives a new best fit .
the fraction of the total flux contributed by the de vaucouleurs component is the parameter fracdev , which is constrained to lie in the interval @xmath32 $ ] .
the fracdev parameter is functionally equivalent to the @xcite index @xmath33 in the interval @xmath34 @xcite .
if a galaxy has a srsic index @xmath33 , then @xmath35 corresponds to fracdev = 0 , @xmath36 corresponds to fracdev = 1 , and the dependence of fracdev upon @xmath33 is monotonic in the interval @xmath34 .
following @xcite , we call galaxies with @xmath37 fracdev @xmath38 ( @xmath39 ) _ ex _ galaxies .
galaxies with @xmath40 fracdev @xmath41 ( @xmath42 ) are labeled _ ex / de _ galaxies .
galaxies with @xmath43 fracdev @xmath44 ( @xmath45 ) are labeled _ de / ex _ galaxies .
finally , galaxies with fracdev @xmath46 ( @xmath47 ) are labeled _ de _ galaxies .
the sdss dr4 databases provide many different measures of the apparent axis ratio @xmath48 of each galaxy in each of the five photometric bands . in this paper
, we use the @xmath49 band data , at an effective wavelength of 6165 ; this is the band in which the star / galaxy classification is made .
a useful measure of the apparent shape in the outer regions of galaxies is the axis ratio of the 25 mag arcsec@xmath50 isophote .
the sdss dr4 data pipeline finds the best fitting ellipse to the 25 mag arcsec@xmath50 isophote in each band ; the semimajor axis and semiminor axis of this isophotal ellipse are @xmath51 and @xmath52 .
the isophotal axis ratio @xmath53 then provides a measure of the apparent galaxy shape at a few times the effective radius . for galaxies in our sample with fracdev = 1 , @xmath54 ; for galaxies with fracdev = 0 , @xmath55
. the second measure of the apparent axis ratio that we use is @xmath56 , the axis ratio determined by the use of adaptive moments of the galaxy s surface brightness .
the method of adaptive moments determines the @xmath33th order moments of a galaxy image , using an elliptical weight function whose shape matches that of the image @xcite .
the sdss dr4 adaptive moments use a weight function @xmath57 that is a gaussian matched to the size and ellipticity of the galaxy image @xmath58 . the adaptive first order moments ,
@xmath59 tell us the ` center of light ' of the galaxy s image . with this knowledge , we can compute the adaptive second order moments : @xmath60 and so forth .
the sdss dr4 provides for each image the values of the parameters @xmath61 , @xmath62 , and @xmath63 .
the adaptive second moments can be converted into an axis ratio using the relation @xmath64 where @xmath65 .
the adaptive moments axis ratio @xmath56 is not corrected for the effects of seeing .
the sdss dr4 also provides the fourth order adaptive moments of the galaxy image , and the adaptive moments @xmath66 , @xmath67 , and @xmath68 of the point spread function at the galaxy s location .
these moments can be used to correct for the smearing and shearing due to seeing ; such corrections are essential for studying the small shape changes resulting from weak lensing @xcite . however ,
when we examine the apparent axis ratios of galaxies in section [ sec - axis ] , we will only look at well - resolved galaxies ( @xmath69 ) , for which the seeing corrections are negligible .
we note that , at most , weak lensing shears the shape of a galaxy at the one percent level @xcite .
such minute changes in the shape do not significantly affect our results .
our complete sample of galaxies consists of those objects in the sdss dr4 spectroscopic sample which are flagged as galaxies , which have @xmath70 , and which have spectroscopic redshifts @xmath71 and a redshift confidence parameter @xmath72 .
the absolute magnitude @xmath73 of each galaxy is computed from its sdss cmodel apparent magnitude , assuming a uniform hubble flow with @xmath74 , @xmath75 , and @xmath76 .
the complete sample contains @xmath77 galaxies with a median redshift of @xmath78 .
the morphology of each galaxy in the sample is quantified by the parameter fracdev , and by the two measures of the apparent axis ratio : @xmath79 , which gives the shape in the outer region , and @xmath56 , which is more strongly weighted toward the inner region of the galaxy .
a galaxy with absolute magnitude @xmath73 will enter the sdss spectroscopic survey if its luminosity distance is less than @xmath80 mpc , where the limiting apparent magnitude of the survey is @xmath81 . in our study of the relationship between surface brightness profile type and environment ,
we create six volume - limited samples with absolute magnitude cut - offs @xmath82 = @xmath83 , @xmath84 , @xmath85 , @xmath86 , @xmath87 , and @xmath88 .
the maximum luminosity distance in the volume - limited samples ranges from @xmath89 for the @xmath90 sample ( corresponding to @xmath91 ) , to @xmath92 , for the @xmath93 sample ( corresponding to @xmath94 ) . in each sample
we choose as our ` target galaxies ' those galaxies with @xmath95 $ ] .
the target galaxies are those galaxies for which we will determine the profile type environment dependence .
the total number of target galaxies in each volume - limited sample , and the fraction of target galaxies corresponding to each of our four morphological types , is presented in table [ tab : profile ] .
table [ tab : profile ] also lists , under the heading ` cylinder width ' , the total number of cylinders ( see [ sec - env ] ) that do not intersect the border of the sdss spectrographic survey , which is the same as the number of galaxies that we keep in our analysis ( see [ sec - env ] ) .
the galaxies in each sample are subject to the quality cuts described in [ sec - data ] .
the environment of each target galaxy is determined by a count of neighboring galaxies .
we adhere to the following in determining this count .
a cylinder is created around each target galaxy in redshift space , with the long axis of the cylinder lying along the line of sight .
the cylinder stretches 6 @xmath96 mpc ( with @xmath97 ) in either radial direction from the target galaxy ( i.e. 12 @xmath96 mpc in total ) .
the radius of the cylinder is one of the following : 0.5 @xmath96 mpc , 2 @xmath96 mpc , 8 @xmath96 mpc . counting the number of galaxies with @xmath98 in this cylinder defines our density parameter .
we note that in dense environments , this method undercounts the true number of galaxies in each cylinder .
this is because the spectroscopic fibers on each sdss plate can not be placed more closely than @xmath99 .
this causes @xmath100 of galaxies ( of the total ) to be lost @xcite , undercounting galaxies in high density regions .
fibers are assigned to galaxies with no correlations to galaxy properties @xcite .
hence , there should be no significant effect to our results .
for example , if @xmath101 of each galaxy type in the highest density bin were moved to the second - highest density bin , the relative fraction would not change .
in general , this is the largest source of galaxy loss , as the completeness of the spectroscopic survey is @xmath102 @xcite .
the sdss takes spectra using circular plates whose diameter subtends an angle of @xmath103 on the sky .
the sdss dr4 spectroscopic survey does not completely match the area covered by the dr4 photometric survey .
figure [ fig:1 ] shows the boundary of the spectroscopic survey .
if a target galaxy is within one cylinder width of the border , it is dismissed from our analysis .
in addition , if a galaxy is within one - half cylinder length , @xmath104 , of the maximum luminosity distance , it is dismissed from our analysis . in determining the environment for target galaxies in each magnitude bin we count _ all _ of the galaxies brighter than the magnitude cut - off @xmath82 . in plotting our results , we group together at least 100 galaxies in each bin .
we take the weighted mean of the number of galaxies in each cylinder that falls into a particular bin .
this defines the local density , or environment for all galaxies in each bin .
the highest density bin may have as few as 50 galaxies ( the only excpetion to this rule is the case with @xmath105 and @xmath106 mpc cylinders where the number in each bin is half that just described ) .
we use the same density bins for all profile types so that we may determine the relative fraction of each .
the bins are determined by the binning of the _ ex _ galaxies subject to the previously given conditions .
we changed the possible binning for this analysis , varying it by profile type , and found no substantial change in the results .
many studies ( e.g. @xcite ) have shown that as the density of galaxies in a region increases , the relative fraction of elliptical galaxies to spiral galaxies increases .
instead of using the hubble classification scheme , we classify galaxies based on their surface brightness profile type , using four classes ranging from the highly concentrated _ de _ galaxies to the least concentrated _ ex _ galaxies .
our largest survey volume extends to the distance at which a galaxy with @xmath107 has @xmath81 , corresponding to @xmath108 . @xcite
have shown that in the redshift interval @xmath109 , there is little evolution in the fraction of galaxies with @xmath110 ( corresponding to our classes _ de _ and _ de / ex _ ) . since all our other magnitude
cut - offs correspond to lower maximum redshifts , we do not expect significant redshift evolution effects in our samples .
figures [ fig:2 ] through [ fig:4 ] display our findings on the trends of galaxy profile with environment .
the plots are organized by magnitude cut - off and cylinder radius .
the _ ex _ galaxies are represented by green solid lines and solid circles ; the _ ex / de _ galaxies , by blue dotted lines and solid triangles ; the _ de / ex _ galaxies , by red dashed lines and open circles ; the _ de _ galaxies , by magenta long - dashed lines and open triangles .
the plotted errors are the expected errors in the mean , assuming poisson statistics .
table [ tab : profile ] lists the number of galaxies in each of our volume - limited surveys .
it also lists the fraction of each profile type .
finally , it lists the total number of galaxies that survive all of our quality cuts ( see [ sec - data ] and [ sec - env ] ) . we summarize the results as follows : \1 .
figure [ fig:2 ] displays the results for cylinders of width @xmath111 mpc .
there are four primary results .
first , galaxies with @xmath112 are found to have mostly exponential light profiles in all environments .
second , galaxies with @xmath113 have mostly de vaucouleurs light profiles in all environments .
third , for galaxies with @xmath3 , as local density increases , the fraction of de vaucouleurs galaxies increases monotonically with increasing local density , while the fraction of exponentials decreases .
fourth , there is a crossover of predominant profile type between high and low density environments only for galaxies with @xmath114 .
plotting results for so many magnitude bins clearly shows how the change in dominant profile type depends not only on environment , but also on absolute magnitude .
@xcite examined the morphology environment relation for galaxies with @xmath115 ( assuming @xmath97 ) ; for a typical color @xmath116 , this corresponds roughly to @xmath117 .
figures [ fig:3 ] and [ fig:4 ] display the effect of increasing the cylinder size .
the main result is that trends are washed out on scales larger than about 2 @xmath96 mpc .
there is no plot for the magnitude cut - offs @xmath83 , @xmath84 , and @xmath85 for cylinders that are @xmath118 mpc wide ; the volume of space probed is small enough that once all of the cylinders that intersect with the border are thrown out , there are too few galaxies left over to make the plot statistically interesting .
this section of the paper addresses how the average shape of galaxies , as quantified by their apparent axis ratio , depends on their environment , for a fixed luminosity and profile type .
our analysis of this topic is described below .
we first highlight any differences in the methodology from the preceding section . unless otherwise mentioned
, our methods here remain the same as those described in [ sec - prof ] .
one idiosyncrasy discovered in @xcite is that for fixed luminosity the average axis ratio is not always a monotonic function of srsic index .
subdividing all galaxies into only two groups , one with all galaxies with fracdev @xmath119 0.5 , or @xmath120 , and another with the remaining galaxies , would average over this interesting behavior .
this provides us with further motivation for maintaining our galaxy classification scheme , where we place galaxies in one of the four groups : _ ex _ , _ ex / de _ , _ de / ex _ and _ de _ ( see [ sec - data ] ) .
the survey volume for this analysis is similar to that described in [ sec - vol ] .
an important difference is that we place a stricter cut on the point spread function ( see [ sec - data ] ) and in the process lose many de vaucouleurs galaxies . in @xcite
it was noted that there is a luminosity axis ratio trend . in the range
@xmath121 $ ] the trend is approximately flat . in the range
@xmath122 $ ] the axis ratio increases with increasing luminosity . for galaxies brighter than @xmath88 , the axis ratio seems to level off . in principle
, this correlation could affect our results .
this motivates us to keep luminosity bins as narrow as possible , so as to minimize any effect of this trend within a single bin . on the other hand ,
statistical claims are only useful for large sample sizes . to keep our subsamples of galaxies large enough , we are motivated to keep luminosity bins as wide as possible . as a compromise
, we make luminosity bins two magnitudes wide .
we start by creating three volume - limited surveys with absolute magnitude cut - offs @xmath123 and @xmath88 . for the volumes corresponding to the first two cut - offs , we determine the environment ( see [ sec - env ] ) for target galaxies that have absolute magnitude @xmath124 $ ] , while for the brightest cut - off we find the environment for all galaxies with @xmath6 $ ] .
there are only sixteen galaxies that are brighter than @xmath125 in absolute magnitude in our volume ; therefore , their inclusion does not significantly widen the effective magnitude range of the bin .
for example , in the volume corresponding to the @xmath126 cut - off we find the environment for all galaxies with @xmath127 $ ] where the luminosity distance of a galaxy is no larger than @xmath128 143 mpc , corresponding to the maximum luminosity distance at which a galaxy of absolute magnitude @xmath126 could be observed .
the total number of galaxies in each volume , and the percentage of each profile type is presented in table [ tab : shape ] .
table [ tab : shape ] also lists , under the heading ` cylinder width ' , the total number of cylinders ( see [ sec - env ] ) that do not intersect the border of the sdss spectrographic survey , which is the same as the number of target galaxies that we keep in our analysis .
the galaxies in each sample are subject to quality cuts which are described in [ sec - data ] .
the binning procedure is described in [ sec - env ] .
figures [ fig:5 ] through [ fig:7 ] show the correlations of apparent axis ratio with environment , labeled by profile type .
tables [ tab : shape18 ] through [ tab : shape22 ] contain all related statistical information .
we plot a best fit straight line through the binned data that minimizes the chi - square .
the value of the reduced chi - square for each fit is printed in the tables .
the error bars show the error in the mean assuming poisson statistics .
the choice of colors , line type and point type are the same as previously described in [ sec - profresults ] .
since the average axis ratio depends on the average luminosity , the normalization of each best fit line , for the axis ratio environment trend , is set by the average luminosity for that profile type . also , since the luminosity functions for each profile type are different in the different volume - limited samples , the average luminosity of each profile type is slightly different .
for example , in the @xmath129 $ ] range , the _ ex _ galaxies have an average magnitude of @xmath130 , while the _ de _ galaxies have @xmath131 . from the luminosity axis ratio trend , it is expected that the _ de _ galaxies would be rounder .
this is reflected in figure [ fig:6 ] , say , where the trend for the _ de _ galaxies lies above the trend for the _ ex _ galaxies .
therefore , while it is useful to compare the difference in the axis ratio as determined by the 25 mag arcsec@xmath50 isophote versus the axis ratio as determined by the adaptive moments technique , comparisons of axis ratios between galaxy profile types need to be made with the caveat that they contain different luminosity information . in the absolute
r _ band magnitude range @xmath122 $ ] , it has been shown @xcite that , for all profile types , there is a trend toward roundness as galaxies become brighter . in particular , the average axis ratio may increase by as much as approximately 0.05 when the magnitude decreases by unity @xcite .
however , we find that the luminosity in each local density bin is equal to within one tenth of a magnitude confirming that any trend we see are not due to luminosity .
specifically , for @xmath5 $ ] , the average magnitude for the _ ex _ galaxies is @xmath132 , for the _ ex / de _ galaxies it is @xmath133 , for the _ de / ex _ galaxies it is @xmath134 and for the _ de _ galaxies it is @xmath135 . in other magnitude ranges trends of mean axis ratio with luminosity are weaker and we also find that magnitude does not vary significantly from bin to bin .
table [ tab : shape ] lists the number of galaxies in each of our volume - limited samples .
also , it lists the fraction of galaxies of each profile type .
tables [ tab : shape18 ] through [ tab : shape22 ] list all statistical information related to each profile type and cylinder size : the chi - square of the best fit line ; whether the axis ratio is increasing ( ` + ' ) or decreasing ( ` - ' ) with increasing local density ; our statistic labeled ` significance ' is the ratio of the slope to its standard deviation , characterizing if the slope is significantly different than zero .
we summarize the results .
\1 . in the magnitude range
@xmath121 $ ] the data show one statistically important trend at the @xmath136 level .
the _ ex / de _ galaxies get rounder as the local density of galaxies increases , and the axis ratio , as measured by the axis ratio determined by the moments , for the _ de / ex _ galaxies increases with density .
other profile types in this magnitude range tend to become rounder at higher local density , but at a lower level of statistical significance .
plots are only created for cylinders of width 0.5 @xmath96 mpc and 2 @xmath96 mpc .
when cylinders of width 8 @xmath96 mpc are considered , too many are near the border leaving a statistically uninteresting sample .
\2 . in the magnitude range
@xmath122 $ ] we find the strongest trends between axis ratio and environment . generally , the _ de _ and _ de / ex _ galaxies get rounder with increasing local density .
this trend is more pronounced in the inner regions of these galaxies . on the other hand ,
the _ ex / de _ galaxies tend toward greater flattening as the local density of galaxies increases , with a similar trend , though less statistically significant , observed for the _ ex _ galaxies .
trends of axis ratio with environment are only significant on scales less than or equal to 2 @xmath96 mpc . \3 . for our sample of the brightest galaxies , in the magnitude range
@xmath137 $ ] , we find that trends are more pronounced in the inner regions than the outer regions ; trends in the shape , determined by the moments of the light distribution , are no longer present when we use the axis ratio determined by the 25 mag arcsec@xmath50 isophote .
this is especially true of the _ de _ galaxies .
statistically , they show a strong positive trend in their axis ratio as determined by the moments of their light distribution
. however , the average axis ratio for these galaxies derived from the 25 mag arcsec@xmath50 isophote is essentially constant with environment .
the novel aspect of our study of the surface brightness profile type
environment relationship is our use of a large magnitude range subsequently divided into many volume - limited samples . averaging over a wide magnitude range
blurs out some of the observed trends , resulting in an inability to distinguish them .
using many volume - limited samples allows us to determine that , at absolute _
r _ band magnitudes around @xmath83 , galaxies are almost all described by purely exponential light profiles at all densities , with no strong trends in the fraction of different profile types with environment . for absolute magnitudes around @xmath85 and @xmath86
there is a crossover from low densities to high densities ; galaxies with exponential profiles are more common at low densities , while at high local densities galaxies are most likely to be well described by a de vaucouleurs profile .
finally , for bright galaxies with @xmath138 , most galaxies have de vaucouleurs profiles .
these results fit well into the hierarchical structure formation picture . generally , the result of mergers is to increase entropy . in spiral galaxies , stellar orbits in the disk
are nearly circular and all of the disk stars travel in the same direction , corresponding to a low overall entropy . on the other hand , orbits of stars in elliptical galaxies
are unsystematic , giving rise to a high entropy system .
moreover , each of these two systems have different light profiles .
spiral galaxies tend to be less concentrated and have exponential profiles ; ellipticals have more of their light concentrated near their center and most bright ellipticals have de vaucouleurs profiles .
we find that the dimmer galaxies , which have presumably experienced fewer mergers , are mostly well described by exponential profiles . for galaxies in the middle of our luminosity range , the environment
is strongly correlated with profile type .
galaxies in high density environments experience , on average , more mergers than those in low density environments .
the brightest galaxies , and presumably the ones that have been created through many mergers , are predominantly de vaucouleurs .
the dependences of profile type and apparent axis ratio on environment show statistically interesting trends on scales up to 2 @xmath96 mpc .
this is consistent with other studies of the correlations of galactic properties with environment @xcite .
the properties of a galaxy are generally affected by the presence of other galaxies in the same group or cluster , but not by galaxies in neighboring clusters .
@xcite use the apm bright galaxy survey @xcite and study the axis ratio - environment relationship for elliptical galaxies .
they arrive at the conclusion that `` local density is not an important factor in determining the flattening of ellipticals '' .
the lack of redshift information constrains @xcite to estimate the local density in projection .
this dilutes the signal by inevitably counting galaxies that have similar celestial coordinates as neighbors , while in reality they are separated by a large distance in space .
furthermore , they use a fixed angular scale on the sky within which they count the number of neighbors . for nearby galaxies , this scale corresponds to a larger physical distance than for distant galaxies .
our plots ( figures [ fig:5 ] through [ fig:7 ] ) show that trends tend to get washed out as cylinder size increases .
they are effectively mixing different cylinder sizes and in the process washing out any trend . within this angular area , @xcite
label their galaxies as ` high - density ' if there are more than two neighboring galaxies , and ` low - density ' if there are two or less .
this cut divides their sample into an approximately equal number of high and low density galaxies .
they claim that there is no statistical difference in the observed distribution of their high and low density ellipticals .
we performed a komolgorov - smirnov test on the distribution of axis ratios derived from the moments , for the _ de _ galaxies with @xmath139 $ ] in @xmath111 mpc cylinders .
the ks probability is minimized when we place the cut between low and high densities at eight nearest neighbors ; this results in a probability of @xmath140 . instead of dividing our sample into two equal parts
, this cut corresponds to a low / high density split of @xmath141 .
thus @xcite did not see the trend for two reasons : density in projection is a poor measure , and the difference in shape is most significant in very high density environments .
one open problem is whether or not field ellipticals are created and evolve in a fundamentally different way from those found in high density environments .
@xcite have shown that the massive early type galaxies in low density environments appear to have younger stellar populations and be slightly more metal rich than the ones in high density environments . @xcite
have shown that elliptical galaxies with younger stellar populations tend to be flatter in their inner regions than ones with older stellar populations .
one possible explanation for such behavior is that the central black hole in older ellipticals has had a longer time to randomize the orbits of stars near the center . in light of our result that the inner regions of _ de _ galaxies tend to be more flattened in low density versus high density regions , these two previous studies may point to a more fundamental cause for our observed trends .
however , using an age environment correlation to explain our results is no more fundamental than simply saying that axis ratio and environment are correlated . instead , the missing link is in understanding what processes form centrally concentrated galaxies such as massive ellipticals .
@xcite have used numerical simulations to show that it is possible to form elliptical systems from the merger of a large disk with multiple smaller disks . when a large disk galaxy is merged with one or two small disks ( with the mass ratio of the large disk to the small disk being 1:7 ) it is found that a galaxy of an intermediate type is created .
such a galaxy may correspond to the _ ex / de _ and _ de / ex _ galaxies observed in the sdss .
a well known fact is that the merger of two spirals of roughly equal mass leads to the creation of an elliptical ( e.g. @xcite ; @xcite ; @xcite ) .
however , it has not been shown that the most common way of creating an elliptical is through the merger of two spirals .
other numerical studies @xcite claim to find evidence that there exist elliptical galaxies that appear not to have been formed by a major merger of spirals .
such studies show that different merger histories can lead to the creation of a galaxy with a de vaucouleurs profile .
moreover , different merger histories lead to galaxies with different intrinsic axis ratios @xcite . if it was determined conclusively that certain types of mergers are favored in high versus low density regions
, it would be possible to deduce the average axis ratio of galaxies in each environment .
conversely , if different merger histories produce characteristically different axis ratios , it may be possible to work backwards and deduce which merger histories are more common in a given environment . in any case , the axis ratio of a galaxy , in its outer regions , could also depend on harassment from neighboring galaxies .
however , without knowing which merger histories are favored in different environments , and assuming that the observed environment is similar to the pre - merger environment , it is not possible to determine the ` initial ' axis ratio of a galaxy and thus separate the the effect of harassment from that of the merger history . one intriguing fact uncovered is that the _ ex _ and _ ex / de _ galaxies with @xmath142 are flatter , on average , in denser environments .
these galaxies are primarily flattened disk galaxies , with a past free from major mergers . in a high density environment ,
a galaxy may be harassed by numerous encounters in the past , or may have a tidal tail from a recent close encounter .
these processes may distort the apparent shape of galaxies in dense regions .
however , we visually inspected all the _ ex _ and _ ex / de _ galaxies that have 25 or more neighboring galaxies in a cylinder of radius @xmath143 mpc .
there were 35 such galaxies , of which only one appeared to have a tidal tail and only one was a possible recent merger remnant .
we also inspected 35 _ ex _ and _ ex / de _ galaxies with at most one neighbor ; none of them had obvious tidal tails or asymmetric distortions .
almost all galaxies in our visual samples were separated from their neighbors by a distance significantly larger than the galaxy s visual radius .
if the greater average flattening at higher density is due to tidal effects , then the tidal distortions must be subtle , rather than taking the form of strong tidal tails .
the sdss has recently been used to study the correlation of many galactic properties with each other .
exhaustive studies , such as those produced by @xcite , @xcite and @xcite , arrive at the conclusion that luminosity and color are the fundamental galactic properties from which all correlations of other properties with environment are derived .
such claims are tested in and are in reasonable agreement with numerical simulations as described , for example , in @xcite .
however , we find a trend , namely the apparent axis ratio local galaxy density correlation , for galaxies of the same _ r _ band luminosity and surface brightness profile type , that is not due to a correlation of luminosity or color with environment .
funding for the creation and distribution of the sdss archive has been provided by the alfred p. sloan foundation , the participating institutions , the national aeronautics and space administration , the national science foundation , the u.s .
department of energy , the japanese monbukagakusho , and the max planck society .
the sdss website is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium ( arc ) for the participating institutions .
the participating institutions are the university of chicago , fermilab , the institute for advanced study , the japan participation group , the johns hopkins university , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , university of pittsburgh , princeton university , the united states naval observatory , and the university of washington .
lrccccrrr @xmath145 & 1610 & 0.56 & 0.30 & 0.11 & 0.03 & 1140 & 253 & + @xmath146 & 5915 & 0.50 & 0.31 & 0.14 & 0.05 & 5267 & 2410 & + @xmath147 & 15580 & 0.41 & 0.30 & 0.17 & 0.12 & 14698 & 9192 & + @xmath148 & 46755 & 0.27 & 0.25 & 0.20 & 0.28 & 45549 & 36208 & 8856 + @xmath149 & 81037 & 0.12 & 0.20 & 0.22 & 0.46 & 79988 & 70560 & 26975 + @xmath150 & 57900 & 0.04 & 0.14 & 0.21 & 0.61 & 57482 & 54219 & 30614 + lrccccrrr @xmath151 & 8669 & 0.53 & 0.35 & 0.11 & 0.01 & 7691 & 3459 & + @xmath152 & 39964 & 0.37 & 0.39 & 0.20 & 0.04 & 38901 & 30536 & 7180
+ @xmath153 & 20211 & 0.12 & 0.37 & 0.37 & 0.14 & 20055 & 18731 & 10068 + cccccccc _ ex _ & 0.5 & 0.28 & + & 1.1 & 0.38 & + & 0.96 + _ ex / de _ & 0.5 & 0.50 & + & 2.8 & 0.43 & + & 2.9 + _ de / ex _ & 0.5 & 0.82 & + & 1.2 & 0.35 & + & 2.7 + _ ex _ & 2 & 0.49 & - & 0.36 & 0.41 & + & 0.11 + _ ex / de _ & 2 & 0.40 & + & 2.4 & 0.67 & + & 2.6 + _ de / ex _ & 2 & 0.22 & + & 0.58 & 0.22 & + & 0.86 + cccccccc _ ex _ & 0.5 & 1.2 & - & 2.5 & 0.83 & - & 2.2 + _ ex / de _ & 0.5 & 0.66 & - & 4.2 & 0.74 & - & 3.7 + _ de / ex _ & 0.5 & 1.9 & + & 0.23 & 2.0 & + & 2.4 + _ de _ & 0.5 & 0.59 & + & 3.6 & 0.34 & + & 4.9 + _ ex _ & 2 & 1.4 & - & 3.6 & 1.4 & - & 3.4 + _ ex / de _ & 2 & 1.3 & - & 4.0 & 1.2 & - & 4.0 + _ de / ex _ & 2 & 1.1 & - & 0.52 & 0.93 & + & 1.2 + _ de _ & 2 & 1.3 & + & 1.1 & 0.93 & + & 1.4 + _ ex _ & 8 & 0.75 & - & 1.7 & 0.76 & - & 1.6 + _ ex / de _ & 8 & 1.0 & - & 1.3 & 1.2 & - & 2.1 + _ de / ex _ & 8 & 0.61 & + & 0.88 & 0.77 & - & 0.37 + cccccccc _ ex _ & 0.5 & 0.012 & - & 1.3 & 0.13 & - & 1.6 + _ ex / de _ & 0.5 & 0.60 & - & 3.6 & 0.66 & - & 3.1 + _ de / ex _ & 0.5 & 0.29 & + & 0.23 & 0.45 & + & 0.67 + _ de _ & 0.5 & 0.0010 & + & @xmath154 & 0.55 & + & 5.5 + _ ex _ & 2 & 1.0 & - & 1.4 & 0.99 & - & 0.96 + _ ex / de _ & 2 & 0.75 & - & 3.5 & 1.3 & - & 3.3 + _ de / ex _ & 2 & 0.86 & - & 2.1 & 0.65 & - & 0.83 + _ de _ & 2 & 0.35 & - & 0.86 & 0.45 & + & 5.0 + _ ex _ & 8 & 1.0 & - & 1.1 & 0.57 & - & 1.1 + _ ex / de _ & 8 & 0.91 & - & 1.5 & 0.56 & - & 1.2 + _ de / ex _ & 8 & 1.2 & + & 0.40 & 0.85 & + & 0.25 + _ de _ & 8 & 0.38 & + & 0.35 & 1.8 & + & 2.4 + | using a sample of galaxies from the sloan digital sky survey data ( sdss ) data release 4 , we study the trends relating surface brightness profile type and apparent axis ratio to the local galaxy environment .
we use the sdss parameter ` fracdev ' to quantify the profile type ( fracdev = 1 for a pure de vaucouleurs profile ; fracdev = 0 for a pure exponential profile ) .
we find that galaxies with @xmath0 are mostly described by exponential profiles in all environments .
galaxies with @xmath1 mainly have exponential profiles in low density environments and de vaucouleurs profiles in high density environments .
the most luminous galaxies , with @xmath2 , are mostly described by de vaucouleurs profiles in all environments . for galaxies with @xmath3 ,
the fraction of de vaucouleurs galaxies is a monotonically increasing function of local density , while the fraction of exponential galaxies is monotonically decreasing . for a fixed surface brightness profile type
, apparent axis ratio is frequently correlated with environment .
as the local density of galaxies increases , we find that for @xmath4 $ ] , galaxies of all profile types become slightly rounder , on average ; for @xmath5 $ ] , galaxies with mostly exponential profiles tend to become flatter , while galaxies with de vaucouleurs profiles tend to become rounder ; for @xmath6 $ ] , galaxies with mostly exponential profiles become flatter , while the de vaucouleurs galaxies become rounder in their inner regions , yet exhibit no change in their outer regions .
we comment on how the observed trends relate to the merger history of galaxies . |
in the cold dark matter ( cdm ) paradigm for galaxy formation ( e.g. * ? ? ?
* ) , galaxies are embedded in extended dark matter distributions with a specific and ( nearly ) universal radial profile ( e.g. * ? ? ?
* ; * ? ? ?
measurements of rotation curves from neutral hydrogen ( ) observations in the outer parts of late - type galaxies have provided evidence for the presence of dark matter in these systems already more than two decades ago ( e.g. * ? ? ?
* ) . in the outer parts of early - type galaxies ,
however , cold gas is scarce ( but see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , so evidence for a dark matter halo has to come from other tracers , such as kinematics of stars , planetary nebulae and globular clusters ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) or hot x - ray gas ( e.g. * ? ? ?
however , these tracers are not always ( sufficiently ) available , the observations are often challenging , and the interpretation modeling dependent . also in the inner parts of galaxies the amount , shape and profile of dark matter is still poorly known ( e.g. * ? ? ? * ) , even though stellar kinematics are readily available .
a fundamental problem in using stellar kinematics ( as well as other collisionless kinematic tracers ) for this purpose is the mass - anisotropy degeneracy : a change in the measured line - of - sight velocity dispersion can be due to a change in total mass , but also due to a change in velocity anisotropy .
both effects can be disentangled by measuring also the higher - order velocity moments @xcite , but only the inner parts of nearby galaxies are bright enough to obtain the required high - quality kinematic measurements ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . a unique alternative to break the mass - anisotropy degeneracy is provided by gravitational lensing . in case of strong lensing
, the mass of a foreground galaxy bends the light of a distant bright object behind it , resulting in multiple images . from the separation and fluxes of the images
the total mass distribution of the lens galaxy can be inferred directly .
the velocity anisotropy then can be determined from the observed velocity dispersion , without the need for higher - order velocity moments .
@xcite ( @xcite , and references therein ) have applied this approach to several strong lensing systems , of which 0047 - 281 @xcite is the best constrained case , with three radially separated velocity dispersion measurements extending to about the effective radius of the lens galaxy .
they measure the mass within the einstein radius by fitting a singular isothermal ellipsoid to the positions of the quasar images .
this einstein mass is used to set the amplitude of the total ( stellar and dark ) matter distribution , which they assume to be spherical .
the constant stellar mass - to - light ratio @xmath16 determines the contribution of the stars with a fixed radial profile , with the remainder due to dark matter with a single power - law profile with slope @xmath21 .
they then compare the dispersion profile predicted by the spherical jeans equations , for an ad - hoc assumption of the velocity anisotropy @xmath22 , with the observed dispersion measurements .
based on a reasonably constrained @xmath16 and an upper limit on @xmath21 , they conclude that a significant amount of dark matter is present in the inner parts of the lens galaxy , with a slope shallower than the nearly isothermal total mass distribution . an additional ( external ) constraint on @xmath16 ( or @xmath21 ) is needed to go beyond this limit on the dark matter distribution .
even so , their results are limited by too few kinematic constraints ( which leaves the anisotropy degenerate ) , and by the use of a simple spherical dynamical model .
clearly , most lens galaxies are significantly flattened and so can not be well - described by spherical models .
non - spherical models provide a more realistic description of the lens galaxy , but the increase in freedom requires also ( significantly ) more spatially resolved kinematic measurements to constrain them .
only a few of the known strong gravitational lens systems are close enough to obtain such kinematic data .
even then , one can make ( ad - hoc ) assumptions on the velocity distribution , such as velocity isotropy in the meridional plane of an axisymmetric model .
the latter restriction to a distribution function of two ( instead of three ) integrals of motion allows for the recovery of the intrinsic shape and mass distribution in the presence of two - dimensional kinematic data @xcite , possibly even without the additional constraint provided by gravitational lensing ( e.g. * ? ? ?
* ) . in this paper
, we relax the two - integral assumption and find that one can still derive independent determinations of the intrinsic mass distribution from both gravitational lensing and stellar dynamics with two - dimensional kinematic data .
we have observed the gravitational lens system qso2237 + 0305 , well - known as the einstein cross , with the integral - field spectrograph ` gmos ` on the gemini - north telescope .
we fit the extracted stellar velocity and velocity dispersion maps of the inner parts of the lens galaxy at a redshift @xmath3 with axisymmetric dynamical models .
the resulting intrinsic mass distribution agrees well with that inferred from the lens model that fits the four quasar image positions and relative fluxes .
furthermore , we compare this total mass distribution with the stellar mass distribution , based on an independent estimate of the stellar mass - to - light ratio @xmath16 , to place constraints on the dark matter distribution in the inner parts of the lens galaxy . in section [ sec : observations ] , we briefly describe the einstein cross , and we present the photometric and spectroscopic observations . in section [ sec
: analysis ] , we extract the two - dimensional stellar kinematics , describe the dynamical modeling method , and construct lens models . from the latter we infer the total mass distribution , which we compare in section [ sec : results ] with the luminous mass distribution inferred from the observed surface brightness .
we then build axisymmetric dynamical models , and compare the resulting dynamical mass - to - light ratio with that inferred from the lens models , and in turn with an estimate of the stellar mass - to - light ratio . in section [ sec : discconcl ] we discuss our findings and we summarize our conclusions . throughout we adopt the wmap cosmological parameters for the hubble constant , the matter density and the cosmological constant , of respectively @xmath23kms@xmath6mpc@xmath6 , @xmath24 and @xmath25 @xcite .
the einstein cross is the well - known gravitational lens system qso2237 + 0305 ( ra : 22h40m30.3s , dec : @xmath262131 , j2000 ) . in this system , a distant quasar at redshift @xmath27
is lensed by the bulge of the early - type spiral pgc069457 at @xmath28 ( angular diameter distance @xmath29mpc , @xmath30kpc ) , resulting in a cross of four bright images separated by about 1.8 .
the einstein cross has long been the closest strong gravitational lens system known , and has been very well studied since its discovery by @xcite .
there is a wealth of ground- and space - based imaging data at all wavelengths ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the resulting precise measurements of the positions and relative fluxes of the quasar images can be used to construct a detailed lens model .
in contrast , kinematic data of the lens galaxy is very scarce , with only one measured central stellar velocity dispersion @xcite and two rotation curve measurements in the very outer parts @xcite .
there are several previous integral - field studies of the einstein cross : ` tiger ` : ; ` integral ` : @xcite ; ` cirpass ` : @xcite .
however , none of these studies were concerned with the stellar kinematics of the lens galaxy , but instead investigated the quasar spectra .
we use two wfpc2 images retrieved from the hst - archive ( f555w @xmath31-band image , 1600 seconds , pi : westphal , and f814w @xmath11-band image , 120 seconds , pi : kochanek ; see left and middle panels of fig .
[ fig : mgesurf ] ) to determine the luminosity density of the lens galaxy .
we correct the @xmath11-band image for extinction following @xcite , and we convert to solar units using the wfpc2 calibration of @xcite , while assuming an absolute @xmath11-band magnitude for the sun of 4.08 mag ( table 2 of * ? ? ?
* ) . from a de vaucouleurs @xmath32 profile fit to the @xmath11-band photometry in the inner bulge - dominated region , we obtain an effective radius @xmath2 , which is consistent with previous measurements ( e.g. * ? ? ?
* ) . for the construction of the lens model
, we use the accurate positions of the quasar images from the website of the ` castles ` survey based on _ hubble space telescope _ ( hst ) imaging .
although also optical flux ratios are given on this website , we use the radio fluxes provided by @xcite , because they are in general ( much ) less affected by differential extinction or microlensing .
observations of the einstein cross lens system were carried out using the integral - field unit of the ` gmos`-north spectrograph @xcite on july 17@xmath33 and august 1@xmath34 2005 as part of the program gn-2005a - dd-7 .
the data were obtained using the ifu two - slit mode that provides a field - of - view of 5@xmath357 .
an array of 1500 hexagonal lenslets , of which 500 are located 1 away from the main field to be used for sky subtraction , sets the 02 spatial sampling .
eight individual science exposures of 1895 seconds each were obtained during the two nights , resulting in a total on - source integration time of @xmath36 hours .
an offset of 03 was introduced between exposures to avoid bad ccd regions or lost fibers .
the r400-g5305 grating in combination with the cat - g0309 filter was used to cover a wavelength range of 78009200 with a fwhm spectral resolution of 2.8 . for the data reduction we used an updated version of the officially distributed gemini iraf package .
we applied bias subtraction , flat - fielding and cosmic ray rejection ( using the _ l.a . cosmic _ algorithm by * ? ? ? * ) to each science exposure , and wavelength calibration after the extraction of the data . a careful flat - fielding procedure , crucial for proper removal of fringes in the spectral direction ,
was carried out using the quartz halogen ( qh ) lamp .
wavelength calibration was performed with cuar lamp exposures taken before each science exposure . at the observed wavelengths , a complex spectrum of h@xmath37o absorption features overlaps with the cat lines we are interested in .
we constructed a correction spectrum from observations of the white dwarf star wolf1346 , taken with the same instrumental setup as our science frames .
we checked the range of fiber - to - fiber variations of the spectral resolution of the instrument by measuring the width of the sky lines in each fiber , resulting in the nominal fwhm value of @xmath38 for most of the data cubes . in order to combine all the data , we homogenized the science frames by convolving all spectra to an instrumental fwhm resolution of 3.1 ( or @xmath39kms@xmath6 ) and resampling the spectra to the same range and sampling in wavelength .
after interpolating each science frame to a common spatial grid , taking into account the small spatial offsets applied during the observations , we sum the spectra sharing the same position in the sky to produce the final merged data cube .
the significant contribution from the sky lines to the overall spectrum of the lens galaxy made the sky subtraction the most challenging step in the data reduction process .
although the scatter in the instrumental resolution was small within the individual science frames , the combined effect of other steps in the data reduction ( e.g. fringing , telluric absorption , resampling of data in wavelength ) , introduced alterations in the shape of the sky lines ( from the sky fibers ) that were not equally reproduced in the science fibers . as a consequence
any attempt to use the sky from the sky fibers resulted in serious systematic effects ( i.e. p - cygni like residuals ) in the galaxy spectra underneath . in order to minimize these effects
, we chose to extract the sky spectra from the outer most regions of our fully merged science data cube , where no galaxy light was appreciable .
the sky was then subtracted optimally as described below .
to measure reliable stellar kinematics we first co - add spectra using the adaptive spatial two dimensional binning scheme of @xcite to obtain in each resulting ( voronoi ) bin a minimum signal - to - noise ratio ( s / n ) .
the resulting spectra are spectrally rebinned to steps of constant @xmath40 , equivalent to a sampling of 24kms@xmath6 per pixel .
we adopt the single stellar population ( ssp ) models of @xcite as spectral templates , which are convolved to have the same effective instrumental broadening as the ` gmos ` data .
next , using an updated version ( v4.2 ) of the penalized pixel - fitting ( ppxf ) algorithm of @xcite , a non - negative linear combination of these templates is convolved with a gaussian line - of - sight velocity distribution ( losvd ) and fit to each observed spectrum , to derive the mean line - of - sight velocity @xmath31 and velocity dispersion @xmath41 per bin on the plane of the sky .
this version of the code allows for the inclusion of a sky spectrum into the set of stellar templates , and determines the optimal scaling of the sky to the overall spectrum .
we used this procedure to _ clean _ the individual spectra in our merged data cube , before we derived the final stellar kinematics .
as an example of how this routine performed , we show in fig .
[ fig : spectrum ] the spectrum from the center of the lens galaxy , with the best - fit ssp template overplotted .
we also plot the removed sky spectrum on top , to show that the sky emission - lines can severely affect the ca ii triplet absorption lines that we use to extract the stellar kinematics , especially at larger radii where the the relative contribution from the sky is more significant .
in particular , in our experiments , a poor sky subtraction caused significant systematic effects in the map of @xmath41 at the receding side of the galaxy , where one of the absorption lines is redshifted onto ( the residual of ) a sky emission line .
the map of @xmath31 , however , was ( nearly ) unaffected , showing regular rotation with a straight zero - velocity curve that coincides with the minor axis of the ( bulge ) photometry .
these tests already show that , at least in the inner parts of the lens galaxy , both the photometry and kinematics are consistent with axisymmetry .
if we now _ assume _ this symmetry from the start and apply it during the extracting of the stellar kinematics , we can not only increase the s / n , but at the same time also suppress or even remove systematic effects .
to do this , we make use of a built - in functionality in the ppxf algorithm that allows diametrically opposed spectra to be fitted simultaneously . the symmetry imposed by this technique can be either mirror - symmetric or point - symmetric , corresponding to the behavior of the losvd in axisymmetric or triaxial stellar systems , respectively . in practice , this is done following the method outlined by @xcite , whereby the two spectra from opposed positions in the galaxy are laid end - to - end to create a single vector .
the template library is also doubled in length , where the second half is convolved by the symmetric counterpart of the broadening function used for the first half ( simply reflecting the broadening kernel around the systemic velocity ) .
the two galaxy spectra are therefore fitted simultaneously , with the implied assumption of symmetry . performing a symmetric extraction of the kinematics
presents a number of advantages , and has been used by numerous authors not just as a quality - check ( e.g. * ? ? ?
* ) , but also when the assumption of symmetry is an integral part of the analysis ( e.g. * ? ? ?
the two main advantages for the case presented here are in reducing the impact of systematic effects in the data ( including sky subtraction ) and increasing the effective spatial resolution .
the latter is possible since under the assumption of axisymmetry , all moments of the losvd are symmetric about the galaxy major axis .
we can therefore ` fold ' the data cube about the major axis , combining spectra from the two hemispheres , which increases the effective s / n within a given spatial element , and therefore reduces the required spatial bin size .
we interpolate the merged ` gmos ` data cube onto a new spatial grid with @xmath42 and @xmath43 coordinates parallel to the major and minor photometric axes respectively .
we then fold the data - cube about the major axis by summing together spatial pixels ( ` spaxels ' ) with equal @xmath44 positions . before spatially binning the spectra to a minimum s / n , the folded data cube is folded again , into a single quadrant by combining spaxels with equal @xmath45 positions .
the voronoi bins are derived in this quadrant , and the resulting bin centers are mirrored around @xmath46 .
the spectra are combined according to these voronoi bins to ensure symmetry around the minor axis . at the field perimeter
, it is generally not the case that both sides of the galaxy are sampled at the exact same positions , due to the alignment of the ` gmos ` field with respect to the galaxy s principle axes .
this results in some bins where the s / n of the opposed spectra is far from equal , or in some cases where observations are only present on one side of the galaxy .
the former is taken into account by consideration of the error spectrum of each bin .
for the latter , the bin is fitted as a single spectrum , giving kinematic values only at that position , without a mirrored counterpart . from monte - carlo simulations for data of this spectral range and resolution , a minimum s / n of 20
is deemed optimal .
this results in an average error in both @xmath31 and @xmath41 of about @xmath47kms@xmath6 , while still preserving the spatial resolution of the data .
however , towards the edge of the ` gmos ` field , the errors increase to @xmath48kms@xmath6 in @xmath31 and @xmath49kms@xmath6 in @xmath41 .
the final symmetrically binned data cube consists of @xmath50 bins , with the largest containing @xmath51 individual 0.2 spaxels .
the resulting ( original and symmetrized ) maps of @xmath31 and @xmath41 are presented in fig .
[ fig : eckin ] , with superposed contours of the integrated flux from the data cube .
the velocity map shows regular rotation with amplitude of @xmath31 up to @xmath5kms@xmath6 .
the dispersion map shows somewhat higher values of @xmath41 at the position of the quasar images .
this is caused by imperfect subtraction of the quasar continuum , which acts to strongly dilute the absorption lines and make them appear broader . in our stellar template fit to the observed spectra , we include additive polynomials to account for remaining systematic effects in the data , e.g. due to imperfect flat fielding . at the same time , we use these additive components to mimic the contaminating contribution of the quasar images , which impart a strong featureless continuum on top of the galaxy s integrated stellar spectrum .
this method nicely removes nearly all contribution from the quasar , except where the quasar images are the brightest , in which case @xmath41 can not be reliable measured .
we exclude these regions when we fit dynamical models .
however , between the quasar images , the spectra are dominated by the stellar light of the galaxy , giving a reliable central dispersion @xmath52kms@xmath6 .
likewise , at the edges of the field , the quasar contamination is small , providing a reliable gradient in the velocity dispersion . comparing in fig .
[ fig : eckin ] the symmetrized stellar kinematics with what is obtained without imposing axisymmetry , there is a marked reduction of noise and systematic effects , especially in the velocity dispersion map .
however , the main features of regular rotation , decreasing @xmath41 at larger radii , as well as corresponding velocity and dispersion amplitudes , are present in both approaches , albeit with larger uncertainty and systematics in the non - symmetrized case . given the various advantages , we consider hereafter the symmetrically extracted kinematics . as mentioned above , both the photometry and ( unsymmetrized ) kinematics in the inner parts of the lens galaxy are consistent with axisymmetry .
hence , in constructing dynamical models we consider an axisymmetric stellar system in which both the potential @xmath53 and distribution function ( df ) are independent of azimuth @xmath54 and time .
by jeans ( @xcite ) theorem the df only depends on the isolating integrals of motion : @xmath55 , with energy @xmath56 , angular momentum @xmath57 parallel to the symmetry @xmath58-axis , and a third integral @xmath59 for which in general no explicit expression is known . however , usually may loose this symmetry . ]
@xmath59 is invariant under the change @xmath60 .
this implies that the mean velocity is in the azimuthal direction ( @xmath61 ) and the velocity ellipsoid is aligned with the rotation direction ( @xmath62 ) .
@xcite introduced a method that sidesteps our ignorance about the non - classical integrals of motion .
it finds the set of weights of orbits computed in an arbitrary gravitational potential that best reproduces all available photometric and kinematic data at the same time .
the method has proved to be powerful in building detailed spherical and axisymmetric models of nearby galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) as well as globular clusters , and since recently also triaxial models @xcite .
however , in all cases the stellar systems are significant closer than the lens galaxy in the einstein cross , allowing for ( even ) more and higher - order stellar kinematic measurements necessary to constrain the large freedom in this general modeling method .
instead we construct simpler , but still realistic dynamical models based on the solution of the axisymmetric jeans equations .
when we multiply the collisionless boltzmann equation in cylindrical coordinates by respectively @xmath63 and @xmath64 and integrate over all velocities , we obtain the two jeans equations ( see also * ? ? ?
* eq . 4 - 29 ) @xmath65 where @xmath66 is the intrinsic luminosity density . due to the assumed axisymmetry ,
all terms in the third jeans equation , that follows from multiplying by @xmath67 , vanish .
we are thus left with four unknown second order velocity moments @xmath68 , @xmath69 , @xmath70 and @xmath71 and only two equations .
this means we have to make assumptions about the velocity anisotropy , or in other words the shape and alignment of the velocity ellipsoid . in case
the velocity ellipsoid is aligned with the cylindrical @xmath72 coordinate system @xmath73 , so that we can readily solve equation for @xmath69 . if we next assume a constant flattening of the velocity ellipsoid in the meridional plane , we can write @xmath74 and solve equation for @xmath70 .
this assumption provides in general a good description for the kinematics of real disk galaxies @xcite , such as the lens galaxy in the einstein cross , which is an early - type spiral galaxy .
when @xmath75 , the velocity distribution is isotropic in the meridional plane , corresponding to the well - known case of a two - integral df @xmath76 ( e.g. * ? ?
* ; * ? ? ?
* ) . knowing the intrinsic second - order velocity moments , the line - of - sight second - order velocity moment for a stellar system viewed at an inclination @xmath77 away from the @xmath58-axis follows as @xmath78 \ , { \mathrm{d}}z',\end{aligned}\ ] ] where @xmath79 is the ( observed ) surface brightness with the @xmath42-axis along the projected major axis . for each position
@xmath80 on the sky - plane , @xmath81 yields a prediction of the combination @xmath82 of the ( observed ) mean line - of - sight velocity @xmath31 and dispersion @xmath41 . under the above assumptions , besides the anisotropy parameter @xmath22 ( and possibly the inclination @xmath83 )
, the only unknown quantity is the gravitational potential , which via poisson s equation is related to the total mass density @xmath84 .
we may estimate the latter from the intrinsic luminosity density @xmath66 , derived from deprojecting the observed surface brightness @xmath79 , once we know the total mass - to - light ratio @xmath85 .
it is common in dynamical studies of the inner parts of galaxies to consider @xmath85 as an additional parameter and to assume its value to be constant , i.e. , mass follows light ( e.g. * ? ? ?
since @xmath85 may be larger than the stellar mass - to - light ratio @xmath16 , this still allows for possible dark matter contribution , but with constant fraction . as an alternative to these
standard constant-@xmath85 models , we also construct dynamical models in which we use the strong gravitational lensing to constrain the gravitational potential . in [ sec : lensmodel ] below , we use the positions and relative fluxes of the quasar images to estimate the surface mass density @xmath86 , which we deproject to obtain @xmath84 , so that @xmath85 is not anymore a free parameter . for both type of dynamical models ,
the best - fits to the stellar kinematic data are shown in fig .
[ fig : jeansmodel ] , and discussed further in
[ sec : axijeansmodels ] below .
we use the multi - gaussian expansion method to parameterize both the observed surface brightness @xmath79 as well as the estimated surface mass density @xmath86 as a set of two - dimensional gaussians .
even though gaussians do not form a complete set of functions , in general surface density distributions are accurately reproduced , including deviations from an elliptical distribution and ellipticity variations with radius .
representing also the point - spread function ( psf ) by a sum of gaussians , the convolution with the psf becomes straightforward .
moreover , the mge - parameterization has the advantage that the deprojection can be performed analytically once the viewing angle(s ) are given .
also many intrinsic quantities such as the potential can be calculated by means of simple one - dimensional integrals .
similarly , the calculation of @xmath81 in equation reduces from the ( numerical ) evaluation of in general a triple integral to a straightforward single integral ( see * ? ? ?
* for further details and a comparison with more general dynamical models ) .
the latter integral is given in appendix [ sec : appmge ] , together with expressions for @xmath66 and @xmath53 for the mge - parametrization of @xmath79 and @xmath86 as given in table [ tab : mgepar ] , and further discussed in
[ sec : massdistr ] below . under the thin - lens approximation ,
the gravitational lensing properties of a galaxy are characterized by its potential projected along the line - of - sight , also known as the deflection potential @xmath87 . in case of a mge parametrization of the surface mass density of the galaxy , the projection along the line - of - sight of the corresponding gravitational potential , or , alternatively , solving the two - dimensional poisson equation becomes straightforward .
the calculation of @xmath87 and its derivatives reduce to the ( numerical ) evaluation of a single integral , as shown in appendix [ sec : appmgeaxilens ] for an axisymmetric system .
this mge approach thus provides a powerful technique to build general lens models .
however , since in this case the quasar images only provide a few constraints within a limited radial range , we adopt instead a simpler and less general approach .
we use the algorithm of @xcite to construct a lens model that accurately fits the ( optical ) positions and relative ( radio ) fluxes of the four quasar images in the einstein cross .
the deflection potential is assumed to be a scale - free function @xmath88 of the polar coordinates @xmath89 and @xmath90 in the lens sky - plane , with @xmath91 for realistic models .
the angular part @xmath92 is expanded as a fourier series @xmath93.\ ] ] the positions @xmath94 of the images are related to the position @xmath95 of the source by the lens equation ( e.g. * ? ? ?
* ) @xmath96 , \nonumber \\ \label{eq : lenseq } \eta & = & y_i \left [ 1 - \left ( \alpha f_i - \cot\theta_i \ , f'_i \right ) r_i^{\alpha-2 } \right ] , \end{aligned}\ ] ] where @xmath97 and @xmath98 denotes the derivative to @xmath99 .
the flux ratios of the images follow from their magnifications @xmath100 , which are given by @xmath101 r_i^{2(\alpha-2)},\end{gathered}\ ] ] where @xmath102 is the parity of image @xmath83 . since the fourier coefficients @xmath103 and @xmath104 correspond to a displacement of the source position @xmath95
, we set @xmath105 to remove this degeneracy . for a given slope @xmath91 ,
we are left with @xmath106 free parameters @xmath107 and eleven constraints ( four image positions and three flux ratios ) . for each image
, we obtain a prediction of the source position @xmath108 and magnification @xmath109 through equations and , respectively .
we find the best - fit parameters by minimizing @xmath110 + \sum_{j=1}^3 \frac{(f_j-\tilde{f_j})^2}{\delta f_j^2},\ ] ] with @xmath111 and @xmath112 errors in the observed image positions , and @xmath113 and @xmath114 the observed flux ratios and corresponding errors ( flat rotation curve ) equations and are linear in the free parameters , so that the solution follows by straightforward matrix inversion @xcite . ] . during the minimization we guide the solution towards a smooth , realistic lens model by adding to @xmath115 a constant times @xmath116 , which suppresses ( strong ) deviations from an elliptic shape . for a range of slopes @xmath117
, the fits of the above lens model with @xmath118 fourier terms are shown in fig .
[ fig : ecfit ] . whereas the predictions of the source position remain within the observed errors over the full range of slopes , the recovery of the observed radio fluxes places constraints on the slope . dividing the corresponding @xmath115 values by the ten degrees of freedom ( eleven constraints minus the slope parameter )
provides a quality of fit , shown in the lower - right panel of fig . [
fig : ecfit ] .
the minimum value @xmath119 shows that for @xmath120 a good overall best - fit model is obtained .
the corresponding @xmath121%-confidence interval ( dotted line ) is @xmath122 . with @xmath123
fourier terms the fits are significantly worse ( @xmath124 ) , and with @xmath125 fourier terms the fit does not improve ( for @xmath126 the fitting problem becomes underdetermined ) . if we replace the radio fluxes by the optical fluxes from the ` castles ` website , no acceptable lens model fit is possible ( @xmath127 ) . on the other hand , the mid - infrared fluxes from @xcite yield similar good fits as the radio fluxes .
the best - fit slope @xmath128 is lower because the fluxes of image a and c are respectively about 10% and 30% lower than in the radio .
image c is the faintest in both mid - infrared and radio , while image d is the faintest in the optical .
since the radio is expected to be ( much ) less affected by differential extinction and microlensing , we adopt the corresponding lens models .
c*13c @xmath129 & @xmath130 & @xmath131 & @xmath132 & @xmath133 & @xmath134 & @xmath135 & @xmath136 & @xmath137 & @xmath138 & @xmath14 & @xmath139 & @xmath140 + 0.8 & 0.0824 & -0.0151 & 2.1616 & -0.0590 & 0.0001 & 0.0017 & 0.0593 & 0.0022 & 0.0030 & 0.8858 & 1.5414 & 1.6004 + 0.9 & 0.0736 & -0.0143 & 1.9456 & -0.0494 & 0.0001 & 0.0010 & 0.0500 & 0.0005 & 0.0019 & 0.8862 & 1.5425 & 1.5846 + 1.0 & 0.0651 & -0.0133 & 1.7733 & -0.0415 & 0.0002 & 0.0006 & 0.0423 & -0.0006 & 0.0012 & 0.8866 & 1.5441 & 1.5756 + 1.1 & 0.0570 & -0.0121 & 1.6325 & -0.0349 & 0.0004 & 0.0003 & 0.0357 & -0.0014 & 0.0009 & 0.8872 & 1.5461 & 1.5708 + 1.2 & 0.0492 & -0.0107 & 1.5154 & -0.0293 & 0.0005 & 0.0002 & 0.0299 & -0.0018 & 0.0007 & 0.8879 & 1.5485 & 1.5691 + in the investigation below , we use the lens models with slopes @xmath141 , shown in fig . [
fig : ecmod ] , and with best - fit parameters given in table [ tab : ecmod ] .
the critical curves @xmath142 are obtained by solving the ( quadratic ) equation for infinite magnification , i.e. , for vanishing left - hand - side .
the caustics then follow upon substitution of @xmath142 in the lens equation .
we see in fig .
[ fig : ecmod ] that even the two models outside the @xmath121%-confidence interval for @xmath129 are relatively smooth as expected for a realistic galaxy model .
the surface mass density of the lens models follows from poisson s equation as @xmath143 r^{\alpha-2},\ ] ] with the critical surface mass density defined as @xmath144 where @xmath145 is the speed of light and @xmath146 , @xmath147 and @xmath148 are the ( angular diameter ) distance to the lens galaxy , the quasar source and the distance from lens to source , respectively .
the ( projected ) mass within the critical curve @xmath149 { \ensuremath{r_c}}(\theta)^\alpha { \mathrm{d}}\,\theta,\ ] ] is in general close to the projected mass within the einstein radius .
the latter is the radius @xmath14 for which the projected mass is equal to the einstein mass defined as @xmath150 . because @xmath14 is independent of @xmath90 , all higher order fourier terms average out , and we are left with @xmath151 , so that the einstein mass becomes @xmath152 since we express all ( projected ) coordinates on the plane of the sky in arcseconds , including the radius @xmath89 , we multiply expressions and by @xmath153 to convert from arcseconds to pc for a given ( angular diameter ) distance @xmath146 to the lens galaxy in mpc . from the last two columns in table [ tab : ecmod ] , we see indeed that @xmath154m@xmath13 , nearly independent of the slope @xmath129 .
taking into account an inverse scaling with the hubble constant of @xmath155kms@xmath6mpc@xmath6 , our values are within a few per cent of previous measurements ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
c*3rc*3r & & & + @xmath83 & @xmath156 & @xmath157 & @xmath158 & & @xmath159 & @xmath157 & @xmath158 + 1 & 4.329 & -1.564 & 0.700 & & 5.114 & -1.398 & 0.645 + 2 & 3.935 & -0.942 & 0.700 & & 4.556 & -1.037 & 0.660 + 3 & 3.606 & -0.627 & 0.700 & & 4.238 & -0.779 & 0.662 + 4 & 3.293 & -0.239 & 0.700 & & 3.928 & -0.544 & 0.678 + 5 & 3.005 & -0.043 & 0.700 & & 3.628 & -0.350 & 0.673 + 6 & 2.845 & 0.230 & 0.700 & & 3.459 & -0.186 & 0.675 + 7 & 2.261 & 0.587 & 0.700 & & 3.354 & -0.001 & 0.675 + 8 & 2.160 & 0.917 & 0.414 & & 3.208 & 0.220 & 0.675 + 9 & 1.334 & 1.135 & 0.700 & & 3.069 & 0.496 & 0.675 + 10 & - & - & - & & 2.991 & 1.000 & 0.675 + in the left and middle panel of fig .
[ fig : mgesurf ] we show mge fits ( contours ) to respectively the @xmath31-band and the @xmath11-band hst image , obtained with the software of @xcite , while masking the quasar images . for each fit
we assume the same position angle ( pa ) for the gaussians , consistent with the adopted axisymmetry in the dynamical models .
although the bar and spirals can not be reproduced , it provides a good description of the disk in the outer parts ( left panel ) and reproduces well the bulge in the inner parts ( middle panel ) .
similarly , we derive mge fits to the surface mass density in equation of the best - fit lens models with slopes @xmath129 from 0.8 to 1.2 . in the right panel of fig .
[ fig : mgesurf ] , we show the mge fit ( solid contours ) to the surface mass density ( dashed contours ) of the overall best - fit lens model with slope @xmath160 .
the corresponding parameters are given in table [ tab : mgepar ] ( columns 57 ) , together with the parameters of the mge fit to the @xmath11-band surface brightness ( columns 24 ) .
the latter parameters are `` deconvolved '' using a mge model fitted to the wfpc2 psf as determined with tiny tim @xcite .
the gaussians in the mge fit to the @xmath11-band surface brightness have a pa of @xmath161@xmath162 , which is similar to the pa of @xmath163@xmath162 in the mge fit to the surface mass density of lens models .
both these values are consistent with the measurements by @xcite , who found a pa of @xmath163@xmath162 for the axis through quasar images c and d , bracketed by a pa of @xmath164@xmath162 for the outer disk and a pa of @xmath165@xmath162 for the bar ( see also fig . 1 of * ? ? ?
* ) . as shown in fig .
[ fig : ecmassdistr ] , not only the orientation , but also the shape of the luminous density distribution as inferred from the surface brightness , is very similar to that of the total mass density distribution as derived from the lens models . the left and middle panel in the top row show
respectively the profile and the projected ellipticity @xmath166 of the surface mass density as function of the major axis on the sky - plane .
the thin curves are for the lens models with fixed slope @xmath129 ( as indicated in the bottom - right panel ) .
the thick solid curve is for the @xmath11-band surface brightness , multiplied with a constant mass - to - light ratio @xmath167@xmath10 .
the latter is derived from dividing the projected mass from the lens models ( shown in the top - right panel of fig .
[ fig : ecmassdistr ] ) within the einstein radius , i.e. , the einstein mass @xmath139 , by the projected luminosity within the einstein radius @xmath14 .
we see that around @xmath14 ( indicated by the dashed vertical line ) , where the lens models are best constrained by the observed quasar image positions and relative flux ratios , both the slope and the ellipticity of the _ projected _ luminous and total mass density are nearly the same .
the same holds true for the corresponding _ intrinsic _ mass densities , of which the profiles and ellipticities @xmath168 are shown in the left and middle panel in the bottom row of fig .
[ fig : ecmassdistr ] .
these ( analytic ) deprojections ( see appendix [ sec : appmge ] ) are under the assumption of oblate axisymmetry and for a given inclination which we derive from the flattening of the disk in the outer parts .
we find from the mge fit to the @xmath31-band surface brightness ( left panel of fig . [
fig : mgesurf ] ) , that gaussian components with a measured flattening as small as @xmath169 are required for an acceptable fit .
this sets a lower limit to the inclination of @xmath170@xmath162(significantly larger than @xmath171@xmath162 found by * ? ? ?
adopting a lower limit for the _ intrinsic _ flattening of the disk of @xmath172 ( e.g. * ? ? ?
* ) , we then obtain an inclination of @xmath173@xmath162 . around @xmath14
the luminosity density is close to isothermal ( @xmath174 intrinsic or @xmath175 in projection ) like the mass density of the overall best - fit lens model . towards the center
the slope of the luminosity density becomes shallower .
the ( fixed ) slope of the lens model is not anymore ( well ) constrained by the quasar images towards the center , neither for radii a few times @xmath14 . at these larger radii
also the luminosity density is not only due to the bulge , but the bar and disk start contributing , as can also be seen from the increase in the ellipticity . nevertheless , within a radius @xmath176
, the intrinsic luminous and total mass distribution are very similar , i.e. , mass follows light .
this means we can infer the gravitational potential either from the observed surface brightness adopting a ( constant ) total mass - to - light ratio @xmath85 , or directly from the best - fit lens model ( see also appendix [ sec : appmge ] ) .
the corresponding circular velocity curves , defined as @xmath177 in the equatorial plane , are shown in the bottom - right panel of fig .
[ fig : ecmassdistr ] , using @xmath178@xmath10 as above . in fig .
[ fig : jeansmodel ] , the second panel in the bottom row shows the square root of the combination @xmath179 of the observed ( and symmetrized ) mean line - of - sight velocity @xmath31 and dispersion @xmath41 maps of the lens galaxy ( fig .
[ fig : eckin ] , two panels on the right ) .
the other panels in fig .
[ fig : jeansmodel ] show the line - of - sight second - order velocity moment as predicted by the axisymmetric jeans models , using the solution given in equation of appendix [ sec : appmge ] . with a central velocity dispersion of @xmath180kms@xmath6 ,
the @xmath181 relation as given by @xcite predicts a central black hole of mass @xmath182m@xmath13 for the lens galaxy .
since the corresponding sphere - of - influence @xmath183 is not resolved by our kinematic data , we can not fit for it , but do include the predicted bh in the jeans models ( as an additional gaussian ) .
finally , we convolve the predicted second - order velocity moment with the gaussian psf of the kinematic data with fwhm@xmath184 .
the ( linear ) scale in kms@xmath6 , as indicated by the color bar and limits at the top of the panel with the observations , is the same in all panels of fig .
[ fig : jeansmodel ] . the axisymmetric jeans models in the top row
use the gravitational potential inferred from the @xmath11-band surface brightness , assuming that mass follows light .
the corresponding ( constant ) dynamical mass - to - light ratio @xmath185 that provides the best - fit to the observations is indicated at the top , together with the assumed value for the anisotropy parameter @xmath22 , increasing from left to right . the corresponding increase in the ( constant ) flattening of the velocity ellipsoid in the meridional plane ( or decrease in @xmath186 ) , has two main effects : a less pinched `` butterfly '' shape and a weaker gradient parallel to the ( vertically aligned ) short axis . comparing with the observations
, we see that at lower @xmath22 values the pinching is too strong , while at higher @xmath22 the gradient along the short axis is too shallow .
therefore , it is not unexpected that the `` best - fit '' is when @xmath187 and @xmath188@xmath10 ( in the @xmath11-band ) , but with reduced @xmath189 significantly above unity .
part of the latter mismatch might be due to ( systematic ) discrepancies in the data , in particular due to the challenging measurement of the velocity dispersion .
the average uncertainty in the observed second - order velocity moment is about @xmath190kms@xmath6 , which translates into an error of around @xmath191 per cent in @xmath185 , or about @xmath192@xmath10 .
however , the uncertainties in the observed second - order velocity moment vary quite a lot , from only @xmath193kms@xmath6 in the center to @xmath49kms@xmath6 a the edge of the ` gmos ` field . and
although we were very careful in disentangling the contribution from the quasar and excluded the regions at the quasar images , the measured velocity dispersion around these regions might still be systematically affected by ( broadening due to ) residual light from the quasar source .
this possibly explains the observed excess in the observed second - order velocity moment around the quasar images with respect to the accurately measured central value . at the same time , such an excess might mimic a steeper gradient in the observed second - order velocity moment .
indeed , in all jeans models in the top row of fig .
[ fig : jeansmodel ] the prediction at the center is significantly higher than the observed value . as a result
, we expect that the dynamical mass - to - light ratio is overestimated . if , alternatively , we infer the gravitational potential directly , i.e. , without the need for a mass - to - light ratio , from the ( deprojected ) surface mass density of the best - fit lens models , the axisymmetric jeans models predict maps of the second - order velocity moments that are similar in shape but with lower values near the center .
this is shown in the middle row of fig .
[ fig : jeansmodel ] for the overall best - fit lens model with slope @xmath194 , and for the same range in anisotropy parameters @xmath22 .
the formal best - fit is obtained for @xmath195 , but the observed central value is more closely matched in case of mild velocity anisotropy in the meridional plane with @xmath196 . for @xmath128 .
i.e. , density slopes steeper then isothermal , the observed central value is best matched when @xmath195 , with at the same time a steeper gradient along the short - axis , as shown in the first panel in the bottom row of fig .
[ fig : jeansmodel ] .
the last two panels show that for @xmath197 , velocity isotropy in the meridional plane ( @xmath198 ) results in a central value much lower than the observed second - order velocity moment . we conclude that for the lens models with slopes @xmath199 , as constrained by the lensing geometry , and for a mild anisotropic velocity distribution in the meridional plane with @xmath200 , the predictions of the second - order velocity moments match the accurately measured central value , and follow the gradient within the observational uncertainties outside the region that might be affected by ( residual ) quasar light . the latter results in broadening of the velocity dispersion , so that the best - fit dynamical mass - to - light ratios we derive when we infer the gravitational potential from the surface brightness , are likely overestimated .
fitting instead only the unaffected and accurately measured second - order velocity moment in the center , we find for @xmath201 a best - fit value of @xmath202@xmath10 , with an error around @xmath192@xmath10 . on the other hand ,
both the einstein mass and luminosity within the einstein radius are well constrained , resulting in at most a few per cent error in @xmath178@xmath10 .
therefore , in what follows , we adopt the latter as the total mass - to - light ratio in the inner bulge - dominated region of the lens galaxy . in the above axisymmetric jeans models we fix the inclination to @xmath203@xmath162 determined from the observed ellipticity of the outer disk ( see [ sec : massdistr ] ) .
when we vary the inclination over a reasonable range above the lower limit @xmath170@xmath162 , there is no significant change in the above results , which is expected because the inclination only has a weak effect on the mass - to - light ratio ( see also fig .
4 of * ? ? ?
we have shown that in the bulge - dominated inner region of the lens galaxy , the total mass distribution closely follows the light distribution .
this does not mean there is no dark matter present , since if dark matter follows ( nearly ) the same distribution it might contribute to part of the above estimated total mass - to - light ratio @xmath204@xmath10 . to determine this possible _ constant _ dark matter fraction , we need to measure the stellar mass - to - light ratio @xmath16 . to this end
we use the single stellar population ( ssp ) models of @xcite , adopting as initial reference the @xcite initial mass function ( imf ) with lower mass cut - off @xmath205m@xmath13(faintest star @xmath206m@xmath13 ) and upper mass cut - off of @xmath207m@xmath13 .
we derive a high - s / n spectrum from our data cube by collapsing the central spectra that are unaffected by the quasar images .
even so , the ca ii triplet is equally well fitted by ssp models of nearly all ages and metallicities of about @xmath208}}= \pm 0.2 $ ] around solar , which leaves the stellar mass - to - light ratio nearly unconstrained .
therefore , we also derived colors from archive hst images ( pi : kochanek ) .
after matching all images , using both the quasar positions and isophotes of the lens galaxy , we compute the magnitudes in a circular aperture of radius @xmath209 ( well within the quasar images at @xmath210 ) .
next , we convert the resulting magnitudes in the f555w , f675w , f814w ( wfpc2 ) , f160w and f205w ( nicmos2 ) filters to ( rest frame ) magnitudes in respectively the johnson @xmath31 , @xmath11 , @xmath89 , @xmath211 and @xmath212 filters .
we use the method described in 3 of @xcite to take simultaneously into account the filter responses and the redshift of the lens galaxy .
the resulting colors are best fitted by a ssp model with age @xmath213gyr and metallicity @xmath208}}\simeq 0.2 $ ] , corresponding to an @xmath11-band stellar mass - to - light ratio @xmath214@xmath10 .
the @xmath215 confidence limits yield a range in @xmath216 from 7 to 14gyr , and in @xmath208}}$ ] from @xmath217 to @xmath218 , corresponding to a range in @xmath16 from @xmath17 to @xmath18@xmath10 .
the lower limit implies at most @xmath49 per cent of dark matter within @xmath219 , but the results are fully consistent with no dark matter at all .
the ssp models of @xcite with a kroupa imf yield similar results . adopting a @xcite imf with the same lower and upper mass cut - offs ,
increases @xmath16 by about @xmath220 per cent .
this not only implies no dark matter , but even for the lower limit @xmath221 , which is unphysical .
this is consistent with evidences for both late - type galaxies @xcite and early - type galaxies @xcite that , if the imf is universal , a salpeter imf is excluded , whereas a kroupa imf matches the observations ( see * ? ? ?
* for a review ) .
we used the ` gmos`-north integral - field spectrograph to obtain two - dimensional stellar kinematics of the lens galaxy in the einstein cross .
in addition to the four bright quasar images and the distance of the lens galaxy ( @xmath29mpc ) , in particular the presence of sky lines in the observed ca ii triplet region made the extraction of the absorption line kinematics challenging .
even so , we were able to derive high - quality line - of - sight velocity @xmath31 and dispersion @xmath41 maps of the bulge - dominated inner region @xmath176 , reaching about two - thirds of the effective radius @xmath2 of this early - type spiral galaxy .
the @xmath31 map shows regular rotation up to @xmath5kms@xmath6 around the minor axis of the bulge , consistent with axisymmetry .
the @xmath41 map shows a weak gradient increasing towards a central value of @xmath222kms@xmath6 .
the only other direct ( single ) measurement of the velocity dispersion is @xmath223kms@xmath6 by @xcite . for a singular isothermal sphere lens model
, we can use the relation @xmath224 ( e.g. * ? ? ?
* ) with a separation @xmath225 of the four quasar images , to obtain a simple estimate for the dispersion of @xmath226kms@xmath6 .
taking into account aperture correction and a range in velocity anisotropy , @xcite convert this to a central velocity dispersion of @xmath227kms@xmath6 .
the king and de vaucouleurs models of @xcite predict a similar value of @xmath228kms@xmath6 , and also @xcite find a value of @xmath229kms@xmath6 based on their two rotation curve measurements .
it is likely that the long - slit measurement by @xcite is affected by the bright quasar images , whereas our spatially resolved measurements allow for a clean separation of the quasar contribution .
a large variety of different lens models have been constructed for the einstein cross , most of which fit the positions of the quasar images but not their relative flux ratios . adopting the scale - free lens model of @xcite , we found that fitting at the same time also the ( radio ) flux ratios constrained the slope of the total mass surface density @xmath230 to be @xmath199 .
the total mass within the einstein radius @xmath8 , i.e. , the einstein mass , is @xmath231m@xmath13 , nearly independent of the slope @xmath129 , and consistent with previous measurements in the literature .
dividing by the projected luminosity within @xmath14 , as measured from the observed @xmath11-band surface brightness , we obtained a mass - to - light ratio of @xmath167@xmath10 , with an error of at most a few per cent .
we determined the mass - to - light ratio in an additional , independent way by fitting dynamical models to the ( combined ) observed @xmath31 and @xmath41 maps .
we used the solution of the axisymmetric jeans equations to predict this second - order velocity moment , with the gravitational potential inferred from the deprojected @xmath11-band surface brightness , assuming that mass follows light .
we expect the resulting best - fit constant mass - to - light ratio to be an overestimation due to possible residual contribution from the quasar light .
even so , restricting the fit to the unaffected and accurately measured central second - order velocity moment , we found @xmath232@xmath10 .
when we used instead the gravitational potential that follows directly , i.e. , without mass - to - light ratio conversion , from the surface mass density of the best - fit lens models , we arrived at a similar prediction of the second - order velocity moment .
we showed that the reason is that the luminous and total mass distribution , as inferred from respectively the surface brightness and lens model , are very similar .
this implies that the inner region of the lens galaxy mass ( closely ) follows light , with a total mass - to - light ratio @xmath204@xmath10 . by fitting single stellar population models to measured colors of the center of the lens galaxy , we estimated an @xmath11-band stellar mass - to - light ratio @xmath16 from 2.8 to 4.1@xmath10 .
although a _ constant _ dark matter fraction of 20 per cent is thus not excluded , it is likely that dark matter does not play a significant role in the inner region of this early - type spiral galaxy of luminosity @xmath233 .
this is consistent with indications that less - luminous early - type galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and late - type galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) lack the dark matter that is ubiquitous in dwarf galaxies as well as in ( at least the outer parts of ) giant elliptical and spiral galaxies .
the constraint @xmath234 on the slope of the lens model implies that the intrinsic total mass density is close to isothermal , consistent with previous studies of lens galaxies ( e.g. * ? ? ?
however , one has to be very careful not to over - interpret this result , not only due to assumptions in the modeling , but most of all because the slope is only ( well ) constrained around the einstein radius @xmath14 .
if indeed mass follows light in the inner region of this lens galaxy , the ( deprojected ) surface brightness indicates deviations from isothermal towards the center where the slope becomes shallower as well as a possible steepening at larger radii .
interestingly , the construction of realistic galaxy density profiles with a stellar and dark matter component that are both non - isothermal , shows that the combined slope and corresponding lensing properties are nevertheless consistent with isothermal around @xmath14 @xcite . with more extended images , e.g. in the case of galaxy - galaxy lensing ,
one can place a stronger constraint on the total mass distribution of the lens galaxy ( e.g. * ? ? ?
* ; * ? ? ?
since the extension in these cases is still mostly tangential , constraining a large radial range is in particular possible when lensing occurs in multiple ( redshift ) planes , resulting in images at different einstein radii ( e.g. * ? ? ?
nevertheless , already the quasar images provide an accurate constraint on the total mass within @xmath14 , nearly independent of the details of the lens model ( e.g. * ? ?
* ; * ? ? ?
therefore , at least around this radius no higher - order velocity moments are needed to break the mass - anisotropy degeneracy . towards the center
the surface brightness increases steeply , so that high enough s / n to measure velocity moments beyond @xmath31 and @xmath41 might be achievable even at higher redshift . in the outer parts ,
the degeneracy might be ( partially ) broken by combining the kinematics of stars and/or discrete tracers , such as globular clusters and planetary nebulae , with total mass estimates from hot x - ray gas and/or weak lensing measurement ( e.g. * ? ? ?
* ) . in the current and even more in the upcoming extensive and deep photometric surveys
, numerous ( strong ) gravitational lensing systems will be discovered .
they provide important and independent constraints on the total mass distribution in galaxies , especially in combination with ( resolved ) stellar kinematics , as we showed in this paper ( see also * ? ? ?
* ; * ? ? ?
we are grateful to jean - ren roy and matt mountain for granting us director s discretionary time for this project and for generous hospitality in hilo to tdz .
we thank ed turner for initial support of this project and tracy beck for efficient and cheerful assistance .
gvdv acknowledges support provided by nasa through hubble fellowship grant hst - hf-01202.01-a awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, for nasa , under contract nas 5 - 26555 .
mc acknowledges support from a stfc advanced fellowship ( pp / d005574/1 ) .
based on observations obtained at the gemini observatory , which is operated by the association of universities for research in astronomy , inc .
, under a cooperative agreement with the nsf on behalf of the gemini partnership : the national science foundation ( united states ) , the particle physics and astronomy research council ( united kingdom ) , the national research council ( canada ) , conicyt ( chile ) , the australian research council ( australia ) , cnpq ( brazil ) , and conicet ( argentina ) .
we summarize the ( numerically ) convenient expressions of the axisymmetric intrinsic density and gravitational potential in the case of a multi - gaussian expansion ( mge ) of the surface density .
next , we show that also the gravitational lensing properties can be readily computed , so that the mge method can be used to efficiently construct general lens models . finally , we give the line - of - sight second - order velocity moment as a solution of the axisymmetric jeans equations which is discussed in detail in @xcite .
we parameterize the surface brightness @xmath79 by a sum of gaussian components @xmath235 \right\}.\ ] ] each with three parameters : the central surface brightness @xmath236 , the dispersion @xmath237 along the major @xmath42-axis and the flattening @xmath238 . in case of an ( oblate ) axisymmetric system viewed at an inclination @xmath239 ,
the corresponding intrinsic luminosity density is @xmath240 \right\},\ ] ] with intrinsic dispersion @xmath241 and intrinsic flattening @xmath242 given by @xmath243 .
the intrinsic mass density follows as @xmath244 , where the mass - to - light ratio @xmath245 per gaussian component is a free parameter .
alternatively , if the surface mass density @xmath86 is available ( e.g. from gravitational lensing ) , an mge - parametrization as in directly yields @xmath246 as in , but with @xmath236 in both expressions replaced by the central surface mass density @xmath247 .
the corresponding gravitational potential follows upon ( numerical ) evaluation of @xmath248 where we have introduced @xmath249 \right\ } \frac{1}{\mathcal{q}_j(u)},\ ] ] and @xmath250 .
the total mass per gaussian component is given by @xmath251 , with @xmath252 when the mass density is inferred from the surface brightness .
in the latter case , we have implicitly assumed a _ constant _ mass - to - light ratio @xmath245 per gaussian component by taking it outside the integral .
nevertheless , we can still mimic a ( radially ) varying mass - to - light ratio by considering the @xmath245 of the gaussian components as free parameters .
however , it is common in dynamical studies of the inner parts of galaxies ( e.g. * ? ? ?
* ) to assume the total mass - to - light ratio to be constant , i.e. , @xmath253 for each gaussian component @xmath254 .
since @xmath85 may be larger than the stellar mass - to - light ratio @xmath16 , this still allows for possible dark matter contribution , but with a constant fraction . under the thin - lens approximation
, the gravitational lensing properties of a galaxy are characterized by the deflection potential @xmath87 and its ( partial ) derivatives .
the deflection potential follows from projecting the potential along the line - of - sight or by solving the two - dimensional poisson equation @xmath255 . here
, @xmath256 is the normalized surface mass density with the critical lensing value @xmath257 defined in equation .
similar to the surface brightness in equation , we parameterize this so - called `` convergence '' @xmath258 by a sum of gaussian components @xmath259 \right\}.\ ] ] the corresponding deflection potential is then @xmath260 where @xmath261 is the total mass per gaussian component , and we have introduced @xmath262 \right\ } \frac{1}{{\mathcal{q}'_k}(u)},\ ] ] and @xmath263 . the first - order partial derivatives of @xmath264 follow as @xmath265 whereas the second - order partial derivatives are given by @xmath266 u \ , { \mathrm{d}}u , \nonumber \\ \frac{\partial^2 \phi_k}{\partial { y'}^2 } & = & \frac{m_k}{\pi{\sigma'_k}^2{\ensuremath{\sigma_c } } } \int_0 ^ 1 { \mathcal{f}'_k}(u ) \left [ 1 - \frac{{x'}^2}{{\sigma'_k}^2 } \frac{u^2}{{\mathcal{q}'_k}^2(u ) } \right ] \frac{u \ , { \mathrm{d}}u}{{\mathcal{q}'_k}^2(u ) } , \nonumber \\ \label{eq : mge_dphi2_k } \frac{\partial^2 \phi_k}{\partial { x ' } \partial { y ' } } & = & \frac{m_k}{\pi{\sigma'_k}^2{\ensuremath{\sigma_c } } } \int_0 ^ 1 { \mathcal{f}'_k}(u ) \left [ \frac{{x'}{y'}}{{\sigma'_k}^2 } \ , u^2 \right ] \frac{u \ , { \mathrm{d}}u}{{\mathcal{q}'_k}^2(u)}.\end{aligned}\ ] ] the above single integrals can be readily evaluated numerically , so that the lensing properties follow in a straightforward way .
the image positions @xmath80 are related to the source position @xmath95 by the lens equation @xmath267 given the parity @xmath268 of the image , the magnification @xmath269 is given by @xmath270 the einstein radius @xmath14 follows from solving @xmath271 , where @xmath272 is the average convergence within the projected radius @xmath273 . after substituting @xmath274 and @xmath275 in equation and performing the integral over @xmath273 ,
we are left with @xmath276 \right\ } \frac{{\mathrm{d}}\theta'}{{\mathcal{p}'}(\theta')},\ ] ] where @xmath277 .
the mass within the einstein radius then follows as @xmath150 .
we consider a luminous component with gaussian intrinsic luminosity density @xmath278 given by , of which the observed kinematics trace the underlying gravitational potential @xmath279 , with the sum over all ( luminous and dark ) components @xmath280 given by . assuming that the velocity ellipsoid is aligned with the cylindrical @xmath72 coordinate system , so that @xmath281 , we can readily solve jeans equation as @xmath282_j = 4 \pi g \sum_k \frac{q'_k \sigma_{0,k}}{\sqrt{2\pi } \sigma_k } \int_0 ^ 1 q_j^2 \sigma_j^2 \frac{\mathcal{f}_k(u ) u^2}{1-f_{jk}u^2 } \ ; { \mathrm{d}}u,\ ] ] with @xmath283 defined in and @xmath284 if we next also assume a constant flattening of the velocity ellipsoid in the meridional plane , we can write @xmath285_j = b_j \ , [ \overline{v_z^2}]_j$ ] , and solve jeans equation as @xmath286_j = 4 \pi g \sum_k \frac{q'_k \sigma_{0,k}}{\sqrt{2\pi } \sigma_k } \int_0 ^ 1 \left [ b_j q_j^2 \sigma_j^2 + \mathcal{g}_{jk}(u ) r^2 \right ] \frac{\mathcal{f}_k(u ) u^2}{1-f_{jk}u^2 } \ ; { \mathrm{d}}u,\ ] ] where we have introduced @xmath287u^2.\ ] ] since the constant @xmath288 may be different for each luminous component @xmath254 , the total anisotropy in the meridional plane @xmath289_j } { \sum_j b_j \nu_j [ \overline{v_z^2}]_j},\ ] ] is allowed to vary throughout the system . in case of isotropy in the meridional plane all @xmath290 and thus @xmath75 , corresponding to a two - integral df @xmath76 .
after substitution of these expressions in equation , the line - of - sight integral can be solved , resulting in ( * ? ? ?
* eq . 27 ) @xmath291_j = \frac{4 \pi^{3/2 } g}{i_j(x',y ' ) } \ , \frac{q'_j i_{0,j}}{\sqrt{2\pi}\sigma_j q_j } \sum_k \frac{q'_k \sigma_{0,k}}{\sqrt{2\pi } \sigma_k } \int_0 ^ 1 \frac{q_j^2 \sigma_j^2 ( \cos^2i + b_j\sin^2i ) + \mathcal{g}_{jk}(u ) x'^2\sin^2i } { ( 1-f_{jk}u^2)\sqrt{[1-(1-q_k^2)u^2 ] ( \mathcal{a}+\mathcal{b}\cos^2i ) } } \\ \times
u^2 \exp\left\ { - \mathcal{a } \left [ x'^2 + \frac{(\mathcal{a}+\mathcal{b})y'^2 } { \mathcal{a}+\mathcal{b}\cos^2i } \right ] \right\ } \ , { \mathrm{d}}u,\end{gathered}\ ] ] where @xmath292 and @xmath293 are functions of @xmath294 defined as @xmath295\sigma_k^2 } \right).\end{aligned}\ ] ] the remaining single integral can be readily evaluated numerically .
this provides a direct prediction of the combination @xmath82 of the observed mean line - of - sight velocity @xmath31 and dispersion @xmath41 at a given position @xmath80 on the sky - plane , through the ( luminosity weighted ) sum @xmath296_j / \sum_j i_j$ ] over the corresponding luminous components .
, m. , bacon , r. , bureau , m. , damen , m. c. , davies , r. l. , de zeeuw , p. t. , emsellem , e. , falcn - barroso , j. , krajnovi , d. , kuntschner , h. , mcdermid , r. m. , peletier , r. f. , sarzi , m. , van den bosch , r. c. e. , & van de ven , g. 2006 , , 366 , 1126 , k. , richstone , d. , tremaine , s. , lauer , t. r. , bender , r. , bower , g. , dressler , a. , faber , s. m. , filippenko , a. v. , green , r. , grillmair , c. , ho , l. c. , kormendy , j. , magorrian , j. , & pinkney , j. 2003 , , 583 , 92 , d. n. , bean , r. , dor , o. , nolta , m. r. , bennett , c. l. , dunkley , j. , hinshaw , g. , jarosik , n. , komatsu , e. , page , l. , peiris , h. v. , verde , l. , halpern , m. , hill , r. s. , kogut , a. , limon , m. , meyer , s. s. , odegard , n. , tucker , g. s. , weiland , j. l. , wollack , e. , & wright , e. l. 2007 , , 170 , 377 , s. , gebhardt , k. , bender , r. , bower , g. , dressler , a. , faber , s. m. , filippenko , a. v. , green , r. , grillmair , c. , ho , l. c. , kormendy , j. , lauer , t. r. , magorrian , j. , pinkney , j. , & richstone , d. 2002 , , 574 , 740 | we present two - dimensional line - of - sight stellar kinematics of the lens galaxy in the einstein cross , obtained with the gemini 8 m telescope , using the ` gmos ` integral - field spectrograph .
the stellar kinematics extent to a radius of @xmath0 ( with @xmath1 spaxels ) , covering about two - thirds of the effective ( or half - light ) radius @xmath2 of this early - type spiral galaxy at redshift @xmath3 , of which the bulge is lensing a background quasar at redshift @xmath4 .
the velocity map shows regular rotation up to @xmath5kms@xmath6 around the minor axis of the bulge , consistent with axisymmetry .
the velocity dispersion map shows a weak gradient increasing towards a central value of @xmath7kms@xmath6 .
we deproject the observed surface brightness from hst imaging to obtain a realistic luminosity density of the lens galaxy , which in turn is used to build axisymmetric dynamical models that fit the observed kinematic maps .
we also construct a gravitational lens model that accurately fits the positions and relative fluxes of the four quasar images .
we combine these independent constraints from stellar dynamics and gravitational lensing to study the total mass distribution in the inner parts of the lens galaxy .
we find that the resulting luminous and total mass distribution are nearly identical around the einstein radius @xmath8 , with a slope that is close to isothermal , but which becomes shallower towards the center if indeed mass follows light .
the dynamical model fits to the observed kinematic maps result in a total mass - to - light ratio @xmath9@xmath10 ( in the @xmath11-band ) .
this is consistent with the einstein mass @xmath12m@xmath13 divided by the ( projected ) luminosity within @xmath14 , which yields a total mass - to - light ratio of @xmath15@xmath10 , with an error of at most a few per cent .
we estimate from stellar populations model fits to colors of the lens galaxy a stellar mass - to - light ratio @xmath16 from @xmath17 to @xmath18@xmath10 .
although a _ constant _ dark matter fraction of 20 per cent is not excluded , dark matter may play no significant role in the bulge of this @xmath19l@xmath20 early - type spiral galaxy . |
since the pioneering work of anderson on localization@xcite , it was realized by mott and twose@xcite , that all states in an infinite 1- dimensional chain are localized at arbitrary disorder strength , at zero temperature . they could show that solutions of the corresponding schroedinger equation at points at a distance @xmath4 are with a probability @xmath5 in resonance , where @xmath6 is the mean free path .
thus , its exponential decrease gives a localization length of the order of @xmath7 .
later on thouless argued that this statement can be extended to a thin disordered wire , and he found that the localization length is given by @xmath8 where @xmath7 is the elastic mean free path , and @xmath9 the number of transverse channels , with the fermi wave vector @xmath10 , and the cross section of the wire , @xmath11@xcite .
this was proven rigourously for a matrix of @xmath12 chains , when @xmath12 is small , by anderson et al .
, weller et al . , and then by dorokhov by solution of a fokker- planck equation@xcite , calculating the transmission probability through the wire . in the limit of a thin wire ,
where the motion of the electrons is diffusive in all directions on small length scales , the proof of complete localization at zero temperature was given by efetov and larkin with a field theoretical method , obtaining an exponential decrease of the density - density correlation function in space , in the zero frequency limit .
they discovered in addition that the localization length depends on the global symmetry of the wire@xcite .
the localization length was found to be given by @xmath13 , where @xmath14 for orthogonal , unitary , and symplectic symmetry , which corresponds to no magnetic field , weak magnetic field , and strong spin- orbit interaction , respectively .
the density autocorrelation function was recently studied for the total spectrum as a function of the length of the mesoscopic wire by altland and fuchs @xcite . because of the complexity of the problem , they did not obtain a closed analytical expression for arbitrary frequency , but succeeded to do a numerical analysis in the unitary regime .
new information on the level statistics of the wire as a function of its length was found in the metal - insulator - crossover regime . in this article
we argue that in order to study the level statistics , it is enough to calculate the simpler asd .
this function contains information on the spectrum , but its complexity is reduced so that it can be calculated analytically more easily .
we will show that it provides a usefull tool to study localization , and could be used in situations which have been unaccessible to other analytical methods .
the article is organized as follows . in the first part
the characterization of level statistics by an autocorrelation function is reviewed , and the asd is defined . in the second part
the result for the disorder averaged asd for a quasi-1-dimensional wire in a weak magnetic field is presented and discussed for various regimes in the frequency - length plane .
information on the metal - insulator - crossover is obtained .
we conclude with a discussion of the results and the potential of the asd as a new tool to study anderson localization .
the crossover from a metal to an insulator in a finite coherent , disordered metal particle is accompanied by a change in the statistics of the discrete energy levels@xcite .
this can be studied by calculating a disorder averaged autocorrelation function between two energies at a distance @xmath15 in the energy level spectrum .
then , considering a quasi-1-dim .
disordered metal wire with crossection @xmath11 , a map as a function of its length @xmath16 and the energy @xmath15 can be drawn as in fig . 1 . here
, @xmath17 is the total mean level spacing with the average density of states @xmath18 .
@xmath19 is the number @xmath20 of electrons per volume @xmath21 .
@xmath22 is the fermi energy , and @xmath23 the electron mass .
@xmath24 is the local mean level spacing , when the length of the wire , @xmath16 , exceeds its localization length @xmath1 .
[ fig1 ] the thouless energy @xmath25 is defined through the diffusion time across the length l , @xmath26 when the diffusion is free , as obtained from the classical diffusion equation @xmath27 , where @xmath28 is the electron density .
the classical diffusion constant d in 3 dimensions is related to the elastic mean free time @xmath29 by @xmath30 .
@xmath31 is the energy which limits the universal ( ergodic ) regime of nonintegrable ballistic quantum billiards @xcite , @xcite . since @xmath31 depends on the exact boundary conditions , it may change as a function of l in a continous , but nonmonotonous way as indicated in fig . 1 .
this map has been explored by considering the autocorrelation function of density of states , see @xcite and references therein . here , we will restrict us to the asd , as defined by @xmath32 where @xmath33 @xmath34 is the hamiltonian of the considered system , and @xmath35 a central energy .
it contains only information about the spectrum , not on the wave functions of the mesoscopic system .
this function was recently obtained for the wigner- dyson random matrices ( gue@xcite , goe@xcite , cue , coe@xcite and crossover@xcite ) and successfully used to characterize the spectrum of nonintegrable quantum systems , like a sinai billiard@xcite , for energies @xmath36 , where the average over the energy @xmath35 had to be done in order to obtain a universal function . in the following ,
it is obtained for free electrons in a random potential in a finite system , which can be used for the study of a disordered mesoscopic metal of which at least one dimension exceeds the mean free path l. the asd can be calculated analytically and shown to agree with the one for wigner - dyson random matrices , when the dimensions of the system do not exceed a localization length @xmath1 .
this is expected , since a disordered metal particle is an example of a nonintegrable physical system , and should therefore have the same energy level statistics as f.e . the sinai billiard , for frequencies not exceeding the thouless energy , the ergodic regime , where a particle has time enough to cover the whole sample uniformly@xcite . here , we will derive the asd for the more general case , when one dimension may exceed the localization length @xmath1 , a quasi-1-dim .
conductor , of length @xmath16 and crossection @xmath11 where the number of transverse channels is much larger than 1 , @xmath37 , in order to find out if this relatively simple function does contain information about the metal - insulator crossover .
the hamiltonian is given by , @xmath38 where @xmath39 is the electron charge , c the velocity of light , and @xmath40 the vector potential due to an external magnetic field @xmath41 .
@xmath42 is a gaussian distributed random function @xmath43 which models randomly distributed , uncorrelated impurities in the wire .
the asd can be calulated for such a hamiltonian as a function of the length of the wire @xmath44 and energy difference @xmath15 .
the unitary limit is considered , where the magnetic flux through the wire , @xmath45 , exceeds @xmath46 .
the asd can be written in terms of grassmannian functional integrals .
this allows to perform the impurity averaging as a gaussian integral .
the resulting interacting theory of grassman fields can be decoupled by a transformation , introducing a functional integral over a @xmath47-matrix q. next , the grassmann fields can be integrated out .
the integral over q can be simplified for @xmath48 to an integral over gapless fluctuations around the saddle point which have the action of an o(3 ) nonlinear sigma model @xmath49 = \frac{\pi}{2 } \int \frac{d x}{s l } [ g tr ( \frac{\nabla}{\pi / l } q ) ^2 + i \frac{\omega}{\delta } tr \lambda q ] , \ ] ] with the nonlinear constraint @xmath50 . here ,
@xmath51 which has the physical meaning of the dimensionless conductance @xmath52 of the wire , as long as the einstein relation to the diffusion constant @xmath53 holds , @xmath54 .
the derivation is given in .
a nonlinear sigma model for disordered electron systems had been first derived in @xcite for n replicas using a functional integral over conventional numbers , in @xcite for grassmann variables , and then for superfields @xcite . choosing a representation of the matrix @xmath55
, the integrals can be performed by means of the transfer matrix method .
thus , the problem can be reduced to the solution of the equation@xcite , @xmath56 p_z ( \lambda ) , \ ] ] with the boundary condition @xmath57 . here ,
@xmath58 , @xmath59 , and the hamilton operator is , @xmath60 = - i \pi g \frac { \omega}{\delta } \lambda + \frac{\pi}{2 } \partial_{\lambda } ( 1 - \lambda^2 ) \partial_{\lambda}.\ ] ] the asd is then given by @xmath61 while we did not succeed to find an exact analytical solution of this initial value problem , the function @xmath62 \nonumber \\ & & \exp [ \frac{\pi}{2 } ( \frac{\omega } { \delta } ) ^2 g^2 \int_{1/g}^{x/(l g ) } d s ( \exp ( \pi ( s - 1/g ) ) -1 ) ^2 ] \end{aligned}\ ] ] is a good approximation when @xmath63 , for arbitrary @xmath64 , and becomes exact for @xmath65 when @xmath66 . thus , the asd is obtained as @xmath67 with @xmath68 and @xmath69 fig .
2 shows a plot of the asd as a function of the scaled frequency @xmath4 and the scaled length @xmath70 .
a clear damping of the amplitude of oscillations accompanied by a shift of their phase is seen .
this shows that there is an effect of localization on level correlations . at smaller @xmath64 ,
the oscillations are damped more strongly , and the envelope approaches a gaussian decaying function . [ fig2 ] to see this in more detail , let us consider approximations of eq .
( [ result ] ) in various regimes of interest .
\1 . in the metallic regime @xmath71 for @xmath72
we obtain : @xmath73 where @xmath2 , which for @xmath74 reduces to the result obtained with the unitary wigner - dyson ensemble of random matrices@xcite . \2 . in the strongly localized regime @xmath75 ,
one obtains : @xmath76 rescaling @xmath77 , we note the similarity to eq .
( [ met ] ) .
this result shows clearly that the correlations between energy levels belonging to states which are spatially separated by more than the localization length @xmath1 are weak . as a result
, the asd shows only correlations with the period of twice the effective local energy level spacing @xmath78 of energy levels whose wave functions overlap spatially . as @xmath79 , the function
is dominated by the gaussian factor .
thus , @xmath80 exactly as @xmath81 , and @xmath82 . in fig .
3 a plot of @xmath83 is shown , where @xmath84 . the asd is decaying from @xmath85 to @xmath86 as the frequency is held constant at @xmath87 and one varies the length of the wire @xmath16 or the parameter @xmath88 in fig . 3 ,
compare with fig . 1 . [ fig3 ]
in summary , the asd is established as a tool to study the level statistics of disordered metals .
an analytical expression is obtained for the asd of a quasi-1-dim . disordered mesoscopic wire . at frequencies below the mean level spacing @xmath3
the asd approaches 1 like a gaussian for any value of the conductance g , and there is no information on localization in this regime .
this was pointed out by efetov @xcite when studying the weakening of level repulsion by localization .
it was stressed there , that the noncompact degrees of freedom are needed to describe localization that way . here
, it is shown that the information is rather contained in the large frequency correlations .
the asd shows a crossover to a strong damping of the correlations as the length of the wire exceeds the localization length @xmath1 , accompanied by the convergence of the period of the strongly damped oscillations to the constant @xmath89 .
thus , the wire can be thought of as effectively separated into localization volumes , as obtained earlier in refs . .
one may argue that , since the averaging over the impurity potential was done before normalization , the resulting function might contain different information than the one obtained by normalizing for a given impurity potential before doing the averaging@xcite .
the goal of this article is however to show that level statistics can be characterized with the simplest tool , @xmath90 .
now , it might become possible to address problems analytically , which could not be solved with the methods known so far , due to their complexity .
while the asd can not contain any information on the eigenfunctions of the system , we have seen that it contains enough information to characterize the energy level statistics .
the function @xmath91 may serve as a parameter characterizing localization : it is @xmath85 in the metallic regime and @xmath86 , when all states at all energies are localized .
it decays to approximately @xmath92 when the length of the wire coincides with the localization length , @xmath93 .
in addition , recently it has been shown that the asd can contain information not only on a metal - insulator - crossover , but also on a -transition , as demonstrated with the anderson model on a bethe lattice@xcite .
the author would like to thank uzy smilansky for drawing his attention to the asd as a tool to study level statistics , and thomas dittrich , konstantin efetov , dietrich klakow , d. e. khmelnitskii , igor lerner , daniel miller , vladimir prigodin and klaus ziegler for usefull discussions , and simon villain- guillot for critical reading of the manuscript .
present address , e- mail : [email protected] p. w. anderson , phys .
rev . * 109 * , 1492(1958 ) .
n. f. mott , w. d. twose , adv .
* 10 * , 107(1961 ) .
d. j. thouless , phys . rev . lett . * 39 * , 1167(1977 ) .
p. w. anderson , d. j. thouless , e. abrahams , d. s. fisher phys .
rev . b*22 * , 3519 ( 1980 ) ; w. weller , v. n. prigodin , y. a. firsov , phys .
stat . sol . *
110 * , 143(1982 ) ; n. dorokhov , sov .
jetp lett .
* 36 * , 318(1982 ) ; sov .
jetp * 58 * , 606(1983 ) . k. b. efetov , a. i. larkin , zh .
. fiz . * 85 * , 764(1983 ) ( sov .
jetp * 58 * , 444 ) . k. b. efetov , adv . phys . *
32 * , 53 ( 1983 ) ; _ supersymmetry in disorder and chaos _ ( cambridge university press , cambridge,1997 ) .
a. altland , d. fuchs , phys .
lett . * 74 * , 4269(1995 ) . o. agam , b. l. altshuler and a. v. andreev , phys . rev
. lett . * 75 * , 4389 ( 1995 ) .
a. altland , y. gefen , g. montambaux , phys .
* 76 * , 1130(1996 ) .
a. v. andreev and b. d. simmons , phys
lett . * 75 * , 2304 ( 1995 ) . s. kettemann , d. klakow , u. smilamsky , j. phys .
a,3643(1997 ) .
f. haake , m. kus , h.- j. sommers , h. schomerus , and k. zychowski , j. phys .
a * 29 * , 3641(1996 ) .
o. bohigas , m. giannoni , c. schmit , phys .
* 52 * , 1(1984 ) .
s. kettemann , unpublished(1998 ) .
f. wegner , z. phys .
b * 35 * , 207(1979 ) ; l. schaefer , f. wegner , z. phys .
b * 38 * , 113(1980 ) .
k. b. efetov , a. i. larkin , d. e. khmelnitskii , sov .
jetp * 52 * , 568(1980 ) ( zh . eksp .
. fiz . * 79 * , 1120(1980 ) ) .
s. kettemann , k. b. efetov , phys .
lett . * 74 * , 2547(1995 ) . k. ziegler , i. lerner , private discussion . | the autocorrelation function of spectral determinants ( asd ) is used to characterize the discrete spectrum of a phase coherent quasi- 1- dim . , disordered metal wire as a function of its length l at finite , weak magnetic field .
an analytical function is obtained , depending only on the dimensionless conductance @xmath0 , where @xmath1 is the localization length , the scaled frequency @xmath2 , where @xmath3 is the average level spacing of the wire , and the global symmetry of the system .
a metal - insulator crossover is observed , showing that information on localization is contained in the disorder averaged asd .
pacs- numbers : 72.15.rn,73.20.fz,73.23.-b |
the electronic structure of organic molecular crystals is strikingly different from the conventional inorganic semiconductors , such as si , in that the electronic polarization of the dielectric medium by charge carriers constitutes a major effect , with energy scale greater than transfer integrals or temperature @xcite .
the transport gap @xmath2 for creating a separated electron - hole pair has a substantial ( 12 ev ) polarization energy contribution @xcite and exceeds the optical gap by @xmath3 ev .
limited overlap rationalizes the modest mobilities of organic molecular solids .
devices such as light - emitting diodes , thin film transistors , or photovoltaic cells require charge transport and are consequently based on thin films , quite often deposited on metallic substrates @xcite .
organic electronics relies heavily on controlling films with monolayer precision , on forming structures with several thin films , and on characterizing the interfaces .
the positions of transport states and mechanisms for charge injection are among the outstanding issues for exploiting organic devices .
we focus here on the electronic polarization of crystalline thin films near surfaces and interfaces .
we find that electronic polarization , and hence @xmath2 , in a prototypical organic crystal is significantly different at a free surface , near a metal - organic interface , in thin organic layers , and in the bulk .
weak intermolecular forces characterize organic molecular crystals , whose electronic and vibrational spectra are readily related to gas - phase transitions @xcite . due to small transfer integrals ,
charge carriers are molecular ions embedded in the lattice of neutral molecules .
the transport gap @xmath2 in the crystal is derived from the charge gap for electron transfer in the gas phase , @xmath4 , which is the difference between the ionization potential and the electron affinity .
but crystals have electrostatic interactions even in the limit of no overlap , and charge carriers are surrounded by self - consistent polarization clouds .
in contrast to polaronic effects , electronic polarization is instantaneous and directly affects the positions of energy levels .
formation of polarization clouds is associated with stabilization energy @xmath5 for cations ( the `` holes '' ) and @xmath6 for anions ( the `` electrons '' ) .
the combination @xmath0 occurs in @xmath7 .
since coulomb interactions are long - ranged , polarization clouds extend over many lattice constants and @xmath1 depends on the proximity to surfaces and interfaces . 0.1 in schematic representation of charge - generation processes in crystalline molecular films .
( a ) ups / ipes generates a cation / anion at the outer surface , while charge injection from the substrate involves the layer next to the metal .
( b ) tunneling through a monolayer . dashed ovals in ( a ) and ( b ) represent image charges in the metal
. 0.1 in figure 1(a ) depicts schematically an ultraviolet photoemission ( ups ) process where the ejected electron leaves behind a molecular cation in the outermost layer . in inverse photoelectron spectroscopy ( ipes ) , the surface is irradiated with low - energy electrons and the emitted photon is detected when an electron is captured to form a molecular anion .
ups data is increasingly available from sub - monolayer to @xmath8100 films @xcite , while the ipes data is more limited .
the combination of ups with ipes yields @xmath2 directly , with @xmath1 about 12 ev in representative organic materials used in devices @xcite . as sketched in fig .
1(b ) , tunneling electron spectroscopy gives @xmath2 as the interval between the differential conductance peaks when the potential of the tip matches either the electron or the hole transport levels of the film .
0.1 in ( a ) composite ups / ipes spectra as a function of ptcda thickness on ag .
energy scales are aligned by measuring the fermi energy by ups and ipes on ag prior to ptcda deposition .
( b ) di / dv(v ) sts spectra of filled and empty states recorded for a monolayer ( bottom ) and a 2 - 3 molecular layer ( top ) film of ptcda deposited on au .
the corresponding stm images of the films are shown .
the curves were recorded at the same tunneling setpoints as the corresponding area scans . 0.1 in figure 2(a ) shows ups and ipes spectra of ptcda ( perylenetetracarboxylic acid dianhydride ) on silver .
ptcda is an excellent former of crystalline films whose structures are close , though not identical , to having a ( 102 ) plane of the bulk crystal in contact with the substrate @xcite .
the measured transport gap on thick ( @xmath9 ) films is @xmath10 ev on ag , in excellent agreement with ptcda on au @xcite .
we use ptcda on ag rather than au because the au(5d ) levels interfere with ups of a monolayer .
@xmath11 includes a 0.2 ev intramolecular vibrational contribution @xcite that reduces the ups / ipes gap .
this correction is the same for all ptcda films .
careful analysis of peak positions indicates @xmath8200 mev shifts of both the cation level ( ups ) and anion level ( ipes ) between mono- and multilayer films . figure 2(b ) shows the results of scanning tunneling spectroscopy ( sts ) on a monolayer ( lower spectrum ) and a 2 - 3 layer film ( upper spectrum ) of ptcda on au(111 ) .
these di / dv(v ) spectra represent the density of filled and empty states involved in tunneling out of and into the layer , respectively .
each spectrum is the average of 25 spectra recorded at various points on highly ordered molecular layers .
high - resolution scanning tunneling microscopy ( stm ) images of these layers , taken concomitantly with sts , show the characteristic difference @xcite between monolayers , in which molecules are in contact with , and parallel to , the au surface and appear symmetric , and second and subsequent layers , in which the tilt angle of molecules introduces an asymmetry in their stm image .
the tunneling spectrum of the monolayer shows peaks leading to @xmath12 ev for the energy difference between adding an electron or hole .
remarkably , this energy difference increases by about 0.25 ev on the 2 - 3 layer spectrum , in excellent agreement with the intermediate ups / ipes spectra of a 16 film ( fig .
each peak shifts away from the fermi level by a roughly equal amount with increasing coverage .
we note that @xmath11 and @xmath13 show no dependence on dipoles at the metal - organic interface , which have opposite signs for ptcda on ag and au @xcite .
the dipole at the ptcda / ag interface corresponds to electron transfer from the metal to interface molecules , which gives rise to the filled gap states at @xmath14 and @xmath15 ev on the 4 ups spectrum .
the broad feature around @xmath14 ev on the 4ptcda / au sts spectrum corresponds to the au surface state , which is not eliminated by the deposition of organic molecules @xcite .
reduced polarization energy at surfaces has long been appreciated on general grounds @xcite . as anticipated and found for thin films of anthracene on au , @xmath5 is about 200 mev smaller in the surface layer @xcite . with similar reductions expected for @xmath6 , the transport gap at the surface
is _ increased _ some 400 mev from its bulk value , @xmath2 .
the gap increases near surfaces because vacuum is not polarizable .
conversely , @xmath2 decreases near organic - metal interfaces due to the high polarizability of metals .
these opposite contributions partially cancel in very thin films on metal substrates .
we note that @xmath2 , which has been the focus of theoretical study @xcite and governs bulk transport , is beyond the reach of the surface experimental techniques that access the outermost layers of the material and thus reflect @xmath11 .
similarly , the transport gap @xmath16 of the layer next to the metal is relevant for charge injection ( fig .
a gap reduction of several hundred mev is significant since it directly alters interface barriers .
no theoretical treatment exists for the electronic polarization near surfaces and interfaces , which require an accuracy of @xmath8100 mev .
methods to estimate @xmath17 in the bulk have been developed , primarily for the acenes , based on the microelectrostatics of polarizable points that represent organic molecules @xcite .
dielectric response of a neutral surface has also been studied @xcite .
we have recently developed an approach based on the analysis of charge redistribution in organic molecules @xcite , and demonstrated accurate calculations of @xmath5 , @xmath6 , and energies of ion pairs in bulk ptcda and anthracene crystals @xcite .
we apply here the same approach to calculate polarization in thin organic films . in the zero - overlap limit ,
molecules comprising the crystal are quantum - mechanical objects interacting by classical forces . a self - consistent problem can be formulated @xcite that treats molecules rigorously in the external fields of all other molecules .
we describe charge redistribution in organic molecules in terms of the atom - atom polarizability tensor @xmath18 that relates a change in the partial charge at an atom @xmath19 due to the electrostatic potential @xmath20 at atom @xmath21 : @xmath22 @xmath23 are the atomic charges in an isolated molecule .
the tensor @xmath18 is a natural extension of the similar quantity in @xmath24-electron theory @xcite .
we compute @xmath18 using indo / s @xcite , which is a semiempirical hamiltonian designed for spectroscopic molecular properties .
@xmath18 describes the major , `` charge - induced '' part of molecular polarizability @xmath25 , which is augmented to reflect the actual polarizability @xmath26 by introducing induced atomic dipoles @xmath27 and distributing the difference @xmath28 over 38 atoms of ptcda in the spirit of submolecular methods @xcite .
self - consistent equations for @xmath29 and @xmath27 are then solved for increasing cluster sizes of mesoscopic dimensions ( @xmath8100 ) , and the macroscopic limits are found . for neutral lattices
the approach has yielded accurate anisotropic dielectric tensors of two representative organic crystals @xcite .
we use identical molecular inputs here to model ptcda films as infinite slabs terminated by ( 102 ) planes next to a vacuum and a metal , respectively , as sketched in fig . 1 .
the metal is taken as a constant - potential plane at @xmath30 , a distance @xmath31 from the innermost molecular layer .
image charges and dipoles at @xmath32 ensure that the potential @xmath33 .
any @xmath34 leads to the same result for @xmath0 , since one charge is stabilized and the other is destabilized .
the metal - organic separation @xmath31 is a model parameter that is related to van der waals radii . 0.1 in electronic polarization @xmath0 at the outer layer of an @xmath35-layer film ( fig .
1 ) at separation @xmath31 from the metal .
the @xmath1 values at the free surface ( @xmath36 ) , a monolayer ( @xmath37 ) , free - standing films ( @xmath38 ) , and the bulk are indicated .
a pill - box geometry with @xmath39 is used .
the inset shows the transport gap , @xmath40 , across a 10-layer film with @xmath41 next to the metal and @xmath42 at the outer surface . 0.1 in we use a pill - box geometry , defined by the radius @xmath43 ( sketched in fig .
3 ) with the ion placed in one of the layers near the center of the box .
we initially find self - consistent atomic charges and induced dipoles @xmath44 and @xmath45 , @xmath46 in a neutral film of @xmath35 layers , where translational symmetry gives rapid convergence .
we then consider pill - boxes of increasing diameter @xmath47 to find self - consistent @xmath48 and @xmath49 for every molecule @xmath50 in the pill - box .
the largest systems ( @xmath51 , @xmath52 ) contain @xmath82400 ptcda molecules and their images , and larger values of @xmath43 can be used .
@xmath5 is the energy difference between a neutral film and one containing a cation .
@xmath6 is the corresponding energy for an anion in the same position .
we note that the energy of an infinite film is extensive , but the difference is finite and can be evaluated @xcite .
all data reported below is in the limit @xmath53 .
@xmath1 thus depends on which layer contains the ion .
for the ion in the outermost molecular layer in the limit @xmath36 we obtain the free ( 102 ) surface polarization @xmath54 ev .
this limit does not depend on @xmath31 or the metal .
@xmath55 is 0.41 ev less than the bulk value @xmath56 ev @xcite .
the difference is consistent with experimental estimates @xcite and corresponds to the surface correction @xmath57 , where the value inferred @xcite from ups and ipes spectra was @xmath58 .
figure 3 gives results for ptcda layers of finite thickness .
unlike bulk or free - surface calculations , which are essentially parameter - free , the finite layer data depends on @xmath31 , which is the only way the metal enters our idealized model .
the reasonable value @xmath59 places the metal plane one lattice spacing @xmath60
from the innermost layer , which is also consistent with van der waals radii , and 10 % variations of @xmath31 are shown in fig .
we also computed @xmath1 at @xmath61 , @xmath62 and @xmath63 , and extrapolated as @xmath64 .
the limit @xmath38 gives the polarization at the surface of a free - standing film of @xmath35 layers .
all curves in fig .
3 converge to the free - surface value @xmath65 ev . using the curve @xmath59
we find the single monolayer value @xmath66 ev , which corresponds to the tunneling spectroscopy setup ( fig .
we see that for a monolayer on the metal surface , the polarization energy is indeed close to the bulk value in line with the expected cancellation discussed above .
the difference @xmath67 ev agrees with the experimental @xmath680.50 ev for ptcda on ag or au .
in fact , the agreement is slightly better since ups / ipes data is for the the films of finite thickness @xmath69 .
the free - standing monolayer has @xmath70 ev , which is about @xmath71 .
such a big polarization is consistent with large in - plane polarizability of ptcda molecules .
these results for @xmath1 are summarized for comparison in table i. lr bulk & 1.82 ev + free surface & 1.41 ev + monolayer , free - standing & 0.64 ev + monolayer on metal ( @xmath59 ) & 1.93 ev + surface of a bilayer on metal ( @xmath59 ) & 1.73 ev + layer next to metal , thick film ( @xmath59 ) & 2.21 ev + + analysis of ups data for films of the electron - transport molecule alq@xmath73 [ tris(8-hydroxy - quinoline)-aluminum ] on silver @xcite yields strikingly similar conclusions : the transport gap for a monolayer on metal substrate , @xmath74 , is equal to the inferred bulk value , and is about 400 mev narrower than the gap in the outermost surface layer of a thick film , @xmath75 . also , the inferred gap in the innermost layer of a thick film on metal surface , @xmath16 , is about 400 mev less @xcite than @xmath74 .
figure 4 shows the variation of the transport gap @xmath2 across a 10-layer ptcda film on a metal substrate ( @xmath59 ) .
we note that surface effects extend several layers into the sample .
the long - range nature of surface polarization has been largely neglected .
an influential early ups study of 20 vapor - deposited anthracene films on au ascribed the 200 mev shift of @xmath5 to the single outermost layer @xcite .
the additivity of polarization contributions suggested @xcite for alq@xmath73 is tacitly based on short - range interactions .
greater polarizability next to the metal is consistent with strong charge confinement to interfaces , as inferred recently for pentacene field - effect transistors with remarkable electronic characteristics @xcite . in general , the 400 mev increase of @xmath72 at the metal - organic interface
is not shared equally by the electron and hole .
the stabilization of either carrier by roughly 200 mev is important for matching transport levels in injection .
weak overlap in molecular solids constitutes , in fact , a significant simplification over covalent bonding in inorganic semiconductors . even though charge transport is critical for electronic applications , the picture of localized carriers is the proper zeroth - order approximation , to which the overlap ( i.e. kinetic energy ) should be considered as a perturbation .
this does not by itself rule out the band - like description .
rather , charged quasiparticles are to be understood as surrounded by polarization clouds , likely affecting the bandwidths .
our results for @xmath5 and @xmath6 are exclusively electronic .
lattice relaxation around charges are considered to be small corrections ( @xmath810% ) on general grounds @xcite .
the idealized model of image charges does not depend on the metal s fermi energy or on surface dipoles , whose shifts cancel in @xmath0 .
it also ignores surface states or surface reactions that are known to occur at specific organic / metal inferfaces @xcite .
since the self - consistent calculation requires the film s structure , it is not directly applicable to amorphous or structurally uncharacterized films . | the electronic polarization energies , @xmath0 , of a ptcda ( perylenetetracarboxylic acid dianhydride ) cation and anion in a crystalline thin film on a metallic substrate are computed and compared with measurements of the ptcda transport gap on gold and silver .
both experiments and theory show that @xmath1 is 500 mev larger in a ptcda monolayer than in 50 films .
electronic polarization in systems with surfaces and interfaces are obtained self - consistently in terms of charge redistribution within molecules .
+ 2 [ ] |
we would like to thank giovanni cantatore , aaron chou , marin karuza , axel lindner , giuseppe ruoso , pierre sikivie , and karl van bibber for interesting discussions .
f. wilczek , phys .
* 49 * ( 1982 ) 1549 .
y. chikashige , r. n. mohapatra and r. d. peccei , phys .
b * 98 * ( 1981 ) 265 .
g. b. gelmini and m. roncadelli , phys .
b * 99 * ( 1981 ) 411 .
c. t. hill and g. g. ross , nucl .
b * 311 * ( 1988 ) 253 .
y. liao , arxiv:0704.1961 [ hep - ph ] .
p. sikivie , phys .
* 51 * ( 1983 ) 1415 [ erratum - ibid .
* 52 * ( 1984 ) 695 ] .
g. raffelt and l. stodolsky , phys .
d * 37 * ( 1988 ) 1237 .
a. a. anselm , yad .
* 42 * ( 1985 ) 1480 .
m. gasperini , phys . rev
. lett .
* 59 * ( 1987 ) 396 .
k. van bibber , n. r. dagdeviren , s. e. koonin , a. kerman and h. n. nelson , phys .
lett . * 59 * ( 1987 ) 759
. g. ruoso _ et al . _
[ bfrt collaboration ] , z. phys .
c * 56 * ( 1992 ) 505 .
r. cameron _ et al . _
[ bfrt collaboration ] , phys .
d * 47 * ( 1993 ) 3707 .
k. ehret _ et al .
_ [ alps collaboration ] , `` production and detection of axion - like particles in a hera dipole magnet : letter - of - intent for the alps experiment , '' arxiv : hep - ex/0702023 . c. rizzo for the [ bmv collaboration ] , 2nd ilias - cern - cast axion academic training 2006 , http://cast.mppmu.mpg.de/ s. l. adler , annals phys . * 67 * ( 1971 ) 599 .
s. l. adler , j. phys . a * 40 * ( 2007 ) f143 [ arxiv : hep - ph/0611267 ] .
s. biswas and k. melnikov , phys .
d * 75 * ( 2007 ) 053003 [ arxiv : hep - ph/0611345 ] .
l. maiani , r. petronzio and e. zavattini , phys .
b * 175 * ( 1986 ) 359 .
m. ahlers , h. gies , j. jaeckel and a. ringwald , phys .
d * 75 * ( 2007 ) 035011 [ arxiv : hep - ph/0612098 ] .
e. masso and j. redondo , jcap * 0509 * ( 2005 ) 015 [ arxiv : hep - ph/0504202 ] .
p. jain and s. mandal , int .
j. mod .
d * 15 * ( 2006 ) 2095 [ arxiv : astro - ph/0512155 ] .
e. masso and j. redondo , phys .
* 97 * ( 2006 ) 151802 [ arxiv : hep - ph/0606163 ] .
j. jaeckel , e. masso , j. redondo , a. ringwald and f. takahashi , phys . rev .
d * 75 * ( 2007 ) 013004 [ arxiv : hep - ph/0610203 ] .
r. n. mohapatra and s. nasri , phys .
lett . * 98 * ( 2007 ) 050402 [ arxiv : hep - ph/0610068 ] .
p. jain and s. stokes , arxiv : hep - ph/0611006 .
a. ringwald , arxiv : hep - ph/0112254 .
r. rabadan , a. ringwald and k. sigurdson , phys .
* 96 * ( 2006 ) 110407 [ arxiv : hep - ph/0511103 ] .
u. ktz , a. ringwald and t. tschentscher , arxiv : hep - ex/0606058 .
p. sikivie , d. b. tanner and k. van bibber , phys .
* 98 * ( 2007 ) 172002 [ arxiv : hep - ph/0701198 ] .
a. v. afanasev , o. k. baker and k. w. mcfarlane , arxiv : hep - ph/0605250 . | we present a scheme to extend the sensitivity of axion - photon regeneration experiments towards larger masses with the help of properly chosen and placed phase shift plates .
many proposals to embedd the standard model of particle physics into a more general , unified framework predict a number of new very light particles which are very weakly coupled to ordinary matter .
typically , such light particles arise if there is a global continuous symmetry that is spontaneously broken in the vacuum a notable example being the axion @xcite , a pseudoscalar particle arising from the breaking of a u(1 ) peccei - quinn symmetry @xcite introduced to explain the absence of cp violation in strong interactions .
other examples of light spin - zero bosons beyond the standard model are familons @xcite , majorons @xcite , the dilaton , and moduli , to name just a few .
we will call them axion - like particles , alps , in the following . at low energies , the coupling of such an alp , whose corresponding quantum field we denote by @xmath0 , to photons
is described by an effective lagrangian , @xmath1 where @xmath2 ( @xmath3 ) is the ( dual ) electromagnetic field strength tensor ) applies for a pseudoscalar alp , i.e. a spin - zero boson with negative parity . in the case of a scalar alp , the @xmath4 in eq .
( [ em_pseudoscalar ] ) is replaced by @xmath5 . for the more general case where @xmath0 ceases to be an eigenstate of parity@xcite ,
see ref .
@xcite . ] and @xmath6 is the mass of the alp .
correspondingly , in the presence of an external magnetic field , a photon of energy @xmath7 may oscillate into an alp of small mass @xmath8 , and vice versa @xcite .
the exploitation of this mechanism is the basic idea behind alp - photon regeneration
sometimes also called `` light shining through a wall '' experiments @xcite ( cf . fig . [
fig : ph_reg ] ) .
namely , if a beam of photons is shone across a magnetic field , a fraction of these photons will turn into alps .
this alp beam could then propagate freely through an optical barrier without being absorbed , and finally another magnetic field located on the other side of the wall could transform some of these alps into photons apparently regenerating these photons out of nothing .
a pioneering experiment of this type was carried out in brookhaven by the brookhaven - fermilab - rochester - trieste ( bfrt ) collaboration , using two prototype magnets for the colliding beam accelerator @xcite . presently , there are worldwide several second generation alp - photon regeneration experiments under construction or serious consideration ( cf .
table [ tab : exp ] ; for a review , see refs .
@xcite ) .
these efforts are partially motivated by the report from the pvlas collaboration of evidence for a non - zero apparent rotation of the polarization plane of a laser beam after passage through a magnetic field @xcite . while the size of the observed effect greatly exceeds the expectations from quantum electrodynamics @xcite
, it is compatible with the expectations @xcite arising in the context of a photon - alp oscillation hypothesis .
indeed , the rotation observed by pvlas can be reconciled with the non - observation of a signal by bfrt , if there exists an alp with a mass @xmath9 mev and a coupling @xmath10 gev@xmath11 @xcite . although these parameter values seem to be in serious conflict with bounds coming from astrophysical considerations , there are various ways to evade them @xcite .
therefore , it is extremely important to check the alp interpretation of pvlas by purely laboratory experiments @xcite .
moreover , it would be nice if in this way one might ultimately extend the laboratory search for alps to previously unexplored parameter values ( see also ref .
@xcite ) . in this letter
, we propose a method to extend the sensitivity of the planned photon - regeneration experiments to higher alp masses .
.experimental parameters of upcoming photon regeneration experiments : magnetic fields @xmath12 and their length @xmath13 on production ( @xmath14 ) and regeneration ( @xmath15 ) side ( cf . fig .
[ fig : ph_reg ] ) . [ cols="<,<,^,<",options="header " , ] [ tab : exp ] let us start with an outline of the calculation of the photon @xmath16 alp conversion probability @xmath17 , to lowest order in the coupling @xmath18 . as emphasized in ref .
@xcite , this calculation amounts to solving the classical field equations following from eq .
( [ em_pseudoscalar ] ) , @xmath19 to lowest order in @xmath20 , where @xmath21 is the linear dimension associated with the extent of the magnetic field in eqs .
( [ em_pseudoscalar_fieldeq ] ) , ( [ lead_sol ] ) , and ( [ onedimensional ] ) is replaced by @xmath22 . ] .
this can be done by neglecting the modification of the electromagnetic field due to the presence of the pseudoscalar field ( through the right hand side of the first equation above ) .
solving for @xmath0 in the second equation yields @xcite @xmath23 where the energy @xmath7 and the modulus of the three - momentum @xmath24 are related by @xmath25 .
of the ( pseudo-)scalar versus its mass @xmath6 .
iso - contour of the regeneration probability @xmath26 , for the parameters of the alps experiment , i.e. magnetic fields @xmath27 t , over a length @xmath28 m , exploiting a green ( @xmath29 nm ) photon beam , corresponding to @xmath30 ev , in vacuum . also shown in red
are the 5 sigma allowed regions @xcite from pvlas data on rotation @xcite plus bfrt data on rotation , ellipticity , and regeneration @xcite plus q&a data on rotation @xcite .
[ regprop_stan ] ] this solution simplifies even more if we specialize to the usual experimental configuration of a laser photon beam send along the @xmath31-axis with fixed linear polarization in the @xmath32 direction .
if the transverse extent of the magnetic field is much larger than that of the laser beam , the problem is effectively one - dimensional . in one dimension and taking into account only alps that propagate into the positive @xmath31-direction , eq
. becomes , @xmath33 inserting in eq .
( [ onedimensional ] ) furthermore the appropriate plane wave form @xmath34 for the electric field of the laser beam and assuming , as realized in all the proposed experiments , a magnetic field with fixed direction along the @xmath32-axis and possibly variable ( as a function of @xmath31 ) magnitude , @xmath35 , one ends up with the solution ) applies also in the case of a scalar alp , if the magnetic field direction is chosen to point into the @xmath36 direction , @xmath37 ( or , alternatively , if the polarization of the laser is chosen to point in the @xmath36 direction ) . ]
@xmath38 where @xmath39 is the momentum transfer to the magnetic field , i.e. the modulus of the momentum difference between the photon and the alp .
the probability that a photon converts into an axion - like particle and vice versa can be read off from eq .
( [ sol_stand ] ) and reads @xcite @xmath40 which reduces , for a constant magnetic field , @xmath41 , of linear extension @xmath21 , to @xmath42 clearly , in the experimental setup considered , the maximum conversion probability , @xmath43 , is attained at small momentum transfer , @xmath44 , corresponding to a small alp mass .
for this mass range , the best limits are obtained in a straightforward manner by exploiting strong and long dipole magnets , as they are used for storage rings such as hera @xcite or lhc @xcite , cf .
the experiments alps @xcite and osqar @xcite , respectively ( see table [ tab : exp ] )
. however , for larger masses , the sensitivity of this setup rapidly diminishes .
we illustrate this in fig .
[ regprop_stan ] , which displays an iso - contour of the light shining through a wall probability in the @xmath18-@xmath6 plane , exploiting the experimental parameters of the alps experiment @xcite . clearly , for this setup
, the parameter region in @xmath18 vs. @xmath6 suggested by the combination of bfrt plus q&a exclusion and pvlas evidence can not be probed .
this is even more dramatic for the osqar experiment , which exploits an lhc magnet .
moreover , increasing the refraction index by filling in buffer gas does not help since it works in the wrong direction ( contrary to the claim .
the momentum transfer ( [ mom_trans ] ) reads then @xmath45 .
the second term in this expression has the opposite sign as the corresponding term in ref .
@xcite .
correspondingly , one would need a buffer gas with refraction index less than unity , i.e. a plasma , in order to decrease @xmath46 ( and thereby maximize the conversion probability ( [ sinsuppress ] ) ) rather than to increase it .
a.r .
would like to thank aaron chou for pointing out the correct sign . ] in the alps letter of intent @xcite ) . a simple possibility to probe the mev region in the alps setup is to reduce the effective length of the magnetic field region both on the production and detection side of the magnet by shortening the beam pipe on both sides . as can be seen in fig .
[ regprop_short ] , this possibility enables to extend the mass region probed by the experiment , however at the expense of sensitivity : one looses about one order of magnitude in the light shining through a wall probability . , but with reduced lengths of the magnetic field region .
note , that the regeneration probability is reduced by a factor of @xmath47 .
[ regprop_short ] ] another idea to extend the sensitivity towards larger alp masses was introduced in ref .
@xcite .
there , it was shown that a segmentation of the magnetic field into regions of alternating polarity gives a form factor @xmath48 that peaks at a nonzero value of @xmath46 , thereby giving sensitivity to higher - mass pseudoscalars .
in fact , the conversion probability ( [ prop_stan ] ) reads @xcite , in a magnet with @xmath49 segments of alternating field direction ( but the same magnitude @xmath50 ) , @xmath51 \nonumber & = & \frac{g^2\,b_0 ^ 2}{q^2 } \left\ { \begin{array}{clc } \sin^2\left ( q\ell/2 \right ) \tan^2\left ( q\ell/(2n ) \right ) & { \rm for } & n\ { \rm even } \\ \cos^2\left ( q\ell/2 \right ) \tan^2\left ( q\ell/(2n ) \right ) & { \rm for } & n\ { \rm odd } \end{array } \right .
\,,\end{aligned}\ ] ] where @xmath52 is the length of each of the @xmath49 segments . for @xmath53 ,
this indeed gives rise to more sensitivity at non - zero values of @xmath46 . in this letter
, we will introduce a similar , but more practical possibility based on the use of phase shift plates .
the idea is very simple . from our starting point ,
eq .
( [ onedimensional ] ) , we can see that what counts is actually @xmath54 .
the configuration based on @xmath49 alternating magnetic fields is therefore equivalent to a configuration with non - alternating magnetic field , however with @xmath55 retardation plates with phase shift @xmath56 ( `` @xmath57 '' plates ) inverting the sign of the electric field , placed equidistantly over the length @xmath21 of the magnet . in this case
we have alternating signs of @xmath58 , where @xmath59 is the angle between @xmath60 and @xmath61 , instead of alternating signs of the magnetic field . but both cases have an identical profile of @xmath54 . in fig . [ regprop_ret ]
, we show that with a proper choice of the number and positions of such phase shifters , alps should easily cover the region of parameter space suggested by pvlas + bfrt + q&a .
the same applies for osqar . . here
, we used one phase shift ( `` @xmath57 '' ) plate each in the middle of the generation and the regeneration sides .
[ regprop_ret ] ] let us now get a more intuitive understanding of how this works and see how we can do even better .
the crucial part in eq . is the integral @xmath62 for a constant magnetic field of length @xmath21 the oscillating factor @xmath63 suppresses the integral compared to the massless case with @xmath64 , where the integral is simply @xmath65 this suppression arises because coherent production of alps works only if the alp and the photon are in phase .
the factor @xmath66 accounts for the phase difference between alp and photon .
to improve the situation one would want to bring photon and alp back into phase with each other .
this can be achieved in a simple way by the introduction of phase shift plates . a simplified picture of the phase correction process is given in fig .
[ illu ] . at the beginning ,
photon ( red ) and alp ( black ) are in phase . however , due to its mass , the alp has a slightly larger wavelength than the photon .
after a few oscillations photon and alp are more and more out of phase
. then we insert the phase shift plate ( turquoise ) . with refractive indices @xmath67 we can not make the wavelength larger .
so it is not possible to `` delay '' the photon until the alp has caught up .
what we can do , however , is to is to _ increase _ the phase difference between alp and photon such that it is exactly @xmath68 ( the photon does an extra wiggle in fig .
[ illu ] . ) .
now , a phase shift of @xmath68 is exactly equivalent to a phase shift of @xmath69 .
photon and alp are in phase again .
therefore , we can keep photon and alp in phase over quite long distances simply by inserting a suitable phase shift plate whenever the phase difference becomes too large , and we get coherent production over the whole length of the magnet . .
this is corrected by introduction of a phase shift plate that causes the photon to get an extra phase @xmath70 . in other words
the plate causes the photon to complete the extra wiggle .
[ illu ] ] let us now understand more quantitatively how this works .
to derive eqs .
and we have assumed that the photon is a plane wave .
therefore , we can identify @xmath71 as the phase difference between the photon wave and the alp wave at the point @xmath72 .
in general we should write ( cf .
eq . )
@xmath73 where @xmath74 are the phases of the photon and the alp fields , respectively .
let us imagine a situation where we insert @xmath55 thin , the mixing between photon and alp is effectively switched off compared to the mixing in vacuum .
photon and alp simply propagate through the plate without changing their amplitudes ( the phases change , of course ) .
in other words , the thickness of the plates has to be subtracted from the total length of the production or regeneration region .
that is why we require thin plates . for practical purposes ,
this is a rather mild constraint . for @xmath75 , the thickness of the plates required for a phase shift of
the order of @xmath68 is only @xmath76 , which is tiny compared to the typical lengths of the production / regeneration regions which are of the order of a few m. ] , non - reflective plates that accelerate the photon phase by @xmath77 at equidistant places @xmath78 , @xmath79 , in a constant magnetic field of length @xmath21 .
the plates affect only the photon .
the alp phase remains unaffected .
therefore , we have , @xmath80 inserting this into eq . , we find @xmath81 we can now choose the number of plates @xmath49 and the phase shift @xmath77 according to the recipe described above .
first we choose @xmath49 large enough such that @xmath82 and then we choose @xmath77 such that the phase difference that has accumulated over @xmath83 is `` completed '' to @xmath68 , @xmath84 evaluating eq . in the limit
@xmath85 one finds @xmath86 and we have coherent production over the whole length @xmath21
. the potential of this approach is demonstrated in fig .
[ optim1 ] for the example of the alps experiment . in the optimized mass region
we get more than ten times as many regenerated photons as we would get if the length of the magnet is reduced as in fig .
[ regprop_short ] . .
in the left figure , we have used no phase correction ( red ) , one plate with @xmath87 ( green ) , and one plate with the optimal choice of @xmath77 according to eq . for @xmath88 ( blue ) .
the black curve is for 20 plates with the optimal choice of @xmath77 . in the right figure
, we have the same but with @xmath89 plates for the green and blue curves .
[ optim1],title="fig : " ] . in the left figure , we have used no phase correction ( red ) , one plate with @xmath87 ( green ) , and one plate with the optimal choice of @xmath77 according to eq . for @xmath88 ( blue ) .
the black curve is for 20 plates with the optimal choice of @xmath77 . in the right figure
, we have the same but with @xmath89 plates for the green and blue curves .
[ optim1],title="fig : " ] another practical advantage of this method is that we can scan through a whole mass range . performing several measurements with different phase shift plates we can always choose for each @xmath46 , i.e. for each @xmath90 , plates with an appropriate @xmath77 such that it is close enough to its optimal value , @xmath91 for
an infinite number of plates this would allow to extend the mass range all the way to the frequency @xmath7 of the photons becomes imaginary . ] . in practice
, we can insert only a finite number of phase shift plates and eq .
can not be fulfilled for too large masses .
but , already a small number of plates leads to a remarkable increase of the sensitivity for higher masses , as we can see from fig .
[ optim2 ] . .
this figure demonstrates the potential for scanning through a whole range of masses by choosing the right @xmath77 for each @xmath90 . the red curve is the sensitivity without phase correction .
the black curve is obtained by using three plates but scanning through a whole range of @xmath77 . in other words to obtain this curve one would insert the plates . measure .
change the plates to a slightly different value of @xmath77 and measure again .
this is repeated for all values of @xmath77 in the range @xmath92 $ ] .
[ optim2 ] ] _ in summary : _ so called `` light shining through a wall '' experiments are a promising tool to search for light particles coupled to photons . in this note we have shown how the reach of such an experiment can be extended towards larger masses by inserting properly chosen phase shift plates . although our explicit discussion is for the case of spin-0 axion - like particles the method works in general for particles exhibiting photon - particle - photon oscillations . |
since the beginning of the century the analysis regarding entanglement properties of ground states of condensed matter systems has been on an equal footing with the ginzburg - landau approach to complex quantum many - body systems @xcite .
further analyses have revealed the monotonic scaling of the von neumann entanglement entropy in the ground state of spin models @xcite the so - called _ area - law _ as well as profound relations to conformal field theory ( cft ) @xcite .
the von neumann entropy is the most celebrated member of the renyi entanglement entropies .
for example , the von neumann is so important because of determining the scaling properties of numerical algorithms based on matrix product states @xcite , but in terms of this research line it is particular useful because it lends itself to a useful tool to determine the continuous or discontinuous nature of a phase transition @xcite and to estimate quasi - long - range order in low - dimensional systems @xcite .
above all there is a growing awareness that the entanglement properties the whole is greater than the sum of its parts @xcite may provide the most fundamental characterization of quantum phases of matter .
this is particularly true for cases concerning the presence of phases even in the absence of local order parameters .
examples of this phenomenon are represented by quantum phase transitions in which the `` ordered phase '' are characterized by a non - vanishing topological component of the ground - state entanglement entropies @xcite . in this context
, the entanglement spectrum and the topological component of the rnyi entropies are being actively investigated in relation to bose - hubbard spin liquids @xcite , frustrated models on nontrivial lattice geometries @xcite , non - abelian fractional hall systems @xcite , and low - dimensional gapless models @xcite . in spite of recent exemplary findings concerning the analysis of bipartite entanglement , there still remains a number of unexplored avenues .
firstly , there is an undeniable rich structure to multipartite quantum correlations for which very little is understood .
secondly , since inseparability is known to be an np - hard problem @xcite and is devoid of computable necessary and sufficient criteria , this leaves sufficient - only criteria as the only way to tackle inseparability .
however , note that many examples such as measurement based quantum computation @xcite or many other algorithms ( e.g. ref .
@xcite ) suggest that these properties are nonetheless important . in refs .
@xcite a general framework to construct sufficient - only criteria was introduced ; in particular , certain inequalities that have to be fulfilled in specific types of partially separable states .
if these inequalities are violated they prove partial inseparability , i.e. the presence of entanglement .
this may in turn be used to show the extreme situation of a multipartite state that is not biseparable with respect to any bipartition , commonly referred to as a genuine multipartite entangled state ( detailed definition in the next section ) .
there are only few papers dealing with partial separability of gme in many body system , e.g. @xcite . in ref .
@xcite the authors presented a novel method to explore genuine multipartite entanglement and partial inseparability in many - body - systems by means of macroscopic observables only , such as energy .
the idea is to compute the mean energy value of a given hamiltonian and to compare the value with the mean energy value minimized over all , e.g. , bipartite states or , generally @xmath3-separable states .
since the mean energy value is bound from below , the difference or gap between the energy value minimized over all k - separable states ( choosing , e.g. , @xmath4 equals gme ) and the ground state energy values serves as a detection tool of partial separability .
this introduced _ gme gap _ " and _ k - entanglement gap _ " , detects large areas of genuine multipartite entanglement and partial entanglement in typical many body states , however , it requires non - separable ground states . in another recent paper @xcite the genuine tripartite entanglement of the anisotropic @xmath5 spin model was analyzed by applying proper constructed entanglement witnesses . in this paper
we shall follow a promising path of possibilities by applying the multipartite criterion introduced in ref .
@xcite directly to a subsystem of an one dimensional chain of spin-@xmath6 particles described by the @xmath0 model in the thermodynamic case and the finite size case . in this respect
, the paper takes the first steps to analyzing the properties of quantum multi - body systems via detection tools for multipartite entanglement .
we demonstrate that this method opens new avenues for successfully addressing properties of a condensed matter system such as quantum phase transitions or scaling properties .
note that few days after we put this paper on the net , two other related papers appeared on the arxiv server .
namely , in ref . @xcite the authors investigate quantum discord in the @xmath0-model and in ref .
@xcite the authors investigated also genuine multipartite entanglement in the @xmath7-model , however , different to this present paper their tool for detecting multipartite entanglement is based on a superset of the biseparable states . the paper is organized as follows . in sec .
[ meb ] we recall the definitions of partial separability and genuine multipartite entanglement . in sec .
[ model ] we introduce the one dimensional spin-@xmath6 @xmath0 model in an external magnetic field with periodic boundary conditions .
the principal characteristics are analyzed for both cases , i.e. for a finite size chain and in the thermodynamic limit .
we have chosen to focus our analysis on the @xmath0 model since it represents a milestone in the field of the quantum entanglement relating to many body systems ( being one of the few models for which it is possible to evaluate analytically the reduced density matrix of the ground states regardless the size of the system @xcite and it turns out to be , e.g , a realistic model for helium absorption on metallic surfaces @xcite or can be realized with bosonic atoms loaded in the p - band of an optical lattice in the mott regime @xcite ) . in sec .
[ panoramic ] we analyze the presence of the genuine multipartite entanglement in the @xmath0 model paying particular attention to what happens close to the critical and factorization points .
the theory of quantum entanglement describes the coherence properties of composite quantum systems .
indeed , the subtleties of composite quantum systems are such that there is no simple way to relate coherence properties of a single system to a larger composite system .
however , one elementary method to characterize the coherence properties of composite systems is obtained in terms of the @xmath3-partite non - separability , i.e. the minimum number @xmath3 of entangled components necessary to describe an @xmath8-partite system . * definition : `` k - separability '' * a pure @xmath8-partite quantum state @xmath9 is said to be @xmath3-separable if and only if it can be written as a product of @xmath3 states @xmath10 ( @xmath11 ) each one living on a different and non - overlapped hilbert subspace @xmath12 a mixed @xmath8-partite quantum state is said to be @xmath3-separable if and only if it has a decomposition into @xmath3-separable pure states : @xmath13 from the above definition it immediately follows that if a state is @xmath3-separable , it is automatically also @xmath14-separable for all @xmath15 . in particular , an @xmath8-partite state is called fully separable , if it is @xmath8-separable , whereas is called genuinely @xmath8-partite entangled , if and only if it is not biseparable ( @xmath16-separable ) . in this interesting case
all subsystems `` contribute '' to the entanglement .
if neither of these is the case , the state is called partially entangled or partially separable . while the above intuitive definition has been shown to provide a proper characterization of multipartite systems , it is far from straightforward to determine whether or not a particular state is @xmath3-separable .
this is immediately understood for @xmath3-separable mixed states since they may be @xmath3-separable for various partitions .
for example , a tripartite biseparable state can be given by @xmath17 where @xmath18 describe bipartite states .
consequently , the set of biseparable states are given by the convex hull of the states that are separable for a fixed bipartition . from a physical point of view
, the generation of fully separable or biseparable states does not require the interaction of all parties .
moreover , from the quantum information theoretic and operational perspective genuine multipartite entangled states allow for new applications such as quantum secret sharing ( i.e. by distributing a secret over many different parties the genuine multipartite entanglement assures security against eavesdropping or even unfair parties @xcite ) .
the authors of ref .
@xcite introduced a criterion that has to be satisfied for every @xmath3-separable state . in the instance that it is violated , it serves as a test for multipartite entanglement .
it is obtained by defining a set of permutation operators @xmath19 .
such operators act on two copies of an @xmath8-partite state , swapping the @xmath20-th subsystems between the two copies , i.e. @xmath21 where the @xmath22 and @xmath23 indicate the subsystems of the first and second copy of the state , respectively .
it is evident that if the permuted subsystem is separable from the rest of the state , the state is invariant under such a permutation .
this observation is the key idea to constructing very general separability criteria that are based on the following theorem proven in ref .
@xcite:``_every @xmath3-separable state @xmath24 , regardless its purity , satisfies the following inequality @xmath25 for all fully separable states @xmath26 , where @xmath27 , the sum runs over all possible partitions @xmath28 of the considered system into @xmath3 subsystems , the permutation operators @xmath29 are the operators permuting the two copies of all subsystems contained in the @xmath20-th subset of the partition @xmath28 and is the total permutation operator , permuting the two copies .
_ '' the aim of this paper is to consider this criterion @xmath30 for @xmath4 applied to the ground states of the @xmath0 model in order to prove the existence of genuine multipartite entanglement ( we will always consider the optimized value unitary operators . ] , i.e. optimized over all fully separable states @xmath26 ) .
importantly , this undertaking allows us to prove the existence if a set of genuine multipartite entangled states that do not posses the symmetries given by the vanishing commutators , @xmath31=0 $ ] . obviously
, non - violation of the criterion @xmath32 does not imply that the state of interest is not genuine multipartite entangled since the criterion @xmath32 is only necessary but not sufficient for biseparability @xcite .
let us remark that the obtained optimized value of @xmath32 are also lower bounds on multipartite measures as shown in several refs .
in particular , these multipartite measures are based on the well - known concurrence @xcite applied to @xmath33 subspaces . in the following
we show that for most cases of the @xmath0 model the criterion @xmath34 is helpful and for the few cases where the criterion fails we discuss different strategies .
having defined our machinery for the analysis of the multipartite entanglement we will proceed to introduce the well known xy model .
we consider a one dimensional chain of spin-@xmath6 particles , each localized at side @xmath20 and interacting with its nearest neighbor via an interaction in @xmath35- and @xmath36-direction .
additionally , an external magnetic field in @xmath37-direction can be activated .
we will pay particular attention to the class of translationally invariant states with periodic boundary condition .
this one - dimensional ferromagnetic @xmath0 spin model can be described by the following hamiltonian @xmath38 where @xmath39 @xmath40 stands for the spin-@xmath41 pauli operator on site @xmath20 , @xmath2 is the strength of the external transverse field , and @xmath1 is the anisotropy of the nearest neighbor interaction in @xmath35- and @xmath36-direction , taking values in the interval @xmath42 $ ] .
the extremes correspond , respectively , to energy - gapless isotropic @xmath0 model ( @xmath43 ) and the ising model ( @xmath44 ) .
note that the entanglement properties do not change when passing from the ferromagnetic to the anti - ferromagnetic @xmath0 model ( @xmath45 and @xmath46 ) since only local unitary operators are needed to pass between these two cases .
such models governed by the above defined hamiltonian play a relevant role in the field of quantum statistical mechanics since , firstly , it is one of the few models that can be solved analytically , and , secondly , it is a very good approximation to real condensed matter systems ( e.g. , for neutral atoms loaded on an optical lattice @xcite ) , and , thirdly , it simulates a quantum circuit @xcite .
the analytic solution is obtained by applying the jordan - wigner transformations onto the spin operators obtaining pseudo - fermions @xcite . after these transformations the system can be easily diagonalized . to obtain solutions for various finite sized systems and any value of the external magnetic field @xmath2
we generalize the solutions obtained by lieb _
et al . _
@xcite , whereas for the solutions in the thermodynamic limit one can consult refs .
@xcite .
the model given by the hamiltonian ( [ hamiltonian1 ] ) shows two important physical properties : 1 .
regardless the value of anisotropy @xmath1 and the size of the system , when the external magnetic field @xmath2 becomes equal to @xmath47 the system admits , among others , two fully factorized ground - states , i.e. global states that are products of single - site states @xcite ( in case of @xmath48 into one single state ) .
this case is consequently named `` * * factorization point * * '' .
2 . in the thermodynamic limit for any value of the anisotropy @xmath1 at the critical value of the magnetic field @xmath49 , denoted in the following as the `` * * critical * * '' point
, the system undergoes to a _
quantum phase transition _ : the ordered phase ( @xmath50 ) possesses different properties in dependence of the anisotropy @xmath1 . for @xmath51 the system is characterized by a twofold degenerated ground - state space associated to a gapped energy spectrum ; whereas for @xmath43 the ground state space is not degenerated and the energy spectrum is gapless . however , regardless the value of the anisotropy @xmath1 , in the paramagnetic phase @xmath52 the system is also non - degenerated but the energy spectrum shows a gap between the ground state and the first excited state .
the presence of a degeneracy in the ground - state - space forces us to make a choice among all possible ground - states of the system . to be consistent between regions with and without degeneracy
, we will always take into account that the ground - state preserves all the symmetries of the hamiltonian .
moreover , we focus our interest on the presence of genuine multipartite entanglement between three adjacent spins , i.e. position @xmath53 and @xmath20 and @xmath54 . for such quantities , all the information needed is contained in the reduced density matrix @xmath55 obtained from the ground - state wave function ( after all the spins except those at positions @xmath53 and @xmath20 and @xmath54 have been traced out ) . owning to the invariance of @xmath55 under spatial translation ,
the resulting matrix does not depend on the particular choice of @xmath20 .
we now present genuine multipartite entanglement features originating from the criterion @xmath32 for three adjacent spins of the ground state , i.e. detecting the inseparability of this state into any two parts .
we begin by considering the thermodynamic limit and follow this by an examination of the finite sized systems .
then we discuss scaling properties at the critical point @xmath56 and at the factorization point @xmath57 and last but not least we focus on the case of vanishing magnetic field @xmath58 . for the reduced density matrix @xmath55 as function of the external field @xmath2 for different values of @xmath59 in the thermodynamic limit @xmath60 .
orange solid line @xmath43 ; magenta dotted line @xmath61 ; green dashed line @xmath62 ; red dot - dashed line @xmath63 ; blue dot - dot - dashed line @xmath64 and black dot - dot - dot - dashed line @xmath44 ( ising model ) . in the paramagnetic phase
@xmath65 the criterion @xmath32 detects genuine multipartite entanglement for all values of @xmath1 , except for the isotropic case @xmath43 , which is known to be fully separable ; whereas in the ferromagnetic phase @xmath66 only in the ising case genuine multipartite entanglement is detected for all values of the magnetic field . ] in fig
. [ bisepthermolimit ] the behavior of the criterion @xmath32 applied to @xmath55 in dependence of @xmath1 and @xmath2 is summarized for the thermodynamic limit case ( @xmath67 ) .
for all values of @xmath59 we observe that all the curves follows a very similar behavior : at high values of the external magnetic field @xmath2 the criterion @xmath32 detects genuine multipartite entanglement , i.e. @xmath34 . further lowering the external magnetic field
@xmath2 the amount of the violation of the inequality increases until it reaches a maximum close to the quantum critical point
. then it starts to decrease ( showing a diverge in the derivative exactly at @xmath68 ) until it vanishes at the factorization points @xmath57 .
lowering further the strength of the magnetic field @xmath69 the amount of the violation increases again , reaching a second local maximum before dropping down and even becoming negative before the external field vanishes .
for @xmath43 the behavior is very different reflecting the fact that the factorization field coincides with the critical case as well as the fact that for values greater than the critical field @xmath70 the ground state of the system is always the ferromagnetic state @xmath71 , i.e. a fully separable state .
it is natural to compare the behavior of @xmath32 for the three adjacent spins for the parity preserving ground state of the @xmath0 model in the thermodynamic limit with the behavior of the concurrence @xmath72
@xcite for the two adjacent spins , i.e. the reduced density matrix obtained tracing out the @xmath73-th spin of @xmath55 . the result is summarized in fig .
[ concthermolimit ] . in fig .
[ concnextthermolimit ] we have summarized the result of the concurrence computed for the next nearest spins , i.e. tracing over the @xmath20-th spin . for different value of @xmath59 in the thermodynamic limit @xmath60 : orange solid line @xmath43 ; magenta dotted line @xmath61 ; green dashed line @xmath62 ; red dot - dashed line @xmath63 ; blue dot - dot - dashed line @xmath64 and black dot - dot - dot - dashed line @xmath44 ( ising model ) . ] for different value of @xmath59 in the thermodynamic limit @xmath60 : orange solid line @xmath43 ; magenta dotted line @xmath61 ; green dashed line @xmath62 ; red dot - dashed line @xmath63 ; blue dot - dot - dashed line @xmath64 and black dot - dot - dot - dashed line @xmath44 ( ising model ) . ] comparing the behavior of the quantity @xmath32 ( fig .
[ bisepthermolimit ] ) with the one of the concurrence @xmath72 ( fig .
[ concthermolimit ] ) we observe a similar behavior above the critical point @xmath74
. however , below the critical point the nearest neighbor entanglement increases and becomes maximal at @xmath58 .
different to the behavior of the nearest neighbor concurrence the next - nearest neighbor concurrence becomes zero for small values of @xmath2 ( see fig .
[ concnextthermolimit ] ) . in this respect
it is also interesting to investigate the question : what is the maximum bipartite entanglement in a translational invariant spin chain ?
for finitely correlated qubit - chains the answer was found @xcite , i.e. the concurrence obtains the maximal value of @xmath75 .
we see that in our case ( fig .
[ concthermolimit ] ) that we are below this value . in the sec .
[ vanishing ] we investigate the case close to @xmath58 further .
let us now compare the behavior of the the inseparability in the thermodynamic limit with finite size systems .
[ bisepfinitesize ] one observes three differences : 1 .
the first concerns the behavior of @xmath32 around the quantum critical point @xmath74
. the function @xmath32 does not show ( as expected ) the divergence of the derivative of the external magnetic field @xmath2 .
we discuss the finite size scaling in more details in sec .
[ critical ] .
2 . the second concerns the factorization point @xmath57 , here @xmath32 does not vanishes .
that means that the genuine multipartite entanglement decreases exponentially with the number of interacting spins .
this fact , at first sight , is very surprising because one is always driven to associate factorization with the vanishing of any type of entanglement related quantity .
however , as we discuss in more details in sec .
[ factorizing ] , the factorization point is always , i.e. regardless to the length of the chains , characterized by a two fold degenerate ground space @xcite .
the last difference is represented by a sort of oscillations in the region with @xmath69 that are associated to repetitive level crossing between even and odd states @xcite .
such oscillations are also reflected in the behavior of bipartite entanglement @xcite . for the reduced density matrix obtained tracing out the degree of freedom of each spins in the system except three adjacent spins as function of the external field @xmath2 for @xmath76 for different sizes @xmath77 of the chains ; black dashed line : thermodynamic limit ; red solid line : @xmath78 .
] in the thermodynamic limit at the critical point the derivative @xmath79 diverges . in the case of finite numbers of spins @xmath77 the derivative is presented in fig .
[ derivbisepfinitesize ] .
depending on @xmath2 in the case of @xmath44 ( ising model ) for different length ( n ) : @xmath80 orange dot - dot - dashed line ; @xmath81 green dot - dashed line ; @xmath82 brown dot - dot - dot - dashed line ; @xmath83 blue dotted line ; @xmath84 red dashed line ; @xmath85 black solid line . ]
we observe by increasing the length of the chain and , hence , the system approaches the thermodynamic limit , the derivative shows an increasingly pronounced maximum which becomes even more closer to the critical value @xmath68 .
such qualitative descriptions can be quantified as visualized in fig .
[ analysisofmax ] : in the upper panel we show how the maximum of the derivative of @xmath32 increase with increasing length @xmath77 , whereas in the lower panel we summarize how the external field for which the maximum of @xmath86 is reached
, @xmath87 , converges to the critical point @xmath68 for increasing @xmath77 .
): dependence of the maximum value of @xmath86 on the length of the chain @xmath77 .
lower panel ( @xmath44 ) : dependence of @xmath87 , i.e. the value of the external field in which the maximum of @xmath86 is reached , on the size @xmath77.,title="fig : " ] ): dependence of the maximum value of @xmath86 on the length of the chain @xmath77 .
lower panel ( @xmath44 ) : dependence of @xmath87 , i.e. the value of the external field in which the maximum of @xmath86 is reached , on the size @xmath77.,title="fig : " ] to be more precise the convergence of @xmath87 to @xmath88 is given by @xmath89 , whereas its maximum value diverges logarithmically with increasing system size @xmath77 as : @xmath90 this equation quantifies non - local correlations which are only due to genuine multipartite entanglement in the critical region . depending on the external magnetic field @xmath2 as a function of @xmath91 in the thermodynamic limit . ] in the thermodynamic limit the derivative @xmath92 diverges approaching the value @xmath93 ( as shown in fig .
[ derivativethermo ] ) : @xmath94 in the case of @xmath44 .
the red circles correspond to @xmath78 spins , the blue squares to @xmath84 spins and the green diamonds to @xmath95 spins . ] according to the scaling ansatz @xcite in the case of logarithmic singularities , the ratio between the two prefactors of the logarithm in eq .
( [ finitesizescaling1 ] ) and eq .
( [ finitesizescaling2 ] ) is the exponent @xmath96 that governs the divergence of the correlation length @xmath97 .
consequently , arising from eq .
( [ finitesizescaling1 ] ) and eq .
( [ finitesizescaling2 ] ) , it follows that @xmath98 ( this result is consistent with the solution for the ising model @xcite ) .
this is in agreement with the universality principle , a cornerstone of the theory of critical phenomena , i.e. the critical properties depend only on the size of the system and the broken symmetry in the ordered phase . by proper scaling @xcite and taking into account the distance of the maximum from the critical point
, it is possible to make all the data for different @xmath77 collapse onto a single curve ( see fig .
[ testfinitesize ] ) .
this figure contains the data for chain sizes ranging from @xmath78 up to @xmath95 . as function of the external field @xmath2 for reduced density matrix obtained tracing out all the spins except @xmath99
( black dashed line ) , and @xmath100 ( red solid line ) for @xmath101 ( ising model ) in the thermodynamic limit @xmath60 . ]
one of the more surprising facts discovered by fazio et al .
@xcite regarding the entanglement between pairs of spins is that the range of entanglement remains finite , regardless of however close to the quantum critical point the range of the standard correlations diverges . in fig .
[ biseparability_as_func_of_r_gamma_1 ] we plot the behavior of our genuine multipartite criterion for the ising model for nonadjacent spins , @xmath99 and @xmath100 , respectively .
our criterion does not detect any genuine multipartite entanglement .
even if this does not imply the absence of genuine multipartite entanglement at the critical point except in the case of three adjacent spins , it strongly suggests that the range of entanglement is finite regardless the fact that it is bipartite or not . as function of the external field @xmath2 for reduced density matrix obtained tracing out all the spins except three adjacent or four adjacent spins for different value of @xmath1 in the thermodynamic limit @xmath60 .
red dot - dashed line corresponds to the case of three adjacent spins and @xmath44 , whereas the red solid line stands for the case of four adjacent spins with the same @xmath1 .
the black dotted line stands for the case of three adjacent spins and @xmath62 , whereas the black dashed line corresponds to the case of four adjacent spins with the same @xmath1 . ]
this is in contrast to the case where we compared genuine multipartite entanglement of three and four adjacent spins ( fig .
[ bi_vs_triseparability_gamma_1 ] ) , where we observe the same scaling behavior and a different behavior at @xmath2 close to zero .
let us remark here that it is not known how the detection ability of the criteria scales with the number of particles and with dimension . thus our results demonstrate that all the key ingredients of the finite size scaling are present in the multipartite quantity .
our analysis shows that in the region of the quantum phase transition the bipartite and the multipartite entanglement of three adjacent spins obey the same scaling behavior . in summary ,
the simple quantity @xmath32 turns out to be useful to analyze phase transitions and , as such , offers a promising framework for fully characterizing phase transitions of quantum complex multi - body systems in terms of genuine multipartite entanglement . as we have noticed in sec .
[ model ] , regardless the length of the chain , at the factorization point @xmath102 the system shows a degeneracy of the ground state @xcite that is associated with the existence of two factorized ground state ( for @xmath103 in a single state ) , of the form @xcite @xmath104 where @xmath105 are fixed by the hamiltonian parameters @xmath106 however , for @xmath59 such states are not eigenstates with a fixed parity and hence are not the states on which we are focusing our analysis , i.e. do not allow a coherent analyses . in such cases
, the two eigenstates with fixed parity respect to @xmath107 can be constructed from the two factorized states , that represent a non orthogonal basis for the ground sub - space @xmath108 where @xmath109 ( @xmath110 ) is the normalization coefficient that ensures the normalization of the even ( odd ) ground state . as the two factorized ground states @xmath111 are , in general , not orthogonal to each other ,
the required normalization coefficients are non - trivial and can be deduced as @xmath112 the two factorized ground state are fully separable and hence , for them , any quantity related to the entanglement must vanishes . however , due to the non - orthogonality of the even and odd ground states the resulting state can be entangled . in fig .
[ bisepatfactorization ] we show the detailed behavior of @xmath32 and the nearest neighbor concurrence @xmath113 at factorization point as function of @xmath77 and different @xmath1 s . , and the behavior of the nearest neighbor concurrence @xmath113 at the factorization point as function of the size of the chain @xmath77 for different values of @xmath1 : @xmath61 ( blue ) solid line @xmath113 and dot - dashed line @xmath32 ; @xmath76 ( red ) dashed line @xmath113 and dot - dot - dashed line @xmath32 ; @xmath64 ( black ) dotted line @xmath113 and dot - dot - dot - dashed line @xmath32 . ]
we observe that increasing the size of the chain both entanglement quantities decrease with an exponential law proportional to @xmath114 , where @xmath115 is a function of @xmath1 that is plotted in fig .
[ argofexp ] ( that depends on the inverse of the energy gap between the ground state and the first excited states @xcite ) . as function of @xmath1 . ]
increasing @xmath1 the function @xmath116 increases resulting in a faster and faster exponential decay .
for the ising model @xmath44 it diverges signaling that , in such a case , both @xmath32 and the concurrence @xmath113 of the even and odd ground states vanishes for any @xmath77 .
in particular , the ground state is exactly the ghz state , hence , tracing out any part of the system will reveal the disappearance of all entanglement . for the opposite limit , i.e. when @xmath43 for any @xmath117 and any @xmath77 , the system admits as unique ground state @xmath71 and hence all entanglement vanishes . for all other cases ,
i.e. for @xmath118 , we find that due to the non - orthogonality of @xmath119 and @xmath120 multipartite entanglement disappears only in the thermodynamic limit . in this case
@xmath55 can be regarded as an equally weighted mixture of the two fully separable states , @xmath119 and @xmath120 . as evident from fig .
[ bisepthermolimit ] and fig .
[ bisepfinitesize ] and regardless the size of the system , the criterion @xmath32 becomes negative at sufficiently small value of the external field @xmath2 for @xmath118 . on the other hand ,
for @xmath121 it vanishes exactly only at @xmath58 .
since the positivity of @xmath32 is only a necessary but not sufficient criterion based on permutation of all possible bipartitions of the two - copy state
we can not conclude that the genuine multipartite entanglement vanishes for small @xmath2 .
however , by approaching the problem numerically via monte carlo techniques , we may be able to exploit various known symmetries . taking a monte carlo approach and defining @xmath122
as a generic three spin state obtained as a mixture of fully separable and biseparable state we sought to minimize the hilbert - schmidt distance @xcite defined by @xmath123\;. \nonumber\end{aligned}\ ] ] in all our simulations we choose as starting point a mixture composed of two pure fully separable state with the same weight : and where @xmath124 ( @xmath125 ) is the eigenvalues of @xmath126 with eigenvalues @xmath127 ( @xmath128 ) .
we based this choice on the fact that such states are equal to @xmath55 for @xmath44 ( quantum ising model ) , regardless of the chain length .
therefore @xmath55 for the quantum ising model is always , i.e. for any @xmath77 , fully separable .
our algorithm starts by defining a mixture of @xmath129 fully separable states and @xmath130 biseparable states ( @xmath131 states with entanglement between spin @xmath53 and spin @xmath20 , @xmath131 with entanglement between spin @xmath53 and @xmath54 , and @xmath131 with entanglement between @xmath20 and @xmath54 ) . then the distance is computed and compared to the distance for variations of the initial state .
if the distance is smaller , the varied state becomes the new initial one , otherwise the initial state is kept , however , only with probability equal to the negative exponent of the difference of the two distances .
the results of our montecarlo numerical simulation are summarized in fig .
[ montecarlo ] where the upper ( lower ) panel shows the behavior of the distance @xmath132 as a function of the step of the montecarlo algorithm for the system in the thermodynamic limit ( of @xmath80 spins ) for different values of @xmath133 . as a function of the steps @xmath134 in the montecarlo simulations for different values of @xmath1 at @xmath58 for a chain of infinite length ( upper panel ) and for a chain of @xmath80 spin ( lower panel ) . from the bottom to the top : black lines @xmath135 ; red lines @xmath76 ; blue lines @xmath136 ; green lines @xmath43.,title="fig : " ] as a function of the steps @xmath134 in the montecarlo simulations for different values of @xmath1 at @xmath58 for a chain of infinite length ( upper panel ) and for a chain of @xmath80 spin ( lower panel ) . from the bottom to the top : black lines @xmath135 ; red lines @xmath76 ; blue lines @xmath136 ; green lines @xmath43.,title="fig : " ] all samples follow the same behavior : after a very fast decrease the algorithm reaches a threshold ( that depends strongly on @xmath1 and seems also to depend only weakly on @xmath77 ) that seems impossible to be overcome . even if these numerical simulations do not represent a proof of the presence of the multipartite entanglement in the @xmath0 model , the lack of convergence in our algorithm is a strong suggestion
; in agreement with the results presented in ref .
we also wish to draw the readers attention to fig .
[ bi_vs_triseparability_gamma_1 ] where we present the criterion @xmath32 applied to four adjacent spins and observe genuine multipartite entanglement for vanishing magnetic fields .
in conclusion , in this paper we have analyzed the presence of genuine multipartite entanglement in the ground - state of the @xmath0 model . despite its simplicity
the @xmath0 model being characterized by the anisotropy parameter @xmath1 and an external magnetic field @xmath2 possesses genuine multipartite entanglement for almost all values of @xmath1 and @xmath2 . from a physical point of view
this means that the entanglement present in a subset of a spin chain does require interaction of all particles .
moreover , by analyzing in detail the finite size case and the thermodynamic case we proved that the considered criterion detecting genuine multipartite entanglement being also a good bound on multipartite measures is a proper quantity to explore all the features of phase transitions and scaling properties . consequently , further analyses of distinct types of genuine multipartite entanglement or different multipartite criteria open up a promising toolbox to characterize properties of condensed matter systems . in particular
, our approach may turn out to be extremely useful for systems that show topological order ; in this case the bipartite entanglement is not able to provide a complete characterization of the ground state properties @xcite .
other systems where our approach applies are frustrated quantum spin models @xcite ; here some preliminary results suggest that the transition between a saturated ground - state and a non - saturated ground state is due to an abrupt change of the relative weight in the multipartite entanglement properties .
we want to thank colin wilmott and the referees for many useful suggestions to improve the paper .
smg and bch acknowledge gratefully the austrian science fund ( fwf - p23627-n16 ) . | we analyze the @xmath0 model characterized by an anisotropy @xmath1 in an external magnetic field @xmath2 with respect to its genuine multipartite entanglement content ( in the thermodynamic and finite size case ) . despite its simplicity
we show that the quantity detecting genuine multipartite entanglement through permutation operators and being a lower bound on measures witnesses the presence of genuine multipartite entanglement for nearly all values of @xmath1 and @xmath2 .
we further show that the phase transition and scaling properties are fully characterized by this multipartite quantity .
consequently , we provide a useful toolbox for other condensed matter systems , where bipartite entanglement measures are known to fail . |
it has been widely recognized that the water environment affects the energy landscape and functionality of biomolecules in a profound way .
there is , however , another solvent - related effect that is considered less frequently : the hydrodynamic interactions ( hi ) between individual segments of a biomolecule that moves .
these interactions may affect the dynamics of conformational changes because any motion of one segment generates a local fluid flow which influences another segment .
the presence of hi is known to affects dynamic properties of soft mater .
for instance , hi modify the values of diffusion coefficients in colloidal suspensions @xcite , affect the characteristics of the coil - stretch transition in polymers @xcite , change the kinetic pathways of phase separation in binary mixtures @xcite , and alter the kinetics of macromolecule adsorption on surfaces @xcite .
much less is known about the role of hi in protein folding and unfolding processes .
dickinson @xcite and tanaka @xcite speculated that hi might affect the kinetics of protein folding , but the actual numerical assessment of the role of hi has come with the paper by baumketner and hiwatari @xcite .
they have considered coarse grained models and found that hi delay folding of a @xmath0 hairpin but do not affect folding of the @xmath1-helix . in this paper , we consider mechanical stretching of proteins and study the relevance of hi to the process .
the stretching can be accomplished in several ways and we discuss three modes : at constant speed , at constant force , and through fluid flow .
we chose ubiquitin as a model system , since there is a large body of experimental @xcite and theoretical @xcite data on its unfolding .
a coarse - grained , go - type model @xcite of a protein is used , constructed based on the knowledge of its native state .
the go models have been shown to give surprisingly good agreement with both the experimental results @xcite and all - atom molecular dynamic simulations @xcite when it comes to stretching .
we outline the model in section 2 , then introduce two different ways of tracking the evolution of the system : through langevin dynamics and brownian dynamics in sections 3 and 4 respectively . in the following sections we discuss results pertaining to the three modes of stretching and demonstrate that hi can take many roles : they inhibit unfolding by fluid flow , but make the constant force stretching faster . at constant speed , they reduce the peak force if the speed is sufficiently high . this hi - related reduction in force
may be downplayed in the all - atom simulations of titin by lu and schulten @xcite which would provide part of an explanation for the excessively large forces obtained in these studies .
in our simulations , we use the coarse - grained , go - type model of a protein . in the model , each residue is represented by a single bead centered on the position of the c@xmath2 atom .
the successive beads along the backbone are tethered by harmonic potentials with a minimum at 3.8 .
the other interactions between the residues are split into two classes : native and non - native .
this determination is made by checking for native overlaps of all atoms in aminoacids when represented by enlarged van der waals spheres as proposed in reference @xcite .
the amino acids , @xmath3 and @xmath4 , that do overlap in this sense are endowed with the effective lennard - jones potential @xmath5 $ ] .
the length parameters @xmath6 are chosen so that the potential minima correspond , pair - by - pair , to the experimentally established native distances between the respective aminoacids . in order to prevent emergence of entanglements ,
the non - native contacts are endowed with a hard core repulsion described by the @xmath7 part of the lennard - jones potential combined with a constant shift term that makes the potential vanish smoothly at 4 .
the specificity of a protein is contained in the length parameters @xmath6 .
the energy parameter , @xmath8 , is taken to be uniform .
we take @xmath9 , which correlates well with the data on titin and ubiquitin unfolding @xcite .
thus the reduced temperature , @xmath10 of 0.3 should be close to room temperature .
all of the simulations reported here were performed at this temperature . various simulation methods to study the dynamics of the system
are outlined in the following sections .
in this case , the dynamics of a protein is assumed to be governed by the langevin equation @xmath11 here , @xmath12 is the position of the @xmath3th aminoacid , @xmath13 is the net force on it due to contact potentials , @xmath14 is the friction coefficient , and @xmath15 denotes the solvent flow field . finally , @xmath16 is a white noise term with the dispersion obeying @xmath17 where * i * is the identity matrix .
the white noise term mimicks the effect of the random collisions of the aminoacids with the surrounding solvent at the same time serving as a thermostat of the system
. however , this scheme completely neglect the effects of hi which may exist in a real system , when the motion of one particle induces the flow influencing the dynamics of all the other particles . in the simulations ,
the friction coefficient @xmath14 is taken to be equal to @xmath18 where @xmath19 is the characteristic time scale of oscillations in the lennard - jones well .
the parameter @xmath20=5 used in the above definition is a characteristic value of @xmath6 in the system .
the selected value of @xmath14 corresponds to a situation in which the inertial effects are small @xcite but the damping action is not yet as strong as in water .
the equations of motion are solved by a fifth order predictor - corrector scheme .
let us note , however , that although langevin dynamics is commonly used in simulations of biological systems , the validity of this approach is not always well - established .
namely , as already noted by lorentz @xcite , the langevin equation in the above form may only be used if there is a separation of time scales between the relaxation time of the particle ( i.e. the bead ) velocity @xmath21 and the viscous relaxation time of the solvent , @xmath22 , where @xmath23 is the density of the particle , @xmath24 its radius , and @xmath25 - the density of the solvent ( see also the thorough discussion in the book by mazo @xcite . ) the ratio of those two time scales is proportional to @xmath26 . since the densities of proteins are only about 50% higher than those of the surrounding liquid @xcite , there is no separation of time scales between the relaxation of fluid variables and those of the bead and , strictly speaking , instead of eq.([lang ] ) one should use the generalized langevin equation involving a memory kernel @xmath27 where the noise is again gaussian and related to the dissipative term through the generalized fluctuation - dissipation relation @xmath28 such an approach is naturally much harder to implement ( see , however , @xcite ) and thus the ordinary langevin description as in eq.([lang ] ) is usually resorted to . in this paper , we show that in protein unfolding simulations the results of a simple langevin dynamics ( [ lang ] ) are consistent with those of brownian dynamics ( see sec .
the latter is not affected by the solvent and particle inertia effects , hence the agreement between the two methods seems to imply that non - instantaneous response of the solvent to the change of particle velocity does not play any important role in the protein unfolding processes .
if the momentum relaxation time scale ( @xmath29 ) is small in comparison to the time scales characterizing the conformational evolution of the system ( @xmath30 ) , it is appropriate to describe the dynamics in terms of equilibration of the particle configurations only .
the exact definition of @xmath30 is problem dependent : if the protein is stretched by the flow @xmath31 then @xmath32 , if a force @xmath33 acts on the molecule then @xmath34 , and if the brownian diffusion plays the central role in particle evolution then @xmath35 , ( where @xmath36 is the diffusion constant and @xmath24 the radius of the bead ) .
the algorithm for simulations of the evolution of particle positions in this time regime has been devised by ermak and mccammon @xcite .
the displacement of particle @xmath3 in time step @xmath37 ( in the absence of the flow ) obeys @xmath38 where the index @xmath39 denotes the values of respective quantities at the beginning of the time step , @xmath40 is the force exerted on particle @xmath4 by other particles , @xmath41 is the diffusion tensor and @xmath42 - a random displacement given by a gaussian distribution with an average value of zero and covariance obeying @xmath43 it is nontrivial to generalize the above expression to incorporate the effects of a general external flow field @xcite .
however , in the present case , we will be interested only in the uniform flow , in which case one gets simply @xmath44 where @xmath45 is the flow velocity .
if the diffusion tensor is nondiagonal , there exists a coupling between the force acting on the particle @xmath4 and the displacement of particle @xmath3 ( _ cf .
_ eq.[nar1 ] ) .
this coupling , mediated by the solvent , is commonly referred to as the `` hydrodynamic interactions '' .
note that without the hi , the diffusion tensor is simply @xmath46 and we recover the overdamped limit of eq.(1 ) .
the diffusion tensor @xmath41 depends in a complicated nonlinear way on the instantaneous positions of all particles in the system . for a system of spheres , exact explicit expressions for
the diffusion tensor @xmath47 exist in the form of the power series in interparticle distances , which may be incorporated into the simulation scheme @xcite . here
we adopt a pairwise , far - field approximation of @xmath41 proposed by rotne , prager @xcite and yamakawa @xcite @xmath48 and @xmath49 , & r_{ij } \geq 2a\ \\ \\
\displaystyle \frac{r_{ij}}{2a } \left [ \left ( \frac{8}{3}-\frac{3r_{ij}}{4a } \right ) { \bf i } + \frac { r_{ij}}{4a } { \bf \hat{r}}_{ij } { \bf \hat{r}}_{ij } \right ] , & r_{ij } < 2a \end{array}\right . \label{rp}\ ] ] where @xmath50 and @xmath24 represents the hydrodynamic radius of a bead .
since the above expression is exact only in the large @xmath51 limit , the radius @xmath24 should be taken to be significantly smaller than 1.9 , which is the half of the distance between the succesive beads .
on the other hand , @xmath24 can not be too small , since the space along the chain is densely filled with amino acids .
we take @xmath52 in our simulations , which seems a reasonable starting point for a qualitative assessment of hi impact on protein unfolding .
however , further studies on the impact of @xmath24 on the system dynamics are needed , in particular the hydrodynamic radius might need to vary along the chain , reflecting the different sizes of the residues . in the approximation ( [ rp ] ) , the divergence of the diffusion matrix vanishes ( @xmath53 ) , which further simplifies the numerical scheme .
however , if the full hydrodynamic interactions are included , the divergence term should be taken into account @xcite . the simulation using eq .
( [ nar1 ] ) together with ( [ rp0 ] ) and ( [ rp ] ) will be referred to as brownian dynamics with hydrodynamic interactions ( bdhi ) in contrast to a simple bd with the diagonal diffusion tensor ( [ nohi ] ) .
note that the bd describes configurational evolution of the beads on time scales in which the inertia effects of the beads and solvent molecules are negligible @xcite and , therefore , time scale separation issues discussed in section 3 are not pertinent here .
this feature favours bd as a method of choice when simulating stochastically driven motion of proteins at a coarse - grained level @xcite . in our previous studies on ubiqutin unfolding @xcite
, we have used ld . here ,
on the other hand , we incorporate hi within the bd approach .
this calls for a comparison of the three schemes ( ld , bd , and bdhi ) to distinguish the effects resulting from hi and those from the usage of distinct integration schemes .
fig . 1 presents the force - extension curves for the constant - velocity unfolding at different unfolding speeds . in the simulations ,
both termini of a protein are attached to harmonic springs with the elastic constant @xmath54=0.06 @xmath55@xmath56 .
the other end of the n - terminus spring is fixed whereas the c - terminus spring moves at a speed @xmath57 .
we consider three values of @xmath57 : 0.5 , 0.05 and @xmath58@xmath59 . in the low - speed limit , all the three data sets obtained using the ld , bd and bdhi are seen to converge to a single curve .
in contrast , at large unfolding speeds , the differences between the ld and bd are pronounced . however , strictly speaking , in this time regime bd has a limited validity , because of the lack of separation between the momentum relaxation time ( @xmath60 ) and the characteristic time of the aminoacid movement due to the stretching @xmath61 ( for the highest speed quoted above ) . in the experiments ,
the separation of time scales is huge . in water
@xmath62 g s@xmath63 which leads to @xmath64 ps ( for the typical aminoacid mass of @xmath65 g ) . on the other hand ,
the pulling speeds are of the order of 500 nm / s which gives @xmath66 ms , thus there is a five - order - of - magnitude separation in time scales .
for such a case ,
ld and bd simulations would give exactly the same result . the fact that the differences between the bd and bdhi trajectories disappear in the limit of small @xmath57 is due to the lack of impact of hi at small velocities . to conclude , in the experimentally relevant small speed limit , the effects of hi are expected to be negligible .
an inspection of fig .
1 indicates that in the case of high stretching speeds neglecting hi results in larger peak forces . in the high speed all - atom simulations of titin in water @xcite the forces of stretching
are found to be excessively large .
such all - atom molecular dynamics programs are not geared towards hydrodynamic phenomena and may incorporate hi poorly .
it is possible to consider that the excessive force could be partially due to the missing hi .
in the force - clamp afm unfolding @xcite one applies the stretching force to the protein terminus and monitors the end - to - end distance , @xmath67 .
the experimental data and the numerical simulations @xcite show that proteins unfold in a stepwise manner at a constant force .
this means that a rapid unfolding transition takes place after a certain waiting time .
the smaller the force , the longer the waiting time . in our simulations
, we apply the force to the @xmath68 terminus of the protein , whereas the n - terminus is attached to a harmonic spring of elastic constant @xmath54=0.06 @xmath55@xmath56 .
the unfolding trajectories of ubiquitin are presented in fig . 2 for two values of the force , @xmath69 and @xmath70 . the ld and bd methods essentially coincide for these relatively large forces ( small differences between the trajectories are merely stochastic in nature ) . however , the inclusion of hi changes the physics considerably the waiting times become much smaller and the duration of the unfolding transition itself decreases from @xmath71 to about @xmath72 at both values of the force .
the fact that the hi facilitate protein unfolding may be understood qualitatively when one realizes that an amino acid moving away from the bulk of a protein creates a flow which drags other residues with it ( see fig .
the differences between unfolding with and without hi are further highlighted by analysis of the so called unfolding scenarios @xcite , in which one plots an average time when a given contact is broken against the contact order , i.e. against the sequential distance , @xmath73 , between the amino acids that form a native contact . figs . 4 and 5
compare the unfolding scenarios for ld , bd and bdhi at @xmath69 and @xmath70 respectively .
remarkably , although the differences in time scales between the unfolding with and without hi are considerable , the unfolding scenarios for the smaller force are similar ( fig .
4 ) , which shows that the unfolding pathway of a protein is not affected by the hydrodynamic effects .
however , as the force is increased , both ld and bd scenarios change ( fig .
5 ) : the @xmath0 hairpin structure now unfolds at the end instead of at the beginning of the unfolding process . in contrast , in the case of bdhi , such a switch is not observed : the scenarios for larger and smaller forces look qualitatively similar .
finally , we study the influence of hi on the characteristics of the protein unfolding in a uniform flow with a speed of @xmath31 .
although the process has not yet been realized experimentally , the simulations @xcite seem to suggest that uniform flow unfolding leads to a richer set of metastable conformations than the constant force pulling .
for example , when the n terminus is anchored , ubiquitin stretches in a flow through two distinct intermediate states corresponding to a partial unzipping .
a detailed analysis of the uniform flow unfolding of ubiqutin in the absence of hi , together with the snapshots of intermediate conformations for different anchorings , can be found in @xcite . in that reference , we have mistakenly reported values of the forces in wrong units . instead of the correct unit of @xmath8/ we used @xmath74 , where @xmath75 was equal to 5 .
6 shows examples of unfolding trajectories of ubiquitin in a uniform flow for four different flow velocities , both with and without hi .
we observe that unfolding of the system with hi requires a much larger flow speed than without .
this can be understood qualitatively in terms of the so - called no - draining effect @xcite : the residues hidden inside the protein are shielded from the flow and thus only a small fraction of the residues experience the full drag force of @xmath76 ( see fig .
in contrast , when no hi are present , this drag force is applied to all residues @xcite .
notably , although the time scales and velocities involved in the protein unfolding with and without hi are completely different , the metastable states are nearly identical ( see fig .
this feature is related to the dynamic character of hi - they do not change the potential energy of the system and , therefore , do not affect its stationary properties . in principle ,
however , one could imagine a situation in which , due to the differences in dynamics imposed by hi , a system chooses alternative pathways when unfolding .
this is clearly not the case here which may be related to the _ directed _ character of the disturbance ( flow in this case or the force in force - clamp experiment ) which imposes a prefered direction of unfolding , thus greatly reducing the set of available unfolding pathways .
the similarities in unfolding pathways of ubiquitin are further confirmed by the comparison of unfolding scenarios . as an example , fig .
8 shows the unfolding scenario for the flow @xmath77 @xmath78 with hi compared to @xmath79 @xmath78 without hi ( the mean unfolding times are comparable in both cases ) .
the scenarios are very close to each other , the main difference being that the hi enhance cooperativity by breaking the contacts in a more simultaneous fashion .
in summary , hydrodynamic interactions seem to affect the time scales of unfolding by a constant force and by a fluid flow in opposite way but keep the set of the possible metastable states .
the hi may also reduce peak forces in stretching at a constant speed , although this effect weakens with the diminishing stretching speed .
fig . 1 . : : force - extension curves for the ubiquitin unfolding at a constant speed obtained using langevin dynamics ( dotted line ) and brownian dynamics with ( thick solid line ) and without ( thin solid line ) hydrodynamic interactions .
the succesive panels correspond to the pulling speeds of @xmath80 and @xmath81@xmath59 from top to bottom respectively .
2 . : : the end - to - end distance of the model ubiquitin as a function of time during unfolding at a constant force . the thick solid , thin solid and dotted lines correspond to the bdhi , bd , and ld simulations respectively .
the upper panel corresponds to the force @xmath69 , and the lower one to @xmath70 .
3 . : : the dragging effect : the moving particle creates a flow patern which affects other particles by pulling them in the direction of its motion . fig .
4 . : : unfolding scenarios of ubiquitin at constant force of @xmath82 simulated with ld ( the upper panel ) , bd ( the center panel ) and bdhi ( the lower panel ) .
open circles , triangles , squares , pentagons and solid triangles and squares correspond to contacts ( 36 - 44)(65 - 72 ) , ( 12 - 17)(23 - 34 ) , [ ( 1 - 7),(12 - 17)](65,72 ) , ( 41 - 49)(41 - 49 ) , ( 17 - 27)(51 - 59 ) , ( 1 - 7)(12 - 17 ) respectively .
the crosses denote all other contacts .
the segment ( 23 - 34 ) corresponds to a helix .
the two @xmath0-strands ( 1 - 7 ) and ( ( 12 - 17 ) form a hairpin .
the remaining @xmath0-strands are ( 17 - 27 ) , ( 41 - 49 ) , and ( 51 - 59 ) . fig .
: : the same as in fig . 3 but for @xmath70 .
: : the end - to - end distance of the model ubiquitin as a function of time during unfolding by a uniform fluid flow as illustrated by several trajectories .
the lower panel corresponds to the description with the hydrodynamic interactions and the upper without .
the succesive curves in the upper panel correspond to the flows of @xmath83 and @xmath84 bottom to top respectively . in the lower panel
the flow speeds are @xmath85 @xmath59 in all the simulations , the n - teminus of the protein is fixed .
: : the shielding effect : the particles inside a cluster experience a smaller drag force than those on the surface . fig .
: : the left and right panels show unfolding scenarios without the hydrodynamic interactions ( @xmath79 @xmath59 ) and with the hydrodynamic interactions ( @xmath77 @xmath59 ) respectively . | we incorporate hydrodynamic interactions in a structure - based model of ubiquitin and demonstrate that the hydrodynamic coupling may reduce the peak force when stretching the protein at constant speed , especially at larger speeds .
hydrodynamic interactions are also shown to facilitate unfolding at constant force and inhibit stretching by fluid flows . |
modern researchers have been fascinated by how efficiently solar photons are converted into chemical energy in photosynthesis .
initially , photons are absorbed in so - called light - harvesting complexes ( lhcs ) .
the following steps of the cascading energy transfer within lhcs are remarkably efficient : at low light intensities , 9 out of 10 absorbed photons create a charge separated state in the reaction center as the basis for further energy conversion .
@xcite a thorough understanding of this process and its technological utilization will be a major contribution to a sustainable energy concept . accordingly .
this hope has motivated numerous studies , with the aim of finding and characterizing a bio - inspired , artificial light harvesting systems .
@xcite despite the efforts , artificial systems remain still less efficient and less stable than their natural counterparts .
it is generally accpeted that a detailed picture of the involved electronic energy levels as well as the energy dissipation pathways in both artificial and natural light harvesters is crucial in the realization of an efficient and durable bio - mimetic photosystem . while there is great structural diversity and flexibility in photosynthetic lhcs ,
@xcite essentially only two molecular species serve as pigments , namely carotenoids and ( bacterio)chlorophylls , ( b)chls . carotenoids and ( b)chls are fundamentally different in structure and complementing in physiological function .
carotenoids are linear molecules with varying endgroups .
their optical properties are defined by the @xmath0-conjugated electronic states , extended along the polyene backbone as depicted in fig .
[ sec : figure absorption ] .
the earliest suggested energy flow models used a three - level system with ground state @xmath1 and excited states @xmath2 and @xmath3 @xcite .
the linear absorption spectra of carotenoids stem from transitions between the electronic ground state @xmath1 and the `` bright '' electronic excited state @xmath2 , whereas the lowest - lying excited state @xmath3 is `` dark '' , with the @xmath4 transition being one - photon forbidden .
the properties of @xmath3 are observed either by two - photon absorption from @xmath1 or by using nonlinear spectroscopy , e.g. transient absorption ( ta ) , where the esa @xmath5 is characteristically strong .
@xcite the population flow rate from @xmath2 and @xmath3 strongly depends on the carotenoid s chain length , with typical transfer times of less than 200fs for solvated carotenoids ( _ in vitro_).besides quenching of harmful long - lived triplet states in ( b)chls and structural support of lhcs , the efficient transfer of the excitation energy from carotenoid to the lower lying ( b)chl states is the main functional role of carotenoids .
the most efficient transfer route occurs from carotenoid s @xmath2 state to the energetically closest ( b)chl - state , namely @xmath6 .
transfer from @xmath3 to @xmath7 is found in several antenna complexes and occurs on a longer timescale than @xmath8.@xcite _ in vivo _
, i.e. as part of a lhc , the lifetime of @xmath2 decreases dramatically .
@xcite for example , in lhcii , the main antenna system of higher plants and algae , the lifetime of the @xmath2 state of the involved carotenoids is @xmath9120 fs _ in vitro _ and only @xmath926 fs _ in vivo _ as measured by fluorescence upconversion , which makes it one of the shortest in naturally occurring systems .
@xcite the reason for this dramatic speed - up in carotenoid lifetime is the interaction with the closely neighboring ( b)chls . chemically , the chromophore of ( b)chls is a nitrogen containing tetrapyrrole - macrocycle with one or two reduced pyrrole rings . according to the four orbital picture of porphyrins , @xcite the absorption spectrum of ( b)chls is described by at least two transitions in the so - called soret band ( @xmath10 and @xmath11 for homo - lumo+1 and homo-1 - lumo+1 , respectively ) and two @xmath12-band transitions ( homo - lumo for @xmath7 and homo-1 - lumo for @xmath6 ) . depending on molecular structure , the @xmath13-band is found in the blue to near - uv spectral region .
the @xmath12-bands stem from transitions with orthogonal transition dipole moments and are energetically degenerate for planar and symmetric porphyrins such as metal - centered phthalocyanines.@xcite ( b)chls show a reduced symmetry in comparison , causing a split of the @xmath12-band . for bchl _ a _ in organic solvents , the @xmath6-band peaks near 600 nm , while the maximum of @xmath7 is found just below 800 nm . despite their structural differences , carotenoids and ( b)chls fulfill complementary tasks in photosynthetic antenna complexes . _ in vivo _ , excitonic coupling and the polar protein environment shift the @xmath7 -band of ( b)chls to the red , thus covering the red edge of the visible solar spectrum . in this contribution
we investigate an artificial dyad that mimics _
rhodopseudomonas acidophila _
( lh2 _ rps .
_ ) , and compare it to a naturally abundant lh2 of _ marichromatium purpuratum _
( lh2 _ m. pur .
_ , formerly known as _ chromatium purpuratum _ ) .
the main difference between the artificial dyad and lh2 _ m. pur .
_ is the smaller energy difference between @xmath2 and @xmath6 in the latter .
based on this comparison , along with results reported in the literature for lh2 _ rps .
_ , we develop a novel energy transfer model for @xmath8 energy transfer .
we show that a transfer process mediated by vibronic resonances between @xmath2 and @xmath6 presents a mechanistic alternative to conventional schemes incorporating only electronic coupling .
the dyad - sample was provided by the group around ana moore and synthesized according to previously published procedures .
@xcite the dyad and its constituents were dissolved in spectroscopic grade toluene .
cells of _ marichromatium purpuratum _
strain bn5500 ( also designated dsm1591 or 984 ) were grown anaerobically in the light , harvested by centrifugation and suspended in 20 mm trishcl ph 8.0 .
then , upon the addition of dnase and mgcl@xmath14 , the cells were broken by passage through a french press .
the photosynthetic membranes were collected from the broken - cell mixture by centrifugation and resuspended in the trishcl solution to an optical density ( od ) of 0.6 at 830 nm .
the membranes were then solubilized by the addition of 1%(v / v ) of the detergent n , n - dimethyldodecylamine - n - oxide ( ldao ) .
these solubilized membranes were subjected to sucrose density centrifugation with a step gradient ( as described previously by brotosudarmo _
et al._,@xcite ) in order to fractionate the sample into lh2 and rclh1 complexes .
the lh2 fraction was then further purified by ion - exchange chromatography using diethylaminoethyl cellulose ( de52 , whatman ) .
the bound protein was washed with 20 mm trishcl ph 9.0 containing 0.15% dodecyl maltoside ( dm ) to exchange the ldao detergent and then eluted by adding increasing concentrations of nacl also containing 0.15% dm .
additional purification was achieved by gel - filtration chromatography using a superdex s-200 column ( xk 16/100 , ge healthcare ) which had been equilibrated overnight with 20 mm trishcl ph 9.0 plus 0.15% dm .
the two spectroscopic techniques employed in this article , transient grating ( tg ) and two dimensional electronic spectroscopy ( 2d - es ) are both four wave mixing techniques . in four wave mixing , three excitation pulses interact with the sample to create a non - linear polarization of third order in the field , which acts as a source term for an emerging signal . in the experiment employed here ,
all three excitation fields emerge along individual wavevectors , resulting in a non - linear signal direction which is different from any of the incoming beams .
this allows for a virtually background free signal .
details of the experimental layout are given elsewhere .
@xcite briefly , excitation pulses tunable throughout the visible spectral range are provided by a home - built non collinear optical parametric amplifier ( nopa ) , @xcite pumped by a regenerative titanium - sapphire amplifier system ( rega 9050 , coherent inc . ) at 200 khz repetition rate .
pulse spectra were chosen to overlap with the investigated sample s absorption spectrum ( see fig.[sec : figure absorption ] ) and compressed down to a temporal fwhm of sub-11 fs in each case , determined by intensity autocorrelation .
the pulses were attenuated by a neutral density filter to yield 8.5nj per excitation pulse at the sample .
this corresponds to a fluence of less than @xmath15 photons/@xmath16 per pulse or 0.5% of excited molecules in the focal volume .
the four wave mixing experiment used for both tg and 2d - es relies on a passively phase stabilized setup with a transmission grating @xcite and has a temporal resolution of 5.3as for @xmath17 , i.e. the delay between the first two pulses and 0.67fs for @xmath18 , the delay between second and third excitation pulse .
a detailed description can be found in milota _ _
et al.__@xcite for tg measurements , @xmath17 , was kept at 0fs while @xmath18 was scanned .
the emerging signal was spectrally resolved in @xmath19 by a grating - based spectrograph and recorded with a ccd camera . at a given delay time , spectra were not recorded on a shot - to - shot basis , but averaged over approximately @xmath20 shots per spectrum .
sample handling was accomplished by a wire - guided drop jet @xcite with a flow rate of 20 ml / min and a film thickness of approx .
the fact that the jet operates without the use of cell windows allows us to interpret 2d spectra and tg - signals even during pulse overlap ( @xmath21 ) .
all measurements were performed under ambient temperatures ( 295 k ) . to facilitate the interpretation of the measured 2d - spectra
, results of model calculations were included in the analysis .
it turned out that simulations including only carotenoid allowed most features of the measured 2d spectra to be explained .
therefore the model was based on the properties of the carotenoid component , whereas transfer from carotenoid to bchl only entered in terms of a contribution to lifetime broadening . with regards to electronic transitions it was assumed that the spectral profile of the pulses used in the experiment allows for resonant transitions between @xmath22 and @xmath23 , and from @xmath24 to @xmath25 subsequent to population transfer from @xmath23 to @xmath24 . to account for vibrational effects in excitation processes involving the electronic states @xmath26 and @xmath27 ,
two modes were included as underdamped oscillators with reorganization energies @xmath28 and vibrational frequencies @xmath29 in terms of line shape functions @xcite @xmath30 @xmath31 @xmath32the reorganization energies are connected to the huang - rhys factors @xmath33 via @xmath34 .
damping of the vibrations , which is known to appear in @xmath24 , was included by calculating the line shape function of the respective underdamped oscillator via double integration of a correlation function @xcite , where the real part of the correlation function was approximated by including terms up to a selected order @xmath35 of a series characterized by the matsubara frequencies @xmath36 .
furthermore , during the evolution in electronic states @xmath26 and @xmath27 the influence of the bath was described by two brownian oscillator spectral density components with damping constant @xmath37 and reorganization energy @xmath38 . from the respective spectral density components @xcite @xmath39 line shape functions @xmath40 of the bath
can be calculated by using the general formula for an arbitrary spectral density @xmath41 , which reads @xmath42 the sum of all line shape function components was included in the response functions for the calculation of 2d - spectra .
the response functions contain the transition frequencies @xmath43 between the electronic states @xmath26 and @xmath27 and rate constants of transfer processes .
besides the rate @xmath44 for transfer from @xmath23 to @xmath24 also a transfer rate @xmath45 from @xmath23 to an electronic state @xmath46 of the chlorophyll component of the lh2 complex plays a role in lifetime broadening effects .
the rate of lifetime broadening in @xmath23 consists of the sum @xmath47 and is denoted as @xmath48 .
for lifetime broadening in @xmath24 a rate @xmath49 was introduced .
the response functions can be distinguished by their rephasing or nonrephasing properties and assigned to excitation processes of stimulated emission ( se ) , ground state bleaching ( gsb ) or excited state absorption ( esa ) type , where in the case of se and gsb only the electronic transition dipole moment @xmath50 enters , while for esa components with population transfer the transition dipole moment @xmath51 also appears .
the response functions were formulated in analogy to @xcite and @xcite and are given in the appendix [ sec : response - functions ] .
the sum of all rephasing and nonrephasing response function components yields the third - order response @xmath52 and @xmath53 , respectively , which reflect the excitation dynamics in the limit of infinitely short pulses .
the influence of the finite pulse width pulses was taken into account in analogy to ref .
@xcite , i.e. by convolution of the calculated signal with the pulses . by integration over the time intervals @xmath54 ,
@xmath55 and @xmath56 between the electronic transitions , the resulting third - order polarization is transformed into a dependence on the pulse delays @xmath57 , @xmath58 and @xmath59 .
the rephasing part of the 2d - spectrum is obtained via @xmath60 @xmath61the nonrephasing part results from @xmath62 @xmath63 to simulate the energy transfer rate from the carotenoid @xmath2 state to the @xmath6 of the bchl in lh2 and the dyad we employ a hetero - dimer model .
this approach is natural for the dyad showing now signs of aggregation between the chromophores .
for lh2 the monomeric treatment of @xmath6 and @xmath2 is also justified , given that there are no signs of excitonic splitting for these bands . moreover
the carotenoid predominantly interacts with one of its neighboring b850 bchls with the next nearest neighbor coupling showing only half the interaction strength .
@xcite to model the carotenoid molecule we use an effective two state model which involves two electronic states ( @xmath1 and @xmath2 ) and one effective vibrational mode which replaces the two fast vibrational modes with frequencies @xmath64 @xmath65 and @xmath66 @xmath65 known to be present the carotenoid and it simulates the progression of states @xmath67 , @xmath68 , @xmath69 , @xmath70 which originates from the two modes .
we can group these levels according to vibrational energies equal approximately average of two mode values to @xmath71 , @xmath72 @xmath65 , @xmath73 @xmath65 , etc .
( counted from the zero point energy ) . because the spectral lines corresponding to the transitions to the states listed above are broadened , they act effectively as one mode with a roughly average frequency .
this carotenoid mode ( referred to as primary mode further on in this paper ) is coupled to a bath of harmonic oscillators @xcite ( a secondary modes or secondary bath ) which provides damping and dephasing .
the coupling between the effective mode and the carotenoid electronic transition provides the optical dephasing leading to the broadening of the absorption spectra . in order to calculate correctly the energy transfer rate , we parametrize the fast mode and its coupling to the bath of harmonic oscillators in such a way that it reproduces absorption spectra
this effective model of the bath enables us to limit the bath description to one harmonic mode whose dynamics is treated explicitly by a master equation according to @xcite . in this way the mode effectively provides a correct bath spectral density for the energy transfer between the carotenoid and bchl while allowing for a non - perturbative treatment .
fit of the absorption spectra of the carotenoid in the dyad yields the carotenoid effective vibrational frequency @xmath74 @xmath65 , hr factor @xmath75 , the reorganization energy of the secondary bath @xmath76 @xmath65 and the bath correlation time @xmath77 fs .
for the lh2 _ m. pur .
_ we obtained @xmath78 @xmath65 , @xmath79 , @xmath80 @xmath65 and @xmath81 fs .
the @xmath6 states of the bchl and purpurin are treated by the same model .
the representative mode on the bchl and purpurin molecules are chosen so that they fit absorption spectra including the significant vibrational side bands of the molecules .
the side band for the purpurin can be easily located in the monomeric absorption spectrum given in fig .
[ sec : figure absorption ] . by fitting the purpurin monomeric absorption spectrum we get the effective mode frequency @xmath82 @xmath65 , hr factor @xmath83 , reorganization energy @xmath84 @xmath65 and bath correlation time of @xmath85 fs .
bchl absorption spectrum exhibits a much weaker vibrational side band and leads to a smaller hr factor of @xmath86 .
other parameters are as follows : @xmath87 , @xmath88 @xmath65 and @xmath89 fs .
the effective vibrational modes represent the overdamped part of the spectral density found in bchls ( see e.g. ref .
@xcite ) by the term discussed in section [ sub : vibronic - coupling ] and the side band by its first replica .
the strength of this replica is given by the hr factor .
the parameters obtained by the fit of the monomers are slightly readjusted in a fit of the aggregate spectra , either the purpurine - carotenoid dyad or the lh2 complex .
energy transfer rates are then estimated with no further adjustment of the parameters . to estimate the energy transfer rate , the population dynamics of the system
is calculated by the master equation from ref .
@xcite with the initial population corresponding to the laser pulse excitation of the carotenoid .
the total time dependent population of the @xmath6 state which is initially zero is calculated and the population transfer time is defined as the time @xmath90 when the population of @xmath6 state is equal to @xmath91 .
[ sec : figure absorption](a ) shows the absorption spectrum of the dyad ( black ) and its constituents , i.e. the carotenoid donor ( green ) , and the purpurin acceptor ( red ) . ]
details on the dyads structure and dynamics are given in detail elsewhere .
@xcite it is a donor - acceptor system , consisting of a @xmath92-carotene derivative ( donor , green in the molecular structure in fig.[sec : figure absorption](a ) ) , and a tetrapyrrole - macrocycle as an acceptor , referred to as purpurin ( red ) .
the dyad - carotenoid and purpurin are linked by an amide bond , providing structural rigidity but only partial conjugation between donor and acceptor .
the dyad was synthesized to mimic lh2 _ rps .
_ , whose structure is depicted in fig.[sec : figure absorption](b ) , as taken from the papiz _ _ et al.__@xcite lh2 _ rps .
_ consists of nine pairs of pigment - protein subunits ( @xmath93-subunits ) , forming two concentric cylinders , to which carotenoids and bchl _ a _ molecules are bound non - covalently . as can be seen in fig .
[ sec : figure absorption](b ) , the bchls are arranged in two ring - like structures ( red and yellow ) with different numbers of bchls ( 18 for red , 9 for yellow ) and ring diameters , leading to a weakly and strongly interacting set of bchls ( yellow and red , respectively ) . as a result , the excitonic part of the absorption spectrum of lh2 _ rps .
_ , shown in fig.[sec : figure absorption](c ) , exhibits two peaks , one near the monomeric @xmath7-transition at 800 nm and a red - shifted peak near 850 nm .
the red and yellow sets of bchls in fig .
[ sec : figure absorption](b ) are therefore referred to as b850 and b800 , respectively .
the carotenoid is in van der waals contact ( < 3.5 ) with both the b800 and the b850 ring.@xcite the dyad lacks such excitonic bands , as its structure is monomeric and there are no interacting bchl - moieties .
an interesting difference between the absorption spectrum of the dyad and lh2 _ rps .
_ arises in the carotenoid region of the spectrum .
the vibronic structure of the monomeric carotenoids absorption spectrum , shown in green in fig.[sec : figure absorption](a ) , is more pronounced than in the dyad , shown in black .
this can be understood by invoking vibronic coupling between donor and acceptor , as will be addressed in detail in section [ sec : discussion ] .
the absorption spectrum of lh2 _ m. pur . _
( fig.[sec : figure absorption](d ) ) exhibits substantially less vibronic modulation in carotenoid region of the spectrum .
this is attributed to the polar endgroup of the bound carotenoid ( okenone ) , interacting strongly with the polar _ in vivo _ protein environment .
as the bright @xmath2-state carries ionic character,@xcite its energy relative to @xmath1 is strongly affected by the interaction with the protein and shifts to the red in comparison the carotenoid ( rhodopin glucoside ) of lh2 _ rps .
ac . _ the bchl _ a _ molecules in lh2 _ m. pur .
_ do nt show similar bathochromic shifts , which leads to a greatly enhanced overlap of the carotenoid- and @xmath6-band in this system .
the excitonic region of the spectrum of lh2 _ m. pur . _ exhibits a less pronounced red shift with respect to the monomeric transition as compared to lh2 _ rps .
_ recent preliminary x - ray structure analysis@xcite explains this by a reduced number of @xmath93-subunits ( eight for lh2 _ m. pur .
_ and nine in the case of lh2 _ rps .
ac . _ ) fig .
[ sec : figure tg](a ) shows the emission frequency ( @xmath94 ) resolved transient grating ( tg ) signal of lh2 _
m. pur_. the signal shows a broad frequency range , spanning the entire spectrum of the excitation pulse , shown as an orange line in fig .
[ sec : figure tg](b ) , in comparison to the absorption spectrum of lh2 _
m. pur_. depicted as a grey area .
the double - peaked structure of the excitation spectrum explains the overall shape of the tg - signal .
the dashed lines mark frequencies characteristic for @xmath2 ( dashed green line ) and @xmath6 ( dashed dark red line ) .
figures [ sec : figure tg](c ) and ( d ) show the respective transients . for both detection frequencies , we observe a fast decaying peak around @xmath18=0fs , followed by a slower mono - exponential decay with a decay constant of 33050fs according to least square fit analysis
. the signal does not decay to zero within the 1000fs time window of the experiment .
the observed decay behavior is , within error margins , identical for both detection frequencies .
tg probes the sum of the absorptive and dispersive part of the induced third order signal , while pump - probe selects only the absorptive part , meaning that the retrieved time constants can not be directly compared to pump - probe measurements .
@xcite vibrational dynamics however will yield similar frequencies between the two techniques ( see .
i.e. vibrational dynamics in @xmath92-carotene probed by pump - probe @xcite and by tg @xcite ) .
the vibrational response manifests as an oscillatory signal , superimposed on a slowly varying background .
after subtraction of the latter and fourier transformation of the remaining signal along @xmath18 for every detection frequency , the ( @xmath95 , @xmath19 ) frequency map in fig.[sec : figure tg](e ) is retrieved .
a cut along a @xmath2-specific detection frequency ( green dashed line ) yields the well - known carotenoid frequencies as depicted in fig.[sec : figure tg](f ) near 1000@xmath65 ( methyl - rocking motion ) , 1160@xmath65 ( carbon single bond stretching ) and 1530@xmath65 ( carbon double bond stretching ) , and are in agreement with resonance raman spectra measured for okenone .
@xcite the same set of frequencies is observed when detecting at the @xmath6-band , as shown in fig.[sec : figure tg](g ) .
this is interesting because bchl - specific modes near 1600@xmath65 or 700@xmath65 as know from resonance raman measurements @xcite are not found .
the absence of bchl vibrational signatures is readily explained by vibrational - electronic coupling in bchl , which is small in comparison to carotenoids.@xcite ) resolved transient grating signal of lh2 _ m. pur . _
( b ) absorption spectrum in gray in comparison to the spectrum of the excitation pulse in orange .
temporal cuts at @xmath2-specific detection frequencies specific for @xmath2 and @xmath6 in ( c ) and ( d ) , respectively .
( e ) fourier - transform spectra along @xmath18 , dispersed along @xmath94 .
( f ) and ( g ) show cuts through ( e ) for @xmath2- and @xmath6-specific detection frequencies , respectively . [
sec : figure tg ] ] as mentioned in section [ sec : methods ] , 2d - es and tg are related in the sense that they are both four wave mixing techniques .
the difference lies in the treatment of the first inter - pulse delay @xmath17 , which is kept at zero in tg , but scanned and fourier - transformed ( @xmath96 ) in 2d - es .
the emerging ( @xmath97 , @xmath19)-2d plots , recorded at a fixed second inter - pulse delay @xmath18 , allow for a correlation of excitation ( @xmath97 ) and emission frequencies ( @xmath19 ) .
fig.[sec:2d - dyad_vs_lh2 ] compares electronic 2d spectra at @xmath18=0fs for the dyad and lh2 _ m. pur_. = 0fs .
positive ( negative ) signals are drawn a red ( blue ) full lines at 5% steps .
each figure is normalized to its respective maximum .
the dashed grey lines mark the nodal line at zero intensity .
the lower panels in each graph show the overlap between absorption ( dark red line ) and excitation spectrum ( green line).[sec:2d - dyad_vs_lh2 ] ] at @xmath18=0fs , electronic population is given no time to relax to lower lying states , within experimental time resolution .
both spectra in fig.[sec:2d - dyad_vs_lh2 ] show strong positive ( red ) peaks along the diagonal ( @xmath97=@xmath19 ) .
comparison to the absorption spectra , given in the lower panels in fig.[sec:2d - dyad_vs_lh2](a ) and ( b ) , show that the strong diagonal peaks stem from the carotenoid s @xmath98 transition . the dyad s spectrum in fig.[sec:2d - dyad_vs_lh2](a ) shows two well resolved peaks corresponding to the @xmath6-transition at approximately 17200@xmath65 and a carotenoid peak near 19000@xmath65 .
the off - diagonal or cross peaks ( @xmath99@xmath19 ) with negative amplitude in fig.[sec:2d - dyad_vs_lh2 ] stem from esa - pathways . for carotenoids ,
esa from @xmath2 in the visible has been observed .
@xcite the esa of the dyad s porphyrinic acceptor was described previously as broad and featureless .
while the combination of donor and acceptor esa explains the negative features in the dyad s 2d - spectrum , we note that at @xmath18=0fs , negative signals do not necessarily have to be related to esa , as they already occur in a two level system at this waiting time .
@xcite positive off - diagonal peaks can stem from vibrational energy levels,@xcite stimulated emission ( se ) or coupling between two electronic oscillators.@xcite the latter would be the most interesting scenario , as electronic coupling between donor and acceptor would suggest an excitonic energy level structure and the associated transfer mechanisms.@xcite upon visual inspection , the location of the positive cross peaks in the dyad s 2d signal suggests electronic coupling as a likely explanation .
the stronger of the two positive cross peaks , found below the diagonal ( @xmath100 ) , peaks at a value of @xmath19that coincides with the absorption frequency of the @xmath6-band .
there is also a corresponding peak above the diagonal ( @xmath101 ) , which peaks blue detuned with respect to @xmath6 and with weaker intensity than its counterpart below the diagonal .
such differences in intensity may however be attributed to the ultrafast population transfer from @xmath2 to @xmath6 , the onset of which takes place within the experimental time resolution of 11fs .
the blue detuned maximum of the upper cross peak can be attributed to finite pulse width or effects of chirp in the excitation pulses .
@xcite inspection of the 2d signal of lh2 _ m. pur _ disproves the assumption of electronic coupling .
the major difference between the dyad and lh2 _ m. pur _ is the position of the @xmath2-transition . while the dyad s carotenoid exhibits maximum absorption above 19000@xmath65 , the carotenoid in lh2 _ m. pur
_ , okenone , has a carbonyl - containing endgroup as shown in fig.[sec : figure absorption](d ) , shifting the 0 - 0 transition of okenone _ in vivo _ down to 17570@xmath65 , @xcite while the energetic position of the @xmath6-band remains unchanged between dyad and lh2 _
m. pur_. hence , the energy difference between @xmath2 and @xmath6 should be greatly reduced in lh2 _
m. pur _ , and the corresponding cross peaks should be shifted , leading to a large and fairly featureless electronic 2d - signal .
in contrast to this expectation , the lower cross peak in lh2 _
m. pur _ shows roughly the same energy difference from the diagonal of approximately 1500@xmath65 .
this strongly suggest a vibrational origin , given that the prominent c = c stretching mode of carotenoids is found at this energy .
the peak above the diagonal in lh2 _
m. pur _ is harder to explain .
the esa - signal in fig.[sec:2d - dyad_vs_lh2](b ) is much weaker than for the dyad case , which is readily explained by a different esa - transition energy from @xmath2 for okenone _ in vivo_. @xcite we can thus observe a weak diagonal peak at @xmath10216480@xmath65 , with an excitation frequency @xmath97 corresponding to the upper cross peak .
this energetic position excludes coupling to @xmath6 as a possible explanation , peaking at 17035@xmath65 in lh2 _
m. pur_. the physical origin of the upper cross peak in fig.[sec:2d - dyad_vs_lh2](b ) becomes more obvious when examining the peak shape evolution along @xmath18 , as shown in fig.[sec : figure 2d_experimental ] .
the upper diagonal peak is discernible up to 30fs , but drops to zero by 80fs .
this behavior is unexpected for an electronic coupling peak , whose amplitude should only be a function of coupling constant _ j _ ,
but not of waiting time @xmath18 .
the rapid decay of the upper cross peak rather suggests that it is an artifact , related to the shape of the excitation pulses .
the excitation spectra are clearly non - gaussian ( see lower panel of fig.[sec:2d - dyad_vs_lh2](b ) ) , making limited pulse width or chirp related effects @xcite the most likely explanation for the upper diagonal cross peak for lh2 _ m. pur_. another interesting feature is observed at ( @xmath97=-17500@xmath65 , @xmath19=16480@xmath65 ) , i.e. an energetic position indicative of @xmath8 energy transfer pathways .
the peak builds up rapidly within the first 30fs , which corresponds well with the 55fs transfer time reported for this process.@xcite after 50fs , negative esa - signals dominate the spectra .
this is explained by the intramolecular @xmath103 energy transfer , occurring at a 95fs build up rate as reported previously.@xcite we support this assignment by calculating the electronic 2d - spectra only for okenone , the carotenoid of lh2 _
m. pur_. details of the calculations can be found in section [ sec : methods]c .
spectroscopic experiments reveal the transfer from car @xmath2 and @xmath3 states to chls and bchls as fast and relatively efficient despite the rather short @xmath2 state life time of the monomeric carotenoid ( at least in comparison with chls and bchls ) . both @xmath6 and @xmath2 states
are optically allowed , and they can therefore interact by the resonance coupling mechanism .
because of the proximity of the two molecules in the lh2 structure , one can not exclude an electron exchange ( dexter ) mechanism . however , quantum chemical calculations suggest coulomb interaction to be dominant here @xcite . in this case ,
theoretical description the carotenoid - bchl interaction and modeling of the corresponding excitation energy dynamics remain within the well understood framework of excitonic description of photosynthetic antennae @xcite . in the standard framework
, we assume the electronic coupling element between the two electronic states to be independent of the vibrational dof of the molecular system .
electronic excitation is transferred between the two molecules due to an interplay of the resonance coupling and the fast fluctuations of the molecules respective energy gaps .
the energy gap fluctuations are caused by their interaction with the nuclear degrees of freedom ( dof ) which involve both the intra- as well as intermolecular vibrations .
these are considered a thermodynamic heat bath characterized by certain temperature and spectral density .
two regimes of the relative coupling strengths are usually distinguished .
the _ delocalized regime _ in which the system - bath interaction is weak in comparison with the resonance interaction between the molecules , and the _ localized regime _ in which the system - bath coupling is considered strong .
the terms localized and delocalized refer to the effective electronic eigenstates of the interacting system . for the delocalized regime
the coupling between the two molecules is strong enough to sustain correlation between the electronic excitations on different molecules so that delocalized excitons are formed . in the localized regime ( also referred to as frster regime here ) , the system - bath interactions induce energy gap fluctuations which prevent any such correlation , and the excitations appear to be localized on the molecules .
in fact , even in relatively weakly excitonically coupled systems one can often detect signs of delocalization ( see e.g. ref .
@xcite ) , because the actual apparent eigenstates are always somewhere in between the two theoretical limits . for
photosynthetic aggregates formed by bchl and chl molecules , where the protein bath can be to some extent assumed as unstructured , modern simulation methods such as the hierarchical equations of motion ( heom ) enable us to determine the excited state dynamics beyond the two limits mentioned above.@xcite however , even in the case of aggregates composed of bchl or chl molecules only , some clearly visible spectroscopic features can result from the involvement of pronounced vibrational modes .
@xcite in the discussion that follows , we want to argue that the involvement of the vibrational dof of the carotenoid molecules is crucial for the energy transfer dynamics between the @xmath2 and the @xmath6 states .
we will compare two theoretical approaches to the calculation of the energy transfer rates , both motivated by the observed features of the spectra and known spectroscopic behavior of the system .
we will argue that both support the crucial role of the fast vibrational modes , in particular the carotenoid ground state vibration levels .
the hamiltonian of the relevant part of the system ( state @xmath104 of the bchl and the @xmath2 state of the carotenoid ) can be written in the following way @xmath105 @xmath106 @xmath107 here , @xmath108 and @xmath109 are the optical electronic transition energies of the bchl @xmath6 state and the carotenoid @xmath2 state , respectively ( including the reorganization energy of the bath ) , @xmath110 and @xmath111 are the hamiltonian operators of the nuclear dof on the bchl and carotenoid , respectively .
these two sets on nuclear dof described by macroscopic sets of coordinates @xmath112 and @xmath113 , where @xmath35 is a macroscopically large number , form the bath interacting through operators @xmath114 and @xmath115 with the electronic transitions on the bchl and carotenoid , respectively .
the electronic coupling element @xmath116 describes the resonance interaction between the collective excited state @xmath117 in which the bchl is excited to the state @xmath6 and the carotenoid is in its electronic ground state and the collective excited state @xmath118 in which the bchl is in its electronic ground state and the carotenoid is excited to its excited state @xmath2 . to complete the hamiltonian
we should also to include the state @xmath119 in which both car and bchl are excited .
however it will be shown later that the features related to this state are negligible as the resonance coupling will be shown to be weak . in the experimental 2d spectra , shown in fig .
[ sec : figure 2d_experimental ] , one can easily identify the fast rising esa component , which is usually assigned to the strong absorption from the @xmath3 states of the carotenoid . contribution of this state is taken into account by including a separate partial hamiltonian and @xmath103 relaxation rate into the response functions in appendix [ sec : response - functions ] all other states with no spectroscopic contributions are taken into account in the simulations by introducing decay rates for the bright states .
for the subsequent discussion it will be useful to introduce a bare electronic ( single exciton ) hamiltonian which reads as @xmath120 @xmath121 this hamiltonian represents the basic three level ( ground state energy was chosen to be zero ) structure of the problem .
as discussed above , the interplay of energy gap fluctuations and the resonance coupling energy determines which theoretical limit applies .
the usual discussion of the relative strengths of the two interactions is made based on the relative values of the reorganization energy @xmath122 which describes the magnitude of the energy gap fluctuations and the value of the resonance coupling @xmath116 @xcite .
our system is characterized by reorganization energies @xmath123 of the bchl and @xmath124 of the car which should be compared to the resonance coupling energy @xmath116 . for the validity of the weak resonance coupling limit
it is usually required that @xmath125 . in our case , assuming @xmath126 @xmath65 ( see ref .
@xcite ) confirms the validity of this regime because the total @xmath124 is in the order of thousands of @xmath65 as it involves two fast vibrational modes with huang - rhys factors close to @xmath127 and vibrational frequencies above @xmath128 @xmath65 as shown in fig .
[ sec : figure tg ] .
the @xmath7 state of bchl in the protein environment of the lh2 complex exhibits @xmath129 @xmath65 ( see e.g. ref .
@xcite ) and we can assume a similar value @xmath130 @xmath65 as for the @xmath6 state . here , the reorganization energy of the chromophore is only comparable , not larger , than the resonance coupling element .
we note that conventionally the role of temperature is completely neglected in this discussion .
the magnitude of the energy gap correlation function , which provides a true measure of the energy gap fluctuations , depends on temperature and the dependency is linear in the high temperature limit ( in the case of brownian oscillator model , see ref .
@xcite ) . it would therefore be more suitable to compare the value of @xmath131 and @xmath132 . at room temperature ,
@xmath133 @xmath65 and @xmath134 . yet another problem arises when considering the role of the relative energy gap @xmath135 of the @xmath6 and @xmath2 states . in an ideal dimer
, the delocalization is determined by the so - called mixing angle @xmath136@xcite the delocalization decreases with the increasing @xmath137 .
the interplay of the relative energy gap , temperature , reorganization energy and resonance energy is a subject of numerous theoretical studies in the context of photosynthesis .
however , no simple formula taking into account the influences of all these parameters on the degree of delocalization exists . after the discussion of the parameters of the carotenoid - bchl dimer
, we have many reasons to believe that the interaction of the @xmath2 and @xmath6 states can be treated in the weak coupling limit .
the issues is however more complicated because of the nature of the system - bath interaction in the carotenoid .
most of its reorganization energy can be assigned to the fast intramolecular vibrational modes .
the spacing between the vibrational levels and the relative electronic energy gap @xmath137 are such that the @xmath6 state can be viewed as effectively interacting with the nearest vibrational levels and not with the carotenoid electronic state as a whole .
this is a situation similar to the one studied in refs .
@xcite where special resonances between vibrational and electronic energy gaps lead to pronounced effects despite the small huang - rhys factors of the bchls .
clearly , treating the carotenoid vibrations as a thermodynamic bath is a questionable approximation . in the following subsections
we will briefly discuss the carotenoid - bchl energy transfer rates based on the weak coupling frster theory , and compare it with an explicit treatment of the carotenoid vibrations in the hamiltonian , along with refs . @xcite .
the energy level structure characteristic for the two situations is presented in fig .
[ fig : interaction - schemes ] . interaction schemes of the electronic and vibrational levels in ( a ) localized ( frster ) and ( b ) delocalized ( vibronic ) regime . in the case
( a ) the transition energy from the non - equilibrated state of the carotenoid excited state vibrations has to be resonant with the @xmath6 transition energy .
the relative energy gap @xmath137 between the @xmath2 and @xmath6 transitions requires involvement of more than one vibrational quanta of the carotenoid .
the vibronic case ( b ) also involves excitonic mixing of the vibrational levels of the carotenoid with the @xmath6 transition .
the collective single excited states of the carotenoid - purpurin or carotenoid - bchl dimer involve states in which the @xmath6 transition is excited and simultaneously the electronic ground state vibrations of the carotenoid are excited .
the states can excitonically mix with the carotenoid electronically excited state with a different number of vibrational quanta . ] in the weak resonance coupling limit , the parameter @xmath116 of the hamiltonian , eq .
( [ eq : qxs2_ham ] ) , is assumed small and perturbation theory to the second order can be performed . for two transitions
( donor and acceptor ) , the transition rate is given by the fermi golden rule as@xmath138 it is possible to view the donor molecule as a set of many available de - excitation transition ( the molecule is excited electronically and vibrationally ) while the acceptor molecule can be viewed as a set of transitions ready to be excited . in a realistic case
it is therefore necessary to integrate the fermi formula over all acceptor and donor transitions @xmath139 @xmath140 where @xmath141 and @xmath142 correspond to the normalized distribution of the transition energies available for excitation on the acceptor and the normalized distribution of the transition energies available for the deexcitation on the donor .
the frster rate can be derived directly from eq .
( [ eq : kad_distrib ] ) by noticing that the acceptor absorption spectrum is related to the distribution of transition frequencies @xmath143 as @xmath144 and similarly for the donor fluorescence spectrum @xmath145 ( see e.g. @xcite ) or directly by perturbation theory with respect to @xmath116 and a cumulant expansion for the bath dof as in ref .
@xcite . in both cases there is a direct relation between the rate and the overlap of absorption spectrum of the acceptor with the emission spectrum of the donor @xmath146 it is important to note that the frster theory does not in general depend on the dipole - dipole coupling approximation although the specific dependence of the rate in this approximation is often used . from the point of view of the present manuscript ,
the coupling energy @xmath116 can be calculated by any type of advanced methods of quantum chemistry which takes into account all the details of the electronic structure of the excited states of the interacting molecules @xcite . from the theoretical point of view
, the frster rate can be derived from hamiltonian , eq .
( [ eq : qxs2_ham ] ) , under various assumptions .
usually one assumes relaxed state of the environment and the intramolecular vibrational modes .
however , frster theory is very versatile and allows many generalizations @xcite , most importantly it is still valid for the case that the excited state of the donor is unrelaxed @xcite .
( [ eq : kad_distrib ] ) is then still valid , but it does not result in eq .
( [ eq : foerster_final_relaxed ] ) .
frster theory therefore also allows for a direct experimental estimation of the resonance coupling if the fluorescence spectrum of the donor and the absorption spectrum of the acceptor are known , even in case that the fluorescence does not start from a relaxed state . for the present dyad , the coupling @xmath116 was determined to be @xmath147 @xmath65 , within the framework of eq .
( [ eq : foerster_final_relaxed ] ) .
@xcite fig .
[ fig : interaction - schemes](a ) describes a typical situation for the short time dynamics of the @xmath148 energy transfer .
the higher vibrationally excited levels of the @xmath2 are populated , and in order to achieve resonance with the @xmath6 transition , even higher vibrational levels on the carotenoid ground state are excited after deexcitation of the donor ( @xmath2 ) .
the energy transfer is enabled by the fact that the energy corresponding to the @xmath137 is accepted by the ground state vibrational levels of the carotenoid molecule as depicted in fig .
[ fig : interaction - schemes](a ) .
the same statement can be rephrased in terms of the spectral overlap .
because the rate depends on the overlap of the donor emission and the acceptor absorption spectra , the effective broadening of the carotenoid spectrum due to the fast vibrational modes increases significantly the energy transfer rate .
frster theory thus assigns the electronic ground state vibrational levels of the carotenoid a crucial importance in the energy transfer process .
the main conclusion of the preceding section is that the vibrational modes of the carotenoid are the main driving element of the energy transfer from @xmath2 to @xmath6 .
frster theory shows that these modes provide resonance for the energy transfer and that this resonance involves multiple vibrational quanta . as one can learn from fig .
[ sec : figure absorption](a ) , absorption lineshape undergoes a slight change in the carotenoid region suggesting weak delocalization between @xmath6 and @xmath2 states , at least for dyad .
it is therefore natural to include the important vibrational modes explicitly into the hamiltonian , and to treat only the remaining overdamped modes as a bath .
this leads directly to the model of vibronic excitons ( see refs .
such a model automatically treats the selected vibrational modes non - perturbatively .
we split the nuclear part of the carotenoid hamiltonian in eq .
( [ eq : qxs2_ham ] ) into some selected modes denoted by coordinates @xmath149 and referred to later in the text as primary modes .
the rest of the modes which we denote by @xmath150 ( see sec . [ sub : carotenoid - chlorophyll - interaction ] for the notation ) will be referred to as secondary modes or secondary bath .
we have @xmath151 where @xmath152 is the hamiltonian of the secondary bath and the hamiltonian of the primary modes and their interaction with the secondary bath read as @xmath153 and @xmath154 here , @xmath155 are the frequencies of the primary modes to be treated explicitly , and @xmath156 are the coordinates from the set @xmath150 coupled to the primary modes by some coupling constants @xmath157 . in the excited state of the carotenoid , the bath hamiltonian is now split in the following way @xmath158 @xmath159 here , we assume that the same secondary bath which drives the ground state vibrations into the equilibrium , drives also the excited state vibrations to their corresponding equilibrium . correspondingly , the energy gap operator reads as @xmath160 @xmath161 we assumed that the spectral densities describing the interaction of the selected modes with the rest of the bath are the same . the energy gap operator , eq .
( [ eq : dvs_with_explicit ] ) , is composed of three parts , @xmath162 representing the overdamped bath , the part representing the direct influence of the selected modes on the energy gap fluctuations @xmath163 and the part describing the influence of the overdamped bath on the electronic energy gap functions via the coupling with the selected modes .
the strength of the latter part of the interaction hamiltonian is given by the shift @xmath164 of the excited state vibrational potential with respect to the ground state potential .
the essence of the vibronic model is to include @xmath165 to the hamiltonian which we treat explicitly and only an effective energy gap operator @xmath166 is treated by perturbation theory .
it leads to the bath correlation function @xmath167 where the last term on the right hand side corresponds to the direct contribution of the secondary bath to the dephasing .
the damping of the @xmath35th primary modes is governed by the bath correlation function originating from the term , eq .
( [ eq : hvib - b ] ) , @xmath168 while the direct contribution of the secondary bath to the dephasing is expressed by the same correlation function scaled by the dimensionless shift of the effective vibrational mode @xmath169 the calculation of the energy transfer dynamics in the vibronic model thus starts with a diagonalization of the hamiltonian @xmath170 the diagonalization results in eigenstates which combine the electronic and vibrational states into the mixed vibronic states .
in a second step the rates of population transfer and coherence dephasing are calculated for the set of eigenstates by second order perturbation theory and cumulant expansion @xcite using correlation function , eq . ( [ eq : cbn ] ) , for the damping of the selected mode and the correlation function @xmath171 only to simulate electronic dephasing .
this means that we assume the correlation function @xmath172 , eq .
( [ eq : corr_fce_tot ] ) , to be composed of the contribution of the secondary bath only .
this choice decreases the number of degrees of freedom in fitting and effectively ties strength of the coupling between the primary mode and the secondary bath to the strengths of the interaction between the electronic transition and the primary mode . because only effective spectral density is important for the calculation of the transition rates
, we believe that the advantage in limiting the number of parameters outweighs the restriction put on our model . in principle one could estimate the overall rate of energy transfer directly by weighting the calculated rates between individual levels .
we have however chosen the more practical approach of calculating the transfer time as described in section [ sub : calculation - of - transfer ] .
because of the rather large vibrational frequency and moderate coupling , the @xmath6 transition can be viewed as interacting only with the energetically nearest transition .
this situation is depicted in fig .
[ fig : interaction - schemes](b ) . for the sake of clarity of the graphical presentation ,
we depict a situation of perfect resonance between the relative energy gap @xmath137 and the vibrational frequency and show energy level splitting of the energetically nearest levels .
[ fig : interaction - schemes]b depicts the relevant carotenoid and bchl energy levels as two separate entities ( in gray ) and in terms of collective states of the carotenoid - bchl dimer in the case of @xmath173 .
conceptually , the ground state vibrational levels of the carotenoid contribute to an excited collective state if the @xmath6 state is excited .
this is a consequence of the fact in a singly excited aggregate state all molecules except one are in their ground state ( carotenoid in this case ) .
the ground state vibrational levels of the carotenoid thus interact resonantly with the vibrational levels of the carotenoid in the electronically excited state @xmath2 .
this interaction corresponds to the energy transfer from @xmath2 to @xmath6 with a simultaneous deposition of the excess energy to the ground state vibrations of the carotenoid .
the complete picture of the energy levels of the system is much more complicated because the excitonic mixing involves all available states .
the details of this mixing , which are included in the basis transformation from the localized states to the energetic eigenstates , lead to discernible changes in the lineshape of the carotenoid molecule . as will be demonstrated in the following sections , the vibronic model provides an explanation for the observed spectral changes as shown in fig .
[ sec : figure absorption](a ) .
also , the detailed treatment of the interaction between carotenoid vibrational levels and the @xmath6 state leads to the experimentally observed energy transfer rates with a significantly weaker resonance coupling than predicted by eq .
( [ eq : foerster_final_relaxed ] ) . the vibronic model discussed in the previous section
was implemented as described in section [ sub : calculation - of - transfer ] .
the absorption spectra of monomeric carotenoid was fitted with a model hamiltonian involving one effective high frequency mode to obtain starting values of the effective mode parameters for the subsequent fitting of the absorption spectra of the dyad and the lh2 complex .
we found the frequency of the vibrational mode in the dyad to be @xmath74@xmath65 and its huang - rhys factor was determined to be @xmath75 .
all monomeric parameters are discussed in section [ sub : calculation - of - transfer ] . in fig .
[ fig : absorption - spectra - dyad ] we summarize our fitting efforts for the dyad .
we chose the dyad because the same solvent can be used for the dyad and its components avoiding solvent related bathochromic shifts .
the absorption spectra of lh2 _ m. pur . _ and lh2 _ rps .
_ were equally well reproduced with the employed vibronic model .
[ fig : absorption - spectra - dyad](a ) presents the result of summation of the dyad s monomer spectra .
this is the spectrum corresponding to the weak resonance coupling case ( frster regime ) and it is presented to highlight the changes between monomeric and dyad spectra . in fig .
[ fig : absorption - spectra - dyad]b , we present a result of the fitting in which the huang - rhys factor of the carotenoid , the @xmath6 transition frequency , the relative energy gap @xmath137 between @xmath2 and @xmath6 and the resonance coupling @xmath116 were allowed to change .
the fitting achieves good agreement with the experimental dyad spectrum , most importantly in the carotenoid related part .
there the relative hight of the absorption maxima of different vibrational peaks changes between the monomer and the dyad .
the estimated coupling is @xmath174@xmath65 while @xmath175 , @xmath176@xmath65 and @xmath177@xmath65 .
the minor change with respect to the parameters of the monomers , such as the change of the transition energy to the excited state , have to be assigned to the chemical change upon formation of the dyad.@xcite the same fitting procedure with the fixed resonance coupling @xmath116 estimated from measured fluorescence spectrum by frster theory , @xmath147 @xmath65 @xcite , results in a less satisfactory agreement as can be seen in fig .
[ fig : absorption - spectra - dyad](c ) . in order to demonstrate that lineshape changes in all cases are not only due to the change of the monomeric huang - rhys factor of the carotenoid molecule
, we show the spectrum of the dyad with the parameters resulting from the same parameters but with @xmath173 @xmath65 in fig .
[ fig : absorption - spectra - dyad](d ) .
interestingly , the best fit is achieved with a change of the huang - rhys factor of less than @xmath127 % and the change in the line shape can be therefore mostly assigned to the resonance coupling .
absorption spectra of the dyad .
panel ( a ) shows the sum of the monomeric spectra , which corresponds to the weak coupling case ( no changes to the lineshape except of lifetime broadening ) ; ( b ) presents the best fit of the dyad spectrum with the @xmath6 transition energy , relative energy gap @xmath137 , carotenoid hr factors and the resonance coupling free to be adjusted .
panel ( c ) shows the best fit with the resonance coupling fixed at @xmath147 @xmath65 and panel ( d ) presents the sum of the monomers with the parameters corresponding to the best fit from panel ( b ) , but with @xmath173 @xmath65 . ]
the simulation of the absorption spectrum has resulted in a value of @xmath116 which is by @xmath178 % smaller than the one estimated from the experiment , suggesting that higher value is not compatible with the spectral changes with respect to the monomers seen in the dyad .
the obvious question is whether this lower value for the resonance coupling is compatible with the measured ultrafast rates . to this end
we perform calculations of the @xmath2 to @xmath6 transfer dynamics to estimate the transfer time introduced in section [ sub : calculation - of - transfer ] for various values of @xmath116 and @xmath137 and keeping other parameters of the model fixed to those from the best fit from fig . [
fig : absorption - spectra - dyad](b ) .
the plot of transfer time as a function of @xmath116 and @xmath137 is presented in fig .
[ fig : transfer - time ] .
the calculations yield a transfer time of @xmath179 fs for the values of the best fit which is in good agreement with the experimental value of @xmath180 fs.@xcite moreover the 2d plot in fig .
[ fig : transfer - time ] and the corresponding cut at the value of @xmath174 @xmath65reveals a periodic modulation of the transfer time with the period corresponding to the effective spacing between vibrational levels .
the transfer time decreases monotonically with increasing value of @xmath116 , as shown in the lower panes of fig .
[ fig : transfer - time ] representing cuts along the experimental vales of @xmath137 .
the periodic modulation of the transfer time is in agreement with the resonant involvement of the carotenoid vibrational modes in the energy transfer . as in the frster case
this implies the involvement of the carotenoid vibrational levels from the electronic ground state .
conservation of energy during the energy transfer requires the excess energy to be deposited to the vibrational energy of the carotenoid .
transfer time as a function of @xmath116 and @xmath137 for the dyad ( a ) and for lh2 _ m. pur . _
( b ) . at the experimental value of @xmath137 and the values of resonance coupling determined by fitting absorption spectrum , transfer time for the dyad and lh2 _ m. pur
. _ are 55 fs and 40 fs , respectively , in agreement with the values reported in pump probe measurements . ] in order to assign the observed features of the experiment 2d - spectrum we performed simulations of the carotenoid 2d spectra .
this is a first step towards a complete simulation of the interacting car - bchl system , which allows us to assign car only features and possibly identify 2d spectral components originating in @xmath6 state .
such a full calculation will be a subject of our future work .
calculating the 2d spectrum in a spectral region covering the edge of the studied molecule is an extremely difficult task given the fact that a good fit on the outer region of ( even the absorption ) spectrum depends on many rarely studied details of the chromophore interaction with its environment . for instance , assumptions about the gaussian disorder of the electronic transitions energies and typical estimates for bath spectral densities are well suited for explaning the features around the maximum of absorption and do not cover well the situation at the edges of the spectra . on the other hand , introducing more fitting freedom by assuming arbitrarily some other types of spectral densities and disorder distributions seems not to yield more theoretical insight , as the new parameters can not be fixed by a limited set of experiments .
below we therefore describe a calculation of the car 2d - spectra in which the car molecule should well represent the car component of the lh2 .
correspondingly a limited fitting to the experiment 2d - spectra from fig .
[ sec : figure 2d_experimental ] is done , keeping in mind that the two systems which are compared in such a fitting are different .
large part of the car parameters are therefore motivated by the values from literature . from the linear absorption spectrum of the lh2 complex the electronic excitation energy @xmath181{cm^{-1}}$ ]
was obtained ( see fig . [
sec : figure absorption ] ) .
because of the line shape function approach , which relies on second - order cumulant expansion with the electronic ground state as the reference state , this value corresponds to the vertical transition energy , i.e. the sum of the electronic excitation energy and the reorganization energies of the spectral density components .
the transfer rates were chosen under the assumptions @xmath182{fs}$ ] and @xmath183{fs}$ ] in agreement with @xcite .
because of the slow transfer from @xmath24 to other electronic states , the lifetime broadening constant @xmath49 was taken as zero .
even though some spectral density parameters for okenone are available from the literature @xcite , the respective parameters for @xmath92-carotene have been reported in more detail .
they can at least provide some orientation for the choice of the parameters to model the okenone component of the investigated lh2 complex . according to @xcite
the frequencies of the included vibrational modes are @xmath184{cm^{-1}}$ ] and @xmath185{cm^{-1}}$ ] .
the huang - rhys factors of the vibrational modes in @xmath23 were assumed as @xmath186{}$ ] and @xmath187{}$ ] . while these values are somewhat smaller than the ones for @xmath92-carotene given in ref .
@xcite , their ratio is similar . according to the tendency reported in the literature
, the huang - rhys factors in @xmath24 were assumed to be larger than those in @xmath23 by a scaling factor chosen as . to describe the damping of the vibrational modes in @xmath24 with a time constant of @xmath188{fs}$ ]
, ten matsubara terms were taken into account in the calculation of the lineshape function .
for one of the brownian oscillator spectral density components a fast decay with damping constant @xmath189{fs}$ ] and a reorganization energy @xmath190{cm^{-1}}$ ] in agreement with ref .
@xcite was assumed . to reproduce the inhomogeneous broadening effects in the absorption spectrum of the lh2 complex , the second brownian oscillator spectral density component with very slow decay constant of @xmath191{ps}$ ]
@xcite was included in the model . for a reorganization energy of @xmath192{cm^{-1}}$ ] comparable inhomogeneous broadening effects as in a measured linear absorption spectrum could be obtained and features in the measured 2d - spectra could be qualitatively reproduced . in ref . @xcite it has been reported that coherence gets lost during this population transfer process , so that the fluctuations in @xmath23 and @xmath24 can be considered as uncorrelated .
accordingly , @xmath193 and @xmath194 were taken as zero .
the huang - rhys factors for excitation from @xmath24 to @xmath25 are known to be much smaller than the ones of the excitation from @xmath22 to @xmath23 . in ref.@xcite a value of @xmath195{}$ ]
is given , which was assumed for both modes in our calculation . according to our interpretation
, this value of the huang - rhys factor is given with respect to @xmath22 , as the vibrational energy for the transition between @xmath24 and @xmath25 reported in ref .
@xcite can not be explained by such a small huang - rhys factor .
however , the much larger huang - rhys factors of @xmath24 with respect to @xmath22 allows for a larger vibrational energy than expected from the value of @xmath196 for the huang - rhys factor . regarding the brownian oscillator modes ,
it has been reported in @xcite that for double excitation from @xmath24 the reorganization energy corresponds to a fraction of only @xmath197{}$ ] of the corresponding reorganization energy in @xmath24 from the electronic ground state , which was taken as the same as for @xmath23 in our description .
the vertical transition energy between @xmath24 and @xmath25 was chosen as @xmath198{cm^{-1}}$ ] in agreement with ref .
@xcite . in the prefactors of the esa response functions @xmath199{}$ ] entered , following the values given in @xcite for other carotenoid derivatives .
the finite pulse widths were taken into account in terms of gaussian profiles determined from a fit of the measured local oscillator spectrum . in this way
the central frequency @xmath200{cm^{-1}}$ ] and a fwhm of @xmath201{cm^{-1}}$ ] were obtained . as the fitted curve corresponds to the squared pulse profile , multiplication of the fwhm with a factor of @xmath202
was required to obtain the fwhm of the single pulses with a value of @xmath203{cm^{-1}}$ ] . in the real part of the sum of all calculated response function contributions
after convolution with the pulses an intensive , positive - valued diagonal peak in the region of ( @xmath204{cm^{-1}}$],@xmath205{cm^{-1}}$ ] ) appears at all population times up to @xmath206 fs , starting from @xmath207{fs}$ ] .
this peak mainly stems from the rephasing gsb contribution .
furthermore , below the diagonal at ( @xmath204{cm^{-1}}$],@xmath208{cm^{-1}}$])a crosspeak from the rephasing gsb contribution is found , which stems from vibrational effects .
this initially oval peak becomes butterfly - shaped with a nodal line separating positive an negative region at @xmath209{fs}$ ] , whereas at @xmath210{fs}$ ] it becomes completely positive again . at @xmath211{fs}$ ] a modified shape of the vibrational crosspeak below the diagonal from the rephasing gsb contribution in combination with a rise of of the rephasing se contribution and
the negative - valued rephasing esa contribution leads to a localization of the crosspeak close to ( @xmath212{cm^{-1}}$],@xmath213{cm^{-1}}$ ] ) . from @xmath214{fs}$ ]
the esa contribution starts to obscure the latter . at @xmath215{fs}$ ]
the negative esa peak covers a broad energetic range below the diagonal .
this tendency becomes even more pronounced at @xmath216{fs}$ ] and @xmath217{fs}$ ] , where the esa peak pushes the diagonal peak almost completely above the diagonal .
different from the measured 2d - spectra , the upper diagonal peak at @xmath18=0 to 30fs is missing in the calculated 2d - spectra shown in fig.[fig : simulated - electronic-2d ] , most likely due to the fact that the spectral phase of the pulses was considered as flat in the calculations .
when comparing the spectra from experiment ( fig.[sec : figure 2d_experimental ] ) and simulation ( fig.[fig : simulated - electronic-2d ] ) at @xmath18=30fs , it becomes apparent that the experimentally obtained cross peak , which we identified as an indication of @xmath8 energy transfer , is missing in simulations .
considering that the simulations only incorporate the okenone - part of the spectrum , this missing cross peak confirms our assignment to energy transfer to @xmath6 .
2d - es elucidates why carotenoid to bacteriochlorophyll energy transfer pathways are difficult to analyze in pump - probe spectroscopy .
@xcite the according cross peak coincides with a recurring negative carotenoid feature ( see @xmath18=30fs and @xmath18=10fs spectrum in fig.[fig : simulated - electronic-2d ] ) . additionally , @xmath8 energy transfer is overlaid with esa from @xmath3 , making it only observable within the ultrashort lifetime of @xmath2 , i.e. only during 55fs in lh2 _ m. pur_.
in this work we have shown that a vibronic coupling mechanism describes several central aspects of the carotenoid to bchl energy transfer dynamics , yielding moderate coupling constants @xmath116 close to 100@xmath65 .
this value is in excellent agreement with structure - based calculations and reproduces experimental absorption spectra well .
the low values of @xmath116 obtained from the vibronic model suffice for explaining the experimentally observed ultrafast transfer rates .
this is explained by vibronic resonances , which have the potential of dramatically speeding up energy transfer .
@xcite hence , the vibronic coupling mechanism circumvents the shortcomings of frster theory and explains static as well as dynamical properties of natural light harvesters .
the investigated carotenoid - bchl system is an extreme case of a heterodimer , meaning that donor ( carotenoid ) and acceptor ( bchl ) differ in transition energies and hr - factor .
future studies will test the vibronic coupling mechanism on energy transferring dimers where the spectroscopic properties are more alike , such as perylene bisimide dyads.@xcite in such systems , both donor and acceptor show strong electron - phonon coupling , which should make vibronic effects in energy transfer dynamics even more pronounced .
another highly promising class of systems is bulk heterojunction solar cells , where vibronic coupling between donor and acceptor was recently suggested to be the mechanism behind ultrafast electron transfer.@xcite the authors wish to thank ana moore for providing the dyad sample and heiko lokstein for fruitful discussion .
funding by the austrian science fund ( fwf ) : start project y631-n27 ( j. h. and c. n. l. ) , the cze - aus mobility grant ( grant no .
7amb14at007 , cz 05/2014 ) ( j. h. and c. n. l. , v. p. , j. s. , f. s and t.m . ) and by the czech science foundation ( gacr ) grant .
14 - 25752s ( v. p. , j. s. , f. s and t.m . ) is acknowledged .
r. j. c. and l. j. c acknowledge funding by bbsrc .
in this appendix , the response functions used to calculated the car 2d - spectrum are specified . the rephasing gsb term with population of @xmath23 and a decay due intramolecular transfer to @xmath24
is given as v. sundstrom , in _ biophotonics : spectroscopy , imaging , sensing , and manipulation _ , _ nato science for peace and security series b - physics and biophysics _
, edited by b. dibartolo and j. collins ( publisher , address , 2011 ) , pp .
219236 .
j. savolainen , r. fanciulli , n. dijkhuizen , a. l. moore , j. hauer , t. buckup , m. motzkus , and j. l. herek , proceedings of the national academy of sciences of the united states of america * 105 * , 7641 ( 2008 ) . | the initial energy transfer in photosynthesis occurs between the light - harvesting pigments and on ultrafast timescales .
we analyze the carotenoid to bacteriochlorophyll energy transfer in lh2 _ marichromatium purpuratum _ as well as in an artificial light - harvesting dyad system by using transient grating and two - dimensional electronic spectroscopy with 10fs time resolution .
we find that frster - type models reproduce the experimentally observed 60fs transfer times , but overestimate coupling constants , which leads to a disagreement with both linear absorption and electronic 2d - spectra .
we show that a vibronic model , which treats carotenoid vibrations on both electronic ground _ and _ excited state as part of the systems hamiltonian , reproduces all measured quantities .
importantly , the vibronic model presented here can explain the fast energy transfer rates with only moderate coupling constants , which are in agreement with structure based calculations .
counterintuitively , the vibrational levels on the carotenoid electronic ground state play a central role in the excited state population transfer to bacteriochlorophyll as the resonance between the donor - acceptor energy gap and vibrational ground state energies is the physical basis of the ultrafast energy transfer rates in these systems . |
particle identification in the belle experiment is based upon a composite system of subdetectors , as illustrated in fig .
[ belle_pid ] .
this hybrid system consists of ionization loss measurements ( de / dx ) in the central drift chamber ( cdc ) , cherenkov light emission measurement in the barrel and endcap aerogel chernkov counters ( acc ) , and flight time measurement in the time of flight ( tof ) system . as indicated in the lower section of this figure
, the three systems work together to cover the momentum range of interest . of these recording systems ,
the tof system makes the most severe demands on time resolution .
indeed , given the 2ns spacing between rf buckets ( and possible collisions ) , it is not known at recording time to which collision a given particle interaction in the tof system corresponds .
= 3.2 in precision time recording in a very high rate environment requires an encoding scheme capable of continuous recording , with a minimum of deadtime per logged event . at the time of the construction of the belle experiment @xcite at the kek
b - factory @xcite , a decision was made to unify the entire detector readout ( except for the silicon vertex detector ) on the lecroy 1877 multi - hit tdc module .
this fastbus module is based upon the mtd132a asic @xcite , which has a 0.5ns resolution encoding , comparable to a number of similar devices @xcite .
given the limited manpower for daq system development and maintenance , this proved to be a wise choice .
the intrinsic time resolution was quite adequate for recording the timing information from the cdc , as well as the amplitude information ( through use of a charge - to - time converter ) for the cdc and acc .
the challenge then was to be able to record pmt hits with 20ps resolution , using a multi - hit tdc having 500ps least count , and for collisions potentially separated by only 2ns .
this latter constraint meant that traditional techniques using a common start or stop could not be applied , since the bunch collision of interest was not known at the time at which the hits need to be recorded .
moreover , in order to avoid incurring additional error due to comparing a separate fiducial time , it is desirable to directly reference all time measurements to the accelerator rf clock .
the solution adopted was a so - called time stretcher circuit , developed by one of the authors in conjunction with the lecroy corporation @xcite .
this work built upon valuable lessons learned in developing a similar recording system for the particle identification detector system of the cplear experiment @xcite .
the principle of operation is seen in fig .
hits are time - dilated with respect to the accelerator clock and recorded at coarser resolution , but in direct proportion to the stretch factor employed .
statistically , by also logging the raw hits , this stretch factor can be determined from the data .
= 3.2 in as seen in fig .
[ ts ] , four timing edges are recorded for each hit signal .
the leading edge corresponds to the actual output time of the discriminator .
this rising edge is paired with a falling edge , corresponding to the 2nd accelerator reference clock ( rf clock divided by 16 ) occuring after the initial hit timing .
the interval of interest is then bounded to be between about 16 - 32 ns . with a tdc least count of 0.5ns , a factor of twenty time expansion
is needed the stretch factor . in the figure
the third edge corresponds to the time - expanded version of the interval between the rising and falling edges .
a benefit of this technique is that it provides self - calibration . by recording a large number of events
, the stretch factor can be extracted from the data itself since the raw and expanded signals are recorded .
a 4th edge is provided , two clock rising edges after the 3rd edge , to provide a return to known state before next pulse .
an obvious drawback in this scheme is that the deadtime for each hit will be something like 320 - 640ns , as will be discussed later . in more detail
, the signal chain of the current belle tof electronics @xcite , is sketched in fig .
[ toffee ] .
= 3.2 in high and low level discriminators are used , with the high - level used to reject background photon hits in the tof and a low - level threshold used to provide the best possible leading edge timing .
the charge of triggered events is also recorded with a charge to time ( q - to - t ) asic , which is recorded with the same common tdc module .
charge recording is needed to correct for amplitude dependent timing effects in the discriminator itself .
the tof readout system has worked well for almost a decade . increased luminosity ( already 60% over design ) has lead to much higher single channel rates than had been specified in the design . from the beginning ,
the maximum design specification was 70khz of single particle interaction rate for each channel . at this rate
the expected inefficiency would be a few percent , comparable to the geometric inefficiency ( due to cracks between scintillators ) . already the world s highest luminosity collider , the kekb accelerator @xcite can now produce in excess of one million b meson pairs per day . upgrade
plans call for increasing this luminosity by a factor of 30 - 50 , providing huge data samples of 3rd generation quark and lepton decays . precise interrogation of standard model predictions will be possible , if a clean operating environment can be maintained .
extrapolation of current occupancies to this higher luminosity mandates an upgrade of the readout electronics .
the current system already suffers from significant loss of efficiency with higher background rates , as may be seen in fig .
[ occ ] .
in considering an upgrade to the tof readout electronics , it is worthwhile to consider the needs of an upgraded pid system for belle . a comparative study of the belle system , as depicted in fig .
[ belle_pid ] , with that of babar @xcite pid system is informative .
it is clear in fig .
[ pidbakeoff ] the direct internally reflected cherenkov ( dirc ) detector of babar has a higher efficiency and lower fake rate than the hybrid tof / acc scheme used by belle .
= 5.5 in indeed , it was realized in the construction stage of belle that such a dirc - type detector would have merits , and prototypes were explored @xcite . while these results were very promising ,
the schedule risks led the collaboration to stick with technologies in which significant time and effort had already been invested .
thinking about an upgrade , it is reasonable to revisit the choice of technology . in the intervening decade
, significant progress has been made in the development of ring imaging cherenkov ( rich ) detectors @xcite , as well as detectors based upon the arrival time of the cherenkov photons , such at the correlated cherenkov timing ( cct ) @xcite and time of propagation ( top ) @xcite counters . because of the great cost encumbered in the procurement and construction of the csi crystal calorimeter , it is planned not to upgrade the barrel section . as a consequence , the volume available for the tof / acc replacement detector is rather limited .
therefore a rich type detector has not been pursued .
the most promising technologies to date are those illustrated in fig .
[ pidconcept ] .
the top concept uses timing in place of one of the projected spatial dimensions to reconstruct the cherenkov emission ring .
a focusing dirc is principally using geometry to reconstruct the cherenkov ring segments .
however , in this case precision timing is still very useful for two important reasons .
first it allows for the possibility of using timing to correct for chromatic dispersion in the emission angle of the cherenkov photon . and
second , fine timing allows time of flight to be measured using the quartz radiator bar .
therefore , in both of the viable detector options considered , a large number of fine timing resolution recording channels are required . in the case of a finely segmented focusing dirc @xcite option
, the number of readout channels could be comparable to that of the current silicon vertex detector .
clearly if such a detector is to be viable , significant integration of the readout electronics will be essential .
not shown is a proposal for a multi - segmented tof detector consisting of short scintillator bars .
while this option remains viable ( and the electronics presented would work well with such a system ) , the pid performance degradation of such a system is probably unacceptable . of the choices listed ,
the most attractive in terms of performance is a focusing dirc detector , if the issues of the photodetector and readout can be addressed .
either as an upgrade of only the readout electronics or as a prototype for a higher channel count pid detector , it is worth considering improvements to the existing readout .
the time stretcher technique has worked very well and belle has been able to maintain approximately 100ps resolution performance with the tof system .
a slow degradation with time is consistent with loss of light output .
detailed monte - carlo simulation @xcite has been able to reproduce much of the performance of the tof system and the degradation is consistent with light loss due to crazing of the scintillator surface .
a larger concern is the significant degradation of tof system performance due to high hit rates .
while the multi - hit tdc is capable of keeping up with high rates ( though the limited number of recorded edges ( 16 ) also leads to inefficiency ) , by its very nature , the time stretcher output can not be significantly reduced .
recently , the clock speed was doubled , to help reduce this effect . nevertheless , at ever higher hit rates , the deadtime leads to ever increasing inefficiency .
= 5.5 in a logical solution to this problem is to introduce a device which has buffering .
also , while taking the effort to reduce the deadtime , it makes sense to consider a much more compact form - factor .
this was done with the thought toward moving to a larger number of readout channels in a future belle pid upgrade @xcite , as mentioned earlier .
one proposed solution is the monolithic time stretcher ( mts ) chip , a prototype of which is shown in fig .
[ mts1 ] .
the fundamental logic of the device is identical to that currently in use with two major changes : 1 .
high density 2 .
multi - hit high density is achieved by replacing discrete emitter - coupled logic components on daughter cards with a full custom integrated circuit .
this higher integration permits having multiple time stretcher channels for each input . by toggling to a secondary output channel
, the deadtime can be significantly reduced . once a hit is processed in one output channel ,
the next is armed to process a subsequent hit . in fig .
[ mts1 ] the 8 channel repeating structure of each time stretcher circuit is clearly seen in the die photograph .
the basics of the time - stretcher circuit are visible in fig .
[ ts_ckt ] .
a one - shot circuit at the upper left leads to an immediate output signal , as well as starts charging current @xmath0 .
pipelining of the hit signal continues for two clock cycles after which current @xmath0 is switched off and discharge current @xmath1 is switched on .
a comparator monitors the voltage induced on the storage capacitor due to charging and discharging , providing an output signal to indicate the stretched time when the voltage is discharged .
= 5.8 in = 5.8 in the stretch factor is given by the ratio of the two currents : @xmath2 .
each input channel of the mts1 has two time stretcher circuits , the second corresponding to the secondary output when the primary channel is active .
each output is recorded by a separate tdc channel . with this configuration at 10% deadtime for a single channel of time stretcher
can be reduced to 1% .
as the the incremental cost of additional tdc channels is rather low , it is possible consider additional buffering depths , which would reduce the deadtime by the @xmath3 , where n is the buffer depth , though that was not explored beyond a depth of two in this device .
reduction of cross - talk and electro - magnetic interference is enhanced by the use of low voltage differential signalling ( lvds ) @xcite .
mts1 is fabricated in the taiwan semiconductor manufacturing corporation @xmath4 m cmos process .
when considering a photodetector with a large number of channels , the form factor of this device is very attractive , as shown for comparison in fig .
[ mts_comp ] , a substantial reduction in size has been achieved . on the left
is a 16-channel fastbus - sized time stretcher card used currently in the belle experiment .
inset is a test board with one of the mts1 packaged devices for comparison , where a dime has been placed on the board for scale .
= 3.2 in with this level of integration it becomes feasible to consider integration of the time stretcher and tdc electronics on detector , as is being done for detector subsystems in the lhc experiments . in order to test the performance of the mts1 , a multi - hit tdc should be used . as a demonstration of the power of this time stretching technique , an field programmable gate array ( fpga )
can be used as this tdc @xcite , where the results from a simple gray - code counter implementation of the hit time recording may be seen in fig .
[ timing_resol ] .
the rms of the distribution is about 840ps for the xilinx spartan-3 device used .
this resolution could be improved by use of a faster fpga , though is sufficient to obtain the test results shown below .
indeed , it is worth noting that this combined time - stetcher + fpga technique is very powerful for two important reasons : 1 . low - cost , high - density tdc implementation 2 .
deep and flexible hit buffering and trigger matching logic = 3.2 in a test sweep of the mts1 input is shown in fig . [ mts1_linear ] , where it should be noted that due to the encoding scheme it is only meaningful to scan within a time expansion clock cycle period .
a scan of expansion ratios was performed and the best results were obtained for stretch factors of 40 - 50 .
= 3.2 in as can be seen , there is some non - linearity in the expanded time .
this is more clearly seen when a plot of the residual distribution is made by subtracting off the linear fit , as shown in fig .
[ mts1_resid ] .
a periodic structure is seen , roughly consistent with the expansion clock period , if the negative timing dips are correlated to transition edges .
= 3.2 in as with the hptdc device @xcite developed at cern for the alice detector , a fine calibration is needed to obtain a precision comparable to the current belle system .
applying such a calibration , determined in a separate data set , significantly improved linearity and residuals are obtained .
the subsequent results are histogrammed in fig .
[ mts1_resol ] .
= 3.2 in as can be seen , the timing resolution fits well to a double gaussian , with a narrow sigma less than 20ps , which is comparable to ( and actually slightly better than ) the existing belle system .
this result is consistent with the expectation from the fpga tdc used , where @xmath5 and the measured sigma is about 15ps .
it is possible that a finer resolution fpga tdc would allow for an even more precise timing determination . in practice
the systematic effects of the upstream discriminator and its amplitude dependent threshold crossing ( and comparator overdrive ) dependence make any further improvements difficult .
nevertheless it is an interesting question for future exploration .
this timing resolution is comparable to that obtained with the hptdc after careful non - linearity calibration .
the broader gaussian distribution and significant non - gaussian tails are correlated with expansion clock feedthrough to the ramping circuit .
this could be improved in a future version with better layout isolation .
the @xmath6 m process used only had 3 metal routing layers available and migration to a finer feature size process would allow for dedicated shields and better power routing . in order to reduce deadtime a second time - stretcher circuit , with a separate output ,
is provided for each input channel .
this second circuit becomes armed when the primary stretcher circuit is running .
use of such a scheme can significantly reduce data loss due to arrival of subsequent hit during operation of the first stretcher circuit .
the factor may be expressed as @xmath7 where n is the number of buffer stages .
for n=2 , the case prototyped here , a large existing deadtime of 20% could be reduced to 4% . moreover
, this technique can be extended to an even larger number of buffer channels , a realistic possibility when using a low cost fpga - based tdc . in the case of 4 outputs ,
a 20% single time stretcher deadtime would become a completely negligible @xmath8 .
apart from the arming circuitry , the second time stretcher channel is identical to the primary .
testing was performed with double - pulse events and the result for the second channel is seen in fig .
[ mts1_buff ] .
= 3.2 in note that these secondary channels have a time - stretch factor that is systematically smaller .
as the same reference currents are mirrored in all channels , it is believed that this is due to ramp window reduction due to latency in the arming logic .
an important check of performance of the mts1 is the impact of time stretcher operation on one channel while another is operating .
this has been performed in fig .
[ nts32_crosstalk ] , where the timing of the first channel is fixed and the timing relation of the signal in channel 2 is varied .
= 3.2 in the impact of operation of this second channel is clear during the ramping portion of the readout cycle , as well as the threshold crossing at the end of the ramping interval .
while this effect can be calibrated out to some extent , just like effects of the clock feedthrough , this perturbation to the circuit would be better mitigated through better isolation in the ic layout .
an improved layout paired with future , higher clock frequency fpgas could open the possibility of very dense channel count , sub-10ps resolution tdc recording . for many applications
the hptdc is perfectly suitable and gives comparable time resolution to the mts1 + fpga tdc . in both cases
a non - linearity correction is required to obtain this resolution . however the time encoding itself is only part of the issue for obtaining excellent timing resolution from a detector output .
correction for time slew in the discriminator threshold crossing is critical .
moreover the addition of many channels of high - speed discriminator inside a detector is a noise and power concern .
compact , high - speed waveform recording @xcite may be a promising next evolutionary step in the readout of precision timing detectors .
this work was supported by the us - japan foundation and the department of energy advanced detector research award number de - fg02 - 06er41424 .
g. blanar and r. sumner , _
new time digitizer applications in particle physics experiments _ , proc . of the first international conference on electronics for future colliders ,
editors g. blanar and r. sumner , lecroy corporation , new york ( 1991 ) ; technical data sheet , mtd132a , from lecroy corporation , 700 chestnut ridge road , chestnut ridge , ny 11977 . + _ a short list , not meant to be comprehensive : _
w. earle , e. hazen , b. safford , g. varner , `` a deadtimeless multihit tdc for the new g-2 experiment '' , prepared for 4th international conference on electronics for future colliders , chestnut ridge , ny , 11 - 12 may 1994 , pp 223 - 232 .
m. passaseo , e. petrolo , s. veneziano , `` design of a multihit tdc integrated circuit for drift chamber readout '' , prepared for 5th international conference on electronics for future colliders , chestnut ridge , ny , 10 - 11 may 1995 , pp 139 - 142 .
+ h. kichimi , y. yoshimura , t. browder , b. casey , m. jones , s. olsen , m. peters , j. rodriguez , g. varner , y. zheng , y. choi , d. kim , j. nam , _ `` the belle tof system '' _ , nucl .
. meth . * a453*:315 - 320 , 2000 .
nam , y.i .
choi , d.w .
kim , j.h .
kim , b.c.k .
casey , m. jones , s.l .
olsen , m. peters , j.l .
rodriguez , g. varner , y. zheng , n. gabyshev , h. kichimi , j. yashima , j. zhang , t.h .
kim and y.j .
kwon , `` a detailed monte carlo simulation for the belle tof system '' , nucl .
. meth . * a 491 * 54 - 68 ( 2002 ) .
low - voltage differential signaling ( lvds ) uses high - speed analog circuit techniques to provide multi - gigabit data transfers on copper interconnects .
it is defined under the electronic industries alliance ( eia)-644 - 1995 standard . | identifying light mesons which contain only up / down quarks ( pions ) from those containing a strange quark ( kaons ) over the typical meter length scales of a particle physics detector requires instrumentation capable of measuring flight times with a resolution on the order of 20ps . in the last few years a large number of inexpensive , multi - channel time - to - digital converter ( tdc ) chips have become available .
these devices typically have timing resolution performance in the hundreds of ps regime .
a technique is presented that is a monolithic version of `` time stretcher '' solution adopted for the belle time - of - flight system to address this gap between resolution need and intrinsic multi - hit tdc performance . |
coherent control of quantum dynamics via a periodically oscillating external field has been one of the subjects of long - lasting interest in diverse branches of physics and chemistry @xcite .
one seminal result of the control is coherent destruction of tunneling ( cdt ) , a phenomenon originally discovered by grossmann _
et al . _ in 1991 for a periodically driven double - well system @xcite , upon the occurrence of which the tunneling can be brought to a standstill provided that the system parameters are carefully chosen . due to its importance for understanding many fundamental time - dependent processes and its potential application in quantum motor @xcite and quantum - information processing romero , cdt
has received growing attention from both theoretical and experimental studies .
theoretically , it has been extended in various forms such as nondegenerate cdt @xcite , selective cdt @xcite , nonlinear cdt @xcite , many - body cdt @xcite , instantaneous cdt @xcite and multiphoton cdt @xcite ; experimentally , it has been observed in many different physical systems like modulated optical coupler @xcite , driven double - well potentials for single - particle tunneling @xcite , three - dimensional photonic lattices @xcite , a single electron spin in diamond @xcite , bose - einstein condensates in shaken optical lattices @xcite , and chaotic microcavity @xcite .
in addition to conventional approach of modulating a lattice in a uniform fashion , modulating some certain lattice sites selectively also provides an attractive alternative for generation of cdt , in which the rescaled tunneling amplitudes for different sites can be shut off effectively , thereby some intriguing quantum manipulations are realizable such as dissipationless directed transport @xcite and beam splitter @xcite .
cdt was traditionally thought to occur only at the isolated degeneracy point of the quasi - energies and is related to dynamical localizationdunlap .
however , recently a novel quantum phenomenon called dark cdt occurring over a wide range of system parameters has been introduced in odd-@xmath0-state systems @xcite which is demonstrated to be caused by localized dark floquet state that has zero quasi - energy and negligible population at the intermediate state , rather than the superposition of degenerate floquet states .
those advances on cdt studies offer benefits for coherent control of quantum dynamics . in this paper , we propose a scheme for the control of quantum tunneling in a periodically driven lattice array with controllable boundary through combination of characteristics of both normal cdt and dark cdt .
we consider the coherent motion of a quantum particle in a lattice array , in which the harmonic oscillating external fields act on only the two boundary sites of chain .
we have found an interesting result that a single particle initially occupying the left boundary site experiences coherent destruction of tunneling to coherent tunneling ( cdt - ct ) transition or coherent tunneling to coherent destruction of tunneling ( ct - cdt ) transition , when the driving amplitude of the external periodic field applied to the right boundary site is increased from zero .
we have also revealed that the cdt - ct transition in three - site lattice is robust against the second order coupling ( soc ) between next - nearest - neighbor sites , whereas the ct - cdt transition in four - site system is significantly affected by soc .
moreover , we present a good understanding of the results with help of the high - frequency approximation analysis as well as numerical computation of the corresponding floquet states and quasi - energies .
we consider a single particle in an array of lattice sites with only two boundary sites driven by external periodic field , as shown in fig .
( [ fig : figure1 ] ) . in our model ,
the left boundary site is driven with fixed driving amplitude and driving frequency , while the right boundary site is driven with varied driving amplitude and fixed driving frequency . in the tight - binding approximation and assuming a coherent dynamics , the single - particle motion can be generally described by the tight - binding hamiltonian @xmath1 where @xmath2 represents the wannier state localized in the @xmath3th site , @xmath4 is the coupling strength connecting nearest - neighboring sites , and @xmath5 is the second - order coupling ( soc ) strength between next - nearest - neighboring sites .
instead of modulating a lattice in a uniform fashion , we modulate the on - site energies selectively . in this scheme , we assume a harmonic oscillating field applied in form of @xmath6 for site 1 , and @xmath7 for site @xmath8 . here
@xmath9 and @xmath10 are the driving amplitudes and @xmath11 is the driving frequency , respectively .
we expand the quantum state of system ( [ equ : h ] ) as @xmath12 , where @xmath13 represents occupation probability amplitudes at the @xmath3th site , with the normalization condition @xmath14 . from the schrdinger equation @xmath15 , the evolution equation for the probability amplitudes @xmath13 reads
@xmath16 ) . here
the left boundary site is driven with fixed amplitude @xmath9 and frequency @xmath17 , and the right boundary site driven with varied amplitude @xmath10 and fixed frequency @xmath18 .
all other sites are undriven.,scaledwidth=48.0% ] in this work , we mainly illustrate how the quantum dynamics of a single particle initially localized at the left boundary site can be controlled by driving the other boundary site , for both zero and non - zero soc cases .
we start our considerations from the three - site lattice array , the minimal one for odd-@xmath8-site system . according to eq .
( [ equ : a ] ) , the evolution equations for the probability amplitudes read , @xmath19 we numerically solve the time - dependent schrdinger equation ( equ : a1 ) with @xmath20 , i.e. , neglecting next - nearest - neighbor tunneling in the chain .
the initial state is @xmath21 , and the driving parameters of site 1 fixed as @xmath22 .
the evolution of the probability distribution @xmath23 is presented in figure ( [ fig : figure2 ] ) for three typical driving conditions of site 3 . for @xmath24
, @xmath25 remains near unity , signaling suppression of tunneling .
this is the quantum phenomenon well known as cdt . for @xmath26
, @xmath25 oscillates between 1 and @xmath27 , showing partial suppression of tunneling . at @xmath28 = 2.4 ,
@xmath25 oscillates between zero and one , demonstrating no suppression of tunneling .
the numerical results clearly indicate that the system undergoes a cdt - ct transition when @xmath29 is increased from zero .
such a cdt - ct transition is more clearly demonstrated in fig .
( fig : figure2)(b ) , where the minimum value of @xmath25 is used to measure the suppression of tunneling .
when there is large suppression of tunneling , @xmath30 is close to 1 ; when there is no suppression , @xmath31 is zero . as clearly shown in fig .
( [ fig : figure2])(b ) , @xmath31 slowly falls from its initial value to zero with @xmath29 increasing from zero to @xmath32 .
it is observed that the value of @xmath31 drops to zero in narrow intervals around the zeros of @xmath33 , indicating that the particle is able to tunnel freely from site to site in these regimes . ) .
( a ) time evolution of the probability at site 1 , @xmath23 , with different values of @xmath10 ; ( b ) the minimum value of @xmath25 as a function of @xmath34 ; ( c ) numerical results of probability distribution @xmath25 versus the driving amplitude @xmath10 and time , obtained from the original model ( [ equ : a1 ] ) ; ( d ) analytical results of probability distribution @xmath25 versus the driving amplitude @xmath10 and time , given by the formula ( [ equ : p1 ] ) . the initial condition is @xmath35 .
the other parameters are chosen as @xmath36 , @xmath37 , @xmath38.,scaledwidth=48.0% ] we explain the above numerical results through some analytic deduction based on high - frequency approximation . in the high - frequency limit , we introduce the transformation @xmath39a_1 $ ] , @xmath40 , @xmath41a_3 $ ] , where @xmath42 are slowly varying functions . using the expansion @xmath43 = \sum\nolimits_k { j_k \left ( { a/\omega } \right)\exp \left ( { \ \pm ik\omega t } \right)}$ ] in terms of bessel functions and neglecting all orders except @xmath44 in the high frequency region , we arrive at the effective equations of motion , @xmath45 here we have dropped the soc terms .
it is easy to derive the analytical solutions of eq .
( [ equ : q1 ] ) , @xmath46 \nonumber \\ & b_2 = c_2 \cos \left ( { kt } \right ) + c_3 \sin \left ( { kt } \right ) \nonumber \\ & b_3 = c_1 - i\frac{{j_{02 } } } { { \sqrt { j_{01}^2 + j_{02}^2 } } } \left [ { c_2 \sin \left ( { kt } \right ) - c_3 \cos \left ( { kt } \right ) } \right],\end{aligned}\ ] ] where @xmath47 are constants to be determined by the initial states and normalization condition , @xmath48 , @xmath49 and @xmath50 . applying the initial conditions @xmath51 , @xmath52 , @xmath53 to eq .
( [ equ : q2 ] ) yields the undetermined constants @xmath54 in the forms @xmath55 , @xmath56 , @xmath57 and the occupation probability at site 1 as @xmath58 from expression ( [ equ : p1 ] ) , we immediately have two observations : ( i ) when @xmath59 , i.e. , the zeros of @xmath60 , the minimal value of @xmath61 is zero ; ( ii ) when @xmath62 is tuned near to the zeros of @xmath63 and @xmath64 tuned far away from those values , latexmath:[$% @xmath25 based on formula ( [ equ : p1 ] ) are plotted in fig .
( [ fig : figure2])(d ) , which agrees well with the the numerical results obtained from the original model ( [ equ : a1 ] ) as shown in fig .
( [ fig : figure2])(c ) .
the floquet theory provides a powerful tool for understanding the tunneling dynamics obtained in fig .
( fig : figure2 ) .
since hamiltonian ( [ equ : h ] ) is periodic in time , @xmath66 , where @xmath67 is the period of the driving , the floquet theorem allows us to write solutions of the schrdinger equation ( [ equ : a ] ) in the form @xmath68 . here
@xmath69 is the quasi - energy , and @xmath70 is the floquet state .
the floquet states inherit the period of the hamiltonian , and are eigenstates of the time evolution operator over one period of the driving @xmath71,\end{aligned}\ ] ] where @xmath72 is the time - ordering operator . noticing that eigenvalues of @xmath73 are @xmath74 ,
the quasi - energies of this system can be numerically computed by direct diagonalization of @xmath73 .
our numerical results of the quasi - energies and the floquet states for the modulated three - site system ( [ equ : a1 ] ) are plotted in fig .
( [ fig : figure3 ] ) .
there are three floquet states with quasi - energies @xmath75 , @xmath76 and @xmath77 .
we immediately notice from fig .
( [ fig : figure3])(a ) that there exists a dark floquet state with zero quasi - energy for all of the values of @xmath29 .
for this three - site system , there is no degeneracy in the quasi - energy levels .
therefore , the occurrence of suppression of tunneling should be further explored through probe into the floquet states .
we display the time - averaged population probability @xmath78 for a given floquet state @xmath79 in figs .
( fig : figure3)(b)-(d ) . the floquet state with @xmath80
is generally regarded as a state localized at the @xmath3th site . as seen in figs .
( [ fig : figure3])(c ) , the dark floquet state has negligible population at site 2 while the population @xmath81 holds for all values of @xmath29 except those in the vicinity of zeros of @xmath82 .
the other two floquet states are not localized at site 1 since their populations @xmath83 are lower than 0.5 . in a word ,
the cdt - ct transition shown in fig .
( [ fig : figure2 ] ) comes from the dark floquet state , whose population @xmath84 undergoes a localization - delocalization transition . ) .
( a ) quasi - energies versus @xmath85 . the time - averaged populations for the floquet state in the quasi - energy level ( b ) @xmath86 , ( c ) @xmath87 and ( d ) @xmath88 .
the other parameters are @xmath89 .
, scaledwidth=50.0% ] we are now in the position to investigate the quantum control of the four - site system . however , similar behaviors discussed in next subsection will appear in six - site lattice system as well . for a four - site system ,
the coupled equations ( [ equ : a ] ) read , @xmath90 we plot in fig .
( [ fig : figure4])(a ) the minimum value of @xmath25 versus @xmath28 by direct numerical simulations of the schrdinger equation ( [ equ : q3 ] ) with @xmath91 .
we start a particle at site 1 and fix the driving parameters of site 1 as @xmath22 . as shown in fig .
( fig : figure4)(a ) , @xmath92 takes extremely low values about zero except at a series of very sharp peaks .
the quasi - energies of this system are shown in fig .
( [ fig : figure4])(b ) , where we find that a pair of quasi - energies cross at the zeros of @xmath93 .
this is the origin of the extremely sharp peaks in localization seen in fig .
( [ fig : figure4])(a ) . figs .
( [ fig : figure4])(c)-(f ) show that there is no localized floquet states .
these numerical results demonstrate the possibility of inducing ct - cdt transition in four - site system through tuning the rescaled driving amplitude @xmath29 from zero to the points of quasi - energy crossing . as a function of @xmath94 .
the extremely narrow peaks appear at the zeros of @xmath95 .
( b ) quasi - energies versus @xmath85 . at the zeros of @xmath96 ,
a pair of quasi - energies degenerate .
( c)-(f ) the time - averaged populations @xmath97 of the floquet states corresponding to @xmath86-@xmath98 .
the other parameters are @xmath36 , @xmath37 , @xmath99 , @xmath100.,title="fig:",scaledwidth=50.0% ] . in the above discussion , the influence of second order coupling ( soc ) , generally thought to be detrimental to cdt , is neglected . in this subsection , we have checked the robustness of our proposed scheme by direct numerical simulations of the schrdinger equation ( [ equ : a ] ) in the presence of soc effects .
as before , the left boundary site is initially occupied and its driving parameters is fixed as @xmath22 . ) with second order coupling .
( a ) the time evolution of @xmath25 with different driving amplitude @xmath10 .
( b ) the minimum value of @xmath25 as a function of @xmath101 .
( c ) quasi - energies versus @xmath85 .
( d)-(f ) the time - averaged populations @xmath97 of the floquet states corresponding to quasi - energies @xmath86-@xmath102 .
the other parameters are @xmath36 , @xmath37 , @xmath99 and @xmath103.,title="fig:",scaledwidth=50.0% ] . the influence of second order coupling on three - site system is plotted in fig .
( [ fig : figure5 ] ) . when soc effects are taken into account , the numerical results in figs .
( [ fig : figure5])(a)-(b ) show that the three - site system displays similar dynamical behaviors ( cdt - ct transitions ) as the case without considering soc .
the numerically computed quasi - energies and floquet states are depicted in figs .
( [ fig : figure5])(c)-(f ) .
correspondingly , a localization - delocalization transition can be seen for the population distribution @xmath84 of the floquet state corresponding to a quasi - energy @xmath76 close to zero , when @xmath29 is increased from zero .
we have also investigated the impact of soc on the dynamics of four - site system , as shown in fig .
( [ fig : figure6 ] ) . fig .
( [ fig : figure6])(a ) shows the minimum value of @xmath25 as a function of the driving amplitude @xmath28 . as @xmath29
is increased from zero , @xmath92 steadily decreases from a value of @xmath104 to zero before it peaks at @xmath105 .
contrary to our expectation , soc facilitates rather than hindering the localization .
this result is somewhat counter - intuitive . from fig .
( [ fig : figure6])(b ) , it can be seen that a pair of the quasi - energies make a series of close approaches to each other as @xmath106 increases .
detailed examination of the close approaches reveals that they are in fact avoided crossings . at the points of close approach
, the tunneling is suppressed and the localization peaks . compared with the zero soc case shown in fig .
( [ fig : figure4 ] ) , the population distributions of floquet states change dramatically for four - site system in the presence of soc effects ; see figs .
( [ fig : figure6])(c)-(f ) .
the modification of the four - site - system s dynamics due to soc , such as enhancement of localization , is the consequence of the floquet states localized at site 1 instead of the quasi - energy degeneracy . ) with second order coupling .
( a ) the minimum value of @xmath25 as a function of @xmath94 . ( b )
quasi - energies versus @xmath85 .
( c)-(f ) the time - averaged populations @xmath97 of the floquet states corresponding to quasi - energies @xmath86-@xmath98 .
the other parameters are @xmath36 , @xmath37 , @xmath99 and @xmath103.,scaledwidth=50.0% ] our analysis above is given for three- and four - site systems , but similar behavior can be obtained for other finite number of sites when the soc effects are not considered .
the quantum dynamics of the driven @xmath8-site systems is investigated by direct integration of the time - dependent schrdinger equation ( [ equ : a ] ) with the particle initially localized at site 1 . in what follows
, we will present the numerical results for @xmath107 and @xmath108 .
dynamics for five sites is depicted in fig .
( [ fig : figure7 ] ) , which shows a similar cdt - ct transition as that of three - site system . like the case of three - site system , this five - site system possesses a dark floquet state with zero quasi - energy and negligible population at all of the even @xmath3th sites , as illustrated in figs .
( [ fig : figure7 ] ) ( c)-(d ) .
the reason for cdt - ct transition in the five - site system lies in the fact that population distribution @xmath84 for the dark floquet state also experiences a localization - delocalization transition ( see fig .
( fig : figure7)(d ) ) . when @xmath109 , the cdt occurs only at the isolated points of parameters as shown in fig .
( fig : figure8)(a ) , where a pair of quasi - energies become degeneracy ( fig .
( [ fig : figure8])(b ) ) .
this is exactly the same as in four - site system .
it is demonstrated that the ct - cdt transition can be induced in six - site system by increasing @xmath29 from zero . , with different values of @xmath10 ; ( b ) the minimum value of @xmath25 as a function of @xmath85 ; ( c ) quasi - energies versus @xmath85 ; ( d ) the time - averaged populations @xmath97 of the dark floquet state corresponding to @xmath110 .
the other parameters are @xmath36 , @xmath37 , @xmath99 , @xmath100.,title="fig:",scaledwidth=50.0% ] . as a function of @xmath85 ; ( b ) quasi - energies versus @xmath94 .
the other parameters are @xmath36 , @xmath37 , @xmath99 , @xmath100.,title="fig:",scaledwidth=50.0% ] .
in conclusion , we have studied how the dynamics of a single quantum particle initially localized in the left boundary site under periodic driving can be controlled by only driving the right boundary site of a lattice array . in studying the number of lattice sites @xmath111 and @xmath108
, we have found that ( i ) a dark floquet state with a zero quasi - energy exists in the three- and five - site systems . by raising the driving amplitude of the external periodic field applied to the right boundary site from zero , we can induce a cdt - ct transition caused by a localization - delocalization transition of the dark floquet state ; ( ii ) no localized floquet state occurs in the four- and six - site systems .
however , we can realize a ct - cdt transition in the same way , thanks to the fact that a pair of quasi - energies degenerate at isolate points of parameters .
we have also revealed that the cdt - ct transition in three - site lattice persists when the soc effects are considered .
long - range interactions ( non - nearest - neighbor couplings ) in the lattice are important and non - negligible in some systems like biomolecules @xcite , polymer chains @xcite , coupled waveguides @xcite , and charge transport in a quantum dot array @xcite .
therefore , our proposed scheme provides a new route to possible application of cdt in such systems .
nevertheless , the ct - cdt transition in four - site system is significantly affected by soc .
in such a system , it is found , there exists a counter - intuitive phenomenon that soc can enhance rather than hinder localization . finally , we discuss the experimental possibility of observing our theoretical predictions . the hamiltonians ( [ equ : h ] ) can be realized in different physical systems , for example using cold atoms or trapped ions in optical lattices @xcite , electron transport in quantum dot chains@xcite , and light propagation in an array of coupled optical waveguides with harmonic modulation of the refractive index of the selected waveguide along the propagation direction @xcite .
recently , several zigzag - type waveguide systems were experimentally realized longhi3,szameit , which provides simpler realization of a one - dimensional lattice with controllable first- and second - order ( that is , next - nearest neighbors ) couplings .
though experimental observation of cdt has been reported in many different physical systems , the proofs presented are not so rigorous as claimed , because cdt occurs only at isolated system parameters . since it is impossible to precisely determine a parameter point through experiment ,
what is claimed is possibly not a genuine cdt which can be testified only after infinite time of evolution , something unrealizable in real experiment . due to the limited time of evolution in experiment , the cdt
as called may be a pseudo - cdt occurring around the surrounding regime adjacent to an isolated parameter , instead of a real one .
hence , observation of cdt , such as the one in our considered four - site and six - site systems , still present a generally unsolved challenge in current experimental setups the work is supported in part by the national fundamental research program of china ( grant no .
2012cb922103 ) , the national science foundation ( nsf ) of china ( grant nos . 11375067 , 11275074 , 11374116 , 11204096 and 11405061 ) , the fundamental research funds for the central universities , hust ( grant no .
2014qn193 ) .
x. luo is supported by the nsf of china under grants 11465009 , 11165009 , 10965001 , the program for new century excellent talents in university of ministry of education of china ( ncet-13 - 0836 ) , the financial support from china scholarship council , and scientific and technological research fund of jiangxi provincial education department under grant no .
x. luo also would like to acknowledge his debt to congjun wu for providing him with an opportunity of visiting ucsd where part of this work is carried out . | we have investigated how the dynamics of a quantum particle initially localized in the left boundary site under periodic driving can be manipulated via control of the right boundary site of a lattice array .
because of the adjustable coupling between the right boundary site and its nearest - neighbor , we can realize either coherent destruction of tunneling to coherent tunneling ( cdt - ct ) transition or coherent tunneling to coherent destruction of tunneling ( ct - cdt ) transition , by driving the right boundary site while keeping the left boundary site driven by a periodically oscillating field with a fixed driving parameter .
in addition , we have also revealed that our proposed cdt - ct transition is robust against the second order coupling ( soc ) between next - nearest - neighbor sites in three - site system , whereas localization can be significantly enhanced by soc in four - site system . |
large - scale spiral density waves commonly arise in a protoplanetary disc due to the gravitational instabilities at the early stage of disc life , when it is sufficiently massive .
it has been known for some time that the density waves can act to transport angular momentum outwards , allowing mass to accrete onto the protostar , and could well be the primary transport mechanism provided thus by disc self - gravity during the earliest stages of star formation ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . the other perturbation type in discs that has received much attention , partly in connection with planetesimal formation ,
are vortices .
they can arise in both self - gravitating and non - self - gravitating discs by several means .
a localized radial structure in the disc can trigger the rossby wave instability , which in turn results in the formation of vortices ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
vortices can also be generated as a result of other hydrodynamic instabilities , such as the baroclinic instability @xcite or the kelvin - helmholtz instability of seed vorticity strips @xcite .
although the dynamics and evolution of vortices in non - self - gravitating discs has been under active investigation for a while , since the first idea of their relevance to planet formation by @xcite , systematic studies of vortex dynamics in self - gravitating discs are quite recent @xcite .
it was commonly thought that spiral density waves are the only perturbation type present in self - gravitating discs , and the existence of other modes also participating in the overall disc dynamics had been ignored in many studies of self - gravitating discs .
developments in the theory of non - self - gravitating discs , however , showed that vortices can be ( linearly ) coupled with , and excite , spiral density waves due to the disc s differential rotation @xcite .
this effect was demonstrated to be even more efficient in the presence of self - gravity @xcite .
moreover , a linear analysis by @xcite showed that in fact vortical mode is also subject to the influence of self - gravity and can exhibit gravitational instability with amplification factors comparable to that of density waves .
this clearly implies that vortices are as important as density waves in forming a complete dynamical picture of self - gravitating discs .
subsequent nonlinear dynamics and evolution of vortices varies greatly in non - self - gravitating and self - gravitating systems . in non - self - gravitating discs , small - scale ( anticyclonic ) vortices , soon after formation , smoothly merge into each other and increase in size ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
when the vortices reach a few scale height , compressibility becomes important , causing them to emit spiral density waves - due to the above - mentioned mode coupling phenomenon - and , consequently , slowly decay on the time - scale of several hundred orbital periods @xcite . in spite of this shock dissipation ,
anticyclonic vortices can be considered long - lived , coherent structures in discs .
however , they can be subject to radial migration @xcite and/or to the elliptical instability @xcite , the typical time - scale of which is larger than vortex turnover / orbital time , as well as to destruction by the magnetorotational instability in magnetized regions @xcite .
the vortex merging is resisted under the influence of disc self - gravity and smaller size vortices are favoured @xcite . in some sense , self - gravity acts to oppose the inverse cascade of spectral energy to larger scales , as occurs in the non - self - gravitating case , and , instead , scatters it to smaller scales . the local shearing sheet simulations of radially unstructured ( i.e. , in the absence of baroclinic or rossby wave instabilities ) discs by @xcite demonstrate that vortices in a quasi - steady gravitoturbulence are transient , short - lived structures undergoing repeating cycles of formation , growth to sizes comparable to the local jeans length , and eventual shearing and destruction due to the effects of self - gravity and keplerian shear .
this process lasts a few orbital periods , and results in a very different , compared to the non - self - gravitating case , evolutionary picture , with many small , less organized vortices ( eddies ) in various stages of evolution , rather than the relatively larger scale well organized vortices gradually growing via slow mergers . on the other hand , in global simulations by @xcite discs
are radially structured and , although they remain laminar ( without gravitoturbulence ) , develop vortices via rossby wave instability at vortensity minimum . due to self - gravity ,
these vortices have a smaller azimuthal extent ( i.e. , higher azimuthal wavenumbers ) compared with similar ones without self - gravity . what is noteworthy , however
, vortices in both local and global simulations share a common feature they produce over - densities at their centres coinciding with the minima of toomre s parameter @xmath0 .
these over - densities are imposed on the density variations due to shocks of density waves emitted by these vortices . in massive discs ,
the gravitational instabilities have been shown to cause the disc to fragment , possibly leading to the direct formation of gas giant planets @xcite .
although a massive protoplanetary disc is thought to be present at early times in the star formation process , it is not clear that such discs can cool fast enough due to radiative transfer to directly form giant planets via this mechanism @xcite .
however , even if the effective cooling times present within discs are too long to allow giant planets to form via fragmentation , as mentioned above , protoplanetary discs are very likely self - gravitating in their early stages @xcite , sometimes even at the t tauri stage @xcite . in this case , indirect means of planet formation accumulation of dust particles in gaseous density structures that arise as a result of the gravitational instability
could be at work .
as outlined above , such structures can be either due to density waves or to smaller scale vortices . in low mass discs , where self - gravity is negligible
, vortices have been shown to have a significant influence on the local evolution of dust particles ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in these simulations ,
not including self - gravity , dust grains with a wide range of sizes were shown to be trapped within the centres of the vortex structures with the strongest concentration occurring for particles with friction / stopping times comparable to the local orbital period . in other words ,
a smooth , sufficiently long - lived vortex is indeed able to effectively trap dust particles in its core , possibly accelerating planetesimal formation . to date
, however , the evolution of dust particles in self - gravitating discs has been primarily studied in a large scale setting , concentrating on the effect spiral density waves , in the gaseous component of the disc , have on dust particles .
@xcite used global simulations to show that the presence of spiral density waves in the disc can have a dramatic effect on the particle layer . in @xcite , the dust particles are assumed to interact with the disc solely via an aerodynamic drag force .
the particles , especially those with stopping times comparable to the local orbital period , are shown to become trapped within the density wave structures that form in the disc due to self - gravity .
this causes the local density of solid material within the density waves to rise by an order of magnitude or more compared to the average .
the gas pressure gradient changes from positive to negative across the density wave structure , creating sub - keplerian velocities on one side of the wave , and super - keplerian on the other . as a result
, the drag force causes dust grains to drift toward the density / pressure maxima at the crest of density waves .
@xcite generalized these findings by including the gravitational effects of the solid particles themselves , which allow these particle enhancements to become even more pronounced , resulting in gravitationally bound clumps , or planetesimals , forming in the disc .
more recently , these results were expanded on in @xcite using higher resolution two - dimensional ( 2d ) local shearing sheet simulations of self - gravitating gaseous discs , treated as in related local studies @xcite , with added numerical super - particles to model dust - particle dynamics .
these studies found that the spiral density waves are very efficient at trapping relatively small , from several centimeter up to meter - sized particles , whose stopping times are comparable to the local orbital period .
very tightly bound clumps of particles can form in the crests of these spiral waves due to the strong gravitational attraction between particles at such high concentrations .
these processes can result in a substantial population of large , planetesimal - sized solid objects being rapidly formed in the early stages of disc evolution , paving the way for terrestrial and gas giant planet formation via the traditional core - accretion model ( e.g. , * ? ? ? * ) .
dust particles can also become trapped by pressure maxima associated with ( anticyclonic ) vortices and , as mentioned above , this process has been extensively studied in non - self - gravitating discs . the role of vortices in trapping dust in self - gravitating discs is less understood and needs further investigation , because of the different character of vortex evolution with self - gravity .
particle concentration by vortices , as well as vortex dynamics itself in the presence of self - gravity , was first addressed by @xcite via global disc simulations . however , in their set - up vortices are generated by the rossby wave instability in an otherwise laminar disc and have a long - lived regular nature - as in the non - self - gravitating case - though smaller azimuthal size due to the effect of self - gravity . here , we consider a different situation a self - gravitating disc in a quasi - steady gravitoturbulent state , without continuous driving by global baroclinic or rossby wave instabilities , and analyse the behaviour of embedded dust particles using the shearing sheet approximation .
we would like to note that on shorter ( comparable to the disc scale height or less ) length - scales just such a gravitoturbulent state , rather than a regular / coherent one , is more common in self - gravitating discs ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . as shown by @xcite , this state consists of transient short - lived small - scale vortices producing over - dense regions overlaid on density wave structures .
in other words , we extend this study by including dust particles with a goal of understanding the role these vortical structures play in dust evolution .
dust dynamics in the presence of only density waves in the shearing sheet has been explored in our previous papers @xcite , where larger domain sizes were used . here , we take a smaller size domain that enables us to zoom in on smaller scale vortical structures and examine their dust trapping capability against the backdrop of density variations due to the density waves ( shocks ) .
this will help us to establish the role of vortices in planet formation process in self - gravitating discs .
the paper is organized as follows . in section 2 ,
we outline the physical set - up and initial conditions adopted in our simulations . in section 3 , we discuss the evolution of the dust particles in the the vortex structures in the gas .
finally , summary and discussions are given in section 4 .
to investigate the dynamics of solid particles embedded in a self - gravitating protoplanetary disc , we solve the 2d local shearing sheet equations for gas on a fixed grid , including disc self - gravity , together with the equations of motion of solid particles coupled to the gas through aerodynamic drag force .
we also include the self - gravity of particles necessary to examine their collapse properties . as a main numerical tool
, we employ the .
it is a sixth order spatial and third order temporal finite difference code ( see @xcite for full details ) .
the treats solids as numerical super - particles @xcite . in the shearing sheet approximation , disc dynamics is studied in the local cartesian coordinate frame centred at some fiducial radius , @xmath1 , from the central object , and rotating with the disc s angular frequency , @xmath2 , at this radius . in this frame with the @xmath3-and @xmath4axis pointing , respectively , in the radial and azimuthal directions , the disc s differential rotation manifests itself as an azimuthal parallel flow characterized by a linear shear , @xmath5 , of background velocity along the @xmath6axis , @xmath7 .
the equilibrium surface densities of gas , @xmath8 , and particles , @xmath9 , are spatially uniform . since the disc is cool , and therefore thin , the aspect ratio is small , @xmath10 , where @xmath11 is the disc scale height , and @xmath12 is the gas sound speed .
the local shearing sheet model is based on the expansion of the basic 2d hydrodynamic equations of motion to the lowest order in this small parameter , assuming that the disc is also razor thin ( see e.g. , * ? ? ?
our simulation domain spans the region @xmath13 , @xmath14 .
as is customary , we adopt the standard shear - periodic boundary conditions well approbated in shearing sheet studies of discs ( e.g. , * ? ? ?
* ; * ? ? ?
namely , for any variable @xmath15 , including azimuthal velocity with background flow subtracted , we have @xmath16 @xmath17 the shear parameter is @xmath18 for the keplerian rotation profile adopted in this paper . in this local model ,
the continuity equation for the vertically integrated gas density @xmath19 is @xmath20 where @xmath21 is the gas velocity relative to the background keplerian shear flow @xmath22 . due to the high - order numerical scheme of the
it also includes a diffusion term , @xmath23 , to ensure numerical stability and capture shocks , @xmath24 here the quantity @xmath25 is the shock diffusion coefficient defined as @xmath26 \rangle(\delta x)^2 \label{shock}\ ] ] where @xmath27 is a constant , characterizing the strength of shock diffusion as outlined in appendix b of @xcite , and @xmath28 is the grid cell size .
the equation of motion for the gas relative to the unperturbed keplerian flow takes the form @xmath29 where @xmath30 is the vertically integrated pressure , @xmath31 is the surface density of particles , @xmath32 is the gravitational potential produced jointly by the gas and particle surface densities ( see equation 6 ) .
the left hand side of equation ( 2 ) describes the advection by the velocity field , * u * , itself and by the mean keplerian flow . on the right hand side ,
the first term is the pressure force , the second and third terms represent the coriolis force and the effect of shear , respectively , and the fourth term describes the back - reaction exerted on the gas by the dust particles due to aerodynamic drag force ( see e.g. , * ? ? ?
* ; * ? ? ?
this force is proportional to @xmath31 and to the difference between the velocity of particles @xmath33 and the gas velocity @xmath34 and is inversely proportional to the stopping , or friction time , @xmath35 , of particles .
the fifth term represents the force due to self - gravity of the system .
finally , the code includes an explicit viscosity term , @xmath36 , @xmath37,\end{aligned}\ ] ] which contains both shear viscosity and a bulk viscosity for resolving shocks . here * s * is the traceless rate - of - strain tensor @xmath38 and @xmath39 is the shock viscosity coefficient analogous to the shock diffusion coefficient @xmath25 defined above , but with @xmath27 replaced by @xmath40 .
the uses entropy , @xmath41 , as its main thermodynamic variable , rather than internal energy , @xmath42 .
the equation for entropy evolution is @xmath43 where the first term on the right hand side is the viscous heating term and the second term is an explicit cooling .
following @xcite , we assume the effective cooling time @xmath44 to be constant throughout the simulation domain and take its value @xmath45 , which is sufficiently large for the disc to avoid fragmentation and settle into a quasi - steady self - regulated state .
the final term on the right hand side , @xmath46 , is a shock dissipation term analogous to that outlined for the density .
the dust particles are treated as a large number of numerical super - particles @xcite with positions @xmath47 on the grid and velocities @xmath48 relative to the unperturbed keplerian rotation velocity , @xmath49 , of particles in the local cartesian frame .
these are evolved according to @xmath50 @xmath51 the first two terms on the right hand side of equation ( 5 ) represent the coriolis force and the non - inertial force due to shear .
the third term is the force exerted on the particles due to the common gravitational potential @xmath32 .
the fourth term describes the drag force exerted by the gas on the particles which arises from the velocity difference between the two . unlike the gas
, the particles do not feel the pressure force .
in the code , the drag force on the particles from the gas is calculated by interpolating the gas velocity field to the position of the particle , using the second order spline interpolation outlined in appendix a of @xcite .
the back - reaction on the gas from particles in equation ( 2 ) is calculated by the scheme outlined in @xcite .
the gravitational potential in the dynamical equations ( 2 ) and ( 5 ) is calculated by inverting poisson equation for it , which contains on the right hand side the gas plus particle surface densities in a razor thin disc ( e.g. , * ? ? ?
* ) @xmath52 using the fast fourier transform ( fft ) method outlined in the supplementary material of @xcite .
note that the perturbed gas , @xmath53 , and particle , @xmath54 , surface densities enter equation ( 6 ) , since only the gravitational potential associated with the perturbed motion ( and hence density perturbation ) of both the gaseous and solid components determine gravity force in equations ( 2 ) and ( 5 ) . here ,
the surface density is fourier transformed from the @xmath55-plane to the @xmath56-plane without the intermediate co - ordinate transformation performed by @xcite .
for this purpose , a standard fft method has been adapted to allow for the fact that the radial wavenumber @xmath57 of each spatial fourier harmonic depends on time as @xmath58 in order to satisfy the shearing sheet boundary conditions ( see also * ? ? ?
we normalise the quantities by setting @xmath59 . the time and velocity units are @xmath60 = \omega^{-1}$ ] and @xmath61 = c_{s0}$ ] , resulting in the orbital period , @xmath62 .
the unit of length is the scale height , @xmath63 = h = c_{s0}/\omega$ ] .
the initial toomre parameter @xmath64 is equal to 1 throughout the domain .
this sets the gravitational constant @xmath65 = @xmath66 .
the surface density of gas is initially set to unity everywhere in the sheet .
the simulation domain is a square with dimensions @xmath67 and is divided into a grid of @xmath68 cells .
this choice of units sets the domain length @xmath69 and the corresponding grid cell size @xmath70 .
the gas velocity field is initially perturbed by some small random fluctuations with the rms amplitude @xmath71 .
we take the viscosity and diffusion coefficients to be @xmath72 and
@xmath73 , this ensures numerical stability across the shock fronts , without washing them out .
we use @xmath74 particles , split evenly between five friction times , @xmath75\omega^{-1}$ ] .
the dust particles are initially randomly placed throughout the domain with zero velocity and do not interact with the gas or other dust particles for the first 10 orbital periods . at this point ,
once the gas has settled into a quasi - steady state , which is independent of its initial conditions , the drag forces between the dust particles and gas are introduced .
the system is then evolved for a further 30 orbits , before the particles self - gravity is introduced . the initial value for the dust - to - gas surface density ratio , @xmath76 , is taken to be 0.01 for each particle species .
the evolution of the gaseous component of the disc is in good agreement with that observed in previous analogous studies of self - gravitating discs in the shearing sheet with an imposed constant cooling time .
the initial perturbations grow and develop into nonlinear fluctuations in velocity , surface density and potential .
shocks then form , which proceed to heat the gas , while the cooling acts to reduce the internal energy and entropy of the gas .
density structures develop , which are sheared out by differential rotation .
once the system has completed several orbits , the heating generated by the shocks is balanced by the cooling term , and the system settles into a quasi - steady , self - regulated state . in this gravitoturbulent state ,
the thermal , kinetic and gravitational energies of the disc are on average constant over time .
the saturated value of the shear stress parameter @xmath77 is proportional to the inverse cooling time , while the amplitude of the density fluctuations to the square root of the inverse cooling time @xcite .
figure [ gas_dens ] shows the surface density of the gas after 40 orbits in the fully developed quasi - steady gravitoturbulence . here ,
the large - scale shocks due to density waves are visible along with the smaller scale structures associated with vortices .
the dynamics of these vortices in the gravitoturbulent state was investigated by @xcite , which we recap below .
the vortical structures in the gas are characterized by the potential vorticity ( pv ) , @xmath78 during the burst phase , initial small - scale positive and negative pv regions are strongly sheared into strips , but negative pv ( anticyclonic ) regions start to wrap up into vortex - like structures due to the nonlinear kelvin - helmholtz instability @xcite .
the positive pv regions remain sheared into strips , showing no signs of vortex formation during the entire course of evolution .
only these anticyclonic regions , having negative pv , are able to survive in the flow by taking the form of vortices , though they are not as regular / coherent as those occurring in non - self - gravitating discs ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these smaller scale , irregular anticyclonic vortices , or eddies , characterized by negative pv , which have developed in the quasi - steady state are clearly seen in the pv field in fig .
[ vor_field ] .
they induce spatial variations ( structures ) with a similar form in the gas surface density due to compressibility and self - gravity , as clearly seen in fig .
[ gas_dens ] .
one can say that in the presence of self - gravity , the pv field makes its `` imprint '' in the density field .
this is in contrast to the case with multiple nonlinearly interacting vortices in the non - self - gravitating shearing sheet , where only density waves ( shocks ) shed by them are seen in the gas density field @xcite . comparing figs .
[ gas_dens ] and [ vor_field ] , we notice that those vortices with smaller , by absolute value , pv give rise to higher over - densities than those with larger , by absolute value , pv ( for example , small white region in the surface density map near the lower left corner ) .
these over - densities at some locations are even larger than the density variations near the shock fronts
. + + in this quasi - steady gravitoturbulent state , the vortices are transient / unsteady structures undergoing recurring phases of formation , growth to sizes comparable to a local jeans scale and eventual shearing and destruction due to the combined effects of self - gravity ( gravitational instability ) and background keplerian shear .
each phase typically lasts about two orbital periods or less . as a result , in self - gravitating discs ,
the overall dynamical picture of vortex evolution is irregular consisting of many transient vortices at different evolutionary stages and , therefore , with various sizes up to the local jeans scale .
by contrast , in the non - self - gravitating case , long - lived vortex structures persist for hundreds of orbits via merging of smaller vortices into larger ones until eventually their size reaches the disc scale height .
it should be noted that the disc cooling , which we have described using a simple constant cooling time prescription , in reality is due to radiative losses from the disc and depends on its physical properties ( e.g. , * ? ? ?
* ; * ? ? ?
obviously , cooling affects disc thermodynamics , which , in turn , through the baroclinic source , determines vortex ( potential vorticity ) dynamics .
however , in the presence of self - gravity the situation is somewhat different . in the case of a constant cooling time , despite being a simple approximation , the disc settles into the quasi - steady gravitoturbulent state characterized by a local balance between heating and cooling , which is qualitatively similar to that with a more realistic radiative cooling ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
so , it is expected that the evolution of pv in the presence of realistic cooling will also be similar to that with a constant cooling time as long as the disc resides in the quasi - steady state and the global baroclinic or rossby wave instabilities do not intervene in the dynamics . in non - self - gravitating discs ,
regular and long - lived anticyclonic vortices have been shown to be efficient at trapping dust particles in their cores ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in section [ particle_conc ]
we examine the particle trapping capabilities of these unsteady and irregular anticyclonic vortices . once the gas has reached a quasi - steady gravitoturbulent state , the aerodynamic drag force and the corresponding back - reaction terms are introduced into the particle and gas evolution , respectively .
the simulation is then evolved for a further 30 orbits , at which time the particles have basically also settled into a quasi - steady state .
the particle self - gravity is then introduced and the simulation was evolved for another 6 orbits .
after this point , the high particle densities achieved in bound clumps ( see below ) result in the self - gravity term becoming dominant in the particle momentum equation .
consequently , the time step of the simulation becomes very small such that it is impractical to evolve the particles further .
however , inspection of the particle density field up to this point indicates that the particles have been sufficiently evolved to draw some conclusions about their further behaviour . at this time , particle dynamics is practically no longer affected by gas and one must also incorporate particle collisions , which become more significant at high particle concentrations within each clump , to correctly follow the subsequent dynamics and contraction of the latter @xcite .
the surface density of the particles at the end of the run with only the drag force ( i.e. , before introducing particle self - gravity ) is plotted in fig .
[ surf_density ] . as shown in @xcite , the small- and intermediate - sized particles with friction times @xmath79\omega^{-1}$ ]
are efficiently trapped in the density wave structures in the gas , with the particles of stopping times @xmath80 exhibiting the highest concentration .
of particular interest for the present work is the behaviour of the dust particles in the vicinity of smaller scale vortices that form in and around the density wave structures .
the dynamical behaviour of particles in the vicinity of these anticyclonic vortices appears to depend on their stopping time , @xmath35 .
the particles with @xmath81\omega^{-1}$ ] tend to accumulate around the central region of anticyclonic vortices , but do not drift further into their centres , leaving noticeable voids at the centres , which correspond to the local minima of pv ( fig .
[ vor_field ] ) .
these particles in fact trace out the over - dense structures around vortices , which , as a result , appear to be surrounded by a spiral , or ring , of dust particles .
these rings also tend to contain some of the highest concentrations of particles in the domain , though not significantly larger than the concentrations found in the crests of density waves .
thus , the smaller particles with @xmath81\omega^{-1}$ ] preferentially map out the vortical structures . on the other hand ,
the intermediate - sized particles with friction times @xmath82 get trapped into crests of density waves and even more in negative pv regions with relatively low by absolute value pv ( curly areas in fig .
[ vor_field ] ) , which produce higher over - densities in the gas ( fig .
[ gas_dens ] ) .
as opposed to the smaller friction time case , these particles fill the central parts of over - dense regions and do not leave voids . the larger particles with @xmath83\omega^{-1}$ ]
accumulate only in density wave crests ; vortices do not seem to affect their motion , as is clear from the surface density of these particles . this differential concentration of particles in the vortical structures depending on their friction time can be explained as follows . as described above
, these small - scale vortices are transient structures , which undergo cycles of formation , amplification and shearing away over a few orbits @xcite .
the small particles , with friction times less than the orbital time , are tightly coupled to the gas and therefore closely follow the density variations induced by vortices . during short times , these transient vortices are approximately in balance by self - gravity , pressure and rotation , forming under - dense core , coinciding with the minimum of pv , surrounded by over - dense ring . as a result ,
the small particles , rapidly adjusting to the vortical motions , get mostly concentrated in these over - dense rings , leaving the vortex centre practically empty . on the other hand , during orbital time the vortices evolve , their pv reduces by absolute value and the corresponding over - density becomes stronger and no longer has an under - dense core .
the particles with friction times comparable to the orbital time tend to accumulate in these over - densities repeating their shape , though with a little delay .
larger particles are not affected by these transient vortices , because their friction times are longer than the orbital time the characteristic time - scale of vortices .
+ + figure [ sigma_t ] shows the evolution of the maximum surface density ( relative to the mean ) of solids in the simulation domain .
when the drag force between the particles and the gas is introduced after 10 orbits , the maximum surface density rapidly jumps , increasing by a factor of @xmath84 , from a few times to almost 100 times the mean over the course of a few orbits .
then , the particles remain in a quasi - steady state , where they are trapped in vortices and/or density waves , but these gaseous structures get continuously sheared away and reappear again over few orbital times , so the particles leaving a gaseous perturbation are swept up by next one and so on .
this cycle continues until the self - gravity of the particles is introduced after 40 orbits . at this point ,
the particle concentrations with large enough densities that have formed before this time start to undergo rapid gravitational collapse , forming many gravitationally bound clumps of particles , which undergo further mergers .
the largest of these objects exhibit local surface densities thousands of times the mean surface density .
these objects are sufficiently stable to remain bound after the parent density wave and/or vortical perturbation they formed in had been sheared away .
the boundedness of the clumps is established by calculating total , kinetic plus gravitational energies inside an individual clump and , if the latter is negative , the clump is classified as gravitationally bound .
the map of the surface density of the solids at the end of the simulation is shown in fig .
[ surf_density_sg ] , where we clearly see bound clumps of particles formed due to their own self - gravity as white dots .
these clumps are mostly made up of intermediate - sized particles with @xmath85\omega^{-1}$ ] , which tend to concentrate most effectively ( see also fig .
[ surf_density ] ) .
the masses of these clumps are typically of the order of @xmath86 @xcite . at this time
, the gas density and pv fields are shown in figs .
[ gas_dens_sg ] and [ vor_field_sg ] , respectively .
remarkably , majority of these clumps are formed within the vortices .
consistent with the above case before switching on particle self - gravity ( fig .
[ surf_density ] ) , the particles with @xmath87 experience clumping in over - dense rings around vortices , whereas larger particles with @xmath82 in more diffuse vortices with negative and smaller by absolute value pv ( curly areas in fig .
[ vor_field_sg ] ) . as mentioned above , these evolved
vortices tend to give rise to higher over - densities in the gas than the density waves do , as evident from fig .
[ gas_dens_sg ] ( especially the one in the central - lower region ) .
thus , irregular , smaller vortices appear to be more efficient at trapping particles than relatively large scale density waves .
this suggests that vortical perturbations can be as important as density waves in accumulating solids in self - gravitating discs .
in this paper , we investigated the particle trapping properties of vortices and the potential role they can play in the planet formation process at the early stage of protoplanetary disc life , when most of its interior is dense and cool and self - gravity is a dominant agent governing the dynamics and evolution of the disc .
however , mass accretion due to the gravitational instability can raise the temperature and activate the magnetorotational instability in the inner disc ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , but this effect is beyond the scope of the present analysis , which focuses on the effects of the gravitational instability only .
we used a similar simulation set - up as adopted in our previous work @xcite , but took a smaller simulation domain , which allowed us to capture small - scale vortical structures against a backdrop of larger scale density waves .
the large - scale picture of the early stages of planetesimal and planet formation described in the above papers remains unchanged : dust particles accumulate within a quasi - steady gaseous spiral structure produced by a combined effect of disc self - gravity and cooling .
the work performed here gives us deeper insight into how smaller scale processes affect the dynamics and evolution of particles in the quasi - steady gravitoturbulence . in this state , the smaller scale turbulent vortical motions ( eddies ) leave their traces in the evolution of the gas density by producing over - dense regions along with spiral density waves .
particles are accumulated both in crests of density waves and in the anticyclonic irregularly - shaped unsteady vortices .
vortices with small , by absolute value , pv are characterized by higher over - densities than those with larger , by absolute value , pv and consequently the particle concentrations in the former turn out to be accordingly higher , even larger than in the density waves for smaller and intermediate - sized particles ( fig .
[ surf_density ] ) .
when the self - gravity of the solid component of the disc is introduced , the particle over - densities that form within these irregular vortices and wave crests rapidly undergo gravitational collapse , forming bound clumps with local densities hundreds of times larger than the typical gas densities within the sheet .
remarkably , the number of clumps formed in the vortices is somewhat larger than that formed in the wave crests ( fig .
[ surf_density_sg ] ) .
it is noteworthy that the degree of particle concentration in these structures and hence their clumping property depend on their friction time .
we found that small and intermediate - sized dust particles corresponding to friction times comparable to , or less than , the local orbital period of the disc , @xmath88 , become especially tightly trapped .
this suggests that vortical perturbations , which generally are an essential ingredient , together with density waves , in the dynamics of self - gravitating discs can also play an important role in the trapping of dust particles and planetesimal formation at early stages of disc life .
we would like to stress once again that the principal reason for high densities and clumping of particles occurring in this problem is their self - gravity . here
, we have not included particle collisions and one might ask how this effect could modify the results . @xcite investigated the effect of collisions on the particle dynamics in gaseous discs . they found that collisions , which become more important at high particle concentrations , actually tend to promote particle concentration via damping their rms velocities : the maximum densities of particles including collisions turned out to be more than a factor of three higher compared to those without collisions .
the collisions also seem to have no apparent effect on the further clumping / collapse process of these particle concentrations due to their own self - gravity .
the masses of the most massive particle clumps ( planetesimals ) formed are relatively insensitive to the presence or absence of collisions . based on these results
, we anticipate that inclusion of collisions in our problem would not much alter a basic dynamical behavior of particles , although extension of this study with treatment of collisions between numerical super - particles ( for example , such as in * ? ? ?
* ) is needed to support this conclusion . in this study , we have considered the simplified case of a razor - thin ( 2d ) disc with a simple cooling law in order to gain first insight into the effects of vortices on dust particles in the presence of self - gravity . obviously , for a fuller understanding , three - dimensional ( 3d ) treatment of vortices , with embedded particles , in self - gravitating discs is necessary , which , as far as we are aware , has not been done yet .
the dynamics of 3d vortices in non - self - gravitating discs has been investigated in a number of studies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in the 3d case , in contrast to the 2d one , they are subject to the elliptical instability , which eventually destroys vortices
however , stratification and baroclinic / rossby wave driving can counteract the elliptical instability , saving vortices from destruction .
a self - gravitating 3d disc is clearly stratified in the vertical direction .
on the other hand , self - gravity does not favour long - lived regular vortices .
but in the 3d case , the effect of self - gravity is somewhat reduced compared to that in the 2d case ( e.g. , * ? ? ?
* ; * ? ? ?
thus , it remains to be seen in future numerical studies how the combination of the above factors shapes the dynamics of 3d vortices in self - gravitating discs .
when analyzing behaviour of particles in such 3d vortices in self - gravitating discs , one must also take into account the basic effects in the particle dynamics being at work in 3d sedimentation due to vertical gravity and stir up and clumping of the dust layer due to the kelvin - helmholtz and streaming instabilities , respectively ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the work presented here expands on the picture of the early stages of planetesimal formation built up by @xcite , which still presents an attractive method for the rapid creation of a large reservoir of planetesimals , along with several massive objects of mass @xmath89 .
these objects are then likely to continue to grow via the traditional core accretion process ( e.g. , * ? ? ?
* ) . further work , which allows the replacement of large , gravitationally bound accumulations of particles with ` sink ' particles , interacting with the disc and with each other , could expand the scope of simulations such as those presented above .
this would allow for the subsequent evolution of these planetesimal - sized objects and could proved further insight into how quickly the core accretion process can work in self - gravitating discs .
this work made use of the facilities of hector the uks national high - performance computing service , which is provided by uoe hpcx ltd at the university of edinburgh , cray inc and nag ltd , and funded by the office of science and technology through epsrcs high end computing programme .
grm acknowledges financial support from the rustaveli national science foundation ( georgia ) grant pg/52/07 .
wkmr acknowledges support from stfc grant st / m001229/1 .
we thank an anonymous reviewer for constructive comments that improved the presentation of our work .
99 andrews s. m. , williams j. p. , 2007 ,
apj , 671 , 1800 ataiee s. , dullemond c. p. , kley w. , regaly z. , meheut h. , 2014 , a&a , 572 , a61 bai x .-
n . , stone j. , 2010 , apj , 722 , 1437 barge p. , sommeria j. , 1995 , a&a , 295 , l1 barranco j. , marcus p. , 2005
, apj , 623 , 1157 bodo g. , tevzadze a. , chagelishvili g. , mignone a. , rossi p. , ferrari a. , 2007 , a&a , 475 , 51 bodo g. , chagelishvili g. , murante g. , tevzadze a. , rossi p. , ferrari a. , 2005 , a&a , 437 , 9 boley a. c. , mejia a. , durisen r. , cai k. , pickett m. , dalessio p. , 2006 ,
apj , 651 , 517 boss a. p. , 1998 ,
nature , 393 , 141 brandenburg a. , 2003 , in advances in nonlinear dynamos , ed .
a. ferriz - mas & m. nunez ( london : taylor & francis ) , 269 clarke c.j .
, 2009 , mnras , 396 , 1066 cossins p. , lodato g. , clarke c. , 2009 , mnras , 396 , 1157 durisen r. h. , boss a. p. , mayer l. , nelson a. f. , quinn t. , rice , w. k. m. , protostars and planets v , b. reipurth , d. jewitt , and k. keil ( eds . ) , university of arizona press , tucson , 951 , pp . , 2007 .
, p.607 - 622 eisner j. , hillenbrand l. , carpenter j. , wolf s. , 2005 , apj , 635 , 396 forgan d. , rice w. k. m. , cossins p. , lodato g. , 2011 , mnras , 410 , 994 fu w. , li h. , lubow s. , li s. , liang e. , 2014 , apj , 795 , 39 gammie c. f. , 2001 , apj , 553 , 174 gibbons p. g. , mamatsashvili g. r. , rice w. k. m. , 2014 , mnras , 442 , 361 gibbons p. g. , rice w. k. m. , mamatsashvili g. r. , 2012 , mnras , 426 , 1444 godon p. , livio m. , 2000 , apj , 537 , 396 godon p. , livio m. , 1999 , apj , 523 , 350 goldreich p. , lynden - bell d. , 1965 , mnras , 130 , 97 guan x. , gammie c. , simon j. , johnson b. , 2009 , apj , 694 , 1010 heinemann t. , papaloizou j. c. b. , 2009 , mnras , 397 , 52 johansen a. , youdin a. , lithwick y. , 2012 , a&a , 537 , a125 johansen a. , klahr h. , henning th . , 2011 , a&a , 529 , a62 johansen a. , youdin a. , mac low m .-
, 2009 , a&a , 704 , l75 johansen a. , oishi j. s. , mac low m. , klahr h. , henning th . , youdin a. , 2007 , nature , 448 , 1022 johansen a. , youdin a. , 2007 , apj , 662 , 627 johansen a. , klahr h. , henning t. , 2006 , apj , 643 , 1219 johansen a. , andersen a. c. , brandenburg a. , 2004 , a&a , 417 , 361 johnson b.m . , gammie c. f. , 2005 , apj , 635 , 149 johnson b. m. , gammie c. f. , 2003 , apj , 597 , 131 klahr h. h. , bodenheimer p. , 2006 ,
apj , 639 , 432 klahr h. , bodenheimer p. , 2003 ,
apj , 582 , 869 lesur g. , papaloizou j. c. b. , 2010 , mnras , 513 , 60 lesur g. , papaloizou j. c. b. , 2009 , a&a , 498 , 1 li h. , colgate s. a. , wendroff b. , liska r. , 2001 , apj , 551 , 874 li h. , finn j. m. , lovelace r. v. e. , colgate s. a. , 2000 , apj , 533 , 1023 lin d. n. c. , pringle j. e. , 1990 , apj , 358 , 515 lin d. n. c. , pringle j. e. , 1987 , mnras , 225 , 607 lin m .- k . , 2012a , apj , 754 , 21 lin m .- k . , 2012b , mnras , 426 , 3211 lithwick y. , 2007 , apj , 670 , 789 lovelace r. v. e. , hohlfeld r. g. , 2013 , mnras , 429 , 529 lovelace r. v. e. , li h. , colgate s. a. , nelson a. f. , 1999 , apj , 513 , 805 lyra w. , lin m .- k .
, 2013 , apj , 775 , 17 lyra w. , klahr h. , 2011 , a&a , 527 , 138 lyra w. , johansen a. , zsom a. , klahr h. , piskunov n. 2009 , a&a , 497 , 869 lyra w. , johansen a. , khlar h. , piskunov n. , 2008 , a&a , 479 , 883 mamatsashvili g. r. , rice w. k. m. , 2010 , mnras , 406 , 2050 mamatsashvili g. r. , rice w. k. m. , 2009 , mnras , 394 , 2153 mamatsashvili g. r. , chagelishvili g. d. , 2007 , mnras , 381 , 809 meheut h. , keppens r. , casse f. , benz w. , 2012a , a&a , 542 , 9 meheut h. , meliani z. , varniere p. , benz w. , 2012b , a&a , 545 , 134 meheut h. , casse f. , varniere p. , tagger m. , 2010 , a&a , 516 , 31 paardekooper s .- j . , 2012 , mnras , 421 , 3286 paardekooper s .- j . , lesur g. , papaloizou j. c. b. , 2010 , apj , 725 , 146 petersen m. r. , julien k. , stewart g. r. , 2007 , apj , 658 , 1236 pollack j. b. , hubickyj o. , bodenheimer p. , lissauer j. , podolak m. , greenzweig y. , 1996 , icarus , 124 , 62 raettig n. , lyra w. , klahr h. , 2013 , apj , 765 , 115 rafikov r. r. , 2005 , apj , 621 , l69 regaly zs . , juhasz a. , sandor zs . , dullemond c. p. , 2012 ,
mnras , 419 , 1701 rice w. k. m. , armitage p. j. , mamatsashvili g. r. , lodato g. , clarke c. j. , 2011 , mnras , 418 , 1356 rice w. k. m. , mayo j. h. , armitage p. j. , 2010 , mnras , 402 , 1740 rice w. k. m. , armitage p. j. , 2009 , mnras , 396 , 2228 rice w. k. m. , lodato g. , pringle j. e. , armitage p. j. , bonnell i. a. , 2006 , mnras , 372 , l9 rice w. k. m. , lodato g. , pringle j. e. , armitage p. j. , bonnell i. a. , 2004 , mnras , 355 , 543 rice w. k. m. , armitage p. j. , bate m. r. , bonnell i. a. , 2003 , mnras , 338 , 227 richard s. , barge p. , le dizes s. , 2013 , a&a , 559 , 30 shen y. , stone j. m. , gardiner t. a. , 2006 , apj , 653 , 513 umurhan o. m. , regev o. , 2004 , a&a , 427 , 855 youdin a. , johansen a. , 2007 , apj , 662 , 613 youdin a. , shu f. , 2002 , apj , 580 , 494 zhu z. , stone j. m. , rafikov r. r. , bai x .-
n . , 2014 , apj , 785 , 122 zhu z. , hartmann l. , gammie c. , book l. , simon j. , engelhard e. , 2010 , apj , 713 , 1134 zhu z. , hartmann l. , gammie c. , 2009 , apj , 694 , 1045
to investigate the effects of resolution on the dynamics of vortices in the presence of self - gravity , we ran simulations for a domain with half size , @xmath90 ( in units of @xmath91 , initial @xmath92 ) at three different resolutions @xmath93 , including the same number of particles and starting with the same initial conditions as in the paper .
this is equivalent to running the fiducial model with @xmath94 considered in this work at doubled resolutions @xmath95 and @xmath96 , respectively , since the corresponding grid cell sizes , @xmath97 would be the same in both cases .
so far as numerical convergence is concerned , grid cell size is a central length - scale , since measuring various physical lengths of the system against it allows us to find out whether a simulation is resolved .
therefore , due to the same grid cell sizes , a resolution study we do for the half - size model also establishes whether vortices in the fiducial model are sufficiently resolved . in the quasi - steady state , the pv fields from all these three runs at the same time are shown in fig .
a1 . these fields , being at different resolutions ,
display essentially identical structures , indicating that they are resolved . to put this conclusion on more quantitative grounds , as in @xcite
, we consider the autocorrelation function for the fluctuating part of pv , @xmath98 , given by @xmath99 where the angle brackets denote average over the whole domain .
figure a2 shows the autocorrelation functions corresponding to the pv fields in fig .
a1 . they do appear similar and their cores have nearly an elliptical shape with minor and major principal axis . following the method of @xcite , we calculated related correlation lengths along these axes by @xmath100 where @xmath101 is the distance from @xmath102 along the principal axes given by the unit vectors @xmath103 and @xmath104 .
these correlation lengths can be used as a measure for the typical size of the pv structures .
the smallest correlation lengths along the minor axis are @xmath105 , for the lowest , intermediate and highest resolution runs , respectively
. the ratios of the minimum correlation length to the grid cell size at these resolutions are , respectively , @xmath106 . this ratio increases with resolution which serves as an indication for convergence and , as a matter of fact , implies that the vortices ( eddies ) can be considered to be sufficiently resolved in the fiducial model which has @xmath107 . in our previous papers @xcite ,
the domain size is @xmath108 and resolution @xmath109 for the same other parameters , giving @xmath110 and hence @xmath111 .
thus , although this resolution is quite suitable to resolve larger scale density waves in those studies , it is still coarse to detect and examine vortical structures at such larger domain sizes in contrast to that at four times smaller domain size used in the present paper . | the mechanism through which meter - sized boulders grow to km - sized planetesimals in protoplanetary discs is a subject of active research , since it is critical for planet formation . to avoid spiralling into the protostar due to aerodynamic drag
, objects must rapidly grow from cm - sized pebbles , which are tightly coupled to the gas , to large boulders of 1 - 100 m in diameter .
it is already well known that over - densities in the gaseous component of the disc provide potential sites for the collection of solids , and that significant density structures in the gaseous component of the disc ( e.g. , spiral density waves ) can trap solids efficiently enough for the solid component of the disc to undergo further gravitational collapse due to their own self - gravity . in this work ,
we employ the pencil code to conduct local shearing sheet simulations of massive self - gravitating protoplanetary discs , to study the effect of anticyclonic transient vortices , or eddies , on the evolution of solids in these discs .
we find that these types of structures are extremely efficient at concentrating small and intermediate - sized dust particles with friction times comparable to , or less than , the local orbital period of the disc .
this can lead to significant over - densities in the solid component of the disc , with density enhancements comparable to , and even higher , than those within spiral density waves ; increasing the rate of gravitational collapse of solids into bound structures .
[ firstpage ] accretion , accretion discs - gravitation - hydrodynamics - instabilities - planets and satellites : formation |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.