article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the experimental verification of many scalar - isoscalar resonances , @xmath6 , up to the energy of @xmath7 gev @xcite reenforced the idea that the scalar glueball lies among them .
however , the remaining open question is which of these scalars - isoscalars is predominantly the glueball.in this paper we address this issue by studying the vacuum phenomenology of scalar - isoscalar states within the framework of the so called extended linear sigma model ( elsm ) @xcite .
our numerical calculations of the effective lagrangian ( [ lagrangian ] ) are explicitly done in the case of three quark flavours . for @xmath8 two scalar - isoscalar @xmath5 states
exist , the nonstrange @xmath9 and the strange @xmath10 mesons .
the scalar glueball @xmath11 is implemented in the elsm from the very beginning as the fluctuation of the dilaton field .
we describe @xmath12 as well as the generation of the scale anomaly of the pure yang - mills lagrangian at the quantum level by the usual logarithmic dilaton lagrangian , first introduced in ref . @xcite . due to the same quantum numbers a mixing between the scalar - isoscalar quarkonia and glueball in our model takes place.in this study we focus on the determination of the mixing matrix by using the mass eigenvalues of the scalar - isoscalar states and identify them with the physical masses of the resonances @xmath3 , @xmath4 and @xmath2 @xcite.the outline of this proceeding is as follows : in sec .
2 we present the lagrangian of the extended linear sigma model with a scalar glueball . in sec .
3 we discuss our results and finally in sec .
4 we give a summary as well as an outlook for further research .
in order to study the vacuum phenomenology of the scalar - isoscalar resonances @xmath13 and @xmath2 we use the elsm including a scalar glueball @xcite .
its compact representation for an optional number of flavours is as follows : @xmath14\nonumber\\ & -\lambda_{1}(\text{\textrm{tr}}% [ \phi^{\dag}\phi])^{2}+c_{1}(\text{\textrm{det}}\phi^{\dag}-\text{\textrm{det } } \phi)^{2}+\mathrm{tr}[h(\phi^{\dag}+\phi)]\nonumber\\ & + \text{\textrm{tr}}\left [ \left ( \frac{m_{1}^{2}}{2}\left ( \frac { g}{g_{0}}\right ) ^{2}+\delta\right ) \left(l^{\mu2}+r^{\mu2}\right ) \right ] \nonumber\\ & -\frac{1}{4}\text{\textrm{tr}}\left [ l^{\mu\nu2}+r^{\mu\nu2}\right ] + \frac{h_{1}}{2}\text{\textrm{tr}% } [ \phi^{\dag}\phi]\text{\textrm{tr}}[l_{\mu}l^{\mu}+r_{\mu}r^{\mu}]\nonumber\\ & + h_{2}\text{\textrm{tr}}[\phi^{\dag}l_{\mu}l^{\mu}\phi+\phi r_{\mu}r^{\mu } \phi^{\dag}]+2h_{3}\text{\textrm{tr}}[\phi r_{\mu}\phi^{\dag}l^{\mu}]\text { + ... , } \label{lagrangian}\end{aligned}\ ] ] where @xmath15 and @xmath16the dilaton lagrangian ( [ ldil ] ) describes the scale anomaly of the pure yang - mills sector where the logarithmic term with the energy scale @xmath17 breaks the dilatation symmetry explicitly .
the field @xmath12 is the scalar dilaton and after performing the shift @xmath18 a particle with the mass @xmath19 arises which we interpret as the scalar glueball.our model underlies the following symmetries and their several breakings : i ) the global chiral symmetry , @xmath20 , which is broken explicitly due to the quantum effects and the nonvanishing quark masses as well as spontaneously . as a consequence of breaking of the latter one a nonvanishing quark condensate , @xmath21 , arises .
ii ) a crucial symmetry in our model is the already mentioned dilatation symmetry or scale invariance , @xmath22 , which is realized at the classical level of the yang - mills sector of qcd , but explicitly broken by the loop corrections . this
is known as scale or trace anomaly , respectively and leads to the nonvanishing gluon condensate , @xmath23 . taking into account the dilatation symmetry we can constrain the number of possible terms in our model
this implies that in the chiral limit with the exception of the logarithmic term in eq .
( [ ldil ] ) and the term generating the @xmath24 anomaly , all parameters entering in the model are dimensionless .
( note , using the chiral anomaly it was also possible to couple the pseudoscalar glueball to the model , for details and results see refs.@xcite).iii ) our model is also in agreement with discrete symmetries of qcd e.g. parity @xmath25 and charge conjugation @xmath26.the multiplet of the ordinary scalar and pseudoscalar mesons in the case of @xmath8 , containing the two out of three bare scalar - isoscalar states @xmath27 and @xmath28 , reads @xcite : @xmath29{ccc}% \frac{(\sigma_{n}+a_{0}^{0})+i(\eta_{n}+\pi^{0})}{\sqrt{2 } } & a_{0}^{+}% + i\pi^{+ } & k_{0}^{\star+}+ik^{+}\\ a_{0}^{-}+i\pi^{- } & \frac{(\sigma_{n}-a_{0}^{0})+i(\eta_{n}-\pi^{0})}% { \sqrt{2 } } & k_{0}^{\star0}+ik^{0}\\ k_{0}^{\star-}+ik^{- } & \bar{k}_{0}^{\star0}+i\bar{k}^{0 } & \sigma_{s}+i\eta_{s}% \end{array } \right ) .
\label{phimatex}%\ ] ] the explicit form of the left - handed and the right - handed ( axial)vector mesons multiplets , @xmath30 and @xmath31 can be found in ref . @xcite .
a reasonable approach to study the scalar - isoscalar sector of the elsm with a scalar glueball is to use the values of the global fit performed in ref .
@xcite and only to determine the additional three free parameters which enter in this sector .
two of them arise directly from the dilaton lagrangian ( [ ldil ] ) , namely the bare mass of the scalar glueball @xmath32 and the energy scale @xmath17 .
the third one is @xmath33 which couples the ordinary ( pseudo)scalar mesons and was in ref .
@xcite only determined as a part of combination.in order to obtain the numerical values of the corresponding parameters we use the bare masses of the scalar - isoscalar states and the experimental masses of the resonances @xmath1 and @xmath2 .
accordingly we consider the potential of the corresponding states in the matrix representation : @xmath34{ccc}% m_{\sigma_{n}}^{2 } & z_{g\sigma_{n } } & z_{\sigma_{s}\sigma_{n}}\\ z_{g\sigma_{n } } & m_{g}^{2 } & z_{g\sigma_{s}}\\ z_{\sigma_{s}\sigma_{n } } & z_{g\sigma_{s } } & m_{\sigma_{s}}^{2}% \end{array } \right ) \left ( \begin{array } [ c]{c}% \sigma_{n}\\ g\\ \sigma_{s}% \end{array } \right)\label{pot}.%\ ] ] the corresponding bare mass equations are : @xmath35 @xmath36 @xmath37 after diagonalization of the bare mass matrix ( see eq .
( [ pot ] ) ) we obtain the matrix containing the physical masses of the scalar - isoscalar states , @xmath38 , where @xmath39 is the mixing matrix which transforms the bare states into the physical ones and vice versa : @xmath40{c}% f_{0}(1370)\\ f_{0}(1710)\\ f_{0}(1500)% \end{array } \right)=b \left ( \begin{array } [ c]{c}% \sigma_{n}\equiv\left(\bar{u}u+\bar{d}d\right ) /\sqrt{2}\\
g\equiv gg\\ \sigma_{s}\equiv\bar{s}s\\ \end{array } \right ) \text { .}\label{trafo}%\ ] ] the numerical values of the free parameters are presented in table [ table1 ] .
they have been obtained by requiring that the three following masses of the scalar - isoscalar states @xcite hold : @xmath41 mev , @xmath42 mev and @xmath43 mev .
[ c]|c|c|c|parameter & value + @xmath33 & @xmath44 + @xmath32 & @xmath45 [ mev ] + @xmath17 & @xmath46 [ mev ] + the numerical solution of the mixing matrix reads : @xmath47{ccc}% 0.92 & -0.39 & 0.05\\ -0.33 & -0.83 & -0.45\\ -0.22 & -0.40 & 0.89% \end{array } \right ) \text { .}\label{bnum}\ ] ] it turns out that the resonances @xmath3 and @xmath4 are predominantly nonstrange and strange @xmath5 states , and that the resonance @xmath2 is predominantly a scalar glueball . for other results of the mixing matrix @xmath39 using different theoretical models , we refer to refs .
@xcite and refs . therein .
the solution in eq .
[ bnum ] relies only on calculations of masses of the scalar - isoscalar states .
in order to give final statements regarding the composition of the @xmath48 resonances evaluation of the corresponding decays is essential .
nevertheless , it is astonishing that by increasing the number of flavours in our model , @xmath49 , the phenomenology of the @xmath48 resonances seems to change ( see ref .
we have evaluated the numerical values of the mixing matrix @xmath39 by using the masses of the scalar - isoscalar states between @xmath50 and @xmath7 gev within a chiral model , the so called extended linear sigma model with a scalar glueball @xcite .
we have found that the resonances @xmath3 and @xmath4 are predominantly nonstrange and strange @xmath5 states , and the resonance @xmath2 is predominantly a scalar glueball.study in progress are calculations of the decay processes appropriate to the solution presented in this work and further search for a solution with the assignment where @xmath4 is predominantly a glueball and @xmath2 predominantly a strange @xmath5 state @xcite .
the author thanks f. giacosa , d. parganlija and d. h. rischke for cooperation and useful discussions and acknowledges support from h - qm and hgs - hire .
s. gallas , f. giacosa and d. h. rischke , phys .
d * 82 * , 014004 ( 2010 ) [ arxiv:0907.5084 [ hep - ph ] ] ; d. parganlija , f. giacosa and d. h. rischke , phys .
d * 82 * , 054024 ( 2010 ) [ arxiv:1003.4934 [ hep - ph ] ] .
c. rosenzweig , a. salomone and j. schechter , phys .
d * 24 * , 2545 ( 1981 ) ; a. salomone , j. schechter and t. tudron , phys .
d * 23 * , 1143 ( 1981 ) ; c. rosenzweig , a. salomone and j. schechter , nucl .
b * 206 * , 12 ( 1982 ) [ erratum - ibid .
b * 207 * , 546 ( 1982 ) ] ; h. gomm and j. schechter , phys .
b * 158 * , 449 ( 1985 ) ; r. gomm , p. jain , r. johnson and j. schechter , phys .
d * 33 * , 801 ( 1986 ) . c. amsler and f. e. close , phys .
d * 53*,295 ( 1996 ) [ arxiv : hep - ph/9507326 ] ; w. j. lee and d. weingarten , phys . rev .
d * 61 * , 014015 ( 2000 ) [ arxiv : hep - lat/9910008 ] ; f. e. close and a. kirk , eur .
j. c * 21 * , 531 ( 2001 ) [ arxiv : hep - ph/0103173 ] ; f. giacosa , t. gutsche , v. e. lyubovitskij and a. faessler , phys .
d * 72 * , 094006 ( 2005 ) [ arxiv : hep - ph/0509247 ] ; f. giacosa , t. gutsche and a. faessler , phys .
c * 71 * , 025202 ( 2005 ) [ arxiv : hep - ph/0408085 ] ; h. y. cheng , c. k. chua and k. f. liu , phys .
d * 74 * , 094005 ( 2006 ) [ arxiv : hep - ph/0607206 ] ; v. mathieu , n. kochelev and v. vento , int .
j. mod .
e * 18 * , 1 ( 2009 ) [ arxiv:0810.4453 [ hep - ph ] ] .
w. i. eshraim , s. janowski , f. giacosa and d. h. rischke , phys .
d * 87 * , 054036 ( 2013 ) [ arxiv:1208.6474 [ hep - ph ] ] ; w. i. eshraim , s. janowski , a. peters , k. neuschwander and f. giacosa , acta phys .
supp . * 5 * , 1101 ( 2012 ) [ arxiv:1209.3976 [ hep - ph ] ] ; w. i. eshraim and s. janowski , j. phys .
426 * , 012018 ( 2013 ) [ arxiv:1211.7323 [ hep - ph ] ] ; w. i. eshraim and s. janowski , arxiv:1301.3345 [ hep - ph ] . | using the @xmath0 extended linear sigma model with the ordinary ( pseudo)scalar and ( axial)vector mesons as well as a scalar glueball , we study the vacuum phenomenology of the scalar - isoscalar resonances @xmath1 and @xmath2 .
we present here a solution , based only on the masses and not yet on decays , in which the resonances @xmath3 and @xmath4 are predominantly nonstrange and strange @xmath5 states respectively , and the resonance @xmath2 is predominantly a scalar glueball . |
dissipation , while generally seen as detrimental for quantum computers , can nevertheless be a useful resource if suitably engineered . appropriately chosen system - bath couplings can result in a non - equilibrium dynamics where initial states converge towards some dynamical steady state .
this kind of ` quantum reservoir engineering ' has been proposed as a viable approach towards the experimental preparation of interesting many - body states @xcite .
remarkable examples include the preparation of pure states with long - range order in bose - einstein - condensates @xcite , photonic arrays @xcite , as well as topologically ordered states @xcite .
more generally , verstraete et al .
@xcite argued that , at least in principle , an arbitrary quantum computation can be realized by dissipation .
the corresponding process is similar to feynman s clock construction and has the final state of the computation as its steady state .
subsequent work @xcite following this program proposed dissipative gadgets allowing to realize different dissipative dynamics during subsequent time - intervals . here we examine the dissipative preparation of specific topologically ordered states .
while not realizing a fully dissipative computation , this basic primitive could act as a building block in a hybrid scheme where initial states for quantum computation are prepared by thermalization and subsequent computations are performed in the usual framework of topological quantum computation .
we ask whether dissipative processes can be used to realize an _ encoder _ , i.e. , a map which turns states on individual physical qubits into encoded ( many - qubit ) states .
in contrast , previous work only considered the dissipative preparation of _ some _ ground state without guarantees on the logical information .
it is worth mentioning that various unitary encoders are known for topologically ordered systems . for the toric code
@xcite on an @xmath0 lattice , dennis et al .
@xcite gave a unitary circuit with two - local controlled - not ( cnot ) gates of depth @xmath3 acting as an encoder .
bravyi et al .
@xcite showed for any evolution under a local time - dependent hamiltonian acting as an encoder requires time at least @xmath4 . in turn , brown et al .
@xcite present a duality transformation from a 2d cluster state to a topologically ordered state which can be interpreted as a geometrically local quantum circuit of matching depth . dropping the requirement of locality , aguado and vidal @xcite gave an encoder with depth @xmath5 with geometrically non - local two - qubit gates .
the dissipative encoder considered here may be realized by designing suitable system - environment interactions .
this is to be contrasted with schemes involving error correction , which generally consist of syndrome extraction by measurement and associated correction operations . for the toric code , an encoding procedure of this form
was given @xcite .
it involves active error correction operations similar to the minimal matching technique used in @xcite . to define the notion of an encoder in more detail ,
consider a quantum error - correcting code @xmath6 encoding @xmath7 logical qubits into @xmath8 physical qubits .
informally , an encoder is a map taking any state @xmath9 into its encoded version @xmath10 .
( this notion implicitly assumes a choice of basis of @xmath11 ) .
since we are interested in a physical system of @xmath8 qubits , we will require the encoder to convert a ` simple ' unencoded initial state into an encoded logical state .
that is , we ask that for a fixed subset @xmath12 of qubits and a fixed product state @xmath13 on the remaining qubits , the encoder maps @xmath14 we are interested in encoders realized by evolution under a markovian master equation @xmath15 here the liouvillian has lindblad form @xmath16 with lindblad operators @xmath17 acting locally on a constant number of qubits .
we ask whether the completely positive trace - preserving map ( cptpm ) @xmath18 generated by @xmath19 is an ( approximate ) encoder for sufficiently large times @xmath20 , i.e. , whether it can realize the map ( [ eq : encodermap ] ) .
our result is the following .
[ thm : encoder ]
let @xmath21 be the toric code consisting of @xmath22 qubits on the edges of a periodic @xmath0-lattice .
consider the partition of the qubits into disjoint sets @xmath23 shown in figure [ fig : encoderqubits ] .
that is , * @xmath24 are two neighboring qubits having a common adjacent vertex @xmath25 and plaquette @xmath26 .
* @xmath27 and @xmath28 are located along a vertical respectively horizontal line passing through @xmath29 . *
@xmath30 and @xmath31 are located along a horizontal respectively vertical line passing through @xmath32 .
* @xmath33 are the remaining @xmath34 qubits .
there is a geometrically local liouvillian @xmath19 ( with @xmath35-qubit lindblad operators ) such that the following holds : for any state @xmath36 on @xmath37 and @xmath38 , and for any @xmath39 , we have @xmath40 whenever @xmath41 in this expression , we use the trace norm @xmath42 for hermitian operators .
we will give a detailed description of the relevant liouvillian in section [ sec : schemedescription ] .
, and their appropriate initialization for encoding : qubits @xmath43 are initialized in the state to be encoded .
each qubit in @xmath44 is in the state @xmath45 , while qubits in @xmath46 are in the state @xmath47 .
the state of the remaining qubits @xmath33 can be an arbitrary ( mixed ) state . in the figure
, we also illustrate the support of possible realizations for logical operators @xmath48 and @xmath49 associated with the first and second logical qubit , respectively .
our encoding procedure requires choosing one such realization for each logical pauli generator such that they overlap only at the initial unencoded qubits .
[ fig : encoderqubits ] , width=377 ] the evolution @xmath18 implements a continuous - time version of a local error correction process somewhat analogous to toom s rule : excitations move towards a single plaquette / vertex where they annihilate .
an analogous ground - state preparation scheme for more general frustration - free hamiltonians was discussed in @xcite .
however , guarantees about encoded information appear to be harder to obtain in their generic scheme . furthermore because their construction requires the injectivity @xcite property of the associated projected entangled pair state ( peps ) description ( and hence blocking of sites ) , the resulting locality of the lindblad operators will be slightly worse .
in contrast , our scheme directly exploits the stabilizer structure of the underlying code , resulting in a comparatively simple liouvillian .
indeed , our construction is optimal in terms of locality , i.e. , the number of particles involved in each lindblad term @xcite .
related locality constraints for pure steady states were derived in @xcite . in our setup
, the initial product state determines in a transparent way which code state is prepared .
our work goes beyond the mere characterization of steady states by adding two key ingredients : the consideration of logical observables ( which are preserved ) and an analysis of the convergence towards the ground space of the toric code .
the bound @xmath1 on the convergence time established by theorem [ thm : encoder ] improves on the @xmath50 upper bound predicted for the analogous construction in @xcite , without a guarantee on the logical information .
in fact , it is tight : there are initial states @xmath51 which thermalize slowly , i.e. , @xmath52 is far away from the code space @xmath11 for any time @xmath53 . in separate work @xcite
, we provide a general no - go theorem in this direction : dissipative state preparation of topologically ordered states requires at least a linear amount of time in @xmath54 if the liouvillian is local . combined with theorem [ thm : encoder ] , this implies that the construction presented here is optimal in terms of preparation time among the entire class of local liouvillians . in summary , our work shows that dissipative processes can be used to implement an encoder for the toric code .
intriguingly , this encoder is more time - efficient than the best known unitary circuit .
we stress , however , that both types of encoders need to be supplemented with additional mechanisms in the presence of noise , especially if the encoded information is further processed . as discussed in @xcite , local liouvillians such as the one considered here are not suitable for the preservation of encoded information .
we first describe a generic liouvillian associated with a stabilizer code @xmath11 with stabilizer generators @xmath55
. we will subsequently specialize this to the toric code .
let @xmath56 be the projections onto the @xmath57 eigenspaces of the stabilizer @xmath58 .
the code space @xmath11 is the ground space of the hamiltonian @xmath59 ( the global energy shift is introduced for convenience ) .
our goal is to implement a local error - correction strategy by a dissipative evolution .
concretely , we associate a unitary pauli correction operator @xmath60 to each stabilizer generator @xmath58 such that @xmath60 and @xmath58 anticommute , i.e. , @xmath61 .
we define the cptpm @xmath62 note that by definition , @xmath63 lowers the energy of the term @xmath64 in the hamiltonian , that is , @xmath65 while after application of @xmath63 , the stabilizer constraint defined by @xmath58 is satisfied , its application may create a non - trivial syndrome ( excitation ) for a neighboring stabilizer @xmath66 , @xmath67 . by design ( i.e. , the choice of correction operators for the toric code discussed below ) , repeated application of all @xmath68 eventually removes all excitations , resulting in a state supported on the code space @xmath11
. it will be convenient to introduce the averaged cptpm @xmath69 which randomly chooses a syndrome and applies the associated correction map . by the correspondence discussed in @xcite
, each local cptpm @xmath63 defines a local liouvillian @xmath70 .
we are interested in the evolution under @xmath71 , which , by definition , is a sum of constant - strength local terms .
observe that @xmath72 .
we now consider the toric code with qubits on the edges of an @xmath0 ( periodic ) square lattice .
we separate the hamiltonian into @xmath73 where the former includes all plaquette- and the latter includes all vertex terms . here , we have taken @xmath74 and @xmath75 to correspond to plaquette and vertex stabilizers respectively .
@xmath76{plaquettestabilizer}}}\qquad\qquad s_v = x^{\otimes 4}={\raisebox{-0.75em}{\includegraphics[width=2.5em]{vertexstabilizer}}},\ ] ] in other words , @xmath74 is the product of @xmath77-type pauli operators acting on the four qubits on the edges bounding plaquette @xmath78 and @xmath75 is the product of four @xmath79-type pauli operators acting on the qubits incident to vertex @xmath80 . to define the associated pauli correction operators @xmath81 and @xmath82 , let us partition the set of vertices and plaquettes into @xmath83 here @xmath26 is a single plaquette , @xmath84 consists of the @xmath85 plaquettes lying on a fundamental cycle of the torus ( which we refer to the ` equator ' , running ` horizontally ' or ` east - west ' along the torus ) on which @xmath26 is located , whereas @xmath86 are the remaining @xmath87 plaquettes .
the vertex @xmath25 as well as the sets @xmath88 and @xmath89 are defined similarly on the dual lattice , see figure [ fig : liouvilleandef ] .
the subscript associated with these sets indicates the direction of movement of an excitation under application of the local correction map .
for example , a magnetic excitations ( caused by an @xmath79-error ) on a plaquette @xmath90 will move to the neighboring plaquette to the north of @xmath91 under application of the correction map @xmath92 .
that is , we define @xmath93{singleplaqueteanyonmoverup}}}\qquad\textrm { for } \qquad p\in{\mathcal{s}}^{(p)}_\uparrow\qquad c_p={\raisebox{-0.5em}{\includegraphics[width=1.5em]{singleplaqueteanyonmoverright}}}\qquad\textrm { for } \qquad p\in{\mathcal{s}}^{(p)}_\rightarrow\nonumber\\ c_v&={\raisebox{-0.5em}{\includegraphics[width=1.5em]{singlevertexanyonmoverdown}}}\qquad\textrm { for } \qquad v\in{\mathcal{s}}^{(v)}_\downarrow\qquad c_v={\raisebox{-0.5em}{\includegraphics[width=1.5em]{singlevertexanyonmoverleft}}}\qquad\textrm { for } \qquad v\in{\mathcal{s}}^{(v)}_\leftarrow\ .\nonumber\end{aligned}\ ] ] we will set @xmath94 corresponding to a trivial correction operation ( this is simply done for convenience ) .
the liouvillian @xmath19 is then defined as in section [ sec : generalstabilizercodes ] .
the proof of theorem [ thm : encoder ] relies on two basic statements .
the first one concerns the logical information encoded in the state : for suitable initial states , this information is preserved along the evolution . recall that the toric code has two encoded qubits .
let @xmath95 be the two - qubit logical pauli operators defined by figure [ fig : encoderqubits ] , and let us set @xmath96 where @xmath97 is the identity operator .
these operators play a crucial role in the following statement .
[ lem : logicalinfopres ] for any two - qubit pauli operator @xmath98 , @xmath99 , let @xmath100 , @xmath101 be its corresponding logical counterpart .
consider an initial state of the form @xmath102 where the state @xmath36 is arbitrary on @xmath103 .
then @xmath104 let @xmath98 be arbitrary .
observe first that the rhs of ( [ eq : toproveeqtimeevolvelog ] ) is equal to @xmath105 because of the definition of @xmath106 ( cf .
[ fig : liouvilleandef ] ) and the form of @xmath107 , i.e. , the fact that the qubits in @xmath108 and @xmath109 are @xmath110-eigenstates of the single - qubit @xmath79 and @xmath77- operators , respectively . in other words , it suffices to show the expectation value @xmath111 is time - independent for initial states @xmath107 of the form ( [ eq : specialinitialstate ] ) .
we claim that an even stronger statement holds : we have @xmath112 i.e. , any observables of the form @xmath106 does not evolve in the heisenberg picture . to prove ( [ eq : constantnonevolution ] ) , observe that the one - parameter family of unital maps @xmath113 is generated by the adjoint @xmath114 of the liouvillian , hence ( [ eq : constantnonevolution ] ) is equivalent to @xmath115 since @xmath116 is a sum of liouvillians @xmath70 associated with stabilizers @xmath117 , it suffices to verify that for every @xmath117 , we have @xmath118 or @xmath119 because @xmath106 is a logical operator , it commutes with each stabilizer , and hence also the projections @xmath120 .
equality , ( [ eq : pjp ] ) is in fact implied by @xmath121=0 $ ] , which we can verify by a case - by - case analysis for single - qubit ( logical ) operators @xmath122 ( the general case then follows since a product @xmath123 commutes with @xmath124 if each factor @xmath125 does ) .
consider for example the case where @xmath126 .
we then have to consider two cases : a. the correction @xmath127 is associated with a plaquette . in this case
@xmath92 is a pauli-@xmath79 ( or the identity ) .
however , inspection of the support of @xmath128 ( see fig . [
fig : encoderqubits ] ) and the location of the correction operators ( see fig . [ it : plaquettesfig ] ) reveals that no correction operation @xmath92 acts on the support of @xmath128 , hence @xmath129=0 $ ] as desired .
b. the correction @xmath130 is associated with a vertex . in this case
@xmath131 is a pauli-@xmath77 ( or the identity ) , hence @xmath132=0 $ ] holds trivially .
the second fundamental statement is about convergence to the ground space .
[ lem : convergencetimebound ] let @xmath133 be the projection onto the code space of the toric code , and let @xmath134 the projection onto the orthogonal complement .
let @xmath51 be an arbitrary initial state .
then we have @xmath135 we have not optimized the constants in this bound as we are interested in the overall ( linear ) scaling in @xmath54 .
the proof strategy is different from the arguments in @xcite and may be of independent interest . consider the function @xmath136 defined as follows : @xmath137 in other words , the function expresses the axial distance to @xmath138 and @xmath139 , respectively . for a vertex @xmath140 ( plaquette @xmath91 ) , the quantity @xmath141 ( @xmath142 ) is the number of vertices ( plaquettes ) traversed by an electric ( magnetic ) excitation on the primary ( dual ) lattice before reaching @xmath25 ( @xmath26 ) along a path @xmath143 ( or @xmath144 ) .
for the special vertex @xmath138 and the plaquette @xmath139 , we set @xmath145 for convenience ( alternatively , we could omit the discussion of the corresponding trivial correction operations altogether as the stabilizer generators are linearly dependent ) .
the key property of the function @xmath146 is the fact that it is compatible with the way excitations are propagated under the correction operations .
more precisely , for each stabilizer @xmath58 , let @xmath147 be the set of correction operations that anticommute with it , i.e. , @xmath148 then @xmath146 has the property that @xmath149 namely , whenever an excitation is created by @xmath150 at @xmath58 , one can be certain that a higher valued excitation has been removed at @xmath151 .
we will argue that for any @xmath152 , we have @xmath153 where @xmath154 is the maximal number of stabilizers a single correction operator can excite .
note that eq .
( [ eq : upperboundonorthogprojection ] ) holds for arbitrary stabilizer codes with the corresponding definitions . specializing to the toric code , where @xmath155 , we obtain @xmath156 ( implying the claim ) by choosing @xmath157 and observing that @xmath158 to prove ( [ eq : upperboundonorthogprojection ] ) , consider the observable @xmath159 clearly @xmath160 for any @xmath152 , hence it suffices to show that the expectation value @xmath161 is upper bounded by the rhs . of ( [ eq : upperboundonorthogprojection ] ) , that is @xmath162 consider the heisenberg evolution of the projection operators @xmath163 .
since pauli operators either commute or anticommute , a straightforward calculation gives @xmath164=0$\\ ( i-{{\bf p}}_j^-){{\bf p}}_k^--{{\bf p}}_j^-{{\bf p}}_k^- & if $ \{c_k , s_j\}=0$.\\}\nonumber \ ] ] in particular , the expectation values behave classically under the designed liouvillian , that is ( writing @xmath165 for brevity ) , @xmath166 according to ( [ ineq : anyonoccupations ] ) , we have @xmath167 where we used property ( [ eq : monotonicitypropertyf ] ) on the function @xmath146 .
according to the definition of @xmath168 and @xmath169 , this implies @xmath170 i.e. , the expectation value decays exponentially .
the claim ( [ eq : ddecay ] ) then follows from the fact that @xmath171 for all @xmath172 since these are projections , hence @xmath173 . with lemma [ lem : logicalinfopres ] and lemma [ lem : convergencetimebound ] , we are ready to prove our main result .
consider an initial state @xmath107 of the form ( [ eq : specialinitialstate ] ) and assume that @xmath174 . by lemma [ lem : convergencetimebound ]
, we have @xmath175 . with the gentle measurement lemma ( see e.g. , ( * ? ?
* lemma 9.4.1 ) ) , this implies @xmath176 where @xmath177 is defined as @xmath178 .
observe that the state @xmath177 is supported entirely on the code space @xmath11 .
note that for any two states @xmath179 we have @xmath180 and therefore @xmath181 for any normalized operator @xmath182 . hence ( [ eq : tclrhoevolve ] ) implies that for any normalized logical operator @xmath106 , we have @xmath183 let @xmath184 be the target encoded state , i.e. , the state satisfying ( cf .
( [ eq : logicalopeval ] ) and ( [ eq : constantnonevolution ] ) ) @xmath185 combining this with ( [ eq : overlpexp ] ) shows that @xmath186 and @xmath177 have approximately the same expectation values for logical operators . since both states are supported on the ground space @xmath11 for which we have access to a full algebra of logical observables ( linear combinations of logical pauli observables ) , one may choose is the normalized local observable which optimally distinguishes between the unencoded states @xmath187 and @xmath188 , then @xmath189 is the required logical operator . ]
a normalized logical operator @xmath106 such that @xmath190 which achieves the maximization defining @xmath191-norm distance @xmath192= \big|{\mathop{\mathrm{tr}}\nolimits}(\overline{p}({|\overline{\psi}\rangle\langle\overline{\psi}|}-\bar{\rho}'_t))\big| \leq 2\sqrt{\epsilon}\ .\label{eq : closetpsi}\end{aligned}\ ] ] combining ( [ eq : closetpsi ] ) with ( [ eq : tclrhoevolve ] ) and using the triangle inequality , we conclude that @xmath193 the claim of theorem [ thm : encoder ] follows immediately from this statement .
we have presented a quasi - local time - independent liouvillian which generates a dissipative encoder : it encodes two physical qubits into the ground space of the toric code .
its key features are translation - invariance in the bulk , optimal locality of the lindblad operators , as well as optimal , i.e. , linear convergence ( encoding ) time of the resulting dissipative encoder as a function of the lattice size .
this illustrates the power of engineered dissipation for state preparation in a non - trivial setting with degeneracy .
since the liouvillian is constituted of quasi - local cptpm associated with each stabilizer , one may interpret the resulting evolution @xmath18 as the repeated application of the ` average ' correction map ( cf . ( [ eq : averagebtsuperop ] ) ) .
this viewpoint immediately results in a preparation algorithm for surface code states .
the latter inherits the properties of the dissipative evolution : it is translation - invariant up to boundary conditions , composed of feedback - free quasi - local maps , and requires only a minimal number of iterations to converge .
these features make it an attractive candidate for potential experimental realizations .
the most pressing open question is to incorporate noise in the form of control and initialization imperfections into the design and analysis of such an encoder .
the holy grail is to engineer a reasonable , possibly time - dependent liouvillian master equation such that after a specified initial period physical qubits are encoded with only a constant , lattice - size independent probability of suffering uncorrectable errors .
expecting further reduction seems impossible without changing the assumptions due to the possibility of an error occuring on the unencoded qubit .
as it stands , our protocol @xmath54-fold amplifies such an initial error probability as certain errors on the qubits in @xmath194 are propagated into logical errors .
indeed , we suspect that a fully fault - tolerant encoding scheme would require progressively growing the code s physical support while simultaneously introducing dissipative terms responsible for self - correction within this support . as in @xcite
such terms would be responsible for ensuring that the rate of occurence for uncorrectable errors decreases with the code size .
jd and rk gratefully acknowledge funding provided by nserc .
rk would like to thank john preskill for discussions , and the institute for quantum information and matter for their hospitality .
fp would like to thank iman marvian , ignacio cirac and john preskill for helpful discussions and comments .
fp acknowledges funding provided by the institute for quantum information and matter , a nsf physics frontiers center with support of the gordon and betty moore foundation ( grants no . phy-0803371 and phy-1125565 ) . | we consider the problem of preparing specific encoded resource states for the toric code by local , time - independent interactions with a memoryless environment .
we propose a construction of such a dissipative encoder which converts product states to topologically ordered ones while preserving logical information .
the corresponding liouvillian is made up of four - local lindblad operators . for a qubit lattice of size @xmath0 , we show that this process prepares encoded states in time @xmath1 , which is optimal .
this scaling compares favorably with known local unitary encoders for the toric code which take time of order @xmath2 and require active time - dependent control . |
the 2df qso redshift ( 2qz ) survey has established that the qso sample is a useful probe of cosmological models as a tracer of the large scale distribution of mass ( croom , et al .
2001 ; , hoyle , et al .
2002 ) . in general ,
constraints on cosmological parameters from qso sample are not very tight .
however , the cosmological parameters estimated from the qso sample have a unique implication for cosmology ( outram , et al . 2003 ; yamamoto 2003a ) .
for example , the cosmological principle can be tested by comparing with the result from other observations such as galaxy redshift survey and cosmic microwave background anisotropies .
the pioneering work on the qso power spectrum analysis was done by hoyle et al .
( 2002 ) with the 2qz 10000 catalogue .
recently outram et al .
have reported the result of the similar analysis with the final 2qz catalogue containing 22652 qsos ( 2003 ) .
they have shown that the qso power spectrum is consistent with the hubble volume @xmath0cdm simulation . furthermore , by fitting the power spectrum with the @xmath0cdm model within linear theory of density perturbations , they obtained a constraint on the cosmological density parameters . in the modeling of the qso power spectrum in outram et al ( 2003 ) , however , the light - cone effect ( matarrese et al . 1997 ; matsubara , suto & szapdi 1997 ; yamamoto & suto 1999 ) , the geometric distortion ( ballinger , peacock & heavens 1996 ; matsubara & suto 1996 ) and the nonlinear effects ( mo , jing & b@xmath3rner 1997 ; magira , jing & suto 2000 ) are not taken into account .
the neglect of these effects might fail to estimate the correct cosmological parameters . to test this point
, we revisit the power spectrum analysis of the 2qz sample .
we have independently performed the power spectrum analysis of clustering with the complete 2qz sample .
then we fit the 2qz power spectrum with theoretical template incorporating the effects , which are not considered in the work by outram et al .
the methodology in the present paper is almost same as that in the reference ( yamamoto 2002 ) , in which the fitting of the 2df qso power spectrum from the 10000 catalogue was investigated using an analytic approach .
thus the primary purpose of the present paper is to test the robustness of the result by outram et al .
( 2003 ) for independent determination of the power spectrum and for more careful modeling of the theoretical power spectrum , including the nonlinear effects , the geometric distortion and the light - cone effect . on the other hand , calvao et al .
( 2002 ) claimed that the equation of state of the dark energy @xmath2 might be constrained from the 2df qso sample . due to the geometric distortion effect ,
the qso redshift - space power spectrum may depend on @xmath2 even if the original matter power spectrum ( or the transfer function ) does not depend on @xmath2 ( yamamoto 2003b ) .
the strategy in the present paper is not the one proposed by calvao et al .
( 2002 ) , however , we check a constraint on @xmath2 by considering how the estimated density parameters depends on @xmath2 by the fitting of the power spectrum .
the second purpose of this paper is to test the equation of state of the dark energy @xmath2 using the qso power spectrum .
this paper is organized as follows : in section 2 , we describe our power spectrum analysis . in section 3 , our theoretical modeling of the qso power spectrum is explained . in section 4 ,
constraint on the density parameters is discussed by fitting the 2qz power spectrum .
section 5 is devoted to summary and conclusions . throughout this paper
we use the unit in which the light velocity equals @xmath4 .
in our power spectrum analysis , we use the complete sample of the full 2qz survey , which is publicly available @xmath5 .
the 2qz survey covers two area of @xmath6 deg@xmath7 , one in the south galactic cap ( sgc ) and the other in the north galactic cap ( ngc ) , respectively , in the range of redshift less than 3 .
the survey area is defined by the equatorial coordinates from @xmath8 to @xmath9 and @xmath10 in the sgc , and @xmath11 and @xmath12 in the ngc , respectively .
the survey area of the ngc is jagged and we select a simple rectangle area in our power spectrum analysis
. then we use 10713 and 8443 qsos in the sgc and the ngc , respectively , in the range of redshift @xmath13 , , which is slightly different from our choice .
this difference does not alter our result .
] incorporating the hole information publicly available .
we describe the estimator of the power spectrum adopted here .
three dimensional map is constructed by introducing the distance @xmath14 where @xmath15 is the comoving distance of the @xmath0 cold dark matter model with the density parameter @xmath16 .
we denote the density field by @xmath17 and the mean number density by @xmath18 , where @xmath19 with @xmath20 specifying the direction . introducing a random synthetic density field @xmath21 , which has mean number density @xmath22 times that of @xmath17
, we define the fourier coefficient @xmath23 e^{i{\bfk}\cdot\bfs } \label{fouk}\end{aligned}\ ] ] with @xmath24 .
the estimator of the power spectrum is defined @xmath25 where @xmath26 is the volume of a thin shell in the @xmath27-space with the radius @xmath28 . in the case
@xmath29 , the variance of the power spectrum is @xmath30 note that we have not used the optimal weighting scheme by setting the optimal weight factor being constant ( feldman , kaiser and peacock 1994 , tegmark et al .
1998 , yamamoto 2003b ) . this choice does not alter the result of the qso power spectrum analysis because the qso is sparse and @xmath31 . instead of equation ( [ fouk ] )
, the discrete density field can be rephrased as @xmath32 , \label{foukdis}\end{aligned}\ ] ] where @xmath33 and @xmath34 are the position of the @xmath35-th qsos and the @xmath36-th random objects .
figure 1 plots the power spectrum ( filled squares ) , which is obtained by combining results from the sgc and ngc data sets .
fig.1 shows a good agreement with the result ( open squares ) by outram et al .
however , the error bar of our power spectrum is larger than that of 2qz group .
this can originate from the difference of the error estimator adopted , which we describe in more details below .
the solid curve in fig.1 is the theoretical curve of the @xmath0 cold dark matter ( cdm ) model with the cosmological parameter , @xmath37 , @xmath38 , @xmath39 , @xmath40 and @xmath41 , motivated by the wmap result ( spergel et al .
2003 , see also next section ) .
finally in this section , we compare the error estimator of the power spectrum adopted in the present paper with that of the 2qz group .
there are two differences .
first , they used the equation 2.3.2 in paper by feldman et al .
( 1994 ) . in the limit that @xmath42 is a constant , there is the discrepancy of the factor @xmath43 , if @xmath26 , the volume of a thin shell in the @xmath44-space , is same . in our analysis
we used @xmath45 , where @xmath46 is the width of the @xmath28 bin .
our error estimator is consistent with that formulated by tegmark et al .
( 1998 ; see equation 62 in their paper ) .
second , the 2qz group adopted the different method for @xmath26 , estimated by @xmath47 with @xmath48 the number of independent modes in the @xmath28-shell and @xmath49 the volume of one @xmath28-mode ( for details see also hoyle , et al .
2002 ; hoyle 2000 ) .
the author considered a constraint on the density parameters , @xmath50 and @xmath51 , by fitting the 2df qso power spectrum found by hoyle et al .
( 2002 ) , in a previous investigation ( yamamoto 2002 ) .
we here adopt the similar methodology to apply it to the 2qz power spectrum obtained in the previous section .
the important improvement of the present work is the inclusion of nonlinear effects in modeling the theoretical power spectrum and a systematic uncertainty in measuring the redshift in the 2qz survey , as well as the inclusion of an arbitrary equation of state parameter @xmath2 of the dark energy . in this section
we briefly explain the modeling , following the same notation in the previous paper . we restrict ourselves to a spatially flat frw universe and follow the quintessential cold dark matter ( qcdm ) model .
the effective equation of state of the dark energy , @xmath2 , can be a function of redshift , however , we consider the qcdm model with a constant equation of state for simplicity ( wang et al .
2000 ) . in this case
the relation between the comoving distance and the redshift is @xmath52 where @xmath53km / s / mpc is the hubble constant .
wang & steinhardt ( 1998 ) have given a useful approximate formula for the linear growth index in the qcdm model , which we adopt in the present work .
for nonlinear modeling of the mass density perturbation , we adopt the simple fitting formulae for the nonlinear mass perturbation power spectrum presented by ma et al .
( 1999 ) for the qcdm model , combined with the transfer function by eisenstein & hu ( 1998 ) , which is robust even when the baryon fraction is large . to incorporate the nonlinear ( finger - of - god ) effect due to the random motion of objects
, we consider the power spectrum multiplied by the damping factor @xmath54 $ ] , where @xmath55 is the pairwise velocity dispersion and @xmath56 is the comoving wave number parallel to the line - of - sight of an observer in real space .
we adopt the expression assuming an exponential distribution function for the pairwise velocity , and the corresponding damping factor is ( mo , et al . 1997 ; magira , et al .
2000 ) @xmath57={1\over 1 + q_{||}^2\sigma_{\rm p}^2/2}. \label{dsigmap}\end{aligned}\ ] ] the pairwise velocity dispersion is the function of redshift @xmath58 .
we computed @xmath59 by the approximate formula for the mean squire velocity dispersion at the large separation determined through the cosmic energy equation ( mo , et al .
1997 ; magira , et al .
2000 ; suto et al .
2000 ) . on the other hand ,
as described in the paper by outram et al .
( 2003 ) , systematic uncertainty in measurement of the redshift causes an apparent velocity dispersion .
we incorporate the uncertainty in redshift - measurement @xmath60 , based on the following consideration .
we write the density fluctuation field @xmath61 in a fourier expansion form as @xmath62 where @xmath63 represents the error in measuring the position , which is related with @xmath60 by @xmath64
. then an ensemble average of @xmath65 can be written as @xmath66 where we have assumed that @xmath67 and @xmath68 are independent probability variables and @xmath69 .
we further assume that the angular coordinates are well determined and @xmath70 . in this case ,
denoting the wave number of the line - of - sight direction by @xmath71 , we have @xmath72 where we have assumed that @xmath73 and @xmath74 are independent gaussian probability variables with the variance @xmath75 . therefore the damping factor due to the error in redshift - measurement is written @xmath76= \exp\left[-k_{||}^2 \left({ds\over dz}\right)^2 \langle\delta z^2\rangle \right ] , \label{delz}\end{aligned}\ ] ] where @xmath77 is the variance of the error in measuring the redshift . in the present paper
we adopt @xmath78 following croom et al .
figure 2 shows the damping factor @xmath79 $ ] ( solid curves ) in the @xmath0cdm model$ ] denotes the damping factor @xmath54 $ ] with replacing @xmath56 with @xmath80 . ] , as function of the wave number @xmath71 with the redshift fixed as @xmath81 , @xmath82 and @xmath83 for three curves from top to bottom .
the dashed curves show the damping factor @xmath79 $ ] multiplied by @xmath84 $ ] . because we are considering the angular averaged power spectrum , and this figure
does not exactly express the damping factor of our power spectrum .
however , this figure indicates that the damping factor have a substantial effect on the power spectrum shape .
thus the uncertainty in redshift - measurement can be an important factor in the power spectrum analysis .
concerning the modeling of the bias , we assume the scale independent model .
following the previous work ( yamamoto 2002 ) , we consider the model @xmath85 , where @xmath86 and @xmath87 are the constant parameters and @xmath88 is the linear growth rate normalized as @xmath89 .
we determine @xmath86 to minimize @xmath90 , which we define in the next section . in the present paper
, we show the result adopting @xmath91 .
our result slightly depends on @xmath87 , however , plausible alternation of @xmath87 does not alter our conclusion qualitatively .
( see also yamamoto 2002 ) .
we consistently use the number density per unit redshift per unit solid angle @xmath92 from the catalogue
. it will be useful to show how the theoretical power spectrum depends on the equation of state @xmath2 of the dark energy .
curves in figure 3 are theoretical power spectra with the various equation of state @xmath93(solid curve ) , @xmath94(long dashed curve ) and @xmath95(dashed curve ) .
the other cosmological parameter is fixed as @xmath37 , @xmath96 , @xmath39 , @xmath40 and @xmath41 .
the difference of the three curves comes from the geometric distortion effect : the comoving distance in real space is given by equation ( [ defrz ] ) , while the power spectrum is measured in redshift space with the radial coordinate @xmath97 defined by equation ( [ defsz ] ) .
therefore the power spectrum in redshift space is distorted compared with that in real space . denoting the wave number in real space by @xmath98 ,
the geometric distortion is described by the scaling of wave numbers @xmath99 , where @xmath100 is the ratio @xmath101 and @xmath102 for the component parallel and perpendicular to the line - of - sight direction , respectively ( for details see e.g. , ballinger et al .
1996 , yamamoto 2002 ; 2003b ) .
for the model with @xmath103 , the factor @xmath100 becomes less than unity .
this shifts the power spectrum from right ( large @xmath28 ) to left ( small @xmath28 ) for the model with @xmath103 , as is clearly shown in fig .
thus @xmath2 can be measured from the redshift power spectrum due to the scaling effect of the geometric distortion .
such a test using the scaling effect is distinguished from the alcock - paczynski test which relies on the anisotropy of clustering in redshift space ( e.g. , ryden 1995 ; alcock & paczynski 1979 ) .
in this section we compare our observational power spectrum from the 2qz catalogue with the theoretical model to constrain cosmological parameters .
we introduce @xmath104 ^ 2\over \delta p(k_i)^2 } , \end{aligned}\ ] ] where @xmath105 is the value of the power spectrum from the theoretical model at the wave number @xmath106 , @xmath107 is the observational data and @xmath108 is the variance of errors , for which we use the @xmath109 data points obtained in fig . 1 . following the analysis by outram et al .
( 2003 ) , we consider the likelihood function @xmath110 .
figure 4 shows contours of the likelihood function with @xmath90 found when different values of the cosmological density parameters , @xmath50 and @xmath111 are used in the theoretical modeling .
each panel corresponds to constraints for the difference model of @xmath112 : the panels ( a ) , ( c ) and ( d ) assume @xmath93 , @xmath113 and @xmath114 , respectively .
the panel ( b ) is same as ( a ) , but switched off the damping factor due to the error in the redshift - measurement by setting @xmath115 in equation ( [ delz ] ) . in fig . 4
, we fix the other parameters @xmath39 , @xmath41 and @xmath40 .
the curves are contours of confidence @xmath116 , @xmath117 and @xmath118 on the plane .
first , the panel ( b ) in fig .
4 shows the effect of the error in the redshift - measurement in modeling the power spectrum .
the peak value of ( a ) is located at @xmath119 , while that of ( b ) is located at @xmath120 .
thus the effect of @xmath60 can be of importance for determining @xmath50 , precisely .
second , fig .
4 demonstrates the constraint on the density parameters depends on the equation of state parameter @xmath2 of the dark energy .
it shows that the preferable value of @xmath50 and @xmath51 increases as @xmath2 becomes large .
figure 5 shows the contour of the likelihood function with @xmath90 found when different values of @xmath50 and @xmath2 are used in the modeling the power spectrum , where we fixed @xmath38 , motivated from the wmap result ( spergel et al .
it is shown that the density parameters in the range @xmath121 is preferable , which is not very sensitive to the value of @xmath2 .
therefore the constraint on @xmath2 is not very tight , and @xmath122 is only excluded at the one ( two ) sigma level .
in summary we have revisited the power spectrum analysis of the complete 2df qso sample , as a complementary test of the work by outram et al .
our analysis has reproduced a power spectrum , which is consistent with the 2qz group .
we have investigated the fitting of the power spectrum including the light - cone effect , the geometric distortion and the nonlinear effects .
it is shown that the qso spatial power spectrum is consistent with the @xmath0cdm model with @xmath123 , which is effectively consistent with the 2qz group
. however , our constraint is weaker than that of the 2qz group .
this will be traced back to the difference of the error estimation of the power spectrum . in the present paper , we have emphasized the importance of the error in measurement of the redshift @xmath60 because it may have influence in estimating the density parameters .
we have also investigated the effect of the equation of state of the dark energy @xmath2 . for models with large value of @xmath103
, the preferable value of the density parameters becomes large .
when we fix the baryon density parameter as that found by the wmap team , the matter density parameter @xmath121 is preferable .
no tight constraint on @xmath2 is obtained , @xmath122 is only excluded at the one ( two ) sigma level . in general qso sample is sparse and the shot noise contribution is substantial .
therefore the constraint from it is not very tight .
however , such a power spectrum analysis will provide more useful constraint on @xmath2 when applied to the sdss luminous red galaxy sample ( yamamoto 2003b ) and next generation redshift survey such as the kaos project ( seo and eisenstein 2003 ) .
the author thanks anonymous referee for useful comments on earlier version of the manuscript , which helped improve it .
this work is supported in part by grant - in - aid for scientific research of japanese ministry of education , culture , sports , science and technology , no.15740155 .
99 alcock , c. , & paczynski , b. 1979 , nature , 281 , 358 ballinger , w. e. , peacock , j. a. , & heavens , a. f. 1996 , mnras , 282 , 877 calvao , m. o. , de mello neto , j. r. t. , & waga , i. 2002 , phys .
88 , 091302 croom , s. m. , shanks , t. , boyle , b. j. , smith , r.j . , miller , l. , loaring , n. , & hoyle , f. 2001 , mnras , 325 , 483 croom , s. m. , smith , r. j. , boyle , b. j. , shanks , t. , miller , l. , outram , p. j. , & loaring , n. s. 2003 , preprint eisenstein , d. j. , & hu , w. 1998 , apj , 496 , 605 feldman , h. a. , kaiser , n. , & peacock , j. a. 1994 , apj , 426 , 23 hoyle , f. , 2000 , phd thesis , univ .
durham hoyle , f. , outram , p.j .
, shanks , t. , croom , s.m . , boyle , b.j .
, loaring , n. , miller , l. , & smith , r.j .
2002 , mnras , 329 , 336 ma , c .- p . , calldwell , r. r. , bode , p. , & wang , l. 1999 , apj , 521 , l1 matarrese , s. , coles , p. , lucchin , f. , & moscardini , l. 1997 , mnras , 286 , 115 matsubara , t. , & suto , y. 1996 , apj , 470 , l1 matsubara , t. , suto , y. , & szapdi , i. 1997 , apj , 491 , l1 magira , h. , jing , y. p. , & suto , y. 2000 , apj , 528 , 30 mo , j. , jing , y. p. , & b@xmath3rner , g. 1997 , mnras , 286 , 979 nishioka , h. , & yamamoto , k. 1999 , apj , 520 , 426 outram , p.j . , hoyle , f. , shanks , t. , croom , s. m. , boyle , b. j. , miller , l. , smith , r. j. , & myers , a. d. 2003 , mnras , 342 , 483 ryden , b. s. 1995 , apj , 452 , 25 seo , h - j , & eisenstein d. j. 2003 , astro - ph/0307460 spergel , d. n. , et al .
2003 , apjs , 148 , 175 suto , y. , magira , h. , jing , y. p. , matsubara , t. , & yamamoto , k. 2000 , prog .
133 , 183 tegmark , m. , hamilton , a. j. s. , strauss , m. a. , vogeley , & m. s. , szalay , a. s. 1998 , apj , 499 , 555 wang , l. , caldwell , r. r. , ostriker , j. p. , & steinhardt , p. j. 2000 , apj , 530 , 17 wang , l. , & steinhardt , p. j. 1998 , apj , 508 , 483 yamamoto , k. , & suto , y. 1999 , apj , 517 , 1 yamamoto , k. 2002 , mnras , 334 , 958 yamamoto , k. 2003a , mnras , 341 , 1199 yamamoto , k. 2003b , apj , 595 , 577 | we revisit the power spectrum analysis of the complete sample of the two degree field ( 2df ) qso redshift ( 2qz ) survey , as a complementary test of the work by outram et al .
( 2003 ) .
a power spectrum consistent with that of the 2qz group is obtained . differently from their approach , fitting of the power spectrum is investigated incorporating the nonlinear effects , the geometric distortion and the light - cone effect .
it is shown that the qso power spectrum is consistent with the @xmath0 cold dark matter ( cdm ) model with the matter density parameter @xmath1 .
our constraint on the density parameter is rather weaker than that of the 2qz group .
we also show that the constraint slightly depends on the equation of state parameter @xmath2 of the dark energy . the constraint on @xmath2 from the qso power spectrum
is demonstrated , though it is not very tight . |
the classical wks sampling theorem has been extended to the case of nonuniform sampling by numerous authors . for detailed information on the theory and its numerous applications ,
we refer to the book @xcite .
most known irregular sampling results deal with paley wiener functions which have @xmath2 restrictions on the real line .
it seems that the best known nonuniform wks sampling results for entire functions in @xmath3spaces were given in @xcite .
there are no multidimensional @xmath3wks sampling theorems with precise truncation error estimates in open literature
. however , explicit truncation error upper bounds in multidimensional wks reconstructions are of great importance in signal and image processing .
alternative reconstruction approaches for irregular sampling problems were developed in @xcite .
however , due to long traditions the wks type reconstructions still play key role in applied signal and image processing .
new numerical methods for quick sinc series summations ( see , for example ,
@xcite ) let more efficient usage of wks formulae than before . on the other hand wks type results are important not only because of signal processing applications .
wks theorems are equivalent to many key results in mathematics , see , e.g. @xcite . therefore they are also valuable for theoretical studies
. it explains why in spite of substantial progress in modern approximation methods wks type reconstructions are still of great importance and vast amount of new refine results are published regularly .
in this paper we use approaches and methods developed in @xcite to investigate multidimensional irregular sampling in @xmath3-spaces .
let @xmath4 be a normed space and assume that the structure of @xmath4 admits the sampling approximation procedure @xmath5 where @xmath6 ; @xmath7 and @xmath8 are the sampling set and the sampling function respectively .
this formula is one of the basic tools in signal processing . in direct numerical implementations
we consider the truncated variant of ( [ 1 ] ) , reads as follows : @xmath9 where the index set @xmath10 is finite by application reasons .
namely , recovering a continuous signal from discrete samples or assessing the information lost in the sampling process are the fundamental problems in sampling and interpolation theory .
the usual procedure is to estimate the truncation error @xmath11 where @xmath12 could stay for different norms on different sides of this relation and @xmath13 denotes the truncation error upper bound .
simple truncation error upper bounds are the main tools in numerical implementations , when they do not contain infinite products and/or unknown function values .
however , the suitably sharp truncation error upper bound enables pointwise , almost sure , uniform _ etc . _ type convergence of the approximating sequence @xmath14 to initial @xmath15 .
the main aim of this paper is for @xmath16 to discuss @xmath17 by getting pointwise upper bounds valid for all @xmath18 on the whole range of the signal domain without strong assumptions on the decay rate of @xmath19 see @xcite .
this kind of upper bounds we call _
universal_. the most frequent rearrangement of in the literature is of the form @xmath20 @xmath21 being a conjugated hlder pair , i.e. @xmath22 . to make the approximant @xmath14 more precise let us assume @xmath23 , that is , that the sampling index set @xmath24 depends on the location of time parameter @xmath18 with respect to the behaviour of @xmath25 in estimating @xmath26 see @xcite .
that means @xmath27 is _ a fortiori _ time shifted and possesses time adapted sampling size . to obtain a class of truncation error upper bounds when
the decay rate of the initial signal function is not known one operates with the straightforward @xmath28 where @xmath29 is suitable absolute constant .
thus becomes @xmath30 we are interested in estimates for @xmath31 such that vanish with @xmath32 therefore , the obtained upper bounds really will be universal for wide classes of @xmath33 and @xmath34 .
denote @xmath35 the @xmath36-norm in finite case ( while @xmath37 and @xmath38 the set of all functions whose restrictions to @xmath39 have finite @xmath36-norm .
assume @xmath40 be of exponential type @xmath41 and let @xmath42 be a separated real sequence , i.e. such that @xmath43 then it holds true ( * ? ? ?
* eq . ( 76 ) ) @xmath44 where @xmath45 the display ( [ pp ] ) exposes the celebrated plancherel
plya inequality , see @xcite .
it could be mentioned that boas @xcite has been established another estimate in one dimensional case under different assumptions on @xmath34 , and recently lindner publishes an estimate in onedimensional case when @xmath46 @xcite . here
, we will give the multidimensional analog of the plancherel plya inequality .
hereinafter @xmath47 denotes the bernstein class @xcite of @xmath48variable entire functions of exponential type at most @xmath49 coordinatewise whose restriction to @xmath50 is in @xmath51 .
[ th1 ] let @xmath52 be real separated sequence , i.e. @xmath53 .
let @xmath54 .
then @xmath55 where @xmath56 take @xmath57 ; the proof will be identical in the case @xmath58 . from the assumption @xmath59 and the inequality of different dimensions @xcite which holds for @xmath60 follow that @xmath61 and @xmath62 therefore we can apply ( [ pp ] ) coordinatewise .
because @xmath63 are separated with @xmath64 there holds @xmath65 second subsequent application of ( [ pp ] ) to ( [ delta2 ] ) yields @xmath66 where in ( [ delta1 ] ) the fubini s theorem is used . therefore , the assertion of theorem is proved .
the case of wks sampling restoration by nonequidistantly spaced nodes @xmath67 is the so called _ irregular sampling procedure_. in this case there is a doubt on the reality of analogue of regular sampling results on sharp estimates for truncation error upper bounds in uniform sampling ( for example , see @xcite ) , since @xmath68 is wandering together with @xmath18 , and therefore different time values @xmath18 generate windows with substantially different sampling sets .
of course , this results in increasing magnitude of the truncation error @xcite .
let @xmath69 . in @xcite
it was considered a case when outside @xmath24 the time - jitter @xmath70 vanishes .
these results can be generalized to multidimensional case .
namely , having on mind the infinite product representation of sine function and replacing the equidistant nodes having indices inside @xmath71 with the disturbed @xmath72 ones @xmath73 we get the so called _ window canonical product sampling function _
@xmath74 where @xmath75 denotes a derivative with respect to @xmath76 , being @xmath77 so , when @xmath34 is separated , one can deduce @xmath78 where the convergence is uniform on bounded subsets of @xmath50 under additional suitable conditions upon @xmath25 and @xmath34 .
we will discuss the conditions for multidimensional case later ( see @xcite for onedimensional case ) . in this framework
the sampling restoration procedure becomes of lagrange yen type @xcite .
for the general case of multidimensional irregular sampling with window canonical product sampling function @xmath79 we have time jittered nodes outside @xmath24 as well .
nonvanishing time - jitter @xmath70 outside @xmath24 leads to functions @xmath80 given by formulae different from ( [ 20 ] ) , see @xcite .
however , in irregular sampling applications we would like to approximate @xmath33 using only it s values from sample nodes @xmath81 indexed by @xmath82 therefore we can try to use the truncated to @xmath83 sampling approximation sum @xmath84 with @xmath85 ( such that is given by ) even for arbitrary sample nodes outside @xmath24 . under such assumptions the _ truncation error _
@xmath86 coincides with truncation error for the case given by .
it would seem that in this framework @xmath87 depends on samples in @xmath88 , but it does not depend on samples in @xmath89 , which are used to build approximation sum @xmath90 . however , it is not true , because @xmath91 depends on nonvanishing @xmath70 in @xmath24 due to multiplicative form of ( [ 21 ] ) . therefore , the multidimensional sampling problems are more difficult than the one dimensional ones . in this section
we propose a way to obtain universal truncation bounds for multidimensional irregular sampling restoration procedure .
first of all , we will give the multidimensional sampling theorem for @xmath92 functional class .
denote @xmath93 assume that @xmath94 ; for all @xmath95 .
[ th_ms ] let @xmath96 @xmath97 @xmath98 for all @xmath99 @xmath100 be real separated sequence with @xmath101 then the sampling expansion ( [ 21 ] ) holds uniformly on each bounded @xmath18subset of @xmath102 .
moreover , the series in ( [ 21 ] ) converges absolutely too .
here , in , @xmath103 stands for the characteristic function of the random event @xmath104 .
first @xmath34 is obviously separated since @xmath105 now , by the assumption @xmath106 and by the inequality of different dimensions @xcite we conclude @xmath107 . on the other hand we rewrite the function @xmath80 ( defined by ) into @xmath108 where @xmath109
because @xmath80 has representation ( [ g ] ) , it is clear that all results of @xcite remain valid if one uses @xmath110 instead of @xmath111 , and make use of @xmath80 instead of @xmath112 , considered by hinsen @xcite .
therefore , if ( [ m ] ) holds , we have @xmath113 we can subsequently apply the same sampling expansion formula ( [ s1 ] ) with respect to all variables @xmath114 on the right side of ( [ s1 ] ) , being ( [ s1 ] ) absolutely convergent for @xmath115 @xcite .
this procedure results in absolute convergent @xmath48tuple series on the right in ( [ 21 ] ) .
the belonging truncation error upper bound result reads as follows . [ th_ir ]
let @xmath116 satisfy , @xmath117 for all @xmath73
. then we have @xmath118 where @xmath119 and @xmath120 as we mentioned earlier in section 3 , the function @xmath90 does not depend on samples in @xmath88 and we assume that outside @xmath24 it is @xmath121 therefore , the structure of @xmath122 becomes uniform @xmath123 . in this case theorem [ th_ms ]
guarantees that @xmath33 admits the representation ( [ 21 ] ) , and the so evaluated model ( [ 4 ] ) gives us @xmath124 the multiplicative structure of @xmath125 enables to estimate @xmath31 in the following way @xmath126 let us estimate @xmath127 note , that due to our assumptions @xmath128 for all @xmath129 .
hence @xmath130 where @xmath131 denotes the index closest to @xmath132 , i.e. @xmath133 due to @xmath134 we have @xmath135 and @xmath136 then , being @xmath137 let us estimate the first sum : @xmath138 and the second one as @xmath139 collecting all estimates ( [ est1])-([est5 ] ) , we conclude @xmath140 and hence @xmath141 thus , we proceed evaluating @xmath142 such that gives @xmath143 now , let us evaluate @xmath144 , the second addend in .
note , that due to the declared assumptions of the theorem , we have @xmath145 let us consider the case @xmath146 .
then @xmath147 as sinc@xmath148 decreases in @xmath149 $ ] , we get @xmath150 and the inequality ( [ est1 ] ) is valid in our case too .
together with further necessary estimates , such as @xmath151 we deduce @xmath152 let us estimate the first sum as @xmath153 making use of to the second sum in we have @xmath154 combining all estimates ( [ est1])-([est5 ] ) , we obtain @xmath155 let us consider the case @xmath156 as sinc@xmath148 decreases , by ( [ est1 ] ) it follows @xmath157 hence , for @xmath158 we get @xmath159 because ( [ est3])-([est5 ] ) are still valid in this case as well . combining all upper bounds concerning @xmath160 we have @xmath161
it remains to realize the estimate @xmath162 such that gives @xmath163 therefore , by ( [ 26 ] ) @xmath164 to estimate @xmath165 we use theorem [ th1 ] with @xmath166 .
this results in @xmath167 now , collecting all these involved estimates , we arrive at ( [ 22 ] ) and ( [ 23 ] ) .
denote @xmath168 and @xmath169 let us consider the paley - wiener space @xmath170 of all complex - valued @xmath171functions whose fourier spectrum is bandlimited to @xmath172^d$ ] .
[ cor ] let @xmath173 @xmath97 @xmath116 satisfy ( [ m ] )
. then we have @xmath174 where @xmath175 * proof : * we can use the estimate ( [ ap ] ) for @xmath31 from theorems [ th_ir ] with @xmath116 instead of all @xmath176 and @xmath177 instead of all @xmath178 ( in this case @xmath34 could contain some additional @xmath81 which might results only in increasing @xmath31 )
. from @xmath179 follows that @xmath180 decreases when @xmath181 and increases for @xmath182 , where @xmath183 , @xmath184 to find the numerical value of @xmath185 we use the _ mathematica s _ in built numerical routine productlog[z ] for the computation of @xmath186 function which is the inverse function of @xmath187 , @xcite .
namely , we use that branch of @xmath186 which gives the nonzero solution of the equation @xmath188 .
thus , @xmath189 . therefore by ( [ b_destimate ] ) we have the estimate @xmath190 for @xmath191 it holds @xmath192 let @xmath173 @xmath97 @xmath193 @xmath194 and @xmath195 in such way that @xmath196
. then it holds @xmath197 by definitions for constants @xmath198 from theorem [ th_ir ] we get @xmath199 when @xmath200 therefore by ( [ k ] ) @xmath201 when @xmath202 if@xmath203 application of corollary [ cor ] and ( [ m ] ) finishes proof and gives us conditions ( [ m1 ] ) .
in most papers known in literature ( see , for example , @xcite , @xcite , @xcite , and references in them ) to obtain kotelnikov - shannons theorems and upper bounds for approximation errors in @xmath38 spaces various authors considered particular classes of functions with prescribed decay conditions which gave opportunity to estimated @xmath165 in ( [ 4 ] ) in such manner that the estimate vanishes when @xmath204 tends to infinity .
the another term @xmath31 in ( [ 4 ] ) usually was estimated by some constant which did not depend on @xmath205 in this paper we propose estimates for @xmath31 which depends on @xmath204 and vanishes when @xmath204 tends to infinity .
this approach gives opportunity to consider and obtain approximation errors estimates for wide functional classes without strong assumptions on decay behaviour .
new upper error bounds in @xmath206 norm in sampling theorem were obtained .
the case of multidimensional irregular sampling was considered .
results presented in the paper and numerical simulation done by the authors ( consult @xcite ) arise some new open problems : 1 . to obtain sharp estimates in theorem [ th1 ] ; 2 . to obtain sharp estimates in theorem [ th_ir ] ( for uniform sampling and @xmath207 such sharp estimates
were derived in @xcite ) ; 3 . to apply obtained results to stochastic case , see @xcite . in the case
( 3 ) we point out two approaches .
the first one concerns the so called spectral representation @xmath208 of given stochastic process @xmath209 , with deterministic kernel function @xmath210 coming from function space with respect to the first argument , e.g. paley wiener class @xmath211 @xcite , @xcite , @xcite , bernstein class @xmath212 @xcite . under suitable conditions the kernel function possesses wks sampling sum expansion in the related @xmath36norm .
again , by the quoted spectral representation formula one deduces the related wks sampling restoration sum for the initial stochastic signal @xmath213 in the mean
square and/or almost sure sense .
second , when @xmath213 weak sense stationary , it is enough to consider the wks sampling restoration sum for the covariance function @xmath214 of the considered stochastic signal .
then we deduce the final mean square wks sampling restoration sum for @xmath213 . in both cases
the main mathematical tool will be the celebrated karhunen
cramr theorem on integral representation of stochastic processes ( * ? ? ?
* , p. 156 ) however , the structure of considered stochastic signals should mainly influence the convergence conditions and rates of derived wks sampling sums .
the authors were supported by the la trobe university research grant-501821 `` sampling , wavelets and optimal stochastic modelling '' . 00 aldroubi a. and feichtinger h. exact iterative reconstruction algorithm for multivariate irregularly sampled functions in spline - like spaces : the @xmath3-theory , _ proc .
soc . _ * 126 , * no . * 9*(1998 ) , 26772686 .
yaglom , a.m. _ correlation theory of stationary and related random functions ii .
supplementary notes and references . _ springer
verlag , new york @xmath217 berlin @xmath217 heidelberg @xmath217 london @xmath217 paris @xmath217 tokyo , 1986 . | universal ( pointwise uniform and time shifted ) truncation error upper bounds are presented in whittaker kotelnikov shannon ( wks ) sampling restoration sum for bernstein function class @xmath0 @xmath1 when the sampled functions decay rate is unknown .
the case of multidimensional irregular sampling is discussed .
whittaker kotelnikov shannon sampling restoration formula ; approximation / interpolation error level ; nikolski placherel plya inequality ; truncation error upper bound ; irregular sampling ; multidimensional sampling primary 94a20 , 26d15 ; secondary 30d15 , 41a05 |
the formation of electromagnetic showers in aligned single crystals was actively studied during the last decade .
the first experimental investigation of such showers has been performed in @xcite at very high energy of incident electrons .
corresponding theoretical studies were started with @xcite where an analytic solution of the problem was obtained , assuming that energies of all charged particles and photons involved are very high .
this limitation was surmounted in @xcite by suggesting a specific form of the radiation spectrum at axial alignment and performing corresponding simulations . using this approach ,
the results of @xcite for ge crystal had been reproduced in@xcite .
the results of @xcite are consistent with those of @xcite where another approach was used to obtain the radiation spectrum . in @xcite,@xcite , and @xcite ,
the shower characteristics , such as spectral - angular distributions of photons and positrons as well as the amount of energy deposition have been obtained depending on the kind and thickness of crystal targets .
investigations of the specific shower formation give good grounds for the idea proposed in @xcite , to substitute in a positron source an axially aligned crystal target for an amorphous one , as the enhancement of the radiation intensity is maximal just at the axial alignment .
in further experiments ( see [ 8 - 13 ] ) using 1.2 - 10 gev electrons aligned to the @xmath0- axis of tungsten crystals , measurements of some shower characteristics were already aimed to the development of a crystal - assisted positron source .
theoretical estimations performed in @xcite display a rather good agreement with results of recent experiments [ 10 - 13 ] .
so , we can rely on our understanding of the physics of shower formation and on numerical results , at least for tungsten crystals in the energy range of incident electrons below 10 gev .
note that just this energy range is proposed in future linear collider projects ( 2 , 6.2 , and 10 gev correspondingly for clic @xcite , nlc @xcite , and jlc @xcite ) and is considered here .
let us define the positron yield as the number of accepted positrons per one incident electron and the optimal target thickness as that providing the maximal yield .
according to @xcite , @xcite , @xcite , the maximal yield from a crystal target is always higher than that from an amorphous one and the excess goes up when the electron energy increases .
however , the magnitude of such an enhancement is small , less than 14@xmath2 even at 10 gev .
the more pronounced advantage of crystal targets appear in a considerable ( by a factor of two at 10 gev ) decrease of the energy deposition . indeed , the thermal effects caused by the energy deposited in a target are a critical issue for any powerful positron source based on the conventional scheme .
we dwell mainly on this issue in the present paper .
below qualitative arguments are given explaining the lower energy deposition in crystals .
the total deposited energy and the distribution of its density over the target volume are calculated for crystal and amorphous tungsten targets using the parameters of clic , nlc , and jlc .
thereby , a possible gain for these projects resulting from the use of crystal targets in the positron source is estimated .
for accurate studies of thermal effects , some improvements have been performed in the computer code developed in @xcite , @xcite . the updated version of the code is used to study both crystal and amorphous cases .
in the energy range under consideration we are dealing with a `` soft''(see @xcite ) shower when pair production is entirely due to the conventional bethe - heitler mechanism , while the crystal structure reveals in a considerable enhancement of the radiation intensity and a softness of the photon spectra .
remember that this enhancement decreases when the particle energy does so as the shower develops . starting with some depth @xmath3
( see discussion in @xcite , @xcite ) , further development of the shower proceeds more or less in the same way for any ( crystal or amorphous ) type of the remaining part of the target . for the sake of simplicity
, calculations are performed here for the all - crystal targets .
however , they may serve as a good estimate for hybrid targets of the same total thickness and with a crystal - part length of the order of @xmath3 .
let us remind that a hybrid target consists of a photon radiator made of a crystal followed by a pair converter made of an amorphous piece . from fig.[fig
: enbal1 ] , a value of @xmath4 can be chosen for 10-gev electrons , since the fraction of the total energy carried by photons ( @xmath5 ) has been already saturated at this depth and the mean energy of the primary electron is sufficiently low to eliminate crystal effects in the last part .
such a saturation takes place in amorphous targets as well , but with a lower conversion level ( @xmath6 ) and at substantially larger depth ( @xmath7 at 10 gev ) . only a small part ( less than 0.4@xmath2 in the above example ) of the beam power
is deposited over @xmath3 and the energy - deposition density is much less ( about 8 times at 10 gev ) than its peak value .
so , the crystal part of a hybrid target is not exposed to serious thermal effects which appear at larger depths in the later stage of the shower development
. from calculations performed in @xcite , @xcite , the energy deposition in equivalent ( providing the same positron yield ) targets is always less in the crystal case .
let us present some qualitative arguments to explain this fact .
the main process leading to the energy deposition is the ionization loss , the rate of which , @xmath8 , reads approximately as @xmath9 , where @xmath10 is the number of charged particles at the depth @xmath11 . strictly speaking ,
the coefficient @xmath12 may depend on @xmath11 but its small variation as well as a small difference of @xmath12-values in crystal and amorphous cases are neglected in our estimation .
the total energy , @xmath13 , deposited over the thickness @xmath14 reads @xmath15 or , going over to the variable @xmath10 @xmath16 for sufficiently large @xmath14 , the positron yield is roughly proportional to the total number of charged particles , @xmath17 , i.e.,for equivalent targets , the integrals in ( [ qen2 ] ) are taken over the same region of variable @xmath18 in both cases . to prove our statement , it remains only to verify that the logarithmic derivative which appears in the denominator of the integrand in ( [ qen2 ] ) is larger for crystals .
this derivative , or logarithmic increment , characterizes the growth of the number of charged particles . as seen in fig.[fig : enbal1 ] , the conversion of the initial electron energy into photons is going faster for crystals , where , correspondingly , the pair - production process starts earlier and is more intensive , resulting in a larger increment . for the purposes of illustration ,
the energy - deposition rate per charged particle and the logarithmic increment are shown in fig.[fig : rate2 ] .
we emphasize that the energy - deposition rate is practically the same for crystal and amorphous samples , being almost independent of the electron energy .
the non - constancy of this rate is mainly due to the contributions of other processes like photon absorbtion and annihilation of positrons which are taken into account by our simulations but were ignored in the estimation of @xmath8 used in ( [ qen1 ] ) .
evidently , the role of these processes increases with growing depth .
a slow decrease of the rate at comparatively small depth is due to that of the mean particle energy .
the point is that we use the so - called non - restricted energy loss description , where the rate diminishes when the particle energy does so . on the whole , the results of simulations presented in fig.[fig : rate2 ] confirm the above qualitative considerations .
as shower develops , particles of sufficiently low energy may substantially change their direction of propagation due to single or multiple scattering . so that there even appears a `` backward '' flux of particles moving upstream
; they do not increase the positron yield , however , they heat the target . this contribution is taken into account in our simulations , where we are able to trace separately effects from `` forward '' and `` backward '' particles and photons . to study the shower characteristics depending on depth ( @xmath11 ) , the target is divided into slices by planes perpendicular to the incident beam direction , taken as @xmath11-axis .
the spacing @xmath19 is used ( remember that @xmath20 = 3.5 mm for tungsten ) .
energy deposition is simulated within each slice and various distributions are recorded at the right - hand boundary ( plane ) of the slice .
in particular , the development of momentum distributions for positrons and photons is obtained as well as that of the beam spot size . shown in fig.[fig : edep3](a ) is the energy , @xmath21 , deposited in slices in units of @xmath22 .
this quantity is higher for the lower energy .
note that the deposited power,@xmath23 , reads as @xmath24 where @xmath25 is the incident beam power .
the contribution of the `` backward '' particles to @xmath26 increases with the target thickness @xmath14 ; it amounts typically to about 20@xmath2 at @xmath27 and thereby is not negligible .
the results of simulations presented in fig.[fig : edep3 ] ( b ) clearly confirm once more ( cf .
fig.9 in @xcite ) the statement concerning the comparison between the values of @xmath26 in the equivalent targets .
the positron yield is calculated using the `` theoretical '' acceptance conditions from @xcite ( see figs.11 and 12b in @xcite ) in all three cases .
the angular spread of the incident beam is neglected and the transverse size of the beam , @xmath28 is set to 1.6 mm at 2 and 6.2 gev , and to 2.5 mm at 10 gev . at @xmath22=2 gev , a simulation for @xmath29=2.0 mm
was performed as well .
corresponding results , not shown on graphs , are presented in the table below .
let us remind now that , at equal depths and initial energies , charged particles are softer and have a larger angular spread for crystal targets ( see , e.g. , figs . 2 - 5 in @xcite and corresponding discussion ) .
all other things being equal , positron spectra are softer at a lower initial energy . as an illustration of these features , the mean energy and the transverse momentum , @xmath30 , of `` forward ''
positrons are shown in fig.[fig : menspo4](a ) .
we emphasize that , starting with @xmath31 , @xmath30 is almost constant and practically independent of the initial electron energy and the type of the target ( note the merging of six different curves on fig.[fig : menspo4](a ) ) .
such result confirms and extends those concerning an amorphous target ; it is essentially due to a counterbalance between the increase of the angle as a consequence of the multiple scattering and the decrease of the positron energy with increased thickness @xcite . at the same time
, the larger angular spread leads in the crystal case to a larger beam spot area , @xmath32 , as seen in fig.[fig : menspo4](b ) where @xmath33 is plotted for `` forward '' charged particles . for evident reasons , the spot area of the `` backward '' charged flux ( not shown in fig.[fig : menspo4 ] )
is somewhat larger ( typically by 20 - 25 @xmath2 ) than that of the `` forward '' one . whereas the average deposited power can be handled somehow ( e.g. , by rotating the target and removing the excess heat through water cooling as used at slc ) , the local and nearly instantaneous energy deposition is unavoidable , while being of critical concern for target damage .
basing on the analysis of the slc damaged target ( see @xcite ) , it is now adopted that the peak energy - deposition density should not exceed 35 j / g to ensure a sufficiently long term of safe operation . in our simulations ,
the total scanned volume is a cylinder coaxial with the incident beam direction .
the radius , @xmath34 , of the cylinder was about @xmath35 , in which case less than one thousandth of initial electrons does not hit the cylinder at entry .
this cylinder is sliced into disks of the thickness @xmath19 , i.e. , @xmath11-spacing is the same as in above calculations , allowing a mutual checking . in turn , each disk is divided by circles of uniformly increasing radii with the step @xmath36 into 50 parts - one internal disk ( altogether , such disks form the internal cylinder ) and 49 rings .
more precisely , we have @xmath37=1.6mm)=0.12 mm , @xmath37=2.0mm)= 0.15 mm , and @xmath37=2.5mm)=0.185 mm .
the energy deposition and the number of charged particles is simulated in each meshed volume providing corresponding transverse distributions for each @xmath11-slice .
it is noteworthy that @xmath11-dependencies of shower characteristics derived from 3-dimensional distributions coincide with those obtained in direct calculations , thereby verifying the consistency of the two essentially different forms of the output which we have used . from fig.[fig : endede ] ( b ) , the energy - deposition density ( edd ) drops at large cell radius , being maximal in the internal disk of each slice . shown in fig.[fig : endede ] ( a ) is the edd per one incident electron in the internal cylinder , versus the depth . like the positron yield and the total number of charged particles , the peak values of edd ( pedd ) in crystals
are somewhat higher and are reached at smaller depths , compared with amorphous targets . at equal beam size , pedd is higher for the higher energy ( cf .
pairs of curves in fig.[fig : endede](a ) calculated at 2 and 6.2 gev for the same @xmath381.6 mm ) .
however , by increasing the beam size , lower values of pedd may be obtained even at higher energy .
this is seen in fig.[fig : endede ] ( a ) if we compare the curves calculated at 10 gev ( @xmath29 = 2.5 mm ) with those at 6.2 gev .
examples of transverse distributions of edd are plotted in fig.[fig : endede](b ) for the same depth @xmath39 .
the distribution of charged particle density ( not shown ) is similar to that of edd with a ratio edd / density being almost constant within the first ten cells and roughly equal to the energy - deposition rate . in fig.[fig : endede](b ) this similarity reveals in larger width of distributions for crystal targets ( cf .
fig.[fig : menspo4](b ) ) .
the edd is presented in fig.[fig : endede ] in units of @xmath40 which for tungsten corresponds to @xmath41 .
due to an extremely short duration of the pulse , contributions to edd from all the incident electrons are added and the resulting edd - value is simply the product of the obtained edd - value per @xmath42 and the number of electrons per pulse , provided that the target reverts to initial thermal conditions .*energy deposition in crystal and amorphous targets*. the fraction of an incident beam power deposited in a target , @xmath43 ; the peak energy deposition density , @xmath44 , in units of j / g ; the energy,@xmath22 , and transverse size of an incident beam,@xmath45 ; target thickness , @xmath14 , measured in conventional radiation lengths , @xmath20 ; gain in the total deposited power , @xmath46 , from comparison of @xmath47 and @xmath48 targets [ cols="^,^,^,^,^,^,^,^,^,^",options="header " , ] during the repetition period .
some results concerning the energy deposition in crystal and amorphous targets are presented in table [ tab : compar ] . for the sake of comparison , the thicknesses of amorphous tungsten targets ( @xmath49 ) are the same as proposed for @xmath50 targets in the projects @xcite , @xcite , and @xcite . for crystal targets ( @xmath51 ) ,
the corresponding thicknesses are determined from the equivalence rule ( the same positron yield as for @xmath49 target ) .
our values for @xmath52 given in table [ tab : compar ] are obtained using the same numbers of electrons per pulse , @xmath53 , @xmath54 , and @xmath55 , as in @xcite , @xcite , and @xcite respectively . since the parameters of the two amorphous targets such as radiation lengths and densities almost coincide , the same is expected for shower characteristics at equal depths .
however , one should bear in mind that two different computer codes were used in simulations , egs4 for @xmath50 targets , and our code for @xmath49 targets . from table [
tab : compar ] , the concordance of the two approaches is almost perfect for the fraction of the deposited power , @xmath56 . concerning the peak energy - deposition density , @xmath52 , the values obtained for the @xmath49 target at 2 and 10 gev are somewhat smaller than those for the @xmath50 target and the distinctions are not too big .
on the contrary , our value for the @xmath49 target is substantially larger at 6.2 gev .
let us argue the point in detail . at equal beam size
, the peak edd / electron is expected to be roughly proportional to the initial electron energy , @xmath22 .
this assertion follows from the qualitative consideration performed above and is verified by calculations .
for example , if we compare the curves in fig.[fig : endede](a ) calculated at 2 and 6.2 gev for the same @xmath381.6 mm , we obtain @xmath572.7 as the peak edd ratio instead of @xmath573.1 from the rough estimate where @xmath58 . at @xmath59 , which corresponds to nlc conditions
, this ratio further diminishes up to 2.55 since the peak edd value is achieved for @xmath22 = 6.2 gev at a larger depth .
so , at equal beam size , the relationship between @xmath52 values at different energies reads roughly as @xmath60 the @xmath52 in @xmath50 target at @xmath22 = 2 gev was calculated in @xcite not only for @xmath382.0 mm , but also for @xmath381.6 mm , where the value of 53.1 j / g has been obtained ( at these conditions we have 47.7 j / g for the @xmath49 target ) . using the estimate ( [ est3 ] ) with @xmath61=53.1 j / g and @xmath572.55 , we obtain for nlc conditions @xmath62=62.5 j / g which is consistent with our result for the @xmath49 target .
conversely , starting with results of @xcite for the @xmath50 target , we obtain the estimate ( @xmath61=35 j / g , @xmath63= 10 gev , @xmath570.62 ) @xmath64 33 j / g at @xmath382.5 mm , @xmath65=6.2 gev .
so , the @xmath52 value of 35 j / g at @xmath22=6.2 gev is more consistent with the beam size of @xmath382.5 mm than with @xmath381.6 mm indicated in @xcite . comparing the magnitude of thermal effects in @xmath49 and @xmath51 targets providing the same positron yield ,
we conclude that , using crystal targets , the total deposited energy can be considerably diminished while the peak value of the energy - deposition density is kept approximately on the same level as in the amorphous case .
one of us ( v.s . ) is thankful for kind hospitality during his stay at lal where a part of this work has been done .
he is also grateful to the russian fund of basic research for partial support of this work by grants 01 - 02 - 16926 and 03 - 02 - 16510 . | the characteristics of the electron - photon showers initiated by 2 to 10 gev electrons aligned along the @xmath0 - axis of tungsten crystals are compared with those for the amorphous tungsten . in this energy range , as known , the positron yield at the optimal target thicknesses is larger in a crystal case only by several percent . however , the amount of the energy deposition in a crystal turns out to be considerably ( by 20 - 50@xmath1 ) lower than in an amorphous target providing the same positron yield , while the peak energy - deposition density is approximately of the same magnitude in the both cases .
+ + pacs numbers : 07.77.ka , 12.20.ds , 03.65.sq + |
the four - electron li@xmath0 ion is interesting because of the significant role played by electron correlation in the binding of the outermost electron in this weakly bound system .
the major contribution to the correlation energy arises from the interaction of the two valence electrons . beyond the three - body h@xmath0 system ,
the li@xmath0 ion is most tractable to theory . in the frozen core approximation , for example
, the li@xmath0 ion becomes an effective three - body system consisting of a pair of highly correlated electrons interacting weakly with an inert core .
such a model lends itself well to semi - empirical model potential calculations in which the potential experienced by the valence electrons is obtained via spectral information on the eigenvalue spectrum of the li atom .
most calculations of the electron affinity of li to date are of this type , e.g. moccia et al . @xcite or graham et al .
@xcite ( and references therein ) .
recently , however , accurate measurements of this quantity have stimulated ab initio calculations @xcite of comparable accuracy .
the most direct , and potentially the most accurate , method of measuring electron affinities is to use the laser threshold photodetachment ( ltp ) method @xcite . here
one records , as a function of the wavelength of a tunable laser , the onset of production of either photoelectrons or residual atoms in the vicinity of a detachment threshold .
to date , three ltp measurements of the electron affinity of li have been reported .
the earliest such experiment was a crossed laser and ion beam experiment by feldman @xcite .
he utilized an infrared laser to study the li@xmath4 photodetachment threshold .
an accurate threshold energy for a @xmath5-wave detachment was difficult to determine because the cross section rises , according to the wigner law @xcite , only slowly from zero .
bae and peterson @xcite used collinear laser and ion beams to investigate the total photodetachment cross section around the li(@xmath6 ) cusp situated at the opening of the li@xmath7 channel . from a careful analysis of this sharp structure they obtained , as one result , an electron affinity value of about the same accuracy as feldman .
the measurement of dellwo et al .
@xcite was a direct investigation of the resolved li@xmath7 channel threshold using photoelectron spectroscopy . in this experiment , however , doppler broadening associated with the use of crossed laser and ion beams limited the attainable energy resolution .
the electron affinity of li determined in the present measurement is an order of magnitude more accurate than previous lpt measurements .
we utilized resonance ionization @xcite combined with a collinear laser - ion beam geometry to measure the threshold of the li@xmath7 partial photodetachment cross section .
the state selectivity of the resonance ionization method leads to an excellent signal - to - background ratio .
this in turn enabled us to attain a resolution limited only by the laser bandwidth of about 0.2 @xmath8 .
the present threshold energy measurement clearly demonstrates the potential of the method .
the concept of combining collinear laser - ion beam spectroscopy with resonance ionization detection was first proposed by kudriatsev and letokhov @xcite and later applied to isotope detection measurements by the same authors @xcite .
balling et al .
@xcite and petrunin et al .
@xcite have recently used the same technique in photodetachment measurements .
the two - color state selective photodetachment experiment described in the present paper is simple in concept .
one laser of frequency @xmath9 is used to photodetach li@xmath0 ions producing an excited li atom and a free electron ( fig . 1 ) .
a second laser of frequency @xmath10 resonantely photoexcites li atoms left in the @xmath6 state to a rydberg state which subsequently is field ionized .
hence , the entire process can be represented by the following steps : @xmath11 where @xmath12 denoted field ionization and li(@xmath13 ) corresponds to a highly excited rydberg atom in either a @xmath14 or @xmath15 state .
state selectivity is accomplished in the resonant ionization step since only li(@xmath6 ) atoms can be ionized via the intermediate rydberg state . in this manner
we were able to isolate a particular photodetachment channel , in this case the li(@xmath6 ) channel , and investigate the partial photodetachment cross section by measuring the yield of li@xmath2 ions .
the @xmath16li@xmath0 ion beam was produced by charge exchange in a cesium vapor cell of a mass selected li@xmath2 beam from a plasma ion source .
an ion current of typically a few na was obtained in the interaction region .
the beam energy was approximately 4 kev . in the interaction chamber ( fig .
2 ) the negative ions interacted with laser light in a region defined by two apertures with a diameter of 3 mm placed 0.5 m apart . the ions were deflected in and out of the laser beam by means of two electrostatic quadrupole deflectors whose symmetry axes were perpendicular to the laser and ion beams .
the ion current in the interaction region was monitored with a faraday cup placed after the second quadrupole deflector .
rydberg atoms formed in the interaction region travel to the second quadrupole where they are ionized by the same electric field that deflects the negative ion beam into the faraday cup .
positive ions formed in this process were deflected in the opposite direction into a positive ion detector . in this detector
the fast positive ions impinged on a conducting glass plate producing secondary electrons that were detected with a channel electron multiplier ( cem ) . a metal grid connected to a voltage supply
was placed between the glass plate and the cem .
this made it possible to either allow or prevent the secondary electrons from reaching the cem .
the detection efficiency of the positive ion detector was close to unity .
light of frequency @xmath9 was generated by a dye laser operated with coumarine 307 and light of frequency @xmath10 was produced by a dye laser operated with bmq .
both dye lasers were pumped by a common xecl excimer laser .
the pulse duration of both lasers was about 15 ns .
the maximum energy in a laser pulse delivered into the interaction region was 1.5 mj for the radiation of frequency @xmath9 and 200 @xmath17j for the radiation of frequency @xmath10 . during the experiment both lasers were attenuated , as will be discussed below .
the two laser pulses counter propagated in the interaction region and arrived simultaneously in the interaction region .
a shift of a few ns between the pulses , occurring at both ends of the interaction region , is negligible since the li(@xmath1 ) radiative lifetime is 26.99(16 ) ns @xcite .
the frequency @xmath9 was determined by combining fabry - prot fringes with optogalvanic spectroscopy .
the fabry - prot fringes served as frequency markers whereas transitions of ne or ar in a hollow cathode lamp provided an absolute calibration of the energy scale .
first we had to find an appropriate resonance transition for the laser of frequency @xmath10 . a scan of @xmath10 , during which @xmath9 is set to a frequency far above the li@xmath18 threshold , is shown in fig . [ uv2 ] .
below the ionization limit we found a wealth of lines . due to strong saturation ,
the intensities of the lines are not proportional to the corresponding transition probabilities ; close to the ionization limit the detection system became overloaded such that it only recorded one count per laser shot thus attenuating the strongest lines . for most of the lines we also expect the transition itself to be saturated .
the rydberg series in fig .
[ uv2 ] exhibits a cut off at 28270 @xmath8 indicating that the rydberg atoms were mainly field ionized in an electric field of about 200 kv / m in the second quadrupole deflector .
we interpret the lines below the cut off as being due to photoionization .
a detailed scan of the transition used for the photodetachment threshold measurements is shown in fig .
we identified the two lines as corresponding to transitions from the two fine structure levels @xmath19 to the same rydberg state .
the measured separation of 0.33(4 ) @xmath8 between the two lines in fig .
[ fein ] corresponds very well to the fine structure splitting of 0.337 @xmath8 of the @xmath1 term @xcite .
the resolved fine structure shown in fig . [ fein ] demonstrates that the energy resolution of the present collinear beam apparatus is approximately 0.2 @xmath8 , which is of the same order as the laser bandwidth . a typical measurement of the li(2@xmath5 ) photodetachment threshold with anti - parallel laser and ion beams
is shown in fig .
the laser frequency @xmath10 was held constant and the intensity was set to saturate the transition to the rydberg state .
the frequency @xmath9 was then scanned over the li(@xmath6 ) threshold .
it was established that two processes contributed to the background , namely ( a ) : @xmath20 and ( b ) two - electron collisional ionization .
process ( a ) produced nearly 80% of the background , even though the intensity of the laser light with frequency @xmath10 was attenuated .
process ( b ) , two - electron collisional ionization , contributed the remaining background at the operating pressure of @xmath21 mbar ( @xmath22 pa ) .
this contribution was found to be proportional to the pressure and is thus dominated by single collision double detachment , as has been discussed previously by bae and peterson @xcite .
the laser intensities were too low for other processes , such as direct two - electron multiphoton detachment , to influence the experiment .
close to the photodetachment threshold the cross section @xmath23 is well represented by the wigner law @xcite ( eq . [ eqwig ] ) .
if @xmath24 is the threshold energy and @xmath25 the photon energy , the wigner law for electrons detached with different angular momenta @xmath26 can be written as @xmath27 for the desired first step process in eq .
[ stepone ] , both @xmath28- and @xmath29-wave final states for the detached electron are allowed by parity and total angular momentum conservation . at the threshold
, however , the @xmath28-wave cross section , starting with an infinite slope , completely dominates over the @xmath29-wave cross section .
we fitted the wigner law ( eq .
[ eqwig ] ) for @xmath28-wave detachment ( @xmath30 ) to the recorded positive ion signal .
the data shown in fig . [ wig ] was recorded with the laser ( @xmath9 ) anti - parallel to the ion beam .
the fit window is the range of the plotted line .
the fit yields a value for the red - shifted threshold energy , @xmath31 . in order to eliminate the doppler shift we repeated the measurements using parallel laser and ion beams to determine the blue - shifted threshold energy , @xmath32 the threshold energy corrected for the doppler to all orders , @xmath24 , is given by the geometric mean of the two measurements : @xmath33 the interaction of the induced dipole moment of the atom with the electron will limit the range of validity of the wigner law . omalley
@xcite obtained an expression which accounts for the residual induced - dipole point - charge interaction , @xmath34\quad , \ ] ] where @xmath35 and @xmath36 is in atomic units .
this equation can be used to estimate an approximate range of validity for the wigner law by determining the conditions under which the second term , and therefore higher terms , become negligible . with a dipole polarizibility of @xmath37 for the li @xmath1 state @xcite ,
the second term is much less than unity up to 1 @xmath8 above the threshold , and consequently the wigner law should be applicable over this range .
this , of course , assumes that there are no resonances in this region , a condition which has been previously established by dellwo et al.@xcite .
saturation of the signal can originate in the photodetachment process or the detection system . in the data shown in fig .
[ wig ] the detection system starts to saturate at about 250 counts ( for 100 laser pulses ) .
we attenuated the intensity of the laser of frequency @xmath9 sufficiently to avoid any saturation of the signal within the first 1.0 @xmath8 above the threshold . to determine the electron affinity of li we had to subtract the well known @xmath38 transition energy of 14903.648130(14 ) @xmath8 @xcite from the measured doppler- .comparison of different measurements and calculations of the li electron affinity .
[ cols="<,^ " , ] corrected threshold energy @xmath24 ( eq . [ schwelle ] ) .
we determined the li electron affinity to be 4984.90(17 ) @xmath8 . to the error
there are two major contributions : 0.13 @xmath8 is due to a calibration uncertainty of the hollow cathode lamps , and the rest is due to statistical scattering of the fitted threshold values . in order to convert our electron affinity in @xmath8
to ev we used the recommended factor of ( 1/8065.5410 ) @xmath39 $ ] @xcite and obtained 0.618049(20 ) ev . in fig .
[ ealithium ] this value is compared with previous experiments .
these values are also compiled in tab .
[ tab1 ] together with some recently calculated li electron affinities .
we have demonstrated that photodetachment spectroscopy combined with resonance ionization is a powerful method for studying partial photodetachment cross sections of negative ions .
the collinear beam geometry simultaneously provides high sensitivity , due to the large interaction volume , and excellent resolution , due to velocity compression .
i addition , removal of the doppler shift to all orders can be achieved by use of two separate measurements involving parallel and anti - parallel laser and ion beams .
both merits can be fully retained for channel specific photodetachment investigations with the excitation scheme used in this experiment .
our improved li electron affinity of 0.618049(20 ) ev reveals the potential of this method , and this method can , in principle , be extended to essentially all elements .
we are currently developing this type of state - selective detection scheme in connection with on - going studies of high lying doubly excited states of the li@xmath0 ion @xcite .
financial support for this research has been obtained from the swedish natural science research council ( nfr ) .
personal support was received from the wenner - gren center foundation for igor kiyan .
david pegg acknowledge the support from the swedish institute and and the us department of energy , office of basic energy sciences , division of chemical sciences . | we have investigated the threshold of photodetachment of li@xmath0 leading to the formation of the residual li atom in the @xmath1 state .
the excited residual atom was selectively photoionized via an intermediate rydberg state and the resulting li@xmath2 ion was detected .
a collinear laser - ion beam geometry enabled both high resolution and sensitivity to be attained .
we have demonstrated the potential of this state selective photodetachment spectroscopic method by improving the accuracy of li electron affinity measurements an order of magnitude . from a fit to the wigner law in the threshold region , we obtained a li electron affinity of @xmath3 ev . |
it is now widely believed that the afterglow emission of gamma
ray bursts is due to the deceleration of the relativistic fireball in the circum burst matter ( for reviews see piran 1999 ; 1999 ) .
this produces a shock that accelerates electrons to random relativistic energies and probably enhances the magnetic field , leading to the production of synchrotron emission .
if the magnetic field is completely tangled over the entire emission seen by the observer , the resulting synchrotron emission is unpolarized . on the other hand
a very high degree of linear polarization can be expected if a fraction of the magnetic field is well ordered , reaching 6070% in the case of a completely ordered field .
polarization values in the optical band in the range 330% have been indeed observed in cosmic sources , like bl lac objects and high polarization quasars ( see e.g. angel & stockman 1980 ; impey & tapia 1990 ) , whose radiation is believed to be produced by the synchrotron process .
one therefore expects that also in gamma ray burst afterglows the emission is polarized , and attempts were made to measure it .
after the upper limit ( 2.3% ) found for grb 990123 ( hjorth et al .
1999 ) , covino et al .
( 1999 ) detected linear polarization in the afterglow of grb 990510 , at the small but significant level of [email protected]% .
this detection was then confirmed by wijers et al .
( 1999 ) who detected similar polarization values two hours and one day later . on the theoretical side , gruzinov & waxman ( 1999 , hereafter gw99 ) and gruzinov ( 1999 ) predict values around 10% , significantly larger than observed .
this estimate is based on the assumption that the overall emission reaching the observer is produced in a finite number @xmath2 of regions causally disconnected , each of which is embedded in a completely ordered magnetic field .
the predicted total polarization level is @xmath3 , equal to @xmath410% for @xmath2 .
gw99 discuss how the coherence length of the magnetic field generated at the external shock front of a grb fireball grows with time .
if , however , the magnetic field is generated at the collisionless shock front , which is extremely local , it is not clear why the magnetic field embedded in the newly swept matter should be linked to the field in the regions behind the shock .
an alternative magnetic field generation process ( and hence geometry ) has been discussed by medvedev & loeb ( 1999 , ml99 hereafter ) , who consider a magnetic field completely tangled in the plane of the shock front , but with a high degree of coherence in the orthogonal direction . in the case of a spherical fireball
this geometry produces no polarization unless a part of the fireball emission is amplified and part is obscured , as is the case of interstellar scintillation . in this case , however , the resulting polarization can be much better observed at radio wavelengths and should show a rapid and erratic change of the position angle .
we here propose an alternative model , in which the magnetic field geometry is analogous to that of ml99 , but in a fireball that is collimated in a cone and observed slightly off axis . in this case
the circular symmetry is broken and net polarization can be observed ( see e.g. hjorth et al . 1999 , covino et al . 1999 ,
wijers et al . 1999 )
. evidences for beaming of the fireball of grb 990510 from the anomalous decay of the optical lightcurve has been discussed in many recent papers ( harrison et al . 1999 ,
israel et al .
1999 , stanek et al .
1999b ) .
the key assumption of our model is that the fireball is collimated in a cone , observed slightly off axis .
the key result we obtain is the polarization lightcurve , its connection with the flux behavior and a characteristic change of 90@xmath5 in the polarization angle , making the model very testable .
assume a slab of magnetized plasma , in which the configuration of the magnetic field is completely tangled if the slab is observed face on , while it has some some degree of alignment if the slab is observed edge on .
such a field can be produced by compression in one direction of a volume of 3d tangled magnetic field ( laing 1980 , hereafter l80 ) or by weibel instability ( ml99 ) .
if the slab is observed edge on , the radiation is therefore polarized at a level , @xmath6 , which depends on the degree of order of the field in the plane . at the angle @xmath7 from the normal of the slab , the degree of polarization can be expressed by
, following l80 : @xmath8 if the emitting slab moves in the direction normal to its plane with a bulk lorentz factor @xmath9 , we have to take into account the relativistic aberration of photons .
this effect causes photons emitted at @xmath10 in the ( primed ) comoving frame @xmath11 to be observed at @xmath12 ( see also ml99 ) .
we assume that in gamma ray burst fireballs the emitting region is a slab expanding radially and relativistically , compressed along the direction of motion .
we assume also that the fireball is collimated into a cone of semi aperture angle @xmath13 , and that the line of sight makes an angle @xmath14 with the jet axis ( upper panel of fig .
[ fig : geom ] ) .
as long as @xmath15 , the observer receives photons from a circle of semi - aperture angle @xmath16 around @xmath14 ( i.e. within the grey shaded area of fig .
[ fig : front ] ) . consider the edge of this circle : radiation coming from each sector is highly polarized , with the electric field oscillating in radial direction ( see also ml99 ) . as long as we observe the entire circle , the configuration is symmetrical , making the total polarization to vanish .
however , if the observer does not see part of the circle , some net polarization survives in the observed radiation .
this happens if a beamed fireball is observed off
axis when @xmath17 .
the probability to observe a cone along its axis is vanishingly small , since it corresponds to a small solid angle ; it is therefore more likely to observe the collimated fireball off axis . if the cone angle @xmath13 is small , the probability @xmath18 is approximately distributed as : @xmath19 where @xmath20 is the average off axis angle .
assume therefore @xmath21 ( fig .
[ fig : front ] ) . at the beginning of the afterglow ,
when @xmath9 is large , the observer sees only a small fraction of the fireball ( grey shaded region in fig .
[ fig : front ] ) and no polarization is observed . at later times ,
when @xmath9 becomes smaller than @xmath22 , the observer will see only part of the circle centered in @xmath14 : there is then an asymmetry , and a corresponding net polarization flux ( horizontally line shaded region of fig .
[ fig : front ] ) . to understand why the polarization angle in this configuration is horizontal ,
consider that the part of the circle which is not observed would have contributed to the polarization in the vertical direction .
this missing fraction of vertical polarization does not cancel out the corresponding horizontal one , which therefore survives . at later times
, as the fireball slows down even more , a larger area becomes visible .
when @xmath23 , the dominant contribution to the flux comes from the upper regions of the fireball ( see fig . [ fig : front ] ) , which are vertically polarized .
the change of the position angle happens when the contributions from horizontal and vertical polarization are equal , resulting in a vanishing net polarization . at still later times , when @xmath24 , light aberration vanishes , the observed magnetic field is completely tangled and the polarization disappears .
we therefore expect two maxima in the polarization lightcurve , the first for the horizontal component and the second for the vertical one .
the quantitative calculation of the lightcurve of the polarized fraction has been worked out as follows .
we assume that the emission of each small volume of the fireball is isotropic in the comoving frame @xmath11 .
we also assume that in this frame each small element emits the same instantaneous intensity .
for each ring of radius @xmath25 and width @xmath26 of fireball material around the line of sight ( see fig .
[ fig : inte ] ) the relativistically enhanced monochromatic intensity has been computed as @xmath27 , where @xmath28^{-1}$ ] is the relativistic doppler factor , and @xmath29 . for @xmath30
we then have @xmath31 .
3 shows the geometrical set up of the system . in our calculations , the distances @xmath32 , @xmath33 and @xmath25 are considered as angles , since all of them scale with the distance @xmath34 of the fireball from the center of explosion ( see fig .
1 ) . in this case
, @xmath25 is equivalent to the angle @xmath7 between the velocity vector of each element and the line of sight .
therefore the intensity observed from each element is a function of @xmath9 and @xmath25 .
the total intensity is obtained by integrating over the entire surface of the fireball , taking into account that each ring is observed at a different angle , and then characterized by a different @xmath35 .
we assume a constant spectral index @xmath36 throughout our calculations .
the total intensity from the entire fireball at a given time ( and hence at a given @xmath9 ) is : @xmath37 d\rho \nonumber\ ] ] the first integral corresponds to the grey shaded area of fig . 2 , and to @xmath38 ( see fig .
3 ) . the second integral corresponds to @xmath39 . in this case
we receive radiation only from a sector of the ring , between the angles @xmath40 and @xmath41 , with @xmath40 given by : @xmath42\ ] ] we now compute the polarized intensity . to this aim
, we consider again each element of the ring of radius @xmath25 . after calculating the corresponding viewing angle in the comoving frame
, we apply eq . 1 to derive the intensity of the linearly polarized light .
the position angle of the polarization produced by each element is in the radial direction .
we then calculate the polarization of each ring by summing the polarization vectors .
finally , we integrate over @xmath25 .
it is convenient to write the polarization vector as a complex number @xmath43 and integrate it between @xmath40 and @xmath44 ( see fig .
[ fig : inte ] ) . here
@xmath45 is the position angle of the linear polarization in the observer frame .
the polarization of a generic ring is : @xmath46 since the result is a real number , the polarization direction can lie either in the plane that contains both the line of sight and the cone axis ( we call it vertical " polarization angle ) or in the orthogonal plane ( i.e. horizontal " polarization angle ) .
no intermediate values are possible .
integration over @xmath25 then yields : @xmath47 \ , d\rho \label{eq : pola}\end{aligned}\ ] ] for simplicity , we neglected the light travel time effects introduced by the overall curvature of the emitting regions ( see fig .
[ fig : geom ] ) , approximating the emitting volume with a slab . since the degree of polarization of each element
is divided by the total intensity of the same element , we expect these effects to be small in our case
. note that to include this effect requires to assign a specific relation between @xmath9 and the observed time @xmath48 , which can be different according to different models ( due to e.g. adiabatic vs radiative evolution and/or gradients in the interstellar density ) .
the result of the numerical integration of eq .
[ eq : pola ] is shown in the lower panel of fig .
[ fig : lcur ] for four different values of the off axis ratio @xmath49 . for the specific cases shown in the figure we have assumed @xmath50 , but the general properties of the polarization lightcurve
are unaffected by the particular choice of @xmath13 .
all the lightcurves ( except the one with the lower off axis ratio , which shows almost zero polarization ) , are characterized by two maxima in the polarized fraction . as discussed above , at the beginning the polarization is horizontal , it reaches a first maximum and then , when the polarized fraction vanishes , the angle abruptly changes by @xmath51 .
the polarized fraction raises again and then finally declines to zero at late times .
the second peak has a value always larger than the first one . by interpolating the results of the numerical integrations ( assuming @xmath36 )
, we can express the maximum of the polarization lightcurve as a function of the off axis angle : @xmath52 \nonumber\end{aligned}\ ] ] which is accurate within @xmath53 in the specified @xmath14 and @xmath13 ranges .
this maximum always corresponds to the second peak .
the scenario proposed above for the polarization behavior has a strict and direct link with the behavior of the lightcurve of the total flux .
the exact connection will depend on the specific model assumed for the deceleration of the fireball ( i.e. homogeneous or radially distributed density of the interstellar medium , adiabatic or radiative regime , and so on ) , but there are general properties that can be incorporated in any model . for illustration ,
assume the simplest model of a spherical fireball expanding adiabatically in a homogeneous medium , predicting that the bulk lorentz factor @xmath54 , and predicting a decay law for the flux density @xmath55 ( & rees 1997 ) , where @xmath56 is the spectral index of the radiation spectrum , i.e. @xmath57 .
this law assumes that the flux is proportional to the source solid angle : if the fireball has reached a region of size @xmath34 , then the accessible solid angle @xmath58 . in the case of a fireball collimated in a cone of constant opening angle ( e.g. neglecting for simplicity the possible lateral spreading of the fireball : rhoads 1997 , 1999 ; see see also moderski , sikora & bulik 1999 ) , we will have the above relation as long as @xmath59 . on the other hand , when @xmath16 becomes greater than @xmath60 , all the cone front becomes visible , and therefore @xmath61 .
this produces a steepening of the power law lightcurve decay , which now becomes @xmath62 ( see also mszros & rees , 1999 ) . at intermediate times , for which @xmath63 , the accessible solid angle increases with a law intermediate between @xmath64 and @xmath65 , producing a @xmath66 steepening in the lightcurve . assuming again @xmath54 , and calling @xmath67 , @xmath68 and @xmath69 , @xmath70 the corresponding times , we simply have @xmath71 in the case of the grb 990510 afterglow , the decay laws have been @xmath72 at early times ( galama et al .
1999 ) up to @xmath4 half a day after the burst event , and @xmath73 after @xmath45 days ( israel et al . 1999 ) .
therefore the ratio @xmath74 is of the order of 1015 , implying @xmath49 between 0.4 and 0.45 .
however , the observed steepening was larger than the value ( 3/4 ) derived above .
another cause of steepening can be the lateral spreading of the fireball , as suggested by rhoads ( 1997 ) , when @xmath75 ( but see panaitescu & 1999 , who suggest that this phase should occur later ) .
in addition , some steepening of the lightcurve decay may be due to a curved synchrotron spectrum : in fact at early times the spectral index derived on the basis of @xmath76 photometric observations , de reddened with @xmath77 , was flat ( @xmath78 , 21.5 hours after the burst , stanek et al .
such a flat spectral index in the optical band must necessarily steepen at higher frequencies , to limit the emitted power .
an estimate of such a steepening will come from the analysis of the x ray afterglow flux , observed from 8 to 44.5 hours after the burst ( kuulkers et al .
1999 ) .
we conclude that the observed steepening of the lightcurve decay may then be the combined result of a curved synchrotron spectrum and of a collimated fireball . in this scenario , the spectral index of the late ( after 5 days )
optical spectrum should be @xmath79 1.11.2 .
we have proposed a model to derive the amount of linear polarization observable from a collimated fireball .
we have shown that some degree of polarization can be observed even if the magnetic field is completely tangled in the plane of the fireball , as long as the fireball is observed slightly off axis .
one of the main virtues of the proposed model is in its easily testable predictions : i ) the lightcurve of the degree of polarization has two maxima ; ii ) the observable polarization position angles are fixed between the first and the second maximum , being orthogonal between each other ; iii ) there is a strong link with the lightcurve of the total flux .
the degree of polarization is predicted to be moderate , reaching 10% only if we are observing a collimated fireball at its edge , and only for a short period of time
. a larger degree of polarization would then suggest that the magnetic field is not completely tangled in the plane orthogonal to the line of sight , as suggested by gruzinov & waxman ( 1999 ) . up to
now very few attempts have been done to measure linear polarization in optical afterglows , and is therefore premature to draw any firm conclusion from the upper limit detected in grb 990123 ( hjorth et al .
1999 ) and from the positive detection in the case of grb 990510 ( covino et al .
1999 , wijers et al .
note however that the lightcurve behavior of grb 990510 matches our predictions , as well as the constant position angle of the observed polarization .
at the time of the two first polarization measurements of grb 990510 ( made two hours apart ) the lightcurve of the total flux was decaying as @xmath80 ( stanek et al .
1999a ) , i.e. it was already steepening , in agreement with our model . unfortunately , the polarization value measured one day later ( wijers et al .
1999 ) was not precise enough to further constrain the proposed scenario .
it will be very interesting to explore in the future the association of a gradually steepening lightcurve and the presence of polarization .
we can not exclude the possibility that some of the already observed afterglows can indeed be fitted by steepening power laws , but that the lack of data makes such an attempt meaningless .
in addition , there are some optical afterglows ( e.g. grb 980326 , groot et al . , 1998 ;
grb 980519 , halpern et al . , 1999 ) which showed a rapid decay . in these cases
we may have observed only that part of the lightcurve corresponding to @xmath81 . from these considerations
we conclude that grb 990510 may not be unique in its category and that a large fraction of gamma ray burst afterglows can have some degree of optical linear polarization .
99 angel , j.r.p .
& stockman , h.s . , 1980 , ara&a , 18 , 321 covino , s. et al . , 1999 , a&a in press galama , t.j .
et al . , 1999 ,
gcn 313 groot , p.j .
et al . , 1998 ,
apj , 502 , l123 gruzinov , a. & waxman , e. 1999 , apj , 511 , 852 ( gw99 ) gruzinov , a. , 1999 , apj submitted ( astro ph/9905272 ) halpern , j.p . , kemp , j. , piran , t. & bershady , m.a .
, 1999 , apj submitted ( astro ph/9903418 ) harrison , f.a . ,
et al . , 1999 , apj submitted ( astro ph/9905306 ) hjorth , j. et al . , 1999 ,
science , 283 , 2073 kuulkers , e. et al .
, 1999 , gcn 326 impey , c.d . &
tapia , s. , 1990 , apj , 354 , 124 israel , g.l . , et al . , 1999 ,
a&a in press ( astro ph/9906409 ) laing , r.a .
1980 , mnras , 193 , 439 ( l80 ) medvedev , m.v . &
loeb , a. , 1999 , apj in press ( astro ph/9904363 ) , p. , 1999 nuclear phys .
b in press ( astro ph/9904038 ) mszros , p. & rees , m.j .
, 1997 , apj , 476 , 232 mszros , p. & rees , m.j . , 1999 , mnras submitted ( astro ph/9902367 ) moderski , r. , sikora , m. & bulik , t. , 1999 , apj submitted ( astro ph/9904310 ) panaitescu , a. & mszros , p. , 1999
, apj submitted ( astro ph/9806016 ) piran , t. , 1999 , physics reports , in press ( astro ph/9810256 ) rhoads , j.e . , 1997 , apj , 487 , l1 rhoads , j.e . , 1999 , apj submitted ( astro ph/9903399 ) stanek , k.z . , garnavich , p.m. , kaluzny , j. , pych , w. , thompson , i. 1999a , gcn 318 stanek , k.z .
, garnavich , p.m. , kaluzny , j. , pych , w. , thompson , i. 1999b , apj submitted ( astro ph/9905304 ) wijers , r.a.m.j . , et al . 1999 , apj submitted ( astro ph/9906346 ) | the recently detected linear polarization in the optical lightcurve of grb 990510 renewed the interest on how polarization can be produced in gamma ray burst fireballs . here
we present a model based on the assumption that we are seeing a collimated fireball , observed slightly off axis .
this introduces some degree of anisotropy , and makes it possible to observe a linearly polarized flux even if the magnetic field is completely tangled in the plane orthogonal to the line of sight .
we construct the lightcurve of the polarization flux , showing that it is always characterized by two maxima , with the polarization position angle changing by @xmath0 between the first and the second maximum .
the very same geometry here assumed implies that the total flux initially decays in time as a power law , but gradually steepens as the bulk lorentz factor of the fireball decreases . #
1#1 gamma rays : bursts polarization radiation mechanisms : non thermal |
heavy - fermion systems @xcite are characterized by a hierarchy of distinctive energy scales @xcite .
the kondo scale , @xmath1 with bandwidth @xmath2 and superexchange @xmath3 , marks the screening of local magnetic moments .
this screening is a many - body effect which entangles the spins of the conduction electrons and local moments @xcite . below the coherence temperature , which is believed to track the kondo scale @xcite , the paramagnetic ( pm ) heavy - fermion liquid
@xcite emerges and corresponds to a coherent , bloch like , superposition of the screening clouds of the individual magnetic moments . even in the kondo limit ,
where charge fluctuations of the impurity spins are completely suppressed , this paramagnetic state is characterized by a large fermi surface with luttinger volume including both the magnetic moments and conduction electrons @xcite .
the coherence temperature of this metallic state is small or , equivalently , the effective mass large .
+ kondo screening competes with the ruderman - kittel - kasuya - yosida ( rkky ) interaction , which indirectly couples the local moments via the magnetic polarization of the conduction electrons .
the rkky energy scale is set by @xmath4 where @xmath5 corresponds to the spin susceptibility of the conduction electrons @xcite .
+ the competition between kondo screening - favoring paramagnetic ground states - and the rkky interaction - favoring magnetically ordered states - is at the heart of quantum phase transitions @xcite , the detailed understanding of which is still under debate ( for recent reviews see ref . ) . + here , two radically different scenarios have been put forward to describe this quantum phase transition . in the _ standard _ hertz - millis picture @xcite ,
the quasi - particles of the heavy - fermion liquid remain intact across the transition and undergo a spin - density wave transition . in particular , neutron scattering experiments of the heavy - fermion system @xmath6
show that fluctuations of the antiferromagnetic order parameter are responsible for the magnetic phase transition and that the transition is well understood in terms of the hertz - millis approach @xcite .
+ on the other hand , since many experimental observations such as the almost wave vector independent spin susceptibility in @xmath7 @xcite , or the jump in the low - temperature hall coefficient in @xmath8 @xcite are not accounted for by this theory alternative scenarios have been put forward @xcite . in those scenarios ,
the quantum critical point is linked to the very breakdown of the quasi - particle of the heavy - fermion state @xcite , and a topological reorganization of the fermi surface across the transition is expected @xcite .
+ recent experiments on @xmath9 @xcite or @xmath10 @xcite show that a change in fermi surface ( fs ) topology must not necessarily occur only at the magnetic order - disorder quantum critical point ( qcp ) .
in fact , even in ybrh@xmath11si@xmath11 it has since been shown that the fermi surface reconstruction can be shifted to either side of the qcp via application of positive or negative chemical pressure @xcite . in this paper , we address the above questions through an explicit calculation of the fermi surface topology in the framework of the kondo lattice model ( klm ) . in its simplest form
the klm describes an array of localized magnetic moments of spin @xmath12 , arising from atomic @xmath13-orbitals , that are coupled antiferromagnetically ( af ) via the exchange interaction @xmath3 to a metallic host of mobile conduction electrons .
+ we present detailed dynamical cluster approximation ( dca ) calculations aimed at the investigation of the klm ground state . for the simulations within the magnetically ordered phase ,
we have extended the dca to allow for symmetry breaking antiferromagnetic order .
we map out the magnetic phase diagram as a function of @xmath14 and conduction electron density @xmath15 , with particular interest in the single - particle spectral function and the evolution of the fermi surface .
the outline is as follows .
the model and the dca implementation is discussed in sec .
[ sec : section2 ] .
results for the case of half - band filling and hole - doping are discussed in sec .
[ sec : section3 ] and [ sec : section4 ] .
section [ sec : section5 ] is devoted to a summary .
this paper is an extension to our previous work , where part of the results have already been published @xcite .
the kondo lattice model ( klm ) we consider reads @xmath16 the operator @xmath17 denotes creation of an electron in a bloch state with wave vector @xmath18 and a z - component of spin @xmath19 .
the spin @xmath12 degrees of freedom , coupled via @xmath20 , are represented with the aid of the pauli spin matrices @xmath21 by @xmath22 and the equivalent definition for @xmath23 using the localized orbital creation operators @xmath24 .
the chemical potential is denoted by @xmath25 .
the definition of the klm excludes charge fluctuations on the @xmath13-orbitals and as such a strict constraint of one electron per localized @xmath13-orbital has to be included . for an extensive review of this model we refer the reader to ref . .
+ particle - hole symmetry at half - filling is given if hopping is restricted to nearest neighbors on the square lattice and the chemical potential is set to zero .
we introduce a next - nearest neighbor hopping with matrix element @xmath26 to give a modified dispersion @xmath27 -2 t ' \left [ \cos(k_{x}+k_{y } ) + \cos(k_{x}-k_{y})\right]$ ] .
as we will see , the suppression of particle - hole symmetry due to a finite value of @xmath26 leads to dramatic changes in the scaling of the quasi - particle gap at low values of @xmath14 and at half - band filling .
we have considered the value @xmath28 .
this choice guarantees that the spin susceptibility of the host metallic state at small dopings away from half - filling is peaked at wave vector @xmath29 .
thus , antiferromagnetic order as opposed to an incommensurate spin state is favored .
+ to solve the model we have used the dca @xcite approach which retains _ temporal _ fluctuations and hence accounts for the kondo effect but neglects spatial fluctuations on a length scale larger than the cluster size .
the approach relies on the coarse - graining of momentum space , and momentum conservation holds only between the @xmath18-space patches . by gradually defining smaller sized patches the dca allows to restore the @xmath18-dependency of the self - energy .
+ the standard formulation of dca is naturally translationally invariant . to measure orders beyond translation invariance
the dca can be generalized to lattices with a supercell containing @xmath30-unit cells of the original lattice @xcite .
a general unit cell is addressed by @xmath31 , @xmath32 denoting the supercell and @xmath33 the relative points with @xmath34 . in this work we opted for @xmath35 which is the minimum requirement for the formation of antiferromagnetic order . the reduced brillouin zone ( rbz )
is then spanned by @xmath36 and @xmath37 and we have set the lattice constant to unity .
this supercell as well as the rbz ( also coined magnetic brillouin zone ( mbz ) ) are plotted in fig .
[ fig : fig1 ] . + the self - energy and green function are to be understood as spin dependent matrix functions , with an index for the unit cell within the supercell as well as an orbital index for the @xmath38- and @xmath13-orbitals in each unit cell .
the @xmath18-space discretization into @xmath39 patches with momentum conservation only between patches yields the coarse - grained lattice green function @xmath40 and ensures that the self - energy is only dependent on the coarse - grained momentum @xmath41 : @xmath42 . here
, the reciprocal vector @xmath43 denotes the center of a patch and the original @xmath18 vectors are given by @xmath44 .
the self - energy is extracted from a real - space cluster calculation with periodic boundaries yielding the quantized @xmath43 values .
let @xmath45 be the bath ( non - interacting ) green function of the cluster problem and @xmath46 the full cluster green function .
hence , @xmath47 and self - consistency requires that : @xmath48 + and @xmath49 connect the af unit cells .
the two inequivalent @xmath38-orbitals are denoted by filled ( empty ) circles and the localized @xmath13-orbitals with arrows .
the reciprocal - space lattice vectors @xmath50 and @xmath51 span the magnetic brillouin zone .
the mbz is uniformly discretized in @xmath52 patches , corresponding to real - space clusters each containing @xmath39 magnetic unit cells . the color coding refers to constant momentum dependency of the self - energy and cluster green function .
the dashed square denotes the extended brillouin zone . ]
the non - interacting lattice green function is denoted by @xmath53 .
+ to summarize , our implementation of dca approximates the fourier space of a lattice with two - point basis by patching . regarding the klm this leads to real - space clusters of size @xmath39 , each encompassing @xmath54 @xmath38- and @xmath13-orbitals .
in the present work cluster sizes @xmath55 and @xmath52 are considered ( fig .
[ fig : fig1 ] ) .
+ the lattice green function @xmath56 with @xmath57 can equally be expressed as @xmath58 in an extended brillouin zone scheme , with @xmath59 and @xmath60 : @xmath61_{\mu \nu}\;. \label{eqn : eqn4}\end{aligned}\ ] ] in order to implement the dca one has to be able to solve , for a given bath , the kondo model on the effective cluster .
the method we have chosen is the auxiliary field quantum monte carlo ( qmc ) version of the hirsch - fye algorithm , following precisely the same realization as in ref . .
the implementation details of the self - consistency cycle can be found in ref . .
the performance of the qmc cluster solver has been enhanced by almost an order of magnitude by implementing the method of delayed updates @xcite . during the qmc markov process
the full updates of the green function are delayed until a sufficiently large number of local changes in the auxiliary field have been accumulated .
this enhances the performance of the whole qmc algorithm since considerable cpu time is spent in the updating section . to efficiently extract spectral information from the imaginary time discrete qmc data a stochastic analytic continuation scheme
is employed @xcite .
+ ground state properties of the model are accessed by extrapolation of the inverse temperatures @xmath62 to infinity .
this is a demanding task since all relevant energy scales , the kondo scale , the coherence scale and the rkky scale , become dramatically smaller with decreasing @xmath14 . in the paramagnetic phase and below the coherence
scale a clear hybridization gap is apparent in the single particle spectral function and we use this criterion to estimate the coherence temperature . since the computational time required by the qmc cluster solver @xcite increases proportional to @xmath63 , this limits us in the resolution of energy scales to values of @xmath64 .
it is important to note that for small dopings away from half - band filling and for the considered cluster sizes , the negative sign problem is not severe and hence is not the limiting factor .
we consider the klm at half - filling in two cases : either with or without particle - hole symmetry .
that is @xmath65 and @xmath66 respectively . at @xmath67 lattice qmc simulations
do not suffer from the negative sign problem , and it is well established that the kondo screened phase gives way to an af ordered phase at @xmath68 @xcite . this magnetic order - disorder transition occurs between insulating states and it is reasonable to assume that it belongs to the three - dimensional @xmath69 universality class @xcite . here , the dynamical exponent takes the value of unity such that the correlation lengths in imaginary time and in real space are locked in together and diverge at the critical point .
due to the very small cluster sizes considered in the dca it is clear that we will not capture the physics of this transition .
in fact as soon as the correlation length exceeds the size of the dca cluster , symmetry breaking signaled by a finite value of the staggered moment @xmath70 sets in and mean - field exponents are expected .
the normalization is chosen such that the staggered magnetization of the fully polarized state takes the value of unity . the above quantity is plotted in fig .
[ fig : fig2 ] at @xmath65 and @xmath55 as apparent , the critical value of @xmath14 overestimates ( @xmath71 . )
the _ exact _ lattice qmc result .
switching off nesting by including a finite value of @xmath72 shifts the magnetic order - disorder transition to lower values of @xmath14 .
+ to improve on this result , we can systematically enhance the cluster size . however , our major interest lies in the single - particle spectral function . as we will see below , and well within the magnetically ordered phase , this quantity compares very well with the _ exact _ lattice qmc results . of the local moment spins in the ground state as a function of coupling @xmath14 with next - nearest neighbor hopping @xmath73 and @xmath74 , respectively .
the low temperature limit is reached by performing simulations at various inverse temperatures @xmath62 .
convergence to the ground state has been achieved for the following inverse temperatures : @xmath75 ( @xmath76 ) , @xmath77 ( @xmath78 ) , @xmath79 ( @xmath80 ) and @xmath81 ( @xmath82 ) . ]
we have calculated the single - particle spectral function @xmath83 $ ] of the conduction electrons along a path of high symmetry in the extended brillouin zone . in all our spectral plots , on the energy axis , @xmath84 values are given relative to the chemical potential @xmath25 .
+ in fig .
[ fig : fig3 ] we plot the spectrum for @xmath85 which was seen to be in the paramagnetic phase ( see fig .
[ fig : fig2 ] ) .
the lower band runs very flat around @xmath86 , with relatively low spectral weight ( note the logarithmic scale of the color chart ) in comparison to the other parts of the band which are mostly unchanged from the non - interacting case .
this feature is associated with kondo screening of the impurity spins and the resultant large effective mass of the composite quasi - particles . since no band crosses the fermi energy ( @xmath87 ) we classify this region of parameter space as a kondo insulator .
the observed dispersion relation is well described already in the framework of the large-@xmath88 mean - field theory of the kondo lattice model @xcite . at the particle - hole symmetric point
this approximation , which in contrast to the dca+qmc results of fig .
[ fig : fig3 ] neglects the constraint of no double occupancy of the @xmath13-orbitals , yields the dispersion relation of hybridized bands : @xmath89 .
\label{eqn : eqn6}\ ] ] within the mean - field theory the quasi - particle gap ( see eq .
[ eqn : eqn7 ] ) , @xmath90 , tracks the kondo scale . + at half - filling ( @xmath91 ) and with particle - hole symmetry ( @xmath74 ) close to the magnetic phase transition on the paramagnetic side , @xmath85 . ] at @xmath92 we have measured a non - vanishing staggered magnetization @xmath93 . the spectral function ( fig .
[ fig : fig4 ] ) for this lightly af ordered simulation now includes additional low - energy band structures : in the upper band around @xmath86 and in the lower band around @xmath94 .
these _ shadow bands _ arise due to the scattering of the heavy quasi - particle off the magnetic fluctuations centered at wave vector @xmath95 . + at half - filling ( @xmath91 ) and with particle - hole symmetry ( @xmath74 ) close to the magnetic phase transition on the af side , @xmath92 ( a ) and closeup ( b ) .
the onset of magnetic order and concomitant emergence of the mbz is signaled by the appearance of shadow bands .
the staggered magnetization for this point is measured to be @xmath93 . ] at @xmath96 we make a comparison with the spectrum obtained via qmc lattice simulations using the projective auxiliary field algorithm of ref .
for the same parameters .
we see in fig .
[ fig : fig5 ] the qmc lattice simulation results @xcite on the left and our dca result on the right , both only for the photoemission spectrum @xmath97 .
the agreement of the dca spectrum with the blankenbecler - scalapino - sugar ( bss ) result is excellent . as in the bss lattice calculations and for the particle - hole symmetric case , the conduction ( valence ) band maximum ( minimum )
is located at @xmath98 ( @xmath99 ) for all considered values of @xmath14 .
+ with the @xmath100 spectral function . data ( a ) taken from ref .
which used a projective auxiliary field monte carlo ( bss ) method for the klm on a @xmath101 lattice and our dca results ( b ) at @xmath77 , both for the same parameter set - @xmath74 , @xmath102 , @xmath96.,title="fig : " ] + it is noteworthy that already with the smallest possible cluster capable of capturing af order , @xmath55 , we are able to produce a single - particle spectrum which is essentially the same as the lattice qmc result .
this confirms that the dca is indeed a well suited approximation for use with the klm : the essence of the competition between rkky - mediated spacial magnetic order and the time - displaced correlations responsible for kondo screening is successfully distilled to a small cluster dynamically embedded in the mean - field of the remaining bath electrons .
we now set @xmath73 , but remain at half - filling via careful adjustment of the chemical potential @xmath25 . with decreasing @xmath14 we can divide the phase diagram into four regions on the basis of the characteristic spectrum in each region . for @xmath103 ( fig .
[ fig : fig6](a ) ) the spectrum is qualitatively identical to the pm phase in the @xmath74 case such that the model is a kondo insulator with an indirect gap .
the minimum of the valence band lies at @xmath94 and the maximum of the conduction band is at @xmath86 .
+ for @xmath104 ( fig .
[ fig : fig6](b ) ) the system is weakly af ordered , and the spectrum reflects this in the development of shadow bands , but is otherwise qualitatively the same as before . at @xmath105 ( fig .
[ fig : fig6](c ) ) , well inside the af phase , the shadow bands are more pronounced and now the minimum of the valence band has shifted to @xmath106 . the final plot in the series ( fig .
[ fig : fig6](d ) ) , with @xmath107 shows that the form of the lower band has also changed such that the maximum of this band now lies at @xmath108 .
+ in figs .
[ fig : fig7](a - h ) we show the evolution of this lower band between @xmath94 and @xmath86 with an enlargement of the energy axis around the fermi energy .
the local minimum energy dip at @xmath108 becomes less pronounced as the heavy - fermion band flattens until by the time we reach @xmath96 this dip has turned into a bump such that the lower band maximum has shifted from @xmath86 to @xmath108 .
this important change in band structure appears to be continuous . for smaller coupling , @xmath109 , @xmath110 , and @xmath111
the lower band maximum remains at @xmath108 and becomes more pronounced .
assuming a rigid band picture , this evolution of the band structure maps onto a topology change of the fermi surface at small dopings away from half filling .
we will see by explicit calculations at finite dopings that this topology change indeed occurs .
+ the position of the minima and maxima in the valence and conduction bands of the spectra for @xmath73 are summarized schematically in fig .
[ fig : fig8 ] . and for half - filling .
the squares represent the first brillouin zone ( for the pm phase ) or the extended brillouin zone ( for the af phase ) , with the bottom left corner and top right corner of each square given by @xmath112 and @xmath86 , respectively.,title="fig : " ] + the quasi - particle gap corresponds to the energy difference between the top of the conduction band and the bottom of the valence band . to be more precise , @xmath113 here , @xmath114 corresponds to the ground state energy of the @xmath115 particle system with total momentum @xmath18 and @xmath116 .
technically , the energy differences are at best extracted from the low temperature single - particle green functions : @xmath117 as a function of kondo coupling @xmath14 . at @xmath118 , the magnetic order - disorder transition takes place at @xmath119 . in the coupling range @xmath120 kondo screening coexists with magnetism and is at the origin of the quasiparticle gap . ] in the large @xmath14 limit , each impurity spin traps a conduction electron in a kondo singlet .
the wave function corresponds to a direct product of such kondo singlets and the quasi - particle gap is set by the energy scale @xmath121 required to break a kondo singlet .
hence , in the large @xmath14 limit the quasi - particle gap does not depend on the details of the band structure .
the situation is more subtle in the weak - coupling limit , where the underlying nesting properties of the fermi surface play a crucial role . assuming static impurity spins locked into an antiferromagnetic order , @xmath122 one obtains the single particle dispersion relation @xmath123 with @xmath124 . in one dimension and at @xmath74
nesting leads to @xmath125 and magnetic order opens a quasi - particle gap set by @xmath126 .
both the dca results presented in fig . [
fig : fig9 ] as well as the bss results of ref .
support this point of view . in one - dimension ,
this scaling of the quasi - particle gap is also observed @xcite . at @xmath66 nesting
is not present and magnetic ordering can only partially gap the fermi surface . since the dca results of fig .
( [ fig : fig9 ] ) support a finite quasi - particle gap in the magnetically ordered phase , the above frozen - spin ansatz fails and points to one of the main result of our work , namely that within the magnetically ordered phase kondo screening and the heavy quasi - particles are still present .
even though we can not track the functional form of the gap at @xmath127 and at weak couplings , we interpret the dca results in terms of a quasi - particle gap which tracks the kondo scale from weak to strong coupling as in the large-@xmath88 mean - field calculations @xcite .
in this section we concentrate on the hole - doped klm at @xmath66 .
we will first map out the magnetic phase diagram in the doping versus coupling plane and then study the evolution of single - particle spectral function from the previously discussed half - filled case to the heavily doped paramagnetic heavy - fermion metallic state . of the @xmath13-electrons as a function of @xmath15 at different constant couplings @xmath14 . ]
[ fig : fig10 ] shows the staggered magnetization as a function of conduction electron density for @xmath82 , @xmath110 , @xmath128 and @xmath129 .
the magnetically ordered state found at half - filling initially survives when doping with holes . at all coupling values a continuous magnetic phase transition
is observed with the af order decreasing gradually as the system is doped and vanishing smoothly at a quantum critical point .
the results are summarized in the magnetic phase diagram shown in fig .
[ fig : fig11 ] .
with decreasing values of @xmath14 and increasing dopings the rkky interaction progressively dominates over the kondo scale and the magnetic metallic state is stabilized .
checks were made by varying the temperature , @xmath130 , of the simulations to ensure that the results can be considered to be ground state .
below @xmath131 we found the coherence scale to be too low to guarantee convergence .
as pointed out previously , the limiting computational factor is the cubic scaling of the hirsch - fye @xcite algorithm , @xmath132 and not the negative sign problem .
( color - coded ) as a function of coupling @xmath14 and conduction electron occupancy @xmath15 .
triangles : pm region , large fs .
squares : af , large fs . circles : af , small fs . here @xmath66 and the calculations
are carried out with the @xmath133 cluster .
below , the fs topologies corresponding to the numbered regions are shown schematically.,title="fig : " ] + we have calculated and followed the evolution of the single - particle spectrum , plotted in an extended brillouin zone scheme . to begin with we set @xmath80 and show results in fig .
[ fig : fig12 ] for the spectral function , as it is here that a change in the fermi surface topology becomes evident .
starting from the previously discussed half - filled case ( fig .
[ fig : fig12](a ) ) a rigid band approximation produces hole pockets around the @xmath134 points in the brillouin zone .
this is confirmed by explicit calculations at small dopings away from half - filling as demonstrated in fig .
[ fig : fig12](b ) . with further doping
this low energy band flattens out progressively becoming almost flat by @xmath135 ( fig .
[ fig : fig12](d ) ) .
doping further gives rise to a fermi surface with holes centered around @xmath136 .
since at @xmath137 ( fig .
[ fig : fig12](e ) ) and @xmath138 ( fig .
[ fig : fig12](f ) ) we still have non - zero magnetizations of @xmath139 and @xmath140 , respectively , shadow features centered around @xmath94 are expected .
the weight of those shadow features progressively diminishes as the staggered magnetization vanishes .
in particular , in fig . [
fig : fig12](g ) the magnetization is very small ( @xmath141 ) and indistinguishable from zero in the final figure of the series . + across the transition from @xmath142 to @xmath143 the quasiparticle weight @xmath144 as obtained from the behaviour of @xmath145 at large imaginary times decreases albeit it shows no sign of singularity .
this observation stands in agreement with results obtained at half - band filling @xcite and excludes the occurrence of a kondo breakdown . +
this evolution of the single - particle spectral function at @xmath146 points to three distinct fermi surface topologies , sketched in fig .
[ fig : fig11 ] . in the paramagnetic phase ( fermi surface ( 1 ) in fig .
[ fig : fig11 ] ) the fermi surface consists of hole pockets around the @xmath147 points in the brillouin zone .
even though in our simulations charge fluctuations of the @xmath13-sites are completely prohibited , the fermi surface volume accounts for both the conduction electrons and impurity spins .
this fermi surface topology maps onto that of the corresponding non - interacting periodic anderson model with total particle density given by @xmath148 and is coined large fermi surface . in the antiferromagnetic metallic phase close to the magnetic order - disorder transition , ( fermi surface ( 2 ) in fig .
[ fig : fig11 ] ) the fermi surface merely corresponds to a backfolding of the paramagnetic fermi surface as expected in a generic spin - density wave transition . here
, a heavy quasi - particle with crystal momentum @xmath18 can scatter off a magnon with momentum @xmath149 to produce a shadow feature at @xmath150 .
it is only within the magnetically ordered phase that we observe the topology change of the fermi surface to hole pockets centered around @xmath151 . due to the antiferromagnetic order and the accompanied reduced brillouin zone
, this fermi surface topology satisfies the luttinger sum rule .
we coin it a small fermi surface in the sense that it is linked to that obtained in a frozen @xmath13-spin mean - field calculation as presented in eq .
( [ eqn : eqn9 ] ) . of the conduction electrons for @xmath105 and @xmath73 , at @xmath152 .
the shadow features present in ( a ) vanish as the occupation number @xmath15 is reduced . ]
we have equally plotted the single - particle spectral function at higher values of @xmath105 as a function of doping ( see fig . [
fig : fig13 ] ) . for this coupling strength , and at half - band filling ,
the conduction band maximum is located at @xmath86 with accompanying shadow bands at @xmath94 .
the fermi surface obtained upon doping follows from a rigid band approximation , and yields the topology shown in fig .
[ fig : fig11 ] ( 2 ) .
the transition to the paramagnetic state shows up in the vanishing of the shadow features at @xmath153 .
+ we note that our results are confirmed by simulations carried out using the larger cluster size @xmath52 , see fig .
[ fig : fig14 ] .
in particular , we observe the small fermi surface topology in the strong af region with small coupling @xmath14 and evidence for a large fermi surface in the weakly ordered region ( @xmath154 ) . of the conduction electrons obtained using cluster size @xmath52 . confirming that the small fermi surface topology in the lightly doped af phase with coupling @xmath80 is not an artifact of the smaller cluster results .
the staggered magnetisation of the @xmath13-electrons is @xmath155 . ]
we resort to a mean - field modelling in order to propose a scenario for the detailed nature of the topological transition of the fs within the ordered phase . within this model
the topology change will correspond to two lifshitz transitions .
aspects of the single - particle spectral function can be well accounted for within the following mean - field hamiltonian @xcite : @xmath156 here @xmath157 , ( @xmath25 ) are lagrange multipliers fixing the @xmath13- ( @xmath38- ) particle number to unity ( @xmath15 ) and @xmath158 , @xmath159 corresponds to the staggered magnetization of the conduction and @xmath13-electrons .
the order parameter for kondo screening is @xmath160 .
the large-@xmath88 mean - field saddle point corresponds to the choice @xmath161 and @xmath162 .
as already mentioned this saddle point gives a good account of the hybridized bands we observe numerically in the paramagnetic phase . in the magnetic phase
, one could speculate that kondo screening breaks down such that the @xmath13-spins are frozen and do not participate in the luttinger volume .
this corresponds to the parameter set @xmath163 , @xmath164 and @xmath165 and leads to the dispersion relation of eq .
( [ eqn : eqn8 ] ) . throughout the phase diagram of fig .
[ fig : fig11 ] the single - particle spectral function never shows features following this scenario . in fact , many aspects of the spectral function in the magnetically ordered phase can be understood by choosing @xmath166 , @xmath167 and @xmath168 .
this explicitly accounts for a heavy - fermion band in the ordered phase .
it is in this sense that we claim the absence of kondo breakdown within our model calculations .
[ fig : fig15 ] plots the four - band energy dispersion relation @xmath169 obtained from the mean - field calculation in the magnetically ordered phase at @xmath73 , @xmath80 and at band fillings @xmath170 ( fig .
[ fig : fig15](a , b ) ) and @xmath171 ( fig .
[ fig : fig15](c , d ) ) corresponding to the fermi surface topologies labeled ( 2 ) and ( 3 ) , respectively , in fig .
[ fig : fig11 ] . in each case
we have used the dca value for the magnetization and varied @xmath172 to obtain the best qualitative fit to the respective dca spectrum .
the change in the topology of the fermi surface stems from a delicate interplay of the relative magnitudes of the magnetization @xmath173 and _ kondo screening _
parameter @xmath172 . + with @xmath80 ( @xmath73 ) .
( a ) @xmath174 , @xmath170 .
the line intensity is proportional to the quasi - particle spectral weight , @xmath175 .
the fermi surface topology ( see closeup ( b ) ) results from backfolding of the large-@xmath88 mean - field bands .
this corresponds to the fermi surface ( 2 ) in fig.[fig : fig11 ] .
+ ( d ) @xmath176 , @xmath171 .
the fermi surface topology ( see closeup ( c ) ) corresponds to that of hole pockets centered around @xmath177 as depicted by the fermi surface ( 3 ) in fig .
[ fig : fig11 ] . ] in the dca calculations , we are unable to study the precise nature of the topology change of the fermi surface within the magnetically ordered phase since the energy scales are too small .
however we can draw on the mean - field model to gain some insight .
[ fig : fig16 ] plots the dispersion relation along the @xmath94 to @xmath86 line in the brillouin zone as a function of @xmath172 .
we would like to emphasize the following points .
i ) the transition from hole pockets around the wave vector @xmath178 to @xmath179 corresponds to two lifshitz transitions in which the hole pockets around @xmath178 disappears and those around @xmath180 emerge .
the luttinger sum rule requires an intermediate phase with the presence of pockets both at @xmath180 and at @xmath178 .
this situation is explicitly seen in fig .
[ fig : fig16](b ) at @xmath181 .
ii ) starting from the antiferromagnetic state with hole pockets at @xmath177 there is an overall flatting of the band prior to the change in the fermi surface topology .
hence on both sides of the transition , an enhanced effective mass is expected . owing to point i )
the effective mass does not diverge .
iii ) the two lifshitz transition scenario suggested by this mean - field modeling , leads to a continuous change of the hall coefficient . however , since a very _ small _ variation of the hybridization @xmath172 suffices to change the fermi surface topology one can expect , as a function of this control parameter , a rapid variation of the hall coefficient as it is observed in experiment @xcite . of the dca results ( @xmath80 , @xmath73 @xmath182 , @xmath183 , @xmath170 ) .
as apparent on the scale of the closeup ( b ) , hole pockets around @xmath180 as well as around @xmath178 are present at @xmath184 . ]
the evolution of the fermi surface across the magnetic order - disorder transition has been investigated by other groups and especially in the framework of a gutzwiller projected mean - field wave function corresponding to the ground state of the single particle hamiltonian of eq .
[ eqn : eqn9 ] . both a variational quantum monte carlo calculation @xcite as well as a gutzwiller approximation @xcite show a rich phase diagram which bears some similarities but also important differences with the present dca calculation . at small doping away from half - filling ,
the magnetic transition as a function of the coupling @xmath14 is continuous and of sdw type as marked by the back - folding of the fermi surface . within the magnetically ordered phase a first order transition occurs between hole ( af@xmath185 ) and electron ( af@xmath186 ) like fermi surfaces . at larger hole dopings ,
the magnetic transition is of first order and the fermi surface abruptly changes from large to small ( af@xmath186 ) .
the calculations in refs . are carried out at @xmath187 . at finite values of @xmath188
one expects the af@xmath186 fermi - surface topology to correspond to hole pockets centered around @xmath189 as shown in fig .
[ fig : fig12 ] . on the other hand ,
the af@xmath185 fermi surface arises from a back - folding of the large fermi surface ( fig .
[ fig : fig13 ] ) . with this identification ,
our dca results bear some similarity with the variational calculation in the sense that the same fermi surface topologies are realized .
however , in our calculations we see no sign of first order transitions and no direct transitions from the large paramagnetic fermi - surface topology to the af@xmath186 phase .
it is equally important to note that in both the variational and dca approaches no kondo breakdown is apparent .
in particular the variational parameter @xmath190 which encodes hybridization between the @xmath13- and @xmath38-electrons never vanishes @xcite . on the other hand , @xmath100 cdmft calculations of the three - dimensional periodic anderson model on a two site cluster and finite sized baths @xcite have put forward the idea that the magnetic order - disorder transition is driven by an orbital selective mott transition or in other words a kondo breakdown @xcite .
in particular even in the presence of spin symmetry broken baths allowing for antiferromagnetic ordering a big enhancement of the @xmath13-effective mass is apparent in the vicinity of the magnetic transition . in the paramagnetic phase ,
the low energy decoupling of the @xmath13- and @xmath38-electrons stem from the vanishing of the hybridization function at low energies .
our results , which makes no approximation of the number of bath degrees of freedom and which allow to treat larger cluster sizes do not support this point of view .
in particular as shown in ref .
no singularity in the quasiparticle weight is apparent across the transition .
equally , fig .
[ fig : fig12 ] which plots the single particle spectral function at finite doping shows no singularity in @xmath191 .
we have presented detailed dca+qmc results for the kondo lattice model on a square lattice .
our dca approach allows for antiferromagnetic order such that spectral functions can be computed across magnetic transitions and in magnetically ordered phases . at the particle - hole symmetric point ( half - band filling and @xmath67 )
we can compare this approach to previous _
lattice qmc auxiliary field simulations @xcite .
unsurprisingly , with the _ small _ cluster sizes considered in the dca approach the critical value of @xmath14 at which the magnetic order - disorder transition occurs is substantially overestimated .
the important point however is that the dca+qmc approximation gives an extremely good account of the single - particle spectral function both in the paramagnetic and antiferromagnetic phases .
hence , and as far as we can test against benchmark results , the combination of static magnetic ordering and dynamical kondo screening , as realized in the dca , provides a very good approximation of the underlying physics . + as opposed to _ exact _
lattice qmc auxiliary field simulations , which fail away from the particle - hole symmetric point due to the so - called negative sign problem , the dca+qmc approach allows us to take steps away from this symmetry either by doping or by introducing a finite value of @xmath26 .
it is worthwhile noting that for cluster sizes up to @xmath0 orbitals the limiting factor is the cubic scaling of the hirsch - fye algorithm rather than the negative sign problem which turns out to be very mild in the considered parameter range .
+ our major findings are summarized by the following points .
+ i ) we observe no kondo breakdown throughout the phase diagram even deep in the antiferromagnetic ordered state where the staggered magnetization takes large values .
this has the most dramatic effect at half - band filling away from the particle - hole symmetric point and at small values of @xmath14 where magnetic order is robust . here
, magnetism alone will not account for the insulating state and the observed quasi - particle gap can only be interpreted in terms of kondo screening .
the absence of kondo breakdown throughout the phase diagram is equally confirmed by the single - particle spectral function which always shows a feature reminiscent of the heavy - fermion band.@xcite + ii ) the transition from the paramagnetic to the antiferromagnetic state is continuous and is associated with the backfolding of the large heavy fermion fermi surface .
+ iii ) within the antiferromagnetic metallic phase there is a fermi surface topology change from hole pockets centered around @xmath86 at _ small _ values of the magnetization @xmath159 to one centered around @xmath192 at _ larger _ values of @xmath159 .
the latter fermi surface is adiabatically connected to one where the @xmath13-spins are frozen and in which kondo screening is completely absent .
@xcite this transition comes about by _ continuously deforming
_ a heavy - fermion band such that the effective mass grows substantially on both sides of the transition . within a mean - field modelling of the quantum monte carlo
results the transition in the fermi surface topology corresponds to two continuous lifshitz transitions . as a function of decreasing mean - field kondo screening parameter @xmath172 and at constant hole doping , the volume of the hole pocket at @xmath86 decreases at the expense of the increase of volume of the hole pocket at @xmath108 .
owing to the luttinger sum rule , the total volume of the hole pockets remains constant .
it is important to note that the energy scales required to resolve the nature of this transition are presently orders of magnitudes smaller than accessible within the dca with a quantum monte carlo solver .
+ + we would like to thank the dfg for financial support under the grant numbers as120/6 - 1 ( for1162 ) and as120/4 - 3 .
the numerical calculations were carried out at the lrz - munich as well as at the jsc - jlich .
we thank those institutions for generous allocation of cpu time . | we report the results of extensive dynamical cluster approximation calculations , based on a quantum monte carlo solver , for the two - dimensional kondo lattice model .
our particular cluster implementation renders possible the simulation of spontaneous antiferromagnetic symmetry breaking . by explicitly computing the single - particle spectral function both in the paramagnetic and antiferromagnetic phases
, we follow the evolution of the fermi surface across this magnetic transition .
the results , computed for clusters up to @xmath0 orbitals , show clear evidence for the existence of three distinct fermi surface topologies .
the transition from the paramagnetic metallic phase to the antiferromagnetic metal is continuous ; kondo screening does not break down and we observe a back - folding of the paramagnetic heavy fermion band . within the antiferromagnetic phase and
when the ordered moment becomes _ large _ the fermi surface evolves to one which is adiabatically connected to a fermi surface where the local moments are frozen in an antiferromagnetic order . |
challenge when determining the performance of a block error - correcting code is to measure its bit - error rate ( ber ) , which quantifies the reliability of the system . in practice ,
the ber estimation for a single code is simple , just send data and divide the errors committed among the total number of bits .
however , it would be too costly and time - consuming if a comparative between several codes is required .
mathematical software packages for encoding and decoding are very limited and restricted to specific codes and simulations would consume a huge amount of time when dealing with low bit - error rates .
for this reason , a theoretical approach to the measurement of the ber is proposed by several authors in the literature , see for instance @xcite .
all these papers follow this scheme of codification : let @xmath0 be a code of length @xmath1 and dimension @xmath2 over the field with @xmath3 elements , being @xmath4 .
an @xmath1-tuple is transmitted through a @xmath3-ary symmetric discrete memoryless channel . in this step , there are two possibilities , the transmission is right or failed in some symbols . in a second step ,
the code corrects the @xmath1-tuple , detects an erroneous transmission but does not correct it , or asserts that the transmitted @xmath1-tuple is a codeword .
finally , there is a comparison between the encoded and decoded @xmath1-tuples , see figure [ f1 ] .
@xmath5 { encoded $ n$-tuple } ; \draw ( 5,0 ) node(dos ) [ draw , shape = rectangle , rounded corners , fill = gray!20 ] { transmitted $ n$-tuple } ; \draw ( 0,-1 ) node(tres ) [ draw , shape = rectangle , rounded corners , fill = gray!20 ] { decoded $ n$-tuple } ; \draw[decorate , decoration={expanding waves , angle=5 } ] ( uno ) to node[snake = snake , midway , sloped , above , pos=0.5 ] { \small transmission } ( dos ) ; \draw [ - > ] ( dos ) to node[midway , sloped , below , pos=0.5 ] { \small decoding } ( tres ) ; \draw [ < - > , dashed ] ( uno ) to node[midway , right ] { \small comparison } ( tres ) ; \end{tikzpicture}\ ] ] when we run over all the possibilities in each step ( of course , not all combinations are possible ) , this yields five disjoint cases : 1 . a correct transmission ( ct ) , i.e. , all the symbols are correctly received .
2 . a right correction ( rc ) , i.e.
, some of the symbols are incorrectly received but the decoding algorithm corrects them .
an error detection ( ed ) , i.e. , the number of errors exceeds the error - correction capability of the code , the block is not corrected and the bad transmission is detected .
4 . a wrong correction ( wc ) , i.e.
, some errors occur ( beyond the error capability of the code ) , there is a code - correction , but nevertheless , the encoded block differs from the decoded block .
5 . a false negative ( fn ) , i.e. , some symbols are incorrectly received but the whole block is a codeword , so , from the receiver s point of view , the block is correctly received .
cases fn and wc are called undetected errors in @xcite , and it is proven that , for maximum - distance - separable ( mds ) codes , the probability of an undetected error decreases monotonically as the channel symbol error decreases , that is , mds codes are proper in the terminology of @xcite . hence , the performance of an mds code is characterized by the probability that an erroneous transmission will remain undetected . in @xcite , as a performance criterion , the probability of an ed is added and an exhaustive calculus of the word , symbol , and bit error rate of ed , wc , and fn is made . in this paper , we propose a refinement in the calculi of the probability of an fn , a wc , and an ed .
consequently , we get a better approximation of the ber for a @xmath3-ary mds code . as in the above references , we consider a bounded distance reproducing decoder , i.e.
, it reproduces the received word whenever there are uncorrectable errors .
the underlying idea consists in removing the symbol errors produced in the redundancy part of the decoded @xmath1-tuple , that is , following the nomenclature of @xcite , unlike the aforementioned papers , we estimate the information bit error rate ( iber ) , sometimes also called post - decoding bit error rate .
more formally , let us assume , without loss of generality , that the codification is systematic and the first @xmath2 symbols of the @xmath1-tuples form an information set .
hence , following the above scheme , after the comparison step , if there are @xmath6 errors , the symbol - error proportion is @xmath7 . nevertheless , some of these errors belong to the redundancy part and they will not spread in the final post - decoded @xmath2-tuple . in other words , a new variable should be considered : the comparison between the original block and the final @xmath2-tuple obtained after decoding , see figure [ f2 ] .
@xmath8 { original $ k$-tuple } ; \draw ( 0,1 ) node(uno ) [ draw , shape = rectangle , rounded corners , fill = gray!20 ] { encoded $ n$-tuple } ; \draw ( 3,0 ) node(dos ) [ draw , shape = rectangle , rounded corners , fill = gray!20 ] { transmitted $ n$-tuple } ; \draw ( 0,-1 ) node(tres ) [ draw , shape = rectangle , rounded corners , fill = gray!20 ] { decoded $ n$-tuple } ; \draw ( -3,0 ) node(cuatro ) [ draw , shape = rectangle , rounded corners , fill = gray!20 ] { target $ k$-tuple } ; \draw [ decorate , decoration={expanding waves , angle=4 } ] ( uno ) to [ bend left ] node[midway , sloped , above , pos=0.5 ] { \small transmission } ( dos ) ; \draw [ - > ] ( dos ) to [ bend left ] node[midway , sloped , below , pos=0.4 ] { \small decoding } ( tres ) ; \draw [ < - > , dashed ] ( uno ) to node[midway ] { \small comparison } ( tres ) ; \draw [ - > ] ( cero ) to[bend left ] node[midway , sloped , above ] { \small encoding } ( uno ) ; \draw [ - > ] ( tres ) to[bend left ] node[midway , sloped , below ] { \small post - decoding } ( cuatro ) ; \draw [ < - > , dashed ] ( cero ) to node[midway ] { \small comparison } ( cuatro ) ; \end{tikzpicture}\ ] ] attending to this new variable , we may split the ed into two disjoint cases : 1 . a pure error detection ( ped ) , i.e.
, some errors affect the information set .
2 . a false positive ( fp ) , i.e.
, all errors belong to the redundancy part .
so , from the point of view of the receiver , there are uncorrectable errors but , indeed , the post - decoded block is right
. then , a study of the ber should consider the probability of obtaining a false positive after the post - decoding process and , hence , the criterion for measuring the performance of the code should be the probability of undetected and ped errors . all along this paper
we shall assume that the code under consideration is a @xmath3-ary @xmath9 $ ] mds code in a bounded distance reproducing encoder , where @xmath10 for some natural @xmath11 and @xmath12 , with minimum distance @xmath13 and error - correcting capacity up to @xmath14 errors .
furthermore , we shall assume , without loss of generality , that the generator matrix is systematic in its first @xmath2 coordinates .
so since it is well known that , in an mds code , any subset of @xmath2 coordinates form an information set , see e.g. ( * ? ? ?
* theorem 5.3.4 ) .
the reorganization of these components to the first @xmath2 does not affect our calculations and makes the text much more readable .
the channel ber shall be denoted by @xmath15 , and by @xmath16 , the probability of a bit of being correctly received .
the code is @xmath17-ary , so the probability for a symbol of being correctly received is @xmath18 .
therefore , the expected probability of transition between any two symbols , the channel symbol - error rate , is the average of all possible cases , i.e. , @xmath19 .
finally , if we want to know the probability for a bit to be erroneous inside an erroneous symbol , we simply give the conditional probability @xmath20 .
we will denote by @xmath21 and @xmath22 the minimum and the maximum of a set of integers .
the paper is structured as follows . in section [ s2 ]
, we count the codewords of a @xmath3-ary mds code .
the number of codewords of a fixed weight has been calculated before ( see @xcite and @xcite ) ; nevertheless , we need a finer approach . in @xcite and @xcite a formula for the number of codewords of fixed weight with respect to a given partition
is provided . by means of arguments of linear algebra
, we shall give an equivalent formula for counting the number of codewords of a fixed weight both in the information set and in the redundancy part .
this allows us to calculate the iber of an fn . in section [ s3 ] , we shall deal with the decoding failures , that is , when the received word is corrected to a different codeword to the original one .
we shall follow the style of @xcite and calculate the words inside the sphere of a codeword of a given information and redundancy weight .
we will make use of these calculations in section [ s4 ] in order to give an expression of the probability that an fp occurs and obtain the desired approximation of the ber of an mds code . in order to make the paper self - contained , in the appendix , we add some of the combinatorial formulae needed all along the text . figures [ graphfn ] to [ graphped0 ] have been made with @xcite .
let @xmath0 be a @xmath3-ary @xmath9 $ ] mds code generated by a systematic matrix @xmath23 .
the aim of this section is to compute @xmath24 , the number of codewords in @xmath0 with @xmath25 non - zero information symbols and @xmath26 non - zero redundancy symbols , where @xmath27 and @xmath28 .
this is called the _ input - redundancy weight enumerator _ ( irwe ) in the literature ( see e.g. @xcite ) .
in fact the weight enumerator of any partition @xmath29 on @xmath30 is computed in ( * ? ? ?
* theorem 1 ) and ( * ? ? ?
* theorem 3.1 ) , hence @xmath24 can be obtained as a particular case of those theorems .
we propose a new way to compute @xmath24 involving linear algebra techniques .
we shall need the following lemmata .
[ matrixmdscfr ] any square submatrix of @xmath31 is regular .
let @xmath32 and let @xmath33 and @xmath34 be the sets of indices corresponding to the rows and columns of @xmath31 that form an @xmath35 submatrix , where @xmath36 .
we call such submatrix @xmath37 .
if @xmath38 , there exists a non - zero vector @xmath39 such that @xmath40 let @xmath41 be the vector in which @xmath42 is in the @xmath43-th coordinate for all @xmath44 and zero otherwise .
since @xmath45 , there are @xmath46 non - zero coordinates in the first @xmath2 coordinates of @xmath47 .
now , for any @xmath44 @xmath48 by , so , in the last @xmath49 coordinates of @xmath47 , there are no more than @xmath50 non - zero coordinates .
then , @xmath47 has weight @xmath49 at most .
since @xmath51 and the minimum distance of @xmath0 is @xmath13 , we get a contradiction .
consequently , @xmath37 is regular . given a system of linear equations , we say that a solution is totally non -
zero if all the coordinates of the vector are other than zero .
we say that a matrix has totally full rank if any submatrix has full rank . by lemma [ matrixmdscfr ]
, @xmath31 has totally full rank .
let us denote by @xmath52 the number of totally non - zero solutions of a homogeneous linear system , over the field @xmath53 , with @xmath25 variables and @xmath26 equations whose coefficient matrix has totally full rank .
[ fqijrecursiva ] for any integers @xmath54 , @xmath52 is given by the following recurrence : @xmath55 if @xmath56 , there are , at least , as many equations as variables .
since its coefficient matrix has full rank , the zero vector is the only solution
. then @xmath57 . if @xmath58 , since its coefficient matrix has full rank , the system is undetermined
. specifically , whenever we fix @xmath59 coordinates , we will find a single unique solution .
then , there are @xmath60 solutions whose first @xmath59 coordinates are non - zero . in order to calculate @xmath52 ,
it is enough to subtract those solutions for which some of the remaining @xmath26 coordinates are zero .
for any @xmath61 , the solutions with exactly @xmath62 coordinates being zero may be obtained choosing @xmath62 coordinates from @xmath26 , and calculating the number of totally non - zero solutions in a linear system of @xmath63 variables and @xmath26 equations , that is , @xmath64 .
therefore , @xmath65 [ fqijdirecta ] for any integers @xmath66 , @xmath67 by lemma [ fqijrecursiva ] , it is enough to check that the expression @xmath68 satisfies .
observe that if @xmath69 , then @xmath70 , since the sum runs over the empty set .
suppose that @xmath58 .
since @xmath71 for any @xmath72 , we obtain @xmath73 we substitute @xmath74 by @xmath75 in , and using , we get @xmath76 this expression is a polynomial in @xmath77 .
let us now calculate the coefficients of @xmath78 for @xmath79 .
we group those coefficients in which @xmath80 .
if @xmath81 , then @xmath82 , and therefore the coefficient of @xmath60 is @xmath83 suppose that @xmath84 .
the coefficient of @xmath78 is given by @xmath85 where the last equality comes from lemma [ diffin ] , since @xmath86 is a polynomial in @xmath62 of degree @xmath87 . because of @xmath88 , then @xmath89 .
as a result the case when @xmath90 is left to be analyzed . in such case , the coefficient of @xmath78 is given by @xmath91 where , again , the last equality follows from lemma [ diffin ] since @xmath92 is a polynomial in @xmath62 of degree @xmath93
. then , @xmath94 by , @xmath95 so @xmath96 .
we recall that @xmath24 is the number of codewords of @xmath0 with weight @xmath25 in the first @xmath2 coordinates ( the information set ) and @xmath26 , in the remaining @xmath49 coordinates ( the redundancy part ) .
[ ankijviafqij ] let @xmath97 .
if @xmath98 , then @xmath99 if @xmath100 , then @xmath101 , where @xmath102 is the kroneker s delta .
the case @xmath100 is evident because the minimum distance of @xmath0 is @xmath13 .
suppose that @xmath103 .
a codeword with weight @xmath25 in the information set and @xmath26 in the redundancy part must be obtained as a linear combination of @xmath25 rows of @xmath104 with non zero coefficients such that exactly @xmath105 coordinates of the redundancy part become zero , i.e. it is a totally non - zero solution of a homogeneous linear system with @xmath105 equation and @xmath25 variables . by lemma [ matrixmdscfr ] , this number is given by @xmath106 . however some of the solutions counted in @xmath106 can turn zero some of the remaining @xmath26 coordinates .
hence , the lemma is obtained as a consequence of the inclusion exclusion principle .
[ ankij=0 ] observe that if @xmath107 , then @xmath108 , so @xmath109 .
this agrees with the fact that the minimum weight of a non - zero codeword must be @xmath110 .
let us now give a neater description of @xmath24 . by mixing proposition [ fqijdirecta ] and lemma [ ankijviafqij ]
, we have @xmath111 we proceed analogously to the proof of proposition [ fqijdirecta ] .
let us examine the coefficient of @xmath112 in , where @xmath80 . since @xmath113 , it is necessary that @xmath114 .
we distinguish two cases . on the one hand , if @xmath115 , then @xmath116 , and then , the coefficient is given by @xmath117 where the last equality is a consequence of lemma [ chuvandermonde ] .
on the other hand , if @xmath118 , the coefficient of @xmath112 is given by @xmath119 where @xmath120 is given by lemma [ chuvandermonde ] and lemma [ extbinom ] , and @xmath121 is a consequence of lemma [ prodbinom ] .
that is , by and , we have proven the main result of this section : [ ankij ] @xmath122 recall from ( * ? ? ?
* theorem 1 ) and ( * ? ? ?
* theorem 3.1 ) that for any @xmath123 partition @xmath29 on @xmath30 , the partition weight enumerator is @xmath124 [ pwe = ankij ] let @xmath29 be the @xmath125 partition on @xmath30 associated to the @xmath2 information symbols and @xmath49 redundancy symbols .
then @xmath126 for all @xmath127 and @xmath128 .
the proof of proposition [ pwe = ankij ] is in appendix [ pf : pwe = ankij ] . as a consequence of theorem [ ankij ]
, we may recover the well known formula of the weight distribution of an mds code as it appears in @xcite .
let @xmath129 denotes the weight distribution of @xmath0 .
then @xmath130 since a codeword of weight @xmath131 distributes its weight between the first @xmath2 and the last @xmath49 coordinates .
[ mdswd ] the weight distribution of an mds code is given by the following formula : @xmath132\ ] ] for @xmath133 , @xmath134 and @xmath135 if @xmath136 .
the case @xmath137 is evident .
if @xmath138 , as we have pointed out in remark [ ankij=0 ] , @xmath139 .
let us now suppose that @xmath140 . by theorem [ ankij ] , as @xmath141 if @xmath142 , @xmath143 we calculate @xmath144
. since @xmath145 , then @xmath146 and @xmath147 if @xmath148 , then @xmath149 for all @xmath150 , so , for any @xmath2 and @xmath131 , @xmath151 by lemma [ chuvandermonde ] , we obtain that @xmath152 so @xmath153 , and hence @xmath154 we drop @xmath155 and develop the rest of the previous formula , @xmath156 where we have split the case @xmath157 . by lemma [ binomsumprod ] for @xmath158 , @xmath159 , @xmath160 and @xmath161 , @xmath162 and this yields @xmath163 the independent term in @xmath3 of this expression
can be rewritten as @xmath164 where the last equality is given by lemma [ binomsumprod ] with @xmath165 , @xmath166 , @xmath167 and @xmath168 . by lemma [ altbinom ] with @xmath169 , @xmath170 and @xmath171
we have @xmath172 and consequently @xmath173 .
\end{split}\ ] ] this finishes the proof . as a corollary of the above proof , we find a new description of the weight distribution of an mds code .
for any @xmath174 , the number of weight @xmath131 codewords of an mds code is given by the formula @xmath175 now , we are in the position to describe the information bit and information symbol error rate of a false negative , concretely , @xmath176 the iber of a false negative can be compared for different codes in figure [ graphfn ] .
observe that it increases monotonically as the channel ber increases .
the iber of a fn is significantly smaller than the channel ber , at least for dimensions less or equal than @xmath177 . of length 127 and dimension @xmath2 .
in this section , we shall make use of the calculus of the values @xmath24 in order to obtain the information errors of a decoding failure . for simplicity
, we may assume that the zero codeword is transmitted .
suppose that we have received an erroneous transmission with @xmath178 non - zero coordinates in the information set and @xmath179 in the redundancy .
all along the paper , we shall say that the weight of this error is @xmath180 .
if the received word is corrected by the code , then it is at a maximum distance @xmath181 of a codeword .
obviously , if @xmath182 , the word is properly corrected , so we may assume that @xmath183 .
in this case , the correction is always wrong and we have a decoding failure .
our aim now is to count these words , highlighting the number of wrong information symbols and the errors belonging to the redundancy , i.e. the words of weight @xmath180 that decode to a codeword @xmath184 of weight @xmath185 .
the reasoning is as follows : for any codeword @xmath184 of such weight , we calculate @xmath186 , the number of words of weight @xmath180 which belong to its sphere of radius @xmath181 .
this can be carried out by provoking up to @xmath181 changes in @xmath184 .
our reasoning is analogous to the one in @xcite .
firstly , there is a minimum number of symbols that must be changed , either to zero or to something non - zero depending on the sign of @xmath187 and @xmath188 in order to obtain the correct weight .
if @xmath181 is large enough , we can use the remaining correction capacity to modify an additional number of symbols to zero , and the same number of symbols to a non - zero element of @xmath189 in order to keep the weight unchanged .
finally , the remaining possible symbol modifications can be used to change some non - zero symbols into other non - zero symbols , without affecting the weight of the word .
let @xmath190 where @xmath191 denotes the absolute value of @xmath1 .
we may distinguish four cases : [ [ r_1-leq - c_1-and - r_2-leq - c_2 ] ] @xmath192 and @xmath193 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in the @xmath194 non - zero information coordinates , @xmath195 of them should be changed to zero .
additionally , we also allow @xmath196 more . in the same way , on the @xmath197
non - zero coordinates of the redundancy , we must change @xmath198 and @xmath199 of them . therefore
, we have the following number of possibilities : @xmath200 now , we should give a non - zero value to @xmath196 coordinates between the @xmath201 remaining information symbols , and @xmath199 coordinates between the @xmath202 remaining redundancy symbols .
thus , the number of possible changes is as follows : @xmath203 since the changes can not exceed @xmath181 , the admissible quantities for @xmath196 and @xmath199 satisfy @xmath204 and hence @xmath205 finally , we may change some of the remaining non - zero @xmath206 and @xmath207 coordinates to another non - zero symbol .
if we change the @xmath208 and @xmath209 coordinates , respectively , we obtain @xmath210 changes , where @xmath208 and @xmath209 satisfy @xmath211 therefore , the total number of words is the following : @xmath212 if we denote @xmath213 and @xmath214 , by lemma [ prodbinom ] , we have @xmath215 [ [ r_1-leq - c_1-y - r_2-c_2 ] ] @xmath192 y @xmath216 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we proceed as in the above case with the information symbols .
so we can make @xmath217 changes in the information set . in the redundancy part
, we must give a non - zero symbol to @xmath218 coordinates , and , therefore , change @xmath199 of @xmath197 coordinates to zero . in that way
, we have @xmath219 possible changes , where @xmath220 and then @xmath205 finally , changing the value of @xmath208 and @xmath209 of the remaining @xmath206 and @xmath221 non - zero coordinates , we have @xmath222 changes , where @xmath211 thus , the total number of words is @xmath223 again , we denote @xmath213 and @xmath214 , and , by lemma [ prodbinom ] , we simplify the expression to @xmath224 if we change @xmath225 and we take in care that binomial coefficients of negative integers are @xmath226 , we get @xmath227 [ [ r_1-c_1-y - r_2-leq - c_2 ] ] @xmath228 y @xmath193 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in this case , @xmath229 as before .
if we make the change of variable @xmath230 we can proceed as in the previous case and we get the formula @xmath231 where @xmath232 and @xmath233 .
[ [ r_1c_1-y - r_2c_2 ] ] @xmath234 y @xmath235 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in this case , @xmath236 we proceed as in the two previous cases and we obtain the formula @xmath237 where @xmath225 , @xmath238 , @xmath239 and @xmath240 . [ nrc ] let @xmath183 .
for any codeword @xmath184 of weight @xmath185 the number @xmath186 of words of weight @xmath180 which belong to its sphere of radius @xmath181 is @xmath241 where @xmath242 and @xmath243 .
it is a direct consequence of equations , , and than @xmath244 by lemma [ prodbinom ] @xmath245 hence @xmath246 lemma [ chuvandermonde ] provides the result . in order to calculate the iber in a wrong correction , we follow again the ideology of @xcite .
indeed , the probability of getting a wrong bit is different if it is due to the channel or due to the encoder . in the first case ,
it is given by the channel ber whilst , in the second case , it is assumed to be @xmath247 .
hence , we should calculate the rate of errors in the information set committed by the decoder , that we shall call @xmath248 .
this calculation is similar to the one for numbers @xmath186 . concretely , if @xmath192 , for each @xmath25 , @xmath249 , @xmath26 , and @xmath250 , in equation , the number of changes due to the decoder is @xmath251 . in order to simplify the resulting expression , note that @xmath252 by lemmas [ extbinom ] and [ chuvandermonde ] , where @xmath253 .
hence , number @xmath248 , for the case @xmath254 , is derived from equation @xmath255 if @xmath228 the number of changes due to the decoder is @xmath256 . in this case
we check as before that @xmath257 so @xmath258where @xmath259 and @xmath260 are as in theorem [ nrc ] .
hence , the information bit and symbol error rate of a wrong correction is as follows . observe from figure [ graphwc ] that the iber of a wc also increases monotonically , so mds codes can be said `` proper '' with respect to the iber , see @xcite .
@xmath261 of length 127 and dimension @xmath2 . ]
as we pointed out in the introduction , there exists the possibility of occurrence of an fp . up to our knowledge , this has not been treated before in the literature . in this section ,
we calculate the probability that a ped and an fp occur , finishing our estimation of the iber of an mds code .
as we noticed above , without loss of generality , we may suppose that the zero word is transmitted and we want to analyze the behaviour of the received word .
our purpose now is to count the words whose weight decomposes as @xmath262 , i.e. there are no non - zero information symbols , which are not corrected by the decoder . obviously ,
if @xmath263 , the word shall be properly corrected , so we assume that @xmath264 .
two disjoint cases can take place : the error is detected but not corrected , producing an fp , or the error is ( wrongly ) corrected to a codeword . since the total number of such words is given by @xmath265 it is enough to calculate the words corresponding to one of the two cases .
we can make use of the calculations in section [ s3 ] and give an expression of the words belonging to the second case .
indeed , the number of words of weight @xmath131 with @xmath266 that are corrected to a codeword is as follows : @xmath267 hence , the number of false positives of weight @xmath131 is given by @xmath268 and the probability of producing a false positive is given by the following formula @xmath269 it can be observed in figure [ graphfp ] that the probability of a fp has a maximum for each code .
when the channel ber is high enough this probability increases as the error correction capability of the code increases , see figure [ graphfp2 ] . 0.45 of length 127 and dimension @xmath2.,title="fig : " ] 0.45 of length 127 and dimension @xmath2.,title="fig : " ] we may now give an estimation of the iber of a ped . indeed , when the received word has a weight greater than @xmath181 , the error - correcting capability of the code , three disjoint cases can take place : an undetected error , an fp , or a ped .
hence , for a given weight @xmath270 , the number of words producing a ped is given by @xmath271 whenever @xmath272 and zero otherwise .
therefore , @xmath273 the reader may observe from figures [ graphped0 ] that , for high channel ber s , the behaviour of the iber of a ped becomes almost linear . actually , the curves approximate to the line @xmath274 according the dimension of the code diminishes .
0.45 of length 127 and dimension @xmath2.,title="fig : " ] 0.45 of length 127 and dimension @xmath2.,title="fig : " ]
for the convenience of the reader and in order to make the paper self - contained , we add the combinatorial identities that have been referenced all along the paper . [ chuvandermonde ] @xmath275 see e.g. ( * ? ? ?
* , equation ( 3 ) ) .
[ prodbinom][extbinom ] @xmath276 it follows directly from the definition .
[ altbinom ] @xmath277 an easy induction on @xmath278 using pascal s rule .
[ binomsumprod ] @xmath279 ( * ? ? ?
* pg.7 , equation ( 2 ) ) [ diffin ] let @xmath280 $ ] be a polynomial whose degree is less that @xmath259
. then @xmath281 direct consequence of the calculus of finite differences , see e.g. @xcite .
let s develop the formula of theorem [ ankij ] : @xmath282 let s compute the coefficients of each degree in @xmath3 .
let @xmath283 .
the coefficient of @xmath284 is obtained when @xmath285 : @xmath286 where the last equality follows from lemma [ binomsumprod ] . if @xmath287 , the coefficient of @xmath288 is @xmath289 where we have applied lemma [ binomsumprod ] again in the equality .
therefore @xmath290 we proceed with @xmath291 is a similar way .
by @xmath292 where @xmath293 and @xmath294 are consequence of lemma [ chuvandermonde ] .
making @xmath295 we are done .
the authors wish to thank dr .
pascal vontobel for pointing out the references @xcite and @xcite and for its comments of the manuscript .
the authors also wish to thank dr .
javier lpez for his assistance in performing the graphics .
m. el - khamy and r.j .
mceliece , `` the partition weight enumerator of mds codes and its applications , '' _ information theory , 2005 .
isit 2005 . proceedings .
international symposium on _ , pp .
926 - 930 , 4 - 9 sept . | in this paper , a computation of the input - redundancy weight enumerator is presented .
this is used to improve the theoretical approximation of the information bit and symbol error rate , in terms of the channel bit - error rate , in a block transmission through a discrete memoryless channel .
since a bounded distance reproducing encoder is assumed , the computation of the here - called false positive ( a decoding failure with no information - symbol error ) is provided . as a consequence , a new performance analysis of an mds code is proposed .
lobillo , navarro and gmez - torrecillas mds code , bit - error rate ( ber ) , block error - correcting code , information bit error rate ( iber ) , false positive . |
the cabibbo - kobayashi - maskawa ( ckm ) paradigm of quark mixing in the standard model ( sm ) is yet to be accurately tested in the @xmath2 sector , and it is quite possible that the np can affect the @xmath2 system while keeping the @xmath9 system untouched . indeed , for most of the flavor - dependent np models , the couplings relevant for the second and third generations of sm fermions are much less constrained than those for the first generation fermions , allowing the np to play a significant role in the @xmath2 mixing , in principle .
over the last few years , the tevatron experiments cdf and d , and to a smaller extent the b factories belle and babar , have provided a lot of data on the @xmath10 meson , most of which are consistent with the sm .
there are some measurements , though , which show a significant deviation from the sm expectations , and hence point towards new physics ( np ) .
the major ones among these are the following .
( i ) measurements in the decay mode @xmath0 yield a large cp - violating phase @xmath11 @xcite .
in addition , though the difference @xmath12 between the decay widths of the mass eigenstates measured in this decay is consistent with the sm , it allows @xmath6 values that are almost twice the sm prediction , and also opposite in sign @xcite .
( ii ) the like - sign dimuon asymmetry @xmath1 in the combined @xmath13 data at d @xcite is almost @xmath14 away from the sm expectation .
the resolutions of the above anomalies , separately or simultaneously , have been discussed in the context of specific np models : a scalar leptoquark model @xcite , models with an extra flavor - changing neutral gauge boson @xmath15 or r - parity violating supersymmetry @xcite , two - higgs doublet model @xcite , models with a fourth generation of fermions @xcite , supersymmetric grand unified models @xcite , supersymmetric models with split sfermion generations @xcite or models with a very light spin-1 particle @xcite .
possible four - fermion effective interactions that are consistent with the data have been analyzed by @xcite and the results are consistent with @xcite .
similar studies , based on the minimal flavor violating ( mfv ) models @xcite , and the randall - sundrum model @xcite , have been carried out . in this paper
, we try to determine , in a model - independent way , which kind of np would be able to account for both the above anomalies simultaneously .
we take a somewhat different approach than the references cited above . rather than confining ourselves to specific models , we assume that the np responsible for the anomalies contributes entirely through the @xmath2 mixing , and parameterize it in a model independent manner through the effective hamiltonian for the @xmath2 mixing .
this effective hamiltonian @xmath16 is a @xmath17 matrix in the flavor basis , and the relevant np contribution appears in its off - diagonal elements .
the np can then be parameterized by using four parameters : the magnitudes and phases of the dispersive part and the absorptive part of the np contribution to @xmath16 .
a `` scatter - plot '' analysis that constrained these four new parameters using only @xmath1 has been carried out in @xcite .
we perform a @xmath18 fit to the @xmath2 mixing observables and obtain a quantitative measure for which kind of np is preferred by the data .
this would lead us to shortlist specific np models that have the desired properties , which can give testable predictions for other experiments .
it is found that the np needs to contribute to both the dispersive as well as absorptive part of the hamiltonian in order to avoid any tension with the data .
we also extend our framework to include possible cpt violation in the @xmath2 mixing , parameterized through the difference in diagonal elements of @xmath16 .
the motivation is to check if this can obviate the need for an absorptive contribution from the np .
such an analysis to constrain cpt and lorentz violating parameters was carried out in @xcite .
however they have used only @xmath1 and not @xmath7 in their analysis , and their parameters are only indirectly connected to the elements of @xmath16 .
we try to account for the two anomalies above with only cpt violation as the source of np , and with a combination of cpt violation and the np contribution to the off - diagonal elements of @xmath16 .
as we will show , nothing improves the fit significantly from the sm unless there is a nonzero absorptive part in the @xmath2 mixing amplitude .
the paper is organized as follows . in sec .
[ hamiltonian ] , we introduce our formalism for the four np parameters . in sec .
[ measurements ] , we summarize the experimental measurements and theoretical predictions for the observables relevant for @xmath2 mixing . in sec . [ fit ] , we present the results of our fits , and their implications for np models are discussed in sec . [ preferred ] . in sec .
[ cptv - formalism ] , we introduce the formalism for introducing cpt violation and in sec .
[ cptv - fit ] we explore the extent to which it can help resolving the anomalies . sec .
[ concl ] summarizes our results and concludes .
the evolution of a @xmath2 state can be described by the effective hamiltonian = m_11 & m_12 + m_12^ * & m_22 - _ 11 & _
12 + _ 12^ * & _ 22 [ eff - h ] in the flavor basis , where @xmath19 and @xmath20 are its dispersive and absorptive parts , respectively . when cpt is conserved , @xmath21 and @xmath22 .
the eigenstates of this hamiltonian are @xmath23 and @xmath24 , with masses @xmath25 and @xmath26 respectively , and decay widths @xmath27 and @xmath28 respectively .
the difference in the masses and decay widths can be written in terms of the elements of the hamiltonian as m_s & & m_sh - m_sl 2 |m_12| , [ dms - def ] + _ s & & _ sl - _ sh 2 |_12| .
[ dgs - def ] the above expressions are valid as long as @xmath29 , which is indeed the case here .
since cpt is conserved , the effect of np can be felt only through the off - diagonal elements of @xmath16 .
we separate the sm and np contributions to these terms via m_12 & = & m_12^sm + m_12^np , + _ 12 & = & _ 12^sm + _ 12^np .
[ sm - np ] the np can then be completely parameterized in terms of four real numbers : @xmath30 and @xmath31 .
we take the phases @xmath32 and @xmath31 to lie in the range 0 - 2@xmath33 . in a large class of models ,
including the minimal flavor violation ( mfv ) models , the np contribution has no absorptive part , i.e. @xmath34 .
this is true for a lot of non - mfv models too .
this occurs when np does not give rise to any new intermediate light states to which @xmath10 or @xmath35 can decay . for such models , eq .
( [ dgs - def ] ) implies that @xmath36 , i.e. the value of @xmath12 is always less than its sm prediction @xcite . in such models
, the np is parameterized by only two parameters : @xmath37 and @xmath32 .
an analysis restricted to this class of models was performed in @xcite .
however there exists a complementary class of viable models where the np contributes to @xmath4 substantially .
these include models with leptoquarks , r - parity violating supersymmetry , a light gauge boson , etc .
it has been pointed out in @xcite that such a nonzero absorptive part that arises naturally in these class of models can enhance @xmath6 significantly above its sm value , contrary to the popular expectations based on @xcite .
one notes that a new absorptive part in the mixing amplitude necessarily means new final states that can be accessed by both @xmath10 and @xmath35 .
the data from the direct measurements of branching ratios is extremely restrictive @xcite , apart from that for a few final states like @xmath38 @xcite .
as we shall see later in this paper , such models are favored by the @xmath2 mixing data .
the importance of @xmath39 final states from @xmath40 and @xmath10 decays has also been pointed out in @xcite .
the @xmath2 oscillation and cp violation therein can be quantified by four observables , viz .
the mass difference @xmath5 , the decay width difference @xmath6 , the cp - violating phase @xmath7 , and the semileptonic asymmetry @xmath41 .
the mass difference is measured to be m_s = ( 17.770.100.07 ) ps^-1 , [ dm - value ] which is consistent with the sm expectation @xcite m_s(sm ) = ( 17.32.6 ) ps^-1 .
[ dm - sm ] however measurements in the @xmath0 decay mode show a hint of some deviation from the sm .
the cp - violating phase @xmath7 in this decay is _
s^j/ = ( - ) , [ betas - def ] whose average value measured at the tevatron experiments @xcite is _ s^j/ = ( 0.41^+0.18_-0.15 ) ( 1.16^+0.15_-0.18 ) .
[ betas - value ] in the sm , _
s^j/(sm ) = arg ( - ) 0.0190.001 .
[ betas - sm ] thus , the measured value of @xmath11 is more than @xmath42 away from the sm expectation .
on the other hand , the difference in the decay widths of the mass eigenstates @xmath43 and @xmath44 is measured to be @xcite _ s = ( 0.154^+0.054_-0.070 ) ps^-1 , [ dgs - value ] while the sm expectation is @xcite _ s ( sm ) = ( 0.0870.021 ) ps^-1 .
[ dgs - sm ] the measurement is consistent with the sm expectation to @xmath45 , however it allows for @xmath6 values that are almost twice the sm prediction .
note that the sign of @xmath6 is undetermined experimentally and this gives us more room to play with the np parameters .
cdf has recently announced its new results , based on 5.2 fb@xmath46 of data @xcite : _ s&= & ( 0.075 0.035 0.010 ) ps^-1 , + _ s^j/ & = & ( 0.02 - 0.52 ) ( 1.08 - 1.55 ) [ expdata - cdf ] to 68% c.l .. while we note that the results are consistent with the sm , the final tevatron averages are still awaited .
therefore , we use the values in eq .
( [ dgs - value ] ) in our analysis .
the other anomalous measurement is the like - sign dimuon asymmetry . averaging the 9.0 fb@xmath46 data of d @xcite and 1.6 fb@xmath46 data of cdf @xcite , and adding the errors in quadrature and treating them as gaussian
, we get a^b_sl = -(7.411.93)10 ^ -3 , [ ab - value ] which differs by more than @xmath47 from its sm prediction a^b_sl ( sm ) = ( -0.23^+0.05_-0.06)10 ^ -3 .
[ ab - sm ] note that for @xmath48 , cdf has a poorer statistics than d and therefore the average value is dominated by the d data . even in the presence of new physics ,
the sm relationship holds : a^b_sl = ( 0.506 0.043 ) a^d_sl + ( 0.4940.043 ) a^s_sl , [ ab - ad - as ] where @xmath41 and @xmath49 are the semileptonic asymmetries for the @xmath2 and the @xmath9 systems , respectively .
the former is related to the @xmath2 mixing observables through a^s_sl = _ s [ asl - def ] where @xmath50 .
the latter is defined analogously .
the coefficients in eq .
( [ ab - ad - as ] ) are experimentally measured , and contain information about @xmath51 , @xmath52 , and production fractions of @xmath40 and @xmath10 mesons . using @xmath53 @xcite , this leads to a^s_sl = - 0.010 0.006 , [ asl - value ] which is about @xmath54 away from the sm prediction a^s_sl(sm ) = ( 2.060.57)10 ^ -5 .
[ asl - sm ] the value of @xmath49 depends on @xmath55 and @xmath56 , the parameters in the @xmath40 sector anologous to those in eq .
( [ asl - def ] ) .
these parameters depend on the np in the @xmath40 sector , which is independent of the np parameters in the @xmath10 sector that we are considering .
we therefore do not consider the measured values of @xmath49 as a direct constraint , but express it in terms of @xmath57 , @xmath58 , and @xmath56 , whose experimental values are taken as inputs . in the sm
, we have @xmath59 @xcite .
note that if the dominating contribution to @xmath60 were from a pair of intermediate @xmath61 quarks , @xmath62 would have been equal to @xmath63 .
since the intermediate @xmath64 and @xmath65 quark states give comparable contributions to @xmath60 , we have @xmath66 @xcite .
we perform a @xmath18 fit to the observed quantities @xmath67 and @xmath41 , using the np parameters @xmath30 and @xmath31 .
we assume all the measurements to be independent for simplicity , though the measurements of @xmath6 and @xmath7 are somewhat correlated .
the values of all the observables and their sm values are as given in sec . [ measurements ] . in order to express them in terms of @xmath68 and @xmath69
, one has to use eq .
( [ dgs - def ] ) in addition . in order to take into account the errors on the sm parameters , we add the theoretical and experimental errors on our observed quantities in quadrature . note
that since we have four observable quantities and four parameters , it is not surprising that we obtain the global minimum value of @xmath18 as @xmath70 when all the np parameters are allowed to vary .
the questions we address here are ( i ) what the preferred values of the np parameters are , and ( ii ) to what confidence level ( c.l . ) a given set of np parameters ( or sm , which is a special case of np with @xmath71 ) is allowed .
the latter is obtained assuming all errors to be gaussian . here
we give our results in terms of the goodness - of - fit contours for the joint estimations of two parameters at a time .
the @xmath72 contours , that are equivalent to @xmath73-values of @xmath74 , or confidence levels of @xmath75 , correspond to @xmath76 , respectively . in fig .
[ fig12 ] , we show the @xmath77 contours in the @xmath78 plane , where the other np parameters are marginalized over . clearly , we see a preference towards nonzero @xmath79 as well as nonzero @xmath32 values .
there are two best - fit points with @xmath80 , one at @xmath81 ps@xmath46 and the other at @xmath82 ps@xmath46 , shown with crosses in fig .
[ fig12 ] . actually , each of these crosses is a superimposed double , with two values of @xmath83 , as shown in fig .
[ fig34 ] .
the points correspond to the constructive and destructive interference between the sm and np amplitudes in order to give the measured central values of @xmath5 .
the region with @xmath84 , i.e. the @xmath85-axis , is outside the @xmath42 region , indicating that it will be rather difficult to fit the current data without some np contribution to the dispersive part of the @xmath2 mixing .
the contours also imply that @xmath86 ps@xmath46 to @xmath47 . in fig .
[ fig34 ] , we show the goodness - of - fit contours in the @xmath87 plane , marginalizing over other two np parameters . as the measurements
do not determine the sign of @xmath6 , for any particular value of @xmath88 , we perform the @xmath18 fit for both positive and negative values , and keep the minimum @xmath18 of the two .
this doubles the number of best - fit solutions , and the two best - fit points of fig .
[ fig12 ] now split into four . for @xmath89 , the solutions are @xmath90 or @xmath91 , and for @xmath92 , the corresponding solutions are @xmath93 or @xmath94 ( both @xmath95 and @xmath83 are in ps@xmath46 , here , and also later where not mentioned explicitly ) .
note that there is a reflection symmetry about @xmath96 .
again , a preference for nonzero values of @xmath97 is indicated , though @xmath31 may vanish .
the region with @xmath98 , i.e. the @xmath85-axis , is outside the @xmath14 allowed region , indicating that np contribution to the absorptive part of the effective hamiltonian is highly favored .
the contours also imply that @xmath99 at @xmath47 .
[ fig13 ] displays the contours in the @xmath100 plane , and the two np phases are marginalized over .
not only does it show a preference for nonzero values of @xmath95 and @xmath83 , but the @xmath84 axis is outside the @xmath42 allowed region and the @xmath98 axis is outside the @xmath14 allowed region .
the best fit points are again superimposed doubles , whose values can be read off from the discussion above . the origin in this figure
is the sm , which has @xmath101 , and lies even outside the @xmath14 allowed region .
this dramatically quantifies the failure of the sm to accommodate the current data .
the reason is evident from eqs .
( [ betas - value ] ) and ( [ asl - def ] ) ; while @xmath102 prefers @xmath103 close to @xmath104 or @xmath105 , with a probability minimum near @xmath106 , the measurement of @xmath48 , and hence that of @xmath41 , prefers large @xmath107 , forcing @xmath7 close to @xmath108 .
this creates the tension between these two measurements .
[ fig13 ] also tells us that the models for which @xmath98 , like r - parity conserving supersymmetry , universal extra dimension , and extra scalars , fermions , or gauge bosons , can not bring the tension down even to the @xmath14 range , unless the data moves towards the sm expectations ( and unless the new bosons are flavor - changing so as to generate a nonzero @xmath83 ) .
the best fit point with @xmath98 has @xmath109 and corresponds to @xmath110 .
this is further emphasized in fig .
[ fig120 ] , which shows the @xmath111 ( @xmath73 value of @xmath112 , @xmath113 ) contour for those np models where @xmath83 is set to vanish ( within the closed contour above and under the open contour below ) .
one may question the optimistic sm uncertainty for @xmath114 as quoted in eq .
( [ dgs - sm ] ) .
however , this has an almost negligible effect .
for example , if we increase the uncertainty by 50% , neither the best fit points nor the confidence levels change significantly . the best fit point in fig .
[ fig120 ] has a @xmath18 minimum of 20.58 instead of 20.75 .
the reason is the large deviation of @xmath41 from its sm value , to explain which we need a significant enhancement in @xmath107 .
from the results and discussion in the previous section , it appears that : + ( i ) the sm by itself is strongly disfavored . either @xmath95 or @xmath83 should be nonzero .
+ ( ii ) @xmath115 but @xmath98 is also not allowed at @xmath14 , but the fit is marginally better than the sm .
+ ( iii ) the hypothetical case where @xmath116 but @xmath84 is also disfavored to more than @xmath42 .
( this is a rather natural condition , since any interaction that contributes to @xmath83 will necessarily contribute to @xmath95 . )
most of the np models can contribute significantly to @xmath95 .
leading examples are the mfv models like minimal supersymmetry , universal extra dimensions , little - higgs with t - parity , etc .
non - mfv models like a fourth chiral generation , supersymmetry with r - parity violation , two - higgs doublet models , models with extra @xmath15 , etc . can also contribute significantly to @xmath95 .
the np models that can contribute significantly to @xmath83 , however , are rather rare .
this is because the np contribution to the absorptive part needs light particles in the final state , and there are strong limits on the decays of @xmath10 to most of the possible light final state particles .
one of the few exceptions is the mode @xmath117 , on which there is no available bound at this moment .
thus , the np that contributes to @xmath83 has to do so via the interaction @xmath118 , but without affecting related decays like @xmath119 or @xmath120 @xcite .
this can be achieved only in a limited subset of models , for example those with second and third generation scalar leptoquarks , r - parity violating supersymmetry @xcite , or extra @xmath15 bosons @xcite .
it turns out that the former can provide enough contribution to @xmath83 to increase @xmath6 up to its current experimental upper bound @xcite .
the amount of np required for this is consistent with the difference between the decay widths of @xmath40 and @xmath10 mesons ( @xmath121 @xcite ) , and the recent measurement of the branching ratio of @xmath122 , which is less than @xmath123 at 90% c.l .
@xcite .
one should note here that if the d results on the dimuon charge asymmetry survive the test of time , it will be a clear indication of the presence of a nonzero @xmath124 .
such models are also favoured from the cdf and d combined result on the allowed contours for @xmath7 and @xmath125 , but we need to wait for the final tevatron average .
the analysis till now is valid only if we assume cpt - invariance .
however , the cpt symmetry may be violated in theories that break lorentz invariance @xcite .
indeed for local field theories , cpt violation requires lorentz violation @xcite .
( this need not be true for nonlocal field theories as well as for theories with noncommutative space - time geometry , see @xcite . ) in general , cpt violation should result in differences in masses and decay widths between particle - antiparticles pairs
. however it may be easier to identify even through oscillation experiments , which typically are sensitive to an interference between the cpt - conserving and cpt - violating interactions .
while cpt violation in the @xmath126 system is severely constrained through the mass difference between the neutral kaons @xcite , the bounds on the cpt violating parameters in the @xmath40 and @xmath10 systems are rather weak .
in fact , the bounds for the @xmath40 sector are about three orders of magnitude weaker than those for the @xmath126 sector @xcite .
the bounds on lorentz - violating parameters using the data on b mesons can be found in @xcite and references therein .
here we use a model - independent parameterization , like the one earlier followed in @xcite and recently used by two of us @xcite , and determine the preferred parameter space using the data on @xmath2 oscillations . unlike @xcite , we take both @xmath48 and @xmath103 data into account . one should note that as a new physics option , cpt violation is not exactly at the same footing as the models mentioned before . however in the language of the effective hamiltonian @xmath16 , the cpt violation manifests itself naturally through in a difference between the diagonal elements of @xmath16 .
it is therefore interesting to see if the constraints on the np coming from @xmath127 and @xmath114 can be relaxed at all with these additional degrees of freedom .
a posteriori , we will justify the discussion on cpt violation by showing that if the new physics indeed turns out to be without an absorptive part , cpt violation might help to explain the @xmath2 mixing data , albeit only marginally .
the cpt violation manifests itself in the effective hamiltonian through the difference in the diagonal elements .
we write the effective hamiltonian in eq .
( [ eff - h ] ) as h = m_0 - _ 0 - & m_12 - _ 12 + m_12^ * - _ 12^ * & m_0 - _ 0 + , [ h - with - cpt ] and define the dimensionless cpt - violating complex parameter @xmath128 as = , where @xmath129 .
the eigenvalues of @xmath16 are = ( m_0 - _ 0 ) y h_12 , [ evalues ] where @xmath130 and @xmath131 .
the corresponding mass eigenstates are |b_sh > & = & p_1|b_s > + q_1|_s > , + @xmath132 , and _ 1 & & = ( + ) , + _ 2 & & = ( - ) .
clearly , cpt invariance corresponds to @xmath133 .
let us now determine the dependence of our four observables on the cpt - violating parameters .
the differences in masses and decay widths of the eigenstates are related to the difference in eigenvalues as _
1-_2 = m + , where @xmath134 and @xmath135 are ordered such that @xmath136 . from eq .
( [ evalues ] ) , m & = & m_1 - m_2 = 2 re ( y h_12 ) , [ dm - cpt ] + & = & _ 2 - _ 1 = 4 i m ( y h_12 ) [ dgs - cpt ] . since @xmath137 , we can write h_12 & = & |m_12| ^ + & & |m_12| .
then eqs .
( [ dm - cpt ] ) and ( [ dgs - cpt ] ) yield m & & |m_12| , [ dm - cpt - approx ] + & & |m_12| .
[ dgs - cpt - approx ] the dependence on the cpt - violating parameter @xmath128 appears entirely through @xmath138 .
let us pause here for a moment and find what the above two equations tell us about the allowed parameter space .
let us first focus on the best constraint , @xmath5 , and work in the limit where @xmath139 is negligible .
@xmath140 , and hence @xmath95 , can be arbitrarily large , as re(@xmath138 ) can be made arbitrarily small by an appropriate choice of @xmath128 .
similarly , re(@xmath138 ) can be quite large ( albeit compatible with other constraints ) as long as there is a near - perfect cancellation between the sm and np mixing amplitudes , making @xmath140 small .
however , the smallness of @xmath141 constrains @xmath142 to be small , thus indicating that @xmath138 is almost real . since @xmath143
, this implies that @xmath144 is almost real and @xmath145 .
therefore , one would expect that @xmath128 is either almost real , or it is almost imaginary , but with @xmath146 .
now let us consider the cp - violating observables @xmath7 and @xmath41 .
the effective value of the former may be obtained in the presence of cpt violation by considering the decay rates of @xmath10 and @xmath35 to a final cp eigenstate @xmath147 as @xcite : ( b_s(t ) f_cp ) & = & |a_f|^2 , [ evol - b ] + ( _ s(t ) f_cp ) & = & ||^2 , [ evol - bbar ] with _
f_1 _ 1 , _ f_2 _ 2 , . here
@xmath148 and @xmath149 are the amplitudes for the processes @xmath150 and @xmath151 , respectively .
the time evolutions are given by f_-(t ) & = & ( e^-i _ 1 t - e^-i _ 2 t ) , + f_+(t ) & = & ( e^-i _ 1 t + e^-i _ 2 t ) , + @xmath0 is not a cp eigenstate , but a combination of cp - even and cp - odd final states , which may be separated using angular distributions . with the transversity angle distribution @xcite
, the time - dependent decay rate to the cp - even state is given by the coefficient of @xmath152 , while the time - dependent decay rate to the cp - odd state is given by the coefficient of @xmath153 .
the value of effective @xmath7 in this process is determined by writing the time evolutions ( [ evol - b ] ) and ( [ evol - bbar ] ) in the form ( b_s(t ) f_cp ) & = & c_1 ( _ s t/2 ) + c_2 ( _ s t/2 ) + + & & c_3 ( m_st ) + c_4 ( m_st ) , + ( _ s(t ) f_cp ) & = & |c_1 ( _ s t/2 ) + |c_2 ( _ s t/2 ) + + & & |c_3 ( m_st ) + |c_4 ( m_st ) .
[ trig - form ] the direct cp violation in @xmath0 is negligible ; i.e. @xmath154 .
also , @xmath155 , so that in the absence of cpt violation , @xmath156 .
then in terms of @xmath157 , one can write = - = - _ cp ( 2_s^j/ ) , where @xmath158 is the cp eigenvalue of @xmath147 .
when cpt is violated , the effective phases @xmath7 and @xmath159 measured through @xmath160 and @xmath161 decays , respectively , will turn out to be different . indeed , the difference between these effective phases will be a clean signal of cpt violation .
( 2_s^j/ ) & = & -_cp , [ betaeff - b ] + ( 2 |_s^j/ ) & = & -_cp , [ betaeff - bbar ] though the analysis of the @xmath10 and @xmath35 modes needs to be performed separately , here we assume identical detection and tagging efficiencies for both , and use the average of eq .
( [ betaeff - b ] ) and eq .
( [ betaeff - bbar ] ) for our fit . the semileptonic cp asymmetry @xmath41 is measured through the `` wrong - sign '' lepton signal : a^s_sl = .
[ aslq1 ] here , ( b_s(t)^- x ) & = & |_1 f_- a(b_s ^+ x)|^2 , + ( _ s(t)^+ x ) & = & |(f_-/_2 ) a(_s ^+ x)|^2 , and since @xmath162 , a^s_sl = = , [ asl - cpt ] which is independent of the cpt - violating parameter @xmath128 . that the semileptonic asymmetry does not contain a cpt violating term in the leading order
was also noted earlier @xcite .
in this section , we perform a @xmath18-fit to the observables @xmath5 , @xmath6 , the effective phase @xmath103 , and @xmath41 .
let us first assume that there is no cpt - conserving np contribution coming from @xmath95 and @xmath83 , so that the only relevant np contribution is cpt violating , and is parameterized by @xmath163 and @xmath164 . the allowed parameter space is shown in fig .
[ fig - cpt - only ] .
it turns out that in this case , the value of @xmath165 is @xmath166 16.4 ( at @xmath167 and @xmath168 ) , marginally better than the one obtained in the @xmath169 case discussed above in fig .
[ fig120 ] .
there are some , albeit small , regions in the parameter space that are allowed to @xmath14 .
however a fit good to @xmath47 or better is still not possible .
we therefore need to add the cpt - conserving np to the cpt - violating contribution .
however we have already seen in the preceding section that @xmath95 and @xmath83 together are capable of explaining the data by themselves . therefore the fit using @xmath128 , @xmath95 as well as @xmath83 is redundant . with six independent parameters and only four observables , not only is @xmath170 guaranteed , but no effective limits on cpt - conserving and cpt - violating parameters are generated .
we , therefore , go directly to the possibility where there is cpt - conserving np , but without an absorptive part : @xmath98 .
we have already observed ( fig .
[ fig120 ] ) that the entire region in the @xmath171 is outside the @xmath14 region in such a scenario .
we would now ask what happens if we enhance the two - parameter np with two more cpt violating parameters , _ viz .
_ , re(@xmath128 ) and im(@xmath128 ) .
this scenario is interesting because , as we have seen before , only very specific kind of np can contribute to @xmath83 , which would be tested severely in near future . in case no evidence for the relevant np
is found ( e.g. the branching ratio of @xmath8 is observed to be the same as its sm prediction ) , the next step would be to check if cpt violation , along with the np contribution through @xmath95 , would be able to account for the anomalies .
for example , one may want to determine @xmath172 and @xmath173 of eqs .
( [ betaeff - b ] ) and ( [ betaeff - bbar ] ) separately and see whether they are different .
[ m12-phase-0 ] shows the situation in the @xmath171 plane . as
compared to fig .
[ fig120 ] , one can see that once we marginalize over @xmath128 , we now have some regions allowed to within @xmath14 ( within the closed contour above and below the open contour ) , but none within @xmath47 . indeed , @xmath174 at @xmath175 .
this clearly does not improve the goodness - of - fit substantially , indicating that there is no good alternative for @xmath83 .
[ red - rem-0 ] shows the situation in the complex @xmath128 plane , when @xmath95 has been marginalized over .
the best - fit point corresponds to @xmath176 , which gives @xmath174 as mentioned earlier .
the cpt conserving point ( @xmath177 ) lies outside the @xmath14 region . as expected from the discussion in sec .
[ cptv - formalism ] , the allowed values of @xmath128 are close to the @xmath163 or @xmath164 axis , with @xmath178 restricted to 2 .
one observes that the current data allows rather large ( @xmath179 ) positive values of @xmath164 at @xmath14 .
any flavor - dependent new physics model can in general affect both mass and width differences in the @xmath10@xmath35 system .
it can also affect the cp - violating phase , as well as the dimuon asymmetry , which was found by the d collaboration to have an anomalously large value . with these four observables
, one can constrain the free parameters of the new physics model .
we have used the model independent approach where we consider the effective @xmath2 mixing hamiltonian @xmath16 and parameterize the np through its contribution to @xmath16 .
we quantify the goodness - of - fit for the sm and np parameter values by performing a combined @xmath18-fit to all the four measurements .
the tension of the data with the sm is clear by the high value of @xmath18 at the sm .
moreover , it is observed that we need np to contribute to the dispersive as well as absorptive part of the off - diagonal elements of @xmath16 in order for the current data to be explained . the absorptive contribution , in particular ,
can be obtained from a very limited set of models , which will be severely tested in near future .
we also introduce the possibility of cpt violation by adding unequal np contributions to the diagonal elements of @xmath16 .
we explicitly show how cpt violation might affect the observables , especially dwelling on the effect on @xmath11 . taken alone
, the cpt violation can not affect the dimuon asymmetry , and it can make the fit to the @xmath2 mixing data only marginally better . in combination with a cpt conserving np , it can enhance the allowed parameter space for that np , however it does not seem to be able to obviate the need of an absorptive contribution from np . the data on all the observables considered in this paper is still relatively preliminary , the deviations from the sm are only at about 2 - 3@xmath180 level , and future data may either confirm these deviations or expose them as statistical fluctuations . if the errors and uncertainties shrink keeping the central values more or less intact , this will mean : * the sm is strongly disfavored . moreover , the relevant np should be flavor - dependent , as we do not see much deviation in the @xmath9 sector .
* the np models that do not contribute to the absorptive amplitude of the @xmath2 mixing are also strongly disfavored if cpt is conserved .
the best bets are those np models that provide both dispersive and absorptive amplitudes in the @xmath2 mixing .
this also gives rise to new decay channels for @xmath10 .
for example , one might find the branching ratio of @xmath38 enhanced significantly from its sm expectation .
* without any cpt - conserving np , only cpt violation is only of marginal help , as it can not enhance the semileptonic asymmetry . even in combination with the cpt - conserving dispersive np
, it can not allow regions in the parameter space to better than @xmath47 . to summarize , the np models that contribute an absorptive part to @xmath2 mixing seem to be essential if one wants to explain the data on @xmath181 and @xmath1 simultaneously .
there is only a limited set of such models , and they will be severely tested in near future . in the scenario that such an absorptive np contribution is ruled out ,
one may have to resort to cpt violation in order to explain the data .
a prominent signature of such a cpt violation would be a difference in @xmath182 and @xmath159 as shown in eqs .
( [ betaeff - b ] ) and ( [ betaeff - bbar ] ) .
skp acknowledges csir , government of india , for a research fellowship .
the work of ak was supported by csir , government of india , and the drs programme of the university grants commission .
v. m. abazov _ et al . _
[ d0 collaboration ] , phys .
d * 82 * , 032001 ( 2010 ) [ arxiv:1005.2757 [ hep - ex ] ] ; phys .
lett . * 105 * , 081801 ( 2010 ) [ arxiv:1007.0395 [ hep - ex ] ] ; for the 9.0 fb@xmath46 data , see v. m. abazov _ et al . _
[ d0 collaboration ] , arxiv:1106.6308[hep - ex ] .
a. dighe , a. kundu and s. nandi , phys .
d * 76 * , 054005 ( 2007 ) [ arxiv:0705.4547 [ hep - ph ] ] .
a. dighe , a. kundu and s. nandi , phys .
d * 82 * , 031502 ( 2010 ) [ arxiv:1005.4051 [ hep - ph ] ] .
n. g. deshpande , x. g. he and g. valencia , phys .
d * 82 * , 056013 ( 2010 ) [ arxiv:1006.1682 [ hep - ph ] ] ; r. -m .
wang , y. -g .
xu , m. -l .
liu , b. -z .
li , jhep * 1012 * , 034 ( 2010 ) [ arxiv:1007.2944 [ hep - ph ] ] .
a. k. alok , s. baek and d. london , jhep * 1107 * , 111 ( 2011 ) [ arxiv:1010.1333 [ hep - ph ] ] . c. w. bauer , n. d. dunn , phys .
lett . * b696 * , 362 - 366 ( 2011 ) [ arxiv:1006.1629 [ hep - ph ] ] . k. blum , y. hochberg , y. nir , jhep * 1009 * , 035 ( 2010 ) [ arxiv:1007.1872 [ hep - ph ] ] .
a. datta , m. duraisamy and s. khalil , phys .
d * 83 * , 094501 ( 2011 ) [ arxiv:1011.5979 [ hep - ph ] ] .
a. kostelecky , r. van kooten , phys . rev . * d82 * , 101702 ( 2010 ) [ arxiv:1007.5312 [ hep - ph ] ] .
z. ligeti , m. papucci , g. perez , j. zupan , phys .
lett . * 105 * , 131601 ( 2010 ) [ arxiv:1006.0432 [ hep - ph ] ] .
y. grossman , z. ligeti and e. nardi , phys .
d * 55 * , 2768 ( 1997 ) [ arxiv : hep - ph/9607473 ] .
a. lenz and u. nierste , arxiv:1102.4274 [ hep - ph ] ; jhep * 0706 * ( 2007 ) 072 [ arxiv : hep - ph/0612167 ] .
a. datta , e. a. paschos and l. p. singh , phys .
b * 548 * , 146 ( 2002 ) ; k. r. s. balaji , w. horn and e. a. paschos , phys .
d * 68 * , 076004 ( 2003 ) .
a. kundu , s. nandi and s. k. patra , phys .
d * 81 * , 076010 ( 2010 ) [ arxiv:1002.1134 [ hep - ph ] ] .
a. s. dighe , i. dunietz , h. j. lipkin and j. l. rosner , phys .
b * 369 * , 144 ( 1996 ) [ arxiv : hep - ph/9511363 ] ; a. s. dighe , i. dunietz and r. fleischer , eur .
j. c * 6 * , 647 ( 1999 ) [ arxiv : hep - ph/9804253 ] .
a. pais and s. b. treiman , phys .
d * 12 * , 2744 ( 1975 ) [ erratum - ibid .
d * 16 * , 2390 ( 1977 ) ] ; s. bar - shalom , g. eilam , m. gronau , j. l. rosner , phys .
* b694 * , 374 - 379 ( 2011 ) [ arxiv:1008.4354 [ hep - ph ] ] . | recently observed anomalies in the @xmath0 decay and the like - sign dimuon asymmetry @xmath1 hint at possible new physics ( np ) in the @xmath2 mixing .
we parameterize the np with four model - independent quantities : the magnitudes and phases of the dispersive part @xmath3 and the absorptive part @xmath4 of the np contribution to the effective hamiltonian .
we constrain these parameters using the four observables @xmath5 , @xmath6 , the mixing phase @xmath7 , and @xmath1 .
our quantitative fit indicates that the np should contribute a significant dispersive as well as absorptive part .
in fact , models that do not contribute a new absorptive part are disfavored at more than 99% confidence level .
we extend this formalism to include cpt violation , and show that cpt violation by itself , or even in presence of cpt - conserving new physics without an absorptive part , helps only marginally in the simultaneous resolution of these anomalies .
the np absorptive contribution to @xmath2 mixing therefore seems to be essential , and would imply a large branching fraction for channels like @xmath8 . |
this paper tries to understand whether concentrations of emission in observations of position - velocity space represent real clumps of gas in molecular clouds ; whether the properties inferred for the observed objects describe the actual clouds ; and what we can say about the mechanisms that drive interstellar turbulence by observing molecular clouds with different tracers .
several papers have tried to address these points before .
@xcite show that , in a galactic disk model , gas complexes in a longitude - velocity diagram do not necessarily correspond to real complexes .
more recently , based on numerical simulations , @xcite showed that observed clumps frequently come from the superposition of several physically disconnected regions in the line of sight at the same radial velocity ( see also @xcite , ) but not necessarily at the same three - dimensional velocity .
moreover , the morphology seen in observed position - position - velocity maps seems to be more representative of the spatial distribution of the velocity field in the line of sight than of the distribution of the density field , as has been shown in 3d numerical simulations by @xcite .
similarly , @xcite shown that only the velocity field is responsible for the structure existing in the channel maps of observational data cubes .
if observed structures do not match actual structures , it is valid to ask whether statistical properties of observed clouds , such as the @xcite relationships , are still valid for actual clouds .
as @xcite mentions , they may instead be artifacts of the observations .
in fact , @xcite argues that the mean density - size relationship is an artifact of the limited dynamical range of the observations ( see also @xcite . ) @xcite used two - dimensional simulations of the diffuse interstellar medium to suggest that there is no relationship between mean density and size of physical clouds ( defined as a connected set of pixels in their simulations ) , but that there is a relationship , although with strong scatter , between the velocity dispersion and size , a result confirmed by @xcite .
although as we mentioned , several papers have studied the same questions as the present paper , they addressed them only partially , studying for example only objects in physical space @xcite , or in observational space @xcite ; or giving physical arguments to explain the observed @xcite relationships ( e.g. , @xcite itself ; @xcite ; @xcite ) . in the present paper
we use three - dimensional numerical models of molecular clouds to directly compare structures in simulated observations to the actual physical structures that produce them .
our goals are : to give observational predictions about the nature of the sources of the turbulence ; to disentangle what observed properties are representative of the true physical entities ; and what are more likely to just be the result of superposition of multiple structures along the line of sight . another way of asking this is , what are the differences between the observational space and the physical space ?
the plan of the paper is as follows : in [ simulationssec ] we describe the numerical models used here , and the methods used to define clumps and to generate simulated observations from model density and velocity cubes . in
[ role ] we study the role of the luminosity @xmath0 , wavenumber @xmath1 and initial magnetic field @xmath2 in the production of density fluctuations , and give observational criteria to infer the properties of the sources of turbulence .
we discuss the superposition effects , and analyze the relative importance of the parameters of the simulations in the superposition effects ( [ superposition ] ) .
section [ larsonsec ] studies larson s relations ( mean density vs. size , velocity dispersion vs. size and mass spectrum ) and discusses the importance of each one of them , by comparing physical and observational space , and [ summary ] draws the main conclusions .
in the present work we analyze driven , supersonic , hydro- and magnetohydrodynamical ( mhd ) models described by @xcite .
details of the models and forcing can be found there .
here we just mention that the models are solutions of the mass , momentum and induction equations using zeus-3d @xcite at a resolution of @xmath3 and @xmath4 zones .
they include periodic boundary conditions in each direction and an isothermal equation of state , implying that the simulations are scale free .
they are forced at a characteristic scale given by the wavenumber @xmath1 and at a constant energy injection rate @xmath0 , with an initial magnetic field intensity @xmath2 .
table [ simulations ] shows the parameters of those simulations .
the first column shows the name of the run , following the notation in @xcite of the alfvn velocity to the sound speed , with x representing 10 . ] in columns 2 , 3 and 4 we show the driving luminosity , wavenumber and initial intensity of the magnetic field intensities .
column 5 shows the value of the root mean square of the density fluctuations @xmath5^{1/2}.\ ] ] these simulations were performed to model the interior structure of molecular clouds , at scales of 0.01 - 1 pc , densities of @xmath6 cm3 , and temperatures of @xmath7 k. under these conditions , the isothermal equation of state is a reasonable approximation to the full thermodynamics of the cloud .
other works focussing on larger scales ( 10 - 1000 pc ) have explicitly included heating and cooling , a star formation scheme , and so forth @xcite , but these are not necessary to study the scales we are concerned with .
a bigger omission in our simulations is the lack of self - gravity , which will play a role in the production of dense cores @xcite .
nevertheless , we find the non - self - gravitating simulations useful to study superposition effects , and their consequences in the statistical properties observed for clumps .
even more , since our results are consistent with other works that include self - gravity , we conclude that its lack in our simulations is not relevant for the conclusions stressed here .
a similar conclusion has recently been reached by @xcite , who studied wavelet transforms of self - gravitating mhd turbulence and found that the overall structure was decoupled from that of the densest cores .
we analyze the numerical data first by analyzing the physical variables such as density , velocity , and magnetic fields themselves , and second by simulating observations of the models . in previous work @xcite
we represented the line profiles of the simulations as histograms of the velocity weighted by density .
however , those line profiles did not include the effect of temperature in broadening the profiles , and did not account for optical depth effects .
therefore we calculate the radiative transfer here , assuming local thermodynamic equilibrium ( lte ) for the population of the molecular energy levels , which is a sufficiently good approximation to study qualitatively the effects of projection and superposition ( velocity crowding ) of structure in the cloud . in order to model differences
when observing the same region with different tracers , we took into account that at low densities the molecules will be underexcited . to do this
we made the very simple assumption that the lines are excited only if the density is above 650 cm3 for the @xmath8co(1 - 0 ) transition , 6200 cm3 for @xmath8co(2 - 1 ) , and 1.8@xmath94 for cs(1 - 0 ) @xcite .
in other words , we assume lte if the local density is larger or equal to these density thresholds for each transition . because the simulations are scale - free , the units of density , temperature and length are arbitrary , but the included physics is adequate to represent the behavior of molecular gas with length scales between thousands of astronomical units for the densest cores , up to several parsecs . for the construction of the @xmath8co(1 - 0 ) ,
( 2 - 1 ) and cs(1 - 0 ) line profiles , we adopted solar abundances ; and typical scales for the density and length of @xmath10 cm3 and @xmath11 pc , respectively . since the simulations are isothermal , we choose the temperature equal to 10 k. these choices produce regions that are mostly optically thin in both @xmath8co and cs . in the production of line profiles
, we fixed the number of velocity channels at 32 .
this choice results in a velocity resolution of 0.06 km s@xmath12 for the less energetic simulations , and of 0.2 km s@xmath12 for the more energetic cases .
these resolutions are comparable to the velocity resolution of typical observations of molecular clouds at fcrao which is 0.08 km s@xmath12 , or in large - scale surveys like the full galactic co survey by @xcite which reaches resolutions as high as 0.26 km s@xmath12 . since we know the detailed density and velocity distribution , and assume a constant temperature , in order to determine the emission from any given molecular transition we just have to integrate the corresponding transfer equation , @xmath13 where @xmath14 is the intensity
; @xmath15 is the source function , which we assume to be a blackbody at @xmath16 k ; @xmath17 is the distance along the ray ; and @xmath18 is the mass absorption coefficient of the line , and has units of [ @xmath19 gr@xmath12 ] . in order to integrate equation ( [ transfer ] ) along the line of sight we solve , in practice , the following equations : @xmath21 and @xmath22 where @xmath23 represents the background intensity and @xmath24 is the total optical depth of the cloud , and @xmath25 is the local mass density . since the models have constant elements of volume ( zones )
, the integration over @xmath26 becomes a summation with constant step equal to the zone size .
the absorption coefficient @xmath18 is _
= a_ij c^28 ^ 2 g_ig_j ( v ) [ kappa ] where @xmath27 is the einstein coefficient for the @xmath28 transition if the density is larger than the threshold density for excitation , and zero if the density is smaller ; @xmath29 and @xmath30 are the statistical weights of the upper and lower states respectively ; @xmath31 is the numerical density of molecules in the lower state ; and @xmath32 is the line profile , which is assumed to be thermal .
these last two parameters take the form n_j = g_j q^-1 , and ( v)=c(m_mol2kt)^1/2 ( -m_mol ( v - v_0)^22kt ) , where @xmath33 is the mean molecular mass of the gas ( assumed to be 2.5 times the mass of the hydrogen atom ) , @xmath34 is the mass of the emitting molecule , @xmath35 is its molecular abundance relative to the hydrogen ( assumed to be solar ) , @xmath36 is the partition function , and @xmath37 is the velocity of the zone , which is known .
the molecular parameters are given in table [ molecules ] in order to represent and compare the physical and observational spaces in two - dimensional plots we integrate over one of the dimensions .
in particular , when we integrate in the @xmath38 direction the physical space , the corresponding observational cube is integrated in the @xmath39 direction .
this will allow us to compare , whether a clump in a position - velocity diagram has a counterpart in the position - position diagram .
then , when we label a plot as @xmath40 or @xmath41 , we show the column density field integrated in the @xmath42 direction , after zeroing all those points where the density is smaller than the density threshold ( see table [ molecules ] . )
the corresponding observational plot will be @xmath43 or @xmath44 respectively , and then represents the simulated observation of the @xmath45 map as seen by an observer located at the left of the plot , assuming @xmath8co(1 - 0 ) or @xmath8co(2 - 1 ) emission as described above .
the first problem that appears when we study the properties of clouds and clumps is how to define them .
for instance , if we are interested in studying large complexes and their substructure , we define a complex as a connected set of points with intensity above an arbitrary threshold @xcite .
when we work in physical space , the `` intensity '' is the actual density .
if we work in the observational space , the `` intensity '' is the brightness temperature . if instead , we are interested in studying individual clumps , we define a clump as a connected set of points below a local maximum following the intensity _ only _ downwards until the threshold is reached
. this scheme has been implemented by @xcite in a code called clumpfind .
a single large cloud with internal structure in the first scheme may be counted as several smaller clumps in the second scheme .
other schemes are also used .
for example , one can assume that clumps have a gaussian distribution of intensity @xcite , so that a clump defined as single by clumpfind can be decomposed into several gaussian substructures . in the present work we will use clumpfindfor simplicity .
differences between methods will be discussed in future contributions , but it is worth noting that , as @xcite has shown , these methods can not identify clumps larger than a few resolution elements .
we analyze how the parameters of the simulations ( driving strength and wavelength , magnetic field strength ) affect the structure in both observational and physical space . in particular , we attempt to give observational criteria to help determine the equivalent parameters of the physical mechanism driving the observed turbulence .
we begin by comparing models with the same driving wavenumber and magnetic field , but different driving strengths . in table [ simulations ] we see that the larger the driving strength @xmath0 , the larger the density fluctuations .
this occurs because the rms mach number in the simulations depends on the driving strength and the wavenumber as @xmath46 @xcite , and the density contrast across an isothermal shock depends on the mach number as @xmath47 .
figure [ maps_l ] shows maps of both the physical column density and the observed emission for two hydrodynamical models with the same wavenumber ( @xmath48 ) , but driving strength different by a factor of 100 : model ha8 ( @xmath49 ) and model he8 ( @xmath50 ) .
the main difference between these extreme cases is that ha8 has larger column densities of low - density tracers , and lower column densities of high - density tracers .
another clear difference is that in high - density tracers ha8 shows well - separated , roundish cores , while he8 shows a more filamentary structure with few compact cores .
these differences can be understood as follows . when the turbulence is weakly driven ( ha8 ) ,
density fluctuations are small .
there will be few regions with density below the density threshold for the low density tracer ( @xmath8co(1 - 0 ) ) , so most of the cube will emit in that line , but conversely only a few places will have densities higher than the threshold for the high - density tracer ( @xmath8co(2 - 1 ) ) , so only a small column density is seen in that line .
conversely , for strong driving ( he8 ) , the fluctuations around the mean are larger , and more regions will have densities either below the threshold for @xmath8co(1 - 0 ) or above the threshold for @xmath8co(2 - 1 ) , so that the mean column density for @xmath8co(1 - 0 ) is lower , but the mean column density for @xmath8co(2 - 1 ) is larger .
stronger driving leads not just to higher densities , but more extreme fluctuations , as indicated by a greater fraction of material at both high and low densities . since structures in the simulations are produced by the convergence of turbulent flows , the smaller the driving wavenumber , the larger the dominant structures .
we can see this effect in figure [ maps_k ] , where we show four maps for hc2 and four maps for hc8 , both with the same driving strength ( @xmath51 ) , but different wavenumber ( @xmath52 and 8 respectively ) . from these plots
we can see that the density structure is more homogeneously distributed in the case of high wavenumbers ( hc8 ) , in both physical and observational spaces . on larger scales , this suggests that _ large molecular clouds might be the result of turbulence driven at large scales in the diffuse interstellar gas ( see , e.g. , @xcite ) , and not the agglomeration of small clouds by the action of the self - gravity _ , as in some models of molecular clouds ( e.g. , @xcite ) the effect of magnetic fields in our models is consistent with previous work using simulations of the diffuse interstellar medium by @xcite , who showed that weak magnetic fields decrease the size of the density fluctuations compared to the hydrodynamic case . for intermediate magnetic fields ,
the density fluctuations may increase above the hydrodynamic and weak field cases , and finally , for very strong magnetic fields , the density fluctuations decrease again .
( the actual diagnostic they used , as shown in their fig .
8 , is the star formation rate as a function of the magnetic field strength ; but in their scheme the star formation rate depends directly on the size of the density fluctuations ) . in figure [ magnetic ]
we demonstrate this behavior .
the simulations used have intermediate driving strength ( @xmath51 ) and wavenumber ( @xmath53 ) , and three different initial magnetic field strengths of @xmath54 ( hc4 ) , @xmath55 ( mc41 ) , @xmath56 ( mc45 ) .
we can understand the dependence of the density fluctuations on the magnetic field strength as follows : weak magnetic fields are strongly tangled by the flow , so that they have a nearly isotropic magnetic pressure that prevents compressions and large density fluctuations , while stronger fields have an anisotropic magnetic pressure that allows larger compressions along the mean field lines , and thus larger density fluctuations .
magnetic fields do not noticeably affect the structure of the maps unless they are strong enough to introduce some anisotropy .
for example , in figure [ maps_b_10 ] we show @xmath8co(1 - 0 ) maps of three different simulations , integrated parallel ( @xmath39 ) , and perpendicular ( @xmath38 ) to the initial field direction . from bottom to top , the relative importance of the magnetic field compared to the luminosity of the simulation increases .
we show : me21 ( large driving strength , small wavenumber and small magnetic field ) ; ma81 ( low driving strength , large wavenumber and small magnetic field ) ; and ma4x ( small driving strength , intermediate wavenumber and large magnetic field ) .
only the strong field case ma4x ( top panels ) shows a morphology different than the hydrodynamic case . in this case , a marked anisotropy is visible , with the projection perpendicular to the magnetic field ( upper left panel ) , showing low - density structures aligned mainly in the direction of the field lines while the projection along the field line ( upper right panel ) presents regular structures , with no preferred direction .
interestingly , in a higher density tracer , there is no clear morphological difference between projections along or perpendicular to the initial field ( see fig [ maps_b_21 ] . )
the maximum values of the column density are similar in both physical and observational projections for small magnetic field intensity . nevertheless , in the strong - field case ( ma4x , see upper panels in fig .
[ maps_b_10 ] and [ maps_b_21 ] ) the column density integrated perpendicular to the mean field ( right panels ) is higher than the column density integrated parallel to it ( left panels ) , as can be seen in the grayscale bars .
this effect can be explained in the same way as that shown in figure [ magnetic ] : strong magnetic fields only allow compressions parallel to the field .
therefore , when we observe along the magnetic field lines we do not see high column density contrasts , while when we observe perpendicular to the field lines , we see higher contrasts in the compressions along the field lines . as a summary , _ strong fields can produce anisotropy and compressed structures perpendicular to the field lines , but weaker fields merely reduce the maximum density contrast somewhat .
_ velocity crowding contributes substantially to the generation of clumps in observational space , so observed clumps frequently contain emission from physically separated regions @xcite .
we demonstrate this effect using a typical mhd simulation : mc41 .
we choose this run because it has intermediate values of driving strength ( @xmath51 ) and wavenumber ( @xmath53 ) , and it clearly shows the effects of superposition .
figure [ map_mc41 ] shows physical and observed maps for @xmath8co(1 - 0 ) and ( 2 - 1 ) .
we see that clumps in real space ( letters a , b , and c in panel @xmath57 ) do not necessarily have a counterpart in observational space .
clumps in observational space , on the other hand , ( letter d in panel @xmath58 ) do not necessarily come from isolated regions in real space , but have contributions from many different regions along the same line of sight . in figure [ map_mc41 ]
we plot dotted lines that show the places where the emission of a physical clump lies in observational space , and where the emission from an observational clump is generated in physical space . for reference
, we use the same lines in the @xmath8co(1 - 0 ) map as in the @xmath8co(2 - 1 ) map .
for the same data set we made another set of maps using higher density tracers .
figure [ mc41_cs10 ] shows the emission of cs(1 - 0 ) .
the same lines are shown as in figure [ map_mc41 ] , to indicate the location of the large @xmath8co(1 - 0 ) core . in this case , the illusory observational clump d has almost disappeared , and the physical clump a is observed more clearly in its observational counterpart .
we also note that the position that we would attribute to the cores in different tracers is slightly different . in order to quantify the effects of superposition
, we now compute the typical number @xmath59 of physically disconnected regions that contribute to the emission of a single observed clump .
to do so , we developed an algorithm that brackets each observed clump in position and velocity and calculates the number of connected regions in physical space within the bracketed region in phase space . in figure
[ subclumps]_a _ we show the influence of driving strength and wavenumber on the amount of superposition .
we find that high - density tracers behave differently than low - density tracers . in the case of a high - density tracer such as cs , @xmath60 23 , and is almost constant for different driving strengths and wavenumbers . on the other hand , in the case of a low - density tracer such as @xmath8co , which emits from most of the cube , superposition increases with driving wavenumber .
this is a natural consequence of the way we forced the models , since short - wavelength driving produces multiple physical regions at the same velocity .
there is also a marginal tendency for strongly - driven models to exhibit more superposition than weakly - driven ones .
this occurs because the greater velocity dispersion in the strongly - driven model produces clumps with greater internal velocity dispersion that more easily overlap .
we conclude that higher - density tracers will better disentangle velocity crowding , as is shown by the results using cs(1 - 0 ) , though still by no means completely . in figure
[ subclumps]_b _ we show the relative influence of the magnetic field by examining @xmath59 as a function of the ratio of the driving strength to the initial magnetic field strength @xmath61 .
we find that the amount of superposition is a bit larger when we observe along the mean field direction ( @xmath39 ) than when we observe perpendicular to it .
the effect is small , but present , and it is a consequence of the greater compressions parallel to the field lines producing more condensed clumps with smaller cross - sections for superposition when observed perpendicular to the field lines .
a final word concerning superposition is in order : although it is true that high spatial resolution helps to segregate large clouds into their clumps , high velocity resolution does not necessarily help to disentangle superposition .
it only helps if there is a monotonic velocity gradient along the line of sight , such that two clumps with similar velocities merge at low velocity resolution , and can be separated at higher velocity resolution .
if clumps are produced in regions of converging flow in a larger turbulent region , however , as is likely the case in real clouds , then the velocity gradients will not be monotonic , and physically disconnected regions will often share the same line - of - sight velocity , so increasing velocity resolution will not help . as a summary ,
we conclude that velocity crowding is likely to always occur , implying that cores in observational space are not single entities in real space , and that it is larger for low - density tracers , large driving wavenumbers , and when we look along the mean field lines .
if observed clumps have contributions from multiple physical regions , it is important to understand whether the relationships reported for observed clumps in molecular clouds have anything to do with the properties of the actual physical clumps .
@xcite studied the dependence on size of the mean density , velocity dispersion , and mass spectrum of the clouds in a sample of observational data taken from the literature .
he found @xmath62 @xmath63 and @xmath64 the most commonly quoted values in the literature are @xmath65 , @xmath66 , and @xmath67 .
in particular , values of @xmath68 and @xmath69 have been attributed to virial equilibrium @xcite .
however , there is some discrepancy in the values reported @xcite , especially for high latitude clouds , where the velocity dispersion - size relationship has been questioned @xcite .
@xcite himself already suggested that the relationships he identified might be due to the limitations of the observations , but it was @xcite who first demonstrated that the observed mean density - size relationship could be due to observational effects , and that the observed and physical properties such as radius or volumetric density may be quite different .
@xcite showed that co and extinction saturate at roughly the same column density , which forms the upper envelope of the mean density - size relationship .
he also discussed how the surveys are biased due to the limitations of instrumental sensitivity and saturation .
@xcite reproduced the mass spectrum , as well as the velocity dispersion - size relationship ( although with large scatter ) for clumps in physical space , using two - dimensional models .
they reported the lack of a mean density - size relationship , confirming numerically the analysis by kegel ( 1989 , see also the discussions in @xcite and @xcite ) in the sense that there are clouds with small sizes and low column density that will be undetected in observational surveys .
more recently , @xcite reported a slope close to @xmath70 for the mean density - size relationship derived from simulated observations of their models , and a very flat correlation between velocity dispersion and size ( @xmath71 ) .
as they mentioned , both results are a consequence of their clump finding method ( based on what they call regions of contrast ) , which samples the entire line of sight .
none of these studies attempting to explain the origin of the @xcite relationships have compared simulated observations to the actual physical density and velocity distributions for the same numerical model .
we now do that , using the three - dimensional simulations of molecular clouds that we have described above .
we must first define the size of a clump , both in observational and in physical space . for clumps drawn from simulated observations
we take the circular radius , defined as @xmath72 , where @xmath73 is the projected area of the clump on the plane of the sky , for comparison with wgb94 . for physical clumps
we take the geometric radius , defined as @xmath74 .
other definitions are equally valid , but can lead to differences of a factor up to half an order of magnitude when comparing different size definitions , and differences of as much as a factor of 100 when calculating mean densities by dividing the total mass by the size . finally , since the results discussed in the present section do not appear to depend on the parameters of the runs we analyze or ( in the magnetic case ) on the projection with respect to the mean magnetic field , we only present results for run hc8 - 256 ( see table [ simulations ] ) , an available high - resolution run ( @xmath4 zones ) .
we search for clumps in physical space using clumpfind , considering only regions with a density larger than 650 cm3 , the critical density for @xmath8co(1 - 0 ) emission .
the results do seem to depend on the clumpfinding scheme used and on its parameters .
for example , the number and size of the clumps found in clumpfinddepends on the level of refinement used in clumpfind : at high refinement , large clumps can be split into multiple smaller clumps , changing not only the slope of the mass spectrum , but also the dynamical range of size in the scaling relations . in fig . [ rhomeanfig ] ( upper panel ) , we show the mean density - size relationship for clumps in three - dimensional physical space .
we note three points about this figure .
first , the relation between mean density and size given by equation ( [ rhomean ] ) is not seen at all , confirming the results of vbr97 from two - dimensional models .
second , there is a minimum density below which there are no clumps identified .
this minimum is just given by the density threshold we used in clumpfind .
third , even though the simulations exhibit a large dynamical range in density ( @xmath75 ) , the dynamical range in the mean density - size relationship is small , because in constructing such a plot we choose the clumps around the local density maxima . in the middle and lower panels of fig .
[ rhomeanfig ] we show the mean density - size relationship for clumps in simulated observational maps of the same run , integrated along the @xmath38 axis .
the middle panel is the one obtained by using cs(1 - 0 ) , and the lower panel is the relationship obtained by using @xmath8co(1 - 0 ) .
the observed clumps do exhibit approximately the relationship given by @xcite , despite the lack of correlation exhibited by the physical clumps in this model .
we conclude from this demonstration that the observed density - size relationship ( eq . [ [ rhomean ] ] ) is an observational artifact .
two mechanisms have been suggested to explain density - size relationship with slope @xmath76 in observations .
first , it might be due to the selection effect described by @xcite : only clouds with intensity exceeding the noise threshold determined by the instrumentation will be detected , effectively setting a column density cutoff , rather than the physical density cutoff imposed in the physical density - size relationship .
a constant column density cutoff produces a cutoff with slope @xmath70 in the mean density - size plane ( middle and lower panels ) , just as the constant physical density cutoff in physical space produces a flat cutoff .
second , it might be due to the limited dynamical range of the observations @xcite .
to understand this , lets consider the simulations : if we had clumps with constant column density @xmath77 , we would infer a mean density @xmath78 , giving exactly a slope of @xmath70 . in the simulation analyzed here
we do not have a constant column density , but the column density varies by only a factor of 30 ( see burkert & mac low 2001 ) , far less than the variation in physical density .
even worse , from this narrow distribution , the selected clumps have an even narrower column density distribution : on the one hand , as we already mentioned , there is a column density cutoff given by the minimum intensity used in clumpfind ; on the other hand , even the emission of the brightest clump does not come from the whole line of sight , and its column density is not the largest column density available in the simulation .
the end result is that selected clumps naturally end up with column density constant to within an order of magnitude , consistent with the observed scatter in the middle and lower panels of fig .
[ rhomeanfig ] .
similar arguments can be applied to the observational data .
for example , when observing emission lines from molecules , there is a limited dynamical range in density available in which any particular tracer can be excited : it will not be excited at too low density , and it will become optically thick at too high a density @xcite . at iras wavelengths , @xcite claim constant column densities in iras observations of dust , where the dynamical range of the data should have allowed detection of any significant variations .
however , @xcite showed that for iras observations , only the external layers of the clouds are heated by the diffuse uv radiation , so that the emission at 60 and 100 @xmath79 m actually comes from surface layers that are indeed expected to have nearly constant column density . finally , we point out that deviations from the mean density - size relationship have also been found in observations @xcite , suggesting that with sufficient care , the trap of apparently constant column density can be overcome . in figure [ deltavfig ]
we show the velocity dispersion - size relationship for clumps in physical and observational space . in the case of physical space ( upper panel )
the typical velocity dispersion is of order of 0.2 km s@xmath12 , a value that coincides with the value of the fluctuations of the overall velocity field for the run analyzed here ( hc8 - 256 ) .
the slope fit to the data is 0.3 @xmath80 0.38 , smaller than the value of 1/2 suggested for compressible turbulent behavior ( see , e.g. , vbr97 ) , but with a strong scatter .
in fact , in order to reduce substantially the uncertainty in the determination of the slope , we should have at least two orders of magnitude of dynamical range in sizes if the scatter is about one order of magnitude , as it is the case .
but with clumpfind , it is difficult to have a large dynamical range ( see , e.g. , upper right panel of fig 7 in @xcite ) , since a reasonable refinement of the intensity contouring makes large clumps split up into several pieces @xcite . on the other hand , when we analyze observational data as in fig [ deltavfig ] ( lower panel ) , additional problems play a role .
we now have the possibility that substantial superposition is occurring , in particular for low - density tracers .
if this is so , the velocity dispersion of a single observed clump will reflect the velocity dispersion of several , physically disconnected regions in the same line of sight .
this will also help to produce both large mean values of @xmath81 ( 0.8 km s@xmath12 , compared to the 0.2 km s@xmath12corresponding to the mean velocity fluctuations in the simulation ) and low values of the slope in the velocity dispersion - size relationship ( fig .
[ deltavfig ] , lower panel ) .
in contrast , for high density tracers ( cs(1 - 0 ) , fig.[deltavfig ] , middle panel ) , we obtain a steeper relationship ( 0.46 ) , and the mean velocity dispersion drops back to @xmath82 km s@xmath12 .
three reasons why this occurs are : in the case of high density tracers there is weaker superposition as shown in fig .
[ subclumps ] ; clumps are substantially smaller than in the low - density case , giving smaller velocity dispersion ; and the sizes span about one and a half orders of magnitude , favoring disentanglement between the scatter and the actual trend .
now we turn to the clump mass spectrum . before showing the mass spectra for the simulations
, we have to think about the nature of the turbulent density structure and how clumpfind ( or any other clumpfinding algorithm ) works .
for instance , in a driven turbulent medium , such as the simulations we examine here , we expect density fluctuations at all scales above the dissipation scale .
it is not at all clear when to stop counting substructure inside structure . in other words
, should we look for all the stones in the small hills in every single mountain of the mountain chain , or only count the larger mountains ?
if only the larger mountains , how do we define a lower limit ?
intuitively , we might think that the dependence of clump numbers on clumpfind parameters would be reduced in regions where clumps are well defined and reasonably isolated , as in an isolated star - formation region where cores have suffered some amount of collapse . to show that this is _
not _ the case , we show in figure [ mapa_hc8 ] a contour map of a slice through the center of model hc8 .
we use five and ten isocontours between an arbitrary density threshold of @xmath83 and the maximum density @xmath84 . as we can see , the density structures seem to be well - separated and reasonably well defined . in figure [ closeup_hc8 ]
we present an enlargement of the lower - right corner of figure [ mapa_hc8 ] , again using five and ten isocontours .
we see that , even for the same density threshold , by changing the number of isocontours we find five more peaks ( see right panel in fig . [ closeup_hc8 ] ) that will result in five new clumps in clumpfind .
this happens in just a small region containing what we would think of as well - defined structures , at a single slice in @xmath39 .
but in practice , we have 3d structures , so that every structure is far more complicated than apparent at first glance . even without considering
whether observational and physical clumps are the same or not , the mass spectrum appears to be more an artifact of the manner in which we count clumps than an actual physical characterization of the structure , unless we count them exhaustively all the way down to the numerical or physical dissipation scales below which there is no longer well - defined structure .
we note also that in the case of self - gravitating simulations the result is the same ( ossenkopf et al .
2001 ) , and that similar results will hold for observations : at low resolution the structures look soft , but once we increase the resolution and the sensitivity , structure emerges inside the structure , as is clear from figure 4 in dame et al .
. preliminary work by @xcite shows that , by setting the density threshold in physical space to the average density , a large enough number of contour levels gives a log - normal mass spectrum , independently of the number of contours used .
we compute the mass distribution using this prescription for run hc8 smoothed at a resolution of @xmath85 for physical space , and @xmath86 velocity channels for observational space in order to avoid excessive computation time and memory usage by clumpfind .
the results are shown in fig .
[ masspec ] .
the upper panel corresponds to physical space , and the lower panel corresponds to observational space , where instead of density , we used intensity contouring .
the results are clear : for large number of isocontours the shape of the mass spectrum converges .
good agreement is observed in the large - mass end of the distribution , although slower convergence is observed at the low - mass end .
convergence to a factor of two for the full spectrum appears to be reached for 64 or more contours . in fig .
[ lognormal ] we compare the mass spectrum calculated in physical space ( bold line ) to that from observational space ( thin line ) using 512 isocontours .
the dotted parabola shows a log - normal fit that reproduces fairly well the large - mass end of the spectra
. such log - normal behavior of the observed spectrum might explain observations obtained for dense cores in the last few years .
in particular , @xcite found a power - law mass spectrum with slope of @xmath87 for star - forming cores in serpens , substantially larger than the standard @xmath88 value .
moreover , @xcite found a change in slope at low mass : the less massive cores exhibit slope of @xmath88 , considerably shallower than the @xmath89 slope found for the more massive cores .
finally , we note that in these works , the mass spectrum was obtained using dust continuum observations with high signal - to - noise ratios . for present - day molecular - line observations , typical signal - to - noise ratios of 10 or less do not allow small contour separation , since contouring at less than the signal - to - noise ratio is meaningless @xcite .
mass spectra derived from such observations are likely to be far from physical ( fig .
[ masspec ] ) .
we analyze a set of three - dimensional mhd numerical simulations at intermediate ( @xmath3 zones ) and higher ( @xmath4 zones ) resolution described by @xcite , in which the parameters driving strength ( @xmath0 ) , driving wavenumber ( @xmath1 ) , and initial magnetic field ( @xmath2 ) are varied .
we show that : * density fluctuations are primarily controlled by the driving luminosity : the larger the luminosity , the larger the fluctuations seen in the maps .
morphology of the density structure , on the other hand , is primarily controlled by the driving wavelength : the larger the driving wavelength , the larger the structures seen in the maps , and vice - versa . *
the magnetic field only produces detectable effects in our simulated molecular line observations if the fields are strong and in the plane of the sky , and the structure is observed in low density tracers . when observing magnetized clouds , it is slightly easier to obtain larger column densities if the field is in the plane of the sky than if it is perpendicular to it . *
simulated observational maps of the models show strong superposition of structures along the line of sight .
this implies that single clumps in observational space are frequently the superposition of multiple clumps in physical space .
the apparent properties of observed clumps are poorly related to the physical properties of their constituent objects if strong superposition occurs .
* we explore the effect of the simulation parameters on the generation of strong velocity crowding , finding that large driving strength and large wavenumber tend to favor velocity crowding .
so far , the only more - or - less reliable method to ensure there is no substantial superposition of clumps is to use high - density ( close to the maximum density ) tracers .
nevertheless , even in those situations superposition can occur .
* larson s ( 1981 ) mean - density to size relationship does not exist in physical space , but does occur in our simulated observations .
it appears to be an observational artifact , whose explanation we discuss .
* larson s velocity - dispersion to size relationship is reproduced in both physical clumps and simulated observations , though with substantial scatter in the slope .
more dynamical range in the simulations would probably help , although our clumps already span an order of magnitude in size with half an order of magnitude in scatter .
* the mass spectra converge to a particular shape only when we use large number of contours ( above 32 for the high mass end , but above 64 for the whole spectrum ) , corresponding to an extraordinarily large signal - to - noise ratio in the observations . rather than a single power law
, they seem to follow a log - normal distribution .
ballesteros - paredes , j. & vzquez - semadeni , e. 1997 , in star formation near and far , eds .
holt & l.g .
aip conference series 393 , 81 ballesteros - paredes , j. , vzquez - semadeni , e. , & scalo , j. 1999 , , 515 , 286 sheth , k. , vogel , s. n. , wilson , c. d. , and dame , t. m. 2000 , in `` the interstellar medium in m31 and m33 '' , proceedings 232 .
we - heraeus seminar , 22 - 25 may 2000 , bad honnef , germany . edited by elly m. berkhuijsen , rainer beck , and rene a. m. walterbos .
( shaker : aachen ) 2000 , p. 37 .
vazquez - semadeni , e. , ostriker , e. c. , passot , t. , gammie , c. f. , & stone , j. m. in 2000 , protostars and planets iv , eds mannings , v. , boss , a.p . , russell , s. s. ( tucson : university of arizona press ) , p. 3
williams , j. p. , de geus , e. j. , & blitz , l. 1994 . , 428 , 693 rrrrr ma81 & 0.1 & 8 & 0.1 & 0.68 + mc81 & 1 & 8 & 0.1 & 1.21 + mc85 & 1 & 8 & 0.5 & 1.56 + mc45 & 1 & 4 & 0.5 & 1.83 + ma4x & 0.1 & 4 & 1 & 1.27 + mc4x & 1 & 4 & 1 & 2.21 + mc41 & 1 & 4 & 0.1 & 1.35 + me21 & 10 & 2 & 0.1 & 2.32 + hc4 & 1 & 4 & 0 & 1.97 + hc8 & 1 & 8 & 0 & 1.40 + hc8 - 256 & 1 & 8 & 0 & 2.00 + ha8 & 0.1 & 8 & 0 & .71 + hb8 & 0.3 & 8 & 0 & 1.05 + hd8 & 3 & 8 & 0 & 1.82 + he8 & 10 & 8 & 0 & 1.97 + he4 & 10 & 4 & 0 & 2.39 + hc2 & 1 & 2 & 0 & 2.21 + he2 & 10 & 2 & 0 & 2.50 r c c c r r c @xmath8co(1 - 0 ) & @xmath90 & @xmath91 & 0.112 & 1.43 @xmath92 & 650 cm3 & 1,2 + @xmath8co(2 - 1 ) & @xmath93 & @xmath94 & 0.112 & 1.43@xmath95 & 6200 cm3 & 1,2 + cs(1 - 0 ) & @xmath96 & @xmath97 & 1.957 & 1.0 @xmath98 & 18000 cm3 & 1,2 + | we examine the question of how well the physical properties of clumps in turbulent molecular clouds can be determined by measurements of observed clump structures . to do this ,
we compare simulated observations of three - dimensional numerical models of isothermal , magnetized , supersonic turbulence to the actual physical structure of the models .
we begin by determining how changing the parameters of the turbulence changes the structure of the simulations .
stronger driving produces greater density fluctuations , and longer wavelength driving produces larger structures .
magnetic fields have a less pronounced effect on structure , and one that is not monotonic with field strength .
aligned structures are seen only with low - density tracers , and when the intensity of the field is large . comparing different regions with the same tracers ( or conversely
, the same region with different tracers ) can give information about the physical conditions of the region .
in particular , different density tracers can help determine the size of the density fluctuations and thus the strength of the driving .
nevertheless , velocity superposition of multiple physical clumps can fully obscure the physical properties of those clumps , and short wavelength ( compared to the size of the region under analysis ) driving worsens this effect .
we then compare larson s relationships and mass spectra in physical and observational space for the same structure dataset .
we confirm previous claims that the mean density - size relationship is an observational artifact due to limited dynamical range in column density : it is the inevitable consequence presence of a lower cutoff in column density . the velocity dispersion - size relationship , on the other hand ,
is reproduced in both physical and observed clumps , although with substantial scatter in the derived slope , consistent with observations .
finally , we compute the mass spectra for the models and compare them to mass spectra derived from simulated observations of the models .
we show that , when we look for clumps with high enough resolution , they both converge to the same shape .
this shape appears to be log - normal , however , rather than the power - law function usually used in the literature . |
a scale - invariant spectrum of gravitational radiation is a key prediction of inflation @xcite . measuring the ratio of the amplitude of gravitational radiation to the amplitude of density perturbations , @xmath0 , would be a direct probe of the inflationary energy scale . such a measurement has eluded observation in the cosmic microwave background ( cmb ) thus far @xcite and represents one of the key goals of future cmb observational missions @xcite . on the other hand ,
some authors have argued that phase transitions can mimic the scale - invariant inflationary signal @xcite .
the mechanism through which this is accomplished is not trivial .
the phase transition itself is not the source .
rather , energy is deposited into gravitational radiation via the self - ordering of fields as regions of spacetime become causally connected .
the process begins with @xmath1 scalar fields , @xmath2 , subject to a temperature dependent potential , @xmath3 . at high temperatures
the potential looks quadratic with a single minimum at the origin .
this state has @xmath4 symmetry since the state is symmetric any any rotation in the @xmath1-dimensional field space .
once the potential drops below a critical temperature , the field acquires a vacuum expectation value ( vev ) .
this process spontaneously breaks the @xmath5 to a @xmath6 symmetry .
the transition can be very fast , with a duration , @xmath7 , that is much smaller than the hubble time , @xmath8 .
the field configuration of each hubble volume will lie somewhere in the vacuum state , however each causally disconnected volume will be independent of the others
. this process can be quite smooth , and there is no _ a priori _ reason to believe such a process radiates .
however , if the universe is radiation ( and later matter ) dominated after the transition , the growth of the comoving hubble scale continuously brings previously causally disconnected regions into contact .
regions thus acquire field gradients on the scale of the hubble horizon .
these gradients generate anisotropic stresses that source gravitational waves .
the growth of the horizon acts as a high - pass filter on the spectrum of gravitational waves , _ freezing out _ large - wavelength modes until they enter the horizon and become dynamical . as the horizon grows power
is distributed at all scales between the physical horizon size at the time of the transition , @xmath9 , and the horizon size at the time of observation . in all previous studies
, authors have relied on large-@xmath1 approximations in order to calculate the gravitational wave signal .
here we make no such approximation and approach the problem numerically .
we place @xmath1 scalar fields on a lattice evolving in a friedmann - lematre - robertson - walker background .
each lattice point is initialized with a field value derived from thermal initial conditions .
if we allow the lattice spacing to be @xmath10 at the beginning of the simulation each lattice point will settle into an independent position in the vacuum manifold .
as the simulation progresses , the fields will evolve and align themselves .
this self - ordering produces anisotropic stresses that source gravitational waves .
this paper is organized as follows . in section [ gsb ]
we will introduce a toy model of spontaneous symmetry breaking .
we outline our computational methods in section [ gwp ] and present the spectra we calculate in our simulations . in section [ discussion ] we will comment on the differences between the spectra we predict from self - ordering and the spectrum produced during inflation .
the gut scale naturally arises from particle physics as the scale at which the electroweak and strong couplings are of the same order of magnitude .
it is likely that these forces are unified under some larger symmetry above the gut scale and that this symmetry is broken as the universe cools .
nevertheless , the nature of the gut symmetry and how it is broken remains unknown
. the transition could be first - order , in which case bubbles of the broken phase nucleate and coalesce . in this model
the phase transition happens very rapidly the entirety of the universe can end up in a unique state in less than a hubble time .
this process is likely to produce gravitational radiation @xcite as bubbles collide and coalesce .
conversely , the phase transition could be second - order . in this case the field smoothly transitions to the broken phase as the temperature of the universe drops .
if the broken phase is not unique , that is if the vacuum state has some symmetry with respect to the field configuration , the effects of the _ existence _ of this phase transition can lead to observational effects for many hubble times .
we begin with two assumptions : ( 1 ) that the universe is radiation dominated at the time when the phase transition occurs , and ( 2 ) that the energy associated with the fields undergoing the phase transition is some small fraction , @xmath11 , of the total energy density at the time of the transition , @xmath12 . the total density , at any time , is @xmath13 where @xmath14 and @xmath15 since the universe is necessarily dominated by the radiation energy - density , we will only consider cases where @xmath16 so that the universe is well described by assuming @xmath17 . the potential in eq .
( [ tdeppot ] ) is temperature dependent . to leading order in temperature , @xmath18 where @xmath19 .
the temperature dependent effective mass can be parameterized by @xmath20 at temperatures higher than the critical temperature , @xmath21 , the effective mass is positive , the potential has a unique minimum at the origin , and this minimum has full @xmath4 symmetry , and at the origin @xmath22 after the phase transition the potential has an @xmath6 symmetric vev @xmath23 the phase transition occurs at the critical temperature , @xmath21 , when the effective mass of the field vanishes .
although the field has a mean value , @xmath24 , there is a variance associated with this value , @xmath25 that sets the distribution of field values at the time of the transition .
we can assume that at this time , each hubble volume ( sphere of radius @xmath9 ) has a homogeneous field value , drawn from a gaussian distribution ( see appendix [ thermal ] ) @xmath26 and @xmath27 where the second equality comes from setting @xmath28 .
the temperature of the universe at the beginning of the simulation , @xmath21 , is related to the energy density of the universe at that time , @xmath29 where @xmath30 is the number of ultra - relativistic degrees of freedom at the time of the phase transition .
we take @xmath31 . the average energy density in the field is well approximated by setting @xmath24 in eq .
( [ potential ] ) , @xmath32 which is some fraction , @xmath11 , of the total energy density @xmath33 this constrains the value of the vev , @xmath34 this provides us with a good self - consistency check , namely , that the variance of the fluctuations of the filed are small compared to the vev , @xmath35 .
this ratio is given by @xmath36 which is , in general , less that one if @xmath37 is significantly below @xmath38 and @xmath11 and @xmath39 are of similar order .
evolving scalar fields on a discrete lattice is now a mature field of study .
scalar fields were first introduced to the lattice , in a cosmological context , by latticeeasy @xcite and later by defrost @xcite to study the non - linear dynamics of preheating after inflation .
more recently the authors of @xcite re - framed the question by moving the fundamental description of the fields from configuration space to momentum space .
even more recently the author of @xcite introduced a versatile code that allows the user more control over the integrating scheme .
here we chose to use latticeeasy since we are interested in sub- and super - horizon scales , large lattices with efficient storage and a specific associated potential , eq .
( [ potential ] ) .
this software natively evolves scalar fields according to the klein - gordon equation in an expanding background , @xmath40 where we work in units where @xmath41 .
the homogeneous background evolution is determined by @xmath42 where @xmath43 is the homogeneous , average , energy density at time @xmath44 .
we couple latticeeasy to a code that evolves the metric perturbation using the methods of @xcite .
since the lattice realizes the fields at discrete values of time , it is most convenient to perturb the metric , @xmath45 , in a synchronous gauge @xmath46dx^idx^j.\ ] ] additionally , the radiative part of @xmath45 obeys the transverse - traceless conditions @xmath47 the radiative perturbations obey sourced klein - gordon equations @xmath48 where the source term , @xmath49 is the transverse - traceless projection of the anisotropic stress tensor , @xmath50 we specify our model eq . ( [ potential ] ) and allow latticeeasy to evolve the fields and the scale factor .
we can then calculate the source term of eq .
( [ equationofmotion ] ) , and evolve the six metric perturbations , @xmath51 .
we can always check our numerical stability by checking to see if the metric perturbations are still transverse - traceless ; transverse - traceless metric perturbations require _ both _ a transverse - traceless source and accurate evolution . at any point during the simulation
, we can calculate the power spectrum of gravitational radiation .
the stress - energy associated with metric perturbations is @xcite , @xmath52 where the brackets denote a spatial average over at least a few wavelengths .
the @xmath53 component is the energy density , @xmath54 where @xmath55 .
finally , we can invoke parseval s theorem ( see @xcite ) to rewrite eq .
( [ gwdensity ] ) as @xmath56 where @xmath3 is the comoving volume over which the spatial average is being performed .
we can then write @xmath57 which can be transferred to present - day amplitude and frequency by @xcite , @xmath58 where the @xmath59 and @xmath60 subscripts denote quantities defined today and the end of our simulations , respectively .
we also keep the convention that @xmath61 absorbs the uncertainty in the present value of the hubble parameter , @xmath62 is the current fraction of the energy density in the form of radiation , and @xmath63 is the total energy density at the end of our simulations . the ratio , @xmath64 , is the number of degrees of freedom today to the number of degrees of freedom at matter / radiation equality .
we approximate @xmath65 .
the first major difference between the _ structure _ of the gravitational - wave spectrum from self - ordering and that predicted by inflation is the lack of power at high - frequencies .
this cut - off feature exists because we only considering larger then hubble length fluctuations as sources of gravitational waves as there will be no / short lived gradient terms to source the gravitational waves inside the hubble volumes by second order phase transitions .
.,width=288 ] , @xmath66 , @xmath67.,width=288 ] we see in @xcite that the cut - off frequency is related to the hubble length at the time when the gravitational wave is generated @xmath68 this , along with and the first friedmann equation , @xmath69 allows us to determine where the cut - off should appear @xmath70 for example , with @xmath71 we expect the cutoff to be at @xmath72 which agrees with the results of our simulation , shown in fig .
[ fig : ncomparison ] .
the characteristic amplitude and scale invariant nature of the spectrum was expected from the analytical approach of @xcite and @xcite .
it is interesting to note that the spectrum is scale invariant for all the cases in fig .
[ fig : ncomparison ] , including those for @xmath73 ; the analytical methods employed by @xcite assumed that @xmath1 is large . in fig .
[ fig : soucomparison ] we observe two important scaling effects .
first , we recover the fact that the high - frequency cutoff predicted by eq . [ peakeq ] scales with @xmath74 .
we also see that the amplitude of the signal is proportional to the energy density of the universe at the time of the transition , @xmath75 this scaling is also suggested predicted in @xcite and @xcite .
a final step is to compare the numerical results here with the analytic arguments of @xcite and @xcite . in both cases
the authors use the model presented here to estimate the gravitational wave signal from reordering .
the two sets of authors use slightly different parameterizations of the model ; however , all authors arrive at the conclusion that there should be a scale - invariant gravitational wave spectrum from this transition in @xcite , the authors estimate the power in gravitational waves from field reordering to be ( eq . (
10 ) in @xcite ) @xmath76 where @xmath77 is calculated numerically , the hubble constant today is @xmath78 , @xmath79 is conformal time today and we ve modified the normalization constant @xmath80 to reflect a convention choice in @xcite and a typographical error in @xcite .
the strain power , @xmath81 , ends up being a function only of @xmath82 .
it peaks around @xmath83 , although this corresponds to modes that entered the horizon during matter domination . to ensure that we re identifying modes that entered during radiation domination
, we choose a mode , @xmath84 , @xmath85 where we note that gravitational waves scale as a constant fraction of @xmath86 after matter - radiation equality , the subscript @xmath87 indicates evaluating quantities at the time of radiation - matter equality . we can read off @xmath88 from fig . 1 in @xcite , @xmath89 so we can estimate the total gravitational wave energy per octave , @xmath90 where the physical wavevector now is @xmath91 larger than the physical wavevector at the time of radiation - matter equality .
we can make a crude estimate of the value of conformal time at radiation - matter equality , @xmath92 where the fractional energy densities come from @xcite .
putting eq .
( [ maxpowerkrauss ] ) together with eqs .
( [ estimate2],[conftimenow ] ) , we get an estimate , @xmath93 or , imposing our parameterization of the current - day hubble constant , setting @xmath94 and using eqs .
( [ defofvev],[firstfried ] ) @xmath95 in @xcite , the authors predict a scale - invariant power spectrum ( eq .
( 5.2 ) of @xcite ) @xmath96 where we use our definition of @xmath97 and our parameterization @xmath98 . using eqs . (
[ defofvev],[firstfried ] ) along with @xmath99 , the expression in eq .
( [ fenuresult ] ) reduces to @xmath100 these two estimates should vary from our simulations by one important factor . in both cases ,
the universe is comprised only of the scalar fields . to preserve a radiation - dominated phase during and after the phase transition
, we have , inherently , diluted the source by a factor of @xmath11 which dilutes the analytic estimates eqs .
( [ jestimate],[festimate ] ) by a factor of @xmath101 .
it is worth pointing out that some of the phase transitions we have simulated result in the production of global topological defects . specifically , global strings for @xmath102 , global monopoles @xmath103 , and global textures for @xmath104 . surprisingly , we find that the gravitational radiation produced is consistent with the large @xmath1 approximation even for low values of @xmath1 , where the approximation is not valid ( see our analytic estimates above ) .
the gravitational wave backgrounds produced by global strings and monopoles are larger than those produced by textures .
we will investigate these cases in more detail in a future publication .
phase transitions at high energies are a generic consequence of ( almost ) all models of high energy physics .
since there is no unique model of physics at this scale , we are forced to look for generic observational consequences at this scale .
second order phase transitions will not produce gravitational radiation over one hubble time , yet the reordering of the fields that mediate the transition could produce characteristic gravitational radiation over a wide range of scales . although this signature could be misinterpreted as the gravitational radiation from primordial quantum fluctuations , it might be possible to distinguish the two at very high frequencies
. such a high - frequency detection would also carry information about the energy scale at which the phase transition occurred .
.[tab : specn ] spectral amplitudes as a function of number of fields for simulations with @xmath105 .
the numerical values , @xmath106 , are an average value taken from the simulations , while the values in the second two columns are obtained from eq .
( [ jestimate ] ) or eq .
( [ festimate ] ) . [ cols="^,^,^,^",options="header " , ] we considered the phenomenological model of @xcite and @xcite in which an @xmath107 symmetric false vacuum is dynamically broken into a @xmath108 true vacuum , we find this produces a scale invariant gravitational wave spectrum whose amplitude depends inversely on number of fields .
these results are summarized in table [ tab : specn ] .
our numerical results suggest that the large @xmath1 is not needed to make a scale invariant spectrum .
since the results of @xcite are derived using a large @xmath1 approximation for the amplitude , one does not expect these estimates to be a perfect estimator of the amplitude of the gravitational waves in the low-@xmath1 limit . additionally , the results for varying values of @xmath12 are given in table [ tab : specr ] .
we find that our simulations differ from analytic estimates by only a small amount and are more consistent at large @xmath1 , consistent with the fact that analytic methods are derived from a large @xmath109 expansion .
we thank latham boyle , andrew tolley , harsh mathur and katherine jones - smith for useful discussions .
jtg is supported by the national science foundation , phy-1068080 , and a cottrell college science award from the research corporation .
the work of xs and bv is supported by national science foundation grants phy-0970074 , phy-0955929 , and phy-0758155 , and the university of wisconsin milwaukee research growth initiative .
research at the perimeter institute for theoretical physics is supported by the government of canada through industry canada and by the province of ontario through the ministry of research & innovation .
jtg would also like to thank the university of wisconsin - milwaukee for its generous hospitality while some of this research was completed .
lrp would like to thank the perimeter institute for theoretical physics for their hospitality while some of this work was being completed there .
the probability of the scalar field taking some average value , @xmath110 , is given by @xmath111 \delta(\phi_v - \bar{\phi } ) = \langle \delta(\phi_v-\bar{\phi})\rangle,\ ] ] where @xmath112 $ ] is the probability functional and we ve introduced the volume - averaged field @xmath113 where @xmath114 is a window function introduced for later computational convenience .
we compute eq .
( [ fullprob ] ) using a gaussian approximation @xmath115 with @xmath116 and @xmath117 .
the problem is now to compute the moments of the field @xmath118\left(\frac{1}{v}\int_v d^3x\ , \phi(x ) \right ) \label{firstmoment}\\ \langle\phi_v^2\rangle & = & \int d\phi \mathcal{p}[\phi ] \left(\frac{1}{v^2}\int_v d^3xd^3y\ , \phi(x)\phi(y ) \right)\label{secondmoment } \end{aligned}\ ] ] the full probability functional can be found in , e.g. @xcite .
here we consider only the leading order temperature dependence , which modifies the potential by replacing @xmath119 with a temperature dependent term , @xmath120 , where @xmath121 .
choosing a temperature above @xmath122 puts us in a symmetric phase of the effective potential .
we then have @xmath123 & \approx & \frac{1}{z } \exp{\left(-\beta h_{\mathrm{eff}}[\phi ] \right)}\\ & = & \frac{1}{z } \exp{\left(\frac{-\beta}{2}\int d^3x\ , \phi(x)(\nabla + m_{\mathrm{eff}}(\beta)^2)\phi(x)\right)}.\end{aligned}\ ] ] henceforth we write @xmath124 . defining @xmath125 & = & \int d\phi \exp\bigl(\frac{-\beta}{2}\int d^3x\ , \phi(x)(\nabla + m_{\rm eff}^2)\phi(x)\\ & \phantom{=}&\phantom{\int d\phi\exp -\beta\int d^3x } + j(x)\phi(x)\bigr ) \\ & = & z \exp{\left(\frac{1}{2 \beta } \int \int d^3x d^3y j(x ) k^{-1}(x - y ) j(y)\right ) } , \end{aligned}\ ] ] with @xmath126 allows us to write the moments of the field as @xmath127\bigr|_{j=0 } \\
\langle\phi_v^2\rangle & = & \frac{1}{v}\int_{-\infty}^{\infty } d^3xd^3y\ , i(x)i(y)\nonumber\\ & \phantom{= } & \phantom{\frac{1}{v}\int_{-\infty}^{\infty}}\times\frac{\delta}{\delta j(x)}\frac{\delta}{\delta j(y)}w[j]\bigr|_{j=0}.\end{aligned}\ ] ] choosing a gaussian window such that @xmath128 and evaluating the integrals leads to @xmath129 putting this all together we have @xmath130 a. a. starobinsky , jetp lett . *
30 * , 682 ( 1979 ) [ pisma zh .
fiz . * 30 * , 719 ( 1979 ) ] .
v. a. rubakov , m. v. sazhin and a. v. veryaskin , phys .
b * 115 * , 189 ( 1982 ) .
e. komatsu _ et al .
_ [ wmap collaboration ] , astrophys . j. suppl . * 192 * , 18 ( 2011 ) [ arxiv:1001.4538 [ astro-ph.co ] ] .
[ planck collaboration ] , arxiv : astro - ph/0604069 .
l. m. krauss , phys .
b * 284 * , 229 ( 1992 ) .
k. jones - smith , l. m. krauss and h. mathur , phys .
* 100 * , 131302 ( 2008 ) [ arxiv:0712.0778 [ astro - ph ] ] .
e. fenu , d. g. figueroa , r. durrer and j. garcia - bellido , jcap * 0910 * , 005 ( 2009 ) [ arxiv:0908.0425 [ astro-ph.co ] ] .
e. witten , phys .
d * 30 * , 272 ( 1984 ) .
a. kosowsky , m. s. turner and r. watkins , phys .
d * 45 * , 4514 ( 1992 ) .
m. kamionkowski , a. kosowsky , m. s. turner , phys .
rev . * d49 * , 2837 - 2851 ( 1994 ) .
[ astro - ph/9310044 ] .
r. easther , j. t. giblin , jr and e. a. lim , phys . rev
. lett . * 99 * , 221301 ( 2007 ) [ arxiv : astro - ph/0612294 ] .
r. easther , j. t. giblin , jr and e. a. lim , phys . rev .
d * 77 * , 103519 ( 2008 ) [ arxiv:0712.2991 [ astro - ph ] ] . c. w. misner , k. s. thorne , j. a. wheeler , san francisco 1973 , 1279p .
l. m. krauss , k. jones - smith , h. mathur and j. dent , phys .
d * 82 * , 044001 ( 2010 ) [ arxiv:1003.1735 [ astro-ph.co ] ] .
m. hindmarsh , r. j. rivers , nucl .
phys . * b417 * , 506 - 526 ( 1994 ) .
[ hep - ph/9307253 ] . | global second - order phase transitions are expected to produce scale - invariant gravitational wave spectra . in this manuscript
we explore the dynamics of a symmetry - breaking phase transition using lattice simulations .
we explicitly calculate the stochastic gravitational wave background produced during the transition and subsequent self - ordering phase .
we comment on this signal as it compares to the scale - invariant spectrum produced during inflation . |
interference plays a crucial role for both quantum physics and classical wave theories . in quantum optics
quantum interferences and superimposed electromagnetic field components occur simultaneously .
therefore , a proper analysis and determination of the character of interference patterns is indispensable for separating classical wave phenomena from quantum effects .
as photons reflect the particle nature of the electromagnetic field , their generation and detection are of fundamental interest @xcite . in the low photon number regime , detectors are often based on avalanche photodiodes ( apd ) in the geiger mode .
apds produce a `` click '' for any number of absorbed photons and remain silent otherwise , _
i.e. _ `` no - click '' .
the incident light field can be split into fields with equal intensities , each being measured with an apd , to extract information beyond the binary one .
realizations are time- and spatial - multiplexed detectors and equally illuminated arrays of apds ; _ cf .
_ @xcite for recent experiments . in fig .
[ fig : setup ] , two spatial - multiplexing schemes labeled with `` 1 '' and `` 2 '' are considered .
click counting devices have been applied in quantum metrology @xcite , or for the determination of entanglement @xcite .
recently , weak - field homodyning with multiplexing detectors has been implemented to probe quantum features of light @xcite .
the click counting statistics @xmath0 for a measurement with @xmath1 apds is of the form @xmath2 where @xmath3 is the number of clicks , @xmath1 the number of apds , and @xmath4 the normal ordering prescription @xcite .
this quantum version of a binomial statistics is described by the operator @xmath5{:}$ ] , whose expectation value is the probability to record a click with a single apd . herein @xmath6 is the photon number operator , @xmath7 the quantum efficiency , and @xmath8 the dark count rate of each apd .
quantum properties , _
e.g. _ sub - binomial light , can be verified with such detection schemes @xcite . moreover , higher order correlations , nonlinear absorption processes , multi - time correlations , and state engineering protocols have been investigated for these click counters @xcite . in the high intensity regime , _
i.e. _ a few - photon approximation is not valid , the wave nature of the electromagnetic field is often studied by balanced homodyne detection ( bhd ) which is based on the photoelectric detection theory @xcite . a signal ( si )
is mixed on a beam splitter with a much stronger local oscillator ( lo ) while controlling the relative phase .
interferometric measurements , such as bhd , have been used to verify phase dependent nonclassical phenomena @xcite . in this scenario
, one would typically not apply single - photon counters because the total intensity of lo and si is considered to exceed the capabilities of click detection . instead
, the outputs fields are measured with a detector which produces an electric current being proportional to the intensity of the incident light field .
applications have been established in quantum metrology @xcite and weak lo homodyning @xcite .
a nonclassical light field may be characterized by the non - existence a positive semi - definite glauber - sudarshan @xmath9 representation @xcite .
moment based criteria in terms of field quadratures have been derived to verify such quantum correlations using bhd @xcite .
nonclassical light fields characterized in such a form , such as squeezed states , serve as a fundamental resource in applications which require a superior phase determination , for example , for gravitational wave detectors @xcite . in order to verify other correlations , _
e.g. _ , entanglement , it is indispensable to perform a careful detector analysis @xcite .
a phase sensitive detection theory does not exist for click counting devices , so that phase sensitive effects are nonaccessible in this regime . in the present contribution
we formulate a bhd - type phase resolving click counting theory , which unifies the detection of the particle and the wave nature of quantum radiation fields .
this yields a new observable a nonlinear quadrature operator as the key element of our approach .
we provide a hierarchy of nonclassicality criteria to uncover phase sensitive quantum effects using on - off detector systems .
direct sampling formulas are given which allow to implement our theory in experiments .
in fig . [ fig : setup ] we describe the bhd setup using click counting devices .
the si field and the lo are combined on a 50:50 beam splitter .
we assume that the lo and the si are single mode fields with a perfect overlap . for the time being , we also say that the lo is in a perfect coherent state given by the coherent amplitude @xmath10 .
the output beams are individually detected with a system of on - off detectors , which yields a joint click counting statistics @xmath11 @xcite . for simplicity , we assume that both detector systems , 1 and 2 , have identical characteristics , _
i.e. _ , number of apds @xmath12 , quantum efficiencies @xmath13 , and dark count rates @xmath14 .
finally , the difference count rate @xmath15 can be obtained as @xmath16 this difference click counting statistics
depending on the phase of the lo , @xmath17 is shown for a coherent si state @xmath18 in fig .
[ fig : clickstatistics ] .
note that a low intensity approximation with photoelectric detectors , having a poissonian form of statistics , is impossible in this case since the mean number of photons from si and lo is in the same order as the number of diodes @xcite .
in general , we will refer to a high intensity if the total intensity of lo and si exceeds the total number of apds , @xmath19 . in this high intensity regime ,
the click counting statistics is valid even if the probability to have multiple photons at one detector significantly differs from zero .
now , the question arises how to infer quantum properties from this measurement .
since moment - based criteria turn out to be a fruitful approach , we may initially define a nonlinear quadrature operator @xmath20 through @xmath21 where @xmath22{:}$ ] describe the first(second ) detector , respectively .
normally ordered powers of the operator @xmath20 will be used later to uncover quantum effects . using the beam splitter transform , @xmath23 and @xmath24 , we rewrite eq . for the coherent lo , @xmath10 , as @xmath25\!\ !
{ -}\exp\!\left[-\!\left(\!\frac{\eta}{n}\hat a_1^\dagger\hat a_1{+}\nu\!\right)\!\right ] \!{:}\!\right\rangle \\{=}&2 n{\rm e}^{-\frac{\eta r^2}{2n}-\nu } \left\langle\!{:}\,{\rm e}^{-\frac{\eta}{2n}\hat n}\sinh\left[\frac{\eta r}{2n}\hat x(\varphi)\right]{:}\!\right\rangle , \end{aligned}\ ] ] with the photon number @xmath26 and the linear quadrature @xmath27 of the signal field .
we observe several features of the click quadrature operator @xmath20 .
first , it has the intensity dependent contribution , @xmath28 $ ] , which is limiting the range of possible expectation values .
this makes sense because for any si and lo power the expectation value must not exceed the values @xmath29 , _ cf .
_ the right hand side in .
second , the quadrature @xmath20 is nonlinearly related to the si s true quadrature @xmath30 in terms of a hyperbolic sine . in fig .
[ fig : nonlinquadrature ] , we plot the expectation value for a coherent si with different intensities to show this nonlinear behavior .
finally we can analyze the limit @xmath31 , @xmath32 which corresponds to the linear quadrature being the result of bhd with photoelectric detection theory .
alternatively , a first order taylor expansion in terms of low efficiencies @xmath33 yields the same result .
moreover , the second order term includes intensity field correlations , @xmath34 , which have been studied in the context of nonclassicality determination @xcite . the fact that we can get these number - amplitude cross correlation highlights the nonlinear character of @xmath20 in eq . .
let us consider a nonclassical signal of @xmath35-photons superimposed with vacuum , @xmath36 being the single mode version of the so - called n00n state @xcite .
figure [ fig : photonquadrature ] shows the phase dependency of these states , @xmath37 , with mean photon numbers @xmath38 . the periodicity depends on the number of photons , which is of a particular interest for quantum metrology @xcite .
is shown for the states in eq .
, @xmath37 ( solid , dashed , dotted ) . for a better comparison
the curves are scaled with @xmath39 : @xmath40 for @xmath37 , respectively .
the periods for increasing @xmath35 reduce to @xmath41 for odd @xmath35 .
the detection parameters are @xmath42 , @xmath43 , @xmath44 , and @xmath45 .
, width=245 ]
in order to formulate nonclassicality conditions , we may define the matrix of click quadrature operator moments , @xmath46 with @xmath47 being the floor function . since this matrix of moments is formulated in terms of normally ordered operator powers , it has to be positive semi - definite for any classical light field ; _ cf .
e.g. _ , @xcite .
therefore , we can formulate sufficient nonclassicality criteria as follows .
a light field is nonclassical , if for a choice of indices holds that @xmath48<0 , \end{aligned}\ ] ] where the set @xmath49 describes the rows and columns of the considered minor . for states with a positive semi - definite glauber - sudarshan @xmath9 functions ,
these minors are necessarily non - negative .
note that @xmath50 and @xmath51 . to guarantee the direct applicability in experiments , we formulate sampling formulas to determine the moments @xmath52 directly from the measured joint click counting statistics @xmath53 . using the generating function approach in ref .
@xcite , @xmath54 we obtain @xmath55^m{:}\rangle \\ = & \sum_{j=0}^m\binom{m}{j}(-1)^{m - j}n^m\langle { : } \hat\pi_1^j\hat\pi_2^{m - j}{:}\rangle \\\nonumber=&\sum_{j=0}^m\sum_{k_1=j}^n\sum_{k_2=m - j}^n(-1)^{m - j}n^m\frac{\binom{m}{j}\binom{k_1}{j}\binom{k_2}{m - j}}{\binom{n}{j}\binom{n}{m - j } } c_{k_1,k_2}. \end{aligned}\ ] ] this renders it possible to certify nonclassical effects from the experimentally obtained counting statistics without time consuming data post processing . because of its relevance in physics , we select from the hierarchy of nonclassicality conditions the second order one . for this case ,
@xmath56 , we get @xmath57 ^ 2{:}\rangle \end{pmatrix}=\langle{:}[\delta\hat x(\varphi)]^2{:}\rangle . \end{aligned}\ ] ] the vacuum field yields @xmath58 .
hence , inequality can be formally expressed in the same form as in the case of standard bhd , _
i.e. _ : _ whenever the variance is below the variance of vacuum fluctuations , we have a nonclassical light field _ ; or , equivalently , @xmath59
^ 2{:}\rangle<\langle{:}[\delta\hat x(\varphi)]^2{:}\rangle_{\rm vac}$ ] . .
vacuum fluctuations yield the bound @xmath59 ^ 2{:}\rangle_{\rm vac}=0 $ ] .
the detector characteristics is @xmath42 , @xmath43 , and @xmath44 .
the lo intensity is fixed , @xmath45 , whereas the si intensity goes up as @xmath60 increases .
eventually , the signal intensity even exceeds the lo intensity . still , nonlinear squeezing can be observed .
, width=302 ] as an example , we study a squeezed vacuum si state , @xmath61|{\rm vac}\rangle \end{aligned}\ ] ] characterized by the squeezing parameter @xmath62 .
figure [ fig : variancesqueezed ] shows the nonlinear quadrature variance for @xmath63(@xmath64 ) corresponding to the squeezed(anti - squeezed ) quadrature , respectively .
we observe that even in the case of imperfect detection and a small number of apds , we can identify squeezing through condition for all parameters @xmath60 .
it is also worth mentioning that the minimal normally ordered variance is @xmath65 . due to continuity
, the verification of quantum features for @xmath43 implies that this nonclassicality probe is also applicable for efficiencies below this value .
for instance , we identified a relatively small squeezing for all @xmath60 values even for @xmath66 , with a minimum of @xmath59 ^ 2{:}\rangle_{\rm min}\approx -7\cdot10^{-4}$ ] .
another interesting feature is a saturation effect in fig .
[ fig : variancesqueezed ] , @xmath67 yields @xmath59 ^ 2{:}\rangle=0 $ ] .
this is due to the fact that the intensity in such a case is so high that all apds click all the time , which can be also achieved with a strong coherent si , @xmath18 with @xmath68 , being a classical state .
hence the limits of strong coherent light and infinitively squeezed vacuum can not be discriminated from each other .
figure [ fig : variancesqueezed ] suggests that bounds to the linear squeezing , @xmath69 ^ 2\rangle={\rm e}^{-2\xi}<1=\langle [ \delta\hat x(\varphi)]^2\rangle_{\rm vac } , \end{aligned}\ ] ] might be obtained from the nonlinear squeezing value . at least in the considered case of the pure states
, we can use the plot to retrieve the parameter @xmath60 from the nonlinear squeezing and anti - squeezing .
this enables us to indirectly infer the linear quadrature squeezing levels using click counters .
however , the case of mixed states requires further studies .
another example is presented for the state in eq . using a higher - order nonclassicality condition ; here @xmath70 . in fig .
[ fig : highercorrelationefficiency ] , we plotted this minor depending on the phase @xmath71 and the efficiency @xmath7 . since the variance of this state is non - negative , @xmath59 ^ 2{:}\rangle\geq0 $ ] ,
this scenario verifies that higher order correlations are useful to detect nonclassicality .
moreover , for all efficiencies , @xmath72 , exist phase intervals with negativities which increase with increasing efficiency . , is shown depending on the quantum efficiency @xmath7 and the phase @xmath71 .
the other detection parameters are @xmath44 , @xmath45 , and @xmath42 . negativities ( red )
verify the nonclassicality , for some phases and all @xmath7 values .
, width=302 ] so far we considered imperfections of the click detector itself , but not of the lo . in case of the standard bhd , classical lo fluctuation do basically not occur in the difference current of both detectors
. here such fluctuations can be crucial due to the nonlinear structure , _ cf . _
thus we introduce a convolution of the unperturbed quadrature moments @xmath52 with the lo noise distribution @xmath73 and @xmath10 @xmath74 for a fundamental study , we infer gaussian fluctuations , _ cf .
_ @xcite , being subdivided into phase noise @xmath75 and amplitude noise @xmath76 .
the decomposition @xmath77 in a @xmath71-rotated frame yields a gaussian fluctuation distribution as @xmath78}{2\pi\sigma_x\sigma_p } , \end{aligned}\ ] ] having a mean coherent amplitude @xmath79 .
using polar coordinates , the phase noise variance is given by @xmath80 and the amplitude noise variance by @xmath81 . the influence of the lo fluctuations to the verification of nonlinear squeezing is shown in fig .
[ fig : variancephase ] . the amplitude noise ( dotted curve )
solely affects the value of the negativities in comparison with the unperturbed variance ( solid curve ) .
the phase noise ( dashed curve ) additionally diminishes the intervals of certified squeezing .
typically , the phase is much better controlled than assumed for the plot .
however , such high phase diffusion shall underline the applicability in extreme scenarios .
naturally both effects add up as shown by the dot - dashed curve .
, @xmath43 , @xmath44 , @xmath45 , and @xmath82 for a squeezed vacuum state .
the solid line shows @xmath59 ^ 2{:}\rangle$ ] without lo fluctuations .
the dashed graph includes a significant amount of phase noise , @xmath83 ( @xmath84 ) , and the dotted one shows the effect of amplitude noise ( @xmath85 ) of the same order of magnitude as the mean lo amplitude .
the dot - dashed curve combines amplitude and phase noise .
, width=283 ] let us outline the treatment of other possible sources of imperfections , which are not studied in detail in this work .
initially , we assumed a perfect mode overlap between the lo and the si .
however , a mode mismatch might occur in our scheme . in ref .
@xcite , the characterization of the click counting devices has been generalized to a multimode description ( section iv b ) , which can be used to model an incomplete overlap between lo and si . in the same work ,
time - dependent correlations in click detection have been elaborated ( section iv a ) .
this technique is also useful to consider temporal drifts in the relative phase between lo and si , or , in case of time - bin multiplexing , it renders it possible to treat temporal overlaps of wave - packets in different time bins . based on these approaches ,
the influence of a manifold of experimental imperfection can be rigorously studied in addition to the perturbations presented above .
techniques for measuring the particle and the wave nature of quantum light have been unified in the theory of balanced homodyne detection with click counting devices .
its consistent formulation leads to a nonlinear quadrature operator as the basic observable .
its features have been studied for local oscillators in the weak and intermediate intensity regime , for detector imperfections , and fluctuations of the local oscillator . for consistency
, we showed that standard balanced homodyne detection is recovered in proper limits .
a hierarchy of conditions has been derived for determining phase sensitive nonclassical effects .
the second order criterion applies to verify nonlinear squeezing of a squeezed vacuum state for arbitrary squeezing strength .
higher order criteria identify phase - sensitive nonclassicality of a superposition of @xmath35 photons with vacuum . even in the case of quantum efficiencies below @xmath86
, it has been demonstrated that nonclassicality can be uncovered in terms of higher - order correlations . by applying more than one phase sensitive click counting device , it is possible to infer phase sensitive correlations between multimode radiation fields .
our techniques can also be further developed for the aim of applications in quantum information and metrology .
js and wv acknowledge financial support by deutsche forschungsgemeinschaft through sfb 652 . , _ nature photon . _ , * 4 * ( 2010 ) 243 .
a _ , * 83 * ( 2011 ) 031806(r ) .
a _ , * 90 * ( 2014 ) 042105 .
_ , * 104 * ( 2010 ) 063602 . , _ nature photon .
_ , * 6 * ( 2012 ) 364 . , _ phys . rev .
_ , * 105 * ( 2010 ) 253603 . ,
_ nature commun .
_ , * 5 * ( 2014 ) 5584 .
_ , * 99 * ( 2007 ) 223602 . , _ phys . rev
_ , * 101 * ( 2008 ) 233604 . ,
_ phys . rev .
a _ , * 83 * ( 2011 ) 063836 . , _
_ , * 3 * ( 2013 ) 2684 .
* 85 * ( 2000 ) 1349 . ,
_ new j. phys .
_ , * 12 * ( 2010 ) 033042 . ,
_ , * 113 * ( 2014 ) 160401 . ,
a _ , * 85 * ( 2012 ) 023820 .
_ , * 109 * ( 2012 ) 093601 . ,
_ , * 110 * ( 2013 ) 173602 . , _ josa b _ , * 31 * ( 2014 ) b8 . , _ phys
a _ , * 88 * ( 2013 ) 043821 .
a _ , * 89 * ( 2014 ) 043829 .
_ , * 39 * ( 1999 ) 63 . , _ rev .
_ , * 81 * ( 2009 ) 299 . , _ quantum optics _ , ( cambridge university press , cambridge ) 2013 . , _ phys .
a _ , * 79 * ( 2009 ) 022122 . ,
_ new j. phys .
_ , * 12 * ( 2010 ) 113025 . ,
_ opt . express _ , * 22 * ( 2014 ) 11945 .
* 47 * ( 1993 ) 4227 .
_ , * 10 * ( 1963 ) 84 ; sudarshan e. c. g. , _ phys . rev .
_ , * 10 * ( 1963 ) 277 . , _ phys
_ , * 54 * ( 1985 ) 323 . , _ phys
a _ , * 36 * ( 1987 ) 3796 .
_ , * 95 * ( 1993 ) 109 .
a _ , * 71 * ( 2005 ) 011802(r ) ; shchukin e. v. and vogel w. , _ phys .
a _ , * 72 * ( 2005 ) 043808 .
d _ , * 23 * ( 1981 ) 1693 .
, _ nature phys .
_ , * 7 * ( 2011 ) 962 . , _ phys
a _ , * 83 * ( 2011 ) 032119 .
_ , * 67 * ( 1991 ) 2450 ; vogel w. , _ phys . rev .
_ , * 100 * ( 2008 ) 013605 . ,
_ , * 85 * ( 2000 ) 1855 ; foster g. t. , orozco l. a. , castro - beltran h. m. and carmichael h. j. , _ phys .
_ , * 85 * ( 2000 ) 3149 . , _ phys . rev
_ , * 85 * ( 2000 ) 2733 . ,
_ , * 34 * ( 1987 ) 909 . ,
_ phys . rev .
a _ , * 90 * ( 2014 ) 010301(r ) . | driven by single photon detection requirements , the theory of arrays of off - on detectors has been developed and applied .
however for a comprehensive characterization of nonclassicality one also needs phase sensitive properties .
this missing link is introduced with the derived theory of phase sensitive click counting measurements .
it unifies the balanced homodyne detection for high intensities with the click detection in the few photon regime .
we formulate a hierarchy of nonlinear squeezing conditions to probe quantum effects beyond standard squeezing .
imperfections stemming from fluctuations , detector efficiency , and dark count rates are considered .
experimentally accessible sampling formulas are given .
our theory paves the way towards novel applications of light in quantum metrology . |
the idea of universality plays a major role in the modern understanding of critical phenomena .
all the physical systems undergoing a continuous phase transition are believed to fall in a certain number of universality classes , depending on the dimensionality of the system and its symmetry group , but not on the details of the microscopic interactions ( as long as these are short ranged ) .
all the systems in a given universality class display the same critical behavior , meaning that certain dimensionless quantities have the same value for all the physical systems in a given class .
critical indices and universal amplitude ratios are examples of these universal quantities . for gauge theories with a high temperature deconfinement transition ,
the universality hypothesis takes the form of the svetitsky yaffe conjecture , @xcite , which can be formulated as follows : suppose a @xmath6dimensional gauge theory with gauge group @xmath7 has a second order deconfinement transition at a certain temperature @xmath8 ; consider the @xmath9dimensional statistical model with global symmetry group coinciding with the center of the gauge group : if also this model displays a second order phase transition , then the two models belong to the same universality class .
the validity of the conjecture has been well established in several monte carlo analyses ( see _ e.g. _ @xcite and references therein ) .
for the case we are interested in here , namely @xmath10 and @xmath11 , a precise numerical test of the svetitsky
yaffe conjecture can be found in ref.@xcite .
the most obvious application of universality arguments , and in particular of the svetitsky yaffe conjecture , is the prediction of the critical indices .
for example , consider @xmath1 gauge theory ; it undergoes a high temperature deconfinement transition which is known to be second order in both three and four space time dimensions .
the center of @xmath1 is @xmath12 , therefore the dimensionally reduced statistical model is the ising model , which has a second order phase transition both in @xmath11 and @xmath13 . hence the svetitsky yaffe conjecture applies , and we can predict the critical indices of the @xmath1 deconfinement transition in @xmath6 dimensions to coincide with the ones of the ising model in @xmath9 dimensions .
however , the predictive power of universality is certainly not limited to the values of the critical indices . in ref.@xcite
a program has been initiated of systematic exploitation of universality arguments in studying the non perturbative physics of gauge theories .
for example , it was shown that non
trivial results on finite size effects and correlation functions at the deconfinement point can be obtained from universality arguments . in this way it has been possible to evaluate exactly the expectation value of the plaquette operator in presence of static sources , giving some new insight into the structure of the color flux tube for mesons and baryons . in this paper
we continue the program by analysing the finite size scaling behavior of the plaquette operator in @xmath0 @xmath1 gauge theory at the deconfinement temperature . since the @xmath2 ising model is exactly solved , the svetitsky
yaffe conjecture gives in this case exact predictions on finite
size scaling effects .
we write down these predictions for the expectation value of the plaquette operator and we compare them with monte carlo results . the same analysis was performed in ref.@xcite for @xmath12 gauge theory .
the svetitsky yaffe conjecture can be seen as a mapping between observables of the @xmath0 @xmath1 gauge theory at finite temperature and operators of the @xmath2 ising model . the polyakov loop is mapped into the magnetization , while the plaquette operator is mapped into a linear combination of the identity and the energy operators of the statistical model @xcite : @xmath14 where the expectation value in the l.h.s . is taken in the gauge theory , while the ones in the r.h.s .
refer to the two dimensional ising model .
the dots represent contributions from secondary fields in the conformal families of the identity and energy operators , whose contributions are subleading for asymptotically large lattices .
the finite size dependence of the energy expectation value in the two dimensional ising model on a torus is @xcite @xmath15 where @xmath16 and @xmath17 are respectively the area and the modular parameter of the torus .
@xmath18 is the partition function of the ising model at the critical point : @xmath19 ( we follow the notations of ref.@xcite for the jacobi theta functions @xmath20 and the dedekind function @xmath21 ) .
consider now @xmath0 @xmath1 lattice gauge theory regularized on a @xmath22 lattice , with @xmath23 .
for a given @xmath24 the gauge coupling @xmath25 can be tuned to a critical value @xmath26 to simulate the theory at the finite temperature deconfinement phase transition .
precise evaluations of @xmath26 for various values of @xmath24 are available in the literature @xcite .
the universality argument gives us the following prediction for the finite size scaling behavior of the plaquette operator at the deconfinement point @xmath27 where @xmath28 is a function of the modular parameter @xmath17 only : @xmath29 here @xmath30 and @xmath31 are non
universal constants which depend on @xmath24 and must be determined numerically .
once these have been determined , eq .
( [ fseplaq ] ) predicts the expectation value of the plaquette operator for all sizes and shapes of the lattice , _ i.e. _ for all values of @xmath32 and @xmath33 .
the @xmath34 corrections represent the contribution of secondary fields .
therefore eq .
( [ fseplaq ] ) is valid asymptotically for large lattices .
once the constants @xmath30 and @xmath31 have been determined at a given value of @xmath24 , for example through the finite size scaling analysis presented here , all the correlation functions of the plaquette operator at criticality are in principle exactly computable , since they can be readily derived from the corresponding correlators of the energy operator in the @xmath35 ising model .
consider for example the plaquette expectation value in presence of static sources , _
i.e. _ the correlation function of the plaquette with @xmath36 polyakov loops : @xmath37 this is the typical quantity to study if one is interested in the structure of the color flux tube , since it represents the density of action in presence of external sources .
universality then tells us that the connected part of @xmath38 is given by @xmath39 where the correlator on the r.h.s .
is computed in the critical @xmath2 ising model , @xmath40 is the spin operator and @xmath41 is the constant which relates it to the polyakov loop ; @xmath41 can be determined with methods similar to the ones we used here to determine @xmath31 .
therefore the svetitsky yaffe conjecture , together with the exact results of @xmath2 conformal field theory , gives us complete control over the critical behavior of @xmath0 @xmath1 gauge theory at the finite temperature deconfinement transition .
more generally , this holds for every @xmath0 gauge theory whose deconfinement transition is second order , including @xmath42 .
some studies of the flux tube structure following this line were presented in ref .
three dimensional finite temperature gauge theories are simulated by using @xmath43 lattices with @xmath44 .
we have performed our simulations of @xmath1 pure gauge theory with @xmath45 and @xmath46 varying between @xmath47 and @xmath48 , and modular parameter @xmath49 between 1 and 3 .
the boundary conditions are periodic in all three directions . for each value of @xmath24
we have chosen the coupling @xmath26 corresponding to the deconfinement transition point : from ref.@xcite we have @xmath50 the only free parameters in eq .
( [ fseplaq ] ) are the non universal constants @xmath30 and @xmath31 , which depend on the value on @xmath24 and on the kind of plaquette we are considering ( time like or space like ) .
therefore we have a total of four sets of data to be compared with the theoretical prediction : for each of these sets we can perform a two parameter fit with eq .
( [ fseplaq ] ) .
it turns out however that finite size effects on space like plaquettes are of the same order of magnitude than the typical statistical uncertainties of our simulations .
therefore from now on we will consider time like plaquettes only .
it is important to stress that for each value of @xmath24 all the data , corresponding to different values of @xmath51 , are included in the same two
parameter fit .
specifically , for each lattice with sides @xmath52 we define an `` effective area '' @xmath53 so that eq .
( [ fseplaq ] ) becomes @xmath54 and we can fit the plaquette expectation value to a linear function of @xmath55 . the monte carlo results for the plaquette expectation values are reported in tab.1 and tab.2 while the results of the fits for @xmath56 and @xmath57 are reported in tab.3 .
the values of the reduced @xmath58 show that the agreement is very satisfactory . in figs . 1 and 2 the plaquette expectation values are plotted against @xmath55 together with the best fit line .
the same data are plotted against @xmath59 in fig . 3 for @xmath56 .
this shows the crucial importance of including the non trivial part of eq .
( [ fseplaq ] ) , namely its @xmath60dependence .
the same analysis was performed in ref.@xcite for @xmath0 @xmath12 gauge theory , whose finite temperature deconfinement transition is also in the universality class of the @xmath2 ising model .
reported in ref.@xcite for @xmath12 gauge theory are negative : this is due to the fact that the simulation was actually performed in the @xmath0 spin ising model , using @xmath0 duality .
as a consequence , the svetitsky - yaffe mapping between @xmath0 observables and @xmath2 ising operators at criticality includes a duality transformation , which changes the sign of @xmath61 . ]
we have shown that the svetitsky yaffe conjecture , _
i.e. _ universality applied to the deconfinement transition , provides an exact description of finite size effects for @xmath0 @xmath1 gauge theory at the deconfinement point .
this description is obtained by mapping the gauge theory into the @xmath2 ising model , which is in the same universality class , and is exactly solved .
we have compared the predictions obtained from this mapping with monte carlo results , finding complete agreement .
the main motivation for this work was not to verify the validity of the svetitsky
yaffe conjecture , which is by now well established .
our intent is rather to stress that universality arguments provide a powerful , analytical approach to a deeply non perturbative region , namely the deconfinement transition and its neighborhood .
this is especially true for @xmath0 gauge theories , since critical behavior in @xmath2 is completely understood with the techniques of conformal field theory .
1.5 cm : we thank p. cea and l. cosmai for providing us with their efficient monte carlo code
. one of us ( r.f ) thanks also a. papa for helpful comments .
999 b. svetitsky and l. g. yaffe , nucl .
b210 ( 1982 ) , 423 .
j. engels , j. fingberg and m. weber , nucl .
b332 ( 1990 ) 737 .
j. engels , j. fingberg and d. e. miller , nucl .
b387 ( 1992 ) 501 . m. caselle and m. hasenbusch , nucl.phys
. b470 ( 1996 ) 435 . k. pinn and s. vinti , nucl . phys .
( proc . suppl . ) 63 ( 1998 ) 457 , phys .
d57 ( 1998 ) 3923 .
m. teper , phys .
b313 ( 1993 ) 417 . f. gliozzi and s. vinti , nucl .
b ( proc . suppl . )
53 ( 1997 ) 593 . f. gliozzi , in _ quark confinement and the hadron spectrum ii _ ,
como 1996 , n. brambilla and g. m. prosperi eds .
, world scientific ( hep - lat/9609040 ) .
f. gliozzi and p. provero , phys .
d56 ( 1997 ) 1131 .
a. e. ferdinand and h. g. fisher , phys .
185 ( 1969 ) 832 .
p. di francesco , h. saleur and j. b. zuber , nucl .
b290 ( 1987 ) 527 . c. itzykson and j. drouffe , ``
statistical field theory '' , cambridge 1989 , chap . | we show that universality arguments , namely the svetitsky yaffe conjecture , allow one to obtain exact results on the critical behavior of @xmath0 @xmath1 gauge theory at the finite temperature deconfinement transition , through a mapping into the @xmath2 ising model . in particular , we consider the finite size scaling behavior of the plaquette operator , which can be mapped into the energy operator of the @xmath2 ising model . we obtain exact predictions for the dependence of the plaquette expectation value on the size and shape of the lattice and
we compare them to monte carlo results , finding complete agreement .
we discuss the application of this method to the computation of more general correlators of the plaquette operator at criticality , and its relevance to the study of the color flux tube structure .
dftt 30/98 + dfcal - th 4/98 + june 1998 0.5 cm * critical behavior of @xmath0 @xmath1 gauge theory at finite temperature : exact results from universality * + 0.6 cm r. fiore@xmath3 , f. gliozzi@xmath4 and p. provero@xmath4 0.6 cm _ @xmath3 dipartimento di fisica , universit della calabria _ _ istituto nazionale di fisica nucleare , gruppo collegato di cosenza _
_ rende , i87030 cosenza , italy _ 0.2 cm _ @xmath4 dipartimento di fisica teorica delluniversit di torino _ _ istituto nazionale di fisica nucleare , sezione di torino _ _ via p.giuria 1 , i10125 torino , italy _ 0.6 cm .5 cm ' '' '' .3 cm @xmath5_work supported by the ministero italiano delluniversit e della ricerca scientifica e tecnologica _ |
spin - polarized electron beams are useful for a number of applications in various fields of physics @xcite .
they are utilized , for example , to study magnetic properties in condensed matter systems @xcite , electron - exchange processes in atomic collisions @xcite , and the inner structure of the nucleon in deep - inelastic scattering .
spin - polarized electron ( and positron ) beams can also enhance the sensitivity of high - energy physics experiments @xcite .
there are various methods to generate spin - polarized electrons @xcite . at first sight , the most straightforward way would be to split an electron beam into its spin components , just as a stern - gerlach setup does with a beam of atoms .
however , for charged particles , the spin - separating mechanism in an inhomogeneous static magnetic field is hindered by the influence of the lorentz force . instead ,
spin - polarized electrons are produced , for instance , by elastic scattering from unpolarized high-@xmath0 atomic targets .
due to spin - orbit coupling , considerable degrees of polarization are attainable this way though at moderate intensities , because the polarization maxima lie near minima of the scattering cross section .
alternatively , one may exploit the spin - orbit interaction in bound states and photoionize polarized atoms . of great practical importance
is photoelectron emission from gaas cathodes after selective excitation into the conduction band with circularly polarized light . in principle
, also macroscopic laser fields can affect the electron spin . while laser - electron interactions usually are dominated by the coupling of the field to the electron charge , under suitable conditions the electron spin may play a role as well .
spin effects have theoretically been predicted , for instance , in strong - field photoionization of atoms @xcite and , very recently , also observed in experiment for the first time @xcite .
besides , spin - flip transitions were studied theoretically in laser - assisted mott @xcite and multiphoton compton scattering @xcite . in general ,
laser - induced spin effects were found to be rather small , unless the field frequency or intensity is very high . to the best of our knowledge ,
a spin polarizer for free electrons based on laser fields has not been designed yet . in the present paper
, we describe a new method to generate spin - polarized electron beams .
our method relies on coherent electron scattering from laser fields and quantum pathway interferences . in the ideal case ,
the setup is capable of perfectly splitting an incident electron beam into its spin components along the laser magnetic field direction .
thus , the electromagnetic field configuration acts as a stern - gerlach device for free electrons .
the scheme of the spin polarizer is shown in fig .
1 . scheme of the spin - polarizing interferometric beam splitter .
an incident electron beam is first coherently bragg scattered off a bichromatic laser field with frequencies @xmath1 ( red ) and @xmath2 ( blue ) , splitting the beam into two portions .
afterwards , the latter are coherently superposed via scattering from monochromatic standing laser waves . due to quantum interference , the outgoing electron beam is separated into its spin components along the laser magnetic field . ]
coherent electron scattering through the kapitza - dirac effect on the periodic potential generated by laser waves resembles the diffraction of light on a grating , but with the roles of light and matter interchanged @xcite .
the effect has been confirmed experimentally both in the bragg @xcite and diffraction @xcite regimes .
related experiments observed the kapitza - dirac effect on atoms @xcite .
kapitza - dirac scattering can be sensitive to the electron spin @xcite .
in contrast to the original work @xcite , this spin - dependent kapitza - dirac effect relies on a 3-photon process , where two photons are absorbed and one photon is emitted ( or vice versa ) . the interaction may be considered as arising from an @xmath3 term in the hamiltonian , in combination with a @xmath4 term ( which , in general , has to compete with the spin - preserving @xmath5 term @xcite ) .
when the incident electron momentum has no component along the field polarization direction , the three - photon process is rendered possible only by the nonzero spin of the electron .
the latter thus attains a crucial role which is exploited in the current study .
the quantum dynamics of electrons in the nonrelativistic regime , including their spin degree of freedom , is described by pauli s equation . in the presence of an electromagnetic field , described by a vector potential @xmath6 in radiation gauge
, it reads @xmath7 where @xmath8 is the electron wave function as a pauli spinor , @xmath9 the electron mass and @xmath10 its charge .
@xmath11 denotes the 3-vector of pauli matricies .
the monochromatic kapitza - dirac diffraction in the so - called bragg regime can be formulated by a vector potential for a standing wave in the form of @xmath12 the standing wave amplitude is given by @xmath13 , its wave number by @xmath14 and its frequency by @xmath15 .
the phase parameter @xmath16 allows to adjust the positions of the field nodes along the @xmath17-axis .
the polarisation is chosen in @xmath18-direction .
a slowly varying envelope function @xmath19 is introduced to model switching on and off of the laser field .
in contrast , the bichromatic spin - dependent kapitza - dirac effect relies on electron scattering from two counterpropagating linearly polarized waves with frequencies @xmath1 and @xmath20 .
this corresponds to a vector potential @xmath21 \ , .
\label{eqn : vector_potential_bi}\ ] ] here and henceforth , the incident electron momentum @xmath22 is assumed to lie in the @xmath23-@xmath17-plane , being orthogonal to the laser polarization , with @xmath24 .
we shall solve eq .
for the field configuration of fig .
[ fig : skizze ] both by analytical methods and direct numerical integration .
having only @xmath17-dependence in the potentials , pauli s equation becomes effectively one - dimensional in space .
moreover , by taking a temporal average of the pauli hamiltonian for the monochromatic vector potential ( with @xmath25 ) and neglecting constant terms , one obtains a ponderomotive potential @xcite @xmath26 it represents a periodic grating from which the electron beam diffracts .
similarly , by means of a magnus expansion to third order of the pauli hamiltonian with the bichromatic vector potential , one finds a spin - dependent effective ponderomotive potential @xcite @xmath27 the electron wave function @xmath28 may be expanded into momentum eigenstates .
the time evolution is encoded in the pauli spinors @xmath29 $ ] which are quantized along the @xmath17-axis . by imposing energy conservation
, the bragg condition allows only the two momentum eigenmodes with @xmath30 , lying @xmath31 apart , to interact with each other .
we can therefore reduce the ansatz to @xmath32 due to the interaction with the laser fields , a characteristic rabi oscillation dynamics will occur between the two momentum modes . in the relevant four - dimensional subspace
, the potentials are represented by the block matricies @xmath33 with the corresponding rabi frequencies @xmath34 and @xmath35 . since all the involved momentum eigenstates share the same kinetic energy
, we can remove the latter by a gauge transformation .
this corresponds to switching to the interaction picture .
the interaction times with the laser potentials are chosen to yield a quarter ( or half ) rabi cycle in each stage of fig .
[ fig : skizze ] . with @xmath36
, the corresponding time evolution operators thus read @xmath37 for a @xmath38-pulse with the potential @xmath39 , @xmath40 for a @xmath41-pulse and @xmath42 for a @xmath38-pulse with @xmath43 . combining these three interactions in the order sketched by fig .
[ fig : skizze ] results in the time evolution operator @xmath44 for the whole process .
the desired spin filtering effect is achieved when the relative phase is set to @xmath45 . in this case
, the total time evolution operator becomes @xmath46 when acting on initially @xmath23-polarized electron states @xmath47 with @xmath24 , we obtain @xmath48 thus , incident electrons with spin - up along the @xmath23-axis go through the beam splitter with unchanged momentum , whereas spin - down electrons are scattered to the mirrored momentum state @xmath49 .
we can also consider an incident electron with arbitrary spin polarisation , described by a density matrix with vector @xmath50 in the bloch sphere .
then we find @xmath51 where @xmath52 is the projection of @xmath53 along the @xmath23-axis onto the @xmath18-@xmath17-plane . from the upper left and the lower right component we can read off , that the incident electron is deflected with exactly the same probability @xmath54 as having its spin in negative @xmath23-direction .
also the spin for each separate momentum is in a correspondingly @xmath23-polarized pure state @xmath55 . for an unpolarized ensemble of incident electrons ( @xmath56 ) one
obtains @xmath57 thus , the outgoing electrons are divided into one half with @xmath58 and spin - up ( along the @xmath23-axis ) , and another half with @xmath59 and spin - down .
eq . may be interpreted that the initial spin uncertainty is translated into momentum uncertainty , whereas the certainty of momentum is transformed into certainty of a specific spin state for each given momentum state .
in order to corroborate our analytical considerations , we have performed a real - space simulation of an electron wave packet travelling through the electromagnetic field configuration of fig .
[ fig : skizze ] .
initially , the wave packet has a central longitudinal momentum of @xmath60 and a spatial width of @xmath61 ; its spin is oriented along the positive @xmath17-axis ( @xmath62 ) .
note that the spin expectation value along the @xmath23-axis thus vanishes .
figures [ fig : wavepacket_z ] and [ fig : wavepacket_y ] show the time evolution of the wave packet , involving different spin projections . in the first interaction , a vector potential [ see eq . ] with @xmath63 and @xmath64 is switched on for a duration of @xmath65 .
accordingly , the wave packet is partly reflected by the bichromatic kapitza - dirac effect forcing a selective spin flip ( see lower panel in fig . [
fig : wavepacket_z ] ) on the scattered part . at the second interaction , both partial beams
are reflected by a @xmath41-pulse of the type with @xmath66 , @xmath64 and duration @xmath67 without altering their spin state .
after closing the diamond shape , a third kapitza - dirac diffraction ( a @xmath68-pulse of the same kind as before ) acts as a spin - insensitive beam splitter to coherently mix the two partial beams in two outgoing momentum channels . , width @xmath61 and spin along the positive @xmath17-axis . the projections on @xmath62 in the upper and @xmath69 in the lower panel are shown .
at the bottom , the interaction periods with the lasers are marked .
[ fig : wavepacket_z ] ] it is instructive to consider the same time evolution projected on the @xmath70-eigenstates .
one can see in fig .
[ fig : wavepacket_y ] , that the two output channels of fig .
[ fig : wavepacket_z ] are subject to constructive or destructive interference depending on the spin content .
because the corresponding parts of the wave function are superposed coherently , each momentum state is prepared in a specific @xmath70 spin state , in perfect agreement with eq . . , but with the wave function projected on the @xmath70 eigenstates @xmath71 in the upper panel , and @xmath72 in the lower panel .
[ fig : wavepacket_y ] ]
the results presented above have shown that the laser field configuration of fig . [
fig : skizze ] acts as a stern - gerlach magnet for free electrons .
this is an important finding because it fills a gap in the literature on generation of spin - polarized electron beams . a stern - gerlach - like device for free electrons based on macroscopic electromagnetic fields
whose realizability sometimes is even questioned at all provided by laser waves has so far not been known . in view of an experimental implementation , we note that the single - color kapitza - dirac effect in the bragg regime , as required in the second and third stage of fig .
[ fig : skizze ] , has successfully been observed @xcite .
wavelength and @xmath73 duration , focused to @xmath74 height and @xmath75 width , resulting in an intensity of @xmath76 . ]
the most demanding step is the spin - dependent bichromatic kapitza - dirac scattering in the first stage . nevertheless , despite the rather weak spin interaction
, it can be accomplished utilizing nonrelativistic laser intensities . in the example of figs .
[ fig : wavepacket_z ] and [ fig : wavepacket_y ] , the xuv field intensities amount to @xmath77 and @xmath78 , corresponding to small values of the relativistic parameter @xmath79 .
the bichromatic rabi frequency amounts to @xmath80 ( @xmath81 ) , implying a necessary interaction time of @xmath82 .
the latter translates into a laser beam width of @xmath83 , where @xmath84 denotes the transverse component of the electron velocity , and also determines the minimum laser pulse duration @xmath85 .
an electron energy of 30ev is assumed ( @xmath86 ) . if the laser beams have spherical cross section with @xmath87 , we find that the pulse energy is on the order of 50mj .
thus , the required laser parameters for the first scattering step are challenging , but lie rather close to the performance values of present x - ray free - electron lasers such as the lcls ( stanford , california ) .
currently , xuv pulses with @xmath88 duration , up to @xmath89 intensity and few mj pulse energy are available there @xcite .
besides , improved x - ray focussing techniques to reach sub-@xmath90 beam waists have been proposed @xcite . in light of this , while being demanding , the first stage in fig .
[ fig : skizze ] seems to be within experimental reach .
the same conclusion was drawn in @xcite considering laser fields in the optical domain .
the requirements for the second and third scattering stage are much more relaxed . here
the required intensity in the above example is @xmath91 .
it could be obtained by outcoupling a small portion from the main laser beam .
we point out that laser fields of circular polarization might also appear attractive to coherently control the electron spin . in this case , however , the spin - flip transitions still compete with spin - preserving electron scattering because the @xmath5 interaction term can not be avoided . as a result
, the desired spin flips are suppressed . before passing on to the conclusion
, we wish to mention that a spin - polarizing wien filter for electron vortex beams was put forward in @xcite .
it relies on multipolar electric and magnetic field geometries with cylindrical symmetry and a conversion of orbital to spin angular momentum . besides , it is interesting to note that a spin - insensitive , but otherwise similar laser - based interferometric setup as in fig .
[ fig : skizze ] is applied to atomic beams by the gain project for gravitational wave studies @xcite . with regard to electron beams ,
a spin - insensitive interferometric beam splitter exploiting the kapitza - dirac effect has been proposed in @xcite .
a new way of generating spin - polarized electron beams by interaction with intense laser fields has been put forward .
our scheme relies on coherent electron scattering through the kapitza - dirac effect in an interferometric setup of laser beams .
it acts as a stern - gerlach device for free electrons , splitting the incident beam into spin components along the laser magnetic field direction .
free - electron laser facilities appear particularly promising for an experimental test of our predictions .
since the time scale for the spin - sensitive electron interaction is set by a rather slow bichromatic rabi dynamics , low electron energies are advantageous to limit the required spatial extent of the laser fields .
the following strengths of the setup are noteworthy : ( i ) after an electron beam has been sent through , both partial beams are spin - polarized and separatly available for further use .
( ii ) it may equally well serve as spin filter and spin analyzer . (
iii ) by proper synchronization , the scheme offers a novel method of creating and characterizing ultrashort ( e.g. , femtosecond ) spin - polarized electron pulses @xcite .
we finally note that analogies of the kapitza - dirac effect exist in other areas of physics , such as electron scattering from travelling waves in dielectric media @xcite or from optical near - fields in transmission microscopes @xcite .
it is an interesting question whether the spin - polarizing electron beam splitter presented here can be transferred to these related phenomena as well .
we thank s. ahrens for stimulating discussions at the onset of this project . this study was supported by sfb tr18 of the german research foundation ( dfg ) under project no . b11 . | a spin - polarizing electron beam splitter is described which relies on an arrangement of linearly polarized laser waves of nonrelativistic intensity .
an incident electron beam is first coherently scattered off a bichromatic laser field , splitting the beam into two portions , with electron spin and momentum being entangled .
afterwards , the partial beams are coherently superposed in an interferometric setup formed by standing laser waves . as a result
, the outgoing electron beam is separated into its spin components along the laser magnetic field , which is shown by both analytical and numerical solutions of pauli s equation .
the proposed laser field configuration thus exerts the same effect on free electrons like an ordinary stern - gerlach magnet does on atoms . |
thanks to gravitational amplification , galaxy clusters are a powerful tool to probe distant galaxies . indeed , one of the most distant galaxy detected is a gravitational arc at @xmath1=4.92 in the cluster cl1358 + 62 ( @xcite ) .
other very distant lensed sources ( @xmath23 ) have been identified behind cl0939 + 4713 ( @xcite ) and a2390 ( @xcite , @xcite ) leading to important results on the formation history and evolution of galaxies at large redshift .
determining the redshift of arcs and arclets is of great importance as it fixes the angular scales of the optical configuration , hence giving an absolute cluster mass estimate within the arc radius ( _ e.g. _ @xcite , @xcite ) . but
despite the cluster magnification , measuring arclets redshifts is a difficult observational task due to their low surface brightness ( @xcite , @xcite ) , and the lack of strong spectral features in the optical domain for galaxies with @xmath3 .
accurate modeling of cluster potentials based on the analysis of multiple images and weak shear distortions has shown that cluster mass distributions are best described by the sum of a smooth and large scale component ( the cluster ) and the contribution of cluster galaxy halos ( @xcite , @xcite , @xcite ) . for
a given mass distribution , kneib et al ( 1996 ) demonstrated that a redshift can be estimated if one can measure accurately the shape of an individual arclet . in order to validate this method and
study its biases , extensive programs of gravitational arclet spectroscopy have been undertaken .
in particular , ebbels et al .
( 1998 ) have measured 19 redshifts of lensed objects behind abell 2218 .
most of them confirm the lens redshift prediction , and allow the accuracy of the mass model to be improved .
similarly , bzecourt and soucail ( 1997 ) have started the spectroscopy of arclets in abell 2390 , which has been used to constrain the mass distribution in this cluster with a great accuracy ( @xcite ) .
using these accurate cluster mass models and a spectrophotometric description of galaxy evolution ( bruzual & charlot 1993 , @xcite ) , bzecourt et al .
( 1998a ) have predicted the expected arclet number counts and their redshift distribution .
this model presents many improvements with respect to previous work ( _ e.g. _ @xcite , @xcite , @xcite , @xcite ) as it includes many observational limits such as magnitude ranges , surface brightness cut - off or a choice of the optical waveband , and this for any mass distribution , regardless of its complexity .
abell 2218 is the first cluster where the number counts and redshift distribution of the background arclets have been examined in detail ( @xcite , @xcite ) .
we propose in this paper to further extend this study to another well known cluster lens , namely abell 370 .
its mass distribution was first accurately derived by kneib et al .
( 1993 ) [ hereafter k93 ] who showed that the mass model has to be bimodal in shape in order to accommodate the gravitational pair b2b3 .
this was later confirmed by the hst / wfpc1 observation described in smail et al .
( 1996 ) and the x - ray map of the cluster , displaying a bimodal shape compatible with the lens model ( @xcite ) .
we present new observations of the cluster abell 370 in section 2 : a deep hst / wfpc2 image and spectroscopic data .
section 3 discusses the new lensing mass model . in section 4 ,
we use an improved version of the code developed by bzecourt et al .
( 1998a ) to determine the expected counts and redshift distribution of arclets in abell 370 .
our analysis explores two different scenarios of galaxy evolution to study their differences , and compute the depletion curves of background number counts at different wavebands .
discussion , conclusions and further prospects to constrain galaxy evolution through lenses are presented in section 5 . throughout the paper , we use a hubble constant of h@xmath4 , with @xmath5= 0 or 1 and @xmath6 .
abell 370 was observed with the hst / wfpc2 camera with the f675w filter [ i d : 6003 , p.i.:r.p .
saglia ] , resulting in a reasonably deep image with a total exposure time of @xmath7 = 5.6 ksec .
these data were provided by the eso / st - ecf science archive facility , garching , germany and were reduced with standard reduction procedures .
we used the iraf / stsdas packages for the image combinations after centering and cosmic ray cleaning .
the absolute photometry was obtained using magnitude zero - points given in holtzman et al .
the photometry of the field was obtained with the sextractor package ( @xcite ) .
a criterion of 12 contiguous pixels above the given detection threshold was used to identify an object .
the 1 @xmath8 detection limit is @xmath9 mag arcsec@xmath10 . from the magnitude histogram obtained from the catalog
, we estimate the completeness limit to be @xmath11 .
we also built a sample of arclets for the purpose of our study of their photometric and statistical properties . to define an arclet we imposed the following selection criteria : at least 12 contiguous pixels above 2@xmath8 of the local sky level , an axis ratio greater than 2 , a central surface brightness lower than @xmath12 mag .
arcsec@xmath10 and a magnitude range @xmath13 .
the final sample contains 81 arclets and their magnitude histogram is discussed in section 4.2 .
abell 370 ( @xmath1=0.37 ) is a rich optical cluster dominated by two giant elliptical galaxies identified as # 20 and # 35 , following the numbering of mellier et al .
a set of multiple image candidates and gravitational arcs are identified on the wfpc2/f675w image and are displayed in figure [ ima_arclets ] .
their photometric and geometrical properties are also summarized in table [ table_arclets ] .
we now discuss them in detail : * a0 :* near galaxy # 35 is located the spectacular giant arc ( @xmath1=0.724 ) initially detected by soucail et al .
( 1987 ) . from the high resolution
wfpc2 image more details on its morphology are clearly visible , suggesting that it is a gravitationally lensed image of a spiral galaxy ( @xcite ) . * a1 to a6 :* a first set of arclets ( labelled a1 to a6 ) was detected from ground based images by fort et al .
( 1988 ) and further discussed by kneib et al .
most of them are blue objects , but none have a spectroscopic redshift yet .
a5 is the most extended one and presents very blue colors and a strong dimming in the reddest bands , suggesting a young star forming galaxy . despite deep spectroscopic data for a5 ( @xcite )
no significant emission line has been identified suggesting that @xmath14 .
arclets a3 to a6 are probably single images of background lensed galaxies in view of their location in the lens plane .
arclets a1 and a2 may be multiple images of the same source .
* the multiple images b , c and d :* k93 demonstrated that the b2b3 objects ( fort et al . 1988 ) correspond to a gravitational pair with a counter image labeled as b4 . the b2b3 pair and the arc a0
were used to constrain the k93 model which showed a bimodal mass distribution .
this model proposed that objects c1c2c3 and d1d2 , identified on high quality ground - based images , were also multiple image systems . a redshift estimate based on the mass model was proposed for each of the pairs : @xmath15 , @xmath16 and @xmath17 .
the reality of b4 was confirmed _ a posteriori _ by the hst / wfpc1 data ( smail et al .
c3 is likely to be a wrong identification of the counter image of c1c2 , and we denote c4 the correct counter image used in the present model . * the radial arc r :* in the hst / wfpc1 image , smail et al .
( 1996 ) discovered a radial arc candidate r ( r1-r2 ) located close to galaxy # 35 .
they modeled it as a 5-image configuration , and predicted a redshift of @xmath18 using the k93 model .
this radial arc is well identified in the f675w image and is clearly the merging of 2 images . * new multiple images
:* the detailed insight of the mass model and the exquisite hst resolution allow the identification of other multiple image candidates .
we only discuss here the e1e2e3 multiple configuration as it presents a characteristic inversion of parity as expected from lensing theory ( see also @xcite and @xcite for other spectacular examples ) .
other multiple image candidates will be discussed in a forthcoming paper .
crrcrc object & x ( @xmath19 ) & y ( @xmath19 ) & @xmath20 & @xmath21 & r@xmath22 + a1 & 16.35 & 57.30 & 0.74 & -23.8 & 24.83 + a2 & 5.56 & 60.43 & 0.75 & -4.6 & 24.38 + b2 & 8.46 & 21.14 & 0.76 & -0.2 & 23.08 + b3 & 13.80 & 20.58 & 0.74 & -15.7 & 23.23 + b4 & -19.45 & 20.98 & 0.42 & -41.6 & 23.82 + c1 & 6.46 & 6.50 & 0.17 & 4.6 & 24.05 + c2 & 10.19 & 4.96 & 0.10 & 63.2 & 24.00 + c3 & -21.82 & 2.78 & 0.14 & -60.9 & 23.99 + r1/r2 & 5.27 & 7.03 & 0.76 & 76.4 & 23.98 + e1 & 32.86 & 18.62 & 0.51 & 74.6 & 24.83 + e2 & -31.22 & 19.15 & 0.68 & 64.4 & 24.53 + e3 & 0.54 & 22.21 & 0.66 & -69.8 & 24.44 + # 20 & 3.09 & 37.85 & 0.36 & -70@xmath23 & 16.9 + # 35 & 0.00 & 0.00 & 0.45 & -77@xmath24 & 17.1 + spectroscopic data were acquired at the 3.6 m telescope of la silla ( eso ) with efosc on october 1996 .
a long slit of 1.5 width was positioned along the two objects b2 and b3 for a total integration time of 3 hours with an average seeing of 1.9 . following the predictions of k93
, the [ oii ] emission line was expected to lie at @xmath25 .
thus the r300 grism was used , providing a useful spectral range 60009000 and a dispersion of 7.5 / pixel .
the data were reduced with standard procedures for flat fielding , wavelength and flux calibration in the iraf environment .
sky subtraction was performed on the 2d image , and the final spectra were extracted with an optimal extraction algorithm .
both objects b2 and b3 show a prominent emission line at @xmath26 and @xmath27 respectively on the two dimensional spectra , which we identify with [ oii ] 3727(figure [ fig - b2b3 ] and [ fig - specb2b3 ] ) .
another feature appears at @xmath28 and @xmath29 respectively , identified with [ oiii ] 5007 thus confirming the redshifts .
we find @xmath30 and @xmath31 , clearly demonstrating the similarity of the two objects within the error bars , and in good agreement with the @xmath32 prediction .
this definitely confirms the nature of these objects as a multiply imaged galaxy .
although , arclets a1 and a2 were observed during 4.5 hours , no continuum nor emission line could be detected .
this multiple image is one of the bluest arclets observed in this field .
accurate photometry is available in many filters extending from @xmath33 to @xmath34 with data coming from a u hst image with the f336w filter ( i d : 5709 , p.i .
j.m deharveng , @xcite , hereafter paper ii ) , @xmath35 , @xmath36 and @xmath37 images from cfht ( @xcite ) and unpublished cfht infrared images in @xmath38 and @xmath34 bands ( le borgne and picat , private communication ) .
b2 lies too close to galaxy # 20 and is contaminated by its envelope at long wavelengths , hence we only concentrate on b3 .
we compute the spectral energy distribution ( sed ) of b3 , corrected from its redshift , in order to study the stellar content of the galaxy .
we compare and fit the b3 sed with synthetic ones from the bruzual and charlot spectrophotometric evolutionary code ( gissel96 ) .
this code can provide seds for different metallicities ranging from z@xmath39/50 to 5 z@xmath39 , different spectroscopic galaxy type determined by the star formation history ( burst , elliptical , spiral and irregular ) , and different internal absorption by dust can be studied .
the extinction law used is the one proposed by calzetti ( 1997 ) which is similar to what is observed in our galaxy without the 2175 bump .
we explore the 3-dimensional parameter space ( spectroscopic type , metallicity and absorption , e(@xmath40 ) from 0 to 0.3 mag . ) to fit the b3 photometry by the model seds .
the main constraint comes from the high uv flux which requires a weak internal absorption ( @xmath41mag ) and a low metallicity ( @xmath42 ) ( figure [ fig - sed - b3 ] ) . despite this , the fit does not reproduce the uv part very well .
a possibility is that we are seeing the superposition of a very recent burst of star formation on an older stellar population .
a strong starburst alone would show a much fainter flux at red and infrared wavelengths , hence the observed ir emission must come from an already existing old population .
the b multiple image is similar to arclet a5 which is also very bright in the uv but much fainter in the ir ( @xmath43 for b3 and @xmath44 for a5 ) .
more precisely , assuming a redshift of @xmath1=1.3 ( mellier et al .
1991 ) , we have similarly fitted the a5 sed .
it shows evidences for a low absorption ( @xmath45 mag . ) and a very low metallicity ( @xmath46 ) ( figure [ fig - sed - a5 ] ) .
the fact that the best - fit metallicity ( @xmath47 ) is the lower bound allowed by the code should not be a problem as a satisfying fit is also obtained for @xmath48 with comparable significance .
the trend towards low @xmath49 is then stable and not too fast .
the corresponding age is 0.3 gyr , drawing the portrait of a very young object in an active phase of star formation .
an additional old population is not required for a5 which is the major difference with object b. these two objects b3 and a5 are typical examples of the population revealed by uv imaging .
the uv selection exhibits young objects with low metallicity and low absorption in the redshift range [ 0.5,2.0 ] .
such detailed stellar population study , could in principle be applied on any other arclets in the field , for which redshifts have been spectroscopically confirmed or are strongly constrained by lensing and photometric redshift techniques ( pell et al .
using both the giant arc a0 and the b2b3b4 triple system , k93 showed that a bimodal mass distribution was an adequate fit to these multiple image constraints ( see also abdelsalam et al .
1998 for a different approach ) . here , we improve this gross picture by taking into account the contribution of the cluster galaxies ( mainly e / s0 galaxies in the cluster core ) in a similar way as kneib et al .
( 1996 ) and natarajan et al .
all mass components ( galaxies+clusters ) have been assumed to follow a truncated pseudo isothermal mass distribution ( piemd , kassiola and kovner 1993 , hjorth & kneib 1998 ) .
following kneib et al .
( 1996 ) , for each galaxy halo , the velocity dispersion @xmath50 , the truncation radius @xmath51 and the core radius @xmath52 are scaled to the galaxy luminosity computed from the observed f675w magnitude .
the scaling relations used for the galaxy halos are : @xmath53 @xmath54 @xmath55 the scaling relations adopted are motivated by the properties of the fundamental plane ( fp ) and are similar to the one used by brainerd et al .
in particular , the exponent 0.8 in eq .
( 2 ) leads to a total mass - to - light ratio that scales with @xmath56 in agreement with the observed correlation of the fp ( _ e.g. _ jorgensen , franx & kjaergaard 1996 ) .
brighter galaxies have more extended and therefore more massive dark halos .
the orientation and ellipticity of the galaxy halos are taken from the observed values of the light distribution while @xmath57 and @xmath58 are optimized .
@xmath59 is fixed at 0.15 @xmath60 . to model the cluster components we considered two ` large scale ' mass distributions ( represented by 2 truncated piemd mass distribution ) centered on galaxies
# 20 and # 35 .
their orientations , ellipticities , velocity dispersions and core radius are left as free parameters .
the constraints imposed in the optimization procedure are the multiple images described in section 2.2 , along with the redshifts of a0 and of the b2b3b4 triplet .
the best fit parameters are summarized in table [ hst ] ( @xmath61=4.5 ) .
we also derive a `` lensing '' redshift estimate for the multiple images .
the c system is at @xmath62 , d is at @xmath63 and e is at @xmath64 .
r is moved to larger redshift with @xmath65 and a1/a2 could form a gravitational pair at @xmath66 .
we computed the total projected mass in different apertures centered at the barycentre of the # 20 and # 35 galaxies .
we find that within 75 , 150 and 300 kpc the total mass is , respectively , for this modeling and the previous one ( k93 ) : 0.5 ( 0.45 ) @xmath670.05 10@xmath68 m@xmath69 , 1.8 ( 1.6 ) @xmath670.1 10@xmath68 m@xmath69 and 4.8 ( 4.3 ) @xmath670.15 10@xmath68 m@xmath69 .
furthermore , 5% of the total mass is retained in cluster galaxy halos .
the total mass - to - light ratio is @xmath70180 ( m / l@xmath71)@xmath69 ( out to 300kpc , 160(m / l@xmath71)@xmath69 for the k93 modeling ) , and @xmath709 ( m / l@xmath71)@xmath69 for a @xmath72 galaxy halos .
these results are similar to the one found in other cluster lenses like abell 2218 , 2390 and ac114 ( kneib et al 1996 , 1998 ; natarajan et al 1998 ) .
.model parameters for a370 potential [ cols="^,^,^,^,^,^,^ " , ]
the number counts of gravitational arclets in the field of a massive cluster of galaxies are a competition between the magnification of the luminosity by the cluster potential that makes more objects visible and the surface dilution that decreases the surface density of arclets by the same factor as for the magnification .
the number of arcs brighter than magnitude @xmath73 with an axis ratio greater than @xmath74 and a surface brigthness brighter than @xmath75 within the field of a cluster of galaxies is : @xmath76 the sum is over the different morphological types @xmath77 .
@xmath78 is the lens redshift and @xmath79 is the redshift cutoff corresponding to the limit in central surface brightness @xmath75 .
@xmath80 is the angular area in the source plane ( at redshift @xmath1 ) that gives arcs with an axis ratio between @xmath81 and @xmath82 .
@xmath83 is the luminosity function at redshift @xmath1 for each morphological type .
it was required as a preliminary step that counts in wavebands from @xmath33 to @xmath84 and redshift distributions in @xmath35 and @xmath37 in empty field are correctly reproduced .
we used the model for galaxy evolution of bruzual and charlot ( 1993 ) with the prescriptions of pozzetti et al .
( 1996 ) for the optimisation of the parameters representing the field galaxy distributions ( see bzecourt et al . 1998a for details ) . the weak point in the bzecourt
et al . 1998a model , is the approximation of circular sources used for the calculation of arclets axis ratio .
indeed , this may underestimate the number of galaxies with axis ratio larger than a certain threshold especially at low redshifts and then bias the predicted redshift distribution . to provide a more accurate and reliable redshift distribution ,
we now take into account the galaxy ellipticity distribution as given by ebbels ( 1998 ) ( this distribution is based on the analysis of hst / mds - like images , and is therefore appropriate to our study ) : @xmath85 \label{ell}\ ] ] where @xmath86 for an elliptical object with semi major axis @xmath87 and semi minor axis @xmath88 .
the complex deformation @xmath89 is defined by : @xmath90 where @xmath91 is the orientation of the major axis . at each point in the image plane
( and at each redshift step for the sources ) , we estimate the fraction of the background galaxies which can give a lensed image more distorted than a fixed limit .
we first consider the effect of the lens deformation through the complex relation given by kneib et al .
( 1996 ) that relates the image deformation @xmath92 as a function of the source deformation @xmath93 and the strength of the potential @xmath94 : @xmath95 where @xmath96 ( see kneib et al 1996 for a full description of this formalism ) .
the term @xmath97 is a correction for strong lensing only .
a lower threshold in the observed axis ratio ( @xmath98 here ) corresponds to a lower threshold in the deformation ( @xmath99 ) whatever the position angle of the image . within this limit
, we then scan all the allowed solutions for @xmath100 and the source position angle @xmath101 .
the fraction @xmath102 of galaxies at the location @xmath103 in the image plane and at redshift @xmath1 which fill the condition is computed using the probability distribution of equation ( 4 ) for @xmath100 and an average over the position angle @xmath101 .
the magnitude histogram of the arclets observed in a370 is shown in figure [ histo_f675w ] .
as already stressed in bzecourt et al .
( 1998a ) , we find that the addition of galaxy scale components and the effect of the source ellipticity distribution significantly increase arclets counts by a factor of 1.4 ( @xmath105 ) and 1.7 ( @xmath106 ) with respect to a k93 model with the assumption of circular sources .
such increase in the arclets counts was already observed in bzecourt et al .
( 1998a ) when using a detailed mass distribution but only with circular sources .
considering an ellipticity distribution for the galaxies also increases the arclet counts and strongly modifies the redshift distribution as discussed below . with this refinement ,
the arclet counts prediction is now more consistent with the data over the whole magnitude range
. clearly , detailed lens models and proper assumptions on the source ellipticity distribution are mandatory to explore arclets counts .
the next step is to consider the redshift distribution of the arclets .
figure [ redshift_675 ] shows the predicted one with the selection criteria close to the observational limits for faint object spectroscopy ( @xmath107 , @xmath108 and @xmath98 ) .
the comparison with real data is not easy at present because no well defined sample of arclets , complete in magnitude , has yet been spectroscopically explored .
the redshift distributions displayed in figure [ redshift_675 ] present a prominent peak at @xmath109 and a secondary one at @xmath110 due essentially to the contribution of bright and young elliptical - type galaxies in their process of strong star formation .
this second peak is strongly attenuated compared to what was discussed in bzecourt et al .
( 1998a ) where there was clearly a strong excess in the high redshift tail .
taking into account the ellipticity distribution of the sources has fixed this problem . in order to see how stable the redshift distribution is with respect to the ellipticity distribution ,
another @xmath111 is considered following ebbels ( 1998 ) : @xmath112 \label{ell_2}\ ] ] representative of objects with @xmath113 while equation [ ell ] was derived with bright objects ( @xmath114 ) . as
this distribution corresponds to more elongated objects than in equation [ ell ] , more arclets appear at low @xmath1 ( figure [ redshift_675 ] , dotted line ) .
however , the total number of arclets is increased by only 11% .
analysis of figure [ redshift_675 ] suggests several comments : * compared with the redshift distribution of field galaxies within the same observational limits , the arclet distribution is biased towards more distant objects .
we can take advantage of this modification of the redshift distribution to select more distant galaxies in a redshift survey sample .
this is particularly true near the central part of the lens where very high redshift galaxies ( @xmath115 ) have already been found ( @xcite , @xcite , @xcite , @xcite , @xcite ) . * the two galaxy evolution models presented in this paper ( @xmath106 and @xmath105 ) have been validated in empty fields where the number counts and redshift distribution of galaxies were compatible with the observed ones .
however they present significant differences at high redshift for arclets .
the @xmath106 model with pure luminosity evolution " ( ple ) predicts more high redshift arclets than the @xmath105 model where number density evolution is included .
this different behavior of the redshift distribution at @xmath116 is encouraging as it could be a way to distinguish the two scenarios by analyzing the redshift distribution of arclets with @xmath116 .
although it is presently difficult to speculate on the true fraction of very high redshift objects in well defined samples of arclets , this population of very high redshift arclets does not seem to dominate the actual samples .
hence we favor the @xmath105 model corresponding to a merging scenario of galaxy formation and evolution . *
because the high redshift domain is sensitive to many uncertainties included in the evolutionary code ( dust obscuration in the uv , uncertainties in the slope of the imf for massive stars or in their uv tracks , influence of the short time scale phenomena ) , we must be cautious on the above conclusions .
furthermore , the galaxy evolution models assume that all the galaxies form instantaneously at a given redshift , which is a clear limitation of these models ( we have @xmath117 for @xmath106 , which gives for the present day galaxies an age of 16 gyr , and @xmath118 for @xmath105 , corresponding to an age of 12.2 gyr ) . a more phenomenological model of galaxy formation that follows in detail
the mass evolution of galaxy sub units ( @xcite ) may improve this simple description , and would likely be the best to be compared with the current data . for a simple description of the background population , it is easy to show that the number density of objects brighter than a given magnitude @xmath73 behind a lens is : @xmath119 where @xmath120 is the number density in blank field , @xmath121 is the magnification and @xmath122 is the logarithmic slope of field number counts . an excess or a lack of objects
is then expected for a slope steeper or shallower than 0.4 respectively . in practice ,
optical number counts of field galaxies show a slope smaller than 0.4 in nearly all wavebands ( except in u and b at relatively bright magnitudes ) , so a``depletion '' is expected in most cases , being more pronounced at longer wavelengths ( fort et al .
1996 , taylor et al .
1998 ) . with our model
, it is possible to compute radial depletion curves instead of global number counts , for any filter or magnitude range .
a clear illustration of the wavelength dependence of the magnification bias is given by the ratio of the number of lensed objects expected in a given area in the field of abell 370 over the number of field galaxies in the same area ( figure [ depletion ] ) .
one can note that the predicted intensity of the depletion is higher at longer wavelengths , as expected , because of the shallower slope of the field galaxies counts .
moreover , because of the flattening of the counts at faint magnitudes , the depletion is also very sensitive to the magnitude threshold ( see the i curve in figure [ depletion ] ) .
uv counts have a slope larger than 0.4 at bright magnitudes ( @xmath123 ) but it quickly decreases at fainter levels .
this fast flattening is due to the lyman break that goes through the red limit of the u filter at redshift @xmath124 .
hence , the lack of objects at faint magnitudes in @xmath33 produces the depletion curve shown in figure [ depletion ] ( see paper ii for more details of the uv modeling ) .
the detection of this magnification bias is highly dependant on the poissonian noise of the background sources and the contamination by cluster members and foreground objects .
this is critical in cluster cores where the density of objects is very low , of the order of a few units per arcmin@xmath125 .
hence , these very poor statistics can not bring valuable information on the sources redshift distribution as most of the information is expected in the dip of the depletion curve .
only massive clusters are able to show such a depletion curve but the recovery of the sources redshift distribution seems out of reach with present datasets .
in this paper , we have presented a new analysis of the cluster lens abell 370 .
thanks to the measurement of the redshift for the b2b3 gravitational pair ( @xmath126 ) and the identification of several new multiple arclets in the wfpc2 image , a more accurate and well constrained model of the mass distribution in abell 370 is proposed , including galaxy scale components .
we studied the spectral energy distribution ( sed ) of arclets b3 and a5 which are found to be both young , low metallicity star forming objects without strong interstellar extinction .
the lens model has been used to study the background population of galaxies .
taking the galaxy ellipticity distribution into account induces significant changes in the predicted redshift distribution of arclets .
the excess of very distant sources found in previous analyses is now strongly attenuated and arclets at low redshift are recovered .
for the two prescriptions used on galaxy evolution , although they both reproduce well the number counts in empty fields and in cluster lenses , the `` pure luminosity evolution '' model in a low density universe over predicts the number of high redshift ( @xmath116 ) galaxies , compared to what is currently observed .
if this effect is observationally confirmed it may constrain the number density evolution of galaxies in order to interpret the redshift distribution of arclets . to understand further the properties of the background population through cluster lenses ,
deep dedicated cluster surveys are needed in order to enlarge the number of arclets and have significant statistical properties .
a multi - color approach is favored as it can help in constraining the redshift distribution .
this is particularly true for those unresolved sources for which no shape information is available and only the spectral energy distribution can discriminate between background objects and cluster members .
in addition , detailed study of their spectral energy distribution will be useful to characterize properties like the luminosity function , dust extinction and star formation histories of distant galaxies ( e.g. pell et al .
1998b ) .
we thank r. pell , r.s .
ellis and y. mellier for many fruitful discussions and encouragements .
this research has been conducted under the auspices of a european tmr network programme made possible via generous financial support from the european commission ( http://www.ast.cam.ac.uk/ioa/lensnet/ ) .
this work was also supported by the programme national de cosmologie and cnrs .
soucail g. , kneib j.p .
, picat j.p . , ellis r.s . , 1998a , in `` the young universe galaxy formation and evolution at intermediate and high redshift '' , monteporzio , september 1997 .
conference series , eds s. dodorico , a. fontana , e. giallongo . | we present new observations of the cluster - lens abell 370 : a deep hst / wfpc2 f675w image and eso 3.6 m spectroscopy of faint galaxies .
these observations shed new light on the statistical properties of faint lensed galaxies .
in particular , we spectroscopically confirm the multiple image nature of the b2b3 gravitational pair ( @xcite ) , and determine a redshift of @xmath0 which is in very good agreement with earlier predictions . a refined mass model of the cluster core ( that includes cluster galaxy halos )
is presented , based on a number of newly identified multiple images .
following bzecourt et al .
( 1998a ) , we combine the new cluster mass model with a spectrophotometric prescription for galaxy evolution to predict the arclets number counts and redshift distribution in the hst image .
in particular , the ellipticity distribution of background sources is taken into account , in order to properly estimate the statistical number and redshift distribution of arclets .
we show that the redshift distribution of arclets , and particularly its high redshift tail can be used as a strong constraint to disentangle different galaxy evolution scenarios . a hierarchical model which includes number density evolution is favored by our analysis .
finally , we compute the depletion curves for the faint galaxy number counts and discuss their wavelength dependence . |
sex , which involves the alternation of meiosis and gamete fusion , is a rather inefficient means of self - propagation as compared to asexual reproduction , where offspring stem only from a mitotically produced cells .
one of the most common reasons used to explain the origin and maintenance of sex is its ability to reduce the mutation load if consecutive mutations lead to an increasing decline in relative fitness , although it is not clear _ a priori _ that the heritable variance in fitness is significantly increased by sex . despite decades of developing theoretical models to explain why sex is such a widespread phenomenon and how sexual reproduction may confer advantages that outweigh its disadvantages , until now no such general clear advantage has been found .
investigations of evolutionary problems by physicists have in fact boomed in the last few years . since computer simulations of natural systems can provide much insight into their fundamental mechanisms , they can be used to test theoretical ideas that could be otherwise viewed as too vague to deserve the status of scientific knowledge @xcite . in this way , many computer models in population dynamics have been proposed to investigate the evolution of sex and its justification , as well as the comparison between sexual and asexual reproduction , for instance , the redfield model @xcite , the penna bit - string model @xcite , a genomic bit - string model without aging @xcite and stauffer model @xcite . of particular interest here is the heumann - htzel model @xcite , which originally simulated the evolution of asexual population , composed of haploid individuals , without recombination .
thus now we introduce the recombination in this model , in order to find out if the sexual reproduction ( with males and females ) can produce better results than the simple asexual reproduction @xcite . in the next section ,
we describe the standard and the modified heumann - htzel model , in section 3 , we present our results and in section 4 , our conclusions .
since it has been proposed by michael heumann and michael htzel in @xmath0 , the heumann - htzel model @xcite , which was an unsuccessful attempt to introduce more ages in the dasgupta model @xcite , has remained forgotten due to the fact that after many generations it reduces to the two - age model of partridge - barton @xcite .
the dasgupta model consists in taking into account some modifications such as hereditary mutations and food restrictions in the partridge - barton model .
in fact , the heumann - htzel paper @xcite , treats basically the computer simulations using the generalized dasgupta aging model proposed by michael htzel in his work , under the supervision of dietrich stauffer , in order to obtain the license to teach in german secondary school @xcite .
michael heumann , who was another teacher s candidate , worked only on the inclusion of the `` dauer '' state in the dasgupta model @xcite .
recently , the heumann - htzel model was reinvestigated and , according to the authors , with `` simple and minor change in the original model '' this incapacity to describe populations with many ages seems to be surmounted @xcite . in the original version of the heumann - htzel , the genome of each ( haploid ) individual
is represented by one set of probabilities @xmath1 , where @xmath2 is the survival probability that an individual has to reach age @xmath3 from age @xmath4 . at every time step @xmath5 , @xmath6
individuals are chosen randomly to have their survival probability @xmath2 altered by mutations to @xmath7 , where the age @xmath4 is also randomly chosen .
@xmath8 is the size of the population at time @xmath5 and @xmath9 is the maximum age one individual can live , which is set up in the beginning of the simulation . the quantity @xmath10 is chosen randomly as any number between @xmath11 and @xmath12 and when it is negative ( positive ) it corresponds to a deleterious ( beneficial ) mutation .
the effect of food and space restrictions is taken account by an age - independent verhulst factor , which gives to each individual a probability @xmath13 $ ] of staying alive ; @xmath14 represents the maximum possible size of the population .
this mean - field probability of death for the computer simulations has the benefit of limiting the size of population to be dealt with .
the passage of time is represented by the reading of a new locus in the genome of each individual in the population , and the increase of its age by @xmath15 . after taking account the natural selection and the action of verhulst dagger , at the completion of each period of life , each individual gives birth to one baby ( age=0 ) which inherits its set of probabilities ( @xmath16 ) . in the recent reinvestigation of this model@xcite ,
individuals with age @xmath4 in the interval @xmath17 will generate @xmath18 offspring and the mutations are allowed only on a fraction @xmath19 ( @xmath20 ) of the babies .
= 12,5 cm in the sexual version , each ( diploid ) individual of the population , which consists of males and females , is genetically represented now by two sets of survival probabilities , @xmath21 and @xmath22 , to be read in parallel . in this way
, we have studied the following cases ( see below ) : @xmath23 @xmath24 @xmath25 * case ( a ) * - the effective survival probability in some age will be the arithmetic average of the values present in both sets at that age : @xmath26 @xmath25 * case ( b ) * - the effective survival probability in some age will be the maximum value between the values present in both sets at that age : @xmath27 , \max\left[p_1^{1},p_1^{2}\right ] , ... , \max\left[p_{{\rm maxage}-1}^{1},p_{{\rm maxage}-1}^{2}\right])\ ] ] = 12,5 cm = 12,5 cm if the female succeeds in surviving until the minimum reproduction age @xmath28 , it chooses , at random , an able male to mate ( @xmath29 ) and it generates , with probability @xmath30 , @xmath18 offspring every iteration until the maximum age of reproduction @xmath31 .
the offspring inherits its set of survival probabilities from its parents in the following way : the two sets of survival probabilities of the male , for instance , are broken in the same random position , and the complementary pieces , belonging to different strings , are joined to form two male gametes .
one of the gametes is then randomly chosen to be passed to the offspring .
after that , @xmath32 random mutations are introduced into this gamete , and the final result corresponds to one string of the baby genome .
the same process occurs with the female genome , generating the second string of the baby , with @xmath33 mutations . at the end
the offspring genome contains a total of @xmath34 mutations . finally , the sex of the baby is randomly chosen , each one with probability @xmath35 .
this procedure is repeated for each of the @xmath18 offspring .
the simulation starts with @xmath36 individuals ( half for each sex ) and runs for @xmath37 time steps , at the end of which ( the last @xmath38 steps , when the population was stabilized ) averages are taken over the population .
the parameters of the simulations are : @xmath25 initial population @xmath39 ; @xmath25 maximum population size @xmath40 ; @xmath25 probability to give birth @xmath41 ; @xmath25 birth rate @xmath42 ; @xmath25 mutation rate @xmath43 per gamete ; @xmath25 @xmath44 and @xmath45 ( the same values used in the original htzel model ) .
our figures with @xmath39 are confirmed by larger simulations with @xmath46 , and also by larger simulations with @xmath47 time steps .
= 12,5 cm from figure 1 and its inset we can see that the diploid sexual population is not only larger than the haploid asexual one , but also presents a higher survival probability .
= 12,5 cm = 12,5 cm in figure 2 ( case ( a ) ) and figure 3 ( case ( b ) ) , we present the survival probability as a function of age for different period of reproduction ( @xmath48 ) . _ aging starts with reproduction _ :
the survival rate decays as soon as reproduction age is reached .
there are no individuals alive older than the maximum reproduction age @xmath49 .
figure 4 corresponds to the case in which all the individuals of the population reproduces only once - the so - called catastrophic senescence effect @xcite . in this way
, two rules of reproducing were adopted : 1 ) the reproduction age is the same for all individuals ( @xmath50 ) , 2 ) the reproduction age is randomly chosen between @xmath51 and @xmath52 .
we can noticed that this effect is more pronounced for the former @xcite , as well that the responsible for that are both breeding once and breeding for all individuals at the same age .
the explanation for these effects observed in figures 2 - 4 is based on the darwin theory : individuals must stay alive in order to reproduce and perpetuate the species .
if they can no longer generate offspring but remain in the population , they are killed by the accumulation of deleterious mutations @xcite .
in fact , real mutations can be divided into the common recessive ( almost @xmath53 of the real mutations are recessive ) and the rare dominant mutations . in this way , if among the many genes of a species , one of the father s genes differs from the corresponding one of the mother , then it adversely affects the child only if the mutation is dominant .
recessive mutations affect the child only if both the father and mother have them . in order to take into account this aspect in the sexual version of htzel , at the beginning of the simulation we choose randomly @xmath54 dominant positions and keep them fixed during the whole simulation . the effective survival probability in the dominant positions ( ages ) will be the smallest value of the two located in the same position in both strands , and for recessives ones the effective survival probability will be the arithmetic average of them . in figure 5 , we can see that the inclusion of dominance does not alter the lifespan of the population , although it has been observed that population evolved without dominance is larger than the other without dominance due the deaths in the former being bigger than the latter , since in these dominant positions the effective survival probability is the minimum value between the values present in both sets at that age . in the particular simulation shown , for @xmath55
it is noticed a decrease in the survival probability when the dominance is considered , since the ages @xmath56 , @xmath57 , @xmath58 , @xmath59 were dominant positions .
figure 6 shows the time evolution of the total population of each age for sexual reproduction when the mutations are exclusively harmful . from this figure
we can notice that a stable population for ages @xmath60 is obtained , in contrast to the original model in which there are no individuals alive older than age @xmath61 , even if beneficial mutation and also a deleterious mutation rate @xmath62 times smaller have been assumed .
the result obtained here ( fig .
1 ) , introducing sex in the original model , was found with the asexual htzel model @xcite only when mutations were allowed on a very small fraction @xmath63 of the babies and also a minimum age of reproduction @xmath64 was considered . in our simulations , @xmath65 , @xmath66 and @xmath67 .
we have shown that main problem related to the htzel model , which was its incapacity to treat populations with many age intervals , has been overcome by introducing recombination ( with males and females ) in this model , without any other assumptions . as well as , with the inclusion of sex in the model , the population meltdown observed in the asexual version , when only deleterious mutations are considered , has been avoided . moreover , in agreement with some earlier models , we have also obtained that the sexual reproduction ( with males and females ) produces better results than the asexual one .
m. eigen ( 1971 ) naturwissenchaften * 58 * , 465 ; d. charlesworth , m.t .
morgan and b. charlesworth ( 1992 ) genet .
camb . * 59 * , 49 ; l.d .
mueller and m.r . rose ( 1996 ) proc .
usa * 93 * , 15249 ; b. charlesworth ( 2001 ) j. theor .
biology * 210 * , 47 .
redfield ( 1994 ) nature * 369 * , 145 ; s. siller , nature * 411 * , 689 ( 2001 ) ; r.s .
howard and c.m .
lively , nature ( london ) * 367 * , 554 ( 1994 ) ; c. zeyl , t. vanderford and m. carter , science * 299 * , 555 ( 2003 ) . | why sex evolved and it prevails in nature remains one of the great puzzles of evolution .
most biologists would explain that it promotes genetic variability , however this explanation suffers from several difficulties .
what advantages might sex confer ?
the present communication aims at certain investigations related to this question , in this way we introduce sexual recombination on the htzel model ( with males and females ) and we compare these results with those from asexual reproduction without recombination . * sex and recombination in the htzel aging model * a.o .
sousa _ institute for theoretical physics , cologne university , d-50937 k " oln , germany _ * keywords*:population dynamics ; aging ; monte carlo simulations ; evolution ; recombination |
hbt ( hanbury - brown - twiss ) interferometry is an indispensable tool to study the space - time structure of the particle - emitting source produced in high energy heavy ion collisions@xcite .
the experimental pion hbt measurements at rhic give the ratio of @xmath4@xcite , which is much smaller than many earlier theoretical expectations .
such a discrepancy between theory and experiment is referred to as the rhic hbt puzzle@xcite . on the other hand ,
hydrodynamical calculations give reasonably good descriptions of the elliptic flow , which has been considered as an evidence for a strongly - coupled quark - gluon plasma@xcite .
the resolution of the hbt puzzle is important in finding out why traditional hydrodynamics succeed in explaining the elliptic flow but fails in explaining the hbt radii .
traditional studies of the hydrodynamics of the evolving fluid assume a single contiguous blob of matter under expansion , with a relatively smooth initial and final density distributions .
initial transverse density fluctuations and hydrodynamical instabilities have been neglected but their inclusion may lead to `` multi - fragmentation '' in the form of large scale final - state density fluctuations and the formation of granular droplets .
it is useful to explore the consequences of the occurrence of granular droplets .
previously we propose a granular model to explain the hbt puzzle@xcite .
we would like to review here the important ingredients which enters into the resolution of the puzzle .
further suggestions of using single - event hbt interferometry to search for signatures of the granular source can be found in refs ..
based on the recent results of high - energy heavy - ion collisions at rhic , the early matter produced in the collisions may be a strongly - coupled qgp ( sqgp ) , which has a very high energy density and reaches local thermalization within about 1 fm / c@xcite .
the expansion of the matter after that time may be unstable .
many effects , such as the large fluctuations of the initial transverse energy density@xcite , the sausage instability@xcite , and possible phase transition@xcite , may lead to the fragmentation of the system and the formation of many spherical droplets due to the surface tension of the qgp @xcite . to describe the evolution of a droplet , we use relativistic hydrodynamics where the energy momentum tensor of a thermalized fluid element in the center - of - mass frame of the droplet is@xcite @xmath5 u^{\mu}(x ' ) u^{\nu}(x ' ) - p(x ' ) g^{\mu \nu } \ , , \ ]
] @xmath6 is the space - time coordinate of the fluid element in the center - of - mass frame , @xmath7 , @xmath8 , and @xmath9 are the energy density , pressure , and 4-velocity of the element , and @xmath10 is the metric tensor . with the local conservation of energy and momentum ,
one can obtain the equations for spherical geometry as@xcite @xmath11 = - f \ , , \end{aligned}\ ] ] @xmath12 where @xmath13 , @xmath14 , @xmath15 , @xmath16
. in the equations of motion ( [ eqe ] ) and ( [ eqm ] ) there are three unknown functions @xmath7 , @xmath8 , @xmath17 . in order to obtain the solution of the equations of motion , we need an equation of state which gives a relation @xmath18 between @xmath8 and @xmath7 [ @xcite ] . at rhic energy ,
the system undergoes a transition from the qgp phase to hadronic phase . as the net baryon density in the central rapidity region is much smaller than the energy density of the produced matter ( here presumed to be qgp ) ,
the baryon density of the system in the center rapidity region can be neglected .
lattice gauge results suggest the entropy density of the system as a function of temperature as@xcite @xmath19\,,\ ] ] where @xmath20 is the entropy density at the transition temperature @xmath21 , @xmath22 and @xmath23 are the degrees of freedom in the qgp phase and the hadronic phase , and @xmath24 is the width of the transition .
the thermodynamical relations among @xmath8 , @xmath7 , and @xmath25 in this case are @xmath26 from these thermodynamical relations and eq .
( [ eos ] ) , we can obtain the equation of state @xmath18 .
= 0 = 2.85 in fig
. 1 . ( a ) temperature profile and ( b ) isotherms for the droplet . here , @xmath27 and @xmath28 .
= -16 = -2.3 in using the hlle scheme@xcite and sod s operator splitting method@xcite , one can obtain the solution of eqs .
( [ eqe ] ) and ( [ eqm])@xcite , after knowing the equation of state and initial conditions .
we assume that the droplet has a uniform initial energy density @xmath29 within a sphere with radius @xmath30 , and has a zero initial velocity in its center - of - mass frame .
1(a ) and ( b ) show the temperature profiles and isotherms for the droplet .
in our calculations , we take the parameters of the equation of state as @xmath31 , @xmath32 , @xmath33 mev , and @xmath34 , and take the initial energy density @xmath35 , which is about two times of the density of quark matter at @xmath21 [ @xcite ] .
= -3 = -2.3 in if we assume that the final pions are emitted from the droplet at the freeze - out configuration characterized by a freeze - out temperature @xmath0 , we can see from figure 1(b ) that the the average particle emission time scales with the initial radius of the droplet @xmath30 . in hbt interferometry
, the radius @xmath2 is related to the spatial size of the particle - emitting source and the radius @xmath1 is related not only to the source spatial size but also to the lifetime of the source@xcite .
a long lifetime of the source will lead to a large @xmath1@xcite . from the hydrodynamical solution in figure 1(b ) , both the average freeze - out time and freeze - out radial distance increase with @xmath30 for a single droplet source . as a consequence
, @xmath36 is insensitive@xcite to the values @xmath30 .
the value of @xmath37 for the single droplet source@xcite is about 3 [ @xcite ] , much larger than the observed values@xcite .
the rhic hbt puzzle of @xmath38 suggests that the pion emitting time may be very short . in order to explain the hbt puzzle
, we consider a granular source of @xmath39 qgp droplets distributed in a distribution @xmath40 as illustrated in fig .
we assume that the evolution of the granular source is simply the superposition of the evolutions of individual droplets with the same initial conditions .
as the average freeze - out time is proportional to the initial radius of the droplet , the freeze - out time and @xmath41 decreases if the initial radius of the droplet decreases . on the other hand , @xmath42 increases if the width of the droplet spatial distribution @xmath40 increases .
a variation of the droplet size and the width of droplet spatial distribution can result in @xmath1 nearly equal to @xmath42 [ @xcite ] .
furthermore , if the granular source has a collective expansion the hbt will measure the size of the region of the ellipse in figure 2(b)@xcite . in this case
the value of @xmath43 decreases even further and may be smaller than unity@xcite .
we assume that the droplets are initially distributed in a short cylinder of length @xmath44 along the beam direction ( @xmath45 direction ) with an initial transverse spatial distribution up to a radius @xmath46 , @xmath47 \theta({\cal r}_t -\rho ) \,\theta({\cal r}_z -|z| ) \,,\end{aligned}\ ] ] where @xmath48 and @xmath45 are the coordinates of the center of a droplet , and @xmath49 describes a shell - type radial distribution . because of the early thermalization and the anisotropic pressure gradient , the droplets will acquire anisotropic initial velocities .
we therefore consider a granular source of the the qgp droplets with an anisotropic velocity distribution and the velocity of a droplet depends on the initial coordinates of the droplet center , @xmath50 , in the form @xmath51 where @xmath52 describes the magnitude of the anisotropic expansion , @xmath53 denotes the sign of @xmath54 , and @xmath55 are the exponential power parameters that describe the variation of the velocities with @xmath54 . in our calculations
the velocity parameters are taken as @xmath56 , @xmath57 , @xmath58 , @xmath59 , and @xmath60 , by comparing with the experimental data of pion transverse momentum spectral@xcite and elliptic flow@xcite in @xmath61 gev au + au collisions at rhic .
we take the freeze - out temperature as @xmath62 in our calculations .
the two - particle bose - einstein correlation function is defined as the ratio of the two - particle momentum distribution @xmath63 relative to the the product of the single - particle momentum distribution @xmath64 . for a chaotic pion - emitting source , @xmath65 , and @xmath63
can be expressed as@xcite @xmath66 where @xmath67 is the magnitude of the amplitude for emitting a pion with 4-momentum @xmath68 in the laboratory frame at @xmath69 and is given by the bose - einstein distribution with freeze - out temperature @xmath0 in the local rest frame of the source point .
@xmath70 is the two - pion wave function . neglecting the absorption of the emitted pions by other droplets ,
@xmath71 is simply @xmath72\ , .\end{aligned}\ ] ] using the components of out " , side " , and long"@xcite of the relative momentum of the two pions , @xmath73 , as variables , we can construct the correlation function @xmath74 from @xmath63 and @xmath75 by picking pion pairs from the granular source and summing over @xmath76 and @xmath77 for each @xmath78 bin@xcite .
the hbt radii @xmath41 , @xmath42 , and @xmath79 can then be extracted by fitting the calculated correlation function @xmath80 with the following parametrized correlation function @xmath81 figure 3 shows the theoretical two - pion correlation functions for the granular source .
the top figures give the results for the average pion transverse momentum of a pion pair , @xmath82 , less than 400 mev / c , and the bottom figures for @xmath83 mev / c . in our calculations
the size parameters of the granular sources are taken to be @xmath84 fm , @xmath85 fm , @xmath86 fm , and @xmath87 fm .
the number of droplet @xmath39 is taken to be 40 in our calculations .
the left panels in fig . 4
give the extracted two - pion hbt radii for the granular source as a function of @xmath82 .
the symbols of circle and down - triangle are for @xmath88 fm and @xmath89 fm , respectively .
the experimental phenix results@xcite , and star results@xcite are shown on the right panels .
the curve gives the theoretical results for @xmath88 fm . for our theoretical hbt calculations
, we use a cut for particle pseudo - rapidity region @xmath90 , the same as in the phenix experiments@xcite .
we find that if we increase the parameter @xmath30 , the hbt radii @xmath1 and @xmath91 will increase . and if we increase the parameter @xmath92 , the variation of hbt radii @xmath1 and @xmath42 with @xmath82 will become steep . as can be inferred from fig . 4 , the hbt results of granular source for @xmath85 fm agree quite well with experimental data , although the case with @xmath93 also give almost as good an agreement .
the expansion of the dense matter ( sqgp ) produced in high - energy heavy - ion collisions at rhic may be unstable
. the initial transverse density may also be highly fluctuation . the unstable expansion and fluctuating initial transverse density may lead to a fragmentation of the system and the formation of a granular source of qgp droplets .
although a granular structure was suggested earlier as the signature of first - order phase transitions@xcite , the occurrence of granular structure may not be limited to the occurrence of first - order phase transitions .
there are additional effects which may lead to the dynamical formation of granular droplets@xcite . for a granular source of the droplets ,
the average particle emission time scales with the initial radius of the droplet .
pions will be emitted earlier if the radius is smaller .
an earlier emission time will lead to a smaller extracted hbt radius @xmath1 , while the extracted hbt radius @xmath42 is determined by the scale of the distribution of the droplets .
a collective expansion of the droplets can further decrease @xmath1 and the value of @xmath94 can be close to , or even less than 1 for an expanding granular source of qgp droplets . in an event - mixing experimental analysis ,
some signals of the granular source such as the correlation function fluctuations may likely be suppressed after averaging over many events@xcite .
however , the property of an earlier average emission time of the granular source will remain as it is common to all events and not averaged out by event mixing .
this property leads to a smaller @xmath43 in mixed - event hbt analysis . in conclusion
, a granular source model can explain the rhic hbt puzzle and reproduce the data of pion hbt radii , as well as the data of pion transverse momentum spectra and elliptic flow@xcite in @xmath61 gev au + au collisions at rhic .
it is of great interest to find direct evidences of the granular structure and to study the mechanics of granular source formation@xcite .
this research was supported by the national natural science foundation of china under contracts no . 10575024 , and by the division of nuclear physics , us doe , under contract no .
de - ac05 - 00or22725 managed by ut - battle , lc . | we present a review on the explanation of the rhic hbt puzzle by a granular pion - emitting source of quark - gluon plasma droplets .
the evolution of the droplet is described by relativistic hydrodynamics with an equation of state suggested by lattice gauge results .
the granular source evolution is obtained by superposing all of the evolutions of individual droplets .
pions are assumed to be emitted thermally from the droplets at the freeze - out configuration characterized by a freeze - out temperature @xmath0 .
we find that the average particle emission time scales with the initial radius of the droplet .
pions will be emitted earlier if the droplet radius is smaller .
an earlier emission time will lead to a smaller extracted hbt radius @xmath1 , while the extracted hbt radius @xmath2 is determined by the scale of the distribution of the droplet centers .
however , a collective expansion of the droplets can further decrease @xmath1 . as a result ,
the value of @xmath3 can be close to , or even less than 1 for the granular source of qgp droplets . |
we investigate different effects contributing to the determination of the optimal angle of release at shot put .
standard text - book wisdom tells us that the optimal angle is @xmath0 , while measurements of world - class athletes @xcite typically give values of below @xmath3 . in table
[ tab1 ] we show the data from the olympic games in 1972 given by kuhlow ( 1975 ) @xcite with an average angle of release of about @xmath4 .
the measurements of dessureault ( 1978 ) @xcite , mccoy et al .
( 1984 ) @xcite , susanaka and stepanek ( 1988 ) @xcite , bartonietz and borgstrm ( 1995 ) @xcite , tsirakos et al .
( 1995 ) @xcite and luhtanen et al .
( 1997 ) @xcite give an average angle of release of about @xmath5 . +
this obvious deviation triggered already considerable interest in the literature @xcite .
most of these investigations obtained values below @xmath0 but still considerably above the measured values .
e.g. in the classical work of lichtenberg and wills ( 1976 ) @xcite optimal release angles of about @xmath2 were found by including the effect of the height of an athlete . + we start by redoing the analysis of lichtenberg and wills ( 1976 ) @xcite .
next we investigate the effect of air resistance . here
we find as expected @xcite that in the case of shot put air resistance gives a negligible contribution instead of @xmath6 ( @xmath7 diameter , @xmath8 radius ) and have therefore a force that is four times as large as the correct one . ] .
if the initial velocity @xmath9 , the release height @xmath10 and the release angle @xmath11 are known , the results obtained up to that point are exact .
we provide a computer program to determine graphically the trajectory of the shot for a given set of @xmath9 , @xmath10 and @xmath11 including air resistance and wind .
+ coming back to the question of the optimal angle of release we give up the assumption of lichtenberg and wills ( 1976 ) @xcite , that the initial velocity , the release height and the release angle are uncorrelated .
this was suggested earlier in the literature @xcite .
we include three correlations : * the angle dependence of the release height ; this was discussed in detail by de luca ( 2005 ) @xcite .
* the angle dependence of the force of the athlete ; this was suggested for javeline throw by red and zogaib ( 1977 ) @xcite . in particular a inverse proportionality between the initial velocity and the angle of release was found .
this effect was discussed for the case of shot put in mcwatt ( 1982)@xcite , mccoy ( 1984)@xcite , gregor ( 1990)@xcite and linthorne ( 2001)@xcite . *
the angle dependence of the initial velocity due to the effect of gravity during the period of release ; this was discussed e.g. in tricker and tricker ( 1967 ) @xcite , zatsiorski and matveev ( 1969 ) @xcite , hay ( 1973 ) @xcite and linthorne ( 2001)@xcite . to include these three correlations
we still need information about the angle dependence of the force of the athlete . in principle
this has to be obtained by measurements with each invididual athlete . to show the validity of our approach we use a simple model for the angle dependence of the force and obtain realistic values for the optimal angle of release . + our strategy is in parts similar to the nice and extensive work of linthorne ( 2001 ) @xcite .
while linthorne s approach is based on experimental data on @xmath12 and @xmath13 our approach is more theoretical .
we present some toy models that predict the relation @xmath14 found by red and zogaib ( 1977 ) @xcite .
+ we do not discuss possible deviations between the flight distance of the shot and the official distance . here
were refer the interested reader to the work of linthorne ( 2001 ) @xcite .
let us start with the simplest model for shot put .
the shot is released from a horizontal plane with an initial velocity @xmath9 under the angle @xmath11 relative to the plane .
we denote the horizontal distance with @xmath15 and the vertical distance with @xmath16 . the maximal height of the shot is denoted by @xmath17 ; the shot lands again on the plane after travelling the horizontal distance @xmath18 , see fig.[fig1 ] . solving the equations of motions @xmath19 with the initial condition @xmath20 one obtains @xmath21 the maximal horizontal distance is obtained by setting @xmath22 equal to zero @xmath23 from this result
we can read off that the optimal angle is @xmath24 - this is the result that is obtained in many undergraduate textbooks .
it is however considerably above the measured values of top athletes .
moreover , eq.([xm0 ] ) shows that the maximal range at shot put depends quadratically on the initial velocity of the shot .
next we take the height of the athlete into account , this was described first in lichtenberg and wills ( 1976 ) @xcite .
( [ eom1 ] ) still holds for that case .
we denote the height at which the shot is released with @xmath10 .
the maximal horizontal distance is now obtained by setting @xmath22 equal to @xmath25 .
@xmath26 this equation holds exactly if the parameters @xmath27 , @xmath10 and @xmath11 are known and if the air resistance is neglected . + assuming that the parameters @xmath27 , @xmath10 and @xmath11 are independent of each other we can determine the optimal angle of release by setting the derivative of @xmath18 with respect to @xmath11 to zero . @xmath28 the optimal angle is now always smaller than @xmath0 . with increasing @xmath10 the optimal angle is getting smaller , therefore taller athletes have to release the shot more flat .
the dependence on the initial velocity @xmath27 is more complicated .
larger values of @xmath27 favor larger values of @xmath11 .
we show the optimal angle for three fixed values of @xmath29 m , @xmath30 m and @xmath31 m in dependence of @xmath27 in fig.[fig2 ] . with the average values from table [ tab1 ] for @xmath32 m and @xmath33
m / s we obtain an optimal angle of @xmath34 , while the average of the measured angles from table [ tab1 ] is about @xmath4 .
we conclude therefore that the effect of including the height of the athlete goes in the right direction , but the final values for the optimal angle are still larger than the measured ones . in our example
the initial discrepancy between theory and measurement of @xmath35 is reduced to @xmath36 .
these findings coincide with the ones from lichtenberg and wills ( 1976 ) @xcite .
+ for @xmath37 m and @xmath38 m / s ( @xmath39 ) we can also expand the expression for the optimal angle @xmath40 = \frac{1}{\sqrt{2 } } \left [ 1 - \frac{1}{4 } \frac{e_{pot}}{e_{kin } } + { \cal o } \left ( \frac{hg}{v_0 ^ 2 } \right)^2 \right ] \ , , \ ] ] with the kinetic energy @xmath41 and the potential energy @xmath42 .
@xmath43 denotes the mass of the shot . + by eliminating different variables from the problem , lichtenberg and wills ( 1976 ) @xcite derived several expressions for the maximum range at shot put :
@xmath44 expanding the expression in eq.([xmthel ] ) in @xmath45 one gets @xmath46 here we can make several interesting conclusions * to zeroth order in @xmath45 the maximal horizontal distance is simply given by @xmath47 .
this can also be read off from eq .
( [ xm0 ] ) with @xmath48 . *
to first order in @xmath45 the maximal horizontal distance is @xmath49 . releasing the shot from 10 cm more height results in a 10 cm longer horizontal distance . *
the kinetic energy is two times more important than the potential energy .
if an athlete has the additional energy @xmath50 at his disposal , it would be advantageous to put the complete amount @xmath50 in kinetic energy compared to potential energy .
* effects of small deviations from the optimal angle are not large , since @xmath18 is stationary at the optimal angle .
next we investigate the effect of the air resistance .
this was considered in tutevich ( 1969 ) @xcite , lichtenberg and wills ( 1976 ) @xcite .
the effect of the air resistance is described by the following force @xmath51 with the density of air @xmath52 , the drag coefficient of the sphere @xmath53 ( about @xmath54 ) , the radius of the sphere @xmath8 and the velocity of the shot @xmath55 .
the maximum of @xmath27 in our calculations is about @xmath56 m / s which results in very small accelerations @xmath57 .
in addition to the air resistance we included the wind velocity in our calculations . + we confirm the results of tutevich ( 1969 ) @xcite .
as expected the effect of the air resistance turns out to be very small .
tutevich stated that for headwind with a velocity of @xmath58 m / s the shot is about @xmath59 cm reduced for @xmath60 m / s compared to the value of @xmath18 without wind .
he also stated that for tailwind one will find an increased value of @xmath61 cm at @xmath60 m / s compared to a windless environment .
we could verify the calculations of tutevich ( 1969 ) @xcite and obtain some additional information as listed in table [ tab2 ] by incorporating these effects in a small computer program that can be downloaded from the internet , see rappl ( 2010)@xcite . an interesting fact is that headwind reduces the shot more than direct wind from above ( which could be seen as small @xmath62 factor corrections ) .
if the values of @xmath27 , @xmath10 and @xmath11 are known ( measured ) precisely then the results of our program are exact .
+ now one can try to find again the optimal angle of release .
lichtenberg and wills ( 1978 ) @xcite find that the optimal angle is reduced compared to our previous determination by about @xmath63 for some typical values of @xmath27 and @xmath10 and by still assuming that @xmath27 , @xmath10 and @xmath11 are independent of each other .
+ due to the smallness of this effect compared to the remaining discrepancy of about @xmath64 between the predicted optimal angle of release and the measurements we neglect air resistance in the following .
next we give up the assumption that the parameters @xmath11 , @xmath27 and @xmath10 are independent variables .
this was suggested e.g. in tricker and tricker ( 1967 ) @xcite , hay ( 1973 ) @xcite , dyson ( 1986 ) @xcite , hubbard ( 1988 ) @xcite , de mestre ( 1990 ) @xcite , de mestre et al .
( 1998 ) @xcite , maheras ( 1998 ) @xcite , yeadon ( 1998 ) @xcite , linthorne ( 2001 ) @xcite and de luca ( 2005 ) @xcite .
we will include three effects : the dependence of the height of release from the angle of release , the angle dependence of the force of the athlete and the effect of gravity during the delivery phase .
the height of the point , where the shot is released depends obviously on the arm length and on the angle @xmath65 with the height of the shoulder @xmath66 and the length of the arm @xmath67 .
clearly this effect will tend to enhance the value of the optimal angle of release , since a larger angle will give a larger value of @xmath10 and this will result in a larger value of @xmath18 .
this effect was studied in detail e.g. in de luca ( 2005)@xcite .
we redid that analysis and confirm the main result of that work .
should have a different sign . ]
the optimal angle can be expanded in @xmath68 @xmath69 as expected above we can read off from this formula that the optimal angle of release is now enhanced compared to the analysis of lichtenberg and wills ( 1976 ) , @xmath70 for typical values of @xmath27 , @xmath66 and @xmath67 de luca ( 2005)@xcite gets an increase of the optimal angle of release in the range of @xmath71 to @xmath72 . with the inclusion of this effect
the problem of predicting the optimal angle of release has become even more severe .
the world records in bench - pressing are considerably higher than the world records in clean and jerk .
this hints to the fact that athletes have typically most power at the angle @xmath73 compared to larger values of @xmath11 .
this effect that is also confirmed by experience in weight training , was suggested and investigated e.g. by mcwatt ( 1982)@xcite , mccoy ( 1984)@xcite , gregor ( 1990)@xcite and linthorne ( 2001)@xcite . the angle dependence of the force of the athlete can be measured and then be used as an input in the theoretical investigation . below we will use a very simple model for the dependence to explain the consequences .
this effect now tends to favor smaller values for the optimal angle of release .
finally one has to take into account the fact , that the energy the athlete can produce is split up in potential energy and in kinetic energy .
@xmath74 where @xmath75 .
hence , the higher the athlete throws the lower will be the velocity of the shot .
since the achieved distance at shot put depends stronger on @xmath27 than on @xmath10 this effect will also tend to giver smaller values for the optimal angle of release .
this was investigated e.g. in tricker and tricker ( 1967 ) @xcite , zatsiorski and matveev ( 1969 ) @xcite , hay ( 1973 ) @xcite and linthorne ( 2001)@xcite .
now we put all effects together . from eq.([arm ] ) and eq.([energy ] ) we get @xmath76 the angle dependence of the force of the athlete will result in an angle dependence of the energy an athlete is able to transmit to the put @xmath77 the function @xmath78 can in principle be determined by measurements with individual athletes . from these two equations and from eq.([arm ] ) we get @xmath79 , \label{v0th } \\
h ( \theta ) & = & h_s + b \sin \theta .
\end{aligned}\ ] ] inserting these two @xmath11-dependent functions in eq.([xm ] ) we get the full @xmath11-dependence of the maximum distance at shot put @xmath80 the optimal angle of release is obtained as the root of the derivative of @xmath81 with respect to @xmath11 . to obtain numerical values for the optimal angle
we need to know @xmath78 . in principle
this function is different for different athletes and it can be determined from measurements with the athlete . to make some general statements we present two simple toy models for @xmath82 .
we use the following two simple toy models for @xmath78 @xmath83 this choice results in @xmath84 for @xmath73 and @xmath85 for @xmath86 , which looks reasonable . at this stage
we want to remind the reader again : this ansatz is just supposed to be a toy model , a decisive analysis of the optimal angle of release will have to be done with the measured values for @xmath78 .
we extract the normalization @xmath87 from measurements @xmath88 , \\ \frac{e_{2,0}}{m } & = & \frac{e_2(\theta)}{m f_2(\theta ) } = \frac{1}{1- \frac{2}{3 } \frac{\theta}{\pi } } \left [ g b \sin \theta + \frac{v_0 ^ 2}{2 } \right].\end{aligned}\ ] ] with the average values of table [ tab1 ] ( @xmath89 m , @xmath90 m / s and @xmath91 ) and @xmath92 m we get @xmath93 now all parameters in @xmath81 are known .
looking for the maximum of @xmath94 we obtain the optimal angle of release to be @xmath95 which lie now perfectly in the measured range ! + next we can also test the findings of maheras ( 1998)@xcite that @xmath27 decreases linearly with @xmath11 by plotting @xmath96 from eq.([v0th ] ) against @xmath11 .
we find our toy model 1 gives an almost linear decrease , while the decrease of toy model looks exactly linear .
+ our simple but reasonable toy models for the angle dependence of the force of the athlete give us values for optimal release angle of about @xmath97 , which coincide perfectly with the measured values .
moreover they predict the linear decrease of @xmath27 with increasing @xmath11 as found by maheras ( 1998)@xcite
in this paper we have reinvestigated the biomechanics of shot put in order to determine the optimal angle of release .
standard text - book wisdom tells us that the optimal angle is @xmath24 , while measurements of top athletes tend to give values around @xmath98 .
including the effect of the height of the athlete reduces the theory prediction for the optimal angle to values of about @xmath2 ( lichtenberg and wills ( 1978 ) @xcite ) . as the next step
we take the correlation between the initial velocity @xmath9 , the height of release @xmath10 and the angle of release @xmath11 into account .
therefore we include three effects : 1 .
the dependence of the height of release from the angle of release is a simple geometrical relation .
it was investigated in detail by de luca ( 2005 ) @xcite .
we confirm the result and correct a misprint in the final formula of luca ( 2005 ) @xcite .
this effect favors larger values for the optimal angle of release .
the energy the athlete can transmit to the shot is split up in a kinetic part and a potential energy part .
this effect favors smaller values for the optimal angle of release .
3 . the force the athlete can exert to the shot depends also on the angle of release .
this effect favors smaller values for the optimal angle of release .
the third effect depends on the individual athlete . to make decisive statements the angle dependence of the angular dependence of the force has to be measured first and then the formalism presented in this paper
can be used to determine the optimal angle of release for an individual athlete .
+ to make nethertheless some general statements we investigate two simple , reasonable toy models for the angle dependence of the force of an athlete . with these toy models
we obtain theoretical predictions for the optimal angle of @xmath98 , which coincide exactly with the measured values . for our predictions
we do not need initial measurements of @xmath12 and @xmath13 over a wide range of release angles . in that respect our work represents a further developement of linthorne ( 2001)@xcite .
moreover our simple toy models predict the linear decrease of @xmath27 with increasing @xmath11 as found by maheras ( 1998)@xcite .
we thank philipp weishaupt for enlightening discussions , klaus wirth for providing us literature and jrgen rohrwild for pointing out a typo in one of our formulas .
5 a. kuhlow , `` die technik des kugelstoens der mnner bei den olympischen spielen 1972 in mnchen , '' leistungssport * 2 * , ( 1975 ) j. dessureault , `` selected kinetic and kinematic factors involved in shot putting , '' in biomechanics vi - b ( editeb by e. asmussen and k. jrgensen ) , p 51 , baltimore , md : university park press .
m. w. mccoy , r. j. gregor , w.c .
whiting , r.c .
rich and p. e. ward , `` kinematic analysis of elite shotputters , '' track technique * 90 * , 2868 ( 1984 ) .
p. susanka and j. stepanek , `` biomechanical analysis of the shot put , '' scientific report on the second iaaf world chamionships in athletics , 2nd edn , i/1 ( 1988 ) , monaco : iaaf .
k. bartonietz and a. borgstrm , `` the throwing events at the world championships in athletics 1995 , gteborg - technique of the world s best athletes .
part 1 : shot put and hammer throw , '' new studies in athletics * 10 * ( 4 ) , 43 ( 1995 ) .
d. k. tsirakos , r. m. bartlett and i. a. kollias , `` a comparative study of the release and temporal characteristics of shot put , '' journal of human movement studies * 28 * , 227 ( 1995 ) .
p. luhtanen , m. blomquist and t. vnttinen , `` a comparison of two elite putters using the rotational technique , '' new studies in athletics * 12 * ( 4 ) , 25 ( 1997 ) .
r. a. r. tricker and b. j. k. tricker , `` the science of movement , '' london : mills and @xmath99 boon ( 1967 ) .
b. p. garfoot , `` analysis of the trajectory of the shot , '' track technique * 32 * , 1003 ( 1968 ) .
v. m. zatsiorski and e. i. matveev , `` investigation of training level factor structure in throwing events , '' theory and practice of physical culture ( moscow ) * 10 * , 9 ( 1969 ) ( in russian ) .
v. n. tutevich , `` teoria sportivnykh metanii , '' fizkultura i sport , moscow ( in russian ) 1969 . j. g. hay , `` biomechanics of sports techniques , '' englewood cliffs , nj : prentice - hall 1973 ( 4th edn 1993 ) .
d. b. lichtenberg and j. g. wells , `` maximizing the range of the shot put , '' am .
* 46(5 ) * , 546 ( 1978 ) . v. m. zatsiorsky , g. e. lanka and a. a. shalmanov , `` biomechanical analysis of shot putting technique , '' exercise sports sci
* 9 * , 353 ( 1981 ) .
b. mcwatt , `` angles of release in the shot put , '' modern athlete and coach * 20 * ( 4 ) , 17 ( 1982 )
. m. s. townend , `` mathematics in sport , '' 1984 , ellis horwood ltd . ,
chichester , isbn 0 - 853 - 12717 - 4 g. h. g. dyson , `` dyson s mechanics of athletics , '' london : hodder @xmath99 stoughton ( 8th edn .
1986 ) .
m. hubbard , `` the throwing events in track and field , '' in the biomechanics of sport ( edited by c. l. vaughan ) , 213 ( 1988 ) .
boca raton , fl : crc press . n. j. de mestre , `` the mathematics of projectiles in sport , '' 1990 , cambridge university press , cambridge , isbn 0 - 521 - 39857 - 6 r. j. gregor , r. mccoy and w. c. whiting , `` biomechanics of the throwing events in athletics , '' proceedings of the first international conference on techniques in athletics , vol . 1 ( edited by g .-
brggemann and j. k. ruhl ) , 100 ( 1990 ) .
kln : deutsche sporthochschule .
n. j. de mestre , m. hubbard and j. scott , `` optimizing the shot put , '' in proceedings of the fourth conference on mathematics and computers in sport ( edited by n. de mestre and k. kumar ) , 249 ( 1998 ) .
gold coast , qld : bond university .
a. v. maheras , `` shot - put : optimal angles of release , '' track @xmath99 fields coaches review , * 72 * ( 2 ) , 24 ( 1998 ) .
f. r. yeadon , `` computer simulations in sports biomechanics . '' in proceedings of the xvi international symposium on biomechanics in sports ( ed . by h.j .
riehle and m.m .
vieten ) , ( 1998 ) 309 .
kln : deutsche sporthochschule .
n. p. linthorne , `` optimum release angle in the shot put , '' j. sports sci .
* 19 * , 359 ( 2001 ) .
m. hubbard , n. j. de mestre and j. scott , `` dependence of release variables in the shot put , '' j. biomech .
* 34 * , 449 ( 2001 ) .
m. bace , s. illijic and z. narancic , `` maximizing the range of a projectile , '' eur .
* 23 * , 409 ( 2002 ) .
f. mizera and g. horvth , `` influence of environmental factors on shot put and hammer throw range , '' j. biophys .
* 35 * , 785 ( 2002 ) .
r. cross , `` physics of overarm throwing , '' am .
* 72(3 ) * , 305 ( 2004 ) .
u. oswald and h. r. schneebeli , `` mathematische modelle zum kugelstossen , '' vsmp bulletin 94 ( 2004 ) .
www.vsmp.ch/de/bulletins/no94/oswald.pdf r. de luca , `` shot - put kinematics , '' eur
. j. phys .
* 26 * , 1031 ( 2005 ) .
nicholas p. linthorne , `` throwing and jumping for maximum horizontal range , '' arxive : physics/0601148 .
w. e. red and a. j. zogaib , `` javelin dynamics including body interaction , '' journal of applied mechanics * 44*,496 ( 1977 ) .
.[tab1]compendium of some data measured during the summer olympic games 1972 from kuhlow@xcite [ cols="<,^,^,^,^,^,^",options="header " , ] | we determine the optimal angle of release in shot put .
the simplest model - mostly used in textbooks - gives a value of @xmath0 , while measurements of top athletes cluster around @xmath1 .
including simply the height of the athlete the theory prediction goes down to about @xmath2 for typical parameters of top athletes . taking further the correlations of the initial velocity of the shot , the angle of release and the height of release into account we predict values around @xmath1 , which coincide perfectly with the measurements .
do - th 10/12 |
the future power system will be modernized with advanced metering infrastructure , bilateral information communication network , and intelligent monitoring and control system to enable a smarter operation @xcite .
the transformation to the smart grid is expected to facilitate the deep integration of renewable energy , improve the reliability and stability of the power transmission and distribution system , and increase the efficiency of power generation and energy consumption .
demand response program is a core subsystem of the smart grid , which can be employed as a resource option for system operators and planners to balance the power supply and demand .
the demand side control of responsive loads has attracted considerable attention in recent years @xcite .
an intelligent load control scheme should deliver a reliable power resource to the grid , while maintaining a satisfactory level of power usage to the end - user .
one of the greatest technical challenges of engaging responsive loads to provide grid services is to develop control schemes that can balance the aforementioned two objectives @xcite .
to achieve such an objective , a hierarchical load control structure via aggregators is suggested to better integrate the demand - side resources into the power system operation and control @xcite . in the hierarchical scheme
, the aggregator performs as an interface between the loads and the system operator .
it aggregates the flexibility of responsive loads and offers it to the system operator . in the meantime
, it receives dispatch signals from the system operator , and execute appropriate control to the loads to track the dispatch signal .
therefore , an aggregate flexibility model is fundamentally important to the design of a reliable and effective demand response program . it should be detailed enough to capture the individual constraints while simple enough to facilitate control and optimization tasks . among various modeling options for the adjustable loads such as thermostatically controlled loads ( tcls )
, the average thermal battery model @xcite aims to quantify the aggregate flexibility , which is the set of the aggregate power profiles that are admissible to the load group .
it offers a simple and compact model to the system operator for the provision of various ancillary services .
apart from the adjustable loads , deferrable loads such as pools and plug - in vehicles ( pevs ) can also provide significant power flexibility by shifting their power demands to different time periods .
however , different from the adjustable loads , it is more difficult to characterize the flexibility of deferrable loads due to the heterogeneity in their time constraints . in this paper , we focus on modeling the aggregate flexibility for control and planning of a large number of deferrable loads .
there is an ongoing effort on the characterization of the aggregate flexibility of deferrable loads @xcite .
an empirical model based on the statistics of the simulation results was proposed in @xcite .
a necessary characterization was obtained in @xcite and further improved in @xcite . for a group of deferrable loads with homogeneous power , arrival time , and departure time ,
a majorization type exact characterization was reported in @xcite . with heterogeneous departure times and energy requirements ,
a tractable sufficient and necessary condition was obtained in @xcite , and was further utilized to implement the associated energy service market @xcite . despite these efforts , a sufficient characterization of the aggregate flexibility for general heterogeneous deferrable loads remains a challenge . to address this issue
, we propose a novel geometric approach to extract the aggregate flexibility of heterogeneous deferrable loads .
geometrically , the aggregate flexibility modeling amounts to computing the minkowski sum of multiple polytopes , of which each polytope represents the flexibility of individual load .
however , calculating the minkowski sum of polytopes under facet representation is generally np - hard @xcite .
interestingly , we are able to show that for a group of loads with general heterogeneity , the exact aggregate flexibility can be characterized analytically .
but the problem remains in the sense that there are generally exponentially many inequalities with respect to the number of loads and the length of the time horizons , which can be intractable when the load population size or the number of steps in the considered time horizon is large .
therefore , a tractable characterization of the aggregate flexibility is desired . for deferrable loads with heterogeneous arrival and departure times , the constraint sets are polytopes that are contained in different subspaces .
alternative to the original definition of the minkowski sum , we find it beneficial to regard it as a projection operation . from the latter perspective
, the aggregate flexibility is considered as the projection of a higher dimensional polytope to the subspace representing the aggregate power of the deferrable loads .
therefore , instead of approximating the minkowski sum directly by its definition , we turn to approximating the associated projection operation . to this end , we formulate an optimization problem which approximates the projection of a full dimensional polytope via finding the maximum homothet of a given polytope , i.e. , the dilation and translate of that polytope .
the optimization problem can be solved very efficiently by solving an equivalent linear program .
furthermore , we propose a `` divide and conquer '' strategy which enables efficient and parallel computation of the aggregate flexibility of the load group .
the scheduling policy for each individual load is derived simultaneously along the aggregation process .
finally , we apply our model to the pev energy arbitrage problem , where given predicted day - ahead energy prices , the optimal power profile consumed by the load group is calculated to minimize the total energy cost .
the simulation results demonstrate that our approach is very effective at characterizing the feasible aggregate power flexibility set , and facilitating finding the optimal power profile .
there are several closely - related literature on characterizing flexibility of flexible loads . in our previous work
@xcite , a geometric approach was proposed to optimally extract the aggregate flexibility of heterogeneous tcls based on the given individual physical models .
the simulation demonstrated accurate characterization of the aggregate flexibility which was very close to the exact one .
however , this approach can not be applied to the deferrable loads directly .
similar to @xcite which sought a special class of polytopes to facilitate fast calculation of minkowski sum , the authors in @xcite proposed to characterize the power flexibility using zonotopes .
different from @xcite , this method could deal with the time heterogeneity as appeared in the deferrable loads .
in addition , both approaches extracted the flexibility of _ individual _ load . * * in comparison , the approach proposed in this paper is a _ batch _ processing method : it directly approximates the aggregate flexibility of a group of loads , which could mitigate the losses caused by the individual approximation as emphasized in @xcite .
* notation * : the facet representation of a polytope @xmath0 is a bounded solution set of a system of finite linear inequalities @xcite : @xmath1 , where throughout this paper @xmath2 ( or @xmath3 , @xmath4 , @xmath5 ) means elementwise inequality . a polytope @xmath0 is called full dimensional , if it contains an interior point in @xmath6 .
given a full dimensional polytope @xmath7 in @xmath6 , a scale factor @xmath8 , and a translate factor @xmath9 , the set @xmath10 is called a homothet of @xmath11 .
we use @xmath12 to denote the minkowski sum of multiple sets , and @xmath13 of two sets .
we use @xmath14 to represent the @xmath15 dimensional column vector of all ones , @xmath16 the @xmath15 dimensional identity matrix , and @xmath17 the indicator function of the set @xmath18 .
the bold @xmath19 denotes the column vector of all @xmath20 s with appropriate dimension . for two column vectors
@xmath21 and @xmath22 , we write @xmath23 for the column vector @xmath24^{t}$ ] where no confusion shall arise .
we consider the problem of charging a group of @xmath25 pevs .
the energy state of each pev can be described by a discrete time difference equation on a finite time horizon @xmath26 $ ] , @xmath27 where @xmath28 is the state of charge ( soc ) with initial condition @xmath29 , and @xmath30 $ ] with @xmath31 is the charging power supplied to the vehicle during @xmath32 .
let @xmath33 denote the time interval @xmath32 , where without loss of generality , we will assume the time unit @xmath34 hour in the sequel .
moreover , the pev must be charged during a time window @xmath35 where @xmath36 is its arrival time , @xmath37 is its departure time , and @xmath38 .. at the deadline @xmath37 , the pev is supposed to be charged with an soc @xmath39 $ ] , where we assume that @xmath40 .
the load is called deferrable if @xmath41 .
a charging power profile @xmath42^{t}$ ] is called admissible if the load is only charged within @xmath43 and its soc at @xmath44 is within @xmath45 $ ] .
we differentiate the @xmath46 pev ( @xmath47 ) by using a superscript @xmath48 on the variables introduced above .
the charging task of the @xmath46 pev is determined by @xmath49 .
let @xmath50 be the set of all the admissible power profiles of the @xmath46 load .
it can be described as , @xmath51,\forall t\in\mathbb{a}^{i},\\ u_{t}^{i}=0,\forall i\in\mathrm{\mathbb{t}}\backslash\mathbb{\mathbb{a}}^{i}\mbox { and } { \bf 1}_{m}^{t}u^{i}\in[\underline{e}^{i},\bar{e}^{i } ] \end{array}\right\ } \label{eq : pi}\ ] ] it is straightforward to see that each @xmath50 is a convex polytope .
in addition , we say @xmath50 is of codimension @xmath52 if its affine hull is a @xmath53 dimensional subspace of @xmath6 . in the smart grid ,
the aggregator is responsible for procuring a generation profile from the whole market to service a group of loads .
we define the generation profile that meets the charging requirements of all pevs as follows . a generation profile @xmath21 is called _ adequate _ if there exists a decomposition @xmath54 , such that @xmath55 is an admissible power profile for the @xmath46 load , i.e. , @xmath56 .
we call the set of all the adequate generation profiles the _ aggregate flexibility _ of the load group
. it can be defined as the minkowski sum of the admissible power sets of each load , @xmath57 it is straightforward to show that @xmath7 is also a convex polytope whose codimension is to be determined by the parameters @xmath58 of the deferrable loads .
the numerical complexity of the existing algorithms for calculating the minkowski sum is rather expensive ( see @xcite for some numerical results ) . in general , calculating @xmath59 when @xmath60 and @xmath61 are polytopes specified by their facets is np - hard @xcite .
however , for the particular problem of pev charging , it is possible to characterize the exact aggregate flexibility @xmath7 analytically .
such characterization is built on the results from the matrix feasibility problem and from the network flow theory , both of which are intrinsically connected with the pev charging problem .
[ thm : iff]consider a group of pevs or deferrable loads with heterogeneous parameters @xmath62 , @xmath63 .
then the set @xmath7 of adequate generation profiles consists of those @xmath64^{t}$ ] which satisfy @xmath65 for all subsets @xmath66 and @xmath67 , where @xmath68 and @xmath69 are the complement sets of @xmath70 and @xmath71 in @xmath72 and @xmath73 , respectively .
we first interpret the characterization of @xmath7 as a matrix feasibility problem . by definition ,
if a generation profile @xmath21 is adequate , then there exists a decomposition @xmath54 such that @xmath55 completes the @xmath46 pev s charging task .
this is equivalent to the existence of a @xmath74 matrix @xmath75 , the @xmath46 row of which is an admissible power profile of the @xmath46 pev .
the matrix @xmath75 will be referred to as the charging matrix .
given @xmath76 , let @xmath77 denote the set of all such charging matrices .
these matrices have special structures : the columns indexed by @xmath78 in the @xmath46 row are the free positions which can be filled with a real number in @xmath79 $ ] , while the rest of the positions in this row are forbidden positions that can only be filled with @xmath20 s .
moreover , @xmath80 , it has the @xmath81 column sum @xmath82 and the @xmath46 row sum in the interval @xmath83 $ ] .
clearly , the non - emptiness of the set gives the condition for being an adequate generation profiles , i.e , @xmath84 furthermore , the condition for @xmath85 can be derived by applying ( * ? ? ?
* theorem 2.7 ) to the matrix case . by the definition in @xcite ,
partitions @xmath86 and @xmath87 of a sequence are said to be orthogonal , if @xmath88 , @xmath89 and @xmath90 . for a matrix ,
clearly the rows and columns constitute such orthogonal partitions .
this is the only condition required by ( * ? ? ?
* theorem 2.7 ) .
then by a direct calculation of the summation on the right hand side of ( * ? ? ?
* ( 2.8 ) ) , we can obtain ( [ eq : fullhetero ] ) .
this completes the proof . in general , there are @xmath91 inequalities in ( [ eq : fullhetero ] ) , which will be intractable if @xmath25 is several thousand .
when the pevs are fully homogeneous , i.e. , they share the same set of parameters @xmath58 , the above result reduces to the well - known majorization condition @xcite , which consists of only @xmath92 inequalities . in the linear algebra literature ,
studies on the matrix feasibility problem are also focused on finding tractable conditions under limited heterogeneities in the parameters . adapted to the pev charging scenario , if the arrival time @xmath36 and the charging rate @xmath93 are homogeneous , @xmath94 , and under certain monotonicity condition on @xmath21 , the number of inequalities in ( [ eq : fullhetero ] )
can be significantly reduced @xcite .
the author in @xcite obtained a simple majorization type result under a special monotonicity condition on @xmath21 . under this condition , the charging rate @xmath93 can be relaxed to be heterogeneous both among different pevs and at different time instances .
note that theorem [ thm : iff ] also applies to the case where the charging rate takes integer values ( see ( * ? ? ?
* remark ( 2.19 ) ) ) .
since the condition ( [ eq : fullhetero ] ) is very difficult to check in practice , the goal of this paper is to find a sufficient approximation of it using much fewer inequalities that does not depend on @xmath25 .
the direct approximation from ( [ eq : fullhetero ] ) could be difficult .
however , it is possible to start with the definition ( [ eq : p_af ] ) . an interesting perspective is to view the minkowski sum as a projection operation .
clearly , from ( [ eq : p_af ] ) , we see that @xmath7 is the projection of a higher dimensional polytope @xmath95 onto the @xmath21 subspace , i.e. @xmath96 where @xmath97 @xmath98 and @xmath99 is the projection onto the @xmath21 subspace .
in fact , the minkowski sum of two sets is often calculated via projection .
note that the number of the facets of the polytope @xmath95 is of @xmath100 , as compared to @xmath101 of its projection .
the relation ( [ eq : prj ] ) inspires us to approximate @xmath7 based on only the expression of @xmath95 .
the specific approximation method will be described in the next section .
we will first present our method in a general setting of computing the maximum homothet of a polytope included in a polytopic projection , and then apply it to the pev charging scenario .
our approximation method is inspired by @xcite , where the ellipsoidal approximation of a polytopic projection is addressed resorting to the robust optimization technique @xcite .
given full dimensional polytopes @xmath102\leq c\},\end{aligned}\ ] ] we want to find its maximum homothet of @xmath11 contained in the projection of @xmath95 onto the @xmath21 subspace .
it can be formulated as the following optimization problem @xmath103 to facilitate the later formulation of the optimization problems as linear programs , we perform a change of variables @xmath104 and @xmath105 .
thus problem ( [ eq : opp ] ) is equivalent to finding the minimum homothet of @xmath106 that contains @xmath11 , i.e. @xmath107 furthermore , since orthogonal projection is a linear operation , we have @xmath108 where @xmath109 $ ] is the lift of the vector @xmath110 in @xmath111 by setting the additional dimensions to @xmath19 .
hence , @xmath112 for some @xmath113 implies the constraint in ( [ eq : opp1 ] ) .
therefore , it is sufficient to pose a more restrictive constraint to obtain a suboptimal solution , and we have @xmath114 where the constraint can be expressed as @xmath115\leq sc+b\tilde{r}\right\ } .\ ] ] by applying the farkas s lemma ( see the appendix [ subsec : farkaslemma ] ) , the above optimization problem can be transformed into the following linear programming problem , @xmath116,\\ & gh\leq b\left[\begin{array}{c } r\\ -\bar{u}_{0 } \end{array}\right]+sc .
\end{array}\label{eq : opp3}\ ] ] before proceed , we illustrate the formulation ( [ eq : opp2 ] ) using a simple example borrowed from @xcite .
[ exa : test]let @xmath95 be given by @xmath117 , and @xmath118 .
the polytope @xmath95 is plotted in fig .
[ fig : example ] .
we solve the problem ( [ eq : opp3 ] ) to find a sufficient approximation of @xmath119 , and obtain @xmath120 , @xmath121 , @xmath122 .
the corresponding scale factor is @xmath123 , and translate factor is @xmath124 . from these data
we can have that @xmath125^{t}\subset\tilde{\mathcal{p}}$ ] .
this corresponds to the fact that the longest horizontal line segment that is contained in @xmath95 is at @xmath126 .
the approximation of the projection of @xmath95 via problem ( [ eq : opp2 ] ) . ] clearly , the formulation of problem ( [ eq : opp2 ] ) is very conservative .
it actually requires the homothet of the polytope @xmath11 be entirely contained in @xmath95 .
it amounts to each time fixing @xmath127 , and then measuring the cross section of @xmath95 .
however , for approximating the projection of @xmath95 , we only need @xmath128 where @xmath129 is a function of @xmath21 , while in problem ( [ eq : opp2 ] ) @xmath21 is determined by @xmath129 .
the relation ( [ eq : dependence ] ) can be interpreted in the context of the adjustable robust optimization problem @xcite , where @xmath21 is the so called non - adjustable variable , and @xmath129 is the adjustable variable . the function @xmath130 is called the decision rule .
solving ( [ eq : opp2 ] ) over all possible choices of @xmath130 is intractable . an efficient way to overcome
this is to restrict the choice of @xmath130 to be the affine decision rules , @xmath131 where @xmath132 , and @xmath133 . using ( [ eq : decision ] ) in problem ( [ eq : opp2 ] ) and by some manipulation , we can obtain @xmath134u\leq b\left[\begin{array}{c } r\\ -v \end{array}\right]+sc\right\ } .
\end{array}\label{eq : oppapp}\ ] ] using the farkas s lemma , problem ( [ eq : oppapp ] ) can be solved by the following linear programming @xmath135,\\ & gh\leq b\left[\begin{array}{c } r\\ -v \end{array}\right]+sc . \end{array}\ ] ] we test the above formulation by computing the approximation of the polytopic projection in example [ exa : test ] .
we solve the problem ( app ) and obtain @xmath136 , @xmath137 .
this corresponds to the interval @xmath138 $ ] , which is exactly @xmath119 . in general
, the problem ( app ) gives a suboptimal solution for the approximation of @xmath119 with respective to @xmath11 . a possible way to reduce
the conservativeness is to employ the quadratic decision rule or other nonlinear decision rules as reported in @xcite . in this subsection
, the polytopic projection approximation developed in the above section will be employed to aggregate the pevs flexibility .
we will discuss several issues including the choice of the nominal model @xmath11 , the preprocessing of the charging constraints , and the strategy for parallel computation .
finally , the explicit formulae for the flexibility model and the corresponding scheduling policy are derived .
intuitively , one can choose the nominal polytope @xmath11 to be of the similar form of ( [ eq : pi ] ) , and the parameters can be taken as the mean values of the pev group . more generally , we can define the virtual battery model as follows .
the set @xmath139 is called a @xmath15-horizon discrete time virtual battery with parameters @xmath140 , if @xmath141\}.\ ] ] @xmath139 is called a sufficient battery if @xmath142 .
conceptually , the virtual battery model mimics the charging / discharging dynamics of a battery .
we can regard @xmath21 as the power draw of the battery , @xmath143 and @xmath144 as its discharging / charging power limits , and @xmath145 and @xmath146 as the energy capacity limits .
geometrically , it is a polytope in @xmath6 with @xmath147 facets , which is computationally very efficient when posed as the constraint in various optimization problems .
note that the original high - dimensional polytope @xmath95 defined in ( [ eq : ptilde ] ) contains equality constraints , which is not full dimensional .
therefore , first we have to remove the equalities by substituting the variables . for simplicity , assume that @xmath40 .
more explicitly , let @xmath148 be the @xmath46 pev s charging profile at time @xmath149 , @xmath89 , and @xmath150 , @xmath82 be the generation profile at time @xmath149 .
the overall charging constraints of the pev group can be written as follows @xmath151 which is a polytope in @xmath152 , and the coordinate is designated to be @xmath153 .
we first need to eliminate the equality constraints containing @xmath154 , i.e. , the first line of ( [ eq : original ] ) .
to standardize the elimination process , define @xmath155 , which is the index set of the pevs that can be charged at time @xmath149 . without loss of generality , assume that @xmath156 , @xmath150 , and we substitute @xmath157 , @xmath150 in ( [ eq : original ] ) , where @xmath158 is the first pev in the set @xmath159 .
let @xmath160 be the set of time instants at which the substitution of @xmath148 is made .
further , we remove the coordinate @xmath161 and obtain , @xmath162 where the new coordinate becomes @xmath163 with @xmath164 and @xmath165_{t\in\mathbb{a}^{i}\backslash s_{i}}$ ] , @xmath63 .
clearly , @xmath166 has a dimension @xmath167 , and there are a number of @xmath168 linear inequalities in ( [ eq : eliminated ] ) .
we denote it by the matrix form , @xmath169\leq c,\label{eq : bm}\ ] ] where note that @xmath170 is a sparse matrix and has the structure , @xmath171.\ ] ] for a fixed time horizon @xmath15 , both the number of the decision variables and the number of inequality constraints of problem ( app ) increase linearly with @xmath172 .
when the number of pevs to be aggregated is too large , solving problem ( app ) would be intractable . to address the increasing numerical complexity
, we propose to divide the pevs into small groups , and solve problem ( app ) for each group with respect to the same nominal model @xmath11 . denoting the solutions of problem ( app ) for the @xmath173 group by @xmath174 , the aggregate flexibility of the @xmath173 group
is given by @xmath175 .
then the flexibility of the overall pev group can be calculated directly based on the following lemma . [
lem : mksksumscale]let @xmath11 be a polytope , and @xmath176 , @xmath177 be non - negative scalars
. then @xmath178 .
the above result can be easily verified .
a more general proof for the convex body can be found in ( * ? ? ? * remark 1.1.1 ) ) . by this `` divide and conquer '' strategy
, the original highly complex optimization problem can be solved very efficiently in a parallel fashion . however , this increases the conservativeness of the approximation , which is a result of the trade - off between the tractability and the optimality . in case
that different nominal models @xmath179 are used for each group , we can perform the aggregation again over the obtained groups . repeating this after several stages , we can arrive one virtual battery model for the overall pev group .
even though we have to spread the computation over time in different stages , in practice this process terminates soon since the number of stages is of order @xmath180 when @xmath181 pevs / groups are processed at each run of problem ( app ) . combining the above development
, the following explicit formulae for designing the sufficient virtual battery can be derived . the scheduling policy for each individual pevs
can also be obtained . without loss of generality ,
these results are stated for the case where only one stage aggregation is executed .
the formulae for multi - stage aggregation can be obtained analogously .
for convenience , let us denote the solutions of the problem ( app ) by the output of the function @xmath182 where @xmath183 is the high - dimensional polytope associated with the @xmath173 group of pevs , and @xmath11 is a given nominal model parameterized by @xmath184 .
the proof of the following theorem can be found in the appendix [ subsec : proof - of - theorem ] .
[ thm : suffvb]suppose @xmath185 , @xmath186 .
then @xmath187 is a sufficient battery parameterized by @xmath188 where @xmath189 , @xmath190 and @xmath191 , @xmath192 .
furthermore , @xmath193 the scheduling policy is given by @xmath194 where @xmath195 denotes the charging profiles of the number of @xmath196 pevs in the @xmath173 group .
the maximum and the minimum charging rates of the designed virtual battery model . ] aggregate charging profile versus the price . ]
in this section , we consider coordination of a group of 1000 pevs for energy arbitrage .
the considered time horizon is 24 hours , and the price is taken as the day - ahead energy market locational marginal pricing ( lmp ) @xcite .
the parameters of pevs are randomly generated by their types and the corresponding probability distributions ( see @xcite for more details ) .
since most of the pevs arrive during the afternoon to midnight and leave during the next 12 hours , we choose to simulate from 12:00 noon to the same time on the next day .
in addition , we assume a @xmath197 total charging energy flexibility around the nominal energy requirement .
the ( app ) problem is solved using the glpk linear programming solver @xcite interfaced with yalmip @xcite . at the first stage ,
we randomly divide the 1000 pevs into 100 groups , where each group contains 10 pevs .
the aggregate flexibility approximation is thus solved for 10 pevs in one group .
this number is chosen according to the numerical efficiency of the solver .
the parameters of the nominal battery model @xmath11 for each group are chosen as the average values of the group .
for those groups having the same minimum arrival time and maximum departure time , we take the average values between them and approximate their flexibilities using the same nominal model . for example , by setting the elements in the charging matrix @xmath75 by their maximum values , the upper charging limits @xmath144 for the nominal battery model are calculated as the column averages of @xmath75 .
since these groups share the same nominal model , their approximated aggregated flexibilities can be calculated very easily based on lemma [ lem : mksksumscale ] . in our simulation , after the first stage , the 100 groups are merged into 22 collections which are represented by polytopes of different codimensions
. then we repeat the process in the first stage to approximate the flexibility of these 22 collections , where each time we aggregate 11 collections .
finally a sufficient battery model @xmath198 is obtained for the characterization of the aggregate flexibility of the entire loads group .
the dynamic charging limits of the obtained battery model are illustrated in fig .
[ fig : bounds ] .
the total charging energy bounds are obtained as @xmath199 mwh and @xmath200 mwh , which lies in the true aggregate energy consumption interval @xmath201 $ ] mwh . from fig .
[ fig : bounds ] , we can see that around the midnight ( the @xmath202 hour ) , the charging flexibility of the pevs are the largest in terms of the difference of the charging rate bounds . denoting the energy price by @xmath203 , and the planned energy by @xmath204 , the energy arbitrage problem can be formulated as a linear programming problem as follows , @xmath205 clearly , the above optimization problem can be solved much more efficiently than directly optimizing the power profiles subject to the constraints of 1000 pevs .
we plot the obtained power profiles against the price changes in fig .
[ fig : planing ] .
it can be observed that most of the energy demand are consumed during @xmath206 @xmath207 to @xmath208 @xmath209 in the morning , when the prices are at its lowest .
the same curve of the planned power is also plotted in fig . [
fig : bounds ] ( the dotted line ) , where note we assume that the time discretization unit is @xmath210 hour .
we see that the planned charging rate lies in the charging bounds of @xmath198 , and almost always matches the maximum / minimum bound . using this charging profile ,
the total energy being charged to the pevs is @xmath211 mwh which lies in the interval @xmath212 .
hence , it is adequate and the charging requirement of individual pev can be guaranteed by using the scheduling policy ( [ eq : schdplcy ] ) .
we choose the immediate charging policy as the baseline and use it to compare with the obtained optimal charging profile in fig .
[ fig : bounds ] . to ensure a fair comparison
, we impose an additional constraint that the total energies consumed by both profiles are the same .
the total energy cost for the baseline charging profile is @xmath213 , while the cost for the optimal charging profile is @xmath214 , which reduces the baseline cost by about @xmath215 .
this paper proposed a novel polytopic projection approximation method for extracting the aggregate flexibility of a group of heterogeneous deferrable loads .
the aggregate flexibility of the entire load group could be extracted parallelly and in multiple stages by solving a number of linear programming problems .
the scheduling policy for individual load was simultaneously derived from the aggregation process .
finally , a pev energy arbitrage problem was solved to demonstrate the effectiveness of our approach at characterizing the feasible aggregate power flexibility set , and facilitating finding the optimal power profile .
our future work includes studying the performances of using other decision rules such as the quadratic decision rule and the nonlinear decision rule , and as compared to the method using zonotopes .
in addition , it is interesting to consider a probabilistic description of the aggregate flexibility as in practice the uncertainty of the loads parameters must be considered to reduce the risk of over - estimating or under - estimating the aggregated flexibility . for the sake of completeness , we restate the following version of farkas s lemma as used in @xcite , which will be used to derive our algorithm for approximating the polytopic projection . its proof can be found in @xcite .
[ lem : farkas](farkas lemma ) suppose that the system of inequalities @xmath216 @xmath217 has a solution and that every solution satisfies @xmath218 @xmath219 .
then there exists @xmath220 , @xmath221 , such that @xmath222 and @xmath223 .
\(1 ) sufficient battery : since @xmath224 is the solution of the app problem , we have @xmath225 , where @xmath226 and @xmath227 .
let @xmath228 and it can be shown that , @xmath229 where @xmath230 , and the last equality is derived by using lemma [ lem : mksksumscale ] . since @xmath231
we see @xmath198 is sufficient .
now suppose @xmath232 , then @xmath233 .
denoting @xmath11 by its facet representation , we have @xmath234 , where @xmath235^{t},\\ h= & ( \bar{u},-\underline{u},\bar{e},-\underline{e}),\end{aligned}\ ] ] and then parameters of @xmath198 can be obtained from @xmath236
\(2 ) scheduling policy : given a generation profile @xmath21 in @xmath198 , we can decompose it into the individual admissible power profile @xmath166 through two steps .
first , we decompose it into the generation profiles for each groups : @xmath237 , by part ( 1 ) we know @xmath238 and further more @xmath239 denoting the generation profile for the @xmath173 group by @xmath240 and hence , @xmath241 which is the aggregate flexibility extracted from the @xmath173 group .
it can be further decomposed into each pevs in the @xmath173 group .
now we need to use the linear decision rule in ( [ eq : decision ] ) .
note that the decision rule ( [ eq : decision ] ) is applied in ( [ eq : oppapp ] ) which actually maps from @xmath11 to @xmath242 , while the decomposition mapping we need is actually from @xmath243 to @xmath244 .
the mappings between these four polytopes form a commutative diagram ( see below ) .
observe that the linear ratio of the decomposition mapping does not change , and only the translate vector needs to be calculated .
therefore , assume that the decomposition takes the form @xmath245 where @xmath246 is the translate vector to be determined . from the above commutative diagram , we must have @xmath247 , @xmath248 solve the above equality and we will have @xmath249 and the overall scheduling policy ( [ eq : schdplcy ] ) follows from the composition of @xmath250 and @xmath251 .
j. bhatt , v. shah , and o. jani , `` an instrumentation engineer s review on smart grid : critical applications and parameters , '' _ renewable and sustainable energy reviews _ , vol .
12171239 , dec . 2014 .
d. s. callaway , `` tapping the energy storage potential in electric loads to deliver load following and regulation with application to wind energy , '' _ energy conversion and management _ , vol .
50 , no . 5 , pp .
1389 1400 , may 2009 .
m. henk , j. richter - gebert , and g. m. ziegler , `` basic properties of convex polytopes , '' in _ handbook of discrete and computational geometry _
, j. e. goodman and j. orourke , eds.1em plus 0.5em minus 0.4emboca raton , fl , usa : crc press , inc . , 1997 , ch .
243270 .
d. hershkowitz , a. j. hoffman , and h. schneider , `` on the existence of sequences and matrices with prescribed partial sums of elements , '' _ linear algebra and its applications _ , vol .
265 , pp . 7192 ,
m. kamgarpour , c. ellen , s. soudjani , s. gerwinn , j. mathieu , n. mullner , a. abate , d. callaway , m. franzle , and j. lygeros , `` modeling options for demand side participation of thermostatically controlled loads , '' in _ bulk power system dynamics and control - ix optimization , security and control of the emerging power grid , irep symposium _ , aug .
2013 , pp . 115 .
s. koch , j. mathieu , and d. callaway , `` modeling and control of aggregated heterogeneous thermostatically controlled loads for ancillary services , '' in _
17th power system computation conference _ , stockholm , sweden , august 2011 .
j. liu , s. li , w. zhang , j. l. mathieu , and g. rizzoni , `` planning and control of electric vehicles using dynamic energy capacity models , '' in _ the 52nd ieee annual conference on decision and control _ , dec .
2013 , pp . 379384 .
a. nayyar , j. taylor , a. subramanian , k. poolla , and p. varaiya , `` aggregate flexibility of a collection of loads , '' in _ the 52nd ieee annual conference on decision and control _ , dec .
2013 , pp . 56005607 . | aggregation of a large number of responsive loads presents great power flexibility for demand response . an effective control and coordination scheme of flexible loads requires an accurate and tractable model that captures their aggregate flexibility .
this paper proposes a novel approach to extract the aggregate flexibility of deferrable loads with heterogeneous parameters using polytopic projection approximation .
first , an exact characterization of their aggregate flexibility is derived analytically , which in general contains exponentially many inequality constraints with respect to the number of loads . in order to have a tractable solution
, we develop a numerical algorithm that gives a sufficient approximation of the exact aggregate flexibility .
geometrically , the flexibility of each individual load is a polytope , and their aggregation is the minkowski sum of these polytopes .
our method originates from an alternative interpretation of the minkowski sum as projection .
the aggregate flexibility can be viewed as the projection of a high - dimensional polytope onto the subspace representing the aggregate power .
we formulate a robust optimization problem to optimally approximate the polytopic projection with respect to the homothet of a given polytope . to enable efficient and parallel computation of the aggregate flexibility for a large number of loads , a muti - stage aggregation strategy is proposed .
the scheduling policy for individual loads is also derived .
finally , an energy arbitrage problem is solved to demonstrate the effectiveness of the proposed method . |
quadratic gravitational lagrangians were proposed shortly after the formulation of general relativity ( gr ) as alternatives to einstein s theory .
gravity modifications in the form of higher - order curvature invariants in the lagrangian are generally known as higher - order gravity ( hog ) theories .
they arise in string - theoretic considerations , e.g. , brane models with gauss - bonett terms @xcite or models with a scalar field coupled to the gauss - bonett invariant @xcite ( see @xcite for a review ) and generally involve linear combinations of all possible second order invariants that can be formed from the riemann , ricci and scalar curvatures . a quarter of a century ago there was a resurgence of interest in such theories in an effort to explain inflation .
the reasons for considering hog theories were multiple .
firstly , it was hoped that higher order lagrangians would create a first approximation to quantum gravity , due to their better renormalisation properties than gr @xcite .
secondly , it was reasonable to expect that on approach to a spacetime singularity , curvature invariants of all orders ought to play an important dynamical role .
far from the singularity , when higher order corrections become negligible , one should recover gr .
furthermore , it was hoped that these generalized theories of gravity might exhibit better behavior near singularities .
thirdly , inflation emerges in these theories in a most direct way . in one of the first inflationary models , proposed in 1980 by starobinsky @xcite , inflation is due to the @xmath0 correction term in a gravitational lagrangian @xmath1 where @xmath2 is a constant .
recently there is a revival of interest in hog theories in an effort to explain the accelerating expansion of the universe @xcite .
the general idea is to add an @xmath3 term to the einstein - hilbert lagrangian or more generally to consider @xmath4 lagrangians @xcite .
as the universe expands , one expects that the inverse curvature terms will dominate and produce the late time accelerating expansion .
most studies are restricted to simple friedmann - robertson - walker ( frw ) models because of the complexity of the field equations . at present , the observational viability of these models is a subject of active research ( see @xcite and references therein ) .
however , it seems that a large class of lagrangians may fit the observational data , but simple models based on the frw metric are insufficient to pick the correct lagrangian ( see for example @xcite for the reconstruction of the @xmath5 theory which best reproduces the observed cosmological data ) . for more general spacetimes it may even be meaningless to say that @xmath6 is small in some epoch of cosmic evolution and large in some other one ( for a thorough critic see @xcite ) . in this paper
we investigate the late time evolution of flat and positively curved frw models with a perfect fluid in the @xmath7 theory .
this is the simplest generalization of the einstein - hilbert lagrangian and the addition of the quadratic term represents a correction to general relativity .
the simple vacuum case was studied in @xcite , where oscillatory behaviour of the solutions of closed models was found . since hog theories in vacuum are conformally equivalent to gr with a scalar field , it is tempting to say that the @xmath0 contribution has predictable cosmological consequences @xcite .
however , this is an oversimplification of the picture ( see @xcite for specific examples ) .
the two frames are mathematically equivalent , but physically they provide different theories . in the jordan frame , gravity is described entirely by the metric @xmath8 in the einstein frame , the scalar field exhibits a non - metric aspect of the gravitational interaction , reflecting the additional degree of freedom due to the higher order of the field equations in the jordan frame .
inclusion of additional matter fields , further complicates the situation and while the field equations in the einstein frame are formally the einstein equations , nevertheless this theory is not physically equivalent to gr .
there is no universally acceptable answer to the issue which conformal frame is physical @xcite ( see @xcite for a thorough analysis of different views ) .
the plan of the paper is as follows .
next section contains a short comment on the stability of well - known power - law solutions .
the field equations are written as a constrained four - dimensional polynomial dynamical system .
section 3 contains the analysis of the flat case .
the so - called normal form of the dynamical system greatly simplifies the problem , since two of the equations decouple . in section 4
we study the qualitative behaviour of the solutions near the equilibrium points of positively curved models and analyse their late time evolution .
it is shown that an initially expanding closed frw universe may exhibit oscillatory behaviour .
the general gravitational lagrangian in four - dimensional spacetimes contains curvature invariants of all orders , @xmath9 .
the term @xmath10 is omitted because of the gauss - bonnet theorem .
a further simplification can be done in homogeneous and isotropic spacetimes where the variation of @xmath11 with respect to the metric is proportional to the variation of @xmath0 @xcite .
we conclude that for isotropic cosmologies the gravitational lagrangian contains only powers of the scalar curvature and we may consider hog theories derived from lagrangians of the form @xmath12 where @xmath13 is an arbitrary smooth function .
it is well - known that the corresponding field equations are fourth - order and take the form @xmath14 where @xmath15 and a prime ( @xmath16 ) denotes differentiation with respect to @xmath17 the generalised bianchi identities imply that @xmath18 contraction of ( [ hoge ] ) yields the trace equation @xmath19 in contrast to gr where the relation of @xmath6 to @xmath20 is algebraic , in hog theories the trace equation ( [ trace ] ) is a differential equation for @xmath21 with @xmath20 as source term @xcite .
this suggests that in hog theories both the metric and the scalar curvature are dynamical fields .
in the following , we consider a quadratic lagrangian without cosmological constant , i.e. @xmath22 and confine our attention to cosmologies with a perfect fluid with energy density @xmath23 and pressure @xmath24 of the form @xmath25 for homogeneous and isotropic spacetimes is the scale factor , an overdot denotes differentiation with respect to time @xmath26 and units have been chosen so that @xmath27 described by the standard frw metric we need the following useful relations ( @xmath28 and @xmath29 ) @xmath30 the @xmath31 component of ( [ hoge ] ) is @xmath32 = \frac{1}{3}\rho , \label{0 - 0}%\ ] ] where @xmath33 is the hubble function . at this point
we make a digression . for flat , @xmath34 models , differentiating the relation @xmath35 ( which comes from the third of ( [ isot ] ) ) with respect to @xmath36 ,
equation ( [ 0 - 0 ] ) takes the form@xmath37 for radiation , @xmath38 , there exists a special solution @xmath39 as @xmath40 as barrow and middleton point out @xcite , this is also an exact vacuum solution of the purely quadratic theory , in the sense that it solves @xmath41 however , this solution is unstable as we shall see in a moment ( see @xcite for a detailed stability analysis of isotropic models in general @xmath42 theories ) . following carroll et al @xcite ,
we reduce the order of ( [ 00flat ] ) in vacuum by defining@xmath43 then , the asymptotic values of the function @xmath44 as @xmath45 correspond to the exponents @xmath46 for power - law solutions @xmath47 .
we apply this technique to ( [ 00flat ] ) and we find the first - order equation@xmath48 with the corresponding direction field shown in fig .
we see that the solution @xmath49 is a past attractor ( @xmath50 ) , but becomes unstable as @xmath45 in agreement with @xcite , and that @xmath51 corresponding to the singularity @xmath52 .
this is also evident by studying the asymptotic behaviour of solutions of the linearised equation near the constant solutions @xmath53 and @xmath54 ) ] we now continue our discussion about the choice of variables . setting @xmath55
the @xmath31 equation becomes @xmath56 = \frac{1}{3}\rho , \label{00}%\ ] ] and the evolution equation for @xmath57 is @xmath58 the evolution equation for @xmath59 comes from the third of ( [ isot ] ) and takes the form@xmath60 the conservation equation@xmath61 is a consequence of the bianchi identities . with the relation @xmath62 equation ( [ trace ] ) becomes @xmath63 this equation is usually considered as superfluous , since it follows from the differentiation of the equation ( [ 00 ] ) with respect to @xmath36 ; for example equation ( [ trac ] ) was used in @xcite as a control of the accuracy of numerical investigations of bianchi type i and ix models .
thus , one can chose @xmath64and @xmath23 as dynamical variables which obey the evolution equations ( [ 00 ] ) , ( [ chi ] ) , ( [ hubb ] ) and ( [ cons ] ) and constitute a four - dimensional dynamical system .
the form of equation ( [ 00 ] ) suggest the choice of expansion normalized variables of the type @xmath65 detailed studies using this approach for general @xmath42 theories can be found in @xcite .
however , even in the simplest case of flat models in vacuum , numerical investigation of the system with initial values @xmath66 shows that @xmath67 exhibit damped oscillations , with almost zero minima .
therefore although permissible , the transformation ( [ tran ] ) may induce fake singularities to the solutions .
moreover , as mentioned in @xcite , a drawback of this choice of variables is that it does not give a complete description of the evolution for bouncing or recollapsing models .
if the hubble parameter passes through zero the logarithmic time coordinate is ill - defined and the transformation ( [ tran ] ) is singular ( see @xcite for a comparison of compact and non - compact variables ) . in order to circumvent these difficulties , we introduce one more degree of freedom by using ( [ trac ] ) , thus augmenting the dimension of the dynamical system .
the state @xmath68 of the system lies on the hypersurface of @xmath69 defined by the constraint ( [ 00 ] ) .
the presence of @xmath70 in the state vector reflects the fact that there are additional degrees of freedom in hog theories than in gr ( cf . the remark after equation ( [ trace ] ) ) .
we define @xmath71 @xmath72 and use ( [ 00 ] ) to eliminate @xmath73 so that equations ( [ trac ] ) , ( [ chi ] ) and ( [ hubb ] ) constitute a four - dimensional system .
the parameter @xmath2 may be used to define dimensionless variables by rescaling@xmath74 and our system becomes@xmath75 with @xmath76 $ ] .
* remark . *
the system ( [ full ] ) is not an arbitrary free four - dimensional system . in view of ( [ 00 ] ) the initial conditions have to satisfy the condition @xmath77 with a little manipulation of the equations ( [ full ] ) it can be shown that , once we start with initial conditions satisfying ( [ ineq ] ) at time @xmath78 , the solutions of the system satisfy this inequality for all @xmath79 thus the field equations share the general property of the einstein equations , namely that the subsequent evolution of the system is such that the solutions respect the constraint .
in the flat , ( @xmath80 ) , case the dimension of the dynamical system is reduced by one , since the evolution equation for @xmath57 decouples from the remaining equations .
the corresponding system is@xmath81 i.e. , the vector field does not depend on the dynamical variable @xmath82 vacuum models are significantly simpler to analyse , but we do not study them separately as they arise formally by setting @xmath38 in all the equations , while ( [ ineq ] ) describing the phase space becomes equality . the only equilibrium point of ( [ flat ] ) is the origin and corresponds to flat empty spacetime .
the eigenvalues of the jacobian matrix at the origin are @xmath83 and therefore we can not infer about stability using the linearisation theorem . for nonhyperbolic equilibrium points there exist no general methods for studying their stability
. the normal form of the system may provide some information about the behaviour of the solutions near the equilibrium .
the normal form theory consists in a nonlinear coordinate transformation that allows to simplify the nonlinear part of the system ( cf .
@xcite for a brief introduction ) . this task will be accomplished in three steps in some detail for the convenience of readers with no previous knowledge of the method .
let @xmath84 be the matrix formed from the eigenvectors which transforms the linear part of the vector field into jordan canonical form .
we write ( [ flat ] ) in vector notation ( with @xmath85 ) as @xmath86 where @xmath87 is the linear part of the vector field and @xmath88 .
\2 . using the matrix @xmath84
, we define new variables , @xmath89 , by the equations@xmath90 or in vector notation @xmath91 so that ( [ sys3a ] ) becomes @xmath92 denoting the canonical form of @xmath87 by @xmath93 we finally obtain the system @xmath94 where @xmath95 in components system ( [ sys3b ] ) is@xmath96{c}% \dot{y}_{1}\\ \dot{y}_{2}\\ \dot{y}% \end{array } \right]=\left [ \begin{array } [ c]{ccc}% 0 & -1 & 0\\ 1 & 0 & 0\\ 0 & 0 & 0 \end{array } \right ] \left [ \begin{array } [ c]{c}% y_{1}\\ y_{2}\\ y \end{array } \right]+ \\ & & \left [ \begin{array } [ c]{c}% 0\\ \left ( 4 - 3\gamma\right ) y_{1}^{2}-\left ( 3\gamma+2\right ) y_{2}^{2}% -\frac{4 - 3\gamma}{4}y^{2}-3yy_{2}-\left ( 4 - 3\gamma\right ) y_{1}\left ( 2y_{2}+y\right ) ^{2}\\ -2\left ( 4 - 3\gamma\right ) y_{1}^{2}+2\left ( 3\gamma-2\right ) y_{2}% ^{2}-\frac{3\gamma}{2}y^{2}-2yy_{2}+2\left ( 4 - 3\gamma\right ) y_{1}\left ( 2y_{2}+y\right ) ^{2}% \end{array } \right ] .\end{aligned}\ ] ] inequality ( [ ineq ] ) imposes the constraint @xmath97 3 . under the non - linear change of variables@xmath98 and keeping only terms up to second order , the system transforms to @xmath99 finally , defining cylindrical coordinates @xmath100 we obtain@xmath101 we may continue to simplify the third order terms and the result should be @xmath102 this is the normal form in cylindrical coordinates of every three - dimensional vector field with linear part @xmath103{ccc}% 0 & -1 & 0\\ 1 & 0 & 0\\ 0 & 0 & 0 \end{array } \right ] \left [ \begin{array } [ c]{c}% y_{1}\\ y_{2}\\ y \end{array } \right ] , \ ] ] ( see @xcite p. 377 ) . however , since we are interested on the behaviour of the solutions only near the origin , we truncate the vector field at @xmath104 we note that the @xmath105 dependence of the vector field has been eliminated , so that we can study the system in the @xmath106 space .
the second equation of ( [ cylifl ] ) implies that the trajectory in the @xmath107 plane spirals with angular velocity @xmath108 .
the projection of ( [ cylifl ] ) on the @xmath109 plane is @xmath110 this system belongs to a family of systems studied in 1974 by takens @xcite ( see also @xcite for a description of all phase portraits for different values of the parameters )
. system ( [ 2dfl ] ) is invariant under the transformation @xmath111 @xmath112 ( which implies that all trajectories are symmetric with respect to the @xmath113 axis ) and the line @xmath114 is invariant .
note also that the system ( [ 2dfl ] ) has invariant lines @xmath115 ( compare with the similar system in @xcite ) .
the behaviour of the solutions depends on the parameter @xmath116 and as we shall see , @xmath117 is a bifurcation value . ) for @xmath118 and @xmath119 .
the invariant lines @xmath115 are shown with heavy lines and @xmath120 by the dotted line . ]
case i , @xmath121 we observe that @xmath122 is always decreasing along the orbits while @xmath113 is decreasing in the first quadrant .
since no trajectory can cross the line @xmath123 all trajectories starting above this line , approach the origin asymptotically .
the phase space of the dynamical system ( [ 2dfl ] ) is not the whole @xmath109 plane , because of the constraint ( [ ineq1 ] ) . in terms of the variables ( [ nonl ] ) and neglecting fourth - order terms the constraint becomes@xmath124
since we are interested on trajectories starting close to the origin , for initial values of @xmath125 satisfying @xmath126 the first of ( [ 2dfl ] ) guarantees that @xmath127 remains less than @xmath128 , so that ( [ ineq2 ] ) implies @xmath129 therefore , we should consider only trajectories starting above the line @xmath130 and according to the previous discussion all these trajectories asymptotically approach the origin .
case ii , @xmath131 in the first quadrant @xmath113 is decreasing along the orbits and , that @xmath132 vanishes along the line @xmath133 once a trajectory crosses the line @xmath134 it is trapped between the lines @xmath135 and @xmath123 and since @xmath136 it approaches the origin asymptotically .
case iii , @xmath137 it is evident that all trajectories are straight lines approaching asymptotically the origin .
we conclude that the late time behaviour of flat models is similar to the future predicted by gr .
more precisely , all initially expanding flat models close to the state @xmath138 asymptotically approach the flat empty spacetime .
in this section we consider an initially expanding closed universe described by the full four - dimensional system ( [ full ] ) with @xmath139 .
there are two equilibria : @xmath140 this corresponds to the limiting state of an almost empty , slowly varying universe with @xmath141 while the scale factor goes to infinity .
the point @xmath142 which resembles to the minkowski solution , is located at the boundary of the phase space .
@xmath143 with @xmath144 it is a static solution and the eigenvalues of the jacobian matrix at @xmath145 are@xmath146 the real parts of the eigenvalues are nonzero for almost all ] and nonzero in the rest of the interval @xmath147 permissible values of @xmath116 , and we conclude that the local stable and unstable manifolds through @xmath145 are both two - dimensional .
the point @xmath145 corresponds to the einstein static universe , where the effective cosmological constant is provided by the curvature equilibrium @xmath148 to see this , it is sufficient to write equation ( [ 00 ] ) in the original variables at the equilibrium point as @xmath149 with @xmath150 the cosmological constant depends on both the parameters @xmath2 and @xmath151 for example , for @xmath152 @xmath153 static solutions have little interest as cosmological models and we turn our attention to the other equilibrium , @xmath142 .
the point @xmath142 is a nonhyperbolic equilibrium and we find again the normal form of the system , which is given by ( [ normalf ] ) in the appendix .
defining cylindrical coordinates @xmath154 we obtain@xmath155 we truncate the vector field at @xmath156 and we note again that the @xmath105 dependence of the vector field has been eliminated , so that we can study the system in the @xmath157 space .
we write the first and third of ( [ cyli ] ) as a differential equation@xmath158 which has the general solution@xmath159 substitution of ( [ r - equ ] ) into the fourth equation of ( [ cyli ] ) yields the projection of the fourth - dimensional system on the @xmath160 plane , namely@xmath161 some general properties of the solutions of ( [ 2dim ] ) follow by inspection . by standard arguments all trajectories are symmetric with respect to the @xmath57 axis .
note that the new @xmath57 defined by ( [ nonl1 ] ) remains non - negative for initial values @xmath162 sufficiently small and , since the line @xmath163 is invariant , any trajectory starting at the half plane @xmath164 remains there for all @xmath165 system ( [ 2dim ] ) has two equilibrium points , the origin @xmath166 and @xmath167 , where @xmath168 and therefore @xmath169 for @xmath170 or @xmath131 the origin is again a nonhyperbolic equilibrium point .
computation of the jacobian matrix at the equilibrium @xmath171 shows that for the linearised system , this point is a saddle for @xmath172 and a center for @xmath173 .
for @xmath174 it is easy to see that @xmath57 is decreasing in the first quadrant and @xmath122 is decreasing along the orbits in the strip @xmath175 . on
any orbit starting in the first quadrant with @xmath176 @xmath122 becomes zero at some time and the trajectory crosses vertically the @xmath177axis . once the trajectory enters the second quadrant , @xmath57 increases and @xmath122 decreases . for @xmath178 ,
all trajectories starting in the first quadrant follow the same pattern . system ( [ 2dim ] ) has a first integral , _
viz_.@xmath179 this can be seen by writing ( [ 2dim ] ) as @xmath180 setting @xmath181 we obtain a linear differential equation for @xmath182 which is easily integrable . the level curves of @xmath183 are the trajectories of the system
. we shall show for the system ( [ 2dim ] ) that : _
( i ) for every _
@xmath184 $ ] _ there are no solutions asymptotically approaching the origin ( ii ) for _ @xmath185 $ ] _ there are no periodic solutions and ( iii ) for _ @xmath186 $ ] _ there exist periodic solutions and the basin of every periodic trajectory is the set _ @xmath187
the proof mimics that found in @xcite for the simple case @xmath188 let @xmath189 .
we have@xmath190 which implies that the function @xmath191 must be non - negative .
we consider two cases . \1 . @xmath192
. then @xmath13 is strictly increasing for @xmath193 and @xmath194 , thus @xmath13 has a unique root @xmath195 depending on @xmath196 .
it follows that for @xmath192 any orbit starting in the first quadrant satisfies@xmath197 i.e. , there are no solutions approaching the axis @xmath163 .
these solutions are not closed since they intersect the @xmath177axis only once at @xmath198 .
if @xmath178 , then @xmath13 has again a unique root hence the trajectories follow the same pattern as in case 1 . if @xmath119 , then @xmath13 has two zeros , say @xmath200 .
this means that @xmath201 , i.e. , @xmath57 is bounded , and by ( [ xy ] ) , so is @xmath122 .
thus , an orbit of ( [ 2dim ] ) starting in the first quadrant crosses the @xmath177axis at @xmath202 and re - enters in the first quadrant crossing the @xmath177axis at @xmath203 i.e. it is a closed curve and represents a periodic solution .
the curve corresponding to @xmath204 separates the phase space into two disjoint regions i and ii . in region
ii , ( @xmath199 ) , every trajectory corresponds to a periodic solution and we conclude that the basin of every periodic trajectory is the set @xmath205 * remark . * the mere existence of closed orbits around the equilibrium point @xmath206 could be inferred from the following theorem : if an equilibrium point @xmath207 is a center for the linearised system and all trajectories are symmetric with respect to the @xmath57 axis , then @xmath207 is also a center for the nonlinear system ( [ 2dim ] ) ( cf .
@xcite theorem 6 , page 141 ) . in the above proof
we also determine the subset of the phase space which contains all periodic orbits . using all this information we may sketch the phase portrait of the system ( fig .
3 ) . ) for @xmath208 , @xmath209 and @xmath131 ] for @xmath170 the homoclinic curves to the origin as well as the curves approaching the origin indicate that an initially expanding closed universe may avoid recollapse .
this result is also valid in gr when matter fields violate the strong energy condition @xcite . for @xmath185
$ ] every solution curve becomes unbounded and we may interpret this as an indication that this universe recollapses .
the range @xmath211 $ ] for @xmath116 is the more interesting because of the periodic orbits in region ii .
the phase portrait in fig .
3 may lead to the conclusion that the periodic orbits are far from the origin .
however , the position of the cycles in the phase space depends on the constants @xmath212 in ( [ 2dim ] ) and @xmath196 in ( [ xy ] ) and therefore , for suitable values of @xmath212 and @xmath213 there exist periodic orbits arbitrary close to the origin .
note that the periodic solutions of ( [ 2dim ] ) induce periodicity to the full four - dimensional system ( [ cyli ] ) or ( [ normalf ] ) .
in fact , if @xmath214 and @xmath215 are periodic solutions , then ( [ r - equ ] ) implies that the solutions @xmath216 oscillate in the @xmath107 plane with a periodic amplitude @xmath217 note also that the periodic motion in the @xmath160 plane is independent from the rotation of @xmath113 in the @xmath107 plane . obviously we can not assign a physical meaning to the new variables @xmath218 since the transformations ( [ line ] ) and ( [ nonl1 ] ) have mixed the original variables of ( [ full ] ) in a nontrivial way .
however , the periodic character of the solutions of ( [ normalf ] ) whatever the physical meaning of the variables be , has the following interpretation .
_ close to the equilibrium of the original system ( [ full ] ) , there exist periodic solutions_. this result is in agreement with @xcite where it is shown that for lagrangians @xmath219 bounces of closed models are allowed for every integer value of @xmath220 .
* remark*. as mentioned in section 2 numerical experiments show that the solutions of the system have oscillatory behavior .
this property is intuitively evident by looking at the harmonic oscillator , equation ( [ trac ] ) . using the rescalings of the dynamical variables along with the scaling @xmath221 equation ( [ trac ] )
becomes @xmath222 which is the equation of a forced , damped harmonic oscillator with unit angular frequency .
qualitative arguments supported by numerical solutions indicate that oscillatory motion , possibly slightly damped , is essentially independent of @xmath223 and @xmath116 and becomes the late time behaviour .
this is revealed in the normal forms of the systems , ( [ cylifl ] ) and ( [ cyli ] ) , where the @xmath105 equation shows oscillatory motion with unit angular frequency . we emphasize again that this motion is independent from the periodic motion in the @xmath160 plane .
the normal form analysis reveals both kinds of oscillatory behaviour and distinguishes the models which actually exhibit undamped oscillations at late time .
we analysed the qualitative behaviour of flat and positively curved frw models filled with ordinary matter described by a perfect fluid in the jordan frame of the @xmath7 theory .
we have shown that initially expanding flat models close to the equilibrium @xmath224 are ever expanding and asymptotically approach flat empty spacetime .
therefore , the late time behaviour of flat models is a common property of quadratic gravity and gr .
closed models can avoid recollapse for @xmath170 , but not for @xmath116 in the range @xmath225 , $ ] thereby also behaving similarly to the corresponding general - relativistic cosmologies .
the interesting feature is the existence of periodic solutions near the origin for @xmath226 .
this is not revealed in the einstein frame ( see miri2 ) , possibly because in that investigation , the scalar field related to the conformal transformation is not coupled to matter , i.e. , the matter lagrangian is added after performing the conformal transformation .
were the two frames physically equivalent then , a naive physical explanation of the cyclic behavior could rely on the scalar field which behaves like a cosmological constant in the high curvature limit .
the perfect fluid which is also present dominates in the low curvature regime allowing for a recollapse , but then the effective cosmological constant induces a bounce in the high - curvature regime .
the periodic solutions imply that an initially expanding closed universe can avoid recollapse through an infinite sequence of successive expansions and contractions .
the oscillatory open model proposed by steinhardt and turok @xcite has renewed interest in cyclic universes .
however , observations do not exclude @xmath227 to be slightly larger than one @xcite .
oscillatory closed models were considered in the context of loop quantum cosmology @xcite and oscillatory ( but not periodic ) solutions in gr were found in @xcite for closed models containing radiation and dust or scalar field . in fig .
3 the basin of all periodic trajectories of the ( [ 2dim ] ) is the domain on the right of the @xmath204 curve and since it is an open subset of the phase space , we conclude that there is enough room in the set of initial data of ( [ full ] ) which lead to an oscillating scale factor .
the @xmath7 theory has offered a successful inflationary model , but whether it is capable to explain the acceleration of the universe is an open question that needs to be studied in more detail . in particular
, the observed slow acceleration must be related to the periods of the closed curves of ( [ 2dim ] ) .
the normal form theory is a powerful method for determining the qualitative behaviour of a dynamical system near a nonhyperbolic equilibrium point , but does not give any information about the structure of the solutions far from this equilibrium .
our results are based on an analysis of the behaviour of the dynamical system ( [ full ] ) only near the equilibrium solutions .
the geometry of the trajectories of a four - dimensional dynamical system may be quite complicated , e.g. strange attractors may be present . for the system ( [ full ] ) the whole picture may come in view only from the investigation of the global structure of the solutions .
the study of this question is an interesting challenge for mathematical relativity .
i thank spiros cotsakis and alan rendall for useful comments .
i am grateful to the referee for comments from the physical perspective .
following the usual algorithm , we use the matrix which transforms the linear part of ( [ full ] ) into jordan canonical form and define new variables by@xmath228 so that ( [ full ] ) becomes@xmath96{c}\dot{y}_{1}\\ \dot{y}_{2}\\ \dot{x}\\ \dot{y}\end{array } \right]=\left [ \begin{array } [ c]{cccc}0 & -1 & 0 & 0\\ 1 & 0 & 0 & 0\\ 0 & 0 & 0 & 0\\ 0 & 0 & 0 & 0 \end{array } \right ] \left [ \begin{array } [ c]{c}y_{1}\\ y_{2}\\ x\\ y \end{array } \right]+ \\ & & \left [ \begin{array } [ c]{c}0\\ \left ( 4 - 3\gamma\right ) y_{1}^{2}-\left ( 3\gamma+2\right ) y_{2}^{2}-\frac{4 - 3\gamma}{4}\left ( kx^{2}+y^{2}\right ) -3yy_{2}-\left ( 4 - 3\gamma\right ) z_{3}\\ -2xy_{2}-xy\\ -2\left ( 4 - 3\gamma\right ) y_{1}^{2}+2\left ( 3\gamma-2\right ) y_{2}^{2}-\frac{3\gamma-2}{2}kx^{2}-\frac{3\gamma}{2}y^{2}-2yy_{2}+2\left ( 4 - 3\gamma\right ) z_{3}\end{array } \right ] , \end{aligned}\ ] ] with @xmath229 .$ ] under the non - linear change of variables@xmath230 and keeping only terms up to second order , the system transforms to @xmath231 brans , c.h . :
quantum grav .
* 5 * l197 ( 1998 ) ; ferraris , m. , francaviglia , m. , magnano , g. : class .
quantum grav .
* 7 * 261 ( 1990 ) ; cotsakis , s. : phys
. rev . * * * * d * 47 * 1437 ( 1993 ) ; erratum phys . rev . * * * * d * 49 * 1145 ( 1994 ) | homogeneous and isotropic models are studied in the jordan frame of the second order gravity theory .
the late time evolution of the models is analysed with the methods of the dynamical systems .
the normal form of the dynamical system has periodic solutions for a large set of initial conditions .
this implies that an initially expanding closed isotropic universe may exhibit oscillatory behaviour . |
as we know , the universe is expanding and this expanding is undergoing an accelerating phase .
this issue discovered in type ia supernova @xcite , associated with large scale structure @xcite and cosmic microwave background @xcite .
it is noteworthy that the accelerated expansion of the universe is created from a mysterious energy called dark energy , which is about two thirds the total energy of universe . by using the einstein field equation , the accelerated expansion described by a small positive cosmological constant .
also this discovery demonstrates that geometry of the universe is very close to flat space @xcite .
various candidates have been introduced to describe dark energy scenario , so that the universe dominates with a perfect fluid by a negative pressure and the eos parameter which is less than @xmath4 , the so - called phantom phase .
we can introduce some of these models such as the cosmological constant @xcite , the scalar fields ( including quintessence , @xmath5-essence , phantom , tachyon , etc ) @xcite , the vector field @xcite , holographic @xcite , interacting model @xcite and braneworld models @xcite . moreover , there are some other methods to describe the universe accelerated expansion . one of these theories is modified gravity theories that from the physical point of view , this prevents the complexities of the previous models especially the complicated computation of numerical solutions .
another benefit of the modified gravity theories is consistent with recent observations for late accelerating universe and dark energy .
some of these modified gravity theories can directly be achieved by replacing the ricci scalar @xmath6 by @xmath7 , @xmath8 and @xmath9 with an arbitrary function in the einstein - hilbert action .
these theories are introduced as @xmath10 gravity @xcite , @xmath8 gravity @xcite and @xmath9 gravity @xcite which respectively an arbitrary function of the ricci scalar @xmath6 , the torsion scalar @xmath11 and the gauss - bonnet term @xmath12 .
therefore , we can tell that modified gravitational theories are a generalization of general relativity .
we also note that @xmath8 gravity is a generalized version of teleparallel gravity originally proposed by einstein @xcite , he had tried to redefine the unification of gravity and electromagnetism on the mathematical structure of distant parallelism by introducing of a tetrad or vierbein field , the so - called teleparallelism .
thus , instead of using the levi - civita connection in the framework of general relativity , we use the weitzenbck connection in teleparallelism . in that case , the four - dimensional space - time manifold must be a parallelizable manifold @xcite . in this paper , we will explain the late time accelerated expansion of the universe with the unification of @xmath7 and @xmath13 gravity theories as @xmath0 gravity which one is an arbitrary function of curvature scalar @xmath6 and torsion scalar @xmath11 @xcite .
the existence of both scalars @xmath6 and @xmath11 into @xmath0 gravity is effective on only geometry of universe no on matter source .
this means that universe is included the effects of curvature and torsion .
as we discussed , the acceleration can be consequence of the dark energy influence that this leads to some other models called chaplygin gas @xcite .
chaplygin gas is a fluid with negative pressure that begins to dominate the matter content and , at the end , the process of structure formation is driven by cold dark matter without affecting the previous history of the universe .
this kind of chaplygin gas cosmology has an interesting connection to string theory via the nambu - goto action for a d - brane moving in a @xmath14-dimensional space - time , feature than can be regarded to the tachyonic panorama @xcite .
the main objective of this paper is that we will consider the @xmath0 gravity model as a source of dark energy .
therefore , we intend to investigate the interacting model between @xmath0 gravity and modified chaplygin gas with this motivation that we can describe the accelerated expansion of the universe . the paper is organized as follows : in sec . [ s2 ] , we review @xmath0 gravity model and obtain the friedmann equations by using the corresponding connections . in sec .
[ s3 ] , we introduce the basic setup of the modified chaplygin gas , and then will interact @xmath0 gravity with modified chaplygin gas .
thereinafter , total energy density and total pressure of universe will be written in the form of a combination of modified chaplygin gas and dark energy . in sec .
[ s4 ] , we reconstruct the current model with a source of scalar field by quintessence model , and then the cosmological parameters will be written in terms of redshift and @xmath2-folding number .
also we investigate stability of the model and present corresponding graphs in terms of redshift and @xmath2-folding number . finally , a short summary is given in sec .
the action of @xmath0 gravity theory coupled with matter is given by @xcite @xmath15 where @xmath16 , in which @xmath17 is the determinant of the metric tensor @xmath18 , and @xmath19 is the matter lagrangian .
the @xmath0 is an arbitrary function of curvature scalar @xmath6 and torsion scalar @xmath11 .
we note that curvature scalar represents gravity in general relativity , and torsion scalar represents gravity in teleparallel gravity by a different mathematical notations as levi - civita connection and wienzbck connection , respectively . despite this difference ,
independently both theories have similar result for equivalent descriptions of gravitation .
therefore , we consider a vierbein field @xmath20 with index @xmath21 running from @xmath22 to @xmath23 , which one is an orthonormal basis for the tangent space at each point @xmath24 of the manifold in wienzbck connection .
then , we can relate vierbein field to the metric as @xmath25 in which the minkowski metric @xmath26 .
we consider homogeneous and isotropic universe in the flat flrw metric in the following form @xmath27 where @xmath28 is scale factor , and one is the function of cosmic time .
we can write the vierbein field in accordance with as follows @xmath29 now we intend to find the corresponding lagrangian of the action as @xmath30 in that case , we expand expression @xmath0 by maclaurin series , so that the lagrangian is simplicity yielded as @xmath31 where indices denote derivative with respect to @xmath6 and @xmath11 in corresponding locations , and the point denotes derivative with respect to cosmic time
. the curvature and torsion scalars are find by [ rt1 ] @xmath32 where @xmath33 is the hubble parameter , and , @xmath34 and @xmath35 be defined as the two arbitrary functions in terms of @xmath36 and @xmath37 , and , @xmath38 , @xmath39 and @xmath40 are the ricci , antisymmetric and torsion tensors respectively , in the form [ rst ] @xmath41 where the contortion tensor @xmath42 is @xmath43 by using the eq .
, we can obtain the friedmann equations in the following form [ rhop ] @xmath44 where @xmath45 and @xmath46 are total energy density and total pressure of an universe dominated with a perfect fluid , respectively .
the continuity equation of the model becomes @xmath47 in this paper , we consider a simple particular model of @xmath0 gravity as @xmath48 where @xmath49 and @xmath50 are constants . now for solving the model ,
we choose a power form for functions @xmath51 and @xmath52 as [ uv1 ] @xmath53 to substitute eqs . into eqs .
, the friedmann equations are rewritten by [ rho2 ] @xmath54
one of the recent cosmological models which is based on the use of exotic type of perfect fluid suggests that our universe filled with the chaplygin gas with the following equation of state @xmath55 where @xmath56 is a positive constant .
the equation of state generalized as @xmath57 which is known as generalized chaplygin gas .
therefore , modified chaplygin gas with the following equation of state introduced @xmath58 where @xmath59 is a positive constant , and @xmath60 and @xmath61 are the pressure and energy density in which @xmath62 .
this model is more appropriate choice to have constant negative pressure at low energy density and high pressure at high energy density .
the special case of @xmath63 is the best fitted value to describe evolution of the universe from radiation regime to the @xmath64 regime .
also , the modified chaplygin gas has been investigated in refs .
@xcite .
therefore , we consider the total energy density and pressure of universe as the combination of components of dark energy and chaplygin gas in the following form [ rhoptot ] @xmath65 now , we take an energy flow between the dark energy and modified chaplygin gas , so the energy flow is introduced as an interaction between two components . in this way , the continuity equation be written to two separate continuity equations in the following form [ rhopq ] @xmath66 where @xmath67 is the interaction term between the dark energy and modified chaplygin gas , that one be usually considered as : @xmath68 , @xmath69 and @xmath70 in which @xmath71 is the coupling parameter or transfer strength @xcite .
this choice is based on positive motivation @xmath67 , because from the observational data at the four years @xmath72 implies that the coupling parameter must be a small positive value @xcite . in this paper , we consider @xmath69 , and by substituting @xmath67 into eq .
we can obtain energy density of modified chaplygin gas as @xmath73^{\frac{1}{1+\gamma}},\ ] ] where @xmath74 , and then @xmath75^{\frac{1}{1+\gamma}}-\frac{b}{\left[\frac{b}{\eta}+c_0 a^{-3 \eta ( 1+\gamma)}\right]^{\frac{\gamma}{1+\gamma}}},\end{aligned}\ ] ] where @xmath76 is a constant of integration . the @xmath77 and @xmath78 obtain by eqs . and in the following form [ prhode ] @xmath79^{\frac{1}{1+\gamma}},\label{prhode-1}\\ & p_{de}=-(\mu+\nu)\left(2\dot{h}+3h^2\right)-\frac{1}{6 } \mu\,\alpha(n+3)\ , a^n-\frac{1}{6 } \nu\,\beta ( m+3)\ , a^m \label{prhode-2}\\ \nonumber & -a \left[\frac{b}{\eta}+c_0 a^{-3 \eta ( 1+\gamma)}\right]^{\frac{1}{1+\gamma}}+ \frac{b}{\left[\frac{b}{\eta}+c_0 a^{-3 \eta ( 1+\gamma)}\right]^{\frac{\gamma}{1+\gamma}}}.\end{aligned}\ ] ] the dark energy eos becomes @xmath80
in this section , we first will reconstruct the model as a scalar field and a potential entitled the quintessence model .
for this purpose , we suppose that the origin of interacted @xmath0 gravity with chaplygin gas is a real scalar field in which the corresponding friedmann equations is written in the following form @xcite [ rhopphi ] @xmath81 the continuity equation and the eos parameter are given by [ rhoeos ] @xmath82 now let s replace eqs . by transforming @xmath83 and @xmath84 into , so that we find kinetic energy term and potential as [ phidotpphi ]
@xmath85^{\frac{1}{1+\gamma}}\left(1+a-\frac{b}{\frac{b}{\eta}+c_0 a^{-3 \eta ( 1+\gamma)}}\right),\\ & v(\phi ) = \frac{1}{2}(\rho_{\phi } -p_{\phi } ) = ( \mu+\nu ) \left(\dot{h}+3 h^2\right)+\frac{1}{12 } \mu \alpha ( n+6 ) a^n + \frac{1}{12 } \nu \beta ( m+6 ) a^m \label{phidotpphi-2}\\ \nonumber & -\frac{1}{2}\left[\frac{b}{\eta}+c_0 a^{-3 \eta ( 1+\gamma)}\right]^{\frac{1}{1+\gamma}}\left(1-a+\frac{b}{\frac{b}{\eta}+c_0 a^{-3 \eta ( 1+\gamma)}}\right),\end{aligned}\ ] ] the right hand side of eqs .
are determined as a function of the scale factor . in that case , we take a particular form of universe evolution by power law for the scale factor in terms of cosmic time as follows : @xmath86 where @xmath87 is the value of the scale factor at the present universe , and @xmath88 is consistent the accelerated expansion of universe . to substitute the aforesaid scale factor into eqs . , we can obtain the following relationships [ phidotpphi1 ] @xmath89^{\frac{1}{1+\gamma}}\left(1+a-\frac{b}{\frac{b}{\eta}+c_0 ( a_0\ , t^s)^{-3 \eta ( 1+\gamma)}}\right),\\ & v(\phi ) = ( \mu+\nu ) \frac{s(3s-1)}{t^2}+\frac{1}{12 } \mu \,\alpha\ , ( n+6)\ , a_0^n\,t^{ns } + \frac{1}{12 } \nu\ , \beta\ , ( m+6)\ , a_0^m\,t^{ms } \label{phidotpphi1 - 2}\\ \nonumber & -\frac{1}{2}\left[\frac{b}{\eta}+c_0 ( a_0\ , t^s)^{-3 \eta ( 1+\gamma)}\right]^{\frac{1}{1+\gamma}}\left(1-a+\frac{b}{\frac{b}{\eta}+c_0 ( a_0\ , t^s)^{-3 \eta ( 1+\gamma)}}\right).\end{aligned}\ ] ] the eq .
is a reconstructed potential .
this means that universe is described by the quintessence model with the field @xmath90 and the potential @xmath91 instead of @xmath0 gravity .
here we realize that interacting @xmath0 gravity with modified chaplygin gas is a good alternative to quintessence model .
also this reconstruction demonstrates the reconstructed eos of scaler field in fig .
[ omegaphi2 t ] .
we note that choice of free parameters play a very important role in describing this model , so the motivation for these choices is based on crossing over phantom divide line . in terms of cosmic time for @xmath92 , @xmath93 , @xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 , @xmath99 , @xmath100 , @xmath101 , @xmath102 , @xmath103 and @xmath104 .
] in what follows , we turn our attention to study our model by parametrization . in the sense that the energy density and pressure of dark energy are written in terms of redshift , @xmath105 , in which @xmath87 is the value of the scale factor at the present universe , in other words , @xmath106 corresponds the today universe .
we introduce parameter , @xmath107 in which @xmath108 is the value of hubble parameter of current universe . in that case
, we find @xmath109 then , we can rewrite the eqs . and in terms of @xmath1 as [ prhode1 ] @xmath110^{\frac{1}{1+\gamma}},\\ & p_{de}=-h_0 ^ 2 ( \mu+\nu)\left(-(1+z ) r'+3 r\right)-\frac{1}{6 } \mu\,\alpha(n+3)\ , \left(\frac{a_0}{1+z}\right)^n-\frac{1}{6 } \nu\,\beta ( m+3)\ , \left(\frac{a_0}{1+z}\right)^m \label{prhode1 - 2}\\ \nonumber & -a \left[\frac{b}{\eta}+c_0 \left(\frac{a_0}{1+z}\right)^{-3 \eta ( 1+\gamma)}\right]^{\frac{1}{1+\gamma}}+ \frac{b}{\left[\frac{b}{\eta}+c_0 \left(\frac{a_0}{1+z}\right)^{-3 \eta ( 1+\gamma)}\right]^{\frac{\gamma}{1+\gamma}}},\\ & \omega_{de}=-1 + \label{prhode1 - 3}\\ \nonumber & \frac{h_0 ^ 2 ( \mu+\nu ) ( 1+z ) r'-\frac{1}{6 } \mu \alpha n \left(\frac{a_0}{1+z}\right)^n-\frac{1}{6 } \nu \beta m \left(\frac{a_0}{1+z}\right)^m-(1+a ) \left[\frac{b}{\eta}+c_0 \left(\frac{a_0}{1+z}\right)^{-3 \eta ( 1+\gamma)}\right]^{\frac{1}{1+\gamma}}+\frac{b}{\left[\frac{b}{\eta}+c_0 \left(\frac{a_0}{1+z}\right)^{-3 \eta ( 1+\gamma)}\right]^{\frac{\gamma}{1+\gamma}}}}{3 h_0 ^ 2 ( \mu+\nu ) r+\frac{1}{2 } \mu \alpha \left(\frac{a_0}{1+z}\right)^n+\frac{1}{2 } \nu \beta \left(\frac{a_0}{1+z}\right)^m-\left[\frac{b}{\eta}+c_0 \left(\frac{a_0}{1+z}\right)^{-3 \eta ( 1+\gamma)}\right]^{\frac{1}{1+\gamma}}},\end{aligned}\ ] ] where the prime denotes a derivative with respect to @xmath1 .
now , in order to describe the model as a origin of dark energy , we intend to reconstruct the eos of the model by parametrizing the hubble parameter in terms of the @xmath1 .
we note that reconstruction of dark energy models was fulfilled in refs .
@xcite for a minimally scalar field with a potential .
therefore , we take the following parametrization as @xmath111 where values of the corresponding coefficients are @xmath112 , @xmath113 , @xmath114 and @xmath115 , which these obtained the results of refs . @xcite . by inserting eq . into eqs .
, expressions @xmath77 , @xmath78 and @xmath116 will obtain in terms of redshift @xmath1 .
then , we plot the cosmological parameters in figs .
[ rhopde ] and [ omegade ] by the natural units @xmath117 .
we note that free parameters play the role of an important to plot the cosmological parameters , in which the motivation of the selections is based on positivity energy density and negativity pressure .
now for that free parameters correspond with observable quantities , we plot the eos of the model versus the @xmath2-folding number , n = ln(a ) as the time variable in fig .
[ omegade ] .
we can see the values of eos in three cases @xmath118 , @xmath119 ( late time ) and @xmath120 respectively with values @xmath22 , @xmath121 and @xmath122
. therefore , fig .
[ omegade ] shows us that variation of the eos from early time till future time is @xmath22 till @xmath122 , and this confirms the existence of an accelerated expansion of the universe .
the value of eos of late time confirms the result of @xcite in a flat universe . in terms of @xmath2-folding number @xmath123 for @xmath92 , @xmath124 , @xmath125 , @xmath126 , @xmath127 , @xmath128 , @xmath129 , @xmath130 , @xmath131 , @xmath132 , @xmath133 and @xmath134 .
] in order to have a more complete discussion , we intend to discuss the stability of the model . for this purpose
, we will use an useful function @xmath135 ( in which @xmath136 ) as a thermodynamic system , so this function represents sound speed in a perfect fluid .
therefore , the stability condition occurs when the function @xmath137 becomes bigger than zero .
it should be noted that a thermodynamic system can be described with adiabatic and non - adiabatic perturbations .
this means that the aforesaid function @xmath138 be related to an adiabatic sound speed . for making derivative the eqs .
and with respect to redshift @xmath1 , the speed sound function be plotted in terms of @xmath2-folding number @xmath123 in fig .
[ c2s2n ] .
we can see in the corresponding figure the values of @xmath137 in three cases @xmath118 , late time ( @xmath119 ) and @xmath120 respectively with values @xmath22 , @xmath139 and @xmath122 .
therefore , the fig .
[ c2s2n ] shows us that there is a stability in late time , because value @xmath137 is positive for case @xmath140 . in terms of @xmath2-folding number @xmath123 for @xmath92 , @xmath124 , @xmath125 , @xmath126 , @xmath127 , @xmath128 , @xmath129 , @xmath130 , @xmath131 , @xmath132 , @xmath133 and @xmath134 .
in this paper , we have studied @xmath0 gravity in terms of an arbitrary function of curvature and torsion scalars . for simplicity
, the function @xmath0 has been gotten as a linear function of the curvature and torsion scalars , and have computed in flrw metric .
the motivation of this research is based on existence of both cases curvature and torsion in the universe , while so far the research carried out on just implications existence of the curvature in the universe .
afterwards , we have considered the @xmath0 gravity model by interacting with modified chaplygin gas .
we wrote the corresponding action as the combination of @xmath0 gravity and a matter lagrangian . in order to obtain the friedmann equations
, we used a different mathematical notations for curvature scalar and torsion scalar by levi - civita connection and wienzbck connection , respectively . in
what follows , we considered total energy density and total pressure of universe as dominated with a perfect fluid , in which dark energy contribution is taken the combination of @xmath0 gravity and modified chaplygin gas as eq . .
then , by dividing these two functions we obtained the eos of dark energy for the scenario . subsequently , we reconstructed the model by an origin of a scalar field entitled quintessence model with a field potential . with correspondence between our model and quintessence model
, we plotted the eos of scalar field in terms of cosmic time by taking the scale factor as power law .
the graph of corresponding eos described the accelerated expanding of universe .
next , we used a parametrization , and the cosmological parameters have been written in terms of redshift @xmath1 . in that case , we drew them with respect to @xmath1 and @xmath2-folding number @xmath123 in figs .
[ rhopde ] and [ omegade ] .
these figures showed us the existence of an accelerating universe , especially graph of eos showed us that its value in late time is @xmath121 which one is consistent with ref .
we noted that the free parameters play an important role in plotting figures , the motivation of these selections are based on positivity energy density and negativity pressure .
finally , in order to investigate the stability , we tried to compute speed sound in this scenario in terms of @xmath2-folding number , so we plotted the @xmath141 with respect to @xmath2-folding number and have obtained its corresponding values in three status @xmath118 , late time ( @xmath119 ) and @xmath120 .
therefore , the fig .
[ c2s2n ] showed us that there is the stability in late time .
abazajian , k. , adelman - mccarthy , j. k. , ageros , m. a. , allam , s. s. , anderson , k. s. , anderson , s. f. , ... and hogan , c. j. ( 2004 ) .
the second data release of the sloan digital sky survey .
the astronomical journal , 128(1 ) , 502 .
pope , a. c. , matsubara , t. , szalay , a. s. , blanton , m. r. , eisenstein , d. j. , gray , j. , ... and sdss collaboration .
cosmological parameters from eigenmode analysis of sloan digital sky survey galaxy redshifts .
the astrophysical journal , 607(2 ) , 655 .
spergel , d. n. , verde , l. , peiris , h. v. , komatsu , e. , nolta , m. r. , bennett , c. l. , ... and wright , e. l. ( 2003 ) .
first - year wilkinson microwave anisotropy probe ( wmap ) observations : determination of cosmological parameters . the astrophysical journal supplement series , 148(1 ) , 175 . farooq , m. u. , jamil , m. , and rashid , m. a. ( 2010 ) . interacting entropy - corrected new agegraphic k - essence , tachyon and dilaton scalar field models in non - flat universe
. international journal of theoretical physics , 49(9 ) , 2278 - 2287 .
setare , m. r. , and saridakis , e. n. ( 2009 ) .
braneworld models with a non - minimally coupled phantom bulk field : a simple way to obtain the1-crossing at late times .
journal of cosmology and astroparticle physics , 2009(03 ) , 002 .
einstein , a. ( 1928 ) .
new possibility for a unified field theory of gravitation and electricity .
sitzungsberichte der preussischen akademie der wissenschaften , physikalisch - mathematische klasse , 17 , 224227 .
lu , j. , xu , l. , li , j. , chang , b. , gui , y. , and liu , h. ( 2008 ) .
constraints on modified chaplygin gas from recent observations and a comparison of its status with other models .
physics letters b , 662(2 ) , 87 - 91 .
khurshudyan , m. , kahya , e. o. , pasqua , a. , and pourhassan , b. ( 2014 ) .
higher derivative corrections of @xmath146 gravity with varying equation of state in the case of variable @xmath147 and @xmath148 .
arxiv preprint arxiv:1401.6630 .
amanullah , r. , lidman , c. , rubin , d. , aldering , g. , astier , p. , barbary , k. , ... and yasuda , n. ( 2010 ) .
spectra and hubble space telescope light curves of six type ia supernovae at 0.511 < z < 1.12 and the union2 compilation . the astrophysical journal , 716(1 ) , 712 . | in this paper , we have studied @xmath0 gravity as an arbitrary function of curvature and torsion scalars in friedmann lematre robertson
walker ( flrw ) background .
then , we have considered interacting model between @xmath0 gravity and modified chaplygin gas .
the novelty of this model is that the universe includes both cases curvature and torsion , and one dominated by a chaplygin gas . in order to calculate cosmological solutions
, we obtained friedmann equations and also equation of state ( eos ) parameter of dark energy . by employing interacting model we considered the total energy density and the total pressure of universe as the combination of components of dark energy and chaplygin gas .
subsequently , we reconstructed the model by an origin of a scalar field entitled quintessence model with a field potential .
the field potential has been calculated in terms of free parameters of @xmath0 gravity and modified chaplygin gas . in what follows , we used a parametrization , and
the cosmological parameters have been written in terms of redshift @xmath1 .
next , we plotted cosmological parameters with respect to three variable of cosmic time , redshift @xmath1 and @xmath2-folding number @xmath3 , and the figures showed us an accelerated expansion of universe . also , we have described the scenario in three status early time , late time and future time by @xmath2-folding number .
finally , the stability of scenario has been investigated by a useful function named sound speed , and the graph of sound speed versus @xmath2-folding number has been showed us that there is the stability in late time . |
the recent discovery at the large hadron collider ( lhc ) of a resonance at 125 gev compatible with the expectations for the higgs particle @xcite represents a major step towards understanding the origin of the mass of fundamental particles .
eventually , this should also affect the other subfield in which mass has a pivotal role , _
i.e. _ gravitation .
this is particularly relevant in models in which the higgs field has nonminimal coupling to the general relativity sector , as invoked in various extensions of the standard model .
nonminimal coupling between the higgs and spacetime curvature may be beneficial to have the higgs responsible for inflation @xcite , and as a suppression mechanism for the contribution to dark energy expected from quantum fields @xcite .
upper bounds on the gravitational interaction of higgs bosons from the lhc experiments have been recently discussed @xcite .
bounds on the crosstalk between the higgs particle and gravity may also be obtained by considering strong - gravity astrophysical objects , as proposed in @xcite in the case of active galactic nuclei ( agn ) and primordial black holes .
the presence of a strong spacetime curvature deforms the vacuum expectaction value of the higgs field and therefore the mass of fundamental particles such as the electron .
nucleons instead should be minimally affected by the strong curvature since most of their mass arises from the gluonic fields that , being massless , are not coupled to the higgs field at tree level .
peculiar wavelength shifts are therefore predicted which should be present for electronic transitions and strongly suppressed for molecular transitions in which the main role is played by the nuclei themselves , such as in vibrational or rotational spectroscopy . due to the vanishing of the ricci scalar for
spherically symmetric objects , attention was focused on the possibility of couplings to the only non - null curvature invariant , the kreschmann invariant , defined as @xmath0 , where @xmath1 is the riemann curvature tensor .
this invariant plays an important role in quadratic theories of gravity @xcite , and more in general in modified @xmath2 theories @xcite and einstein - gauss - bonnet models of gravity @xcite .
while agns would provide a strong - gravity setting near their black holes , their complex structure and the presence of turbulence and high - energy interactions near the accretion region induce uncontrollable systematic effects which hinder the possibility for extracting bounds on a higgs - kreschmann coupling as this relies upon the simultaneous observation of atomic and molecular transitions . to our knowledge ,
no neutron stars appear to show both molecular and atomic lines in their spectra , while white dwarfs have both .
although their surface gravity is much weaker than around agns and neutron stars , many features can be controlled more precisely , thus providing a quieter environment to search for the putative higgs shift .
white dwarfs have been known since the 19th century and in addition to their interest for astronomical and cosmological problems including understanding the late stages of stellar evolution , determining the galaxy s age , and the nature of ia supernovae , they have had a prominent role in fundamental physics since the early 20th century .
@xcite made the first attempt to verify general relativity by measuring the gravitational redshift of sirius b. @xcite studied the consequences of fermi - dirac statistics for stars , introducing his celebrated limit .
bounds on the distance dependence of the newtonian gravitational constant have been discussed comparing observations and models for the white dwarf sirius b @xcite and those in the hyades @xcite .
more recently @xcite proposed using white dwarfs to study the dependence of the fine structure constant on gravity . here
we show that white dwarfs can be used to obtain limits on the coupling of the higgs field to a specific curvature invariant , by means of spectroscopic observations of a carbon - rich white dwarf , bpm 27606 , using the southern african large telescope ( salt ) .
the analysis is complemented by considering data taken from the hst archive on a second white dwarf , procyon b , in which caii and mgii lines , in addition to the c@xmath3 bands , are also present .
the search for coupling between the higgs ( or any scalar field permeating the whole universe ) and spacetime curvature arises naturally within the framework of field theory in curved spacetime @xcite . the lagrangian density for an interacting scalar field in a generic spacetime characterized by the metric tensor @xmath4 is written as @xcite : @xmath5,\ ] ] where @xmath6 and @xmath7 are the mass parameter and the self - coupling quartic coefficient of the higgs field , respectively .
1 we have also introduced the determinant of the metric @xmath8 as @xmath9 , and @xmath10 , the coupling constant between the higgs field @xmath11 and the ricci scalar @xmath12 . the coupling constant @xmath10 is a free parameter in any model so far imagined to describe scenarios of scalar fields coupled to gravity , and it is therefore important to extract this coefficient , or upper bounds , from phenomenological analyses .
the higgs field develops , under spontaneous symmetry breaking , a vacuum expectation value @xmath13 in flat spacetime , and the masses of the fundamental fermions are proportional to @xmath14 via the yukawa coefficients of the fermion - higgs lagrangian density term , @xmath15 .
the effective mass parameter of the higgs field gets an extra - term due to the scalar curvature as @xmath16 , and the vacuum expectation value of the higgs field will become spacetime dependent through the curvature scalar as : @xmath17 where the approximation holds in a weak - curvature limit .
this implies that the mass @xmath18 of fundamental fermions , such as the electron , will be simply changed proportionally to the higgs vacuum expectation value @xmath19 in other words , the presence of coupling of the higgs field to space - time curvature adds to its inertial mass a contribution , which acts as a mass renormalization due to curved space - time ( however , for an opposite interpretation of this mass shift see @xcite ) .
this mass shift is not present for protons and neutrons , due to the fact that their mass is primarily due to the gluonic fields which , being massless , are unaffected by the higgs field .
as discussed in more detail in @xcite , this implies that all molecular transitions only depending on the nuclei mass , such as vibrational and rotational spectra , should be unaffected by the higgs - curvature coupling at leading order .
unfortunately , the ricci scalar outside spherically symmetric masses is zero , so we can not use this coupling to infer possible mass shifts for the electrons .
the only non - zero curvature scalar outside spherically symmetric masses is the kreschmann invariant , and we will assume in the following considerations bounds to the kreschmann coupling * @xmath20 * to the higgs field in a lagrangian of the form @xmath21,\ ] ] where @xmath22 is the planck length , whose value is @xmath23 m in conventional quantum gravity models , or larger values such as the one corresponding to models with _ early _ unifications of gravity to the other fundamental interactions @xcite .
in the latter case the planck length occurs at the tev scale via extra - dimensions , @xmath24 m , and in the following we will consider both these extreme situations .
analogously to the case of the ricci scalar , the mass parameter in the higgs term then gets normalized as @xmath25 , where we have introduced the compton wavelength corresponding to the higgs mass , @xmath26 , equal to @xmath27 m if we assume @xmath28 gev . notice that , due to the subattometer scale values of @xmath29 and @xmath30 , an extremely large value of @xmath20 is necessary for having mass shifts of order unity or lower to compensate for kreschmann invariants due to macroscopic curvature of any spacetime of astrophysical interest .
in fact , we get a relative mass shift , for instance in the case of the electron , equal to @xmath31 in the case of the schwarzschild metric the kretschmann invariant is @xmath32 , with @xmath33 the schwarzschild radius @xmath34 , and @xmath35 the distance from the center of the mass @xmath36 . as a benchmark , for a solar mass white dwarf , @xmath37 with a earth radius @xmath38 , we get @xmath39 m and @xmath40 . with @xmath23 m and the abovementioned value of @xmath29
we obtain @xmath41 in mksa units . for atomic transitions due to relocations of the electron in states with different principal quantum number
, we expect that the mass shift affects the spectroscopy with a scaling of the transition wavelengths as @xmath42 , and therefore any evidence for a wavelength shift in the electronic transitions not accompanied by the same shift for transitions determined by the mass of the nuclei may be a distintive signature of higgs - shifts .
therefore we need to detect emission or absorption wavelengths of both electronic and nuclear nature from a strong - gravity source , and make a comparison with either laboratory spectra or spectra gathered from weak - gravity astrophysical sources .
the spectrum of the c@xmath3 molecule has been the subject of extensive experimental and theoretical studies in molecular spectroscopy , and has been found in various astrophysical contexts including white dwarfs . in the visible region ,
the most prominent features of the c@xmath3 spectrum are the swan bands , involving vibronic transitions between the electronic states d@xmath43-a@xmath44 @xcite . for these transitions , in the presence of a higgs - shift the electronic energy levels , proportional to the electron mass , should be shifted , while the vibrational levels , proportional to the nucleon mass , should stay constant .
we therefore expect that the separation between different terms of the same swan band should stay constant , the only effect of the higgs shift being an overall shift of all the wavelengths . in the following we therefore focus on these specific spectra as gathered from two white dwarfs .
in the bruce proper motion survey , @xcite found bpm 27606 , which has a dme common proper motion companion .
@xcite indicated that it is a white dwarf from _
ubv _ photometry .
@xcite discovered the strong c@xmath3 swan bands which establish its spectral class as dq in the current classification scheme , and discussed its kinematical properties @xcite .
more detailed descriptions of the spectrum were given by @xcite and @xcite .
atmospheric analyses @xcite showed that bpm 27606 has a helium dominated atmosphere ( c : he @xmath45 ) and effective temperature @xmath46 k and more recent studies of dq atmospheres ( _ e.g. _ @xcite ) leave this essentially unchanged . in recent years
many new dq white dwarfs have been found ( _ e.g. _ @xcite ) and several new properties about them have been discovered .
these include rapid variability of some of the hot ones @xcite , rotation @xcite , and strong magnetic fields @xcite .
an additional problem that remains unsolved is the physics of the blueshifts in the swan bands of the dqs .
these were already measured in the milder cases of bpm 27606 and l 879 - 14 @xcite , but for cooler dqs this becomes more extreme , such as for lhs1126 @xcite . @xcite and @xcite have reviewed possible mechanisms , but this subject is clearly hampered by the lack of information on the behaviour of the spectra of ci and c@xmath3 under white dwarf conditions .
bpm 27606 is a particularly good star for studying these effects because it is in the common proper motion system with cd-51@xmath47 13128 .
this allows its true velocity to be known within fairly restrictive limits and the pair has a known distance from the trigonometric parallax .
in addition , the shifts in the lines are relatively small ( @xmath48 ) and its magnetic field is not large .
@xcite have reported a magnetic field measurement of @xmath49 mg from circular polarization of the ch bands near @xmath74300 .
however the magnetic field may not be this high . from our new spectra
this seems inconsistent with the lack of splitting of the @xmath74771 ci line and h@xmath50 , which indicates a magnetic field @xmath51 g. although variability is possible , the line is single on all of our spectra as it was in the 1978 spectra of @xcite .
the observations were made with the robert stobie spectrograph ( rss ) attached to the southern african large telescope ( salt ) , which is described by @xcite .
the rss @xcite ) employs volume space holographic transmission gratings ( vphgs ) and three e2v44 - 82 2048 @xmath52 4096 ccds with 15 @xmath6
m pixels separated by gaps of 1.5 mm width , or about 10 - 15 at the dispersions used here .
all of the observations of bpm 27606 were obtained using a 1.0 arcsec @xmath52 8 arcmin slit rotated by 71@xmath47 so that both the white dwarf and its bright companion could be observed simultaneously .
a 2 @xmath52 2 pixel binning with a gain of 1.0 e@xmath53/adu gives a readout noise of 3.3 e@xmath53/pixel .
the h@xmath54 region ( @xmath55 6085 - 6925 ) which also covers the @xmath56 bandhead was observed 2013 may 2 with the pg2300 grating and a pc04600 filter . three 900 seconds exposures were obtained along with ar comparison spectra .
the fwhm measured from comparison lines is 1.3 .
an identical set of exposures were secured 2014 may 12 to improve the signal - to - noise ratio . the @xmath55 4700 - 5380 wavelength region which covers the 5135 bandhead was observed 2013 may 19 ( two exposures ) and 2013 may 21 ( four exposures ) .
all exposure times were 812 seconds with the pg3000 grating , no filter , and cuar comparisons were used , giving a fwhm of comparison lines of 1.0 .
the 4737 and 4382 bandheads ( @xmath55 4326 - 5004 ) were observed 2013 may 5 and three 1042 second exposures using the pg3000 grating and no filter with a cuar wavelength comparison .
the resulting comparison line has a fwhm=1.1 . in 2014 , two additional spectral regions were observed .
three 900 seconds exposures were secured 2014 may 13 of @xmath553540 - 4323 with the pg3000 grating , no filter and a thar comparison producing a fwhm resolution of 1.3 from companion lines .
three 663 seconds exposures covering @xmath555050 - 6010 were made 2014 may 22 using the pg2300 grating with an ar comparison giving a fwhm of 1.7 for the comparison lines .
starting from the bias subtracted and flattened images provided by the salt pipeline @xcite , the spectra were wavelength calibrated using the longslit menu in iraf .
fifth order polynomials were used for the wavelength calibrations in two dimensions , background subtraction , and the spectra were extracted using apsum .
special care was taken to use portions of the frames adjacent to the spectra , to minimize the effects of the curved lines in the rss .
the final individual spectra were summed using imcombine and the ccdsum option to remove cosmic rays .
examples of the resulting spectra are shown in figures 1 - 5 .
the heliocentric velocity of the m2ve companion , measured from our spectra using the h@xmath57 emission lines , was @xmath58 km / s , which we adopt here .
this can be compared with @xmath59 km / s @xcite and @xmath60 km / s @xcite .
the observed wavelengths of the major bandheads of c@xmath3 are shown in table 1 where they are compared with the wavelengths given by @xcite .
these wavelengths refer to the minimum intensity at each bandhead from our spectra , and have been corrected to heliocentric values .
we have analyzed the distance between two consecutive lines corresponding to the same @xmath61 , for the observed wavelengths @xmath62 and for the laboratory measured wavelengths @xmath63 , evaluated both the overall average separation as @xmath64 $ ] and its dispersion as @xmath65 ^ 2 $ ] , obtaining the values of @xmath66 and @xmath67 .
the fact that @xmath68 shows that , within the instrumental error , the spacing of the vibrational transitions is the same for the laboratory and the observed wavelengths .
this provides a reliable anchor to study possible shifts in the bandheads due to the electron mass shift induced by the higgs field .
the stability of the vibrational transitions is further assured by their insensitivity to pressure - induced shifts since these are expected to be the same , in a linear approximation suitable for low pressures , for different vibronic bands @xcite . due to the lack of recent detailed spectral scans of bpm 27606
, its effective temperature is estimated from photometry .
@xcite used intermediate band strmgren uby data from @xcite and @xmath69 models which yield an effective temperature @xmath70 k and @xmath71 . using models of @xcite these colours give @xmath72 and @xmath73 .
@xcite found @xmath74 k , @xmath75 , and @xmath76 in line , within few standard deviations , with our findings .
broadband ubv photometry , 2mass near infrared colours @xcite and an atmosphere model by @xcite bracket these values , so for the present estimates we adopt @xmath77 k and @xmath78 . to estimate the mass , we first determine the white dwarf radius using the relationship @xmath79 with @xmath80 from @xcite .
the visual magnitude @xmath81 is the mean of @xcite and @xcite . for the parallax we adopt @xmath82 , the mean of three parallax measurements : from hipparcos @xcite , the yale parallax catalogue @xcite , and the photometric parallax using the colour magnitude diagrams of reid ( http://www.stsci.edu ~ inr / cmd.html ) for the dm companion
this gives @xmath83 which implies a mass @xmath84 according to the hamada and salpeter carbon or chandrashekhar @xmath85 mass - radius relations .
these values give a surface gravity of @xmath86 or @xmath87 .
the corresponding gravitational redshift would be @xmath88 .
the leading source of systematic effects in the c@xmath3 bands is pressure shifts in the dominant he background atmosphere , and in this section we estimate their order of magnitude .
we interpolate in the dq models of @xcite for the atmospheric parameters at rosseland mean optical depth @xmath89 for @xmath90 k and @xmath91 for the order of magnitude of conditions in the line forming region .
this is @xmath92 k and @xmath93 .
as the models use @xmath94 , and for bpm 27606 @xmath95 , the number density is scaled to be @xmath96 .
@xcite measured pressure shifts for the five @xmath97 bandheads observed here in he under conditions close to those in bpm 27606 ( @xmath98 k and @xmath99 ) . although these measurements can not explain the large blueshifts @xmath100 -200 in the cooler peculiar dq stars @xcite , the shifts in bpm 27606 are much smaller than this and the laboratory measurements are of similar size , so it seems reasonable that for this star they can be used to estimate the pressure shifts .
if the line broadening in the swan bands resembles van der waals broadening , the pressure shifts measured by @xcite , @xmath101 , would scale as @xmath102 with @xmath103 expressed in k and @xmath104 in @xmath105 .
for the estimated conditions at @xmath89 , @xmath106 , which is used to correct the measured wavelengths of the @xmath97 bandheads given in table 2 as @xmath107 , where @xmath108 is the gravitational redshift .
the major source of systematic error in equation 6 is the choice of @xmath89 for the model atmosphere .
this turns out to be relatively insensitive , as increasing the rosseland optical depth to @xmath109 would multiply @xmath110 in tables 2 and 3 by a factor 1.2 .
we also detect a weak feature at @xmath111 .
if this is the @xmath112 line , @xmath113 , no pressure shifts are expected , and this gives a gravitational redshift @xmath114 km / s .
we do not detect the @xmath115 line in our spectra .
the ch(0,0 ) band at 4314.2 appears to be present as a weak feature at @xmath116 and helps to confirm the presence of hydrogen .
a weak ci line can be seen in figure 2 ( see also fig . 3 for details ) .
the measured heliocentric wavelength neglecting pressure shifts is 4771.1 .
the laboratory wavelength of the strongest component of the @xmath117 multiplet is 4771.75 @xcite .
there are currently no data on pressure shifts of ci lines .
the magnetic field on the surface of bpm 27606 is estimated looking at the resolution of the ci line .
the classical zeeman effect yields the relationship @xmath118 ( in cgs units ) .
the fact that only a single line with fwhm @xmath119 is visible ( although with a low signal - to - noise ratio ) , yields an upper bound on the magnetic field of @xmath120 g , which makes negligible any correction to the estimates above . using
detailed calculations by @xcite of the magnetic splitting of this line also suggest such a low magnetic field .
the orbital motion is estimated considering the presence of the common proper motion companion to bpm 27606 with @xmath121 , of spectral type m2ve ( @xmath122 ) at a projected separation of 28.45 arcsec , which at 15 pc yields a separation of @xmath123 au . with a white dwarf mass of @xmath124 and taking this to be the semimajor axis of a circular orbit
, the orbital period would be @xmath125 years and the projected orbital velocity would be @xmath126 km / s , which is the order of magnitude of an additional source of uncertainty in the gravitational redshift determination . for bpm 27606 we conclude that an upper bound to the wavelength difference between the molecular c@xmath3 bands and the h@xmath127 and ci atomic lines is 1.3 .
this is obtained by summing up the variance of the estimated pressure shifts ( the square of the error in the second column of table 2 , corrected by the 1.44 hammond scaling factor ) and the square of the final difference between the processed and the laboratory wavelength ( last column in table 2 ) , all other systematic errors being negligible with respect to these sources .
if we disqualify the troubling ci line , it is 0.7 .
although the corrections for pressure shifts in the c@xmath3 bands seems satisfactory , one must consider the uncertainties in their measurements which dominate the error budget in our estimate of the difference and is of the order of @xmath128 2 .
procyon b is another white dwarf that has both atomic and molecular features in its spectrum and is in a well known binary .
its orbit and mass have been long studied ( _ e.g. _ @xcite ; @xcite ; @xcite ) .
@xcite obtained masses of @xmath129 and @xmath130 for the two components .
although procyon b was long known to be a white dwarf , its spectrum could not be studied due to its proximity to its bright ( @xmath131 + 0.34 ) primary ( separation @xmath132 5 arcsec ) .
@xcite secured spectra using the stis instrument on the hst in 1998 february ( proposal 7398 ; pi h.l .
shipman ) where the observations and data reductions are detailed .
here we adopt the atmospheric analysis in @xcite which gives @xmath133 k , @xmath134 and @xmath135 .
these parameters show that procyon b lies close to the carbon white dwarf mass - radius relation and predict a surface gravity of @xmath136 and a gravitational redshift of @xmath137 . from the hst archives we used images
04g802010 , 04g802020 , 04g8020j0 , and 04g802090 which have suitable signal to noise ratio .
we measured the wavelengths for @xmath97 bands , mgii and caii lines .
figure 6 shows the @xmath97 bands near 4,737 and figure 7 shows the caii lines .
the remainder of the spectrum is shown in @xcite . the orbital velocity from @xcite for the radial motion of procyon b in 1998 is @xmath138 .
the models of @xcite using the @xmath139 and @xmath9 above indicate at @xmath89 , t@xmath140=6,750 k , and n@xmath141 .
the @xmath97 pressure shifts of @xcite discussed in section 3.2 are thus multiplied by 1.3 .
the pressure shifts of the observed caii and mg lines produced by he are available and in both cases are redshifts . for caii , theoretical values @xcite and laboratory measurements @xcite agree relatively well .
@xcite calculated pressure shifts for mgii and these were scaled to the above atmospheric parameters assuming van der waals broadening .
table 3 summarizes the laboratory and measured wavelengths of features in the spectrum of procyon b along with the corrections due to the orbital motion , @xmath142 , the estimated pressure shifts @xmath110 , and the gravitational redshift , @xmath108 .
the resulting corrected wavelength of each line , @xmath143 , and the residual @xmath144 are given .
these are consistent with the results for bpm 27606 . by repeating the analysis as for the latter , and using the hammond scaling factor of 1.3 ,
the difference between atomic and molecular lines is formally 0.5 , with the same uncertainties due to pressure shifts as before .
based on the absence of relative shifts between the electronic and the vibrational transitions of the swan bands , we are able to assess upper bounds on the higgs - kreschmann coupling . the expected wavelength shift is @xmath145 where the kreschmann invariant is , for the estimated values of the mass and radius of bpm 27606 , @xmath146 . by assuming @xmath147 m , and the value of @xmath148 quoted in section 2 , we obtain @xmath149 , which may be inverted yielding an upper bound , for an average wavelength of @xmath150 @xmath151 where @xmath152 ( in ) is the estimated wavelength resolution .
if instead we use @xmath153 m as in models with extra dimension and quantum gravity at the fermi scale @xcite , the bounds are a bit more constraining , as @xmath154 at @xmath155 .
both examples of upper bounds are expressed in meter - kilogram - second - ampere ( mksa ) units of the systeme international ( si ) , and it is worth converting them into natural units for the benefit of a comparison to the bounds already estimated from the observation of the higgs particle at lhc @xcite . the action term for the higgs field including its coupling to spacetime , taking into account explicitly @xmath156 and @xmath157 , is expressed in si units in terms of , for instance , the mass term as @xmath158
. then the scalar field has dimensions @xmath159=m^{1/2 } t^{-1/2}$ ] , and the higgs - curvature term satisfying @xmath160 implies that @xmath10 is dimensionless . if the analysis is repeated for the action term expressed in natural units ( nu ) , with @xmath161 , the dimensions of the scalar field change accordingly , and from the mass term it is simple to infer that @xmath162 , implying the following relationship between the @xmath10 parameter evaluated in the two units systems , @xmath163 .
analogous considerations may be repeated for the higgs - kreschmann coupling in which the curvature term has the same dimensions since @xmath12 is replaced by @xmath164 .
this implies that in natural units the bounds on the higgs - kreschmann coupling become respectively , for the two extreme choices of @xmath30 , @xmath165 ( for @xmath147 m ) and @xmath166 ( for @xmath153 m ) .
this second bound on the higgs - kreschmann coupling is quantitatively comparable to the ones assessed on the higgs - ricci coupling through measurements at the lhc as reported in @xcite and @xcite , although in their analyses the usual @xmath147 m is assumed . with respect to bounds from table - top experiments based on tests of the superposition principle for gravitational interactions
as discussed in @xcite , the bounds derived in this paper represent an improvement by ten orders of magnitude , as shown in table 4 .
we have shown that observations in a somewhat controlled environment like the one provided by carbon - rich white dwarfs may be used to give upper bounds on the coupling between the higgs field and a specific invariant of the curvature of the spacetime , such as the kreschmann invariant .
the existence of nonminimal couplings between the higgs and spacetime curvature is crucial to various proposals in which the higgs also plays the role of the inflaton @xcite , and as a mechanism to suppress the dark energy contribution of quantum fields to the level compatible with the astrophysical observations based on snia @xcite .
this methodology is complementary to the upper bounds recently discussed arising from the lhc experiments and their degree of agreement with the standard model of elementary particle physics @xcite , and can be adopted also to search for couplings between generic scalar fields , not necessarily directly related to the higgs vacuum , and space - time curvature , which may be competitive with bounds arising from the analysis of the cosmic background radiation as reported in @xcite .
scalar fields , even if not directly interacting among themselves at level of their classical lagrangian , will have a crosstalk once quantum radiative corrections are considered - the very origin of the hierarchy problem in grand unified theories - so any scalar field will be then coupled with the higgs field and will indirectly affect the mass of elementary particles .
bounds based on this analysis could be even more stringent as many of the proposed candidates , for instance scalar fields invoked to accommodate the acceleration of the universe @xcite have a compton wavelength greatly exceeding the one associated to the higgs field , provided that compact astrophysical sources with non - zero ricci scalar may be found .
the spectroscopic analysis presented here could be improved in a number of ways in the near future .
measurements and calculations of the pressure shifts of c i lines and the @xmath97 swan bands under white dwarf conditions are needed .
observations of the ultraviolet carbon and other metal lines which are from lower energy levels would help disentangle the pressure shifts from the gravitational redshift measurement . a detailed scan of the white dwarf s spectral energy distribution combined with an updated atmosphere model would help understanding the details of the molecular and atomic carbon features .
additional objects of this type that are in binary pairs would help in the assessment of the local gravity , further diversifying the sample to counterbalance peculiar systematic effects .
we are grateful to susanne yelin for a critical reading of the manuscript .
some of the observations reported in this paper were obtained with the southern african large telescope ( salt ) under proposals 2013 - 1-dc-001 and 2014 - 1-dc-001 .
additional data used observations made with the nasa / esa hubble space telescope , and were obtained from the hubble legacy archive , which is a collaboration between the space telescope science institute ( stsci / nasa ) , the space telescope european coordinating facility ( st - ecf / esa ) and the canadian astronomy data centre ( cadc / nrc / csa ) .
this work was also partially funded by the national science foundation through a grant for the institute for theoretical atomic , molecular , and optical physics at harvard university , and the smithsonian astrophysical laboratory .
cccc @xmath167 & @xmath168 & @xmath169 & @xmath170 + 0,2 & 6179.90 & 6191.2 & -11.3 + 1,3 & 6114.74 & 6122.1 & -7.4 + 0,1 & 5630.35 & 5635.5 & -5.15 + 1,2 & 5580.07 & 5585.5 & -5.4 + 2,3 & 5536.00 & 5540.7 & -4.7 + 3,4 & 5495.64 & 5501.9 & -6.3 + 4,5 & 5467.73 & 5470.3 & -2.6 + 0,0 & 5159.97 & 5165.2 & -5.2 + 1,1 & 5125.64 & 5129.3 & -3.7 + 2,2 & 5094.45 & 5097.7 & -3.3 + ccccccc line & @xmath101 & @xmath110 & @xmath171 & @xmath172 & @xmath108 & @xmath173 + ( 1 ) & ( 2 ) & ( 3 ) & ( 4 ) & ( 5 ) & ( 6 ) & ( 7 ) + c@xmath3(0,2 ) & -8.4 @xmath128 0.3 & -12.1 & -11.3 & + 0.18 & -1.0 & 0.0 + c@xmath3(0,1 ) & -3.2 @xmath128 0.5 & -4.6 & -5.15 & + 0.17 & -0.9 & -1.3 + c@xmath3(0,0 ) & -3.2 @xmath128 0.5 & -4.6 & -5.2 & + 0.16 & -0.8 & + 0.0 + c@xmath3(1,0 ) & -3.8 @xmath128 1.0 & -5.5 & -4.0 & + 0.14 & -0.7 & + 0.9 + c@xmath3(2,0 ) & -3.1 @xmath128 1.0 & -4.5 & -3.6 & + 0.13 & -0.7 & + 0.3 + ci(@xmath74771 ) & - & - & -0.65 & + 0.15 & -0.75 & -1.2 + h@xmath127 & - & - & + 0.5 & + 0.20 & -1.0 & -0.3 + cccccccc line & @xmath169 & @xmath168 & @xmath142 & @xmath110 & @xmath108 & @xmath143 & @xmath173 + ( 1 ) & ( 2 ) & ( 3 ) & ( 4 ) & ( 5 ) & ( 6 ) & ( 7 ) & ( 8) + c@xmath3(0,0 ) & 5165.2 & 5161.8 & -0.15 & + 4.2 & -0.53 & 5165.3 & + 0.1 + c@xmath3(1,1 ) & 5129.3 & 5126.5 & -0.15 & & & & + c@xmath3(1,0 ) & 4737.1 & 4733.1 & -0.14 & + 4.9 & -0.49 & 4737.4 & + 0.3 + c@xmath3(2,1 ) & 4715.2 & 4714.5 & -0.14 & & -0.49 & & + c@xmath3(3,2 ) & 4697.6 & 4696.1 & -0.14 & & -0.48 & & + caii h & 3968.47 & 3969.84 & -0.12 & -1.4 & -0.41 & 3967.9 & -0.6 + caii k & 3933.66 & 3935.18 & -0.11 & -0.4 & -0.41 & 3934.3 & + 0.7 + mgii & 2802.70 & 2803.10 & -0.08 & -0.5 & -0.29 & 2802.2 & -0.5 + mgii & 2795.53 & 2796.85 & -0.08 & -0.6 & -0.29 & 2795.9 & + 0.4 + cccccc system & @xmath174 & @xmath175 & @xmath176 ( ) & @xmath177(@xmath178 m ) & @xmath177(@xmath179 m ) + bpm 27606 & 0.0105 & 0.78 & 1.3 & @xmath180 & @xmath181 + procyon b & 0.0124 & 0.602 & 0.5 & @xmath182 & @xmath183 + table - top experiments & & & & @xmath184 & @xmath185 + | we report on a search for differential shifts between electronic and vibronic transitions in carbon - rich white dwarfs bpm 27606 and procyon b. the absence of differential shifts within the spectral resolution and taking into account systematic effects such as space motion and pressure shifts allows us to set the first upper bound of astrophysical origin on the coupling between the higgs field and the kreschmann curvature invariant .
our analysis provides the basis for a more general methodology to derive bounds to the coupling of long - range scalar fields to curvature invariants in an astrophysical setting complementary to the ones available from high - energy physics or table - top experiments . |
galactic winds that extend on a scale comparable to the host galaxies are now known to be a common feature both locally and at high redshifts .
these winds are significant carriers of mass , momentum , and energies that may impact the formation and evolution of the host galaxies and the surrounding intergalactic medium . given the scope of this conference , the present review focusses exclusively on starburst - driven winds .
agn - driven galactic winds , perhaps a very important phenomenon in the lives of galaxies with spheroids ( kormendy & gebhardt 2001 ) , are not discussed here ( see , e.g. , veilleux et al .
2002a for a recent review of this topic ) .
due to space limitations , the emphasis of this review is on the recent ( @xmath0 1998 ) literature .
readers interested in results from earlier studies may refer to the reviews by strickland ( 2002 ) and heckman ( 2002 ) .
first , the basic physics of starburst - driven winds is described briefly in 2 .
an observational summary of the properties of local winds is given in the preamble to 3 .
the remainder of 3 describes detailed data on three well - studied cases of local starburst - driven winds , and summarizes the evidence for winds in luminous and ultraluminous infrared galaxies and distant lyman break galaxies .
this section often emphasizes the importance of using multiwavelength data to draw a complete picture of this complex multi - phase phenomenon .
the impact of starburst - driven winds on the host galaxies and their environment is discussed briefly in 4 . here
the focus of the discussion is on the existence and properties of the wind fluid and on the size of the `` zone of influence '' of these winds .
a summary is given in 5 .
the driving force behind starburst - driven winds is the mechanical energy from stellar winds and supernova events ( e.g. , chevalier & clegg 1985 ) .
this mechanical energy is quickly thermalized to produce a hot cavity with a temperature @xmath1 k , where @xmath2 is the mass - loading term .
this over - pressured cavity expands through the ambient medium , sweeping this material up in the process to produce a bubble - like structure .
the complex interaction between the wind and the ism of the host galaxy has been the subject of several numerical simulations ( e.g. , maclow & mccray 1988 ; suchkov et al . 1994 , 1996 ; maclow & ferrara 1999 ; dercole & brighenti 1999 ; strickland & stevens 2000 ; silich & tenorio - tagle 2001 ) .
if radiative energy losses are negligible ( probably a good assumption in some objects ; e.g. , heckman et al . 2001 ) , the bubble expands adiabatically through the galaxy ism with a velocity @xmath3 km s@xmath4 , where @xmath5 is the ambient nucleon density in @xmath6 , @xmath7 is the rate of deposition of mechanical energy in 10@xmath8 erg s@xmath4 , and @xmath9 is the age of the bubble in 10@xmath10 years ( e.g. , weaver et al . 1977 ) .
a powerful starburst may inject enough energy to produce a cavity of hot gas that can burst out of the disk ism , at which point the dense walls of the bubble start accelerating outward , become rayleigh - taylor unstable , and break up into cloudlets and filaments .
if halo drag is negligible ( probably _ not _ a good assumption in general ) , the wind fluid may reach terminal velocities as high as @xmath11 3000 @xmath12 km s@xmath4 , well in excess of the escape velocity of the host galaxy .
in contrast , the terminal velocities of clouds accelerated by the wind are more modest , of order @xmath13 , where @xmath14 is the wind momentum flux in 10@xmath15 dynes , @xmath16 is the solid angle of the wind in steradians , @xmath17 is the initial position of the cloud in kpc , and @xmath18 is the column density of the cloud in 10@xmath19 @xmath20 ( strelnitskii & sunyaev 1973 ; heckman et al .
2000 ) .
a critical quantity in all of these calculations is the thermalization efficiency , or the percentage of the mechanical energy from the starburst that goes into heating the gas .
unfortunately , this quantity is poorly constrained observationally .
most simulations assume a thermalization efficiency of 100% , i.e. none of the energy injected by the starburst is radiated away . in reality , this efficiency depends critically on the environment , and is likely to be significantly less than 100% in the high - density environment of powerful nuclear starbursts ( e.g. , thornton et al .
1998 ; strickland & stevens 2000 ; silich , tenorio - tagle , & muoz - tun 2003 ) .
galactic magnetic fields may also `` cushion '' the effects of the starburst on the ism , and reduce the impact of the galactic wind on the host galaxy and its environment ( e.g. , tomisaka 1990 ; ferrire et al .
1991 ; slavin & cox 1992 ; mineshinge et al .
1993 ; ferrire 1998 ) .
a great number of surveys have provided important statistical information on galactic winds in the local universe ( e.g. , heckman , armus , & miley 1990 ; veilleux et al .
1995 ; lehnert & heckman 1995 , 1996 ; gonzalez delgado et al .
1998 ; heckman et al .
2000 , rupke , veilleux , & sanders 2002 , 2003 , in prep . ) .
galaxy - scale winds are common among galaxies with global star formation rates per unit area @xmath21
m@xmath22 yr@xmath4 kpc@xmath23 , where @xmath24 is the optical radius .
this general rule - of - thumb also appears to apply to ultra / luminous infrared galaxies ( see 3.3 ) and distant lyman break galaxies ( see 3.4 ) .
`` quiescent '' galaxies with global star formation rates per unit area below this threshold often show signs of galactic fountaining in the forms of warm , ionized extraplanar material a few kpc above or below the galactic disks ( e.g. , miller & veilleux 2003a , 2003b and references therein ) .
the energy input from stellar winds and supernovae in these objects elevates some of the ism above the disk plane , but is not sufficient to produce large - scale winds .
this rule - of - thumb is conservative since a number of known wind galaxies , including our own galaxy ( 3.1 ) and several dwarf galaxies , have @xmath25 0.1 m@xmath22 yr@xmath4 kpc@xmath23 ( e.g. , hunter & gallagher 1990 , 1997 ; meurer et al .
1992 ; marlowe et al . 1995 ; kunth et al . 1998
; martin 1998 , 1999 ; see kunth s and martin s contributions at this conference ) .
the production of detectable winds probably depends not only on the characteristics of the starburst ( global _ and _ local @xmath26 , starburst age ) , but also on the detailed properties of the ism in the host galaxies ( e.g. , see the theoretical blowout criterion of maclow & mccray 1988 ) .
the winds in actively star - forming galaxies in the local universe show a very broad range of properties , with opening angles of @xmath11 0.1 0.5 @xmath27 ( 4@xmath28 sr ) , radii ranging from @xmath29 1 kpc to several 10s of kpc , outflow velocities of a few 10s of km s@xmath4 to more than 1000 km s@xmath4 ( with clear evidence for a positive correlation with the temperature of the gas phase ) , total ( kinetic and thermal ) outflow energies of @xmath11 10@xmath30 10@xmath31 ergs or 5 20% of the total mechanical energy returned to the ism by the starburst , and mass outflow rates ranging from @xmath29 1 m@xmath22 yr@xmath4 to @xmath32 100 m@xmath22 and scaling roughly with the star formation rates ( see 3.3 below ) . in the remainder of this section
, we discuss a few well - studied cases of galactic winds in the local universe and summarize the evidence for winds in luminous and ultraluminous infrared galaxies at low and moderate redshifts as well as in distant lyman break galaxies . by far
the closest case of a large - scale outflow is the wind in our own galaxy .
evidence for a dusty bipolar wind extending @xmath11 350 pc ( @xmath11 1@xmath33 ) above and below the disk of our galaxy has recently been reported by bland - hawthorn & cohen ( 2003 ) based on data from the midcourse space experiment ( msx ) .
the position of the warm dust structure coincides closely with the well - known galactic center lobe detected at radio wavelengths ( e.g. , sofue 2000 and references therein ) .
simple arguments suggest that the energy requirement for this structure is of order @xmath11 10@xmath34 ergs with a dynamical time scale of @xmath11 1 myr .
bland - hawthorn & cohen ( 2003 ) also argue that the north polar spur , a thermal x - ray / radio loop that extends from the galactic plane to @xmath35 ( e.g. , sofue 2000 ) , can naturally be explained as an open - ended bipolar wind , when viewed in projection in the near field .
this structure extends on a scale of 10 20 kpc and implies an energy requirement of @xmath11 ( 1 30 ) @xmath27 10@xmath34 ergs and a dynamical timescale of @xmath11 15 myr , i.e. considerably longer than that of the smaller structure seen in the msx maps .
if confirmed , this may indicate that the milky way galaxy has gone through multiple galactic wind episodes .
bland - hawthorn & cohen ( 2003 ) point out that the north polar spur would escape detection in external galaxies ; it is therefore possible that the number of galaxies with large - scale winds has been ( severely ? ) underestimated .
two classic examples of starburst - driven outflows are described in this section to illustrate the wide variety of processes taking place in these objects . 0.1 in * m 82 .
* this archetype starburst galaxy hosts arguably the best studied galactic wind .
some of the strongest evidence for the wind is found at optical wavelengths , where long - slit and fabry - perot spectroscopy of the warm ionized filaments above and below the disk shows line splittings of up to @xmath11 250 km s@xmath4 , corresponding to deprojected velocities of order 525 655 km s@xmath4 ( e.g. , mckeith et al .
1995 ; shopbell & bland - hawthorn 1998 ) . combining these velocities with estimates for the ionized masses of the outflowing filamentary complex ,
the kinetic energy involved in the warm ionized outflow is of order @xmath11 2 @xmath27 10@xmath34 ergs or @xmath11 1% of the total mechanical energy input from the starburst .
the ionized filaments are found to lie on the surface of cones with relatively narrow opening angles ( @xmath11 5 25@xmath33 ) slightly tilted ( @xmath11 5 15@xmath33 ) with respect to the spin axis of the galaxy .
deep narrow - band images of m82 have shown that the outflow extends out to at least 12 kpc on one side ( e.g. , devine & bally 1999 ) , coincident with x - ray emitting material seen by @xmath36 ( lehnert , heckman , & weaver 1999 ) and @xmath37-newton ( stevens , read , & bravo - guerrero 2003 ) .
the wind fluid in this object has apparently been detected by both @xmath38 ( griffiths et al .
2000 ) and @xmath37-newton ( stevens et al .
2003 ) . the well - known h i complex around this system ( e.g. , yun et al .
1994 ) may be taking part , and perhaps even focussing , the outflow on scales of a few kpc ( stevens et al .
recently published high - quality co maps of this object now indicate that some of the molecular material in this system is also involved in the large - scale outflow ( walter , weiss , & scoville 2002 ; see also garcia - burillo et al .
the outflow velocities derived from the co data ( @xmath11 100 km s@xmath4 on average ) are considerably lower than the velocities of the warm ionized gas , but the mass involved in the molecular outflow is substantially larger ( @xmath11 3 x 10@xmath39 m@xmath22 ) , implying a kinetic energy ( @xmath11 3 @xmath27 10@xmath34 ergs ) that is comparable if not larger than that involved in the warm ionized filaments . the molecular gas is clearly a very important dynamical component of this outflow . 0.1 in * ngc 3079 .
* an outstanding example of starburst - driven superbubble is present in the edge - on disk galaxy , ngc 3079 .
high - resolution @xmath40 h@xmath41 maps of this object show that the bubble is made of four separate bundles of ionized filaments ( cecil et al .
the two - dimensional velocity field of the ionized bubble material derived from fabry - perot data ( veilleux et al .
1994 ) indicates that the ionized bubble material is entrained in a mushroom vortex above the disk with velocities of up to @xmath11 1500 km s@xmath4 ( cecil et al .
2001 ) . a recently published x - ray map obtained with the @xmath38 ( cecil , bland - hawthorn , & veilleux 2002 ) reveals excellent spatial correlation between the hot x - ray emitting gas and the warm optical line - emitting material of the bubble , suggesting that the x - rays are being emitted either as upstream , standoff bow shocks or by cooling at cloud / wind conductive interfaces .
this good spatial correlation between the hot and warm gas phases appears to be common in galactic winds ( strickland et al . 2000 , 2002 ; veilleux et al .
2003 , and references therein ) .
the total energy involved in the outflow of ngc 3079 appears to be slightly smaller than that in m 82 , although it is a lower limit since the total extent of the x - ray emitting material beyond the nuclear bubble of ngc 3079 is not well constrained ( cecil et al .
contrary to m 82 , the hot wind fluid that drives the outflow in ngc 3079 has not yet been detected , and evidence for entrained molecular gas is sparse and controversial ( e.g. , irwin & sofue 1992 ; baan & irwin 1995 ; israel et al .
1998 ; but see koda et al . 2002 ) .
given that the far - infrared energy output of a ( dusty ) galaxy is a direct measure of its star formation rate , it is not surprising _ a posteriori _ to find evidence for large - scale galactic winds in several luminous and ultraluminous infrared galaxies ( lirgs and ulirgs ; e.g. , heckman et al .
1990 ; veilleux et al .
systematic searches for winds have been carried out in recent years in these objects to look for the unambiguous wind signature of blueshifted absorbing material in front of the continuum source ( heckman et al .
2000 ; rupke et al .
2002 ) . the feature of choice to search for outflowing neutral material in galaxies of moderate redshifts ( @xmath42 0.6 )
is the na i d interstellar absorption doublet at 5890 , 5896 .
the wind detection frequency derived from a set of 44 starburst - dominated lirgs and ulirgs is high , of order @xmath11 70 80% ( rupke
et al . 2002 , 2003 in prep . ;
also see rupke s and martin s contributions at this conference ) .
the outflow velocities reach values in excess of 1700 km s@xmath4 ( even more extreme velocities are found in some agn - dominated ulirgs ) . a simple model of a mass - conserving free wind ( details of the model
are given in rupke et al .
2002 ) is used to infer mass outflow rates in the range @xmath43(h)@xmath44 for galaxies hosting a wind .
these values of @xmath43 , normalized to the corresponding global star formation rates inferred from infrared luminosities , are in the range @xmath45 .
the parameter @xmath46 , often called the `` mass entrainment efficiency '' or `` reheating efficiency '' shows no dependence on the mass of the host ( parameterized by host galaxy kinematics and absolute @xmath47- and @xmath48-band magnitudes ) , but there is a possible tendency for @xmath46 to decrease with increasing infrared luminosities ( i.e. star formation rates ) .
the large molecular gas content in ulirgs may impede the formation of large - scale winds and reduce @xmath46 in these objects . a lower thermalization efficiency ( i.e. higher radiative efficiency ) in these dense gas - rich systems may also help explain the lower @xmath46 ( rupke et al .
2003 , in prep . ;
see rupke s contribution at this conference ) .
evidence for galactic winds has now been found in a number of z @xmath11 3 5 galaxies , including an important fraction of lyman break galaxies ( lbgs ; e.g. , franx et al .
1997 ; pettini et al .
2000 , 2002 ; frye , broadhurst , & benitez 2002 ; dawson et al .
2002 ; ajiki et al .
2002 ; adelberger et al . 2003 ; shapley et al .
the best studied wind at high redshift is that of the gravitationally lensed lbg ms 1512-cb58 ( pettini et al . 2000 , 2002 ) .
an outflow velocity of @xmath11 255 km s@xmath4 is derived in this object , based on the positions of the low - ionization absorption lines relative to the rest - frame optical emission lines ( ly@xmath41 is to be avoided for this purpose since resonant scattering and selective dust absorption of the ly@xmath41 photons may severely distort the profile of this line ; e.g. , tenorio - tagle et al . 1999 ) .
the mass - conserving free wind model of rupke et al .
( 2002 ) applied to ms 1512-cb58 ( for consistency ) results in a mass outflow rate of @xmath11 20 @xmath49 yr@xmath4 , equivalent to about 50% the star formation rate of this galaxy based on the dust - corrected uv continuum level .
similar outflow velocities are derived in other lbgs ( pettini et al .
the possibly strong impact of these lbg winds on the environment at high @xmath50 is discussed in the next section ( 4.2 ) .
starburst - driven winds may have a strong influence on structure formation at high redshifts , on the `` porosity '' of star - forming galaxies ( i.e. the probability for ionizing photons to escape their host galaxies ) , hence the nature of the extragalactic uv and infrared background , and on the chemical and thermal evolution of galaxies and their environment . due to space limitations
, this review only addresses the last issue .
* hot metal - enriched gas . * nuclear starbursts inject both mechanical energy and metals in the centers of galaxies .
this hot , chemically - enriched material , the driving engine of galactic winds , is eventually deposited on the outskirts of the host galaxies , and contributes to the heating and metal enrichment of galaxy halos and the igm .
surprisingly little evidence exists for the presence of this enriched wind fluid .
this is due to the fact that the wind fluid is tenuous and hot and therefore very hard to detect in the x - rays .
the current best evidence for the existence of the wind fluid is found in m 82 ( griffiths et al .
2000 ; stevens et al . 2003 ) , ngc 1569 ( martin , kobulnicky , & heckman 2002 ) , and possibly the milky way ( e.g. , koyama et al . 1989 ; yamauchi et al . 1990 ) .
the ratio of alpha elements to iron appears to be slightly super - solar in the winds of both ngc 1569 and m 82 , as expected if the stellar ejecta from sne ii are providing some , but not all of the wind fluid . 0.1 in * selective loss of metals .
* the outflow velocities in lirgs and ulirgs do not appear to be correlated with the rotation velocity ( or equivalently , the escape velocity ) of the host galaxy , implying selective loss of metal - enriched gas from shallower potentials ( heckman et al .
2000 ; rupke et al .
if confirmed over a broader range of galaxy masses ( e.g. , martin 1999 ; but see the contribution by martin at this conference for a word of warning ) , this result may help explain the mass - metallicity relation and radial metallicity gradients in elliptical galaxies and galaxy bulges and disks ( e.g. , bender , burstein , & faber 1993 ; franx & illingworth 1990 ; carollo & danziger 1994 ; zaritsky et al .
1994 ; trager et al .
the ejected gas may also contribute to the heating and chemical enrichment of the icm in galaxy clusters ( e.g. , dupke & arnaud 2001 ; finoguenov et al .
2002 , and references therein ) . 0.1 in * dust outflows .
* galactic winds also act as conveyor belts for the dust in the hosts .
the evidence for a large - scale dusty outflow in our own galaxy has already been mentioned in 3.1 ( bland - hawthorn & cohen 2003 ) .
far - infrared maps of external galaxies with known galactic winds show extended dust emission along the galaxy minor axis , suggestive of dust entrainment in the outflow ( e.g. , hughes , gear , & robson 1994 ; alton et al .
1998 , 1999 ; radovich , kahanp , & lemke 2001 ) .
direct evidence is also found at optical wavelengths in the form of elevated dust filaments in a few galaxies ( e.g. , ngc 1808 , phillips 1993 ; ngc 3079 , cecil et al . 2001 ) . a strong correlation between color excesses , @xmath51 , and the equivalent widths of the blueshifted low - ionization lines in star - forming galaxies at low ( e.g. , armus , heckman , & miley 1989 ; veilleux et al .
1995 ; heckman et al .
2000 ; rupke et al .
2003 ) and moderate - to - high redshifts ( e.g. , rupke et al . 2003 ; shapley et al . 2003 ) provides additional support for the prevalence of dust outflows . assuming a galactic dust - to - gas ratio , heckman et al .
( 2000 ) estimate that the dust outflow rate is about 1% of the total mass outflow rate in lirgs .
dust ejected from galaxies may help feed the reservoir of intergalactic dust ( e.g. , coma cluster ; stickel et al .
1998 ) .
the impact of galactic winds on the host galaxies and the environment depends sensitively on the size of the `` zone of influence '' of these winds , i.e. the region affected either directly ( e.g. , heating , metals ) or indirectly ( e.g. , ionizing radiation ) by these winds . this section summarizes the methods used to estimate this quantity . 0.1 in * indirect measurements based on estimates of the escape velocity .
* the true extent of galactic winds is difficult to determine in practice due to the steeply declining density profile of both the wind material and the host ism .
the zone of influence of galactic winds is therefore often estimated using indirect means which rely on a number of assumptions .
a popular method is to use the measured velocity of the outflow and compare it with the local escape velocity derived from some model for the gravitational potential of the host galaxy .
if the measured outflow velocity exceeds the predicted escape velocity _ and _ if the halo drag is negligible , then the outflowing material is presumed to escape the host galaxy and be deposited in the igm on scales @xmath0 50 100 kpc .
this method was used for instance by rupke et al .
( 2002 ) to estimate the average escape fraction @xmath52 and `` ejection efficiency '' @xmath53 for 12 ulirgs , which were found to be @xmath54 and @xmath55 , respectively .
these calculations assumed that the host galaxy could be modeled as a singular isothermal sphere truncated at some radius @xmath56 .
neither the escape fraction nor the ejection efficiency were found to be sensitive to the exact value of @xmath56 .
other strong cases for escaping material include the `` h@xmath41 cap '' of m 82 and the @xmath11 1500 km s@xmath4 line - emitting material in the superbubble of ngc 3079 ; both objects were discussed in 3.2 .
note that the outflow velocities measured by rupke et al .
( 2002 ) refer to the neutral component of the outflow , not the hot enriched wind fluid .
unfortunately , direct measurements of the wind velocity are not yet technically possible so one generally relies on the expected terminal velocity of an adiabatic wind at the measured x - ray temperature @xmath57 [ @xmath58 , where @xmath59 is the mean mass per particle ] to provide a lower limit to the velocity of the wind fluid ( this is a lower limit because it only takes into account the thermal energy of this gas and neglects any bulk motion ; e.g. , chevalier & clegg 1985 ; martin 1999 ; heckman et al . 2000 ) .
arguably the single most important assumption made to determine the fate of the outflowing gas is that halo drag is negligible .
silich & tenorio - tagle ( 2001 ) have argued that halo drag may severely limit the extent of the wind and the escape fraction .
drag by a dense halo or a complex of tidal debris may be particularly important in ulirgs if they are created by galaxy interactions ( e.g. , veilleux , kim , & sanders 2002b ) . 0.1 in * deep x - ray and optical maps of local starbursts . * the fundamental limitation in directly measuring the zone of influence of winds is the sensitivity of the instruments .
fortunately , @xmath38 and @xmath37-newton now provide powerful tools to better constrain the extent of the hot medium ( e.g. , m 82 , stevens et al .
2003 ; ngc 3079 , cecil et al . 2002 ; ngc 6240 , komossa et al . 2003 ; veilleux et al . 2003 ; ngc 1511 , dahlem et 2003 ) .
the reader should refer to the contribution of m. ehle at this conference for a summary of recent x - ray results ( see also strickland et al .
2003 and references therein ) .
the present discussion focusses on optical constraints derived from the detection of warm ionized gas on the outskirts of wind hosts .
progress in this area of research has been possible thanks to advances in the fabrication of low - order fabry - perot etalons which are used as tunable filters to provide monochromatic images over a large fraction of the field of view of the imager . the central wavelength ( 3500
1.0 @xmath59 m ) is tuned to the emission - line feature of interest and the bandwidth ( 10 100 ) is chosen to minimize the sky background . continuum and emission - line images are produced nearly simultaneously thanks to a `` charge shuffling / frequency switching '' mode , where the charges are moved up and down within the detector at the same time as switching between two discrete frequencies with the tunable filter , therefore averaging out temporal variations associated with atmospheric lines and transparency , seeing , instrument and detector instabilities .
the narrow - band images are obtained in a straddle mode , where the off - band image is made up of a pair of images that `` straddle '' the on - band image in wavelength ( e.g. , @xmath60 = 6500 and @xmath61 = 6625 for rest - frame h@xmath41 ) ; this greatly improves the accuracy of the continuum removal since it corrects for slopes in the continuum and underlying absorption features .
these techniques have been used with the taurus tunable filter ( ttf ; bland - hawthorn & jones 1998 ; bland - hawthorn & kedziora - chudczer 2003 ) on the aat and wht to produce emission - line images of several `` quiescent '' disk galaxies ( miller & veilleux 2003a ) and a few starburst galaxies ( veilleux et al .
2003 ) down to unprecedented emission - line fluxes . gaseous complexes or filaments larger than @xmath11 20 kpc have been discovered or confirmed in a number of wind hosts ( e.g. , ngc 1482 and ngc 6240 ; the presence of warm ionized gas at @xmath11 12 kpc from the center of m 82 was discussed in 3.2 ) .
multi - line imaging and long - slit spectroscopy of the gas found on large scale reveal line ratios which are generally not h ii region - like .
shocks often contribute significantly to the ionization of the outflowing gas on the outskirts of starburst galaxies .
as expected from shock models ( e.g. , dopita & sutherland 1995 ) , the importance of shocks over photoionization by ob stars appears to scale with the velocity of the outflowing gas ( e.g. , ngc 1482 , ngc 6240 , or eso484-g036 versus ngc 1705 ; ngc 3079 is an extreme example of a shock - excited wind nebula ; veilleux et al . 1994 ) , although other factors like the starburst age , star formation rate , and the dynamical state of the outflowing structure ( e.g. , pre- or post - blowout ) must also be important in determining the excitation properties of the gas at these large radii ( e.g. , shopbell & bland - hawthorn 1998 and veilleux & rupke 2002 ) . 0.1 in * influence of the wind on companion galaxies .
* companion galaxies located within the zone of influence of the wind will be affected by the wind ram pressure .
irwin et al . ( 1987 ) noticed that the dwarf s0 galaxy ngc 3073 exhibits an elongated h i tail that is remarkably aligned with the nucleus of ngc 3079 .
irwin et al . have argued that ram pressure due to the outflowing gas of ngc 3079 is responsible for this tail .
if that is the case , the wind of ngc 3079 must extent to at least @xmath11 50 kpc .
this is the only system known so far where this phenomenon is suspected to take place . 0.1 in * absorption - line studies .
* absorption - line spectroscopy of bright background galaxies ( e.g. , high-@xmath50 quasars , lyman break galaxies ) can provide direct constraints on the zone of influence of galactic winds .
norman et al .
( 1996 ) have used this method to estimate the extent of the wind in ngc 520 .
a strong and possibly complex mg ii , mg i , and fe ii absorption - line system was found near the systemic velocity of ngc 520 at a distance from the galactic nucleus of 24 @xmath62 kpc . a weaker system at a distance of 52 @xmath62 kpc is also possibly present .
unfortunately , ngc 520 is undergoing a tidal interaction so the absorption may arise from tidally disrupted gas rather than material in the purported wind .
norman et al . also looked for absorption - line systems associated with the wind of ngc 253 , but the proximity of this system to our own galaxy and to the line of sight to the magellanic stream makes the identification of the absorption - line systems ambiguous .
no other local wind galaxy has been studied using this technique .
large absorption - line data sets collected on high-@xmath50 galaxies provide new constraints on the zone of influence of winds in the early universe .
adelberger et al .
( 2003 ) have recently presented tantalizing evidence for a deficit of neutral hydrogen clouds within a comoving radius of @xmath11 0.5 @xmath62 mpc from @xmath63 lbgs .
the uncertainties are large and the results are significant at less than the @xmath11 2@xmath64 level .
adelberger et al .
( 2003 ) argue that this deficit , if real , is unlikely to be due solely to the ionizing radiation from lbgs ( e.g. , steidel et al .
2001 ; giallongo et al .
they favor a scenario in which the winds in lbgs directly influence the surrounding igm .
they also argue that the excess of absorption - line systems with large civ column densities near lbgs is evidence for chemical enrichment of the igm by the lbg winds .
tremendous progress has been made over the past five years in understanding the physics and impact of starburst - driven winds in the local and distant universe .
we now know that starburst - driven winds are common in galaxies with high star formation rates per unit area , both locally ( nearby starbursts , luminous and ultraluminous infrared galaxies ) and at high redshifts ( e.g. , lyman break galaxies ) .
there is _ direct _ evidence that starburst - driven winds have had a strong influence on the chemical evolution of the host ism and possibly also that of the igm : ( 1 ) enriched wind fluid has been detected in a few nearby galaxies .
( 2 ) approximately half of the outflowing material in powerful starburst galaxies ( ulirgs ) have velocities in excess of the escape velocities . ( 3 ) deep emission - line maps of local wind galaxies indicate that the zone of influence of the wind often extends beyond @xmath11 10 kpc .
( 4 ) recent results on lyman break galaxies suggest tentatively that the zone of influence of lbg winds may extent out to @xmath11 500 @xmath62 kpc .
nevertheless , much work remains to be done in this area of research ; the next five years promise to be equally exciting as the last five ! it is a pleasure to thank s. aalto for organizing a stimulating conference .
some of the results presented in this paper are part of a long - term effort involving many collaborators , including j. bland - hawthorn , g. cecil , p. l. shopbell , and r. b. tully and maryland graduate students s. t. miller and d. s. rupke .
this article was written while the author was on sabbatical at the california institute of technology and the observatories of the carnegie institution of washington ; the author thanks both of these institutions for their hospitality .
the author acknowledges partial support of this research by a cottrell scholarship awarded by the research corporation , nasa / ltsa grant nag 56547 , and nsf / career grant ast-9874973 .
adelberger , k. l. , et al .
2003 , , 584 , 45 ajiki , m. , et al .
2002 , , 576 , l25 alton , p. b. , davies , j. i. , & bianchi , s. 1999 , , 343 , 51 alton , p. b. , et al .
1998 , , 507 , l125 armus , l. , heckman , t. , & miley , g. 1989 , , 347 , 727 baan , w. a. , & irwin , j. a. 1995 , , 446 , 602 bender , r. , burstein , d. , & faber , s. m. 1993 , , 411 , 153 bland - hawthorn , j. , & cohen , m. 2003 , , 582 , 246 bland - hawthorn , j. , & jones , d. h. 1998 , pasa , 15 , 44 bland - hawthorn , j. & kedziora - chudczer , l. 2003 , pasa , in press ( astro - ph/0305032 ) carollo , c. m. , & danziger , i. j. 1994 , , 270 , 523 cecil , g. , bland - hawthorn , j. , & veilleux , s. 2002 , , 576 , 745 cecil , g. , et al .
2001 , , 555 , 338 chevalier , r. a. , & clegg , a. w. 1985 , nature , 317 , 44 dahlem , m. , et al .
2003 , , 403 , 547 dawson , s. , et al .
2002 , , 570 , 92 dercole , a. , & brighenti , f. , 1999 , , 309 , 941 devine , d. , & bally , j. 1999 , , 510 , 197 dopita , m. a. , & sutherland , r. s. 1995 , , 455 , 468 dupke , r. a. , & arnaud , k. a. 2001 , , 548 , 141 ferrire , k. 1998 , , 503 , 700 ferrire , k. m. , maclow , m .- m .
, & zweibel , e. g. , 375 , 239 finoguenov , a. , et al .
2002 , , 381 , 21 franx , m. , & illingworth , g. 1990 , , 359 , l41 franx , m. , et al .
1997 , , 486 , l75 frye , b. , broadhurst , t. , & benitez , n. 2002 , , 568 , 558 garcia - burillo , s. , et al .
2001 , , 563 , l27 giallongo , e. , et al .
2002 , , 568 , l9 gonzalez delgado , r. m. , et al .
1998 , , 495 , 698 griffiths , r. e. , et al .
2000 , science , 290 , 1325 heckman , t. m. 2002 , in asp conf .
254 , extragalactic gas at low redshift , eds .
j. mulchaey and j. stocke ( san francisco : asp ) , 292 heckman , t. m. , armus , l. , & miley , g. k. 1990 , , 74 , 833 heckman , t. m. , et al .
2000 , , 129 , 493 . 2001 , , 554 , 1021 hughes , d. h. , gear , w. k. , & robson , e. i. 1994 , , 270 , 641 hunter , d. a. , & gallagher , j. s. iii 1990 , , 362 , 480 . 1997 , , 475 , 65 irwin , j. a. , & sofue , y. 1992 , , 396 , l75 irwin , j. a , et al .
1987 , , 313 , l91 israel , f. p. , et al .
1998 , , 336 , 433 koda , j. , et al .
2002 , , 573 , 105 komossa , st .
, et al . 2003 , , 582 , l15 kormendy , j. , & gebhardt , k. 2001 , in the 20th texas symposium on relativistic astrophysics , eds . j. craig wheeler and hugo martel , p.363 koyama , k. , et al .
1989 , nature , 339 , 603 kunth , d. , et al .
1998 , , 334 , 11 lehnert , m. d. , & heckman , t. m. 1995 , , 97 , 89 . 1996 , , 462 , 651 lehnert , m. d. , heckman , t. m. , & weaver , k. a. 1999 , , 523 , 575 maclow , m .- m . , & ferrara , a. 1999 , , 513 , 142 maclow , m .-
m . , & mccray , r. 1988 , , 324 , 776 marlowe , a. t. , et al .
1995 , , 438 , 285 martin , c. l. 1998 , , 506 , 222 . 1999 , , 513 , 156 martin , c. l. , kobulnicky , h. a. , & heckman , t. m. 2002 , , 574 , 663 mckeith , c. d. , et al .
1995 , , 293 , 703 meurer , g. r. , et al .
1992 , , 103 , 60 miller , s. t. , & veilleux , s. 2003a , apjs , 148 , 000 . 2003b , apj , 592 , 79 mineshige , s. , shibata , k. , & shapiro , p. 1993
, , 409 , 663 norman , c. a. , et al .
1996 , , 472 , 73 pettini , m. , et al .
2000 , , 528 , 96 . 2001 , , 554 , 981 . 2002 , , 569 , 742 phillips , a. c. 1993 , , 105 , 486 radovich , m. , kahanp , j. , & lemke , d. 2001 , , 377 , 73 rupke , d. s , veilleux , s. , & sanders , d. b. 2002 , , 570 , 588 shapley , a. e. , et al .
2003 , , 588 , 65 shopbell , p. l. , & bland - hawthorn , j. 1998 , , 493 , 129 silich , s. a. , & tenorio - tagle , g. 2001 , , 552 , 91 silich , s. tenorio - tagle , g. , muoz - tun , c. 2003 , , 590 , 791 slavin , j. d. , & cox , d. p. 1992
, , 392 , 131 sofue , y. 2000 , , 540 , 224 steidel , c. c. , pettini , m. , & adelberger , k. l. 2001 , apj , 546 , 665 stevens , i. r. , read , a. m. , & bravo - guerrero , j. 2003 , , preprint ( astro - ph/0306334 ) stickel , m. , et al .
1998 , , 329 , 55 strelnitskii , v. s. , & sunyaev , r. 1973 , soviet astron .
, 16 , 579 strickland , d. k. 2002 , in asp conf .
253 , chemical enrichment of the intracluster and intergalactic medium , eds .
r. fusco - femiano and f. matteucci ( san francisco : asp ) , 387 strickland , d. k. , ponman , t. j. , & stevens , i. r. 1997 , , 320 , 378 strickland , d. k. , & stevens , i. r. 2000 , , 314 , 511 strickland , d. k. , et al .
2000 , , 120 , 2965 . 2002 , , 568 , 689 . 2003 , apjs , preprint ( astro - ph/0306592 ) suchkov , a. , et al .
1994 , , 430 , 511 suchkov , a. , et al .
1996 , , 463 , 528 tenorio - tagle , g. , et al .
1999 , , 309 , 332 thornton , k. , et al .
1998 , , 500 , 95 tomisaka , k. 1990 , , 361 , l5 trager , s. c. , et al .
1998 , , 116 , 1 veilleux , s. , kim , d .-
, & sanders , d. b. 2002b , , 143 , 315 veilleux , s. , & rupke , d. s. 2002 , , 565 , l63 veilleux , s. , et al .
1994 , , 433 , 48 . 1995 , , 98 , 171 . 2002a , rmxac , 13 , 222 . 2003 , , in press ( astro - ph/0308330 ) walter , f. , weiss , a. , & scoville , n. 2002 , , 580 , l21 weaver , r. , et al . 1977 , , 218 , 377 yamauchi , s. , et al .
1990 , , 365 , 532 yun , m. s. , ho , p. t. p. , & lo , k. y. 1994 , nature , 372 , 530 zaritsky , d. , kennicutt ,
r. c. jr . , & huchra , j. p. 1994
, , 420 , 87 | multiwavelength data on star - forming galaxies provide strong evidence for large - scale galactic winds in both nearby and distant objects . the results from recent ground - based and space - borne programs are reviewed .
the impact of these winds on the host galaxies and the surrounding environment is discussed in the context of galaxy evolution .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
throughout their development , information theory , and more generally , probability theory , have benefitted from non - negative measures of dissimilarity , or loosely speaking , distances , between pairs of probability measures defined on the same measurable space ( see , e.g. , @xcite ) .
notable among those measures are ( see section [ sec : preliminaries ] for definitions ) : * total variation distance @xmath3 ; * relative entropy @xmath4 ; * @xmath1-divergence @xmath5 ; * hellinger divergence @xmath6 ; * rnyi divergence @xmath7 .
it is useful , particularly in proving convergence results , to give bounds of one measure of dissimilarity in terms of another .
the most celebrated among those bounds is pinsker s inequality : . ]
@xmath8 proved by csiszr @xcite and kullback @xcite , with kemperman @xcite independently a bit later . improved and generalized versions of pinsker s inequality have been studied , among others , in @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite . in this work , we find relationships among the various divergence measures outlined above as well as a number of other measures of dissimilarity between probability measures .
the framework of @xmath0-divergences , which encompasses the foregoing measures ( rnyi divergence is a one - to - one transformation of the hellinger divergence ) serves as a convenient playground .
the rich structure of the total variation distance as well as its importance in both theory and practice merits placing special attention on bounding the rest of the distance measures in terms of @xmath9 .
relationships among measures of distances between probability measures have long been a focus of interest in probability theory and statistics ( e.g. , for studying the rate of convergence of measures ) .
the reader is referred to surveys in ( * ? ? ?
* section 3 ) , ( * ? ? ?
* chapter 2 ) , @xcite and ( * ? ? ?
* appendix 3 ) , which provide several relationships among useful @xmath0-divergences and other measures of dissimilarity between probability measures .
some notable existing bounds among @xmath0-divergences include , in addition to : * ( * ? ? ?
* lemma 1 ) , @xcite @xmath10 * ( * ? ? ? * ( 2.2 ) ) @xmath11 * ( * ? ? ?
* theorem 5 ) , ( * ? ? ?
* theorem 4 ) , @xcite @xmath12 * ( * ? ? ?
* corollary 5.6 ) for all @xmath13 @xmath14 the inequality in is reversed if @xmath15 $ ] , and it holds with equality if @xmath16 . * @xcite , ( * ? ? ?
* ( 58 ) ) @xmath17 * @xcite @xmath18 * ( * ? ? ?
* ( 2.8 ) ) @xmath19 * @xcite , ( * ? ? ?
* corollary 32 ) , @xcite @xmath20 * @xcite ( cf . a generalized form in ( * ? ? ?
* lemma a.3.5 ) ) @xmath21 a generalization of is provided in ( * ? ? ?
* proposition 2.15 ) : @xmath22 * ( * ? ? ?
* proposition 2.35 ) if @xmath23 and @xmath24 , then @xmath25 + ( * ? ? ?
* theorems 3 and 16 ) the rnyi divergence @xmath7 is monotonically increasing in @xmath26 , and @xmath27 is monotonically decreasing in @xmath28 $ ] ; the same monotonicity properties also hold for @xmath6 ( * ? ? ?
* proposition 2.7 ) .
* @xcite if @xmath28 $ ] , then @xmath29 * a reverse pinsker inequality " , providing an upper bound on the relative entropy in terms of the total variation distance , does not exist in general since we can find distributions which are arbitrarily close in total variation but with arbitrarily high relative entropy . nevertheless , it is possible to introduce constraints under which such reverse pinsker inequalities hold . in the special case of a finite alphabet @xmath30 , csiszr and
talata @xcite showed that @xmath31 when @xmath32 is positive . *
* theorem 3.1 ) yields that if @xmath33 is a strictly convex function , then there exists a real - valued function @xmath34 which depends on @xmath0 such that @xmath35 and and @xmath36 . ]
@xmath37 is a sequence of pairs of probability measures , then @xmath38 the numerical optimization of an @xmath0-divergence subject to simultaneous constraints on @xmath39-divergences @xmath40 was recently studied in @xcite , which showed that for that purpose it is enough to restrict attention to alphabets of cardinality @xmath41 .
earlier , @xcite showed that if @xmath42 , then either the solution is obtained by a pair @xmath43 on a binary alphabet , or it is a deflated version of such a point .
therefore , from a purely numerical standpoint , the minimization of @xmath44 such that @xmath45 can be accomplished by a grid search on @xmath46 ^ 2 $ ] .
occasionally , as in the case where @xmath47 and @xmath48 , it is actually possible to determine analytically the locus of @xmath49 ( see @xcite ) .
in fact , as shown in ( * ? ? ?
* ( 22 ) ) , a binary alphabet suffices if the single constraint is on the total variation distance .
the same conclusion holds when minimizing the rnyi divergence @xcite .
the rest of the paper is structured as follows : section [ sec : preliminaries ] introduces the basic definitions needed and in particular the various measures of dissimilarity between probability measures used throughout .
based on functional domination , section [ sec : functional domination ] provides a basic tool for the derivation of bounds among @xmath0-divergences . under mild regularity conditions , this approach further enables to prove the optimality of constants in those bounds .
we also show instances where such optimality can be shown in the absence of regularity conditions .
the basic tool used in section [ sec : functional domination ] is exemplified in obtaining relationships among important @xmath0-divergences such as @xmath50 and @xmath9 .
this approach is also useful in providing an alternative proof of samson s inequality @xcite ( an analog to pinsker s inequality , useful in proving certain concentration of measure results @xcite ) , whose constant we show can not be improved . section [ sec : bounded ] provides an approach for bounding @xmath0-divergence in terms of @xmath51-divergence , assuming that the relative information is lower and/or upper bounded with probability one .
the approach is exemplified in bounding @xmath52 , @xmath53 , @xmath54 , and the local behavior of @xmath0-divergence ratios .
we also show that bounded relative information leads to a strengthened version of jensen s inequality , which , in turn , results in upper and lower bounds on the ratio of the non - negative difference @xmath55 to @xmath56 .
we also show a reverse version of samson s inequality .
section [ sec : tv - ri ] gives several useful identities linking the total variation distance with the relative information spectrum , which result in a number of upper and lower bounds on @xmath9 , some of which are tighter than pinsker s inequality in various ranges of the parameters .
it also provides refined bounds on @xmath57 as a function of @xmath1 divergences and the total variation distance .
section [ sec : reversep ] is devoted to proving
reverse pinsker inequalities , " namely , lower bounds on @xmath58 as a function of @xmath59 involving either ( a ) bounds on the relative information , ( b ) lipschitz constants , or ( c ) the minimum mass of the reference measure ( in the finite alphabet case ) . in the latter case
, we also examine the relationship between entropy and the total variation distance from the equiprobable distribution , as well as the exponential decay of the probability that an independent identically distributed sequence is not strongly typical .
section [ sec : eg ] focuses on the @xmath2 divergence .
this @xmath0-divergence generalizes the total variation distance , and its utility in information theory has been exemplified in @xcite .
section [ sec : eg ] starts by providing an integral representation of @xmath0-divergences as a function of the @xmath2 divergence ; this representation shows that @xmath60 uniquely determines @xmath61 and @xmath6 , as well as any other @xmath0-divergence with twice differentiable @xmath0 .
in addition , we show an extension of pinsker s inequality to @xmath2 divergence , which leads to a relationship between the relative information spectrum and relative entropy .
section [ sec : rd ] expresses the rnyi divergence in terms of the relative information spectrum , and gives upper and lower bounds on the rnyi divergence in terms of either the variational distance or relative entropy .
particular attention is placed on the finite alphabet case .
we assume throughout that the probability measures @xmath62 and @xmath63 are defined on a common measurable space @xmath64 , and @xmath65 denotes that @xmath62 is _ absolutely continuous _ with respect to @xmath63 , namely there is no event @xmath66 such that @xmath67 .
[ def : ri ] if @xmath65 , the _ relative information _ provided by @xmath68 according to @xmath43 is given by denotes the radon - nikodym derivative ( or density ) of @xmath62 with respect to @xmath63 .
logarithms have an arbitrary common base , and the exponent indicates the inverse function of the logarithm with that base . ]
@xmath69 when the argument of the relative information is distributed according to @xmath62 , the resulting real - valued random variable is of particular interest .
its cumulative distribution function and expected value are known as follows .
[ def : ris ] if @xmath65 , the _ relative information spectrum _ is the cumulative distribution function @xmath70,\end{aligned}\ ] ] with means that @xmath71 = p ( { \ensuremath{\mathcal}}{f } ) $ ] for any event @xmath66 . ]
@xmath72 . furthermore , if @xmath73 , then the relative entropy of @xmath62 with respect to @xmath63 is @xmath74 \\
\label{eq2:re1 } & = \mathbb{e } \bigl [ \imath_{p \| q}(y ) \ ,
\exp\bigl(\imath_{p \| q}(y ) \bigr ) \bigr].\end{aligned}\ ] ] introduced by ali - silvey @xcite and csiszr ( @xcite ) , a useful generalization of the relative entropy , which retains some of its major properties ( and , in particular , the data processing inequality @xcite ) , is the class of @xmath0-divergences . a general definition of an @xmath0-divergence is given in @xcite , specialized next to the case where @xmath65 .
[ def : fd ] let @xmath33 be a convex function , and suppose that @xmath65 .
the _ @xmath0-divergence _ from @xmath62 to @xmath63 is given by @xmath75\end{aligned}\ ] ] with @xmath76 in , we take the continuous extension implies its continuity on @xmath77 . ]
@xmath78.\end{aligned}\ ] ] if @xmath79 and @xmath80 denote , respectively , the densities of @xmath62 and @xmath63 with respect to a @xmath81-finite measure @xmath82 ( i.e. , @xmath83 , @xmath84 ) , then we can write as @xmath85 [ remark : equivalence - fd ] different functions may lead to the same @xmath0-divergence for all @xmath43 : if for an arbitrary @xmath86 , we have @xmath87 then @xmath88 the following key property of @xmath0-divergences follows from jensen s inequality .
[ prop : fgibbs ] if @xmath33 is convex and @xmath89 , @xmath65 , then @xmath90 if , furthermore , @xmath0 is strictly convex at @xmath91 , then equality in holds if and only if @xmath92 . the reader is referred to @xcite for surveys on general properties of @xmath0-divergences , and also to the textbook by liese and vajda @xcite which provides a systematic and rigorous study of @xmath0-divergences .
the assumptions of proposition [ prop : fgibbs ] are satisfied by many interesting measures of dissimilarity between probability measures . in particular
, we give the following examples which receive particular attention in this paper . as per definition
[ def : fd ] , in each case the function @xmath0 is defined on @xmath77 . 1 . _ relative entropy _
@xcite : @xmath93 , @xmath94 with @xmath95 defined as @xmath96 2 . _ relative entropy _ : ( @xmath97 ) @xmath98 , @xmath99 3 .
_ jeffrey s symmetrized divergence _
@xcite : ( @xmath100 ) @xmath101 , @xmath102 4 .
_ @xmath1-divergence _ @xcite : @xmath103 or @xmath104 , @xmath105 consequently , the @xmath1-divergence can be expressed in the following equivalent forms : @xmath106 - 1 \\ \label{eq : chi - square 4 } & = \mathbb{e } \bigl [ \exp\bigl(\imath_{p\|q}(x)\bigr ) \bigr ] - 1\end{aligned}\ ] ] with @xmath72 and @xmath73 .
note that if @xmath107 , then from the right side of , we obtain @xmath108 with @xmath109 .
_ hellinger divergence of order @xmath110 _ @xcite , ( * ? ? ?
* definition 2.10 ) ( a.k.a .
tsallis divergence , e.g. , ( * ? ? ?
* ( 49 ) ) ) : @xmath111 with @xmath112 the @xmath1-divergence is the hellinger divergence of order 2 , while @xmath113 is usually referred to as the _ squared hellinger distance_. the analytic extension of @xmath6 at @xmath114 yields @xmath115 6 . _
total variation distance _ : setting @xmath116 results in @xmath117 \label{eq3 : tv distance } & = 2 \ , \sup_{{\ensuremath{\mathcal}}{f } \in \mathscr{f } } \bigl(p({\ensuremath{\mathcal}}{f } ) - q({\ensuremath{\mathcal}}{f})\bigr).\end{aligned}\ ] ] 7 . _ triangular discrimination ( a.k.a .
the vincze - le cam distance ) _
@xcite : @xmath118 with @xmath119 note that @xmath120 8 .
_ jensen - shannon divergence ( a.k.a .
capacitory discrimination ) _
@xcite : @xmath121 with @xmath122 9 . _
@xmath2 divergence _
( see , e.g. , @xcite ) : for @xmath123 , @xmath124 with @xmath125 where @xmath126 .
@xmath2 is sometimes called hockey - stick divergence " because of the shape of @xmath127 . if @xmath128 , then @xmath129 10 .
_ degroot statistical information _
@xcite : for @xmath130 , @xmath131 with @xmath132 note that , from , it follows that ( cf .
* ( 77 ) ) ) @xmath133 + this measure was first proposed by degroot @xcite due to its operational meaning in bayesian statistical hypothesis testing ( see section [ saens ] ) , and it was later identified as an @xmath0-divergence ( see ( * ? ? ?
* theorem 10 ) ) .
_ marton s divergence _
@xcite : @xmath134 \right]\end{aligned}\ ] ] where the minimum is over all probability measures @xmath135 with respective marginals @xmath136 and @xmath137 . from @xcite @xmath138 with @xmath139 note
that marton s divergence satisfies the triangle inequality ( * ? ? ?
* lemma 3.1 ) , and @xmath140 implies @xmath92 ; however , due to its asymmetry , it is not a distance measure .
another generalization of relative entropy was introduced by rnyi @xcite in the special case of finite alphabets .
the general definition ( assuming @xmath65 ) is the following .
[ def : rd ] let @xmath65 .
rnyi divergence of order @xmath141 _ from @xmath62 to @xmath63 is given as follows : * if @xmath142 , then @xmath143 \bigr ) \\[0.1 cm ] \label{eq : rd3 } & = \frac1{\alpha-1 } \ ; \log \bigl ( \mathbb{e } \bigl [ \exp \bigl((\alpha-1 ) \ , \imath_{p\|q}(x)\bigr)\bigr ] \bigr)\end{aligned}\ ] ] with @xmath72 and @xmath73 .
* if @xmath144 , then .
rnyi @xcite defined @xmath145 , while we have followed @xcite defining it instead as @xmath146 . ]
@xmath147 * if @xmath148 , then @xmath149 which is the analytic extension of @xmath150 at @xmath114 ( if @xmath151 , it can be verified by lhpital s rule that @xmath152 ) .
* if @xmath153 then @xmath154 with @xmath73 .
rnyi divergence is a one - to - one transformation of hellinger divergence of the same order @xmath155 : @xmath156 which , when particularized to order 2 becomes @xmath157 note that , , follow from and the monotonicity of the rnyi divergence in its order , which in turn yields . the bhattacharyya distance has been introduced in @xcite , and it has been popularized in the engineering literature in @xcite . [ definition : b distance ] the bhattacharyya distance between @xmath62 and @xmath63 , denoted by @xmath158 , is given by @xmath159 note that , if @xmath107 , then @xmath160 and @xmath161 if and only if @xmath92 , though @xmath162 does not satisfy the triangle inequality .
let @xmath0 and @xmath51 be convex functions on @xmath77 with @xmath163 , and let @xmath62 and @xmath63 be probability measures defined on a measurable space @xmath64 .
if , for @xmath26 , @xmath164 for all @xmath165 then , it follows from the definition of an @xmath0-divergence that @xmath166 this simple observation leads to a proof of , for example , and the left inequality in with the aid of remark [ remark : equivalence - fd ] .
we start this section by proving a general result , which will be helpful in proving various tight bounds among @xmath0-divergences . [
theorem : tight bound ] let @xmath65 , and let @xmath0 , @xmath51 be convex and non - negative functions on @xmath77 with @xmath163 . denote the function with domain @xmath77 @xmath167 and @xmath168 then , a. @xmath169 b. [ theorem : tight bound : partb ] if , in addition , @xmath51 is strictly convex at 1 , and @xmath170 , then @xmath171 a. the bound in follows from and @xmath172 for all @xmath173 . b. since @xmath51 is strictly convex at 1 , proposition [ prop : fgibbs ] implies that @xmath174 if @xmath175 .
the convexity of @xmath176 on @xmath77 implies their continuity ; since @xmath51 is non - negative on @xmath77 and strictly convex at 1 with @xmath177 then @xmath178 for all @xmath179 , which implies that @xmath180 in and is well defined , continuous on @xmath181 , and right continuous at 1 .
+ to show , we fix an arbitrary @xmath182 and construct a sequence of pairs of probability measures whose ratio of @xmath0-divergence to @xmath51-divergence converges to @xmath183 . to that end , for sufficiently small @xmath184 , let @xmath185 and @xmath186 be parametric probability measures defined on the set @xmath187 with @xmath188 and @xmath189 .
then , @xmath190 & = \lim_{\varepsilon \to 0 } \ , \frac{\varepsilon \ , f(\nu ) + ( 1-\varepsilon ) \ , f\left(\frac{1-\nu \varepsilon}{1-\varepsilon}\right)}{\varepsilon \ , g(\nu ) + ( 1-\varepsilon ) \ , g\left(\frac{1-\nu \varepsilon}{1-\varepsilon}\right ) } \\[0.1 cm ] & = \lim_{\alpha \to 0 } \frac{f(\nu ) + \frac{\nu -1}{\alpha } \ ; f(1 - \alpha)}{g(\nu ) + \frac{\nu -1}{\alpha } \ ; g(1-\alpha ) } \label{arlington } \\[0.1 cm ] & = \kappa(\nu ) \label{cun}\end{aligned}\ ] ] where holds by change of variable @xmath191 , and holds by the assumption on the derivatives of @xmath0 and @xmath51 at 1 . if @xmath192 we are done .
if the supremum in is not attained on @xmath193 ( e.g. , it is equal to either @xmath194 or @xmath195 ) , then due to the continuity of @xmath196 on @xmath181 and right continuity at 1 , the right side of can be made arbitrarily close to @xmath197 by an appropriate choice of @xmath198 . beyond the restrictions in proposition
[ prop : fgibbs ] , the only operative restriction imposed by theorem [ theorem : tight bound][theorem : tight bound : partb ] ) is the differentiability of the functions @xmath0 and @xmath51 at @xmath91 .
indeed , we can invoke remark [ remark : equivalence - fd ] and add @xmath199 to @xmath200 , without changing @xmath201 ( and likewise with @xmath51 ) and thereby satisfying the condition in theorem [ theorem : tight bound][theorem : tight bound : partb ] ) ; the stationary point at 1 must be a minimum of both @xmath0 and @xmath51 because of the assumed convexity , which implies their non - negativity on @xmath77 .
it is useful to generalize theorem [ theorem : tight bound][theorem : tight bound : partb ] ) by dropping the assumption on the existence of the derivatives at 1 . to that end , note that the inverse transformation used for the transition to is given by @xmath202 where @xmath184 is sufficiently small , so if @xmath203 ( resp .
@xmath204 ) , then @xmath26 ( resp .
@xmath205 ) .
consequently , it is easy to see from that if @xmath206 , the construction in the proof can restrict to @xmath203 , in which case it is enough to require that the left derivatives of @xmath0 and @xmath51 at 1 be equal to @xmath207 . analogously , if @xmath208 , it is enough to require that the right derivatives of @xmath0 and @xmath51 at 1 be equal to @xmath207 .
since the rnyi divergence of order @xmath210 is monotonically increasing in @xmath211 , yields @xmath212 inequality , which can be found in @xcite and ( * ? ? ?
* theorem 5 ) , is sharpened in theorem [ thm : d - chi ] under the assumption of bounded relative information . in view of ,
an alternative way to sharpen is @xmath213 for @xmath214 , which is tight as @xmath215 .
relationships between the relative entropy , total variation distance and @xmath1 divergence are derived next .
a. if @xmath65 and @xmath216 , then @xmath217 holds if @xmath218 and @xmath219 .
furthermore , if @xmath220 then @xmath221 is optimal , and if @xmath222 then @xmath223 is optimal .
b. if @xmath107 and @xmath175 , then @xmath224 and the constant in the right side of is the best possible .
a. the satisfiability of with @xmath218 is equivalent to . + let @xmath65 , and @xmath73 . then , @xmath225 \\ \label{eq2 : improved diaconiss96 } & \leq \tfrac12 \ , \mathbb{e } \left [ \left ( 1 - \frac{\text{d}p}{\text{d}q } \ , ( y ) \right)^+ + \left ( \frac{\text{d}p}{\text{d}q } \ , ( y ) - 1 \right)^2 \right ] \ , \log e \\
\label{eq3 : improved diaconiss96 } & = \left ( \tfrac14 \ , |p - q| + \tfrac12 \ , \chi^2(p\|q ) \right ) \ , \log e\end{aligned}\ ] ] where follows from the definition of relative entropy with the function @xmath226 defined in ; holds since for @xmath165 @xmath227 \ , \log e\end{aligned}\ ] ] and follows from , , and the identity @xmath228 $ ] .
this proves with @xmath219 .
+ next , we show that if @xmath220 then @xmath229 is the best possible constant in . to that end , let @xmath230 , and let @xmath231 .
it can be verified that the function @xmath196 is monotonically decreasing on @xmath77 , so @xmath232 since @xmath233 and @xmath234 , and @xmath235 , the desired result follows from theorem [ theorem : tight bound][theorem : tight bound : partb ] ) .
+ to show that @xmath223 is the best possible constant in if @xmath222 , we let and @xmath236 $ ] .
theorem [ theorem : tight bound][theorem : tight bound : partb ] ) does not apply here since @xmath237 is not differentiable at @xmath91 .
however , we can still construct probability measures for proving the optimality of the point @xmath238 . to that end , let @xmath239 , and define probability measures @xmath240 and @xmath241 on the set @xmath242 with @xmath243 and @xmath244 . since @xmath245 and @xmath246 , @xmath247 where holds since @xmath248 and the result follows since , in the right side of , @xmath249 .
b. we have @xmath250 with @xmath251 where is easy to verify since @xmath196 is monotonically increasing on @xmath252 , and monotonically decreasing on @xmath253 .
the desired result follows since the conditions of theorem [ theorem : tight bound][theorem : tight bound : partb ] ) apply .
inequality strengthens the bound in ( * ? ? ? * ( 2.8 ) ) , @xmath254 note that the short outline of the suggested proof in @xcite leads to the weaker upper bound @xmath255 nats .
note that implies a looser result in where the constant in the right side of is doubled .
furthermore , pinsker s inequality and result in the bound @xmath256 which , although weaker than , has the same behavior for @xmath257 .
the next result shows three upper bounds on the vincze - le cam distance in terms of the relative entropy ( and total variation distance ) .
[ thm : re - delta - tv ] let @xmath226 be the function defined in , and let @xmath259 denote the function @xmath33 defined in .
a. if @xmath65 , then @xmath260 where @xmath261 , computed with @xmath262 furthermore , the constant @xmath263 in is the best possible .
b. if @xmath65 , then @xmath264 where @xmath265 is defined by @xmath266 furthermore , the constant @xmath267 in is the best possible .
c. if @xmath107 , then @xmath268 and the bound in is tight .
a. let @xmath269 and @xmath270 .
these functions satisfy the conditions in theorem [ theorem : tight bound][theorem : tight bound : partb ] ) , and result in the function @xmath196 defined in .
the function @xmath196 is maximal at @xmath271 with @xmath272 . since @xmath273 and @xmath274 , theorem [ theorem : tight bound][theorem : tight bound : partb ] ) results in with the optimality of its constant @xmath263 .
b. by the definition of @xmath267 in , it follows that for all @xmath275 @xmath276 note that holds for all @xmath277 ( i.e. , not only in @xmath252 as in ) since , for @xmath278 , is equivalent to @xmath279 which indeed holds for all @xmath280 .
hence , the bound in follows from , , and . it can be verified from that @xmath265 , the maximal value of @xmath281 on @xmath282 is attained at @xmath283 and @xmath284 .
the optimality of the constant @xmath267 in is shown next ( note that theorem [ theorem : tight bound][theorem : tight bound : partb ] ) can not be used here , since the right side of is not differentiable at @xmath91 ) .
let @xmath185 and @xmath186 be probability measures defined on @xmath187 with @xmath285 and @xmath189 for @xmath239 .
then , it follows that @xmath286 \label{eq d2 : re - delta - tv } & = \lim_{\varepsilon \to 0 } \frac{f_{\delta}(t^\star ) \ , \varepsilon \ , \log e + o(\varepsilon)}{\bigl [ r(t^\star ) + 2c_2 \ , ( 1-t^\star ) \ , \log e \bigr ] \ , \varepsilon + o(\varepsilon ) } \\
\label{eq d3 : re - delta - tv } & = \kappa_{c_2}(t^\star ) = 1\end{aligned}\ ] ] where follows by the construction of @xmath185 and @xmath186 , and the use of taylor series expansions ; follows from ( note that @xmath287 ) , which yields the required optimality result in .
c. let @xmath269 and @xmath288 be defined by @xmath289 for all @xmath275 .
these functions , which satisfy the conditions in theorem [ theorem : tight bound][theorem : tight bound : partb ] ) , yield @xmath290 and @xmath291 . theorem [ theorem : tight bound][theorem : tight bound : partb ] ) yields the desired result with @xmath292 we can generalize and to @xmath293 where @xmath216 and @xmath107 . in view of theorem [
theorem : tight bound ] , the locus of the allowable constant pairs @xmath294 in can be evaluated . to that end ,
theorem [ theorem : tight bound][theorem : tight bound : partb ] ) can be used with @xmath295 , and @xmath296 for all @xmath173 .
this results in a convex region , which is symmetric with respect to the straight line @xmath297 ( since @xmath298 ) .
note that , and the symmetry property of this region identify , respectively , the points @xmath299 , @xmath300 and @xmath301 on its boundary ; since the sum of their coordinates is nearly constant , the boundary of this subset of the positive quadrant is nearly a straight line of slope @xmath302 .
an analog of pinsker s inequality , which comes in handy for the proof of marton s conditional transportation inequality ( * ? ? ? * lemma 8.4 ) , is the following bound due to samson ( * ? ? ?
* lemma 2 ) : [ thm : samson ] if @xmath65 , then @xmath303 where @xmath304 is the distance measure defined in .
we provide next an alternative proof of theorem [ thm : samson ] , in view of theorem [ theorem : tight bound][theorem : tight bound : partb ] ) , with the following advantages : a. this proof yields the optimality of the constant in , i.e. , we prove that @xmath305 where the supremum is over all probability measures @xmath306 such that @xmath175 and @xmath107 .
b. a simple adaptation of this proof enables to derive a reverse inequality to , which holds under the boundedness assumption of the relative information ( see section [ subsec : reverse samson s inequality ] ) .
@xmath307 where , from , @xmath308 is the convex function @xmath309 and , from and , the non - negative and convex function @xmath33 in is given by @xmath310 for all @xmath275 .
let @xmath311 be the non - negative and convex function @xmath226 defined in , which yields @xmath312 .
note that @xmath313 , and the functions @xmath0 and @xmath311 are both differentiable at @xmath91 with @xmath314 .
let @xmath315 \to { \ensuremath{\mathbb{r}}}$ ] be the continuous extension of @xmath316 .
the desired result follows from theorem [ theorem : tight bound][theorem : tight bound : partb ] ) since in this case @xmath317 as can be verified from the monotonicity of @xmath196 on @xmath252 ( increasing ) and @xmath253 ( decreasing ) . as mentioned in @xcite , samson s inequality
strengthens the pinsker - type inequality in ( * ? ? ?
* lemma 3.2 ) : @xmath318 nevertheless , similarly to our alternative proof of theorem [ thm : samson ] , one can verify that theorem [ theorem : tight bound][theorem : tight bound : partb ] ) yields the optimality of the constant in .
let @xmath33 be a convex function with @xmath89 , and let @xmath319 be given by @xmath320 for all @xmath275 . note that @xmath321 is also convex , @xmath322 , and @xmath323 if @xmath107 . by definition ,
we take @xmath324 vajda ( * ? ? ?
* theorem 2 ) showed that the range of an @xmath0-divergence is given by @xmath325 where every value in this range is attainable by a suitable pair of probability measures @xmath65 .
recalling remark [ remark : equivalence - fd ] , note that @xmath326 with @xmath327 defined in .
* lemma 11.1 ) strengthened , showing that @xmath328 note that , provided @xmath329 and @xmath330 are finite , yields a counterpart to .
next , we show that the constant in can not be improved . [
theorem : strengthened basusp ] if @xmath33 is convex with @xmath89 , then @xmath331 where the supremum is over all probability measures @xmath332 such that @xmath65 and @xmath175 . as the first step
, we give a simplified proof of ( cf .
@xcite ) . in view of remark [
remark : equivalence - fd ] , it is sufficient to show that for all @xmath173 , @xmath333 if @xmath334 , reduces to @xmath335 , which holds in view of the convexity of @xmath0 and @xmath336 .
if @xmath278 , we can readily check , with the aid of , that reduces to @xmath337 , which , in turn , holds because @xmath321 is convex and @xmath338 . for the second part of the proof of
, we construct a pair of probability measures @xmath240 and @xmath241 such that , for a sufficiently small @xmath184 , @xmath339 can be made arbitrarily close to the right side of . to that end , let @xmath340 $ ] , and let @xmath240 and @xmath241 be defined on the set @xmath341 with @xmath342 and @xmath343 .
then , @xmath344 where holds since @xmath345 , and follows from and .
csiszr ( * ? ? ?
* theorem 2 ) showed that if @xmath329 and @xmath330 are finite and @xmath65 , then there exists a constant @xmath346 which depends only on @xmath0 such that @xmath347 if @xmath348 , then is superseded by where the constant is not only explicit but is the best possible according to theorem [ theorem : strengthened basusp ] . a direct application of theorem [ theorem : strengthened basusp ] yields @xmath349 \label{eq : sup2 } & \sup_{p \neq q } \frac{\delta(p \| q)}{|p - q| } = 1 , \\[0.1 cm ] \label{eq : sup3 } & \sup_{p \neq q } \frac{\mathrm{js}(p \| q)}{|p - q| } = \log 2 , \\[0.1 cm ] \label{eq : sup4 } & \sup_{p \neq q } \frac{d_2 ^ 2(p , q)}{|p - q| } = \frac12 , \\[0.1 cm ] \label{eq : sup5 } & \sup_{p \neq q } \frac{d_2
^ 2(p , q ) + d_2 ^ 2(q , p)}{|p - q| } = 1\end{aligned}\ ] ] where the suprema in are over all @xmath65 with @xmath175 , and the supremum in is over all @xmath107 with @xmath175 . the results in , and strengthen , respectively , the inequalities in ( * ? ? ?
* proposition 2.35 ) , ( * ? ? ? * ( 11 ) ) and ( * ? ? ? * theorem 2 ) . the results in and form counterparts of .
as the following result shows , it is possible to find bounds among @xmath0-divergences without requiring a strong condition of functional domination ( see section [ sec : functional domination ] ) as long as the relative information is upper and/or lower bounded almost surely . assuming @xmath65 and @xmath73 , denote @xmath350 $ ] by @xmath351 \label{eq : beta2 } \beta_2 & \triangleq \operatorname*{ess\,inf}\frac{\text{d}p}{\text{d}q } \ , ( y)\end{aligned}\ ] ] with the convention that if the right side of is @xmath352 , then @xmath353 . note that if @xmath107 , then @xmath354 \label{eq : beta1-alt } \beta_1 & = \operatorname*{ess\,inf}\frac{\text{d}q}{\text{d}p } \ , ( y).\end{aligned}\ ] ] [ thm : fd1 ] let @xmath65 , and let @xmath33 and @xmath355 be convex functions with @xmath163 .
let @xmath180 be defined as in theorem [ theorem : tight bound ] .
then , @xmath356 denote @xmath357 for @xmath358 , and @xmath359 .
then , @xmath360 where follows from the definition of @xmath361 , the non - negativity of @xmath51 on @xmath77 , and the fact that @xmath362 has no mass outside @xmath363 $ ] . to justify , we will consider three separate cases : * if @xmath364 or @xmath365 , then the first term in the right side of is 0 and is satisfied with equality ; * if @xmath366 , @xmath367 , then @xmath368 and is trivially true ; * if @xmath366 , @xmath369 , and @xmath370 , then @xmath371 which , upon substitution in the left side of yields the desired result .
an analogous reasoning yields .
[ example : non - monotonic ] if @xmath103 and @xmath372 , we get @xmath373 [ remark : play ] in the application of theorem [ thm : fd1 ] , it is often convenient to make use of the freedom afforded by remark [ remark : equivalence - fd ] and choose the corresponding offsets such that * the non - negativity property of @xmath51 , which is required by theorem [ thm : fd1 ] , is satisfied ; * @xmath374 as a necessary condition for the finiteness of the right side of
. [ remark : tight constants ] similarly to the proof of theorem [ theorem : tight bound][theorem : tight bound : partb ] ) , under the conditions therein , one can verify that the constants in theorem [ thm : fd1 ] are the best possible among all probability measures @xmath306 with given @xmath375 ^ 2 $ ] .
if @xmath196 is monotonically increasing , then theorem [ thm : fd1 ] becomes @xmath376 in this case , if @xmath377 , sometimes it is convenient to replace @xmath377 and @xmath378 in with @xmath379 and @xmath380 $ ] at the expense of loosening the bounds . in the following examples , which illustrate important cases in which @xmath381 and @xmath382 are positive
, we exclude the trivial case where @xmath383 .
( _ gaussian distributions . _ ) [ example : gaussians ] if the variances are equal , then @xmath384 . if the means are equal and @xmath385 , then @xmath386 and @xmath366 . if the means are equal and @xmath387 , then @xmath353 and @xmath388 .
( _ shifted laplace distributions . _ ) [ example : two laplacians ] let @xmath62 and @xmath63 be the probability measures whose probability density functions ( with respect to lebesgue measure ) are , respectively , given by @xmath389 and @xmath390 with @xmath391 where @xmath392 . in this case , gives @xmath393 which yields @xmath394 ( _ cramr distributions .
_ ) [ example : two cramer distributions ] suppose that @xmath62 and @xmath63 have cramr probability density functions @xmath395 and @xmath396 , respectively , with @xmath397 where @xmath398 and @xmath399 . in this case , we have @xmath400 since the ratio of the probability density functions , @xmath401 , tends to @xmath402 in the limit where @xmath403 . in the special case where @xmath404 , the ratio of these probability density functions is @xmath405 at @xmath406 ; due also to the symmetry of the probability density functions around @xmath407 , it can be verified that in this special case @xmath408 ( _ cauchy distributions . _ ) [ example : two cauchy distributions ] suppose that @xmath62 and @xmath63 have cauchy probability density functions @xmath409 and @xmath410 , respectively , @xmath411 and @xmath412^{-1 } , \quad x \in { \ensuremath{\mathbb{r}}}\end{aligned}\ ] ] where @xmath413 . in this case
, we also have @xmath400 since the ratio of the probability density functions tends to @xmath414 in the limit where @xmath403 . in the special case where @xmath415 , @xmath416 the remaining part of this section is devoted to various applications of theorem [ thm : fd1 ] . from this point ,
we make use of the definition of @xmath95 in
. an illustrative application of theorem [ thm : fd1 ] gives upper and lower bounds on the ratio of relative entropies .
[ thm : bounds re and dual ] let @xmath107 , @xmath175 , and let @xmath350 $ ] be given in - .
let @xmath315 \to [ 0 , \infty]$ ] be the continuous extension to @xmath417 $ ] of @xmath418 defined on @xmath179 , namely , @xmath419 , @xmath420 , and @xmath421 .
then , @xmath422 for @xmath173 , let @xmath423 then @xmath424 , @xmath425 , and @xmath245 in view of .
the desired result follows from theorem [ thm : fd1 ] and the monotonicity of @xmath426 shown in appendix [ appendix : properties of kappa ] .
the following result bounds the ratio of relative entropy to hellinger divergence of an arbitrary positive order @xmath428 .
theorem [ thm : improved hausslero ] extends and strengthens a result for @xmath23 by haussler and opper ( * ? ? ?
* lemma 4 and ( 6 ) ) ( see also @xcite ) , which in turn generalizes the special case for @xmath429 obtained simultaneously and independently by birg and massart ( * ? ? ?
* ( 7.6 ) ) .
[ thm : improved hausslero ] let @xmath65 , @xmath175 , and let @xmath350 $ ] be given in -
. for @xmath110 , let @xmath430 \to [ 0 , \infty]$ ] be the continuous extension to @xmath417 $ ] of @xmath431 defined on @xmath181 , namely @xmath432 , @xmath433 , and @xmath434 is respectively @xmath352 or @xmath207 if @xmath23 or @xmath435 .
then , for @xmath23 , @xmath436 and , for @xmath437 , @xmath438 from , @xmath312 .
let , for @xmath110 , @xmath439 defined for @xmath165 . since @xmath440
then , for every @xmath110 , the function @xmath441 is monotonically decreasing on @xmath442 $ ] and monotonically increasing on @xmath443 ; hence , @xmath444 implies that @xmath441 is a non - negative function as required in theorem [ thm : fd1 ] . in view of remark [ remark : equivalence - fd ] and , we have @xmath445 . the function @xmath430 \to [ 0 , \infty]$ ] in is the continuous extension of @xmath446 . as shown in appendix [
appendix : monotonicity of kappa_alpha ] , @xmath447 is monotonically increasing on @xmath417 $ ] if @xmath23 , and it is monotonically decreasing on @xmath417 $ ] if @xmath435 . the bounds in and follow , respectively , from theorem [ thm : fd1 ] for @xmath23 and @xmath437 .
theorem [ thm : improved hausslero ] is particularly of interest for @xmath448 . in this case since , from , @xmath449 \to [ 0 , \infty]$ ] is monotonically increasing with @xmath450 , the left inequality in yields . for large arguments , @xmath451 grows logarithmically .
for example , if @xmath452 , it follows from that @xmath453 which is strictly smaller than the upper bound on the relative entropy in ( * ? ? ? * theorem 5 ) , given not in terms of @xmath381 but in terms of another more cumbersome quantity that controls the mass that @xmath362 may have at large values .
let @xmath65 and @xmath175 .
* theorem 2 ) and ( * ? ? ? * ( 11 ) ) show , respectively , @xmath457 the following result suggests a refinement in the upper bounds of and .
[ thm : supertopsoell ] let @xmath65 , @xmath175 , and let @xmath350 $ ] be given in - .
then , @xmath458 \label{eq1 :
superll2 } \frac{\delta(p\|q)}{|p - q| } & \leq \max \bigl\{\kappa_2(\beta_1^{-1 } ) , \ , \kappa_2(\beta_2 ) \bigr\}\end{aligned}\ ] ] with @xmath459 \to [ 0 , \log 2]$ ] and @xmath460 \to [ 0 , 1]$ ] defined as the continuous extensions of @xmath461 , \\[0.1 cm ]
\label{eq2 : superll2 } & \kappa_2(t ) = \frac{|t-1|}{t+1}\end{aligned}\ ] ] i.e. , @xmath462 let @xmath463 , @xmath464 denote the functions @xmath33 in , and , respectively ; these functions yield @xmath9 , @xmath465 and @xmath466 as @xmath0-divergences .
the functions @xmath467 and @xmath468 , as introduced in , are the continuous extensions to @xmath417 $ ] of @xmath469 it can be verified by and that @xmath470 for @xmath471 $ ] and @xmath472 ; furthermore , @xmath467 and @xmath468 are both monotonically decreasing on @xmath46 $ ] , and monotonically increasing on @xmath473 $ ] .
consequently , if @xmath474 $ ] and @xmath472 , @xmath475 in view of theorem [ thm : fd1 ] and remark [ remark : equivalence - fd ] , follows from , and ; follows from and .
if @xmath476 or @xmath477 , referring to an unbounded relative information @xmath478 , the right inequalities in , and , respectively coincide ( due to , , and since @xmath479 ) ; otherwise , the right inequalities in , provide , respectively , sharper upper bounds than those in , .
[ example : supertopsoe ] suppose that @xmath65 , @xmath175 and @xmath480 for all @xmath68 ( e.g. , @xmath481 and @xmath482 ) ; substituting @xmath483 in gives @xmath484 \label{eq4 : superll2 } \frac{\delta(p\|q)}{|p - q| } & \leq \tfrac13\end{aligned}\ ] ] improving the upper bounds on @xmath454 and @xmath456 which are @xmath485 and 1 , respectively , according to the right inequalities in , . for finite alphabets , ( * ? ? ?
* theorem 7 ) shows @xmath486 the following theorem extends the validity of for hellinger divergences of an arbitrary order @xmath487 and for a general alphabet , while also providing a refinement of both upper and lower bounds in when the relative information @xmath478 is bounded .
[ thm : supperll ] let @xmath65 , @xmath175 , @xmath155 , and let @xmath350 $ ] be given in - .
let @xmath488 \to [ 0 , \infty)$ ] be the continuous extension of @xmath489}{t^{\alpha}+\alpha-1-\alpha t } \end{aligned}\ ] ] i.e. , @xmath490 then , if @xmath23 , @xmath491 } \kappa_{\alpha}(\beta)\end{aligned}\ ] ] and , if @xmath437 , @xmath492 let @xmath493 and @xmath494 denote , respectively , the functions @xmath33 in and which yield the hellinger divergence , @xmath495 , and the jensen - shannon divergence , @xmath466 , as @xmath0-divergences . from
, @xmath447 is the continuous extension to @xmath417 $ ] of @xmath496 as shown in the proof of theorem [ thm : improved hausslero ] , for every @xmath110 and @xmath165 , we have @xmath497 ( i.e. , the denominator of @xmath498 in the right side of is non - negative , as required in theorem [ thm : fd1 ] ) .
it can be verified that @xmath430 \to { \ensuremath{\mathbb{r}}}$ ] has the following monotonicity properties : * if @xmath23 , there exists @xmath499 such that @xmath500 is monotonically increasing on @xmath501 $ ] , and it is monotonically decreasing on @xmath502 $ ] ; * if @xmath437 , @xmath447 is monotonically decreasing on @xmath417 $ ] .
based on these properties of @xmath500 , for @xmath474 $ ] : * if @xmath23 @xmath503 } \kappa_{\alpha}(t);\end{aligned}\ ] ] * if @xmath435 , @xmath504 in view of theorem [ thm : fd1 ] , remark [ remark : equivalence - fd ] and , the bounds in and follow respectively from and .
specializing theorem [ thm : supperll ] to @xmath505 , since @xmath506 for @xmath165 and @xmath507 achieves its global maximum at @xmath91 with @xmath508 , and imply that @xmath509 under the assumptions in example [ example : supertopsoe ] , it follows from that @xmath510 which improves the lower bound @xmath485 in the left side of .
specializing theorem [ thm : supperll ] to @xmath16 implies that ( cf . and )
@xmath511 under the assumptions in example
[ example : supertopsoe ] , it follows from that @xmath512 while without any boundedness assumption , yields the weaker upper and lower bounds @xmath513 another application of theorem [ thm : fd1 ] shows that the local behavior of @xmath0-divergences differs by only a constant , provided that the first distribution approaches the reference measure in a certain strong sense .
[ thm : local behavior fd ]
suppose that @xmath514 , a sequence of probability measures defined on a measurable space @xmath64 , converges to @xmath63 ( another probability measure on the same space ) in the sense that , for @xmath73 , @xmath515 where it is assumed that @xmath516 for all sufficiently large @xmath517 . under the assumptions on the functions @xmath0 , @xmath51 and @xmath196 in theorem [ thm : fd1 ] , we have @xmath518 and @xmath519 since @xmath89 , @xmath520 } f(\beta)\end{aligned}\ ] ] where we have abbreviated @xmath521 the condition in yields ( see appendix [ appendix : proof of 2nd condition ] ) @xmath522 hence , follows from , , , , the continuity of @xmath0 at 1 ( due to its convexity ) , and since also @xmath51 has the same properties . from - , we also have @xmath523 } \kappa(\beta ) \cdot d_g(p_n \|
q ) & \leq d_f(p_n \| q ) \\ \label{eq2 : bounds fd - pn to q } & \leq \sup_{\beta
\in [ \beta_{2,n } , \ , \beta_{1,n}^{-1 } ] } \kappa(\beta ) \cdot d_g(p_n \| q).\end{aligned}\ ] ] finally , follows from , , , and the continuity of @xmath196 . continuing with examples [ example : two laplacians ] , [ example : two cramer distributions ] and [ example : two cauchy distributions ] , it is easy to check that the conditions and are satisfied in the following cases .
[ example : convergent laplacians ] a sequence of laplacian probability density functions with common variance and converging means : @xmath524 a sequence of converging cramr probability density functions : [ example : convergent cramer pdfs ] @xmath525 [ example : convergent cauchy pdfs ] a sequence of converging cauchy probability density functions : @xmath526^{-1 } , \quad x \in { \ensuremath{\mathbb{r}}}\\ & \lim_{n \to \infty } m_n = m \in { \ensuremath{\mathbb{r}}}\\ & \lim_{n \to \infty } \gamma_n = \gamma > 0.\end{aligned}\ ] ] we will have opportunity to use the following specialization of theorem [ thm : local behavior fd ] .
[ cor : ratios ] let @xmath527 converge to @xmath63 in the sense of . then , @xmath528 and @xmath529 vanish as @xmath530 with @xmath531 the continuous extension of @xmath196 in to @xmath417 $ ] yields @xmath532 . as mentioned in section
[ section : fd ] , @xmath533 is equal to the hellinger divergence of order 2 . specializing theorem [ thm : improved hausslero ] to the case @xmath534 results in : [ cor : re and chi - square ]
let @xmath468 be the continuous extension to @xmath535 $ ] of the function @xmath536 namely , @xmath537 , @xmath538 , and @xmath539 .
then , @xmath540 corollary [ cor : re and chi - square ] improves the upper and lower bounds in ( * ? ? ?
* proposition 2 ) : @xmath541 for example , if @xmath542 , gives @xmath543 ( in nates ) whereas gives @xmath544 .
note that if @xmath477 , then the upper bound in is @xmath537 whereas it is @xmath352 according to ( * ? ? ?
* proposition 2 ) . in view of remark [
remark : tight constants ] , the bounds in are the best possible among all probability measures @xmath332 with given @xmath375 ^ 2 $ ] . specializing theorem [ thm : local behavior fd ] results in ( cf .
* theorem 4.1 ) ) : [ cor : ratios : chi ] let @xmath527 converge to @xmath63 in the sense of . then , @xmath545 and @xmath528 vanish as @xmath530 with @xmath546 the continuous extension of @xmath468 in to @xmath417 $ ] yields @xmath547 .
the next result sharpens inequality under the assumption of bounded relative information .
[ thm : d - chi ] let @xmath107 and @xmath350 $ ] be defined as in .
if @xmath377 , then @xmath548 the main tool for the proof of theorem [ thm : d - chi ] is the following strengthened version of jensen s inequality , which generalizes a result in ( * ? ?
* theorem 1 ) .
[ lemma : superjensen ] let @xmath549 be probability measures defined on a measurable space @xmath64 , and fix an arbitrary random transformation @xmath550
. denote means that the marginal probability measures of the joint distribution @xmath551 are @xmath552 and @xmath553 . ]
@xmath554 , and @xmath555 .
let @xmath556 be a convex function , then @xmath557 ) \right ] - f ( \mathbb{e } [ z_\mathtt{0 } ] ) \bigr ) \leq \mathbb{e } [ f ( \mathbb{e } [ z_\mathtt{1 } | x_\mathtt{1 } ] ) ] - f ( \mathbb{e } [ z_\mathtt{1 } ] ) \end{aligned}\ ] ] where @xmath558 , @xmath559 , and @xmath560 see appendix [ appendix : proof superjensen ] . letting @xmath561 , and choosing @xmath552 so that @xmath562 ( e.g.
, @xmath563 is a restriction of @xmath552 to an event of @xmath552-probability less than 1 ) , becomes jensen s inequality @xmath564 ) \leq \mathbb{e}[f(x_\mathtt{1})]$ ] .
theorem [ thm : d - chi ] is a special case of the following result with @xmath98 @xmath565 .
[ thm : gi fd ] let @xmath33 be a convex function with @xmath89 .
assume that @xmath107 , let @xmath350 $ ] be defined as in , and let @xmath72 .
then , @xmath566 - f\bigl ( 1 + \chi^2(p\|q ) \bigr ) \\ \label{eq : gi fd } & \leq \beta_1^{-1 } \ , d_f(p \| q).\end{aligned}\ ] ] we invoke lemma [ lemma : superjensen ] with @xmath567 that is given by the deterministic transformation @xmath568 . then , @xmath569 = \exp \left(\imath_{p\|q } ( x_\mathtt{0 } ) \right ) $ ] .
if , moreover , we let @xmath570 , we obtain @xmath571 & = 1 , \\ \mathbb{e } \left [ f ( \mathbb{e } [ z_\mathtt{0 } | x_\mathtt{0 } ] ) \right ] & = d_f ( p \| q ) \end{aligned}\ ] ] and if we let @xmath572 , we have ( see ) @xmath573 & = 1 + \chi^2 ( p \| q ) , \\ \mathbb{e } \left [ f ( \mathbb{e } [ z_\mathtt{1 } | x_\mathtt{1 } ] ) \right ] & = \mathbb{e } \left [ f \left ( \exp ( \imath_{p\|q } ( x ) ) \right ) \right].\end{aligned}\ ] ] therefore , follows from lemma [ lemma : superjensen ] .
to show we use with @xmath574 and @xmath575 , and follows from lemma [ lemma : superjensen ] as well . in view of the alternative proof of theorem [ thm : samson ] , and the basic tool in theorem [ thm : fd1 ] , our next result forms a reverse inequality to .
[ thm : samson - refined ] let @xmath65 , and @xmath350 $ ] be given by .
then , @xmath576 & \leq \tfrac{2}{\log e } \ ; d(p\|q ) \label{eq : samson - refined - 2}\end{aligned}\ ] ] where @xmath577 \to \bigl[0 , \tfrac{2}{\log e } \bigr]$ ] is defined in .
we rely on our alternative proof of theorem [ thm : samson ] ( which already obtains the upper bound on @xmath578 ) .
recall that the function @xmath196 , which is the continuous extension of to @xmath417 $ ] , is monotonically increasing on @xmath46 $ ] , and monotonically decreasing on @xmath473 $ ] . hence , @xmath579 } \kappa(t ) & = \min \bigl\{\kappa(\beta_1^{-1 } ) , \ , \kappa(\beta_2 ) \bigr\ } , \\ \sup_{t
\in [ \beta_2 , \beta_1^{-1 } ] } \kappa(t ) & = \kappa(1 ) = \tfrac{2}{\log e}\end{aligned}\ ] ] and the upper and lower bounds on @xmath578 ( see and ) are obtained from theorem [ thm : fd1 ] . from the continuous extension of
to @xmath417 $ ] @xmath580 which yields that , if @xmath175 , then the left side of provides a positive lower bound on @xmath578 as long as @xmath377 .
the following theorem provides several expressions of the total variation distance in terms of the relative information .
these expressions prove to be useful later in this paper .
[ thm : tv ] let @xmath65 , and let @xmath72 and @xmath73 be defined on a measurable space @xmath64 .
then , the following equalities hold for the total variation distance between @xmath62 and @xmath63 : , and @xmath581 @xmath582 \\ \label{eq : tv3 } & = 2 \ , \mathbb{e } \bigl [ \bigl ( 1 - \exp(\imath_{p\|q}(y ) ) \bigr)^- \bigr ] \\ \label{eq : tv4 } & = 2 \ , \mathbb{e } \bigl [ \bigl ( 1 - \exp(-\imath_{p\|q}(x ) ) \bigr)^+ \bigr ] \\[0.1 cm ] \label{eq : tv4a } & = 2 \left ( \mathbb{p}\bigl[\imath_{p\|q}(x ) > 0\bigr ] - \mathbb{p}\bigl[\imath_{p\|q}(y ) > 0\bigr ] \right ) \\ \label{eq : tv4b } & = 2 \left ( \mathbb{p}\bigl[\imath_{p\|q}(y ) \leq 0\bigr ] - \mathbb{p}\bigl[\imath_{p\|q}(x ) \leq 0\bigr ] \right ) \\ \label{eq : tv5 } & = 2 \ , \int_0 ^ 1 \mathbb{p}\bigl[\imath_{p \| q}(y )
< \log \beta\bigr ] \ , \text{d}\beta \\ \label{eq : tv6 } & = 2 \ , \int_0 ^ 1 \mathbb{p}\bigl[\imath_{p \| q}(x ) > \log \frac1{\beta } \bigr ] \ ,
\text{d}\beta \\ \label{eq : tv100 } & = 2 \ , \int_1^{\beta_1^{-1 } } \beta^{-2 } \left [ 1 -\mathds{f}_{p\|q}(\log \beta ) \right ] \ , \text{d}\beta.\end{aligned}\ ] ] furthermore , if @xmath107 , then @xmath583.\end{aligned}\ ] ] eq
. follows from the definitions in and . since @xmath584 and @xmath585 , for all @xmath586 , and follow from and @xmath587 = \int \left ( 1 - \frac{\mathrm{d}p}{\mathrm{d}q } \right ) \ , \mathrm{d}q = 0.\end{aligned}\ ] ] by change of measure , for any measurable function @xmath588 with @xmath589 < \infty$ ] and @xmath590 < \infty$ ] , @xmath591 & = \mathbb{e } \bigl[\frac{\text{d}p}{\text{d}q}(y ) \ , f(y)\bigr ] \\ & = \mathbb{e } \bigl [ \exp\bigl(\imath_{p\|q}(y)\bigr ) \ , f(y)\bigr ] .
\end{split}\end{aligned}\ ] ] hence , it follows from that @xmath592 & = \mathbb{e } \bigl[1\bigl\{\imath_{p\|q}(x ) > 0\bigr\}\bigr ] \\ \label{eq : cm1 } & = \mathbb{e } \bigl[\exp(\imath_{p\|q}(y ) ) \ ; 1\bigl\{\imath_{p\|q}(y ) > 0\bigr\ } \bigr]\end{aligned}\ ] ] and @xmath593 & = \mathbb{e } \bigl[1\bigl\{\imath_{p\|q}(x ) \leq 0\bigr\}\bigr ] \\ \label{eq : cm2 } & = \mathbb{e } \bigl[\exp(\imath_{p\|q}(y ) ) \ ; 1\bigl\{\imath_{p\|q}(y ) \leq 0\bigr\ } \bigr].\end{aligned}\ ] ] to show , note that from and the change of measure in , we get @xmath594 \\ & = \mathbb{e } \bigl [ \bigl ( \exp(\imath_{p\|q}(y ) ) - 1 \bigr ) \ ; 1\bigl\ { \imath_{p \| q}(y ) > 0 \bigr\ } \bigr ] \\ & = \mathbb{e } \bigl [ \bigl ( 1 - \exp(-\imath_{p\|q}(x ) ) \bigr ) \ ; 1\bigl\ { \imath_{p \| q}(x ) > 0 \bigr\ } \bigr ] \\ & = \mathbb{e } \bigl [ \bigl ( 1 - \exp(-\imath_{p\|q}(x ) ) \bigr)^{+ } \bigr].\end{aligned}\ ] ] to show and , we get from and @xmath595 \\
\label{eq2 : tv4a } & = \mathbb{e } \bigl [ \bigl(\exp(\imath_{p\|q}(y ) ) - 1 \bigr ) \ ; 1\bigl\{\imath_{p\|q}(y ) > 0\bigr\ } \bigr ] \\ \label{eq3 : tv4a } & = \mathbb{p } \bigl[\imath_{p\|q}(x ) > 0\bigr ] - \mathbb{p } \bigl[\imath_{p\|q}(y ) > 0\bigr]\end{aligned}\ ] ] where is , and is equivalent to .
to show , we use and the notation in in order to write @xmath596 \\ & = \mathbb{e } \bigl [ ( z-1 ) \ , 1\{z>1\ } \bigr ] \\ \label{zzb } & = \int_0^{\infty } \mathbb{p } \bigl [ ( z-1 ) \ , 1\{z>1\ } \geq \beta \bigr ] \ , \text{d } \beta \\ & = \int_1^{\infty } \mathbb{p } \bigl [ z \geq \beta \bigr ] \ , \text{d } \beta \\ \label{zzc } & = \int_0 ^ 1 \mathbb{p } \bigl [ z < \beta \bigr ] \ , \text{d } \beta\end{aligned}\ ] ] where follows from with @xmath597 in ; exploits the fact that the expectation of a non - negative random variable is the integral of its complementary cumulative distribution function ; and holds since @xmath597 is non - negative with @xmath598 = 1 $ ] . to show , we use to write @xmath599 \\ & = \int_0^\infty \mathbb{p } \bigl [ \bigl ( 1 - \exp(-\imath_{p\|q}(x ) ) \bigr)^{+ } > \beta \bigr ] \ , \text{d } \beta \\ & = \int_0 ^ 1 \mathbb{p } \bigl [ \bigl ( 1 - \exp(-\imath_{p\|q}(x ) ) \bigr)^{+ } > \beta \bigr ] \ , \text{d } \beta \\ & = \int_0 ^ 1 \mathbb{p } \left [ \imath_{p\|q}(x ) > \log \frac1{1-\beta } \right ] \ , \text{d } \beta \\ & = \int_0 ^ 1 \mathbb{p } \left [ \imath_{p\|q}(x ) > \log \frac1{\beta } \right ] \ , \text{d } \beta.\end{aligned}\ ] ] to prove , a change of variable of integration in , and the fact that @xmath600 for @xmath601 give @xmath602 \ , \text{d}t \\[0.1 cm ] & = \int_0 ^ 1 \bigl [ 1 - \mathds{f}_{p\|q}\bigl(\log \frac1{t}\bigr ) \bigr ] \text{d}t \\[0.1 cm ] & = \int_1^{\infty } \frac{1-\mathds{f}_{p\|q}(\log \beta)}{\beta^2 } \ ; \text{d}\beta \\[0.1 cm ] & = \int_1^{\beta_1^{-1 } } \frac{1-\mathds{f}_{p\|q}(\log \beta)}{\beta^2 } \ ; \text{d}\beta\end{aligned}\ ] ] with the convention that @xmath603 if @xmath353 .
assume that @xmath107 .
to show simply note that , the symmetry of the total variation distance , and the anti - symmetry of the relative information where @xmath604 enable to conclude that @xmath605.\end{aligned}\ ] ] similarly , switching @xmath62 and @xmath63 in results in @xmath606 \\ & = \mathbb{e } \bigl [ \bigl ( 1 - \exp\bigl(-\imath_{p \| q}(x)\bigr ) \bigr)^{- } \bigr]\end{aligned}\ ] ] which proves . in view of , if @xmath65 , the supremum in is a maximum which is achieved by the event @xmath607 similarly to theorem [ thm : tv ] , the following theorem provides several expressions of the relative entropy in terms of the relative information spectrum . [ theorem : relative entropy - exact expressions ] if @xmath151 , then @xmath608 \label{eq2 : re } & = \int_1^\infty \frac{1 - \mathds{f}_{p\|q}(\log \beta)}{\beta } \ , \mathrm{d}\beta - \int_0 ^ 1 \frac{\mathds{f}_{p\|q}(\log \beta)}{\beta } \ , \mathrm{d}\beta \\[0.1 cm ] \label{eq3 : re } & = \int_0^\infty \pr\bigl[\imath_{p\|q}(y ) > \alpha \bigr ] \
, \alpha e^{\alpha } \ , \mathrm{d}\alpha - \int_{-\infty}^0 \pr\bigl[\imath_{p\|q}(y )
< \alpha \bigr ] \ , \alpha e^{\alpha } \ , \mathrm{d}\alpha\end{aligned}\
] ] where @xmath73 , and for convenience and assume that the relative information and the resulting relative entropy are in nats .
the expectation of a real - valued random variable @xmath609 is equal to @xmath610 = \int_0^\infty \pr[v > t ] \ , \mathrm{d}t - \int_{-\infty}^0 \pr[v < t ] \ , \mathrm{d}t\end{aligned}\ ] ] where we are free to substitute @xmath611 by @xmath612 , and @xmath613 by @xmath614 .
if we let @xmath615 with @xmath72 , then yields provided that @xmath616 < \infty$ ] .
. follows from by using the substitution @xmath617 when the relative entropy is expressed in nats . to prove , let @xmath618 with @xmath73 , and let @xmath619 where @xmath95 is given in with natural logarithm .
the function @xmath311 is strictly monotonically increasing on @xmath443 , on which interval we define its inverse by @xmath620 ; it is also strictly monotonically decreasing on @xmath442 $ ] , on which interval we define its inverse by @xmath621 \to ( 0,1]$ ] .
then , only the first integral on the right side of can be non - zero , and we decompose it as @xmath622 \ , \mathrm{d}t + \int_0^{\infty } \pr\bigl[z < 1 , \ , r(z ) > t\bigr ] \ , \mathrm{d}t \\ & = \int_0^{\infty } \pr\bigl[z > s_1(t)\bigr ] \ , \mathrm{d}t + \int_0 ^ 1 \pr\bigl[z < s_2(t ) \bigr ] \ , \mathrm{d}t \\ \label{eq : change of var } & = \int_1^{\infty } \pr[z > v ] \
, \log_e v \ , \mathrm{d}v - \int_0 ^ 1 \pr[z < v ] \ , \log_e v \ , \mathrm{d}v\end{aligned}\ ] ] where follows from the change of variable of integration @xmath623 . upon taking @xmath624 on both sides of the inequalities inside the probabilities in , and
a further change of the variable of integration @xmath625 , is seen to be equal to . in this section ,
we provide three upper bounds on @xmath9 which complement .
[ thm : ubtv ] if @xmath65 and @xmath72 , then @xmath626 $ ] , @xmath627 substituting @xmath628 , taking expectation of both sides of , and using give @xmath629 \\ & = \mathbb{e } \bigl [ \imath_{p\|q}(x ) + | \imath_{p\|q}(x ) | \bigr ] \\ & = d(p \| q ) + \mathbb{e } \bigl[|\imath_{p\|q}(x)| \bigr].\end{aligned}\ ] ] theorem [ thm : ubtv ] is tighter than pinsker s bound in ( * ? ? ?
* ( 2.3.14 ) ) : latexmath:[\ ] ] where follows from , and the monotonicity of @xmath51 , and is due to .
discussions with jingbo liu , vincent tan and mark m. wilde are gratefully acknowledged .
e. abbe and c. sandon , `` community detection in general stochastic block models : fundamental limits and efficient recovery algorithms , '' march 2015 .
[ online ] .
available at http://arxiv.org/abs/1503.00609 . to appear in the _
56th annual ieee symposium on foundations of computer science _ , berkeley , california , usa , october 2015 .
e. abbe and c. sandon , `` recovering communities in the general stochastic block model without knowing the parameters , '' june 2015 .
[ online ] .
available at http://arxiv.org/abs/1506.03729 . to appear in the _ 29th annual conference on neural information processing systems _ , montral , canada , december 2015 .
a. barron , `` information theory and martingales , '' presented at the _ 1991 ieee international symposium on information theory _
( recent results session ) , budapest , hungary , june 2329 , 1991 .
[ online ] .
available at http://www.stat.yale.edu/~arb4/publications_files/information theory and maringales.pdf[http://www.stat.yale.edu/~arb4/publications_files/information theory and maringales.pdf ] .
a. basu , h. shioya and c. park , `` statistical inference : the minimum distance approach , '' _ chapman & hall/ crc monographs on statistics and applied probability _ ,
vol . 120 , crc press , boca raton , florida , usa , june 2011 . c. d. charalambous , i. tzortzis , s. loyka and t. charalambous , `` extremum problems with total variation distance and their applications , '' _ ieee trans .
on automatic control _ , vol .
59 , no . 9 , pp .
23532368 , september 2014 .
i. csiszr , `` eine informationstheoretische ungleichung und ihre anwendung auf den bewis der ergodizitt von markhoffschen ketten , '' _ publ . math .
_ , vol .
8 , pp . 85108 , january 1963 .
s. s. dragomir , `` upper and lower bounds for csiszr @xmath0-divergence in terms of the kullback - leibler distance and applications , '' _
inequalities for csiszr f - divergence in information theory , rgmia monographs _ , edited by s. s. dragomir and t. m. rassias , victoria university , australia , 2000 .
s. s. dragomir , `` an upper bound for the csiszr @xmath0-divergence in terms of the variational distance and applications , '' _
inequalities for csiszr f - divergence in information theory , rgmia monographs _ , edited by s. s. dragomir and t. m. rassias , victoria university , australia , 2000
. s. s. dragomir and v. glucevi , `` some inequalities for the kullback - leibler and @xmath1-distances in information theory and applications , '' _ tamsui oxford journal of mathematical sciences _ ,
17 , no . 2 , pp .
97111 , 2001 .
e. even - dar , s. m. kakade and y. mansour , `` the value of observation for monitoring dynamical systems , '' _ proceedings of the international joint conference on artificial intelligence _ , pp . 24742479 , hyderabad , india , january 2007 .
h. jefrreys , `` an invariant form for the prior probability in estimation problems , '' _ proceedings of the royal society of london .
series a , mathematical and physical sciences _ , vol .
186 , no .
1007 , pp . 453461 , september 1946 .
m. krajci , c. f. liu , l. mikes and s. m. moser , `` performance analysis of fano coding , '' _ proceedings of the ieee 2015 international symposium on information theory _ , hong kong , pp .
17461750 , june 1419 , 2015 .
j. liu , p. cuff and s. verd , `` resolvability in @xmath2 with applications to lossy compression and wiretap channels , '' _ proceedings of the 2015 ieee international symposium on information theory _ , pp . 755759 , hong kong , june 1419 , 2015 .
j. liu , p. cuff and s. verd , `` one - shot mutual covering lemma and marton s inner bound with a common message , '' _ proceedings of the 2015 ieee international symposium on information theory _ , pp . 14571461 , hong kong , june 1419 , 2015 .
k. pearson , `` on the criterion that a given system of deviations from the probable in the case of a correlated system of variables is such that it can be reasonably supposed to have arisen from random sampling , '' _ the london , edinburgh , and dublin philosophical magazine and journal of science _ ,
50 , no .
302 , pp . 157175 , 1900 .
m. raginsky and i. sason , `` concentration of measure inequalities in information theory , communications and coding , '' _ foundations and trends in communications and information theory _ ,
10 , no . 1 - 2 ,
pp . 1250 , 2013 .
( second edition , 2014 ) .
i. sason and s. verd , `` upper bounds on the relative entropy and rnyi divergence as a function of total variation distance for finite alphabets , '' _ proceedings of the 2015 ieee information theory workshop _ , pp .
214218 , jeju island , south korea , october 1115 , 2015 .
m. tomamichel and v. y. f. tan , `` a tight upper bound for the third - order asymptotics for most discrete memoryless channels , '' _ ieee trans . on information
59 , no . 11 , pp . 70417051 ,
november 2013 .
s. verd , `` total variation distance and the distribution of the relative information , '' _ proceedings of the 2014 information theory and applications workshop _ , pp .
499501 , san - diego , california , usa , february 2014 . | this paper develops methods to obtain bounds on an @xmath0-divergence based on one or several other @xmath0-divergences , dealing with pairs of probability measures defined on arbitrary alphabets .
functional domination is one such approach , where special emphasis is placed on finding the best possible constant upper bounding a ratio of @xmath0-divergences .
another approach used for the derivation of bounds among @xmath0-divergences relies on moment inequalities and the logarithmic - convexity property , which is studied in this paper to obtain tight bounds on the relative entropy and bhattacharyya distance in terms of @xmath1 divergences . a rich variety of bounds
are shown to hold under boundedness assumptions on the relative information .
special attention is on the total variation distance and its relation to the relative information and relative entropy , including
reverse pinsker inequalities , " as well as on the @xmath2 divergence , which generalizes the total variation distance .
pinsker s inequality is extended for this type of @xmath0-divergence , a result which leads to an inequality that links the relative entropy and relative information spectrum .
integral expressions of the rnyi divergence in terms of the relative information spectrum are derived , leading to bounds on the rnyi divergence in terms of either the variational distance or relative entropy . * keywords * : relative entropy , total variation distance , @xmath0-divergence , rnyi divergence , pinsker s inequality , relative information . |
organic field - effect transistors ( ofets ) are the elementary building blocks of `` plastic electronics''.@xcite in these devices , charges are induced in a thin conducting channel at the interface between an organic semiconductor and a gate insulator .
while commercial applications have already appeared , mostly based on low - cost organic thin - films , the fundamental physical mechanisms governing the charge dynamics in these devices are not fully understood .
as it has become clear in recent years , even when high purity crystalline semiconductors are used in order to minimize the effects of structural and chemical disorder , the electronic characteristics of ofets are strongly affected by interactions taking place both within the semiconducting material as well as with its close environment .
this can be traced back to the extremely narrow electronic bands resulting from the weak van der waals intermolecular bonding , that make organic semiconductors much more sensitive to electronic interactions than their inorganic counterparts .
for this reason , polaronic effects have been shown to play an important role in these devices .
several electron - lattice interaction mechanisms relevant to organic molecular crystals have been identified and studied in the past .
these include holstein - like couplings to intra - molecular vibrations , leading to a local reorganization of the molecular energy levels@xcite as well as su - schrieffer - heeger ( ssh ) or peierls - like couplings where the transfer integrals between the molecules are modulated by the inter - molecular motion.@xcite the latter mechanism couples directly the electronic motion to the strong lattice fluctuations arising from the mechanical softness of these compounds , and has been recently identified as the dominant limiting factor of the charge mobility in pure crystalline semiconductors .
@xcite in addition to such intrinsic mechanisms , interface - related effects have also been demonstrated in several systematic studies of organic fets using different gate dielectrics .
sources of charge trapping have been identified , either related to the interface quality@xcite or to the long - range forces arising from polar impurities at the organic / dielectric interface .
@xcite furthermore , it has been shown that the long - range polarization induced in high-@xmath0 gate materials by the charge carriers themselves can lead to self - trapping of the carriers .
even when sufficient care is taken in order to minimize extrinsic sources of disorder , frhlich polarons are formed due to the polar long - range interaction of the carriers in the organic semiconductor with the phonons at the semiconductor / dielectric interface .
the strength of this remote electron - phonon ( e - ph ) interaction can be tuned by an appropriate choice of the gate dielectric .
the `` metallic - like '' carrier mobility characteristic of crystalline organic semiconductors can be converted into a much lower , thermally activated mobility.@xcite while there is a vast theoretical literature dedicated to the different electron - lattice interaction mechanisms mentioned above , up to now there have been no attempts to study the interplay between bulk and interface e - ph couplings . to fill this gap , in this work
we analyze a model that includes both bulk and interface effects as can be realized in organic fets , and treat them in a common framework .
we focus on the combined effects of the ssh coupling of the electronic transfer integrals to the inter - molecular vibrations and of a frhlich long - range e - ph interaction as induced by the presence of the gate dielectric in a field - effect geometry .
apart from its relevance to organic / dielectric interfaces , from the fundamental viewpoint this model presents an interesting combination of two _ a priori _ competitive mechanisms .
our results show that a rather weak ssh coupling strength as estimated in refs .
[ ] can have an unexpectedly strong effect in stabilizing a polaronic state when a moderate long - range polar coupling is present .
therefore , self - trapped states of small radius can exist at common organic / dielectric interfaces , even in such cases where the carrier - interface phonon interaction alone would not be sufficient to produce polaronic self - localization @xcite .
this provides a microscopic theoretical basis for the experimental results of ref . , where a finite activation energy indicative of self - trapped states was observed using gate materials such as al@xmath1o@xmath2 and ta@xmath1o@xmath3 .
this paper is organized as follows .
the model under study is introduced in section [ sec : model ] .
the two methods of solution that will be used are described in section [ sec : methods ] . in section [ sec : results ] we introduce the main quantities of interest and present the results of our calculations .
section [ sec : conclusions ] is devoted to the conclusions .
a detailed derivation of the form of the long - range interaction with the gate material is presented in appendix [ sec : appendix ] .
we consider the following one dimensional tight - binding model [ sec : model ] @xmath4 \ ; ( c^\dagger_{i } c_{i+1 } + h.c . ) \nonumber\\ & + & \sum_{i , q } c^\dagger_i c_i ( m_q e^{iqr_i}b_q+h.c . )
\nonumber\\ & + & \omega_{ssh}\sum_i a^\dagger_i a_i + \omega_{lr}\sum_q b^\dagger_{q } b_{q } \label{eq : ssh}\end{aligned}\ ] ] where electrons or holes described by the creation and destruction operators @xmath5 move on a lattice labelled by site index @xmath6 .
these interact with molecular displacements @xmath7 via a ssh interaction , describing the transfer integral modulation on the distance between nearest neighbors with strength @xmath8 , as well as with optical modes @xmath9 at the polar interface via a coupling @xmath10 . to keep the discussion as simple as possible and illustrate the main consequences of the interplay between the `` off - diagonal '' short range ssh coupling and the `` diagonal '' long - range polar coupling
, we shall restrict the electronic motion to a one - dimensional chain . yet , our results can be considered as being qualitatively representative of organic / dielectric interfaces in the more realistic two - dimensional case .
as it is discussed in appendix a we consider the one dimensional interface model introduced in refs .
the e - ph interaction matrix element reads in this case @xmath11 with @xmath12 a cut - off length of the order of the lattice spacing @xmath13 and @xmath14 a strictly one - dimensional vector . as it is clear from eq .
( [ eq : mq ] ) , the interaction between the electron charge and the associated lattice polarization is long ranged , with a power law decay in real space .
we define two dimensionless coupling constants for the two different el - ph couplings : @xmath15 and @xmath16 we shall study the system in the range of parameters typical of single crystal organic semiconductors , taking rubrene as a case study .
we set @xmath17 as an average value for the interface phonons in the case of common polarizable gate materials@xcite and @xmath18 for the relevant intermolecular vibrations@xcite ( @xmath19 throughout the paper ) .
the lattice cutoff parameter entering in eq .
( [ eq : mq ] ) is chosen to be @xmath20 .
@xcite in view of the small values of @xmath21 and @xmath22 , we will study the system by using two approaches : the adiabatic approximation , valid for vanishing @xmath21 and @xmath22 , and a variational exact diagonalization ( ved ) method able to include non adiabatic contributions .
for values of @xmath21 and @xmath22 that are much smaller than @xmath23 , the lattice polarization can not follow the electronic oscillations and the wave function of the system can be factorized@xcite into a product of normalized variational functions @xmath24 and @xmath25 depending on the electron and phonon coordinates respectively : @xmath26 where @xmath27 this factorization becomes exact when we neglect completely the ionic kinetic energy i.e. @xmath28 . in eq.([ad2 ] ) , @xmath29 is the electron vacuum state and @xmath30 are parameters that satisfy the relation : @xmath31 the expectation value of the hamiltonian ( [ eq : ssh ] ) on the state @xmath32 provides an effective hamiltonian for the lattice fields : @xmath33 \nonumber \\ & & + \sum_{q } \left [ \omega_{lr } b^{\dagger}_{q } b_{q}+ \left ( \rho_{q } b_{q}+h.c.\right ) \right]\nonumber \\ & & + \sum_{m } \left [ \omega_{ssh } a^{\dagger}_{m } a_{m}+ \left ( v_{m } a_{m}+h.c.\right ) \right ] \label{ad4}\end{aligned}\ ] ] with @xmath34 and @xmath35 .
\label{ad6}\ ] ] it can be proven that the ground state of this hamiltonian is a product of two coherent states : @xmath36 where @xmath37 and @xmath38 are the vacuum states of the two lattice fields . we stress that the parameters @xmath39 are real [ see eq.([ad6 ] ) ] . at this stage
, the phonon state @xmath40 can be used to obtain an effective hamiltonian for the charge carrier : @xmath41 \ ; ( c^{\dagger}_i c_{i+1 } + c^{\dagger}_{i+1 } c_{i } ) \nonumber\\ & -&\sum_{i } c^{\dagger}_i c_i \sum_{q } ( m_q e^{iqr_i } \rho^*_{q}+h.c.)+k \label{elec}\end{aligned}\ ] ] where @xmath42 the present adiabatic approximation breaks translational invariance .
the problem is reduced to studying an electron , confined by the potential well raised by the lr interaction .
the hopping amplitude from site to site within the well depends on the specific site pair , via the ssh coupling .
the wavefunction parameters @xmath30 are determined variationally . in the systems
on which we are focusing our attention the phonon frequencies @xmath22 and @xmath21 are small compared to the transfer integrals.@xcite the adiabatic approach discussed above should therefore provide a reasonably appropriate description of the ground state properties .
it is known , however , that a proper inclusion of non adiabatic terms in the calculation removes the fictitious translational symmetry breaking , which is found within the pure adiabatic approximation , and can introduce significative changes at any non - zero phonon frequencies . furthermore physical quantities related to the quantum nature of the phonons ( number of phonons in the ground state ) or the delocalization of the electronic wave function ( effective mass and ground state spectral weight ) can be accessed only via a non adiabatic approach . to this aim
we implement here a method , combining a variational approach and an exact diagonalization technique , which is able to provide an accurate description of the ground state properties at all values of the system parameters .
we shall first analyze the two interactions separately , and then introduce the method for treating the full hamiltonian eq .
( [ eq : ssh ] ) .
it has been shown@xcite that the ground state features in the case of `` diagonal '' electron - lattice interactions , such as the lr term eq .
( [ eq : mq ] ) , are very well reproduced by methods assuming as starting point the approach introduced by toyozawa.@xcite the idea is to restore a translationally invariant bloch state using the wavefunctions provided by the adiabatic approximation , i.e. by taking a superposition of localized states centered on different lattice sites .
the trial wave function that accounts for the translational symmetry reads : @xmath43 where @xmath44 and @xmath45 in eqs.([1na]-[3na ] ) the index @xmath46 denotes the distance between the electron position and the lattice modes .
the next step is to generalize this wavefunction to correctly recover the weak coupling regime where the electron is weakly dressed by lattice oscillations , as predicted by perturbation theory .
this can be achieved by treating the parameters entering into the definition of the function @xmath47 as independent variational parameters,@xcite rather than being constrained by the shape of the electron function as in eq .
( [ ad5 ] ) . actually , also in the perturbative regime the lattice displacement is described by a coherent state , but with a different function @xmath47 including an electron recoil term due to the scattering with phonons.@xcite the parameters @xmath48 have to be determined via numerical diagonalization of the corresponding hamiltonian , defining the variational exact diagonalization method described below . in this way
we obtain a wavefunction that is able to capture the weak and strong coupling limits and interpolates between them in the intermediate regime . in practice
we use a basis where each vector is a linear superposition , with appropriate phase factors , of the translational copies of a state having a given electron position and lattice configurations ( phonons and electron are translated together , with their relative position being held fixed ) .
we choose as phonon states those appearing in eq .
( [ 3na ] ) where @xmath47 is a variational function .
there are then @xmath49 basis vectors , labelled by the @xmath46 index appearing in eq .
( [ 3na ] ) , and each of them includes the variational function @xmath47 .
the coefficients @xmath50 in eq.([2na ] ) are determined for each fixed set of the parameters defining the variational function @xmath47 by diagonalizing the hamiltonian matrix in the chosen basis .
these do not coincide , in general , with the coefficients of the function @xmath47 in eq.([ad5 ] ) .
summarizing , we have to diagonalize a @xmath51 matrix for each value of the variational function @xmath47 . finally , in the following , we assume that the variational parameters @xmath52 , entering into the definition of the function @xmath47 in eq .
( [ ad5 ] ) , are different from zero up to 3rd nearest neighbors .
we have checked that this assumption allows to obtain convergency for all the values investigated of the model parameters .
the procedure described above could be straightforwardly generalized to include also the interaction with ssh phonons .
unfortunately , it can be shown that it does not capture the correct physical features of the ssh hamiltonian in the absence of lr interaction.@xcite to treat the ssh term we therefore apply a different approach based on an exact diagonalization procedure , where we do not assume as basis set the coherent states of the adiabatic approach . in this approach the phonon variables
are efficiently limited to a new set which is able to describe successfully the ground state properties , as it will be shown below .
the real bottleneck comes from the phonon hilbert space , that is infinite . to circumvent this difficulty , we keep states where the lattice deformations involve only the sites up to 3rd nearest neighbors of the lattice site where electron is located , all the others being undeformed @xmath53 where @xmath46 indicates the electron position .
we apply a truncation procedure considering only the states with at most @xmath54 quanta , i.e. @xmath55 , where @xmath54 is fixed by requiring convergency of the ground state energy .
these states provide a correct physical description of the ground state properties , since the lattice deformations are effectively localized around the site where the electron is located .
furthermore , in analogy with the lr coupling , we use the translational symmetry associated to periodic boundary conditions to reduce the size of the sparse hamiltonian matrix , requiring that the states have a definite momentum . when both lr and ssh couplings are considered , we adopt a diagonalization technique on a truncated phonon basis that combines the two subspaces introduced separately for the @xmath56 and @xmath57 cases .
this basis is given by the product of the vectors describing the two different couplings separately , eqs .
( [ 3na ] ) and ( [ ph ] ) .
the ground state is found by using the modified lanczos algorithm.@xcite we emphasize that this approach retains the full translational invariance requiring that each basis vector has a well defined momentum .
due to their different microscopic nature , the two interaction terms in eq .
( 1 ) induce different types of polaron states : the lr coupling to the electron density favors localized states centered on the lattice sites ( site polaron ) , while the ssh coupling to the electronic transfer integrals tends to localize the polarons on inter - molecular bonds , i.e. with the electronic charge centered between two neighboring molecular sites ( bond polaron ) .
this competition results in the phase diagrams reported in fig .
[ fig : phasediag ] , which constitute the central results of this paper .
the most notable feature is the existence of a reentrant phase where a bond polaron state is stabilized by the interplay between the two microscopic interaction mechanisms .
this result is robust against the presence of phonon quantum fluctuations , as is apparent from the comparison of the phase diagram obtained from the adiabatic approximation , valid in the limit of vanishing frequencies @xmath28 [ fig.[fig : phasediag ] ( a ) ] , with the results at finite phonon frequencies [ fig.[fig : phasediag ] ( b ) ] . in the adiabatic approximation the ground state electronic wave function is always localized .
however , for small values of @xmath58 , a crossover from a large to a small site - centered polaron occurs upon increasing the lr interaction strength @xmath59 [ blue dashed line in fig .
[ fig : phasediag ] ( a ) ] : the wavefunction extension is progressively reduced , being always centered on a given molecular site .
the locus of the crossover can be determined by analyzing the qualitative changes induced on the phonon wavefunction , as described at the end of this section .
the large to small polaron crossover is only weakly affected by the presence of the ssh interaction . as @xmath58 increases , the symmetry of the electronic wave function changes continuously from site - centered to bond - centered .
the breaking of translational invariance allows us to analyze this transition by direct inspection of the electronic wavefunction , which is shown in fig .
[ fig : adiabatic - wf ] together with its associated distortions in the bond - centered ( left panel ) and site - centered polaron ( right panel ) states .
the transition between the two phases is indicated by the green dashed line in fig.[fig : phasediag]a .
contrary to the large / small polaron crossover , that is rather insensitive to the value of @xmath58 , the site to bond - polaron transition is significantly modified by the presence of the lr interaction , which strongly stabilizes the bond polaron state .
the boundary of the bond - centered polaron phase shows a markedly reentrant behavior , reaching a minimum @xmath60 at @xmath61 .
@xcite + it is instructive to understand the stabilization of the bond polaron phase induced by the long - range interface interactions from the following energy balance argument . in the absence of intermolecular coupling ( @xmath62 ) the energy of the large polaron can be written in general as @xmath63 where the second term can be expanded for small @xmath59 as @xmath64 .
on the other hand we see from fig .
[ fig : adiabatic - wf ] that in the bond polaron phase the adiabatic electronic wave function is mainly localized on two neighboring sites , with exponentially vanishing tails on the adjacent molecules . neglecting such tails
provides the following estimate for the bond polaron energy : @xmath65 where the two contributions come respectively from ssh and lr interactions .
the parameter @xmath66 ( @xmath13 the lattice spacing , @xmath67 ) takes into account the spatial variation of the long - range potential - well between two neighboring sites . comparing the two energies we obtain a transition when
@xmath68\lambda_{lr},\ ] ] and a reentrant behavior is found when @xmath69 . observing that @xmath70 is largest for slowly varying potentials , we recognize that the reentrance is strongly favored by the long range nature of the polar dielectric interaction .
the reason is that such long - ranged polar interaction , for which the preferred polaronic state would be site - centered , is not extremely sensitive to the local details of the wavefunction , so that it can still provide a sizable energy stabilization in the case of a dimer structure .
it should be stressed that the arguments presented here are only qualitative . for a quantitative analysis of the large polaron to bond polaron phase boundary
, the extended nature of the bond polaron wavefunction must be taken into account , which further enhances the reentrant behavior as compared to eq .
( [ eq : reentr ] ) ( see fig.[fig : phasediag ] ) . phonon
related characteristics also change across the bond polaron transition , following the changes of the electronic wavefunction . as can be seen from fig .
[ fig : adiabatic - wf ] , the average ssh distortions within a bond polaron are opposite , since this configuration minimizes the ssh interaction energy . within the entire bond polaron region of the phase diagram , even at the relatively large value of the lr coupling @xmath71 illustrated in fig .
[ fig : adiabatic - wf ] , the lr distortions follow the charge distribution determined by minimizing the ssh interaction energy . the phonon probability distribution functions ( pdf ) provide an alternative tool to characterize the ground state properties,@xcite with the advantage of being directly generalizable to the translationally invariant ground states examined in the next section .
the ssh - pdf , which represents the distribution of ssh displacements at site @xmath72 given an electron on site @xmath6 , is defined as @xmath73 where @xmath74 is the coordinate eigenstate corresponding to the displacement of ssh phonons on site @xmath72 .
the lr - pdf is defined by substituting @xmath75 with @xmath76 in eq .
( [ eq : pdf ] ) .
it is worth noting that the pdf is closely related to the distributions of interatomic distances that can be directly measured by neutron spectroscopy such as pulsed neutron diffraction.@xcite the local ( @xmath77 ) part of these functions is plotted in fig .
[ fig : adiabatic - ppdf ] for different polaronic ground states . in the small @xmath58 region of the phase diagram
[ see fig .
[ fig : phasediag ] a ) ] , the large - to - small polaron crossover is highlighted by the bimodality in the lr - pdf [ panel ( a ) of fig .
[ fig : adiabatic - ppdf ] ] . on the other hand ,
when the bond polaron state is _ well _ formed , the ssh - pdf shows two symmetric peaks [ panel ( b ) ] , whose distance increases with @xmath58 .
this provides an alternative method to locate the bond polaron region , which is shown in fig .
[ fig : phasediag ] ( a ) as a red continuous line .
the resulting phase diagram shows that a broad crossover region emerges ( between the green and red lines in fig .
[ fig : phasediag ] where the electronic wavefunction has already undergone the bond transition but a real bimodality of the ssh - pdf is still not observed , with the ssh - pdf showing large non - gaussian fluctuations .
let us note that in the bond polaron region of the phase diagram , the analysis of the lr - pdf does not show any evidence of a large to small bond polaron crossover , at least for values of @xmath78
. + summarizing , the adiabatic phase diagram shows three distinct regimes : small site - centered polaron , large site - centered polaron and bond polaron . the bond polaron regime , with the electron wavefunction mostly localized on a molecular dimer , is stabilized by the presence of even a moderate long range coupling .
[ fig : phasediag ] ( b ) illustrates the phase diagram obtained using the scheme described in section [ section : vedmethod ] .
translational symmetry breaking is prevented by the quantum fluctuations of both ssh and lr phonons .
accordingly , all the lines shown in fig . [ fig : phasediag ] ( b ) correspond to _ crossover _ lines separating translationally invariant states with different characteristics , as obtained by analyzing the phonon pdfs .
@xcite + + in fig .
[ fig : pdfnormal ] we plot the ssh - pdf and lr - pdf for @xmath79 , in the small @xmath58 region ( we use @xmath80 ) of the phase diagram , across the large - to - small polaron crossover [ see fig .
[ fig : phasediag ] ( b ) ] .
this crossover is evinced by the bimodality in the lr - pdf , occurring at @xmath81 . both the on - site ( @xmath77 ) and the inter - site lr - pdf ( @xmath82 , which are equal by symmetry ) , become bimodal at the crossover . in this region ,
the ssh phonons present a nearly gaussian distribution with opposite distortions around the polaron localization site .
it is interesting to note that the average value of the ssh displacement has a non - monotonic dependence on the lr interaction and it is largest for intermediate values of @xmath59 , signaling the proximity to the bond polaron regime .
+ in the bond polaron phase both the lr and the ssh - pdf have a bimodal character , as shown in fig .
[ fig : pdfintermediate ] .
the ssh - pdf displays bimodality in both the on - site ( @xmath83 ) and nearest neighbor ( @xmath84 ) contributions .
the site where the polaron is located belongs to either a left or right - centered dimer , a situation which is consistent with the translational invariance of the ground state .
the lr - pdf instead is bimodal only on the nearest neighbor ( @xmath85 ) , with two peaks corresponding to distorted and non - distorted sites respectively .
the presence of a distortion with the same amplitude on nearest neighboring sites is another signal of the tendency to form a dimer .
this tendency is also displayed by the lr distortions which , in this region of the phase space , follow the charge distribution dictated by the ssh interaction .
let us note that the reentrant bond polaron phase at @xmath86 has a different symmetry than that at @xmath87 .
indeed , according to ref .
[ ] , the transition to the bond polaron phase induces the ground state wavenumber to change continuously from @xmath88 to @xmath89 in a narrow region around @xmath90 .
as soon as @xmath59 is different from zero , we do not find any evidence of a @xmath91 ground state for @xmath92 . to illustrate this point , the electron dispersion characteristic of this phase , where the minimum is located at @xmath88 ,
is plotted in fig .
[ fig : band ] .
we conclude that in the reentrant bond polaron phase , the lr coupling stabilizes a bloch wavefunction having @xmath88 . in fig .
[ fig : massvslambda ] we plot the inverse effective mass and the quasiparticle weight , defined as @xmath93 , as a function of @xmath59 for different values of the ssh coupling @xmath58 . at small vaues of @xmath58 the inverse effective mass
first decreases smoothly with @xmath59 and then drops at the large - to - small polaron crossover around @xmath94 .
the quasiparticle spectral weight shows a similar dependence on @xmath59 , although with a steeper reduction which is characteristic of non - local e - ph interactions.@xcite in contrast with the rather smooth @xmath59 dependence , a much stronger dependence on @xmath58 can be seen in fig .
[ fig : massvslambda ] , signaling the crossover into the reentrant bond polaron regime at @xmath92 .
we note that the quasiparticle spectral weight and inverse effective mass have in this case the same behavior with @xmath59 as it occurs for local e - ph interactions , @xcite i.e. they tend to be mutually proportional
. complementary insight on the electron dynamics in the different phases can be obtained from the analysis of the total kinetic energy , @xmath95 , shown in fig .
[ fig : ekinvslambda ] . unlike the quasiparticle spectral weight @xmath96 and effective mass @xmath97
, this quantity gets contributions from both the coherent long distance motion of the polaron , and from the internal motion of the electron within the lattice potential - well .
the large - to - small polaron crossover at small @xmath58 causes a sharp drop of @xmath98 around @xmath99 . for @xmath100 a region
emerges where the kinetic energy stays relatively constant over a broad range of @xmath59 .
this plateau is another clear signature of the bond polaron phase . here
the effective mass and the quasiparticle spectral weight are very small , indicative of polaron self - trapping , yet a large contribution to the kinetic energy arises from the internal dimer structure of the polaron , with the electron moving freely along the intermolecular localization bond . the large number of ssh phonons participating in the polaronic state in the reentrant phase , shown in fig .
[ fig : nphvslambda ] , indicates that a strong coupling with intermolecular modes in the organic crystal can be effectively triggered even at relatively weak values of @xmath58 , by suitably turning on the lr interaction . at the same time
, the number of lr phonons stays relatively small , exhibiting only a weak dependence on @xmath58 ( see lower panel of fig .
[ fig : nphvslambda ] ) .
in this paper , we have investigated a microscopic model that describes the interplay between the inter - molecular electron - phonon coupling and the additional interaction arising in crystalline organic semiconductors due to the long range polarization at the interface with a gate dielectric , two ingredients that are commonly found in organic field effect transistors .
the ground state properties obtained by two complementary approaches the adiabatic approximation and a more refined variational method that correctly preserves translational invariance in the presence of phonon quantum fluctuations demonstrate the synergistic cooperation between bulk and interfacial interactions .
in particular , we have found that a large - mass polaronic state with a dimer structure and strong inter - molecular distance fluctuations is stabilized at relatively low values of both the long - range polar coupling @xmath59 and the intermolecular coupling @xmath58 .
such _ bond polaron _
reentrant phase arises in a region of the phase diagram where the two interactions taken separately would not give rise to strong polaronic effects .
the long range nature of the polar interaction is crucial for the establishment of this cooperative interplay .
the latter occurs because the long range interface polarization is able to provide a sizable energy gain without much perturbing the local dimer structure dictated by the ssh intermolecular coupling , so that the system can take advantage of both microscopic interaction mechanisms simultaneously . from the device perspective ,
our findings provide a microscopic basis to explain the existence of small - radius self - trapped states without invoking unrealistic values of the polar coupling .
indeed , the electrical characteristics of high-@xmath0 organic fets have been analyzed in ref .
[ ] in terms of the polar interface interaction alone , and the observed thermally activated mobilities are not entirely consistent with the estimated values of @xmath59 , as has been pointed out in ref .
the present results show that the proper inclusion of a bulk electron - lattice interaction mechanism such as the intermolecular ssh coupling extends the region of small - radius polaronic self - trapping down to much lower values of @xmath59 .
let us stress that an analogous interplay as the one studied here can be expected when combining the effects of a bulk inter - molecular electron - lattice coupling and the interaction with localized polarizable impurities such as those considered in refs .
[ , ] . comparing the inter - molecular coupling strength estimated for rubrene , @xmath101 , @xcite with the phase diagrams of fig .
[ fig : phasediag ] , we conclude that interfaces involving this organic material and the aforementioned gate materials lie in a region of the phase diagram where the intermolecular / gate synergistic effects are expected to be of relevance .
more importantly , our results could explain why at present there are only few organic semiconductors ( among which , rubrene ) where metallic - like mobilities have been measured in a fet configuration .
other materials with slightly larger values of the inter - molecular coupling strength , @xmath58 , would rapidly enter the bond - polaron phase , causing the carriers to undergo a self - trapping transition as soon as a moderate interaction with the dielectric environment is present .
although the model investigated in this paper includes some drastic approximations , i.e. the model is one - dimensional and the real lattice structure of the organic fet is not taken into account , we believe that the main conclusions provided here will not be qualitatively modified by the inclusion of more realistic lattice structures and interactions . the detailed study of the transport characteristics in the bond polaron phase identified here remains as an open question for future work .
in this work we studied the interplay between ssh and frhlich e - ph interactions in one - dimension . while the definition of the ssh coupling in one - dimension is unambiguous,@xcite the polar e - ph interaction term depends on the geometry of the interface .
we now examine two physically different cases .
the case that is studied in the paper is a truly one - dimensional interface , derived for a wire of square cross - section and permittivity @xmath102 embedded in a three - dimensional polar dielectric with permittivity @xmath103.@xcite we specialize the results of ref .
[ ] by allowing the motion of the charges only on the symmetry axis of the interface @xmath104 with @xmath105 , at a distance @xmath12 from the polar dielectric with permittivity @xmath103 .
the medium with permittivity @xmath102 is taken to be the vacuum . in this geometry
the interaction of electrons with the interface phonons reads @xmath106 with @xmath107 or , in real space , @xmath108 notice that here the phononic modes are purely one dimensional modes , i.e. they are propagating waves only along the wire ( direction @xmath104 ) .
it is interesting to compare the one dimensional interaction discussed above with a two dimensional model .
the model that we consider is the generalization of the case studied previously by mori and ando,@xcite that describes the coupling between electrons and two - dimensional interface phonons . in this case
the interaction matrix element in momentum space reads @xmath109 , with the distance @xmath104 between the electrons and the interface acting as a short - distance cutoff . performing the fourier
transform we find the hypergeometric function @xmath110 this scales as @xmath111 for large @xmath112 , and tends to a constant @xmath113 for @xmath114 .
this function is well approximated by the power - law form @xmath115 where @xmath116 and @xmath117 .
it can be easily checked by retransforming back to k - space that this power - law form gives rise to the correct @xmath118 limit and lies within @xmath119 of the original one for all @xmath120 .
then , in both one and two - dimensional cases , eqs .
( [ eq:1d],[eq:2d ] ) give rise to long range interactions , describing qualitatively similar physical situations .
we have chosen to study explicitly the first model in the body of the paper , as in that case the properties of the phononic degrees of freedom can be directly accessed by the solution of a genuinely one dimensional model .
10 we note that the site - polaron to bond - polaron transition sharpens upon reducing the lr coupling strength , until it approaches @xmath90 at @xmath87 .
this value marks the onset of the `` pathological '' localization found in ref .
[ ] in the static phonon limit , where the ssh coupling leads to a two - site polaron dimer that is perfectly isolated from the rest of the system , due to the vanishing of the corresponding transfer integrals .
the model remains however well - defined even for larger values @xmath58 , @xcite whose analysis is beyond the scope of the present paper . | a model describing the low density carrier state in an organic single crystal fet with high-@xmath0 gate dielectrics is studied . the interplay between charge carrier coupling with inter - molecular vibrations in the bulk of the organic material and the long - range interaction induced at the interface with a polar dielectric
is investigated .
this interplay is responsible for the stabilization of a polaronic state with an internal structure extending on few lattice sites , at much lower coupling strengths than expected from the polar interaction alone .
this effect could give rise to polaron self - trapping in high-@xmath0 organic fet s without invoking unphysically large values of the carrier interface interaction . |
atlas will be a particle physics experiment at the future large hadron collider ( lhc ) , which is being built at cern and is expected to start operation in 2007 .
the pixel detector is the innermost component of the atlas inner tracker . in the barrel
the detector modules are mounted on staves , while the modules in the end caps are organized in disk sectors .
the pixel detector consists of 1774 detector modules ( barrel : 1456 modules ; discs : 318 ) .
@xmath0 + the most important components of a detector module are : * 46080 individual pixel sensors with size of @xmath1 * 16 front end read out chips * 1 module controller chip
the pixel stave is the module local support unit of the barrel section of the pixel detector .
the components of a stave are the thermal management tile ( tmt ) , the stave carbon structure , and the stave cooling circuit .
additionally , every stave has two geometrical reference marks ( ruby balls ) .
the stave coordinate system is specified in figure [ fig : altube ] .
the tmt itself consists of two parts .
both parts have a shingled design with an angle of 1.1 degrees which are glued together . as material for the tmts , carbon - carbon ( c - c ) has been chosen .
the reason for this is a thermal conductivity which is 10 - 100 times better than standard carbon fiber reinforced plastic ( cfrp ) , even in the transverse direction to the fibres .
it has excellent mechanical properties , stability , and transparency to particles .
the tmt is made of 1502 zv 22 c - c material from sgl ( augsburg , germany ) .
the raw material is in the form of plates of about 6 mm thickness .
the plates consist of 2-d roving fabric carbon fibres layers overlapped in a phenolic resin matrix , densified , and graphitized at high temperature to enhance the thermal properties .
the raw tmts are machined to the final stepping shape with a cnc milling machine equipped with high speed spindle and diamond coated millers .
the specific properties of the material are summarized in table [ tab : propertiesofsglcc1502zv22 ] .
.properties of sgl cc 1502 zv 22 [ cols="<,<,<",options="header " , ] the stave cooling circuit is made of a thin aluminum tube ( see figure [ fig : altube ] ) , shaped to fit inside the inner cross section of the stave carbon structure .
the material chosen is 6060 aluminum alloy .
this material shows good extrusion properties at small thickness .
the cooling system is based on the evaporation of fluorocarbon and provides an operating temperature of about @xmath2 . the stave carbon structure ( `` omega profile '' )
is made of three layers of unidirectional ultra high modulus ( uhm ) carbon fibre reinforced cyanate ester resin pre - preg ( preimpregnated ) .
the adopted lay - up ( 0 - 90 - 0 ) , 0.3 mm thick , has been optimised through an extensive design and test program . the choice of the pre - preg material and of the lay - up with a central cross - layer has been made in order to match the longitudinal cte to that of the impregnated c - c , thus minimizing distortions resulting during cool - down .
the 13 modules are arranged one after the other along the stave axis and they are partially overlapped in order to achieve the required coverage in the axial direction .
thus , the surface of the stave in contact with the modules is stepped and the modules are arranged in a shingled layout . however , c - c materials have two main technological drawbacks limiting their application range : porosity and difficulty to achieve complex and accurate geometries due to the high temperature manufacturing process .
to overcome the porosity of the c - c material , it was impregnated with resin such that infiltration by the thermal greases and carbon dust release could be avoided .
for the assembly of modules on staves a custom made module loading machine was developed and built in wuppertal .
the requirements of this machine are : * handling the modules with minimal stress * positioning of modules on the stave with an accurancy better than 50 microns * regular glue deposition + to control the applied stress the bow of each module is measured before and after loading .
the main components of the module loading machine are the granite base of @xmath3 and a flatness of @xmath4 from johan fischer . on this base
several linear guideways from schneebergertype monorail bm are mounted to allow movement of the central measurement unit , the microscope m 420 from leica .
the microscope itself is connected to owismicrometric tables and allows movements in all dimensions .
the movements are controlled by heidenhainsealed linear encoders .
heidenhain digital readouts type nd 760 are used for displaying and storing the position of the m 420 using a personal computer .
the module mounting head is connected to several linear guides , goniometer and rotary tables to reach any position .
the module is fixed by vacuum to the mounting head and its position is always monitored by the microscope . for the deposition of the glue a computer controlled dispenser from efdtype 1502
the assembly time of each module is about 1 hour , the curing time of the glue is 2 hours .
this leads to a production rate of 1 stave per week .
the x , y , and z position of the glued modules of each stave is controlled with respect to the stave ruby balls .
these provide the reference system for the module position and allow one to assess the accuracy of the loading procedure . as a typical result the z position measurement for 13 staves is provided in figure [ fig : z - deviation_modwise ] . for each stave the deviation from the nominal position is shown for each of the 13 modules . the stated module
i d s are equivalent to defined locations on the stave .
the tolerances are @xmath5 and are indicated in the figure by thicker lines .
the plot shows that the accuracy of the z positioning is always within the tolerances of @xmath6 . in figure
[ fig : z - deviation_distribution ] , the distribution of the z - deviation is given and demonstrates that 50% of the modules are glued with an accuracy which is even better than @xmath7 .
one can also see that there is a systematic shift of 6 - 7@xmath8 . as mentioned previously , the difference between the bow of the module before and after loading is an indicator as to whether stress has been applied during the loading procedure .
figure [ fig : bow ] shows the mean bow difference ( mean value ) , averaged over all 13 modules of one stave , as well as the standard deviation ( sd ) and the average of the absolute values ( mean of amounts ) .
one can see , that the bow difference is always better than @xmath9 .
this is interpreted to show that a minimum of stress is being applied to the modules during the loading procedure .
a module loading machine for the atlas pixel detector has been successfully realized in wuppertal .
all requirements of the mounting precision are fulfilled .
the position of the modules after loading are well within the tolerances .
the applied stress during loading is negligible .
this work has been supported by the _ bundes - ministerium fr bildung , wissenschaft , forschung und technologie _ ( bmbf ) under grant number05 ha4px1/9 . atlas technical proposal cern / lhcc/94 - 43 , 15 december 1994 .
pixel detector technical design report cern / lhcc/98 - 13 .
atlas detector and physics performance technical design report cern / lhhc/99 - 14 .
b - layer ( atl - ip - cs-0009 ) , layer 1 and 2 specifications ( atl - ip - cs-0007 ) . pitch based carbon fibre ys80 ( data sheet ) , nippon graphite fiber ( ngf ) .
available from : http://plaza6.mbn.or.jp/~gf/english/products/space/ysa.html .
cyanate ester resin system ex1515 ( data sheet ) , bryte technologies inc .
available from : http://www.brytetech.com/pdf-ds/ex-1515.pdf . | the barrel part of the atlas pixel detector will consist of 112 carbon - carbon structures called `` staves '' with 13 hybrid detector modules being glued on each stave .
the demands on the glue joints are high , both in terms of mechanical precision and thermal contact . to achieve this precision
a custom - made semi - automated mounting machine has been constructed in wuppertal , which provides a precision in the order of tens of microns . as this is the last stage of the detector assembly providing an opportunity for stringent tests ,
a detailed procedure has been defined for assessing both mechanical and electrical properties .
this note gives an overview of the procedure for affixation and tests , and summarizes the first results of the production .
, , , , , , , , pixel detector , stave , mounting machine |
this is the third in a series of papers analyzing the structure and physical parameters of the shapley supercluster ( ssc ) , based on galaxy redshifts .
the first ( quintana et al . 1995 ; hereafter _ paper i _ ) presents and gives an initial analysis of the results of spectroscopic observations of the central region .
the second ( quintana , carrasco , & reisenegger 2000 , hereafter _ paper ii _ ) presents a much extended sample of galaxy redshifts and gives a qualitative discussion of the ssc s morphology . here
, we use dynamical collapse models applied to this sample , in order to obtain the mass of the central region of the ssc . an upcoming paper ( carrasco , quintana , & reisenegger 2000 , hereafter _ paper iv _ ) will analyze the individual clusters of galaxies contained in the sample , to obtain their physical parameters ( velocity dispersion , size , mass ) , search for substructures within the clusters , and determine the total mass contained within the virialized regions of clusters in the whole shapley area .
the shapley concentration ( shapley 1930 ) is the richest supercluster in the local universe ( zucca et al .
1993 , einasto et al . 1997 , but see also batuski et al .
this makes its study important for three main reasons .
first , its high density of mass and of clusters of galaxies provides an extreme environment in which to study galaxy and cluster evolution .
second , its existence and the fact that it is the richest supercluster in a given volume constrain theories of structure formation , and particularly the cosmological parameters and power spectrum in the standard model of hierarchical structure formation by gravitational instability ( e.g. , ettori , fabian , & white 1997 ; bardelli et al . 2000 ) . finally , it is located near the apex of the motion of the local group with respect to the cosmic microwave background .
thus it is intriguing whether the ssc s gravitational pull may contribute significantly to this motion , although most mass estimates ( e.g. , raychaudhury 1989 ; raychaudhury et al .
1991 ; paper i ; ettori et al . 1997 ; bardelli et al .
2000 ) make a contribution beyond a 10 % level very unlikely .
galaxy counts in redshift space ( bardelli et al .
2000 ) suggest that most of the supercluster has a density several times the cosmic average , while the two complexes within @xmath7 mpc of clusters a 3558 and a 3528 have overdensities @xmath8 and @xmath9 , respectively .
these regions are therefore far outside the `` linear regime '' of small density perturbations , but still far from being virialized after full gravitational collapse .
the same conclusions are easily reached by even a casual glance at the redshift structure presented in paper ii .
the density of these complexes indicates that they should be presently collapsing ( e.g. , bardelli et al . 2000 ) , and in the present paper we study this hypothesis for the main complex , around a 3558 , where substantially more data , with better areal coverage , are available ( see paper ii ) .
it should be pointed out that within each of these complexes we expect very large peculiar velocities , which dominate by far over their hubble expansion . in this case
, redshift differences among objects within each complex can give information about its dynamics , which we will analyze below , but essentially no information about relative positions along the line of sight ( except , perhaps , non - trivial information within a given dynamical model ) .
while this is generally acknowledged to be true within clusters of galaxies , where galaxy motions have been randomized by the collapse , it is sometimes overlooked on somewhat larger , but still nonlinear scales .
for instance , ettori et al .
( 1997 ) calculate three - dimensional distances between clusters in the shapley region on the basis of their angular separations and redshifts , and conclude that the group sc 1327 - 312 and the cluster a 3562 , at projected distances @xmath10 and @xmath11 mpc from the central cluster a 3558 , are between 5 and @xmath12 mpc from it in three - dimensional space , because of moderate differences in redshift .
however , there is evidence for interactions between these clusters and groups ( venturi et al .
1999 ) , suggesting true distances much closer to the projected distances .
the discrepancy is naturally explained by quite modest peculiar velocities of several hundreds of km s@xmath3 , easily caused by the large mass concentration . in the present paper ,
we analyze the region around a 3558 in terms of an idealized , spherical collapse model , which is used both in its pristine , but undoubtedly oversimplified , original form ( regs & geller 1989 ) , seen from a slightly different point of view , and in its less appealing , but possibly more accurate , modern fine - tuning calibrated by simulations ( diaferio & geller 1997 ; diaferio 1999 ) . section 2 explains the models , argues for the presence of velocity caustics , and gives the equations relating the caustics position to the mass distribution .
( some mathematical remarks regarding this relation are given in the appendix . ) in 3 we present the data , argue that velocity caustics are indeed present , and explain how we locate their position quantitatively .
section 4 presents and discusses our results , and 5 contains our main conclusions .
in this approach , we consider a spherical structure in which matter at any radius @xmath13 moves radially , with its acceleration determined by the enclosed mass @xmath14 . at a given time @xmath15 of observation , the infall velocity @xmath16 ( traced by galaxies participating in the mass inflow ) can give direct information about the mass profile ( kaiser 1987 ; regs & geller 1989 ) .
of course , @xmath17 is not directly observable . instead , for each galaxy we only observe its position on the sky , which translates into a projected distance from the assumed center of the structure , @xmath18 , and its redshift , which can be translated into a line - of - sight velocity @xmath19 with respect to the same center .
the `` fundamental '' variables @xmath13 and @xmath20 and the `` observed '' variables @xmath21 and @xmath19 ( with the line - of - sight velocity of the structure s center already subtracted ) are related by @xmath22^{1\over 2}u(r).\ ] ] contrary to the hubble flow observed on large scales , @xmath19 is negative ( approaching ) for the more distant galaxies on the back side of each shell , and positive ( receding ) for the closer galaxies on the front side . at
any given projected distance @xmath21 , one observes galaxies at many different true distances @xmath13 from the center .
the infall velocity @xmath17 decreases at large enough distances @xmath13 , reaching zero at a finite ( `` turnaround '' ) radius @xmath23 , and matching onto the hubble flow , @xmath24 for @xmath25 .
the projection factor increases from zero at @xmath26 ( galaxies moving perpendicularly to the line of sight ) , asymptotically approaching unity for @xmath27 .
therefore , there will be some maximum projected velocity @xmath28^{1\over 2}u(r),\ ] ] which is a monotonically decreasing function of @xmath21 ( see appendix ) , giving rise to caustics in the @xmath29 diagram with the characteristic `` trumpet shape '' described by kaiser ( 1987 ) and by regs & geller ( 1989 ) . in order to obtain the infall velocity @xmath17 from the galaxy redshifts ,
we first identify the caustics amplitude @xmath30 from the @xmath29 diagram by the procedure outlined in 3 .
given this relation , we can invert eq .
( 2 ) to obtain @xmath31^{1\over 2}}.\ ] ] this is an _
in_equality rather than an equality because for an arbitrary shape of @xmath17 it is _ not _ guaranteed that the shells at _ every _ @xmath13 will correspond to a maximum amplitude for some @xmath21 ( see appendix for a detailed mathematical discussion ) .
if the mass density decreases outwards , then the collapse occurs from the inside out , with innermost mass shells first reaching turnaround and recontraction , and outer shells following in succession .
any given shell will enclose the same mass @xmath32 at all times until it starts encountering matter that has already passed through the center of the structure and is again moving outwards .
the latter is only expected to happen in the very central , nearly virialized , part of the supercluster .
elsewhere , the dynamics of any given shell is described by the well - known parameterized solution @xmath33 ( e.g. , peebles 1993 , chapter 20 ) . here ,
@xmath34 and @xmath35 are constants for any given shell ( related to each other by the enclosed mass @xmath32 ) , and @xmath36 labels the `` phase '' of the shell s evolution ( initial `` explosion '' at @xmath37 , maximum radius or `` turnaround '' at @xmath38 , collapse at @xmath39 ) .
as we are observing many shells at one given cosmic time @xmath15 ( measured from the big bang , at which all shells started expanding , to the moment at which the structure emitted the light currently being observed ) , for each shell we can write @xmath40 where @xmath41 is the current value of the hubble parameter .
we can determine an upper bound on @xmath42 , and therefore on @xmath43 , by the procedure described above ( note that @xmath13 itself depends on the uncertainty in the cosmic distance scale ) .
the combination @xmath44 is a dimensionless constant , dependent on the cosmological model ( identified by dimensionless constants such as the density parameters @xmath45 ) . at the redshift of the ssc ( @xmath46 ) , it is likely to lie in the range @xmath47 , with the lower limit corresponding to an einstein - de sitter universe ( with critical matter density , @xmath48 , and no other ingredients ) , and the upper limit corresponding to an empty universe ( @xmath49 ) or a flat , low - density , @xmath50-dominated universe ( @xmath51 ) ( e.g. , peebles 1993 , chapter 13 ) . for an assumed value of this parameter ,
the left - hand side becomes fully determined , and the equation can be solved for the value of @xmath36 for each shell .
the equations can also be combined to yield @xmath52 the mass enclosed within the shell ( of current radius @xmath13 ) . for two reasons , _ a mass estimate obtained by applying this model to real data should be regarded as an upper bound on the true mass_. first , the model gives @xmath53 , an upper bound on @xmath17 , and this upper bound is used to derive the mass .
second , the caustics amplitude @xmath30 is amplified through random motions due to substructure within the infalling matter ( diaferio & geller 1997 ) . keeping this in mind , we will use the model to put a bound on the mass of the collapsing region around a 3558 . on the other hand ,
diaferio & geller ( 1997 ; see also diaferio 1999 ) have shown that the mass profile of structures forming in numerical simulations can be recovered to good precision from the formula @xmath54 there is no rigorous derivation for this result , although it can be justified heuristically by assuming that @xmath55 reflects the escape velocity at different radii , i.e. , that all galaxies within the caustics are gravitationally bound to the structure .
one has to assume further that the radial density profile lies between @xmath56 and @xmath57 ( diaferio & geller 1997 ; diaferio 1999 ) , as in the outskirts of simulated clusters of galaxies ( e.g. , navarro , frenk , & white 1997 ) .
this mass estimate is independent of the parameter @xmath44 , since no dynamical evolution is involved .
it has already been applied to the coma cluster ( geller , diaferio , & kurtz 1999 ) .
the @xmath29 diagram for the galaxies with available redshifts in the shapley region ( see paper ii ) , shown in fig . 1 for an adopted center at the main cluster a 3558 ( @xmath58 km s@xmath3 ) , indeed shows the predicted `` trumpet shape '' , with a maximum half - width @xmath59 km s@xmath3 , and extending cleanly out to @xmath60 mpc , and less cleanly to perhaps @xmath61 mpc from the center .
, assumed to be of purely cosmological origin , relative line - of - sight velocities are given by @xmath62 ( harrison & noonan 1979 ) . in an einstein - de sitter universe ( @xmath63 ) , physical distances projected on the sky are given by @xmath64/[h_0(1+z_c)]$ ] ( e.g. , peebles 1993 ) , where @xmath65 is the angular distance in radians .
again at the redshift of a 3558 , @xmath66 corresponds to @xmath67 mpc .
the latter result is fairly insensitive to the choice of cosmological model , therefore we adopt it in general . ] thus , there is indeed a coherent structure ( _ `` extended core '' _ ) , enclosing at least ( i.e. , within @xmath1 mpc ) 11 abell clusters ( abell , corwin , & olowin 1989 ) and three rich groups not included in the abell catalog ( see table 1 ) .
all of these have relative velocities with respect to a 3558 smaller than 1200 km s@xmath3 , with a median @xmath68 km s@xmath3 .
given its shape , this extremely cluster - rich structure is most likely due to gravitational collapse .
of course , the morphology of the ssc or even of its `` extended core '' does not support the assumption of spherical symmetry . the inner core ( formed by a 3558 , a 3556 , a 3562 , sc 1327 - 312 , and sc 1329 - 314 ) is clearly elongated , and there is evidence that both the galaxies and the clusters in the region of interest tend to form a planar structure ( paper i ; bardelli et al .
2000 ; paper ii ) .
this is supported by the elementary fact that ( within the extended core and a fair distance beyond it ) all the clusters in the north ( a 3559 , a 3557a , a 3555 , further away a 1736b ) have lower redshifts than a 3558 , while those in the south ( a 3560 , a 3554 , others at larger distances ) and west ( particulary those belonging to the _ nw filament , _ which extends towards a 3528 and is described in paper ii ) have higher redshifts .
it is also hinted at by the slight upturn of the structure in fig . 1 at @xmath69 ( @xmath70 mpc ) , which is due to the concentrations of clusters to the south - west and south - east of the central concentration .
( the downward - pointing arm at @xmath71 is due to the clusters a 1736a , b and a 3571 . ) however , 1 ) the extreme simplicity of this symmetry assumption compared to any possible improvement to it , 2 ) our ignorance of the true 3-d distribution of _ matter _ in the supercluster , and 3 ) the fact that the gravitational potential is generally more spherical than the mass distribution originating it , motivate us to still apply a spherical model .
note , in fact , that the model requires the gravitational potential to be spherically symmetric , but no symmetry assumptions are being made about the distribution of galaxies as tracer particles .
although the structure appears well - defined to the eye , it does not have a perfectly sharp edge , as it would be expected to have in the perfect spherical infall model .
departures from spherical symmetry , substructure such as clusters , the finite number of `` test particles , '' and ( to a much lesser degree ) observational uncertainties wash out and distort the structure .
this makes a determination of the caustics non - trivial . in the rest of this section ,
we follow diaferio s ( 1999 ) general approach in first smoothing the data , i.e. , obtaining a smooth estimate @xmath72 for the density of observed galaxies on the @xmath29 plane , and then applying a cut at some density contour which is taken to correspond to the caustics .
the details of how each of these steps is carried out differ slightly from diaferio s approach , and are discussed in the rest of this section . for a global analysis of the central ( collapsing ) region of the ssc
, we need to obtain a smooth estimate @xmath72 of the density of galaxies in the @xmath29 diagram .
density estimation has been discussed by many authors , such as silverman ( 1986 ) , and in the astronomical context pisani ( 1993 ; 1996 ) and merritt & tremblay ( 1994 ) .
diaferio ( 1999 ) applied density estimation to the particular problem of interest here . for @xmath73 data points ( galaxies ) with coordinates @xmath74 , he adopts the estimate @xmath75 where @xmath76 is a smooth , but centrally peaked , kernel function .
the ratio of smoothing lengths , @xmath77 is fixed , approximately equal to the ratio of observational uncertainties ( @xmath78 ) , and the individual values of , say , @xmath79 are chosen by an adaptive algorithm . any choice of smoothing lengths is a compromise between keeping as much structure as possible ( favoring small smoothing lengths ) while eliminating as much noise as possible ( favoring large values ) . the ideal compromise , though quite subjective in any case , depends on the density of data points , which generally varies over the volume being studied , motivating the choice of a different smoothing length for each data point . for our particular application ,
the density of points does not vary enormously over the area of interest , and we are only interested in the overall envelope of the structure , not in fine details .
therefore , we consider the additional computational effort of adaptive smoothing with different local smoothing lengths unjustified .
we apply fixed , overall smoothing lengths @xmath80 mpc , @xmath81 500 km s@xmath3 ( giving @xmath82 ) , chosen by eye to preserve the overall shape while minimizing the noise , and each corresponding to about 1/8 of the total extension of the structure studied . changing
either of the two lengths by a factor of 2 either way does not substantially change our results .
we note also that the chosen lengths are much larger than the respective uncertainties in the data , which therefore become irrelevant in determining the detected structures .
one problem with the smoothing kernel given above is that the data have a natural cutoff at @xmath83 , where they go abruptly from a fairly high ( near maximum ) density ( at @xmath84 ) to zero ( at @xmath85 ) .
when applied to points of small ( positive ) @xmath21 , the smoothing kernel extends to negative values ( where there are no data points ) , producing a decrease in @xmath72 when approaching @xmath83 from above .
this causes isodensity contours to narrow as @xmath86 , as seen , e.g. , in fig . 1 of geller , diaferio , & kurtz ( 1999 ) , in figs . 4 and 5 of diaferio ( 1999 ) , and in fig .
2a of the present paper .
this can be cured , e.g. , by making a mirror image of the data at @xmath85 and letting the smoothing kernel integrate over both the real data and their image ( fig .
2b ) . aside from correcting for the `` misbehavior '' at @xmath83
, this procedure gives results very similar to that of diaferio ( 1999 ) .
a more rigorous approach is suggested by merritt & tremblay ( 1994 ) , who deal with density estimation in circularly symmetric structures .
they focus on the surface density @xmath87 ( number per unit area ) rather than the radial density @xmath88 ( number per unit radial coordinate ) , and estimate @xmath87 with a circularly averaged kernel . in our case ( with one additional coordinate @xmath19 , which is not affected by this problem ) , we can define a number density per unit ( projected ) area per unit line - of - sight velocity as @xmath89 and estimate it through a kernel which is a product of a standard , one - dimensional kernel for @xmath19 and a circularly averaged kernel for @xmath21 . in particularly , fig . 2c shows the results of applying to our data a one - dimensional quadratic ( _ epanechnikov _ ) kernel for @xmath19 , and an annularly averaged , two - dimensional quadratic kernel for @xmath21 .
( see merritt & tremblay 1994 , eqs . 9a and 28a for explicit formulae . )
the density @xmath90 obtained from this procedure is well - behaved in all respects , decreasing from @xmath83 outwards .
indeed , it decreases so quickly that the density contours tend to close at fairly small radii , contrary to the visual impression from the data .
therefore , these contours are unlikely to be realistic representations of the velocity caustics .
a final alternative ( with the added virtue of reducing biases due to non - uniform spatial sampling ) is to normalize @xmath72 at each given @xmath21 with respect to the value at @xmath91 , i.e. , take a density estimate @xmath92 with the `` original '' @xmath72 determined by any of the other methods ( the case shown in fig .
2d is based on diaferio s estimator ) .
this estimator gives results very similar to the first two .
overall , we consider that the second procedure ( the `` mirror image '' density estimate ) is the one that most closely represents the visual appearance of the data , while at the same time having mathematically desirable properties ( smooth density contours slowly narrowing with increasing @xmath21 ) , and being close enough to diaferio s to permit a direct comparison of results .
therefore , we use the `` mirror image '' density estimation for the analysis that follows .
however , we stress that it is a very arbitrary choice , and that other choices may give quite different final results .
discarding the very different result based on the `` surface density '' scheme , the other procedures discussed above give masses that , at any given radius , differ by less than 10 % from the one obtained by the `` mirror image '' procedure .
given the estimated density @xmath93 , we now turn to finding the caustics which separate the collapsing structure from the galaxies in the foreground and background .
we are again inspired by diaferio ( 1999 ) , who uses a fixed density cutoff , @xmath94 , with the value of @xmath95 fixed by virial arguments applied to the most central region . in principle , taking a fixed value is somewhat arbitrary . by plotting in @xmath29 space , we are summing galaxies over annuli , and therefore a uniform background of galaxies would result in a density increasing @xmath96 , and therefore a fixed cutoff might include more and more of the background as @xmath21 increases .
the problem is worsened by the non - uniformly sampled data in our particular case .
therefore , a more natural and in principle better way to distinguish the structure from the background might be to fix on maxima of @xmath97 for each @xmath21 .
however , taking and maximizing a _ derivative _ of the numerically determined function is much noisier than just imposing a fixed cutoff .
therefore , we follow diaferio in adopting the latter approach . in order to choose the value of the cutoff , diaferio ( 1999 ) uses the virial theorem to relate the escape velocity ( in his interpretation represented by the velocity amplitude @xmath55 ) and the velocity dispersion @xmath98 of the central cluster , therefore writing @xmath99 , where the average is a galaxy number - weighted average over the region enclosed by the virial radius @xmath100 of the central cluster , for a given density cutoff @xmath95 . for a 3558
, the velocity dispersion is a relatively robust number ( @xmath101 km s@xmath3 ) , not sensitive to the radius of the sphere to be averaged over , and in the central region @xmath55 ( determined with any reasonable cutoff ) is also fairly radius - independent .
therefore , the dependence on the ( poorly determined ) virial radius is weak , and we arbitrarily choose it as @xmath102 mpc , and do a straight radial average ( _ not _ number - weighted ) to calculate @xmath103 and determine @xmath95 by diaferio s condition .
we tested the validity of diaferio s condition by the following procedure .
figure 3 shows the area of the @xmath29 diagram enclosed by contour levels with different @xmath95 .
we clearly distinguish 3 regimes : * at very low densities , the enclosed area is most of the diagram , therefore enclosing much of the background , not belonging to the structure .
the area rapidly decreases as the threshold density is increased . * at intermediate densities , the decreasing curve becomes much flatter ( @xmath104 constant ) , and we interpret this as having most of the background excluded and probing progressively denser parts of the structure .
this is confirmed by watching the contour plots , which indeed trace the boundaries of the structure , and become progressively tighter . *
finally , at high values of @xmath95 , only a few isolated peaks in the structure are left enclosed , and these finally disappear when @xmath95 reaches the maximum density present .
this analysis suggests choosing the threshold at the transition between regimes ( a ) and ( b ) .
this falls close to the threshold value chosen by diaferio s ( 1999 ) condition as outlined above , marked by the vertical line .
this strengthens the argument for diaferio s choice of cutoff , already used to choose the contours in fig . 2 , and adopted hereafter .
the chosen density contour of course gives two values of @xmath19 ( one positive , @xmath105 , and one negative , @xmath106 ) for each value of @xmath21 . in the ssc
, it turns out that the upper contour is much `` cleaner '' ( separating a dense region from a nearly empty one ) , therefore we simply adopt @xmath107 rather than diaferio s prescription @xmath108
fig . 4 shows the enclosed mass as a function of radius , @xmath14 , as determined by the two methods discussed in 2 , together with a third determination , namely the cumulative mass of the clusters enclosed in the given radius , given in table 1 .
note that the mass estimates @xmath109 , taken from ettori et al .
( 1997 ) for the most important clusters , are masses within a radius enclosing an average density 500 times the critical density @xmath110 .
this is substantially higher than the standard `` virialization density '' of @xmath111 , and therefore gives a conservative lower limit to the total virialized mass , which may be increased by a factor @xmath112 for a more realistic estimate .
several comments are in order : \1 ) as discussed above , the pure spherical infall model is highly idealized and , even if correct , can only give an upper bound on the mass within any given radius .
therefore , the upper ( dot - dashed ) curve , corresponding to pure spherical collapse in a cosmological model with @xmath113 should be regarded as a fairly robust upper limit to the mass within any given radius .
\2 ) the model of diaferio & geller ( 1997 ) has been calibrated against simulations . applied to the infall regions around clusters of galaxies ,
it should in principle give the correct mass to within about @xmath114 ( geller et al .
however , it assumes a density profile decreasing at least as fast as @xmath115 , which may not apply to the very noisy region around a 3558 , which contains a number of other , fairly rich clusters .
it appears surprising that the mass profile it gives for this region is quite similar ( both in shape and in amplitude ) to that obtained by geller et al .
( 1997 ) , considering that coma is a quite massive cluster ( as massive or perhaps even more massive than a 3558 ) , but is _ not _ surrounded by any other massive clusters .
\3 ) the mass estimate based on individual cluster masses is uncertain for two reasons .
first , of course it does not consider the mass in the non - virialized outskirts of the clusters or not associated with clusters at all , and therefore it would be expected to underestimate the total mass . on the other hand , in the absence of information on the three - dimensional distance @xmath13 of each cluster to a 3558 , and given that the velocity does not give reliable distance information within the collapsing structure , each cluster was put at its projected radius @xmath18 , and therefore contributes to the enclosed mass already at radii smaller than its true position .
therefore , the mass in virialized clusters within any given radius @xmath14 is overestimated by the projection into radius @xmath13 of clusters actually at larger radii . given these caveats
, there seems to be fair agreement among the different mass determinations , and it seems safe to say that the mass enclosed by radius @xmath116 mpc lies between @xmath117 and @xmath118 .
it is interesting , nevertheless , that diaferio s method gives results that differ so little from the lower limit to the virialized mass in clusters .
therefore , if diaferio s method is applicable to the ssc , the either there would be very little mass outside the inner , virialized parts of clusters of galaxies in this region , or the cluster mass estimates would have to be systematically high . for comparison , ettori et al .
( 1997 ) used three different mass estimates , namely : 1 ) the sum of the gravitational masses of clusters as obtained from their x - ray emission profile , @xmath119 , 2 ) the total mass expected to be associated with the baryons observed in clusters , @xmath120 , and 3 ) the mass obtained from applying the virial theorem to the enclosed clusters , used as test particles , @xmath121 .
they applied these methods to four progressively larger structures , each enclosing the previous one .
the one most similar to our @xmath1 mpc sphere appears to be the second , enclosing 12 clusters , and with a nominal 3-d radius of @xmath122 mpc , obtained by treating redshift as a third coordinate , which we have argued to give an overestimated depth in the collapsing region . for this region , they find @xmath123 , @xmath124 and @xmath125 , all in units of @xmath126 .
the first two are likely underestimates ( as they consider only the matter in the virialized regions of clusters observed in x - rays ) , and the third is completely uncertain , given the uncertain distances along the line of sight and the fact that the ssc is not virialized ( but see small et al .
1998 for a modern , more careful application of the virial theorem to the corona borealis supercluster ) .
thus , it is reasonable that we find a somewhat higher mass for the ( optically observed ) clusters , and a possibly much higher total dynamical mass , as suggested by the spherical collapse model . the average enclosed density ( see fig .
5 ) drops from a value 400 500 times the critical density within @xmath127 mpc ( consistent with the presence of a massive , already collapsed and virialized cluster ) to a value still several times critical ( @xmath128 to 23 times , depending on the model ) within our outermost radius , @xmath1 mpc . from galaxy counts in redshift space ,
bardelli et al . ( 2000 ) find an overdensity @xmath129 within a region of equivalent radius @xmath130 mpc . assuming that galaxies trace mass , this might in principle allow us to determine the universal matter density parameter @xmath131 in practice , however , the uncertainty in this estimate is still much too large to put a useful constraint on @xmath132 .
6 shows that , in the spherical collapse model , the whole structure within @xmath1 mpc has already been contracting for more than 1/3 of its lifetime , and the inner regions are in the final stages of collapse , consistent with the presence of a massive cluster .
even if there were no additional mass beyond @xmath1 mpc , the current turnaround radius would be at @xmath133 mpc , and the bound region ( to collapse eventually ) would extend to @xmath134 mpc , enclosing essentially the whole supercluster , including the strong concentration around a 3528 , a 3530 , and a 3532 .
however , the much lower enclosed densities in the diaferio & geller model would imply that the @xmath1 region around a 3558 is ( at best ) only now reaching turnaround , and has another hubble time to go for final collapse .
as discussed in paper i , the mass required at the distance of the ssc to produce the observed motion of the local group with respect to the cosmic microwave background is given by @xmath135 the mass within @xmath136 mpc can therefore produce at most @xmath137 of the observed local group motion , which makes it unlikely that even the whole ssc would dominate its gravitational acceleration .
unfortunately , not much can be said about the regions beyond a radius of @xmath138 mpc from the center .
it seems safe to assert , though , that the collapsing region does not extend far beyond , say , @xmath61 mpc .
this gives an upper bound on the average density enclosed in larger radii , @xmath139 .
thus , in order to produce the peculiar velocity of the local group , one would need a region of radius @xmath140 which , for the range of cosmological values considered before , corresponds to a lower limit @xmath141 mpc .
therefore , in the unlikely case that the whole ssc ( characterized in paper ii ) were on the verge of gravitational collapse , it would be able to produce on its own the observed peculiar velocity of the local group .
this statement ignores , of course , that the apex of the local motion does not point exactly at the ssc , and therefore some additional contribution is necessary in any case .
we have presented the ( to our knowledge ) first application of a plausible dynamical model to a supercluster of galaxies , containing a substantial number of clusters .
the central @xmath1 mpc region of the shapley spercluster ( and probably a much more extended region surrounding it ) is argued to be currently collapsing under the effect of its own gravity .
its mass , although uncertain due to idealizations in the model , indicates a large enhancement over the average density of the universe , although still far from that required to produce the local group s observed motion with respect to the cosmic microwave background .
the authors thank a. diaferio for interesting discussions at the first princeton - u .
catlica astrophysics workshop , _ the cosmological parameters @xmath45 , _ held in pucn , chile , in january 1999 , and for extensive e - mail exchanges thereafter .
a previous version of the kernel smoothing program used here was due to a. meza .
we also thank him and r. benguria for useful discussions .
this work was financially supported by fondecyt grant 8970009 ( _ proyecto de lneas complementarias _ ) , and by a presidential chair in science awarded to h. quintana .
e.r.c . was funded by fapesp ph.d .
fellowship 96/04246 - 7 .
in order to derive and understand intuitively the properties of the observed velocity amplitude @xmath30 and the infall velocity @xmath17 producing it in the spherical model , it is convenient to define new variables @xmath142 , @xmath143 , @xmath144 , and @xmath145 .
then , note that , seen as a function of @xmath148 for given @xmath100 , @xmath149 is a straight line intersecting the horizontal axis at @xmath150 and the vertical axis at @xmath151 .
therefore , the function @xmath152 defines a set of straight lines whose upper envelope gives the function @xmath153 ( see fig .
the relation between the functions @xmath152 and @xmath153 is very similar to the _ legendre transformation _
( e.g. , courant & hilbert 1989 ,
i.6 ) , and shares many of its properties . since @xmath154 for all @xmath100 , *
@xmath153 is positive ( @xmath155 ) , monotonically decreasing ( @xmath156 ) , and convex ( @xmath157 ) . *
8 shows the function @xmath153 obtained from the data in the way described in this paper .
it is clear that it does not strictly satisfy the conditions of monotonicity and convexity , indicating that , as expected , the pure spherical infall model does not exactly represent the data .
@xmath158 is a strictly increasing function , but it is not necessarily continuous .
for example , it can happen that , for some point @xmath164 , the maximum occurs at the intersection of two straight lines labeled by @xmath166 and @xmath167 ( see fig .
9 ) , with the lines corresponding to all other values of @xmath100 lying below it , i.e. , this is an equality for those @xmath100 which correspond to a maximum @xmath149 for some @xmath148 , and a strict inequality in all other cases .
the latter can in principle be diagnosed by realizing that , in the case discussed above , so both @xmath158 and @xmath172 are discontinuous at @xmath159 . in practice , with noisy data , a discontinuity in @xmath172 is difficult to detect , and the inequality , eq .
( a5 ) , has to be used as such .
it is interesting to note that , taking the reciprocal value of all variables , one can write this has the same form as eq .
( a1 ) . we can conclude that @xmath174 has the same properties as @xmath153 , being positive , monotonically decreasing , and convex , in fact , it is the convex and decreasing lower envelope of @xmath175 . as long as the latter is itself convex and decreasing , then @xmath176 , while in general @xmath177 .
10 shows @xmath174 as obtained from the data .
its curved parts ( here absent ) and `` corners '' are expected ( within the pure spherical collapse model ) to correctly estimate @xmath175 , while the straight segments are lower bounds .
venturi , t. , morganti , r. , bardelli , s. , dallacasa , d. , & hunstead , r. w. 1999 , in _
observational cosmology : the development of galaxy systems , _
g. giuricin , m. mezzetti , and p. salucci , eds .
, astronomical society of the pacific , vol .
176 , p. 256
cccccc a 3558 & 0.00 & 0 & 6.09 & 6.09 & 1 + sc 1327 - 312 & 0.94 & + 700 & 2.14 & 8.23 & 1 + a 3556 & 1.92 & -16 & 1.82 & 10.05 & 1 + sc 1329 - 314 & 1.93 & -956 & 1.03 & 11.08 & 1 + a 3562 & 2.94 & -12 & 3.06 & 14.14 & 1 + a 3560 & 4.36 & + 144 & 2.72 & 16.86 & 1 + a 3552 & 4.48 & + 1153 & 0.36 & 17.22 & 2 + a 3559 & 4.65 & -376 & 0.83 & 18.05 & 1 + a 3554 & 6.13 & + 99 & 1.26 & 19.31 & 2 + a 3557a & 6.18 & -114 & 0.47 & 19.78 & 2 + a 3555 & 7.07 & -419 & 0.10 & 19.88 & 2 + sc 1342 - 302 & 7.77 & + 119 & 0.20 & 20.08 & 2 + a 724s & 7.86 & + 367 & 0.88 & 20.96 & 2 + a 726s & 7.89 & + 206 & 0.59 & 21.55 & 2 + | we present the first application of a spherical collapse model to a supercluster of galaxies .
positions and redshifts of @xmath0 galaxies in the shapley supercluster ( ssc ) are used to define velocity caustics that limit the gravitationally collapsing structure in its central part .
this is found to extend at least to @xmath1 mpc of the central cluster , a 3558 , enclosing 11 aco clusters .
infall velocities reach @xmath2 km s@xmath3
. dynamical models of the collapsing region are used to estimate its mass profile .
an upper bound on the mass , based on a pure spherical infall model , gives @xmath4 for an einstein - de sitter ( critical ) universe and @xmath5 for an empty universe .
the model of diaferio & geller ( 1997 ) , based on estimating the escape velocity , gives a significantly lower value , @xmath6 , very similar to the mass found around the coma cluster by the same method ( geller et al .
1999 ) , and comparable to or slightly lower than the dynamical mass in the virialized regions of clusters enclosed in the same region of the ssc . in both models ,
the overdensity in this region is substantial , but far from the value required to account for the peculiar motion of the local group with respect to the cosmic microwave background . |
the tumor suppressor p53 plays a central role in cellular responses to various stress , including oxidative stress , hypoxia , telomere erosion and dna damage @xcite . in unstressed cells ,
p53 is kept at low level via its negative regulator mdm2 @xcite . under stressed conditions , such as dna damage
, p53 is stabilized and activated to induce the express of downstream genes , including p21/waf1/cip1 and gadd45 that are involved in cell cycle arrest , and puma , bax and pig3 that can induce apoptosis @xcite .
the cell fate decisions after dna damage are closely related to the p53 dynamics that is regulated by p53-mdm2 interactions @xcite .
oscillations of p53 level have been observed upon ir induced dna damage at the population level in several human cell lines and transgenic mice @xcite .
more interestingly , pulses of p53 level were revealed in individual mcf7 cells , and it was suggested that the cell fate is governed by the number of p53 pulses , i.e. , few pulses promote cell survival , whereas sustained pulses induce apoptosis @xcite .
fine control of p53 dynamics is crucial for proper cellular response .
mutations and deregulation of p53 expression have been found to associate with various cancer types @xcite . programmed cell death 5 ( pdcd5 ) , formerly referred to as tfar19 ( tf-1 cell apoptosis - related gene 19 ) , is known to promote apoptosis in different cell types in response to various stimuli @xcite . decreased expression of pdcd5 has been detected in various human tumors @xcite , and restoration of pdcd5 with recombinant protein or an adenovirus expression vector can significantly sensitive different cancers to chemotherapies @xcite .
pdcd5 is rapidly upregulated after dna damage , interacts with the apoptosis pathway , and translocates from the cytoplasm to nucleus during the early stages of apoptosis @xcite . recently
, novel evidence indicates that pdcd5 is a p53 regulator during gene expression and cell cycle @xcite .
it was shown that pdcd5 interacts with p53 by inhibiting the mdm2-mediated ubiquitination and accelerating the mdm2 degradation .
hence , upon dna damage , pdcd5 can function as a co - activator of p53 to regulate cell cycle arrest and apoptosis .
a series of computational models have been constructed to investigate the mechanism of p53-mediated cell - fate decision @xcite . in these models
the p53-mdm2 oscillation was considered to be crucial for cell - fate decision after dna damage . after dna damage , such as double strand breaks ( dsbs ) , active atm monomer ( atm * )
become dominant . in the nucleus , atm * active p53 in two ways : atm * promotes the phosphorylation of p53 at ser-15 @xcite and accelerates the degradation of mdm2 in nucleus ( @xmath0 ) @xcite . in the cytoplasm
, active p53 induces the production of @xmath1 , which in turn promotes the translation of _ p53 _ mrna to form a positive feedback .
the p53-mdm2 oscillation is a consequence of the two coupled feedback loops : the negative - feedback between p53 and @xmath0 , and the positive - feedback between p53 and @xmath1 .
recently , mathematical models of how pdcd5 interacts with the dna damage response pathway to regulate cell fate decisions have been developed @xcite . in our previous study
@xcite , a model for the effect of pdcd5 to the p53 pathway has been established through a nonlinear dynamics model that is developed based on the module of p53-mdm2 oscillator in @xcite and experimental findings in @xcite .
it was shown that the p53 activity can display different dynamics after dna damage depending on the pdcd5 level .
the p53 protein shows low activity in case of pdcd5 deletion , sustain intermediate level for medial pdcd5 expression , and pulses when the pdcd5 level is upregulated @xcite .
nevertheless , little is known about the global p53 dynamics upon pdcd5 interactions with changes in the expression levels of p53 and pdcd5 , which is often seen in tumors . here ,
we investigate , based on the mathematical model proposed in @xcite , how the dynamics of p53 activity after dna damage depend on changes in the levels of p53 production and pdcd5 .
a global bifurcation analysis shows that p53 activity can display various dynamics depending on the p53 production and pdcd5 levels , including monostability , bistability with two stable steady states , oscillations , and co - existence of a stable steady state and an oscillatory state .
these dynamics are further investigated through the method of potential landscape .
the stability of the oscillation states are characterized by the potential force and the probability flux .
we also discuss the effect of pdcd5 efficiency on p53 dynamics .
pdcd5 efficiency can induce p53 oscillation by hopf bifurcation , and the transition of p53 dynamics near the hopf bifurcation are charaterized by barrier height and energy dissipation .
figure . [ fig:1 ] illustrates the model of p53-mdm2 oscillator with pdcd5 regulation studied in this paper . here
we summarize the model equations and refer @xcite for details . , which in turn promote the translation of _ p53 _ mrna . in the nucleus
, active p53 is degraded by binding to @xmath0 , and the interaction is disrupted by pdcd5 . both active atm and pdcd5 are able to accelerate the degradation of @xmath0 .
mdm2 in the nucleus and cytoplasm can shuttle between the two compartments .
refer to the text and @xcite for details.,width=302 ] the model equations are composed of three components : active p53 in the nucleus @xmath2 $ ] , mdm2 in the nucleus @xmath3 $ ] and in cytoplasm @xmath4 $ ] . after dna damage
, atm is activated and promotes the degradation of @xmath0 so that the p53 level is unregulated .
active p53 promotes the production of @xmath1 , which promotes the translation of _ p53 _ mrna to produce p53 and hence form a positive - feedback loop . in the nucleus , active p53 is degraded slowly at a rate @xmath5 by weakly binding to @xmath0 , and the interaction is disrupted by pdcd5 .
moreover , pdcd5 can also accelerate the degradation of @xmath0 .
mdm2 in the nucleus and cytoplasm can shuttle between the two compartments at rates @xmath6 and @xmath7 , respectively .
these interactions give the following differential equations : @xmath8}{dt}&=&v_{\mathrm{p53}}([\mathrm{mdm2}_{\mathrm{cyt}}])-d_{\mathrm{p53}}([\mathrm{mdm2}_{\mathrm{nuc}}])\,[\mathrm{p53}]\nonumber\\ & = & f_{1}([\mathrm{p53}],[\mathrm{mdm2}_{\mathrm{cyt}}],[\mathrm{mdm2}_{\mathrm{nuc } } ] ) , \\ \label{eq : mc } \frac{d[\mathrm{mdm2}_{\mathrm{cyt}}]}{dt}&=&v_{\mathrm{mdm2}}([\mathrm{p53}])-k_{\mathrm{in}}\,[\mathrm{mdm2}_{\mathrm{cyt}}]\nonumber\\ & & { } + k_{\mathrm{out}}\,[\mathrm{mdm2}_{\mathrm{nuc}}]-d_{\mathrm{mdm2}}\,[\mathrm{mdm2}_{\mathrm{cyt}}]\nonumber\\ & = & f_{2}([\mathrm{p53}],[\mathrm{mdm2}_{\mathrm{cyt}}],[\mathrm{mdm2}_{\mathrm{nuc } } ] ) , \\ \label{eq : mn } \frac{d[\mathrm{mdm2}_{\mathrm{nuc}}]}{dt}&=&k_{\mathrm{in}}\,[\mathrm{mdm2}_{\mathrm{cyt}}]-k_{\mathrm{out}}\,[\mathrm{mdm2}_{\mathrm{nuc}}]\nonumber\\ & & { } -f(t)\ , d_{\mathrm{mdm2}}\,[\mathrm{mdm2}_{\mathrm{nuc}}]\nonumber\\ & = & f_{3}([\mathrm{p53}],[\mathrm{mdm2}_{\mathrm{cyt}}],[\mathrm{mdm2}_{\mathrm{nuc } } ] ) .
\end{aligned}\ ] ] here @xmath9 are constants , and other rate functions are given by @xmath10)&= & \bar{v}_{\mathrm{\mathrm{p53}}}\,\big[(1-\rho_1 ) + \rho_1 \,\dfrac{{[\mathrm{mdm2_{cyt}}]}^{s_1}}{k_1^{s_1}+ {
[ \mathrm{mdm2_{cyt}}]}^{s_1}}\big],\\ d_{\mathrm{p53 } } ( { [ \mathrm{mdm2_{nuc } } ] } ) & = & \bar{d}_{\mathrm{p53}}\,\big[(1-\rho_2)\nonumber\\ & & { } \quad\quad+ \rho_2\,\dfrac{{[\mathrm{mdm2_{nuc}}]}^{s_2}}{k_2(p(t))^{s_2}+ [ \mathrm{mdm2_{nuc}}]^{{s_2}}}\big],\\ v_{\mathrm{mdm2}}({[\mathrm{p53 } ] } ) & = & \bar{v}_{\mathrm{mdm2}}\,\big[(1-\rho_3 ) + \rho_3\,\dfrac{{[\mathrm{p53}]}^{s_3}}{k_3^{s_3}+ { [ \mathrm{p53}]}^{s_3}}\big],\\ f(t ) & = & \bar{f}\,\big[(1-\rho_{4}-\rho_{5 } ) + \rho_{4}\,\dfrac{p(t)^{s_{4}}}{k_4^{s_{4}}+p(t)^{s_{4}}}\\ & & { } + \rho_{5}\ , \dfrac{a(t)^{s_{5}}}{k_5^{s_{5}}+a(t)^{s_{5}}}\big ] , \end{aligned}\ ] ] where @xmath11 here @xmath12 and @xmath13 are time dependent functions for the levels of active atm and pdcd5 in nucleus , respectively . in this study
, we only consider the process of dna repair after dna damage that can persist for as long as 10 hours before the onset of apoptosis @xcite . both active atm and pdcd5 level
are upregulated after dna damage .
hence , we set @xmath14 to mimic this process in accordance with the simulation given by @xcite .
similarly , during dna repair , we assume @xmath15 with @xmath16 represents the pdcd5 expression level .
we refer table [ tab:1 ] for parameter values in this study .
.typical parameter values ( the unit for time is ` min ` , and arbitrary unit for the concentration)(ref .
@xcite ) . [ cols= " < ,
< , < , < , < , < , < " , ] [ tab:1 ] the ideas of potential landscape have been introduced for uncovering global principles in biology for protein dynamics @xcite , interactions @xcite , and gene networks @xcite . for a non - equilibrium open systems such as the p53-mdm2 oscillator that exchanges energies and informations with outside environments , the potential and non - equilibrium probability flux are able to reveal insights for the global robustness and physical mechanisms of the non - equilibrium interactions . here
we summarize the potential landscape and flux framework that were introduced by wang _
et al . _ in @xcite .
we write the dynamical equation of the p53-mdm2 oscillator as @xmath17 , where @xmath18,[\mathrm{mdm2_{cyt}}],[\mathrm{mdm2_{nuc}}])$ ] and @xmath19 represents the right hand size of - .
the above equation can be extended to include fluctuations by a probability approach : @xmath20 , where @xmath21 is the noise perturbation .
the statistical nature of the noise is often assumed as gaussian ( large number theorem ) and white ( no memory ) : @xmath22 and @xmath23 , where @xmath24 is the correlation tensor ( matrix ) measuring the noise strength .
the probability of system states @xmath25 evolves following the fokker - plank equation @xmath26 , where the flux vector @xmath27 is defined as @xmath28 . the flux @xmath27 measures the speed of the flow in the concentration space .
at the stationary state , @xmath29 , then @xmath30 .
there are two possibilities : one is @xmath31 , which implies detailed balance , and the other is @xmath32 so that the detailed balance is broken and the system is at the non - equilibrium state . at the stationary state ,
the probability flux is defined by @xmath33 where @xmath34 is the probability density at stationary state .
hence , @xmath35 here @xmath36 is defined as the non - equilibrium potential . from ,
we have divided the force @xmath37 into two parts : the potential force @xmath38 and curl flux force @xmath39 . at detail balance , the curl flux force is zero . for a non - equilibrium open system ,
both potential landscape and the associated flux are essential in characterizing the global stationary state properties and the system dynamics .
to investigate how pdcd5 interacts with p53 to regulate cell fate decision dynamics after dna damage , we performed codimension - two bifurcation analysis with respect to the two parameters @xmath40 and @xmath41 .
the bifurcation diagrams were computed with auto incorporated in xppaut @xcite .
the main bifurcation diagram is shown at fig . [
fig : bif2 ] and detailed below , with fig .
[ fig : bif2]b the enlarge of the dashed square region in fig .
[ fig : bif2]a .
the parameter plane @xmath42 is divided into ten regions labelled _ i__x _ each with different dynamical profiles and marked with different colors in fig .
[ fig : bif2 ] ( refer fig . [
fig : bif2]b for regions @xmath43-@xmath44 ) .
typical dynamics of each region is shown at fig .
[ fig : dyn ] and described below : 1 .
i _ corresponds to monostability with a single stable steady state of low p53 activity despite the expression level of pdcd5 .
2 . in region _ ii _
there is a stable steady state and two unstable steady states .
i _ and _ ii _ are separated by a curve @xmath45 of saddle - node bifurcation , across which an unstable node and a saddle appear due to fold bifurcation .
3 . region _ iii _
is on the right of region _
ii_. crossing the curve @xmath46 from region _ ii _ , the stable node and the saddle collide and disappear , accompanied by the emergence of a stable limit cycle that surrounds an unstable steady state .
meanwhile , parameters can also change from region _
i _ to _ iii _ crossing the boundary @xmath47 or @xmath48 , respectively ( fig . [
fig : bif2]b ) .
the curve @xmath47 represents supercritical hopf bifurcation at which the stable steady state in region
_ i _ becomes unstable and a stable limit cycle appears .
the curve @xmath48 represents subcritical hopf bifurcation . before crossing the curve @xmath49 , two limit cycles , stable and unstable ones ,
come out due to fold bifurcation of limit cycles that is represented by the curve @xmath50 ( fig .
[ fig : bif2]b ) .
next , the unstable limit cycle disappear and the stable steady state lose its stability when at the curve @xmath49 .
4 . regions _ iii _ and _ iv _ are separated by a subcritical hopf bifurcation curve @xmath51 by which an unstable limit cycle appears and locates between the stable limit cycle and the steady state , and the steady state that is unstable at region _
iii _ becomes stable at region _
region _ v _
, similar to region _
i _ , has a stable steady state which corresponds to a state of high p53 concentration .
the two limit cycles at region _ iv _ collide and disappear when parameters cross the fold bifurcation curve @xmath52 that separates regions _ iv _ and _ v_. 6 .
three steady states arise in region _
vi _ , two of them are unstable and appear while crossing the fold bifurcation curve @xmath53 of equilibria between regions _ vi _ and _ v_. 7 .
region _ vii
_ gives bistability in p53 activity , including two stable steady states with low or high p53 concentrations , and an unstable steady state .
there are two ways to reach this bistable region . from region _
vi _ and crossing the subcritical hopf bifurcation @xmath49 , an unstable limit cycle emerges and the unstable steady state with low p53 level becomes stable , but the unstable limit cycle immediately disappear due to homoclinic bifurcation ( refer the curve @xmath54 in fig .
[ fig : bif2]a which is very close to @xmath49 ) . from region _
i _ and crossing the boundary @xmath55 , a pair of stable and unstable steady states appear due to fold bifurcation of equilibria . 8 .
there are three regions ( _ viii_-_x _ ) around @xmath56 which share the same phase diagram with a stable limit cycle and three steady states ( fig .
[ fig : bif2]b ) . in region _
the three steady states are unstable , two of them ( a saddle and an unstable focus ) come from the fold bifurcation ( @xmath57 ) that separates regions _ iii _ and _ viii_. in region _
ix _ the steady state with high p53 level becomes stable and is surrounded by an unstable limit cycle .
the change is originated from the subcritical hopf bifurcation @xmath51 from region _
viii_. finally , the unstable limit cycle disappear in region _ x _ due to the homoclinic bifurcation @xmath58 between regions _ ix _ and _ x_. we note that in regions _ ix _ and _ x _ , there are bistable states of a stable steady state and a stable limit cycle , similar to the region _ iv_. in addition to the above ten regions , there are four codimension - two bifurcation points denoted by black dots in fig .
[ fig : bif2 ] : two _ cusp _ points ( @xmath59 and @xmath60 ) , one generalized hopf bifurcation ( gh ) and one bogdanov - takens bifurcation ( bt ) .
the _ cusp _
point @xmath59 locates at @xmath61 where the fold bifurcation curve @xmath62 and the saddle - node homoclinic bifurcation curve @xmath46 meet tangentially .
the other _ cusp _ point @xmath60 is given by @xmath63 by the two fold bifurcation curves @xmath64 and @xmath53 , where two fold bifurcations coalesce and disappear .
the point gh at @xmath65 corresponds to the meeting point of @xmath66 and @xmath49 from where a fold bifurcation curve of the limit cycle ( @xmath50 ) occurs . at the bt point
@xmath67 the fold bifurcation curve @xmath64 and @xmath51 meet tangentially to give a homoclinic bifurcation curve @xmath58 .
the bt point separates @xmath64 into two segments @xmath55 and @xmath57 . crossing @xmath55 from left to right gives a stable equilibrium and a saddle , while crossing @xmath57 gives an unstable equilibrium and a saddle @xcite . in summary , by manipulating the expression level of pdcd5 and the maximum production rate of p53 ,
the system displays four types of stable dynamics : a single stable steady state ( regions _ i _ , _ ii _ , _ v _ , and _ vi _ ) , two stable steady states ( region _ vii _ ) , a stable limit cycle ( regions _ iii _ and _ viii _ ) , and coexistence of a stable steady state and a stable limit cycle ( regions _ iv _ , _ ix _ , and _ x _ ) .
the bifurcation diagram is zoomed out by codimension - one bifurcations with seven @xmath40 values marked by @xmath68-@xmath69 in fig .
[ fig : bif2 ] and are detailed at the next section . from fig .
[ fig : bif2 ] , during dna damage and when the pdcd5 is upregulated ( @xmath70 ) , the system shows either low p53 activity , p53 oscillation , or sustained high p53 level with the increasing of p53 production rate .
p53 oscillations have been known to be essential for cellular response to dna damage including dna repair and apoptosis @xcite .
our analyses suggest that proper cell response to dna damage with p53 oscillation is possible only when pdcd5 is highly expressed and the p53 production rate takes proper value . to further investigate the oscillation dynamics , fig .
[ fig:4 ] shows the period and amplitude of the stable limit cycles corresponding to regions _ iii _ and _ iv _ in fig .
[ fig : bif2 ] .
the results show that the oscillation periods decrease with the maximum p53 production rate ( @xmath41 ) , while the amplitudes increase with @xmath41 .
this is consistent with the experimental observations @xcite .
however both periods and amplitudes are insensitive with the pdcd5 level over a wide parameter range .
these results indicates that upregulated pdcd5 and a proper p53 expression level are essential for producing proper p53 oscillation that regulates the cellular response to dna damage .
when pdcd5 is downregulated ( @xmath71 ) , the system has either a single steady state with low p53 activity or bistability with either low or high p53 activity after dna damage depending on the maximum p53 production rate .
experiments have shown that cells exposed to sustained p53 signaling frequently underwent senescence @xcite . .
, width=340 ] to get a clear insight into the codimension - two bifurcation diagram in fig .
[ fig : bif2 ] , we considered a codimension - one bifurcation of the concentration of p53 with respect to the p53 production rate @xmath72 . the bifurcation diagram is shown at fig .
[ fig:5 ] , with panels a - g for given values @xmath73 , respectively , as marked at fig .
[ fig : bif2 ] . diagrams for each @xmath16 value is detailed below . 1 . for low pdcd5 level ( @xmath74 ) ,
the equilibrium show a s - shaped bifurcation diagram with bistability ( fig . [
fig:5]a ) .
there are three branches of the equilibrium depending on @xmath41 .
the upper branch is composed of stable nodes , states at the middle branch are saddles , while the lower branch consists of stable nodes and foci which are separated near the subcritical hopf bifurcation point @xmath75 ( @xmath76 ) .
there is a fold bifurcation of equilibria @xmath77 ( @xmath78 ) where a stable node and a saddle appear , so that bistability occurs with @xmath41 between @xmath79 and @xmath80 . with the increasing of @xmath72 ,
an unstable limit cycle appears from a homoclinic bifurcation point @xmath81 at a saddle ( @xmath82 ) , and the unstable limit cycle disappears via the subcritical hopf bifurcation @xmath75 . crossing @xmath80 , stable focus at the lower branch change to unstable one .
the saddle and the lower unstable focus collide and disappear via another fold bifurcation @xmath83 at @xmath84 .
when @xmath40 increases ( @xmath85 ) ( fig .
[ fig:5]b ) , the fold bifurcation @xmath77 becomes larger than the subcritical hopf bifurcation @xmath75 so that the bistability of equilibrium vanishes . with the increasing of @xmath41 ,
an unstable limit cycle arises from the hopf bifurcation @xmath80 .
the limit cycle becomes stable at the fold bifurcation @xmath86 , and then vanishes at the second fold bifurcation @xmath87 .
different from the case @xmath88 , now the fold bifurcation @xmath79 gives a saddle ( the middle branch ) and an unstable focus ( the upper branch ) ( fig .
[ fig:5]b inset ) .
the unstable focus on the upper branch becomes stable and an unstable limit cycle appears via the subcritical hop bifurcation @xmath89 .
the unstable limit cycle develops until it meet the saddle at the homoclinic bifurcation point @xmath90 .
the stable focus and stable limit cycle coexist with @xmath41 between @xmath91 and @xmath87 , which gives the region _
x _ in fig .
[ fig : bif2 ] .
3 . when @xmath40 increases to @xmath92 ( fig . [
fig:5]c ) , the diagram is similar but the hopf bifurcation @xmath80 changes from subcritical to supercritical , so that stable limit cycles exist with @xmath41 taken values over a wide range from @xmath80 to @xmath87 .
4 . when @xmath93
, @xmath77 becomes too close to @xmath83 so that the unstable limit cycle arise from @xmath89 does not collide with a saddle but ends up at the fold bifurcation point of limit cycles @xmath94 ( fig .
[ fig:5]d ) . 5 .
for a larger @xmath40 ( fig .
[ fig:5]e ) , the two fold bifurcation points @xmath77 and @xmath83 coalesce and disappear due to a codimension - two _ cusp _ point @xmath95 in fig .
[ fig : bif2 ] . 6 . with the further increasing of @xmath40 ( fig .
[ fig:5]f ) , another two fold bifurcation points ( @xmath96 and @xmath97 ) arise from the hopf bifurcation @xmath75 due to the _ cusp _ point @xmath98 in fig . [
fig : bif2 ]
. 7 . further increasing @xmath40 to higher level ( @xmath99 ) , the supercritical hopf bifurcation point @xmath75 moves to the left and
collides with @xmath97 to give a fold - hopf bifurcation point ( _ zh _ ) , and then @xmath97 changes to a saddle - node homoclinic bifurcation point _ snic _ , from which the saddle and the node collide and disappear to generate a stable limit cycle .
the above bifurcation analyses present the global dynamics of the whole system , which is further explored below through the underlying potential landscape @xcite . to explore the global dynamics from the potential perspective , we projected the potential function to two independent variables , the p53 concentration @xmath2 $ ] , and the total mdm2 concentration @xmath100=[\mathrm{mdm2}_{\mathrm{cyt}}]+[\mathrm{mdm2}_{\mathrm{nuc}}]$ ]
. the potential landscapes for parameter values taken from the 10 typical regions are shown at fig .
[ fig:6 ] .
the potential landscapes show that when there is a single stable steady state ( _ i _ , _ ii _ , _ v _ , _ vi _ ) , the potential is funnelled towards a global minimum which corresponds to the global stable steady state . in region _
vii _ , there are two stable steady states , and the potential has two local minimum , corresponding to high or low p53 levels , respectively .
the landscape at low p53 state has wide attractive region and shallow slop , while at the high p53 state has small attractive region and deep slope .
these suggest that a cell from randomly select initial state is more likely to response with low p53 state because of the wider attractive region . in other regions with p53/mdm2 oscillations ( _ iii _ , _ iv _ , _ viii _ , _ ix _ , _ x _ )
, the potential landscapes show an irregular and inhomogeneous closed ring valley that corresponds to the deterministic stable limit cycle trajectory .
it is obvious that the landscape is not uniformly distributed along the limit cycle path due to the inhomogeneous speed on the limit cycle @xcite .
we also note that in the landscapes for regions _ iv _ , _ ix _ and _ x _ , in addition to the closed valley , there is a deep funnel towards a local minimum .
these local minimum show the coexistence of the stable steady state ( marked by @xmath101 and the arrows in fig .
[ fig:6 ] ) .
the regions with p53 oscillations are the most interested because the oscillation dynamics are essential for cell fate decision in response to various stresses .
to further analyze the potential landscape when the system display stable p53 oscillations , we examined the case at region _
iii _ by calculating the potential force ( @xmath102 ) and the probability flux ( @xmath103 ) ( see fig . [ fig:7 ] ) . as shown at fig .
[ fig:7 ] , the potential has local minimum valley along the deterministic oscillation trajectory .
the values of potential are not uniformly distributed along the cycle .
there is a global minimal potential at a state around @xmath104)=(1 , 0.02)$ ] , which with small curl flux forces are large potential forces .
the states with larger @xmath2 $ ] or @xmath100 $ ] have higher potential , larger curl flux forces and smaller potential forces .
we note that the potential forces are mostly vertical to the cycle path , and the curl flux forces are along the cycle .
these results suggest that potential forces tend to attract the system states to the oscillation trajectory , while the curl flux forces drive the oscillation along the cyclical trajectory .
the obvious low potential value at the global minimum indicates a pulse dynamics in each cycle the system states mostly stay at the state of global minimum while go through the other part of the limit cycle quickly .
this pulse dynamics is in agree with previous studies in @xcite .
( blue arrows ) , stationary flux @xmath103 ( white arrows ) and the deterministic oscillation trajectory(black line ) .
here @xmath105 , and the noise strength @xmath106 .
, width=302 ] pdcd5 interacts with the p53 pathway by promoting p53 stability through the disruption of the reaction between p53 and @xmath0 . in our model , the p53 degradation rate function is given by a hill - type function and the half - maximal effective concentration ( ec50 ) is dependent on pdcd5 concentration given by eq .
( [ eq : k2 ] ) . in this function ,
the efficiency of pdcd5 is represented by the coefficient @xmath107 , larger @xmath107 indicates higher efficiency . to investigate the effects of pdcd5 regulation efficiency
, we took @xmath108 and changed @xmath107 from zero to triple the default value to examine its effect on system dynamics . .
here @xmath109 marks the supercritical hopf bifurcation point , red solid lines represent stable steady states , black dashed lines are unstable steady states , green solid dots are the maxima and minima of stable limit cycles .
( b ) period of stable limit cycles .
( c ) the barrier height of stable limit cycles ( bh ) .
( d ) the energy dissipation ( @xmath110 ) . here
@xmath105 , and the noise strength @xmath106 . ,
width=302 ] fig .
[ fig:8]a shows the bifurcation diagram of p53 with respect to the efficiency @xmath107 . there is a supercritical hopf bifurcation ( @xmath111 ) as @xmath107 increases , where a stable limit cycle appears to yield p53 oscillations .
the oscillation amplitude is insensitive with @xmath112 .
the period increase rapidly near the hopf bifurcation transition , and then decreases to @xmath113 when @xmath112 increases away the bifurcation point ( fig .
[ fig:8]b ) . to further study the stability of p53 oscillation , we calculated the barrier height in the potential landscape , which is a good measure of the stability of limit cycles @xcite .
the barrier height ( bh ) is defined by the potential function as @xmath114 , where @xmath115 is the maximum potential inside the stable limit cycle , and @xmath116 is the minimum potential along the cycle @xcite . higher barrier height means more stable .
[ fig:8]c shows the barrier height versus @xmath107 . in our simulation ,
the barrier height is highly fluctuated near the hopf bifurcation ( @xmath117 ) , and then decreases towards a value of about @xmath118 .
next , we considered the effect of @xmath107 to the dissipation of energy , which measures the inevitability for an open nonequilibrium system due to exchanging information and energies with its surrounding
. the mean rate of energy dissipation ( @xmath119 ) is give by @xmath120 , where @xmath37 is the driving force , and @xmath27 the probability flux vector . according to @xcite ,
the energy dissipation is equal to the entropy production rate ( @xmath121 ) at equilibrium state .
therefore , the energy dissipation can be used to obtain a global physical characterization of the nonequilibrium systems .
[ fig:8]d shows the dissipation versus @xmath107 .
the dissipation equals zero when @xmath107 is smaller than the hopf bifurcation value and there is a single steady state .
this shows the detail balance at equilibrium state , in consistent with the fact that the least dissipation cost gives a more stable system in general @xcite .
when @xmath107 increases beyond the hopf bifurcation ( @xmath122 ) , the energy dissipation increases rapidly to maximum at @xmath123 , and then decreases to and sustain at @xmath124 with @xmath107 increasing . in summary
, these results show clearly transition dynamics when the system switches from a single stable steady state to a stable limit cycle with the increasing of pdcd5 efficiency . during the transition
, the steady state is destabilized to yield p53 oscillations , both oscillation periods and energy dissipations first increase rapidly and then decrease to a sustain level , and the barrier height shows large fluctuation in the transition region .
after the pdcd5 efficiency increases beyond the transition region , the system dynamics is insensitive to the efficiency .
the dynamics of tumor suppressor p53 plays important roles in the regulation of cell fate decision in response to various stresses@xcite .
pdcd5 is known to interact with the p53 pathway and functions as a co - activator in p53 regulations . in p53 pathways ,
the p53-mdm2 oscillator is crucial for cell response to dna damage . in this study , we have systematically studied effects of pdcd5 to the p53-mdm2 oscillator by methods of bifurcation analyses and potential landscapes .
our results reveal that the p53-mdm2 oscillator can display monostability and bistability under low pdcd5 expression . when pdcd5 level is upregulated ( for example , by dna damage ) ,
p53 oscillations emerge by hopf bifurcation , and bistability with the coexistence of both stable oscillation and a stable steady state is possible for proper pdcd5 level and p53 synthesis rate .
these results were further verified by the potential landscapes , which clearly show the transition of landscapes with changing parameter values .
we have also investigate the effects of pdcd5 efficiency in the interaction with p53 pathway , we showed that p53 oscillations can only be induced only when the efficiency is larger than a critical values of hopf bifurcation , and the system dynamics show clear transition features in both barrier height and energy dissipation when the efficiency is close to the bifurcation point .
such abnormal behaviours at the transition region have been highlighted in recent years in the application of prediction and early diagnosis of complex diseases @xcite . in this study
we have focused on the pdcd5-regulated p53 dynamics in the p53-mdm2 oscillator .
our results reveal global p53 dynamics mediated by pdcd5 and the levels of p53 production . however , a more complete p53 pathway is certainly necessary in understanding cell fate decisions in response to dna damage .
further consideration of the effects of pdcd5 on a complete p53 network is certainly required in future studies .
this work is supported by the national natural science foundation of china ( nos . 91430101 , 11272169 , 11372017 ) and the scientific research project of inner mongolia colleges and universities ( no .
njzy14130 ) . 47ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ , vol .
( , ) @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | dynamics of p53 is known to play important roles in the regulation of cell fate decisions in response to various stresses , and pdcd5 functions as a co - activator of p53 to modulate the p53 dynamics . in the present paper , we investigate how p53 dynamics are modulated by pdcd5 during the dna damage response using methods of bifurcation analysis and potential landscape .
our results reveal that p53 activities can display rich dynamics under different pdcd5 levels , including monostability , bistability with two stable steady states , oscillations , and co - existence of a stable steady state and an oscillatory state .
physical properties of the p53 oscillations are further shown by the potential landscape , in which the potential force attracts the system state to the limit cycle attractor , and the curl flux force drives the coherent oscillation along the cyclic .
we also investigate the effect of pdcd5 efficiency on inducing the p53 oscillations .
we show that hopf bifurcation is induced by increasing the pdcd5 efficiency , and the system dynamics show clear transition features in both barrier height and energy dissipation when the efficiency is close to the bifurcation point .
this study provides a global picture of how pdcd5 regulates p53 dynamics via the interaction with the p53-mdm2 oscillator and can be helpful in understanding the complicate p53 dynamics in a more complete p53 pathway . |
in his seminal work , luneburg @xcite derived a spherical optical lens with radially varying refractive index that focused a beam of parallel rays to a point at the opposite side of the lens ; a two dimensional variant is straightforward to deduce .
of course this relies upon the governing equation being the helmholtz equation , which the full ela in the model configuration presented here the elastic energy is primarily carried by rayleigh surface waves ; they are a particular solution of navier s equation for elastodynamics for a half - space bounded by a traction - free surface , e.g. the earth s surface .
well known in seismology , for the idealised situation of isotropic and homogeneous media they are non - dispersive , elliptically polarized and in practical terms @xcite they have a velocity very close to that of shear waves : @xmath6 where @xmath7 is the shear modulus and @xmath8 the density @xcite so for simplicity we will simply use the shear wave speed in our analysis .
shear horizontally polarized waves ( sh ) are also present in our numerical model , and they also propagate with wavespeed @xmath9 ; notably sh waves are governed by a helmholtz equation without any approximation .
we do not consider love waves here , which can also be important is seismology , as they only exist for stratified layered media and we assume that our elastic half space is vertically homogeneous , that is , the material parameters do not vary with depth .
in cartesian coordinates we take @xmath10 to be the depth coordinate and @xmath11 to be in the plane of the surface , then the rayleigh waves can be represented using a helmholtz equation on the surface and we consider a circular lens on the @xmath12 plane as in fig .
1c , is characterized by a radially varying refraction profile @xcite .
this lens , and the associated material variation , then extends downwards and the material is considered vertical homogeneous ; we distinguish the material outside the lens to have parameters with a subscript @xmath13 and that inside to have subscript @xmath14 .
the refraction index @xmath15 between two media , say , material 0 and material 1 can be formulated in terms of the ratio of velocity contrast @xmath16 . for a luneburg lens
we require the refractive index , @xmath17 , to be : @xmath18 where @xmath19 is the radial coordinate and @xmath20 the outer radius of the lens ( fig .
we tune the material velocity within the lens to reproduce the index given in [ eq : ref_lune ] so @xmath21 taking a continual material variation is perfect for theory , but from a practical perspective it is not possible to realize a circular structure 10 s of meters in depth and radius , whose soil properties change smoothly ( e.g. on the scale of fig .
instead we create a composite soil made of bimaterial cells such that their effective material properties have the variation we desire , this provides a realistic lens using actual soil parameters that could be created using conventional geotechnical techniques @xcite . in fig .
1c the circular surface of the lens is discretized using equally spaced cells on a periodic square lattice .
each cell contains an inclusion of softer material that , in our illustration , is represented by a pillar extending down into the soil ; the exponential decay of the rayleigh wave amplitude with depth means that for the computational model we can truncate this and a depth of 30 m is more than sufficient .
the diameter of each pillar is determined using the effective velocity prescribed for each cell based upon its radial position ( @xmath22 ) from the center of the lens .
assuming a square section cell of width @xmath23 on the @xmath12 plane the filling fraction is defined using the surface area occupied by the pillar in the cell . for cylindrical pillars with diameter @xmath24 ( fig .
1c ) we have a geometrical filling fraction , @xmath25 , with @xmath26 .
the maxwell - garnet formula @xcite , derived for composites , relates the filling fraction with the corresponding effective property : @xmath27 where @xmath28 is the effective shear velocity in the cell and @xmath29 is the shear velocity of the inclusion ( the pillar ) .
we combine the geometrical definition of @xmath25 with ( [ eq : garnett ] ) to obtain the effective velocity as a function of inclusion size .
hence , by tuning the pillar diameter we obtain the required velocity variation desired in eq .
[ eq : vel_profi ] and use this to define the structure and variation for each of the luneburg lenses one of which is shown in ( fig .
we now place four luneburg lenses as shown in fig .
1b and use these to protect an object placed in between them .
the idea is simply that a plane wave incident along either the @xmath4 or @xmath5 axes will be focussed by the lens to a single point , the point at which the cylinder touches its neighbour , which will then act as source into the next luneburg lens and the plane wave will then reemerge unscathed ; the building to be protected should , in this perfect scheme , be untouched .
we are aiming to demonstrate the concept not in a perfect scenario , but using realistic parameters and a setting in which the effective medium approach provide a discrete velocity profile , yet the protection achieved is considerable . to construct the luneburg lenses , to reach the minimum @xmath9 prescribed in eq .
[ fig:1 ] , @xmath29 needs be lower than 350 m / s . we choose a @xmath29 of 200 m / s which is a value that is realistic for poorly consolidated soil ( sand or water filled sediments ) @xcite . in the lens configuration depicted in figs .
1b and c for each lens there are 26 elementary cells ( @xmath30 m ) along the radial axis of the lens and the diameter of the pillars increases towards the center of the lens as discussed earlier . in the frequency range we investigate ( 3 - 8 hz ) , the inclusion is deeply subwavelength and non - resonant .
the only parameter of interest for the lens design using composite soil is the filling fraction , so there is no bound on the size of the elementary cell so long as it remains subwavelength to avoid bragg scattering phenomena ; in our simulations the minimum size of each cell is bounded for numerical reasons ( explained in the next section ) . for an actual implementation of the lens
the cell could be chosen to be even smaller than our choice here with corresponding decrease in the pillar diameters .
a maximum diameter of approximately 2 m would permit the pillars to be realised with existing geotechnical machineries @xcite .
three dimensional numerical simulations of seismic elastic surface waves are implemented using specfem3d : a parallelized , time domain , and spectral element solver for 3d elastodynamic problems widely used in the seismology community @xcite .
the reference computational domain , depicted in fig .
1a , is a 40 m depth halfspace 350 x 500 m wide of homogeneous sedimentary material with background velocity @xmath9 set to 400 m / s .
the computational elastic region , apart from the surface itself , is surrounded by perfectly matched layers ( pmls ) @xcite that mimic infinity and prevent unwanted reflections from the computational boundaries and these are standard in elastic wave simulation @xcite .
a small building with a flexural fundamental mode of approximately 4 hz is located at the center of the model .
we place a line of broadband ricker source time functions @xcite centered at 5 hz to generate an almost perfectly plane wavefront ( figs .
1a and b ) .
the driving force * f * has equal components in all 3 orthogonal directions and hence the resulting excitation is not only made of rayleigh but also of sh waves .
body waves leave the computational domain and pass into the pml at the bottom and side boundaries of the computational domain ; their interaction with the structure and the lens is negligible .
this configuration could represent the upper layer of a deep sedimentary basin with some strategic structure located at the surface ( power plants , data centers , hospitals ) desired to be shielded from a seismic energy source . in fig .
1b representing the building surrounded by the square array of lenses , the pillars are inserted as softer inclusions in the background velocity model .
this approach is commonly used ( e.g. * ? ? ?
* ; * ? ? ?
* ) to simplify the discretization of complex models .
we use very fine meshing ( down to 2 m ) and this , combined with @xmath31 order accuracy in space of the spectral element method , allows us to accurately model pillars down to a smallest diameter of 0.3 m. this approach was validated against a model where the pillars were meshed explicitly ; the only notable difference was a factor of 5 increase in the runtime for the explicit case vis - a - vis the regular mesh .
specfem3d is a parallel code and simulations are run on 64 cores for a total of approx .
30 corehours for a simulated time of 1.5 seconds with the 3d wavefield saved for post - processing ; specfem3d is the standard geophysics code used in academic and industry applications with a long history of development and application @xcite .
this simulation is shown for wave - field snapshots for different propagation distances in fig .
2 both with , and without the lenses . the sources generating the plane wavefront in fig .
1 are located at the surface and so most of the seismic energy propagates as rayleigh and sh waves .
the vertical component of the displacement @xmath32 shown in fig .
2 , is dominated by the elliptically polarized motion of the rayleigh waves .
although not visible , sh waves behave very similarly to rayleigh waves for the model here discussed , body waves have far lower amplitude and are not relevant in our analysis . fig .
2a shows , as one would expect , the strong scattering by the building and its highly oscillatory motion .
when the luneburg lenses are inserted in the model ( fig .
2b ) the simulation shows that the rayleigh wave front splits and then progressively converges to the focal points of lenses l1 and l2 .
given the square layout , the focal points lie exactly at the touching points of the l1-l3 and l2-l4 lenses .
this lens does not support any subwavelength phenomena ( the evanescent part is lost ) hence , the size of the focal spot is diffraction limited at @xmath33 and some energy is backscattered during the focusing process creating some low amplitude reflections .
the second lenses ( l3 and l4 ) behave in a reciprocal manner converting the two point ( secondary ) source - like wavefields back into a plane wavefront . during the entire process , the inner region where
the building is placed has experienced a very low seismic excitation as compared to the reference unprotected case .
3 presents the motion of the roof of the reference building on a db scale and it shows the vibration is drastically reduced . the snapshots in the bottom row of fig .
2 showing the wavefront as it emerges from the lenses shows that despite the strong alteration of the ray - path , the reconstruction of the wavefront after the lenses is surprisingly good .
hence this device combines the some cloaking behaviour with the seismic protection . considering the broad frequency bandwidth of the input signal
this is an interesting result as most cloaks so far proposed have problems with broadband excitation .
the velocity structure of the lenses is such that the propagation of the wavefront in fig .
2b is slightly slower than the reference configuration of fig
thus , we observe cloaking functionality to be valid for the wave envelope but not for the phase .
this is not particularly relevant in the present seismic context where the only application of cloaking is to avoid very directive scattering , it would be unfortunate and undesirable to scatter or refocus the signal to a neighbouring building , while simultaneously realising seismic protection .
a quantitative analysis of the wavefield is presented in figs .
3a and 3b which show the energy maps for the reference and protected cases .
the energy is calculated at the surface ( @xmath34 ) taking the @xmath35 norm of the three components of the displacement field @xmath36 . in the homogeneous case ( fig .
3a ) of the unprotected building the energy is almost uniform across the whole computational domain making the building resonate . in the protected case , the energy is focused towards the axes of symmetry of the lenses , leaving a central region relatively undisturbed ; in the two stripes shown in fig . 3b
the energy ( and equally the amplitude ) is much higher than elsewhere .
3c shows the frequency response function of the building with , and without , the lenses . the rooftop horizontal displacement ( roof drift )
@xcite is a diagnostic of the amplification phenomena due to the resonance frequency of the structure . over the whole spectrum an average amplitude reduction of 6 db
is achieved which is reduction of almost an order of magnitude in the vibration .
complete cancellation of the wavefield is not achieved primarily because the evanescent field slightly couples with the building and as we focus the wavefield to a point source we introduce some back scattering that also interacts with the building ( fig .
nonetheless the concept is demonstrated and shows that one can successfully translate ideas from optics into the field of elastic seismic waves ; figs . 2 and 3
should inspire the use of these concepts to both reroute surface waves and to reduce their impact on surface structures .
we have combined concepts from transformation optics and plasmonics , composite media , and elastic wave theory to create an arrangement of seismic luneburg lenses that can reroute and reconstruct plane seismic surface waves around a region that must remain protected .
the lens is made with a composite soil obtained with columns of _ softer _ soil material with varying diameter distributed on a regular lattice .
the use of softer soil inclusions emphasises that the methodology we propose is not reflection of waves , or absorption , or damping for which rigid or viscoelastic columns might be more intuitive ; the softer inclusions are designed to progressively alter the material itself so that waves are `` steered '' and the reconstruction of the wavefronts after exiting the arrangement illustrates the mechanism .
the luneberg lens arrangement proposed here could be tested in a small scale experiment using elastic materials , such as metals , as the concept itself is very versatile or on larger scale test areas where one could then evaluate effects such as nonlinearity .
although presented in the context of seismic engineering there are everyday ground vibration topics that could benefit from this design .
the damping of anthropic vibration sources ( e.g. train - lines , subway , heavy industry ) is very important for high precision manufacturing process , to reduce structural damages due to fatigue @xcite or simply to decrease domestic or commercial building vibrations .
our aim here has been to present the concept of steering elastic surface waves completely in context and with a design that could be built , this should motivate experiments , further design and the implementation of ideas now widely appreciated in electromagnetism and acoustics to this field .
one important practical point regarding our design is that we have presented normal incidence to the four luneberg lens and this is practical , of course , if the position of the vibration ( a railway line for instance ) is known .
the protection progressively deteriorates as the incidence angle of the plane wave increases ; at @xmath37 , the focal point is in the centre of the region ( fig .
4 ) and the energy is steered to this point .
the concept of using soil mediation to steer surface elastic waves is clear and the luneburg arrangement is a clear exemplar of the ideas .
the proposed design is easily adapted to higher frequency bands and smaller regions .
if one is only interested in seismic protection , and less in the wavefront reconstruction , the lenses can be structured differently with only one or two lenses .
the other well - known grin lenses offer different extents and types of wave control ( e.g. maxwell , eaton @xcite ) can provide an isotropic wave shielding .
other types of grin lenses and layout ( for instance using 4 half - maxwell lenses as shown in optics @xcite ) can be utilised ; however the luneburg lens requires the lowest velocity contrast between the lens and exterior region and we choose to use it as practically it could be implemented .
the main practical difficulty is that these lenses are either singular ( an eaton lens has @xmath38 m / s in the center ) or prescribe stronger velocity contrast ( maxwell ) requiring difficult ( or not yet available ) soil engineering solutions .
we are particularly grateful to the specfem3d development team led by dimitri komatitsch , who helped us with the implementation of the pml conditions essential in these simulations .
the authors also thank the epsrc , erc and cnrs for their financial support .
ac was supported by a marie curie fellowship .
84ifxundefined [ 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 * * , ( ) link:\doibase 10.1126/science.1125907 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) link:\doibase 10.1103/physrevlett.103.024301 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.108.014301 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop _ _ , ultrasonics technologies ( , ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop _ _ ( , ) @noop _ _ ( , ) link:\doibase 10.1103/physrevlett.85.3966 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) \doibase http://dx.doi.org/10.1063/1.3068491 [ * * , ( ) , http://dx.doi.org/10.1063/1.3068491 ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) http://stacks.iop.org/2040-8986/14/i=7/a=075705 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) \doibase http://dx.doi.org/10.1063/1.4923040 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.4893153 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.4818716 [ * * , ( ) ] @noop * * , ( ) @noop _ _ ( , ) p. @noop * * , ( ) @noop * * , ( ) https://books.google.co.uk/books?id=moj_v-8u0tec[__ ] ( , ) @noop _ _ ( , ) in @noop _ _ ( , ) pp .
@noop _ _ ( , ) @noop _ _ ( , ) link:\doibase 10.1785/0120020140 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1111/j.1365 - 246x.2011.05044.x [ * * , ( ) ] in link:\doibase 10.1109/sc.2003.10023 [ _ _ ] ( ) pp . in @noop
_ _ , ( , ) pp .
link:\doibase 10.1785/0120130106 [ * * , ( ) ] link:\doibase 10.1046/j.1365 - 246x.1999.00967.x [ * * , ( ) ] @noop ( ) @noop * * , ( ) link:\doibase 10.1190/1.2757586 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/j.jseaes.2014.02.009 [ * * , ( ) ] @noop _ _ ( , , ) https://books.google.fr/books?id=ogsvy9tbmbmc[__ ] , proceedings and monographs in engineering , water and earth sciences ( , ) @noop * * , ( ) @noop * * ( ) | metamaterials are artificially structured media that can focus ( lensing ) or reroute ( cloaking ) waves , and typically this is developed for electromagnetic waves at millimetric down to nanometric scales or for acoustics or thin elastic plates at centimeter scales . extending the concepts of @xcite we show that the underlying ideas are generic across wave systems and scales by generalizing these concepts to seismic waves at frequencies , and lengthscales of the order of hundreds of meters , relevant to civil engineering . by applying ideas from transformation optics we can manipulate and steer rayleigh surface wave solutions of the vector navier equations of elastodynamics ; this is unexpected as this vector system is , unlike maxwell s electromagnetic equations , not form invariant under transformations . as a paradigm of the conformal geophysics that we are creating , we design a square arrangement of luneburg lenses to reroute and then refocus rayleigh waves around a building with the dual aim of protection and minimizing the effect on the wavefront ( cloaking ) after exiting the lenses . to show that this is practically realisable we deliberately choose to use material parameters readily available and this metalens
consists of a composite soil structured with buried pillars made of softer material .
the regular lattice of inclusions is homogenized to give an effective material with a radially varying velocity profile that can be directly interpreted as a lens refractive index .
we develop the theory and then use full 3d time domain numerical simulations to conclusively demonstrate the validity of the transformation seismology ideas : we demonstrate , at frequencies of seismological relevance @xmath0 hz , and for low speed sedimentary soil ( @xmath1 m / s ) , that the vibration of a structure is reduced by up to 6 db at its resonance frequency .
this invites experimental study and opens the way to translating much of the current metamaterial literature into that of seismic surface waves .
mathematicians and physicists have long studied the physics of waves at structured interfaces : back in 1898 , lamb wrote a seminal paper on reflection and transmission through metallic gratings @xcite that then inspired numerous studies on the control of surface electromagnetic waves .
the concept that periodic surface variations could create and guide surface waves has emerged in a variety of guises : rayleigh - bloch waves for diffraction gratings @xcite , yagi - uda antenna theory @xcite and even edge waves localised to coastlines @xcite for analogous water wave systems .
most notably in the last decade , the discovery of spoof plasmon polaritons ( spps ) @xcite has motivated research not only in plasmonics @xcite but also in the neighbouring , younger , field of platonics @xcite , devoted to the control of flexural ( lamb ) waves in structured thin elastic plates .
the extension of ideas such as cloaking @xcite to plasmonics @xcite is highly motivational as it suggests that one can take the concepts of wave physics from bulk media , as reviewed in @xcite , and apply them to surfaces . in platonics , implementations of ideas from transformation optics
@xcite show how , all be it for the limiting cases of thin elastic plates , similar concepts can take place in another arena of physics , in this case mechanics and elasticity . unfortunately in much of elastic wave theory , and for many applications ,
one actually has the opposite physical situation to that of a thin plate , that is , one has an elastic material with infinite depth ; on the surface of such a half - space elastic surface waves , rayleigh waves , exist @xcite that exponentially decay with depth and have much in common conceptually with surface plasmons in electromagnetism .
it is therefore attractive to investigate whether concepts proven in plasmonics can be translated across despite the underlying governing equations having many fundamental differences .
some experimental work has taken place in broadly related scenarios such as the attenuation of rayleigh waves at high frequencies in a marble quarry with cylindrical holes @xcite and in piezo - electric substrates with pillars @xcite , but with differing aims and not at frequencies relevant for seismic waves .
some work on structured elastic media , going beyond elastic plates , to try and create seismic metamaterials @xcite is underway with some success either with subwavelength resonators @xcite or with periodic structuring within the soil @xcite and trying to still utilise flexural wave modelling .
our aim here is complementary in that we want to implement ideas from transformation optics into this elastic surface wave area and investigate what can be achieved . the desire , and need , to control the flow of waves is common across wave physics in electromagnetic , acoustic and elastic wave systems @xcite .
acoustics is , mathematically , a simplified version of full elasticity with only compressional waves present : full elasticity has both shear and compression , with different wave speeds , and full coupling between - leading to a formidable vector system .
the quest for a perfect flat lens @xcite with unlimited resolution and an invisibility cloak @xcite that could conceal an object from incident light via transformation optics @xcite are exemplars for the level of control that can , at least , in theory be achieved for light @xcite or sound @xcite .
an invisibility cloak for mechanical waves is also envisioned @xcite for bulk waves , and experiments using latest developments in nanotechnology support these theoretical concepts @xcite .
however , some form of negative refractive index appears to be necessary to achieve an almost perfect cloak .
unfortunately , a negative index material emerges from a very complex microstructure that is not feasible for the physical length - scales , for seismic waves at frequency lower than 10 hz this means wavelengths of the order of a hundred meters , in geophysical applications .
notably one can design locally resonant metamaterials that feature deeply subwavelength bandgaps @xcite and so could be used in seismic and acoustic contexts , but these effects are observed over a limited frequency band ; this is being improved to allow for simultaneous protection and cloaking @xcite and has potential . however
, here we look at protection and cloaking , applied to low frequency seismic waves , using the concept of lensing . elastic waves , in the same way as light , are subject to snell s law of refraction , and an appropriate choice of material properties leads to spectacular effects like those created by gradient index ( grin ) lenses .
compared to classic lens where ray - paths are bent through discontinuous interfaces causing losses and aberrations , grin lenses are obtained with a smooth refractive index transition .
rayleigh and maxwell themselves studied grin lenses , notably the maxwell s fisheye whose spatially varying refractive index was later shown to be associated with a stereographic projection of a sphere on a plane @xcite . as noted in @xcite grin lenses
have been mainly studied and implemented for optical applications , or wave systems governed by the helmholtz equation .
the ideas behind transformation optics are not limited to metamaterial - like applications and have contributed to recent advances in plasmonic light confinement by touching metallic nanostructures @xcite .
a classical example of a grin lens is the circular luneburg lens @xcite , once a plane wave enters the lens , its radially varying refraction index steers the ray - path towards a focal point located on the opposite side of the lens @xcite .
the eponymous fisheye lens of maxwell @xcite is another well documented and fascinating grin lens @xcite and it has been proposed as a non - euclidean cloak @xcite . to date , applications have been mainly limited to scalar wave systems governed by helmholtz s type operators for transversely polarized electromagnetic waves @xcite , for pressure waves @xcite or platonics where composite and thickness modulated plates have recently been proposed @xcite . in full elasticity , where the propagation is described by the vector navier equation that is not simply a helmholtz equation or scalar , a proof of concept is still missing . experimentally , for seismic waves , the realization of any proposed lensing arrangement becomes a real challenge because wavelengths range from @xmath2 m to @xmath3 m. for real applications wave control must be achieved over a broad frequency band , to cover the various wavelengths of interest ; unlike other metamaterial cloak designs such as those based around subwavelength resonators @xcite , the lens proposed here is effective across a very broad spectrum of frequencies . in civil engineering
the structuring or reinforcement of soils is commonplace with geotechnical solutions aimed at improving soil seismic performances @xcite
( e.g. soil improvement by dynamic compaction , deep mixing and jet grouting ) , typically implemented prior to construction of structures , aimed to rigidify or decouple the building response and not at rerouting the seismic input . in the conceptual lens we design , the structuring of the soil is feasible and we use material parameters typical of poorly compacted sediments . using the ideas of transformation optics , and detailed 3d numerical simulations , we show that a square arrangement of four luneburg lenses can completely reroute waves around an area for seismic waves coming with perpendicular incident directions ( e.g. @xmath4 and @xmath5 in fig .
1b ) . not only are the waves rerouted leaving the inner region protected , but the wave - front is reconstructed coherently after leaving the arrangement of lenses . |
this review is about mesoscopic conductors , statistics of related physical quantities and their scaling behavior under change of the size of the conductor .
a few words about the terms used here shall tell the reader what to expect .
mesoscopic systems are intermediate between microscopic and macroscopic systems .
the term goes back to van kampen and was widely used after a work on `` physics of mesoscopic systems '' by imry @xcite .
mesoscopic conductors contain many ( e.g. @xmath1 ) elementary objects like electrons and atoms . on the other hand their properties
can not be obtained by using the _ thermodynamic limit _
( volume and particle number going to infinity with fixed particle density ) .
the thermodynamic limit is a convenient mathematical device in solid state theory if one is interested in material constants .
in particular , in macroscopic conductors the conductivity , @xmath2 , is such a material constant and the conductance , @xmath3 , is then given by ohm s law , @xmath4 , where @xmath5 is the cross section of the conductor and @xmath6 its length .
this ohmic behavior is to be expected once the system size ( denoted by one length @xmath6 ) is much larger than characteristic length scales ( see fig . 1 ) .
these are ( i ) the kinematic scale set by the de broglie wavelength @xmath7 or , for low temperatures , by the _ fermi wavelength _
@xmath8 , ( ii ) the _ elastic mean free path _
@xmath9 , which is the average distance an electron travels before its initial momentum relaxes and ( iii ) the _ phase coherence length _
@xmath10 , which is the average distance an individual electron travels before its initial and final phases become incoherent .
once the phase coherence length @xmath10 becomes _ larger _ than the system size the conductor can no longer be described by material constants .
microscopic details of the conductor will then influence even global physical quantities like conductance such that measurements yield _ finger prints _ of the mesoscopic conductor .
these finger prints are due to quantum mechanical interference effects . since mesoscopic conductors contain many electrons one has to use a quantum statistical description for their thermodynamic and transport quantities . however , this is not the only aspect of statistics in mesoscopic conductors .
apart from the electronic degrees of freedom the conductor contains other degrees of freedom which are responsible for resistivity to applied voltages . in the simplest case
these are static fields caused by impurities , vacancies and dislocations in an otherwise ideal crystal lattice .
also static inhomogeneous electric fields caused by surrounding media are responsible for resistivity .
one refers to these static fields as _
static disorder potentials_. for different realizations of a static disorder potential the global conductance of an otherwise identical mesoscopic conductor will differ sometimes significantly .
therefore , one considers ensembles of disorder potentials characterized by a few parameters , e.g. by the average mean free path on short scales .
the task of the theory is then to determine the _ probability distribution _ of physical quantities like conductance for such ensembles .
thus , on top of the quantum statistical treatment for an individual disorder potential the statistical properties of a whole ensemble of disorder potentials is addressed in mesoscopic physics .
typically , the phase coherence length is of the order of a few microns for metals and semi - conductors below liquid - helium temperatures .
this is the reason that technological advances , starting in the 80 s , were needed in order to study mesoscopic conductors systematically .
we will discuss some of the fundamental experiments in section [ mes ] . although it is true that technological advances were needed to bring mesoscopic physics to the attention of a broader community in solid state physics , a great deal of the physics had been discussed much earlier .
for example , anderson had introduced the idea of _ localization _ already in 1958 @xcite . by localization
it is meant that strong disorder can trap electrons by quantum interference to a finite region such that the conductor actually behaves as an insulator .
also , landauer had presented a formula in 1957 @xcite that describes the conductance of a phase coherent conductor in terms of pure quantum mechanical transmission amplitudes , and a similar important approach to the conductance of mesoscopic systems goes back to works in 1972 by edwards and thouless @xcite .
the works by edwards and thouless @xcite and by wegner @xcite form the starting point of the _ scaling theory _ for mesoscopic conductors . later in 1979
it was formulated in a simple and predictive way by abrahams , anderson , licciardello and ramakrishnan @xcite . the scaling theory for disordered mesoscopic conductors
is based on the assumption that the transport properties of systems probed at scales much larger than the mean free path should become insensitive to the microscopic origin of the disorder .
the scaling theory states that once we know the conductance for a given system size we can obtain the conductance for an even larger system from a universal flow equation .
now , since we have already pointed out that the conductance is a random variable which depends in a complicated manner on the particular disorder realization this assumption needs further explanation . as has been stressed by anderson _ et al .
_ @xcite and later by shapiro @xcite it can be correct only in a probabilistic sense .
the whole probability distribution of the conductance approaches a universal distribution function which depends only on very few characteristic system parameters .
the scaling theory leads to a classification of mesoscopic electron systems into three classes : depending on the initial values of the characteristic system parameters ( called scaling variables ) for a system size @xmath11 the system will flow under increase of system size to an insulating state ( localized phase ) , or to an conducting state ( delocalized phase ) , or it stays scale independent at a _ critical state _ .
thus a disorder induced transition from a localized to a delocalized phase can occur which resembles a critical phenomenon .
this transition is frequently denoted as anderson transition or metal - insulator transition ( mit ) . to distinguish this disorder induced mit from other mits occurring in solid state physics
we adopt the notion of _ localization - delocalization _ ( ld ) transition .
accordingly , the critical point is referred to as the ld transition point .
the central aim of this article is to provide a unique picture about the ld transition in mesoscopic disordered electron systems . to achieve this aim
we will introduce the basic physical quantities , scales and theoretical models and rely on simple calculations avoiding involved technical details .
we stress general concepts and methods such as quantum composition rules for conductors , fokker planck equations , and scaling laws for distribution functions .
important results that are based on technically more elaborate methods will only be quoted without detailed description .
nevertheless , we hope that the present article will help to find a way through the original literature .
a number of reviews and books exist on related subjects discussed in this article .
the following short list shall help for further reading .
the review by y. imry @xcite is an introduction to mesoscopic physics .
a review on the subject of mesoscopic transport that contains also information about the technological aspect of devices was written by beenakker and van houten @xcite .
the book by datta @xcite on `` electronic transport in mesoscopic physics '' covers many subjects that have proved to be relevant in the field . as review on localization and scaling theory for mesoscopic systems
we recommend the ones by lee and ramakrishnan @xcite and by kramer and mackinnon @xcite .
the book by altshuler and lee @xcite on `` mesoscopic phenomena in solids '' contains review articles by inventors in the field .
the present article is organized as follows . in sect .
[ mes ] fundamental experiments will be discussed .
the electronic aharonov - bohm - effect makes the definition of mesoscopic conductors more precise .
universal conductance fluctuations demonstrate the _ finger - print _ character of mesoscopic conductance measurements and experiments on weak and strong localization point to the necessity of a critical phenomenon description of the ld transitions .
basic physical quantities are the linear response quantities such as conductance . these and the corresponding relevant energy and length scales are the subject of sect .
[ bas ] . to discuss the statistical problem inherent in the problem of mesoscopic conductors we introduce hamiltonian models in sect .
there we will see that the green s function contains the essential information .
only the asymptotic behavior of the green s function is needed for describing global conductances resulting in a scattering theoretical formulation of mesoscopic conductors ( sect .
[ sca ] ) .
this point of view has been particular useful in describing strong localization in so - called quasi - one - dimensional systems .
however , the scattering theoretical modeling is not restricted to this case but can be extended to more general geometries by considering networks of scattering matrices . after the basic quantities and tools have been presented we discuss the statistical properties of mesoscopic conductors in the idealized localized and delocalized phase , respectively ( sect .
[ staide ] ) . to prepare for the statistics in more realistic systems we introduce the concept of the scaling theory in sect .
the flow of the relevant scaling variables is controlled by so - called @xmath12-functions from which a critical exponent for the ld transitions can be obtained .
it describes how the correlation length of this critical phenomenon diverges on approaching the ld transition point .
based on a pedagogical model , introduced by shapiro ( cf .
@xcite ) , we emphasize that typical values of the conductance distribution can be good candidates for scaling variables .
we pay special attention to the statistics of local quantities ( like the local density of states ) that show power law scaling in non - localized phases .
we come to the conclusion that , at the ld transition , a spectrum of power law exponents ( called multifractal spectrum ) is needed in order to describe the distribution .
the more realistic models are the subject of sect .
[ starea ] .
the onset of broad distributions characterizing global and local quantities at large , but finite , values of typical conductance leaves open questions about their interpretation . to clarify some of them the conductance distribution in an idealized quasi - one - dimensional geometry is studied .
the idealization is due to the assumption of absence of transversal diffusion in such systems .
the calculations give useful insight into the physics of universal conductance fluctuations and lead to the interpretation that the onset of broad tails in the conductance distribution is intimately connected to the critical regime of the ld transition .
such regime , as well as long tails in the conductance distribution , is absent in the idealized quasi - one - dimensional conductor .
giving up the idealization , one can study the ld transition in quasi - one - dimensional systems by the so - called finite size scaling method . due to the complexity of calculations for non - ideal quasi - one - dimensional systems the finite size scaling method
is often restricted to numerical calculations . however , such calculations are very successful and provide a lot of insight into the physics as well as quantitative results for universal properties .
the investigation of statistics and scaling of mesoscopic disordered electron systems close to the ld transition in sect .
[ statra ] reveals the multifractal properties of electronic states and opens the possibility to consider the local density of states as an appropriate order parameter field for the ld transition .
the typical value , as given by the geometric mean , is the global order parameter .
similarly , we collect arguments that allow to view the typical conductance as the _ only _ relevant scaling variable .
our conclusions are summarized in sect .
[ sum ] and a brief account of linear response theory is given in the appendix sect .
in this section we will discuss some fundamental mesoscopic experiments without paying much attention to historical order and completeness .
rather , we wish to give motivation to the following sections which are based on a theoretical physicists point of view . the electronic aharonov - bohm - effect can be used for a definition of mesoscopic conductors .
it is our starting point of investigations .
it points to the necessity for a quantum mechanical description of transport .
the need to describe physical quantities in terms of distribution functions rather than in terms of material constants became evident after the discovery of universal conductance fluctuations .
that one conductor can behave both , as a perfect conductor and as a perfect insulator , when changing system parameters is the insight coming from experiments on weak and strong localization .
weak localization refers to a situation where the system is metallic , but the conductivity is reduced as compared to its classical value .
this behavior is interpreted as a precursor of the strong localization phenomenon where the system is insulating due to spatial localization of electronic states .
weak localization effects have been found in various materials .
the experimental verification of strong localization is most clear - cut in the quantum hall effect , and first experimental investigations of fluctuations at the ld transition point have been performed . a ring - shaped conductor shown schematically in fig .
2 with wire thickness of @xmath13 nm and a diameter of @xmath14 nm that was etched out of a high quality polycrystalline gold film of thickness @xmath13 nm was used to perform the first realization of the most fundamental experiment in mesoscopic physics , the electronic aharonov - bohm - effect . at temperatures less than @xmath15
k the resistance oscillates as a function of the magnetic flux enclosed by the ring @xcite .
figure 3 shows the result for the resistance oscillations found in a ring - shaped conductor fabricated in high mobility gaas - al gaas heterostructures performed by timp _
_ @xcite . here
the diameter of the ring is @xmath16 m. the period in the magnetic field ( @xmath17 mt ) corresponds to a periodicity in the magnetic flux enclosed by the ring of one flux quantum @xmath18 where @xmath19 is the elementary charge of an electron and @xmath20 is planck s constant .
the qualitative explanation of the oscillations relies on the concept that electrons in a magnetic field @xmath21 , where @xmath22 is the vector potential , pick up a quantum mechanical phase , @xmath23 when they move along a classical path @xmath24 that connects two points in space .
now , for an electron coming from the current source the path splits up into the upper and lower branch of the ring and the phase difference @xmath25 between both paths is given by the total magnetic flux @xmath26 enclosed by the ring @xmath27 in terms of the the flux quantum @xmath28 the phase difference is @xmath29 . taking for granted that the resistance is sensitive to quantum interference of electronic waves
the periodicity of the resistance is given by @xmath30 .
the electronic aharonov - bohm - effect demonstrates that the electrons are to be treated quantum mechanically along the ring and that their initial phase is not randomized before leaving the ring .
this phenomenon is called phase coherence and the aharonov - bohm - effect gives rise to an operative definition of the phase coherence length @xmath10 ( see also sect .
[ subrel ] ) . on increasing the temperature or the diameter of the ring
the phase coherence gets lost and the aharanov - bohm effect disappears .
the system size at which the aharanov - bohm effect ( gradually ) disappears quantifies the temperature dependent phase coherence length @xmath10 .
the disappearance of the aharanov - bohm effect is caused by an increasing number of inelastic scattering events which destroy the phase coherence .
the average distance of inelastic scattering events can be identified with @xmath10 .
measurements of conductance in mesoscopic systems that have been performed since the 80 s , showed sample specific , reproducible , statistical fluctuations when system parameters like fermi energy , magnetic field or disorder configurations were changed @xcite . in figure 4
the results of a recent experiment are shown .
gold - nanowires of very high purity with cross - section @xmath31 nm@xmath32 and lengths between @xmath33 and @xmath34 nm were used .
the conductance @xmath3 was measured as a function of applied magnetic field below @xmath35 t in steps of @xmath15 mt at @xmath36 mk .
although the average conductance varies from @xmath37 @xmath38 ( for @xmath39 nm ) to @xmath40 ( for @xmath41 nm ) , the variance of the conductance stayed constant and was @xmath42 .
this phenomenon is referred to as _ universal conductance fluctuations _ ( ucf ) . in a mesoscopic conductor with high conductance ( in atomic units @xmath38 )
the conductance fluctuations are reproducible _ finger - prints _ of the conductor . however , the variance turns out to be approximately a constant @xmath43 two aspects of ucf are interesting : reproducibility and universality . that the fluctuations are reproducible can be understood again by a semiclassical description of conducting electrons .
consider electrons moving along classical paths between elastic collisions , but having a quantum mechanical phase attached .
since the phase information is not lost over distances of the phase coherence length ( mesoscopic regime ) a sample specific interference pattern arises .
the interference pattern will be altered e.g. by applying a magnetic field .
this gives rise to reproducible conductance fluctuations .
the fact that the order of magnitude of these fluctuations turns out to be independent of the average conductance and does only depend weakly on dimensionality and symmetry properties of the conductor is , however , not easy to understand without having a transport theory of mesoscopic conductors .
a striking consequence of the phenomenon is that mesoscopic conductors can not be characterized by an average conductivity being a material constant , even if the system size @xmath44 becomes very large .
conductivity @xmath2 is related to conductance by ohm s law . for a cubic geometry
the conductance behaves as @xmath45 .
for ucf the relative fluctuations of conductance are @xmath46 which is in contrast to the classical behavior @xmath47 .
thus , even in 3d metallic systems the relative fluctuations are much stronger than @xmath48 .
early measurements of conductivity of amorphous silicon @xcite showed that , at low enough temperatures , the temperature behavior of conductivity follows mott s @xmath49-law , @xmath50 over two orders of magnitude with some constants @xmath51 and @xmath52 ( see fig .
this behavior of the conductivity could be understood when assuming that the transport is mediated by phonon assisted hopping processes between localized states @xcite .
localized states are spatially localized to some finite volume within the conductor .
the finite volumes are characterized by a localization radius @xmath53 .
the phonon assisted hopping is possible because localized states can be energetically close to each other if the corresponding localization centers are separated by distances much larger than the localization radius .
the observation of mott s law made it clear that electronic states can indeed be localized by disorder which has been pointed out by anderson @xcite much earlier .
the effect is referred to as _
strong localization _ since the localization radius is smaller than system size .
a further experimental hint to localization came from weakly disordered metallic films at low temperatures .
they showed a logarithmic increase of resistance when the temperature is decreased ( see fig .
although the system behaves metallic with large conductance the conductance was lower than its classical value .
the effect was interpreted as being due to quantum interference corrections to classical transport and is referred to as _
weak localization_. in weak localization the localization length is larger than system size but the effect is interpreted as a precursor of strong localization , i.e. under further increase of system size ( while still being mesoscopic , i.e @xmath44 ) the system would show strong localization .
this is , of course , in practice very hard to realize when localization lengths are larger than microns . the weak localization effect ( predicted by gorkov _ et al .
@xcite ) can be explained by the phenomenon of _ enhanced backscattering _ due to quantum interference .
it relies on the fact that the quantum coherent probability of the sum of two probability amplitudes @xmath54 is two times larger than the incoherent sum of the corresponding probabilities , @xmath55 .
consider the chance for an electron traveling in a phase coherent conductor ( see fig . 1 )
to return back to some point within the conductor . in fig
7 a possible path is drawn for such a path ( straight line ) .
if the motion is time reversal symmetric also the time reversed path ( broken line ) is possible .
the amplitudes of both paths add coherently in a semiclassical calculation of the return probability .
thus , the quantum mechanical return probability is twice the classical value . without explicitly calculating the conductivity it is obvious from this observation that the conductivity must be reduced as compared to its classical value .
the coherence is limited by the phase coherence length @xmath10 which varies as some inverse power of temperature , @xmath56 for system sizes @xmath6 _ larger _ than the phase coherence length @xmath57 the logarithmic corrections to the classical conductivity could be interpreted as being due to the quantity @xmath58 where @xmath9 is the average elastic mean free path .
magnetic fields destroy the time reversal symmetry of the electronic motion and thus the weak localization effect vanishes ( see upper set of fig .
8) a very interesting effect is due to spin - orbit scattering processes that are time reversal symmetric . in this case
the time reversed path , however , gives rise to destructive interference in the return probability . as a result
the sign of the conductance corrections is reversed ( see lower set of fig .
8) and the enhancement of conductivity above its classical value is denoted as _ weak anti - localization _
( first predicted by hikami _
@xcite ) .
weak localization experiments are an important tool e.g. to determine the phase coherence length @xmath10 and the spin - orbit scattering length . for more details
we refer to the review by bergmann @xcite . to reach experimentally the regime of strong localization where the localization radius is less than the system size @xmath44 one has to use strongly disordered systems or work in energy regions with very low density of states .
localization - delocalization ( ld ) transitions have been observed in several experiments ( for reviews see @xcite ) .
figure 9 shows the finding of an experiment by stupp _
@xcite indicating that the conductivity vanishes on approaching a critical charge carrier concentration .
a major problem in comparing theoretical models with the experiments comes from the fact that in the low density of states regimes coulomb interaction between electrons becomes essential , but most theoretical models are based on an independent electron picture . the most clear - cut experiment for the occurrence of ld transitions is the quantum hall effect discovered by von klitzing @xcite .
the quantum hall effect that occurs in two - dimensional electron gases in the presence of strong perpendicular magnetic fields is characterized by a step - function like behavior of the hall conductivity @xmath59 as a function of the so called filling factor ( a dimensionless quantity proportional to carrier concentration and inverse magnetic field ) and by a vanishing dissipative conductivity @xmath2 in the hall plateau regimes ( see fig .
10 ; note , that for non - vanishing hall conductivity a vanishing dissipative conductivity causes also a vanishing longitudinal resistivity . ) .
the peaks in the dissipative conductivity as a function of the filling factor have a clear interpretation in terms of ld transitions .
however , no metallic phase exists .
all states , except for those at _ critical fillings _ where the localization radius is larger than system size , are strongly localized .
the systems showing the quantum hall effect ( quantum hall systems ) are well suited to study the properties of critical states .
recently , in an experiment by cobden and kogan @xcite it was possible to extract the whole conductance distribution at criticality for a truly mesoscopic quantum hall system .
the main finding was a conductance distribution which is independent of system size ( within the mesoscopic regime ) , almost uniform between @xmath60 and @xmath38 .
this means in particular that the fluctuations are of the same order as the average value .
thus , fluctuations in the conductance can become much stronger than one could expect from classical transport theory where the relative fluctuations scale like the square root of the inverse volume .
this section prepares for the discussion of transport theory in mesoscopic electron systems .
we introduce the basic physical quantities of disordered mesoscopic systems for which the phase coherence length is larger than the system size .
they are introduced here on a phenomenological level as response quantities to applied fields .
the modeling of mesoscopic systems and the tools to calculate these quantities is the subject of the following sect .
[ mod ] ( and the appendix [ bri ] ) .
the basic physical quantities are the thermodynamic density of states and the transport quantities resistance , conductance and diffusion .
related physical scales are thermodynamic and kinematic scales like level spacing , fermi energy , fermi wavelength , magnetic length and transport scales like thouless energy , elastic mean free path and the localization length .
the thermodynamic density of states @xmath61 determines the change of the global particle density @xmath62 with respect to a change in the chemical potential @xmath63 .
the local thermodynamic density of states @xmath64 is defined similarly when replacing @xmath62 by the local particle density @xmath65 . in charge transport
one considers a stationary though non - equilibrium situation .
global transport quantities are the resistances in a multi - probe setup @xmath66 where probes are characterized by electro - chemical potentials @xmath67 . the current @xmath68 is driven from the current source ( probe @xmath69 ) to the current sink contact ( probe @xmath70 ) ( cf .
the resistances @xmath71 are determined by the conductance coefficients @xmath72 @xcite @xmath73 with the condition of total current conservation @xmath74 the conductance coefficients @xmath72 fulfill the relations,[2.4 ] ) can also be considered for time dependent transport with the quantities @xmath75 , @xmath76 being the fourier components for frequency @xmath77 . in that case @xmath71 and @xmath78 will depend on @xmath77 , too . the total current conservation and eq.([2.5 ] ) will still be valid as long as electric fields which are generated when applying voltages to the system ( including the coherent probe region ) have their sources and sinks within the system ( see e.g. @xcite ) . ]
@xmath79 in a two - probe geometry the transport is determined by the two - probe conductance @xmath3 ( ohm s law ) @xmath80 where @xmath81 and @xmath82 and @xmath83 .
a full description of the local charge transport is possible with the help of the _ conductivity tensor _ , @xmath84 , @xmath85 relating local electric field , @xmath86 , to the local current density , @xmath87 .
both can here be considered as the fourier components for frequency @xmath77 of the corresponding time - dependent quantities ( @xmath88 .
local currents can be also generated by charge density gradients under conditions of local equilibrium , i.e. the chemical potential @xmath63 is a function of position .
the relevant transport coefficient in this case is the _ diffusion function _ , @xmath89 , @xmath90 this name
is justified since the continuity equation between current and charge density , @xmath91 , @xmath92 tells that @xmath91 obeys a non - local diffusion equation @xmath93 under the conditions of local equilibrium the charge density can be related to the gradient of the chemical potential via the thermodynamic charge response , @xmath94 , @xmath95 since the linear response quantities conductivity and diffusion belong to the same system and describe currents in response to local electric and chemical potential gradients , respectively , they can be related by identifying the currents and driving fields .
this constitutes a generalization of the important _ einstein relation _ between conductivity and diffusivity , @xmath96 equation ( [ 2.7 ] ) reduces to ohm s law for spatially averaged currents and electric fields @xmath97 and for the _ global conductivity tensor _
, @xmath98 , eq . (
[ 2.12 ] ) reduces to the global einstein relation between conductivity and diffusion @xmath99 in the following we will concentrate on the d.c .
limit of transport ( @xmath100 ) and on a longitudinal ( dissipative ) component of the conductivity tensor , say @xmath101 . in the case
where off - diagonal components of conductivity vanish , e.g. in the absence of magnetic field , the two - probe conductance @xmath3 of a system with box geometry ( characterized by length @xmath6 and cross section @xmath102 , see fig .
12 ) is related to the conductivity @xmath2 by @xmath103 equation ( [ 2.16 ] ) is also referred to as ohm s law . whenever off - diagonal components of conductivity are present eq .
( [ 2.16 ] ) refers to the dissipative conductance which can be measured in an appropriate four - probe setup .
we stress that the global conductivity is treated here as a spatial average over the conductivity tensor , @xmath104 , and may depend on @xmath105 . the einstein relation , eq .
( [ 2.15 ] ) , and ohm s law , eq . ( [ 2.16 ] ) allow for a very interesting interpretation of the conductance measured in atomic units @xmath38 . defining a diffusion time @xmath106 by @xmath107 and a quantum mechanically related energy scale @xmath108 called thouless energy , and the thermodynamic level spacing , @xmath109 , by @xmath110 we obtain the thouless formula @xcite for the dissipative conductance , @xmath111 , @xmath112
this means that the dimensionless conductance , @xmath113 , can be expressed as the ratio of two characteristic energy scales of the problem : a transport related energy scale @xmath114 and a thermodynamically defined energy scale @xmath115 . as we will see later this expression
will help a lot in classifying mesoscopic electron systems .
so far we have not introduced any particular dynamical properties of the electron system . in a linear response treatment as outlined in the appendix sect .
[ bri ] thermodynamic as well as dynamic response quantities can be calculated by appropriate correlation functions . in a first attempt we make a phenomenological ansatz to describe the global dissipative conductivity @xmath2 thereby making contact with standard macroscopic transport theory . in the absence of magnetic fields and in the presence of static disorder one
can make the plausible assumption that an electron starting with velocity @xmath116 will decelerate owing to scattering by impurities .
this process can be modeled in the time dependent velocity correlator ( see eq .
( [ 4.4],[4.7b ] ) @xmath117 where @xmath118 is a phenomenological momentum relaxation time and @xmath119 the static velocity auto - correlator for a given direction and serves also as a measure of the kinetic energy of the initial state . with the help of the kubo relation , eq .
( [ 4.10 ] ) , this correlator is found to be @xmath120 ( @xmath121 being the number of electrons , @xmath122 the inverse temperature @xmath123 and @xmath124 being the electrons mass ) .
the global conductivity is proportional to the laplace transform of the velocity correlator ( see eq .
( [ 4.14b ] ) ) which yields the drude formula for the d.c .
conductivity @xmath125 for non - interacting electrons , at @xmath126 , eq .
( [ dru3 ] ) can be interpreted as @xmath127 where @xmath128 is the fermi energy , @xmath129 the fermi velocity and @xmath130 the density of states . in a finite system of volume
@xmath131 the density of states defines the level spacing @xmath132 which is the smallest quantum kinematic energy scale of the system .
the fermi velocity sets a quantum kinematic length scale , the _ fermi wavelength _
@xmath133 the momentum relaxation time is equivalent to an elastic _ mean free path _ , @xmath9 , that describes a typical distance between elastic collisions @xmath134 the corresponding diffusion constant follows from the einstein relation , eq.([2.15 ] ) , @xmath135 assuming that in a large but finite system the diffusion constant is indeed a constant the thouless energy , eq .
( [ 2.17 ] ) , and the conductance , eq .
( [ 2.18 ] ) , can be determined as @xmath136 here @xmath137 is a geometrical number , the surface area of the unit sphere in @xmath138 dimensions , e.g. @xmath139 in 2d .
the situation with @xmath140 corresponds to large @xmath113 and will be denoted as weak scattering case . within macroscopic transport theory
the transport is determined by material constants , e.g. the diffusion constant , and the basic task is to calculate them from microscopic models .
the situation changes when the system becomes smaller than the phase coherence length @xmath141 .
consider a given conducting system with fixed geometry and at fixed temperature .
imagine that we can pierce a hole in the conductor and put a magnetic flux through it . as quantum mechanics tells , such flux can only influence the phase of the electrons wave functions . in case that one observes periodic oscillations in the resistances when
changing the flux over several flux quanta , @xmath18 , ( electronic aharonov - bohm effect , see sect.[subele ] ) the electronic system is mesoscopic , i.e. the electronic wave functions are phase coherent over the entire conducting system .
as soon as the charge carrying wave packets experience scattering processes with dynamical degrees of freedom , e.g. phonons , that can dissipate energy from the electronic system the phase coherence will be destroyed and the electronic aharonov - bohm effect gets lost .
it is worth noticing that elastic scattering can not destroy the phase coherence .
thus , the criterion for an electron system to be mesoscopic is set by the _ phase coherence length _
@xmath10 being larger than the system size @xmath6 , @xmath142 here @xmath10 can be considered as a typical distance before the charge carrying states experience an inelastic scattering event .
it is evident that @xmath10 will usually decrease with increasing temperature . in practice
one has to go below liquid helium temperatures and use micro - structured electron systems to reach the mesoscopic regime .
the main consequence of the phase coherence in mesoscopic conductors is the fact that quantum interference plays a dominant role for physical quantities to be observed . as we have already seen in sec .
[ mes ] it can cause two important effects : unexpectedly large fluctuations and localization . the first effect can be qualitatively described as follows .
a wave packet entering the mesoscopic conductor will be scattered by some elastic collisions . after a while it will form a complicated interference pattern . if one changes an external parameter only slightly , e.g. magnetic field , one has to be aware of drastic effects on the measured physical quantity since the interference pattern changes .
a similar effect can be caused by a slight modification in the arrangement of elastic scatterers .
at least , one can not expect that a measured physical quantity is close to an average value obtained from a large number of measurements under slightly different conditions .
instead , the quantities will show fluctuations which can become unexpectedly large . however , such fluctuations are reproducible
. the magnitude of mesoscopic fluctuation effects will depend on system specific parameters .
this will be the topic of secs .
[ con],[starea],[statra ] . what can be learned by this reasoning
is that a characterization of a mesoscopic sample is incomplete unless the whole distribution function of the physical quantity of interest is obtained .
it is no longer possible to describe it by material constants ( see sect .
[ subuni ] ) .
this insight forms the motivation to ask for the statistical properties of mesoscopic electron systems .
the second effect , localization , was pointed out by anderson in a by now famous paper @xcite .
he demonstrated in a model calculation that the conductance of a phase coherent disordered ( mesoscopic ) system can decrease exponentially with increasing system size . denoting @xmath6 as the linear system size
, the conductance can scale as @xmath143 where @xmath144 is called the _ localization length _ is due to conventions which will be discussed later . ] .
the origin of this phenomenon lies in the multiple interference of the electrons wave packet after several elastic collisions that can lead to destructive interference such that the wave packet is bound to a finite region , the localization volume , with exponentially decreasing amplitude outside of this region . for system sizes that are several times larger than the localization length the conductor actually behaves as an insulator
the effect becomes even more interesting since it is possible that one system can change from the insulating , localized , phase to a conducting , delocalized , phase when parameters are changed continuously .
such parameters can be the degree of disorder , electron density or external fields like pressure , electric or magnetic field ( see sect .
[ subwea ] ) .
thus , mesoscopic systems can undergo a disorder induced metal - insulator transition which we called localization - delocalization transition ( ld transition ) .
in this section we introduce model systems appropriate for disordered mesoscopic systems and describe how to obtain relevant physical quantities . since we do not want to describe coupling of electrons to dynamic degrees of freedom that can dissipate energy from the electronic system we ignore from the very beginning electron - phonon interactions and restrict ourself to zero temperature , unless otherwise stated .
in addition , we will treat electrons mainly as independent additive systems characterized by effective mass , only .
this is a drastic restriction and it is in general _ not _ justified to incorporate electron - electron interactions into such effective one - particle models .
however , with one - particle models one can already gain some insight into the mechanism of ld transitions and fluctuation effects .
the interplay of disorder and electron - electron interaction forms a challenging research field which has still to be developed .
furthermore , internal degrees of freedom of the electrons will be ignored to a large extent , leading to models of non - interacting spin - less fermions in the presence of static disorder potentials .
independent electrons of effective mass @xmath124 moving in a random one - particle potential @xmath145 are described by the hamiltonian @xmath146 where the kinetic momentum @xmath147 may include magnetic fields @xmath148 .
randomness of the disorder potential has to be specified by a distribution function .
it is often assumed that the essential parameters of the distribution are contained in the first and second cumulant , i.e. a gaussian distribution , fixed by the mean value @xmath149 and a finite range correlation function @xmath150 , will be general enough to cover the essential physics .
it is reasonable to assume that neither specific points , nor specific directions are preferred by the whole ensemble of potentials .
then one can characterize the ensemble by two parameters , a potential correlation strength , @xmath151 , and a potential correlation length , @xmath152 , @xmath153 in case that @xmath152 can be considered to be smaller than other relevant microscopic scales , it can be set to zero , leading to a gaussian white - noise potential @xmath154 where @xmath155 corresponds to @xmath156 . within the gaussian white - noise potential model the disorder is characterized by one parameter , @xmath155 , of dimension @xmath157 $ ] .
we will see later , eq .
( [ 3.26 ] ) , how this parameter corresponds to an elastic mean free path , @xmath9 .
a homogeneous magnetic field pointing in a fixed direction , say @xmath158 , introduces a quantum mechanical length scale , the _ magnetic length @xmath159 _ defined by the size of an area which is penetrated by one flux quantum @xmath18 @xmath160 the magnetic length comes along with chirality due to the axial vector character of magnetic field .
as long as the magnetic length is larger than the actual system size , the effect of the magnetic field on the classical paths of electrons can be neglected and it will only influence the phases of wave functions .
as soon as the magnetic length is of the order of the mean free path @xmath161 the dynamics is , to a large extent , determined by the magnetic field and we refer to this situation as strong magnetic field case . a simple modeling to account for possible spin - orbit scattering events that can flip the electron s spin is given by the addition of @xmath162 to the hamiltonian in eq .
( [ 3.1 ] ) @xcite . here
@xmath163 is the spin operator and @xmath164 denotes an average spin - flip potential the gradient of which is taken to be a constant into all directions @xmath165 depending on the kinematic properties of the electron s state the inverse of @xmath166 gives rise to a distance , the _ spin - orbit scattering length _
@xmath167 , after which the electrons spin relaxes on average .
so far , we have considered a hamiltonian modeling in a continuous real space representation .
the randomness is characterized completely by a diagonal potential , eqs .
( [ 3.2a],[3.2b],[3.3 ] ) .
it is instructive to consider the same model in a discrete matrix representation and to ask for the resulting statistical properties . in a finite cubic system ( with volume @xmath131 ) with periodic boundary conditions we can use the momentum representation , @xmath168 , which diagonalizes the kinetic energy in eq .
( [ 3.1])(@xmath169 ) @xmath170 where @xmath171 is the fourier transform of @xmath145 .
the gaussian white noise results in the correlation @xmath172 thus , the random matrix @xmath173 has a diagonal deterministic part and a random part the average value of which vanishes , but with strong statistical correlations among those elements which fulfill momentum conservation .
what can be seen here is that the statistical properties of a representing matrix for the hamiltonian do , in general , depend on the chosen basis in hilbert space .
the matrix @xmath173 can be taken to be finite dimensional if one restricts the wave lengths to be larger than a convenient microscopic scale , @xmath174 .
then the matrix dimension is @xmath175 and the system can be studied on a computer . a direct modeling of disordered mesoscopic systems by a finite dimensional hamiltonian matrix is given by a tight - binding version of the gaussian white - noise potential and was introduced in the work of anderson @xcite .
this anderson model is defined on a cubic lattice with lattice constant @xmath174 @xmath176 where @xmath177 denote tight - binding states situated at lattice point @xmath178 and @xmath179 means nearest neighbors only .
the site energies @xmath180 correspond to the ( random ) potential energy and the ( deterministic ) hopping matrix elements @xmath181 to the kinetic energy .
again , the matrix dimension is @xmath175 for a cubic system of linear size @xmath6 .
however , the matrix is sparse since the nearest neighbor condition leaves most elements vanishing . in the absence of magnetic field
the hopping strength is taken to be equal for all nearest neighbors ( with coordination number @xmath182 ) and set to @xmath15 .
this defines the kinetic energy band to be of width @xmath183 .
a homogeneous magnetic field @xmath158 can be included in the kinetic energy by the peierls substitution @xcite . in the landau gauge ,
@xmath184 , it reads in 2d @xmath185 here @xmath186 is the number of flux quanta @xmath18 per unit cell .
in contrast to continuum models a feature of commensurability appears . requiring commensurability between the lattice constant and the magnetic length restricts @xmath187 to take rational values only @xmath188 furthermore , in this model the magnetic length
is restricted to be larger than the lattice constant since the effect of magnetic field is assumed to be weak on the size of a unit lattice cell , i.e. only its influence on phases is taken into account by the peierls substitution .
the statistical properties of the model are fixed once a distribution of the independent diagonal energies is chosen .
a convenient choice is a box distribution on the interval @xmath189 $ ] with correlation @xmath190 the parameter @xmath191 can be identified with @xmath192 ( @xmath193 taken from the white - noise model of eq .
( [ 3.3 ] ) ) and determines the mean free path @xmath9 ( see eq .
( [ 3.26 ] ) ) .
we mention that the inclusion of spin - orbit scattering into the anderson model is also possible @xcite and gives rise to a spin - orbit scattering length @xmath167 .
equations ( [ 3.1],[3.9 ] ) correspond to the most prominent models of disordered mesoscopic electron systems . in the following
we want to discuss , how one can use these models to calculate some of the basic physical quantities .
as outlined in the appendix , for non - interacting fermion models we can concentrate on zero temperature ( see eq . ( [ 4.22 ] ) ) and only need the resolvent of the hamiltonian @xmath194 to calculate the basic physical quantities ( eqs .
( [ 4.19 ] [ 4.27 ] ) ) .
choosing a convenient basis one has to perform traces involving @xmath195 .
the knowledge of @xmath195 is equivalent to finding the eigenvalues @xmath196 and eigenvectors @xmath197 of @xmath198 since @xmath199 the dynamic content of the resolvent is expressed by its relation to the time evolution operator @xmath200 @xmath201 with this it can be concluded that the long time average of the time dependent probability , @xmath202 , to find an electron at point @xmath203 that was created at point @xmath204 is related to the green s function , @xmath205 , by @xmath206 that leads to the important interpretation that @xmath207 defines the long time transition probability from @xmath204 to @xmath203 for a given energy @xmath208 . the local density of states ( ldos ) at @xmath126
, @xmath209 can be calculated by ( see eq .
( [ 4.17 ] ) ) @xmath210 and the global ( averaged over volume @xmath131 ) density of states ( dos ) by @xmath211 for a finite closed system the dos consists of isolated @xmath212 peaks which have to be smeared out by attaching a finite width of the scale of the level spacing to the peaks or by averaging over the ensemble of disorder realizations . the average level spacing @xmath115 at @xmath213 can thus be obtained by @xmath214 furthermore , it is of conceptual importance to mention that the green s function @xmath215 can be represented by a path integral , since @xmath216 \exp ( i/\h)s[\r(t ) ] \ , , \label{3.19b}\ ] ] where the integration runs over all paths @xmath217 connecting initial ( @xmath204 ) and final ( @xmath203 ) points and @xmath218 $ ] denotes the classical action associated to the hamiltonian of the electron
. the stationary path of @xmath218 $ ] yields the solution of the classical equation of motion . as a first instructive example we calculate the ensemble average of the green s function in the continuous model without magnetic field , using the momentum representation , eqs .
( [ 3.7],[3.8 ] ) .
since we can not solve the schrdinger equation for arbitrary disorder potential we rely on a perturbative analysis , @xmath219 where @xmath220 is the green s function of the kinetic energy term . due to the property eq .
( [ 3.8 ] ) the ensemble averaged green s function is diagonal in momentum representation , too .
it can be written in the form @xmath221 where @xmath222 is called the self - energy the imaginary part of which defines an inverse time scale which can be interpreted as an elastic mean free time @xmath118 @xmath223 we mention that , in general , this @xmath118 must not be identical to the average momentum relaxation time due to random collisions . the self - energy can now be calculated order by order in perturbation theory ( for details cf .
@xcite ) . for weak scattering ,
i.e. for @xmath140 , it turns out to be sufficient to take the first non - vanishing contribution which yields @xmath224 and improve it by replacing @xmath225 with the full resolvent @xmath226 in eq .
( [ 3.23 ] ) making the approximation for the self energy self - consistent .
it leads to @xmath227 where @xmath228 is the average dos which , for fermi energies far off the kinetic energy band edges , is not changed as compared to the dos without disorder , i.e. @xmath229 consequently , the elastic mean free path is related to @xmath193 and @xmath8 by @xmath230 calculating the disorder average of the conductivity requires already the average over a product of two green s functions ( called two - particle green s function ) which is already a non - trivial task .
starting from the kubo formula , eq .
( [ 4.23 ] ) , for the velocity correlator , using the momentum representation and replacing the matrix elements of the resolvent by their average values leads back to the drude formula with the momentum relaxation time given by eq .
( [ 3.24 ] ) .
however , this procedure ignores completely that the average of two green s functions does not decouple into a product of averages . in the context of the path integral representation for the green s functions , eq .
( [ 3.19b ] ) , it becomes obvious that this method can not cope with quantum interference effects .
thus , the mean free time as given by eq .
( [ 3.24 ] ) corresponds to a drude like conductivity when quantum interference is suppressed .
the symmetry properties of the disorder ensemble yield a translational and rotational invariant two - particle green s function @xmath231 with this function the longitudinal conductivity , as given by eq .
( [ 4.27 ] ) , is @xmath232 the diffusion constant @xmath233 is defined by the long time limit of the square displacement with respect to the probability distribution @xmath234 as @xmath235 where @xmath236 is a growing time scale as @xmath237 is sent to @xmath238 .
thus , eq . (
[ 3.28 ] ) demonstrates the validity of the einstein relation .
furthermore , it allows for a first definition of a localization length by means of the green s function , i.e. if @xmath239 stays finite for arbitrary finite @xmath204 , but falls off as @xmath240 the average conductance of a large volume @xmath131 will show the localization phenomenon introduced in eq .
( [ 2.21 ] ) with @xmath241 defining a localization length .
note , that this length is defined with respect to the average of @xmath242 and must not be identical to just @xmath243 of the exponential fall off of averages of powers @xmath244 . nevertheless , the factor of @xmath245 in eq .
( [ 3.30 ] ) follows the convention to associate the localization length with the modulus of the green s function .
we are now in the position to formulate the statistical problem of calculating physical quantities in mesoscopic electron systems . the system is described by an ensemble of hamiltonian matrices in a certain matrix representation .
the relevant physical quantities can be obtained from the green s function @xmath246 .
the average value will only determine the average dos and the short distance mean free path , however not the global transport behavior . to account for the latter , the averaged two - particle green s function is needed in order to see localization effects .
fluctuation effects in global transport can not be obtained on the basis of averaged two - particle green s functions ; at least four - particle green s functions have to be considered .
we see that this situation calls for a general approach to the distribution functions of @xmath246 . the field theoretic approaches to mesoscopic systems commonly known as non - linear sigma - models are constructed to fulfill this purpose@xcite .
we will not enter this subject because of its technical complexity , but will here focus on some general considerations .
owing to eq .
( [ 3.14 ] ) the statistics of @xmath246 is contained in the joint probability distribution of eigenvalues , @xmath196 , and eigenvectors , @xmath247 , @xmath248 of course , it seems hopeless to find a general method to determine this function for any given ensemble of hamiltonian matrices .
to get some feeling about the nature of the statistical problem we consider the hamiltonian in a finite basis @xmath249 as hermitean @xmath250 matrix @xmath251 .
the diagonalizing unitary matrix @xmath252 with matrix elements @xmath253 fulfills @xmath254 it is related to the amplitude of an eigenstate @xmath247 in the @xmath255 basis by @xmath256 we can think of the randomness of @xmath198 as being controlled by a large number of independent parameters , e.g. strength and position of point scatterers in space , or the on - site energies in the anderson model .
each realization of the hamiltonian represents one point in this high dimensional parameter space .
we can ask what happens for a certain stochastic process @xmath257 in this parameter space where @xmath258 denotes a fictitious time and each element @xmath257 is weighted by the corresponding probability .
the original @xmath198 corresponds to some arbitrary point , say @xmath259 .
we can increase @xmath258 in small steps @xmath260 which leads to @xmath261 the corresponding unitary matrix @xmath262 behaves then as @xmath263 where @xmath264 denotes the unitary matrix that diagonalizes the hamiltonian @xmath265 with respect to the eigenvector basis of @xmath257 .
thus , @xmath262 evolves in a _ multiplicative _ manner .
one could think of treating @xmath264 by perturbation theory with respect to @xmath260 to derive closed evolution equations for the probability distribution of @xmath266 and the corresponding set of eigenvalues @xmath267 .
this has , so far , not been undertaken .
however , recently chalker et al .
@xcite adopted a related picture to investigate the eigenvalue statistics separately by relying on simplifying approximations . here , we only wanted to stress , that the problem of the statistics of eigenvectors can be viewed as the problem of a random multiplicative process for unitary matrices .
as to the problem of the statistics of eigenvectors one can easily imagine two extreme situations .
in the first situation the probability distribution of @xmath268 is peaked at a single fixed matrix . this matrix singles out a certain basis of eigenstates .
we choose this matrix as the unit matrix , such that the hamiltonian is diagonal in the initial basis @xmath269 the corresponding eigenstates are localized at certain ` sites ' , i.e. @xmath270 the second extreme situation corresponds to an _ isotropic _ distribution for the unitary matrix . by this
we mean that the probability density , @xmath271 , to find a certain unitary matrix @xmath268 within the volume element , @xmath272 $ ] , is equal for all elements @xmath273 .
the volume element itself stays invariant under the action of group transformations ( invariant measure ) @xmath272 $ ] , i.e. it does not single out any particular element .
the corresponding eigenstates are also isotropically distributed among all possible eigenstates , no basis is preferred .
of course , both extreme situations are not generic ones and we have to see for which values of physical scales they may appear .
for mesoscopic conductors with several probes the d.c .
conductance coefficients @xmath78 are determined by asymptotic current correlators , eq .
( [ 4.15 ] ) , and hence for independent fermions at @xmath126 by asymptotic green s functions , eq . ( [ 4.26 ] ) . in a geometry where probes are represented as infinite leads attached to the conductor
the asymptotic green s functions determine the matrix of transmission amplitudes,@xmath274 , to scatter from channel @xmath275 in lead @xmath69 to channel @xmath122 in lead @xmath124 and the bttiker formula relates them to the conductance , eq .
( [ 4.27b ] ) , @xmath276 this formula tells that the dimensionless conductance , especially in a two - probe geometry , is the number of effective transmitting modes , @xmath277 and it is thus of similar conceptual importance as the thouless formula , eq .
( [ 2.19 ] ) . the transmission matrix can be calculated by asymptotic green s functions .
these have to be determined under the requirement of attaching the leads to the conductor . for this purpose
it is advantageous to divide the hilbert space into two parts by the projection operator @xmath278 onto the hilbert space of the conductor and its counterpart @xmath279 that projects onto the leads . denoting projected operators @xmath280 by @xmath281 , etc .
the general algebraic solution of the problem to calculate the projected green s function reads @xmath282 where the effective ( in general energy dependent and non - hermitean ) , hamiltonian @xmath283 is defined via the couplings and the green s function of the leads @xmath284 @xmath285 since eq .
( [ 3.41 ] ) is defined for a finite system and the infinite clean leads green s function can be determined analytically for not too complicated geometries the method of effective hamiltonian also helps in doing the calculations numerically .
the green s function evaluated at the surface between leads and conductor will then determine the conductance coefficients .
the bttiker formula points out that the mesoscopic conductor has very much in common with optical wave guides where transmission probabilities between incoming and outgoing wave modes are the central physical quantity of interest .
it is thus tempting to model mesoscopic conductors by scattering matrices in analogy to optical wave guides . as a guiding example
let us consider a box shaped system ( see fig .
12 with two semi - infinite leads attached serving as particle reservoirs ( contacts ) ) .
the cross section @xmath102 is kept fixed while the the length @xmath6 is treated variable .
such systems are denoted as quasi - one - dimensional ( quasi-1d ) conductors .
the leads are characterized by discrete branches of energy dispersion ( see fig .
13 ) due to the finite extension in transverse direction . for a given value of the fermi energy there exist @xmath286 quantum states of incoming and outgoing waves ( channels ) . as a rough estimate one may think of @xmath286 being the number of lattice points one can put on the cross section of the conductor with lattice spacing of half a fermi wavelength , @xmath287 .
a direct consequence of the bttiker formula is the integer quantization of dimensionless conductance in ideal ballistic conductors where no scattering occurs , or the mean free path @xmath9 is much larger than @xmath288 .
then @xmath289 and the conductance displays quantized plateaus as a function of the fermi energy ( or particle density ) .
jumps from one plateau to another occur whenever a new channel is occupied , i.e. the fermi energy crosses a new branch of the leads energy dispersion .
in contrast to classical physics the conductance of an ideal conductor is not infinite but bounded by the finite number of quantum channels .
the classical behavior can only be recovered for vanishing fermi wavelength or , equivalently , for infinite number of channels .
the finite resistance of ideal ballistic conductors can be attributed to the contacts where the equilibration of the electrons take place .
therefore , one may think of a decomposition of the resistance into a _ contact resistance _
, @xmath290 , and an _ intrinsic resistance _ , @xmath291 , such that @xmath292 .
this yields for the corresponding intrinsic conductance @xmath293 , a result that was obtained by langreth and abrahams @xcite generalizing the one - channel version of landauer s pioneering works @xcite .
however , one should keep in mind that the notion of intrinsic conductance is a theoretical construction and not a quantity to be measured in a two - probe experiment . such experiment corresponds to the conductance as described by the bttiker formula . in general ,
the quasi-1d conductor can be described by a @xmath294 scattering matrix @xmath295 @xmath296 connecting incoming and outgoing channels left and right of a scatterer , @xmath297 , @xmath298 , @xmath299 and @xmath300 being @xmath301 matrices of transmission and reflection coefficients for scattering from left to right and vice versa , respectively ( see fig .
14 ) . here
@xmath297 is precisely the matrix of transmission amplitudes that appear in the bttiker formula . due to probability - flux conservation @xmath295
is unitary , @xmath302 to model a long quasi-1d system one can now add several conductors in series . however , the total @xmath295-matrix is not multiplicative . instead of @xmath303
we are seeking a _ transfer matrix _
@xmath304 with the property @xmath305 , i.e. @xmath306 a straightforward calculation yields @xmath307 making use of probability - flux conservation @xmath308 where @xmath15 stands for the @xmath301 unit matrix we can rewrite the dimensionless two - probe conductance as @xmath309 a modeling of a mesoscopic quasi-1d conductor can now be based directly on the scattering matrix or , equivalently , on the transfer matrix .
this can be done by fixing the statistical properties of the @xmath295-matrix corresponding to a small strip of length @xmath310 , denoted as strip @xmath295-matrix @xmath311 and compose the whole conductor by putting statistically independent strip @xmath295-matrices in series .
the total @xmath295-matrix follows then from the multiplication of the corresponding strip transfer matrices @xmath312 .
the assumption of statistical independence is justified if the strip length @xmath310 is larger than the microscopic potential correlation length @xmath152 introduced in sect .
[ subrel ] .
furthermore , this modeling allows for a simple description of the mean free path corresponding to the strip @xmath295-matrix .
as long as the corresponding reflection probabilities @xmath313 are small compared to @xmath15 the mean free path @xmath9 is large compared to @xmath310 and can be defined as follows @xmath314 the @xmath295-matrix fulfills the requirement of unitarity . to model systems with specific symmetry properties
one can impose further symmetry constraints on @xmath295 .
the most important case is that of time reversal symmetry .
this symmetry can be broken by magnetic fields . to model systems with magnetic field
one faces the problem that the magnetic length ( see eq .
( [ 3.4 ] ) ) as a chiral length scale has to be incorporated into the model .
it is not obvious how to do that in the scheme described here .
however , weak magnetic fields only break time reversal symmetry ( the magnetic length being larger than system size ) and can be simply incorporated in the symmetry properties of @xmath295 . for example , time reversal operates by interchanging incoming and outgoing channels and complex conjugation of amplitudes . in case of time
inversion symmetry the @xmath295-matrix and transfer matrix @xmath315 fulfill @xmath316 internal degrees of freedom , such as spin , can also be incorporated by taken the corresponding time reversal transformation into account .
so far , the modeling rests on a microscopic length scale , the mean free path @xmath9 , and symmetry properties .
the statistical problem is thus defined by ( i ) fixing the distribution of the strip @xmath295-matrix @xmath311 and ( ii ) by applying the composition law for the corresponding transfer matrices @xmath317 this defines a stochastic multiplicative _ process _ where the _ time variable _ is the system length .
this resembles eq .
( [ 3.35 ] ) .
( iii ) the statistical properties of the increment @xmath318 are known by construction . following the steps ( i ) ( iii ) it is possible to construct a stochastic differential equation ( fokker - planck equation ) for the distribution function @xmath319 ( see e.g. @xcite ) .
still , a general solution of this equation is presently not available and we will not discuss it in detail .
a related approach , based on simplifying assumptions , will be discussed in sect .
[ substaqua ] .
the matrices @xmath320 and @xmath321 occurring in the conductance formulas , eqs .
( [ 3.38 ] , [ 3.47 ] ) , are hermitean and , thus , can be diagonalized .
the positive eigenvalues are denoted as @xmath322 ( for @xmath320 ) and as @xmath323 ( for @xmath321 ) .
the eigenvalues of @xmath321 appear in inverse pairs and we can restrict to those with @xmath324 .
the conductance reads in the corresponding eigenvalue representation @xmath325 here the eigenvalues @xmath326 have to be distinguished from the total probability to be scattered from a channel @xmath275 into forward direction , @xmath327 , that occurs in eq .
( [ 3.39 ] ) . a related _ polar parameterization _ ( cf .
@xcite ) of the transfer matrix @xmath315 in terms of _ radial coordinates _ @xmath328 and four unitary @xmath301 matrices @xmath329 is also used frequently @xmath330 here @xmath331 and @xmath332 it worth noticing that in these @xmath295-matrix models the conductance itself appears as a linear statistics of the random parameters , @xmath333 . ] of eigenvalues of a certain random matrix .
however , one should keep in mind that the statistical properties of these eigenvalues are , in general , not independent of those of the associated eigenvectors .
before we introduce generalizations of the @xmath295-matrix modeling we will show that the present one is well suited to discuss the problem of localization in quasi-1d conductors .
an instructive example is the 1d conductor for which @xmath334 .
the composition law eq .
( [ 3.50 ] ) tells that the composition law for the transmission reads ( see fig .
15 ) @xmath335 exploiting unitarity of @xmath295 yields for the transmission probability @xmath336 , @xmath337 , @xmath338 where @xmath26 is the sum of phases of @xmath339 and @xmath340 . based on eq .
( [ 3.54 ] ) one can derive an evolution equation for the probability distribution of @xmath341 with increasing length @xmath6 .
we will postpone this to sect .
[ subscadis ] .
here we will take advantage of the fact that an average of @xmath342 over a uniform distribution of phases @xmath26 yields the simple expression @xcite @xmath343 thus , the phase - average of the logarithm of the transmission turns out to be additive .
iterating this procedure tells that the logarithm of the transmission will be distributed in a gaussian way according to the central limit theorem for independent additive random numbers .
this motivates to call @xmath344 the _ typical _ transmission , and to write down a _ quantum _ series composition law for conductance @xmath345 in 1d conductors @xmath346 this composition law is in striking contrast to the macroscopic composition law for incoherent conductors , @xmath347 .
equation ( [ 3.56 ] ) immediately tells that the conductance will exponentially decrease with increasing system length @xmath6 , which means localization . to determine the localization length @xmath144 we calculate @xmath348 thus , @xmath349 with @xmath350 , and by eq .
( [ 3.48 ] ) @xmath351 this result demonstrates that in one - dimensional conductors all states are localized .
furthermore , the localization already occurs on the scale of the mean free path .
this is a crucial mesoscopic effect as it is a consequence of the multiplicative quantum series composition , eq .
( [ 3.56 ] ) . for arbitrary channel numbers
@xmath286 the above 1d result does not apply .
however , one can make an educated guess what might happen to the localization length , if localization occurs at all . to this end
, we notice that the conductance in 1d for length @xmath6 much smaller than @xmath9 is given by @xmath352 in accordance with our definition of @xmath9 . since the result @xmath353 corresponds to ideal transmission of the wave guide an ohmic series decomposition into _ contact resistance _ @xmath15 and _ intrinsic resistance _ ,
@xmath354 , is meaningful . assuming that for @xmath355 and @xmath356 ohm s parallel composition law is applicable yields the _ classical conductance _
@xmath357 and the natural limit for this law to hold , @xmath358 , yields an estimate of @xmath359 @xmath360 we are now going to discuss rigid results that confirm the localization effect for arbitrary @xmath286 .
in sect . [ substaqua ]
we will see that our estimate of the quasi-1d localization length is in accordance with known results .
a crucial point in the 1d calculation was the multiplicative composition law for typical conductance .
the transfer matrix obeys also a multiplicative composition law and the question arises what can be said about the statistical properties of a large product of random transfer matrices .
recall that the product of independently distributed positive random numbers @xmath361 , @xmath362 gives rise to a central limit theorem for its logarithm : for large @xmath121 , the distribution of @xmath363 is well approximated by a gaussian with mean value @xmath364 and variance @xmath365 for the original variable @xmath366 the distribution is called _ log - normal distribution _ and reads @xmath367 furthermore , one can conclude that for almost all realizations the geometric mean asymptotically coincides with the typical value @xmath368 there exist some extensions of the law of large numbers to the case of products of independent random transfer matrices ( see @xcite ) . a version of the theorem of oseledec @xcite guarantees the existence ( with probability @xmath15 ) of eigenvalues of the diagonalizable limiting matrix @xcite @xmath369 ( @xmath370 , @xmath11 some fixed reference length ) of the form @xmath371 motivated by the analogy between transfer matrices and time evolution matrices considered in chaotic dynamics the quantities @xmath372 are called _
lyapunov exponents_. the lyapunov exponents @xmath373 are asymptotic non - random numbers and are the so - called self averaging limit of the random numbers @xmath374 , the exponents of eigenvalues of @xmath375 , divided by @xmath376 @xmath377 a system with channel number @xmath286 has thus @xmath286 characteristic length scales @xmath378 according to the conductance formula eq .
( [ 3.51 ] ) the inverse of the smallest lyapunov exponent yields the relevant localization length @xmath379 a theorem by frstenberg @xcite states that the largest lyapunov exponent is always positive .
this proves localization in strictly 1d , but is meaningless for @xmath355 .
virster @xcite has later proofed that @xmath380 is strictly positive , as long as @xmath286 is finite .
thus , in the quasi-1d limit @xmath381 the localization length @xmath144 is finite .
we postpone the discussion of the statistics of @xmath382 to sect .
[ substaqua ] and proceed by introducing more general @xmath295-matrix models .
so far , the modeling of mesoscopic systems by scattering matrices was restricted to the quasi-1d case .
more general @xmath138-dimensional mesoscopic systems can be described by a model introduced by shapiro @xcite . instead of taking a strip @xmath295-matrix
one can consider a regular network ( lattice spacing @xmath174 ) consisting of @xmath383 sites and @xmath384 bonds ( see figs .
16 , 17 ) .
each site represents a scatterer and is described by a unitary @xmath385 scattering matrix @xmath295 .
the bonds represent free propagation between the blocks .
each bond carries two waves propagating in opposite directions .
each @xmath295-matrix transforms @xmath386 incoming amplitudes into @xmath386 outgoing amplitudes . to each bond amplitude a random phase is added which simulates an irregular lattice of scatterers which are distributed between distance @xmath174 and @xmath387 where @xmath8 is the wavelength of scattered waves . by specifying appropriate boundary conditions one can describe multi - probe conductors within this network model . consequently , physical quantities like conductance coefficients and even channel dependent transmission amplitudes can be studied with less effort than in systems described by a hamiltonian .
interesting is the fact that the @xmath295-matrix modeling does also allow to determine energy eigenstates and eigenvalues .
this comes along with the following interpretation : mapping incoming to outgoing states by scattering matrices usually refers to a process consuming infinite time in order to guarantee energy conservation .
however , if one allows for a certain uncertainty in the energy , a finite time is enough to follow the scattering process
. therefore , one can take the point of view that the scattering matrices at each site describe a unitary time evolution of states in one unit of time . by this interpretation
the finite energy uncertainty is treated as negligible .
accordingly we define the unitary matrix of time evolution in a unit of time by @xcite @xmath388 here @xmath389 forms , for each time @xmath297 , a set of @xmath390 complex bond amplitudes .
normalizing them , @xmath391 matrix @xmath268 is uniquely determined by the network of scattering matrices since it summarizes nothing but the scattering conditions at the sites .
energy eigenstates @xmath392 are stationary solutions of the unitary operator @xmath268 @xmath393 and thus correspond to the eigenstates of @xmath268 with eigenvalue @xmath15 .
the energy of the underlying electron system enters here only parametrically through the scattering amplitudes . for simplicity ,
let us assume that the scattering strengths are equal at all nodes ( abbreviated by one symbol @xmath258 ) and randomness enters only through the bond phases .
then @xmath268 will have an eigenstate to eigenvalue @xmath15 only for a discrete set of scattering strengths , which correspond to a discrete set of energy eigenvalues of the underlying electron system . instead of looking for the discrete set of energies one can also fix the energy and ask for the more general eigenvalue problem @xmath394 @xmath395 are , for fixed @xmath208 , a set of @xmath390 eigenstates at energy @xmath208 of slightly modified disorder configurations , each of the modifications being an overall shift in random phases on the links .
the eigenphases @xmath396 can be considered as an energy excitation spectrum at energy @xmath208 .
thus , the network models allow to investigate energy eigenvector statistics and , via the phases @xmath396 , energy eigenvalue statistics .
the network models have a further advantage as compared to hamiltonian models concerning numerical calculations . in case that only a finite resolution of energy scales is required
one does not need to fully diagonalize the unitary operator .
instead , one iterates the time evolution step by step until a certain time scale is reached , the inverse of which yields the required energy resolution . finally , we like to point out that within the network models there are possibilities not only to include global symmetries but also characteristic length scales .
for example by allowing for internal degrees of freedom ( e.g. spin ) the number of amplitudes on each bond can be increased and scattering strengths can be introduced that take the change of internal degrees of freedom into account .
the inclusion of strong magnetic fields with their characteristic chiral character is also possible .
no commensurability features like in the anderson model with peierls substitution do occur .
especially , one model of this category has become widely known as the chalker - coddington model @xcite in 2d which only allows for scattering to the left and right ( see figs . 18 , 19 ) .
in this section we consider idealized mesoscopic systems and ask for the statistics of energy eigenvalues and eigenvectors .
the idealization is due to the assumption that the statistical properties of both can be studied separately .
we distinguish between the localized situation imitated by a diagonal hamiltonian and the delocalized situation imitated by a standard random hermitean matrix having independent gaussian distributed random entries .
we will review the energy level statistics in these systems and the statistics of the local box - probability @xmath397 for an electron that is in an eigenstate @xmath398 to be found in a certain box of volume @xmath399 , @xmath400 the complete information about the ( energy ) level statistics of a @xmath250 hamiltonian is contained in the joint probability density @xmath401 to simplify notations we will assume throughout that any reordering of levels in @xmath402 does not change the probability . the simplest statistical quantity to be extracted is the average level density @xmath403 defining the average level spacing @xmath404 .
the two - level correlation function is @xmath405 for simplicity we assumed that it is translational invariant within a band of states with @xmath115 being independent of energy . by writing the energy separation in units of average level spacing ( @xmath258 ) one
focuses on those properties that are independent of the chosen energy units .
quantities which contain information about @xmath62-level correlation functions are the @xmath62-level spacing distribution @xmath406 which yields the probability to find exactly @xmath62 levels in an energy interval of width @xmath258 ( the center of which can be arbitrary ) .
the probability to find exactly no level inside the interval defines the so - called level spacing distribution @xmath407 which yields the probability density to find a level spacing @xmath258 with respect to @xmath408 .
the average number @xmath409 of levels in an interval @xmath258 and its variance @xmath410 are also of interest .
the derivative of @xmath411 with respect to @xmath412 defines ( for large @xmath413 ) a _ level compressibility _ @xmath414 we consider @xmath121 orthogonal energy states @xmath415 each of which is localized to one of @xmath121 sites of a regular lattice ( lattice spacing @xmath174 ) in an arbitrary dimension @xmath138 , leading to a _ vanishing localization length_. the corresponding energy eigenvalues are independent random variables taken from a distribution @xmath416 the corresponding random hamiltonian is diagonal in site representation @xmath417 with eigenstates @xmath418 this situation was already addressed in sect .
[ subthe ] as one of two extreme cases for the eigenvector statistics . as compared to the anderson tight - binding hamiltonian , eq .
( [ 3.9 ] ) , the hamiltonian in eq . ( [ 5.6 ] ) ignores the kinetic energy completely .
the energy level statistics is contained in the joint probability distribution @xmath419 consequently , the average level density is @xmath420 and the average level spacing within the band ( @xmath421 ) is @xmath422 , @xmath423 .
the two - level correlation function @xmath424 reflects the fact that there is only auto - correlation between levels ( up to @xmath425 corrections ) .
@xmath426 the level spacing distribution follows the poisson law @xmath427 ( that is why the ensemble is called _ poisson ensemble _ ) and the characteristic relation between number variance and average number of levels can be obtained @xmath428 indicating that the levels are compressible with compressibility @xmath15 . the probability to find an electron in a box of size @xmath174 in the eigenstate @xmath247
is @xmath429 increasing box sizes to @xmath430 does not change the result as long as the boxes are chosen to be non - overlapping and centered at the original point . independently of the box size the probability is either zero , if the state is not located in the box or @xmath15 in the opposite case .
for a given energy @xmath208 we consider the microcanonical average of @xmath431 @xmath432 where a smearing - out of @xmath212-functions is understood .
the smearing width @xmath433 has to be adjusted such that isolated @xmath212-peaks are broadened , but not over many neighboring levels .
a convenient choice is to take @xmath433 of the order of the actual level spacing at @xmath208 in a given realization .
the denominator in eq .
( [ 5.13b ] ) is then @xmath434 .
since the eigenvalues are independently distributed , the centers of the corresponding eigenstates are randomly distributed over the lattice points .
therefore , in almost any case one will find zero box - probability at a certain site @xmath435 , unless the energy @xmath208 coincides ( within the scale of the smearing @xmath433 ) with eigenvalue @xmath436 , in which case the box - probability is @xmath15 .
the average box - probability is finite and given by @xmath437 .
thus , the probability distribution @xmath438 of the box - probability is obtained as @xmath439 the average value is not characteristic for this distribution , since the main weight is given to @xmath440 .
however , e.g. the geometric mean @xmath441 reflects a _ typical _ box - probability .
alternatively , one may consider the median of the distribution ( that value up to which half of the total weight is used up ) which vanishes also and could be termed _
typical _ as well .
the ldos @xmath442 with a smearing - out of @xmath212-functions on the scale of @xmath433 yields for the ldos in a box , @xmath443 @xmath444 however , we must mention that this notion of ldos in a box differs from that used by other authors ( see e.g. @xcite ) .
if one considers the ldos with a parametric lorentzian smearing - out of @xmath212-functions , @xmath445 where @xmath446 can vary from @xmath60 to @xmath447 one can investigate parametrically the interplay of level and eigenvector statistics . to simplify the discussion
we will be often more restrictive and call @xmath448 the _ reduced _ ldos in a box .
this definition concentrates on the eigenvector statistics .
consequently , the distribution function reads @xmath449 in summary , ideal localized systems ( with vanishing localization length @xmath144 ) are characterized by a compressible spectrum of uncorrelated energies and a ldos in a box that typically vanishes , although the average dos is finite . for the idealized delocalized phase we make the assumption that eigenstates are distributed isotropically within the space of all possible eigenstates for a @xmath250 hamiltonian matrix , @xmath450 . in other words the diagonalizing unitary matrices @xmath253 , eq .
( [ 3.32 ] ) , are uniformly distributed with respect to the invariant measure @xmath272 $ ] .
this situation was already addressed in sect .
[ subthe ] as the second extreme case for the eigenvector statistics .
therefore , we assume the distribution of the hamiltonian matrix to be unitarily invariant .
one of the simplest possible choices is the so - called gaussian unitary ensemble ( gue ) introduced by wigner @xcite when studying the level statistics of complex nuclei , @xmath451 here @xmath452 is a normalization constant , @xmath453 some arbitrary energy scale and the volume element is defined in terms of the independent matrix elements of @xmath198 as @xmath454 since @xmath455 the gue describes a random matrix with all its elements uncorrelated .
each of the elements vanishes on average and its absolute value fluctuates , @xmath456 .
if we think of the model in site - representation and compare it to the anderson model we see that by eq .
( [ 5.17a ] ) hopping terms are dominant and hops from one site to another are equally probable independent of the spatial distance .
this means that we can associate a _ vanishing time scale _ @xmath106 to travel through the system .
introducing eigenvalues @xmath196 and eigenvectors @xmath457 one can transform the probability @xmath458 to these variables at the expense of introducing a jacobian between the set of variables .
fortunately , the jacobian can be calculated , the integration over @xmath272 $ ] being a trivial normalization ( since @xmath459 is unitarily invariant ) , and one obtains for the joint probability of eigenvalues @xmath460 the factors in front of the exponential are due to the jacobian and ensure that the probability to find two levels close to each other vanishes ; a phenomenon which is denoted as _ level repulsion_. the factors can be rewritten as @xmath461 such that the joint probability describes a classical gibbs ensemble , @xmath462 of a gas of _
particles _ with coordinates @xmath196 and a hamiltonian @xmath463 that contains a logarithmic two - body interaction @xmath464 and a one - body ( confining ) potential @xmath465 .
the _ inverse temperature _ is @xmath466 for the gue .
a related ensemble of real symmetric matrices , reflecting time inversion symmetry , denoted as gaussian orthogonal ensemble ( goe ) , gives rise to the same gibbs ensemble , the only change being @xmath467 ( @xmath468 corresponds to spin systems with time reversal symmetry , for more details and review see @xcite ) .
the interpretation of the joint probability distribution of levels as a gibbs - ensemble has led to an important mean - field approach to level statistics @xcite which we will use later in the context of quasi-1d conductance ( cf . sect . [ substaqua ] ) . in the standard random matrix of gue and
goe a number of results are well known : 1 .
the average dos is ( for @xmath469 ) given by wigner s _ semi - circle law _
, @xmath470 with average level spacing ( @xmath471 ) @xmath472 .
2 . the two - level correlation function shows level - repulsion for @xmath473 and decays @xmath474 for @xmath475 .
in the gue it reads ( @xmath476 , @xmath477 ) @xmath478 3 .
the level spacing distribution is very well approximated by wigner s surmise @xmath479 where e.g. @xmath480 .
the characteristic small @xmath258 behavior @xmath481 reflects the power of level - repulsion characteristic of the symmetry of the hamiltonian .
4 . the level number variance @xmath411 depends only logarithmically on the average level number , @xmath482 which yields a vanishing level compressibility ( @xmath483 ) .
eigenvectors @xmath457 are isotropically distributed uncorrelated gaussian variables up to @xmath425 corrections . more precisely , in the gue the eigenvector components are complex with uncorrelated gaussian distributed real and imaginary parts .
the average of each is zero while the variance is , due to normalization , given as @xmath484 for the goe the components are real with vanishing average and variance @xmath485 .
this fixes the distribution of the probability @xmath486 to find an electron in a box of size @xmath174 ( @xmath174 lattice spacing ) . in the gue
one finds the rayleigh distribution @xmath487 and in the goe the porter - thomas distribution @xmath488 both distributions follow immediately once an intensity of waves @xmath392 is considered which are large sums of either complex or real numbers .
this leads , by the law of large numbers , to gaussian distributions for @xmath392 .
both distributions can be generalized to arbitrary box size @xmath489 by replacing @xmath490 with @xmath491 .
characteristic of these distributions is the exponential tail and the fact that no parameter besides @xmath492 occurs .
still , the average value is a typical value for the distribution ; e.g. the median differs from the average value by only a factor of @xmath493 . the statistics for the reduced ldos in a box , @xmath494 ,
is then straightforward . in summary , ideal delocalized systems ( with vanishing time scale @xmath106 ) are characterized by an incompressible spectrum of correlated energies and a ldos in a box the distribution of which develops an exponential tail .
still , the average value is typical .
the level and eigenvector statistics of the idealized systems show a remarkable feature : no physical scale besides the kinematic scales , level spacing @xmath115 , total volume @xmath495 and box volume @xmath399 enters the statistical quantities .
the parameters of the models , @xmath496 ( for the localized phase ) , and @xmath453 only define the energy unit for @xmath115 . no transport scale enters the models .
this is consistent with the following interpretation : in the localized phase the only relevant scale is the localization length , which was treated as zero . in the idealized delocalized phase
the only relevant scale is the time to cross the system , or equivalently the corresponding thouless energy @xmath497 which was treated as infinite .
we may thus expect that the statistics described in the foregoing section correspond to asymptotic situations for which the dimensionless conductance is either zero ( localized phase ) or infinite ( delocalized ) phase .
the question then arises how the statistical properties do change for finite conductance .
of particular importance are the statistical properties when approaching the ld transition .
which physical scales determine the statistical properties ? a related question concerns the change of the statistical properties with increasing system size .
looking at the conductance formulas by thouless and bttiker , eqs .
( [ 2.19],[3.39 ] ) , it is tempting to assume that the conductance itself is the ( perhaps only ) relevant quantity to distinguish localized and delocalized phases and to characterize the ld transition . 1 .
the localized phase with @xmath498 corresponds to _ only closed _
transport modes ( channels ) ( bttiker formula with @xmath499 ) and the existence of a finite localization length @xmath500 , @xmath501 .
the delocalized phase with @xmath502 corresponds to many _ open channels _ ( bttiker formula with @xmath503 modes of @xmath504 ) and a thouless energy much larger than the level spacing , @xmath505 .
the ld transition is expected to occur when scales match , i.e. @xmath506 according to the bttiker formula only a number of @xmath507 modes are open and the localization length diverges .
this divergence turns out to be algebraically on approaching the transition point , @xmath508 where @xmath118 measures the deviation of a tuning parameter ( e.g. fermi energy ) from its _ critical value _ at the ld transition .
the exponent @xmath228 is called _ critical exponent _ of the localization length . by
this reasoning a relevant length scale @xmath0 can be defined via the conductance : @xmath0 is that fictitious system size for which @xmath509 in the localized phase @xmath0 can therefore be identified with the localization length , while in the delocalized phase @xmath510 . as our considerations call for a description of the ld transition as a critical phenomenon we briefly review some aspects of the latter .
a thermodynamic system is described in terms of the coordinates of its equilibrium states manifold @xmath511 ( e.g. volume @xmath512 , particle number @xmath121 and temperature @xmath341 ) .
any thermodynamic quantity of interest can be calculated from a thermodynamic potential ( e.g. the free energy @xmath513 ) which is an analytic function of these coordinates ( e.g. the specific heat @xmath514 ) .
generally , one can change the parameterization of @xmath511 ( e.g. volume @xmath512 @xmath515 pressure @xmath516 ) to describe the same equilibrium state at least locally . in the thermodynamic limit where @xmath121 and @xmath512 tend to infinity simultaneously leaving the density @xmath517 constant
, it may happen that the thermodynamic potential is no longer an analytic function of its coordinates .
in such a situation the state of the system is not uniquely described by only one set of coordinates manifesting in phases . when crossing the coexistence curve the thermodynamic potential jumps .
such a process is called a first - order phase transition .
the jump is quantitatively described by the value of the latent heat .
assume that the coexistence curve ends in some point of the @xmath518 diagram which is called the critical point . here
the latent heat vanishes , the thermodynamic potential is continuous and the difference of densities between the phases vanishes . yet , at the critical point , the thermodynamic potential is not analytic since its second derivatives are discontinuous .
it is then helpful to look for a quantity which serves as an _ order parameter _ of the transition .
it should be finite in one phase and vanishes in the other phase .
a more restrictive requirement is that an order parameter should correspond to a local quantity @xmath519 the average value of which @xmath520 vanishes with some power law at the critical point ( e.g. the difference of the corresponding densities in the transition ) , @xmath521 the exponent @xmath12 is called the critical exponent of the order parameter and @xmath522 .
it is a necessary condition for such power law behavior that the phase transition is not of first order since a power law is the paradigm of a physical law with no scales involved . in a first order transition , however , the jump in the thermodynamic potential sets a physical energy scale .
a defining characteristic of a critical phenomenon is that there is also no relevant length scale at the critical point .
thus , the _ correlation length _ @xmath0 of the order parameter , @xmath523 , defined by the statistical properties of the local order parameter field @xmath524 has to diverge at @xmath525 with a power law with some critical exponent @xmath228 , @xmath526 the crucial assumption in any scaling approach to critical phenomena is that the critical exponents @xmath527 , @xmath228 have their origin in the divergence of a single relevant length scale , the correlation length @xmath0 . at the critical point where @xmath0 diverges
the correlation function obeys a power law @xmath528 using @xmath0 as a cutoff when calculating the global susceptibility we find a _ scaling relation _
@xmath529 let us now return to finite systems of linear size @xmath6 and assume that both the distance @xmath298 and the finite system size @xmath6 are within the regime without length scales and that the function obeys a power law with respect to both lengths @xmath530 the new exponent @xmath531 describes the system length dependence and @xmath532 describes the distance dependence . to consider this case turns out to be relevant in the ld transition problem as will be discussed in more detail in sect .
[ submul ] . a similar reasoning as above yields now @xmath533 notice that the quantity @xmath534 is not determined by @xmath535 and @xmath228 alone but requires the additional knowledge of the unusual exponent @xmath531 .
far away from the critical regime there may still exist a _ scaling variable _
@xmath536 which replaces the role of @xmath341 in the following sense .
the state of the system is described for a finite size @xmath6 already by the quantity @xmath536 and the flow with increasing system size @xmath6 is determined by a single function of this variable , the so called @xmath12-function @xmath537 which should be a function of @xmath538 alone .
provided the @xmath12-function is smooth , the flow is determined by the solution of the differential equation , eq .
( [ 6.13 ] ) , frequently called a _ renormalization group equation_. the regime where @xmath539 can be linearized around a fixed point @xmath540 of the flow ( @xmath541 ) , @xmath542 with @xmath543 being the slope of the @xmath12-function at @xmath540 , is called the critical regime .
it is in this regime where power law scaling can be expected .
if we start with a system of size @xmath544 , @xmath545 chosen close to @xmath540 , and turn on the renormalization flow until the system reaches a size @xmath0 determined by the range of validity of eq .
( [ 6.15 ] ) ( which is here by definition the correlation length ) we get @xmath546 assuming that the critical regime is narrow , we can expand @xmath547 and @xmath548 around @xmath540 and get for the correlation length @xmath549 the width of the critical regime is determined by @xmath118 such that @xmath550 thus , the exponent @xmath228 is given by the inverse slope of the @xmath12-function at the fixed point @xmath551 the behavior of the scaling variable close to the transition is therefore of the form @xmath552 the coefficient @xmath280 can be brought to the form @xmath553 where @xmath554 is of order @xmath555 .
note , that , in general , the coefficient @xmath554 can differ on both sides of the transition , while the critical exponent @xmath228 is the same .
it is crucial that any other scaling variable @xmath556 which is a smooth function of the previous one leads to the same value of @xmath228 , although the overall form of its @xmath12-function can differ .
this reflects the fact that critical exponents are universal numbers characterizing a whole class of systems . such systems may differ by microscopic details , however the critical behavior is characterized by universal critical exponents of a few relevant quantities . in sect .
[ subphy ] we have learned that knowing the green s function is , in principle , enough to describe the transport properties including the ld transition .
a natural guess for the order parameter field is the ldos since it typically vanishes in the localized phase and is finite in the delocalized phase .
consequently , the average dos ( which is determined by the one - particle green s function ) appears as the natural ( formal ) order parameter . since the average dos does not show the ld transition but is a smooth function of the energy ( which , in the problem of the ld transition , is the analog of temperature in equilibrium phase transitions ) the _ critical exponent _
@xmath12 of this formal order parameter vanishes @xmath557 the ld transition is not manifested in the one - particle green s function , but in transport related quantities such as the two - particle green s function .
we like to mention that this fact rules out the possibility of a straightforward mean - field theory for the ld transition .
furthermore , as the scaling relation eq .
( [ 6.10 ] ) tells us the correlation exponent of the local susceptibility ( a two - particle green s function ) vanishes , @xmath558 this result seems to rule out a power law behavior with finite power for the two particle green s function ( say the density correlator ) at criticality . in spite of
that a finite power has been observed @xcite for the density correlator at a ld transition .
however , it turned out that it corresponds to a regime where the correlation function shows a power law ( with unusual exponent @xmath531 ) also with respect to the system size @xmath6 .
a further unusual feature of the ld transition , which at first glance seems to be independent of the previous ones , was first pointed out by altshuler _
@xcite when reexamining the phenomenological scaling theory of abrahams et al .
@xcite .
this theory picks up the ideas that we introduced in the beginning of this section .
the conductance @xmath559 of a cube with linear dimension @xmath6 is treated as a scaling variable , i.e. @xmath113 obeys a differential equation in terms of a @xmath12-function which , for a certain _ universality class _ , is a unique function of @xmath560 .
the fact that one scaling variable is enough to describe the transition is referred to as _ one - parameter _ scaling theory .
that such treatment makes sense can be seen when calculating the @xmath12-function for extreme situations . in a good metal with a conductivity that is a material constant , ohm s law , eq .
( [ 2.16 ] ) , yields @xmath561 while in the exponential localized phase one obtains @xmath562 where @xmath563 is a constant for each dimension @xmath138 . assuming @xmath564 to be a smooth function one may interpolate between the asymptotic branches ( see fig .
20 ) . from this picture one concludes : in 1d all states are localized , in 3d a generic ld transition takes place and for 2d the situation depends on the sign of weak - localization effects for large @xmath113 .
if these tend to localize , the @xmath12-function stays negative and all states will localize in the large @xmath6 limit . for
weak anti - localization ( found in 2d systems with spin - orbit interaction @xcite ) a ld transition should occur .
it has been observed by numerical model calculations ( see e.g. @xcite ) . in an experiment by kravchenko _
@xcite on a strongly interacting 2d electron system a ld transition was observed . at present
it is not clear if it can be due to strong spin - orbit scattering .
the quantum hall effect , as mentioned in the introduction , is exceptional in this picture .
this is due to the absence of a metallic phase .
the corresponding @xmath122-function terminates from below at the critical point . although the one - parameter scaling theory by abrahams _ et al . _
@xcite is simple , predictive and qualitatively correct , it has an obvious shortcoming as was pointed out by altshuler _
the conductance of a mesoscopic system strongly depends on the individual system properties ( e.g. on the given disorder potential ) . as a result , an ensemble of different systems exhibits conductance fluctuations which can become so strong that the corresponding distribution can not be characterized by the mean value alone
the latter as well as higher moments , can be drastically influenced by the far tails of the distribution .
it can thus happen that the mean value of the conductance @xmath565 can _ not _ serve as a scaling variable .
instead , an appropriate scaling approach to the ld transition has to be set up in terms of the complete distribution function of the conductance .
however , as pointed out by shapiro @xcite it may be possible to apply the renormalization group approach to certain parameters occurring in that distribution function . if an appropriate parameter of the distribution exists which is less influenced by the far tails and which can serve to define a _ typical _ conductance , @xmath566 , it may be possible to find a flow equation in terms of a @xmath122-function for this typical conductance . in this sense
the theory by abrahams _
_ can still be complete .
for this scenario to be realistic , at least in the vicinity of the transition point , only _
one _ length scale @xmath0 should diverge at criticality .
then the ld transition obeys one - parameter scaling and the exponent @xmath228 of the correlation length @xmath0 is given by the @xmath12-function of @xmath566 .
it is worth mentioning that the conductivity , defined by the conductance as @xmath567 , is often referred to as an order parameter of the transition .
it vanishes at the transition point in @xmath568 and @xmath569 defines an critical exponent @xmath258 .
however , this notion of conductivity gives no further information than that already contained in the conductance .
only for @xmath568 the conductivity does vanish at the transition point and the exponent @xmath258 is trivially related to @xmath228 by @xmath570 since @xmath571 .
this relation is sometimes referred to as a scaling relation between critical exponents @xmath258 and @xmath228 suggesting that this replaces the scaling relation eq .
( [ 6.10 ] ) in common critical phenomena . to the author
this interpretation is misleading .
firstly , the conductivity is defined via the conductance , which is the scaling variable of the problem . for any critical phenomenon with a scaling variable @xmath536
one can define similar _ order parameters _ by simply multiplying the scaling variable by a negative power of @xmath6 .
it is obvious that this does not give further insight .
secondly , for ld transitions occurring in @xmath572 the conductivity stays finite at the transition , which invalidates the notion of an order parameter for @xmath2 .
thirdly , the conductivity is not related to a local field of the problem . in sect.[statra
] we will see that the choice of an appropriate order parameter and the occurrence of broad distributions of physical quantities are intimately interrelated .
in this section we will exploit our experience with the notion of typical conductance in 1d and make an extension to higher dimensions following a work by shapiro @xcite .
the extension relies on uncontrolled approximations and may thus be only of pedagogical interest .
however , the results are surprisingly far - reaching and consistent with commonly accepted results about ld transitions .
recall that the typical transmission probability in 1d followed a simple multiplicative composition law which under a scale transformation @xmath573 leads to the scaling law in 1d @xmath574 we would like to generalize this law to higher dimensions .
the multiplicative nature of the composition law is a consequence of the multiplicative nature of transfer matrices when put in series .
if the cross section of the conductor is finite the transfer matrix acquires more modes .
the parallel composition law for classical conductance treats all modes equal and so do we . to this end , we first have to define an appropriate conductance for a quasi-1d system for which the parallel composition makes sense .
the notion of intrinsic resistance @xmath291 and conductance @xmath575 , respectively , is helpful . in 1d
we have @xmath576 which can take all positive values . by eq .
( [ 6.24 ] ) the series composition law reads @xmath577 we adopt this law for any series composition in quasi-1d when we put @xmath578 blocks of size @xmath131 in series . that this makes sense
can be seen by considering the cases with @xmath579 and @xmath580 , respectively . for weak resistivity of each block ,
the resistances behave additive like in classical resistor networks .
however , for strong resistivity of each block we recover the multiplication law leading to strong localization which is consistent with the finding of strong localization in quasi-1d . once we have put the @xmath578 blocks in series , forming a quasi-1d conductor of cross section @xmath581 and length @xmath582 we put @xmath583 of such conductors in parallel , resulting in a cubic conductor of total volume @xmath584 . for this composition
we assume the classical parallel composition law to be valid , @xmath585 the application of parallel composition to arbitrary scaling factors can become quite inadequate
. however , we will now restrict to very small scaling factors @xmath586 , @xmath587 since we are only interested in the change of @xmath291 under a small increase of system size . by this
we can easily derive the corresponding @xmath12-function for @xmath291 , @xmath588 for easier comparison with the scaling theory of abrahams _ et al .
we reformulate the result in terms of the intrinsic conductance @xmath589 one sees immediately that this @xmath122-function has the same asymptotic behavior as in the scaling theory by abrahams _
_ , i.e. for @xmath590 , @xmath591 , and for @xmath592 , @xmath593 .
furthermore , it has no zero for @xmath594 ( see fig .
21 ) and for @xmath595 with @xmath596 one recovers the results @xmath597 of a much more involved field theoretic calculation @xcite which is expected to be exact in the limit @xmath598 .
in addition , one finds for @xmath599 a value of @xmath600 which is close to known results ( @xmath601 @xcite ) of numerical calculations for 3d models .
we are now going to address the flow of distribution functions . as a guiding example we consider the 1d conductor . here
localization on the scale of the mean free path @xmath9 dominates and we can expect to get an almost complete description . the composition law for transmission , eq .
( [ 3.54 ] ) , translates to that of intrinsic resistance , @xmath291 , @xmath602 we can use this to derive an evolution equation for the distribution function @xmath603 . since the procedure is of general importance we describe the essential steps ( cf .
@xcite ) : a stochastic quantity @xmath604 depending on time @xmath297 may increase in time @xmath605 by @xmath606 . if the stochastic process is markovian , i.e. it is determined by transition probabilities from one time step to the next and is not influenced by the history of the process the essential information is contained in the moments , @xmath607 . if , in addition , only the first two moments scale linear with @xmath605 ( while all higher moments scale with higher powers ) a fokker - planck equation ( fpe ) results @xmath608 where @xmath609 and @xmath610 ( for @xmath611 ) are called _ drift _ and _ diffusion _ function , respectively . in our problem ,
the stochastic quantity is @xmath612 and by eq .
( [ 6.29 ] ) we obtain @xmath613 the average is to be taken over random phases @xmath26 and over the strip - resistance @xmath614 which introduces the mean free path @xmath615 .
this definition is often referred to as _ local weak scattering condition _ and is the basis of a fpe approach to the statistics .
equation ( [ 6.31 ] ) shows that only the first and second moments are linear in @xmath310 while higher moments scale with higher powers .
thus , a fpe can be derived for @xmath616 with drift and diffusion given by @xmath617 the fpe can be brought to the form @xcite @xmath618 this distribution depends on only one parameter , the mean free path @xmath9 .
two limiting cases can be considered now : ( i ) @xmath619 and ( ii ) @xmath620 . for case ( i )
the solution is an exponential distribution @xmath621 with average value @xmath622 .
this solution corresponds to system sizes @xmath623 . for case ( ii )
the solution is a log - normal distribution @xmath624 with a typical value @xmath625 .
this solution corresponds to localization with @xmath626 , and a typical localization length of @xmath627 , consistent with our previous finding ( see eq .
( [ 3.58 ] ) ) .
interesting is the fact that the second parameter of the log - normal distribution , the log - variance @xmath628 , is simply related to the log - average value @xmath629 the log - normal form of the distribution for @xmath630 is also consistent with the general statement , eq . ( [ 3.67 ] ) , that the localization length ( @xmath631 ) corresponds to normally distributed quantities that become self averaging in the limit @xmath632 .
the distribution @xmath633 is broad even for @xmath634 ( since an exponential tail already introduces a strong growth of moments @xmath635 ) and causes large fluctuations for @xmath291 .
the observation of broad distributions is central to mesoscopic systems and therefore we make some general statements about them . following shapiro @xcite we formulate the concept of one - parameter scaling for broad distributions as follows .
a distribution of a physical quantity @xmath636 that depends on system size @xmath6 and a set of initial parameters @xmath637 , @xmath638 , obeys one - parameter scaling if ( for large @xmath6 ) it is approximately a function of only @xmath636 and one scale dependent parameter @xmath639 , @xmath640 and the flow of the parameter @xmath639 is determined by a @xmath122-function @xmath641 that means that the whole set of initial parameters @xmath637 , needed to specify the initial small - scale distribution , disappear into the single parameter @xmath642 .
the exact distribution _ does _
contain all the information about the initial set of parameters ; however , the information can be accumulated in the very far tails of the distribution . by _
very far tail _ of a distribution we denote that range of values @xmath643 , @xmath644 that is not statistically significant to some small but fixed probability @xmath212 @xmath645 by eq .
( [ 6.37 ] ) it is meant that the approximation for the main part of the distribution is exact up to corrections of order @xmath212 . for sufficiently large @xmath6
it can be taken to be arbitrarily small .
still , the moments of the distribution might be dominated by the very far tails of the exact distribution .
therefore , one can not expect that in general the parameter @xmath639 is simply related to moments of the distribution . as
a consequence of one - parameter scaling there should exist a universal limiting distribution at a critical point @xmath646 which is fixed by the critical value @xmath647 of the scaling variable .
a distribution is said to be broad if its moments @xmath648 grow like @xmath649 with @xmath650 increasing stronger than linear with @xmath62 , or if its moments might even diverge .
this notion refers to the fact that the corresponding distribution of the sum of @xmath121 independent realizations of the variable @xmath636 will still deviate substantially from a gaussian , even for very large @xmath121 . according to this definition of _ broadness _ a weak case of a broad distribution
is realized by the exponential distribution ( @xmath651 ) @xmath652 for which the moments grow as @xmath653 .
the prototype of a broad distribution is the log - normal distribution @xmath654 the moments of which grow as @xmath655 .
it represents the generic distribution for random multiplicative variables .
after these general remarks we would like to calculate the conductance distribution in higher dimensions including the ld transition .
however , this is difficult .
an interesting attack has been undertaken by shapiro @xcite .
he used the classical parallel composition law to derive fokker - planck equations for the conductance distribution in higher dimensions .
however , the uncontrolled approximations involved overestimate fluctuations .
analytical approaches are usually restricted to dimension @xmath595 where the critical conductance is of order @xmath656 and perturbation theory together with renormalization techniques can be used .
we do not enter this technically involved subject and we will briefly review the main conclusions in sects .
[ starea],[statra ] . here , we will turn to local quantities and address the question of power law scaling relevant to the delocalized phase and the ld transition . in the delocalized phase wavefunctions
are extended all over the system .
the correlation length @xmath0 in this phase can be estimated from @xmath657 taking the metallic limit , eq .
( [ dru7 ] ) , one finds that ( @xmath568 ) @xmath658 is a truly microscopic scale .
thus , one can expect a power law scaling of moments of the box probability @xmath659 for @xmath660 . in the critical regime of the ld transition
the correlation length @xmath0 is larger than system size l and one can thus also expect a power law scaling of moments of the box probability @xmath659 for @xmath661 ( where @xmath70 stands for microscopic scales ) .
therefore , we will now consider distribution functions the moments of which display power law scaling . to simplify notations
we consider a random variable @xmath662 $ ] and a scale @xmath663 $ ] with distribution @xmath664 .
we assume that for sufficiently small @xmath665 the moments obey power law scaling @xmath666 reflecting that scales in the system are separated . here
@xmath667 and @xmath668 are treated as functions of arbitrary real values @xmath669 .
the power law scaling condition is very restrictive and allows for a number of conclusions : 1 .
normalization requires @xmath670 2 .
since @xmath668 is independent of @xmath7 one can calculate its derivatives and send @xmath671 . by this one
can conclude two strong inequalities @xmath672 where the prime means differentiation with respect to @xmath669 .
3 . introducing the variable @xmath673 one can see that the moments are the laplace transform of the function @xmath674 , @xmath675 thus , one may reconstruct the distribution from the moments by inverse laplace transformation ( if the latter can be carried out ) .
the case of strictly linear @xmath676 ( @xmath677 is not interesting ) allows for a _
global _ rescaling of the distribution function by its first moment . @xmath678 and
the coefficients @xmath667 are determined by the scale independent distribution @xmath679 @xmath680 a linear behavior of @xmath681 seems to be natural for power law scaling ; all moments scale as powers of the first and only one critical exponent describes the scaling of all of them , including that of the distribution function .
the rescaled distribution function is now scale independent and can take any form , e.g. it can be a broad distribution in the sense defined above .
examples are the rayleigh and porter - thomas distributions found for the box - probability in the ideal delocalized phase ( see eqs .
( [ 5.25],[5.26 ] ) ) .
we refer to the situation of linear @xmath682 as the _ gap scaling _ situation .
the linear case is not the most general one . for finite curvature in @xmath668 , @xmath683 ,
the coefficients @xmath667 can be treated as constant since the essential @xmath669 dependence that dominates for small @xmath7 comes from @xmath682 .
we will see in the following that the distribution function @xmath684 can be essentially reconstructed from @xmath668 by a legendre transformation and it is an almost log - normal distribution with special features .
one can write @xmath685 where the normalization @xmath686 has only a weak @xmath7 dependence ( e.g. @xmath687 ) . the essential information is now contained in the function @xmath688 which is the legendre transform of @xmath682 , @xmath689 before we outline a derivation of the statement we make further comments . 6 .
in contrast to the gap scaling situation with arbitrary form of the scale independent distribution we have now the situation that the form of the distribution is fixed by a spectrum of scaling exponents . this situation has been phrased ( in a bit more restricted context ) _
multifractal scaling _ situation ( for a review see @xcite ) .
the spectrum has several analytical properties following from the legendre transformation .
it has a unique maximum at @xmath690 and a slope of @xmath15 at @xmath691 .
the derivatives fulfill @xmath692 close to the typical value @xmath693 the function @xmath688 is well approximated by a parabola and @xmath694 is approximated by a log - normal distribution @xmath695 where @xmath696 is a positive constant .
the derivation of eq.([6.51 ] ) relies on the fact that a finite curvature of @xmath682 allows for a stationary point evaluation of the integral in eq .
( [ 6.45 ] ) for each moment @xmath697 in the limit @xmath671 .
writing @xmath698 and transforming to the variable @xmath174 , with the distribution @xmath699 , yields the integral relation @xmath700 with @xmath701 the function @xmath702 can have a maximum with respect to @xmath174 ( for @xmath703 ) that is only weakly dependent on @xmath7 .
it yields a sharp maximum in the integrand ( due to the large pre - factor -@xmath704 ) .
then , the integrand is determined up to @xmath705 by the value of the integrand at this maximum and one finds that @xmath668 is the legendre transform of @xmath706 .
consequently , @xmath706 must have a very weak @xmath7 dependence ( e.g. @xmath707 ) . in case that @xmath702 has no maximum with respect to @xmath174 that is only weakly dependent on @xmath7
the corresponding @xmath668 will not have finite curvature , but the gap scaling situation applies . the phrase _
multifractality _ has its origin in the interpretation of @xmath708 as a local measure defined for a geometrical object embedded in a @xmath138-dimensional cube of linear size @xmath6 .
the scale @xmath7 is given by the ratio @xmath709 where @xmath489 is the linear size of a box on which the measure is calculated .
due to normalization @xmath710 the average of @xmath708 is the inverse of the number of non - empty boxes and thus yields , for @xmath703 , the fractal dimension @xmath711 of the geometrical object @xmath712 the fact that higher moments @xmath697 may scale not simply as @xmath713 but as @xmath714 where @xmath715 defines a set of @xmath669-dependent _ generalized dimension _
@xmath716 by @xmath717 , has led to the notion of multifractality .
especially , one has @xmath718 , @xmath719 .
the legendre transform of @xmath715 then defines the so - called multifractal @xmath720-spectrum .
it can be interpreted as the fractal dimension of that subset of the object where the measure scales locally as @xmath721 ( see @xcite ) .
the general analytic properties of multifractal exponents are visualized in fig .
22 we see that , for measures @xmath708 , we have the following identification @xmath722 under quite general conditions , @xmath688 ( or @xmath720 ) is also bounded from below by @xmath723 ( or @xmath719 ) and terminates with infinite slope at the endpoints of the finite interval @xmath724 $ ] ( or @xmath725 $ ] ) . in many cases
the parabolic approximation of eq .
( [ 6.55 ] ) extends to values @xmath726 and the constraint on the slope of @xmath720 at @xmath727 allows to write a parabolic approximation ( pa ) for @xmath720 that depends only on one - parameter ( besides @xmath711 ) @xmath728 @xmath729 for example , the box - probability @xmath730 considered in the ld - transition problem ( eq .
( [ 5.13b ] ) ) is of this type of variable , since the squared amplitude of the wave function defines a local measure .
the fractal dimension @xmath731 is always trivial . in the idealized delocalized phase we found the gap scaling class and @xmath732 . the interesting information is contained in the shape of the corresponding scale independent distribution . for the ld transition ,
however , it is known from a calculation by wegner@xcite for @xmath733 dimensions that the multifractal scaling class applies and @xmath734 .
the pa he calculated yields @xmath735 in summary , we found that the restriction to power - law scaling leads to two _ distinct _ classes of distribution functions .
the gap scaling class is characterized by a single scaling exponent and a related scale independent distribution of arbitrary shape .
the multifractal scaling class is characterized by a spectrum of scaling exponents and a shape of the distribution close to log - normal .
the distribution of local box - probability in the ideal delocalized phase falls into the gap scaling class .
the possibility of the multifractal scaling class is interesting for the ld transition point , because it opens a possibility to find an appropriate order parameter in terms of the ldos .
if , as it can happen in the multifractal scaling class , average and typical dos scale differently with respect to system size , the inability of the average dos to work as an order parameter must not be shared by the typical dos . in sect .
[ sublocden ] we will see that , indeed , this is the case .
in this section we will review known facts about the statistics of conductance , energy levels and local quantities for the delocalized phase with large typical conductance @xmath736 and for the localized phase with finite typical localization length @xmath737 .
the onset of broad distributions with log - normal tails in the delocalized phase is most interesting here . in 2d
it leads to multifractality of the ldos . in order to interpret this phenomenon
we consider the conductance distribution for quasi-1d systems for which the number of channels is large , but the width is still of the order of a microscopic scale , e.g. the mean free path . under these conditions a fokker - planck approach works that reproduces the ucf phenomenon , but
the long tails are absent .
we come to the conclusion that the log - normal tails are a precursor of the ld transition and do only occur if the relevant scaling variable is not infinite . to study more realistic quasi-1d systems the method of finite size scaling ( fss ) will be outlined . in this method
a scaling variable is defined which is close to the definition of typical conductance . in the delocalized phase of a finite mesoscopic conductor where the average conductance is large
@xmath738 the bttiker formula tells that many ( @xmath739 ) open modes are occupied .
thus , the conductance is a sum of many random numbers @xmath740 .
naive application of the central limit theorem suggests that the conductance distribution will be a gaussian centered around @xmath741 with a relative deviation @xmath742 where @xmath743 .
however , this reasoning ignores the strong correlations between the transmission eigenvalues @xmath326 .
this correlation leads to the occurrence of universal conductance fluctuations ( ucf ) ( cf .
[ subuni ] ) , i.e. the variance is @xmath744 and does neither depend on the actual value of @xmath745 nor on the system size @xmath6 .
taking ucf into account the conductance distribution seems to follow one - parameter scaling : it is gaussian with scale dependent average value ( which is also typical here ) and a universal variance .
@xmath746 this was the common belief until the work of altshuler _ et al .
_ @xcite . in this work
it was shown by a combination of diagrammatic perturbation theory and renormalization group techniques that the conductance distribution deviates from a gaussian for @xmath747 .
for these tails the distribution crosses over to a log - normal form ( see fig .
this was concluded from the calculation of high moments which showed the characteristic dependence @xmath748 here @xmath749 indicating that a finite value of conductance is responsible for the effect .
the calculations were carried out for 2d ( with localization lengths much larger than system size ) and @xmath595 ( in the delocalized phase ) . in the work by altshuler _
_ it was conjectured that their finding will penetrate to the ld transition point , i.e. that the conductance distribution will become very broad .
the observations of these log - normal tails led to questioning the whole idea of one - parameter scaling .
later , shapiro and coworkers @xcite showed that these tails are not in conflict with one - parameter scaling theory as defined in secs .
[ subscadis ] . on the basis of the @xmath733 results they concluded that a critical conductance distribution will appear with long power - law tails .
such tails lead to divergence of low moments and exclude any one - parameter scaling theory in terms of moments of the conductance .
later works @xcite confirmed the calculation of altshuler _ et al . _ and showed its principle validity for 3d as well .
the 1d and quasi-1d cases need special attention since strong localization dominates . in the localized phase the bttiker formula tells that the conductance is dictated by the smallest value @xmath750 corresponding to eigenvalues of @xmath751 ( cf .
( [ 3.51 ] ) ) .
this value is known to be gaussian distributed and thus it is no surprise that the conductance distribution , for @xmath752 , @xmath753 , is of the log - normal form .
this reflects the gaussian distribution of inverse localization lengths .
the typical localization length ( @xmath754 ) defines the typical conductance as the geometric mean @xmath755 . here , in principle a second parameter besides @xmath756 might occur in the distribution , namely the variance of @xmath750 .
however , as happened already in the 1d case ( cf .
[ subscadis ] ) , all known results indicate that this variance is simply related to the average value ( cf .
e.g. @xcite ) and thus only one - parameter seems to describe the distribution accurately , @xmath757 with @xmath758 the behavior of the conductance distribution close to the ld transition will be discussed in sect .
[ subcond ] .
concerning the level statistics we restrict our discussion to the two - level correlation function @xmath424 .
the energy separation @xmath759 can be associated with an inverse time scale .
the behavior of @xmath424 is related to the question how a wave packet evolves in time @xmath760 @xcite . for times larger than the diffusion time
@xmath106 the wave packet explores the whole system several times and it is not surprising that here the standard random matrix theory results are approximately valid . in the standard random matrix theory for the idealized delocalized phase
@xmath424 only depends on the level distance @xmath258 measured in average level spacing units . a finite , but large ,
conductance is related to the diffusion time via the thouless energy @xmath761 , the relevant transport energy scale .
consequently , the dimensionless conductance @xmath113 ( here @xmath113 can be taken as the average value ) sets a scale on the @xmath258-line .
indeed , one finds different level correlation functions for @xmath762 and for @xmath763 . in the first case ,
as expected , the function @xmath424 is approximately given by the standard random matrix theory result and only small corrections ( proportional to powers of @xmath764 ) occur . for example
, the result for systems with broken time reversal symmetry reads @xcite @xmath765 ( @xmath563 a constant depending only on dimension and symmetry ) .
the case @xmath763 has been investigated by altshuler and shklovskii @xcite and the finding is ( for @xmath258 not exceeding the inverse of the microscopic mean free time @xmath118 and possible oscillations on the scale @xmath766 are ignored ) @xmath767 where @xmath768 is a numerical coefficient that depends only on d and on the symmetry class .
the result can be understood by supposing diffusive motion of wave packets in the system . in 2d there
exists a broad cross over regime between the two asymptotic regimes of eqs .
( [ 7.4],[7.5 ] ) @xcite .
this can be understood as follows .
the 2d metal resembles a critical system in as much as the corresponding @xmath122-function is close to @xmath60 .
the strong localization that will finally dominate comes along with very large localization lengths .
it has to be expected that the level statistics will change on approaching the ld transition . ] .
a similar behavior can be expected to occur in the 2d case .
the level - correlation function for the localized phase with vanishing localization length is given by the poisson statistics , eq .
( [ 5.10 ] ) , i.e. no - correlation for @xmath476 . to take a finite localization length into account which is , however , much smaller than the system size @xmath6 , @xmath752
, it is helpful to think of dividing the system into subsystems of volume @xmath769 . the level - correlation function r(s;l ) will then obey a scaling form @xcite @xmath770 in the thermodynamic limit the level correlations vanish , @xmath771 , in accordance with the poisson statistics being the limiting case .
the scaling form is in agreement with the one - parameter scaling theory since the variable @xmath772 is related to the typical conductance in the localized phase , @xmath773 .
the form of the scaling function @xmath774 can be obtained for @xmath775 being large or small compared to @xmath15 . for @xmath776 , the corresponding wave - packet dynamics probes only the region within one localization volume and is , thus , diffusive .
consequently one recovers the altshuler - shklovskii behavior @xcite @xmath777 for @xmath778 but @xmath779 regions much larger than one localization volume are probed and the only correlation between levels comes from weak couplings between such volumes @xcite resulting in a logarithmic dependence on @xmath775 @xmath780 we now turn to the statistics of local quantities .
again , for a good metal with @xmath736 the distribution of local quantities is essentially given by the standard random matrix theory ( cf .
( [ 5.25],[5.26 ] ) ) .
however , the long tails are different and the finite value of @xmath113 becomes important .
we concentrate on the box - probability @xmath730 in the following .
this generalizes to the reduced ldos @xmath781 .
we introduce the normalized quantity @xmath782 and consider the case @xmath736 .
the scale @xmath70 stands for microscopic box size .
the threshold for the standard random matrix result turns out to be @xmath783 . for @xmath784
the rayleigh and porter - thomas distributions are valid up to small corrections of @xmath785 ( cf .
@xcite ) . for @xmath786
a number of authors @xcite have found the onset of broad distributions which in 2d and 3d develop long tails of the form @xmath787 here @xmath563 depends on the dimensionality @xmath138 and on the disorder strength and @xmath788 . for @xmath789 values of @xmath245 and @xmath790 have been predicted .
the value @xmath791 is a typical value serving as reference point .
the precise expressions are currently under debate ( because the authors of @xcite use different approximative calculation schemes ) .
however , what seems to be clear is the occurrence of long - tails of the general form eq .
( [ 7.10 ] ) . for the 2d case
there is also agreement that @xmath792 and also @xmath793 . due to this scaling behavior
the result can ( for @xmath794 ) be identified with a multifractal distribution within the parabolic approximation ( pa , cf .
( [ parabol ] ) ) @xmath795 where the essential parameter @xmath796 is here ( d=2 ) proportional to the inverse of the conductance @xmath797 thus , states in a good 2d metal ( with the localization length being much larger than the system size ) are multifractally distributed . since the metallic situation in 2d resembles a critical regime of a ld transition
the finding is , after wegner s pioneering work @xcite and the work by altshuler _
@xcite a further analytical hint at the multifractality of critical eigenstates .
the case of 1d and quasi-1d conductors need special attention . in 1d
the localization length @xmath144 is of the order of the mean free path @xmath798 and thus there is no room for diffusive behavior . for the quasi-1d case one
has to be a bit more precise in fixing the sizes .
the length @xmath6 is treated variable and the cross - section @xmath102 is kept fixed . however , the dimensionality is in general essential for the determination of physical quantities such as the localization length .
a strong simplification arises for the case when @xmath799 but a huge number @xmath800 of channels is occupied . under these conditions
the dimensionality becomes inessential and the only kinematic quantity besides @xmath6 is the channel number @xmath286 .
we refer to this case as the _ idealized _ quasi-1d case . for the idealized quasi-1d system
one can show that @xmath801 with @xmath280 a numerical constant of order @xmath15 .
we have anticipated this result already in sect .
[ subloc ] and will sketch a more reliable derivation in the following subsection .
the fact that one can reach a metallic regime by @xmath802 makes the idealized quasi-1d system more interesting than strictly 1d . on the other hand , the idealized quasi-1d system is still tractable by analytical methods ( cf .
e.g. @xcite ) . in metallic idealized quasi-1d systems
one has an average conductance ( which is also typical ) @xmath803 the distribution of @xmath297 is , for @xmath784 , again given by the standard random matrix theory with small corrections of @xmath804 ( cf .
. however , for @xmath786 the situation is different as compared to the 2d and 3d case .
one finds no log - normal tails but stretched exponential tails ( at least when the box is shrinking to a point ) @xmath805 note , that @xmath806 is the quasi-1d conductance . in the localized regime of idealized quasi-1d systems
, @xmath807 where the typical conductance is @xmath808 , one finds a similar behavior for @xmath809 @xmath810 note , that @xmath772 is related to the typical conductance here .
both equations , ( [ 7.15],[7.17 ] ) , have @xmath772 as the reference point . in order to have @xmath809 one
has to have @xmath811 .
thus , the stretched exponential tail does occur when the box - size is much less the localization length . in the case where the box - size becomes much larger than the localization length
the previously obtained results of the idealized localized phase apply ( cf .
[ sublocpha ] ) .
the tails of the distribution @xmath812 that have been observed in idealized quasi-1d ( stretched exponential ) , in 2d ( log - normal ) and in 3d ( log - cube ) have all been interpreted in terms of _ anomalously localized states _ ( or _ pre - localized states _ ) @xcite .
the notion refers to anomalously rare states with sharp peak in the amplitude on top of an extended backround ( in 3d delocalized phases ) or localized states with anomalously short localization radius ( in @xmath594 ) .
the question arises what the tails might say in view of the ld transition .
the 2d case seems to be clear . here
the log - normal tails correspond to multifractal scaling to be expected once the correlation length is much larger than system size .
this interpretation is in accordance with the finding of log - normal tails in @xmath595 .
the distributions become multifractal when the conductance is close to its critical value . in 2d
the correlation length is the localization length which is much larger than the system size in the @xmath736 regime , and thus the states are _ critical _ in the sense that the system size is less the correlation length . in 3d with @xmath736 the correlation length is ( see eq .
( [ 6.43 ] ) ) microscopic and there is no reason to expect multifractal scaling . in quasi-1d with @xmath736
the quasi-1d localization length is much larger than the system length @xmath6 , however due to the strong difference between @xmath799 and @xmath6 , this length can not be considered as the correlation length . in strict 1d the correlation length is the 1d localization length which is again microscopic and no _ critical _ behavior of eigenstates can be expected on scales larger than @xmath9 . in this context it is worthwhile to discuss the findings about the ldos for _ open _ systems .
these are defined by surrounding the system with an ideally conducting medium . in that case
we can not talk about eigenstates of the system .
still the ldos as given by the imaginary part of the green s function is a meaningful object .
the system itself does not conserve total probability which formally breaks hermiticity of the effective hamiltonian for the system .
for the 2d and 3d case , however , the results for the distribution of ldos are qualitatively the same as for the distribution of the restricted ldos in the closed systems .
a qualitative difference occurs for the quasi-1d case . in the open quasi-1d conductor
the ldos shows log - normal tails , @xmath813 where @xmath228 is the average dos , @xmath280 is of order @xmath15 , and @xmath814 .
this equation is also valid in strict 1d .
the log - normal tails are _ not _ due to the statistics of the eigenstates of the corresponding closed system , but can rather be interpreted as reflecting the distribution of current - relaxation times for the open system .
the latter have a similar log - normal distribution .
the broadness of this distribution is due to the presence of anomalously long - living states @xcite . in order to better understand the idealized quasi-1d system we will focus in the following subsection on the description of transport in idealized quasi-1d systems .
one particular question we like to address@xcite : is it possible to obtain similar log - normal tails for the conductance distribution in the metallic regime as those found in the @xmath595 case by altshuler _
@xcite ? the transport properties of a quasi-1d system can be obtained within a transfer matrix modeling as outlined in sect .
[ subqua ] .
the characteristic parameters are the length @xmath6 , the number of channels @xmath286 related to the cross - section @xmath102 and the mean free path @xmath9 . in close analogy to the procedure that led to a fpe for the intrinsic resistance in 1d ( eq .
( [ 6.33 ] ) one can derive a fpe for the whole transfer matrix @xmath315 .
the essential assumptions for this to be possible are : 1 . the locally weak scattering condition @xmath815 2 . the statistical independence of strip transfer matrices defined for adjacent strips . fixing the statistical properties of a single strip transfer matrix yields a fpe for @xmath319 .
one particular model turned out to be analytically tractable to a large extent , the so - called _ isotropic _ quasi-1d model . in this model
@xcite it is assumed that forward and backward scattering within a strip @xmath295-matrix is equal for all channels .
this crucial simplification means that there is no diffusion in transversal direction .
the modeling thus restricts the width @xmath816 to be less than the mean free path @xmath9 @xcite .
therefore , an idealized quasi-1d model in the sense of the foregoing section results .
furthermore , the most explicit results can be obtained in the limit of large channel numbers , or more precisely when @xmath817 but @xmath818 remaining fixed and finite . in terms of the polar parameterization eq.([3.52 ] ) the isotropy assumption is equivalent to the assumption that the unitary matrices @xmath329 are distributed isotropically within the unitary group .
similar to the standard random matrix theory for hamiltonians the statistics of eigenvectors is treated as being trivial and decoupled from the statistics of the radial parameters @xmath819 .
a consequence of the isotropy assumption is the fact that a closed fpe can be found for the distribution of the radial parameters @xmath819 alone .
these radial parameters determine the conductance ( eq . ( [ 3.51 ] ) ) @xmath820 the derivation follows the same principle steps as outlined in sect .
[ subscadis ] and the final result is the well - known dmpk - equation @xcite @xmath821 here @xmath822 characterizes different symmetry classes ( systems with time reversal symmetry @xmath467 , systems with broken time reversal symmetry @xmath823 , systems with time reversal symmetry and spin - orbit interaction @xmath468 ) and @xmath824 is the jacobian for the transformation from the cartesian parameterization @xmath825 to the polar parameterization of the transfer matrix .
the dmpk equation has been solved exactly by beenakker and rejaei @xcite for @xmath826 and for @xmath827 formal solutions exist @xcite .
common to the solutions is that they represent a gibbs - ensemble of the form already encountered in the standard random matrix theory for energy levels , eqs .
( [ 5.19a],[5.19b ] ) , @xmath828 for @xmath466 the explicit form of the two- and one - body potentials read @xmath829 it is worth mentioning that , for @xmath830 ( open channels @xmath831 ) , the two - body potential reduces to @xmath832 which has been used to model quasi-1d wires , before the exact solution was found ( see e.g. @xcite ) .
a general feature of the gibbs - ensembles is that physical parameters , such as @xmath833 and @xmath6 do only occur in the one - body potential @xmath834 in the combination @xmath835 . in contrast
, the two - body potential is a
_ universal function _ reflecting universal level repulsion . without solving the dmpk equation
explicitly one can study the resulting equations for the moments @xmath836 by integrating on both sides of the dmpk equation over @xmath837 . a perturbative analysis
as a series expansion in powers of the single parameter @xmath838 is possible in the limit @xmath839 , i.e. in the diffusive limit ( mello _ et al . _ in ref .
this yields the result for the average conductance @xmath840 the constant @xmath841 contains corrections to the classical result ( for @xmath842 these are negative weak localization corrections ) .
thus , the leading term is the one we have already anticipated in eq .
( [ 3.59 ] ) .
the leading term for the variance @xmath843 turns out to be universal @xmath844 and proofs the ucf effect for isotropic quasi-1d .
the effort to derive results for higher moments by the perturbative method increases drastically . in the limit @xmath845
it is advantageous to transform to the parameters @xmath846 ( eq . ( [ 3.51 ] ) ) since they become the lyapunov exponents in the thermodynamic limit .
their interaction can then be neglected and one ends up with an effective joint probability density of the form @xcite @xmath847 demonstrating again the self - averaging feature of lyapunov exponents . in that limit
the conductance is @xmath848 since all higher @xmath846 contribute much less .
the conductance distribution is then found to be of the log - normal form .
the typical value and the log - variance are related by ( @xmath849 ) between variance and average value instead of a factor @xmath245 occuring in eq .
( [ 7.25c ] ) . ]
@xmath850 the localization length is then found to be @xmath851 before we look at the conductance distribution in more detail we report on the findings concerning the distribution of transmittances @xmath852 and @xmath853 .
they are related to the eigenvalues @xmath854 of @xmath855 by unitary matrices @xmath856 and @xmath857
@xmath858 these equations point out that transmittances are analogs of wave function amplitudes @xmath859 . however , the transmittances do depend also explicitly on the eigenvalues @xmath860 . the statistics of the unitary matrices was assumed to be isotropic in the dmpk approach and the eigenvalues follow the dmpk equation .
it turned out that the distribution of transmittances can be calculated exactly @xcite .
actually , only @xmath861 needs to be calculated in the large @xmath286 limit . in a work by kogan and kaveh
@xcite it was shown that , for isotropic scattering in the large @xmath286 limit , the distributions of @xmath327 and @xmath862 are connected by @xmath863 in the metallic regime @xmath864 , @xmath738 , the distribution of @xmath865 is of the rayleigh type for the bulk of the distribution ( @xmath466 ) and develops stretched exponential tails ( first obtained by nieuwenhuizen and van rossum @xcite ) @xmath866 it is therefore completely analogous to the distribution of wave function amplitudes in _ closed _
quasi-1d systems ( see eq .
( [ 7.15 ] ) ) .
the corresponding distribution of @xmath327 is gaussian in the bulk and develops exponential tails @xmath867 these results depend heavily on the isotropy assumption and they indicate that it is very unlikely that the distribution of conductance , @xmath868 , in the metallic regime can develop log - normal tails in isotropic quasi-1d .
however , so far the extension to conductance is not easy since the correlation between eigenvalues @xmath326 have to be taken into account . in the localized regime @xmath869
, the statistics is dominated again by the smallest value @xmath750 of @xmath846 , i.e. by the largest value @xmath870 of @xmath326 .
this value is log - normally distributed ( see eq .
( [ 7.25c ] ) ) and accordingly the transmittances are log - normally distributed , too . since the conductance is a linear statistics of the transfer matrix radial coordinates @xmath819 @xmath871 the problem of a complete statistical description of @xmath113 reduces to the problem of a linear statistics @xmath872 in a gibbs - ensemble of a classical gas determined by a universal two - body potential ( which can be chosen to be symmetric ) and a parameter - dependent one - body ( confining ) potential @xmath873 beenakker pointed out @xcite that the one - body potential can be viewed as a source term in the _ partition sum _ @xmath874
all cumulants are linear combinations of moments of order @xmath875 .
while moments can be generated from a partition sum @xmath182 , the corresponding cumulants are generated by @xmath876 .
the gaussian distribution is characterized by vanishing cumulants for @xmath877 of the level - density , @xmath878 can be obtained by functional derivatives @xmath879 cumulants of a linear statistics @xmath872 are given by integration @xmath880 the whole distribution @xmath881 can be obtained from a modified partition sum @xmath882 , @xmath883 where @xmath882 follows from @xmath182 through a simple shift in the one - body potential @xmath884 in order to work with functional derivatives one has to know @xmath182 ( or @xmath882 ) as an explicit functional of the one - body - potential . alternatively , the knowledge of the average level density @xmath885 as a functional of the one - body - potential would be enough .
such representation of the average level - density @xmath886 was developed by dyson @xcite . here
we outline a derivation @xcite which allows for a systematic check of the range of validity of the results .
we start from the partition sum @xmath182 , or @xmath882 and proceed by the following steps : 1 .
the hamiltonian is expressed by the level - density @xmath887 , replacing sums by integration . here
the absence of self - interaction is ignored , i.e @xmath888 is replaced by @xmath889 .
2 . with the help of a @xmath212-functional
we introduce a field @xmath890 that takes the role of the level - density .
we represent the @xmath212-functional by its fourier representation which introduces a conjugate field @xmath891 .
4 . now the original integration over the set @xmath892 can be carried out leaving a field theoretical partition function in terms of two field degrees of freedom , @xmath890 and @xmath891 . 5 . due to the two - body character of the original @xmath893
the field @xmath890 can be integrated out by a gaussian integration .
the final result is a path integral representation of the partition sum @xmath182 or of the complete distribution function @xmath881 , where the integration runs over field configurations of @xmath891 .
we concentrate on @xmath881 which reads @xmath894 here we use a scalar product notation , @xmath895 @xmath896 denotes the inverse operator of @xmath268 ( @xmath897 ) , the functional @xmath898 is defined as @xmath899 and the functional @xmath900 is a free energy of @xmath286 independent particles with one - body - potential @xmath901 , @xmath902 the omission of the self - interaction can be cured on an effective potential level . by shifting the one - body - potential @xmath834 to @xmath903 where @xmath904 is the average level spacing at @xmath775 ,
the theory is able to account for those quantities that are smooth on the scale of @xmath904 .
for example , with the logarithmic interaction ,
@xmath905 the effective potential reads @xmath906 . in the following , we will denote @xmath907 by @xmath512 . the path integral in eq .
( [ 7.35 ] ) can not be calculated exactly . however , for large @xmath286 one can use the method of stationary point . to be more explicit
we will now use the conductance @xmath113 as linear statistics variable and recall that @xmath908 contains the pre - factor @xmath909 which we assume to be large : @xmath910 since we are looking for the conductance distribution in the metallic regime . introducing the _ mean - field _ level density @xmath911 as @xmath912 corresponding to the stationary point , @xmath913 the mean - field equation reads @xmath914 here the kernel @xmath915 is defined as @xmath916 and @xmath917 as @xmath918 with @xmath919 in deriving these equations it has been used that the mean - field level density is normalized @xmath920 and yields the current @xmath113 as an expectation value @xmath921 .
now one can draw the following conclusions : 1 .
for @xmath922 the expression @xmath923 equals the average value of @xmath113 , independently of current @xmath113 .
since @xmath512 contains the large factor @xmath924 the inequality is satisfied as long as @xmath925 2 . the stationary point of @xmath295
is then given by @xmath926 where @xmath927 is independent of current @xmath113 .
3 . one can also analyze fluctuations around the stationary point and show that they give sub - leading contributions to the path integral .
finally , one arrives at the conclusion , that the distribution of conductance is , for @xmath928 , @xmath929 up to @xmath930 corrections in the exponent . therefore , our conclusion is that the conductance distribution in isotropic quasi-1d does not show long - tails in the regime @xmath928 , but is gaussian with universal variance .
the universality is due to the universal two - body potential @xmath931 which reflects essentially the universal level repulsion expressed by the jacobian in eq .
( [ 7.20 ] ) .
recall , that the occurrence of log - normal tails in @xmath595 happened already for @xmath932 . the absence of log - normal tails in the conductance distribution for isotropic quasi-1d can be interpreted as being caused by the isotropy assumption .
relaxing from the isotropy assumption the resulting fpe for the radial parameters @xmath819 is not closed .
rather the diffusion function depends on the statistics of eigenvectors of @xmath751 .
the latter is presently unknown , but one can conclude in general @xcite that long tails in the distribution of the eigenvectors will lead to long - tails in the distribution of @xmath113 .
in addition , one can conclude that in such case the probability density of the radial parameters is no longer given by a gibbs - ensemble with only one - body and two - body potentials . also the results for transmittances
indicate that the isotropy assumption expels any log - normal tails for the conductance distribution . on the other hand
, the isotropy assumption can only be justified for transversal width @xmath816 being less the mean free path @xmath9 . it is instructive to look at the correlation length @xmath0 relevant to such idealized quasi-1d systems .
recall , that the correlation length , in the spirit of one - parameter scaling theory , is defined as that fictitious system size @xmath0 for which the conductance of a @xmath138-dimensional cubic system is of order @xmath15 . to attach a correlation length to the quasi-1d system requires a meaningful definition of a conductance in a cubic system .
the easiest way is to use a parallel composition law and define @xmath933 where @xmath934 is the conductance of the quasi-1d system with length @xmath6 and cross - section @xmath102 . for the isotropic quasi-1d system with @xmath935 and @xmath936 , @xmath937
, we thus have @xmath938 which is extremely large .
consequently , the correlation length is @xmath939 thus , even for @xmath572 the isotropic quasi-1d model corresponds to microscopic correlation lengths .
one should note that it is _ not _ the quasi-1d localization length @xmath940 which plays the role of the correlation length , but @xmath0 . both coincide only for strict 1d where @xmath334 .
we interpret the occurrence of log - normal tails in the conductance distribution as precursors of the ld transition .
for @xmath595 they are controlled by the one - parameter scaling variable @xmath941 . with this interpretation
the absence of such tails in isotropic quasi-1d is no surprise since the appropriate scaling variable @xmath942 is sent to infinity .
giving up the isotropy assumption allows to study more realistic systems by a quasi-1d set - up .
true dimensionality effects should enter the physical properties .
an analytical treatment becomes much more difficult , but a numerical treatment turns out to be very efficient .
the green s function can be studied by recursive methods @xcite , putting strips of cross - section @xmath102 together .
alternatively , strip transfer matrices can be multiplied @xcite .
that quantity which can be obtained with high precision ( due to its self - averaging property ) is the quasi-1d localization length @xmath943 . by studying @xmath943 as a function of @xmath816
one should be able to study the localization behavior in a large ( cubic ) @xmath138-dimensional system .
our first task is to construct a suitable scaling variable which can be derived from @xmath943 .
when we increase @xmath816 and the system flows towards the insulating ( localized ) phase , @xmath943 will stop growing as soon as @xmath816 is comparable with the ( finite ! ) localization length of the infinite @xmath138-dimensional system : @xmath944 on the other hand , @xmath943 will grow forever , if the system floats to the metallic ( delocalized ) phase . at the critical point
, however , we expect the system to be scale invariant ( diverging correlation length ) , i.e. @xmath943 should scale exactly as @xmath816 as soon as finite size effects have died out .
so one tries @xmath945 as a scaling variable . a further reason for this choice comes from the interpretation that , in the metallic regime , an appropriate conductance for the corresponding cubic system is given by eq .
( [ 7.21 ] ) where @xmath946 yielding @xmath947 in the localized regime , where @xmath943 approaches @xmath948 , @xmath949 represents the inverse logarithm of the corresponding conductance @xmath950 now , one - parameter scaling means that away from the critical point ( where @xmath951 is a constant ) @xmath952 does not depend on the correlation length @xmath0 ( which can be identified with @xmath948 in the localized phase ) and @xmath953 separately , but only on their dimensionless ratio @xmath954 here @xmath955 is a called the scaling function . in the presence of ld transitions , @xmath955 is a two - valued function .
close to a critical point ( where linearizing of the corresponding @xmath12-function is appropriate ) the scaling function is of the form ( cf .
( [ 6.19b ] ) ) @xmath956 where the positive ( negative ) sign marks the delocalized ( localized ) branch of @xmath955 : @xmath957 ( @xmath958 ) . the general procedure to extract the scaling function from a large number of calculated localization lengths @xmath943 is to find a quantity @xmath0 independent of @xmath953 such that the entire data set @xmath959 ( where @xmath118 is a system parameter related to e.g. the disorder or fermi energy with critical value @xmath960 ) collapses onto a single curve .
this curve then defines @xmath955 . in practice , this can be done by plotting the @xmath961 curves on transparencies and moving them around by hand until they fall on top of each other or by a least squares fitting procedure . a typical example for a scaling function is shown in fig .
24 . after determining @xmath555
one can determine the critical exponent @xmath228 from a fit to eq .
( [ fcrit ] ) .
it may , however , turn out to be difficult to find the critical value @xmath555 .
for example , it can happen that the curves @xmath962 , for different choices of @xmath816 , do not intersect at one point @xmath963 .
this can be due to the fact that the system size is not large enough to reach the asymptotic scaling regime , i.e. it can be a finite size effect .
assume that , at the _ true _ critical point @xmath964 , the scaling variable still shows a dependence on @xmath816 of the form @xmath965 here @xmath966 is a critical exponent .
more convenient is @xmath967 with @xmath968 of order @xmath15 and an explicit scale @xmath969 called _ irrelevant _ length .
since @xmath969 stays finite the correction to @xmath555 decreases when @xmath816 becomes larger than @xmath969 .
@xmath970 in this sense , the length @xmath969 becomes irrelevant in the large @xmath816 limit .
however , in a finite size calculation one can never be sure that microscopic scales are already much smaller than @xmath816 and one has to face the fact that corrections to the ideal scaling , eq .
( [ fcrit ] ) , can occur .
they can then be analyzed by means of irrelevant length scales , eq .
( [ 8.10a ] ) .
the best studied system is the quantum hall system in 2d .
although no metallic phase exists , the position of the critical point on the energy scale is precisely known for certain models due to spectral symmetries .
this fact and the two - dimensionality enables to get results for critical quantities with high precision ( see e.g. @xcite ) .
also a 3d analog of the quantum hall system has been studied recently @xcite .
further results are known for 2d systems with spin - orbit scattering ( see e.g. @xcite ) and for a number of 3d systems ( see e.g. @xcite ) .
the finite size scaling method made it possible to analyze a self - averaging quantity serving as scaling variable in the sense of one - parameter scaling theory .
it demonstrates not only that one - parameter scaling does work , but also allows for explicit calculations of the critical exponent of the correlation length
. however , not very much information comes out of this for the statistical properties of non - self - averaging quantities like conductance and ldos . at least , by eqs .
( [ ohm1],[loc2 ] ) one can expect that the scaling variable @xmath971 of the finite size scaling method is related to the typical conductance @xmath941 .
the latter can be identified as the mean value in the delocalized phase and as the geometric mean in the localized phase . since , in the delocalized phase the geometric mean is very close to the mean ( the distribution is gaussian with @xmath972 ) it is reasonable to expect that the geometric mean @xmath973 can serve as a typical conductance which in turn is a scaling variable . in this section
we will firstly consider the statistics of critical eigenstates .
after the findings of the previous section we already expect them to show the multifractal scaling property .
we will discuss the implications of this for the ldos .
several arguments will point out that the typical conductance will determine the distributions of local quantities .
after a brief discussion of critical energy level statistics we collect results for the critical conductance distribution . in finite size systems the correlation length @xmath0 of the electronic states is larger than the system size @xmath6 for a certain parameter range , @xmath974 , around the critical parameter value @xmath963 .
these states are called _
critical states_. in the thermodynamic limit @xmath975 where @xmath228 is the critical exponent of @xmath0 .
after the pioneering works by wegner @xcite and aoki @xcite it became clear that the critical wave functions have a multifractal structure ( for a review see @xcite and references therein ) .
the entire distribution of local amplitudes and its scaling behavior is encoded in the multifractal @xmath976 spectrum , as outlined in sect .
[ subscadis ] .
the distribution @xmath977 is broad on all length scales and close to a log - normal distribution .
the most important quantity is the maximum position , @xmath728 , of @xmath976 .
it describes the scaling behavior of the geometric mean of what serves as a _
typical _ amplitude of a critical wave function . in early works
aoki @xcite gave a nice argument for the multifractal behavior of critical wave functions ( although at that time the phrase _
multifractality _ was not yet common ) .
his argument goes as follows .
consider the inverse participation number defined by @xmath978 where @xmath979 denotes a @xmath138-dimensional region with linear size @xmath6 .
if the wave function @xmath980 is uniformly distributed like in the delocalized phase then @xmath981 and the participation ratio @xmath982 is constant . in the localized regime @xmath983 and @xmath516
vanishes in the thermodynamic limit . at the transition point where the wave function is extended
the participation ratio still has to vanish in the thermodynamic limit if the ld phenomenon is similar to a second order phase transition for which the participation ratio acts as an order parameter .
consequently , @xmath984 scales with a power @xmath985 .
wegner had already calculated the whole spectrum of exponents for generalized inverse participation numbers within the non - linear sigma - model @xcite .
this spectrum was interpreted as a multifractal spectrum by castellani and peliti @xcite .
after extensive numerical work ( e.g. @xcite ) the following description of the statistics of critical wave functions @xmath986 is now established : at the ld transition the distribution function @xmath977 gives rise to the power law scaling for the moments , @xmath987 where @xmath988 is a non - linear function of @xmath669 .
this non - linearity is a direct consequence of aoki s observation that @xmath989 . in practice
it turns out that , to a good accuracy , the disorder average can often be substituted by the spatial average over one wave function for a given configuration . within numerical accuracy
the resulting spectra are identical .
the corresponding ( universal ) distribution function can be described in terms of the single - humped , positive @xmath976 spectrum , @xmath990 where @xmath991 ; as shown in sect .
[ subscadis ] @xmath976 is related to @xmath715 by a legendre transformation @xmath992 the parabolic approximation ( pa ) , eq .
( [ parabol ] ) , contains @xmath728 as the only parameter besides @xmath138 .
this is due to the assumed validity of the pa at least up to @xmath993 .
equation ( [ parabol ] ) corresponds to a log - normal distribution centered around the typical value @xmath994 with log - variance proportional to @xmath796 . a simple one - parameter approximation for @xmath976 which takes into account that the support @xmath995 $ ] of @xmath976 is finite , is the semi - elliptic approximation ( sea ) @xmath996 to demonstrate that the distribution of local amplitudes of critical eigenstates is encoded in @xmath976 we discuss numerical results for a quantum hall system ( qhs ) @xcite .
the quantum hall system is the most convenient to study critical eigenstates since one knows the critical point exactly and it is only two - dimensional which keeps the numerical effort low .
the wave functions are calculated for the model of independent ( spin - less ) electrons subject to strong magnetic field and disorder .
the disorder was implemented by a set of @xmath212-impurities with random positions and random strengths symmetric around zero .
the microscopic length scale of the problem is the magnetic length @xmath159 defined by the size of an area penetrated by a single flux - quantum , @xmath997 .
the representing hamiltonian matrix was worked out in the landau representation diagonalizing the clean system ( no disorder ) .
the clean case has the famous landau level spectrum of highly degenerate and equally spaced levels at discrete energy values .
the spacing is given by the cyclotron frequency @xmath998 .
disorder ( symmetric around zero energy ) broadens the landau levels to landau bands of non - degenerate states .
for strong magnetic fields the broadened landau bands are still separeted , the band - width @xmath999 being smaller than @xmath1000 ( see fig .
25 ) . restricting to one landau band
( it is most convenient to take the lowest ) the ld transition takes place precisely at the center of the symmetric landau band ( cf .
@xcite,@xcite ) .
a system of area @xmath1001 is represented by a finite matrix of dimension @xmath1002 .
the numerical diagonalization yields the eigenvalues and eigenstates for any desired energy window within the lowest landau band . in fig .
26 the squared amplitudes of a wave function from the center of the landau band are shown together with the @xmath976 spectrum calculated from these amplitudes .
the corresponding histogram of the logarithm of amplitudes ( measured on a box of size @xmath1003 ) is displayed in fig .
27 together with the distribution function calculated from the @xmath976 spectrum using eq .
( [ 8.3 ] ) .
these figures demonstrate that the distribution of amplitudes is ( i ) encoded in the @xmath976 spectrum and ( ii ) is close to a log - normal distribution characterized by one critical exponent @xmath1004 ( the average over 130 critical states ) .
figures 26 , 27 correspond to a model of @xmath212-impurities resulting in a short range potential correlation .
similar calculations for a long range potential correlation ( the potential correlation length is much larger than the magnetic length ) have been performed within the chalker - coddington network model @xcite .
a corresponding critical wave function and its histogram are shown in figs .
they demonstrate the universality of the multifractal properties with respect to the potential correlation length . to study the spatial correlations of amplitudes for a fixed energy consider the @xmath1005dependent correlations @xmath1006}(r , l_b , l):=\left\langle
\lbrack p_i(l_b)\rbrack^q \lbrack p_{i+s}(l_b)\rbrack^q\right\rangle_l\label{8.7}\ ] ] where the average is to be taken over all pairs of boxes with fixed distance @xmath1007 . for critical states where the microscopic scale ( @xmath159 in our case ) and the macroscopic scale ( the localization length @xmath144 in our case ) are separated , one can expect power law behavior of @xmath1008}$ ] in the regime @xmath1009 usually , in critical phenomena one studies correlations for infinite system size ( and @xmath489 being microscopic ) as a function of @xmath298 alone .
this is justified if , for large enough system sizes @xmath6 , the correlation function is independent of @xmath6 ( for simplicity we neglect any trivial @xmath6-dependence due to pre - factors in the definition of the observable @xmath1010 ) . however
, this is not true in the multifractal case .
multifractality reflects broadness of the distribution function @xmath1011 on all length scales .
the local box observable @xmath1012 depends on a large number of conditions for the entire system of linear size @xmath6 , simultaneously . in the context of the ld transition coherence at zero temperature
is due to quantum mechanical phase coherence of the electron s wave function , and disorder introduces a huge number of parameters , e.g. the position of point - scatterers . in the multifractal scaling case one has to face the fact that @xmath1008}$ ] depends non - trivially on @xmath6 , even for @xmath381 .
we have incorporated such behavior already in our general discussion on correlation functions in sect .
[ subrelbet ] . therefore , we consider the regime @xmath1013 and make the ansatz @xmath1006}(r , l_b , l)\propto l_b^{x_2(q)}l^{-y_2(q)}r^{-z(q)}\ , .\label{8.10}\ ] ] the task is now to find the scaling relations between the set of exponents @xmath1014 and the @xmath715 function .
these scaling relations can be derived by requiring consistency with the limiting situations ( i ) where @xmath298 is of the order of @xmath489 and ( ii ) where @xmath298 is of the order of @xmath6 .
we find @xcite @xmath1015 it is worth mentioning that the sum @xmath1016 vanishes due to the normalization of the wave function . the analytic behavior of @xmath1017 according to eq .
( [ 8.11c ] ) is shown in fig .
30 . in general
, it is non - negative and asymptotically bounded by the dimension @xmath138 . to check on the validity of eq .
( [ 8.11c ] ) pracz _ et al .
_ took @xmath1018 critical states of a system with @xmath1019 and calculated @xmath1008}(l_b , r , l)$ ] with fixed values @xmath1020 ; @xmath1021 . within the errors
the validity of the scaling relations in eqs .
( [ 8.11c ] , [ 8.11b ] ) could be confirmed . to include aspects of local energy statistics as well ,
let us consider the @xmath669-dependent correlation of box probabilities corresponding to two different eigenstates with energies @xmath208 and @xmath1022 @xmath1023}(r , l_b , l):=\left\langle \lbrack p_i(e;l_b)\rbrack^q \lbrack p_{i+s}(e+\omega;l_b)\rbrack^q\right\rangle_l \ , .\label{8.13}\ ] ] to understand the correlation behavior of non - localized states with respect to the energy separation one has to compare the relevant energy scales of the problem .
these are the average level spacing @xmath115 and the ( frequency dependent ) thouless energy energy @xmath1024 corresponding to the time a wave packet ( formed from states within an energy window of width @xmath77 ) needs to diffuse through the system , @xmath1025 . here
@xmath1026 is the corresponding diffusion constant .
according to chalker @xcite , these scales give rise to the definition of two length scales depending on the energy separation @xmath77 : @xmath1027 the first length scale , @xmath1028 , is the typical distance a wave packet will travel diffusively in a time @xmath1029 . from this
it is natural to assume that correlations between @xmath1030 and @xmath1031 will be present at least for distances @xmath1032 whereas for larger distances the amplitudes are uncorrelated .
such uncorrelated behavior is typical for the standard random matrix theory approach to extended states in random systems .
recall , that in this theory it is assumed that the unitary matrices that diagonalize the hamiltonian are distributed uniformly in the unitary group and no correlations ( apart from the unitarity property ) between different matrix elements occur .
thus , the presence of correlations here is an explicit breakdown of the _ no - preferential basis _ assumption . in electron systems with spatial disorder
a preference to some basis is always given .
this preference can not be seen in @xmath1008}_\omega$ ] for distances @xmath1033 .
the second length scale , @xmath1034 , is the linear size of a system with level spacing @xmath77 .
two wave functions with energetic separation smaller than the level spacing show a spatial correlation behavior of its amplitudes similar to that corresponding to one of those wave functions , i.e. they are statistically indistinguishable . at the critical point of the ld transition
the typical conductance becomes a size independent quantity , @xmath1035 , and with the help of the einstein relation between conductivity and diffusion one finds @xmath1036 @xcite .
since @xmath1037 is of @xmath507 the two length scales coincide at the ld transition .
therefore , one can focus on the role of @xmath1034 . asking for the scaling properties of @xmath1008}_\omega$ ] in the regime @xmath1038 we make the ansatz ( cf .
( [ 8.10 ] ) ) @xmath1006}_\omega\propto l_b^{x_2(q)}r^{-z(q)}l_\omega^{z(q)}l^{-y_2(q)}\ , .\label{8.16}\ ] ] here we have already anticipated that the exponent with respect to @xmath298 is @xmath1017 , as given before . as in the case of @xmath1008}$ ] for a fixed energy by considering limiting situations one finds @xmath1039 now the following conclusions can be drawn : 1 .
the result for @xmath1017 is the same as in the case of zero energy separation .
the energy separation @xmath77 is not an independent scaling parameter but appears only in the combination @xmath1040 .
the exponent corresponding to the box size , @xmath1041 , is not affected by a finite energy separation but the exponent corresponding to the system size @xmath6 ( which is @xmath1042 for zero energy separation ) splits up into the exponents @xmath1043 ( corresponding to @xmath1034 ) and @xmath1044 ( corresponding to @xmath6 for finite energy separation ) . in the work by pracz _
_ it was demonstrated that the scaling relations are consistent with numerical results .
similar findings have been obtained by metzler @xcite within the scattering matrix network model ( chalker coddington model ) .
having established the role of @xmath1045 as the relevant scaling parameter for correlations of eigenstate amplitudes with universal exponent @xmath1017 ( related to @xmath715 by a scaling relation ) let us now discuss the consequences of this for the interpretation of the local density of states being an order parameter of the ld transition . with the smearing - out of the @xmath212-functions
the ldos is given by ( see eq .
( [ 3.17 ] ) ) @xmath1046 where @xmath1047 stands for the microcanonical average of squared amplitudes at a given energy @xmath208 . since @xmath433 behaves as @xmath1048 , the scaling behavior of the ldos
is determined by that of the wave function .
consequently , we have @xmath1049 and for the typical value @xmath1050 \propto l^{d-\alpha_0 } \label{8.25}\ ] ] which _ does _ reflect the ld transition . scaling @xmath6 with the correlation length @xmath1051 the typical ldos vanishes on approaching the critical point with exponent @xmath1052 ( @xmath1053 in quantum hall systems ) .
the behavior of the typical ldos is , according to the argument presented in the beginning of this section , similar to that of the participation ratio introduced by wegner @xcite .
the participation ratio was based on the second moment of local probability .
higher moments will work as well and our notion of typical ldos is just a bit more convenient since it focuses on a typical value rather than on some specific moments .
our interpretation is also consistent with the findings for the ldos in the localized and delocalized phase as discussed in sect .
[ staide ] .
therefore , we summarize the findings : 1 . in the localized phase the typical local density of states vanishes
. 2 . in the delocalized phase the typical local density of states is finite and positive .
3 . approaching the ld transition from the delocalized phase the typical ldos vanishes with a positive critical exponent , @xmath1054 a schematic picture for the typical ldos that stresses its role as order parameter is shown in fig .
the unconventional feature as compared to ordinary critical phenomena lies in the facts that 1 .
the order parameter field has a broad distribution resulting in a non - linear dependence of exponents on the degree of moments considered ( multifractality ) .
2 . the average value shows a vanishing scaling exponent while the typical value gives rise to a positive scaling exponent the scaling relations that we derived for the wave functions amplitudes transform to scaling relations of the ldos since each box amplitude has to be multiplied by a constant factor of @xmath1055 , @xmath1056 with @xmath1057 these scaling relations form , in the sense of sect .
[ subrelbet ] , the appropriate scaling relations connecting the spatial correlations of the local order parameter field to its scaling dimensions ( cf .
( [ 8.24]),([8.26]),([8.27 ] ) ) .
we mention that @xmath1058 ( for the correlator of the density of states in quantum hall systems ) with @xmath1040 forming the scaling parameter is equivalent ( cf .
@xcite ) to the phenomenon of _ anomalous diffusion _ found by chalker and daniell @xcite .
as pointed out in @xcite , the anomalous character of diffusion lies in the non - gaussian dispersion of a wave packet in time @xmath297 despite the fact that the average diameter grows like @xmath1059 .
this non - gaussian time dispersion is caused by the multifractal character of eigenstates .
a further support for the order parameter interpretation comes from conformal mapping ideas for 2d systems . in ordinary critical phenomena theory
, the following assumption for critical correlation functions @xmath1060 seems plausible : scale invariance , reflected by power laws , should hold also for local scale transformations which preserve angles but may change scales locally ( called conformal mappings ) @xcite . thus , the main idea behind conformal mapping arguments in scaling theory is an extension of a homogeneity law for correlation functions with respect to rescaling .
such a law exists also in the multifractal case , since any rescaling of _ all _ length scales in the ldos correlator ( for a fixed energy ) by the same scaling factor @xmath258 leads to @xmath1061 where @xmath1062 is given by eq .
( [ 8.27 ] ) . extending this law to conformal mappings of a geometry @xmath979 to geometry @xmath1063 leads , for large but finite 2d systems , to @xmath1064 where @xmath1065 is any holomorphic function of complex coordinate @xmath1066 and @xmath1067 denotes the derivative . by choosing @xmath1068
which maps the plane onto a strip one can show ( cf .
@xcite ) that the correlator in the strip is characterized by a @xmath669-dependent quasi-1d localization length @xmath1069 which is related to @xmath1062 by @xmath1070 in fss calculations it is rather the average of the logarithm of the correlation function which is calculated by knowing that this quantity defines a typical localization length . therefore , the result of eq .
( [ conf1 ] ) leads in particular ( consider @xmath669 close to @xmath60 ) to the relation @xmath1071 here @xmath555 appears since it is just the typical localization length at criticality divided by the width @xmath816 .
equation ( [ conf1 ] ) has been confirmed in numerical calculations by dohmen _
@xcite , and eq . ( [ conf2 ] ) is , so far , in accordance with all known numerical results .
the fact that the critical value @xmath796 is related to the critical value of the scaling variable ( at least in 2d ) is an important hint for the typical conductance @xmath756 being a one - parameter scaling variable .
the reasoning behind this statement goes as follows : as pointed out in sect .
[ subfin ] the scaling variable @xmath1072 of finite size scaling in quasi-1d can be related to the typical conductance of a cube @xmath1073 . from the @xmath733 field
theoretic calculations at the ld transition @xcite as well as from the multifractal ldos found in 2d @xcite we can conclude : if the correlation length is larger than system size and the typical conductance is still large , the eigenstates behave critical and are multifractal . the multifractal exponent @xmath796 is determined by @xmath756 and @xmath1074 at a generic critical point with @xmath1075 the conformal mapping relation tells , that also here the value of @xmath796 is determined by @xmath941 .
this means that the bulk of the ldos distribution function is , in the critical regime , already determined by the value of typical conductance .
before we collect results for the conductance distribution at criticality we discuss a similar object related to the level statistics .
not only the ldos and conductance distributions should be universal functions at criticality ( see eq .
( [ limitdis ] ) ) , but also e.g. the level spacing distribution @xmath1076 and the two - level correlation function @xmath1077 . to study these quantities in view of the one - parameter scaling picture
was put forward in a work by shklovskii _
these authors pointed out that the study of the level spacing distribution @xmath1076 is , starting from a hamiltonian modeling , numerically much easier to perform than calculating conductances .
only the spectrum has to be calculated . according to the one - parameter scaling picture
@xmath1076 should obey the following scaling behavior @xmath1078 here @xmath1079 is a function that does no longer depend on microscopic details , but only on one scale dependent parameter @xmath639 . depending on the initial value of @xmath639 the system flows to either the localized fixed point ( l ) , the delocalized fixed point ( d ) or it stays at the critical fixed point ( c ) . in terms of the typical conductance , @xmath1080 , the fixed points are characterized by @xmath1081 the corresponding asymptotic level - spacing distributions @xmath1082 are given by the wigner surmise , eq .
( [ 5.22 ] ) , @xmath1083 the poisson distribution , eq . ( [ 5.11 ] ) , @xmath1084 and by a , yet unknown , critical distribution @xmath1085 .
numerical studies by several authors ( e.g. @xcite , @xcite , @xcite , @xcite , @xcite ) have demonstrated that a size independent critical distribution @xmath1085 distribution exists , once the parameter @xmath118 triggering the ld transition is put to its critical value @xmath963 .
it is also known that , qualitatively , the distribution resembles the wigner surmise for values @xmath1086 and the poisson law for @xmath1087 . at @xmath1088
the distributions intersect .
furthermore , the way the critical distribution is approached allows for a calculation of the critical exponent of the correlation length @xmath228 .
this can be done in several ways .
the authors of @xcite took as a scaling variable the integral over @xmath1076 up to @xmath1088 ( one could use also some other point ) @xmath1089 and analyzed it according to the general ansatz , ( eq . ( [ 6.19b ] ) , @xmath1090 the values for @xmath228 found by this method do agree with those found in fss analysis within the errors . as to the precise form of @xmath1085 there exists no conclusive theory , except that the very small @xmath258 behavior is again dictated by the level repulsion @xmath1091 .
for example , the tail of @xmath1092 is not exactly known .
it seems now to be established that there is a leading exponential tail with sub - leading stretched exponential tail contributions .
the question about the form of the tail is related @xcite to the question about the relation between the number variance @xmath411 and the average level number @xmath413 , ( see sect .
[ staide ] ) . based on one - parameter scaling arguments aronov _ et al . _
@xcite had predicted ( for the ld transition point without magnetic fields ) the following scaling law , for large @xmath413 , @xmath1093 however , in this work the anomalous diffusion at criticality ( or in other words the multifractality ) was not taken into account . correcting this omission leads to an additional linear term , characterized by a finite level compressibility @xcite @xmath1094 in view of the finite level compressibility eq .
( [ 10.6 ] ) describes sub - leading contributions to @xmath411 .
it is very interesting that the analysis of @xmath411 as a function of @xmath413 seems to allow for computing the correlation length exponent @xmath228 and the fractal dimension @xmath1095 related to the order parameter exponent @xmath1096 ( within the parabolic approximation to @xmath976 @xmath1097 ) . as to the critical conductance distribution the general considerations formulated in eqs .
( [ 10.1],[10.2 ] ) for the level spacing distribution do apply as well .
the precise form of the critical conductance distribution @xmath1098 is not known .
based on the @xmath733 calculations by altshuler et al .
@xcite , shapiro and coworkers @xcite predicted that power law tails should occur .
so far , this could not neither be proved nor disproved by numerical means . however , in a work by markos and kramer @xcite it was shown that the critical conductance distribution is universal for a 3d system without magnetic field , i.e. it does not depend on system size @xmath6 and does not depend on which parameter @xmath118 is taken to its critical point value .
the distribution showed fluctuations @xmath1099 that are of the same order ( @xmath507 ) as the mean value of the distribution .
the precise form of the tails could not be determined .
the scaling of the conductance distribution close to criticality follows from expanding the scaling variable around criticality which brings the exponent @xmath228 into play , @xmath1100 where @xmath1101 denotes the derivative of the function @xmath1102 with respect to the scaling variable @xmath639 . to analyze the behavior of @xmath1103 at one particular point
is numerically not advantageous .
therefore some integrated quantities work better .
however , as explained earlier , the chosen quantity should not be sensible to the tails of the distribution .
therefore , one should refrain from taking moments .
unfortunately , not much has been done in this direction .
fastenrath _ et al .
_ @xcite have investigated a quantity which is similar to the dissipative conductance , the so - called thouless number , for quantum hall systems and studied the distribution function .
also in this case it turned out that , at the critical point , the distribution function was scale independent with fluctuations of the same order as the typical value .
they explicitly took the typical thouless number @xmath1104 and analyzed it according to @xmath1105 they found a value for @xmath228 ( @xmath1106 ) which is in good agreement with the value obtained by fss .
thus , their observation gives further support that the typical conductance is an appropriate scaling variable .
cho and fisher @xcite have performed a calculation of a true two - probe conductance within the chalker - coddington network model for the quantum hall effect .
there findings are consistent with the two - probe experiment by cobden and kogan @xcite mentioned in the introduction : the distribution is almost uniform between @xmath60 and @xmath15 .
this means in particular that the fluctuations are of the same order as the average value . for this uniform distribution on a finite interval
the average value is already a convenient choice for the scaling variable and cho and fisher were able to show that the average value shows scaling of the form written in eq .
( [ 10.8 ] ) with the correct critical exponent @xmath1107 .
similar results were also obtained by wang _
_ @xcite .
thus , although the precise form of critical conductance distributions is not known , there are a number of results which confirm that it is universal and that the notion of typical conductance allows for studying the critical exponent of the correlation length .
in this review we have discussed a few fundamental experiments that show the importance of quantum mechanical interference effects for transport measurements in mesoscopic systems .
static disorder leads to complicated interference patterns of electronic wave functions which can cause large fluctuations in physical quantities .
in addition , they can lead to localization of states to a finite volume within the mesoscopic system .
mesoscopic systems can be experimentally realized at very low temperatures ( @xmath1108k ) and with small devices ( @xmath1109 m ) such that the phase coherence length @xmath10 is larger than system sizes @xmath6 .
we introduced the basic physical quantities which are the local and global density of states ( ldos and dos ) , the global conductance which is ( in atomic units @xmath38 ) given by a dimensionless quantity @xmath113 , and the conductivity tensor @xmath1110 ( or equivalently the diffusion function @xmath1111 ) , related to local charge transport .
the relevant physical energy scales are the quantum kinematic scales of level spacing @xmath115 and fermi energy @xmath213 , and the transport energy scale set by the thouless energy @xmath114 .
the conductance is given by the thouless formula , @xmath1112 .
a quantum kinematic length scales is the fermi wavelength @xmath8 , and transport length scales are the microscopic mean free path @xmath9 and the localization length @xmath144 for localized states .
we discussed models for disordered mesoscopic electron systems in terms of hamiltonians and scattering @xmath295-matrices . from the hamiltonian models energy eigenvalues @xmath196 and eigenvectors
@xmath247 can be calculated .
their statistical properties determine the statistics of the most important quantity , the green s function which determines all of the relevant quantities and scales . within the @xmath295-matrix modeling the transmission strengths @xmath865
are the most important quantities .
they directly determine the conductance via the bttiker formula , @xmath1113 .
in addition , certain network @xmath295-matrix models also allow for a determination of energy eigenvalues and eigenvectors . for the statistical and scaling properties of mesoscopic electron systems we collected the following insights . 1 . ideal localized systems ( with vanishing localization length @xmath144 , i.e. vanishing conductance @xmath1114 ) are characterized by a compressible spectrum of uncorrelated energies and a ldos that typically vanishes , although the average dos is finite .
2 . ideal delocalized systems ( with infinite thouless energy @xmath114 , i.e. infinite conductance @xmath1115 )
are characterized by an incompressible spectrum of correlated energies and gaussian distributed wavefunctions which leads to an exponential tail for the ldos probability distribution .
3 . changing control parameters , @xmath118 , of the mesoscopic system like particle density , disorder strength , pressure or applied fields can drive the system through a disorder induced localization - delocalization ( ld ) transition ( at @xmath963 ) which resembles a critical phenomenon . due to the randomness of disorder the transition has to be described in terms of distribution functions the flow of which with respect to system size ( scaling ) forms the object of interest in a critical phenomenon description .
4 . a one - parameter scaling theory for distribution functions of global quantities like conductance and energy - level spacing seems to work well .
it states that the distributions are essentially determined by one scale dependent parameter @xmath639 , called scaling variable , that determines the flow , @xmath1116 .
5 . a possible scaling variable for the ld transition
is given by the geometric mean of the conductance distribution which we called typical conductance @xmath756 .
with increasing system size @xmath756 can approach the idealized localized phase ( @xmath1117 ) , or it can approach the idealized delocalized phase ( @xmath1118 ) .
corrections to distribution functions as compared to the ideal situations are controlled by finite values of @xmath756 .
right at the ld transition @xmath756 will become a size independent universal quantity @xmath1119 that is of order @xmath15 in generic ld transitions .
the corresponding distribution functions of global quantities like conductance and level - spacing are universal scale - independent functions .
the fictitious system size @xmath0 for which @xmath1120 defines the correlation length of the ld transition .
in the localized phase it can be identified with the localization length @xmath144 .
close to the transition point , where it diverges , it scales with a critical exponent , @xmath1121 .
while in the localized phase the typical ( not the average ) ldos vanishes , the ldos shows power law scaling in non - localized phases . the average value is always scale - independent . in the ideal delocalized phase one exponent , simply related to dimensionality @xmath138 , controls the scaling of the whole distribution function . at the ld transition
a spectrum of scaling exponents is necessary to describe the scaling of the distribution which is close to a log - normal distribution .
the spectrum has an interpretation in terms of a spectrum of fractal dimensions ( multifractal spectrum ) .
still , there is one particular exponent , related to the typical value of the ldos , which determines the log - normal approximation for the distribution .
it is called @xmath1122 and is determined by the value of the scaling variable at criticality .
thus , multifractality is not in conflict with one - parameter scaling theory .
8 . on the contrary
, the multifractal scaling of the ldos opens the possibility to consider the typical value of the ldos as an appropriate order parameter for the ld transition .
the corresponding critical exponent is given by @xmath1123 .
further support to this idea comes from the observation that a number of scaling relations for the ldos can be formulated that are in close analogy to similar relations in ordinary critical phenomena . in short ,
the occurrence of broad distributions and multifractality in disordered mesoscopic electron systems is by no means in contradiction to the one - parameter scaling theory , but points out that only typical rather than average quantities can serve as order parameter and scaling variable in this theory . *
acknowledgments * i gratefully acknowledge helpful discussions with a. altland , y. avron , m. backhaus , j. chalker , a. dohmen , p. freche , j. hajdu , b. huckestein , r. klesse , i. lerner , m. metzler , d. polyakov , k. pracz , b. shapiro and m. zirnbauer .
this work was supported by sfb 341 of the deutsche forschungsgemeinschaft and the minerva foundation .
i thank the department of theoretical physics in oxford , u.k . , and the institute of theoretical physics at the technion , haifa , for the kind hospitality during visits in which part of the work was carried out .
we start with a general outline of how to calculate linear response quantities when the system is specified by a hamiltonian . we then specialize to additive systems where the resolvent @xmath1124 of the corresponding one - particle hamiltonian ( taken at complex energies @xmath1066 ) yields all the basic physical quantities .
studying transport in multi - probe geometries will lead to the concept of the scattering matrix . consider a many - particle system characterized by the hamilton operator @xmath198 coupled to a linear increment of a force field , @xmath1102 , @xmath1125 where @xmath636 is a hermitean operator of the system .
the system is assumed to be in an equilibrium state at temperature @xmath341 described by the ( grand ) canonical density operator @xmath1126 .
the thermodynamic susceptibility @xmath1127 is defined as @xmath1128 linearizing @xmath1126 with respect to the increment @xmath1129 the thermodynamic susceptibility reads @xmath1130 where the bilinear expression ( kubo scalar product ) @xmath1131 is called the canonical correlator of @xmath636 and @xmath1132 with @xmath1133 , @xmath1134 , and @xmath1135 is the time evolution of @xmath636 in the heisenberg picture .
the isothermal susceptibility describes an idealized linear response situation where the force field does not disturb equilibrium , but is just a parameter field of the system .
we turn now to the dynamic linear response problem @xcite where the following ( kubo ) process is considered : the increment in an applied force field @xmath1136 depends on time @xmath297 and is switched on adiabatically at @xmath1137 .
the total hamilton operator is @xmath1138 where the slowness parameter @xmath534 has to be much smaller than the inverse of a typical positive time scale @xmath1139 at which the response to the field is measured the system is assumed to be in equilibrium for @xmath1137 described by the density operator @xmath1140 and isolated for @xmath1141 . restricting our consideration to one fourier component of the force field , @xmath1142 the resulting linear increment in the fourier component of the mean value of an observable @xmath1132 is @xmath1143 where the dynamic susceptibility @xmath1144 is given by the kubo formula @xmath1145 here @xmath1146 denotes the time derivative of @xmath636 and the dynamic correlator @xmath1147 appears as the laplace transform ( @xmath1148 ) of the time dependent canonical correlator . integrating by parts in eq .
( [ 4.7b ] ) yields the fundamental equation of motion for dynamic correlators @xmath1149 another important equation of motion , referred to as kubo relation , holds for the canonical correlator @xmath1150 where @xmath1151 denotes the commutator of @xmath636 and @xmath1132 .
it is worth mentioning that the second term in eq .
( [ 4.9 ] ) can be finite in the case of zero frequencies , @xmath1152 , since it defines the long time average of @xmath1153 .
@xmath1154 equations ( [ 4.3],[4.7b],[4.9],[4.10 ] ) form the backbone of any linear response theory ( cf .
@xcite ) . with these
we can now write down the expressions for the physical quantities considered in sect .
[ bas ] .
the local charge density , @xmath91 , can be related to local electric potentials , @xmath1155 , by the charge response function , @xmath1156 , @xmath1157 charge response and conductivity ( eq . ( [ 2.7 ] ) ) are related due to the continuity equation @xmath1158 the particle density @xmath65 couples to the local chemical potential @xmath1159 in the same way as the local charge @xmath1160 couples to the local electric potential @xmath1155 and thus the thermodynamic charge response reads @xmath1161 while the dynamic charge response reads @xmath1162 note , that the equilibrium charge response @xmath1163 is not equal to the @xmath100 limit of the dynamical charge response @xmath1156 .
this can be also seen when considering the response to homogeneous potentials . while a homogeneous electric potential does not create a charge response , and hence @xmath1164 , shifting the overall chemical potential by a constant usually leads to a change in the particle density , i.e. @xmath1165 . according to the general scheme described above
, the conductivity tensor appears as the dynamic current correlator @xmath1166 and the relation in eq .
( [ 2.9a ] ) can be viewed as a special case of eq .
( [ 4.9 ] ) .
the corresponding global conductivity tensor reads in the thermodynamic limit @xmath1167 @xmath1168 where @xmath116 denotes the velocity operator of electrons .
more delicate is the representation of the global conductance coefficients @xmath1169 , since one has to know the local potential distribution within the conductor as a functional of applied electro - chemical potentials @xmath75 and we do not go into the detail of this interesting problem .
however , due to the continuity equation the situation simplifies for the d.c .
situation : the current in any lead is independent of the actual potential distribution @xmath1155 within the conductor ( cf .
@xcite ) and @xmath78 is determined by the dynamical correlator of asymptotic current operators which are the time derivatives of the total charge operator , @xmath1170 , @xmath1171 where omission of the argument @xmath1172 $ ] refers to the d.c .
situation . assuming additive electron hamilton operators like the ones introduced in sect .
[ mod ] the correlators can be written in closed form with the help of the one - particle resolvent ( green s function in real space representation ) @xmath1174 of the corresponding one - particle hamiltonian @xmath198 .
it defines a spectral resolution via @xmath1175 with the help of the fermi distribution function @xmath1176 ( where @xmath63 denotes the chemical potential ) average values can be calculated by @xmath1177 the canonical correlator reads @xmath1178 and identifying @xmath534 with @xmath1179 yields @xmath1180 for the dynamical correlator , provided the equilibrium mean values of @xmath636 or @xmath1132 vanish .
since the fermi function is a step function at @xmath126 the general structure @xmath1181 occurring in eqs .
( [ 4.19][4.21 ] ) allows to conclude e.g. for the canonical correlator @xmath1182 a similar equation holds for the dynamic correlator .
therefore , for additive ( non - interacting ) fermion systems we can always restrict the discussion to the zero temperature limit . in the zero frequency limit
further simplifications are possible .
for example , any auto - correlator is essentially determined by states at the fermi level @xmath213 , @xmath1183 which shows that the longitudinal conductivity is a fermi level quantity . to simplify other correlators
it is convenient to use the equation of motion on the level of the one - particle resolvent operator @xmath1184 an important example for an ( essentially ) fermi level quantity is the zero frequency dynamic correlator of two operators one of which is a time derivative @xmath1185 from eq .
( [ 4.26 ] ) one can conclude that the conductance coefficients are determined by asymptotic green s functions , and it is tempting to formulate a scattering theoretical description .
this is indeed possible by decomposing the total hamiltonian into two parts , one of which describes the leads , @xmath1189 , and a second part which describes the conductor plus its coupling to the leads , @xmath1190 , @xmath1191 the total resolvent @xmath1192 can then be decomposed according to dyson s formula on introducing the @xmath341-matrix of scattering theory @xmath1193 the eigenstates of the leads are scattering states , denoted as @xmath1194 , where @xmath69 labels the lead and @xmath275 the quantum number ( channel ) of a scattering state at energy @xmath208 .
the matrix elements @xmath1195 define transmission amplitudes to scatter from channel @xmath275 in lead @xmath69 to channel @xmath122 in lead @xmath124 and form the _ matrix of transmission amplitudes _ @xmath1196 .
applying eq .
( [ 4.25 ] ) and eq . ( [ 4.26 ] ) and using the fact that the asymptotic charge operators are nothing but @xmath19 times the projector onto the hilbert space of the leads , the important bttiker formula ( @xmath1197)@xcite @xmath1198 can be obtained .
fyodorov , a.d .
mirlin , phys .
67 , 2405 ( 1991 ) ; phys .
69 , 1093 ( 1992 ) ; int .
b 8 3795 ( 1994 ) ; phys .
b 51 , 13403 ( 1995 ) ; a.d .
mirlin y.v .
fyodorov , phys .
72 526 ( 1994 ) ; j. phys .
france i , 4 , 665 ( 1994 ) .
figure 4 : universal conductance fluctuations as a function of applied magnetic field .
different curves correspond to gold - nanowires of different length .
the average value of conductance was shifted for convinience ( after @xcite ) .
figure 5 : hopping transport in amorphous silicon .
the logarithm of the conductivity is plotted versus @xmath49 showing mott s law .
the different curves are for different deposition temperatures ( after @xcite ) .
figure 7 : diffusion path in a disordered system .
the electron propagates in both directions ( time reversal symmetry ) . due to constructive interference
the quantum return probability is twice as great as the classical return probability .
figure 8 : the resistance curves of a thin mg - film ( upper set of curves ) as function of magnetic field ( magneto - resistance ) .
the weak localization occurs for zero field . after a superposition with @xmath1200 atomic layer of au the magneto - resistance changes drastically .
the au introduces a rather pronounced amount of spin - orbit scattering which rotates the spin in the complementary scatterd waves .
this changes the interference from constructive to destructive ( after @xcite ) .
figure 10 : the quantum hall effect : the hall resistance @xmath1201 and the magnetoresistance @xmath1202 as function of the magnetic field measured in medium mobility gaas heterostructures at @xmath36mk ( after @xcite ) .
figure 13 : schematic view of energy branches for semi - infinite leads . due to transversal quantization
a discrete number of modes exist ( characterized by a wave number @xmath69 ) for a given value of the fermi energy @xmath1203 .
figure 16 : the scattering element of a 2d network . for unit amplitude on the ingoing link from the left and all other ingoing links being empty
the amplitudes on the outgoing links are @xmath1204 , @xmath1205 and @xmath298 , respectively .
figure 20 : qualitative behavior of the @xmath122-function for disordered electrons .
for @xmath572 the full line corresponds to absence of magnetic fields and spin - orbit scattering while the dotted line corresponds to the presence of spin - orbit scattering . in @xmath599
a ld transition occurs quite generally at some critical value @xmath1035 which forms an unstable fixed point of the flow determined by the @xmath122-function .
figure 22 : qualitative behavior of multifractal exponents .
the top figure shows the exponents @xmath715 corresponding to moments of a multifractal distribution , the middle shows the resulting generalized dimensions @xmath1207 and the bottom shows the @xmath720-spectrum being the legendre transform of @xmath715 .
@xmath720 describes the scaling of the whole distribution .
figure 23 : qualitative behavior of the conductance distribution in a metal . within a range of width @xmath1208 around the average value @xmath745
the distribution is gaussian with universal variance of order @xmath15 . outside this region
the distribution develops log - normal tails .
figure 24 : the scaling function for a 3d quantum hall system showing a ld transition .
@xmath1209 denotes the scaling variable as a function of energy @xmath208 and system size @xmath315 .
@xmath1210 denotes the correlation length obtained from a fitting procedure .
different data symbols correspond to different energies and system sizes ( after @xcite ) .
figure 25 : qualitative picture visualizing the explanation of the quantum hall effect being due to a sequence of localization - delocalization transitions occurring at the landau energies .
@xmath1211 denote the density of states , the dissipative conductivity and the hall conductivity , respectively . in finite systems
the range of extended states ( grey ) on the energy scale has finite width .
this width is believed to shrink to zero in the thermodynamic limit .
figure 26 : squared amplitudes of a critical wave function for a system of linear size @xmath1212 magnetic lengths . the corresponding @xmath976 spectrum ( @xmath1213 ) together with the parabolic approximation ( @xmath1214 ) and a semi - elliptic approximation ( )
are also shown ( after @xcite ) .
figure 27 : histogram of the logarithm of squared amplitudes shown in fig . 26 .
the continuous curve is the distribution function following from the corresponding @xmath976 spectrum with @xmath1215 ( after @xcite ) .
figure 29 : histogram of the logarithm of squared amplitudes shown in fig .
the continuous curve is the distribution function following from the parabolic approximation to the @xmath976 spectrum with @xmath1216 .
the inset shows the histogram for the squared amplitudes itself ( after @xcite ) . | = 0.75this review is intended to give a pedagogical and unified view on the subject of the statistics and scaling of physical quantities in disordered electron systems at very low temperatures .
quantum coherence at low temperatures and randomness of microscopic details can cause large fluctuations of physical quantities . in such mesoscopic systems
a localization - delocalization transition can occur which forms a critical phenomenon .
accordingly , a one - parameter scaling theory was formulated stressing the role of conductance as the ( one - parameter ) scaling variable .
the localized and delocalized phases are separated by a critical point determined by a critical value of conductance .
however , the notion of an order parameter was not fully clarified in this theory .
the one - parameter scaling theory has been questioned once it was noticed that physical quantities are broadly distributed and that average values are not characteristic for the distributions .
based on presently available analytical and numerical results we focus here on the description of the total distribution functions and their flow with increasing system size .
still , one - parameter scaling theory does work in terms of typical values of the local density of states and of the conductance which serve as order parameter and scaling variable of the localization - delocalization transition , respectively . below a certain length scale ,
@xmath0 , related to the value of the typical conductance , local quantities are multifractally distributed .
this multifractal behavior becomes universal on approaching the localization - delocalization transition with @xmath0 playing the role of a correlation length .
# 1#2#1 ; # 2 # 1#2 ( # 1 # 2 ) # 1#2 ( # 1 # 2 ) z * s * = 0.7 |
kamioka observatory , institute for cosmic ray research , university of tokyo , is located in kamioka - town in the northern part of the gifu prefecture in japan .
the location of the laboratories is under the peak of mt .
ikenoyama providing 2,700 meters water equivalent ( or 1000 m of rock ) overburden .
this observatory was established in 1995 to push forward with the super - kamiokande experiment and has become a world frontier center of neutrino physics .
many studies complementary to the flagship sk are also ongoing as illustrated in figure [ map ] .
experiments to search for dark matter and double beta decay are very active now .
further a new 3 km gravitational wave telescope is being constructed and the next generation flagship experiment : hyper - kamiokande is seriously discussed in japanese physics community .
super - kamiokande ( sk ) is a water cherenkov detector containing 50,000 tons of pure water viewed by 11,129 ( inner detector:32kton ) and 1,885 ( outer detector:18kton ) photomultiplier tubes ( pmts ) .
a broad range of energies is studied with the sk detector , from a few mev up to @xmath6(tev ) . in the low energy region below @xmath720mev ,
solar neutrino interactions are detected by neutrino - electron scattering . at higher energies , atmospheric neutrino interactions covering 5 orders of magnitude
are detectable by neutral- and charged - current neutrino - nucleon interactions . in this presentation , we focus on the lower energy physics , solar neutrino .
sk detects solar neutrinos through neutrino - electron elastic scattering , where the energy , direction , and time of the recoil electron are measured . due to its large ( 22.5kton )
fiducial mass sk makes precise measurements of @xmath8b solar neutrinos , including precision information on their energy spectrum and its time variation .
since sk - iii started , we have made many efforts to reduce backgrounds .
the most serious background comes from the beta decay of @xmath9bi , which is produced in the decays of radon in the air and detector materials as well as from radium in the water . in order to reduce the @xmath9bi background ,
the sk water system has been upgraded .
first , the cooling power of the heat exchanger for the supply water was increased so as not to induce convection in the tank , which transports radon near the pmts into the fiducial volume .
second , new membrane degasifier units were added to increase the efficiency of radon removal from the supply water .
third , the water flow in the detector was precisely investigated and optimized to reduce the background contamination in the fiducial volume as much as possible . during the sk - iv period
we have introduced a precise temperature control system for the inlet water to further reduce convection in the water . as a result of these improvements ,
the background rate in the lower energy region in sk - iv has been reduced by a factor of more than three compared to the sk - i period .
further , it has allowed a lower analysis energy threshold . until the end of march 2012 ,
1069.3 days of sk - iv solar neutrino data for analysis was taken .
a clear solar neutrino signal in the 3.5 - 4.0mev kinetic energy region was observed at more than @xmath10 .
in addition , we developed a new analysis method for these low energy regions based on the amount of multiple coulomb scattering of electrons using the pmt hit pattern of the cherenkov cone .
lower energy electron hit patterns ( i.e. the @xmath9bi background ) are more isotropic than those of the higher energy solar neutrino recoil electrons .
this analysis improves the statistical uncertainty of the number of signal events by about 10% .
we use this new method for recoil energies below 7.5mev .
figure [ sk : sol03 ] shows sk - i to sk - iv combined energy spectrum with expectations from the solar global and solar+kamland as well as flat reduction of the neutrino spectrum .
the vertical axis shows the ratio of the observed energy spectrum to the expectation from the unoscillated mc simulation assuming a @xmath8b flux of 5.25@xmath11@xmath12s@xmath13 .
the combined energy spectrum is consistent with the flat prediction , but the level of favoring flat over the upturn is [email protected]@xmath14 level .
b solar neutrino spectrum .
predictions for ( 1 ) @xmath15 and @xmath16ev@xmath17(blue ) , ( 2 ) @xmath18 and @xmath19ev@xmath17(light blue ) , ( 3 ) flat probability ( black ) , and ( 4 ) flat probability and @xmath20 shape for pure @xmath21 scattering ( gold ) are also shown . , width=377 ] concerning differences in the day and night fluxes the expected flux asymmetry , defined as @xmath22 , is about @xmath23 based on current understanding of neutrino oscillation parameters .
although this is not a large effect , long term observations by sk enable discussion of a finite value of the day - night asymmetry .
the @xmath24 value using the combined sk - i to sk - iv data is @xmath25 , which is a 2.3@xmath14 difference from zero .
figure.[sk : sol05 ] shows @xmath24 as a function of @xmath26 together with the expectation .
the observed @xmath24 is consistent with the expectation using the best fit @xmath26 from both kamland and the global solar analysis . .
a thick black line shows the result of the sk fit and the surrounding gray band indicates statistical and systematic uncertainty ( thin black lines for statistical error only ) .
the red curve shows the expectation assuming @xmath27 .
the light blue and light green vertical bands show the allowed range of @xmath26 from the kamland reactor and solar global , respectively .
, width=377 ] we performed a global solar neutrino oscillation analysis including all sk data ( sk - i@xcite , sk - ii@xcite , sk - iii@xcite , and sk - iv ) as well as the most recent results from sno@xcite , the radiochemical experiments@xcite and the latest @xmath28be flux measurement from borexino@xcite .
this analysis was then compared and combined with the reactor neutrino results from kamland@xcite .
the obtained range of sin@xmath29 from the solar global analysis is @xmath30 , while the value of the kamland reactor analysis is @xmath31 . in the combined fit sin@xmath29
was found to be @xmath32 , which is a roughly 2 @xmath14 hint that @xmath33 is different from zero .
this has been already discussed prior to 2010@xcite .
after 2011 , the t2k and reactor experiments presented indications and later evidence for a finite @xmath33 . the combination of their measurements yields @xmath34 and the result of the combined analysis of the global solar and kamland reactor data is consistent with this value .
assuming @xmath35 is fixed at 0.025 , the obtained parameters from the global solar analysis are @xmath36ev@xmath37 and sin@xmath38 . comparing these values with those from kamland , ( @xmath39ev@xmath37 and sin@xmath40 ) , there is a 1.8 @xmath14 level tension in the @xmath41 results . combining the global solar data with kamland ,
the oscillation parameters become @xmath42ev@xmath37 and sin@xmath43 .
among the most plausible dark matter candidates , weakly interacting massive particles ( wimps ) , which are expected to couple to ordinary matter primarily through the weak force , can be detected directly through observation of nuclear recoils produced in their elastic scattering interactions with detector nuclei @xcite .
the most convincing signature of the wimps appears in the directions of nuclear recoils .
it is provided by the earth s velocity through the galactic halo ( @xmath7230 km / s ) and the distribution of the nuclear recoil direction shows a large asymmetry .
hence , detectors sensitive to the direction of the recoil nucleus would have a great potential to identify wimps .
time projection chambers ( tpcs ) with fine spacial resolutions are among such devices , and we are developing a micro - tpc @xcite , which can detect three - dimensional fine tracks of charged particles .
since the energy deposits of wimps to nuclei are only a few tens of kev and the range of nuclei is limited , the micro - tpc should be operated at low pressures .
we are interested in operating the micro - tpc with cf@xmath44 because @xmath45f has a special sensitivity to sd interactions for its unique spin structure .
we performed our first dark matter search in 2010 at kamioka laboratory using newage-0.3a detector @xcite .
the exposure was 0.524kg@xmath5days .
the gas pressure was 152torr and the energy threshold was 100kev .
then , second search was conducted from january 2012 to may 2012 .
this exposure was 0.140kg@xmath5days with the same gas pressure as before and the energy threshold was lowered to 50kev .
figure [ newlimit ] shows the energy spectra and exclusion limits obtained through these searches .
we set a limit of 800 pb for 100 gev wimps , however it is indispensable to lower the threshold and background . in this presentation
, we report on the study of lowering the energy threshold via operating the detector in lower pressure .
the low pressure gas study was performed using newage-0.3b detector in the surface laboratory .
a schematic drawing of newage-0.3b is shown in figure [ newdet ] the size of the @xmath46-pic ( no.sn070309-3 ) is @xmath47@xmath48 cm , and the number of pixels is 768@xmath49768 .
we supply a positive high voltage to the anode electrodes of the @xmath46-pic ( @xmath50 ) .
a large area gem , consisting of a 50mm - thick polyimide layer sandwiched between 5mm - thick copper electrodes is set at 5 mm above the @xmath46-pic to obtain sufficient gain .
the size of the gem is 23cm@xmath4928 cm , and this gem is segmented into 8 parts to reduce the risk by discharges .
we supply negative high voltages at the top ( gas - volume side : @xmath52 ) and bottom ( @xmath46-pic side : @xmath53 ) of the gem .
the drift length is 50 cm and the detection volume is surrounded by a drift cage to form the electric field .
the size of the drift cage is @xmath54 cm .
copper electrodes are formed on the inner surface of the drift cage by every 1 cm .
we supply a positive high voltage ( @xmath55 ) to the drift plane .
these detector components are set in a chamber made of sus304 stainless steel whose inner surface was electro - polished to reduce out - gassing .
the chamber was filled with cf@xmath44 gas at 76torr .
a dedicated electronics system was used for the measurement .
details of the electronics are described in @xcite and the references therein .
three - dimensional tracks were detected by the discriminated hit signals of 768 anode strips and 768 cathode strips and the anode - cathode coincidence - timing .
energy of each track was determined by the summed waveform .
-pic and gem , is 5 mm .
, width=302 ] as a beginning , the gain dependence on the bias voltages was measured to optimize operation parameters for cf@xmath56 at 76torr .
we had operated the previous newage-0.3a at a gas gain of 700 with 152torr gas .
accordingly , the optimum gas gain should be 1400 at 76torr , because the primary electron density became 1/2 of the previous detector . in figure
[ gain ] , the obtained gas gain against @xmath57 , @xmath58 , and the induction field ( @xmath59 ) .
the other two parameters are fixed in each plot .
the gas gain increased with @xmath57 until the voltage was limited by discharged .
as for @xmath60 and @xmath59 , gain curves saturate below 2000 .
this gain suppression is due to the operation in low gas pressure .
the mean free path of an electron in 76torr of cf@xmath56 gas is several mm , and this length is close to the thickness of the gem ( 50 mm ) , therefore the gas amplification at the gem is to be suppressed .
although it turned out that the gain is saturated , we have achieved the required gas gain of 1400 .
the optimized parameters are @xmath61v , @xmath62v , @xmath63v and @xmath64kv . , @xmath60 and @xmath59 ( or @xmath65.,width=188 ] then , the angular resolution of newage-0.3b was evaluated using nuclear tracks by neutrons from a @xmath66cf source .
we determined the angular resolution by comparison of measured and simulated distributions of the recoil angle .
the results are shown in figure [ angle ] .
we plan to move the newage-0.3b detector to kamioka underground in 2013 . in order to operate it with lower pressure gas ,
we need an even higher gas gain . as shown in above the gain
is saturated at around 2000 in the current set up , therefore we have started to test a new gem which has a thickness of 100 mm instead of 50 mm .
we are also improving the data acquisition system ( daq ) for recording the time - over - threshold ( tot ) of the @xmath46-pic signals in order that we can measure the energy deposition at each strip .
over the past 10 years , some experiments indicate a possible wimp signal @xcite with a lighter mass of @xmath67@xmath68 and with a spin - independent cross section of the order of @xmath69 .
these positive signals have come predominantly from experiments without the capability to discriminate between electromagnetic and nuclear recoils , while other experiments that have this ability have excluded light wimps at these cross - sections @xcite . in this presentation , a search for light wimps using a 5591.4kg@xmath5day exposure of the xmass experiment without nuclear recoil discrimination is presented .
xmass is a single phase liquid xenon scintillator detector containing 1050 kg of xe in an ofhc copper vessel . as shown in figure .
[ xmass_detector ] xenon scintillation light is collected by 630 hexagonal and 12 cylindrical inward - pointing hamamatsu r10789 series photomultiplier tubes ( pmts ) arranged on an 80 cm diameter pentakis - dodecahedron support structure within the vessel to give a total photocathode coverage of 62.4% .
these pmts view an active target region containing 835 kg of liquid xenon . in order to monitor the pmt stability and measure the trigger efficiency , eight blue leds with teflon diffusers are mounted to the support structure .
there are six leds arranged along the equator and one each at the top and the bottom of the pentakis - dodecahedron . to shield the scintillator volume from external gammas , neutrons , and muon - induced backgrounds , the copper vessel is placed at the center of a @xmath7010 m @xmath49 11 m cylindrical tank filled with pure water .
this volume is viewed by 72 hamamatsu r3600 20-inch pmts to provide both an active muon veto and passive shielding against these backgrounds .
this is the first water cherenkov shield used in a dark matter search experiment . to perform energy and position reconstruction calibrations
a portal has been prepared along the central vertical axis ( @xmath71-axis ) of the pmt support structure through which an ofhc copper rod can be inserted directly into the target volume .
thin cylindrical calibration sources containing either of @xmath72fe , @xmath73co , @xmath74cd , or @xmath75am are placed at the tip of this rod to perform detector calibrations .
a more detailed description of the xmass detector is presented in @xcite .
the data used for this analysis , corresponding to 6.70 days of livetime , was taken in february 2012 with a low trigger threshold of four pmt hits @xcite . using 122kev gammas from the @xmath73co calibration source
the xenon light yield was found to be 14.7photoelectrons / kevee .
this large light yield allows the analysis threshold to be lowered sufficiently for sensitivity to low mass wimps . in order to achieve optimal sensitivity ,
the entire detector volume is used because fiducialization is increasingly difficult at these low energies .
a sequence of data reduction is applied to remove events caused by the tail of the scintillation light distribution after energetic events ; ( 1 ) events triggered only with the liquid xenon detector are selected , ( 2 ) events that occurred within 10ms of the previous event are rejected , and ( 3 ) events whose timing distribution has an rms greater than 100ns are removed .
the last cut is applied to remove cherenkov events originated from @xmath76k contamination in the pmt photocathodes ; events with more than 60% of their pmt hits occurring within the first 20ns of the event window are removed as cherenkov - like .
figure [ fig : spe ] shows simulated wimps energy spectra overlaid on the observed spectrum after the data reduction was applied .
wimps are assumed to be distributed in an isothermal halo with @xmath77km / s , a galactic escape velocity of @xmath78km / s , and an average density of 0.3gev / cm@xmath4 . in order to set a conservative upper bound on the spin - independent wimp - nucleon cross section
, the cross section is adjusted until the expected event rate in xmass does not exceed the observed one in any energy bin above the analysis threshold .
it is chosen as the energy at which the trigger efficiency is greater than 50% for 5gev wimps and corresponds to 0.3kevee . the resulting 90% confidence level ( c.l . )
limit is shown in figure [ fig : limit ] .
the impact of the uncertainty from @xmath80 @xcite is large in this analysis , so its effect on the limit is shown separately in the figure .
are taken into account in the xmass 90% c.l .
limit line .
the effect of the @xmath80 uncertainty on the limit is shown in the band .
limits from other experiments and favored regions are also shown @xcite.,width=302 ] after careful study of the events surviving the analysis cuts , their origins are not completely understood .
contamination of @xmath81c in the gore - tex@xmath82 sheets between the pmts and the support structure may explain a fraction of the events .
light leaks through this material are also suspect .
nonetheless , the possible existence of a wimp signal hidden under these and other backgrounds can not be excluded .
although no discrimination has been made between nuclear - recoil and electronic events , and many events remain in the analysis sample , the present result excludes part of the parameter space favored by other measurements @xcite when those data are interpreted as a signal for light mass wimps .
the fiducial volume analysis is ongiong for more sensitive searches .
in addition , we are working on modifications to the inner surface of xmass , especially around the pmts , to improve the detector performance . finally ,
a r&d project aiming at modification to single phase liquid xenon tpc with charge readout has been started .
9 j. hosaka et al . 2006 _ phys . rev .
d _ * 73 * 112001 . | at kamioka observatory many activities for low energy rare event search are ongoing .
super - kamiokande(sk ) , the largest water cherenkov neutrino detector , currently continues data taking as the fourth phase of the experiment ( sk - iv ) . in sk - iv
, we have upgraded the water purification system and tuned water flow in the sk tank .
consequently the background level was lowered significantly .
this allowed sk - iv to derive solar neutrino results down to 3.5 mev energy region . with these data ,
neutrino oscillation parameters are updated from global fit ; @xmath0 , @xmath1 , @xmath2 .
newage , the directional sensitive dark matter search experiment , is currently operated as `` newage-0.3a '' which is a @xmath3 m@xmath4 micro - tpc filled with cf4 gas at 152 torr .
recently we have developed `` newage-0.3b '' .
it was succeeded to lower the operation pressure down to 76 torr and the threshold down to 50 kev ( f recoils ) .
xmass experiment is looking for scintillation signals from dark matter interaction in 1 ton of liquid xenon .
it was designed utilizing its self - shielding capability with fiducial volume confinement .
however , we could lower the analysis threshold down to 0.3 kevee using whole volume of the detector . in february 2012 ,
low threshold and very large exposure data ( 5591 kg@xmath5days ) were collected . with these data ,
we have excluded some part of the parameter spaces claimed by dama / libra and cogent experiments . |
xb1254690 is a persistently bright , low - mass x - ray binary ( lmxb ) .
it was identified with a faint blue star ( grmus , _ v _ = 19.1 ) , which exhibited the bowen blend of niii and ciii in emission @xcite .
the object produced type 1 x - ray bursts @xcite , indicating the presence of a neutron star as the compact object .
dips of up to 95% of the 110 kev flux , with a recurrence period of _ @xmath4
_ = 3.88 @xmath5 0.15 hr and lasting @xmath0 0.8 hr per cycle , were discovered in _
data by @xcite .
the dips are caused by obscuration of the central source by a bulge on the outer edge of the accretion disc , implying a moderately high inclination . from _
v_-band observations of the optical counterpart , @xcite determined an optical ephemeris where minimum light occurs at _ @xmath6 _ = jd 2,445,735.693 ( 4 ) + 0.163890 ( 9 ) days , equivalent to a period of _ @xmath7 _ = 3.9334 @xmath5 0.0002 hr .
this broad optical modulation is probably due to the changing visibility of the heated face of the secondary star in the system , with an additional contribution from the x - ray heated bulge , although other explanations such as an asymmetrical disc or occultation of the disc by the companion can not be ruled out . within the errors ,
the optical and x - ray period measurements are consistent , with later x - ray observations indicating that the optical modulation provides a more accurate measurement of the orbital period @xcite .
a short section of simultaneous x - ray and optical coverage showed that the optical minimum occurs @xmath0 0.16 in phase after the centre of the x - ray dips @xcite .
simple geometrical modelling is possible by considering the x - ray illuminated area of the companion star , disc radius , disc opening angle , and inclination at given distances which are compatible with the observed optical flux and the amplitude of the light curve .
this leads to constraints on the source inclination of 65@xmath873@xmath8 , and on the distance of 8 - 15 kpc @xcite .
the first mass constraints for this system were based upon a unique relation between the orbital period and the mass of the companion , assuming that the companion is a low mass , zero - age main sequence star filling its roche lobe @xcite . using this approximation , @xcite estimate the mass of the companion to be in the region of @xmath9 .
the first kinematic mass constraints were proposed by @xcite , who obtained intermediate resolution ( @xmath0 3 ) spectra using the 4 m telescope at the cerro tololo inter - american observatory . assuming that velocities derived from the heii @xmath14686 emission line centroid represent the motion of the compact star and not , for example , the streaming of gas between the stars or out of the system
, then approximate stellar masses can be derived using the mass function .
the result is consistent with a low - mass main - sequence star ( _ m _ @xmath10 0.5 - 0.8 @xmath11 ) and a canonical 1.4 @xmath11 neutron star . in the remainder of this paper
we will attempt to confirm and better constrain these mass estimates using high - resolution blue optical spectroscopy .
we will measure the wings of the heii @xmath14686 emission line to gain an estimate for the compact object velocity ( section 3.3 ) , and use the bowen fluorescence technique @xcite to measure the donor star velocity ( section 3.4 ) .
we also obtained simultaneous x - ray observations ( see section 3.1 ) in order to re - establish the ephemeris of @xcite .
this can not be extended to the observations described herein because the propagated uncertainty in phase is now comparable to the orbital period itself .
we have obtained simultaneous optical and x - ray data for grmus / xb1254690 .
the journal of observations is given in table [ tabobsmus ] .
.log of observations of grmus / xb1254690 [ cols="<,^,^,^",options="header " , ] phase - resolved spectroscopy of grmus was obtained from 26 to 28 may 2004 using the fors2 spectrograph mounted on the vlt / ut4 at eso s paranal observatory . during each night
we observed grmus for @xmath10 one full orbit , resulting in a total of 38 spectra with an integration time of 900s each .
we used the 1400v volume - phased holographic grism , and a slit width of 0.7@xmath12 , giving a wavelength coverage of @xmath1@xmath14513 - 5814 with a resolution of 70 kms@xmath13 ( fwhm ) .
the seeing during these observations varied between 0.5 and 2.1 arcsec .
the slit was orientated at a position angle of 88.15@xmath8 to include a comparison star in order to calibrate slit losses .
we observed the flux standard feige 110 with the same instrumental set - up in order to correct for the instrumental response .
he , ne , hg and cd arc lamp exposures were taken during daytime for the wavelength calibration scale .
we de - biased and flat - fielded all of the images and used optimal extraction techniques to maximise the signal - to - noise ratio of the extracted spectra @xcite .
the pixel - to - wavelength scale was derived through polynomial fits to a large number of identified reference lines resulting in a dispersion of 0.64 pix@xmath13 .
velocity drifts due to instrumental flexure ( found to be always @xmath145 kms@xmath13 ) were corrected by cross - correlation with the sky spectra .
simultaneous x - ray data from xb1254690 were obtained using the proportional counter array ( pca ) on board the _ rossi x - ray timing explorer _ ( _ rxte _ ) . the proportional counter array ( pca ; for a detailed description , see @xcite ) onboard the _ rxte _ satellite consists of an array of 5 co - aligned proportional counter units ( pcus ) that are sensitive to photons of energy 2 to 60 kev with a total collecting area of 6500 @xmath15 .
we obtained 36 ks of total coverage during our optical observations . for our analysis
we only used the data from the standard 2 configuration , with a time resolution of 16 seconds .
the ephemeris of @xcite is now out of date because the propagated uncertainty in phase is over a factor of 2 greater than the orbital period .
we therefore need to make use of our contemporaneous _ rxte _
observations to set a new zero point using the x - ray dips ( note that we still use the @xcite orbital period ) .
this is not quite as simple as it would first appear since there are large variations in dip depth and structure that are apparent from observation to observation , as well as cycle to cycle @xcite .
additionally , a note of caution must be sounded since in a number of earlier _ rxte _ and _ bepposax _ observations @xcite , x - ray dipping was not observed at all , with an accompanying decrease in the mean optical variability of the source . with the combination of these effects , it was suggested that the vertical structure on the accretion disc edge , associated with the impact point of the accretion stream , had decreased in angular size at the time of these earlier observations .
the cause of this intermittent reduction in opening angle is not understood .
our observations include just one deep dipping episode ( fig .
[ xray ] ) , co - incident with our first night of optical observations .
xb1254690 typically exhibits ` shoulders ' of dipping , i.e. regions both before and after the deep dipping in which there is a small decrease of intensity .
each shoulder of the dip lasts @xmath0650925 seconds , during which the total count rate in the 2 - 25 kev band drops by @xmath015% with no strong changes in the hardness ratio .
the shoulder is followed by an interval of strongly variable deep dipping lasting @xmath01100 seconds @xcite , with associated strong changes in the hardness ratio ( fig . [ hard ] ) .
the hardening of the spectra in the dips is most likely to be due to absorption by cold matter .
our observation of a dipping episode opens with the system already in the first shoulder phase , the onset of deep dipping occurring approximately 200250 seconds into the observation ( before a brief return of the source intensity to the non - dip level ) , and then continuing to capture the full egress from deep dipping , into the shoulder and then returning to the continuum level .
note that the return of the source intensity to the non - dip level during this episode is not unusual , and the light curve can in fact be interpreted as superpositions of small micro - dips ( e.g. * ? ? ?
unfortunately , the lack of information regarding the full ingress to deep dipping through the preceding shoulder introduces a degree of uncertainty into our estimate of the dip centre .
for example we could potentially be observing only a small section of an extended deep dipping episode ( e.g. * ? ? ?
* report dips lasting for @xmath040 minutes ) .
if this is the case , the absence of the beginning of the dip will cause the estimate for dip centre to be incorrect .
nevertheless , we do have cause for optimism . the deep dipping which we observe
is approximately 1100 seconds in length , equivalent to the typical period of a dipping episode in xb1254690 ( see above ) .
this is a strong indicator that we have full , or nearly - full , coverage of the deep dipping phase .
we estimate the dip centre by eye , taking the mid - point between the onset of and exit from deep dipping ( which can be determined with the help of the associated hardness ratio changes , see fig . [ hard ] ) .
this occurs at @xmath16 = jd 2453151.647 ( 3 ) . from @xcite
we know that the dip centre has an average orbital phase of 0.84 ( with phase 0 occurring at optical minimum ) .
we therefore estimate a new value for @xmath17 of jd 2453151.509 ( 3 ) .
there are problems with the reliability of an ephemeris generated from just a single dipping episode , not least because we do not have full coverage either side of the dip .
it is therefore hard to estimate the dip centre with absolute confidence .
however , we are encouraged by the fact that our dip lies within 0.05 phase of the 1999 dips ( the propagated phase uncertainty from these dips is of the order @xmath00.6 ) reported by @xcite .
more importantly , the continuum lightcurve from the resulting ephemeris possesses the same phasing relation reported by @xcite ( fig .
[ lcurve ] ) . in light of this evidence
we are confident that we have identified the dip centre correctly and we will use this zero - point to calculate orbital phases throughout the remainder of this paper .
the continuum subtracted average of the 38 individual optical spectra , corresponding to an exposure time of nearly 10 hours , is given in fig .
the overall spectrum resembles that of other lmxbs , with high excitation lines superposed on a blue continuum .
this is typical of strongly x - ray heated accretion discs ( e.g. * ? ? ?
the blue end of the optical spectrum in this system is dominated by the bowen emission blend of ciii and niii .
heii @xmath14686 is also seen strongly in emission , in addition to weaker h@xmath18 and heii @xmath15411 .
the absorption component longward of h@xmath18 is observed in a few other lmxbs such as x1822371 @xcite , xtej2123058 @xcite , gx3394 @xcite and mmser @xcite .
the narrow emission feature at 5577 is [ oi ] , a strong night - sky line useful for confirming the veracity of our wavelength calibration .
the broad features between 49005300 remain unidentified .
the absence of these features in our comparison star lead us to believe that they are real .
we also see some of these ` bumps ' in other lmxbs ( e.g. gx9 + 9 , x1822371 , v801ara , v926sco ) , and perhaps they are blends of a number of different weak emission lines
. part of this could be hei @xmath14922 and @xmath15016 , and also feii @xmath14924 , @xmath15018 , @xmath15169 and @xmath15297 .
there are also several oii and oiii lines between 48915006 which may explain the large bump at @xmath04930 .
note also that the bowen blend ( and to a lesser extent the heii @xmath14686 emission line ) appears to be sitting upon a broad emission ` hump ' .
this has been noted in other lmxbs ( e.g. v801ara ; * ? ? ?
* ) , and attributed to a blend of feii emission lines @xcite . in fig .
[ trail ] , we display the bowen blend and heii @xmath14686 in the form of a trailed spectrogram .
all of the individual spectra were first normalized to the continuum using a third - order polynomial fit , and then phase - folded into 15 bins .
the spectra were then smoothed using a gaussian filter with a fwhm of two pixels .
we see a clear s - wave structure modulated on the orbital period in the heii @xmath14686 emission .
the blue - to - red crossing phase is @xmath00.6 and the semi - amplitude @xmath0250 kms@xmath13 .
this behaviour indicates a close association with the accretion disc or compact object .
there is also an additional , weaker , component present .
again , this is modulated on the orbital period , approximately a quarter cycle out of phase with the more prominent component .
there also appear to be faint traces of two anti - phased sharp components in the bowen blend though the broad underlying component moves in phase with the accretion disc s - wave ( maximum blue - shift at phase 0.25 ) .
the tomograms discussed in section 3.4 will help to characterise these spectral trails more thoroughly .
the h@xmath18 line is shown in the right - hand panel in fig .
[ trail ] , along with the broad emission feature at @xmath04930 .
the absorption component longward of h@xmath18 is present at all phases , and in common with the emission component displays an s - wave approximately in phase with the expected motion of the accretion disc .
the broad feature at @xmath04930 also displays a s - wave .
the heii @xmath15411 emission line is too weak in the individual spectra to produce a trailed spectrogram .
early attempts to determine any kinematic constraints on this system met with limited success .
@xcite noted that the relative weakness of the emission lines in their spectra did not allow for the study of their variation with orbital phase .
however , they did exclude the possibility of orbital motion of heii @xmath14686 ( which requires a source of highly ionising radiation , and is typically seen in accretion discs ) in excess of 300 kms@xmath13 .
later observations by @xcite did result in a measurement of the same heii line s core velocity , which possessed a semi - amplitude of 114 @xmath5 13 kms@xmath13 .
however , this is not a particularly reliable measurement of the compact object velocity .
the line core can often be heavily contaminated by other emission sites ( see fig .
[ profile ] ) , for example from the accretion stream or ` hot - spots ' on the accretion disc itself . instead
, the compact object velocity can be better estimated by examining the wings of the emission line .
these wings find their origin in the high - velocity gas of the inner accretion disc , very close to the compact object itself .
presumably this gas will closely replicate the motion of the compact object , and it is far less likely that alternative emission sites will provide a strong contaminating effect .
however , this method is not immune to contamination and we must be cautious in accepting the results of this analysis . we have followed the double - gaussian technique @xcite to estimate both the velocity semi - amplitude @xmath19 and the systemic velocity @xmath20 of the compact object from heii @xmath14686 .
we employed a double gaussian bandpass with fwhm of 100 kms@xmath13 and gaussian separations of 3001700 kms@xmath13 in steps of 100 kms@xmath13 .
radial velocity curves of line sections with separations of 7001400 kms@xmath13 yield consistent results with a median blue - to - red crossing phase of 0.58 @xmath5 0.03 , median velocity semi - amplitude @xmath19 = 130 @xmath5 16 kms@xmath13 and median systemic velocity @xmath20 = 173 @xmath5 12 kms@xmath13 ( fig .
[ diag ] ) .
the errors are conservatively taken from the maximum error obtained from sine wave fits to the results from each individual line section .
separations below 700 kms@xmath13 suffer badly from contamination by the line core , whilst separations over 1400 kms@xmath13 become corrupted by continuum noise and the bowen emission blend . the small difference in phasing ( 0.08 orbits ; slightly larger than the ephemeris uncertainty of @xmath00.02 orbits ) indicates that even the line wings may be contaminated to some extent by inhomogeneities in the disc .
this contamination will also affect the measurements of @xmath19 and @xmath20 to some extent .
nevertheless , this remains the best method for estimating @xmath19 from our own and other currently available data .
unfortunately , the heii @xmath15411 and h@xmath18 emission lines , which are clearly seen in our averaged spectrum ( fig . [ spec ] ) , are too weak in the individual spectra for a similar analysis .
the fitting procedures are simply dominated by noise .
we can use doppler tomography @xcite to measure the radial velocity amplitudes of any weak , narrow , emission lines that may be present . by mapping the observed data onto a velocity coordinate frame
, doppler tomography makes use of all data at once and can thus be used to search for features that are too weak to be distinguished in each individual spectrum .
one effectively resolves the distribution of line emission in the co - rotating frame of the binary system , providing an excellent tool for identifying the origin and kinematics of the various emission components .
it also allows us to map the emission line distribution from the broader heii and h@xmath18 lines .
secondary star emission is readily identifiable in doppler tomograms since the solid body rotation of the roche lobe is mapped to a corresponding roche lobe area along the positive v@xmath21-axis . for reference , in figs .
[ dheii][dniii ] we overplot the area of the roche lobe and the centres of mass of the donor star ( uppermost cross ) , full system ( map centre ) and the compact object ( lower cross ) assuming a mass ratio of @xmath00.4 ( equivalent to the mass ratio proposed by * ? ? ?
* ) and compact object velocity of 130 kms@xmath13
. additionally , we have overplotted the ballistic accretion stream for such a system .
the circle , centred upon the compact object and with radius 600 kms@xmath13 , indicates the common emission regions from a generic lmxb accretion disc .
we employed a maximum entropy implementation of doppler tomography whereby the data are fitted under the added constraint of maximizing the entropy ( i.e. image smoothness ) of the tomogram @xcite .
this reduces the presence of noise artifacts in the recovered tomograms and allows for a simultaneous fit to a number of heavily blended lines . in fig .
[ dheii ] we display a tomogram of the heii @xmath14686 emission line .
this displays a clear enhancement in the lower left quadrant and an additional component close to the accretion stream .
we do not observe the classic ring - like distribution of the accretion disc , presumably due to the strength of the line core features vastly overwhelming the high velocity inner disc emission .
for this reason we can not be fully confident in the results of the double gaussian technique ( see section 3.3 ) , though this remains at present the best method for estimating @xmath19 .
the enhancement in the ( lower ) left quadrant is also seen in the tomograms of h@xmath18 ( fig .
[ dhb ] ) and heii @xmath15411 ( fig .
[ d5411 ] ) . whilst this phenomenon is a common feature in the remarkable swsex - type systems @xcite ,
it has also been noted in some other lmxbs , notably xtej2123058 @xcite , gx9 + 9 @xcite and j0422 + 32 @xcite .
however , in this case ( and in the case of x1636536 and x1735444 , * ? ? ?
* ) the emission is not quite so localised in the lower - left quadrant , but is present in a broad band on the left - hand side of the maps . rather than being an analogue to the highly magnetic swsex - type systems ,
it is likely that this emission is produced in an extended disc bulge . in most lmxbs
, the heii emission is predominantly produced in a hotspot where the accretion stream impacts the disc ( thus producing a spot in the upper left quadrant of a doppler map , e.g. * ? ? ? * ) or is more evenly distributed around the bulk of the accretion disc . if the emitting gas was symmetrically distributed around the primary , the doppler image would reveal a circular distribution centred on @xmath22 = 0 , @xmath23 = -@xmath19 .
this would have been a useful way of confirming the @xmath19 estimate derived from the wings of the heii @xmath14686 emission line . like heii @xmath14686
, the heii @xmath15411 tomogram also indicates the presence of some diffuse emission at velocities which partially correspond to the roche lobe of the donor star .
however , as this emission is diffuse in the doppler map , extending over a broad range of velocities , it is difficult to directly associate this extended emission with the donor star .
fluorescent emission from the donor in this case should manifest itself as a tightly concentrated spot situated on the positive @xmath23 axis . in fig .
[ dniii ] we present a tomogram of the niii @xmath14640 emission line .
the maximum entropy method of producing doppler tomograms allows for a simultaneous fit to a number of different lines to produce one map .
however , in this case the niii @xmath14640 line strongly dominates over any others in the bowen blend .
we see two regions of enhanced emission .
the brighter of the two spots occurs at a combination of phasing and velocity where we would expect the donor star to be fluorescing .
the small horizontal offset of @xmath22@xmath1035@xmath530 kms@xmath13 is consistent with the errors in our ephemeris ( @xmath5 0.02 in phase ) .
the second spot occurs at a much lower velocity , and is 12 @xmath24 weaker ( relative to the local noise level ) than the first .
it is difficult to provide a truly satisfactory explanation for its origin .
perhaps it traces some form of low velocity outflow ( e.g. disc wind ) , though why it should be preferentially observed at an apparent phase of @xmath00.25 remains a mystery ( orbital phase can be considered to increase in a clockwise - fashion around a doppler map ) . if we presume that the bright spot in fig .
[ dniii ] is caused by the irradiated inner face of the donor star , this is an excellent tracer for the kinematics of the system .
the precise location of the spot can be measured using a two - dimensional gaussian fit ( though bearing in mind that the actual shape of the spot is not expected to be gaussian ) , providing a velocity of @xmath25 = 245 @xmath5 30 kms@xmath13 . using this value
we can shift all of our spectra into the rest - frame of the ( irradiated face of the ) donor . a sharp niii @xmath14640 component is clearly resolved ( fig . [ donorspec ] ) . however , since this is a measure of the motion of the irradiated inner face of the donor star , it only provides us with a lower - limit on the velocity of the centre of mass ( see e.g. * ? ? ?
* ; * ? ? ?
one final piece of information remains to be extracted from the doppler tomograms ; assigning the correct systemic velocity should produce the most sharply defined features in a doppler map .
we used the heii and niii doppler maps to verify the systemic velocity .
the @xmath26 value of the map was calculated for a range of @xmath20 , and the best fit in terms of minimal @xmath26 was achieved for @xmath20 = 192 @xmath5 2 km s@xmath13 for the niii map and @xmath20 = 185 @xmath5 2 km s@xmath13 for heii @xmath14686 .
these are both consistent with , if slightly higher than , the results derived from the heii @xmath14686 wings in section 3.3 .
the complex , asymmetric core profile of the heii emission line could be contaminating one wing more than the other , leading to this small offset .
we heretofore adopt the value for @xmath20 derived from the heii @xmath14686 doppler map .
this should be more reliable than the niii map since only a single , isolated emission line is involved in the fitting process , rather than the more complicated blend of bowen lines .
whilst the systemic velocities inferred from both the wings of the heii @xmath14686 line and the doppler tomograms are self - consistent ( @xmath0 185 kms@xmath13 ) , they are considerably higher than the earlier estimate by @xcite ( 98 @xmath5 10 kms@xmath13 ) . this measurement was obtained from rather low quality data , with poor spectral and temporal resolution .
additionally , they based their measurements on the motion of the heii @xmath14686 line core .
there are a number of different emitting regions which contribute to this line , leading to a complex core structure ( see e.g. figs .
[ profile ] , [ dheii ] ) . a basic measurement of the line centre under the assumption that this will accurately trace
the motion of the compact object will therefore be subject to large systematic errors . in light of this , we believe that our estimate for the systemic velocity will be a better reflection of the true value .
although large , our measurements do seem reasonable in the context of the high galactic latitude ( @xmath27 = -6.42 ) of grmus ; the neutron star is likely to have received a substantial ` kick ' out of the galactic plane at birth .
a number of statistical studies on pulsar velocities have been carried out .
these studies give a mean birth velocity of 100500 km s@xmath13 , with some possessing velocities of @xmath281000 km s@xmath13 @xcite . the niii doppler map ( fig .
[ dniii ] ) appears to display enhanced emission in the region of and at the phasing where we would expect to see the companion star roche lobe , in addition to an abundance of noisy structure in the outer regions of the map .
a useful way to confirm the validity of the doppler map is to observe the behaviour of sharp components in our trailed spectrogram ( fig .
[ trail ] ) .
this is not trivial due to the faintness of grmus compared to a system such as scox-1 , which clearly features narrow bowen lines moving in phase with the donor star @xcite .
4686 emission line and the left hand panel shows the ideal trailed spectrogram reconstructed from the doppler tomogram shown in fig .
[ dheii ] .
orbital phase is plotted along the @xmath29-axis . ]
we can reconstruct an ideal trailed spectrogram from the doppler tomograms to test for systematic residuals in the maps . as a test , we first apply this technique to the heii @xmath14686 emission line .
the two distinct components we see in the doppler map ( fig .
[ dheii ] ) are reproduced clearly in the predicted trailed spectrogram ( left hand panel , ( fig .
[ pheii ] ) , and can be easily identified in the observed data . for the case of the niii @xmath14640 emission line ( fig .
[ pniii ] ) we again see two components in the reconstructed spectrogram , though not quite so distinctly as for the case of heii . comparing this idealised image to the real data it is indeed possible to detect faint traces of these two narrow , approximately anti - phased lines which produce the two distinct spots in fig .
[ dniii ] .
the second spot is therefore unlikely to be a systematic residual produced when creating the doppler map , and its unusual phasing / positioning could perhaps therefore be attributed to a region which violates one of the axioms of doppler tomography , for example that all motion should be parallel to the orbital plane .
note that a violation of these axioms does not mean that the map is erroneous , but it does complicate the successful interpretation of the map . 4640 emission line , part of the bowen emission complex .
next to this is plotted the ideal trailed spectrogram reconstructed from the doppler tomogram in fig .
[ dniii ] .
orbital phase is plotted along the @xmath29-axis . ]
@xcite estimated an opening ( semi- ) angle for the accretion disc in grmus of @xmath30 , using a simple geometric model for the reprocessing of x - rays in lmxbs .
it is important to note that this value is hostage to the implicit assumptions underlying these models , for example the geometric description of the disc and of the x - ray emitter .
this will entail a systematic uncertainty that likely exceeds the internal model accuracy of less than a degree .
nevertheless , this value agrees well with @xcite who estimated an opening angle of @xmath31 in order to explain the observed amplitude of the orbital photometric modulation . we can use this information to derive a mass estimate for the individual components by using the `
k - correction ' algorithm developed by @xcite .
this measures the deviation between the reprocessed light centre of the fluorescence emission lines and the centre of mass of a roche lobe filling star in a persistent lmxb , including the screening effects by a flared accretion disk . using the fourth order polynomial fits given by @xcite for disc opening angles of 8@xmath8 , 10@xmath8 , 12@xmath8 and 14@xmath8 gives a total mass range for the neutron star of 1.20 @xmath2 2.64 .
since the neutron star mass and the mass ratio are related by the mass function ] , we obtain 0.32 @xmath32 0.43 , and limits on the companion mass of 0.45 @xmath3 0.85 . this correction is dependent upon the opening angle of the accretion disc , monotonically decreasing with opening angle such that the largest opening angle gives the lowest neutron star mass .
since the largest opening angle we have used exceeds the estimates of both @xcite and @xcite , this gives a very conservative lower limit to the neutron star mass , and the true value is unlikely to be so low .
equally , the upper limit is derived using an opening angle lower than that suggested by @xcite and as such the neutron star is unlikely to be quite so massive .
lclc parameter & & parameter & + @xmath16 ( hjd ) & 2453151.647 ( 3 ) & @xmath20 ( kms@xmath13 ) & 185 @xmath5 2 + @xmath17 ( hjd ) & 2453151.509 ( 3 ) & @xmath33 & 0.320.43 + @xmath19 ( kms@xmath13 ) & 130 @xmath5 16 & @xmath34 & 1.202.64 + @xmath25 ( kms@xmath13 ) & 245 @xmath5 30 & @xmath35 & 0.450.85 + limiting ourselves to a disc opening angle of exactly 12@xmath8 @xcite , the mass constraints upon the neutron star tighten up to 1.35 @xmath2 2.32 , with 0.33 @xmath32 0.41 and 0.49 @xmath3 0.79 .
these limits are not a vast improvement over the ones derived above using a range of opening angles from 814@xmath8 .
this indicates that the major source of uncertainty in this case is still incurred from the measurements of @xmath25 and particularly of @xmath19 ( see section 3.3 ) , in addition to the system inclination angle .
the choice of disc opening angle ( within a reasonable range ) has only a small effect in comparison .
nevertheless , we feel it is prudent not to restrict our final mass estimate by the assumption of a disc opening angle of exactly 12@xmath8 , preferring the more conservative range of 814@xmath8 .
the derived parameters for grmus are listed in table [ tabparmus ] .
we know that the companion must be filling its roche lobe in order to permit persistent accretion . there is a well known and useful relationship between the fractional roche lobe radius and the mass - ratio @xcite which can be combined with kepler s 3rd law to provide an estimate of the size of the roche lobe as a function of period . in the case of this system ( with a 3.9 hour period ) , the relationship may be expressed as : @xmath36 .
we can thus derive the range of potential roche lobe radii consistent with the mass range of the companion star that we have derived kinematically .
this turns out to be 0.44 @xmath37 0.55 .
whilst at the lower end of our mass constraints the donor would be consistent with a main - sequence - like object as proposed by @xcite , we can not rule out the possibility that grmus harbours an over - massive evolved companion that has been stripped of its outer layers ( e.g. * ? ? ?
we have used the bowen emission blend and doppler tomography to trace for the first time the motion of the donor star in the persistent lmxb grmus .
this has allowed us to derive kinematical mass limits for this system . in combination with estimates for the opening angle of the accretion disc , and new measurements of the heii @xmath14686 emission line ( in an attempt to trace the motion of the compact object )
, we have derived tentative new mass constraints of 1.20 @xmath2 2.64 for the compact object and 0.45 @xmath3 0.85 for the companion star .
we can not rule out the possibility that grmus harbours an over - massive evolved companion star , or more intriguingly an over - massive neutron star .
tighter constraints on the system parameters may therefore have important implications for our knowledge of the equation of state of nuclear matter ( e.g. * ? ? ? * ) or indeed on the formation scenarios for typical lmxbs ( e.g. * ? ? ?
future investigations should also concentrate on determining the origin of the unusual second spot in the niii doppler map ( fig .
[ dniii ] ) .
this feature is hard to explain , and perhaps could be caused by a violation of the axioms of doppler tomography .
related to this point , the broad emission band seen in the left hand side of the doppler maps for heii and h@xmath18 ( in this system and a number of others ) is of uncertain origin .
we believe this emission to be produced in an extended disc bulge , but physical modelling will be necessary to confirm this .
adb was supported by a pparc research studentship .
ds acknowledges a smithsonian astrophysical observatory clay fellowship as well as support through nasa go grant nng06gc05 g .
based on observations made with eso telescopes at the paranal observatory under programme i d 073.d-0819 .
we gratefully acknowledge the use of the molly and doppler packages developed by t.r . marsh .
, schenker k. , 2002 , in ` the physics of cataclysmic variables and related objects ' , asp conference proceedings , vol . 261 . edited by b. t. gnsicke , k. beuermann , and k. reinsch . ,
san francisco : astronomical society of the pacific , p.233 van paradijs j. , mcclintock j.e . , 1995 , optical and ultraviolet properties of x - ray binaries ( x - ray binaries , cambridge university press , edited by lewin , w.h.g .
, van paradijs , j. & van der heuvel , p.j . ) | we present simultaneous high - resolution optical spectroscopy and x - ray data of the x - ray binary system grmus ( xb1254690 ) , obtained over a full range of orbital phases .
the x - ray observations are used to re - establish the orbital ephemeris for this source .
the optical data includes the first spectroscopic detection of the donor star in this system , through the use of the doppler tomography technique on the bowen fluorescence blend ( @xmath04630 - 4650 ) . in combination with an estimate for the orbital parameters of the compact object using the wings of the heii @xmath14686 emission line , dynamical mass constraints of 1.20 @xmath2 2.64 for the neutron star and 0.45 @xmath3 0.85 for the companion are derived .
[ firstpage ] binaries : close - stars : binaries : individual : grmus , xb1254690 - binaries : spectroscopic - x - rays : binaries . |
white dwarf ( wd ) stars represent the end points of stellar evolution for all low - mass stars , and are the fate of more than 97% of all stars in our galaxy .
roughly 80% of wds belong to the spectral class da , with atmospheres characteristically dominated by hydrogen @xcite . when da wds cool to the appropriate temperature to foster a hydrogen partial ionization zone , they begin their journey through the zz ceti ( or dav ) instability strip , where global pulsations are driven to observable amplitudes and their fundamental parameters can be determined using asteroseismology ( see reviews by @xcite , @xcite and @xcite ) . aside from their variability , the zz ceti stars discovered to date appear to be otherwise normal wds , and are therefore believed to be a natural phase in the evolution of all das . although some das within the empirical instability strip have been observed not to vary to modest limits ( e.g. , @xcite ) , follow - up observations have shown that some of these stars really do pulsate at low amplitude ( e.g. , @xcite ) .
higher - quality optical and uv spectra have also moved some of these non - variable interlopers out of the instability strip @xcite .
thus , it is currently believed that the zz ceti instability strip is pure , and that all da wds will at some point pass through it and pulsate @xcite .
much work has been devoted to observationally mapping the zz ceti instability strip , which runs in temperature from roughly @xmath11 k for standard @xmath4 = 8.0 wds @xcite .
there is also a dependence on surface gravity , such that wds with lower @xmath4 pulsate at lower effective temperatures .
this trend has been observed for wds with masses from 1.1 @xmath1 down to 0.5 @xmath1 @xcite .
the blue edge of the zz ceti instability strip , where pulsations are turning on , has been successfully estimated by both convective period arguments @xcite and full non - adiabatic calculations @xcite .
a slightly more efficient prescription for convection has to be assumed , by increasing the value of the mixing - length theory parameter ml2/@xmath12 , to make the theory match the observed blue edge , which was most recently mapped empirically by @xcite .
however , estimating the temperature at which pulsations should shut down has remained a challenge .
modern non - adiabatic calculations do not predict a red edge until around 5600 k @xcite , more than 5000 k cooler than the empirical red edge @xcite .
@xcite argue that a surface reflection criterion can be enforced to limit the maximum mode period , which may push a theoretical red edge to hotter temperatures , nearer what is observed in zz ceti stars @xcite .
the recent discovery of pulsating extremely low - mass ( elm , @xmath13 0.25 @xmath1 ) wds provides us with an exciting new opportunity to explore the nature of the physics of wd pulsations at cooler temperatures and much lower masses . since the first discovery by @xcite , more than 160 zz ceti stars have been found , all of which have masses @xmath14 0.5 @xmath1 and thus likely harbor carbon - oxygen ( co ) cores . that changed with the discovery of the first three pulsating elm wds @xcite .
these elm wds are likely the product of binary evolution , since the galaxy is not old enough to produce such low - mass wds through single - star evolution @xcite . during a common - envelope phase ,
the elm wds were most likely stripped of enough mass to prevent helium ignition , suggesting they harbor he cores .
the pulsating elm wds will be incredibly useful in constraining the interior composition , hydrogen - layer mass , overall mass , rotation rate , and the behavior of convection in these low - mass wds , which may derive a majority of their luminosities from stable hydrogen burning for the lowest - mass objects @xcite .
several groups have recently investigated the pulsation properties of he - core wds , and non - adiabatic calculations have shown that non - radial @xmath15- and @xmath10-modes should be unstable and thus observable in these objects @xcite .
pulsating elm wds will also extend our empirical studies of the zz ceti instability strip to significantly lower surface gravities . boosted by the many new elm wds catalogued by the elm survey , a targeted spectroscopic search for elm wds @xcite , we have looked for additional pulsating elm wds throughout a large part of parameter space .
the first three pulsating elm wds all have effective temperatures below @xmath16 k , much cooler than any previously known co - core zz ceti star @xcite , which makes up the coolest class of pulsating wds .
we now add to that list the two coolest pulsating wds ever found , sdss j161431.28 + 191219.4 ( @xmath17 mag , hereafter j1614 ) and sdss j222859.93 + 362359.6 ( @xmath18 mag , hereafter j2228 ) , bringing to five the number of elm wds known to pulsate .
in section [ sec : j1614 ] we detail our discovery of pulsations in j1614 and outline our new spectroscopic observations of this elm wd . in section [ sec : j2228 ]
we describe the discovery of multi - periodic variability in the elm wd j2228 and update its determined atmospheric parameters .
we conclude with a discussion of these discoveries , and update the observed da wd instability strip .
@xcite found that j1614 had @xmath2 @xmath19 k and @xmath4 @xmath20 , based on a single spectrum of this @xmath17 mag wd from the flwo 1.5 m telescope using the fast spectrograph @xcite .
we have obtained an additional 51 spectra using the same instrument and setup .
we have co - added our spectroscopic observations to determine the atmospheric parameters of the elm wd j1614 ( figure [ fig : j1614spec ] ) .
our observations cover a wavelength range from @xmath21 .
the model atmospheres used for this analysis are described at length in @xcite and employ the new stark broadening profiles from @xcite .
models where convective energy transport becomes important are computed using the ml2/@xmath12 = 0.8 prescription of the mixing - length theory ( see * ? ? ?
a discussion of our extension of these models to lower surface gravities and more details of our fitting method can be found in section 2.1.1 of @xcite .
our final fit to the phased and co - added spectrum of j1614 is shown in the top panel of figure [ fig : j1614spec ] and yields @xmath2 @xmath3 k and @xmath4 @xmath5 .
this corresponds to a mass of @xmath60.20 @xmath1 using the he - core wd models of @xcite , if we assume the wd is in its final cooling stage .
the more recent models of @xcite predict a mass of 0.19 @xmath1 given the atmospheric parameters , which we adopt .
we have also performed our fit without using the low s / n lines h11@xmath22h12 , but this marginally affects our solution : using only the h@xmath23@xmath22h10 lines of the balmer series , we find @xmath2 @xmath24 k and @xmath4 @xmath25 . to remain consistent with our previous pulsating elm wd atmospheric determinations @xcite , we will include the h11@xmath22h12 lines in our adopted solution for j1614 .
@lcccc@ run & ut date & length & seeing & exp .
+ & & ( hr ) & ( ) & ( s ) + + a2690 & 2012 jun 21 & 2.6 & 1.7 & 5 + a2692 & 2012 jun 22 & 2.0 & 1.8 & 5 + a2695 & 2012 jun 23 & 3.7 & 1.2 & 5 + a2697 & 2012 jun 24 & 3.5 & 1.4 & 5 + a2699 & 2012 jun 25 & 3.6 & 1.3 & 5 + + a2521 & 2011 nov 28 & 3.5 & 1.5 & 10 + a2524 & 2011 nov 29 & 1.6 & 2.5 & 10 + a2528 & 2011 nov 30 & 1.9 & 2.2 & 10 + a2707 & 2012 jul 13 & 2.3 & 1.1 & 5 + a2710 & 2012 sep 17 & 2.8 & 2.4 & 10 + a2719 & 2012 sep 20 & 6.4 & 1.6 & 15 + a2721 & 2012 sep 21 & 7.4 & 1.4 & 10 + elm wds are typically found in close binary systems ; these companions are necessary to strip the progenitor of enough mass to form such a low - mass wd within the age of the universe @xcite . however , using the code of @xcite , we do not detect any significant radial velocity variability in our observations of j1614 . the r.m.s .
scatter gives us an upper limit on the rv semi - amplitude : @xmath26 km s@xmath27 .
the systemic velocity is @xmath28 km s@xmath27 .
we note that this non - detection does not require the lack of a companion to the elm wd in j1614 .
rather , the system may be inclined nearly face - on to our line of sight , or the companion may be a much cooler low - mass wd
. if the inclination is @xmath29 , which is more than 85% likely if the orientation of the system with respect to the earth is drawn from a random distribution , the companion has @xmath30 @xmath1 if the system has a 7 hr orbital period , the median for elm wd binaries in the elm survey @xcite . empirically , there are similarly low - mass wds in the elm survey with no significant radial velocity variability @xcite .
we obtained high - speed photometric observations of j1614 at the mcdonald observatory over five consecutive nights in 2012 june for a total of nearly 15.4 hr of coverage .
we used the argos instrument , a frame - transfer ccd mounted at the prime focus of the 2.1 m otto struve telescope @xcite , to obtain @xmath31 s exposures on j1614 .
a full journal of observations can be found in table [ tab : jour ] .
observations were obtained through a 3 mm bg40 filter to reduce sky noise .
we performed weighted , circular , aperture photometry on the calibrated frames using the external iraf package @xmath32 written by antonio kanaan @xcite .
we divided the sky - subtracted light curves by the brightest comparison star in the field , sdss j161433.39 + 191058.3 ( @xmath33 mag ) , to correct for transparency variations , and applied a timing correction to each observation to account for the motion of the earth around the barycenter of the solar system @xcite .
the top panel of figure [ fig : j1614 ] shows a portion of a typical light curve for j1614 , obtained on 2012 june 23 , and includes the brightest comparison star in the field over the same period .
the bottom panel of this figure shows a fourier transform ( ft ) utilizing all @xmath34 light curve points collected thus far .
we display the 4@xmath35 significance line at the bottom of figure [ fig : j1614 ] , calculated from the average amplitude , @xmath35 , of an ft within a 1000 @xmath36hz region in steps of 200 @xmath36hz , after pre - whitening by the two highest - amplitude periodicities .
.frequency solution for sdss j161431.28 + 191219.4 [ tab : j1614freq ] [ cols="<,^,^,^,^ " , ] we may compare the first five pulsating elm wds to the previously known zz ceti stars by placing them in a @xmath4@xmath2 diagram , shown in figure [ fig : search ] .
doing so , we discover there are at least six elm wds with temperatures and surface gravities between the newfound pulsating elm wd j2228 and the other four known pulsating elm wds .
these non - variable elm wds have been observed extensively and do not show significant evidence of pulsations to at least 1% relative amplitude .
we have excellent limits on the lack of variability in four of these six , ruling out pulsations larger than 0.3% amplitude .
we have put limits on three of these new non - detections , detailed in table [ tab : null ] .
we note that @xcite previously observed psr 1012 + 5307 , but we have put much more stringent limits on a lack of variability on this faint elm wd with 7 hr of observations in excellent conditions .
the other three interlopers have been detailed in previous studies .
sdss j0822 + 2753 is a @xmath2 @xmath37 k , @xmath4 @xmath38 wd observed not to vary to 0.2% @xcite .
sdss j1443 + 1509 is a @xmath2 @xmath39 k , @xmath4 @xmath40 wd with exquisite limits on lack of variability , to @xmath41% @xcite . finally , nltt 11748 is the @xmath2 @xmath42 k , @xmath4 @xmath43 primary wd in an eclipsing wd+wd binary @xcite .
it was shown by @xcite not to vary out of eclipse to above 0.5% .
we have obtained an additional 8 hr of photometry of nltt 11748 out of eclipse at mcdonald observatory and can independently rule out variability larger than 0.3% . the discovery of pulsations in j2228 , which is considerably cooler than at least a half - dozen other photometrically constant elm wds , questions the purity of the instability strip for he - core wds and confuses the location of an empirical red edge .
however , there is no a priori reason to expect the elm wd instability strip to be pure , or for there to exist a connected low - mass extension of the classical co - core zz ceti instability strip ; evolution through a specific temperature - gravity region is not well established for the elm wds , and they may not all cool through the instability strip in as simple a manner as the co - core zz ceti stars .
in fact , some of these elm wds may indeed be in the throws of unstable hydrogen shell burning episodes ; they may not be cooling at all , but rather looping through the hr diagram prior to settling on a final cooling track ( e.g. , @xcite ) .
such excursions are not expected for co - core zz ceti stars , which are expected to monotonically cool through an observationally pure instability strip .
we have plotted the evolution of theoretical cooling tracks for several different wd masses through the effective temperatures and surface gravities in figure [ fig : search ] .
we plot the 0.16 @xmath1 , 0.17 @xmath1 , 0.18 @xmath1 , 0.20 @xmath1 , 0.25 @xmath1 , and 0.35 @xmath1 he - core models of @xcite as dotted magenta lines .
we have also used the stellar evolution code mesa @xcite to model the evolution of 0.15 @xmath1 , 0.20 @xmath1 , and 0.25 @xmath1 he - core wds , shown as solid cyan lines in figure [ fig : search ] . for reference
, we have also included 0.6 @xmath1 , 0.8 @xmath1 , and 1.0 @xmath1 co - core cooling tracks @xcite . where the lowest - mass wd models enter this diagram depends on how we artificially remove mass from the models , and there is a very noticeable discrepancy between the 0.16 @xmath1 @xcite wd models and our 0.15 @xmath1 wd models using mesa . as an added complication , except for the lowest - mass elm wds ( below roughly 0.18 @xmath1 ) , recurrent hydrogen shell flashes cause the elm wd model to loop many times through this @xmath2-@xmath4 plane , further confusing the picture @xcite .
thus , it is not entirely surprising to find non - variable elm wds between j2228 and the four warmer pulsating elm wds .
further empirical exploration of the entire elm wd instability strip offers a unique opportunity to constrain physical and evolution models of elm wds , specifically these late thermal pulses and the mass boundary for the occurrence of these episodes .
long - term monitoring of the rate of period change of pulsating elm wds also affords an opportunity to constrain the cooling ( or heating ) rate of these objects @xcite .
in contrast to the confusion along the red edge of the instability strip , the blue edge is more reliably predicted by theory .
the theoretical blue edge ( dotted blue line ) in figure [ fig : search ] has been calculated following @xcite and @xcite .
we use the criterion that @xmath44 for the longest period mode that is excited , where @xmath45 is the mode period and the timescale @xmath46 describes the heat capacity of the convection zone as a function of the local photospheric flux , which we compute from a grid of models ( see @xcite for further details ) .
we use the criterion @xmath47 s , with the convective prescription ml2/@xmath12=1.5 .
we also include the theoretical blue edge of @xcite , which uses a slightly less efficient prescription for convection , ml2/@xmath12=1.0 .
we have discovered pulsations in two new extremely low - mass , putatively he - core wds using optical facilities at the mcdonald observatory .
spectral fits show that these two elm wds , j1614 and j2228 , are the coolest pulsating wds ever found .
this brings to five the total number of pulsating elm wds known , establishing them as a new class of pulsating wd . as with the more than 160 co - core zz ceti stars that have been known for more than four decades , the luminosity variations in these elm wds is so far consistent with surface temperature variations caused by non - radial @xmath10-mode pulsations driven to observability by a hydrogen partial ionization zone .
the coolest pulsating elm wd , j2228 , has a considerably lower effective temperature than six similar - gravity elm wds that are photometrically constant to good limits .
in contrast to the co - core zz ceti stars , which are believed to represent a stage in the evolution of all such wds , elm wds may not all evolve through an instability strip in the same way , and thus we may not observe their instability strip to be pure .
theoretical he - core wd models predict multiple unstable hydrogen - burning episodes , which complicates the evolution of an elm wd through a simple instability strip . empirically discovering elm wds in this space that do or do not pulsate opens the possibility to use the presence of pulsations in elm wds to constrain the
binary and stellar evolution models used for low - mass wds , which may better constrain these poorly understood cno - flashing episodes .
we acknowledge the anonymous referee for valuable suggestions that greatly improved this manuscript .
j.j.h . , m.h.m . and d.e.w .
acknowledge the support of the nsf under grant ast-0909107 and the norman hackerman advanced research program under grant 003658 - 0252 - 2009 .
m.h.m . additionally acknowledges the support of nasa under grant nnx12ac96 g . b.g.c .
thanks the support from cnpq and fapergs - pronex - brazil .
the authors are grateful to the essential assistance of the mcdonald observatory support staff , especially dave doss and john kuehne , and to fergal mullally for developing some of the data analysis pipeline used here . | we report the discovery of two new pulsating extremely low - mass ( elm ) white dwarfs ( wds ) , sdss j161431.28 + 191219.4 ( hereafter j1614 ) and sdss j222859.93 + 362359.6 ( hereafter j2228 ) .
both wds have masses @xmath0 0.25 @xmath1 and thus likely harbor helium cores .
spectral fits indicate these are the two coolest pulsating wds ever found .
j1614 has @xmath2 @xmath3 k and @xmath4 @xmath5 , which corresponds to a @xmath60.19 @xmath1 wd .
j2228 is considerably cooler , with a @xmath2 @xmath7 k and @xmath4 @xmath8 , which corresponds to a @xmath60.16 @xmath1 wd , making it the coolest and lowest - mass pulsating wd known .
there are multiple elm wds with effective temperatures between the warmest and coolest known elm pulsators that do not pulsate to observable amplitudes , which questions the purity of the instability strip for low - mass wds .
in contrast to the co - core zz ceti stars , which are believed to represent a stage in the evolution of all such wds , elm wds may not all evolve as a simple cooling sequence through an instability strip . both stars
exhibit long - period variability ( @xmath9 s ) consistent with non - radial @xmath10-mode pulsations .
although elm wds are preferentially found in close binary systems , both j1614 and j2228 do not exhibit significant radial - velocity variability , and are perhaps in low - inclination systems or have low - mass companions .
these are the fourth and fifth pulsating elm wds known , all of which have hydrogen - dominated atmospheres , establishing these objects as a new class of pulsating wd .
[ firstpage ] stars : white dwarfs stars : oscillations ( including pulsations ) galaxy : stellar content stars : individual : sdss j161431.28 + 191219.4 , sdss j222859.93 + 362359.6 |
giant molecular clouds ( gmcs ; @xmath4 ) in the galaxy are highly inhomogeneous : they are often filamentary in appearance , consisting of discrete clumps , or cores ( @xmath5 ) , which contain most of the mass of a cloud ( including any young stars ) and are surrounded by a more diffuse component of predominantly atomic gas ( e.g. , williams , blitz , & stark 1995 )
. indeed , gmcs are clumpy on all scales observed , and are possibly even fractal in nature ( falgarone , phillips , & walker 1991 ) .
remarkably enough , however , the gross properties of cloud complexes are rather simply interrelated .
total masses , mean densities , and average velocity dispersions vary with sizes ( effective radii ) roughly as @xmath6 , @xmath7 , and @xmath8 ( larson 1981 ; sanders , scoville , & solomon 1985 ; solomon et al .
1987 ) , with uncertainties in the exponents typically of order @xmath90.1 .
the seeming universality of these results demands a physical explanation .
the relationship between size and linewidth ( which term we use interchangeably with velocity dispersion ) is further interesting because @xmath10 is observed to decrease towards smaller radii _ inside _ gmcs , and within individual dense cores ( larson 1981 ; miesch & bally 1994 ; fuller & myers 1992 ; caselli & myers 1995 ) .
since the total linewidths of gmcs are mostly nonthermal ( the clouds are stable against gravitational collapse on the largest scales , but have masses several orders of magnitude above the thermal jeans value , so that their support must come largely from nonmagnetic or , very likely , mhd turbulence ) , this decrease reflects a move towards domination by thermal motions on the smallest scales .
we should therefore expect the linewidths of low - mass cores to have a smaller turbulent component than those of high - mass cores .
this is indeed the case ; in fact , the velocities nearest the centers of small cores are almost ( to within a few percent ) wholly thermal .
however , it also happens ( caselli & myers 1995 ) that the nonthermal velocity dispersion shows _ a stronger dependence on radius in low - mass cores _
( where @xmath11 ) than in massive ones ( @xmath12 ) . because star formation is localized in the cores of gmcs , their overall structure
and this aspect specifically bears strongly on our understanding of this most fundamental process .
the goal of this paper is to find a model for the internal structure of molecular cores ( low- and high - mass both ) which quantitatively matches their observed , internal velocity - dispersion profiles , and is also consistent with global properties ( such as the observed mass - radius - linewidth relations ) of large , self - gravitating clumps and even whole gmcs .
our approach is to reduce this problem to the specification of a ( total ) pressure - density relation an equation of state that , when used to solve the equation of hydrostatic equilibrium , results in a gas cloud with the required features .
it is significant that the linewidth profiles of cores are insensitive to the presence or absence of young stars , and thus may be viewed as one of the prerequisites for star formation ( fuller & myers 1992 ; caselli & myers 1995 ) .
this justifies our focus on the structure of purely gaseous clouds .
to proceed , we shall resort to a virial - theorem treatment of molecular clouds which idealizes them as spheres of gas in hydrostatic equilibrium and satisfying poisson s equation . this is appropriate enough for dense cores , which in many cases are roughly spheroidal ( probably prolate : myers et al .
1991 ) , or even near spherical ( e.g. , williams et al .
1995 ) , overall .
( although their internal density distributions may not be especially smooth on very small scales , we concern ourselves here with a description of their _ bulk _ structure . ) in addition , observations of very massive cores imply that they are self - gravitating and in approximate `` gravitational virial equilibrium , ''
this is not the case for low - mass cores , but these still appear to satisfy the full virial theorem if surface - pressure terms are included ( bertoldi & mckee 1992 ) . finally , even the total masses , radii , and linewidths of entire gmc complexes are generally consistent with virial - equilibrium models of spheres ( e.g. , solomon et al . 1987 ; elmegreen 1989 )
and it has been repeatedly confirmed that whole clouds tend to comply with @xmath14 ( larson 1981 ; solomon et al .
1987 ; myers & goodman 1988b ) . it should be noted that the correct , average mass - radius and size - linewidth relations can be recovered in a purely scale - free description of gmcs ( e.g. , henriksen 1991 ) .
but the distinctly different scalings of velocity dispersion with radius inside low- and high - mass clumps is one indication that molecular clouds are in fact not entirely featureless .
other current models for these objects ( such as isothermal spheres or negative - index polytropes ) face similar difficulties in accounting simultaneously for their global properties and their internal structures .
magnetic fields are an important presence in regions of star formation ( e.g. , heiles et al .
1993 ) , so we begin in 2 by writing down the virial - theorem ( or bonnor - ebert ) relations between the masses and total linewidths including turbulent velocities of magnetized spheres , truncated at radii such that the internal pressure just balances that of a diffuse surrounding medium .
we then develop a stability criterion for such clouds which depends only on the assumption that their thermal linewidths ( i.e. , kinetic temperatures ) are invariant .
when combined with the circumstantial evidence for equipartition between the kinetic and mean - field magnetic energies in gmcs and massive cores ( myers & goodman 1988a , b ; bertoldi & mckee 1992 ) , this stability criterion leads to mass - radius - linewidth relations between critical - mass objects that agree with the observed scalings ( both the exponents and the coefficients ) among gmcs .
our analysis is therefore similar to those of , e.g. , chize ( 1987 ) , fleck ( 1988 ) , and especially elmegreen ( 1989 ) , but ours holds for clouds with an _ arbitrary _ equation of state .
in addition , we find that critically stable clouds in magnetic equipartition should all satisfy @xmath15 .
we therefore conclude that gmcs and massive cores are approximately at their critical masses and magnetically supercritical , with @xmath16 ( see also mckee 1989 ; bertoldi & mckee 1992 ) .
the generality of these results allows for the investigation of essentially any pressure - density relation as a potential description of the interiors of molecular cores , just so long as critically stable configurations are at all possible . in 3
, we discuss polytropic equations of state : @xmath17 , with @xmath18 any positive number . clearly , if @xmath19 , then the total velocity dispersion ( @xmath20 ) stays constant or decreases with decreasing density ( increasing radius ) inside a cloud .
if instead @xmath18 is allowed to be less than 1 ( e.g. , maloney 1988 ) , then @xmath21 increases with radius , as required . however , in our analysis , any such `` negative - index '' polytrope would be unconditionally stable against gravitational collapse ; or equivalently ( but independently of any stability criterion ) , it could not self - consistently satisfy the relation @xmath22 .
while this is not a problem for low - mass cores , it is inconsistent with what we know of very large clumps , and gmcs overall .
thus , we argue that polytropic pressure - density relations give an incomplete picture of interstellar clouds .
a corollary to this is that weakly damped alfvn waves , for which @xmath23 ( mckee & zweibel 1995 ) , can not be invoked as the _ sole _ explanation of nonthermal linewidths in gmcs .
section 4 describes what is , in our view , a more suitable alternative .
there we consider the possibility that pressure varies only logarithmically with density : @xmath24 .
we refer to the resulting gas cloud as a `` pure '' logotrope .
this term was introduced by lizano & shu ( 1989 ; see also mckee 1989 ) , who actually added a logarithmic term to an otherwise isothermal equation of state in an attempt to account for turbulent linewidths .
such models have also been studied in detail by gehman et al .
( 1996 ) . here
, however , we dispense with the explicitly isothermal component , for two reasons : ( 1 ) assuming @xmath25 , with @xmath26 the thermal velocity dispersion and @xmath27 some reference density ( lizano & shu 1989 ) , the observational inference that linewidths should be essentially purely thermal at the centers of cores requires that @xmath28 .
but then the internal @xmath20 _ decreases _ with increasing radius .
( 2 ) if now @xmath29 , then for large values of @xmath30 such as those suggested by gehman et al .
( 1996 ) , @xmath31 vanishes for @xmath32 rather near unity .
thus , real clouds would have to be almost uniform - density , which again is not observed .
these difficulties do not extend to the specific equation of state which we examine .
instead , an outwards - increasing velocity - dispersion profile obtains for an equilibrium pure logotrope .
such a model can moreover account for the observed mass - radius and size - linewidth relations between gmcs .
we also demonstrate that the linewidth measurements in both low- and high - mass cores , from a variety of molecular clouds , are quantitatively reproduced if @xmath3 .
the argument makes explicit use of the fact that small cores are _ not _ at their critical masses ( while large ones generally are ) , but that they are still in approximate virial equilibrium when the effects of surface pressures are considered .
although the logarithmic @xmath33 relation we advocate is phenomenological , its overall viability , along with the failings of other models , ultimately makes for a useful description of molecular clouds and their cores .
interstellar clouds can be viewed as essentially `` pressure - truncated '' bodies of gas .
( this term is meant to imply the existence , not of some radius where a gas cloud suddenly ends , but of one where it `` blends in '' with an ambient medium . )
thus , the boundary of a gmc is set by pressure balance with the surrounding , hot ism .
note that even if there is no such balance initially , it will eventually obtain after an overall expansion or contraction of the cloud complex . the extent of a core within a gmc is similarly limited by the pressure of a tenuous interclump medium , as has been argued by bertoldi & mckee ( 1992 )
. the equilibrium structure of pressure - truncated isothermal spheres was first described by ebert ( 1955 ) and bonnor ( 1956 ) ( see also mccrea 1957 ) .
their expressions connecting the masses , radii , and linewidths of the spheres follow from the assumption of hydrostatic equilibrium and poisson s equation , and therefore are written in terms of an internal pressure profile and gravitational potential . in appendix
a , we re - derive these bonnor - ebert relations , but for clouds satisfying an arbitrary gas equation of state ( eqs .
[ [ eq:4a]][[eq:4d ] ] ) .
we also provide a connection ( eqs .
[ [ eq:13a]][[eq:13 ] ] ) with the more transparent and observationally convenient virial - theorem formulation given by equations ( [ eq:12 ] ) below .
as various authors have noted ( chize 1987 ; fleck 1988 ; maloney 1988 ; elmegreen 1989 ) , these relations in either guise provide a framework for an understanding of the standard `` larson s laws '' for gmcs ( @xmath6 and @xmath34 ; larson 1981 ) .
the development of appendix a refers specifically to `` nonmagnetic '' clouds , meaning only that no _ ordered _ ( mean ) magnetic field is considered to be present .
we work on the assumption that the effects of disordered fields ( mhd turbulence ) can be separately dealt with , in an equivalent hydrostatic problem that makes use of an effective equation of state to describe _ all _ of the contributions to gas pressure as a function of density .
still , the effects of mean fields must also be considered in any applications to real interstellar clouds . in the absence of any analytic models for magnetohydrostatic clouds ,
we proceed by assuming spherical symmetry and turning to the scalar virial theorem : @xmath35 here the mass - averaged , total one - dimensional velocity dispersion of a cloud with radius @xmath36 is related to its mean pressure and density by @xmath37 so that the kinetic ( or internal ) , mean - field magnetic , and gravitational energies are given by the usual @xmath38 @xmath39 and @xmath40 the parameter @xmath41 is essentially a measure of the non - uniformity of a gas sphere , and as such depends on the equation of state and the truncation radius ( in terms of a fixed scale @xmath42 ; eq . [ [ eq:13a ] ] ) . however , it is generally of order unity : for a power - law density profile @xmath43 , equation ( [ eq:9 ] ) gives @xmath44 .
we expect @xmath45 in a realistic gmc or core , and thus @xmath46 .
it is often more useful to work in terms of the virial parameter of bertoldi & mckee ( 1992 ) : @xmath47 this _ observable _ quantity can also be written ( cf .
mckee & zweibel 1992 ) as @xmath48 although this form of the virial theorem is appropriate for clouds of any shape , the corrective factors for spheroidal clouds are rather near unity ( bertoldi & mckee 1992 ) , and we still allow only for spherical symmetry here .
it is further convenient to distinguish between the virial parameters which would obtain for a cloud with and without a mean magnetic field ; these are related by @xmath49 again , @xmath50 refers specifically to the combination ( [ eq:10 ] ) of observables . on the other hand
, @xmath51 applies to the @xmath52 ( no mean field ) counterpart of a given cloud ; it is directly observable only in this special case ( since then @xmath53 ) , but may always be calculated for a given gas equation of state , as outlined in appendix a. once this is known , @xmath50 follows with the specification of a mean - field configuration ( see 2.3 ) .
the connection ( [ eq:11 ] ) between @xmath50 and @xmath51 is valid insofar as the ratio @xmath54 does not change drastically upon the `` addition '' of a mean magnetic field to a cloud which is already in hydrostatic equilibrium .
this must hold along field lines anyway ( force balance is required in that direction ) , and thus everywhere on the surface of a roughly spherical cloud . defining @xmath55
, such reasoning implies that @xmath56 , so manipulation of the virial theorem gives the following : @xmath57 @xmath58 @xmath59 and @xmath60 these relations , of which only two are independent ( they are contained , for example , in eqs .
[ 9 ] and [ 10 ] of elmegreen 1989 ) , are also given by harris & pudritz ( 1994 ) for the specific case of critically stable isothermal spheres ( as defined in 2.2 ; @xmath61 and @xmath62 at a cloud radius @xmath63 ) .
as written here , they apply to any generic gas cloud , stable or unstable , isothermal or not .
any cloud in hydrostatic equilibrium and satisfying poisson s equation has a central region where the potential , density , and velocity dispersion are very nearly constant with radius .
equations ( [ eq:13a ] ) and ( [ eq:13b ] ) show that spheres truncated in or just outside this central region ( i.e. , at radii small enough that @xmath64 ) have @xmath65 , and thus @xmath66 by equations ( [ eq:12]a , b ) .
roughly this scaling has been observed for the virial parameters of low - mass , high-@xmath67 cores in several gmcs , leading bertoldi & mckee ( 1992 ) to argue that such clumps can be viewed as truncated spheroids which are essentially in pressure equilibrium with an intracloud medium . given equations ( [ eq:12 ] ) for the equilibrium structure of pressure - truncated gas spheres , we are in a position to question their stability : under what conditions will they be able to withstand the combined effects of self - gravity and surface pressure , and when will they be unstable to wholesale gravitational collapse ?
the answer to this depends , of course , on any boundary conditions attached to a perturbation of the cloud . obviously , the total mass should be unchanged by a contraction or expansion of the entire structure .
in addition , following maloney ( 1988 ) , we suppose that the central velocity dispersion remains constant as the cloud radius , or the surface pressure , is varied .
this stipulation is meant to reflect the fact that the turbulent linewidth decreases steadily towards smaller scales in cores and in entire gmcs .
we therefore identify the central velocity dispersion with the thermal part of the total linewidth : @xmath68 . insisting that this be invariant amounts to recognizing the rough uniformity of kinetic temperatures @xmath69 ( which to first order can be understood as a consequence of the competition between cosmic - ray heating and co cooling ) over a large range of scales in interstellar clouds .
the stability criterion that follows ultimately leads to a set of results which self - consistently explain some important observational features of gmcs and massive cores .
a gas cloud will be stable against radial perturbations if the derivative @xmath70 ( taken with @xmath71 and @xmath72 held fixed ) is @xmath73 : a slight decrease in the cloud radius then leads to an increase in the pressure just inside its boundary , which in turn leads to reexpansion .
appendix b shows that , for any equation of state , this condition is just @xmath74\leq0\ , \label{eq:14}\ ] ] where @xmath75 is the internal density at the edge of the cloud .
although this stability criterion has been derived without explicitly considering the effects of mean magnetic fields , we expect that it should not be greatly altered by their inclusion .
( this is again implied by our assumption of approximate spherical symmetry , since eq .
[ [ eq:14 ] ] must at least be satisfied along field lines at the boundary of a magnetized cloud . ) depending on the equation of state , there may exist a radius for which a pressure - truncated cloud is marginally stable ( the expression [ [ eq:14 ] ] is just 0 ) , and beyond which it is unstable . it is these critical equilibria , which must have @xmath76 that are of particular interest here . once @xmath51 and @xmath41 are known as functions of radius ( as in appendix a ) , the satisfaction , if possible , of equation ( [ eq:15 ] ) sets the boundary @xmath77 of the cloud .
this in turn allows for evaluation of the coefficients in equations ( [ eq:12 ] ) . quite generally , @xmath78
@xmath79 @xmath80 and @xmath81 equation ( [ eq:16]c ) shows that a nonmagnetic cloud ( @xmath53 ) on the verge of gravitational collapse has a mean column density which is fixed by the pressure of the surrounding medium , independently of any gas equation of state . in general ,
the virial parameter of a given cloud is sensitive to its internal structure ( through the equation of state ) and its total radius .
however , for a _ critically stable _ cloud we always have @xmath82 . since @xmath41 is typically of order ( but slightly greater than ) unity , this implies @xmath83 and @xmath84 . the virial theorem ( [ eq:5 ] )
then becomes @xmath85 if there is equipartition @xmath86 between the magnetic and kinematic energies in such a cloud , then @xmath13 as well ; thus , @xmath87 , and the critical @xmath50 is expected to be of order unity ( cf .
eq . [ [ eq:11 ] ] ; see also elmegreen 1989 and mckee & zweibel 1992 ) . observations of molecular clouds show that @xmath13 ( or equivalently , @xmath88 ) , and are _ consistent _ with @xmath86 ( myers & goodman 1988a , b ) , although actual magnetic - field measurements are few and uncertain .
the most massive cores in gmcs similarly tend to show @xmath22 and @xmath50 near 1 ( this is not the case for low - mass cores , however : e.g. , williams , de geus , & blitz 1994 ; williams et al . 1995 ; see also 4.1 below ) .
observations of them are also indicative of magnetic equipartition ( bertoldi & mckee 1992 ) , though again the evidence is rather indirect , and not necessarily conclusive .
having said this , it does seem that gmcs and massive cores both satisfy equation ( [ eq:17a ] ) , which would imply that they are at or near their critical masses
. thus arrived at , this conclusion depends on the criterion one adopts for cloud stability , i.e. , it follows from the assumption that the thermal linewidth @xmath71 is held fixed during any radial perturbation ( and from the additional proviso that @xmath86 ) .
nevertheless , our analysis which specifies no equation of state provides a natural explanation for the fact that so many gmcs and large cores appear to be in simple `` gravitational virial equilibrium '' ( i.e. , @xmath89 ) , even though they are threaded by appreciably strong mean magnetic fields .
there is also separate evidence for the criticality of massive molecular cores .
for instance , bertoldi & mckee ( 1992 ) , in their study of the cores in four gmcs , argue that the most massive are at least magnetically supercritical , a necessary condition for gravitational instability .
more fundamentally , star - forming regions clearly must be susceptible to gravitational collapse ; but in the rosette gmc at least , those cores which are most obviously associated with iras sources are also among the heaviest ( williams et al .
as mentioned above , the largest cores in several gmcs also have the smallest virial parameters ( as low as 1 ) , so that this would seem to be a feature of clouds which are close to instability .
the observation of a mean @xmath90 for gmcs then implies that they , too , are near some critical mass , and certainly in excess of the nonmagnetic jeans or bonnor - ebert value ( for which @xmath52 implies @xmath91 and @xmath92 ) .
mckee ( 1989 ) has further argued that gmcs on the whole are magnetically supercritical , and of course they must be strongly self - gravitating in order to be molecular at all ( e.g. , elmegreen 1985 ) . on a related note , mckee ( 1989 ) points out that gmcs must generally be near criticality because they show a @xmath93 which is typically an order of magnitude larger than the total ( thermal plus turbulent ) pressure in the hot ism .
given our stability criterion , equation ( [ eq:16]d ) shows that self - gravity can supply a maximal pressure enhancement @xmath94 of a spherical cloud over its surrounding medium , and this only for a critical - mass body .
even putting the nonsphericity of gmcs aside , however , it is important to note that the analysis here speaks only to the molecular , self - gravitating parts of gmcs , and not to their diffuse , low-@xmath95 components .
elmegreen ( 1989 ) has shown that the weight of these atomic `` envelopes '' can easily increase the pressure at the boundaries of the molecular parts of a cloud complex by a factor of 5 or more above the value in the ism at large ; overall , then , @xmath96 .
if gmcs and their most massive cores are indeed critical - mass objects , then they must all have the same dimensionless radii ( although the physical scale @xmath42 will generally vary ) , and the same virial parameters .
aside from possible variations in @xmath97 , which are discussed in detail by elmegreen ( 1989 ) , this causes the coefficients in equations ( [ eq:16 ] ) or ( [ eq:12 ] ) to be roughly constant , and allows for well defined mass - radius - linewidth relations between clouds .
moreover , we have argued that @xmath98 and , in the event of magnetic equipartition , @xmath99 for critical clouds , regardless of the underlying equation of state .
the coefficients in the @xmath100 scalings are then independent of this detail , and the average properties of gmcs can shed no light on their internal structure .
this both explains why critical , magnetized isothermal - sphere models are successful in quantitatively accounting for the observed scalings ( with @xmath101 and @xmath99 : elmegreen 1989 ; harris & pudritz 1994 ) , and implies that the same agreement with the data comes with _ any _ model which provides for the existence of a critical mass ( see eqs .
[ [ eq:411 ] ] below ) .
finally , equations ( [ eq:16 ] ) as written would suggest that all gmcs must be under similar surface pressures @xmath97 if , for example , @xmath102 is to be roughly the same among them .
other authors ( e.g. , mckee 1989 ; mouschovias 1987 ; myers & goodman 1988b ) have argued instead that either @xmath102 itself ( and hence @xmath93 ) , or the mean field strength @xmath103 , is the more fundamentally invariant attribute of gmcs .
if so , then @xmath97 could be eliminated from the critical bonnor - ebert relations in favor of any of these quantities , and our approach does not preclude the others . consider a cloud of radius @xmath36 , threaded by a mean magnetic field approximated as uniform and of magnitude @xmath103 . outside the cloud , let @xmath104 fall off as @xmath105 to a radius @xmath106 , where it matches onto an ambient , uniform field of strength @xmath107 ( nakano 1984 ) .
conservation of flux ( or continuity of the normal component of * b * across the boundary of the cloud ) demands @xmath108 , and evaluating equation ( [ eq:8 ] ) at the surface @xmath109 gives @xmath110 defining @xmath111 and @xmath112 , we have @xmath113 so that equation ( [ eq:11 ] ) gives @xmath114 with @xmath115 g and @xmath116 g for gmcs ( myers & goodman 1988b ) , flux conservation gives @xmath117 ; and @xmath118 implies @xmath119 .
thus , @xmath88 when @xmath83 , as expected . some indication of the reliability of equations ( [ eq:16 ] ) and ( [ eq:17 ] ) , and of the approximations leading to them , can be had by comparing the critical masses they predict for magnetized , isothermal spheres ( @xmath62 ) with those obtained from self - consistent , axisymmetric numerical calculations . in particular , the mass @xmath120 which separates magnetically sub- and supercritical clouds , and for which @xmath121 , is given by equation ( [ eq:17 ] ) as @xmath122 , only a 50% overestimate of the exact result @xmath123 ( mouschovias & spitzer 1976 ; tomisaka , ikeuchi , & nakamura 1988 ) .
further , for @xmath124 , the critical masses we obtain by using equation ( [ eq:11 ] ) in ( [ eq:16]a ) lie within a factor 2 of those found by tomisaka et al .
( 1988 ; see their eq . [ 4.7 ] ) .
( in fact , our formula is more accurate in the weak - field limit because it approaches the correct @xmath125 for the nonmagnetic , @xmath126 isothermal sphere . ) and for the equipartition @xmath127 seen in gmcs , setting @xmath90 in equation ( [ eq:17 ] ) implies @xmath128 .
this is roughly 1.6 times our ( approximate ) @xmath120 , and 2.4 times the exact value , which level of agreement is quite acceptable . in any case , we are led to expect that critical - mass gmcs and cores are strongly magnetically supercritical , with @xmath129 a result which has also been argued by mckee ( 1989 ) and bertoldi & mckee ( 1992 ) .
the outwards increase of linewidth both within giant molecular clouds as a whole ( larson 1981 ; miesch & bally 1994 ) and within individual dense cores ( fuller & myers 1992 ; caselli & myers 1995 ) , immediately suggests the class of negative - index polytropes as possible models for these structures .
that is , if @xmath130 , then for a polytropic index @xmath131 we have @xmath132 increasing for decreasing @xmath133 , as required .
such models have been studied by , e.g. , viala & horedt ( 1974 ) and maloney ( 1988 ) , and we derive the generalized bonnor - ebert relations for them in appendix c. although there are various physical arguments to support their use ( e.g. , shu et al .
1972 ; de jong , dalgarno , & boland 1980 ; mckee & zweibel 1995 ) , these polytropes turn out to be unlikely descriptions of real gmcs ( and of course , a positive polytropic index is undesirable because it is inconsistent with a velocity dispersion that increases with radius ) . here
we define @xmath134 , so that @xmath135 and @xmath136 for @xmath131 . from appendix c , a nonmagnetic , negative - index polytrope which is truncated anywhere outside of its constant - density central region ( inside which , @xmath137 and @xmath51 increases without bound towards @xmath138 )
will then satisfy @xmath139 thus , an @xmath136 polytrope has @xmath140 ; and since its density profile is @xmath141 , with @xmath142 everywhere for any @xmath143 ( appendix c ) , we also find @xmath144 . according to equation ( [ eq:14 ] ) , then ,
_ a truncated , negative - index polytrope will always be stable , and never critically so _
( as was also noted by maloney 1988 ) . ultimately , the same steady increase of linewidth with radius which would recommend the polytropic equation of state in the first place also proves to be its undoing : the internal pressure gradient which results is so shallow as to stabilize a cloud under any external pressure .
the concept of a critical mass is then irrelevant for these models , which casts doubt on their utility in describing real gmcs or high - mass cores .
any additional support from a mean magnetic field in the cloud obviously serves only to exacerbate this problem .
again , this result follows to some extent from the constraint that the thermal linewidth @xmath71 be fixed during a perturbation of the cloud .
by contrast , both viala & horedt ( 1974 ) and chize ( 1987 ) consider the possibility that the constant of proportionality in the relation @xmath135 is invariant , and find that polytropes can become unstable for certain truncation radii .
still , our approach is closely related to a point which is _ independent of any rule for cloud stability _ :
the minimum value of the nonmagnetic virial parameter for a truncated polytrope is ( eq . [ [ eq : b7 ] ] ) @xmath145 as usual , this is a lower limit because @xmath51 can be very large indeed if the polytrope is truncated at a very small radius . in the event of equipartition @xmath146 between kinematic and magnetic energies , equation ( [ eq:11 ] ) and the identity @xmath147 lead to @xmath148 in which case equation ( [ eq : b11 ] ) and @xmath136 finally imply that @xmath149 thus , regardless of whether there exist any unstable modes for truncated polytropes , the virial parameters of such clouds are significantly larger than the @xmath99 ( @xmath13 ) seen in gmcs ( myers & goodman 1988b ) and in the most massive molecular cores ( bertoldi & mckee 1992 ; williams et al . 1994 , 1995 ) , even if dynamically significant mean magnetic fields are allowed to be present .
it is a property of our specific stability criterion that this fact implies the absence of critically stable equilibria .
the virial parameters of magnetized polytropes could be reduced to @xmath99 , for any @xmath143 , if the mean field were such that @xmath150 . in this case , however , using equation ( [ eq : b11 ] ) for @xmath54 in the virial theorem ( [ eq:5 ] ) leads to @xmath151 for @xmath152 , which is difficult to reconcile with the rather higher degree of magnetic supercriticality that is observationally inferred for gmcs and massive cores ( @xmath153 , and @xmath154 ; see 2.2 , 2.3 ) .
one consequence of all of this is that the nonthermal linewidths in gmcs , @xmath155 , can not be attributed entirely to the pressure of weakly damped alfvn waves , for which @xmath156 ( @xmath157 , or @xmath158 ; mckee & zweibel 1995 ) , and thus @xmath159 under magnetic equipartition .
it seems almost certain that alfvn waves do play a significant role in the support of gmcs and cores ( e.g. , arons & max 1975 ; pudritz 1990 ) ; but , as mckee & zweibel ( 1995 ) also note , they can not be uniquely responsible for their large - scale stability . in this context
, we note that the size - linewidth relation between clouds can be expressed in terms of a mean magnetic field strength , as in , e.g. , myers & goodman ( 1988a ) .
specifically , equation ( [ eq:16]d ) can be used to write ( [ eq:16]b ) in terms of @xmath93 rather than @xmath97 , and the definition of @xmath160 ( 2.3 ) relates @xmath93 to @xmath103 . then , with a mean mass per particle @xmath161 and a kinetic temperature @xmath162 k , we have @xmath163 the equality is not quite exact here , because the scalings with @xmath103 and @xmath36 strictly apply to the total @xmath164 .
still , this relation is accurate in the typical case , @xmath165 ; and for @xmath166 , it is consistent with available data ( see myers & goodman 1988a ) .
thus , although mouschovias & psaltis ( 1995 ) argue for an interpretation of this result in terms of alfvn waves , we see here that it is independent of _ any _ assumptions on the physical origin of the nonthermal motions in gmcs ( in principle , they need not even derive from magnetic fields ) . finally , it is clear that any simple `` mixing '' of two polytropes with different indices will still preclude the existence of a critical ( or low-@xmath50 ) cloud ; even the superposition of an isothermal part , i.e. , @xmath167 , only admits one if it is essentially isothermal anyway ( @xmath168 ) .
thus , we now turn to a different equation of state , in which the gas pressure varies only logarithmically with density .
many theoretical models of star - forming clouds employ the singular isothermal sphere , in which @xmath169 . however , there is some evidence that regions are concentrated towards the co centroids of their parent gmcs such that their three - dimensional number density is most consistent with an @xmath170 fall - off ( waller et al .
1987 ; scoville et al .
this suggests the possibility that , in a heavily smoothed , average sense , the internal density structure of molecular clouds is essentially @xmath171 ( see also solomon et al .
further , extinction measurements indicate @xmath171 or so in the outer parts of cores as well ( cernicharo , bachiller , & duvert 1985 ; stwe 1990 ) .
lizano & shu ( 1989 ; see also gehman et al .
1996 ) introduced the so - called logotropic equation of state for gmcs , @xmath172 , with @xmath173 and @xmath174 .
if the central linewidth of a cloud is to be entirely thermal in origin , the reference density in @xmath175 must be @xmath28 .
however , @xmath132 then _ decreases _ with radius , which is incompatible with the observations .
we suggest that a more complete description of gmcs and cores is given instead by a `` pure '' logotrope : @xmath176\ , \label{eq:41}\ ] ] with @xmath177 a parameter to be adjusted .
the assumption here is that any nonthermal motions , which presumably arise from mhd turbulence , add to the thermal pressure such that both are fully accounted for by the equation of state ( [ eq:41 ] ) . this relation is shown schematically in fig .
[ fig0 ] , along with an isothermal - sphere combination and the polytropic @xmath23 for alfvn waves . in the rest of our discussion ,
equation ( [ eq:41 ] ) is referred to simply as a logotrope , with the understanding that no further consideration is given to any mixed - isothermal version . given equation ( [ eq:41 ] ) , the equation of hydrostatic equilibrium ( [ eq:1 ] ) integrates to @xmath178 where @xmath179 is a dimensionless gravitational potential .
substitution of this expression into poisson s equation ( [ eq:2 ] ) shows the existence of a singular solution , @xmath180 which describes the density profile of the bounded ( finite @xmath181 ) solution at large radii .
( the scale radius @xmath182^{1/2}$ ] . ) in spite of such a slowly decreasing internal density , a logotrope has only a finite extent , since for any positive @xmath2 the pressure will eventually vanish . assuming that @xmath2 is _ small _ enough for the singular density profile to apply at the edge of the cloud , we have @xmath183 the velocity dispersion , @xmath20 , increases with radius inside the cloud until it reaches a maximum of @xmath184 at @xmath185 , beyond which it falls again ( to 0 , at @xmath186 ) .
a logotrope truncated by the pressure @xmath97 of an external medium has precisely one critical mode for any given value of @xmath2 .
as discussed in appendix b , if a cloud is truncated at a very small radius @xmath187 , it will always be stable .
but in the power - law part of the logotrope , equations ( [ eq:42 ] ) and ( [ eq:43 ] ) show that @xmath188 , so once @xmath187 is so large that @xmath189 , the cloud is unstable ( cf .
[ [ eq : a5a ] ] ) .
thus , @xmath190 where we have again assumed that @xmath2 is relatively small and the singular solution ( [ eq:43 ] ) holds at @xmath191 .
( this expression shows that @xmath10 achieves its maximum _ inside _ @xmath191 , and the surface of the critical logotrope is cooler than its interior .
mckee [ 1989 ] has suggested that such behavior might arise from the radiation of alfvn waves into the ambient medium . ) now , evaluating the pressure ( [ eq:41 ] ) and density ( [ eq:43 ] ) at the radius ( [ eq:45 ] ) yields , for any critically stable logotrope , @xmath192 evidently , self - consistency in the use of equation ( [ eq:43 ] ) here requires @xmath193 .
equations ( [ eq:4d ] ) and ( [ eq:6 ] ) can then be used to find @xmath194 independently of @xmath2 .
a numerical solution for our logotrope with @xmath195 ( which we justify shortly ) yields the internal density , pressure , and velocity - dispersion profiles shown in fig .
[ fig1 ] . equations ( [ eq:44 ] ) and ( [ eq:45 ] ) are valid here , as an @xmath170 density profile
is realized well before the cloud ends ( the vertical line in all four panels of the figure marks the radius @xmath191 ) .
it is the steep pressure gradient near @xmath186 which sets the logotrope apart from the negative - index polytropes and allows for a critical cloud with a small virial parameter .
the _ total _ velocity dispersion @xmath21 in fig .
[ fig1 ] is roughly constant near the center of the cloud ; and its rise with radius further on is consistent with @xmath196 , which is essentially the scaling originally found for gmcs by larson ( 1981 ) .
identifying the central dispersion @xmath71 with the thermal part of the linewidth ( so that thermal motions dominate on the smallest scales , as is observed ) , the nonthermal contribution to cloud support is @xmath197 , which is also shown in fig .
[ fig1 ] . at small to moderate radii
, @xmath198 naturally rises more steeply than the total @xmath10 , while at larger @xmath187 the two are more comparable in magnitude . roughly , then
, our model has @xmath199 over some range in radius before it flattens to @xmath200 and eventually turns over ( the maximum @xmath201 is just @xmath202 , and occurs at @xmath203 ) .
it is found that @xmath11 within gmcs ( e.g. , myers & goodman 1988b ) , and low - mass molecular cores show the same general trend ; high - mass cores are more suggestive of @xmath204 ( fuller & myers 1992 ; caselli & myers 1995 ) .
a logotropic equation of state both for individual cores and for entire cloud complexes is consistent with these observations , provided that low - mass cores can be viewed as spheres which are pressure - truncated at radii @xmath205 .
we discuss this point further in 4.1 below .
as always , though , no matter how @xmath10 varies with @xmath206 internally , the _ average _ relation @xmath207 still holds for critical - mass clouds , i.e. , for gmcs . a handle on an appropriate value of the parameter @xmath2 for gmcs as a whole , can be gotten by reinventing them as smoothed - out spheres with a number of distinct overdense regions ( the cores ) sprinkled throughout . the observed contrasts in mean density and velocity dispersion between gmcs and cores
can then be related to the @xmath2 of a model logotrope .
harris & pudritz ( 1994 ; see their table 2 ) list each of @xmath208 , @xmath164 , and @xmath102 for a gmc of mass @xmath209 and a core with @xmath210 .
( such values are `` typical '' in the sense that , _ by mass _ , half of all gmcs [ cores ] are larger than @xmath211 [ @xmath212 . ) defining @xmath213 , @xmath214 , and @xmath215 as the core - to - gmc ratios of mean volume densities , linewidths , and column densities , these data show that @xmath216 cores with @xmath217 are among the most massive observed , and are likely near critical . if they are also logotropes , then their @xmath218 , etc .
, should be roughly given by equation ( [ eq:48 ] ) , regardless of whether @xmath219 .
such large cores might further be expected to lie in very dense , cold , highly pressured regions of their parent clouds .
then the pressure @xmath97 at the surface of a core is comparable to the gmc s @xmath220 , and similarly for the density and velocity dispersion .
these assumptions , together with the relation @xmath221 ( eq . [ [ eq:16]c ] ) , result in @xmath222 comparison of equations ( [ eq:46 ] ) and ( [ eq:47 ] ) then shows that the observed @xmath213 , @xmath223 , and @xmath224 ( only two of which are actually independent ) are realized in a logotropic gmc with @xmath225 although this procedure is highly idealized , it is perhaps telling that an @xmath2 exists at all which gives at once the correct @xmath226 and @xmath214 .
the logotrope appears to be the simplest barotropic equation of state which can account for these ratios _ and _ the small virial parameters ( or the critical equilibrium ) of gmcs and massive cores .
for example , if gmcs and large cores were both @xmath157 polytropes truncated at @xmath227 , then equation ( [ eq:46 ] ) could be roughly satisfied , but they would also have a large @xmath228 , and hence @xmath229 for magnetic equipartition ( again , @xmath150 is required to reduce @xmath50 to unity for a negative - index polytrope , but this is not wholly satisfactory for other reasons ; see 3 ) . alternatively , adding an explicit @xmath230 term to equation ( [ eq:41 ] ) destroys the _ simultaneous _ correspondence between @xmath213 and @xmath223 for critical clouds . to this point , we have not accounted for any mean magnetic fields in gmcs and cores .
however , these should have no effect on our estimate of @xmath231 insofar as very massive clumps have the same @xmath88 as their parent clouds .
more important is the assumption that the cores here lie at the rare , highest - pressure peaks of cloud complexes ; if instead they are found in less extreme regions , then equation ( [ eq:410 ] ) is an upper limit to @xmath231 .
even so , more direct estimates of @xmath231 are in fairly good agreement with the representative value quoted above .
for example , equations ( [ eq:47 ] ) and ( [ eq:48 ] ) can be combined to obtain an expression for @xmath232 ; comparison with observed values of the total @xmath164 and the thermal @xmath71 in gmcs then suggests @xmath233 . on a final note , since @xmath171 at the edges of such logotropes , we have @xmath234 . with the ratio of mean kinematic and magnetic pressures @xmath119
, equations ( [ eq:15 ] ) and ( [ eq:18 ] ) then give for the critical virial parameters , @xmath235 these are near 2 and 1 , as expected from general arguments , and not far from the values 2.054 and 1.15 for a bonnor - ebert isothermal sphere .
as alluded to earlier ( 2.2 ) , the mass - radius and size - linewidth relations among real gmcs can therefore be understood in terms of this model : equations ( [ eq:16]b , c ) become @xmath236 and @xmath237 which agree well with observations ( cf .
elmegreen 1989 ) . here
@xmath97 is the surface pressure on the _ molecular part _ of the gmc , while @xmath238 is the total pressure of the hot , intercloud medium .
the ratio of these two is typically around @xmath239 , due to the weight placed on a gmc by its uv - shielding atomic layer ( elmegreen 1989 ; also 2.2 above ) . as a class ,
dense molecular cores seem somewhat more heterogeneous than their parent gmcs .
first , many studies ( carr 1987 ; loren 1989 ; stutzki & gsten 1990 ; lada , bally , & stark 1991 ) show that the size - linewidth relation between cores can be significantly weaker than the virial , @xmath8 scaling among gmcs , and in some cases is difficult to discern at all .
second , in the l1630 gmc at least , the size - linewidth relation is considerably better defined for clumps which stand out more strongly against the interclump medium ( see the comparison in fig .
13 of lada et al . between their `` 5@xmath10 '' and `` 3@xmath10 '' cores ) . and third
, the virial parameters @xmath50 of gmc cores are not restricted to values near 1 , but rather can range as high as @xmath240100 ( bertoldi & mckee 1992 ; williams et al .
1994 , 1995 ) .
this last point is particularly significant because the virial parameters of cores in a number of different clouds are seen to correlate with their size : low mass goes along with rather large @xmath50 , and high mass with more moderate @xmath88 .
thus , as bertoldi & mckee ( 1992 ) argue , the smallest clumps in gmcs are far removed from gravitational instability ( surface pressure is more important than self - gravity in their confinement ) , while the largest are actually near criticality ( strongly self - gravitating ; see also 2.2 above ) .
equivalently , a range in the masses of cores can be viewed , at least roughly , as a range in the dimensionless radii @xmath187 small for low mass , and near @xmath191 for high mass at which they are truncated by the ambient pressure in a gmc ( virial parameters generally decrease with @xmath187 for any equation of state ; appendix a ) .
this notion could account for the rather confused size - linewidth relation _ between _ cores ( if @xmath187 does not have a fixed value , then neither do @xmath51 and @xmath50 in eq .
[ [ eq:12]b ] ) , and ultimately must also have consequences for the interpretation of data on the radial variation of velocity dispersion _ inside _ cores . if indeed very low - mass cores differ from very high - mass ones mainly in being truncated at @xmath241 , then the former must be less centrally condensed , have ratios @xmath242 nearer unity , and be generally less distinguishable from the intercore ( gmc ) gas than the latter .
this results in the scale radius @xmath243 being _ larger _ for less massive clumps . to see this ,
compare the @xmath42 of a low - mass core , truncated at @xmath187 small enough that @xmath64 , with the @xmath42 of a near - critical , high - mass core . on average
, any two cores should be under similar surface pressures @xmath97 ( this essentially being set by the internal @xmath93 of a parent gmc ) , so that @xmath244\ \simeq\ \sqrt{{a\over{4}}}\ \left[{{\sigma_c^2{\rm{(high)}}}\over{\sigma_c^2{\rm{(low)}}}}\right]\ , \label{eq:412}\ ] ] where the second step follows from equation ( [ eq:47 ] ) if a logotropic equation of state applies .
thus , for @xmath71 not too widely different between the two cores , and for @xmath2 small enough , the ratio ( [ eq:412 ] ) will be @xmath245 .
the same _ physical _ radius @xmath206 in high- and low - mass cores then corresponds to respectively larger and smaller @xmath246 .
this result , which just reflects the fact that more strongly self - gravitating structures are more centrally concentrated , must be considered when attempting to match any model to any observations .
returning now to the issue of internal velocity - dispersion profiles , fuller & myers ( 1992 ; hereafter fm ) and caselli & myers ( 1995 ; hereafter cm ) have compiled linewidth measurements in at least three different molecular lines for each of 14 low - mass and 24 massive cores .
the resulting @xmath10 vs. @xmath206 profiles indicate that , over similar ranges @xmath247 pc in all the clumps , and independently of whether or not stars are present within them , the velocity dispersion rises more steeply with radius in low - mass cores than in high - mass ones .
in particular , nonthermal linewidths in the former are consistent with @xmath248 ; in the latter , @xmath249 ( cm ) . in our view , this situation is a direct result of the effect summarized by equation ( [ eq:412 ] ) .
that is , we consider the low - mass cores of fm to be eminently stable logotropes truncated at small @xmath187 , while the high - mass clumps of cm are near critical stability and hence have smaller @xmath42 .
the two surveys target similar @xmath206 within each type of core , so the massive - core data apply to larger dimensionless @xmath187 , and must sample the shallower parts of velocity - dispersion profiles like those in fig . [ fig1 ] . to actually confront the predictions of our logotropic equation of state with the observations of fm and cm
, we again identify the ( model ) central velocity dispersion with the ( observed ) thermal linewidth : @xmath250 , where @xmath251 is taken from cm , and @xmath252 .
cm assume @xmath162 k for all the low - mass cores , and we further assume them all to have @xmath64 ; they should therefore have a common @xmath42 , which may be easily estimated .
five of the fm cores are observed at radii @xmath253 such that @xmath254 ; among these , @xmath255 pc ( see table 3 of cm ) .
the related quantity @xmath256 in a model logotrope is just that point where @xmath257 .
if , for example , @xmath258 in equation ( [ eq:41 ] ) , we find @xmath259 , and thus @xmath260 this result is then used in equation ( [ eq:412 ] ) , along with @xmath261 and @xmath258 , to give @xmath262 where @xmath263 for the massive cores ( cm ) .
the data may now be appropriately scaled and compared to theoretical @xmath264 and @xmath265 curves .
figure [ fig2 ] is the result of such a comparison with three logotropes of different @xmath2 .
the good overall agreement between the models and data suggests that gmc cores are well described by @xmath266 which is encouragingly close to @xmath231 as given in equation ( [ eq:410 ] ) above . in table
[ tbl1 ] we list the radii and other properties ( as calculated from eqs .
[ [ eq:44 ] ] to [ [ eq:47 ] ] above ) of critical logotropes with @xmath2 near 0.2 .
some of the scatter in fig .
[ fig2 ] could arise if @xmath2 does not have a common value for all cores , and any embedded stars could further complicate the situation .
( notably , fm and cm both find that starless cores alone show a tighter but not different linewidth - radius relation than do cores with stars , suggesting that the basic form seen here is one of the initial conditions for star formation .
) however , some of the scatter is surely due to the fact that the cores in this sample do not reside in a single gmc , but come from diverse environments .
( for instance , the low - mass cores here tend to be found in dark clouds of mass @xmath267 , and the high - mass ones in somehat larger gmcs . )
thus , there is no guarantee that every low - mass core observes @xmath64 , or that every high - mass core is at its critical mass .
it is the large number of cores available here which allows the mean trend in fig .
[ fig2 ] , and the implied structural dichotomy between small and large clumps , to emerge . as with our earlier consideration of core - to - gmc density and linewidth ratios , a fit of similar quality to the data in fig .
[ fig2 ] can be obtained with the model velocity - dispersion profile of an @xmath157 polytrope which has a scale @xmath42 about a factor of two smaller than that used here . as we have stressed , however , there are other difficulties with the application of polytropic equations of state to gmcs and cores . in summary ,
the approach we have taken self - consistently allows for purely thermal motions to dominate on the smallest scales in cores ( and in gmcs ) ; for an internal density structure ( @xmath171 in the outer parts of cores , and much shallower near their centers ) which is consistent with at least some observations ( cernicharo et al .
1985 ; stwe 1990 ; andr , ward - thompson , & barsony 1993 ) ; and for a unified treatment of low- and high - mass cores .
this is in contrast to the models of fm and cm ( also myers & fuller 1992 ) , which assume two distinct components of isothermal and nonthermal gas in cores and predict a @xmath169 singularity at their centers .
it should also be noted that the assumption @xmath64 in low - mass cores still allows for these noncritical clumps to show a fair degree of central concentration .
for example , an @xmath258 logotrope which is truncated at @xmath268 has @xmath269 , but @xmath270 ; thanks to the relatively weak dependence of @xmath31 on @xmath133 , we do not require , nor even expect , that all low - mass cores be uniform - density spheres .
our model is therefore consistent with the significantly non - uniform density profiles of low - mass cores ( @xmath271 ) in the rosette gmc ( williams et al .
we have generalized the classic bonnor - ebert relations for isothermal spheres , to give expressions for the masses , radii , and total velocity dispersions of magnetized , pressure - truncated gas clouds with any internal pressure - density relation .
the analysis combines an exact approach , based on solving the equation of hydrostatic equilibrium , with a virial - theorem treatment to incorporate mean magnetic fields into the clouds .
a stability criterion has been developed that relies only on the assumption of an invariant central velocity dispersion @xmath71 , and is independent of the gas equation of state .
clouds which are just _ critically _ stable under this condition have mass - radius and size - linewidth relations which are effectively oblivious to their internal structure , i.e. , to their equation of state . these results have been applied to molecular cloud complexes and dense cores , leading to three main conclusions : \(1 ) in spite of their nonspherical geometry , the gross features of gmcs are consistent with those of critical - mass clouds in our analysis .
( in the context of gmcs and cores , we identify the central velocity dispersion with the _ thermal _ part of the total [ thermal plus turbulent ] linewidth . ) _ independently of the exact equation of state _ , if equipartition between magnetic and kinetic energies obtains in a critically stable cloud , then it will appear to be in simple gravitational equilibrium , @xmath89 ( equivalently , it will have a virial parameter @xmath272 )
. these are observed features of real gmcs and their most massive cores .
in addition , any of our critical - mass spheres obey the same mass - radius and size - linewidth relations which hold among real gmcs .
all of this implies that gmcs , and large clumps , are near criticality .
\(2 ) no polytropic ( power - law ) pressure - density relation can fully characterize gmcs or cores .
a positive polytropic index leads to a velocity dispersion which decreases outwards within a cloud , opposite to what is seen both in individual cores and in entire cloud complexes .
alternatively , a negative index ( i.e. , @xmath17 with @xmath273 ) gives a linewidth which increases with radius , as required .
however , such clouds do not show virial parameters near unity unless their mean fields are so strong that their magnetic and gravitational energies are comparable in magnitude , and this is again in conflict with observations of gmcs and large cores .
given our stability criterion , an equivalent statement is that pressure - truncated polytropes are unconditionally stable against gravitational collapse .
since weakly damped alfvn waves satisfy @xmath23 ( mckee & zweibel 1995 ) , this shows explicitly that they can not fully explain the observed nonthermal support in molecular clouds .
\(3 ) the most successful model for the internal structure of molecular cores , and one which is also consistent with global properties of gmcs , is a `` pure logotrope : '' @xmath24 .
this relation is meant to account for _ all _ contributions to the total gas pressure , including the effects of disordered magnetic fields ( mhd turbulence ) .
once it is recognized that low - mass cores are far below their critical masses ( though still in virial equilibrium ; they are essentially pressure - confined ) , and that high - mass cores are near criticality , the internal velocity - dispersion profiles of clumps from a variety of environments are seen to be consistent with a logotropic model with @xmath3 .
a similar value of @xmath2 can also explain characteristic core - to - gmc ratios of mean densities and linewidths , and any @xmath2 is consistent with the observed size - linewidth relation between gmcs because the logotropic equation of state allows for critical equilibria . finally , the equilibrium density profile of a logotrope has @xmath171 , outside of a constant - density central region .
there is some observational support to be found for this prediction , in gmcs and cores both .
we would like to thank phil myers for his comments on an earlier version of this paper , and charles curry for valuable discussions .
r.e.p . also acknowledges conversations with paola caselli and gary fuller .
this work was supported in part by the natural sciences and engineering research council of canada .
for a spherical cloud in hydrostatic equilibrium , with kinematic ( thermal plus turbulent ) pressure @xmath274 ( so that @xmath10 is the one - dimensional velocity dispersion ) and a self - gravitational potential @xmath275 , we define @xmath276 where a subscript @xmath277 denotes evaluation of a quantity at the cloud center .
the characteristic scale of a cloud is set by @xmath42 ; the factor of 9 in its definition identifies it with the projected half - power radius of an isothermal sphere ( e.g. , binney & tremaine 1987 ) .
the equation of hydrostatic equilibrium then reads @xmath278 and poisson s equation becomes @xmath279 ( note that @xmath280 at @xmath281 , and @xmath282 for @xmath283 . ) the mass enclosed within radius @xmath187 is given by @xmath284 so that , with the help of equation ( [ eq:2 ] ) , @xmath285 following ebert ( 1955 ) and bonnor ( 1956 ) , the surface of the cloud is defined by that radius @xmath187 at which the internal pressure @xmath286 just equals a confining pressure @xmath97 due to a surrounding medium of negligible gravity .
equation ( [ eq:3 ] ) , the definition of @xmath42 , and the identity @xmath287 then yield , for any equation of state relating @xmath31 and @xmath133 , @xmath288 the physical radius of the cloud is @xmath289 : @xmath290 these relations combine to give @xmath291 and @xmath292 given an equation of state , the integration of equations ( [ eq:1 ] ) and ( [ eq:2 ] ) determines @xmath293 and @xmath242 at any truncation radius @xmath187 .
this fixes the coefficients in equations ( [ eq:4a])([eq:4d ] ) and affords mass - radius and size - linewidth relations for pressure - truncated clouds .
( note that the scalings in these relations @xmath294 , @xmath295 , etc .
are _ global _ ones , and do not necessarily reflect the radial dependence of any quantity _ inside _ a cloud . )
the definition of the non - uniformity parameter @xmath41 ( eq . [ [ eq:9 ] ] ) allows it to be found as a function of radius : @xmath296 and equation ( [ eq:3 ] ) implies that the virial parameter @xmath297 for nonmagnetic clouds ( i.e. , @xmath53 ) is just @xmath298 which tends to be a nonincreasing function of @xmath187 .
alternatively , equations ( [ eq:4d ] ) and ( [ eq:12]d ) give @xmath299 these results can be used to rewrite equations ( [ eq:4a])([eq:4d ] ) in terms of @xmath51 , @xmath41 , and @xmath164 ; equations ( [ eq:12 ] ) ( with @xmath300 ) are then obtained .
also , once @xmath51 is known as a function of radius in any cloud , the observable @xmath50 for a given magnetic field configuration may be calculated ( e.g. , 2.3 ) .
we begin by recalling the definition of the scale radius @xmath42 from equation ( [ eq:3 ] ) : @xmath301 and rearranging equation ( [ eq:4a ] ) : @xmath302 the stability of a cloud truncated at radius @xmath36 by the external pressure @xmath97 is determined by the sign of the derivative @xmath303 : this is negative for a stable equilibrium , 0 for the `` critical '' cloud , and positive for instability .
we now apply a perturbation @xmath304 , while keeping the mass @xmath72 and central ( i.e. , thermal ) velocity dispersion @xmath71 of the cloud fixed , so that @xmath305 then @xmath289 and equation ( [ eq : a1 ] ) give us @xmath306^{-1 } \nonumber \\ & = & { 1\over{r_0}}\left[1-\xi\left(\xi^2\dpsi\right)^{-1 } { d\over{d\xi}}\left(\xi^2\dpsi\right)\right]^{-1 } \nonumber \\ & = & { 1\over{r_0}}\left[1 - 9\xi\left(\dpsi\right)^{-1}{{\rho_s}\over{\rho_c } } \right]^{-1 } , \label{eq : a4}\end{aligned}\ ] ] where @xmath75 is the density just inside the cloud edge and we have used poisson s equation ( [ eq:2 ] ) .
we also have , from equation ( [ eq : a2 ] ) , @xmath307 \nonumber \\ & = & p_s\left(\dpsi\right)^{-1}\left({{\rho_s}\over{\rho_c}}\right)\left [ 18-\left(\dpsi\right)^2\left({{p_s}\over{p_c}}\right)^{-1}\right ] , \label{eq : a5}\end{aligned}\ ] ] where both equations ( [ eq:1 ] ) and ( [ eq:2 ] ) have been used .
thus , equations ( [ eq : a3 ] ) , ( [ eq : a4 ] ) , and ( [ eq : a5 ] ) show that @xmath308 .
\label{eq : a5a}\ ] ] now , from equations ( [ eq:13 ] ) and ( [ eq:4d ] ) , we have @xmath309 so that finally , @xmath310 .
\label{eq : a7}\ ] ] alternatively , for a _ nonmagnetic _
cloud , we can make use of equation ( [ eq:12]c ) , with @xmath300 , to write @xmath311 .
\label{eq : a6}\ ] ] this treatment is valid for _ any _ gas equation of state , so long as the cloud radius is perturbed in such a way that @xmath72 and @xmath71 are unaffected .
the result ( [ eq : a6 ] ) was also obtained by ebert ( 1955 ) and bonnor ( 1956 ) , who specifically considered the stability of truncated isothermal spheres , and by maloney ( 1988 ) in his study of negative - index polytropes . any virialized cloud truncated at small @xmath187 ( i.e. , where @xmath133 and @xmath21 are approximately constant ) will be stable .
this is because , regardless of the equation of state , the potential @xmath312 at small radii can always be expanded in a power series of the form : @xmath313 ( see appendix c for an example ) .
thus , referring to equation ( [ eq : a5a ] ) with @xmath314 , we have @xmath315 .
depending on the equation of state , there may or may not be a truncation at larger @xmath187 which results in @xmath316 and an unstable equilibrium .
consider the ( nonmagnetic ) equation of state @xmath317 where the polytropic index @xmath131 .
then we define @xmath134 and write @xmath318 with @xmath319 corresponding to isothermality . in order to evaluate the coefficients in equations ( [ eq:4a])([eq:4d ] ) ( or [ [ eq:12 ] ] , with @xmath300 )
, we first use the relation ( [ eq : b1 ] ) to integrate the equation of hydrostatic equilibrium ( [ eq:1 ] ) and find @xmath320^{n/(1-n ) } , \label{eq : b2}\ ] ] so that poisson s equation ( [ eq:2 ] ) becomes @xmath321^{n/(1-n)}. \label{eq : b3}\ ] ] ( note that in the limit @xmath322 , the right - hand side of eq . [ [ eq : b3 ] ] becomes @xmath323 , just what is required for the isothermal sphere . ) at the cloud center , @xmath280 , so for small @xmath187 the potential @xmath312 can be expanded in a power series , @xmath324 allowing for an integration of equation ( [ eq : b3 ] ) .
the result of this is a core - halo structure to the cloud : at very small radii , the solution is indistinguishable from that for an isothermal sphere , and the density and velocity dispersion are essentially constant with @xmath187 , independently of @xmath143 .
once @xmath325 a few , however , a power - law density profile obtains ; for each @xmath143 , there exists a singular solution to which the equilibrium cloud tends at large radii .
these singular solutions are readily found by substituting the prescription @xmath326 into equation ( [ eq : b2 ] ) and solving ( [ eq : b3 ] ) for @xmath327 and @xmath328 .
this gives @xmath329^{p/2 } , \label{eq : b5}\ ] ] so that with @xmath330 we have @xmath331 , and these clouds are diffuse enough that they have no `` natural '' edge .
that is , the density never vanishes , and any boundary to the cloud must be defined by the point at which the internal pressure matches that of an external medium .
( it is again worth noting the exact correspondence to the singular isothermal sphere when @xmath319 in eq .
[ [ eq : b5 ] ] ; cf .
chandrasekhar 1967 . )
consider now a polytrope of negative index , which is truncated at radius @xmath187 large enough that the structure of the cloud is given by the singular solution .
then @xmath293 and @xmath242 can be calculated analytically , and equation ( [ eq:13 ] ) yields @xmath332 we also know that @xmath44 , so @xmath333 and we have the following : @xmath334 @xmath335 @xmath336 and @xmath337 thus , regardless of where they are truncated by the surface pressure @xmath97 , these model clouds can only ever represent modest enhancements over the intercloud medium . in particular , @xmath338 has a maximum of 3 ( for isothermality , @xmath319 ) , but even for @xmath157 ( which gives the @xmath133 dependence of alfvn wave pressure ; mckee & zweibel 1995 ) is reduced to just 9/7 .
as discussed in 3 , any truncation of a negative - index polytrope results in a stable cloud .
this is because , with @xmath141 , we have that @xmath339 , and the stability criterion ( [ eq : a7 ] ) becomes ( with the help of [ [ eq : b6 ] ] ) @xmath340 . in particular , this holds in the limit @xmath322 , when the numerator and denominator of ( [ eq : a7 ] ) both vanish .
what allows for the existence of critical and unstable equilibria for the bounded isothermal sphere , then , is the fact that its density profile actually oscillates about the singular solution @xmath169 at large radii ( chandrasekhar 1967 ) .
this does _ not _ occur for those polytropes with @xmath136 ; instead , the bounded spheres follow the singular solutions exactly at large @xmath187 . finally , it is worth noting that these results do not contradict the well known instability ( e.g. , shu 1977 ) of truly singular spheres , for which @xmath141 all the way to the center and @xmath341 . in such cases , perturbation of the truncated cloud radius
is performed with @xmath72 , @xmath71 , _ and _
@xmath181 all held fixed .
stability then depends only on the sign of @xmath342 ( eq .
[ [ eq : a5 ] ] ) , which is always positive for these polytropes .
andr , p. , ward - thompson , d. , & barsony , m. 1993 , apj , 406 , 122 arons , j. , & max , c. e. 1975 , apj , 196 , l77 binney , j. , & tremaine , s. 1987 , galactic dynamics ( princeton : princeton univ . press ) bertoldi , f. , & mckee , c. f. 1992 , apj , 395 , 140 bonnor , w. b. 1956 , mnras , 116 , 351 carr , j. s. 1987 , apj , 323 , 170 caselli , p. , & myers , p. c. 1995 , apj , 446 , 665 ( cm ) cernicharo , j. , bachiller , r. , & duvert , g. 1985 , a&a , 149 , 273 chandrasekhar , s. 1967 , an introduction to the study of stellar structure ( new york : dover ) chize , j. p. 1987
, a&a , 171 , 225 de jong , t. , dalgarno , a. , & boland , w. 1980 , a&a 91 , 68 ebert , r. 1955 , zap , 37 , 217 elmegreen , b. g. 1985 , in protostars & planets ii , ed . d. black & m. matthews ( tucson : univ . of arizona press ) , p. 33 elmegreen , b. g. 1989 , apj , 338 , 178 falgarone , e. , phillips , t. g. , & walker , c. k. 1991 , apj , 378 , 186 fleck , r. c. , jr . 1988 , apj , 328 , 299 fuller , g. a. , & myers , p. c. 1992 , apj , 384 , 523 ( fm ) gehman , c. s. , adams , f. c. , fatuzzo , m. , & watkins , r. 1996 , apj , 457 , 718 harris , w. e. , & pudritz , r. e. 1994 , apj , 429 , 177 heiles , c. , goodman , a. , mckee , c. f. , & zweibel , e. g. 1993 , in protostars and planets iii , ed .
e. h. levy & j. i. lunine ( tucson : univ . of arizona press ) , p. 279
henriksen , r. n. 1991 , apj , 377 , 500 lada , e. a. , bally , j. , & stark , a. a. 1991 , apj , 368 , 432 larson , r. b. 1981 , mnras , 194 , 809 lizano , s. , & shu , f. h. 1989 , apj , 342 , 834 loren , r. b. 1989 , apj , 338 , 925 maloney , p. 1988 ,
apj , 334 , 761 mccrea , w. h. 1957 , mnras , 117 , 562 mckee , c. f. 1989 , apj , 345 , 782 mckee , c. f. , & zweibel , e. g. 1992 , apj , 399 , 551 mckee , c. f. , & zweibel , e. g. 1995 , apj , 440 , 686 miesch , m. s. , & bally , j. 1994 , apj , 429 , 645 mouschovias , t. ch .
1987 , in physical processes in interstellar clouds , ed .
g. e. morfill & m. scholer ( dordrecht : reidel ) , p. 453
mouschovias , t. ch . ,
& psaltis , d. 1995 , apj , 444 , l105 mouschovias , t. ch . , & spitzer , l. 1976 , apj , 210 , 326 myers , p. c. , & fuller , g. a. 1992 , apj , 396 , 631 myers , p. c. , & goodman , a. a. 1988a , apj , 326 , l27 myers , p. c. , & goodman , a. a. 1988b , apj , 329 , 392 myers , p. c. , fuller , g. a. , goodman , a. a. , & benson , p. j. 1991 , apj , 376 , 561 nakano , t. 1984 , fund .
cosmic phys . , 9 , 139 pudritz , r. e. 1990 , apj , 350 , 195 sanders , d. b. , scoville , n. z. , & solomon , p. m. 1985 , apj , 289 , 373 scoville , n. z. , yun , m. s. , clemens , d. p. , sanders , d. b. , & waller , w. h. 1987 , apjs , 63 , 821 shu , f. h. 1977 , apj , 214 , 488 shu , f. h. , milione , v. , gebel , w. , yuan , c. , goldsmith , d. w. , & roberts , w. w. 1972 , apj , 173 , 557 solomon , p. m. , rivolo , a. r. , barrett , j. , & yahil , a. 1987 , apj , 319 , 730 stutzki , j. , & gsten , r. 1990 , apj , 356 , 513 stwe , j. a. 1990 , a&a , 237 , 178 tomisaka , k. , ikeuchi , s. , & nakamura , t. 1988 , apj , 335 , 239 viala , y. p. , & horedt , g. p. 1974 , a&a , 33 , 195 waller , w. h. , clemens , d. p. , sanders , d. b. , & scoville , n. 1987 , apj , 314 , 397 williams , j. p. , blitz , l. , & stark , a. a. 1995 , apj , 451 , 252 williams , j. p. , de geus , e. j. , & blitz , l. 1994 , apj , 428 , 693 | we generalize the classic bonnor - ebert stability analysis of pressure - truncated , self - gravitating gas spheres , to include clouds with arbitrary equations of state .
a virial - theorem analysis is also used to incorporate mean magnetic fields into such structures .
the results are applied to giant molecular clouds ( gmcs ) , and to individual dense cores , with an eye to accounting for recent observations of the internal velocity - dispersion profiles of the cores in particular .
we argue that gmcs and massive cores are at or near their critical mass , and that in such a case the size - linewidth and mass - radius relations between them are only weakly dependent on their internal structures ; any gas equation of state leads to essentially the same relations .
we briefly consider the possibility that molecular clouds can be described by polytropic pressure - density relations ( of either positive or negative index ) , but show that these are inconsistent with the apparent gravitational virial equilibrium , @xmath0 , of gmcs and of massive cores .
this class of models would include clouds whose nonthermal support comes entirely from alfvn wave pressure .
the simplest model consistent with all the salient features of gmcs and cores is a `` pure logotrope , '' in which @xmath1 .
detailed comparisons with data are made to estimate the value of @xmath2 , and an excellent fit to the observed dependence of velocity dispersion on radius in cores is obtained with @xmath3 . |
as the mechanical properties of dna play an important role in its biological capacities , there has been much activity to accurately characterize these properties , not only in the elastic regime of small to moderate fluctuations but also for more strongly stressed systems .
for example , dna is found to overstretch beyond a salt - dependent critical force @xcite .
similarly , in response to twist dna forms plectonemes beyond a critical buckling superhelical density @xcite . here , we are interested in the response of dna to strong bending .
the worm - like chain ( wlc ) model provides a good description of small to moderate bending fluctuations in dna @xcite .
however , although there is a consensus that for sufficiently strong bending the stress will be localized within small regions , often termed `` kinks '' , much about this crossover to non - wlc behaviour remains controversial .
for example , a recent review by vologodskii _
et al . _
@xcite highlighted a number of open questions , including what is the free energy cost of kink formation , how does the free - energy of a kink depend on bend angle , what is the critical curvature that causes the double helix to kink ?
one way to address these questions is with molecular simulations of a coarse - grained dna model , as such simulations are able to directly probe the relevant factors . in this and an accompanying paper @xcite we have begun this task for the oxdna model @xcite , which is able to describe well the thermodynamics of dna hybridization and basic mechanical properties such as the persistence length and the torsional modulus @xcite and which has been productively applied to study a wide variety of biophysical and nanotechnological systems @xcite . in the first paper , we addressed the thermodynamics of kink formation and one particular experimental system , a `` molecular vice '' @xcite , that probed dna strong bending . here
, we consider dna cyclization , in particular looking at the length of dna molecules at which duplex kinking begins to play a role in this process .
dna cyclization is a convenient model system used to probe dna bending .
cyclization experiments were first reported in 1966 , albeit on 48500 base pair ( bp ) @xmath1-dna @xcite . in 1981 , shore _
_ developed a method to probe the bending of shorter fragments @xcite , later noting periodicity in the cyclization efficiency of fragments @xcite .
more recently , there has been a particular interest in probing the cyclization of sub - persistence length dna , to explore whether this regime is accurately described by the wlc model .
for example , in 2004 cloutier & widom ( c&w ) @xcite challenged the conventional wisdom of wlc flexibility established by shore _
et al . _
@xcite , claiming much greater than predicted cyclization efficiency in dna fragments .
this controversial finding spurred debate on the characteristic length at which dna cyclization efficiency deviates from the predictions of the wlc model . despite much experimental @xcite and theoretical effort @xcite ,
a consensus has not yet been established . a typical cyclization experiment , as depicted in ( a ) , uses a cyclization substrate with complementary sticky ends , @xmath2 bases in length , on both ends of a @xmath3 base - pair duplex .
cyclization leads to the formation of a @xmath4-base - pair duplex , where @xmath5 .
the resultant structure is not a closed minicircle two backbone ` nicks ' are present at either end of the sticky ends . either the forward rate or equilibrium constant of the cyclization reaction is reported .
experiments differ in how exactly they probe cyclization : methods based on ligation @xcite , fret @xcite and multimerization @xcite have been reported .
and @xmath6 are the forward and reverse rate constants respectively , and * ( b ) * dimerization where the rate constants are @xmath7 and @xmath8 .
note that for the dimerization system there is only one complementary sticky end per monomer , the other end being blunt to allow for only one reaction product , a linear dimer .
figures are oxdna representations for monomers of length @xmath9 , including complementary sticky ends of length @xmath10 .
( dimer length is @xmath11 ) . ,
width=321 ] in ligase - based experiments , cyclized molecules are fixed in the cyclized state by ligation of the two backbone nicks .
the open and ligated cyclized molecules can then be resolved by gel electrophoresis and the concentration of different products measured .
fret - based experiments can be performed in equilibrium , with the molecules allowed to cyclize and uncyclize indefinitely .
fluorophores are attached to both ends of the molecule as fret reporters : a high fret signal will be reported when the duplex ends are in close proximity ( cyclized ) , low fret when apart ( open ) .
although non - wlc behaviour has been suggested by the ligase - based experiments of c&w @xcite and the fret - based experiments of vafabakhsh & ha ( v&h ) @xcite , these results have been contested @xcite .
there is not yet a consensus on their interpretation .
cyclization efficiency is traditionally reported in terms of a @xmath0-factor , first introduced by jacobson & stockmayer @xcite , which is a measure of the effective local concentration of duplex ends at zero end - to - end separation .
the @xmath0-factor enables the ring closure probability to be calculated , and importantly , may be related to a ratio of equilibrium constants : @xmath12 where @xmath13 and @xmath14 are the equilibrium constants for cyclization and dimerization , respectively .
multimerization of the cyclization substrate yields a mixture of linear and circular products @xcite . to avoid this complication ,
a separate dimerization substrate may be prepared with the same sequence as the cyclization substrate , but with only one @xmath2-base complementary sticky end per molecule @xcite ( ( b ) ) .
following hybridization , the total length of the system is then @xmath15 base pairs , with blunt as opposed to sticky ends .
the consequences of this choice for @xmath14 and hence @xmath16 are discussed in supplementary . by assuming the contribution from base pairing to @xmath13 and @xmath14 is the same , the wlc model can be used to estimate @xmath16 .
a common assumption is that the cyclized state is fully stacked , with coaxial stacking across the two nicks .
for this situation , the analytic expression derived by shimada & yamakawa
( sy ) @xcite , which includes both the bending energy cost of bringing the two ends together and the twist energy cost of bringing the two helix ends into register , is the most appropriate . some of the confusion surrounding claims of cyclization efficiency greater than that predicted by the wlc model revolves around the use and interpretation of @xmath0-factors .
while the @xmath0-factor relation using the ratio of cyclization to dimerization equilibrium constants is well established , experimental studies usually report the ratio of forward rate constants . in the case of ligase - based assays ( reviewed in reference @xcite ) : @xmath17 where @xmath18 and @xmath19 are the forward rate constants for the formation of the _ ligated _ circle and dimer , respectively . in the experimental limit where the ligation rate is very slow compared to the rate constants for uncyclization ( @xmath6 ) and undimerization ( @xmath8 ) , the concentrations of unligated circles and dimers will reach an equilibrium with reactants .
if this condition is met , @xmath20 should be equivalent to @xmath16 . in practice
, this limit is valid for low ligase concentrations ( @xmath21 } \ll \ce{[dna]}$ ] ) and short sticky ends @xcite .
the importance of this condition is illustrated by du _
et al . _
@xcite , who suggested that the apparent non - wlc behaviour in the c&w experiments was rather due to an insufficiently low ligase concentration ( reviewed in reference @xcite ) .
an additional complication with ligase experiments relates to the structure of the substrate , specifically the dna surrounding the nick .
it is unclear whether the ligase will act equally on all nicked duplexes , or only the subset that happen to adopt the right configuration at the nick , be that coaxially stacked or kinked , to allow the ligase to bind . as we show in this work , to alleviate stress , cyclized systems are more likely than dimers to break coaxial stacking at a nick
. the ligation rate of hybridized complementary sticky ends may therefore vary substantially depending on the system .
the fret experiments of v&h @xcite have the advantage of directly monitoring the transition between cyclized and open states . while it is possible to report thermodynamics from fret experiments , v&h report dynamics : @xmath22 where @xmath23 and @xmath7 are the forward rate constants for the formation of the _ unligated _ circle and dimer , respectively .
dynamic @xmath0-factors extracted from fret - based experiments must also be interpreted with care . making a comparison between @xmath24 and the sy prediction for @xmath25 , v&h
make a claim of much greater than wlc flexibility at @xmath26 .
however , @xmath27 only in the limit where @xmath28 , a condition that v&h , as well as another more recent fret experiment @xcite , have shown not to be met . given that @xmath29 , one should not expect @xmath27 at short @xmath4 .
thus , the observed deviation of @xmath24 from @xmath25 is not necessarily an indication of non - wlc behaviour .
the interpretation of cyclization experiments is not straightforward ; in particular , the microscopic states , responsible for the putative non - wlc flexibility , can not be directly observed .
additionally , non - wlc behaviour has been reported in a number of other systems , including dna minicircles @xcite , a `` molecular vice '' @xcite , and a `` strained duplex '' @xcite . as we emphasised in the accompanying paper @xcite , establishing whether these observations are mutually consistent is a challenging task due to the distinct interplay of mechanics , geometry and topology inherent in each experimental system .
fortunately , simulation can help bridge this gap .
models that reproduce the basic mechanical , structural and thermodynamic properties of dna can be used to simulate the relevant systems .
the results can be used to establish : ( i ) whether experimental results are truly indicative of non - wlc behaviour ; ( ii ) whether the behaviour can be attributed to certain types of structure , such as ` kinks ' @xcite ; ( iii ) whether different systems present consistent evidence for non - wlc behaviour ; and ( iv ) whether the inferred occurrence of any disruptions to duplex structure is consistent with our current understanding of dna thermodynamics . here
we use oxdna to probe the cyclization equilibrium as a function of duplex length .
this approach provides direct access to microscopic states , enabling a structural interpretation of experimental observations .
oxdna is particularly well - suited to this task , as it provides a good description of both single- and double - stranded dna , including hybridization thermodynamics , persistence length , torsional modulus and basic structure , and has previously been shown to reproduce a number of stress - induced transitions in dna @xcite .
oxdna @xcite is a nucleotide - level coarse - grained model of dna that has been employed successfully for a wide variety of systems @xcite , beginning with the thermodynamic and structural characterization of dna nanotweezers @xcite .
briefly , the model consists of rigid nucleotides with three interaction sites per nucleotide , interacting via watson - crick base pairing , base stacking , excluded volume , and a potential to represent backbone connectivity .
the model is parameterized to reproduce the thermodynamics of duplex melting at high - salt ( ) , where backbone - backbone electrostatic repulsion is short - ranged due to counter - ion screening .
two parameterizations of the model are available .
we use the the average - base parameterization @xcite , in which the strength of the base - pairing and stacking interactions are independent of the identity of the bases , to highlight the basic thermodynamics of dna cyclization .
by contrast , the sequence - dependent parameterization @xcite has stacking and base - pairing interactions that depend on the base identity and is used to compare more directly to the v&h experiments . simulations of the cyclization equilibrium were performed with a virtual - move monte carlo ( vmmc ) algorithm @xcite at .
as the free - energy barrier between typical open and cyclized states is large , the transition between the two macrostates constitutes a rare - event .
umbrella sampling , a technique that allows for the biased sampling of states with respect to an order parameter @xcite , was employed to sample the barrier crossing in reasonable computational time .
we use a two - dimensional order parameter @xmath30 to characterize the transition .
@xmath31 is a discretized measure of the distance of closest approach between the complementary sticky ends .
@xmath32 is the number of base pairs formed between complementary sticky ends , where @xmath33 . to further improve computational efficiency , umbrella sampling was windowed to separately sample the open and cyclized states of each molecule . for the window associated with the open state ,
the system was restricted to @xmath34 ; for the window associated with the cyclized state , the system was restricted to @xmath35 ( the value corresponding to the shortest distances between sticky ends ) .
simulations were run until convergence to within @xmath36 for each window .
the sampling windows overlap at @xmath37 ; results were combined by normalizing each window so that the free energies were equal for this value of the order parameter . as there is only one well - defined overlap between the values of the order parameters for both windows ,
more complex approaches , such as the weighted histogram analysis method @xcite , were unnecessary . to further simplify sampling
, we forbade the formation of base pairs that are not intended in the design of the system ( non - native base pairs ) .
further details of the cyclization simulations can be found in supplementary .
the simulation of dimerization equilibrium is roughly analogous to cyclization , and is elaborated in supplementary . to compute the equilibrium constants , we deemed all states with @xmath38 to contribute to the cyclized and dimerized states ( details in ) .
a complete list of sequences is available in .
error bars represent the standard error of the mean from 5 independent simulations . for sufficiently strong bending stress ,
localized structural disruptions to the dna double - helical structure are expected .
we define three such disruptions , namely fraying , bubble formation and kinking , which are elaborated in detail in the accompanying paper @xcite .
briefly , both fraying and bubble formation involve the breaking of base pairs ; the difference is in the location along the duplex .
fraying involves disruption of base pairing at the duplex ends , while bubbles occur in otherwise fully base - paired contiguous stretches away from the duplex ends .
conceptually , a kink is an area of strong bending localized to a small segment of dna , and can occur both at a nick and within the duplex .
when kinking occurs in a duplex region , it is nearly always accompanied by bubble formation .
similarly , kinking at a nick can be accompanied by fraying .
we do not attempt to distinguish different types of kinks , as has been done when analysing atomistic molecular dynamics simulations @xcite .
instead , we simply define a kink as present within a duplex region when there is a change in orientation of consecutive bases of greater than 90@xmath39 on either strand . for nicked regions ,
only consecutive bases on the intact ( unnicked ) strand are considered .
although this cutoff is somewhat arbitrary , since in the current system , kinks arise to localize bending stress , they are usually very sharply bent and the criterion works well @xcite . sometimes , however , it gives rise to false negatives , particularly in the case of kinks at nick sites , and when fraying is present .
nonetheless , it is satisfactory as an indicator of behaviour for our purposes .
more details , along with subtleties related to kinking at a nick , are discussed in supplementary .
we simulate a large range of system sizes ; some illustrative configurations are shown in .
quantitatively , we first consider the behaviour of @xmath16 as a function of length . for the dimerization system we only computed the equilibrium constant @xmath14 for a few lengths ( monomers @xmath40 ) .
as expected , we found @xmath14 to be length - independent to within numerical error ( supplementary , ) ; therefore , we use an average value of @xmath41 in our @xmath0-factor calculation . in contrast
, we found @xmath13 , and thereby @xmath16 ( ) , to vary substantially with length . in
, we show @xmath42 values , calculated from our measured values of @xmath13 and @xmath14 using , for 81 different lengths in the range @xmath43 for fixed @xmath44 using the average - base parameterization of oxdna .
these results are compared to @xmath25 predictions based on the shimada & yamakawa ( sy ) expression @xcite using previously calculated values for the relevant structural and mechanical properties of oxdna @xcite .
the sy expression is appropriate for the formation of a fully stacked `` circle '' configuration with coaxial stacking at both nicks ( ( a ) ) .
note that the comparison in is fit - free .
the effect of varying @xmath2 as well as the role of nicks and mismatches are discussed in supplementary .
the behaviour of the sy expression is well understood . in the regime of interest
, shortening @xmath4 tends to make cyclization less favourable as a result of increased bending stress within the duplex .
this effect becomes particularly acute for dna lengths below the persistence length ( / for the curve plotted in ) . on top of this systematic behaviour
, a periodic oscillation is associated with the need to over- or under - twist the duplex when the natural twist is not commensurate with that required to form a closed circle .
the magnitude of this oscillation increases at shorter @xmath4 because a stronger twist per base pair is required . when considering the behaviour of @xmath42 in light of the sy expression , three length - scale dependent regimes become apparent : long ( @xmath45 ) , intermediate ( @xmath46 ) and short ( @xmath47 ) . in the long length regime ( @xmath45 )
, oxdna reproduces the periodic oscillations predicted by the sy expression , and values of @xmath16 coincide at the maxima of these oscillations .
however , even at the longest lengths we consider , the magnitude of the oscillation is smaller for oxdna than predicted by the sy expression . at shorter
@xmath4 , the magnitude of the oscillation decreases for oxdna , at odds with the sy expression .
the difference stems from the possibility of adopting alternative `` teardrop '' configurations ( ( b ) ) , in which most of the twisting stress , and some of the bending stress , can be relieved by kinking at one of the nicks , thus breaking coaxial stacking . in these configurations
there is still the constraint of dna continuity at the nick and this can now be more easily satisfied , rather than through over- or undertwist , by out - of - plane bending ( ( d ) ) .
the possibility of adopting this alternative teardrop configuration reduces the free - energy penalty for incommensurate values of @xmath4 ( i.e.@xmath48 ) , thus suppressing the oscillations .
further , at shorter @xmath4 , the bending stress increases more slowly for these kinked teardrop configurations than for the coaxially stacked circles .
thus , the difference in @xmath16 between the `` on - register '' ( coaxially stacked ) and `` off - register '' ( kinked at a nick ) molecules decreases at shorter @xmath4 , rather than increasing as predicted by the sy model . although not taken into account for most analyses of cyclization , the possibility of the cyclized molecule exhibiting a `` teardrop '' configuration has previously been suggested by vologodskii _
et al . _
@xcite . as a function of length @xmath4 , in the duplex , at either nick and at both nicks . *
( b ) * probability of broken base pairs @xmath49 as a function of length @xmath4 for fraying ( base - pairing disruption at either nick ) and bubble formation ( base - pairing disruption in the duplex region ) .
, width=321 ] further evidence in support of this analysis is given in ( a ) , which shows the probability of kinking as a function of @xmath4 . at long lengths ( @xmath45 )
, kinking does not occur in the duplex regions , but can occur at the nick sites .
there is clear periodicity in kinking at a nick .
for example , at @xmath50 , the probability of kinking at either of the nicks is negligible and the system virtually always adopts a coaxially stacked circle configuration ; however , at the longer @xmath51 , the probability is @xmath52 .
as @xmath4 is shortened , the probability of kinking at a nick gradually increases for these `` off - register '' lengths ; it is not until @xmath53 that the bending stress along the duplex is sufficient to cause @xmath54 of molecules to kink at a nick for an `` on - register '' length . in the intermediate - length regime ( @xmath46 ) , we observe enhanced cyclization efficiency compared to the sy prediction . although @xmath42 continues to decrease with @xmath4 , it does so more gradually than the sy expression would predict ; consequently , @xmath42 is in excess of the peak envelope of the sy expression . in this regime ,
bending stress in the circular coaxially stacked configuration is sufficiently large that kinking at one of the nicks occurs for even the on - register systems ( ( a ) ) .
oscillations due to on - register effects also seem to contribute to @xmath42 in the upper end of the intermediate - length regime , with shallow maxima occurring at @xmath55 and @xmath56 , approximately 6 and 7 times the pitch length .
however , these sizes no longer correspond to minima in the probability of kinking at a nick ( ( a ) ) so the structural underpinnings of these variations is less clear . for shorter lengths , although there are size - dependent variations in @xmath42 ( e.g. maxima at @xmath57 and 49 ) , there is no longer a simple relationship to the pitch length , instead reflecting more complex geometric compatibilities that allow cyclized states at these lengths to be particularly stable compared to nearby lengths . in the intermediate regime ,
most cyclized molecules are kinked at one of the two nicks . at
shorter @xmath4 , the bending stress in the duplex region of the teardrop configurations increases , with the highest curvature being localized opposite the nick that is kinked .
consequently , it becomes increasingly favourable to localize bending stress into a kink in the duplex ( unnicked ) region , with the probability of this duplex kinking ( ( e ) ) increasing from near zero at @xmath58 to near one at @xmath59 ( ( a ) ) .
a kink in the duplex will generally be located opposite a kink at a nick because this arrangement minimizes the residual bending stress in the unkinked portions of the duplex by equalizing the lengths of the double - helical segments between the two kinks .
in addition to a loss of stacking , kinking in the duplex typically involves breaking base pairs @xcite .
a typical configuration for such a kinked duplex state is shown in ( e ) . as with kinking at the nick
, the availability of this configuration lowers the free - energy cost of cyclization , raising @xmath42 further above the sy prediction . at the lower end of the intermediate - length regime ,
the cost of bending without kinking in the duplex region is so high that molecules with duplex kinking dominate the cyclized state .
states with two kinks localize the majority of the bending stress at the kinking sites , as shown in ( e ) , with long duplex sections relatively relaxed .
consequently , the free - energy cost of looping is largely independent of @xmath4 in this regime , causing @xmath42 to level off and reach an approximately constant value .
in contrast , the sy @xmath25 prediction decreases very rapidly .
for example , at @xmath59 , @xmath42 is @xmath60 times greater than the sy prediction
. there exists a rich landscape of structures informed by slight differences in local geometries as a function of @xmath4 .
in addition to the canonical two - kink structures , containing one kink in the duplex and one kink at a nick ( ( e ) ) , we observe several non - trivial arrangements , albeit with relatively low probability .
for example , the configuration in ( f ) contains a kink at both nicks as well as in the duplex . in the short - length regime ( @xmath61 )
, we observe an increase in @xmath42 , in stark contrast to the rapidly decreasing @xmath25 predicted by the sy expression .
given the many wlc assumptions that are violated at this length scale , a deviation is unsurprising ; however , we did not anticipate an increase in @xmath42 . as @xmath3 is now not that much larger than @xmath2
, kinking at both nicks allows the single - stranded sticky ends to hybridize without duplex kinking .
the system now generally adopts a conformation of two parallel duplexes , with the stress now borne by a mixture of continuous bending ( ( g ) ) and fraying of a few base pairs at the ends of the duplexes ( ( h ) ) .
( a ) shows a very clear crossover to cyclized states with kinks at both nicks and no kinks present in the duplex , occurring abruptly at @xmath62 . at the same time , fraying at the duplex ends also increases ( ( b ) ) . in this regime ,
as @xmath4 is shortened , the difference in length of the duplexes decreases and the stress in the system tends to drop , leading to higher @xmath16 ( ) . overlaid on this overall trend are non - trivial geometric effects associated with whether the lengths of the two duplexes relative to the pitch length is convenient for connecting them ( the thickness of the double helix is now significant compared to the duplex lengths ) , which leads to non - monotonic behaviour of @xmath16 and kinking .
note that the value of @xmath4 at the crossover between the short- and intermediate - length regimes is expected to be very dependent on @xmath2 , occurring at smaller @xmath4 for smaller @xmath2 .
so far we have only reported results using the oxdna average - base parameterization , in which the strength of base pairing and stacking interactions are independent of base identity .
as the free - energy cost of disrupting a duplex to form a kink is sequence - dependent , it is important to consider how sequence might perturb the general trends we have elaborated thus far .
we therefore studied a variety of sequences used in the v&h @xcite experiments including six at @xmath63 using the parameterization of oxdna that includes sequence - dependent thermodynamics ( ) .
as duplex kinking for oxdna typically involves the breaking of base pairs , we observe that this kinking preferentially occurs at a t base pairs .
base pairs that are weaker than average introduce preferred locations for kinking ; thus , when duplex kinking is relevant , @xmath42 is expected to be larger for the sequence - dependent than for the average - base parameterization .
this is indeed the case for @xmath64 .
the sequence - induced variation for the v&h @xmath64 sequences is a factor of @xmath65 ( gc - content ) .
this compares to a factor of @xmath66 between the extrema in gc - content , poly(at ) and poly(gc ) . that we find sequence heterogeneity generally makes duplex kinking easier is consistent with our explicit investigation of the free energy of duplex kinking in the accompanying paper @xcite .
our results imply that the crossover to cyclized configurations with duplex kinking occurs at slightly longer @xmath4 when sequence - dependence is included than for the average - base parameterization
. system ( @xmath44 , @xmath67 ) .
the activation free - energy barriers , @xmath68 for the forward ( cyclization , dimerization ) and reverse ( uncyclization , undimerization ) reactions are labelled .
@xmath69 reflects the free - energy cost of bending to form the first base pair in a cyclization system , whereas @xmath70 reflects the entropic cost of bringing two monomers together within the simulation volume .
the dimerization simulations are for a cubic simulation box of dimension , corresponding to a duplex concentration of @xmath71 .
note that for clarity , we have depicted @xmath72 and @xmath73 as the free energy difference between the fully base - paired closed state @xmath74 and the transition state @xmath75 ; however , in practice , frayed states @xmath76 $ ] do contribute to the closed state . while the distinction does not significantly impact our results ( @xmath77 )
, we do include the contribution of frayed states in both @xmath72 and @xmath73 .
, width=321 ] as well as equilibrium constants , free - energy profiles as a function of the number of bases pairs were computed for each system we considered , as these can give more insight into the pathway for association .
example profiles for cyclization and dimerization at @xmath78 are illustrated in .
one interesting feature of the profiles is that the free - energy gain from hybridizing the complementary sticky ends once an initial base pair has formed is less for cyclization than for dimerization : @xmath79 .
physically , this indicates that substantial additional bending stress develops as subsequent base pairs form for the cyclization system , reducing the free - energy gain upon zippering of the sticky ends relative to dimerization .
for example , the @xmath80 configuration in ( c ) is clearly less bent than the @xmath81 configuration in ( b ) ( supplementary ) .
the activation energies derivable from these profiles will be particularly useful in the next section when we consider the dynamic @xmath0-factor . as noted in the introduction
, the results of ligase experiments performed in the low ligase concentration limit should , in principle , be comparable to equilibrium @xmath0-factors ( although there are subtleties related to the ensemble of states actually detected by the ligation enzymes ) .
indeed , for long dna molecules ( @xmath4 much longer than the persistence length ) and low ligase concentrations , there has been consistent agreement between experiment and the wlc model @xcite . however , cloutier & widom ( c&w ) @xcite reported results for @xmath4 shorter than the persistence length ( @xmath82 ) , showing an apparent deviation from wlc behaviour , with @xmath20 ( ) enhanced over the sy wlc prediction @xmath25 @xcite by a factor of @xmath83 .
the differences between the maxima in @xmath42 and @xmath25 in this size range are much smaller ( ) and so can not account for this discrepancy .
in contrast , du _
et al . _
@xcite found no deviation from wlc behaviour for @xmath84 .
furthermore , they presented evidence suggesting that the c&w experiments used too high a ligase concentration to enable @xmath20 to be compared with @xmath25 .
the results of du _
et al . _
are in good agreement with the sy wlc expression , albeit with somewhat different materials parameters ( torsional stiffness , persistence length and pitch length ) than for oxdna , owing in part to different buffer conditions ( supplementary , ( a ) ) .
fret measurements on dna cyclization , as pioneered by vafabakhsh & ha ( v&h ) @xcite , provide a more direct measure of cyclization because @xmath23 , @xmath6 and @xmath13 are obtainable , although v&h mostly report @xmath23 .
v&h claim enhanced cyclization at @xmath85 , based on a comparison between their @xmath24 ( ) and the sy wlc expression @xmath25 .
however , as noted earlier , this is only a fair comparison if @xmath86 , which , as @xmath8 is expected to be length independent , also implies that @xmath6 should be independent of @xmath4 .
however , since both v&h @xcite and more recent fret measurements @xcite suggest that @xmath6 increases with @xmath4 , this condition is not met .
although we do not directly simulate the dynamics of cyclization , we can estimate the relative rates of processes using free - energy profiles such as those in and unimolecular rate theory for activated processes @xcite . in agreement with experimental investigations ,
previous work on oxdna has shown that duplex formation has an effective `` transition state '' involving a very small number of base pairs @xcite .
we therefore make the assumption that uncyclization and undimerization rates are given by : @xmath87 where @xmath88 is a constant for dna melting , and @xmath72 and @xmath73 are defined in .
the physical content of this assumption is that an increased favourability of base - pair formation is manifested in slower unbinding rates .
this is important because bending stress in cyclized systems reduces @xmath72 ( )
. the rate constants will be equal ( @xmath86 ) in the very long length limit ( @xmath4 much longer than the persistence length ) .
@xmath69 and @xmath72 are plotted as a function of @xmath4 in .
in particular there is a general decrease in @xmath72 at shorter lengths , suggesting @xmath6 increases with shorter @xmath4 , in agreement with experimental results @xcite .
additionally , any torsional stress in the cyclized state is relieved as base pairs are disrupted , leading to the oscillations in @xmath72 at long lengths , with the minima occurring at the more torsionally stressed off - register lengths . * ( b ) * uncyclization @xmath72 ( ) and undimerization ( , average highlighted with dashed line ) , all computed using the oxdna average - base parameterization . , width=321 ] to compare with v&h s results for @xmath24 , we note that @xmath89 therefore , using , our approximation for @xmath90 is @xmath91 at each @xmath4 .
we are now in a position to estimate @xmath90 from oxdna s equilibrium constants and activation free - energy barriers .
we expect @xmath14 and @xmath73 to be length - independent as the excluded volume of the duplex far from the complementary single - stranded sticky ends is likely to have little effect on the dimerization process .
indeed , this appears to be the case to within less than in @xmath92 ( ( b ) , supplementary ) .
our resulting @xmath90 values are plotted in , where they are compared to our @xmath42 values , and v&h s @xmath24 .
we observe that oxdna s @xmath90 values lie substantially above @xmath42 , except in the limit of long @xmath4 for on - register lengths .
only if all the stress - induced destabilization of the fully cyclized state is exhibited in the forward rate would @xmath93 .
although the activation free - energy for cyclization increases as @xmath4 decreases ( ( a ) ) , the full bending and twisting stress present in the cyclized state is not yet present in the transition state where the two complementary sticky ends have formed their first base pair , e.g. ( c ) , and so @xmath94 . instead , stress which is not present in the transition state leads to a decrease in the free - energy barrier for uncyclization ( ( b ) ) and accelerates uncyclization relative to undimerization
alternatively , one can consider that the stress in the cyclized state subjects the duplex formed between the sticky ends to a shear force @xcite which is well known to lead to a more rapid rupture of a duplex @xcite .
the observed behaviour is similar to that seen by v&h ; indeed , @xmath90 and @xmath24 agree remarkably well in the range @xmath95 , with reasonable agreement extending to @xmath96 .
our results suggest that variation of uncyclization rates with @xmath4 may be an important contribution to apparent non - wlc behaviour in @xmath24 . to begin to understand the difference between v&h s dynamic @xmath0-factor and @xmath25 a number of authors have tried to account for the role of the single - stranded tails in cyclization by incorporating a `` capture radius '' into the wlc @xmath0-factor calculation , as an approximation for how close the duplex ends must be in order for the sticky ends to hybridize @xcite .
indeed , this approach leads to a significantly enhanced @xmath0-factor when a value that is taken to be roughly appropriate for the 10-base sticky ends of the v&h experiment , namely , is used @xcite . assuming this approach is an attempt to capture the weaker constraints at the transition state , and hence to estimate the activation free - energy barrier relevant to @xmath97 ( note @xmath97 and @xmath16 are often not clearly differentiated in discussions of v&h s results )
, we can use our oxdna results to test the reasonableness of this approximation by measuring the separation of the duplex ends in the transition state ensemble . at long lengths
our measured capture radius is approximately constant with a value just under , but increases at shorter lengths because some of the stress at the transition state is partitioned into stretching the single - stranded tails , reaching a maximum of about at about @xmath98 ( supplementary ) .
thus , although the value used previously @xcite is not unreasonable , unsurprisingly this approach does not capture the full complexity of the transition state to cyclization , and nor does it account for the strain present in the sticky ends for them to achieve contact .
therefore , although explaining a dynamic @xmath0-factor in terms of a capture radius is physically well - motivated , it does not provide a full understanding .
moreover , as its value is not tightly constrained , and certain contributions to stability are neglected , it is difficult to judge whether wlc behaviour is violated or not by comparing a calculated @xmath0-factor curve to an experimental @xmath97 .
we note that @xmath90 , in contrast to @xmath42 , varies comparatively smoothly with @xmath4 , with only a weak periodicity on the length scale of the pitch length .
the shallow maxima at larger lengths occur at @xmath99 , because the two sticky ends are then on the same side of a torsionally unstressed duplex .
consistent with this smooth variation , @xmath100 also varies relatively smoothly with @xmath4 .
the strong periodicity in @xmath42 comes from the free - energy gain when zippering up the complementary sticky ends ( @xmath101 ) , which is greater when @xmath4 is an integer multiple of the pitch length , allowing the formation of a relatively relaxed coaxially stacked circle . v&h
suggest that their @xmath24 data at @xmath102 displays a strong oscillation with a period of about one pitch length ( ) . in agreement with vologodskii _
et al . _ , @xcite we find no physical mechanism for such a strong oscillation and would suggest that the experimental evidence for this oscillation is not compelling .
at the shortest lengths investigated , @xmath24 still lies above @xmath90 .
some of this difference might be due to our use of the oxdna average - base parameterization .
sequence does play a role in dna flexibility , as noted in the accompanying paper @xcite , but the impact of sequence variation is non - trivial .
in particular , because kinks tend to localize to at base pairs , kinking in the duplex is easier for a sequence with gc - content than reported for our average - base parametrization . using the oxdna parameterization with sequence - dependent thermodynamics we found @xmath97 to increase compared to the values for the average - base parameterization .
consistent with our results for @xmath42 , we find a sequence - induced variation in @xmath90 of a factor of @xmath65 for the six v&h @xmath64 sequences ( gc - content ) @xcite compared to a factor of @xmath66 between the extrema in gc - content , poly(at ) and poly(gc ) .
this compares to a factor of @xmath103 for experimental looping rates , a discrepancy which may be explained by v&h s use of poly(a ) tracts , a sequence - motif well - known to introduce intrinsic curvature in duplex dna @xcite .
as oxdna s sequence - dependent parameterization is based on the nearest - neighbour thermodynamics of santalucia _ et al . _
@xcite , alternative structural motifs such as poly(a ) tracts are outside the scope of the model .
oxdna also does not reproduce sequence - dependent structural ( e.g. the difference in size between purine and pyrimidine ) or mechanical ( e.g. flexibility ) properties .
the origin of the remaining discrepancy between @xmath24 and @xmath90 at the shortest @xmath4 is not yet clear , but may indicate enhanced flexibility for v&h versus oxdna due to wlc ( lower persistence length ) or non - wlc behaviour ( kinking within the duplex at slightly longer @xmath4 ) .
the oxdna persistence length of is within the range of experimental observations at , but smaller values at are not implausible @xcite .
it is also possible that oxdna slightly underestimates the prevalence of kinking within duplex regions @xcite ; an onset of kinking at slightly lower stress ( longer @xmath4 ) would make cyclization at shorter lengths more favourable .
finally , it is worth noting that the presence of fluorophores may cause perturbations in the v&h experiments .
v&h do report @xmath13 for some systems ; oxdna results are in good agreement for those lengths ( supplementary ) .
cyclization is a system dependent manifestation of the general thermodynamics of strong dna bending , elaborated in the accompanying paper @xcite .
the remarkable range of behaviour in cyclized systems is explicable by the interplay between three specific deformation modes of stressed duplexes : continuous bending , kinking and fraying .
oxdna reveals that each of these modes is present at a characteristic length - scale with respect to cyclization : continuous bending at long lengths ( @xmath45 ) , duplex kinking at intermediate lengths ( @xmath46 ) and fraying at short lengths ( @xmath104 ) . in addition , as @xmath4 is shortened , there is an increase in kinking at the two nicks that remain after the hybridization of the sticky ends . at longer lengths , kinking at a nick
is only observed for `` off - register '' molecules that can not form torsionally relaxed coaxially stacked circles .
the ability of said nicks to relax bending , as well as torsional stress , means that they become increasingly prevalent for shorter @xmath4 . at the shortest lengths , kinking at both nicks
is dominant .
we use oxdna to probe the reported observation of non - wlc behaviour in fret - based cyclization experiments @xcite . in agreement with experiment
, we observe that for shorter values of @xmath4 , the apparent @xmath0-factor lies substantially above the predictions of the shimada & yamakawa ( sy ) @xcite wlc model .
we also observe that the periodic oscillations predicted by the sy model are suppressed .
this behaviour arises from two conceptually distinct phenomena .
firstly , highly stressed cyclized systems can adopt configurations that relax stress more effectively than through continuous bending , thereby reducing the overall free - energy cost of cyclization relative to a direct estimate based on a simple wlc - based model . at various values of @xmath4
, oxdna identifies kinking at nicks , kinking within the duplex region and fraying of base pairs as key relaxation modes .
secondly , oxdna suggests that not all of the reduction in @xmath13 relative to @xmath14 is due to stress manifest in the cyclization rate ; uncyclization rates are also substantially increased relative to undimerization rates .
the result is that dynamic @xmath0-factors based on the ratio of cyclization and dimerization rates lie even further above the sy prediction than their equilibrium @xmath0-factor counterpart . of the above effects , only kinking within the duplex can reasonably be described as truly non - wlc behaviour .
wlc and related statistical models do not predict absolute rates directly .
kinking at nicks and fraying can only occur when the dna backbone is discontinuous , and any resultant effects are unrelated to whether wlc models accurately describes the body of the dna duplex . in oxdna
( with the average - base parameterization ) , kinking in the duplex only has a substantial effect for @xmath105 .
note , if sequence dependence is taken into account , then kinking may occur for slightly longer lengths .
our results suggest that much of the apparent `` extreme bendability '' reported by vafabakhsh and ha @xcite can be attributed to factors that are not strictly speaking evidence of non - wlc behaviour
. we can not account for deviations at their very shortest lengths ( @xmath106 ) . in oxdna , kinking within the duplex region is present , but not completely dominant .
it is possible that oxdna slightly overestimates the difficulty of kinking within a duplex
if this is the case , the data for the very smallest values of @xmath4 studied by v&h may be indicative of duplex flexibility over and above that predicted by the wlc , although we note that to give a substantial effect on @xmath16 , kinking must not only be present , but must dominate the ensemble .
kinking dominating the ensemble at @xmath107 is inconsistent with oxdna predictions , and available experimental evidence for a `` molecular vice '' @xcite and dna minicircles @xcite .
various authors have incorporated a `` capture radius '' into the wlc @xmath0-factor calculation to capture phenomenologically some of the effects listed above @xcite .
indeed , this approach leads to an enhanced @xmath0-factor , but the choice of capture radius is somewhat arbitrary and imprecise , making it difficult to assess whether non - wlc behaviour is present . nonetheless , we find that the capture radius that has been used in the interpretation of v&h s experiments is not unreasonable .
oxdna is only a model , and good correspondence with experimental results should not be over - interpreted .
nonetheless , the relaxation mechanisms identified are clearly physically plausible . for example , enhanced uncyclization rates have previously been noted in the literature @xcite ( and also for transcription - factor mediated looping @xcite ) .
it is clear that much of the apparent discrepancy between the data of v&h and the predictions of wlc - based models are due to effects that are not true violations of the wlc model of duplex dna flexibility .
our study also helps to reconcile the results of v&h with previous ligase - based assays which saw no evidence of enhanced flexibility at @xmath108 @xcite , and previous studies of minicircles which detected no evidence of duplex disruption at these length scales @xcite . to explore whether the shortest lengths studied by v&h do show evidence of kinking and enhanced flexibility , we would propose experiments of shorter sequences and systematic collection of both dynamic and equilibrium data .
the latter is extremely important ; statistical wlc models make equilibrium predictions , and so the breakdown of a wlc description can only be confirmed with equilibrium data .
indeed , elucidating the subtleties of dynamic and quasi - dynamic ( c&w ) @xmath0-factors is one of the key issues addressed in this work .
this work was supported by the engineering and physical sciences research council [ ep / i001352/1 ] , the national science foundation graduate research fellowship program , the national institutes of health national heart , lung and blood institute , wolfson college , oxford , and university college , oxford .
the authors acknowledge the computing facilities of the oxford advanced research computing and the e - infrastructure south iridis high performance computing facility . 10 smith , s. b. , cui , y. , and bustamante , c. ( 1996 ) overstretching b - dna : the elastic response of individual double - stranded and single - stranded dna molecules .
_ science , _ * 271 * , 795799 .
strick , t. r. , allemand , j .- f . ,
bensimon , d. , bensimon , a. , and croquette , v. ( 1996 ) the elasticity of a single supercoiled dna molecule .
_ science , _ * 271 * , 18351837 .
shore , d. and baldwin , r. l. ( 1983 ) energetics of dna twisting : i. relation between twist and cyclization probability .
biol . , _ * 170 * , 957981 .
bustamante , c. , marko , j. f. , siggia , e. d. , and smith , s. b. ( 1994 ) entropic elasticity of @xmath1-phage dna .
_ science , _ * 265 * , 15991600 .
du , q. , smith , c. , shiffeldrim , n. , vologodskaia , m. , and vologodskii , a. ( 2005 ) cyclization of short dna fragments and bending fluctuations of the double helix .
usa , _ * 102 * , 53975402
. mazur , a. k. and maaloum , m. ( 2014 ) dna flexibility on short length scales probed by atomic force microscopy .
, _ * 112 * , 068104 .
vologodskii , a. and frank - kamenetskii , m. d. ( 2013 ) strong bending of the dna double helix .
acids res . , _ * 41 * , 67856792 .
harrison , r. m. , romano , f. , ouldridge , t. e. , louis , a. a. , and doye , j. p. k. ( 2015 ) coarse - grained modelling of strong dna bending i : thermodynamics and comparison to an experimental `` molecular vice '' .
acids res . , _ submitted .
ouldridge , t. e. , louis , a. a. , and doye , j. p. k. ( 2011 ) structural , mechanical , and thermodynamic properties of a coarse - grained dna model .
_ j. chem .
_ * 134 * , 085101 .
ouldridge , t. e. ( 2012 ) coarse - grained modelling of dna and dna self - assembly , springer , berlin .
ulc , p. , romano , f. , ouldridge , t. e. , rovigatti , l. , doye , j. p. k. , and louis , a. a. ( 2012 ) sequence - dependent thermodynamics of a coarse - grained dna model . _
_ * 137 * , 135101 .
doye , j. p. k. , ouldridge , t. e. , louis , a. a. , romano , f. , ulc , p. , matek , c. , snodin , b. e. k. , rovigatti , l. , schreck , j. s. , harrison , r. m. , and smith , w. p. j. ( 2013 ) coarse - graining dna for simulations of dna nanotechnology .
phys . , _ * 15 * ,
2039520414 . fields , a. p. , meyer , e. a. , and cohen , a. e. ( 2013 ) euler buckling and nonlinear kinking of double - stranded dna .
acids res . , _ * 41 * , 98819890 .
wang , j. c. and davidson , n. ( 1966 ) thermodynamic and kinetic studies on the interconversion between the linear and circular forms of phage lambda dna .
_ * 15 * , 111123 .
wang , j. c. and davidson , n. ( 1966 ) on the probability of ring closure of lambda dna . _
_ * 19 * , 469482 .
shore , d. , langowski , j. , and baldwin , r. l. ( 1981 ) dna flexibility studied by covalent closure of short fragments into circles .
usa , _ * 78 * , 48334837 .
cloutier , t. e. and widom , j. ( 2004 ) spontaneous sharp bending of double - stranded dna .
cell , _ * 14 * , 355362
. zhang , y. and crothers , d. m. ( 2003 ) high - throughput approach for detection of dna bending and flexibility based on cyclization .
usa , _ * 100 * , 31613166 .
vafabakhsh , r. and ha , t. ( 2012 ) extreme bendability of dna less than 100 base pairs long revealed by single - molecule cyclization .
_ science , _ * 337 * , 10971101
. le , t. t. and kim , h. d. ( 2013 ) measuring shape - dependent looping probability . _ biophys .
j. , _ * 104 * , 20682076 .
le , t. t. and kim , h. d. ( 2014 ) probing the elastic limit of dna bending . _ nucl .
acids res . , _ * 42 * , 1078610794 .
zhang , y. and crothers , d. m. ( 2003 ) statistical mechanics of sequence - dependent circular dna and its application for dna cyclization . _
j. , _ * 84 * , 136153 .
yan , j. and marko , j. f. ( 2004 ) localized single - stranded bubble mechanism for cyclization of short double helix dna . _
, _ * 93 * , 108108 .
travers , a. ( 2005 ) dna dynamics : bubble n flip for dna cyclisation ? .
_ current biology , _ * 15 * , r377r379 .
wiggins , p. a. , phillips , r. , and nelson , p. c. ( 2005 ) exact theory of kinkable elastic polymers .
_ physical review e , _ * 71 * , 021909 .
forties , r. a. , bundschuh , r. , and poirier , m. g. ( 2009 ) the flexibility of locally melted dna . _ nucl .
acids res . ,
_ * 37 * , 45804586 .
wilson , d. p. , tkachenko , a. v. , and meiners , j .- c .
( 2010 ) a generalized theory of dna looping and cyclization .
_ epl , _ * 89 * , 58005
. sivak , d. a. and geissler , p. l. ( 2012 ) consequences of local inter - strand dehybridization for large - amplitude bending fluctuations of double - stranded dna
. _ j. chem .
phys . , _ * 136 * , 045102 .
xu , x. l. , thio , b. j. r. , and cao , j. s. ( 2014 ) correlated local bending of a dna double helix and its effect on dna flexibility in the sub - persistence - length regime .
_ j. phys .
lett . , _ * 5 * , 28682873 .
taranova , m. , hirsh , a. d. , perkins , n. c. , and andricioaei , i. ( 2014 ) role of microscopic flexibility in tightly curved dna .
_ j. phys .
b , _ * 118 * , 1102811036 . chen , b. and dong , c. ( 2014 ) modeling deoxyribose nucleic acid as an elastic rod inlaid with fibrils . _
mech . , _ * 8 * , 071005 .
pollak , y. , goldberg , s. , and amit , r. ( 2014 ) self - avoiding wormlike chain model for double - stranded - dna loop formation .
e , _ * 90 * , 052602 .
shin , j. , lee , o .- c . , and sung , w. ( 2015 ) how a short double - stranded dna bends . _ j. chem .
phys . , _ * 142 * , 155101 .
nam , a. , laaksonen , a. , and vercauteren , d. p. ( 2015 ) a coarse - grained simulation study of the structures , energetics and dynamics of linear and circular dna with its ions .
_ j. chem .
theory comput . , _ * 11 * , 28132826 .
salari , h. , eslami - mossallam , b. , naderi , m. s. , and ejtehadi , m. r. ( 2015 ) extreme bendability of dna double helix due to bending asymmetry .
_ arxiv:1505.03138_. podtelezhnikov , a. a. , mao , c. , seeman , n. c. , and vologodskii , a. ( 2000 ) multimerization - cyclization of dna fragments as a method of conformational analysis .
_ biophys .
j. , _ * 79 * , 26922704 .
vologodskii , a. , du , q. , and frank - kamenetskii , m. d. ( 2013 ) bending of short dna helices . _ artificial dna : pna & xna , _ * 4 * , 13 .
jacobson , h. and stockmayer , w. h. ( 1950 ) intramolecular reaction in polycondensations .
i. the theory of linear systems . _
phys . , _ * 18 * , 16001606 . shimada , j. and yamakawa , h. ( 1984 ) ring - closure probabilities for twisted wormlike chains . application to dna . _
macromolecules , _ * 17 * , 689698 .
peters , j. p. and maher , l. j. ( 2010 ) dna curvature and flexibility in vitro and in vivo . _ q. rev .
biophys . , _ * 43 * , 2363 .
du , q. , kotlyar , a. , and vologodskii , a. ( 2008 ) kinking the double helix by bending deformation .
acids res . , _ * 36 * , 11201128 .
demurtas , d. , amzallag , a. , rawdon , e. j. , maddocks , j. h. , dubochet , j. , and stasiak , a. ( 2009 ) bending modes of dna directly addressed by cryo - electron microscopy of dna minicircles .
acids res . ,
_ * 37 * , 112 .
qu , h. , tseng , c .- y . , wang , y. , levine , a. j. , and zocchi , g. ( 2010 ) the elastic energy of sharply bent nicked dna .
_ epl , _ * 90 * , 18003 .
qu , h. and zocchi , g. ( 2011 ) the complete bending energy function for nicked dna .
_ epl , _ * 94 * , 18003 .
kim , c. , lee , o .- c . , kim , j .- y . , sung , w. , and lee , n. k. ( 2015 ) dynamic release of bending stress in short dsdna by formation of a kink and forks .
int . , _ * 54 * , 16 .
lanka , f. , lavery , r. , and maddocks , j. h. ( 2006 ) kinking occurs during molecular dynamics simulations of small dna minicircles .
_ structure , _ * 14 * , 15271534 .
mitchell , j. s. , laughton , c. a. , and harris , s. a. ( 2011 ) atomistic simulations reveal bubbles , kinks and wrinkles in supercoiled dna .
acids res . , _ * 39 * , 39283938 .
spiriti , j. , kamberaj , h. , de graff , a. m. r. , thorpe , m. f. , and van der vaart , a. ( 2012 ) dna bending through large angles is aided by ionic screening .
_ j. chem .
theory comput . ,
_ * 8 * , 21452156 .
matek , c. , ouldridge , t. e. , levy , a. , doye , j. p. k. , and louis , a. a. ( 2012 ) dna cruciform arms nucleate through a correlated but asynchronous cooperative mechanism .
_ j. phys .
b , _ * 116 * , 1161611625 .
romano , f. , chakraborty , d. , doye , j. p. k. , ouldridge , t. e. , and louis , a. a. ( 2013 ) coarse - grained simulations of dna overstretching .
_ j. chem .
, _ * 138 * , 085101 . matek , c. , ouldridge , t. e. , doye , j. p. k. , and louis , a. a. ( 2015 ) plectoneme tip bubbles : coupled denaturation and writhing in supercoiled dna . _
scientific reports , _ * 5 * , 7655 .
mosayebi , m. , louis , a. a. , doye , j. p. k. , and ouldridge , t. e. ( 2015 ) force - induced rupture of a dna duplex . _ arxiv:1502.03623,_. ouldridge , t. e. , louis , a. a. , and doye , j. p. k. ( 2010 ) dna nanotweezers studied with a coarse - grained model of dna .
lett . , _ * 104 * , 178101 .
whitelam , s. and geissler , p. l. ( 2007 ) avoiding unphysical kinetic traps in monte carlo simulations of strongly attractive particles .
_ j. chem .
phys . , _ * 127 * , 154101 .
torrie , g. and valleau , j. ( 1977 ) nonphysical sampling distributions in monte carlo free - energy estimation : umbrella sampling
. _ j. comput .
* 23 * , 187199 .
kumar , s. , rosenberg , j. m. , bouzida , d. , swendsen , r. h. , and kollman , p. a. ( 1992 ) the weighted histogram analysis method for free - energy calculations on biomolecules .
i. the method . _
j. comput .
, _ * 13 * , 10111021 .
chodera , j. d. , swope , w. c. , pitera , j. w. , seok , c. , and dill , k. a. ( 2007 ) use of the weighted histogram analysis method for the analysis of simulated and parallel tempering simulations .
_ j. chem .
theory comput . ,
_ * 3 * , 2641 .
zeida , a. , machado , m. r. , dans , p. d. , and pantano , s. ( 2012 ) breathing , bubbling , and bending : dna flexibility from multimicrosecond simulations .
e , _ * 86 * , 021903 .
hnggi , p. , talkner , p. , and borkovec , m. ( 1990 ) reaction - rate theory : fifty years after kramers .
_ * 62 * , 251341 .
ouldridge , t. e. , sulc , p. , romano , f. , doye , j. p. k. , and louis , a. a. ( 2013 ) dna hybridization kinetics : zippering , internal displacement and sequence dependence .
acids res . , _ * 41 * , 88868895 .
hatch , k. , danilowicz , c. , coljee , v. w. , and prentiss , m. ( 2008 ) demonstration that the shear force required to separate short double - stranded dna does not increase significantly with sequence length for sequences longer than 25 base pairs .
e , _ * 78 * , 011920 .
rivetti , c. , walker , c. , and bustamante , c. ( 1998 ) polymer chain statistics and conformational analysis of dna molecules with bends or sections of different flexibility .
, _ * 280 * , 4159 .
santalucia , j. ( 1998 ) a unified view of polymer , dumbbell , and oligonucleotide dna nearest - neighbor thermodynamics .
usa , _ * 95 * , 14601465
. santalucia , j. and hicks , d. ( 2004 ) the thermodynamics of dna structural motifs .
biom . , _ * 33 * , 415440 .
savelyev , a. ( 2012 ) do monovalent mobile ions affect dna s flexibility at high salt content ? .
phys . , _ * 14 * , 2250 .
herrero - galn , e. , fuentes - perez , m. e. , carrasco , c. , valpuesta , j. m. , carrscosa , j. l. , moreno - herrero , f. , and arias - gonzalez , j. r. ( 2013 ) mechanical identities of rna and dna double helices unveiled at the single - molecule level .
soc , _ * 135 * , 122131 .
harrison , r. m. , romano , f. , ouldridge , t. e. , louis , a. a. , and doye , j. p. k. ( 2015 ) coarse - grained modelling of dna minicircles .
_ to be submitted,_. chen , y .- j . , johnson , s. , mulligan , p. , spakowitz , a. j. , and phillips , r. ( 2014 ) modulation of dna loop lifetimes by the free energy of loop formation .
usa , _ * 111 * , 1739617401 . * supplementary material for `` coarse - grained modelling of strong dna bending ii : cyclization '' *
cyclization simulations were performed in three phases : exploratory , equilibration and production . in all cases , we use a virtual - move monte carlo ( vmmc ) algorithm @xcite in combination with umbrella sampling @xcite . in the exploratory phase
, we iteratively adjusted the umbrella sampling bias for the windows associated with the open and cyclized states , yielding a flat population of states for both windows .
we use discrete potentials for both dimensions of the order parameter ( see main text ) . for @xmath31 , we use variable width distance increments : , 3.4072 , 5.1108 , 8.518 , 12.777 , 17.036 , 21.295 , 25.554 , 34.072 , 42.59 , 51.108 and @xmath109 . for @xmath32 , the increment is fixed , and is simply the number of base pairs formed ( @xmath33 ) . as
the weighting iterations were semi - automated , simulation times varied in the range vmmc steps per particle .
simulations were performed with one molecule in a cubic box of dimension , which corresponds to a unimolecular concentration of . to further simplify sampling , we forbid the formation of base pairs that are not intended in the design of the system ( non - native base pairs ) . in the equilibration phase , we equilibrated the system for vmmc steps per particle using the aforementioned umbrella weights .
adequate sampling ( number of transitions in @xmath31 and @xmath32 ) and decorrelation ( via a block average decorrelation method ) of potential energy , bubble size , fraying and structural kinking were checked . for the production phase ,
each simulation ( five independent simulations per measurement ) was initialized with a statistically independent starting configuration and a unique random seed .
this is accomplished by randomly drawing starting configurations from the equilibration phase , one for each production trial , ignoring the first vmmc steps per particle of equilibration . the production trial run time is vmmc steps per particle . for reference ,
the characteristic decorrelation time for the potential energy is @xmath110 vmmc steps per particle , while the decorrelation time for kinking is @xmath111 vmmc steps per particle .
the `` seed moves '' used to build clusters in the vmmc algorithm were : * rotation of a nucleotide about its backbone site , with an axis chosen uniformly on the unit sphere , and with an angle drawn from a normal distribution with a mean of zero and a standard deviation of 0.10 radians . *
translation of a nucleotide , where the displacement along each cartesian axis is drawn from a normal distribution with a mean of zero and a standard deviation of .
dimerization simulations follow a very similar procedure as cyclization simulations ( ) , but adapted for a bimolecular system . with the exception of the starting configuration ( bimolecular ) , exploratory ,
equilibration and production procedures are identical .
as discussed in the main text , each molecule has one complementary sticky end and one blunt end .
this precludes the formation of multimers and circular dimers , allowing only the formation of linear dimers .
simulations were performed with these two complementary , non - palindromic , sticky - ended duplexes , in a cubic box of dimension , which corresponds to a dimer concentration of . to estimate the impact of excluded volume effects , dimerization systems were also performed in a box of half the size and therefore @xmath112 times the concentration ( per side , which corresponds to a dimer concentration of ) . as for the cyclization simulations ,
non - native base pairing was disallowed .
analogously to the cyclization simulations , the dimerization simulations are windowed using the order parameters @xmath31 and @xmath32 , respectively the distance of closest approach and the number of base pairs formed between complementary sticky ends . the window associated with the dimerized state
is given by @xmath113 ( @xmath35 , the small separation between base sites of the base paired nucleotides ) , while the window associated with the undimerized state is given by @xmath114 .
the cyclization reaction is unimolecular ( ( a ) ) : @xmath115[k_\textnormal{uncyc } ] } \ce{b},\ ] ] where and are the open and cyclized states respectively , @xmath23 is the forward rate constant ( cyclization ) and @xmath6 is the reverse rate constant ( uncyclization ) .
the equilibrium constant for cyclization , @xmath13 , can be estimated directly from simulations of a unimolecular , isolated system as @xmath116 here @xmath117 and @xmath118 are the probabilities with which uncyclized and cyclized systems are observed in simulation ( @xmath119 ) .
the dimerization reaction is a bimolecular association of distinct molecules : @xmath120[k_\textnormal{undim } ] } \ce{ab}.\ ] ] in this case , the bimolecular equilibrium constant can be inferred from a simulation of a single pair of dimerizing monomers via @xmath121 } } , \end{aligned}\ ] ] in which @xmath122 $ ] is the total concentration of strand type a in simulation , and @xmath123 and @xmath124 are the probabilities with which monomers and dimers are observed in simulation .
this expression follows from eq .
7 of ref .
the resultant @xmath14 obeys the standard relation for bulk systems in equilibrium , @xmath125 } { [ \ce{a } ] [ \ce{b } ] } .\end{aligned}\ ] ] recall from the main text that we use a definition for the equilibrium @xmath0-factor @xmath16 that is consistent with vafabakhsh and ha ( v&h ) @xcite : @xmath126 this is a measure of the effective concentration of one sticky end in the vicinity of the other when the dimerization reaction involves distinct monomers , forming a heterodimer , with only one sticky end per monomer .
researchers often estimate @xmath14 using the dimerization of identical monomers , forming a homodimer , with palindromic sticky ends . in this case ,
given the same underlying interaction strength between sticky ends ( and hence the same @xmath13 ) , @xmath127 would be increased by a factor of two due to combinatorial ( 4 times the number of possible dimers ) and symmetry effects ( a factor of 1/2 for a homodimer rather than a heterodimer ) @xcite . as a result , with this approach the equilibrium @xmath0-factor is estimated as @xmath128 in the case of identical monomers with two non - palindromic sticky ends , there is no pre - factor in the @xmath0-factor estimate , as the combinatorial ( twice the number of possible dimers ) and symmetry effects ( a factor of 1/2 for a homodimer ) cancel . overall , under the approximation of ideal behaviour of separate complexes , these three approaches are equivalent . while it is often visually straightforward to identify kinks , automating their detection is not without difficulty .
we have developed both energetic and structural criteria for identifying kinks , which are based on disruption of stacking interactions and changes in base orientation , respectively . in this study , we rely on the structural criterion ; we have discussed the differences between the two criteria in detail in the supplementary material of the accompanying paper @xcite .
structurally , we define a kink using the relative orientation of consecutive nucleotides . in oxdna ,
the orientation of each nucleotide is unequivocally determined by two orthogonal unit vectors , the base - backbone vector and the base - normal vector .
the base - backbone vector connects the backbone and base interaction sites of each nucleotide .
the model is designed such that , in a relaxed duplex , the base - normal vector at index @xmath129 , @xmath130 , and at the consecutive index @xmath131 , @xmath132 , are approximately parallel ; that is , @xmath133 .
a kink is defined to be present if @xmath134 , a condition implying a more than 90@xmath39 change in orientation of consecutive nucleotides along one strand .
naturally , for duplex regions , we consider kinking along either strand ; although , usually if a kink is present in the duplex , then both strands are kinked .
to limit false positives due to fraying , we do not include the first and last 3 pairs of nucleotides in the duplex region in our analysis .
in the fully cyclized configuration there are two `` nicks '' where the two ends of each strand meet .
kinks in nicked regions are treated slightly differently , as our criterion is only well defined along an intact strand .
a kink at a nick is detected on the opposite strand .
as kinks at a nick may diffuse slightly , we define a region of on either side of the nick on the intact strand .
if the intact strand is kinked in this region , then the molecule is considered kinked at that nick . since
distinguishing fraying from kinking at low @xmath32 is problematic , we compute kinking only for the most probable values , e.g. @xmath135 for the @xmath44 molecules depicted in .
the lower @xmath32 configurations appear in our simulations only because of biased sampling and have a negligible contribution to the equilibrium kinking probability .
the equilibrium constant for dimerization @xmath14 is length - independent to within two times the standard error of the mean for the oxdna average - base parameterization ( ) .
varying concentration suggests that the role of excluded volume effects is small between and . for the sequence - dependent parameterization
, there is some variation in @xmath14 , but this does not represent length - dependence per - se , rather it is most likely due to sequence variation in the bases at the interface between the duplexes and the sticky ends . for the computation of @xmath16 and @xmath97
, we use average values for . for sequence - dependent results at @xmath136 , a length where the dimerization equilibrium constant was computed , we use the appropriate length - specific value instead of the average .
p0.5inp0.5 in ccc p0.1 in ccc & & & & + ( l)3 - 5 ( r)7 - 9 & & & & & & & & + 20;20 & 336 & @xmath137 & @xmath138 & @xmath139 & & @xmath140 & @xmath141 & @xmath142 + 57;57 & 336 & @xmath143 & @xmath144 & @xmath145 & & @xmath146 & @xmath147 & @xmath148 + 63;63 & 336 & @xmath149 & @xmath150 & @xmath151 & & @xmath152 & @xmath153 & @xmath154 + 91;91 & 336 & @xmath155 & @xmath156 & @xmath157 & & @xmath158 & @xmath159 & @xmath160 + avg & 336 & @xmath161 & @xmath162 & @xmath163 & & @xmath164 & @xmath165 & @xmath166 + + 10;10 & 2690 & @xmath167 & @xmath168 & @xmath169 & & @xmath170 & @xmath171 & @xmath172 + 28;29 & 2690 & @xmath173 & @xmath174 & @xmath175 & & @xmath176 & @xmath177 & @xmath178 + 30;33 & 2690 & @xmath179 & @xmath180 & @xmath181 & & @xmath182 & @xmath183 & @xmath184 + 45;46 & 2690 & @xmath185 & @xmath186 & @xmath187 & & @xmath188 & @xmath189 & @xmath190 + avg & 2690 & @xmath191 & @xmath192 & @xmath193 & & @xmath194 & @xmath195 & @xmath196 + there is a general trend towards a larger activation free - energy barrier to cyclization ( @xmath69 ) at shorter @xmath4 .
this is due to the bending stress imposed upon the system by the formation of the initial base pair ( @xmath75 ) . at the very shortest lengths
@xmath197 , this trend is reversed because the complementary single - stranded sticky ends ( of length @xmath44 ) are sufficiently long relative to @xmath3 that an initial base pair can form with less bending stress in the duplex region .
states at the top of the free - energy barrier ( i.e. @xmath75 ) for cyclization and dimerization at @xmath198 are compared in , clearly showing bending in the cyclized molecules , but not the dimerized molecules .
importantly , at @xmath199 , the relatively long single - stranded region ( @xmath200 ) reduces the requirement to bend the duplex ( ( c ) ) , compared to the strong bending necessary at @xmath78 ( ( a ) ) .
states for cyclized and dimerized configurations , highlighting the bending required to form the first base pair .
dimerized configurations in ( b ) and ( d ) are shown for @xmath201 , instead of length @xmath4 monomers . for ( b ) , @xmath202 , @xmath203 and @xmath204 . for ( d ) , @xmath205 and @xmath204 .
, width=321 ] we compare oxdna to available v&h equilibrium data @xcite for @xmath206 , @xmath207 . specifically , v&h report the fractional occupancy of the cyclized state @xmath208 .
akin to the nomenclature in , @xmath209 , where @xmath210 is the fractional occupancy of the open ( uncyclized ) state . .
* ( a ) * comparison between oxdna and v&h @xcite for @xmath211 , the free energy of the cyclized compared to the open state . * ( b ) * free - energy profiles for cyclization . the reference state corresponding to @xmath212 is set to be at @xmath213 .
all results are for @xmath214 , and using the oxdna average - base parameterization , unless stated . , width=321 ]
fractional occupancy can be a somewhat misleading measure by which to compare systems , because the entire transition region @xmath215 is covered by a free - energy difference of @xmath216 , while a similar free - energy difference at the extrema ( @xmath217 ) would be indistinguishable .
we therefore make our equilibrium comparison in terms of @xmath211 , where @xmath218 and @xmath219 oxdna appears to be in good agreement with experiment ; although , this may be partly coincidental as v&h do not report either the temperature ( presumed to be ) or a salt concentration ( v&h give their imaging buffer cation concentration as @xmath220 } = \sirange{500}{1000}{\milli\molar}$ ] or @xmath221 } = \sirange{10}{30}{\milli\molar}$ ] ; recall that oxdna is parameterized to @xmath220 } = \si{500}{\milli\molar}$ ] ) . at @xmath222 and @xmath44 , oxdna and experiment differ by @xmath223 , while at @xmath224 , oxdna under - reports experiment by @xmath225 ( cyclized state is less likely in oxdna than experiment ) .
as expected from the santalucia model @xcite of nearest - neighbour thermodynamics , the linear relationship between oxdna results in ( a ) is reflective of the @xmath225 stabilization of each additional base pair .
a possible explanation for the discrepancy between oxdna and v&h @xcite is that hairpin formation in the @xmath2 region could decrease the fraction of cyclized molecules ; but given the sequence , hairpin formation should be negligible . in particular , we can not explain the discrepancy between oxdna and the v&h data for @xmath224 , although we note that identifying a very low yield in experiment can be challenging , for instance due to the presence of impurities inducing large changes in @xmath211 .
we examine the role of structural defects , namely nicks and mismatches , on cyclization using the average - base oxdna parameterization .
unsurprisingly , we find that nicked and mismatches molecules cyclized more readily than their intact counterparts ( ) .
v&h briefly examine the impact of nicks and mismatches on the rate of cyclization , recording @xmath208 versus time ( figure s3 in ref .
@xcite ) . it does not appear that all molecules have reached an equilibrium plateau in @xmath208 , therefore a concrete equilibrium comparison to oxdna is difficult .
we do , however , observe good agreement between the oxdna and the experimental observations of v&h regarding the stabilizing effect of various motifs ( ) .
however , we do not reproduce the unexpected similarity between the experimental values for intact sequences at @xmath226 and @xmath227 .
this is consistent with our inability to reproduce v&h
s @xmath0-factors for their shortest sequences .
l c c c c c & & & + ( l)2 - 3 ( l)5 - 6 & & & & & + @xmath226 + intact & @xmath228 & & & 1.10 + mismatch & @xmath229 & -0.6 & & 3.14 & -1.1 + + @xmath227 + intact & @xmath230 & & & 0.93 + nick 1 & @xmath231 & -1.8 & & 1.73 & -0.6 + nick 2 & @xmath232 & -1.7 + double nick & @xmath233 & -2.8 & & 8.78 & -2.3 + comparison to the ligase experiments of du _
et al . _
@xcite and cloutier & widom ( c&w ) @xcite highlight some interesting features of the oxdna model ( ) . for example , while both the du _ et al . _
experimental and oxdna model results are similar to the sy wlc expression for @xmath234 , there is a significant absolute deviation in @xmath0-factor due to differences in the relevant mechanical properties for oxdna and those seemingly appropriate for the conditions of the experiment of du _
et al_. specifically , they use neb t4-ligase buffer , which has a salt concentration of @xmath221 } = \si{10}{\milli\molar}$ ] , while oxdna is parametrized to @xmath220 } = \si{500}{\milli\molar}$ ] .
s fit to their data gives a weaker torsional stiffness ( versus ) , longer persistence length ( versus ) and longer pitch length ( versus ) than for oxdna .
the comparison also highlights how relatively small deviations in dna mechanical properties ( in particular a @xmath235 change in the persistence length ) may shift the apparent @xmath16 by an order of magnitude .
the differences in sy wlc expression parameters are reasonable given the different solution conditions ( the persistence length decreases with increasing salt concentration @xcite and temperatures ( e.g. @xmath16 is enhanced by a factor of @xmath236 between and @xcite ) .
we note that the oxdna persistence length is reasonable for the high salt conditions used by v&h @xcite .
-factor via a ligase - based assay @xmath237 ( ) compared to the sy wlc expression @xmath25 with du _
. _ parameters ( ) and the oxdna model @xmath42 ( ) . the top envelop of @xmath25 with oxdna parameterized is dashed ( ) * ( b ) * c&w ligase - based @xmath0-factor compared to @xmath90 ( ) and @xmath42 ( ) .
the maxima envelope of @xmath42 is given with a gray dashed line .
, width=321 ] du _ et al .
_ only investigated lengths near the maxima in @xmath25 . as a consequence ,
their experiments are unable to confirm our prediction of a decrease in the magnitude of the oscillations in @xmath42 , which we attribute to the affect of teardrop configurations .
whether the experiments could potentially see the effects of teardrop configurations also depends on how the ligase activity depends on the nature of the dna confirmation .
indeed , vologodskii _ et al .
_ mention the possibility of poor ligation efficiency for teardrop configurations @xcite .
we also note that because of the potential role of teardrop configurations , the torsional modulus extracted from fits to the sy expression should be interpreted with caution , and it is interesting to note that the values typically obtained from these fits @xcite are significantly lower than those from single - molecule experiments with magnetic tweezers @xcite . since it is likely that c&w conducted their experiment at too high a ligase concentration , it is difficult to characterize their @xmath0-factor in terms of either @xmath97 or @xmath16 . to isolate high ligase concentration as the culprit behind an apparent deviation from @xmath25 , du
et al . _
used conditions identical to c&w , namely buffer ( @xmath221 } = \si{10}{\milli\molar}$ ] ) and temperature ( @xmath238 for du _ et al .
_ , @xmath239 for c&w and @xmath240 for oxdna ) . intriguingly , c&w s results mainly lie inbetween @xmath90 and @xmath42 . comparing results for @xmath97 with a wlc model
requires that the finite dimensions of the sticky ends be taken into account . defining a `` capture radius '' ,
a measure of the distance within which the sticky ends can hybridize , is a common way of accounting for the finite size of dna sticky ends .
as oxdna naturally accounts for the finite - size of the sticky ends , we have the ability to test the capture radius assumption . in oxdna
, we define the capture radius @xmath241 as the distance between the first and last bases of the duplex region ( @xmath3 ) at @xmath213 ( ( a , c ) ) .
nucleotide positions are used to report distance , instead of the mid - point along the helical axis , because fraying may occur at the ends of the duplex .
the choice of nucleotide strand does not impact our results .
our ensemble average capture radius , @xmath242 is shown in .
@xmath243 is roughly constant at @xmath244 , increases slowly between @xmath245 , and then more rapidly for @xmath246 , before decreasing for @xmath247 .
the saturation value ( @xmath248 ) is quite close to the capture radius of that has been used to interpret v&h s experiments with 10-base sticky ends @xcite .
the free - energy cost associated with forming the initial contact is partitioned between that for bending the duplex and that for stretching the single - stranded sticky - ends . at shorter
@xmath4 , the amount of bending required for initial binding increases , as does the stretching of the single - stranded sticky - ends . at short lengths ,
this trend is reversed when the length of the unperturbed duplex approaches the capture radius at @xmath213 .
interestingly , the distribution of @xmath241 is roughly gaussian , with similar standard deviation for all lengths . while we explicitly disallow misbonding , allowing misbonding may slightly increase the capture radius , an effect that should be more prominent for longer sticky - ends .
overall , the value for the capture radius used in the interpretation of v&h s experiments appears not unreasonable .
taylor , w. h. and hagerman , p. j. ( 1990 ) application of the method of phage t4 dna ligase - catalyzed ring - closure to the study of dna structure : ii .
nacl - dependence of dna flexibility and helical repeat .
, _ * 212 * , 363376
. harrison , r. m. , romano , f. , ouldridge , t. e. , louis , a. a. , and doye , j. p. k. ( 2015 ) coarse - grained modelling of strong dna bending i : thermodynamics and comparison to an experimental `` molecular vice '' .
acids res . ,
_ submitted .
du , q. , smith , c. , shiffeldrim , n. , vologodskaia , m. , and vologodskii , a. ( 2005 ) cyclization of short dna fragments and bending fluctuations of the double helix .
usa , _ * 102 * , 53975402
. herrero - galn , e. , fuentes - perez , m. e. , carrasco , c. , valpuesta , j. m. , carrscosa , j. l. , moreno - herrero , f. , and arias - gonzalez , j. r. ( 2013 ) mechanical identities of rna and dna double helices unveiled at the single - molecule level .
soc , _ * 135 * , 122131 .
janssen , x. j. a. , lipfert , j. , jager , t. , daudey , r. , beekman , j. , and dekker , n. h. ( 2012 ) electromagnetic torque tweezers : a versatile approach for measurement of single - molecule twist and torque .
_ nano lett . , _
* 12 * , 36343639 . | dna cyclization is a powerful technique to gain insight into the nature of dna bending .
the worm - like chain model provides a good description of small to moderate bending fluctuations , but some experiments on strongly - bent shorter molecules suggest enhanced flexibility over and above that expected from the worm - like chain . here
, we use a coarse - grained model of dna to investigate the thermodynamics of dna cyclization for molecules with less than 210 base pairs .
as the molecules get shorter we find increasing deviations between our computed equilibrium @xmath0-factor and the worm - like chain predictions of shimada and yamakawa .
these deviations are due to sharp kinking , first at nicks , and only subsequently in the body of the duplex . at the shortest lengths , substantial fraying at the ends of duplex domains
is the dominant method of relaxation .
we also estimate the dynamic @xmath0-factor measured in recent fret experiments .
we find that the dynamic @xmath0-factor is systematically larger than its equilibrium counterpart , with the deviation larger for shorter molecules , because not all the stress present in the fully cyclized state is present in the transition state .
these observations are important for the interpretation of recent experiments , as only kinking within the body of the duplex is genuinely indicative of non - worm - like chain behaviour . |
several ideas have been put forward for dealing with strongly coupled theories .
an early example , that works surprisingly well for qcd , is the notion of hidden local symmetry ( hls ) .
take a theory with flavour symmetry @xmath9 spontaneously broken to a subgroup @xmath10 .
the strategy is to first use low energy theorems describing the associated nambu - goldstone bosons ( ngbs ) to construct an effective , nonlinear sigma model description on the manifold @xmath11 @xcite . as it happens , this model is gauge equivalent to a linear model with flavour symmetry @xmath9 and a broken _ gauge _ symmetry @xmath10 , thus providing an alternative effective description @xcite .
the broken gauge symmetry is the hls , in the sense that it was not present in the original theory but `` emerges '' in the low energy physics .
a somewhat more modern tool for tackling strongly coupled theories is seiberg duality , applicable to certain @xmath12 supersymmetric gauge theories @xcite . in its default form
, the duality links the low energy physics of @xmath0 sqcd with @xmath13 colours and @xmath14 flavours , with that of a second sqcd theory with @xmath15 colours , @xmath14 flavours of quark , some elementary singlets ( identified as bound state mesons ) and a yukawa coupling between them all in the superpotential .
the duality also extends to @xmath6 and @xmath7 gauge groups @xcite as well as a veritable zoo of other , more intricate theories @xcite .
the key feature of seiberg duality is that it is a strong - weak duality which can , in certain circumstances , yield calculable results in strongly coupled theories . in particular , choosing @xmath16 renders the original , electric theory asymptotically free and the dual , magnetic theory ir free .
despite the fact that seiberg duality is well tested and well understood from a practical point of view , the question of _ why _ it exists has remained only partially answered .
@xcite and , more recently , refs .
@xcite reinterpreted the duality by proposing that the magnetic gauge group is in fact an hls of the electric theory . by analogy with qcd the magnetic gauge fields
are then related to @xmath2-mesons of the electric theory .
indeed , hints of this idea are already apparent in the flavour symmetry breaking pattern of sqcd : at a generic point in moduli space su(n+n)_lsu(n+n)_rsu(n)_lsu(n)_r , and the surviving non - abelian factors clearly have the same form as the magnetic gauge group .
previous investigations of this interpretation focused mainly on the phase structure of the theory , or on the noether currents and comparison with qcd . in this article
we will instead return to the full , supersymmetric hls formalism of refs .
this approach yields many new results , all supporting the idea that seiberg duality is just the way that hls manifests itself in supersymmetry .
a difference between seiberg duality and hls that will be central to our discussion is that , whereas the former is a duality that exists between unbroken theories , the hls is defined as a spontaneously broken symmetry .
the property of supersymmetric theories that allows us to reconcile this difference is that their potentials have enhanced complex flavour symmetries . as a consequence
their moduli spaces contain both ngbs _ and _ quasi - ngbs , and there is a tendency for flavour symmetries to be broken by expectations of the latter .
one can therefore change the pattern of flavour symmetry breaking simply by moving around the moduli space .
it is then interesting to investigate what happens in the hls description at points with enhanced flavour symmetry .
we will see that the previously broken hls can be ( partially ) restored .
the newly massless gauge fields correspond to combinations of the ngbs that can no longer be considered ngbs at such points .
this is the general mechanism connecting hls to seiberg duality , and is the focus of section [ sec : hl ] . applying the hls formalism to sqcd ( section [ sec : hq ] ) provides an explicit realisation .
we initiate the procedure at a generic point in the electric theory s moduli space , where the flavour symmetry is maximally broken , and use the hls formalism to recover the usual magnetic dual in a confined phase .
the magnetic superpotential is instrumental in this result .
it is required to avoid the double - counting of certain degrees of freedom , but it can also be considered as a uv completion that drives the breaking of the hls ( i.e. the magnetic gauge symmetry ) via confinement .
we find it an appealing feature of sqcd that one can identify the dynamical mechanism that breaks the hls .
equivalently , one can start from the magnetic theory , drive it into a confined phase via a meson expectation and recover the electric theory as the hls description instead .
regardless of which gauge group is taken to be the hls , one can smoothly scale all symmetry breaking expectations to zero by moving along the quasi - ngb direction associated with spontaneously broken @xmath1-symmetry . in this limit gauge and flavour symmetry
are restored in both theories , thus recovering full seiberg duality . after re - establishing the hls interpretation of seiberg duality
we derive several new results .
the mapping of electric @xmath2-mesons to magnetic gauge fields is immediate and explicit . comparing this result with a second derivation ( using noether currents and colour - flavour locking as in ref .
@xcite ) allows us to fix the value of the parameter @xmath3 , analogous to that appearing in the chiral lagrangian of real world qcd . on a baryonic branch of sqcd we find @xmath4 ( as in ref .
@xcite ) and on a mesonic branch @xmath5 ( consistent with ref .
@xcite ) .
electric quark mass terms are easily accommodated .
they reduce the size of the unbroken flavour symmetry leading to a higgsing of the magnetic gauge group .
we can also fix the duality scale , which would otherwise be a free parameter , at particular points in moduli space .
these results are presented in section [ sec : ch ] . in sections 5 and [ sec :
vs ] we extend the discussion in various novel directions ; section 5 discusses what can be learnt by gauging @xmath1-symmetry , and in section [ sec : vs ] we show that the hls interpretation can be straightforwardly extended to the @xmath6 and @xmath7 versions of seiberg duality , as well as adjoint sqcd .
in addition to its theoretical interest , a better understanding of seiberg duality opens up several areas of application . including operators charged under the magnetic gauge group in the duality s dictionary allows us to discuss dynamical processes .
this can lead to a semi - calculable description of the unitarisation of composite @xmath8 scattering . by gauging the flavour symmetry one can also interpolate continuously between higgsing / technicolour descriptions of ( supersymmetric ) electroweak symmetry breaking .
lessons learnt from applying the hls formalism to sqcd may help understand the reason @xmath4 is selected in real world qcd . finally we ( recklessly ) speculate that the whole procedure could be implemented as a systematic way of finding non - supersymmetric dualities .
these issues are discussed in section [ sec : ap ] .
consider a theory with a flavour symmetry @xmath9 broken to some subgroup @xmath10 .
low energy theorems tell us that the behaviour of the associated ngbs depends not on the specifics of the theory , but only on the symmetry breaking pattern @xmath17 .
any effective lagrangian that realises this provides a valid description of the underlying theory s ir physics @xcite . in non - supersymmetric theories
a general approach is to realise the flavour symmetry nonlinearly , via a sigma model description on the manifold @xmath11 @xcite .
it turns out that this description is gauge equivalent to one with a linearly realised flavour symmetry @xmath9 and a broken gauge symmetry @xmath10 .
the gauge symmetry , which is not present originally , is said to emerge as a hidden local symmetry of the underlying theory . ref .
@xcite contains a comprehensive review of this idea and its application to supersymmetric theories .
we begin by briefly summarising this latter aspect .
susy tells us that each real ngb comes with two massless superpartners : one real scalar and one weyl fermion .
the extra light scalars can be considered a direct consequence of the holomorphy of the superpotential , which elevates real constants parameterising flavour transformations to complex ones . at the superpotential level
the original flavour symmetry is therefore enhanced to its complex extension @xmath18 .
typically this results in more symmetry generators being broken , hence more massless scalars . while the genuine ngbs are coordinates for the real manifold @xmath11 , the full set of massless scalars spans the larger , complex manifold @xmath19 .
here @xmath20 is the complex symmetry group preserved by the moduli space of the theory .
that it contains @xmath21 follows from the fact that generators of @xmath21 are constrained to be hermitian , whereas those of @xmath22 are not . thus a supersymmetric theory with flavour symmetry breaking @xmath17 can be described by a sigma model on the manifold @xmath19 .
the ngbs are , of course , the usual massless scalars corresponding to the @xmath17 part of the symmetry breaking .
any other massless scalars are known as quasi - ngbs and are forbidden from getting mass terms only by susy .
the precise number of quasi - ngbs depends on how much bigger @xmath19 is than @xmath11 and is given by [ eq : hlnm ] n_m=-=+- .
we count independent real dimensions such that @xmath23}=2\dim{[g]}$ ] and so on . ref .
@xcite mainly studied the limiting case of @xmath24 , whereupon @xmath25 and all massless scalars are ngbs .
the other extreme is @xmath26 , whereupon @xmath27}$ ] and there is a one to one correspondence between ngbs and superfields .
more generally , one can separate chiral superfields into p and m - types .
p - type ( or pure ) superfields have ngbs for both scalar components .
m - type ( or mixed ) superfields contain one ngb and one quasi - ngb .
hence @xmath28 also gives the number of m - type superfields .
whatever the value of @xmath28 , one can define dimensionless superfields @xmath29 to accommodate the normalised ngbs .
these are assembled into a chiral superfield matrix is used as shorthand for all superspace coordinates ] [ eq : hlxipi ] ( ) = e^(x),where(x)=^a(x)^a .
the @xmath30 are the broken generators of @xmath18 and we have chosen a basis such that @xmath31 } & = \frac{1}{2}\delta^{ab } & { \mathop{\rm tr}\left[\hat{s}^{\alpha\dag}\hat{s}^\beta\right ] } & = \frac{1}{2}\delta^{\alpha\beta } & { \mathop{\rm tr}\left[\hat{t}^{a\dag}\hat{s}^\alpha\right ] } & = 0\end{aligned}\ ] ] for generators @xmath32 of @xmath22 . note that these generators are not necessarily hermitian for arbitrary complex groups .
the scalar components of @xmath33 are by definition elements of @xmath18 , and provide standard representatives of each coset in the left coset space @xmath19 . acting on @xmath33 with a group element @xmath34 does not usually preserve this parameterisation , but instead mixes in components involving the @xmath32 via g^()=(^)^-1(,g ) for some element @xmath35 that depends on @xmath36 and @xmath37 in some complicated way .
this means that @xmath33 transforms under flavour transformations according to [ eq : hlgnls ] ( ) ( ^)=g()^-1(,g ) so the flavour symmetry is realised nonlinearly .
we could just as well have chosen @xmath33 to provide representatives of the _ right _ coset space @xmath19 instead , whereupon @xmath38 acts from the left and @xmath39 from the right in the above expression . before continuing ,
we briefly consider the expansion of the scalar component [ eq : hlxipis ] ( ) e^(x)e^i(x ) where @xmath40 and @xmath41 are hermitian scalar matrices . roughly speaking ,
the anti - hermitian scalar part of @xmath36 contains the ngbs and provides the phase factor @xmath42 .
this parameterises a nonlinear sigma model on @xmath11 ; which would have been constructed in a non - supersymmetric theory .
it satisfies the constraint @xmath43 with the non - zero right hand side being a direct consequence of the symmetry breaking .
meanwhile the hermitian scalar parts of @xmath36 provide @xmath40 , modifying this non - supersymmetric constraint to @xmath44 . therefore @xmath40 parameterises fluctuations in the order parameters of the symmetry breaking .
that order parameters can appear as low energy degrees of freedom is a key feature of susy and , in principle , allows some of the symmetry breaking to be dialled down .
it occurs because said order parameters often arise from expectations of quasi - ngbs , leading to rich structure in the low energy theory .
this feature will be important when we come to discuss seiberg duality as it enables the duality to be established for unbroken , not just broken , gauge symmetries . moving back to the task
at hand we require any effective lagrangian to be invariant under the nonlinear transformation .
building one is slightly trickier than in non - supersymmetric theories , mainly because @xmath45 for an arbitrary complex group .
one proceeds by defining projection operators @xmath46 satisfying @xmath47 for every @xmath48 .
the first two expressions are generic features of such operators .
the third one states that the @xmath46 projected subspace ( with @xmath46 acting from the right ) is closed under @xmath22 , i.e.[eq : hlxiheta ] _ ( ^)=(^)=g_()^-1_(,g),where^-1_(,g)=^-1(,g ) .
there is one projection operator for each @xmath10-irreducible block in @xmath9 .
we are now able to write down a khler potential [ eq : hlk0 ] k^s_=v_^2 for a real dimension 1 parameter @xmath49 associated with the scale of symmetry breaking in the underlying theory . because of the projection of @xmath50 and @xmath51 it is not possible to split the holomorphic and anti - holomorphic factors in the determinant.s as the resulting term in the khler potential does not contribute to the metric . ]
this khler potential transforms according to @xmath52 } } } \nonumber\\ & = v_\eta^2\ln{\det{{\left[\hat{h}^{\dag-1}_\eta(\pi^\dag , g^\dag)\xi^\dag_\eta(\pi^\dag)g^\dag g\xi_\eta(\pi)\hat{h}^{-1}_\eta(\pi , g)\right ] } } } \nonumber\\ & = k^s_\eta+v_\eta^2\ln{\det{[\hat{h}^{\dag-1}_\eta(\pi^\dag , g^\dag)]}}+v_\eta^2\ln{\det{[\hat{h}^{-1}_\eta(\pi , g)]}}.\end{aligned}\ ] ] since @xmath53 is a holomorphic function the last two terms have no d - term
. therefore they do not contribute to the action which is consequently invariant .
any linear combination of khler potentials of this form for different projection operators thus produces a suitable effective lagrangian .
the resulting description is the expected nonlinear sigma model on the complex manifold @xmath19 .
we now turn to an hls description , which is a linear description of the same system based on a theory with a flavour symmetry @xmath9 and a complex gauge symmetry @xmath22 . in the wess - zumino gauge , @xmath10 gauge transformations mix with susy transformations
so that the full theory has a complexified gauge group @xmath21 in superspace . ] in the hls theory , a chiral superfield @xmath33 is defined to live in a matrix representation of @xmath9 transforming as ( x)g(x)^-1(x ) .
this is essentially the same variable used above , although it is now considered to be an elementary chiral superfield rather than a function of @xmath36 .
the same projection operators are introduced such that _
( x)g_(x)^-1_(x),where^-1_(x)=^-1(x)and we define an @xmath46 projected vector superfield @xmath54 , transforming in the usual way e^-v__(x)e^-v_^_(x ) under @xmath46 projected gauge transformations .
we then write down a khler potential [ eq : hlkveta ] k^v_=v_^2 i.e. a gauge theory with an fi - term @xmath55 and @xmath22 invariant auxiliary chiral superfield @xmath56 . one role of @xmath56 will be to ensure that any quasi - ngbs associated with the breaking of @xmath57 symmetries remain unfixed by the vector superfield equations of motion .
a secondary role is clear from the fact that shifts in @xmath56 correspond to ( complexified ) gauge shifts in @xmath58 , therefore restoring the radial " degrees of freedom that are absent in the non - linear sigma model .
we shall comment more on this later .
currently there are no kinetic terms for the vector superfield .
it is an auxiliary degree of freedom and one can solve the equations of motion to find [ eq : hlvsolns ] d_^(x)d_(x)e^v_=_^(x)_(x ) .
( note that the equations of motion still give @xmath59 even if @xmath60 does not contain a @xmath57 subgroup such that the fi - term vanishes . ) substituting back into eq .
, ignoring constant terms and a term proportional to @xmath61 ( which vanish after the superspace integration ) gives k^v_=v_^2=k^s_. note that this expression is gauge invariant even though it has no @xmath58 because of the discussion around eq . .
hence any khler potential [ eq : hlketa2 ] k_=(1-a_)v_^2+a_v_^2 for an arbitrary , real constant @xmath62 reduces to @xmath63 after solving the vector superfield equations of motion .
this arbitrary constant relates the vector superfield mass to the gauge coupling @xmath64 via @xmath65 .
it is analogous to the @xmath3 parameter appearing in the chiral lagrangian of regular qcd , where dynamics seems to fix its value to @xmath66 .
finally , observe that the transformation properties of @xmath33 mean that it can be written in the form [ eq : hlxisigma ] ( x)=e^(x)e^(x),where(x)=^(x)^. fixing gauge such that @xmath67 ( the unitary gauge ) , @xmath33 becomes a function of @xmath36 only and is identical to its counterpart in eq . .
however , flavour transformations do not respect this choice as ( ( x))g((x))=e^^(x)e^(x ) . to maintain the gauge fixing they
must be accompanied by gauge transformations ( ( x))g((x))^-1((x),g)=(^(x ) ) exactly reproducing the nonlinear transformation properties of eq . .
despite the current lack of kinetic terms for the vector superfields many examples are known in which they develop dynamically , elevating the auxiliary vector superfields into true gauge fields .
alternatively one might think of the hls description without kinetic terms as being strongly coupled , with the kinetic terms being suppressed by the gauge coupling . in sqcd the appropriate interpretation will depend on where one is on the moduli space .
that said , there is no symmetry stopping us adding gauge field kinetic terms to eq . by hand
this is the approach usually taken for @xmath2-mesons in real world qcd , where it is anyway expected that kinetic terms are generated dynamically .
note that , for the sake of pedagogy , we have been treating the whole of @xmath22 democratically .
often , different parts of @xmath22 will have different properties .
for example , some of @xmath22 may be anomalous , as will turn out to be the case in sqcd .
the hls formalism remains applicable but the anomaly tells us that that part of the gauge symmetry can not be restored .
in other cases kinetic terms may fail to arise dynamically for some components of @xmath58 which remain as auxiliary fields .
we shall continue to refer to the `` @xmath22 gauge theory '' , but bearing this in mind .
in summary , the two candidate low energy descriptions , a nonlinear sigma model on the complex manifold @xmath19 and an @xmath22 gauge theory with flavour symmetry @xmath9 , are gauge equivalent to one another .
the @xmath22 gauge symmetry arises as an hls and there is one vector superfield for each projection operator satisfying eq .
( or @xmath10-irreducible block of @xmath9 ) .
many supersymmetric theories break their flavour symmetry via quasi - ngb expectations .
it is then possible to smoothly take a limit in which the symmetry is ( partially ) restored . here
, we investigate in general terms how the hls description behaves in this limit
. it will be an important component of our discussion of sqcd in the following section .
we begin by saying a little more about the auxiliary chiral superfield @xmath56 . in the wess - zumino gauge ,
the solution to the vector superfield equations of motion demands vacuum expectations [ eq : hlxiex ] _ ^(x)_(x)=d_^(x)d_(x)for the scalar components of @xmath33 .
gauge symmetry breaking is thus manifest for @xmath68 : the scalar expectations provide a mass term for @xmath58 .
returning to eq .
, suppose there is a broken generator @xmath69 satisfying @xmath70\eta & = 0.\end{aligned}\ ] ] equivalently , the restriction of @xmath69 to the @xmath46 projected subspace generates a complexified @xmath57 symmetry .
we can pull out the associated ngb superfield to write [ eq : hlpikappa ] _ ^()_()=e^2|_(x)e^_a1^a(x)^ae^_a1^a(x)^aso
the scale of @xmath71 is able to fluctuate , as parameterised by the real scalar [ eq : hlkappadef ] |_(x)=. this is simply the superpartner of the ngb for the broken @xmath57 , and evidently a quasi - ngb of the theory .
as already stated , one job of @xmath56 is to ensure that the expectation of this quasi - ngb is not determined by the equations of motion . to this end , eq .
fixes @xmath72 in the wess - zumino gauge .
the solution to the vector superfield equations of motion becomes [ eq : hlxiex2 ] e^2|_(x)e^v_=_^(x)_(x ) with the scalar components getting expectations @xmath73 in the wess - zumino gauge as required .
note that this issue is restricted to quasi - ngbs associated with complexified @xmath57 symmetries in @xmath22 , since the fi - term in the khler potential is only relevant to them . in the absence of @xmath57 symmetries in @xmath22
the vector superfield equations of motion give @xmath74 but leave its overall scale unfixed .
our general strategy is now to define the hls description at a point in moduli space where its gauge symmetry is broken , then use the quasi - ngb @xmath75 to move to where gauge symmetry is restored .
one can not go directly to an hls description with unbroken gauge symmetry as the sigma model description , from which the hls description is derived , is not well defined at these points in moduli space .
the hls description can , in fact , be considered to resolve the behaviour of the sigma model description as such points .
. implies that the expectation of @xmath50 vanishes and gauge symmetry is restored in the limit @xmath76 , i.e. @xmath77 .
however , the khler potential blows up in this limit unless we simultaneously take @xmath78 .
actually , this is precisely what we should do .
recall from the discussion around eq . that @xmath75 parameterises fluctuations in the order parameter for the @xmath46 projected part of the symmetry breaking .
since @xmath49 corresponds to the scale of this symmetry breaking , any scaling of @xmath79 is matched by a scaling of @xmath49 .
specifically , the combination @xmath80 remains fixed .
if we now define canonically normalised , dimensionful chiral superfields @xmath81 the khler potential becomes k_=(1-a_)v_^2 + up to terms that vanish after the superspace integral . in the @xmath77 limit with constant @xmath80 it dramatically simplifies to k_=.
this khler potential is smooth everywhere and describes an unbroken gauge theory with massless matter , in which the @xmath46 projected part of @xmath82 is realised linearly .
the normalisation of @xmath83 is determined at the point in moduli space where the hls description is first defined , then remains constant as the symmetry restoring limit is taken .
hence many different normalisations lead to the same unbroken description .
we will see a manifestation of this effect in seiberg duality : the duality scale @xmath84 . also note that , as mentioned earlier , the massless degree of freedom that was absent from the sigma model , corresponding to the radial mode @xmath49 , can be associated with the @xmath58 field .
regardless of whether @xmath58 becomes a dynamical gauge field or remains as an auxiliary field , this degree of freedom passes to @xmath83 and @xmath85 .
( in the former case we can always go to the wz gauge to make the bottom component of @xmath58 zero . ) meanwhile , in the sigma model description , the khler metric based on eq .
vanishes when @xmath78 . since part of the flavour symmetry
is restored in this limit , it comes as no surprise to find that a description built around a broken flavour symmetry breaks down , and one has to revert to the underlying theory with restored flavour symmetry .
indeed when @xmath78 the @xmath57 symmetry is restored so there is no longer a ngb , but the scalar partner of @xmath75 stays massless due to continuity of the moduli space .
the same thing happens to the ngbs of any other part of the flavour symmetry restored when @xmath78 . although the total number of massless degrees of freedom is unchanged the parameterisation used in eq .
no longer makes sense .
not all of the light states can be attached to a broken symmetry generator so the description is invalid .
so we see that in both descriptions there are ` new ' massless degrees of freedom in the limit @xmath78 . in the hls description
they are vector superfields , in the sigma model description they are former ngbs . comparing the two we therefore conjecture that hls gauge fields emerge from ngbs that can no longer be considered such at points of enhanced symmetry in the underlying theory . using the equations of motion for the vector superfields
we can actually be a little more specific .
they are easily rearranged to read [ eq : hlvsolns2 ] v_== upon substitution of eq . .
expanding to leading order in the ngb superfields gives v_(+^)+^where we use the fact that @xmath86 is simply the identity element of the @xmath46 projected subspace and the implied sum in @xmath36 is understood .
hence @xmath58 can indeed be related to the @xmath46 projection of the ngb superfields : precisely those whose scalar components are no longer ngbs when @xmath78 .
finally , we can make a stronger claim for the vector superfields concerning the kinetic terms .
returning to eq .
we can rewrite the gauge coupling in the khler potential as k_a_v_^2 by factoring out the @xmath87 part of @xmath50 . as the dynamical degree of freedom @xmath87 appears to mix with the trace of the vector superfield , it is plausible that the vector superfield becomes dynamical via a mixing with this quasi - ngb .
we will see more examples of the equivalence between vector superfields and quasi - ngbs later on .
many of the abstract ideas of the previous section can be crystallised by considering the example of sqcd . indeed
, we will provide further arguments that the magnetic dual can be interpreted as the hls description as suggested in refs .
@xcite . to do
so we will first derive the appropriate sigma model description , then go onto show that it is gauge equivalent to the usual magnetic theory .
we take an electric theory with @xmath13 colours and @xmath14 flavours .
the anomaly free flavour symmetry is g = su(n+n)_lsu(n+n)_ru(1)_bu(1)_r under which electric quarks @xmath88 and @xmath89 transform as per table [ tab : hqel ] . gauge invariant meson and baryon operators @xmath90 parameterise the theory s moduli space . at a generic point
they pick up expectations @xmath91 up to symmetry transformations .
d - flatness demands that the difference @xmath92 is a constant but the parameters are otherwise free . for non - zero @xmath93 s and @xmath94 s the gauge symmetry
is completely broken .
since the electric quark matrices are rank @xmath13 , the flavour symmetry breaking is limited to [ eq : hqsympatt ] h = su(n)_lsu(n)_ru(1)_b^u(1)_r^ where the unbroken @xmath57 symmetries are a mixture of the original ones with @xmath95 transformations .
the order parameters are conveniently organised by defining @xmath96 and there is a constraint on the moduli space b_1 n^1 n - det_n(m)=0 .
q & & & 1 & 1/n & n/(n+n ) + & & 1 & & -1/n & n/(n+n ) + let us now look at one sector of the symmetry breaking in detail , e.g. @xmath97 .
without loss of generality we consider this to be triggered by an expectation of the @xmath98 quark matrix [ eq : hqqvev ] q= v & 0 , where v = diag(v_1, ,v_n ) .
the broken and unbroken generators acting ( somewhat confusingly ) on the right are complex matrices with the forms @xmath99 up to unimportant normalisation factors .
both @xmath100 and @xmath101 are traceless .
an @xmath102 subgroup is evidently generated by @xmath101 but there remain @xmath103 additional unbroken generators . therefore = 2(n^2 - 1)+2nn and eq . tells us that there are @xmath104 quasi - ngbs .
in other words there are @xmath104 m - type and @xmath105 p - type superfields associated with the @xmath106 factor , saturating the degrees of freedom available in @xmath88 .
similar reasoning applies for the @xmath107 factor where @xmath108 for traceless matrices @xmath109 and @xmath110 acting on the left ( for consistent confusion ) .
in addition to these generators there are identity matrices from each of the two @xmath57 factors .
only a linear combination of the original baryon number generator with @xmath111 and @xmath112 is broken .
it can be absorbed into the existing generators by redefining @xmath113 and @xmath114 with the difference in sign for the @xmath115 component arising from @xmath88 and @xmath89 having equal and opposite baryon number .
on the other hand , we include the broken @xmath1-symmetry generator in its entirety .
unbroken symmetry transformations therefore have the forms @xmath116 where @xmath117 , in addition to a separate @xmath118 transformation .
including the full @xmath1-symmetry generator in the @xmath119 s enables us to avoid complications inherent to gauged @xmath1-symmetries ( discussed in section [ sec : gr ] ) .
however , it also leads to different representatives for the @xmath19 coset space relative to eq .
, as we do not remove all contributions from the @xmath22 generators .
one can think of this as leaving more gauge redundancy in the @xmath33 s than is usual in the hls formalism . even though it is nt explicitly part of the hls
, we will still see gauge - like properties in the @xmath1-symmetry sector of the hls description . at this point
we have to decide whether we want the sigma model description s variables to live in left or right cosets of @xmath19 .
the obvious choice is for @xmath120 to act on the right and @xmath121 on the left ( and vice - versa for @xmath122 and @xmath123 ) , mirroring the original flavour transformations of the quarks .
it is then possible to find a unique projection operator satisfying eq . for @xmath120 , and @xmath124 for @xmath121 : [ eq : hqetadef ]
= 0 & 0 + 0 & .
the action of @xmath120 and @xmath121 on the @xmath46 projected subspace @xmath125 is simply an @xmath126 transformation . using eqs . , and we can then define chiral superfields @xmath127 where @xmath128 , and the independent superfield @xmath129 comes from the broken @xmath130 generator . applying the projection operator we find a low energy sigma model description in terms of chiral superfields @xmath131 and @xmath132 transforming as @xmath133 and with equal and opposite charge under @xmath118 .
the nonlinear dependence of @xmath134 on @xmath36 and @xmath122 is understood .
both @xmath135 and @xmath136 contain @xmath105 chiral superfield degrees of freedom generated by off diagonal components of the broken generators .
these are the p - type superfields of the flavour symmetry breaking .
meanwhile @xmath129 is an m - type superfield .
the khler potential for this description follows straight from eq . and
is [ eq : hqksigma ] k^s=. alternatively we can choose @xmath120 to act on the left and @xmath121 on the right ( and versa - vice for @xmath122 and @xmath123 ) .
is mostly unchanged but the projection operator should now satisfy @xmath137 and @xmath138 ; the unique solution being ^= & 0 + 0 & 0 .
this is actually a special choice because it satisfies @xmath139 .
hence the @xmath140 projected subspace is invariant under @xmath22 , not just closed .
the unbroken baryon number symmetry is also projected out .
this already suggests that flipping the cosets should result in a description of quasi - ngbs acting as order parameters , which are invariant under @xmath22 by definition .
we are thus able to define a second set of chiral superfields @xmath141 and @xmath142 with transformation properties @xmath143 which are conveniently assembled into a single variable [ eq : hqmdef ] _ ^_^=e^2_r _ n_n & _ n_u
+ _ l_n & _
l_u g_r[_^_^]g_l^. due to its invariance under @xmath22 , the khler potential for @xmath144 is constrained only by the flavour symmetry .
any real function invariant under @xmath145 will do .
note that this superfield contains the remaining @xmath104 m - type superfields , thus encapsulating the other low energy quasi - ngb degrees of freedom ( although there are @xmath104 from each half of the flavour symmetry breaking , half of them are eaten by the electric gauge fields and one accounted for by @xmath129 ) .
it also contains a copy of the superfield degrees of freedom that are already present in @xmath50 and @xmath146 , although now of course transforming in a different way .
thus we must ensure that these degrees of freedom are included only once if we wish to use both standard and flipped coset descriptions at the same time .
we now seek an hls description .
following section [ sec : hl ] we consider a theory based on an @xmath147 gauge group .
it is immediately apparent that ( individually at least ) both @xmath148 factors have cubic anomalies for the anticipated matter content . in practise
there is nothing stopping us implementing an anomalous gauge symmetry in an effective theory , as long as its gauge fields are massive ( which of course they are by construction in the hls description ) .
clearly a limit in which _ all _ gauge fields become massless can not exist .
nonetheless , the approach still gives a valid description of the low energy physics and , moreover , there is nothing forbidding limits in which the gauge fields of an anomaly free subgroup become massless . & su(n+n)_r + _ & & 1 & & & + 1 & & 1 + _ & 1 & & & & -1 & 1 & + guided by eq .
we therefore write down a khler potential @xmath149}+ \nonumber\\ & { \mathop{\rm tr}\left[av^2{\left({\left(\frac{\xi_\eta^\dag\xi_\eta}{d_\eta^\dag d_\eta}\right)}e^{v_{b^\prime}-v_l}-v_{b^\prime}\right)}+\tilde{a}\tilde{v}^2{\left({\left(\frac{\tilde{\xi}_\eta\tilde{\xi}_\eta^\dag}{d_\eta^\dag d_\eta}\right)}e^{v_r - v_{b^\prime}}+v_{b^\prime}\right)}\right]}\end{aligned}\ ] ] for the standard coset description .
all superfields are functions of superspace coordinates and transform as in table [ tab : hqxi ] .
the vector superfields @xmath150 and @xmath151 are constructed from the @xmath46 projected generators @xmath152 and @xmath153 respectively .
the fi - terms pick out @xmath154 , the @xmath118 vector superfield being the only one with non - zero trace . solving the equations of motion one can easily show that this theory is gauge equivalent to the sigma model description of eq . . instead of sticking with the original @xmath155 symmetry ,
consider taking the linear combination @xmath156 defined in table [ tab : hqxi ] .
the associated vector superfields are @xmath157 the gauge symmetry associated with @xmath158 is the anomaly free , diagonal combination with @xmath159 . in terms of these new vector superfields the khler potential is @xmath160}+ \nonumber\\ & { \mathop{\rm tr}\left[av^2{\left({\left(\frac{\xi_\eta^\dag\xi_\eta}{d_\eta^\dag d_\eta}\right)}e^{v_{b^\prime}-v - v^\prime}-v_{b^\prime}\right)}+\tilde{a}\tilde{v}^2{\left({\left(\frac{\tilde{\xi}_\eta\tilde{\xi}_\eta^\dag}{d_\eta^\dag d_\eta}\right)}e^{v - v_{b^\prime}-v^\prime}+v_{b^\prime}\right)}\right]}.\end{aligned}\ ] ] all vector superfield equations of motion are solved for @xmath161 after substituting @xmath162 for @xmath163 and defining @xmath164 as per section [ sec : rgs ] . at this point we chose to fix the gauge for @xmath165 and @xmath118 by absorbing the vector superfields @xmath166 and @xmath154 into @xmath33 and @xmath167
conversely , we see that the chiral superfield @xmath129 , related to the spontaneously broken @xmath1-symmetry , could instead be considered as parameterising a gauge transformation . specifically , it would realise an hls corresponding to the unbroken @xmath168 symmetry , that has thus far been omitted from the hls gauge group .
one can quite generally choose the wess - zumino gauge for subgroups of @xmath22 , then trade the scalar components of vector superfields for the corresponding quasi - ngbs ( or vice - versa ) in the above manner .
we therefore rewrite the vector superfields in the original khler potential as a sum of chiral superfields , then absorb them into dimensionful degrees of freedom @xmath169 where @xmath170 transforms in the fundamental of @xmath165 such that @xmath171 has the correct transformation properties .
note that since @xmath165 is anomalous @xmath166 would only ever be expected to play the role of an auxiliary field in the low energy theory anyway .
only @xmath158 can survive to become a true gauge field .
. allows us to extract @xmath170 and @xmath172 from @xmath83 and @xmath85 via the baryon expectations @xmath173 hence the erstwhile gauge fields @xmath166 and @xmath154 can be replaced by @xmath174 and @xmath175 . that these expressions are given in terms of particular baryonic degrees of freedom
is a side effect of our particular choice of electric quark expectation . in full generality
, one has @xmath176 and @xmath83 and @xmath85 are @xmath165 singlets transforming as @xmath177 under @xmath178 .
upon substituting in all new degrees of freedom we find the final form for the khler potential [ eq : hqkfin ] k=+v^2+^2 .
explicit dependence on @xmath3 and @xmath179 is removed after eliminating @xmath154 using the vector superfield equations of motion .
the first terms here are simply the canonical khler potential of an @xmath148 gauge theory , under which @xmath83 and @xmath85 transform in the antifundamental and fundamental representations respectively .
this is precisely what we expect from the seiberg dual , where the magnetic gauge field @xmath180 is identified with that of the diagonal @xmath148 gauge symmetry @xmath158 .
all symmetry breaking is then driven by the remaining terms . from eqs . and
with all ngb expectations rotated to zero , the expectations of @xmath83 and @xmath85 are found to be @xmath181 an important observation is that the hls description exhibits colour - flavour locking .
the @xmath83 and @xmath85 expectations break @xmath178 to @xmath182 , where @xmath183 is the diagonal combination of @xmath184 with the gauged @xmath148 . the orthogonal @xmath165 gauge symmetry , whose erstwhile gauge fields were absorbed into @xmath83 and @xmath85 , does not mix with the flavour symmetry in this way . we have just shown how to derive the quark sector of the seiberg dual theory , but the meson superfield and associated superpotential remain absent .
to find them , we consider the flipped coset sigma model description and use eq . to define [ eq : hqmdef2 ] m = ve^2_r _ n_n & _
n_u + _ l_n & _ l_u which is a gauge singlet in the @xmath185 representation of @xmath95 , and is considered an elementary chiral superfield . for simplicity we have chosen electric quark expectations with @xmath186 and @xmath187 .
prefactor is removed and one makes the replacements @xmath188 and @xmath189 , with similar generalisations in what follows . ]
q & & & 1 & 1/n & n/(n+n ) + & & 1 & & -1/n & n/(n+n ) + m & 1 & & & 0 & 2n/(n+n ) + although it has been provocatively labelled with an `` @xmath190 '' we should confirm that this object , derived from the flipped coset description , really can be interpreted as the superfield corresponding to @xmath191 . to do so , we expand the quarks in the broken electric theory around their expectations as @xmath192 using components @xmath193 and @xmath194 .
normalised ngb superfields are then given explicitly by @xmath195 using this basis to parameterise the goldstone manifold we find from eq .
that @xmath196 plugging into eq .
and expanding to leading order gives m= v+v+q & p + v & p = q as required
. note also that the meson expectation breaks the @xmath197 factor of the flavour symmetry that is not broken by the expectations of @xmath83 and @xmath85 .
having identified the meson superfield we define a real duality scale @xmath84 for normalisation , whereupon the superpotential [ eq : hqwdef ] w= is the unique choice compatible with eq . .
it is then straightforward to reconstruct the anomaly free @xmath57 symmetries as in table [ tab : hqmg ] .
of course the superpotential is not merely allowed : it is required . simultaneously using normal and flipped coset descriptions means some of the degrees of freedom in @xmath88 and @xmath89 have been counted twice . however ,
in conjunction with the expectation of @xmath190 , the superpotential gives a mass @xmath198 to @xmath13 flavours of @xmath83 and @xmath85 .
hence they are integrated out and the double counted degrees of freedom are removed . solving the equations of motion for the massive flavours and substituting back in
, @xmath8 disappears and the sigma model description is again recovered .
therefore the superpotential does not prevent the hls and sigma model descriptions coinciding at low energy .
a second way of thinking about the superpotential is to recall that everything takes place on top of a non - zero meson background .
ergo we can consider @xmath190 as as a source for @xmath199 .
it is then interesting to comment on the parallels between the superpotential and gauge sectors .
one can think of the superpotential as a sort - of chiral ` gauge ' coupling with the meson as its ` gauge ' field . just as the khler potential part of the theory deals with redundancies inherent to the ngb sector , so the superpotential deals with those in the quasi - ngb sector .
indeed , the corresponding ` gauge ' transformations map @xmath200 , thereby moving one around the moduli space . for appropriate supermultiplets ( perhaps those of @xmath201 susy )
, it seems likely that the two sectors could be unified into a single current interaction .
we will not explore this direction here .
either way , integrating out the massive degrees of freedom leaves an sqcd - like theory with @xmath15 colours and @xmath15 flavours .
this is well known to confine @xcite due to its quantum deformed moduli space [ eq : hqqdefms ] b^n+1
n+n_n+1 n+n = m_q^n_mg^2n - n=(v)^n^-n_mg^2n - n . here , @xmath174 and @xmath175 are the usual baryon degrees of freedom , we have set @xmath202 and @xmath203 is the dynamical scale of the hls description s gauge group .
we thus see that the superpotential drives confinement in the hls description , and is therefore responsible for breaking the gauge symmetry .
so in addition to ensuring that the hls description ultimately has the correct degrees of freedom , the superpotential also allows for a uv completion in which the gauge symmetry breaking is not simply encoded in the field definitions . from the hls point of view
this confinement is inevitable because , once the magnetic quark degrees of freedom are made massive ( as they must be in order to avoid double counting ) , it is the only way to reproduce the @xmath204 symmetry breaking of the original nonlinear sigma model .
sqcd provides a perfect illustration of the discussion of symmetry restoration in section [ sec : rgs ] . to restore the gauge symmetry
one takes the limit @xmath205 holding @xmath206 and @xmath207 fixed .
it is the @xmath130 symmetry and its associated quasi - ngb @xmath208 that enables magnetic quarks to be scaled in this way .
this is perhaps unsurprising , given that we know @xmath1-charges are generally related to scaling dimensions in superconformal field theories .
the decay constants , hls gauge field masses and meson expectation all vanish in this limit , with the magnetic quark masses following suit .
the gauge symmetry is therefore restored ( as is the full flavour symmetry ) and all additional terms vanish from eq . , leaving @xmath148 sqcd+m with massless quarks and a canonical quark khler potential . to recast the meaning of the limit in the language of seiberg duality , recall that all order parameters arise from quasi - ngb expectations in the electric theory . by choosing non - zero values for @xmath93 and @xmath94
we therefore define the duality at a particular point in moduli space , where the flavour symmetry is maximally broken .
travelling away from this point in the magnetic theory , along the quasi - ngb direction that breaks @xmath130 , all fields ( and consequently order parameters ) are scaled to zero .
duality implies that one should simultaneously move to the same point in the electric theory s moduli space , hence all order parameters vanish there as well . in the full theory
this would , of course , restore the corresponding electric gauge symmetry .
note that the process is insensitive to the initial values of @xmath93 and @xmath94 so they remain independent parameters . before taking the @xmath205
limit the expectations of @xmath88 and @xmath89 saturated their ranks , forbidding expectations for the components of the electric quarks @xmath194 and @xmath209 ( which instead contain ngbs of the symmetry breaking ) . as the quark expectations start to vanish the constraint is relaxed and new flat directions open up ; precisely those parameterised by the @xmath2-mesons constructed from @xmath194 and @xmath209 . at the same time
the hls gauge fields become massless .
already we are starting to see hints that the magnetic gauge fields can be identified with electric @xmath2-mesons .
note also that the magnetic quarks become massless as the meson expectation vanishes .
initially , one may think this leads to double counted degrees of freedom in the hls description .
however , the electric gauge fields become massless in this limit as well ( besides which , the sigma model description breaks down so its degrees of freedom are ill defined ) . indeed ,
if we are to identify magnetic gauge fields with electric @xmath2-mesons , duality suggests we should also identify electric gauge fields with magnetic @xmath2-mesons .
thus the newly massless magnetic quarks represent electric gauge fields rather than double counted degrees of freedom .
another interesting limit to consider is the baryonic branch of the theory . here
, one has a vanishing expectation for @xmath89 but q= v & 0 i.e. only the baryon direction of moduli space has non - zero expectation .
the unbroken flavour symmetry is then h = su(n)_lsu(n)_lsu(n+n)_ru(1)_b^u(1)_r^. hence all breaking is confined to the @xmath106 factor .
following the standard procedure one does not expect magnetic antiquarks @xmath85 to appear in the hls description at all .
the same is true for the meson superfield , for which only the components arising from @xmath210 are present .
specifically , one would write down [ eq : hqkq ] k= in place of eq . .
the rest of the reasoning of section [ sec : rm ] remains unchanged and yields a theory with [ eq : hqkq2 ] k=+v^2 and no superpotential .
however , this theory clearly fails to capture all of the low energy physics as there are no degrees of freedom corresponding to the massless chiral superfields @xmath89 .
this is because no scalar components of @xmath89 are ngbs on the baryonic branch so are not represented in a description based on ngb superfields .
suppose we include the magnetic antiquarks , meson superfield and @xmath151 anyway , and continue to use the khler potential with superpotential .
expanding around the vacuum using eq .
for the expectation of @xmath83 , the superpotential becomes w = b^1/n(+y)+ ,where = & , and m= x & + z & y .
hence all components of @xmath85 are massive . upon integrating them out the khler potential reduces back to eq . and the superpotential vanishes .
one can therefore include the magnetic antiquarks even though they are not explicitly generated by ngb superfields .
once again the superpotential has proven itself instrumental in removing superfluous degrees of freedom from the hls description .
furthermore , the meson is now expanded to leading order as m = q = v & 0 + & 0 .
its massless components @xmath211 and @xmath212 are directly related to @xmath89 ; degrees of freedom that would otherwise be missing .
rather than arising from the flipped coset hls description , the meson is now an extra matter superfield added by hand ( or indeed just left alone for the whole procedure ) .
meanwhile the superpotential ensures that the components @xmath213 and @xmath214 , that correspond to degrees of freedom already accounted for in the quark sector , are massive and so removed from the low energy theory .
we therefore conclude that the hls description derived in section [ sec : rm ] also describes the baryonic branch of the theory , even though the symmetry breaking pattern is different there .
while the branch with @xmath215 is the only one discussed in this section , it is clear that the same reasoning applies to @xmath216 . in this section
we have argued that the magnetic gauge group of seiberg duality arises as an hls , corresponding to the diagonal subgroup of the @xmath155 flavour symmetry preserved at a generic point in moduli space .
the magnetic quarks and mesons are constructed from ngbs associated with the flavour symmetry breaking , with the superpotential @xmath217}$ ] ensuring that no degrees of freedom are counted twice .
equivalently , the combination of meson and superpotential can be thought of a source for @xmath199 due to the fact that there is a non - zero meson background , or as a facilitator for a consistent uv completion of the hls description .
the hls formalism only defines the duality at points in moduli space where both electric and magnetic gauge groups are completely broken .
however , the presence of a spontaneously broken @xmath1-symmetry allows a limit to be taken in which both are restored .
the limit is smoothly attained by travelling along the quasi - ngb direction responsible for breaking @xmath1-symmetry until @xmath1-symmetry , and all other symmetries , are restored . although there are no explicit kinetic terms for the hls gauge fields , we assume they are generated dynamically ( as argued in ref .
@xcite ) when the magnetic theory is in a higgsed phase , or when its gauge group is unbroken . when it is in a confined phase we instead consider the kinetic terms to be suppressed by the divergent gauge coupling . in either case
, they are not forbidden by any symmetry so we are anyway free to add them by hand . , as discussed at the end of section [ sec : rgs ] . to check for this
, one could consider the full unbroken @xmath218 generator when constructing @xmath33 and @xmath167 then factor out a chiral superfield @xmath219 , as was done with @xmath129 .
an @xmath220 term would appear alongside the vector superfield terms in the khler potential , facilitating the mixing of the two . ]
now that the hls description of electric sqcd has been formally identified with the magnetic dual , we would like to use the interpretation to try and learn some more about seiberg duality .
our main results will be to confirm the identification of @xmath2-mesons with magnetic gauge fields ( proposed in ref .
@xcite ) , to improve our understanding of the effect of electric quark masses and to determine the value of the duality scale @xmath84 in certain regimes .
the quickest way to understand the origin of the magnetic gauge fields is to consult the solution to their equations of motion .
these imply v_mg=. at leading order in @xmath221 and @xmath222 , eq
. then gives v_mg - or , in terms of components [ eq : chvmg ] v_mg^. this unambiguously identifies the gauge fields of the magnetic theory with the @xmath2-mesons of the electric theory .
an orthogonal approach , using the colour - flavour locking observed in sqcd , is to examine the noether currents attached to the unbroken @xmath223 and @xmath224 parts of the flavour symmetry . in electric variables they are @xmath225 } & \cj^\alpha_r & = { \mathop{\rm tr}\left[\tilde{p}^\dag s^\alpha\tilde{p}e^{-v_{\rm el}}\right]}\end{aligned}\ ] ] for an electric gauge superfield @xmath226
. in the magnetic theory one has @xmath227}+\mbox{mesons } & \cj^\alpha_r & = -{\mathop{\rm tr}\left[\tilde{p}s^\alpha\tilde{p}^\dag e^{v_{\rm mg}}\right]}+\mbox{mesons}\end{aligned}\ ] ] where @xmath83 and @xmath85 have been parameterised using @xmath228 matrices @xmath229 and @xmath230 : @xmath231 for small fluctuations in @xmath226 and @xmath180 these currents can be expanded around the vacuum of the electric theory as @xmath232}+\mbox{3 particles } & \cj^\alpha_r & \approx{\mathop{\rm tr}\left[s^\alpha\tilde{p}\tilde{p}^\dag\right]}+\mbox{3 particles}\end{aligned}\ ] ] or , expanding around magnetic expectations @xmath233 and @xmath234 , @xmath235 equating the two we can therefore write [ eq : chvcfl ] v_mg^-_l^-_r^ for any constant @xmath236 .
this result is identical to that of eq . for expectations @xmath237 and @xmath238 .
invoking the baryon map familiar from seiberg duality we also have @xmath239 hence @xmath236 is determined solely by the electric quark expectations : [ eq : chcdef ] ( ) ^n=()^2n-2n .
all conclusions reached in this section are perturbative in nature and only apply when the fluctuations in @xmath194 and @xmath209 can be considered small .
these are generically of order @xmath240 , the dynamical scale of the electric theory , so one requires @xmath241 .
this limit will be emphasised when we consider the behaviour of electric and magnetic theories in section [ sec : mu ] . on a mesonic branch of the theory @xmath242 and we have @xmath243 , which fixes @xmath244 .
hence the magnetic quarks are normalised so that their expectations , and therefore the symmetry breaking scale , correspond exactly to their electric counterparts ( this idea will be useful when discussing the duality scale ) .
referring back to eq . and setting the expectations of all fluctuations around the vacuum to zero then implies that @xmath245 , as proposed for the mesonic branch in ref .
@xcite .
baryonic branches have one of @xmath93 of @xmath94 equal to zero and so @xmath246 . setting @xmath215 , for example
, implies that @xmath247 , whereupon the magnetic gauge field is given by @xmath248}$ ] .
the absence of @xmath209 is is to be expected since the electric antiquarks no longer take part in the flavour symmetry breaking .
furthermore , the magnetic quarks pick up a factor of @xmath249 in their normalisation implying that @xmath4 is now the correct choice , as proposed for the baryonic branch in ref .
@xcite . adding electric quark masses
is well known to provide another way to higgs the magnetic gauge group .
the superpotential deformation [ eq : chwel ] w_el=- for a rank @xmath250 matrix @xmath251 gives masses to @xmath252 flavours .
it also explicitly breaks the flavour symmetry to su(n+n - k)_lsu(n+n - k)_ru(1)_b . in the magnetic theory one
adds the corresponding linear meson deformation w_mg=- for the @xmath253 component of the meson @xmath211 .
the f - terms for @xmath211 fix @xmath254 , higgsing the magnetic gauge group to @xmath255 at the origin of moduli space and giving mass to @xmath252 flavours of magnetic quark .
duality is thus preserved at low energy , where massive flavours are integrated out of both descriptions . from an hls point of view
we can understand this effect by varying the ratio @xmath256 , where @xmath257 denotes a typical electric quark mass . if @xmath258 the electric theory retains an approximate @xmath95 flavour symmetry at the electric higgsing scale .
the hls interpretation thus results in a magnetic theory based on a broken @xmath148 gauge group . in line with our earlier discussion
, the gauge group in this regime is broken by the confinement occurring once when the heavy magnetic quarks are integrated out .
the difference is a small correction to the gauge field masses from @xmath259 .
this only mildly breaks their degeneracy and is in accord with the approximate nature of the original @xmath95 flavour symmetry . if @xmath260 and @xmath261 the electric theory has @xmath252 fewer flavours at the electric higgsing scale .
this limits the possible symmetry breaking to h = su(n - k)_lsu(n - k)_ru(1)_b^. accordingly , the hls description s gauge group is diminished to @xmath255 . below the magnetic higgsing scale @xmath262 this matches the magnetic gauge group found through seiberg duality . further still into the ir
the magnetic theory again confines and breaks the residual gauge symmetry , as is usual in the hls interpretation . that the magnetic gauge group is completed to @xmath148 above the higgsing scale
is necessary for continuity on the moduli space ; as @xmath93 is increased we must recover the @xmath263 scenario .
equivalently , the electric quark mass term lifts some of the quasi - ngbs parameterised by @xmath2-mesons . indeed
, one expects @xmath2-mesons composed of quarks with mass @xmath264 to themselves have mass greater than @xmath93 .
due to their identification with these states , one expects the masses of the magnetic gauge fields to be similarly raised . of particular interest is the choice @xmath265 .
now the electric theory has @xmath13 colours and @xmath13 flavours below the scale @xmath257 , so confines with its own quantum deformed moduli space [ eq : chqdefel ] -b = m^n_el^2n - n . by adjusting @xmath256 we can then vary between higgsed and confined electric phases . at the same time the magnetic theory varies between confined and higgsed phases .
we can exploit this effect to see that a _ higgsed _
@xmath148 hls emerges from a _ confining _ @xmath0 gauge theory as we shall see in the following section .
in fact , the choice @xmath260 and @xmath265 can be recast as a reversal of the hls interpretation .
another interesting corollary is the ability to fix the duality scale for a given electric quark expectation in certain regimes .
we generically find that @xmath84 should be chosen such that the scale of symmetry breaking is the same in both electric and magnetic theories . in the following ,
we choose a specific , single point in moduli space with @xmath266 .
consider first the case with no electric quark mass terms .
using eq . when @xmath267 we have @xmath243 , ergo @xmath244 ( up to phase factors ) .
can then be solved for @xmath84 :
^n = v^2(n - n)_mg^2n - n .
this should be compared with the relationship between electric and magnetic dynamical scales ^n+n=_el^2n - n_mg^2n - n usually found in seiberg duality .
solving for @xmath84 we find [ eq : chmu ] = _ el()^2(n - n)/n = v()^(n-2n)/n and therefore _
mg=_el()^2(n^2-n^2)/n(n-2n)=v()^n(2n - n)/n(n-2n ) for a given choice of @xmath240 and @xmath93 .
the magnitude of @xmath203 can also be compared to the magnetic quark mass m_q==_el()^2n / n = v()^(2n - n)/n generated by the meson expectation , as well as the confinement scale of the magnetic theory @xmath93 .
( as a function of logarithmic rg scale @xmath268 ) in massless sqcd when @xmath269 . blue denotes the electric theory , red the magnetic and dashed lines a higgsed gauge group .
_ left _ : the conformal window for @xmath270 and @xmath271 .
_ right _ : the free magnetic phase for @xmath272 and @xmath271.[fig : chrgv],title="fig:",scaledwidth=40.0% ] ( as a function of logarithmic rg scale @xmath268 ) in massless sqcd when @xmath269 .
blue denotes the electric theory , red the magnetic and dashed lines a higgsed gauge group .
_ left _ : the conformal window for @xmath270 and @xmath271 .
_ right _ : the free magnetic phase for @xmath272 and @xmath271.[fig : chrgv],title="fig:",scaledwidth=40.0% ] these expressions are only valid when the perturbative arguments of section [ sec : rho ] are valid , so can reliably be used to set @xmath273 and @xmath274 ; i.e. @xmath241 such that fluctuations in @xmath194 and @xmath209 are small .
choosing @xmath275 puts the theory in the conformal window .
inspection of the above formulae then gives @xmath276 .
alternatively , choosing @xmath16 puts us in the free magnetic phase and yields @xmath277 . in both cases
the low energy behaviour is the same .
the magnetic theory confines at the scale @xmath93 whereas the electric theory is higgsed and provides a perturbative , low energy description .
some example rg flows are shown in figure [ fig : chrgv ] .
trying to take @xmath93 below @xmath240 in this framework entails higgsing the electric theory in a strongly coupled electric regime , hence one looses some degree of theoretical control .
the magnetic theory also appears to confine below its dynamical scale in the conformal window , and above it in the free magnetic phase .
while the former is conceptually okay ( despite not yielding a useful low energy description ) the latter does not make physical sense .
this tells us that the assumptions used to fix @xmath84 break down , as expected .
it seems likely that , rather than being determined by eq .
, @xmath84 is fixed at @xmath240 for all choices @xmath278 .
now consider turning on an electric quark mass @xmath279 , with @xmath267 hence @xmath243 maintained .
when @xmath280 there are no qualitative changes to the above ; the higgsing scale of the electric theory remains higher than the confinement scale induced by the quark mass terms , i.e. @xmath281 .
similarly , the confinement scale of the magnetic theory remains higher than the higgsing scale induced by the quark mass terms .
however , when @xmath282 ( corresponding to @xmath283 ) the confinement and higgsing scales of electric and magnetic theories all become equal .
one can not decrease @xmath93 any further than this ; the electric quark expectations are fixed at @xmath262 by the geometry of the quantum deformed moduli space .
thus the scale appearing in eq .
should be replaced accordingly .
in addition , the magnetic quark expectations are no longer determined by confinement ; they are fixed at @xmath262 by the higgsing superpotential .
thus the scale appearing in eq .
should also be replaced . upon updating both scales ,
the two expressions for the magnetic gauge field remain consistent provided @xmath284 .
equivalently , the magnetic quarks are still normalised such that magnetic and electric symmetry breaking scales are equal .
this is analogous to the result @xmath244 found in section [ sec : rho ] on the mesonic branch , the implication being that the duality scale should always be chosen so as to enact this normalisation .
consequently , one has @xmath285 and therefore @xmath286 from the usual relationship between electric and magnetic scales .
note that the parameter @xmath93 is now redundant .
( as a function of logarithmic rg scale @xmath268 ) in massive sqcd when @xmath216 and @xmath287 . blue denotes the electric theory , red the magnetic and dashed lines a higgsed gauge group .
_ left _ : the conformal window for @xmath270 and @xmath271 .
_ right _ : the free magnetic phase for @xmath272 and @xmath271.[fig : chrgm],title="fig:",scaledwidth=40.0% ] ( as a function of logarithmic rg scale @xmath268 ) in massive sqcd when @xmath216 and @xmath287 .
blue denotes the electric theory , red the magnetic and dashed lines a higgsed gauge group .
_ left _ : the conformal window for @xmath270 and @xmath271 .
_ right _ : the free magnetic phase for @xmath272 and @xmath271.[fig : chrgm],title="fig:",scaledwidth=40.0% ] choosing @xmath288 one thus has @xmath289 in the conformal window and @xmath290 in the free magnetic phase .
the roles of electric and magnetic theories are then reversed : the electric theory confines in the ir , while the magnetic theory is higgsed and provides a perturbative , low energy description .
some example rg flows are shown in figure [ fig : chrgm ] .
just like @xmath93 in the massless case , trying to take @xmath257 below @xmath240 leads to ill defined low energy physics in this framework .
this is because the expansion scale in eq .
is now @xmath262 .
when @xmath288 , fluctuations of @xmath194 and @xmath209 are therefore sufficiently small for the expansion to be trusted . at smaller values of @xmath257 the perturbative approximation is invalidated . variation between confined and higgsed phases in the magnetic theory is clear .
the most useful regime is the clearly latter , where the electric theory confines and can be described at low energy by a higgsed magnetic theory .
however , seiberg duality is ( of course ) a duality , so the whole hls interpretation should work both ways .
we will shortly see that this is indeed true , but a brief examination of figures [ fig : chrgv ] and [ fig : chrgm ] already reveals an electric / magnetic exchange symmetry with @xmath291 and @xmath292 .
so far we have been focussing on how the hls maps onto the magnetic description in seiberg duality .
the hls theory describes the same low energy physics as the @xmath19 nonlinear sigma model of broken @xmath0 sqcd , which can be thought of as the linear uv completion .
however , it is interesting to ask what , given the magnetic hls theory , one can learn about the full @xmath0 electric theory if its form is not assumed a priori .
consider the khler potential of the electric theory in the symmetry restoring limit , @xmath293 , where we will take as read appropriate factors of @xmath294 to make it gauge invariant .
this term is proportional to the anomalous current @xmath295 , which is in turn related to the global @xmath1-current supermultiplet ( as described in @xcite ) . indeed , as we have seen , the modulus @xmath129 for the breaking of @xmath130 can play a role in the restoration of symmetry .
it can also be recast as a gauge field . therefore in order to answer this question one is motivated to first consider gauging @xmath130 .
this has to be carried out within the framework of superconformal @xmath296 supergravity @xcite .
the additional fields of interest for our discussion are the @xmath130 gauge field @xmath151 and the conformal compensator @xmath297 . assuming the usual @xmath1-charge for the gauginos of @xmath298 , the fields transform under a @xmath130 transformation as @xmath299 where @xmath300 are generic matter superfields ( i.e. quarks and mesons ) with superconformal @xmath1-charges @xmath301 and @xmath302 is a chiral superfield . note that the extra scaling symmetry introduced by superconformality changes @xmath129 from an m - type to a p - type superfield .
the general lagrangian is of the form @xmath303 for the superpotential @xmath8 to give an invariant term it should transform as @xmath304 , i.e. it has @xmath1-charge @xmath305 as usual .. these will be set to one here ( they are discussed in ref .
@xcite ) . also note this approach is not quite that taken in ref .
@xcite the @xmath151 here is the actual @xmath1-gauge field rather than the gauge field of a normal anomaly free gauge group .
] thus the relevant terms in the magnetic theory take the form @xmath306}+ \nonumber\\ { \int\mathrm{d}^{2}\theta\,}\phi^3 & { \mathop{\rm tr}\left[\phi q\tilde{q}\right]}+\mbox{h.c.}\end{aligned}\ ] ] where we have defined the canonically normalised meson by @xmath307 and the trace is over the @xmath308 flavour indices ( with contraction over @xmath148 colour being understood )
. we have taken degenerate expectations in the fi - term .
the @xmath1-charges @xmath309 are fixed by absence of @xmath310 anomalies , while @xmath311 is fixed by the superpotential coupling in eq .
: they are as given in table [ tab : hqxi ] . as per the seiberg magnetic dual
this theory has an _ anomalous _ global @xmath295 symmetry , under which the quarks and antiquarks have charge @xmath298 and the meson has charge @xmath312 .
in the normal hls fashion one solves the equation of motion of the @xmath151 field [ zebra ] ( r_q-)q^qe^r_qv_r+(r_-)^e^r_v_r+(r_-)^e^r_m v_r = v^2 which gives ( q^q+^)e^r_qv_r-2^e^r_m v_r = v^2 .
the lhs of this equation is the current @xmath313 , hence we can rewrite the equation as [ aardvark ] _ a = v^2 .
as we have seen , the magnetic theory spontaneously breaks the @xmath95 flavour symmetry to the colour - flavour locked @xmath314 because @xmath315 and @xmath316 have rank @xmath15 .
thus the quarks contribute only rank @xmath15 to the lhs of eq . , while the colour singlet mesons contribute the remaining rank @xmath13 expectation .
this breaks the original symmetry as @xmath317 .
however , the rhs of eq . is flavour symmetric and so , therefore , is the lhs . indeed
the expectations ( setting elements @xmath318 ) have to satisfy @xmath319 we shall return to this result below . now suppose that we knew only this magnetic theory , and wanted to infer the properties of the electric one by mapping the equations above .
more precisely , let us assume that there are no colour singlet states in the uv completion , and determine the form that eq . must take in terms of electric degrees of freedom . by definition
there is no @xmath148 hls in the electric theory to mix with the original @xmath155 flavour symmetry .
hence a necessary requirement for the @xmath10 subgroup to remain unbroken is that the lhs of eq .
maps in the electric theory to the sum of two hermitian @xmath320 matrices , whose ranks are given by @xmath321/2=n$ ] ( one to break @xmath322 and one to break @xmath323 ) .
it is a theorem that every rank @xmath13 , hermitian , @xmath320 matrix @xmath324 has a rank factorisation @xmath325 , where @xmath88 is an @xmath98 matrix .
therefore even when the electric theory is not weakly coupled , the lhs of eq .
is proportional to where @xmath88 and @xmath326 are some @xmath98 matrices . for a weakly coupled uv completion ,
@xmath88 and @xmath89 are of course identified with the electric quarks transforming as fundamental and anti - fundamental under @xmath0 .
. written in terms of weakly coupled canonically normalised electric degrees of freedom must take the form [ coyote ] ( r_q-)q^qe^r_qv_r+(r_-)^e^r_v_r = v^2 . equating the lhs of eqs . with the lhs of we find consistency
with the mapping of non - holomorphic operators discussed in @xcite which , among other things , enables one to track soft terms through strong coupling .
absence of @xmath327 anomalies then fixes @xmath328 , with the result that @xmath329 .
furthermore the other expectations in eq .
are consistent with the mapping @xmath330 and with the classical constraints @xmath331 .
of course , we knew from seiberg duality that this had to be the case since these identifications are known to be consistent with @xmath1-symmetry .
however , since a field s @xmath1-charge directly determines its expectation when one solves the @xmath151 equation of motion , the role of @xmath1-symmetry in the matching of moduli spaces is self - evident .
finally , if we begin with no meson in the magnetic theory ( and no superpotential there ) , there is clearly a role reversal : the rank condition on the magnetic quarks leaves an @xmath197 flavour symmetry unbroken , with the @xmath0 electric gauge symmetry acting as the hls for the anomaly free diagonal subgroup .
seiberg duality also exists for gauge theories based on @xmath6 and @xmath7 groups , so the hls interpretation ought to work for these as well . to show this ,
it is convenient to first invert the hls interpretation so as to derive the electric description instead .
dual gauge groups for @xmath6 and @xmath7 theories are then determined by the rank of meson expectation that causes the magnetic theory to confine . in some senses
this is actually a more natural way to think of the hls interpretation .
there is no need to introduce a superpotential and one automatically finds a confining original theory being described by a higgsed hls theory .
consider starting from a magnetic @xmath332 theory with @xmath15 colours and @xmath14 flavours , then giving an expectation to @xmath190 to provide a mass term for some of the magnetic quarks .
the maximum rank for this expectation is @xmath13 .
larger values lead to an effective theory with fewer flavours than colours , in which case an ads superpotential @xcite is generated and the theory no longer has a vacuum . when the rank condition is saturated the maximum possible flavour symmetry is preserved by choosing m = v & 0 + 0 & 0 whereupon @xmath145 is broken to @xmath333 .
integrating out the @xmath13 massive flavours the theory confines ( as described by eq . ) , further breaking the flavour symmetry to @xmath334 and completely breaking the gauge symmetry .
both surviving @xmath148 factors can be thought of as a mixture of flavour and global gauge transformations .
note that it is vital that all gauge fields are rendered massive if we are to find a sigma model ( and therefore hls ) description at low energy .
this will be important when we move onto the other gauge groups .
splitting generators into those that are broken and those that are not we find broken generators @xmath335 for a traceless , anti - hermitian matrix @xmath336 and arbitrary complex matrices @xmath337 and @xmath338 .
there are also identity matrices for the two broken @xmath57 symmetries .
the unbroken generators are @xmath339 for traceless , hermitian matrices @xmath340 , @xmath101 and @xmath110
. therefore @xmath341 where @xmath342 .
the chiral superfields @xmath129 and @xmath219 originate from the broken @xmath1- and baryon number symmetries respectively .
there are two projection operators : @xmath343 each commutes with @xmath344 so we are free to work with left or right cosets for @xmath33 and @xmath167 .
the @xmath140 projection of @xmath344 would lead to an @xmath148 gauge theory , simply reproducing the magnetic theory we first thought of .
we therefore consider the @xmath46 projection instead , which projects @xmath344 down to the generators of @xmath345 .
we choose cosets for @xmath33 and @xmath167 such that @xmath346 under nonlinear flavour transformations and define @xmath347 the @xmath155 part of the unbroken flavour symmetry is realised linearly in @xmath122 and @xmath123 .
this leads to a sigma model description with k_s = v^2 for a polynomial @xmath348 , which is the most general khler potential invariant under eq .
( and includes terms of the form ) . for the hls description
we write down a trial khler potential k_v = v^2 .
there is no fi - term as @xmath349 is traceless .
solving the equations of motion for @xmath158 gives _
_ ^e^v=_^_e^-v and so k_v=2v^2 . to reproduce the sigma model description we therefore write down a full hls description khler potential k = v^2+av^2 . defining dimensionful chiral superfields @xmath350 and @xmath351 , and absorbing all non - gauge terms in @xmath348 this becomes k=+v^2 i.e. an @xmath0 gauge theory with @xmath14 flavours of electric quark @xmath88 and @xmath89 .
the definitions of @xmath88 and @xmath89 , along with the constraint @xmath342 , determine their expectations to be rank @xmath13 and of order @xmath352 .
the electric gauge group is thus completely higgsed , as required .
seiberg duality for @xmath6 gauge groups @xcite can be summarised in table [ tab : vsso ] . starting from a magnetic theory with @xmath353 colours ( i.e. gauge group @xmath354 ) and @xmath14 flavours
, we again consider giving an expectation to the magnetic meson .
because of the superpotential this still give masses to magnetic quarks .
q & & & ( n-2)/(n+n ) + m & 1 & & 2(n+2)/(n+n ) + & so(n ) & su(n+n ) & u(1)_r + q & & & ( n+2)/(n+n ) + the maximum rank allowed for the meson expectation is @xmath13 , above which the theory has no vacuum due to the generation of an ads superpotential . assuming the rank condition is saturated , the flavour symmetry is broken to @xmath355 as the meson is in the symmetric representation of @xmath356 . at the same time
the gauge sector of the magnetic theory confines after integrating out the massive quarks , so a low energy sigma model description is appropriate .
since the generators of @xmath354 are antisymmetric tensors one can define the hybrid baryon b^_j_1 j_n=__1 _n+4w_^_1_2w_^_3_4q^_5_j_1
q^_n+4_j_n where @xmath357 ( being careful not to confuse the spinor index @xmath358 with the gauge indices @xmath359 ) is the antisymmetric field strength superfield .
the f - terms for @xmath190 fix the expectation of @xmath83 to be at most rank @xmath15 , so this is the only baryon - like operator whose expectation does not vanish .
it is a singlet under the residual @xmath148 so a non - zero expectation for @xmath360 leaves this part of the flavour symmetry intact . at the quark level , this can be understood as a mixing between real @xmath148 flavour transformations and complexified , global @xmath361 gauge transformations .
we then proceed exactly as before to find an hls description with gauge group @xmath362 , where the unbroken @xmath148 part of the flavour symmetry is realised linearly . since there is only one @xmath356 factor in the flavour symmetry
there are no @xmath89 superfields and the khler potential is [ eq : vsk ] k = v^2 + for @xmath362 gauge field @xmath158 , and @xmath88 transforming as in table [ tab : vsso ] .
this is exactly the electric theory anticipated .
note that the associated metric is smooth everywhere so there is no problem in taking the symmetry restoring limit @xmath363 .
furthermore , both the gauge and @xmath0 part of the flavour symmetry are broken by a maximal , rank @xmath13 expectation of @xmath88 following from the usual definition of @xmath88 .
this in turn implies a rank @xmath13 expectation for the electric meson @xmath364 and baryon @xmath365 .
the mapping of these expectations to their magnetic counterparts @xmath366 agrees with that expected from seiberg duality .
@xmath7 theories are extremely similar to @xmath6 theories from an hls point of view .
duality for them @xcite is summarised in table [ tab : vssp ] .
starting from a magnetic theory with @xmath367 colours ( i.e. gauge group @xmath368 the factor of 2 ensuring that the number of colours is even ) and @xmath369 flavours , we take the familiar approach of giving an expectation to the magnetic meson .
magnetic quarks are again rendered massive and , for this class of theory , the maximum allowed rank for the meson expectation is @xmath370 .
this leads to confinement with total flavour symmetry breaking , ergo no @xmath10 to form the basis of an hls description . taking the next largest rank of @xmath371 results in flavour symmetry breaking @xmath372 , as the meson is in the antisymmetric representation of @xmath373
the theory goes on to confine after integrating out the massive quarks , but with no further flavour symmetry breaking .
the hls description should therefore have gauge group @xmath374 and the @xmath375 part of the unbroken symmetry is realised linearly .
again , we explicitly point out that the confinement is vital in order for a sigma model description to apply at low energy .
the khler potential is that of eq . , but with @xmath158 an @xmath374 gauge superfield , and @xmath88 transforming as in table [ tab : vssp ] .
the conventional seiberg dual electric theory is therefore recovered . a rank @xmath371 expectation for @xmath88 breaks the gauge and @xmath376 part of the flavour symmetry . as for the @xmath6 version
there are no problems in the symmetry restoring limit @xmath363 and the mapping m^ijq^iq^j is in agreement with that predicted by seiberg duality .
q & & & ( n+1)/(n+n ) + m & 1 & & 2(n-1)/(n+n ) + & sp(2n ) & su(2n+2n ) & u(1)_r + q & & & ( n-1)/(n+n ) + theories with colour adjoints @xmath211 were examined in @xcite .
there is convincing evidence that in the presence of a superpotential for the adjoints w= an @xmath0 electric theory with @xmath377 flavours of quarks and anti - quarks maps to a magnetic @xmath378 theory with its own adjoint @xmath379 and superpotential @xmath380}$ ] .
the particle content in both theories is shown in table [ tab : kssem ] .
@xmath381 in order to understand this duality we first consider the moduli space of the electric theory . here
the adjoint field enhances the moduli space .
indeed the mesons are given by @xmath382 , where @xmath383 .
the @xmath211 equation of motion ensures that the chiral ring is truncated as @xmath384 .
flavour can in principle be broken by giving expectations to `` dressed '' quarks @xmath385 .
since these directions are not distinguished in the moduli space , their expectations break the enhanced complex flavour group @xmath386 . at this point
the rank condition in the electric theory still holds : we can only assign a rank @xmath13 expectation to the combined @xmath387 and @xmath388 system , and the flavour group is broken to @xmath389 . note that there are @xmath390 degrees of freedom available for the dressed quarks , and we require @xmath391 degrees of freedom for the p and m - type superfields .
hence we require @xmath392 which we can write as [ banana ] n_f+n>(2k-1)n_f .
the free magnetic window of adjoint sqcd is @xmath393 , with @xmath394 signalling @xmath395-confinement .
since by assumption @xmath396 , the condition in eq .
defining when we are able to use the hls formalism also seems to define the upper edge of the free magnetic window .
the hls description now closely follows the discussion in section 3 , but with the flavour symmetries enhanced to @xmath397 factors , and with the hls being assigned to the anomaly free diagonal @xmath398 group . in the standard coset description ,
@xmath135 and @xmath136 are @xmath399 matrices , with the latter index being identified with magnetic colour .
the full @xmath50 transforms as @xmath400 and similar for @xmath146 , where @xmath122 are flavour rotations in @xmath397 , giving precisely the degrees of freedom of the dressed magnetic quarks .
the flipped coset description producing the mesons goes over all @xmath397 indices and hence reproduces all @xmath401 mesons .
the discussion extends quite directly to the more complicated @xmath0 duals discussed in refs .
@xcite for example the sqcd model presented there with two adjoints has @xmath402 flavours of dressed quarks , and the magnetic gauge group is accordingly @xmath403 .
our main aim in this work has been to place seiberg duality on a more dynamical footing . in this way
we hope ultimately to use the duality to learn more about dynamical processes in strongly coupled theories , rather than just the properties of the lagrangians and vacua .
this section briefly summarises some of the applications we have in mind .
detailed investigations are left for future work .
an obvious application is composite gauge field scattering , as illustrated in figure [ fig : chrgm ] .
an electric theory where @xmath15 flavours have equal mass @xmath288 is taken to be the ` true ' theory , but it confines in the ir meaning that the low energy physics is obscured . fortunately , seiberg duality steps in and provides an alternative , perturbative description in the shape of a higgsed magnetic theory .
the magnetic gauge group is an emergent symmetry , with the massive gauge fields originating purely from composite operators .
if this were true , one would expect to see effects from the underlying electric theory near to the confinement scale
. in particular , signs of compositeness should start to appear in magnetic gauge field scattering amplitudes . using the hls interpretation
we can start to quantify such phenomena . as discussed in sections
[ sec : rho ] and [ sec : mu ] , magnetic gauge fields are explicitly related to electric @xmath2-mesons via v_mg^ where = _
el()^n/2n is the higgsing scale of the magnetic theory , and also the confinement scale of the electric theory .
schematically , one expects elastic scattering amplitudes for the longitudinal components of magnetic gauge bosons to grow with the centre of mass energy squared @xmath395 .
this divergence is the standard unitarity violation problem .
it is addressed in the current framework at the higgsing scale of the magnetic theory , whereupon higgs boson exchange unitarises the scattering .
the associated higgs field is itself a composite object so , in effect , we have a composite higgs model .
however , the higgsing description is short lived and the composite nature of the gauge bosons immediately becomes apparent .
their amplitudes are instead mapped onto the equivalent @xmath2-meson scattering amplitudes in the electric theory . in the perturbative regime
the leading order contribution of figure [ fig : parhos ] dominates , scaling as @xmath404 for running electric gauge coupling @xmath405 .
asymptotic freedom ensures that these amplitudes remain under control as the centre of mass energy is increased further .
the overall situation is illustrated in figure [ fig : pacs ] .
approaching from below the higgsing scale we anticipate the appearance of a resonance in @xmath406 scattering , corresponding to the magnetic higgs boson .
approaching from above we expect the @xmath407 amplitude to diverge as the electric theory becomes strongly coupled .
interpolating between the two therefore suggests a top heavy , broadened ` resonance ' around the higgsing / confinement scale @xmath262 . ) for the longitudinal components of magnetic gauge bosons . below the higgsing scale
the amplitude grows with @xmath395 before hitting a higgs resonance at @xmath262 . above the higgsing scale it matches onto the amplitude for elastic @xmath2-meson scattering in the electric theory , which goes like @xmath404 in the perturbative regime .
the dashed line crudely interpolates between the two perturbative regimes , tracing out a top heavy , broadened ` resonance'.[fig : pacs],scaledwidth=40.0% ] this could , for example , be the case in the electroweak sector of the standard model .
all we really know is that low energy physics is well described by a broken @xmath408 gauge theory
. we do not know that this is ever an actual symmetry of nature .
it could merely emerge as an effective description of some other , strongly coupled theory .
one can also consider weakly gauging the original diagonal @xmath148 flavour factor . in this case
we denote the full gauge group as @xmath409 .
@xmath410 is the usual , composite , magnetic colour factor and @xmath411 is the elementary factor from the gauged flavour symmetry .
the only difference in the resulting hls description is that the surviving diagonal subgroup su(n)_csu(n)_esu(n ) is now a gauge symmetry , which comes with _ partially _ composite gauge fields .
taking gauge couplings @xmath412 and @xmath413 for each factor , the tree level mass eigenstates are @xmath414 with mass squared @xmath415 and zero respectively .
the running gauge couplings are evaluated at the higgsing scale @xmath262 so the composition of the mass eigenstates can vary . in the free magnetic phase , @xmath412 increases and the heavy state becomes more composite for larger values of @xmath262 .
when @xmath416 , the coupling @xmath412 hits its landau pole and the heavy state is fully composite .
models with partially composite gauge fields arising from seiberg duality have recently been studied in the context of the electroweak sector of the mssm @xcite .
it has been suggested that they have several phenomenological advantages , including an increased higgs mass and a `` natural '' superpartner spectrum with light stops . in these models
the symmetry breaking typically occurs in two phases , corresponding to two different electric quark masses @xmath417 . at the higher scale @xmath418 the @xmath419 gauge symmetry
is broken to its diagonal subgroup , which is identified with the @xmath420 of the mssm . at @xmath421
this is further broken by the usual ( composite ) higgs fields . for @xmath422 the two processes
are well separated and the final @xmath420 breaking looks like a vanilla higgs mechanism ( albeit with a potential different from that usually found in the mssm ) . for @xmath423
this separation does not exist and the underlying strong coupling has a large effect .
the final @xmath148 breaking starts to look more like that of a technicolour model , being driven by confinement in the electric theory . between these two regimes
one has a composite higgs model , with the compositeness becoming more noticeable as the ratio @xmath424 is increased .
hence this framework can be used to continuously interpolate between higgsing and technicolour descriptions of ( supersymmetric ) electroweak symmetry breaking .
the hls interpretation allows a concrete uv completion ( i.e. electric theory ) to be defined for these models .
it also allows for the exploration of phenomenology away from the far ir or uv , such as the gauge boson scattering amplitudes discussed above .
much of the original work on hidden local symmetries was motivated by trying to understand the chiral lagrangian of low energy qcd . indeed ,
this was also the primary phenomenological focus of refs . @xcite .
consequently , our discussion would not really be complete without some brief comments on lessons that might be learned in this area .
real world qcd is not supersymmetric so there is a limit to how trustworthy any insights derived from seiberg duality can be . in particular , many of our conclusions rely on the existence of quasi - ngbs that simply do not exist in the absence of susy .
one could , for example , attempt to construct a chiral lagrangian accounting for heavy quarks by writing down an electric theory with @xmath425 .
three flavours ( top , bottom and charm ) have masses way above the electric ( i.e. qcd ) scale so are integrated out . the electric theory
subsequently confines at a scale @xmath426 . since @xmath427 for this theory the magnetic description has the same gauge group .
in addition , section [ sec : mu ] suggests @xmath428 for such theories .
the gauge group is higgsed to @xmath420 at a scale @xmath429 , which is further higgsed to nothing at a scale @xmath430 .
we therefore have magnetic gauge fields with masses @xmath431 for running gauge coupling @xmath432 , evaluated at each of the two higgsing scales . a qcd scale of @xmath433 mev would give @xmath434 gev and @xmath435 gev : too small to correspond to any of the vector mesons in the heavy quark sector of qcd .
however , this is to be expected .
the magnetic gauge fields arise from quasi - ngbs in sqcd , which have no right to remain massless without susy . it is therefore unsurprising that they end up with much larger masses in qcd , where susy is taken away .
an aspect of the above discussion _ not _ strongly affected by susy is the argument used to fix the value of @xmath3 in section [ sec : rho ] .
the general strategy was to derive two separate expressions for the magnetic gauge fields : one from the equations of motion turned out by the hls formalism , and one from the noether currents for the unbroken flavour symmetry . by comparing them
we were able to fix the normalisation of the magnetic quarks and , consequently , the value of @xmath3 .
the same idea ought to work for real world qcd . in sqcd
we knew exactly what the electric and magnetic quark expectations were , and could easily write down explicit expressions for the ngbs . in qcd
the moduli space is less well understood , with fermion , rather than scalar , expectations providing the order parameters for the symmetry breaking .
hence a direct translation of our results is not possible .
nonetheless , the hls formalism has been applied to massless , two flavour qcd @xcite where the quark condensate is assumed to break the chiral @xmath436 flavour symmetry to its diagonal subgroup .
the formalism results in hls gauge field equations of motion that set v^_=-^^_^for pions @xmath437 .
one can also write down the current for the unbroken flavour symmetry in the hls description . in the unitary gauge it is @xcite generators
are chosen to satisfy @xmath438}=2\delta^{\alpha\beta}$ ] . ]
[ eq : apj ] j^_=2av^2v^_+2v^2(a-2)^^_^+ if we then consider a pure gauge field state the current is given by @xmath439 .
substituting in the gauge field equations of motion simplifies the expression to @xmath440 . on the other hand
, one can calculate the current directly from the sigma model description to find @xmath441 .
consistency thus requires @xmath4 .
of course , there remains the question of how a pure gauge field state can exist when the equations of motion fix @xmath442 . in a sense
it can not .
but as far as the physical currents are concerned it can , provided @xmath4 such that the pion contribution disappears .
so @xmath4 can be considered a direct consequence of allowing ` pure ' gauge field states ( i.e. @xmath2-mesons ) to exist in the hls description of qcd .
this discussion brings us finally to non - supersymmetric dualities which , of course , it would be very interesting to establish . in the past
there have been various attempts in this direction .
none of them are quite as compelling as seiberg duality itself , mainly because non - supersymmetric theories do not usually have interesting moduli spaces .
other matching tests such as t hooft anomaly matching are significantly weaker , often not uniquely pinning down the dual description . moreover the most constraining anomalies involve the @xmath1-symmetry , which is not available in non - supersymmetric theories .
we have seen that the notion of an hls provides a somewhat more mechanical route to dualities . to establish a pair of seiberg duals one begins with the electric theory of interest .
its higgsing leads to a nonlinear sigma model that can in turn be expressed as a linearised hls theory .
the existence of @xmath1-symmetry then guarantees a modulus along which the gauge symmetries in both descriptions are smoothly restored .
this procedure does not rely on the usual tests of moduli space or anomaly matching . in principle at least , it is clear what would be required in order to establish such a duality without susy .
first one requires an electric theory that can be higgsed or confined to give a nonlinear sigma model .
this is linearised in the hls formalism , and the identification of magnetic degrees of freedom made in the same manner as for the susy theories . in order to be able to take a symmetry restoring limit one
would then need a scaling symmetry to be present .
thus the theories of interest would most likely be of the kind discussed in @xcite , in which scale invariance is spontaneously broken by a `` dilaton '' playing much the same role as the conformal compensator ( equivalently @xmath129 ) in the susy case .
having established a candidate duality in this way one could , of course , still apply all the usual tests .
any duality constructed via the hls formalism that passed them all would surely be on firm footing indeed .
s. r. coleman , j. wess and b. zumino , _ structure of phenomenological lagrangians . 1 .
_ , _ phys.rev .
_ * 177 * ( 1969 ) 22392247 .
j. callan , curtis g. , s. r. coleman , j. wess and b. zumino , _ structure of phenomenological lagrangians .
_ , _ phys.rev .
_ * 177 * ( 1969 ) 22472250 .
s. weinberg , _ the quantum theory of fields .
2 : modern applications_. cambridge university press , 1996 .
m. bando , t. kugo , s. uehara , k. yamawaki and t. yanagida , _ is the @xmath2 meson a dynamical gauge boson of hidden local symmetry ? _ , _ phys.rev.lett . _ * 54 * ( 1985 ) 1215 . m.
bando , t. kugo and k. yamawaki , _ nonlinear realization and hidden local symmetries _ , _ phys .
* 164 * ( 1988 ) 217314 .
n. seiberg , _ electric - magnetic duality in supersymmetric non - abelian gauge theories _ , _ nucl .
* b435 * ( 1995 ) 129146 [ http://arxiv.org/abs/hep-th/9411149[hep-th/9411149 ] ] .
k. a. intriligator and n. seiberg , _ lectures on supersymmetric gauge theories and electric - magnetic duality _ , _ nucl .
suppl . _ * 45bc * ( 1996 ) 128 [ http://arxiv.org/abs/hep-th/9509066 [ hep - th/9509066 ] ] .
k. a. intriligator and p. pouliot , _ exact superpotentials , quantum vacua and duality in supersymmetric @xmath443 gauge theories _ , _
_ * b353 * ( 1995 ) 471476 [ http://arxiv.org/abs/hep-th/9505006 [ hep - th/9505006 ] ] .
d. kutasov , _ a comment on duality in @xmath444 supersymmetric non - abelian gauge theories _ , _ phys .
* b351 * ( 1995 ) 230234 [ http://arxiv.org/abs/hep-th/9503086[hep-th/9503086 ] ] .
d. kutasov and a. schwimmer , _ on duality in supersymmetric yang - mills theory _ , _ phys .
* b354 * ( 1995 ) 315321 [ http://arxiv.org/abs/hep-th/9505004[hep-th/9505004 ] ] .
k. a. intriligator , r. g. leigh and m. j. strassler , _ new examples of duality in chiral and nonchiral supersymmetric gauge theories _ , _ nucl .
phys . _ * b456 * ( 1995 ) 567621 [ http://arxiv.org/abs/hep-th/9506148[hep-th/9506148 ] ] .
p. pouliot , _ chiral duals of nonchiral susy gauge theories _ , _ phys.lett .
_ * b359 * ( 1995 ) 108113 [ http://arxiv.org/abs/hep-th/9507018[hep-th/9507018 ] ] .
d. kutasov , a. schwimmer and n. seiberg , _ chiral rings , singularity theory and electric - magnetic duality _ , _ nucl .
phys . _ * b459 * ( 1996 ) 455496 [ http://arxiv.org/abs/hep-th/9510222[hep-th/9510222 ] ] .
pouliot and m. strassler , _ a chiral @xmath148 gauge theory and its nonchiral spin(8 ) dual _ , _ phys.lett . _
* b370 * ( 1996 ) 7682 [ http://arxiv.org/abs/hep-th/9510228[hep-th/9510228 ] ] .
j. h. brodie , _ duality in supersymmetric su(@xmath445 ) gauge theory with two adjoint chiral superfields _ , _ nucl .
phys . _ * b478 * ( 1996 ) 123140 [ http://arxiv.org/abs/hep-th/9605232[hep-th/9605232 ] ] .
j. h. brodie and m. j. strassler , _ patterns of duality in @xmath444 susy gauge theories or : seating preferences of theater - going non - abelian dualities _ , _ nucl .
* b524 * ( 1998 ) 224250 [ http://arxiv.org/abs/hep-th/9611197[hep-th/9611197 ] ] .
s. abel and j. barnard , _ electric / magnetic duality with gauge singlets _ , _ jhep _ * 0905 * ( 2009 ) 080 [ http://arxiv.org/abs/0903.1313 [ 0903.1313 ] ] .
n. craig , r. essig , a. hook and g. torroba , _ new dynamics and dualities in supersymmetric chiral gauge theories _ , _ jhep _ * 1109 * ( 2011 ) 046 [ http://arxiv.org/abs/1106.5051[1106.5051 ] ] .
n. craig , r. essig , a. hook and g. torroba , _ phases of n=1 supersymmetric chiral gauge theories _ , _ jhep _ * 1112 * ( 2011 ) 074 [ http://arxiv.org/abs/1110.5905[1110.5905 ] ] . m. harada and k. yamawaki , _ conformal phase transition and fate of the hidden local symmetry in large @xmath377 qcd _ , _ phys.rev.lett . _
* 83 * ( 1999 ) 33743377 [ http://arxiv.org/abs/hep-ph/9906445 [ hep - ph/9906445 ] ] .
m. harada and k. yamawaki , _ hidden local symmetry at loop : a new perspective of composite gauge boson and chiral phase transition _ , _ phys.rept . _ * 381 * ( 2003 ) 1233 [ http://arxiv.org/abs/hep-ph/0302103[hep-ph/0302103 ] ] .
z. komargodski , _ vector mesons and an interpretation of seiberg duality _ , _ jhep _ * 02 * ( 2011 ) 019 [ http://arxiv.org/abs/1010.4105 [ 1010.4105 ] ] .
r. kitano , _ hidden local symmetry and color confinement _ , _ jhep _ * 1111 * ( 2011 ) 124 [ http://arxiv.org/abs/1109.6158 [ 1109.6158 ] ] .
m. bando , t. kuramoto , t. maskawa and s. uehara , _ nonlinear realization in supersymmetric theories _ , _
prog.theor.phys . _
* 72 * ( 1984 ) 313 .
s. abel , m. buican and z. komargodski , _ mapping anomalous currents in supersymmetric dualities _ , _ phys.rev .
_ * d84 * ( 2011 ) 045005 [ http://arxiv.org/abs/1105.2885[1105.2885 ] ] .
s. ferrara , l. girardello , t. kugo and a. van proeyen , _ relation between different auxiliary field formulations of @xmath444 supergravity coupled to matter _ , _
_ * b223 * ( 1983 ) 191 .
r. barbieri , s. ferrara , d. v. nanopoulos and k. stelle , _ supergravity , r invariance and spontaneous supersymmetry breaking _ , _
phys.lett . _ * b113 * ( 1982 ) 219 .
a. h. chamseddine and h. k. dreiner , _ anomaly free gauged r - symmetry in local supersymmetry _ , _ nucl.phys . _
* b458 * ( 1996 ) 6589 [ http://arxiv.org/abs/hep-ph/9504337[hep-ph/9504337 ] ] . m. a. luty and r. rattazzi , _ soft supersymmetry breaking in deformed moduli spaces , conformal theories , and @xmath446 yang - mills theory _ , _ jhep _ * 9911 * ( 1999 ) 001 [ http://arxiv.org/abs/hep-th/9908085[hep-th/9908085 ] ] .
i. affleck , m. dine and n. seiberg , _ dynamical supersymmetry breaking in supersymmetric qcd _ , _ nucl .
* b241 * ( 1984 ) 493534 .
i. affleck , m. dine and n. seiberg , _ dynamical supersymmetry breaking in four - dimensions and its phenomenological implications _ , _ nucl .
phys . _ * b256 * ( 1985 ) 557 . k. a. intriligator and n. seiberg , _ duality , monopoles , dyons , confinement and oblique confinement in supersymmetric @xmath447 gauge theories _ , _
_ * b444 * ( 1995 ) 125160 [ http://arxiv.org/abs/hep-th/9503179[hep-th/9503179 ] ] .
s. abel and t. gherghetta , _ a slice of @xmath448 as the large @xmath13 limit of seiberg duality _ , _ jhep _ * 1012 * ( 2010 ) 091 [ http://arxiv.org/abs/1010.5655[1010.5655 ] ] .
n. craig , d. stolarski and j. thaler , _ a fat higgs with a magnetic personality _ , _ jhep _ * 1111 * ( 2011 ) 145 [ http://arxiv.org/abs/1106.2164[1106.2164 ] ] . c. csaki , y. shirman and j. terning , _ a seiberg dual for the mssm : partially composite @xmath8 and @xmath212 _ , _ phys.rev .
_ * d84 * ( 2011 ) 095011 [ http://arxiv.org/abs/1106.3074[1106.3074 ] ] . c. csaki , l. randall and j. terning , _ light stops from seiberg duality _ , http://arxiv.org/abs/1201.1293[1201.1293 ] . m. shaposhnikov and d. zenhausern , _ quantum scale invariance , cosmological constant and hierarchy problem _ , _ phys.lett .
_ * b671 * ( 2009 ) 162166 [ http://arxiv.org/abs/0809.3406[0809.3406 ] ] . | it is widely believed that the emergent magnetic gauge symmetry of sqcd is analogous to a hidden local symmetry ( hls ) .
we explore this idea in detail , deriving the entire ( spontaneously broken ) magnetic theory by applying the hls formalism to spontaneously broken @xmath0 sqcd .
we deduce the khler potential in the hls description , and show that gauge and flavour symmetry are smoothly restored along certain scaling directions in moduli space .
we propose that it is these symmetry restoring directions , associated with the @xmath1-symmetry of the theory , that allow full seiberg duality . reconsidering the origin of the magnetic gauge bosons as the @xmath2-mesons of the electric theory ,
colour - flavour locking allows a simple determination of the parameter @xmath3 .
its value continuously interpolates between @xmath4 on the baryonic branch of moduli space corresponding to vector meson dominance " and @xmath5 on the mesonic branch .
both limiting values are consistent with previous results in the literature .
the hls formalism is further applied to @xmath6 and @xmath7 groups , where the usual seiberg duals are recovered , as well as adjoint sqcd .
finally we discuss some possible future applications , including ( naturally ) the unitarisation of composite @xmath8 scattering , blended higgs / technicolour models , real world qcd and non - supersymmetric dualities . |
it seems plausible that the true origin of high - temperature superconductivity is found in a proper combination of the finite - range coulomb repulsion with a significant finite - range epi as suggested by a growing number of experimental and theoretical studies @xcite . in highly polarizable ionic lattices like cuprate superconductors both interactions are quite strong ( of the order of 1 ev ) compared with the low fermi energy of doped carriers because of a poor screening by non- or near - adiabatic carriers @xcite . in those conditions
the bcs - eliashberg theory @xcite breaks down because of the polaronic collapse of the electron bandwidth @xcite .
the many - body theory for polarons has been developed for extremely weak and strong epi . in the weak - coupling limit this problem
is reduced to the study of a structure factor of the uniform large polaron gas @xcite . for strong coupling
the problem is reduced to on - site @xcite or inter - site @xcite small bipolarons on a lattice .
a strong enhancement of t@xmath3 was predicted in the crossover region from the bcs - like polaronic to bec - like bipolaronic superconductivity due to a sharp increase of the density of states in a narrow polaronic band @xcite , which is missing in the so - called _ negative _ hubbard @xmath4 model .
nevertheless the theory of dense polaronic systems in the intermediate coupling regime remains highly cumbersome , in particular , when epi competes with strong electron correlations .
corresponding microscopic models with the on - site hubbard repulsion and the short - range holstein epi have been studied using powerful numerical techniques @xcite . in most analytical and numerical studies
mentioned above and many others both interactions are introduced as input parameters not directly related to the material .
quantitative calculations of the interaction matrix elements can be performed from pseudopotentials using the density functional theory ( dft ) @xcite . on the other hand
, one can express the bare coulomb repulsion and epi through material parameters rather than computing them from first principles in many physically important cases @xcite .
in particular , for a polar coupling to longitudinal optical phonons ( the frhlich epi ) , which is the major epi in polar crystals , both the momentum dependence of the matrix element , @xmath5 , and its magnitude are well known , @xmath6 with a dimensionless @xmath7 , where @xmath8 is a unit cell volume , @xmath9 is the number of unit cells in a crystal , @xmath10 is the optical phonon frequency , and @xmath11 . the high - frequency , @xmath12 and the static , @xmath13 dielectric constants are both measurable in a parent polar insulator .
as is well known , a two - particle bound state exists even in the weak - coupling regime , @xmath14 , due to a quantum ( exchange ) interaction between two large polarons forming a _ large bipolaron _
@xcite(@xmath15 is the familiar epi constant of the bcs - eliashberg theory ) .
these weakly coupled large pairs overlap in dense systems , so that their many - particle ground state is a bcs - like superconductor with cooper pairs ( see below ) . here
the analytical multi - polaron theory is given in the strong - coupling regime for highly polarizable lattices with @xmath16 .
the dielectric response function of strongly correlated electrons is _ apriori _ unknown .
hence one has to start with a generic hamiltonian including _ unscreened _ coulomb and frhlich interactions operating on the same scale since any ad - hoc assumption on their range and relative magnitude might fail , @xmath17+h_{ph}. \label{hamiltonian}\end{aligned}\ ] ] here @xmath18 is the bare hopping integral , @xmath19 is the chemical potential , @xmath20 and @xmath21 include both site @xmath22 and spin @xmath23 states , @xmath24 , @xmath25 are electron and phonon operators , respectively , @xmath26 is a site occupation operator , and @xmath27 is the polar vibration energy . in highly polarizable lattices with @xmath28 the familiar lang - firsov ( lf ) @xcite canonical transformation @xmath29 is particulary instrumental with @xmath30 $ ] .
it shifts the ions to new equilibrium positions changing the phonon vacuum , and removes most of _ both _ interactions from the transformed hamiltonian , @xmath31 , @xmath32where @xmath33 is the renormalised hopping integral involving the multi - phonon transitions described with @xmath34 $ ] , and @xmath35 is the chemical potential shifted by the polaron level shift , @xmath36here , the integration goes over the brillouin zone ( bz ) and @xmath37 ev in la@xmath38cuo@xmath39 @xcite .
the electron - phonon coupling constant is defined as @xmath40 . in the case of 2d carriers with a constant bare density of states ,
@xmath41 per spin , eq.([shift ] ) places cuprates in the strong - coupling regime , @xmath42 , if the bare band mass @xmath43 ( here @xmath44 is the in - plane lattice constant ) .
for comparison , let us first consider the weak - coupling limit , where not only @xmath45 but also the number of phonons dressing the carrier is small , @xmath46 . in this limit one can expand @xmath47 in eq.([trans ] ) in powers of @xmath48 keeping just single - phonon transitions so that ( in the momentum representation ) @xmath49 where @xmath50 , @xmath51 is the bare band dispersion , and @xmath52/\sqrt{2n}$ ] is the transformed epi matrix element , renormalised by the coulomb repulsion .
there are no other interactions left in the transformed hamiltonian since the bare coulomb repulsion is nullified by the frhlich epi . applying the bcs - eliashberg formalism @xcite yields the master equation for the superconducting order parameter , @xmath53 , @xmath54 where @xmath55 $ ] is the phonon propagator and @xmath56 are the matsubara frequencies ( @xmath57 ) .
depending on a particular shape of the band dispersion , eq.([master ] ) allows for different symmetries of the order parameter since epi is not local @xcite .
here we confine our analysis to a simple estimate of t@xmath3 by assuming a @xmath58-independent gap function , @xmath59 . then factorizing the kernel in eq.([master ] ) on the `` mass shell '' , @xmath60 and linearizing eq.([master ] ) with respect to the gap function one obtains the familiar estimate of the critical temperature , @xmath61 $ ] , where @xmath62 is the coulomb pseudopotential . in our case the weak - coupling bcs superconductivity with @xmath63 exists _ exclusively _ due to the `` tolmachev - morel - anderson '' logarithm @xmath64 , if the epi is retarded ( i.e. @xmath65 ) .
actually the number of virtual phonons in the polaron cloud is large in oxides and some other polar lattices , @xmath66 with the characteristic ( oxygen ) optical phonon frequency @xmath67 mev , so that multi - phonon vertexes are essential in the expansion of the hopping operator @xmath68 . to deal with this challenging problem
let us single out the coherent hopping in eq.([trans ] ) averaging @xmath69 with respect to the phonon vacuum , and consider the remaining terms as perturbation , @xmath70 . here
@xmath71 describes free phonons and polarons coherently propagating in a narrow band with the exponentially diminished hopping integral , @xmath72 $ ] , @xmath73 , \label{t}\ ] ] and @xmath74 is the residual polaron - multiphonon interaction , which is a perturbation at large @xmath75 . in the diagrammatic technique
the corresponding vertexes have any number of phonon lines as shown in fig.sigma for the second - order in @xmath76 polaron self - energy ( @xmath77 ) and the phonon self - energy ( @xmath78 ) @xcite , where @xmath79 is the lattice coordination number and @xmath80 is the atomic density of carriers .
hence the perturbation expansion in @xmath81 is applied if @xmath82 @xcite .
importantly there is no structural instability in the strong coupling regime since @xmath83 @xcite .
the lf transformation , eq.([trans ] ) is exact for any adiabatic ratio @xmath84 .
however , if the perturbation expansion in @xmath85 is restricted by lowest orders , then it significantly overestimates polaron masses in the adiabatic regime , @xmath86 , for the case of the short - range ( holstein ) epi @xcite ( here @xmath87 is the nearest - neighbor bare hopping integral ) .
the polaronic band narrowing factor , @xmath88 becomes very small for this epi in the strong - coupling regime , which would eliminate any possibility of high temperature superconductivity and even metallicity of the small hosltein polarons . in our case of the long - range ( frhlich ) epi ,
quantum monte - carlo simulations @xcite show that the lf transformation provides numerically accurate polaron masses already in the zero order of the inverse - coupling expansion both in the adiabatic regime as well as in the non - adiabatic one for _ any _ strength of the frhlich epi
. moreover , such small polarons @xcite and small bipolarons @xcite are perfectly mobile in the relevant range of the coupling and the adiabatic ratio @xcite .
the perturbation @xmath76 has no diagonal matrix elements with respect to phonon occupation numbers .
hence it can be removed from the hamiltonian in the first order using a second canonical transformation @xmath89 with @xmath90 here @xmath91 and @xmath92 , respectively .
taking into account that the polaron fermi energy is small compared with the phonon energy at strong coupling and/or sufficiently low doping @xcite , one can neglect its contribution to @xmath93 and project the second - order in @xmath85 hamiltonian @xmath94 onto the phonon vacuum @xmath95 with the following result @xmath96 where @xmath97\hat{x}^{\dagger}_{i^\prime } \hat{x}% _ { j^\prime}|0\rangle , \label{int}\end{aligned}\ ] ] and @xmath98 is the heisenberg multi - phonon operator obtained by replacing @xmath99 in @xmath100 with @xmath101 . calculating the integral , eq.([int ] ) with @xmath102 yields @xmath103 where @xmath104 $ ] .
all matrix elements , eq.([v2 ] ) , of the polaron - polaron interaction are small compared with the polaron kinetic energy except the _ exchange _ interaction , @xmath106 such that @xmath107 . using @xmath108 with @xmath109 and @xmath110 ( for large @xmath111 ) one obtains a substantial @xmath112 , which is much larger than the polaron hopping integral , @xmath113 in the strong coupling limit . here
@xmath1 is the nearest - neighbor polaron hopping integrals . keeping only this exchange we finally arrive with the polaronic `` t - j@xmath0 '' hamiltonian , @xmath114 where @xmath115 is the spin 1/2 operator ( @xmath116 are the pauli matrices ) , @xmath117 , and @xmath118 is the chemical potential further renormalized by @xmath76 .
there is a striking difference between this polaronic t - j@xmath0 hamiltonian and the familiar t - j model derived from the repulsive hubbard u hamiltonian in the limit @xmath119 omitting the so - called three - site hoppings and epi @xcite .
the latter model acts in a projected hilbert space constrained to no double occupancy . within this standard t - j model the bare transfer amplitude of electrons ( @xmath1 )
sets the energy scale for incoherent transport , while the heisenberg interaction ( @xmath120 ) allows for spin flips leading to coherent hole motion with an effective bandwidth determined by @xmath121 . using the gutzwiller - type approximation to remove the constraint results in an unconstrained t - j model also containing a band narrowing , but purely electronic rather than phononic origin @xcite . on the contrary in our polaronic t - j@xmath0 hamiltonian , eq.([tj ] ) there is no constraint on the double on - site occupancy since the coulomb repulsion is negated by the frhlich epi .
the polaronic hopping integral @xmath1 leads to the coherent ( bi)polaron band and the antiferromagnetic exchange of purely phononic origin @xmath122 bounds polarons into small superlight inter - site bipolarons .
last but not least the difference is in the `` + '' sign in the last term of eq.([tj ] ) proportional to @xmath123 , which protects the ground superconducting state from the bipolaron clustering , in contrast with the `` - '' sign in the similar term of the standard t - j model , where the phase separation is expected at sufficiently large j @xcite .
-0.5 mm the cancelation of the bare coulomb repulsion by the frhlich epi is accurate up to a @xmath124 correction .
this correction produces a long - range residual repulsion of ( bi)polarons in the transformed hamiltonian , eq.([tj ] ) , which is small as soon as @xmath125 . the residual repulsion results in some screening of the coulomb interactions responsible for the doping dependence of the bipolaron binding energy and of the ( bi)polaron mass @xcite .
in layered polar insulators the static dielectric constant could be anisotropic , which together with local field corrections might result in different epi matrix elements for in - plane and out - of - plane polarised optical phonons , respectively .
the difference is not considered here .
the polaronic t - j@xmath0 hamiltonian , eq.([tj ] ) is analytically solvable in the limit of sufficiently low atomic density of carriers , @xmath126 . neglecting the first term in @xmath94 , which is the polaron kinetic energy proportional to @xmath127
, one can readily diagonalise the remaining spin - exchange part of the hamiltonian .
its ground state is an ensemble of inter - site singlet bipolarons with the binding energy @xmath128 localised on nearest neighbor sites .
such small bipolarons repel each other and single polarons via a short - range repulsion of about @xmath129 .
the kinetic energy operator in eq.([tj ] ) connects singlet configurations in the first and higher orders with respect to the polaronic hopping integrals . taking into account only the lowest - energy degenerate singlet configurations and discarding all other configurations one can project the t - j@xmath0 hamiltonian onto the inter - site bipolaronic hamiltonian using the bipolaron annihilation operators @xmath130 , where @xmath131 connects nearest neighbors @xcite .
these operators are similar to the bond - order operators introduced later by newns and tsuei @xcite , which are weakly coupled in their model with the single - plane lattice vibrations via a nonlinear ( two - phonon ) epi [ eq.(9 ) in ref .
actually it is well known that the nonlinear anharmonic corrections to epi are small compared with the linear frhlich interaction in real three - dimensional solids , which makes the two - dimensional model of ref.@xcite of cuprate superconductors unfeasible .
our strong - coupling projection is illustrated using a polar lattice , sketched in fig.[toy]a , of anion - cation triangular planes ( the in - plane lattice constant is @xmath44 and the nearest - neighbor hopping distance is @xmath132 ) separated by the out - of - plane lattice constant @xmath133 . for a zig - zag ladder - fragment of the lattice , fig.toyb , the projected bipolaronic hamiltonian in the nearest - neighbor hopping approximation is @xmath134 where @xmath135 are annihilation operators of the four degenerate singlets .
fourier transformation @xmath136 yields four bipolaronic bands , @xmath137 $ ] , @xmath138 $ ] with the center - of - mass momentum @xmath139 . expanding in powers of @xmath140 one obtains the effective mass of these small singlets , @xmath141 , where @xmath142 is the polaron mass .
a similar hamiltonian can be derived also for a square lattice , if next - nearest - neighbor hopping integrals are taking into account @xcite .
small bipolarons are hard - core bosons with the short - range repulsion of the radius @xmath143 and a huge anisotropy of their effective mass since their inter - plane hopping is possible only in the second order of @xmath1 @xcite .
the occurrence of superconductivity in bipolaronic strong coupling systems is not controlled by the pairing strength , but by the phase coherence among the electron pairs @xcite .
while in two dimensions bose condensation does not occur in either the ideal or the interacting system , there is a phase transition to a superfluid state at @xmath144}}\ ] ] in the dilute bose gas @xcite ( here @xmath145 is the boson density per unit area ) . using eqs.([shift ] , [ t ] )
we obtain @xmath146 and @xmath147 , allowing for a quantitative estimate of t@xmath3 ( here @xmath148 ) . with typical values of @xmath149 nm , @xmath150 nm , @xmath151 , the bare band mass @xmath152 , @xmath153 mev and the moderate atomic density of polarons , @xmath154 ( avoiding an overlap of bipolarons ) one obtains @xmath155 ev , @xmath156 , and @xmath157k .
importantly , the projection procedure of reducing eq.([tj ] ) to eq.([bip ] ) is well justified since the ratio @xmath158 is small and @xmath159 , so that only the lowest singlet configurations can be included while discarding the others . in conclusion , it seems very likely that a peculiar cancelation of the long - range coulomb repulsion by the long - range frhlich epi can help much in producing high - temperature superconductivity in doped polar insulators such as cuprates and other oxides , for instance bakbio .
the polaronic t - j@xmath0 hamiltonian , eq.([tj ] ) derived here from the bare long - range coulomb interactions could provide a novel avenue for analytical and computational studies of superconductivity in complex ionic lattices since the repulsive hubbard u model and its strong - coupling t - j projection do not explain high t@xmath3 @xcite .
the author thanks alexander bratkovsky , janez bona , jozef devreese , holger fehske , viktor kabanov , dragan mihailovi , peter prelovsek , john samson and g. sica for helpful discussions .
this work was partially supported by the royal society ( london ) .
alexandrov a. s. , physica c ( amsterdam ) * 182 * , 327 ( 1991 ) ; phys .
b * 53 * , 2863 ( 1996 ) .
aubry s. , j. physique ( france ) iv colloque c**2 * * , 349 ( 1993 ) ; bona j. and trugman s. a. , phys .
b * 64 * , 094507 ( 2001 ) ) ; alexandrov a. s. and kornilovitch p. e. , j. phys . : condens .
matter * 14 * , 5337 ( 2002 ) ; macridin a. , sawatzky g. a. and jarrell m. , phys . rev .
b * * 69 * * , 245111 ( 2004 ) : hague j. p. , kornilovitch p. e. , samson j. , and alexandrov a. s. , phys . rev . lett . * 98 * , 037002 ( 2007 ) .
alexandrov a. s. and kornilovitch p. e. , j. phys . : condens .
matter * 14 * , 5337 ( 2002 ) .
alexandrov a. s. , kabanov v. v. and mott n. f. , phys .
lett . * 77 * , 4796 ( 1996 ) popov n. , theor .
phys . * 11 * , 565 ( 1972 ) . | many high - temperature superconductors are highly polarizable ionic lattices where the frhlich electron - phonon interaction ( epi ) with longitudinal optical phonons creates an effective attraction of doped carriers virtually equal to their coulomb repulsion .
the general multi - polaron theory is given with both interactions being strong compared with the carrier kinetic energy so that the conventional bcs - eliashberg approximation is inapplicable .
the many - electron system is described by the polaronic t - j@xmath0 hamiltonian with reduced hopping integral , @xmath1 , allowed double on - site occupancy , large phonon - induced antiferromagnetic exchange , @xmath2 , and a high - temperature superconducting state of small superlight bipolarons protected from clustering . |
vw ari ( hd 15165 , bds 1269 ) is a remarkable visual binary system consisting of two components : vw ari a ( v=6.@xmath671 , a - type ) and its companion ( v=8.@xmath633 , f - type ) . the primary vw ari a is a multiperiodic pulsating star ( probably of @xmath7 sct - type ) having non - radial modes .
this star shows the spectrum typical of very metal - deficient stars .
the rather high @xmath8 value found for this star , makes it difficult to derive accurate elemental abundances .
a first attempt was undertaken by andrievsky et al .
( 1995 ) , who showed that calcium and iron are strongly deficient in the atmosphere of vw ari a , while the secondary component possesses a solar - like chemical composition .
such a strange discrepancy between the metallicities of the two components can be explained by several hypotheses .
for example , these stars possibly do not constitute a physical pair or , in case they do , such an unusual stellar system could be formed as a result of stellar capture .
nevertheless , taking into account that 1 ) with a rather high probability vw ari is a binary system and 2 ) the probability of stellar capture in the field is too small , we propose that the difference in chemical composition of both components could appear simply due to the peculiar evolution of vw ari a as a @xmath2 boo - type star . the atmospheres of this type of stars are known to be strongly deficient in some heavy metals , while cno - elements exhibit solar - like abundances ( see e.g. strenburg , 1993 ) . to check this hypothesis , we performed a detailed spectroscopic analysis of vw ari ( primary component of the system ) based on the spectral synthesis technique .
seven ccd spectra have been obtained on 21 november 1994 with the chelle spectrometer lynx ( modified version : 29 spectral orders with the length of each order @xmath9 60 ) on the 6-m telescope ( special astrophysical observatory of the russian academy of sciences , russia , northern caucasus ) . the detailed description of the spectrometer is given by panchuk et al .
the resolving power was 24000 , s / n @xmath9 100 .
the spectral region was 5035 - 7185 .
the epochs at mid - exposures were the following : jd 2449670 + 1 ) 8.158 , 2 ) 8.165 , 3 ) 8.186 , 4 ) 8.215 , 5 ) 8.232 , 6 ) 8.247 , 7 ) 8.263 .
all spectra have been reduced using the dech20 code ( galazutdinov , 1992 ) , which includes extraction of spectra from images , dark and cosmic hits subtraction , flat - field correction , wavelength calibration , etc .
the effective temperature and gravity for vw ari a ( t@xmath0=7200 k , @xmath1=3.7 ) were estimated using the photometric indices @xmath10 and @xmath11 , and the calibration by kurucz ( 1991 ) .
we adopted a microturbulent velocity of 3 kms@xmath12 , which is appropriate for a - f main - sequence stars , and @xmath13 kms@xmath12 was taken from abt ( 1980 ) .
the starsp code ( tsymbal , 1996 ) was applied to derive the elemental abundances .
the atmosphere model was interpolated from kurucz s ( 1992 ) grid .
the input oscillator strengths of the investigated lines and blends were initially corrected by comparison of the solar synthetic spectrum ( solar model from kurucz s grid , @xmath14 kms@xmath12 and solar abundances from grevesse and noels , 1993 ) with the solar flux spectrum ( kurucz et al . 1984 ) .
the resulting abundances were found by means of the optimal fitting of the synthetic spectrum to the observed one .
they are given in table 1 .
.abundances for vw ari a [ cols="<,^,^,^,^,^,^,^,^,^,^,^,^,^ " , ]
the abundance pattern in the atmosphere of vw ari resembles that of @xmath2 boo - type stars ( see , e.g. strenburg , 1993 , andrievsky et al . , 1998 ) : normal abundances ( or slight underabundances ) of carbon and oxygen and strong deficiency of other elements .
an additional confirmation that vw ari could be a @xmath2 boo star is its position in @xmath15 diagrams .
this star possesses photometric characteristics which place it exactly in the region occupied by @xmath2 boo stars . supposing that vw ari belongs to the @xmath2 boo group , one can also easily explain the remarkable difference between the metallicities of this star and of its companion f - star with solar - like abundances ( andrievsky et al . , 1995 ) .
abt , h.a . : 1980 , _ publ .
_ , * 92 * , 796 andrievsky , s.m . ,
chernyshova , i.v . ,
usenko , i.a . ,
kovtyukh , v.v . ,
panchuk , v.e . ,
galazutdinov , g.a . : 1995 , _ publ .
_ , * 107 * , 219 andrievsky , s.m . ,
chernyshova , i.v . ,
klochkova , v.g . ,
panchuk v.e . , 1998 , these proceedings galazutdinov , g.a .
: 1992 , _ prepr .
sao ras _ , * no.92 * , 1 grevesse , n. , noels , a. : 1993 , _ origin and evolution of the elements _ , eds . n. prantzos , e. vangioni - flam and m. cass , _
cambridge univ .
press _ , 14 kurucz , r.l . : 1991 , _ precision photometry : astrophysics of the galaxy _ , + eds . a.g.d .
philip , a.r .
upgren and k.a .
l. davis press _ , 1 kurucz , r.l . : 1992 , _ the stellar populations of galaxies _ , eds .
b. barbuy , a. renzini , _
iau symp .
149 _ , 225 kurucz , r.l . ,
furenlid , i. , brault , i. and testerman l. : 1984 , _ the solar flux atlas from 296 nm to 1300 nm _ , national solar observatory panchuk , v.e . ,
klochkova , v.g . ,
galazutdinov , g.a . ,
ryadchenko , v.p . and chentsov , e.l .
: 1993 , _ sov .
_ , * 19 * , 11 paunzen , e. , weiss , w.w . , heiter , u. and north , p. : 1997
, _ astron .
_ , * 123 * , 93 strenburg s. , 1993 , a&a 277 , 139 tsymbal , v. : 1996 , _ model atmospheres and spectrum synthesis _ , eds . s.j .
adelman , f. kupka and w.w .
weiss , _ asp conf .
_ , * 108 * , 198 | seven high - resolution and high s / n ccd spectra were used to derive elemental abundances in the atmosphere of vw ari a ( t@xmath0=7200 , @xmath1=3.7 ) which is the primary component of a visual binary system . the synthetic spectrum technique applied in the analysis allowed to reveal the following feature : the atmosphere of this star is strongly deficient in some metals , while light elements have solar - like abundances .
taking into account these results , one can suggest that vw ari a is a @xmath2 boo - type star . another argument supporting
this supposition is the following : on the diagrams `` @xmath3 '' , `` @xmath4 '' and `` @xmath5 '' ( paunzen et al .
1997 ) vw ari a falls exactly in the region occupied by the @xmath2 boo stars .
note also , that a previous analysis ( andrievsky et al .
1995 ) has shown that the secondary component of vw ari has a normal metallicity .
differences in chemical compositions of the two components appear to be due to the specific evolution of the primary vw ari a. |
the interstellar medium ( ism ) of a typical disk galaxy is divided into variety of distinct phases , generally classified into hot ionized ( @xmath4 ) gas , the warm neutral medium ( wnm ; @xmath5 ) , and the cold neutral medium ( cnm ; @xmath6 ) .
star formation , however , appears to be restricted to cold dense gas with a star formation rate ( sfr ) that strongly correlates with the molecular gas content @xcite .
this link between star formation and molecular gas is probably not causal , in the sense that the presence of molecules simply marks those parts of the ism that are cold and dense enough to undergo gravitational collapse @xcite , but the molecular gas nonetheless represents an indispensable tracer of star forming regions in the local and distant universe .
understanding these crucial links between star formation , the ism , and galaxy evolution , all intricately coupled through a variety of energetic stellar feedback mechanisms , must rely on identifying the pathways and conditions under which cold , molecular gas is able to develop
. furthermore , the gas temperature sets the characteristic size ( @xmath7 ) and mass ( @xmath8 ) of prestellar cores that develop in post - shock regions within magnetized , star - forming clouds ( e.g. , * ? ? ?
* ; * ? ? ?
since gas cooled by rotational transitions of co is able to reach lower temperature than that cooled by fine - structure lines of atomic carbon , tracking molecular chemistry is crucial to accurately represent small - scale fragmentation in numerical simulations of star - forming clouds .
young , massive stars are the main source of far - ultraviolet ( fuv ) photons that permeate the ism and readily photodissociate interstellar molecules , such as molecular hydrogen ( @xmath0 ) and carbon monoxide ( co ) .
molecular gas is thus found predominantly in dense , cold regions where dust shielding and self - shielding by the molecules themselves have attenuated the interstellar radiation field ( isrf ) intensity far below its mean .
the transition zones separating atomic and molecular gas , so called photodissociation regions ( pdrs ) , have been studied extensively both numerically and theoretically ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
typically , the outer , unshielded layers of a pdr are composed of atomic hydrogen and singly ionized carbon .
as column density increases the atomic - to - molecular transition begins to occur , with the c - co transition occurring interior to the @xmath9-@xmath0 one .
deeper still , when external fuv photons have been entirely absorbed , the chemistry and thermodynamics becomes dominated by the presence of deep penetrating cosmic rays and the freeze out of molecules onto dust grains .
simply understanding the structure of a non - dynamic , one - dimensional pdr is of limited utility .
the ism is a dynamic environment where a variety of non - linear , coupled processes determine the global distribution and state of gas .
star forming complexes are continually assembled through self - gravity , while stellar feedback acts disperse the structures .
turbulent motions , ubiquitous in the ism , dissipate through radiative shocks , but are continually replenished through a combination of supernovae and stellar winds .
thermal processes , such as dust grain photoelectric heating and molecular line cooling , further influence the state of the ism and are sensitive functions of ambient radiation fields , column density , and the non - equilibrium chemical state of the gas .
implementing these processes in a numerical simulation is highly non - trivial , a challenge further exacerbated by the enormous range of spatial scales involved ; the star forming region of a galactic disk can extend radially for tens of kiloparsecs while individual star forming disks are typically hundreds of aus and smaller , making this a computational _ tour de force _ for even adaptively refined grids .
nevertheless , three - dimensional simulations of finite , but representative , portions of galactic disks have attempted to replicate a supernova - driven ism , finding success in reproducing a multiphase ism @xcite , or demonstrating the regulation of the sfr by stellar feedback @xcite .
studies that have examined the atomic - to - molecular transition within large - scale , three - dimensional simulations has received far less attention . @xcite and
@xcite simulated the conversion of the atomic ism to molecular form in isolated periodic boxes , without explicit feedback but including driven or decaying turbulence . @xcite and @xcite included non - equilibrium @xmath0 formation in simulations of galactic discs , but neglected self - gravity and supernova feedback .
these studies did include an @xmath0-dissociating fuv photo - background , though assumed it to be constant in both space and time .
@xcite recently conducted a series of galactic scale simulations including supernova feedback , non - equilibrium chemistry , and radiative shielding based on the treecol algorithm @xcite . here
, the strength of the fuv photodissociating background scales linearly with the instantaneous sfr , though is spatially uniform .
@xcite finds the amount and distribution of molecular gas depends sensitively on a number of parameters , in particular the precise rhythm and spatial distribution of supernovae .
while these simulations have provided insight into the relationship between dynamics and chemistry , the price of simulating 3d time - dependent chemistry is that their treatment of photodissociation , the crucial process for regulating the chemical state of the ism , is extremely primitive .
the true photodissociation and photoheating rate at any point depends on the flux of fuv radiation integrated over all solid angles and from sources at all distances .
in contrast , most of the chemodynamical simulations conducted to date have relied on _ ad hoc _ prescriptions for the dust- and self - shielding - attenuated photodissociation rates in which a single , uniform , permeating radiation field is assumed to exist in all unshielded regions , and the degree of attenuation at a point is characterized by a single characteristic column density .
these prescriptions are for the most part untested , and are of unknown accuracy . here
our goal is to perform a detailed study on the chemical state of gas in a self - consistently simulated ism , placing special emphasis on accurately computing the molecular content with an rigorous treatment of the generation and attenuation of photodissociating fuv radiation .
we do this by post - processing the supernovae driven galactic disk simulations of @xcite with multifrequency ray tracing and a physically motivated spatial distribution of stellar sources .
this approach has the price that it discards time - dependent dynamical effects .
however , it does not rely on untested approximations for the radiative transfer problem .
it therefore provides a useful complement to the more approximate dynamical simulations that have appeared in the literature thus far .
in particular , our approach enables tests to evaluate the accuracy of simplified shielding schemes that have been previously used . in this paper ,
our main goal is to compare a number of commonly used local approximations for the degree of radiative shielding to full solutions of the radiative transfer equation , in order to gauge the effect that different approaches to radiative shielding have on the molecular abundances and temperature . for our tests
, we use a density structure obtained from large - scale ism simulations with turbulence self - consistently driven by star formation feedback , which naturally includes vertical stratification and strong density contrasts between warm and cold gas phases .
we are also able to test how the distribution of molecular complexes is sensitive to parameters such as the stellar distribution and overall fuv intensity , and the relative importance of dust- and self - shielding .
our comparison of different shielding prescriptions will advise us as to the validity and best use of local shielding approximations in upcoming multidimensional simulations of star formation that we shall perform .
we organize this paper as follows . in section [ sec : method ] we describe our methodology , including a description of the simulation we apply our post - processing to , the details of our chemical network , and the numerical approach to solving the equation of radiative transfer . in section [ sec : results ] we display our results .
finally , we discuss our results and conclude in section [ sec : discussion ] .
in this paper we apply radiative and chemical post - processing to the large - scale galactic disk simulations of @xcite to compute the equilibrium chemical and thermal state of the disk given a realistic distribution of stellar sources , a self - consistently derived morphological structure , and multi - frequency , long - characteristics radiative transfer .
our methodology broadly consists of two parts .
first , we perform frequency dependent radiative transfer through the simulation volume , which we describe in detail in section [ sec : raytrace ] , supplying us with the shielding attenuated chemical photorates . next
, these photorates are passed to a chemical network , described in section [ sec : chemistry ] , which is then run to equilibrium , on a cell - by - cell basis , to produce a three - dimensional datacube of chemical abundances and , in a subset of our models , gas temperature ( section [ sec : thermal ] ) . in this procedure , the ray trace and chemical network integration are intricately linked via the photodissociation rates .
the chemical abundances have an effect on the global photodissociation rates , which in turn determine the chemical abundances . given the non - local coupling between these two steps , chemical network integration and ray tracing must be carried out iteratively until convergence is obtained in the chemical abundances .
the simulations of @xcite ( hereafter k15 ) were run with the athena code @xcite which solves the equations of ideal magnetohydrodynamics on a uniform grid . the models also include self- and external - gravity , heating and cooling , thermal conductivity , coriolis forces , and tidal gravity in the shearing box approximation to model the effect of differential galactic rotation . while k15 ran a suite of simulations with a range of magnetic field strengths and configurations , here we focus exclusively on their mb10 ( solar neighborhood analog ) model that has the following initial properties : gas surface density of @xmath10 , midplane density of stars plus dark matter @xmath11 , galactic rotation angular speed @xmath12 , uniform azimuthal magnetic field of @xmath13 , and box size @xmath14 , @xmath15 , where the @xmath16 direction is perpendicular to the disk .
the computational box had a uniform resolution of @xmath17 in each direction such that the number of grid cells along each direction were @xmath18 and @xmath19 .
gas cooling , dominated by ly@xmath20 emission at high temperatures and @xmath21 fine structure emission at lower temperatures , is included by use of the fitting formula of @xcite , @xmath22 where @xmath23 is the gas temperature in kelvin . treating radiation from young , massive stars to be the main source of heating via the dust grain photoelectric effect ,
the gas heating rate is set to be proportional to the recent , global sfr surface density @xmath24 , @xmath25\,{\rm erg}\ , { \rm s}^{-1}\ , , \label{eq : heating}\ ] ] where @xmath26 is the solar neighborhood heating rate and @xmath27 is the sfr surface density in the solar neighborhood . the term @xmath28 accounts for a metagalactic fuv radiation field .
radiative shielding of fuv photons , an important effect at high gas column densities , is not included . supernova feedback is included in the form of instantaneous momentum injection at a rate proportional to the local star formation rate , as set by the free - fall time of gas with an efficiency of @xmath29 above a critical density threshold . the momentum from each supernova is set to @xmath30 representing the value at the radiative stage of a single supernova remnant ( see @xcite and references therein ) .
this approach does not form or follow the hot ( @xmath31 ) ism phase , but instead focuses on the self - consistent turbulent driving , dissipation , and gravitational collapse of the atomic medium , the main gas reservoir of the ism .
the absence of a hot phase will have only a minimal effect on our results , since we are mainly concerned with molecule formation processes that are restricted to much lower temperatures .
we utilize the chemical network described in @xcite which we briefly summarize here .
the full chemical network consists of 32 atomic and molecular species and @xmath32 chemical and photoreactions .
the abundances of only 14 of these species ( h@xmath33 , @xmath0 , he@xmath33 , c@xmath33 , o@xmath33 , oh , h@xmath34o , co , c@xmath34 , o@xmath34 , hco@xmath33 , ch , ch@xmath34 , and ch@xmath35 ) are computed by formal integration , i.e. , via solving a stiff set of coupled ordinary differential equations .
the abundances of the remaining species are computed by either assuming instantaneous chemical equilibrium ( as in the case of h@xmath36 , @xmath37 , h@xmath35 , ch@xmath33 , ch@xmath38 , oh@xmath33 , h@xmath34o@xmath33 , and h@xmath39o@xmath33 , species that react so rapidly as to always be close to their equilibrium abundance values ) or by utilizing conservation laws ( for e@xmath36 , h , he , c , and o ) , thus reducing the number of coupled differential equations , and computational cost , that must be integrated to solve the full chemical network . to model photodissociation and photoionization , we take the shape and strength of the interstellar radiation field ( isrf ) to follow the standard @xcite field . in habing units , the strength of the draine field is @xmath40 , where the @xcite radiation field is defined to be @xmath41 where @xmath42 is the angle averaged specific intensity and the integral runs over the fuv range , from @xmath43 .
the photodissociation and photoionization rate for any given chemical species can be written as @xmath44 where @xmath45 is the photorate in optically thin gas ( see table b2 in @xcite ) , and @xmath46 is the degree by which the photorate is reduced as compared with the optically thin rate .
we defer a discussion of how the shielding factors @xmath46 and unattenuated radiative intensity @xmath47 are determined to section [ sec : raytrace ] . in a subset of our models
we evolve the temperature to equilibrium along with the chemical abundances . in this section
we describe these heating and cooling processes .
in the low temperature , moderate density , molecular ism , line emission originating from the rotational transitions of co is a major source of radiative cooling .
the volumetric cooling rate from co can be written as @xmath48 .
in general , the co cooling function @xmath49 is a complicated function of temperature , density , column density , and velocity dispersion , and in a multidimensional simulation it is not feasible to self - consistently solve for the level populations in every computational cell .
hence , we employ the despotic code @xcite to pre - compute the co cooling function . utilizing the large velocity gradient approximation
, we tabulate @xmath49 as a function of @xmath50 , temperature , and , following @xcite , an effective column density per velocity @xmath51 defined such that @xmath52 where @xmath53 is the co number density . the local velocity divergence , @xmath54 ,
is computed using a three - point stencil around the cell of interest . during the computation we read in the despotic - computed cooling tables and employ tri - linear interpolation to compute the co cooling rate as needed .
in addition to co line cooling , we also consider fine - structure line emission from @xmath55 $ ] , @xmath56 $ ] , @xmath57 $ ] .
in the case of these species , because their lines remain optically thin under all the conditions found in our calculations , it is straightforward to solve for the level populations and compute the cooling rate directly .
we use the collisional rate coefficients , atomic data , and methodology presented in @xcite .
the rate of energy exchange between gas and dust is @xmath58 where @xmath59 is the dust - gas coupling coefficient for mliky way dust in @xmath0 dominated regions @xcite and @xmath60 is the dust temperature . in principle
@xmath60 can be determined self - consistently by considering dust to be in thermal equilibrium between absorption and emission , but for simplicity here we assume a constant dust temperature of @xmath61 .
this should be an acceptable approximation since gas - dust coupling does not typically become important until @xmath62 , a regime not probed here .
cooling due to the collisional excitation of atomic hydrogen , or lyman-@xmath63 , is given by @xcite , @xmath64 where @xmath65 , and @xmath66 and @xmath67 are the electron and neutral hydrogen densities , respectively .
the photoelectric heating rate , including recombination cooling , is given by @xcite , @xmath68 where @xmath69 is the attenuated radiation field strength described in section [ sec : raytrace ] , @xmath70 , @xmath71 , and @xmath72 is the photoelectric heating efficiency .
the heating rate due to cosmic ray ionization is given by @xmath73 where @xmath74 is the cosmic ray ionization rate per hydrogen nucleus which we take to be @xmath75 , a relatively low value that aids in the production of co. the energy added per cosmic ray ionization @xmath76 depends on the chemical composition of the gas . for purely atomic gas
we use the recommendation from @xcite , @xmath77 while we use a piecewise fit given in ( * ? ? ? * equation b3 ) of numerical data from @xcite for @xmath78 . note that we do not include any sort of heating due to turbulent dissipation or ambipolar diffusion , two related effects whose combined heating rate can be comparable to the cosmic ray heating rate for the low cosmic ray ionization rate we adopt here ( e.g. , * ? ? ?
modeling radiation transport in full generality within a three - dimensional simulation presents a significant numerical challenge .
as noted by @xcite , the cost of multifrequency radiation transfer scales as @xmath79 , where @xmath80 is the number of discrete frequency bins and @xmath81 is the number of computational elements . even with moderate resolution , this is orders - of - magnitude larger than the hydrodynamic evolution , which scales as @xmath82 , and generally intractable . while a number of approximate methods exist for solving the equation of radiative transfer in a numerical simulation ( e.g. ,
* ; * ? ? ?
* ; * ? ? ?
* ) , these approaches are not suitable for line transfer calculations , which are crucial for determination of the photoreaction rates . here , we elect to solve for the radiation field , and chemical state , by direct ray - tracing applied to static simulation snapshots .
-axis with a separation equal to the cell spacing @xmath83 . the next set of rays ( green arrows ) are launched with an angle of @xmath84 with respect to the @xmath85-axis and maintain a perpendicular separation @xmath83 with the first set of rays .
this choice guarantees the computational domain is evenly sampled by rays for each angular set of rays . only when the rays are physically located within the computational grid ( solid boxes ) is the ray tracing performed.,scaledwidth=45.0% ] the radiation intensity at any point along a ray is governed by the time - independent radiative transfer equation , @xmath86 where @xmath87 is the specific intensity , @xmath88 is the absorption coefficient , @xmath89 is the emissivity , and @xmath90 is the distance along the ray .
we are particularly concerned with fuv photons whose dominant sources are stars with ages of @xmath91 myr or less . since the simulation does not follow young stars explicitly , and since by this age stars are generally no longer predominantly found in clusters ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , we approximate their distribution as a smooth field that varies with height @xmath16 alone .
we therefore assign every grid cell a frequency - dependent fuv emissivity @xmath92 , where the normalization @xmath93 is fixed by the requirement that the unattenuated midplane isrf intensity , @xmath94 , is in accordance with the current sfr surface density : @xmath95 where @xmath24 and @xmath96 retain their definitions from equation [ eq : heating ] . for the vertical distribution of the fuv emissivity
we adopt the profile @xmath97 appropriate for an isothermal disc , where @xmath98 is the stellar scale height . for our fiducial model
we adopt a scale height of @xmath99 , though we do explore how variations in @xmath98 affect our results . even in a static snapshot ,
solving the transfer equation in three dimensions with enough frequency resolution to resolve the dominant photodissociation lines of every important species is prohibitively expensive .
fortunately , it is also unnecessary for the purpose of computing photoreaction rates , which is our goal .
photodissociation for important interstellar species occurs via resonant absorption of an fuv photon in one or more bands that occupy a relatively narrow range of frequencies .
one - dimensional simulations that include full frequency - dependent transfer show that the photodissociation rate produced by the isrf decreases as the radiation propagates through increasing columns of material due to two main effects : continuum absorption by dust , and resonant absorption of photons near the centers of the dissociating absorption lines @xcite . for dust ,
the reduction in the photodissociation rate is given by @xmath100 where @xmath101 is the scaling between the dust opacity at the photodissociation absorption frequency and the v - band . here
@xmath102 is the dust opacity evaluated at the absorption band relevant for that species , which is taken to be at a single characteristic frequency , a good approximation since @xmath102 varies slowly with @xmath103 over the band ; @xmath104 is the v - band dust opacity .
for the chemical species we consider , @xmath105 ranges from @xmath106 to @xmath107 .
the visual extinction in magnitudes @xmath108 is directly proportional to the total column density of h nuclei , @xmath109 , @xmath110 where @xmath111 @xcite is the total dust cross section in the v - band per hydrogen nucleus . to properly model the photodissociation of @xmath0 and co we must also consider self - shielding in which the molecules themselves , in addition to dust grains , contribute to the attenuation of fuv photons .
self - shielding of @xmath0 can be well approximated with the following analytic expression from @xcite , @xmath112 where @xmath113 , @xmath114 , and @xmath115 is the doppler broadening parameter for @xmath0 .
the total shielding factor for @xmath0 will be @xmath116 .
co not only experiences self - shielding and dust shielding , but is additionally shielded by molecular hydrogen .
the photodissociation rate for co can be written as @xmath117 , where @xmath118 is the co self - shielding function and @xmath119 accounts for the cross - shielding of co by the overlapping lyman - werner lines of @xmath0 .
both @xmath120 and @xmath121 are taken from @xcite , while @xmath122 .
a critical point to realize is that , because the co and h@xmath34 self - shielding factors result from a rearrangement of the photon frequency distribution as radiation propagates along a ray , they may be applied independently on every ray .
this means that , rather than requiring many frequency bins to resolve the dissociating lines of h@xmath34 and co , we can approximate the effect by considering a single frequency bin for h@xmath34-dissociating photons , and similarly for co - dissociating ones , and use the pre - tabulated shielding functions to model the reduction in photodissociation rate along a ray .
the main limitation in this approach is that the shielding functions have been tabulated for gas with a fixed velocity dispersion and zero bulk velocity , whereas in our simulations the velocity fields are significantly more complex .
however , we show below that our results are quite insensitive to exactly how we estimate the velocity - dispersion that enters the shielding factors . based on the discussion in the preceding section , we now discretize the equation of radiative transfer ( equation [ eq : rad_transfer ] ) by defining three distinct frequency bins : one describing radiation in the fuv dust continuum , @xmath123 , one that corresponds to the lyman - werner bands for @xmath0 , @xmath124 , and one describing the photodissociation of co , @xmath125 .
the three frequency bins can be formally defined through the following filters : @xmath126 defined such that @xmath127 and @xmath128 equal unity in the frequency ranges that overlap with their photodissociation absorption lines , and are zero otherwise .
likewise , @xmath129 in the fuv frequency range , excluding the photoabsorption lines of @xmath0 and co , and is zero otherwise .
the intensity at any point @xmath90 along a ray is given by the formal solution to equation ( [ eq : rad_transfer ] ) , @xmath130 identifying @xmath131 to be the shielding factor @xmath46 ( equation [ eq : photo_shield ] ) and integrating over the frequency filters defined in equation ( [ eq : filters ] ) gives us the intensity in our three frequency bins : @xmath132 @xmath133 @xmath134 in the above equations , the shielding factors are evaluated using the total column densities between points @xmath90 and @xmath135 , @xmath136 , @xmath137 , and similarly for @xmath138 and @xmath139 we are now in a position to discretize the equations ( [ eq : ih2 ] ) , ( [ eq : ico ] ) , and ( [ eq : icont ] ) in a way suitable for numerical calculation . consider a ray passing through our cartesian grid at some angle , and consider the @xmath140th cell along that ray .
let the sequence of cells through which the ray passes on its way to that cell @xmath140 be numbered @xmath141 , and let @xmath142 be the path length of the ray through cell @xmath143 ; let @xmath144 be the corresponding number density of species @xmath145 ( e.g. , h@xmath34 ) , and similarly for all other quantities . in this case equation ( [ eq : ih2 ] )
can be discretized to @xmath146 where the shielding factor , which includes both dust- and self - shielding , is evaluated using the total column @xmath147 between the cells @xmath140 and @xmath143 .
similar results hold for equations ( [ eq : ico ] ) and ( [ eq : icont ] ) .
we have now written down a discretized version of the radiative transfer equation for a particular direction of propagation .
the remaining step is to discretize the problem in angle .
we do so by drawing rays in @xmath148 directions ( parametrized by @xmath84 and @xmath149 ) selected via the healpix algorithm for equal area spherical discretization @xcite .
all our models use healpix level @xmath150 , corresponding @xmath151 , which we have found provides sufficient resolution to properly sample the radiation field . for each direction
, we draw through the computational domain a series of rays separated by one cell spacing @xmath83 in the direction perpendicular to the rays , such that each ray represents a solid angle @xmath152 and area @xmath153 .
this choice guarantees that each cell is intersected by at least @xmath148 rays in total .
we show a schematic diagram of our ray tracing procedure in figure [ fig : gridfig ] .
the radiation intensity in any given cell is then able to be computed through an angle average of all intersecting rays .
we are ultimately interested in photodissociation rates which are in turn proportional to the angle - averaged intensity , @xmath154 and similarly for all other species . evaluating this
integral requires some subtlety ; although all our rays represent an equal amount of solid angle , not all rays intersecting a cell have the same path length @xmath155 through it , even rays that represent the same direction of radiation propagation .
the contribution to the mean intensity in a cell from a given ray @xmath140 will be proportional to the time that the photons propagating along that ray spend in the cell , and thus proportional to @xmath156 .
we therefore compute the mean intensity by weighting each ray s contribution by @xmath155 : @xmath157 where @xmath158 is the total number of rays that intersected the cell .
the conversion between @xmath159 and @xmath160 is implicitly performed through a proper choice of the normalization of the fuv cell emissivity , @xmath161 .
we are interested in comparing the results obtained with the full angle - dependent ray - trace in the previous section to those produced by various local approximations for the degree of radiative shielding that have been proposed . in these approximations ,
the factor by which a chemical species photodissociation rate is reduced relative to its optically thin rate , @xmath46 , is solely dependent on an effective column density : @xmath162 in the case of dust shielding , and @xmath163 in the case of @xmath0 and co shielding . working under the approximation of uniform density and chemical composition , we can rewrite column density as the product of number density and some shielding length scale , @xmath164 .
this significantly reduces the complexity and computational expense of multidimensional , multifrequency radiative transfer to the relatively simple task of computing an appropriate , physically motivated shielding length @xmath165 .
various physically motivated expressions for @xmath165 have been proposed and utilized , to varying degrees of success .
the sobolev approximation , alternatively known as the large velocity gradient ( lvg ) approach , was originally devised by @xcite to study expanding stellar envelopes , but has been used extensively in simulations to model the photodissociation of @xmath0 ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
it relies on the observation that in regions with a constant velocity gradient @xmath166 , a photon will see doppler shifted absorption lines with respect to its emission frame .
@xcite proposed that once a photon is doppler shifted by one thermal linewidth , the photon should be free to escape .
thus , we can define the sobolev length to be @xmath167 where @xmath168 is the local sound speed and the local velocity divergence @xmath54 serves as a multidimensional analog to @xmath166 .
the jeans length @xmath169 is the critical length scale under which the force of gravity overwhelms outward thermal pressure allowing gravitational collapse to proceed .
the jeans length is a common approximation for @xmath165 , with the physical justification that at the center of a gravitationally collapsing core , the radiative background will be attenuated by a column of gas with a length scale approximately equal to @xmath170 .
we also investigate a shielding length based on the local density and its gradient such that @xmath171 the density gradient approach has been used by @xcite to account for the self - shielding of @xmath0 from photodissociating radiation and provided reasonable accuracy in the column density range @xmath172 .
an additional method to account for radiative shielding is to utilize the results of a full ray trace run to calibrate a relationship between density and column density ( or equivalently , visual extinction ) . we we will show in section [ sec : results ] , this relationship can be approximated as @xmath173 which holds above a density of @xmath174 and becomes increasingly accurate with increasing density .
this method naturally has the disadvantage that it is perhaps unique to the particular physical system one is examining which detracts form its generality .
a highly simplified approach to account for radiative shielding is to assume that all shielding in a given cell is due to matter in the cell in question , such that @xmath175 pc .
this single - cell approximation will surely underestimate the degree of radiative shielding , and thus molecular chemical abundances , as compared with the fiducial ray - tracing model , allowing limits to be placed on the effectiveness of radiative shielding . finally ,
while not a strictly local approach , we perform a model where ray tracing is only performed over the six directions aligned with the cartesian axes , instead of the @xmath176 healpix selected directions in the ray trace models .
this approach , known as the six - ray approximation , can be computationally efficient since the transport of rays exploits the alignment of the grid .
it is a compromise between a local approach and full - fledged , multi - angle radiative transfer . for this to be a clean comparison ,
the local and ray - trace models need to utilize the same unattenuated radiation field @xmath47 so that the sole difference between the various methodologies is in the computation of @xmath46 . to achieve this , in the local models we begin by performing the multiangle ray - trace procedure described in section [ sec : raytrace ] , but with all radiative shielding turned off , @xmath177 , providing us with the unattenuated radiation field @xmath47 in each grid cell . this unattenuated field
is then used in equation [ eq : photo_shield ] , along with the locally determined @xmath46 , to compute the chemical photodissociation rates . ,
top - right ) , co fraction versus density ( bottom left ) , and co fraction versus @xmath178 ( bottom - right ) .
color corresponds to the mass - fraction that particular region of phase space . for comparison ,
the overplotted black lines are approximate analytic expressions for the @xmath0 fraction @xcite , mass - weighted in density ( left panel ) or @xmath179 ( right panel ) bins .
the shaded region in the top - right panel denotes the analytic prediction from ( * ? ? ?
* equation 40 ) for the maximum visual extinction before the h@xmath180-@xmath0 transition , for densities from @xmath181 to @xmath182.,scaledwidth=48.0% ] [ cols="<,^,^,^,^,^,^,^,^ " , ] _ notes _ numbers in parenthesis represent the scientific notation exponent , i.e. , 0.19(-3 ) @xmath183 .
( 1 ) the total gas mass fraction with an @xmath0 ( first column ) or co ( second column ) fraction greater than @xmath106 .
the second number , after the slash , is the equivalent quantity when only considering cold , dense gas ( @xmath184 , @xmath185 ) .
( 2 ) the total mass in the fiducial simulation ( f1 ) is @xmath186 .
( 3 ) see text surrounding equation [ eq : err ] for the description of the @xmath0 and co error measurements . to model the spatial distribution of fuv emissivity ( as required by equation [ eq : soln ] )
we treat massive ob stars as the primary emitters of fuv radiation whose vertical density obeys the functional form of equation [ eq : fz ] .
the fiducial model f1 utilized a vertical scale height of @xmath187pc , roughly consistent with the measured mean half - width of @xmath0 in the inner galaxy of @xmath188pc @xcite .
models v1 and v2 , which use scale heights of @xmath189 and @xmath190pc , respectively , demonstrate the relatively insensitivity of our results to @xmath98 .
inspecting table [ tab : results ] , model v1 has @xmath191 more @xmath0 and @xmath192 more co than f1 , reasonable considering that a reduction ( increase ) in @xmath98 lowers ( increases ) the overall global number of fuv photodissociating photons .
model v2 has a corresponding decrease in amount of @xmath0 , though less than a @xmath29 reduction in the total co mass as compared with f1 .
morphologically , models v1 and v2 are virtually indistinguishable from f1 .
models v3 through v6 , in which one or a number of shielding mechanisms are artificially switched off , are designed to assess the relative importance of self- and dust - shielding in determining the molecular abundances .
unsurprisingly , each of these models results in overall less @xmath0 and co mass than the comparison model f1 .
models v3 , v4 , v5 , v6 have , respectively , @xmath193 , @xmath193 , @xmath194 , and @xmath195% the mass of @xmath0 as model f1 . as for the co mass ,
these percentages become @xmath196 , @xmath197 , @xmath198 , and @xmath199% ( table [ tab : results ] ) .
figures [ fig : proj.shielding.h2 ] and [ fig : proj.shielding.co ] show mass - weighted projections of the @xmath0 and co fractions for models f1 , v3 , v4 , v5 , and v6 .
a comparison of models f1 with v3 and v4 in figure [ fig : proj.shielding.h2 ] highlights the importance of @xmath0 self - shielding in attenuating the isrf and reducing the @xmath0 photodissociation rate .
the @xmath0 fractions in models v3 and v4 , which do not include @xmath0 self - shielding ( equation [ eq : fh2 ] ; @xmath200 ) , are substantially suppressed ( by nearly a factor of @xmath201 see table [ tab : results ] ) compared with models that do include self - shielding . in these models ,
@xmath0 fractions of unity are only realized in the highest density gas that approaches the resolution of this study .
switching off co self- and cross - shielding , model v5 , unsurprisingly has no effect on the @xmath0 abundance . in model v6 dust
shielding is deactivated in an effort to isolate its role in reducing the direct photodissociation rates of @xmath0 and co. importantly , dust shielding remains active for all other species besides @xmath0 and co. this choice removes the indirect effect that increased photodissociation rates for other chemical species would have in altering the chemical formation pathways of co and , to a much lesser extent , @xmath0 . as shown in the far right panel of figure [ fig : proj.shielding.h2 ] , the removal of dust shielding has a fairly small , though non - zero , effect on the @xmath0 fraction .
in contrast to @xmath0 , self - shielding , cross - shielding by @xmath0 , though whether this stems from @xmath0 cross - shielding or reduced co formation rates is a subtle issue we explore later in this section ] , and dust shielding are all comparably important in photoshielding co molecules .
this can be seen in figure [ fig : proj.shielding.co ] where each model results in a significant decrease in the amount of co as compared with model f1 .
model v4 , in which @xmath0 self - shielding is switched - off and the global @xmath0 fractions are significantly reduced , demonstrates the importance of @xmath0 shielding co via their overlapping photodissociation lines in the lyman - werner bands .
models v5 and v6 , meanwhile , respectively underscore the importance of self- and dust shielding . however , the inclusion of _ any _ form of radiative shielding , whether from self- , cross- , or dust - shielding , results in the formation of gas with @xmath202 at the highest densities , @xmath203 .
model v10 ( not shown ) , in which all radiative shielding is turned off , results in no gas with @xmath204 , a statement that additionally applies to @xmath0 .
we can explore this further by examining the cumulative distribution functions ( cdf ) of @xmath205 and @xmath206 for models v3-v6 and f1 , which we plot in figure [ fig : shielding.cdf.cd ] . here , we restrict this analysis to cold , dense gas as this makes for a more potent demonstration of the competing factors at work ; qualitatively identical conclusions would be drawn from the cdfs that included all gas .
focusing first on @xmath0 ( left panels ) , we see the removal of dust shielding only reduces the gas fraction with @xmath207 as compared with model f1 .
turning off @xmath0 self - shielding ( models v3 and v4 ) , conversely , suppresses the @xmath0 mass @xmath208 for gas with @xmath209 .
here it is clear that by and large the dominant mechanism by which @xmath0 is shielded from the fuv isrf is self - shielding , at least for the range of isrf properties probed in these simulations . in the analytic models of @xcite , the relative importance of self - shielding and dust shielding depends on the value of @xmath210 , equation [ eq : chi ] . at milky way values of @xmath210 the two processes
are about equally important , but as noted above , the simulations we analyze here have radiation field intensities that are @xmath211 times smaller than those found in the solar neighborhood , so the relative dominance of self - shielding is not surprising .
( left panels ) and co fraction ( right panels ) .
this analysis is restricted to cold , dense gas ( @xmath184 , @xmath185 ) .
lines of different color denote models with different treatments of dust- and self - shielding as described in the text.,scaledwidth=53.0% ] as for co , a great deal of information can be extracted from a comparison of the various cdfs , shown in the right panels of figure [ fig : shielding.cdf.cd ] .
model v4 does not include @xmath0 self - shielding ( @xmath200 ) and as a consequence has a highly suppressed @xmath0 fraction .
this choice encapsulates two combined effects , one chemical and one radiative , both of which act to decrease the amount of co. first , @xmath212 ( the cross - shielding factor ) is reduced as the global @xmath0 column densities are significantly reduced .
second , the two predominant co formation pathways ( one involving oh , and the other ch ) both rely on the availability of @xmath0 , thus reducing the overall speed of the chemical formation pathways that lead to co. these effects combined result in the co mass being suppressed by a factor of @xmath213 for @xmath214 compared with model f1 the suppression increases with increasing co fraction . from this model alone , however , is it unclear which of these two effects , chemical or radiative , is dominant .
model v5 does not include co self - shielding nor cross - shielding from @xmath0 . as can be seen , this has little effect in gas with a low co fraction ( @xmath215 ) .
evidently for this diffuse co gas , co self - shielding and @xmath0 cross - shielding is relatively unimportant . for co rich gas ( @xmath216 ) ,
the mass in co is suppressed by a factor of @xmath217 .
chemically , there is no suppression of the @xmath0 abundance in this model so any reduction in @xmath206 is entirely due to the lack of shielding from itself and @xmath0 . the drop in the co fraction between models v5 and v3 is a chemical , not radiative , effect and highlights the role of @xmath0 in the co formation pathways .
observe that the sole difference between models v5 and v3 is that the @xmath0 fraction in model v3 is highly suppressed due to the lack of @xmath0 self - shielding . with regards to co
, this isolates the effect that the @xmath0 abundance has on the efficiency of the chemical pathways that lead to co. quantitatively , this purely chemical effect reduces the total co mass by @xmath218% ( see table [ tab : results ] ) . furthermore , at the highest co fractions , there is no difference between model v5 and v3 in terms of the cdf , while model v4 still displays a roughly order - of - magnitude drop .
this implies that a reduction in the @xmath0 fraction ( via removing @xmath0 self - shielding ) decreases the co fraction predominantly through less @xmath0 cross - shielding and a larger co photodissociation rate , as opposed to a reduction in the co formation rate .
the co fractions realized in the absence of dust shielding ( model v6 ) are similar to that of models v3 and v4 for @xmath219 , underscoring the importance of dust shielding for co. in other words , the consequence of removing dust shielding for co is roughly comparable to the combined removal of co self - shielding , @xmath0 cross - shielding , and decreased co formation efficiency due to a suppressed @xmath0 abundance . evidently , dust shielding is critical for the formation of co fractions approaching unity . in the absence of dust
shielding , no gas is able to form with @xmath220 .
models v7 and v8 consider the scenario where the overall strength of the fuv radiative intensity is , respectively , increased or decreased by a factor of @xmath221 . increasing the isrf by a factor of @xmath221 ( v7 ) results in a @xmath222 decrease in both the total @xmath0 and co mass , while decreasing the radiative intensity tenfold ( v8 ) results in a factor of @xmath223 more @xmath0 and @xmath224 times more co mass . in model v9
, the cosmic - ray ionization rate is increased from @xmath225 to @xmath226s@xmath227 to gauge its impact on molecular chemistry .
this has a much more severe effect on the co abundance than the @xmath0 abundance .
as seen in table [ tab : results ] , increasing @xmath74 reduced the global @xmath0 mass by roughly a factor of @xmath150 , while the co mass fell to roughly @xmath228 its value in model f1 .
this strong dependance of the co abundance on the cosmic ray ionization rate is an indirect effect , stemming mainly from the cosmic ray ionization of neutral helium and the subsequent charge - exchange destructive reaction between he@xmath33 and co. finally , in model v10 , all radiative shielding is switched off , causing a catastrophic decrease in the abundances of @xmath0 and co. as can be seen in table [ tab : results ] , this results in zero gas with an @xmath0 or co fraction higher than @xmath106 .
the total molecular mass is also significantly reduced ; the total mass of @xmath0 and co is @xmath229 and @xmath230 , respectively , as compared to model f1 .
this result only underscores the importance of radiative shielding in permitting the formation of significant molecular abundances .
however , as will be shown in section [ sec : temperature ] , the removal of all radiative shielding has a relatively small effect on the temperature structure of the gas . is entirely due to the vertical dependance of the unattenuated radiation field .
the primary effect of changing the fuv intensity ( models t3 and t4 ) is a shift in the density at which gas transitions from the warm to cold neutral medium .
the blue shaded region in the top panel denotes this paper s definition of cold , dense gas.,scaledwidth=50.0% ] in models t1-t4 , the gas temperature is evolved to equilibrium along with the chemical abundances , utilizing the thermal processes described in section [ sec : thermal ] .
model t1 is identical to model f1 save for the temperature evolution , while model t2 operates under optically thin conditions , i.e. , no radiative shielding . in models t3 and t4 the mean fuv intensity is altered , though are otherwise identical to model t1 . in figure
[ fig : dts.compare ] we show density - temperature diagrams for models f1 ( whose density and temperature are unchanged from the values initially supplied by k15 ) , t1 , t2 , t3 , and t4 . at any given density ,
the spread in temperature for model t1 is significantly reduced as compared with model f1 .
in fact , any spread in temperature in model t1 , at a given density , is overwhelmingly due to spatial variations in the photoelectric heating rate ( equation [ eq : photoelectric ] ) which arises due to differences in the degree of radiative shielding ( via @xmath231 ) and the vertical dependance of the fuv emissivity ( equation [ eq : fz ] ) and consequently the unattenuated radiation field @xmath47 . aside from this reduction in the temperature spread , the density - temperature relationship between models f1 and t1 agree remarkably well below a density of @xmath232 . above @xmath233 ,
the temperature in model t1 begins to drop below that of f1 , reaching @xmath234 by @xmath235 as a result of co line emission , a cooling pathway not included in k15 .
the relationship between density and temperature in model t2 , which does not include the effect of radiative shielding , is similar to that of model t1 below @xmath236 .
the only significant difference between the two models is the absence of @xmath237 degree gas at the highest densities , due to the complete lack of any significant co cooling , in model t2 . even by increasing the midplane @xmath47 by a factor of @xmath221 to @xmath238 ( model t3 )
, a cold phase still develops , though at a slightly larger density and equilibrium temperature of @xmath239 before co cooling cools the gas below @xmath240 .
model t4 , with its tenfold reduction in the strength of the midplane fuv intensity , exhibits the emergence of a molecular cold phase at the relatively low density of @xmath241 .
the molecular nature of this cold phase should be viewed as an artifact of our equilibrium assumption , given the chemical time at @xmath242 is of order @xmath221 gyr see section [ sec : discussion ] .
the minimum temperature obtained in model t4 is also lower than any other model , nearly reaching the cosmic microwave background ( cmb ) temperature floor ( @xmath243 ) around @xmath244 .
while radiative shielding is crucial for the formation of significant molecule fractions ( section [ sec : variations ] ) and for reaching the lowest temperatures observed in the ism , the ubiquitous appearance of a marginally cold phase ( where @xmath6 ) in all these models supports the important point that shielding is not strictly required for the transition to a cold phase , at least in the regime where the background fuv radiation field strength is relatively modest , @xmath245 .
the primary objective in this section is to assess the validity and accuracy of utilizing a locally computed shielding length to model radiative attenuation and thus the fuv - radiation regulated chemical state .
we will do this by examining how closely the local models l1-l6 reproduce the @xmath0 and co abundances of the fiducial model f1 using a variety of qualitative and quantitative measures .
considering first the total mass of @xmath0 ( @xmath246 , table [ tab : results ] column @xmath247 ) , we see all the local models come within a factor of @xmath150 of f1 , with the exception of model l2 ( sobolev approximation ) which results in @xmath248% less @xmath0 .
the total mass of co is a bit more scattered than @xmath246 , though the four best performing local models ( l1a , l3 , l4 , and l6 ) have @xmath249 within @xmath250% of model f1 s value .
any given local approximation model always produces an increase ( or decrease ) between both the total @xmath0 and co masses , with the exception of model l3 ( density gradient ) which , due to the peculiarity of using density gradients to estimate a global column density , results in @xmath251% less @xmath0 , and nearly @xmath252% _ more _ co , than model f1 .
the morphology and distribution of @xmath0 between models l1-l6 and f1 ( figure [ fig : proj.local_compare.h2 ] ) highlights the relative similarity between the distribution of molecular hydrogen in many of the local approximation models compared with the fiducial model .
evidently the exact prescription used to account for the attenuation of the isrf is a relatively unimportant factor in modeling the overall distribution of @xmath0 gas .
the inclusion of _ some _ form of radiative shielding , however , is a necessary ingredient to form any significant @xmath0 fraction not shown here is model v10 ( no shielding ) in which no cell of gas achieves an @xmath0 ( or co ) fraction greater than @xmath253 and results in a factor @xmath254 less @xmath0 and co mass .
it is important , however , to scrutinize any variation between the various local models and model f1 in figure [ fig : proj.local_compare.h2 ] .
models in which radiative shielding is based on the local jeans length ( l1 and l1a ) or a power - law fit between @xmath255 and @xmath178 ( l4 ) appear to slightly overestimate the @xmath0 fraction , while a shielding length computed via the velocity divergence ( sobolev length ; l2 ) , the density gradient ( l3 ) , or a single computational cell ( l5 ) , tend to underpredict @xmath205 . by eye , the best level of agreement
is achieved by the six - ray approximation ( model l6 ) ; this is not a strictly a local approximation , but is instead effectively the result of reducing the number of ray - tracing propagation directions from @xmath176 to @xmath247 , now all aligned along the cartesian axes .
in contrast to @xmath0 , the spatial distribution and amount of co ( figure [ fig : proj.local_compare.co ] ) shows a great deal more sensitivity to the radiative shielding prescription .
as compared with model f1 , models l2 and l5 fail to produce any gas with visually discernible co. in model l3 there exists what appears to be an unphysical scattering of cells with @xmath206 approaching unity and a paucity of diffuse co , @xmath256 . as with @xmath0 , the six - ray approximation ( model l6 ) appears to do the best job in reproducing @xmath206 . it should be noted , though , that large concentrations of co can be difficult to discern given the small spatial extent of these concentrations .
model l1 , where @xmath257 , seems to perform reasonably well in matching @xmath206 from model f1 , except for the appearance of two large ( @xmath258pc ) structures with @xmath259 that do not exist in model f1 .
we attribute this to out - of - equilibrium hot gas with density @xmath260 and temperature @xmath261 . since the jeans length @xmath262 , these cells have an unphysical elevated shielding length which reduces the fuv radiation intensity and boosts the co abundance .
this is an undesired effect of modeling @xmath165 with @xmath263 these cells do not represent gravitationally collapsing cores and physically should not experience a higher degree of radiative shielding .
this shortcoming can be addressed quite simply , however , by introducing a temperature ceiling into the calculation of @xmath263 . in model l1a the shielding length is given by this temperature capped jeans length with an empirically chosen ceiling temperature of @xmath264 .
an inspection of figure [ fig : proj.local_compare.co ] confirms that these unphysical co rich bubbles are strongly suppressed in model l1a . to better understand the relationship amongst the local models , in figure [ fig : local_compare.dens.av.phase ] we plot
the mass - weighted effective extinction ( equation [ eq : aveff ] ) as a function of density for each local model as compared with model f1 ( previously plotted in figure [ fig : dens_av_phase ] ) . here
it is clear how well models l1 , l1a , l4 , and l6 perform in matching the effective visual extinction as computed from detailed ray - tracing , at least at high densities , @xmath265 .
models based on local derivatives of grid - based quantities severely overpredict ( l3 ) or underpredict ( l2 ) the true column density .
caution should be used in interpreting figure [ fig : local_compare.dens.av.phase ] ; in doing so , one would be led to conclude that model l4 should be the best performing local model since it was explicitly calibrated to match the @xmath266 relationship from model f1 . as discussed in section [ sec : variations ] , @xmath0 is primarily shielded by itself , the degree of which is a function of the @xmath0 column density @xmath267 , not @xmath162 . thus the @xmath268 relationship is not the ideal proxy for how well a model will predict the @xmath0 abundance . it is important to have a single number that characterizes the level of agreement , with regards to the spatial distribution of @xmath0 and co , between each local approximation model and the fiducial model f1 .
many such quantitative metrics could be constructed which effectively compress a three - dimensional error distribution to a single value . for our purposes here , we favor such a metric that represents the ability of a local approximation to accurately model the atomic - to - molecular transition for both @xmath0 and co , rather than one that is sensitive to diffuse molecules and low density gas , since this is most relevant in regards to both observations and molecule - mediated thermo - energetics . to this end , we have found the best metric to be the mass - weighted fractional error of @xmath269 which can be expressed as @xmath270 where @xmath271 , @xmath272 is the density of molecule @xmath85 in model f1 , @xmath273 is the density of molecule @xmath85 in any separate model , @xmath85 denotes either @xmath0 or co , and the sum runs over all cells @xmath274 .
equation [ eq : err ] is formulated such that perfect agreement between the molecular abundances would produce @xmath275 , while @xmath276 signifies the local approximation model formed no molecules of species @xmath85 whatsoever . to prevent extremely small , negligible abundances from skewing this error metric and to better quantify the significant differences amongst the local models ,
we omit cells with concentrations so low where co would not be a significant coolant , adopting a value of @xmath277 for both co and @xmath0 .
we compute the above error metric between model f1 and each other model ( see table [ tab : models ] ) except for f2 and f3 , whose different density maps make this error metric meaningless . , equation [ eq : aveff ] ) vs. density .
lines of different color denote models with different treatments of radiative shielding ( models l1-l6 ) while the black shaded region represents the @xmath181-@xmath278 dispersion of @xmath178 for model f1.,scaledwidth=42.0% ] an examination of table [ tab : results ] shows that , amongst the local models , the six - ray approximation possesses the smallest @xmath0 weighted error ( l6 , err@xmath279 ) , followed by the temperature capped - jeans length ( l1a , @xmath280 ) and density gradient ( l3 , @xmath281 ) . as for co , model l1a has the lowest error ( err@xmath282 ) followed by the power - law fit ( l4 , @xmath283 ) and single cell shielding ( l5 , @xmath284 ) .
this error metric does have limitations , and should be considered along with the other comparison measures .
for instance , the relatively low err@xmath285 value of model l5 would suggest it does an excellent job matching the co abundances of f1 , though inspecting its co spatial distribution ( figure [ fig : proj.local_compare.co ] ; see also figure [ fig : xco_fiducial_compare ] ) suggests otherwise .
this discrepancy is due to the co error metric ( err@xmath285 , equation [ eq : err ] ) being weighted by @xmath206 , while the logarithmically scaled projections allow the detection of much more diffuse ( @xmath286 ) co. as a final comparison between models l1-l6 and
f1 , we show cell - by - cell , mass - weighted abundance comparisons of @xmath0 and co in figures [ fig : xh2_fiducial_compare ] and [ fig : xco_fiducial_compare ] .
these plots are generated by comparing the @xmath0 fraction between model f1 ( @xmath287 ) and each local model ( @xmath288 ) on a cell - by - cell basis ( and similarly for co ) .
the color scale represents the mass fraction in a particular region of the @xmath289 phase space .
perfect agreement between a local model and model f1 would exclusively lie on the diagonal dashed line defined by @xmath290 .
local models based on the jeans length ( l1 , l1a ) or a power - law fit ( l4 ) tend to slightly overpredict the @xmath0 abundance by @xmath291 ( see table [ tab : results ] ) . as expected , when all radiative shielding is assumed to take place in the immediate vicinity of a computational cell ( model l5 ) the @xmath0 abundance is slightly underpredicted by roughly a factor of @xmath150 .
models in which the local shielding length requires a computation of a discrete gradient or divergence of a grid variable ( l2 and l3 ) tend to exhibit a larger degree of scatter in the @xmath0 abundance as compared to other local models ( figure [ fig : xh2_fiducial_compare ] ) .
as before , model l6 performs the best in matching the results of the fiducial model , overpredicting the total @xmath0 mass by only @xmath292 . turning to co ( figure [ fig : xco_fiducial_compare ] ) , we see a much larger scatter and discrepancy amongst the local models .
models l1 and l3 tend to significantly overpredict the co abundance by up to four orders of magnitude . comparing models l1 and l1a , however , shows how effectively this disagreement can be greatly reduced by imposing a temperature ceiling of @xmath3 in the calculation of the jeans , and thus shielding , length .
model l2 is unique in that effectively no gas exists with @xmath259 .
single cell radiative shielding ( model l5 ) results in relatively little scatter but an underprediction of the co fraction in all cells save for those with co fractions approaching unity .
model l5 and l1a also have a lower @xmath293 ( 1d error representation ) than model l6 , though the 2d error distribution displayed in figure [ fig : xco_fiducial_compare ] seems to suggest l6 to be better performing than l5 and on - par with model l1a , highlighting the utility of multiple error comparisons to gauge the effectiveness of each model in reproducing the molecular abundances of the fiducial model . finally , it is interesting to note the similarity between models l1a ( jeans with temperature ceiling ) and l4 ( power - law fit ) in their prediction of both the @xmath0 and co abundances .
this can be attributed to their respective density scalings : model l4 utilizes a @xmath294 relationship calibrated from model f1 ( @xmath295 ) , while the jeans length approximation can be shown to obey a @xmath296 scaling .
the similarity of these exponents should be unsurprising , though , given the physical principle underlying the jeans length : @xmath263 is the appropriate shielding length to utilize when radiative shielding is due to surrounding material in a dense clump undergoing gravitational instability , exactly the physical entities that are able to achieve @xmath297 and @xmath204 in the simulations we analyze here .
in this paper we have applied ray - tracing and chemical network integration to static simulation snapshots in an effort to model the distribution , morphology , and amount of @xmath0 and co gas in a finite , but representative , portion of a galactic disk .
we have additionally performed calculations where local approximations , instead of ray - tracing , were employed to account for radiative shielding in an effort to assess the validity and accuracy of such radiative transfer alternatives .
we find significant concentrations of @xmath0 gas ( @xmath297 ) are vertically restricted to within @xmath298pc of the galactic midplane and possess a filamentary morphology which traces that of the underlying gas distribution .
gas with significant co concentrations ( @xmath259 ) is further confined to only the highest density clumps , located interior to @xmath0 rich gas .
radiative shielding is crucial for the development of any gas with an @xmath0 or co fraction @xmath299 . when the gas temperature is permitted to reach equilibrium along with the chemical abundances , we find little change from the k15 temperature values ( except for gas at the highest densities , @xmath300 , where co cooling became effective ) , suggesting the bulk of the gas in k15 to exist in thermal equilibrium , and an overall insensitivity of temperature to radiative shielding .
different physical mechanisms are responsible for the radiative shielding of @xmath0 and co. for @xmath0 , self - shielding is far dominant over dust shielding in reducing the @xmath0 photodissociation rate in all regimes explored here . the dominant shielding mechanisms for co , on the other hand , are regime dependent .
more diffuse co ( @xmath301 ) is shielded primarily by dust and @xmath0 while co rich gas ( @xmath259 ) additionally depends on the presence of self - shielding by the molecule itself .
however , this regime dependence for co shielding may be sensitive to the large - scale gas distribution in this particular simulation .
we have compared a number of commonly used local approximations for radiative shielding to ray - tracing based solutions of the radiative transfer equation .
overall , it is promising how well many of the local models reproduce the @xmath0 and co abundances as compared with a more accurate ray - tracing based approach .
one of the main objectives of this work is to inform multidimensional simulations as the validity of these local approximations to account for fuv radiative shielding .
based on our analysis , the six - ray approximation ( l6 ) , temperature - capped jeans length ( l1a ) , and power - law ( l4 ) perform the best in matching the @xmath0 and co abundances of model f1 . among the approximations where the effective shielding length for a given cell ( or , potentially , sph particle ) is exclusively computed from
local _ quantities ( thus excluding model l6 ) , the temperature - capped ( at @xmath3 ) jeans length performed the best , an assessment based on its superiority in terms of matching the total @xmath0 and co masses of model f1 , and possessing the smallest @xmath0 and co weighted errors ( equation [ eq : err ] ) .
we caution , however , that we have only explored one regime , that of a turbulent galactic disk , with relatively coarse ( @xmath302pc ) resolution .
while this regime does include both the h i-@xmath0 and , at the highest densities , c ii - co atomic - to - molecular transitions , there are undoubtedly physical environments to which this analysis applies poorly .
furthermore , the use of any such an approximation for @xmath165 critically depends on the physics a multidimensional simulation is attempting to probe . if one simply wishes to model the thermal impact of the h i-@xmath0 and c ii - co transitions , then coarse approximations , like those discussed above , are likely physically appropriate . on the other hand
, we do not recommend such simple approximations when , for example , attempting to reproduce the detailed distribution and concentration of a large number of chemical species for observational comparison . in these cases , performing detailed , multi - frequency radiative transfer in post - processing , or utilizing advanced radiative transfer modules that operate on - the - fly , is a much more appropriate approach .
all the models presented here operate under the assumption of steady - state ( equilibrium ) chemistry , the validity of which can be assessed by comparing the chemical and dynamical timescales .
taking @xmath303 to be the rate coefficient for @xmath0 formation on grain surfaces , and @xmath304 to be the gas clumping factor on unresolved scales , the chemical time , the timescale for fully atomic gas to convert to molecular form , is given by @xmath305myr .
the dynamical time , roughly the timescale for turbulent motions to displace molecular gas , can be expressed as @xmath306 , which assumes a linewidth - size relation @xmath307 @xcite .
the assumption of equilibrium is violated when @xmath308 , or equivalently when @xmath309 .
thus the equilibrium chemistry is reasonably justified given we focus our analysis on @xmath0 within the high density h i - h@xmath34 transition .
non - equilibrium effects , however , are likely important at lower densities where the @xmath0 formation timescale is much longer . this back - of - the - envelope argument ,
though , should be compared with actual findings from multidimensional simulations that included non - equilibrium chemistry .
the driven turbulence simulations of @xcite identified moderate amounts of molecular hydrogen to be out of chemical equilibrium , a finding they attributed to turbulent transport of @xmath0 from high- to low - density gas that occurs on a shorter timescale than the chemistry can readjust .
this same type of mass transport would likely occur within other large - scale simulations .
similarly , the colliding flow simulations of @xcite show the assumption of chemical equilibrium can under- or over - predict the true @xmath0 abundance depending on density range and integration time , with the largest discrepancies occurring at early times and , interestingly , high densities .
while the assumption of chemical equilibrium likely overestimates the @xmath0 fraction , particularly at low densities , the amount of co is likely _ underestimated _ due to the supernovae feedback prescription in k15 . as discussed in section [ sec : fiducial ] , supernovae events in k15 were modeled by injecting momentum the instant density exceeds the star formation threshold , rather than allowing any time delay . for the original purposes of k15 , which focused on the properties of diffuse atomic gas that is the main ism mass reservoir , the exact timing of supernovae events was unimportant compared to the ability for the supernovae rate to change in time in response to the ism state . however , for the present purposes , this lack of a time delay tends to suppress the formation of large , dense structures where large co abundances would be expected . nonetheless ,
since our primary objective is to understand the role of radiative shielding and numerical approximations to it in determining the chemical balance of the ism , and not to study the chemical balance itself , this limitation of the simulations does not significantly interfere with our goals . in future work
, we plan to utilize our best performing local model , one in which the temperature - capped jeans length is used as a proxy for @xmath165 , to account for radiative shielding in time - dependent radiation - magneto - hydrodynamic amr zoom - in simulations from galactic disk scales down to stellar cluster scales that will achieve significantly higher concentrations of dense , molecule rich gas than is present in the k15 snapshots .
post - processing these snapshots with the tools developed here will further validate the utility of such a local approximation to accurately model fuv radiative shielding and photo - regulated chemistry .
we additionally plan to extend the radiative transfer post - processing analysis performed here to different physical systems , notably isolated , dwarf galaxies with sub - parsec resolution and densities approaching @xmath310 ( e.g. , * ? ? ?
* ) . in these targets of future analyses ,
the presence of more and much larger co rich structures will permit the measurement of integrated line intensities , the x - factor , and their variation with metallicity and observation line - of - sight .
this work was supported by nasa tcan grant nnx-14ab52 g ( for ctss , mrk , ck , eco , jms , sl , cfm , and rik ) .
rik , mrk and cfm acknowledge support from nasa atp grant nnx-13ab84 g .
cfm and rik acknowledge support from nsf grant ast-1211729 .
rik acknowledges support from the us department of energy at the lawrence livermore national laboratory under contract de - ac52 - 07na27344 . | to understand the conditions under which dense , molecular gas is able to form within a galaxy , we post - process a series of three - dimensional galactic - disk - scale simulations with ray - tracing based radiative transfer and chemical network integration to compute the equilibrium chemical and thermal state of the gas .
in performing these simulations we vary a number of parameters , such as the isrf strength , vertical scale height of stellar sources , cosmic ray flux , to gauge the sensitivity of our results to these variations .
self - shielding permits significant molecular hydrogen ( @xmath0 ) abundances in dense filaments around the disk midplane , accounting for approximately @xmath1% of the total gas mass .
significant co fractions only form in the densest , @xmath2 , gas where a combination of dust , @xmath0 , and self - shielding attenuate the fuv background .
we additionally compare these ray - tracing based solutions to photochemistry with complementary models where photo - shielding is accounted for with locally computed prescriptions . with some exceptions , these local models for the radiative shielding length perform reasonably well at reproducing the distribution and amount of molecular gas as compared with a detailed , global ray tracing calculation .
specifically , an approach based on the jeans length with a @xmath3 temperature cap performs the best in regards to a number of different quantitative measures based on the @xmath0 and co abundances .
[ firstpage ] -1 cm stars : formation stars : mass function |
driven many - particle systems are often affected by strongly nonlinear fluctuations which can lead to dramatic phenomena such as self - organized density waves @xcite , jamming @xcite , and spontaneous segregation @xcite . in inhomogeneous type - ii superconductors , nonlinear phenomena resulting from the motion of magnetic flux quanta ( vortices )
have been investigated , for instance , in the context of dendritic flux instabilities stemming from the breakdown of the bean critical state @xcite . in a superconductor , vortices are subjected to the lorentz force induced by an applied current density @xmath0 and to the pinning force produced by sample inhomogeneities . a dissipative dynamical state of moving vortices is established at the critical current @xmath1 when the lorentz force overcomes pinning.@xcite just above @xmath1 , motion is usually plastic with vortices moving at different velocities , which results in a nonlinear velocity - force ( voltage - current ) characteristic . at a stronger drive ,
the pinning potential is washed out and the corresponding dynamical state , coined _ flux flow _ ,
is essentially linear .
these dynamical regimes have been extensively investigated in disordered superconductors @xcite and , more recently , in films with periodic arrays of nanoengineered pinning centers.@xcite however , much less attention has been given to the dynamics of vortices in arrays of obstacles ( or antipinning centers).@xcite because in this configuration vortices are not trapped individually , it renders a very small critical current , which , in principle , has limited interest for applications . on the other hand a small @xmath1 allows for investigation of vortex motion in a much wider current range .
moreover , vortices may be forced into meandering paths inducing strong lateral fluctuations that can give rise to novel dynamical phases . in this paper
, we demonstrate the breakdown of the flux - flow state in a superconducting film with an array of obstacles into a vortex - density wave state and , subsequently , into a fully jammed phase .
evidence of density waves in vortex systems have previously been found in the context of current - depaired vortices and antivortices in clean superconductors @xcite and turbulence in superfluids @xcite .
in contrast , the phenomena we present here are collective states of the moving vortex matter resulting from the interplay of vortex - vortex interactions and the friction induced by the obstacles .
as we shall see in this paper , these competing interactions lead to a critical profile of the density waves where distinct dynamical phases , as well as a jammed phase , coexist .
the paper is organized as follows . in sec
[ sec.model ] , we give the details of our model and numerical procedure . in sec .
[ sec.dynph ] , we present the main results of our molecular dynamics simulations and discuss the dynamical phases . an analytical model is proposed in sec .
[ sec.mf ] to explain the conditions for formation and stabilization of vortex - density waves and how this leads to a constant voltage regime .
[ sec.mf ] is devoted to illustrating an application of the interplay between the moving and jammed phases , namely , the negative absolute resistance effect .
finally , our main findings and final remarks are summarized in sec .
[ sec.concl ] .
we consider a set of @xmath2 vortices generated in the film by a perpendicular magnetic field @xmath3 .
the dynamics of a vortex @xmath4 is modeled by the bardeen - stephen equation , @xmath5 where @xmath6 is the viscous drag coefficient , @xmath7 is the lorentz force induced by the applied current , and @xmath8 is the vortex pair potential , modeled here as @xmath9 ( @xmath10 and @xmath11 is the flux quantum ) .
@xmath12 is the effective penetration depth measuring the range of vortex - vortex interactions . for @xmath12
much larger than all length scales of the system @xmath8 asymptotically reduces to a logarithmic potential . in this investigation
, we analyze the effect of short and long range interactions by conducting simulations for @xmath13 .
hereafter we adopt the following units : @xmath14 ( obstacle lattice constant ) for length , @xmath15 for time , and @xmath16 for energy .
@xmath17 is the potential resulting from a triangular array of cross - shaped obstacles ( fig .
[ fig.pot ] ) . generated by the array .
each cross comprises two legs perpendicular to one another , of length @xmath18 , width @xmath19 , and height @xmath20 . to prevent proximity effects ,
the ferromagnets are assumed to be separated from the superconductor by a @xmath21 thick insulating layer . ] to be specific we have chosen as such obstacles small ferromagnets ( fms ) , each with a uniform permanent magnetization @xmath22 antiparallel to @xmath23 .
this choice is motivated by the well - known properties of fms to either attract or repel vortices , depending on their magnetic orientation @xcite .
other possible choices include higher-@xmath24 superconducting dots @xcite and pillars @xcite .
each fm repels a vortex @xmath25 via a potential given by @xmath26 , where @xmath27 is the position of the volume element of the micromagnet @xmath28 and @xmath29 is the local flux density of vortex @xmath25 at @xmath30 calculated within the london approximation @xcite . in the results presented here we used @xmath31 ( @xmath32 , assuming a coherence length @xmath33 ) . for this value the ferromagnets do not induce any vortex - antivortex pairs @xcite .
@xmath17 results from the superposition of all fms in the array ( fig . [ fig.pot ] )
it introduces a current - induced trapping mechanism between two critical drive values : @xmath34 , at which a vortex can get trapped at a cross corner , and @xmath35 , above which a vortex surmounts the barrier .
these properties provide the main ingredient for the phenomena we investigate here , namely , _ a non - linear , drive - dependent friction_. eqs .
[ eq.motion ] are solved numerically via molecular dynamics ( md ) simulations in a cell of size @xmath36 with periodic boundary conditions .
the values of @xmath37 and @xmath38 were chosen after a careful finite size analysis .
we verified that the transitions between the observed dynamical phases become unaltered for system sizes typically larger or equal to 24@xmath3948 or 12@xmath3996 fms .
therefore , @xmath2 typically ranges up to a few thousands of vortices .
the simulation procedure is as follows .
first , the vortex system is equilibrated at zero current via a standard simulated annealing scheme @xcite . then
, the driving force @xmath40 , applied along the @xmath41 axis is slowly increased . for each @xmath40 ,
time series and averages are calculated on an interval @xmath42 of typically @xmath43 time steps after a stationary state has settled .
the main physical quantities are the center - of - mass velocity @xmath44 of the vortex array along the drive direction and its time average .
we conducted an extensive series of simulations of the model described above for densities ranging from @xmath45 to 2.0 vortices per obstacle and different values of @xmath12 .
[ fig.vi]-(a ) as a function of the driving force @xmath40 for logarithmic ( @xmath46 ) vortex - vortex interaction and occupation numbers ranging from @xmath47 to 1.5 .
the roman numbers correspond to the phases depicted in the @xmath40-@xmath48 diagram ( b ) and explained in the text .
( c ) @xmath49 versus @xmath40 for @xmath47 and different @xmath12 values : @xmath50 to 4 ( symbols ) and @xmath46 ( line ) . ] presents @xmath49-@xmath40 curves for a few @xmath48 values .
the different dynamical behaviors are arranged in the diagram of fig .
[ fig.vi]-(b ) . in general , for @xmath51 , all vortices meander through the potential channels leading to a linear ( flux - flow ) phase , which we call phase i. this phase has an effective viscous drag coefficient , defined as @xmath52 , approximately force independent and smaller than @xmath6 .
this is consistent with the fact that the meandering motion of vortices dissipates more energy as compared to the case of conventional flux flow . above @xmath53
, we observed a remarkable plateau in @xmath54 for long - ranged vortex - vortex interactions .
this constant velocity state , phase ii , persists in a wide range of force values and ends drastically in a fully jammed ( @xmath49=0 ) state for @xmath55 , phase iii , or in a moving linear phase iv for larger @xmath48 . at even higher drives ( @xmath56 ) ,
all vortices surmount the barriers leading to a new flux - flow regime ( phase v ) .
phase ii dominates the diagram in the range of parameters studied . to check the stability of this phase with respect to the kind of vortex - vortex interaction
, we fixed @xmath47 and run simulations for @xmath12 values down to 0.5 .
as observed in fig .
[ fig.vi]-(c ) , the plateau shrinks as @xmath12 decreases and is substituted by a smooth decrease of @xmath49 for @xmath57 . as we shall see below , this region of decreasing @xmath49 have similar properties to the plateau region of the @xmath58 systems .
here we analyze phase ii in more detail .
for simplicity we focus only on the vortex density @xmath47 . in fig .
[ fig.flux ] , we plot snapshots of the flux density distribution @xmath59 for @xmath60 ( long range ) and @xmath61 ( short range ) .
the central panels correspond to density plots of @xmath59 whereas the panels on the right are flux profiles averaged over the @xmath62 direction , i.e. , @xmath63 .
the left panels depict the snapshots of vortex positions in the areas indicated .
these plots reveal a spontaneous breakdown of the vortex distribution when going from phase i [ panels ( a ) and ( f ) ] to phase ii [ ( b)-(d ) , for @xmath64 , and ( g)-(i ) , for @xmath65 .
the resulting patterns comprise stripes of high flux density oriented perpendicularly to the drive direction and traveling at constant speed as soliton - like vortex - density waves ( vdw ) . for long range interactions ,
the waves have a peculiar triangular flux profile , which changes shape as the force is increased , and are intercalated by narrow regions of sparsely distributed vortices trapped by the cross corners . by carefully analyzing the dynamics in different regions of the sample , we found out that the left side of the wave is characterized by meandering motion of vortices , similar to the dynamics at @xmath66 ( phase i ) , while on the right side vortices will rather assist each other to jump over the barrier , which is the same dynamics found at @xmath67 ( phase v ) .
therefore , a single density wave corresponds to a moving region where two distinct dynamical phases coexist .
such phase separation was also observed for short range interactions , but not so clearly because in this case a density wave spans only a few lattice spacings of the fm array .
in addition , the pinned vortex regions for small @xmath12 are considerably larger and increases with force , which is accompanied by a reduction of the mean vortex speed .
a detailed analysis of the morphology of the vdws and its dependence on the driving force and other parameters is given in sec .
[ sec.mf ] . although the vortex wave patterns observed in our simulations are not space periodic , they do present time order . a convenient way to search for time correlations in the vortex density @xmath68 is by studying the power spectrum of its fluctuations , @xmath69 , with @xmath70 .
@xcite we perform this task by recording the density @xmath68 integrated over a @xmath71 region of the film every other 10 time steps during a series of @xmath43 time steps .
@xmath72 is then estimated by averaging @xmath73 over @xmath74 segments of the time series .
plots of typical @xmath72 , for both short ( @xmath75 ) and long ( @xmath64 ) range cases , are depicted in fig .
[ fig.power ] . and @xmath75 ( a ) and @xmath64 ( b ) .
the curves are displaced from each other by two decades for better visualization . in both panels @xmath76 curves ( gray line )
are plotted as guides to the eye ( see text ) .
insets : driving force dependence of the mean vortex velocity and the density wave velocity for @xmath75 and 4.0 . ]
as it is clear in this figure , @xmath72 has a long @xmath76 tail for all drive intensities and interaction ranges studied .
this high frequency behavior is associated with brownian noise resulting mainly from the meandering dynamics , where a vortex colliding with a barrier has to choose randomly whether to dodge left or right .
the low frequency region of @xmath72 , however , changes dramatically from a flat behavior in phase i to a series of strong peaks in phase ii .
these peaks correspond to the harmonics associated with the coherent passage of high flux density domains at constant speed .
therefore , the power spectra provide an efficient means of determining the density - wave velocity : @xmath77 , where @xmath78 is the first harmonic frequency .
we have calculated the wave speed for different force values both for @xmath75 and @xmath79 . for better accuracy
, we calculate @xmath80 from a ( sharper ) higher harmonic @xmath81 ( typically , @xmath82 or 4 ) .
the results are plotted in the insets of fig .
[ fig.power ] .
remarkably , @xmath80 is always larger than the mean vortex speed for both kinds of interactions . for short - range interaction
, @xmath80 increases with the driving force , whereas @xmath83 decreases .
such high speeds of the density waves as compared to @xmath83 can be explained by the presence of slow moving vortices between adjacent waves .
in fact , because vortices in a density valley are diluted , they can easily be trapped at barrier corners ( thus contributing to decrease @xmath83 ) until being catch by a wave front . in opposition , vortices in the density peaks assist each other to overcome the traps .
put in another way , the vdws observed here are _
anti - jamming waves _ self - assembled in a way as to avoid a full stop of the vortex flow .
density waves have been observed in other classical many - particle systems , for instance , car traffic @xcite , granular flow @xcite and galaxies @xcite , and their origin is still unclear .
in contrast to the anti - jamming waves observed in our simulations , the density peaks in these systems are jamming domains intercalated by free motion regions .
accordingly , their wave velocities are usually smaller than the mean flow velocity and can even become negative @xcite .
however , in all these systems , ours included , density waves can be understood as a dynamical , self - organized phase separation , similar to equilibrium pattern formation in systems with competing interactions @xcite .
for instance , in the case of granular media flowing through narrow tubes @xcite density waves form spontaneously and generate a spectrum of density fluctuations very similar to those presented here in fig .
[ fig.power ] , in spite of the very different types of interaction in these systems and different nature of the waves .
to understand how the vortex - density waves are formed and why they lead to a constant voltage regime for large @xmath12 , we propose an analytical model based on a mean - field analysis of eq .
[ eq.motion ] .
we start from the fokker - planck equation for the vortex distribution function at zero temperature and coarse - grain it over scales larger than @xmath12 .
this procedure leads to the following equation for the vortex density @xmath84:@xcite @xmath85.\ ] ] @xmath86 is the lowest order ( long - wavelength ) contribution of vortex - vortex interactions , with @xmath87 .
this indicates that , as expected , a local flux gradient introduces an additional term to the local driving force , that is @xmath88 .
@xmath89 is the coarse - grained force produced by the pattern , which can be understood as a drive - dependent friction force . to account for a dynamical trapping mechanism similar to that induced by the cross - shaped barriers , we assume that @xmath90 represents a static friction in the range @xmath91 , for which no motion occurs . for @xmath92 and @xmath93
, motion takes place and , accordingly , @xmath90 must represent the corresponding kinetic frictions @xmath94 and @xmath95 , respectively .
these properties can be arranged by modeling the friction force as @xmath96 , with : @xmath97 we proceed by assuming a one - dimensional density wave is formed at a given @xmath98 , with @xmath99 , and search for the conditions for such a wave to exist with a fixed shape and a constant velocity @xmath100 directed parallel to @xmath101 . in this case
, we can transform into a reference frame moving with the wave , @xmath102 , and rewrite eq .
[ eq.f - p ] as @xmath103 where @xmath84 is assumed to be a function of @xmath104 only and @xmath105 is the density gradient in the moving frame . to determine the shape of the wave , it is convenient to choose the position of the wave maximum as the origin of @xmath104 and analyze separately the regions @xmath106 and @xmath107 . further , for simplicity
, we assume the wave is single - peaked and define @xmath108 , such that @xmath109 , for @xmath110 , and @xmath111 , for @xmath112 . with the help of eq .
[ eq.friction ] , we can state the following conditions of motion of the wave : @xmath113 where we used the fact that , for any @xmath114 $ ] , the positive slope on the right side of the wave rules out the possibility @xmath115 , whereas the negative slope on the right rules out @xmath116 .
this naturally accounts for the dynamical phase separation observed in our md simulations and discussed in sec .
[ sec.dynph-vdw ] .
a direct consequence of the above condition is that if the vortex distribution peaks up at a certain point , the corresponding bump can only turn into a traveling wave if the gradients in both sides of the bump fulfill eqs .
[ eq.ineq1 ] and [ eq.ineq2 ] .
subcritical bumps , i.e. those with @xmath117 and @xmath118 , would simply be trapped by the pattern . on the other hand ,
an increase in @xmath119 above its critical value leads to an increase in the right side velocity while an increase in @xmath120 above @xmath121 reduces velocity on the left side .
therefore , if the vortex distribution peaks up too sharply the wave will be unstable and will rapidly decay to a profile with gentler slopes so that the velocities on the left and right sides of the wave come to a common value .
it is clear then that the condition for the stability of a wave traveling at constant speed and fixed shape is that _ the density gradient must be everywhere critical _ , that is @xmath122 this result is analogous to the build up of the bean critical state in the magnetization of hard type - ii superconductors , where the flux gradient ( induced by the external flux penetrating the material ) exactly balances the critical force induced by pinning centers . here
the critical forces are exactly balanced by the overall driving force , which accounts for the applied lorentz force and the flux gradient .
[ eq.critl ] and [ eq.critr ] describe a triangular shape for the density wave which switches its left ( right ) side slope from gentle ( steep ) to steep ( gentle ) as the lorentz force is increased from @xmath53 towards @xmath123 . to contrast this result with our simulation data ,
we take @xmath34 and @xmath124 , obtained from the @xmath125 characteristic for @xmath60 and @xmath47 , and calculate the parameter @xmath126 for bessel - like vortex - vortex interactions , which gives @xmath127 .
then , we calculate the critical gradients from eqs .
[ eq.critl ] and [ eq.critr ] and compare with the flux profiles obtained from the md simulations for @xmath128 , 7.0 , and 9.0 .
the results , presented in fig . [ fig.flux ] ( b)-(d ) , demonstrate a good agreement between the numerics and the mean - field model . to estimate the wave velocity , we first impose that the velocity in the left side of the wave ( @xmath129 ) must be consistent with the linear , meandering dynamics expected to occur for @xmath116 . for this kind of motion ,
an increase in the drive leads to a proportional increase in the velocity , i.e. , @xmath130 .
then , we notice that , from eq .
[ eq.critl ] , @xmath131 . hence
, the wave velocity does not depend on @xmath98 : @xmath132 this is in excellent agreement with the md results for large @xmath12 if one assumes that all vortices participate on a wave , that is @xmath133 .
in fact , as revealed by the simulations , there are narrow , low - density regions of trapped vortices coexisting with the dynamical phases for @xmath47 and large @xmath12 .
for that reason , @xmath80 is typically slightly larger than @xmath134 .
it is worth noticing that in the limit @xmath135 a moving wave solution is not possible ( at least in the framework of the present model ) .
therefore , a condition for the occurrence of vortex - density waves is that vortices interact with a potential which is able to trap them only when the applied current surpass a certain critical value .
this is probably the reason why vortex - density waves have never been detected or predicted in , for instance , arrays of pinning centers . in these systems , in general , there is no moving phase preceding the pinned phase .
it is also important to emphasize that the mean - field approximation ( eq . [ eq.f - p ] ) is strictly valid for vortex densities @xmath136 . in the simulations performed for @xmath137 and @xmath138 we have @xmath139 , which can be considered inside the range of validity of the model
indeed , the main results obtained in the md calculations for these parameters were correctly reproduced by the mean - field approach . for @xmath140 , however , the vortex distribution is too diluted .
in addition , in order to treat the interaction with the barriers macroscopically , @xmath12 should span several lattice spacings .
for these reasons , the analytical results presented here are not strictly valid for short range interactions .
notwithstanding , our simple analysis ignores the details of the barriers , taking into account only its main macroscopic properties .
therefore , it can be applied to other barrier configurations , as long as their coarse - grained properties can be expressed in a way similar to eq .
[ eq.friction ] .
the dynamical transition between the vdw and jamming phases opens the possibility for an interesting application : the construction of an active device ( made of superconducting material ) exhibiting _ negative absolute resistance _ ( nar ) .
the working principle of this device is based on the negative absolute mobility ( nam ) effect , which corresponds to motion in a direction that opposes the driving force , irrespective of the drive direction .
this phenomenon was predicted to occur in the transport of a single classical brownian particle through a symmetric , periodic substrate @xcite and was experimentally corroborated in a system of colloidal spheres in a microfluidic device @xcite and subsequently in a josephson junction @xcite . in the later , the phase dynamics , which can be mapped into the problem of a single brownian particle , lead to negative resistance .
quite generally , the main ingredients for the observation of nam are : ( i ) a medium that allows easy motion of the particles at low drives and strongly suppresses mobility at high drives ; ( ii ) a fluctuating ( nonequilibrium ) force , for instance an ac excitation , superposed to the dc drive .
the first requirement implies necessarily that , for a certain force range , the mean velocity of the driven system must decrease as the driving force increases , that is , negative differential mobility .
this phenomenon have been experimentally demonstrated for vortices driven in superconducting films with periodic arrays of pinning centers at magnetic fields close to one flux quantum per pinning site ( @xmath141).@xcite previous numerical calculations have pointed out that such negative differential mobility of vortices near @xmath141 is a result of a transition from a highly - dissipative , disordered regime to a filamentary state with somewhat smaller but non - vanishing mean velocities.@xcite as shown in fig .
[ fig.vi ] , the @xmath142 characteristics of the vortex system under investigation here do present negative differential resistance , in a broad field range , as a result of the transition from the vdw to the jammed state . here
, however , the voltage drop corresponds to a strong suppression of vortex mobility , which , as discussed below , is a crucial feature for the observation of nam .
here we demonstrate how the nam effect can be applied to abrikosov vortices in patterned superconductors in order to achieve negative resistance .
we subject the vortices to a lorentz force @xmath143 , with @xmath144 and an ac amplitude @xmath145 fixed at a value chosen just above the vdw - jamming ( ii - iii ) transition , and measure @xmath49 , which is proportional to the dc voltage , as a function of the dc drive @xmath146 , proportional to the dc current .
the ac term of @xmath40 provide the extra energy source required for the observation of nam .
the results are shown in fig .
[ fig.nar ] in the presence of a sinusoidal excitation with frequency @xmath147 and amplitudes @xmath148,@xmath149 , @xmath150 and @xmath151 corresponding to @xmath152 , respectivelly .
all curves exhibit absolute negative mobility of vortices at small @xmath146 . ] for several @xmath12 values . at zero dc current ( @xmath153 )
the vortex dynamics cycle symmetrically through phases i , ii , iii , and back , producing zero dc voltage
. however , for small positive ( negative ) @xmath146 , vortices are most of the time in the pinned phase iii during the positive ( negative ) half - cycle of @xmath154 and always in one of the moving phases i or ii in the negative ( positive ) half - cycle .
this leads to a net motion contrary to the dc force , that is , the mean electric field induced by vortex motion points antiparallel to the applied dc current , thus generating the negative absolute resistance effect .
in conclusion , we studied novel dynamical phases of vortices in a patterned superconducting film . by molecular dynamics calculations we demonstrated , for a wide range of parameters , the existence of vortex - density waves propagating at a constant speed @xmath80 larger than the mean vortex velocity @xmath134 and essentially force - independent for long - range vortex - vortex interactions .
the waves consist of well - defined regions of coexisting moving phases , where vortices assist each other to either dodge or overcome the barriers , and are intercalated by regions of trapped vortices .
our mean - field analysis revealed that these waves stem from fluctuations in the vortex distribution induced by a highly nonlinear friction force .
a density bump produced by such fluctuations turns into a stable moving wave when it reaches a certain critical profile .
it is precisely this critical shape of the waves that results in a force independent wave velocity , thereby naturally accounting for the voltage plateau observed in the md simulations .
we have also demonstrated the feasibility of a superconducting device which exhibits negative absolute resistance .
this is achieved by exploring a combination of ac and dc excitations in such a way as to conveniently switch the system between the moving and fully - jammed phases .
this prediction could be promptly tested using conventional transport measurement techniques on a nanostructured sample with a pattern similar to that proposed here .
finally , it is worth pointing out that , given the generality of our mean field model , the main results predicted here can also be applied to other systems of interacting particles , such as colloids and pedestrians .
an advantage of the vortex system proposed here is that interactions can easily be tuned . in typical nanostructured superconducting films
, @xmath12 can be varied from a fraction to several lattice spacings by controlling the film temperature near @xmath24 @xcite , thereby allowing for experimentally accessing our predictions in both long and short range cases . in additional modern imaging techniques
could be used to identify the vortex density waves .
we would like to thank alejandro v. silhanek , giovani l. vasconcelos , and leonardo r. e. cabral for useful suggestions and enlightening discussions .
this work was supported by the brazilian science agencies cnpq and facepe ( grant no .
apq-0589 - 1.05/08 ) . | we study theoretically dynamical phases of vortices in superconducting films with arrays of obstacles . by performing a series of molecular dynamics simulations and analytical calculations , we demonstrate the existence of a phase of soliton - like vortex - density waves existing in a wide range of parameters .
these waves are formed by a self - assembled phase separation process induced by strongly nonlinear density fluctuations of the moving vortex matter above a certain critical driving current . at high vortex concentrations ,
the waves move at an approximately current - independent speed resulting in a wide plateau in the voltage - current characteristics . at stronger drives ,
the vortex system enters into a fully jammed ( zero - voltage ) phase . by combining ac and dc drives
, the interplay between the vortex - density - wave and jammed phases leads to the observation of negative absolute mobility of vortices , which induces the superconducting film into a negative resistance state . |
the series @xmath2 where @xmath3 , can be expressed as @xmath4 this identity ( [ binomial_thm ] ) is called the binomial theorem . the binomial theorem ( [ binomial_thm ] )
has many applications .
for example , because @xmath5 from ( [ binomial_thm ] ) , we can perform numerical calculation of @xmath6 effectively ( cf .
292 page in chapter 4 in part ii of [ eu ] ) : @xmath7 using ( [ binomial_thm ] ) like this , we are able to do numerical calculations of algebraic numbers .
moreover , we obtain , for non - negative integer @xmath8 , @xmath9 thus , we can obtain combinatorial identities from ( [ binomial_thm ] ) . as seen above
, the binomial theorem is very useful .
analogically , if hypergeometric series which are generalizations of @xmath10 can be expressed in terms of well - known functions , then it is expected that those hypergeometric identities have many applications .
indeed , transcendental numbers are expressed in term of hypergeometric series as we see later .
further , it is known that most of combinatorial identities appearing in [ goul ] are special cases of those identities ( cf .
[ an ] ) .
that is why many methods for obtaining hypergeometric identities have been constructed . in particular , in the last several decades , many methods that exploit progress in computer technology have been formulated : gosper s algorithm , the w - z method , zeilberger s algorithm , etc .
[ ko ] and [ pwz ] ) .
these algorithms have been used to obtain and prove hypergeometric identities . in this article
, we present a new method for finding identities for hypergeoemtric series , such as the ( gauss ) hypergeometric series , the generalized hypergeometric series and the appell - lauricella hypergeometric series .
moreover , as an example of the application of this method , we give identities for the hypergeometric series ; we see that values of the hypergeometric series at some arguments can be expressed in terms of gamma functions , together with certain elementary functions . because our method systematically yields almost all known identities of this type besides many new identities , for completeness , we tabulate the entire set . in the appendix of this paper
, we also make use of our method to have some hypergeometric identities for generalized hypergeometric series and appell - lauricella hypergeometric series . in particular ,
hypergeometric identities for latter series have never been obtained .
the reason comes from the fact that it is difficult to apply known methods directly to appell - lauricella hypergeometric series because these are multiple series .
therefore , our method will become a powerful tool for investigating hypergeometric identities for many kinds of hypergeometric series , especially for appell - lauricella hypergeometric series .
now , we present a new method for finding identities for hypergeoemtric series . to begin with , we give a proof of the binomial theorem ( [ binomial_thm ] ) , which is the model of our method . put @xmath11 .
recalling that @xmath12 , we see @xmath13 we also get @xmath14 the above two formulae lead to @xmath15 thus , we find that @xmath10 and @xmath16 are linearly related over rational functions .
furthermore , we can regard ( [ proof_binom ] ) as a first order difference equation with rational functions coefficients .
now , we solve this difference equation ( [ proof_binom ] ) . because @xmath17 holds for any positive integer @xmath18 from ( [ proof_binom ] ) ,
we see @xmath19 substituting @xmath20 in ( [ proof_binom2 ] ) , we get the following identity : @xmath21 since the identity ( [ proof_binom3 ] ) is valid for any non - negative integer @xmath18 , we find that ( [ proof_binom3 ] ) holds for any complex number @xmath18 by virtue of carlson s theorem ( cf . 5.3 in [ ba ] or lemma 3.1 in this article ) .
thus , we get the binomial theorem .
our method for finding hypergeometric identities is a generalization of this procedure .
it is known that several hypergeometric series with the same parameters up to additive integers are linearly related over rational functions .
suppose that this linear relation degenerates into a linear relation between two hypergeometric series under a condition , and further , this degenerate relation can be regarded as the first order difference equation with rational functions coefficients .
then , solving this difference equation as we get the binomial theorem , we will obtain a hypergeometric identity . in this paper , applying this method to the hypergeometric series @xmath22 we actually obtain almost all previously known hypergeometric identities and , of course , many new hypergeometric identities . as stated previously
, we will also make use of this method to yield some hypergeometric identities for generalized hypergeometric series and appell - lauricella hypergeometric series in the appendix of this paper .
it is known that for a given triple of integers @xmath23 there exists a unique pair of rational functions @xmath24 satisfying @xmath25 where @xmath26 is the field generated over @xmath27 by @xmath28 , @xmath29 , @xmath30 and @xmath1 ( cf .
1 in [ eb1 ] ) . this relation is called the * three term relation * of the hypergeometric series . from this , we obtain @xmath31 where @xmath32 let @xmath33 be a solution of the system @xmath34 then , for this @xmath33 , we have the first order difference equation @xmath35 or we have @xmath36 solving ( [ degenerated0 ] ) or ( [ degenerated1 ] ) , we obtain hypergeometric identities ; for example , if @xmath37 , then by substituting @xmath38 , where @xmath39 is an integer satisfying @xmath40 , into ( [ degenerated1 ] ) , we find the hypergeometric identity @xmath41 in this way , we can find values of @xmath0 . in this paper , we call values of the hypergeometric series obtained by our method * special values * of the hypergeometric series .
we summarize how to get special values of the hypergeometric series : * step 1 : give @xmath42
. * step 2 : obtain @xmath33 satisfying the system ( [ condition ] ) .
* step 3 : solve ( [ degenerated0 ] ) or ( [ degenerated1 ] ) .
we remark that the numerator of @xmath43 is a polynomial in @xmath18 over @xmath44 $ ] for a given @xmath42 .
therefore , in order to obtain @xmath33 satisfying the system ( [ condition ] ) , we only have to seek the @xmath33 that eliminate all of the coefficients of this polynomial .
namely , we solve polynomial systems in @xmath45 to implement step 2 .
the grbner basis theory is useful for this .
thus , we see that ( [ degenerated1 ] ) and ( [ -nk - k ] ) are valid for any integer @xmath18 from the constitution method of @xmath33 satisfying ( [ condition ] ) . as an example , we give a special value for @xmath46 . because the three term relation for @xmath46 is @xmath47 we have @xmath48 hence , @xmath33 satisfying ( [ condition ] ) are @xmath49 .
however , since the special values in the cases that @xmath50 are obvious from the definition of @xmath0 , we only consider the special value in the case that @xmath51 . in this case , we have the first order difference equation @xmath52 or we have @xmath53 for any positive number @xmath18 , where we assume @xmath54 which is the convergence condition of @xmath55 . noticing that @xmath56 we obtain @xmath57 for @xmath54 .
this is called the gauss summation formula ( cf . [
48 ] in 24 of [ ga ] ) .
moreover , as a particular case of ( [ gauss ] ) , we have @xmath58 where @xmath59 .
though ( [ chu ] ) has been known since the 13th century , today , this is called the chu - vandermonde equality ( cf .
corollary 2.2.3 in [ aar ] ) .
the special values for other lattice points @xmath60 are considered in sections 3 and 4 . in [ 48 ]
in 24 of [ ga ] , gauss obtains ( [ gauss ] ) using ( [ gauss2 ] ) and ( [ gauss3 ] ) which are got by substituting @xmath61 into the relation @xmath62 therefore , our method is applied by gauss in the special case . in the above example , we obtained the gauss summation formula ( [ gauss ] ) .
this is the general expression of @xmath0 at @xmath61 .
hence , we only have to consider values of @xmath0 at other points .
so , in step 2 , we seek @xmath33 satisfying the following : let @xmath63 be a triple such that the number of solutions of ( [ condition ] ) is finite , and let @xmath1 @xmath64 be one of its solutions ( cf .
remark 2.4 ) .
we call such a quadruple @xmath33 an * admissible quadruple*. although we need to investigate @xmath42 for obtaining the special values of @xmath0 as seen in step 1 , we actually only have to investigate @xmath60 contained in @xmath65 we close this section by giving an outline of this statement ( see section 2 for details ) . the hypergeometric equation @xmath66 , where @xmath67 admits 23 hypergeometric solutions in addition to @xmath0 . for each of these solutions
, there exists a relation similar to ( [ degenerated1 ] ) .
for example , for the solution @xmath68 , we have @xmath69 where @xmath70 and @xmath71 appear in ( [ three_term ] ) .
thus , because @xmath72 for an admissible quadruple @xmath33 , we can get special values of @xmath73 with this quadruple . the same
can be done for the other 22 solutions .
thus , for the lattice point @xmath60 , we are able to obtain special values of 24 hypergeometric series with the above quadruple .
the identity ( [ degenerated2 ] ) implies that the special values of the 24 hypergeometric series mentioned above for the lattice points @xmath74 coincide with those for the lattice point @xmath60 ( see subsection 2.3 for details ) .
in other words , @xmath74 is equivalent to @xmath60 with respect to the obtained special values . in fact , this holds generally , as all 24 lattice points represented by triples corresponding to the 24 hypergeomtric solutions are equivalent with respect to these special values .
in addition , from the relation @xmath75 , which implies that @xmath60 is equivalent to @xmath76 , it can also be shown that the @xmath77 @xmath78 lattice points are equivalent with respect to these special values .
these 48 lattice points form the orbit of @xmath60 under the action of the group @xmath79 on @xmath80 , where @xmath79 is the group generated by following mappings @xmath81 we remark that @xmath82 , where @xmath83 is the symmetric group of degree @xmath18 . from this , regarding these 48 lattice points as equivalent , we can take ( [ quotient ] ) as a complete system of representatives of the quotient @xmath84 of this action ( see subsection 2.3 for details ) .
thus , we only need to investigate the lattice points contained in ( [ quotient ] ) to obtain the special values of the hypergeometric series . in this paper
, we tabulate the special values for @xmath60 satisfying @xmath85 .
as stated in the previous section , for a given @xmath60 , the 24 hypergeometric solutions of @xmath86 with an admissible quadruple @xmath33 have two term relations as ( [ degenerated1 ] ) and ( [ degenerated2 ] ) , say degenerate relations . in this section , we list these degenerate relations explicitly .
this list will be used subsequently , when we evaluate special values of the corresponding hypergeometric series . after presenting this list ,
we give a proof that ( [ quotient ] ) can be taken as a complete system of representatives of @xmath84 . in this subsection
, we introduce contiguity operators .
the contiguity operators of the hypergeometric series are defined as @xmath87 where @xmath88 .
these operators satisfy the following relations : @xmath89 ( cf .
theorems 2.1.1 and 2.1.3 in [ iksy ] ) .
let @xmath90 be a composition of these contiguity operators such that @xmath91 also , @xmath90 can be expressed as @xmath92 where @xmath93 is an element of the ring of the differential operators in @xmath1 over @xmath26 , and @xmath94 ( cf .
( 2.4 ) in [ eb1 ] ) . in this subsection
, we list degenerate relations .
@xmath86 admits the following 24 hypergeometric solutions ( cf
. 2.9 in [ erd ] ) : @xmath95 in the above , we must take the appropriate branches of @xmath96 and @xmath97 .
we now apply both sides of ( [ composition ] ) to @xmath98 @xmath99 .
first , noting the relation @xmath100 we have @xmath101 from ( [ three_term_f ] ) and ( [ composition ] ) .
this implies that @xmath102 next , we apply both sides of ( [ composition ] ) to @xmath103 .
this yields @xmath104 ( cf . (
3.2 ) and lemma 2.2 in [ eb1 ] ) . combining this and ( [ qr ] )
, we obtain @xmath105 finally , applying both sides of ( [ composition ] ) to @xmath98 @xmath106 , we have the following : we define @xmath70 and @xmath71 as ( [ three_term ] ) .
then , we have @xmath107 using lemmas 2.1 and 2.2 , we are able to obtain 24 degenerate relations . for example , substituting @xmath108 into the top formula in lemma 2.2 , we obtain @xmath109 therefore , defining @xmath110 where @xmath111 , we find that @xmath112 for an admissible quadruple @xmath33 .
the following is obtained similarly from lemmas 2.1 and 2.2 .
fix @xmath42 . for an admissible quadruple @xmath33
, we obtain the following 24 degenerate relations : @xmath113 from the treatment presented in section 1 , it is understood that @xmath1 in an admissible quadruple is not a free parameter .
the reason for this is the following .
if @xmath1 were a free parameter , then some of the 24 degenerate relations may be incorrect , while the special values obtained from the remaining correct relations would be trivial .
( here ` trivial ' means obvious from the definition of @xmath0 . ) for example , in the case @xmath114 , we have @xmath115 this implies that @xmath116 and , therefore , the @xmath33 satisfying ( [ condition ] ) are @xmath117 and @xmath118 . in the former case , we find that the 12 degenerate relations ( xiii)-(xxiv ) are incorrect because the denominator of the coefficient of @xmath119 @xmath120 in lemma 2.2 contains @xmath28 as a factor .
in addition , the special values obtained from the 12 correct degenerate relations are trivial .
for this reason , we exclude cases in which @xmath1 is a free parameter from consideration . in this subsection
, we prove that we only need to investigate the lattice points contained in ( [ quotient ] ) to obtain the special values of the hypergeometric series .
first , we show that the 48 lattice points that form the orbit of @xmath60 under the action of @xmath79 on @xmath80 ( cf .
section 1 ) are equivalent with respect to the obtained special values .
we do this by considering two particular examples , @xmath122 and @xmath123 , and demonstrating that they are equivalent to @xmath60 with respect to the special values .
the general result follows by analogy . to begin with
, we consider the case of @xmath122 .
we start by replacing @xmath33 with @xmath124 .
then , the three term relation for @xmath122 is expressed as @xmath125 and that for @xmath126 is @xmath127 these two three term relations lead to @xmath128 equating ( [ ( m - k , m - l , m ) ] ) with ( [ ( m - k , m - l , m)-2 ] ) , we have @xmath129 hence , the admissible quadruples for @xmath60 coincide with those for @xmath122 .
this implies that the special values for @xmath60 coincide with those for @xmath122 ; that is , @xmath122 is equivalent to @xmath60 with respect to the obtained special values .
we next show that @xmath123 is equivalent to @xmath60 with respect to the obtained special values . from the top formula in lemma 2.2
, we have @xmath130 also , making the replacement @xmath131 , we find that the three term relation for @xmath123 is expressible as @xmath132 then , using @xmath133 and ( [ k , m - l , m-2 ] ) , we get @xmath134 finally , equating ( [ k , m - l , m ] ) and ( [ k , k - m , m-2 ] ) , we obtain @xmath135 therefore , the admissible quadruples for @xmath60 coincide with those for @xmath123 .
this means that @xmath123 is equivalent to @xmath60 with respect to the obtained special values .
we can show that the other 46 lattice points are equivalent to @xmath60 with respect to the obtained special values analogously .
the reason that we assumed @xmath136 in the definition of an admissible quadruple @xmath33 is that if we do not make this assumption , there are cases in which the admissible quadruples for @xmath60 do not coincide with those for the other 47 points ( cf .
( [ qq ] ) , ( [ k , k - m , m-3 ] ) and subsection 3.1 ) .
next , we determine a complete system of representatives of the quotient of the action of @xmath79 on @xmath80 .
recall that @xmath137 . to begin with
, we find a complete system of representatives of the quotient of the action of @xmath138 on @xmath80 . from @xmath139 , we see that @xmath140 is identical to @xmath60 .
hence , we can assume @xmath141 .
further , because @xmath122 is identical to @xmath60 by @xmath142 , we can assume @xmath143 .
in addition , because @xmath76 is identical to @xmath60 by @xmath144 , we can assume @xmath145 .
thus , we can take @xmath146 , which we call @xmath147 , as a complete system of representatives of the quotient of the action of @xmath138 on @xmath80 .
we now proceed to obtain a complete system of representatives of the quotient of the action of @xmath79 on @xmath80 .
if @xmath148 , then @xmath147 also contains @xmath149 , @xmath150 , @xmath151 , @xmath152 and @xmath153 . then , with @xmath154 we have the following : @xmath155 together , these imply that we are able to adopt ( [ quotient ] ) as a complete system of the quotient of the action of @xmath79 on @xmath80 .
thus , we only have to investigate the lattice points contained in ( [ quotient ] ) to obtain the special values of the hypergeometric series .
in this section , we derive the special values for some lattice points as examples .
although we implicitly assumed that the parameter @xmath30 of @xmath0 is an element of @xmath156 , for the rest of paper , we expand the definition of @xmath0 .
from this point , even if the parameter @xmath30 is a non - positive integer , we define @xmath0 as follows if the parameter @xmath28 is a non - positive integer satisfying @xmath157 : @xmath158 we first consider the case @xmath159 .
then , the only @xmath60 satisfying ( [ quotient ] ) is @xmath160 .
for this point , the corresponding three term relation is ( [ ( 0,1,1 ) ] ) , and this leads to @xmath161 hence , @xmath51 satisfies ( [ condition ] ) ( and so does @xmath162 , but for the reason explained in remark 2.4 , we do not consider this case ) . in the case
@xmath51 , because @xmath163 we have @xmath164 from this , substituting @xmath165 , where @xmath166 , we get the chu - vandermonde equality , ( [ chu ] ) .
note that this equality holds even if @xmath30 is a non - positive integer with @xmath167 . because @xmath118 is not an admissible quadruple
, we should investigate the other 47 points .
for example , considering the case @xmath46 , we get the gauss summation formula , ( [ gauss ] ) .
similarly , we only get the chu - vandermonde equality and the gauss summation formula from the other 46 points . in the case
@xmath168 , because @xmath169 the numerator of @xmath43 is @xmath170 therefore , the admissible quadruple @xmath33 is @xmath171 , and @xmath172 where we denote @xmath173 by @xmath174 , and we also denote thus for the rest of this article .
hence , ( i ) in proposition 2.3 leads to @xmath175 although ( [ ( 1,2,2 ) ] ) is valid by virtue of analytic continuation , this equality regarded as an infinite series expression does not make sense .
for this reason , we carry out the degeneration of this into a finite series expression that does make sense .
this is done separately in the following cases : @xmath176 , @xmath177 and @xmath178 , where @xmath166 .
if @xmath176 , then the right hand side of ( [ ( 1,2,2 ) ] ) becomes @xmath179 if @xmath177 , then the right hand side of ( [ ( 1,2,2 ) ] ) is @xmath180 if @xmath178 , then the right hand side of ( [ ( 1,2,2 ) ] ) is equal to @xmath181 the special values obtained from ( ii ) , ( iii ) and ( iv ) are identical to the above .
next , using ( v ) in proposition 2.3 , we have @xmath182 substituting @xmath20 into ( [ ( 1,2,2)-v ] ) , this becomes @xmath183 note that the validity of ( [ ( 1,2,2)-v1 ] ) for any integer @xmath18 follows from the constitution method of @xmath33 satisfying ( [ condition ] ) ( cf .
( [ consti_method ] ) ) . the following lemma is known ( cf .
5.3 in [ ba ] ) : ( carlson s theorem ) we assume that @xmath184 and @xmath185 are regular and of the form @xmath186 , where @xmath187 , for @xmath188 , and @xmath189 for @xmath190
. then , @xmath189 on @xmath191 .
it is easily confirmed that both sides of ( [ ( 1,2,2)-v1 ] ) satisfy the assumption of the above lemma .
hence , ( [ ( 1,2,2)-v1 ] ) holds for any complex number @xmath18 for which the left hand side is meaningful .
resultingly , substituting @xmath192 into ( [ ( 1,2,2)-v1 ] ) , we find @xmath193 now , we consider the case @xmath28 , @xmath194 with @xmath195 . specifically , we consider the case @xmath196 , where @xmath18 , @xmath197 .
then , ( [ ( 1,2,2)-v ] ) leads to @xmath198 next , we consider the case @xmath29 , @xmath194 with @xmath199 .
then , if @xmath200 , @xmath201 , where @xmath202 , @xmath166 , ( [ ( 1,2,2)-v ] ) implies @xmath203 if @xmath200 , @xmath204 , where @xmath202 , @xmath166 , we find @xmath205 from ( [ ( 1,2,2)-v ] ) .
we are able to obtain special values from the other 19 degenerate relations similarly .
those values are tabulated in the following section .
for an admissible quadruple @xmath207 , we have @xmath208 now , we consider the special values obtained from @xmath209 in proposition 2.3 . because @xmath209 in this case is @xmath210 we find @xmath211 however , we can not directly apply lemma 3.1 to the above identity , because its left hand side does not satisfy the assumption of that lemma .
for this reason , we get the special value of @xmath212 using the following algebraic transformation of the hypergeometric series : @xmath213 with @xmath214 ( see formula ( 130 ) in [ gour ] ) .
we remark that this holds near @xmath215 .
now , we carry out an analytic continuation of each side of ( [ algebraic ] ) along a curve starting at @xmath215 and ending at @xmath216 , as depicted in figure 1 .
-plane , width=264 ] -plane , width=264 ] noting that @xmath217 encircles @xmath218 once counterclockwise , as shown in figure 2 , we obtain @xmath219 using the following lemma ( cf .
theorem 4.7.2 in [ iksy ] ) : let @xmath220 be a loop starting and ending at @xmath221 , where @xmath222 , and encircling @xmath61 once in the counterclockwise direction .
then , analytic continuation of @xmath223 along @xmath220 is given by @xmath224 where @xmath225,\ a=\left [ \begin { array}{cc } 1&0\\ \noalign{\medskip}0&{{\rm e}^{2\,i \pi \ , \left ( c - a - b \right ) } } \end { array } \right ] .\end{gathered}\ ] ]
in this section , using proposition 3.2.3 , we tabulate the special values for @xmath60 satisfying @xmath226 .
however , we exclude the chu - vandermonde equality , ( [ chu ] ) and the gauss summation formula , ( [ gauss ] ) .
let @xmath18 , @xmath202 , @xmath227 .
then , recall that @xmath228 when @xmath28 and @xmath30 are non - positive integers satisfying @xmath229 .
further , recall that we denote @xmath230 by @xmath231 , and so on . for a given @xmath60
, we use the expression ( * * ) @xmath232 ( * ) to indicate that the values obtained from ( * * ) in proposition 3.2.3 coincide with or are contained in those obtained from ( * ) . in this case
, there is no admissible quadruple ( cf .
subsection 3.1 ) . in the case that @xmath168 , we obtain @xmath234 [ [ section ] ] ( 1,2,2 - 1 ) + + + + + + + + + & ( i ) f(a , b;2a;2)= , + , + 0 & ( the first case is identical to ( 4.11 ) in [ ge ] ) .
we find that ( ii ) , ( iii ) , ( iv ) @xmath232 ( i ) . &
( v ) f(a , b;b+1-a;-1 ) & + & = , + , + 0 , + & ( the first case is identical to 2.8(47 ) in [ erd ] ) .
we find that ( vi)@xmath232(v ) . &
( vii ) ( a,1-a;b+1-a;1/2 ) & + & = , + , + & ( the first case is identical to 2.8(51 ) in [ erd ] ) . &
( viii ) ( b , b+1 - 2a;b+1-a;1/2 ) & + & = , + , + 0 & ( the first case is identical to 2.8(50 ) in [ erd ] ) .
the special values obtained from ( ix)-(xxiv ) are contained in the above . when @xmath235 , we obtain @xmath236 [ [ section-1 ] ] ( 0,2,2 - 1 ) + + + + + + + + + the special values obtained from ( 0,2,2 - 1 ) are contained in those obtained from ( 1,2,2 - 1 ) . in this case , there is no admissible quadruple . in this case
, we have @xmath239 [ [ section-2 ] ] ( 1,3,2 - 1 ) + + + + + + + + + & ( i ) ( a,3a-1;2a;1/2 + 1/2i ) & + & = , + & ( the first case is identical to 2.8(55 ) in [ erd ] and the second case is identical to theorem 11 in [ ek ] ) . &
( ii ) ( a,1-a;2a;1/2 + 1/2i ) & + & = , + . & & ( iii ) ( a,1-a;2a;1/2 - 1/2i ) & + & = , + . & & ( iv)(a,3a-1;2a;1/2 - 1/2i ) & + & = , + & ( the first case is identical to 2.8(56 ) in [ erd ] and the second case is identical to theorem 11 in [ ek ] ) .
the special values obtained from ( v)-(xxiv ) are contained in the above .
[ [ section-3 ] ] ( 1,3,2 - 2 ) + + + + + + + + + the special values obtained from ( 1,3,2 - 2 ) coincide with those obtained from ( 1,3,2 - 1 ) . in this case , we get @xmath242 [ [ section-4 ] ] ( 0,3,3 - 1 ) + + + + + + + + + the special values obtained from ( 0,3,3 - 1 ) except trivial values are special cases of @xmath243 [ [ section-5 ] ] ( 0,3,3 - 2 ) , ( 0,3,3 - 3 ) , ( 0,3,3 - 4 ) + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the special values obtained from ( 0,3,3 - 2 ) , ( 0,3,3 - 3 ) and ( 0,3,3 - 4 ) coincide with those obtained from ( 0,3,3 - 1 ) . in this case , we have @xmath244 [ [ section-6 ] ] ( 1,2,3 - 1 ) + + + + + + + + + & ( i ) ( a,2a-1/3;3a;9 ) = , + , + & ( the first case is identical to theorem 17 in [ ek ] ) . &
( ii ) ( 2a , a+1/3;3a;9 ) = , + 0 , + & ( the third case is identical to theorem 18 in [ ek ] ) . &
( iii ) ( a , a+1/3;3a;9/8 ) = , + .
& & ( iv ) ( 2a,2a-1/3;3a;9/8 ) = , + 0 , + , + .
+ & & ( v ) ( a,2a-1/3;2/3;-8 ) = ( -3)^3n , + ( -3)^3n , + ( -3 ) ^3n+1 & ( the second case is identical to ( 3.12 ) in [ gs ] and the first case is identical to ( 5.23 ) in [ gs ] ) .
we find that ( vi)@xmath245(v ) . &
( vii ) f(a,1 - 2a;2/3;8/9)= 2 ^ -a ( ( 5/6-a ) ) & ( the above is identical to ( 8/9.1 ) in [ gos ] and ( 3.2 ) in [ ka ] ) .
we derive ( vii ) using ( [ algebraic ] ) and connection formulae for the hypergeometric series(see ( 25)(44 ) in 2.9 in [ erd ] ) .
we find that ( viii)@xmath245(vii ) . &
( ix ) ( a,1 - 2a;4/3-a;1/9)= & ( the above is identical to ( 1/9.4 ) in [ gos ] and ( 1.2 ) in [ ka ] ) . &
( x ) ( 4/3 - 2a , a+1/3;4/3-a;1/9)= & & ( the above is identical to ( 1/9.5 ) in [ gos ] and ( 1.3 ) in [ ka ] ) . & ( xi ) ( a , a+1/3;4/3-a;-1/8)= & . &
( xii ) ( 1 - 2a,4/3 - 2a;4/3-a;-1/8)= & ( the above is a generalization of theorem 32 in [ ek ] ) .
we find ( xiii)@xmath232(ix ) , ( xiv)@xmath232(x ) , ( xv)@xmath232(xii ) , ( xvi)@xmath232(xi ) , ( xvii)@xmath232(i ) , ( xviii)@xmath232(ii ) , ( xix)@xmath232(iv ) , ( xx)@xmath232(iii ) . &
( xxi ) ( 2a , a+1/3;4/3;-8 ) = , + 0 , + + & ( the third case is identical to ( 3.7 ) in [ gs ] ) .
we find ( xxii)@xmath232(xxi ) . &
( xxiii ) ( 2a,1-a;4/3;8/9 ) = & ( the above is identical to ( 8/9.2 ) in [ gos ] , ( 5.24 ) in [ ds ] and ( 3.3 ) in [ ka ] ) .
we derive ( xxiii ) by using formula ( 132 ) in [ gour ] and connection formulae for the hypergeometric series(see ( 25)(44 ) in 2.9 in [ erd ] ) .
we find ( xxiv)@xmath232(xxiii ) .
[ [ section-7 ] ] ( 1,2,3 - 2 ) + + + + + + + + + the special values obtained from ( 1,2,3 - 2 ) coincide with those obtained from ( 1,2,3 - 1 ) . in this case , we get @xmath247 [ [ section-8 ] ] ( 1,3,3 - 1 ) + + + + + + + + + & ( i ) ( a,3a-1/2;3a;-3 ) = , + , + , + 0 . + & & ( ii ) ( 1/2,2a;3a;-3 ) = , + 0 . + & & ( iii ) ( 1/2,a;3a;3/4 ) = , + & ( the first case is identical to ( 3/4.1 ) in [ gos ] ) . &
( iv ) ( 2a,3a-1/2;3a;3/4 ) = , + .
& & ( v ) ( a,3a-1/2;a+1/2;4 ) = ( -3)^3n , + ( -3)^3n , + - ( -3 ) ^3n+1 & ( the first case is identical to theorem 12 in [ ek ] ) . &
( vi ) ( 1/2,1 - 2a;a+1/2;4 ) = 1 , + -1/3 . & & ( vii ) ( a,1 - 2a;a+1/2;4/3 ) = 3 ^ 2n , + 3 ^ -2n , + 3 ^ -2n-2 .
& & ( viii ) ( 1/2,3a-1/2;a+1/2;4/3 ) = 1 , + -1 .
+ & & ( ix ) ( a,1 - 2a;3/2 - 2a;-1/3 ) = . & & ( x ) ( 1/2,3/2 - 3a;3/2 - 2a;-1/3 ) = & ( the above is a generalization of ( 28.1 ) in [ ge ] ) . &
( xi ) ( 1/2,a;3/2 - 2a;1/4 ) = & ( the above is identical to ( 1/4.1 ) in [ gos ] and ( 5.22 ) in [ gs ] ) . &
( xii ) ( 1 - 2a,3/2 - 3a;3/2 - 2a;1/4 ) = & ( the above is a generalization of theorem 30 in [ ek ] ) . &
( xiii ) ( 1/2,3a-1/2;2a+1/2;-1/3 ) = . & & ( xiv ) ( 2a,1-a;2a+1/2;-1/3 ) = . & & ( xv ) ( 2a,3a-1/2;2a+1/2;1/4 ) = . & & ( xvi ) ( 1/2,1-a;2a+1/2;1/4 ) = & ( the above is identical to ( 1/4.2 ) in [ gos ] ) . &
( xvii ) ( 1/2,1 - 2a;2 - 3a;-3 ) = , + - . & & ( xviii)(1-a,3/2 - 3a;2 - 3a;-3 ) & + & = , + , + 0 , + 0 . & & ( xix)(1 - 2a,3/2 - 3a;2 - 3a;3/4 ) & + & = , + . + & & ( xx ) ( 1/2,1-a;2 - 3a;3/4 ) & + & = , + & ( the first case is identical to ( 3/4.2 ) in [ gos ] ) . &
( xxi ) ( 1/2,2a;3/2-a;4 ) = , + 0 & ( the above are identical to ( 5.25 ) in [ gs ] ) . &
( xxii ) ( 1-a,3/2 - 3a;3/2-a;4 ) & + & = , + 0 , + 0 & ( the first case is identical to theorem 13 in [ ek ] ) . &
( xxiii ) ( 2a,1-a;3/2-a;4/3 ) = , + 0 , + .
& & ( xxiv ) ( 1/2,3/2 - 3a;3/2-a;4/3 ) = 0 , + 0 . &
[ [ section-9 ] ] ( 1,3,3 - 2 ) + + + + + + + + + the special values obtained from ( 1,3,3 - 2 ) coincide with those obtained from ( 1,3,3 - 1 ) .
[ [ section-10 ] ] ( 1,3,3 - 3 ) + + + + + + + + + & ( i ) ( a,3a+1;3a;3/2 ) = 0 , + , + . &
the special values obtained from ( ii ) and ( iii ) are trivial . &
( iv ) ( 2a,3a+1;3a;3 ) = 0 , + , + .
+ & & ( v ) ( a,3a+1;a+2;-1/2 ) = 2 ^ 3a3 ^ -3a ( a+1 ) & ( the above is a special case of @xmath248 this is given in ( a / b.1 ) in [ gos ] and ( 1.6 ) in [ eb2 ] ) .
& ( vi ) ( 2,1 - 2a;a+2;-1/2 ) = 2/3a+2/3 & ( the above is a special case of @xmath249 this is given in ( 1.5 ) in [ eb2 ] ) .
& ( vii ) ( a,1 - 2a;a+2;1/3 ) = 2 ^ 2a3 ^ -2a ( a+1 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( viii ) ( 2,3a+1;a+2;1/3 ) = 3/2a+3/2 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( ix)(a,1 - 2a;-2a;2/3 ) = 0 , + , + .
+ & the special values obtained from ( x ) and ( xi ) are trivial . & ( xii)(-3a,1 - 2a;-2a;-2 )
= 0 , + 0 , + , + .
+ & & ( xiii ) ( 2,3a+1;2a+2;2/3 ) = 6a+3 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( xiv ) ( 2a,1-a;2a+2;2/3 ) = 3 ^ -a ( 2a+1 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( xv ) ( 2a,3a+1;2a+2;-2)= 3 ^ 3n ( 1 - 6n ) , + 3 ^ 3n+1 ( -1 - 6n ) & ( the above are special cases of ( [ 1.6eb2 ] ) ) .
& ( xvi ) ( 2,1-a;2a+2;-2 ) = 2/3n+1 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( xvii ) ( 2,1 - 2a;2 - 3a;3/2 ) = 6n+1 , + 6n+10 & ( the above are special cases of ( [ 1.5eb2 ] ) ) .
& ( xviii ) ( -3a,1-a;2 - 3a;3/2 ) = ( -2 ) ^-1-n ( -3n-2 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . & ( xix ) ( -3a,1 - 2a;2 - 3a;3 ) = 2 ^ 2n ( 6n+1 ) , + - 2 ^ 2n+3 ( 6n+10 ) & ( the above are special cases of ( [ 1.6eb2 ] ) ) . &
( xx ) ( 2,1-a;2 - 3a;3 ) = 3/2n+1 & the above is a special case of ( [ 1.5eb2 ] ) ) .
the special values obtained from ( xxi ) are trivial . &
( xxii ) ( -3a,1-a ;- a;-1/2 ) = 0 , + .
& & ( xxiii ) ( 2a,1-a ;- a;1/3 ) = 0 , + . &
the special values obtained from ( xxiv ) are trivial .
[ [ section-11 ] ] ( 1,3,3 - 4 ) + + + + + + + + + the special values obtained from ( 1,3,3 - 4 ) coincide with those obtained from ( 1,3,3 - 3 ) . in this case
, we have & ( a , b , c , x)=(a , b , b+1-a,-1 ) , + & ( a , b , c , x)=(a,4a-1/2,3a,-1 ) , s^(n)= , & + & ( a , b , c , x)=(a,4a-5/2,3a-1,-1 ) , s^(n)=. & [ [ section-12 ] ] ( 1,4,3 - 1 ) + + + + + + + + + the special values obtained from ( 1,4,3 - 1 ) are evaluated in paragraphs ( 1,2,2 - 1 ) and ( 0,2,2 - 1 ) , [ [ section-13 ] ] ( 1,4,3 - 2 ) + + + + + + + + + & ( i ) ( a,4a-1/2;3a;-1 ) & + & = , + & ( the first case is a generalization of theorem 1 in [ ek ] ) . &
( ii ) ( 2a,1/2-a;3a;-1 ) & + & = , + . & & ( iii)(a,1/2-a;3a;1/2 ) + & = , + . & & ( iv)(2a,4a-1/2;3a;1/2 ) + & = , + . & & ( v ) ( a,4a-1/2;2a+1/2;2 ) & + & = , + , + , + 0 , + 0 . & & ( vi ) ( 1 - 2a , a+1/2;2a+1/2;2 ) & + & = , + ,
+ . & & ( vii)(a,1 - 2a;2a+1/2;2 ) = , + , + . & & ( viii)(4a-1/2,a+1/2;2a+1/2;2 ) & + & = , + , + 0 , + 0 , + . + & we find that ( ix)@xmath232(ii ) , ( x)@xmath232(i ) , ( xi)@xmath232(iii ) , ( xii)@xmath232(iv ) . & ( xiii)(4a-1/2,a+1/2;3a+1/2;-1 ) & + & = , + & ( the second case is identical to theorem 2 in [ ek ] ) . & ( xiv)(2a,1-a;3a+1/2;-1 ) & + & = , + . & & ( xv)(2a,4a-1/2;3a+1/2;1/2 ) & + & = , + . & & ( xvi)(1-a , a+1/2;3a+1/2;1/2 ) & + & = , + . & the special values obtained from ( xvii)-(xxiv ) are contained in the above .
[ [ section-14 ] ] ( 1,4,3 - 3 ) + + + + + + + + + the special values obtained from ( 1,4,3 - 3 ) coincide with those obtained from ( 1,4,3 - 2 ) . in this case
, we have @xmath253 where , @xmath254 is a solution of @xmath255 .
[ [ section-15 ] ] ( 0,4,4 - 1 ) + + + + + + + + + the special values obtained from ( 0,4,4 - 1 ) are evaluated in the case ( 0,2,2 - 1 ) .
[ [ section-16 ] ] ( 0,4,4 - 2 ) + + + + + + + + + the special values obtained from ( 0,4,4 - 2 ) except trivial values are the special cases of ( [ f(1,b;2;x ) ] ) .
[ [ section-17 ] ] ( 0,4,4 - 3 ) + + + + + + + + + the special values obtained from ( 0,4,4 - 3 ) coincide with those obtained from ( 0,4,4 - 2 ) in this case , we have @xmath257 [ [ section-18 ] ] ( 1,3,4 - 1 ) + + + + + + + + + & ( i ) ( a,3a-1/2;4a;4 ) = , + , + 0 , + & ( the first case is identical to theorem 3 in [ ek ] ) . &
( ii)(3a , a+1/2;4a;4 ) = , + , + 0 , + + & ( the fourth case is identical to theorem 9 in [ ek ] ) . &
( iii)(a , a+1/2;4a;4/3 ) = , + . & & ( iv)(3a,3a-1/2;4a;4/3 ) & + & = , + , + 0 , + , + 0 , + - , + & & ( v)(a,3a-1/2;1/2;-3 ) = , + , + 0 , + .
+ & we find ( vi)@xmath245(v ) . &
( vii)(a,1 - 3a;1/2;3/4 ) = & ( the above is identical to ( 3/4.4 ) in [ gos ] )
. we derive ( vii ) by using the algebraic transformation formula ( 44 ) in 2.11 in [ erd ] and connection formulae for the hypergeometric series(see ( 25)(44 ) in 2.9 in [ erd ] ) .
we find ( viii)@xmath245(vii ) . &
( ix)(a,1 - 3a;3/2 - 2a;1/4 ) = . & & ( x)(a+1/2,3/2 - 3a;3/2 - 2a;1/4 ) = . & & ( xi)(a , a+1/2;3/2 - 2a;-1/3 ) = . & & ( xii)(1 - 3a,3/2 - 3a;3/2 - 2a;-1/3 ) = . &
the special values obtained from ( xiii)-(xx ) are contained in the above those . &
( xxi)(3a , a+1/2;3/2;-3 ) = , + , + 0 , + . &
we find ( xxii)@xmath245(xxi ) . &
( xxiii)(3a,1-a;3/2;3/4 ) = . &
( the above is identical to ( 3/4.3 ) in [ gos ] ) .
we derive ( xxiii ) by using the algebraic transformation formula ( 46 ) in 2.11 in [ erd ] and connection formulae for the hypergeometric series(see ( 25)(44 ) in 2.9 in [ erd ] ) .
we find ( xxiv)@xmath245(xxiii ) .
[ [ section-19 ] ] ( 1,3,4 - 2 ) + + + + + + + + + the special values obtained from ( 1,3,4 - 2 ) coincide with those obtained from ( 1,3,4 - 1 ) .
[ [ section-20 ] ] ( 1,3,4 - 3 ) + + + + + + + + + & ( i)(a,3a-1/2;4a;-8 ) = , + , + , + & ( the first case is identical to theorem 7 in [ ek ] ) . &
( ii)(3a , a+1/2;4a;-8 ) = , + 0 , + 0 , + & ( the fourth case is identical to theorem 5 in [ ek ] ) . &
( iii)(a , a+1/2;4a;8/9 ) = , + , + . & & ( iv)(3a,3a-1/2;4a ; 8/9 ) = , + , + & ( the first case is identical to ( 3.1 ) in [ ka ] ) . &
( v)(a,3a-1/2;1/2;9 ) = 2 ^ 6n , + 2 ^ 6n , + 2 ^ 6n+2 , + -2 ^ 6 n+3 . &
we find ( vi)@xmath245(v ) . &
( vii)(a,1 - 3a;1/2;9/8 ) = ( -2 ) ^3n , + ( -2 ) ^-3n , + ( -2 ) ^-3n-1 , + ( -2 ) ^-3n-3 . &
we find ( viii)@xmath245(vii ) . &
( ix)(a,1 - 3a;3/2 - 2a;-1/8 ) = ( ) ^1 - 3a . & & ( x)(a+1/2,3/2 - 3a;3/2 - 2a;-1/8 ) = . & & ( xi)(a ,
a+1/2;3/2 - 2a;1/9 ) = & ( the above is identical to ( 1/9.1 ) in [ gos ] ) . &
( xii)(1 - 3a,3/2 - 3a;3/2 - 2a;1/9 ) = ( ) ^3a-1 & ( the above is identical to ( 1.1 ) in [ ka ] ) .
the special values obtained from ( xiii)-(xx ) are contained in the above those . & ( xxi)(3a , a+1/2;3/2;9 ) = , + 0 , + 0 , + , + & we find ( xxii)@xmath245(xxi ) . & ( xxiii)(3a,1-a;3/2 ; 9/8 ) = , + 0 , + 0 , + .
& we find ( xxiv)@xmath245(xxiii ) .
[ [ section-21 ] ] ( 1,3,4 - 4 ) + + + + + + + + + & ( i)(a,3a-1/4;4a;-8 ) = , + , + , + 0 + & ( the first case is identical to theorem 6 in [ ek ] ) . &
( ii)(3a , a+1/4;4a;-8 ) = , + , + 0 , + + & ( the fourth case is identical to theorem 4 in [ ek ] ) . &
( iii)(a , a+1/4;4a;8/9 ) = , + , + . + & & ( iv)(3a,3a-1/4;4a;8/9 ) & + & = , + , + + & ( the first case is identical to ( 3.4 ) in [ ka ] ) . & ( v)(a,3a-1/4;3/4;9 )
= , + , + , + 0 . + & & ( vi)(1 - 3a,3/4-a;3/4;9 ) = , + 0 , + , + . & & ( vii)(a,1 - 3a;3/4;9/8 ) = , + , + 0 , + .
+ & & ( viii)(3a-1/4,3/4-a;3/4;9/8 ) & + & = , + , + 0 , + . & & ( ix)(a,1 - 3a;5/4 - 2a;-1/8 ) & + & = . & & ( x)(5/4 - 3a , a+1/4;5/4 - 2a;-1/8 ) & + & = . & & ( xi)(a , a+1/4;5/4 - 2a;1/9 ) & + & = & ( the above is identical to ( 1/9.2 ) in [ gos ] , ( 6.5 ) in [ gs ] and ( 1.4 ) in [ ka ] ) . &
( xii)(1 - 3a,5/4 - 3a;5/4 - 2a;1/9 ) & + & = . & & ( xiii)(3/4-a,3a-1/4;2a+3/4;-1/8 ) & + & = . & & ( xiv ) ( 3a,1-a;2a+3/4;-1/8 ) & + & = . & & ( xv ) ( 3a,3a-1/4;2a+3/4;1/9 ) & + & = . & & ( xvi)(1-a,3/4-a;2a+3/4;1/9 ) & + & = & ( the above is identical to ( 1/9.3 ) in [ gos ] , ( 6.6 ) in [ gs ] and ( 1.5 ) in [ ka ] ) . &
( xvii)(1 - 3a,3/4-a;2 - 4a;-8 ) & + & = , + 0 , + , + + & ( the fourth case is identical to theorem 8 in [ ek ] ) . &
( xviii)(1-a,5/4 - 3a;2 - 4a;-8 ) & + & = , + , + - , + 0 ( the first case is identical to theorem 10 in [ ek ] ) . &
( xix)(1 - 3a,5/4 - 3a;2 - 4a ; 8/9 ) & + & = , + , + - + & ( the first case is identical to ( 3.5 ) in [ ka ] ) . &
( xx ) ( 1-a,3/4-a;2 - 4a;8/9 ) & + & = , + , + .
+ & & ( xxi)(3a , a+1/4;5/4;9 ) = , + , + 0 , + .
+ & & ( xxii ) ( 1-a,5/4 - 3a;5/4;9 ) = , + , + , + 0 . + & & ( xxiii)(3a,1-a;5/4 ; 9/8 ) = , + , + 0 , + . + & & ( xxiv)(5/4 - 3a , a+1/4;5/4;9/8 ) & + & = , + , + 0 , + . + & [ [ section-22 ] ] ( 1,3,4 - 5 ) , ( 1,3,4 - 6 ) , ( 1,3,4 - 8 ) + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the special values obtained from ( 1,3,4 - 5 ) , ( 1,3,4 - 6 ) and ( 1,3,4 - 8 ) coincide with those obtained from ( 1,3,4 - 4 ) . [ [ section-23 ] ] ( 1,3,4 - 7 ) + + + + + + + + + the special values obtained from ( 1,3,4 - 7 ) coincide with those obtained from ( 1,3,4 - 3 ) . in this case , we have @xmath259 [ [ section-24 ] ] ( 1,4,4 - 1 ) + + + + + + + + + & ( i ) ( a,4a+1;4a;4/3 ) & + & = 0 , + , + , + . &
the special values obtained from ( ii ) and ( iii ) are trivial . &
( iv ) ( 3a,4a+1;4a;4 ) & + & = 0 , + , + , + . &
( v ) ( a,4a+1;a+2;-1/3 ) = 2 ^ -8a3 ^ 4a ( a+1 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( vi)(2,1 - 3a;a+2;-1/3 ) = 3/4a+3/4 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( vii)(a,1 - 3a;a+2;1/4 ) = 2 ^ -6a3 ^ 3a ( a+1 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( viii ) ( 2,4a+1;a+2;1/4 ) = 4/3a+4/3 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( ix ) ( a,1 - 3a;-3a;3/4 ) = 0 , + , + , + .
+ & the special values obtained from ( x ) and ( xi ) are trivial . &
( xii ) ( -4a,1 - 3a;-3a;-3 ) = 0 , + 0 , + 0 , + , + , + .
+ & & ( xiii ) ( 2,4a+1;3a+2;3/4 ) = 12a+4 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( xiv)(3a,1-a;3a+2;3/4 ) = 2 ^ -2a ( 3a+1 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) .
& ( xv ) ( 3a,4a+1;3a+2;-3 ) = 2 ^ 8n ( 1 - 12n ) , + 2 ^ 8n+2 ( -2 - 12n ) , + 2 ^ 8n+4 ( -5 - 12n ) & ( the above are special cases of ( [ 1.6eb2 ] ) ) .
& ( xvi ) ( 2,1-a;3a+2;-3 ) = 3/4n+1 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( xvii ) ( 2,1 - 3a;2 - 4a;4/3 ) = 4n+1 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( xviii ) ( -4a,1-a;2 - 4a;4/3 ) = - ( -3 ) ^-n-1 ( 4n+3 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( xix ) ( -4a,1 - 3a;2 - 4a;4 ) = ( -3 ) ^n ( 4n + 1 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( xx ) ( 2,1-a;2 - 4a;4 ) = 4/3n+1 & ( the above is a special case of ( [ 1.5eb2 ] ) ) .
the special values obtained from ( xxi ) are trivial . &
( xxii ) ( -4a,1-a ;- a;-1/3 ) = 0 , + . + & & ( xxiii ) ( 3a,1-a ;- a;1/4 ) = 0 , + & ( the first case is identical to ( 29.3 ) in [ ge ] ) .
the special values obtained from ( xxiv ) are trivial .
[ [ section-25 ] ] ( 1,4,4 - 2 ) + + + + + + + + + the special values obtained from ( 1,4,4 - 2 ) coincide with those obtained from ( 1,4,4 - 1 ) . in this case
, there is no admissible quadruple . in this case
, we have @xmath262 [ [ section-26 ] ] ( 2,2,4 - 1 ) , ( 2,2,4 - 2 ) + + + + + + + + + + + + + + + + + + + + the special values obtained from ( 2,2,4 - 1 ) and ( 2,2,4 - 2 ) coincide with those obtained from ( 1,2,2 - 1 ) . in this case , we have @xmath264 [ [ section-27 ] ] ( 2,3,4 - 1 ) + + + + + + + + + & ( i)(a,3/2a-1/4;2a;8 + 4 ) & + & = , + , + , + , + 0 + & ( the first case is identical to theorem 28 in [ ek ] ) . & ( ii)(a,1/2a+1/4;2a;8 + 4 ) & + & = , + , + + & ( the third case is identical to theorem 16 in [ ek ] ) . & ( iii)(a,1/2a+1/4;2a;8 - 4 ) & + & = , + , + & ( the third case is identical to theorem 16 in [ ek ] ) . & ( iv)(a,3/2a-1/4;2a;8 - 4 ) & + & = , + , + , + , + 0 + & ( the first case is identical to theorem 28 in [ ek ] ) . & ( v)(a,3/2a-1/4;1/2a+3/4;-7 - 4 ) & + & = , + , + , + , + 0 + & ( the first case is identical to theorem 31 in [ ek ] ) . & ( vi ) ( 1-a,3/4 - 1/2a;1/2a+3/4;-7 - 4 ) & + & = , + , + .
+ & & ( vii ) ( a,1-a;1/2a+3/4;1/2 + 1/4 ) & + & = & ( the above is a generalization of theorem 37 in [ ek ] ) .
we derive the above formula using the algebraic transformation @xmath265 and connection formulae for the hypergeometric series ( see ( 25)(44 ) in 2.9 in [ erd ] ) . &
( viii ) ( 3/4 - 1/2a,3/2a-1/4;1/2a+3/4;1/2 + 1/4 ) & + & = . & & ( ix ) ( a,1-a;5/4 - 1/2a;1/2 - 1/4 ) & + & = & ( the above is a generalization of theorem 37 in [ ek ] ) . &
( x ) ( 5/4 - 3/2a,1/2a+1/4;5/4 - 1/2a;1/2 - 1/4 ) + & = . & & ( xi)(a,1/2a+1/4;5/4 - 1/2a;-7 + 4 ) + & = . & & ( xii ) ( 1-a,5/4 - 3/2a;5/4 - 1/2a;-7 + 4 ) + & = & ( the above is a generalization of theorem 31 in [ ek ] ) . the special values obtained from ( xiii)-(xxiv )
are contained in the above those .
[ [ section-28 ] ] ( 2,3,4 - 2 ) + + + + + + + + + the special values obtained from ( 2,3,4 - 2 ) coincide with those obtained from ( 2,3,4 - 1 ) . in this case
, we have @xmath267 [ [ section-29 ] ] ( 2,4,4 - 1 ) + + + + + + + + + the special values obtained from ( 2,4,4 - 1 ) are evaluated in the case ( 1,2,2 - 1 ) .
[ [ section-30 ] ] ( 2,4,4 - 2 ) + + + + + + + + + & ( i)(a,2a-1/2;2a;-2 + 2 ) & + & = , + , + . & & ( ii)(1/2,a;2a;-2 + 2 ) & + & = , + , + & ( the second case is identical to theorem 33 in [ ek ] ) . &
( iii)(1/2,a;2a;-2 - 2 ) & + & = , + & ( the first case is identical to theorem 33 in [ ek ] ) . &
( iv)(a,2a-1/2;2a;-2 - 2 ) & + & = , + , + , + , + 0 , + 0 . & & ( v)(a,2a-1/2;a+1/2;3 - 2 ) = & ( the above is a generalization of theorem 25 in [ ek ] ) . &
( vi)(1/2,1-a;a+1/2;3 - 2 ) = . & & ( vii)(a,1-a;a+1/2;1/2 - 1/2 ) = & ( the above is a generalization of theorem 36 in [ ek ] ) . &
( viii)(1/2,2a-1/2;a+1/2;1/2 - 1/2 ) = . & & ( ix ) ( a,1-a;3/2-a;1/2 + 1/2 ) & + & = , + , + , + & ( the third case is identical to theorem 36 in [ ek ] ) . &
( x ) ( 1/2,3/2 - 2a;3/2-a;1/2 + 1/2 ) & + & = , + , + 0 , + 0 . & & ( xi)(1/2,a;3/2-a;3 + 2 ) & + & = , + . & & ( xii ) ( 1-a,3/2 - 2a;3/2-a;3 + 2 ) & + & = , + , + , + , + 0 , + 0 + & ( the first case is identical to theorem 25 in [ ek ] ) . the special values obtained from ( xiii)-(xxiv ) coincide with the above .
[ [ section-31 ] ] ( 2,4,4 - 3 ) + + + + + + + + + the special values obtained from ( 2,4,4 - 3 ) coincide with those obtained from ( 2,4,4 - 2 ) . in this case , we have @xmath269 [ [ section-32 ] ] ( 2,5,4 - 1 ) + + + + + + + + + & ( i ) ( a,5/2a-1;2a;-1/2 + 1/2 ) & + & = , + , + . & & ( ii ) ( a,1 - 1/2a;2a;-1/2 + 1/2 ) & + & = , + , + & ( the first case is a generalization of theorem 24 in [ ek ] ) . &
( iii)(a,1 - 1/2a;2a;-1/2 - 1/2 ) & + & = , + , + . & & ( iv ) ( a,5/2a-1;2a;-1/2 - 1/2 ) & + & = , + , + , + , + 0 , + 0 . & & ( v ) ( a,5/2a-1;3/2a;3/2 - 1/2 ) & + & = , + . & & ( vi ) ( 1-a,1/2a;3/2a;3/2 - 1/2 ) & + & = , + & ( the second case is identical to theorem 21 in [ ek ] ) . &
( vii ) ( a,1-a;3/2a;1/2 - 1/2 ) & + & = , + . & & ( viii ) ( 1/2a,5/2a-1;3/2a;1/2 - 1/2 ) & + & = , + . & & ( ix ) ( a,1-a;2 - 3/2a;1/2 + 1/2 ) & + & = , + , + , + & ( the third case is identical to theorem 34 in [ ek ] ) . &
( x)(2 - 5/2a,1 - 1/2a;2 - 3/2a;1/2 + 1/2 ) & + & = , + , + 0 , + , + 0 . & & ( xi ) ( a,1 - 1/2a;2 - 3/2a;3/2 + 1/2 ) & + & = , + , + + & ( the third case is identical to theorem 22 in [ ek ] ) .
& ( xii ) ( 1-a,2 - 5/2a;2 - 3/2a;3/2 + 1/2 ) & + & = , + , + , + , + 0 , + 0 . + & & ( xiii ) ( 1/2a,5/2a-1;3/2a;1/2 + 1/2 ) & + & = , + , + , + 0 , + 0 . & & ( xiv ) ( a,1-a;3/2a;1/2 + 1/2 ) & + & = , + , + , + + & ( the first case is identical to theorem 35 in [ ek ] ) . &
( xv ) ( a,5/2a-1;3/2a;3/2 + 1/2 ) & + & = , + , + , + , + 0 , + 0 . + & & ( xvi ) ( 1/2a,1-a;3/2a;3/2 + 1/2 ) & + & = , + , + + & ( the first case is identical to theorem 21 in [ ek ] ) . & ( xvii ) ( 1/2a,1-a;2 - 2a;-1/2 + 1/2 ) & + & = , + , + + & ( the first case is a generalization of theorem 23 in [ ek ] ) . &
( xviii ) ( 1-a,2 - 5/2a;2 - 2a;-1/2 + 1/2 ) & + & = , + , + , + & & ( xix ) ( 1-a,2 - 5/2a;2 - 2a;-1/2 - 1/2 ) & + & = , + , + , + , + 0 , + 0 . & & ( xx)(1/2a,1-a;2 - 2a;-1/2 - 1/2 ) & + & = , + , + . & & ( xxi ) ( a,1 - 1/2a;2 - 3/2a;3/2 - 1/2 ) & + & = , + & ( the second case is identical to theorem 22 in [ ek ] ) . &
( xxii ) ( 1-a,2 - 5/2a;2 - 3/2a;3/2 - 1/2 ) & + & = , + . & & ( xxiii ) ( a,1-a;2 - 3/2a;1/2 - 1/2 ) & + & = , + . & & ( xxiv ) ( 2 - 5/2a,1 - 1/2a;2 - 3/2a;1/2 - 1/2 ) & + & = , + . &
[ [ section-33 ] ] ( 2,5,4 - 2 ) , ( 2,5,4 - 3 ) , ( 2,5,4 - 4 ) + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the special values obtained from ( 2,5,4 - 2 ) , ( 2,5,4 - 3 ) and ( 2,5,4 - 4 ) coincide with those obtained from ( 2,5,4 - 1 ) . in this case , we have @xmath271 [ [ section-34 ] ] ( 2,6,4 - 1 ) , ( 2,6,4 - 2 ) , ( 2,6,4 - 3 ) + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the special values obtained from ( 2,6,4 - 1 ) , ( 2,6,4 - 2 ) and ( 2,6,4 - 3 ) are evaluated in paragraphs ( 1,2,2 - 1 ) and ( 0,2,2 - 1 ) .
[ [ section-35 ] ] ( 2,6,4 - 4 ) , ( 2,6,4 - 5 ) + + + + + + + + + + + + + + + + + + + + the special values obtained from ( 2,6,4 - 4 ) and ( 2,6,4 - 5 ) coincide with those obtained from ( 1,3,2 - 1 ) . in this case
, we have @xmath274 where @xmath254 is a solution of @xmath275 .
[ [ section-36 ] ] ( 0,5,5 - 1 ) + + + + + + + + + the special values obtained from ( 0,5,5 - 1 ) except trivial values are the special cases of ( [ f(1,b;2;x ) ] ) .
[ [ section-37 ] ] ( 0,5,5 - 2 ) + + + + + + + + + the special values obtained from ( 0,5,5 - 2 ) coincide with those obtained from ( 0,5,5 - 1 ) . in this case
, there is no admissible quadruple . in this case
, we have @xmath278 [ [ section-38 ] ] ( 1,5,5 - 1 ) + + + + + + + + + & ( i)(a,5a+1;5a;5/4 ) = 0 , + , + , + , + .
+ & the special values obtained from ( ii ) and ( iii ) are trivial . &
( iv)(4a,5a+1;5a;5 ) = 0 , + , + , + , + . & & ( v)(a,5a+1;a+2;-1/4 ) = 2 ^ 10a5 ^ -5a ( a+1 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( vi ) ( 2,1 - 4a;a+2;-1/4 ) = 4/5a+4/5 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( vii ) ( a,1 - 4a;a+2;1/5 ) = 2 ^ 8a5 ^ -4a ( a+1 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( viii ) ( 2,5a+1;a+2;1/5 ) = 5/4a+5/4 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( ix ) ( a,1 - 4a;-4a;4/5 ) = 0 , + , + , + , + .
+ & the special values obtained from ( x ) and ( xi ) are trivial . &
( xii)(-5a,1 - 4a;-4a;-4 ) = 0 , + 0 , + 0 , + 0 , + , + , + , + .
+ & & ( xiii ) ( 2,5a+1;4a+2;4/5)=5(4a+1 ) & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( xiv ) ( 4a,1-a;4a+2;4/5)=5 ^ -a ( 4a+1 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( xv ) ( 4a,5a+1;4a+2;-4 ) = -5 ^ 5n ( 20n -1 ) , + -5 ^ 5n+1 ( 20n+3 ) , + -5 ^ 5n+2 ( 20n + 7 ) , + -5 ^ 5n+3 ( 20n + 11 ) + & ( the above are special cases of ( [ 1.6eb2 ] ) ) . &
( xvi ) ( 2,1-a;4a+2;-4 ) = 4/5n+1 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( xvii)(2,1 - 4a;2 - 5a;5/4 ) = 5n+1 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( xviii ) ( -5a,1-a;2 - 5a;5/4 ) = - ( -4 ) ^-n-1 ( 5n + 4 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( xix ) ( -5a,1 - 4a;2 - 5a;5 ) = 2 ^ 8n ( 20n+1 ) , + -2 ^ 8n+2 ( 20n+6 ) , + 2 ^ 8n+4 ( 20n+11 ) , + -2 ^ 8n+14 ( 20n+36 ) & ( the above are special cases of ( [ 1.6eb2 ] ) ) . &
( xx ) ( 2,1-a;2 - 5a;5 ) = 5/4n+1 & ( the above is a special case of ( [ 1.5eb2 ] ) ) .
the special values obtained from ( xxi ) are trivial . &
( xxii ) ( -5a,1-a ;- a;-1/4 ) = 0 , + + & ( the first case is identical to ( 29.6 ) in [ ge ] ) . &
( xxiii ) ( 4a,1-a ;- a;1/5 ) = 0 , +
. + & the special values obtained from ( xxiv ) are trivial .
[ [ section-39 ] ] ( 1,5,5 - 2 ) + + + + + + + + + the special values obtained from ( 1,5,5 - 2 ) coincide with those obtained from ( 1,5,5 - 1 ) . in this case
, we have @xmath280 [ [ section-40 ] ] ( 1,6,5 - 1 ) + + + + + + + + + the special values obtained from ( 1,6,5 - 1 ) are evaluated in paragraphs ( 1,2,2 - 1 ) and ( 0,2,2 - 1 ) . in this case
, there is no admissible quadruple . in this case
, there is no admissible quadruple . in this case , we have @xmath284 [ [ section-41 ] ] ( 2,5,5 - 1 ) + + + + + + + + + & ( i ) ( a,5/2a+1;5/2a;5/3 ) & + & = 0 , + , + , + , + .
+ & the special values obtained from ( ii ) and ( iii ) are trivial . &
( iv ) ( 3/2a,5/2a+1;5/2a;5/2 ) & + & = 0 , + , + , + , + . + & & ( v ) ( a,5/2a+1;a+2;-2/3 ) = 3 ^ 5/2a5 ^ -5/2a ( a+1 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( vi ) ( 2,1 - 3/2a;a+2;-2/3 ) = 3/5a+3/5 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( vii ) ( a,1 - 3/2a;a+2;2/5 ) = 3 ^ 3/2a5 ^ -3/2a ( a+1 ) & ( the above is a special case of ( [ 1.6eb2 ] ) and is a generalization of theorem 38 in [ ek ] ) . &
( viii ) ( 2,5/2a+1;a+2;2/5 ) = 5/3a+5/3 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( ix ) ( a,1 - 3/2a;-3/2a;3/5 ) & + & = 0 , + , + , + . &
the special values obtained from ( x ) and ( xi ) are trivial . &
( xii ) ( -5/2a,1 - 3/2a;-3/2a;-3/2 ) & + & = 0 , & + 0 , & + 0 , & + 0 , & + , & + , & + . & & & ( xiii)(2,5/2a+1;3/2a+2;3/5 ) = 15/4a+5/2 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . & ( xiv ) ( 3/2a,1-a;3/2a+2;3/5 ) = 2^a-15 ^ -a ( 3a+2 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( xv ) ( 3/2a,5/2a+1;3/2a+2;-3/2 ) & + & = 2 ^ -5n-15 ^ 5n ( 2 - 15n ) , & + 2 ^ -5n-25 ^ 5n+1 ( -1 - 15n ) , & + 2 ^ -5n-35 ^ 5n+2 ( -4 - 15n ) , & + 2 ^ -5n-45 ^ 5n+3 ( -7 - 15n ) & + & ( the above are special cases of ( [ 1.6eb2 ] ) ) . &
( xvi ) ( 2,1-a;3/2a+2;-3/2 ) = 3/5n+1 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( xvii ) ( 2,1 - 3/2a;2 - 5/2a;5/3 ) = 5/2n+1 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( xviii ) ( -5/2a,1-a;2 - 5/2a;5/3 ) = ( -2)^n3 ^ -n-1(5n+3 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( xix ) ( -5/2a,1 - 3/2a;2 - 5/2a;5/2 ) & + & = 2 ^ -n-1 ( -3 ) ^n ( 5n + 2 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( xx ) ( 2,1-a;2 - 5/2a;5/2 ) = 5/3n+1 & ( the above is a special case of ( [ 1.5eb2 ] ) ) .
the special values obtained from ( xxi ) are trivial . &
( xxii ) ( -5/2a,1-a ;- a;-2/3 ) = 0 , + , & + . & + & & ( xxiii ) ( 3/2a,1-a ;- a;2/5 ) = 0 , + , & + . &
+ & the special values obtained from ( xxiv ) are trivial .
[ [ section-42 ] ] ( 2,5,5 - 2 ) + + + + + + + + + the special values obtained from ( 2,5,5 - 2 ) coincide with those obtained from ( 2,5,5 - 1 ) . in this case
, there is no admissible quadruple . in this case
, there is no admissible quadruple . in this case
, we have @xmath289 where @xmath254 and @xmath290 are solutions of @xmath291 and @xmath292 , respectively .
[ [ section-43 ] ] ( 0,6,6 - 1 ) + + + + + + + + + the special values obtained from ( 0,6,6 - 1 ) are contained in those from ( 1,2,2 - 1 ) and ( 0,2,2 - 1 ) .
[ [ section-44 ] ] ( 0,6,6 - 2 ) + + + + + + + + + the special values obtained from ( 0,6,6 - 2 ) except trivial values are special cases of ( [ f(1,b;2;x ) ] ) .
[ [ section-45 ] ] ( 0,6,6 - 3 ) + + + + + + + + + the special values obtained from ( 0,6,6 - 3 ) coincide with those obtained from ( 0,6,6 - 2 ) .
[ [ section-46 ] ] ( 0,6,6 - 4 ) + + + + + + + + + the special values obtained from ( 0,6,6 - 4 ) except trivial values are special cases of ( [ f(1,b;2;x ) ] ) .
[ [ section-47 ] ] ( 0,6,6 - 5 ) + + + + + + + + + the special values obtained from ( 0,6,6 - 5 ) coincide with those obtained from ( 0,6,6 - 4 ) . in this case
, we have @xmath294 [ [ section-48 ] ] ( 1,5,6 - 1 ) + + + + + + + + + & ( i ) ( a,5a-1/2;6a;-4 ) & + & = , & + , & + , & + 0 , & + , & + 0 . & + & & ( ii ) ( 5a , a+1/2;6a;-4 ) & + & = , & + 0 , & + , & + , & + 0 , & + . & + & & ( iii ) ( a , a+1/2;6a;4/5 ) & + & , + , & + . & & & ( iv ) ( 5a,5a-1/2;6a;4/5 ) & + & , + , & + . &
+ & & ( v ) ( a,5a-1/2;1/2;5 ) = , + , + , + 0 , + , + 0 .
+ & & ( vi ) ( 1 - 5a,1/2-a;1/2;5 ) = , + 0 , + 0 , + , + , + . + & & ( vii ) ( a,1 - 5a;1/2;5/4 ) & + & = , + , + 0 , + 0 , + , + . + & & ( viii ) ( 5a-1/2,1/2-a;1/2;5/4 ) & + & = , + , + 0 , + , + 0 , + .
+ & & ( ix ) ( a,1 - 5a;3/2 - 4a;-1/4 ) & + & = . & & ( x ) ( 3/2 - 5a , a+1/2;3/2 - 4a;-1/4 ) & + & = . & & ( xi ) ( a , a+1/2;3/2 - 4a;1/5)= & ( the above is a generalization of theorem 20 in [ ek ] ) . &
( xii ) ( 1 - 5a,3/2 - 5a;3/2 - 4a;1/5 ) & + & = . & & ( xiii ) ( 5a-1/2,1/2-a;4a+1/2;-1/4 ) & + & = . & & ( xiv ) ( 5a,1-a;4a+1/2;-1/4)= . & & ( xv ) ( 5a,5a-1/2;4a+1/2;1/5)= . & & ( xvi ) ( 1-a,1/2-a;4a+1/2;1/5)= & ( the above is a generalization of theorem 19 in [ ek ] ) . & ( xvii ) ( 1 - 5a,1/2-a;2 - 6a;-4 ) & + & = , + 0 , + 0 , + , + , + .
+ & & ( xviii ) ( 1-a,3/2 - 5a;2 - 6a;-4 ) & + & = , + , + , + , + 0 , + 0 . + & & ( xix ) ( 1 - 5a,3/2 - 5a;2 - 6a;4/5 ) & + & = , + , + . + & & ( xx ) ( 1-a,1/2-a;2 - 6a;4/5 ) & + & = , + , + . + & & ( xxi ) ( 5a , a+1/2;3/2;5 ) = , + 0 , + , + , + 0 , + . & & ( xxii ) ( 1-a,3/2 - 5a;3/2;5 ) = , + , + , + , + 0 , + 0 . + & & ( xxiii ) ( 5a,1-a;3/2;5/4 ) & + & = , + 0 , + , + , + 0 , + . + & & ( xxiv ) ( a+1/2,3/2 - 5a;3/2;5/4 ) & + & = , + , + , + , + 0 , + 0 . &
[ [ section-49 ] ] ( 1,5,6 - 2 ) , ( 1,5,6 - 3 ) , ( 1,5,6 - 4 ) + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the special values obtained from ( 1,5,6 - 2 ) , ( 1,5,6 - 3 ) and ( 1,5,6 - 4 ) coincide with those obtained from ( 1,5,6 - 1 ) . in this case , we have @xmath296 [ [ section-50 ] ] ( 1,6,6 - 1 ) + + + + + + + + + & ( i)(a,6a+1;6a;6/5 ) & + & = 0 , + , + , + , + , + .
+ & the special values obtained from ( ii ) and ( iii ) are trivial . &
( iv)(5a,6a+1;6a;6 ) & + & = 0 , + , + , + , + , + . & & ( v ) ( a,6a+1;a+2;-1/5 ) = 5 ^ 6a6 ^ -6a ( a+1 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( vi ) ( 2,1 - 5a;a+2;-1/5 ) = 5/6a+5/6 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( vii ) ( a,1 - 5a;a+2;1/6 ) = 5 ^ 5a6 ^ -5a ( a+1 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . & ( viii)(2,6a+1;a+2;1/6 ) = 6/5a+6/5 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( ix ) ( a,1 - 5a;-5a;5/6 ) & + & = 0 , + , + , + , + , + . &
the special values obtained from ( x ) and ( xi ) are trivial . &
( xii ) ( -6a,1 - 5a;-5a;-5 ) = 0 , + 0 , + 0 , + 0 , + 0 , + , + , + , + , + . & & ( xiii)(2,6a+1;5a+2;5/6)= 6(5a+1 ) & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( xiv ) ( 5a,1-a;5a+2;5/6 ) = 6 ^ -a ( 5a+1 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( xv ) ( 5a,6a+1;5a+2;-5 ) = 6 ^ 6n(1 - 30n ) , + 6 ^ 6n+1(-4 - 30n ) , + 6 ^ 6n+2(-9 - 30n ) , + 6 ^ 6n+3(-14 - 30n ) , + 6 ^ 6n+4(-19 - 30n ) & ( the above are special cases of ( [ 1.6eb2 ] ) ) . &
( xvi ) ( 2,1-a;5a+2;-5 ) = 5/6n+1 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( xvii ) ( 2,1 - 5a;2 - 6a;6/5 ) = 6n+1 & ( the above is a special case of ( [ 1.5eb2 ] ) ) . &
( xviii ) ( -6a,1-a;2 - 6a;6/5 ) = -5 ^ -n-1(6n+5 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( xix ) ( -6a,1 - 5a;2 - 6a;6 ) = ( -5)^n(6n+1 ) & ( the above is a special case of ( [ 1.6eb2 ] ) ) . &
( xx ) ( 2,1-a;2 - 6a;6 ) = 6/5n+1 & ( the above is a special case of ( [ 1.5eb2 ] ) ) .
the special values obtained from ( xxi ) are trivial . &
( xxii ) ( -6a,1-a ;- a;-1/5 ) = 0 , + . & & ( xxiii ) ( 5a,1-a ;- a;1/6 ) = 0 , + . &
the special values obtained from ( xxiv ) are trivial .
[ [ section-51 ] ] ( 1,6,6 - 2 ) + + + + + + + + + the special values obtained from ( 1,6,6 - 2 ) coincide with those obtained from ( 1,6,6 - 1 ) . in this case
, there is no admissible quadruple . in this case
, we have @xmath299 [ [ section-52 ] ] ( 2,4,6 - 1 ) , ( 2,4,6 - 2 ) + + + + + + + + + + + + + + + + + + + + the special values obtained from ( 2,4,6 - 1 ) and ( 2,4,6 - 2 ) coincide with those obtained from ( 1,2,3 - 1 ) .
[ [ section-53 ] ] ( 2,4,6 - 3 ) + + + + + + + + + & ( i ) ( a,2a-1/3;3a;-9 + 6 ) & + & = , + , + , + + 0 , + . & & ( ii ) ( 2a , a+1/3;3a;-9 + 6 ) & + & = , + , + , + 0 , + , + . & & ( iii ) ( a , a+1/3;3a;9/4 + 3/4 ) & + & = , + , + , + .
+ & & ( iv ) ( 2a,2a-1/3;3a;9/4 + 3/4 ) & + & = , + , + , + 0 , + , + , + 0 , + . + & & ( v ) ( a,2a-1/3;2/3;10 - 6 ) = & ( the above is a generalization of theorem 26 in [ ek ] )
. we derive ( v ) by using ( 131 ) in [ gour ] and connection formulae for the hypergeometric series(see ( 25)(44 ) in 2.9 in [ erd ] ) .
we find ( vi)@xmath232(v ) . &
( vii ) ( a,1 - 2a;2/3;2/3 - 2/9 ) = & ( the above is a generalization of theorem 40 in [ ek ] ) . we derive ( vii ) by using ( 130 ) in [ gour ] and connection formulae for the hypergeometric series(see ( 25)(44 ) in 2.9 in [ erd ] ) .
we find ( viii)@xmath232(vii ) . &
( ix ) ( a,1 - 2a;4/3-a;1/3 + 2/9 ) & + & = . & & ( x ) ( 4/3 - 2a , a+1/3;4/3-a;1/3 + 2/9 ) & + & = . & we derive ( ix ) and ( x ) using formula ( 129 ) in [ gour ] and connection formulae for the hypergeometric series ( see ( 25)(44 ) in 2.9 in [ erd ] ) . &
( xi ) ( a , a+1/3;4/3-a;-5/4 - 3/4 ) & + & = , + , + , + .
+ & & ( xii ) ( 1 - 2a,4/3 - 2a;4/3-a;-5/4 - 3/4 ) & + & = , + , + 0 , + , + , + , + , + 0 . & the special values obtained from ( xiii)-(xx ) are contained in the above . &
( xxi ) ( 2a , a+1/3;4/3;10 - 6 ) & + & = .
( the above is a generalization of theorem 27 in [ ek ] ) .
we derive ( xxi ) using formula ( 133 ) in [ gour ] and connection formulae for the hypergeometric series ( see ( 25)(44 ) in 2.9 in [ erd ] ) .
we find ( xxii)@xmath232 ( xxi ) . &
( xxiii ) ( 2a,1-a;4/3;2/3 - 2/9 ) & + & = ( the above is a generalization of theorem 39 in [ ek ] ) .
we derive ( xxiii ) using formula ( 132 ) in [ gour ] and connection formulae for the hypergeometric series ( see ( 25)(44 ) in 2.9 in [ erd ] ) .
we find ( xxiv)@xmath232 ( xxiii ) .
[ [ section-54 ] ] ( 2,4,6 - 4 ) + + + + + + + + + & ( i ) ( a,2a-1/3;3a;-9 - 6 ) & + & = , + , + , + + 0 , + . & & ( ii ) ( 2a , a+1/3;3a;-9 - 6 ) & + & = , + , + , + 0 , + , + . & & ( iii ) ( a , a+1/3;3a;9/4 - 3/4 ) & + & = , + , + , + , + . + & ( iv ) ( 2a,2a-1/3;3a;9/4 - 3/4 ) & + & = , + , + , + , + .
+ & & ( v ) ( a,2a-1/3;2/3;10 + 6 ) & + & = , + , + , + , + 0 , + . & ( the first case is identical to theorem 26 in [ ek ] ) .
we find ( vi)@xmath232(v ) . &
( vii ) ( a,1 - 2a;2/3;2/3 + 2/9 ) & + & = , + , + , + , + 0 , + & ( the first case is identical to theorem 40 in [ ek ] ) .
we find ( viii)@xmath232(vii ) . &
( ix ) ( a,1 - 2a;4/3-a;1/3 - 2/9 ) & + & = . &
( x ) ( a+1/3,4/3 - 2a;4/3-a;1/3 - 2/9 ) & + & = . &
( xi ) ( a , a+1/3;4/3-a;-5/4 + 3/4 ) & + & = . &
( xii ) ( 1 - 2a,4/3 - 2a;4/3-a;-5/4 + 3/4 ) & + & = .
the special values obtained from ( xiii)-(xx ) are contained in the above . &
( xxi ) ( 2a , a+1/3;4/3;10 + 6 ) & + & = , + , + , + 0 , + , + & ( the fifth case is identical to theorem 27 in [ ek ] ) . we find ( xxii)@xmath232(xxi ) . &
( xxiii ) ( 2a,1-a;4/3;2/3 + 2/9 ) & + & = , + , + , + 0 , + , + & ( the fifth case is identical to theorem 39 in [ ek ] ) .
we find ( xxiv)@xmath232(xxiii ) .
[ [ section-55 ] ] ( 2,4,6 - 5 ) + + + + + + + + + the special values obtained from ( 2,4,6 - 5 ) coincide with those ( 2,4,6 - 3 ) .
[ [ section-56 ] ] ( 2,4,6 - 6 ) + + + + + + + + + the special values obtained from ( 2,4,6 - 6 ) coincide with those ( 2,4,6 - 4 ) . in this case
, there is no admissible quadruple . in this case , we get @xmath302 [ [ section-57 ] ] ( 2,6,6 - 1 ) , ( 2,6,6 - 2 ) + + + + + + + + + + + + + + + + + + + + the special values obtained from ( 2,6,6 - 1 ) and ( 2,6,6 - 2 ) coincide those obtained from ( 1,3,3 - 1 ) . [ [ section-58 ] ] ( 2,6,6 - 3 ) , ( 2,6,6 - 4 ) +
+ + + + + + + + + + + + + + + + + + + the special values obtained from ( 2,6,6 - 3 ) and ( 2,6,6 - 4 ) coincide those obtained from ( 1,3,3 - 3 ) . in this case , there is no admissible quadruple . in this case , we have @xmath305 [ [ section-59 ] ] ( 2,8,6 - 1 ) + + + + + + + + + the special values obtained from ( 2,8,6 - 1 ) are evaluated in paragraphs ( 1,2,2 - 1 ) and ( 0,2,2 - 1 ) .
[ [ section-60 ] ] ( 2,8,6 - 2 ) , ( 2,8,6 - 3 ) + + + + + + + + + + + + + + + + + + + + the special values obtained from ( 2,8,6 - 2 ) and ( 2,8,6 - 3 ) coincide with those obtained from ( 1,4,3 - 2 ) . in this case , we have @xmath307 [ [ section-61 ] ] ( 3,3,6 - 1 ) + + + + + + + + + & ( i)(a , a-1/2;2a;4 ) = 1 , + -2 , + 1 , + 2 , + -1 , + -1 . & & ( ii)(a ,
a+1/2;2a;4 ) = 1 , + 0 , + -1 , + 1 , + -1 , + 0 . & & ( iii ) ( a , a+1/2;2a;4/3 ) = ( -3 ) ^-3n-3 , + -2 ( -3 ) ^-3n-1 , + ( -3 ) ^-3n-2 , + ( -3 ) ^-3n , + - ( -3 ) ^-3n-1 , + 0 . & & ( iv ) ( a , a-1/2;2a;4/3 ) = ( -3 ) ^-3n-3 , + 0 , + - ( -3 ) ^-3n-2 , + 2 ( -3 ) ^-3n-3 , + - ( -3 ) ^-3n-4 , + - ( -3 ) ^-3n-2 . & & ( v ) ( a , a-1/2;1/2;-3 ) = 2 ^ 6n , + -2 ^ 6n+3 , + 2 ^ 6n+4 , + 2 ^ 6n , + -2 ^ 6n+1 , + -2 ^ 6n+3 . & we find ( vi)@xmath232(v ) . &
( vii ) ( a,1-a;1/2;3/4)= 2 ( 1/6 ( 4a+1 ) ) . & & ( viii ) ( 1/2-a , a-1/2;1/2;3/4 ) =( 1/6 ( 4a+1 ) ) . & & ( ix ) ( a,1-a;3/2;1/4 ) = . & & ( x ) ( 3/2-a , a+1/2;3/2;1/4 ) = . & & ( xi ) ( a , a+1/2;3/2;-1/3 ) = . &
we find ( xii)@xmath232(xi ) . &
( xiii ) ( 1/2-a , a-1/2;1/2;1/4 ) =( 1/3 ( 2-a ) ) . & & ( xiv ) ( a,1-a;1/2;1/4 ) =( 1/3 ( 2-a ) ) . & & ( xv ) ( a , a-1/2;1/2;-1/3 ) = . &
we find ( xvi)@xmath232(xv ) , ( xvii)@xmath232(i ) , ( xviii)@xmath232(ii ) , ( xix)@xmath232(iii ) and ( xx)@xmath232(iv ) . &
( xxi ) ( a , a+1/2;3/2;-3 ) = , + 0 , + , + , + , + 0 . &
we find ( xxii)@xmath232(xxi ) . &
( xxiii ) ( a,1-a;3/2;3/4 ) = .
& & ( xxiv ) ( 3/2-a , a+1/2;3/2;3/4 ) = . & [ [ section-62 ] ] ( 3,3,6 - 2 ) , ( 3,3,6 - 3 ) , ( 3,3,6 - 4 ) , ( 3,3,6 - 9 ) , ( 3,3,6 - 10 ) , ( 3,3,6 - 11 ) , ( 3,3,6 - 12 ) + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the special values obtained from ( 3,3,6 - 2 ) , ( 3,3,6 - 3 ) , ( 3,3,6 - 4 ) , ( 3,3,6 - 9 ) , ( 3,3,6 - 10 ) , ( 3,3,6 - 11 ) and ( 3,3,6 - 12 ) coincide with those obtained from ( 3,3,6 - 1 ) .
[ [ section-63 ] ] ( 3,3,6 - 5 ) + + + + + + + + + & ( i ) ( a , a+3/2;2a-1;4 ) = - , + , + 0 , + , + - , + 0 . & & ( ii ) ( a-1,a-5/2;2a-1;4 ) = - , + , + , + , + - , + - . & & ( iii ) ( a , a-5/2;2a-1;4/3 ) = - , + - , + 0 , + , + , + . & & ( iv ) ( a-1,a+3/2;2a-1;4/3 ) = - , + , + - , + , + , + 0 . & & ( v)(a , a+3/2;7/2;-3 ) = , + , + 0 , + , + , + 0 .
+ & we find ( vi)@xmath232(v ) . &
( vii ) ( a,2-a;7/2;3/4 ) = . & & ( viii ) ( a+3/2,7/2-a;7/2;3/4 ) = . & & ( ix ) ( a,2-a;-1/2;1/4 ) = . & & ( x ) ( a-5/2,-a-1/2;-1/2;1/4 ) = . & & ( xi ) ( a , a-5/2;-1/2;-1/3 ) = . &
we find ( xii)@xmath232(xi ) . &
( xiii ) ( a+3/2,7/2-a;5/2;1/4 ) = . & & ( xiv ) ( a-1,1-a;5/2;1/4 ) = . & & ( xv ) ( a-1,a+3/2;5/2;-1/3 ) = . &
we find ( xvi)@xmath232(xv ) , ( xvii)@xmath232(i ) , ( xviii)@xmath232(ii ) , ( xix)@xmath232(iii ) , ( xx)@xmath232(iv ) . &
( xxi ) ( a-1,a-5/2;-3/2;-3 ) & + & = -2 ^ 6n+6 ( 6n+5 ) ( 6n+7 ) , + 2 ^ 6n+1 ( 6n+1 ) ( 6n+3 ) , + 2 ^ 6n+3 ( 6n+3 ) ( 6n+5 ) , + 2 ^ 6n-3 ( 6n-2 ) ( 6n-4 ) , + -2 ^ 6n+4 ( 6n+4 ) ( 6n+6 ) , + -2 ^ 6n+6 ( 6n+6 ) ( 6n+8 ) . &
we find ( xxii)@xmath232(xxi ) . &
( xxiii ) ( a-1,1-a;-3/2;3/4 ) = - ( 2a-1 ) ( 2a-3 ) ( 1/6 ( 4a-1 ) ) .
& & ( xxiv ) ( a-5/2,-a-1/2;-3/2;3/4 ) & + & = -1/8 ( 2a-1 ) ( 2a-3 ) ( 1/6 ( 4a-1 ) ) . &
[ [ section-64 ] ] ( 3,3,6 - 6 ) , ( 3,3,6 - 7 ) , ( 3,3,6 - 8 ) , ( 3,3,6 - 13 ) , ( 3,3,6 - 14 ) , ( 3,3,6 - 15 ) , ( 3,3,6 - 16 ) + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the special values obtained from ( 3,3,6 - 6 ) , ( 3,3,6,-7 ) , ( 3,3,6 - 8 ) , ( 3,3,6 - 13 ) , ( 3,3,6 - 14 ) , ( 3,3,6 - 15 ) and ( 3,3,6 - 16 ) coincide with those obtained from ( 3,3,6,5 ) . in this case , we have @xmath309 [ [ section-65 ] ] ( 3,4,6 - 1 ) + + + + + + + + + the special values obtained from ( 3,4,6 - 1 ) are evaluated in paragraphs ( 1,2,2 - 1 ) and ( 0,2,2 - 1 ) . in this case
, we have @xmath311 [ [ section-66 ] ] ( 3,5,6 - 1 ) + + + + + + + + + & ( i ) ( a,5/3a-1/2;2a;-8 + 4 ) & + & = , + , + , + , + , + & & ( ii ) ( a,1/3a+1/2;2a;-8 + 4 ) & + & = , + , + , + , + . & & ( iii ) ( a,1/3a+1/2;2a;-8 - 4 ) & + & = , + , + , + . & & ( iv ) ( a,5/3a-1/2;2a;-8 - 4 ) & + & = , + , + , + , + , + 0 , + , + 0 . & & ( v ) ( a,5/3a-1/2;2/3a+1/2;9 - 4 ) & + & = . & & ( vi ) ( 1-a,1/2 - 1/3a;2/3a+1/2;9 - 4 ) & + & = & ( the above is a generalization of theorem 14 in [ ek ] ) . &
( vii ) ( a,1-a;2/3a+1/2;1/2 - 1/4 ) & + & = . & & ( viii ) ( 1/2 - 1/3a,5/3a-1/2;2/3a+1/2;1/2 - 1/4 ) & + & = . & & ( ix ) ( a,1-a;3/2 - 2/3a;1/2 + 1/4 ) & + & = , + , + , + , + , + . & & ( x ) ( 1/3a+1/2,3/2 - 5/3a;3/2 - 2/3a;1/2 + 1/4 ) & + & = , + , + , + , + 0 , + 0 . & & ( xi ) ( a,1/3a+1/2;3/2 - 2/3a;9 + 4 ) & + & = , + , + , + & ( the fourth case is identical to theorem 15 in [ ek ] ) . &
( xii ) ( 1-a,3/2 - 5/3a;3/2 - 2/3a;9 + 4 ) & + & = , + , + , + , + , + , + 0 , + 0 . & & ( xiii ) ( 1/2 - 1/3a,5/3a-1/2;2/3a+1/2;1/2 + 1/4 ) & + & = , + , + , + 0 , + , + 0 . & & ( xiv ) ( a,1-a;2/3a+1/2;1/2 + 1/4 ) & + & = , + , + , + , + , + . & & ( xv ) ( a,5/3a-1/2;2/3a+1/2;9 + 4 ) & + & = , + , + , + , + , + 0 , + , + 0 . & & ( xvi ) ( 1-a,1/2 - 1/3a;2/3a+1/2;9 + 4 ) & + & = , + , + , + & ( the fourth case is identical to theorem 14 in [ ek ] ) . &
( xvii ) ( 1-a,1/2 - 1/3a;2 - 2a;-8 + 4 ) & + & = , + , + , + , + . & & ( xviii ) ( 1-a,3/2 - 5/3a;2 - 2a;-8 + 4 ) & + & = , + , + , + , + . & & ( xix ) ( 1-a,3/2 - 5/3a;2 - 2a;-8 - 4 ) & + & = , + , + , + , + , + , + 0 , + 0 . & & ( xx ) ( 1-a,1/2 - 1/3a;2 - 2a;-8 - 4 ) & + & = , + , + , + . & & ( xxi ) ( a,1/3a+1/2;3/2 - 2/3a;9 - 4 ) & + & = & ( the above is a generalization of theorem 15 in [ ek ] ) . &
( xxii ) ( 1-a,3/2 - 5/3a;3/2 - 2/3a;9 - 4 ) & + & = . & & ( xxiii ) ( a,1-a;3/2 - 2/3a;1/2 - 1/4 ) & + & = . & & ( xxiv ) ( 1/3a+1/2,3/2 - 5/3a;3/2 - 2/3a;1/2 - 1/4 ) & + & = . &
[ [ section-67 ] ] ( 3,5,6 - 2 ) , ( 3,5,6 - 3 ) , ( 3,5,6 - 4 ) + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the special values obtained from ( 3,5,6 - 2 ) , ( 3,5,6 - 3 ) , ( 3,5,6 - 4 ) coincide with those obtained from ( 3,5,6 - 1 ) . in this case , we have @xmath313 [ [ section-68 ] ] ( 3,6,6 - 1 ) + + + + + + + + + the special values obtained from ( 3,6,6 - 1 ) are evaluated in paragraphs ( 1,2,2 - 1 ) and ( 0,2,2 - 1 ) . in this case
, there is no admissible quadruple . in this case
, we have @xmath316 [ [ section-69 ] ] ( 3,8,6 - 1 ) + + + + + + + + + the special values obtained from ( 3,8,6 - 1 ) are evaluated in paragraphs ( 1,2,2 - 1 ) and ( 0,2,2 - 1 ) . in this case , we have @xmath318 [ [ section-70 ] ] ( 3,9,6 - 1 ) , ( 3,9,6 - 2 ) + + + + + + + + + + + + + + + + + + + + the special values obtained from ( 3,9,6 - 1 ) and ( 3,9,6 - 2 ) coincide with those obtained from ( 1,3,2 - 1 ) .
in this appendix , we give some identities for generalized hypergeometric series and appell - lauricella hypergeometric series using the method introduced in section 1 .
many identities for generalized hypergeometric series @xmath319 have been discovered by making use of various methods .
for example , see `` hypergeometric database '' in 36 page , table 6.1 in 108 page and table 8.1 in 158 page of [ ko ] .
the author feels that applying the method introduced in section 1 of this paper to generalized hypergeometric series , we can obtain all identities appearing in those tables of [ ko ] and , of course , also new identities .
so , the author is planning to tabulate hypergeometric identities for generalized hypergeometric series obtained from our method . here
, we only rediscover some identities appearing in tables of [ ko ] , since the purpose of this article is not to tabulate hypergeomeric identities for generalized hypergeometric series but to tabulate those for the hypergeometric series .
now , we apply our method to generalized hypergeometric series @xmath320 .
it is known that for a given quintuple of integers @xmath321 , there exists a unique triple of rational functions @xmath322 satisfying @xmath323 this relation is called the contiguity relation of @xmath320 ( or the four term relation of @xmath320 ) .
we remark that coefficients @xmath324 , @xmath325 , @xmath326 can be exactly computed for a given @xmath327 . as to a calculation method of these coefficients , for example ,
see [ ta1 ] . from ( [ four_term_g ] )
, we have @xmath328 where @xmath329 for @xmath330 .
let sextuple @xmath331 satisfy @xmath332 then , for this sextuple , we have the first order difference equation @xmath333 or @xmath334 finally , solving ( [ diff_eq_g ] ) or ( [ diff_eq_g1 ] ) , we will yield hypergeometric identity for @xmath320 .
we summarize the whole process : thus , the method introduced in the main part of section 1 has broad utility . although stated in section 1 , we remark once again that implementing step 2 is equivalent to solving polynomial systems in @xmath336 .
we now do the above process for two quintuple of integers @xmath337 to give examples for the generalized hypergeometric series obtained from our method . when @xmath338 , @xmath339 is given as one of solutions of the system ( [ case_g ] ) .
in this case , we have @xmath340 by solving this equation ( [ ex_eq_g1 ] ) , we obtain @xmath341 this was discovered by dixon ( see no.4 of `` hypergeometric database '' in page 36 of [ ko ] ) .
moreover , remarking that @xmath342 equation ( [ ex_eq_g1 ] ) degenerates into @xmath343 ( the equation with equivalent to ( [ red_ex_eq_g1 ] ) also appears in the process of getting formulae ( 1,2,2 - 1)(v ) ) . by solving this equation ( [ red_ex_eq_g1 ] ) , we can find formulae which are equivalent to ( 1,2,2 - 1)(v ) .
we also note that taking @xmath344 in formula ( [ ex_g1 ] ) , the first case of ( 1,2,2 - 1)(v ) is obtained . when @xmath345 , we yield @xmath346 as one of solutions of the system ( [ case_g ] ) . in this case , we have @xmath347 solving this equation ( [ ex_eq_g1 ] ) , we yield @xmath348 this formula is due to watson and whipple ( see no.5 of `` hypergeometric database '' in page 36 of [ ko ] ) . moreover , paying attention to @xmath349 equation ( [ ex_eq_g2 ] ) degenerates into @xmath350 ( the equation with equivalent to ( [ red_ex_eq_g2 ] ) also appears in the process of getting formulae ( 1,2,2 - 1)(viii ) ) . solving this equation ( [ red_ex_eq_g2 ] )
, we can find formulae which are equivalent to ( 1,2,2 - 1)(viii ) .
we also remark that taking @xmath351 in formula ( [ ex_g2 ] ) , the first case of ( 1,2,2 - 1)(viii ) is obtained .
hypergeometric identities for appell - lauricella hypergeometric series have not been studied yet , while many transformation formulae for these series have been given ( for example , see formulae ( 8.3.1)(8.4.1.2 ) in [ sl ] ) .
the reason comes from the fact that it is difficult to use previously known methods for these series , because these are multiple series .
however , our method can be applied to these series , because these have contiguity relations like generalized hypergeometric series .
therefore , our method will become a powerful tool for investigating hypergeometric identities for appell - lauricella hypereometric series .
we now apply our method to appell hypergeometric series @xmath353 we discuss the detail of applying our method to @xmath354 , although we perform in much the same way as @xmath320 case .
it is known that for a given quadruple of integers @xmath355 , there exists a unique triple of rational functions @xmath356 satisfying @xmath357 this relation is called the contiguity relation of @xmath354 ( or the four term relation of @xmath354 ) . as to a calculation method of coefficients for the above relation , for example , see [ ta1 ] and [ ta2 ] .
we define @xmath358 by @xmath359 let sextuple @xmath360 satisfy @xmath361 then , for this sextuple , we have the first order difference equation @xmath362 or @xmath363 solving ( [ diff_eq_a ] ) or ( [ diff_eq_a1 ] ) as the final step , we will find hypergeometric identity for @xmath354 .
we summarize the whole process : when @xmath366 , we have @xmath367 as one of solutions of the system ( [ case_a ] ) . in this case
, the equation ( [ diff_eq_a1 ] ) becomes @xmath368 by solving this equation ( [ ex_eq_a1 ] ) , we find @xmath369 pay attention to the fact that @xmath370 from the fact , we can degenerate formula ( [ ex_a1 ] ) into the following two hypergeometric identities for @xmath371 : @xmath372 formulae ( [ ex_a1_d1 ] ) and ( [ ex_a1_d2 ] ) are equivalent to formulae ( 1,2,3 - 1)(x ) and ( 1,3,3 - 3)(viii ) , respectively .
hence , we can regard ( [ ex_a1 ] ) as a fusion of hypergeometric identities for the hypergeometric series ( 1,2,3 - 1)(x ) and ( 1,3,3 - 3)(viii ) . when @xmath373 , we have @xmath374 as one of solutions of the system ( [ case_a ] ) . in this case
, the equation ( [ diff_eq_a1 ] ) becomes @xmath375 by solving this equation ( [ ex_eq_a2 ] ) , we find @xmath376 in a similar manner to example 1 , we can degenerate formula ( [ ex_a2 ] ) into the following two hypergeometric identities for @xmath371 : @xmath377 here , we assume that @xmath378 in formula ( [ ex_a2_d1 ] ) , because we have extended @xmath0 as ( [ extend_hgf ] ) . formulae ( [ ex_a2_d1 ] ) and ( [ ex_a2_d2 ] ) are equivalent to the above case of formula ( 1,3,3 - 1)(xix ) and formula ( 1,3,3 - 1)(xii ) , respectively .
hence , we can regard ( [ ex_a2 ] ) as a fusion of hypergeometric identities for the hypergeometric series ( 1,3,3 - 1)(xix ) and ( 1,3,3 - 1)(xii ) , while we are not able to obtain the case below of formula ( 1,3,3 - 1)(xix ) from equation ( [ ex_eq_a2 ] ) or formula ( [ ex_a2 ] ) . in a similar manner , we can make use of our method for appell series @xmath379 , @xmath380 , @xmath381 and lauricella series @xmath382 , @xmath383 , @xmath384 , @xmath385 .
we have never investigated hypergeometric identities for appell - lauricella series , whereas we have advanced the study of those for generalized hypergeometric series so far .
this reason is due to the difficulty in applying previously known methods to the former series directly .
however , our method can be applied to appell - lauricella series as well as generalized hypergeometric series .
the author feels worthy enough to launch the study of hypergeometric identities for appell - lauricella series from the point of view of fusions ( or generalizations ) of those identities for the hypergeometric series .
kr g.e.andrews , r.askey and r.roy , _ special functions _ , cambridge university press , cambridge , 1999 .
g.e.andrews , _ applications of basic hypergeometric functions _ , siam review , * 16 * 1974 , no . * 4 * , 441484 .
m.apagodu and d.zeilberger , _ searching for strange hypergeometric identities by sheer brute force _ , integers , * 8 * ( 2008 ) , a36 , 6 pp .
w.n.bailey , _ generalized hypergeometric series _ , cambridge mathematical tract no .
32 , cambridge university press , ( 1935 ) .
a.ebisu , _ three term relations for the hypergeometric series _ , funkcial .
ekvac . , * 55*(2012 ) , no . * 2 * , 255283 .
a.ebisu , _ on a strange evaluation of the hypergeometric series by gosper _ , ramanujan j. , * 32(2013 ) * no . * 1 , 101108*. s.b.ekhad , _ forty `` strange '' computer - discovered[and computer - proved(of course ) ] hypergeometric series evaluations _ , the personal journal of ekhad and zeilberger , + http://www.math.rutgers.edu/~zeilberg/mamarim/mamarimhtml/strange.html , ( 2004 ) .
a.erdlyi ( editor ) , _ higher transcendental functions vol.1 _ , mcgraw - hill , ( 1953 ) .
l.euler , _ institutiones calculi differentialis cum eius usu in analysi finitorum ac doctrina serierum _ , typographeo petri galeatii , 1755 : available at http://www.math.dartmouth.edu/~euler/docs/originals/e212sec2ch4.pdf .
c.f.gauss , _ disquisitiones generales circa seriem infinitam _
ii , werke , * 3 * , 123162 , ( 1812 ) .
i.gessel , _ finding identities with the wz method _ , j.symbolic computation , * 20*(1995 ) , 537566 .
r.w.gosper , _ a letter to d. stanton _ , xerox palo alto research center , 21 december 1977 .
h.w.gould , _ combinatorial identities _ ,
morgantown , west virginia , 1972 .
.goursat , _ sur lquation diffrentielle linaire , qui admet pour intgrale la srie hypergomtrique _ ,
cole norm .
( * 2 * ) * 10*(1881 ) , 3142 .
i.gessel and d.stanton , _ strange evaluations of hypergeometric series _ , siam j. math .
, * 13*(1982 ) , no .
* 2 * , 295308 .
k.iwasaki , h.kimura , s.shimomura and m.yoshida , _ from gauss to painlev a modern theory of special functions _ , vieweg verlag , wiesbaden(1991 ) .
p.w.karlsson , _ on two hypergeometric summation formulas conjectured by gosper _ , simon stevin * 60 * ( 1986 ) , no .
4 , 329337 .
w.koepf , _ hypergeometric summation an algorithmic approach to summation and special function identities , second edition _ ,
universitext , springer(2014 ) .
m.petkovek , h.wilf and d.zeilberger , _ a = b _ , a.k.peters , wellesley(1996 ) .
l.j.slater , _ generalized hypergeometric functions _ , cambridge university press , cambridge(1966 ) .
n.takayama , _ grbner basis for rings of differential operators and applications _ , in _ grbner bases : statistics and software systems _ , 279344 , springer(2013 ) .
n.takayama , _ grbner basis and the problem of contiguous relations _ , japan j. appl . math . *
6*(1989 ) , no .
* 1 * , 147160 . r.vidunas , _ algebraic transformations of gauss hypergeometric functions _ , funkcial .
, * 52*(2009 ) , no . * 2 * , 139180 . | in this paper , we present a new method for finding identities for hypergeoemtric series , such as the ( gauss ) hypergeometric series , the generalized hypergeometric series and the appell - lauricella hypergeometric series .
furthermore , using this method , we get identities for the hypergeometric series @xmath0 ; we show that values of @xmath0 at some points @xmath1 can be expressed in terms of gamma functions , together with certain elementary functions .
we tabulate the values of @xmath0 that can be obtained with this method .
we find that this set includes almost all previously known values and many previously unknown values .
key words and phrases : hypergeometric series , three term relation , special value , solving polynomial systems .
2010 mathematics subject classification numbers : primary 33c05 secondary 13p10 , 13p15 , 33c20 , 33c65 . |
claims of room temperature ferromagnetism ( fm ) have been made for several semiconducting oxides that do not contain any magnetic ions . the phenomenon has been observed in a variety of binary oxides prepared in nanoparticle@xcite and thin - film form @xcite .
there has been considerable interest in these otherwise non - magnetic oxides as a part of the effort to develop suitable materials for spintronic devices .
this work is concerned with the origin of room temperature magnetic hysteres is observed in single crystal srtio@xmath0 ( 001 ) substrates ( sto ) purchased from a variety of suppliers .
the hysteresis , which can not be explained by non - interacting paramagnetic impurities , is interesting in light of recent work which highlights the considerable potential of sto surfaces for electronic and spintronic devices .
consider , for example , the two dimensional electron gas formed at the interfaces of laalo@xmath0/sto heterostructures @xcite as well as other phenomena such as blue luminescence @xcite and high conductivity @xcite produced by ar@xmath5 irradiation .
the observation of fm hysteresis in commercial sto raises several questions , which include whether the hysteresis is a bulk or surface effect , whether or not it can be associated with an impurity phase , or whether or not it can be associated with a particular kind of point or extended defect that can be controlled through materials processing .
it has been established that the surfaces of sto can be quite complex .
for example , slicing a verneuil grown sto crystal into substrates suitable for thin film growth produces a large concentration of dislocations and voids @xcite near the cut surface while annealing in oxidizing or reducing conditions can promote segregation into ti - rich or sr - rich regions in the surface skin layer@xcite . in the experiments described below ,
particular attention is paid to the effect of annealing in reducing and oxidizing atmospheres since it has been found in other oxides where fm has been observed , for example tio@xmath6@xcite and hfo@xmath6@xcite , that the magnitude of the remanent moment can be altered by vacuum annealing .
it is interesting to note that in sto oxygen vacancies donate free electrons to the hybridized ti(3d)-o(2p ) conduction band and can produce superconductivity with maximum t@xmath7 0.3 k @xcite .
the possibility of the coexistence of ferromagnetism and superconductivity is a compelling reason to study the magnetic properties of reduced sto .
finally , sto is a common substrate for metal oxide thin film growth . since certain thin film materials with interesting magnetic and charge ordering properties such as lavo@xmath0 @xcite
can only be produced in reducing atmospheres @xcite , it is important to be aware of how reduction at elevated temperatures affects the magnetic properties of the substrate .
sto ( 100 ) single crystal substrates were purchased from a variety of suppliers : alfa aesar , crystec , mti corporation and semiconductor wafer inc .
substrates were purchased in both one side polished ( 1sp ) and two side polished ( 2sp ) form .
extreme care was taken - for example always using teflon tweezers that had never been in contact with stainless steel - to avoid contamination of the samples with magnetic elements@xcite .
magnetic moment measurements were performed using a quantum design squid magnetometer .
some small degree of magnetic hysteresis at 300 k was measured in every single one of the over 50 substrates tested . to establish whether the hysteresis was a bulk or a surface effect , two sorts of tests were done .
hcl and hno@xmath0 acids were both found to reduce the size of the remanent moment while polishing the unpolished surfaces in stages down to mirror - like smoothness was found to completely remove all traces of fm hysteresis .
[ polishing](b ) compares measurements of moment versus field on sto substrates with one side polished ( 1sp ) - the as purchased state - with substrates with both @xmath8 mm surfaces polished ( 2sp ) as well as substrates with all six sides of the substrate polished ( asp ) .
each round of polishing removed a surface layer approximately 10 to 15 @xmath9 thick .
the hysteresis is clearly associated with the unpolished surfaces of the substrate .
the saturation magnetization can be extracted by subtracting out the diamagnetic contribution .
[ polishing](a ) summarizes the whole set of measurements of the saturation moment versus unpolished surface area .
energy dispersive x - ray spectroscopy ( edx ) and x - ray diffraction spectra ( xrd ) were measured on both the unpolished and polished surfaces of the sto substrates .
these measurements revealed no significant difference between the polished and unpolished surfaces except that the xrd lines were considerably wider for the stressed and disordered unpolished side of the substrates , as expected@xcite .
data for the unpolished surfaces are shown in fig .
[ edx_xrd ] which reveal no impurity phases or magnetic elements to the level of sensitivity of these techniques .
the peak near 3.7 kev indicates the presence of a small amount of calcium in these substrates .
there are at least two possibilities for the fm hysteresis of the unpolished surfaces .
firstly , it could be due to particles of fm material left from either the diamond saw or wire saw cuts used to produce substrates from the crystal boule , or from handling by the suppliers of the unpolished edges using stainless steel tweezers . at low enough concentration , these particles
would not be detected by the edx and xrd measurements illustrated in fig .
[ edx_xrd ] .
secondly , the hysteresis could be associated with extended or point defects such as oxygen vacancies in the vicinity of the cut surface . two different approaches to this question
were taken : ( i ) a systematic study was undertaken of the effect of annealing atmosphere and temperature on both the room temperature magnetic hysteresis as well as the width of the xrd lines .
the reasoning was that annealing should heal some of the disorder simultaneously reducing the width of the xrd lines and remanent moment .
in addition , annealing in vacuum allows one to introduce a controlled density of oxygen vacancies via the annealing temperature .
( ii ) artificial disorder was introduced by ar@xmath5 or ar@xmath10 bombardment of polished sto surfaces to see if a magnetic moment could be induced which would clearly be associated with disorder .
the goal was to find a way of introducing controlled amounts of disorder associated with surface ferromagnetism .
annealing of the sto substrates was performed at different temperatures in either a vacuum of approximately @xmath11 torr ( reducing environment ) or in air ( oxidizing environment ) by placing the substrate in an al@xmath6o@xmath0 crucible inside a quartz evacuation chamber in the furnace .
a pattern emerged which is illustrated in fig .
[ treatments ] .
annealing in air , for @xmath12c , destroyed or significantly decreased the room temperature remanent magnetization while annealing in vacuum for @xmath13c did not destroy the remanent magnetization . the vacuum annealing experiments produced samples with a wide range of oxygen vacancy density .
increasing the annealing temperature from 800@xmath1c to 1100@xmath1c systematically decreased the dc resistivity ( measured using the van der pauw technique ) and increased the dark colouration of the sto substrates .
it is possible to produce darker , even less resistive samples by either annealing the substrate along with powdered ti in a sealed , evacuated tube or by intentionally growing grossly oxygen deficient crystals @xcite .
sample magnetization data for one of the reduced substrates as well as that of a srtio@xmath14 crystal are shown in fig .
[ vary_reduction ] ( a ) .
note that data of fig .
[ vary_reduction ] ( a ) are for samples that have no unpolished surfaces .
the m vs. h relations of all the oxygen deficient samples were linear , while only two such curves are shown in fig .
[ vary_reduction ] ( a ) for simplicity .
this demonstrates clearly that oxgyen vacancies do not produce fm in sto .
the vacancy densities were not directly measured but there is a well known correlation between the room temperature dc resisitivity and free carrier concentration@xcite where , in the simplest picture , each oxygen vacancy donates two free electrons to the conduction band . in fig .
[ vary_reduction](b ) , the susceptibility of pure sto , which is due to the diamagnetic response of the closed shell ions plus a van vleck term@xcite , is @xmath15 emu / g - g and becomes less negative with the positive pauli contribution of an increasing number of free carriers .
this was the explanation given by frederikse and candela@xcite and the data illustrated in fig .
[ vary_reduction](b ) agrees to within 10% with their earlier measurements@xcite .
almost all of the samples included in fig .
[ vary_reduction](b ) have carrier concentrations compatible with low temperature superconductivity @xcite .
because the samples which have carrier density ( @xmath16@xmath4 ) compatible with superconductivity exhibit linear magnetic susceptibilities , it appears that sto is not a system where fm and superconductivity coexist .
having concluded that oxgyen vacancies do not produce the fm hysteresis , one might hypothesize that other kinds of disorder such as dislocations and interstitials produced by the diamond saw cuts may be associated with the fm hysteresis .
x - ray diffraction reveals that annealing does indeed relieve the strain on the unpolished surfaces as shown in fig .
[ linewidth](a ) which illustrates how the linewidth of the ( 003 ) diffraction peak from the unpolished side decreases with annealing temperature .
the line on the polished side is also illustrated , for comparison .
the decrease in linewidth with annealing temperature is seen in all diffraction peaks for both vacuum and air annealing . however , fig .
[ linewidth](b ) which is a plot of of the remanent moment versus linewidth of the x - ray diffraction peaks , suggests that the correlation between remanent moment and surface strain and disorder is not strong not only because of the scatter in the data but because air annealing and vacuum annealing have such different effects on magnetization but produce similar decreases in the linewidth of the x - ray diffraction peaks as mentioned earlier , ar ion bombardment was utilized to produce surface disorder in an attempt to mimic the disorder observed in the x - ray spectra of the unpolished surfaces .
this was done by taking samples with polished surfaces exhibiting linear magnetization versus applied field relations and mounting them with non - ferromagnetic adhesive onto 5-cm diameter si wafers and placing them in an ar ion beam to receive a dose of 10@xmath17 ions .
two beam energies were attempted : either 10 kev ( range 92 straggle 42 estimated using srim software ( stopping range of ions in matter@xcite ) ) or 90 kev ( range 570 straggle 250 ) .
a dose of 10@xmath17 90 kev ions produces a huge amount of damage , milling away part of the surface and producing an amorphous layer than can be observed using x - ray diffraction .
[ argon](b ) contrasts the ( 002 ) diffraction peaks from both irradiated and non - irradiated surfaces of the same substrate .
note the creation of a broad peak at lower diffraction angle which is associated with the amorphous layer of thickness roughly 60 nm . for comparison ,
diffraction peaks in amorphous si can have widths well over one degree @xcite .
note that the main ( 002 ) peak of the ar bombarded side of the crystal is narrower than that of the non - irradiated side . on the non - irradiated but polished surface ,
x - rays are scattered from both a surface strained layer plus a less strained region deeper in the crystal . on the irradiated side , the narrow peak represents diffraction from deeper in the crystal than the surface amorphous layer .
more importantly , a lack of magnetic hysteresis was measured after ion bombardment as shown in fig .
[ argon ] ( a ) for both incident ar ion kinetic energies .
it has been conclusively shown that fm is not associated with oxgyen vacancies , nor with an amorphous layer produced by ar ion bombardment while fig .
[ linewidth](b ) offers only a weak correlation between surface strain and disorder and the observed saturation moment .
it is difficult to prove the second hypothesis , in which the fm hysteresis of the unpolished surfaces is due to contamination by fm particles , because of the low particle density . recall that neither xrd nor edx spectra illustrated in fig .
[ edx_xrd ] showed any evidence of impurity elements or phases .
the approach taken here was to see whether artificially contaminated samples produced the same annealing behaviour as the raw substrates .
for the final set of experiments performed samples were prepared with no moment by soaking a 1sp substrate in nitric acid for 10 minutes in an ultrasonic bath as shown in fig .
[ spiked](a ) .
subsequently such samples were artificially contaminated with iron by rubbing the unpolished surface on stainless steel .
comparison of figs .
[ spiked](a),(b ) and ( c ) demonstrates that artificial contamination creates a moment of roughly the same magnitude as observed in the as purchased 1sp substrates .
finally the iron contaminated samples were subject to the same annealing treatments as untreated 1sp substrates .
comparison of fig .
[ treatments ] and fig .
[ spiked](b ) and ( c ) illustrates that the iron contaminated samples demonstrate the same annealing behaviour as the untreated 1sp substrates . a natural explanation of fig . [ treatments ] and fig
[ spiked ] comes from a comparison of the magnetic properties of iron and the iron oxides which are listed in table [ iron_oxides ] .
iron particles on the rough surfaces of the substrates coming from either diamond saw cuts or handling with stainless steel tweezers , are the source of the fm hysteresis observed on the unpolished surfaces .
below 900@xmath1c , vacuum annealing would either maintain these particles in the fe state , or at 1100@xmath1c there may oxidation to antiferromagnetic wustite feo whose presence is indicative of a highly reducing environment @xcite . on
the other hand air annealing would oxidize the iron particles to antiferromagnetic @xmath18fe@xmath6o@xmath0 at all temperatures .
@xmath18fe@xmath6o@xmath0 is more thermodynamically stable than @xmath19fe@xmath6o@xmath0 @xcite .
definite confirmation of the presence of iron on unpolished surfaces awaits detection by more sensitive surface elemental probes like pixe ( particle induced x - ray emission ) .
it is of interest to compare the present work with the only other systematic study of the ferromagnetic properties of commercial substrates by salzer _
et al . _ who considered sapphire ( al@xmath6o@xmath0 ) substrates purchased from crystec@xcite .
the authors studied the magnetic hysteresis curves of both 1sp and 2sp samples of a variety of orientations and sizes .
significantly they also tried to measure surface magnetic impurity concentration using pixe ( proton irradiated x - ray emission ) measurements .
the pixe measurements revealed that fe was the main impurity and that the average fe concentration on unpolished c - faces was 230 ng/@xmath20 while on polished c - faces the concentration was 10 ng/@xmath20 .
it is very interesting to note that a linear fit of fig .
[ polishing](a ) yields fe density of 190@xmath21 ng/@xmath20 if one assumes that the moment of the substrates is produced by pure fe with a room temperature magnetization of 220 emu / g .
there has been an on - going discussion surrounding ferromagnetism in magnetic ion free oxides .
on one side are those who argue that ferromagnetism is produced by defects in the thin flims@xcite , while on the other side are those who argue that it is a spurious effect produced by iron contamination@xcite . since some of the strongest claims linking oxygen vacancies and ferromagnetism have been made for tio@xmath22 thin films , it is useful to compare the tio@xmath22 data with the present measurements on single crystal srtio@xmath0 .
a summary of the properties of tio@xmath22 thin films grown by a number of techniques on a variety of substrates appears in table [ tio2_films ] .
the most notable feature of the data is that several groups have found that the saturation magnetization depends on the oxygen partial pressure during film deposition and have also observed that air annealing destroys the ferromagnetism .
all these authors have concluded that this is evidence that oxygen vacancies are producing the ferromagnetic hysteresis of the films and one has cited theoretical support@xcite for oxygen - vacancy - induced - ferromagnetism in tio@xmath6 .
however , no measurements of oxygen vacancy were actually made in these works .
on the other hand , in the present work , reasonable estimates of oxygen vacancy density can be inferred from the free carrier density measurements made and fig .
[ vary_reduction ] shows that sample magnetization is a linear function of applied field even in samples with a large density of bulk oxygen vacancies .
it seems very clear that oxygen vacancies do not produce ferromagnetism in sto .
the effect of vacuum annealing on sto and tio@xmath6 is quite similar where oxygen vacancies produce an increase in the electrical conductivity .
hence , the differences in ferromagnetism associated with oxygen vacancies in these two materials must be associated with differences in the ti - o network where sto contains a network of corner shared tio@xmath23 octahedra , while there is a combination of edge and corner sharing in rutile .
this is not the first work to have shown that commercial substrates without films themselves exhibit ferromagnetic hysteresis .
saturation moments for various substrates as measured by a number of groups are listed in table [ commercial_substrates ] .
it is interesting to note that aside from the anomalously large moment measured for laalo@xmath0 by hong _
et al._@xcite , there appears to be a rough correlation between substrate moment and the hardness of the substrate which are all harder than stainless steel .
it is difficult to compare the moments of the tio@xmath22 films with the substrates because usually it is the magnetization of the samples , rather than the raw moments that are listed in the papers .
however , it is curious that both yoon _ et al._@xcite and sudakar _ et al .
_ @xcite find that the magnetization to be a decreasing function of film thickness .
this data could be partially explained by iron contamination on the unpolished substrate surfaces .
it is also puzzling that sudakar _
_ measure magnetization of films produced by reactive rf sputtering to be ten times less than films produced by pulsed laser deposition .
they suggest that it is because the polycrystalline sputtered films on fused quartz may have less defects than the epitaxial pld films on laalo@xmath0 , but this does not seem convincing . a simpler explanation could be that different substrates have different amounts of surface contamination .
it is also interesting that yoon _
_ , found the curie temperature of tio@xmath22 thin film samples to be 880 k which is close to the transition temperature of @xmath19fe@xmath6o@xmath0 as shown in table [ iron_oxides ] .
the present annealing experiments illustrated in fig . [ treatments ] , which can be explained by the presence of iron on the unpolished surfaces of substrates , raise the possibility that the supposed dependence of saturation moment on oxygen partial pressure in the tio@xmath6 films might be the response of the substrates themselves .
ferromagnetic hysteresis with a saturation moment of the order of @xmath24 emu/@xmath20 is regularly observed in sto ( 001 ) substrates with 1 or 2 sides polished as purchased from a variety of suppliers .
oxygen vacancies do not appear to be associated with ferromagnetism in sto nor can a magnetic moment be created by ar ion bombardment .
reduced sto is not a system where fm and superconductivity coexist . in the as purchased samples ,
the hysteresis is associated with the unpolished surfaces and is likely due to fe contamination from handling with stainless steel tweezers or the diamond or wire saw cuts rather than oxygen vacancies or other lattice defects .
it is possible that all single crystal substrates used for thin film growth contain ferromagnetic iron or iron oxide on the unpolished surfaces .
polishing and simple acid washes can remove the weak ferromagnetic signal .
extreme care must be taken in handling samples to avoid contamination and to interpret magnetic data measured on thin films grown on sto substrates with unpolished surfaces .
one possible application of the results of this paper is for thin film growers using sto substrates . by annealing at high temperature before film growth , all traces of the ferromagnetic signal of the substrate
can be removed giving film growers confidence that any subsequent ferromagnetic signals are due to the films themselves .
it may also be of interest to further study iron implantation as a way to produce ferromagnetic surface layers in a controlled manner .
a. c. mofor , a. el - shaer , a. bakin , a. waag , h. ahlers , u. siegner , s. sievers , m. albrecht , w. schoch , n. izyumskaya , v. avrutin , s. sorokin , s. ivanov , j. stoimenos , appl .
87 * 062501 ( 2005 ) .
.properties of iron and iron oxides .
@xmath25 feo is typically formed in reducing atmospheres .
@xmath26the magnetization of @xmath27-fe@xmath6o@xmath0 does not saturate .
the value listed is for t=295 k and h=1 t , well above any fields used in this study . [
cols="^,^,^,^,^",options="header " , ] , which is for 1sp substrates 0.5 mm thick , the data are for 19 different samples from two different suppliers .
for the point where the unpolished surface area is approximately 0.2 @xmath20 which is for 2sp substrates 1 mm thick , the data are for 15 different samples from four different suppliers .
( b ) magnetization versus applied field at 300 k for a 5@xmath285 mm sto substrate polished on one side ( open squares marked 1sp ) , hand polished on both 5@xmath285 mm faces ( open triangles marked 2sp ) , polished on all six faces of the parallelpiped ( closed circles marked asp ) . ] and for a sto substrate that has been annealed at 1100c for 10 h in a vacuum of @xmath29 torr .
the m vs. h curves of substrates with no unpolished surfaces annealed between 800 and 1100 @xmath1c in vacuum are linear , but only one representative curve ( 1100@xmath1c ) is shown for clarity .
( b ) room temperature magnetic susceptibility of srtio@xmath30 as a function of free carrier concentration - measured via the room temperature dc resistivity as discussed in ref . - which in a first approximation is proportional to oxygen vacancy density .
points are included for a series of sto substrates annealed for 10 h in vacuum at 800,900,1000 , and 1100c .
the point marked ` ti ' is for a substrate annealed in a vacuum sealed quartz tube containing powdered ti .
the point marked srtio@xmath14 is for a grossly non - stoichiometric single crystal .
the solid horizontal line indicates the measured room temperature susceptibility of pure srtio@xmath0 . ]
torr at various temperatures or as purchased . except for the long dash curve which is for a polished side ,
all the spectra are for unpolished surfaces .
the data are similar for air - annealed samples .
lower panel ( b ) : remanent moment at 295k versus the full width at half maximum of the ( 003 ) line of the xrd spectrum taken on unpolished surfaces unnannealed and annealed ( air or vacuum ) at a variety of temperatures .
the uncertainty in moment is for a single measurement of the hysteresis curve on a single sample . ]
mm sto substrates .
the unfilled symbols show data before the acid or annealing treatment while the filled symbols show the data after the treatment .
panel ( a ) : effect of nitric acid treatment on as purchased substrates .
panel ( b ) : effect of annealing at 600 @xmath1c in a vacuum of @xmath11 torr on sto substrate artificially spiked with fe .
panel ( c ) : effect of high temperature air annealing ( 1100 @xmath1c ) on sto artificially spiked with fe . ] | room temperature ferromagnetic hysteresis is observed in commercial srtio@xmath0 substrates purchased from a variety of suppliers .
it is shown that the ferromagnetic signal comes from the unpolished surfaces .
surface impurity phases can not be detected using either x - ray diffraction or energy dispersive x - ray spectra on the unpolished surfaces . however , a possible correlation between surface disorder ( xray diffraction peak linewidth ) and ferromagnetism is observed .
ar ion bombardment ( 10kev-90 kev ) can be used to produce surface layer disorder but is not found to induce ferromagnetism .
annealing of the substrates at temperatures ranging from 600 to 1100 @xmath1c is found to alter the hysteresis curves differently depending on whether the annealing is performed in air or in vacuum .
identical annealing behaviour is observed if the substrates are artificially spiked with iron .
this suggests that the ferromagnetic hysteresis of as purchased srtio@xmath0 could be due to fe contamination of the unpolished surfaces .
in addition , it is shown that no ferromagnetism is observed in samples that contain 10@xmath2 - 10@xmath3 @xmath4 of oxygen vacancies if all the faces are polished . |
the recent discovery of an overdensity of stars in the color - magnitude diagram ( cmd ) of the large magellanic cloud ( lmc ) having nearly the same color as the `` red clump '' of core he - burning stars but extending @xmath10.9 mag brighter has been interpreted as an intervening population of stars at @xmath2 kpc that may represent a dwarf galaxy or tidal debris sheared from a small milky way satellite ( zaritsky & lin 1997 , hereafter zl ) .
zaritsky & lin label this overdensity the vrc ( vertical extension of the red clump ) , and reject other possible explanations to conclude that the vrc represents a massive foreground population with about 5% of angular surface density of the lmc itself .
if true , this conclusion would have profound consequences for the interpretation of galactic microlensing studies ( renault 1997 , alcock 1997a ) since such debris could , in principle , be responsible for a sizable fraction of the microlensing signal toward the lmc ( zhao 1996 , 1998 ) that is generally attributed to microlensing by compact objects in the smoothly - distributed halo of the milky way itself .
this particular stellar foreground population as an explanation for the lmc microlensing optical depth has been challenged on several grounds .
the macho team find no evidence for a foreground population at @xmath3 kpc in their extensive photometric database , confirming the lmc membership of their cepheids ( alcock 1997b , minniti 1997 ) .
they do find an overdensity of stars in a composite macho @xmath4 versus @xmath5 color - magnitude diagram ( cmd ) , but conclude that the _ redder _ color of this feature is incompatible with the hypothesis of a foreground clump population .
( the feature found by macho is unlikely to be the vrc , but rather another stage of stellar evolution associated with the asymptotic giant branch . )
gould ( 1997 ) argues on the basis of surface photometry of lmc performed by devaucouleurs ( 1957 ) that one of the following is true about any luminous foreground population : ( 1 ) it does not extend more than 5 from the lmc center , ( 2 ) is smooth on 15 scales , ( 3 ) has a stellar mass - to - light ratio 10 times that of known populations , or ( 4 ) provides only a small fraction of the microlensing optical depth . using a semi - analytic method to determine the phase space distribution of tidal debris , johnston ( 1998 )
has analyzed the zhao ( 1998 ) proposition , concluding that an ad hoc tidal streamer to explain the microlensing optical depth toward the lmc would cause unobserved overdensities of 10 - 100% in star counts elsewhere in the magellanic plane or would require disruption precisely aligned with the lmc within the last @xmath6 years .
bennett ( 1997 ) argues that a recently - determined stellar mass function combined with the assumption that the putative foreground population has a star formation history similar to the lmc results in an implied microlensing optical depth from the vrc that is only a small fraction of that determined by microlensing observations .
we will argue that the vrc feature observed by zl in color - magnitude diagrams of the lmc originates in the lmc itself .
using bvr ccd photometry of several fields at different locations in the lmc , we confirm the presence of substructure in lmc red clump morphology corresponding to the vrc .
in contrast to zl , however , we argue that the origin is likely to be due to stellar evolution , not an intervening population .
we begin by illustrating that the vrc is seen in all our fields . because the red clump morphology varies slightly in color and magnitude over the face of the lmc , interpretation of composite cmds
is complicated by the superposition of different features .
we therefore focus on individual lmc fields , overlaying isochrones and evolutionary tracks of the appropriate metallicity and age in order to demonstrate that the vrc corresponds precisely in magnitude and color to the so called `` blue loops '' experienced by aging intermediate - mass core he - burning stars .
we then show that similar red clump morphology is present in the cmd of hipparcos , which probes stellar populations on scales of @xmath7 pc from the sun , where intervening dwarf galaxies or tidal debris can not be invoked .
finally , we analyze the argument used by zl to reject stellar evolution as the cause of the vrc , and show that a more realistic model for the star formation history in the lmc is not only consistent with the vrc , but also provides a better fit to the data .
in january 1994 , bessel bvr photometry was performed with the danish 1.5 m telescope at eso la silla on the eros#1 , eros#2 and macho#1 microlensing candidates and a fourth field was taken far from the bar ; we will refer to these fields as f1 , f2 , f3 and f4 respectively .
the detector was a thinned , back - illuminated , ar - coated tektronix @xmath8 ccd with a nominal gain of 3.47 e-/adu , readout noise of 5.25 e- rms , and pixel size of @xmath9 corresponding to 0.38 on the sky .
the detector is linear to better than 1% over the whole dynamic range and is not affected by any large cosmetic defects .
observational and field characteristics are listed in table i. the cmd of these fields have been used to calibrate data obtained by the eros microlensing survey , further details can be found in beaulieu ( 1995 ) . we have performed a reanalysis of these bvr data with ( schechter , mateo & saha 1993 ) .
typical -reported errors on relative photometry are 0.02 mag at v = 19 ( typical for the clump stars ) for the cosmetically superior ( type 1 ) stars used throughout this analysis .
absolute calibration was performed using graham ( 1982 ) and vigneau & azzopardi ( 1982 ) .
foreground extinction was estimated using and iras maps ( schwering & israel 1991 ) ; these corrections are listed in table 1 for each field . beginning with this foreground extinction and assuming a metallicity of @xmath10 for the lmc and a helium abundance of @xmath11 , we then vary the internal extinction to achieve the best fit of the main sequence to the isochrones of bertelli ( 1994 ) . in fields f1 and f4 , this produces a total extinction equal to that of the foreground ( no internal reddening ) ; in fields f2 and f3 , the internal extinction results in an additional @xmath12 .
we will show that although correcting for extinction is important , the difference between the foreground reddening and the reddening determined from main sequence fitting does not affect our conclusions .
unless otherwise stated , we assume a distance modulus of @xmath13 for the lmc .
this choice of distance modulus produces the best fits to the isochrones along the main sequence and in the red clump .
a calibrated composite @xmath14 cmd for all four fields is shown in fig . 1 both with and without extinction corrections for the individual fields .
comparable cmd have been presented and analyzed by vallenari ( 1996 , and references therein ) .
the red clump is the most notable feature in the lmc other than the main sequence itself and is clearly visible in the composite cmd at about @xmath15 and @xmath16 ( @xmath17 ) .
first identified by cannon ( 1970 ) , red clump stars are the counterpart of the older horizontal branch ( hb ) in globulars , and represent a post helium - flash stage of stellar evolution ( for a review , see chiosi , bertelli & bressan 1992 ) .
differential reddening may be responsible for the elongated , tilted red clump in the dustier bar fields f1 and f3 , although differences in the age distribution and contamination by the red giant branch bump ( see 2.2.4 ) are also likely to play a role .
the morphology of the red clump itself may also be shaped by mass - loss ( @xcite , @xcite ) .
our analysis is aimed at understanding that portion of red clump morphology relevant to testing the hypothesis that the vrc is due to an intervening stellar population .
a narrow vertical extension of the clump having nearly the same color as the peak red clump density but extending up to @xmath10.8 mag brighter can also be seen .
a second source of substructure , or `` supra - clump , '' is apparent at a position @xmath10.8 mag brighter in @xmath18 and @xmath10.1 mag redder in @xmath19 than the peak of the primary red clump .
the positions of these features are marked in fig . 1 .
this substructure can be seen not only in the composite cmd , but also in the calibrated , extinction - corrected @xmath14 cmds for each of the individual four lmc fields presented in fig .
2 . fields f1 and f3 are located close to the lmc bar ; fields f2 and f4 further away at about @xmath20 from the center of the lmc . the magnitude and direction of the reddening vector is shown for each field .
examination of fig .
2 indicates that although the overall morphology of the cmd is the same in each field , field - to - field variations can be seen in the extent of the red clump and in relative stellar densities along the main sequence and between the main sequence and the red clump .
some of this variation is due to differences in crowding ; less severe crowding clearly results in deeper photometry for the f2 field , for example , and thus a larger number of main sequence stars .
since histograms reveal a small relative excess of bright main sequence stars in the outer compared to the inner lmc fields , the star formation history of our fields may also be somewhat different , complicating any analysis that rests on a composite cmd drawn from several regions of the lmc .
we therefore choose to analyze the fields independently . using contour plots of our four cmd
, we determine the position of the red clump at peak density and the extent of other substructures relative to the clump .
the results are summarized in table 2 .
the _ relative _ position of the vertical extension is remarkably constant in each of our fields and is also consistent with that found by zl : we find that the vertical extension has a @xmath5 color that varies no more than 0.03 mag from that of the primary clump peak and that it extends at least 0.85 mag brighter in @xmath18 than the red clump peak .
the second , redder substructure also maintains a constant relative position to the red clump from field to field . to within 0.02 mag in every field
, this supra - clump is 0.85 mag brighter in @xmath18 and 0.10 mag redder in @xmath21 than the peak density of the red clump . to test whether the relative stellar density in our cmd within the region of the vrc is consistent with that found by zl , we return to our composite and individual dereddened @xmath22 cmd . in color - magnitude space
we define the vertical extension ( @xmath23 , @xmath24 ) , the supra - clump ( @xmath25 , @xmath24 ) and primary clump ( @xmath26 , @xmath27 ) and count the stars within these regions .
these counts are summarized in table 3 , indicating that the vrc represents @xmath18% and the supra - clump @xmath114% of the primary clump in the composite cmd . in individual cmds ,
the vrc to red clump fraction vrc / rc varies from 6% ( in fields f1 and f3 ) to 15% in field f2 .
the supra - clump to red clump fraction sc / rc varies from 12% to 16% .
slightly different choices for the relevant boxes , for example narrowing them to reduce contamination from stars in other stages of stellar evolution , yield very similar fractions . since our estimate from the composite cmd for the vrc / rc fraction
has an uncertainty of @xmath28 from counting statistics alone , it is consistent with the zl estimate of @xmath29 for the relative projected angular surface density of the vrc . to summarize , the characteristics that we measure for vertical extension to the red clump and the redder supra - clump peak are identical within the uncertainties to those found by zl and the macho team respectively .
we therefore identify the vertical feature seen in our data with the vrc discussed by zl and the redder supra - clump with the overdensity discussed by alcock in the following section we propose stellar evolutionary origins for each of these features .
the position of red clump stars in a color - magnitude diagram depends on their mass , and thus their age , as they pass through this evolutionary stage . using the isochrones of bertelli ( 1994 ) , we plot in fig . 3 the mean locus of the core helium - burning phase for a variety of stellar ages in a @xmath30 cmd .
this time - averaged locus marks where stars in core helium - burning phase are likely to be found . we have used tracks with metallicity and helium abundance appropriate to the lmc , namely @xmath31 and @xmath32 . note that clump stars with ages in the range @xmath33 ( 0.4 to 1 gyr ) exhibit color changes smaller than @xmath34mag in @xmath35 while differing in @xmath18 magnitude by 0.79 mag .
we now compare these theoretical expectations with the vrc in our own lmc data .
we begin by focusing on the field f2 , since the low level of crowding in this outer field has resulted in the best photometry of our four fields .
the luminosity function for main sequence stars in this field peaks at about @xmath36 ; in the bar fields this occurs about 0.5 mag sooner .
the @xmath30 cmd for field f2 , dereddened by @xmath37 is shown in the upper panel of fig . 4 , for two different choices for the lmc distance modulus .
overplotted are theoretical isochrones from bertelli ( 1994 ) computed with new radiative opacities ( opal ) for metallicities appropriate to the lmc ( @xmath10 and @xmath11 ) and ages corresponding to 0.25 , 0.40 , 0.63 , 1.0 and 2.5 gyr ( log(age ) = 8.4 , 8.6 , 8.8 , 9.0 and 9.4 ) .
the lower panel enlarges the region of the cmd near the red clump region , which is now plotted as contours under the mean locus of the core he - burning phase from fig .
both the fit to the main sequence and the red clump is significantly improved for the smaller distance modulus of @xmath0 . since this improvement was apparent in all our fields , we fixed @xmath38 at this value .
using the same isochrones from bertelli ( 1994 ) , zl concluded that stars of age 2.5 gyr provided the best fit to the red clump morphology seen in their lmc data .
if , following zl , we do not apply an extinction correction , the 2.5 gyr isochrone does indeed provide a plausible fit to the red giant branch . with our extinction correction
, however , this isochrone falls at the very reddest ( or oldest ) edge of the red clump in field f2 . as fig .
5 illustrates , this is true for all of our fields .
note that if we had applied only foreground extinction corrections due to the milky way itself , the discrepancy with the 2.5 gyr isochrone would still be present : for two fields the internal extinction ( as determined by our main sequence fitting ) is negligible , for two others @xmath39 is increased by only 25 - 33% .
furthermore , younger isochrones fit the red clump of each field similarly despite the fact that different external extinction corrections ( based on estimates from and iras data ) were made .
we take this as an indication that these extinction corrections are reasonable and necessary for the proper interpretation of the stellar evolution .
the he - core burning phase is associated with the horizontal branch in old , metal - poor globulars , but in systems such as the lmc containing stars with a variety of ages and metallicities the horizontal branch becomes blurred into the red clump in cmd .
furthermore , as fig .
3 makes clear , the horizontal branch actually becomes _ vertical _ for core he - burning stars of intermediate masses and ages between @xmath10.4 and @xmath11.0 gyr .
stars of these masses experience the well - known `` blue loops '' ( see sweigart 1987 , chiosi 1992 ) that are caused by the increasing temperature of the outward expanding h - burning shell as the he - burning core gains mass . when the hydrogen in the shell is exhausted and helium begins to burn in the shell , the star moves redward again in the cmd to quickly join the asymptotic giant branch ( agb ) . while burning helium in their cores and hydrogen in their envelopes , stars spend most of their time near the bluest end of the blue loops .
the position of the most blueward extension of the loops for @xmath40 stars differs substantially in luminosity , but very little in color ( fagotto 1994 ) ; in cmd stars of these masses thus create a vertical extension brightward of the blue end of the red clump . as can be seen in fig .
5 , all of our fields contain main sequence stars as young as 250 myr .
since their lifetime on the main sequence is relatively short , intermediate mass stars should also be passing through the core he - burning phase .
indeed , stars with ages between @xmath10.4 0.8@xmath41gyr lie on a sequence of blue loops whose densely populated blue edge corresponds to the position of the vrc .
this is demonstrated clearly in fig . 6 , where the time - weighted centroid of core he - burning stars from fig . 3 is overplotted on contours of the stellar density for each of our fields .
the locus of core he - burning stars agrees with the position of the primary red clump and the vertical extension .
the agreement is remarkable , especially considering that we have not adjusted the metallicity of the isochrones to achieve a better fit .
we therefore identify the vrc with intermediate mass stars in the lmc currently undergoing core he - burning . in section
2.3 we show that the relative density of stars in the vrc is also consistent with a stellar evolutionary origin .
the spatial coincidence of the red and asymptotic giant branches with the supra - clump seen both in our data and in the composite cmd of the macho database ( alcock 1997b ) suggests that this second substructure in the cmd may be associated with giant evolution .
indeed , zaritsky & lin ( 1997 ) identify the supra - clump with the so - called `` red giant branch bump '' ( rgbb , see rood 1972 , sweigart , greggio & renzini 1989 , and fusi pecci 1990 ) . during the red giant branch ( rgb ) phase
, color and luminosity evolution pauses as the h - burning shell passes through a discontinuity left by the maximum penetration of the convective envelope .
this pause in the ascension results in an overdensity along the rgb . using an lmc distance modulus of 18.3 , the apparent magnitude of rgbb stars of 2@xmath42 and lmc metallicities ( sweigart , greggio & renzini 1989 ) agrees with the position and extent of the supra - clump in our data .
the same is true using the parameterized models of fusi pecci ( 1990 ) .
the position of the rgbb is however quite sensitive to stellar age ; older stars be fainter when in the rgbb phase . the mean age ( and metallicity ) of the stars
will thus determine the position of the rgbb for a mixed stellar population .
recent simulations based on isochrones from opal opacities indicate that the rgbb may lie at the same magnitude as the red clump itself in the lmc ( gallart & bertelli 1998 ) and thus significantly below the position of the observed supra - clump .
an alternative explanation for the supra - clump is offered by these simulations : a stalling of the evolution at the base of the asymptotic giant branch ( agb ) as the star struggles to reach thermal equilibrium after the helium exhaustion in the core ( gallart 1998 ) .
this early phase of agb evolution creates an over - density or bump ( denoted the agbb by gallart ) in the cmd since subsequent evolution proceeds much faster along the agb . both secondary features in the cmd are thus explained by stellar evolution : the vrc by young stars of intermediate - mass experiencing core he - burning and the supra - clump by a relatively long - lived phase of he - core exhausted stars at the base of the asymptotic giant branch .
the position of the vertical extension to the red clump corresponds to the end of the blue loops experienced by intermediate age stars in their he - core burning phase .
this argues against the need for an intervening population to explain the vrc , but is the relative stellar density of the vrc consistent with that expected for a stellar population with the star formation history of the lmc ?
comparisons of the stellar density in different parts of the cmd must be done with extreme care due to the selection effects induced by crowding .
in addition , comparing regions of the cmd that vary significantly in age increases the chances of error due to uncertainties in the star formation history and the increase of metallicity with decreasing age .
for example , comparing the stellar density of the vrc to that of the primary clump not only suffers confusion from the overlapping red giant branch , but also from the tracks of low metallicity old ( low mass ) clump stars that intersect those of higher metallicity younger ( higher mass ) clump stars .
for these all reasons , we choose to compare the stellar density of stars in the vrc with stars on the main sequence of comparable age and brightness .
our goal is to determine whether simple , reasonable assumptions about the imf and star formation history of the lmc can reproduce not only the position of the vrc , but its relative stellar density as well .
detailed models for the stellar density in a particular region of the cmd must combine stellar evolution with assumptions about the star formation history and the initial mass function ( imf ) of the population . the lmc is generally believed to be forming stars continuously and to have undergone more star formation in the last few gyr than previously ( see olszewski , suntzeff & mateo 1996 for a review ) , both on the basis of ground - based studies ( bertelli 1994 , girardi 1995 , and vallenari 1996 ) and deeper studies using hubble space telescope ( hst ) cmd ( gallagher 1996 , holtzman 1997 , elson 1997 , and geha 1997 ) . as fig .
7 demonstrates , in each of our four fields , a constant star formation rate with a salpeter imf ( salpeter 1955 ) described by @xmath43 with @xmath44 provides a good approximation to the relative densities of stars on the main sequence down to the completeness limit of @xmath118.5 19 .
field f4 appears to contain a slight excess of very bright main sequence stars ( fig . 5 & 7 ) , which may be indicative of a recent burst of star formation activity .
these results are consistent with ground - based and deeper hst studies of the lmc , although most recent work indicates that a slightly steeper imf provides a better match to the luminosity function at the expense of some relative density ratios ( @xcite , @xcite , @xcite ) . for the order of magnitude estimates presented here
, we will therefore assume that star formation has proceeded at some relatively constant and arbitrary rate between the epochs @xmath45 ( @xmath46 gyr ) , with @xmath47 . because we will count stars on the main sequence and in the vrc corresponding to the same initial mass
, our density analysis is actually independent of the imf as long as it remains unchanged over this same period .
only when we examine the effects of crowding will slope @xmath48 of the imf play a role . according to the models of bertelli ( 1994 ) , stars with initial masses between 2.6 and 3.0@xmath42
will be @xmath1500 myr old in the red clump where they will have the colors and magnitudes that place them in the center of the vrc .
regardless of its slope , if the imf is constant over the last gyr in the lmc , then the ratio of density of these stars in the vrc to their density on the main sequence will depend only on the ratio of lifetimes in these two phases . here ,
we define the lifetime in the vrc to be the length of time that these @xmath49 stars burn helium in their cores and have @xmath50 .
we use the evolutionary models of fagotto ( 1994 ) to estimate the lifetime on the main sequence , which we define as the time during which the fraction of hydrogen in the core is above 0.1% .
this definition is operationally convenient since it allows us to count all stars blueward of the subgiant branch . with these definitions , the ratio of vrc to main sequence lifetimes
and thus the ratio of corresponding vrc to main sequence counts is 23% . using these definitions , stars of initial mass in the range @xmath49 have @xmath51 and @xmath52 on the main sequence , and @xmath53 in the vrc where their colors are defined as above .
counts in these regions of the cmd yield vrc to main sequence fractions of 98% , 42% , 61% , and 46% for fields
f1 through f4 , respectively .
note that the bar fields appear to have the highest vrc fraction . however , since all of our fields are complete at the faintest end of the vrc , but none are complete at @xmath54 on the main sequence , these ratios are overestimated and must be corrected for incompleteness . to estimate the completeness of each field , we compare the actual number of stars found in the main sequence bin to that predicted by a salpeter imf normalized to provide good fits to the brighter portion of the main sequence histograms .
( for this purpose , main sequence was defined empirically to mean all stars with @xmath55 . )
the estimates of the completeness _
c@xmath56 _ at magnitudes where @xmath49 main sequence stars would be found are given in fig . 7 for each field ; they range from 42% to 61% .
it is difficult to estimate the uncertainty in these completeness estimates , but they are more dominated by counting statistics , which are typically on the order of 15% , than by choice of bins .
the resulting vrc to main sequence fractions , corrected for incompleteness , are 59% , 25% , 27% and 19% respectively ; all except bar field f1 are in 1.5@xmath57 agreement with the prediction of 23% based only on the assumption that the star formation rate and imf slope have remained constant over the last 1 gyr in the lmc .
the composite completeness - corrected vrc to main sequence ratio is 30% ; if the anomalously high field f1 is excluded , this drops to 25% .
the completeness correction assumed a salpeter slope of @xmath44 . a steeper imf would place more stars on the fainter part of the main sequence and
would thus result in larger completeness corrections and smaller vrc ratios .
the converse is true for shallow imf slopes .
how well does stellar evolution fare at predicting the 23% vrc to main sequence ratios for @xmath49 stars if the imf is modified ? redoing the completeness analysis with a shallow imf slope of @xmath58 yields a composite vrc to main sequence fraction of 38% . a steeper slope of @xmath59 , as preferred by many lmc luminosity function studies ( vallenari 1996 , and references therein , holtzman 1997 , and references therein ) , yields a smaller value of 21% , in excellent agreement with the predicted 23% .
finally , note that stars with masses of @xmath14 @xmath42 , corresponding to ages of 200 myr in the core he - burning phase evolve so quickly that they would not be expected to be detected in the clump region of the cmd .
this may explain the upper magnitude cutoff of the vrc . in conclusion ,
stellar evolution combined with a salpeter imf and constant star formation over the last 1 gyr can account for @xmath175% of the observed stellar density in the vrc .
for the somewhat steeper slopes preferred by most recent lmc studies , the fraction rises to 100% .
given the simplicity of the assumptions , counting procedure and incompleteness corrections , this agreement is remarkable and leaves little room for an intervening population of non - lmc stars in this region of the lmc color - magnitude diagram .
the parallaxes obtained by the hipparcos mission have allowed the determination of distances to stars brighter @xmath60 with accuracies on the order of 10% ( perryman 1997 ) .
this has resulted in a more accurately - determined color - absolute magnitude diagram for stars within @xmath1100 pc than has heretofore been possible , resulting in the first clear detection of `` clump giants '' in the solar neighborhood ( perryman 1995 ) .
paczyski & stanek ( 1998 ) conclude from their analysis that the hipparcos red clump has a mean distance of 105 pc and a geometrical mean distance of 98 pc , making it unlikely that extinction significantly affects the magnitude or color of the clump . in fig .
8 , we show the color - absolute magnitude diagram for 16229 stars from the hipparcos catalog with relative distances determined to within 10% and colors determined to within 2.5% ( perryman 1995 , and references therein ) .
the red clump is clearly visible as a highly concentrated collection of stars with a density peaking near @xmath61 and @xmath62 .
the fainter , redward tail of the clump seen most clearly in the contours may be due to the evolving core he - burning stars overlapping the giant branch ( pre - helium flash giants ) . also visible in the hipparcos cmd
is a vertical extension of the red clump toward brighter magnitudes ; its color is indistinguishable from that of the peak of the red clump . centered near @xmath63
, this feature is clearly discernible by eye at @xmath64 and @xmath65 .
its presence and location is robust to a variety of different choices for the smoothing and contouring of the cmd .
a second overdensity at similar magnitudes ( @xmath66 ) but redder colors ( @xmath67 ) is also apparent , and may correspond to the position of the agbb . the isochrones overplotted in the upper panel of fig .
8 assume a solar metallicity of @xmath68 for stars of ages 0.4 , 1.0 , 2.5 , 6.3 and 10.0 gyr .
it thus appears that the solar neighborhood has undergone some star formation in the last 400 myr , but that the bulk of the stars in the red clump in the solar neighborhood are older than those of the lmc .
the horizontal extent of the hipparcos red clump is due to older stars that begin burning helium in their cores at fainter , redder positions in the cmd and experience less severe blue loops . in the lower panel of fig . 8 ,
the area around the clump is enlarged and the stellar density plotted as contours .
the vrc in the solar neighborhood can be seen clearly .
both the isochrones overplotted in the upper panel of fig . 8 and
the theoretical locus of core he - burning stars ( of appropriate metallicity ) overplotted in the lower panel make it clear that the vrc seen in the hipparcos data is due to younger clump stars with ages between 300 myr and 1 gyr .
both our cmd and those of zl exhibit the same vertical extension to the red clump , yet we reach different conclusions as to the origin of the vrc despite using the same isochrones from bertelli ( 1994 ) with the same metallicity @xmath10 and helium abundance @xmath11 .
why do our conclusions differ ?
zl choose an isochrone with an age of log(age)=9.4 as the best match to the red giant morphology of their un - dereddened cmd .
they then note that although younger isochrones with log(age)=8.6 can reproduce the increased luminosity of 0.9 mag observed in the vrc , the evolutionary models of bertelli predict a difference in color of the mean locus of the core he - burning phase of @xmath69mag , @xmath70 mag between these two isochrones . since they observe a color difference @xmath71 in @xmath72 between the vrc and the centroid of the red clump , they dismiss stellar evolution as the origin of the vrc .
they further note that the maximum change in luminosity predicted by sweigart ( 1987 ) for stars evolving in the clump is 0.6 mag compared with the 0.9 mag that they observe in their cmd .
if the cmd are first dereddened before comparing to model isochrones , and the lmc is assumed not to be coeval but to have resulted from star formation that has been relatively constant over the last few gyr , one reaches a different conclusion .
fig . 3 shows how the mean @xmath18 magnitude of the core he - burning clump varies as a function of its @xmath73 color for stars of different ages using the models of bertelli ( 1994 ) .
indeed , as noted by zl , a measurable color difference is expected between stars of log(age)=8.6 and log(age)=9.4 . however , as demonstrated by our figs . 4 and 5 , stars as old as 2.5 gyr ( log(age)=9.4 ) do not fit the dereddened red clump at all , and for some of our fields appear to be too red to fit the bulk of red giant branch as well .
dereddening by even 0.12 mag ( comparable to the average _ foreground _ reddening of our lmc fields ) has a dramatic effect on the conclusions since this shifts the mean age of the clump to younger ages for which the luminosity of the blue loops is a very strong function of age , while the color does not change appreciably .
thus , allowing for a small range in the ages of young stars in the red clump is sufficient to reproduce the vertical extent of the vrc without significantly modifying its color .
note that this color insensitivity of the blue loops to age for stars between about 400 myr and 1 gyr is also true for the color defined by zl ( @xmath74 , as shown in fig . 9 for regions of the cmd comparable to that shown by zl in their fig . 1 .
zaritsky & lin note that if the foreground population hypothesis is correct for the origin of the vrc , then the entire lmc cmd should show traces of a parallel cmd shifted by 0.9 mag .
they can find no such traces in their own data , but do point to the hst color - magnitude diagram of holtzman ( 1997 ) as corroborating evidence that such a parallel feature may be present in the lower main sequence of the lmc .
they rightly note that binary stars would be a natural explanation for this excess of stars displaced by about @xmath75 mag from the primary main sequence , but dismiss such an explanation as unappealing since it can not simultaneously explain the vrc , where fine - tuning would be required to ensure that both members of the binary arrive in this region of the cmd at the same time .
zl conclude that if binaries are invoked as an explanation for the excess population in the lower main sequence , another explanation is necessary for the vrc .
while we make no claims as to the origin of a parallel main sequence in the lmc , we do proffer stellar evolution as the required alternate explanation for the vrc .
using bvr color magnitude diagrams obtained in four different regions in the lmc , we confirm the presence of a vertical extension of the red clump having the same color as the clump peak , but extending to brighter magnitudes , as first mentioned by zaritsky & lin ( 1997 ) . unlike zl , however , we conclude that this feature is due to stellar evolution , not a foreground population . our argument is based on noting that the feature ( 1 ) is present in all our lmc fields , ( 2 ) is in precise agreement with the time - averaged locus the blue loops that represent a relatively long - lived phase in the stellar evolution of younger core he - burning stars in the clump , ( 3 ) has a relative stellar density consistent with the density on the main sequence and the assumption of continuous and constant star formation history in the lmc over the past 1 gyr , and ( 4 ) is present in the solar neighborhood as demonstrated by the hipparcos color - magnitude diagram .
the difference in our conclusions despite using the same model isochrones ( bertelli 1994 ) rests on differences in our reddening corrections and assumptions about the ages of lmc stars . zl perform no reddening correction , whereas we correct for reddening using well - determined foreground estimates with a small correction for internal reddening based on main sequence fitting .
this results in an average inferred @xmath76 , only 0.02 mag greater than the average deduced for foreground extinction alone .
zl consider a single age of 2.5 gyr for the clump with one possible burst at 400 myr ; we consider a population more distributed in age and weighted toward younger ages than zl .
an isochrone of 2.5 gyr does not provide a good fit to the red clump in any of our fields , whereas the presence of younger stars in all our lmc fields is clearly demonstrated by comparison of isochrones with the position of the clump and the stellar density of the upper main sequence .
we find that stars of ages @xmath77 gyr are responsible for nearly the full extent of the vrc .
the best fit to both the main sequence and red clump of our cmd require a distance modulus to the lmc of @xmath0 , assuming that the agreement of evolutionary models to the cmd of solar neighborhood can be taken to imply that no large systematic offset in luminosity is present in the theory . with our justified assumption of continuous and constant star formation over the last gyr in the lmc ,
current understanding of stellar evolution predicts with precision not only the position of the vrc seen in both our data and those of zl , but also its relative stellar density .
the same stellar evolutionary models ( with appropriate assumptions for solar - metallicities ) reproduce the vrc feature in the solar neighborhood cmd measured by the hipparcos satellite .
these models when applied to the lmc with a salpeter imf can account for @xmath175% of stellar density of the vrc compared to that of the upper main sequence . a slightly steeper imf slope , such as preferred by many recent luminosity function studies of the lmc , reproduces the observed vrc stellar density exactly .
even with the salpeter imf , only one of our four fields ( f1 ) contains vrc stars 1.5@xmath57 in excess of expectations from stellar evolution and simple assumptions for the star formation history of the lmc .
since microlensing events are currently being reported both in the bar and outer regions of the lmc , the 4@xmath57 discrepant bar field alone can not contain a significant fraction of the microlensing optical depth to the lmc in the form of intervening stellar population .
indeed three of our four fields ( f1 , f2 and f3 ) contain a known microlensing candidate and thus have non - negligible optical depth to microlensing .
we conclude that no foreground population need be invoked to explain the presence of the vertical extension to the red clump in the lmc . if present , such a population is unlikely to account for more than @xmath125% of the vrc since the presence of intermediate mass lmc stars in the vrc are _ required _ by stellar evolutionary models in the observed numbers . whatever the primary source of the measured microlensing optical depth toward the lmc , it is unlikely to be due to a new foreground population that has made its presence evident in this vertical extension of the red clump .
we thank konrad kuijken for useful discussion and a prompt yet careful reading of the manuscript .
we are grateful to andrew cole for pointing out an error in an early version of this work and to carme gallart and alvio renzini for discussions about the agb .
we also thank eric maurice and louis prvot for advice on calibration procedures and richard naber for with subtleties .
this work is based on observations carried out at the european southern observatory , la silla .
work by pds while at the institute for advanced study in princeton was supported by the nsf grant ast- 92 - 15485 .
note added in proof : the red clump is an increasingly popular distance indicator ; two recent studies also find a smaller lmc distance ( udalski 1998 , stanek 1998 ) .
alcock ( the macho collaboration ) 1997a , , 486 , 697 alcock ( the macho collaboration ) 1997b , , 490 , l59 beaulieu j.p .
( the eros collaboration ) 1995 , , 299 , 168 bertelli , g. , bressan , a. , chiosi , c. , fagotto , f. & nasi , e. 1994 , , 106 , 275 bennett , d. 1997 , , 493 , l79 cannon , r. d. 1970 , , 150 , 111 chiosi , c. , bertelli , g. & bressan , a. 1992 , , 30 , 235 de vaucouleurs , g. 1957 , , 62 , 69 elson , r.a.w . ,
gilmore , g.f . &
santiago , b.x .
1997 , , 289 , 157 fusi pecci , f. , ferraro , f.r . ,
crocker , d.a . ,
rood , r.t .
& buonanno , r. 1990 , , 238 , 95 fagotto , f. , bressan , a. , bertelli , g. & chiosi , c. 1994 , , 105 , 29 [ gallagher 1997]galhst96 gallagher j. 1996 , , 466 , 732 [ gallart 1998]gallart98 gallart , c. 1998 , , 495 , l43 [ gallart & bertelli 1998]galbert98 gallart , c. & bertelli , g. 1998 , proceedings of the bonn / bochum graduiertenkolleg workshop `` the magellanic clouds and other dwarf galaxies , '' eds . t. richtler & j.m .
braun , in press .
[ geha 1997]geha97 geha , m.c .
1997 , preprint , astro - ph/9711144 girardi , l. , chiosi , c. , bertelli , g. & bressan , a. 1995 , , 298 , 87 gould , a. 1997 , preprint , astro - ph/9709263 [ graham 1982]gra82 graham j.a . 1982 , , 94 , 244 [ holtzman 1997]hol97 holtzman , j. 1997 , aj , 113 , 656 [ jimenez , flynn & kotoneva 1997]jim97 jimenez , r. , flynn , c. & kotoneva , e. 1997 , submitted to , also astro - ph/9709056 [ johnston 1998]kvj98 johnston , k.v .
1998 , , 495 , 297 minniti d. 1997 , in the proceedings of the uc santa cruz conference on _ galactic halos _ , ed .
d. zaritsky , to appear in olszewski , e.w . ,
suntzeff , n.b . & mateo , m. 1996 , , 24 , 511 paczyski , b. & stanek , k.z .
1998 , , 494l , 219 perryman 1995 , , 304 , 69 perryman 1997 , , 323 , l49 renault c. ( the eros collaboration ) 1997 , , 324 , l69 renzini , a. & fusi pecci , f. 1988 , , 26 , 199 rood , r.t . 1972 , 177 , 681 salpeter , e.e . 1955 , , 121 , 161 schechter p. , mateo m. & saha a , 1993 , , 105 , 1342 schwering , p.b.w .
& israel , f. 1991 , , 246 , 231 stanek , k.z . , zaritsky , d. & harris , j. 1998 , submitted to , astro - ph/9803181 .
sweigart , a.v .
, greggio , l. & renzini , a. 1989 , , 69 , 911 sweigart , a.v .
1987 , , 65 , 95 udalski a. , szymanski , m. , kubiak , m. , pietrzynski , g. , woziak , p. & zebrun , k. 1998 , acta astron . , in press , astro - ph/9803035 .
vallenari a. , 1996 , , 309 , 367 [ vigneau & azzopardi 1982]va82 vigneau j. & azzopardi m. 1982 , , 50 , 119 zaritsky , d. & lin , d.n.c .
1997 , , 114 , 2545 ( zl ) zhao , h .- s .
1996 preprint , submitted to , and astro - ph/9606166 zhao , h .- s .
1998 , , 294 , 139
fig . 1 calibrated , composite @xmath22 color - magnitude diagrams for all four lmc fields described in the text .
only the 16,139 stars with cosmetically superior point spread functions ( type 1 ) in @xmath18 and @xmath4 are plotted . both uncorrected ( left ) and dereddened ( right ) cmd
are shown .
positions of the red clump ( rc ) , vertical extension to the red clump ( vrc ) , and the supra - clump are indicated .
calibrated , dereddened @xmath22 color - magnitude diagrams for the individual four lmc fields .
only stars with cosmetically superior point spread functions in @xmath18 and @xmath4 are plotted .
the number of stars plotted is 3614 , 4241 , 5591 , 2693 in @xmath21 for fields f1 through f4 respectively .
centroid of the @xmath18-band luminosity of the core he - burning phase as a function of @xmath73 for stars of differing ages with lmc metallicity and helium abundance of z=0.008 and y = 0.25 ( bertelli 1994 ) .
note that no significant color change occurs for stars with ages between 400 myr and 600 myr ; these stars correspond to the bulk vrc in the lmc . a similar sequence for z=0.02 and y=0.28 ( appropriate to the solar neighborhood )
yields a @xmath78 color redder by @xmath79mag and is shown in fig .
8 . in the solar neighborhood ,
older stars are also present , causing the horizontal dispersion in color on the fainter edge of the red clump in the hipparcos data .
fig . 4 * top panel : * calibrated , dereddened @xmath30 color - magnitude diagram for our least crowded field f2 .
4666 stars with cosmetically superior point spread functions in @xmath80 and @xmath18 are shown .
isochrones with lmc metallicity ( @xmath10 ) and helium abundance ( @xmath11 ) with ages @xmath818.4 , 8.6 , 8.8 , 9.0 , and 9.4 from bertelli ( 1994 ) are shown superposed for two different assumptions for the distance modulus , @xmath38 to the lmc . *
bottom panel : * contour representation of the cmd density for field f2 in region of the red clump shown superposed on the mean locus of the core helium burning phase as a function of age as in fig .
separate scales for the abscissa of the top and bottom panels are indicated .
the region of cmd shown has been divided into 25 bins along both the magnitude and color axes . for this binning , contour levels displayed are 12 , 10 , 8 , 6 , 4 and 2.5 stars .
a shorter distance modulus of @xmath0 provides improved fits to the main sequence and red clump .
same as the bottom panel of fig 4 , but for all of our lmc fields .
the region of cmd shown has been divided into 25 bins along both the magnitude and color axes .
for this binning , contour levels displayed are 25 , 20 , 15 , 12 , 10 , 8 , 6 , 4 and 2.5 stars .
the highest contour levels are not always present in the outer fields .
7 histograms of main sequence stars for each of our fields fields are show on a logarithmic scale .
the number of stars in each magnitude bin expected from a salpeter imf , taking into account the main sequence lifetime and assumed a constant star formation rate over the last 1 gyr , is shown as the solid sloping line .
the normalization has been adjusted to achieve a good fit to the bulk of the main sequence stars between @xmath82 .
the faintest vrc and main sequence stars used for the counting arguments of 2.3.2 are indicated by the vertical lines at @xmath8318.5 and 19.5 , respectively .
the main sequence counts clearly suffer from some incompleteness , whereas the vrc counts do not .
comparison with expectations from the salpeter imf produce the main sequence completeness estimates _
c@xmath56 _ indicated for each field .
fig . 8 * top panel : * hipparcos @xmath30 color - magnitude diagram containing over 16000 stars with accurate distances and photometry .
a vertical extension can be seen clearly on the blueward side of the red clump with nearly the same color as the peak clump density and extending about a magnitude brighter .
the redder supra - clump is also visible .
isochrones with solar metallicity ( z = 0.02 ) and @xmath84 8.6 , 9.0 , 9.4 , 9.8 and 10.0 are shown overplotted .
note that positions of the blue loops again coincide with the vertical extension .
* bottom panel : * contour representation of the cmd density in region of the red clump for the solar neighborhood as measured by hipparcos is shown superposed on the mean locus of the core he - burning phase as a function of age for stars with metallicity and helium abundance appropriate to the solar neighborhood ( @xmath85 and @xmath86 ) .
separate scales for the abscissa of the top and bottom panels are indicated .
the region of cmd shown has been divided into 20 bins along both the magnitude and color axes . for this binning ,
contour levels displayed are 35 , 25 , 15 , 10 , 5 and 2.5 stars .
fig . 9 the same isochrones of lmc metallicity from fig .
3 are plotted in color - magnitude diagrams for the region of the red clump .
the color @xmath74 is that used by zaritsky & lin ( 1997 ) as are the bands on the vertical axis : u ( top ) , b ( middle ) , and i ( bottom ) respectively .
reddening vectors assuming an average @xmath87 are shown in each panel .
the blue loops correspond with the feature seen by zl at a dereddened color of @xmath88 . | we examine the morphology of the color - magnitude diagram ( cmd ) for core helium - burning ( red clump ) stars to test the recent suggestion by zaritsky & lin ( 1997 ) that an extension of the red clump in the large magellanic cloud ( lmc ) toward brighter magnitudes is due to an intervening population of stars that is responsible for a significant fraction of the observed microlensing toward the lmc . using our own ccd photometry of several fields across the lmc
, we confirm the presence of this additional red clump feature , but conclude that it is caused by stellar evolution rather than a foreground population .
we do this by demonstrating that the feature ( 1 ) is present in all our lmc fields , ( 2 ) is in precise agreement with the location of the blue loops in the isochrones of intermediate age red clump stars with the metallicity and age of the lmc , ( 3 ) has a relative density consistent with stellar evolution and lmc star formation history , and ( 4 ) is present in the hipparcos cmd for the solar neighborhood where an intervening population can not be invoked .
assuming there is no systematic shift in the model isochrones , which fit the hipparcos data in detail , a distance modulus of @xmath0 provides the best fit to our dereddened cmd .
_ accepted for publication in the astronomical journal _ |
in many concrete situations the statistician observes a finite path @xmath0 of a real temporal phenomena .
a common modeling is to assume that the observation is a finite path of a second order weak stationary process @xmath1 ( see , for example , @xcite ) .
this means that the random variable ( r.v . )
@xmath2 is , for any @xmath3 , square integrable and that the mean and the covariance structure of the process is invariant by any translation on the time index .
that is , for any @xmath4 @xmath5 does not depend on @xmath6 and @xmath7 only depends on the distance between @xmath6 and @xmath8 .
a more popular frame is the gaussian case where the additional gaussianity assumption on all finite marginal distributions of the process @xmath9 is added . in this case , as the multidimensional gaussian distribution only depends on moments of order one and two , the process is also strongly stationary .
this means that the law of all finite dimensional marginal distributions are invariant if the time is shifted : @xmath10 gaussian stationary process are very popular because they share plenty of very nice properties concerning their statistics or prediction ( see , for example , @xcite or @xcite ) . hence , an important topic in the field of stationary process is the implementation of a statistical procedure that allows to assess gaussianity . in the last three decades
, many works have been developed to build such methods .
for example , in @xcite a test based on the analysis of the empirical characteristic function is performed . in @xcite
based on the skewness and kurtosis test or also called jarque - bera test . in @xcite
based on both , empirical characteristic function and skewness and kurtosis . in @xcite
we can find another test , this based on the bispectral density function .
an important drawback of these tests is that they only consider a finite order marginal of the process ( generally the order one marginal ! ) .
obviously , this provides tests at the right level for the intended problem ; but these tests could be at the nominal power against some non - gaussian alternatives .
for example , in the case of a strictly stationary non - gaussian process having one - dimensional gaussian marginal . in this paper
, we propose a procedure to assess that a strictly stationary process is gaussian .
our test is consistent against every strictly stationary alternative satisfying regularity assumptions .
the procedure is a combination of the random projection method ( see @xcite and @xcite ) and classical methods that allow to assess that the one - dimensional marginal of a stationary process is gaussian ( see the previous discussion ) .
regarding the random projection method , we follow the same methodology as the one proposed in @xcite . roughly speaking ,
it is shown therein that ( only ) a random projection characterizes a probability distribution .
in particular , we employ the results of @xcite where the main result of @xcite is generalized to obtain goodness - of - fit tests for families of distributions , and in particular for gaussian families .
therefore , given a strictly stationary process , @xmath11 , we are interested in constructing a test for the null hypothesis @xmath12 .
notice that @xmath13 holds if , and only if , @xmath14 is gaussian . so that , using the random projection method , @xcite , this is , roughly speaking , equivalent to that a ( one - dimensional ) randomly chosen projection of @xmath14 is gaussian .
this idea allows to translate the problem into another one consisting on checking when the one - dimensional marginal of a random transformation of @xmath11 is gaussian .
this can be tested using a usual procedure . here
, we will employ the well - known epps test , @xcite , and lobato and velasco skewness - kurtosis test , @xcite .
we also use a combination of them as a way to alleviate some problems that those tests present . furthermore , epps test checks whether the characteristic function of the one - dimensional marginal of a strictly stationary process coincides with the one of a gaussian distribution .
this checking is performed on a fixed finite set of points . as a consequence
, it can not be consistent against every possible non - gaussian alternative with non - gaussian marginal .
however , in our work , the points employed in epps test will be also drawn at random .
this will provide the consistency of the whole test . regarding lobato and velasco skewness - kurtosis test
we will prove the consistency of the test under different hypothesis than those in @xcite .
the paper is organized as follows . in the next section
we will give some basic definitions and notations .
in section [ juan ] , we discuss some useful known results .
one concerns the random projection method , some gaussianity tests for strictly stationary processes and another a procedure for multiple testing .
it also contains a new result characterizing gaussian distributions . in section [ sectionmodelo ]
we introduce our procedure and analyze its asymptotic behavior .
section [ simulations ] contains some details on the practical application of the method and section [ simu ] includes the results of the simulations .
the paper ends with a discussion . in the whole paper
all the processes are assumed to be integrable .
if @xmath15 is a random variable , we denote by @xmath16 its characteristic function ; @xmath17 denotes the characteristic function of the gaussian distribution with mean @xmath18 and variance @xmath19 . denotes a separable hilbert space with inner product @xmath20 and norm @xmath21 .
@xmath22 denotes a generic orthonormal basis of @xmath23 and @xmath24 the @xmath25-dimensional subspace spanned by @xmath26 . for any subspace , @xmath27 we write @xmath28 for its orthogonal complement .
if @xmath29 is an -valued random element , then @xmath30 denotes the projection of @xmath29 on the subspace @xmath31 of .
@xmath32 and @xmath33 denote indistinctly a process . through the following , when we say that a process is stationary we mean that it is strictly stationary .
given a stationary process @xmath34 let us denote , if they exists , @xmath35 $ ] the mean and @xmath36,$ ] with @xmath37 the centered moment of order @xmath38 further , let @xmath39,$ ] with @xmath40 be the autocovariance of order @xmath6 .
let @xmath41 @xmath42 be a sample of equally spaced observations of the random process @xmath43 let @xmath44 be its sample mean , @xmath45 for @xmath37 its sample centered moment of order @xmath46 and @xmath47 for @xmath48 the sample autocovariance of order @xmath49 when it is clear to which process they are referring we suppress the subindex @xmath50 note that then we write @xmath51 as @xmath52
. for the sake of simplicity , let us denote @xmath53 and analogously @xmath54 finally , by i.i.d.r.vs .
we mean independent and identically distributed random variables .
we assume that all the random elements are defined on the same , rich enough , probability space @xmath55 .
in this section we discuss both a characterization of gaussian distributions in infinite dimensional spaces , a characterization of the one - dimensional gaussian distributions and two tests of gaussianity for stationary processes .
we also recall some facts on multiple testing procedure .
all this material are tools for our results . excluding the characterization of the one - dimensional gaussian distributions ( proposition [ theo1dimens ] ) , the results in this section are well known and they are included here for the sake of completeness .
the result of this subsection comes from @xcite .
it is based on the use of _ dissipative distributions _ which are defined next .
[ dede ] let @xmath29 be an @xmath23-valued random element .
we will say that its distribution is dissipative if there exists an orthonormal basis @xmath56 of @xmath57 such that 1 .
@xmath58 , for all @xmath59 ( see section [ snoba ] for the definition of @xmath24 ) .
2 . the conditional distribution of @xmath60 given @xmath61 is absolutely continuous with respect to the @xmath25-dimensional lebesgue measure .
theorem 3.6 in @xcite states the following : [ theogausianas ] let @xmath62 be a dissipative distribution on @xmath23 . if @xmath63 is an -valued random element and @xmath64 then @xmath65 is gaussian .
the importance of this result relies on the fact that if @xmath62 is dissipative then the following @xmath66 law holds @xmath67moreover , @xmath63 is not gaussian if , and only if , @xmath68 in other words , if we ask if the distribution of @xmath63 is gaussian , then the only thing we have to do is to select at random a point @xmath69 using a dissipative distribution and check if the real - valued random variable @xmath70 is gaussian . we will obtain the right answer with probability one .
we start this subsection by stating the definition of analytic characteristic function which has been taken from @xcite .
a characteristic function @xmath71 is said to be analytic if there exist * a complex valued function , @xmath72 , of the complex variable @xmath73 which is holomorphic in a circle @xmath74 where @xmath75 * a positive real number @xmath76 such that @xmath77 for @xmath78 that is , an analytic characteristic function is a characteristic function which coincides with a holomorphic function in some neighborhood of zero .
some properties on analytic characteristic functions may be found in @xcite . in particular , it is proved therein that the characteristic function of a gaussian distribution is analytic ( this is a well known fact ) .
some other well - known distributions having analytic characteristic function are the binomial , poisson and gamma distributions but not the cauchy one .
the following result will be useful to assess that our goodness of fit test will work with all non - gaussian alternatives .
[ theo1dimens ] let @xmath79 be a borel probability measure defined on .
assume that @xmath79 is absolutely continuous with respect to the lebesgue measure .
let @xmath15 be a r.v .
having an analytic characteristic function @xmath16 . then , @xmath15 is gaussian if , and only if , @xmath80 + necessary part is obvious .
let us show the sufficiency .
as @xmath15 satisfies ( [ equnid1 ] ) , and @xmath79 is absolutely continuous , we have that the set @xmath81 is infinite and not denumerable .
thus , it contains at least an accumulation point . furthermore , the function @xmath82 is analytic , and it vanishes on @xmath83 .
therefore , this function has a non - isolated zero but the only analytical function with at least a non - isolated zero is the null function which proves the result ( see for example @xcite ) .
proposition [ theo1dimens ] may be seen as a spectral counterpart of theorem [ theogausianas ] . through this section we present some useful popular tests for checking whether a stationary random process @xmath84 is gaussian .
the test discussed in this section is a particular case of the one studied in section 3 of @xcite .
we begin with some notations and definitions .
given @xmath85 , let us define @xmath86 where @xmath87 denotes transposition .
+ let @xmath88 and let @xmath89 be the @xmath90-dimensional column vector composed by the real and complex parts of the empirical characteristic function computed at @xmath91 . that is @xmath92
further let , for @xmath93 real and @xmath94 , the @xmath90-dimensional vector composed by the real and complex parts of @xmath95 computed at @xmath91 : @xmath96 we denote by @xmath97 the spectral density matrix ( see for example @xcite ) of the process @xmath98 at frequency @xmath99 . notice that if we assume that @xmath100 is a gaussian stationary process with @xmath101 then the existence of @xmath97 is one of the conclusions of lemma 2.1 in @xcite .
for the construction of the test statistic we will use the following estimator of @xmath97 : @xmath102 where @xmath103 and @xmath104 denotes the integer part .
the estimator ( [ fhat ] ) was used in @xcite , but with @xmath105 replaced by a general constant in the interval @xmath106 notice also that it is a particular case of the one proposed in @xcite . in @xcite
it is proved that if @xmath100 is gaussian , stationary and satisfies ( [ h ] ) , then @xmath107 converges almost surely to @xmath108 let @xmath109 be the generalized inverse of @xmath110 and let @xmath111 be the quadratic form @xmath112 let @xmath113 be an open bounded subset of @xmath114 and let @xmath115 .
we state two assumptions .
: : the set @xmath116 is nowhere dense in @xmath117 h2 . : : for each @xmath118 we have , @xmath119 and @xmath120 theorem [ theoepps ] below describes the gaussianity test studied in @xcite .
[ theoepps ] let @xmath100 be a stationary gaussian process satisfying ( [ h ] ) .
let @xmath113 be an open and bounded subset of @xmath114 and @xmath115 such that _ * h1 .
* _ and _ * h2 . * _ hold .
further , let @xmath121 be the minimizer on @xmath113 nearest to @xmath122 of the map @xmath123 assume further that @xmath124 is positive definite . then , for each fixed @xmath125 , @xmath126 converges in distribution to @xmath127 .
obviously a test based on theorem [ theoepps ] may be not consistent .
indeed , it only focuses on the values of the characteristic function at some points . in other words ,
the test could not detect some alternatives with gaussian one - dimensional marginal .
even the test fails against alternatives with non - gaussian one - dimensional marginal but that satisfy that the characteristic functions of the one - dimensional marginal coincides with the one of the corresponding gaussian at the selected points .
the test to assess normality of time series discussed in this subsection was introduced in @xcite .
it uses the skewness - kurtosis test statistic , also called jarque - bera test ( see @xcite and @xcite ) , but improves previous tests of this kind because the statistic is studentized by standard error estimators .
given a process @xmath128 let us denote @xmath129 .
this is an estimator of @xmath130 the test proposed in @xcite handles the statistic : @xmath131 [ love ] let @xmath100 be an ergodic stationary process . *
[ 11 ] if @xmath100 is gaussian and satisfies @xmath132 then @xmath133 in distribution . *
[ nada ] if @xmath100 satisfies * * @xmath134<\infty$ ] , * * @xmath135 for q=2, ... ,16 , where @xmath136 denotes the @xmath137th - order cumulant of @xmath138 * * @xmath139^{1/2}<\infty,$ ] for @xmath140 where @xmath141 denotes the @xmath142-field generated by @xmath143 , @xmath144 and * * @xmath145 ^ 2 + 2\sum_{t_1=\infty}^\infty { \mbox{$\mathbb{e}$}}([(y_0-\mu)^k-\mu_k][(y_t-\mu)^k-\mu_k])>0,$ ] for @xmath140 + then the statistic @xmath146 diverges to infinity whenever @xmath147 or @xmath148 in section [ sectionmodelo ] we will prove this theorem under lighter assumptions on the alternative .
we will need the following recent result taken from @xcite .
it is an improvement of the well - known result in @xcite .
[ teokava ] let @xmath100 be a stationary process with the representation @xmath149=0 , \mbox{where } ( \epsilon_t ) \mbox { are i.i.d.r.vs .. }\end{aligned}\ ] ] assume that @xmath150<\infty$ ] for some @xmath151 if @xmath152 for @xmath153 and @xmath154 then @xmath155 in section [ simulations ] we will propose to use several tests to assess the gaussianity of a process .
thus we obtain several @xmath156-values @xmath157 , where @xmath46 is the number of procedures used . the most popular way to handle several @xmath156-values is to use the bonferroni correction .
however , it is very well - known that this procedure is too conservative .
several alternatives have been proposed in the literature in order to alleviate this problem . here
, we will employ the _ false discovery rate _
the fdr is the expected proportion of wrongly rejected hypotheses along the @xmath46 tests .
taking into account that all the hypothesis we have are equivalent , the fdr coincides with the level of the procedure .
the fdr was introduced in benjamini and hochberg @xcite for independent tests . here
, we employ the improvement proposed in @xcite that does not require dependence assumptions among the tests .
this procedure , when applied to our case , works as follows : [ theofdr ] let us assume that we apply @xmath46 statistical tests to check the same null hypothesis and that the ordered @xmath156-values that we obtain are @xmath158 where @xmath159 .
let @xmath160 . the fdr of the test which rejects if the set @xmath161 is not empty is , at most , @xmath162
. therefore , according to the previous theorem , if we denote @xmath163 we can reject at any level @xmath164 and then , we can take @xmath165 as the resulting @xmath156-value of the procedure .
in this section we present a universal test to check if the distribution of a stationary process is gaussian . thus , given @xmath166 a stationary process of real - valued random variables we are interested in constructing a test for the null hypothesis * @xmath167 against the alternative * @xmath168 notice @xmath13 holds if , and only if , for all @xmath169 @xmath170 is a gaussian vector . as * x
* is stationary , it is equivalent to the distribution of @xmath14 is gaussian .
in addition , it is the sames as the gaussianity of the process @xmath171 for any @xmath172 to check whether @xmath173 is gaussian , we only need to include @xmath173 in an appropriate hilbert space , then select a vector @xmath174 using a dissipative distribution , and compute the scalar product @xmath175 because , according to theorem [ theogausianas ] , almost surely , @xmath173 is gaussian if , and only if , @xmath175 is gaussian . concerning the hilbert space in which the process is included , let us consider the space of sequences @xmath176 with @xmath177 and @xmath178 endowed with the scalar product @xmath179 it is easy to see that if @xmath63 is a stationary process and if the variance of @xmath2 is finite , then , almost surely , @xmath180 and that the gaussianity in this space is equivalent to the ( usual sense ) gaussianity of @xmath181 the reason is that @xmath182 $ ] is finite if it is so the variance of @xmath183 now we need a dissipative distribution on @xmath184 we will use the so - called dirichlet distribution ( see @xcite ) .
we build this distribution using the so - called stick breaking method .
that is , let @xmath185 and consider the probability distribution which selects a random point in @xmath186 according to the following iterative procedure : * @xmath187 $ ] is chosen with the beta distribution of parameters @xmath188 and @xmath189 * given @xmath190 , @xmath191 $ ] is chosen with the beta distribution of parameters @xmath188 and @xmath192 times @xmath193 let us define @xmath194 for @xmath195 and take @xmath196 it can be easily checked that the distribution of @xmath197 is dissipative ( see definition [ dede ] ) .
moreover , @xmath198 almost surely because , as shown in proposition [ unop ] , @xmath199 , almost surely . [ unop ]
let @xmath200 be a stochastic process constructed as described above .
let @xmath201 be the mean of the beta distribution of parameters @xmath188 and @xmath189 then , we have that 1 .
[ el ] @xmath202=\alpha(1-\alpha)^n , \mbox { for every } n\in{\mbox{$\mathbb{n}$}}^*$ ] .
[ el.2 ] @xmath203 , almost surely .
+ obviously _ [ el ] .
_ holds for @xmath204 .
thus , let us assume that _ [ el ] .
_ is satisfied for @xmath205 and let us show that it also holds for @xmath206 . by construction
, we have that if @xmath207 is a random variable with beta distribution of parameters @xmath188 and @xmath208 , then @xmath209 = { \mbox{$\mathbb{e}$}}[\beta ] \left ( 1 - \sum_{i=0}^n { \mbox{$\mathbb{e}$}}[l_{i } ] \right ) = \alpha \left ( 1 - \sum_{i=0}^n \alpha(1-\alpha)^i \right ) = \alpha(1-\alpha)^{n+1 } , \ ] ] where last equality comes from the application of the formula giving the sum of @xmath25 numbers in a geometric progression . concerning _ [ el.2 ] .
_ , we have that @xmath210 indeed , by construction , for every , @xmath211 . however , applying _ [ el ]
. _ , we have that @xmath212 = \sum_{i=0}^\infty \alpha(1-\alpha)^{i}= 1.\ ] ] so that , by ( [ eqel1 ] ) we obtain _ [ el.2 ] .
_ now , let @xmath213 be a fixed realization of the random element @xmath214 drawn independently of the process @xmath215 .
let us consider the process @xmath216 given by the projections of @xmath217 on the one - dimensional subspace generated by @xmath218 i.e. @xmath219 as we will see in proposition [ prop ] , the properties of the process * x * are inherited by the process * y*. moreover , according to theorem [ theogausianas ] , to assess the gaussianity of * x * is enough to do the job on the one - dimensional marginal distributions of * y*. this can be done for instance with epps or lobato and velasco tests presented in section [ subspro ] whenever @xmath220 satisfies the appropriate assumptions .
the following subsections are devoted to this task .
we begin by proving lemma [ ll ] which is necessary for proposition [ prop ] .
[ ll ] let @xmath63 be an ergodic and stationary process such that @xmath221 if we select @xmath222 as described above , then , 1 .
[ un]@xmath223 almost surely .
2 . [ do ] almost surely , the random variable @xmath224 is conditionally integrable given @xmath225 .
+ _ [ un ] .
_ it is straightforward because the cauchy - schwartz inequality gives that @xmath226 where last equality comes from proposition [ unop ] .
+ to prove _ [ do ] .
_ , let @xmath227 be a fixed realization of @xmath225 .
we have that @xmath228 & = & \sum_{i , j=0}^\infty h_ih_ja_ia_j{\mbox{$\mathbb{e}$}}[|x_{-i}-\mu_x||x_{t - j}-\mu_x| ] \\ & \leq & \sum_{i , j=0}^\infty h_ih_ja_ia_j({\mbox{$\mathbb{e}$}}[(x_{-i}-\mu_x)^2])^{1/2}({\mbox{$\mathbb{e}$}}[(x_{t - j}-\mu_x)^2])^{1/2}=\gamma_x\left(\sum_{i}^\infty h_ia_i\right)^2.\end{aligned}\ ] ] thus , @xmath229 is conditionally integrable thanks to _ [ un ] . _ and that @xmath230 . in the sequel
@xmath231 denotes the conditional autocovariance of order @xmath6 of @xmath15 given @xmath232 . that is , denoting by @xmath233 the conditional expectation of @xmath234 given @xmath235 , @xmath236.\ ] ] [ prop ] let @xmath237 be an ergodic and stationary process such that @xmath238 with @xmath239 then
, conditionally on @xmath232 , the process @xmath240 defined in ( [ y ] ) is ergodic and stationary . in addition , @xmath241 $ ] and @xmath242 are finite . + @xmath237 is a stationary ergodic process .
so that , conditionally on @xmath232 , @xmath240 is also a stationary ergodic process ( see @xcite page 458 ) . using the definition of the process * y * we have @xmath243 \leq { \mbox{$\mathbb{e}$}}\left .
\left[\sum_{i=0}^\infty h_ia_i |x_{-i}| \right| \textbf{h } \right]={\mbox{$\mathbb{e}$}}[|x_0|]\sum_{i=0}^\infty h_ia_i<\infty , \
\mbox { a.s.}\ ] ] because of _ 1 . _ in lemma [ ll ] . by _ [ do ] . _ in lemma [ ll ]
, we have that @xmath244\ ] ] exists .
so that , using the dominated convergence theorem , we obtain that @xmath245 and @xmath246 obviously , @xmath247 , where * @xmath248 * @xmath249 * @xmath250 if @xmath251 as @xmath252 and @xmath253 , we have @xmath254 thus , @xmath255 because @xmath256 then , @xmath257 and so , _ 1 . _ in lemma [ ll ] implies @xmath258 concerning @xmath259 as @xmath260 and @xmath261 , we can apply the @xmath262inequality ( see @xcite p.157 ) to @xmath263 to obtain that there exists @xmath264 such that @xmath265 thus , @xmath266 then , using the same tricks as for @xmath267 we obtain that @xmath268 + for @xmath269 , the fact that @xmath257 , implies that there exists an @xmath270 such that @xmath271 for all @xmath272 .
therefore , @xmath273 by _ 1 . _ in lemma [ ll ] , to show that @xmath274 , we only need to prove that @xmath275 . furthermore , applying jensen inequality and _ [ el ] .
_ in proposition [ unop ] , we have that @xmath276\leq \sum_{j=0}^\infty a_j^{1/2 } ( 2j)^\zeta ( 2j+1 ) \alpha^{1/2 } ( 1- \alpha)^{j/2}.\ ] ] this last series is convergent ( @xmath160 ) .
hence , @xmath277 is finite almost surely and the proof is ended . in this subsection
we analyze the theoretical behavior of epps test when applied to the randomly projected process ( see theorem [ theoconsist ] ) .
moreover , in a corollary ( corollary [ coroepps ] ) we will show that if @xmath91 is drawn randomly , then the epps test is consistent against many more alternatives .
let us first state lemma [ cs ] that gives the consistency for the estimator of the spectral density function at zero defined in ( [ fhat ] ) .
let us denote by @xmath278 the fourth - order cumulant of @xmath279 @xmath280 @xmath281 and @xmath282 where , for instance , @xmath283 is the @xmath284-th component of the vector @xmath285 ( see subsection [ subsepps ] ) .
[ cs ] let @xmath88 . if @xmath286 is a stationary process such that @xmath287 then , @xmath288 + it is straightforward from the proof of lemma 2.2 in @xcite but substituting by ( [ cumu ] ) the use of ( [ h ] ) and gebelein inequality , @xcite , for gaussian processes .
gebelein inequality says that the autocovariance of a multidimensional process is smaller or equal than the product of variances of the marginals .
lemma 3.1 in @xcite proves that if @xmath286 is a stationary gaussian process that satisfies ( [ h ] ) , then ( [ cumu ] ) holds . in @xcite ,
gebelein inequality is extended to two - dimensional vectorial processes with diagonal densities .
so that , any stationary process that satisfies ( [ h ] ) and whose two - dimensional marginal has diagonal density , also satisfies ( [ cumu ] ) .
let @xmath113 be an open and bounded subset of @xmath289 in @xcite , it is proved that * h1 * and * h2 * ( see subsection [ subsepps ] ) are satisfied if @xmath290 is equal to a rational number times @xmath291 @xmath292 now , thanks to lemma [ la ] below , we have that @xmath293 can be taken at random and still fulfill * h1 * and * h2*. [ la ] assume that @xmath294 ( @xmath85 ) is drawn randomly and has distribution @xmath295 having the following properties .
first @xmath295 is such that @xmath296 and @xmath297 are independent and identically distributed and have a density .
further , for @xmath298 @xmath290 is a rational number times @xmath299 then , * h1 * and * h2 * are fulfilled almost surely .
+ proceeding as in @xcite we have that @xmath300 now , in order to get that the cardinal of @xmath301 is larger than one , we need @xmath297 is equal to a rational number times @xmath299 however , this happens with probability zero and so , with probability one @xmath302 thus , * h1 * and * h2 * follow directly .
note that in case @xmath85 lemma [ la ] remains valid if we draw independently at random @xmath303 @xmath304 in addition , thanks to this lemma we have the following corollary of theorem [ theoepps ] .
[ teo ] let @xmath100 be a stationary gaussian process which satisfies ( [ h ] ) and @xmath293 be as in lemma [ la ] .
let @xmath121 be the minimizer on @xmath113 of the map @xmath305 nearest to @xmath306 if we assume that @xmath124 is positive definite , then @xmath126 converges in distribution to @xmath307 in the next theorem , the function @xmath308 also depends on the random @xmath232 .
however , for the sake of simplicity we have not express the functional relationship .
[ theoconsist ] let * x * be an ergodic stationary process satisfying ( [ h ] ) .
draw respectively @xmath293 as in lemma [ la ] and @xmath309 independently of @xmath293 using @xmath310 ( as described above ) .
assume that , conditionally on @xmath232 , @xmath311 defined in ( [ y ] ) satisfies ( [ cumu ] ) , that the characteristic function of its one - dimensional marginal is analytic and that @xmath312 exists and is positive definite for almost every @xmath232 .
let @xmath313 be the quadratic form defined in ( [ eqfcuadrat ] ) applied to * y * and @xmath121 its minimizer on @xmath113 nearest to @xmath314 let further @xmath315 then , * x * is gaussian if , and only if , @xmath316>0.$ ] + necessary part is obvious , because if * x * is gaussian , then * y * also is gaussian and proposition [ prop ] implies that @xmath286 satisfies the assumptions of corollary [ teo ] . + let us show the sufficient part . as @xmath316>0 $ ] we have that there exist @xmath317 and @xmath293 with @xmath318 and @xmath319 such that @xmath126 converges in law to a non - degenerated distribution .
in addition , we may assume without loss of generality that @xmath320 and @xmath321 indeed , as @xmath322 is an analytic characteristic function it has only isolated zeros .
+ therefore , @xmath323 converges in probability to zero . by lemma [ cs ]
, @xmath107 converges to @xmath324 thus , @xmath325 is positive definite because it is the inverse of @xmath326 .
this together with ( [ eqfcuadrat ] ) gives that @xmath327 since @xmath328 is an ergodic stationary process , by @xcite page 458 we have that @xmath329 is also an ergodic stationary process .
thus , as @xmath330 and @xmath331 for all @xmath332 we have by theorem 2 of chapter iv in @xcite that @xmath333.\ ] ] from this and ( [ 2 g ] ) we may conclude that @xmath334 converges in probability to @xmath335 ( @xmath336 ) . + let us see how this implies that the sequence @xmath337 converges .
we have that latexmath:[\[\lim_{n\rightarrow\infty}|\phi_{\mu_n , \gamma_n}({\mbox{$\lambda$}}_1 ) | = \lim_{n\rightarrow\infty } e^{-{\mbox{$\lambda$}}_1 ^ 2\gamma_n/2 } = since @xmath318 and @xmath339 , this implies that there exists @xmath340 such that @xmath341 in probability .
note that there exists @xmath342 such that @xmath343 as @xmath318 , if we take @xmath344 , then , we have that @xmath345 + analogously , we have that @xmath346 in probability , and as @xmath341 we obtain @xmath347 denoting @xmath348 , we obtain that @xmath349 together with ( [ ese ] ) this gives @xmath350 as @xmath351 was drawn independently from @xmath352 with a distribution absolutely continuous with respect to the lebesgue measure and as @xmath322 is analytic , by proposition [ theo1dimens ] we get that @xmath234 is gaussian
. then , by theorem [ theogausianas ] , we obtain that the process @xmath63 is gaussian .
it is only necessary to assume that * x * is ergodic to prove the inverse part in theorem [ theoconsist ] since every stationary gaussian process which satisfies ( [ h ] ) is ergodic .
( see for example @xcite ) applying the arguments of theorem [ theoconsist ] directly to the process @xmath353 , we obtain the following corollary .
it gives a modification of epps test with better consistency properties .
[ coroepps ] let * x * be an ergodic stationary process .
assume that the characteristic function of its one - dimensional marginal is analytic .
assume further that ( [ h ] ) holds .
let us take @xmath293 as in lemma [ la ] , @xmath313 as in ( [ eqfcuadrat ] ) , let @xmath121 be its minimizer on @xmath113 nearest to @xmath354 and @xmath355 if we assume that @xmath356 exists and is positive definite , then , * x * is gaussian if , and only if , @xmath357 remark below can be obviously deduced from theorems [ theoepps ] and [ theoconsist ] .
this remark allows to perform a test based on the asymptotic distribution of @xmath358 [ c ] theorem [ theoconsist ] and corollary [ coroepps ] remain valid if we change in the definition of sets @xmath359 and @xmath360 non - degenerated distribution " by chi - squared distribution with @xmath361 degrees of freedom " .
in addition , we have the following corollary . under the assumptions of theorem [ theoconsist ] , @xmath316\in\{0,1\}$ ] and *
x * is gaussian if , and only if , @xmath316=1.$ ] analogously , under the assumptions of corollary [ coroepps ] , @xmath362 and * x * is gaussian if , and only if , @xmath363 in this subsection we show that a slight modification of the statistic @xmath146 satisfies theorem [ love ] under different assumptions than the ones used in @xcite .
the test statistic is @xmath364 with @xmath365 where , according to theorem [ teokava ] , we take @xmath366 for @xmath367 @xmath368 and @xmath151 thus , the differences between @xmath369 and @xmath146 are the absolute values in the denominator and the number of terms involved in @xmath370 [ lovenuestro ] let @xmath237 be an ergodic and stationary process such that @xmath371 we have that 1 . [ 1 ] if @xmath237 is a gaussian process , then @xmath372 2 .
[ 2 ] if @xmath373 can be written as ( [ kava ] ) and @xmath374<\infty,$ ] then , conditionally on @xmath232 , @xmath369 diverges almost surely to infinity whenever @xmath375 or @xmath376 + using proposition [ prop ] for @xmath377 we get that @xmath240 is an ergodic and stationary process with @xmath378 if @xmath237 is gaussian , the process @xmath240 is also gaussian .
thus , assumptions of the first part of theorem [ love ] hold for the process @xmath240 and so @xmath379 now , as * y * is gaussian , by @xcite page 568 , we have that @xmath380 for @xmath381 repeating the proof of lemma 1 in @xcite , we have that @xmath382 and so , we may conclude that @xmath383 which shows _
1_. + let us prove now statement _ [ 2 ] .
_ first , let us show that @xmath384<\infty$ ] , almost surely , for @xmath385 by hlder inequality we have that @xmath386 and , as by proposition [ unop ] @xmath387 almost surely , we can apply jensen inequality .
we obtain that @xmath388 thus , @xmath389<\infty,$ ] almost surely , for @xmath385 by @xcite , page 458 , we have that @xmath390 is stationary and ergodic , for all @xmath385 therefore , theorem 2 of chapter iv in @xcite implies that @xmath391 further , let us prove that @xmath392 for almost every * h * and @xmath381 we have @xmath393 taking into account that @xmath394 with @xmath37 @xmath395 and @xmath396 we have @xmath397 and then we obtain @xmath398 let us prove now that @xmath399 note that as @xmath374<\infty,$ ] we also have @xmath400 & = & \sum_{j_1, ... ,j_4=1}^\infty\prod_{r=1}^4k(j_r)e\left[\prod_{r=1}^4 \epsilon_{n - j_r}\right ] \\ & = & { \mbox{$\mathbb{e}$}}[\epsilon_{1}^4]\sum_{j=1}^\infty k(j)^4 + { \mbox{$\mathbb{e}$}}[\epsilon_{1}^2]^2\sum_{i , j=1 , i\neq j}^\infty k(i)^2k(j)^2,\end{aligned}\ ] ] because @xmath401 are i.i.d.r.vs . with @xmath402=0.$ ]
so that @xmath403<\infty.$ ] further , using theorem [ teokava ] we obtain that @xmath404 thus , @xmath405 then , by proceeding similarly as in the proof of proposition [ prop ] , we get @xmath406 and so , @xmath392 for @xmath381 using ( [ mu ] ) we may conclude that _ 2 .
_ holds . finally ,
applying theorem [ lovenuestro ] directly to the process @xmath353 , we obtain the following corollary .
[ lovenuestroc ] let @xmath237 be an ergodic and stationary process such that @xmath371 we have that 1 .
[ 111 ] if @xmath237 is a gaussian process , then @xmath407 2 .
[ 22 ] if @xmath373 can be written as ( [ kava ] ) and @xmath374<\infty,$ ] then , conditionally on @xmath232 , @xmath408 diverges almost surely to infinity whenever @xmath409 or @xmath410
in this section we discuss the practical implementation of the gaussianity test .
we start doing some remarks on epps test .
although theorem [ theoepps ] works for any @xmath411 with @xmath412 which satisfying * h1 * and * h2 * , in @xcite it is stated that : _ _ * when either @xmath413 is large or the spacing between the @xmath414 is small , relative to the scale of the data , the matrix @xmath415 often appears computationally singular . * also , values of @xmath414 which are large , relative to the scale of the data , makes difficult to find a minimum of @xmath416 with much precision .
epps suggests to take @xmath417 recall that @xmath418 denotes the sample variance of the process .
he proved that theorem [ theoepps ] works taking such @xmath419 in the simulations of epps , and also in the ones of @xcite , @xmath420 and @xmath421 however , we need to draw @xmath293 randomly in order to have a consistent test ( theorem [ theoconsist ] ) .
so , we take @xmath422 @xmath423 distributed as the absolute value of a standard normal distribution and @xmath424 distributed as the absolute value of a normal distribution with mean zero and variance @xmath425 with this selection , although seldom , we have found that @xmath426 could be singular .
this is the main reason to choose @xmath109 as the generalized inverse of @xmath427 another important practical issue is the procedure used to find the minimizer nearest to @xmath306 of the map @xmath428 in the simulations of @xcite and @xcite they use the simplex method developed in @xcite .
we did the same .
the code of such method can be found in @xcite under the name amoeba . the theoretical development of section [ sectionmodelo ]
was carried out assuming that the observed sample is infinite . however , in practice , only a finite number of measurements @xmath429 are available .
so that , only a finite number of components of @xmath309 are computed .
this last difficulty is handled by fixing a small @xmath430 ( equal to @xmath431 in the simulations that we present in section [ simu ] ) and by taking @xmath432 with @xmath433 where @xmath434 are drawn by the stick breaking procedure described in section [ sectionmodelo ] .
further , @xmath435 is fixed such that @xmath436 concerning the projected process , some possibilities are available , but here we use @xmath437 let us give a short comment on the choice of the parameters @xmath438 of the beta distribution used to generate @xmath439 here we have to deal with the following situation : if @xmath284 is large , then the random variables @xmath440 are linear combinations of many random variables from the first sample and then , because of the central limit theorem , the distribution of the random variables @xmath440 will become closer to a normal law . that will cause some loss of power when the marginal of * x * is not gaussian .
thus , in order to detect a non - gaussian marginal , it is wise to select @xmath188 and @xmath208 in such a way that @xmath284 is small or even @xmath99 or @xmath441 .
this goal is achieved if we take @xmath442 and @xmath443 .
our selection in section [ simu ] is @xmath444 .
however , in this case the samples @xmath445 and @xmath446 are quite similar . so that , the test will not be good in detecting non - gaussian alternatives with gaussian marginal . in order to overcome this problem we should take @xmath309 in such a way that the projections mix several variables from the initial sample . to achieve this goal we need to take @xmath447 but with @xmath448 being not too big to avoid the effect of the central limit theorem . in this case
, a selection like @xmath449 and @xmath450 seems appropriate .
therefore , it seems that in a practical situation we should decide which alternative is more plausible and then , select the appropriate parameters .
however , there is another possibility : select two projections ( one with each pair of parameters ) and apply theorem [ theofdr ] to mix the @xmath156-values .
this is our proposal .
finally , we need a gaussianity test for the one dimensional marginal of @xmath451 . we have seen two such tests ( which have some advantages and disadvantages discussed in section [ simu ] ) and we can also mix them . having all these requirements in mind , we propose the following procedure : 1 .
draw @xmath452 with the @xmath453 distribution and apply epps test to the projections to obtain the @xmath156-value @xmath454 .
2 . draw @xmath455 ( independently of @xmath452 ) with the @xmath453 distribution and apply lobato and velasco test to the projections to obtain the @xmath156-value @xmath456 .
3 . draw @xmath457 ( independently of @xmath452 and @xmath455 ) with the @xmath458 distribution and apply epps test to the projections to obtain the @xmath156-value @xmath459 .
4 . draw @xmath460 ( independently of @xmath452 , @xmath455 and @xmath457 ) with the @xmath458 distribution and apply lobato and velasco test to the projections to obtain the @xmath156-value @xmath461 .
combine the @xmath156-values @xmath462 using the procedure described in section [ mt ] to decide the gaussianity hypothesis at the level @xmath162 .
thus , ordering these four @xmath156-values such that @xmath463 we obtain that the @xmath156-value of the random projection test is equal to @xmath464
in this section we study the behavior of the proposed procedure in different situations .
we have used the same distributions as in @xcite , in order to perform comparisons .
further , we will study a situation where the process has gaussian marginal but is not gaussian ( see section [ non - gaussian ] ) .
in addition , in subsection [ real ] we apply the random projection test to real data .
the authors of @xcite study the case of an ar(1 ) process depending on a parameter @xmath137 defined by @xmath465 where @xmath466 @xmath467 and @xmath468 are i.i.d .
random variables with distribution @xmath469 which may be any of the following ones : * standard normal ( n(0,1 ) ) , * standard log - normal ( log n ) , * student @xmath6 with 10 degrees of freedom , ( @xmath470 ) , * chi - squared with @xmath441 ( @xmath471 ) and @xmath472 degrees of freedom ( @xmath473 ) , * uniform on @xmath474 $ ] ( @xmath475 ) , * beta with parameters @xmath476 ( @xmath477 ) .
it is obvious that if @xmath482 , this process is not stationary .
for instance , @xmath483=\mbox{var}[\varepsilon_1](1-q^{2t})/(1-q^2 ) $ ] which is not constant and , obviously , the differences increase with @xmath484 . in order to alleviate this problem
, we disposed a certain number , _ past _ , of observations .
we have taken @xmath485 equal to @xmath486 and @xmath487 equal to @xmath488 which are the sample sizes handled in @xcite .
we have performed 5000 simulations in each situation . in every run we have computed the @xmath156-values using the asymptotic distributions .
this could have caused that sometimes the rejection rates under the null hypothesis are far away from the nominal level ( mostly for the lowest sample size @xmath489 ) and that they decrease under some alternatives with the sample size ( mostly for high values of @xmath484 ) .
there are some differences between our rates and those published in @xcite .
we think they could be due to the fact that the _ past _ taken in @xcite is not large enough .
for example , in the case @xmath489 , @xmath490 and @xmath469 being @xmath477 we obtain a rejection rate of @xmath491 when using epps test while in @xcite they obtain one of @xmath492 which is appreciatively worse .
as explained before , our simulations were made with _
past_@xmath493 , but from table [ past ] we see that @xmath494 is a rejection rate reasonable for _ past _ @xmath495 and that the rejection rates increase with _ past _ , approaching to the value we have obtained .
furthermore , another difference to bold between what we do here and @xcite is that in subsection [ conlv ] a sum until @xmath500 is involved in the estimation of @xmath501 while in @xcite the sum goes until @xmath502 where @xmath25 is the sample size . here
, we have to take @xmath503 where @xmath504 with @xmath505 and @xmath368 .
thus , @xmath506 may be as close as desired to .5 and so , we have decide to fix its value at @xmath507 for the simulations . in order to select the right value of @xmath499
, we have made a small analysis to see how sensitive is the method to this parameter .
we run lobato and velasco test under the null hypothesis for all values of @xmath137 and @xmath508 where @xmath509 and @xmath510 therefore , @xmath511 @xmath512 and @xmath513 the results suggest that the value of @xmath499 has not much influence in the rejection rates and so , we choose @xmath514 the results for the cases @xmath515 and @xmath498 appear in figure [ c5 ] . it is worth saying that the situation for @xmath516 is a bit different than for the other values , as with @xmath516 the rejection rates look constant till a point in which those rates strongly decrease
. tables [ t1 ] , [ t2 ] and [ t3 ] contain the rejection rates for several procedures when applied at the level @xmath517 .
next we mention the procedures we have selected and make some comments on the results of our simulations . 1 .
[ e ] * epps test , e - test*. we take @xmath518 in ( [ beta ] ) .
+ it seems that this test behaves poorly when @xmath469 is @xmath470 .
moreover , broadly speaking , its power decreases for the considered alternative distributions when @xmath484 increases , having low powers when @xmath519 .
note also that under the null hypothesis ( excepting the case @xmath515 with @xmath520 ) , the rejection rates are above the level of the test and that they increase with @xmath521 + the power decreases when the sample size increases in the cases in which @xmath519 and the alternative is @xmath522 @xmath523 @xmath475 or @xmath477 ( and even with @xmath524 when @xmath525 ) .
* lobato and velasco test , g - test . *
the rejection rates displayed have been simulated using the statistic @xmath408 .
however , they are similar to those obtained using @xmath526 .
+ the * g*-test has very low powers when @xmath484 is large , sometimes even lower than those of the * e*-test .
in addition it suffers from a lack of power when @xmath527 is @xmath475 or @xmath477 .
the rejections under the null hypothesis are above the level of the test only in 4 cases out of 24 .
in contrast with the * e*-test , here the rejection rates under the null hypothesis decrease when @xmath137 increases .
3 . * combined epps and lobato and velasco test , ge - test . * in previous paragraphs we have commented some problems of the * e * and * g * tests which go , let us say , in opposite directions .
in order to solve these problems we combine both tests , using the multiple testing procedure presented in section [ mt ] . + as stated in subsection [ subepps ] , the * ge*-test have been obtained by drawing independently @xmath423 with absolute value of a standard normal distribution and @xmath424 with absolute value of a normal distribution with mean zero and standard deviation @xmath528 however , it is worth noting that the rejection rates we have obtained have been a bit larger than in the case we take @xmath421 + we can observe from tables [ t1 ] , [ t2 ] and [ t3 ] that this combination gives rejection rates between those of the * e * and * g*-tests although closest to the highest one , and , sometimes , even above .
this is due to , as we have previously said , the rejection rates of * e * are here a bit larger than when @xmath421 4 . * random projection test , rp - test .
* we apply this test following the guidelines provided in subsection [ subsrptest ] .
+ when @xmath137 is negative and we are under the alternative , we always get the highest rejection rates with the * rp*-test .
the most striking behavior of this test happens for @xmath529 and @xmath530 and @xmath531 where the * rp*-test obtains rejection rates larger than @xmath532 while the second more successful test remains below @xmath533 . for the remaining values
, it happens that the rejection rates using the * rp*-test are between the rates obtained with the * e , g * and * ge * tests but closer to the highest than to the lowest .
* acknowledgment .
* this research has been carried out partially during a stay of j.a .
cuesta - albertos at universit paul sabatier , toulouse , france , supported by the spanish ministerio de educacin under grant pr2009 - 0355 .
( 2006 ) . _ combinatorial stochastic processes .
lectures from the 32nd summer school on probability theory held in saint - flour . _ springer . , teukolsky , s.a . ,
vetterling , w.t . and flannery , b.p .
_ numerical recipes . the art of scientific computing .
_ cambridge university press . | in this paper we address the statistical problem of testing if a stationary process is gaussian .
the observation consists in a finite sample path of the process . using a random projection technique introduced and studied in @xcite in the frame of goodness of fit test for functional data ,
we perform some decision rules . these rules really stand on the whole distribution of the process and not only on its marginal distribution at a fixed order .
the main idea is to test the gaussianity on the marginal distribution of some random linear combinations of the process .
this leads to consistent decision rules .
some numerical simulations show the pertinence of our approach .
[ section ] [ prop ] lemma [ prop]theorem [ prop ] corollary [ prop ] claim [ prop ] [ prop ] example [ prop]definition [ prop]figure [ prop]table _ key words and phrases : gaussianity test , strictly stationary random process , random projection , consistent test _ _ a.m.s .
1980 subject classification : a60 g _ |
lattice models of interacting bosons and fermions such as hubbard model and its various generalizations are usually strongly correlated systems exhibiting various phases with competing orders , which are of fundamental interest in fields of both condensed matter and cold atomic physics .
interests on both types of hubbard models are renewed recently , since they can be realized in cold atomic gases loaded in optical lattices ( for a review see refs .
[ ] and [ ] and references therein ) . unlike fermions
, there is a natural superfluid order for free bosons at zero temperature driven by the kinetic energy .
when the interaction is switched on , the bosons are likely to be localized in various crystalline patterns , which may coexist with superfluid order @xcite to give a realization of intriguing `` supersolid '' state that has been pursued for decades since 1950s @xcite .
recently , people have observed the non - classical rotational inertia in solidified @xmath3he @xcite implying a possible supersolid state , which , in spite of the controversy over this topic , also triggers extensive studies on various boson hubbard models .
experimentally , the boson hubbard model can be used to mimic the granular superconductors , where the cooper pairs are described as bosons , which has been studied by fisher _
[ ] two decades ago , where with only on - site repulsive interaction they showed that bosons can form either mott insulating state with integer filling or superfluid state .
recent experimental progress in cold atomic system provides another realization of boson hubbard model by loading atoms into an optical lattice with possible long range interactions through dipole interaction @xcite , or mediated by other intermediate states or fermions @xcite .
in addition , the boson models also share similarities with quantum magnets , e.g. , the uniaxial magnetization corresponds to insulating states of boson hubbard model ( e.g. ref . ) , while the easy - plane magnetization corresponds to the superfluid state .
hence , the studies on the boson hubbard model may shed light on some common issues of strongly correlated lattice models .
generally speaking , boson models with interactions at zero temperature have two principal phases : ( i ) the superfluid and ( ii ) the incompressible insulating state , which are favored respectively by kinetic and interaction energies , and can coexist . depending on the features of interaction terms , there are several types of insulating phases , such as mott , valence bond crystal , and charge density wave ( cdw ) .
note that we in this article define the incompressible states with oscillating density profile as cdw , though the bosons may not carry charges .
the extended boson hubbard ( ebh ) model with onsite ( @xmath4 ) and nearest neighbor ( @xmath1 ) interactions is a minimal model in favor of cdw and supersolid phases , which has the form of @xmath5 where @xmath6 ( @xmath7 ) is the creation ( annihilation ) bosonic operator at site @xmath8 , @xmath9 is the hopping amplitude , @xmath10 is the particle number , @xmath0 is the chemical potential , and @xmath11 runs over all nearest neighbors .
recently , hamiltonian eq . and its hard - core version ( equivalent to the quantum spin-1/2 xxz model ) with different underlying lattices have been extensively studied in different parameter regimes @xcite . however , a global phase diagram of the three - dimensional ( 3d ) ebh model [ eq . ] is still of lack . as there is no sign problem for the ebh model ,
the quantum monte carlo ( qmc ) simulation is the most convenient tool for this purpose .
the worm algorithm @xcite will be invoked to study hamiltonian on a simple cubic lattice , together with other perturbation and mean - field approaches .
the system described by the ebh model can give rise to a charge ordered crystal at commensurate fillings .
the first one is for half filling @xmath12 , and the corresponding solid state is labeled as cdw i. doping
_ particles _ into this state can lead to a supersolid state @xcite .
however , as shown in ref .
@xcite , doping _
holes _ into it acts quite differently , which may not result in a supersolid state with @xmath13 , but a phase separation between superfluid and cdw i states , which signals a first - order phase transition .
their argument is based upon the following two observations .
( i ) taking one particle out of a perfect cdw crystal with half filling costs almost no potential energy , but only chemical potential . at the same time
, the hopping hole also gains a kinetic energy which is quadratic in @xmath9 @xmath14 .
for a perfect cdw crystal , these three processes are balanced , so one can not take one particle out .
( ii ) the cdw phase breaks the translational symmetry , leading to a two - fold degenerate ground state .
if holes are doped into the domain wall between these two degenerate phases , the kinetic energy gained is proportional to @xmath9 .
hence , the cdw phase is unstable toward the domain wall formation if the hole density exceeds @xmath15 for @xmath16 lattice , though it is still stable against losing one particle .
this argument perfectly explains the first - order phase transition from the cdw i to superfluid state with @xmath17 , but it fails in two circumstances .
the first is that in one dimension the kinetic energy is always linear in @xmath9 , and the corresponding transition is of the kosterlitz - thouless type @xcite .
the other is that if @xmath1 is comparable to @xmath9 the kinetic energy of holes is also linear in @xmath9 , which may result in the supersolid phase with the particle density less than half filling ( see sec .
[ sec : case - finite - t ] ) . this can be verified by the mean - field calculations @xcite . at unitary
filling , the ground state can be either a uniform mott insulator with one particle per site or a charge ordered crystal with two particles on one sublattice and empty on the other one which is labeled as cdw ii .
there is a critical region around @xmath18 , where the two states with different translation symmetries become degenerate , and however , they are separated thermodynamically , i.e. , any local perturbation can not take one to the other .
correspondingly , the transition between them is a first - order one .
it is noted that the aforementioned transition from the superfluid to cdw i state by tuning the chemical potential is of the _ weak _ first - order @xcite .
far less attention has been paid to the region with @xmath19 by now , of which the details are given as part of the phase diagram in this article .
to plot the ground state phase diagram , we focus on the case with small hopping and average particle density around or smaller than 1 . for larger @xmath9 or @xmath20 , we expect no essentially new physics .
this article is organized as follows . in sec .
[ sec : ground - state - phase ] , we shall present the global phase diagram .
the details of the order parameters will be discussed in sec .
[ sec : order - parameters ] . the conclusion will be given in last section .
we start from the classical case without hopping .
the energy per site of ground state is a function of the particle numbers on the two sublattices , @xmath22 and @xmath23 , @xmath24\;,\end{aligned}\ ] ] where the coordination number @xmath25 for the simple cubic lattice .
the states can be labeled by @xmath26 .
the mott states correspond to @xmath27 , and the cdw states with @xmath28 break the translational symmetry , which is two - fold degenerate . in this article
, we define the state @xmath29 as cdw i state , and @xmath30 as cdw ii state and we only consider @xmath31 for the cdw states for convenience .
( color online ) the phase diagram with zero hopping @xmath21 , where the states are labeled by the particle numbers on two sublattices @xmath26 with the assumption @xmath32 .
the states with @xmath33 can be obtained by inversion .
the cdw states with one sublattice empty on the right side .
some states can only exist on the green thick solid line , e.g. , the states @xmath34 and @xmath35.,width=226 ] for @xmath36 the ground state is a vacuum without any particles . as the chemical potential is increased @xmath37 , the particles are loaded into one sublattice to form a charge ordered pattern with @xmath38 and @xmath39 , i.e. , cdw i state .
if we further increase the chemical potential , more particles are loaded into the cubic lattice , which fill either the empty sites if @xmath40 to form a uniform mott state , or the occupied sites if @xmath41 leading to a cdw ii state . in the mott state ,
each particle interacts with its nearest neighbors , which effectively lowers the chemical potential to be @xmath42 , and then the critical line between cdw i and mott states is simply @xmath43 . while that between cdw i and ii states is a horizontal line @xmath44 because the chemical potential only needs to compensate the on - site interaction @xmath4 for adding new particles .
similarly , by studying the instability of adding particle to a state @xmath26 , one can determine all the phase boundaries between different classical insulating states , as shown in fig .
[ fig : zerot ] .
there is a special vertical line @xmath45 , on which many states can coexist with the same free energy .
for example , on the boundary between mott and cdw ii states , there are actually three macroscopic states , which are ( 1,1 ) , ( 2,0 ) and ( 0,2 ) . in fact , some of them only exist on this line in the absence of hopping terms , e.g. states ( 3,1 ) and ( 1,3 ) on the boundary between mott ( 2,2 ) and cdw ( 4,0 ) states .
these degenerate states are macroscopic which can not be transformed to each other smoothly by local perturbations , i.e. , there are infinitely high barriers between these macroscopic states .
for the case with a finite hopping @xmath9 , the particles ( holes ) adding to an insulating state can gain a kinetic energy , which results in the shrinking of insulating areas in the phase diagram ( fig .
[ fig : phasediagram ] ) comparing with the classical case ( fig .
[ fig : zerot ] ) . in three dimensions , these mobile bosons condense at low temperature leading to the superfluidity , which enriches the phase diagram .
there are four phases for the ebh model characterized by the following three quantities : the particle density @xmath20 , superfluid density @xmath46 , and static structure factor @xmath47 at momentum @xmath48 with @xmath49 where @xmath50 are the winding numbers along @xmath51 and @xmath52 directions , @xmath53 is the inverse temperature , and @xmath54 is the effective mass of the bosons . in the insulating mott and cdw states , particles are localized by the interaction and the local particle number is quantized as in the classical case .
the pure superfluid state has nonzero superfluid density @xmath46 and a vanishing static structure factor @xmath47 , while both are finite in the supersolid phase .
( color online ) the phase diagram in the plane of @xmath55 _ vs_. @xmath56 for the extended bose hubbard model on a simple cubic lattice , where @xmath57 and the lattice size is @xmath58 .
four kinds of phases including the superfluid ( sf ) , supersolid ( ss ) , mott and cdw states are identified .
the cdw states are further classified as i and ii by the filling numbers .
the solid lines with circles are the phase boundaries calculated by qmc simulations , and the dotted lines are from the perturbation calculations .
the green lines are the second - order phase boundaries , and the red ones are of the first - order .
the inset is a zoom near the tip of cdw i lobe , where one can find a narrow region below the lobe corresponding to a supersolid phase with the filling less than one half .
for details of order parameters one may refer to figs .
[ fig : extrapolation ] , [ fig : fixedmu ] and [ fig : fixmurho].,width=302 ] fig .
[ fig : phasediagram ] is the phase diagram determined by the qmc simulation with the worm update algorithm , where the solid lines with circles are the qmc results and the dotted lines are from the perturbation expansion in the strong coupling limit @xcite where the insulating states become unstable against adding or removing particles .
it is seen that the perturbation results agree quite well with those of the qmc simulation in part of the phase diagram , but it is still not applicable in some regions since it can not deal with the superfluid order .
comparing with the classical case in fig .
[ fig : zerot ] , the cdw i state is detached from its insulating neighbors , i.e. , the mott , cdw ii and vacuum states . the upper boundary of vacuum state
is actually lowered below @xmath59 due to the hopping of bosons , which is not shown in fig .
[ fig : zerot ] .
the gaps between different insulating states are filled with the superfluid and supersolid phases .
the lower boundary of cdw i state is a critical line on which there occurs a phase separation between the superfluid and cdw i phases , which breaks the u(1 ) gauge symmetry and the translational symmetry , respectively .
the transition between them is of the weak first - order across which the particle density and superfluid density have a jump . considering the correspondence between the ebh model and spin models ,
this transition is similar to the spin - flopping process in the two- or three - dimensional anisotropic xxz model in the presence of magnetic field pointing in @xmath52-axis , which is equivalent to the ebh model in the hard - core limit @xcite . as explained in ref .
[ ] , this first - order phase transition with a particle number jump is due to the fact that the cdw i phase is unstable toward the domain wall formation if the filling number exceeds @xmath15 though it is still stable against doping one hole . doping particles upon the cdw i phase by increasing the chemical potential does not lead to a first - order transition as in the hard - core ebh model ( or equivalently the xxz model ) , where the particle - hole symmetry makes the upper and lower boundaries of cdw i phase identical . in case of the soft - core bosons ,
these additional particles can move upon the alternating charge ordered background with the effective hopping amplitude @xmath60 , which can bose condense at zero temperature without destroying the staggered density order , and thus leads to a supersolid state with @xmath61 .
this transition is of a second - order , as shown by a green solid line in fig .
[ fig : phasediagram ] where all the second - order phase boundaries are in green that are distinct from the first - order ones which are colored in red . continuously increasing the chemical potential upon supersolid phase ,
two different situations occur .
( 1 ) for @xmath62 , the particles like to occupy the empty sites which weakens the cdw order accompanied by the occurrence of superfluid order . until some critical filling @xmath63 , the cdw order is completely destroyed and a pure superfluid state appears with the translational symmetry restored .
the transition is of the second - order .
( 2 ) for @xmath64 , the additional particles are added to the occupied sites so that the cdw order is actually enhanced until finally entering into another insulating state , cdw ii , through a _ first_-order phase transition ,
of which the reason is the same as that from the superfluid with @xmath13 to cdw i. in this sense , the staggered density order in supersolid phase is inherited from the cdw i state , not related to the cdw ii state , since doping holes to the cdw ii state can not _ smoothly _ result in the supersolid state .
( color online ) the extrapolation of the order parameters in the supersolid phase with @xmath13 . in the thermodynamic limit , the staggered structure factor @xmath65 , the superfluid density @xmath66 and the particle density @xmath67 , which indicates a supersolid state less than the half filling.,width=264 ] as shown in the previous study @xcite , doping _ holes _ into the cdw i state may not lead to supersolid state with @xmath13 on a two - dimensional square lattice , contrary to the case of doping _
particles_. the reason is that the cdw i state has two - fold degeneracy .
as long as enough particles are removed , the insulating state becomes unstable towards the formation of domain walls between the two degenerate states , but can still be stable against losing _
one _ particle .
this explains the particle density jump across the first - order phase boundary between the superfluid and cdw i phases and that between the supersolid and cdw ii phases .
however this argument is invalid for @xmath1 close to @xmath9 that is around the tip of the cdw i lobe , where we show below that the kinetic energy gain by doping _
one _ hole into the cdw i state is also linear in @xmath9 , that can cause instability of cdw i state towards a supersolid state without the formation of domain walls .
suppose that one particle is taken from a cdw i state , the hole leaving behind moves in an effective staggered potential , roughly speaking , @xmath68 in one sublattice and @xmath69 in the other . solving this single particle problem , one estimates the kinetic energy gain @xmath70 is @xmath71^{-1},\end{aligned}\ ] ] which is about @xmath72 in the cubic lattice near the tip of cdw i lobe where @xmath73 ( see fig . [
fig : phasediagram ] ) . as a consequence ,
a supersolid phase with @xmath13 occurs , whose boundary is plotted in the inset of fig .
[ fig : phasediagram ] . to confirm
, we also extrapolate the static structural factor , superfluid order and particle density at point @xmath74 and @xmath75 to the thermodynamic limit ( see fig . [
fig : extrapolation ] ) , which indicates a supersolid state with the particle density less than half filling . for more information
, one can refer to next section where we shall examine the order parameters in details . in the presence of a kinetic term ,
the vertical boundary at @xmath19 between mott and cdw ii states does not yet split but moves slightly to the cdw side . on this boundary ,
the free energies of both states are equal .
however , they have different symmetries and are separated from each other thermodynamically .
the transition between these two insulating states is of the first - order , which is similar to the conventional liquid - solid phase transition .
further doping particles into the mott and cdw ii states leads to the reentrance of superfluid and supersolid states , respectively , with density @xmath76 . because the particles or holes can hop between nearest neighbors on a mott background , which leads to a kinetic energy gain linear in @xmath9
, however they can only hop among next nearest neighbors in a staggered cdw background that gives rise to a hopping energy proportional to @xmath77 . then , as @xmath9 increases the mott region shrinks much faster than the cdw ii region , which results in the mismatch of phase boundary between them as @xmath19(see fig .
[ fig : phasediagram ] ) . since doping
the mott state with holes leads to a superfluid state , one expects a boundary between the superfluid state and the cdw ii state in the critical region at the lower end of the vertical boundary .
this phase boundary is again of the first - order . at the upper end ,
the extension of the vertical boundary separates the superfluid and supersolid phases in the way of the second - order phase transition .
as a function of chemical potential @xmath0 for different @xmath1 and @xmath57 .
the first plateau at @xmath12 corresponds to the cdw i phase , and the second one corresponds to the mott phase for @xmath78 and the cdw ii phase for @xmath79 .
the system size is @xmath80.,width=264 ] in this section we give the details of the parameters @xmath20 , @xmath46 and @xmath47 for different @xmath0 and @xmath1 .
[ fig : density ] is the density profile as we vary the chemical potential for several fixed values of @xmath1 , where the plateaus correspond to the incompressible states , i.e. , the mott , cdw i and ii states , with vanishing isothermal compressibility @xmath81 .
note that these insulating states correspond to two single points @xmath12 and @xmath82 in fig .
[ fig : sf ] , which is the plot of superfluid density @xmath46 and @xmath47 as functions of particle density @xmath20 . for @xmath13
, the particles bose condense to form a pure superfluid state with a nonzero @xmath46 but vanishing @xmath47 as shown in fig . [
fig : sf ] .
when the particle density reaches a commensurate value @xmath12 , a plateau appears implying @xmath83 which corresponds to the incompressible cdw i state with translational invariance broken .
this transition is of the first - order since the particle density , as a first - order derivative of free energy with respect to the chemical potential , is discontinuous .
this is also reflected in fig .
[ fig : sf ] as the fact that a segment of particle density below the half filling is inaccessible , and @xmath47 jumps at @xmath12 to a finite value from zero . as the particle density exceeds @xmath84
, @xmath85 becomes a finite positive value again . at the same time the superfluidity appears continuously upon the cdw i state to form a supersolid .
the corresponding transition is of the second - order , since the particle density @xmath20 is continuous but @xmath85 jumps from zero to a finite value . and staggered structure factor @xmath47 as functions of the particle density @xmath20 at @xmath86 .
note that the plateaus of constant @xmath20 in fig .
[ fig : density ] are only single points here . ] between the two plateaus at @xmath12 and @xmath82 , the slope of @xmath87 curve for @xmath88 and @xmath89 , i.e. @xmath62 , shows another jump , implying that @xmath85 is discontinuous and a second - order phase transition from the supersolid to superfluid phase occurs .
this transition is manifested in fig .
[ fig : sf ] , where @xmath46 keeps finite in the whose region @xmath90 , but @xmath47 vanishes at some critical values in between . further increasing the chemical potential
, the system enters into the mott phase , which corresponds to the plateau with @xmath82 in fig .
[ fig : density ] , through a second - order phase transition . at the transition point
@xmath46 vanishes as shown in fig .
[ fig : sf ] . for @xmath91 , i.e. , @xmath41 , the second plateaus in fig .
[ fig : density ] corresponds to the cdw ii state , into which the system enters directly from the supersolid phase as @xmath0 is increased .
the transition is a first - order one as shown in fig .
[ fig : density ] as the jump of particle density .
it is also reflected in fig .
[ fig : sf ] where a segment of particle density @xmath20 is not accessible before @xmath20 reaches 1 . the normalized order parameters @xmath92 and @xmath93 as functions of @xmath56 .
the upper , middle , and bottom panels correspond to @xmath94 , @xmath95 and @xmath96 , respectively .
the supersolid phases occur in shaded areas .
the lattice size is @xmath80.,width=302 ] in fig .
[ fig : fixedmu ] , we plot the normalized order parameters @xmath97 and @xmath98 as functions of @xmath56 around the cdw i lobe , where @xmath99 and @xmath100 . in the plot , we take three characteristic values of the chemical potential @xmath94 , @xmath95 and @xmath96 from the top to the bottom panel . in the shaded areas , when @xmath1 increases , the superfluid density @xmath46 decreases while the static structure factor @xmath47 increases , and both @xmath46 and @xmath47 are nonzero in this region indicating a supersolid state .
the supersolid area becomes more and more narrow as @xmath0 decreases , until finally shrinks to a point in the @xmath56 axis when @xmath101 , which implies that two second - order phase boundaries merge into one first - order phase boundary across which the order parameters @xmath46 and @xmath47 are both discontinuous .
the corresponding density profiles are plotted in fig .
[ fig : fixmurho ] , where we observe that the particle density is always smaller than @xmath84 for @xmath102 corresponding to the bottom panel in fig .
[ fig : fixedmu ] until it enters into the cdw i phase , that implies the supersolid state can exist below the half filling . for @xmath94 ,
the situation is different where @xmath20 decreases as @xmath1 is increasing , which can be intuitively attributed to the loss of effective chemical potential due to nearest neighbor interaction @xmath1 in the mean - field level .
the case of @xmath103 shows an intermediate behavior , for which the particle number first decreases , then increases slightly larger than @xmath84 and finally reaches the cdw i state . the particle density vs. @xmath56 for @xmath94 , @xmath95 and @xmath96.,width=302 ]
in this article , we present the global phase diagram ( fig . [ fig : phasediagram ] ) of the 3d extended bose hubbard model .
the ebh model exhibits four kinds of ground states , including ( 1 ) the mott state without breaking any symmetry , ( 2 ) the cdw i and ii states with translational symmetry broken , ( 3 ) the superfluid with u(1 ) gauge symmetry broken , and ( 4 ) the supersolid with both symmetries broken . by using the qmc simulation as well as other analytical tools ,
we identify the transition type between these phases . among them
, the first - order phase boundary includes those between the superfluid and cdw ( i and ii ) states , and between the mott and cdw ii states .
the other boundaries are all continuous .
the critical regions for @xmath104 and the tip of the cdw i lobe are examined in detail .
we demonstrate that in the present 3d ebh model , the supersolid phase with @xmath13 can appear in a small region near the cdw i lobe where the hopping amplitude @xmath9 is comparable to the nearest neighbor interaction @xmath1 . in this region ,
the general `` domain wall '' argument for the nonexistence of supersolid state with @xmath13 is no longer applicable , since it is based on the assumption of @xmath105 .
we acknowledge the helpful discussions with m. ma . this work is supported by nsfc grant no .
10904081 , 10934008 , 90922033 , hksar rgc grant no .
hku 701009 and chinese academy of sciences .
34ifxundefined [ 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 * * , ( ) w. li , s. s. gong , y. zhao , and g. su , phys .
b * 81 * , 184427 ( 2010 ) . @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | we present the global phase diagram of the extended boson hubbard model on a simple cubic lattice by quantum monte carlo simulation with worm update algorithm .
four kinds of phases are supported by this model , including superfluid , supersolid , mott , and charge density wave ( cdw ) states , which are identified in the phase diagram of chemical potential @xmath0 versus nearest neighbor interaction @xmath1 . by changing the chemical potential ,
a continuous transition is found from the mott phase to a superfluid phase without breaking the translational symmetry . for an insulating cdw state ,
adding particles to it gives rise to a continuous transition to a supersolid phase , while removing particles usually leads to a first - order one to either supersolid or superfluid phase . by tuning the nearest neighbor interaction
, one can realize the transition between two insulating phases , mott and cdw with the same particle density , which turns out to be of the first - order .
we also demonstrate that a supersolid phase with average particle density less than 1/2 can exist in a small region of @xmath2 phase diagram . |
the radio far infrared ( fir ) correlation is one of the tightest observed correlations in astrophysics that connects several independent physical parameters in the interstellar medium ( ism ) .
the radio luminosity and the fir luminosity of star - forming galaxies are observed to be correlated over five orders of magnitude for the global scale @xcite with dispersion less than a factor of 2 .
the radio luminosity is typically measured at 1.4 ghz and the fir luminosity can be both monochromatic ( at 24 , 60 or 70@xmath1 m ) or bolometric ( between 40 and 120@xmath1 m or between 8 and 1000@xmath1 m ) . the radio fir correlation is well studied for galaxies in the local universe for several classes of galaxy morphology like spirals , ellipticals , dwarf irregulars , etc .
it is known to hold good at global @xcite as well as at local scales ( few 100 pc to few kpc ) within galaxies @xcite . at the brightest end of fir luminosity ,
the relationship is observed to hold for ( ultra ) luminous infrared galaxies [ ( u)lirg ] and star - burst galaxies . at the faintest end it holds in dwarf galaxies @xcite .
it is believed that star - formation connects the two regimes of emission .
synchrotron ( also referred to as non - thermal ) emission in the radio band is caused by acceleration of cosmic ray electrons ( cres ) in the galactic magnetic field produced by supernova explosions of massive stars . in the fir
, the emission originates due to dust re - radiation , heated by ultraviolet ( uv ) photons from massive ( @xmath13 ) , short lived ( @xmath14 yrs ) stars .
however , the tightness seen in the correlation needs to be explained , as a number of independent physical quantities are responsible for the emission in each regime like , the magnetic field , number density of cres , energy losses of cres , star formation history , dust / gas density , dust absorption efficiency , etc .
several models have been proposed to explain the tightness seen in the radio fir correlation ( see e.g. ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
more recent models by @xcite and @xcite have shown that the above mentioned factors conspire to maintain the tightness observed for the global radio fir correlation .
observationally , it is important to assess the form of the radio fir correlation at high redshifts as it might depend on the evolution of ism parameters with redshift ( @xmath0 ) like synchrotron and inverse - compton losses , dust content , star formation rate , magnetic field strength and overall sed ( see e.g. ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
recently , @xcite predicted a modification of the form of the radio fir correlation , based on the observed relationship between magnetic field strength and star formation rate caused due to turbulent amplification of the magnetic field .
a breakdown in the correlation is expected depending on the dominant energy loss mechanism of the cres in the radio domain , i.e. , synchrotron , inverse - compton , bremsstrahlung and/or ionization losses .
typical ( 1@xmath15 ) sensitivity of most of the existing deep radio surveys are limited only to few tens of @xmath1jy ( see e.g. , @xcite [ vla - vvds ] ; @xcite [ vla - cosmos ] ; @xcite [ evla - stripe82 ] , etc . ) .
however , a few deeper surveys exists reaching 1@xmath15 sensitivity @xmath16jy ( see e.g. , @xcite [ e - cdfs ] ; @xcite [ goods - n ] ) .
these observations can detect normal galaxies ( @xmath17 ) up to redshift of @xmath18 at 1.4 ghz with @xmath19 sensitivity , making it difficult to study the radio fir correlation for such galaxies at higher redshifts .
the correlation has been studied for ( u)lirgs with higher luminosity ( @xmath20 ) up to redshifts of @xmath21 @xcite .
such galaxies can have significant contamination due to agns and compact nuclear starbursts . even in the case of relatively low optical depth , starburst related free
free absorption can give rise to substantial obscuration @xcite that can affect the form of the correlation .
it is therefore imperative to study the radio fir correlation for less extreme star - forming galaxies at _ high redshifts _ where the bulk of the radio and fir emission originates from star formation . in this paper
, we study the properties of the radio fir correlation , both the slope and the traditionally defined ` @xmath22 ' parameter , for a flux limited and color selected sample in the xmm - lss field .
we explore the correlation for blue star - forming galaxies up to @xmath3 employing the technique of image stacking . due to the inherent flux limitation of the parent sample ,
we detect normal star - forming galaxies up to @xmath23 and more luminous galaxies above that . for comparison , we study the correlation for luminous galaxies that are directly detected in this field up to @xmath24 0.95 . the paper is organized as follows : in section 2 , we describe our sample selection and data .
we discuss the technique of image stacking at 0.325 ghz and 1.4 ghz in the radio and at 24 , 70 , 160 , 250 , 350 and 500 @xmath1 m in the fir and the @xmath5correction method in section 3 .
we present our results in section 4 and discuss them in section 5 . throughout this paper , we assume a flat @xmath25cdm model with @xmath26 , @xmath27 and @xmath28 . | we study the radio far infrared ( fir ) correlation in `` blue cloud '' galaxies chosen from the prism multiobject survey ( primus ) up to redshift ( @xmath0 ) of 1.2 in the xmm - lss field .
we use rest - frame emission at 1.4 ghz in the radio and both monochromatic ( at 70@xmath1 m ) and bolometric ( between @xmath2 m ) emission in the fir . to probe the nature of the correlation up to @xmath3 , where direct detection of blue star - forming galaxies is impossible with current technology
, we employ the technique of image stacking at 0.325 and 1.4 ghz in the radio and in six infrared bands , viz .
24 , 70 , 160 , 250 , 350 and @xmath4 m . for comparison
, we also study the correlation for more luminous galaxies that are directly detected .
the stacking analysis allows us to probe the radio fir correlation for galaxies that are up to 2 orders of magnitude fainter than the ones detected directly .
the @xmath5correction in the infrared wavebands is obtained by fitting the observed spectral energy distribution ( sed ) with a composite mid - ir power law and a single temperature greybody model .
we find that the radio luminosity at 1.4 ghz ( @xmath6 ) is strongly correlated with monochromatic fir luminosity at 70 @xmath1 m ( @xmath7 ) having slope @xmath8 and with bolometric luminosity ( @xmath9 ) having slope @xmath10 .
the quantity @xmath11)$ ] is observed to decrease with redshift as @xmath12 probably caused due to the non - linear slope of the radio fir correlation . within the uncertainties of our measurement and the limitations of our flux - limited and color - selected sample
, we do not find any evolution of the radio fir correlation with redshift . |
it is well - established that the fraction of star - forming galaxies declines as a function of increasing local galaxy density in the low redshift universe .
also known as the star formation - density relation @xcite , this correlation has been confirmed in many studies , primarily using optical and uv data to trace star formation in massive galaxy clusters and field environments .
mid - infrared data from the infrared satellite observatory ( iso ) and the multi- band imaging photometer for spitzer ( mips ) have also revealed the presence of highly obscured , dusty star forming galaxies , previously undetected by optical or uv surveys ( e.g. * ? ? ?
* ; * ? ? ?
* ) . while the sensitivity of mips has enabled detailed studies of obscured star formation in individual local and distant galaxy clusters ( e.g. *
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , there are still only a small number of low redshift clusters that have been systematically surveyed for dusty star - forming galaxies out to the virial radius .
there remain many uncertainties in the relationship between star formation in clusters and their global cluster properties . in particular ,
several studies have tried to understand the correlation between cluster mass and the mass - normalized cluster star formation rate ( sfr ) .
while results from @xcite suggest that there is no strong correlation between cluster specific sfr and cluster mass , others such as @xcite , @xcite , and @xcite argue that cluster specific sfr decreases with cluster mass .
the large spatial coverage required to observe dusty star - forming galaxies in low redshift clusters out to the cluster infall regions has thus far hindered our ability to understand how star formation is affected by global cluster properties such as cluster mass . in this paper
we exploit data from the wide - field infrared survey explorer ( wise ; * ? ? ?
* ) to overcome this observational challenge and present results on obscured star formation and how it relates to cluster mass and radius out to 3 in a sample of 69 clusters at @xmath1 . and are commonly used interchangeably with virial radius and total cluster mass , respectively .
is the radius within which the average density is 200 times the critical density of the universe and is the mass enclosed within that radius .
wise is a medium - class explorer mission funded by nasa and has completed observations of the entire sky in four infrared bands : 3.4 , 4.6 , 12 , and 22 ( w1 to w4 , respectively ) .
wise scanned the sky with 8.8 second exposures , each with a 47 arcmin field of view , providing at least eight exposures per position on the ecliptic and increasing depth towards the ecliptic poles .
the individual frames were combined into coadded images with a pixel scale of 1.375 arcsec per pixel .
cosmic - rays and other transient features were removed via outlier pixel rejection .
the photometry used for our analyses is point spread function ( psf ) fitted magnitudes from the `` first - pass operations coadd source working database '' created by the wise data reduction pipeline .
galaxies in our cluster sample have a diffraction limited resolution of @xmath2 ( full width half maximum ) in the 22@xmath0 m band .
we have confirmed from w4 coadded images that all star - forming galaxies considered in our analyses appear unresolved in the 22 band , and have psf photometry reduced @xmath3 values less than 1.5
. therefore we use the psf magnitudes from the first - pass photometric catalog to obtain estimates of total flux . for the minimum coverage of 8 overlapping frames ,
the sensitivity for @xmath4 in the w4 band is 6 mjy , including uncertainty due to source confusion @xcite . to ensure an unbiased comparison of global sfrs and total ir luminosities of clusters at different redshifts , we impose a lower limit of sfr=4.6 on our entire cluster sample , which is equivalent to a total ir luminosity of @xmath5 , and corresponds to the 6 mjy flux limit at @xmath6 .
we hereafter refer to our sample of star - forming galaxies as demi - lirgs , which have nearly half the total ir luminosity of a luminous infrared galaxy or lirg .
however we note for future extragalactic studies using wise data that most coadded observations will have at least 12 overlapping frames and hence better sensitivity than the conservative 6 mjy limit we adopt in this paper .
additional information regarding wise data processing is available from the preliminary data release explanatory supplement .
we use the cluster infall regions ( cirs ; rines & diaferio 2006 ) sample because it provides high - fidelity mass estimates , is at sufficiently low redshift to enable detection with wise of strongly star - forming galaxies , and has extensive spectroscopy for membership determination .
the cirs sample consists of 72 low - redshift x - ray galaxy clusters identified from the rosat all - sky survey that are within the spectroscopic footprint of sdss data release 4 .
the redshift range of the cirs clusters is @xmath7 , with a median of @xmath8 .
cluster masses are available from @xcite , who utilize the caustics infall pattern to determine total dynamical cluster mass and @xcite . the clusters in this paper consist of the entire cirs sample , excluding three clusters at @xmath9 , which leaves 69 remaining clusters with a minimum redshift of @xmath10 .
optical photometric and spectroscopic data are obtained from the sloan digital sky survey data release 7 ( sdss dr7 ) @xcite , which are 90% spectroscopically complete for galaxies with @xmath11 and half - light surface brightness @xmath12 mag arcsec@xmath13 .
however , the spectroscopic completeness is lower in high - density regions such as in the core of galaxy clusters , due to fiber collisions .
adjacent fibers can not be placed closer than 55 arcsec from each other , which corresponds to a separation of 63 kpc at @xmath8 . to verify that the spectroscopic incompleteness arising from constraints on fiber placement has a negligible effect on radial trends of star formation in clusters , we look at the fraction of w4-bright sources as a function of distance from the cluster center . among the @xmath11 sample , the spectroscopic completeness of w4-bright sources ( w4 @xmath14 ) is @xmath1570% within the central 5 arcmin radial bin , and increases to @xmath1580% in the 5 - 10 arcmin bin . in other words
, we are not missing a significant fraction of star - forming galaxies in the cluster core due to fiber collisions .
the spectroscopic completeness for all galaxies ( regardless of ir emission ) with @xmath11 shows a similar behavior with distance from the cluster center , and on average reaches at least @xmath1580% completeness beyond the central 5 arcmin radius . since the spectroscopic incompleteness of both w4-bright and optically bright galaxies reaches @xmath1580% beyond the cluster core , our results on radial trends of star formation are negligibly affected by spectroscopic incompleteness .
in addition , we are measuring trends on scales of hundreds of kpc , and therefore not significantly affected by small differences in spectroscopic completeness near the cluster core .
we use photometric data from sdss dr7 to estimate stellar masses for cluster members , using the tight correlation between stellar mass - to - light ( m / l ) ratio and optical colors , as determined by @xcite . with a sample of more than 10,000 optically bright galaxies ( @xmath16 ) , @xcite construct a grid of stellar population models with a range of metallicities and star formation histories , then compare the best - fit galaxy templates with evolved zero redshift templates to determine present - day m / l ratios .
we use the relation between @xmath17 color and @xmath18-band stellar m / l ratio to derive stellar masses for our sample .
the estimated uncertainty of the color - based stellar m / l ratios , including random and systematic errors , is @xmath1545% @xcite .
spectroscopic redshifts from sdss dr7 are used to determine membership for each of our clusters .
figure [ fig : caustic ] illustrates an example of using the caustics infall method to determine cluster membership for abell 1377 , the most massive cluster in our sample .
it shows the difference in radial velocity ( @xmath19 ) of galaxies in the cluster field with respect to the cluster systemic velocity , as a function of projected distance from the cluster center .
galaxies that are dynamically bound to the cluster form a well - defined region that decreases in velocity offset as a function of projected radius .
we define the galaxies that are within the edge of this envelope and within a projected radius of @xmath203 , as cluster members . while the average turnaround radius for the cirs sample is @xmath155 , we restrict our cluster galaxy sample to within 3 , since cluster infall patterns are generally better defined closer in to the cluster center .
only galaxies with sdss spectroscopic redshifts are considered in the following analyses .
we also limit all cluster and field galaxies to be brighter than @xmath21 , which corresponds to the 90% spectroscopic completeness limit of @xmath22 at @xmath6 .
star formation rates and total infrared luminosities ( ) are determined from the wise 22 photometry , using the relations presented in @xcite that are calibrated for mips 24 data .
@xcite constructed model average spectral energy distribution ( sed ) templates from a sample of local lirgs and ulirgs and derive correlations between 24 flux , sfr , and .
the flux - sfr relation given in their equation 14 can be used for both mips 24 and wise 22 data @xcite .
the similarity of the two bandpasses are confirmed by @xcite who quantify the correlation between and the inferred mips 24 and wise 22 luminosities .
the mips and wise luminosities are inferred by applying a color - correction to the 18 akari flux , and is derived from fitting the ir sed templates of @xcite to all six akari bandpasses ( 9 , 18 , 65 , 90 , 140 , and 160 ) for @xmath15600 galaxies at @xmath1 .
the resulting correlations between the @xmath23 and @xmath24 are nearly identical , with only a @xmath154% offset .
therefore we proceed by using the flux - sfr and @xmath23 relations of @xcite , assuming the calibration determined from mips 24 data .
the total error attributed to the sfr is @xmath150.2 dex , which is dominated by scatter in the @xmath25- relation , and does not include uncertainties inherent in the assumed stellar initial mass function ( imf ) .
the imf adopted by @xcite is similar to the @xcite and @xcite imfs , which have relatively fewer low mass stars compared to a salpeter imf , and is more applicable for extragalactic star - forming regions @xcite .
a robust estimate of the cluster star formation rates requires that we first identify and exclude all 22@xmath0 m sources that are dominated by agn rather than star formation .
as demonstrated in @xcite , even a few ir - bright agn can significantly bias the inferred global star formation rate of a cluster .
there are several methods by which we can identify agn in our data set .
the first method relies upon wise colors .
similar to the agn wedge in spitzer / irac data @xcite , agn - dominated sources will have red colors in w1-w2 ( [ 3.4]-[4.6 ] ) .
specifically , we use the criteria [email protected] ( vega ) to identify candidate agn in the wise data set .
this color selection is similar to , though slightly bluer than , the agn selection determined in jarrett et al .
2011 ( submitted ) and stern et al .
2011 ( in prep ) .
while this single color cut is in general less robust than the full agn wedge , at the low redshifts that are the focus of this work , star - forming galaxies should have colors uniformly blueward of this threshold .
we illustrate in figure [ fig : bpt ] that our [email protected] criterion works well in selecting out agn from star - forming galaxies .
profile - fit photometry is used to determine the w1-w2 colors and agn exclusion . while some of the nearest galaxies will be resolved in these bands , we are interested only in the w1-w2 color , rather than in single - band photometry .
comparison of agn selection based on w1-w2 color from aperture photometry and profile - fit photometry show that the two methods are similarly successful in isolating agn .
however , the color cut based on profile - fit magnitudes has a slightly better overlap with the optically detected agn , discussed below . a second approach to agn identification is use of optical spectroscopy to identify sources that lie in the agn region of the baldwin - phillips - terlevich ( bpt ) diagram ( bpt ; * ?
here we use the bpt diagram as a cross - check on the wise color selection for the subset of sources .
figure [ fig : bpt ] shows the emission line ratios of [ oiii]/h@xmath27 and [ sii]/h@xmath28 , for a sample of 136 cluster members with @xmath29 , of which 27 have [email protected] .
boundaries from @xcite are shown as dotted lines and separate regions where narrow emission lines arise from the presence of seyferts , liners ( low ionization narrow emission line regions ) , and hii regions ( star - forming galaxies ) .
galaxies that are selected as agn candidates based purely on having a red wise color ( [email protected] ) are indicated as filled ( red ) circles .
figure [ fig : bpt ] shows that out of 22 optically identified seyferts , fourteen ( 65% ) are also identified as agn based on the wise [email protected] selection .
sources flagged as quasars ( or qsos ) in the sdss catalog are highlighted with ( purple ) triangle symbols , with all ten quasars being independently identified as agn using the wise color selection . since the bpt diagnostic is meant to classify galaxies based only on _ narrow _ emission line ratios , it is not surprising that nearly all of the sdss quasars are not properly diagnosed in the bpt diagram .
the h@xmath30 and h@xmath31 emission lines in eight out of the ten sdss quasars have broad wings relative to their neighboring [ oiii ] or [ sii ] emission lines .
overall , we conclude from figure [ fig : bpt ] that the wise color selection is an effective method of excluding agn from our sample of bright w4 sources , identifying 85% of the optically detected seyferts and quasars .
the two central questions that we aim to address are how the mean specific star formation rate of cluster galaxies ( mssfr ) depends upon location within a cluster , and how the total integrated star formation rate of a cluster depends upon cluster mass .
specifically , the mssfr is defined as the total star formation rate in cluster galaxies _ at _ a given projected radius as inferred from 22@xmath0 m photometry , divided by the total stellar mass of all cluster galaxies _ at _ that radius . for probing the total integrated star formation rate
we also define the integrated cluster quantity @xmath32 which is a useful mass - normalized measure of the total star formation rate within the infall region , and referred to as the cluster specific sfr ( cssfr ) . among the 69 clusters ,
a total of 136 demi - lirgs are detected within 3 , of which 27 are determined to be agn based on their w1-w2 color . in the following sections , all sfr quantities are determined from the remaining 109 star - forming demi - lirgs .
eight seyferts identified from the bpt diagram are included in the cluster star - forming galaxy sample because their w1-w2 colors are not indicative of agn activity , and we prefer to maintain a uniform wise selection of agn among all field and cluster galaxies .
we note that all results presented in this paper are negligibly affected by the exclusion / inclusion of these eight seyferts .
the coordinates , redshift , w4 magnitude , sfr , , projected distance from cluster center , and agn flags are listed for the 136 demi - lirgs in table 1 .
there is a long history in the literature demonstrating the existence of a strong radial dependence for star formation ( or color ) in galaxy clusters ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the main strengths of the current analysis are the uniform 22@xmath0 m wise data , spectroscopic completeness , and existence of measurements for the full sample , which together provide us with an infrared - based view of star formation for a homogeneous sample extending well beyond the virial radius .
a common approach in the literature has been to look at the radial dependence of the fraction of star - forming galaxies .
here we investigate both the dependence of the star - forming fraction , and also the mssfr out to 3 . for comparison ,
the identical quantities for a field population are calculated , with the field sample chosen from a catalog of galaxies located within a projected radius of 5 mpc from the cluster center , at the cluster redshift . among these galaxies , we choose those with a radial velocity greater than 5000 away from the cluster systemic velocity , redshift @xmath1 , and absolute r magnitude @xmath33 .
these are the same redshift and magnitude limits of the cluster galaxy sample .
the highest velocity dispersion of our cluster sample is @xmath15960 , which means that the chosen field galaxies are more than @xmath34 away from the cluster redshift .
figure [ fig : caustic ] shows delta - velocity ( @xmath19 ) versus projected distance from the cluster center for galaxies within a 5 mpc projected radius of abell 119 .
the field galaxies indicated with ( blue ) cross symbols are clearly not associated with the cluster galaxies , which appear distinctly confined to a trumpet - shaped region . by gathering field galaxies within a projected 5 mpc radius from 69 different regions of the sky ,
we have compiled a large enough sample to obtain a representative field value of mssfr and the fraction of star - forming galaxies with @xmath29 .
there are a total of 11180 field galaxies , of which 566 are demi - lirgs .
figure [ fig : ssfr_radius ] shows the mssfr as a function of r/ for our ensemble of 69 clusters , including star formation only for galaxies with @xmath29 l@xmath35 ( sfr@xmath36 m@xmath35 yr@xmath37 ) .
we emphasize that by construction our sfr limit therefore means that the observed mssfr is a lower bound but a consistent lower bound across the sample . as expected , the mean specific sfr increases with projected radius , with the most central bin containing the fewest star - forming demi - lirgs twelve out of a total of 109 demi - lirgs found in 69 clusters .
the mssfr displays a steep increase from the central bin to , then continues to increase monotonically and nearly flattens out below the field value at larger radii .
this is consistent with a low redshift study of h@xmath30 star - forming galaxies by @xcite , who found that the median sfr ( normalized by l@xmath38 ) of cluster galaxies reaches the field value beyond 3 .
such a radial trend can be driven by two factors an increase in the ssfr of the sub - population of star - forming galaxies , or an increase in the fraction of star - forming galaxies with radius .
we find that there is no statistically significant change in the ssfr of individual star - forming galaxies with radius , implying that the trend is driven primarily by a radial gradient in the star forming fraction .
this can be seen in figure [ fig : lirg_fraction ] , where we directly plot the fraction of star - forming galaxies as a function of radius .
more than one third of the star - forming galaxies reside at 2@xmath39r@xmath393 .
what is striking is that even at these large radii the mssfr and star - forming fraction remain below the field values . in figure
[ fig : lirg_fraction ] , the demi - lirg fraction is higher in the field relative to the cluster population at 3 by a factor of @xmath151.5 . one possible interpretation of this result is that these are infalling galaxies that may already be `` pre - processed '' in intermediate - density environments , such as galaxy groups or filaments ( e.g. zabludoff & mulchaey 1998 ) , which would explain why they are suppressed relative to a true field population .
there is both observational and theoretical support in favor of the suppressed star - forming galaxy fraction out to 3 .
@xcite found that the mean cluster sfr , which they derived from [ oii ] emission of galaxies in 15 clusters , is lower than the mean field sfr by more than a factor of two at 2 .
results from @xcite and @xcite similarly conclude that cluster star formation is suppressed at several times the virial radius relative to the field .
in addition to galaxies being pre - processed at large radii , another explanation for the suppressed demi - lirg fraction in figure [ fig : lirg_fraction ] could be the contribution of a `` backsplash '' population beyond .
simulations from @xcite and @xcite show that up to @xmath1550% of galaxies currently located between and 2 may have previously traveled inward of the virial radius of the cluster .
therefore the suppressed demi - lirg fraction at large radii may , to some degree , reflect quenched star formation that occurred within the cluster core .
however our galaxy sample , which has an absolute magnitude limit of @xmath33 , is sensitive to massive galaxies brighter than m@xmath40 @xcite , whereas backsplash galaxies are on average significantly less massive than first infall galaxies in the cluster outskirts @xcite . as such
, our sample is biased against the lower mass backsplash galaxies relative to first infall galaxies .
nonetheless , a contribution of backsplash galaxies may indeed be partially responsible for the suppressed star formation activity at large distances from the cluster core .
total mass is , both for astrophysics and cosmology , the most fundamental physical galaxy cluster parameter .
there are multiple physical processes that depend on the depth of the cluster potential well ( e.g. ram pressure , harassment , tidal interactions ) which can significantly alter the morphologies , gas content , and star formation rates in cluster galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
furthermore , it has been observed that while the total baryon fraction ( stellar and gas content ) remains roughly constant with cluster mass , the stellar mass decreases and gas content increases in galaxy clusters as a function of cluster mass @xcite .
this implies that the integrated star formation efficiency of the history of a cluster is directly tied to cluster mass . in this section
we examine the relation between cluster mass and total current cluster star formation to better understand how strongly the present - day star formation rate depends upon cluster mass .
our approach is to compute the cluster sfr normalized by cluster mass ( cssfr ; equation 1 ) as a function of cluster mass . for each cluster ,
the cluster sfr is the sum of sfrs for all member galaxies that have [email protected] and lie within a projected radius of 3 .
the left panel of figure [ fig : ssfr_m200 ] shows the cssfr for 62 clusters , which consists of the sample of 69 clusters , excluding seven clusters that have incomplete spatial coverage in the sdss dr7 spectroscopic survey within 3 .
the dotted curve shows the limiting detectable cssfr as a function of corresponding to a single demi - lirg with a sfr of 4.6 in a cluster .
clusters that have no members above the [email protected] limit are assigned a cssfr of zero .
the binned data ( large squares ) indicate the average cssfr in each bin , including clusters with a cssfr of zero .
thus the cssfr of the lowest bin is strongly biased by incompleteness , since nearly all of the clusters at low mass fall below the [email protected] limit and have a cluster sfr set to zero . to ensure that our results are not significantly biased by this incompleteness at low cluster mass
, we also show the cluster specific sfr versus for a sub - sample of clusters at @xmath41 in the right panel of figure [ fig : ssfr_m200 ] . for these 22 low - z clusters , we are sensitive to star - forming galaxies down to [email protected] , which corresponds to the w4 band 5@xmath42 detection limit of 6 mjy at @xmath8 , and is illustrated with a dotted curve .
using the sample of 62 clusters , the left panel of figure [ fig : ssfr_m200 ] shows no significant correlation between cluster specific sfr and cluster mass , over more than one order of magnitude in @xmath43 .
the first pass coadd processing used here has the same calibration as in the wise preliminary data release , but was run every other day using only one day of data , creating gaps in coverage at low ecliptic latitude .
these gaps should not significantly impact this result , since location of the gaps are random with respect to the cluster centers .
we confirm this expectation by examining the cluster specific sfrs of a sub - sample of 24 clusters with ecliptic latitude @xmath44 deg .
these clusters are at sufficiently high ecliptic latitude such that they have fairly complete spatial coverage with the first - pass wise coadd data .
the average cluster specific sfrs of the high ecliptic latitude sub - sample are overplotted in figure [ fig : ssfr_m200 ] in small purple diamonds .
these high ecliptic latitude clusters show no significant correlation between cluster specific sfr and cluster mass , similar to the lack of correlation seen with the full cluster sample . the lack of correlation between cssfr and is also confirmed with the @xmath41 sub - sample , for which we have applied a significantly lower detection limit of [email protected] ( figure [ fig : ssfr_m200 ] , right panel ) . with the lower sfr limit ,
the clusters in figure [ fig : ssfr_m200 ] have an average of 5.2 star - forming galaxies within r@xmath393 , with a large scatter .
five of the low redshift clusters have between 10 to 16 star - forming galaxies .
in contrast , for the sample of 62 clusters with the demi - lirg cut ( [email protected] ) , there is an average of 1.6 star - forming galaxies per cluster within r@xmath393 . while the average number of star - forming galaxies per cluster is small in this case , the cssfrs calculated using only the demi - lirgs are well correlated with the cssfrs using the [email protected] cut , with a linear pearson correlation coefficient of 0.9
. therefore , even with an average of only 1.6 star - forming galaxies per cluster , the cssfrs of the full cluster sample are qualitatively representative of cssfrs that sample the fainter end of the infrared luminosity function .
the binned data in both panels of figure [ fig : ssfr_m200 ] show cluster specific sfrs that are consistent within 1@xmath42 of each other , across the full range of .
the error bars associated with the average cssfr are the poisson uncertainty on the number of star - forming galaxies detected above the sfr limit in each bin .
the lack of a significant trend between cssfr and indicates that transformation mechanisms which scale strongly with cluster mass may not play a dominant role in the evolution of star formation in clusters .
while figure [ fig : ssfr_m200 ] illustrates the integrated cluster sfr per by taking the sum of all star - forming galaxies within r@xmath393 , our results remain robust when considering only those galaxies within r@xmath39 .
since galaxies within are more susceptible to global cluster processes such as ram pressure , the lack of correlation between cssfr ( r@xmath39 ) and cluster mass re - affirms the main conclusion drawn from figure [ fig : ssfr_m200 ] mechanisms that are strongly dependent on cluster mass do not play a significant role in the evolution of star formation in clusters .
two cluster processes that strongly scale with cluster mass are galaxy harassment and ram pressure .
ram pressure is a mechanism that acts to compress and/or strip cold gas reservoirs in galaxies .
the impact of ram pressure is proportional to @xmath45 , where @xmath46 is the density of the intracluster medium ( icm ) and @xmath47 is the velocity of a galaxy with respect to the icm @xcite .
galaxy harassment refers to the cumulative effect of high velocity close encounters between galaxies and has the potential to disturb morphologies and quench star formation @xcite .
the effects of galaxy harassment , like ram pressure , are expected to scale up with cluster mass , since both cluster velocity dispersion and icm gas density increase with cluster mass . while there has been evidence in support of both ram pressure and harassment having an impact on star formation in individual cluster galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , the lack of a correlation between cluster specific sfr and cluster mass in figure [ fig : ssfr_m200 ] indicates that neither of these mechanisms significantly trigger / quench star formation in local galaxy clusters , within the mass range of @xmath48 to @xmath49 .
our results are consistent with those of @xcite and @xcite , who use a sample of @xmath15100 clusters at @xmath1 and find no significant correlation between h@xmath30-derived cluster specific sfr and cluster mass . both @xcite and @xcite use sdss dr2 data , with the same limiting r - band magnitude as used in this paper .
the limiting h@xmath30 sfr is @xmath153 .
similarly , @xcite demonstrate that the passive galaxy fraction is roughly constant as a function of system mass , from @xmath50 to @xmath51 .
assuming that the non - passive galaxy fraction is dominated by star - forming galaxies rather than agn , the results of @xcite support our finding that cluster specific sfr is not strongly dependent on cluster mass .
in contrast , several other studies have found evidence for an anti - correlation between cluster specific sfr and cluster mass . @xcite and
@xcite obtain cluster sfrs by integrating h@xmath30 derived sfrs for members out to 0.5 , and cluster masses from observed velocity dispersions for a sample of eight and nine clusters , respectively .
their clusters span an approximate redshift range of 0.2 to 0.8 , and the h@xmath30 sfrs are sensitive to @xmath151 . the anti - correlation observed between cluster specific sfr and cluster mass
is significantly weakened when only clusters of the same epoch are considered , indicating that evolutionary effects may be the dominating factor .
in addition to interactions between galaxies and the cluster environment , studies have demonstrated that tidal interactions between galaxies or galaxy - galaxy mergers can trigger a burst of star formation ( e.g. * ? ? ?
* ; * ? ? ?
galaxy - galaxy interactions are generally optimized in intermediate density regions with low velocity dispersions , such as in galaxy groups , whereas the more massive clusters are less likely to host galaxy mergers .
figure [ fig : ssfr_m200 ] includes a substantial number of low mass systems ( 57 clusters with @xmath52 ) , yet shows no sign of significantly elevated cluster specific sfr in low mass clusters relative to intermediate mass clusters .
any signature of enhanced star formation in low mass clusters due to galaxy - galaxy interactions should be particularly evident in our data , because we integrate the sfr of cluster members out to three times the virial radius , where galaxy - galaxy mergers and tidal interactions are more like to occur .
a comparison of the distance to nearest neighbor among the demi - lirg population and the general cluster population as a function of projected radius shows that the two populations have mean nearest neighbor distances that are consistent within 1@xmath42 .
this indicates that galaxy - galaxy mergers or close tidal interactions do not play a dominant role in the star formation properties of these clusters .
our results imply that even in the low mass @xmath53 groups , star formation has already been quenched to similar levels observed in clusters that are more massive by over an order of magnitude .
the lack of correlation between cssfr and ( figure [ fig : ssfr_m200 ] ) , combined with the suppressed demi - lirg fraction at 3 relative to the field ( figure [ fig : lirg_fraction ] , would suggest a scenario in which star formation is quenched in a significant population of galaxies within small groups or filaments .
then once the remaining star - forming galaxies are accreted into the cluster environment , a mechanism that is not strongly dependent on cluster mass must operate .
we suggest strangulation as a plausible candidate for quenching star formation in cluster star - forming galaxies . as a galaxy enters the cluster icm , its hot halo gas is removed , thereby cutting off its resource for future cold gas supplies , since the halo gas would otherwise eventually cool and settle onto the disk @xcite .
the effectiveness of strangulation is less dependent on cluster mass than ram pressure or harassment .
simulations from @xcite have shown that strangulation can occur in low mass groups , where the icm - galaxy halo interaction is weak relative to clusters .
strangulation has been suggested by several cluster studies as an important cluster mechanism ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , including by @xcite who discovered `` mild gradients '' in the morphological fractions of galaxies in cl0024 + 16 out to large cluster radii , which were best explained by a slow - working gentle mechanism such as strangulation . while strangulation may quench star formation in massive clusters ,
our results suggest that the dominant process for regulating star formation in dense environments does not depend upon cluster mass , and hence must be efficient in small groups or filaments .
we use data from wise and sdss dr7 to study the dependence of star formation on cluster mass and location within a cluster in 69 local ( @xmath1 ) galaxy clusters .
cluster membership is determined from sdss dr7 spectroscopic redshifts for galaxies brighter than @xmath21 , and star formation rates are determined with 22 photometry from wise using the relations outlined in @xcite .
out of 69 clusters , we find a total of 109 star - forming demi - lirgs with sfr@xmath36 within 3 .
both the fraction of demi - lirgs and the mean specific sfr of cluster galaxies increases with projected distance from the cluster center .
however , the fraction of demi - lirgs remains below the field value even at 3 times the cluster virial radius .
one plausible explanation for the suppressed demi - lirg fraction at 3 is that even galaxies that reside significantly beyond the virial radius have already been quenched in their previous environments , such as small groups or filaments .
we also investigate the impact of cluster mass on star formation by presenting the total cluster sfr normalized by as a function of .
we find no evidence of a correlation between the cluster specific sfr and cluster mass in this first uniform dataset to detect obscured star formation out to several times the virial radius in a large sample of low redshift clusters .
our result indicates that cluster mechanisms which scale with cluster mass , such as ram pressure or harassment , are not likely to play a dominant role in the evolution of star formation in local clusters .
this publication makes use of data products from the wide - field infrared survey explorer , which is a joint project of the university of california , los angeles , and the jet propulsion laboratory / california institute of technology , funded by the national aeronautics and space administration .
the authors would like to thank emilio donoso for his help and advice on navigating the sdss database .
we also thank the anonymous referee for a careful reading and comments which improved the paper .
wisepc j235654.30 - 101605.2 & 359.22626 & -10.26812 & 0.074 & 4.12 & 0.025 & 90.3 & 7.2e+11 & 1.1 & agn + wisepc j155850.42 + 272324.5 & 239.71010 & 27.39013 & 0.093 & 4.76 & 0.024 & 86.7 & 7.0e+11 & 1.7 & agn + wisepc j121635.79 + 040709.2 & 184.14912 & 4.11922 & 0.076 & 4.62 & 0.027 & 58.5 & 4.9e+11 & 2.1 & agn + wisepc j130534.25 - 021119.1 & 196.39270 & -2.18864 & 0.088 & 5.41 & 0.039 & 38.8 & 3.4e+11 & 0.2 & agn + wisepc j121742.00 + 034631.2 & 184.42499 & 3.77535 & 0.080 & 5.25 & 0.032 & 35.6 & 3.1e+11 & 1.8 & agn + wisepc j160003.13 + 263707.8 & 240.01305 & 26.61884 & 0.093 & 5.67 & 0.033 & 35.1 & 3.1e+11 & 2.0 & agn + wisepc j111519.95 + 542316.6 & 168.83313 & 54.38794 & 0.070 & 4.98 & 0.026 & 34.2 & 3.0e+11 & 1.5 & agn + wisepc j171031.49 + 643914.6 & 257.63123 & 64.65405 & 0.079 & 5.41 & 0.026 & 29.7 & 2.6e+11 & 0.3 & sf + wisepc j134236.22 + 592324.7 & 205.65091 & 59.39019 & 0.071 & 5.15 & 0.030 & 29.7 & 2.6e+11 & 0.5 & agn + wisepc j170226.20 + 341117.5 & 255.60916 & 34.18821 & 0.105 & 6.16 & 0.035 & 28.9 & 2.6e+11 & 0.1 & agn + wisepc j111448.38 + 401749.3 & 168.70160 & 40.29702 & 0.076 & 5.47 & 0.032 & 25.8 & 2.3e+11 & 1.2 & sf * + wisepc j162021.16 + 301020.7 & 245.08818 & 30.17243 & 0.096 & 6.14 & 0.037 & 23.7 & 2.2e+11 & 2.6 & agn + wisepc j160105.12 + 272539.2 & 240.27135 & 27.42756 & 0.087 & 5.91 & 0.034 & 23.3 & 2.1e+11 & 2.0 & sf + wisepc j170100.39 + 342042.9 & 255.25163 & 34.34524 & 0.101 & 6.27 & 0.039 & 23.1 & 2.1e+11 & 0.2 & sf + wisepc j232514.19 + 151442.1 & 351.30911 & 15.24503 & 0.043 & 4.28 & 0.021 & 21.8 & 2.0e+11 & 0.3 & sf + wisepc j165948.48 + 335944.1 & 254.95201 & 33.99559 & 0.085 & 5.96 & 0.036 & 20.4 & 1.9e+11 & 0.1 & sf + wisepc j152138.79 + 305037.4 & 230.41161 & 30.84372 & 0.081 & 6.07 & 0.040 & 16.2 & 1.5e+11 & 1.0 & sf + wisepc j104319.01 + 050818.0 & 160.82922 & 5.13833 & 0.068 & 5.70 & 0.041 & 15.4 & 1.5e+11 & 0.8 & sf
+ wisepc j151941.93 + 312905.5 & 229.92471 & 31.48486 & 0.080 & 6.11 & 0.031 & 15.2 & 1.4e+11 & 1.1 & sf + wisepc j221445.86 - 092300.8 & 333.69110 & -9.38356 & 0.082 & 6.24 & 0.051 & 14.2 & 1.4e+11 & 0.4 & agn + wisepc j102200.74 + 382914.4 & 155.50308 & 38.48734 & 0.057 & 5.39 & 0.030 & 14.0 & 1.3e+11 & 0.5 & sf + wisepc j130330.80 - 021400.0 & 195.87834 & -2.23334 & 0.084 & 6.38 & 0.049 & 13.2 & 1.3e+11 & 0.1 & agn + wisepc j152021.87 + 485222.0 & 230.09113 & 48.87279 & 0.074 & 6.09 & 0.033 & 12.8 & 1.2e+11 & 1.5 & sf + wisepc j170957.92 + 335507.3 & 257.49133 & 33.91869 & 0.084 & 6.42 & 0.042 & 12.5 & 1.2e+11 & 0.2 & sf + wisepc j004336.31 - 092547.6 & 10.90130 & -9.42988 & 0.050 & 5.29 & 0.030 & 11.3 & 1.1e+11 & 0.3 & sf + wisepc j155857.04 + 273758.0 & 239.73769 & 27.63278 & 0.090 & 6.70 & 0.056 & 11.2 & 1.1e+11 & 1.7 & sf + wisepc j170231.67 + 335135.6 & 255.63196 & 33.85988 & 0.087 & 6.61 & 0.051 & 11.2 & 1.1e+11 & 2.6 & sf + wisepc j110018.01 + 100256.9 & 165.07504 & 10.04914 & 0.036 & 4.57 & 0.027 & 10.7 & 1.1e+11 & 0.8 & agn + wisepc j101346.84 - 005450.9 & 153.44518 & -0.91414 & 0.042 & 4.96 & 0.028 & 10.6 & 1.0e+11 & 0.7 & agn + wisepc j163032.67 + 392303.2 & 247.63612 & 39.38421 & 0.030 & 4.21 & 0.024 & 10.5 & 1.0e+11 & 2.2 & agn + wisepc j122940.27 + 121743.6 & 187.41780 & 12.29546 & 0.087 & 6.72 & 0.155 & 10.2 & 1.0e+11 & 0.1 & sf + wisepc j114623.86 + 552422.2 & 176.59941 & 55.40616 & 0.053 & 5.53 & 0.024 & 10.0 & 9.9e+10 & 1.7 & sf + wisepc j005656.90 - 011242.4 & 14.23707 & -1.21176 & 0.050 & 5.45 & 0.036 & 9.6 & 9.6e+10 & 0.3 & sf + wisepc j162143.25 + 294332.6 & 245.43019 & 29.72572 & 0.098 & 7.09 & 0.070 & 9.5 & 9.4e+10 & 2.8 & sf + wisepc j102126.42 + 381747.5 & 155.36008 & 38.29652 & 0.055 & 5.72 & 0.038 & 9.1 & 9.1e+10 & 1.2 & sf * + wisepc j165903.80 + 334854.6 & 254.76585 & 33.81516 & 0.086 & 6.81 & 0.062 & 8.9 & 8.8e+10 & 0.2 & sf + wisepc j171447.38 + 643541.1 & 258.69742 & 64.59475 & 0.080 & 6.67 & 0.050 & 8.7 & 8.7e+10 & 0.4 & sf + wisepc j132632.23 + 002800.7 & 201.63428 & 0.46685 & 0.086 & 6.84 & 0.060 & 8.7 & 8.7e+10 & 0.3 & sf + wisepc j075225.73 + 283040.0 & 118.10722 & 28.51112 & 0.062 & 6.08 & 0.044 & 8.5 & 8.5e+10 & 0.3 & agn + wisepc j103342.72 + 392926.9 & 158.42801 & 39.49082 & 0.068 & 6.32 & 0.049 & 8.3 & 8.4e+10 & 1.2 & sf + wisepc j132944.92 - 014239.7 & 202.43716 & -1.71104 & 0.089 & 6.98 & 0.077 & 8.2 & 8.2e+10 & 0.8 & sf + wisepc j162345.89 + 410456.5 & 245.94122 & 41.08235 & 0.034 & 4.70 & 0.016 & 8.1 & 8.1e+10 & 2.1 & agn + wisepc j125830.21 - 015835.4 & 194.62585 & -1.97650 & 0.080 & 6.75 & 0.073 & 8.1 & 8.1e+10 & 0.2 & sf * + wisepc j115720.93 + 051506.0 & 179.33719 & 5.25168 & 0.081 & 6.80 & 0.064 & 8.0 & 8.0e+10 & 1.6 & sf + wisepc j152003.58 + 304350.6 & 230.01491 & 30.73073 & 0.082 & 6.84 & 0.208 & 7.9 & 7.9e+10 & 1.7 & sf + wisepc
j155637.07 + 270043.1 & 239.15446 & 27.01196 & 0.091 & 7.11 & 0.079 & 7.7 & 7.7e+10 & 1.6 & sf + wisepc j154351.50 + 363136.9 & 235.96458 & 36.52691 & 0.067 & 6.38 & 0.033 & 7.6 & 7.7e+10 & 1.4 & agn + wisepc j121754.98 + 040117.5 & 184.47910 & 4.02153 & 0.082 & 6.87 & 0.108 & 7.6 & 7.7e+10 & 2.1 & sf + wisepc j102946.81 + 401913.7 & 157.44505 & 40.32048 & 0.067 & 6.39 & 0.047 & 7.6 & 7.6e+10 & 1.3 & agn + wisepc j162637.09 + 390739.3 & 246.65456 & 39.12757 & 0.035 & 4.89 & 0.016 & 7.5 & 7.6e+10 & 2.6 & sf + wisepc j123011.93 + 120632.8 & 187.54971 & 12.10911 & 0.085 & 6.96 & 0.180 & 7.5 & 7.6e+10 & 0.1 & sf + wisepc j112344.58 + 030018.9 & 170.93575 & 3.00525 & 0.051 & 5.75 & 0.036 & 7.5 & 7.5e+10 & 1.8 & sf + wisepc j152255.30 + 305905.4 & 230.73042 & 30.98484 & 0.080 & 6.83 & 0.073 & 7.4 & 7.5e+10 & 1.4 & sf * + wisepc j170102.30 + 340400.7 & 255.25958 & 34.06686 & 0.094 & 7.23 & 0.069 & 7.4 & 7.5e+10 & 0.2 & agn + wisepc j155842.84 + 270736.5 & 239.67848 & 27.12680 & 0.086 & 7.03 & 0.060 & 7.3 & 7.4e+10 & 1.6 & sf + wisepc j115604.23 + 050150.7 & 179.01761 & 5.03075 & 0.079 & 6.81 & 0.075 & 7.3 & 7.4e+10 & 2.1 & sf * + wisepc j152420.43 + 295719.7 & 231.08514 & 29.95546 & 0.076 & 6.72 & 0.049 & 7.3 & 7.3e+10 & 1.5 & sf + wisepc j075213.93 + 292023.5 & 118.05803 & 29.33986 & 0.061 & 6.22 & 0.034 & 7.2 & 7.3e+10 & 0.4 & sf + wisepc j133742.55 + 585209.8 & 204.42731 & 58.86938 & 0.074 & 6.68 & 0.078 & 7.2 & 7.2e+10 & 0.9 & sf + wisepc j130344.14 - 030652.2 & 195.93390 & -3.11451 & 0.081 & 6.91 & 0.081 & 7.1 & 7.2e+10 & 0.4 & sf + wisepc j123018.51 + 113811.4 & 187.57713 & 11.63651 & 0.083 & 6.99 & 0.101 & 7.0 & 7.1e+10 & 2.5 & sf + wisepc j155711.60 + 274455.2 & 239.29832 & 27.74868 & 0.088 & 7.12 & 0.073 & 7.0 & 7.1e+10 & 2.1 & sf + wisepc j011242.10 + 010839.7 & 18.17543 & 1.14435 & 0.044 & 5.51 & 0.033 & 6.9 & 7.0e+10 & 0.5 & sf + wisepc j161241.86 + 484805.8 & 243.17441 & 48.80160 & 0.059 & 6.16 & 0.034 & 6.9 & 7.0e+10 & 2.0 & sf + wisepc j133940.58 + 590307.8 & 204.91910 & 59.05216 & 0.073 & 6.70 & 0.081 & 6.9 & 6.9e+10 & 0.8 & sf + wisepc j165915.56 + 331019.5 & 254.81482 & 33.17208 & 0.087 & 7.13 & 0.072 & 6.8 & 6.9e+10 & 0.0 & sf + wisepc j171424.22 + 633937.8 & 258.60092 & 63.66050 & 0.081 & 6.95 & 0.110 & 6.8 & 6.9e+10 & 0.1 & sf + wisepc j135135.91 + 045100.4 & 207.89963 & 4.85010 & 0.076 & 6.78 & 0.069 & 6.8 & 6.9e+10 & 1.1 & sf + wisepc j132633.84 + 000924.7 & 201.64101 & 0.15686 & 0.086 & 7.09 & 0.061 & 6.8 & 6.9e+10 & 0.5 & sf + wisepc j111459.27 + 402142.0 & 168.74695 & 40.36167 & 0.073 & 6.70 & 0.060 & 6.8 & 6.9e+10 & 0.8 & sf + wisepc j112016.54
+ 540624.1 & 170.06892 & 54.10668 & 0.072 & 6.68 & 0.047 & 6.7 & 6.8e+10 & 1.4 & sf + wisepc j170522.43 + 334145.0 & 256.34344 & 33.69582 & 0.090 & 7.23 & 0.077 & 6.6 & 6.7e+10 & 0.0 & sf + wisepc j152043.23 + 304122.7 & 230.18015 & 30.68964 & 0.077 & 6.86 & 0.058 & 6.5 & 6.6e+10 & 0.9 & agn + wisepc j154256.63 + 351602.8 & 235.73595 & 35.26744 & 0.070 & 6.64 & 0.041 & 6.5 & 6.6e+10 & 1.7 & sf + wisepc j170132.94 + 331901.1 & 255.38724 & 33.31697 & 0.088 & 7.18 & 0.069 & 6.4 & 6.6e+10 & 2.5 & agn + wisepc j152227.36 + 303446.0 & 230.61401 & 30.57945 & 0.078 & 6.90 & 0.051 & 6.4 & 6.6e+10 & 1.6 & sf + wisepc j134227.74 + 585930.3 & 205.61560 & 58.99174 & 0.072 & 6.73 & 0.057 & 6.3 & 6.5e+10 & 0.6 & sf + wisepc j235425.88 - 095923.4 & 358.60782 & -9.98984 & 0.076 & 6.88 & 0.067 & 6.2 & 6.4e+10 & 0.9 & sf + wisepc j151933.94 + 303057.6 & 229.89142 & 30.51601 & 0.078 & 6.93 & 0.071 & 6.2 & 6.3e+10 & 1.0 & sf + wisepc j152418.93 + 295638.8 & 231.07889 & 29.94411 & 0.076 & 6.89 & 0.056 & 6.1 & 6.3e+10 & 1.2 & sf + wisepc j111250.48 + 015616.5 & 168.21034 & 1.93791 & 0.076 & 6.91 & 0.077 & 6.1 & 6.2e+10 & 1.5 & sf + wisepc j114951.78 + 560852.8 & 177.46574 & 56.14800 & 0.050 & 5.93 & 0.040 & 6.0 & 6.2e+10 & 1.5 & sf + wisepc j162549.24 + 402042.8 & 246.45515 & 40.34522 & 0.029 & 4.70 & 0.028 & 5.9 & 6.0e+10 & 2.6 & sf + wisepc j111925.05 + 545849.5 & 169.85439 & 54.98043 & 0.071 & 6.78 & 0.056 & 5.9 & 6.0e+10 & 1.6 & sf + wisepc j122747.50 + 120322.3 & 186.94794 & 12.05620 & 0.088 & 7.28 & 0.073 & 5.9 & 6.0e+10 & 0.1 & sf + wisepc j111459.91 + 024551.0 & 168.74963 & 2.76418 & 0.077 & 6.95 & 0.056 & 5.9 & 6.0e+10 & 1.3 & sf + wisepc j162253.90 + 402947.0 & 245.72458 & 40.49639 & 0.031 & 4.85 & 0.026 & 5.9 & 6.0e+10 & 2.6 & sf + wisepc j155849.70 + 272639.2 & 239.70708 & 27.44421 & 0.085 & 7.20 & 0.076 & 5.9 & 6.0e+10 & 1.8 & sf + wisepc j152308.37 + 304847.1 & 230.78488 & 30.81309 & 0.072 & 6.82 & 0.064 & 5.9 & 6.0e+10 & 1.4 & sf + wisepc j102815.22 + 400800.0 & 157.06342 & 40.13334 & 0.067 & 6.65 & 0.064 & 5.8 & 5.9e+10 & 0.8 & sf + wisepc j141309.46 + 442850.4 & 213.28941 & 44.48066 & 0.090 & 7.37 & 0.082 & 5.7 & 5.9e+10 & 1.1 & sf + wisepc j011703.58 + 000027.4 & 19.26492 & 0.00762 & 0.046 & 5.76 & 0.029 & 5.7 & 5.9e+10 & 0.5 & agn + wisepc j111341.36 + 412319.9 & 168.42235 & 41.38885 & 0.073 & 6.88 & 0.062 & 5.7 & 5.9e+10 & 0.8 & sf + wisepc j102134.50 + 410605.2 & 155.39377 & 41.10143 & 0.092 & 7.42 & 0.103 & 5.7 & 5.8e+10 & 0.5 & sf + wisepc j132459.25 + 110431.7 & 201.24689 & 11.07547 & 0.084 & 7.21 & 0.142 & 5.7 & 5.8e+10 & 0.5 & sf + wisepc j111512.08 + 542741.4 & 168.80035 & 54.46151 & 0.066 & 6.64 & 0.057 & 5.7 & 5.8e+10 & 1.5 & sf + wisepc j133450.77 + 593713.3 & 203.71155 & 59.62035 & 0.076 & 6.99 & 0.075 & 5.6 & 5.8e+10 & 0.9 & sf + wisepc j132054.04 + 113912.6 & 200.22517 & 11.65351 & 0.095 & 7.51 & 0.147 & 5.6 & 5.8e+10 & 0.6 & sf + wisepc j103130.85
+ 401450.9 & 157.87852 & 40.24746 & 0.066 & 6.66 & 0.058 & 5.6 & 5.7e+10 & 1.3 & sf + wisepc j155844.97 + 270812.4 & 239.68739 & 27.13678 & 0.095 & 7.53 & 0.095 & 5.5 &
5.7e+10 & 1.8 & sf + wisepc j104115.63 + 045313.8 & 160.31512 & 4.88717 & 0.068 & 6.76 & 0.063 & 5.5 & 5.6e+10 & 1.3 & agn + wisepc j162036.15 + 335643.1 & 245.15063 & 33.94529 & 0.031 & 4.97 & 0.026 & 5.4 & 5.6e+10 & 2.1 & sf + wisepc
j130306.96 - 031846.9 & 195.77899 & -3.31304 & 0.086 & 7.30 & 0.112 & 5.4 & 5.6e+10 & 0.2 & sf + wisepc j115809.58 + 555140.0 & 179.53992 & 55.86110 & 0.063 & 6.57 & 0.084 & 5.4 & 5.5e+10 & 1.8 & sf + wisepc j171248.41 + 634258.0 & 258.20169 & 63.71612 & 0.075 & 7.01 & 0.067 & 5.4 & 5.5e+10 & 0.3 & sf + wisepc j115938.06 + 561702.5 & 179.90857 & 56.28402 & 0.070 & 6.85 & 0.080 & 5.3 & 5.4e+10 & 2.2 & sf * + wisepc j121734.75 + 035019.8 & 184.39478 & 3.83884 & 0.073 & 6.96 & 0.110 & 5.3 & 5.4e+10 & 1.9 & sf + wisepc j155745.93 + 274043.6 & 239.44136 & 27.67878 & 0.090 & 7.46 & 0.096 & 5.2 & 5.4e+10 & 1.7 & sf + wisepc j101051.19 - 003842.5 & 152.71327 & -0.64514 & 0.043 & 5.75 & 0.076 & 5.2 & 5.4e+10 & 0.8 & sf * + wisepc j130510.25 - 023516.0 & 196.29272 & -2.58778 & 0.088 & 7.41 & 0.133 & 5.2 & 5.4e+10 & 0.4 & sf + wisepc j162858.00 + 391909.3 & 247.24168 & 39.31924 & 0.034 & 5.18 & 0.027 & 5.2 & 5.3e+10 & 2.4 & sf + wisepc j151848.57 + 305449.2 & 229.70239 & 30.91367 & 0.078 & 7.13 & 0.064 & 5.2 & 5.3e+10 & 1.2 & sf + wisepc j123001.03 + 113940.5 & 187.50427 & 11.66124 & 0.082 & 7.24 & 0.124 & 5.2 & 5.3e+10 & 2.2 & sf + wisepc j133105.70 + 584939.1 & 202.77376 & 58.82752 & 0.073 & 6.97 & 0.065 & 5.1 & 5.3e+10 & 0.8 & sf + wisepc j152428.76 + 310750.4 & 231.11983 & 31.13067 & 0.078 & 7.14 & 0.094 & 5.1 & 5.2e+10 & 1.1 & agn + wisepc j103954.63 + 051546.3 & 159.97762 & 5.26285 & 0.064 & 6.69 & 0.069 & 5.1 & 5.2e+10 & 1.7 & sf + wisepc j132422.33 + 105047.7 & 201.09305 & 10.84658 & 0.093 & 7.59 & 0.175 & 5.0 & 5.2e+10 & 0.7 & sf + wisepc j082926.09 + 302528.8 & 127.35872 & 30.42466 & 0.051 & 6.17 & 0.047 & 5.0 & 5.2e+10 & 0.7 & sf + wisepc j111503.18 + 542856.5 & 168.76323 & 54.48237 & 0.067 & 6.80 & 0.076 & 5.0 & 5.2e+10 & 1.5 & sf + wisepc j155957.42 + 272745.9 & 239.98924 & 27.46274 & 0.095 & 7.63 & 0.124 & 5.0 & 5.1e+10 & 1.8 & agn + wisepc j171351.15 + 640012.7 & 258.46313 & 64.00354 & 0.088 & 7.45 & 0.179 & 5.0 & 5.1e+10 & 0.2 & sf + wisepc j155930.98 + 273257.4 & 239.87907 & 27.54927 & 0.091 & 7.53 & 0.132 & 4.9 & 5.1e+10 & 1.9 & sf + wisepc j235345.66 - 103353.6 & 358.44025 & -10.56490 & 0.078 & 7.19 & 0.083 & 4.9 & 5.1e+10 & 0.7 & sf + wisepc j115610.78 + 045943.3 & 179.04491 & 4.99536 & 0.079 & 7.24 & 0.104 & 4.8 & 5.0e+10 & 1.6 & agn + wisepc j122915.88 + 113858.6 & 187.31615 & 11.64962 & 0.088 & 7.49 & 0.160 & 4.8 & 4.9e+10 & 2.5 & sf + wisepc j235514.75 - 102946.9 & 358.81146 & -10.49636 & 0.082 & 7.33 & 0.105 & 4.7 & 4.9e+10 & 0.6 & sf + wisepc j133111.02 - 014338.6 & 202.79590 & -1.72740 & 0.084 & 7.39 & 0.096 & 4.7 & 4.9e+10 & 0.7 & sf + wisepc j235208.99 - 095434.4 & 358.03748 & -9.90957 & 0.077 & 7.19 & 0.082 & 4.7 & 4.9e+10 & 0.6 & sf * + wisepc j134033.01 + 015016.0 & 205.13754 & 1.83778 & 0.078 & 7.24 & 0.094 & 4.7 & 4.9e+10 & 0.5 & sf + wisepc j130335.10 - 025417.0 & 195.89626 & -2.90471 & 0.083 & 7.38 & 0.110 & 4.7 & 4.8e+10 & 0.2 & sf + wisepc j130329.04 - 025337.2 & 195.87102 & -2.89366 & 0.079 & 7.27 & 0.092 & 4.7 & 4.8e+10 & 0.3 & sf + wisepc j122902.48 + 114340.3 & 187.26031 & 11.72786 & 0.086 & 7.49 & 0.243 & 4.6 & 4.8e+10 & 2.3 & sf + wisepc j135143.98 + 045833.5 & 207.93326 & 4.97596 & 0.077 & 7.20 & 0.089 & 4.6 & 4.8e+10 & 0.5 & sf + wisepc j170436.61 + 334454.2 & 256.15256 & 33.74839 & 0.090 & 7.60 & 0.099 & 4.6 & 4.8e+10 & 3.0 & sf + wisepc j121734.57 + 033450.9 & 184.39404 & 3.58081 & 0.074 & 7.13 & 0.089 & 4.6 & 4.8e+10 & 1.8 & sf + wisepc j123102.64 + 113929.5 & 187.76102 & 11.65819 & 0.087 & 7.52 & 0.126 & 4.6 & 4.7e+10 & 2.8 & sf + [ table : demi_lirgs ] | we present results from a systematic study of star formation in local galaxy clusters using 22@xmath0 m data from the wide - field infrared survey explorer ( wise ) .
the 69 systems in our sample are drawn from the cluster infall regions survey ( cirs ) , and all have robust mass determinations .
the all - sky wise data enables us to quantify the amount of star formation , as traced by 22@xmath0 m , as a function of radius well beyond , and investigate the dependence of total star formation rate upon cluster mass .
we find that the fraction of star - forming galaxies increases with cluster radius but remains below the field value even at 3 .
we also find that there is no strong correlation between the mass - normalized total specific star formation rate and cluster mass , indicating that the mass of the host cluster does not strongly influence the total star formation rate of cluster members . |
asteroseismology was born as a natural extension of helioseismology ( for a recent review see @xcite ) .
the success of helioseismology results from the ability to extract information from the sun s interior through accurately measured frequencies of more than 7000 acoustic oscillation modes ( e.g. , @xcite ; @xcite ; @xcite ) .
the development of this discipline was instrumental to improve the physics of the solar standard model that led to the resolution of the solar neutrino problem ( e.g. , ; @xcite ) .
similarly , asteroseismology is expected to extract information from the interiors of stars at different stages of their evolution , through the study of their observed oscillation spectrum . however , in this case , only global , low degree modes are expected to be detected , due to observational limitations . in the next decade , an unprecedented progress is expected in this research field , in part due to the large amount of data that has been made available by the missions _
@xcite and _ kepler _ @xcite .
up until now these missions have already detected oscillations in more than 500 stars @xcite .
these seismic data are obtained with an unprecedented high precision , which opens the possibility of studying new and old pulsating stars , as well as to test the theories of stellar evolution and probe for the existence of new physics inside stars @xcite .
a primordial target in asteroseismic studies is the class of sun - like stars , since these stars are expected to have an internal structure and acoustic spectrum of oscillations identical to the sun .
there are a few sun - like pulsations for which a few dozen global acoustic modes were successfully identified , typically acoustic modes with a degree smaller than @xmath3 .
in fact , these global oscillations are particularly interesting to study because these modes travel from the surface up to the core of the star , and thus these acoustic modes carry information about the entire structure of the star . moreover , the interest of studying this class of stars is twofold : ( 1 ) to better understand the physics behind stellar evolution in general ; and ( 2 ) to better describe the physics occurring inside our own sun . like the sun , in many
sun - like stars , the intricate surface structure where there are strong interactions among the magnetic fields , the turbulent convection and the rotation , as well the steady flows of meridional circulation , allow the possibility of local glitches that can produce visible perturbations in the spectrum of stellar oscillations . it is our intent to test this theoretical scenario thoroughly and without the loss of generality .
we have focused the work in a sun - like star which is the well known @xmath1 centauri a. we may have found some evidence of a perturbation of this type occurring in the oscillation spectrum of this star . in this work , we discuss the structure of the upper layers of a sun - like star , as it is possible to be inferred from the most complete observational table of frequencies .
in particular , we apply the acoustic phase shift method ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* for detaills ) to probe the physical mechanisms operating in these upper layers .
in this sun - like star , we show that the existence of a rapid variation ( rv ) layer in the external region of the star will produce a significant amount of scattering of acoustic waves .
this relatively thin layer is quite likely produced by an rv in the sound speed , which is similar to variations of sound speed that can occur in transition layers such as the well known examples in the sun of the bottom of the convection zone and the he ii ionization zone .
the seismic diagnostic is done by means of the seismic parameter , which allows us to infer the scattering of acoustic waves in the surface of the star using the available table of observational frequencies . in section 2 ,
different methods are shown to compute the scattering of acoustic waves in the upper layer of the star , and the impact of a rv of sound speed produced in the acoustic potential and consequently in the acoustic phase shift . in section 3
we apply this method to a toy model of a sun - like star , @xmath1 centauri a , and discuss the seismic diagnostics obtained from the observational data available for this star .
a comparison is made between the theoretical data and the observational data .
this comparison suggests that the internal structure of the stars could be quite different from what is predicted by the theoretical evolutionary models .
it specifically indicates the presence of a thin layer where the sound speed undergoes an abrupt variation .
finally , in section 4 , we present a summary and the conclusions of this work .
the asymptotic theory of adiabatic , non - radial oscillations has been used and developed in the past , based on the fact that the oscillation modes have wavelengths shorter than the scale of local variations of the background state . however , in the regions near the stellar surface , this approximation is no longer valid .
in fact , the thermal and dynamical time scales in the stellar surface are of the same order of magnitude , contrary to what happens in the interior of a sun - like star , for which the former is much larger than the latter .
consequently , the classical asymptotic theory gives a poor account of the propagation of waves in these more external layers .
strictly speaking , this means that the oscillations are no longer adiabatic , and an important exchange of energy occurs between the propagating wave and the background state . to solve this problem ,
@xcite has introduced a frequency dependency on phase shift to take into account the scattering of waves in these external layers .
this phase shift frequency dependency is an obvious generalization of the reflection condition that occurs for all waves when reflected in a simple stellar isothermal atmosphere @xcite .
the phase shift @xmath1 rather than being a constant value for all acoustic modes , such as @xmath4 where @xmath5 is the polytropic index of an isothermal atmosphere of a star like the sun for which @xmath6 , now becomes dependent on the frequency of the mode , such as @xmath7 where @xmath8 is the frequency of the mode of oscillation .
furthermore , it was shown that this phase shift can be partly inferred by means of the seismic observable @xmath9 , a proxy of the phase shift @xmath10 that is computed either from the phase shift scattering occurring in the stellar envelope , or from a table of frequencies of the modes that propagate through the same region .
several methods were developed to determine the scattering of acoustic waves in the outer layers of the sun and stars .
a detailed discussion can be found in the literature .
the fourth - order system of ordinary differential equations describing the linear adiabatic oscillations in the cowling approximation ( the gravitational effects are small in the outer layers ) can be reduced to a second - order schrdinger - type equation , considering that the trajectories of the acoustic waves in the surface layers are vertical .
it follows that the equation of acoustic oscillations reads @xmath11 where @xmath12 is the acoustic depth , @xmath13 the acoustic potential , @xmath14 the cyclic frequency and @xmath15 the wave function .
the acoustic depth is given by @xmath16 where @xmath17 is the adiabatic sound speed , and @xmath18 is the stellar radius .
the acoustic potential @xmath13 is defined by @xmath19 with @xmath20 and @xmath21 where @xmath22 is the radial gravity acceleration .
the method of the phase function @xcite was used to represent the eigenfunction @xmath15 @xmath23 where @xmath24 , with an additional condition on the amplitude function @xmath25 the outer phase shift of acoustic waves @xmath1 is computed from the stellar envelope , by numerically solving the following equation @xmath26 this equation is solved numerically with the proper initial boundary condition at @xmath27 .
the details of the numerical procedure can be found in @xcite .
@xmath9 is obtained from numerical differentiation of @xmath28 as @xmath29 the properties of the observable @xmath9 were largely discussed by @xcite , among others .
these techniques , based on an analysis of the phase shift of acoustic waves , led to the measurement of solar helium abundance by @xcite and @xcite and to a contribution of the same authors to the calibration of the equation of state @xcite .
the acoustic potential @xmath13 used in this calculation was determined based on a detailed stellar structure model of a sun - like star specifically computed for this analysis .
the stellar evolution code cesam was used to compute this structure model .
this code has an up - to - date and refined microscopic physics ( updated equation of state , opacities , nuclear reactions rates , and an accurate treatment of the microscopic diffusion of heavy elements ) , including the solar mixture of @xcite .
the basic input physics considered are identical to the solar standard model , as discussed in @xcite and @xcite .
figure 1 shows the value of @xmath30 computed using this method .
the solid line represents the model being studied in this work , a star with 1.1 @xmath31 . in this figure
we have also included the parameter @xmath30 calculated for a set of sun - like stars with masses from @xmath32 to @xmath33 .
[ table : fits ] with the location of an abrupt variation of the sound speed .
this location is expressed in terms of the acoustic depth .
the deeper the glitch , the higher the frequency of the oscillation induced in the phase @xmath0 by the glitch .
the influence of the second ionization zone of helium around the @xmath34 s is visible . in this region , we have a slight increase of the frequency which contradicts the general tendency .
vertical bars show the location of a pretended glitch to induce an oscillation with the same period that we obtained from the observational @xmath0 .
the dashed line corresponds to the modes with degree @xmath35 ( magenta line ) and the dot - dashed line corresponds to modes with @xmath36 ( green line ) .
, scaledwidth=50.0% ] the seismic parameter @xmath30 is the best proxy of the phase shift of acoustic waves that occurs as a consequence of the scattering of these waves in the upper layers of the star .
@xmath30 , as previously mentioned , can be computed from a table ( theoretical or observational ) of frequencies .
the algorithm reads @xmath37 where @xmath38 and @xmath39 and @xmath40 are the degree and radial order of the oscillation mode .
we developed an algorithm that allows the calculation of these derivatives using the maximum of points available .
this algorithm is based upon the formula proposed by @xcite , to study the acoustic phase shift in the sun .
the partial derivatives in @xmath41 should be treated carefully due to the small number of points available usually in observational frequency tables . in the specific case of this work we decided to use only the modes with degrees @xmath35 and @xmath36 .
this is due to the fact that the modes with degrees @xmath35 and @xmath36 are the modes with the biggest number of observational frequencies available , and thus , the modes that lead to a more reliable result . the richness and properties of the observable @xmath9
were largely discussed by lopes and collaborators .
@xmath41 computed as indicated in equation ( 10 ) allows a direct comparison between the star and its stellar model , specifically , highlighting the physical process differences in the upper layers of the model and the star .
we computed @xmath41 using a theoretical table for the same stellar structure model used in the computation of @xmath30 from the stellar envelope ( see the previous section ) .
the frequencies were obtained using a pulsation code that computes the adiabatic non - radial oscillations of the stellar structure model .
the theoretical table of frequencies obtained comprises around @xmath42 frequencies that correspond to acoustic modes with degree @xmath43 , and @xmath40 , taking values between @xmath44 and @xmath45 .
figure 1 ( b ) shows the computation of the @xmath0 in the case of a sun - like star with a mass of @xmath46 .
two methods of computing @xmath0 are represented ; the two theoretical descriptions of @xmath41 .
the computation of @xmath47 from the theoretical frequencies were performed over a range of frequencies between 1mhz and 5mhz .
the @xmath30 computed from the stellar envelope has a very good overall agreement with the @xmath47 calculated from the theoretical table of frequencies .
the general form of @xmath0 is very identical to the one obtained in the case of the sun @xcite .
however , @xmath30 , in the case of the sun spreads throughtout a range of frequencies between @xmath44 to @xmath48 mhz . in the case of this sun - like star
, the equivalent range is between @xmath44 to @xmath49 mhz .
the shape of @xmath30 is especially influenced by two factors : the superadiabatic region with a contribution of the hydrogen ionization zone , which defines the parabola form , and the zone of the partial ionization of helium which is responsible for the production of a sinusoidal component .
this helium signature is evidenced in figure 1 , the larger the mass of the star the stronger the sinusoidal component .
@xmath50 also has a weak dependence on the initial phase .
furthermore , in the case of low degree acoustic modes , it is known that there is a strong dependence of @xmath47 on the degree of the mode .
this effect occurs due to the contribution of eulerian perturbation of the gravitational field to define eigen - frequencies of oscillation .
this leads to @xmath51 presenting a dependence with the degree @xmath39 .
this effect is very visible in the case of the sun @xcite .
this sun - like star also shows the same effect . however , it is less visible ( see figure 1 ) .
this , in part , explains the difference between @xmath0 computed from the stellar envelope ( the solid line ) and @xmath0 computed from the theoretical table of frequencies for @xmath52 ( dotted lines ) . in the internal structure of stars ,
there are some regions where the sound speed changes rapidly due to changes in the structure .
these regions are usually called glitches and they introduce an oscillatory behavior in the frequencies proportional to @xmath53 where @xmath54 gives the location of the glitch @xcite . as an example , in the sun , the base of the convective zone and the helium ionization zone and their oscillatory signatures were studied in detail by @xcite , and @xcite .
application to sun - like stars was a natural step forward : methods to locate the base of the convective zone and the he ii ionization zone were studied and developed by , @xcite , @xcite , @xcite , @xcite , , and .
all these regions are in fact transition layers , where a sharp variation of the sound speed occurs .
acoustic sound waves are reflected and refracted in the surface of the star and by these layers .
the phase shift generated by these regions was estimated by @xcite .
considering that the acoustic potential has a variation of @xmath55 within an interval @xmath56 , there will be a variation of the eigenfuncion @xmath57 . from equation ( 1 )
we can reach the following wronskian @xmath58^{\tau_2}_{\tau_1 } = \int_{\tau_1}^{\tau_2 } \delta u^2 \psi_1\ , \psi_2\ , { d}\tau\ ] ] where @xmath59 the eigenfunction @xmath15 given by equation ( 6 ) allows us to rewrite the wronskian @xmath58^{\tau_2}_{\tau_1 } = \omega a_1(\tau_2 , \omega ) a_2(\tau_2 , \omega ) \sin(\pi \alpha_0).\ ] ] this is obtained considering the outer boundary condition at @xmath60 ( @xmath61 ) and @xmath62 . at a first approximation , and not taking into account the variation of the amplitude for @xmath56 , we have @xmath63 d \tau,\ ] ] where @xmath64 is the phase shift of the unperturbed eigenfunction .
this is a general expression which approximately estimates the phase shift @xmath65 induced by a variation of the acoustic potential within an interval in the acoustic depth for some value of the frequency @xmath66 .
as we can see , this induced phase shift will depend on the particular feature that is described by @xmath67 .
discontinuities in the first derivative of the sound speed produce glitches in the acoustic potential and can be represented by a dirac @xmath68 function .
we will see that this perturbation in the potential will produce an additional periodic component .
indeed , if we represent the acoustic potential as @xmath69 where @xmath70 is the unperturbed potential , @xmath71 is the contribution to the acoustic potential from the glitch ( i.e from the discontinuity in the first derivative of the sound speed ) and @xmath68 is the dirac @xmath68-function , we obtain directly from integral ( 14 ) @xmath72.\ ] ] this @xmath73 can be seen as the quasi - periodic contribution to the phase shift which originates from the discontinuity in the first derivative of the sound speed .
we numerically simulated a small variation of sound speed , typically smaller than @xmath74 , and realized that small variations of sound speed of such magnitudes can originate acoustical glitches in the potential . in this procedure , several different locations for @xmath54 were tested , namely between the acoustic depth of @xmath75s ( 0.99@xmath76 ) and @xmath77s ( 0.88@xmath76 ) .
figure 2 illustrates these tests made with the modified acoustic potential and the correspondent effects on the seismic parameter @xmath30 .
the period " of the observed quasi - periodic signal is determined by the location of the rv region and its amplitude is determined by the magnitude of the discontinuity in the acoustic potential @xcite .
table 1 shows the different locations tested and the corresponding periods of the oscillation generated by the rv region . in figure 3 , we fitted the values of table 1 with a piecewise cubic hermite interpolating polynomial .
we clearly see the decreasing tendency of the period of the oscillatory feature , except in the he ii ionization region . in particular ,
we want to focus now on the two cases of figure 4 .
this figure shows two of the previous scenarios : the a3 ( blue ) and the b3 ( red ) layers represent the positions where the glitch is located .
figure 4(b ) shows @xmath30 computed for these two new scenarios , as well as @xmath30 corresponding to the non - modified potential ( see figure 1 ) .
a detailed analysis of the observed @xmath0 suggests that the real @xmath1 centauri a could have an identical potential feature located in the external layers of the star .
however , a recent study by @xcite of the overshoot region at the bottom of sun s convective zone , suggests that the regions of transition in real stars might be smoother than generally predicted by theoretical models .
the stars of the @xmath1 centauri stellar binary system are among the most studied and observed sun - like stars .
being this binary system the closest to the earth , it makes it so its stars have a well known set of observational constraints .
this work was applied to the study of the upper layers of @xmath1 centauri a , the most massive star in this binary system .
this was done using the acoustic phase shift and some seismic properties of this star .
a large number of stellar evolution models have been developed for this star .
the first model was computed by @xcite , where they assumed a solar composition for the star @xmath1 cen a , but were not able to obtain the precise astrometric and photometric parameters . until the year 2000 , several models were proposed , based exclusively on non - asteroseismic constraints : , @xcite , , @xcite , , , and @xcite .
all these theoretical models were unable to give an unambiguous description of the stellar parameters . in 2002 , , obtained a definitive identification of _
p_-mode observational frequencies .
in the same year , computed , with improved precision , the masses of the @xmath1 centauri system .
their work was based on the accurate estimate of the parallax by .
one year later , using new interferometric measurements made for this star , provided accurate measurements of the angular diameters for the @xmath1 cen system , which allowed the precise determination of the radii .
these new constraints generated a new set of models for the stars @xmath1 centauri : , , , and .
these models were able to fit some of the asteroseismic constraints , but they left some questions unanswered , such as the nature of the core of the star .
our @xmath1 centari a model was calibrated to have the present stellar radius @xmath78 , luminosity @xmath79 , mass @xmath80 and age @xmath81 .
the model was also required to have a fixed value of the photospheric ratio @xmath82 , where _ x _ and _ z _ are , respectively , the mass fraction of hydrogen and the mass fraction of elements heavier than helium .
the value of @xmath82 we adopted was proposed by .
our model is identical to other models of @xmath1 centauri a found in the literature .
it is specifically similar to the model a3 with a radiative core from .
recently , published an observational table of frequencies that unified the data from coralie and from uves and ucles @xcite .
the goal was to compute a combined velocity time series in order to reduce the daily aliases in the power spectrum .
they were able to present a frequency table in which they identify 44 acoustic modes .
these 44 acoustic modes allow a reliable seismic study of the acoustic phase shift , especially for the modes with degrees @xmath52 . .
the continuous curves are the theoretical curves ( @xmath83 @xmath35 , o @xmath36 ) and the dashed curves represent the @xmath0 calculated from an observational frequency table for the modes with the degrees ( @xmath83 @xmath35 , o @xmath36 ) .
, scaledwidth=50.0% ] our aim now is to recover @xmath84 , using equation ( 10 ) from the observational table of frequencies of the star @xmath1 centauri a. we used the frequency table of which is so far the most updated table for this star .
the derivatives in equation ( 10 ) were calculated using up to five - point central differences formulas .
the problem with any numerical differentiation is directly connected with the structure of the experimental data : small perturbations on the function to be differentiated may cause large errors in the numerical derivatives . in our case , the frequencies used to evaluate the derivatives differ by a value of @xmath85 and the uncertainties in the frequencies are , in the worse case , given by @xmath86 .
this means that the error in calculating the derivatives will never exceed a few percents .
several tests were performed to verify the quality and stability of the algorithm to compute @xmath0 from this observational frequency table . in particular , we have repeatedly added a random white noise on the observed frequencies ( @xmath87 ) .
this white noise was added within the larger value of the frequency measurement error ( @xmath88 ) and the procedure was executed 400 times . at the same time , we used the propagation error theory to calculate the error bars associated with the exact value of the uncertainties of the observational frequencies .
the results are presented in figure 5 for @xmath52 over a range of frequencies from @xmath89 1.8mhz to @xmath89 3mhz and confirm that the algorithm is stable . in figure 6
we superimposed the theoretical signatures with the observational @xmath0 .
we can notice a dispersion in the observational modes that is not present in the theoretical modes .
this dispersion , as we will see , carries information about the oscillatory character present in each set of individual acoustic modes with the same degree .
the possibility of using this observable as a diagnostic tool and its limitations in the case of sun - like stars was discussed by @xcite .
a detailed analysis of the observed @xmath0 suggests that the real star could have an identical potential feature located in the external layers of the star . to illustrate the origin of the oscillatory character of the observational @xmath84
, we compare the @xmath9 of the ( b3 ) scenario ( red ) in figure 3 with @xmath84 computed from the observational frequency table , for which the @xmath84 of acoustic modes with the same degree are connected with a single line ( see figure 7 ) .
we also include in this comparison the ( a3 ) scenario ( blue ) and the @xmath30 calculated from the unmodified acoustic potential , the black line in figure 4 .
these tests with the modified acoustic potential are able to reproduce the oscillatory behavior of @xmath84 ( set of frequencies with the same @xmath39 are connected with a single dashed line ) better than the unmodified potential . for the `` period '' of the observable @xmath0 we estimated an error of @xmath90 using error propagation theory .
this means that for @xmath35 we will have an error of @xmath91 and for @xmath36 this same error is @xmath92 in the respective periods " .
we found that the location of the _ rv layer _ that best fits the observational @xmath84(with fix @xmath39 ) is located around @xmath93s ( 0.94 @xmath76 ) .
this feature is not present in the unmodified @xmath84 computed from the theoretical table and is not predicted by theoretical models . had noticed the presence of an oscillation corresponding to an acoustic depth of @xmath94 in this star , but considered it as an artifact . the range of acoustic depths available to perform a study with the characteristics above
is determined by the value of the large separation of the star
. we should be aware if the order of magnitude of the separation of the points , used to evaluate the observable @xmath0 is similar to the period we want to determine .
indeed , in this case , the ability to measure the period of the signature @xmath0 will be lost , since we have reached the limit of resolution of the method which can be expressed by the sampling theorem ( kotelnikov / nyquist - shannon ) .
the values of table 1 together with figures 2 , 4(b ) and 5 show that the deeper we move inside the star the closer we are to the resolution limit .
[ table : fits ] when an abrupt change occurs in the internal stratification of a star , the frequencies of oscillation will show a characteristic periodic signal .
these are regions of rv of the sound speed , and the observable seismic parameters are consequently affected by the oscillations in the frequencies .
the most studied example of such a region is the base of the convective zone in the sun .
the rv of the sound speed associated with the base of the convective zone in the sun was studied in detail by several authors in the past .
specifically , it was found that the location of the convective region can be extracted from the frequencies .
@xcite was the first to associate the period of such an oscillation with the acoustic depth of the glitch .
although using a different approach also showed that these oscillations can be found in the large separation and in the second difference .
higher order differences were discussed by @xcite and . in this work ,
we are mainly interested in the seismic parameter @xmath0 .
this phase - shift derivative was investigated in detail by @xcite .
the phase shift @xmath95 is known to vary quasi - periodically with the frequency @xmath8 with a period of @xmath96 where @xmath54 is the acoustic depth of the location of the rv region @xcite .
there is no reason to think that these techniques can not be applied to other sun - like stars .
based on the previously mentioned studies we made a test to the observational @xmath84 with the intent to isolate this new oscillation feature . in this approach , we start to compute the difference between the observational and the theoretical values of the seismic parameter @xmath0 , i.e. @xmath97 . the @xmath98 was obtained by making a spline fit to the theoretical values of @xmath84 ( see the black dots in figure 1 ) .
we think this is a more accurate representation of the theoretical @xmath99 of the star rather than using the @xmath99 computed from the stellar envelope .
this procedure allows us to eliminate any contribution from the superadiabatic region and the second helium ionization zone from the observations .
following such procedure , we performed a sinusoidal least - squares fit to each @xmath100 with fixed @xmath39 .
the results are shown in table 2 and the coefficients of the fits were determined with 95% confidence intervals . here again , if we use the _ period _ of each oscillation to estimate the location of the acoustic waves associated with _ rv layer _ , we find the following results : for @xmath100 computed for modes with degree @xmath35 and @xmath36 , the angular frequencies are @xmath101 and @xmath102 respectively , which implies the frequency will be @xmath103 .
next , we can calculate the period " @xmath104 and this period gives us the acoustic depth @xmath105 .
these results suggest the existence of the _ rv layer _ below the region of partial ionization of helium .
however , its exact location seems to be more difficult to determine .
it is well known that the upper turning points for low frequency modes are located deeper in the surface layers than the turning points for modes with higher frequencies .
considering the observational frequencies , a typical mode of 2.4 mhz , will have its upper turning point at @xmath106 whereas a mode with a lower frequency of 1.7 mhz has its upper turning point at @xmath107 .
we took into account this shift and found no significative differences for the location of the glitch in terms of normalized radius @xmath108 ( in the case of the mode with degree @xmath35 we found @xmath109 instead of @xmath110 listed in table 2 and no differences in the case of the mode with degree @xmath36 ) .
we think it is possible to say that the radial location of the glitch in table 2 is not affected by the vertical stratification in the surface layers and the upper reflection boundary is well approximated by the photospheric radius .
all the values are shown in table 2 with the respective uncertainty .
to have an idea of the expression of this uncertainty we represented all the range of this uncertainty with a grey double arrow in figure 4 .
some seismic indicators frequently used to identify sharp variations of the sound speed in a sun - like star , display an effect known as the aliasing problem .
this effect states that for modes of pulsation with the same value of the degree @xmath39 it is not possible to distinguish from discontinuities located at an acoustic depth of @xmath12 or @xmath111 . here
, @xmath112 is the total acoustic radius .
this effect was first noticed by and then by @xcite in the case of gravity modes .
we think that the method described in this paper , which uses the seismic observable @xmath0 , is safe from this aliasing effect . by definition , @xmath113 ( see equation ( 10 ) ) is computed considering the maximum number of modes available ( with @xmath114 ) .
nevertheless , this method should , whenever possible , be used in combination with other seismic methods to validate the results .
in this work , we explore the diagnostic capabilities of `` the acoustic phase shift method '' to probe the detailed structure of stellar envelopes . we specifically develop a technique to infer the impact that a small rv layer ( located in the upper layers of the star ) and a glitch of the sound speed have on the acoustic phase shift , or equivalently , on the frequencies of the acoustic modes .
in particular , we test our method in the case of the spectrum of acoustic oscillations of a @xmath115 sun - like star . the diagnostic of the acoustic phase shift is obtained by means of a functional constructed from the table of frequencies , usually known as the seismic parameter @xmath9 .
this observable @xmath0 , that is a derivative of the phase of the acoustic modes , is therefore used to deduce the properties of the physical processes occurring in the upper layers of a sun - like star . following
that , we decided to test our technique in the case of the star @xmath1 centauri a. we have computed @xmath30 for the theoretical and observational tables of frequencies .
we noticed that there is a disagreement between the theoretical @xmath0 ( obtained numerically by integrating the wave equation through the external layers and from a theoretical frequency table ) and the observational @xmath0 .
this disagreement is due to a sinusoidal - like feature found in the observational @xmath0 , but absent in the theoretical one . such surprising result obtained in this preliminary study of @xmath1 centauri a , motivated us to inquire further about this observed feature and further test the validity of our result by computing @xmath99 for several stellar envelopes and by comparing @xmath0 with other seismic diagnostics . using @xmath30 as a probe tool
, we have constructed several stellar envelope models with a local perturbed sound speed profile .
we found that only a small rv on the sound speed ( or equivalently in the acoustic potential ) , i.e. , a glitch , could explain the origin of this observed sinusoidal - like feature in @xmath0 .
we also found that the period and amplitude of this sinusoidal - like feature gives the location , thickness , and height of the glitch . in this specific case of @xmath1 centauri
a , the glitch is located around an acoustical depth of @xmath116 s ( @xmath89 0.940@xmath76 ) .
the existence of a helium ii ionization layer in the envelope of these types of stars , is among the physical processes that best explain such types of sinusoidal - like features in the stellar acoustic spectrum , or @xmath0 , including the star used in our case study - @xmath1 centauri a. indeed , the fact that a glitch located in the envelope of a sun - like star could produce a sinusoidal - like signature in the acoustic modes was first pointed out by @xcite .
we believe that even if this is not the only way to obtain this type of sinusoidal like feature , this physical process is among the best candidates to explain it . in particular , in the case of @xmath1 centauri a , this could explain the existence of the rv layer located 6% below the surface of the star . however , stellar evolution models of these types of sun - like stars , predict the location of the helium ii ionization around 2% below the surface , which is well above the value found from the observational data .
the exact physical mechanism responsible for the scattering of the acoustic waves in the sun - like stars is still unknown and needs further investigation .
nevertheless , there are a few indications from theoretical and observational studies that favor the existence of these rapid transition layers in the interior of sun - like stars . among others ,
we could mention the partial ionization of heavy elements ( for example carbon , nitrogen and oxygen ) , which in some stars such as sub dwarf b stars @xcite excited oscillations by the opacity mechanism , near surface rotational shear layer connected with meridional flows within the upper part of convection zone @xcite , and physical processes such as penetrative convection .
it is worth mentioning that the existence of convective cores in stars , especially , if their mass is above the critical mass value of @xmath117 @xcite , could increase the amplitude of this observed sinusoidal - like feature .
indeed , we made several tests and found that in the case of a sun - like star with a convective core , it is still possible to observe a sinusoidal - like feature in @xmath0 , identical to the one observed in figures 3 and 5 .
it is worth pointing out that the period associated with this feature rules out the possibility of it being created by the convective core .
however , we think that waves traveling from the core toward the surface , when perturbed by a rv layer of a glitch located in the stellar envelope , will experiment a wave interference process that could reinforce or even amplify the observed sinusoidal - like feature produced by the glitch . finally , it is important to highlight that this method , based on the scattering of the acoustic waves in the outer layers of the sun - like stars , can also be used to study others stars , and thus can contribute to and increase our knowledge of their interiors . the observable @xmath30 is a natural theoretical - observable of the acoustic phase shift which can become a powerful tool to extract information on the stellar structure from the acoustic global modes .
we want to thank to the referee for his comments that led to a deepening of the theoretical aspects of the relation between a region of rapid variation of the sound speed inside a star and its signature in the phase shift .
referee comments also resulted in a more accurate and robust manuscript . , m. , grevesse , n. , & sauval , a. j. 2005 , in astronomical society of the pacific conference series , vol .
336 , cosmic abundances as records of stellar evolution and nucleosynthesis , ed .
t. g. barnes , iii & f. n. bash , 25 , w. j. , kjeldsen , h. , christensen - dalsgaard , j. , basu , s. , miglio , a. , appourchaux , t. , bedding , t. r. , elsworth , y. , garca , r. a. , gilliland , r. l. , girardi , l. , houdek , g. , karoff , c. , kawaler , s. d. , metcalfe , t. s. , molenda - akowicz , j. , monteiro , m. j. p. f. g. , thompson , m. j. , verner , g. a. , ballot , j. , bonanno , a. , brando , i. m. , broomhall , a .- m . , bruntt , h. , campante , t. l. , corsaro , e. , creevey , o. l. , doan , g. , esch , l. , gai , n. , gaulme , p. , hale , s. j. , handberg , r. , hekker , s. , huber , d. , jimnez , a. , mathur , s. , mazumdar , a. , mosser , b. , new , r. , pinsonneault , m. h. , pricopi , d. , quirion , p .- o .
, rgulo , c. , salabert , d. , serenelli , a. m. , silva aguirre , v. , sousa , s. g. , stello , d. , stevens , i. r. , suran , m. d. , uytterhoeven , k. , white , t. r. , borucki , w. j. , brown , t. m. , jenkins , j. m. , kinemuchi , k. , van cleve , j. , & klaus , t. c. 2011 , science , 332 , 213 , r. l. , brown , t. m. , christensen - dalsgaard , j. , kjeldsen , h. , aerts , c. , appourchaux , t. , basu , s. , bedding , t. r. , chaplin , w. j. , cunha , m. s. , de cat , p. , de ridder , j. , guzik , j. a. , handler , g. , kawaler , s. , kiss , l. , kolenberg , k. , kurtz , d. w. , metcalfe , t. s. , monteiro , m. j. p. f. g. , szab , r. , arentoft , t. , balona , l. , debosscher , j. , elsworth , y. p. , quirion , p .- o . , stello , d. , surez , j. c. , borucki , w. j. , jenkins , j. m. , koch , d. , kondo , y. , latham , d. w. , rowe , j. f. , & steffen , j. h. 2010 , , 122 , 131 , e. , baglin , a. , auvergne , m. , catala , c. , samadi , r. , baudin , f. , appourchaux , t. , barban , c. , weiss , w. w. , berthomieu , g. , boumier , p. , dupret , m .- a . , garcia , r. a. , fridlund , m. , garrido , r. , goupil , m .- j . , kjeldsen , h. , lebreton , y. , mosser , b. , grotsch - noels , a. , janot - pacheco , e. , provost , j. , roxburgh , i. w. , thoul , a. , toutain , t. , tiphne , d. , turck - chieze , s. , vauclair , s. d. , vauclair , g. p. , aerts , c. , alecian , g. , ballot , j. , charpinet , s. , hubert , a .- m . ,
lignires , f. , mathias , p. , monteiro , m. j. p. f. g. , neiner , c. , poretti , e. , renan de medeiros , j. , ribas , i. , rieutord , m. l. , corts , t. r. , & zwintz , k. 2008 , science , 322 , 558 , d. , nidever , d. , mccarthy , c. , butler , r. p. , tinney , c. g. , marcy , g. w. , jones , h. r. a. , penny , a. j. , carter , b. d. , bouchy , f. , pepe , f. , hearnshaw , j. b. , skuljan , j. , ramm , d. , & kent , d. 2002 , , 386 , 280 , v. , basu , s. , brando , i. m. , christensen - dalsgaard , j. , deheuvels , s. , doan , g. , metcalfe , t. s. , serenelli , a. m. , ballot , j. , chaplin , w. j. , cunha , m. s. , weiss , a. , appourchaux , t. , casagrande , l. , cassisi , s. , creevey , o. l. , garca , r. a. , lebreton , y. , noels , a. , sousa , s. g. , stello , d. , white , t. r. , kawaler , s. d. , & kjeldsen , h. 2013 , , 769 , 141 | in the last decade the quality and the amount of observational asteroseismic data that has been made available by space based missions had a tremendous upgrowth .
the determination of asteroseismic parameters to estimate the fundamental physical processes occurring in stars interiors , can be done today in a way that has never been possible before . in this work
we propose to compute the seismic observable @xmath0 , which is a proxy of the phase shift of the acoustic modes propagating in the envelope of the sun - like stars .
this seismic parameter @xmath0 can be used to identify rapid variation regions usually known as glitches .
we show that a small variation in the structure , smaller than 1% in the sound speed , produces a glitch in the acoustic potential that could explain the oscillatory character of @xmath0 .
this method allows us to determine the location and the thickness of the glitch with precision .
we applied this idea to the sun - like star @xmath1 centauri a and found a glitch located at approximately @xmath2 ( 1400 s ) and with a thickness of 0.2% of the stars radius .
this is fully consistent with the data and also validates other seismic tests . |
the gaussian free field is widely considered to be a universal object describing the flucutations of heights of random surfaces .
previous work has rigorously shown this to be the case in specific models ( @xcite,@xcite ) . in this paper , we show that if an interlacing particle system in two dimensions can be described as a determinantal point process , and mild assumptions are made about the correlation kernel , then the covariances of the fluctuations of the height function are governed by a particular green s function .
a general formula for the green s function is given .
in particular , we will use this general theorem to show that in an interlacing particle system that arises from the representation theory of the orthogonal groups , the green s function is @xmath0 note that @xmath1 is the green s function for the laplace operator with dirichlet boundary conditions on the set @xmath2 .
it turns out that there is a map @xmath3 from the surface to @xmath4 .
we will show that the fluctuations of the height function converge to a gaussian process whose covariance is given by the pullback of @xmath1 under @xmath3 .
* particle system .
* now let us describe this particle system , which was the initial motivation for this paper .
introduce coordinates on the plane as shown in figure [ configur ] .
denote the horizontal coordinates of all particles with vertical coordinate @xmath5 by @xmath6 , where @xmath7 .
there is a wall on the left side , which forces @xmath8 for @xmath5 odd and @xmath9 for @xmath5 even .
the particles must also satisfy the interlacing conditions @xmath10 for all meaningful values of @xmath11 and @xmath5 .
[ configur ] by visually observing figure [ configur ] , one can see that the particle system can be interpreted as a stepped surface .
we thus define the height function at a point to be the number of particles to the right of that point .
define a continuous time markov chain as follows .
the initial condition is a single particle configuration where all the particles are as much to the left as possible , i.e. @xmath12 for all @xmath13 .
this is illustrated in the left - most iamge in figure [ jumps ] .
now let us describe the evolution .
we say that a particle @xmath14 is blocked on the right if @xmath15 , and it is blocked on the left if @xmath16 ( if the corresponding particle @xmath17 or @xmath18 does not exist , then @xmath14 is not blocked ) .
each particle has two exponential clocks of rate @xmath19 ; all clocks are independent .
one clock is responsible for the right jumps , while the other is responsible for the left jumps .
when the clock rings , the particle tries to jump by 1 in the corresponding direction . if the particle is blocked , then it stays still . if the particle is against the wall ( i.e. @xmath20}=0 $ ] ) and the left jump clock rings ,
the particle is reflected , and it tries to jump to the right instead . when @xmath14 tries to jump to the right ( and is not blocked on the right ) , we find the largest @xmath21 such that @xmath22 for @xmath23 , and the jump consists of all particles @xmath24 moving to the right by 1 .
similarly , when @xmath14 tries to jump to the left ( and is not blocked on the left ) , we find the largest @xmath25 such that @xmath26 for @xmath27 , and the jump consists of all particles @xmath28 moving to the left by 1 . in other words , the particles with smaller upper indices can be thought of as heavier than those with larger upper indices , and the heavier particles block and push the lighter ones so that the interlacing conditions are preserved .
[ jumps ] figure [ jumps ] depicts three possible first jumps : left clock of @xmath29 rings first ( it gets reflected by the wall ) , then right clock of @xmath30 rings , and then left clock of @xmath29 again . in terms of the underlying stepped surface
, the evolution can be described by saying that we add possible `` sticks '' with base @xmath31 and arbitrary length of a fixed orientation with rate 1/2 , remove possible `` sticks '' with base @xmath31 and a different orientation with rate 1/2 , and the rate of removing sticks that touch the left border is doubled . cube .
if one uses the dual convention that this is a cube - shaped hole then the orientations of the sticks to be added and removed have to be interchanged , and the tiling representations of the sticks change as well . ] a computer simulation of this dynamics can be found at + http://www.math.caltech.edu/papers/orth_planch.html[@xmath32 .
this particle system falls in the universality class of the anisotropic kardar - parisi - zhang ( akpz ) equation with a wall .
the kpz equation was first introduced in @xcite and is of interest to physicists , see @xcite .
similar markov chains have been previously studied in @xcite without the wall , and in @xcite with a different ( `` symplectic '' ) interaction with the wall .
* more general particle systems . * by following the proof , the author realized that a more general statement could be proved .
if a point process in a two - dimensional lattice is determinantal , and the correlation kernel can be expressed as a double integral with certain technical assumptions ( see definition [ normal ] below ) , then the moments of the fluctuations of the height function can be governed by a green s function .
the exact statement is theorem [ maintheorem ] .
we then can use this theorem to determine the green s function for the specific point process described above .
* motivations .
* there are three reasons for proving the results in the more general case .
the first reason is that the proofs are not much more difficult .
the second reason is that it is easier to check the conditions for the general case than it is to repeat the full calculations .
the third reason is that the general result tells us that the formula for the green s function only depends on @xmath33 , and therefore should only depend on the horizontal movement of the particles .
for instance , in the model of @xcite , the expression for @xmath1 was explicitly computed . in the model of @xcite ,
the particles had two different jump rates , and the expression for @xmath1 was the same .
our general results should imply , for example , that @xmath1 will be the same for any number of different jump rates . in order for a model to have a different expression for @xmath1
, there should be some change along the horizontal direction .
for example , in the model considered in this paper , a reflecting wall is added to the model of @xcite , and we see that @xmath1 is changed .
the correlation kernels in previous problems ( @xcite,@xcite ) should satisfy the conditions of the more general case , although the author has not rigorously checked the details . *
* there are several statements which should be true , but are not pursued in this paper .
one of them concerns the one - point fluctuations .
namely , after a scaling of @xmath34 , the fluctuations of the height function at a single point should converge in moments to a standard gaussian .
the proof would consist mainly in proving that the variance is proportional to @xmath35 , which by @xcite immediately implies the convergence to a gaussian .
the appropriate scaling was first demonstrated in @xcite using a renormalization group argument . using this convergence
, one can then show that the convergence to the gaussian free field occurs in distribution , not just in moments .
analogous ( rigorous ) statements are found as theorem 1.2 and proposition 5.6 in @xcite .
it should also be possible to prove similar results when @xmath36 , but this was also not pursued .
* organization of paper . * in section
[ statement of the main theorem ] , the theorem for more general particle systems ( theorem [ maintheorem ] ) is stated . in sections [
algebrageneral ] and [ analysisgeneral ] , the algebraic and analytic steps of the proof are given . in section [ specific results ] , it is shown that the particle system described above satisfies the conditions of theorem [ maintheorem ] .
section [ asymptotics ] deals with the necessary asymptotic analysis .
* acknowledgments . *
the author would like to thank alexei borodin and maurice duits for insightful comments .
the author was supported by the national science foundation s graduate research fellowship program .
suppose we have a family of point processes on @xmath37 which runs over time @xmath38 .
( note that these are different co - ordinates from the introduction ) .
in other words , at any time @xmath39 , the system selects a random subset @xmath40 .
if @xmath41 , then we say that there is a particle at @xmath42 .
for any @xmath43 and @xmath44 , let @xmath45 $ ] be defined by @xmath46 assume that there is a map @xmath47 on @xmath48 such that @xmath49_{1\leq i , j\leq k}.\ ] ] the maps @xmath50 and @xmath47 are called the _ kth correlation function _ and the _ correlation kernel _ , respectively . a function @xmath51 on @xmath52
is called a _ conjugating factor _ if there exists another function @xmath53 on @xmath54 such that @xmath55 note that if @xmath51 is a conjugating factor , then @xmath56_{1\leq i , j\leq k } = \det[c(x_i , n_i , x_j , n_j)k(x_i , n_i , x_j , n_j , t)]_{1\leq i , j\leq k}.\ ] ] two kernels @xmath47 and @xmath57 are called _ conjugate _ if @xmath58 for some conjugating factor @xmath51 . if a correlation kernel exists , the point process is called _ determinantal_. on a discrete space , a point process is determined uniquely by its correlation functions ( see e.g. @xcite ) . therefore ,
if we are given two determinantal point process on a discrete space with conjugate kernels , they must have the same law .
the set @xmath59 is called the _
nth level_. given a subset @xmath40 , let @xmath60 be the cardinality of the set @xmath61 .
assume that the numbers @xmath60 take constant finite values which are independent of the time parameter @xmath39 . in words
, this means that the number of particles on the @xmath62th level is always @xmath60 .
further assume that @xmath63 for all @xmath62 .
let @xmath64 denote the elements of @xmath61 .
a subset @xmath65 is called _ interlacing _ if @xmath66 @xmath67 assume that at any time @xmath39 , the system almost surely selects an interlacing subset .
let @xmath68 equal @xmath69 .
define the random _ height function _ by @xmath70 @xmath71 in words , @xmath72 counts the number of particles to the right of @xmath42 at time @xmath39 .
we wish to study the large - time asymptotics of this particle system .
let @xmath73 , n=[n\eta ] , t = n\tau$ ] , where @xmath74 is a large parameter .
define @xmath75 to be @xmath76 let @xmath77 be defined by @xmath78 @xmath79 in words , @xmath77 is the fluctuation of the height function around its expectation . before stating the theorem
, we need to state some more assumptions on the kernel .
suppose the kernel @xmath47 is conjugate to a kernel @xmath57 such that @xmath57 satisfies the following property : there is a number @xmath80 such that whenever @xmath81 , @xmath82 @xmath83 further suppose that for @xmath84 , @xmath85 suppose @xmath86 is a complex - valued function on @xmath87 . to save space , we will sometimes write @xmath88
. expressions such as @xmath89 will be shorthand for @xmath90 , @xmath91 and @xmath92 , respectively .
assume @xmath93.also suppose there exists a differentiable map @xmath94 from @xmath95 to the upper half - plane @xmath96 such that @xmath94 is a critical point of @xmath97 .
in other words , @xmath98 note that @xmath3 need not be onto . for any @xmath99 ,
if the set @xmath100 is nonempty , let @xmath101 denote its supremum . [ normal ] with the notation above , a determinantal point process on @xmath102 is _ normal _ if all of the following hold : * for all @xmath103 , the limit @xmath104 exists and is a positive real number .
* for all @xmath103 , as @xmath105 approaches @xmath101 from the left , @xmath106 . * @xmath47 is conjugate to some @xmath57 such that - hold for some integer @xmath80 . *
set @xmath107 , @xmath108 $ ] and @xmath109 $ ] for @xmath110 , where @xmath111 .
let @xmath112 denote @xmath113 and let @xmath114 denote @xmath115 . for finite integers @xmath116 and @xmath117 , as @xmath118 @xmath119 where @xmath120 and @xmath121 are steepest descent paths , @xmath122 , and @xmath123 are complex - valued meromorphic functions satisfying the identity @xmath124 here , we have written @xmath125 for @xmath126 . * for any @xmath127 , the following indefinite integral satisfies @xmath128 where the sum is taken over all @xmath129-cycles in @xmath130 and the indices are taken cyclically . * for any finite interval @xmath131 $ ] , @xmath132 $ ] and the lesbesgue measure of the set @xmath133 : i'(x)\in 2\pi{\mathbb{z}}+[-\delta,\delta]\}$ ] is @xmath134 for some positive @xmath135 .
the following remarks will help explain the definition .
[ normalexplanation ] ( 1 ) the assumption that @xmath136 occurs naturally .
one often finds that for @xmath137 , @xmath138+k_1,[n\eta],[n\nu]+k_2,[n\eta],n\tau)=\frac{1}{2\pi i}\int_{\bar{{\omega}}}^{{\omega } } \frac{dz}{z^{k_1-k_2 + 1}}=\frac{\im({\omega}^{k_2-k_1})}{\pi(k_2-k_1)},\ ] ] where the contour crosses the positive real line . by setting @xmath139
, we see that the right hand side reduces to the ubiquitous sine kernel . when @xmath140 , we see that @xmath141,[n\eta])=\frac{1}{2\pi}(\log{\omega}-\log\bar{{\omega}})=\frac{\arg{\omega}(\nu,\eta,\tau)}{\pi}.\ ] ] since the left hand side equals zero , we expect @xmath142 .
\(2 ) since @xmath93 , this means that @xmath97 has a critical point at both @xmath3 and @xmath143 . as @xmath3 approaches the real line ,
the two critical points coalesce into a triple zero , so @xmath144 converges to @xmath145 as @xmath39 approaches @xmath101 .
we need a control for how quickly this convergence to @xmath145 occurs , in order to order to control the behavior near the boundary of @xmath95 .
more specifically , it controls the bound in proposition [ edgeasymptotics ] .
there is a heuristic understanding for why ( 2 ) should hold .
the function @xmath97 has two critical points which coalesce into a triple zero .
the simplest example of such a function is @xmath146 as @xmath39 approaches @xmath145 . in this case
, the solution to @xmath147 is @xmath148
. then @xmath149 .
\(3 ) assumptions will be elucidated when we interpret the particles as lozenges . in particular , see remark [ lozengeremark ] .
\(4 ) it is common for the kernel to be expressed in this form ; previous examples are @xcite and @xcite .
if the kernel has a different expression with the same asymptotics as in propositions [ prop6.9 ] and [ edgeasymptotics ] , the results still hold .
\(5 ) in particular , holds if there always exist @xmath150-substitutions and an expression @xmath151 such that @xmath152 where @xmath153 and @xmath154 .
this is because of lemma 7.3 in @xcite , which refers back to @xcite , which says that @xmath155 \(6 ) this is a technical lemma which allows lemma [ zerolemma ] to be applied .
we can now state the main theorem .
[ maintheorem ] suppose we are given a normal determinantal point process .
for @xmath156 , let @xmath157 be distinct points in @xmath95 , and let @xmath158 .
define the function @xmath1 on the upper half - plane to be @xmath159 then @xmath160 where @xmath161 is the set of all involutions in @xmath162 without fixed points .
we note that these are the moments of a linear family of gaussian random variables : see appendix a. using the results of @xcite , it should be possible to show that @xmath163 converges to a gaussian , but this was not pursued . the most natural way to view
@xmath54 is as a square lattice .
however , it turns out that a hexagonal lattice is more useful . to obtain the hexagonal lattice ,
take the @xmath62th level and shift it to the right by @xmath164 . see figure [ particles ] .
figure [ particles ] also shows that the particle system can be interpreted as lozenges .
each lozenge is a pair of adjacent equilateral triangles .
see figure [ lozenges ] . by setting the location of each triangle to be the midpoint of its horizontal side
, each lozenge can be viewed as a pair @xmath165 , where the black triangle is located at @xmath42 and the white triangle is located at @xmath166 .
for example , in figure [ particles ] there are lozenges @xmath167 and @xmath168 .
the three types of lozenges can be described as follows . for lozenges of type i , @xmath169 . for lozenges of type ii , @xmath170 .
for lozenges of type iii , @xmath171 . note that a lozenge of type i is just a particle .
we say that @xmath172 is _ viable _ if @xmath173 , or @xmath174 .
a sequence @xmath175 of viable elements is _ non - overlapping _
if @xmath176 are all distinct from each other and @xmath177 are also all distinct from each other .
we do , however , allow the possibility of @xmath178 .
equal @xmath179 .
the black line on the left represents the points where @xmath180 .
examples of @xmath181 are @xmath182.,height=192 ] [ particles ] [ lozenges ] the statement and proof of the next proposition are similar to theorem 5.1 of .
[ lozengekernel ] suppose the kernel @xmath47 is conjugate to some @xmath57 such that - hold for some @xmath80 . if @xmath183 and @xmath184 @xmath185 is a sequence of non - overlapping viable elements of @xmath52 , then @xmath186_{1\leq i , j\leq k}.\end{gathered}\ ] ] [ lozengeremark ] the equations can now be intuitively understood .
equation says that each black triangle is located in exactly one of the three lozenges around it , and equation makes an identical statement for white triangles .
equations and say that lozenges of type ii almost surely do not occur far to the right of the particles , with controlling the off - diagonal entries in the determinant and controllling the diagonal entries . similarly , equations and says that lozenges of type iii almost surely do occur far to the right of the particles . this intuition will be exploited in the proof of thereom [ lozengekernel ] .
we proceed by induction on the number of lozenges that are not of type i. when this number is zero , the statement reduces to and . for any set @xmath187 of non - overlapping , viable elements ,
let @xmath188 and @xmath189 denote the left and right hand sides of , respectively .
first , as a preliminary statement , we prove that if @xmath190 for @xmath191 , then @xmath192 note that if @xmath193 is replaced by @xmath194 in , the statement is immediate , since the black triangle at @xmath195 must be contained in exactly one lozenge .
expand the determinants into a sum over permutations @xmath196 . now split the sum into two parts , one part where the @xmath197 fixes @xmath198 and one part where @xmath197 does not fix @xmath198 .
the first part equals @xmath199,\end{gathered}\ ] ] which , by , equals @xmath189 .
the second part equals @xmath200.\end{gathered}\ ] ] since @xmath201 by assumption , implies that this expression equals @xmath145 .
therefore holds . in a similar manner ,
if @xmath202 for @xmath191 , then implies that @xmath203 again , the statement holds if @xmath193 is replaced by @xmath194 .
now let us prove that @xmath193 and @xmath194 still agree if we add a lozenge of type ii to @xmath204 .
suppose that @xmath205 is viable and that @xmath206 is non - overlapping .
then equation is equivalent to @xmath207 and the same holds for @xmath194 instead of @xmath193 . by the induction hypothesis
, @xmath208 thus , implies @xmath209 assume for now that @xmath210 for @xmath211k .
then we cam apply equation , which implies that @xmath212 and the same statement holds for @xmath194 .
thus , @xmath213 if @xmath214 is non - overlappinng , then is again applicable .
we repeatedly apply and as often as possible .
first , suppose that this can be done indefinitely .
then @xmath215 since lozenges of type ii almost surely do not appear when we look far to the right of the particles , @xmath216 by expanding the determinant into a sum over @xmath217 , and imply that @xmath218 now suppose that and can only be applied finitely many times .
this means that @xmath219 equals either @xmath220 or @xmath221 in the first case , @xmath222 is non non - overlapping .
this implies @xmath223 ( because two of the rows are idential ) and @xmath224 ( because a triangle can not be in two different lozenges at the same time ) .
thus , @xmath193 and @xmath194 agree .
a similar argument holds in the second case .
thus , @xmath193 and @xmath194 agree whenever a lozenge of type ii is added to @xmath204 .
an identical holds for type iii lozenges , except that we use and instead of and .
we have been describing a lozenge as a pair @xmath165 .
it can also be described as @xmath225 , where @xmath166 is the location of the white triangle and @xmath226 is the type of the loznege .
thus the proposition can be restated as the following statement .
[ alternative ] for any non - overlappign @xmath227 , @xmath228_{1\leq i , j\leq k},\end{gathered}\ ] ] where @xmath229 this is a result of the correspondences @xmath230 there are two different formulas for the height function .
one formula is @xmath231 it is possible to only use to complete the proof .
however , when there are multiple points on one level , i.e. not all @xmath232 are distinct , the computation becomes much more complicated .
this is because lozenges of type i will appear in multiple sums of the form .
we can avoid this difficulty by introducing another formula for the height function : @xmath233 where , for @xmath234 , @xmath235 therefore , the expression @xmath236\right)\ ] ] can be expressed as a sum of terms of the form @xmath237\prod_{l = k'+1}^k [ h_{n_l , n_l'}(x_l)-\mathbb{e}(h_{n_l , n_l'}(x_l))]\right).\ ] ] assume that the following sets are disjoint : @xmath238 then @xmath239,\ ] ] where the matrix blocks are : @xmath240_{1\leq i , j\leq k ' } \\ a_{12 } & = [ \tilde{k}(s_i , n_i , x_j , p_j ,
t)]_{1\leq i\leq k',\ k'+1\leq j\leq k } \\
a_{21 } & = [ -\tilde{k}(x_i-\delta_{p_i-1},p_i-1,s_j , n_j , t)]_{k'+1\leq
i\leq k,\ 1\leq j\leq k'}\\ a_{22 } & = [ -(1-\delta_{ij})\tilde{k}(x_i-\delta_{p_i-1},p_i-1,x_j , p_j , t)]_{k'+1\leq i , j\leq k}\end{aligned}\ ] ] by applying corollary [ alternative ] to and , we see that @xmath241 equals the right hand side of with the @xmath242 terms removed .
it is well - known that subtracting the expectation corresponds to putting zeroes on the diagonal .
for example , this is noticed in the proof of theorem 7.2 of @xcite .
write the determinant in as a sum over permutations @xmath197 in @xmath162 .
if the cycle decomposition of @xmath197 contains the cycle @xmath243 of length @xmath244 and @xmath245 denotes the matrix in the right hand side of , then the contribution from @xmath197 is @xmath246 where @xmath247 correspond to other cycles of @xmath197 .
let @xmath248 denote @xmath249 if @xmath250 , and @xmath251 if @xmath252 . since the sum over @xmath248 only affects the matrix terms @xmath253 and @xmath254 , the contribution from @xmath197 is @xmath255 where @xmath256 denote other cycles .
in other words , the contribution from @xmath197 can be expressed as a product over the cycles in the cycle decomposition of @xmath197 . note that if @xmath197 fixes any points , then the correponding contribution is zero because all the diagonal entries are zero . in , set @xmath108 , n_l=[n\eta_l],$ ] and @xmath107 .
our goal is to find the limit of as @xmath257 .
expanding the determinant into a sum over @xmath258 , we just saw that the contribution from a fixed @xmath197 is of the form . first note that if any of the @xmath259 denotes @xmath260 , then @xmath261 this is because each @xmath262 is proportional to @xmath263 ( by proposition [ prop6.9 ] , so @xmath264 is proportional to @xmath265 , but the sum is only taken over @xmath266 terms .
therefore , can be expressed as a single term of the form in , and in this term @xmath267 .
now we will prove ( stated as theorem [ goalinanalysissection ] below ) that @xmath268 once this is proven , implies that the total contribution from @xmath269 equals zero .
when @xmath270 , then the right hand side is just @xmath271 , completing the proof of theorem [ maintheorem ] .
recall the definitions of @xmath97 and @xmath3 from section [ statement of the main theorem ]
. set @xmath272 to be @xmath273 [ productofkernels ] for @xmath274 , let @xmath275 , @xmath276 $ ] , @xmath277 $ ] and @xmath107 . for @xmath278 ,
let @xmath279 denote @xmath280 , let @xmath281 denote @xmath282 and let @xmath283 denote @xmath284 .
let @xmath285 and @xmath286 .
let @xmath287 .
then @xmath288 } k(x_1,n_1,y , n_2,t)k(y , n_2,x_3,n_3,t )
\\ = o\left(\frac{1}{n}\right)+\frac{e^{n\re((g_1({\omega}_1)-g_3({\omega}_3)))}}{2\pi n\sqrt{\vert g_1''({\omega}_1)\vert}\sqrt{\vert g_3''({\omega}_3)\vert}}\int_{\gamma_+\cup\gamma_-}\frac{dz}{2\pi g'_{\nu_1}(z)}\\ \times \big [ f({\omega}_1,z)f(z,{\omega}_3)\frac{e^{in\im(g_1({\omega}_1))-i{\theta}_1}}{e^{in\im(g_3({\omega}_3))+i{\theta}_3 } } + f({\bar{\omega}}_1,z)f(z,{\omega}_3)\frac{e^{-in\im(g_1({\omega}_1))+i{\theta}_1}}{e^{in\im(g_3({\omega}_3))+i{\theta}_3}}\\ + f({\omega}_1,z)f(z,{\bar{\omega}}_3)\frac{e^{in\im(g_1({\omega}_1))-i{\theta}_1}}{e^{-in\im(g_3({\omega}_3))-i{\theta}_3 } } + f({\bar{\omega}}_1,z)f(z,{\bar{\omega}}_3)\frac{e^{-in\im(g_1({\omega}_1))+i{\theta}_1}}{e^{-in\im(g_3({\omega}_3))-i{\theta}_3}}\big].\end{gathered}\ ] ] let @xmath289 denote @xmath290,\eta_2,\tau , z)$ ] , let @xmath291 denote @xmath292,\eta_2,\tau)$ ] and @xmath293 denote @xmath294 . fix some @xmath295 and split up the sum into two parts : the first part is from @xmath296 to @xmath297 , while the second sum is from @xmath297 to @xmath298 . since there are no particles to the right of @xmath299 in the limit @xmath257 , the sum from @xmath299 to @xmath300 can be ignored .
it is common to refer to the first sum as the bulk and the second sum as the edge .
first examine the bulk . by proposition [ prop6.9 ] , @xmath301\\ + \mathcal{o}(g_2''({\omega}_2)^{-4}n^{-3 } ) + \mathcal{o}(g_2''({\omega}_2)^{-7}n^{-4}),\end{gathered}\ ] ] where @xmath302 denotes the other fifteen terms that occur in the sum .
first let us examine the error term in the bulk .
by ( 2 ) of definition [ normal ] , each term in the error is bounded by @xmath303 and @xmath304 , respectively .
there are @xmath305 terms , and since @xmath306 , we must have @xmath307 and @xmath308 .
therefore the sum is @xmath309 .
now let us return to the main term in the bulk . for eight of the sixteen terms in @xmath302 ,
the expression @xmath310 cancels in the numerator and the denominator . by propositions
[ oneproposition ] , these eight terms are @xmath309 . by [ riemannapproximation ] , the other eight terms equal @xmath311 d\nu+o\left(\frac{1}{n}\right),\end{gathered}\ ] ] where @xmath312 represent the other seven terms . of the eight total terms , four have @xmath313 and four have @xmath314 . for the four terms with the expression @xmath293 , make the substitution @xmath315 .
the new integration path is @xmath316 . by taking the partial of with respect to @xmath105 and using the chain rule , @xmath317 which implies @xmath318 for the four terms with @xmath319 , make the substitution @xmath320 .
the path of integration is @xmath321 .
finally , the integral becomes @xmath322.\end{gathered}\ ] ] now we sum over the edge . by proposition [ edgeasymptotics ] and ( 2 ) of definition [ normal ] , the sum is bounded above by @xmath323 as long as @xmath324 , the sum over the edge is also @xmath309 .
[ goalinanalysissection ] for @xmath325 , let @xmath275 and set @xmath276,n_i = n\eta_i$ ] . for @xmath325 ,
let @xmath279 denote @xmath326 , let @xmath281 denote @xmath282 and let @xmath283 denote @xmath284 .
let @xmath327 and @xmath328 .
then @xmath329 } \cdots \displaystyle\sum_{y_l>[n\nu_l ] } \prod_{i=1}^l k(y_i , x_i , y_{i+1},x_{i+1},t ) \\ \rightarrow \left(\frac{1}{2\pi}\right)^l \int_{\gamma_1^{+}\cup\gamma_1^{- } } dz_1 \cdots \int_{\gamma^{+}_l\cup\gamma^{-}_l } dz_l
\frac{f(z_1,z_2)}{g'_{\nu}(z_1 ) } \ldots \frac{f(z_l , z_1)}{g'_{\nu}(z_l)}.\end{gathered}\ ] ] the indices are taken cyclically . by proposition [ prop6.9 ]
, the product has @xmath330 terms .
each application of proposition [ productofkernels ] decreases the number of terms by a factor of 4 , so repeated applications of proposition [ productofkernels ] yields the result .
we now return to the particle system with a reflecting wall described in the introduction . for notational reasons , it is more convenient to use different co - ordinates . instead of labeling the levels as @xmath331 , it is more convenient to label them as @xmath332 .
if the @xmath333 is at least as high as the @xmath334 level , then this will be denoted as @xmath335 .
this happens if and only if @xmath336 .
using the notation of section [ algebrageneral ] , @xmath337 and @xmath338 . along the horizontal direction , we will use a square lattice , so that the particles live on @xmath339 instead of @xmath340 or @xmath341 .
let @xmath342 be defined by @xmath343 let @xmath344 denote the ( normalized ) jacobi polynomial with parameters @xmath345 .
the normalization is set so that for any nonzero complex number @xmath346 , @xmath344 satisfies @xmath347 let @xmath348 be defined for nonnegative integers @xmath349 by @xmath350 note that for @xmath351 , @xmath352 by theorem 4.1 of @xcite , the correlation functions are determinantal with kernel @xmath353 @xmath354 where the @xmath346-contour is the unit circle and the @xmath355-contour is a circle centered at the origin with radius bigger than @xmath356 .
[ isnormal ] the determinantal point process is normal .
the green s function is given by @xmath0 once we prove the point process is normal , the expression for the green s function follows from theorem [ maintheorem ] with @xmath357 @xmath358 in section [ specificalgebra ] , we show that the third condition in definition [ normal ] is satisfied . in section [ specificanalysis ] , we show that the fourth and second conditions are satisfied .
since these are conditions are the hardest to prove , we will focus mainly on their proofs .
the fifth conditions follows from the substitution @xmath359 and ( 5 ) of remark [ normalexplanation ] .
let @xmath360 equal @xmath361 and @xmath362 .
then @xmath363 satisfies for @xmath364 . using and the orthogonality relation , it is straightforward to check that and hold .
what happens is that in the left hand side of or , one obtains six terms , three of which come from and three of which coe from .
the three terms from always sum to @xmath145 , while the three terms from sum to @xmath145 or @xmath356 .
now we will prove - when @xmath365 .
the term equals zero , so we only need to look at . explicitly , the expression is @xmath366 and we want the asymptotic result when @xmath367 in such a way that @xmath368 is @xmath145 or @xmath356 . expand the paranthetical expression @xmath369 to get two terms , each of which is a double integral .
since @xmath370 , the term with @xmath371 goes to zero . for the remaining term , expand @xmath372 to get two terms .
for the term with @xmath373 , make the substitution @xmath374 .
what remains is @xmath375 now deform the @xmath346-contour to the circle @xmath376 and the @xmath355-contour to the circle @xmath377 , where @xmath378 . with these deformations , @xmath379
, so the double integral goes to zero .
however , residues are picked up when the contours pass through each other .
these residues equal @xmath380 there is a residue at @xmath381 which equals @xmath382 , and a residue at @xmath383 which equals @xmath145 for @xmath384 and @xmath385 for @xmath386 . since @xmath387 , this proves and when @xmath365 .
the case when @xmath388 is similar .
it remains to show and .
when considering the product of two kernels , we obtain a quadruple integral .
after the substitutions @xmath389 and @xmath390 , the part of the integrand that depends on @xmath349 is just @xmath391 .
therefore , deforming contours so that @xmath392 gives and . for this section ,
we need a slightly different expression for the kernel . by ( 40)(42 ) of @xcite , the kernel equals @xmath393 @xmath394 @xmath395 where @xmath396 is any real number , and the arc from @xmath397 to @xmath398 is outside the unit circle and does not cross @xmath399 $ ] . set @xmath400 by proposition 5.1.1 of @xcite , we can take @xmath95 to be @xmath401 where @xmath402 [ inverseomega ] let @xmath403 denote @xmath404
. then @xmath405 in general , @xmath406 where @xmath407 and @xmath244 are @xmath408 let @xmath409 denote the polynomial @xmath410 corresponding to @xmath411 .
note that @xmath412 . by definition
, @xmath403 is the zero of @xmath413 that is in the upper half - plane .
therefore , @xmath414 .
now let us return to the proof of the fourth condition in definition [ normal ] . start by examining . expanding the parantheses , we obtain four terms corresponding to @xmath415 , and @xmath416 .
for the two terms with @xmath417 , make the substitution @xmath418 .
what remains are two terms , corresponding to @xmath416 and @xmath419 .
therefore , equals @xmath420 we now need to deform the contours in to steepest descent paths . in other words , we need @xmath421 for all @xmath346 on the @xmath346-contour and @xmath422 for all @xmath355 on the @xmath355-contour . by lemma [ inverseomega ] and the definition of @xmath97 , we see that @xmath423 .
if @xmath424 , then @xmath425 . since the steepest descent paths can go completely outside the unit circle ( see proposition 5.1.2 of @xcite ) , follows from .
, and on the right is @xmath426 .
white regions indicate @xmath427 and shaded regions indicate @xmath428 .
the double zero occurs at @xmath113 .
the arc @xmath355 goes from @xmath397 to @xmath398 .
the unit circle has been drawn on the right.,title="fig : " ] [ contours ] if we deform the contours to the steepest descent paths @xmath120 and @xmath121 in figure [ contours ] , we get that asymptotically becomes @xmath429 @xmath430 plus possibly the residues at @xmath431 .
since @xmath121 goes outside the unit circle and the critical point of @xmath432 lies inside the unit circle , the second double integral is negligible .
now we need to compute the possible residues at @xmath431 .
if the contours pass through each other , then the residues at @xmath431 equal @xmath433 @xmath434 where @xmath435 is any complex number satisfying and .
see figure [ residuecontours ] .
if the contours do not pass through each other , then there is no contribution from the residues . for notational convenience ,
set @xmath436 and @xmath355 contours from figure [ contours ] .
they intersect at @xmath435.,title="fig : " ] [ residuecontours ] it is important to note that @xmath437 is arbitrarily selected .
the only requirement on @xmath435 is that it satisfies the inequalities and , and the only requirement on @xmath398 is that @xmath438 .
so there exists @xmath378 such that if @xmath439 , then @xmath440 also satisfies those inequalities .
now we need to compute and . expanding the parantheses , we get four terms corresponding to @xmath441 . for the terms corresponding @xmath442 and @xmath443 , make the substitution @xmath418 .
therefore , the sum of , , , equals @xmath444 where the contour crosses @xmath445 if @xmath446 , and it crosses @xmath447 if @xmath448 . for each integral , deform the contour to a circular arc of constant radius .
then the absolute value of the integrand is maximized at the endpoints . to see this , note that @xmath449 is constant on the arc
, so it suffices to analyze @xmath450 now note that @xmath451 therefore , @xmath452 increases as we move counterclockwise along the arc in the upper half - plane , and @xmath452 decreases as we move counterclockwise along the arc in the lower half - plane . if @xmath446 , then @xmath453 is maximized when @xmath452 is maximized , which occurs at the endpoints since the contour crosses @xmath445 .
likewise , if @xmath448 , then @xmath452 is minimized at the endpoints .
thus , in both cases , @xmath453 is maximized at the endpoints . using a standard asymptotic analysis ( see e.g. chapter 3 of @xcite ) , we get that the asymptotic expansion of is @xmath454 for some constants @xmath455 . to complete the proof ,
notice that if @xmath456 for some selection of @xmath457 , then the asymptotic expansion of the kernel would depend on @xmath437 .
but @xmath437 was arbitrarily selected , so this is impossible . now that the fourth condition has been proved , it remains to show that the second condition in definition [ normal ] holds .
recall that @xmath458 is the root of @xmath459 that lies in the upper half - plane , where @xmath407 is the polynomial from lemma [ inverseomega ] .
we thus need to solve @xmath460 since @xmath461 is a double zero of @xmath462 , we thus have to solve @xmath463 which implies that @xmath464 . in other words ,
as @xmath105 approahces @xmath101 , @xmath465 . plugging this into the expression for @xmath466 gives the result .
[ zerolemma ] suppose that @xmath467 $ ] and @xmath468 $ ] .
suppose that as @xmath469 , the lesbesgue measure of the set @xmath133 : i'(x)\in 2\pi{\mathbb{z}}+[-\delta,\delta]\}$ ] is @xmath134 for some positive @xmath135 .
let @xmath470 $ ] depend on @xmath74 .
then @xmath471 let @xmath472 denote @xmath473 .
note that @xmath474 .
fix some @xmath475 and consider @xmath476 we bound this sum in two cases .
* case 1*. assume @xmath477 $ ] . for @xmath478 , taylor
s theorem says that @xmath479+[\frac{1}{2}i''(c_k)(t_k - t_s)^2]=:[i_1(t_k)]+[i_2(t_k)]\ ] ] for some @xmath480 between @xmath481 and @xmath472 . we will prove that @xmath482 using the inequality @xmath483 we have that @xmath484 furthermore , for @xmath485 , @xmath486 also , @xmath487 using , the definition of @xmath488 and @xmath489 , and respectively , @xmath490 * case 2*. assume that @xmath491 . in this case ,
only a simple estimate is needed : @xmath492 since the estimate in case 1 is much better than the estimate in case 2 , we need an upper bound on how frequently case 2 can occur . in other words , we need an upper bound on the measure of the set @xmath133 : i'(x)\in 2\pi\mathbb{z}+(\delta,\delta)\}$ ] .
we assumed that @xmath493 : i'(x)\in 2\pi\mathbb{z}+[\delta,\delta]\}\vert = { \mathcal{o}}(\delta^a).\ ] ] now we need to sum over all @xmath349 in the set @xmath494 .
there are @xmath495 terms for which case 2 applies
. therefore , @xmath496 and setting @xmath497 yields the result .
[ oneproposition ] suppose @xmath498 is a function such that for each @xmath499 , @xmath500 where @xmath501 and @xmath502 satisfy the same assumptions as in lemma [ zerolemma ] .
further suppose that the error term @xmath503 is uniform , i.e. @xmath504.\ ] ] then as @xmath257 , @xmath505 let @xmath378 . by assumption
, there exists @xmath506 such that for @xmath507 and @xmath508 $ ] @xmath509 by summing over @xmath510 , we see that @xmath511 the proposition now follows from lemma [ zerolemma ] . [ bvlemma ] suppose @xmath501 is a function on @xmath131 $ ] with bounded variation
. then as @xmath257 , @xmath512 let @xmath378 .
by the definition of a riemann sum , there exists @xmath245 such that @xmath513 furthermore , for @xmath514 , @xmath515 finally , for @xmath516 , @xmath517 [ riemannapproximation ] suppose @xmath518 is a function such that for each @xmath499 , @xmath519 where @xmath501 is a function on @xmath131 $ ] of bounded variation .
further suppose that the error term @xmath503 is uniform , i.e. @xmath520.\ ] ] then @xmath521 let @xmath378 . by assumption
, there exists @xmath506 such that for @xmath507 and @xmath508 $ ] @xmath522 we can see this by noting that when @xmath523 , @xmath524 .
furthermore , increasing @xmath510 by @xmath356 causes @xmath39 to increase by @xmath263 , which causes @xmath525 to increase by @xmath356 . by summing over @xmath510 , we see that @xmath526 the proposition now follows from lemma [ bvlemma ] .
[ prop6.9 ] for @xmath110 , let @xmath111 , @xmath112 denote @xmath113 , @xmath114 denote @xmath115 , and @xmath527 denote @xmath528 . with the assumptions from section [ statement of the main theorem ] , @xmath529.\end{gathered}\ ] ] first , we show that the main term is correct . by assumption
, we can deform @xmath120 and @xmath121 so that @xmath530 passes through @xmath531 for @xmath110 .
the contributions to the integral away from @xmath531 are exponentially small , so we can replace @xmath530 with @xmath532 , where @xmath533 and @xmath534 are steepest descent paths near @xmath112 and @xmath535 , respectively .
the integration over @xmath536 expands into four integrations corresponding to @xmath537 .
we explicitly do the calculation for @xmath538 .
the other three calculations are essentially identical .
make the substitutions @xmath539 and @xmath540 . in the neighborhood of @xmath541 and @xmath542
, we have @xmath543 which imply @xmath544 then we get @xmath545 , \end{aligned}\ ] ] where the last equality follows from @xmath546 .
the @xmath547 appears because the maps @xmath548 and @xmath549 are both two - to - one .
it still remains to show that the error term is correct .
the remainder of this section is devoted to proving this .
the idea is to reduce the double integral to progressively simpler forms .
first , by a reparametrization , the integral over two arcs in @xmath550 can be written as a integral in @xmath551 .
second , by using a taylor approximation , the integral in @xmath551 can be written as a product of two integrals in @xmath552 , each of which is of the form @xmath553 , where @xmath554 has a maximum @xmath555 in the interval of integration .
third , by using the implicit function theorem , this integral reduces to the form @xmath556 , where the interval of integration is a small neighbourhood @xmath555 .
fourth , this last integral is a slight generalization of @xmath557 , which is dealt with by the well - known watson s lemma ( lemma [ watsonslemma ] below ) .
since the first two steps have been done before ( see chapters 3 and 4 of @xcite ) , we will focus mostly on the third and fourth steps .
[ watsonslemma ] suppose @xmath558 and @xmath559 is a complex valued , absolutely integrable function on @xmath560 $ ] : @xmath561 also suppose that @xmath562 , where @xmath563 and @xmath501 is continuously differentiable in some neighbourhood of @xmath564
. then for any @xmath565 and @xmath566 $ ] , @xmath567 see proposition 2.1 of @xcite .
suppose @xmath558 and @xmath559 is a complex valued , absolutely integrable function on @xmath560 $ ] : @xmath561 also suppose that @xmath568 is twice continuously differentiable in some neighbourhood of @xmath564 . then for any @xmath565 and @xmath569 $ ] , @xmath570 where @xmath571 and @xmath572 see pages 5860 of @xcite .
[ 4.8 ] suppose that @xmath452 and @xmath568 are infinitely continuously differentiable in some neighbourhood of @xmath573 .
also suppose that @xmath555 is a local maximum of @xmath452 and @xmath574 .
then for any @xmath565 and @xmath569 $ ] , @xmath575 where @xmath576 @xmath577 where @xmath578 is an infinitely differentiable function solving @xmath579 before continuing , a few estimates on @xmath578 are needed .
[ estimateonv ] let @xmath578 be as in .
\(a ) @xmath580 @xmath581
\(b ) set @xmath582 .
then @xmath583 for @xmath584 in particular , @xmath585 and @xmath586 .
\(c ) let @xmath587 then @xmath588 @xmath589 \(d ) with the same bounds on @xmath590 , @xmath591 \(a ) the proof comes from page 69 of @xcite .
it follows immediately from using implicit differentiation of and setting @xmath592 .
\(b ) first notice that if @xmath593 with @xmath594 and @xmath595 are the solutions to @xmath596 , then @xmath597
. we will use @xmath598 therefore , we obtain bounds on @xmath578 by solving @xmath596 , which is equivalent to solving @xmath599 in other words @xmath600 .
we will use the intermediate value theorem to estimate roots of @xmath601 . for @xmath602 and @xmath603 where @xmath604 is any real number
, we have @xmath605 where the @xmath606 are polynomials which satisfy the following inequalities when @xmath607 @xmath608 now setting @xmath609 where @xmath610 , @xmath611 since @xmath612 this implies that @xmath613 for @xmath614 . by an identical argument , for @xmath615 and @xmath603 where @xmath607 , @xmath616 and setting @xmath609 where @xmath617 , @xmath618 for @xmath619 and @xmath620 , @xmath621 therefore @xmath622 for @xmath619 . for @xmath623 and @xmath624 , @xmath625 so @xmath626 .
thus the lower bound holds in both cases .
the last statement follows because @xmath627 \(c ) differentiating yields @xmath628 or equivalently @xmath629 to estimate @xmath630 , let us first estimate @xmath631 . by a taylor expansion , @xmath632 by the triangle inequality and part ( b ) , @xmath633 which , by and , implies @xmath634 to estimate the inverse of @xmath631 , use @xmath635 which , by setting @xmath636 and @xmath637 , implies that @xmath638 multiplying by @xmath639 finishes the proof of ( c ) .
\(d ) differentiating twice yields @xmath640 from part ( c ) and a taylor approximation for @xmath641 , @xmath642 since @xmath643 , @xmath644 by and the estimates on @xmath590 , @xmath645 combining , and , @xmath646 and using @xmath647 on the second term gives the result .
[ eh ] suppose that @xmath452 and @xmath568 are infinitely continuously differentiable in some neighbourhood of @xmath573 .
also suppose that @xmath555 is a local maximum of @xmath452 and @xmath574 .
let @xmath648 and @xmath649 be positive numbers such that @xmath650 and assume @xmath651 equals the right - hand side of .
let @xmath652 @xmath653 then for any @xmath565 , @xmath654 use lemma [ 4.8 ] . by part ( a ) of lemma [ estimateonv ] , @xmath655 by parts ( c ) and ( d ) of lemma [ estimateonv ] , @xmath656 when @xmath657 , @xmath658 @xmath659 @xmath660 @xmath661 when @xmath662 , @xmath663 @xmath664 @xmath665 @xmath666 thus @xmath667 , and @xmath668 we can finally wrap up the proof of proposition [ prop6.9 ] . since @xmath349 is not too small ( polynomial in @xmath669 ) , the exponential terms are small enough to be ignored .
therefore the error term is @xmath670 .
the main term is of order @xmath671 .
thus , when multiplying two integrals of the form in corollary [ eh ] , we get @xmath672 as needed . in proposition [ prop6.9 ] ,
the error term blows up at the edge .
therefore a better bound is needed . to get this bound
, we simply use the first term in watson s lemma , as opposed to using two terms .
since the method of the proof is identical as before and the details are simpler , the proof will be omitted .
the exact statement is the following .
[ edgeasymptotics ] for @xmath110 , let @xmath111 , @xmath112 denote @xmath113 , @xmath114 denote @xmath115 , and @xmath527 denote @xmath528 . with the assumptions in section [ statement of the main theorem ] , @xmath673 \end{gathered}\ ] ]
this section is adapted from @xcite .
given a domain @xmath674 , let @xmath675 be the space of smooth , real - valued functions on @xmath676 that are supported on a compact subset of @xmath193 .
let @xmath677 be the dirichlet inner product on @xmath675 .
let @xmath678 be the completion of @xmath675 under the dirichlet inner product .
note that @xmath678 is a hilbert space .
a * gaussian free field * is any gaussian hilbert space @xmath679 of random variables denoted by @xmath680 one random variable for each @xmath681 such that @xmath682=(a , b)_{\nabla}.\ ] ] in other words , the map from @xmath678 to @xmath679 defined by sending @xmath125 to the random variable @xmath680 preserves the inner product .
let @xmath685 be the eigenfunctions of the dirichlet laplacian on @xmath193 which form an orthonormal basis for @xmath686 , and let @xmath687 denote the corresponding eigenvalues . by the spectral theorem for compact self - adjoint operators ,
all the @xmath688 are negative .
suppose the ordering satisfies @xmath689 .
define the operator on @xmath686 : @xmath690 define @xmath691 to be the set of all @xmath692 such that @xmath693 .
we may view @xmath694 has a hilbert space with inner product @xmath695 .
note that when @xmath696 , the sum @xmath697 may not converge in @xmath686 , but it converges in the space of distributions .
note that by integration by parts , @xmath698 .
therefore , instead of considering @xmath678 as the hilbert space , we can also consider @xmath699 .
then @xmath700=(f , g)_{1/2}=((-\delta)^{-1/2 } f,(-\delta)^{-1/2 } g)=((-\delta)^{-1}f , g).\ ] ] if @xmath703 are mean zero random variables such that @xmath704 is gaussian for all @xmath705 , then @xmath706= \begin{cases } \displaystyle\sum_{\sigma}\prod_{j=1}^{k/2 } \mathbb{e}[x_{\sigma(j)}x_{\sigma(j+1 ) } ] , \ \
k\ \text{even}\\ 0 , \ \
k \ \text{odd } , \end{cases}\ ] ] where the sum is over all involutions on @xmath707 with no fixed point .
\99 ) c. boutillier , modles de dimres : comportements limites , ph.d . thesis , universit e de paris - sud , 2005 .
a. borodin and p. ferrari , _ anisotropic growth of random surfaces in 2 + 1 dimensions _
( 2008 ) , http://arxiv.org/abs/0804.3035[arxiv:0804.3035v1 ] a. borodin and j. kuan , _ random surface growth with a wall and plancherel measures for _ @xmath708 , comm .
math , volume 63 , issue 7 , pages 831 - 894 , july 2010 .
http://arxiv.org/abs/0904.2607[arxiv:0904.2607 ] a. borodin and j. kuan , _ asymptotics of plancherel measures for the infinite - dimensional unitary group _ , adv . in math ,
* 219 * ( 2008 ) , 894931 .
http://arxiv.org/abs/0712.1848[arxiv:0712.1848v1 ] a.l .
barabsi and h.e .
stanley , _ fractal concepts in surface growth _
, cambridge university press , cambridge , 1995 .
m. duits , _ the gaussian free field in an interlacing particle system with two jump rates _ , http://arxiv.org/abs/1105.4656[arxiv:1105.4656v1 ] .
r. kenyon , _ height fluctuations in the honeycomb dimer model _ , comm .
* 281 * ( 2008 ) , 675709 .
m. kardar , g. parisi and y. zhang , _ dynamic scaling of growing interfaces _ , phys . rev .
lett , volume 56 , issue 9 ( 1986 ) , 889892 .
a. lenard , _ correlation functions and the uniqueness of the state in classical statistical mechanics _ ,
, volume 30 , number 1 , 3544 .
miller , peter .
_ applied asymptotic analysis_. american mathematical society : providence , rhode island , 2006 .
s. sheffield , _ gaussian free field for mathematicians _ , probab . theory related fields 139 ( 2007 ) , no .
3 - 4 , 521541 .
http://arxiv.org/abs/0812.0022[arxiv:0812.0022v1 ] .
soshnikov , gaussian fluctuation for the number of particles in airy , bessel , sine and other determinantal random point fields , j. stat . phys .
* 100 * ( 2004 ) , 491522 .
http://arxiv.org/math-ph/9907012v2[arxiv:math-ph/9907012v2 ] j. warren and p. windridge , _ some examples of dynamics for gelfand tsetlin patterns _ , electronic j. of probability * 14 * ( 2009 ) , no .
59 , 17451769 .
http://arxiv.org/abs/0812.0022[arxiv:0812.0022v1 ] .
wolf , kinetic roughening of vicinal surfaces , phys .
* 67 * ( 1991 ) , 1783-1786 . | we show that if an interlacing particle system in a two - dimensional lattice is a determinantal point process , and the correlation kernel can be expressed as a double integral with certain technical assumptions , then the moments of the fluctuations of the height function converge to that of the gaussian free field . in particular , we examine a specific particle system that was previously studied in @xcite and compute the corresponding green s function . |
at optical and near infrared wavelengths , interstellar dust depresses the observed flux of galaxy disks by both scattering and absorption .
several authors have proposed that , even in the face on perspective , normal , non starburst disks are optically thick at optical wavelengths , while others have argued for substantial transparency ( see the volume edited by davies & burstein 1995 and the review by calzetti 2001 for details ) . using a sample of spiral galaxies within @xmath6 , we statistically derived photometric solutions for the degree of internal extinction at i band as a function of disk inclination ( giovanelli 1994 ) , finding a difference of more than a magnitude of flux between face on and edge on systems . at the same time , we found evidence for transparency of the outer parts of disks , at radii @xmath7 disk scale lengths from the center .
we later reported that the amount of internal extinction is luminosity dependent : more luminous disks being more opaque than less luminous ones ( giovanelli 1995 ) .
these results have been confirmed by more recent analyses ( e.g. tully 1998 ; wang & heckman 1996 ) .
purely photometric techniques are subject to a peculiar set of selection effects , that can severely affect quantitative conclusions on internal extinction , as witnessed by the liveliness of the debate over the last decade . in 1981 , goad & roberts noted a kinematic effect which can provide an independent test for disk extinction .
let @xmath8 be the rotation curve of a disk ( @xmath9 being the radial coordinate along the disk s projected major axis ) , assumed to have axial symmetry .
let @xmath10 be a set of cartesian coordinates _ in the plane of the disk_. if the disk is thin and it is observed at inclination @xmath11 ( @xmath12 for edge on ) , the component of velocity along the line of sight which intercepts the disk at @xmath10 is v_= v(r ) x i + v_turb , where @xmath13 accounts for turbulence and @xmath14 is oriented along the disk s apparent major axis . an observed rotation curve , as derived for example from a long slit spectrum positioned along @xmath14 , is smeared by seeing , instrumental resolution , averaging across the slit width , the finite thickness of the disk and extinction occurring within the disk itself . as realistically thick disks approach the edge on perspective , lines of sight along the major axis sample regions of increasingly broad range in @xmath15 , yielding a velocity distribution with a peak velocity contributed by parcels of gas at @xmath16 and a low velocity wing contributed by parcels at @xmath17 .
if extinction is important , only foreground parts of the disk contribute to the emission and the factor @xmath18 depresses the velocity distribution observed at @xmath19 . goad & roberts noted how this tapering effect may , in opaque edge on disks , produce observed rotation curves resembling solid body behavior , independently of the true shape of @xmath8 . bosma ( 1992 ) applied this technique to two edge on systems , ngc 100 and ngc 891 , by comparing hi synthesis and observations .
the evidence led them to conclude that the disk of ngc 100 is transparent , while in the case of ngc 891 they could not exclude the possibility of extinction in the inner parts of the disk .
prada ( 1994 ) compared long slit spectra of ngc 2146 in the optical and near ir and reported evidence for extinction in the inner parts of the galaxy . here , we apply the goad & roberts test in a statistically convincing manner to a sample of more than 2000 / [ ] rotation curves .
our results clearly indicate substantial opacity in the inner disks of spiral disks .
they also show the effect to be luminosity dependent , in a manner very much in agreement with our previous photometric determination . in section 2
we present our data sample and describe our rotation curve fits .
results on extinction are discussed in section 3 . in section 4 , we discuss an interesting serendipitous finding : the apparent constancy of the mean value of the kinematic scale length of / [ ] rotation curves , across populations of different luminosity class . throughout this paper
, distances and luminosities are obtained from redshifts in the cosmic microwave background reference frame and scaled according to a hubble parameter @xmath20 mpc@xmath21 .
as an effort to map the peculiar velocity field of the local universe , we have assembled a sample of several thousand observations of i band photometry and rotational width data of spiral galaxies .
our observations , including those listed in dale & giovanelli ( 2000 and refs . therein ) as well as several hundred obtained at the hale 5 m telescope after that report and to be presented elsewhere , have been complemented by the samples observed in the southern hemisphere by mathewson and co workers ( mathewson 1992 ; mathewson & ford 1996 ) .
the kinematic information for a substantial subset of these data is in the form of rotation curves in electronic form , derived from long slit / [ ] spectra .
rotation curves can be fitted by a variety of parametric models , some of which are motivated by the physical expectation of contributions by a baryonic component with a mass distribution mimicking that of the light plus a dark matter spheroidal halo .
other models rely purely on the versatility of a mathematical form in fitting effectively the observed rotation curves with a minimum of free parameters . here
we report on fits to 2246 rotation curves with a parametric model of the form v_pe(r ) = v_(1-e^-r / r_pe)(1+r / r_pe ) where @xmath22 regulates the overall amplitude of the rotation curve , @xmath23 yields a scale length for the inner steep rise and @xmath24 sets the slope of the slowly changing outer part .
this simple model , which we refer to as _
polyex _ , has been found to be very `` plastic '' .
it fits effectively both the steeply rising inner parts of rotation curves , as well as varying outer slopes , and it can be used to advantage in estimating velocity widths at specific radial distances from the galactic center as required by applications of the luminosity linewidth relation . in particular , it provides a very useful estimate of the inner slope of the rotation curve as given by @xmath23 . of the 2246 rotation curves fitted by the model shown in equation ( 2 )
, 425 have been rejected due to poor quality of the data or of the fit or because @xmath23 projects to less than 3 " and may thus be affected by poor angular resolution .
mean values of @xmath24 , the value of @xmath23 as seen in face on systems , @xmath25 , and of the ratio between @xmath23 and the scale length of the disk light for face on systems are given in table 1 , for different luminosity bins ; mean values of @xmath23 are shown in figure 1 and discussed in the next section , where they are used to probe disk extinction .
contents of columns 5 and 6 of table 1 are discussed in section 4 .
cccccc @xmath26 & @xmath27 & @xmath28 & @xmath29 & @xmath30 & @xmath31 @xmath32 & @xmath33 & @xmath34 & @xmath35 & @xmath36 & @xmath37 @xmath38 & @xmath39 & @xmath40 & @xmath41 & @xmath42 & @xmath31 @xmath43 & @xmath44 & @xmath45 & @xmath46 & @xmath47 & @xmath48 @xmath49 & @xmath50 & @xmath51 & @xmath52 & @xmath53 & @xmath54 @xmath55 & @xmath56 & @xmath57 & @xmath58 & @xmath59 & @xmath60
in figure 1 , the mean value of @xmath23 is shown separately for different luminosity classes , as a function of the disk inclination as expressed by @xmath61 , where @xmath62 is the apparent axial ratio .
each symbol is the average within a bin including between 20 and 30 galaxies .
the horizontal dotted line in each panel indicates an arbitrary scale of @xmath63 kpc . the total number of galaxies and the absolute magnitude range is indicated within each panel of the figure .
the parameter @xmath23 gives an indication of the radial distance at which the rotational velocity is @xmath64 of @xmath65 ( the asymptotic velocity for flat rotation curves , for which @xmath66 ) .
figure 1 shows that , as @xmath61 approaches 0.45 ( @xmath67 ) , the kinematic scale length @xmath23 of observed rotation curves starts increasing .
the effect is more marked for more luminous disks , for which edge on systems exhibit an average value of @xmath23 nearly three times as large as that of systems with @xmath68 .
the effect is not evident in the least luminous systems ( those with @xmath69 fainter than -19.5 ) .
this result is in agreement with the expectations based on the photometric determination of giovanelli ( 1995 ) .
no axial ratio dependence is observed for @xmath24 , the outer slope of the rotation curves .
we have carried out preliminary modelling of the observations shown in figure 1 by simulating the dust and distributions as exponential disks both in @xmath9 and @xmath70 . assuming the same distribution for dust and tracers , with a ratio between scale height and scale length of 0.06 ( as found by xilouris 1997 ; 1998 ) , we obtain approximate match between models and figure 1 for values of the the face on ( @xmath71 ) , central ( @xmath72 ) optical depth of the model at the wavelegth of ( @xmath73 ) , @xmath74 , of about 4 .
these modelling results are preliminary and depend very sensitively on the assumed thickness of disks and on the relative scales of dust and sources .
we will report more extensively on the results of this effort at a later stage .
figure 1 shows that , for low axial ratios ( where the effect of internal extinction is negligible ) , the average value of the kinematical scale length @xmath23 is approximately the same for disks of all luminosities .
is this result to be expected ?
first , we exclude that the angular resolution of spectroscopic observations has a significant impact on the reported findings . not using galaxies for which @xmath23 projects to an angular size @xmath75 rules out bias due to extreme cases of poor angular resolution .
since we use rotation curves from different sources , which have been produced with different angular sampling intervals , we have also verified that there is no systematic difference in the fitted values of @xmath76 among different data sources , taking advantage of significant sample overlaps .
we have done so for several dozen galaxies appearing in at least two samples among those : of our own , of mathewson et al .
, of courteau ( 1997 ) and of rubin et al .
( 1999 and refs . therein ) .
second , we explore the effect of sample selection criteria .
the values of @xmath77 displayed in figure 1 are affected by malmquist bias , which does not impact on the trends produced by extinction but alters the relative values of @xmath77 among different luminosity classes : since more luminous populations have larger average distance , for them the angular size limit of 3 `` produces an @xmath77 biased high . in order to remove this bias , we draw a subsample which is volume limited , by including in it only systems with @xmath78 , @xmath79 kpc and within the distance range @xmath80 to @xmath81 mpc . at @xmath81 mpc , an angular size limit of 3 '' translates into a linear size limit of about 1 kpc ; the reason for a lower limit in distance is to reduce the uncertainty introduced by peculiar velocities .
we further restrict the subsample to low inclination disks ( @xmath82 , @xmath83 , @xmath84 respectively for @xmath85 , @xmath86 and @xmath87 ) in order to get around the extinction effect discussed in the previous section and compute the mean values of @xmath23 and @xmath76 , now identified by a subscript `` 1 '' ; those values are listed in columns 5 and 6 of table 1 , for each luminosity class .
the exclusion of galaxies with @xmath88 kpc biases high the values of @xmath89 ; it does so however equally for all luminosity classes , through which @xmath89 remains approximately constant .
the mean values @xmath90 , which increase with decreasing luminosity , illustrate the fact that less luminous populations are on the average more nearby .
note that the values of @xmath90 are fairly large in comparison with the typical seeing , excluding the likelihood of an angular resolution bias .
next , we inquire on the implications of a constant @xmath23 on the total mass distribution
. assume the total mass of the galaxy @xmath91 is that comprised within a limiting radius @xmath92 , defined as that within which the halo mean density is 200 times the critical density of the universe .
then , m(r_pe)=m_200|(r_pe)|_200 ( r_per_200)^3 where @xmath93 is the mass within @xmath23 , and @xmath94 and @xmath95 are respectively the mean density within @xmath23 and within @xmath92 .
since @xmath96 , we can also express the ratio between the rotational velocity at @xmath23 and the asymptotic value @xmath97 as ^2 = |(r_pe)|_200 ( r_per_200)^2 for a flat rotation curve , the definition of @xmath23 yields @xmath98 ^ 2\simeq 0.4 $ ] ; then m(r_pe ) 0.4 m_200r_per_200 it can be shown that @xmath99 ; thus , for a constant value of @xmath23 : m(r_pe ) m_200 ^ 2/3 since @xmath65 does not generally coincide with @xmath97 , the effect of the luminosity dependence of @xmath24 needs to be taken into account : it can be parametrized by writing @xmath100 , where @xmath101 is small and positive . in general , equation ( 6
) can then be rewritten as m(r_pe ) m_200 ^ 2/3 + we now inquire on the dependence of the halo mass within @xmath23 on the total mass .
we assume a halo density form as proposed by navarro , frenk & white ( 1997 ) : _
h(r ) = _ crit_(r
/ r_s)(1+r / r_s)^2 where @xmath102 is the cosmological critical density , and the scaling parameters @xmath103 and @xmath104 can be expressed in terms of @xmath92 and a concentration index @xmath105 , with @xmath106^{-1}$ ] . for small values of @xmath9 ,
the halo mass within radius @xmath9 is m_h(r ) = 400c^2 _ crit r_200(1+c)-c/(1+c ) r^2 where @xmath107 . for a constant @xmath23 , m_h(r_pe )
m_200 ^ 1/3 numerical simulations indicate that @xmath108 is mildly dependent on halo mass , _ decreasing _ approximately as @xmath109 with @xmath110 ( navarro , frenk & white 1997 ) .
a comparison of equations ( 7 ) and ( 10 ) thus indicates that the mass in the halo can not account for the inner shape of the rotation curve , unless the concentration index _ increases _ with halo mass , opposite to what numerical simulations suggest .
an alternative interpretation is that the contribution of disk and bulge to @xmath93 increases with galaxy luminosity ( and mass ) , a well exercised idea in the current literature ( e.g. burstein & rubin 1985 ; broeils 1992 ; moriondo , giovanardi & hunt 1998 ; sellwood 1999 and refs .
therein ) .
if we assume that @xmath111 , @xmath112 and neglect the dependence of @xmath108 on @xmath91 , 1+f_d m_200 ^ 1/3 , suggesting that as the galaxy mass increases , the fraction of baryonic material in the inner few kpc grows rather quickly . for these systems , the baryonic mass fraction may saturate at its cosmological level ( i.e. these systems retain all their baryons ) and their inner rotation curves may be completely determined by it .
the spirals in our sample have small bulges ( they are mostly of type sbc and sc ) . within that context ,
note that in an exponential disk dominated rotation curve a rotational velocity @xmath64 of the maximum is obtained at a radial distance of about 0.6 disk scale lengths . for galaxies in our high luminosity bin , the mean ratio between @xmath23 and the scale length of the disk light is @xmath29 , as shown in table 1 , while it rises to twice that value for the less luminous galaxies in our sample .
we conclude that the simplest and most likely explanation of the near constancy of @xmath77 among luminosity classes is that the baryonic mass fraction within the inner few kpc of spirals increases with total luminosity , and that within that region the most luminous spirals are entirely dominated by baryonic matter .
we integrate photometric profiles to @xmath113 in order to obtain i band luminosities within that radius , assuming a solar absolute magnitude of + 3.94 ( livingston 2000 ) .
estimating mass from @xmath114 , for galaxies brighter than @xmath115 we obtain a mean baryonic mass to light ratio @xmath116 @xmath3 within @xmath23 .
individual colors for galaxies in our sample are not available ; however , if we assume that spiral galaxies radiate approximately like blackbodies with @xmath117 , then a mean bolometric @xmath118 @xmath119 can be estimated . the coarse analysis presented here
does not take into consideration the dynamical effect of a central bulge with a low spin parameter and we have not attempted to separate disk and bulge contributions to the light . it is interesting to compare our result with that of bottema ( 1999 ) for the sb galaxy ngc 7331 : from measurements of stellar velocity dispersions he obtains @xmath120 and @xmath121 .
we have shown that the / [ ] rotation curves of spiral galaxies exhibit a dependence of the inner slope of their rotation curves on the galaxy axial ratio .
we have interpreted this result as induced by extinction occurring within disks .
the effect is luminosity dependent , being stronger for the most luminous systems as previously indicated by giovanelli ( 1995 ) on photometric grounds .
serendipitously , it was also found that rotation curves of galaxies of different luminosity classes exhibit approximately the same average exponential scale length in the inner regions , of about @xmath122 kpc ( the result has potential in the determination of redshift independent distances : we will report elsewhere on this application ) .
this is in agreement with the idea that in the inner parts of spiral galaxies the baryonic component contributes a rapidly increasing fraction of the mass , with increasing total luminosity of systems .
in fact , the most luminous galaxies in our sample appear to be completely baryon dominated within the inner few kpc .
we use this inference and the combination of spectroscopic and photometric data to estimate a mean baryonic mass to light ratio for that region @xmath123 @xmath3 in the i band . | we use the tapering effect of / [ ] rotation curves of spiral galaxies first noted by goad & roberts ( 1981 ) to investigate the internal extinction in disks .
the scale length of exponential fits to the inner part of rotation curves depends strongly on the disk axial ratio .
preliminary modelling of the effect implies substantial opacity of the central parts of disks at a wavelength of 0.66 @xmath0 .
in addition , the average kinematic scale length of rotation curves , when corrected to face on perspective , has a nearly constant value of about @xmath1 kpc , for all luminosity classes .
the interpretation of that effect , as the result of the increasing dominance of the baryonic mass in the inner parts of galaxies , yields a mean baryonic mass to light ratio in the i band @xmath2 @xmath3 , within the inner @xmath1 kpc of disks .
83@xmath4 50w@xmath5 = cmr8 = cmbx8 = cmti8 = cmsl8 = cmmi8 + 584 _
use346 6.5 truein |
both photon beams and electron beams can carry orbital angular momentum . in the case of photon beams ,
the orbital angular momentum is carried by the spiral wavefront @xcite .
photon beams that carry orbital angular momentum have important applications @xcite in micro - machines , atom manipulation , and quantum information . in the case of electron beams ,
the orbital angular momentum was also shown @xcite to be carried by the spiral structure of the beam s wavefront when the spin polarization is uniformly distributed .
electron beams that carry orbital angular momentum were generated very recently in experiments @xcite by making use of techniques that are similar to those @xcite used for generating photon beams of orbital angular momentum .
apart from the orbital - angular - momentum photon beams which are approximately uniformly polarized @xcite , another kind of photon beams that are rotationally invariant in polarization , the so - called cylindrical - vector beams @xcite , were also predicted @xcite and experimentally generated @xcite .
the cylindrical - vector photon beams have drawn much attention in diverse areas of application @xcite .
now that the local state of quantum electron waves can be described in terms of the concept of spin polarization @xcite , an much more interesting question arises naturally as to whether there exist electron beams that are cylindrically symmetric in spin polarization . to the best of our knowledge , no theoretical investigations of such electron beams
have been reported .
an electron beam of cylindrically symmetric spin polarization means that it can not be an eigen state of the spin operator in a particular direction , @xmath1 , where @xmath2 is the vector of pauli matrix and @xmath3 is a fixed unit vector . upon noticing that the cylindrical - vector photon beams are eigen states of the total angular momentum in the propagation direction @xcite ,
denoted as the @xmath0 axis , our aim is therefore to look for free electron beams that are eigen states of total angular momentum in the @xmath0 direction , but not eigen states of @xmath4 . for this purpose , we will follow a method that is similar to the method @xcite for constructing the cylindrical - vector photon beams .
astonishingly , the results of our calculations reveal that so constructed free electron beams have spin skyrmion textures @xcite . for simplicity only monochromatic beams are considered . the wave function in the position representation , @xmath5 ,
is related to the wave function in the momentum representation , @xmath6 , via @xmath7 in circular cylindrical coordinates , where @xmath8 and @xmath9 . in non - relativistic quantum frame ,
the hamiltonian of free electrons is @xmath10 when the spin is omitted .
a complete set of mutually compatible observables in the monochromatic case can be chosen to be composed of @xmath11 and @xmath12 , the @xmath0 components of the linear and orbital angular momenta , respectively .
the corresponding complete orthonormal set of eigen functions in the momentum representation is given by @xmath13 where @xmath14 .
substituting it into eq .
( [ integral ] ) yields the orthonormal eigen functions in the position representation , @xmath15 where @xmath16 is the bessel function of the first kind and @xmath17 .
obviously , the eigen values of @xmath11 are @xmath18 ( we choose @xmath19 ) and the eigen values of @xmath12 are @xmath20 .
they are the scalar non - diffractive beams @xcite in the sense that the probability distribution does not change along with the propagation . equipped with the scalar basis ( [ basis without spin ] )
, we proceed to construct spinor states that are cylindrically symmetric in spin polarization . in analogy with the case of photon beams ,
there exist two distinct configurations .
the transverse component is either in radial direction or in azimuthal direction .
when the spin is taken into account , the free hamiltonian is @xmath21 .
we first introduce a unit - vector operator that is defined @xcite in terms of @xmath22 and the unit vector @xmath23 along the @xmath0 axis as @xmath24 . in the momentum representation
, it shows up as @xmath25
. then we pay attention to operator @xmath26 .
its normalized eigen spinors corresponding to eigen values @xmath27 have the forms of @xmath28 respectively .
it constitutes a complete set of mutually compatible observables together with @xmath11 and the total angular momentum @xmath29 in the @xmath0 direction . their common normalized eigen functions in the momentum representation
are proven to be @xmath30 the eigen value of @xmath31 is @xmath32 . substituting eq .
( [ basis v ] ) into eq .
( [ integral ] ) and replacing @xmath20 with @xmath33 , one obtains the corresponding normalized eigen functions in the position representation , @xmath34 let us prove that the spin polarization in these states is indeed cylindrically symmetric . according to definition @xcite @xmath35 , where @xmath36 is the probability density , the transverse and longitudinal components of the spin polarization are given by @xmath37 respectively .
they clearly show that ( a ) the transverse component is in radial direction and both the transverse and longitudinal components are rotationally invariant , indicating that the spin polarization is cylindrically symmetric ; ( b ) both the transverse and longitudinal components do not change along with the propagation , so that the beam described by eq . (
[ nd beam v ] ) is non - diffractive in spin - polarization distribution as well as in probability distribution .
( [ t - component v ] ) means that the transverse component of spin polarization at @xmath38 vanishes .
that is to say , the spin polarization at the center is always longitudinal
. based on the following asymptotic behavior of the bessel function , @xmath39 it is deduced from eq .
( [ l - component v ] ) that @xmath40 when @xmath41 and @xmath42 when @xmath43 .
in addition , when either equation @xmath44 or equation @xmath45 is satisfied , the spin polarization is also purely longitudinal . furthermore , when equation @xmath46 holds , the spin polarization is purely transverse . consequently , as @xmath47 increases continuously from the center , the spin polarization varies smoothly between transverse and longitudinal directions alternatively , starting from a purely longitudinal one . in fig .
[ nondiffraction ] are schematically shown the spin - polarization distributions of non - diffractive beams of @xmath48 , where the value of @xmath49 is taken from @xmath50 to @xmath51 , the first zero point of @xmath52 at which the spin polarization is opposite to that at the center .
, @xmath53 , ( b ) @xmath54 , @xmath55 , ( c ) @xmath56 , @xmath53 , and ( d ) @xmath56 , @xmath57.,width=453 ] now we are ready to construct finite beams that are cylindrically symmetric in spin polarization .
the eigen functions ( [ basis v ] ) or ( [ nd beam v ] ) constitute an orthonormal basis for monochromatic beams that have energy @xmath58 .
it is a direct calculation to show that the following wave functions in the momentum representation represent eigen states of @xmath31 , @xmath59 which is @xmath60 , where the scalar spectral function @xmath61 is square integrable and satisfies the normalization condition @xmath62 . as a superposition of eigen functions @xmath63 over the eigen value @xmath18 ,
the wave function @xmath64 is still an eigen state of operators @xmath65 and @xmath31 . substituting eq .
( [ angular spectrum ] ) into eq .
( [ integral ] ) and noticing the correspondence between eqs .
( [ basis v ] ) and ( [ nd beam v ] ) , one arrives at the corresponding wave functions in the position representation , @xmath66 where @xmath67 straightforward calculations give for the transverse and longitudinal components of spin polarization , [ sp ] @xmath68 respectively . since
function @xmath69 is axially symmetric , it is clearly seen from eqs .
( [ sp ] ) that the spin polarization is cylindrically symmetric . from eqs .
( [ fn ] ) and ( [ t - component ] ) one may deduce that the spin polarization at @xmath38 has no transverse component . the same as before ,
whether it is in positive or negative @xmath0 direction depends on the sign of eigen value @xmath70 .
furthermore , it can be seen from eq .
( [ fn ] ) that when the spectral function @xmath61 is a real - valued function , @xmath69 is also real - valued at plane @xmath71 , the plane of beam waist . in this case , the transverse component of spin polarization at plane @xmath71 is strictly in radial direction .
it is noted that the expectation of the spin in the state ( [ wave function ] ) vanishes , @xmath72 this result can be understood as follows .
unit - vector operator @xmath73 is an orbital vector operator in the sense @xcite that it satisfies commutation relations @xmath74
= \sum_{k } i \epsilon_{zjk } v_k$ ] , where @xmath75 is the levi - civit pseudo tensor .
so operator @xmath26 signifies a coupling between the spin and orbital angular momentum .
consequently , in its eigen state ( [ wave function ] ) there exists such a coupling which is introduced through the wave - vector dependence of the eigen spinors ( [ eigen spinor v ] ) . as a result of this specific spin - orbit coupling @xmath26
, we have the vanishing spin angular momentum . as an example
, we consider a paraxial beam which has a spectral function of gaussian type , @xmath76 , where @xmath77 describes the divergence angle satisfying @xmath78 . the integral formula @xmath79 makes us to cope with the cases of @xmath80 and @xmath81 separately , since it holds under the conditions @xcite of @xmath82 as well as @xmath83 .
due to the relation @xmath84 , we will consider only the case of @xmath80 in the next of this section .
substituting @xmath61 into eq .
( [ fn ] ) , making use of paraxial approximation @xmath85 in the factor @xmath86 , and extending the upper integral limit to infinity , one finds @xmath87 where @xmath88 , @xmath89 , and @xmath90 . since expression ( [ distribution ] ) contains factors of gaussian and modified bessel functions , the beams described by eqs .
( [ wave function ] ) and ( [ distribution ] ) are referred to as modified - bessel - gaussian beams . in fig .
[ skyrmion ] are schematically shown the spin - polarization distributions at the plane of beam waist for ( a ) @xmath54 , @xmath53 and ( b ) @xmath54 , @xmath55 , where @xmath91 is taken from the center to @xmath92 . ,
@xmath53 and ( b ) @xmath54 , @xmath57.,width=453 ] an interesting phenomenon that can be seen from fig .
[ skyrmion ] as well as eqs .
( [ sp ] ) is that the distribution of spin polarization found here shows up as a skyrmion texture @xcite .
the spin polarization at a cross section of the beam is rotationally invariant about the center and varies smoothly along with the increase of the radius .
therefore the topological charge of such a skyrmion texture @xcite , @xmath93 , is found from eqs .
( [ l - component ] ) and ( [ distribution ] ) to be @xmath94 which does not change along with the beam propagation .
it is dependent only on @xmath70 , the quantum number of @xmath31 .
when @xmath95 , we have @xmath96 as is clearly shown in fig .
[ skyrmion ] .
when @xmath97 , the topological charge is a fractional number and approaches @xmath98 in the limit @xmath99 .
apart from the above mentioned beam the transverse spin polarization of which is in radial direction at the plane of beam waist , there exists another kind of beams the transverse spin polarization of which is in azimuthal direction at the same plane . for the sake of completeness
, we briefly summarize the essential procedures of their construction and put forward the corresponding non - diffractive beams . from unit - vector operator @xmath73 ,
we define another momentum - dependent unit - vector operator @xmath100 .
the eigen spinors of operator @xmath101 are as follows , @xmath102 which correspond to eigen values @xmath103 , respectively , where @xmath104 .
one may check that @xmath105 is the common eigen function of operators @xmath101 , @xmath11 , and @xmath31 in the momentum representation . the eigen value of @xmath31 is @xmath106 . substituting eq .
( [ basis u ] ) into eq .
( [ integral ] ) and replacing @xmath20 with @xmath33 , one obtains the corresponding eigen functions in the position representation , @xmath107 for @xmath103 , respectively .
the transverse and longitudinal components of spin polarization in state @xmath108 are given by @xmath109 respectively , where @xmath110 . and
the transverse and longitudinal components of spin polarization in state @xmath111 are given by @xmath112 respectively .
obviously , the spin - polarization distributions in both situations are cylindrically symmetric .
but now the transverse components are all in azimuthal direction . in much the same way as used before , one is easy to construct finite beams the transverse spin polarization of which is in azimuthal direction at the plane of beam waist .
in a word , we have found solutions of quantum electron beams that bear close analogy to cylindrical - vector photon beams to show cylindrical symmetry in spin polarization .
these beams might be generated in experiments that are designed specially in much the similar way to those @xcite that are designed to generate cylindrical - vector photon beams .
for example , one might consider the transmission of uniformly polarized beams through a magnetic thin film that has a stable two - dimensional skyrmion texture in magnetization @xcite .
another possibility might be the electron emission from a source that has a stable two - dimensional skyrmion texture in magnetization .
cfl is indebted to ulrich k. rler for his helpful suggestions .
this work was supported in part by the national natural science foundation of china ( 60877055 and 60806041 ) and the shanghai leading academic discipline project ( s30105 ) .
l. allen , m. w. beijersbergen , r. j. c. spreeuw , and j. p. woerdman , phys .
a 45 , 8185 ( 1992 ) .
s. franke - arnold , l. allen , and m. padgett , laser photon .
rev . 2 , 299 ( 2008 ) .
k. y. bliokh , y. p. bliokh , s. savelev , and f. nori , phys .
99 , 190404 ( 2007 ) .
m. uchida and a. tonomura , nature 464 , 737 ( 2010 ) .
j. verbeeck , h. tian , and p. schattschneider , nature 467 , 301 ( 2010 ) .
beijersbergen , r.p.c .
coerwinkel , m. kristensen , and j.p .
woerdman , opt .
112 , 321 ( 1994 ) .
li , phys .
a 80 , 063814 ( 2009 ) .
k. youngworth and t. brown , opt .
express 7 , 77 ( 2000 ) .
r. h. jordan and d. g. hall , opt .
19 , 427 ( 1994 ) .
d. g. hall , opt .
21 , 9 ( 1996 ) .
t. v. higgins , laser focus world 28 , 18 ( 1992 ) .
m. stalder and m. schadt , opt .
lett . 21 , 1948 ( 1996 ) . y. kozawa and s. sato , opt . lett . 30 , 3063 ( 2005 ) .
q. zhan , adv .
photon . 1 , 1 ( 2009 ) .
e. merzbacher , quantum mechanics , 3rd ed .
( john wiley & sons , new york , 1998 ) . s. j. van enk and g. nienhuis , j. mod
41 , 963 ( 1994 ) .
sondhi , a. karlhede , s.a .
kivelson , and e.h .
rezayi , phys .
b 47 , 16419 ( 1993 ) .
a.n . bogdanov and d.a .
yablonskii , zh .
fiz . * 95 * , 178 ( 1989)[sov . phys .
jetp * 68 * , 101 ( 1989 ) ] ; a.n . bogdanov and a. hubert , j. magn .
magn . mater . *
138 * , 255 ( 1994 ) .
nature 465 , 901 ( 2010 ) .
durnin , j. opt .
a 4 , 651 ( 1987 ) .
i.s . gradshteyn and i.m .
ryzhik , table of integrals , series , and products , 7th ed .
( elsevier , amsterdam , 2007 ) .
k. moon et .
al . , phys . rev .
b 51 , 5138 ( 1995 ) .
rler , a.n .
bogdanov , and c. pfleiderer , nature * 442 * , 797 ( 2006 ) . x.z .
nature materials 10 , 106 ( 2011 ) . | cylindrically symmetric electron beams in spin polarization are reported for the first time .
they are shown to be the eigen states of total angular momentum in the @xmath0 direction .
but they are neither the eigen states of spin nor the eigen states of orbital angular momentum in that direction . |
x - ray studies of millisecond pulsars ( msps ) can constrain fundamental properties of their emission regions and , when combined with radio timing studies , their underlying neutron stars ( nss ) . in globular clusters
both msps and low mass x - ray binaries ( lmxbs ) , their likely progenitors , are significantly enhanced ( per unit mass ) over their values in the galactic disk by stellar and binary interactions .
the dense cluster ( core ) environment needed for their excess formation may also alter their evolution . thus cluster vs. field msps , as studied in x - rays and radio , can constrain intrinsic vs. extrinsic ( evolutionary ) properties of these oldest ns systems .
we have conducted a deep _ chandra _
survey for msps as well as quiescent lmxbs and cataclysmic variables ( cvs ) in the globular clusters 47 tuc ( grindlay et al . 2001a ; ghe01a ) and ngc 6397 ( grindlay et al . 2001b ; ghe01b ) .
the full details of the msp survey are given in grindlay et al .
( 2001c ; gch01 ) . here
we present the highlights of this study , focusing on just the x - ray properties of the 16 msps with radio timing positions in 47 tuc ( freire et al .
2001a , freire 2001 ) and the one in ngc 6397 ( damico et al .
2001 ; dpm ) as well as their comparison with the field msp population ( cf . becker & trumper 1997 , 1999 ; bt97 , bt99 ) .
we defer to the full paper the discussion of the total msp populations and spatial distributions , which probe cluster dynamics .
the 47 tuc msps were found initially ( ghe01a ) to be soft sources . in gch01
we give the detected counts in 3 bands : softcts ( 0.21kev ) , mediumcts ( 12kev ) and hardcts ( 28kev ) for each of the 14 resolved msps , with counts for 47 tuc - g and -i ( unresolved ) estimated . from these bands ,
we form the hardness ratios hr1 = mediumcts / softcts and hr2 = hardcts / mediumcts and plot the msps , with counting statistics errors , in the color - color diagram shown in figure 1 ( left ) .
the msp colors are clustered in a relatively narrow range of hr1 and hr2 with 47 tuc - j clearly harder , as was evident in the xcolor distributions in ghe01a . using the pimms tool ,
we construct values of hr1 and hr2 for 3 simple models : thermal bremsstrahlung ( tb ) , blackbody ( bb ) and power law ( pl ) , with index values ( kt or photon index ) given in the caption of figure 1 ( left ) .
the observed range of hr1-hr2 is roughly consistent with tb spectra with kt 1kev , bb spectra with kt 0.20.3kev ( except for 47 tuc - j ) or pl spectra with photon index 3 .
the weighted mean colors for all but 47 tuc - j are consistent with a bb spectrum with kt 0.22kev , giving x - ray luminosities ( 0.5 - 2.5kev ) 1 - 4 10@xmath0 erg s^-1erg s@xmath1 and thus mean bolometric l@xmath9 = 2.6 10@xmath0 erg s^-1erg s@xmath1 .
the x - ray colors rule out tb models ( surrounding column densities inconsistent with the msp dispersion measures ; dm ) and pl fits ( spectral indices implausible ) .
simple bb fits for l@xmath9 give emission radii of only 0.1 km whereas h ( or he)-atmosphere models ( rajagopal & romani 1996 ) typically give temperatures reduced ( from bb ) by a factor of 2 and thus radii increased to 0.4 km .
either case suggests soft x - ray emission from a region smaller than the entire polar cap , as predicted in recent models of harding & muslimov ( 2001 ) for polar cap heating .
although the 3.2s temporal resolution of _ chandra_-acis prevents a pulsation analysis , the small thermal emission area suggests the emission would be pulsed , with a sinusoidal pulse shape appropriate to the fractional visibility of the isotropically radiating thermal polar cap .
in contrast , the narrower pulse duty cycles of 10% for some field msps ( and one in the globular cluster m28 ; bt99 ) are probably due to non - thermal beamed emission .
a key question for this rich _ chandra _ dataset is the correlation of x - ray luminosity and pulsar spindown luminosity @xmath10 , which is found for field msps ( with much more uncertain distances ) to scale as ( 0.1 - 2.4kev ) 10@xmath7 ( bt97 ) and with a possibly steeper logarithmic slope ( 1.4 ) for in the 2 - 10kev band ( possenti et al .
2001 ; pcc ) .
we derive instrinsic period derivatives , @xmath11 , corrected for the cluster acceleration by estimating the 3d positions of each msp in the cluster from the observed dm value and the observed hot gas and thus electron density in the cluster ( freire et al .
2001b ) and then subtracting the cluster acceleration using a king model with cluster parameters derived by meylan & mayor ( 1986 ) . using a standard ns moment of inertia @xmath12g@xmath13 , we then derive = @xmath14 for each msp and plot them vs. ( 0.5 - 2.5kev ) in figure 1 ( right ) .
uncertainties in the values are typically 0.20.5 in the log but are not shown for clarity ; uncertainties in log(l@xmath15 ) are typically 0.2 , and extrapolating to the _ rosat _ band , 0.1 - 2.4kev , would increase log(l@xmath15 ) only by 0.1 . for comparison with 47 tuc
, we plot the msp in ngc 6397 ( ghe01b ) , for which the uncertainty is small , and updated values ( cf .
gch01 ) for the 10 field msps previously detected in x - rays as well as in the globular cluster m28 .
whereas the msps in the field and m28 show ( figure 1 , right ) a correlation log(0.1 - 2.4kev ) = ( 1.13@xmath16)log- 7.5@xmath17 , the msps in 47 tuc appear to have a weaker dependence : log(0.5 - 2.5kev ) = ( @xmath18)log+ @xmath19 for the nominal cluster model with central velocity dispersion = 11.6kms@xmath1 , where the errors ( @xmath201@xmath21 ) in both correlations are due to just the scatter in the points .
allowing for uncertainties in the cluster model and distance gives slope @xmath22 and intercept @xmath23 . including the errors for the values estimated for the 47 tuc msps , but with the approximation that unequal errors ( on @xmath10 ) are simply averaged ( which biases the slope to steeper values , since the unequal errors are much larger for smaller values of @xmath10 ) , increases the logarithmic slope to @xmath24 and offset to @xmath25 .
the best ( median ) estimate for the 47 tuc msps alone is thus @xmath4 @xmath26 .
apart from the uncertain detection of 47 tuc - c ( gch01 ) , the msps in 47 tuc have ( 0.52.5kev ) values within a factor of 4 despite a range of 25 in @xmath10 .
figure 1 ( right ) shows that 6397-a , the msp in ngc 6397 , is consistent with the - correlation shown by the 47 tuc msps .
including 6397-a in the fit yields @xmath27 ( scatter only ) or 0.55@xmath16 ( with averaged @xmath28 errors for the 47 tuc values ) .
the smaller @xmath4 values for 47 tuc and ngc 6397 may be due to the different formation histories and possibly different physical parameters of their msps vs. the field objects . in figure 2 ( left )
we plot vs. spindown ages , @xmath29 , for msps in the field ( and m28 ) vs. the 47 tuc and ngc 6397 sample .
error bars on the age parameter are not shown , for clarity , but are typically @xmath30 in the log and primarily due to the 47 tuc acceleration model . despite the uncertainties ,
the correlation is striking : the field msps show a declining with `` age , '' and the 47 tuc msps appear to fall on this trend but with a flatter vs. age slope .
however , spindown ages correspond only approximately to actual ages , and while these estimates are consistent with the formation of most 47 tuc msps early in the cluster history , this would not by itself provide an explanation for the different vs. correlations found for msps in 47 tuc and ngc 6397 vs. the field .
another possible physical difference between the 47 tuc and field msps might be magnetic field strength at the light cylinder ( at which the corotation speed equals @xmath31 ) , since this likely affects the relative importance of non - thermal ( magnetospheric ) emission . for an assumed dipole field , this is given by @xmath32 g , with @xmath33 in units of 10@xmath34 . in figure 2 ( right ) we plot vs. @xmath35 . again , considered as a homogeneous group , the msps in the field ( and m28 ) lie on a steeper slope than the 47 tuc msps . in support of the hypothesis that @xmath35 , rather than the field @xmath36 ( @xmath37)@xmath38 g at the ns surface , is more closely related to
is the fact that the correlation of with @xmath39 is less defined and with larger scatter , and with logarithmic slopes differing even more : 0.05@xmath40 for the 47 tuc msps vs. 2.80@xmath41 for the field msps .
we note that 3 out of the 4 msps with @xmath42 g display x - ray emission that is seemingly magnetospheric , with the nature of emission from the 4th ( the eclipsing msp b1957 + 20 ) indeterminate ( bt99 ; takahashi et al .
conversely , field pulsars with @xmath43 g have x - ray emission that is typically either thermal or of indeterminate character ( bt99 ) .
considering the small numbers of such pulsars studied , and that most of them have highly uncertain distances , it seems possible that field pulsars with @xmath44 g may show an l@xmath15@xmath35 trend that is fairly flat and roughly consistent with the better determined relation for the 47 tuc msps .
however for this interpretation to hold , a few field msps with relatively well determined distances ( e.g. 0437@xmath454715 and 1744@xmath451134 ; cf .
gch01 ) must be accounted for and the even larger deviations of 0751 + 1807 and 1024@xmath450719 from the 47 tuc correlation line would require factor 3 adjustments in these msp distances .
the msps in 47 tuc are primarily very soft x - ray sources , consistent with thermal emission from the pulsar polar caps .
the msps in both 47 tuc and ngc 6397 seem to have a less efficient conversion of rotational spindown energy ( @xmath10 ) into soft x - rays ( l@xmath15 ) than most field msps , even those with correspondingly low values of their magnetic field at the light cylinder . for @xmath2 @xmath3 ,
the 47 tuc - ngc 6397 samples are fit by @xmath46 whereas the updated ( gch01 ) field ( and m28 ) sample are consistent with the value @xmath47 found previously ( bt97 ; pcc and references therein ) .
the soft colors for the 47 tuc msps and the extended emission from 6397-a ( gch01 ) indicate a lack of beamed magnetospheric emission , which may suggest they have different @xmath35 values or configurations .
we speculate their b fields may have a multipole field geometry ( and thus lower dipole component ) and be less efficient particle accelerators . unlike msps in the field , those in dense cluster cores have a possibility of being driven back into contact and an accretion phase , as a re - cycled lmxb ( from a msp ) ! renewed accretion ( and 6397-a is presently close to filling its roche lobe ; cf .
dpm ) would likely continue the b - field burial process thought to be responsible ( e.g. romani 1990 ) for field decay from the @xmath48 g fields at ns birth to the @xmath49 g values typical of msps .
the 47 tuc msps have spent their @xmath50gyr lifetime in a dense ( @xmath51 ) cluster core , where they undergo scattering interactions with both single stars and other binaries .
such scattering causes the binding energy @xmath52 of a hard binary to secularly increase .
thus , while angular momentum transfer to the secondary acts to detach the binary , scattering events tend to drive it back toward contact .
a complication in this picture is that since msps are extremely hard binaries , the secular increase in @xmath52 is largely the result of infrequent strong scattering events , which are likely to eject the binary from the cluster core or even the cluster .
nevertheless , it appears plausible that a typical old ( @xmath50gyr ) msp in a _ dense _
cluster core might undergo one or more msp - lmxb transformation cycles .
thus a few percent of the msps at any one time may be in ( or near ) this recurrent lmxb phase .
this might explain the puzzling luminous qlmxbs x5 and x7 ( ghe01a and heinke et al .
2002 ) in 47 tuc : they might be recently ( @xmath53yr ) detached lmxbs , in which their underlying msp nature is hidden by their evaporating ( from msp - driven winds ? ) fossil disks , thus explaining their lack of any detectable low - level accretion signatures .
the younger system in ngc 6397 has the advantages of having been recently scattered out of a still higher density ( 10@xmath54 pc@xmath7 ) core collapsed cluster core and possibly having exchanged its companion ( ferraro et al .
2001 , ghe01b ) .
either or both would likely have restored an accretion phase .
thus 6397-a need not be just `` born , '' as suggested by ferraro et al .
( 2001 ) ; it may instead have just been reborn .
in contrast , the msp in m28 is both 10 younger , single and in a lower density core and so is unlikely to have gone through a renewed accretion phase . | we summarize the x - ray properties of the complete samples of millisecond pulsars ( msps ) detected in our _ chandra _ observations of the globular clusters 47 tuc and ngc 6397 .
the 47 tuc msps are predominantly soft sources suggestive of thermal emission from the neutron star polar cap and have x - ray luminosities in a surprisingly narrow range ( 14 10@xmath0 erg s^-1erg s@xmath1 ) .
the single msp in ngc 6397 is both hard and apparently extended , probably due to shocked hot gas evaporating from its main sequence companion .
in contrast to msps in the field and the cluster m28 , which show correlation between x - ray luminosity and spindown luminosity @xmath2 @xmath3 with @xmath4 1 - 1.4 , the 47 tuc ( and ngc 6397 ) sample display a relatively tight correlation with @xmath5 .
the correlations of vs. and light cylinder magnetic field values are also different .
it is possible the magnetic field configuration has been altered ( by episodic accretion ) for old msps in dense cluster cores . /cm2secerg
@xmath6 s@xmath1 2erg @xmath6 3pc@xmath7 -3cm@xmath7 /skm s@xmath1 a&a 1.25 in .125 in .25 in |
in recent few years new data from tjinaf on the @xmath6 elastic scattering were reported .
they include the electric structure function , @xmath0 , measured with high precision up to @xmath11 @xcite and measurements of tensor polarization observables , @xmath2 , @xmath3 and @xmath4 , up to @xmath12 @xcite .
this data , together with data on the magnetic structure function , @xmath1 @xcite , restrict the deuteron structure at scales where quark - gluon degrees of freedom are expected to become defrozen .
for example , according to optimistic estimations pqcd should start from @xmath13 of order of few @xmath14 @xcite .
it is nice that this prediction was confirmed by analysis of tjinaf data on @xmath0 at @xmath15 @xcite . for further conclusions one also should consider the spin structure of the deuteron from pqcd point of view .
however data on polarization observables , as well as on @xmath1 , correspond to @xmath16 , which is not enough for pqcd .
this is a typical intermediate region between nucleon - meson and quark - gluon pictures , where isobar configurations , meson exchange currents and constituent quark degrees of freedom are all important @xcite .
the purpose of this work is to investigate phenomenologically a smooth connection between nucleon - meson and pqcd regions and make predictions for @xmath1 and the polarization observables at higher @xmath13 , where pqcd should work . a parameterization which connects these two regions
was proposed earlier by one of the authors ( a.p .
k. ) and a.i .
syamtomov @xcite .
it assumes power fall off of helicity spin amplitudes at asymptotically high @xmath13 coming from quark counting rules . a new analysis of the parameterization @xcite which includes the recent tjinaf data was provided in @xcite .
now we study logarithmic corrections to the power behavior .
such corrections are shown to be important for the structure function @xmath0 at the highest region of tjinaf energy @xcite .
the paper is organized as follows . in sect . 2 we discuss the general structure of the helicity amplitudes for the elastic @xmath6 scattering in the light cone frame ( lcf ) and pqcd predictions for the helicity amplitudes at high @xmath13 .
parameterization of the helicity amplitudes which smoothly connects regions of low and high @xmath13 is given in sect .
3 . then , in sect . 4 ,
the data base and fitting procedure are summarized .
discussions and summary are given in sect .
the main object of our analysis is the helicity amplitudes of the @xmath17 transition @xmath18 where @xmath19 and @xmath20 are momenta and @xmath21 and @xmath22 are helicities of the deuteron in the initial and final states , respectively . due to gauge invariance , covariance and discrete symmetries only three of the 36 helicity amplitudes ( [ 1 ] ) are independent and one can choose different sets of independent helicity amplitudes .
direct calculations , however , demonstrate that it is not so in dynamics at lcf @xcite .
this phenomena was shown to come from the incompatibility of transformation properties of approximate current and a deuteron wave function used in practical calculations @xcite . as a result
a non - physical dependence on orientation of light - front plane appears .
thus the choice of the independent amplitudes becomes of great importance in pqcd calculations where lcf is often used .
first let us define lcf as a limiting reference system where the @xmath23-projection of the incoming and outgoing deuteron is close to infinity . in lcf the momenta of the incoming and outgoing deuteron
are given as follows @xmath24 with @xmath25 ( @xmath26 is the deuteron mass ) .
the momentum of the virtual photon is given by @xmath27 and the polarization vectors for the deuteron in the initial and final states , respectively , read @xmath28 here we put @xmath29 . using the standard expression for the e.m .
current matrix element @xmath30-\right.\nonumber\\ & -&\left.g_3(q^2)(p+p')^{\mu } \frac{\left(\varepsilon^{\ast}(\lambda',p')q\right ) \left(\varepsilon(\lambda , p)q\right)}{2m^2 } \right\ } \label{current_std}\end{aligned}\ ] ] one gets the following expressions for the current plus - component @xmath31 where @xmath32 .
it is easy to show that they satisfy the so - called angular condition @xmath33 and thus there are only three independent helicity amplitudes between the @xmath34 , @xmath35 , @xmath36 and @xmath37 @xcite .
alternatively the angular condition ( [ angular_c ] ) teaches us that even at pqcd extreme there appears ( through dimensionless ratio @xmath38 ) an additional scale parameter @xmath39 , apart from the pqcd parameter @xmath40 .
the charge , @xmath41 , magnetic , @xmath42 , and quadruple , @xmath43 , form factors are connected with the form factors @xmath44 , @xmath45 and @xmath46 as follows @xmath47 using ( [ 00])-([11 ] ) one expresses @xmath41 , @xmath42 and @xmath43 in terms of any three helicity amplitudes @xmath48 , for example @xmath49,\\ \nonumber g_m & = & \frac{1}{2{{\cal p}}(2\eta+1)}\left [ j^{+}_{00}+\frac{(2\eta-1)}{\sqrt{2\eta } } j^{+}_{10}-j^{+}_{1 - 1}\right],\\ \nonumber g_q & = & \frac{1}{2{{\cal p}}(2\eta+1)}\left [ -\frac12j^{+}_{00}+ \sqrt{\frac{1}{2\eta}}j^{+}_{10}- \frac{\eta+1}{2\eta}j^{+}_{1 - 1}\right ] .
\label{bh_set}\end{aligned}\ ] ] in turn , the @xmath0 and @xmath1 structure functions and the @xmath50 polarizations read @xmath51 \right\ } , \label{t20}\\ t_{21 } & = & \frac{2}{\sqrt{3}\,{\cal s}\cos\frac{\theta}{2 } } \eta\left ( \eta+\eta^2\sin^2\frac{\theta}{2 } \right)^{\frac12}g_m g_q , \label{t21 } \\ t_{22 } & = & -\frac{1}{2\sqrt{3}\,{\cal s}}g_m^2 , \label{t22}\end{aligned}\ ] ] where @xmath52 .
from pqcd arguments one gets very simple rules to determine the power behavior of the helicity amplitudes @xmath37 , @xmath36 and @xmath35 @xcite .
for example , it follows that the amplitude @xmath37 is a leading amplitude with an asymptotic fall off @xmath53 up to logarithmic corrections .
it was also argued that in lcf the helicity flip amplitudes @xmath36 and @xmath35 are suppressed as @xcite @xmath54 similar considerations give that @xmath55 which agrees with the angular condition ( [ angular_c ] ) , but only at extremely high @xmath13 , @xmath56 . in our analysis , following arguments of ref .
@xcite , we consider the set @xmath37 , @xmath36 and @xmath35 as main amplitudes , where the behavior ( [ 17]),([18 ] ) is regulated in the intermediate region only by the @xmath57 . in turn , the amplitude @xmath34 must be determined from the angular condition ( [ angular_c ] ) and thus it depends on the two scale parameters , @xmath58 and @xmath39 .
following the idea of reduced nuclear amplitudes in qcd @xcite , we define the reduced helicity transition amplitudes @xmath59 , @xmath60 and @xmath61 as follows : @xmath62 where @xmath63 is a three - quark - cluster ( nucleon ) form factor . for the @xmath63
we assume dipole behavior @xmath64^{-2}$ ] , but with the parameter @xmath65 is different from that for a free nucleon @xmath66 .
we consider separately two kinematical regions , large @xmath67 region , @xmath68 , and low @xmath67 region , @xmath69 .
the parameter @xmath70 is expected to be of order of few @xmath14 .
its exact value will be determined from a fit to experimental data .
pqcd predicts that at asymptoticaly high @xmath67 the reduced transition amplitudes behave as follows @xmath71 in ( [ 21 ] ) the factor @xmath72 takes into account logarithmic corrections @xmath73^{5}}{[\alpha_{s}(q^{2}/4)]^{4 } } \frac{(\lg q^{2}/\lambda_{\rm qcd}^{2})^{\gamma^{d } } } { [ \lg q^{2}/(4\lambda_{\rm qcd}^{2})]^{\gamma^{n } } } \label{22}\ ] ] and @xmath74 and @xmath75 are leadings anomalous dimensions for the deuteron and the nucleon , respectively , @xmath76 where @xmath77 , @xmath78 , @xmath79 is number of quark colors and @xmath80 is the number of flavors , @xmath81 is the running quark - gluon coupling @xcite . at @xmath82 the following parameterization for the reduced amplitudes
is assumed @xmath83 where @xmath84 ^ 2 $ ] .
one imposes the constrains ( see , e.q . ,
@xcite ) : @xmath85 on the coefficients @xmath86 , @xmath87 and @xmath88 to demand form factor normalization at @xmath89 . in ( [ 25 ] ) @xmath90 is the deuteron magnetic moment in `` deuteron magnetons '' and @xmath91 is the deuteron quadrupole momentum in @xmath92 .
the coefficients @xmath93 , @xmath94 and @xmath95 appearing in ( [ 21 ] ) can not be calculated from pqcd and we determine them from the smooth connection of the two parametrizations at the point @xmath96
in out fit we use the following data : for @xmath0 from @xcite , for @xmath1 from @xcite , and for @xmath2 from @xcite and for @xmath4 , @xmath98 from @xcite .
but it must be noted that due to large errors @xmath4 , @xmath3 data are practically not informative for the fit .
data for @xmath99 from @xcite are in sharp contradiction with all the wold data set and therefore were omitted from the fit .
we also do not include in the data base results from @xcite because they have large errors and practically do not change the obtained results .
data @xcite are somewhat lower that the world data , but due to large their uncertainties the fit is insensitive to this data .
data for the @xmath100 structure function were normalized , if necessary , to our convention of the magnetic form factor .
@xmath101 was considered as a parameter of the model , but the qcd cutoff parameter @xmath58 was fixed at 200 mev . in ( [ 25 ] ) we chose @xmath102 , so that the the model has 10 independent fit parameters .
the fit parameters summarized in table 1 were obtained with @xmath103 for 200 data points .
figs . 1 and 2 display comparison of the model with @xmath0 data . to show how far experimental data are from pqcd results in preasymptotic region we continue asymptotic behavior given by ( [ 21 ] ) to region lower than @xmath104 ( dotted line on fig . 2 and figs . 3 - 7 ) .
one concludes that for @xmath0 pqcd works from @xmath13-region between 1 and 2 @xmath105 .
comparison with data for @xmath1 and the polarization observables are given on figs . 3 5 .
one sees that for the magnetic form factor pqcd should start from @xmath13 between 2 and 3 @xmath106 , but for the polarization observable it should start somewhat earlier , near 2 @xmath105 .
7 shows results for for the charge and quadrupole form factors , @xmath107 and @xmath108 . in summary
, we give a parameterization of the deuteron form factors up to @xmath109 .
asymptotic behavior of the form factors is dictated by quark counting rules and pqcd helicity rules and therefore one may hope that this parameterization can be extrapolated for higher transferred momentum .
for example , the model predicts behavior of the magnetic structure function , @xmath1 , at @xmath110 which can be studied in future experiments .
we thank m. garon , g.g . petratos and e.a .
strokovsky for number of helpful discussions .
one of the authors ( a.k .
k. ) acknowledge the hospitality of rcnp , where this work was carried out with a center of excellence grant from the ministry of education , culture , sports , science and technology ( monbu - kagaku - sho ) , japan .
alexa et al . , phys . rev .
lett . * 82 * , 1374 ( 1999 ) .
d. abbott et al .
. lett . * 82 * , 1379 ( 1999 ) .
d. abbott et al .
lett . * 84 * , 5053 ( 2000 ) .
brodsky , c .- r .
ji and g.p .
lepage , phys .
lett . * 51 * , 83 ( 1983 ) .
bosted et al . , phys .
c * 42 * , 38 ( 1990 ) .
e. lomon , phys .
c * 64 * , 035204 ( 2001 ) .
a.p . kobushkin and a.i .
syamtomov , yad . fiz . * 58 * , 1565 ( 1995 ) [ phys .
. nucl . * 58 * , 1477 ( 1995 ) ] .
d. abbott et al .
j. a * 7 * , 421 ( 2000 ) .
i.l . grach and l.a .
kondratyuk , yad . fiz . * 39 * , 316 ( 1984 ) [ sov .
j. nucl .
phys . * 39 * , 198 ( 1984 ) ] .
karmanov and a.v .
smirnov , nucl.phys .
a * 546 * , 691 ( 1992 ) .
karmanov , nucl .
phys . a * 608 * , 316 ( 1996 ) .
chung , f. coester , b.d .
keister , and w.n .
polyzou , phys .
c * 37 * , 2000 ( 1988 ) .
a.i . vainshtein and v.i .
zakharov , phys .
b * 72 * 368 ( 1978 ) . c. carlson and f. gross , phys . rev .
lett . * 53 * 127 ( 1984 ) .
s.j . brodsky and j.r .
hiller , phys .
d * 46 * , 2141 ( 1992 ) .
a.p . kobushkin and a.i .
syamtomov , phys .
d * 49 * , 1637 ( 1994 ) .
s.j . brodsky and b.t .
chertok , phys .
lett . * 37 * , 269 ( 1976 ) ; phys .
d * 14 * , 3003 ( 1976 ) .
arnold et al . , phys .
. lett . * 35 * , 776 ( 1975 ) .
d. benaksas et al . , phys .
rev . * 148 * , 1327 ( 1966 ) .
c.d . buchanan and r. yearian ,
lett . * 15 * , 303 ( 1965 ) .
r. cramer et al . , z. phys .. c * 29 * , 513 ( 1985 ) . d.j .
drikley and l.n . hand .
* 9 * , 521 ( 1962 ) .
elias et al . , phys . rev . * 177 * , 2075 ( 1969 ) .
s. galster et al . , nucl .
b * 32 * , 221 ( 1971 ) . s. platchkov et al . ,
a * 510 * , 740 ( 1990 ) .
s. auffret et al . , phys . rev .
* 54 * , 649 ( 1985 ) .
simon et al . , nucl .
a * 364 * , 285 ( 1981 ) .
m. bouwhuis et al .
lett . * 82 * , 3755 ( 1999 ) .
dmitriev et al . , phys .
* 157b * , 143 ( 1985 ) .
m. ferro - luzzi et al .
lett . * 77 * , 2630 ( 1996 ) .
m. garon et al . , phys .
c * 49 * , 2516 ( 1994 ) .
r. gilman et al . , phys .
65 * , 1733 ( 1990 ) .
schulze et al . , phys . rev .
lett . * 52 * , 597 ( 1984 ) .
voitsekhovskii et al . , pisma v zh .
. fiz . * 43 * , 567 ( 1986 ) [ jetp lett . * 43 * , 733 ( 1986 ) ] .
akimov et al . , yad fiz . *
29 * , 649 ( 1979 ) [ sov . j. nucl . phys . * 29 * , 335 ( 1979 ) ] .
+ * figure 1 .
* comparison of the model fit with data for @xmath0 at @xmath111 . +
* figure 2 .
* comparison of the model fit ( solid line ) with data for @xmath0 at @xmath112 .
the dotted line is asymptotic behavior given by ( [ 21 ] ) extrapolated to lower transfer momentum . + * figure 3 .
* comparison of the model fit with data for @xmath1 .
the lines are the same as at fig . 2 . +
* figure 4 .
* comparison of the model fit with data for @xmath2 .
the lines are the same as at fig . 2 . +
* figure 5 .
* comparison of the model fit with data for @xmath3 .
the lines are the same as at fig . 2 . +
* figure 6 .
* comparison of the model fit with data for @xmath4 .
the lines are the same as at fig . 2 . +
* figure 7 .
* comparison of the model fit with data for @xmath113 ( upper figure ) and @xmath108 ( lower figure ) .
data are from @xcite .
for the last two points the two solutions ( filled and open points ) are shown , see @xcite .
the lines are the same as at fig . 2 . | electron - deuteron elastic scattering data ( @xmath0 and @xmath1 structure functions and polarization observables @xmath2 , @xmath3 and @xmath4 ) are fit with a model that respects asymptotic properties of pqcd at high momentum transfer .
the data analysis shows that pqcd starts from @xmath5 .
predictions for the magnetic structure function @xmath1 and the polarization observables at high momentum transfer are given .
+ * pacs * number(s ) : 13.40.fn , 12.38.bx , 25.30.bf * pqcd phenomenology of elastic @xmath6 scattering * a.p .
kobushkin @xmath7 and ya.d .
krivenko @xmath8 + @xmath9_n n bogolyubov institute for theoretical physics _ + _ metrologicheskaya str . 14b , kiev , 03143 , ukraine _ + @xmath10_research center for nuclear physics , osaka university ( suita campus ) _
+ _ 10 - 1 , mihogaoka ibaraki osaka 567 - 0047 , japan _ + @xmath8_institute for nuclear research _
+ _ prospekt nauki 47,kiev , 03028 , ukraine _ |
the cosmic microwave background radiation ( cmbr ) measurements play a crucial and irreplaceable role in establishing the favored cosmological model , that is , a flat cosmological constant - dominated , cold dark matter model ( @xmath2cdm ) , and constraining the cosmological parameters .
it is important , however , to bear in mind that cmbr observations predominantly probe the early universe at high redshift ( @xmath3 ) . as a result
, a projection within a given cosmological model is needed when we interpret these observations in terms of the standard cosmological parameters defined at @xmath4 , for instance , the hubble constant , @xmath5 , and the matter density parameter , @xmath0 , which provide basic information and are key parameters of the universe .
recently , one of the most exciting events is the release of scientific findings based on data from the first 15.5 months of _ planck _ operations @xcite . because of the high precision , the new _
data could constrain several cosmological parameters at few percent level @xcite . within the context of the spatially flat @xmath2cdm cosmology ,
a low value of the hubble constant , @xmath6 , and a high value of the matter density parameter , @xmath7 , are obtained .
these are seemingly in tension with the measurements of the magnitude - redshift relation for type ia supernova ( sne ia ) @xcite , but are entirely consistent with geometrical constraints from baryonic acoustic oscillation ( bao ) surveys @xcite .
this inconsistency between fundamental cosmological parameters constrained from the high redshift cmbr measurements and those from the observations at relatively low redshifts may indicate the existence of defects in the cosmological model where we project constraints on the standard cosmological parameters from these observations to @xmath4 , since projected parameters should presumably be the same from measurements at all @xmath8 in a given model .
thus , after _ planck _ , attempts have been made to resolve this tension @xcite .
for instance , the cosmic variance has been suggested to account for the discrepancy in @xmath5 @xcite and an extension of the friedmann - lem@xmath9itre - robertson - walker ( flrw ) metric to the reputed `` swiss - cheese '' model for the background has been proposed to alleviate the tension of @xmath0 @xcite . here ,
we take a different approach to the issue .
we show that if we first calibrate , with the help of the distance - duality relation , the light - curve fitting parameters in the distance estimation of the sne ia using the data on angular diameter distance of the galaxy clusters so as to eliminate the cosmological model - dependence that exists in the global fit to the hubble diagram where the light - curve fitting parameters are treated free on the same footing as cosmological parameters , then a higher value of the matter density parameter @xmath0 can be obtained from snls3 .
this is consistent with the _ planck _ at the 68.3% confidence , thereby alleviating the tension .
furthermore , with the light - curve fitting parameters and the absolute magnitude of a fiducial sne ia calibrated first , a low value of the hubble constant @xmath5 which is consistent with _ planck _ can also be obtained .
note that in parallel with cmbr measurements at high redshift , accurate distance estimation to celestial objects at relatively low redshift is another key tool in observational cosmology .
some fundamental changes in our understanding of the universe have resulted from such distance measurements .
for example , brahe s supernova and hubble s cepheids completely reconstructed our understanding of the cosmos @xcite .
almost five years after the sne ia were shown to be accurate standard candles , distance measurements for them have directly led to the discovery of the cosmic acceleration @xcite .
after several decades of intensive study , sne ia remain , at present , the most direct and mature portal to explore the essence of the accelerated expansion @xcite . in the past decade or so ,
several supernova data sets with hundreds of well - measured sne ia were released , such as essence " @xcite , constitution " @xcite , sdss - ii " @xcite , and union2.1 " @xcite . since the sne ia has been proposed as a distance indicator , various empirical approaches ( known as light - curve fitters ) to distance estimation , using light - curve shape parameters ( @xmath10 or a stretch factor ) @xcite or color information @xcite , or both @xcite ,
have been advanced .
currently , the distance of the sne ia is usually estimated by expressing it as an empirical function of the observable quantities because of the variability of the large spectra features .
taking the salt2 light - curve fitter @xcite as an example , the distance estimator ( distance modulus : @xmath11 + 25 $ ] ) of the sne ia is given by a linear combination of @xmath12 , @xmath13 , and @xmath14 : @xmath15 where @xmath13 is the stretch ( a measurement of the shape of the sne light curve ) and @xmath14 is the color measurement for the sne .
@xmath16 is the rest - frame peak magnitude of an sne . @xmath17 and @xmath18 are nuissance parameters which characterize the stretch - luminosity and color - luminosity relationships , reflecting the well - known broader - brighter and bluer - brighter relationships , respectively .
the value of @xmath1 is another nuissance parameter representing the absolute magnitude of a fiducial sne .
in general , in salt2 ( similar for sifto @xcite , or salt2/sifto combined @xcite ) , @xmath17 and @xmath18 are left as free parameters ( on the same weight as cosmological parameters ) that are determined in the global fit to the hubble diagram .
this treatment results in the dependence of distance estimation on cosmological model .
thus , cosmological implications derived from the distance estimation of the sne ia with the light - curve fitting parameters determined in the global fit to the hubble diagram are somewhat cosmological - model - dependent . on the other hand , besides the luminosity distance , @xmath19 , measurement for the standard candle such as sne ia , distance estimation for objects with known size ( that is , standard ruler ) named as angular diameter distance ( add ) , @xmath20 , is also often employed in astronomy .
recently , an add sample of 25 galaxy clusters ( @xmath21 ) has been obtained by combining the x - ray brightness and sunyaev - zeldovich temperature decrements ( sz effect @xcite ) observations @xcite .
in addition , the three - dimensional structure of galaxy cluster was also minutely studied in this work and it was found that the spherical hypothesis for geometry of cluster is generally rejected .
the luminosity distance , @xmath19 , and add , @xmath20 , may be measured independently by different astronomical observations from different celestial objects , but they relate to each other by means of the etherington s reciprocity relation @xcite : @xmath22 this relation , sometimes referred as the distance - duality ( dd ) relation , is completely general and valid for all cosmological models based on the riemannian geometry .
that is , the validity is dependent neither on the einstein field equation for gravity nor on the nature of the matter - energy content of the universe .
it only requires that the source and observer be connected by null geodesic in a riemannian spacetime and that the number of photons be conserved .
the fundamental dd relation has played an essential role in modern observational cosmology , for instance , gravitational - lensing studies @xcite , the plethora of cosmic consequences from primary and secondary temperature anisotropies of the cmbr observations @xcite and analysis from galaxy cluster observations @xcite .
thus , the dd relation , the validity of which is a seemingly reasonable assumption without new physics , along with the add data of galaxy clusters , provides us a natural possibility to calibrate the light - curve fitting parameters , @xmath17 and @xmath18 , for distance estimation in the sne ia observation in a cosmological - model - independent manner before being used to estimate the distances of the sne ia for cosmological analysis . in the following
, we will demonstrate that if we use @xmath17 and @xmath18 corrected this way to re - estimate the luminosity distances of the sne ia and explore cosmological implications in the framework of the spatially flat @xmath2cdm cosmology , we can obtain a higher value of matter density parameter , @xmath23 ( the original snls3 gives @xmath24 ) , which is in good agreement with that obtained from the _ planck _ observations . thus , tension regarding @xmath0 can be alleviated without invoking new physics or resorting to extensions of the standard cosmological model .
furthermore , a low value of hubble constant , @xmath25 , can also be obtained from snls3 , which is consistent with _ planck_.
in order to place cosmological - model - independent constraints on @xmath17 and @xmath18 with the aid of the reciprocity relation in eq .
( [ eq2 ] ) , the data pairs of observed @xmath19 and @xmath20 almost at the same redshift should be provided . for the observed @xmath19 ,
the snls3 sn ia sample compiled with salt2/sifto combined fitter @xcite is considered .
galaxy clusters sample , where an elliptical geometry is supposed for the morphology of clusters and the adds are obtained by combining the sze+x - ray brightness measurements @xcite , is responsible for providing the observed @xmath20 .
since the sample size of the sn ia is much larger than that of the galaxy clusters , we bin the observed @xmath19 from the data points of the snls3 , with the corresponding redshifts satisfying the selecting criteria , @xmath26 , only two sne ia , 04d1jd ( @xmath27 ) and 05d4cs ( @xmath28 ) , satisfy @xmath29 . for the sake of completeness of the galaxy clusters sample , these two sne ia are binned for matching the very galaxy cluster . ] , to match the observational data of the add sample , @xmath30 it should be noted that both binned distance modulus @xmath31 , and corresponding uncertainties @xmath32 are functions of @xmath17 and @xmath18 .
in addition , we have to express the observed distances in terms of the distance modulus , that is , @xmath33 for the sne ia observations and @xmath34 + 25 $ ] for the galaxy clusters sample , to marginalize the absolute magnitude of a fiducial sne ia , @xmath1 , when @xmath17 and @xmath18 are fitted using the standard minimum-@xmath35 route , @xmath36 where @xmath37 ^ 2}{\sigma_{\mathrm{tot}}^2(\alpha , \beta)}\;,\\ b(\alpha , \beta)&=&\sum_{i=1}^{25}\frac{[\mu^{\mathrm{sn}}(z_i ; \alpha , \beta , m=0)-\mu^{\mathrm{cluster}}(z_i)]}{\sigma_{\mathrm{tot}}^2(\alpha , \beta)}\;,\\ c(\alpha , \beta)&=&\sum_{i=1}^{25}\frac{1}{\sigma_{\mathrm{tot}}^2(\alpha , \beta)}\;.\end{aligned}\ ] ] here @xmath38 are propagated from both the statistical uncertainties in sne ia and that in galaxy clusters observations .
( [ chi22 ] ) has a minimum at @xmath39 , and it is @xmath40 different from the marginalization of a combination of the absolute magnitude of a fiducial sne ia and the hubble constant in the global fit to the hubble diagram , the analysis performed here can give an estimation for the absolute magnitude of a fiducial sne ia and thus break the degeneracy between them .
unfortunately , the systematic uncertainties of the sne ia ( in terms of the covariance matrix ) are difficult to be included when we bin the selected sne ia for the corresponding galaxy cluster to obtain our data pairs . however , the systematic errors are taken into consideration in our following cosmological implication analysis . the cosmological - model - independent constraint on @xmath17 and @xmath18 is shown in fig .
[ fig1 ] . compared to the light - curve fitting parameters determined from the global fit to the hubble diagram in the framework of
the constant @xmath41 dark energy model ( marked as the red star ) @xcite , the result derived from our cosmological - model - independent analysis ( indicated by the blue cross ) favors a larger @xmath17 and a smaller @xmath18 . along with these two light - curve fitting parameters , a model - independent estimation for the absolute magnitude of a fiducial sne ia @xmath42
is also achieved , which is in good agreement with what obtained from photometric measurements @xcite ( refer to review elsewhere @xcite ) .
this may be seen as an indication of reliability of our proposal to calibrate the light - curve fitting parameters in the distance estimation for sne ia using the add data .
it is worth noting that an estimation for @xmath1 can not be accomplished without any assumption prior for the hubble constant in previous global fit procedure . with this estimation for the absolute magnitude of a fiducial sne ia
, a constraint on hubble constant from snls3 alone can be obtained in our following analysis for cosmological implications .
now let us explore the cosmological implications of the corrected distance for the snls3 using the best fit values for the light - curve fitting parameters @xmath17 and @xmath18 constrained from our model - independent analysis .
following the minimum-@xmath35 route presented in appendix c of ref .
@xcite , we place constraints from the corrected snls3 sn ia on the spatially flat @xmath2cdm cosmology .
the results are shown in fig .
( [ fig2 ] , [ fig3 ] ) . from fig .
[ fig2 ] , we find that , compared to the original snls3 sn ia , the corrected one yields a higher value of the matter density parameter , @xmath23 .
this agrees with that obtained from _
observations very well .
moreover , with the previously determined @xmath42 , we can also derive a constraint on the hubble constant from the corrected snls3 .
as shown in fig .
[ fig3 ] , we obtain a low value of @xmath5 ( @xmath43 ) which is also in agreement with that from _ planck _ observations at 68.3% confidence . cosmological - model - independent constraint on light - curve fitting parameters , @xmath17 and @xmath18 , from snls3 sn ia analyzed with salt2/sifto combined fitter and the galaxy clusters sample.,scaledwidth=50.0%,scaledwidth=45.0% ] constraints on the matter density parameter , @xmath0 , in the context of spatially flat @xmath2cdm cosmology.,scaledwidth=75.0%,scaledwidth=45.0% ] constraints on the hubble constant , @xmath5 , at 68.3% confidence , in the context of the spatially flat @xmath2cdm cosmology .
, scaledwidth=75.0%,scaledwidth=45.0% ]
we have demonstrated that the tension between _ planck _ measurements and the observed magnitude - redshift relation for the sne ia may be alleviated if we first calibrate , with help of the distance - duality relation , the light - curve fitting parameters @xmath17 and @xmath18 in the distance estimation of the sne ia using the data on angular diameter distance of the galaxy clusters .
this eliminates the cosmological model - dependence that exists in the global fit to the hubble diagram where the parameters are treated free on the same footing as cosmological parameters .
we can use @xmath17 and @xmath18 corrected in this manner to re - estimate the luminosity distances of the snls3 sne ia and explore their cosmological implications in the framework of the spatially flat @xmath2cdm cosmology , and so a higher value of the matter density parameter , @xmath23 , can be obtained .
this alleviates the tension between _ planck _ and sne ia observations regarding @xmath0 significantly ( rendering them be consistent at 68.3% confidence ) .
another unusual feature in our approach is that the estimation for the absolute magnitude of a fiducial sne ia , @xmath42 , can simultaneously be obtained from the standard minimum-@xmath35 fitting route for @xmath17 and @xmath18 without any assumption for the hubble constant prior .
this makes constraining the hubble constant with snls3 alone possible and a low value of @xmath5 ( @xmath43 ) is obtained , which is also in good agreement with what obtained from _
planck _ observations .
however , the globally averaged hubble constant we obtained here from snls3 , although consistent with _ planck _ , is in tension with the locally measured expansion rate of the universe .
this might be a result of the cosmic variance @xcite , or even more speculatively , a dilute local environment @xcite .
finally , we must note that although the method for determining the light - curve fitting parameters proposed here can remove the cosmological model - dependence that is present in the global fit to the hubble diagram , yielding a reasonable absolute magnitude of a fiducial sne ia , and thereby reducing the tensions between _ planck _ measurements at high red - shift and the observed magnitude - redshift relation for the sne ia at relatively low red - shifts , the presence of systematic uncertainties in measurements using sze+x - ray surface brightness observations and the limited samples of the galaxy clusters may lead to biases of our result .
thus , add data of more samples of galaxy clusters with greater precision are needed to increase the statistical power of our result .
this being considered , the analysis herein presented suggests a possibility to reconcile the _ planck _ and sne ia observations without invoking new physics or resorting to extension of the standard cosmological model thus giving a direction for future observational endeavors .
we would like to thank a. conley and a. avgoustidis for helpful discussions .
hy thanks the kavli institute for theoretical physics china where part of this was done for hospitality and support .
this work was supported by the ministry of science and technology national basic science program ( project 973 ) under grant no.2012cb821804 , the national natural science foundation of china under grants nos .
10935013 , 11075083 , 11175093 , 11222545 and 11375092 , zhejiang provincial natural science foundation of china under grants nos .
z6100077 and r6110518 , the fanedd under grant no . 200922 , the hunan provincial natural science foundation of china under grant no .
11jj7001 , and the srfdp under grant no.20124306110001 .
zl was partially supported by china postdoc grant no.2013m530541 .
99 planck collaboration i , p. a. r. ade , _ et al .
_ , arxiv:1303.5062 , ( 2013 ) .
planck collaboration xvi , p. bielewicz , _ et al .
_ , arxiv:1303.5076 , ( 2013 ) .
a. g. riess , l. macri , s. casertano , _ et al .
_ , apj , 730 , 119 ( 2011 ) . w. l. freedman , b. f. madore , v. scowcroft , _ et al .
_ , apj , 758 , 24 ( 2012 ) .
a. conley , _ et al .
_ , apjs , 192 , 1 ( 2011 ) . w. j. percival , _ et al .
_ , mnras , 401 , 2148 ( 2010 ) .
c. blake , _ et al .
_ , mnras , 418 , 1707 ( 2011 ) .
v. marra , l. amendola , i. sawicki , and w. valkenburg , prl , 110 , 241305 ( 2013 ) s .- n .
zhang , and y .- z .
ma , arxiv:1303.6124v3 , ( 2013 ) .
l. verde , p. protopapas , and r. jimenez , arxiv:1306.6766 , ( 2013 ) .
xia , h. li , and x. zhang , prd , 88 , 063501 ( 2013 ) .
p. fleury , h. dupuy , and j .-
uzan , prl , 111 , 091302 ( 2013 ) . q. gao , and y. gong , arxiv : 1308.5627 , ( 2013 ) .
a. marchini , and v. salvatelli , prd , 88 , 027502 ( 2013 ) .
v. salvatelli , a. marchini , l. lopez - honorez , and o. mena , prd , 88 , 023531 ( 2013 ) .
e. hubble , proc .
, 15 , 168 ( 1929 ) .
a. g. riess , _ et al .
_ , aj , 116 , 1009 ( 1998 ) . s. perlmutter , _ et al .
_ , apj , 517 , 565 ( 1999 ) .
d. a. howell , nat .
comm . , 2 , 350 ( 2011 ) . w. m. wood - vasey , _ et al .
_ , apj , 666 , 694 ( 2007 ) .
m. hicken , _ et al .
_ , apj , 700 , 1097 ( 2009 ) .
r. kessler , _ et al .
_ , apjs , 185 , 32 ( 2009 ) .
n. suzuki , _ et al .
_ , apj , 746 , 85 ( 2012 ) . m. m. phillips , apj , 413 , l105 ( 1993 ) .
a. g. riess , w. h. press , and r. p. kirshner , apj , 438 , l17 ( 1995 ) .
s. perlmutter , _ et al .
_ , apj , 483 , 565 ( 1997 ) .
l. wang , g. goldhaber , g. aldering , and s. perlmutter , apj , 590 , 944 ( 2003 ) .
x. wang , l. wang , x. zhou , y. lou , and z. li , apj , 620 , l87 ( 2005 ) .
a. g. riess , w. h. press , and r. p. kirshner , apj , 473 , 88 ( 1996 ) .
r. tripp , a&a , 331 , 815 ( 1998 ) .
j. guy , p. astier , s. nobili , n. regnault , and r. pain , a&a , 443 , 781 ( 2005 ) . j. guy , _ et al .
_ , a&a , 466 , 11 ( 2007 ) .
a. conley , _ et al .
_ , apj , 681 , 482 ( 2008 ) .
r. a. sunyaev and ya .
b. zeldovich , comments astrophys .
space phys . , 4 , 173 ( 1972 ) . e. de filippis , m. sereno , w. bautz , and g. longo , apj , 625 , 108 ( 2005 ) . i. m. h. etherington , phil . mag . ,
15 , 761 ( 1933 ) .
i. m. h. etherington , gen .
grav . , 39 , 1055 ( 2007 ) .
g. f. r. ellis , gen .
grav . , 39 , 1047 ( 2007 ) .
p. schneider , j. ehlers and e. e. falco , gravitational lenses ( new york : springer , 1999 ) .
e. komatsu , _ et al .
_ , apjs , 192 , 18 ( 2011 ) . j. v. cunha , l. marassi and j. a. s. lima , mnras , 379 , l1 ( 2007 ) .
a. mantz , _ et al .
_ , mnras , 406 , 1773 ( 2010 ) .
a. g. riess , _ et al .
_ , aj , 118 , 2675 ( 1999 ) . w. hillebrandt , and j. c. niemeyer , araa , 38 , 191 ( 2000 ) . s. jha , a. riess , and r. kirshner , apj , 659 , 122 ( 2007 ) .
t. haugboelle , _ et al .
_ , apj , 661 , 650 ( 2007 ) .
a. conley , _ et al .
_ , apj , 664 , l13 ( 2007 ) . | there is an apparent tension between cosmological parameters obtained from _ planck _ cosmic microwave background radiation observations and that derived from the observed magnitude - redshift relation for the type ia supernova ( sne ia ) . here
, we show that the tension can be alleviated , if we first calibrate , with the help of the distance - duality relation , the light - curve fitting parameters in the distance estimation in sne ia observations with the angular diameter distance data of the galaxy clusters and then re - estimate the distances for the sne ia with the corrected fitting parameters .
this was used to explore their cosmological implications in the context of the spatially flat cosmology .
we find a higher value for the matter density parameter , @xmath0 , as compared to that from the original snls3 , which is in agreement with _ planck _
observations at 68.3% confidence .
therefore , the tension between _ planck _ measurements and sne ia observations regarding @xmath0 can be effectively alleviated without invoking new physics or resorting to extensions for the standard concordance model .
moreover , with the absolute magnitude of a fiducial sne ia , @xmath1 , determined first , we obtained a constraint on the hubble constant with snls3 alone , which is also consistent with _ planck_. |
let @xmath0 and @xmath1 be banach spaces over a real or complex field @xmath6 .
we use the traditional notations @xmath7 and @xmath8 for the unit sphere and the closed unit ball of the space @xmath0 , respectively .
the banach space of all bounded linear operators @xmath9 will be represented by @xmath10 .
in particular , when @xmath11 we denote @xmath12 simply by putting @xmath13 called the dual space of @xmath0 .
we say that an operator @xmath14 attains its norm if there exists @xmath15 such that @xmath16 . in this case
, we say that @xmath17 is a norm attaining operator and it attains its norm at @xmath18 .
the subset of @xmath10 of all norm attaining operators is denoted by @xmath19 .
we recall that a bounded linear operator is compact if the closure of the image of the unit ball is compact . for @xmath20
we denote by @xmath21 the euclidean space @xmath22 endowed with the @xmath23-norm @xmath24 with @xmath25 and @xmath26 and @xmath27 endowed with the sup - norm @xmath28 . to simplify the notation we put just @xmath29 when the field that we are working is specified .
the bishop - phelps theorem @xcite says that every bounded linear functional can be approximated by norm attaining functionals .
in other words , this means that the set of all norm attaining functionals on a banach space @xmath0 is dense in its dual space @xmath13 .
it was proved by lindenstrauss @xcite in 1963 that , in general , the same result does not work for bounded linear operators .
more precisely , he exhibited a banach space @xmath0 such that the set @xmath30 is not dense in @xmath31 .
seven years later , bollobs proved a numerical version of the bishop - phelps theorem which nowdays is known as the bishop - phelps - bollobs theorem @xcite . as a consequence of (
* theorem 2.1 ) we may enunciate this theorem as follows . _ ( bishop - phelps - bollobs theorem , @xcite , @xcite ) _ let @xmath32 and suppose that @xmath33 and @xmath34 satisfy @xmath35 then , there are @xmath36 and @xmath37 such that @xmath38 since we do not have a bishop - phelps version for bounded linear operators , we can not expect a bishop - phelps - bollobs version for this type of functions either .
so it is natural to study the conditions that the banach spaces @xmath0 and @xmath1 must satisfy to get a theorem of this nature . in 2008 , acosta , aron , garca and maestre introduced a definition in order to attain this problem .
[ bpbp ] ( bishop - phelps - bollobs property , @xcite ) we say that a pair of banach spaces @xmath4 satisfies the _ bishop - phelps - bollobs property _
( bpbp , for short ) when given @xmath39 , there exists @xmath40 such that whenever @xmath41 and @xmath15 are such that @xmath42 there are @xmath43 and @xmath36 such that @xmath44 there are many classical banach spaces that satisfy the bpbp .
for example , when @xmath0 and @xmath1 are finite dimensional spaces , the pair @xmath4 has this property ( * ? ? ? * proposition 2.4 ) .
also , if @xmath1 has the property @xmath45 of lindenstrauss , as @xmath46 and @xmath47 do , then the pair @xmath4 satisfies the bpbp for all banach space @xmath0 ( * ? ? ?
* theorem 2.2 ) .
more positive results appear when we assume that the range space @xmath1 is uniformly convex : the pairs @xmath48 , @xmath49 and @xmath50 all satisfy the bpbp ( see @xcite , @xcite and @xcite , respectively ) .
also if @xmath0 is a uniformly convex banach space , then the pair @xmath4 has the bpbp for all banach space @xmath1 ( * ? ? ? * theorem 3.1 ) .
just to help make the paper entirely accessible , we remember the concept of uniform convexity .
a banach space @xmath0 is uniformly convex if given @xmath39 , there exists @xmath51 such that whenever @xmath52 satisfy @xmath53 , then @xmath54
. we recall that if @xmath55 then @xmath56 is uniformly convex . in 2014 ,
kim and lee ( * ? ? ?
* theorem 2.1 ) given a characterization for uniformly convex banach spaces that associate this type of spaces with a peculiarity on the bishop - phelps - bollobs property .
more precisely , they proved that a banach space @xmath0 is uniformly convex if and only if given @xmath39 then we are able to find a positive number @xmath40 such that whenever @xmath57 and @xmath15 satisfy the relation @xmath58 there exists a vector @xmath36 such that @xmath59 note that the theorem says that a banach space @xmath0 is uniformly convex if and only if the pair @xmath60 satisfies the bishop - phelps - bollobs property without changing the initial functional @xmath61 , that is , the functional that almost attains its norm at some point @xmath18 is the same that attains its norm at the new vector that is close to @xmath18 . in this paper , we study this last result for bounded linear operators and we call it as the strong bishop - phels - bollobs property ( sbpbp ) .
first , we study it in the case that the real number @xmath62 depends of @xmath39 and also of a fixed operator @xmath17 , and we get some positives results about it . after that , we study the property in the uniform case , that is , when the number @xmath63 depends only of @xmath39 as we are used to work when we work with the bpbp . as we will see in the next section , we get many negatives results about the uniform case and we use these results to prove that there are uniformly convex banach spaces @xmath0 and infinite dimensional banach spaces @xmath1 such that the pair @xmath4 fails the sbpbp . finally , we give a complete characterization to the sbpbp for the pair @xmath5 which describes when these pairs satisfy the property .
in this section we study the strong bishop - phelps - bollobs property .
namely , we study the conditions that the banach spaces @xmath0 and @xmath1 must have and the hypothesis that we have to add to get a kim - lee type theorem for bounded linear operators . the kim - lee theorem is enunciated as follows .
[ kim - lee ] _
( kim - lee theorem , @xcite ) _ a banach space @xmath0 is uniformly convex if and only if given @xmath39 , there exists @xmath40 such that whenever @xmath57 and @xmath15 satisfy @xmath58 there is @xmath36 such that @xmath64 as we mentioned before , it is like a bishop - phelps - bollobs property without changing the initial functional @xmath61 .
a natural question arises : the result is true for other pairs of banach spaces @xmath4 with @xmath0 and @xmath1 having additional hypothesis considering bounded linear operators instead bounded linear functionals ?
although it is more natural put the question just like that , it is seems to us to be a strong problem in the sense that it will be hard to find concrete pairs of banach spaces @xmath4 satisfying the kim - lee theorem for bounded linear operators .
so we start by considering that the positive real number @xmath65 that appears in theorem [ kim - lee ] does not depend only of @xmath39 but also of a given operator @xmath17 fixed .
before we do that , we want to comment that carando , lassalle and mazzitelli @xcite defined a weak bpbp for ideals of multilinear mappings where the real positive number @xmath66 in the definition of the bpbp in this context depends of a given @xmath39 and also of the ideal norm of the operator defined on a normed ideal of @xmath67-linear mappings . in other words , a normed ideal of @xmath67-linear mappings @xmath68 where @xmath69 are banach spaces
has the weak bpbp if for each @xmath70 with @xmath71 and @xmath39 , there exists @xmath72 depending also of @xmath73 such that if @xmath74 satisfies @xmath75 , then there exist @xmath76 with @xmath77 and @xmath78 such that @xmath79 , @xmath80 and @xmath81 .
they proved that if @xmath82 are uniformly convex banach spaces then @xmath83 has the weak bpbp for ideals of multilinear mappings for all banach space @xmath1 . here
we will work on a different context where @xmath62 depends of a fixed operator not of the norm of the operators ideal as we may see in the following definition .
[ sbpbp ] a pair of banach spaces @xmath4 has the _ strong bishop - phelps - bollobs property _ ( or sbpbp for short ) if given @xmath39 and @xmath41 , there exists @xmath84 such that whenever @xmath15 satisfies @xmath85 there is @xmath36 such that @xmath86 if @xmath14 is compact and the pair @xmath4 has the sbpbp , then we say that the pair @xmath4 has the sbpbp for compact operators . in the next proposition , we assume that the domain space @xmath0 is finite dimensional to get the sbpbp first result .
in fact , as we mentianed in the previous paragraph , we get a positive real number @xmath84 that depends of @xmath39 and also of the operator @xmath41 fixed at the beggining of the proof .
the proof is by contradiction .
let @xmath41 .
if the result is false for some @xmath87 , then for all @xmath88 , there exists @xmath89 such that @xmath90 but @xmath91 , where @xmath92 for @xmath88 .
since @xmath0 is finite dimensional , there exists a subsequence @xmath93 of @xmath94 such that @xmath95 for some @xmath96 .
this implies that @xmath97 and since @xmath98 we get that @xmath99 and so @xmath100 .
then @xmath101 and @xmath102 which is a contradiction .
the proof is again by contradiction .
let @xmath41 be a compact operator .
if the result is false , for some @xmath87 and for all @xmath88 , there exists @xmath89 such that @xmath104 but @xmath91 for all @xmath88 .
then @xmath105 as @xmath106
. since @xmath0 is uniformly convex , @xmath0 reflexive and then by the smulian theorem there exists a subsequence @xmath93 of @xmath94 and @xmath96 such that @xmath93 converges weakly to @xmath18 . since @xmath17 is completely continuous , @xmath107 converges in norm to @xmath108 as @xmath109 .
therefore @xmath110 and so @xmath100 .
thus @xmath101 and @xmath111 this implies that @xmath112 and , using again that @xmath0 is uniformly convex , we get that @xmath113 which is a contradiction because of the following inequalities : @xmath114 consequently we get two more positives results about the sbpbp . in the next corollary
we prove that the pair @xmath4 has the sbpbp whenever @xmath0 is uniformly convex and @xmath1 has the schur s property which implies as a particular case that the pair @xmath115 satisfies the property .
corollary [ sbpbp3 ] shows that whenever @xmath0 is uniformly convex and @xmath1 has finite dimension , the pair @xmath4 has the sbpbp by using the fact that every bounded linear operator with finite dimensional range is compact .
we apply theorem [ sbpbp2 ] . to do this , we prove that every bounded linear operator @xmath103 is compact .
indeed , since @xmath17 is continuous , @xmath17 is @xmath116-@xmath116 continuous .
let @xmath117 .
since @xmath0 is reflexive , by the smulian theorem , there are a subsequence of @xmath118 ( which we denote again by @xmath118 ) and @xmath96 such that @xmath119 .
so @xmath120 .
now , since @xmath1 has the schur s property , @xmath121 in norm .
so @xmath17 is compact . by theorem [ sbpbp2 ]
the pair @xmath4 has the sbpbp .
[ sbpbp5 ] note that in definition [ sbpbp ] the operator @xmath103 must attains its norm if the pair @xmath4 has the sbpbp .
so if @xmath0 is not reflexive , then the pair @xmath4 fails the sbpbp for all banach space @xmath1 .
indeed , since @xmath0 is not reflexive , by the james theorem , there is a linear continuous functional @xmath122 such that @xmath123 for all @xmath124 .
let @xmath125 and define @xmath103 by @xmath126
. then @xmath127 and @xmath128 for all @xmath124 .
this implies that @xmath17 never attains its norm and then the pair @xmath4 can not have the sbpbp .
note that in the classic definition of the bishop - phelps - bollobs property the number @xmath62 depends only of @xmath39 .
so what happen if we ask for more in the definition of the strong bishop - phelps - bollobs property ? as we will see below , when we put @xmath129 to depends only of @xmath39 we get negative results . however , we use them to get examples of pairs of banach spaces that fail the sbpbp when the domain space is reflexive or when the range space has infinite dimension .
just to help to make reference we put a name of it .
we say that a pair of banach spaces @xmath4 has the _ uniform strong bishop - phelps - bollobs property _
( uniform sbpbp , for short ) if given @xmath39 , there exists @xmath40 such that whenever @xmath41 and @xmath15 are such that @xmath42 there is @xmath36 such that @xmath86 we observe that the kim - lee theorem says that a banach space @xmath0 is uniformly convex if and only if the pair @xmath60 has the uniform sbpbp where @xmath130 or @xmath131 . note also that if the pair @xmath4 satisfies the uniform sbpbp then the pair @xmath4 satisfies the bpbp .
the first thing that we notice is that if @xmath4 has the uniform sbpbp for some banach space @xmath1 , then @xmath0 must be uniformly convex . given @xmath132 , consider @xmath40 the positive real number that satisfies the uniform sbpbp for the pair @xmath4 .
we prove that the pair @xmath60 has the uniform sbpbp with the same @xmath133 .
indeed , let @xmath57 and @xmath15 be such that @xmath134 let @xmath125 and define @xmath14 by @xmath135 for all @xmath136 .
so @xmath137 and @xmath138 since the pair @xmath4 has the uniform sbpbp with @xmath133 , there exists @xmath36 such that @xmath139 and @xmath140 .
since @xmath141 the proof is complete . what about the reciprocal of corollary [ corsbpbp ] ?
the first thing that come to mind , since every hilbert space is uniformly convex , is to assume that the domain space @xmath0 is a hilbert space and try to find some banach space @xmath1 such that the pair @xmath4 satisfies the property .
but even in the simplest situation the result fails as we may see in the following example .
[ exsbpbp ] this example works for both real and complex cases . for a given @xmath39 , suppose that there exists @xmath40 satisfying the uniform sbpbp for the pair @xmath144 .
let @xmath145 be defined by @xmath146 for every @xmath147 .
for every @xmath148 , we have @xmath149 since @xmath150 , we obtain @xmath151 .
moreover , @xmath152 we prove now that every @xmath153 such that @xmath154 assumes the form @xmath155 for @xmath156 .
indeed , since @xmath157 and @xmath154 , we have @xmath158 .
since @xmath159 , we have @xmath160 and @xmath161 with @xmath156 . in summary
, we have a unit operator @xmath17 and a unit vector @xmath162 satisfying @xmath163 but if @xmath17 attains its norm at some point @xmath164 then @xmath165 with @xmath156 .
this contradicts the assumption that the pair @xmath144 has the uniform sbpbp since @xmath166 is far from @xmath162 in view of the fact that @xmath167 .
this shows that the pair @xmath168 fails to have the uniform sbpbp for @xmath130 or @xmath131 .
now what if we add on the hypothesis that both @xmath0 and @xmath1 are hilbert spaces ?
the answer for this question is still no as we can see below .
[ sbpbp8 ] this example works for both real and complex cases . for a given @xmath39 , suppose that there exists @xmath40 satisfying the uniform sbpbp for the pair @xmath169 .
define @xmath170 by @xmath171 for every @xmath147 .
then for every @xmath148 , @xmath172 and since @xmath173 , we have @xmath151 . also , we see that @xmath174 . now ,
if @xmath153 is such that @xmath175 , then @xmath176 and since @xmath177 , we get @xmath178 which implies that @xmath160 and using again that @xmath159 , we obtain @xmath179 with @xmath156 and so @xmath180 which contradicts the hypothesis that the pair @xmath169 has the uniform sbpbp . [ paramiguel1 ] let @xmath181 and @xmath182 be the space @xmath183 endowed with the @xmath23-norm and the @xmath184-norm , respectively , with @xmath185 ( or @xmath186 ) . given @xmath187 , there exists a bounded linear operator @xmath188 with @xmath189 such that let @xmath187 and @xmath185 .
define @xmath188 by @xmath194 for every @xmath195 .
if @xmath196 , since @xmath197 , we get @xmath198 which implies that @xmath199 .
since @xmath200 , we have @xmath189 .
now , let @xmath201 be such that @xmath202 .
we prove that @xmath179 with @xmath156 .
indeed , the equality @xmath203 implies that @xmath204 and since @xmath205 , we do the difference between these two equalities to get @xmath206 since @xmath197 and @xmath207 , @xmath208 and @xmath209 .
because of the above equality , we get that @xmath210 . but
@xmath211 which implies that @xmath212 thus @xmath160 and then @xmath179 with @xmath156 as desired .
so if @xmath191 is such that @xmath192 , then @xmath213 which completes the proof . as a consequence of this last result , we get that all the pairs @xmath214 fail the uniform sbpbp for @xmath185 for @xmath130 or @xmath131 .
now we study the pair @xmath215 in the real case when we put the sum norm on the range space .
unfortunatly , we can construct a bounded linear operator to get the same contradiction as before .
let @xmath216 and @xmath142 be the banach spaces @xmath217 endowed with the @xmath218-norm and the sum - norm , respectively .
given @xmath187 , there exists a bounded linear operator @xmath219 with @xmath189 such that let @xmath187 .
define @xmath219 by @xmath223 for all @xmath224 .
then @xmath225 so @xmath226 and @xmath227
. also , we have @xmath189 .
indeed , let @xmath148 .
then @xmath228 recall that for every @xmath229 , we may write @xmath230 . with this in mind , if @xmath231 , then @xmath232 and if @xmath233 likewise we have @xmath234 + \frac{1}{2 } |\beta x - y| = \max \ { -\beta x , -y \ }
\leq 1.\ ] ] since @xmath235 , we have @xmath189 . now
, suppose that @xmath153 is such that @xmath221 .
so , @xmath236 again , we have two cases . if @xmath237 , then @xmath238 which implies @xmath239 since @xmath240 , and if @xmath241 , then @xmath242 which implies @xmath243 . since @xmath244
, we obtain @xmath160 and @xmath245 .
this means that if @xmath246 attains its norm at some @xmath247 , then @xmath180 .
[ paramiguel2 ] let @xmath216 and @xmath182 be the space @xmath217 endowed with the 2-norm and the @xmath184-norm , respectively , with @xmath250 .
given @xmath187 , there exists @xmath251 with @xmath189 such that let @xmath250 and define @xmath254 by @xmath255 for every @xmath147 .
first of all , note that @xmath256 i.e. , @xmath257 .
this shows that @xmath258 .
next , we show that @xmath259 and also that the only points which @xmath17 attains its norm are at @xmath260 and @xmath261 .
to do so , we study the norm of the operator @xmath17 by using the following compact set : @xmath262 by symmetry , the norm of @xmath17 is the maximum of @xmath263 with @xmath166 in @xmath264 .
let @xmath265 a point of @xmath264 such that @xmath17 attains its norm at @xmath266 , that is , @xmath267 .
we consider @xmath268 as the segment that connect @xmath269 with @xmath162 , @xmath270 as the segment that connect @xmath269 with @xmath271 and @xmath272 as the arc that connect @xmath162 with @xmath271 .
see figure [ fig : maximum ] .
it is enough to study the values of @xmath273 on the set @xmath274 since the operator @xmath17 attains its norm at elements of the sphere and @xmath275 .
we have @xmath276 with @xmath277 defined as @xmath278 and @xmath279 defined as @xmath280 . since @xmath281 for every @xmath250 , then @xmath282 .
( on the other hand observe that if @xmath283 , then @xmath284 and if @xmath285 , then @xmath286 . )
thus if @xmath287 , then either @xmath288 and then @xmath289 would be a critical point of @xmath290 in @xmath291 , where @xmath292 or @xmath293 and in this case @xmath294 would be a critical point of @xmath295 in @xmath296 , where @xmath297.\ ] ] but , as we will see in the next lines , these can not happen because @xmath298 and @xmath299 for all @xmath300 and then @xmath301 .
indeed , we consider first the case that @xmath302 for every @xmath302 , we get @xmath303.\ ] ] for @xmath302 , we have that @xmath304 and since @xmath305 for every @xmath306 on this interval , we obtain that @xmath307 for every @xmath308 . a simply change of the letter @xmath290 by @xmath295 and @xmath306 by @xmath309 implies that @xmath299 for every @xmath310 .
everything we did so far was to prove that @xmath311 and that @xmath17 attains its norm on @xmath264 only at @xmath312 and @xmath313 .
therefore , we may conclude that @xmath17 attains its maximum at @xmath260 and at @xmath261 .
in other words , we proved that @xmath314 .
now , for @xmath315 , define @xmath251 by @xmath316 for every @xmath147 .
note that @xmath317 and @xmath318 . since @xmath319 , then @xmath199 . also , using that @xmath314 and that @xmath320 , we have that @xmath321 .
this implies that if @xmath164 is such that @xmath192 , then @xmath322 and therefore @xmath180 as we wanted .
what about the case that @xmath325 and @xmath250
? we will study the real case of this right now .
consider @xmath325 .
define @xmath326 by @xmath327 for every @xmath328 . then @xmath329 and @xmath330 .
also , if @xmath331 , then @xmath332 and so @xmath333 .
this implies that @xmath334 and @xmath335 , and therefore @xmath336 thus @xmath337 .
given @xmath315 , let @xmath338 be as in the proposition [ paramiguel2 ] with @xmath250 .
now , define @xmath339 by @xmath340 .
then @xmath341 .
also , @xmath342 and @xmath343 .
suppose that there exists @xmath344 such that @xmath345 .
then @xmath346 and then , as we can see in the proof of the proposition [ paramiguel2 ] , @xmath166 must be equals to @xmath271 or @xmath347 . in both cases
, we have that @xmath348 .
we just have proved the following result .
suppose that there exists @xmath40 that depends only of a given @xmath39 satisfying the property .
let @xmath352 and @xmath52 be such that @xmath353 for @xmath354 .
define @xmath355 by @xmath356 for all @xmath136 . then @xmath357 and @xmath358 .
moreover , since @xmath359 , we have that @xmath259 .
this shows that @xmath151 .
therefore , there exists @xmath360 such that @xmath361 and @xmath362 . since @xmath363 and @xmath364
, we have that @xmath365 . on the other hand , since @xmath366 we get a contradiction , since @xmath367 it is a consequence of the fact that the pair @xmath144 fails the uniform sbpbp that the pair @xmath368)$ ] also fails this property .
indeed , consider @xmath369 $ ] positive functions defined on @xmath370 $ ] such that @xmath371 , @xmath372 $ ] and @xmath373 $ ] .
define @xmath374 $ ] by @xmath375 for all @xmath376 $ ] and @xmath377 .
then @xmath378 is linear .
since for @xmath379 we have @xmath380 and for @xmath381 $ ] we have @xmath382 , we get that @xmath383 } | \varphi(x , y)(t)| = \max \ { |x| , |y| \ } = \|(x , y)\|_{\infty}\ ] ] for all @xmath377 , using the fact that @xmath371 .
this shows that @xmath378 is a linear isometry between @xmath384 and the closed subspace of @xmath385 $ ] generated by @xmath386 and @xmath387 .
now suppose by contradiction that the pair @xmath368)$ ] has the uniform sbpbp .
then there exists @xmath40 for all @xmath132 , the modulus of the uniform sbpbp for this pair .
let @xmath388 with @xmath151 and @xmath389 be such that @xmath390 define @xmath391 $ ] by @xmath392 .
then for all @xmath148 , we have that @xmath393 and since @xmath378 is an isometry , we get that @xmath394 which implies that @xmath395 . also , @xmath396 so , there exists @xmath397 such that @xmath398 and @xmath399 . since @xmath400 , we just have proved that the pair @xmath144 also has the uniform sbpbp which contradicts example [ exsbpbp ] and proposition [ sbpbp1 ] .
next we show that if @xmath1 is a @xmath218-dimensional banach space , then the pair @xmath401 does not have the uniform sbpbp . to do so
, we use the existence of the auerbach base for a finite dimensional banach space ( see , for example , ( * ? ? ?
* proposition 20.21 ) ) .
let @xmath1 be an @xmath402-dimensional banach space .
then there are elements @xmath403 of @xmath1 and @xmath404 of @xmath405 such that @xmath406 for all @xmath407 and @xmath408 for @xmath409 .
in fact , @xmath410 is a base of @xmath1 called the auerbach base of @xmath1 .
let @xmath411 and @xmath412 satisfying @xmath406 for @xmath413 and @xmath408 for @xmath354 . since @xmath414 is a base for @xmath1 , every @xmath415 has an expression in terms of @xmath416 and @xmath417 given by @xmath418 .
given @xmath187 , define the continuous linear operator @xmath419 by @xmath420 for all @xmath421 .
then for all @xmath422 , we have that @xmath423 then @xmath199 . also ,
note that @xmath424 .
so @xmath189 .
now let @xmath125 be such that @xmath425 .
then , using that @xmath426 we get that @xmath427 and therefore @xmath428 .
finally , if the pair @xmath401 has the uniform sbpbp , there exists @xmath40 satisfying the property .
if we put @xmath429 , there exists an operator @xmath430 such that @xmath151 , @xmath163 and for all @xmath125 which satisfies @xmath431 is such that @xmath432 .
this is a contradiction and the pair @xmath401 fails the uniform sbpbp .
it is clear but it is worth mentioning that if the pair @xmath4 has the uniform sbpbp , then the pair @xmath433 also has this property for all closed subspace @xmath434 of @xmath1 .
this implies , by the above proposition , that the pair @xmath401 fails the uniform sbpbp for all @xmath402-dimensional finite space @xmath1 .
anyway , just for curiosity , if @xmath435 , the proof of proposition [ yy ] works as well in this situation .
indeed , for @xmath187 we define @xmath436 by @xmath437 for all @xmath415 , where @xmath438 and @xmath439 is given by the auerbach basis .
then @xmath440 for @xmath441 and @xmath442 . to prove that @xmath199 , we add and subtract @xmath443 in @xmath444 where @xmath422 , to get @xmath445 .
now , it is clear that if @xmath246 attains its norm at some @xmath125 then @xmath446 for all @xmath441 .
if @xmath1 is a banach space which contains strictly convex @xmath218-dimensional subspaces , then there exists a uniformly convex banach space @xmath0 such that the pair @xmath4 fails the uniform sbpbp .
indeed , let @xmath434 be a subspace of @xmath1 such that @xmath434 is stricly convex and @xmath447 .
then @xmath448 is uniformly convex , since @xmath434 is finite dimensional . by proposition [ yy ]
, the pair @xmath433 fails the uniform sbpbp and by the above observation the pair @xmath4 can not have this property .
next we use the negatives results that we got so far about the uniform sbpbp to get examples of pairs @xmath4 that fail the strong bishop - phelps - bollobs prooperty . in remark
[ sbpbp5 ] we noted that if @xmath0 is not reflexive , then the pair @xmath4 does not have the sbpbp . in what follows we get examples of reflexive banach spaces @xmath0 such that the pair @xmath4 fails the sbpbp .
we also present a complete characterization for the pairs @xmath5 concerning this property by showing that there are cases that these pairs satisfy the property and other cases not ( see theorem [ sbpbp9 ] ) .
first of all we use the fact , which is showed in the next remark , that the pair @xmath449 fails the uniform sbpbp to get examples of banach spaces @xmath1 such that the pair @xmath450 fails the sbpbp ( see corollary [ sbpbp7 ] ) . [ sbpbp4 ] by corollary [ sbpbp3 ] , the pair @xmath451 has the sbpbp if @xmath452 .
but in the case of the uniform sbpbp , we get a negative result .
we note that the pair @xmath449 fails the uniform sbpbp . indeed ,
suppose by contradiction that this pair satisfies the property .
then given @xmath39 there exists @xmath40 such that whenever @xmath453 and @xmath454 are such that @xmath455 , there is @xmath456 such that @xmath139 and @xmath457 .
since the pair @xmath169 fails the uniform sbpbp , there exists some @xmath87 , a norm one linear operator @xmath458 and a norm one vector @xmath459 with @xmath460 such that there is no point @xmath461 such that @xmath462 and @xmath463 .
let @xmath464 be the projection on the first two coordinates , i.e. , @xmath465 for all @xmath466 . then @xmath467 .
define @xmath468 by @xmath469 .
then @xmath470 . let @xmath471 .
we have that @xmath472 then there exists @xmath473 such that @xmath474 and @xmath475 . since @xmath476 we get that @xmath477 . on the other hand , @xmath478 .
this is a contradiction and then the pair @xmath449 fails the sbpbp as desired .
the next theorem connects both sbpbp and uniform sbpbp in order to get examples of pairs of banach spaces @xmath4 that not satisfy the sbpbp by using the examples of the pairs that fail the uniform sbpbp .
suppose that the pair @xmath4 fails the uniform sbpbp .
then for each @xmath88 , there are @xmath480 , @xmath481 and @xmath89 with @xmath482 such that whenever @xmath124 satisfies @xmath483 we have @xmath484 .
define @xmath485 by @xmath486 for every @xmath487 .
since @xmath488 for all @xmath88 , we get that @xmath259 . on the other hand ,
@xmath489 so @xmath151 . we suppose that there is @xmath490 such that the pair @xmath491 has the sbpbp with this constant .
denote by @xmath492 the element of @xmath493 such that in the @xmath402-th position is @xmath166 and in the rest is zero for all @xmath494 .
we take @xmath88 to be such that @xmath495 observe that @xmath496 thus there is @xmath497 such that @xmath498 by the second inequality , we get that @xmath499 for all @xmath500 and @xmath501 . moreover , by the equality @xmath502 we get that @xmath503 since @xmath504 for all @xmath500 .
so @xmath505 .
this contradicts the beggining of the proof and we conclude that the pair @xmath491 fails the sbpbp whenever @xmath4 fails the uniform sbpbp .
* @xmath144 , * @xmath169 , * @xmath510 for @xmath185 .
* @xmath215 , * @xmath511 , for @xmath250 .
* @xmath512 for @xmath324 . * @xmath513 for @xmath325 and @xmath350 .
* @xmath368)$ ] , * @xmath401 , where @xmath514 .
\(i ) by the pitt s theorem every bounded linear operator from @xmath56 into @xmath520 with @xmath519 is compact . by theorem [ sbpbp2 ]
the pair @xmath5 has the sbpbp since @xmath56 is uniformly convex for @xmath521 .
\(ii ) let @xmath132 and @xmath185 .
consider @xmath56 and @xmath520 as the banach spaces @xmath522 and @xmath523 , respectively .
for each @xmath88 define @xmath524 by @xmath525 for each @xmath195 .
let @xmath526 and let @xmath527 be defined by @xmath528 for each @xmath529 we have that @xmath530 and then @xmath531 so @xmath259 . we consider the vectors @xmath532 thus we get that @xmath533 .
so @xmath151 .
suppose that there exists @xmath84 such that the pair @xmath5 has the sbpbp .
let @xmath88 be such that @xmath534 .
so since @xmath535 and @xmath536 there exists @xmath537 such that @xmath538 next we claim that @xmath539 for all @xmath540 .
indeed , suppose that there exists some @xmath541 such that @xmath542 .
thus @xmath543 which is a contradiction .
then @xmath539 for all @xmath540 .
because of that , we have @xmath544 this new contradiction shows that the pair @xmath5 fails the sbpbp for @xmath185
. * acknowledgement .
* this article is based on parts of my ph.d .
thesis at universidad de valencia and i would like to thank domingo garca , manuel maestre and miguel martn for their wonderful supervision on this work . | in this paper we introduce the strong bishop - phelps - bollobs property ( sbpbp ) for bounded linear operators between two banach spaces @xmath0 and @xmath1 .
this property is motivated by a kim - lee result which states , under our notation , that a banach space @xmath0 is uniformly convex if and only if the pair @xmath2 satisfies the sbpbp .
positive results of pairs of banach spaces @xmath3 satisfying this property are given and concrete pairs of banach spaces @xmath4 failing it are exhibited . a complete characterization of the sbpbp for the pairs @xmath5 is also provided . |
the cosmic star formation history ( sfh ) is an important test for galaxy formation models .
experimentally our knowledge of the sfh comes from @xcite up to @xmath5 and from observations of color - selected lyman break galaxies @xcite , ly@xmath6 emitters @xcite , uv+ir measurements @xcite , and grb observations @xcite at higher @xmath4 ( hereafter , these will be refereed to as h2006 , m2007 , b2008 , o2008 , r2008 , c2007 , y2008 and w2009 , respectively ) .
however , direct high-@xmath4 measurements constitute an extreme challenge even for the most powerful telescopes and remain sparse . due to their high luminosity grbs
can be used as sfh probes into the very distant universe @xcite , potentially to higher redshifts than allowed by galaxies alone .
for example , grb 090429b at @xmath7 @xcite is the current record - holder object , followed by a @xmath8 galaxy @xcite and grb 090423 , at @xmath9 @xcite . in principle
, the redshift distribution of grbs can give us important clues on the early stages of cosmic history .
in practice , the connection between grbs and the underlying host galaxy star formation mode is far from trivial ( see e.g. @xcite ) , thus making the probe value subject to uncertainties .
the purpose of this work is to show how _ principal component analysis _
( pca ) can be used to map the grb redshift distribution onto the cosmic sfh in a model - independent way .
pca has already been applied to other astrophysical and cosmological contexts @xcite , and recognized as a useful tool to reconstruct parameters without the introduction of _ ad hoc _ parameterizationscdm model with wmap7 best fit parameters @xcite , @xmath10 , and @xmath11 km s@xmath2 mpc @xmath2 . ] .
we assume that the formation rate of long grbs ( duration longer than 2 sec ) follows closely the sfh ( e.g. , @xcite ) .
hence , the comoving rate of observable grbs , @xmath12 is @xmath13 where @xmath14 is the field of view of the experiment , @xmath15 is the grb formation efficiency , @xmath16 is the beaming factor of the burst , @xmath17 is the fraction of grbs with measured redshift and @xmath18 is the luminosity threshold of a given experiment . in the following we discuss in detail each of the terms in eq .
( [ psigrb ] ) .
we define telescope - related quantities in eq .
( [ psigrb ] ) after _ swift _ specs , i.e. @xmath19 @xcite and @xmath20 .. ] the overall grb rate depends on @xmath21 , where @xmath22 is the opening angle of the jet . according to @xcite the average value of @xmath23 . using a radio transients survey , @xcite placed the upper limit @xmath24 .
we set @xmath25 as a fiducial value . the stellar initial mass function ( imf ) , @xmath26 , determines the fraction of stars massive enough to leave a black hole remnant .
current theories indicate that the threshold mass to trigger a grb is @xmath27@xcite .
however , only a fraction @xmath28 @xcite of black holes resulting from supernova explosion actually gives rise to a grb ; to be conservative , we considered @xmath29 .
hence , the grb formation efficiency factor per stellar mass is @xmath30 for simplicity , we assume a standard " salpeter imf , @xmath31 , with @xmath32 @xcite . the number of detectable grbs depends on the instrument sensitivity and intrinsic isotropic grb luminosity function . for the latter ,
we adopt a power - law distribution function of @xcite @xmath33 with @xmath34 erg s@xmath2 , @xmath35 and @xmath36 .
the luminosity threshold is then @xmath37 , where @xmath38 is the luminosity distance and @xmath39 is the bolometric energy flux limit of the instrument . in what follows we set @xmath40 @xcite . from eq .
( [ psigrb ] ) we can determine the number of observed grbs with redshift in ( @xmath4 , @xmath41 ) over a time interval , @xmath42 , in the observer rest frame : @xmath43 where @xmath44 is the comoving volume element per unit of redshift .
our observable , i.e. the cumulative number @xmath45 of grbs up to redshift @xmath4 , is given by @xmath46 so far we have discussed the physical meaning of all terms in eq .
( [ psigrb ] ) except the sfh .
since our aim is to build a model which is independent from the specific form of @xmath47 , we avoid making hypothesis about this quantity . instead , our intention is to derive @xmath47 by using pca and a ( mock or real ) data set whose data points are @xmath48 pairs .
the main goal of pca is the dimensionality reduction of an initial parameter space through the analysis of its internal correlations .
suppose we have a model composed by @xmath49 parameters .
if two of them are highly correlated , they are actually providing the same information .
this means that it is possible to rewrite the data in a new parameter space consisting of @xmath50 terms , with minimum loss of information ( for a complete review see @xcite ) .
this new set of parameters are recognized as the principal components ( pcs ) , or the eigenvectors of the fisher information matrix , .
we postulate that the data set is composed by @xmath51 independent observations , each one characterized by a gaussian probability density function , @xmath52 $ ] . in our notation , @xmath53 is a measurement of an independent variable ; @xmath54 represent the measurements of a quantity @xmath55 depending on @xmath53 ; @xmath56 is the uncertainty associated with the measures , and @xmath57 is the parameter vector of the theoretical model . in other words ,
we investigate a specific quantity , @xmath58 , which can be written as a function of the parameters @xmath59 . in this context , the likelihood function is given by @xmath60 and the fisher matrix is defined as @xmath61 brackets in eq .
( [ eq : fmdefinicao ] ) represent the expectation value .
we can now diagonalize * * , and determine the set of its eigenvectors / pcs , @xmath62 , and eigenvalues , @xmath63 . following the standard convention
, we enumerate @xmath64 from the largest to the smallest associated eigenvalue . our ability to determine the form of each pc
is given by @xmath65 .
the set @xmath62 forms a complete base of uncorrelated vectors .
this allows us to use a subspace of @xmath62 , @xmath66 , to rewrite @xmath58 as a linear combination of all the elements in @xmath66 , @xmath67 .
the data is then used to find the values of the linear expansion coefficients , @xmath68 ( for a detailed discussion see i2011 ) .
the question of how many pcs should be used in the final reconstruction , or how to choose the dimensionality of @xmath66 , depends on the particular data set analyzed and our expectation towards them .
to provide an idea of how much of the initial information ( variance ) is included in our plots , we shall order them following their _ cumulative percentage of total variance_. a reconstruction with the first @xmath69 pcs encloses a percentage of this value @xmath70 it is important to emphasize that each added pc brings its associated uncertainty ( @xmath71 ) into the reconstruction .
so , although the best - fitted reconstruction converges to the real " function as m increases , the uncertainty associated also raises . as a consequence , the question of how many pcs turns into a matter of what percentage of total variance we are willing to enclose . to specify our method of sfh reconstruction from grb data ,
let us consider a data set formed by @xmath72 measurements of the cumulative number of grb up to redshift @xmath73 , @xmath74 , and its corresponding uncertainty ( @xmath75 ) .
the likelihood is given by @xmath76.\end{aligned}\ ] ] using equation ( [ eq : fmdefinicao ] ) , the fisher matrix components are @xmath77 the fisher matrix determination is now a matter of calculating the derivatives of @xmath78 , for which we use the theoretical prescriptions of sec .
[ sec_grb_rate ] .
aiming at model independence and simplicity , we model the sfh as @xmath79 where @xmath59 are constants , @xmath80 is the total number of redshift bins , and @xmath81 is a window function which returns 1 if @xmath82 and 0 otherwise . using this description , we may write any functional form with resolution limited by our computational power . the derivatives of @xmath45 can be computed analytically : @xmath83 , \label{eq : nderiv}\end{aligned}\ ] ] where @xmath84 is a step function which returns 0 if @xmath85 and 1 otherwise , @xmath86 corresponds to the number of integer bins up to redshift @xmath4 , @xmath87 is the kronecker delta function , @xmath88 is the lower bound of the @xmath89-th redshift bin , and @xmath90 from these relations the fisher matrix can be computed and the functional form of the sfh reconstructed through pca . .
the black - thick line is the final reconstruction for each case and the red - dashed - thick lines corresponds to @xmath91 confidence levels .
the inset shows the cumulative percentage of total variance , @xmath92 . ]
[ [ sec : mock ] ] mock data + + + + + + + + + the mock sample is composed of data pairs @xmath93 , distributed in redshift bins of width @xmath94 , where @xmath95 represents the middle of each bin .
the fiducial model used for the sfh is a simple double - exponential function , @xmath96 , fitted to numerical results by @xcite , @xmath97 with @xmath98 for @xmath99 0.993 ; @xmath100 for @xmath101 and @xmath102 for @xmath103 . we generated 500 simulations , each realization with uncertainty in the determination of @xmath45 set to unit and containing 65 redshift bins .
after we generated the mock sample , the information about @xmath104 is discarded .
our goal from now on is to re - obtain the functional form of eq .
( [ rho_star ] ) from pca .
the fisher matrix is obtained as described above , assuming an observing time @xmath105 yr . as the mock sample purpose is to test the procedure under an ideal scenario , we did not include uncertainties in the parameters of eq .
( [ psigrb ] ) . having obtained the pcs , we can rewrite the sfh as @xmath106 where @xmath107 and @xmath108 are constants to be determined and @xmath69 is the number of pcs we choose to use in the reconstruction .
the simulated data points are then used to find the appropriate values for the parameters @xmath107 and @xmath108 as those that minimize the expression @xmath109 where @xmath110 for all redshift bins .
the reconstructions obtained using 1 to 6 pcs are shown in fig .
[ fig : rec_sim_err10_all ] .
the uncertainty in the final reconstruction was calculated by a quadrature sum that includes the parameters @xmath111 and the uncertainty in the determination of parameters @xmath112 and @xmath113 . from fig . [
fig : rec_sim_err10_all ] we can appreciate the success of the procedure in reconstructing the underlying unknown sfh in an ideal scenario , with increasing agreement as the number of pcs raises . confidence levels also become wider as @xmath69 increases , with the only exception of the reconstruction with 1 pc . since @xmath114 dominates the errors due to the limited freedom to fit the second peak of the fiducial model with only 1 pc . with 2 pcs fitting
the second peak becomes easier , and as a consequence , the magnitude of @xmath114 decreases to levels below those of @xmath111 .
[ [ sec : swift ] ] _ swift _ data + + + + + + + + + + + + after validating pca reconstruction under ideal conditions , we turn to the use of currently available _ swift _ data , and compare these results with independent measurements of sfh from the literature .
first , we need to properly choose our data set .
since only grbs with measured redshifts can be used in our analysis , the question of how the redshift measurements were obtained must be examined carefully .
grbs redshifts are generally obtained from optical afterglow spectra using absorption lines or photometry , or from the spectrum of the host galaxy using emission lines . as pointed by @xcite ,
different methods yield different redshift distributions : a visual inspection of fig .
[ fig : swift_hist ] illustrates this point .
most noticeably , the grb redshift distribution determined from their hosts lacks very high-@xmath4 events .
moreover , emission ( and to a lesser extent , absorption ) lines are susceptible to a selection effect known as the redshift desert " in the range @xmath115 @xcite .
additional bias sources are preliminary discussed by @xcite . to avoid systematic errors affecting the overall redshift distribution , our data sample
is composed by 120 _ swift _ grbs with redshift determined from absorption lines and photometry ( gray region in fig .
[ fig : swift_hist ] , top panel ) .
the next step is to choose the appropriate redshift bin width . in principle
, the quality of the reconstruction should increase with the number of bins .
however , as grb are discrete events , if we pick a bin width based on the available data ( for example , in such a way that each bin has at least one grb ) , the bins will be too wide ( @xmath116 ) . in this case , the assumption that the sfh is constant inside the bin will not hold , leading to reconstructions with bad resolution . to overcome this limitation we performed a gaussian kernel fit for a kernel @xmath117 , bin width @xmath118 and a total of @xmath119 bins . ] to the data ( black line in fig .
[ fig : swift_hist ] , top panel ) .
now we have a continuous probability distribution function ( pdf ) for @xmath120 , which follows the real data distribution and allows us to set the bin width as small as required .
we kept @xmath94 and use the pdf to calculate the cumulative number of observed grbs in each bin .
the comparison between the real data cumulative distribution and the one calculated via the fitted pdf are shown in fig .
[ fig : swift_hist ] , bottom panel .
the fisher matrix is calculated using @xmath121 yr of observation time .
the parameters @xmath75 were obtained by summing in quadrature the uncertainties in the quantities involved in eq .
( [ psigrb ] ) .
[ fig : rec_data_err10_all ] shows the first 2 pcs and the corresponding reconstruction using both of them , which already encloses more than @xmath122 of total variance . in the lower panel ,
the points correspond to completely independent measurements from the literature .
these data points are shown only for comparison purposes and have not been used in our calculations .
confidence levels .
the inset shows the cumulative percentage of total variance , @xmath92 . ]
we have proposed the use of pca as a powerful tool to reconstruct the cosmic star formation history exploiting the measured gamma - ray burst redshift distribution .
the procedure was successfully validated using synthetic data and next applied to actual _ swift _ data ( fig .
3 ) . it is important to highlight that the approach is completely independent of the initial choice of the theoretical model parameter vector , @xmath57 .
this has the obvious advantage of avoiding any a priori hypothesis on the sfh , @xmath123 .
however , the degeneracy between @xmath123 and any other factor we are failing to take into account can not be removed , i.e. the reconstructed sfh contains also the behavior of all the agents influencing the determination of @xmath45 and not included in the model .
for example , @xcite have argued that grb progenitors will have a low metallicity .
such an effect would be a consequence of the mass and angular momentum loss caused by winds in high - metallicity stars .
this would prevent such stars of becoming grbs and consequently , change their expected redshift distribution @xcite .
we implicitly considered that this and other such effects will span within the error bars in our analysis . in spite of the remaining uncertainties , which are probably less severe than those affecting other methods aimed at recovering the high-@xmath4 tail of the sfh
, there are robust indications that we can gather from the analysis of our results .
the first is that the combination of grb data and pca suggest that the level of star formation activity at @xmath124 could have been already as high as the present - day one ( @xmath1 yr@xmath2 mpc@xmath3 ) .
this is a factor 3 - 5 times higher than deduced from high-@xmath4 galaxy searches through drop - out techniques , similarly to the trend found by @xcite .
if true , this might alleviate the long - standing problem of a photon - starving reionization ; it might also indicate that galaxies accounting for most of the star formation activity at high redshift go undetected by even the most deep searches .
finally it is worth noticing that a sustained high-@xmath4 star formation activity is consistent with predictions of reionization models that simultaneously match a number of observable experimental constraints as the gunn - peterson effect , thomson free - electron optical depth , lyman limit systems statistics etc .
( @xcite , @xcite ) .
given the expected large growth of grb detections from the next generation of instruments , the method proposed here promises to become one of the most suitable and reliable tools to constrain the star formation activity in the young universe .
we thank k. ioka , r. salvaterra and n. yoshida for useful comments .
e.e.o.i . thanks capes ( 1313 - 10 - 0 ) for financial support . r.s.s .
thanks cnpq ( 200297/2010 - 4 ) for financial support .
af acknowledges support from ipmu where this research started .
this work was supported by wpi initiative , mext , japan . | we propose a novel approach , based on principal components analysis , to the use of gamma - ray bursts ( grbs ) as probes of cosmic star formation history ( sfh ) up to very high redshifts .
the main advantage of such approach is to avoid the necessity of assuming an _ ad hoc _ parameterization of the sfh .
we first validate the method by reconstructing a known sfh from monte carlo - generated mock data .
we then apply the method to the most recent _ swift _ data of grbs with known redshift and compare it against the sfh obtained by independent methods .
the main conclusion is that the level of star formation activity at @xmath0 could have been already as high as the present - day one ( @xmath1 yr@xmath2 mpc@xmath3 ) .
this is a factor 3 - 5 times higher than deduced from high-@xmath4 galaxy searches through drop - out techniques .
if true , this might alleviate the long - standing problem of a photon - starving reionization ; it might also indicate that galaxies accounting for most of the star formation activity at high redshift go undetected by even the most deep searches .
[ firstpage ] methods : statistical , gamma - ray burst , star formation |
in 2012 , h. amann introduced a class of ( possibly noncompact ) manifolds , called . roughly speaking , an @xmath0-dimensional riemannian manifold @xmath1 is _ uniformly regular _
if its differentiable structure is induced by an atlas such that all its local patches are of approximately the same size , all derivatives of the transition maps are bounded , and the pull - back metric of @xmath2 in every local coordinate is comparable to the euclidean metric @xmath3 . the precise definition of will be presented in section 2 below . in the sequel , we understand all our manifolds to be smooth and without boundary , unless stated otherwise .
the main objective of this short note is to prove that the family of coincides with the class of .
a manifold is said to be of bounded geometry if it has positive injectivity radius , and all covariant derivatives of the curvature tensor are bounded .
the precise definition of bounded geometry will be given later in this introductory section .
nowadays , there is rising interest in studying differential equations on non - compact manifolds , see @xcite , for instance .
it is a well - known fact that many well established analytic tools in euclidean space fail , in general , on riemannian manifolds .
for instance , for @xmath4 with @xmath5 , we can always find a sequence @xmath6 such that @xmath7 nevertheless , it is known that this maximum principle does not always hold true on non - compact riemannian manifolds . indeed , there are counterexamples provided by h. omori @xcite .
a cornerstone in the study of differential equations is the theory of function spaces . in order to study this theory on riemannian manifolds ,
it is natural to impose extra geometric conditions , most likely certain restrictions on the curvatures . among all efforts made to find proper assumptions , one extensively studied category is the class of .
a manifold @xmath1 is said to have a positive injectivity radius if there exists a positive number @xmath8 such that the exponential map at @xmath9 , @xmath10 , is a diffeomorphism from @xmath11 for all @xmath9 .
a manifold @xmath1 is of bounded geometry if it has a positive injectivity radius and all covariant derivatives of the curvature tensor are bounded , i.e. , @xmath12 where @xmath13 is the riemannian curvature tensor , and @xmath14 denotes the extension of the levi - civita connection over @xmath15 with @xmath16 .
let @xmath17 and @xmath18 be the tangent and the cotangent bundle , respectively . then @xmath15 is the @xmath19-module of all smooth sections of @xmath20 , the @xmath21-tensor bundles .
@xmath22 is the ( vector bundle ) norm induced by the extension of the riemannian metric @xmath2 from the tangent bundle to @xmath23 .
condition can be replaced by the boundedness of all covariant derivatives of the sectional curvature .
see ( * ? ? ?
* section 5.14 ) for a justification .
this condition can also be formulated equivalently by asking that in all normal geodesic coordinates of every local chart @xmath24 , we have @xmath25\geq c,\quad \max_{|\alpha|\leq k}\|\partial^\alpha g_{ij}\|_\infty\leq c(k ) , \quad k\in\nz,\ ] ] where @xmath26 is the local matrix expression of the metric tensor @xmath2 , for some constants @xmath27 .
see ( * ? ? ?
* section 7.2.1 ) .
the amount of literature on differential equations on is vast .
most of the work concerns heat kernel estimates and spectral theory .
see , for example , @xcite and the references therein . with additional restrictions like nonnegative ricci curvature , @xmath28-@xmath29 maximal regularity theory is established for second order elliptic operators .
see @xcite . to illustrate some of our recent results for differential equations on
, we look at linear differential operators @xmath30 of order @xmath31 acting on @xmath21-tensor fields , defined by @xmath32 where @xmath33 , the set of all sections of @xmath34 , and @xmath35 denotes complete contraction
. see ( * ? ? ?
* section 2 ) for a detailed discussion .
we consider the following initial value problem on a @xmath1 in hlder spaces : @xmath36 here @xmath37 for @xmath38 $ ] . for @xmath39 and @xmath16 , we define @xmath40 and @xmath41 we also set @xmath42 endowed with the conventional projective topology
. then @xmath43 the closure of @xmath44 in @xmath45 .
let @xmath46 .
now the _ little hlder _
space @xmath47 is defined by @xmath48 here @xmath49 is the continuous interpolation method , see ( * ? ? ?
* example i.2.4.4 ) . in @xcite , the theory of function spaces , including the _ little hlder _ space , is studied . a linear differential operator @xmath50 is said to be _ normally elliptic _
if there exists some constant @xmath51 such that for every pair @xmath52 with @xmath53 for all @xmath9 , the principal symbol of @xmath54 defined by @xmath55 satisfies @xmath56 and @xmath57 in the above , @xmath58 is the contravariant metric induced by @xmath2 .
we readily check that a normally elliptic operator must be of even order .
@xmath54 is called @xmath59-_regular _ if @xmath60 the following continuous maximal regularity theorem has been established by two of the authors .
[ y. shao , g. simonett @xcite ] [ s1 : cont - mr ] let @xmath1 be a uniformly regular riemannian manifold , @xmath61 $ ] , and @xmath62 .
suppose that @xmath54 is a @xmath63-th order normally elliptic and @xmath59-regular differential operator acting on @xmath21-tensor fields .
then for any @xmath64 equation has a unique solution @xmath65 equivalently , @xmath54 generates an analytic semigroup on @xmath66 and has the property of continuous maximal regularity .
here @xmath67 is some weighted @xmath68-space for a given banach space @xmath69 , see ( * ? ? ?
* section 3 ) for a precise definition .
one may observe from the statement of theorem [ s1 : cont - mr ] that no additional geometric assumption is needed .
so it generalizes the existing results on , see @xcite . by means of results of g.
da prato , p. grisvard @xcite , s. angenent @xcite and p. clment , g. simonett @xcite , theorem [ s1 : cont - mr ] gives rise to existence and uniqueness of solutions to many quasilinear or even fully nonlinear differential equations on .
see @xcite for example . a similar result to theorem [ s1 : cont - mr ] in an @xmath29-framework can be found in @xcite for second order initial boundary value problems . in (
* theorem 4.1 ) , it is shown that a manifold with bounded geometry is _ uniformly regular_. we aim at establishing the other implication , i.e. , a is of bounded geometry . in section 3
, we prove the following theorem .
[ s1:main thm ] suppose that @xmath1 is a uniformly regular riemannian manifold .
then * all covariant derivatives of the curvature tensor are bounded , i.e. , @xmath70 where @xmath13 is the riemannian curvature tensor ; * @xmath1 has positive injectivity radius ; * @xmath1 is complete .
[ s1 : rmk ] it is known that part ( b ) of theorem [ s1:main thm ] actually implies ( c ) .
see ( * ? ? ?
* proposition 1.2 ) .
we remind that a manifold @xmath1 is ( geodesically ) complete if all geodesics are infinitely extendible with respect to arc length .
the hopf - rinow theorem states that this is equivalent to asserting that @xmath71 is complete as a metric space with respect to the intrinsic metric induced by @xmath2 .
we find it illustrative to present two direct proofs of completeness of that do not use positive injectivity radius a priori .
a riemannian manifold is uniformly regular , if and only if it has a bounded geometry .
this article is organized as follows . in section 2
, we give the precise definition of and present several examples of this class .
this concept can be extended to manifolds with boundary , see @xcite . in section 3
, we provide a proof for theorem [ s1:main thm ] .
let @xmath1 be a @xmath72-riemannian manifold of dimension @xmath0 endowed with @xmath2 as its riemannian metric such that its underlying topological space is separable .
an atlas @xmath73 for @xmath71 is said to be normalized if @xmath74 . here
@xmath75 is the unit ball centered at the origin in @xmath76 .
we put @xmath77 .
the atlas @xmath78 is said to have _ finite multiplicity _ if there exists @xmath79 such that any intersection of more than @xmath80 coordinate patches is empty . put @xmath81 the _ finite multiplicity _ of @xmath78 and the separability of @xmath71 imply that @xmath78 is countable .
an atlas @xmath78 is said to fulfil the _ uniformly shrinkable _ condition , if it is normalized and there exists @xmath82 such that @xmath83 is a cover for @xmath84 .
following h. amann @xcite , we say that @xmath1 is a if it admits an atlas @xmath78 such that * @xmath78 is uniformly shrinkable and has finite multiplicity . if @xmath71 is oriented , then @xmath78 is orientation preserving . *
@xmath85 , @xmath86 , @xmath87 , and @xmath88 . *
@xmath89 , @xmath86 . here
@xmath3 denotes the euclidean metric on @xmath90 and @xmath91 denotes the pull - back metric of @xmath2 by @xmath92 .
* @xmath93 , @xmath86 and @xmath94 .
here @xmath95 , and it is understood that a constant @xmath96 , like in ( r2 ) , depends only on @xmath97 .
an atlas @xmath78 satisfying ( r1 ) and ( r2 ) is called a _ uniformly regular atlas_. ( r3 ) reads as @xmath98 * * suppose that @xmath1 and @xmath99 are both uniformly regular riemannian manifolds . then so is @xmath100 .
* let @xmath101 be a diffeomorphism of manifolds .
if @xmath1 is a uniformly regular riemannian manifold , then so is @xmath102 .
* @xmath103 is uniformly regular . *
every compact manifold is uniformly regular . *
every manifold with bounded geometry is uniformly regular .
* let @xmath104 , and @xmath105 : @xmath106 for @xmath107 $ ] .
assume that @xmath108 is a @xmath109-dimensional compact submanifold of @xmath110 .
the ( model ) @xmath111-funnel @xmath112 on @xmath113 is defined by @xmath114 it is a @xmath115-dimensional submanifold of @xmath116 .
the map @xmath117 is a diffeomorphism .
suppose that @xmath118 is an open covering of @xmath1 such that @xmath119 is isometric to some @xmath111-funnel , @xmath120 , and @xmath121 , @xmath122 are relatively compact in @xmath71 .
then @xmath1 is uniformly regular .
* @xmath1 is a stretched corner manifold , that is , @xmath118 is an open covering of @xmath1 with @xmath121 , @xmath122 relatively compact in @xmath71 , and @xmath119 is isometric to @xmath123 . here
@xmath108 is an @xmath109-dimensional compact submanifold embedded in the unit sphere of @xmath124 .
both @xmath125 and @xmath126 denote the standard metric of @xmath127 $ ] . the stretched ( model )
corner end @xmath128 is defined by @xmath129\times b \times ( 0,1 ] \}\subset \r^{d+2}.\ ] ] then @xmath1 is uniformly regular . * the open unit ball @xmath130 in the euclidean space @xmath76 equipped with the poincar disk metric ,
that is , @xmath131 is uniformly regular .
* @xmath1 is a @xmath132-riemannian manifold " . to be more precise , let @xmath133 be a compact manifold with boundary @xmath134 .
we extend @xmath134 by a stretched conic end , that is , @xmath135 , where @xmath69 is diffeomorphic to @xmath127\times \partial b$ ] and equipped with the metric @xmath136 .
this metric @xmath2 is called an exact @xmath132-metric .
then @xmath1 is uniformly regular . with the help of the first two examples
, we can construct more complicated based on example [ s2 : exp of urm](c)-(i ) . in order to prove some of the statements in examples [ s2 : exp of urm ] , it is convenient to introduce the concept of _ singular manifolds_. roughly speaking , a manifold @xmath137 is called a _ singular manifold _ in the sense defined in ( * ? ? ?
* section 2 ) , if there exists some @xmath138 such that @xmath139 is _ uniformly regular_. the function @xmath140 is called a _
singularity datum _ of @xmath141 .
a _ singular manifold _ is _ uniformly regular _ if @xmath142 . if two real - valued functions @xmath143 and @xmath2 are equivalent in the sense that @xmath144 for some @xmath145 , then we write @xmath146 .
* * the concept of stretched corner manifolds is used in @xcite . note that the conventional corner manifolds , see @xcite , are not _ uniformly regular _ , but indeed are _
singular manifolds_. * the concept of @xmath132-riemannian manifolds " was introduced by r.b .
melrose , see ( * ? ? ?
* chapter 2 ) for a detailed discussion .
example [ s2 : exp of urm](f ) implies that theorem [ s1 : cont - mr ] on can be considered as some extension of the theory of @xmath132-calculus .
[ * proof of example [ s2 : exp of urm ] * ] * \(a ) example [ s2 : exp of urm](a ) is a special case of ( * ? ? ? * theorem 3.1 ) .
\(b ) ( * ? ? ?
* lemma 3.4 ) implies example [ s2 : exp of urm](b ) .
\(c ) example [ s2 : exp of urm](c ) follows from ( * ? ? ?
* formula ( 3.3 ) ) .
\(d ) see ( * ? ? ?
* corollary 4.3 ) .
\(e ) see ( * ? ? ?
* theorem 4.1 ) , and also ( * ? ? ?
* lemma 2.2.6 ) and @xcite .
\(f ) see ( * ? ? ?
* theorem 1.2 ) .
\(g ) ( * ? ? ?
* example 5.1 , lemmas 5.2 and 6.1 ) imply that @xmath147 is a _ singular manifold _ with _ singularity datum _ @xmath148 .
then the assertion follows from ( * ? ? ?
* lemma 3.3 ) .
\(h ) we put @xmath149 . by (
* lemma 5.2 ) , we have @xmath150 ; dt^2)$ ] is a _ singular manifold _ with singular datum @xmath151 .
* lemma 3.4 ) implies that @xmath152 is a _ singular manifold _ with singular datum @xmath153 .
the unit sphere @xmath154 is _ uniformly regular _ by example [ s2 : exp of urm](d ) .
then by ( * ? ? ?
* theorems 3.1 , 8.1 and lemma 3.4 ) , @xmath155 is a _ singular manifold _ with _ singularity datum _ @xmath153 , where @xmath156 is the metric on @xmath154 induced by @xmath3 .
it is easy to see that @xmath157 is diffeomorphic to @xmath158 .
now the statment of example [ s2 : exp of urm](h ) follows from the definition of _ singular manifolds _ and ( * ? ? ? * lemma 3.4 ) .
\(i ) to see that example [ s2 : exp of urm](i ) holds true , by ( * ? ? ?
* lemma 3.3 ) we only need to show that @xmath159 is _ uniformly regular_. ( * ? ? ?
* example 5.1 , lemmas 5.2 and 6.1 ) imply that @xmath160 is a _ singular manifold _ with _ singularity datum _ @xmath161 . by definition
, @xmath159 is _ uniformly regular_.
[ * proof of theorem [ s1:main thm](a)-(b ) * ] * \(a ) to prove part ( a ) , we need to first introduce some notations .
let @xmath16 . for abbreviation , we set @xmath162 , and @xmath163 is defined alike .
given local coordinates @xmath164 , @xmath165 and @xmath166 , we set @xmath167 with @xmath168 .
the local representation of @xmath169 with respect to these coordinates is given by @xmath170 with coefficients @xmath171 defined on @xmath172 .
we can also extend the riemannian metric @xmath173 from the tangent bundle to any @xmath21-tensor bundle @xmath23 such that @xmath174 by @xmath175 in every coordinate with @xmath176 , @xmath177 and @xmath178 in addition , @xmath179 , @xmath180 is called the ( vector bundle ) norm induced by @xmath2 .
the riemannian curvature tensor , in every coordinate , reads as @xmath181 where @xmath182 , and @xmath183 are the christoffel symbols with respect to the levi - civita connection @xmath14 , see @xcite . for @xmath184 , @xmath185 and @xmath186 , @xmath187 where @xmath188 with @xmath189 a linear combination of the elements of @xmath190 the coefficients being polynomials in the derivatives of the christoffel symbols of order at most @xmath191
. see ( * ? ? ? * formula ( 3.18 ) ) .
it follows from ( * ? ? ?
* formula ( 3.19 ) ) that @xmath192 one can easily infer from ( r4 ) that @xmath193 where @xmath58 is the induced contravariant metric .
now combing with ( r4 ) , - imply the asserted statement .
\(b ) it is instructive to comment on the intuition behind our proof of positive injective radius , as well as some of its geometric underpinnings .
the breakdown of injectivity of the exponential map at a point @xmath194 will happen , as it is known , at the first time when any two geodesics leaving @xmath194 intersect each other . whether or not this happens , and when it happens , is dictated by the curvature of the manifold , which governs the focusing of such geodesics , and with large curvature values contributing for potentially smaller injectivity radius . in our case ,
the uniform bounds on derivatives of the metric assure uniform bounds on the curvature .
thus , intuitively , it is not unexpected that uniformly regular manifolds posses a lower bound on the injectivity radius .
we can turn this idea into a more geometric appealing proof , which we now state .
the _ uniformly shrinkable _ condition implies that there exists a number @xmath82 such that @xmath195 still forms an open cover of @xmath71 .
we claim that there exists some constant @xmath145 such that @xmath196 for any @xmath86 , @xmath197 , and @xmath198 . here
@xmath199 denotes all the points on @xmath71 with distance less than @xmath200 to @xmath194 with respect to the intrinsic metric .
indeed , put @xmath201 , and let @xmath202 denote the distance function , that is , @xmath203\to
\m \text { piecewise $ c^\infty$-curve with } \gamma(0)=\p , \
, \gamma(1)={\sf q } \},\ ] ] where @xmath204 is the length of @xmath205 with respect to the intrinsic metric induced by @xmath2 , see @xcite .
for any @xmath206 and on any piece - wise smooth curve @xmath207\rightarrow \m$ ] connecting @xmath208 and @xmath194 , i.e. , @xmath209 and @xmath210 , we take @xmath211 to be the first escape time of @xmath205 out of @xmath212 . then letting @xmath213
, one can compute by means of ( r3 ) @xmath214 here the constant @xmath215 is the same as in , and @xmath207\rightarrow \m$ ] runs over all piecewise smooth curves connecting @xmath208 and @xmath194 .
this proves the first part of inequality .
the second part of follows in a similar manner .
given any point @xmath9 , there exists some @xmath86 such that @xmath216 .
set @xmath217 and @xmath218 .
@xmath219 can be identified with a collar neighborhood of @xmath220 by a diffeomorphism @xmath221 : \
, \p \mapsto ( \psk(r\vpk(\p)/ |\vpk(\p)| ) , |\vpk(\p)|/(r+\delta ) ) .\ ] ] we can attach a neck , @xmath222 $ ] , to @xmath223 along its boundary , obtaining the new manifold @xmath224),\ ] ] see figure [ fig ] . here
the notation @xmath225 means that we take the union of @xmath223 and @xmath226 $ ] , and @xmath227 defines an equivalence relation between @xmath228 and one component of the boundary of @xmath226 $ ] , i.e. , @xmath229 .
let @xmath230 be a copy of @xmath223 . as before
, we can identify @xmath231 with @xmath232 by a diffeomorphism @xmath233 .
attach @xmath230 to the other end of the neck , @xmath222 $ ] , to construct a new compact manifold @xmath234 ( see figure [ fig ] again ) , i.e. , @xmath235 ) \cup_{\psi_\kappa } \ohkc.\ ] ] by means of the diffeomorphism @xmath233 , we can identify @xmath236 with @xmath237 .
based on this observation , we can introduce a differentiable structure on @xmath234 .
such a smooth differentiable structure on @xmath234 exists as long as there is a diffeomorphism @xmath238 .
the reason for us to introduce the maps @xmath227 and @xmath233 explicitly is to show that we can construct a metric on the new manifold @xmath234 satisfying some uniform boundedness conditions on several geometric quantities of @xmath234 , as shown in the condition below .
is constructed out of gluing the boundary of two balls to a common cylindrical neck .
] choose smooth cutoff functions @xmath239 , \quad \chi|_{\psk(\b(0,r+\delta/2))}\equiv 1,\ ] ] and @xmath240 , \quad \zeta|_{\partial\otk \times [ ( 1 + 3r)/2(1+r),3-(1 + 3r)/2(1+r)]}\equiv 1.\ ] ] we denote the copy of @xmath241 on @xmath230 by @xmath242 .
endow @xmath234 with the metric @xmath243 here , @xmath244 is the standard product metric on @xmath245 .
we can choose @xmath246 and @xmath247 to be independent of @xmath194 and @xmath86 .
then there exists a constant @xmath248 satisfying @xmath249 here @xmath250 , @xmath251 and @xmath252 are the diameter , volume and sectional curvature of @xmath253 , respectively .
indeed , the first two inequalities are straightforward . to verify the third inequality
, we first check that , in the neighborhood @xmath254 of @xmath220 , the metric @xmath2 satisfies @xmath255\geq m,\quad \|\widetilde{\phi}_\kappa^ * g|_{\ohk}\|_{k,\infty}\leq m(k ) , \quad k\in \nz , \end{aligned}\ ] ] with @xmath256 and @xmath257 independent of @xmath194 and @xmath86 . here
@xmath258 : \
, \p \mapsto ( r\vpk(\p)/ |\vpk(\p)| , |\vpk(\p)|/(r+\delta ) ) .\ ] ] this assertion follows easily from ( r3 ) , ( r4 ) and direct computation .
since the pull back metric of @xmath244 under the map @xmath259 satisfies an analog of , the boundedness of @xmath252 follows from a similar argument to the proof of part ( a ) .
now it follows from ( * ? ? ?
* theorem 5.8 ) that @xmath234 has a lower bound on its injectivity radius , i.e. , @xmath260 for some @xmath261 . restricting back to @xmath262 ,
it follows from that there exists some @xmath263 independent of @xmath194 such that any normal geodesic , i.e. , geodesics with unit speed , starting from @xmath194 can not leave @xmath223 in @xmath264 , and thus will not close in @xmath71 within @xmath265 , where @xmath266 .
this proves the injectivity radius of @xmath194 is larger than or equal to @xmath267
. see ( * ? ? ?
* section 6.6 ) .
since the constant @xmath248 is independent of @xmath194 and @xmath86 , so is @xmath268 .
therefore we have proved the asserted statement . as mentioned in remark [ s1 : rmk ] , we will give two independent proofs for theorem [ s1:main thm](c ) .
( i ) by the hopf - rinow theorem , it suffices to show that @xmath1 is complete as a metric space in the intrinsic metric induced by @xmath2 .
as in the proof of theorem [ s1:main thm](b ) , there exists a number @xmath82 such that @xmath195 still forms an open cover of @xmath71 .
assume that @xmath269 is a cauchy sequence in the intrinsic metric .
then for any @xmath270 , where @xmath198 is fixed and @xmath215 is the constant in , there exists some @xmath271 such that @xmath272 for all @xmath273 .
since @xmath274 covers @xmath71 , for any @xmath275 , we can find some @xmath276 with @xmath277 .
then for any @xmath278 , we have @xmath279 since the maps @xmath280 and @xmath92 are continuous between @xmath281 and @xmath282 , it is easy to see that we actually have @xmath283 hence there exists a convergent subsequence @xmath284 for some @xmath285 .
since @xmath269 is cauchy , this implies that @xmath286 in the intrinsic metric .
( ii ) geodesic completeness of a @xmath1 can also be established directly by considering the equations for geodesics . given any @xmath9 and @xmath287
, we assume that @xmath216 .
then the equation of the geodesic starting from @xmath288 with initial velocity @xmath287 in the local coordinate @xmath289 reads as @xmath290 where @xmath291 . without loss of generality
, we may assume that @xmath292 .
setting @xmath293 , equation can be rewritten as follows .
@xmath294 let @xmath295 , and @xmath296 be so defined that @xmath297 equation is equivalent to @xmath298 by , there exists a constant @xmath256 uniform in all indices and @xmath299 , such that @xmath300 we fix some @xmath301 .
then @xmath302 [ s3 : unif exist of geo ] given any , @xmath303 , has a unique nonextendible solution @xmath304 on @xmath305 , where @xmath306 . by the picard - lindelf theorem , has a unique local in time solution on a maximal interval of existence @xmath305 with @xmath307 .
suppose that for some @xmath308 , @xmath309 . for @xmath310 , we have @xmath311 let @xmath312 } |z^i(s)|$ ] for @xmath313
. then we infer from the above equation that @xmath314 let @xmath315 be the solution to the equation @xmath316 .
it is easy to see that @xmath317 implies that @xmath318\cup [ \alpha_2(t),\infty)$ ] .
we will show that in fact @xmath318 $ ] .
let @xmath319 at @xmath320 , we have @xmath321 .
hence , @xmath322 is nonempty . by the continuity of @xmath323 and @xmath324 in @xmath325
, @xmath322 is closed in @xmath113 .
for the same reason , @xmath326 is also closed in @xmath113 .
then @xmath322 is open , and thus @xmath327 .
therefore @xmath328 for @xmath329 .
it implies that latexmath:[\[\begin{aligned } \label{s3 : bd of z } @xmath329 , this yields @xmath331 we thus infer that @xmath332 and contradict ( * ? ? ?
* theorem 7.6 ) .
this proves the uniform lower bound for @xmath333 , i.e. , @xmath334 .
lemma [ s3 : unif exist of geo ] implies that , given any positive constant @xmath248 , for any initial velocity @xmath335 , the maximal interval of existence @xmath336 for the solution to is uniform , i.e. , @xmath337 for some @xmath338 independent of @xmath308 .
given @xmath9 , any geodesic @xmath339 starting from @xmath194 with initial velocity @xmath287 fulfilling @xmath340 satisfies equation in the local coordinate @xmath341 with @xmath342 .
in view of ( r3 ) , @xmath343 fulfils @xmath344 for some @xmath248 independent of @xmath345 . therefore , @xmath339 exists on some @xmath346 $ ] , where @xmath347 is independent of @xmath345 . since geodesics are parameterized with respect to arc length , any geodesic is infinitely extendible .
this gives an alternative proof for the geodesic completeness of @xmath1 .
the concept of with boundary is defined by modifying the _ normalization _ condition of the atlas @xmath78 as follows . for those local patches
@xmath348 , @xmath349 , where @xmath350 is the closed half space @xmath351 . see @xcite .
the proof of theorem [ s1:main thm](a ) and ( c ) still works for with non - empty boundary . however , the concept of positive injectivity radius for manifolds with boundary needs to be defined separately for @xmath352 and @xmath353 .
see @xcite for precise definitions of interior and boundary injectivity radius .
the idea of the proof for theorem [ s1:main thm](b ) can still be applied to in the case of non - empty boundary with necessary modifications .
but , in this case , these manifolds are no longer geodesically complete , but only complete as a metric space .
therefore , roughly speaking , with boundary are complete as a metric space , with positive interior and boundary injectivity radius , and have bounded derivatives of the curvature tensor .
the second and third authors would like to thank stefan meyer for helpful discussions .
we would like to express our appreciation to the reviewer for insightful remarks .
mazzucato , v. nistor , mapping properties of heat kernels , maximal regularity , and semi - linear parabolic equations on noncompact manifolds . j. hyperbolic differ . equ . * 3 * , no . 4 , 599 - 629 ( 2006 ) .
melrose , the atiyah - patodi - singer index theorem .
research notes in mathematics , 4 .
( a k peters , ltd . , wellesley , ma , 1993 ) .
y. shao , a family of parameter - dependent diffeomorphisms acting on function spaces over a riemannian manifold and applications to geometric flows .
3.nodea nonlinear differential equations appl . *
22 * , no . 1 , 45 - 85 ( 2015 ) . | we establish the equivalence between the family of uniformly regular riemannian manifolds without boundary and the class of manifolds with bounded geometry . |
if borrowers have only assets that when liquidated generate cash in a local currency different from the currency in which their debt is due , their default risk will be higher than in the one currency case , as a consequence of the additional exchange rate risk .
the increase in default risk is reflected both in higher probabilities of default ( pds ) as well as in higher asset correlations between the borrowers . in this note , by modifying merton s model of the default of a firm , we derive some simple relations between the pds without and with exchange rate risk , between the borrowers asset correlations without and with exchange rate risk , and pds and asset correlations when taking account of exchange rate risk . in general , the formulae we derive include as parameters the borrowers asset volatilities , the exchange rate volatility , and the mean logarithmic ratio of the exchange rates at times 1 and 0 .
however , assuming independence of the exchange rate and the borrowers asset values as well as zero mean logarithmic ratio of exchange rates at times 1 and 0 yields a relation between the borrowers asset correlation without and with exchange rate risk and the borrowers pds without and with exchange rate risk that does not require knowledge of additional parameters ( see equation ) .
in the special case of borrowers with identical individual risk characteristics (= pds ) , relation can be stated as follows : @xmath0 where @xmath1 and @xmath2 denote the original pd and asset correlation without exchange rate risk and @xmath3 and @xmath4 denote the pd and asset correlation when there is additional exchange rate risk .
both and can be understood as consistency conditions that should be satisfied when the risk parameters pd and asset correlation are to be adjusted for incorporating exchange rate risk .
we describe in section [ sec : just ] the background of the model we use . in section [
se : metho ] , it is shown how the results are derived from the model .
the note concludes with a brief discussion of what has been reached .
as in merton s model for the default of a firm @xcite , we assume that @xmath5 , the borrower s asset value as a function of time , can be described by a geometric brownian motion , i.e. @xmath6 where @xmath7 is the asset value at time @xmath8 ( today ) , @xmath9 is the drift of the asset value process , @xmath10 is its volatility , and @xmath11 denotes a standard brownian motion that explains the randomness of the future asset values . similar to , we assume that @xmath12 , the exchange rate of the two currencies at time @xmath13 , can be described as another geometric brownian motion @xcite , i.e. @xmath14 where @xmath15 is the exchange rate at time @xmath8 , @xmath16 is the drift of the exchange rate process , @xmath17 is its volatility , and @xmath18 denotes another standard brownian motion that explains the randomness of the future exchange rates .
the brownian motions @xmath19 are correlated with correlation parameter @xmath20 , i.e. @xmath21\ = \ r , \quad 0 \le s < t.\ ] ] as in merton s model of the default of a firm , the borrower defaults after one year ( i.e. @xmath22 ) if her or his asset value by then has fallen below her or his level of due debt @xmath23 . however , debt is due in a currency different from the currency in which the asset value is denominated .
hence the asset value must be multiplied with the exchange rate at time 1 : @xmath24 from an economic point of view , it is convenient to divide both sides of by @xmath15 .
this leads to @xmath25 with @xmath26 the advantage of compared to is the fact that on the one hand the debt level is expressed as a value in the local currency of the borrower s assets with an exchange rate as observed today . on the other hand ,
compared to the one currency case the volatility of the left - hand side of is higher because it includes the factor @xmath27 that reflects the change of the exchange rate between today and time 1 .
this effect might be mitigated to some extent by the difference of the interest rates in the two countries .
for the purpose of this note , however , it is assumed that mitigation by interest rates differences can be neglected .
this assumption seems justified in particular when the debt is composed of fixed rate loans or is short - term . taking the logarithm of both sides of and standardisation of the random variable @xmath28 lead to @xmath29
define now @xmath30 , @xmath31 , and @xmath32 to arrive at @xmath33 in , @xmath34 is the logarithmic ratio of the exchange rates at times 1 and 0 and is jointly normally distributed with @xmath35 . as a consequence from ,
the correlation of @xmath36 and @xmath37 is given by @xmath38\ = \
r.\ ] ] note that , due to the convexity of the exponential function , @xmath39= 1 $ ] is * not * equivalent to @xmath40 = 0 $ ] but to @xmath40 = - \tau^2/2 $ ] . if @xmath40 = 0 $ ] on the other hand , then @xmath39= \tau^2/2 $ ] .
* see also the references therein ) suggested to take model of a borrower s asset value as the basis for a model of the joint default behaviour of several borrowers . as we consider the evolution in one period only , standardisation similar to the one used for deriving shows that the pds @xmath41 , @xmath42 of borrowers 1 and 2 can be characterised by the equation @xmath43 where @xmath44 .
the numbers @xmath45 are called _
default thresholds_. the correlation @xmath2 is called _ asset correlation _ because it can be interpreted as the correlation of the changes in asset values of the two borrowers .
equation does not yet include the case of assets and debt being denominated in different currencies .
assume that borrowers 1 and 2 receive their revenues mainly from one country and , hence , have their assets denominated in the local currency .
assume , however , that their debt is due in another currency .
combining the reasoning that led to and with the reasoning for then yields a description of the joint default behaviour under exchange rate uncertainty as follows : the default events for borrowers 1 and 2 are characterised by the equation @xmath46 where @xmath47 and @xmath48 are as in . should be the same as in might be surprising at first glance .
it follows , however , from the fact that in the foreign currency debt amount has been converted to a local currency amount at today s exchange rate .
additional uncertainty and higher default risk are then caused by the fact that the exchange rate in one year may differ from today s rate .
] @xmath36 is related to the annual change in the exchange rate between the both currencies ( see ) .
moreover , @xmath34 is assumed to be jointly normally distributed with @xmath47 : @xmath49\ = \
r_i , \quad i=1,2.\ ] ] @xmath50 denotes the volatility of the asset value process of borrower @xmath51 as in .
equation and the assumptions on the joint normal distribution of exchange rate and asset values readily imply the following result on the change in pd and asset correlation when exchange risk for the foreign currency denominated risk is taken into account .
[ pr:1 ] when the borrowers assets and debt are denominated in different currencies , their probabilities of default should be adjusted the following way : @xmath52\ = \ \phi\left(\frac{c_i - \nu/\sigma_i}{\sqrt{\tau^2/\sigma_i^2 + 1 + 2\,r_i\,\tau/\sigma_i}}\right).\ ] ] in , @xmath53 depends on the one - currency - only pd @xmath54 according to , @xmath16 and @xmath17 denote the exchange rate drift and volatility according to , @xmath55 stands for the volatility of borrower @xmath51 s assets according to , and @xmath56 is the correlation of asset value changes and exchange rate changes according to . additionally , the borrowers asset correlation should be adjusted : @xmath57 \ =
\ \frac{\varrho + r_1\,\tau/\sigma_1 + r_2\,\tau/\sigma_2 + \tau^2/(\sigma_1\,\sigma_2 ) } { \sqrt{\tau^2/\sigma_1 ^ 2 + 1 + 2\,r_1\,\tau/\sigma_1}\,\sqrt{\tau^2/\sigma_2 ^ 2 + 1 + 2\,r_2\,\tau/\sigma_2}},\ ] ] where @xmath2 denotes the one - currency - only asset correlation of the borrowers according to and all other quantities are as in . in order to apply adjustments to pd and asset correlation
according to proposition [ pr:1 ] , in particular estimates of the asset volatilities @xmath58 must be provided . as in general the asset value processes can not be directly observed , such estimates are notoriously hard to obtain . under additional simplifying assumptions ,
proposition [ pr:1 ] nonetheless implies the following useful consistency condition that might be taken into account when exchange rate risk adjustments are applied to pds and asset correlations .
[ co:1 ] if the exchange rate @xmath36 and the asset values @xmath47 are uncorrelated ( i.e. @xmath59 in ) and the mean logarithmic ratio of the exchange rates at times 1 and 0 can be neglected ( i.e. @xmath40 = \nu = 0 $ ] ) , then the following equation links up the adjusted pds and asset correlations : @xmath60 to see that corollary [ co:1 ] follows from proposition [ pr:1 ] , solve for @xmath61 and insert the result into .
equation can be generalised easily for @xmath62 .
generalisation for the case @xmath63 , however , would again involve asset volatilities and exchange rate volatility as additional parameters .
considerable simplification can be reached in when it is assumed that the individual risk characteristics of the borrowers are identical ( homogeneous portfolio assumption ) , i.e.@xmath64 in this case can be equivalently written as .
figure [ fig:1 ] presents for some fixed combinations of original pds and original asset correlations the relationship of adjusted pd and adjusted asset correlation .
note that implies that @xmath65 as long as @xmath66 if @xmath67 and @xmath68 .
hence the curves in figure [ fig:1 ] start in the original pds .
moreover , all the curves converge towards correlation 1 when the original pd approaches 0.5 .
in general , the curves are strictly concave such that the asset correlations grow over - proportionally as long as the adjusted pds are small , and under - proportionally when the adjusted pds are high ( close to 0.5 ) .
[ fig:1 ] + 270
taking recourse to well - known models by @xcite , @xcite , and @xcite we have derived a simple model for incorporating exchange rate risk into the pds and asset correlations of borrowers whose assets and debt are denominated in different currencies . on principle , the results can be used to derive values of pds and asset correlations that are adjusted to take account of exchange rate risk .
however , this requires knowledge of parameters like the borrowers asset value volatility that are hard to estimate .
another type of result is potentially more useful because thanks to additional assumptions like the independence of exchange rate and asset value processes it represents a consistency condition for the joint change of pds and asset correlations when exchange rate risk is taken into account .
this latter result does not involve asset value volatilities or other inaccessible parameters . | intuitively , the default risk of a single borrower is higher when her or his assets and debt are denominated in different currencies .
additionally , the default dependence of borrowers with assets and debt in different currencies should be stronger than in the one - currency case . by combining well - known models by @xcite , @xcite , and
@xcite we develop simple representations of pds and asset correlations that take into account exchange rate risk . from these results
, consistency conditions can be derived that link the changes in pd and asset correlation and do not require knowledge of hard - to - estimate parameters like asset value volatility . |
in this article , we consider the classic problem : find the gravitational field which is produced by a point massive particle . if particle is at rest , then the gravitational field is spherically symmetric and static .
the spherically symmetric solution of the vacuum einstein s equations is well known : it is the schwarzschild solution @xcite .
therefore , it is often stated that the schwarzschild solution ( in the schwarzschild coordinates ) describes gravitational field of a point particle .
this statement is incorrect because there is no @xmath0-type energy - momentum tensor corresponding to a particle on the right hand side of einstein s equations .
on the other hand , the solution of einstein s equations outside point massive particle must be isometric to the schwarzschild solution . therefore the natural question arises : `` where is the @xmath0-function ? '' .
the answer turned out to be unexpected : @xmath0-function corresponds to infinite value of the schwarzschild radial coordinate .
namely , we prove in this paper , that the schwarzschild solution in isotropic coordinates is the solution of einstein s equations in a topologically trivial space - time @xmath1 with @xmath0-type source .
the solution is understood in the generalized sense after integration with a test function . at the same time , the metric components are locally integrable functions for which the nonlinear einstein s equations are mathematically defined . note that the schwarzschild solution in schwarzschild coordinates is locally isometric to the schwarzschild solution in isotropic coordinates .
nevertheless global structure of space - time is totally different .
the maximal extension of space - time along geodesics for the schwarzschild solution in schwarzschild coordinates is the topologically nontrivial manifold which is equal to the topological product of the well known carter
penrose diagram with the sphere .
this maximally extended solution describes white and black holes and is not related to the gravitational field of a point particle . at the same time
the schwarzschild solution in isotropic coordinates corresponds to the topologically trivial space - time .
severe mathematical difficulties arise during solution of this problem .
a solution of einstein s equations must be understood in a generalized sense after integration with test functions because the @xmath0-function stands on the right hand side .
but there is no multiplication in the space of generalized functions ( distributions ) and the question arises what is the mathematical meaning of the left hand side of einstein s equations which are nonlinear . besides , locally nonintegrable functions arise during solution of equations , and some functionals must be attributed to them . in other words , regularization is needed . in the preset paper ,
the solution of einstein s equations is found in a generalized sense : the equations are satisfied after integration with test functions .
we choose the usual space @xmath2 of infinitely differentiable functions on @xmath3 with compact support as the space of test functions .
metric components for the obtained solution are locally integrable functions and therefore belong to the conjugate space @xmath4 .
though no multiplication is defined in @xmath5 , the left hand side of einstein s equations is well defined for the derived solution .
here we use the analytic regularization for exponential spherically symmetric functionals .
the obtained solution turned out to be the well known schwarzschild solution in isotropic coordinates .
it appears after gluing together two exterior solutions of a black hole along the horizon .
this solution is isometric to einstein
rosen bridge and is asymptotically flat not only at large distances but near the particle itself where the curvature of space - time tends also to zero ( gravitational asymptotic freedom ) . at large distances ,
the gravitational field is attractive . under the horizon attraction
changes to repulsion .
this repulsion results in geodesic completeness of space - time near the particle .
attempts to interpret the schwarzschild solution in terms of generalized functions were made earlier [ 26 ] .
papers [ 24 ] are related to our approach and discussed in some detail in section [ seldef ] . in reference
@xcite , the energy - momentum tensor of matter was taken to have spherically symmetric gaussian distribution around the origin of the coordinate system , and exact solution to einstein s equation is found . in the limit of zero distribution radius ,
the energy - momentum tensor is proportional to the @xmath6-function .
this energy - momentum tensor differs from that in our paper and does not correspond to a point particle .
another approach was adopted in @xcite .
roughly speaking , the schwarzschild sphere was shrank to a point .
the energy - momentum tensor was shown to be proportional to the @xmath6-function at this point in a sense of a distribution , though not all einstein s equations were actually solved .
again , the energy - momentum tensor is different from ours .
the authors of @xcite regularized the schwarzschild metric in schwarzschild coordinates .
afterwards they calculated the energy momentum tensor and take off the regularization .
the resulting energy - momentum tensor is proportional to the @xmath6-function located at the origin . in this case
, the energy - momentum tensor also differs from ours .
let us consider topologically trivial manifold @xmath7 ( space - time ) with cartesian coordinates @xmath8 , @xmath9 , and metric @xmath10 of lorentzian signature @xmath11 .
we denote the world line of a point particle by @xmath12 where @xmath13 is a parameter along the world line .
we use the following notations for geometric notions : @xmath14 where @xmath15 are christoffel s symbols , @xmath16 is the curvature tensor , @xmath17 is the ricci tensor , and @xmath18 is the scalar curvature . in general relativity ,
a point particle of mass @xmath19 is described by the following action @xmath20 where @xmath21 and @xmath22 .
variations of action ( [ eacwek ] ) with respect to metric components @xmath23 and coordinates of a particle @xmath24 yield einstein s equations of motion and equations for extremals ( geodesics ) : @xmath25 where @xmath26 is the particle energy - momentum tensor , @xmath27 , and @xmath28 is the three - dimensional @xmath0-function on sections @xmath29 .
we assume that for particle trajectory @xmath30 . to analyze system of equations ( [ eiepop])([extwel ] ) , we rewrite it in the hamiltonian form .
to this end , the adm parameterization of the metric is used @xmath31 where @xmath32 is the metric on space like sections of the space - time @xmath29 . in this parameterization , @xmath33 functions : the lapse function @xmath34 and shift functions @xmath35 , are introduced instead @xmath33 components @xmath36 and @xmath37 of the metric .
here we use notation @xmath38 where @xmath39 is the @xmath40 matrix which is inverse to @xmath32 : @xmath41 in what follows , raising of space indexes which are denoted by greek letters from the middle of the alphabet ( @xmath42 ) is performed using the inverse three - dimensional metric marked with a hat .
we assume that all sections @xmath29 are space like , and therefore the metric @xmath32 is negative definite .
let @xmath43 and @xmath44 be momenta conjugate to generalized coordinates @xmath32 and @xmath24 . the action for a point particle is invariant with respect to reparameterization of the world line . to simplify formulae
, we fix the gauge @xmath45
. then einstein s equations ( [ eiepop ] ) in the hamiltonian form reduce to constraint and dynamical equations .
constraints have the form @xmath46 where @xmath47 , @xmath48 , @xmath49 is the three dimensional scalar curvature for metric @xmath32 , and @xmath50 is the three dimensional covariant derivative . constraints ( [ edycgw ] ) and ( [ ekicop ] ) are called dynamical and kinematical , respectively .
dynamical hamiltonian einstein s equations are @xmath51 where @xmath52 is the three - dimensional laplace beltrami operator , @xmath53 , and @xmath54 .
hamiltonian equations for geodesic have the form @xmath55_{{\boldsymbol{x}}={\boldsymbol{q}}}.\end{aligned}\ ] ] transition from lagrangian equations of motion ( [ eiepop ] ) , ( [ extwel ] ) to hamiltonian ( [ edycgw])([eqmopg ] ) is complicated .
hamiltonian formulation of general relativity was given in @xcite .
for general relativity and a point particle it can be found i.e. in @xcite .
combined hamiltonian system of equations of motion for a point particle in general relativity was considered in @xcite .
now we are prepared to solve the full system of equations ( [ edycgw])([eqmopg ] ) for a point particle at rest .
we assume that the particle is located at the origin of the coordinate system : @xmath56 we are seeking spherically symmetric static solution of the system of equations of motion ( [ edycgw])([eqmopg ] ) . in this case
, the coordinate system can be chosen in such a way that shift functions are zero , @xmath57 .
we choose also the weyl flat gauge for the spatial metric @xmath58 where @xmath59 is the euclidean metric and @xmath60 . in the static case ,
all momenta vanish : @xmath61 and @xmath62 .
derivatives @xmath63 are also zero .
thus there are two unknown functions in this case which depend only on radius in spherical coordinates : @xmath64 and @xmath65 . if the above assumptions are fulfilled then equations ( [ ekicop ] ) , ( [ edefmk ] ) , and ( [ eqmoqf ] ) are identically satisfied
. equations ( [ edycgw ] ) , ( [ ecamod ] ) , and ( [ eqmopg ] ) take the form @xmath66 where we divided equation ( [ eqmopg ] ) by @xmath19 .
it is important that equation ( [ qsemby ] ) can not be divided by @xmath67 , and indices @xmath68 can not be lowered because metric have singularity at @xmath69 , and the ricci tensor and scalar curvature contain the @xmath0-function .
equation ( [ qdytwv ] ) is the main equation which is to be solved .
the three - dimensional volume element @xmath67 on the left hand side is needed for covariance because @xmath0-function is the scalar density with respect to coordinate transformations .
this equation is the simplest covariant second order differential equation for metric with the @xmath0-function on the right hand side . in two dimensions in weyl flat gauge
, it reduces to the poisson equation and yields the fundamental solution for two - dimensional laplace operator . in higher dimensions it becomes nonlinear , and
its solution is not known up to now .
we believe that this equation may have its own applications in differential geometry which are not discussed here .
we are seeking spherically symmetric solutions of equation ( [ qdytwv ] ) , and that is in agreement with the symmetry of the right hand side .
any spherically symmetric metric is weyl flat , which means that there exists the coordinate system where the metric has the form ( [ qfsdok ] )
. then main equation ( [ qdytwv ] ) for the weyl flat spherically symmetric metric takes the form @xmath70 where @xmath71 is the flat laplace operator and we introduced notation @xmath72 . equation ( [ qfilht ] )
is nonlinear because of the second term on the left hand side .
solution of this equation should be understood in the generalized sense after integration with a test function ( see , i.e.@xcite ) because there is the @xmath0-function on the right hand side . if we look for a solution in the space of functionals ( generalized functions ) @xmath4 then two serious problems arise . since the equation is nonlinear
, the generalized functions must be multiplied , but multiplication in @xmath5 is absent .
second , we shall see in what follows that some terms in equation ( [ qfilht ] ) are not locally integrable , and hence a regularization of the appearing integrals is necessary . though multiplication of functionals in @xmath5 is not defined in general , for some particular functionals , i.e.sufficiently smooth , the left hand side has nevertheless definite meaning .
we look for spherically symmetric solution of equation ( [ qfilht ] ) in the spherical coordinate system @xmath73 .
[ toyfrd ] the function @xmath74 satisfies equation ( [ qfilht ] ) which solution is understood in the generalized sense after integration with a test function .
we use the analytic regularization for the generalized function @xmath75 , @xmath76 @xcite . for @xmath77
the function @xmath75 is locally integrable , and the functional @xmath78 is defined for all test functions @xmath79 ( smooth functions with compact support ) .
this functional is analytically continued on the whole complex plane @xmath76 except simple poles located on the real line @xmath80 @xcite .
the poles does not matter because we do not fall into them in the considered case .
consider the functional @xmath81 depending on parameter @xmath82 .
simple calculations show that the following identity holds @xmath83 where prime denotes differentiation with respect to the radius .
it is clearly well defined for @xmath84 .
this identity can be analytically continued to @xmath85 because we do not fall into the poles described above . as a result ,
we get the following expression for the second term on the left hand side of equation ( [ qfilht ] ) @xmath86 this equality is understood in the generalized sense after integration with test functions .
therefore equation ( [ qfilht ] ) for the generalized function ( [ qsolco ] ) takes the form @xmath87 the function @xmath88 is the unique fundamental solution of the laplace equation decreasing at infinity ( see .
, for example , @xcite ) : @xmath89 the function @xmath90 is not locally integrable and requires regularization .
we understand it as analytic continuation of the functional @xmath75 to @xmath91 , which is mentioned at the beginning .
the equality @xmath92 is simply verified for @xmath93 .
it remains valid also after analytic continuation to the point @xmath94 . for @xmath94 , it looks as follows @xmath95 thus , for analytic regularization , we attribute to the locally nonintegrable function @xmath96 the functional @xmath97 the last functional is well defined .
we see that the nonlinear term on the left hand side of the equation ( [ qfilht ] ) is canceled after integration with a test function @xmath98 if the analytic regularization is used for the function @xmath90 .
equation ( [ qdytwv ] ) for a massive point particle was considered in [ 24 ] for weyl euclidean form of the metric .
the authors used parameterization @xmath99 and obtained nonlinear equation @xmath100 which is equivalent to equation ( [ qfilht ] ) .
they proposed the following solution @xmath101 differing from that corresponding to solution ( [ qsolco ] ) by the essential factor @xmath102 .
this factor is obtained from the regularized ( smeared ) @xmath0-function and depends on small parameter @xmath103 , @xmath104 which is the `` radius '' of the regularized @xmath0-function . in the limit @xmath105 ,
solution ( [ qrqwsd ] ) becomes trivial , @xmath106 , and satisfies equation @xmath107 corresponding to a vacuum rather than to a point particle .
so the solution proposed in [ 24 ] does not depend on mass @xmath19 .
the generalization of equation ( [ qteral ] ) for a charged massive particle interacting with electromagnetic field was also considered . in this case , the factor @xmath102 becomes nontrivial , and the total mass of a particle is proportional to its charge .
this effect is interpreted as the regularization of the self - energy of a point charge by gravitational interaction . to show the difference in the approaches ,
let us consider equation ( [ qteral ] ) for @xmath108 which is different from ( [ qrqwsd ] ) .
substitution of this solution into equation ( [ qteral ] ) yields @xmath109 expression @xmath110 is not defined .
we define it as follows .
relation @xmath111 can be easily verified for @xmath112 .
we analytically continue it to the point @xmath113 where it can be rewritten as @xmath114 so , we define @xmath115 for analytically regularized @xmath116 the right hand side is zero , and we get relation @xmath117 which is valid in a generalized sense .
in other words , the relation @xmath118 which is valid for @xmath119 is analytically continued to the point @xmath113 .
we see that solution ( [ qsolco ] ) can be obtained either by solving equation ( [ qfilht ] ) or ( [ qteral ] ) . in both cases , we used the analytic regularization and get the same solution . our solution ( [ qsolco ] ) does depend on mass @xmath19 .
it is nontrivial for neutral particle and differs essentially from solution ( [ qrqwsd ] ) considered in [ 24 ] .
thus we have found the generalized solution of equation ( [ qfilht ] ) in the space of functionals @xmath4 .
for this solution the left hand side of the main equation is well defined .
unfortunately , we did not describe explicitly that subspace in @xmath4 , for which the left hand side of the main equation is defined in general .
this question is complicated and related to the uniqueness of the obtained solution . we leave it for future work
we are left with equations ( [ qsemby ] ) and ( [ qlpojg ] ) which are to be solved . as for the case of equation ( [ qdytwv ] )
, we can not explicitly describe the subspace in @xmath4 for which the nonlinear part of equation ( [ qsemby ] ) is defined .
instead , we write down the solution in explicit form and check that all terms are really defined for it .
let three dimensional metric has the form ( [ qfsdok ] ) where the conformal factor is given by equation ( [ qsolco ] ) . then lapse function @xmath120 satisfies equation ( [ qsemby ] ) in @xmath3 in a generalized sense .
the lapse function is positive for @xmath121 and negative for @xmath122 .
it has finite limit at zero , @xmath123 and is infinitely differentiable for @xmath124 .
therefore the lapse function is locally integrable and lies in @xmath4 .
one can easily verify that the laplacian of the lapse function for @xmath124 is equal to zero , @xmath125 function @xmath126 is continuous , and its derivatives have discontinuities at zero .
since @xmath127 then the weak derivatives are defined everywhere , and equality @xmath128 is fulfilled in a generalized sense in the whole @xmath3 because the lapse function has finite limit at zero .
this equality can be multiplied by @xmath129 because near zero @xmath130 thus the first term on the left hand side of equation ( [ qsemby ] ) is equal to zero in the generalized sense in the whole @xmath3 .
one can verify that for @xmath124 the following equality holds @xmath131 hence we are left only with @xmath0-functions which are contained in the ricci tensor and scalar curvature .
they do not cancel , and equation ( [ qsemby ] ) takes the form @xmath132 this equation is fulfilled because near zero @xmath133 thus equation ( [ qsemby ] ) is fulfilled .
we are left only with equation ( [ qlpojg ] ) for the point particle . since @xmath134 it is not satisfied . at this point , we invoke physical arguments .
the right hand side of equation ( [ eqmopg ] ) is the force acting on the particle from the gravitational field which is produced by the particle itself .
therefore equation ( [ eqmopg ] ) is understood after averaging over a sphere surrounding a particle .
then it is fulfilled because of the spherical symmetry . in other words ,
a particle does not move under the action of its own gravitational field .
the same situation happens in newton s gravity and electrodynamics .
for example , for gravitational potential @xmath135 in newton s theory , the force acting on a particle from its own gravitational field is equal to @xmath136 and equations of motion are not satisfied .
nevertheless they are fulfilled after averaging over a sphere .
taking together the laps function ( [ qshuyr ] ) and the conformal factor for the three - dimensional space part of metric ( [ qsolco ] ) we obtain the metric @xmath137.\ ] ] it is the well known schwarzschild metric in isotropic coordinates .
isotropic coordinates for the schwarzschild metric are well known for a long time , see , i.e. @xcite .
the new result is not the schwarzschild metric in isotropic coordinates ( [ qschre ] ) , but the statement that it satisfies equations ( [ qdytwv ] ) and ( [ qsemby ] ) with @xmath0-type source .
this fact has principal meaning and far reaching physical consequences .
we shall see that gravitational attraction to the point mass @xmath19 at large distances is replaced by repulsion at distances @xmath138 . for @xmath139 ,
metric ( [ qschre ] ) is defined everywhere in @xmath1 except the world line of the origin of the spherical coordinate system : @xmath140 let us compare the schwarzschild metric in isotropic coordinates ( [ qschre ] ) and the schwarzschild metric in schwarzschild coordinates @xcite @xmath141 the schwarzschild radial coordinate is denoted here by @xmath142 .
the last metric is defined for @xmath143 ( outside the horizon @xmath144 ) and for @xmath145 ( under the horizon ) .
it is asymptotically flat and tends to the lorentz metric when @xmath146 .
it was obtained without @xmath0-function source , and the constant @xmath19 appears in it as the integration constant . from mathematical point of view , it may take arbitrary values @xmath147 .
however , if we assume that the schwarzschild solution describes the gravitational field outside a point particle , then comparison with the newton gravitational law at large distances ( see , i.e. @xcite ) tells us that the integration constant @xmath19 is the mass of a particle and therefore must be positive .
the transformation of radial coordinate which brings metric ( [ qschrt ] ) into the form ( [ qschre ] ) is well known @xmath148 it is shown in fig.[fcotra ] . when radius @xmath149 decreases from infinity to the critical value @xmath150 , the schwarzschild radial coordinate decreases from @xmath151 to the horizon @xmath152 .
afterwards the radius @xmath142 increases from @xmath153 to @xmath151 as the radius @xmath149 decreases from @xmath154 to zero .
thus two copies of the schwarzschild metric outside horizon @xmath143 are mapped on two distinct domains in @xmath3 : @xmath155 and @xmath156 . on the critical sphere
@xmath157 they are smoothly glued together .
the spatial part of the metric is not degenerate here . note that area of a sphere surrounding a particle tends to infinity as it gets closer to it .
this is related to the fact that components of the spatial part of the metric ( [ qschre ] ) diverge for @xmath158 .
one can easily calculate the asymptotic of the zeroth component of the metric ( [ qschre ] ) @xmath159 it is the same as for the schwarzschild metric in the schwarzschild coordinates .
this is not surprising because @xmath160 for @xmath161 .
it means that there is no problem with newton s gravitational law for metric ( [ qschre ] ) because it is defined by the asymptotic at @xmath161 .
we see that point massive particle is located at that site where the schwarzschild radial coordinate equals infinity .
this phenomena may be called gravitational asymptotic freedom because near the particle , @xmath162 the space - time becomes asymptotically flat .
the determinant of metric ( [ qschre ] ) is equal to @xmath163 consequently , the metric is degenerate on the sphere of radius @xmath164 which corresponds to the horizon @xmath144 for the schwarzschild solution and on the @xmath165-axis @xmath166 .
degeneracy on the @xmath165-axis is related to the spherical coordinate system and location of a point particle at the origin of the coordinate system for @xmath167 .
it is well known that all geometric invariants for the schwarzschild metric constructed from the curvature tensor have no singularities on the horizon .
thus , we have no curvature singularities at the critical sphere @xmath168 .
nevertheless , there is a problem . to describe global structure of the obtained space - time we have to analyze behaviour of extremals ( geodesics ) for schwarzschild solution in isotropic coordinates which describe motion of probe particles .
the behavior of geodesics in schwarzschild coordinates is well known and can be recalculated in isotropic coordinates .
therefore we mention here only some points of their qualitative behavior .
the space - time at @xmath69 ( near the particle ) and at @xmath169 ( space infinity ) is geodesically complete , because both regions correspond to infinite value of the schwarzschild radial coordinate @xmath170 . at critical value of the radius
@xmath157 corresponding to the schwarzschild radius @xmath153 geodesics are well known to be incomplete .
it means that at @xmath157 the space - time in isotropic coordinates is geodesically incomplete , and we have to extend it .
the best way for isotropic coordinates is to identify geodesics from outside and inside the critical sphere @xmath154 .
afterwards we would get geodesically complete space - time .
the problem is how to do this in a mathematically consistent way .
this is a hard problem , and we hope to address it in the future .
another particular property of isotropic coordinates is that gravitational attraction at large distances changes to repulsion .
it is clear because in schwarzschild coordinates , probe particles always have acceleration directed towards horizon .
it means that probe particles located inside the critical sphere move outside the center .
thus the gravitational attraction outside @xmath154 changes to repulsion in the region @xmath155 .
as far as we know this is a novel feature of the obtained solution .
in the present paper , we prove that the schwarzschild solution in isotropic coordinates satisfies the system of einstein s and geodesic equations for a point massive particle .
nontrivial energy - momentum tensor appears on the right hand side of einstein s equations which is proportional to three - dimensional @xmath0-function .
rewriting the schwarzschild solution in isotropic coordinates is by itself a student s problem .
the result of the paper is that we attribute mathematical meaning for einstein s equations with @xmath0-type source and provide explicit solution .
the mass @xmath19 appears in the schwarzschild solution as an integration constant and may be arbitrary . in our approach
, the mass @xmath19 of a point particle enters the action from the very beginning .
the obtained solution is isometric to einstein
rosen bridge @xcite . in the original paper ,
the bridge was attributed to an elementary particle which in our notations corresponds to the critical sphere @xmath157 .
the space - time is described by two sheets .
we have shown that the particle described by a @xmath0-function is located not at @xmath157 , but at geodesically complete `` infinity '' of one of the sheets corresponding to @xmath69 .
geodesically complete infinity of the second sheet lies at infinity @xmath169 and corresponds to asymptotically flat space - time .
both sheets are glued together at the critical sphere @xmath157 where metric becomes degenerate and gravitational attraction at large distances is replaced by repulsion at small distances .
this effect is of primary importance and makes us to reconsider our approach to many gravitational phenomena . at present , the einstein
rosen bridge has another interpretation .
two sheets are considered as two different universes which are connected by a worm hole at @xmath157 .
it was noted recently @xcite that nontrivial energy - momentum tensor appears on the right hand side of einstein s equations at @xmath154 . in our approach ,
the right hand side of einstein s equations is equal to zero at @xmath157 .
the difference comes from choosing the solution to equation @xmath171 . for a given metric we chose solution ( [ qshuyr ] ) which is positive for @xmath172 and negative at @xmath173 .
the other solution @xmath174 is chosen in @xcite , the modulus sign resulting in the appearance of the singularity on the right hand side of einstein s equations at @xmath157 .
the obtained space - time is topologically four - dimensional euclidean space with removed world line of the particle @xmath69 .
this space - time is geodesically complete at @xmath161 as well as at @xmath175 . at the critical sphere @xmath154 geodesics are incomplete .
the problem of how to continue them remains open .
the effect of the transformation of gravitational attraction into repulsion is the straightforward consequence of einstein s equations and leads to many questions .
for example , how can a black hole be formed if particles can not come close one to another ?
one can make a lot of speculations at this point but we do not consider them here .
the author is grateful to i. v. volovich , a. k. gushchin , yu . n. drozhzhinov , , and v. p. mikhailov for discussions and valuable comments .
the work is supported by rffi ( grants 11 - 01 - 00828-a and 13 - 01 - 12424-ofi - m ) , the program for supporting leading scientific schools ( grant nsh-2928.2012.1 ) , and the program `` modern problems of theoretical mathematics '' by ras .
r. arnowitt , s. deser , and c. w. misner .
the dynamics of general general relativity . in _ gravitation : an introduction to current research _ , ed .
l. witten .
john wiley & sons , inc .
london , 1962 , gr - qc/0405109 . | it is well known that the schwarzschild solution describes the gravitational field outside compact spherically symmetric mass distribution in general relativity . in particular
, it describes the gravitational field outside a point particle .
nevertheless , what is the exact solution of einstein s equations with @xmath0-type source corresponding to a point particle is not known . in the present paper , we prove that the schwarzschild solution in isotropic coordinates is the asymptotically flat static spherically symmetric solution of einstein s equations with @xmath0-type energy - momentum tensor corresponding to a point particle .
solution of einstein s equations is understood in the generalized sense after integration with a test function .
metric components are locally integrable functions for which nonlinear einstein s equations are mathematically defined .
the schwarzschild solution in isotropic coordinates is locally isometric to the schwarzschild solution in schwarzschild coordinates but differs essentially globally .
it is topologically trivial neglecting the world line of a point particle
. gravity attraction at large distances is replaced by repulsion at the particle neighborhood . |
one of the implications from the data collected by the iras mission , combined with extensive radio surveys , is a nearly universal and tight correlation between the far - infrared thermal dust emission and the optically - thin total radio continuum emission ( mixing thermal and synchrotron components ) in normal star - forming galaxies . @xcite
have reached this conclusion comparing 60@xmath1 m fluxes with 6.3 cm fluxes from the effelsberg dish , @xcite comparing 40120@xmath1 m fluxes with 21 cm fluxes from westerbork .
although the correlation is generally expressed between luminosities , it holds when they are normalized to cancel size and distance effects , and is valid for a very broad range of star formation rate densities ( review by @xcite ) . at centimeter wavelengths ,
the radio emission is dominated by non - thermal processes .
all models accounting for the constancy of the infrared to radio flux ratios invoke the star formation process as the indirect source of both types of emission , via a production rate of cosmic rays by type - ii supernov proportional to the production rate of heating photons .
additional coupling mechanisms are needed , such as between the interstellar medium density and the magnetic field intensity @xcite , or between the star formation rate and the magnetic field amplification @xcite .
various physical mechanisms can lead to these couplings , among which differential rotation , or turbulence driven by massive star formation amplifying the magnetic field .
in addition , given that the various timescales involved in the underlying mechanisms are very different ( heating of dust by massive and then intermediate - mass stars , explosion of supernov , acceleration , diffusion and decay of the relativistic electrons , variations of the magnetic field strength ) , some variation in infrared to radio flux ratios with the star formation recent history is to be expected
. there are naturally known exceptions to the general correlation .
a fraction of galaxies hosting non - stellar nuclear activity have excess radio continuum emission , from cosmic rays accelerated in jets from the nucleus @xcite .
star - forming galaxies in cluster environments can also have radio excess @xcite , perhaps triggered by external magnetic field compression due to the intracluster gas pressure or to interactions with neighbor galaxies .
two pairs of colliding galaxies with excess synchrotron emission originating from a bridge between the disks were studied by @xcite and @xcite , to illustrate the importance to the correlation of regulated cosmic ray escape .
@xcite also discovered excess synchrotron in an interacting galaxy , due to external compression of the magnetic field .
some statistical studies found high far - infrared to radio ratios in compact galaxy groups @xcite and in lenticular galaxies @xcite , but may suffer from photometric errors ( confusion problem for iras fluxes in galaxy groups , as pointed out by @xcite , and extended radio emission missed by interferometric observations ) . for dwarf galaxies
, metallicity effects can lower the dust emission , and the radio spectra of many are dominated by thermal emission .
@xcite found that the far - infrared to radio ratio tends to be lower in blue compact galaxies than in spirals , while @xcite found similar ratios in both .
using the radio fluxes of @xcite and new iras fluxes measured with _
@xcite , it seems in fact that there is a continuum of infrared to radio ratios from very low to very high values , i.e. with a much larger dispersion than in spirals , perhaps depending on two competing effects : reduction of the dust emission by metal deficiency , and reduction of the synchrotron emission by a lack of supernov or by fast escape of cosmic rays . variations also exist inside spiral galaxies : @xcite found higher 60@xmath1 m to 11 cm flux ratios in the central regions of large angular size spirals than in their disks , following the @xmath2 ratio , an average dust temperature indicator .
this is likely because the scale length of the synchrotron emission is larger than in the infrared , due to cosmic ray propagation far from the site of their acceleration , located in star formation regions hosting hot dust @xcite . in this paper
, we study three nearby galaxies ( an amorphous galaxy , an early - type spiral and a late - type spiral ) which are strongly deficient in radio emission with respect to their dust emission , trying to identify the cause of their peculiarity . in normal galaxies ,
the synchrotron radiation comes mainly from electrons having left the supernova remnants which have accelerated them , and propagating in the interstellar magnetic field @xcite .
a lack of synchrotron emission may thus have its origin in either an unusually low strength of the magnetic field , or a reduced supernova rate in the recent past . with non - detections in the synchrotron emission ,
the magnetic field strength is extremely difficult to estimate .
however , we address the star formation properties using an array of new observations , and thereby approach indirectly the issue of magnetic fields .
we will argue that these galaxies have a high star formation rate associated with a low supernova rate , reflecting a starburst in a very early stage occurring after a long period of quiescence , since supernov begin to explode a few myr after the birth of their progenitors .
if this interpretation is valid , these galaxies could provide the opportunity to observe the properties of the earliest stages of a starburst , and to understand better the regulation of the galactic infrared and radio emission .
although such galaxies are apparently quite rare , they may exist in greater numbers at more primitive epochs of star formation .
they must therefore be taken into account when using high - redshift radio surveys to study the star formation history , and when using high - resolution radio observations to identify faint infrared or submillimeter sources .
the paper is structured as follows : in section [ review ] we discuss existing data and assemble comprehensive information about ngc1377 , ic1953 and ngc4491 ; section [ effelsberg ] presents and analyses new radio continuum observations ; in section [ mid_ir ] we describe and exploit near - infrared spectroscopic observations , extract estimates of the star formation rate from the dust emission and discuss the excitation and apparent cooling rate of the interstellar medium ; section [ sest ] examines results from single - dish molecular gas observations in ngc1377 ; section [ n4418 ] is a brief discussion of some common properties of ngc4418 with the three galaxies studied above ; section [ scenarios ] lists and discusses the various physical scenarios which could account for the observed synchrotron deficiency .
these scenarios are related either to a very low cosmic ray density stemming from a low supernova rate , or to a very low magnetic field intensity ; we conclude in favor of the former , as would be expected in a nascent starburst .
section [ stat_highq ] then compares the statistics of synchrotron - deficient galaxies to those of nascent starbursts , and discusses possible triggering mechanisms .
section [ summary ] contains a summary of our conclusions .
general information about the three galaxies under consideration can be found in table [ tab_gen ] .
ngc1377 and ic1953 were identified as radio synchrotron - deficient galaxies among the iras bright galaxy sample mapped with the vla @xcite .
ngc4491 , which is fainter , was also found to be synchrotron - deficient in a search among the small sample of @xcite , based on mid - infrared colors ( see sect .
[ irsed ] ) .
all three galaxies are low - mass and quite unremarkable at first sight , if it were not for their high dust temperatures .
being both faint and hot in their dust emission , they are very peculiar with respect to the average local @xmath3@xmath2 function , especially ngc1377 @xcite .
the average @xmath2 ratio at their infrared luminosities is only @xmath4 , and they deviate from this by about @xmath5 and @xmath6 .
although ic1953 seems to have a very close companion ( fig . [ images ] ) , this galaxy is at much higher redshift ( see section [ envir ] ) .
ngc1377 is classified as a lenticular or amorphous galaxy , and does not present any apparent disturbance .
its only peculiarity , noted by @xcite , is a central dust lane along the southern part of the minor axis , visible in the b - band isophotes .
ngc4491 is a barred dwarf sa galaxy @xcite whose shape is also regular , with a tenuous outer disk .
ic1953 is a strongly barred sd galaxy ( the bar is short but straight and narrow , connecting at right angles to the spiral arms ) .
it has an unsually bright and well - defined core for its late type ( very prominent in the mid - infrared dust emission ) .
one of the arms winds to join the opposite end of the bar , and the other arm , winding for more than 360 , offers the aspect of three distinct branches .
the mid - infrared emission of ngc1377 and ngc4491 , in the isocam maps at 7@xmath1 m and 15@xmath1 m , is dominated by the central regions , of apparent sizes respectively about 2.3kpc and 800pc , and the dust emission of ic1953 is also more concentrated ( in a 1.5kpc region ) than in other galaxies of its late - type morphology ( fig .
[ images ] ) . decomposing the surface brightness profile of ngc1377 at 7@xmath1 m
, we find an unresolved source ( at the resolution of isocam , with a point spread function of @xmath7 ) contributing about 60% of the total flux or more .
for ngc4491 , the central region is also unresolved in the mid - infrared , and superimposed onto a very diffuse component .
the 7@xmath1 m brightness profile decomposition yields a point source accounting for about 65% of the flux of the infrared core , and 38% of the total flux .
an unresolved source is also seen in the 7@xmath1 m brightness profile of ic1953 , containing about 65% of the flux of the circumnuclear regions .
ngc1377 and ic1953 are both members of the eridanus galaxy group . according to @xcite ,
this is a small group ( 54 members brighter than @xmath8 ) in a young dynamical state , since it has an irregular and subclustered aspect , is rich in spiral galaxies , and forms with the fornax cluster a bound system which is not yet virialized . in the soft x - ray maps of @xcite ( 0.10.9kev ) , eridanus is not detected .
ngc1377 and ic1953 do not belong to any subcondensation , but to the same loose sub - group lying in projection in the center , and are close to each other both spatially and in radial velocity .
ngc4491 seems also to be part of a dynamically young system , the core named group a in the virgo cluster , containing m87 and 100 to 300 members , and which is described by @xcite as a `` turbulent core with a long tail of low velocities '' .
ngc4491 belongs to this tail , with a large velocity with respect to the systemic velocity of @xmath9kms@xmath10 .
none of the three galaxies shows any morphological disturbance , but they have likely interacted with the ambiant medium , since they are deficient in atomic gas .
we did not find any published hi mass measurement for ngc1377 , but a spectrum from the hipass survey , conducted with the parkes telescope @xcite .
this spectrum has a sensitivity of about 10mjy per spectral channel , and implies that m(hi@xmath11m@xmath12 .
compared with the s0/a - sab galaxies in @xcite , the logarithmic hi deficiency of ngc1377 , normalized by the dispersion in the field reference sample , is higher than 1.0 .
this value is @xmath13 for ic1953 and @xmath14 for ngc4491 @xcite , and a galaxy can be considered hi - deficient when @xmath15 .
ic1953 has two small - mass neighbors , eso548-g040 , a very low - surface brightness galaxy at about 2.6 arcmin in projected distance to the nw , and eso548-g036 , at about 6 arcmin to the se .
eso548-g040 has a much higher receding velocity ( 4061kms@xmath10 ) than the members of the eridanus group ( @xmath16kms@xmath10 ) , and is therefore most likely a background galaxy . on the contrary
, the velocity of eso548-g036 ( 1480kms@xmath10 ) identifies it as a true neighbor of ic1953 , with a velocity difference of 400kms@xmath10 and a physical distance @xmath17kpc .
it is thus tempting to consider eso548-g036 a candidate for tidal interaction with ic1953 , which could be responsible for the peculiar arm structure of ic1953 .
the closest neighbor to ngc4491 is the blue compact dwarf ic3446 at 6.3 arcmin to the w , and the closest galaxy of comparable mass is ngc4497 , at about 12 arcmin to the nw .
their velocities are 760 and 610kms@xmath10 larger than that of ngc4491 . in order to build a reference for far - infrared to radio flux ratios , obtained with homogeneous iras and vla data , we chose the sample of 69 nearby spiral galaxies studied by @xcite , with mid - infrared and other data already at hand , and which comprises two of the three galaxies examined here , ngc4491 and ic1953 .
we measured iras fluxes using the _ scanpi _ tool @xcite with default parameters . only for ngc4491 at 100@xmath1 m was the baseline fitted over a 30 range replaced by a baseline fitted over the central 14 only , where it is reasonably flat .
given the brightness of the three target galaxies , we used as a flux estimator for them the template amplitude , except for ic1953 which is slightly extended at 25 and 60@xmath1 m ( we used the flux integration of the scan between the first zero crossings ) .
radio continuum fluxes or upper limits at 21 cm were measured from nvss maps @xcite , with a beam fwhm of 45 and a detection limit of the order of 1mjy .
we obtain an upper limit for ngc1377 which is equal to that given by the more sensitive survey of @xcite in a 21-beam .
@xcite defined a logarithmic ratio of far - infrared to radio flux densities as : + @xmath18 $ ] , where all flux densities , @xmath19 , @xmath20 and @xmath21(21 cm ) , are expressed in the same unit . in the reference sample , excluding radio upper and lower limits , we find an average ratio @xmath22 .
the dispersion is the same as that obtained by @xcite for the iras bright galaxy sample .
the relationship between far - infrared and 21 cm fluxes , both normalized by the flux or the flux density in the blue band , is shown in fig .
[ rad_fir_b ] .
five galaxies turn out to have far - infrared to radio flux ratios @xmath23 : ngc1377 ( @xmath24 ) , ngc4491 ( @xmath25 ) and ic1953 ( @xmath26 ) , with far - infrared to blue luminosity ratios higher than 1.5 ; ngc4580 ( @xmath27 ) and ngc4394 ( @xmath26 ) , two virgo galaxies having @xmath28 . the latter two
are most clearly separated from the first three by their average dust temperature ( fig .
[ q_temp ] ) : they have @xmath2 of respectively 0.26 and 0.23 , and also low mid - infrared colors @xmath29 of 0.9 to 1 , whereas ngc4491 and ic1953 have @xmath30 , and the flux density spectrum of ngc1377 peaks at 60@xmath1 m . the lack of radio synchrotron emission can be accounted for easily in the cool - dust galaxies .
if their star formation rate is presently very low , as indicated by their dust emission properties ( low colors and low surface brightnesses ) , then the far - infrared - emitting dust can be expected to be heated mainly by low - mass stars , well in excess of the emission produced by the star formation activity .
after the production of cosmic rays has ceased , the timescale for synchrotron emission decay is less than @xmath31years @xcite , so that the @xmath32 ratio should increase on this timescale after an exhaustion of the star formation .
the whole disk of ngc4394 is extremely diffuse in h@xmath33 @xcite and in the mid - infrared as well @xcite .
this is true also of ngc4580 outside its inner pseudo - ring .
however , the high @xmath2 ratios of ngc1377 , ngc4491 and ic1953 exclude such an explanation for their lack of synchrotron emission . the correlation between far - infrared and radio continuum luminosities
is known to be non - linear , the radio emission rising faster or decreasing faster than the dust emission @xcite .
faint galaxies such as those studied here are thus expected to have higher @xmath32 ratios than brighter galaxies .
however , @xcite found evidence of a small decrease of @xmath34(20 cm ) when the average dust temperature @xmath2 rises , along with an increase of the radio brightness temperature . to linearize and tighten the relationship between infrared and radio fluxes , @xcite apply an empirical correction which assumes that the radio emission is a good measure of the star formation rate , and decreases the @xmath32 ratio in proportion to @xmath35(20 cm ) , the blue to radio flux ratio .
although they find that variations in @xmath2 can not account for the dispersion in @xmath32 , @xmath2 is correlated with @xmath36 , and the trend of decreasing @xmath32 with increasing @xmath2 exists in our sample also ( particularly illustrated in fig .
[ q_temp ] by the behavior of ngc4580 and ngc4394 ) .
ngc1377 , ngc4491 and ic1953 manifestly break this trend .
if we apply the correction of @xcite to our data , as shown in fig .
[ q_qb ] , then the peculiarity of ngc1377 , ngc4491 and ic1953 with respect to the other galaxies is not destroyed : they still lie at @xmath37 , @xmath38 and @xmath39 above the new mean @xmath40 .
the only other galaxy having @xmath41 is ngc1022 ( @xmath42 ) , an amorphous starburst with @xmath43 .
we note that the correction proposed by @xcite , based on the blue to infrared luminosity ratio , can not be appropriate , since @xmath32 as a function of @xmath44 shows no defined trend .
the high @xmath2 and low @xmath45 ratios signal the existence of a dust component with high temperatures and thus high heating intensities , which could be due to an extensive star formation burst or non - stellar activity , or both @xcite . in this respect , the spectral indices between 25 and 60@xmath1 m of ngc1377 and ngc4491 , -1.5 and -2.1 , are ambiguous @xcite .
the broadband spectral energy distributions of the three galaxies are shown in fig .
j , h and ks - band fluxes are derived from images built for the 2mass large galaxy atlas @xcite , with a preliminary photometric calibration valid for the cit system .
the u , v and i - band fluxes of ngc1377 come from the colors given by @xcite .
other optical data were found in the ned database .
we measured the 45@xmath1 m fluxes of ngc1377 and ic1953 in isocam maps .
it appears that the dust emission spectrum of ngc1377 can not be reproduced by any empirical sed of @xcite ( fig .
[ sed ] ) , ajusted to a sample of normal galaxies , and parametrized by the @xmath2 ratio ( note however that the color bin of ngc1377 comprises only two galaxies ) .
the emission from transiently - heated dust particles , from 7 to 25@xmath1 m , is well in excess of what is expected for the large grain temperature ( as estimated by the @xmath2 ratio ) , by up to 8@xmath46 .
@xmath47 is expected to decrease steadily with rising dust temperature , and then starts to increase again at the highest @xmath2 ratios , but ngc1377 overdevelops this trend ( fig .
[ mirsfir ] ) . for @xmath48 , which in normal galaxies shows less dispersion than @xmath49 , the same remark applies .
the mid - infrared spectral shape between 2.5 and 11@xmath1 m , observed with isophot in @xmath50 @xcite , is quite unusual : instead of the typical spectrum of aromatic bands observed in most star - forming galaxies @xcite , a broad asymmetric feature between 6 and 8.5@xmath1 m contributes the bulk of the emission .
this is reminiscent of some ultraluminous galaxies @xcite .
such a broad feature , accompanied by very weak 6.2 and 11.3@xmath1 m bands , is also observed in the peak e of the smc hii region n66 by @xcite , in proto - planetary nebul@xcite , and bears some resemblance with the signature of anthracite , used by @xcite to reproduce proto - planetary nebula spectra .
all this suggests that the aromatic band carriers have not been processed in the same way as in normal star - forming galaxies , and that the dust responsible for the emission is subjected to hard radiation .
since it contributes the same part in the infrared energy budget as in quiescent galaxies , this dust must be more resilient in order not to be destroyed .
the abrupt fall - off of the spectrum from 9 to 11@xmath1 m further suggests that the absorption by silicates may be substantial , but such an interpretation is very hazardous , because optically - thin spectra of normal galaxies also have a minimum in this region .
dust heating can not be provided by a shock , in particular because one would then observe either a prominent silicate emission feature , or a negligible mid - infrared emission with respect to the far - infrared emission , depending on the gas density @xcite .
@xcite , on the basis of this spectrum , propose that ngc1377 contains a buried seyfert nucleus . however , several arguments suggest otherwise : + * 1 ) * the continuum emission at 34@xmath1 m and 911@xmath1 m is near zero , contrary to what is observed in seyfert nuclei . the short - wavelength continuum , in particular , which is not affected as much by extinction , is now routinely used as a diagnostic of the contribution of an active nucleus to the energy budget @xcite . among the sample of @xcite ,
the spectrum of ngc1377 resembles much more that of iras02530 + 0211 , a galaxy of starburst type , than that of any of the seyfert galaxies .
one could argue that the broad feature seen between 6 and 8.5@xmath1 m is similar to that seen in centaurusa @xcite , and is produced by a continuum strongly absorbed longward of 8@xmath1 m by silicates ; however , the similarity is superficial , since the spectra of centaurusa and other seyfert nuclei rise steeply longward of 10@xmath1 m and are brighter at 11.5@xmath1 m than at 8@xmath1 m , whereas the 11.5@xmath1 m flux of ngc1377 is near zero .
this is a strong argument against pure silicate absorption , because the red wing of the absorption band would otherwise be exceptionally wide . trying a quantitative fit of the 8@xmath1 m to 100@xmath1 m data , with the extinction law of @xcite (
some variants exist , but introduce differences essentially shortward of 8@xmath1 m , which is why we exclude short wavelengths ) , @xmath51 is required in order to reproduce the 810@xmath1 m isophot spectrum with an absorbed dust continuum , but no good fit can be performed : the solution is too high in the 1011.5@xmath1 m range ; one can obtain a simultaneous fit of the 12@xmath1 m , 25@xmath1 m , 60@xmath1 m and 100@xmath1 m data with the sum of two blackbodies of 370k and 100k , but the fit at 15@xmath1 m is then much higher than the observed flux , which is impossible to fix by adding a new dust component . we thus conclude that the broad feature between 6 and 8.5@xmath1 m may contain a genuine emission band , and may not be exclusively caused by absorption is adopted , then the emission of hot small grains up to @xmath52@xmath1 m would have to be boosted by a factor five to obtain the intrinsic small grain luminosity . since the complementary power observed in the far - infrared is only 1.2 times the observed power of small grains , at least @xmath53 would have to be re - emitted at longer wavelengths , in the submillimeter , which is inconceivable . and this would only exacerbate the difficulty to account for the dust luminosity of ngc1377 in the total absence of activity in the radio window ( this will be developed in sect .
[ seyfert ] ) . ] . + * 2 ) * the j , h and k - band surface brightness profiles computed by @xcite are nearly flat inside the central 34 , significantly below the best @xmath54 fit , whereas most seyfert galaxies with @xmath55 require an extra nuclear point source .
a high central absorption could flatten the profiles , but this is unlikely , because 2mass data indicate that the j - h and h - k colors of the central 10 are not different from the global colors within the error bars .
h - k is relatively `` blue '' compared to the other galaxies in @xcite , and j - k is very blue compared to the ultraluminous galaxies in @xcite ( like the bluest starbursts and liners ) , or to the 2mass extended sources brighter than @xmath56 @xcite . the power emitted by dust between 4.5 and 100@xmath1 m is only 1.8 times the total stellar power between 3600 and 4.5@xmath1 m , and about twice more considering the center alone .
this means that the luminosity - weighted extinction can not be high enough to hide a deep stellar potential well .
+ * 3 ) * seyfert galaxies either emit excess radio emission with respect to their infrared emission , or follow the same correlation as normal galaxies @xcite .
although no starburst galaxy is known to be as radio - deficient as ngc1377 , no active nucleus either has been shown to be synchrotron - deficient so far .
x - ray data would be useful to rule out seyfert activity , but to our knowledge , none exist for ngc1377 .
the 7 to 25@xmath1 m emission of ngc4491 is somewhat in excess of the corresponding sed of @xcite , but not significantly .
the blue - band surface brightness profile of @xcite is exponential , with no central peak , and so are the near - infrared profiles .
the j - h and h - k colors are close to those of ngc1377 , and the color differences between the circumnuclear region and the whole galaxy are small
. the same conclusions thus apply . as to ic1953 ,
only the 7@xmath1 m emission is 2@xmath57 in excess of the average empirical sed .
its global near - infrared colors are extremely blue , and its circumnuclear colors are slightly bluer than those of ngc1377 and ngc4491 .
since the disk infrared emission in ic1953 is significant , and well resolved in isocam maps , we decomposed the spectral energy distribution into a disk component and a circumnuclear region ( cnr ) component , which we identify to the core with a high @xmath29 ratio . to do this , we estimated typical disk infrared colors using the isocam and iras fluxes of ngc4027 @xcite , which has high signal to noise ratios and whose entire body can be assimilated to a disk : @xmath58 ( the small 7@xmath1 m excess in the disk of ic1953 being ascribed to stellar emission ) , @xmath59 , @xmath60 , @xmath61 and @xmath62 .
the result , shown in fig .
[ sed]d , outlines the fact that part of the peculiarity of ic1953 is hidden by normal disk properties : the center has hotter dust and higher infrared to optical ratios than the galaxy as a whole .
the radio deficiency is likely more severe in the center ; using the radio measurement of @xcite inside 21 and the above decomposition , we obtain @xmath63 . the position of a galaxy in a mid - infrared surface brightness color diagram ( @xmath64(15@xmath1 m ) versus @xmath29 ) can give some sense of the relative contribution of stellar populations of different ages , on a timescale of several @xmath31 years , in the dust heating @xcite .
the circumnuclear regions of ngc4491 and ic1953 populate a part of the diagram otherwise empty , and have a central @xmath29 color more than twice the average value corresponding to their surface brightness .
this is fully consistent with the hypothesis that they are undergoing a starburst in an early phase , which is massive with respect to the mass of the underlying intermediate stellar population . as for ngc1377
, its @xmath29 excess is only moderate .
however , the emission in the 7@xmath1 m bandpass being due to an unusual dust species , the interpretation of its @xmath29 color is not straightforward .
we mapped the galaxies with the effelsberg 100 m antenna , at 6.2 cm and 3.6 cm , during six observation sessions between april and august 2002 .
these observations made use of the 4850mhz hemt dual - horn receiver ( 30k system temperature , 500mhz bandwidth ) and the 8350mhz hemt single - horn receiver ( 25k system temperature , 1100mhz bandwidth ) , both installed in the secondary focus of the telescope .
the beamwidths are 147 at 6.2 cm and 82at 3.6 cm , thus comparable to the galactic disk sizes . between 10 and 20 coverages per galaxy were obtained , scanned in azimuthal direction at 6.2 cm and in right ascension and declination , alternatively , at 3.6 cm . 3c138 and
3c286 were used as flux calibration sources .
the data reduction was performed using the nod2 software package .
the coverages were combined by applying the `` basket - weaving '' technique @xcite .
the final 3.6 cm maps were smoothed to a beamwidth of 90 to increase the signal to noise ratio .
the measured flux densities are contained in table [ tabradio ] and the contour maps shown in fig .
[ contours ] .
significantly polarized emission was not detected in any galaxy .
figure [ sed_radio ] compares the radio measurements with the free - free and synchrotron spectra of normal galaxies matched to the far - infrared fluxes of ngc1377 , ic1953 and ngc4491 .
the total flux expected at 21 cm was derived from the average @xmath32 ratio , the free - free component from the ionizing photon flux estimate ( assuming @xmath65k ) , and we adopted the spectral indices @xmath66 and @xmath67 , unless otherwise constrained for the synchrotron component . ngc1377 is not detected at any radio frequency , and the upper limits even lie below fluxes expected from a mere free - free component matched to the ionizing photon flux derived from the dust emission ( fig .
[ sed_radio ] ) .
this fact suggests to us not only that ngc1377 is mostly devoid of synchrotron radiation , but also that the star formation episode is so young that most massive stars are still embedded in their parent molecular clouds , which would cause a large fraction of the ionizing photons to be intercepted by dust before being able to ionize the gas .
ngc1377 was reobserved in 1990 at 3.55 cm with the vla in the a configuration , yielding a beam of @xmath68 ( j.j .
condon , unpublished ) .
the galaxy is still undetected at this high resolution , and the @xmath69 upper limit is 0.13mjy per beam , i.e. 15 times less than the 3.6 cm upper limit from effelsberg . interpreting this limit
, we have to take into account the possibility that the star - forming region be more extended than 0.3 . assuming a size of 100pc or @xmath70(see the next to last paragraph in this section ) , the equivalent upper limit on the total 3.55 cm flux is of the order of 1.3mjy .
@xcite derive from a `` snapshot '' observation with the vla @xmath21(6@xmath71mjy for ngc1377 , in a @xmath72-beam , which brings a stronger constraint .
in such conditions , the data require a reduction of ionizing photons by about 70% or more ( also by 70% using instead the 21 cm upper limit ) .
if we adopt the limit in a 0.3-beam , then the ionizing photons have to be suppressed by more than 95% .
if indeed a larger fraction than usual of the ionizing photons are absorbed by dust in ngc1377 , then this will have the simultaneous effect of decreasing the free - free emission and enhancing the infrared emission , with respect to a normal galaxy with the same star formation rate .
thermal opacity is an alternative explanation for the non - detection of the free - free emission . the upper limit at 3.55 cm however places very stringent a constraint in this case : with @xmath73mjy , the effective opacity is @xmath74(3.6@xmath75 , or @xmath74(21@xmath76 .
assuming a simple geometry and uniform electronic density , such a value would imply that the scale of the emitting region be less than 28pc , and the electronic density higher than 2700@xmath77 .
since this is unrealistic , the free - free opacity explanation is dismissed . assuming a normal initial mass function and no dust emission excess , the starburst must have produced approximately @xmath78 o stars ( see section [ sfrate ] ) .
assuming that the dust emission at 60 and 100@xmath1 m is due to a pure blackbody of temperature 80k ( which fits the 25 to 100@xmath1 m data with only a small observed excess of 25% at 25@xmath1 m ) , its diameter is about 37pc . using the model of @xcite for o5 stars ,
the size of the starburst region is rather of the order of 100pc ( 1 ) or larger .
the emission region may thus be truly more extended than the vla a - configuration beam . to trigger star
formation quasi - instantaneously over such a scale perhaps needs a merger with a satellite galaxy , which does not necessarily produce morphological disturbances , but should be visible in the gas kinematics . the dust mass required to produce the observed far - infrared emission is @xmath79m@xmath12 , according to the dust emissivity derived by @xcite for the galaxy , with @xmath80k ( which produces an emission excess from small dust grains of 76% at 25@xmath1 m ) .
using instead a blackbody fit , only @xmath81m@xmath12 of dust are needed . assuming that the gas to dust mass ratio is similar to the galactic value of @xmath0 @xcite , the gas mass associated with the starburst in ngc1377 is then of the order of @xmath82 to @xmath83m@xmath12 , which represents much less than 10% of the observed stellar mass in the whole central region .
since this mass ratio is reasonable , the hypothesis that the dust is heated by a starburst is plausible , but still needs to be confirmed by observations of the molecular gas : see section [ sest ] .
the upper limit on the atomic gas mass , m(hi@xmath11m@xmath12 ( section [ envir ] ) , is not in contradiction with the starburst hypothesis .
we will discuss the possibility that a radio - quiet active nucleus heats the dust in section [ seyfert ] .
the entire radio spectrum lies well below a normal synchrotron spectrum , but contrary to ngc1377 , it is higher than the expected free - free component alone . at 21 cm , @xcite measure a flux density of @xmath84mjy inside a 21-beam , which is a bit larger than the size of the circumnuclear regions seen in the mid - infrared .
this value is consistent with the free - free emission derived from the infrared spectrum of the central regions .
as explained in section [ irsed ] , we performed a decomposition of the infrared spectral energy distribution into disk and circumnuclear ( cnr ) components , based on an empirical color model and on the true distribution observed at 7 and 15@xmath1 m .
this allows us to estimate the free - free and synchrotron emission separately for the disk and the galactic center , as shown in fig .
[ sed_radio ] ( note that no component has been tuned in any way to the radio data in this graphic ) .
the central regions thus appear almost devoid of any synchrotron emission ; otherwise , the 21 measurement would be higher .
this finds support in the total flux densities at 21 cm , 6.2 cm and 3.6 cm being in agreement with the synchrotron radiation derived for the disk alone ( with a reasonable spectral index assumed for illustration purposes to be -0.75 ) .
extremely diffuse extended emission from the disk is detected at all three wavelengths .
the 3.6 cm map contains extentions to the south , which do not correspond to any structure at other wavelengths and are thus suspect ; if the flux measurement is limited to the beam area rather than the total extention , however , then one obtains @xmath85mjy instead of @xmath86mjy .
the data therefore are consistent with free - free emission and disk synchrotron emission , but not with the central synchrotron component expected from a normal supernova rate .
one may question this result , since a minimum level of emission would be produced by cosmic rays originating in the disk and diffusing into the central region . in @xmath87 years , cosmic rays may propagate to about 1kpc from their source , using a diffusion coefficient of a few @xmath88m@xmath89s@xmath10 , as found by e.g. @xcite and @xcite in the galaxy .
scaling the disk synchrotron brightness , which is very low , by a reasonably large size of 5 for the starburst , and by an increase of the magnetic field strength from 10@xmath1 g to 50@xmath1 g ( see section [ magfield ] for a justification of these numbers ) , we estimate the synchrotron emission from disk cosmic rays to be at most of the same order as the free - free emission within the central region , even at 20 cm .
furthermore , if the central magnetic field is much larger than the disk magnetic field , then it will reflect cosmic rays coming from the disk @xcite .
ic1953 was reobserved at 3.55 cm with the vla in the a configuration ( @xmath68 beam ) .
the detected flux is @xmath90mjy , and the size of the emitting region of the order of @xmath91 , with a position angle of 11(j.j .
condon , unpublished ) .
this compact component accounts for less than half the peak flux density measured at 3.6 cm with the effelsberg antenna , which is @xmath92mjy .
it is consistent with normal free - free emission provided that about 65% of the flux comes from a region larger than 0.3 , i.e. 30pc , which is very likely , since the lower limit on the size of the starburst , obtained for a blackbody of 70k matched to the cnr far - infrared emission , is 46pc . only for ngc4491
does a flux measurement at another radio wavelength exist .
@xcite measured @xmath21(2.8@xmath93mjy , with the effelsberg telescope , in a 69-beam .
since the observed flux density at 3.6 cm in ngc4491 ( @xmath94mjy ) , confirming the measurement of @xcite at 2.8 cm , is in good agreement with what is expected from the spectra of normal galaxies ( 4.9mjy for a non - thermal index of -0.75 between 3.6 cm and 21 cm ) , and the free - free emission can contribute only of the order of 20% of the total emission at this wavelength , we have evidence that ngc4491 produces synchrotron emission .
however , this emission seems to vanish at 21 cm : the nvss map indicates that @xmath95mjy .
ngc4491 was not detected either in the ( much less sensitive ) 12.6 cm survey of @xcite .
source confusion could be responsible for this inconsistency . using the 8.4ghz source counts of @xcite , which indicate @xmath96 , and a beam of 80
, the probability of observing a confusing source at 3.6 cm lies between 1/360 and 1/800 .
given that the radio positions at 3.6 cm and 6.2 cm are in good agreement with the optical center , as well as the position at 6.2 cm of a bright quasar seen to the sw in the nvss map , and that the observed flux coincides with the flux expected from the infrared emission , confusion is unlikely .
as ngc4491 lies only 54.5 away from m87 , which emits about 220jy at 20 cm , it is possible that the nvss map does not retrieve the correct flux density of ngc4491 , due to imperfect cleaning of the beam sidelobes which can generate negative sources . to evaluate this effect , we selected a random sample of 50 nvss fields ( centered on ned extragalactic sources ) , at distances to m87 between 52.5 and 56.5 , and looked in each one for the minimum brightness .
the histogram of the minima forms a smooth distribution down to -6.52mjy per beam , and two outliers at @xmath97 and -16mjy per beam .
the latter are characteristically associated with a chain of other deep negative points , of amplitude scaled by about one third . since
this pattern is not present in the ngc4491 map , we are confident that such a deep negative component can not be superimposed onto ngc4491 ( furthermore , it would have to coincide accurately with the galaxy s position , the probability of which is low ) .
we thus adopt as an absolute upper limit to any missing flux density the value 6.52mjy , and we derive @xmath98mjy .
we will use this limit in the following discussion , but we insist that deeper observations at 20 cm are needed to set any more significant limit . the new infrared to radio flux ratio is only constrained to be @xmath99 .
however , the spectral indices are very flat : @xmath21([3.66.2cm]@xmath100 and @xmath21([6.220cm]@xmath101 with @xmath102 .
the second value is very uncertain , but the [ 3.66.2 cm ] index is more reliable . from radio continuum studies carried out at high angular resolution @xcite ,
several starburst galaxies , e.g. ngc2146 , ngc4214 , ngc5253 , he2 - 10 for the most part with wolf - rayet signatures and being either irregular , amorphous , blue compact or merging dwarf galaxies have been found to contain sources with flat or inverted radio spectra , revealing optically - thick free - free emission regions .
however , we find that considering _ total _ infrared and radio fluxes , all of these galaxies have normal @xmath32 ratios ( between 2.23 and 2.67 ) . moreover , ngc4491 emitting synchrotron emission at 3.6 cm , it is much more difficult to explain the shape of its radio spectrum by free - free absorption , for the thermal electrons would have to fill a very compact distribution , and at the same time encompass the relativistic electrons . assuming nevertheless that this is the case
, the effective opacity at 20 cm would have to be larger than 0.7 .
@xcite found three ultraluminous galaxies with @xmath103 , which are argued to be compact starbursts in merging systems , and whose radio - quiescence is best explained by high free - free opacity .
however , these very luminous objects are not comparable at all with the galaxies studied here . for two of them
do stellar continuum observations exist , and their 60@xmath1 m to near - infrared power ratios are more than 20 to 40 times greater than in ngc1377 , ngc4491 and ic1953 .
furthermore , the required free - free opacities are only 0.50.7 at 20 cm .
thus , ngc4491 is more likely intrinsically deficient in synchrotron emission rather than opaque at 20 cm .
the hypothesis of non - stellar activity powering the infrared and radio emission will be discussed in section [ seyfert ] .
another possibility is that the population of relativistic electrons is very young , so that no significant energy losses occurred yet .
we address here the likelihood of such a scenario .
the median spectral index of radio supernova remnants , which are the likely sites of cosmic ray acceleration , is about @xmath104 ( with a total distribution between @xmath105 and @xmath106 ) , independent of age @xcite .
this is significantly flatter than the spectral index of diffuse synchrotron emission in galactic disks , @xmath106 on average @xcite .
very flat indices of supernova remnants can be reproduced in diffusive shock acceleration models , when second - order fermi acceleration is also included @xcite or when backreaction on the shock due to the pressure of the relativistic particles is taken into account @xcite . since the lifetime of radio supernova remnants is short , @xmath107 years @xcite , and they normally account for a small fraction of the total synchrotron emission in a galaxy , it is unrealistic to assume that the radio continuum emission of ngc4491 be due predominantly to flat - spectrum supernova remnants .
steepening of the energy spectrum is a natural effect of synchrotron and inverse compton losses , once the relativistic electrons have escaped their acceleration sites , but it gradually takes place over their whole lifetime of several @xmath87 years . cosmic - ray diffusion also contributes to steepen the radio spectrum : with a diffusion length scaling as @xmath108 for @xmath109gev @xcite , the spectral index is decreased by @xmath110 . the flat radio spectrum of ngc4491 can therefore be explained if any old population of cosmic rays has completely decayed , and a fresh population has started building up within the last few myr .
this is consistent with the observations presented in section [ n4491_pifs ] . assuming constant spectral indices between each measurement pair , the total radio power of ngc4491 between 3.6 cm and 20 cm
is of the order of 74% of the radio power expected of a normal spectrum , or less , owing to the 20 cm upper limit and to the fact that unobserved frequencies lower than 1.4ghz were not included in the computed power .
it is thus conceivable that not all cosmic rays were yet released by supernova remnants and that the interstellar medium of ngc4491 is out of equilibrium ( in which case the assumption of equipartition between cosmic ray and magnetic energy densities is invalidated ) .
the h@xmath33 recombination line , classically used as tracer of massive stars , is not suitable to estimate the star formation rates of the three galaxies , because nebular extinction in the circumnuclear core can be expected to be high . in a 2-wide slit placed on the nucleus of ngc1377 ,
@xcite detect [ nii ] ( 6583 ) and [ sii ] ( 6716 and 6731 ) in emission , but not h@xmath33 , and furthermore a high [ sii]@xmath111 to [ sii]@xmath112 ratio , indicating that the detected nebular emission arises in low - excitation and low electronic density regions , presumably in shocks . modelling the starburst nucleus of ngc4569
, @xcite found that the [ sii ] and [ nii ] emission arises predominantly from low - density gas at large distances from the starburst core ; the same may apply to ngc1377 .
ic1953 was observed in ( h@xmath33 + [ nii ] ) by @xcite , who kindly provided their map to us . in the h@xmath33 light ,
the circumnuclear region is fainter than several star - formation complexes in the disk , whereas it is of much higher surface brightness than any other part of the galaxy in the 7 and 15@xmath1 m dust emission .
the bar is not detected in h@xmath33 , which is not customary in late - type galaxies , but more characteristic of strong bars .
the three galaxies do not feature in the optical range any indication of a starburst , but have starburst - like far - infrared and mid - infrared colors . in order to better constrain the star formation and supernova rates , we undertook an observation program of near - infrared emission lines .
we observed ngc4491 on february 1 and 2 , 2002 , at the 5-m hale telescope at the palomar observatory , with the palomar integral field spectrograph , pifs @xcite .
the hydrogen recombination lines pa@xmath113 and br@xmath114 were chosen to estimate the star formation rate and the extinction in the circumnuclear region , and the [ feii]1.644 line was used to set a constraint on the supernova rate .
we also observed ngc4102 , as a bright comparison galaxy with a very young circumnuclear starburst @xcite , but normal infrared and radio properties ; its @xmath32 ratio is 2.37 , and its central mid - infrared color , @xmath115 , is comparable to those of the three galaxies studied here . on october 17 - 19 , 2002 , we pursued this program by observing ngc1377 in pa@xmath113 , br@xmath114 , [ feii]1.644 and the h@xmath116(1 - 0)s(1 ) line at 2.12@xmath1 m .
we chose as a second comparison galaxy ngc1022 , a bright starburst galaxy with a moderately high @xmath32 ratio ( sect . [ general ] ) . in view of the raw data
, we decided to observe also h@xmath116(2 - 1)s(1 ) at 2.25@xmath1 m and h@xmath116(1 - 0)s(0 ) at 2.22@xmath1 m .
the observations and data reduction are described in appendix [ app_pifs ] .
the spectra are shown in fig .
[ spec ] and the images and velocity maps in fig .
[ pifsima ] . in both ngc4491 and ngc1377 ,
the velocity gradient derived from the line centroids is much smaller than the velocity resolution . using the pa@xmath113/br@xmath114 decrement as an extinction estimator
, we obtain @xmath117(pa@xmath118 in the central @xmath119 of ngc4491 , and @xmath117(pa@xmath120 in the central @xmath121 of ngc4102 .
this is assuming an intrinsic pa@xmath113/br@xmath114 ratio of 5.58 , valid for @xmath65k and @xmath122@xmath77 @xcite , and adopting the extinction law of @xcite .
these extinctions are equivalent , in the h@xmath33 line , to respectively 2.4 and 7.3 .
although the absolute central star formation rate of ngc4491 , @[email protected]@xmath10 using the calibration of @xcite , is much less than that of ngc4102 , @[email protected]@xmath10 , the pa@xmath113 equivalent widths of the two galaxies are more comparable : we measure respectively 6.65 in @xmath119 ( 2.24in @xmath125 ) and 9.16 .
ngc4102 is experiencing a powerful outflow ( accounting for about one third of the total h recombination line fluxes ) , at a velocity of @xmath126km.s@xmath10 , which confirms the youth of its starburst .
the [ feii]1.644 emission is thought to be due to shock excitation in young supernova remnants @xcite . as supernov
appear only a few myr after the starburst onset , the ratio of the [ feii ] line to the hydrogen recombination lines , which are produced almost instantaneously , rises as a function of time elapsed since the beginning of star formation .
the [ feii]/br@xmath114 flux ratio in the center of ngc4102 , corrected for extinction , amounts to about 1.3 , which is in the low range of values found in spiral starbursts , between 0.9 and 3 @xcite . in the central @xmath119 of ngc4491 ,
[ feii]/br@xmath114 is constrained to be less than 0.5 at the @xmath127 level . comparing this to what is expected from the synthesis model of @xcite for continuous star formation over a short period , with default parameters ( solar metallicity and salpeter imf between 1 and 100m@xmath12 ) ,
the last star formation episode is less than 7.1myr old . in the same model
, the starburst of ngc4102 is at least twice older ( @xmath128myr ) , but has not reached steady - state either .
ngc4491 is very faint , so our observations are not sensitive enough to produce more stringent a constraint
. however , the data are consistent with the idea that ngc4491 is observed very shortly after the onset of a new episode of star formation , even more recent than in ngc4102 . for the central @xmath119 of ngc1022 , the galaxy used for comparison with ngc1377 , we derive @xmath117(h@xmath129 , @xmath130(pa@xmath131 , and [ feii]/br@xmath132 .
this galaxy has not reached steady - state either , due to its moderately high infrared to radio flux ratio ( section [ general ] ) , and hosts an extranuclear star - formation complex about 2 to the north of the nuclear complex .
if most ionizing photons are intercepted by dust as suggested by the radio spectrum of ngc1377 , then the hydrogen recombination lines severely underestimate the star formation rate . in agreement with the lack of free - free emission
, we detect none of the hydrogen lines . in an area of @xmath119 centered on the nucleus , the @xmath69 upper limits are @xmath133([email protected]@xmath135 and @xmath133([email protected]@xmath135 .
we do not detect either the shock tracer [ feii ] , but a strong h@xmath116(1 - 0)s(1 ) emission line is observed .
the h@xmath116(2 - 1)s(1 ) to h@xmath116(1 - 0)s(1 ) ratio is less than 0.03 and the h@xmath116(1 - 0)s(0 ) to h@xmath116(1 - 0)s(1 ) ratio is less than 0.07 , which points to low - velocity shocks as the main mechanism exciting the molecular gas , rather than fluorescence , unless the gas density is high enough to de - excite the molecules by collisions @xcite ( but the hydrogen nucleus density would have to be higher than @xmath137@xmath77 ) .
the emission arises in a region smaller than 2 , or 200pc . in the hypothesis of very high nebular extinction ,
the h@xmath116 emission is probably foreground , but it could also originate from the molecular material embedding the newly - formed stars .
the young starburst ngc1022 emits much less flux in h@xmath116 than in any other line ( table [ pifs ] ) and has a h@xmath116(2 - 1)s(1 ) to h@xmath116(1 - 0)s(1 ) ratio of the order of 0.2 , so that both shocks and ultraviolet photons can contribute to the molecule excitation .
the lack of hydrogen emission in ngc1377 may be due partly to lack of ionizing photons and partly to nebular extinction . in the assumption that dust absorbs as much as @xmath138% of the ionizing photons , as suggested by the lowest radio continuum upper limit ( sect .
[ n1377 ] ) , then the dust emission could easily overestimate the star formation rate by a factor of two . with @xmath139 ,
our most stringent limit on the recombination rate still implies a nebular absorption @xmath140 , or equivalently @xmath141 . if only 70% of the ionizing photons are absorbed by dust ( sect .
[ n1377 ] ) , then @xmath142 , or @xmath143 .
either the nebular extinction is very high , or the quasi - totality of the ionizing photons are absorbed by dust , i.e. no hii regions have yet appeared . to prevent the development of hii regions , massive stars have to still undergo an active accretion - outflow phase , which lasts less than @xmath144years @xcite .
this phase could be responsible for shocking the molecular gas in ngc1377 .
@xcite discuss the properties of a galactic iras source which is a massive star - forming region , still accreting material , and with no free - free emission , hence in a pre - hii region stage .
@xcite modelled the dust emission from an accreting ob star , and obtained spectral energy distributions strikingly similar to the 3100@xmath1 m spectrum of ngc1377 in its qualitative features .
we thus propose that ngc1377 may host a proto - cluster of this type of object , accounting for an unknown fraction of its infrared luminosity . in this case , a large part of dust heating might not be provided by the direct radiation from massive stars , but by the accretion luminosity @xcite , and in view of the luminosities computed for a range of model parameters , of the order of @xmath145 o stars would be needed to generate the total infrared luminosity of ngc1377 , i.e. a few times more than for main - sequence stars .
the continuum emission in the pifs field of view has the same morphology and orientation as the large - scale stellar emission ( fig .
[ images ] ) and shows no disturbance .
the velocity gradient in the h@xmath116(1 - 0)s(1 ) line is much less than the velocity resolution , and the continuum emission does not indicate any stellar density cusp .
however , the continuum map ratios reveal an emission excess in the ks band , in the central 12(slightly elongated parallel to the major axis ) . the color excesses with respect to the surrounding bluer area , @xmath146 and @xmath147 , are not compatible with pure extinction ( @xmath148 using the extinction law of @xcite ) , but consistent with the presence of hot dust contributing to the ks - band emission . assuming that the hot dust contributes only to the ks - band flux , and not to the j and h fluxes
, we find that @xmath149 , the flux density ratio of the mid - infrared peak at 7.7@xmath1 m to the ks - band excess , is of the order of 500 . as a comparison ,
@xcite found an average ratio of 150 in a sample of normal star - forming galaxies ( see their figure 6a ) .
the 68@xmath1 m mid - infrared emission in ngc1377 is thus abnormally high not only with respect to the far - infrared emission , but also with respect to the hot dust emission in the ks band , although its ( h - k ) color excess is more extreme than in all but one of the galaxies studied by @xcite , drawn from the same sample as the galaxies in @xcite .
table [ sfr ] lists the _ total _
star formation rates based on an extrapolation of far - infrared fluxes to the 81000@xmath1 m range @xcite , using the observed spectral energy distribution to compute the 842@xmath1 m flux and the model of @xcite to estimate the contribution from 122 to 1000@xmath1 m , as a function of @xmath2 .
star formation rates in the circumnuclear regions can be derived from the spatially - decomposed mid - infrared fluxes , applying the calibration established in @xcite .
ic1953 and ngc4491 are depleted in 7@xmath1 m emission with respect to the far - infrared , which could mean that the aromatic band carriers are destroyed in these high dust - temperature galaxies ; the 7@xmath1 m calibration is therefore not applicable to them without correcting for the grain destruction effect .
considering that galaxies from which the calibration was derived have @xmath150 ratios between 0.12 and 0.25 , and that this quantity is about 0.02 in the center of ic1953 and 0.04 in ngc4491 , we estimate star formation rates of respectively @[email protected]@xmath10 and @[email protected]@xmath10 . these numbers are not far below the _ total _ star formation rates .
ngc1377 , on the contrary , has a 7@xmath1 m to far - infrared flux ratio similar to normal and disk - dominated galaxies ( fig .
[ mirsfir ] ) .
although the 611@xmath1 m spectrum of ngc1377 reveals that the dust responsible for the emission in this range is peculiar , the star formation rate derived from @xmath153 agrees to within 3% with that derived from the total dust emission for ngc1377 ( table [ sfr ] ) . to further characterize the star formation activities in the central regions of ic1953 and ngc4491 , a comparison of their mid - infrared data with those of circumnuclear starbursts and merging galaxies , for which ionizing photon flux densities derived independently are available , can be instructive . using the circumnuclear regions of the galaxy sample of @xcite , the starburst galaxies m82 , ngc253 and ngc1808 , and the merging systems arp220 and ngc6240 to span a range of more than five orders of magnitude in star formation surface density , we found a well - defined relationship between the mid - infrared color and the ionizing photon flux density @xcite .
7@xmath1 m fluxes show no appearance of saturation and remain valid star formation tracers up to ionizing photon flux densities as high as @xmath154s@xmath10pc@xmath135 , but decline steadily above that threshold @xcite .
the colors of ic1953 and ngc4491 suggest that their star formation is more compact than in the m82 core
. if this assumption is correct , this corresponds to ionizing photon flux densities above @xmath155s@xmath10m@xmath135 . combined with star formation rates derived from the total infrared emission ,
upper limits on the sizes of the star - forming cores can be inferred : diameters of less than 125pc and 70pc are respectively obtained . in the blackbody assumption , we obtain lower limits of 46pc and 35pc .
the derived star formation rates are only of the order of [email protected]@xmath10 . however , the three galaxies have low masses . comparing these rates with the underlying stellar mass , as traced by the ks - band light , we find that the timescale to build up an equivalent mass is only of the order of 0.52gyr in the central @xmath156 of ngc1377 , ic1953 and ngc4491 ( using @xmath157 , as found by @xcite , with ks@xmath158mag , from @xcite ) . as a comparison ,
this timescale is similar for the starburst in ngc4102 ( 1.7gyr ) and much longer in the bright star - forming center of the normal spiral ngc4535 ( of the order of 10gyr ) .
the star formation activity in the three galaxies can thus be truly identified with starbursts . using the star formation rate range derived from dust emission , and comparing it with the extinguished star formation rate derived from the h@xmath33 line , we derive an absorption in h@xmath33 between 4.3 and 4.9 for the circumnuclear region of ic1953 .
if the [ nii ] to h@xmath33 ratio were unusually high such as in ngc1377 , then the nebular absorption could be even higher . for ngc4491
, we have measured the star formation rate from near - infrared hydrogen recombination lines in the pifs field of view .
the 7@xmath1m - emitting region may be larger than this , but the strong discrepancy between the recombination estimate and the dust estimate ( table [ sfr ] ) points to either a major fraction of the nebular emission being completely screened , or an overestimation of the star formation rate by the infrared emission , unless photospheric absorption from a young stellar population strongly affects the hydrogen line strengths .
far - infrared lines due to collisional excitation in photodissociation regions and ionized regions were observed by iso - lws with an effective beam of 78 ( lws handbook ) .
this beam is much larger than the size of the infrared emission of ngc1377 and ngc4491 , but encompasses only 60% of the total 7@xmath1 m emission of ic1953 .
fluxes are collected in table [ lws ] . the [ cii]157.7 line is the major coolant of photodissociation regions . since aromatic band carriers dominating the emission in the 7@xmath1 m bandpass are of planar structure
, they are thought to be more efficient sources of photoelectric heating than three - dimensional dust grains @xcite .
this finds a confirmation in the constancy of the [ cii ] to 7@xmath1 m flux ratio found by @xcite in star - forming spiral galaxies . ngc4491 and ic1953
behave exactly as this sample : their @xmath159}/f_7 $ ] ratios are within @xmath160 of the average value .
their @xmath159}/f_{\rm co}$ ] ratio is of the order of 20004000 , depending on the assumed co spatial distribution , or less , owing to the smaller co beams in the observations of @xcite and @xcite ( respectively 33 and 44 ) .
these values are typical of normal star - forming galaxies , and argue for near - solar metallicities , since metal - deficient galaxies will have much higher ratios @xcite .
however , ngc1377 appears deficient in [ cii ] by at least an order of magnitude and lies more than @xmath161 ( in dex units ) below the average @xmath159}/f_7 $ ] .
interestingly , the 7@xmath1 m flux in ngc1377 , which is not produced by classical aromatic particles , is also 0.85dex higher than expected from its 60 and 100@xmath1 m fluxes ( fig .
[ sed ] ) .
the species responsible for the 7@xmath1 m emission in ngc1377 is thus either not easily photoionized , because already highly ionized in its equilibrium state , or is heated mainly by low - energy photons , which is unlikely owing to the observed high dust temperatures in this galaxy , or is not an efficient purveyor of photoelectrons if its structure is not planar .
c@xmath162 is collisionally de - excited above relatively low densities , such that the [ cii ] deficiency of ngc1377 could be due to increased density ; but [ oi]63 remains an efficient cooling agent at densities 100 times higher . since the @xmath163}/f_{\rm fir}$ ] ratio of ngc1377 is also low compared to the values it takes in @xcite , this explanation can be ruled out .
@xcite also discard high optical depths in c@xmath162 as the cause of the [ cii ] deficiency ( relative to the far - infrared flux ) in their high-@xmath2 galaxies , because they would expect [ oi ] to become optically thick before [ cii ] , which they do not observe .
a plausible origin of the [ cii ] deficiency in ngc1377 is thus a radiation field intense enough to highly ionize the small dust particles emitting in the mid - infrared , which then can not heat the interstellar medium , and different structural properties of these particles than in normal galaxies . alternatively , since hii regions seem to be absent in ngc1377 ( sect .
[ n1377 ] and [ n1377_pifs ] ) , c@xmath162 ions may not exist and thus be unable to act as a coolant . in any case ,
the [ cii ] deficiency is unusual and puzzling .
the upper limits on @xmath164}$ ] are within the range of detections at lower dust temperatures of @xcite , when normalized by the far - infrared flux .
ngc1377 was observed in the co(1 - 0 ) and co(2 - 1 ) lines at the sest in january 2003 , during three sessions of 7h each , as part of the eso time share .
important details on the observations can be found in appendix [ app_sest ] .
the measurements and results are given in table [ tabco ] .
we measure @xmath165kkms@xmath10 for the co(1 - 0 ) line and a co(2 - 1 ) to co(1 - 0 ) brightness temperature ratio of @xmath166 .
the spectra are shown in figure [ co_maps ] .
the averages of the spectra of the five centermost pointings are shown in figure [ co_tot ] .
the co(2 - 1 ) line profile is perfectly fitted by a gaussian of fwhm 66kms@xmath10 ; the co(1 - 0 ) line profile can be fitted by a gaussian of the same width and centered at the same velocity , but shows a wing toward high velocities .
this excess is only at the @xmath167 level . for lack of adequate constraints on the h@xmath116 mass to co luminosity ratio , we adopt a standard value of @xmath168 moleculesm@xmath135(kkms@xmath10)@xmath10 @xcite , which is uncertain by a factor of at least a few . for infrared - luminous starbursts
in which co is subthermally excited , assuming virialized clouds , @xcite derive factors between about 2 and @xmath169 , in the same unit . due to its low co(2 - 1 ) to co(1 - 0 ) brightness ratio
, collisional excitation of the molecular gas in ngc1377 seems to be also subthermal .
@xcite similarly found very low co(2 - 1 ) to co(1 - 0 ) brightness ratios in starburst galaxies , that suggest that most of the emission comes from low - density and very clumpy gas @xcite , while the star - forming cores are probably hidden by their envelopes .
we derive a hydrogen mass of the order of @xmath170m@xmath12 .
this is at least ten times more than the minimum mass required by the starburst hypothesis ( sect .
[ n1377 ] ) .
the far - infrared to co flux ratio is about seven times higher than the average for normal galaxies , but within the broad range observed for starbursts @xcite , which is not surprising owing to the high dust temperature and emissivity in ngc1377 .
the total gas mass to blue luminosity ratio lies at the upper bound of the range measured by @xcite for a volumed - limited sample of lenticular galaxies .
the timescale to consume all the molecular gas , including the mass of helium and metals , at the current star formation rate derived from the far - infrared emission , can be estimated to only @xmath171myr .
the co(1 - 0 ) line was also observed in ic1953 , ngc4491 and ngc4418 ( see below a discussion of this galaxy ) , with smaller signal to noise ratios @xcite . in all cases , the velocity profile is very narrow ( approximate fwhm of 120 , 80 and 120kms@xmath10 ) , and the direct effect of rotation invisible , which argues for a concentration of the molecular gas in a small central area .
the size of the molecular disk , even if it can not be directly estimated , can be coarsely constrained by indirect arguments .
if one assumes that ngc1377 must follow the same schmidt law as characterized by @xcite from normal spirals to starbursts , within the observed scatter , then the diameter of the gas disk should be ( @xmath172)pc .
dynamics can be used to set an independent constraint , but one has to make the unverified assumption that the co disk is coplanar with the stellar disk , which could be false if the gas was accreted from another galaxy .
the maximum velocity , corrected for inclination , is about 90kms@xmath10 . in case
the gas is decoupled from stars ( i.e. the stars are distributed in a much more extended structure with a significant plane - perpendicular scale height ) , the diameter is about 330pc .
if the gas is confined within this region , then it accounts for a large fraction of the total enclosed mass : at least a quarter , estimating the stellar mass from the ks image .
we now compute the toomre parameter for local gravitational stability , @xmath173 @xcite , in the same manner as usually done for galactic centers , i.e. assuming a gas density @xmath174 equal to the average , and the epicyclic angular velocity @xmath175 in the linear part of the rotation curve .
the sound speed @xmath176 is chosen of the order of 10kms@xmath10 , and the diameter 4 ( 400pc ) . in these conditions , @xmath177 , which means that the gas is locally very unstable .
we neglected stars , although they influence the gas stability , but they are expected to only decrease it @xcite .
the parameters entering @xmath178 are all ill - constrained , and there is no guarantee that the gas is distributed in a thin disk , but if it is the case , since values close to unity are routinely found in the literature ( indicating gravitational stability allowing the mild growth of perturbations ) , it is at least likely that the molecular gas disk of ngc1377 be overcritical .
this would explain the instantaneous trigger of a starburst throughout @xmath0pc .
sound evidence clearly requires more detailed data , difficult to obtain owing to the low brightness of ngc1377 . finally , the co flux can be coupled with the h@xmath116 line measurements described in section [ n1377_pifs ] , to estimate the mass fraction of hot molecular gas .
the h@xmath116(1 - 0)s(0 ) upper limit does not bring any useful constraint , but the lower limit on the h@xmath116(1 - 0)s(1 ) / h@xmath116(2 - 1)s(1 ) flux ratio implies temperatures below 1500k . with standard assumptions ( due to lack of constraints by the observations ) ,
the hot h@xmath116 mass is then above 250m@xmath12 , and the fraction of hot gas is above @xmath179 .
such a fraction is in the normal range found for bright starburst galaxies and can not be further constrained .
ngc4418 , mentioned by @xcite as a high-@xmath32 galaxy , is an interesting case for which very valuable data exist .
its far - infrared luminosity is @xmath180l@xmath12 , with @xmath181 , @xmath182 and @xmath183 .
the nvss map shows a second source without any optical counterpart about 1.25 to the east , which is certainly included in the iras fluxes ; depending if this source is included or not in the 21 cm flux , the @xmath32 ratio of ngc4418 is 3.02 or 3.09 , between 3.5 and @xmath184 above the average ratio .
@xcite show that its mid - infrared spectrum presents a broad feature similar to that observed in ngc1377 , but with strong ice absorption bands superposed , which are almost never observed .
other peculiarities of this galaxy are the detection of the h@xmath116 line at 2.12@xmath1 m in emission @xcite and of an oh megamaser @xcite .
@xcite suggested that , like in arp220 , the infrared emission arises from an extremely compact energy source .
ngc4418 is also deficient in [ cii ] , but @xmath159}/f_7 $ ] is only constrained to be more than @xmath185 below the average ratio in @xcite .
there are two other major differences with respect to ngc1377 : ngc4418 has much lower mid - infrared to far - infrared flux ratios , in agreement with the predictions from its high dust temperature , and a very compact radio source is detected at its nucleus , with a brightness temperature of @xmath186@xmath187k both at 2.3ghz inside 0.1 @xcite and at 1.49ghz inside @xmath188 @xcite .
ngc4418 is not detected in x - rays @xcite .
although the compactness and extreme extinction in ngc4418 suggest the presence of a seyfert nucleus , a highly compact starburst is also plausible , and the nuclear 2.2@xmath1 m emission is shown to be more extended than the psf by @xcite .
the 3.55 cm radio continuum , observed with the vla in a @xmath68 ( j.j .
condon , unpublished ) confirms that the radio source is extended , with a size of @xmath189 . during our radio continuum observations of ngc1377 ,
ngc4491 and ic1953 at effelsberg , we also mapped ngc4418 ( fig .
[ sed_radio ] ) .
we measured ( @xmath190)mjy at 6.2 cm and ( @xmath191)mjy at 3.6 cm ( the beam including the source located 1.25 eastward ) .
combined with the nvss flux at 21 cm , ( @xmath192mjy ( with the contaminating eastward source added to ngc4418 ) , this indicates a pronounced spectral bend from a steep index at high frequencies ( @xmath193 } \propto \nu^{-0.94}$ ] ) to a flat index at low frequencies ( @xmath194 } \propto \nu^{-0.30}$ ] ) .
the implied high free - free opacity is consistent with the high extinction and compactness of ngc4418 .
we find that the combined effect of both thermal opacity and a reduction of the intrinsic synchrotron emission ( with respect to the level expected from the far - infrared emission ) provides the best fit to the spectrum .
the lowest possible opacity and most reasonable non - thermal spectral index , @xmath74(21@xmath195 and @xmath196 , are obtained for an intrinsic deficit of synchrotron emission of about 50% .
diminishing the synchrotron deficit in the model causes the opacity to rise and the spectral index to become steeper ; assuming no deficit , @xmath74(21 cm ) would have to be as high as 1.6 and @xmath197 as low as -1.57 .
condon obtained with the vla - a a higher flux at 3.6 cm : 27.4mjy , including the contaminating eastward source . using this measurement instead of the single - dish flux ,
we obtain a maximum opacity @xmath74(21@xmath198 , the steepest possible non - thermal spectral index is @xmath199 , and in such conditions , the intrinsic synchrotron emission is reduced by 70% with respect to a galaxy with normal @xmath32 ratio . based on this new result , and on the other existing evidence , we argue that the infrared emission of ngc4418 originates from a very young and compact starburst .
seyfert activity is not necessary to explain the data .
the emission line properties of ngc4418 are also strikingly similar to those of ngc1377 .
it was observed with the pifs spectrograph by @xcite , who detected neither pa@xmath113 , br@xmath114 nor [ feii]1.644 , but a clear h@xmath116(1 - 0)s(1 ) line .
this fact strengthens the similarity between the two galaxies , and suggests a common excitation mechanism ; since ngc4418 is a synchrotron source , though , it must have reached a later evolutionary stage than ngc1377 , and is possibly a composite source containing both a nascent starburst and a more evolved component .
we now discuss plausible physical scenarios to account for the high infrared to radio ratios of the three galaxies we have studied .
if they are due to a deficit of radio emission , then the primary possibility is a lack of cosmic rays , which would follow either from a nascent starburst or from an initial mass function biased towards low - mass stars ; the alternative is a very weak magnetic field .
if the radio emission is not depressed , then it might be absorbed , pointing to optically - thick 20 cm emission .
finally , we discuss the possibility of non - stellar activity providing dust heating but remaining radio - quiet .
the hypothesis that we favor to explain the infrared activity and the radio deficiency of ngc1377 , ngc4491 and ic1953 is the onset of a star formation burst in their centers . in this view
, the synchrotron weakness derives from a lack of relativistic electrons , which can happen if the previous major star formation episode occurred more than @xmath200 years ago , and the current episode started less than a few ( 1 to 7 ) myr before the epoch of observation . in ngc1377 and the central region of ic1953 ,
no synchrotron component is detected , which implies in this scenario that no or very few supernov have yet been produced . in ngc4491 ,
the synchrotron component does exist but has a much flatter than usual spectral index .
we interpret this as due to an energy distribution of cosmic rays inflated with high energies , rather than the effect of free - free opacity ; the number of relativistic electrons released by supernov is significant , but may still be rising , and the near - infrared spectroscopic results indicate that shocks are weaker than expected from a steady - state situation .
other observations consistent with this scenario are : the high dust temperatures , indicating starburst - like radiation intensities ; the extreme @xmath29 flux ratios of ngc4491 and ic1953 , suggesting that very young stellar populations are dominating the dust heating ; the mid - infrared spectrum of ngc1377 , which suggests that dust nanoparticles have not been processed into classical aromatic components radiating in normal star - forming galaxies .
since the mass spectrum of stars able to heat the dust extends to much lower values than that of stars producing supernov , high infrared to radio flux ratios could be achieved if the initial mass function had a very steep slope , making high - mass stars rarer than in normal galaxies . in this hypothesis
, the extreme dust temperatures would however require extreme stellar densities . to be compatible with the 21 cm upper limit of ngc1377
, the cutoff of a truncated salpeter imf would have to be less than @xmath201 , since supernov have to be totally suppressed .
however , according to the model of @xcite , the @xmath2 ratio of ngc1377 implies an average radiation field about 400 times higher than the solar neighborhood field throughout the dust emission region . reaching this radiation field with only @xmath202 stars is impossible to reconcile with the data , as we will show with a simplistic model assuming a uniform stellar distribution .
the ks - band image gives us a constraint on the stellar mass density .
the h@xmath203[nii ] map of @xcite , of better resolution than the infrared maps , delineates a central region of radius @xmath156 . using the model of @xcite , we computed the bolometric power per unit of stellar mass produced by a star formation episode lasting 10myr ( imf - truncated ) and by an older population of 1gyr , as well as their mass to light ratios in the ks band ( close to 1m@xmath204 ) , and then the resulting radiation field in the center as a function of the mass fraction of the younger population . using the value of the solar neighborhood radiation field given by @xcite , we find that the radiation field is high enough only if more than 80% of the mass is locked in the imf - truncated starburst , hence the mass of any old stellar population is negligible .
since this is unrealistic , an abnormal imf is unlikely to cause the observed synchrotron deficiency in ngc1377 . in ic1953 , the average radiation field intensity implied by the dust temperature is only 100 times the solar neighborhood value ( in fact more than 200 times in the circumnuclear region , according to the decomposition shown in fig .
[ sed ] ) , but the required stellar density is still much higher than that observed , unless more than 65% of the central mass had been generated by a recent starburst .
the hypothesis of a weak magnetic field remains difficult to rule out .
however , if the magnetic energy density of some galaxies happened to be 1040 times lower than in normal galaxies , it would be difficult to explain why this is so rare and why this deficiency is connected with high dust temperatures , which are observed in all high-@xmath32 galaxies , unless it is causally linked with the occurrence of a starburst .
if the magnetic pressure is unusually low , then cloud collapse is more efficient and a starburst can be triggered at relatively low gas densities ; however , the question remains of what caused the low magnetic pressure , and why the starburst was not triggered before .
it is also possible , if magnetic fields are usually amplified by a starburst due to increased turbulence , that there is a time delay for the adjustment of the magnetic field ; this would become a likely scenario if we detected significant shock excitation by supernov by means of the [ feii ] emission line .
note however that the radio spectra of ngc4491 , ngc4418 and the disk of ic1953 bring evidence of a normal magnetic field strength , because they show substantial synchrotron emission , and that no signs of supernova shocks exist yet for ngc1377 .
random magnetic fields of a few @xmath1 g , about half the typical strength of total magnetic fields in spiral galaxies , are expected from turbulence alone , even in the absence of large - scale dynamo action .
field strengths of several @xmath1 g are easily obtained in elliptical galaxies although they are devoid of rotation @xcite , comparable to strengths in spirals .
unbiased statistical information about synchrotron emission of ellipticals is uneasy to find , but they are much weaker radio emitters than spirals , and their activity is mostly due to seyfert nuclei @xcite ; their star formation rates are too low to produce enough cosmic rays .
magnetic field amplification by large - scale dynamo processes occurs on timescales as large as @xmath205 years @xcite .
however , the growth of random fields requires much less time . depending on the turbulence characteristics , in particular the cut - off scale , the fastest - growing modes in a predominantly neutral medium may develop on timescales as small as @xmath78 years , according to @xcite . as pointed out by @xcite , the fact that quiescent galaxies and starbursts follow the same infrared - radio correlation , despite the large expected differences in inverse compton losses , suggests that the total magnetic field grows subsequent to a starburst .
if inverse compton losses are dominant over synchrotron losses , the lifetime of relativistic electrons is inversely proportional to the radiation energy density . assuming a size of 100pc for the infrared - emitting region of ngc1377 ( see sect .
[ n1377 ] ) , the radiation energy density integrated between 3 and 120@xmath1 m , at the surface of the region , is of the order of @xmath206jm@xmath207 . in the ultraluminous compact starbursts studied by @xcite ,
the typical radiation energy density is estimated to be six times higher than this , and yet they are strong radio continuum emitters , so that synchrotron losses must at least balance inverse compton losses : mechanisms at work to amplify magnetic fields seem to conspire to produce a constant ratio of magnetic energy density to radiation energy density .
this must occur on timescales comparable to or shorter than the growth time of the supernova rate , otherwise significant deviations from the correlation would be observed .
the issue of magnetic field strength is unsettled for ngc1377 , but we would like to stress that a weak field is not required to explain the synchrotron deficiency , since it is easily achieved by lack of cosmic rays , consistent with the fact that no supernova activity is observed .
furthermore , a separate cause has to be sought to explain the absence of free - free emission , which is unaffected by the magnetic field strength .
alternatively , if the magnetic field in a galaxy were amplified beyond normal values , it would cause more intense radio emission for the same injected population of cosmic rays , since the synchrotron emissivity scales as a large power of the field strength ( index between 2 and 4 ) . as a result ,
the infrared - to - radio ratio would decrease , but the radio emission would fade faster after an episode of star formation . if the magnetic field became so strong that the cosmic rays decayed very fast , the synchrotron emission would appear in intense bursts followed by periods of radio quiescence . in a steady - state situation , allowing some uncertainty range for the imf slope , one would expect from the infrared luminosity of ngc1377 a type - ii supernova explosion , on average , at least every 100 years . thus , cosmic rays should be injected into the interstellar medium , after a time delay , at this average rate . for the lifetime of cosmic - ray electrons to be reduced to 100 years or less at ghz frequencies , a magnetic field of the order of 50 mg would be required over large portions of the interstellar medium @xcite , that is to say @xmath208 times higher than the typical strength in spiral galaxies @xcite .
no amplification mechanism is known to achieve such a result on large spatial scales , and such a strong field would clearly not be confined @xcite . invoking such absurd values of the magnetic field is much less credible than the nascent starburst hypothesis . as a comparison , @xcite estimate fields strengths up to 350@xmath1 g in the radio - loud galaxy cygnus a , using the assumption of minimum energy density for the sum of cosmic rays and magnetic fields .
starburst galaxies may have magnetic fields of a few 10@xmath1 g , generally calculated under the same mimimum - energy assumption ; for instance , @xcite derived an equipartition total field of 50@xmath1 g in m82 .
we can discard high opacity at 21 cm as the reason for radio weakness , because it would require ngc1377 and ic1953 to have respectively @xmath74(21@xmath209 and @xmath74(21@xmath210 , only in order to have an intrinsic @xmath32 of @xmath211 , and ngc4491 would have to feature an unrealistic geometry .
high opacities are very unlikely , since the high-@xmath32 ultraluminous compact starbursts studied by @xcite ( with @xmath212 ) , much more active and dense , have estimated opacities of only 0.50.7 .
the infrared - radio properties of ngc1377 and ngc4491 may be suggestive of the presence of active nuclei in these galaxies .
however , spectral diagnostics do not support this view , since emission lines are either absent or very weak and narrow .
two indirect arguments can be put forward to support the presence of a seyfert nucleus : warm infrared colors , and the compactness of the infrared emission .
we consider the argument of deep silicate absorption in the 911@xmath1 m range in ngc1377 speculative , especially since the continuum emission longward of 11@xmath1 m has not been observed , and because it would be difficult to reconcile with the moderate dust to stellar emission power ratio ; diagnostics based on optical spectroscopy can not be applied .
however , the above two characteristics support equally well the hypothesis of a young starburst .
the mid - infrared spectrum is an extremely ambiguous diagnostic , since it does not show the aromatic bands associated with star formation , but does show an unusual emission feature , in place of the continuum expected in a pure seyfert nucleus . concerning the infrared broadband colors , @xcite , for instance , have shown that they are not valid discriminants between starburst and seyfert activities even if a larger percentage of energetically - dominating seyfert galaxies is found in any sample pre - selected to have warm colors , than in a sample with cooler colors .
the concentration factor is not a strong argument either , since a young starburst will be compact ( to trigger a star formation burst instantaneously over a large area requires uncommon conditions ) ; the data for ngc1377 suggest that the size could be as small as @xmath0pc , i.e. 1(sect .
[ n1377 ] ) , and is smaller than @xmath213 , i.e. 200pc ( sect . [ n1377_pifs ] ) .
in addition , seyfert nuclei powerful enough to dominate dust heating tend to occur in massive galaxies with a deep potential well , whereas ngc1377 is a low - mass galaxy and with a shallow potential .
but perhaps the most problematic fact for seyfert activity is the radio deficiency .
many active nuclei are known to be radio - quiet . however , their @xmath32 ratios are the same as in normal galaxies .
for instance , the infrared - loud and radio - quiet quasar studied by @xcite has an observed @xmath214 , even if it is strongly absorbed at 20 cm . in the sample of @xcite , containing a large number of radio - quiet active nuclei ,
no nucleus has a @xmath32 ratio significantly higher than the average for normal galaxies ; all the seyfert - type galaxies which are not detected at 20 cm are not detected by iras either .
we have further verified this fact by selecting in the ned database all galaxies in the quasar category with @xmath215jy and no mention of a radio association ; on the 53 objects thus selected , 7 were excluded for lack of any radio observations , and all the others have either normal or low @xmath32 ratios .
the infrared brightness of ngc1377 thus presents serious difficulties for the hypothesis of a radio - quiet active nucleus .
an examination of the energetics provides another argument against the production of the infrared power of ngc1377 by an active nucleus . assuming that the radiated power is equal to the eddington limit , @xmath216 ( m@xmath12 ) , and that it goes entirely into dust heating , an upper limit on the black hole mass provides an upper limit on the infrared power .
for a given black hole mass , the radio loudness can vary considerably ( e.g. @xcite ) ; but for radio - quiet active nuclei exclusively , estimating black hole masses from reverberation mapping techniques , @xcite found a correlation between the black hole mass and the radio power , with a lower envelope of all their data of the form @xmath217 . using the 20 cm upper limit for ngc1377
, the alleged black hole can not be more massive than @xmath218m@xmath12 .
using alternatively the kinematic information derived from the h@xmath116 emission line ( sect .
[ n1377_pifs ] ) , another limit can be set . with a velocity gradient below 40kms@xmath10 ,
the inclination - corrected rotation velocity is less than 25kms@xmath10 ( with a signal to noise ratio of @xmath156 throughout the central 2 ) .
assuming virial equilibrium , and provided the h@xmath116 lies in the plane of the galactic disk , we then do not expect more than @xmath219m@xmath12 to be enclosed in the central parsec .
using the first upper limit , we thus find that the maximum infrared power ( in extremely unrealistic conditions ) radiated by an active nucleus is @xmath220w .
in fact , it is expected to be a factor of several below this value , because the dust heating efficiency can not be 100% .
the infrared luminosity of ngc1377 being @xmath221w , an active nucleus can not easily account for it .
the absence of synchrotron emission might be attributed to lack of magnetic field ( sect .
[ magfield ] ) , or extreme inverse compton losses , due to the high radiation field intensity in ngc1377 .
nevertheless , an active nucleus must also produce some free - free emission , which is not detected either . inside the region where dust grains are sublimated ,
the lyman continuum photons are free to ionize the gas , contrary to the situation in an embedded starburst .
@xcite have compiled spectral energy distributions of quasars and estimated , in particular , the ionizing photon flux and the total power between 1000 and 100@xmath1 m .
the ratio between these two quantities is @xmath222s@xmath10w@xmath10 ( or @xmath223 ) . taking the infrared luminosity of ngc1377 as a lower limit to @xmath224 and assuming that the ionizing photon flux available in the dust - free region is equal to @xmath225 , the free - free emission at 3.6 cm , where thermal opacity is negligible , should amount to @xmath226mjy ( with @xmath227k ) .
it would thus have been detected , especially by the sensitive observation with the vla in the a configuration .
the single - dish upper limit is 2mjy , and the vla upper limit is 0.13mjy .
although we can not rule out completely the presence of an active nucleus , because the properties of ngc1377 remain challenging , the above arguments rather support our hypothesis of a nascent starburst .
the properties of ngc4491 may in turn be indicative of non - stellar activity , because some active nuclei are known to have flat radio spectra .
dust heating by an active nucleus is however very unlikely , owing to the following facts : ngc4491 has no bulge ; the velocity gradient inferred from the pa@xmath113 line ( see section [ nirspec ] ) is much less than 100kms@xmath10 , which confirms that there is no central gravitational cusp ( the inclination on the line of sight is about 62 ) ; optical and near - infrared emission lines do not give any indication of an active nucleus ; the dust spectral energy distribution is fully consistent with the properties of star - forming galaxies , whereas one would expect a high mid - infrared to far - infrared power ratio if dust were heated by an active nucleus .
other high-@xmath32 galaxies , deviating by more than @xmath127 from the average infrared to radio flux ratio , are known , though ngc1377 is the most extreme case . among the sample studied by @xcite in a series of papers , comprised of galaxies
whose flux density is higher at 60@xmath1 m than at 100@xmath1 m , 31 lie in the nvss sky coverage ( including ngc1377 ) ; of these 31 , we find that five galaxies satisfy the above high-@xmath32 criterion .
of the 40 very luminous galaxies belonging to the iras bright galaxy sample ( @xmath228jy and log(@xmath229 ) , @xcite have found three more high-@xmath32 galaxies , which are distant ( 200 to 300mpc ) and whose flux densities peak at 60@xmath1 m ; they classify them as compact starbursts .
the activity of two of them is obviously driven by on - going mergers , with double nuclei separated by less than 5 @xcite , and the third galaxy presents a strong warp @xcite .
however , @xcite claim that their radio deficiency is best explained by high free - free opacity at 20 cm , rather than a lack of cosmic rays .
nine galaxies with @xmath230 are mentioned by @xcite in a complete sample of 1809 galaxies with @xmath231jy ( i.e. with an occurrence rate of 0.5% ) , two of which in common with @xcite . however , reexamining these galaxies , we find that five of them are false high-@xmath32 detections , due partly to confusion affecting iras fluxes , and partly to nvss catalog fluxes missing low - surface brightness or extended structures . on the other hand , @xcite have missed ngc1377 , ngc4491 and two galaxies of the sample of @xcite , which , according to the selection criteria of @xcite , must be part of their sample and have @xmath230 .
because of inevitable errors due to automatic handling of very large samples , the census of high-@xmath32 galaxies they provide is therefore uncertain .
the two new high-@xmath32 galaxies that they found have @xmath2 ratios of @xmath232 and 1.44 . in view of the above numbers
, it seems that the most efficient way of finding synchrotron - deficient galaxies is to select high dust temperatures ( @xmath233 for all known high-@xmath32 galaxies ) , and that the infrared luminosity matters much less .
the occurrence rate in the color - selected sample of @xcite is as high as 16% .
60@xmath1m - peakers being scarce objects , however , the implied probability for a random local galaxy to be synchrotron - deficient must be very low .
the flux - limited sample of @xcite and @xcite ( 500 galaxies with @xmath228jy ) contains at least 6 high-@xmath32 galaxies identified in the above various samples , which implies a rate of the order of 1% .
the incidence of high dust temperatures is an additional argument in favor of the nascent starburst hypothesis that we have developed for ngc1377 , ic1953 and ngc4491 . in this scenario ,
the scarcity of high-@xmath32 galaxies derives naturally from the timescales involved .
the likelihood for a galaxy to have a high @xmath32 ratio is the product of the probabilities of two occurrences : it has to have been quiescent for at least @xmath31 years , and a starburst has to have started less than 34myr before the epoch of observation .
less than @xmath234% of optically - selected galaxies have infrared color excesses indicative of on - going intense star formation , as apparent from the analysis of @xcite ( of their `` hp sample '' ) ; from this we may infer that galaxies spend , on average , less than 15% of the time in starburst events .
given a typical starburst duration of @xmath235myr , with @xmath236 , the second probability listed above is necessarily lower than @xmath237 .
while the probability of the first occurrence ( quiescence for @xmath31 years ) is difficult to assess , it is sufficiently unlikely that high-@xmath32 systems should be rare , consistent with the statistics .
however , at high redshifts , before the epoch of maximum star formation density , the probability of observing radio - deficient galaxies may be higher .
it is unknown beyond what redshift metallicity effects , reducing the amount of dust and thus of infrared emission , will start to compete with the nascent starburst effect .
@xcite recently estimated @xmath32 ratios of distant galaxies ( @xmath238 up to 1.3 ) and found no significant deviation of the average from the local value .
the data shown in this paper are consistent with the high infrared to radio flux ratios being caused by a delay in either the magnetic field strength or the cosmic ray density , in either case a consequence of a nascent starburst .
in contrast to the above infrared - luminous galaxies , ngc1377 , ngc4491 and ic1953 have a shallow potential well , such that the triggering mechanism of a central starburst is not obvious . in ic1953 , tidal interaction ,
likely at work , could be responsible for strengthening the bar perturbation ( which seems unusually strong for the morphological type of ic1953 ) and causing massive gas inflow due to the bar potential .
in this view , we would be witnessing the first burst caused by the bar , and ic1953 will likely have an earlier - type aspect afterwards .
ngc4491 is also strongly barred , and this bar may also have grown due to tidal forces with other virgo galaxies .
however , no evidence of a bar nor of an intrusion exists in the most extreme galaxy of the three , ngc1377 .
are tidal interactions experienced by galaxies in clusters and groups efficient at producing gas inflows inside them , even in the absence of any visible potential perturbation ?
other hi - deficient and non - seyfert virgo galaxies , besides ngc4491 , indeed have surprisingly high mid - infrared colors in their centers : n4293 , n4569 , n4579 @xcite .
note that all these galaxies either lie close to m87 in projection and have radial velocities more than 600kms@xmath10 away from the velocity centroid @xcite , or lie at the periphery of the cluster in projection , which could mean that they are part of a very turbulent subsystem .
@xcite found such an environmental effect in the virgo unrelaxed subsystem called w , where blue compact dwarf galaxies are more numerous and have hotter dust than in the rest of the cluster .
hidden perturbations are best revealed by observing the hi morphology and velocity structure , but ngc1377 seems to have too little hi gas for such a diagnostic to be applied .
an assessment of the nature and cause of the activity seen in the three galaxies will have to wait for complementary data .
in order to understand the cause of the synchrotron deficiency observed in the three closest high-@xmath32 galaxies , we have presented and analysed extensive data in several windows : broadband dust emission , radio continuum at several frequencies , emission line diagnostics of photon ionization , supernova shocks and hot molecular gas excitation , and emission of the cold molecular gas .
we argued that the galaxies are seen at the very onset of a starburst episode , following a long period of quiescence , based on these data , and consistent with the statistics of high-@xmath32 galaxies inferred from starburst timescales . in this scenario ,
cosmic rays generated by past star formation had the time to decay , and massive stars produced by the current star formation episode have not evolved into supernov yet ( ngc1377 , central region of ic1953 ) or , in the case of ngc4491 , the supernova rate is still rising and the cosmic ray population is in its formative stages . all three galaxies host unusually hot dust , requiring an intense radiation field consistent with that produced by a young starburst .
ngc1377 , whose properties are the most peculiar , is not only devoid of synchrotron radiation , but also of free - free emission .
we propose that the starburst is so young that massive stars are still accreting material , preventing the development of hii regions . along the same line
, dust emission in the mid - infrared has an unusual spectral signature , suggesting that grains were processed differently from grains found in almost every other star formation environment ; the molecular gas disk may be overcritical , which would explain the quasi - instantaneous trigger of a starburst . while this is the most likely explanation ,
the properties of ngc1377 remain very intriguing .
we can rule out abnormal initial mass function , electron opacity and fast synchrotron losses as causes of the depressed radio continuum .
other scenarios are not formally ruled out , but create serious difficulties : radio - quiet seyfert activity is unlikely owing to the energetics arguments discussed in section [ seyfert ] ; a weak magnetic field ( relevant only for ngc1377 ) is possible , but difficult to explain in light of what is known of turbulent field amplification , and still requires a young starburst after a period of quiescence .
we would like to stress that the galaxies studied in this paper are not akin to irregular or blue compact dwarf galaxies , which would at first sight be easily susceptible of synchrotron deficiency for various reasons : fast cosmic ray escape and interstellar medium disruption and loss ; evidence of optically - thick free - free emission in some of them .
although these galaxies seem to have much more dispersed infrared to radio flux ratios than spirals , they follow the same average relationship and none of them is known to deviate from it significantly . among galaxies of the iras bright galaxy sample , selected by their 60@xmath1 m flux ,
the occurrence rate of high-@xmath32 galaxies ( in our interpretation nascent starbursts ) is of the order of 1% or slightly more .
the best way to select these galaxies is to search for high dust temperatures ; no other simple criterion is evident .
although very few examples are known , from this phenomenon arises a potential problem for the identification of distant infrared galaxies with high - resolution radio observations , especially with possibly changing star formation timescales at high redshifts .
the galaxies studied here offer a unique glimpse at starbursts during a critical period of their evolution , and will allow us to address questions concerning the pre - starburst accumulation of molecular material , the triggering mechanisms , the spatial development of a starburst , and the start , build - up and evolution of a cosmic ray population .
we are grateful to antonio garca - barreto for providing us his ( h@xmath33+[nii ] ) and i - band maps of ic1953 , to sangeeta malhotra for mentioning the peculiar infrared properties of ngc4418 , thereby arousing our curiousity , to the persons who assisted us in the observations at effelsberg , palomar and la silla , and to james brauher for communicating us some fluxes in advance of publication .
this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and ipac / caltech , funded by nasa and the national science foundation .
the isocam data presented in this paper were analyzed using and adapting the cia package , a joint development by the esa astrophysics division and the isocam consortium .
this research also benefited from the nasa / ipac extragalactic database ( ned ) , which is operated by the jet propulsion laboratory of caltech .
in february , exposures times were 300s for all lines but [ feii ] in ngc4491 , for which it was increased to 400s .
except for the beginning of feb .
1 , the sky was very clear , and the observations were photometric .
star number 9148 in @xcite was used as a photometric standard . in october , exposure times for ngc1377 and ngc1022 were respectively 260 and 220s for all lines ; the flux calibration was performed with 2mass images .
the field of view of pifs is sliced into eight slits , 0.67-wide and 9.7-long .
we dithered to reconstruct cross - slit spatial information , and obtained maps of @xmath239 , with a pixel size of 0.167 .
the high - resolution grating yields resolutions between 950 and 1500 at the wavelengths of interest .
observations were carried out as outlined in the pifs manual ( which can be found in _
btsoifer / pifs/ _ ) .
the flat - field and atmospheric transmission function was calibrated by sweeping g0v
g8v stars across the array with the chopping secondary , using a fast triangle wave , at the same airmass as the galaxy to within 10% .
the same stars were chopped in the cross - slit direction in order to establish a flux reference and to calibrate spatial curvature .
noble gas spectra served to correct for spectral distorsion and fit the wavelength function .
the data reduction consists of the following steps : ( 1 ) define a mask for bad pixels and cosmic rays ; ( 2 ) subtract the bias for each quadrant of the array ; ( 3 ) subtract the sky , tying the oh line intensities in the off - source spectra to their intensities in the on - source spectra ( in the edges of the field , where the contribution from the galaxy is minimal ) ; ( 4 ) divide spectra by the flat - field and atmospheric function ; ( 5 ) construct a wavevelength map using the noble gas spectra ; ( 6 ) correct for the spectral and spatial distortions , and combine the rectified spectra in a cube . table [ pifs ]
offers a summary of observing parameters and measured fluxes . in order to minimize residuals from oh sky lines in the faint spectra of ngc4491
, we subtracted the spectral variations of an off - center region around its average flux density .
we used dual beam switching , and connected the iram115/230 receivers to the low - resolution acousto - optical spectrometers .
the initial velocity resolution was 3.6kms@xmath10 and 1.8kms@xmath10 , respectively .
system temperatures ranged between 200k and 400k at 230ghz and between 230k and 460k at 115ghz .
the beam width is 45at 115ghz and 23 at 230ghz .
baseline fitting and removal in each separate spectrum , and averaging of spectra obtained at the same position were performed with the class software , which is developed by the observatoire de grenoble and iram .
we used main beam efficiencies of 0.7 at 115ghz and 0.5 at 230ghz , and a forward efficiency of 0.9 .
pointing corrections were determined on the closest bright sio masers , o ceti and orion sio , twice per observing session , at the beginning and after sunset .
these sources are respectively about 15 and 30away from ngc1377 in azimuth , and about 15 away to the north . due to these large distances ,
the pointing accuracy is likely degraded with respect to the quoted value of 3 rms in azimuth and elevation .
furthermore , a hardware limit obliged us to rotate the antenna by 360 in azimuth once per night ; because the co(2 - 1 ) spectrum obtained on the center of ngc1377 before the rotation was not reproduced within the error bar after the rotation , we suspect that this introduced additional pointing errors , and we discarded the redundant data obtained after rotating the antenna . in order to know whether the co source is extended , we performed minimal mapping , integrating on the galaxy center and eight adjacent positions stepped by half the co(2 - 1 ) beam width .
the spectra are shown in figure [ co_maps ] .
the resulting eighteen total flux measurements were least - squares - fitted with a gaussian source of adjustable width , allowing for some random pointing errors of up to 3 at each position , plus some systematic larger pointing errors after each large rotation of the antenna . if no systematic errors are allowed , then the fit is not acceptable ( normalized @xmath240 ) .
we can reproduce the data provided these systematic pointing errors are of the order of 5 or slightly more .
the best fit is then always obtained for the smallest source size of the explored grid . due to the large beams ,
this parameter is not well constrained , but the derived total fluxes are robust results .
barnes d.g . , staveley - smith l. , de blok w.j .
et al . , 2001 ,
mnras 322 , 486 beck r. , 2001 , ssrv 99 , 243 beck r. , brandenburg a. , moss d. , shukurov a. & sokoloff d. , 1996 , ara&a 34 , 155 beck r. & golla g. , 1988 , a&a 191 , l9 beck s.c . , turner j.l . & kovo o. , 2000 , aj 120 , 244 beichman c.a . , soifer b.t .
, helou g. et al . , 1986 , apj 308 , l1 bianchi s. , davies j.i . & alton p.b . , 1999 ,
a&a 344 , l1 bicay m.d . & helou g. , 1990 , apj 362 , 59 binggeli b. & cameron l.m . , 1993 ,
a&as 98 , 297 binggeli b. , popescu c.c . & tammann g.a . , 1993 ,
a&as 98 , 275 boselli a. , casoli f. & lequeux j. , 1995 , a&as 110 , 521 brauher j. , 2002 , in preparation burrows d.n . , singh k.p .
, nousek j.a . , garmire g.p . & good j. , 1993 , apj 406 , 97 cagnoni i. , della ceca r. & maccacaro t. , 1998 , apj 493 , 54 cardelli j.a . ,
clayton g.c . &
mathis j.s . , 1989 , apj 345 , 245 carilli c.l .
& barthel p.d . , 1996 ,
a&arv 7 , 1 cayatte v. , van gorkom j.h . , balkowski c. & kotanyi c. , 1990 , aj 100 , 604 chapman s.c . , helou g. , lewis g.f .
& dale d.a . , 2002 ,
submitted to apj clark d.h . & caswell j.l . , 1976 ,
mnras 174 , 267 colina l. , bohlin r.c . & castelli f. , 1996 , aj 112 , 307 combes f. , prugniel p. , rampazzo r. & sulentic j.w . , 1994 , a&a 281 , 725 condon j.j .
, cotton w.d . &
broderick j.j . , 2002a , aj 124 , 675 condon j.j . ,
helou g. & jarrett t.h .
, 2002b , aj 123 , 1881 condon j.j . , cotton w.d . , greisen e.w .
et al . , 1998 , aj 115 , 1693 condon j.j .
helou g. , sanders d.b .
& soifer b.t . , 1996 ,
apjs 103 , 81 condon j.j .
, helou g. , sanders d.b .
& soifer b.t . , 1993 , aj 105 , 1730 condon j.j . , 1992 , ara&a 30 , 575 condon j.j . , huang z.p . , yin q.f . & thuan t.x . , 1991 ,
apj 378 , 65 condon j.j . , anderson m.l .
& helou g. , 1991 , apj 376 , 95 condon j.j .
helou g. , sanders d.b .
& soifer b.t . , 1990 ,
apjs 73 , 359 condon j.j . &
yin q.f . , 1990 ,
apj 357 , 97 contursi a. , lequeux j. , cesarsky d. et al . , 2000 ,
a&a 362 , 310 dale d.a . ,
roussel h. , contursi a. et al .
, in preparation dale d.a .
, helou g. , contursi a. , silbermann n.a . & kolhatkar s. , 2001 , apj 549 , 215 dsert f.x .
, boulanger f. & puget j.l . , 1990 ,
a&a 237 , 215 devereux n.a . & eales s.a . , 1989 ,
apj 340 , 708 draine b.t . &
lee h.m . , 1984 ,
apj 285 , 89 draine b.t . , 1981 , apj 245 , 880 dressel l.l . & condon j.j . , 1978 , apjs 36 , 53 drge w. , lerche i. & schlickeiser r. , 1987 , a&a 178 , 252 dudley c.c . & wynn - williams c.g . , 1997 ,
apj 488 , 720 dwek e. , foster s.m .
& vancura o. , 1996 , apj 457 , 244 eales s.a .
, becklin e.e . ,
hodapp k.w . , simons d.a .
& wynn - williams c.g . , 1990 , apj 365 , 478 elvis m. , wilkes b.j . , mcdowell j.c .
et al . , 1994 , apjs 95 , 1 emerson d.t . & grve r. , 1988 , a&a 190 , 353 engelmann j.j . , ferrando p. , soutoul a. et al . , 1990 ,
a&a 233 , 96 fermi e. , 1954 , apj 119 , 1 fomalont e.b . ,
kellermann k.i . ,
partridge r.b . ,
windhorst r.a . &
richards e.a . , 2002 , aj 123 , 2402 frster - schreiber n.m .
& roussel h. , in preparation gabel j.r . & bruhweiler f.c . , 2002 , aj 124
, 737 garay g. , brooks k.j . , mardones d. , norris r.p .
& burton m.g . , 2002 , apj 579 , 678 garay g. , mardones d. & mirabel i.f . , 1993 , a&a 277 , 405 garca - barreto j.a . , franco j. , carrillo r. , venegas s. & escalante - ramrez b. , 1996 , revmexaa 32 , 89 garrett m.a . , 2002 ,
a&a 384 , l19 genzel r. , lutz d. , sturm e. et al . , 1998 ,
apj 498 , 579 giovanelli r. , haynes m.p .
, herter t. et al . , 1997 , aj 113 , 22 gordon s.m . , duric n. , kirshner r.p .
, goss w.m . & viallefond f. , 1999 , apjs 120 , 247 de grijp m.h . , miley g.k .
& lub j. , 1987 , a&as 70 , 95 grosbol p.j . & patsis p.a . , 1998 , a&a 336 , 840 guiderdoni b. & rocca - volmerange b. , 1985 ,
a&a 151 , 108 guillois o. , nenner i. , papoular r. & reynaud c. , 1996 , apj 464 , 810 heisler c.a .
, de robertis m.m . &
nadeau d. , 1996 , mnras 280 , 579 heisler c.a .
& vader j.p . , 1995 , aj 110 , 87 heisler c.a . &
vader j.p . , 1994 , aj 107 , 35 helou g. , malhotra s. , hollenbach d.j .
et al . , 2001 ,
apj 548 , l73 helou g. , lu n.y .
, werner m.w . ,
malhotra s & silbermann n , 2000 , apj 532 , l21 helou g. & bicay m.d .
, 1993 , apj 415 , 93 helou g. , hoffman g.l . &
salpeter e.e . , 1989 , in _ the world of galaxies _ , workshop help april 1988 in paris , ed .
corwin & bottinelli , p. 496 helou g. , khan i.r .
, malek l. & boehmer l. , 1988 , apjs 68 , 151 helou g. , 1986 , apj 311 , l33 helou g. , soifer b.t . & rowan - robinson m. , 1985 , apj 298 , l7 dhendecourt l.b .
& lger a. , 1987 , a&a 180 , l9 ho l.c . , 2002 ,
apj 564 , 12 hollenbach d. & natta a. , 1995 , apj 455 , 133 huang j.h . , gu q.s . , su h.j . ,
hawarden t.g .
, liao x.h . & wu g.x . , 1996 , a&a 313 , 13 hummel e. & beck r. , 1995 , a&a 303 , 691 hummel e. , davies r.d . , wolstencroft r.d . , van der hulst j.m .
& pedlar a. , 1988 , a&a 199 , 91 hunt l.k .
, giovanardi c. & helou g. , 2002 , a&a 394 , 873 hunter , d.a . , kaufman m. , hollenbach d.j .
et al . , 2001 ,
apj 553 , 121 jarrett t.h . , chester t. , cutri r. , schneider s. & huchra j. , 2003 , `` the 2mass large galaxy atlas '' , aj , in press jarrett t.h . ,
2000 , pasp , 112 , 1008 jog c.j .
& solomon p.m. , 1984 , apj 276 , 114 jogee s. & kenney j.d.p . , 1995 , baas 27 , 1356 de jong t. , klein u. , wielebinski r. & wunderlich e. , 1985 , a&a 147 , l6 kennicutt r.c . , 1998 , ara&a 36 , 189 kewley l.j . , heisler c.a .
, dopita m.a .
2000 , apj 530 , 704 kim d.c .
, sanders d.b . , veilleux s. , mazzrella j.m . &
soifer b.t .
, 1995 , apjs 98 , 129 klaas u. , haas m. , mller s.a .
et al . , 2001 , a&a 379 , 823 klein u. , weiland h. & brinks e. , 1991 , a&a 246 , 323 klein u. , wielebinski r. & morsi h.w . , 1988 , a&a 190 , 41 kobulnicky h.a .
& johnson k.e . , 1999 ,
apj 527 , 154 kodaira k. , watanabe m. & okamura s. , 1986 , apjs 62 , 703 koopmann r.a .
, kenney j.d . &
young j. , 2001 , apjs 135 , 125 kwok s. , volk k. & bernath p. , 2001
, apj 554 , l87 laureijs r.j . , watson d. , metcalfe l. et al . ,
2000 , a&a 359 , 900 laurent o. , mirabel i.f . , charmandaris v. et al . ,
2000 , a&a 359 , 887 leech k.j .
, vlk h.j . , heinrichsen i. et al . , 1999 ,
mnras 310 , 317 lehnert m.d .
& heckman t.m . , 1995 , apjs 97 , 89 leitherer c. , schaerer d. , goldader j.d .
et al . , 1999 ,
apjs 123 , 3 lesch h. & bender r. , 1990 , a&a 233 , 417 lisenfeld u. , vlk h.j . & xu c. , 1996 , a&a 314 , 745 lu n. , helou g. , werner m.w .
et al . , 2003 ,
apj 588 , 199 madden s.c . ,
2000 , new astr .
review 44 , 249 malhotra s. , kaufman m.j . , hollenbach d. et al .
, 2001 , apj 561 , 766 marsh k.a . & helou g. , 1995 , apj 445 , 599 martin j.m . ,
bottinelli l. , dennefeld m. , gouguenheim l. & le squeren a.m. , 1988 , a&a 201 , l13 mathis j.s .
, mezger p.g .
& panagia n. , 1983 , a&a 128 , 212 menon t.k . , 1991 ,
apj 372 , 419 miller n.a . & owen f.n . , 2001 , aj 121 , 1903 moorwood a.f .
& oliva e. , 1988 , a&a 203 , 278 moss d. & shukurov a. , 1996 , mnras 279 , 229 murphy t.w . , matthews k. & soifer b.t . , 1999 ,
pasp 111 , 1176 nelson c.h . , 2000 , apj 544 , l91 niklas s. , klein u. & wielebinski r. , 1997 , a&a 322 , 19 niklas s. & beck r. , 1997 , a&a 320 , 54 niklas s. , klein u. & wielebinski r. , 1995 , a&a 293 , 56 osorio m. , lizano s. & dalessio p. , 1999 , apj 525 , 808 park y.s . & hong s.s .
1995 , a&a 300 , 890 parker e.n . , 1966 ,
apj 145 , 811 pelletier g. & roland j. , 1988 , a&a 196 , 71 persson s.e .
, murphy d.c .
, krzeminski w. , roth m. & rieke m.j . , 1998 , aj 116 , 2475 radford s.j .
, solomon p.m. & downes d. , 1991 , apj 368 , l15 ridgway s.e . , wynn - williams c.g .
& becklin e.e . , 1994 ,
apj 428 , 609 rigopoulou d. , spoon h.w.w .
, genzel r. , lutz d. , moorwood a.f.m . & tran q.d .
, 1999 , aj 118 , 2625 roussel h. , sauvage m. , vigroux l. & bosma a. , 2001c , a&a 372 , 427 roussel h. , sauvage m. , vigroux l. , bosma a. et al . , 2001b , a&a 372 , 406 roussel h. , vigroux l. , bosma a. , sauvage m. et al . , 2001a , a&a 369 , 473 roy a.l . , norris r.p . , kesteven m.j . , troup e.r
& reynolds j.e . , 1998 ,
mnras 301 , 1019 sanders d.b .
, scoville n.z . &
soifer b.t . , 1991 , apj 370 , 158 sanders d.b .
& mirabel i.f . , 1985 ,
apj 298 , l31 scoville n.z .
, evans a.s . , thompson r. et al . , 2000 ,
aj 119 , 991 sodroski t.j . , bennett c. , boggess n. et al . , 1994 ,
apj 428 , 638 sopp h.m . & alexander p. , 1991
, mnras 251 , sc 14 spoon h.w .
, keane j.v .
, tielens a.g . , lutz d. & moorwood a.f . , 2001 , a&a 365 , l353 storey p.j
& hummer d.g . , 1995 ,
mnras 272 , 41 strong a.w . , bloemen j.b .
, dame t.m .
et al . , 1988 , a&a 207 , 1 strong a.w . , 1977 ,
mnras 181 , 311 subramanian k. , 1998 , mnras 294 , 718 sulentic j.w . & de mello rabaca d.f . , 1993 ,
apj 410 , 520 tarchi a. , neininger n. , greve a. et al . , 2000 , a&a 358 , 95 toomre a. , 1964 , apj 139 , 1217 tran q.d . ,
lutz d. , genzel r. et al . , 2001 , apj 552 , 527 turner j.l .
, ho p.t . & beck s.c . , 1998 , aj 116 , 1212 ulvestad j.s . , 1982 ,
apj 259 , 96 vader j.p . ,
frogel j.a . , terndrup d.m . &
heisler c.a . , 1993 , aj 106
, 1743 vanzi l. & rieke g.h . , 1997 ,
apj 479 , 694 wallace j.m . , 1980 , ap&ss 68 , 27 walmsley m. , 1995 , revmexaa s.c . 1 , 137 walsh d.e . ,
knapp g.r .
, wrobel j.m . & kim d.w . , 1989 ,
apj 337 , 209 welch g.a . &
sage l.j . , 2003 ,
apj 584 , 260 willmer c.n .
, focardi p. , da costa l.n .
& pellegrini p.s . , 1989 , aj 98
, 1531 woo j.h . &
urry c.m . , 2002 ,
apj 579 , 530 wunderlich e. & klein u. , 1988 , a&a 206 , 47 yun m.s . , reddy n.a .
& condon j.j . , 2001 ,
apj 554 , 803 lcccclc ngc1377 & 03 36 39.1 @xmath24120 54 07 & eridanus & 21 & @xmath242 & amorphous & 1.78 + ic1953 & 03 33 41.8 @xmath24121 28 43 & eridanus & 21 & @xmath243 & sbd & 2.75 + ngc4491 & 12 30 57.2 @xmath24411 28 59 & virgoa & 17 & @xmath9 & sba dwarf & 1.70 + lcccccccc ngc1377 & 5.97 & 7.27 & 1.92 & @xmath245 & 4.3 & 3.3 & 1.22 & @xmath246 ( @xmath24 ) + ic1953 & 11.62 & 9.01 & 1.05 & 13.00 & 6.1 & 1.6 & 0.78 & 2.95 ( @xmath26 ) + ngc4491 & 3.45 & 2.68 & 0.43 & @xmath247 & 1.2 & 1.6 & 0.78 & @xmath248 ( @xmath25 ) + lccccccc ngc1377 & @xmath249 & @xmath250 & 0.22 & 0.88 & @xmath251 & 343 & @xmath252 + ic1953 & @xmath253 & @xmath254 & 0.09 & 0.79 & 116 & 34 & 3.4 + ngc4491 & @xmath255 & @xmath256 & 0.10 & 0.46 & 73 & 18 & 4.0 + lllll beam size & 82 & 147 & 45 & + ngc1377 & @xmath257 & @xmath258 & @xmath259 & + ic1953 & @xmath86 & @xmath260 & 13.0 & @xmath261 + ngc4491 & @xmath94 & @xmath262 & @xmath263 & @xmath264 + ngc4418 & @xmath191 & @xmath190 & @xmath265 & @xmath266 + rlccccc + pa@xmath113 & feb . 1 & 80 & @xmath267 & @xmath268 & 18.9 & @xmath269 + pa@xmath113 & & & & @xmath270 & 3.09 & @xmath119 + @xmath271feii@xmath272 $ ] & feb . 2 & 133 & 1.11.3 & @xmath273 & 1.66 & @xmath119 + br@xmath114 & feb . 2 & 40 & 1.42.0 & @xmath274 & 0.95 & @xmath119 + + pa@xmath113 & feb . 1 & 20 & 1.11.2 & @xmath275 & 131.8 & @xmath121 + @xmath271feii@xmath272 $ ] & feb . 1 & 20 & @xmath267 & @xmath276 & 123.6 & @xmath121 + br@xmath114 & feb
2 & 20 & 1.61.8 & @xmath277 & 117.2 & @xmath121 + + pa@xmath113 & oct .
17 & 26 & 1.71.8 & @xmath273 & 4.51 & @xmath119 + @xmath271feii@xmath272 $ ] & oct .
17 & 26 & 1.71.8 & @xmath278 & 3.96 & @xmath119 + br@xmath114 & oct .
18 & 39 & 1.71.9 & @xmath279 & 3.05 & @xmath119 + [email protected] & oct .
18 & 35 & 1.71.8 & @xmath280 & 7.34 & @xmath281 + [email protected] & oct .
19 & 26 & 1.71.8 & @xmath282 & & @xmath119 + [email protected] & oct .
19 & 30 & 1.71.8 & @xmath283 & & @xmath119 + + pa@xmath113 & oct .
17 & 18 & 1.41.5 & @xmath284 & 13.9 & @xmath119 + @xmath271feii@xmath272 $ ] & oct .
17 & 15 & 1.31.4 & @xmath285 & 12.1 & @xmath119 + br@xmath114 & oct .
18 & 15 & 1.71.9 & @xmath286 & 8.46 & @xmath119 + [email protected] & oct .
18 & 40 & 1.41.7 & @xmath287 & & @xmath119 + [email protected] & oct .
19 & 15 & @xmath288 & @xmath289 & & @xmath119 + llllllllll co(1 - 0 ) & @xmath297 & @xmath298 & @xmath299 & @xmath300 + & @xmath301 & @xmath302 & @xmath303 & @xmath304 & @xmath305 + co(2 - 1 ) & @xmath306 & @xmath307 & @xmath308 & @xmath309 + & @xmath310 & @xmath311 & @xmath312 & @xmath313 & @xmath314 + + + + + + | three nearby galaxies which have abnormally high infrared to radio continuum ratios , ngc1377 , ic1953 and ngc4491 , are investigated with a view to understanding the physical origin of their peculiarity .
we review the existing data and present new radio continuum measurements along with near - infrared integral - field spectroscopy and molecular gas observations .
the three galaxies have low luminosities but starburst - like infrared colors ; in ngc1377 , no synchrotron emission is detected at any wavelength ; in ic1953 , the observed synchrotron component is attributable to the spiral disk alone , and is lacking in the central regions ; the radio spectrum of ngc4491 is unusually flat .
we also compare and contrast them with ngc4418 , a heavily extinguished galaxy which shares some attributes with them . after examining various scenarios ,
we conclude that these galaxies are most likely observed within a few myr of the onset of an intense star formation episode after being quiescent for at least @xmath0myr .
this starburst , while heating the dust , has not produced optical signatures nor a normal amount of cosmic rays yet .
we briefly discuss the statistics of such galaxies and what they imply for star formation surveys . |
here we study the behavior of the polyakov loop by treating it as a heavy field that is a singlet under chiral symmetry transformations .
we take the underlying theory to be two colors and two flavors in the fundamental representation .
the degrees of freedom in the chiral sector of the effective theory are @xmath14 goldstone fields @xmath15 and a scalar field @xmath16 . for @xmath17
the potential is @xcite : @xmath18&=&\frac{m^2}{2}{\rm tr } \left[m^{\dagger}m\right]+ { \lambda_1}{\rm tr } \left[m^{\dagger}m\right]^2 + \frac{\lambda_2}{4}{\rm tr } \left[m^{\dagger}mm^{\dagger}m\right ] \label{chiralpot}\end{aligned}\ ] ] with @xmath19 , @xmath20 and @xmath21 .
@xmath22 are the generators provided explicitly in equation ( a.5 ) and ( a.6 ) of @xcite .
the polyakov loop potential in the absence of the @xmath9 symmetry is @xmath23=g_0\chi+\frac{m_\chi^2}{2}\chi^2+\frac{g_3}{3}\chi^3 + \frac{g_4}{4}\chi^4 \ , .
\label{chipot}\end{aligned}\ ] ] the field @xmath24 represents the polyakov loop itself . to complete the effective theory we introduce interaction terms allowed by the chiral symmetry @xmath25&= & \left(g_1\chi + g_2\chi^2\right){\rm tr } \left[m^{\dagger}m\right]=\left(g_1\chi + g_2\chi^2\right)(\sigma^2+\pi^a\pi^a ) \ , .\end{aligned}\ ] ] in the phase with @xmath26 , where chiral symmetry is spontaneously broken , @xmath16 acquires a nonzero expectation value , which in turn induces a modification also for @xmath27 .
the usual choice for vacuum alinement is in the @xmath16 direction , i.e. @xmath28 . the extremum of the linearized potential is at @xmath29 where @xmath30 . here
@xmath31 is the full coefficient of the @xmath32 term in the tree - level lagrangian which , due to the coupling between @xmath24 and @xmath16 , also depends on @xmath27 . spontaneous chiral symmetry breaking appears for @xmath33 . in this regime
the positive mass squared of the @xmath16 is @xmath34 .
the formulae in ( [ vevsigma ] ) hold near the phase transition where @xmath35 is small .
we have ordered the couplings such that @xmath36 and @xmath37 are both much greater than @xmath38 and @xmath39 . near the critical temperature
the mass of the order parameter field is assumed to posses the generic behavior @xmath40 .
equation ( [ vevchi ] ) shows that for @xmath41 and @xmath42 the expectation value of @xmath24 behaves oppositely to that of @xmath43 : as the chiral condensate starts to decrease towards chiral symmetry restoration , the expectation value of the polyakov loop starts to increase , signaling the onset of deconfinement .
this is illustrated in the left panel of figure [ figura1 ] . when applying the analysis presented in @xcite , the general behavior of the spatial two - point correlator of the polyakov loop can be obtained .
near the transition point , in the broken phase , the @xmath24 two - point function is dominated by the infrared divergent @xmath16-loop .
this is so , because the @xmath15 goldstone fields couple only derivatively to @xmath24 , and thus decouple .
we find a drop in the screening mass of the polyakov loop at the phase transition . when approaching the transition from the unbroken phase the goldstone fields do not decouple , but follow the @xmath16 , resulting again in the drop of the screening mass of the polyakov loop close to the phase transition .
we consider the variation of the @xmath24 mass near the phase transition with respect to the tree level mass @xmath44 defined above the chiral phase transition .
we define @xmath45 , where @xmath44 is the mass at a given temperature near the critical point . using a large @xmath46 framework motivated resummation @xcite
we deduce : @xmath47 from the above equations one finds that the screening mass of the polyakov loop is continuous and finite at @xmath48 , and @xmath49 , independent of @xmath50 , the number of pions .
this analysis is not restricted to the chiral / deconfining phase transition .
the entanglement between the order parameter ( the chiral condensate ) and the non - order parameter field ( the polyakov loop ) is universal .
as a second application , consider two color qcd with two massless dirac quark flavors in the adjoint representation . here
the global symmetry is @xmath51 which breaks via a bilinear quark condensate to @xmath52 .
the number of goldstone bosons is @xmath53 .
we take @xmath17 .
there are two exact order parameter fields : the chiral @xmath16 field and the polyakov loop @xmath54 . since the relevant interaction term @xmath55 is now forbidden
, one might expect no efficient information transfer between them .
this naive statement is supported by lattice data @xcite . while respecting general expectations the following analysis suggests the presence of a new and more elaborated structure which lattice data can clarify in the near future .
the chiral part of the potential is given by ( [ chiralpot ] ) with @xmath19 , @xmath56 and @xmath57 .
@xmath22 are the generators provided explicitly in equation ( a.3 ) and ( a.5 ) of @xcite .
the @xmath9 symmetric potential for the polyakov loop is @xmath23=\frac{m_{0\chi}^2}{2}\chi^2+\frac{g_4}{4}\chi^4 \ , , \end{aligned}\ ] ] and the only interaction term allowed by symmetries is @xmath58=g_2\chi^2\,{\rm tr } \left[m^{\dagger}m\right ] = g_2\chi^2(\sigma^2+\pi^a\pi^a ) \ , .\ ] ] here we review the case in which the chiral phase transition happens first , i.e. @xmath59 . for @xmath60 both symmetries are broken , and the expectation values of the two order parameter fields are linked to each other : @xmath61 the coupling @xmath62 is taken to be positive .
the expected behavior of @xmath63 and @xmath64 near @xmath65 and @xmath48 , respectively , combined with the result of eq .
( [ vevad ] ) , yields in the neighborhood of these two transitions the qualitative situation , illustrated in the right panel of figure [ figura1 ] . on both sides of @xmath65 the relevant interaction term @xmath66 emerges , leading to a one - loop contribution to the static two - point function of the @xmath16 field @xmath67 . near the deconfinement transition @xmath68 yielding an infrared sensitive screening mass for @xmath16 .
similarly , on both sides of @xmath48 the interaction term @xmath69 is generated , leading to the infrared sensitive contribution @xmath70 to the @xmath24 two - point function .
we conclude , that when @xmath71 , the two order parameter fields , a priori unrelated , do feel each other near the respective phase transitions .
it is important to emphasize that the effective theory works only in the vicinity of the two phase transitions .
interpolation through the intermediate temperature range is shown by dotted lines in the right panel of figure [ figura1 ] .
possible structures here must be determined via first principle lattice calculations .
the infrared sensitivity leads to a drop in the screening masses of each field in the neighborhood of the transition of the other , which becomes critical , namely of the @xmath16 field close to @xmath65 , and of the @xmath24 field close to @xmath72 .
the resummation procedure outlined in the previous section predicts again a finite drop for the masses : @xmath73 we thus predict the existence of substructures near these transitions , when considering fermions in the adjoint representation . searching for
such hidden behaviors in lattice simulations would help to further understand the nature of phase transitions in qcd .
we have shown how deconfinement ( i.e. a rise in the polyakov loop ) is a consequence of chiral symmetry restoration in the presence of fermions in the fundamental presentation . in nature quarks
have small , but nonzero masses , which makes chiral symmetry only approximate .
nevertheless , the picture presented here still holds : confinement is driven by the dynamics of the chiral transition .
the argument can be extended even further : if quark masses were very large then chiral symmetry would be badly broken , and could not be used to characterize the phase transition .
but in such a case the @xmath9 symmetry becomes more exact , and by reversing the roles of the protagonists in the previous discussion , we would find that the @xmath9 breaking drives the ( approximate ) restoration of chiral symmetry . which of the underlying symmetries demands and which amends can be determined directly from the critical behavior of the spatial correlators of hadrons or of the polyakov loop @xcite . with quarks in the adjoint representation we investigated the physical scenario @xcite in chiral symmetry
is restored after deconfinement set in , i.e. @xmath74 . in this case
we have pointed to the existence of an interesting structure , which was hidden until now : there are still two distinct phase transitions , but since the fields are now entangled , the transitions are not independent .
this entanglement is shown at the level of expectation values and spatial correlators of the fields .
lattice simulations will play an important role in checking these predictions .
the analysis can be extended for phase transitions driven by a chemical potential . for 2 colors
there is a phase transition from a quark - antiquark condensate to a diquark condensate @xcite .
we hence predict , in two color qcd , that when diquarks form for @xmath75 , the polyakov loop also feels the presence of the phase transition exactly in the same manner as it feels when considering the temperature driven phase transition .
such a situation is supported by recent lattice simulations @xcite .
the results presented here are not limited to describing the chiral / deconfining phase transition and can readily be used to understand phase transitions sharing similar features .
f. sannino , phys .
d * 66 * , 034013 ( 2002 ) [ arxiv : hep - ph/0204174 ] .
a. mocsy , f. sannino and k. tuominen , phys .
lett . * 91 * , 092004 ( 2003 ) [ arxiv : hep - ph/0301229 ] .
a. mocsy , f. sannino and k. tuominen , _ induced universal properties and deconfinement _ , arxiv : hep - ph/0306069 .
a. mocsy , f. sannino and k. tuominen , _ confinement versus chiral symmetry _
, arxiv : hep - ph/0308135 .
p. n. meisinger and m. c. ogilvie , phys .
d * 66 * , 105006 ( 2002 ) [ arxiv : hep - ph/0206181 ] .
f. wilczek , int .
j. mod .
phys . a * 7 * , 3911 ( 1992 ) [ erratum - ibid .
a * 7 * , 6951 ( 1992 ) ] .
o.scavenius , a.mocsy , i.n.mishustin and d.h.rischke , phys .
c * 64 * , 045202 ( 2001 ) .
s. holtmann and t. schulze , arxiv : hep - lat/0305019 . f. karsch and m. lutgemeier , nucl .
b * 550 * , 449 ( 1999 ) [ arxiv : hep - lat/9812023 ] . an incomplete list : g. e. brown
, a. d. jackson , h. a. bethe and p. m. pizzochero , nucl .
a * 560 * , 1035 ( 1993 ) ; s. digal , e. laermann and h. satz , nucl .
a * 702 * , 159 ( 2002 ) ; k. fukushima , arxiv : hep - ph/0310121 . n. o. agasian and s. m. fedorov , `` hadron resonance gas and nonperturbative qcd vacuum at finite temperature , '' arxiv : hep - ph/0310249 . t. appelquist , p. s. rodrigues da silva and f. sannino , phys .
d * 60 * , 116007 ( 1999 ) .
j. t. lenaghan , f. sannino and k. splittorff , phys .
d * 65 * , 054002 ( 2002 ) . for the wess - zumino anomaly terms see : z. y. duan , p. s. rodrigues da silva and f. sannino , nucl .
b * 592 * , 371 ( 2001 ) [ arxiv : hep - ph/0001303 ] . | the polyakov loop is the appropriate deconfinement order parameter for yang - mills theories without quarks or with quarks in the adjoint representation of the gauge group .
however it is not a physical state of the theory so the information regarding the center group symmetry must be transferred to the physical states .
we briefly review how this transfer of information takes place . when adding quarks the center group is no longer a symmetry for matter in the fundamental representation .
this feature allows us to explain why color deconfines when chiral symmetry is restored in hot gauge theories with massless quarks . for quarks in the adjoint representation
we show that while deconfinement and the chiral transition do not coincide , entanglement between them is still present . finally , we discuss also the chemical potential driven phase transition . recently in we have analyzed the problem of how , and to what extent the information encoded in the order parameter of a generic theory is transferred to the non - order parameter fields . this is a fundamental problem since in nature most fields are non - order parameter fields .
this problem is especially relevant in qcd and qcd - like theories since there is no physical observable for deconfinement which is directly linked to the order parameter field . at nonzero temperature
the @xmath0 center of @xmath1 is a relevant global symmetry @xcite , and it is possible to construct the polyakov loop @xmath2 operator .
this object is charged with respect to the center @xmath0 of the @xmath3 gauge group @xcite , under which it transforms as @xmath4 with @xmath5 .
a relevant feature of the polyakov loop is that its expectation value vanishes in the low temperature regime , and is non - zero in the high temperature phase .
the polyakov loop is thus a suitable order parameter for the yang - mills temperature driven phase transition @xcite .
the question is : how can the information about the yang - mills phase transition encoded in the @xmath0 global symmetry be communicated to the hadronic states of the theory . in , using an effective lagrangian approach , we have explicitly shown that is always possible on symmetry grounds to couple @xmath2 to the hadronic operators .
effective lagrangians play a relevant role . in @xcite , for example , a number of fundamental properties for qcd were derived .
of course , at a more elementary level the picture is that the gauge invariant operators of the theory carrying center group charges couple to hadronic states .
if not protected by symmetries only an unnatural act would require these relevant couplings to vanish .
such a picture is also confirmed within theoretical models @xcite .
we have also showed @xcite that spatial correlators of the non - critical field are dominated at the phase transition by the critical behavior of the order parameter field . in this way the polyakov loop field imprints the knowledge of the phase transition on the non order parameter spatial correlators ( e.g. the glueball field ) .
our results are applicable to any phase transition and as such are universal .
the picture changes considerably when quarks are added to the theory .
if fermions are in the fundamental and pseudoreal representations for @xmath6 and @xmath7 , respectively , the corresponding @xmath8 or @xmath9 center of the group is never a good symmetry . for massless
quarks the order parameter is the chiral condensate which characterizes the chiral phase transition . for @xmath10=3 and two massless quark flavors at finite temperature and zero baryon density ,
the chiral phase transition is in the same universality class as the three dimensional @xmath11 spin model @xcite , becoming a smooth crossover as small quark masses are accounted for @xcite .
for @xmath12 the relevant universality class is that of @xmath13 both for the fundamental and adjoint representations @xcite .
even if the discrete symmetry is broken , one can still construct the polyakov loop and study the temperature dependence of its properties on the lattice .
one still observes a rise of the polyakov loop from low to high temperatures and naturally , although improperly , one speaks of deconfining phase transition @xcite . for fermions in the adjoint representation
the center of the group remains a symmetry of the theory , and thus besides the chiral condensate , also the polyakov loop is an order parameter .
interestingly , lattice results @xcite indicate that for ordinary qcd with quarks in the fundamental representation , chiral symmetry breaking and confinement ( i.e. a decrease of the polyakov loop ) occur at the same critical temperature .
lattice simulations also indicate that these two transitions do not happen simultaneously when the quarks are in the adjoint representation . despite the attempts to explain these behaviors @xcite , the underlying reasons are still unknown . recently , using an interesting model containing only hadrons this issue has also been addressed for ordinary qcd in @xcite .
in @xcite we have proposed a solution to this puzzle based on the general approach established in @xcite , envisioned first in @xcite .
two general features introduced in @xcite are essential : there exists a relevant trilinear interaction between the light order parameter and the heavy non - order parameter field , singlet under the symmetries of the order parameter field .
this allows for an efficient transfer of information from the order parameter to the fields that are singlets with respect to the symmetry of the theory . as a result , the non - critical fields have infrared dominated spatial correlators . the second feature , also due to the existence of such an interaction , is that the finite expectation value of the order parameter field in the symmetry broken phase induces a variation in the expectation value for the singlet field , whose value generally is non - vanishing in the unbroken phase |
in the search for correlations among physical parameters of early - type galaxies , it has been recognized that the effective radius ( @xmath13 ) , the effective mean surface brightness ( @xmath14 ) , and the central velocity dispersion ( @xmath2 ) form a planar relation ( in logarithmic space ) known as the fundamental plane ( hereafter fp ; @xcite ; @xcite ) , in the form of @xmath15 where @xmath3 and @xmath4 are found to be @xmath10 ( 1.21.5,-0.8 ) at visible wavelengths ( @xcite ; @xcite ) . under the assumption of structural homology and a constant mass - to - light ratio , the virial theorem implies that the fp coefficients should be @xmath16=(2,-1 ) the so - called ` virial expectation ' .
the observed discrepancy , or tilt of the fp with respect to the virial expectation has prompted many discussions to explain its origin ( see @xcite for a review of this field ) .
one of the ideas is that the tilt is caused by the systematic variation in the stellar population as a function of physical parameters such as galaxy luminosity .
@xcite investigated this effect by constructing the fp in @xmath7-band , which is supposedly a good tracer of the stellar mass less affected by age and dust .
meanwhile , @xcite examined the wavelength dependence on the fp coefficients , and concluded that some of the tilt is caused by the stellar population manifested by the color - magnitude relation .
these studies found that the stellar population effect exists , but it can only partially explain the tilt of the fp .
more recent investigations tackle the fp tilt problem using new methods such as gravitational lensing ( @xcite ; @xcite ) or dynamical modeling ( @xcite ; @xcite ) .
such studies suggest that the fp tilt nearly disappears when the fp is constructed in the mass - domain .
their conclusion is that the tilt must arise from the variation in the central mass - to - light ratio ( @xcite ) , but it is not clear whether the variation is dominated by dark matter or by stars @xcite . in this letter , we extend the fp analysis to wavelengths beyond @xmath7-band to further investigate the effect of stellar population on the tilt . by doing
so we aim to provide the fp that possibly better represents stellar mass ( see 5 ) , and to improve the constraints on different scenarios for the fp tilt .
early - type galaxies were chosen from the sample of @xcite , which was used to study the fp of nearby early - type galaxies in visible and near - infrared ( hereafter nir ) passbands .
the sample has the velocity dispersion information necessary for constructing the fp .
we then searched for mid - infrared ( hereafter mir)-band . ] archival images for galaxies in the @xcite sample .
for the mir data , we used the spitzer space telescope , infrared array camera ( hereafter irac ; @xcite ) images , covering four wavelength channels at 3.6 , 4.5 , 5.8 , and 8.0@xmath8 m .
the flux - calibrated , mosaiced irac images were retrieved from the spitzer archive for these objects .
ccccccc 0.55 & 2.1 & 20.1 & 81.1 & -23.4 & -21.4 & -19.4 + 2.2 & 2.3 & 14.8 & 104.0 & -26.7 & -24.5 & -22.6 + 3.6 & 2.0 & 14.3 & 65.3 & -25.8 & -23.6 & -21.8 + 4.5 & 2.2 & 14.7 & 80.5 & -25.1 & -23.0 & -21.2 + 5.8 & 1.2 & 15.2 & 90.0 & -25.1 & -22.7 & -21.0 + 8.0 & 1.1 & 13.6 & 86.8 & -24.2 & -22.3 & -20.5 the surface brightness fitting was performed for these matched galaxies , and the objects satisfying @xmath17 for three or more irac - bands were retained for the fp analysis .
we imposed this size limit in order to work with a sample with reliable @xmath0 values ( see 3.1 ) . after removing a few galaxies (
ngc1275 , ngc4824 , ngc4478 , ngc6166 ) that show peculiar light profiles ( multiple source , close to a bright galaxy or stars ) , we finally identified 56 galaxies with irac data in five clusters ( a0426 , a1656 , a2199 , a2634 , and virgo ) satisfying our selection criteria .
we present a brief summary of the photometric information in table 1 .
the exposure times for the irac data range from 72 to 1000 secs .
the above selection of the sample may introduce a bias in the derived fp coefficients ( @xcite ) .
however , such a bias would not affect our derivation of the wavelength dependence of the fp coefficients , since the multi - wavelength fp coefficients will be derived from the same galaxies for which the same bias would apply .
iraf ellipse was used to obtain surface brightness profiles of our irac sample galaxies .
we restricted the fitting region to @xmath18pixels ( along the semi - major axis ) and discarded regions with s/@xmath191 . during the fit , we held the center , and fixed the ellipticities and the position angles of isophotes to those at the effective radius in the 3.6@xmath8 m band .
in addition , 3@xmath20 clipping was applied to reject outliers such as foreground stars . to subtract the background
, we used the values determined from the sextractor @xcite .
the adaptive background mesh sizes were varied between 16 to 96pixels , and the best mesh was chosen to be the one which flattened the growth curve at the largest isophote ( @xmath2136@xmath22 ) .
after the ellipse photometry , we used the de vaucouleurs r@xmath23 law to fit the observed surface brightness profiles measured along the semi - major axis law but found no difference in the fp coefficients ; we therefore kept the r@xmath23 methodology . ] .
the fitting procedure yields the effective radius ( in @xmath24 ) @xmath25 where @xmath26 is the effective semi major axis and @xmath27 is the axis ratio of the isophote at this position .
we tested the reliability of our fitting procedure using the simulated , psf - convolved galaxies , and found that the surface brightness fitting gives unbiased , reliable results when @xmath28 . at the same time
, we get the mean surface brightness within @xmath0 ( in ab magnitudes ) @xmath29 where @xmath30 is the magnitude of the total flux within the effective isophote defined by @xmath26 and @xmath31 , while cosmological dimming , galactic extinction ( @xmath32 , using the formula of @xcite , and the extinction curve of @xcite ) , and k - correction are taken into account . the k - correction is computed using the spectral energy distribution of a 13gyr age , solar metallicity , and 0.1gyr burst model from @xcite , assuming the salpter initial mass function . the last observable , @xmath2 is a kinematic parameter and is not expected to vary as a function of wavelength ; we consequently use the same data used for the visible and nir bands @xcite . in our analysis ,
angular sizes were converted into physical length units for the fp construction by setting the distance to a1656 as 98.1mpc and calibrating the distances to individual clusters , utilizing the nir fp @xcite as a distance ladder .
we fitted the fp coefficients of the multi - waveband sample in the following manner using a variety of methods : @xmath33 where @xmath34 and @xmath35 are related as @xmath36 .
for the input @xmath0 and @xmath35 , we use our sb - fit results for mir ( 3.1 ) , and those listed in @xcite for v- and k - bands .
we tried five different fitting methods : standard least - squares fit , the inverse least - squares fit , the bisector of the two , the least - squares fit to the orthogonal plane , and the least absolute deviation fit to the orthogonal plane .
these methods are outlined below .
it is natural to think of doing the standard least - squares fit of @xmath37 ( hereafter lsq ; @xcite ; @xcite ) , but early fp work mainly took @xmath38 at the ordinate ( @xcite ; @xcite ; hereafter inverse lsq ) for their purposes .
more recent work prefers the least - squares fitting of @xmath37 by minimizing the variance orthogonal to the fp plane ( hereafter orthogonal least - squares fit , or olsq ; @xcite ) or the least absolute deviations orthogonal to the plane ( hereafter orthogonal least absolute deviation fit , or olad ; @xcite ; @xcite ) .
the orthogonal fitting has an advantage over other methods , reducing the systematic error by treating the variables symmetrically @xcite .
however , the orthogonal methods yield larger measurement errors than the lsq methods , especially for small samples @xcite . therefore , we also estimated the fp coefficients by taking the bisector , or the plane equidistant from the planes obtained through the standard lsq and inverse lsq ( hereafter the lsq bisector ) .
1,000 monte carlo samplings of subsets of early - type galaxies in @xcite were performed to derive the fp coefficient errors on a sample of 50 early types to justify our approach . through the sampling , we found the errors of the fp coefficients to be @xmath39 , best reproduced with the lsq bisector method , while the other orthogonal methods gave overestimated errors ( @xmath4050% for the coefficient @xmath3 ) . aside from the error estimates , all three symmetrized methods reproduce the fp coefficient @xmath3 of @xcite and the @xmath7-band early - type galaxy sample of @xcite within 5% agreement . on the other hand ,
the standard and inverse lsq methods are found to have about minus and plus 20% systematic biases in the coefficient @xmath3 estimates respectively in comparison to the symmetrized methods . given these results
, we adopted the fp coefficients with the lsq bisector method as our base result .
ccccc 0.55 & 1.23 @xmath41 0.10 & -0.86 @xmath41 0.04 & -9.16 @xmath41 0.40 & 0.96 + 2.2 & 1.42 @xmath41 0.11 & -0.81 @xmath41 0.05 & -8.20 @xmath41 0.41 & 0.95 + 3.6 & 1.55 @xmath41 0.11 & -0.89 @xmath41 0.04 & -9.89 @xmath41 0.39 & 0.96 + 4.5 & 1.47 @xmath41 0.11 & -0.92 @xmath41 0.04 & -10.16 @xmath41 0.41 & 0.96 + 5.8 & 1.57 @xmath41 0.13 & -0.92 @xmath41 0.05 & -10.55 @xmath41 0.50 & 0.95 + 8.0 & 1.55 @xmath41 0.14 & -0.75 @xmath41 0.05 & -9.30 @xmath41 0.60 & 0.93 in table 2 , we list the fp coefficients with errors from 1,000 bootstrap resampling ( unless obtained directly from known error estimates , e.g. , lsq methods ) derived from the lsq bisector method , for wavelengths of 0.558.0@xmath8 m .
we further plot the result of the fp fit in figure 1 , overlayed on the data points .
we caution readers to focus less on the absolute values of @xmath16 , but to focus more on the trend of the values with wavelengths or methods ( see discussions at the end of this section and 2 ) .
gathering the outcomes , we are led to the wavelength - dependent nature of the fp coefficients , with @xmath16 values getting close to the virial expectation of ( 2,-1 ) as the wavelength increases . such a tendency has been noted before ( @xcite ; @xcite ) , but our result indicates that it extends to 3.6@xmath8 m , and possibly beyond . when each cluster was analyzed separately
, we also find the trend . the wavelength dependence of the fp coefficients is further presented in figure 2 , where they are plotted using five different fitting methods ( 3.2 ) .
the tilt of the fp is maximally reduced toward the virial expectation at irac - bands , and the thickness of the fp is maintained thin for all but beyond 5.8@xmath8 m . in terms of the methodology , we confirm the analysis of 3.2 the three symmetrized fittings give coefficient @xmath3 values that are fairly consistent with each other ( considering the sample size ) . to quantify the wavelength dependence , we model the change of coefficient @xmath3 as a linear function of wavelength by simultaneously fitting the olsq , olad , and the lsq bisector results as follows : @xmath42 from the visible to 3.6@xmath8 m ( coefficient @xmath3 behaves flat afterward ) .
this relation nicely explains the difference in coefficient @xmath3 of 0.05 in the sdss g*- to z*-bands @xcite .
meanwhile , for @xmath4 , the tendency is not as linear as that for @xmath3 , but has a local maximum near the @xmath7-band , approaches closest to -1 at the irac 3.65.8@xmath8 m bands , and increases again at 8.0@xmath8 m .
we attribute this behavior at 8.0@xmath8 m to the lower s / n , as well as the complexity in the 8.0@xmath8 m emission which can be dominated by the dust emission in some cases ( @xcite ; @xcite ) . indeed
, the 8.0@xmath8 m fp has the largest scatter among irac bands .
the above result , together with the tendency of coefficient @xmath4 from table 2 , implies that the increase in coefficient @xmath3 and @xmath4 are ( @xmath43 ( 0.34,-0.03 ) from @xmath6-band to 3.6@xmath8 m , and ( @xmath43 ( 0.15,-0.08 ) from @xmath7-band to 3.6@xmath8 m .
if we use the fp coefficients from the references in figure 2 as the base values on which to apply equation ( 2 ) , we obtain @xmath10 ( 1.61.8,-0.9 ) at 3.6@xmath8 m , which is quite close to the virial expectation .
the implication of this result is discussed in the next section .
note that our coefficient @xmath3 in @xmath7-band , derived from a subsample of 56 early - types from @xcite is smaller than the value derived from their full sample of 251 early - types by @xmath44 .
the discrepancy should be mostly due to the limited sample size .
more than half of our mir galaxies belong to the coma cluster ( 29 objects ) , and the coma cluster galaxies in @xcite show coefficient @xmath3 in the @xmath7-band ( @xmath45 ) smaller than the total sample result by @xmath46 , consistent with the results of @xcite . apart from the wavelength dependence , our results seem to be skewed to the fp of the coma cluster .
recent studies suggest that the fp tilt originates mostly from a systematic variation in the mass - to - light ratio ( @xcite ; @xcite ) .
however , the cause for the mass - to - light ratio variation is uncertain : it could be due to the stellar population , or the central dark matter fraction @xcite .
also , some studies suggest that the tilt is mostly explained by the non - homology related to the variation in the sersic index n among early - type galaxies @xcite . here , we discuss the implication of our result on these issues . first , we investigated which one of the parameters - size or luminosity - dominates the observed change in the fp coefficients with increasing wavelength .
this was done by deriving the fp coefficients from the mir sample by replacing ( i ) @xmath0 s or ( ii ) @xmath35 s , with those from the shorter wavelength data ( in our case the k - band ) .
the result is presented in figure 3 ( left ) , showing that the luminosity effect is the dominant factor up to 5.8@xmath8 m .
interpretation at 8.0@xmath8 m is difficult due to low s / n and dust emission .
our result suggests that the stellar population effect is significant going from k - band to irac - bands .
next , we examined to what extent the stellar population plays a role in the fp tilt through the dynamical mass - to - light ratio @xmath47 ( e.g. , @xcite ) variation calculated from the fp coefficients . if @xmath48 , then @xmath49
the study of @xcite suggests @xmath50 based on the visible fp .
our result is that the fp coefficient reaches @xmath51@xmath52 at 3.6@xmath8 m ( 4 ) .
in such a case , this relation gives @xmath53 0.060.13 , which enables us to explain more than half of the tilt in the visible fp .
moreover , the reduced tilt in the mass plane ( @xmath54 : @xcite ) is consistent with our @xmath55 from the i - band to the 3.6@xmath8 m in equation ( 2 ) , advocating that the @xmath11 variation is reduced by the regular light distributions in the mir . as for the origin of the fp tilt
, these results add another piece of evidence against the significance of non - homology ( @xcite ; @xcite ; @xcite ) , which predicts no change in the tilt with wavelength .
we also derived the @xmath56 parameter by directly fitting the @xmath11 .
figure 3 ( right ) demonstrates that the observed dependence of @xmath11 upon @xmath12 decreases and becomes flatter at irac - bands , just like the changes in @xmath56 derived from the fp coefficients .
combined with the fact that the change in the fp tilt with wavelength is dominated by the luminosity change , our @xmath11-fit result suggests that the rest - frame mir luminosities at 3.65.8@xmath8 m better represent the stellar mass than the shorter wavelengths , somewhat in agreement with @xcite , but not so with @xcite . among many possibilities , a proper combination of the metallicity and the age variation
can possibly reproduce the observed trend , and we plan to investigate as future work , the physical origin of the @xmath11 - @xmath12 relation as a function of wavelength .
we studied the mir fundamental plane relation of 56 early - type galaxies and derived the wavelength dependence of the fp coefficients . when the fp is expressed as @xmath57
, we found that the exponent on @xmath2 , @xmath3 , increases as a function of wavelength as @xmath58 , while @xmath4 reaches closest to -1 without systematic variation with wavelength .
when the fp coefficients of previous studies are adopted as the starting point to calculate the mir fp coefficients , we found that @xmath10 ( 1.61.8,-0.9 ) which is closer to the virial expectation , and that the change is dominated by the luminosity change .
together with the reduced dependence of the @xmath11 on @xmath12 at mir wavelengths , our outcomes suggest that the variation in the stellar population is responsible for a significant portion of the fp tilt , and that the rest - frame mir better probes the stellar mass of low redshift early - type galaxies than the shorter wavelengths .
this study was supported by a grant ( r01 - 2007 - 000 - 20336 - 0 ) from the basic research program of the korea science and engineering foundation , and by the seoul science fellowship ( hj ) .
we thank the referee for useful comments , and youngmin seo and soonyoung min for algorithmic and technical advice in data analysis .
bernardi , m. , et al .
2003 , , 125 , 1866 bertin , e. , & arnouts , s. 1996 , a&as , 117 , 393 bolton , a. s. , burles , s. , treu , t. , koopmans , l. v. e. , & moustakas , l. a. 2007 , , 665 , l105 bressan , a. , et al .
2006 , , 639 , l55 bruzual , g. , & charlot , s. 2003 , , 344 , 1000 cappellari , m. , et al .
2006 , , 366 , 1126 djorgovski , s. , & davis , m. 1987 , , 313 , 59 donofrio , m. , valentinuzzi , t. , secco , l. , caimmi , r. , & bindoni , d. 2006 , new astronomy review , 50 , 447 dressler , a. , lynden - bell , d. , burstein , d. , davies , r. l. , faber , s. m. , terlevich , r. j. , & wegner , g. 1987 , , 313 , 42 fazio , g. g. , et al .
2004 , , 154 , 10 fitzpatrick , e. l. , & massa , d. 2007 , , 663 , 320 guzmn , r. , lucey , j. r. , & bower , r. g. 1993 , , 265 , 731 isobe , t. , feigelson , e. d. , akritas , m , g. , & babu , g , j. 1990 , , 364 , 104 jrgensen , i. , franx , m. , & kjrgaard , p. 1996
, , 280 , 167 ko , j. , & i m , m. 2007 , baas , 211 , 9603 lacey , c. g. , baugh , c. m. , frenk , c. s. , silva , l. , granato , g. l. , & bressan , a. 2007 , , in press ( astro - ph/0704.1562 ) laureijs , r. j. , helou , g. , & clark , f. o. 1994 , in asp conf . proc .
58 , the first symp . on the infrared cirrus and diffuse interstellar clouds , ed .
r. m. cutri & w. b. latter ( san francisco : asp ) , 133 mobasher , b. , guzman , r. , aragon - salamanca , a. , & zepf , s. 1999 , , 304 , 225 padmanabhan , n. , et al .
2004 , new astronomy review , 9 , 329 pahre , m. a. 1999 , , 124 , 127 pahre , m. a. , djorgovski , s. g. , & de carvalho , r. r. 1998a , , 116 , 1591 pahre , m. a. , de carvalho , r. r. , & djorgovski , s. g. 1998b , , 116 , 1606 robertson , b. , cox , t. j. , hernquist , l. , franx , m. , hopkins , p. f. , martini , p. , & springel , v. 2006 , , 641 , 21 scodeggio , m. , gavazzi , g. , belsole , e. , pierini , d. , & boselli , a. 1998 , , 301 , 1001 temi , p. , brighenti , f. , & mathews , w. g. 2008 , , 672 , 244 treu , t. , koopmans , l. v. e. , bolton , a. s. , burles , s. , & moustakas , l. a. 2006a , , 640 , 662 trujillo , i. , burkert , a. , & bell , e. f. 2004 , , 600 , l39 | three observables of early - type galaxies - size ( @xmath0 ) , surface brightness ( @xmath1 ) , and velocity dispersion ( @xmath2 ) - form a tight planar correlation known as the fundamental plane ( fp ) , which has provided great insights into the galaxy formation and the evolution processes .
however , the fp has been found to be tilted against the simple virial expectation , prompting debates on its origin . in order to investigate the contribution of systematic stellar population variation to the fp tilt
, we study here the fp relations of early - type galaxies in mid - infrared ( mir ) which may represent the stellar mass well .
we examined the wavelength dependence of the fp coefficients , @xmath3 and @xmath4 in @xmath5 , using a sample of 56 early - type galaxies for which visible ( @xmath6-band ) , near - infrared ( @xmath7-band ) , and mir ( spitzer irac , 3.68.0@xmath8 m ) data are available .
we find that the coefficient @xmath3 increases as a function of wavelength as @xmath9 , while the coefficient @xmath4 reaches the closest to -1 at 3.65.8@xmath8 m .
when applied to the visible fp coefficients derived from a larger sample of nearby early - type galaxies , we get the fp relation with @xmath10 ( 1.61.8,-0.9 ) at 3.6@xmath8 m .
our result suggests that the stellar population effect can explain more than half of the fp tilt , closing the gap between the virial expectation and the optical fp .
the reduction in the fp tilt is reflected in the dynamical mass - to - light ratio , @xmath11 , dependence on @xmath12 which decreases toward 3.65.8@xmath8 m , suggesting that the mir light better represents mass than the shorter wavelengths . |
the framework of lp - type problems , invented by sharir and welzl in 1992 @xcite , has become a well - established tool in the field of geometric optimization .
its origins are in linear programming : sharir and welzl developed a randomized variant of the dual simplex algorithm for linear programming and showed that this algorithm actually works for a more general class of problems they called lp - type problems .
for the theory of linear programming , this algorithm constituted an important progress , since it was later shown to be _ subexponential _ in the ram model @xcite . together with a similar result independently obtained by kalai @xcite , this was the first linear programming algorithm provably requiring a number of arithmetic operations subexponential in the dimension and number of constraints ( independent of the precision of the input numbers ) .
for many other geometric optimization problems in fixed dimension , the algorithm by sharir and welzl was the first to achieve expected linear runtime , simply because these problems could be formulated as lp - type problems .
the class of lp - type problems for example includes the problem of computing the minimum - volume ball or ellipsoid enclosing a given point set in @xmath1 , and the problem of finding the distance of two convex polytopes in @xmath1 .
many other problems have been identified as lp - type problems over the years @xcite .
once it is shown that a particular optimization problem is an lp - type problem , and certain algorithmic primitives are implemented for it , several efficient algorithms are immediately at our disposal : the sharir
welzl algorithm , two other randomized optimization algorithms due to clarkson @xcite ( see @xcite for a discussion of how it fits the lp - type framework ) , a deterministic version of it @xcite , an algorithm for computing the minimum solution that violates at most @xmath2 of the given @xmath3 constraints @xcite , and probably more are to come in the future . the framework of lp - type problems is not only a prototype for concrete optimization problems , it also serves as a mathematical tool by itself , in algorithmic @xcite and non - algorithmic contexts @xcite . an ( abstract ) lp - type problem is given by a finite set @xmath4 of _ constraints _ and a _ value _
@xmath5 for every subset @xmath6 .
the values can be real numbers or , for technical convenience , elements of any other linearly ordered set . intuitively , @xmath5 is the minimum value of a solution that satisfies all constraints in @xmath7 .
the assignment @xmath8 has to obey the axioms in the following definition .
[ def : lptype ] an _ abstract lp - type problem _ is a quadruple @xmath9 , where @xmath4 is a finite set , @xmath10 is a set linearly ordered by @xmath11 , and @xmath12 is a mapping satisfying the following two conditions : [ cols= " < , < " , ] [ lem : monotonicity ] assume @xmath13 . then locality yields @xmath14 which contradicts consistency . recall definition [ def : vbasis ] : a basis is a set @xmath15 satisfying @xmath16 for all proper subsets @xmath17 of @xmath15 .
a basis of @xmath7 is an inclusion - minimal subset of @xmath7 with the same violators .
this can be used to prove the following observation , well - known to hold for lp - type problems @xcite .
[ obs : extreme ] let @xmath18 be a violator space . for @xmath19 and all @xmath20 ,
we have 1 . @xmath21 if and only if @xmath22 , and 2 .
@xmath23 if and only if @xmath24 is contained in every basis of @xmath25 .
an element @xmath24 such that ( ii ) holds is called _ extreme _ in @xmath25 .
\(i ) if @xmath26 , we get @xmath27 by lemma [ lem : conseqloc ] . if @xmath28 , then @xmath21 is a consequence of consistency applied to @xmath29 .
( ii ) if @xmath30 , there is a basis @xmath15 of @xmath31 , and this basis is also a basis of @xmath25 not containing @xmath24 .
conversely , if there is some basis @xmath15 of @xmath25 not containing @xmath24 , then @xmath30 follows from monotonicity .
we are particularly interested in violator spaces with small bases .
let @xmath18 be a violator space .
the size of a largest basis is called the _ combinatorial dimension _
@xmath32 of @xmath18 .
observation [ obs : extreme ] implies that in a violator space of combinatorial dimension @xmath33 , every set has at most @xmath33 extreme elements .
this in turn yields a bound for the _ expected _ number of violators of a random subset of constraints , using the _ sampling lemma _ @xcite .
@xcite consider a triple @xmath34 , where @xmath35 is a function mapping subsets of the set @xmath4 to the set @xmath10 ( not necessarily ordered ) .
for @xmath19 , we define @xmath36 for @xmath37 , let @xmath38 be the expected value of @xmath39 , for @xmath25 chosen uniformly at random among all subsets of @xmath4 with @xmath40 elements
. @xmath41 is defined similarly as the expected value of @xmath42
. then for @xmath43 , the following equality holds .
@xmath44 to apply this in our situation , we fix a set @xmath45 , and we define @xmath46 .
since then @xmath47 for all @xmath25 , the following corollary is obtained .
[ cor : sampling ] let @xmath18 be a violator space of combinatorial dimension @xmath33 and @xmath45 some fixed set .
let @xmath38 be the expected number of violators of the set @xmath48 , where @xmath19 is a random subset of size @xmath49 .
then @xmath50 given a violator space @xmath18 of combinatorial dimension @xmath33 , the goal is to find a basis of @xmath4 . for this , we assume availability of the following primitive . [
prim : viol ] given @xmath6 and @xmath51 , decide whether @xmath52 .
given this primitive , the problem can be solved in a brute - force manner by going through all sets of size @xmath53 , testing each of them for being a basis of @xmath4 .
more generally , @xmath54 is a basis of @xmath7 if and only if @xmath55 consequently , the number of times the primitive needs to be invoked in order to find a basis of @xmath4 is at most @xmath56 the next two subsections show that this can be substantially improved .
fix a violator space @xmath18 of combinatorial dimension @xmath33 , implicitly specified through primitive [ prim : viol ] .
clarkson s first algorithm calls clarkson s second algorithm ( ) as a subroutine .
given @xmath6 , both algorithms compute a basis @xmath15 of @xmath7 .
= = = = @xmath57 : + ( * computes a basis @xmath15 of @xmath7 * ) + @xmath58 + @xmath57 + + @xmath59 + @xmath60 + + choose @xmath25 to be a random @xmath40-element subset of @xmath7 , @xmath61 + @xmath62 + @xmath63 + @xmath64 + @xmath65 + + @xmath66 + @xmath67 + + assuming is correct , this algorithm is correct as well : if @xmath15 is a basis of @xmath68 that in addition has no violators in @xmath7 , @xmath15 is a basis of @xmath7 .
moreover , the algorithm augments the working set @xmath10 at most @xmath33 times , which is guaranteed by the following observation .
[ obs : basis ] if @xmath69 and @xmath70 , then @xmath71 contains at least one element from every basis of @xmath7 .
let @xmath15 be a basis of @xmath7 .
assuming @xmath72 consistency yields @xmath73 , implying @xmath74 . from locality and monotonicity ( lemma [ lem : monotonicity ] ) , we get @xmath75 meaning that @xmath76 , a contradiction . it is also clear that is called only with sets of size at most @xmath77 . finally , the expected number of iterations through the loop is bounded by @xmath78 : by corollary [ cor : sampling ] ( applied to @xmath79 ) and the markov inequality , the expected number of calls to before we next augment @xmath10 is bounded by @xmath80 . algorithm computes a basis of @xmath7 with an expected number of at most @xmath81 calls to primitive [ prim : viol ] , and an expected number of at most @xmath78 calls to , with sets of size at most @xmath77 .
this algorithm calls the trivial algorithm as a subroutine . instead of adding
violated constraints to a working set , it gives them larger probability of being selected in further iterations .
technically , this is done by maintaining @xmath7 as a multiset , where @xmath82 denotes the multiplicity of @xmath24 ( we set @xmath83 ) . sampling from @xmath7
is done as before , imagining that @xmath7 contains @xmath82 copies of the element @xmath24 .
= = = = @xmath57 : + ( * computes a basis @xmath15 of @xmath7 * ) + @xmath84 + @xmath57 + + @xmath85 + + choose random @xmath61 + @xmath86 + @xmath63 + @xmath87 + @xmath88 + + @xmath66 + @xmath67 + + invoking corollary [ cor : sampling ] again ( which also applies to multisets as we use them ) , we see that the expected number of calls to before we next reweight elements ( a _ successful _ iteration ) , is bounded by @xmath80 .
it remains to bound the number of successful iterations .
let @xmath2 be a positive integer .
after @xmath89 successful iterations , we have @xmath90 for every basis @xmath15 of @xmath7 .
in particular , @xmath91 .
every successful iteration multiplies the total weight of elements in @xmath7 by at most @xmath92 , which gives the upper bound ( not only for @xmath93 but actually for @xmath94 ) . for the lower bound
, we use observation [ obs : basis ] again to argue that each successful iteration doubles the weight of some element in @xmath15 , meaning that after @xmath89 iterations , one element has been doubled at least @xmath2 times . because the lower bound exceeds the upper bound for @xmath95 , the bound on @xmath2 follows .
summarizing , we get the following lemma .
algorithm computes a basis of @xmath7 with an expected number of at most @xmath96 calls to primitive [ prim : viol ] , and expected number of at most @xmath97 calls to , with sets of size @xmath98 .
[ thm : runtime ] using a combination of the above two algorithms , a basis of @xmath4 in a violator space @xmath18 can be found calling primitive [ prim : viol ] expected @xmath99 many times . using the above bound for the trivial algorithm , can be implemented to require an expected number of at most @xmath100 calls to the primitive . applying this as a subroutine in @xmath101 with @xmath102 , @xmath103
is bounded by @xmath104 , and we get an overall expected complexity of @xmath105 in terms of the number of calls to primitive [ prim : viol ] . the terms @xmath106 and @xmath107 are asymptotically dominated by either @xmath108 or @xmath109 , and we get the simplified bound of @xmath110 .
we show in this section that the problem of finding the sink in a @xmath33-dimensional _ grid unique sink orientation _
@xcite can be reduced to the problem of finding the ( unique ) basis of a violator space of combinatorial dimension @xmath33 .
unique sink orientations of grids arise from various problems , including linear programming over products of simplices and generalized linear complementarity problems ( glcp ) over @xmath0-matrices @xcite .
the glcp has been introduced by cottle and dantzig @xcite as a generalization of the well known lcp @xcite .
there are also applications in game theory ; for instance @xcite show how parity , mean - payoff , and simple stochastic games are related to grid uso .
fix a partition @xmath111 of the set @xmath112 into @xmath33 nonempty subsets , where we refer to @xmath113 as the _ block @xmath114_. a subset @xmath115 is called a _ vertex _ if @xmath116 for all @xmath114 .
the vertices naturally correspond to the cartesian product of the @xmath113 .
let @xmath117 be the set of all vertices . in the following definition ,
we introduce the _ grid spanned _ by subsets @xmath118 whose union is @xmath6 .
the vertex set of this grid contains all vertices @xmath119 ( @xmath120 ) , with two vertices being adjacent whenever they differ in exactly two elements .
the @xmath33-dimensional _ grid _ spanned by @xmath121 is the undirected graph @xmath122 , with @xmath123 here , @xmath124 is the symmetric difference of sets .
@xmath125 is in one - to - one correspondence with the cartesian product @xmath126 and the edges in @xmath127 connect vertices in @xmath125 whose corresponding tuples differ in exactly one coordinate .
see figure [ fig : grid ] left for an example of a grid .
note that @xmath128 is the empty graph whenever @xmath129 for some @xmath114 .
we say that such a @xmath7 is _ not @xmath130-valid _ , and it is _
@xmath130-valid _ otherwise
. a _ subgrid _ of @xmath128 is any graph of the form @xmath131 , for @xmath132 .
an orientation @xmath133 of the graph @xmath134 is called a _ unique sink orientation _ ( uso ) if all nonempty subgrids of @xmath135 have unique sinks w.r.t .
@xmath133 .
we are interested in finding the sink in a uso of @xmath135 as fast as possible , since the sink corresponds to the solution of the underlying problem ( the @xmath0-matrix glcp , for example ) .
our measure of complexity will be the expected number of _ edge evaluations _
, see @xcite .
an edge evaluation returns the orientation of the considered edge and can typically be implemented to run in polynomial time ( depending on the underlying problem ) . in the remainder of this paper , we derive the following theorem .
[ thm : uso_algo ] the sink of a unique sink grid orientation can be found by evaluating expected @xmath110 edges . note that a uso @xmath133 can be cyclic ( see the thick edges in figure [ fig : grid ] right ) .
if @xmath133 induces the directed edge @xmath136 , we also write @xmath137 .
any uso can be specified by associating each vertex @xmath138 with its outgoing edges . given @xmath138 and @xmath139
, we define @xmath140 to be the unique vertex @xmath141 that is different from @xmath138 , and we call @xmath142 the _ neighbor _ of @xmath138 _ in direction _ @xmath143 .
note that @xmath138 is a neighbor of @xmath142 in some direction different from @xmath143 .
given an orientation @xmath133 of @xmath135 , the function @xmath144 , defined by @xmath145 is called the _
outmap _ of @xmath133 . by this definition ,
any sink w.r.t .
@xmath133 has empty outmap value .
let us fix a unique sink orientation @xmath133 of @xmath135 . given a @xmath130-valid subset @xmath6 ,
we define @xmath146 to be the unique sink vertex in @xmath128 . for a subset @xmath7 that is not @xmath130-valid ,
let @xmath147 thus @xmath148 is the set of elements occurring in blocks of @xmath130 disjoint from @xmath7 .
[ def : gridtoviol ] for @xmath6 , define @xmath149 the pair @xmath18 from definition [ def : gridtoviol ] is a violator space of combinatorial dimension @xmath33 .
moreover , for all @xmath130-valid @xmath6 , the unique sink of the subgrid @xmath150 corresponds to the unique basis of @xmath7 in @xmath18 . for every @xmath121
, consistency holds by definition of @xmath151 and @xmath148 . in order to prove locality for @xmath152
, we look at three different cases . [ [ mathbfg - is - not - mathbfpi - valid . ] ] @xmath153 is not @xmath154-valid .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + then , @xmath155 is not @xmath130-valid either .
the condition @xmath156 means that @xmath17 is disjoint from the same blocks as @xmath7 .
this implies @xmath157 , hence @xmath158 .
[ [ mathbfg - and - mathbff - are - both - mathbfpi - valid . ] ] @xmath153 and @xmath159 are both @xmath154-valid .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + then @xmath160 is a nonempty subgrid of @xmath128 , and @xmath161 means that the sink of @xmath160 has no outgoing edges into @xmath128 .
thus the unique sink of @xmath160 is also a sink of @xmath128 and therefore the unique one .
this means that @xmath162 , from which @xmath158 follows .
[ [ mathbfg - is - mathbfpi - valid - mathbff - is - not - mathbfpi - valid . ] ] @xmath153 is @xmath154-valid , @xmath159 is not @xmath154-valid .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + then the condition @xmath161 can never be satisfied since @xmath163 contains at least one full block @xmath113 , and @xmath164 .
next we prove that a largest basis in @xmath18 has at most @xmath33 elements .
for this , let @xmath6 be a set of size larger than @xmath33 . if @xmath7 is @xmath130-valid , we have @xmath165 since @xmath166 for any vertex @xmath138 .
this means that @xmath7 has a subset of size @xmath33 with the same violators , so @xmath7 is not a basis .
if @xmath7 is not @xmath130-valid , we consider some subset @xmath15 that contains exactly one element from every block intersected by @xmath7 . by definition , we have @xmath167 and @xmath168 .
since @xmath15 has less than @xmath33 elements , @xmath7 can not be a basis in this case , either .
it remains to prove that for @xmath7 being @xmath130-valid , the vertex @xmath169 is the unique basis of @xmath7 in @xmath18 .
we have already shown that @xmath170 must hold in this case .
moreover , @xmath171 contains no full block @xmath113 . on the other hand , any proper subset @xmath17 of @xmath169
is not @xmath130-valid , so its violator set _ does _ contain at least one full block .
it follows that @xmath172 , so @xmath169 is a basis of @xmath7 .
the argument is complete when we can prove that no other vertex @xmath119 is a basis of @xmath7 .
indeed , such a vertex @xmath138 is not a sink in @xmath128 , meaning that @xmath173 .
this implies @xmath174 .
note that the global sink of the grid uso corresponds to the unique @xmath33-element ( and @xmath130-valid ) set @xmath15 with @xmath175 .
this is exactly the set output by the call ` basis1`@xmath101 of clarkson s algorithms , when we apply it to the violator space constructed in definition [ def : gridtoviol ] .
primitive [ prim : viol ] corresponds to one edge evaluation in the uso setting . with theorem [ thm : runtime ] ,
we therefore have proved theorem [ thm : uso_algo ] . for small @xmath33 , the running time given in the theorem is faster than the one from the _ product algorithm _
@xcite which needs expected @xmath176 edge evaluations , where @xmath177 is the @xmath3-th harmonic number .
we introduced violator spaces as a new framework for optimization problems and showed that acyclic violator spaces are equivalent to abstract and concrete lp - type problems .
it turned out that the explicit ordering inherent to lp - type problems is not necessary in order to capture the structure of the underlying optimization problem .
violator spaces are more general than lp - type problems , yet clarkson s algorithms still work on them .
the sharir - welzl algorithm is also applicable for violator spaces in a straightforward way .
however , the most obvious translation of this algorithm to the setting of violator spaces is not even guaranteed to finish , since for a general violator space it may run in a cycle and the subexponential analysis thus breaks down .
we have seen that unique sink orientations are models for possibly cyclic violator spaces , and with clarkson s algorithms we therefore have a fast scheme to solve fixed dimensional uso problems like the generalized linear complementarity problem with a @xmath0-matrix .
the glcp with a @xmath0-matrix has in general a cyclic structure and therefore gives rise to a cyclic uso . a violator space obtained from a cyclic
uso is again cyclic .
it is interesting that there are no cycles in a 2-dimensional grid uso @xcite .
whether the same is true for violator spaces of combinatorial dimension 2 is an open question .
we thank an anonymous referee for useful comments .
the second author would like to thank nina amenta for discussions concerning lp - type problems , possibly already forgotten by her as they took place many years ago , but nevertheless helpful for reaching the results in this paper .
m. sharir , e. welzl , a combinatorial bound for linear programming and related problems , in : proc .
9th symposium on theoretical aspects of computer science ( stacs ) , vol .
577 of lecture notes in computer science , springer - verlag , 1992 , pp .
569579 .
h. bjrklund , s. sandberg , s. vorobyov , a discrete subexponential algorithm for parity games , in : proc .
20th annual symposium on theoretical aspects of computer science ( stacs ) , springer - verlag , 2003 , pp .
663674 .
b. grtner , e. welzl , linear programming - randomization and abstract frameworks , in : proc .
13th annual symposium on theoretical aspects of computer science ( stacs ) , springer - verlag , london , uk , 1996 , pp .
669687 .
b. grtner , w. d. morris , jr .
, l. rst , unique sink orientations of grids , in : proc .
11th conference on integer programming and combinatorial optimization ( ipco ) , vol .
3509 of lecture notes in computer science , springer - verlag , 2005 , pp . 210224 .
i. schurr , t. szab , jumping does nt help in abstract cubes , in : proc .
11th conference on integer programming and combinatorial optimization ( ipco ) , vol .
3509 of lecture notes in computer science , springer - verlag , 2005 , pp . 225235 .
h. bjrklund , s. sandberg , s. vorobyov , a combinatorial strongly subexponential strategy improvement algorithm for mean payoff games , in : proc .
29th international symposium on mathematical foundations of computer science ( mfcs ) , vol .
3153 of lecture notes in computer science , springer - verlag , 2004 , pp . 673685 .
b. grtner , l. rst , simple stochastic games and p - matrix generalized linear complementarity problems , in : proc .
15th international symposium on fundamentals of computation theory ( fct ) , vol .
3623 of lecture notes in computer science , springer - verlag , 2005 , pp . 209220 .
p. kovro , generalized linear programming , master s thesis , charles university , prague , faculty of mathematics and physics ( 2002 ) . | sharir and welzl introduced an abstract framework for optimization problems , called _ lp - type problems _ or also _ generalized linear programming problems _ , which proved useful in algorithm design .
we define a new , and as we believe , simpler and more natural framework : _ violator spaces _ , which constitute a proper generalization of lp - type problems .
we show that clarkson s randomized algorithms for low - dimensional linear programming work in the context of violator spaces .
for example , in this way we obtain the fastest known algorithm for the _ @xmath0-matrix generalized linear complementarity problem _ with a constant number of blocks .
we also give two new characterizations of lp - type problems : they are equivalent to _ acyclic _ violator spaces , as well as to _ concrete _ lp - type problems ( informally , the constraints in a concrete lp - type problem are subsets of a linearly ordered ground set , and the value of a set of constraints is the minimum of its intersection ) . , , , lp - type problem , generalized linear programming , violator space , clarkson s algorithms , unique sink orientation , generalized linear complementarity problem |
the availability of quantum - degenerate bosonic atomic systems has recently allowed the extension of atom optics to the nonlinear @xcite and the quantum regimes .
matter - wave four - wave mixing @xcite , coherent matter - wave amplification @xcite and superradiance @xcite , the generation of dark @xcite and bright @xcite atomic solitons and of correlated atomic pairs @xcite have been demonstrated , and so has the matter - wave analog of second - harmonic generation , the creation of a molecular condensate component @xcite .
in contrast , the development of the nonlinear atom optics of fermionic atomic systems is not nearly as far along .
while it has been shown theoretically @xcite that the four - wave mixing of fermionic matter waves is possible in principle , these predictions have not been verified experimentally so far .
still , the recent achievement of temperatures as low as @xmath0 , where @xmath1 is the fermi temperature , for the fermions @xmath2k and @xmath3li @xcite is encouraging , and it is hoped that first experiments on fermionic nonlinear atom optics will take place in the near future . in addition to the fundamental goal of achieving a bcs phase transition into pairing and superfluidity @xcite , research along these lines is also motivated by recent results that hint at the possibility to lower the phase noise in interferometric measurements below the bosonic standard quantum limit by using instead degenerate fermionic beams @xcite .
the first theoretical discussions of fermionic nonlinear atom optics were presented in refs . @xcite and @xcite , which treated the case of a single ` test ' particle scattering off a periodic density grating formed by a degenerate fermi gas .
they showed that for an appropriately prepared grating , the fermionic system can indeed undergo four - wave mixing .
in contrast to the standard interpretation in terms of `` bosonic amplification '' , which clearly is not applicable to fermions , this effect was interpreted in terms of the constructive quantum interference between different paths leading to the same final state .
one important aspect of the fermionic case is that , in contrast to bosons , considerable care must be taken in combining two matter waves to form a `` grating '' , so that their interaction with a third wave can produce a significant four - wave mixing signal .
consider , as we shall do in this paper , counterpropagating matter waves separated in momentum by @xmath4 . in the case of bosons ,
two obvious possibilities correspond to the states @xmath5 @xmath6^n|0\rangle , \ ] ] the @xmath7 being usual bosonic creation operators and @xmath8 the atomic vacuum .
the fist case describes two counterpropagating beams of @xmath9 atoms each and of momenta @xmath10 , while the second state corresponds to a density grating obtained by identically preparing @xmath11 atoms in a coherent superposition of states of momenta @xmath12 .
it is known from the study of atomic diffraction by optical fields that these two states lead to different diffraction patterns , because the first one contains `` which way '' information while the second does nt @xcite .
this difference becomes however insignificant for large gratings .
the situation for fermions is more complicated , since the pauli exclusion principle precludes one from placing more than one atom per state .
one needs instead to consider multimode atomic beams , centered around the mean momenta @xmath12 . in this case
the states @xmath13 and @xmath14 are replaced by @xmath15|0\rangle\ ] ] and @xmath16|0\rangle , \label{fermi - grating}\ ] ] where @xmath17 are fermionic creation operators for atoms of momenta in the vicinity of @xmath12 , the total number of atoms involved being indicated in the appropriate products . from refs .
@xcite , we know that it is the quantum coherence apparent in matter - wave states of the form @xmath18 that is responsible for fermionic four - wave mixing . in order for the required quantum interference to occur , it is essential that every atom be in a coherent superposition of momentum states centered around @xmath19 and @xmath20 .
so far , our discussion has ignored the time dependence of the grating .
but since the atoms forming a fermionic grating all have slightly different kinetic energies , their free evolution results in a dephasing that is expected to eventually lead to the disappearance of the four - wave mixing signal . although the importance of this dephasing was pointed out in ref .
@xcite , no quantitative discussion of its effect has been presented so far .
the present paper addresses this problem quantitatively by a combined analytical and numerical study of the diffraction of a beam of fermionic atoms off a large fermionic grating .
we fully include the dynamics of the atomic beam , but neglect its back - action on the grating dynamics , considering only its free evolution .
this is the matter - wave analog of the undepleted pump approximation in nonlinear optics .
section ii introduces our model and describes its dynamics in terms of equations of motion for particle - hole operators .
the effect of the grating dynamics is characterized in terms of a dephasing time @xmath21 , whose impact is then illustrated in the simple case of raman - nath diffraction .
the bragg regime is analyzed in section iii using a combination of an analytical discussion of a simple two - mode system and full numerical simulations .
we determine the characteristic time scales governing this process , and conclude that four - wave mixing in degenerate fermi gases should barely be observable .
noting that the dephasing of the grating is in principle reversible , we turn in section iv to a discussion of possible ways to achieve such a reversal , based on analogies with the photon echo techniques of quantum optics . since the physical origin of the dephasing is the difference in kinetic energies for atoms with different momenta , one elegant approach exploits the interaction of the grating with a periodic external potential to achieve a negative effective mass of the atoms .
such methods pave the way for the observation of phenomena that seem out of experimental reach at first glance .
finally , section v is a summary and outlook .
a typical atomic four - wave mixing experiment involves three input wave packets that interact , typically via @xmath22-wave collisions , to produce a fourth packet in a new momentum state .
physically , this can be interpreted as the result of the phase - matched scattering of one of the wave packets off the matter - wave grating produced by the superposition of the other two , as discussed in the introduction . as is readily apparent from eq .
( [ fermi - grating ] ) , though , a grating formed by @xmath11 fermions occupies @xmath23 momentum states , with @xmath11 states around the momentum @xmath19 and @xmath11 states near @xmath20 in our case . as a result of the free evolution associated with the spread in kinetic energies of these atoms , the density grating undergoes a dephasing , similar to doppler broadening in a thermal vapor .
the goal of this paper is to investigate the effects of this dephasing and to determine to what extent it can be reduced , or even eliminated .
we consider the situation of a spin - polarized fermionic beam diffracted by an atomic grating of spin - polarized fermions that are either of a different species or in a different spin state , see fig .
[ model ] , the beam and grating atoms interacting coherently via elastic @xmath22-wave scattering .
we neglect @xmath24-wave scattering , whose cross - section at zero temperature is orders of magnitude smaller than that of _ s_-wave scattering .
the diffraction of a fermionic atomic beam by a periodic potential was previously treated for the case of free fermions ( or equivalently a one - dimensional bosonic tonks gas ) @xcite and of cooper pairs @xcite .
these treatments are restricted to the raman - nath regime , where the kinetic energies of the diffracted atoms are negligible compared to the potential energy of the grating .
the scattering of two fermions by an optical standing wave was also considered in ref .
@xcite . also worth mentioning in the present context is ref .
@xcite , which investigates the nonlinear mixing of two bosonic and one fermionic wave . in these cases
pauli blocking has a significant effect on the diffraction pattern , but as the gratings are non - fermionic , the dephasing , which is central to our work , is not addressed .
we restrict the quantized description of the system to the dimension along the axis of the grating only , the `` transverse '' direction .
the dynamics of the system in the direction perpendicular to it is treated classically , and parameterized by the time the atoms in the beam interact with the grating .
note that this approximation neglects both the normal reflection and the bragg reflection of the incident beam off the grating @xcite .
this model is described by the second - quantized hamiltonian @xmath25 where @xmath26 ( @xmath27 ) is the kinetic energy of a grating ( beam ) atom of mass @xmath28 ( @xmath29 ) and transverse momentum @xmath30 , @xmath31 @xmath32 is the reduced mass , @xmath33 is the @xmath22-wave scattering length characterizing the ultracold collisions between grating and beam atoms , @xmath34 is an effective transverse cross - sectional area , and @xmath35 is the quantization length .
the annihilation and creation operators @xmath36 and @xmath37 satisfy the fermionic anticommutation relations @xmath38_+=0 $ ] and @xmath39_+=\delta_{kk'}\delta_{ss'}$ ] , the spin components @xmath40 corresponding to the different spin states or atomic species of the grating and the beam , respectively . since the elementary process underlying matter - wave diffraction consists in annihilating an atom of momentum @xmath30 and creating an atom with momentum @xmath41 , it is useful to describe the dynamics of the atomic beam in terms of `` particle - hole '' operators @xmath42 where the subscripts `` b '' and `` g '' label beam and grating atoms , respectively . the heisenberg equations of motion for these operators
are easily found as @xmath43 with @xmath44 .
a similar equation holds for the grating particle - hole operators , with the substitution @xmath45 .
we consider a beam of @xmath46 atoms propagating toward the grating with some central transverse momentum @xmath47 . as discussed in the introduction the grating is taken to be in a superposition of states of transverse momenta @xmath48 and @xmath49 , with @xmath50 , where @xmath51 is the fermi momentum ( in one dimension ) and @xmath52 is the number of atoms forming the grating .
such a grating can be prepared with two - photon bragg pulses of bandwidth larger than the fermi frequency @xmath53 in a two - step process starting from a stationary homogeneous gas , see e.g. ref .
@xcite . in that scheme , a @xmath54-pulse imparting a transverse momentum @xmath55 to each atom in the static cloud
is followed by a @xmath56 bragg pulse that creates the desired superposition state .
thus the initial state of beam - grating system is @xmath57 where @xmath58 the corresponding initial expectation values of the particle - hole operators are @xmath59 in the following we assume that the matter - wave grating is sufficiently large that one can safely ignore the back - action of the atomic beam on its dynamics .
mathematically , this amounts to neglecting the term proportional to @xmath60 in the grating version of eq .
( [ particleholefullequations ] ) . in this limit ,
the grating simply undergoes a free evolution fully governed by its transverse kinetic energy . assuming , consistently with the assumption of a large grating , that the expectation value of products of beam and grating particle - hole operators can be factorized as @xmath61 making explicit use of the grating initial condition ( [ initialvalues ] ) , and introducing the slowly varying particle - hole operators @xmath62,\ ] ] and similarly for the grating , yields then the beam particle - hole heisenberg equations of motion @xmath63,\end{aligned}\ ] ] where @xmath64,\ ] ] reduces in the continuum limit , @xmath65 , to @xmath66 with @xmath67 . for times larger than the _ dephasing time _
@xmath68 we have that @xmath69 , that is , the diffraction of the beam comes to a stop .
this consequence of the dephasing between different momentum states of the atomic grating resulting from their free evolution , represents an essential difference between fermionic and bosonic four - wave mixing : the free evolution of a bosonic grating optically prepared from a bose - einstein condensate at zero temperature gives rise to a phase factor that can easily be transformed away . for fermions , though , it results in a dephasing and the effective shutting off of its interaction with the atomic beam .
this impact of the dephasing is particularly easy to see in the raman - nath regime of atomic diffraction .
we recall that this is the regime where the kinetic energy of the atoms in the beam is small compared to the potential energy of the density grating .
however , we retain the kinetic energies of the grating atoms , since they are responsible for the dephasing . as such , this model corresponds to the case where the atoms in the incident beam are much heavier than the atoms in the grating .
, in the raman - nath regime .
the initial momentum spread of the beam is @xmath70 .
the effects of the dephasing grating become apparent .
time in units of the inverse mean - field interaction strength @xmath71 and @xmath72.,width=302,height=188 ] approximating the exponential functions by unity in eq .
( [ lin ] ) consistently with the raman - nath approximation gives @xmath73 the solution of these operator equations for @xmath74 constant are known @xcite in terms of bessel functions of the first kind of order @xmath22 as @xmath75 due to conservation of momentum , only states of transverse momenta separated by integer multiples of @xmath4 are dynamically coupled . for a beam of initial momentum
spread @xmath76 , this implies that a linearly increasing sequence of replica of the initial distribution will be generated in time .
things are more complicated if @xmath77 , since pauli blocking leads to a situation where only those transverse modes near the edge of the distribution can initially be diffracted , creating holes into which modes deeper into the initial distribution can then subsequently be coupled .
this results in an effective broadening of the initial distribution in time , as illustrated in the early stages of fig .
[ rn_fermions ] .
since @xmath78 is actually not a constant , the solution ( [ ramansol ] ) is only valid for times @xmath79 . in general , it is necessary to solve the expectation value of the coupled raman - nath equations numerically .
the effect of the grating diffusion , and the resulting vanishing of @xmath78 for long enough times , are readily apparent in fig .
[ rn_fermions ] , which shows that the beam diffraction effectively ceases for @xmath80 .
this is seen even more clearly in fig .
[ raman_gt ] , which compares the number of significantly occupied transverse modes with and without dephasing of the grating . ) as a function of time in the presence ( circles ) and absence ( crosses ) of dephasing .
time is in units of the inverse mean - field frequency @xmath81 and @xmath72 .
, width=302,height=188 ]
we now turn to the analysis of the bragg regime of fermionic atomic diffraction .
we recall that for single atoms and a grating of period @xmath82 , this regime corresponds to the situation where only two degenerate modes , of transverse momenta @xmath12 , are coupled .
the probabilities of occupation of these two states oscillate periodically in time , the so - called _ pendellsung _ @xcite .
bragg scattering is widely used in atomic beam splitters , both for single atoms @xcite and for bose - einstein condensates @xcite .
fermi statistics complicate the situation in two ways : first , as we have seen , the distribution of momentum states in the grating around @xmath12 leads to a dephasing resulting in the time - dependent coupling constant @xmath78 .
in addition , only those fermions with initial transverse momentum sufficiently close to @xmath19 fulfill the bragg condition . hence , it is expected that bragg diffraction will burn a spectral hole in the initial fermi sea of the beam , leaving atoms away from that hole practically untouched .
we can gain some insight into the effects of the dephasing by considering a simple two - state model that retains only the coupling between the transverse modes @xmath12 , neglecting their coupling to higher - order modes .
this approximation is justified if the phase - matching condition is strongly violated for these modes , in other words , if the difference in kinetic energy @xmath83 is much larger than the mean - field energy @xmath84 responsible for mode coupling , i.e. , @xmath85 @xmath86 being the atomic density in the grating .
this situation is described by the closed set of operator equations @xmath87,\nonumber\\ \frac{d}{dt}\hat{\rho}^{(b)}_{-q/2,q}&= & ig(t ) \left [ \hat{n}^{(b)}_{q/2}- \hat{n}^{(b)}_{-q/2}\right],\end{aligned}\ ] ] where @xmath78 is given by eq .
( [ goft ] ) .
these equations are reminiscent of the optical bloch equations that describe the dynamics of a two - level atom driven by a resonant single - mode field of time - dependent amplitude , with the operators @xmath88 corresponding to the populations of the two levels and the particle - hole operator @xmath89 to the atomic polarization .
the evolution of their expectation values can readily be found for the initial conditions @xmath90 and @xmath91 as @xmath92 where @xmath93 similarly to the situation that we encountered in the raman - nath regime , the dephasing of the grating results in the diffraction effectively coming to an end after a time of the order of @xmath21 .
li - beam scattering off a @xmath2k - grating in the bragg regime .
time is in units of the inverse mean - field frequency @xmath94 . in this example
@xmath95 and @xmath96 .
the full momentum widths of the grating and of the beam are @xmath97 and @xmath98 , respectively.,width=302,height=188 ] fig .
[ bwscatter ] summarizes the results of the numerical integration of the full equations of motion ( [ particleholefullequations ] ) in the bragg regime for a @xmath3li beam scattering off a grating formed by @xmath99k atoms .
it shows the population of the transverse modes of the atomic beam as a function of time for the case where the initial full transverse momentum width of the grating and beam are @xmath100 and @xmath101 , respectively . the hole burned in the initial fermi sea is clearly visible , as is the impact of the dephasing of the grating .
the first maximum in the pendellsung oscillations is strongly pronounced for modes near the phase matched mode @xmath19 , but further maxima can not be observed due the dephasing of the fermionic grating . as a function of beam size , @xmath102 , for different sized gratings , with @xmath96 .
the efficiency @xmath103 decreases with separately increasing size of the fermi seas of beam and grating .
nevertheless for reasonably small momentum spreads a decent efficiency is obtained .
, width=302,height=188 ] one can characterize the `` efficiency '' @xmath103 of the four - wave mixing process by the first maximum of the quantity @xmath104 which measures the maximum fraction of atoms whose transverse momentum has been flipped about @xmath105 .
it is plotted in fig .
[ eff_dephase ] as a function of the momentum spread of the incident @xmath3li beam for three widths of the @xmath99k grating .
the behavior of @xmath103 is as expected , since a wider incident beam contains more atoms that do not fulfill the bragg condition , and a larger fermi sea of the grating means a shorter dephasing time .
experiments are likely to be carried out in a three - dimensional geometry , in which case the time - dependent coupling constant ( [ timedepcoupling ] ) becomes @xmath106 where @xmath21 is still given by eq .
( [ taud ] ) .
it is easily seen that @xmath107 decays in a manner similar to @xmath78 , so that the lifetime of the grating is essentially the same as in one - dimension , provided that @xmath108 remains the same .
note however that these two times scale differently with the density of the grating , due to the different mode densities in one and three dimensions . in order to assess the experimental feasibility of a fermionic bragg diffraction experiment it is necessary to compare the dephasing time @xmath21 to the time @xmath109 that it would take to observe a full pendellsung oscillation in the absence of dephasing . with the approximation @xmath110 we have readily that @xmath111 .
hence , the observation of pendellsung oscillations requires @xmath112 where we have used the three - dimensional relationship between the density @xmath86 and @xmath108 , @xmath113 .
consider for example the diffraction of fermionic potassium @xmath114k off a spin - polarized @xmath114k grating of peak number density @xmath115 and a scattering length of @xmath116 . assuming a grating period of 500 nm @xmath117 , so that pendellsung oscillations will not be observable in that case .
the next section discusses possible ways around this difficulty .
the condition ( [ exp - cond ] ) shows that one possible way to increase the ratio of @xmath21 to @xmath109 is to increase the grating density @xmath86 .
this , however , is likely to be impractical .
another solution is to increase the scattering length @xmath33 via a feshbach resonance . noting that @xmath118
, we can also gain an order of magnitude or so in the ratio ( [ exp - cond ] ) by diffracting light fermions such as @xmath119li off a potassium grating .
ketterle _ et . al _
@xcite also proposed a direct way to increase the decay time in three dimensions by reducing the effective fermi wave vector in the direction of the grating by increasing the size of the fermi sea in the other directions .
an alternative method that allows one to completely reverse the effects of the dephasing is based on the observation that this dephasing is coherent , simply resulting from the different kinetic energies of the atoms forming the grating .
this is reminiscent of doppler broadening in laser spectroscopy , which leads to a dephasing of the atomic polarization . in that case , the dephasing can easily be reversed by photon echo techniques @xcite .
things are slightly more complicated here , because the phase of the individual atoms is just given by their kinetic energy @xmath120 the only way to reverse the sign of this phase is to reverse the mass of the atoms . this can be achieved by placing the atoms forming the grating in an additional periodic potential that gives them a _ negative effective mass_. k of the first two bands at the center of the first brillouin zone as a function of @xmath121 , in units of the recoil energy @xmath122 for an optical wavelength of 852 nm .
, width=302,height=188 ] ignoring two - body collisions as well as whatever external fields are required to prepare the initial grating ( [ initialconditions ] ) for now , the evolution of the grating atoms in the presence of an optical lattice is governed by the hamiltonian @xmath123 where @xmath121 is the lattice depth and @xmath124 the wave vector of the laser creating the periodic potential .
the band structure associated with this potential is readily obtained numerically , yielding at the center of the first brillouin zone the effective mass @xmath125 where @xmath126 is the band index and @xmath30 is the quasi - momentum .
figure [ effmass ] plots this effective mass of @xmath114k for the first two energy bands . as is well known , @xmath127 is alternatively positive and negative as the band index is increased .
we remark that since @xmath60 is a pseudo - potential adjusted so that the scattering problem associated with two - body collisions asymptotically produces the correct collisional cross - section , the appearance of an effective mass does not change the reduced mass @xmath128 in eq .
( [ u0 ] ) .
( [ exp - cond ] ) becomes @xmath129 from fig .
[ effmass ] , we see that the effective mass can be orders of magnitude larger than @xmath28 , leading to much easier conditions to satisfy to observe fermionic bragg diffraction .
.,width=302,height=415 ] even more interesting is the use of the negative effective mass associated with the second energy band , since it permits a full reversal of the dephasing .
the excitation of the grating atoms to that band can be achieved e.g. via a two - photon raman with co - propagating laser beams , or by modulating the phase of one of the lattice beams .
assuming for simplicity that the effective masses corresponding to the @xmath130 and @xmath131 bands are equal in magnitude , it is easily seen that if the raman pulse is applied at time @xmath132 , then @xmath133 , and the dephasing has been completely reversed .
clearly , a sequence of pulses alternatively transferring the grating atoms between the two energy bands will keep alternating the sign of @xmath127 , thereby continuously compensating the doppler broadening . in case
the magnitude of curvatures of the two bands are different , the rephasing time needs to be appropriately adjusted .
[ compare ] shows the result of this atom echo technique .
the second peak of the _ pendellsung _ is now clearly visible , in contrast to the situation in fig .
[ bwscatter ] . for better comparability to the case of uncontrolled dephasing
we considered the same magnitudes for effective and real mass .
we have transferred the atoms to the first band at scaled time @xmath134 and back to the lowest band at @xmath135 .
the results of the dephasing of the grating start again later on , but could be eliminated by transferring the atoms back to the first energy band .
in this paper , we have investigated the behavior of purely fermionic scattering , i.e. a beam of fermions diffracted by a fermionic density grating , in the limit of a large grating .
we explored different regimes which are determined by the relative magnitude of mean field energy and kinetic energy .
fermi statistic shows its distinct signatures in the limited lifetime of the grating and the inability to satisfy the bragg condition for large input beams .
we proposed novel ways to increase the dephasing time of the fermionic grating which is the most severe limitation for fermionic four - wave mixing .
future work will focus on the complex dynamics arising from the back - action of the scattered atoms on the grating .
this work is supported in part by the us office of naval research , by the national science foundation , by the us army research office , by the national aeronautics and space administration , and by the joint services optics program .
h. c. acknowledges the support of the studienstiftung des deutschen volkes .
99 g. lenz , p. meystre , and e. m. wright , phys .
* 71 * , 3271 ( 1993 ) .
l. deng , e. w. hageley , j. wen , m. trippenbach , y. band , p. s. julienne , j. e. simsarian , k. helmerson , s. l. rolston , and w. d. phillips , nature * 398 * , 218 ( 1999 ) .
s. innouye , t. pfau , a. p. chikkatur , a. grlitz , d. e. pritchard and w. ketterle , nature * 402 * , 641 ( 1999 ) ; m. kozuma , y. suzuki , y. torii , t. sugiura , t. kuga , e. w. hagley , and l. deng , science * 286 * , 2309 ( 1999 ) .
s. innouye , a. p. chikkatur , d. m. stamper - kurn , j. stenger , d. e. pritchard , and w. ketterle , science * 281 * , 571 ( 1999 ) .
s. burger , k. bongs , s. dettmer , w. ertmer , k. sengstock , a. sanpera , g. v. shlyapnikov , and m. lewenstein , phys . rev . lett . * 83 * , 5198 ( 1999 ) .
k. e. strecker , g. b. partridge , a. g. truscott , and r. g. hulet , nature * 417 * , 150 ( 2002 ) .
j. m. vogels , k. xu , and w. ketterle , phys . rev . lett . * 89 * , 020401 ( 2002 ) .
r. wynar , r. s. freeland , d. j. han , c. ryu , and d. j. heinzen , science * 287 * , 1016 ( 2000 ) .
c. mckenzie _
et al . _ ,
lett . * 88 * , 120403 ( 2002 ) .
w. ketterle , and s. inouye , phys .
lett . * 86 * , 4203 ( 2001 ) .
m. g. moore , and p. meystre , phys .
* 86 * , 4199 ( 2001 ) .
a. g. truscott , k. e. strecker , w. i. mcalexander , g. b. partridge , r. g. hule , science * 291 * , 2570 ( 2001 ) ; f. schreck , g. ferrari , k. l. cubizolles , l. khaykovich , m.o .
mewes , and c. salomon , phys .
a * 64 * , 011402 ( 2001 ) .
b. demarco , and d. s. jin , science * 285 * , 1703 ( 1999 ) .
z. hadzibabic , c. a. stan , k. dieckmann , s. gupta , m. w. zwierlein , a. grlitz , and w. ketterle , phys .
lett . * 88 * , 160401 ( 2002 ) .
g. roati , f. riboli , g. mondugno , and m. ingusco , phys .
* 89 * , 150403 ( 2002 ) .
k. m. ohara , s. l. hemmer , m. e. gehm , s. r. granade , and j. e. thomas , science express report , november 7 , 2002 , report possible evidence of superfluidity in a strongly correlated quantum - degenerate fermi system .
c. p. search , and p. meystre , cond - mat/0210492 b. w. shore , p. meystre , and s. stenholm , j. opt .
b * 8 * , 903 ( 1991 ) .
h. deng and y. yamamoto , unpublished .
a. g. rojo , j. l. cohen , and p. r. berman , phys .
a * 60 * , 1482 ( 1999 ) . c. p. search , h. pu , w. zhang , and p.meystre , phys .
* 88 * , 110401 ( 2002 ) .
g. lenz , p. pax , and p. meystre , phys .
a * 48 * , 1707 ( 1993 ) .
p. villain , p. hberg , l. santos , and m. lewenstein , phys .
a * 64 * , 023606 ( 2001 ) .
y. torii , y. suzuki , m. kozuma , t. sugiura , t. kuga , l. deng and e. w. hagley , phys .
a * 61 * , 041602 ( 2000 ) .
a. f. bernhardt , and b. w. shore , phys .
a * 23 * , 1290 ( 1981 ) .
p. j. martin , b. g. oldaker , a. h. miklich , and d. e. pritchard , phys .
lett . * 60 * , 515 ( 1988 ) .
m. kozuma , l. deng , e. w. hagley , j. wen , r. lutwak , k. helmerson , s. l. rolston , and w. d. phillips , phys .
lett . * 82 * , 871 ( 1999 ) .
i. d. abella , n. a. kurnit , and s. t. hartmann , phys . rev . * 141 * , 391 ( 1966 ) . | we study the dynamics of a beam of fermions diffracted off a density grating formed by fermionic atoms in the limit of a large grating . an exact description of the system in terms of particle - hole operators is developed .
we use a combination of analytical and numerical methods to quantitatively explore the raman - nath and the bragg regimes of diffraction .
we discuss the limits in diffraction efficiency resulting from the dephasing of the grating due to the distribution of energy states occupied by the fermions .
we propose several methods to overcome these limits , including the novel technique of `` atom echoes '' . |
the observational evidence available to date indicates that a major star - forming epoch in a galaxy s history will generate a new set of star clusters that accompanies its population of ` field ' stars .
thus , it is physically meaningful to think of a _ subsystem _ of star clusters as consisting of all clusters formed in a given starburst , and to treat the clusters as a proxy for the stellar subpopulation formed in the same burst ( see de grijs 2010 and larsen 2010 for more extensive treatments of cluster formation in starburst environments ) .
the huge advantage offered by star clusters is that they are easily bright enough to be measured individually within galaxies as distant as 100 mpc and even beyond , and in giant galaxies particularly , they can be found in large numbers ( see figure 1 ) .
we can then construct _ distribution functions _ of such key parameters as mass , age and heavy - element abundance ( metallicity ) for the clusters , instead of just the luminosity - weighted averages that we get from the unresolved field - star light .
the milky way star cluster system ( our starting point for all such work and the ` baseline ' comparison system for all other galaxies ) separates out rather cleanly into the two classic subsystems : the _ open clusters _ ( found throughout the disc and spiral arms along low - eccentricity orbits ) and the _ globular clusters _ ( gcs , inhabiting the galactic bulge and halo in a roughly isotropic distribution of orbits ) . in addition , the gcs are distinctly older than the open clusters ( although with a small range of overlap around @xmath010 gyr ) , as well as more massive and less enriched in heavy elements , indicating that they belonged to a brief early stage of rapid star formation and chemical enrichment .
the open clusters , like the general population of disc field stars , are found at any age but over a more restricted range of metallicities , marking the more gradual ongoing growth of the galaxy s disc .
but in even the nearest external galaxies ( the magellanic clouds , m31 and the other local group galaxies ) , this convenient dichotomy disappears .
the clouds , for example , contain small numbers of classically old , massive , metal - poor gcs as well as many analogues of open clusters , but we also find numerous examples of high - mass , _ young _ clusters that likely resemble gcs as they would have been closer to their formation time .
investigations of star cluster systems in other galaxies reveal still richer varieties , to the stage where every part of the star cluster age / mass / metallicity three - parameter space is occupied . for thorough reviews and perspectives on the earlier literature up to @xmath12005 , interested readers should see harris & racine ( 1979 ) , harris ( 1991 , 2001 ) , ashman & zepf ( 1998 ) and brodie & strader ( 2006 ) .
the present article will concentrate on recent developments in gc system ( gcs ) studies , leading up to a list of currently challenging questions .
this short and biased discussion unfortunately can not do justice to the diversity and richness of approaches now underway , and ( happily ) it will be doomed to be quickly superseded by the rapid advance of both theory and data .
perhaps the most important single implication of the work in this area , however , is that the old gcs represent a common thread in early galaxy evolution , dating back to the first star formation within the pregalactic gas clouds . ) . _
( left ) _ distribution of the blue , metal - poor gcs and _ ( right ) _ the more centrally concentrated distribution of the red , metal - rich gcs .
the circle in each panel has a radius of 100 kpc , projected at the distance of the virgo cluster . ]
some perspective should be offered at this point about the links between models and observations .
this field is at basis a branch of old stellar populations extended to the full range of galaxy types .
gcs studies began as a strongly data - dominated subject , with most discoveries coming directly from photometric and spectroscopic surveys that were guided primarily by the astrophysical intuition of the observers .
quantitative models tended to follow later .
some branches of astrophysics ( e.g. , big - bang cosmology , stellar structure , stellar dynamics ) have clear mathematical foundations accompanied by small numbers of distinctive ` parameters ' that can be invoked to design new observing programmes .
this near - equal cooperation between models and data is not currently the situation here , and a genuine understanding of the formation and evolution of star cluster systems within their parent galaxies is a considerably more complex issue .
it starts on the broad platform of the hierarchically merging galaxy - formation era , and continues with the operation of gas dynamics at all scales , from protogalaxies down to protostars , simultaneously with other key elements including galaxy mergers , satellite accretions and dynamical evolution of compact stellar systems . in the face of this complexity , making the transition from models which are ` scenarios ' ( even if basically accurate ) to ones that are both quantitative and realistic in detail is a steep challenge .
nevertheless , with the rapid advance of high - performance computation and the ability to simulate these processes at a genuinely deep level , we can look forward to major progress on the theory front .
at least some key features of gcs formation have , however , been isolated . at one extreme
lie the smallest dwarf galaxies that have shallow potential wells which can support only a brief initial starburst or else much slower , less dramatic star formation .
these early , low - metallicity _ pregalactic dwarfs _ were probably the formation sites of the metal - poor blue - sequence gcs ( although also leaving behind an intriguing ` specific - frequency problem ' , see below ) ( searle & zinn 1978 ; harris & pudritz 1994 ; burgarella _ et al .
_ 2001 ; beasley _ et al .
_ 2002 ; brodie & strader 2006 ; mashchenko _ et al .
_ 2008 ) . at the other extreme
are the giant and supergiant ellipticals whose gcss are almost certainly composite populations .
they will have gcs originating from _ dissipational collapse _ with huge amounts of gas undergoing star formation spread over several episodes , _ dissipationless accretion _ of satellites coming in later with their own small populations of gcs and _ major merging _ of progenitor disc or elliptical galaxies , with various mixtures of stars and gas .
how important each of these contributions will be for a given galaxy must be driven by its size , environment and the stochastically governed individual hierarchical - merging history .
major attempts to follow these processes simultaneously have been carried out for giant ellipticals in a semianalytic model ( beasley _ et al .
_ 2003 ) and for the milky way in cosmological simulations ( bekki & chiba 2001 ; kravtsov & gnedin 2005 ) , but these are still exploratory and use prescriptive rules for star and cluster formation .
however , syntheses of these approaches are developing ( bekki _ et al .
_ 2008 ) and hydrodynamic simulations of the important cases of gc formation in major mergers and in pregalactic dwarfs are approaching the dynamical range required to resolve individual clusters ( bournaud _ et al .
_ 2008 ; maschenko _ et al .
a key and ( fortunately ) observationally prominent marker of evolution in a stellar population is its heavy - element abundance or _
metallicity_. the increase in mean metallicity with time driven by steady conversion of gas into stars and the cycle of stellar evolution and supernova - driven enrichment is extremely rapid at the earliest times when the gas fractions are highest
. it also seems to be rapid in some of the most violent and shock - driven star - forming environments .
the _ metallicity distribution function _ ( mdf ) is thus one of the most important directly observable tracers of this history .
a major discovery characterizing gcss is that the whole mdf breaks into a strikingly _ bimodal _ form , with a metal - poor mode centred near [ fe / h](mp ) @xmath2 dex and a metal - rich mode centred near [ fe / h](mr ) @xmath3 dex .
a graphic illustration of this two - subpopulation division , as it is most frequently measured , is shown in figure [ cmd_6gal ] .
broadband photometric indices such as in figure 2 for gcs older than @xmath4 gyr become quite insensitive to age and thus the colour reflects the cluster heavy - element abundance surprisingly well . by convention , the metal - poor gcs
are often called the _ blue sequence _ while the metal - richer ones are the _ red sequence_. the translation from gc colour to metallicity ( which is monotonic , but may be nonlinear ) is ultimately established from spectroscopic abundance measurements , which have now been done in several galaxies ( see the work by brodie & huchra 1990 ; cohen _ et al .
_ 1998 , 2003 ; kissler patig _ et al .
_ 1998 ; barmby _ et al . _ 2000 ; perrett _ et al .
_ 2002 ; puzia _ et al .
_ 2005 ; strader _ et al .
_ 2007 ; cenarro _ et al .
_ 2007 ; woodley _ et al .
_ 2009 ; and the fundamental data for the milky way clusters ) .
these two major gc sequences were first identified definitively in the milky way with the work of zinn ( 1985 ) , who demonstrated that the two subpopulations had distinct _
kinematic systemics and spatial distributions _ as well .
the metal - poor cluster subsystem occupies a more extended spatial distribution through the bulge and halo and is kinematically more like an isotropic orbital distribution , while the metal - rich subsystem defines a more centrally concentrated spatial distribution and kinematically contains a higher component of ordered , rotational motion somewhat resembling the field stars in the galactic ` thick disc ' or bulge . ) .
each point marks the total luminosity , @xmath5 , and colour index , @xmath6 , of one cluster .
note the clearly defined blue ( metal - poor ) and red ( metal - rich ) sequences . _
( right ) _ histograms of the colour indices subdivided into three luminosity zones , @xmath7 mag ( top ) , @xmath8 mag ( middle ) and @xmath9 mag ( bottom ) . note the clear bimodality for the lower - luminosity regime , while in the highest - luminosity range the two modes overlap more strongly . ]
dozens of observational papers establishing the bimodal mdfs in other galaxies exist , but key examples include zepf & ashman ( 1993 ) , zepf _ et al . _
( 1995 ) , whitmore _
( 1995 ) , geisler _ et al . _
( 1996 ) , gebhardt & kissler patig ( 1999 ) , neilsen & tsvetanov ( 1999 ) , larsen _ et al . _
( 2001 ) , kundu & whitmore ( 2001 ) , peng et al .
( 2006 ) and harris ( 2009@xmath10 ) .
the factor - of - ten difference in heavy - element abundance between an average metal - poor and a metal - rich gc becomes more sharply defined if more metallicity - sensitive colour indices are used ( cf .
_ 1992 ; geisler _ et al .
_ 1996 ; harris _ et al .
_ 2004 ; rhode _ et al .
_ 2005 ; peng _ et al .
_ 2006 ; spitler _ et al . _ 2008@xmath11 ; harris 2009@xmath10 ) .
direct spectroscopic abundance measurements are best , although these entail far costlier and more resource - intensive data acquisition .
but the clear advantage of photometric samples is that reliable first - order mdfs can be readily obtained for large numbers of galaxies , offering us the ability to track statistical trends across all types of galaxies and environments .
bimodality is now seen so widely that any clear deviations from it within an old gcs would now be regarded as anomalous . if bimodality is the major _ first - order _ feature of the mdf , there are now three _
second - order _ trends providing intriguing additional links to the early enrichment histories of both the host galaxies and the clusters themselves : 1 .
_ mean metallicity and dispersion versus galaxy size . _
the blue and red sequences individually appear at much the same mean metallicity in all galaxies from giants down to dwarfs , and from spirals to ellipticals .
in addition , the intrinsic _ dispersion _ of each sequence ( that is , the rms range in the cluster - to - cluster differences in metallicity ) is roughly constant at @xmath12[fe / h](blue ) @xmath13 dex and @xmath12[fe / h](red ) @xmath14 dex .
the intrinsic dispersion must represent the overall growth of enrichment during a major star - forming period : a cluster forming slightly later in the sequence will start with a larger amount of pre - enrichment in its parent giant molecular cloud . however , a more subtle trend is that the mean metallicity of each sequence increases steadily with host - galaxy luminosity . expressed in terms of heavy - element abundance @xmath15 ,
the scaling relation is @xmath16 ( forbes _ et al .
_ 1997 ; larsen _ et al .
_ 2001 ; strader _ et al .
_ 2004 ; brodie & strader 2006 ; peng _ et al .
+ in qualitative terms , this mean - metallicity scaling indicates that a random gc ( either metal - poor or metal - rich ) drawn from a _ present - day _ dwarf galaxy has a lower enrichment than one drawn from a giant , thus arguing against pure major - merger or dissipationless accretion scenarios in which ( for example ) the blue - sequence clusters now present within a giant galaxy all formed within smaller progenitors later accreted into the bigger central potential .
caveat _ to this view ( see brodie & strader 2006 ) is that an isolated pregalactic dwarf should have a different and more truncated enrichment history than its counterparts lying within the deeper large - scale potential well of an emerging giant elliptical . the key hint from the data ,
however , is that the _ global _ environment of the galaxy ( on scales of @xmath17 kpc ) has influenced the _
local _ formation conditions of its star clusters ( on scales of a few pc ) .
the metal - poor and metal - rich clusters are shown as the solid dots and open triangles , respectively .
the red clusters form a more centrally concentrated subsystem of the galaxy spheroid . _
( right ) _ distribution of gc colour @xmath6 versus galactocentric distance for the luminous gcs in six supergiant ellipticals .
the best - fit power - law curves for the red and blue sequences are shown as the dashed lines , corresponding to a metallicity scaling of @xmath18 . ] 2 . _ the mass / metallicity relation ( mmr ) . _
an unexpected finding of new high - precision photometric surveys of gcss in giant galaxies is a correlation between the mean colours of the blue - sequence gcs and their luminosities ( e.g. , harris _ et al .
_ 2006 ; strader _ et al .
_ 2006 ; mieske _ et al .
_ 2006 ; spitler _ et al .
_ 2006 ; forte _ et al .
_ 2007 ; harris 2009@xmath10 ; cockcroft _ et al .
_ 2009 ; peng _ et al .
effectively , the blue gcs become progressively more heavy - element enriched at higher cluster mass .
the exact form of the correlation was a matter of debate in the first round of papers , but a consensus has been emerging around two conclusions : ( i ) the blue sequence is essentially vertical for the less massive clusters in the range @xmath19 m@xmath20 , that is , @xmath21 for low - mass gcs .
but at progressively higher @xmath22 ( extending up to the most luminous gcs known at @xmath23 l@xmath20 ) , the blue sequence bends towards redder colours , eventually following a slope near @xmath24 at its top end ( harris 2009@xmath10 ; peng _ et al .
( ii ) no such trend has been found along the _ red _ sequence , which is indistinguishable from vertical in all systems studied so far .
detecting this mmr clearly requires internally precise photometry and very large gc samples because clusters above the @xmath25 m@xmath20 level ( where the mmr becomes most noticeable ) are rare and the changes in mean colour are not large ( see again figure [ cmd_6gal ] ) .
+ the mmr is , at this point , thought to be the result of some form of self - enrichment of massive clusters during their formation period ( strader & smith 2008 ; bailin & harris 2009 ) .
a sufficiently massive and dense protocluster ( @xmath26 m@xmath20 of gas within a 1 pc protocluster core ) can provide a sufficiently deep potential well and enough dense gas to hold back a high fraction of the enriched supernova ejecta from the first round of massive stars within the protocluster , which can then enrich the still - forming lower - main - sequence stars .
for this mechanism to work , the entire starburst needs to take 2030 myr to complete . for the highest - mass gcs , the net enrichment over and above its _ _
pre-__enriched metallicity will be @xmath27 z@xmath20 , large enough to double its initial @xmath28 z@xmath20 along the blue sequence .
but the lower - mass clusters can not be significantly self - enriched because the supernova heating of the surrounding protocluster ends up ejecting almost all gas not converted to stars .
interestingly , this model predicts that the phenomenon should affect the metal - rich gc sequence too , but it would be almost undetectable on the red sequence because the incremental @xmath29 is always small relative to their pre - enriched level , @xmath30 z@xmath20 .
metallicity gradients .
_ the mean metallicity of the entire gcs in large galaxies has long been known to decrease outward in the halo .
however , most of this global gradient is actually caused by the changing proportions of red versus blue clusters with changing galactocentric distance , @xmath31 , that is , it is the result of a _ population gradient _
( e.g. , zinn 1985 ; harris _ et al .
_ 1992 ; geisler _ et al .
_ 1996 ; rhode & zepf 2004 ; harris _ et al .
one of the more well - defined but not atypical examples is shown in figure [ m87rad ] for the virgo giant elliptical m87 ( tamura _ et al .
_ 2006 ; harris 2009@xmath32 ) .
notably , the metal - rich gcs follow a spatial distribution that mimics fairly closely the similarly metal - rich spheroid light of the host galaxy , suggesting that they formed together .
the same view is supported by the details of the mdf for both clusters and field - halo stars ( see below ) . by contrast
, the shallower spatial distribution of the blue gcs more closely resembles that of the isothermal dark - matter potential well , or is intermediate between the dark matter and the halo light , consistent with the interpretation that the blue gcs formed at quite an early stage . + but at a finer level , true metallicity gradients can be found _ within _ each of the two subpopulations .
that is , the red and blue gcs show an intrinsic decrease in mean metallicity with increasing @xmath31 , scaling as @xmath33 ( see figure 3b and geisler _ et al .
_ 1996 ; lee _ et al .
_ 2008 ; harris 2009@xmath10 ) . in the hierarchical - merging picture
, the mean gc metallicity would reflect the depth of the larger - scale potential well within which it was formed , being able to reach a higher level of pre - enrichment deeper in to galaxy centre .
later accretions of dwarf satellites , which have more metal - poor gcs , would add to the population predominantly in the outskirts of the halo .
a potential but important complication is that later major mergers after formation may have diluted the initial metallicity gradient ( or , if the merger brought in a very large amount of gas , it may have helped rebuild the gradient for the metal - richer clusters only ) .
the milky way and its nearby satellites ( the magellanic clouds , sagittarius and fornax ) contain the only populations of gcs for which truly fundamental age calibrations can be achieved , based on isochrone fitting to the deep , unevolved main sequence and even the white - dwarf sequence .
recent published work ( e.g. , gratton _ et al . _ 2003 ; de angeli _ et al .
_ 2005 ; hansen _ et al .
_ 2007 ; marn franch _ et al .
_ 2009 ; among many others ) indicates that the absolute age of the metal - poor clusters is near 13 gyr and shows a cluster - to - cluster dispersion that may be as low as @xmath34 gyr , confirming their classic status as among the first stellar structures to have formed in the galaxy .
the mean age for the metal - richer subsystem is perhaps 2 gyr younger than the extremely old , metal - poor subsystem , and exhibits a higher cluster - to - cluster age scatter , near @xmath35 gyr rms .
direct age measurements for gcs in other , more distant systems are extremely important but inevitably less precise .
a method of attack that has gained some valuable ground in the past decade has been the use of spectroscopic line - strength parameters from the integrated light of gcs , which are sensitive to different combinations of age , @xmath36 , metallicity , [ fe / h ] , and @xmath37-abundance ratio , [ @xmath37/fe ] . because the balmer lines are affected strongly by the faint but numerous stars near the main - sequence turnoff , they are sensitive to age , while the many different metal lines ( fe , ca , mg , etc . ) are more exclusively influenced by metallicity . from high
signal - to - noise ratio spectra at moderate resolution , age determinations precise to @xmath35 gyr ( for the ` old ' age regime , @xmath38 gyr ) can be achieved .
key case studies can be found in proctor _
( 2004 ) , puzia _ et al . _
( 2005 ) , pierce _ et al . _
( 2006 ) , beasley _
( 2008 ) and woodley _
( 2009 ) , among several others .
a recent sample for ngc 5128 is shown in figure [ ages_2gal ] . direct comparison with the milky way clusers
indicates that the ngc 5128 gcs exhibit a comparable @xmath37-abundance distribution , along with the familiar bimodal mdf .
however , the age distribution for the ngc 5128 gcs clearly extends to younger ages , with a hint of major starbursts having occurred at ages of @xmath39 , 9 and 5 gyr .
by contrast , on the same age - calibration scale , the milky way gcs range exclusively from 10 to 13 gyr .
roughly similar age patterns show up in other large ellipticals ( see the references cited above ) .
although the _ absolute _ age scale for these spectroscopic - index programmes is still a work in progress , the data so far strongly confirm that large ellipticals are composite systems which assembled in steps from very high redshift down to @xmath40 .
a surprisingly effective outcome of the _ hubble space telescope ( hst ) _ imaging programmes for nearby galaxies and their gcss has been the ability to measure accurately the _ scale sizes _ of the gcs , i.e. , their effective or half - light radii , @xmath41 or @xmath42 .
this is a dynamically valuable quantity because it stays nearly invariant over many internal relaxation times and thus is one of the best signatures we have of the linear size of the cluster shortly after it formed its stars and ejected the remaining protocluster gas . for galaxies within the local group ,
the structure of the cluster can be resolved to within its core radius , @xmath43 , ( barmby _ et al .
_ 2007 ) and rather complete profile fitting is readily possible . at increasing distances , the linear size of a gc
gradually shrinks relative to the @xmath44 _ hst _ resolution , but the crucial scale radius @xmath42 can be accurately measured for galaxies out to 50 mpc ( major databases of this type are in jordn _
et al . _ 2005
; harris 2009@xmath11 ) by appropriate convolution with the stellar point - spread function .
four representative sets of gc scale - radius data are shown in figure [ sizes ] : gcs in the supergiant elliptical m87 ( madrid _ et al .
_ 2009 ) , the giant sa disc galaxy m104 ( harris _ et al .
_ 2009 ) , a compendium of gcs in nearby dwarf ellipticals ( des , georgiev _ et al .
_ 2008 ) and the milky way clusters ( data from harris 1996 ) .
the fundamental similarities among these very different galaxies are striking ( see , for example , jordn _ et al .
_ 2005 , who explore their potential as a ` standard ruler ' ) . the asymmetric tail to larger
@xmath42 is the most noticeable second - order difference .
clusters with @xmath45 pc are found predominantly in the dwarf galaxies and ( for the milky way ) among the lower - mass clusters in the outskirts of the halo , apparently reflecting their gentler tidal - field environment from birth .
et al . _ ( 2009 ) suggest that the distribution may be bimodal , with a secondary mode at @xmath46 pc found in the dwarfs .
the intrinsic shape of the size distribution has been suggested to result ultimately from a stochastic range in star - formation efficiencies starting from a rather narrow range of initial protocluster sizes ( harris _ et al . _
2009@xmath11 ) , followed by a longer phase of dynamical evolution which shapes both the present - day size and mass distributions ( e.g. , fall & rees 1977 ; gieles & baumgardt 2008 ) .
more quantitatively , gc structural parameters have been used to define a fundamental plane analogous to that for elliptical galaxies ( djorgovski 1995 ) .
their various structural properties are in fact so tightly correlated that their structures are almost completely specified by just two parameters , mass and ( secondarily ) central concentration ( mclaughlin 2000 ) . a simple version of this manifold is shown in figure [ sizes]b , where gc scale size is plotted versus luminosity for three massive disc galaxies .
gcs do not follow the scaling relation @xmath47 that typifies only slightly more massive , compact stellar systems such as de nuclei and ultracompact dwarf ( ucd ) galaxies ( e.g. , evstigneeva _ et al .
_ 2008 ; mieske _ et al .
_ 2008 ) . instead , for gcs less massive than @xmath48 m@xmath20
, we find @xmath49 .
a long - standing question has been whether or not the gcs are truly distinct stellar systems or if there is any kind of ` bridge ' connecting them to the lower end of the de / ucd manifold .
the recent evidence , drawn from galaxies with the richest gc populations in which we can find significant numbers of high - mass clusters , is that the radii start to increase for @xmath50 m@xmath20 and may merge fairly seamlessly onto that upper manifold . , for old gcs in four systems : the giant elliptical m87 , the giant sa m104 , a combination of nearby dwarf galaxies and the milky way . _ ( right ) _
@xmath51 versus cluster luminosity for gcs in three massive disc galaxies : the milky way ( large filled circles ) , m31 ( triangles ) and m104 ( small crosses ) .
the median size , @xmath52 pc , is shown as the dashed line . ]
globular clusters that we see _ today _ in the haloes of galaxies of all types have typical masses of @xmath53 m@xmath20 , but the mass of their original protocluster is likely to have been substantially larger ( and more so at higher mass ) . the raw star - formation efficiency within the protocluster
is expected to be @xmath54% .
after formation it undergoes continual mass loss dominated ( at early times ) by supernovae and stellar winds from its massive stars , and at later times by stellar evaporation and tidal truncation , reducing its total mass after 12 gyr by another factor of two or more depending on initial mass ( baumgardt & makino 2003 ; baumgardt & kroupa 2007 ; mclaughlin & fall 2008 ) . at a broader galaxy - wide level , the high - resolution simulation of bournaud _ et al . _
( 2008 ) indicates that about 4% of the gas in a relatively strong major merger is converted into small , dense clouds that can be plausibly identified as protoglobular clusters . combining these arguments
then suggests that present - day galaxy haloes should have @xmath55% of their stellar mass in the gcs . for comparison , the most thorough observational estimates of the mass fraction @xmath56 ( mclaughlin 1999 ; spitler _ et al .
_ 2008@xmath32 ; peng _ et al .
_ 2008 ) place @xmath57 in the range 0.11% of the total baryonic mass of a large galaxy , correlating weakly with the luminosity of the galaxy itself and also accounting for any nonstellar mass in x - ray halo gas . for dwarf galaxies
@xmath58 m@xmath20 , this mass fraction scatters much more , ranging from zero to above 3% .
spitler & forbes ( 2009 ) argue that @xmath57 may be most tightly determined by the total dark - matter halo mass of the parent galaxy except , again , for the smallest dwarfs .
recent simulations show that protoclusters of the right mass scale to resemble gcs will form in both the low - metallicity environment of pregalactic dwarfs ( mashchenko _ et al .
_ 2008 ) and in gas - rich mergers of large disc galaxies ( bournaud _ et al .
_ 2008 ) . but
all of the contemporary analyses lead to the conclusion that it takes a very large reservoir of gas to make a typical gc .
the actual conversion efficiency ( essentially , the ratio of gc protocluster mass to its local host giant molecular cloud mass ) may depend on merger shock velocities , metallicities and other environmental factors , and needs to be much better understood than it is now . _
specific frequency _
@xmath59 is the number of gcs per unit galaxy luminosity and is the most obvious observational proxy for gc formation efficiency .
much of the original puzzle over the order - of - magnitude differences in @xmath59 between ( say ) spirals and ellipticals , or cd ellipticals versus field ellipticals ( e.g. , harris 2001 ) , has been alleviated by focusing on the mass ratio instead ( see above ) .
the possibility that most of the metal - rich clusters in elliptical galaxies formed through major mergers ( ashman & zepf 1992 ) does not , by itself , solve the original ` @xmath59 problem ' because such large amounts of incoming gas are needed to make the thousands of gcs in giant ellipticals at any plausible formation efficiency that this version of their origin becomes similar to basic hierarchical merging ( cf .
harris 2001 ) .
the @xmath59 values calculated separately for the red and blue gcs vary systematically and nonlinearly with galaxy luminosity , a pattern thoroughly discussed by peng _
( 2008 ) .
a newer ` specific - frequency problem ' is exemplified in figure [ n5128halo ] .
the nearby giant elliptical system ngc 5128 offers a rare opportunity to compare the mdfs of the gcs _ and _ the halo red - giant stars in the same galaxy .
our zeroth - order assumption is that a major formation episode in a galaxy s history should produce gcs in direct proportion to the amount of gas .
if so , the mdfs of both types of stellar populations should have the same shape , being merely scaled versions of each other .
this is plainly not the case ( see figure 6 ) : there are about five times fewer metal - poor halo stars _ per unit gc at the same metallicity _ than there are metal - rich halo stars . the analyses for this and other galaxies ( e.g. , forte _ et al .
_ 2005 ) show that metal - rich gcs match the mean metallicity and dispersion of the halo giants well enough to further support the default interpretation that they formed along with the bulk of the galaxy s spheroid .
but where are the many metal - poor stars that should go along with the metal - poor clusters ?
interestingly , similarly high @xmath59 values can be found in some low - luminosity des , whose gcs are predominantly metal - poor ( peng _ et al . _ 2008 ) .
one possible solution is that the metal - poor halo stars are present , but they occupy a much shallower spatial distribution than the metal - rich spheroid , just as the low - metallicity clusters form a spatially more extended subsystem than the metal - rich clusters .
the metal - poor field halo would become dominant only at larger galactocentric distances than have normally been searched ( see harris _
_ 2007 for some tantalizing evidence along those lines ) .
another is a timing argument : if the massive , dense protoclusters formed earliest in the pregalactic clouds , and if later star formation were partially truncated ( perhaps by the epoch of reionization ) , then @xmath59(blue ) would end up artificially high ( harris & harris 2002 ) .
a step towards a more quantitative interpretation has been developed by peng _
( 2008 ) , who argue that the gc formation rate will scale with both the star - formation rate and the star - formation - rate density , thus boosting cluster formation at high redshifts .
much remains to be discovered about the determining conditions of gc formation .
the study of gc populations in galaxies provides a unique observational window on galaxy formation .
a shortlist of the questions that have emerged from recent work would certainly include the following : 1 . what is the physical cause of bimodality
? it does not automatically emerge from current semianalytic or cosmological simulations without deliberately inserting an external truncation mechanism such as reionization ( beasley _ et al .
_ 2002 ; kravstov & gnedin 2005 ) . was the formation of intermediate - metallicity clusters truncated near @xmath60 , _ in the same way in all major galaxies ? _ or were those clusters present but somehow biased towards smaller masses , which would then have been removed more easily by dynamical evolution ?
2 . why was the formation of metal - poor , blue gcs so efficient relative to the ` normal ' , red gcs ?
do the outermost parts of galaxy haloes possess large numbers of metal - poor stars ? more generally , what features of the gas dynamics and composition inside a giant molecular cloud determine the formation efficiency of massive clusters ?
the formation of a giant elliptical galaxy may have finished with a few major , gas - rich mergers , each of which generated metal - rich clusters .
thus , the red gc ` sequence ' is , in itself , probably a composite population . is there a way to deconvolve the red sequence into its age / metallicity components ? 4 .
` field ' ellipticals typically have low specific frequencies and are the most likely to have formed from late major mergers . but
these are very incompletely surveyed by comparison with the gc - rich ellipticals in the virgo , fornax and abell clusters .
do they show the same features of bimodality , the same gc luminosity functions and the same proportions of red / blue clusters as the rich ellipticals ?
gcs more massive than @xmath25 m@xmath20 belong to an intriguing ` transition region ' characterized by an mmr , increasing scale size , higher mass - to - light ratios and ( as we see in massive milky way gcs ) multiple internal stellar populations .
what drives these systematic changes at the high - mass end of the gc sequence ?
cluster mergers ? increasing importance of dark matter ? can all compact stellar systems , from gcs through ucds and elliptical galaxies , eventually be placed onto a single astrophysical sequence ? 6 .
what is the total evolution of a massive star cluster from birth to death ? a comprehensive evolutionary model needs to include the gas dynamics of its formation within a giant molecular cloud , early rapid gas - mass loss , ongoing internal dynamical evolution , slower secular evolution including the effects of the external tidal field and eventual dissolution into the field .
pieces of these steps are understood , sometimes in much detail , but a comprehensive end - to - end story still awaits assembly .
barmby , p. , mclaughlin , d. e. , harris , w. e. , harris , g. l. h. & forbes , d. a. 2007 structural parameters for globular clusters in m31 and generalizations for the fundamental plane .
j. _ , * 133 * , 27642786 .
beasley , m. a. , bridges , t. , peng , e. , harris , w. e. , harris , g. l. h. , forbes , d. a. & mackie , g. 2008 a 2df spectroscopic study of globular clusters in ngc 5128 : probing the formation history of the nearest giant elliptical , _ mon . not .
r. astron .
_ , * 386 * , 14431463 .
cockcroft , r. , harris , w. e. , wehner , e. m. h. , whitmore , b. c. & rothberg , b. 2009 further definition of the mass - metallicity relation in globular cluster systems around brightest cluster galaxies , _ astron .
j. _ , * 138 * , 758769 .
evstigneeva , e. a. , drinkwater , m. j. , peng , c. y. , hilker , m. , depropris , r. , jones , j. b. , phillipps , s. , gregg , m. d. & karick , a. m. 2008 structural properties of ultra - compact dwarf galaxies in the fornax and virgo clusters . _ astron . j. _ ,
* 136 * , 461478 .
georgiev , i. y. , puzia , t. h. , hilker , m. & goudfrooij , p. 2009
globular cluster systems in nearby dwarf galaxies .
hst_/acs observations and dynamical properties of globular clusters at low environmental density .
_ mon . not .
r. astron .
_ , * 392 * , 879893 .
harris , w. e. , whitmore , b. c. , karakla , d. , okon , w. , baum , w. a. , hanes , d. a. & kavelaars , j. j. 2006 globular cluster systems in brightest cluster galaxies : bimodal metallicity distributions and the nature of the high - luminosity clusters .
_ astrophys .
j. _ , * 636 * , 90114 .
jordn , a. _ et al .
_ 2005 the acs virgo cluster survey .
x. half - light radii of globular clusters in early - type galaxies : environmental dependencies and a standard ruler for distance estimation .
_ astrophys .
j. _ , * 634 * , 10021019 .
lee , m. g. , park , h. s. , kim , e. , hwang , h. s. , kim , s.,c . & geisler , d. 2008 washington ccd photometry of the globular cluster system of the giant elliptical galaxy m60 in virgo .
_ astrophys .
j. _ , * 682 * , 135154 .
peng , e. w. _ et al .
_ 2008 the acs virgo cluster survey .
the formation efficiencies of globular clusters in early - type galaxies : the effects of mass and environment .
_ astrophys .
j. _ , * 681 * , 197224 .
peng , e. w. , jordn , a. , blakeslee , j. p. , mieske , s. , ct , p. , ferrarese , l. , harris , w. e. , madrid , j. p. & meurer , g. r. 2009 the color magnitude relation for metal - poor globular clusters in m87 : confirmation from deep _ hst_/acs imaging . _
astrophys . j. _
, * 703 * , 4251 .
perrett , k. m. , bridges , t. j. , hanes , d. a. , irwin , m. j. , brodie , j. p. , carter , d. , huchra , j. p. & watson , f. g. the kinematics and metallicity of the m31 globular cluster system .
j. _ , * 123 * , 24902510 .
puzia , t. h. , kissler
patig , m. , thomas , d. , maraston , c. , saglia , r. p. , bender , r. , goudfrooij , p. & hempel , m. 2005 vlt spectroscopy of globular cluster systems .
ii . spectroscopic ages , metallicities , and [ @xmath37/fe ] ratios of globular clusters in early - type galaxies .
_ , * 439 * , 9971011 .
spitler , l. r. , forbes , d. a. & beasley , m. a. 2008@xmath11 extending the baseline : _ spitzer _ mid - infrared photometry of globular cluster systems in the centaurus a and sombrero galaxies .
_ mon . not .
r. astron .
_ , * 389 * , 11501162 .
spitler , l. r. , forbes , d. a. , strader , j. , brodie , j. p. & gallagher iii , j. s. 2008@xmath32 the connection between globular cluster systems and their host galaxy and environment : a case study of the isolated elliptical ngc 821 .
_ mon . not .
r. astron .
_ , * 385 * , 361380 .
spitler , l. r. , larsen , s. s. , strader , j. , brodie , j. p. , forbes , d. a. & beasley , m. a. 2006 _ hubble space telescope _
acs wide - field photometry of the sombrero galaxy globular cluster system .
j. _ , * 132 * , 15931609 .
strader , j. , brodie , j. p. , spitler , l. & beasley , m. a. 2006 globular clusters in virgo ellipticals : unexpected results for giants and dwarfs from advanced camera for surveys imaging .
j. _ , * 132 * , 23332345 .
tamura , n. , sharples , r. m. , arimoto , n. , onodera , m. , ohta , k. & yamada , y. 2006 a _ subaru_/suprimecam wide - field survey of globular cluster populations around m87 .
ii . colour and spatial distribution .
_ mon . not .
r. astron .
_ , * 373 * , 601612 .
whitmore , b. c. , sparks , w. b. , lucas , r. a. , macchetto , f. d. & biretta , j. a. 1995 _ hubble space telescope _
observations of globular clusters in m87 and an estimate of @xmath62 , _ astrophys . j. _ ,
* 454 * , l73l76 .
woodley , k. a. , harris , w. e. , gmez , m. , puzia , t. h. , harris , g. l. h. & geisler , d. 2009 the metallicities , ages , and alpha element enhancements of globular clusters in the elliptical ngc 5128 : a homogeneous spectroscopic study with _
gemini_/gmos , _ astrophys .
j. _ , submitted . | * globular clusters ; stellar populations ; galaxy formation * the ensemble of all star clusters in a galaxy constitutes its _ star cluster system_. in this review , the focus of the discussion is on the ability of star clusters , particularly the systems of old massive globular clusters ( gcss ) , to mark the early evolutionary history of galaxies .
i review current themes and key findings in gcs research , and highlight some of the outstanding questions that are emerging from recent work .
[ firstpage ] |
there is much current interest in exploiting morphological instabilities to form periodic nanoscale patterns at crystal surfaces @xcite .
because of their natural in - plane anisotropy , vicinal surfaces @xcite prepared at a miscut relative to a high symmetry orientation provide ideal substrates for the formation of ripple patterns parallel or perpendicular to the mean step orientation @xcite . here
we specifically consider patterns formed by _
step bunching _ , the process in which a train of initially equally spaced ( straight ) steps splits into regions of densely packed steps ( step bunches ) , and almost flat regions @xcite . bunched semiconductor surfaces are promising templates for the growth of metallic nanowires @xcite .
step bunching can be induced by growth @xcite , sublimation @xcite , or surface migration of adatoms driven by an electric current @xcite .
the common feature of the different instability mechanisms @xcite is that they break the symmetry between the ascending and descending steps bordering a terrace .
the appearance of step bunches thus provides information about the asymmetry of the attachment / detachment processes at the steps , as well as about the direction of current - induced adatom migration .
once formed , the shape of a bunch is determined by the balance between the destabilizing forces and the repulsive step - step interactions that act to regularize the step train . as a result ,
the bunch shape displays characteristic scaling laws relating e.g. its slope and width to the number of steps in the bunch @xcite .
these scaling laws are used in the interpretation of experiments to extract the functional form of the step interactions as well as material parameters such as the step interaction strength and the electromigration force @xcite .
the large scale properties of step bunches are captured by continuum evolution equations for the surface profile @xcite , which can be derived from the underlying discrete step dynamics in a systematic manner @xcite .
the analysis of static ( time - independent ) solutions of these equations leads to scaling laws which are in reasonable agreement with numerical simulation of the discrete step dynamics @xcite .
however , in the presence of a non - vanishing sublimation or growth flux , step bunches are moving objects .
because of the high temperatures involved , sublimation and hence , bunch motion is significant also in electromigration experiments , where it is not the primary cause of bunching @xcite . in this letter
we show that bunch motion alters the shape and scaling properties of bunches in a fundamental way .
it removes the artificial symmetry between the in - flow and out - flow regions ( in which steps move into and out of the bunch , respectively ) and the concomitant singularities of the static solutions at the bunch edges @xcite .
we show that the lateral speed of a bunch is inversely proportional to its height for a large class of models , and we identify the size of the largest terrace @xmath0 as a natural scaling parameter , in terms of which other important bunch characteristics are expressed in a simple way .
the maximal terrace size @xmath0 is uniquely defined , in contrast to the number of steps in the bunch , which requires a convention to decide which steps belong to it , and it is directly accessible experimentally by means of reflection electron microscopy ( rem ) @xcite .
we consider a system of non - transparent steps @xcite described on the discrete level by the equations of motion @xmath1 for the step positions @xmath2 , where the time scale has been normalized to the growth or sublimation flux .
the parameter @xmath3 governs the asymmetry between ascending and descending steps , relative to the mean step velocity .
the linear form of the first two terms on the right hand side of ( [ discrete_time_evolution ] ) is characteristic of slow attachment / detachment kinetics , and applies equally to step bunching induced by sublimation , growth or surface electromigration @xcite ; here we will assume a sublimating step train going uphill in the @xmath4 direction .
the last term on the right hand side of ( [ discrete_time_evolution ] ) represents stabilizing step - step interactions of strength @xmath5 . in the usual case of entropic or dipolar elastic interactions @xmath6 where @xmath7 and @xmath8 is the average terrace length @xcite .
explicit expressions for @xmath3 and @xmath5 in terms of physical parameters are given below in ( [ parameters ] ) .
for @xmath9 , ( [ discrete_time_evolution ] ) leads to an instability of the equally spaced step configuration @xmath10 and its segregation into step bunches separated by flat regions
. the bunches coarsen slowly in time by coalescence .
we are interested in the final regime of coarsening with a few big bunches left in the system . in this regime , one can study a periodic array of identical bunches , each containing @xmath11 steps , which satisfy ( [ discrete_time_evolution ] ) with @xmath12 and the helicoidal boundary conditions @xmath13 .
it is convenient to consider the comoving step coordinates @xmath14 , in which the center of mass of the step configuation does not move . in this frame ,
the stationary trajectory of a step is a periodic function with some ( unknown ) period @xmath15 , @xmath16 .
stationarity implies that every step follows the same trajectory , up to a space and time shift , according to @xmath17 with @xmath18 . inserting this into ( [ discrete_time_evolution ] ) and setting @xmath19
we obtain an equation for the stationary step trajectory ( in the following we omit subscripts ) @xmath20 . \label{stationary_discrete}%\ ] ] this is a differential - difference equation for two periodic functions @xmath21 and @xmath22 , which for the time being will be treated as independent . expanding the functions in fourier series with frequencies @xmath23 and coefficients @xmath24 and @xmath25 , respectively , we obtain from ( [ stationary_discrete ] ) @xmath26 .
\label{omega_n}%\]]since ( [ omega_n ] ) is valid for any @xmath27 , we can choose @xmath28 for large @xmath11 and expand ( [ omega_n ] ) to obtain the expression @xmath29+o(m^{-1 } ) , $ ] which in fact determines the dependence of the bunch velocity on @xmath11 : in the laboratory frame we have @xmath30 , which implies that the whole step configuration shifts by @xmath31 to the right in time @xmath32 .
hence the lateral bunch speed @xmath33 is given by @xmath34 where @xmath35 $ ] has to be determined self - consistently for a given form of step - step interaction .
when @xmath36 has the usual form ( [ step - step interaction ] ) with @xmath7 , we find numerically that @xmath37 is a constant proportional to the asymmetry , @xmath38 , provided that the asymmetry is not too large , @xmath39 . in the following
we will see that even though the bunch velocity decreases with increasing bunch size , it can not be neglected . according to a scaling argument due to chernov @xcite
, the scaling @xmath40 implies that the average bunch size should increase with time as @xmath41 , which is consistent with experiments @xcite and numerous discrete simulation @xcite .
the continuum evolution equation corresponding to the discrete dynamics ( [ discrete_time_evolution ] ) reads @xcite @xmath42 + h_0 = 0 , \label{pde}%\ ] ] where @xmath43 is the surface profile , @xmath44 is the slope , and @xmath45 denotes the height of a single step .
a periodic array of bunches moving at lateral speed @xmath33 is obtained by the travelling wave ansatz @xcite @xmath46 , where @xmath47 and the function @xmath48 satisfies the boundary condition @xmath49 . inserting this into ( [ pde ] ) we find that the vertical excess speed @xmath50 is related to @xmath33 by @xmath51 , where @xmath52 is the average slope .
integrating once the ordinary differential equation for @xmath48 then becomes @xmath53 we denote derivatives by primes , and measure lateral distances in units of @xmath8 and heights in units of @xmath54 .
the phase shift @xmath55 is a constant of integration satisfying the condition @xmath56 d\xi=0 $ ] , which follows by multiplying ( [ ode ] ) by @xmath57 , integrating over the bunch period , and using the boundary conditions . for future reference
we note that , for small @xmath3 , @xmath58 because of ( [ vlat ] ) . in fig.[fig_h(ksi ) ] a numerical solution of ( [ ode ] ) obtained via a shooting method is compared to the discrete step dynamics , showing excellent agreement .
the continuum description is generally found to work very well , provided the asymmetry @xmath3 is sufficiently small . in fig.[fig_singular_m ] we compare the corresponding slope profile to the time - independent solution derived in @xcite by setting the terms inside the square brackets in ( [ pde ] ) to zero and neglecting the symmetry - breaking term @xmath59 . as was anticipated in @xcite ,
the moving bunch is distinctly asymmetric .
moreover the moving solution extends smoothly over the whole @xmath60-axis , whereas the static solution has finite support due to singularities at the bunch edges . in the following
we find analytically the asymptotics of the inflow and outflow regions of the moving bunch , corresponding to the extreme left and right parts of fig.[fig_h(ksi ) ] ( see also fig.[fig_singular_m ] ) .
in this region the steps are far apart and their interaction @xmath5 is negligible . it is convenient to perform a lagrange transform from the function @xmath48 to its inverse @xmath63 . after some algebra ( [ ode ] ) with @xmath64 then reduces to the linear equation @xmath65 which has the general solution @xmath66 with @xmath67.$ ] fixing the boundary conditions so that the point of minimal slope @xmath68 is located at @xmath69 we have two boundary conditions @xmath70 and @xmath71 to determine @xmath72 .
recalling that @xmath73 , we see that @xmath74 and @xmath75 for large bunches , so that ( [ complete_outlow_solution ] ) becomes a pure exponential , corresponding to a slope profile @xmath76 .
similar behavior was found in a model with short - range step interaction , however in that case @xmath77 and @xmath78 for large bunches @xcite .
in this region one can neglect the first two terms in ( [ ode ] ) , as can be shown by a careful analysis of ( [ forces_distribution ] ) .
the remaining terms give @xmath81 integrating once , we find @xmath82 by requiring the derivative @xmath83 to vanish at the point with minimal slope @xmath84 . integrating again , we obtain an implicit equation for @xmath85 , @xmath86 = \xi . \label{singularity_inflow_equation_integrated2}%\ ] ] note that ( [ singularity_inflow_equation_integrated2 ] ) is valid for @xmath87 , to avoid the logarithmic singularity at @xmath88 . in reality
there is no singularity , because additional terms from ( [ ode ] ) have to be included when @xmath89 , which however are completely irrelevant in the remaining part of the inflow region .
we are now prepared to investigate the scaling properties of large step bunches ( @xmath90 ) .
\(a ) the easiest is to find the _ size of the first terrace in the bunch _
@xmath91 , defined as in @xcite by @xmath92 , since this region is well described by ( [ singularity_inflow_equation_integrated2 ] ) .
away from the singularity , for @xmath93 , ( [ singularity_inflow_equation_integrated2 ] ) reduces to @xmath94 , which is of a similar form as the pokrovsky - talapov singularity found for static bunches @xcite .
this yields immediately @xmath95 \(b ) to estimate the _ size of the minimal terrace _ in the bunch @xmath96 , we multiply ( [ ode ] ) by @xmath97 and integrate from @xmath98 to @xmath99 : @xmath100 _ { \xi_1}^{\xi_2}= -2\omega% { \displaystyle\int\limits_{\xi_1}^{\xi_2 } } m^{2}m^{\prime}\left ( \xi+\xi_{0}-h\right ) d\xi+\frac{1}{3}% { \displaystyle\int\limits_{\xi_1}^{\xi_2 } } \frac{(m^{\prime})^{2}}{m}d\xi . \label{forces_distribution}%\ ] ] setting @xmath101 , @xmath102 and @xmath103 with @xmath104 , the left hand side gives @xmath105 for @xmath106 .
the second integral on the right hand side can be taken , noting that the function @xmath107 vanishes everywhere except in the narrow inflow region , where ( [ singularity_inflow_equation_integrated2 ] ) holds , yielding @xmath108_{0}^{\xi_{\max}}\approx ( 36u\varepsilon^{2 } ) ^{-1 } \equiv i_{0}$ ] . to estimate the first integral , first set @xmath109 in ( [ forces_distribution ] ) so that the left hand side vanishes due to the periodic boundary conditions .
we obtain then @xmath110 = 2\omega\int_{0}^{m}m^{2}m^{\prime}\left ( \xi+\xi_{0}-h\right ) d\xi\approx i_{0}/3 $ ] . denoting by @xmath111 /i_{\omega}\left [ 0,m\right ]
$ ] the relative contribution to the integral from the segment @xmath112 $ ] for large @xmath11 , we find from ( [ forces_distribution ] ) @xmath113 , or @xmath114 numerically we observe that the value of @xmath115 indeed saturates to a fixed value for large @xmath11 , and depends rather weakly on @xmath3 and @xmath5 .
varying @xmath3 and @xmath5 around physically relevant choices of parameters , e.g. , those in fig.[fig_h(ksi ) ] or in @xcite , we find @xmath116 .
\(c ) _ bunch width _ @xmath117 .
the definition of the bunch width depends on the convention used to assign steps to the bunch @xcite .
here we define the bunch as the collection of terraces with sizes smaller than the mean terrace size @xmath8 .
we can obtain an estimate of @xmath117 integrating the first term on the right hand side of ( [ forces_distribution ] ) by parts : @xmath118 , and arguing that the dominant contribution to the integral comes from the bunch interior .
then @xmath119 , where @xmath120 , and we have used @xmath73 . substituting @xmath121
we get @xmath122 for the parameters of fig.[fig_h(ksi ) ] , @xmath123 for large @xmath11 , and it changes only slightly ( @xmath124 ) under significant variations of @xmath3 and @xmath5 .
\(d ) _ the minimal slope . _
as we have seen , many characteristics of the moving bunch are controlled by the single parameter @xmath84 , which can be defined microscopically as the inverse size of the _ largest terrace in the outflow region _ between two consecutive bunches , @xmath125 . in order to make a connection to earlier studies , we need the dependence of @xmath84 on the bunch size @xmath11 . from the asymptotic slope profile @xmath126 in the outflow region ,
one expects @xmath127 to leading order .
numerical studies suggest however strong finite size corrections even for large bunches , @xmath128 .
the behaviour of @xmath84 is rather well approximated by @xmath129 where @xmath130 and typically @xmath131 is not small , e.g. , for @xmath132 from fig.[fig_h(ksi ) ] , @xmath131 @xmath133 in the range of bunch sizes @xmath134 .
if we nevertheless use the asymptotic expression @xmath135 in ( [ first_terrace_size])-([w ] ) we recover the scaling laws derived in @xcite for static bunches , however with different numerical prefactors .
this provides an _ a posteriori _
justification for the agreement between the predictions for static bunches and the numerical data in @xcite .
noting that the dimensionless parameter @xmath136 introduced in @xcite is given by @xmath137 in the present units , we see that ( [ first_terrace_size ] ) and ( [ minimal_terrace_size ] ) reduce to @xmath138 and @xmath139 , which is to be compared with the expressions @xmath140 and @xmath141 for static bunches . from ( [ w ] )
we find that the ratio @xmath142^{-1 } \approx 1.72 $ ] , which is considerably larger than the corresponding number 1.29 in the static case .
this reflects the fact that moving bunches are considerably broader than their static counterparts , because of the gradual increase of the terrace sizes in the outflow region , and explains the significant discrepancy between numerical and analytic estimates for @xmath117 in @xcite .
finally , we note that for a general step - step interaction exponent @xmath143 in ( [ step - step interaction ] ) we arrive at generalized scaling laws @xmath144 and @xmath145 which are , apart from strong finite size corrections , consistent with the ones derived in @xcite .
an important condition for the applicability of our continuum theory is the smallness of the asymmetry parameter , @xmath39 . for step bunching induced by a conventional ehrlich - schwoebel ( es ) effect during sublimation , @xmath146 , where @xmath147 ( @xmath148 ) is the kinetic coefficient for attachment to a step from the lower ( upper ) terrace @xcite ; keeping @xmath3 small
then simply requires a weak es effect . for current - induced step bunching in the attachment / detachment limited regime
, one finds @xcite @xmath149 where @xmath150 denotes the attachment rate of adatoms to steps , @xmath151 the atomic area , @xmath152 the equilibrium adatom concentration , @xmath153 the electromigration force , @xmath154 the monolayer evaporation time , and @xmath155 the step interaction strength .
the quantity @xmath156 is the mobility of an isolated step .
the model ( [ discrete_time_evolution ] ) of non - transparent steps is expected to apply in two of the four temperature regimes @xcite in which step bunching is observed on si(111 ) , around 900@xmath157 c and around 1250@xmath157 c @xcite .
the material parameters given in @xcite lead to the estimate @xmath158 in the low temperature regime and @xmath159 in the high temperature regime .
the latter is presumably an upper bound , since a recent estimate @xcite of the kinetic length @xcite @xmath160 ( where @xmath161 is the adatom diffusion coefficient ) at 1200@xmath157 c indicates that the step mobility @xmath162 increases less rapidly with increasing temperature than was assumed in @xcite .
the step interaction parameter @xmath5 depends very sensitively on the mean miscut @xmath163 .
taking the interaction strength to be @xmath164 ev / @xmath165 at 1250@xmath157 c , the parameters given in @xcite yield @xmath166 for @xmath167 nm .
in this paper we have shown that the continuum equation ( [ pde ] ) faithfully represents the properties of moving step bunches for sufficiently small values of @xmath3 , and that it can be used to extract accurate analytic expressions for the bunch shape and its various characteristic length scales .
our study reveals a central role of the point of minimal slope ( or maximal terrace size ) where the outflow region of one bunch joins the inflow region of the next .
an important ingredient is a fourier analysis of the discrete equations of step motion , which yields an inverse dependence of the bunch speed on bunch size under rather general conditions .
further work is needed to analytically derive the coefficient @xmath37 in ( [ vlat ] ) , and to understand how to obtain the bunch speed directly from the continuum equation , possibly by exploiting a recently proposed analogy to front propagation problems @xcite .
in addition , it seems desirable to study the regime of large @xmath3 , and to investigate the consequences of bunch motion for bunch coarsening beyond simple scaling arguments @xcite .
the conditions assumed in this paper should be realizable in electromigration experiments on si(111 ) at temperatures around 1250@xmath157 c. the gross features of the morphology , such as the maximal terrace size @xmath168 and the bunch width @xmath117 , could be followed in real time using rem , while for the more delicate measurements of @xmath169 and @xmath170 stm - studies would be preferable . | we study step bunching under conditions of attachment / detachment limited kinetics in the presence of a deposition or sublimation flux , which leads to bunch motion .
analysis of the discrete step dynamics reveals that the bunch velocity is inversely proportional to the bunch size for general step - step interactions .
the shape of steadily moving bunches is studied within a continuum theory , and analytic expressions for the bunch profile are derived . scaling laws obtained previously for non - moving bunches
are recovered asymptotically , but singularities of the static theory are removed and strong corrections to scaling are found .
the size of the largest terrace between two bunches is identified as a central scaling parameter .
our theory applies to a large class of bunching instabilities , including sublimation with attachment asymmetry and surface electromigration in the presence of sublimation or growth . |
confirmation , follow - up , and atmospheric characterization of a large number of super - earths and smaller planets will be among the main challenges facing exoplanet researchers in the next decade , especially once missions such as tess @xcite start identifying numerous candidates around bright , nearby stars .
current follow - up plans include the confirmation of those planet candidates using well tested ultra - high precision radial velocity instruments like harps and harps - n ( see e.g. * ? ? ?
* ) , and future instruments like espresso on the vlt @xcite , g - clef on the gmt @xcite , and codex on the e - elt @xcite and the atmospheric characterization of the most interesting confirmed planets using future ground - based ( e.g. gmt , tmt , and e - elt ) , and space - based ( e.g. jwst ) facilities .
however , not much attention is being paid to the potential role of smaller ground - based telescopes in the confirmation and follow - up of those planets .
while there is some work underway to build small arrays of robotic telescopes for radial velocity and photometric follow - up , e.g. project minerva , the full capacity of moderate - size existing telescopes and instruments for this task has not been explored yet . in this letter
, we report the results of our attempts to detect the shallow transit and the atmosphere of a super - earth around a bright nearby star with a 2-meter class telescope and current instrumentation on the ground . our target is 55 cnc e , the super - earth with @xmath4 = 8.3 @xmath5 0.4 @xmath6 ; @xmath7 = 1.94 @xmath5 0.08 @xmath8 , found to transit by @xcite and @xcite after @xcite provided a revised period of 0.74 days , shorter than the originally reported 2.8-day period derived by @xcite .
the transit depth of 55 cnc e is only @xmath0 0.4 mmag , and it orbits the brightest star known to harbor a transiting planet , 55 cnc ( v = 5.95 ) .
therefore , 55 cnc is , of all currently known planet hosts , the one with characteristics most similar to the targets that tess will find .
55 cnc is a 0.9 @xmath9 star @xcite and it harbors four other planets with masses between 0.14 @xmath10 and 3.8 @xmath10 , and orbital periods between @xmath0 14.6 days and @xmath0 4872 days @xcite .
we observed two transits of 55 cnc e with the alfosc instrument on the 2.5-meter nordic optical telescope ( not ) at the observatorio del roque de los muchachos in la palma , spain , as part of the not fast - track service program .
the observations started at 20:54 on march 28 2013 ut ( hereafter night 1 ) , and lasted four hours , during which we obtained 602 frames .
the exposure time was 10 seconds , and we used the 400kpix / sec readout mode to reduce overheads , providing an average cadence of 24 seconds .
we used the nearby m - star bo cnc as a reference to correct for instrumental and atmospheric effects .
both stars were observed simultaneously in differential spectrophotometry mode with the # 7 grism , which provides a wavelength coverage from 3800@xmath11 to 6800@xmath11 . to prevent slit losses
, we opted to perform these observations without a slit .
we used the 550_275 broadband filter to reduce sky background at the expense of reducing spectral coverage to 4100@xmath11 6800@xmath11 .
note that the transmission profile of the 550_275 filter is not flat , but shows variations of up to 20% over 100@xmath11 wavelength scales .
since the observations were slitless , the spectral resolution was determined by the seeing .
the median seeing was @xmath01 arcsec ( 6 pixels ) during the night , which resulted in a resolution of @xmath0700 . during the observations
the target passed close to the zenith , requiring a very rapid movement by the derotator .
this caused the position angle to vary by more than 0.2@xmath12 for @xmath09 minutes .
the offset in the rotator angle resulted in a blurring of the psf both spatially and spectrally , as well as an offset in the spatial and spectral directions .
we therefore exclude all frames where the offset in position angle was larger than 0.2@xmath12 ; in total 24 frames were removed due to this .
also , many of the frames reached peak counts above the linearity limit of the detector ( @xmath0300,000 adu ) .
a second transit of 55 cnc e was observed between 01:05 and 05:00 february 9 , 2014 ut ( hereafter night 2 ) , resulting in a total of 556 frames .
exposure time , readout mode , and cadence were the same as in night 1 .
in contrast to night 1 , the night 2 observations utilized a 40@xmath13 slit and no filter . the slit width was chosen to prevent slit losses due to seeing fluctuations and guiding errors , while significantly reducing the sky background .
as in night 1 , we used the # 7 grism , which provides spectral coverage between 3800@xmath11 and 6800@xmath11 .
all the observations were obtained after the target s meridian crossing , so we experienced no derotator problems .
the telescope was significantly defocused in order to keep the flux of the target and reference stars within the detector s linear regime .
this resulted in a spatial fwhm of @xmath014 - 15 pixels , corresponding to @xmath02.6 - 2.9 , given the 0.19/pixel platescale of alfosc .
this yields an average resolution of @xmath0250 .
the first 26 frames were not defocused and were discarded .
we reduced the data from both nights using a custom set of idl routines .
we first trimmed the overscan and bias - subtracted the frames . for night 1
we applied a non - linearity correction to account for the effect of count levels in the non - linearity regime .
the linearity correction was based on calibration data obtained by the not staff for the 400 kpix / sec read - out mode .
subsequently , we flatfielded the data and removed the sky background using a slightly modified version of the method proposed by @xcite . for each star
the sky outside the stellar psf was fit with a spline function using bspline from idl utils ; any residual gradient in the background was accounted for by fitting a low - order two dimensional polynomial surface . finally , we extracted the spectra of each star by summing up the flux in the spatial direction in a region of @xmath5110 pixels and @xmath580 pixels from the center of the trace for night 1 and 2 , respectively .
we verified that the aperture choice does not alter the results as long as it encompasses most of the psf .
example extracted spectra for each object , in each night , are shown in figure [ fig : white_lc_indiv ] .
as a final step before generating the light - curves , we re - sampled the spectra onto a common wavelength grid .
this is necessary because no atmospheric dispersion corrector was used for the observations and , to observe 55 cnc and the reference star simultaneously , the slit was not aligned with the paralactic angle .
furthermore , residual guiding errors and telescope drift in the spectral direction , also result in wavelength offsets .
light curves were generated by binning the extracted spectra in wavelength .
for the white - light curves we integrated over 1400 pixels in wavelength , which corresponds to @xmath0214 nm .
the narrower color - channel light curves were generated using 200 pixel bins ( @xmath030 nm ) .
the white - light curves for both nights are shown in figure [ fig : white_lc_indiv ] .
the light curves for the different color channels are shown in figure [ fig : color_lc ] . as revealed in the second row of fig .
[ fig : white_lc_indiv ] , the differential light curves from both nights show significant trends .
we fit those trends together with the parameters of the transit .
the parameters we used to model the trends were the spectral centroid position , airmass , the fwhm in the spatial direction ( seeing ) , and residual sky background .
for the transit , we fit the planet - to - star radius ratio and the mid - transit time , modelled as an offset in phase .
the parameter values for the best baseline models were obtained using a bayesian inference criterium ( bic ) analysis ( e.g. * ? ? ?
* ) , where the best solution is given by the lowest bic value . to model the transits we used the @xcite code , with limb darkening coefficients computed from the @xcite tables , for a star of properties similar to 55 cnc @xcite , and assuming an effective wavelength passbands similar to v - band .
observational noise dominates the errors for these observations , so the limb darkening coefficients can not be reliably constrained .
also , because of the relatively high noise levels , we needed to fix the planet s relative separation a / r@xmath14 and impact parameter b to the values from @xcite .
in addition to fitting the nights individually , we also performed joined fits for the two nights .
we fitted two different models to the combined light curves . in the first model we allowed the mid - transit time for each night to vary independently , while in the second model the phase offset of mid - transit , although allowed to vary , was the same for both nights .
the light curves for the color channels were fit jointly for the two nights , with the time of mid - transit fixed to that obtained from the fit to the white - light curves .
the fitting was done using a markov - chain monte - carlo procedure , for which we ran five chains of 200,000 steps each , discarding the first 20,000 steps to prevent initial parameter biases .
the five chains were then merged , after verifying that they were well mixed @xcite .
the parameters perturbed for all channels and runs are r@xmath15/r@xmath14 and the coefficients of the baseline .
for the white lightcurves we also perturbed t@xmath16 , and included a / r@xmath14 and b as priors .
the best fit parameters were determined using the 50% value of the distribution , and the uncertainties were obtained from the 16 - 84% confidence interval .
the best fit models are overplotted in fig .
[ fig : white_lc_indiv ] for the individual white curves , in fig .
[ fig : color_lc ] for the combined color channel curves , and in figure [ fig : white_lc_comb ] for the combined white curve .
the corresponding parameter values for each fit are listed in table 1 .
ccccc band & wavelength & r@xmath15/r@xmath14 & @xmath17 & @xmath18 + & ( nm ) & & & + + white & 457 - 671 & 0.0190@xmath1 & -0.0026@xmath19 & @xmath20 + white & 457 - 671 & 0.0200@xmath2 & @xmath20 & 0.0048@xmath21 + & & & + + white & 457 - 671 & 0.0198@xmath3 & -0.0020@xmath22 & 0.0044@xmath23 + c1 & 459 - 488 & 0.0180@xmath24 & -0.0020^a^ & 0.0044^a^ + c2 & 488 - 517 & 0.0211@xmath25 & -0.0020^a^ & 0.0044^a^ + c3 & 517 - 547 & 0.0198@xmath25 & -0.0020^a^ & 0.0044^a^ + c4 & 547 - 578 & 0.0212@xmath26 & -0.0020^a^ & 0.0044^a^ + c5 & 578 - 609 & 0.0201@xmath3 & -0.0020^a^ & 0.0044^a^ + c6 & 609 - 640 & 0.0221@xmath27 & -0.0020^a^ & 0.0044^a^ + c7 & 640 - 672 & 0.0211@xmath27 & -0.0020^a^ & 0.0044^a^ + & & & + + white & 457 - 671 & 0.0195@xmath3 & 0.0017@xmath28 & 0.0017@xmath28 + c1 & 459 - 488 & 0.0175@xmath29 & 0.0017^a^ & 0.0017^a^ + c2 & 488 - 517 & 0.0208@xmath25 & 0.0017^a^ & 0.0017^a^ + c3 & 517 - 547 & 0.0198@xmath25 & 0.0017^a^ & 0.0017^a^ + c4 & 547 - 578 & 0.0208@xmath3 & 0.0017^a^ & 0.0017^a^ + c5 & 578 - 609 & 0.0199@xmath3 & 0.0017^a^ & 0.0017^a^ + c6 & 609 - 640 & 0.0220@xmath27 & 0.0017^a^ & 0.0017^a^ + c7 & 640 - 672 & 0.0210@xmath27 & 0.0017^a^ & 0.0017^a^ + ^a^ phase offset fixed to value from white lightcurve .
[ tab : rprs ] we measure a white - light radius ratio for 55 cnc e of 0.0190@xmath1 on night 1 and 0.0200@xmath2 on night 2 . for the combined dataset
, we obtain a white - light radius ratio of 0.0198@xmath3 , and find that a model with a separate t@xmath16 offset for each night is slightly favored over a model where the t@xmath16 offsets between nights are fixed .
however , both t@xmath16 offsets are consisted with zero , revealing no noticeable transit timing variations between the two epochs .
when we force the t@xmath16 offset to be identical for both nights ( model 2 in table [ tab : rprs ] ) , we again find that the phase offset is small , and that the planet - to - star radius - ratio is unaltered from the free phase offset model that provided the lowest bic value .
our white - light radius ratios are fully consistent with both the _ most _ satellite and _ spitzer _ measurements from @xcite and @xcite .
when binning the residuals of the white light curves for each of the nights the rms decreases steeper than 1/sqrt(n ) , with n the number of points in the bin , indicating that any contribution from red - noise is small or hidden by the much larger scintillation effects ( see below ) . the color channel radius ratios summarised in table [ tab : rprs ] are also plotted in fig . [ fig : transmission_spec ] , and compared to the transmission spectra models for 55 cnc e generated by @xcite .
the models are for a @xmath30-dominated atmosphere ( @xmath31 = 0.99 ) and a non-@xmath30-dominated atmosphere ( @xmath31 = 0.5 ) , with two different carbon - to - oxygen ratios : a carbon - rich ( @xmath32 = 2.0 ) , and an oxygen - rich atmosphere ( @xmath32 = 0.5 ) .
our transit depths agree , on average , with the transit depth measured by most @xcite . however , the errorbars introduced by scintillation limit the capability of our data to place tight constraints on the atmospheric composition of 55 cnc e , and it will be necessary to improve the errorbars in each bin by at least a factor of two to be able to differentiate between the @xmath30-dominated and non-@xmath30-dominated models at 3-@xmath33 or larger confidence levels . as revealed in the white - light curve from night 2 ( right side , bottom panels in fig . 1 ) , where the observations span a wider range of airmass than in night 1 , the rms of the light - curve increases rapidly towards the end of the observations .
we attribute this increase in the noise to an increase in scintillation at higher airmass . using the scintillation equation in @xcite , and values of the observatory height above sea level ( 2383 m ) and the telescope diameter ( 256 cm ) for not , the exposure time in seconds , an average airmass @xmath34 , and an standard exponent for the airmass term of @xmath35 = 1.75 , we find the average scintillation noise level is about 1.0 mmag per exposure , while the average differential poisson noise is about 0.05 mmag for the white - light curves , and 0.10.2 mmag for the colour channels light curves .
therefore , finding a way to reduce the scintillation noise levels will have important implications for ground - based atmospheric studies of these planets .
we have presented the first ground - based detection of the shallow transit of the super - earth 55 cnc e. the detection was achieved with a 2m - class telescope , and a standard low resolution spectrograph in slitless and wide - slit , differential spectroscopy mode , to allow for the collection of a large number of photons without saturating .
the precision of our transit detection is comparable to that of the previously reported transit detections of this planet from space , with _ most _ and _ spitzer _ , which reveals the great potential to do this type of science from the ground , specially in the upcoming tess era , where the number of small planet candidates around stars in need of follow - up will largely exceed the capacity of space - based and large ground - based instruments .
the fact that such follow - up can be carried out with existing instruments on moderate - size telescopes also constitutes a great use of those facilities and a cost effective option .
this alternative is , however , not free of limitations , most of which can be overcome . in the case of bright stars ,
scintillation noise is the dominant limitation for small telescopes , but scintillation can be significantly reduced with some intelligent planning . with the current instrumentation ,
the main way to do this is to use longer exposure times and thereby reduce the overheads .
however , longer exposure times will likely result in saturation of the detector , which can be mitigated by either defocusing the telescope ( resulting in a lower spectral resolution , and possible slitlosses ) , using a higher resolution grating ( which will decrease the wavelength coverage ) , or using a neutral density filter to reduce the stellar flux .
the latter option will be offered soon at the william hershel telescope .
another possible solution is to increase default gain levels of standard ccds .
another constraint is the limited number of bright stars that would fall in the field of view of the instruments to be used as comparison objects to minimize the effect of the earth s atmosphere .
however , this problem can be overcome , at least in the case of color channel spectroscopy . in our observations
we find that the scintillation noise between two wavelength channels is highly correlated , and using one wavelength channel as a reference for others significantly reduces the point - to - point scatter , although for our data residual systematic effects dominate the noise level .
another possible solution is to develop a system of pickup mirrors that would redirect the light of comparison stars in the telescope s focal plane in to the detectors .
our result highlights the potential of using ground - based telescope to perform follow - up of bright transiting planet systems , such as those that will be found by the tess mission . for future observations of bright super - earth systems
we recommend that large telescopes are used in combination with both a neutral density filter and a ( small ) defocus in order to allow for more efficient observations and to reduce the impact from scintillation noise .
based on observations made with the nordic optical telescope , operated by the nordic optical telescope scientific association at the observatorio del roque de los muchachos , la palma , spain , of the instituto de astrofisica de canarias .
the data presented here were obtained with alfosc , which is provided by the instituto de astrofisica de andalucia ( iaa ) under a joint agreement with the university of copenhagen and notsa .
edm is supported in part by an ontario postdoctoral fellowship .
this work is supported by grants to rj from the natural sciences and engineering research council of canada .
mlm acknowledges support from a grant from the john templeton foundation .
the opinions expressed in this publication are those of the authors and do not necessarily reflect the views of the john templeton foundation .
we are grateful to the staff of the nordic optical telescope for their help with executing these observations .
we also thank renyu hu for sharing his 55 cnc e atmospheric models with us .
we thank the anonymous referee for constructive comments . , l. , cristiani , s. , garca lpez , r. , et al .
2010 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 7735 , society of photo - optical instrumentation engineers ( spie ) conference series , f. a. , cristiani , s. , rebolo lopez , r. , et al .
2010 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 7735 , society of photo - optical instrumentation engineers ( spie ) conference series , a. , frebel , a. , furesz , g. , et al .
2012 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 8446 , society of photo - optical instrumentation engineers ( spie ) conference series | we report the first ground - based detections of the shallow transit of the super - earth exoplanet 55 cnc e using a 2-meter - class telescope . using differential spectrophotometry , we observed one transit in 2013 and another in 2014 , with average spectral resolutions of @xmath0700 and @xmath0250 , spanning the johnson bvr photometric bands .
we find a white - light planet - to - star radius ratio of 0.0190@xmath1 from the 2013 observations and 0.0200@xmath2 from the 2014 observations .
the two datasets combined results in a radius ratio of 0.0198@xmath3 .
these values are all in agreement with previous space - based results .
scintillation noise in the data prevents us from placing strong constraints on the presence of an extended hydrogen - rich atmosphere .
nevertheless , our detections of 55 cnc e in transit demonstrate that moderate - size telescopes on the ground will be capable of routine follow - up observations of super - earth candidates discovered by the transiting exoplanet survey satellite ( tess ) around bright stars .
we expect it will be also possible to place constraints on the atmospheric characteristics of those planets by devising observational strategies to minimize scintillation noise . |
the possibility to isolate and perform direct transport measurements on few or even single graphite layers @xcite has triggered a large activity in the condensed matter community .
the case of a single layer of carbon atoms , known as graphene , is of particular interest because of its unique electronic structure which , under certain conditions corresponds to massless dirac fermions confined in two dimensions @xcite . on the other hand
, the coupling to a superconductor provides an interesting way to test the electronic properties of graphene . in a recent work by beenakker
@xcite it was shown that for an ideal interface between a superconductor and graphene an unusual type of andreev reflection , in which the hole is specularly reflected , appears .
several other effects involving graphene and superconductors like josephson transport @xcite , re - entrance effect @xcite , and quasiparticle transport mediated by multiple andreev processes @xcite have been theoretically analyzed .
in addition to its effect on the transport properties , the coupling to a superconductor also should produce a change in the electronic spectral properties and the induction of pairing correlations due to the proximity effect .
the recent experimental achievement of good contact between superconducting electrodes and graphene layers @xcite open the possibility to explore the proximity effect on these systems with great detail .
furthermore , experiments were the proximity effect on graphene could be explored even with atomic scale resolution using stm are underway @xcite .
at present only results for the total density of states in superconductor - graphene - superconductor structures have been presented @xcite .
the present work is aimed to study in detail the interface between the superconductor and the graphene sheet . to this end
we shall describe the electronic structure of graphene at the level of the tight - binding approximation .
this description allows us to analyze the superconductor - graphene interface more microscopically as compared to a description where the continuous limit leading to an effective dirac - bogoliubov - de gennes equation is taken from the start @xcite . in the continuous description
it is usually assumed that the presence of the interface do not couple different valleys of the graphene band structure , which could not be the case in an actual experimental situation .
moreover , when the study is focused on finite size graphene sheets , a strong dependence on the geometry of the edges appears .
thus , different symmetry directions will have distinct behavior @xcite .
for _ zigzag _ edges zero - energy surface states appear @xcite which could hide the effects of the coupling to a superconductor . in this work
we will concentrate on interfaces defined along an _ armchair _ edge .
we propose two different models for this interface : the first one assumes that graphene is coupled directly to a bulk superconducting electrode which does not maintain the honeycomb structure of the graphene sheet ; the second model studies the possibility that one superconducting electrode on top of the graphene sheet induces a finite pairing amplitude and shifts the fermi level of the graphene sheet far away from the dirac point . as we discuss below , the two models lead to different behavior of the andreev reflection probability as a function of energy , wave vector and doping level .
we further analyze several aspects of the spectral properties of the graphene layer within the two models both for the finite and the semi - infinite case .
the rest of the paper is organized as follows : in sec .
ii we introduce the tight - binding model for a graphene layer and we show the analytic expressions for the green functions for a semi - infinite and a finite layer . in sec .
iii the two different models for the interface with a superconductor are defined and a general expression for the self - energy , which provides the basis for the calculations of the following sections , is obtained . in sec .
iv we study the model dependence of the andreev reflection processes .
we also study , in sec .
v , the influence of the different interface models on the local density of states of a finite graphene layer coupled to a superconductor , analyzing in particular the minigap which is induced in the case of metallic layers .
results for the spatially resolved density of states for a semi - infinite graphene layer are presented in sec .
the paper is closed with some concluding remarks .
for the description of the electronic states in a defect free graphene layer we shall adopt the tight - binding approximation , i.e. we use a model hamiltonian of the type @xmath0 , where @xmath1 denote the hopping element between nearest neighbors carbon atoms on the hexagonal lattice and @xmath2 is a uniform site energy level which allows to vary the level of doping ( @xmath3 corresponds to the undoped case ) .
the dispersion relation for the translational invariant case is given by @xmath4 , where @xmath5 and @xmath6 denote the wavevector in the @xmath7 and @xmath8 direction respectively and @xmath9 is the lattice parameter defined as shown in fig .
[ figure1 ] ( as can be seen @xmath10 , where @xmath11 is the interatomic distance ) . for the undoped case the fermi surface collapse into two nonequivalent points at the brillouin zone corresponding to @xmath12 .
the dispersion relation close to these points can be linearized with a slope @xmath13 which fixes the fermi velocity , @xmath14 .
an essential ingredient for describing the interface between a graphene layer and other material is a good description of the electronic green functions at the edges of the layer .
we concentrate here in the derivation of the edge green function for a semi - infinite graphene layer with armchair orientation .
we assume that there is translational symmetry in the direction parallel to the edge @xmath15 .
the semi - infinite system can be decomposed into lines of sites in the @xmath8 direction which are coupled by hopping elements with the neighboring lines on the @xmath7 direction .
the unit cell on each line includes two sites corresponding to each hexagonal sublattice that are denoted by a and b ( see fig .
[ figure1 ] ) .
these sites are coupled by a hopping element @xmath1 within the unit cell .
thus , the cell hamiltonian is given by @xmath16 the hopping elements between neighboring lines couple also sites of type a with sites of type b but should include a phase factor @xmath17 due to the displacement of the cells in the @xmath8 direction .
the hopping matrix in the a - b space ( both in the forward and in the backward direction ) can be written as @xmath18 , where @xmath19 the self - similarity of the semi - infinite system with one additional line of sites leads to the following implicit equation for the edge green function @xmath20^{-1 } .\ ] ] hereafter we implicitly assume that @xmath21 stands for @xmath22 and that the limit @xmath23 is taken to obtain the retarded or the advanced component respectively
. we can now define @xmath24 which satisfies the simpler equation @xmath25^{-1 } , \ ] ] where @xmath26 . to obtain an explicit expression for @xmath27
it is useful to perform a basis rotation in order to diagonalize the matrix @xmath28 .
the general form of this rotation is @xmath29 where @xmath30 , @xmath31 being the @xmath32-pauli matrix acting on the sublattice space and @xmath33 where @xmath34 .
the eigenvalues of @xmath28 are @xmath35 .
we thus get @xmath36\\ -ie^{-iqa } \left[e^{i(\alpha+\phi_1 ) } - e^{-i(\alpha+\phi_2 ) } \right ] & e^{i\phi_1}-e^{-i\phi_2 } \end{array } \right ) , \end{aligned}\ ] ] where @xmath37 .
the eigenvalues @xmath38 and @xmath39 have been chosen so that the resulting green functions have the proper behavior when the frequency goes to infinity .
starting from the results of the previous section one can obtain the green functions of a finite graphene layer by introducing a perturbation consisting in breaking the bond between the @xmath40-th line and its neighbors on the @xmath41 line . from dyson s equation
we obtain the following set of coupled equations @xmath42 where the superindex @xmath43 stands for the finite system and the subindexes @xmath44 indicate the lines within the layer . on the other hand
the elements @xmath45 can be expressed as @xmath46 .
we now use that @xmath47 and @xmath48 , where @xmath49 corresponds to the surface green function for the semi - infinite system derived in the previous section .
also we have @xmath50^{-1}$ ] and @xmath51 \left [ \hat{i } - \left(\hat{g } \hat{t } \right)^{2}\right]^{-1 } \hat{g}$ ] , which allows to obtain @xmath52^{-1 } \left [ \hat{i } - \left(\hat{g}\hat{t}\right)^{2(n+1 ) } \right]^{-1 } \left [ \hat{i } - \left(\hat{g}\hat{t}\right)^{2n}\right ] \nonumber \\ & & \times \left [ \hat{i } - \left(\hat{g}\hat{t}\right)^{2\left(n - n+1\right)}\right ] \hat{g } \\ \hat{g}_{n , n}^f & = & \left [ \hat{i } - \left(\hat{g}\hat{t}\right)^{2(n+1 ) } \right]^{-1 } \left [ \hat{i } - \left(\hat{g}\hat{t}\right)^{2n}\right ] \left(\hat{g}\hat{t}\right)^{n - n } \hat{g } \nonumber.\end{aligned}\ ] ] making use of the rotation matrix defined in the previous section these quantities can be written in the following rather simple form @xmath53 one can have the expression for the borders of the layer setting @xmath54 or @xmath55 .
then , the eigenvalues of the green functions become @xmath56 and @xmath57 , with @xmath58 , for the @xmath59 and @xmath60 cases , respectively .
this expressions are equivalent to those for a finite tight - binding chain @xcite .
the poles of these green functions determine the spectral properties of the layer .
these poles are fixed by the condition @xmath61 , which is satisfied by @xmath62 , where @xmath63 is an integer .
one can associate this condition with the quantization of the transverse momentum which is used in the continuous model for describing armchair nanoribbons @xcite . at the charge neutrality condition
the existence of zero energy states requires @xmath64 , which can only be satisfied for @xmath65 ( in a more compact notation for @xmath66 )
. therefore the layers can be classified into metallic , for the @xmath66 case , and insulating for the other cases ( @xmath67 ) . in the insulating cases the gap in the spectrum is @xmath68 , where @xmath69 , @xmath70 being the length of the layer .
it should be noted that electron states in the metallic case are doubly degenerate , while the degeneracy is removed in the insulating cases @xcite .
one of the aims of the present work is to analyze different ways to describe the interface between a graphene layer and a superconductor . in the recent literature
it has been assumed that a superconducting electrode deposited on top of graphene induces a finite pairing amplitude and introduces a finite level of doping which shifts the fermi level of the graphene layer far away from the dirac point @xcite .
this heavily doped superconducting graphene ( hdsc ) model provides a simple boundary condition for the effective dirac - bogoliubov - de gennes equations describing the interface .
one can alternatively imagine that the graphene layer is coupled directly to a bulk superconducting electrode by means of a sharp interface which breaks the coherence between the two graphene sublattices .
we shall refer to this case as the _ bulk - bcs _ model .
we wish to analyze the differences between the two models .
the presence of superconducting correlations requires to introduce the nambu space describing electron and hole propagation within the graphene layer .
all green functions acquire a @xmath71 structure in nambu space . for the uncoupled graphene we have @xmath72 where @xmath73 corresponds to the propagators obtained in the previous section and @xmath74
is obtained from @xmath73 by changing @xmath75 and @xmath76 ( notice that we use the hat symbol to denote the sublattice space while the check symbol indicates the nambu space ) .
the effect of the coupling with the superconducting electrode can be introduced by means of a self - energy @xmath77 which renormalizes the uncoupled green functions .
thus , the local green function on the graphene edge at the interface is determined by @xmath78^{-1}$ ] . in the case of the bulk - bcs model
the self - energy is momentum independent and does not have a structure in the sublattice space ; i.e. @xmath79 where @xmath80 is the @xmath32 pauli matrix in nambu space and @xmath81 is the mean square hopping element between the graphene layer and the superconducting electrode .
this quantity controls the value of the parameter @xmath82 which characterizes the quality of the the interface , @xmath83 being the density of states . on the other hand , @xmath84 stands for the dimensionless bcs green function , i.e. @xmath85 and @xmath86 .
we would like now to derive the expression of the self - energy within the hdsc model .
we first notice that in the heavily doped limit one has latexmath:[$|\epsilon|
> > @xmath6 values .
thus in @xmath88 one has @xmath89 . in the new basis
the system is equivalent to a tight - binding chain with site energies @xmath90 and local pairing fixed by @xmath91 .
although the exact green functions for this system is rather complicated , for low energies it can be approximated by @xmath92 where @xmath93 . by further taking the approximation @xmath94 and transforming back into the site representation we obtain the following self - energy @xmath95 notice that , in contrast to the first model , the self - energy in the hdsc model does exhibit a structure in the sublattice space .
however , it satisfies the condition @xmath96 .
this structure turns out to be of importance in connection to andreev reflection as discussed in the next section . in the following
we will use a general form of the model self - energies , which can be expressed as @xmath97 .
thus , appropriate values for @xmath98 and @xmath99 will correspond to the different models ( i.e. @xmath100 with arbitrary @xmath98 for the bulk - bcs model , while @xmath101 and @xmath102 in units of @xmath1 , for the hdsc model ) .
the andreev reflection is the basic mechanism for the conversion of a quasiparticle current into a supercurrent at the interface between a normal metal and a superconductor . in the case of a graphene - superconductor interface like the one we have described in the previous section
there are two channels for the incident electrons with a given wave vector @xmath6 corresponding to the states which diagonalize the @xmath28 matrix
. the reflected hole can be in either of these two channels .
our microscopic theory can thus describe a more general situation than the idealized model for the interface used in @xcite which assumes only one channel for the reflected hole for a given wavector .
the andreev reflection amplitudes can be expressed in terms of green functions .
generalizing previous works @xcite we can derive the expression @xmath103^{-1}\right\}_{eh } \hat{a}^{1/2}_h , \label{coef - ra1}\ ] ] where @xmath104 . using the general form of the model self - energies discussed in the previous section allows us to reduce the expression of @xmath105 to @xmath106^{-1 } \hat{a}^{1/2}_h .
\label{coef - ra2}\end{aligned}\ ] ] this expression becomes particularly simple when @xmath3 because @xmath107 , @xmath108 and @xmath109 .
so @xmath105 is an scalar quantity given by @xmath110 - ( \beta^2+\gamma^2 ) \det{\hat{g}_{e , h } } } \hat{i } , \label{coef - ra3}\ ] ] where @xmath111 = \left[2 \beta g_s g - \gamma ( f+f^{\prime } ) \right]$ ] and @xmath112 . in general doping conditions ( i.e. when @xmath113 ) @xmath105 is not a scalar within the bulk - bcs model .
the eigenvalues of @xmath114 give the andreev reflection probability decomposed into two eigenchannels for each wavector @xmath6 .
the evolution of these eigenvalues for fixed @xmath21 and increasing @xmath2 as a function of @xmath6 is shown in the left panel of fig .
[ andreev - coef ] .
their maximum value is reached for @xmath115 and it never exceeds @xmath116 at normal incidence . within the hdsc model
, however , @xmath105 remains scalar for arbitrary doping and it always reaches the unitary limit at @xmath117 ( see right panel of fig .
[ andreev - coef ] ) .
it is also interesting to analyze the physical character of the andreev reflection in the two models .
the information on how the eigenchannels of the uncoupled structure are connected by an andreev process is contained in the matrix @xmath118 .
as @xmath105 is a scalar within the hdsc model , the channel mixing is determined by @xmath119 .
this is also the case for the bulk - bcs model at zero doping . in this case
we have @xmath120 and thus electrons injected in one channel emerge as holes in the opposite one .
the momentum in the @xmath8 direction is conserved in this process and therefore this type of reflection corresponds to what has been described in @xcite as _ specular andreev reflection_. on the other hand for @xmath113 and @xmath121 we have @xmath122 which corresponds to the usual ( retro ) reflection where holes are reflected on the same channel as the incident electron . for intermediate doping situations
both type of reflection would be present although with a dominance of specular ( retro ) reflection for @xmath123 ( @xmath124 ) . for fixed energy @xmath125 .
the left panel corresponds to the bulk - bcs model ( with @xmath126 ) and the right one to the hdsc model .
the different curves correspond to different values of the doping level @xmath2 : 0 ( full lines ) , 0.1 ( dashed lines ) , 0.3 ( dotted lines ) and @xmath127 ( dashed - dotted lines ) . ] to complete the analysis of the present section we have computed the conductance per unit length due to andreev processes , given by @xcite @xmath128 .\ ] ] normalized to the conductance per unit length of a ballistic graphene layer @xmath129 .
the results for the hdcs model ( full lines ) and for the bulk - bcs model ( dashed lines ) are compared with increasing doping level @xmath130 and @xmath131 . ] the results for @xmath132 within the two models are shown in fig .
[ conductance ] . as suggested in ref .
@xcite we normalize the result by the conductance per unit length of a ballistic graphene sheet , which in the low energy limit is given by @xmath129 . as can be observed , at the charge neutrality point the hcsd model yields a maximum of ratio @xmath133 at zero voltage , which drops to @xmath134 for @xmath135 , in agreement with the results of ref .
this ratio is of the order of @xmath136 in the bulk - bcs model with @xmath115 regardless of the doping level .
the qualitative behavior of the conductance with @xmath2 is similar in both models and agree with the results of ref .
using the previous results one can analyze the effect of the coupling with the superconductor on the spectral properties of a graphene layer of finite size .
again , we will focus on the differences between the bulk - bcs and the hdsc models for the interface . > from the dyson s equation it is straightforward to obtain the green functions at the edge of the layer ( labeled as @xmath137 ) when the coupling to the superconductor is introduced on the opposite edge ( labeled as @xmath40 ) . then , for an arbitrary line @xmath138 inside the layer we have @xmath139^{-1 } \check{\hat{g}}_{n , n}^f , \label{gnn - coupled - finite}\ ] ] where @xmath77 stands for the general form of the self - energy introduced in sec .
we can further reduce this expression to @xmath140 \hat{d}_e^{-1 } & - \beta f_s \hat{d}_h^{-1 } \\ - \beta f_s \hat{d}_e^{-1 } & \left [ \beta g_s + \gamma \hat{\sigma}_x + \left(\beta^2 + \gamma^2 \right ) \hat{g}^{fe}_{n , n } \right ] \hat{d}_h^{-1 } \end{array } \right ) \check{\hat{g}}_{n , n}^f .\ ] ] the quantities @xmath141 and @xmath142 have the following general form , expressed in the graphene subspace @xmath143 as in the previous section , in the limit where @xmath144 these two denominators become equal and simplify to an scalar , @xmath145 , because @xmath146 $ ] , and @xmath147 .
then @xmath148 - \left(\beta^2 + \gamma^2 \right ) \det{\hat{g}^f_{e , h}}\ ] ] using the expression of the finite layer green functions given in section ii and the rotation given in section i one can easily show that @xmath149 & = & - \frac { 1 } { t_g^2 } \frac { \sin{\left(n \phi_1 \right ) } \sin{\left(n \phi_2 \right)}}{\sin{\left[(n+1 ) \phi_1 \right ] } \sin{\left[(n+1 ) \phi_2 \right ] } } , \nonumber \end{aligned}\ ] ] @xmath150 & = & \frac{1}{t_g \sin{\alpha } } \left\ { \frac { \sin{\left(n \phi_1 \right ) } \sin{\left[(n+1 ) \phi_2 \right ] } \left[\beta g_s - \gamma \sin { \left ( \alpha + q \right ) } \right ] } { \sin{\left[(n+1 ) \phi_1 \right]}\sin{\left[(n+1 ) \phi_2 \right ] } } \right .
\nonumber \\ & - & \left .
\frac { \sin{\left(n \phi_2 \right ) } \sin{\left[(n+1 ) \phi_1 \right ] } \left[\beta g_s + \gamma \sin{\left(\alpha - q \right ) } \right ] } { \sin{\left[(n+1 ) \phi_1 \right]}\sin{\left[(n+1 ) \phi_2 \right ] } } \right\}. \nonumber\end{aligned}\ ] ] the zeroes of @xmath145 determine the poles of the coupled system green functions . from them
one can thus analyze the distortion of the spectrum due to the superconducting proximity effect . for the charge neutrality case ,
this zeroes can be obtained from the expression @xmath151}\sin{\left[(n+1 ) \phi_2 \right ] } + ( \beta^2 + \gamma^2 ) \sin{\left(n \phi_1 \right ) } \sin{\left(n \phi_2 \right ) } } { \sin{\left(n \phi_1 \right ) } \sin{\left[(n+1 ) \phi_2 \right ] } - \sin{\left(n \phi_2 \right ) } \sin{\left[(n+1 ) \phi_1 \right ] } } \nonumber \\ & + & \frac { \gamma } { \sin{\alpha } } \frac { \sin{(\alpha + q ) } \sin{\left(n \phi_1 \right ) } \sin{\left[(n+1 ) \phi_2 \right ] } + \sin{(\alpha - q ) } \sin{\left(n \phi_2 \right ) } \sin{\left[(n+1 ) \phi_1 \right ] } } { \sin{\left(n \phi_1 \right ) } \sin{\left[(n+1 ) \phi_2 \right ] } - \sin{\left(n \phi_2 \right ) } \sin{\left[(n+1 ) \phi_1 \right ] } } .
\label{zeros}\end{aligned}\ ] ] of a finite graphene layer coupled to a superconductor as a function of @xmath98 within the bulk - bcs model for @xmath152 ( left panel ) and with the length of the layer @xmath153 within both the bulk - bcs ( at @xmath126 ) and the hdsc models ( right panel ) . in the case of the bulk - bcs model for arbitrary @xmath98 three different behaviors
are found depending on the @xmath154 value . on the contrary ,
a universal behavior of @xmath155 is found within the hdsc model regardless of the @xmath40 value ( see text ) . ]
the spectrum corresponds to a series of subbands which disperse quadratically as a function of @xmath6 in the small @xmath6 limit . as in the uncoupled case
the precise form of the dispersion relation depends on the value of @xmath154 . in the case where the uncoupled layer is metallic ( @xmath156 ) the coupling to the superconductor induces a _ minigap _ in the lowest band .
the existence of this minigap in the spectrum is similar to what is found for diffusive conductors and can be associated to the pseudo - diffusive behavior of graphene at the charge neutrality point @xcite . for the cases @xmath157 the uncoupled layer is insulating and the coupling of the superconductor just leads to a renormalization of the gap in the spectrum .
we shall denote by @xmath155 the lowest energy level for all three cases .
the dependence @xmath155 as a function of @xmath40 and the interface parameters @xmath98 and @xmath99 can be obtained from eq .
( [ zeros ] ) with @xmath117 . for large
@xmath40 this level decreases as @xmath158 with a prefactor which depends on @xmath159 and the interface parameters .
[ minigap ] describes the behavior of @xmath155 both in the bulk - bcs and in the hdsc models .
the left panel shows the lowest energy state within the bulk - bcs model as a function of the interface transparency parameter ( @xmath98 ) for fixed @xmath40 .
the three cases @xmath160 are shown .
as can be observed , in the metallic case the minigap evolves from zero at @xmath161 to a maximum value at @xmath115 .
on the other hand , the two insulating cases exhibit different behavior .
while the starting value at @xmath161 is fixed by @xmath162 in both cases , in the case @xmath163 it decreases until it reaches the same value as the one of the @xmath164 case for @xmath115 . on the other hand
@xmath155 remains approximately constant for @xmath165 .
this behavior indicates that the proximity effect is almost negligible in this case .
the right panel shows the behavior of the lowest energy state as a function of @xmath40 both in the bulk - bcs model with @xmath115 and in the hdsc .
the results are universal ( i.e. independent of the ratio @xmath166 ) when plotted as a function of @xmath167 , where @xmath168 is the superconducting coherence length .
it is interesting to note that while in the bulk - bcs model two limiting @xmath169 curves , corresponding to @xmath170 and @xmath165 , appear , in the hdsc model @xmath155 lay on the same @xmath169 curve regardless of @xmath40 ( dashed line in fig .
[ minigap ] ) . ) in the hdsc model .
the plots on the top panels correspond to the lines inside the layer with @xmath171 and those on lower ones correspond to @xmath172 .
the uncoupled layer has a metallic behavior ( @xmath173 ) .
the ldos for this case at the edges of the layer ( top left panel ) and at the center of the layer ( lower left panel ) is plotted with dashed lines . when coupled to the superconductor , the ldos is modified by the appearance of a minigap ( denoted as @xmath174 in the pictures ) and with the breaking of the uncoupled bands into a pair of subbands ( @xmath175 and @xmath176 in the pictures ) .
the evolution of the ldos along the layer is shown in the right panels .
the results are normalized to the ldos of a bulk graphene layer with zero doping at @xmath177 , denoted by @xmath178 . ]
we define the electronic local density of states ( ldos ) on a line @xmath138 within the graphene layer as @xmath179,\ ] ] which has been normalized to one electron per site and spin .
the ldos thus defined is measured in units of @xmath180 . however , to study the proximity effect it is more convenient to normalize the ldos with the density of a bulk graphene layer with zero doping at @xmath177 , @xmath178 , which for @xmath181 is given by @xmath182 .
the results thus obtained do not depend on the choice of the ratio @xmath166 used in our tight - binding calculations .
the ldos on a metallic layer ( @xmath183 ) is shown in fig .
[ ldos - finita ] .
the results for the coupled case within the hdcs model are compared with the results for the uncoupled case .
it is typically found that the number of singularities in the ldos ( associated with the number of subbands ) in a given energy interval is doubled as compared to the uncoupled case .
this effect is due to the breaking of the double degeneracy of the bands due to the coupling with the superconductor .
the ldos also exhibits an oscillatory behavior with the position on the layer .
this behavior reflects the properties of the electronic wave functions and , as in the case of isolated nanoribbons @xcite , is distinct for lines with @xmath184 and lines with @xmath185 .
we thus illustrate these cases separately on the top and on the lower panels of fig .
[ ldos - finita ] .
the right panels of fig .
[ ldos - finita ] show the evolution along the layer of the ldos close to the singularities .
we illustrate this evolution at three different energies corresponding to the lowest first singularities , indicated by @xmath186 and 2 in fig .
[ ldos - finita ] . for reference
we also show the spatial variation of the ldos close to the first singularity for the uncoupled case . in this case
the ldos reaches a maximum value at the edges of the layer and a minimum in the middle for lines @xmath171 , while the opposite behavior is found for @xmath172 . in the coupled case
one can still identify the singularities with the oscillation pattern in the ldos but it does no longer reach an extreme value at the edge of the layer in contact with the superconductor . as final remark
we note that in the case of insulating nanoribbons the coupling to the superconductor just induces a shift singularities in the spectrum but do not change their number .
this is due to the nondegenerate character of the bands of the uncoupled layer .
the results of the previous section can be extended to analyze the spectral properties of a semi - infinite graphene layer coupled to a superconductor .
the local green function on a line @xmath138 is given by @xmath187^{-1},\ ] ] where @xmath188 is the green function for a finite graphene layer coupled to a semi - infinite superconducting layer obtained in the previous section and @xmath189 is the green function for the edge of the semi - infinite layer .
for a semi - infinite graphene layer coupled to a superconductor within the bulk - bcs model ( lower panels ) .
the upper panels show the corresponding results for the uncoupled case .
the plots on the left correspond to the undoped case while those on the right correspond to @xmath190 . ] fig . [ dos - semiinfinite ] illustrates the spatial variation of the ldos on a semi - infinite graphene layer and the effect of varying the doping level within the bulk - bcs model . for reference
we show the ldos for the uncoupled case on the upper panels . as can be observed , the uncoupled ldos exhibits long wavelength oscillations on the @xmath191 scale on top of the characteristic v shape behavior .
these oscillations are a surface effect which decreases in amplitude inside the layer as shown in fig .
[ ldos - oscillations ] , where the ldos profile at @xmath192 is plotted on a larger scale .
a similar effect has been shown to occur in the case of nanoribbons with zigzag edges @xcite .
the superconducting proximity effect is manifested by the appearance of sharp peaks in the ldos for energies @xmath193 ( lower panels on fig .
[ dos - semiinfinite ] ) .
these peaks distort the v shape density of states , an effect which decays within a few times the coherence length inside the layer .
the small oscillations on the @xmath194 scale are reduced as compared to the uncoupled case but are still observable within the bulk - bcs model ( indicated by the full line in fig .
[ ldos - oscillations ] ) .
the overall behavior of the ldos within the hdsc model is very similar , although in this last case the @xmath194 oscillations are further suppressed ( dashed line in fig .
[ ldos - oscillations ] ) .
the right panels in fig .
[ dos - semiinfinite ] illustrate the effect on the ldos of a displacement from the charge neutrality condition by applying a gate potential ( finite @xmath2 ) .
it is observed that the v shape is essentially rigidly displaced while the peaks induced by the proximity effect remain fixed at @xmath195 . on the other hand ,
the wavelength of the oscillation pattern is in this case set by @xmath196 . .
we have presented a theoretical analysis of the proximity effect at a graphene - superconductor interface . for this study
we have first derived analytical expressions for the green functions on an armchair edge of a semi - infinite graphene layer and for a finite layer . two models for describing microscopically the coupling to a superconductor have been presented . in the first model a bulk superconducting electrode
is connected directly to the armchair edge of a graphene sheet ( bulk - bcs model ) .
the honeycomb structure of graphene is broken and this is reflected in the different behavior of the andreev reflection probability on the two eigenchannels of the graphene sheet as a function of the parallel momentum @xmath6 . only for the case of zero doping
both eigenchannels are equivalent . within this model
one can study the effect of varying the normal transparency of the interface ( through parameter @xmath98 ) .
the andreev reflection probability at normal incidence never reaches unity within this model but has a maximum value of @xmath116 . in the second model
it is assumed that the superconducting electrode induces a finite order parameter on the graphene regions underneath .
this model thus maintains the graphene sublattice structure and , as we have shown , the andreev reflection amplitude @xmath197 is a scalar quantity for arbitrary doping and always reaches the unitary limit for normal incidence .
we have also studied the effect of the coupling to the superconductor on the spectral properties of finite and semi - infinite graphene layers . for finite layers
we have obtained a simple expression for the energy spectrum of the coupled system which can be easily evaluated numerically .
we have shown that a metallic ribbon develops a minigap whose size decreases inversely with the length of the layer .
this effect can be associated with the pseudo - diffusive behavior of graphene .
the induced minigap is slightly smaller in the bulk - bcs model with @xmath115 than in the hdsc model . for the semi - infinite case the proximity effect manifests in the appearance of peaks in the density of states for frequencies @xmath193 .
these peaks decay rapidly inside the graphene sheet for distances a few times the superconducting coherence length @xmath198 . on the other hand ,
the ldos keeps its characteristic v shape for zero doping for frequencies @xmath199 .
we expect that these findings can be useful to analyze future stm experiments on graphene sheets with superconducting electrodes .
this work has been financed by the spanish cycit ( contract fis2005 - 06255 ) .
discussions with j.c .
cuevas , l. brey and h. le sueur are acknowledged .
25 k. novoselov _
et al . _ ,
science * 306 * , 666 ( 2004 ) g. w. semenoff , phys .
lett . * 53 * , 1449 ( 1984 ) ; d. p. divincenzo , e. j. mele , phys .
b * 29 * , 1685 ( 1984 ) ; t. ando , t. nakanishi , r. saito , j. phys .
japan * 67 * , 2857 ( 1998 ) ; v. p. gusynin , s. g. sharapov , phys .
* 95 * , 146801 ( 2005 ) . c. w. j. beenakker , phys .
lett . * 97 * , 067007 ( 2006 ) m. titov , c. w. j. beenakker , phys .
b * 74 * , 041401 ( 2006 ) ali g. moghaddam , m. zareyan , phys . rev .
b * 74 * , 241403 ( 2006 ) a. ossipov , m. titov , c. w. j. beenakker , phys . rev .
b * 75 * , 241401 ( 2007 ) j. c. cuevas , a. levy yeyati , phys . rev .
b * 74 * 180501 ( 2006 ) h. b. heersche
_ et al . _ , nature ( london ) * 446 * , 56 ( 2007 ) ; h. b. heersche _
et al . _ , solid state comm . * 143 * , 72 ( 2007 ) ; a. shailos _
_ , europhysics letters * 79 * , 57008 ( 2007 ) ; f. miao _ et al . _ , science * 317 * , 1530 ( 2007 )
. h. le sueur ( private communication ) m. titov , a. ossipov and c.w.j .
beenakker , phys .
b * 75 * , 045417 ( 2007 ) .
h. schomerus , phys .
b * 76 * , 045433 ( 2007 ) ; y. m. blanter , i. martin , phys . rev .
b * 76 * , 155433 ( 2007 ) .
l. brey and h.a .
fertig , phys .
b * 73 * , 235411 ( 2006 ) .
a. r. akhmerov , c. w. j. beenakker , phys . rev .
b * 75 * , 045426 ( 2007 ) . e. prada , p. san - jose , b. wunsch , f. guinea , phys . rev .
b * 75 * , 113407 ( 2007 ) . e. vecino , a. martn - rodero , a. levy yeyati , phys . rev .
b * 64 * , 184502 ( 2001 ) . j. c. cuevas , a. martn - rodero , a. levy yeyati , phys .
b * 54 * , 7366 ( 1996 ) . j. c. cuevas , a. martn - rodero , a. levy yeyati , phys .
* 80 * , 1066 ( 1998 ) .
blonder , m. tinkham and t.m .
klapwijk , phys .
b * 25 * , 4515 ( 1982 )
. g. tkachov , phys .
b * 76 * , 235409 ( 2007 ) . | we present a theoretical analysis of the proximity effect at a graphene - superconductor interface .
we use a tight - binding model for the electronic states in this system which allows to describe the interface at the microscopic level .
two different interface models are proposed : one in which the superconductor induces a finite pairing in the graphene regions underneath , thus maintaining the honeycomb structure at the interface and one that assumes that the graphene layer is directly coupled to a bulk superconducting electrode .
we show that properties like the andreev reflection probability and its channel decomposition depend critically on the model used to describe the interface .
we also study the proximity effect on the local density of states on the graphene . for finite layers
we analyze the induced _
minigap _ and how it is reduced when the length of the layer increases .
results for the local density of states profiles for finite and semi - infinite layers are presented . |
the _ heat kernel approach _ ( hka for short ) , which will be presented and discussed in this paper , is a systematic method to construct state price densities which are analytically tractable . to be precise on analytic tractability , we mean with this notion that bond prices can be calculated explicitly , and that caps , swaptions or other derivatives on bonds can be calculated up to one integration with respect to the law of the underlying factor markov process
. therefore such models can be easily calibrated to market data and are useful for pricing and hedging purposes , but also for purposes of risk management .
the original motivation of introducing the hka was in modelling of interest rates with jumps .
in the hjm framework @xcite , the drift condition becomes quite complicated ( see h. shirakawa s pioneering work @xcite , see also @xcite and references therein ) if taking jumps into account , while in the spot rate approach , one will find it hard to obtain explicit expressions of the bond prices ( the affine class is almost the only exception ) . in @xcite , the state price density approach .
] is applied and by means of transition probability densities of some lvy processes explicit expressions of the zero coupon bond prices are obtained .
the hka is actually a full extension of the method , and thanks to the generalization its applications are now not limited to jump term structure models as we will show in the sequel . before the presentation of the theorem , we will give a brief survey of the state price density approaches such as the potential approach by l.c.g .
rogers @xcite or hughston s approach @xcite @xcite , etc , in section [ spda ] and section [ lreview ] .
our models are within the class of _ markov functional models _ proposed by p. hunt , j. kennedy and a. pelsser @xcite , and therefore compatible with their practical implementations .
one of the contributions of the hka to the literature could be to give a systematic way to produce markov functional models . as a whole , this paper is meant to be an introduction to the hka , together with some examples .
topics from practical viewpoint like fitting to the real market , tractable calibration , or econometric empirical studies , are not treated in this paper .
nonetheless we note that hka is an alternative approach to practical as well as theoretical problems in interest rate modeling .
we will start from a brief survey of state price densities . by a _ market _ , we mean a family of price - dividend pairs @xmath0 , @xmath1 , which are adapted processes defined on a filtered probability space @xmath2 .
a strictly positive process @xmath3 is a _ state price density _ with respect to the market if for any @xmath1 , it holds that @xmath4,\ ] ] or for any @xmath5 , @xmath6.\ ] ] in other words , @xmath7 gives a ( random ) discount factor of a cash flow @xmath8 at time @xmath9 multiplied by a radon nikodym derivative .
if we denote by @xmath10 the market value at time @xmath11 of zero - coupon bond with maturity @xmath9 , then the formula ( [ cff2 ] ) gives @xmath12.\ ] ] from a perspective of modeling term structure of interest rates , the formula ( [ spd ] ) says that , given a filtration , each strictly positive process @xmath13 generates an _ arbitrage - free _ interest rate model .
on the basis of this observation , we can construct arbitrage - free interest rate models . notice that we do not assume @xmath14 to be a submartingale ,
i.e. in economic terms _ we do not assume positive short rates_. the rational log - normal model by flesaker and hughston @xcite was a first successful interest rate model derived from the state price density approach .
they put @xmath15 where @xmath16 and @xmath17 are deterministic decreasing process , @xmath18 is a one dimensional standard brownian motion and @xmath19 is a constant .
the model is an analytically tractable , leads to positive - short rates , and is fitting any initial yield curve .
furthermore closed - form solutions for both caps and swaptions are available .
several extensions of this approach are known . in @xcite ,
rogers introduced a more systematic method to construct positive - rate models which he called the _
potential approach_. actually he introduced two approaches ; in the first one , @xmath20 where @xmath21 is the resolvent operator of a markov process on a state space @xmath22 , @xmath23 is a positive function on @xmath22 , and @xmath24 is a positive constant . in the second one
, @xmath23 is specified as a linear combination of the eigenfunctions of the generator of @xmath25 .
note that when @xmath25 is a brownian motion , @xmath26 for any @xmath27 is an eigenfunction of its generator , and the model gives another perspective to the rational lognormal models above . in @xcite l.
hughston and a. rafailidis proposed yet another framework for interest rate modelling based on the state price density approach , which they call a _
chaotic approach to interest rate modelling_. the wiener chaos expansion technique is then used to formulate a systematic analysis of the structure and classification of interest rate models .
actually , m.r .
grasselli and t.r .
hurd @xcite revisit the cox - ingersoll - ross model of interest rates in the view of the chaotic representation and they obtain a simple expression for the fundamental random variable @xmath28 as @xmath29)^2 | \mathcal{f}_t ] $ ] holds . in line with these , a stable property of the second - order chaos is shown in @xcite .
in this section we present the simple concepts of the heat kernel approach ( hka ) and several classes of possibly interesting examples .
the guiding philosophy of hka is the following : if one can easily calculate expectations of the form @xmath30 for some factor markov process @xmath31 , and if one knows _
explicitly _ one additional function @xmath32 depending on time and the state variables , then one can construct explicit formulas for bond prices , and analytically tractable formulas for caps , swaptions , etc ( i.e. , those formulas are evaluated by one numerical integration with respect to the law of the underlying markov process ) . in other words ,
any explicit solution of a well - understood kolmogorov equation with respect to some markov process leads to a new class of interest rate models .
we consider a general markov process @xmath33 on a polish state space @xmath22 and a probability space @xmath34 with filtration @xmath35 .
let @xmath36 defined on @xmath37 be a strictly positive function .
it is said to satisfy the * propagation property with respect to @xmath31 * if @xmath38 for all @xmath39 and @xmath40 .
we assume furthermore that @xmath41 for @xmath42 .
let @xmath43 satisfy the propagation property with respect to the markov process @xmath31 and @xmath44 .
we define then a state price density @xmath14 through @xmath45 we clearly have the following basic assertion : [ hkabond ] bond prices are given through @xmath46 for @xmath47 and @xmath40 .
the bond prices are calculated with respect to the state price density ( [ spdofhka ] ) through @xmath48 actually , by the markov property and by the propagation property we can calculate the right hand side , @xmath49 for all @xmath40 and @xmath47 .
notice that this remarkable simple formula is given with respect to the * historical*/*physical * measure .
a derivative with payoff @xmath50 on @xmath51 for some @xmath47 has at time @xmath52 the price @xmath53,\ ] ] which is analytically tractable if one knows how to calculate expectations of the form @xmath54 .
notice that the initial term structure is given in the previous setting through @xmath55 for @xmath56 .
let @xmath33 with a polish state space @xmath22 on a probability space @xmath34 with filtration @xmath35 . take a measurable , bounded @xmath57
, then @xmath58 satisfies the propagation property ( [ propro ] ) .
this example demonstrates that the previous theory covers the results of @xcite .
indeed , consider a @xmath59-dimensional lvy process @xmath60 starting at @xmath61 in its natural filtration .
we assume that it has a density @xmath62 with respect to lebesgue measure
. then @xmath63 defines a markov process on @xmath64 and we have that @xmath65 for @xmath66 and @xmath67 .
notice that this example can be associated to the previous example [ genericexample ] if one allows @xmath68 in the sense of distributions .
consider an affine markov process @xmath25 on @xmath69 and @xmath70 such that @xmath71 exist .
then we define @xmath72 for some numbers @xmath73 . in the spirit of the previous generic example we obtain @xmath74 for @xmath75 and each initial value @xmath76 .
notice that the `` laplace transforms '' @xmath77 are easily calculated for affine processes , which yields a concrete representation of the prices in the physical measure .
these models are completely new interest models constructed from affine models . for instance , the derived short rate models have interesting ( econometric ) features .
an example driven by the affine cox - ingersoll - ross process is visualized in figure 1 .
the pay - off ( at its maturity @xmath78 ) of a ( unit of ) swaption is @xmath79 where @xmath80 with @xmath81 is the tenor structure of the swap contract , and @xmath82 is the swap rate ( see e.g. @xcite*chapter 1 , section 6 ) . note that ( [ pay - off1 ] ) can be rewritten as @xmath83 therefore in the state price density approach , the quantity @xmath84 we get @xmath85 by hka , namely @xmath86 where @xmath32 has the propagation property w.r.t . a markov process @xmath25 .
we have @xmath87 , \end{split}\ ] ] which is again up to integration with respect to the law of @xmath25 explicit .
assume that @xmath36 is differentiable with respect to @xmath88 .
then the short rate @xmath89 of the previous interest rate model is given through @xmath90 since @xmath91 the sign of the short rate is determined through @xmath92 notice that we do not necessarily have a short rate process .
take a finite dimensional manifold @xmath93 and take @xmath25 a markov process with values in @xmath93 .
a non - vanishing function @xmath94 with @xmath95 for some @xmath96 is called eigenfunction of @xmath31 .
let @xmath97 , @xmath98 be eigenfuntions with eigenvalues @xmath99 , and let @xmath100 be decreasing functions .
set @xmath101 then we notice that @xmath102 is a supermartingale , too .
if @xmath103 is strictly positive and @xmath104 , then we can construct an arbitrage - free positive interest rate model by setting @xmath105 as a state price density .
we call such models the _ eigenfunction models_. the class includes the generic approach ii in the potential approach by rogers @xcite , and is included in the flesaker - hughston s general framework since @xmath106 becomes a martingale ( see section [ lreview ] ) .
the formula ( [ swaption : formula ] ) becomes extremely simple when we use the _
eigenfunction model_. @xmath107 where @xmath23 satisfies ( [ eigenf ] ) . by applying ( [ eigens ] ) on ( [ swaption : formula ] )
, we obtain the following result
. [ swpformula ] the pricing formula of the swaption is @xmath108 where @xmath109 and @xmath110 let @xmath31 be a brownian motion with drift : whose generator is @xmath111 where @xmath112 . for @xmath113 with @xmath114 , we have @xmath115 therefore , if @xmath116 is such that @xmath117 ( this is always possible , take @xmath118 for example ) , then the state price density @xmath119 where @xmath120 , gives a positive rate model . in this case @xmath121 where @xmath122 notice that the black - scholes formula for the price at time @xmath11 of the call option with the strike @xmath123 and the maturity @xmath9 is given by @xmath124
this example corresponds to the flesaker - hughston s rational log - normal model @xcite .
let @xmath31 be an ornstein - uhlenbeck process on @xmath125 , whose generator @xmath126 is given by @xmath127 where @xmath128 .
setting @xmath129 , we have @xmath130 therefore we have that @xmath131 is an eigenfunction with the eigenvalue @xmath132 . setting @xmath133 and @xmath134 , we obtain @xmath135 where @xmath136 , @xmath137 and @xmath138 for the proof see the appendix [ ap : ou ] .
as we have pointed out the propagation property of @xmath32 does not ensure that the process @xmath139 is a supermartingale .
thus , the interest rate model with @xmath140 does not exclude the possibility of negative rates . in this section
we provide models based on hka and leading to positive short rates .
let @xmath32 be a positive function with propagation property with respect to a markov process @xmath25 , and a weight @xmath141 such that @xmath142 for arbitrary @xmath143 and @xmath144 .
set @xmath145 then we have [ potential1 ] the process @xmath146 is a supermartingale .
for @xmath147 and @xmath40 , we have @xmath148 here we have used the propagation property of @xmath32 . since @xmath149 we have the desired result .
thus interest rate models with @xmath150 have positive short rates .
the class is an extension of the potential models defined in @xcite , since we can take @xmath32 as in and @xmath151 .
let @xmath152 be a non - negative , measurable function defined on @xmath22 . put @xmath153 and @xmath154 then the bond market spanned by @xmath155 is arbitrage - free with respect to the state price density @xmath156 indeed , this corresponds to an additional factor @xmath157 for @xmath158 added to the markov process @xmath31 . since @xmath159 is again a markov process on @xmath160 , we notice that @xmath161 defined on @xmath162 satisfies the propagation property
. therefore the previously described bond prices are arbitrage - free with respect to the stated state price densities for initial values @xmath163 .
since @xmath164 is decreasing in @xmath88 the short rates are again positive .
we give a third , more involved method to obtain positive rate models .
though it is only valid for symmetric lvy processes at this stage , again the class is completely new .
let @xmath165 be a @xmath166-dimensional lvy process , which is symmetric in the sense that @xmath167 .
we define @xmath168 .
note that in this case its lvy symbol is positive and symmetric , i.e. @xmath169 for some @xmath170 which is symmetric in the sense that @xmath171 .
let @xmath172 be a measure on @xmath64 such that @xmath173 for any borel set @xmath16 .
put @xmath174 assuming it is finite ( except at @xmath175 ) for each @xmath67 .
the function @xmath176 is non - negative and has the propagation property with respect to @xmath177 .
first , notice that by the symmetry of @xmath172 we have @xmath178 and @xmath179 for any @xmath67 .
then by bochner s theorem we know that @xmath180 is non - negative definite and we see that @xmath181 for any @xmath67 . since @xmath182 , @xmath183 \leq 1 $ ] for any @xmath184 and @xmath67 , we have @xmath185 if the fourier transform @xmath186 exists and is bounded in @xmath187 , then by the symmetry of @xmath172 it holds that @xmath188 , and @xmath176 has the following expression : @xmath189 on the other hand , let @xmath172 be the lebesgue measure of @xmath64 .
if the density @xmath190 of @xmath191 exists in @xmath192 , we have @xmath193 that is , @xmath176 is the heat kernel .
we are now able to prove the following theorem on supermartingales constructed from heat kernels : [ mainth ] let @xmath116 be a positive constant greater than @xmath194 and let @xmath195 be any positive constant
. set @xmath196 .
then @xmath197 is a supermartingale with respect to the natural filtration of @xmath25 . by the expression ( [ realu ] ) , we have @xmath198 since @xmath199 and @xmath200 is positive , @xmath201 thus we obtain @xmath202 and then @xmath203 on the other hand , by an elementary inequality we have @xmath204 and since @xmath200 is positive , we obtain @xmath205 combining ( [ 2ndineq ] ) and ( [ 3rdineq ] ) , we have @xmath206 the right - hand - side of ( [ 4thineq ] ) is non - negative since @xmath207 and @xmath208 is decreasing in @xmath11 . finally , since we have @xmath209 = u ( \lambda+2t - t , x^x_{t } ) + c u ( t+\lambda,0),\ ] ] we now see that @xmath210 is a supermartingale . by the above theorem , we can now construct a positive interest rate model by setting @xmath211 to be a state price density , or equivalently @xmath212 here we give some explicit examples . in all examples , @xmath116 is a constant greater than @xmath194 and @xmath195 some positive constant .
a simple example is obtained by letting @xmath213 where the bond prices are expressed as @xmath214 a similar model can be obtained by setting , for @xmath215 , @xmath216 a visualization of this model is shown in figure 2 .
let @xmath217 be a cauchy process in @xmath125 starting from @xmath187 .
the explicit form of the transition density of @xmath218 is known to be @xmath219 where @xmath220 and @xmath221 ( see e.g.@xcite ) .
note that cauchy processes are strictly stable ( or self - similar ) process with parameter @xmath222 ; namely , @xmath223 this property is seen from their lvy symbol : actually we have @xmath224 = e^ { - t ( \theta |\xi| - i \langle \xi , \gamma \rangle ) } .\ ] ] if we set @xmath225 , the density turns symmetric and therefore , setting @xmath226 , we obtain an explicit positive term structure model .
we give the explicit form of the bond prices in the case of @xmath227 : @xmath228 the model is a modification of the cauchy tsms given in @xcite .
let @xmath229 be lvy process and @xmath230 be a subordinator independent of @xmath231 .
recall that the process @xmath232 is usually called the subordination of @xmath25 by the subordinator @xmath9 . among the subordinations of the wiener process , two classes are often used in the financial context ; one is the _ variance gamma _ processes and the other is _ normal inverse gaussian _ processes , since they are analytically tractable ( see @xcite and references therein ) .
the subordinator of a variance gamma process ( resp .
normal inverse gaussian process ) is a gamma process ( resp .
inverse gaussian process ) .
if we let the drift of the wiener process be zero , then the densities of the subordinated processes satisfies the condition for theorem [ mainth ] .
the gamma subordinator @xmath233 is a lvy process with @xmath234 where @xmath235 and @xmath236 are positive constants .
then the heat kernel of the subordinated wiener process @xmath237 is @xmath238 from this expression we have @xmath239 the heat kernel ( [ vgdensity ] ) can be expressed in terms of the modified bessel functions @xmath240 , @xmath241 , which has an integral representation @xmath242 actually we have @xmath243 the heat kernel of the inverse gaussian subordinator @xmath244 is @xmath245 where @xmath235 and @xmath246 are positive constants .
the heat kernel of the subordinated wiener process @xmath247 is @xmath248 applying ( [ mbessel ] ) , we obtain @xmath249 in particular , we have @xmath250
[ a1 ] for @xmath251 , @xmath252 , and @xmath253 , it holds @xmath254 where @xmath255 , _ lvy processes and infinitely divisible distributions _ , vol .
68 of _ cambridge studies in advanced mathematics _ , cambridge university press , 1999 .
translated from the 1990 japanese original ; revised by the author . | we construct default - free interest rate models in the spirit of the well - known markov funcional models : our focus is analytic tractability of the models and generality of the approach .
we work in the setting of state price densities and construct models by means of the so called propagation property .
the propagation property can be found implicitly in all of the popular state price density approaches , in particular heat kernels share the propagation property ( wherefrom we deduced the name of the approach ) . as a related matter
, an interesting property of heat kernels is presented , too . * key wordes * : interest rate models , markov - functional , state price density , heat kernel . |
in quantum field theory , unlike in classical physics , the energy density may be unboundedly negative at a spacetime point .
such situations entail violations of all the known classical pointwise energy conditions , such as the weak energy condition @xcite .
this fact has been known for quite sometime @xcite .
specific examples include the casimir effect @xcite and squeezed states of light @xcite , both of which have observational support .
the theoretical prediction of black hole evaporation @xcite also involves negative energy densities and fluxes in a crucial way . on the other hand ,
if the laws of quantum field theory place no restrictions on negative energy , then it might be possible to produce gross macroscopic effects such as : violation of the second law of thermodynamics @xcite or of cosmic censorship @xcite , traversable wormholes @xcite , `` warp drive''@xcite , and possibly time machines @xcite . as a result , much effort has been recently directed toward determining what constraints , if any , the laws of quantum field theory place on negative energy density .
one approach involves so - called `` averaged energy conditions '' ( see , for example , @xcite-@xcite ) , i.e. , averaging the local energy conditions over timelike or null geodesics .
another method employs `` quantum inequalities '' ( qi s ) @xcite , which are constraints on the magnitude and duration of negative energy fluxes and densities .
the current paper is another in a series which is exploring the ramifications of this approach @xcite-@xcite .
( for a more comprehensive discussion of the history of these topics , see the introductions of refs.@xcite and the references therein . )
the qi s have the general form of an inverse relation between an integral involving the the energy density or flux over a finite time interval and a power of that interval .
more precise forms of the inequality were originally derived for negative energy fluxes @xcite , and later for negative energy density @xcite .
this form of qi s involves `` folding '' the stress energy tensor into a `` sampling function '' , i.e. , a peaked function of time whose time integral is unity .
for example , it was shown in ref.@xcite that for the free quantized massless scalar field in four - dimensional minkowski spacetime , @xmath0 for all choices of the sampling time , @xmath1 . here
@xmath2 is the renormalized expectation value of the energy density evaluated in an arbitrary quantum state @xmath3 , in the frame of an arbitrary inertial observer whose proper time coordinate is @xmath1 .
the physical implication of this qi is that such an observer can not see unboundedly large negative energy densities which persist for arbitrarily long periods of time .
the qi constraints can be considered to be midway between the local energy conditions , which are applied at a single spacetime point , and the averaged energy conditions which are global , in the sense that they involve averaging over complete or half - complete geodesics .
the qi s place bounds on the magnitude and duration of the negative energy density in a _ finite neighborhood _ of a spacetime point along an observer s worldline .
these inequalities were derived for minkowski spacetime , in the absence of boundaries .
however , we recently argued @xcite that if one is willing to restrict the choice of sampling time , then the bound should also hold in curved spacetime and/or one with boundaries .
for example , we proved that the inequality eq .
( [ eq : inten ] ) holds in the case of the casimir effect for sampling times much smaller than the distance between the plates .
it turns out that this observation has some interesting implications for traversable wormholes @xcite .
quantum inequalities in particular curved spacetimes , which reduce to eq .
( [ eq : inten ] ) in the short sampling time limit , are given in ref .
@xcite . in the original derivation of eq .
( [ eq : inten ] ) , we used a rather cumbersome expansion of the mode functions of the quantum field in terms of spherical waves .
the goal of the present paper is to present a much more transparent derivation of qi bounds , based on a plane wave mode expansion . in so doing ,
we prove new qi constraints on negative energy density for the quantized electromagnetic and massive scalar fields . in sec . [ sec : massive ] , we derive a qi bound for the massive scalar field in both four and two - dimensional minkowski spacetime .
our earlier result , eq . ( [ eq : inten ] ) , is recovered as a special case when the mass @xmath4 goes to zero . a similar bound
is obtained for the electromagnetic field in sec .
[ sec : em ] .
our results , and their implications for the existence of traversable wormholes , are discussed in sec .
[ sec : summary ] .
our metric sign convention is @xmath5 .
in this section we derive a qi - bound on the energy density of a quantized uncharged massive scalar field in four - dimensional flat spacetime .
the wave equation for the field is @xmath6 where @xmath7 .
we can expand the field operator in terms of creation and annihilation operators as @xmath8 here the mode functions are taken to be @xmath9 where @xmath10 @xmath4 is the rest mass , and @xmath11 is the normalization volume .
the stress tensor for the massive scalar field is @xmath12 the renormalized expectation value of the energy density , in an arbitrary quantum state @xmath3 , is @xmath13 \nonumber\\ & + & { { \rm re}\over { 2v } } { \sum_{\bf k',k } } { { m^2 } \over { \sqrt{\omega ' \omega } } } \ , \left[\langle a^\dagger_{\bf k'}{a_{\bf k } } \rangle \ , e^{i(\omega'-\omega)t } - \langle { a_{\bf k'}}{a_{\bf k } } \rangle \ , e^{-i(\omega'+\omega)t } \right ] \ , .
\label{eq : menden } \end{aligned}\ ] ] here the energy density is evaluated in the reference frame of an inertial observer , at an arbitrary spatial point which we choose to be @xmath14 .
the time coordinate @xmath15 is the proper time of this observer . because of the underlying lorentz invariance of the field theory , we may of course choose the frame of any inertial observer in minkowski spacetime .
following refs .
@xcite , we multiply @xmath2 by a peaked function of time whose time integral is unity and whose characteristic width is @xmath1 .
a convenient choice of such a function is @xmath16 $ ] .
define the integrated energy density to be @xmath17 substitution of eq .
( [ eq : menden ] ) into eq .
( [ eq : defrhohat ] ) and performance of the integration yields @xmath18 \nonumber\\ & + & { { \rm re}\over { 2v } } { \sum_{\bf k',k } } { { m^2 } \over { \sqrt{\omega ' \omega } } } \ , \left[\langle a^\dagger_{\bf k'}{a_{\bf k } } \rangle \ , e^{-|\omega'-\omega|t_0 } - \langle { a_{\bf k'}}{a_{\bf k } } \rangle \ , e^{-(\omega'+\omega)t_0 } \right ] \ , .
\label{eq : mrhohat } \end{aligned}\ ] ] now write @xmath19 , and apply the lemma in appendix b to each of the sums in eq .
( [ eq : mrhohat ] ) , e.g. , @xmath20 we then obtain @xmath21 \nonumber\\ & + & { { \rm re}\over { 2v } } { \sum_{\bf k',k } } { { m^2 } \over { \sqrt{\omega ' \omega } } } \ , e^{-(\omega'+\omega)t_0 } \ , \left[\langle a^\dagger_{\bf k'}{a_{\bf k } } \rangle \ , - \langle { a_{\bf k'}}{a_{\bf k } } \rangle \ , \right ] \ , .
\label{eq : tempsum } \end{aligned}\ ] ] for the first sum ( i.e. , the one involving @xmath22 ) , apply the first lemma in appendix a , where we take @xmath23 .
apply the same lemma to each term in the sum which involves @xmath24 , taking @xmath25 , with @xmath26 . for the last sum , involving @xmath27 , use the second lemma in appendix a choosing @xmath28 .
the result is @xmath29 where we have used @xmath30 .
now let @xmath31 , in which limit @xmath32 .
performing the angular integrations and changing variables we get @xmath33 let @xmath34 .
we may then rewrite the integral in eq .
( [ eq : mrhoint ] ) as @xmath35 if we now let @xmath36 then our qi - bound may be written as @xmath37 for all @xmath1 . a plot of @xmath38 versus @xmath39 is given in fig . 1 . [ figure 1 ] the graph of @xmath38 versus @xmath39 .
we see that as @xmath40 ( for fixed @xmath1 ) , @xmath41 and @xmath42 , hence in this case our bound reduces to @xmath43 for all @xmath1 , which is the qi for the massless scalar field - originally obtained in @xcite by a much more complicated method . as @xmath44 ( for fixed @xmath1 ) ,
@xmath45 , and we see from the graph that @xmath46 .
note that our bound becomes more stringent as @xmath4 increases , and hence it becomes increasingly difficult to produce large negative energy densities .
this result is not surprising , since now one has to overcome the positive rest mass energy of the field quanta . for @xmath47 ( with fixed @xmath4 )
, corresponding to sampling times much larger than the compton wavelength of the particle , we also have @xmath45 and @xmath46 .
note that , due to the factor of @xmath38 , the right - hand side of eq .
( [ eq : mqi/4d ] ) vanishes more rapidly than the right - hand side of eq .
( [ eq : qi ] ) in the @xmath48 limit .
this can be interpreted as showing that negative energy due to a massive scalar field must be more highly localized than that due to a massless field .
the derivation of a qi bound on the energy density of a massive scalar field in two - dimensional spacetime is very similar to the discussion given in the previous section for the four - dimensional case .
one follows the same steps as in that treatment , but replacing the normalization volume @xmath11 with a periodicity length @xmath49 , and @xmath50 and @xmath51 by @xmath52 and @xmath53 , respectively .
this procedure yields @xmath54 where we have used @xmath55 .
it is at this stage that the 2d analysis differs from the 4d case .
now let @xmath56 , in which limit @xmath57 . rewriting the integral in terms of @xmath58 , one obtains @xmath59 let @xmath60 this expression
can be rewritten @xcite in terms of modified bessel functions of the second kind , @xmath61 , as follows @xmath62 where the prime denotes differentiation with respect to the argument .
we now employ the relation @xcite @xmath63 \ , , \label{eq : recrel}\ ] ] where again @xmath64 .
the integral @xmath65 can then be written as @xmath66 \nonumber\\ & = & { 1 \over { 4 { t_0}^2 } } \ , f(y ) \ , .
\label{eq : i / f}\end{aligned}\ ] ] thus our qi - bound may be written as @xmath67 for all @xmath1 .
the function @xmath68 is plotted against @xmath39 in fig .
2 . the graph of @xmath68 versus @xmath39 . [ figure 2 ] as @xmath40 ( for fixed @xmath1 ) , @xmath69 , and our inequality becomes @xmath70 for all @xmath1 , which is the qi for the massless scalar field in 2d @xcite . for large @xmath39 ,
the behavior of the graph is qualitatively similar to the 4d case .
however , for @xmath39 in the approximate range @xmath71 , there appears to be a small peak in the value of @xmath68 . in the case of flat spacetime
, this seems to be an artifact of 2d , since no corresponding peak occurs in the 4d case @xcite . rather than indicating that one could actually supercede the @xmath72 bound in the 2d massive case for certain values of @xmath39 , this result
may perhaps imply that the bound given in eq .
( [ eq : mqi/2d ] ) is not optimal over this range . in either case
, the relative height of the peak above the value of @xmath73 ( corresponding to the massless case ) is too small to produce a dramatic change in the qi bound . as we argued in ref.@xcite
, the constant on the right - hand side of the inequality would typically have to be larger by many orders of magnitude to result in large macroscopic effects .
in this section , we derive a qi - bound for the energy density of the free quantized electromagnetic field in 4d minkowski spacetime .
the absence of source charges and the imposition of the coulomb gauge condition imply @xmath74 where @xmath75 is the vector potential , and @xmath76 is the scalar potential .
the wave equation is @xmath77 the vector potential can be expanded in terms of creation and annihilation operators as @xmath78 where the @xmath79 are unit linear polarization vectors and thus @xmath80 . here
the sum @xmath81 .
the mode functions @xmath82 are again given by @xmath83 i.e. , we assume periodic boundary conditions in a box with normalization volume @xmath11
. we will adopt the convention that one of the polarization vectors changes direction as @xmath84 , but the other does not .
that is , we have @xmath85 so that @xmath86 .
the commutation relations for the creation and annihilation operators are @xmath87 = \delta _ { { \bf k } { \bf k ' } } \ , \delta_{\lambda \lambda ' } \ , , \label{eq : cr1}\ ] ] @xmath88 = 0 \ , , \label{eq : cr2}\ ] ] @xmath89 = 0 \ , .
\label{eq : cr}\ ] ] with the gauge choice made in eq .
( [ eq : gauge ] ) , the electric field operator becomes @xmath90 similarly the magnetic field operator is given by @xmath91 here we define @xmath92 so that @xmath93 @xmath94 and therefore @xmath95 and @xmath96 the stress energy tensor for the electromagnetic field is @xmath97 if we compute the renormalized expectation value @xmath98 , in an arbitrary quantum state @xmath3 , evaluated at @xmath99 , and fold this into our sampling function we obtain @xmath100 \ , \nonumber\\ & \qquad & \times \left[\langle a^\dagger _ { { \bf k } \lambda } { a _ { { \bf k ' } \lambda ' } } \rangle \ , e^{-|\omega-\omega'|t_0 } + \langle { a _ { { \bf k } \lambda } } { a _ { { \bf k ' } \lambda ' } } \rangle \ , e^{-(\omega+\omega')t_0 } \right ] \ , .
\label{eq : emrhohat } \end{aligned}\ ] ] now write @xmath101 , and similarly for the dot product involving the @xmath102 vectors . expanding eq .
( [ eq : emrhohat ] ) in this fashion , one gets six terms .
to each of these terms we apply the lemma in appendix b , e.g. , @xmath103 recombining the terms , we then obtain @xmath104 \ , \nonumber\\ & \qquad & \times \left[\langle a^\dagger _ { { \bf k } \lambda } { a _ { { \bf k ' } \lambda ' } } \rangle \ , + \langle { a _ { { \bf k } \lambda } } { a _ { { \bf k ' } \lambda ' } } \rangle \right]\ , e^{-(\omega+\omega')t_0 } \ , .
\label{eq : emrhohat2}\end{aligned}\ ] ] an analogous separation of the dot products in eq .
( [ eq : emrhohat2 ] ) into components , repeated application of the first lemma in appendix a , and a recombination of terms results in @xmath105 where we have used @xmath106 and @xmath107 , with @xmath26 . evaluation of the sum over @xmath108 gives @xmath109 now let @xmath31 , so @xmath110 an evaluation of the integral gives us our desired result : @xmath111 for all @xmath1 .
comparison with the qi - bound for the massless scalar field , eq .
( [ eq : qi ] ) , shows that the right - hand side of the bound in the electromagnetic field case is smaller ( i.e. , more negative ) by a factor of @xmath112 .
this is just what one would expect due to the two polarization degrees of freedom .
in the current paper , we have derived `` quantum inequality '' ( qi ) bounds on negative energy density for the quantized ( uncharged ) massive scalar and electromagnetic fields in minkowski spacetime .
the bounds take the form of uncertainty principle - type inequalities on the magnitude and duration of the negative energy density seen by an arbitrary inertial observer .
( note however that the energy - time uncertainty principle is _ not _ used as input to derive the qi bounds . )
originally we had derived such a bound for the massless scalar field in both two and four - dimensional minkowski spacetime @xcite .
our earlier four - dimensional derivation was performed using a rather complicated expansion of the mode functions in spherical waves .
the present treatment is considerably simpler , as the analysis is done in a plane wave mode representation .
we recover our previous results for the massless scalar field as the @xmath72 case of the massive scalar field . for the massive scalar field
, we find that in general it is more difficult to obtain sustained negative energy densities than in the massless case .
this is not surprising , as now one must overcome the positive rest mass energy . in the case of the electromagnetic field , the right - hand side of the bound is slightly weaker , i.e. , by a factor of @xmath112 , than in the massless scalar field case .
this is also to be expected , given the two polarization degrees of freedom in the former case .
for all of our qi bounds on energy density , in the infinite sampling time limit we sample over the entire timelike geodesic of the observer and obtain the averaged weak energy condition ( awec ) : @xmath113 where @xmath114 is the unit tangent vector to the geodesic and @xmath115 is the observer s proper time .
hence , in minkowski spacetime , the awec can be derived from the qi bound .
note that in minkowski spacetime , our bounds hold for an arbitrary choice of the sampling time , @xmath1 .
recently we argued that such bounds should also hold in a curved spacetime and/or one with boundaries , if the sampling time is restricted to be much smaller than the smallest local radius of curvature and/or the distance to any boundaries in the spacetime @xcite .
an explicit example of the validity of this assumption has been given in ref .
there a qi is derived in the static closed and open robertson - walker universe models . for a choice of sampling time which is small compared to the local radius of curvature
, the qi reduces to our flat spacetime bound .
the application of our bound to morris - thorne wormhole spacetimes @xcite , in the small sampling time limit , implied that typically either a wormhole must be only slightly larger than the planck size , or that the negative energy density must be concentrated in an extraordinarily thin band near the throat . as an example , in one case for a wormhole with a throat radius of @xmath116 , the negative energy density must be concentrated in a band _ no thicker _ than about a millionth of a proton radius . even for a wormhole with a throat radius the size of a galaxy
, the band of negative energy must be no thicker than about @xmath117 proton radii ( for this particular type of wormhole ) @xcite .
that analysis assumed that the stress energy tensor which maintained the wormhole consisted of quantized massless scalar fields .
more specifically , the @xmath118 which generates the wormhole geometry was assumed to be the expectation value of the stress energy tensor operator of the massless scalar field in a suitable quantum state .
the results of the present paper indicate that the same restrictions on the wormhole length scales will also apply if one tries to make the stress energy of the wormhole out of electromagnetic or massive scalar fields .
one might hope that the constraints imposed by these bounds could be circumvented by the superposition of many fields , each of which individually satisfies a qi bound .
however , we showed @xcite that in practice to achieve significant macroscopic effects , one needs either on the order of @xmath119 fields or a few fields for which the numerical constant on the right - hand side of the qi bound is many orders of magnitude larger than in the cases examined to date .
neither of these possibilities seem very likely .
it therefore appears probable that nature will always prevent us from producing gross macroscopic effects with negative energy . 0.2 in * acknowledgements
* we would like to thank michael pfenning for useful discussions and for help with the graphics .
tar would like to thank the members of the tufts institute of cosmology for their warm hospitality while this work was being done .
this research was supported in part by nsf grant no .
phy-9507351 and by a ccsu / aaup faculty research grant .
in this appendix we will establish two lemmas on sums of the expectation values of products of creation and annihilation operators .
the discussion is a slight generalization of the argument presented in the appendix of ref.@xcite .
the idea for this method of proof was originally suggested to us by flanagan @xcite .
the first lemma was proven by a more complicated argument in appendix a of ref.@xcite . 0.2 in * lemma 1 * : 0.1 in define the operator @xmath120 by @xmath121 where @xmath122 is a generalized mode label , and the @xmath123 s are assumed to be real . then it is easily seen that @xmath120 is a hermitian operator .
we may form the expectation value @xmath124 this follows from the fact that the left - hand side is simply the norm of the state vector @xmath125 , where @xmath3 is the quantum state in which the expectation value is taken .
we may expand this expression as @xmath126\ , + \ , \sum_j \ , { ( h_j)}^2 \ , .
\label{eq : osum1}\ ] ] use the fact that @xmath127 to write @xmath128 \ ,
+ \ , \sum_j \ , { ( h_j)}^2 \ , .
\label{eq : osum2}\ ] ] it follows from eq .
( [ eq : odago ] ) that @xmath129 \ , \geq \ , - \sum_j \ , { ( h_j)}^2 \ , .
\label{eq : l1}\ ] ] 0.2 in * lemma 2 * : 0.1 in define the operator @xmath130 by @xmath131 where @xmath122 is a generalized mode label , and the @xmath123 s are again assumed to be real .
then the operator @xmath130 is also hermitian . utilizing the same chain of reasoning used to establish lemma 1
, one can show that @xmath132 \ , \geq \ , - \sum_j \ , { ( h_j)}^2 \ , .
\label{eq : l2}\ ] ]
in this appendix , we wish to prove a lemma which generalizes the inequality proven in appendix b of ref .
consider the sum @xmath133 here the summation is over a finite set of modes with frequencies @xmath134 , with @xmath135 , and @xmath136 is an arbitrary complex function of the mode label @xmath122 .
@xcite , it was assumed that the @xmath137 . that assumption is not needed here . )
we first note that @xmath138 is real as a consequence of the fact that @xmath139 .
next we define the sum @xmath140 which is also real .
we wish to prove that @xmath141 first note that @xmath142 this follows from the fact that any sum of the form @xmath143 is non - negative as a consequence of its being the norm of the state vector @xmath144 where @xmath3 is the quantum state in which the expectation values in eqs .
( [ eq : s ] ) and ( [ eq : s_tilde ] ) are to be taken .
let us assume that the eigenfrequencies are discrete , as will be the case with a finite quantization volume , and that the lowest frequency is greater than zero .
we can then order the modes in increasing frequency so that @xmath145 .
in general , there can be several modes with the same frequency .
let @xmath146 label the distinct frequencies . and let @xmath147 label the various modes all having frequency @xmath148 . then @xmath149 . define @xmath150 and @xmath151 so that @xmath152 note that @xmath153 where @xmath154 is the lesser of @xmath134 and @xmath155 .
we can then write @xmath156 if we let @xmath157 and define @xmath158 and @xmath159 then we have that @xmath160 our goal will be to prove that the right - hand side of this expression is non - negative . as a prelude ,
let us note that any sum of the form @xmath161 , where @xmath146 and @xmath162 range over the same set of values , is non - negative as a consequence of its being the norm of a state vector .
in particular , @xmath163 . the quantity @xmath164 is the sum of all of the elements of an @xmath165 matrix .
our plan is to prove the positivity of this sum by working our way through the matrix , beginning in the lower right - hand corner .
the element in this corner is @xmath166 next consider the @xmath167 matrix formed by the four elements in this corner .
the sum over these elements is @xmath168 here we have used the facts that @xmath163 , @xmath169 , and that @xmath170 .
now suppose that we have established that @xmath171 we wish to show that this implies that @xmath172 the additional terms which are added in going from eq .
( [ eq : suml1 ] ) to eq .
( [ eq : suml ] ) are those which lie in the same row as and to the right of @xmath173 and in the same column as and below @xmath173 , as illustrated below : @xmath174 \ ] ] note that @xmath175 increases to the right and @xmath176 increases downward .
thus @xmath177 for all of these elements .
the sum of these terms is @xmath178 now we have that @xmath179 we have established that eq .
( [ eq : suml ] ) holds for @xmath180 and for @xmath181 .
we have further established that if it holds for one value of @xmath49 , then it holds for the next smaller value of @xmath49 .
it now follows by induction that @xmath182 and hence that @xmath141 recall that the sums in both @xmath138 and @xmath183 are over a finite set of modes with maximum frequency @xmath184 .
however , we may now take the limit in which both the number of modes and @xmath184 become infinite .
so long as the resulting sums are convergent , which is the case for the systems with which we are concerned , then we have that @xmath185 which is our final result . | in a previous paper , a bound on the negative energy density seen by an arbitrary inertial observer was derived for the free massless , quantized scalar field in four - dimensional minkowski spacetime .
this constraint has the form of an uncertainty principle - type limitation on the magnitude and duration of the negative energy density .
that result was obtained after a somewhat complicated analysis .
the goal of the current paper is to present a much simpler method for obtaining such constraints .
similar `` quantum inequality '' bounds on negative energy density are derived for the electromagnetic field , and for the massive scalar field in both two and four - dimensional minkowski spacetime .
= cmbx10 scaled 4 -36pt 0.64 cm 0.64 cm gr - qc/9607003 + tutp-96 - 2 + july 1 , 1996 + revised oct . 2 , 1996 restrictions + + + + .7 in l.h .
ford and thomas a. roman + .2 in institute of cosmology + department of physics and astronomy + tufts university + medford , massachusetts 02155 + |
there is a serious question as to whether the specified gradients of rf cavities operating under vacuum would operate in the specified magnetic fields .
this is under study by nfmcc collaboration @xcite and alternative designs using high pressure hydrogen gas , or open cell rf with solenoids in the irises , are being considered .
the bunching and phase rotation will be optimized for the muon collider , instead of being copied from a neutrino factory . instead of the slow helices
, a planar wiggler lattice is being studied that would cool both muon signs simultaneously , thus greatly simplifying the system .
the use of more , but lower field ( e.g. , 35 t ) final cooling solenoids is also under study .
experiments are underway to demonstrate two of the new technologies : mercury target @xcite , ionization cooling @xcite .
further experimental studies are needed .
although much work remains to be done , the scenario outlined here appears to be a plausible solution to the problems of capturing , manipulating , and cooling muons to the specifications for muon colliders with useful luminosities and energies , even up to 8 tev in the center of mass .
99 g. i. budker , quoted and translated in @xcite . d. cline , in _ physics potential and development of @xmath4 colliders : second workshop ; sausalito , ca 1994 _ , edited by d. cline ( american institute of physics , woodbury , ny , 1996 ) , pp . 36 .
a. n. skrinsky , quoted and translated in @xcite ; v. v. parkhomchuk and a. n. skrinsky , in _ proceedings of the 12th international conference on high energy accelerators , 1983 _ , edited by f. t. cole and r. donaldson , 485 ; d. neuffer , fermilab report fn-319 , 1979 ; d. neuffer , part .
* 14 * , 75 ( 1983 ) ; d. v. neuffer and r. b. palmer , in _ fourth european particle accelerator conference _ , edited by v. suller and ch . petit - jean - genaz ( world scientific , singapore , 1994 ) , p. 52 . the @xmath4 collaboration , reports bnl-52503 , fermi lab - conf .
96/092 , lbnl-38946 , 1996 .
http://www.muonsinc.com h. g. kirk _
_ , _ a high - power target experiment at the cern ps _ pac07 , albuquerque , nm , ( mopas094 ) .
y. alexahin and e. gianfelice - wendt , fermilab report beams - doc-2724-v1 , 2007 .
url : http://beamdocs.fnal.gov/adpublic/docdb/documentdatabase d. neuffer , report nfmcc - doc-269-v1 , 2003 .
url : http://nfmcc-docdb.fnal.gov/cgi-bin/documentdatabase/ icool , http://pubweb.bnl.gov/people/fernow/icool/readme.html j. s. berg _ et al .
_ , phys . rev .
beams * 9 * , 011001 ( 2006 ) .
r. palmer _ et al .
_ , phys .
st accel .
beams * 8 * , 061003 ( 2005 ) .
amit klier , presentation at the low emittance muon collider workshop , fermilab ( 6 - 10 feb 2006 ) .
url : http://www.muonsinc.com/mcwfeb06/ g. b. lubkin , phys . today 47n12 , 21 ( 1994 ) .
http://www.superpower-inc.com/20070807.aspx s. a. kahn _ et al .
_ , _ a high field hts solenoid for muon cooling _
pac07 , albuquerque nm , ( mopan118 ) .
summers _ et al .
_ , _ muon acceleration to 750 gev in the tevatron tunnel for a 1.5 tev @xmath5 collider _ , pac07 , albuquerque nm , ( thpms082 ) ; also arxiv:0707.0302
. m. zisman , _ status of the international muon ionization cooling experiment _ ,
pac07 , albuquerque nm , ( thpmn119 ) ; p. drumm , _ mice : the international muon ionisation cooling experiment _ , pac07 , albuquerque nm , ( roaa004 ) . | a complete scheme for production , cooling , acceleration , and ring for a 1.5 tev center of mass muon collider is presented , together with parameters for two higher energy machines .
the schemes starts with the front end of a proposed neutrino factory that yields bunch trains of both muon signs .
six dimensional cooling in long - period helical lattices reduces the longitudinal emittance until it becomes possible to merge the trains into single bunches , one of each sign .
further cooling in all dimensions is applied to the single bunches in further helical lattices .
final transverse cooling to the required parameters is achieved in 50 t solenoids .
.5 in .parameters of three muon colliders .
note 1 : depth is relative to any nearby low land , e.g. fox river at fnal .
note 2 : survival is from the end of phase rotation to the collider ring .
[ cols="<,^,^,^ " , ] [ tuneshift ] note that @xmath0 is larger at earlier cooling stages to allow for losses .
the first order shifts can be corrected by increasing the focus strength , but tune spreads of half the shifts can not be corrected . before the merge ,
the shifts are small because the numbers of muons per bunch are small .
the only 6d cooling stage with significant tune shift is the last ( # 8) .
its tune accepted @xmath1 which is 5 times the calculated maximum full tune spread of @xmath2 7.3% , and is not expected to be a problem .
the tune shifts in the 50 t cooling will be significant only during the reaccelerations , where we have assumed @xmath3s corresponding to 3 t focusing fields .
the design of these lattices to accept such tune shifts appears possible , although we are clearly nearing the limit . |
in the standard model , cp violation is characterized by a single phase in the cabibbo - kobayashi - maskawa ( ckm ) matrix , rendering its elements complex .
the ckm matrix is also unitary , so that determining whether or not this is empirically so is a central test of the standard model s veracity @xcite .
ascertaining whether the angles of the unitarity triangle , @xmath2 , @xmath8 , and @xmath9 , empirically sum to @xmath10 and whether its angles are compatible with the measured lengths of its sides lie at the heart of these tests of the standard model .
we study the impact of isospin violation on the extraction of @xmath1 from an isospin analysis in decays @xcite .
isospin is broken not only by electroweak effects but also by the @xmath4 and @xmath5 quark mass difference .
the latter drives @xmath6 mixing @xcite , which , in turn , generates an amplitude in not included in the isospin analysis .
thus , although the effect of electroweak penguins is estimated to be small @xcite , when all the effects of isospin violation are included , the error in the extracted value of @xmath1 can be significant @xcite . to review
the isospin analysis in @xmath7 decays , due to gronau and london @xcite , let us consider the time - dependent asymmetry @xmath11 @xcite : a(t ) = ( 1 - |r_f_cp|^2 ) ( 1 + | r_f_cp|^2 ) ( m t ) - 2 ( i m r_f_cp ) ( 1 + | r_f_cp|^2 ) ( m t ) , where @xmath12 , @xmath13 , and @xmath14 @xcite .
denoting the amplitudes @xmath15 , @xmath16 , and @xmath17 by @xmath18 , @xmath19 , and @xmath20 , respectively , and introducing @xmath21 to denote an amplitude of final - state isospin @xmath22 , we have @xcite a^+- = a_2 - a_0 ; a^00 = 2a_2 + a_0 ; a^+0 =3 a_2 , [ tridef ] where analogous relations exist for @xmath23 , @xmath24 , and @xmath25 in terms of @xmath26 and @xmath27 .
if isospin were perfect , then the bose symmetry of the @xmath28 @xmath29 state would permit amplitudes merely of @xmath30 , so that the amplitudes @xmath31 would be purely @xmath32 . in this limit the penguin contributions are strictly of @xmath33 character , so that they can not contribute to the @xmath32 amplitude : no cp violation is possible in the @xmath34 final states . the penguin contribution in @xmath17 , or in @xmath35 , can then be isolated and removed by determining the relative magnitude and phase of the @xmath36 to @xmath32 amplitudes .
we have r_^+^-= e^-2i_m ( a_2 - a_0 ) ( a_2 - a_0 ) = e^2i ( 1 - z)(1 - z ) , [ rdef ] where @xmath37 and @xmath38 with @xmath39 and @xmath40 in the standard model @xcite .
given @xmath41 , @xmath42 , @xmath43 , and their charge conjugates , the measurement of @xmath44 determines @xmath1 , modulo discrete ambiguities in @xmath45 , which correspond geometrically to the orientation of the `` triangle '' of amplitudes associated with eq .
( [ tridef ] ) , namely a^+- + 2a^00 = a^+0 , [ triangle ] with respect to @xmath46 and that of its charge conjugate .
the triangles relative orientation can be resolved via a measurement of @xmath47 as well @xcite , and thus @xmath48 is determined uniquely .
we examine the manner in which isospin - violating effects impact the extraction of @xmath1 , for isospin is merely an approximate symmetry .
the charge difference between the @xmath4 and @xmath5 quarks engenders a @xmath49 electroweak penguin contribution , which is outside the scope of the delineated isospin analysis @xcite , although methods have been suggested to include them @xcite .
this is the only manner in which the @xmath4-@xmath5 charge difference enters our analysis , so that we term this source of isospin breaking an `` electroweak effect . ''
the @xmath4-@xmath5 quark mass difference can also engender a @xmath49 strong penguin contribution through isospin - breaking in the hadronic matrix elements .
moreover , strong - interaction isospin violation drives @xmath6 mixing @xcite , admitting an @xmath50 amplitude .
although electroweak penguin contributions are estimated to be small @xcite , other isospin - violating effects , such as mixing , can be important @xcite . to include the effects of mixing , we write the pion mass eigenstate @xmath51 in terms of the su(3)@xmath52-perfect states @xmath53 , @xmath54 , and @xmath55 , where , in the quark model , @xmath56 , @xmath57 , and latexmath:[$|\phi_0\rangle= explicit relations between the physical and su(3)@xmath52-perfect states can be realized by expanding qcd to leading order in @xmath59 , momenta , and quark masses to yield a low - energy , effective lagrangian in which the pseudoscalar meson octet and singlet states are treated on the same footing @xcite .
diagonalizing the quadratic terms in @xmath60 , @xmath61 , and @xmath62 of the resulting effective lagrangian determines the mass eigenstates @xmath63 , @xmath64 , and @xmath65 and yields , to leading order in isospin violation @xcite , |^0= |_3 + ( |_8- @xmath66 , and @xmath67 . moreover , @xmath68 and @xmath69 , with @xmath70 , @xmath71 , @xmath72 , and @xmath73 @xcite .
thus the magnitude of isospin breaking is controlled by the su(3)-breaking parameter @xmath74 .
the @xmath64-@xmath65 mixing angle @xmath75 is found to be @xmath76 so that @xmath77 @xcite .
the resulting @xmath78 is comparable to the one - loop - order chiral perturbation theory result of @xmath79 in @xmath80 @xcite .
empirical constraints also exist on the _ sign _ of mixing .
that is , the ratio of the reduced matrix elements in @xmath81 decays , namely , @xmath82 and @xmath83 , is given by @xcite ( ) ^expt = 1.029 0.010 . using the lagrangian of ref .
@xcite and the quark masses @xmath84 of ref .
@xcite yields ( ) = 1 + _ 8 1.018 0.010 , where @xmath85 is the @xmath86 mixing angle , @xmath87 .
note , for comparison , that the one - loop - order chiral perturbation theory result is 1.022 @xcite . in regard to the @xmath1 results to follow ,
it is worth noting that the isospin - violating parameters we have adopted appear conservative with respect to the existing experimental constraints . using @xmath88 of ali _ et al . _
@xcite , we find @xmath89 and @xmath90 ; we use these values in the subsequent calculations . 7.0 truein -.2
cm in the presence of mixing , the amplitudes become a^-0&= & ^- _ 3 | h^eff | b^- + _ 8 ^ - _ 8 | h^eff | b^- + _ 0 ^ - _ 0 | h^eff | b^- + a^00&= & _ 3 _ 3 | h^eff | |b^0 + _ 8_3 _ 8 | h^eff | |b^0 + _ 0_3 _ 0 | h^eff @xmath91 mixing angle , @xmath92 . note that either of the @xmath63 mesons in the @xmath93 amplitude can suffer mixing ; the factor of two associated with this appears as @xmath94 in eq .
( [ newrel ] ) .
the @xmath7 amplitudes satisfy ^+- + 2a^00 & - & a^-0 = 2_8 _ 3 _ 8 | h^eff | |b^0 + 2_0 _ 3 _ 0 | h^eff | |b^0 + & - & _ 8 ^- _ 8 | h^eff | b^- - _ 0 ^- _ 0 | h^eff | b^- , [ newrel ] and thus the triangle relation of eq .
( [ triangle ] ) becomes a quadrilateral .
we ignore the relatively unimportant mass differences @xmath95 and @xmath96 .
we proceed by computing the individual amplitudes using the @xmath97 effective hamiltonian resulting from the operator product expansion in qcd in next - to - leading logarithmic ( nll ) order @xcite , using the factorization approximation for the hadronic matrix elements . in this context , we can then apply the isospin analysis delineated above to infer @xmath1 and thus estimate its theoretical systematic error , incurred through the neglect of isospin violating effects .
the effective hamiltonian @xmath98 for @xmath99 decay can be parametrized as @xcite ^eff= g_f , where @xmath100 and @xmath101 are as per ref .
@xcite ; we also adopt their wilson coefficients @xmath102 and @xmath103 , computed in the naive dimensional regularization scheme at a renormalization scale of @xmath104 gev @xcite . in nll order ,
the wilson coefficients are scheme - dependent ; yet , after computing the hadronic matrix elements to one - loop - order , the matrix elements of the effective hamiltonian are still scheme - independent @xcite .
this can be explicitly realized through the replacement @xmath105 | b \rangle^{\rm tree}$ ] , where `` tree '' denotes a tree - level matrix element and the @xmath106 are from ref .
the @xmath106 are complex @xcite and depend on both the ckm matrix parameters and @xmath107 , where @xmath108 is the momentum transferred to the @xmath109 pair in @xmath99 decay . noting ref .
@xcite we use @xmath110 , @xmath111 , and @xmath112 @xcite unless otherwise stated .
one expects @xmath113
@xcite ; we use @xmath114 in what follows .
we define the decay constants @xmath115 and @xmath116 , and use the flavor content of the su(3)@xmath52-perfect states to relate @xmath117 to @xmath118 . finally , using the quark equations of motion with pcac and introducing @xmath119 for @xmath120 odd and @xmath121 for @xmath120 even , the @xmath122 matrix element in the factorization approximation with use of the fierz relations is .strong phases and inferred values of @xmath123 @xcite from amplitudes in the factorization approximation with @xmath124 and @xmath125 .
the strong phase @xmath126 is the opening angle between the @xmath127 and @xmath128 amplitudes in fig . 1 , whereas 2@xmath129 is the strong phase associated with the closest matching @xmath123 values , denoted @xmath130 , from i m @xmath131/im @xmath132 , respectively .
the bounds @xmath133 and @xmath134 on @xmath126 from eqs .
( 2.12 ) and ( 2.15 ) of ref .
@xcite are also shown .
all angles are in degrees .
we input a ) @xmath135 @xcite , b ) @xmath136 ( @xmath137 ) @xcite , and c ) @xmath138 ( @xmath139 ) . [
cols="^,^,^,^,^,^,^",options="header " , ] [ alpherr ] = 8 pt @xmath140 the discrete ambiguity in the strong phase is resolved wrongly in this case see table i. procedure can also yield the wrong strong phase ; in case c ) of table [ delcf ] with @xmath141 , the triangles of the chosen solutions `` point '' in the same direction , whereas they actually point oppositely . if @xmath42 and @xmath142 are small @xcite the complete isospin analysis may not be possible , so that we also examine the utility of the bounds recently proposed by grossman and quinn @xcite on the strong phase @xmath143 of eq .
( [ rdef ] ) .
the bounds @xmath144 and @xmath145 given by their eqs .
( 2.12 ) and ( 2.15 ) @xcite , respectively , follow from eq .
( [ triangle ] ) , and thus can be broken by isospin - violating effects . as shown in table [ delcf ] ,
the bounds typically are broken , and their efficacy does not improve as the value of @xmath1 to be reconstructed grows large . to conclude , we have considered the role of isospin violation in @xmath146 decays and have found the effects to be significant .
most particularly , the utility of the isospin analysis in determining @xmath1 strongly depends on the value to be reconstructed .
the error in @xmath1 from a i m @xmath131 measurement grows markedly larger as @xmath1 grows small this is the region of @xmath48 currently favored , albeit weakly , by phenomenology @xcite .
the effects found arise in part because the penguin contribution in @xmath147 , e.g. , is itself small ; we estimate @xmath148 . relative to this scale ,
the impact of mixing is significant .
this is displayed in another way in table [ alpherr ] .
the `` penguin pollution '' in @xmath149 is such that were no isospin analysis applied , the error in @xmath2 would be of the order of @xmath150 .
the isospin - violating effects in @xmath151 suggest that the error in @xmath2 is still of the order of @xmath152 after the gronau - london @xcite analysis is applied . yet ,
were the penguin contributions in @xmath0 larger , pressing the need for the corrections of the isospin analysis , the isospin - violating effects considered would still be germane , for not only would the @xmath49 penguin contributions likely be larger , but the @xmath153 and @xmath154 contributions could also be larger as well @xcite . to conclude , we have shown that the presence of @xmath63-@xmath155 mixing breaks the triangle relationship , eq .
( [ triangle ] ) , usually assumed @xcite and can mask the true value of @xmath123 .
i would like to thank kam - biu luk and german valencia for the opportunity to speak at this meeting .
this work was supported by the u.s .
department of energy under de - fg02 - 96er40989 .
99 the expression for @xmath158 assumes that the sign of the @xmath159 mixing parameter is positive , see y. grossman , b. kayser , and y. nir , phys .
b * 415 * , 90 ( 1997 ) .
we ignore the width difference @xmath160 .
@xmath63-@xmath155 mixing also generates significant isospin - violating corrections to @xmath161 , see a.j .
buras and j .- m .
gerard , _ phys .
b _ * 192 * , 156 ( 1987 ) ; j.f .
donoghue , e. golowich , b.r .
holstein , and j. trampetic , phys .
b * 179 * , 361 ( 1986 ) . | an isospin analysis of @xmath0 decays yields @xmath1 , where @xmath2 is the usual ckm angle @xmath3 $ ] without hadronic uncertainty if isospin is a perfect symmetry . yet
isospin symmetry is broken not only by electroweak effects but also by the @xmath4 and @xmath5 quark mass difference the latter drives @xmath6 mixing and converts the isospin - perfect triangle relation between the @xmath7 amplitudes to a quadrilateral . the error incurred in @xmath1 through the neglect of the resulting isospin - violating effects can be significant , particularly if @xmath1 is small . |
integrable wave equations play a key role in understanding numerous phenomena in science . in this connection ,
understanding the long - time asymptotics of solutions is crucial . roughly speaking
, the typical behavior is that any ( sufficiently fast ) decaying initial profile splits into a number of solitons plus a decaying dispersive part . this has been first observed numerically for the korteweg de vries equation @xcite .
corresponding asymptotic formulas were derived and justified with increasing level of rigor over the last thirty years . to date , the most powerful method for deriving such long - time asymptotics is the nonlinear steepest descent method from deift and zhou @xcite , which was inspired by earlier work of manakov @xcite and its @xcite .
more on this method and its history can be found in the survey @xcite ; an expository introduction to this method for the korteweg de vries equation can be found in @xcite .
although this method has found to be applicable to a wide range of integrable wave equations , there are still some exceptions .
the most notable one is the camassa holm equation ,
also known as the dispersive shallow water equation , @xmath0 where @xmath1 is the fluid velocity in the @xmath2 direction , @xmath3 is a constant related to the critical shallow water wave speed , and subscripts denote partial derivatives .
it was first introduced by camassa and holm in @xcite and camassa et al .
@xcite as a model for shallow water waves , but it actually already appeared earlier in a list by fuchssteiner and fokas @xcite .
regarding the hydrodynamical relevance of this equation , let us also mention the more recent articles by johnson @xcite , ionescu - kruse @xcite as well as constantin and lannes @xcite . while in the case @xmath4 there is an underlying rieman hilbert problem which can be analyzed using the nonlinear steepest descent method @xcite , @xcite , @xcite , this breaks down in the limiting case @xmath5 . in this case
, the solitons are no longer smooth but have a single peak and hence are also known as peakons .
nevertheless , it was conjectured by mckean @xcite ( cf .
also @xcite , @xcite ) that solutions split into a train of peakons , in accordance with earlier numerical observations by camassa et al .
@xcite . however , apart from the multi - peakon case @xcite ( and some low - regularity solutions @xcite as well as for a simplified flow @xcite ) , this has been an open problem , resolved only recently by us in @xcite .
the technical problem here stems from the fact that the underlying isospectral problem has purely discrete spectrum and hence it is no longer possible to set up the usual scattering theory .
our approach in @xcite circumvented this shortcoming by a thorough investigation of the associated isospectral problem , which then allowed us to deduce long - time asymptotics .
however , this approach has still some drawbacks .
for example , it is not possible to obtain long - time asymptotics which hold uniformly in sectors .
the aim of the present article is to propose a novel approach to such kind of problems , which seems to be more natural . in some sense
, it can be thought of as a adaptation of the usual riemann
hilbert problem approach .
more precisely , we will replace the riemann
hilbert problem with a certain coupling problem for entire functions .
consequently , we will investigate the asymptotic behavior of solutions to this problem under known behavior of the given data . as a prototypical example
, we will apply our results to derive long - time asymptotics for the dispersionless camassa
holm equation .
however , we expect that this new technique will also work for other equations , whose underlying isospectral problem exhibits purely discrete spectrum . for example
, it can immediately be applied to derive long - time asymptotics for corresponding equations in the whole camassa holm hierarchy .
while for the positive members of this hierarchy one gets qualitatively the same asymptotic picture , the situation is somewhat different for the negative ones ( including for instance the extended dym equation ) . although solutions of negative members of the camassa
holm hierarchy still split into a train of peakons , their speed will be proportional to the modulus of the corresponding eigenvalue .
this causes the larger peakons to be the slower ones and the smaller peakons to be the faster ones , creating a qualitatively different picture .
the purpose of this section is to introduce the notion of a _ coupling problem _ for entire functions . to this end , consider a fixed discrete set @xmath6 such that the sum @xmath7 is finite .
it is well known that under this condition , the infinite product @xmath8 converges locally uniformly to an entire function of exponential type zero ( * ? ? ?
* lemma 2.10.13 ) , ( * ? ? ?
* theorem 5.3 ) .
furthermore , we introduce the quantities @xmath9 for each @xmath10 which are referred to as the _ coupling constants_. a solution of the coupling problem with data @xmath11 is a pair of entire functions @xmath12 of exponential type zero such that the following three conditions are satisfied : 1 .
coupling condition : @xmath13 2 .
growth and positivity condition : @xmath14 3 .
normalization condition : @xmath15 in order to be precise , if @xmath16 for some @xmath10 , then the coupling condition ( c ) in this definition has to be read as @xmath17 .
the growth and positivity condition ( g ) means that the meromorphic function @xmath18 is a so - called herglotz
nevanlinna function , which satisfy certain growth restrictions ( to be seen from their integral representations ; ( * ? ? ?
* chapter 6 ) , ( * ? ? ?
* chapter 5 ) ) .
moreover , let us mention that since the residues of such a function are known to be nonpositive , condition ( g ) also requires the necessary presumption @xmath19 on the sign of all coupling constants which are finite .
thus the coupling constants corresponding to the smallest ( in modulus ) positive and negative element of @xmath20 have to be nonnegative .
the consecutive coupling constants then have to be alternating nonpositive and nonnegative .
furthermore , the condition ( g ) also tells us that the zeros of the numerator and the denominator of the function in are interlacing ( after cancelation ) ( * ? ? ?
* theorem 27.2.1 ) .
in particular , this guarantees that the sums @xmath21 are finite , where @xmath22 denote the sets of all ( necessarily simple ) zeros of the functions @xmath23 . as a consequence
, these functions can be written as the canonical products @xmath24 bearing in mind the normalization condition ( n ) . finally , we mention the bounds @xmath25 upon roughly estimating and employing the interlacing condition once more .
the simplest case of a coupling problem prevails when the set @xmath20 consists of only one point . in this case
, it is indeed possible to write down the solution explicitly in terms of the one single coupling constant .
[ proponesol ] suppose that @xmath26 for some nonzero @xmath27 .
if the coupling constant @xmath28 is not negative , then the coupling problem has a unique solution given by @xmath29 it is readily verified that the given polynomials are indeed a solution of the coupling problem .
conversely , if @xmath30 is another solution , then @xmath31 for some @xmath32 with @xmath33 .
moreover , we infer that @xmath34 in view of the herglotz
nevanlinna property ( more precisely , from the interlacing condition of the poles and zeros ) .
lastly , the coupling condition ( c ) takes the form @xmath35 now if @xmath36 , then necessarily @xmath37
since otherwise we get the contradiction @xmath38 consequently , we may express @xmath39 in terms of the coupling constant using the coupling condition . in much the same manner ,
one may obtain the coefficients @xmath40 if @xmath41 and finally end up with @xmath42 in either case , which finishes the proof .
note that there is no solution of the coupling problem in proposition [ proponesol ] if the coupling constant is negative , since it would violate the positivity condition ( g ) .
we shall now derive a general result on the asymptotic behavior of solutions to the coupling problem . therefore , let @xmath43 be a first - countable topological space ( that is
, every point has a countable neighborhood basis ) and fix some @xmath44 .
again , we denote with @xmath6 a discrete set such that the sum is finite and define the entire function @xmath45 by .
moreover , for every @xmath46 we consider a set of coupling constants @xmath47 indexed by @xmath10 .
[ thmaa ] suppose there is a partition @xmath48 of @xmath20 such that @xmath49 as @xmath50 for each @xmath51 and define the conjugated coupling constants by @xmath52 if the pairs @xmath53 are solutions of the coupling problem with data @xmath54 for every @xmath55 , then @xmath56 as @xmath50 , locally uniformly for all @xmath57 . first consider a sequence @xmath58 , @xmath59 with @xmath60 as @xmath61 such that the entire functions @xmath62 converge locally uniformly as @xmath61 .
the respective limits are entire functions of exponential type zero in view of and will be denoted by @xmath63 . due to assumption and the coupling condition , we conclude that @xmath64 for @xmath51 ( also observe that the quantities @xmath65 are uniformly bounded in @xmath46 ) . as a consequence , the meromorphic herglotz
nevanlinna function @xmath66 has only one pole and thus at most two zeros , which are necessarily simple .
consequently , we may write ( keep in mind that these functions are of exponential type zero and that their zeros have genus zero ) @xmath67 where @xmath68 are polynomials such that @xmath69 has at most one zero , which is simple .
moreover , the pair @xmath70 satisfies the coupling condition @xmath71 where the constant @xmath72 is given as the limit @xmath73 hereby note that the limit is nonnegative because of . in view of proposition [ proponesol ]
we now may write down the polynomials @xmath68 explicitly and conclude that @xmath74 as @xmath61 , locally uniformly in @xmath57 . finally , from the very definition of the constants @xmath75 we may also rewrite this as @xmath76 as @xmath61 , locally uniformly in @xmath57 .
finally , if the claim of the theorem was not true , then there would be a compact set @xmath77 and a subsequence @xmath58 , @xmath59 with @xmath60 as @xmath61 such that @xmath78 for all @xmath79 , @xmath59 and some @xmath80 .
however , a compactness argument ( recall and apply montel s theorem ) shows that there is a subsequence @xmath81 such that @xmath82 converges locally uniformly as @xmath83 . in view of the first part of the proof
, this gives a contradiction to .
the assumptions in theorem [ thmaa ] allow one of the coupling constants to be arbitrary .
this will turn out to be crucial to obtain long - time asymptotics of the camassa
holm equation which are valid uniformly in sectors .
however , in the case when all of the coupling constants are supposed to converge to zero or infinity , one obtains the following result .
[ coraa ] suppose that we have @xmath84 as @xmath50 for every @xmath10 .
if the pairs @xmath53 are solutions of the coupling problem with data @xmath54 for every @xmath55 , then @xmath85 as @xmath50 , locally uniformly for all @xmath57 .
similarly to the first part of the proof of theorem [ thmaa ] , one infers that @xmath86 as @xmath61 , locally uniformly for all @xmath57 as long as the functions @xmath62 are assumed to converge locally uniformly .
in fact , this follows since the function in is now known to have no poles at all .
now the claim follows in much the same manner as in the second part of the proof of theorem [ thmaa ] , invoking a compactness argument .
as anticipated in the introduction , we will now demonstrate that our results provide a powerful tool to derive long - time asymptotics for the dispersionless camassa hom equation . to this end , let @xmath87 be a solution of @xmath88 with decaying spatial asymptotics . to be precise
, we will assume that the quantities @xmath89 are finite signed measure for each time @xmath90 .
these conditions guarantee ( see ( * ? ? ? * theorem 3.1 ) ) that for every time @xmath90 and @xmath91 , there are unique solutions @xmath92 of the differential equation @xmath93 ( the prime denotes spatial differentiation ) with the spatial asymptotics
@xmath94 in view of ( * ? ? ?
* theorem 4.1 ) , it is known that these solutions are real entire and of exponential type zero with respect to the spectral parameter . now the importance of the spectral problems lies in the well - known fact that their spectra are invariant under the camassa holm flow , that is , they are the same for all times @xmath95 ( for example , we refer to ( * ? ? ?
* section 2 ) , @xcite , ( * ? ? ?
* section 3 ) , ( * ? ? ? * theorem 5.1 ) ) .
for this reason , we may simply denote the spectrum of with @xmath20 , which is known to be real and purely discrete such that the sum @xmath96 is finite , in view of ( * ? ? ? * proposition 3.3 ) .
the wronskian of our two solutions @xmath97 turns out to be independent of space @xmath98 and time @xmath95 .
indeed , this function is the characteristic function of the spectral problem , that is , @xmath99 in view of ( * ? ? ?
* corollary 4.2 ) . in order to point out the connection to the coupling problem for entire functions
, one observes that the solutions @xmath100 and @xmath101 are linearly dependent for every eigenvalue @xmath10 and time @xmath90 .
hence there is some nonzero real @xmath102 such that we may write @xmath103 the time evolution for these quantities is known to be given explicitly by @xmath104 more precisely , this follows from the well - known time evolution of the associated norming constants ( for example , see ( * ? ? ?
* section 2 ) , ( * ? ? ?
* section 3 ) , ( * ? ? ? * theorem 5.1 ) ) and the identity in ( * ? ? ?
* lemma 3.2 ) , @xmath105 where the dot denotes differentiation with respect to the spectral parameter .
we have now collected all necessary ingredients to prove the announced long - time asymptotics for the solution @xmath87 of the camassa holm equation .
in fact , the proof of this result is almost immediate from the general results on asymptotic analysis for our coupling problem of entire functions derived in the previous section .
[ thmlt ] let @xmath106 be a closed sector which contains at most finitely many of the rays @xmath107 , @xmath10 given by @xmath108 .
then we have @xmath109 for @xmath110 as @xmath111 , where the phase shifts @xmath112 for each @xmath10 are given by @xmath113 for every @xmath98 and @xmath90 we introduce the entire functions @xmath114 which will turn out to be a solution of a particular coupling problem . in fact , one clearly has the coupling condition ( c ) @xmath115 moreover , due to ( * ? ? ?
* proposition 4.4 ) , the function @xmath116 is a herglotz nevanlinna function , ensuring the growth and positive condition ( g ) . in fact , this can also be verified by a direct calculation , using the differential equation . finally , the normalization condition ( n ) is immediate from the definition .
we will first consider the special case when the sector @xmath117 contains precisely one ray , say @xmath118 for some @xmath119 . upon defining the sets
@xmath120 by @xmath121 one obtains a partition @xmath122 of @xmath20 such that @xmath123 for some @xmath124 , @xmath51 .
therefore , the coupling constants in satisfy @xmath125 for @xmath110 as @xmath111 . in view of theorem [ thmaa ] and ( * ? ? ?
* lemma 3.4 ) this yields @xmath126 for @xmath110 as @xmath111 . but
this proves the claim in this special case , since @xmath127 for @xmath110 as @xmath111 , as lebesgue s dominated convergence theorem shows .
( 8.8 , 1 ) ( 8.8,4.7 )
( 5,4.7 ) cycle ; ( 4,0 )
( -0.8 , 1 ) ( -0.8,4.7 ) ( 3,4.7 ) cycle ; plot ( , 3 + 0.9*exp(-6*abs(-7 . ) ) + 0.5*exp(-6*abs(-5.9 ) ) + 0.3*exp(-6*abs(-4.9 ) ) - 1.4*exp(-6*abs(-1 ) ) - 0.7*exp(-6*abs(-2.2 ) ) - 0.2*exp(-6*abs(-3.1))+ 0.1*exp(-6*abs(-4.5 ) ) - 0.1*exp(-6*abs(-3.6 ) ) ) node[right ] @xmath128 ; plot ( , 3 + 0.9*exp(-6*abs(-7 . ) ) + 0.5*exp(-6*abs(-5.9 ) ) + 0.3*exp(-6*abs(-4.9 ) ) - 1.4*exp(-6*abs(-1 ) ) - 0.7*exp(-6*abs(-2.2 ) ) - 0.2*exp(-6*abs(-3.1))+ 0.1*exp(-6*abs(-4.5 ) ) - 0.1*exp(-6*abs(-3.6 ) ) ) node[right ] ; ( 4,0 ) ( 9,2 ) node[right ] @xmath129 ; ( 4,0 )
( 9,5 ) node[right ] @xmath130 ; ( 4,0 )
( 7,5 ) node[right ] @xmath131 ; ( 4,0 ) ( 5.5,5 ) node[right ] @xmath132 ; ( 4,0 )
( 4.7,5 ) node[right ] ; ( 4,0 )
( 4.3,5 ) node[right ] ; ( 4,0 ) ( 4.12,5 ) node[right ] ; ( 4,0 )
( 4.05,5 ) node[right ] ; ( 4,0 )
( 4.02,5 ) node[right ] ; ( 4,0 ) ( -1,2.4 ) node[left ] @xmath133 ; ( 4,0 )
( -1,4.8 ) node[left ] @xmath134 ; ( 4,0 ) ( 0.8,5 ) node[left ] @xmath135 ; ( 4,0 ) ( 2.4,5 ) node[left ] @xmath136 ; ( 4,0 ) ( 3.5,5 ) node[left ] ; ( 4,0 ) (
3.7,5 ) node[left ] ; ( 4,0 ) ( 3.88,5 ) node[left ] ; ( 4,0 )
( 3.95,5 ) node[left ] ; ( 4,0 ) ( 3.98,5 ) node[left ] ; hereby , the grey areas represent two sectors in which our long - time asymptotics hold uniformly .
each of the rays @xmath107 , accumulating at the @xmath137-axis , corresponds to an eigenvalue @xmath10 of the underlying isospectral problem .
after long time , one can see that the solution @xmath87 splits into a train of single peakons , each of which travels along one of the rays , with height and speed determined by the corresponding eigenvalue . due to the conditions on the sector in theorem [ thmlt ] , we do not obtain long - time asymptotics of solutions , which hold uniformly in sectors around the @xmath137-axis ( as long as the spectrum is not finite , that is , in the multi - peakon case ) . however , we are able to derive long - time asymptotics which hold uniformly in strips near the @xmath137-axis , that is , as long as @xmath2 stays bounded ( cf .
* theorem 1.2 ) ) . with the notation from the proof of theorem [ thmlt ]
we see that the coupling constants in for every @xmath10 satisfy @xmath141 as @xmath111 .
therefore , an application of corollary [ coraa ] shows that @xmath142 as @xmath111 , in view of ( * ? ? ?
* lemma 3.4 ) .
to finish the proof note that the right hand side of converges to zero as @xmath111 , as an application of lebesgue s dominated convergence theorem shows .
* acknowledgments . *
we gratefully acknowledge the kind hospitality of the _ institut mittag - leffler _ ( djursholm , sweden ) during the scientific program on _ inverse problems and applications _ in spring 2013 , where most of this article was written .
a. boutet de monvel and d. shepelsky , _ riemann hilbert problem in the inverse scattering for the camassa holm equation on the line _ , in `` probability , geometry and integrable systems '' , 5375 , math .
publ . , 55 , cambridge univ . press , cambridge , 2008 .
p. a. deift , a. r. its , and x. zhou , _ long - time asymptotics for integrable nonlinear wave equations _ , in `` important developments in soliton theory '' , 181204 , springer ser .
nonlinear dynam . , springer , berlin , 1993 .
a. r. its , _ asymptotic behavior of the solutions to the nonlinear schrdinger equation , and isomonodromic deformations of systems of linear differential equations _ ,
* 24:3 * ( 1981 ) , 452456 . | we propose a novel technique for analyzing the long - time asymptotics of integrable wave equations in the case when the underlying isospectral problem has purely discrete spectrum . to this end
, we introduce a natural coupling problem for entire functions , which serves as a replacement for the usual riemann hilbert problem , which does not apply in these cases . as a prototypical example
, we investigate the long - time asymptotics of the dispersionless camassa holm equation . |
the problem of determining the angle @xmath0 at which a point mass launched from ground level with a given speed @xmath1 will reach a maximum distance is a standard exercise in mechanics .
there are many possible ways of solving this problem @xcite , leading to the well - known answer of @xmath2 , producing a maximum range of @xmath3 , with @xmath4 being the free - fall acceleration .
conceptually and calculationally more difficult problems have been suggested to improve student proficiency in projectile motion @xcite , with the most famous example being the tarzan swing problem @xcite .
the problem of determining the maximum distance of a point mass thrown from constant - speed circular motion is presented and analyzed in detail in this text .
the calculational results confirm several conceptually derived conclusions regarding the initial throw position and provide some details on the angles and the way of throwing ( underhand or overhand ) which produce the maximum throw distance . the situation analyzed in this text
can be defined as follows : _ `` suppose you want to throw a stone ( approximated by a point mass ) as far horizontally as possible .
the stone rotates in a vertical circle with constant speed @xmath1 . at which point during the motion
should the stone be released ?
should it rotate clockwise ( an overhand throw ) or counter - clockwise ( an underhand throw ) ?
the center of rotation is at height @xmath5 , where @xmath6 is the radius of rotation .
the horizontal distance is measured from the point on the ground directly below the center of rotation . ''
_ an illustration of the problem is given in fig.[figure1 ] .
this problem poses several conceptual difficulties . during motion , the initial height , the horizontal distance with respect to the reference point on the ground and the launch angle all change . since all of these influence the final horizontal distance , it is not easy to deduce exactly what kind of throw should be executed to attain the maximum distance for a given speed .
let s assume that the throw is executed to the right ( this does not restrict the solution in any way ) . for an overhand throw ,
the stone needs to be released during movement through the upper part of the circle , since then it is traveling to the right . during the first part of the motion , the angle of the stone velocity with the horizon is positive , so
the stone is thrown upwards , but the initial horizontal distance from the reference point on the ground is negative . during the second part of the motion , the opposite holds true .
it is clear that if the initial speed is nearly zero , the stone should be released when it is as forward as possible , since then it is practically released from rest and it drops vertically and lands at a distance @xmath6 away from the reference point . on the other hand , if the initial speed of the stone is very large , in the sense that the initial displacement from the reference point on the ground is very small compared to the range of the throw , one would expect that the classical result of an angle equal to @xmath2 produces the largest horizontal distance . for an underhand throw ,
the the stone needs to be released during movement through the lower part of the circle , since then it is traveling to the right . in this case , it is more obvious that the release should happen during the second part of the motion , since then the throw is executed upwards and the initial horizontal displacement of the stone is positive .
again , for a low speed , the stone should be released when it is as forward as possible and for a high speed , it should be released when the throw angle with respect to the horizon is equal to @xmath2 .
interestingly , it is unclear whether the throw should be made overhand or underhand to obtain the maximum throw distance for a general value of speed . to answer this question
, some knowledge of elementary kinematics and numerical calculation is necessary .
we can define the coordinate system as in fig.[figure2 ] .
clearly , there are two cases to consider : the overhand and the underhand throw .
marks the angle the stone velocity makes with the horizontal line.,scaledwidth=70.0% ] let @xmath0 mark the angle with the horizontal line when the stone is released ( set equal to @xmath7 ) .
the initial coordinates of the stone are : @xmath8 the moment when the stone hits the ground is found by setting @xmath9 in the general equation of motion @xmath10 , which gives one physical solution for @xmath11 . inserting that solution into @xmath12 ,
the throw horizontal distance becomes : @xmath13 the absolute sign is required to take into account possible negative values of angle @xmath0 .
the notation in which the upper sign represents the overhand throw and the lower represents the underhand throw is introduced .
the trajectory equations here assume no air drag . the maximum distance of the throw can be found numerically or graphically by plotting @xmath14 as a function of the inital speed @xmath1 and
throw angle @xmath0 .
another approach is to set @xmath15 for a certain intial speed @xmath1 , which often has the benefit of emphasizing the dimensionless variables relevant to the problem . taking the derivative one obtains the following condition , after some algebra : @xmath16 where the shorthands @xmath17 and @xmath18 were introduced and @xmath19 denotes the throw angle at which the range is maximal . at this point
, we will use the simplification @xmath20 and note that in that case @xmath21 is twice the ratio of the kinetic energy of the stone to its gravitational potential energy at the lowest point of rotation in the chosen coordinate system . even though numerical solving was skipped in ( [ eq2 ] ) , here it needs to be employed .
the maximum angle results are obtained when varying the value of @xmath21 from 0 to 50 , separately for the overhand and the underhand cases .
the value of @xmath22 corresponds to the case where the stone is released from rest , while @xmath23 corresponds to the case in which the kinetic energy of the stone is much larger than its initial potential energy . at the end of the calculation
, the throw distance can be found for a specific @xmath24 angle value .
the results of numerically solving ( [ eq_condition ] ) are given on fig.[figure3 ] .
as suspected earlier , the angles obtained for an underhand throw approach @xmath25 for small values of @xmath21 and @xmath26 for large values of @xmath21 .
interestingly , the decline from @xmath27 to @xmath28 is not uniform . with @xmath21 increasing from zero , the angle of maximum distance decreases until @xmath21 reaches a minimum value at @xmath29 , corresponding to a minimum angle of @xmath30 , and then starts increasing again until it asymptotically reaches @xmath28 . for an overhand throw , the maximum distance for very small values of @xmath21
is obtained when @xmath31 , as concluded earlier .
with increasing @xmath21 , the throw angle decreases to zero .
a throw at @xmath32 , which happens at the top of the motion , is achieved when @xmath33 , i.e. when @xmath34 . just as in the case of an underhand throw ,
when further increasing @xmath21 , the throw velocity angle approaches @xmath26 .
at which the throw distance is maximum for a given initial speed in cases of an underhand and an overhand throw .
the parameter @xmath35 , where @xmath1 is the initial speed .
the range of @xmath21 in the graphs is from 0.03 to 20.,title="fig:",scaledwidth=45.0% ] at which the throw distance is maximum for a given initial speed in cases of an underhand and an overhand throw .
the parameter @xmath35 , where @xmath1 is the initial speed .
the range of @xmath21 in the graphs is from 0.03 to 20.,title="fig:",scaledwidth=45.0% ] at which the throw distance is maximum for a given initial speed .
right : the scaled maximum horizontal distance @xmath36 covered by the stone when launched with a given initial speed .
the dashed red line represents an underhand throw , the solid blue line an overhand throw . on the right graph
, @xmath21 ranges from 0.03 to 2.,title="fig:",scaledwidth=45.0% ] at which the throw distance is maximum for a given initial speed .
right : the scaled maximum horizontal distance @xmath36 covered by the stone when launched with a given initial speed .
the dashed red line represents an underhand throw , the solid blue line an overhand throw . on the right graph
, @xmath21 ranges from 0.03 to 2.,title="fig:",scaledwidth=45.0% ] using the values for the obtained angles , one can calculate the maximum throw distances , as shown on fig.[figure4 ] .
it is interesting to note that for speeds for which @xmath37 and @xmath22 , the maximum distances are the same in the cases of an overhand and an underhand throw . in the interval @xmath38 , the maximum attainable distance is greater in the case of an overhand throw .
this result could not be simply deduced without calculation , since after an overhand throw for which @xmath39 the stone has an initial velocity pointing downwards , decreasing its fall time and thus its final horizontal distance .
the results of the calculation show several interesting features . for a throw speed equal to zero
, the stone should be held as far forward as possible before release , performing a simple vertical drop .
the distance it reaches in this case is equal to @xmath6 . when increasing the speed in the interval @xmath40 , it is more beneficial to perform an overhand throw , even though the stone release is performed when its velocity is pointing downwards . for @xmath41 ,
the maximum distances for an overhand and an underhand throw become exactly the same , but the throws are performed at different angles .
for example , in the limiting case of @xmath34 , an overhand throw is performed at the top of the motion ( when @xmath42 ) , while an underhand throw is performed when @xmath43 . with increasing speed ,
the maximum distance increases almost linearly with @xmath21 and the angles for both types of throws tend towards @xmath28 . for an underhand throw , the angle at which the maximum distance is attained dips below its asymptotic value to the value of @xmath30 before it starts increasing towards @xmath28 .
this result might present a conceptual difficulty
after all , for a large enough speed , is it not always more beneficial that the throw is performed at @xmath44 , since in that case the stone is both more forward and higher ? even though this statement is true , one should remember that the stone is not thrown from the ground level , and so the angle at which it attains maximal distance can not be exactly @xmath28 .
however , it is worth noting that in the case of an underhand throw , the convergence to @xmath28 seems very rapid ( when compared to the overhand throw ) .
all the results are summarized on fig.[figure5 ] .
the circumference of the circle shows the points through which the stone can move .
the angles given are the angles which the stone has with respect to the horizontal line when released from a certain point .
the top part of the circle represents overhand throws and the bottom part underhand throws . using the color scale , for
a given value of @xmath21 one can find two corresponding angles on the circle which produce the maximum throw distance .
the color scale is logarithmic since large arcs of the circle correspond to small changes in @xmath21 .
it can be seen that for @xmath22 the stone is released when it is furthest to the right . with increasing @xmath21
, the angle grows for an overhead throw , asymptotically reaching @xmath28 . for underhand throws ,
the angle first decreases below @xmath28 and then increases back towards it , as discussed .
the decrease and the increase are drawn separately on the graphic to avoid overlap . for overhand throws ( top part ) and underhand throws ( bottom part of the circle).,scaledwidth=70.0% ]
the problem of throwing a point particle from a vertical circular constant - speed motion was considered .
the optimal throw angle @xmath19 which maximizes the throw distance was found as a function of @xmath17 and of the throw type .
the results confirmed some conceptually derived predictions , but also provided some new insights . to obtain the maximum throw distance
when @xmath39 it is more beneficial to use the overhand throw .
interestingly , for @xmath37 the maximum throw distance becomes the same for an overhand and an underhand throw , although the throws are executed at different angles with respect to horizon .
for very large values of @xmath21 , the optimal angle in both cases equals @xmath28 and the throw distance becomes @xmath45 .
the result is approximate since the throw is not performed from ground level .
if one wants to throw a stone as far as possible in real life , many effects such as air drag and the spin of the stone , among others , need to be considered .
furthermore , vertical circular motion is not a good model for human overhand throwing which is a complex sequence of torques and levers that can generate more initial speed than underhand throwing .
nevertheless , it would be interesting to construct an apparatus to experimentally demonstrate the results outlined here , as well as evaluate the effects due to air drag and other factors , which is considered as a future step of this article .
9 john s. thomsen , `` maxima and minima without calculus '' , am.j.phys . * 52*(10 ) , 881 - 883 ( 1984 ) r. palffy - muhoray and d. balzarini , `` maximizing the range of the shot put without calculus '' , am . j. phys .
* 52 * , 181 - 181 ( 1982 ) william s. porter , `` the range of a projectile '' , phys .
teach * 15 * , 358 - 358 ( 1977 ) harvard u. problem of the week 71 , https://goo.gl/0h6jzx ronald a. brown , `` maximizing the range of a projectile '' , phys .
* 30 * , 344 ( 1992 ) d. bittel , `` maximizing the range of a projectile thrown by a simple pendulum '' , phys .
teach * 43 * , 98 ( 2005 ) m. rave , m. sayers , `` tarzan s dilemma : a challenging problem for introductory physics students '' , phys . teach * 51
* , 456 ( 2013 ) | the problem of determining the angle at which a point mass launched from ground level with a given speed is a standard exercise in mechanics . similar , yet conceptually and calculationally more difficult problems have been suggested to improve student proficiency in projectile motion .
the problem of determining the maximum distance of a rock thrown from a rotating arm motion is presented and analyzed in detail in this text .
the calculational results confirm several conceptually derived conclusions regarding the initial throw position and provide some details on the angles and the way of throwing ( underhand or overhand ) which produce the maximum throw distance .
+ + to be published in * the physics teacher . * |
breadth first search ( bfs ) is a fundamental graph traversal strategy .
it can also be viewed as computing single source shortest paths on unweighted graphs .
it decomposes the input graph @xmath5 of @xmath0 nodes and @xmath6 edges into at most @xmath0 levels where level @xmath7 comprises all nodes that can be reached from a designated source @xmath8 via a path of @xmath7 edges , but can not be reached using less than @xmath7 edges .
the objective of a dynamic graph algorithm is to efficiently process an online sequence of update and query operations ; see @xcite for overviews of classic and recent results . in our case
we consider bfs under a sequence of either @xmath2 edge insertions , but not deletions ( _ incremental _ version ) or @xmath2 edge deletions , but not insertions ( _ decremental _ version ) .
after each edge insertion / deletion the updated bfs level decomposition has to be output .
we consider the commonly accepted external - memory ( em ) model of aggarwal and vitter @xcite .
it assumes a two level memory hierarchy with faster internal memory having a capacity to store @xmath9 vertices / edges . in an i / o operation
, one block of data , which can store @xmath10 vertices / edges , is transferred between disk and internal memory .
the measure of performance of an algorithm is the number of i / os it performs .
the number of i / os needed to read @xmath11 contiguous items from disk is @xmath12 .
the number of i / os required to sort @xmath11 items is @xmath13 . for all realistic values of @xmath11 , @xmath10 , and @xmath9 , @xmath14 .
there has been a significant number of publications on external - memory graph algorithms ; see @xcite for recent overviews . however , we are not aware of any dynamic graph algorithm in the fully external - memory case ( where @xmath15 ) .
we provide the first non - trivial result on dynamic bfs in external - memory . for general sparse undirected graphs of initially @xmath0 nodes and @xmath1 edges and either @xmath2 edge insertions or @xmath2 edge deletions , we prove an amortized high - probability bound of @xmath3 i / os per update .
in contrast , the currently best bound for static bfs on sparse undirected graphs is @xmath16 i / os @xcite .
also note that for general sparse graphs and worst - case monotone sequences of @xmath2 updates in _ internal - memory _ there is asymptotically no better solution than performing @xmath2 runs of the linear - time static bfs algorithm , even if after each update we are just required to report the changes in the bfs tree ( see fig .
[ bezug ] for an example ) . in case
@xmath4 i / os should prove to be a lower bound for static bfs in external - memory , then our result yields an interesting differentiator between static vs. dynamic bfs in internal and external memory . in section [ static ] we will review known bfs algorithms for static undirected graphs .
then we consider traditional and new external - memory methods for graph clustering ( section [ clusters ] ) .
subsequently , in section [ algorithm ] we provide the new algorithm and analyze it in section [ analysis ] .
final remarks concerning extensions and open problems are given in sections [ extensions ] and [ conclusions ] , respectively .
* internal - memory . *
bfs is well - understood in the ram model .
there exists a simple linear time algorithm @xcite ( hereafter referred as im_bfs ) for the bfs traversal in a graph .
im_bfs keeps a set of appropriate candidate nodes for the next vertex to be visited in a fifo queue @xmath17 .
furthermore , in order to find out the unvisited neighbors of a node from its adjacency list , it marks the nodes as either visited or unvisited . unfortunately , as the storage requirements of the graph starts approaching the size of the internal memory , the running time of this algorithm deviates significantly from the predicted @xmath18 asymptotic performance of the ram model : checking whether edges lead to already visited nodes altogether needs @xmath19 i / os in the worst case ; unstructured indexed access to adjacency lists may add another @xmath20 i / os . + * em - bfs for dense undirected graphs . *
the algorithm by munagala and ranade @xcite ( referred as mr_bfs ) ignores the second problem but addresses the first by exploiting the fact that the neighbors of a node in bfs level @xmath21 are all in bfs levels @xmath22 , @xmath21 or @xmath23 .
let @xmath24 denote the set of nodes in bfs level @xmath23 , and let @xmath25 be the multi - set of neighbors of nodes in @xmath26 .
given @xmath26 and @xmath27 , mr_bfs builds @xmath24 as follows : firstly , @xmath25 is created by @xmath28 random accesses to get hold of the adjacency lists of all nodes in @xmath26 .
thereafter , duplicates are removed from @xmath25 to get a sorted set @xmath29 .
this is done by sorting @xmath25 according to node indices , followed by a scan and compaction phase .
the set @xmath30 is computed by scanning `` in parallel '' the sorted sets of @xmath31 , and @xmath27 to filter out the nodes already present in @xmath26 or @xmath27 .
the resulting worst - case i / o - bound is @xmath32 .
the algorithm outputs a bfs - level decomposition of the vertices , which can be easily transformed into a bfs tree using @xmath33 i / os @xcite . + * em - bfs for sparse undirected graphs .
* mehlhorn and meyer suggested another approach @xcite ( mm_bfs ) which involves a preprocessing phase to restructure the adjacency lists of the graph representation .
it groups the vertices of the input graph into disjoint clusters of small diameter in @xmath34 and stores the adjacency lists of the nodes in a cluster contiguously on the disk .
thereafter , an appropriately modified version of mr_bfs is run .
mm_bfs exploits the fact that whenever the first node of a cluster is visited then the remaining nodes of this cluster will be reached soon after . by spending only one random access ( and possibly , some sequential accesses depending on cluster size ) to load the whole cluster and then keeping the cluster data in some efficiently accessible data structure ( pool ) until it is all processed , on sparse graphs the total amount of i / os can be reduced by a factor of up to @xmath35 : the neighboring nodes of a bfs level can be computed simply by scanning the pool and not the whole graph . though some edges may be scanned more often in the pool , unstructured i / os to fetch adjacency lists is considerably reduced , thereby reducing the total number of i / os . +
mehlhorn and meyer @xcite proposed the algorithms mm_bfs_r and mm_bfs_d , out of which the first is randomized and the second is deterministic . in mm_bfs_r ,
the partitioning is generated `` in parallel rounds '' : after choosing master nodes independently and uniformly at random , in each round , each master node tries to capture all unvisited neighbors of its current sub - graph into its partition , with ties being resolved arbitrarily . a similar kind of randomized preprocessing is also applied in parallel @xcite and streaming @xcite settings .
there , however , a dense compressed graph among the master nodes is produced , causing rather high parallel work or large total streaming volume , respectively .
the mm_bfs_d variant first builds a spanning tree @xmath36 for the connected component of @xmath34 that contains the source node .
arge et al .
@xcite show an upper bound of @xmath37 i / os for computing such a spanning tree .
each undirected edge of @xmath36 is then replaced by two oppositely directed edges . note that a bi - directed tree always has at least one euler tour . in order to construct the euler tour around this bi - directed tree ,
each node chooses a cyclic order @xcite of its neighbors .
the successor of an incoming edge is defined to be the outgoing edge to the next node in the cyclic order .
the tour is then broken at the source node and the elements of the resulting list are then stored in consecutive order using an external memory list - ranking algorithm ; chiang et al .
@xcite showed how to do this in sorting complexity .
thereafter , we chop the euler tour into _ chunks _ of @xmath38 nodes and remove duplicates such that each node only remains in the first chunk it originally occurs ; again this requires a couple of sorting steps .
the adjacency lists are then re - ordered based on the position of their corresponding nodes in the chopped duplicate - free euler tour : all adjacency lists for nodes in the same chunks form a cluster and the distance in @xmath34 between any two vertices whose adjacency - lists belong to the same cluster is bounded by @xmath38 .
the preprocessing methods for the static bfs in @xcite may produce very unbalanced clusters : for example , with mm_bfs_d using chunk size @xmath39 there may be @xmath40 clusters being in charge of only @xmath41 adjacency - lists each . for the dynamic version ,
however , we would like to argue that each random access to a cluster not visited so far provides us with @xmath42 new adjacency - lists .
unfortunately , finding such a clustering i / o - efficiently seems to be quite hard .
therefore , we shall already be satisfied with an euler tour based randomized construction ensuring that the _ expected _ number of adjacency - lists kept in all but one vertices where @xmath43 denotes the number of vertices in the connected component of the starting node @xmath8 . ]
clusters is @xmath42 .
the preprocessing from mm_bfs_d is modified as follows : each vertex @xmath44 in the spanning tree @xmath36 is assigned an independent binary random number @xmath45 with @xmath46= \mathbf{p}[r(v)=1]=1/2 $ ] .
when removing duplicates from the euler tour , instead of storing @xmath44 s adjacency - list in the cluster related to the chunk with the _ first _ occurrence of a vertex @xmath44 , now we only stick to its first occurrence iff @xmath47 and otherwise ( @xmath48 ) store @xmath44 s adjacency - list in the cluster that corresponds to the _ last _ chunk of the euler tour @xmath44 appears in . for leaf nodes
@xmath44 , there is only one occurrence on the tour , hence the value of @xmath45 is irrelevant .
obviously , each adjacency - lists is stored only once .
furthermore , the modified procedure maintains all good properties of the standard preprocessing within mm_bfs_d like guaranteed bounded distances of @xmath49 in @xmath34 between the vertices belonging to the same cluster and @xmath50 clusters overall . for chunk size @xmath51 and each but the last chunk , the expected number of adjacency - lists kept is at least @xmath52 .
let @xmath53 be the sequence of vertices visited by an arbitrary chunk @xmath54 of the euler tour @xmath55 , excluding the last chunk .
let @xmath56 be the number of entries in @xmath57 that represent first or last visits of inner - tree vertices from the spanning tree @xmath36 on @xmath55 .
these @xmath56 entries account for an expected number of @xmath58 adjacency - lists actually stored and kept in @xmath54 .
note that if for some vertex @xmath59 both its first and last visit happen within @xmath54 , then @xmath44 s adjacency - list is kept with probability one .
similarly , if there are any visits of leaf nodes from @xmath36 within @xmath54 , then their adjacency - lists are kept for sure ; let @xmath60 denote the number of these leaf node entries in @xmath57 .
what remains are @xmath61 _ intermediate _
( neither first nor last ) visits of vertices within @xmath54 ; they do not contribute any additional adjacency - lists .
we can bound @xmath61 using the observation that any intermediate visit of a tree node @xmath44 on @xmath55 is preceded by a last visit of a child @xmath62 of @xmath44 and proceeded by a first visit of another child @xmath63 of @xmath44 .
thus , @xmath64 , that is @xmath65 , which implies that the expected number of distinct adjacency - lists being kept for @xmath54 is at least @xmath66 .
in this section we concentrate on the incremental version for sparse graphs with @xmath2 updates where each update inserts an edge .
thus , bfs levels can only decrease over time .
before we start , let us fix some notation : for @xmath67 , @xmath68 is to denote the graph after the @xmath7-th update , @xmath69 is the initial graph .
let @xmath70 , @xmath71 , stand for the bfs level of node @xmath44 if it can be reached from the source node @xmath8 in @xmath72 and @xmath0 otherwise .
furthermore , for @xmath73 , let @xmath74 .
the main ideas of our approach are as follows : * checking connectivity ; type a updates . * in order to compute the bfs levels for @xmath72 , @xmath73 , we first run an em connected components algorithm ( for example the one in @xcite taking @xmath75 i / os ) in order to check , whether the insertion of the @xmath7-th edge @xmath76 enlarges the connected component @xmath77 of the source vertex @xmath8 .
if yes ( let us call this a _ type a update _ ) , then w.l.o.g . let @xmath78 and let @xmath79 be the connected component that comprises @xmath44 .
the new edge @xmath76 is then the only connection between the existing bfs - tree for @xmath8 and @xmath79 .
therefore , we can simply run mr_bfs on the subgraph @xmath80 defined by the vertices in @xmath79 with source @xmath44 and add @xmath81 to all distances obtained .
this takes @xmath82 i / os where @xmath83 denotes the number of vertices in @xmath79 . if the @xmath7-th update does not merge @xmath77 with some other connected component but adds an edge within @xmath77 ( _ type b update
_ ) then we need to do something more fancy : * dealing with small changes ; type b updates . * now for computing the bfs levels for @xmath72 , @xmath73 , we pre - feed the adjacency - lists into a sorted pool @xmath84 according to the bfs levels of their respective vertices in @xmath85 using a certain advance @xmath86 , i.e. , the adjacency list for @xmath44 is added to @xmath84 when creating bfs level @xmath87 of @xmath72 .
this can be done i / o - efficiently as follows .
first we extract the adjacency - lists for vertices having bfs levels up to @xmath88 in @xmath85 and put them to @xmath84 where they are kept sorted by node indices . from the remaining adjacency - lists we build a sequence @xmath89 by sorting them according to bfs levels in @xmath85 ( primary criterion ) and node indices ( secondary criterion ) . for the construction of each new bfs level of @xmath72
we merge a subsequence of @xmath89 accounting for one bfs level in @xmath85 with @xmath84 using simple scanning .
therefore , if @xmath90 for all @xmath91 then all adjacency - lists will be added to @xmath84 in time and can be consumed from there without random i / o .
each adjacency - list is scanned at most once in @xmath89 and at most @xmath88 times in @xmath84 .
thus , if @xmath92 this approach causes less i / o than mm_bfs . *
dealing with larger changes . * unfortunately , in general , there may be vertices @xmath44 with @xmath93 .
their adjacency - lists are not prefetched into @xmath84 early enough and therefore have to be imported into @xmath84 using random i / os to whole clusters just like it is done in mm_bfs .
however , we apply the modified clustering procedure described in section [ ssec : modified : preprocessing ] on @xmath85 , the graph without the @xmath7-th new edge ( whose connectivity is the same as that of @xmath72 ) with chunk size @xmath94 .
note that this may result in @xmath95 cluster accesses , which would be prohibitive for small @xmath88 .
therefore we restrict the number of random cluster accesses to @xmath96 .
if the dynamic algorithm does not succeed within these bounds then it increases @xmath88 by a factor of two , computes a new clustering for @xmath85 with larger chunk size and starts a _ new attempt _ by repeating the whole approach with the increased parameters .
note that we do not need to recompute the spanning tree for the for the second , third , @xmath97 attempt . * at most @xmath98 attempts per update . *
the @xmath99-th attempt , @xmath100 , of the dynamic approach to produce the new bfs - level decomposition will apply an advance of @xmath101 and recompute the modified clustering for @xmath85 using chunk size @xmath102 . note that there can be at most @xmath103 failing attempts for each edge update since by
then our approach allows sufficiently many random accesses to clusters so that all of them can be loaded explicitly resulting in an i / o - bound comparable to that of static mm_bfs . in section [ analysis ] , however , we will argue that for most edge updates within a longer sequence , the advance value and the chunk size value for the succeeding attempt are bounded by @xmath104 implying significantly improved i / o performance . *
restricting waiting time in @xmath84 .
* there is one more important detail to take care of : when adjacency - lists are brought into @xmath84 via explicit cluster accesses ( because of insufficient advance @xmath105 in the prefetching ) , these adjacency - lists will re - enter @xmath84 once more later on during the ( for these adjacency - lists by then useless ) prefetching .
thus , in order to make sure that unnecessary adjacency - lists do not stay in @xmath84 forever , each entry in @xmath84 carries a time - stamp ensuring that superfluous adjacency - lists are evicted from @xmath84 after at most @xmath106 bfs levels .
[ lemma : single : typeb ] for sparse graphs with @xmath1 updates , each type b update succeeding during the @xmath99-th attempt requires @xmath107 i / os .
deciding whether a type b update takes place essentially requires a connected components computation , which accounts for @xmath75 i / os .
within this i / o bound we can also compute a spanning tree @xmath36 of the component holding the starting vertex @xmath8 but excluding the new edge .
subsequently , there are @xmath108 attempts , each of which uses @xmath109 i / os to derive a new modified clustering based on an euler tour with increasing chunk sizes around @xmath36 . furthermore , before each attempt we need to initialize @xmath84 and @xmath89 , which takes @xmath109 i / os per attempt .
the worst - case number of i / os to ( re- ) scan adjacency - lists in @xmath84 or to explicitly fetch clusters of adjacency - lists doubles after each attempt .
therefore it asymptotically suffices to consider the ( successful ) last attempt @xmath99 , which causes @xmath110 i / os .
furthermore , each attempt requires another @xmath109 i / os to pre - sort explicitly loaded clusters before they can be merged with @xmath84 using a single scan just like in mm_bfs . adding all contributions yields the claimed i / o bound of @xmath107 for sparse graphs .
we split our analysis of the incremental bfs algorithm into two parts .
the first ( and easy one ) takes care of type a updates : for sparse undirected graphs with @xmath2 updates , there are at most @xmath111 type a updates causing @xmath112 i / os in total .
each type a update starts with an em connected components computation causing @xmath75 i / os per update .
since each node can be added to the connected component @xmath77 holding the starting vertex @xmath8 only once , the total number of i / os spend in calls to the mr - bfs algorithm on components to be merged with @xmath77 is @xmath113 .
producing the output takes another @xmath109 per update .
now we turn to type b updates : for sparse undirected graphs with @xmath2 updates , all type b updates cause @xmath114 i / os in total with high probability .
recall that @xmath70 , @xmath71 , stands for the bfs level of node @xmath44 if it can be reached from the source node @xmath8 in @xmath72 and @xmath0 otherwise .
if upon the @xmath7-th update the dynamic algorithm issues an explicit fetch for the adjacency - lists of some vertex @xmath44 kept in some cluster @xmath115 then this is because @xmath116 for the current advance @xmath88 .
note that for all other vertices @xmath117 , there is a path of length at most @xmath118 in @xmath85 , implying that @xmath119 as well as @xmath120 .
having current chunk size @xmath121 , this implies @xmath122 if the @xmath7-th update needs @xmath99 attempts to succeed then , during the ( failing ) attempt @xmath123 , it has tried to explicitly access @xmath124 distinct clusters . out of these at least @xmath125 clusters
carry an expected amount of at least @xmath126 adjacency - lists each .
this accounts for an expected number of at least @xmath127 distinct vertices , each of them featuring @xmath128 .
with probability at least @xmath129 we actually get at least half of the expected amount of distinct vertices / adjacency - lists , i.e. , @xmath130 . therefore , using the definitions @xmath131 and @xmath132 , if the @xmath7-th update succeeds within the @xmath99-th attempt we have @xmath133 with probability at least @xmath129 .
let us call this event _ a large @xmath99-yield_. since each attempt uses a new clustering with independent choices for @xmath134 , if we consider two updates @xmath135 and @xmath136 that succeed after the same number of attempts @xmath99 , then both @xmath135 and @xmath136 have a large yield with probability at least @xmath129 , independent of each other .
therefore , we can use chernoff bounds @xcite in order to show that out of @xmath137 updates that all succeed within their @xmath99-th attempt , at least @xmath138 of them have a large @xmath99-yield with probability at least @xmath139 for an arbitrary positive constant @xmath140 .
subsequently we will prove an upper bound on the total number of large @xmath99-yields that can occur during the whole update sequence .
the quantity @xmath141 provides a global measure as for how much the bfs levels change after inclusion of the @xmath7-th edge from the update sequence .
if there are @xmath142 edge inserts in total , then @xmath143 a large @xmath99-yield means @xmath144 .
therefore , in the worst case there are at most @xmath145 large @xmath99-yield updates and according to our discussion above it needs at most @xmath146 updates that succeed within the @xmath99-th attempt to have at least @xmath147 large @xmath99-yield updates with high probability .
as observed before , the dynamic algorithm will not increase its advance and chunk size values beyond @xmath148 implying @xmath149 .
but then we have @xmath150 and @xmath151 for sufficiently large @xmath0 . ] . for the last step of our analysis we will distinguish two kinds of type
b updates : those that finish using an advance value @xmath152 ( type b1 ) , and the others ( type b2 ) .
independent of the subtype , an update costs @xmath153 @xmath154 @xmath155 i / os by lemma [ lemma : single : typeb ] .
obviously , for an update sequence of @xmath142 edge insertions there can be at most @xmath2 updates of type b1 , each of them accounting for at most @xmath156 i / os . as for type
b2 updates we have already shown that with high probability there are at most @xmath157 updates that succeed with advance value @xmath158 . therefore , using boole s inequality , the total amount of i / os for all type b2 updates is bounded by @xmath159 @xmath160 with high probability .
combining the two lemmas of this section implies for general sparse undirected graphs of initially @xmath0 nodes and @xmath1 edges and @xmath2 edge insertions , dynamic bfs can be solved using amortized @xmath3 i / os per update with high probability .
having gone through the ideas of the incremental version , it is now close to trivial to come up with a symmetric external - memory dynamic bfs algorithm for a sequence of edge deletions : instead of pre - feeding adjacency - lists into using an _ advance _ of @xmath105 levels , we now apply a _
lag _ of @xmath105 levels .
therefore , the adjacency - list for a vertex @xmath44 is found in @xmath84 as long as the deletion of the @xmath7-th edge does not increase @xmath70 by more than @xmath88 .
otherwise , an explicit random access to the cluster containing @xmath44 s adjacency - list is issued later on .
all previously used amortization arguments and bounds carry through , the only difference being that @xmath161 values may monotonically increase instead of decrease .
better amortized bounds can be obtained if @xmath162 updates take place and/or @xmath69 has @xmath162 edges .
then we have the potential to amortize more random accesses per attempt , which leads to larger @xmath99-yields and reduces the worst - case number of expensive updates .
consequently , we can reduce the defining threshold between type b1 and type b2 updates , thus eventually yielding better amortized i / o bounds .
details we be provided in the full version of this paper .
modifications along similar lines are in order if external - memory is realized by flash disks @xcite : compared to hard disks , flash memory can sustain many more unstructured read i / os per second but on the other hand flash memory usually offers less read / write bandwidth than hard disks . hence , in algorithms like ours that are based on a trade - off between unstructured read
i / os and bulk read / write i / os , performance can be improved by allowing more unstructured read i / os ( fetching clusters ) if this leads to less overall i / o volume ( scanning hot pool entries ) .
we have given the first non - trivial external - memory algorithm for dynamic bfs .
even though we obtain significantly better i / o bounds than for the currently best static algorithm , there are a number of open problems : first of all , our bounds dramatically deteriorate for mixed update sequences ( edge insertions and edge deletions in arbitrary order and proportions ) ; besides oscillation effects , a single edge deletion ( insertion ) may spoil a whole chain of amortizations for previous insertions ( deletions ) . also , it would be interesting to see , whether our bounds can be further improved or also hold for shorter update sequences .
finally , it would be nice to come up with a deterministic version of the modified clustering .
l. arge , g. brodal , and l. toma . on external - memory mst , sssp and multi - way planar graph separation . in _ proc .
8th scand .
workshop on algorithmic theory ( swat ) _ , volume 1851 of _ lncs _ , pages 433447 .
springer , 2000 .
y. j. chiang , m. t. goodrich , e. f. grove , r. tamasia , d. e. vengroff , and j. s. vitter . external memory graph algorithms . in _ proc .
6th ann.symposium on discrete algorithms ( soda ) _ , pages 139149 .
acm - siam , 1995 .
j. s. vitter . external memory algorithms and data
structures : dealing with massive data . , pages 209271 , 2001 . revised version ( august 2007 ) available online at ` http://www.cs.purdue.edu/homes/jsv/papers/vit.io_survey.pdf ` . | we provide the first non - trivial result on dynamic breadth - first search ( bfs ) in external - memory : for general sparse undirected graphs of initially @xmath0 nodes and @xmath1 edges and monotone update sequences of either @xmath2 edge insertions or @xmath2 edge deletions , we prove an amortized high - probability bound of @xmath3 i / os per update .
in contrast , the currently best approach for static bfs on sparse undirected graphs requires @xmath4 i / os .
ulrich meyer |
recently , a great deal of attention has been payed to wireless sensor networks whose nodes sample a physical phenomenon ( hereinafter referred to as field ) , i.e. , air temperature , light intensity , pollution levels or rain falls , and send their measurements to a central processing unit ( or _ sink _ node ) .
the sink is in charge of reconstructing the sensed field : if the field can be approximated as bandlimited in the time and space domain , then an estimate of the discrete spectrum can be obtained .
however , the sensors measurements typically represent an irregular sampling of the field of interest , thus the sink operates based on a set of field samples that are not regularly spaced in the time and space domain .
the reasons for such an irregular sampling are multifold .
( i ) the sensors may be irregularly deployed in the geographical region of interest , either due to the adopted deployment procedure ( e.g. , sensors thrown out of an airplane @xcite ) , or due to the presence of terrain asperities and obstacles .
( ii ) the transmission of the measurements from the sensors to the central controller may fail due to bad channel propagation conditions ( e.g. , fading ) , or because collisions occur among the transmissions by sensors simultaneously attempting to access the channel . in this case , although the sample has been collected by the sensor , it will not be delivered to the central controller .
( iii ) the sensors may enter a low - power operational state ( sleep mode ) , in order to save energy @xcite . while in sleep mode , the nodes neither perform sensing operations nor transmit / receive any measurement .
( iv ) the sensors may be loosely synchronized , hence sense the field at different time instants .
clearly , sampling irregularities may result in a degradation of the reconstructed signal @xcite .
the work in @xcite investigates this issue in the context of sensor networks .
other interesting studies can be found in @xcite and @xcite , just to name a few , which address the perturbations of regular sampling in shift - invariant spaces @xcite and the reconstruction of irregularly sampled images in presence of measure noise @xcite . in this work ,
our objective is to evaluate the performance of the field reconstruction when the coordinates in the @xmath0-dimensional domain of the field samples , which reach the sink node , are randomly , independently distributed and the sensors measurements are noisy . we take as performance metric the mean square error ( mse ) on the reconstructed field . as a reconstruction technique , we use linear filtering and we adopt the filter that minimizes the mse ( i.e. , the lmmse filter ) @xcite . the matrix representing the sampling system , in the following denoted by @xmath2 ,
results to be a @xmath0-fold vandermonde matrix matrix @xmath5 is vandermonde if its @xmath6th entry , @xmath7 can be written as @xmath8 , @xmath9 . ] . by drawing on the results in @xcite
, we derive both the moments and an expression of the limiting spectral distribution ( lsd ) of @xmath3 , as the size of @xmath2 goes to infinity and its aspect ratio has a finite limit bounded away from zero .
then , by using such an asymptotic model , we approximate the mse on the reconstructed field through the @xmath4-transform @xcite of @xmath3 , and derive an expression for it .
we apply our results to the study of network scenarios of practical interest , such as sensor sensor deployments with coverage holes , communication in presence of a fading channel , massively dense networks @xcite , and networks using contention - based channel access techniques @xcite .
the rest of the paper is organized as follows .
section [ sec : related ] reviews previous work , while section [ sec : system ] describes the system model under study . in section [ sec : preliminaries ] , we first provide some useful definitions and introduce our performance metric , then we recall previous results on which we build our analysis . in section [ sec : results - vandermonde ] , we derive asymptotic results concerning the moments and the lsd of @xmath3 .
such results are applied to different practical scenarios in section [ sec : applications ] .
finally , section [ sec : conclusions ] concludes the paper .
in the context of sensor networks , several works @xcite have studied the field reconstruction at the sink node in presence of spatial and temporal correlation among sensor measurements . in particular , in @xcite
the observed field is a discrete vector of target positions and sensor observations are dependent . by modeling the sensor network as a channel encoder and exploiting some concepts from coding theory , the network capacity , defined as the maximum value of the ratio of the target positions to the number of sensors , is studied as a function of the noise , the sensing function and the sensor connectivity level .
the paper by dong and tong @xcite considers a dense sensor network where a mac protocol is responsible to collect samples from network nodes .
the work analyzes the impact of deterministic and random data collection strategies on the quality of field reconstruction . as a performance measure ,
the maximum of the reconstruction square error over the sensed field is employed , as opposed to our work where the mean square error is considered .
also , in @xcite the field is a gaussian random process and the sink always receives a sufficiently large number of samples so as to reconstruct the field with the required accuracy .
the problem of reconstructing a bandlimited field from a set of irregular samples at unknown locations , instead , has been addressed in @xcite .
there , the field is oversampled by irregularly spaced sensors ; sensor positions are unknown but always equal to an integer multiple of the sampling interval .
different solution methods are proposed , and the conditions for which there exist multiple solutions or a unique solution are discussed . differently from @xcite , we assume that the sink can either acquire or estimate the sensor locations and that the coordinates of the sampling points are randomly located over a finite @xmath0-dimensional domain . as for previous results on vandermonde matrices , in @xcite ryan and debbah considered a vandermonde matrix @xmath2 with @xmath10 and complex exponential entries , whose phases are i.i.d . with continuous distribution .
under such hypothesis , they obtained the important results that , given the phases distribution , the moments of @xmath3 can be derived once the moments for the case with uniformly distributed phases are known .
also , a method for computing the moments of sums and products of vandermonde matrices , for the non - folded case ( i.e. , @xmath10 ) , has recently appeared in @xcite ; further insights on the extremal eigenvalues behavior , still for the case of non - folded vandermonde matrices , can be found in @xcite .
moreover , in @xcite it has been shown that the lsd of @xmath3 converges to the marenko - pastur distribution @xcite when @xmath2 is @xmath0-fold vandermonde with uniformly distributed phases and @xmath11 . note that , with respect to previous studies on vandermonde matrices with entries that are randomly distributed on the complex unit circle , in this work we obtain results on the lsd of @xmath3 where the entries of @xmath2 have phases drawn from a _ generic continuous distribution_. by relying on the results in @xcite , we show that such an lsd can be related to that of @xmath3 when the phases of @xmath2 are _ uniformly _ distributed on the complex unit circle .
we also provide some numerical results that show the validity of our analysis . to our knowledge
, these results have not been previously derived .
we then apply them to the study of several practical scenarios in the context of sensor networks , although our findings can be useful for the study of other aspects of communications as well @xcite .
we consider a network composed of @xmath12 wireless sensors , which measure the value of a spatially - finite physical field defined over @xmath0 dimensions , ( @xmath13 ) .
we denote by @xmath14 the hypercube over which the sampling points fall , and we assume that the sampling points are i.i.d . randomly distributed variables , whose value is known to the sink node . note that this is a fair assumption , as one can think of sensor nodes randomly deployed over the geographical region that has to be monitored , or , even in the case where the network topology is intended to have a regular structure , the actual node deployment may turn out to be random due to obstacles or terrain asperities . in addition , now and then the sensors may enter a low - operational mode ( hence become inactive ) in order to save energy , and they may be loosely synchronized .
all the above conditions yield a set of randomly distributed samples of the field under observation , in both the time and the space domain @xcite . by truncating its fourier series expansion , a physical field defined over @xmath0 dimensions and with finite energy
can be approximated in the region @xmath15 as @xcite @xmath16 where @xmath17 is the approximate one - sided bandwidth ( per dimension ) of the field , @xmath18\tran$ ] is a vector of integers , with @xmath19 , @xmath20 .
the coefficient @xmath21 is a normalization factor and the function @xmath22 maps uniquely the vector @xmath23 over @xmath24 $ ] .
@xmath25 denotes the @xmath26-th entry of the vector @xmath27 of size @xmath28 , which represents the approximated field spectrum , while the real vectors @xmath29 , @xmath30 represent the coordinates of the @xmath0-dimensional sampling points . in this work , we assume that @xmath29 , @xmath30 , are i.i.d .
random vectors having a generic continuous distribution @xmath31 , @xmath32 . in the specific case
where @xmath29 are i.i.d with i.i.d .
entries @xmath33 , uniformly distributed in @xmath34 , we denote the distribution of @xmath29 by @xmath35 .
the coordinates of the @xmath0-dimensional sampling points , however , are known to the sink node , because _ ( i ) _ either sensors are located at pre - defined positions or their position can be estimated through a localization technique @xcite , and _
( ii ) _ the sampling time is either periodic or included in the information sent to the sink . now , let @xmath36^{\rm t}$ ] be the values of the samples at @xmath37 $ ] , respectively . following @xcite , we can write the vector @xmath38 as a function of the field spectrum : @xmath39 where @xmath2 is the @xmath40 @xmath0-fold vandermonde matrix with entries @xmath41 randomly distributed on the complex circle of radius @xmath42 , and @xmath43 is the ratio of the rows to the columns of @xmath2 , i.e. , @xmath44 in general
, the entries of @xmath27 can be correlated with covariance matrix @xmath45 $ ] . however , in the following , we restrict our attention to the class of fields characterized by @xmath45= \sigma^2_a{{\bf i}}$ ] . in the case
where the sensor measurements , @xmath46\tran$ ] , are noisy , then the relation between the sensor samples and the approximated field spectrum can be written as : @xmath47 where @xmath48 is a @xmath12-size , zero - mean random vector representing the noise . here
, we assume a white noise , i.e. , with covariance matrix @xmath49 = \sigma^2_n{{\bf i}}_{m}$ ] .
note that the additive white noise affecting the sensor measurements may be due to quantization , round - off errors or quality of the sensing device .
in this section , we report some definitions and previous results that are useful for our study .
let us consider an @xmath50 non - negative definite random matrix @xmath51 , whose eigenvalues are denoted by @xmath52 .
the average empirical cumulative distribution of the eigenvalues of @xmath51 is defined as @xmath53 $ ] , where the superscript @xmath54 indicates that we refer to a system with size @xmath17 and @xmath55 is the indicator function . if @xmath56 converges as @xmath57 , then @xmath58 .
the corresponding limiting probability density function , or limiting spectral distribution ( lsd ) , is denoted by @xmath59 .
the @xmath4-transform of @xmath51 is given by : @xmath60 = { \mathbb{e}}\left[\frac{1}{n}\sum_{i=1}^n\frac{1}{\gamma\lambda_{{{\bf a}},i } + 1}\right]\ ] ] where @xmath61 is the normalized matrix trace operator and @xmath62 is a non - negative real number .
if @xmath63 converges as @xmath57 , then the corresponding limit is @xmath64 $ ] @xcite , where @xmath65 is the generic asymptotic eigenvalue of @xmath51 , whose distribution is @xmath66 , and the average is computed with respect to @xmath65 @xcite . next ,
consider the matrix @xmath2 as defined in ( [ def : v_multifold ] ) and that the lmmse filter is used for field reconstruction . then
, the estimate of the unknown vector @xmath27 in ( [ eq : p ] ) , given @xmath67 and @xmath2 , is obtained by computing @xmath68 { \mathbb{e}}[{{\bf p}}{{\bf p}}\herm]^{-1 } { { \bf p}}$ ] . through easy computations and using the sherman - morrison - woodbury identity , we can obtain the mse as @xmath69 = \eta^{(n)}_{{{\bf v}}{{\bf v}}\herm}\left(\frac{\gamma}{\beta_{n , m}}\right ) \label{eq : mse_n}\ ] ] where @xmath70 denotes the signal - to - noise ratio on the sensor measurements , and we employed the definition of the @xmath4-transform given in ( [ eq : eta - t ] ) .
next , we approximate the mse of the finite size system in ( [ eq : p ] ) through an asymptotic model , which assumes the size of @xmath2 to grow to infinity while the ratio of its number of rows to its number of columns tends to a finite limit , @xmath71 , greater than zero , i.e. , we assume @xmath72 indeed , in our recent works @xcite it was shown that this asymptotic model provides a tight approximation of the mse of the finite size system , already for small values of @xmath17 and @xmath12 . under these conditions
, we therefore define the asymptotic expression of the mse as @xcite : @xmath73 if the limit exists .
vandermonde matrices have been studied in a number of recent works @xcite .
specifically , @xcite considered the case where the vectors @xmath29 are i.i.d .
, for @xmath30 , and their entries , @xmath74 are i.i.d .
random variables with uniform distribution in @xmath34 .
the work there studied the eigenvalue distribution of @xmath3 for both finite and infinite ( i.e. , @xmath75 ) matrix size .
although an explicit expression of such lsd is still unknown , @xcite provided an algorithm to compute its moments of any order in closed form . indeed , as @xmath76 with @xmath77 having a finite limit @xmath78 , in @xcite it was shown that the @xmath79-th moment of the generic asymptotic eigenvalue of @xmath3 , denoted by @xmath80 , is given by @xmath81 where @xmath82 represents the distribution of @xmath80 .
moreover , @xmath83 is the set of partitions of the set @xmath84 in @xmath85 subsets , and @xmath86 $ ] , @xmath87 is a rational number that can be analytically computed from @xmath88 following the procedure described in @xcite . the subscript @xmath89 in @xmath90 and
@xmath82 indicates that a uniform distribution of the entries of @xmath29 is considered in the matrix @xmath2 . in
@xcite it was also shown that when @xmath91 , with @xmath77 having a finite limit @xmath78 , the eigenvalue distribution @xmath82 converges to the marcenko - pastur law @xcite .
a similar result @xcite also applies when the vectors @xmath29 ( @xmath30 ) are independent but not i.i.d .
, with equally spaced averages .
more recently , ryan and debbah in @xcite considered @xmath10 and the case where the random variables @xmath92 , @xmath30 , are i.i.d . with continuous distribution @xmath93 , @xmath94 . under such hypothesis
, it was shown that the asymptotic moments of @xmath3 can be written as @xmath95 where the terms @xmath96 depend on the phase distribution @xmath93 and are given by @xmath97 for @xmath98 .
the subscript @xmath99 in @xmath100 indicates that in the matrix @xmath2 the random variables @xmath92 have a generic continuous distribution @xmath93 .
note that for the uniform distribution we have @xmath101 , for all @xmath85 .
the important result in ( [ eq : moments_1d ] ) states that , given @xmath71 , if the moments of @xmath3 are known for uniformly distributed phases , they can be readily obtained for any continuous phase distribution @xmath93 .
in this work , we extend the above results by considering a sampling system defined over @xmath102 dimensions with nonuniform sample distribution , where samples may be irregularly spaced in the time and spatial domains , as it occurs in wireless sensor networks .
being our goal the estimation of the quality of the reconstructed field , we aim at deriving the asymptotic mse ( i.e. , @xmath103 ) .
we start by considering a generic continuous distribution , @xmath104 , @xmath32 of the samples measured by the sensors over the @xmath0-dimensional domain .
we state the theorem below , which gives the asymptotic expression of the generic moment of @xmath3 , for @xmath105 .
[ th:1 ] let @xmath2 a @xmath0-fold @xmath106 vandermonde matrix with entries given by ( [ def : v_multifold ] ) where the vectors @xmath29 , @xmath30 , are i.i.d . and have continuous distribution @xmath104 .
then , for @xmath76 , with @xmath77 having a finite limit @xmath78 , the @xmath79-th moment of @xmath3 is given by @xmath107 where @xmath108 and the terms @xmath109 are defined as in @xcite . the proof is given in appendix [ app : th1 ] .
using theorem [ th:1 ] and the definition of @xmath96 , it it possible to show the theorem below , which provides the lsd of @xmath3 .
[ th:2 ] let * @xmath2 be a @xmath0-fold @xmath106 vandermonde matrix with entries given by ( [ def : v_multifold ] ) where the vectors @xmath29 , @xmath30 , are i.i.d . and have continuous distribution @xmath104 , @xmath32 * @xmath110 be the set where @xmath104 is strictly positive , i.e. , @xmath111 * the cumulative density function @xmath112 defined denotes the measure of the set @xmath113 for @xmath114 and let @xmath115 be its corresponding probability density function .
then , the lsd of @xmath3 , for @xmath76 with @xmath77 having a finite limit @xmath78 , is given by @xmath116 the proof can be found in appendix [ app : th2 ] . from theorem [ th:2 ] , the corollary below follows .
[ cor - scal ] consider @xmath31 such that @xmath117 @xmath118 .
then , let us denote by @xmath119 a scaled version of this function , so that @xmath120 where @xmath121 .
it can be shown that @xmath122 where @xmath123 .
the proof can be found in appendix [ app : cor - scal ] . as an example of the result given in corollary [ cor - scal ] , consider that a unidimensional ( @xmath10 ) sensor network monitors the segment @xmath124 $ ] . due to terrain irregularities and obstacles ,
nodes are deployed with uniform distribution only in the range @xmath125 $ ] ( with @xmath126 ) .
we therefore have @xmath127 for @xmath128 and 0 elsewhere .
moreover , @xmath129 .
the expression of @xmath130 is given by ( [ eq : f - scal ] ) , by replacing @xmath10 and the subscript @xmath99 with the subscript @xmath89 .
this result is well supported by simulations as shown in figures [ fig : beta08_x08 ] and [ fig : beta02_x05 ] . in the plots ,
we compare the asymptotic empirical spectral distribution ( aesd ) @xmath131 and @xmath132 instead of the lsds @xmath133 and @xmath134 since an analytic expression of @xmath134 is still unknown .
however , in @xcite it is shown that , already for small values of @xmath17 , the aesd @xmath132 appears to rapidly converge to a limiting distribution .
figure [ fig : beta08_x08 ] refers to the case @xmath135 and @xmath136 .
the solid and dashed lines represent , respectively , the functions @xmath137 and @xmath138 , for @xmath139 . note that the probability mass of @xmath140 at @xmath141 is not shown for simplicity .
similarly , figure [ fig : beta02_x05 ] shows the case @xmath142 and @xmath143 . as evident from these plots , the match between the two functions is excellent for any parameter setting , thus supporting our findings .
since we are interested in evaluating the mse , taking into account the result in ( [ eq : mseinf ] ) , we now apply the definition of the @xmath4-transform to ( [ eq : th2 ] ) .
the corollary below immediately follows .
[ cor2 ] the @xmath4-transform of @xmath3 is given by @xmath144 hence , the asymptotic mse on the reconstructed field , defined in ( [ eq : mseinf ] ) , is given by @xmath145 the proof can be found in appendix [ app : cor2 ] . in ( [ eq : eta ] ) , in order to avoid a heavy notation we referred to @xmath146 as @xmath147 when the phases of the entries of @xmath2 follow a generic random continuous distribution , while @xmath148 refers to the case where the phases are uniformly distributed .
since @xmath149 and @xmath150 , the integral in the right hand side of ( [ eq : eta ] ) is positive , then @xmath151 .
it follows that the mse is lower - bounded by the measure of the total area where the probability of finding a sensor is zero .
this clearly suggests that , in order to obtain a good quality of the field reconstructed at the sink node , this area must be a small fraction of the region under observation .
next , we observe that , in the case of massively dense networks where the number of sampling sensors is much larger than the number of harmonics considered in the approximated field , i.e. , @xmath152 , an interesting result holds : [ cor1 ] let @xmath110 be the set where @xmath31 is strictly positive ; then @xmath153 the proof can be found in appendix [ app : cor1 ] .
thus , as evident from corollary [ cor1 ] , for the limit of @xmath154 , the lsd of @xmath3 is the density of the density of the phase distribution @xmath104 .
furthermore , for massively dense networks , we have : [ cor3 ] let @xmath110 be the set where @xmath31 is strictly positive ; then @xmath155 the proof can be found in appendix [ app : cor3 ] .
the result in ( [ eq : cor3.2 ] ) shows that even for massively dense networks @xmath156 is the minimum achievable @xmath157 , when an area @xmath110 can not covered by sensors .
here , we provide examples of how our results can be used in wireless sensor networks to investigate the impact of a random distribution of the coordinates of the sampling points on the quality of the reconstructed field . in particular , we first consider a wireless channel affected by fading , and then the effects of contention - based channel access .
we consider a wireless sensor network whose nodes are uniformly distributed over a geographical region . without loss of generality , we assume a square region of unitary side ( @xmath158 , @xmath159 ^ 2 $ ] ) , where the sink is located at the center and has coordinates @xmath160 .
through direct transmissions , the sensors periodically send messages to the sink , including their measurements . at every sample period
, a sensor message is correctly received at the sink if its signal - to - noise ratio ( snr ) exceeds a threshold @xmath161 .
the communication channel is assumed to be affected by slow fading and to be stationary over the message duration .
let @xmath0 be the distance between a generic sensor and the sink .
then , the signal to noise ratio at the receiver is given by @xmath162 where @xmath163 is a circularly symmetric gaussian complex random variable representing the channel gain , and @xmath164 is the signal to noise ratio in the absence of fading and when the sensor - sink distance is @xmath10 .
the probability that a message is correctly received at the sink is given by @xmath165 with @xmath166 and @xmath167 being the cumulative density function of @xmath168 .
the probability density @xmath169 corresponding to sensors at distance @xmath170 , @xmath171 from the sink and successfully sending a message is then given by @xmath172 where @xmath173 , is the density representing the sensor deployment ( recall that nodes are assumed to be uniformly distributed in the region hence their density is constant and equal to 1 ) . using ( [ eq : snr ] ) , we obtain : @xmath174 where @xmath175 in order to compute ( [ eq : eta ] ) , we need the function @xmath176 , i.e. , the density of @xmath169 . note that @xmath169 is circularly symmetric with respect to @xmath160 .
let @xmath177 be the value of density of the sampling points at distance @xmath178 from the sink .
then , from ( [ eq : f_xi ] ) we obtain @xmath179 , thus the network area where the density is lower than @xmath177 is given by @xmath180 for @xmath181 , i.e. , @xmath182 .
for @xmath183 , it is possible to show that @xmath184 in conclusion , @xmath185 and @xmath186 since @xmath187 , then the asymptotic mse can be obtained by computing @xmath188 for @xmath189 db , i.e. , for different values of the snr threshold @xmath161 . ]
figure [ fig : g ] shows the density @xmath115 for @xmath189 db .
note that @xmath166 , thus for a fixed @xmath164 ( i.e. , the signal to noise ratio at distance @xmath190 in the absence of fading ) the parameter @xmath191 is proportional to the snr threshold @xmath161 .
in particular , as @xmath161 decreases , the probability that a message successfully reaches the destination increases and , thus , the spatial distribution of correctly received samples , @xmath192 , tends to the uniform distribution @xmath193 . as a consequence
, the density of @xmath169 , i.e. , @xmath115 , for @xmath194 and 5 db is concentrated close to @xmath195 . however , for high values of @xmath161 , messages originated from sensor nodes located far from the sink are successfully received with low probability .
thus , @xmath115 shows a significant probability mass around @xmath196 . ) and in presence ( @xmath104 ) of fading , as the signal to noise ratio on the sensor measurements varies . ]
figure [ fig : mse ] shows the effect of the fading channel on the mse of the reconstructed field ( dashed lines ) , and compares the obtained results with the mse obtained in absence of fading ( solid lines ) .
the plot considers different values of @xmath71 , namely , @xmath197 , and @xmath198 db . the mse is plotted versus the signal to noise ratio on the sensor measurements , @xmath62 .
the curves have been obtained by numerically computing ( [ eq : eta_xi_2 ] ) , where @xmath115 is given by ( [ eq : g_xi ] ) and @xmath199 is replaced by @xmath200 , with @xmath201 .
recall that the analytic expression of the lsd @xmath82 is unknown , hence in the numerical results we considered the aesd @xmath202 instead .
we observe that for low values @xmath71 , in spite of the presence of fading , the sink node still receives a large number of samples from the sensors , hence the degradation of the mse shown in figure [ fig : mse ] is negligible . on the contrary , for @xmath203 ( i.e. , for a larger value of the ratio of the number of harmonics composing the approximated field to the number of sensors )
, the reconstruction performance degrades significantly and this is particularly evident in presence of high values of @xmath62 . in presence of fading , with @xmath198 db and @xmath187 .
the curves obtained for different values of @xmath71 are compared with the density @xmath204 . ] in the case of massively dense networks , the lsd of @xmath3 is given by ( [ eq : cor1 ] ) and from ( [ eq : cor3.2 ] ) we know that the mse tends to 0 as @xmath205 . this result
is confirmed by the plot in figure [ fig : dense ] , which shows the aesd @xmath206 , for @xmath187 , @xmath198 db , and @xmath201 .
the behavior of such a function is compared with the density @xmath204 as @xmath71 varies .
we note that , as @xmath71 decreases , the matching between @xmath206 and @xmath204 improves , and the latter represents an excellent approximation already for @xmath207 , as predicted by the result in ( [ cor1 ] ) . in environmental monitoring applications , it is often desirable to vary the resolution level with which the field measurements are taken over the region under observation , depending on the field variations and the interest level of the different locations @xcite .
it follows that the number of samples generated by the sensors network ( i.e. , the offered traffic load ) varies in the spatial domain .
to represent such a scenario , we consider a wireless sensor network whose nodes are uniformly deployed over a square region .
we also identify @xmath208 areas , @xmath209 @xmath210 , each corresponding to a different value of the offered traffic load .
as often assumed in the literature ( see e.g. , @xcite ) and widely applied in the practice , the network is divided into clusters and a hierarchy of clusters is created .
more specifically , at the first hierarchical layer , layer 1 , the sensors are grouped into clusters , each of which is controlled by a cluster - head .
the cluster - head is in charge of handling all traffic packets it receives from the nodes . at a given layer @xmath211 of the hierarchy
, the cluster - heads are grouped into clusters on their turn and forward the traffic to their parent cluster - head . at the highest layer , layer @xmath212 , we have only one cluster whose cluster - head coincides with the sink node . without loss of generality
, we assume that the cluster at the @xmath212-th layer is composed of @xmath208 cluster - heads , each handling the traffic generated within one of the @xmath208 areas defined above . as for the medium access control ( mac ) layer ,
we consider that the nodes implement the ieee 802.15.4 standard specifications for wireless sensor networks @xcite .
in particular , all nodes within a cluster are in radio visibility of each other and use the slotted carrier - sense multiple - access / collision avoidance ( csma / ca ) technique @xcite .
this is a contention - based scheme and transmissions may fail if two or more sensors access the channel at the same time .
inter - cluster interference is instead avoided by assigning different frequency channels to neighboring clusters .
we consider that packets , whose transmission fails , are discarded . in order to derive the probability that a packet transmission fails within a cluster due to collision , we use the markov chain model presented in @xcite .
we denote by @xmath213 the average number of sensors belonging to the generic cluster at the @xmath214-th layer of the hierarchical architecture , in area @xmath209
( @xmath210 and @xmath215 ) .
similarly , we define @xmath216 as the average traffic load per node , again within the generic cluster at the @xmath214-th layer , in area @xmath209 . then , we set the size of the packet payload to 32 bytes , and the value of the other parameters as in @xcite . under this setting , we compute the value of the collision probability within the generic cluster at layer @xmath214 , in area @xmath209 , as a function of @xmath213 and @xmath216 , i.e. , @xmath217 @xcite .
furthermore , we observe that at the generic layer @xmath214 , with @xmath218 , a node , which acts as cluster - head at layer @xmath219 in area @xmath209 , will have a traffic load equal to @xmath220 $ ] .
it follows that the probability that a packet is successfully delivered to the corresponding @xmath214-layer cluster - head within area @xmath209 ( @xmath210 ) can be obtained as @xmath221 .
then , the probability that a measurement generated by a sensor located in @xmath209 ( @xmath210 ) is successfully delivered to the sink is given by : @xmath222 next , denoting by @xmath223 the measure of @xmath209 , we define @xmath224 as the normalized probability that a message is successfully delivered to the sink . then , the spatial density of the sensors successfully sending their message is as follows : @xmath225 the density of @xmath169 is therefore given by @xmath226 and the asymptotic mse is given by @xmath227 ) and the case where all measurements successfully reach the sink ( @xmath35 ) .
the mse is shown as a function of @xmath71 and for different values of signal - to - noise ratio ( @xmath228 , @xmath229 , @xmath230 , @xmath231 , @xmath232 , @xmath233 ) . ] ) and the case where all measurement transmissions are successful ( @xmath35 ) .
the mse is shown as @xmath71 varies and for different values of signal - to - noise ratio ( @xmath228 , @xmath229 , @xmath234 , @xmath235 , @xmath236 , @xmath237 ) . ] figures [ fig : lambda2 ] and [ fig : lambda1 ] show the impact of collisions due to the contention - based channel access , on the quality of the reconstructed field .
in particular , they compare the mse of the reconstructed field when collisions are taken into account ( @xmath104 ) with the one obtained in the idealistic case where all messages ( measurements ) sent by the sensors successfully reach the sink ( @xmath35 ) .
the results refer to a square region of unitary side , where there are four areas of equal size ( @xmath238 , @xmath239 ) but corresponding to different resolution levels in the measurements collection ( i.e. , they are characterized by different traffic loads ) ; the number of hierarchical levels is set to @xmath228 .
we set @xmath230 , @xmath231 , @xmath232 , @xmath233 in figure [ fig : lambda2 ] , and a higher traffic load in figure [ fig : lambda1 ] , i.e. , @xmath234 , @xmath235 , @xmath236 , @xmath237 . looking at the plots , we observe that both @xmath71 and @xmath62 have a significant impact of the obtained mse , with the mse increasing as @xmath71 grows and smaller values of @xmath62 are considered . most interestingly , by comparing the two figures
, we can see that as the traffic load , hence the collision probability , increases , the performance derived taking into account the contention - based channel access significantly differs from the idealistic one .
furthermore , the latter effect is particularly evident as @xmath62 increases , since the higher the signal - to - noise ratio , the more valuable the samples sent by the sensors toward the sink .
we studied the performance of a wireless network whose nodes sense a multi - dimensional field and transfer their measurements to a sink node .
as often happens in practical cases , we assumed the sensors to be randomly deployed over ( the whole or only a portion of ) the region of interest , and that their measurements may be lost due to fading or transmission collisions over the wireless channel .
we modeled the sampling system through a multi - folded vandermonde matrix @xmath2 and , by using asymptotic analysis , we approximated the mse of the field , which the sink node reconstructs from the received sensor measurements with the @xmath4-transform of @xmath3 .
our results clearly indicate that the percentage of region where sensors can not be deployed must be extremely small if an accurate field estimation has to be obtained .
also , the effect of losses due to fading or transmission collisions can be greatly mitigated provided that a suitable value for the ratio between the number of harmonics approximating the field bandwidth and the number of sensors is selected .
the @xmath79-th moment of the asymptotic eigenvalue distribution of @xmath3 can be expressed as @xcite @xmath240\ ] ] where @xmath241 is the normalized matrix trace operator .
the matrix power can be expanded as a multiple sum over the entries of @xmath2 : @xmath242\ ] ] where @xmath243 , @xmath244 are integer indices and @xmath245 , @xmath246\tran$ ] , @xmath247 are the indices identifying the rows of @xmath2 . since , @xmath248 for @xmath249 and the elements of @xmath250 are i.i.d . , we have that @xmath251 \non\ ] ] where the index @xmath252 is to be considered modulo @xmath79 , i.e. , @xmath253 . as for the sum over the indices
@xmath254 we note that any choice of @xmath255\tran$ ] induces a partition @xmath88 of the set , @xmath256 in @xmath85 subsets @xmath257 , @xmath258 , under the equality relation @xcite . in the following ,
we denote by @xmath83 the set of partitions of @xmath259 in @xmath85 subsets , @xmath258 .
since there are @xmath260 possible vectors @xmath261 inducing a given partition @xmath87 , we can write the @xmath79-th moment as @xmath262 \non & = & \lim_{n , m \rightarrow \infty}\sum_{k=1}^p\beta_{n , m}^{p - k}\sum_{{\boldsymbol{\omega}}\in \omega_{p , k } } \frac{{\mathbb{e}}\left[\phi_{{\boldsymbol{\omega } } } ( { { \bf x}}_1,\ldots,{{\bf x}}_k ) \right]}{n^{d(p - k+1 ) } } \label{eq : moment2}\end{aligned}\ ] ] where @xmath263 , @xmath264 , and @xmath265 is the index of the subset of @xmath259 containing @xmath252 . recall that @xmath266 and that @xmath77
. moreover , since the vectors @xmath250 are i.i.d . , we removed the dependence on the subscript @xmath267 .
following the same steps as in ( * ? ? ?
* appendix h ) , we compute the limit @xmath268}{n^{d(p - k+1 ) } } & = & \lim_{n \rightarrow \infty}\int_{{{\cal h}}^k}f_x({{\bf x}}_1)\cdots f_x({{\bf x}}_k)\frac{\phi_{{\boldsymbol{\omega}}}({{\bf x}}_1,\ldots,{{\bf x}}_k)}{n^{d(p+1-k ) } } { { \rm\,d}}{{\bf x}}_1 \cdots { { \rm\,d}}{{\bf x}}_k \non & = & \lim_{n \rightarrow \infty}\int_{{{\cal h}}^k}f_x({{\bf x}}_1)\cdots f_x({{\bf x}}_k)\prod_{j=1}^d \frac{f_{{\boldsymbol{\omega}}}(x_{1j},\ldots , x_{kj})}{n^{p+1-k } } { { \rm\,d}}x_{1j}\cdots { { \rm\,d}}x_{kj } \nonumber\end{aligned}\ ] ] we then define @xmath269 $ ] where @xmath270 $ ] , for @xmath271 and we integrate first with respect to the variables @xmath272 obtaining @xmath268}{n^{d(p - k+1 ) } } & = & \lim_{n \rightarrow \infty}\int_{[-\frac{1}{2},\frac{1}{2}]^{(d-1)k } } g_{{\boldsymbol{\omega}}}({{\bf y}}_1,\ldots,{{\bf y}}_k)\prod_{j=2}^d \frac{f_{{\boldsymbol{\omega}}}(x_{1j},\ldots , x_{kj})}{n^{p - k+1 } } { { \rm\,d}}x_{1j}\cdots { { \rm\,d}}x_{kj } \nonumber\end{aligned}\ ] ] where @xmath273^{k}}\frac{f_{{\boldsymbol{\omega}}}(x_{11},\ldots , x_{k1})}{n^{p - k+1}}f_x([x_{11 } , { { \bf y}}_1])\cdots f_x([x_{k1 } , { { \bf y}}_k]){{\rm\,d}}x_{11}\cdots { { \rm\,d}}x_{k1 } \label{eq : g_omega}\ ] ] in ( * ? ? ?
* appendix h ) it was shown that , because of the properties of @xmath274 , @xmath275 where @xmath276 for any @xmath277 .
this means that the integral in ( [ eq : g_omega ] ) can be limited to the @xmath278 on the diagonal where @xmath279 .
therefore @xmath280^{k}}\frac{f_{{\boldsymbol{\omega}}}(x_{11},\ldots , x_{k1})}{n^{p - k+1 } } f_x([x_{k1 } , { { \bf y}}_1])\cdots f_x([x_{k1 } , { { \bf y}}_k]){{\rm\,d}}x_{11}\cdots { { \rm\,d}}x_{k1 } \non & = & \int_{[-\frac{1}{2},\frac{1}{2 } ] } \prod_{h=1}^k f_x([x_{k1 } , { { \bf y}}_h ] ) \lim_{n \rightarrow \infty}\left(\int_{[-\frac{1}{2},\frac{1}{2}]^{k-1}}\frac{f_{{\boldsymbol{\omega}}}(x_{11},\ldots , x_{k1})}{n^{p - k+1}}\prod_{h=1}^{k-1}{{\rm\,d}}x_{h1}\right ) { { \rm\,d}}x_{k1 } \non & = & v({\boldsymbol{\omega } } ) \int_{[-\frac{1}{2},\frac{1}{2 } ] } \prod_{h=1}^k f_x([x_{k1 } , { { \bf y}}_h ] ) { { \rm\,d}}x_{k1 } \end{aligned}\ ] ] note that the limit @xmath281^{k-1}}\frac{f_{{\boldsymbol{\omega}}}(x_{11},\ldots , x_{k1})}{n^{p - k+1}}\prod_{h=1}^{k-1}{{\rm\,d}}x_{h1}\ ] ] does not depend on @xmath282 and the coefficient @xmath283 $ ] is described in @xcite .
next , iterating this procedure by integrating over the variables , @xmath284 , @xmath285 we finally get @xmath286}{n^{d(p - k+1 ) } } & = & v({\boldsymbol{\omega}})^d \int_{[-\frac{1}{2},\frac{1}{2}]^d } f_x(x_{k1 } , \ldots , x_{kd})^k { { \rm\,d}}x_{k1 } \cdots { { \rm\,d}}x_{kd } \non & = & v({\boldsymbol{\omega}})^d \int_{{{\cal h } } } f_x({{\bf x}}_k)^k { { \rm\,d}}{{\bf x}}_k \non & = & v({\boldsymbol{\omega}})^d i_k\end{aligned}\ ] ] where we defined @xmath287 .
it follows that @xmath288 which proves the theorem .
note that when the entries of @xmath289\tran$ ] are independent with continuous distribution @xmath290 such that @xmath291 , we have @xmath292 with @xmath293 } f_{x_j}(x)^k{{\rm\,d}}x$ ] .
from theorem [ th:1 ] and the definition of @xmath96 , we have that @xmath294 next , we define the set @xmath110 where @xmath104 is strictly positive as @xmath295 note that for @xmath296 the contribution to the integral in ( [ eq : moment_psi ] ) is zero .
thus , @xmath297 where for any @xmath298 , @xmath299 is the @xmath79-th moment of @xmath3 when the phases are uniformly distributed in @xmath15 and the ratio @xmath300 is given by @xmath301 note also that ( [ eq : moments_psi2 ] ) holds for @xmath302 since , by definition , the zero - th moment of any distribution is equal to 1 .
expression ( [ eq : moments_psi2 ] ) allows us to write the moments of @xmath3 for any distribution @xmath104 , given the moments for uniformly distributed phases .
likewise , it is possible to describe the lsd of @xmath3 , for any continuous @xmath104 , in terms of the lsd obtained for uniformly distributed phases . indeed , let us denote the laplace transform of @xmath303 by @xmath304 if it exists .
then , whenever the sum converges @xmath305 since @xmath306 for any distribution , using ( [ eq : moments_psi2 ] ) we obtain : @xmath307 where @xmath308 is the measure of @xmath110 and @xmath309 is the laplace transform of @xmath82 . by using the properties of the laplace
transform and by taking its inverse , we finally get @xmath310 we can rewrite the second term of ( [ eq : distribution_psi ] ) by defining the cumulative density function @xmath311 for @xmath114 . by using the corresponding probability density function , @xmath115 , and lebesgue integration , we can rewrite ( [ eq : distribution_psi ] ) as in ( [ eq : th2 ] ) .
from the result in ( [ eq : th2 ] ) and from the assumption @xmath117 @xmath118 ( i.e. , @xmath187 ) , we have @xmath312 then , from the definition of @xmath119 given in ( [ eq : fxi_scaled ] ) it follows that @xmath313 and , by consequence , @xmath314 .
therefore , from ( [ eq : f - scal ] ) we have : @xmath315
from the expression of the moments given in theorem [ th:1 ] and the results in @xcite , it is easy to show that for uniformly distributed phases , we have : @xmath316 for any @xmath317 .
it immediately follows that @xmath318 where @xmath319 is the dirac s delta function . by applying this result to ( [ eq : th2 ] )
, we get @xmath320
by using the definition of the @xmath4-transform and the result in ( [ eq : th2 ] ) , we obtain : @xmath321 \non & = & \int_0^{\infty } \frac{1}{\gamma z+1}f_{\lambda , x}(d,\beta , z ) { { \rm\,d}}z \non & = & \int_0^{\infty } \frac{1-|{{\cal a}}|}{\gamma z+1}\delta(z ) + |{{\cal a}}| \int_0^{\infty}\frac{g_x(y)}{y } \int_0^{\infty } \frac{1}{\gamma z+1}f_{\lambda , u}\left(d,\frac{\beta}{y},\frac{z}{y}\right){{\rm\,d}}z { { \rm\,d}}y \non & = & 1-|{{\cal a}}|+ |{{\cal a}}| \int_0^{\infty } g_x(y ) \int_0^{\infty } \frac{1}{\gamma y z+1}f_{\lambda , u}\left(d,\frac{\beta}{y } , z\right){{\rm\,d}}z { { \rm\,d}}y \non & = & 1-|{{\cal a}}|+ |{{\cal a}}| \int_0^{\infty } g_x(y ) \eta_u\left(d,\frac{\beta}{y } , \gamma y\right ) { { \rm\,d}}y .\end{aligned}\ ] ] then , by considering that @xmath322 , the expression of the asymptotic mse immediately follows .
in appendix [ cor1 ] we have shown that @xmath323 . from the definition of the @xmath4-transform
, it follows that @xmath324 and @xmath325 .
thus , from ( [ eq : eta ] ) we have : @xmath326 where we defined @xmath327 . as a consequence , @xmath328 m. perillo , z. ignjatovic , and w. heinzelman , `` an energy conservation method for wireless sensor networks employing a blue noise spatial sampling technique , '' _ international symposium on information processing in sensor networks ( ipsn 2004 ) _ , apr.2004
. d.s .
early and d.g .
long , `` image reconstruction and enhanced resolution imaging from irregular samples , '' _ ieee transactions on geoscience and remote sensing , _
vol.39 , no.2 , pp.291302 , feb.2001 .
a. nordio , c .- f .
chiasserini , and e. viterbo , `` performance of linear field reconstruction techniques with noise and uncertain sensor locations , '' _ ieee transactions on signal processing , _ vol.56 , no.8 , pp.35353547 , aug.2008 .
a. nordio , c .- f .
chiasserini , and e. viterbo , `` reconstruction of multidimensional signals from irregular noisy samples , '' _ ieee transactions on signal processing , _
vol.56 , no.9 , pp.42744285 , sept.2008 .
802.15.4 : ieee standard for information technology - telecommunications and information exchange between systems - local and metropolitan area networks - specific requirements part 15.4 : wireless medium access control
( mac ) and physical layer ( phy ) specifications for low - rate wireless personal area networks ( wpans ) , 2009 .
r. cristescu and m. vetterli , `` on the optimal density for real - time data gathering of spatio - temporal processes in sensor networks , '' _ international symposium on information processing in sensor networks ( ipsn 05 ) , _ los angeles , ca , apr.2005 .
y. sung , l. tong , and h. v. poor , `` sensor activation and scheduling for field detection in large sensor arrays '' , _ international symposium on information processing in sensor networks ( ipsn 05 ) , _ los angeles , ca , apr.2005 .
y. rachlin , r. negi , and p. khosla , `` sensing capacity for discrete sensor network applications , '' _ international symposium on information processing in sensor networks ( ipsn05 ) , _ los angeles , ca , apr.2005 .
m. dong , l. tong , and b. m. sadler , `` impact of data retrieval pattern on homogeneous signal field reconstruction in dense sensor networks , '' _ ieee transactions on signal processing _ ,
vol.54 , no.11 , pp.43524364 , nov.2006 .
p. marziliano and m. vetterli , `` reconstruction of irregularly sampled discrete - time bandlimited signals with unknown sampling locations , '' _ ieee transactions on signal processing , _
vol.48 , no.12 , dec.2000 , pp.34623471 .
d. moore , j. leonard , d. rus , and s. teller , `` robust distributed network localization with noisy range measurements , '' _ acm conference on embedded networked sensor systems ( sensys ) _ , baltimore , md , pp.5061 , nov.2004 .
c. y. jung , h. y. hwang , d. k. sung , and g. u. hwang , `` enhanced markov chain model and throughput analysis of the slotted csma / ca for ieee 802.15.4 under unsaturated traffic conditions , '' _ ieee transactions on vehicular technology _ , vol.58 , no.1 , pp.473478 , jan.2009 . | environmental monitoring is often performed through a wireless sensor network , whose nodes are randomly deployed over the geographical region of interest .
sensors sample a physical phenomenon ( the so - called field ) and send their measurements to a _ sink _
node , which is in charge of reconstructing the field from such irregular samples . in this work ,
we focus on scenarios of practical interest where the sensor deployment is unfeasible in certain areas of the geographical region , e.g. , due to terrain asperities , and the delivery of sensor measurements to the sink may fail due to fading or to transmission collisions among sensors simultaneously accessing the wireless medium . under these conditions ,
we carry out an asymptotic analysis and evaluate the quality of the estimation of a @xmath0-dimensional field ( @xmath1 ) when the sink uses linear filtering as a reconstruction technique .
specifically , given the matrix representing the sampling system , @xmath2 , we derive both the moments and an expression of the limiting spectral distribution of @xmath3 , as the size of @xmath2 goes to infinity and its aspect ratio has a finite limit bounded away from zero . by using such asymptotic results ,
we approximate the mean square error on the estimated field through the @xmath4-transform of @xmath3 , and derive the sensor network performance under the conditions described above . |
to provide accurate quantum - electrodynamical ( qed ) evaluations of energy levels for two- and three - electron configurations of highly charged ions ( hci ) become now an urgent problem in atomic physics
. this can be explained by the growing number of experimental data and the necessity to use the energy levels for the evaluation of important characteristics of hci , such as e.g. transition probabilities and recombination cross sections . in the past
an approximate relativistic approach based on variational non - relativistic wave functions has been used for evaluating energy levels @xcite .
numerous theoretical results for few - electron ions have been obtained within the framework of fully relativistic many - body perturbation theory ( rmbpt ) and relativistic all - order many - body theory ( ao ) @xcite . however
, rigorous qed results , which allow for a consequent order - by - order improvement of the accuracy of the calculations become more and more desirable .
the approximation of non - interacting electrons is commonly employed as a zeroth - order approximation in qed perturbation theory for hci in the absence of a quasi - degeneracy of levels .
accordingly , within the zeroth - order the energy of the few - electron configuration appears as the sum of the dirac eigenvalues for the occupied one - electron states .
one - electron corrections ( termed here as generalized lamb shift ) include qed radiative corrections , such as the electron self energy ( se ) and vacuum polarization ( vp ) as well as nuclear corrections , i.e. , nuclear size ( ns ) , nuclear recoil ( nr ) and nuclear polarization ( np ) , respectively .
few years ago a status report of one - electron energy corrections has been presented in detail in @xcite . since
then considerable progress concerning the evaluation of higher - order self - energy corrections has been made @xcite .
the dominant two - electron contribution is due to the interelectron interaction .
ab initio qed results for the first - order interelectron interaction in two - electron ions are known from ref .
@xcite ( see also @xcite ) .
the higher - order corrections are much more intricate .
complete qed calculations of the second - order interelectron interaction have been accomplished for the ground state and for non - degenerate low - lying excited states of he - like and li - like ions @xcite .
the other important two - electron corrections are the screened self - energy and vacuum - polarization corrections , which have been evaluated in @xcite for non - degenerate two- and three - electron configurations in hci .
various general bound - state qed approaches have been employed for the derivation of the energy corrections in hci .
the one most commonly used is the adiabatic s - matrix approach , developed by gell - mann and low @xcite , generalized by sucher @xcite and first applied to bound - state qed corrections in ref .
@xcite ( see also @xcite ) .
the other one is the green s function approach , first developed in @xcite and now applied frequently in a more sophisticated form of the two - time green s function method @xcite .
finally , the line profile approach ( lpa ) is utilized for the same purpose @xcite . in our previous studies on the qed theory of interelectron interaction in hci @xcite
this approach has been applied . in this paper
we generalize the line profile approach to the case of quasi - degenerate electronic states in two - electron hci .
this problem arises , when a complete qed treatment including the relativistic coupling scheme is applied to the fine - structure multiplets of systems with intermediate nuclear charge numbers @xmath4 .
some components of the multiplet with equal relativistic quantum numbers turn out to be close enough to each other ( the @xmath0 , @xmath1 levels are the standard example ) . up to now
the qed theory of the quasi - degenerate levels was considered only within the two - time green s function method for the self - energy screening corrections ( without any numerical applications ) @xcite , for vacuum - polarization screening corrections @xcite and within the covariant evolution - operator procedure @xcite for the second - order interelectron interaction .
numerical calculations for two - electron ions with @xmath5 @xcite are so far the only example where bound - state qed has been applied to the evaluation of the energy of quasi - degenerate levels taking into account the interelectron interaction up to second order . in this work we present an extension of the line profile approach , which is suitable for the evaluation of energies of any number of the non - degenerate or quasi - degenerate levels .
the interelectron interaction corrections up to first- and second - order of qed perturbation theory are evaluated for the @xmath6 and @xmath7 levels in a wide range of @xmath4 values .
the problem of the natural line profile in atomic physics was considered first in terms of quantum mechanics by weisskopf and wigner @xcite . in terms of modern qed
it was first formulated for one - electron atoms by low @xcite . in @xcite
the appearance of the lorentz profile in the resonance approximation within the framework of qed was described and nonresonant corrections were estimated .
later the line profile qed theory was modified also for two - electron atoms @xcite ( see also @xcite ) and applied to the theory of overlapping resonances in two - electron hci @xcite . another application was devoted to the theory of nonresonant corrections @xcite .
it was found in @xcite , that the lpa provides a convenient tool for calculating energy corrections .
moreover , it clearly determines the limit up to which the concept of the energy of the excited states has a physical meaning
that is the resonance approximation .
the exact theoretical values for the energy of the excited states defined , by the poles in the green s function , can be directly compared with measurable quantities only within the resonance approximation , where the line profile is described by the two parameters : energy @xmath8 and width @xmath9 . beyond this approximation
the evaluation of @xmath8 and @xmath9 should be replaced by the evaluation of the line profile for the particular process . moreover , in the case of two - electron atoms the line profile approach was found to be very efficient for the evaluation of the reference state correction ( reducible part of feynman graphs ) for two - electron atoms @xcite .
consider the simplest process of photon scattering on a one - electron ion which is assumed to be in its ground state @xmath10 ( fig .
[ figure06 ] ) . using the standard feynman rules for bound - electron qed @xcite yields the expression for the s - matrix element @xmath11 where @xmath12 is the wave function of the electron in the ground state , @xmath13 is the dirac matrix together with the electron propagator @xmath14 @xmath15 denotes the vector potential of the electromagnetic field ( photon wave function ) .
the notations @xmath16 and @xmath17 indicate `` up '' and `` down '' vertex coordinates , respectively .
insertion of the expressions for the electron propagator and the photon wave function yields @xmath18 \nonumber\\ & & \times e^{it_u({\mathop{\varepsilon}\nolimits}_{{\mathop { a } \nolimits}}+\omega ' ) } e^{-i\omega_{n}(t_u - t_d ) } \frac{i}{2\pi } \sum\limits_{n } \frac{\psi_{n}({{\mathbf}r}_u)\bar{\psi}_{n}({{\mathbf}r}_d ) } { \omega_{n}-{\mathop{\varepsilon}\nolimits}_{n}(1-i0 ) } e^{-it_d({\mathop{\varepsilon}\nolimits}_{{\mathop { a } \nolimits}}+\omega ) } \nonumber\\ & & \times [ \gamma^{\mu_d}a^{(k,\lambda)}_{\mu_d}({{\mathbf}r}_d ) \psi_{{\mathop { a } \nolimits}}({{\mathbf}r}_d ) ] \,.\end{aligned}\ ] ] here @xmath19 and @xmath20 are frequencies of the absorbed and emitted photons , respectively , @xmath21 are the photon momenta and @xmath22 denote the photon polarizations . the summation over @xmath23
is extended over the entire dirac spectrum of electrons in the nuclear coulomb field , @xmath24 are the dirac energy eigenvalues . integrating over time variables ( @xmath25 ) and abbreviating the expressions in the square brackets by @xmath26 and @xmath27 , respectively ,
we can write @xmath28 the function @xmath29 can be considered as a vertex function , which describes the absorption of a photon by an electron in its ground state .
below , we will formulate the resonance approximation , where we can define the energy and width of the level which have the general meaning independent of the features of the considered scattering process .
hence , the energy and the width will not depend on the function @xmath29 and thus we may consider the function @xmath29 as arbitrary . in particular , it can account for the interaction with the free electromagnetic field ( radiative corrections ) .
let us introduce in eq .
( [ lpaeq3 ] ) the matrix @xmath30 and integrate over the frequency @xmath31 . employing the definition of the transition amplitude @xmath32 via @xmath33 we derive the expression for the amplitude @xmath34 we will consider the resonance case , when the frequency @xmath35 is close to the value @xmath36 , where @xmath37 labels one of the exited states of an ion . in the resonance approximation we have to retain in eq .
( [ rsco2 ] ) only the dominant term with @xmath38 in the sum over @xmath23 , i.e. , @xmath39 in order to simplify the application of the line profile approach to the many - electron ions we introduce the abbreviated notations @xmath40 notice , that the function @xmath41 describes the process of scattering . to obtain the lorentz contour one has to insert the electron self - energy part in the internal electron line in fig .
[ figure06 ] .
for simplicity we neglect the vacuum - polarization part . to the lowest order this leads to the graph depicted in fig .
[ figure07 ] and the corresponding expression for the scattering amplitude evaluated within the resonance approximation takes the form @xmath42t\end{aligned}\ ] ] with @xmath43 here @xmath44 is the renormalized electron self - energy operator . the upper index at the function @xmath45 indicates the order of perturbation theory with respect to powers of the fine structure constant @xmath46 for the graphs contributing to this function . repeating these insertions in higher orders we can compose a geometric progression with the @xmath47-th term @xmath48^{l } = t^{\ast}d^{-1}\left[v^{(1)}(\omega)d^{-1}\right]^l t \,.\end{aligned}\ ] ]
the resulting geometric progression is convergent for any @xmath35 except for values within the interval @xmath49 $ ] close to the position of the resonance . applying
the formula for a convergent geometric progression one derives @xmath50^l
t \nonumber \\ \label{convergeometrprogr3 } & = & \frac{t^{\ast}t}{d - v^{(1)}(\omega ) } = \frac{t^{\ast}t}{\omega+{\mathop{\varepsilon}\nolimits}_{{\mathop { a } \nolimits}}-v^{(0)}-v^{(1)}(\omega ) } \,.\end{aligned}\ ] ] hence , the resonance is shifted into the complex plane and eq .
( [ convergeometrprogr3 ] ) is defined for all @xmath35 values on the real axis .
( [ convergeometrprogr3 ] ) presents the analytic continuation of the expansion @xmath51 to the entire complex plane .
taking the square modulus of the amplitude ( [ rscse ] ) , integrating over the directions of absorbed and emitted photons and summing over the polarizations we obtain the lorentz profile for the absorption probability @xmath52 here @xmath53 is the probability for the absorption of a photon within the frequency interval @xmath35 , @xmath54 and @xmath55 is the partial width of the level @xmath37 , associated with the transition @xmath56 .
taking into account the correction depicted in fig .
[ figure07 ] we improve the position of the resonance @xmath57 formula ( [ rscw ] ) defines the line profile of the process of scattering . within the resonance approximation
the line profile can be described by a lorentz contour which is characterized by two parameters : the position of the resonance and the width .
we define the energy shift for the state @xmath37 as the shift of the resonance .
the energy of the state @xmath37 is @xmath58 and the width of the level as the width of the corresponding lorentz contour at the position of the resonance @xmath59 we note that the energy @xmath8 of the level and its width @xmath9 defined in the framework of the resonance approximation do not depend upon the function @xmath41 ( or @xmath60 ) and , therefore , they do not depend upon the type of the scattering process .
for example , we are free to consider not only the scattering of a photon but of some other particle as well , which couples to electrons . going beyond the resonance approximation the line profile
can no longer be described by a lorentz contour and , consequently , the energy level can not be characterized only by two parameters @xmath8 and @xmath9 . in this case the evaluation of the energy levels should be replaced by the evaluation of the particular line profiles which , in general , depend upon the features of scattering process under consideration .
the real part of the matrix element @xmath61 describes the lowest - order contribution to the lamb shift and the imaginary part , which is finite and not subject to renormalization , represents the total radiative ( single - quantum ) width of the level @xmath37 : @xmath62 the other contribution to the lowest - order lamb shift @xmath63 originates from the vacuum polarization .
this correction does not contribute to the width @xmath64 @xcite .
studying the higher - order lamb shift in one - electron atoms within the line profile approach , we have to account for the feynman graph depicted in fig .
[ figure08 ] .
for reason of simplicity , we will not consider the other second - order graphs . in the case @xmath65 and @xmath66 the graph in fig .
[ figure09 ] can be viewed as a second - order self - energy insertion ( loop - after - loop , irreducible part ) in the graph fig . [ figure06 ] within the resonance approximation .
we derive the following expression for the scattering amplitude @xmath67 t \,,\end{aligned}\ ] ] where @xmath68 note , that the singular term @xmath38 is not included here by definition .
this term was taken into account in the geometric progression described above and represents exactly the second term of this progression .
repeating the evaluations leading to eq .
( [ rscw ] ) with @xmath69^l t \,,\end{aligned}\ ] ] we obtain the improved resonance condition @xmath70 solving this equation for @xmath71 up to terms @xmath72 yields @xmath73{\vphantom{\frac{8}{8 } } } _ { \textstyle \omega={\mathop{\varepsilon}\nolimits}_{a}-{\mathop{\varepsilon}\nolimits}_{{\mathop { a } \nolimits } } } \,\ } + { \mathop { \rm o } \nolimits}(\alpha^3 ) \,.\end{aligned}\ ] ] the term @xmath74 is the contribution of the irreducible part of the graph in fig .
[ figure08 ] .
the derivative term corresponds to the reference state ( @xmath38 ) correction .
in particular , it coincides with the reference state correction that arises from the feynman graph in fig .
[ figure08 ] after application of the adiabatic s - matrix method @xcite .
the other second - order electron self - energy ( sese ) corrections are irreducible @xcite . as in the one - electron case
we will consider the process of photon scattering on an ion which is assumed to be in its lowest ( ground ) state . investigating a non - degenerate energy level associated with a configuration containing at least one @xmath75-electron ( such as @xmath76 , @xmath77 , @xmath78 , @xmath79 , @xmath80 , etc ) we can represent the wave function of the ground state via a proper combination of one - electron dirac wave functions .
a procedure based on this approach has been accomplished in @xcite .
however , for the investigation of quasi - degenerate levels or a doubly excited level the interelectron interaction corrections have to be taken into account in the wave function of the ground state . here
we restrict ourselves to two - electron ions .
the generalization to n - electron ions will be presented at the end of this section . in the one - electron case we introduced the function @xmath60 describing the process of scattering under consideration .
to introduce such a function for the two - electron system we may consider first the simplest process of photon scattering on a two - electron ion disregarding the interelectron interaction corrections to the initial ( ground ) state .
this process is depicted in fig .
[ figure27 ] , where the ground state is represented by two non - interacting electrons one of which absorbs ( or emits ) the photon . accordingly , the ground - state wave function is given by @xmath81 where @xmath82 , @xmath83 are the dirac one - electron functions with different projections of the total angular momentum .
the bar over the one - electron functions indicates the dirac conjugation .
the s - matrix element corresponding to the graph in fig . [ figure27 ] can be written as @xmath84 \,.\end{aligned}\ ] ] in order to employ the functions @xmath85 , @xmath86 we introduced additional integrations @xmath87 .
the first and the second term in the square brackets represents graphs , where the photon is absorbed ( emitted ) by the first or by the second electron , respectively .
since the functions @xmath88 , @xmath89 are antisymmetric it would be sufficient to consider one of these terms only .
as in the one - electron case we will look for the position of the resonance and employ the resonance approximation .
it implies the neglect of the non - singular terms ( evaluated at the resonance ) in comparison with singular ones .
the terms in the sum over @xmath90 , @xmath91 in eq .
( [ lpaeq3x1old2 ] ) may contain a singularity at the position of the resonance only if they correspond to the positive - energy part of the dirac spectrum .
accordingly , in eq .
( [ lpaeq3x1old2 ] ) we can restrict to the terms with @xmath92 , @xmath93 . introducing the function
@xmath94 as @xmath95 we can write @xmath96 here we can employ the identity @xmath97 which follows from the sokhotsky formulas @xmath98 in view of the orthogonality of the dirac functions and the asymmetry of the functions @xmath99 and @xmath86 the first term of eq .
( [ polerevers ] ) yields exactly eq .
( [ lpaeq3x1old2 ] ) .
for @xmath100 the second term of ( [ polerevers ] ) does not contribute when inserted in eq .
( [ lpaeq3x1 ] ) . as it was noticed above , for @xmath101 the second term does not develop any singularity at the position of the resonance and can be disregarded .
having performed the integration over the time variables ( @xmath102 ) we arrive at @xmath103 where @xmath104 formally the expression ( [ lpaeq3x4 ] ) is similar to the one in eq .
( [ lpaeq3 ] ) which has been derived in the one - electron case . taking into account interelectron interaction corrections to the ground state the function @xmath105 and
the energy @xmath106 will become more complicated , in particular , the function @xmath105 will depend on @xmath107 , @xmath108 .
nevertheless , the form of the expression ( [ lpaeq3x4 ] ) would remain unchanged . below we will employ the resonance approximation defining the energy and width of the level such that they will not depend upon the features of the particular process of scattering .
since the function @xmath105 carries all information about the process of scattering we can assume it to be arbitrary .
accordingly , it is convenient to introduce a graphical designation : a rectangle with a letter @xmath10 inside ( see fig .
[ figure18 ] ) .
lower and upper rectangles represent the functions @xmath94 and @xmath109 , respectively , which are defined as @xmath110 here @xmath105 denotes a complicated vertex function describing the scattering process under consideration , @xmath106 is the energy of the ground state , @xmath35 , @xmath111 are the frequencies of the absorbed and emitted photons .
we will look for the position of a resonance near @xmath112 , where @xmath113 is the energy of two non - interacting electrons . applying the identity ( [ polerevers ] ) to eq .
( [ lpaeq3x4 ] ) one can see that the delta - function term is singular close to the resonance , while the other term remains regular ( here we assume the function @xmath105 to be arbitrary ) .
the resonance approximation implies the neglect of the non - singular ( at the resonance ) terms in comparison with singular ones .
accordingly , within the framework of the resonance approximation the expression for the s - matrix becomes @xmath114 integrating over @xmath107 , @xmath108 in eq .
( [ lpaeq3x4old ] ) and introducing the notation @xmath115 we can express the corresponding amplitude for the scattering process in a form , similar to eq .
( [ rsco2 ] ) : @xmath116 since we are searching for the position of the resonance near @xmath112 , we have to retain only terms in the denominator of ( [ amplituda2 ] ) for which @xmath117 .
we assume the energy level close to @xmath113 to be non - degenerate and hence , within the resonance approximation the amplitude takes the form @xmath118 here the summations run only over quantum numbers @xmath119 , satisfying the condition @xmath117 .
the matrices @xmath41 and @xmath120 are given by @xmath121 together with @xmath122 as in the one - electron case @xmath41 defines the type of scattering process under consideration .
the interelectron interaction correction in first order is represented by the graph in fig .
[ figure01 ] . in order to evaluate this contribution which also shifts the position the resonance
, one has to consider the graph in fig .
[ figure19 ] . in this paper
we employ the coulomb gauge together with the covariant metric .
the photon propagator can be written as @xmath123 where @xmath124 and @xmath125 the propagator @xmath126 corresponds to coulomb photons , while @xmath127 describes transverse ( breit ) photons .
the neglect of retardation implies the substitution @xmath128 .
we employ also the following notation @xmath129 the lorentz indices @xmath130 should indicate that the dirac matrices @xmath131 act on dirac wave functions depending on variables @xmath132 .
the corresponding s - matrix element reads @xmath133 while the summations over @xmath134,@xmath135,@xmath136,@xmath137 run over the entire dirac spectrum . employing the identity ( [ polerevers ] ) and retaining only terms which are singular near the resonance , i.e. keeping delta - function terms in ( [ polerevers ] ) only ,
we obtain @xmath138 within the resonance approximation we are left with terms satisfying the condition @xmath139 .
the second - order interelectron interaction correction is represented by the graphs in fig .
[ figure02 ] . in order to apply the line profile approach to the contribution of the `` box '' graph of fig .
[ figure02 ] we have to consider the graph depicted in fig .
[ figure23 ] ( a ) .
the corresponding s - matrix element reads @xmath140 employing the identity ( [ polerevers ] ) and retaining only the terms in the summation over @xmath134,@xmath135 and @xmath136,@xmath137 which become singular close to the resonance , we obtain the following expression for the s - matrix element corresponding to the graph in fig .
[ figure02 ] ( a ) @xmath141 again within the resonance approximation only terms satisfying the condition @xmath139 will be kept in the summations over @xmath142,@xmath143,@xmath144,@xmath145 .
let us consider separately the reference states terms , i.e. , for which @xmath146 holds . inserting a similar identity for the energy denominators @xmath147 into eq .
( [ lpaopkhgf6 ] ) , one can verify that the term with the delta - function coincides with the second element of the geometric progression for the graph in fig .
[ figure01 ] . hence
, while generating the geometric progression this term will refer to the second element of the progression . in order to evaluate rigorously the position of the resonance up to second order in @xmath46 we have to consider all corrections of first and second order simultaneously .
up to first - order of perturbation theory we have to account for self - energy ( se ) and vacuum - polarization ( vp ) corrections as well as for the exchange of one coulomb or one breit photon . in second order
we have to account for all one- and two - electron feynman graphs of second order including radiative corrections , screening of the self energy and vacuum polarization and two - photon exchange graphs .
however , the evaluation of the radiative corrections is not the goal of the present work . below we will present the derivation of a formula for the one - photon exchange contribution in fig .
[ figure01 ] , the two - photon exchange `` box '' graph on fig .
[ figure02 ] ( a ) and the three - photon exchange `` box '' graph fig .
[ figure03 ] . the scattering amplitude can be written as @xmath148 t \,,\end{aligned}\ ] ] where @xmath149 corresponds to the one - photon exchange graph fig .
[ figure01 ] ( see eq .
( [ lpaeq3x1v1x2 ] ) ) @xmath150 in contrast to the one - electron radiative corrections ( see formula ( [ v1se ] ) ) , this one - photon exchange correction does not depend on @xmath35 .
taking into account the `` box '' graph in fig .
[ figure02 ] ( a ) we obtain ( see formulas ( [ lpaopkhgf6 ] ) , ( [ lpa0305110953x ] ) ) @xmath151 again the summations over @xmath152 run over scalar ( coulomb ) and transverse ( breit ) photons , respectively . the second term in eq .
( [ v22ph ] ) represents the remainder after subtracting off the reference state singularity .
this subtraction was done at the stage of generating the geometric progression with the one - photon exchange insertion .
in particular , the identity ( [ lpa0305110953x ] ) was employed for deriving the expression for the reference state contribution ( the reducible part ) of the `` box '' graph , where the delta - function term coincides with the second element of geometric progression for the one - photon exchange graph . in the case of one - electron ions the reference state term ( see eq .
( [ v2e1ph2 ] ) ) appeared only via the derivative term in eq .
( [ seseo6nr ] ) .
considering photon exchange in two - electron ions the situation is different . in this case
a reference state contribution appears directly in @xmath153 ( see eq .
( [ v22ph ] ) ) while a derivative term does not arise since @xmath149 does not depend on @xmath35 . nevertheless , if the sets @xmath154 and @xmath155 are equivalent , the term corresponding to the reference states can be expressed as a derivative . at the point of the resonance we can set @xmath156 in eq .
( [ v22ph ] ) .
accordingly , both factors in the denominator will be identical . utilizing the formula @xmath157 and integrating by parts
we can shift the derivative to @xmath158 .
let us now turn to the three - photon exchange correction neglecting retardation effects , crossed - photon graphs and the contribution of the negative energy part of the dirac spectrum . within this approximation
the photon propagator does not depend on the frequency @xmath159 , which yields @xmath160 where the prime at the summation symbol indicates that the reference states ( @xmath161 , @xmath162 ) are omitted .
taking together the contributions of @xmath149 , @xmath163 and @xmath164 , we can generate a geometric progression , where the @xmath47-th term reads @xmath165^{l } t \,.\end{aligned}\ ] ] performing similar steps as in the one - electron case we sum up this progression and derive a condition for the position of the resonance @xmath166 the energy and the width of the level will be equal to ( see eqs .
( [ lpadefe ] ) , ( [ lpadefgamma ] ) ) @xmath167 it is important to emphasize that eq .
( [ defgammawidth2el ] ) has the meaning of the correction to the width of the level only if the full set of feynman graphs of a given order is under consideration .
indeed , the contribution of the graph in fig . [ figure01 ] cancels completely with a part of the contribution of the self - energy correction , while the vacuum polarization gives zero contribution to the width .
such a cancellation is an immediate consequence of the pauli principle according to which transitions of electrons into occupied states are prohibited @xcite .
we note that the two - electron graphs to the first order in @xmath149 does not depend on @xmath35 .
hence , the solution of eq .
( [ v3e2ph3 ] ) together with @xmath149 , @xmath163 , @xmath164 given by eqs .
( [ v1e2ph1x ] ) , ( [ v22ph ] ) , ( [ v3e2ph3x ] ) yields @xmath168{\vphantom{\frac{8}{8 } } } _ { \textstyle \omega={\mathop{e}\nolimits}^{(0)}-{\mathop{e}\nolimits}_{{\mathop { a } \nolimits } } } \,\ } \nonumber\\ & & + { \mathop { \rm o } \nolimits}(\alpha^4 ) \,.\end{aligned}\ ] ] the term @xmath169 represents the contribution of one - photon exchange graph fig
. [ figure01 ] , while the term @xmath74 accounts for the contribution of the two - photon exchange graphs in fig .
[ figure02 ] .
in particular , this term includes the contribution of the reference states occurring in these graphs .
the third - order term @xmath170 contributes to the three - photon exchange graphs in fig .
[ figure03 ] . it does not contain the contribution of the reference states because we disregard the retardation effects considering it within the framework of relativistic many - body perturbation theory ( rmbpt ) . the derivative term in eq .
( [ el2final ] ) as well as that term in eq .
( [ seseo6nr ] ) does not correspond to a certain feynman graph .
similar to eq .
( [ seseo6nr ] ) it can be related to the contribution of the reducible part ( reference states ) of the graph in fig .
[ figure03 ] .
let us mention that if we would take into account in addition the radiative corrections and as well as screening effects , we would obtain a corresponding correction @xmath171 containing the contribution of the electron self energy ( [ v1se ] ) , the one - photon exchange ( [ v1e2ph1x ] ) and the vacuum polarization . for @xmath153 we would similarly obtain the sum of eqs .
( [ v2e1ph2 ] ) , ( [ v22ph ] ) and in addition all the missing radiative effects of second order together with the screened self - energy and vacuum - polarization corrections .
accordingly , instead of eq .
( [ el2final ] ) we would have been left with @xmath172 { \vphantom{\frac{8}{8}}}_{\textstyle \omega={\mathop{e}\nolimits}^{(0)}-{\mathop{e}\nolimits}_{{\mathop { a } \nolimits } } } + v^{(3)}({\mathop{e}\nolimits}^{(0)}-{\mathop{e}\nolimits}_{{\mathop { a } \nolimits } } ) \nonumber\\ & & + \frac{1}{2 } \
, v^{(1)}({\mathop{e}\nolimits}^{(0)}-{\mathop{e}\nolimits}_{{\mathop { a } \nolimits}})^2 \left[\frac{\partial^2 v^{(1)}(\omega)}{\partial\omega^2 } \right]{\vphantom{\frac{8}{8 } } } _ { \textstyle \omega={\mathop{e}\nolimits}^{(0)}-{\mathop{e}\nolimits}_{{\mathop { a } \nolimits } } } \nonumber\\ & & + v^{(1)}({\mathop{e}\nolimits}^{(0)}-{\mathop{e}\nolimits}_{{\mathop { a } \nolimits } } ) \left[\frac{\partial v^{(1)}(\omega)}{\partial\omega } \right]{\vphantom{\frac{8}{8}}}_{\textstyle \omega={\mathop{e}\nolimits}^{(0)}-{\mathop{e}\nolimits}_{{\mathop { a } \nolimits}}}^2 \nonumber\\ & & + v^{(1)}({\mathop{e}\nolimits}^{(0)}-{\mathop{e}\nolimits}_{{\mathop { a } \nolimits } } ) \left[\frac{\partial v^{(2)}(\omega)}{\partial\omega } \right]{\vphantom{\frac{8}{8 } } } _ { \textstyle \omega={\mathop{e}\nolimits}^{(0)}-{\mathop{e}\nolimits}_{{\mathop { a } \nolimits } } } \nonumber\\ & & \left .
+ v^{(2)}({\mathop{e}\nolimits}^{(0)}-{\mathop{e}\nolimits}_{{\mathop { a } \nolimits } } ) \left[\frac{\partial v^{(1)}(\omega)}{\partial\omega } \right]{\vphantom{\frac{8}{8 } } } _ { \textstyle \omega={\mathop{e}\nolimits}^{(0)}-{\mathop{e}\nolimits}_{{\mathop { a } \nolimits } } } \,\right\ } + { \mathop { \rm o } \nolimits}(\alpha^4 ) \,.\end{aligned}\ ] ] formulating the line profile approach for @xmath173-electron ions , it might be convenient to introduce the function @xmath174 which should be depicted graphically by a rectangle with a letter @xmath10 inside and with @xmath173 out - going electron lines . here
@xmath175 describes the @xmath173-electron ions in the lowest ( ground ) state @xmath10 together with the absorbed photon .
accordingly , formula ( [ polerevers ] ) generalizes to @xmath176 \nonumber\\ & & \times \frac{1}{(\omega_{n_n}-{\mathop{\varepsilon}\nolimits}_{n_n}(1-i0 ) ) } \,.\end{aligned}\ ] ] this identity can be written as @xmath177 then the terms in eqs .
( [ lpaeq3x4 ] ) , ( [ lpaeq3x1v1 ] ) and ( [ lpaeq3x1v186 ] ) , which correspond to the function @xmath178 will not contain the singularities and will be omitted within the framework of the resonance approximation . eqs .
( [ el2final ] ) and ( [ finallpa ] ) will remain unchanged , however , now @xmath45 will contain additional contributions of three- and up to @xmath173-electron graphs . in particular , for three - electron ions the functions @xmath153 and @xmath179 will also account for contributions of three - electron graphs ( see @xcite ) .
we now turn to the application of the line profile approach to quasi - degenerate levels . without loss of generality
, we can restrict ourselves to two mixing configurations .
we will search for the positions of the resonances corresponding to these configurations and will construct basic wave functions @xmath180 and @xmath181 within the coupling scheme .
the energies corresponding to these wave functions are denoted by @xmath182 , @xmath183 and they are supposed to be close to the exact energies of the electron configurations under consideration . employing the line profile approach we will consider a scattering of a photon on a two - electron ion in its ground state @xmath184 .
the positions of resonances may be found near the values @xmath185 and @xmath186 , respectively . within the resonance approximation
we will have to retain two terms in the sum ( [ amplituda2 ] ) corresponding to the basic functions @xmath180 and @xmath181 .
the scattering amplitude may be written as @xmath187 t \,,\end{aligned}\ ] ] where @xmath188 is a matrix @xmath189 , defined on the functions @xmath180 , @xmath181 : @xmath190 here @xmath191 , @xmath192 are the one - electron dirac hamiltonians acting on the one - electron dirac wave functions depending on @xmath193 or @xmath194 , respectively . since the functions @xmath180 and @xmath181 are orthogonal
the matrix @xmath188 is diagonal .
accordingly , we now have to compose a geometric matrix progression with the @xmath47-th term @xmath195^{l } t \,\end{aligned}\ ] ] and sum it up employing the formula for a convergent geometric progression .
the expression for the amplitude reads @xmath196^{-1 } t \equiv t^{+ } \frac{1}{d-\delta v } t = t^{+ } \frac{1}{\omega+{\mathop{e}\nolimits}_{{\mathop { a } \nolimits}}- v } t \,.\end{aligned}\ ] ] introducing the function @xmath197 by means of the relation @xmath198 , where the matrix @xmath199 is assumed to diagonalize the matrix , i.e. @xmath200 .
the expression for the amplitude can now be written in the following form @xmath201_{\phi_1 \phi_1 } } t_{\phi_1 { \mathop { a } \nolimits } } + t^{+}_{{\mathop { a } \nolimits}\phi_2 } \frac{1}{\omega+{\mathop{e}\nolimits}_{{\mathop { a } \nolimits}}-[b^{+}vb]_{\phi_2 \phi_2 } } t_{\phi_2 { \mathop { a } \nolimits } } \nonumber \\ \label{ampl86326rfg } & = & t^{+}_{{\mathop { a } \nolimits}\phi_1 } \frac{1 } { \omega+{\mathop{e}\nolimits}_{{\mathop { a } \nolimits}}- v^{\text{diag}}_{\phi_1 \phi_1}(\omega ) } t_{\phi_1 { \mathop { a } \nolimits } } + t^{+}_{{\mathop { a } \nolimits}\phi_2 } \frac{1 } { \omega+{\mathop{e}\nolimits}_{{\mathop { a } \nolimits}}- v^{\text{diag}}_{\phi_2 \phi_2}(\omega ) } t_{\phi_2 { \mathop { a } \nolimits } } \,.\end{aligned}\ ] ] taking a square modulus of the amplitude ( [ ampl86326rfg ] ) and integrating over the directions of the absorbed and emitted photons yields a line profile for the probability of photon absorption . the positions of the resonances are determined by the equations @xmath202 hence , the energy of the configurations are : @xmath203 assuming that the energies of the configurations are close to each other , we can expand eqs .
( [ sdfc1 ] ) , ( [ sdfc2 ] ) into a taylor series around the values @xmath204 and @xmath205 , respectively . as in the case of non - degenerate levels
this can be achieved up to any desired accuracy .
note , that employing the resonance approximation in case of non - degenerate level we have to retain in a corresponding sum ( [ amplituda2 ] ) certain many - electron functions composed within the coupling scheme . indeed , after diagonalization of the matrix @xmath45 all other combinations of one - electron functions will yield zero in view of the antisymmetry of the wave function of the ground state and the symmetry of the matrix @xmath45 .
hence , having constructed a many - electron function in the coupling scheme , eq .
( [ twoelprogr5 ] ) becomes a scalar one .
the line profile approach outlined above can be easily employed for an arbitrary number of degenerate levels .
the generalization of the method to @xmath173-electron ions was described at the end of the previous section .
we will evaluate the interelectron interaction correction for the two - electron configurations @xmath206 and @xmath207 . employing the relativistic coupling scheme these energy levels become quasi - degenerate in the region @xmath3 . to treat such states within the framework of qed
we will apply the line profile approach . within the coupling scheme
the wave function of a two - electron configuration can be represented as @xmath208 \,,\end{aligned}\ ] ] where the normalization constant is @xmath209 for equivalent electrons and @xmath210 for non - equivalent electrons , respectively .
@xmath211 is a clebsch - gordan coefficient .
the one - electron dirac functions @xmath212 are characterized by the standard set of one - electron quantum numbers : total angular momentum @xmath213 , its projection @xmath214 and the orbital angular momentum @xmath47 , that fixes the parity of the state .
for the two - electron wave function the relevant quantum numbers are the total angular momentum @xmath215 and its projection @xmath216 . following the procedure described in section [ lpa3 ] we will construct the matrix @xmath45 ( [ lpadeltav ] ) on the functions ( [ funcxjj ] ) @xmath217 and examine the positions of the resonances close to @xmath218 and @xmath219 , respectively .
as it has been elaborated in section [ lpa ] the operator @xmath45 , in general , depends on @xmath35 .
the position of the resonance can be derived via taylor expansion at the approximate positions of the resonances @xmath220 and @xmath221 ( see eq .
( [ el2final ] ) ) . for the practical calculations it is convenient to expand some matrix elements of @xmath45 at the point @xmath222 and others at the point @xmath223 keeping only terms @xmath224 in both expansions .
the resulting inaccuracy can be referred to corrections @xmath72 @xcite , because at low @xmath4 values the energy difference @xmath225 becomes small , while at large-@xmath4 values the degeneracy of the levels @xmath226 , @xmath227 is nearly negligible .
the interelectron interaction correction is represented by the set of graphs ( figs .
[ figure01 ] and [ figure02 ] ) which is symmetric under interchange of the upper and the lower indices and relabeling of the electrons in the graphs .
accordingly , the operator @xmath45 is given by a symmetric ( and in general complex ) matrix .
however , as a consequence of performing the taylor expansion of the matrix elements of @xmath45 and neglecting third- and higher - orders terms it can lead to a non - symmetrical matrix . to prevent this asymmetry arising due to purely technical reasons
one may symmetrize the matrix @xmath45 by hand .
hence , the matrix elements of the frequency - dependent operator @xmath45 evaluated at the resonances can be written as @xmath228 \
\langle(1s2p_{3/2})|v^{}({\omega^{\text{res}}})|(1s2p_{1/2})\rangle & = & \langle(1s2p_{1/2})|v^{}({\omega^{\text{res}}})|(1s2p_{3/2})\rangle \,.\end{aligned}\ ] ] the operator @xmath229 is defined via its action on the set of the one - electron dirac functions @xmath230 , which in our case consists of @xmath231 . to zeroth - order perturbation theory the operator @xmath229 reads ( see eq .
( [ lpaquasv0 ] ) ) : @xmath232 being interested in ionization energies it is more convenient to introduced a shifted @xmath229 with the zeroth - order matrix element @xmath233 where the electron rest energy @xmath214 ( in relativistic unites ) and @xmath234-electron energy are subtracted . in first - order perturbation theory
the interelectron interaction represented by the graph in fig .
[ figure01 ] can be described by the matrix element @xmath235 since graph in fig .
[ figure01 ] is irreducible @xmath236 coincides with the expression ( eq .
( [ v1e2ph1x ] ) ) for non - degenerate levels . in second - order perturbation
theory we have to account for the two - photon exchange corrections depicted in fig . [ figure02 ] : @xmath237 @xmath238 \nonumber\\ & & + \delta_{{\mathop{e}\nolimits}^{(0)}_{n_1n_2},{\mathop{e}\nolimits}^{(0)}_{a'b ' } } ( 1-\delta_{{\mathop{e}\nolimits}^{(0)}_{n_1n_2},{\mathop{e}\nolimits}^{(0)}_{ab } } ) \nonumber\\ & & \times \left[\vphantom{\frac{8}{8}}\right . \frac
{ i(|{\mathop{\varepsilon}\nolimits}_{n_2}-{\mathop{e}\nolimits}^{(0)}_{ab}+{\mathop{\varepsilon}\nolimits}_{a'}|)_{a'b'n_1n_2 } i(|{\mathop{\varepsilon}\nolimits}_{n_2}-{\mathop{e}\nolimits}^{(0)}_{ab}+{\mathop{\varepsilon}\nolimits}_{a}|)_{n_1n_2ab } } { { \mathop{e}\nolimits}^{(0)}_{ab}-{\mathop{e}\nolimits}^{(0)}_{n_1n_2 } } \nonumber\\ & & + \frac { i(|{\mathop{\varepsilon}\nolimits}_{n_2}-{\mathop{\varepsilon}\nolimits}_{a'}|)_{b'a'n_1n_2 } i(|{\mathop{\varepsilon}\nolimits}_{n_2}-{\mathop{\varepsilon}\nolimits}_{a}|)_{n_1n_2ba } } { { \mathop{e}\nolimits}^{(0)}_{ab}-{\mathop{e}\nolimits}^{(0)}_{n_1n_2 } } \left.\left . \vphantom{\frac{8}{8}}\right]\right\ } \,,\end{aligned}\ ] ] @xmath239 @xmath240 in eqs .
( [ qboxirr ] ) and ( [ qboxred ] ) the notations @xmath241 , @xmath242 , @xmath243 are introduced .
index @xmath244 runs over @xmath245 ( scalar and transverse photons ) . the kronecker symbols
ensure that terms with potentially zero denominators will be omitted in the summation over @xmath246 . note , that the eqs .
( [ qboxirr ] ) , ( [ qcrossirr ] ) , ( [ qcrossred ] ) for the irreducible parts coincide generically with eq .
( [ v22ph ] ) for non - degenerate levels ( see also @xcite ) .
however , for the reducible part of the `` box '' graph additional terms originating from the geometric progression for the one - photon exchange graph ( non - diagonal matrix elements of the second term of the progression ) occur .
it is easy to make sure that the contribution of reference states ( @xmath247 , @xmath248 ) to the exchange of two coulomb photons ( or breit photons with neglect of retardation ) is absent .
the results of the numerical calculations are presented in tables [ txquasidegx1 ] , [ txquasidegx3 ] . to account for nuclear size corrections we solved the dirac equation with the coulomb potential generated by a nuclear charge density described by a fermi distribution .
the parameters of the fermi distribution are taken from ref .
@xcite . in table [ txquasidegx1 ] we present a detailed analysis of our results obtained for the photon - exchange contribution .
the value @xmath249 is the binding energy of the @xmath250 state according to eq .
( [ defusedboundenergy ] ) , the value @xmath149 corresponds to the one - photon exchange contribution eq .
( [ qboxirrfant ] ) and @xmath163 represents the two - photon exchange contributions given by eqs .
( [ qboxirr][qcrossred ] ) .
we note , that , in general , the matrix @xmath45 has complex elements and both their real and imaginary parts contribute to the energy eigenvalues , i.e. , the real part of the diagonalized matrix @xmath45 . in our calculation
the imaginary part of @xmath149 is taken into account , while the imaginary part of the two - photon exchange contribution ( @xmath163 ) is neglected .
the values @xmath251 and @xmath252 denote the photon - exchange contribution to the energies of the corresponding electron configurations ( neglecting the radiative corrections ) .
for @xmath5 we present also the values for the difference between the energies of the levels under consideration reported in ref .
@xcite . in order to analyse the influence of the quasi - degeneracy on qed effects in more detail we compile the corresponding energy shifts of the levels due to the photon - exchange contribution calculated within various approximations .
the differences between the energies of the levels calculated without the approximations and the energies calculated within the framework of the approximations are presented in table [ txquasidegx1 ] [ @xmath253 : appr . 1 - 5 ]
. approximation 2 : we omit the non - diagonal elements only in the matrix @xmath163 .
as stated above the expression for the one - photon exchange correction ( [ qboxirrfant ] ) does not depend on @xmath35 and coincides with the one for the non - degenerate case .
accordingly , the first - order contribution is taken into account just as the solution of the secular equation , i.e. following usual techniques developed in quantum mechanics for treating degenerate levels .
the influence of quasi - degeneracy due to the second - order matrix element @xmath163 is neglected .
approximation 3 : we calculate the matrix elements of @xmath45 within the framework of rmbpt . compared with
the full ab initio qed calculation the following contributions are missing : 1 ) negative - energy intermediate states , 2 ) crossed - photon interaction , 3 ) rigorous treatment of retardation effects . as mentioned above within the framework of rmbpt no contribution due to reference states ( for two - photon exchange ) arises .
accordingly , the energies of the levels just follow as solutions of the secular equation .
approximation 4 : only the matrix elements of @xmath163 are evaluated within the framework of rmbpt . according to the comment made on approximation 2 , this also follows the quantum mechanical treatment for quasi - degeneracy .
approximation 5 : we neglect the imaginary part of the elements of the matrix @xmath45 .
the matrix @xmath45 defined in eq .
( [ lpadeltav ] ) is a complex one .
although the energy of the level is defined as the real part of the diagonalized matrix @xmath255 at the point of the resonance ( @xmath256 ) , the imaginary part of the elements of the matrix @xmath45 ( non - diagonal ) contributes to the energy .
the results in table [ txquasidegx1 ] demonstrate that for the @xmath257 , @xmath258 levels a complete ab initio qed theory for describing the quasi - degeneracy has to be employed only when going beyond the level of second - order corrections . for @xmath259 approximation 3
provides an accuracy of about 1% at the level of second - order perturbation theory .
accordingly , the inaccuracy can be referred to corrections of third order . for @xmath260 approximation 4
leads to an inaccuracy comparable in magnitude with the corrections of third order . for @xmath261 the effect of quasi - degeneracy decreases definitely to the level of third - order corrections .
consequently , approximation 2 can be employed for high-@xmath4 systems . for @xmath262
the quasi - degeneracy becomes completely negligible , i.e. , it will be sufficient to employ approximation 1 .
the contribution of imaginary parts of the matrix elements @xmath45 to the energy levels appears as an effect of quasi - degeneracy , which originates completely from qed .
it is perceptible only for high @xmath263 , which also reveals that the neglect of the imaginary part of @xmath163 has been legitimate . in table
[ txquasidegx3 ] we present the data for the total energies of the @xmath257 and @xmath258 two - electron configurations , respectively .
the numbers present the ionization energy of the @xmath250-electron with the opposite sign .
these data are compared with the results obtained by plante _
et al . _
@xcite and drake @xcite .
two different approximate methods have been employed in these works : the `` relativistic all - order theory '' ( ao ) @xcite and `` the unified theory '' @xcite .
the latter methods account approximately for qed effects , such as retardation , crossed - photon graphs and negative - energy intermediate states , while taking into account partially higher - orders of the perturbation theory . in the present work
the photon exchange is taken into account up to the second order . the self - energy ( se ) and vacuum - polarization ( vp )
corrections are included only in first order .
quantitative results for se and vp corrections are taken from refs .
the se and vp screening corrections , the radiative corrections of the second order and all the corrections of the third and higher orders are omitted .
we note , that the vp screening corrections for the states considered have been evaluated by artemyev _
@xcite , while results for the se screening corrections are not yet available .
since the se and vp screening corrections partially cancel each other , we do not include the results of @xcite in table [ txquasidegx3 ] . in table
[ txquasidegx5 ] we present various theoretical and experimental data for @xmath264 transition energies .
we conclude that the discrepancy between our data and those from other results arising for small values of @xmath4 is caused by third- and higher - orders of the perturbation theory which have not been accounted for in the present paper . for high @xmath4
the major inaccuracy is due to missing self - energy , vacuum - polarization screening corrections and one - electron radiative corrections of second order .
[ acknowledgement ] the authors are indebted to prof .
w. nagel from the center of high - performance computing at tu dresden for providing access to all necessary computer facilities .
o.y.a . is grateful to tu dresden for the hospitality during his visits in 2002 and 2003 and to the dfg for financial support .
the work of o.y.a . and
l.n.l . was supported by the rfbr grant no .
02 - 02 - 16578 and by minobrazovanie grant
e02 - 3.1 - 7 .
g.p . and g.s .
acknowledge financial support from bmbf , dfg and gsi .
10 [ 2]#2 [ 1]#1 [ 1]#1 , , ( ) . , , and , , ( ) . , , and , , ( ) . , , and , , ( )
, , and , ( , bristol and philadelphia , ) .
, , , and , , ( ) .
, , , and , , ( ) .
, , , , , and , , ( ) . and , , ( ) .
, , , and , , ( ) .
, , and , , ( ) .
, , , and , , ( ) . and , , ( ) .
, , , , , and , , ( ) .
, , , , , and , , ( ) .
, , , , , and , , ( ) . and , , ( ) .
, , ( ) . , , ( ) .
, , , , and , , ( ) . , , , , , and , , ( ) . , , and , , ( )
, , , and , , ( ) . and , , ( ) .
, , ( ) . .
, , , , and , , ( ) . , , and , , ( ) . , , and , , ( )
. and , , ( )
. and , , ( ) . ,
( , moscow , ) . , , ( ) . and , , ( )
. and , , ( ) .
, , and , , ( ) .
, , , , , , , and , , ( ) . , , , and , , ( ) .
, , , and , , ( ) .
, , , , and , , ( ) .
.matrix elements of the operator @xmath45 for the two - electron configurations @xmath226 and @xmath227 ( ev ) .
the individual contributions for the dirac - binding energies of @xmath250-electron states ( @xmath249 ) , the one - photon exchange contribution ( @xmath149 ) and the two - photon contribution ( @xmath163 ) are compiled , respectively .
@xmath251 , @xmath252 are the energies of the corresponding configurations , where only the photon - exchange contributions are taken into account ( neglecting radiative corrections ) . [
cols="<,>,>,>,>,>,>,>,>,>,>",options="header " , ] | accurate qed calculations of the interelectron interaction corrections for the @xmath0 , @xmath1 two - electron configurations for ions with nuclear charge numbers @xmath2 are performed within the line profile approach .
total energies of these configurations are evaluated . employing the fully relativistic treatment based on the coupling scheme
these energy levels become quasi - degenerate in the region @xmath3 . to treat such states within the framework of qed we utilize the line profile approach .
the calculations are performed within the coulomb gauge . |
we are grateful to andrew przeworski for very helpful correspondence concerning his work , jeff weeks for many useful comments , and neil cornish for drawing our attention to the reference @xcite .
we also thank faperj and cnpq for the grants under which this work was carried out .
figure 1 : : the solutions curve of @xmath34 , as plots of @xmath0 ( vertical axis ) versus @xmath48 ( horizontal axis ) , with @xmath2 taken as @xmath10 ( upper curve ) and @xmath38 ( lower curve ) , respectively .
the depth of the survey in both cases correspond to a redshift @xmath51 ( cmbr ) .
included also is a dashed rectangular box , representing the relevant part , for our purposes , of the hyperbolic region of the parameter space given by recent observations .
the undetectable region of the parameter space ( @xmath52 ) corresponding to each value of @xmath53 lies above the related curve .
ellis , _ gen .
* 2 * , 7 ( 1971 ) ; d.d .
sokolov & v.f .
shvartsman , _ sov .
jetp _ * 39 * , 196 ( 1974 ) ; g.f.r .
ellis & g. schreiber , _ phys .
a _ * 115 * , 97 ( 1986 ) ; r. lehoucq , m. lachize - rey & j .- p .
luminet , _ astron .
astrophys . _ * 313 * , 339 ( 1996 ) ; b.f .
roukema , _ mon . not .
r. astron .
_ * 283 * , 1147 ( 1996 ) ; g.f.r .
ellis & r. tavakol , _ class .
quantum grav . _
* 11 * , 675 ( 1994 ) ; n.j .
cornish , d.n .
spergel & g.d .
starkman , _ proc .
sci . _ * 95 * , 82 ( 1998 ) ; n.j .
cornish & j.r .
weeks , astro - ph/9807311 ( 1998 ) ; g.i .
gomero , a.f.f .
teixeira , m.j .
rebouas & a. bernui , gr - qc/9811038 ( 1998 ) ; r. lehoucq , j .-
luminet & j .- p .
uzan , _ astron .
astrophys . _ * 344 * , 735 ( 1999 ) ; j.p .
uzan , r. lehoucq & j.p .
luminet , _ astron .
_ * 351 * , 776 ( 1999 ) ; h.v .
fagundes & e. gausmann , _ phys .
lett . a _ * 261 * , 235 ( 1999 ) ; m.j.rebouas _ int . j. mod .
d _ * 9 * , 561 ( 2000 ) g.i .
gomero , m.j .
rebouas & a.f.f .
teixeira , _ phys .
a _ * 275 * , 355 ( 2000 ) ; g.i .
gomero , m.j .
rebouas & a.f.f .
quantum grav . _ * 18 * , 1885 ( 2001 ) ; r. lehoucq , j.p .
uzan & j.p .
luminet , astro - ph/0005515 ( 2000 ) ; v. blanlil & b.f .
roukema , eds .
, in the proceedings of the _ cosmological topology in paris 1998 _ , astro - ph/0010170 .
starkman , _ class .
quantum grav . _
* 15 * , 2529 ( 1998 ) ; m. lachize - rey & j .- p .
luminet , _ phys .
rep . _ * 254 * , 135 ( 1995 ) ; ya .
b. zeldovich & i.d .
novikov , _ the structure and evolution of the universe _ , pages 633 - 640 .
the university of chicago press , chicago ( 1983 ) .
see also on page 637 of this book the references to earlier works by sveges ( 1966 ) , sokolov ( 1970 ) , paal ( 1971 ) , sokolov and shvartsman ( 1974 ) , and starobinsky ( 1975 ) .
d. atkatz & h. pagels , _ phys . rev .
d _ * 25 * , 2065 ( 1982 ) ; ya.b .
zeldovich & a.a .
starobinsky , _ sov .
_ * 10 * , 135 ( 1984 ) ; y.p .
goncharov & a.a .
bystsenko , _ astrophysics _ * 27 * , 422 ( 1989 ) . a.e .
_ , _ phys .
d _ * 63 * , 042001 ( 2001 ) ; p. de
bernardis _ et al .
_ , in proceedings of the capp2000 conference , verbier , 17 - 28 july 2000 , astro - ph/0011469 ( 2000 ) ; j.r .
et al . _ , in proceedings of neutrino 2000 ( elsevier ) , cita-2000 - 63 , eds .
j. law & j. simpson , astro - ph/0011381 ( 2000 ) ; j.r .
et al . _ , in proceedings of iau symposium 201 ( pasp ) , cita-2000 - 65 , astro - ph/0011378 ; a. balbi _
_ , _ astrophys . j. _ * 545 * , l1l4 ( 2000 ) .
_ , _ astrophys .
j. _ * 507 * 46 ( 1998 ) ; a.g .
_ , _ astron .
j. _ * 116 * , 1009 ( 1998 ) ; s. perlmutter _
_ , _ astrophys . j. _ * 517 * , 565 ( 1999 ) ; s. perlmutter , m.s .
turner & m. white , _ phys .
_ * 83 * , 670 ( 1999 ) . | using recent observational constraints on cosmological density parameters , together with recent mathematical results concerning small volume hyperbolic manifolds , we argue that , by employing pattern repetitions , the topology of nearly flat small hyperbolic universes can be observationally undetectable .
this is important in view of the facts that quantum cosmology may favour hyperbolic universes with small volumes , and from the expectation , coming from inflationary scenarios , that @xmath0 is likely to be very close to one . it is well known that general relativity is a local metrical theory and therefore the corresponding einstein field equations do not fix the global topology of spacetime .
this freedom has resulted in a great deal of interest in the possibility that the universe may possess compact spatial sections with a non - trivial topology , which do not necessarily have positive curvature ( see for example @xcite
@xcite and references therein ) .
interest in such spaces has also come from quantum cosmology , where the existence of the wave function of the universe may require finiteness of the spatial sections ( see for example @xcite ) .
also , in the ` tunnelling from nothing ' scenario , manifolds with small compact spatial sections may be more likely to emerge @xcite .
furthermore , there is a precise sense in which most compact @xmath1-manifolds are hyperbolic @xcite .
together these facts have motivated the study of hyperbolic manifolds with non - trivial topology as possible models for our universe .
there has also been two other important developments recently : one observational , the other mathematical . regarding the former
, recent observations seem to indicate that the ratio of the total density to the critical density of the universe , @xmath0 , is likely to be very close to one @xcite , with a significant proportion of this energy being in the form of a dark component with negative pressure @xcite .
regarding the latter , we first of all discuss briefly some preliminaries and recall that even though at present there is no complete classification of hyperbolic manifolds , a number of important results are known about them , including the two important theorems of mostow @xcite and thurston @xcite . according to the former ,
geometrical quantities of orientable hyperbolic manifolds , such as their ( finite ) volumes and the lengths of their closed geodesics , are topological invariants . according to the latter
, there is a countable infinity of sequences of compact orientable hyperbolic manifolds , with the manifolds of each sequence being ordered in terms of their volumes @xcite , with an overall lower bound which is shown to be greater than 0.28151 @xcite .
a natural way to characterize the shape of such manifolds is in terms of the sizes of their closed geodesics .
a particularly useful indicator in this regard is the so called _ injectivity radius _
, @xmath2 , defined as the radius of the smallest sphere inscribable in @xmath3 .
this in turn allows the definition of a related indicator that has often been utilized in most studies regarding searches for topological multiple images ( see , for example @xcite
@xcite ) and references therein ) , namely the ratio of the injectivity radius to the depth @xmath4 of the astronomical survey up to a given redshift @xmath5 [ r_inj ] t_inj=. the crucial point regarding this indicator is that , in any universe for which @xmath6 , there would be no observed multiple images of either cosmic objects or spots of cosmic microwave background radiation ( cmbr ) , and therefore the topology would not be detectable observationally using pattern repetition , no matter how accurate the observations . despite the global inhomogeneity of hyperbolic manifolds
this result is location independent and therefore applicable to any observer in the universe .
similarly the set of universes for which @xmath7 are observationally detectable through pattern repetitions , at least in principle , for some observers .
an important point regarding @xmath2 is that its lower bound in the set of all compact hyperbolic manifolds is zero .
thus , no matter what the cosmological parameters ( and the resulting @xmath4 ) , there will always exist compact hyperbolic universes ( with small enough @xmath2 such that @xmath7 ) such that their topologies are detectable observationally , at least by some observers , even though their number will decrease drastically as @xmath8 @xcite .
now given the potential importance of the small volume hyperbolic manifolds in connection , e.g. , with quantum cosmology , the question arises as to whether the existence of detectable topologies for all values of @xmath9 still holds if we restrict ourselves only to a set of small volume manifolds .
surprisingly , the answer turns out to be in the negative . to see this
, we recall an important set of recent mathematical results which show that very small values of @xmath2 do not occur in small volume hyperbolic manifolds , and that there is in fact a lower bound on the lengths of geodesics in any set of small volume closed orientable hyperbolic 3-manifolds @xcite .
thus , for example , according to a theorem of przeworski @xcite , the shortest geodesic in closed orientable hyperbolic 3-manifolds with volume less than 0.94274 must have length greater than 0.09 , corresponding to a lower bound on @xmath2 of @xmath10 . an important point for our purposes here
is that it can be shown that there are non - zero lower bounds to lengths of shortest geodesics in any set of closed orientable hyperbolic 3-manifolds , whose volumes are smaller than that of the first orientable cusped manifold by any @xmath11 @xcite . to study an important consequence of these results for cosmology ,
let us assume that the universe can be modelled by a @xmath12-manifold @xmath13 , with a locally hyperbolic isotropic and homogeneous friedmann - lematre - robertson - walker metric in the standard form [ flrw1 ] ds^2 = -c^2dt^2 + r^2 ( t ) .
furthermore , let the @xmath1-space be a multiply connected compact quotient manifold of the form @xmath14 , where @xmath15 is a discrete group of isometries of @xmath16 acting freely on @xmath16 .
now in order to have @xmath17 , we consider a survey of depth up to the redshift @xmath5 . in this cosmological setting the depth of the survey expressed in units of the curvature radius ( @xmath18 ) is given by is identified with the curvature radius of the spatial section of the universe at time @xmath19 , and thus @xmath20 can be interpreted as the distance of any point with coordinates @xmath21 to the origin of coordinates ( in @xmath16 ) , in units of curvature radius , which is a natural unit of length and suitable for measuring areas and volumes . throughout this letter
we shall use this natural unit . ]
@xmath22^{-1/2 } dx \ ; , \ ] ] where the current content of the universe is taken to be dust ( of density @xmath23 ) plus a cosmological constant @xmath24 , with @xmath25 , @xmath26 , @xmath27 , and where the index @xmath28 denotes evaluation at present time . now to examine the consequences of the above theorem in the light of recent cosmological observations ,
recall that as @xmath8 , the curvature radius increases , resulting in a decrease in @xmath4 , and as a result the set of topologies that would be observationally detectable would have to possess @xmath2 approaching zero in order to ensure @xmath29 .
the above mathematical results , with their lower bounds on @xmath2 , have the consequence that the topology of small hyperbolic universes would be undetectable for values of @xmath0 close enough to one .
the important point being that the range of @xmath0 for which the topology of such universes are undetectable turns out to be within the range of values of @xmath0 allowed by recent observations , and particularly those suggested by the inflationary scenarios . to quantify this ,
we proceed in the following way .
the above theorem gives the lower bound on @xmath2 in a set of small volume manifolds ( those manifolds with volume less than 0.94274 , which includes at least the weeks manifold ) to be 0.045 .
this allows bounds to be imposed on the ranges of cosmological parameters for which the topology is undetectable .
we note that 0.045 is the best estimate available at present and the true lower bound on @xmath2 may be greater .
a way of dealing with this possibility is to consider the first 51 smallest manifolds of the hodgson - weeks census of closed hyperbolic manifolds .
this set contains all manifolds of the census with volumes smaller than the volume of the first cusped manifold .
we note that the volumes of the last 7 manifolds in this set differ from the volume of the first cusped hyperbolic orientable manifold by a factor which is smaller than @xmath30 , thus giving an idea of the size of the above mentioned @xmath31 for the set of manifolds considered here .
the manifold in this set with the lowest @xmath32 is the eighteenth manifold in the census , denoted by @xmath33 , with volume 1.58865 .
figure 1 gives a plot of the solution curve of equation @xmath34 in the @xmath35@xmath36 plane for @xmath37 and @xmath38 , where a survey of depth @xmath39 ( corresponding to the redshift of the surface of last scattering , cmbr ) was used .
also included in this figure is a dashed rectangular box , representing the relevant part ( for our purposes here ) of the hyperbolic region of the parameter space ( @xmath40 $ ] and @xmath41 $ ] ) given by recent observations @xcite .
for each value of @xmath2 undetectablilty is ensured for the values of cosmological parameters ( region in the @xmath35@xmath42 plane ) which lie above the corresponding curve .
thus for @xmath43 , all closed orientable hyperbolic manifolds ( universes ) with volumes less than 0.94274 would have undetectable topology , if the total density @xmath0 turned out to be higher than @xmath44 . on the other hand , for @xmath45 , the topology of none of the 51 manifolds of the census ( whose volumes range from 0.94271 to 2.02988 ( corresponding to the manifold @xmath46 ) , and which includes the weeks manifold ) would be detectable , if the total density @xmath0 turned out to be higher than @xmath47 . clearly , the actual precise bounds on @xmath0 depend on the precise value of the @xmath48 employed .
the resulting allowed changes in the bounds on @xmath0 due to employing other values in the observed range of @xmath48 are , however , small , as can be seen from figure 1 .
similarly , one can easily find the corresponding ranges of @xmath0 for any other particular manifold or finite set of small manifolds .
the important point is the existence of a lower bound on @xmath2 for any finite set of small volume manifolds ( smaller , by any @xmath11 , than the volume of the first cusped orientable hyperbolic manifold ) , which in turn gives lower bounds on @xmath49 , such that the topology of the universe is not detectable with methods based on the search for pattern repetitions .
, cornish _ et al . _
@xcite have argued that even for these large volume hyperbolic manifolds @xmath3 , it is very unlikely that the earth is in a region whose closed geodesics are as short as the shortest closed geodesics of @xmath3 .
thus , according to @xcite the chances of detecting the topology of these ( large ) hyperbolic universes are also very low according to recent observations ( @xmath50 ) , even if cmbr is used . ] to conclude , if it turns out , as suggested by inflationary scenarios , that @xmath0 is very close to one , then our results are significant in implying that the topology of small hyperbolic universes , suggested , for example , by some arguments based on quantum cosmology , can be undetectable using pattern repetitions .
this motivates the development of new strategies for looking for the topology of the universe , not based on the observation of repeated patterns . |
the core object of our investigation is an _
e. coli _ network representation of its combined gene regulation and metabolism , which can be thought of as functionally divided into three _ domains _ : the representation captures both gene regulatory and metabolic processes , with these processes being connected by an intermediate layer that models both , the enzymatic influence of genes on metabolic processes , as well as signaling - effects of the metabolism on the activation or inhibition of the expression of certain genes .
the underlying interaction graph @xmath1 with its set of nodes ( vertices ) @xmath2 and links ( edges ) @xmath3 consists of three interconnected subgraphs , the gene regulatory domain @xmath4 , the interface domain @xmath5 and the metabolic domain @xmath6 . from the functional perspective , @xmath7 is the union of gene regulatory ( @xmath4 ) and metabolic processes ( @xmath6 ) , and their interactions and preparatory steps form the interface ( @xmath5 )
. figure [ fig : sketch ] shows a sketch of the network model . a sketch of the domain organization of the integrative _ e. coli _ network .
the regulatory domain , @xmath4 , at the top is connected to the metabolic domain , @xmath6 , shown at the bottom via a protein - interface layer , @xmath5 .
the figure shows the number of vertices within and the number of edges within and between the domains . for illustrative purposes snapshots of the largest weakly - connected components of @xmath4 , @xmath5 and @xmath6
are included in the figure . ]
the integrated network representation has been assembled based on the database ( version 18.5 ; @xcite ) which offers both data about metabolic processes and ( gene ) regulatory events incorporated from the corresponding regulondb release 8.7 @xcite .
the extensive metadata allows for the assignment of the vertices to one of the three functional domains .
details of this process and a detailed characterization of the resulting model will be described elsewhere @xcite .
the corresponding graph representation consists of @xmath8 vertices and @xmath9 directed edges .
since we are interested in the propagation of a signal between the domains , in the following we will refer to the domains of the source and target vertices of the edge @xmath10 as _ source domain _ , sd , and _ target domain _ , td , respectively .
the metadata can be used to assign properties to the nodes and edges of the graph beyond the domain structure , some of which are used in the following analysis , namely in the construction of the propagation rules of the system and of the randomization schemes .
we distinguish between biological categories of edges ( capturing the diverse biological roles of the edges ) and the logical categories ( determining the rules of the percolation process ) . according to their biological role in the system , both vertices and edges
are assigned to a _
biological _ category ; we abbreviate the _ biological category of a vertex _ as bcv and the _ biological category of an edge _ as bce ( for details see supplementary materials ) .
each of the eight bces can then be mapped uniquely to one of only three _ logical categories of an edge _ , lce , @xmath11 which are of central importance for the spreading dynamics in our system : c , conjunct : : the target vertex of an edge with this logical ` and ` property depends on the source node , i.e. , it will fail once the source node fails . for example
, for a reaction to take place , all of its educts have to be available .
d , disjunct : : edges with this logical ` or ` property are considered redundant in the sense that a vertex only fails if the source vertices of all of its incoming d - edges fail . for instance , a compound will only become unavailable once all of its producing reactions have been canceled .
r , regulation : : edges of this category cover @xmath12 different kinds of regulatory events ( described in detail in the supplementary materials ) .
as shown below , in terms of the propagation dynamics we treat these edges similar to the conjunct ones .
next we describe the dynamical rules for the propagation of an initial perturbation in the network in terms of the logical categories of an edge ( lce ) , which distinguish between the different roles a given edge has in the update of a target vertex .
every vertex is assigned a boolean state variable ; since we intend to mimic the propagation of a perturbation ( rather than simulate a trajectory of actual biological states ) we identify the state @xmath13 with _ not yet affected by the perturbation _ while the state @xmath14 corresponds to _ affected by the perturbation_. we stress that the trajectory @xmath15 does not correspond to the time evolution of the abundance of gene products and metabolic compounds , but the rules have been chosen such that the final set of affected nodes provides an estimate of all nodes potentially being affected by the initial perturbation .
a node not in this set is topologically very unlikely of being affected by the perturbation at hand ( given the biological processes contained in our model ) .
a stepwise update can now be defined for vertex @xmath16 with in - neighbours @xmath17 in order to study the spreading of perturbations through the system by initially switching off a fraction @xmath18 of vertices : @xmath19{.5\columnwidth}{$ \left[\sum_j c_{ij } ( 1-\sigma_j ) = 0\right ] \wedge \ ; \left[\sum_j d_{ij } \sigma_j > 0 \vee \sum_j d_{ij}=0 \right ] \;\wedge \ ; \left[\sum_j |r_{ij}| ( 1-\sigma_j ) = 0\right ] $ } \\ 0 & \text{otherwise } \end{cases}\end{aligned}\ ] ] where @xmath20 is @xmath13 if @xmath21 is connected to @xmath22 via a c - link and @xmath14 otherwise ; @xmath23 and @xmath24 are defined analogously .
thus , a vertex will be considered unaffected by the perturbation if none of its in - neighbours connected via either a @xmath25 or an @xmath26 edge have failed ( regardless of the sign of the regulatory interaction ) , and at least one of its in - neighbours connected via a @xmath27 edge is still intact . with an additional rule
it is ensured that an initially switched off vertex stays off .
the choice of the update rules ensures that the unperturbed system state is conserved under the dynamics : @xmath28 . as a side remark ,
the spreading of a perturbation according to the rules defined above could also be considered as an epidemic process with one set of connections with a very large , and a second set of connections with a very low probability of infection @xcite .
in systems which can be described without explicit dependencies between its constituents but with a notion of functionality that coincides with connectivity , percolation theory is a method of first choice to investigate the system s response to average perturbations of a given size that can be modelled as failing vertices or edges @xcite the fractional size of the giant connected component as a function of the occupation probability @xmath29 of a constituent typically vanishes at some critical value @xmath30 , the percolation threshold . in the following
, we will mostly use the complementary quantity @xmath31 so that @xmath32 can be interpreted as the critical size of the initial attack or perturbation of the system .
the strong fluctuations of the system s responses in the vicinity of this point can serve as a proxy for the percolation threshold , which is especially useful in finite systems in which the transition appears smoothed out . in our analysis , the susceptibility @xmath33 , where @xmath34 is the size of the largest cluster , is used @xcite . upon the introduction of explicit dependencies between the system s constituents ,
the percolation properties can change dramatically .
the order parameter no longer vanishes continuously but typically jumps at @xmath30 in a discontinuous transition @xcite as cascades of failures fragment the system .
a broader degree distribution now enhances a graph s vulnerability to random failures , in opposition to the behavior in isolated graphs @xcite .
details of the corresponding theoretical framework have been worked out by parshani _ _ et al . _ _
@xcite , son _ _ et al . _
@xcite , baxter _
_ et al . _
_ @xcite and more recently the notion of networks of networks has been included @xcite .
there have also been attempts to integrate this class of models into the framework of multilayer networks @xcite .
in addition to random node failure other procedures for initial node removal have been explored , e.g. , node removal with respect to their degree ( targeted attacks ) @xcite .
currently , two notions of _ localized _ attacks have been described .
attacks of the first sort are defined on spatially embedded networks and are local with respect to a distance in this embedding , i.e. in a geographical sense @xcite .
the second approach considers locality in terms of connectivity : around a randomly chosen seed , neighbours are removed layer by layer @xcite . in contrast , as described below in our approach , attacks are localized with respect to the three network domains , while within the domains nodes are chosen randomly .
at this point we would like to shortly comment on the applicability of the mathematical concepts of interdependent networks to real - world data .
aiming at analytical tractability , typical model systems need to choose a rather high level of abstraction .
while certainly many systems can be accurately addressed in that way , we argue that especially in the case of biological systems the theoretical concepts can require substantial adjustment to cover essential properties of the system at hand .
when asking for the systemic consequences of interdependency , the distinction between several classes of nodes and links may be required .
effectively , some classes of links may then represent simple connectivity , while others can rather be seen as dependence links . in biology ,
such dependencies are typically mediated by specific molecules ( e.g. , a small metabolite affecting a transcription factor , or a gene encoding an enzyme catalyzing a biochemical reaction ) . such implementations of dependence links are no longer just associations and it is hard to formally distinguish them from the functional links . in contrast to the explicitly alternating percolation and dependency steps in typical computational models in which the decoupling of nodes from the largest component yields dependent nodes to fail , in our directed model both , connectivity and dependency links are evaluated in every time step and ( apart from nodes failing due to dependency ) only fully decoupled vertices cause further dependency failures .
due to the functional three - domain partition of our _ e. coli _ gene regulatory and metabolic network reconstruction , we have the possibility to classify perturbations not only according to their size , but also with respect to their localization in one of the domains comprising the full interdependent system , thereby enabling us to address the balance of sensitivity and robustness of the interdependent network of gene regulation and metabolism . here
we introduce the concept of network response to localized perturbations analysis .
this analysis will reveal that perturbations in gene regulation affect the system in a dramatically different way than perturbations in metabolism .
thus we study the response to _ localized _ perturbations .
we denote such perturbations by @xmath35 , where @xmath36 is the domain , in which the perturbation is localized ( @xmath37 , with @xmath38 representing the total network @xmath7 , i.e. , the case of non - localized perturbations ) .
the perturbation size @xmath31 is measured in fractions of the total network size @xmath39 .
a perturbation @xmath40 thus is a perturbation in the metabolic domain with ( on average ) @xmath41 nodes initially affected .
note that sizes @xmath18 of such localized perturbations are limited by the domain sizes , e.g. , @xmath42 for a perturbation in the gene regulatory domain .
after the initial removal of a fraction @xmath18 of the vertices from the domain @xmath36 the stepwise dynamics described above will lead to the deactivation of further nodes and run into a frozen state @xmath43 in which only a fraction @xmath44 of the vertices are unaffected by the perturbation ( i.e. , are still in state @xmath13 ) .
in addition to being directly affected by failing neighbors , in the process of network fragmentation nodes may also become parts of small components disconnected from the network s core , and could in this sense be considered non - functional ; we therefore also monitor the relative sizes of the largest ( weakly ) connected component in the frozen state , @xmath45 , for different initial perturbation sizes . in the limit of infinite system size we could expect a direct investigation of @xmath45 as a function of @xmath18 to yield the critical perturbation size @xmath32 at which @xmath46 vanishes . in our finite system
, however , we have to estimate @xmath47 ; following radicchi and castellano @xcite and radicchi @xcite we measure the fluctuation of @xmath45 which serves as our order parameter and look for the peak position of the susceptibility @xmath48 as a function of parameter @xmath18 in order to estimate the transition point from the finite system data .
supplementary figure [ fig : overview ] schematically illustrates the analysis .
in order to interpret the actual responses of a given network to perturbations , one usually contrasts them to those of suitably randomized versions of the network at hand .
thereby , the often dominant effect of the node degree distribution of a network can be accounted for and the effects of higher - order topological features that shape the response of the network to perturbations can be studied systematically . the same is true for the localized perturbation response analysis introduced here .
in fact , due to the substantially larger number of links from gene regulation to metabolism ( both , directly and via the interface component of the interdependent network ) than from metabolism to gene regulation we can already expect the response to such localized perturbations to vary . here
we employ a sequence of ever more stringent randomization schemes to generate sets of randomized networks serving as null models for the localized perturbation response analysis . in all of the four schemes the edge - switching procedure introduced by maslov and sneppen @xcite
is employed which conserves the in- and out - degrees of all vertices .
our most flexible randomization scheme ( domain ) only considers the domains of the source and target vertices of an edge ( sd and td ) : only pairs of edges are flipped which share both , the source and the target domain ( e.g. , both link a vertex in the metabolic domain to a vertex in the interface ) .
the remaining three randomization schemes all add an additional constraint .
the domain_lce randomization further requires the edges to be of the same logical categories of an edge ( i.e. , c , d , or r ) , while the domain_bcv scheme only switches edges whose target vertices also share the same biological category of a vertex , bcv .
the strictest randomization , domain_bce , finally , only considers edges with , additionally , the same biological category of an edge , bce
. a tabular overview of the four schemes is given in supplementary table [ tab : randschemes ] .
the main feature of our reconstructed network , the three - domain structure based on the biological role of its constituents , allows us to study the influence of localizing the initial perturbation .
thus , although we will not focus on ( topological ) details of the graph here ( which will be presented elsewhere @xcite ) , already from the vertex and edge counts in figure [ fig : sketch ] we see that the domains are of different structure . while the regulatory and the metabolic subgraphs , @xmath4 and @xmath6 have average ( internal ) degrees of about @xmath49 and @xmath50 , the interface subgraph , @xmath5 , is very sparse with @xmath51 and we can expect it to be fragmented .
hence , in the following we decide to only perturb in @xmath4 and @xmath6 . in a first step we sample some full cascade trajectories in order to check our expectation of different responses of the system to small perturbations applied in either @xmath4 or @xmath6 ; two rather large values of @xmath18
are chosen and the raw number of unaffected nodes is logged during the cascade .
indeed , already this first approach implies a different robustness of the gene regulatory and the metabolic domains in terms of the transmission of perturbation cascades to the other domains .
cascades seeded in the metabolic domain of the network tend to be rather restricted to this domain , while the system seems much more susceptible to small perturbations applied in the gene regulatory domain .
this effect can be seen both in the overall sizes of the aggregated cascades as well as in the domain which shows the largest change with respect to the previous time step , which we indicate by black markers in figure [ fig : states ] .
more sample trajectories are shown in the supplementary materials and , although they illustrate occasionally large fluctuations between the behaviors of single trajectories , they are consistent with this first observation . they also show that considerably larger metabolic perturbations are needed for large cascades and back - and - forth propagation between domains to emerge .
after this first glance at the system we aim for a more systematic approach and apply our analysis as described above : we compute cascade steady - states @xmath43 but now we choose the largest ( weakly ) connected component @xmath45 as the order parameter and compute the susceptibility according to equation @xmath52 , the peak - position of which , when considered as a function of @xmath31 , we use as a proxy for the perturbation size at which the interdependent system breaks down .
the results for different initially perturbed domains illustrate that , indeed , a considerably lower @xmath30 ( i.e. , larger critical perturbation size @xmath47 ) is estimated in the case of metabolic perturbations compared to regulatory or non - localized ones ( figure [ fig : suscept ] , panel a ) . for each point
we average @xmath53 runs for the corresponding set of parameters . in order to assess whether the above - described behavior is due to specific properties of the network we use the sets of randomized graphs
for each of the four randomization schemes we prepared @xmath53 graph instances and repeated the analysis for each of them as done before for the single original graph .
the corresponding results for the susceptibility ( figure [ fig : suscept ] , panels b e ) yield two major observations : firstly , metabolic perturbations still lead to , albeit only slightly , higher @xmath32 estimates ( with exception of domain randomization ) .
thus , the system s tendency to be more robust towards metabolic perturbations is largely preserved .
secondly , we see that overall the original network seems to be much more robust than the randomized networks ; very small perturbations are sufficient to break the latter ones .
the robustness of the original graph , thus , can not be solely due to the edge and vertex properties kept in the randomization schemes .
the susceptibility of the integrative _ e. coli _ model and its randomized versions for perturbations in different domains .
the results for the original graph are shown in the _ top _ frame , while the subsequent frames show results for the four different randomization schemes with the least strict on top and the strictest one at the bottom .
the original system is more robust than its randomized versions ; perturbations in the metabolism consistently need to be larger than in the regulatory part in order to reach the maximum susceptibility . ] finally , let us focus on the practical aspect of these findings . beyond the careful statistical analysis described above , a quantity of practical relevance is the average size of the unaffected part of the system under a perturbation . for this purpose
, we examine the fractions of unaffected vertices , @xmath44 , after cascades emanating from perturbations of different sizes and seeded in different domains , regardless of the resulting component structure and for both , the original graph and the shuffled ones ( figure [ fig : frac ] ) .
mean fractions ( averaged over @xmath54 runs ) of unaffected vertices after initial perturbations of different sizes seeded in the gene regulatory ( filled green dots ) and in the metabolic domain ( filled red squares ) .
the curves at the bottom show the averaged results over @xmath53 shuffles for each randomization scheme . ]
the number of unaffected vertices for the real network is much larger than for all four randomization schemes , suggesting a strong overall robustness of the biological system .
distinguishing , however , between the metabolic and the gene regulatory components reveals that the metabolic part is substantially more robust than the regulatory part ( for not too large initial perturbations , @xmath55 ) .
we investigated the spreading of perturbations through the three domains of a graph representation of the integrated system of _
e. coli _ s gene regulation and metabolism .
our results quantify the resulting cascading failures as a function of size and localization of the initial perturbation .
our findings show that the interdependent network of gene regulation and metabolism unites sensitivity and robustness by showing different magnitudes of damage dependent on the site of perturbation .
while the interdependent network of these two domains is in general much more robust than its randomized variants ( retaining domain structure , degree sequence , and major biological aspects of the original system ) , a pronounced difference between the gene regulatory and metabolic domain is found : small perturbations originating in the gene regulatory domain typically trigger far - reaching system - wide cascades , while small perturbations in the metabolic domain tend to remain more local and trigger much smaller cascades of perturbations . in order to arrive at a more mechanistic understanding of this statistical observation ,
we estimated the percolation threshold of the system , @xmath30 , and found that it is much lower ( i.e , larger perturbations , @xmath32 , are required ) for perturbations seeded in the metabolic domain than for those applied to the gene regulatory domain .
this is in accordance with the intuition that the metabolic system is more directly coupled to the environment ( via the uptake and secretion of metabolic compounds ) than the gene regulatory domain .
the distinct perturbation thresholds therefore allow for implementing a functionally relevant balance between robustness and sensitivity : the biological system can achieve a robustness towards environmental changes , while via the more sensitive gene regulatory domain it still reacts flexibly to other systemic perturbations . discovering this design principle of the biological system required establishing a novel method of analyzing the robustness of interdependent networks , the network response to localized perturbations : an interdependent network can have markedly different percolation thresholds , when probed with perturbations localized in one network component compared to another .
lastly , we would like to emphasize that the application of the theoretical concepts of interdependent networks to real - life systems involves several non - trivial decisions : in the vast majority of ( theoretical ) investigations , two definitions of interdependent networks coincide : the one derived from a distinction between dependency links and connectivity - representing links and the one based on two functionally distinguishable , but interconnected subnetworks . here
we have three classes of nodes : those involved in gene regulation , metabolic nodes , and nodes associated with the ( protein ) interface between these two main domains .
these nodes are interconnected with ( functionally ) different classes of links .
these link classes are necessary to define meaningful update rules for perturbations . as a consequence , the notion of dependency links vs. connectivity links is no longer applicable .
we expect that such adjustments of the conceptual framework will often be required when applying the notion of interdependent networks to real - life systems .
as mentioned above , one major task when dealing with biological data is to abstract from the minor but keep the essential details ; we have outlined that in this study we chose to keep a rather high level of detail . with only incomplete information available ,
a challenge is to find the right balance between radical simplifications of systemic descriptions and an appropriate level of detail still allowing for a meaningful evaluation of biological information .
here we incorporate high level of detail in the structural description , distinguishing between a comparatively large number of node and link types .
this rich structural description , together with a set of update rules motivated by general biological knowledge , allows us to assess the dynamical / functional level with the comparatively simple methods derived from percolation theory .
an important question is , whether the analysis of the fragmentation of such a network under random removal of nodes can provide a reliable assessment of functional properties , since the response of such a molecular network clearly follows far more intricate dynamical rules than the percolation of perturbations can suggest .
a future step could include the construction of a boolean network model for the full transcriptional regulatory network and the connection of this model to flux predictions obtained via flux balance analysis , a first attempt of which is given in samal and jain @xcite ( where the model of covert _
_ et al . _
_ @xcite with still fewer interdependence links has been used ) .
our perturbation spreading approach might help bridging the gap between theoretical concepts from statistical physics and biological data integration : integrating diverse biological information into networks , estimating response patterns to systemic perturbations and understanding the multiple systemic manifestations of perturbed , pathological states is perceived as the main challenge in systems medicine ( see , e.g. , bauer _
_ et al . _
_ @xcite ) .
concepts from statistical physics of complex networks may be of enormous importance for this line of research @xcite . while the simulation of the full dynamics is still problematic as our knowledge of the networks is still incomplete , our present strategy extracts first dynamical properties of the interdependent networks . at a later time point
, we can expect qualitatively advances from full dynamical simulations , however , dependent on the quality of the data sets . on the theoretical side
, future studies might shift the focus onto recasting the system into an appropriate spreading model , e.g. , in the form of an unordered binary avalanche @xcite , or as an instance of the linear threshold model @xcite with a set of links with a very high and a second set with a very low transmission probability ( c / r and d - links , respectively ) .
radicchi @xcite presents an approach for the investigation of the percolation properties of finite size interdependent networks with a specific adjacency matrix with the goal of loosening some of the assumptions underlying the usual models ( e.g. , infinite system limit , graphs as instances of network model ) . while this formalism allows for the investigation of many real - world systems there are still restrictions as to the possible level of detail . in our special case ,
for instance , a considerable amount of information would be lost if the system was restricted to vertices with connections in both the c / r- and d - layers .
the existence of different percolation thresholds for localized perturbations in interdependent networks may reveal itself as a universal principle for balancing sensitivity and robustness in complex systems .
the application of these concepts to a wide range of real - life systems is required to make progress in this direction .
s.b . and m.h .
acknowledge the support of deutsche forschungsgemeinschaft ( dfg ) , grants bo 1242/6 and hu 937/9 .
m.h . and s.b .
designed and supervised the study ; d.k . and a.g .
performed the reconstruction , simulations and analyses ; and d.k .
, s.b . and m.h
. wrote the manuscript .
the authors declare no competing financial interests .
10 url # 1`#1`urlprefix[2]#2 [ 2][]#2 . _ _ * * , ( ) . , & . _ _ * * , ( ) . .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , , , &
_ _ * * , ( ) . & .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , &
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . .
_ _ * * , ( ) . , , , &
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) . , , & .
_ _ * * ( ) . & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
, , & ( ) . .
_ et al . _ . _
_ * * , ( ) . .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) .
, , & . _ _ * * , ( ) . , , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ _ * * , ( ) . , , , & .
_ _ * * , ( ) .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . .
_ _ * * , ( ) . & .
_ _ * * , ( ) .
_ et al . _ . _
, & _ _ * * , ( ) . .
_ _ * * , ( ) . & .
_ _ * * , ( ) . .
_ _ * * , ( ) .
* supplementary material to + the interdependent network of gene regulation and metabolism is robust where it needs to be * here , we provide supplementary information to the main manuscript : 1 . a description of the analysis including a schematic overview , 2 .
an explanation of the notation of vertex and edge categories , 3 .
tables summarizing the four custom - built randomization schemes , the quantities and parameters shown / used in the particular figures as well as for the introduced vertex and edge categories 4 . a collection of sample trajectories , @xmath56 , to be compared to fig .
[ fig : states ] in the main manuscript .
the network response to localized perturbations analysis presented here ( see figure [ fig : overview ] ) , is a multi - step method entailing several runs of the perturbation algorithm , @xmath57 , and the statistical evaluation of these runs .
more precisely , for each analysis 500 runs of the perturbation algorithm have been performed and the set of the remaining vertices have been evalueted , e.g. , in terms of the susceptibility , @xmath58 ( part a in figure [ fig : overview ] ) . for a single run of the perturbation algorithm , @xmath35 , a domain @xmath36 has to be chosen where a perturbation of size @xmath31 ( measured as a fraction of vertices ) will be applied , @xmath59 based on @xmath36 and @xmath18 the set of initially perturbed vertices , @xmath60 , is randomly selected .
the state of the system can also be described as a vector of boolean state variables , @xmath61 , @xmath62 where 0 denotes a perturbed vertex and 1 an unaffected one .
running the perturbation dynamics described in the main manuscript will ( probably ) cause the failure of further vertices resulting in a time series of affected vertices , @xmath63 ( or , equivalently , to the trajectory @xmath64 ) .
the size of the affected network after @xmath65 propagation steps can be described as @xmath66 from the set of affected vertices in the in the asymptotic regime , @xmath67 , the size of the unaffected network , @xmath68 , and the size of the largest ( weakly ) connected component ( @xmath69 ) of the unaffected network , @xmath70 are computed ( figure [ fig : overview ] , part b ) , @xmath71 \right|\text{. } \end{aligned}\ ] ] randomized networks ( we used sets of @xmath53 instances for each of the randomization schemes ) can be passed to the algorithm instead ( figure [ fig : overview ] , part c ) .
( t1 ) original network @xmath7 ; ( p1 ) @xmath72 ; ( t2 ) statistical evaluation : ; ( t3 ) * susceptibility @xmath58 + * variance @xmath73 $ ] ; ( d ) run perturbation dynamics ; ( b1 ) select domain @xmath36 and perturbation size @xmath18 + ; ( t4 ) at ( b1.south ) @xmath74 ; ( t4a ) ; ( b2 ) @xmath75 ; ( b3 ) size of potentially unaffected network @xmath76 ; ( b3a ) ; ( b4 ) size of the largest connected component of the unaffected network , @xmath70 ; ( b4a ) ; let 1=(b1.west),2=(b4.south ) in coordinate ( p2c ) at ( 1,2 ) ; ( p2 ) at ( p2c ) @xmath72 ; background ( box ) ; let 1=(t1.west),2=(box.south ) in coordinate ( t5c ) at ( 1,2 ) ; ( t5 ) randomization scheme @xmath16 ; ( t6 ) set of randomized networks @xmath77 ; ( p2 ) @xmath72 ; ( t7 ) statistical evaluation : ; ( t8 ) * susceptibilities @xmath78 + * variances @xmath79 $ ] ; ( l1 ) * ( a ) * ; let 1=(l1),2=(box.north ) in coordinate ( l2c ) at ( 1,2 ) ; ; ; ( p1.south west ) to ( box.north west ) ; ( p1.south east ) to ( box.north east ) ; ( t4a ) to ( t4 ) ; ( t1 ) to ( p1 ) ; ( p1 ) to node[above ] multiple runs ( t2 ) ; ( @xmath80 ) to ( @xmath81 ) to ( d.west ) ; ( b1.north ) to ( d.west ) ; ( d.east ) to ( b2 ) ; ( b2 ) to ( b3 ) ; ( b3 ) to ( b4 ) ; ( b3 ) to ( b3a ) ; ( b4 ) to ( b4a ) ; ( t5 ) to ( t6 ) ; ( t6 ) to ( p2 ) ; ( p2 ) to node[above ] multiple runs ( t7 ) ;
the following notation concerning vertices @xmath82 and edges @xmath83 and their properties have been used . in our integrated network model a vertex
is characterized by its biological function yielding seven unique _ biological categories of a vertex _ ( bcvs ) , reaction ( @xmath84 ) , compound ( @xmath85 ) , gene ( @xmath86 ) , protein monomer ( @xmath87 ) , protein - protein complex ( @xmath88 ) , protein - compound complex ( @xmath89 ) , and protein - rna complex ( @xmath90 ; see table [ tab : bcvs ] ) , @xmath91 introducing an additional vertex classification facilitates the assignment to one of the three functional domains as well as the edge characterization .
domain - related categories of a vertex _ ( dcvs ) are eight - fold : gene ( @xmath92 ) , protein ( @xmath29 ) , complex ( @xmath93 ) , enzyme ( @xmath94 ) , reaction ( @xmath95 ) , compound ( @xmath96 ) , educt ( @xmath97 ) , product ( @xmath98 ) , @xmath99 while the categories @xmath92 and @xmath95 are a one - to - one translations of the corresponding bcvs , i.e. , @xmath86 and @xmath84 , the domain - related category @xmath96 only comprises vertices of bcv @xmath85 but the inverse does not hold .
the remaining five categories are ambiguous assignments for both bcvs to dcvs and vice versa .
the complete mapping of bcvs onto dcvs is given in table [ tab : dcvs ] .
an edge is characterized by its source and target vertices , @xmath100 and @xmath101 , and their corresponding domains ( source domain , sd and target domain , td ) , as well as the edge s _ logical category _ and its _ biological category _ ; an edge is given by @xmath102 thus determining sd and td .
the _ logical category of an edge _ ( lce ) determines , qualitatively speaking , whether a perturbation will propagate along this edge via a logical ` and ` or a logical ` or ` ; the three categories are conjunct ( @xmath25 ) , disjunct ( @xmath27 ) and regulation ( @xmath26 ) , @xmath103 as an illustration of the potential linkages , two case examples are presented in figure [ fig : biologicalcaseexamples ] .
\(e ) ; ( cd ) ; ( c ) ; ( d ) ; ( c ) ; ( d ) ; ( succ ) ; ( coa ) ; ( atp ) ; ( succcoa ) ; ( adp ) ; ( pi ) ; ( c ) to node[right , pos=0.7 ] d ( c ) ; ( c.south ) to node[left , pos=0.2 ] 2 node[above , pos=0.8 ] c ( cd.north ) ; ( d ) to node[left , pos=0.7 ] d ( d ) ; ( d.south ) to node[right , pos=0.2 ] 2 node[above , pos=0.8 ] c ( cd.north ) ; ( cd ) to node[right , pos=0.75 ] d ( e ) ; ( succ.east ) to node[above right , yshift=-0.5em ] c ( e ) ; ( coa.east ) to node[above right , yshift=-0.2em ] c ( e ) ; ( atp.east ) to node[above right , xshift=-0.2em ] c ( e ) ; ( e ) to node[above right , xshift=-0.2em ] d ( succcoa.west ) ; ( e ) to node[above right , yshift=-0.2em ] d ( adp.west ) ; ( e ) to node[above right , yshift=-0.5em ]
d ( pi.west ) ; ( adp.30 ) to[out=5,in=0,looseness=3.5 ] node[above , pos=0.96 ] r ( cd.43 ) ; ( ch ) ; ( e1 ) ; ( i ) ; ( e2 ) ; \(x ) ; ( menf ) ; ( menf ) ; ( entc ) ; ( entc ) ; ( menf ) to node[right , pos=0.7 ] d ( menf ) ; ( menf ) to node[right , pos=0.25 ] 2 node[right , pos=0.8 ] c ( x ) ; ( x ) to node[right , pos=0.75 ] d ( e1 ) ; ( ch.east ) to node[above , pos=0.7 ] c ( e1 ) ; ( e1 ) to node[above , pos=0.8 ] d ( i.west ) ; ( entc ) to node[right , pos=0.7 ]
d ( entc ) ; ( entc ) to node[right , pos=0.75 ] d ( e2 ) ; ( ch.east ) to node[below , pos=0.7 ] c ( e2 ) ; ( e2 ) to node[below , pos=0.8 ] d ( i.west ) ; let 1=(coa.west),2=(menf.north ) in coordinate ( l1 ) at ( 1,2 ) ; ; let 1=(ch.west),2=(menf.north ) in coordinate ( l2 ) at ( 1,2 ) ; ; the biological categories of an edge ( bces ) are derived from combinations of the domain - related categories of a vertex ( dcvs ) , plus transport ( @xmath65 ) and regulation ( @xmath104 ) , @xmath105 the mapping of biological categories of edges onto logical categories of edges is given in table [ tab : bces ] .
.overview of the four custom - built randomization schemes in order of their strictness . in particular , the degree of freedeom is denoted by the possible pairs of edges available for the randomization and by the conserved quantities of the graph : source domain ( sd ) , target domain ( td ) , logical category of an edge ( lce ) , biological category of a vertex ( bcv ) , and biological category of an edge ( bce ) . [ cols="<,>,<",options="header " , ] lcl16em bces & lces & vertex linkages + gene @xmath106 protein ( @xmath107 ) & @xmath27 & ( g);(p)at(0.8,0);(g)to(@xmath108 ) ; + protein @xmath106 complex ( @xmath109 ) & @xmath25 & ( p);(c)at(0.8,0);(@xmath110)to(@xmath111 ) ; ( p);(c)at(0.8,0);(@xmath110)to(@xmath111 ) ; ( p);(c)at(0.8,0);(@xmath110)to(@xmath111 ) ; + compound @xmath106 complex ( @xmath112 ) & @xmath25 & ( c);(p)at(0.8,0);(c)to(@xmath111 ) ; + enzyme @xmath106 reaction ( @xmath113 ) & @xmath27 & ( p);(r)at(0.8,0);(@xmath110)to(r ) ; ( p);(r)at(0.8,0);(@xmath110)to(r ) ; + educt @xmath106 reaction ( @xmath114 ) & @xmath25 & ( p);(r)at(0.8,0);(@xmath110)to(r ) ; ( p);(r)at(0.8,0);(@xmath110)to(r ) ; ( p);(r)at(0.8,0);(@xmath110)to(r ) ; ( c);(r)at(0.8,0);(c)to(r ) ; + reaction @xmath106 product ( @xmath115 ) & @xmath27 & ( r);(c)at(0.8,0);(r)to(c ) ; ( r);(p)at(0.8,0);(r)to(@xmath108 ) ; ( r);(p)at(0.8,0);(r)to(@xmath108 ) ; ( r);(p)at(0.8,0);(r)to(@xmath108 ) ; + transport ( @xmath65 ) & @xmath25 & ( c1);(c2)at(0.8,0);(c1)to(c2 ) ; + regulation ( @xmath104 ) & @xmath26 & ( g1);(g2)at(0.8,0);(g1)to(g2 ) ; ( p);(g)at(0.8,0);(@xmath110)to(g ) ; ( p1);(p2)at(0.8,0);(@xmath116)to(@xmath117 ) ; ( p1);(p2)at(0.8,0);(@xmath116)to(@xmath117 ) ; ( p);(r)at(0.8,0);(@xmath110)to(r ) ; ( c);(g)at(0.8,0);(@xmath118)to(g ) ; ( p1);(p2)at(0.8,0);(@xmath116)to(@xmath117 ) ; ( p1);(p2)at(0.8,0);(@xmath116)to(@xmath117 ) ; ( p);(r)at(0.8,0);(@xmath110)to(r ) ; ( c);(g)at(0.8,0);(@xmath118)to(g ) ; ( c);(p)at(0.8,0);(@xmath118)to(@xmath108 ) ; ( c);(g)at(0.8,0);(@xmath118)to(g ) ; ( c);(g)at(0.8,0);(c)to(g ) ; ( c);(p)at(0.8,0);(c)to(@xmath108 ) ; ( c);(c)at(0.8,0);(c)to(@xmath111 ) ; ( c);(r)at(0.8,0);(c)to(r ) ; +
here , further sample trajectories are given similar to the ones in fig .
@xmath119 in the main manuscript . | the major biochemical networks of the living cell , the network of interacting genes and the network of biochemical reactions , are highly interdependent , however , they have been studied mostly as separate systems so far . in the last years
an appropriate theoretical framework for studying interdependent networks has been developed in the context of statistical physics .
here we study the interdependent network of gene regulation and metabolism of the model organism _ escherichia coli _ using the theoretical framework of interdependent networks .
in particular we aim at understanding how the biological system can consolidate the conflicting tasks of reacting rapidly to ( internal and external ) perturbations , while being robust to minor environmental fluctuations , at the same time . for this purpose
we study the network response to localized perturbations and find that the interdependent network is sensitive to gene regulatory and protein - level perturbations , yet robust against metabolic changes .
this first quantitative application of the theory of interdependent networks to systems biology shows how studying network responses to localized perturbations can serve as a useful strategy for analyzing a wide range of other interdependent networks .
a main conceptual approach of current research in the life sciences is to advance from a detailed analysis of individual molecular components and processes towards a description of biological _
systems _ and to understand the emergence of biological function from the interdependencies on the molecular level . supported by the diverse high - throughput omics technologies ,
the relatively recent discipline of systems biology has been the major driving force behind this new perspective which becomes manifest , for example , in the effort to compile extensive databases of biological information to be used in genome - scale models @xcite . despite its holistic game plan
, however , systems biology frequently operates on the level of subsystems : even when considering cell - wide transcriptional regulatory networks , as , e.g. , in a network motif analysis @xcite , this is only one of the cell s networks .
likewise , the popular approach to studying metabolic networks in systems biology , constraint - based modeling , accounts for steady - state predictions of metabolic fluxes of genome - scale metabolic networks @xcite , which again , is only one of the other networks of the cell . in the analysis of such large networks
, systems biology draws its tools considerably from the science of complex networks which provides a mathematical framework especially suitable for addressing interdisciplinary questions . combining the mathematical subdiscipline of graph theory with methods from statistical physics , this new field greatly contributed to the understanding of , e.g. ,
the percolation properties of networks @xcite , potential processes of network formation @xcite or the spreading of disease on networks @xcite . in the early 2000s , gene regulation and metabolism have been among the first applications of the formalisms of network biology @xcite . among the diverse studies of network structure for these systems , the most prominent ones on the gene regulatory side are the statistical observation and functional interpretation of small over - represented subgraphs ( network motifs ) @xcite and the hierarchical organization of gene regulatory networks @xcite . on the metabolic side
, the broad degree distribution of metabolic networks stands out @xcite , with the caveat , however , that currency metabolites ( like atp or h@xmath0o ) can severely affect network properties @xcite , as well as the hierarchical modular organization of metabolic networks @xcite . over the last decade , the field of complex networks moved its focus from the investigation of single - network representations of systems to the interplay of networks that interact with and/or depend on each other .
strikingly , it turned out that explicit interdependence between network constituents can fundamentally alter the percolation properties of the resulting _ interdependent networks _ , which can show a discontinuous percolation transition in contrast to the continuous behavior in single - network percolation @xcite .
it has also been found that , contrary to the isolated - network case , networks with broader degree distribution become remarkably fragile as interdependent networks @xcite .
however , this set of recent developments in network science still lacks application to systems biology . arguably , the most prominent representative of interdependent networks in a biological cell is the combined system of gene regulation and metabolism which are interconnected by various forms of protein interactions , e.g. , enzyme catalysis of biochemical reactions couples the regulatory to metabolic network , while the activation or deactivation of transcription factors by certain metabolic compounds provides a coupling in the opposite direction . although it is well - known that gene regulatory and metabolic processes are highly dependent on one another only few studies addressed the interplay of gene regulation and metabolism on a larger scale and from a systemic perspective @xcite .
the first two studies have aimed at finding consistent metabolic - regulatory steady states by translating the influence of metabolic processes on gene activity into metabolic flux predicates and incorporating high - throughput gene expression data .
this can be considered as an extension of the constraints - based modeling framework beyond the metabolic network subsystem . in the paper of samal and jain @xcite , on the other hand , the transcriptional regulatory network of _ escherichia coli _
( _ e .
coli _ ) metabolism has been studied as a boolean network model into which flux predicates can be included as additional interactions .
these models were first important steps towards integrating the subsystems of gene regulation and metabolism from a systems perspective .
the formalism of interdependent networks now allows us to go beyond these pioneering works on integrative models , by analyzing the robustness of the combined system in terms of the maximal effect a small perturbation can have on such interdependent systems .
in particular , the findings can be interpreted in the context of cascading failures and percolation theory .
we here undertake a first application of the new methodological perspective to the combined networks of gene regulation and metabolism in _
e. coli_. using various biological databases , particularly as the main core @xcite , we have compiled a graph representation of gene regulatory and metabolic processes of _
e. coli _ including a high level of detail in the structural description , distinguishing between a comparatively large number of node and link types according to their biological functionality .
a structural analysis of this compilation reveals that , in addition to a small set of direct links , the gene - regulatory and the metabolic domains are predominantly coupled via a third network domain consisting of proteins and their interactions .
figure [ fig : sketch ] shows this three - domain functional division .
details about the data compilation , the network reconstruction and the domain - level analysis are given in grimbs _
_ et al . _
_ @xcite .
this rich structural description , together with purpose - built , biologically plausible propagation rules allows us to assess the functional level with methods derived from percolation theory .
more precisely , we will investigate cascading failures in the three - domain system , emanating from small perturbations , localized in one of the domains . by network response to localized perturbations analysis
we will observe below that ( i ) randomized versions of the graph are much less robust than the original graph and ( ii ) that the integrated system is much more susceptible to small perturbations in the gene regulatory domain than in the metabolic one . |
it is a well known fact that the nonrelativistic nucleon - nucleus optical potential is nonlocal and energy dependent feschbach_ann , feschbach_book .
nevertheless , a common approach to studying the elastic @xmath0 scattering is the use of phenomenological local optical models , and a variation of the model parameters with the incident projectile energy can be taken as a sign of the presence of nonlocal effects in the scattering process @xcite .
for example , a recent work presented local and global optical potentials for neutron and proton elastic scattering from nuclei that fall in the mass range @xmath1 @xcite , while nucleon incident energies ranged from 1 kev to 200 mev .
the authors obtained excellent elastic angular distribution fits using local optical potentials each consisting of 20 fit parameters , however , the strength of the central real part showed the largest variation with incident energy .
another measure of nonlocality is given by the size of the perey damping factor , as will be discussed in the present study for the case of the presence of a gradient term in the optical potential .
such a gradient term , also denoted as velocity dependent term , is usually absent in conventional optical model potentials , but has been introduced in a previous study @xcite , and will be the object of further analysis in the present investigation .
the nonrelativistic optical potential contains several sources of nonlocalities .
one such nonlocality is due to the pauli exclusion principle and can be taken into account by antisymmeterizing the wave function as in the hartree - fock theory @xcite .
further , in the hartree - fock equation , the nonlocality due to exchange effects can be expressed in terms of a spatially variable mass @xcite .
more recently , the importance of accounting for the pauli principle at incident energies greater than 25 mev has been established @xcite . at low energies
the importance of the pauli exclusion principle has been investigated in collective - model coupled - channel calculations @xcite .
a second source of nonlocality is due to the coupling of the inelastic excitations to the nuclear ground state during the scattering process @xcite , that gives rise to a nonlocal dynamic polarization potential in the elastic channel .
since this nonlocality is difficult to take into account rigorously , it is usually accounted for by numerically coupling a few important inelastic channels to the elastic channel @xcite .
however , when the energy of the incident nucleon is low enough , the scattering process is strongly affected by the individual discrete states of the target nucleus . a recent study that employed the multichannel algebraic scattering theory ( mcas ) to explore the dynamic polarization potential fraser , canton for low incident energies also showed that the resulting optical potential was strongly nonlocal and energy dependent .
recently , we introduced a novel gradient term in the schrdinger equation describing the @xmath0 elastic scattering process .
we interpreted this term as a change in mass of the incident nucleon when moving inside the nuclear matter of the target nucleus @xcite .
the mass change could be a consequence of the nucleon - nucleon interaction being affected by the presence of surrounding nucleons , like a surface tension which is present only at the surface of the target nucleus .
this view is supported by the fact that our model , in ref .
@xcite , resulted in excellent fits to the @xmath2 elastic scattering data when the coefficient of the derivative term was assumed to be proportional to the gradient of the nuclear density , which is most important at the nuclear surface .
furthermore , our model reproduced well the large angle scattering minima that are usually associated with nonlocalities @xcite .
however , it is still not yet clear whether the gradient term also simulates other sources of nonlocal effects in the phenomenological optical model like , for examples , the pauli exclusion principle and coupling of the elastic channel to the inelastic ones .
we formulated the mass dependence by introducing a kisslinger type potential into the conventional optical model @xcite .
this is achieved by writing the schrdinger operator as @xmath3 where the resulting potential @xmath4 is velocity - dependent and has the form : @xmath5 .
\label{vdp}\end{aligned}\]]the velocity - dependent potential @xmath6 plays the role of a nonlocal potential @xmath7 .
in fact there is an equivalence between a velocity - dependent ( or momentum - dependent ) potential and a nonlocal potential @xmath7 @xcite .
this can be seen by interpreting the gradient term that acts on the wave function as the first term of a taylor series that displaces the wave function from point @xmath8 to a different location .
furthermore , the spatially variable effective mass @xmath9 is assumed to have the @xmath10where @xmath11 is an isotropic function of the radial variable @xmath12 that expresses the change of the projectile s mass with distance @xmath12 from the center of the target nucleus @xcite .
the notion of an effective mass has also been introduced in relation to relativistic optical models that give rise to significantly reduced effective masses serot . for example , the relativistic effects on the nucleon mean free path in the intermediate energy range were investigated in ref .
in addition , the idea of an effective mass was considered for the specification of a microscopic optical model potential ref .
@xcite . at this point
we stress that the gradient term in our model is different from the darwin term present in the nonrelativistic equivalent of the dirac - based relativistic formulation @xcite because , firstly , our effective mass is real while the darwin term is complex . secondly , in the nonrelativistic equivalent of the dirac formulation , the darwin term
is closely coupled to the spin orbit potential while our gradient term is independent of the spin orbit term @xcite .
furthermore , unlike the case for the relativistic models mentioned above , our gradient term is expressed as an effective mass that has a radial dependence , rather than in terms of an additional scalar potential that is added to the mass .
finally , the effective mass formulation has also been introduced in the field of condensed matter physics calculations @xcite .
other phenomenological forms of nonlocalities have also been considered such as a parity - dependent term in the optical potential @xcite for low energy scattering of protons from @xmath13 .
this parity dependence was later justified tentatively in terms of the feshbach channel coupling effect @xcite . a presently much used nonlocality , first introduced by frahn and lemmer @xcite and further developed by perey and buck in the 1960ies , is given in the form of an integral kernel @xmath14 that acts on the wave function .
perey and buck also obtained a local equivalent potential in the corresponding schrdinger equation @xcite .
a non local potential @xmath15 acts on the nonlocal wave function @xmath16 in the form @xmath17 while a local potential @xmath18 acts as a factor @xmath19 , where @xmath20 is a local wave function .
like the local wave function , @xmath16 is a function of one radial coordinate only . in the nuclear interior ,
a local potential leads to local wave functions which are different from the corresponding nonlocal ones obtained using a nonlocal potential , even though both forms may lead to the same asymptotic phase shift .
this is known as the perey effect austern , and may be expressed quantitatively as @xmath21where @xmath20 and @xmath22 are the local and nonlocal wave functions respectively , while the perey damping factor is defined as @xmath23 @xcite . in the nuclear interior , an attractive nonlocal potential results in a reduction in the nonlocal wave function ( @xmath24 ) , while for a repulsive potential the situation is reversed leading to an enhancement of @xmath22 , that is , @xmath25 . as discussed by austern @xcite , nonlocality arises when one singles out from the overall wave function that part which describes the motion in the elastic channel . outside the nucleus
@xmath22 is the full physical wave function of the system .
however , inside the nucleus part of the wave function of the system is hidden in the invisible ( inelastic ) channels , which results in a suppression of the nonlocal elastic wave function compared to the local one .
conversely , for a repulsive nonlocal potential , the particles of the target nucleus are driven away from the incident particle .
however , conservation of the flux demands that the incident particle carries more flux in order to compensate for the part of the channel space which has been suppressed austern . a more recent work employed nonlocal potentials in three - body direct nuclear reaction calculations and an important nonlocality effect in transfer reactions was reported @xcite . in this work
our aim is to show that the presence of a velocity - dependent part in the conventional , phenomenological optical potential simulates a source of nonlocality .
this is achieved by showing that , in the interior of the nucleus , the nonlocal wave functions are different from the corresponding local ones obtained in the absence of a gradient term ( @xmath26 ) in accordance with the perey effect
. we shall also demonstrate that this effect is angular momentum dependent .
in ref .
@xcite , we considered a novel velocity - dependent optical potential of the form given in eq .
( [ vdp ] ) based on the change of mass of the incident nucleon , and obtained excellent fits for the elastic @xmath27 angular distribution data in the low energy range of @xmath28 mev .
most notably the model reproduced well the prominent large - angle , backscattering minima which depend sensitively on the incident energy , and which are a sign of the presence of nonlocalities rawitscher - l-69,clark .
the local potential consisted of central and spin orbit potentials and was taken to have the form : @xmath29where @xmath30 stands for @xmath31 and so on for the rest of the terms .
in addition , the function @xmath32 has the following woods - saxon form : @xmath33},\]]with @xmath34 being the mass number of the target nucleus . in order to reduce the number of fit parameters ( 12 ) we used the same set of parameters for the real and imaginary parts of the spin - orbit term . in what follows the nonlocal wave function
@xmath22 is the solution of the schrdinger equation corresponding to the velocity - dependent potential @xmath35 given in eq .
( [ vdp ] ) .
the potential parameters are adjusted to obtain a best fit to the experimental @xmath2 elastic scattering angular distribution .
the local wave function @xmath20 , however , is the solution of the schrdinger equation corresponding to the equivalent local potential @xmath18 whose parameters are adjusted to ( i ) fit experimental data and ( ii ) produce the same phase shift values as those obtained using the nonlocal potential @xmath35 . in the local equivalent potential case , the effect of nonlocalities in the scattering process are manifested as a variation of the potential parameters with energy .
this procedure is not unique , but is guided by the conventionally used method to fit angular distributions .
the local wave function obtained in this way is not equal to the trivially equivalent local wave function , which is obtained by starting the fit in the presence of a velocity dependent term , and then mathematically transforming the @xmath36 away by a renormalization of the wave function @xmath37 ericson . in sec .
[ rd ] , we demonstrate that the nonlocal wave function @xmath38 is indeed different form the local wave function @xmath39 in the nuclear interior , which supports our hypothesis that the gradient term introduced into the conventional phenomenological optical potential simulates a source of nonlocality . in addition , we find that the enhancement or suppression of the nonlocal wave function relative to the local one to be angular momentum dependent .
this is in line with earlier works that modeled the @xmath0 elastic scattering by introducing a parity - dependent term @xmath40 into the optical potential @xcite .
as described above , the parameters of the equivalent local potential @xmath41 given in eq .
( [ vr ] ) were adjusted until ( i ) a best fit for the elastic @xmath2 angular distribution data was obtained and ( ii ) the resulting phase shifts were almost identical to the ones obtained corresponding to the nonlocal fits in ref .
the fits for the @xmath42 and @xmath43 mev incident nucleon energies are shown in fig .
angularfits1 , where the experimental data are taken from ref .
the equivalent local potential fit parameters as a function of the incident energy and orbital angular momentum are shown in table i. .conventional optical model fit parameters for the @xmath2 elastic scattering angular distribution .
the potential terms are given by eq .
( [ vr ] ) . [ cols="<,^,^,^,^,^,^,^,^,^ " , ]
in a recent work we introduced a velocity - dependent term into the phenomenological , conventional optical potential . this resulted in excellent fits for the @xmath2 elastic angular distributions even at large back scattering angles @xcite , which is a region known to be associated with nonlocalities @xcite . in the present study it has been shown that the velocity - dependent term simulates a source of nonlocality by showing that inside the target nucleus the nonlocal wave function @xmath22 is enhanced or suppressed compared to the corresponding local one @xmath20 obtained in the absence of the velocity - dependent term .
this is in accordance with the perey effect perey . in table
[ table1 ] we show the fit parameters for the equivalent local potential @xmath18 given in eq .
( [ vr ] ) for the @xmath2 elastic scattering process . clearly , the parameters are energy sensitive , and the strength of the local central potential @xmath44 shows the largest variation with incident energy .
this indicates the presence of nonlocal effects in the scattering process . in this work and in ref .
@xcite we proposed that the gradient term simulates a nonlocality , which is due to a change in mass of the incident nucleon when moving inside the nuclear matter of the target nucleus .
the corresponding fit parameters for the nonlocal , velocity - dependent potential @xmath35 ( given in tables i and ii of reference @xcite ) show less variation with incident energy .
this clearly indicates that there are other sources of nonlocalities like , for example , pauli exchange and knock out processes in addition to the detailed structure of the target nucleus that still need to be accounted for .
the local potential angular distribution fits are given in figure [ angularfits1 ] , which show that the details of the angular distributions are less well described compared to the corresponding ones obtained in ref .
@xcite .
we have also shown that the enhancement or suppression effects are angular momentum dependent . in table
[ table2 ] we show the perey factor @xmath23 as a function of the incident energy as well as the orbital angular momentum quantum number @xmath45 .
the angular momentum dependence of @xmath23 is suggestive of the inclusion of a parity dependent term @xmath46 in the conventional optical potential @xcite . since the gradient term simulating the nonlocality is purely real only the real part of @xmath22
is affected and the imaginary part is almost identical with the corresponding one for the local wave function . figs .
wfncs1 and [ wfncs2 ] show the real parts of local ( dotted ) and nonlocal ( solid ) wave functions for different angular momentum quantum numbers @xmath45 all at the same incident energy of 14 mev .
clearly , @xmath47 is a function of orbital angular momentum . finally , we considered two methods to determine the local wave function @xmath48 corresponding to the nonlocal one , @xmath22 , that was obtained in ref .
the first corresponds to obtaining the best fit parameters for an equivalent local potential such that @xmath48 and @xmath22 are phase equivalent .
the other obtains a trivially equivalent local wave function , which starts by fitting the data in the presence of the velocity - dependent term and then mathematically transforming the @xmath36 away by a renormalization of the wave function as given by eq .
( [ trans ] ) . in this case
the nonlocal @xmath49 and trivially local @xmath50 wave functions share the same fit parameters and , by eq .
( [ frrho ] ) , @xmath47 has energy dependence only as shown in table [ table3 ] . since the values of @xmath11 , which is given by eq .
( [ rho ] ) , range from zero to less than unity @xmath49 is enhanced for all energies . in conclusion ,
the non local nature of the derivative term is justified here in terms of the properties of the corresponding perey damping factor , which differs from unity by 20 to 30% .
furthermore , in future works we shall test the applicability of the introduced velocity - dependent potential to other systems like @xmath51 and @xmath52 in order to determine whetherthe surface term for @xmath11 as defined in eq .
( [ rho ] ) is still valid for heavier nuclei , or whether it should be replaced by a volume term that is proportional to the nuclear density itself , rather than its gradient .
, ( b ) @xmath53 and ( c ) @xmath54 angular momentum quantum numbers .
the local and nonlocal wave functions correspond to the local and nonlocal potentials given by eqs .
( [ vdp ] ) and ( [ vr ] ) .
all figures correspond to @xmath55 mev incident neutron energy . , width=325,height=604 ] , and ( b ) @xmath56 . in ( a )
the dash - dotted peak corresponds to @xmath57 defined in eq .
( [ rho ] ) .
the peaks of @xmath58 and the nonlocal wave function almost coincide , which explains the distorted peak of the latter .
the local and nonlocal wave functions correspond to the local and nonlocal potentials given by eqs .
( [ vdp ] ) and ( [ vr ] ) .
both figures correspond to @xmath55 mev incident neutron energy .
, width=325,height=453 ] | we demonstrate that the presence of a velocity - dependent term in the phenomenological optical potential simulates a source of nonlocality .
this is achieved by showing that , in the interior of the nucleus , the nonlocal wave functions are different from the corresponding local ones obtained in the absence of the velocity - dependent term in accordance with the perey effect .
it is also shown that the enhancement or suppression of the nonlocal wave function is energy as well as angular momentum dependent .
the latter is in line with the results of previous works that introduced parity dependent terms in the conventional optical potential .
keywords : damping factor , perey effect , optical potential . |
the analysis of modern accelerator - based neutrino oscillation experiments requires good control over the intermediate - energy neutrino - nucleus scattering cross section @xcite . in particular
the importance of multi - nucleon events has been suggested in many calculations of charge - changing quasielastic cross sections @xmath2 , at typical neutrino energies of @xmath3 gev @xcite .
the contribution of two - particle - two - hole ( 2p-2h ) excitations is now thought to be essential for a proper description of data @xcite .
thus a growing interest has arisen in including 2p-2h models into the monte carlo event generators used by the neutrino collaborations @xcite .
the only 2p-2h model implemented up to date in some of the monte carlo neutrino event generators corresponds to the so - called ific valencia model @xcite , which has been incorporated in genie @xcite .
there are also plans to incorporate the lyon model @xcite in genie , while phenomenological approaches like the effective transverse enhancement model of @xcite are implemented , for instance , in nuwro generator @xcite .
one of the main problems to implementing the 2p-2h models is the high computational time .
this is due to the large number of nested integrals involved in the evaluation of the inclusive hadronic tensor with sums over the final 2p-2h states . to speed up the calculations ,
several approximations can be made , such as choosing an average momentum for the nucleons in the local fermi gas @xcite , neglecting the exchange matrix elements , or reducing the number of integrations to two nested integrals by performing a non - relativistic expansion of the current operators @xcite .
the latter approach is only useful for some pieces of the elementary 2p-2h response . in this work we present a fast and very efficient method to calculate the inclusive 2p-2h responses in the relativistic fermi gas model ( rfg )
this approach , denoted as the frozen nucleon approximation , was first explored in @xcite but restricted to the analysis of the 2p-2h phase - space . here
it is extended to the evaluation of the full hadronic tensor assuming that the initial momenta of the two struck nucleons can be neglected for high enough energy and momentum transfer , @xmath4 .
the frozen nucleon approximation was found to work properly in computing the phase space function for two - particle emission in the range of momentum transfers of interest for neutrino experiments with accelerators .
here we investigate the validity of the frozen approximation beyond the phase - space study by including the electroweak meson - exchange current ( mec ) model of @xcite .
we find that the presence of virtual delta excitations requires one to introduce a `` frozen '' @xmath1-propagator , designed by a convenient average over the fermi sea .
the main advantage of the frozen approximation consists in reducing the number of nested integrals needed to evaluate the inclusive 2p-2h electroweak responses from 7 ( full calculation ) to 1 .
thus it is well - suited to computing the 2p-2h neutrino cross sections folded with the neutrino flux , and it can be of great help in order to implement the 2p-2h models in the monte carlo codes currently available .
the plan of this work is as follows : in section [ sec_form ] we review the formalism of neutrino scattering and describe mathematically the frozen approximation approach . in section [ sec_results ]
we validate the nucleon frozen approximation by computing the 2p-2h response functions and by comparing with the exact calculation .
finally , in section [ sec_conclusions ] we summarize our conclusions .
the double - differential inclusive @xmath5 or @xmath6 cross section is given by @xmath7 \
, , \end{aligned}\ ] ] where the sign @xmath8 is positive for neutrinos and negative for antineutrinos .
the term @xmath9 in eq .
( [ cross ] ) represents the elementary neutrino scattering cross section with a point nucleon , while the @xmath10 are kinematic factors that depend on lepton kinematic variables .
their explicit expressions can be found in @xcite .
the relevant nuclear physics is contained in the five nuclear response functions @xmath11 , where @xmath12 is the momentum transfer , defining the @xmath13 direction , and @xmath14 is the energy transfer .
they are defined as suitable combinations of the hadronic tensor @xmath15 in this work we compute the inclusive hadronic tensor for two - nucleon emission in the relativistic fermi gas , given by @xmath16 where @xmath17 by momentum conservation , @xmath18 is the nucleon mass , @xmath19 is the volume of the system and we have defined the product of step functions @xmath20 with @xmath21 the fermi momentum . finally the function @xmath22 is the elementary hadron tensor for the 2p-2h transition of a nucleon pair with given initial and final momenta , summed up over spin and isospin , @xmath23 which is written in terms of the antisymmetrized two - body current matrix elements @xmath24 the factor @xmath25 in eq .
( [ elementary ] ) accounts for the antisymmetry of the two - body wave function . for the inclusive responses considered in this work
there is a global axial symmetry , so we can fix the azimuthal angle of one of the particles .
we choose @xmath26 , and consequently the integral over @xmath27 gives a factor @xmath28 .
furthermore , the energy delta function enables analytical integration over @xmath29 , and so the integral in eq .
( [ hadronic ] ) can be reduced to 7 dimensions ( 7d ) . in the `` exact '' results shown in the next section , this
7d integral has been computed numerically using the method described in @xcite .
the frozen nucleon approximation consists in assuming that the momenta of the initial nucleons can be neglected for high enough values of the momentum transfer .
thus , in the integrand of eq .
( [ hadronic ] ) , we set @xmath30 , and @xmath31 .
we roughly expect this approximation to become more accurate as the momentum transfer increases .
the integration over @xmath32 is trivially performed and the response function @xmath33 , with @xmath34 , is hence approximated by @xmath35 where @xmath36 and @xmath37 are the elementary response functions for a nucleon pair excitation , which are defined similarly to eqs .
( [ rcc][rtprima ] ) .
the integral over @xmath38 can be done analytically by using the delta function for energy conservation , and the integral over @xmath39 gives again a factor of @xmath28 .
thus only an integral over the polar angle @xmath40 remains : @xmath41 where the sum runs over the , in general two , possible values of the momentum of the first particle for given emission angle @xmath42 .
these are obtained as the positive solutions @xmath43 of the energy conservation equation @xmath44 the explicit values of the solutions of the above equation can be found in the appendix of @xcite .
care is needed in performing the integral over @xmath42 because the denominator inside the integral can be zero for some kinematics .
the quadrature in these cases can be done with the methods explained in @xcite .
to investigate the validity of the frozen nucleon approximation , we have to choose a specific model for the two - body current matrix elements @xmath45 entering in the elementary 2p-2h response functions , eqs .
( [ elementary],[anti ] ) .
here we use the relativistic model of electroweak mec operators developed in @xcite .
the mec model can be summarized by the feynman diagrams depicted in fig .
[ fig_feynman ] .
it comprises several contributions coming from the pion production amplitudes of @xcite .
the seagull current , corresponding to diagrams ( a , b ) , is given by the sum of vector and axial - vector pieces @xmath46_{1'2',12 } \frac{f^2_{\pi nn}}{m^2_\pi } \frac{\bar{u}_{s^\prime_1}({{\bf p}}^\prime_1)\,\gamma_5 { \not{\!k}}_{1 } \ , u_{s_1}({{\bf h}}_1)}{k^2_{1}-m^2_\pi}\ , \nonumber\\ & \times & \bar{u}_{s^\prime_2}({{\bf p}}^\prime_2 ) \left [ f^v_1(q^2)\gamma_5 \gamma^\mu + \frac{f_\rho\left(k_{2}^2\right)}{g_a}\,\gamma^\mu \right ] u_{s_2}({{\bf h}}_2 ) \nonumber\\ & + & ( 1\leftrightarrow2 ) \ , , \label{seacur } \end{aligned}\ ]
] where @xmath47 corresponds to the @xmath8-components of the two - body isovector operator @xmath48 $ ] .
the @xmath49 sign refers to neutrino ( antineutrino ) scattering .
the four - vector @xmath50 is the momentum carried by the exchanged pion and @xmath51 .
the @xmath52 ( @xmath53 ) and axial ( @xmath54 ) couplings , and the form factors ( @xmath55 , @xmath56 ) have been taken from the pion production amplitudes of @xcite .
the pion - in - flight current corresponding to diagram ( c ) is purely vector and is given by @xmath57_{1'2',12 } \frac{f^2_{\pi nn}}{m^2_\pi}\ , \frac { f^v_1(q^2 ) \ , \left(k^\mu_{1}-k^\mu_{2}\right)}{\left(k^2_{1}-m^2_\pi\right ) \left(k^2_{2}-m^2_\pi\right)}\ , \nonumber \\ & \times & \bar{u}_{s^\prime_1}({{\bf p}}^\prime_1)\,\gamma_5 { \not{\!k}}_{1 } \
, u_{s_1}({{\bf h}}_1 ) \bar{u}_{s^\prime_2}({{\bf p}}^\prime_2)\,\gamma_5 { \not{\!k}}_{2 } \ , u_{s_2}({{\bf h}}_2 ) \
, , \end{aligned}\ ] ] where @xmath58 is the momentum of the pion absorbed by the second nucleon .
the pion - pole current corresponds to diagrams ( d , e ) and is purely axial , given by @xmath59_{1'2',12 } \frac{f^2_{\pi nn}}{m^2_\pi}\ , \frac{f_\rho\left(k_{1}^2\right)}{g_a}\;q^\mu \bar{u}_{s^\prime_1}({{\bf p}}^\prime_1){\not{\!q}}u_{s_1}({{\bf h}}_1 ) \nonumber\\ & \times & \frac { \bar{u}_{s^\prime_2}({{\bf p}}^\prime_2)\,\gamma_5 { \not{\!k}}_{2 } \
, u_{s_2}({{\bf h}}_2 ) } { \left(k^2_{2}-m^2_\pi\right)\left(q^2-m^2_\pi\right ) } + ( 1\leftrightarrow2 ) \,.\end{aligned}\ ] ] finally the @xmath1 current corresponds in fig .
[ fig_feynman ] to diagrams ( f , g ) for the forward and ( h , i ) for the backward @xmath1 propagations , respectively .
the current matrix elements are given by @xmath60_{1'2',12 } \frac{\bar{u}_{s^\prime_2}({{\bf p}}^\prime_2)\,\gamma_5 { \not{\!k}}_{2 } \ , u_{s_2}({{\bf h}}_2)}{k^2_{2}-m^2_\pi } \nonumber\\ & \times & k^\alpha_{2}\,\bar{u}_{s^\prime_1}({{\bf p}}^\prime_1)g_{\alpha\beta}(h_1+q ) \gamma^{\beta\mu}(h_1,q)u_{s_1}({{\bf h}}_1 ) \nonumber\\ & + & ( 1\leftrightarrow2 ) \label{delta_forward } \\
j^\mu_{\delta,\rm b}&= & \frac{f^ * f_{\pi nn}}{m^2_\pi}\ , \left[u_{\rm b}^{\pm}\right]_{1'2',12}\ ; \frac{\bar{u}_{s^\prime_2}({{\bf p}}^\prime_2)\,\gamma_5 { \not{\!k}}_{2 } \ , u_{s_2}({{\bf h}}_2)}{k^2_{2}-m^2_\pi } \nonumber\\ & \times & k^\beta_{2}\,\bar{u}_{s^\prime_1}({{\bf p}}^\prime_1 )
\hat{\gamma}^{\mu\alpha}(p^\prime_1,q ) g_{\alpha\beta}(p^\prime_1-q)u_{s_1}({{\bf h}}_1 ) \nonumber\\ & + & ( 1\leftrightarrow2)\label{delta_backward } \ , .\end{aligned}\ ] ] the @xmath61 coupling is @xmath62 .
the forward , @xmath63 , and backward , @xmath64 , isospin transition operators have the following cartesian components @xmath65 where @xmath66 and @xmath67 are the isovector transition operators from isospin @xmath68 to @xmath69 or vice - versa , respectively .
the @xmath70 operator is for neutrino ( antineutrino ) scattering .
the @xmath1-propagator , @xmath71 , is given by @xmath72 where @xmath73 is the projector over spin-@xmath68 on - shell particles , @xmath74\end{aligned}\ ] ] and whose denominator has been obtained from the free propagator for stable particles , @xmath75 , with the replacement @xmath76 to take into account the finite decay width of the @xmath77 .
the tensor @xmath78 in the forward current is the weak @xmath79 transition vertex a combination of gamma matrices with vector and axial - vector contributions : @xmath80\gamma_5 \\ \gamma^{\beta\mu}_a(p , q)&= & \frac{c^a_3}{m_n}\left(g^{\beta\mu}{\not{\!q}}- q^\beta\gamma^\mu\right)\nonumber\\ & + & \frac{c^a_4}{m^2_n}\left(g^{\beta\mu}q\cdot \left(p+q\right)- q^\beta \left(p+q\right)^\mu\right ) \nonumber\\ & + & c^a_5 g^{\beta\mu}+\frac{c^a_6}{m^2_n } q^\beta q^\mu \ , .\end{aligned}\ ] ] for the backward current , we take @xmath81^{\dagger } \gamma^0 \ , .\ ] ] finally , it is worth noting that the form factors @xmath82 are taken from @xcite .
we refer to that work for further details of the model .
the evaluation of the relevant elementary responses requires one to contract the electroweak two - body mec with themselves by spin - isospin summation .
this leads to the squares of each of the diagrams depicted in fig .
[ fig_feynman ] plus all their interferences .
the validity of the frozen nucleon approximation relies on the fact that the integrand inside the 2p-2h response is a function that depends slowly on the momenta of the two initial nucleons inside the fermi sea . in that case
the mean - value theorem applied to the resolution of the integrals provides very precise results .
this is so for all of the diagrams of the mec except for the forward @xmath1 diagram , which shows a sharp maximum for kinematics around the @xmath1 peak for pion emission , located at @xmath83 .
this is due to the denominator in the @xmath1 propagator , @xmath84 where @xmath85 is the momentum of the hole that gets excited to a @xmath1 . in these cases
the integrand changes very significantly with a small variation of the momentum of the holes and consequently , the frozen approximation can not properly describe the integrand . on the contrary , it only provides a general estimation of the order of magnitude .
to get rid of these difficulties we have developed a prescription to deal with the forward @xmath1-propagator appearing in eq .
( [ delta_forward ] ) .
this procedure is based on the use of an effective propagator ( `` frozen '' ) for the @xmath1 , conveniently averaged over the fermi gas .
this average is an analytical complex function , which is used instead of the `` bare '' propagator inside the frozen approximation , recovering the precision of the rest of diagrams .
the `` frozen '' prescription amounts to the replacement : @xmath86 where the frozen denominator is defined by @xmath87 taking the non - relativistic limit for the energies of the holes ( @xmath88 ) , which is justified because hole momenta are below the fermi momentum , itself a value far below the nucleon rest mass , we can write : @xmath89 where @xmath90 assuming the @xmath1 width ( @xmath91 ) to be constant , we can integrate eq .
( [ averaged_prop ] ) over the angles , getting @xmath92 \,.\ ] ] note the complex logarithm inside the integral , which provides the needed kinematical dependence of the averaged propagator , differing from the bare lorentzian shape . finally the integral over the momentum @xmath93 can also be performed , resulting in @xmath94 \right\ } \ , .
\nonumber\end{aligned}\ ] ] by comparing the response functions evaluated in the frozen approximation , _
i.e. , _ substituting the denominator of the @xmath1 propagator in eq .
( [ delta_prop ] ) for the frozen expression in eq .
( [ averaged_denom ] ) , with the exact results , we find that the shapes around the @xmath1 peak are similar , but with slightly different width and position of the center of the peak .
we have checked that the differences can be minimized by changing the parameters @xmath95 with respect to the `` bare '' ones , given by eqs .
( [ a],[b ] ) .
this is because we have computed the averaged denominator without taking into account the current matrix elements appearing in the exact responses , although the functional form and kinematical dependence is the appropriate one . in practice , we adjust @xmath91 and apply a shift in the expression for @xmath96 in eq .
( [ a ] ) in order to obtain the best approximation to the exact results .
the effective `` frozen '' parameters we actually introduce in eq .
( [ averaged_denom ] ) , are given by @xmath97 we consider @xmath98 and the frozen shift , @xmath99 , to be tunable parameters depending on the momentum transfer @xmath100 .
we have adjusted these parameters for different @xmath100-values and we provide them in table [ table1 ] .
.values of the free parameters of the fermi - averaged @xmath1-propagator for different kinematic situations corresponding to different values of the momentum transfer @xmath100 . [ cols="^,^,^",options="header " , ] -propagator in frozen approximation compared to the average propagator . in this evaluation
we have taken @xmath101 mev / c and @xmath102 mev.,width=302 ]
in this section we validate the frozen approximation by computing the approximate 2p-2h response functions and comparing with the exact results in the rfg .
we consider the case of the nucleus @xmath103c with fermi momentum @xmath104 mev / c , and show the different response functions for low to high values of the momentum transfer .
for other nuclei with different @xmath21 the frozen parameters of table 1 should be determined again , and we expect their values change slightly . in fig .
[ fig_denom ] we show the modulus squared of the @xmath1 propagator , given by the @xmath105 function defined in eq .
( [ denominator ] ) , computed for @xmath106 , as a function of @xmath14 for @xmath107 gev / c .
it presents the typical lorentzian shape corresponding to width @xmath102 mev .
we observe a narrow peak around @xmath108 mev .
this corresponds to the @xmath1-peak position for @xmath107 gev / c . in the same figure
we also show the square of the frozen average @xmath109 ( solid line ) .
the resulting peak is quenched and broadened as compared to the lorentzian shape , reducing its strength and enlarging its width .
this behavior of the averaged @xmath1-propagator drives the actual shape of the exact 2p-2h nuclear responses , being more realistic than the simple lorentzian shape of the frozen approximation without the average , as we will see below . of @xmath103c for different momentum transfers @xmath100 .
the exact results are compared to the frozen approximation with and without the averaged @xmath1 propagator .
, width=283 ] in fig .
[ fig_rt ] we show the weak transverse 2p-2h response function of @xmath103c for four different values of the momentum transfer ranging from 300 to 1500 mev / c .
the curves correspond to different calculations or approximations made in the evaluation of the responses , as labeled in the legend .
the solid line corresponds to the seven - dimensional calculation with no approximations .
the other two curves refer to the different frozen nucleon approximations developed in this work : the dashed line is obtained with eq .
( [ frozen_eq ] ) but performing the replacement expressed in ( [ replacement_rule ] ) for the forward @xmath1-excitation terms in the evaluation of the current matrix elements ; on the contrary , the dotted line corresponds to the same frozen nucleon approximation , eq .
( [ frozen_eq ] ) , but without the fermi - average of the @xmath1-propagator in the forward terms . as it can be seen from fig .
[ fig_rt ] , for those values of the momentum transfer for which the @xmath1-peak is not reached ( the panel with @xmath110 mev / c ) , there is really little difference between averaging or not the @xmath1 propagator .
this is certainly not the case when the @xmath1-peak is fully reached , as shown in the other panels .
in this situation there is a dramatic difference between performing the fermi - average of the @xmath1-propagator or not .
this difference is in consonance with the results shown in the previous fig .
[ fig_denom ] , and it shows how crucial is the treatment of the @xmath1-propagator to obtain accurate results for the 2p-2h responses in the frozen nucleon approximation , i.e , with only one integration . the results in fig .
3 have been obtained after fitting the parameters @xmath111 for the fermi - averaged @xmath1-propagator at the different values of the momentum transfer quoted in table [ table1 ] .
it is also worth noting that there is no way of converting the dotted line into the dashed one by only a suitable fitting of these parameters , _
i.e. , _ without averaging the @xmath1-propagator .
c within different models for two values of the momentum transfer .
the exact rfg results and the frozen approximation are compared with the shell model ( sm ) results of @xcite .
the total shell model results ( 1p-1h ) + ( 2p-2h ) are also shown for comparison .
, width=302 ] c. several values of the momentum transfer are displayed : @xmath112 and 2000 mev / c .
, width=302 ] in fig .
[ fig_shellmodel ] we show results for the transverse electromagnetic 2p-2h response function .
the frozen and exact ( 7d ) @xmath113 response of the rfg are compared with the results obtained in the shell model 2p-2h calculation of @xcite .
this was one of the first computations of the 2p-2h response within the nuclear shell model .
the total nuclear response in the shell model , obtained by adding the 1p-1h to the 2p-2h channel , is also shown to appreciate the relative size of the 2p-2h contribution to the total result . as shown in fig .
[ fig_shellmodel ] , the fermi gas results ( either in frozen approximation or not ) are similar to the shell model ones . the small discrepancy between them can not be attributed to relativistic effects because of the low momentum transfer values considered , but to the different coupling constants and form factors used in the model of the @xmath1 meson - exchange current considered in @xcite and the present approach .
we can remark the slightly different threshold effects between both calculations .
these effects are , as expected , very sensitive to the treatment of the nuclear ground state .
note also that the frozen approximation describes reasonably well this low momentum @xmath110 mev / c , considering the simplifications involved . finally in fig .
[ fig5 ] we show that the frozen approximation works notably well in a range of momentum transfer from low to high values of @xmath100 .
we compare the @xmath113 , @xmath114 , and @xmath115 2p-2h responses in frozen approximation with the exact results obtained computing numerically the 7d integral of the hadronic tensor . the accord is particularly good for the two transverse responses which dominate the cross section .
a slight disagreement occurs for very low energy transfer at threshold where the response functions are anyway small . in the case of the cc response function
some tiny differences are observed .
however , note that this response is small because the dominant @xmath1 current is predominantly transverse
. moreover , its global contribution to the cross section is not very significant because it is partially canceled with the contribution of the cl and ll responses .
[ [ physical - interpretation - of - the - frozen - approximation ] ] physical interpretation of the frozen approximation + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the validity of the frozen approximation led us to conclude that , in the inclusive responses for two - particle emission , the detailed information about the momenta carried out by the two nucleons is lost .
this is because the energy and momentum transfer @xmath116 are shared by the two nucleons in multiple ways .
this is reminiscent from the phase - space kinematical dependence ( which can be obtained setting the elementary response @xmath117 to unity ) already seen in @xcite .
the soft dependence of the elementary response on the initial momenta makes the same argument applicable to the full responses with the exception of the @xmath1 forward current that requires one to soften and average the rapid variations of the @xmath1 propagator .
only the low - energy region where the sharing is highly restricted and the cross section is therefore very small , is found to be sensitive to the details of the initial state .
this is also supported by the comparison between the shell model and the rfg . on the other hand , in the 2p-2h model of @xcite , an average momentum @xmath118
was determined by imposing quasi - deuteron kinematics .
note that this condition is similar to the present frozen approach , but this only guarantees that the total momentum of the two holes is zero , corresponding to selecting back - to - back pair configurations in the ground state only .
in this work we have introduced and validated the frozen nucleon approximation for a fast and precise calculation of the inclusive 2p-2h response functions in a relativistic fermi gas model .
this approximation neglects the momentum dependence of the two holes in the ground state and requires the use of an effective propagator for the @xmath1 resonance conveniently averaged over the fermi sphere , for which we have provided a simple analytical expression . for momentum transfers above the fermi momentum
this approximation makes it possible to compute the responses with only a one - dimensional integral . taking into account all the uncertainties in modeling the two - nucleon emission reactions
, this approach can be used instead of the full 7d integral , obtaining very satisfactory results .
although we have used a specific model of mec to prove the validity of the approximation , it is reasonable to expect that the frozen approach is also valid for other 2p-2h models .
this can be of great interest when implementing 2p-2h models in monte carlo event generators , which up to now have relied on parameterizations from external calculations . in summary ,
the frozen approximation enables one to make 2p-2h calculations very efficiently and rapidly , instead of interpolating pre - calculated tables , including allowing the parameters of the models to be modified inside the codes , if desired .
finally , in the near future this study will be extended to an exploration of how the 2p-2h mec responses depend on nuclear species @xcite .
this work has been partially supported by the spanish ministerio de economia y competitividad and erdf ( european regional development fund ) under contracts fis2014 - 59386-p , fis2014 - 53448-c2 - 1 , by the junta de andalucia ( grants no .
fqm-225 , fqm160 ) , by the infn under project manybody , and part ( twd ) by the u.s .
department of energy under cooperative agreement de - fc02 - 94er40818 .
irs acknowledges support from a juan de la cierva fellowship from mineco ( spain ) .
gdm acknowledges support from a junta de andalucia fellowship ( fqm7632 , proyectos de excelencia 2011 ) .
i. ruiz simo , j. e. amaro , m. b. barbaro , a. de pace , j. a. caballero and t. w. donnelly , arxiv:1604.08423 [ nucl - th ] .
j. e. amaro , m. b. barbaro , j. a. caballero , t. w. donnelly and c. maieron , phys .
c * 71 * ( 2005 ) 065501 . | we present a fast and efficient method to compute the inclusive two - particle two - hole ( 2p-2h ) electroweak responses in the neutrino and electron quasielastic inclusive cross sections .
the method is based on two approximations .
the first neglects the motion of the two initial nucleons below the fermi momentum , which are considered to be at rest .
this approximation , which is reasonable for high values of the momentum transfer , turns out also to be quite good for moderate values of the momentum transfer @xmath0 .
the second approximation involves using in the `` frozen '' meson - exchange currents ( mec ) an effective @xmath1-propagator averaged over the fermi sea . within the resulting `` frozen nucleon approximation '' ,
the inclusive 2p-2h responses are accurately calculated with only a one - dimensional integral over the emission angle of one of the final nucleons , thus drastically simplifying the calculation and reducing the computational time .
the latter makes this method especially well - suited for implementation in monte carlo neutrino event generators .
neutrino scattering , meson - exchange currents , 2p-2h .
25.30.pt , 25.40.kv , 24.10.jv |
we assume that the classical limit of the massive gluons is represented by the equation of motion with a mass : - d _ f _ - m^2 a _ = 0 , [ eqm ] where f_^a = _ a_^a - _ a_^a + g f_abc a_^b a_^c , and @xmath0 is the covariant derivative .
the landau - gauge condition @xmath1 follows from this equation by applying @xmath2 on the equation . at the zeroth order in @xmath3 , we find the following solution & & a_0 = 0 , + & & a = ^-1(_2 t + _ 3 t ) h_0(z , r ) , [ h0 ] where h_0 = ( _ x , _ y , _ z - ( m^2 -^2 ) dz ) ( z , r ) , with @xmath4 and @xmath5 .
the electric and magnetic field is e^a & = & -a_0^a - i g f_abca^b a_0^c - ^a , + b^a & = & a^a .
this solution may be interpreted as having two point charges placed at @xmath6 and @xmath7 with the opposite signs .
this electric field has no divergence @xmath8 except for the points on the charges with the magnitude @xmath9 .
the electric flux looks like that of magnetic field in superconducting material .
the color of the magnetic field and the electric field are 90 degrees different in the space of color , and the magnetic flux keeps inducing the electric flux .
the shape of electric flux and magnetic field are shown in figure [ fig : em4 ] .
the electric flux looks like a tube elongated in @xmath10-direction .
when we pull the electric charges apart , the tube will get longer but it wo nt get thicker and be about @xmath11 .
therefore , a constant tension will occur between the charges , which is quite consistent with stringy picture@xcite of confinement . .
the mass @xmath12 .
electric flux on the vertical plane and magnetic flux on a tube for the zero - th order classical solution are shown.,width=340 ] the zeroth - order solution above may be improved to all the orders perturbatively .
the equation of motion ( [ eqm ] ) in a expanded form is a^a - _ _ a^a - m^2 a^a = g f_abc ^ a_^b a_^c + g f_abc a_^b f_^c .
[ eqm2 ] we can perturbatively obtain the solution by repeatedly applying this equation .
in addition , we have @xmath13 not only for the zeroth - order solution ( [ h0 ] ) but also to all the orders in @xmath3 . the right hand side of eq.([eqm2 ] ) gives zero if we apply @xmath14 on it , since the first term gives @xmath15 , and the second term @xmath16 will become ^ j_^a = ( ^ a^a ) ( _ a_^a - _
a_^a ) + a^a^ ( _ a_^a - _
a_^a ) + g ^ a^a
f_abc a_^b a_^c .
using symmetry , the equation of motion and jacobi s identity , we have ^ j_^a & = & g a^a f_abc ( _ a_^ba_^c + a_^b ( _ a_^c - _
a_^c ) + g a^b f_cde
a_^d a_^e + m^2a^a ) & & + g ^ a^a f_abc a_^b a_^c = 0
. then @xmath17 follows at the @xmath18-th order , if @xmath17 holds at the @xmath19-th order in @xmath3 .
this classical solution gives a picture that @xmath20 and @xmath21 are rotating within a color plane that includes @xmath22 and @xmath23 direction , and quark charges are rotating too .
quantum mechanically , a quark must change its color after emitting a gluon .
this should have been the reason why it was difficult to understand the static force between confined quarks in the analogy of electric force .
further this picture clarifies why non - abelian nature is essential for confinement . in the solution of massive yang - mills theory such as weak theory ,
the solution does not form conserved flux tubes and its electric field vanishes at longer distance .
the reason why we do nt have such a decaying solution is because we do nt have current conservation @xmath24 for the theory with broken global symmetry
. this confinement picture will be valid for all the theories that have the same equation of motion classically , including real qcd and lattice qcd .
here we additionally present a toy but quantum model , in which the mechanism presented in the previous section holds and easier to analyze .
we consider a theory with its lagrangian : ( f_^a)^2 + |_- ig a_^a ^a |^2 - ( ^)^2 + ^ , and with a spontaneous symmetry break of _
ij , [ vev ] where @xmath25 is a complex valued @xmath26 matrix field , and its left index couples to the gauge field but its right index does not .
the gauge symmetry is broken , but the global color - rotation symmetry would not be broken under the vacuum expectation value ( [ vev ] ) since the non - gauged ( right ) index of @xmath25 may be rotated together . using the faddeev - popov method ,
the lagrangian is l & - & ( _ a^a - i g m ( - ^ ) ) ^2 & + & i |c^a ( ( d)_ab + g m ( ^a ^b+ ^b^a ^ ) ) c^b .
it is invariant under the brst transformation : _ ba^a _ &
= & _ c^a + g f_abc a^b _ c^c , _ b _ ij & = & i g c^a^a_ik _ kj , _
b c^a & = & - g f_abc c^b c^c/2 , + _ b|c^a & = & i b^a , _ b b^a & = & 0 . this model gives the equation of motion ( [ eqm ] ) as the classical counterpart . under the spontaneous symmetry breakdown of eq.([vev ] ) , the gauge bosons and ghosts acquire mass , and massive scalar particles appear due to higgs mechanism
. we may take the mass of the higgs particle large enough to make it physically irrelevant .
it is easy to show existence of mass gap in this model . here
@xmath27 is the full hamiltonian with @xmath28 shifted variables , and @xmath29 . for any eigenstate @xmath30 with its energy @xmath31 , we have e = e|h + h_m|e > e|h_m|e , and the right - hand side should be a positive value unless the number of the gluon is zero . in general , |e= |n_g = 0 + |n_g = 1 + |n_g = 2 + , where @xmath32 is a state with its gluon number @xmath33 .
any energy eigenstate with quarks should include @xmath34 states .
if it does not , ( h_0 + h_int + h_a)|e= e|e , where @xmath35 includes @xmath36 states and @xmath37 should only include @xmath38 and @xmath39 states but the right - hand side can not include @xmath34 states . due to nonbreaking of the brst symmetry ,
the quarks are confined in this model because no colored state may appear as follows .
the color current is j_^a = f_abc a^b f_0^c(x ) + j_0^a + ( a_b)^a - i ( |cd_c ) + i ( _ |cc ) , where @xmath40 is the color current from the quarks and the higgs field @xmath41 , where @xmath42 . the maxwell equation d^f_^a + g j_^a = _ b^a - i g f_abc ( _ |c^b ) c^c , may be written as @xmath43 . under brst formalism , physical states must be annihilated by the generator @xmath44 @xcite .
therefore @xmath45 for any physical states @xmath46 and @xmath47 , which means that ( ^f_^a + g j_^a ) |f= 0 inside the physical space . the operator on the left - hand side is the generator of color gauge transformation , and then this equation means that only color - neutral state can appear as a physical state .
in this letter , we presented a new classical solution of qcd and discussed possible relation to confinement .
further , we found that confinement and mass gap may occur in a model with a explicit mass introduction with higgs mechanism .
dual meissner effect , i.e. massiveness of magnetic field , has mainly been considered to be the mechanism that confines color charged particles , i.e. quarks .
however , the picture presented here shows that mass acquisition of electric field is rather appropriate to show the confinement string if the vector particle could acquire mass without breaking the conservation of color current .
this mechanism will be valid in the real qcd because the lattice qcd , which is equivalent to the real qcd , dynamically acquires mass from analytical and lattice studies@xcite .
further , we have another possibility that some of higgsless massive vector field theories @xcite works . in that case , the analyses in the latter sections would be useful .
bogolubsky , e .-
ilgenfritz , m. mller - preussker , a. sternbeck , lattice gluodynamics computation of landau - gauge green s functions in the deep infrared , phys.lett.b676:69-73,2009 [ arxiv:0901.0736 ] | recent researches on the solution of schwinger - dyson equations , as well as lattice simulations of pure qcd , suggest that the gluon propagator is massive . in this letter
, we assume that the classical counterpart of this massive gluon field may be represented with the equation of motion for yang - mills theory with a mass term added .
a new classical solution is given for this equation .
it is discussed that this solution may have some role in confinement .
these days , evidences are accumulating that the lattice qcd , which is equivalent to the real qcd , dynamically acquires mass from analytical and lattice studies@xcite ( and see @xcite for a list of references ) .
the analytical studies with the schwinger - dyson equation ( sde ) nicely agree @xcite with the lattice data .
those sde analyses are based on landau gauge . in this letter , the classical counterpart of massive gluons in the landau gauge is considered . in the next section , a new classical solution is given for the equation of motion , and its relation to confinement is discussed .
the mechanism here will equally be valid as far as the theory has a mass but the color symmetry is unbroken .
though this theory is intended to give insight into real qcd , a non - qcd toy model is additionally analyzed in section 2 .
this model is not for the real world but for facilitating analysis in a toy world .
this model shows interesting behavior of the mass gap and absence of colored states . |
we show existence of a weak global dissipative solution of the cauchy problem for the two - component camassa holm ( 2ch ) system with arbitrary @xmath0 and @xmath1 , given by [ eq : chsys2a ] @xmath2 with initial data @xmath3 and @xmath4 .
the initial data may have nonvanishing limits at infinity , that is , @xmath5 the 2ch system was first analyzed by constantin and ivanov @xcite .
global existence , well - posedness and blow - up conditions have been further studied in a sobolev setting in @xcite and in besov spaces in @xcite . the scalar ch equation
( i.e. , with @xmath6 identically equal to zero ) , was introduced and studied by camassa and holm in the fundamental paper @xcite , see also @xcite , and its analysis has been pervasive .
the ch equation possesses many intriguing properties .
here we concentrate on global solutions for the cauchy problem on the line .
the challenge is that the ch equation experiences blow - up in finite time , even for smooth initial data , in the sense that the @xmath7 norm of the solution remains finite while @xmath8 blows up .
continuation of the solution past blow - up is intricate .
it has turned out to be two distinct ways to continue the solution past blow - up , denoted conservative and dissipative solutions , respectively .
conservative solutions are associated with preservation of the @xmath9 norm , while dissipative solutions are characterized by a sudden drop in @xmath9 norm at blow - up .
this dichotomy has consequences for the well - posedness of the initial value problem as the two solutions coincide prior to blow - up .
here we focus on the dissipative case .
three features are novel in this paper : first of all we include distinct and nonvanishing asymptotics of the initial data , and hence of the solution , at infinity .
since prior work has been on solutions in @xmath9 , this has considerable consequences for the analysis .
secondly , we extend previous results for the ch equation to the 2ch system .
it is not at all clear a priori that the highly tailored construction for the ch equation extends to the 2ch system .
finally , we greatly simplify the analysis of two of us @xcite even in the scalar case of the ch equation with vanishing asymptotics .
one advantage of the present approach is that we can use the same change of variables as in the conservative case , in contrast to the approach chosen in @xcite .
we reformulate the 2ch system in terms of lagrangian coordinates , and in this respect it relates to @xcite for the ch equation .
previous work on the ch equation , covering also the periodic case , includes , e.g. , @xcite .
see also @xcite .
the intricate problems regarding wave breaking can best be exemplified in the context of multipeakon solutions of the ch equation with @xmath10 . for reasons of brevity , and since this example has been discussed in detail in @xcite , we omit the discussion here . for additional theory on multipeakons ,
see @xcite .
the continuation of the solution past wave breaking has been studied both in the conservative @xcite and dissipative @xcite case . in both cases
the approach has been to reformulate the partial differential equation as a system of banach space - valued ordinary differential equations , and we follow that approach here .
a different approach , based on vanishing viscosity , has been advocated in @xcite .
if we for a moment assume vanishing asymptotics , the dichotomy can be further understood if one considers the associated energy , that is , the @xmath9 norm of the solution @xmath11 for the ch equation . in the case of a symmetric antipeakon - peakon collision ,
the @xmath9 norm is constant prior to wave breaking . at collision time it vanishes , and remains zero for dissipative solutions , while returning to the previous value in the conservative case .
thus we need to keep the information about the energy in the conservative case and this is handled by augmenting the solution with the energy .
more precisely , we consider as solution the pair @xmath12 where @xmath13 is a radon measure with absolute continuous part @xmath14 .
this allows for energy concentration in terms of dirac masses , while keeping the information about the energy . on the other hand , in the dissipative case ,
energy is not preserved , rather it is strictly decreasing at wave breaking .
the extension from scalar ch equation to the two - component 2ch system follows by augmenting the lagrangian reformulation by an additional variable .
let us now turn to a more detailed description of the results in this paper .
first we observe that we can put @xmath10 and @xmath15 since if @xmath16 solves , then @xmath17 with @xmath18 and @xmath19 will solve with @xmath10 and @xmath15 . note that this only applies since we allow for non decaying initial data at infinity .
furthermore , we assume that @xmath20 .
we reformulate the 2ch system as @xmath21 next we introduce the characteristics @xmath22 , that is , the solution of @xmath23 for a given @xmath11 and initial data @xmath24 .
the lagrangian velocity is given by @xmath25 .
as long as @xmath22 remains strictly increasing as a function of @xmath26 , the solution remains smooth , and , in particular , conservative and dissipative solutions coincide .
thus we introduce the time for wave breaking , or energy dissipation , by @xmath27 we can rewrite the full 2ch system as a system of ordinary differential equations .
first define @xmath28 where @xmath29 with @xmath30 and @xmath31 with @xmath32 .
in addition , @xmath33 . the function @xmath34 is a smooth increasing function that vanishes for large negative arguments and equals one for large positive arguments .
next we find that the system obeys the following system of ordinary differential equations @xmath35 where @xmath36 and @xmath37 are given by and , respectively ( observe the subtle modifications in the dissipative case compared with and , respectively ) .
introduce @xmath38 and @xmath39 for @xmath40 and @xmath41 , respectively .
we find that @xmath42 satisfies the system @xmath43 .
the function @xmath44 takes values in a specific banach space @xmath45 , see .
this system poses two challenges : first of all , due to wave breaking , the right - hand side is discontinuous , and thus existence and uniqueness of solutions can not follow the standard path .
this is the key difficulty compared with the conservative case .
secondly , the system possesses a number of additional constraints in order to be consistent with the original eulerian formulation . for example
, we need to make sure that @xmath46 and @xmath47 are satisfied for positive @xmath48 .
we will also need @xmath49 .
this is secured by only using initial data from a carefully selected set that makes sure that all additional requirements are preserved for the solution .
the set is denoted @xmath50 and is given in definition [ def : g ] .
a key result is the proof of global existence of a solution , @xmath51 with initial data @xmath52 , in the lagrangian variables , theorem [ th : global ] .
next we have to analyze stability of the solution in lagrangian coordinates .
the problem here is to identify a metric that separates conservative solutions ( say @xmath53 ) and dissipative solutions ( say @xmath54 ) near wave breaking .
the key is the behavior of the derivative of the characteristics , @xmath40 . at wave breaking @xmath40 vanishes . for dissipative solutions
@xmath40 remains constant for all later times , while for the conservative solutions it becomes positive immediately after .
for conservative solutions , the metric is induced by the euclidean norm of the banach space @xmath45 .
for dissipative solutions , we need a metric which in addition can separate @xmath55 and @xmath56 after collision .
indeed , if we denote by @xmath57 the collision time , the distance between @xmath58 and @xmath59 with respect with this metric has to be big , for any @xmath60 .
this is taken care of by the introduction of a function @xmath61 in definition [ def : omega ] , and the corresponding metric @xmath62 in definition [ def : metric1 ] .
the metric yields the flow lipschitz continuous in the sense that @xmath63 for initial data in the set @xmath64 , given by , see lemma [ lem : lip ] .
having obtained the solution in lagrangian coordinates , the next task is to transfer the solution back to eulerian variables @xmath16 .
here we are confronted with relabeling issue ; there are several distinct solutions in lagrangian variables corresponding to one and the same eulerian functions @xmath16 .
this is the reminiscent of the fact that there are many distinct ways to parametrize the graph of a given function .
we identify the functions that give the same eulerian solution , and show that the semigroup in lagrangian coordinates respects the relabeling in the sense that @xmath65 where @xmath66 denotes the semigroup of solutions and @xmath67 denotes the label .
we define , for any @xmath26 such that @xmath68 ( cf . theorem [ th : umudef ] ) , @xmath69 we measure the distance between two eulerian solutions by their corresponding lagrangian distance , see .
the interplay between dissipative and conservative solutions is interesting . as shown in @xcite a positive density @xmath70 regularizes the function @xmath11 .
if @xmath70 is positive on the whole line , no wave breaking will take place , and conservative and dissipative solutions coincide . on the other hand ,
if @xmath70 is identically zero , then @xmath11 will satisfy the scalar ch equation , and we will have wave breaking generically .
a local version of this result is that if @xmath70 is positive on an interval , then the solution will remain regular , i.e. , no wave breaking will take place in the interval bounded by the corresponding characteristics . in @xcite
we have shown that one can obtain conservative solutions of the ch equation by considering solutions of the 2ch system with positive density @xmath6 . if one lets the initial density approach zero appropriately , then the solution @xmath11 will converge to the conservative solution of the ch equation .
however , we do show continuity results for dissipative solutions of the 2ch system , cf .
lemmas [ lemma : kont ] and [ lemma : kont1 ] .
these results are discussed in section [ sec : stabil ] .
we consider the cauchy problem for the two component camassa holm system with arbitrary @xmath0 and @xmath1 , given by [ eq : chsys2 ] @xmath71 with initial data @xmath3 and @xmath4 .
we are interested in global solutions for initial data @xmath72 with nonvanishing and possibly distinct limits at infinity , that is , @xmath73 furthermore we assume that the initial density has equal asymptotics which need not to be zero , that is , @xmath74 more precisely , we introduce the spaces @xmath75 where @xmath34 denotes a smooth partition function with support in @xmath76 such that @xmath77 for @xmath78 and @xmath79 for @xmath80 , and @xmath81 subsequently , we will assume that @xmath82 introducing the mapping @xmath83 from @xmath84 into @xmath85 given by @xmath86 for any @xmath87 , yields that any initial condition @xmath88 is defined by an element in @xmath89 through the mapping @xmath90 .
hence we see that @xmath91 is the image of @xmath84 by @xmath90 , that is , @xmath92 .
the linear mapping @xmath90 is injective .
we equip @xmath93 with the norm @xmath94 where @xmath95 .
then @xmath93 is a banach space . given another partition function @xmath96 ,
we define the mapping @xmath97 from @xmath84 to @xmath84 as @xmath98 , @xmath99 and @xmath100 the linear mapping @xmath101 is a continuous bijection . since @xmath102 we can see that the definition of the banach space @xmath93 does not depend on the choice of the partition function @xmath34 .
the norm defined by for different partition functions @xmath34 are all equivalent . similarly , one can associate to any element @xmath103 the unique pair @xmath104 through the mapping @xmath105 from @xmath106 to @xmath107 which is defined as @xmath108 in fact @xmath105 is bijective from @xmath106 to @xmath107 , which allows us to equip @xmath107 with the norm @xmath109 where we decomposed @xmath6 according to @xmath110 .
thus @xmath107 together with the norm defined in is a banach space .
note that for smooth solutions , we have the following conservation law @xmath111 moreover , if @xmath112 is a solutions of the two - component camassa holm system , then , for any constant @xmath113 we easily find that @xmath114 solves the two - component camassa
holm system with @xmath115 replaced by @xmath116 and @xmath15 .
therefore , without loss of generality , we assume in what follows , that @xmath117 and @xmath15 .
in addition , we only consider the case @xmath10 as one can make the same conclusions for @xmath118 with slight modifications .
in this section we will introduce the set of lagrangian coordinates we want to work with and the corresponding banach spaces .
the 2ch system with @xmath10 can be rewritten as the following system nonzero is simply replaced by @xmath119 . ]
@xmath120 where @xmath121 and @xmath122 are given by @xmath123 and @xmath124 a close inspection of @xmath125 , like in @xcite , reveals that the asymptotic behavior has to be preserved .
thus we write here and later @xmath126 and @xmath127 until wave breaking occurs for the first time every dissipative solution coincides with the conservative one , and hence they can be described in the same way . therefore we summarize the derivation of the lagrangian coordinates and the corresponding system of ordinary differential equations , which describe the conservative solutions here . for details
we refer to @xcite .
afterwards , in the next subsection , we will adapt the system describing the time evolution to the dissipative case .
define the characteristics @xmath22 as the solution of @xmath128 for a given @xmath24 .
the lagrangian velocity is given by @xmath25 and we find using that @xmath129 where @xmath130 is given by @xmath131 where we have introduced @xmath132 the time evolution of @xmath133 is given by @xmath134 where , slightly abusing the notation , @xmath135 is given by @xmath136 last but not least , according to , @xmath137 is preserved with respect to time , i.e. , @xmath138 . wave breaking for the 2ch system means that @xmath8 becomes unbounded , which is equivalent , in the lagrangian setting , to saying that @xmath40 becomes zero .
let therefore @xmath139 be the first time when @xmath140 vanishes , i.e. , @xmath141 where @xmath142 .
the dissipative solutions will then be described through the solutions of the following system of ordinary differential equations [ eq : sysdiss ] @xmath35 where @xmath36 and @xmath37 are given by ( observe the subtle modifications in the dissipative case compared with and , respectively ) @xmath143 and @xmath144 respectively .
( the integrals over the real line in and could be replaced by @xmath145 since @xmath146 outside this domain . )
we introduce the following notation for the banach spaces we will often use . let @xmath147 together with the norm @xmath148 and @xmath149 for any function @xmath150 , b)$ ] for @xmath151 and @xmath152 a normed space , we denote @xmath153}{\left\vertf(t,{\ , \cdot\,})\right\vert}_b.\ ] ] [ def : omega ] for @xmath154 , we define the functions @xmath155 by @xmath156 and @xmath157 where @xmath158 is the set where @xmath159 , @xmath160 is negative , and @xmath161 , thus @xmath162 furthermore , we will split up @xmath163 as follows .
@xmath164 is the complement of @xmath158 restricted to the set where @xmath161 , that is , @xmath165 and @xmath166 is the set of all points such that @xmath167 , that is , @xmath168 note the following obvious relations : @xmath169 see figure [ fig : omega_omrl ] .
as long as we are working in lagrangian coordinates we will identify @xmath170 with @xmath171 .
[ def : g ] the set @xmath50 consists of all @xmath172 such that [ eq : lagcoord ] @xmath173 where we denote @xmath174
. the condition will be valid as long as the solutions exist since in that case we must have @xmath175 by construction .
in addition it should be noted that , due to the definition of @xmath176 , is valid for any @xmath44 that satisfies .
making the identifications @xmath177 and @xmath47 , we obtain [ odesys ] @xmath178 where @xmath36 and @xmath37 are given by @xmath179 and @xmath180 respectively .
the definition of @xmath181 given by ( after replacing @xmath40 by the corresponding variable @xmath38 ) is not appropriate for @xmath182,{{l^\infty({\mathbb r})}})$ ] , and , in addition , it is not clear from this definition whether @xmath181 is measurable or not .
that is why we replace this definition by the following one .
let @xmath183 be a dense countable subset of @xmath184 $ ] .
define @xmath185 the sets @xmath186 are measurable for all @xmath48 , and we have @xmath187 for @xmath188 .
we consider a dyadic partition of the interval @xmath184 $ ] ( that is , for each @xmath189 , we consider the set @xmath190 ) and set @xmath191 where @xmath192 is the indicator function of the set @xmath193 .
the function @xmath194 is by construction measurable .
one can check that @xmath195 is increasing with respect to @xmath189 , it is also bounded by @xmath196 .
hence , we can define @xmath197 and @xmath181 is a measurable function .
the next lemma gives the main property of @xmath181 .
[ lem:2.3 ] if , for every @xmath198 , @xmath199 is positive and continuous with respect to time , then @xmath200 that is , we retrieve the definition .
( from @xcite . ) we denote by @xmath201 the right - hand side of , and we want to prove that @xmath202 . we claim that @xmath203 if @xmath204 , then @xmath205}q(t',\xi)>0 $ ] because @xmath38 is continuous in time and positive . hence , there exists an @xmath189 such that @xmath205}q(t',\xi)>\frac1n$ ] , and we find that @xmath206 in order that @xmath207 .
if @xmath208 , then there exists a sequence @xmath209 of elements in the dense family @xmath210 of @xmath184 $ ] such that @xmath211 and @xmath212 . since @xmath199 is continuous , @xmath213 and for any integer @xmath214 , there exists a @xmath215 such @xmath216 and @xmath217 .
hence , for any @xmath214 , @xmath218 and therefore @xmath219 .
when @xmath220 , for any @xmath214 , there exists @xmath221 such that @xmath222 . from
, we infer that @xmath223 .
hence , @xmath224 , so that @xmath225 letting @xmath189 tend to infinity , we conclude that @xmath226 . if @xmath227 , then @xmath219 for all @xmath228 and @xmath229 for all @xmath189
hence , @xmath230 .
so far we have identified @xmath38 with @xmath40 .
however , @xmath40 does not decay fast enough at infinity to belong to @xmath231 , but @xmath232 will be in @xmath231 and we therefore introduce @xmath233 . in the case of conservative solutions , we know that @xmath37 and @xmath36 are lipschitz continuous on bounded sets and that @xmath37 and @xmath36 can be bounded by a constant depending on the bounded set .
a slightly different result is true when describing dissipative solutions .
let @xmath234 according to our system of ordinary differential equations it seems natural to impose for the solution space that if wave breaking occurs , then the functions @xmath38 , @xmath39 , @xmath235 , and @xmath236 should remain unchanged afterwards , that means @xmath237 , @xmath238 , and @xmath239 for all @xmath240 and @xmath241 .
moreover , also the asymptotic behavior is preserved , i.e. , @xmath242 and @xmath243 . in what follows
we will always assume that these properties are fulfilled for any @xmath244 , b_m)$ ] without stating it explicitly .
in addition it should be pointed out that for any @xmath244,b_m)$ ] the set of all points which enjoy wave breaking within a finite time interval @xmath184 $ ] is bounded , since @xmath245 where @xmath246 denotes some constant only depending on @xmath247 .
[ lem : pq ] ( i ) for all @xmath244,b_m)$ ] , we have @xmath248 for a constant @xmath246 which only depends on @xmath247 .
+ ( ii ) for any @xmath44 and @xmath249 in @xmath250,b_m)$ ] , we have @xmath251 where @xmath252 here
@xmath246 denotes a constant which only depends on @xmath247 .
we will only establish the estimates for @xmath36 as the ones for @xmath37 can be obtained using the same methods with only slight modifications .
the main tool for proving the stated estimates will be young s inequality which we recall here for the sake of completeness .
for any @xmath253 and @xmath254 with @xmath255 , we have @xmath256 ( _ i _ ) : let @xmath257
. then we have @xmath258)(t,\xi)\right\vert}_{l^\infty_te}\\ & \qquad\leq\frac12 e^{{\left\vert\zeta\right\vert}_{l^\infty_tl^\infty}}{\left\vert(f\star [ \chi_{\{\tau(\xi)>t\}}e^\zeta(2c\chi\circ y \bar
u q+ \bar u^2q+\frac12 h+k\bar r)])(t,\xi)\right\vert}_{l^\infty_te}\\ & \qquad\leq c(m ) ( { \left\vertf\right\vert}_{l^1}+{\left\vertf\right\vert}_{l^2}){\left\verte^\zeta(2c\chi\circ y \bar uq+\bar u^2q+\frac12 h+k\bar r)\right\vert}_{l^\infty_tl^2}\\ & \qquad\leq c(m).\end{aligned}\ ] ] similarly , it follows that @xmath259\star [ \chi_{\{\tau(\xi)>t\}}e^{-\zeta}(2c\chi\circ y \bar u q+\bar u^2q+\frac12 h+k\bar r)]\big)(t,\xi)\right\vert}_{l^\infty_te}\leq c(m).\ ] ] analogously one can investigate the other integral term . indeed , since @xmath260 , we have @xmath261 .
the support of @xmath262 is contained in @xmath263 $ ] and this means that the support of @xmath264 is contained in the set @xmath265 . inserting this into @xmath261
, we get that @xmath266 . using that we obtain @xmath267 together with the fact that a similar estimate holds for the @xmath231-norm of @xmath268 .
it follows immediately that @xmath269 and @xmath270 both belong to @xmath231 and that they can be bounded by a constant only depending on @xmath247 . this finishes the proof of the first part .
( _ ii _ ) : the only term which can not be investigated like in ( _ i _ ) is given by the integral term with domain of integration @xmath271 .
let @xmath257 as before , and write @xmath272 .
then we can write @xmath273 we estimate each of these terms separately . the first and the third term are similar ,
thus we only treat the first one .
we obtain @xmath274 the second term can be treated in much the same way as the fourth one .
we have @xmath275 .
introduce @xmath276 with @xmath277 and @xmath278 .
then @xmath279 after applying fubini s theorem in the last step , which is possible since the set of points which enjoy wave breaking within the time interval @xmath184 $ ] is bounded .
finally @xmath280 to show the short - time existence of solutions we will use an iteration argument for the following system of ordinary differential equations . denote generically @xmath281 by @xmath44 and @xmath282 by @xmath283 , thus @xmath284 .
then we define the mapping @xmath285 , \bar v)\to c([0,t ] , \bar v)\ ] ] as follows : given @xmath44 in @xmath250 , b_m)$ ] , we can compute @xmath36 and @xmath37 using and .
then @xmath286 is given as the modified solution with @xmath287 of the following system of ordinary differential equations [ eq : sysfix2 ] @xmath288 next , we modify @xmath249 as follows : determine the function @xmath289 according to ( with @xmath44 replaced by @xmath249 ) .
subsequently , we modify the function @xmath249 by setting @xmath290 observe that the first two components of @xmath249 , given by , remain unmodified .
we write @xmath291 .
thus we set @xmath292 for @xmath293 .
we will in the following , to keep the notation reasonably simple , often write @xmath44 for the vector @xmath294 , or even @xmath295 , and correspondingly for @xmath249 .
we will frequently consider the following spaces .
for @xmath296 and @xmath297 , we define @xmath298 the following set @xmath299,\ ] ] will play a key role in the context of wave breaking .
( for a motivation on the set @xmath300 , please see the paragraph before lemma [ lem : g2 ] . ) in particular , we have that @xmath301 and therefore the set @xmath302 has finite measure if we choose @xmath303 $ ] , and , in particular , @xmath304 .
[ lem : g ] given @xmath305 for some constant @xmath306 , given @xmath307 , b_m)$ ] , we denote by @xmath308 with initial data @xmath52 .
let @xmath309 .
then the following statements hold : ( _ i _ ) for all @xmath48 and almost all @xmath26 @xmath310 and @xmath311 thus , @xmath312 implies @xmath313 and @xmath314 .
we recall the notation @xmath315 .
( _ ii _ ) we have @xmath316 and @xmath317 for all @xmath318 $ ] and a constant @xmath319 which depends only on @xmath320 .
in particular , @xmath321 remains bounded strictly away from zero .
( _ iii _ ) there exists a @xmath322 depending only on @xmath320 and @xmath196 such that if @xmath323 , then @xmath324 for all @xmath325 $ ] , @xmath326 is a decreasing function and @xmath327 is an increasing function with respect to time , and therefore we have @xmath328 in addition for @xmath329 sufficiently small , depending only on @xmath320 and @xmath196 , we have @xmath330 ( _ iv
_ ) moreover , for any given @xmath322 , there exists @xmath331 such that @xmath332 ( _ i _ ) since @xmath333 , equations and hold for almost every @xmath198 at @xmath334 .
we consider such a @xmath26 and will drop it in the notation . from , we have , on the one hand , @xmath335 and , on the other hand , @xmath336 thus , @xmath337 , and since @xmath338 , we have @xmath339 for all @xmath318 $ ] .
we have proved . from the definition of @xmath340
, we have that @xmath341 on @xmath342 and by the definition of @xmath343 , we have @xmath344 for @xmath345 .
hence , @xmath346 for @xmath228 . from
, it follows that , for @xmath347 , @xmath348 and therefore @xmath349 . by continuity
( with respect to time ) of @xmath350 , we have @xmath351 and , since the variable does not change for @xmath345 , we have @xmath349 for all @xmath228 .
( _ ii _ ) we consider a fixed @xmath26 that we suppress in the notation .
we denote the euclidean norm of @xmath352 by @xmath353 . since @xmath354
, we have @xmath355 for a constant @xmath319 which depends only @xmath320 . applying gronwall s lemma
, we obtain @xmath356 .
hence , @xmath357 using , we have @xmath358 hence , yields @xmath359 the second claim can be shown similarly .
( _ iii _ ) let us consider a given @xmath323 .
we are going to determine an upper bound on @xmath329 depending only on @xmath320 and @xmath196 such that the conclusions of ( _ iii _ ) hold .
for @xmath329 small enough we have @xmath360 as otherwise @xmath361 and @xmath362 would lead to a contradiction since @xmath363 .
we claim that there exists a constant @xmath364 depending only on @xmath320 and @xmath196 such that for all @xmath365 , @xmath198 , and @xmath325 $ ] , @xmath366 and @xmath367 we consider a fixed @xmath198 and suppress it in the notation .
if @xmath368 , then yields @xmath369 , where we used that @xmath370 and @xmath371 .
thus either @xmath344 or @xmath372 .
assume that @xmath373 , then @xmath372 .
hence @xmath374 , and we are led to a contradiction if we choose @xmath329 small enough .
hence , @xmath375 , and we have proved .
+ if @xmath376 , we have @xmath377 where we used that @xmath370 and @xmath371 .
( recall that we allow for a redefinition of @xmath378 . ) by choosing @xmath379 , we get @xmath380 , and we have proved . + for any @xmath381 , we consider a given @xmath26 in @xmath302 and again suppress it in the notation .
we define @xmath382\mid \frac{\tilde q}{\tilde q+\tilde h}(t')<2\gamma \text { and } \tilde
w(t')<0 \text { for all } t'\leq t\}.\ ] ] let us prove that @xmath383 .
assume the opposite , that is , @xmath384 .
then , we have either @xmath385 or @xmath386 .
we have @xmath387 on @xmath388 $ ] and @xmath389 is decreasing on this interval . hence , @xmath390 , and therefore we must have @xmath391 . then , implies @xmath392 , and therefore @xmath393 , which contradicts our assumption . from , we get , for @xmath329 sufficiently small , @xmath394 and therefore @xmath395 . by taking @xmath329 small enough we can impose @xmath396 , which proves .
it is clear from that @xmath397 is increasing .
assume that @xmath398 leaves @xmath158 for some @xmath48 .
then , we get @xmath399 and , by taking @xmath329 small enough , we are led to a contradiction .
( _ iv _ ) without loss of generality we assume @xmath400 . from ( _ iii _ ) we know that there exists a @xmath401 only depending on @xmath320 and @xmath196 such that for @xmath402 , @xmath403 and in particular we have that the function @xmath404 is decreasing and @xmath397 is an increasing function both with respect to time on @xmath184 $ ] .
let @xmath405 .
we consider a fixed @xmath198 such that @xmath406 ( which means implicitly @xmath407 for all @xmath48 ) , but @xmath408 .
let us introduce @xmath409 since @xmath410 and @xmath411 , the definition of @xmath412 is well - posed when @xmath413 , and we have @xmath414 . by assumption @xmath415 and @xmath391 or @xmath416 .
we can not have @xmath391 , since it would imply , see , that @xmath392 and therefore @xmath393 which is not possible .
thus we must have @xmath417 and in particular @xmath418 . according to the choice of @xmath419 we have that @xmath420 for all @xmath421 and @xmath422 is increasing .
then we have @xmath423 which yields for @xmath424 @xmath425 since @xmath426 , we choose @xmath427 such that @xmath428 . thus @xmath429 and therefore all points which enjoy wave breaking before @xmath427 are contained in @xmath430 , since any point entering @xmath430 at a later time can not reach the origin within the time interval @xmath431 $ ] according to the last estimate .
[ lem : estshort ] given @xmath432 , there exists @xmath433 and @xmath320 such that for all @xmath434 and any initial data @xmath435 , @xmath436 is a mapping from @xmath250,b_{\bar m})$ ] to @xmath250 , b_{\bar m})$ ] . to simplify the notation
, we will generically denote by @xmath437 and @xmath319 increasing functions of @xmath247 and @xmath320 , respectively . without loss of generality , we assume @xmath438 .
let @xmath244 , b_{\bar m})$ ] for a value of @xmath320 that will be determined at the end as a function of @xmath247 .
we assume without loss of generality @xmath439 .
let @xmath286 . from lemma [ lem : pq ]
, we have @xmath440 since @xmath441 and @xmath442 , we get @xmath443 we use that @xmath444 to deduce that @xmath445 since , @xmath446 , we get @xmath447 moreover , @xmath448 , we have @xmath449 by . from , by the minkowsky inequality for integrals ,
we get [ est : l1 ] @xmath450 these inequalities imply that @xmath451 and , applying gronwall s inequality yields @xmath452 gathering , , , and , we get @xmath453 from we get @xmath454 thus we finally obtain @xmath455 for some constants @xmath437 and @xmath319 that only depend on @xmath247 and @xmath320 , respectively .
we now set @xmath456 .
then we can choose @xmath196 so small that @xmath457 and therefore @xmath458 .
given @xmath459 , there exists @xmath320 which depends only on @xmath247 such that @xmath436 is a mapping from @xmath250 , b_{\bar m})$ ] to @xmath250 , b_{\bar m})$ ] for @xmath196 small enough .
therefore we set @xmath460,b_{\bar m})\}.\ ] ] we define the _ discontinuity residual _ as @xmath461 here it should be noted that @xmath462 describes the distance between @xmath181 and @xmath340 as the following estimate shows , @xmath463 according to lemma [ lem : pq ] , we have , @xmath464 in the next lemma we establish some estimates for @xmath462 , @xmath465 and a quasi - contraction property for @xmath436 .
[ lem : contrgamma ] given @xmath44 , @xmath466 and @xmath322 there exists @xmath467 depending on @xmath320 such that the following inequalities hold + ( _ i _ ) @xmath468 ( _ ii _ ) @xmath469 ( _ iii _ ) @xmath470 where @xmath319 denotes some constant which only depends on @xmath320 . denote by @xmath471 and @xmath472 . given @xmath473 we know from lemma [ lem : g ] ( _ iv _ ) that there exists @xmath196 small enough such that @xmath474 and we consider such @xmath196 . without loss of generality we can assume @xmath475 and @xmath476 .
( _ i _ ) let us now consider @xmath477 such that @xmath478 . without loss of generality
we assume @xmath479 . at time @xmath334 ,
@xmath44 and @xmath249 coincide and therefore we can not have @xmath480 because it would imply @xmath481 . hence @xmath482 .
since @xmath483 and @xmath398 both belong to the @xmath484 and @xmath323 , we get that @xmath483 , @xmath324 and especially @xmath485 and @xmath486 .
thus we get from , if @xmath487 , that for @xmath488 $ ] , @xmath489 thus , since @xmath490 and @xmath491 for @xmath240 , we have @xmath492 a similar inequality holds for @xmath493 . since @xmath494 and the only points that contribute to the integral are contained in @xmath302 , we get @xmath495 ( _ ii _ ) we denote @xmath471 and @xmath472 .
let us now consider @xmath477 such that @xmath496 .
without loss of generality we assume @xmath497 . at time @xmath334 ,
@xmath498 and @xmath499 coincide and therefore we can not have @xmath500 because it would imply @xmath501 .
hence @xmath502 .
since @xmath503 and @xmath504 both belong to the @xmath484 and @xmath323 , we get that @xmath503 , @xmath505 and especially @xmath506 and @xmath507 .
thus we get from that for @xmath508 $ ] , @xmath509 since @xmath510 for all @xmath48 . from , we get following @xmath511 where we used that @xmath512 , @xmath513 , and @xmath514 for @xmath515 $ ] .
a corresponding inequality holds for the case @xmath516 .
we have , using again on the interval @xmath517 $ ] , @xmath518 where we used that @xmath519 and @xmath520 .
( _ iii _ ) first we estimate @xmath521 . for @xmath522 , we have @xmath523 and @xmath524 for all @xmath318 $ ] . hence , @xmath525 we have , since @xmath526 for all @xmath48 , @xmath527 and therefore @xmath528 applying gronwall s lemma to , as @xmath529 , we get @xmath530 hence , we get by that @xmath531 thus , we have by that @xmath532 to estimate @xmath533 , we fix @xmath534 and assume without loss of generality that @xmath535 . from lemma
[ lem : g ] , we have that @xmath536 is positive decreasing and @xmath537 is negative decreasing so that @xmath538 for @xmath539 $ ] , and therefore @xmath540 and @xmath541 for @xmath539 $ ] because @xmath542 for @xmath539 $ ] . since @xmath543 , we know that @xmath544 , and therefore @xmath545 implies that @xmath546 decreases on @xmath547 $ ] .
thus @xmath548 for all @xmath549 $ ] since @xmath550 for all @xmath549 $ ] . for @xmath539
$ ] , we have @xmath551 and @xmath552 .
hence , @xmath553 from . for @xmath554 $ ] , we have @xmath555 and @xmath556
. we proceed as in the previous step and in the same way as we obtained , we now obtain @xmath557 and , after using together with , we get @xmath558 combining , , , and , we get @xmath559 for all @xmath318 $ ] .
since @xmath494 , implies @xmath560 combining and , we get @xmath561 from , we obtain @xmath562 and @xmath563 combining the last two inequalities yields @xmath564 finally from we get @xmath565 thus adding up , , , and we have that @xmath566 [ th : short0 ] for any initial data @xmath567 , there exists a time @xmath467 such that there exists a unique solution @xmath568,\bar v)$ ] of with @xmath569 .
moreover @xmath570 for all @xmath318 $ ] . in order to prove the existence and uniqueness of the solution we use an iteration argument .
therefore we set @xmath571 and @xmath572 for all @xmath573 .
this implies that @xmath574 for @xmath575 belongs to @xmath484 .
we have @xmath576 where we used lemma [ lem : contrgamma ] .
hence , for @xmath196 and @xmath329 small enough , we have @xmath577 summation over all @xmath578 on the left - hand side then yields @xmath579 and @xmath580 independently on @xmath581 .
since @xmath582 for all @xmath583 , the series @xmath584 is increasing , bounded from above and hence convergent .
in particular , @xmath585 and therefore @xmath586 is a cauchy sequence and tends to a unique limit @xmath587 .
the continuity of @xmath436 in @xmath588 follows from and .
hence , we obtain that @xmath587 is not only unique but also a fix point of the mapping @xmath436 to the initial condition @xmath52 .
it is left to prove that @xmath589 and @xmath177 .
recall that @xmath37 is defined via and @xmath37 is differentiable if and only if @xmath590 is differentiable .
a formal computation gives us that @xmath591 and @xmath592\times{\mathbb r})$ ] if @xmath593 .
in addition , as @xmath594 and @xmath595 , we have [ eq : qqxi ] @xmath596 this means in particular if @xmath597 and @xmath598 , that @xmath599 and thus using gronwall s inequality yields that @xmath177 and @xmath589 .
let us prove that @xmath570 for all @xmath48 . from and , we get @xmath600 , @xmath601 and @xmath602 for all @xmath48 and almost all @xmath26 and therefore , since @xmath589 and @xmath177 , the conditions and are fulfilled . since @xmath603 , we obtain by the lebesgue dominated convergence theorem that @xmath604 because @xmath605 for @xmath606 and @xmath607 .
hence , since in addition @xmath608 , @xmath587 fulfills all the conditions listed in and @xmath570 .
[ rem : dertime ] the set @xmath609 is closed with respect to the topology of @xmath45 .
we have @xmath610 for all @xmath198 and @xmath611 , since @xmath612 for @xmath613 .
this means in particular , that @xmath40 , @xmath236 , and @xmath235 are differentiable almost everywhere with respect to time ( in the classical sense ) .
we have that @xmath614 is a fixed point of @xmath615 , and the results of lemma [ lem : g ] hold for @xmath616 . since this
lemma is going to be used extensively we rewrite it for the fixed point solution @xmath44 .
for this purpose , we redefine @xmath64 and @xmath302 , see and , as @xmath617 with @xmath618 , @xmath619.\ ] ] note that every condition imposed on points @xmath323 is motivated by what is known about wave breaking .
if wave breaking occurs at some time @xmath620 energy is concentrated on sets of measure zero in eulerian coordinates , which correspond to the sets where @xmath621 in lagrangian coordinates .
furthermore , it is well- known that wave breaking in the context of the 2ch system means that the spatial derivative becomes unbounded from below and hence @xmath622 for @xmath623 for such points , see @xcite .
finally , it has been shown in ( * ? ? ?
* theorem 6.1 ) that wave breaking within finite time can only occur at points @xmath26 where @xmath624 .
recall that @xmath176 denotes @xmath625 .
lemma [ lem : g ] rewrites as follows .
[ lem : g2 ] let @xmath306 be a constant , and consider initial data @xmath626 .
denote the solution of with initial data @xmath52 by @xmath627 , b_m)$ ] .
introduce @xmath628 .
then the following statements hold : ( _ i _ ) we have @xmath629 and @xmath630 for all @xmath318 $ ] and a constant @xmath319 which depends on @xmath320 .
( _ iii _ ) there exists a @xmath322 depending only on @xmath320 and @xmath196 such that if @xmath323 , then @xmath631 for all @xmath325 $ ] , @xmath632 is a decreasing function and @xmath633 is an increasing function , both with respect to time , and therefore we have @xmath634 in addition , for @xmath329 sufficiently small , depending only on @xmath320 and @xmath196 , we have @xmath635 ( _ iv
_ ) moreover , for any given @xmath322 , there exists @xmath331 such that @xmath636 to prove global existence of the solution we will use the estimate contained in the following lemma .
[ lem : globest ] given @xmath637 and @xmath638 , there exists a constant @xmath247 which only depends on @xmath306 and @xmath639 such that , for any @xmath640 , we have @xmath641 for all @xmath318 $ ] , where @xmath587 denotes the short time solution on @xmath184 $ ] with @xmath642 given by theorem [ th : short0 ] for initial data @xmath52 . to simplify the notation we will generically denote by @xmath643 constants which only depend on @xmath34 , @xmath644 , and @xmath645 , and by @xmath646 constants which in addition depend on @xmath306 and @xmath639 .
let us introduce @xmath647 since @xmath648 and @xmath649 , we have @xmath650 .
we can estimate the @xmath651 as follows : [ eq : derivulinf ] @xmath652 where we used that @xmath653 implies @xmath654 and therefore in the first integral in the second line the integrand is zero whenever @xmath655 .
thus it suffices to integrate over @xmath656 which justifies the subsequent estimate . finally , inserting that @xmath657 , we get @xmath658 from , we get @xmath659 similarly one obtains @xmath660 we can now compute the derivative of @xmath661 . from
we get @xmath662 note that we can put the time derivative under the integral , since @xmath40 and @xmath235 are differentiable almost everywhere with respect to time ( cf .
remark [ rem : dertime ] ) .
we estimate each of these integrals separately .
thus @xmath663 after integration by parts in the last step , since @xmath664 .
thus @xmath665 furthermore @xmath666 finally , by adding these estimates , we get @xmath667 by and .
hence , gronwall s lemma implies that @xmath668}\sigma(t)$ ] can be bounded by some constant only depending on @xmath306 and @xmath639 .
using now , , and , we immediately obtain that the same is true for @xmath669 , @xmath670 , and @xmath671 with @xmath672 $ ] . from , we obtain that @xmath673 and hence also @xmath674 can be bounded on @xmath675 $ ] by a constant only depending on @xmath306 and @xmath639 .
applying young s inequality to and and following the proof of lemma [ lem : pq ] we get @xmath676 let @xmath677 then @xmath678 hence gronwall s lemma gives us @xmath679 . similarly , one can show that @xmath680 it remains to prove that @xmath681 can be bounded by some constant depending on @xmath306 and @xmath639 , but this follows immediately form .
this completes the proof .
we can now prove global existence of solutions .
[ th : global ] for any initial data @xmath567 , there exists a unique global solution @xmath682 of with @xmath569 . by assumption @xmath333 , and therefore there exists a constant @xmath306 such that @xmath683 . by theorem
[ th : short0 ] there exists a @xmath467 such that we can find a unique short time solution @xmath684 on @xmath184 $ ] .
moreover , according to lemma [ lem : estshort ] , the length of the time interval for which the solution exists and is unique , is linked to @xmath306 .
thus we can only find a unique global solution if @xmath685 does not blow up within a finite time interval , but this follows from lemma [ lem : globest ] .
[ def : metric1 ] the mapping @xmath686 @xmath687 for @xmath44 , @xmath688 defines a metric on @xmath50 .
the function @xmath689 is defined as follows @xmath690 in what follows we will denote @xmath691 with @xmath692 here it should be noted that for any two solutions @xmath44 and @xmath249 of the 2ch system , @xmath689 and @xmath693 are independent of time , since @xmath138 and @xmath694 .
denote @xmath695 note that @xmath696 ( cf . definition [ def : omega ] ) .
see figure [ fig : omega_omrl ] . and .,title="fig:",width=302 ] [ fig : omega_omrl ] [ lem : g1 ] the restrictions of @xmath61 to @xmath164 , @xmath166 , and @xmath697 are lipschitz on bounded sets .
more precisely , @xmath698 for any @xmath44 , @xmath249 in @xmath699 , @xmath44 , @xmath249 in @xmath700 , or @xmath44 , @xmath249 in @xmath701 .
the cases when both @xmath44 , @xmath702 , @xmath44 , @xmath703 , and @xmath44 , @xmath704 are straightforward .
let us consider the case when @xmath705 and @xmath706 , that is , @xmath707 and @xmath708 , respectively .
we have @xmath709 [ lem : estg ] given @xmath432 , there exist @xmath710 , @xmath467 and @xmath711 which depend only on @xmath247 such that for any @xmath198 satisfying @xmath712 , we have @xmath713 for all @xmath318 $ ] where @xmath587 and @xmath714 are the solutions to with initial data @xmath52 and @xmath715 belonging to @xmath64 . without loss of generality
we assume @xmath475 .
we already know that there exists @xmath320 only depending on @xmath247 such that @xmath587 , @xmath716 for all @xmath717 $ ] .
we consider @xmath198 such that @xmath718 for a @xmath719 that we are going to determine .
for simplicity we drop @xmath720 in the notation from now on .
since @xmath587 and @xmath714 are solutions of , we see that due to lemma [ lem : g1 ] the estimate will be proved if we can show that either @xmath587 and @xmath714 belong to @xmath697 , @xmath587 and @xmath714 belong to @xmath164 , or @xmath587 and @xmath714 belong to @xmath166 . if @xmath721 and @xmath722 , by , we have @xmath723 .
thus by choosing @xmath724 , we impose that either @xmath725 or @xmath726 .
in addition , since @xmath727 and @xmath728 , we can conclude that points which initially lie inside @xmath166 will remain in @xmath166 for all times and points starting outside @xmath166 can never enter @xmath166 . hence in order to verify the claim it is left to show that either @xmath587 and @xmath714 belong to @xmath164 or @xmath587 and @xmath714 belong to @xmath697 . to this end
denote by @xmath729 and @xmath730 the following sets ( cf .
figure [ fig : omega_omrl ] ) @xmath731 we will distinguish three cases : ( _ i _ ) @xmath52 and @xmath715 in @xmath729 : since @xmath732 , we infer that @xmath52 and @xmath715 belong to @xmath697 .
hence @xmath587 and @xmath733 will satisfy as long as both @xmath587 and @xmath714 belong to @xmath697 .
let us prove that @xmath587 ( and , in the same way , @xmath714 ) for @xmath196 small enough remains in @xmath697 .
denote by @xmath412 the first time when @xmath587 leaves @xmath697 . by continuity
we must have @xmath734 .
since @xmath735 for all @xmath48 in @xmath184 $ ] , it implies @xmath736 , by and the definition of @xmath737 .
hence , as the origin belongs to @xmath697 , either @xmath738 or @xmath739 .
points reaching the origin remain there , that is , if @xmath740 for @xmath741 then it remains true for @xmath421 . hence we infer that , @xmath742 and @xmath739 .
let @xmath743 .
we have , by assumption , @xmath744 , because @xmath745 and @xmath746 . using that @xmath747 and @xmath739 , @xmath748 . on the other hand we can compute @xmath749 and we obtain @xmath750 for some constant @xmath751 only depending on @xmath320 and therefore on @xmath247 .
thus @xmath752 .
hence if we choose @xmath196 small enough , that means @xmath753 , we obtain @xmath754 , which is a contradiction and we have proved that @xmath587 remains in @xmath697 .
similarly one proves that @xmath714 remains in @xmath697 .
@xmath755 and @xmath756 : first of all we want to make sure that holds at time @xmath334 .
according to lemma [ lem : g1 ] , this will be the case if we can prove that @xmath757 implies either @xmath52 , @xmath758 or @xmath52 , @xmath759 . if @xmath760 , then holds since @xmath761 . if @xmath762 , we have @xmath763 so that @xmath764 .
assume that @xmath765 .
then @xmath766 implies @xmath767 and @xmath768 , as @xmath769 and @xmath770 have opposite signs . since @xmath755
, we have @xmath771 which implies using that @xmath772 thus @xmath773 , @xmath774 and we have @xmath775 taking @xmath719 sufficiently small , we are led to a contradiction .
hence @xmath776 .
we have already seen in ( _ i _ ) that we can choose @xmath196 so small that @xmath587 remains in @xmath697 for @xmath318 $ ] .
let us denote @xmath777 .
for @xmath778 as defined in ( _ i _ ) , we have @xmath744 . hence @xmath779 and therefore @xmath780 .
let us now consider the first time @xmath412 when @xmath781 leaves @xmath697 .
again , as in ( _ i _ ) , we obtain that @xmath782 and @xmath783 .
in addition we know @xmath784 , which leads to a contradiction if we choose @xmath196 and @xmath719 small enough . ( _ iii _ ) @xmath52 and @xmath715 in @xmath730 : in this case , since @xmath785 , @xmath52 and @xmath715 belong to @xmath164 .
we have @xmath786 .
let us prove that @xmath587 ( and , in the same way , @xmath714 ) for @xmath196 small enough remains in @xmath164 .
note that because the origin is a repulsive point , solutions can not reach the origin from within @xmath164 .
this means in particular that @xmath587 can only leave @xmath164 if it starts in @xmath787 or after entering @xmath788 .
therefore we assume without loss of generality @xmath789 .
denote by @xmath412 the first time when @xmath587 leaves @xmath790 .
then we must have @xmath791 which gives @xmath792 .
hence we obtain after some computations using the latter equation together with and @xmath793 that @xmath794 .
moreover , @xmath795 which implies @xmath796 , which leads to a contradiction if we choose @xmath196 small enough .
we have thus proved that @xmath587 remains in @xmath790 and the same result holds for @xmath714 .
a close look at the proof of the last lemma shows that @xmath719 is only turning up in ( _ ii _ ) and can be bounded by a constant only depending on @xmath320 .
hence , lemma [ lem : estg ] can be extended to the following lemma .
[ lem : est : diffg ] given @xmath432 and @xmath467 , then the solutions @xmath587 and @xmath714 with initial data @xmath52 and @xmath715 in @xmath64 , respectively , belong to @xmath797 for @xmath318 $ ] . for all @xmath798 $ ] there exist @xmath799 $ ] and @xmath719 positive depending on @xmath320 and independent of @xmath800 , such that for any @xmath198 satisfying @xmath801 , we have @xmath802 for all @xmath803\cap [ 0,t]$ ] . the next lemma shows why the function @xmath61 is so important and why we choose exactly the metric @xmath62 instead of the norm @xmath804 we used in the last section .
[ lem : tauttau ] given @xmath432 and @xmath467 , then the solutions @xmath587 and @xmath714 with initial data @xmath52 and @xmath715 in @xmath64 , respectively , belong to @xmath797 and we have for any @xmath198 the following estimates : + ( i ) if @xmath805 @xmath806 ( ii ) if @xmath807 @xmath808 for a constant @xmath319 depending only on @xmath320 .
we assume without loss of generality @xmath479 .
we distinguish three cases .
+ ( _ i _ ) if @xmath809 for @xmath810 $ ] , we know @xmath811 and @xmath812 on @xmath813 $ ] because @xmath814 so that @xmath403 .
thus we get @xmath815 as already pointed out several times before , @xmath714 remains in @xmath166 for all times if it starts in @xmath166 .
it is left to show what happens if @xmath816 for all times .
the result will follow from the following two estimates
. + ( _ ii _ )
if @xmath817 for @xmath818 $ ] , we know @xmath811 and @xmath812 on @xmath813 $ ] .
hence we have @xmath815 ( _ iii _ ) if @xmath819 , we know that @xmath820 . thus implies that @xmath821 since @xmath820 and @xmath822 for all @xmath810 $ ] , we get @xmath823 in general @xmath824 , does not imply that @xmath817 for all @xmath818 $ ] .
therefore we define @xmath825 as @xmath826 \mid x(t',\xi ) \in\omega_+ \text { for all } t'\geq t\}.\ ] ] then , @xmath817 for all @xmath827 $ ] and @xmath828 so that the general case is proved by combining the cases ( _ ii _ ) and ( _ iii _ ) . [
thm : shortstab ] given @xmath432 there exist constants @xmath438 and @xmath829 depending only on @xmath247 such that for any initial data @xmath52 and @xmath715 in @xmath64 @xmath830}d_{\mathbb r}(x(t ) , \tilde x(t))\leq kd_{\mathbb r}(x_0,\tilde x_0).\ ] ] we assume without loss of generality @xmath475 .
we already know that there exists @xmath320 only depending on @xmath247 ( since @xmath475 ) such that @xmath587 and @xmath714 belong to @xmath797 for all @xmath318 $ ] .
let us introduce for the moment the following metric @xmath831 then @xmath832.\ ] ] this means in particular that the theorem is proved once we show that @xmath833 we estimate each of the terms in .
first , we want to show that @xmath834 to this end we first observe that by following closely the proof of lemma [ lem : pq ] one can show that @xmath835 the first step towards proving the claim is to estimate properly the integral term on the right - hand side .
applying lemma [ lem : tauttau ] yields @xmath836 where we in the last step used that @xmath196 is chosen so small that all points such that @xmath837 or @xmath838 belong to @xmath839 .
this is possible according to lemma [ lem : g2 ] ( _ iv _ ) .
fix @xmath198 such that @xmath482
. then @xmath840 and @xmath841 for @xmath842 $ ] .
hence @xmath843 here we used that for @xmath477 , we have @xmath844 . in the last estimate we applied . for the case
@xmath493 , a similar treatment as in yields an estimate of the same form with every @xmath181 replaced by @xmath340 and vice versa . inserting this estimate and the estimate into implies , since @xmath845 , that @xmath846 choosing @xmath329 small enough we find that @xmath847 and especially @xmath848 which is .
we now return to the proof of , where we carefully investigate all terms in @xmath849 separately using .
the equations @xmath850 and @xmath851 imply @xmath852 and @xmath853 yields @xmath854 since @xmath855 and @xmath856 , after combining , and , we obtain @xmath857 to estimate @xmath858 we split @xmath859 into two sets .
let us introduce @xmath860 . for @xmath861 , we have @xmath862 and @xmath863 for all @xmath318 $ ] .
thus @xmath864 , and @xmath865 we get after applying gronwall s lemma @xmath866 for @xmath318 $ ] . by definition , @xmath867 moreover , @xmath40 , @xmath41 , and @xmath236 are bounded by some constants only depending on @xmath320 and @xmath868 , so that , @xmath869 and hence @xmath870 for @xmath871 , we assume without loss of generality @xmath872 . recall that @xmath494 and @xmath873 and hence @xmath874 . for @xmath875 $ ]
, we have @xmath876 and @xmath863 .
thus @xmath877 and after applying gronwall s lemma we obtain @xmath878 for @xmath875 $ ] . in particular , @xmath879 where we used lemma [ lem : tauttau ] in the last step , and we get @xmath880 when @xmath875 $ ] . for @xmath881 , we have @xmath862 and @xmath863 , so that we obtain after applying gronwall s lemma once more @xmath882 finally , combining and we end up with @xmath883.\end{aligned}\ ] ] integrating over the bounded domain @xmath884 and applying minkowski s inequality for integrals , then yields @xmath885 adding up and , we have @xmath886 finally , it is left to estimate @xmath887 .
we have to distinguish several cases and therefore we introduce the set @xmath888 ( _ i _ ) if @xmath889 , we can choose @xmath890 depending on @xmath196 and @xmath247 as in lemma [ lem : est : diffg ] . if @xmath26 is such that @xmath718 , then @xmath891 . on the other hand , if @xmath26 is such that @xmath892 , we have @xmath893 since @xmath894 and @xmath895 can be bounded by a constant only depending on @xmath320 .
thus we get that , since @xmath719 only depends on @xmath320 , @xmath896 note that since @xmath897 by assumption , @xmath898 is square integrable on @xmath899 . hence @xmath900 ( _ ii _ ) if @xmath901 , either @xmath902 or @xmath403 and @xmath903 ( or the symmetric case where either @xmath904 or @xmath905 and @xmath906 ) . if @xmath907 , then @xmath908 and @xmath909 , and we have @xmath910 if @xmath403 , we know that @xmath911 and hence @xmath912 since the set @xmath158 has finite measure , we get @xmath913 combining , , , , and yields @xmath914 which implies , if we choose @xmath196 small enough , that @xmath915 this finishes the proof .
[ thm : shortstab2 ] given @xmath432 and @xmath467 , then the solutions @xmath587 and @xmath714 with initial data @xmath52 and @xmath715 , respectively , in @xmath64 , belong to @xmath797 for @xmath325 $ ] . for any given @xmath916 $ ]
, there exist @xmath829 and @xmath917 depending on @xmath320 and independent of @xmath918 , such that @xmath919\cap[0,t ] } d_{\mathbb r}(x(t),\tilde x(t))\leq kd_{\mathbb r}(x(\tilde t),\tilde x(\tilde t)).\ ] ] a close inspection of the proof of theorem [ thm : shortstab ] shows that all our estimates use upper bounds , and hence we can replace the constants therein depending on @xmath320 by the @xmath320 in the statement of the present theorem . since @xmath587 belongs to @xmath797 for all @xmath318 $ ] , we can apply lemma [ lem : g2 ] ( _ iv _ ) which tells us that for any initial time @xmath918 we can find for any @xmath329 a time interval such that all points enjoying wave breaking are contained in @xmath302 and this time interval is independent of the initial time @xmath918 .
hence , after this observation we can follow the proof of theorem [ thm : shortstab ] . for any time
@xmath467 there exists a constant @xmath829 only depending on @xmath247 and @xmath196 such that @xmath920}d_{\mathbb r}(x(t),\tilde x(t))\leq kd_{\mathbb r}(x_0,\tilde x_0)\ ] ] for any solutions @xmath587 and @xmath714 in @xmath797 , @xmath318 $ ] , with initial data @xmath52 and @xmath715 , respectively , in @xmath64 .
there exists @xmath320 only depending on @xmath247 and @xmath196 such that @xmath587 and @xmath714 belong to @xmath797 for all @xmath318 $ ] , cf .
theorem [ th : global ] .
from the short time stability result theorem [ thm : shortstab2 ] we know that there exist constants @xmath829 and @xmath433 depending only on @xmath320 such that @xmath921 for any @xmath318\cap [ \tilde t , \tilde t+\bar t]$ ] .
to obtain global stability we therefore split up the interval @xmath184 $ ] into smaller time intervals where the last inequality is valid . for any @xmath467 there exists @xmath922 such that @xmath923 and accordingly we define @xmath924 , @xmath925,@xmath926 , @xmath927 , and @xmath928 . hence @xmath929 for all @xmath930 $ ] , due to the last lemma .
hence we finally obtain @xmath931 which proves the claim .
in addition to the last stability result , one can also show the lipschitz continuity of every solution with respect to time .
[ lem : lip ] given @xmath432 and @xmath467 , then we have for any solution @xmath587 with initial data @xmath932 , @xmath63 for a constant @xmath933 which only depends on @xmath247 and @xmath196 .
we already know that for any @xmath467 , we have @xmath934 for @xmath318 $ ] , can be bounded by some constant @xmath935 depending on @xmath247 and @xmath196
. then we have @xmath936 hence @xmath937 it remains to estimate @xmath938 .
if for a given @xmath198 , @xmath631 and @xmath939 , we have @xmath940 where , slightly abusing the notation , @xmath941 denotes the time derivative of @xmath942 .
similarly if @xmath943 and @xmath944 , we get @xmath945 where @xmath946 denotes the time derivative of @xmath947 . if @xmath631 and @xmath944 ( the case @xmath943 , @xmath939 can be treated in much the same way ) , we can find a @xmath948 such that @xmath949 and therefore @xmath950 since @xmath951 and @xmath952 , can be uniformly bounded by a constant @xmath935 for all @xmath953 $ ] , we get after using - together with applying minkowski s inequality for integrals , that @xmath954
so far the derivation of our system of ordinary differential equations in lagrangian coordinates is only valid for initial data @xmath72 in eulerian coordinates , where no concentration of mass takes place .
however , it is well known that in the case of conservative solutions , concentration of mass is linked to wave breaking . since our description of dissipative solutions in lagrangian variables until wave breaking occurs , coincides with the one used in @xcite for conservative solutions , one might hope that the sets of eulerian and lagrangian coordinates can be described in much the same way , and that the mappings from eulerian to lagrangian coordinates and vice versa can be defined using the same ideas .
it turns out that we can do so . for the sake of completeness
we summarize these results here .
we start by introducing the set of eulerian and lagrangian coordinates together with the set of relabeling functions which allows us to identify equivalence classes in the lagrangian variables .
[ eulerian coordinates][def : d ] the set @xmath955 is composed of all triplets @xmath956 such that @xmath957 , @xmath958 and @xmath13 is a positive finite radon measure whose absolutely continuous part , @xmath959 , satisfies @xmath960 [ def : g1 ] we denote by @xmath961 the subgroup of the group of homeomorphisms from @xmath859 to @xmath859 such that [ eq : gcond ] @xmath962 where @xmath963 denotes the identity function .
given @xmath964 , we denote by @xmath965 the subset of @xmath961 defined by @xmath966 the subsets @xmath967 and @xmath968 of @xmath50 are defined as @xmath969 and @xmath970 where @xmath971 is defined by @xmath972 note that @xmath971 is finite , since from , we have @xmath973 and therefore @xmath974 .
in addition it should be pointed out that the condition on @xmath975 is closely linked to @xmath976 as the following lemma shows .
[ lem : charh ] let @xmath977 .
if @xmath67 belongs to @xmath965 , then @xmath978 almost everywhere .
conversely , if @xmath67 is absolutely continuous , @xmath979 , @xmath67 satisfies and there exists @xmath980 such that @xmath981 almost everywhere , then @xmath982 for some @xmath115 depending only on @xmath983 and @xmath984 .
an immediate consequence of is therefore the following result .
[ lem : fpres ] the space @xmath985 is preserved by the governing equations . for the sake of simplicity , for any @xmath986 and any function @xmath987
, we denote @xmath988 by @xmath989 .
the map from @xmath990 to @xmath967 given by @xmath991 defines an action of the group @xmath992 on @xmath967 .
since @xmath992 is acting on @xmath967 , we can consider the quotient space @xmath993 of @xmath967 with respect to the action of the group @xmath961 .
the equivalence relation on @xmath967 is defined as follows : for any @xmath994 , we say that @xmath44 and @xmath995 are equivalent if there exists a relabeling function @xmath987 such that @xmath996 .
we denote by @xmath997 $ ] the projection of @xmath967 into the quotient space @xmath993 , and introduce the mapping @xmath998 given by @xmath999 for any @xmath1000 .
we have @xmath1001 when @xmath1002 .
it is not hard to prove that @xmath1003 is invariant under the action of @xmath992 , that is , @xmath1004 for any @xmath1005 and @xmath987 .
hence , there corresponds to @xmath1003 a mapping @xmath1006 from the quotient space @xmath993 to @xmath1007 given by @xmath1008)=\gamma(x)$ ] where @xmath1009\in{{{\ensuremath{\mathcal{f}}}/{g}}}$ ] denotes the equivalence class of @xmath1005 .
for any @xmath1002 , we have @xmath1010 .
hence , @xmath1011 .
any topology defined on @xmath1007 is naturally transported into @xmath993 by this isomorphism .
we equip @xmath1007 with the metric induced by the @xmath1012-norm , i.e. , @xmath1013 for all @xmath1014 .
since @xmath1007 is closed in @xmath1012 , this metric is complete .
we define the metric on @xmath993 as @xmath1015,[x'])=d_{\mathbb r}(\gamma(x),\gamma(x')),\ ] ] for any @xmath1009,[x']\in{{{\ensuremath{\mathcal{f}}}/{g}}}$ ] .
then , @xmath993 is isometrically isomorphic with @xmath1007 and the metric @xmath1016 is complete . as in @xcite , we can prove the following lemma .
[ lem : picont ] given @xmath1017 .
the restriction of @xmath1003 to @xmath1018 is a continuous mapping from @xmath1018 to @xmath1007 .
the mapping @xmath1003 is not continuous from @xmath967 to @xmath1007 .
the spaces @xmath1018 were precisely introduced in order to make the mapping @xmath1003 continuous .
we denote by @xmath1019 the continuous semigroup which to any initial data @xmath1020 associates the solution @xmath587 of the system of differential equations at time @xmath48 . as indicated earlier , the two - component camassa
holm system is invariant with respect to relabeling .
more precisely , using our terminology , we have the following result .
[ th : sgs ] for any @xmath1021 , the mapping @xmath1022 is @xmath992-equivariant , that is , @xmath1023 for any @xmath1005 and @xmath987 .
hence , the mapping @xmath1024 from @xmath993 to @xmath993 given by @xmath1025)=[s_tx]\ ] ] is well - defined .
it generates a continuous semigroup .
we have the following diagram : @xmath1026^{\pi}&{{{\ensuremath{\mathcal{f}}}/{g}}}\\ { \ensuremath{\mathcal{f}}}_\alpha\ar[u]^{\gamma}&\\ { \ensuremath{\mathcal{f}}}_0\ar[u]^{s_t}\ar[r]^\pi&{{{\ensuremath{\mathcal{f}}}/{g}}}\ar[uu]_{\tilde s_t } } \ ] ] next we describe the correspondence between eulerian coordinates ( functions in @xmath955 ) and lagrangian coordinates ( functions in @xmath993 ) . in order to do so
, we have to take into account the fact that the set @xmath955 allows the energy density to have a singular part and a positive amount of energy can concentrate on a set of lebesgue measure zero .
we first define the mapping @xmath1027 from @xmath955 to @xmath1007 which to any initial data in @xmath955 associates an initial data for the equivalent system in @xmath1007 .
[ th : ldef ] for any @xmath956 in @xmath955 , let [ eq : ldef ] @xmath1028 then @xmath1029 .
we denote by @xmath1030 the mapping which to any element @xmath1031 associates @xmath1032 given by . on the other hand , to any element in @xmath967 there corresponds a unique element in @xmath955 which is given by the mapping @xmath247 defined below .
[ th : umudef ] given any element @xmath1000 .
then , the measure @xmath1033 is absolutely continuous , and we define @xmath956 as follows , for any @xmath26 such that @xmath68 , [ eq : umudef ] @xmath1034 we have that @xmath956 belongs to @xmath955 .
we denote by @xmath1035 the mapping which to any @xmath44 in @xmath967 associates the element @xmath1036 as given by .
in particular , the mapping @xmath247 is invariant under relabeling .
finally , one has to declare the connection between the equivalence classes in lagrangian coordinates and the set of eulerian coordinates .
[ th : lminv]the mappings @xmath247 and @xmath1027 are invertible .
we have @xmath1037
in the last section we defined the connection between eulerian and lagrangian coordinates , which is the main tool when defining weak solutions of the 2ch system . also stability results will heavily depend on this relation since we want to measure distances between solutions of the 2ch system by measuring the distance in lagrangian coordinates rather than in eulerian coordinates .
accordingly , we define @xmath1038 as @xmath1039 the metric @xmath1040 is defined as @xmath1041 [ eq : defweakconssol ] assume that @xmath1042 and @xmath1043 satisfy + ( i ) @xmath1044 and @xmath1045 , + ( ii ) the equations @xmath1046dxdt\\
= \int_{\mathbb r}u(0,x)\phi(0,x)dx , \end{gathered}\ ] ] @xmath1047dxdt=0,\ ] ] and @xmath1048dxdt=\int_{\mathbb r}\rho(0,x)\phi(0,x)dx,\ ] ] hold for all @xmath1049 .
then we say that @xmath16 is a weak global solution of the two - component camassa
holm system .
[ th : mainx ] the mapping @xmath1038 is a continuous semigroup of solutions with respect to the metric @xmath1040 . given any initial data @xmath1050 ,
let @xmath1051 .
then @xmath16 is a weak solution to and @xmath956 is a weak solution to @xmath1052 the function @xmath1053 which is an increasing function , equals the amount of energy that has concentrated at sets of measure zero up to time @xmath48 .
moreover , for every @xmath1054 , we clearly have @xmath1055 \(i ) equation is also valid in the conservative case , cf .
however , note that there is a difference in the definition of the quantity @xmath1056 in the two cases .
+ ( ii ) an example that illustrates this theorem is given by the symmetric peakon
antipeakon collision in case of the ch equation .
consider the case of @xmath1057 in and let @xmath1058 , and @xmath1059 .
then the solution @xmath11 will vanish pointwise at a collision time @xmath1060 when @xmath1061 , that is , @xmath1062 for all @xmath1063 . at time @xmath1064 , the total energy , @xmath1065 , has concentrated at the origin . using our mapping from eulerian to lagrangian coordinates ,
we obtain after collision , that is , for @xmath1066 , that @xmath1067 for @xmath1068 $ ] and @xmath1069 for @xmath1070 $ ] .
going back from lagrangian to eulerian coordinates we have @xmath1071 , while the whole energy still is concentrated at the origin .
recall that @xmath1072 is defined by and since @xmath1073 , we get @xmath1074 thus setting @xmath1075 and performing the change of variables @xmath1076 yields that @xmath1077 coincides with .
similar considerations yield that @xmath1078 coincides with
. we will only prove and here since and can be shown in much the same way . using the change of variables @xmath1075 and @xmath1079 we get @xmath1080 note that we do not have to restrict the domain in lagrangian coordinates to @xmath1081 , since the integrand vanishes whenever @xmath1069 , and therefore integrating in addition over the set @xmath1082 has no influence on the value of the integral .
we now turn to the proof of , which is equivalent to showing @xmath1083 in the sense of distributions . using the change of variables @xmath1075 and @xmath1084 we get @xmath1085 u^2(t,\xi)y_\xi(t,\xi)d\xi dt\\ { \nonumber } & \qquad = -\iint_{{\mathbb r}_+\times{\mathbb r } } ( u^2(t,\xi)y_{t,\xi}(t,\xi)+2u(t,\xi)u_t(t,\xi)y_\xi(t,\xi))\phi(t , y(t,\xi))d\xi dt\\ { \nonumber } & \qquad \quad-\iint_{{\mathbb r}_+\times { \mathbb r } } u^3(t,\xi)\phi_\xi(t , y(t,\xi))d\xi dt\\ { \nonumber } & \qquad = -\iint_{{\mathbb r}_+\times{\mathbb r } } ( u^2(t,\xi)u_\xi(t,\xi)-2u(t,\xi)q(t,\xi)y_\xi(t,\xi))\phi(t , y(t,\xi))d\xi dt\\ { \nonumber } & \qquad \quad-\iint_{{\mathbb r}_+\times{\mathbb r } } u^3(t,\xi)\phi_\xi(t , y(t,\xi))d\xi dt \\ { \nonumber } & \qquad = -\iint_{{\mathbb r}_+\times{\mathbb r}}u^2(t , x)u_x(t , x)\phi(t , x)dxdt -\iint_{{\mathbb r}_+\times{\mathbb r}}u^3(t , x)\phi_x(t , x)dx dt\\ { \nonumber } & \qquad \quad+\iint_{{\mathbb r}_+\times{\mathbb r } } 2u(t,\xi)p_\xi(t,\xi)\phi(t , y(t,\xi))d\xi dt.\end{aligned}\ ] ] next @xmath1086h(t,\xi)d\xi dt\\ { \nonumber } & \qquad = -\iint_{{\mathbb r}_+\times{\mathbb r}}h_t(t,\xi)\phi(t , y(t,\xi))d\xi dt-\iint_{{\mathbb r}_+\times{\mathbb r}}u(t,\xi)h(t,\xi)\phi_x(t , y(t,\xi))d\xi dt\\ { \nonumber } & \qquad = -\iint_{{\mathbb r}_+\times{\mathbb r } } 2(u^2(t,\xi)+\frac12 k^2-p(t,\xi))u_\xi(t,\xi)\phi(t , y(t,\xi))d\xi dt\\ { \nonumber } & \qquad \quad-\iint_{{\mathbb r}_+\times{\mathbb r}}u(t,\xi)h(t,\xi)\phi_x(t , y(t,\xi))d\xi dt\\ { \nonumber } & \qquad = -\iint_{{\mathbb r}_+\times{\mathbb r}}2(u^2+\frac12 k^2-p)(t , x)u_x(t , x)\phi(t , x)dxdt\\ { \nonumber } & \qquad \quad -\iint_{{\mathbb r}_+\times{\mathbb r } } u(t , x)\phi_x(t , x)d\mu(t , x).\end{aligned}\ ] ] since @xmath243 we get @xmath1087 finally @xmath1088d\xi dt\\ { \nonumber } & \qquad = -\iint_{{\mathbb r}_+\times{\mathbb r } } 2k\bar r_t(t,\xi)\phi(t , y(t,\xi))d\xi dt-\iint_{{\mathbb r}_+\times{\mathbb r } } 2k\bar r(t,\xi)u(t,\xi)\phi_x(t , y(t,\xi))d\xi dt\\ { \nonumber } & \qquad = \iint_{{\mathbb r}_+\times{\mathbb r } } 2k^2u_\xi(t,\xi)\phi(t , y(t,\xi))d\xi dt -\iint_{{\mathbb r}_+\times{\mathbb r}}2k\bar\rho(t , x)u(t , x)\phi_x(t , x)dx dt\\ { \nonumber } & \qquad = -\iint_{{\mathbb r}_+\times{\mathbb r } } 2k^2u(t,\xi)\phi_\xi(t , y(t,\xi))d\xi dt -\iint_{{\mathbb r}_+\times{\mathbb r}}2k\bar\rho(t , x)u(t , x)\phi_x(t , x)dx dt\\ { \nonumber } & \qquad = -\iint_{{\mathbb r}_+\times{\mathbb r}}2k^2u(t , x)\phi_x(t , x)dx dt -\iint_{{\mathbb r}_+\times{\mathbb r}}2k\bar\rho(t , x)u(t , x)\phi_x(t , x)dx dt . \end{aligned}\ ] ] adding up the equalities proves . as far as is concerned , we observe that @xmath1089
already in @xcite we showed that @xmath6 has a regularizing effect and that wave breaking is closely linked to the initial value of @xmath6 .
in particular , we showed that discontinuities travel at finite speed although the 2ch system has an infinite speed of propagation , see @xcite .
moreover , we obtained that wave breaking can only occur at points @xmath1090 which satisfy @xmath1091 . for completeness and to motivate assumptions that were made throughout this paper , we recall this result here .
* theorem 6.1 ) [ th : presreg ] we consider the initial data @xmath1096 .
assume that @xmath1096 is p - regular on a given interval @xmath1097 and @xmath1098 for @xmath1099 .
then , for any @xmath611 , @xmath1100 is p - regular on the interval @xmath1101 , where @xmath1102 and @xmath1103 satisfy @xmath1104 and @xmath1105 and are defined as @xmath1106 in the case of conservative solutions we have been able to prove in ( * ? ? ?
* theorem 6.3 ) that any conservative solution @xmath12 of the ch equation with initial data @xmath1107 such that @xmath1108 can be approximated by smooth conservative solutions @xmath1109 of the 2ch system with @xmath1110 for all @xmath80 .
observe that the approximate solutions of the 2ch system do not experience wave breaking .
in the context of dissipative solutions we can not hope that we can approximate dissipative solutions of the ch equation by solutions of the 2ch system which do not enjoy wave breaking according to the definition of our metric in lagrangian coordinates .
this is illustrated in figure [ fig : compsol ] .
note that , in @xcite , we show the regularizing effect of strictly positive @xmath6 .
this effect is also present in the example shown in the figure because , for strictly positive @xmath70 , dissipative and conservative solutions coincide and the second solution ( dashed line ) is @xmath1111 .
however , @xmath70 is small and we also have numerical errors , so that the solution appears as if it contains peaks . in that way , it looks very much like the conservative solution of the scalar ch equation with initial data @xmath72 .
the numerical scheme used to compute these solutions is an adaptation of the scheme studied in @xcite and the code is available at @xcite . and
@xmath1112 for @xmath1113 ( solid line ) and @xmath1114 ( dashed line ) .
the figures display the function @xmath11 only .
when @xmath1113 , the function @xmath11 solves the dissipative ch equation as @xmath6 will remain identically zero . in the case with @xmath60 , the fact that @xmath70 is strictly positive implies that no dissipation of energy takes place . in contrast , in the case @xmath1113 , dissipation starts occurring between the times @xmath1115 and @xmath1116 . prior to that , @xmath70 is so small that we can not see the difference between the two solutions .
however , the solutions look very different for @xmath1117 , after the first solution has experienced dissipation .
this example show that the semigroup of dissipative solution is not continuous with respect to @xmath6 in a standard sense .
this fact is reflected in the construction of our metric , which completely separates initial data for which @xmath70 vanishes and initial data for which @xmath70 is strictly positive in the same region.,title="fig:",width=226 ] and @xmath1112 for @xmath1113 ( solid line ) and @xmath1114 ( dashed line ) .
the figures display the function @xmath11 only .
when @xmath1113 , the function @xmath11 solves the dissipative ch equation as @xmath6 will remain identically zero . in the case with @xmath60 , the fact that @xmath70 is strictly positive implies that no dissipation of energy takes place . in contrast , in the case @xmath1113 , dissipation starts occurring between the times @xmath1115 and @xmath1116 . prior to that , @xmath70 is so small that we can not see the difference between the two solutions .
however , the solutions look very different for @xmath1117 , after the first solution has experienced dissipation .
this example show that the semigroup of dissipative solution is not continuous with respect to @xmath6 in a standard sense .
this fact is reflected in the construction of our metric , which completely separates initial data for which @xmath70 vanishes and initial data for which @xmath70 is strictly positive in the same region.,title="fig:",width=226 ] + and @xmath1112 for @xmath1113 ( solid line ) and @xmath1114 ( dashed line ) .
the figures display the function @xmath11 only .
when @xmath1113 , the function @xmath11 solves the dissipative ch equation as @xmath6 will remain identically zero . in the case with @xmath60 , the fact that @xmath70 is strictly positive implies that no dissipation of energy takes place . in contrast , in the case @xmath1113 , dissipation starts occurring between the times @xmath1115 and @xmath1116 .
prior to that , @xmath70 is so small that we can not see the difference between the two solutions .
however , the solutions look very different for @xmath1117 , after the first solution has experienced dissipation .
this example show that the semigroup of dissipative solution is not continuous with respect to @xmath6 in a standard sense .
this fact is reflected in the construction of our metric , which completely separates initial data for which @xmath70 vanishes and initial data for which @xmath70 is strictly positive in the same region.,title="fig:",width=226 ] and @xmath1112 for @xmath1113 ( solid line ) and @xmath1114 ( dashed line ) . the figures display the function @xmath11 only .
when @xmath1113 , the function @xmath11 solves the dissipative ch equation as @xmath6 will remain identically zero . in the case with @xmath60 , the fact that @xmath70 is strictly positive implies that no dissipation of energy takes place . in contrast , in the case @xmath1113 , dissipation starts occurring between the times @xmath1115 and @xmath1116 .
prior to that , @xmath70 is so small that we can not see the difference between the two solutions .
however , the solutions look very different for @xmath1117 , after the first solution has experienced dissipation .
this example show that the semigroup of dissipative solution is not continuous with respect to @xmath6 in a standard sense .
this fact is reflected in the construction of our metric , which completely separates initial data for which @xmath70 vanishes and initial data for which @xmath70 is strictly positive in the same region.,title="fig:",width=226 ] + and @xmath1112 for @xmath1113 ( solid line ) and @xmath1114 ( dashed line ) .
the figures display the function @xmath11 only .
when @xmath1113 , the function @xmath11 solves the dissipative ch equation as @xmath6 will remain identically zero . in the case with @xmath60 , the fact that @xmath70 is strictly positive implies that no dissipation of energy takes place .
in contrast , in the case @xmath1113 , dissipation starts occurring between the times @xmath1115 and @xmath1116 .
prior to that , @xmath70 is so small that we can not see the difference between the two solutions .
however , the solutions look very different for @xmath1117 , after the first solution has experienced dissipation .
this example show that the semigroup of dissipative solution is not continuous with respect to @xmath6 in a standard sense .
this fact is reflected in the construction of our metric , which completely separates initial data for which @xmath70 vanishes and initial data for which @xmath70 is strictly positive in the same region.,title="fig:",width=226 ] and @xmath1112 for @xmath1113 ( solid line ) and @xmath1114 ( dashed line ) .
the figures display the function @xmath11 only .
when @xmath1113 , the function @xmath11 solves the dissipative ch equation as @xmath6 will remain identically zero . in the case with @xmath60 , the fact that @xmath70 is strictly positive implies that no dissipation of energy takes place .
in contrast , in the case @xmath1113 , dissipation starts occurring between the times @xmath1115 and @xmath1116 .
prior to that , @xmath70 is so small that we can not see the difference between the two solutions .
however , the solutions look very different for @xmath1117 , after the first solution has experienced dissipation .
this example show that the semigroup of dissipative solution is not continuous with respect to @xmath6 in a standard sense .
this fact is reflected in the construction of our metric , which completely separates initial data for which @xmath70 vanishes and initial data for which @xmath70 is strictly positive in the same region.,title="fig:",width=226 ] ( _ i _ ) given two elements @xmath1122 and @xmath1123 in @xmath1124 such that @xmath1125 then @xmath1126 where @xmath643 denotes a constant dependent on the @xmath231-norm of @xmath1127 , @xmath1128 , @xmath1129 , and @xmath1130 .
we will only prove ( _ i _ ) since ( _ ii _ ) is an immediate consequence of ( _ i _ ) .
we will have to consider different cases and therefore we introduce the following sets : @xmath1136 note that @xmath1137 .
in addition we will denote the sets which correspond to @xmath1138 and @xmath1123 by superscripts @xmath1139 and @xmath1140 , respectively .
( _ iv _ ) @xmath1157 : without loss of generality , we assume @xmath1158 ( the other case follows similarly ) . by assumption
we have @xmath1159 and @xmath1160 , which implies that either @xmath1161 where we used in the last step that @xmath1162 since @xmath1163 . thus applying yields
@xmath1164 ( _ v _ ) @xmath1165 : without loss of generality , we assume @xmath1166 ( the other case follows similarly ) . by assumption
we have @xmath1159 and @xmath1167 , which implies that either @xmath1168 where we used in the last step that @xmath1169 .
thus we get @xmath1170 finally adding , we end up with .
first of all note that according to lemma [ lem : urhog ] the function @xmath1173 converges to @xmath1174 in @xmath1135 . since the set of eulerian and lagrangian coordinates and the mappings between them coincide with the ones used in @xcite , one can prove everything as in ( * ? ? ?
* lemma 6.4 ) , except @xmath1175 .
thus we will only show that @xmath1176 in @xmath231 .
let @xmath1177 and @xmath1178 be the representatives in @xmath1007 given by of @xmath1179 and @xmath1180 , respectively and assume that @xmath1181 in @xmath1182 .
abbreviate by @xmath1183 and @xmath1184 and note that @xmath1185 in @xmath1135 . since the measures @xmath1186 and @xmath1187 are purely absolutely continuous , we obtain that @xmath1188 almost everywhere and in particular @xmath1189 this implies in particular that @xmath1190 almost everywhere and @xmath1191 almost everywhere so that @xmath1192 we will study the first term on the right - hand side in detail and explain afterwards how the other terms can be treated similarly .
we have @xmath1193 using now the change of variables @xmath1194 , since @xmath1195 , we get @xmath1196 since @xmath1197 , we can find to any @xmath1198 a continuous function @xmath1199 with compact support such that @xmath1200 .
hence we can decompose the second term on the right - hand side of into @xmath1201 by arguing as in , one can show that @xmath1202 moreover , since @xmath1203 in @xmath1204 and @xmath1199 is continuous with compact support , we obtain by applying the lebesgue dominated convergence theorem , that @xmath1205 in @xmath1135 and thus we can choose @xmath189 big enough so that @xmath1206 hence , we get that @xmath1207 so that @xmath1208 as far as the second term on the right - hand side of is concerned , we observe that @xmath1209 and @xmath1210 , so that we can follow the same procedures as for the first term .
thus we finally obtain @xmath1176 in @xmath1135 and since @xmath1211 and @xmath1212 this implies @xmath1176 in @xmath231 .
[ lemma : kont1 ] let @xmath1213 be a sequence in @xmath955 that converges to @xmath956 in @xmath955 .
then @xmath1214 where @xmath1215 and @xmath1216 are defined by , for @xmath1109 and @xmath956 , respectively .
since the set of eulerian and lagrangian coordinates and the mappings between them coincide with the ones used in @xcite , one can prove everything as in ( * ? ? ?
* lemma 6.4 ) except for @xmath1217 , which we will prove now .
we denote by @xmath1177 and @xmath1178 the representative of @xmath1218 and @xmath1219 given by . by weak - star convergence
we mean that @xmath1220 for all continuous functions with compact support .
it follows from , that @xmath655 implies that @xmath1221 and therefore @xmath1222 similarly one obtains that @xmath1223 . since @xmath1203 in @xmath1204 and @xmath1224 ,
the support of @xmath1225 is contained in some compact set which can be chosen independently of @xmath189 and , from lebesgue s dominated convergence theorem , we have @xmath1226 in @xmath231 .
hence , since @xmath1176 in @xmath231 , @xmath1227 combining all the equalities we obtained so far yields . | we show existence of a global weak dissipative solution of the cauchy problem for the two - component camassa holm ( 2ch ) system on the line with nonvanishing and distinct spatial asymptotics .
the influence from the second component in the 2ch system on the regularity of the solution , and , in particular , the consequences for wave breaking , is discussed .
furthermore , the interplay between dissipative and conservative solutions is treated . |
hgte / cdhgte quantum wells ( qws ) of thickness above a critical value belong to the class of @xmath1 two - dimensional topological insulators characterized by the existence of counter propagating helical edge modes @xcite .
continuous advance in the technology of cdhgte - based heterostructures stimulates experimental and theoretical studies of their electronic properties related to the non - trivial band topology .
the structure of the edge states responsible for the emergence of the quantum spin hall effect was theoretically studied at zero magnetic field @xcite , in the presence of magnetic field lying in the qw plane @xcite , and in magnetic field normal to the qw plane @xcite .
most of the theoretical papers consider simplified models which do not take into account the natural inversion asymmetry of the hgte / cdhgte qws caused by their atomic structure or treat this asymmetry as a small perturbation @xcite .
contrary , atomistic calculations performed recently @xcite have revealed very strong level repulsion in hgte / cdhgte qws , mostly driven by the natural interface inversion asymmetry of the zinc - blende heterostructures , which results in a considerable modification of the `` bulk '' ( two - dimensional ) electron states and dispersion .
the inversion asymmetry also gives rise to a giant photogalvanic response observed in hgte / cdhgte heterostructures @xcite .
in the present paper we study theoretically the electron structure of bulk and helical edge states in hgte / hgcdte qws with strong natural interface inversion asymmetry in external magnetic field .
we find that the interface mixing of the states results in ( i ) a strong anisotropy of the edge - state @xmath0-factor in the in - plane magnetic field and ( ii ) opening of the gap in the edge - state spectrum by an arbitrary small out - of - plane magnetic field .
both effects are absent in centrosymmetric continuum - medium models .
we obtain analytical results for the energy spectrum and wave functions of the edge states in a semi - infinite two - dimensional structure and do numerical calculations of the spectrum of coupled edge states in a strip of a finite width .
the paper is organized as follows . in sec .
[ sec : bulk ] we present the effective hamiltonian of the system and describe the bulk energy spectrum at zero magnetic field and the structure of bulk landau levels . in sec .
[ sec : semi_inf ] we study analytically the helical states in a semi - infinite system with a single edge at zero magnetic field ( sec . [ sec : zerob ] ) , in the in - plane magnetic field ( sec .
[ sec : b_inplane ] ) , and out - of - plane magnetic field ( sec .
[ sec : b_outplane ] ) .
section [ sec : orient ] presents the study of the edge states and the zeeman effect in a semi - infinite structure with an arbitrary orientation of the edge with respect to crystallographic axes . in sec .
[ sec : finite ] we outline the numeric procedure used to calculate the edge states in a strip of a finite width and compare the obtained numerical and analytical results .
[ sec : concl ] summarizes the paper .
we consider hgte / hgcdte qws grown along the @xmath2 $ ] axis ( @xmath3 point group ) with a symmetric heteropotential . in the qws of the critical thickness @xmath4 , where the transition between the trivial and non - trivial topological phases occurs , and in qws of close - to - critical thickness , the dirac states are formed from the electron - like @xmath5 and heavy - hole @xmath6 subbands , @xcite @xmath7 ,
@xmath8 , and @xmath9 are the envelope functions , @xmath10 , @xmath11 , and @xmath12 are the bloch amplitudes of the @xmath13 and @xmath14 bands at the @xmath15 point of the brillouin zone .
symmetry lowering resulting from the anisotropy of the qw interfaces leads to an efficient interface coupling of the light - hole states @xmath16 and heavy - hole states @xmath17 and , hence , to coupling of the electron - like and heavy - hole subbands .
this coupling leads to the level anticrossing at the interfaces and splitting of the dirac cones @xcite .
the effective 4@xmath184 @xmath19@xmath20@xmath21 hamiltonian , which precisely takes into account the real spatial symmetry of the qw structure , can be constructed in the framework of the group representations theory .
the effective hamiltonian can be derived taking into account that , in the @xmath3 point group , the @xmath5 and @xmath22 pairs transform according to the spinor representation @xmath14 while the components @xmath23 of the in - plane wave vector @xmath19 belong to the irreducible representation @xmath24 .
the effective hamiltonian to the second order in the wave vector in the @xmath25 basis has the form ( see also refs . )
@xmath26 here , @xmath27 , @xmath28 , @xmath29 $ ] and @xmath30 $ ] are the in - plane axes , @xmath31 , @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 , @xmath37 , and @xmath38 are the structure parameters .
the parameter @xmath38 determines the energy gap .
it can be tuned from positive to negative values by varying the qw thickness and defines whether the system is in the trivial ( @xmath39 at negative @xmath32 ) or non - trivial ( @xmath40 , @xmath41 ) topological phase .
the parameters @xmath34 and @xmath35 describe contributions to @xmath19-linear splitting of the electron - like and heavy - hole subbands caused by bulk inversion asymmetry .
the parameters @xmath36 and @xmath37 are determined by the interface mixing strength .
atomistic calculations yield the splitting @xmath42 mev at @xmath43 for hgte / hg@xmath44cd@xmath45te qws with atomically sharp interfaces @xcite .
such a strong interface coupling of the states drastically affects the energy spectrum and can not be treated as a small perturbation .
in contrast , the parameters @xmath34 , @xmath35 , and @xmath37 lead only to corrections to the splitting at @xmath46 .
therefore , to simplify calculations we consider the standard bernevig - hughes - zhang hamiltonian of the hgte - based two - dimensional topological insulators @xcite ( @xmath34 , @xmath35 , @xmath47 ) with account for the interface coupling described by @xmath36 . for numerical calculations presented below we use parameters : @xmath48 ev@xmath20 , @xmath49 ev@xmath20@xmath50 , @xmath51 ev@xmath20@xmath50 corresponding to a 7-nm - wide hgte / hg@xmath44cd@xmath45te qw ( the thickness is slightly above @xmath4 ) @xcite .
we assume @xmath36 to be positive and use @xmath52 mev .
diagonalization of the hamiltonian yields the energy spectrum of the bulk dirac fermions @xcite @xmath53 the dispersion curves given by eq .
are depicted in fig .
[ fig : zero_b ] . generally , the spectrum contains four branches with an energy gap @xmath54 , where @xmath55 , at the wave vector @xmath56 . for a gapless structure ( @xmath57 ) ,
the spectrum consists of two massless dirac cones shifted vertically with respect to each other by 2@xmath36 .
energy spectra of the bulk dirac fermions in hgte / cdhgte qws ( a ) with @xmath58 mev , and ( b ) of the critical thickness , @xmath57 , at zero magnetic field .
the level splitting at @xmath59 in the panel ( b ) is caused by the light - hole heavy - hole mixing .
, scaledwidth=48.0% ] external magnetic field @xmath60 is included in the @xmath19@xmath20@xmath21 theory by the peierls substitution @xmath61 in the zero - field hamiltonian and by adding the zeeman hamiltonian @xmath62 where @xmath63 , @xmath64 is the electron charge , @xmath65 is the vector potential of the magnetic field , @xmath66 , @xmath67 is the bohr magneton , @xmath68 , @xmath69 , @xmath70 , and @xmath71 are the contributions to the @xmath0-factors of the @xmath72 and @xmath73 subbands stemming from the bare electron @xmath0-factor and interaction with remote electron and hole subbands , and @xmath74 . the coupling of the @xmath72 and @xmath73 states by the out - of - plane magnetic field is exactly taken into account in the hamiltonian @xmath75 $ ] .
the coupling of these bands by the in - plane magnetic field , which occurs in qws of the @xmath3 symmetry , is small since it requires the consideration of inversion symmetry breaking , and thus , is neglected in the hamiltonian .
we also note that @xmath70 is expected to be small compared to @xmath68 @xcite . for numerical calculations we use @xmath76 , @xmath77 , @xmath78 , and @xmath79 @xcite .
the structure of the bulk landau levels in the perpendicular magnetic field @xmath80 can be readily found by solving the schrdinger equation @xmath81 with the hamiltonian @xmath82 + \mathcal h_{z } \:.\ ] ] we take the vector potential in the landau gauge @xmath83 and following refs .
solve the schrdinger equation by decomposing the four - component wave function @xmath84 in a series of the landau level functions @xmath85 @xmath86 where @xmath87 and @xmath88 are
the quantum numbers , and @xmath89 , @xmath90 , @xmath91 , and @xmath92 are coefficients .
bulk landau levels in a hgte / hgcdte qw with @xmath93 mev .
red curves show the magnetic field dependence of the zero landau levels .
dashed red lines in panel ( a ) show the positions of the zero landau levels in the absence of the light - hole
heavy - hole mixing ( @xmath94 )
. panels ( b ) and ( c ) show the structure of the landau levels at small magnetic fields .
, scaledwidth=45.0% ] bulk landau levels in a hgte / hgcdte qw of the critical thickness , @xmath57 mev .
red lines present the dispersion of `` zero '' landau levels .
, scaledwidth=45.0% ] figure [ fig : ll1 ] shows the calculated energy spectrum for a hgte / cdhgte qw in the topologically non - trivial phase .
landau levels can be divided into two groups .
the first group comprises two `` zero '' modes corresponding to @xmath95 .
these modes are formed by the @xmath96 and @xmath97 subbands only and are decoupled from other landau levels with @xmath98 .
the energies of the zero modes are given by @xmath99 ^ 2 + \gamma^2 } \:,\end{gathered}\ ] ] where @xmath100 is the magnetic length . in the absence of interface mixing ( @xmath94 ) , the zero landau levels cross at the critical field @xmath101 $ ] @xcite .
the mixing of these states due to interface inversion asymmetry ( or bulk inversion asymmetry @xcite ) leads to coupling of the zero modes and anticrossing with the gap @xmath102 at @xmath103 , see fig .
[ fig : ll1 ] . landau levels that belong to the second group are described by the wave functions @xmath104 with @xmath98 .
the corresponding energies are the roots of a fourth - degree polynomial . for @xmath105 and @xmath106 ,
the energies can be found analytically and have the form @xmath107 the structure of bulk landau levels in the qw of the critical thickness is shown in fig .
[ fig : ll2 ] . in this case , each of the dirac cones [ fig .
[ fig : zero_b](b ) ] forms a fan of landau levels .
we consider now a semi - infinite qw structure in the halfspace @xmath108 . in the topologically non - trivial phase @xmath109 ,
the structure supports edge states , which exponentially decay at @xmath110 .
we find the energy spectrum and wave functions of the edge states by solving the schrdinger equation @xmath111 with the open boundary conditions at the edge , @xmath112 .
a more general form of boundary conditions was studied in ref . .
taking into account the translation invariance along the @xmath113 direction we can present the wave function of the edge state in the form @xmath114 where @xmath88 is the wave vector along the edge , @xmath115 is the normalization length , @xmath116 are the coefficients to be determined from the boundary conditions , @xmath117 are the complex - valued reciprocal lengths , and @xmath118 are the position - independent normalized four - component columns .
we note that all components of @xmath118 are generally nonzero since the hamiltonian contains off - diagonal blocks @xmath119 .
this is in contrast to centrosymmetric models ( @xmath94 ) with the decoupled lower and upper blocks of the hamiltonian . for a given wave vector @xmath88 ,
the columns @xmath118 and the relation between the reciprocal lengths @xmath117 and the energy @xmath120 are found from the matrix equation @xmath121 first , we calculate the wave functions at @xmath122 .
we consider the case of `` electron - hole '' symmetry , i.e. , @xmath123 in the hamiltonian , so that the edge states at @xmath124 have the energy @xmath125 .
a more general case of @xmath126 will be analyzed in sec .
[ sec : finite ] in numeric simulations of the electron states in a strip of a finite width . for @xmath125 , the right - hand side of eq .
vanishes and non - trivial solutions for @xmath127 exist for a set of @xmath128 which satisfy the equation @xmath129 .
this equation yields eight reciprocal lengths @xmath117 which are pairwise related to each other by complex conjugation @xmath130 the corresponding normalized null - space vectors of the matrices @xmath131 have the form @xmath132 the wave functions of the edge states are given by eq . .
to satisfy the boundary condition @xmath133 and the wave function decay at @xmath110 ( implying @xmath134 ) one has to set @xmath135 , @xmath136 , and @xmath137 .
finally , the wave functions of the edge states are two - fold degenerate at @xmath124 and can be chosen in the form @xmath138 where @xmath139^{-1/2}\ ] ] is the normalization constant .
we note that in topologically trivial phase ( @xmath140 ) , the boundary conditions yield @xmath141 for all the coefficients and , hence , no edge states emerge .
the dependence of the wave functions on @xmath142 can be further simplified by introducing the effective lengths @xmath143 and @xmath144 and the wave vector @xmath56 .
for the parameters of hgte / cdhgte qws and @xmath145mev , one has @xmath146 , @xmath147 , and @xmath148 .
taking into account that @xmath149 , we obtain @xmath150 note that the wave vector @xmath151 corresponds to the position of the bulk energy gap in the two - dimensional brillouin zone , see sec . [
sec : bulk ] . to obtain the energy spectrum at @xmath152
we construct the effective 2@xmath182 hamiltonian by projecting the hamiltonian onto the basis functions @xmath153 and @xmath154 . to the first order in @xmath88 , the effective hamiltonian of the edge states reads @xmath155 the hamiltonian
is diagonalized by the unitary transformation @xmath156 where @xmath157 .
in the basis of the functions @xmath158 and @xmath159 , it has the form @xmath160 where @xmath161 is the pauli matrix and @xmath162 the @xmath163 parameter describes the group velocity of the edge states .
in contrast to models neglecting the light - hole heavy - hole mixing , the velocity @xmath163 at @xmath164 depends on the qw thickness . at @xmath165 ,
the velocity tends to zero , and the dispersion of the edge states vanishes .
the wave functions @xmath166 can be written in an equivalent form @xmath167\ : , \nonumber \\ \phi_2 & = & \frac { { \mathrm{e}}^{{\mathrm{i}}k_y y}}{\sqrt{l } } \left [ a(x ) \frac{\xi_1 + \xi_2}{\sqrt{2 } } - { \mathrm{i}}b(x ) \frac{\xi_1 - \xi_2}{\sqrt{2 } } \right]\:,\end{aligned}\ ] ] where @xmath168 and @xmath169 are the real functions @xmath170\ : , \nonumber \\
b(x ) = \sqrt{\frac{2}{l_2 } } \left [ { \mathrm{e}}^{-x / l_1 } \sin \frac{\varphi}{2 } + { \mathrm{e}}^{-x / l_2 } \sin \left ( k_0 x - \frac{\varphi}{2 } \right ) \right]\:.\end{aligned}\ ] ] note that @xmath169 is nonzero only at @xmath164 .
the amplitudes @xmath168 and @xmath169 are depicted in fig .
[ fig : fig1 ] . at @xmath171 , @xmath168 and @xmath169 ( i )
oscillate with the wave length @xmath172 and ( ii ) decay with the characteristic length @xmath173 .
the number of oscillations within the decay length is given by the dimensionless parameter @xmath174 which is large for @xmath165 .
functions @xmath168 and @xmath169 which determine the spatial structure of the edge - state wave functions @xmath166 , see eq . , for different values of @xmath175 . ]
we consider now the effect of in - plane magnetic field @xmath176 on the edge states in a semi - infinite structure with the edge perpendicular to the @xmath29 $ ] axis .
the in - plane field couples the electron as well as the hole spin states , see the zeeman hamiltonian . by projecting @xmath177 onto the edge
states @xmath178 , we obtain the effective zeeman hamiltonian @xmath179 with the components of the effective @xmath0-factor tensor of the edge states given by @xmath180 in the hamiltonian , we neglected the term @xmath181 since the off - diagonal component @xmath182 of the edge @xmath0-factor tensor is proportional to small parameters @xmath183 and @xmath184 .
equations , present one of the main results of our work : the light - hole heavy - hole mixing provided by the d@xmath185 point - group symmetry of the qw results in a strong anisotropy of the in - plane zeeman effect for the edge states .
the effective @xmath0-factor for the magnetic field pointing along the edge , @xmath186 $ ] , is reduced by the factor @xmath187 which is determined by the ratio between the energy of the light - hole
heavy - hole mixing and the bulk band gap .
magnetic field opens the gap in the energy spectrum of the edge states @xmath188 where @xmath189 is the angle between the in - plane vector @xmath60 and the edge normal .
due to the @xmath0-factor anisotropy given by eqs . , the gap depends on the magnetic field orientation . magnetic field perpendicular to the qw plane @xmath80 affects the edge states via the coupling of the electron - like and heavy - hole subbands included in the hamiltonian @xmath75 $ ] and the extra zeeman term .
although magnetic field breaks the time reversal symmetry and , consequently , may destroy topological protection of the edge states , it was shown that in centrosymmetric models magnetic field @xmath190 does not open the gap in the spectrum of the edge states preserving their helical structure @xcite .
the gap is opened in high enough fields only ( @xmath191 ) , when the system is in the quantum hall effect regime .
the interface mixing in real qws qualitatively changes the behavior of the edge states in out - of - plane magnetic field and leads to the emergence of the gap at arbitrary small fields . to analyze the edge - state spectrum in the out - of - plane magnetic field
we take the vector potential in the form @xmath192 , where @xmath193 is a constant .
different @xmath193 correspond to different gauges of the magnetic field . by projecting @xmath75 + \mathcal h_{z}$ ] onto the edge
states @xmath178 , we obtain the effective zeeman hamiltonian @xmath194 where the components of the @xmath0-factor tensor at @xmath122 , in the leading order in @xmath183 and @xmath184 , are given by @xmath195 and @xmath196 is the free electron mass .
both @xmath197 and @xmath198 components have large orbital contributions originating from the term @xmath75 $ ] . from the hamiltonians and
we deduce that the diagonal component @xmath197 leads only to a shift of the energy spectrum along @xmath88 without opening the gap .
the shift depends on the magnetic field gauge and , for a single edge , can be excluded by a proper choice of the coordinate frame .
in contrast , the term @xmath199 is gauge independent .
it couples edge states with the opposite pseudospin projections and opens the gap @xmath200 the gap in the edge - state spectrum emerges due to the lack of space inversion asymmetry in the qw and is a non - monotonic function of the bulk band gap @xmath54 .
in above subsections we studied a semi - infinite system with the edge parallel to one of the cubic axes , @xmath30 $ ] .
meanwhile , since the interface inversion asymmetry is associated with the certain crystallographic axes , it is natural to expect that the edge - state @xmath0-factors depend on the orientation of the edge .
now , we consider a semi - infinite structure with the edge of an arbitrary orientation defined by the angle @xmath201 between the edge and the @xmath202 $ ] axis , see the inset in fig .
[ fig : fig4 ] .
we introduce a new coordinate frame ( @xmath203 ) rotated with respect to the frame ( @xmath204 ) by the angle @xmath201 and the corresponding basis states latexmath:[\ ] ] in - plane magnetic field opens the gap in the edge - state spectrum @xmath220 the dependence of the gap on magnetic field orientation at @xmath79 is given by @xmath221 where @xmath222 is the angle between @xmath60 and the edge normal . the gap induced by normal magnetic field @xmath223
is given by eq . and is independent of the edge orientation .
now we present the numerical results for the energy spectrum of the edge and bulk states in the strip of a finite width @xmath224@xmath225 m and compare it with the analytical theory developed in sec .
[ sec : semi_inf ] .
we consider the strips with different crystallographic orientations and analyze the spectrum modification in both in - plane and out - of - plane magnetic fields . to calculate the spectrum we numerically solve the schrdinger equation with the hamiltonian @xmath226 + \mathcal h_z$ ] using the open boundary conditions @xmath227 .
the band structure parameters used in the calculations are listed in sec .
[ sec : bulk ] .
we consider hgte / cd@xmath45hg@xmath44te qws in the regime of topological insulator ( @xmath228 ) .
varying the absolute value of @xmath229 allows us to simulate the qws of different thickness .
small change of the band structure parameters upon variation of the qw thickness in the vicinity of @xmath4 ( see parameterizations for different well thickness in ref . ) is neglected in the calculations .
figure [ fig : fig2 ] shows the calculated zero - field energy spectrum of electron states in the strip with the bulk band gap @xmath230 mev .
one can see that the states with linear dispersion in the vicinity of @xmath124 emerge inside the band gap .
the dispersion of these states is shifted from the band gap center towards positive energies due to the `` electron - hole asymmetry '' , i.e. @xmath231 .
each of the dispersion curves depicts two almost degenerate states corresponding to the pair of spin - polarized states localized at the two strip edges .
the finite width of the strip leads to an inevitable overlap of the states localized at the spatially separated edges and , consequently , to the opening of a small gap at @xmath124 @xcite .
the gap , however , exponentially decreases with the growth of the strip width : for @xmath224@xmath225 m [ see fig . [
fig : fig2](a ) ] the gap is only about 5 @xmath225ev . at @xmath232 the dispersion of the bulk states has pronounced extrema originating from the light - hole heavy - hole mixing [ c.f .
[ fig : zero_b](a ) ] . according to eq .
, such a shape of the bulk energy spectrum leads to a flattening of the edge - state dispersion curves . for the dispersion of the edge states shown in fig .
[ fig : fig2 ] we find the effective velocity @xmath233cm / s whereas calculations with the same parameters but @xmath94 ( not shown ) yield @xmath234cm / s .
electron energy spectrum of the strip made of hgte / cdhgte qw structure in the topologically non - trivial phase ( @xmath235 mev ) .
the width of the strip is @xmath236 m .
, scaledwidth=48.0% ] as it was shown in secs .
[ sec : b_inplane ] and [ sec : orient ] , in - plane magnetic field opens a gap in the energy spectrum of the edge states .
due to the light - hole
heavy - hole mixing , the gap depends on the magnetic field direction and the edge orientation .
numeric calculation of the gap behaviour upon variation of the magnetic field direction is presented in fig .
[ fig : fig3 ] for the edges along [ 010 ] .
the calculated dependence is perfectly described by eq . with the effective @xmath0-factors
@xmath237 , @xmath238 .
however , the obtained @xmath0-factors are far smaller than those estimated from eq . , @xmath239 , @xmath240
, indicating that the `` electron - hole '' asymmetry ( @xmath231 ) neglected in the analytical theory considerably affects the zeeman splitting .
the value of the @xmath0-factor anisotropy @xmath241 is in a much better agreement with the analytical theory which yields @xmath242 .
energy gap in the edge - state spectrum as a function of the angle @xmath222 between the in - plane magnetic field @xmath60 and the edge normal calculated for @xmath235 mev and @xmath243 t. the insets illustrate the geometry under study and the energy spectrum at @xmath244.,scaledwidth=48.0% ] energy gap in the edge - state spectrum as a function of the angle @xmath222 between the in - plane magnetic field @xmath60 and the edge normal for different orientations of the strip edges .
the curves are calculated for @xmath93 mev and @xmath243 t. the inset illustrates the geometry under study.,scaledwidth=48.0% ] figure [ fig : fig4 ] presents the calculated edge - state energy gap as a function of the angle @xmath222 for different crystallographic orientations of the strip edges . in agreement with eq . , the angular dependence of the gap acquires a phase @xmath245 for the edge tilted by the angle @xmath201 with respect to the @xmath202 $ ] axis .
figure [ fig : fig5 ] presents electron energy spectra of the strip subjected to an out - of - plane magnetic field at four increasing fields @xmath246 t , @xmath247 t , @xmath248 t , and @xmath249 t. the spectra are calculated for the bulk gap @xmath250 mev . in accordance with the results of sec . [
sec : b_outplane ] , the out - of - plane magnetic field opens the gap in the edge - state spectrum .
the dependence of the edge - state gap @xmath251 on @xmath223 is shown in fig .
[ fig : fig6 ] . at small magnetic fields ,
this dependence is linear , and for a given magnetic field , @xmath251 is a non - monotonic function of the bulk gap @xmath54 ( see the inset in fig .
[ fig : fig6 ] ) , in agreement with eq . .
electron energy spectra of the strip subject to out - of - plane magnetic field @xmath223 .
the spectra are calculated for @xmath252 @xmath225 m and @xmath93 mev .
solid and dashed dispersion curves in panels ( a ) and ( b ) correspond to the states localized at the left and right edges , respectively .
dashed - dotted vertical lines indicate the positions @xmath253 for each value of magnetic field.,scaledwidth=48.0% ] the out - of - plane magnetic also results in diamagnetic shifts of the dispersion curves , corresponding to the states localized at the opposite edges .
these shifts along @xmath88 are clearly seen at small magnetic field , see figs .
[ fig : fig5](a ) and ( b ) .
we note that the relative shift of the left - edge and right - edge energy spectra is independent of the magnetic field gauge and , for a wide enough strip , is given by @xmath254 . in our calculations
we use the gauge with @xmath255 resulting in the symmetric diamagnetic shifts @xmath256 with respect to @xmath124 . at large magnetic fields , figs .
[ fig : fig5](c ) and ( d ) , the energy levels become flat for all @xmath88 ( except the narrow ranges around @xmath253 ) , which corresponds to the formation of bulk landau levels . in particular , two levels in the vicinity of @xmath125 are the bulk zero modes with the energies @xmath257 given by eq . .
the highly dispersive states at @xmath253 are localized at the strip edges and correspond to the electron and hole chiral edge modes .
thus , the out - of - plane magnetic field finally drives the system into the quantum hall effect phase .
energy gap in the edge - state spectrum of a strip as a function of the out - of - plane magnetic field .
dashed lines depict the linear behaviour in the region of small fields .
the inset shows the gap at @xmath258 t as a function of the bulk gap @xmath54 , dots present the results of numerical calculations , solid curve shows the analytical dependence at @xmath259 calculated after eq . .
, scaledwidth=48.0% ]
to conclude , we have presented the microscopic theory of the electron structure and the zeeman effect for helical edge states emerging in two - dimensional topological insulators based on hgte / hgcdte quantum wells .
the theory takes into account strong natural interface inversion asymmetry of the ( 001)-grown hgte / hgcdte quantum wells which reflects the real spatial symmetry described by the @xmath3 point group .
the interface inversion asymmetry leads to the mixing of the electron - like and heavy - hole subbands forming the helical edge states .
the subband mixing , described by a single parameter @xmath36 , modifies the edge - state dispersion and leads to spatial oscillations of the edge - state wave functions with the wave length controlled by @xmath36 .
external magnetic field applied to the quantum well structure destroys the topological protection of the helical states and opens the gap in the edge - state spectrum .
for the in - plane magnetic field , the subband mixing gives rise to a strong anisotropy of the edge - state effective @xmath0-factor which also becomes dependent on the crystallographic orientation of the edge .
the @xmath0-factor anisotropy results , in turn , in variation of the edge - state gap with magnetic field direction and edge orientation .
weak magnetic field normal to the quantum well plane couples the counter propagating edge states and opens the gap in the edge - state spectrum due to the subband mixing , whereas strong normal magnetic field drives the system into the phase of the quantum hall effect with the formation of chiral electron and hole edge states .
s. a. tarasenko , m. v. durnev , m. o. nestoklon , e. l. ivchenko , j .- w .
luo , and a. zunger , split dirac cones in hgte / cdte quantum wells due to symmetry - enforced level anticrossing at interfaces , phys .
b * 91 * , 081302 ( 2015 ) .
s. a. tarasenko , m. o. nestoklon , m. v. durnev , and e. l. ivchenko , zero - field splitting of dirac cones in hgte / cdhgte quantum wells , proc . of 22nd int .
`` nanostructures : physics and technology '' ( 23 - 27 june , saint - petersburg , russia ) , pp . 174 - 175 ( 2014 ) .
p. olbrich , c. zoth , p. vierling , k .-
dantscher , g. v. budkin , s. a. tarasenko , v. v. belkov , d. a. kozlov , z. d. kvon , n. n. mikhailov , s. a. dvoretsky , and s. d. ganichev , giant photocurrents in a dirac fermion system at cyclotron resonance , phys .
b * 87 * , 235439 ( 2013 ) . k .-
dantscher , d. a. kozlov , p. olbrich , c. zoth , p. faltermeier , m. lindner , g. v. budkin , s. a. tarasenko , v. v. belkov , z. d. kvon , n. n. mikhailov , s. a. dvoretsky , d. weiss , b. jenichen , and s. d. ganichev , cyclotron - resonance - assisted photocurrents in surface states of a three - dimensional topological insulator based on a strained high - mobility hgte film , phys . rev .
b * 92 * , 165314 ( 2015 ) . l. g. gerchikov and a. v. subashiev , non - monotonic behaviour of the energy gap in the film made of a gapless semiconductor , fiz
. tekh . poluprovodn .
* 23 * , 2210 ( 1989 ) [ sov .
. semicond .
* 23 * , 1368 ( 1989 ) ] .
e. i. rashba .
properties of semiconductors with an extremum loop .
i. cyclotron and comninational resonance in a magnetic field perpendicular to the plane of the loop , fiz .
tela * 2 * , 1874 ( 1964 ) .
m. orlita , k. masztalerz , c. faugeras , m. potemski , e. g. novik , c. brne , h. buhmann , and l. w. molenkamp , fine structure of zero - mode landau levels in hgte / hg@xmath260cd@xmath261te quantum wells , phys .
b * 83 * , 115307 ( 2011 ) .
m. zholudev , f. teppe , m. orlita , c. consejo , j. torres , n. dyakonova , m. czapkiewicz , j. wrbel , g. grabecki , n. mikhailov , s. dvoretskii , a. ikonnikov , k. spirin , v. aleshkin , v. gavrilenko , and w. knap , magnetospectroscopy of two - dimensional hgte - based topological insulators around the critical thickness , phys .
b * 86 * , 205420 ( 2012 ) .
b. buttner , c. x. liu , g. tkachov , e. g. novik , c. brune , h. buhmann , e. m. hankiewicz , p. recher , b. trauzettel , s. c. zhang , and l. w. molenkamp , single valley dirac fermions in zero - gap hgte quantum wells , nat .
* 7 * , 418 ( 2011 ) . | we present a theory of the electron structure and the zeeman effect for the helical edge states emerging in two - dimensional topological insulators based on hgte / hgcdte quantum wells with strong natural interface inversion asymmetry .
the interface inversion asymmetry , reflecting the real atomistic structure of the quantum well , drastically modifies both bulk and edge states . for the in - plane magnetic field
, this asymmetry leads to a strong anisotropy of the edge - state effective @xmath0-factor which becomes dependent on the edge orientation .
the interface inversion asymmetry also couples the counter propagating edge states in the out - of - plane magnetic field leading to the opening of the gap in the edge - state spectrum by arbitrary small fields . |
among the most intriguing objects in the x - ray sky are the non - nuclear ultra - luminous x - ray ( ulx ) sources in nearby galaxies .
this name describes those sources considerably more luminous than expected for a spherically - accreting object of typical neutron star mass . here , we define ulx sources to be those with apparent ( i.e. , assumed isotropically emitting ) intrinsic luminosities in excess of @xmath0 erg / s in the 0.5 - 8.0 kev bandpass . through the first 2 years or so of operation , the chandra x - ray observatory ( cxo ) has imaged enough nearby galaxies using the acis ccd imaging spectrometer to undertake a systematic and uniform analysis of their ulx population .
ultimately , we wish to know the full pedigree of these extreme objects : what are their origins and history ; why and how do they differ from their more - common low - lumiosity cousins ; what does the population of ulx sources reveal about the nature of galaxy formation and evolution ; and what influence do ulx sources have on their local environments ? here , we report principally the correlations between ulxs and global properties of their host galaxies .
based on integration time and the best distance estimate available , we selected all those galaxies for which @xmath1100 source counts are expected from a ulx . in integration time units of ks and distances in mpc , this corresponds to observations with @xmath20.12 . to date
, 85 galaxies in the cxc public archive meet this selection criterion . for all candidate galaxy acis images ,
the following steps were taken : * all events within the @xmath3 isophote were extracted from level 2 event files .
* x - ray sources ( to @xmath13.5 s / n ) were located using standard methods .
* source and local background spectra in the 0.5 8.0 kev energy range were extracted .
* source count rates were determined and the source list ordered by decreasing count rate .
* the ( binned ) time series for each source and for the entire image field were constructed ( the latter to help identify and eliminate high noise level intervals ) . *
beginning with highest count - rate sources , simple models were fit to the spectra to establish spectral shapes and source luminosities .
this process was extended to sources well below the @xmath0 erg / s ulx lower limit to ensure completeness of the sample for each galaxy .
sources within 5 of the host galaxy center were omitted from consideration .
* source positions were overlaid on optical images ( usually dss ) to crudely map source locations to morphological features of host galaxies and to help eliminate obvious foreground objects .
similarly , ulx candidate source positions were queried using the nasa / ipac extragalactic database to help eliminate known background qsos .
the cxo sample of galaxies is compared to the tully catalogue of 2368 nearby galaxies in the two figures shown above .
the cxo sample is composed of mostly nearby galaxies ( figure 1 , left , note the large step at the virgo cluster distance ) , with 1/2 the sample within 10 mpc , and they are typically brighter in blue luminosity ( figure 1 , right ) . throughout this work
, the blue luminosity is used as a proxy for galaxy mass .
although blue light is more sensitive to the properties of the stellar population of the host galaxy than is visible light , the b - v colors of the cxo sample all fall within the range of 0.5 to 1.0 magnitudes and so @xmath4 is a good measure of mass .
another means of displaying the dependence of the two samples on blue luminosity is shown in figure 2 .
the mean absolute blue magnitude of the cxo sample is -20.6 or slightly brighter than that of the tully catalogue at -19.4 .
the cxo sample of galaxies spans the entire range of hubble morphological types ( figure 3 ) .
there are 32 elliptical and lenticular galaxies and 50 spiral and irregular galaxies in the sample .
the distribution of the cxo sample of galaxies in blue and far - infrared ( fir ) luminosity space is shown in figure 4 .
the ellipticals are clustered near the upper left indicative of a relatively high mass per unit star - formation - rate ( sfr ) whereas the spiral galaxies generally span from lower - left to upper right in this plot going generally from small , late - type galaxies toward more massive early - type spirals .
note the several exceptions such as the starburst galaxy m82 ( @xmath5@xmath6 , @xmath7@xmath8 ) .
we have searched for correlations between the number and luminosities of ulx sources with numerous galaxy properties .
examples are shown in figures 5 and 6 .
the distribution of x - ray luminosities of individual ulxs against the host galaxy s fir luminosity is shown above and against the b luminosity is shown below .
there is a distinct separation between elliptical and spiral galaxies in fir luminosity but that simply reflects the sample ( see figure 4 ) .
there is also a trend toward brighter ulxs in the spiral galaxies .
however , there is also more ulx candidates in those galaxies and this trend may simply be a statistical fluctuation .
similarly , in blue light , there are more and brighter ulx candidates per galaxy in more massive ( higher @xmath4 luminosity ) galaxies .
one expects to find more ulxs in more massive galaxies and perhaps , again , the trend toward more luminous ulxs is simply a statistical fluctuation .
the possible correlation with fir luminosity is shown in a different way in figure 7 . here , only the ulx sources with @xmath9@xmath10 erg / s are displayed .
the figure at left shows that the brightest ulxs are indeed in the brightest fir galaxies ( although the distribution about the mean is broad , see figure 6 ) .
the mean x - ray luminosity of ulxs in ellipticals is less than that in the spiral galaxies .
at right we show that the number of ulxs per unit galaxy is strongly correlated with fir luminosity . in figure 8
we show the only other strong correlation we have found between the number and luminosities of ulx sources with galaxy properties , namely , the distance to the host galaxy s nearest neighbor ( measured in units of the @xmath3 isophotal diameter ) .
those with seperations @xmath111 are in the process of merging .
they show much more activity ( and high fir luminosity , typically ) and tend to have many ulxs .
ellipticals and spirals with more distant neighbors have much fewer ulxs and they are not as luminous although those in spirals are still more luminous than those in ellipticals .
in figure 9 is shown the distribution of the basic spectral parameters with the ulx source luminosities .
three simple models were used to characterize the 0.5 8.0 kev spectra of the ulx candidates .
statistically , about 50% were best - fit using an absorbed power law ( mean spectral index for the group is 1.67 ) and about 50% are absorbed thermal emission spectra ( mean temperature 2.9 kev ) .
only 5 or about 4% were best represented by an absorbed blackbody spectrum . of course
, many of the higher count sources will require more complex spectral models .
this is work in progress .
so far , no obvious correlation has been found between spectral properties and ulx luminosity nor with galaxy properties .
the luminosity functions of the entire cxo sample of ulx candidates is shown in figure 10 along with the luminosity functions of the spirals and ellipticals . also shown
are a few of the galaxies with particularly high numbers of ulxs per galaxy .
the horizontal scale has been extended to well below the ulx lower limit of @xmath0 erg / s to clearly show the incompleteness at low luminosities that results from our focus on the brightest objects .
there are several features of interest . clearly , the highest ulx luminosities in the spiral galaxies are much higher than those of the ellipticals and spirals dominate the total luminosity function throughout the entire range @xmath9@xmath10 .
( however , only @xmath13/8 of the galaxies are ellipticals and lenticulars , a factor that has not been accounted for here . ) at the very highest end , the spirals are dominated by the single galaxy pair ngc 4038/9 , with about 40% of the ulxs brighter than @xmath12 erg / s within that one system .
the ellipticals exhibit a very different luminosity function .
the slope of the elliptical galaxy luminosity function above @xmath13 erg / s is -2.3 compared to -0.9 for spirals in the range @xmath14 and to -1.0 for the entire sample on that range . above @xmath15 erg / s , there is a steep decline or cutoff in the luminosity function of the sample . there is also a distinct change in slope in the luminosity function of the elliptical galaxies at @xmath13 erg / s .
this is a real feature as opposed to the gradual flattening at low luminosities due to incompleteness ( occuring in most of the curves at around @xmath16 erg / s ) .
because of this incompleteness , it is not possible to distinguish any change in slope at @xmath17 , the eddington limit for typical neutron star accretors .
the correlation of ulx x - ray luminosity with host galaxy fir luminosity discussed previously is evident in the luminosity functions shown in figure 11 . here
, the cxo sample has been partitioned into fir luminosity ranges . with the exception of 2 very bright sources in the @xmath5@xmath18 to @xmath19 erg / s range
, there is a definite trend with the brightest ulxs occuring in the fir - brightest hosts .
( the 2 `` exceptions '' are both located in a galaxy with @xmath5@xmath20 , the brightest in the group and just below the upper limit of the range . )
we have analyzed a sample of 85 galaxies available in the chandra archive of acis images .
about 120 ulx candidates have been identified and their spectral properties determined .
few correlations have been found so far between the number and luminosities of the ulx candidates and properties of the host galaxy population .
two correlations described here are with the galaxy s fir luminosity and with the galaxy s interacting or merging status as measured by the distance to its nearest neighbor . * further searches for correlations with other galaxy properties including the distributions of sources within the galaxy , correlations with spiral arms , bulges , bars , globular clusters , etc . , and agn activity . * refined modeling of the observed x - ray spectra , particularly of the highest count - rate sources where this is justified statistically .
* source position refinement by registering to accurately - known positions of optical objects in the field , although chandra positions are typically known to @xmath10.6 arcsecs .
* extension to other wavelengths , particularly using archival hst and radio data .
* examination of sample biases and completeness . | the more than 80 nearby galaxies imaged with the chandra advanced ccd imaging spectrometer have been analyzed in a search for ultra - luminous x - ray ( ulx ) sources .
the sample of galaxies span the range of hubble morphological types and include galaxies of various mass , gas content , and dynamical state .
x - ray characteristics of the resulting ensemble of ulx candidates are reported and correlations with properties of the host galaxies are presented .
support for this research was provided in part by nasa / chandra grant ar2 - 3008x to d.a.s . |
lower bounds on neutrino masses have not been found yet in direct laboratory experiments . in particular
, in the case of neutrinoless double - beta decay one expects a decrease of an upper bound on the majorana mass @xmath2 only .
this current improvement of upper limits on neutrino masses takes a long time and strong efforts .
however , we can not justify on this way whether neutrino is really the majorana particle or it can be the dirac one , or a mixture of them ( zkm - neutrino ) . in this connection
let us recall the old experiment by davis@xcite who demonstrated that neutrino and antineutrino are indeed different particles if we are using @xmath3 from the beta - decay @xmath4 as the incident `` neutrino '' for the capture process @xmath5 .
negative result of the experiment@xcite , @xmath6 , is not an evidence that @xmath7 and @xmath3 are the dirac neutrinos with the fixed lepton number @xmath8 . in such experiments the
helicity @xmath9 ( upper signs for @xmath7 ) is appropriate quantum number which is conserved due to the @xmath10 law of charged current weak interaction rather the lepton number .
both the right - handed majorana neutrino and the dirac antineutrino with the same helicity @xmath11 could be emitted in the beta decay @xmath4 with the following suppression of the spin - flip @xmath12 in the process of capture in the chlorine detector ( the latter is true to the order of @xmath13 in the cross section ) .
thus this example demonstrates the well - known fact that _ in the massless limit @xmath14 majorana and dirac neutrinos are not distinguishable_. we can not mark any difference between the fully - polarized right - handed majorana neutrino @xmath15 and the right -handed dirac antineutrino @xmath16 as well as between the left - handed majorana neutrino @xmath17 and the dirac left - handed one , @xmath18 ( see below section 3 ) . in turn , if an incident neutrino flux became _ partially - polarized _ this would give a chance to distinguish these particles .
we propose here a new way for distinction of majorana and dirac neutrino in the solar neutrino experiments by studying the profiles of the electron spectra in the low - energy @xmath0-scattering for incident @xmath19 and @xmath20 fluxes .
it seems possible when solar neutrino flux is partially - polarized .
the ultrarelativistic neutrinos produced in thermonuclear reactions within solar interior are evidently the left - handed ones ( fully - polarized @xmath17 ) and one needs to assume some mechanism for their conversion to the right - handed neutrinos . first , the conversions @xmath21 in the majorana case or @xmath22 in the dirac case are obviously based on the assumption of a non - vanishing neutrino mass @xmath23 .
this leads to nonzero neutrino diagonal @xcite and transition @xcite dipole moments and , therefore , to the possible neutrino spin @xcite and spin - flavor precession @xcite in vacuum in the presence of an external magnetic field . in a medium neutrino oscillations and spin - flavor precession
can occur as the resonant conversions @xmath24 @xcite and @xmath25 @xcite .
the spin - flavor conversion in combination with the msw - mechanism can lead to the right - handed majorana neutrino production ( @xmath21 @xcite , see below section 2 ) , i.e. to a mixture of the left - and right - handed active neutrinos as a partially - polarized @xmath26 neutrino flux incident upon underground detectors .
in contrast to the majorana neutrino case , for the same conditions in the sun the right - handed dirac neutrinos produced via the spin - flip @xmath22 or in the cascade conversions @xmath27 appear to be sterile ones with respect to the @xmath0-scattering in detectors .
notice that necessary large values of transition moments ( even without direct proportionality to @xmath28 ) can be obtained in some extended models obeying all known laboratory , astrophysical and cosmological constraints on neutrino masses and on its dipole moments . for all majorana and dirac neutrinos with a mass below @xmath29 the most restrictive limit on dipole or transition magnetic and electric moments @xmath30 arises from the absence of anomalous neutrino emission from the red - giant cores just before helium ignition@xcite .
the condition @xmath31 follows from a kinematic limit on the neutrino energy @xmath32 lost in the plasmon decay @xmath33 since the plasma frequency in a degenerate electron gas of red- giants is bounded in the same region , @xmath34 .
a large dirac neutrino magnetic moment ( both diagonal and transition ones , @xmath35 ) was also excluded from sn1987a neutrino events in the + kamiokande and imb detectors .
this is due to non - observation there of a hard energy tail stipulated by the sterile @xmath36 emission from a hot supernova core@xcite .
these neutrinos could be produced within core via the electromagnetic scattering ( see below eq .
( [ em ] ) ) and then be converted to the active @xmath17 in the intergalactic magnetic field .
the absence of sn1987a constraint in the majorana case means that the model - dependent estimate of @xmath37 @xcite seems to be less consistent even for light neutrinos suitable for the resonant spin - flavor or the msw conversions in the sun .
therefore the laboratory constraint from reactor antineutrino experiments which is common for diagonal and transition magnetic moments , @xmath38 @xcite , remains an upper estimate of majorana neutrino transition moments corresponding to effective neutrino conversions within solar convective zone with magnetic fields of order @xmath39 .
on the other hand , in magnetic hydrodynamics one can not exclude solutions with a strong magnetic field near bottom of the convective zone of the sun , @xmath40@xcite , and even larger within solar core for equilibrium hydromagnetic configuration in the gravitational field , @xmath41@xcite . as a result even for the case when the limit @xmath42@xcite is valid we may apply some mechanisms for effective spin - flavor conversions .
notice also that the most stringent constraints on transition magnetic moments + @xmath43 were derived in @xcite from the primordial nucleosynthesis bound on additional neutrino species and from a supernova energy loss argument .
these bounds were obtained in @xcite for _ active - sterile neutrino conversions neglecting neutrino mixing _ and in a medium with random magnetic fields .
therefore this approach has no relation to our consideration here since we are using the models @xcite _ for active - active neutrino conversions_. in the end of section 2 we consider experimental bounds on @xmath15 and give a theoretical interpretation of some contradictions in the model used .
in particular , we consider an important question how to avoid some known lacks of the spin - flavor scenario @xcite with the use of large neutrino magnetic moments that inevitably leads to contradiction with non - observation of temporal variations of the solar neutrino flux in the most of current experiments .
then in section 3 we show why the scattering of partially - polarized neutrino flux off electrons should be absolutely different for majorana and dirac neutrino provided that their right - handed components interact ( do not interact in the dirac case ) with electrons . in section 4
we discuss results and crucial parameters of the model used .
the main assumption here is the presence of the right - handed neutrinos @xmath15 , @xmath36 produced by some mechanism within the sun . for instance , in the spin - flavor scenario involving majorana neutrinos the sun would be a source for antineutrinos , some of which could be @xmath15 s by a combination of spin - flavor ( @xmath25 ) and flavor ( @xmath44 ) oscillations ( or due to permutation of steps above , @xmath45 ) .
this scenario was in detail elaborated in @xcite for the case of twisting magnetic field that allowed authors to avoid the suppression of the cascade conversion @xmath21 caused by the almost full compensation of the partial ( two - step ) amplitudes , @xmath46 , ( for a small neutron abundance in the sun ) .
moreover , authors @xcite found conditions when either @xmath17- @xmath15-system decouples from @xmath47 and @xmath48 or the triple resonances merge and a resonant transition leads to a complete @xmath21-conversion . in the case of separation of @xmath17- @xmath15-system the adiabaticity condition is not fulfilled ( at least for the convective zone in the sun ) while for the triple resonance the relevant adiabaticity parameters at the merging point depend on neutrino energy . as result
authors@xcite predict that the @xmath15-flux should have a peak at the energy @xmath49 defined by the resonant condition .
explicitely this occurs for @xmath50 at @xmath51 , where @xmath52 is the mixing parameter ; @xmath53 and @xmath54 is the neutrino squared mass difference ; @xmath55 is the neutrino transition magnetic moment ; @xmath56 is the amplitude of the twisting magnetic field * b*@xmath57 which is perpendicular to the momentum * k * of an ultrarelativistic neutrino , @xmath58 . since the position and the width of the peak strongly depend on the magnetic field strength in the region of resonance , the energy spectrum of resonantly emitted solar @xmath15 s
should exhibit characteristic time dependence .
in particular , depending on the signs of @xmath59 in the northern and southern solar hemispheres the semiannual neutrino flux variation ) leads to the minimum of the neutrino interaction term @xmath60 for the two season positions of the earth ( in june and december ) when the solar core is viewed from the earth through the solar equator .
] can be either enhanced or suppressed . moreover ,
the energy at which the @xmath15-spectrum achieves the maximum should vary in time .
note also other ways of the majorana @xmath3-production like @xmath61-decay ( to majoron ) enhanced in solar matter @xcite . in this case
we do not need any strong magnetic field in the sun or the presence of a large neutrino magnetic moment .
however , the @xmath3-spectrum differs in such case from the initial one for left - handed neutrinos that prevents from the approach developed below .
the active dirac antineutrino @xmath62 never can be produced in the sun due to absence there of the annihilation process @xmath63 which is important , for instance , in a supernova .
however , in the same twisting magnetic field the resonant @xmath64-conversions lead to the production of the sterile neutrinos @xmath65 through a diagonal magnetic moment @xmath66@xcite ( neglecting neutrino flavor mixing ) or analogously to the majorana case in the presence of mixing and via a transition magnetic moment for the system of the active @xmath67 , @xmath68 neutrinos and their sterile @xmath69 , @xmath70- components . on the other hand ,
it is well - known that the kamiokande data already yield restrictive limits on a solar @xmath15-flux at the high energy region , @xmath71 @xcite .
this limit was found from an isotropic background accounting also for @xmath15 s due to the relatively large cross - section @xmath72 with the isotropic distribution of positrons , where @xmath73 .
the best relative limit is at @xmath74 where the @xmath15 flux is less than 5.8 % of @xmath75 solar @xmath7 s at 90% cl .
however , since the @xmath21-conversion within the sun depends on neutrino energy , sign of @xmath59 , etc . , and
due to the absence of the kamiokande limit for such process below @xmath76 we can not exclude the presence of right - handed majorana neutrinos emitted by the sun , in particular , for the low - energy beryllium neutrinos ( the line @xmath77 ) .
one expects to observe these neutrinos in the borexino ( 1998 ? ) and the hellaz experiments where a scintillation detector@xcite and a helium gas at azote temperature@xcite should have a low - energy threshold .
notice that for the beryllium neutrino line emitted by the sun the reaction @xmath78 vanishes because of its high threshold @xmath79 , but the parallel branch@xmath80-scattering contributes to the counting of neutrino events starting from a detector threshold . in
anyway , since we assume here the right - handed neutrinos are producing in the sun and the first scenario @xcite realizes , there appear some known objections concerning a large value @xmath81 for an effective neutrino spin reversal or non - observation of the semiannual flux variations in the current neutrino experiments .
really , from the astrophysical bound on any dirac and majorana neutrino magnetic moments @xmath82@xcite it follows that necessary neutrino spin reversal when the dimensionless parameter @xmath83 should be of the order unity , @xmath84 , is provided by the sufficiently strong magnetic fields @xmath85 applied along a half of the width of the convective zone , @xmath86 .
even such magnetic fields would not be in contrary with some estimates in magnetic hydrodynamics@xcite .
meanwhile , a more weakened and model - independent laboratory constraint @xmath87@xcite seems to be reasonable for a majorana neutrino for which the sn1987a energy loss argument@xcite does not work in contrast to the dirac case .
this means that magnetic field strength in the sun could be less than for the dirac case , or quite realistic values @xmath88 would be sufficient for effective spin - flavor conversions .
non - observation of the semiannual flux variations , perhaps , is explained either by unsuccessful sign of the angular velocity @xmath59 for the twisting magnetic field in both solar hemispheres ( in mechanism @xcite under consideration , see above ) ) or even more simpler than in @xcite without any specific geometry of a magnetic field : by the presence of random magnetic fields which always exist in addition to a regular magnetic field and can have the same strength of the r.m.s .
field @xmath89 .
in contrast to regular field , random magnetic fields do not disappear at the equatorial plane while @xmath90 elsewhere and their scale @xmath91 is much less than the regular field one .
thus one can pronounce that experimental bounds at present do not exclude the possible observation of the low - energy @xmath15 s in future solar neutrino experiments .
both right - handed majorana ( active ) neutrinos and right - handed ( sterile ) dirac neutrinos can influence the observed neutrino events but in different ways .
really , below we find that for partially polarized electron neutrino , @xmath92 , where @xmath93 is the spin density matrix of the solar neutrino flux ( @xmath94 ) with the initial momentum directed along z - axis , * k*@xmath95= @xmath96 , the differential cross - sections of weak interactions , @xmath97 , are the different ones for dirac and majorana neutrinos .
the main point is that the difference of the cross - sections gives the different electron energy spectrum profiles for two kinds of neutrino . as well as in the case of a dirac neutrino in @xcite the polarization components of electron majorana neutrinos in eq .
( [ densitymatrix ] ) are ( left - handed majorana or dirac neutrinos ) . ] , @xmath98 with the essential distinction that now the right - handed component @xmath15 is active
. the spin density matrix for the muon neutrinos is given by the same eq.([densitymatrix ] ) with the change of the index @xmath99 to @xmath100 and of the subscripts @xmath101 , @xmath102 to @xmath103 , @xmath104 for the polarization components ( [ components ] ) .
notice also that , in contrast to @xcite , the normalization of the electron spin density matrix eq .
( [ densitymatrix ] ) , in general , is different from unity , @xmath105 using eq .
( [ densitymatrix ] ) and summing over the helicities of the partially polarized initial neutrino flux coming from the sun to the detector , @xmath106 we obtain in the ultrarelativistic limit @xmath107 the well - known density matrix in @xmath108-representation ( compare formula ( d.54 ) in the book @xcite for a= 1 ) @xmath109 which allows us to calculate in the subsection 3.2 the different mean @xmath110-scattering cross - sections averaged over solar neutrino spectrum , @xmath111 and @xmath112 .
recall that the index @xmath113 refers to the rest frame of a particle , @xmath114 , or @xmath115 is the lorentz - invariant due to the normalization of the spin 4-vector @xmath116 , @xmath117 , in the mishel - wightman density matrix @xmath118 which transits to ( [ densitymatrix1 ] ) for @xmath119 in the ultrarelativistic limit @xmath14 @xcite .
there are also two additional density matrices with the mixed flavors in the interference of the weak and electromagnetic amplitudes , @xmath120 where the transversal polarization values are given by @xmath121 we have determined here the mixed transversal polarizations ( [ mixturepol ] ) as well as the electron neutrino polarization components @xmath122 , @xmath123 and the muon neutrino ones @xmath124 , @xmath125 entering eq .
( [ components ] ) as the functions of the survival probability @xmath126 , of the left - handed muon neutrino probability @xmath127 and of the normalization coefficient @xmath128 given by eq .
( [ normaliz ] ) .
all functions @xmath129 and @xmath128 depend on the neutrino energy @xmath49 , the mixing parameters @xmath130 , the magnetic energy @xmath60 and the matter density .
these functions are very complicated in the scenario@xcite .
for instance , the merging of the three transitions , @xmath131 , @xmath132 , @xmath133 , yields the large @xmath134-transition probability given by eq .
( 48 ) in @xcite , @xmath135 , which after the averaging over fast oscillations gives @xmath136 ^ 2}. \label{transition}\ ] ] for the beryllium line @xmath137 the averaging of the cross - sections with the help of the @xmath138-function , @xmath139 , takes off the energy dependence in the probabilities like eq .
[ transition ] , so that the transition probabilities and the polarization vector eq .
[ components ] depend on the fundamental vacuum constants , @xmath140 , @xmath141 and on the magnetic field parameter @xmath60 which is changing very slowly comparing with the event counting at the underground detectors .
use of @xmath142-function means also that we have neglected the thermal and doppler broadening of the beryllium line @xcite . as result in the case of the beryllium neutrinos
we can parameterize the recoil electron spectra over the values @xmath143 and @xmath144 independently on neutrino energy as well as for the dirac neutrino conversions @xmath145@xcite which do not depend on neutrino energy at all . 0.5 cm now using the spin density matrices eq .
( [ densitymatrix1 ] ) , eq .
( [ mixture ] ) with the given polarization components eq .
( [ components ] ) , eq . ( [ mixturepol ] ) let us consider what kind of differences of the majorana and dirac neutrino scattering off electrons can arise in underground experiments .
the key argument here is the difference of the matrix element for the majorana neutrino current , @xmath146 from the corresponding one for the dirac neutrino ( @xmath147 ) or for the dirac antineutrino ( @xmath148 ) .
absence of the vector part in the matrix element eq .
( [ majorcurrent ] ) is crucial for the neutrino interactions with particles in the case of partial polarization , and it is the direct consequence of the charge - conjugation property @xmath149 for the second - quantized majorana field@xcite @xmath150 the matrix element for the @xmath110-scattering @xmath151 consists of the two terms for a majorana neutrino , @xmath152\times \nonumber\\ & & \times \bigl [ \bar{u}_e(p_2)(g_{il}\gamma_{\alpha}\frac{(1 - \gamma_5)}{2 } + g_r\gamma_{\alpha}\frac{(1 + \gamma_5)}{2})u_e(p_1)\bigr ] , \label{weak}\end{aligned}\ ] ] and @xmath153\bigl [ \bar{u}_e(p_2)\gamma^{\alpha}u_e(p_1)\bigr ] .
\label{electromag}\ ] ] here @xmath154 is the momentum transfer ; @xmath155 and @xmath156 are the constants of the standard model with the weinberg angle parameter @xmath157 , where upper ( lower ) sign for @xmath158 corresponds to the subscript @xmath159 for electron neutrino ( @xmath160 for muon neutrinos ) . we have meant here that in the presence of a flavor mixed neutrino flux from the sun described by the spin density matrices ( [ densitymatrix1 ] ) , ( [ mixture ] ) the @xmath161- scattering weak amplitude @xmath162 conserving helicity (= chiralities l or r ) adds the coherent electromagnetic amplitude @xmath163 of the spin - flavor transition @xmath164 in the case of electron neutrinos ( @xmath165 ) or of the inversed process @xmath166 in the case of muon neutrinos , @xmath160 .
notice that we do not distinguish the mass eigen - states in the electromagnetic scattering @xmath167 through the transition dipole moments @xmath168 from the flavor states in the weak amplitude eq .
( [ weak ] ) because of the inclusion of mixing angles into @xmath168 and the ultrarelativistic approximation used in spinors .
thus , we put @xmath169 elsewhere except for @xmath168 themselves or even do not bother about the limit @xmath14 for a large moment @xmath170 in some known extended models where @xmath171 does not depend on neutrino masses .
it is useful to retain @xmath172 in cp - violating term in eq .
( [ electromag ] ) ( @xmath173 ) without the obvious change @xmath174 for the identical chirality and helicity ( in ultrarelativistic approximation used ) because of the mixed polarization eq .
( [ densitymatrix1 ] ) , or due to an arbitrary @xmath175 .
using eq .
( [ densitymatrix1 ] ) , eq .
( [ weak ] ) , eq .
( [ electromag ] ) we obtain the differential cross - section for the majorana neutrino scattering off electrons , @xmath176 where the weak amplitude eq .
( [ weak ] ) leads to the sum of contributions of @xmath0- and @xmath177 -scattering @xmath178 + \nonumber\\ & & + ( a - p)\bigl [ g_r^2 + \bigl ( 1 - \frac{t}{\omega_1}\bigr ) ^2g_{el}^2 - \frac{m_et}{\omega_1 ^ 2}g_{el}g_r\bigr ] + \nonumber\\ & & + p_1\bigl [ g_{\mu l}^2 + g_r^2\bigl ( 1 - \frac{t}{\omega_1}\bigr ) ^2 - \frac{m_et}{\omega_1 ^ 2}g_{\mu l}g_r\bigr ] + \nonumber\\ & & + ( 1 - a - p_1)\bigl [ g_r^2 + \bigl ( 1 - \frac{t}{\omega_1}\bigr ) ^2g_{\mu l}^2 - \frac{m_et}{\omega_1 ^ 2}g_{\mu l}g_r\bigr ] \}~. \label{weak1}\end{aligned}\ ] ] and the electromagnetic amplitude eq .
( [ electromag ] ) gives @xmath179 . \label{em}\ ] ] both these terms in eq .
( [ total ] ) do not depend on the azimuthal angle @xmath180 .
however , the last interference term in eq .
( [ total ] ) , @xmath181 - \nonumber\\ & & - \bigl ( \frac{d_{12}}{\mu_b}\bigr ) \bigl [ ( g_{el } + g_{\mu l } + 2g_r)\bigl ( \hat{\vec{k}}_1 \cdot [ \vec{p}_2\times ( \vec{\xi}^{(\mu \tilde{e})}_{\perp } + \vec{\xi}^{(e \tilde{\mu } ) } _ { \perp } ) \bigr ) \left ( 2 - \frac{t}{\omega_1}\right ) + \nonumber\\ & & + ( g_{\mu l } - g_{e l})\bigl ( \hat{\vec{k}}_1 \cdot [ \vec{p}_2\times ( \vec{\xi}^{(\mu \tilde{e})}_{\perp } - \vec{\xi}^{(e \tilde{\mu } ) } _ { \perp } ) ] \bigr ) \frac{t}{\omega_1}\bigr ] \bigr \}~ , \label{int}\end{aligned}\ ] ] depends on that angle via the sum in braces , or the cross - section can be rewritten as the product @xmath182\times \nonumber\\ & & \times \bigl [ ( g_{el } + g_{\mu
l } + 2g_r)\bigl ( 2 - \frac{t}{\omega_1}\bigr ) \bigl ( \mid \vec{\xi}_{\perp}^{(\mu \tilde{e } ) } + \vec{\xi}_{\perp}^{(e\tilde{\mu})}\mid \bigr ) + \nonumber\\ & & + ( g_{\mu l } - g_{el})\frac{t}{\omega_1}\bigl ( \mid \vec{\xi}_{\perp}^{(\mu \tilde{e } ) } - \vec{\xi}_{\perp}^{(e\tilde{\mu})}\mid \bigr ) \bigr ] , \label{sum}\end{aligned}\ ] ] where @xmath183 , @xmath184 is the maximum kinetic energy of the recoil electron , and the transversal polarization values eq .
( [ mixturepol ] ) , @xmath185 , @xmath186 , are the external parameters which are given by a neutrino dynamics within solar interior . the polarization vectors @xmath187 should follow the solar magnetic field direction at the resonant point along neutrino trajectory obeying the extremum conditions @xmath188 , @xmath189 or @xmath190 , and the bargman - mishel - telegdi equation for the electroneutral fermion spin motion in the external magnetic field * b*@xcite , @xmath191\bigr ) ~.\ ] ] here the unit vector * n * = * k*@xmath192 and the neutrino momentum * k*@xmath193 are orthogonal to the transversal polarization , @xmath194 , and @xmath195 is the flavor index .
since all transversal polarization vectors are aligned along the magnetic field * b * we may change the modulus of the vector difference and sum @xmath196 in the interference cross - section ( [ sum ] ) to the difference and sum of the vector values @xmath197 given by eq .
( [ mixturepol ] ) .
it is obvious that the interference term ( [ int ] ) vanishes for the separated electron ( @xmath119 , @xmath198 and muon ( @xmath199 ) systems .
let us recall that , in general , the electron neutrino survival probability @xmath200 , the muon neutrino parameter @xmath201 and the normalization factor @xmath202 appearing in eq .
( [ weak1 ] ) eq .
( [ int ] ) , are some functions of the incident neutrino energy @xmath203 , of the solar magnetic field @xmath56 multiplied by @xmath55 , and of the fundamental vacuum parameters @xmath204 , @xmath205 .
analogously to eq .
( [ em ] ) which is proportional to the factor @xmath206 in the interference cross section eq .
( [ int ] ) both terms that are linear in @xmath207 and @xmath208 have the same cp - signs while the intrinsic cp - signs of these dipole moments are opposite . actually , in eq .
( [ int ] ) this coincidence of cp - signs is due to the different cp - signs of the kinematic factors before transition moments , or due to the presence of the third additional 3-vector @xmath209 before @xmath208 . on the other hand , if cp - holds : @xmath210@xmath211@xmath212@xmath213 , one term survives in eq .
( [ em ] ) and eq .
( [ int ] ) only : either @xmath214 or @xmath173 .
notice that for the case @xmath215 the cross - section eq .
( 12 ) coincides after a change of notations with the recent result @xcite obtained by i.v .
gaidaenko .
let us compare the result eq .
( [ total ] ) with the analogous eq.(11 ) obtained in @xcite for the dirac neutrino conversions @xmath216 in the sun with the following neutrino scattering off electrons in underground detectors .
the electromagnetic term eq .
( [ em ] ) is the same as in the dirac case with the trivial change @xmath217 for a majorana neutrino .
however , the main term , eq .
( [ weak1 ] ) , and the interference one eq .
( [ int ] ) have electron energy spectra which are different from the spectra for the corresponding terms in @xcite . really , repeating calculations of @xcite with use of eq .
( [ densitymatrix1 ] ) for the separated @xmath218-system ( @xmath119 ) we find @xmath219 , \label{weak2}\ ] ] and @xmath220\right ) , \label{int1}\ ] ] where the last interference term is exactly eq .
( 11c ) in @xcite , but the weak interaction term eq .
( [ weak2 ] ) is two times bigger than eq .
( 11a ) in @xcite .
our cross - sections @xmath221 eq .
( [ weak1 ] ) , eq .
( [ weak2 ] ) coincide with the standard results @xcite for the scattering of the massless , fully - polarized neutrinos . for instance , for the separated @xmath222-system ( a = 1 )
this coincidence , ( @xmath223 , is shown in fig . 1 by the common line `` a '' which describes both the cross section of the @xmath224-scattering for the left - handed majorana neutrinos with the substitution @xmath225 ( a = p = 1 ) to eq . ( [ weak1 ] ) and the cross section eq .
( [ weak2 ] ) for the fully - polarized dirac neutrinos with the substitution p = 1 .
one can also check from eq .
( [ weak1 ] ) that the right - handed fully polarized majorana neutrinos , @xmath226 , a = 1 , p = 0 , are described by the standard cross - section for the dirac antineutrinos ( @xmath227 and @xmath228 in eq .
( [ weak2 ] ) which , in contrast to the majorana @xmath15 , can not be observed in solar neutrino flux ( line `` d '' in fig .
moreover , in the particular case @xmath119 and after a trivial transition from cm to the laboratory frame of reference we have found the full coincidence of our eq .
( [ weak1 ] ) with the old result eq .
( 3.9 ) in @xcite .
if a sharp difference in the slope of lines for the majorana neutrinos ( `` b '' , `` c '' , `` d '' lines in fig.1 ) and for the dirac ones ( `` e '' , `` f '' ) was marked in the borexino or hellaz experiment this would be a test whether the solar neutrinos are the majorana particles or they are the dirac ones .
_ it does not matter which concrete value of unknown parameter @xmath60 influencing the survival probability p occurs within the sun _ : all corresponding lines given in fig.1 for the same probabilities ( pairs ( `` b '' , `` e '' ) and ( `` c '' , `` f '' ) ) have quite different slopes . emphasize that we have simplified the observable averaged cross sections + @xmath229 using @xmath142-function for the beryllium neutrinos .
all lines are interrupted at @xmath230 ( see also fig.2,3 ) and authors@xcite are planning to distinguish the `` signature edge '' of the recoil spectrum from the monoenergetic be neutrino flux .
this difference appears also for a non - separated system ( merging of resonances in the mechanism@xcite ) shown in fig .
2 for the particular equipartition case @xmath231 when the @xmath177-scattering is taken into account .
the similar difference between majorana and dirac neutrino cases can be found from the total cross section behavior shown in fig.3 for the separated electron neutrino system ( a=1 ) .
these cross sections @xmath232 contribute to the total number of neutrino events which can be observed in some different energy bins separated by the different thresholds @xmath233 .
for a low energy region all lines depending on the different parameters p which correspond to the majorana neutrino cross sections become parallel to each other ( lines `` b '' , `` c '' in fig .
3 ) but the analogous dirac neutrino cross - sections ( lines `` d '' , `` e '' with the same p as in the majorana case ) change their slope .
thus , if the magnetic field value @xmath56 in the sun is somehow changing and , therefore , influences the total cross section @xmath234 ( i.e. leads to the transition @xmath235 , or @xmath236 ) we can distinguish majorana and dirac neutrinos due to different slopes of the event spectra over @xmath233 , @xmath237 . the resonant spin precession can appear for another changing magnetic field configuration , for instance , for the _ linear -polarized alfven wave _ @xcite when a magnetic field direction ( transversal to the incident neutrino momentum ) could be fixed .
however , in contrast to @xcite , this mechanism is still not elaborated for @xmath25 and @xmath238-conversions .
1truecm the partially polarized electron neutrino flux provided by @xmath238 conversions within the sun can give an opportunity to distinguish majorana and dirac neutrinos in the borexino and in the hellaz experiments due to quite different profiles of recoil electron spectra(fig.1 - 3 ) .
notice that measurements of @xmath239 and @xmath240 would allow authors of the project@xcite to determine the neutrino energy @xmath203 .
the inclusion of other neutrino fluxes with the different @xmath241 is necessary for comparison with real electron spectra although we expect the same qualitative difference of the spectra for the majorana and the dirac cases .
really , it is obvious that @xmath242 exceeds @xmath243 in a wide energy region due to additional interaction of right - handed _ active _ majorana neutrinos with electrons in contrast to the _ sterile _ @xmath244 .
probably , in the superkamiokande or in the sno detectors this difference can be marked too . in spite of unknown dependence of the survival probability @xmath245 on the incident neutrino energy @xmath203 both the differential @xmath229 and the total @xmath246 cross - sections averaged over the boron neutrino spectrum @xmath247@xcite
should have quite different slopes depending on @xmath239 or @xmath233 .
one can easily check this statement substituting a fixed probe value of @xmath248 and integrating eq .
( [ weak1 ] ) , eq .
( [ weak2 ] ) over the spectrum @xmath247@xcite .
the absolute value of the averages @xmath229 , @xmath246 changes when @xmath249 varies that corresponds to some different levels of the electron neutrino deficit .
meanwhile both sets of lines ( for majorana or dirac neutrinos ) conserve different slopes for the two kinds of particles . another model independent way to distinguish majorana and dirac neutrinos was proposed in @xcite .
authors @xcite showed that nc events at the sno detector would not oscillate over time for @xmath19 and would oscillate in the case of @xmath20 .
the method is based on any active - active transitions @xmath250 in the solar magnetic field for majorana neutrinos and on active - sterile conversions in the dirac case .
a changing magnetic field @xmath251 influences neutrino flux variations so that one predicts @xmath252 for @xmath19 and @xmath253 for @xmath20 . in conclusion , without any knowledge of solar magnetic field direction and even for a magnetic moment obeying the astrophysical constraint @xmath257
@xcite one can distinguish majorana and dirac neutrino measuring the recoil electron spectra in the low energy @xmath0-scattering in underground detectors and comparing slopes of their profiles .
last remark concerns a possibility to observe some variations of neutrino flux in detectors with large statistics of events .
let us assume that the unknown parameter @xmath60 can change somehow in the sun influencing @xmath258 .
in this case the spectra @xmath255 will be displaced parallel to the initial one without change of a slope in the contrary to the spectra @xmath256 that should change its slope .
this behaviour of spectra would be a strong argument in favour of the mechanism@xcite as solution of the solar neutrino problem .
0.5 cm * acknowledgment * + i acknowledge george zatsepin for useful discussion and for pointing out refs .
@xcite , @xcite and correspondence with robert shrock and samoil bilenky regarding the refs .
@xcite , @xcite and the ref .
i am grateful to i.v .
gaidaenko who was first obtaining correct cross - section of the @xmath110-scattering of partially polarized majorana neutrinos and indebted to him for sending of his preprint @xcite that allowed me to improve a wrong previous version of eq .
this work was supported in part from rffr under grants n. 95 - 02 - 03724 and n. 97 - 02 - 16501 .
bilenky , introduction in feynmann diagrams and physics of the electroweak interaction ( energoatomizdat ( in russian ) , moscow , 1990 ) formula ( d.54 ) .
translated in basics of physics , frontiers ( 1994 ) .
the differential spectra eqs .
( [ weak1 ] ) and ( [ weak2 ] ) for the beryllium neutrino in the case of the separated @xmath222system ( the normalization parameter a in eq .
( [ normaliz ] ) equals to unity , @xmath259 ) : the lines `` b '' , `` c '' , `` d '' correspond to an incident majorana neutrino with the survival probability @xmath260 correspondingly , and the lines `` e '' , `` f '' describe cross sections for a dirac neutrino in the cases @xmath261 . the line `` a '' is the common one for majorana and dirac fully - polarized left - handed neutrinos with @xmath262 . the same spectra in the case of the equipartition @xmath263 , the parameter @xmath264 : the line `` a '' describes majorana neutrino , the line `` b'' dirac neutrino .
the total cross sections for the beryllium neutrino @xmath265 in dependence on the threshold energy @xmath233 . for a liquid scintillator
it assumes a low threshold @xmath233 of order @xmath266 a few hundreds kev .
the line `` a '' corresponds to the fully polarized left - handed neutrinos ( @xmath267 ) with @xmath268 ( ssm prediction without @xmath269- deficit ) ; lines `` b '' , `` c '' are plotted for majorana neutrino with the survival probabilities @xmath270 and @xmath271 correspondingly ; lines `` d '' , `` e '' are plotted for dirac neutrino with the same @xmath270 and @xmath272 . | if neutrino conversions within the sun result in partial polarization of initial solar neutrino fluxes then a new opportunity to distinguish majorana and dirac neutrinos by measuring the differential @xmath0-scattering cross section in solar neutrino detectors arises . the experiment like hellaz would be preferable in testing of recoil electron spectra differences initiated by majorana and dirac neutrinos since low energy recoil electrons ( @xmath1 ) can be detected and electron energy and direction can be determined with good precision .
pacs codes : 13.10.+q ; 13.15.-f ; 13.40.fn ; 14.60.gh ; 96.60.kx .
* key words * : neutrino , magnetic moment , magnetic fields , polarization . -1 cm 8.5 in 0.3 cm * can we distinguish majorana and dirac neutrinos in solar neutrino experiments ? * * v.b .
semikoz * |
gw vir variables belong to the spectroscopic class of the pg 1159 stars ( wesemael , green & liebert 1985 ) , which is named after the prototype pg 1159@xmath0035 .
these objects are strongly hydrogen - deficient post - agb stars which pass through the hottest stage of stellar evolution .
their effective temperatures range between 75 000 and 200 000k , surface gravities vary from @xmath1 $ ] .
the so - called born - again scenario ( a late thermal pulse which transferred these objects back to the agb followed by a second post - agb evolution , iben et al .
1983 ) is mainly accepted as an explanation for the h - deficiency and can reproduce well the observed abundances .
pg 1159 stars have spectra which are dominated by lines of , , and ( werner et al .
2004 ) , their atmospheres show a typical surface composition of he : c : o = 33:50:17 by mass . beside these main constituents
there are several lines of trace elements , such as neon , nitrogen , silicon , sulfur , phosphorus , and fluorine ( reiff et al .
2005 , werner et al . 2005 ) .
presently 37 pg 1159 stars are known , eleven of them proved to be pulsators .
the pulsating members of the pg 1159 class are referred as gw vir variables .
they are non - radial g - mode pulsators with periods from 300s up to 1000s , in some cases exceeding even 2000s ( nagel & werner 2004 ) .
the favored excitation mechanism for the pulsations is the @xmath2-mechanism associated with cyclic ionization of carbon and oxygen ( quirion , fontaine & brassard 2004 ) . in the @xmath3 diagram the gw vir variables are located among the pg 1159 stars in the so - called gw vir instability strip .
spectral analyses of pulsating and non - pulsating pg 1159 stars were used by dreizler & heber ( 1998 ) to define empirically the edges of this instability strip .
but it is still puzzling that also non - pulsating pg 1159 stars are located within the instability strip . in our analysis
we try to find more characteristic properties to distinguish between pulsating and non - pulsating members of this class .
for our analysis we selected pulsating and non - pulsating pg1159 stars for which high resolution ( r @xmath4 20 000 ) fuv spectra obtained with the far ultraviolet spectroscopic explorer ( fuse ) are available .
the resulting sample comprises eleven objects .
the fuse spectra are processed within the standard calfuse pipeline process .
a log of all observations used for this analysis is listed in table [ tab : log ] . besides the fuv spectra we also used spectra obtained with stis , ghrs and iue as well as optical spectra .
the model atmospheres and synthetic line profiles are computed with the tbingen model atmosphere package ( werner et al .
2003 , rauch & deetjen 2003 ) .
the line - blanketed nlte model atmospheres are in radiative and in hydrostatic equilibrium . besides the main constituents of the atmospheres of pg 1159 stars , helium , carbon , and oxygen , our model atmospheres also contain neon and nitrogen . for the abundances of these elements we use atmospheric parameters taken from the literature which are summarized in table [ tab : parameters ] . for neon an abundance of 2% mass fraction was assumed for all models , according to werner & rauch ( 1994 ) and werner et al .
although the abundances in the literature were mostly determined in analyses of optical spectra the synthetic spectra can also fit the fuv spectra well , which confirms the literature values for abundances in most cases . in fig.[fig : pg1159 ] we display the fuse spectrum of pg1159@xmath0035 together with our synthetic spectrum . as lines of sulfur and silicon were also identified in several objects we included those elements in the synthetic spectra , too .
both were treated with line formation calculations without back - reaction on the atmospheric structure .
we assumed solar abundances for both elements .
l l r c object & observation i d & & aperture + rxj2117.1 + 3412 & p1320501 & 8232s & lwrs + pg1144 + 005 & p1320201 & 6859s & lwrs + pg1520 + 525 & p1320101 & 3648s & lwrs + pg1159@xmath0035 & q1090101 & 6321s & lwrs + k1@xmath016 & m1031010 & 11271s & hirs + hs2324 + 3944 & p1320601 & 4004s & lwrs + abell 78 & b1100101 & 9972s & lwrs + & b1100102 & 7894s & lwrs + ngc 7094 & p1043701 & 23183s & lwrs + abell 43 & b0520202 & 12150s & lwrs + pg1424 + 535 & p1320301 & 11132s & lwrs + pg1707 + 427 & p1320401 & 14599s & lwrs + silicon is detectable in at least three objects , which are pg 1159@xmath0035 , and the two cooler stars pg 1424 + 535 and pg 1707 + 427 .
models with a solar si abundance can fit the doublets at 1122/1128 and 1393/1402 well . in all spectra sulfur lines
are detected , but our preliminary fits also suggest abundances less than solar . in fig .
[ fig : sis_lines ] we display part of the fuse spectrum of pg 1424 + 535 with a preliminary fit of the sulfur and silicon lines , both with solar abundances . in former analyses by dreizler & heber ( 1998 ) it was suggested that the nitrogen abundance is a characteristic difference between pulsating and non - pulsating pg 1159 stars , as nitrogen was detected in all gw vir pulsators with a rather high abundance of 1% by mass , while in stable pg 1159 stars no nitrogen could be detected , except for pg 1144 + 005 ( which is considered outside the instability strip ) . in order to confirm
previously determined n abundances we tried to fit the n resonance doublet at 1238/1242 . for this purpose
we also analysed the stis spectrum of pg 1159@xmath0035 , which has a high resolution ( 0.1 ) and high s / n . in this spectrum
the interstellar component of the resonance doublet is clearly separated from the photospheric component .
this allows to determine the n abundance much more precisely than before and it seems to turn out that the n abundance is also significantly lower , about 0.1% by mass , than suggested by dreizler & heber ( 1998 ) .
[ fig : n_comparison ] shows the n resonance doublets of three objects , the pulsators pg 1159@xmath0035 and pg 1707 + 427 and the non - pulsator pg 1424 + 535 .
while the comparison of pg 1707 + 427 and pg 1424 + 535 seems to confirm the characteristic difference in the n abundance , the new fit to the photospheric components in the stis spectrum of pg 1159@xmath0035 shows that the n abundance is only 0.1% by mass , but still higher than in the non - pulsator pg 1424 + 535 .
further analyses are necessary to confirm these preliminary results .
l r c r c r r r c object & @xmath5 & @xmath6 & & he & & & & puls .
+ & @xmath7kk@xmath8 $ ] & ( cgs ) & & + rxj2117.1 + 3412 & 170 & 6.0 & & 38.0 & 56.0 & & 6.0 & @xmath9 + pg1144 + 005 & 150 & 6.5 & & 39.0 & 58.0 & 1.5 & 1.6 & + pg1520 + 525 & 150 & 7.5 & & 44.0 & 39.0 & & 17.0 & + pg1159@xmath0035 & 140 & 7.0 & & 33.0 & 49.0 & 1.0 & 17.0 & @xmath9 + k 1@xmath016 & 140 & 6.4 & & 33.0 & 50.0 & & 17.0 & @xmath9 + hs2324 + 3944 & 130 & 6.2 & 21.0 & 41.0 & 37.0 & & 1.0 & @xmath9 + abell 78 & 110 & 5.5 & & 33.0 & 50.0 & 2.0 & 15.0 & + ngc 7094 & 110 & 5.7 & 36.0 & 43.0 & 21.0 & & & + abell 43 & 110 & 5.7 & 36.0 & 43.0 & 21.0 & & & @xmath9 + pg1424 + 535 & 110 & 7.0 & & 50.0 & 44.0 & & 6.0 & + pg1707 + 427 & 85 & 7.5 & & 43.0 & 38.5 & 1.5 & 17.0 & @xmath9 + | gw vir variables are the pulsating members in the spectroscopic class of the pg 1159 stars . in order to understand the characteristic differences between pulsating and non - pulsating pg1159 stars , we analyse fuse spectra of eleven objects , of which six are pulsating , by means of state - of - the - art nlte model atmospheres . the numerous metal lines in the fuv spectra of these stars allow a precise determination of the photospheric parameters .
we present here preliminary results of our analysis . |
it is well known that collisional hydrodynamic viscosity is too small to explain the inferred rate of angular momentum transport in accretion disks ( shakura & sunyaev 1973 ) .
one possible source of anomalous viscosity is turbulence resulting from the magneto - rotational instability , or mri ( velikhov 1959 ; balbus & hawley 1991 ) . in the linear phase , this instability derives its source of free energy from the radial gradient of the rotational flow in the presence of a _ weak _ magnetic field , i.e. , the mri requires the presence of a magnetic field , but is linearly stable if the field is too large .
the properties of mri turbulence , and its effect on angular momentum transport , depend in a fundamental way on the amplitude of the fluctuations in the non - linear state .
we expect this amplitude to be related to the non - linear saturation mechanism of a _ single _ mri mode .
a common saturation mechanism for linear instabilities is the relaxation of the free energy source through quasi - linear processes . for the case of the mri
, this would imply modification of the radial profile of the rotational velocity , but this mechanism may not be available in accretion disks and other astrophysical settings because of strong gravitational constraints that maintain a keplerian profile ( @xmath0 ) . instead , the mri may saturate ( cease linear growth ) if the non - linear evolution of the instability generates a _
mean _ component of the magnetic field that is sufficient to stabilize the mode .
we remark that for most plasma instabilities , if the energy source is fixed and unvarying ( and nonlinear coupling to other modes is ignored ) , the mode will grow without bound , until limited by dissipation or some physical dimension of the system . in this sense
, the mri might be atypical . in this paper
we investigate the linear growth and non - linear saturation mechanism and amplitude of a _ single _ mri mode in thick - disk geometry , i.e. , @xmath1 .
our primary tool is linear and non - linear mhd computation in cylindrical ( @xmath2 ) geometry which is periodic in the azimuthal @xmath3 and axial ( @xmath4 ) directiona .
we solve a model initial value problem in which the inner and outer radial boundaries are impermeable , perfectly conducting , concentric cylinders that can rotate independently at specified rates .
they are coupled to the internal flow by hydrodynamic viscosity .
the initial mean ( azimuthally and axially averaged ) profile @xmath5 is keplerian , and in most cases is assumed to be maintained for all times by external forces .
perturbations to the equilibrium that depend on ( @xmath2 ) are then introduced , and evolve dynamically according to the single fluid mhd equations .
linear growth rates and eigenfunctions are found by integrating the linearized , single fluid , visco - resistive mhd equations forward in time until an exponentially growing solution is obtained . the saturation mechanism and amplitude of the mode
are determined by solving the non - linear mhd equations for a single mode , with azimuthal mode number @xmath6 and axial mode number @xmath7 , beginning from a small initial perturbation . at finite amplitude
, the mode will interact with itself to modify the mean background state , which in turn alters both its temporal evolution and its radial dependence . in this sense
it differs from a quasi - linear calculation , in which only the effect of the _ linear _ eigenfunction on the background is accounted for .
our _ primary result _ is that , when the _ mean _ flow profile is maintained as keplerian , the mri can saturate due to the generation of a mean component of the magnetic field that is sufficiently large to linearly stabilize the instability .
the mechanism for field generation is by means of a _
mean _ electric field @xmath8 = @xmath9 = @xmath10 ( where @xmath11 denotes the azimuthal and axial average ; and @xmath12 and @xmath13 are the velocity and magnetic fluctuations ) that is produced by the nonlinear interaction of the finite amplitude mode with itself , and is parallel to the mean magnetic field .
this is the well known @xmath14-effect , the correlated product of velocity and magnetic field fluctuations ( moffatt 1978 ) .
it is a fundamental dynamo mechanism .
this suggests that the mri may operate as a dynamo for the generation magnetic fields in accretion disks and other rotating systems .
this is consistent with earlier results indicating that the magnetic energy increases during the onset of mri turbulence ( hawley , gammie & balbus 1996 ) .
finally , when the mean flow profile is allowed to evolve , a single mri mode saturates by relaxation of the rotation profile , as expected .
this is also accompanied by the generation of a mean magnetic field .
thus , the mri dynamo mechanism ( the alpha effect ) may be robust .
our computational results are supported by an analytic theory in which the quasi - linear stresses and mean parallel electric field are calculated explicitly in terms of the linear eigenfunctions of the mri .
the paper is organized as follows .
previous work and the model are described in sections [ sec : pwrk ] and [ sec : model ] respectively .
nonlinear evolution of a single axisymmetric mri mode is presented in section [ sec : nl ] , for fixed keplerian flow ( section [ sec : nl1 ] ) and for evolving flow ( section [ sec : nl2 ] ) . using nonlinear single mode computations , the alpha effect for non - axisymmetric mri modes
is examined in section [ sec : asynl ] . in section [ sec : ql ] , the alpha effect for axisymmetric and non - axisymmetric mri is obtained using quasilinear calculations .
we summarize in section [ sec : summ ] .
in early work , the linear theory of the mri has been investigated carefully , and in the past decade nonlinear mhd simulations of the mri have been performed ( see , e. g. , balbus & hawley 1998 ; hawley & balbus 1999 and the references therein ) , including evolution of an mri - unstable system to a turbulent state .
an interesting feature of the turbulent state is that small - scale magnetic fluctuations arise , representing an increase in magnetic energy ( hawley , gammie & balbus 1996 ) .
turbulence generated by mri in global disk simulations is one of the current mri research topics ( armitage 1998 ; machida , hayashi & matsumoto 2000 ; hawley 2000 ; hawley 2001 ; machida & matsumoto 2003 ; reynolds & armitage 2001 ; armitage & reynolds 2003 ; mckinney & gammie 2002 ; nishikori , machida & matsumoto 2006 ; fromang & nelson 2006 ; lyra et al .
2008 ) .
recently the issue of numerical convergence has been raised in the simulations of the mri .
it was found that the results of some simulations are not independent of the numerical dissipation scale , and physical dissipations should be used ( fromang & papaloizou 2007 ) .
numerical simulations of the mri - induced turbulence in a shearing box with nonzero net vertical flux ( lesure & longaretti 2007 ) and zero net flux ( fromang , papaloizou , lesur & heinemann 2007 ) showed that the physical dissipation , the viscosity and the resistivity , affect the saturated level of mri - driven mhd turbulence , and thus the amount of angular momentum transport .
it was found that the turbulent activity is an increasing function of the magnetic prandtl number .
the dependence of the saturation level of the mri on gas pressure has been investigated by sano et al .
2004 using the local approximation .
based on series of local numerical simulations , a scaling law was also provided that describes the transport of angular momentum in shearing mhd boxes ( pessah , chan & psaltis 2007 ) .
in addition to numerical simulations , analytical methods were used to study some nonlinear aspects of the mri by goodman & xu 1994 .
goodman & xu introduced the notion of whether the mri can amplify very weak initial fields until they reach equipartition with the gas pressure .
they showed that mri axisymmetric modes are exact nonlinear solutions of the local incompressible mhd equations , suggesting that the mri perturbed magnetic field may grow to an amplitude many times larger than the initial unperturbed field .
their analytical calculations assumed a very special form for the axisymmetric mri mode , with a purely vertical wavevector whose field perturbations are purely horizontal ( perpendicular to the rotation axis ) . in this paper , we solve the full nonlinear mhd equations for a single mri mode with perturbations in all three directions ( radial , azimuthal and vertical ) .
we find that large - scale mean magnetic field is amplified through the alpha effect that causes the mode to saturate .
this effect was not studied by goodman & xu .
in fact , the assumption made by goodman & xu may have limited the possibility of the mri mode saturation through the alpha effect , which is the central result of our paper . in the limit of large amplification of mri perturbation ,
goodman & xu also showed that the mri mode could themselves be subject to additional parasitic instabilities ( such as the kelvin - helmholtz instability ) which may destroy the primary mri modes .
the growth rate of parasitic instabilities is proportional to the mri perturbation amplitude .
simulations of these instabilities would require multiple wavenumbers , and can not be captured in single mode ( single @xmath15 ) computations .
however , the saturation _ mechanism _ , a fundamental feature of mri , is not fully understood .
most of the previous studies on the mri saturation have dealt with what controls the saturated level of the mri in a turbulent state .
the most recent results show the effect of physical dissipation parameters on the saturation level using an unstratified shearing box model .
however , these studies have not fundamentally investigated the cause of the mode saturation ( lesure & longaretti 2007 ; fromang , papaloizou , lesur & heinemann 2007 ; pessah , chan & psaltis 2007 ; sano et al . 2004 ) . in this paper , we identify and explain the saturation mechanism of the mri at the level of a single mode . through the nonlinear evolution of a single mode ,
we simplify the problem to better identify the root of mri saturation . using this simple dynamical model
, we find that the instability itself generates a large - scale magnetic field ( mean magnetic field ) through the alpha effect that causes the mode to saturate .
we employ the mhd equations in doubly periodic @xmath16 cylindrical geometry for both the analytic ( quasi - linear ) and computational studies .
all variables are decomposed as @xmath17 , where @xmath18 is the _
@xmath19 component , and @xmath20 is the fluctuating component ( i.e. , all other fourier components with @xmath21 and @xmath22 ) . note that the mean component is a function of radius .
we consider an azimuthal equilibrium flow @xmath23 and vertical magnetic field @xmath24 .
( here , @xmath25 is the equilibrium value of the mean azimuthal flow @xmath26 , and @xmath27 is a constant . ) the single fluid mhd equations are , @xmath28 where the variables , @xmath29 , and @xmath30 are the density , pressure , velocity , magnetic field , current , ratio of the specific heats , and gravitational potential , respectively . time and radius are normalized to the resistive diffusion time @xmath31 and the outer radius @xmath32 , making the normalized outer radius unity .
velocity * v * and magnetic field * b * are normalized to the alfvn velocity @xmath33 , and the magnetic field on axis @xmath27 , respectively .
the parameter @xmath34 is the lundquist number ( where @xmath35 ) , and @xmath36 measures the ratio of characteristic viscosity @xmath37 to resistivity @xmath38 ( the magnetic prandtl number ) , where @xmath39 is the viscous diffusion time .
the factor @xmath40 is the beta normalized to the axis value .
the resistivity profile @xmath41 is uniform .
the boundary conditions in the radial direction are as is appropriate to dissipative mhd with a perfectly conducting boundary : the tangential electric field , the normal component of the magnetic field , and the normal component of the velocity vanish , and the tangential component of the velocity is the rotational velocity of the wall .
the azimuthal @xmath3 and axial @xmath42 directions are periodic .
we pose an initial value problem that consists of the equilibrium plus a perturbation of the form @xmath43 .
equations ( 1 - 6 ) are then integrated forward in time using the debs code .
the debs code uses a finite difference method with a staggered grid for radial discretization and pseudospectral method for azimuthal and vertical coordinates . in this decomposition
, each mode satisfies a separate equation of the form @xmath44 , where @xmath45 is a linear operator that depends on @xmath46 , and @xmath47 is a nonlinear term that represents the coupling of the mode @xmath48 to all other modes @xmath49 .
( this latter term is evaluated pseudospectrally . )
the time advance is a combination of the leapfrog and semi - implicit methods ( schnack et al .
1987 ) .
we distinguish between three types of initial value computations . in _ linear _ computations ,
the initial conditions consist of an equilibrium @xmath50 plus a single mode @xmath51 , as described above , with @xmath52 .
the initial amplitude @xmath53 is a polynomial in @xmath54 that satisfies the boundary conditions at @xmath55 and @xmath56 .
the initial amplitudes of all other modes are set to zero , and are constrained to remain zero for all time . only the mode @xmath48 is then evolved ; in particular , the equilibrium ( the @xmath57 , @xmath58 mode , or @xmath18 ) is _ not _ evolved , and
remains fixed in time .
the mode @xmath48 then satisfies the linear equation @xmath59 , where @xmath60 is independent of time ; an exponentially growing solution indicates linear instability , and the solution @xmath61 is the radial profile of the unstable eigenfunction . in _ nonlinear single mode _
computations , the problem is initialized in the same manner .
however , the @xmath57 , @xmath58 component ( the background ) is allowed to evolve self - consistently .
the term @xmath62 contains the nonlinear contribution from the mode @xmath48 interacting with itself , and the operator @xmath63 , which depends on @xmath64 , can evolve with time .
all other modes @xmath49 have and retain zero amplitude .
when the mode amplitude is small ( @xmath52 ) , the nonlinear term is negligible and the linear solution is again obtained .
as the unstable mode grows and @xmath65 becomes finite , the evolution of the background profile @xmath64 can modify the term @xmath66 and this can affect the evolution the mode @xmath48 .
if , as a result , the mode @xmath48 ceases to grow exponentially it is said to _ saturate_. ( in most cases , this means that the mode ceases continual growth altogether . )
we remark that this differs from _ quasi - linear _ saturation , because in the saturated state the radial profile of the finite amplitude solution @xmath61 can differ significantly from that of the linearly unstable eigenfunction . in
a _ fully nonlinear _
computation , all modes are intitialized with small random amplitude and are evolved in time , including the full nonlinear term @xmath47 . in this case , saturation can occur as a result of both background modification and coupling to other modes .
if there is sufficient numerical resolution , this can lead to a turbulent state . in this work we consider both linear and nonlinear single mode computations of the mri .
the latter begin from a state in which the plasma rotates azimuthally with a mean keplerian flow @xmath67 . the initial ( i.e. , at @xmath68 ) radial _ equilibrium _ force balance ( eq .
2 ) is satisfied by @xmath69 , where @xmath70 , and the initial pressure and density profiles are assumed to be radially uniform .
the azimuthal flow profile is the source of free energy for the mri .
we might expect that the mri would saturate by altering this profile .
however , in astrophysical settings the keplerian profile is fixed by gravitational constraints , so this saturation mechanism may not be available .
we investigate the saturation process in this case by performing computations in which the _ mean _ flow profile is fixed in time ( see section [ sec : nl1 ] ) . in these computations ,
the mean keplerian profile is maintained in time by an external force ; an ad - hoc force ( @xmath71 ) is added in the momentum equation to fix the mean flow profile . in other computations
we allow the mean flow profile to evolve in time ( see section [ sec : nl2 ] ) ; the initial keplerian flow consistent with the angular velocity at the inner and outer boundaries is allowed to relax self - consistently as the mode reaches finite amplitude .
this is the usual case of relaxation of the source of free energy for the mode . in both cases ,
all other mean variables ( e.g. , the mean magnetic field profile ) are allowed to evolve .
pressure and density are evolved ; however , they remain fairly uniform during the computations .
( note that _ perturbations _ to the azimuthal velocity @xmath72 are allowed to evolve . ) in all cases the mri is found to cease exponential growth , and the background profile @xmath50 is modified .
the saturation mechanism for a single mri mode is then investigated by performing _ linear _
computations in which the initial background state @xmath73 is the saturated background profile obtained from the nonlinear single mode calculations , @xmath50 .
we find that this new linear solution does _ not _ grow exponentially , indicating that , as a result of the nonlinear single mode evolution , the background state has become linearly stable
. in particular , this state has a modified mean magnetic field profile that has sufficient magnitude to stabilize the mode .
fully nonlinear calculations of mri saturation will be reported in a future publication .
we consider a cylindrical disk - shaped plasma with aspect ratio l/(@xmath74 ) ( where l is the vertical height , and @xmath75 and @xmath32 are the inner and outer radii , respectively ) .
the inner and outer radial boundaries are perfectly conducting , concentric cylinders that can rotate independently at specified rates .
periodic boundary conditions are used in the vertical and azimuthal directions .
the aspect ratio ( l/(@xmath74 ) ) used in the nonlinear computations is 1.3 .
the nonlinear computations are performed in a thick - disk approximation where vertical and radial distances are of the same order .
the range of parameters used in the computations is @xmath76 , @xmath77 , @xmath78 .
we present the results of single mode nonlinear computations with both fixed and evolving mean azimuthal velocity profiles for an axisymmetric mri mode ( i.e. , @xmath57 ) , as these modes have the largest linear growth rate ( see , e.g. , balbus & hawley 1998 ) . the fixed mean flow profile is relevant to the astrophysical setting ; evolving mean flow is the usual case of relaxation of the free energy source . in both cases
the initial condition consists of the equilibrium plus a perturbation with single vales of @xmath6 ( in this case @xmath57 ) and @xmath7 . in the first , we maintain a fixed _
mean _ profile @xmath79 , so as to isolate the effect of magnetic field growth on mode saturation .
in the second , the _ mean _ azimuthal flow profile is permitted to evolve in time .
( in both cases , _ perturbations _ to the azimuthal flow @xmath72 are permitted and evolve . )
we find that mean magnetic field growth plays an important role in the saturation of the mode , although the effect is somewhat stronger when the _ mean _ azimuthal flow is fixed in time .
we computationally solve the mhd equations ( section [ sec : model ] ) with initial conditions of keplerian flow ( @xmath80 ) and a weak vertical magnetic field @xmath81 .
we first , consider only axisymmetric modes ( @xmath57 ) , which are the fastest growing modes and will be treated analytically in section [ sec : ql ] . since the computations are single mode ( single @xmath6 and @xmath7 ) , the spatial resolution is only in the radial direction .
the number of radial grid points used is 250 .
the solutions are converged in timestep and radial spatial resolution .
we have performed nonlinear computations with different axial wave number and different lundquist numbers , @xmath82 , and with different equilibrium azimuthal flow magnitudes @xmath83 @xmath33 .
we find that the dynamics of mode saturation ( and field amplification ) is independent of axial wave number @xmath7 and flow amplitudes , and very similar for the three different lundquist numbers .
therefore , we only present nonlinear computation for the axisymmetric @xmath84 mode with @xmath85 , at @xmath86 and @xmath87 .
figure [ fig : fig1 ] shows the components of the magnetic modal energy during the nonlinear evolution when flow is fixed for @xmath86 and @xmath87 .
the axisymmetric mode grows with the growth rate @xmath88 and starts to saturate around t=0.016 ( fig .
[ fig : fig1](a ) ) . kinetic energy at t=0 is about two order of magnitude ( 50 times ) larger than magnetic energy ; however after saturation of the mode , total magnetic energy increases , and it becomes of the same order as the kinetic energy .
mean vertical magnetic field at t=0 is @xmath89 , and is shown with the dashed - dotted line in fig .
[ fig : fig2](a ) . as
the mode starts to saturate , the vertical magnetic field starts to increase as the result of the alpha effect @xmath90 .
figure [ fig : fig2](a ) shows the mean vertical magnetic field at time t=0.016 for @xmath86 ( around the time of mode saturation ) . at this time the vertical mean magnetic field in the core ( r=0.2 - 0.4 ) is increased by a factor of nine .
the azimuthal component of the alpha effect which causes the generation of the vertical magnetic field is large in the core ( fig .
[ fig : fig2](b ) ) . the mean vertical magnetic field and the azimuthal component of the alpha effect
are also shown in fig . [
fig : fig2 ] for @xmath87 ( dashed line ) around the time of mode saturation . similar radial structures and magnetic field amplification
are observed for @xmath87 .
the radial structure of the nonlinear alpha effect ( fig . [ fig : fig2](b ) ) is consistent with the quasilinear calculations ( section 6 , fig .
[ fig : fig1 ] ) and results in the amplification of magnetic field .
therefore , the nonlinear computation shows that the fluctuation - induced alpha effect amplifies the mean magnetic field in the core and causes the saturation of the mode .
we verify the stabilizing effect of the modified mean magnetic field profiles by performing linear computations of the saturated state , as described in section 3 .
these computations use the _ background _ profiles near the time of saturation . in this way we are able to calculate the linear growth rates of any resulting instabilities .
each linear computation is initialized with perturbation of a single fourier mode ( m=0 and k=6 ) and the equilibrium profiles .
for each linear computation , the equilibrium profiles are the mean ( m=0 and k=0 component ) magnetic and velocity fields taken from the single mode nonlinear computation at a specific time near the saturated state ( e. g. @xmath91 marked in fig .
[ fig : fig3](a ) ) .
we perform four separate linear computations for m=0 , k=6 mode which are initialized with four different equilibrium profiles .
four set of equilibrium profiles are the mean fields taken from the single mode nonlinear computation at four times during the saturation .
figure [ fig : fig3](a ) shows the radial magnetic energy vs. time for the nonlinear computation with fixed mean flow ; the four times at which the linear computations are performed are also marked .
the linear growth rates calculated from the four linear computations decrease as the mode saturates from @xmath92 to @xmath93 .
the linear growth rate decreases from 0.58 ( the value before the saturation at t=@xmath94 ) to 0.048 at @xmath93 . the mean vertical magnetic field at these times is also shown in fig .
[ fig : fig3](b ) .
as the mean vertical magnetic field amplifies from @xmath92 to @xmath93 by the alpha effect in the plasma core , the resulting mode growth rate obtained from the linear computation substantially decreases ( fig.3 ( c ) ) .
we find that the dynamics of the saturation and the resulting alpha effect is independent of axial wave number k. to illustrate this effect , we have performed nonlinear computation of a single mode with higher k ( k=30 ) shown in fig.3 ( d ) .
the mode with k=30 grows faster than k=6 with a growth rate @xmath95 and starts to saturate around t=0.007 .
the saturation level for this mode is lower than k=6 mode .
similarly , we initialize and perform linear computations with the saturated profiles from nonlinear computations at times @xmath96 @xmath97 . the saturated mean vertical magnetic fields which are used as the equilibrium profiles for the linear computations
are also shown in fig .
three separate linear computations are initialized at times @xmath96@xmath97 shown in fig .
because the linear computations have been started with the saturated mean fields , the linear growth rates obtained from these three computations are zero or negative , as seen in fig .
the linear stability computations thus show that the amplification of the mean vertical magnetic field by the azimuthal component of the alpha effect @xmath90 suppresses the mode growth and thus causes the mode saturation .
the @xmath14-effect of the saturated mri therefore appears to be independent of the axial mode number @xmath7 . in the computation with a fixed mean azimuthal flow profile ( section 4.1 ) ,
the mean magnetic field profile could be modified by the fluctuation - induced alpha effect . in the case when mean flow is allowed to evolve , the mean azimuthal flow profile will also be modified by the mhd stresses .
figure [ fig : fig4 ] shows the nonlinear evolution of m=0 , @xmath7=6 mode for evolving flow .
similar to the computation with fixed flow ( fig .
[ fig : fig1 ] ) , axisymmetric mode grows and starts to saturate around t=0.016 .
figure [ fig : fig5 ] shows the evolution of magnetic field and flow during the nonlinear computations at times @xmath98 ( initial ) , @xmath99 ( near saturation ) and @xmath97 ( final saturated ) . as is seen , the mean flow gradient is reduced in the core during the saturated state .
the initial keplerian mean flow profile flattens in the core ( fig . [
fig : fig5 ] ( b ) ) .
thus , with smaller mean flow gradient ( less free energy for the mode growth ) , the mode starts to saturate .
in addition to the mean flow modification , magnetic field generation still occurs .
the magnetic field amplifies by a factor five in the core ( fig.[fig : fig5](c ) ) as the azimuthal component of the alpha effect ( fig.[fig : fig5](d ) ) increases .
since the mean flow is allowed to evolve , the amplification of the magnetic field is smaller than the computation with fixed mean flow . the alteration in mean flow and mean magnetic field profiles both contribute to mode saturation .
the reynolds and maxwell stresses responsible for the momentum transport that flattens the flow profile are shown in fig .
[ fig : fig6](a ) .
both stress terms contribute to the flow relaxation and outward momentum transport ( consistent with eqs .
[ eq : maxwell ] , [ eq : reynold ] in section 6 ) .
the ratio of the maxwell and reynolds stresses depends upon the magnetic prandtl number .
[ fig : fig6](b ) shows the profiles of maxwell and reynolds stresses for the same computations but with pm=0.1 ( final saturated state ) .
it is seen that the reynolds stress is larger than the maxwell stress in the core for computation with lower prandtl number .
final relaxed flow profiles for these computation with pm=1 and pm=0.1 is slightly different in the core ( fig .
[ fig : fig7 ] ) .
finally , we investigate the generation of magnetic field by the alpha effect from non - axisymmetric mri modes . nonlinear computation for an mri mode with @xmath100
is performed with fixed keplerian mean flow profile with the same equilibrium and parameters as the @xmath57 mode .
the non - axisymmetric m=1 mode grows more slowly than the @xmath57 mode , but saturates at the same level ( fig .
[ fig : fig8 ] ) .
we find that the saturation mechanism for the m=1 mode is very similar to that of the m=0 mode .
however for the @xmath100 mode , the vertical component of the alpha effect , @xmath101 , is also nonzero and becomes important in the process of mode saturation .
the vertical component of the alpha effect will be calculated in the next section ( eq . [ eq : vxbt ] , [ eq : vxbz ] ) and shown to be nonzero for the non - axisymmetric mri modes .
figure [ fig : fig8](b ) shows the radial structure of both vertical and azimuthal components of the alpha effect for the m=1 mode during the linear growth . because of
the nonzero vertical alpha effect , an azimuthal magnetic field @xmath102 ( which is zero at t=0 ) is generated and is shown in fig .
[ fig : fig9 ] .
thus , the saturation of m=1 mode occurs through amplification of both vertical and azimuthal magnetic field through the alpha effect .
to evaluate the influence of the instability ( fluctuations ) on the mean magnetic field , we calculate the alpha effect the mean fluctuation - induced electromotive force @xmath103 , where @xmath104 and @xmath105 are the velocity and magnetic fluctuations , and @xmath106 denotes an average over the azimuthal and vertical directions .
perturbed quantities are assumed in the form of @xmath107 .
the alpha effect is obtained for a single mri fourier mode ( single m and @xmath7 ) .
the linearized momentum equations in terms of the lagrangian displacement @xmath108 are ( chandrasekhar 1961 ) , @xmath109 where @xmath110 , @xmath111 , @xmath112 and @xmath113 is angular velocity .
the velocity fluctuations are related to the lagrangian displacement @xmath114 , @xmath115 , @xmath116 \xi_r $ ] and @xmath117 . in an ideal plasma ( zero resistivity )
magnetic field perturbations are also directly related to the displacement , @xmath118 .
using incompressibility ( @xmath119 ) and eqs .
[ eq : mri1]-[eq : mri3 ] , azimuthal and vertical displacements are obtained in terms of @xmath120 , @xmath121 \label{eq : xiphi}\ ] ] @xmath122 \label{eq : xiphi}\ ] ] we then construct the quasilinear alpha effect , @xmath123 and @xmath124 , in terms of the radial displacement @xmath120 , @xmath125 @xmath126 where , @xmath127 for axisymmetric m=0 modes , the alpha effect reduces to a simple form . the vertical component of the alpha effect ( eq . [ eq : vxbz ] ) vanishes . the azimuthal component ( eq .
[ eq : vxbt ] ) is nonzero and becomes , + @xmath128 thus , the instability is accompanied by an alpha effect that generates magnetic field . to obtain the radial structure of the alpha effect , linear computation ( eqs . 1 - 6 )
is performed for axisymmetric modes .
we construct the alpha effect using linear eigenfunctions obtained from single mode computation .
figure [ fig : fig10](a ) ( solid line ) shows the azimuthal component of the alpha effect for the m=0 mode with @xmath129 .
the azimuthal component generates axial magnetic field in the inner region of the disk , a stabilizing influence . at higher k values ( fig .
[ fig : fig10](a ) , dashed line ) the mode structure and alpha effect becomes more localized radially .
linear computations are also performed for different lundquist numbers , @xmath130 ( fig .
[ fig : fig10](b ) ) .
the growth rate of the mri mode does not change significantly with lundquist number ( since mri is an ideal mode ) .
it is seen that the structure is slightly broader for lower s , which is expected because of higher dissipation at lower s. similar radial mode structure ( @xmath131 ) with peaked structure near the inner radius has also been obtained in earlier work by curry , pudritz & sutherland 1994 and blokland et al .
we can similarly evaluate the fluid stresses that alter the flow .
the mhd stresses for axisymmetric mri mode are given by balbus and hawley 2002 .
here we also calculate the quasilinear stresses for a non - axisymmetric mri mode .
the nonlinear evolution of this mode was examined in section [ sec : asynl ] .
using eqs [ eq : mri1]-[eq : mri3 ] , the maxwell stresses ( for both axisymmetric and non - axisymmetric modes ) are calculated in terms of radial displacement @xmath132 @xmath133 it is seen that the vertical maxwell stress is zero for the axisymmetric modes and the azimuthal lorentz force @xmath134 is obtained @xmath135 @xmath136 similarly , the reynolds stresses are , @xmath137 @xmath138 again , the only nonzero component for m=0 modes is azimuthal , and @xmath139 is calculated to be , @xmath140 where the coefficient d is defined in eq .
[ eq : diff ] . adding eqns [ eq : maxwell ] and [ eq : reynold ] , the total azimuthal fluid force for m=0 modes becomes , @xmath141 the fluid force consists of a simple diffusion - like part ( the first term ) and a more complex part ( the second term ) that also depends on flow nonuniformity ( @xmath142 ) .
in summary , we have shown that the mri will generate large - scale magnetic field as a mechanism for saturation .
computation of the fully evolved , nonlinear steady state of a single mode establishes that the mechanism that determines mode saturation is as described by quasilinear theory , and that magnetic field generation persists .
this is an alpha effect , and can be understood through the stabilizing effect of the magnetic field .
the analytic and computational work presented here treats one mode , as a means to clarify the effect .
the influence of multiple modes and mode coupling will be of considerable importance . in a future work
we will investigated the mri saturation via the alpha effect using the full nonlinear computations with multiple modes .
the field generated is large - scale in that it is symmetric in toroidal and axial directions .
however , it varies radially such that the total toroidal and axial magnetic flux within the plasma does not change significantly in time .
thus , the examples studied here would not be defined as a `` large - scale dynamo . ''
future work also will investigate whether there are geometric variations for which the generation of large - scale field would be accompanied by the generation of total magnetic flux .
+ we acknowledge useful discussions with ellen zweibel .
we also thank the referee for useful comments .
this work is supported by national science foundation .
30 armitage , p. j. 1998 , , 501 , l189 armitage , p. j. , & reynolds , c. s. 2003 , mon . not .
r. astron .
soc . , 341 , 1041 balbus , s. a , & hawley , j. f. 1991 , , 376 , 214 balbus , s. a , & hawley , j. f. 1998 , rev . mod .
, 70,1 balbus , s. a , & hawley , j. f. 2002 , , 573 , 749 blockland , j. w. s. , swaluw , e. , keppens , r. , & goedbloed , j. p. 2005 , astronomy & astrophysics , 444 , 337 chandrasekhar , s. 1961 , hydrodynamic and hydromagnetic stability ( oxford , new york ) .
curry , c. , pudritz , r. e. , & sutherland , p. g. 1994 , , 434 , 206 fromang , s. , & nelson , r. p. 2006 , astronomy & astrophysics , 457 , 343 fromang , s. , & papaloizou , j. 2007 , astronomy & astrophysics , 476 , 1113 fromang , s. , papaloizou , j. , lesur , g. , & heinemann , t. 2007 , astronomy & astrophysics , 476 , 1123 goodman , j. , & xu , g. 1994 , , 432 , 213 hawley , j. f. , & balbus , s. a , 1999 , phys . of plasmas , 6 , 4444 hawley , j. f. , gammie , c. f. , & balbus , s. a , 1996 , apj , 464 , 690 hawley , j. f. 2000 , , 528 , 462 hawley , j. f. 2001 , , 554 , 534 lesur , g. , & longaretti , p .- y . , 2007 mon . not .
r. astron .
, 378 , 1471 lyra , w. , johansen , a. , klahr , h. , & piskunov , n. 2008 , astronomy & astrophysics , 479 , 883 machida , m. , & matsumoto , r. 2003 , apj , 585 , 429 machida , m. , hayashi , m. r. & matsumoto , r. 2000 , , 532 , l67 mckinney , j. c. , gammie , c. f. 2002 , , 573 , 728 moffatt , h. k. 1978 , _ magnetic field generation in electrically conducting fluids _ ( cambridge university press , cambridge ) .
nishikori , h. , machida , m. , & matsumoto , r. 2006 , apj , 641 , 862 pessah , m. e. , chan , c. , & psaltis , d. 2007 , , 668 , l51 reynolds , c. s , & , armitage , p. j. 2001 , , 561 , l81 sano , t. , inutsuka , s. i. , turner , n. j , & stone , j. m. 2004 , apj , 605 , 321 schnack , d. d. , et al .
1987 , j. comput .
phys . , 70 , 330 shakura , n. i. , & sunyaev , r. a. 1973 , astronomy & astrophysics , 24 , 337 velikhov , e. p. 1959
soviet phys . ,
jetp , 36 , 995 ) calculated from the linear m=0 , k=6 computations .
( d ) radial magnetic energy during the nonlinear evolution of mri mode ( m=0 , k=30 ) with fixed mean flow . three linear computations ( at @xmath96@xmath97 )
show no growth for k=30 mode .
( e ) mean vertical magnetic field profiles for @xmath96@xmath97 times during the nonlinear saturation used in the linear computations ( m=0 , k=30 ) .
] vs time for computation when mean flow profile is fixed for m=1 ( solid line ) and m=0 ( dashed line ) ( @xmath145 ) .
( b ) azimuthal and vertical components of the alpha effect during the linear growth of m=1 mode . ] for the m=0 mr mode ( a ) for two wave numbers @xmath146 ( @xmath87 ) , ( b ) for @xmath147 with @xmath129 .
@xmath148 , @xmath149 .
the growth rates are @xmath150 and @xmath151 for @xmath84 and @xmath152 respectively ( @xmath153 ) . ] | the saturation mechanism of magneto - rotational instability ( mri ) is examined through analytical quasilinear theory and through nonlinear computation of a single mode in a rotating disk .
we find that large - scale magnetic field is generated through the alpha effect ( the correlated product of velocity and magnetic field fluctuations ) and causes the mri mode to saturate .
if the large - scale plasma flow is allowed to evolve , the mode can also saturate through its flow relaxation . in astrophysical plasmas , for which the flow can not relax because of gravitational constraints , the mode saturates through field generation only . |
accretion of gas onto a super massive black hole ( smbh ) in the nucleus of galaxies is believed to be the source of activity in quasars and seyfert galaxies ( commonly known as active galactic nuclei ( agns ) ; cf .
rees 1984 ) .
several studies have suggested that the mass of the smbh in these objects is correlated with the luminosity , mass and velocity dispersion of the stellar spheroid of the galaxies ( kormendy & richstone 1995 ; magorrian et al .
1998 ; ferrarese & merritt 2000 ; gebhardt et al . 2000 ; marconi & hunt 2003 ; hring & rix 2004 ) . such correlations may imply an evolutionary relation between the growth of the smbh and the host galaxy itself ( e.g. somerville et al .
2008 ; shankar et al . 2009
; hopkins & hernquist 2009 ) . in order to study the dependence of the various observed phenomena of agns on the black hole mass and the cosmic evolution of the black holes , independent and reliable estimates of the mass of the black holes are required ( e.g. , goulding et al .
2010 ; rafter , crenshaw & wiita 2009 ) . one independent method to estimate the mass of the black hole is using the reverberation mapping technique ( blandford & mckee 1982 ; peterson 1993 ) . in the optical bands , the continuum flux of some agns ,
is known to vary on timescales as short as hours ( e.g. , miller , carini & goodrich 1989 ; stalin et al . 2004 ) .
if the main source of ionization of the broad line region ( blr ) is the continuum itself , any variation of the continuum emission can also be seen in the broad emission lines . however , the variations in the broad line flux will have a time lag ( @xmath6 ) relative to the continuum variations , which can be interpreted as the light travel time across the blr . as a first approximation , therefore , the size of the blr is @xmath7 , where @xmath8 is the velocity of light .
once the @xmath9 is obtained , the mass of the black hole can also be estimated , using the velocity dispersion of the broad component of the emission lines , @xmath10 , and assuming virial equilibrium ( peterson et al . 2004 ; p04 ) ; see peterson 2010 , for a recent review ) .
the reverberation mapping technique has been used to make estimates of smbh masses over a large range of redshift . however , because the technique is observationally taxing , as it demands an enormous amount of telescope time , to date the blr radius of only about three dozen agns ( seyfert 1 galaxies and quasars ) have been determined ( p04 ; kaspi et al . 2007 ; bentz et al . 2009a ; denney et al .
2009 , 2010 ) .
nevertheless , using these estimates a correlation was found between @xmath9 and the optical continuum luminosity at 5100 ( kaspi et al .
2000 ; kaspi et al .
2007 ; p04 ; denney et al . 2009 ; bentz et al .
the r@xmath11@xmath12@xmath13l@xmath14 relation can be considered well constrained between the luminosities 10@xmath15 erg sec@xmath16l@xmath17 erg sec@xmath18 . on the other hand , for luminosities below 10@xmath15 erg sec@xmath18 ,
only a handful of sources are observed , and the estimated values of @xmath9 could also indicate a flattening of the relation ( see fig . 2 of kaspi et al .
this flattening would suggest a lower limit in the possible masses of smbhs in galaxies .
although recent revisions of a few sources made by bentz et al .
( 2006 ) and denney et al .
( 2009;2010 ) are consistent with a continuation of the @xmath19l@xmath14 relation to lower luminosities , and consequently with no lower limit in the mass for the smbh , the correlation is still sparsely sampled . moreover
, the @xmath19l@xmath14 relation is very useful for estimating the smbh masses from single - epoch spectra and calibrating other surrogate relations used for black hole mass estimates ( vestergaard 2004 ; shen et al .
therefore , estimates of @xmath9 for a larger number of sources are required .
the extrapolation of the known @xmath19l@xmath14 relation to low luminosities suggests that the time lag between the variations of the broad line and that of the continuum will be of the order of hours to days , as compared to several months for high luminosity sources .
thus , monitoring programs of short durations , but fast sampling , are required to estimate the reverberation time lags for low luminosity sources .
in this paper , we present the optical spectroscopic and photometric observations of a new low luminosity agn , the x - ray source and seyfert 1.5 galaxy h 0507 + 164 . based on a reverberation mapping campaign that lasted for about a month , during november - december 2007 ,
we have obtained @xmath9 and estimated the mass of the smbh . in section 2 ,
the observations and data reductions are described .
the results of the analysis are given in section 3 , and the conclusions are presented in section 4 .
using the vron - cetty & vron catalogue of quasars and active galactic nuclei ( 12th ed . ;
vron - cetty & vron 2006 ) , we have compiled a list of nearby seyfert 1 galaxies , which , based on the available spectra , have a luminosity at @xmath20 of the order of 10@xmath21 erg sec@xmath18 or lower .
very few candidates were found ( mostly because of the absence of available spectra ) .
the source , h 0507 + 164 , that we selected for our campaign is identified in the catalogue of vron - cetty & vron as an x - ray source , with coordinates @xmath22 , and is classified as a seyfert 1.5 galaxy at a redshift of @xmath0 .
optical spectroscopic and photometric observations of h 0507 + 164 were carried out in 2007 between 21 of november and 26 of december at the 2 m himalayan chandra telescope ( hct ) , operated by the indian institute of astrophysics , bangalore .
the telescope is equipped with a @xmath23 ccd , coupled to the himalayan faint object spectrograph and camera ( hfosc ) . in imaging mode , only the central @xmath24 pixels region of the ccd is used .
the camera has a plate scale of @xmath25 arcsecond / pixel , which yields a field of view of @xmath26 square arcmin .
medium resolution spectra of the nucleus were obtained using a 11 arcmin @xmath27 1.92 arcsec wide slit and a grism .
the spectra have a spectral range of 3800@xmath126700 with a resolution of @xmath288 .
the exposure time varied between 900 and 1000 seconds .
the spectra were reduced using standard procedures in iraf .
after bias subtraction and flat fielding , one dimensional spectra were extracted and calibrated , in wavelength using an fear lamp , and in flux using various observations of the spectrophotometric standard star feige 34 .
since the observed spectra were of low s / n , for further analysis all the spectra were smoothed to a resolution of @xmath2815 .
the standard technique of spectral flux calibration is not sufficiently precise to study the variability of agns .
since even under good photometric conditions the accuracy of spectrophotometry is not better than 10% ( shapovalova et al .
2008 ) , we used a relative calibration procedure .
a first order flux calibration was first obtained in the normal way using the standard star .
then all the spectra were inter - calibrated relative to the spectra of one night ( we choose the 21 of november ) , assuming the flux of the narrow line [ o iii]@xmath29 is constant .
this is justified , because the narrow line region ( nlr ) is much more extended ( of the order of a few hundred parsecs ) than the blr ( much less than a parsec ) and flux variation can not be observed in this region over short time scales ( cf .
osterbrock 1989 ) .
each spectra were scaled relative to the reference spectra using the scaling algorithm devised by van groningen & wanders ( 1992 ) .
this algorithm uses a chi - square , to minimize the residuals of the [ o iii]@xmath135007 line after subtraction from the reference spectrum . the mean spectrum , averaging the 22 nights of observations , is shown in fig . [ fig : meanspec ] . in parallel to the spectroscopic observations ,
r - band images were also obtained with the hfosc .
unfortunately , some observations turned out to be contaminated by the light of the extremely bright stars located near the source . as a consequence ,
our r - band photometry covers only 13 of the 22 nights of the campaign .
the images were bias subtracted and flat fielded using iraf packages . for the rest of the reduction , calibration and analysis , packages in midas were used . after removing the cosmic rays , profile fitting photometry
was done using the daophot and allstar packages .
the observed r - band frame is shown in fig .
[ fig : image ] .
north is up and east is to the left .
note the presence of an extremely bright star near h 0507 + 164 .
[ fig : image ] , width=377,height=264 ] .continuum and @xmath30 fluxes for the source h 0507 + 164 [ table : fluxes ] [ cols=">,^,^ , > " , ]
the lightcurves of the h@xmath2 flux and the continuum at 5100 were obtained using the final inter - calibrated spectra .
the continuum flux in the rest - frame of the galaxy at 5100 was obtained using the mean flux within the observed band from 5172 to 5200 .
the h@xmath2 emission line fluxes were obtained by integrating the emission profile in the band spanning 4884@xmath125012 , after subtracting a continuum .
an average of the mean fluxes in the regions on the blue ( 4808@xmath124852 ) and red ( 5012@xmath125024 ) sides of the h@xmath2 line was used as the continuum below the line . although the measured line fluxes include both the narrow and broad components , any variation observed in the line fluxes can be attributed to the broad component only , since the narrow component is not expected to vary during the period of our observations .
the lightcurves for the continuum at 5100 , for the h@xmath2 and for the [ o iii]@xmath135007 lines are shown in fig .
[ fig : lightcurve ] .
the corresponding fluxes for the continuum and h@xmath2 are listed in table [ table : fluxes ] . as expected from the inter - calibration procedure ,
the lightcurve for [ o iii]@xmath135007 is nearly constant . on the other hand , both
the continuum at 5100 and the h@xmath2 flux are observed to vary .
@xmath135007 , h@xmath2 and continuum at 5100 are plotted .
the fluxes are in units of 10@xmath31 erg s@xmath18 @xmath32 @xmath18 for the [ o iii]@xmath135007 line , 10@xmath33 erg s@xmath18 @xmath32 @xmath18 for the h@xmath34 line and 10@xmath35 erg s@xmath18 @xmath32 @xmath18 for the continuum .
[ fig : lightcurve ] , width=302 ] the observed r - band differential instrumental magnitudes between the galaxy and the comparison star ( marked in fig .
[ fig : lightcurve ] ) are also given in table [ table : fluxes ] .
the r - band differential lightcurve plotted in fig .
[ fig : lightcurve ] ( top panel ) closely follows the lightcurves of the continuum at 5100 and h@xmath2 . the average flux at 5100 is @xmath36 erg @xmath32s@xmath18@xmath18 , which corresponds to @xmath13l@xmath14 of @xmath37 erg s@xmath18 , for the cosmological parameters h@xmath38 km s@xmath18mpc@xmath18 , @xmath39 and @xmath40 .
the variability of the light curves are characterised by the parameters , excess variance , f@xmath41 and the ratio between the maximum and minimum flux of the light curves , r@xmath42 ( rodriguez - pascual et al .
1997 ; edelson et al .
the continuum and h@xmath2 line have f@xmath41 of 0.16 and 0.18 and r@xmath42 of [email protected] and [email protected] respectively .
these values are within the range of values found by other variability studies of agns ( cf .
p04 ; denney et al .
2010 ) .
the time lag between the variations of the continuum flux and the variations of the h@xmath2 emission can be determined by cross - correlating the two light curves . for cross correlation analysis , the method of interpolated cross - correlation function ( iccf ; gaskell & sparke 1986 ;
gaskell & peterson 1987 ) and the method of discrete correlation function ( dcf ; edelson & korlik 1998 ) were used .
although both iccf and dcf methods produce similar results ( white & peterson 1994 ) , the interpolation of the light curve during the period of gaps required by the iccf method might not be a reasonable approximation to the behavior of the light curves .
thus the dcf method is preferable for data with large gaps ( denney et al . 2009 ) . for comparison , we obtain the results using both the methods . the results of the cross correlation analysis are shown in fig .
[ fig : ccf ] .
the cross correlation function ( ccf ) obtained using the iccf method is plotted as a thick solid line . the auto correlation functions ( acfs ) of the continuum at 5100 and the @xmath30 line
are also shown in fig .
[ fig : ccf ] as dashed and dotted lines respectively . for comparison
the cross correlation function obtained using the dcf method is also plotted as a thin solid line . as expected , the auto correlations have zero time lags . on the other hand ,
the time lag in the cross correlation curve is clearly noticeable as an overall shift to the right .
the position of the maximum in the ccf provides an estimate of the time lag between the continuum and the h@xmath34 line .
the maximum was however determined using the centroid , which gives a better estimate for noisy ccfs , rather than the peak , using the formula : @xmath44 the estimate of the centroid includes all the points that are within 50% of the peak value of the ccf . based on this cross correlation analysis , a statistically significant centroid and the associated uncertainty
were obtained using a bootstrap technique that introduces effects of randomness in fluxes and sampling of the light curve ( cf .
p04 ) . a method to carry out monte - carlo ( mc ) simulation using the combined effects of flux randomization ( fr ) and the random subset selection ( rss ) procedures
is described in peterson et al .
additional improvements as suggested by welsh ( 1999 ) are summarised in p04 , which we use for our analysis .
lightcurves , as obtained using the iccf method , is plotted as as a thick solid line .
the auto correlation functions of the continuum at 5100 and the @xmath30 line are plotted as dashed and dotted lines respectively .
the cross correlation function obtained using the dcf method is plotted as a thin solid line .
, width=302,height=302 ] first an rss procedure was applied by randomly selecting 22 observations from the light curve .
the flux uncertainties of the multiply selected observations were weighted according to welsh ( 1999 ) .
this light curve was given as input to the fr procedure , where each measured value of the fluxes are modified by adding the measured flux uncertainties multiplied with a random gaussian value .
the modified light curves were then cross - correlated and the centroids were determined as outlined above using ccf values above 50% of the peak value .
this procedure was repeated for @xmath45 times , retaining only those ccfs whose maximum cross - correlation coefficient is large enough such that the correlation is significant with a confidence level of 95% or larger . a cross - correlation centroid distribution ( cccd )
was built using the above centroids and is shown in fig .
[ fig : ccfdist ] .
the average value of cccd was taken to be @xmath46 . since the cccd is non - gaussian ( cf .
peterson et al .
1998 ) , the upper and lower uncertainties in @xmath46 were determined such that 15.87% of cccd realizations have @xmath47 and 15.87% realizations have @xmath48 .
this error in @xmath46 corresponds to @xmath49 errors for a gaussian distribution .
lccccc & + method & & 2 & 2.5 & 3 & 3.5 & 4 + iccf & 3.14@xmath50 & 3.28@xmath51 & 3.27@xmath52 & 3.06@xmath53 & 3.05@xmath54 + + dcf & 3.91@xmath55 & 4.30@xmath56 & 4.19@xmath57 & 4.11@xmath58 & 4.03@xmath59 + the centroid time lags obtained using different cross correlation methods for different bin sizes are given in table [ table : centroid ] .
the variations due to different bin sizes are within the error bars .
the dcf method gives a mean time lag of about 4 days , whereas the iccf method gives a mean value of about 3 days .
this difference is of the order of the estimated errors on the time lags . considering these uncertainties ,
both the dcf and iccf methods give time delays that are consistent with each other .
this suggests that the estimated time lag is not a spurious result for the sampling of the light curves presented here , and the results of the iccf methods are reliable . to be conservative ,
for further calculations we use the time lag with the largest scatter corresponding to a bin size of 3.5 days , obtained using the iccf method . based on this analysis the average observed frame time lag between the h@xmath60 and the @xmath135100 continuum light - curves
was found to be @xmath61 days . after correcting for the time dilation effects using the redshift of the source , we found a time lag of @xmath62 days in the rest frame of the source .
the wavelength coverage of our observations , also include the h@xmath63 line , and a time lag using the h@xmath63 line can also be estimated
. however , by repeating the analysis procedure presented here , a reliable time lag using h@xmath63 line could not be found , because the correlation curves are too noisy .
this may be due to the shorter duration of the observations and the relative flux calibration procedure using the [ o iii ] @xmath29 line situated much farther apart in wavelength from h@xmath63 being unreliable ( grier et al . 2008 ) .
unfortunately , the nearby doublet [ s ii ] @xmath64 6716,6731 narrow lines , which could be used for the relative calibration of h@xmath63 , are too weak .
thus we do not estimate the time lag using the h@xmath63 line .
, width=302,height=302 ] in order to relate the time lag to the mass of the black hole , an estimate of the line width , the line dispersion @xmath10 , of the broad emission component of h@xmath2 , is required .
following p04 , it is relatively straight forward and more practical to measure @xmath10 , the second moment of the profile , directly from the root mean square ( rms ) spectrum . indeed , in the rms spectrum the constant components , or those that vary on timescales much longer than the duration of the observation vanish , thus largely obviating the problem of de - blending the lines . to obtain the rms spectrum ,
all the observed spectra were combined using the formula : @xmath65}^2 \right\}^{1 \over 2}\ ] ] in fig .
[ fig : rmsspec ] , we show the root mean square ( rms ) spectrum .
it can be seen that the two narrow [ o iii ] lines have almost completely disappeared in the rms spectrum .
the mean value of @xmath10 corrected for the instrumental response of the spectrograph and the associated uncertainty were obtained following the bootstrap method described in p04 . from our observed 22 observed spectra
, we randomly selected 22 spectra , irrespective of whether a particular spectrum has already been selected or not . since some of the spectra were selected multiple times , the mean value of the resultant number of spectra were smaller by 8 .
these randomly selected spectra were then used to construct an rms spectra from which @xmath10 was measured and corrected for the instrumental resolution of the spectrograph .
this procedure was repeated 10000 times and the mean and standard deviation of these realisations are taken as @xmath10 and its uncertainty respectively .
a distribution of @xmath10 values obtained from the bootstrap method is also shown in fig [ fig : sigmadist ] .
we thus estimate a line dispersion of @xmath66 km sec@xmath18 . using the rest frame time delay , the radius of the blr is estimated to be @xmath67 pc . in fig .
[ fig : rlum ] we show the measurement of @xmath9 and @xmath13l@xmath14 luminosity of the source presented here along with the most updated data set given by bentz et al .
( 2009b ) and the additional source , mrk 290 , given by denney et al .
the solid line is the relation obtained by bentz et al .
( 2009b ) . from this figure
it can be seen that our h@xmath60 measurement lags are in agreement with the known @xmath68 relationship .
the mass of the black hole was estimated using the formula in p04 : @xmath69 where @xmath70 is the width of the line and g is the gravitational constant .
the parameter @xmath71 is a scaling factor , which takes into account the geometry and kinematics of the blr .
onken et al . (
2004 ) found an empirical value of @xmath72 , using a sample of agns having both reverberation based black hole masses and host galaxy bulge velocity dispersion ( @xmath73 ) estimates .
this value relies on the assumption that both agns and quiescent galaxies follow the same m@xmath74-@xmath73 relationship ( ferrarese & merritt 2000 ; gebhardt et al .
2000 ) . for this particular
scaling , the appropriate velocity width @xmath70 is the line dispersion in the rms spectrum @xmath10 ( bentz et al .
2008 ) . adopting the onken et al .
( 2004 ) scaling factor and the @xmath10 measured from our observations , we estimated the mass of the smbh in h 0507 + 164 to be @xmath75 m@xmath5 . using the bootstrap method described in the text .
[ fig : sigmadist ] , width=302,height=302 ] , width=355,height=264 ]
we present for the first time monitoring observations of the x - ray source and sy 1.5 galaxy h 0507 + 164 , spanning a time period of about one month .
we have obtained 22 nights of spectra during this period , with a mean sampling time of about 1.6 days .
we measured an observed frame time lag of about @xmath61 days between the changes in the h@xmath2 emission line flux and the changes in the continuum flux at 5100 . after correcting for the redshift
, we find a corresponding time lag of @xmath62 days in the rest frame of the source . from this measured time lag
we deduced a size for the blr of @xmath67 parsec and estimated a black hole mass of @xmath76 m@xmath5 .
our estimate of r@xmath11 using the measured lag of h@xmath60is in agreement with the r@xmath11@xmath12@xmath13l@xmath14 relationship shown by bentz et al .
( 2009b ) .
we thank the anonymous referee for his / her valuable comments that helped to improve the presentation significantly .
we also thank b. m. peterson and r. w. pogge for kindly providing us with the relative flux scaling program . the support provided by the staff at the indian astronomical observatory , hanle and crest , hoskote is also acknowledged .
99 bentz m.c .
2006 , apj , 651 , 775 bentz m.c .
2008 , apj , 689 , l21 bentz m.c .
. 2009a , apj , 705 , 199 bentz m.c .
2009b , apj , 697 , 160 blandford r. d. , mckee , c. f. 1982 , apj , 255 , 419 denney k. d. et al .
2009 , apj , 702 , 1353 denney k. d. et al .
2010 , apj , 721 , 715 edelson r.a .
, krolik j.h .
1988 , apj , 333 , 646 edelson r. , turner t. j. , pounds k. , vaughan s. , markowitz a. , marshall , h. , dobbie p. , & warwick , r. 2002 , apj , 568 , 610 ferrarese l. , merritt d. 2000 , apj , 539 , l9 gaskell c. m. , sparke l. s. , 1986 , apj , 305 , 175 gaskell c. m. , peterson b.m .
, 1987 , apjs , 65 , 1 gebhardt k. et al .
2000 , apj , 539 , l13 goulding a. d. , alexander d. m. , lehmer b. d. , mullaney j. r. 2010 , mnras , 406 , 597 grier c. j. , et al .
2008 , apj , 688 , 837 hring n. , rix h .- w .
2004 , apj , 604 , l89 hopkins p. f. , hernquist l. 2009 , apj , 694 , 599 kaspi s. , smith p.s . , netzer h. , maoz d. , jannuzi b.t . , giveon u. 2000 , apj , 533 , 631 kaspi s. , maoz d. , netzer h. , peterson b.m . , vestergaard m. , jannuzi b.t .
2005 , apj , 629 , 61 kaspi s. , brandt w. n. , maoz d. , netzer h. , schneider d. p. , shemmer o. 2007 , apj , 659 , 997 kormendy j. , richstone d. 1995 , ara&a , 33 , 581 magorrian j. et al .
1998 , aj , 115 , 2285 marconi a. , hunt l. k. 2003 , apj , 589 , l21 miller h.r .
, carini m.t . , goodrich b.d .
1989 , nature , 337 , 627 onken c.a . , ferrarese l. , merritt d. , peterson b.m . , pogge r.w . , vestergaard m. , wandel a. 2004 , apj , 615 , 645 osterbrock d. e 1989 , in astrophysics of gaseous nebulae and active galactic nuclei , university science book peterson b. m. , wanders i. , horne k. , collier s. , alexander t. , kaspi s. , maoz d. , 1998 , pasp , 110 , 660 peterson b. m. 1993 , pasp , 105 , 247 peterson b. m. , et al .
2004 , apj , 613 , 682 peterson b.m . , 2010 , iau symposium , 267 , 151 rafter s. e. , crenshaw d. m. , wiita p. j. 2009 , aj , 137 , 42 rees m. j. 1984 , ara&a , 22 , 471 rodriguez - pascual p. m. , et al .
1997 , apjs , 110 , 9 shankar , f. , weinberg , d. h. , miralda escud , j. 2009 , apj , 690 , 20 shapovalova a.i .
2008 , a&a , 486 , 99 shen y. , greene j.e . ,
strauss m.a . , richards g.t . , schneider d. p. 2008
, apj , 680 , 169 somerville r. s. , hopkins p. f. , cox t. j. , robertson b. e. , hernquist l. 2008 , mnras , 391 , 481 stalin c.s . , gopal - krishna , sagar r. , witta p.j .
2004 , japa , 25 , 1 van groningen e. , wanders i. 1992 , pasp , 104 , 700 vron - cetty m.p .
, vron p. 2006 ,
a&a , 455 , 776 vestergaard m. 2004 , apj , 601 , 676 welsh w. f. , 1999 , pasp , 111 , 1347 white r. j. , peterson b. m. , 1994 , pasp , 106 , 879 | we present the results of our optical monitoring campaign of the x - ray source h 0507 + 164 , a low luminosity seyfert 1.5 galaxy at a redshift , @xmath0 .
spectroscopic observations were carried out during 22 nights in 2007 , from the 21 of november to the 26 of december .
photometric observations in the r - band for 13 nights were also obtained during the same period . the continuum and broad line fluxes of the galaxy were found to vary during our monitoring period .
the r - band differential light curve with respect to a companion star also shows a similar variability . using cross correlation analysis , we estimated a time delay of @xmath1 days ( in the rest frame ) , of the response of the broad h@xmath2 line fluxes to the variations in the optical continuum at 5100 . using this time delay and the width of the h@xmath2 line , we estimated the radius for the broad line region ( blr ) of @xmath3 parsec , and a black hole mass of @xmath4m@xmath5 .
[ firstpage ] galaxies : active - galaxies : individual ( h 0507 + 164 ) - galaxies : seyfert |
the broad emission iron lines are well - known features found in about two dozens of spectra of active galactic nuclei and black hole binaries .
they are supposed to originate close to the black hole by the reflection of the primary radiation on the accretion disc .
the spin of the black hole plays an important role in the forming of the line shape .
especially , it determines the position of the marginally stable orbit which is supposed to confine the inner edge of the accretion disc ( see figure [ intro ] ) .
the innermost stable orbit occurs closer to a black hole with a higher spin value .
however , the spin affects also the overall shape of the line . over almost two decades
the most widely used model of the relativistic disc spectral line has been the one by , which includes the effects of a maximally rotating kerr black hole . in other words
, the laor model sets the dimensionless angular momentum @xmath1 to the canonical value of @xmath0 so that it can not be subject of the data fitting procedure .
have relaxed this limitation and allowed @xmath1 to be fitted in the suite of ky models .
other numerical codes have been developed independently by several groups ( , , ) and equipped with similar functionality .
however , the laor model can still be used for evaluation of the spin if one identifies the inner edge of the disc with the marginally stable orbit . in this case
the spin is actually estimated from the lower boundary of the broad line .
the comparison of the laor and model is shown in the right panel of figure [ intro ] .
the other parameters of the relativistic line models are inclination angle @xmath2 , rest energy of the line @xmath3 , inner radius of the disc @xmath4 , outer radius of the disc @xmath5 , emissivity parameters @xmath6 , @xmath7 with the break radius @xmath8 .
the emissivity of the line is given by @xmath9 for @xmath10 and @xmath11 for @xmath12 .
the angular dependence of the emissivity is characterized by limb darkening profile @xmath13 in the laor model .
the model enables to switch between different emission laws .
we used further two extreme cases , the with the same limb - darkening law as in the laor model and * with the limb - brightening law @xmath14 .
the aim of this paper is to compare the two models applied to the current data provided by the xmm - newton satellite , and to the artificial data generated for the on - coming x - ray mission . for this purpose
we have chosen two sources , mcg-6 - 30 - 15 and gx 339 - 4 , which exhibit an extremely skewed iron line according to recently published papers ( ) . and marginally stable orbit @xmath15 .
right : comparison of the laor ( black , solid ) and ( red , dashed ) model for two values of the spin @xmath16 ( top ) and @xmath17 ( bottom ) .
the other parameters of the line are @xmath18kev , @xmath19 , @xmath20.,scaledwidth=98.0% ] and marginally stable orbit @xmath15 .
right : comparison of the laor ( black , solid ) and ( red , dashed ) model for two values of the spin @xmath16 ( top ) and @xmath17 ( bottom ) .
the other parameters of the line are @xmath18kev , @xmath19 , @xmath20.,title="fig:",scaledwidth=98.0% ] + and marginally stable orbit @xmath15 .
right : comparison of the laor ( black , solid ) and ( red , dashed ) model for two values of the spin @xmath16 ( top ) and @xmath17 ( bottom ) .
the other parameters of the line are @xmath18kev , @xmath19 , @xmath20.,title="fig:",scaledwidth=98.0% ]
we used the sas software version 7.1.2 ( http://xmm.esac.esa.int/sas ) to reduce the xmm - newton data of the sources .
further , we used standard tools for preparing and fitting the data available at http://heasarc.gsfc.nasa.gov ( ftools , xspec ) the galaxy mcg-6 - 30 - 15 is a nearby seyfert 1 galaxy ( @xmath21 ) .
the skewed iron line has been revealed in the x - ray spectra by all recent satellites .
the xmm - newton observed mcg-6 - 30 - 15 for a long 350ks exposure time during summer 2001 ( revolutions 301 , 302 , 303 ) .
the spectral results are described in .
we joined the three spectra into one using the ftool mathpha .
the black hole binary gx 339 - 4 exhibited a strong broadened line in the 76ks observation in 2002 ( ) when the source was in the very high state ( for a description of the different states see ) .
the observation was made in the _ burst mode _ due to a very high source flux .
the 97@xmath22 of photons are lost during the reading cycle in this mode , which results into 2.25ks total exposure time .
we rebinned all the data channels in order to oversample the instrumental energy resolution maximally by a factor of 3 and to have at least 20 counts per bin .
the first condition is much stronger with respect to the total flux of the sources
@xmath23erg@xmath24s@xmath25 in 210kev ( @xmath26cts ) for mcg-6 - 30 - 15 and @xmath27erg@xmath24s@xmath25 in 210kev ( @xmath28cts ) for gx 339 - 4 .
[ cols="^,^,^ " , ]
we investigated the iron line band for two representative sources
mcg-6 - 30 - 15 ( active galaxy ) and gx 339 - 4 ( x - ray binary ) .
the iron line is statistically better constrained for the active galaxy mcg-6 - 30 - 15 due to a significantly longer exposure time of the available observations for comparison of count rates of the sources see table 3 .
the spectra of both sources are well described by a continuum model plus a broad iron line model .
we compared modeling of the broad iron line by the two relativistic models , laor and .
the model leads to a better defined minimum of @xmath29 for the best fit value .
the confidence contour plots for @xmath30 versus other model parameters are more regularly shaped .
this indicates that the model has a smoother adjustment between the different points in the parameter space allowing for more reliable constraints on @xmath30 .
the laor model has a less accurate grid and is strictly limited to the extreme kerr metric .
the discrepancies between the and laor results are within the general uncertainties of the spin determination using the skewed line profile when applied to the current data .
however , the results are apparently distinguishable for higher quality data , as those simulated for the xeus mission .
we find that the laor model tends to overestimate the spin value and furthermore , it has insufficient energy resolution which affects the correct determination of the high - energy edge of the broad line .
the discrepancies in the overall shape of the line are more visible especially for lower values of the spin @xmath30 . as a side - product
, we have found that the correct re - binning of the data with respect to the instrumental energy resolution is crucial to obtain statistically the most relevant results .
arnaud , m. , barcons , x. , barret , d. , bautz , m. , bellazzini , r. , xeus : the physics of the hot evolving universe , _ experimental astronomy _ , tmp ,
24a , 2008 beckwith , k. , done , c. , iron line profiles in strong gravity , _ monthly notices of royal astronomical society _ , 352 , 353 , 2004 brenneman , l. w. , reynolds , c. s. , constraining black hole spin via x - ray spectroscopy , _ astrophysical journal _ , 652 , 1028 , 2006 ade , a. , calvani , m. , relativistic emission lines from accretion discs around black holes , _ monthly notices of royal astronomical society _ , 363 , 177 , 2005 doviak , m. , karas , v. , yaqoob , t. , an extended scheme for fitting x - ray data with accretion disc spectra in the strong gravity regime , _ astrophysical journal supplements _ , 153 , 205 , 2004 fabian , a. c. , vaughan , s. nandra , k. , iwasawa , k. , ballantyne , d. r. et al . ,
a long hard look at mcg-6 - 30 - 15 with xmm - newton , _ monthly notices of royal astronomical society _ , 335 , l1 , 2002 gallo e. , corbel s. , fender r. p. , a transient large - scale relativistic radio jet from gx 339 - 4 , _ monthly notices of royal astronomical society _ , 347l , 52 g , 2004 laor , a. , line profiles from a disk around a rotating black hole , _ astrophysical journal _ , 376 , 90 , 1991 miller , j. m. , fabian , a. c. , reynolds , c. s. , nowak , m. a. , homan , j. et al . ,
evidence of black hole spin in gx 339 - 4 : xmm - newton / epic - pn and rxte spectroscopy of the very high state , _ astrophysical journal _ , 606 , l131 , 2004 miller , j. m. , homan , j. , steeghs , d. , rupen , m. , hunstead , r. w. et al .
, a long , hard look at the low / hard state in accreting black holes , _ astrophysical journal _
, 653 , 525 , 2006 miller , l. , turner , t. j. , reeves , j. n. , an absorption origin for the x - ray spectral variability of mcg-6 - 30 - 15 , _ astronomy & astrophysics _
, 483 , 437 , 2008 reis , r. c. , fabian , a. c. , ross , r. , miniutti , g. , miller , j. m. , reynolds , c. s. , a systematic look at the very high and low / hard state of gx 339 - 4 : constraining the black hole spin with a new reflection model , _ arxiv:0804.0238 _ , 2008 remillard , r. a. and mcclintock , j. e. , x - ray properties of black hole binaries , _ annual review of astronomy & astrophysics _
, 49 , 2006 vaughan , s. , fabian , a. c. , a long hard look at mcg-6 - 30 - 15 with xmm - newton - ii .
detailed epic analysis and modelling , _ monthly notices of royal astronomical society _ , 348 , 1415 , 2004 | the analysis of the broad iron line profile in the x - ray spectra of active galactic nuclei and black hole x - ray binaries allows us to constrain the spin parameter of the black hole .
we compare the constraints on the spin value for two x - ray sources , mcg-6 - 30 - 15 and gx 339 - 4 , with a broad iron line using present relativistic line models in xspec laor and .
the laor model has the spin value set to the extremal value @xmath0 , while the model enables direct fitting of the spin parameter .
the spin value is constrained mainly by the lower boundary of the broad line , which depends on the inner boundary of the disc emission where the gravitational redshift is maximal .
the position of the inner disc boundary is usually identified with the marginally stable orbit which is related to the spin value . in this way
the laor model can be used to estimate the spin value .
we investigate the consistency of the laor and models .
we find that the spin values evaluated by both models agree within the general uncertainties when applied on the current data .
however , the results are apparently distinguishable for higher quality data , such as those simulated for the international x - ray observatory ( ixo ) mission .
we find that the laor model tends to overestimate the spin value and furthermore , it has insufficient resolution which affects the correct determination of the high - energy edge of the broad line . + |
solvable lattice models related to the lie superalgebras @xcite have attracted a great deal of attention @xcite .
for example , the supersymmetric @xmath3 model in strongly correlated electron system has received much attentions in relation with the high @xmath4 superconductivity .
these models have both fermionic and bosonic degree of freedom , and are given as solutions of the graded yang - baxter equations @xcite . to solve such models , the bethe ansatz is widely used in many literatures ( see , for example , @xcite and references therein ) . however , many of them deal only with models related to simple representations like fundamental ones ; there was few _ systematic _
study by the bethe ansatz on more complicated models such as fusion models @xcite . in view of such situations ,
we have recently carried out @xcite _ systematically _ an analytic bethe ansatz @xcite related to the lie superalgebras @xmath5 and @xmath6 .
namely , we have proposed a class of dressed vacuum forms ( dvfs ) labeled by young ( super ) diagrams and a set of fusion relations ( @xmath1-system ) among them . besides the eigenvalue formulae of the transfer matrices ,
the thermodynamics have also been discussed by several people .
particularly , the thermodynamic bethe ansatz ( tba ) @xcite equation was proposed for the supersymmetric @xmath3 model @xcite , which is related to @xmath7 , and the supersymmetric extended hubbard model @xcite , which is related to @xmath8 .
moreover , there is a paper @xcite on the excited state tba equation for these two important models of 1d highly correlated electron systems from the point of view of the quantum transfer matrix ( qtm ) method .
in addition , the tba equation for @xmath9 model was presented @xcite in relation with the continuum limit of the integrable super spin chains .
however , the thermodynamics of the quantum spin model related to the orthosymplectic lie superalgebra @xmath10 is not so understood as @xmath9 case .
in particular , as far as we know , there have been no literatures on the tba equation even for the simplest orthosymplectic @xmath0 integrable spin chain @xcite .
this is regrettable because this model may be related to interesting topics such as @xmath11 superconformal - symmetry in field theory , and the loop model , which will describe statistical properties of polymers in condensed matter physics @xcite . in view of these situation
, we have recently proposed the tba equation for the @xmath0 integrable spin chain @xcite by using the string hypothesis @xcite . though we expect that the resultant tba equation describes the free energy correctly , there exists the deviation from the string hypothesis @xcite .
in addition , it is difficult to evaluate other physical quantities such as correlation length . as an alternative method which overcomes such difficulties , the qtm method has been proposed @xcite .
now we shall briefly sketch the qtm method . utilizing the general equivalence theorem @xcite ,
one transforms the 1d quantum system into the 2d classical counterpart and defines the qtm on such a fictitious system of size @xmath12 ( referred to as the trotter number , which should be taken @xmath13 ) . since the qtm has a finite gap , the original problem for the calculation of the partition function reduces to finding the single largest eigenvalue of the qtm . to evaluate it actually ,
we utilize the underlying integrable structure , which admits introduction of the commuting " qtm with a complex parameter @xmath14 @xcite .
furthermore , we introduce some auxiliary functions including the qtm itself , which satisfy functional relations .
we select these auxiliary functions such that they are _
a_nalytic , _
n_on_z_ero and have _ c_onstant asymptotics in appropriate strips on the complex @xmath14-plane ( we call this property anzc ) .
thus we can transform the functional relations into the non - linear integral equations ( nlie ) which describe the free energy . in these nlie
, we can take the trotter limit @xmath13 . adopting a subset of fusion hierarchy as auxiliary functions
, we find that these nlie are equivalent to the tba equation based on the string hypothesis @xcite .
in general , a set of fusion hierarchy satisfies the functional relations called @xmath1-system , and this @xmath1-system is transformed to the @xmath2-system . and @xmath2-system are given in @xcite and @xcite ( see also @xcite ) .
] by selecting them so that they have anzc property in appropriate strips , one derives the nlie which will be identical to the tba equation .
furthermore , considering the sub - leading eigenvalues of the qtm , we can derive systematically the excited state " tba equations which provide the correlation length at any finite temperatures .
the purpose of this paper is to apply our recent results @xcite to the @xmath0 integrable spin chain , and to construct the tba equation and it s excited state version from the point of view of the above - mentioned qtm method .
we have also confirmed the fact that our tba equation coincides with the one @xcite from the string hypothesis .
we believe that this paper yields a basis of future studies of the thermodynamics by the qtm method for more general models such as the @xmath10 model .
the layout of this paper is as follows . in section 2
we formulate the @xmath0 integrable spin chain at finite temperatures in terms of the commuting qtm . in section 3 , to evaluate its eigenvalue , we utilize the fusion hierarchy of the qtm
. it is given as a set of the dressed vacuum forms ( dvfs ) which is a summation over tableaux labeled by a young ( super ) diagram with one column .
these dvfs and their certain combinations ( @xmath2-functions ) satisfy the @xmath1-system and the @xmath2-system , respectively . the formulation of the dvfs and the functional relations are essentially independent of the vacuum part .
thus we can utilize the results in ref .
@xcite only by replacing the vacuum parts of the dvfs defined in ref .
@xcite with those of the qtm . based on the analytical property of the dvfs and the @xmath2-functions , in section 4
we derive the non - linear integral equations for the free energy , and in section 5 for the correlation length .
section 6 is devoted to the summary and the discussion .
in this section , we introduce the quantum spin chain related to the lie superalgebra @xmath0 and formulate the commuting qtm for this model .
the classical counterpart of this model is given by the @xmath15-matrix ( see for example , ref .
@xcite ) @xmath16 which is a rational solution of the graded yang - baxter equation @xcite associated with the three dimensional representation of the lie superalgebra @xmath0 ( to be precise , super yangian " @xmath17 ) , whose basis is @xmath18-graded and labeled by the parameters @xmath19 , @xmath20 . here
@xmath21 denotes the graded permutation operator which has @xmath22 matrix elements @xmath23 and @xmath24 is a @xmath22 matrix whose matrix elements are @xmath25 where @xmath26 is a @xmath27 matrix @xmath28 and @xmath29 with the total order @xmath30 .
the row - to - row transfer matrix @xmath31 is defined by @xmath32 , \label{rtransf}\ ] ] where @xmath33 is the number of lattice sites and @xmath34 denotes @xmath35 which acts non - trivially on the auxiliary space @xmath36 and the @xmath37-th site of the quantum space .
note that @xmath38 is the ( non - graded ) permutation operator .
the hamiltonian of the corresponding quantum system is given by taking the logarithmic derivative of above transfer matrix ( [ rtransf ] ) at @xmath39 , @xmath40 where we assume a periodic boundary condition . here
@xmath41 is a real coupling constant which determines the phase of this model ; the ferromagnetic and antiferromagnetic regimes correspond @xmath42 and @xmath43 , respectively ( see for example , @xcite ) . to consider the finite temperature property of the model ( [ hami ] )
, we shall introduce another transfer matrix @xmath44 constructed by the @xmath45-matrix @xmath46 , which is defined by @xmath47 rotation of @xmath48 , i.e. , @xmath49 ( @xmath50 means the transposition for the @xmath45-matrix in the @xmath51-th space ) : @xmath52 . \label{ltransf}\ ] ] we can see that the logarithmic derivative of above transfer matrix ( [ ltransf ] ) also represents the hamiltonian ( [ hami ] ) .
thus the expansion of the transfer matrices ( [ rtransf ] ) and ( [ ltransf ] ) are expressed as @xmath53 combining above relations and using the fact @xmath54 ( note that @xmath55 and @xmath56 denote the right and left sift operators , respectively ) , we have @xmath57 consequently , the partition function @xmath58 of the model ( [ hami ] ) can be expressed as @xmath59 where @xmath12 is an even number : it is called the trotter number , which represents the number of fictitious sites in the trotter direction ; @xmath60 ( @xmath1 is the temperature ) .
thus the free energy per site @xmath61 is given by @xmath62 however , in the antiferromagnetic case @xmath43 , the eigenvalues of @xmath63 will be infinitely degenerate in the limit @xmath64 . therefore taking the trace in above expression ( [ free1 ] )
is a serious problem . to avoid this difficulty
, we transform the term @xmath63 as follows : @xmath65 now we introduce the qtm , which plays the fundamental role to describe the thermal quantities @xmath66 due to the yang - baxter equation , we see that the qtm is commutative : @xmath67=0.\ ] ] hereafter we write the @xmath51-th largest eigenvalue of the qtm @xmath68 as @xmath69 .
since the two limits in ( [ free1 ] ) will be exchangeable as proved in @xcite , we shall take the limit @xmath70 first .
since there is a finite gap between @xmath71 and @xmath72 even in the trotter limit @xmath13 , we have @xmath73 the thermodynamical completeness @xmath74 follows from @xmath75 , which is clear from @xmath76 . taking the ratio of the largest eigenvalue and the second largest one , we can systematically calculate the correlation length @xmath77 : @xmath78
in this section , we present the eigenvalue formulae of the fusion qtm @xmath79 as the eigenvalues of the fusion qtm . ] and the functional relations among them by using the results for the row - to - row transfer matrix @xcite . although the quantity we want to evaluate is only @xmath80 , consideration on all @xmath81 plays an essential role in our formulation . for @xmath82 , we define the function @xmath83 as @xmath84 where @xmath85 ; @xmath86 is a quantum number ; @xmath87 .
the vacuum parts of the functions @xmath83 ( [ box ] ) are given as follows : @xmath88 where @xmath89 the eigenvalue formula of the qtm is given as the summation over these boxes ( [ box ] ) : @xmath90 we have denotes the residue of a function @xmath91 at @xmath92 . ]
@xmath93 under the following type of the bethe ansatz equation ( bae ) : @xmath94 thus @xmath95 ( [ pikata ] ) is free of poles under this bae ( [ bae ] ) .
we observe , from numerical analysis , that the largest and the second largest eigenvalues lie in the sector @xmath96 and @xmath97 , respectively .
we shall introduce a function @xmath98 with a spectral parameter @xmath99 and a young ( super ) diagram . ]
@xmath100 , which is a candidate of the dvf for the fusion qtm .
from now on , we often abbreviate the parameter @xmath101 to simplify the notation . for @xmath102
, we define @xmath103 as a summation over products of the boxes in ( [ box ] ) : @xmath104 where the spectral parameter @xmath14 is shifted as @xmath105 from the top to the bottom .
@xmath106 is a set of tableaux @xmath107 ( @xmath108 ) obeying the following rule ( admissibility conditions ) @xmath109 the normalization function @xmath110 is introduced so that the degree of @xmath103 is @xmath111 with respect to @xmath14 : @xmath112 the dvf ( [ dvf - tableau ] ) is free of poles under the bae ( [ bae ] ) for any @xmath113 . note that @xmath103 ( [ dvf - tableau ] ) for @xmath114 reduces to ( [ pikata ] ) since @xmath115 .
the bae ( [ bae ] ) for the qtm has the same form as the one @xcite for the row to row transfer matrix except the vacuum part .
thus the dress part of the function @xmath116 ( [ dvf - tableau ] ) is same as that in @xcite .
therefore the dvfs @xmath116 ( @xmath102 ) satisfy the following functional relations - system ( [ t - sys ] ) is one of them . ]
( @xmath0 version of the fusion relations or the @xmath1-system in ( [ t - sys ] ) slightly different from that in @xcite for practical reason : the variable @xmath117 in ( [ t - sys ] ) corresponds to @xmath118 in @xcite . ] ) , which has essentially the same form in @xcite : @xmath119 where @xmath120 we can prove these functional relations by the jacobi identity and a duality among the dvfs @xcite . using @xmath116 ( [ dvf - tableau ] ) ,
we introduce the following functions ( @xmath2-functions ) : @xmath121 these functions ( [ y - fun ] ) satisfy a so - called @xmath2-system : @xmath122 where @xmath123 . in the subsequent sections ,
these functional relations ( [ y - sys ] ) will be transformed into the tba equations .
let @xmath124 be the dvfs @xmath125 constructed by the roots of bae ( [ bae ] ) which provide the @xmath51-th largest eigenvalue of the qtm @xmath95 , and @xmath126 be the @xmath2-functions constructed by @xmath127 as in ( [ y - fun ] ) . in this section
, we study the analyticity of the dvfs @xmath128 and the @xmath2-functions @xmath129 in the complex @xmath14-plane .
then we derive the non - linear integral equations ( nlie ) which provide the free energy . for this purpose , by keeping the trotter number @xmath12 finite , we have performed numerical analyses with various values of @xmath130 and @xmath12 in investigating the location of zeros of @xmath128 . as seen in fig.[roots ] , there are @xmath131 two - string solutions of the bae ( [ bae ] ) , which will provide the largest eigenvalue of the qtm .
these roots locate symmetrically with respect to the imaginary axis and the line @xmath132 . from the definition of the dvfs ( [ dvf - tableau ] ) , they have @xmath111 zeros on the complex @xmath14-plane .
for example , the location of zeros of @xmath133 for @xmath134 , @xmath135 is plotted in fig.[zeroslgt12 ] . according to them ,
we observe that @xmath136 have @xmath12 zeros on the smooth curve near the line @xmath137 and @xmath12 zeros on the smooth curve near the line @xmath138 .
thus we expect that the following conjecture is valid even in the limit @xmath64 .
[ conj - lar ] all the zeros of @xmath136 for small @xmath139 are located outside of the strip @xmath140 $ ] . from the definition ( [ dvf - tableau ] ) ,
one observes the fact that all the zeros are symmetric with respect to both real and imaginary axes .
as seen in the figure , the deviation from the line is very small , which will be smaller as @xmath141 .
once this conjecture is assumed , we can identify the strip where the @xmath2-functions @xmath129 have the property of _
a_nalytic _ n_on_z_ero and _ c_onstant asymptotics in the limit @xmath142 ( we call it anzc property ) . using the definition of the dvfs ( [ dvf - tableau ] ) and the @xmath2-functions ( [ y - fun ] ) , we can obtain the asymptotic value of the @xmath2-functions in the limit @xmath143 : @xmath144 from conjecture [ conj - lar ] and above asymptotics ( [ yasympt ] ) , the functions @xmath145 and @xmath146 have anzc property in the strip @xmath147 $ ] ( @xmath148 ) . on the other hand , the functions @xmath149 have the anzc property in the strip @xmath150 $ ] ( we call this strip physical strip ) , except for @xmath151 possesses poles ( resp .
zeros ) of order @xmath131 at @xmath152 ( resp .
@xmath153 ) in the physical strip for @xmath42 ( resp .
@xmath43 ) . note that @xmath154 is a small quantity given in ( [ part ] ) .
using cauchy s theorem , we can transform the @xmath2-system ( [ y - sys ] ) into the nlie in the following way .
let us consider the @xmath2-system ( [ y - sys ] ) for @xmath155 .
first , we take the logarithmic derivative and perform the fourier transformation on both side of the eq.([y - sys ] ) .
second , by cauchy s theorem , the fourier integral for the logarithmic derivative of @xmath156 is represented by that of @xmath157 and @xmath158 .
finally , performing the inverse fourier transformation and integrating over @xmath14 , we obtain the desired nlie .
the integral constants are determined by the asymptotic value in ( [ yasympt ] ) . in
the case @xmath114 in ( [ y - sys ] ) , we have to modify @xmath159 to transform @xmath2-system into the nlie since @xmath159 have the poles ( or zeros ) in the physical strip .
@xmath160 where the @xmath161 and @xmath162 signs in front of @xmath101 and @xmath131 should be chosen according as @xmath42 and @xmath43 , respectively .
according to the identity @xmath163 , @xmath164 in the lhs of the eq.([y - sys ] ) can be replaced by @xmath165 in ( [ mody ] ) .
thus the @xmath2-system ( [ y - sys ] ) can be transformed to the nlie in the similar way as mentioned above .
consequently the resultant nlie are represented as @xmath166 where @xmath167 denotes the convolution @xmath168 and @xmath169 in above nlie ( [ goro ] ) , the trotter limit @xmath170 can be calculated : @xmath171 we thus arrive at the nlie for @xmath156 which are independent of the trotter number @xmath12 .
@xmath172 under the identification @xmath173 , above equations ( [ tba ] ) are identical to the tba equation ( 30 ) in @xcite , which was derived by the thermodynamic bethe ansatz based on the string hypothesis . from above equations and the asymptotics ( [ yasympt ] )
, we can determine the @xmath2-functions @xmath156 uniquely . to obtain the free energy per site
, we shall modify @xmath174 as @xmath175 from the definition of the @xmath2-function ( [ y - fun ] ) and the @xmath1-system ( [ t - sys ] ) @xmath176 satisfies the following functional relation .
@xmath177 where @xmath178 anzc property of the both sides leads to @xmath179 where @xmath180 denotes the kernel , @xmath181 calculating the trotter limit @xmath13 , we obtain the free energy per site ( [ free2 ] ) @xmath182 this representation coincides with eq.(33 ) in @xcite .
in this section , we develop the analysis for the largest eigenvalue of the qtm to the the analysis of the second largest one .
namely we derive the nlie ( excited state tba equation ) which provide the correlation length at finite temperature .
we assume that the coupling constant @xmath41 is negative : we consider only the antiferromagnetic regime in this section .
after some numerical analyses , we observe that @xmath97 roots @xmath183 of the bae ( [ bae ] ) , which will bring us the second largest eigenvalue of the qtm , form @xmath184 two - strings and one three - string if @xmath185 ( cf . fig .
[ roots122nd ] ) ; @xmath186 two - strings and one one - string if @xmath187 ( cf .
[ roots142nd ] ) . as in the largest eigenvalue case ,
the distribution of these roots are symmetric with respect to the imaginary axis and the line @xmath132 . from the definition of the dvf ( [ dvf - tableau ] ) , @xmath188 has @xmath111 zeros on the complex @xmath14-plain . from the numerical analyses , we observe that @xmath188 has @xmath189 zeros on the smooth curve near the line @xmath137 , @xmath189 zeros on the smooth curve near the line @xmath138 , two zeros on the imaginary axis near the points @xmath190 and two zeros @xmath191 ( @xmath192 ) on the real axis .
for example , the location of zeros of @xmath193 for @xmath134 and @xmath194 are plotted in figs .
[ zeroslgt122nd],[zeroslgt142nd ] .
we expect that the following conjecture is valid even in the limit @xmath195 .
[ conj2 ] @xmath188 has two real zeros @xmath191 .
every other zero is located outside of the physical strip @xmath150 $ ] . as in the previous section
, we shall identify the strip where the @xmath2-functions @xmath196 have the anzc property . from the definition of the dvfs ( [ dvf - tableau ] ) and the @xmath2-functions ( [ y - fun ] ) , we find the asymptotic value of @xmath197 : @xmath198 from the definition of ( [ y - fun ] ) and conjecture [ conj2 ] , we find @xmath199 has the anzc property in the strip @xmath200 $ ] ( @xmath201 ) , while @xmath197 has zeros of order 1 at @xmath202 and poles of order 1 at @xmath203 if @xmath204 ; @xmath205 has zeros of order @xmath131 at @xmath206 , zeros of order 1 at @xmath207 and poles of order 1 at @xmath208 , and that every other singular point locates outside of the physical strip .
thus we have to modify @xmath197 as follows : @xmath209 these @xmath210 ( [ es - y - fun ] ) have the anzc property in the physical strip .
after similar calculation in the previous section , we obtain the excited state tba equation : @xmath211 where the integral kernel is defined in ( [ kernel ] ) .
in addition @xmath212 is determined by the condition @xmath213 . through
the above excited state tba equation ( [ es - tba2 ] ) , the second largest eigenvalue is given as follows : @xmath214 this second largest eigenvalue ( [ second ] ) together with the largest one ( [ largest ] ) describe the correlation length at any finite temperature ( see , ( [ corr ] ) ) .
these excited state tba equation ( [ es - tba2 ] ) and eq .
( [ second ] ) will be difficult to obtain by the string hypothesis ; they exemplify the efficiency of the qtm method .
we have applied the qtm method to the @xmath0 integrable spin chain . making use of the @xmath1-system and the @xmath2-system , we derived the infinite set of nlie which provide the free energy and the correlation length at any finite temperatures .
as concerns the free energy , these nlie are identical to the tba equation based on the string hypothesis .
one can say that the validity of the @xmath0 @xmath1-system in ref .
@xcite has been confirmed from the point of view of the qtm method . to the author s knowledge , this paper is the first explicit derivation of the excited state tba equation for the @xmath0 spin chain .
we comment on the ferromagnetic regime @xmath42 in the second largest eigenvalue sector , which we have not considered in section 5 . in the antiferromagnetic regime @xmath43 ,
the correlation length are characterized by the additional zeros of the fusion qtms .
these zeros are real and symmetric with respect to the imaginary axis and this symmetry will never break at any finite temperature .
while in the ferromagnetic regime ( @xmath42 ) , the level crossing " will occur successively .
this attributes to the change of distribution patterns of the additional zeros as follows .
first at high temperature , two additional zeros are on the imaginary axis in the physical strip . however , at low temperature , this pattern will no longer provide the second largest eigenvalue .
there is another definition of the transfer matrix for the @xmath45-matrix ( [ rmat2 ] ) , which is based on the _ graded _ formalism @xcite of the quantum inverse scattering method . in this case , the @xmath45-matrix ( [ rmat2 ] ) is defined as @xmath215 , where @xmath21 is the _ graded _ permutation operator . as a result
, the ordinary row - to - row transfer matrix is defined as the _ super_-trace of a monodromy matrix , and the phase factor ( cf.@xcite ) of the bae disappears @xcite . in the qtm for the graded case , we are not sure how the phase factor corresponding to the right hand side in ( [ bae ] ) changes
. one might suspect that our tba equation ( [ tba ] ) is related to the one @xcite for the izergin - korepin model since the representation space of @xmath216 has close resemblance to that of the affine lie algebra @xmath217 .
however , to our knowledge , the tba equation for the izergin - korepin model has been constructed @xcite so that it reduces to the one for a @xmath218-invariant model in the rational limit
. there will be another branch of the rational limit for the izergin - korepin model @xcite .
together with the above - mentioned graded case , the comparison with our results will be interesting .
it is an interesting problem to extend a similar analysis to more general @xmath10 model .
unfortunately , for @xmath219 case , we have case is an exception to this statement @xcite . ] only the subset of the @xmath1-system @xcite . while for @xmath220 case , we have already a complete set of the @xmath1-system @xcite , which will be relevant for the qtm method .
so we hope to report , at the beginning , the tba equation for the @xmath221 model in the near future .
finally we refer to another formulation of nlie based on the finite number of the auxiliary functions ( see for example , @xcite ) .
although there seems to be no connection between the tba equation and such nlie , this formulation will be advantageous to the numerical calculation of the physical quantities at finite temperature .
zt would like to thank professor j. suzuki for discussions at the earlier stage of the work @xcite , which gives a motivation of the present work .
the authors also thank him for his useful comments .
they would like to thank professor a. kuniba for his encouragement .
this work is supported in part by a grant - in - aid for jsps research fellows from the ministry of education , science , culture and sports of japan .
80 v. kac , adv . math . * 26 * , 8 ( 1977 ) .
kulish and e. k. sklyanin , j. sov .
math * 19 * , 1596 ( 1982 ) .
kulish , j. sov . math . * 35 * , 2648 ( 1986 ) .
v. v. bazhanov and a. g. shadrikov , theor .
phys . * 73 * , 1302 ( 1988 ) .
kulish and n. yu .
reshetikhin , lett . math
. phys . * 18 * , 143 ( 1989 ) .
t. deguchi , a. fujii and k. ito , phys .
b * 238 * , 242 ( 1990 ) .
h. saleur , nucl .
b * 336 * , 363 ( 1990 ) .
r. b. zhang , a. j. bracken and m. d. gould , phys .
b * 257 * , 133 ( 1991 ) .
m. j. martins and p. b. ramos , j. phys .
a * 27 * , l703 ( 1994 ) .
p. schlottmann , phys . rev .
b * 36 * , 5177 ( 1987 ) .
f. h. l. essler and v. e. korepin , phys .
b * 46 * , 9147 ( 1992 )
. f. h. l. essler , v. e. korepin and k. schoutens , phys .
lett . * 68 * , 2960 ( 1992 ) .
a. foerster and m. karowski , nucl .
b * 396 * , 611 ( 1993 ) .
f. h. l. essler and v. e. korepin , int .
b * 8 * , 3243 ( 1994 ) . z. maassarani , j. phys .
a * 28 * , 1305 ( 1995 ) .
m. j. martins , phys .
* 74 * , 3316 ( 1995 ) .
m. j. martins , nucl .
b * 450 * , 768 ( 1995 ) . p. b. ramos and m. j. martins , nucl .
b * 474 * , 678 ( 1996 ) .
m. p. pfannmller and h. frahm , nucl .
b * 479 * , 575 ( 1996 ) .
m. j. martins and p. b. ramos , nucl .
b * 500 * , 579 ( 1997 ) .
p. p. kulish , n. yu . reshetikhin and e. k. sklyanin lett .
* 5 * , 393 ( 1981 ) .
z. tsuboi , j. phys .
a * 30 * , 7975 ( 1997 ) .
z. tsuboi , physica a * 252 * , 565 ( 1998 ) .
z. tsuboi , j. phys .
a * 31 * , 5485 ( 1998 ) .
z. tsuboi , physica a * 267 * , 173 ( 1999 ) . z. tsuboi , j. phys .
a * 32 * , 7175 ( 1999 ) .
n. yu . reshetikhin , sov .
jetp * 57 * , 691 ( 1983 ) .
reshetikhin , lett .
* 14 * , 235 ( 1987 ) . v. v. bazhanov and n.yu .
reshetikhin , j. phys .
a * 23 * , 1477 ( 1990 ) .
j. suzuki , _ phys .
lett . _ a * 195 * , 190 ( 1994 ) .
a. kuniba and j. suzuki , commun . math . phys . *
173 * , 225 ( 1995 ) .
a. kuniba and j. suzuki , j. phys .
a * 28 * , 711 ( 1995 ) .
a. kuniba , y. ohta and j. suzuki , j. phys .
a * 28 * , 6211 ( 1995 ) .
yang and c.p .
yang , j. math . phys .
* 10 * , 1115 ( 1969 ) .
g. jttner , a. klmper and j. suzuki , nucl .
b * 512 * , 581 ( 1998 ) .
h. saleur , the continuum limit of @xmath222 integrable super spin chains , solv - int/9905007 .
m. j. martins , b. nienhuis and r. rietman , phys .
rev . lett . * 81 * , 504 ( 1998 ) . k. sakai and z. tsuboi , thermodynamic bethe ansatz equation for @xmath0 integrable spin chain , math - ph/9911010 ; mod .
. a , * 14 * , 2427 ( 1999 ) . m. gaudin , phys .
* 26 * , 1301 ( 1971 ) .
m. takahashi , prog .
. phys . * 46 * , 401 ( 1971 ) .
m. takahashi and m. suzuki , prog .
* 48 * , 2187 ( 1972 ) .
alcaraz and m.j .
martins , j. phys .
a * 21 * , l381 , ibid . 4397 ( 1988 ) .
h. l. eler , v. e. korepin and k. schoutens , j. phys .
a * 25 * , 4115 ( 1992 ) .
g. jttner and b.d .
drfel , j. phys .
a * 26 * , 3105 ( 1993 ) .
m. suzuki , prog .
* 56 * , 1454 ( 1976 ) .
m. suzuki , phys .
b * 31 * , 2957 ( 1985 ) .
m. suzuki and m. inoue , prog .
theor . phys .
* 78 * , 787 ( 1987 ) .
m. inoue and m. suzuki , prog .
* 79 * , 645 ( 1988 ) .
t. koma , prog .
. phys . * 78 * , 1213 ( 1987 ) .
j. suzuki , y. akutsu and m. wadati , j. phys .
jpn * 59 * , 2667 ( 1990 ) .
j. suzuki , t. nagao and m. wadati , int .
b * 6 * , 1119 ( 1992 )
. a. klmper , ann .
physik * 1 * , 540 ( 1992 ) .
a. klmper , z. phys .
b * 91 * , 507 ( 1993 ) .
g. jttner and a. klmper , euro .
lett . * 37 * , 335 ( 1997 ) .
g. jttner , a. klmper and j. suzuki , nucl .
b * 487 * , 650 ( 1997 ) .
a. fujii and a. klmper , nucl .
b * 546 * , 751 ( 1999 ) .
k. sakai , m. shiroishi , j. suzuki and y. umeno , phys .
b * 60 * , 5186 ( 1999 ) .
a. kuniba , k. sakai and j. suzuki , nucl
b * 525 * , 597 ( 1998 ) .
j. suzuki , nucl .
b * 528 * , 683 ( 1998 ) .
k. sakai , j. phys .
* 68 * , 1789 ( 1999 ) .
a. kuniba , t. nakanishi and j. suzuki , int . j. mod
. phys . a * 9 * , 5215 ( 1994 ) .
a. kuniba , t. nakanishi and j. suzuki , int . j. mod
. phys . a * 9 * , 5267 ( 1994 ) . ) for @xmath134 , @xmath135 and @xmath224 , which gives the second largest eigenvalue of the qtm .
there are four two - strings and one three - string .
they are symmetrically distributed with respect to the line @xmath132 and the imaginary axis .
, scaledwidth=95.0% ] ) for @xmath134 , @xmath225 and @xmath226 , which gives the second largest eigenvalue of the qtm .
there are six two - strings and one one - string .
they are symmetrically distributed with respect to the line @xmath132 and the imaginary axis .
, scaledwidth=95.0% ] | the thermodynamic bethe ansatz ( tba ) and the excited state tba equations for an integrable spin chain related to the lie superalgebra @xmath0 are proposed by the quantum transfer matrix ( qtm ) method .
we introduce the fusion hierarchy of the qtm and derive the functional relations among them ( @xmath1-system ) and their certain combinations ( @xmath2-system ) .
their analytical property leads to the non - linear integral equations which describe the free energy and the correlation length at any finite temperatures . with regard to the free energy , they coincide with the tba equation based on the string hypothesis .
int .
j. mod .
phys . a , vol .
15 , p. 2329 - 2346
( 2000 ) |
the rather faint ( @xmath1 ) star sdss j142625.71 + 575218.3 ( referred to hereafter as sdss j1426 + 5752 ) is a fascinating object in several aspects .
first , it belongs to the newly - discovered type of carbon - atmosphere white dwarfs , also known as hot dq stars ( dufour et al .
2007 , 2008a ) .
these are exceedingly rare stars whose unexpected existence was revealed thanks to the availability of some of the data products that came out of the sloan digital sky survey ( e.g. , liebert et al .
2003 and eisenstein et al .
dufour et al . ( 2008b ) found only nine such objects out of a total of about 10,000 white dwarfs identified spectroscopically .
their preliminary atmospheric analysis revealed that all the hot dq white dwarfs fall in a narrow range of effective temperature , between about 18,000 and 24,000 k , and that they have atmospheric carbon - to - helium number ratios ranging from 1 to upward of 100 .
dufour et al . suggested that these stars could be the cooled - down versions of the , so far , unique and very hot ( @xmath2 @xmath3 200,000 k ) carbon - rich pg 1159 star h1504 ( see , e.g. , werner & herwig 2006 ) and form a new family of hydrogen- and helium - deficient objects following the post - agb phase . in this scenario , residual helium would float rapidly to the surface after the pg 1159 phase of evolution , and the descendants of h1504-like stars would thus `` disguise '' themselves as helium - atmosphere white dwarfs ( of the do and db spectral types ) .
this would last until convective mixing dilutes the thin outermost layer of helium in the effective temperature range where substantial subphotospheric convection due to carbon recombination develops in models of these stars .
hence , a dramatic change in the atmospheres of such stars , from helium - dominated to carbon - dominated , would occur in the range of temperature where the hot dq s are actually found .
further evolution would slowly restore the dominance of helium in the atmosphere of these objects as a result of diffusion .
although quite a bit of work needs to be done to establish quantitatively the foundations of this scenario , the preliminary investigations of althaus et al .
( 2009 ) indicate that it is quite viable
. an updated discussion of the properties of hot dq stars has been presented by dufour et al .
( 2009 ) . the second interesting development concerning sdss j1426 + 5752 was the important discovery by montgomery et al . ( 2008 ) that it is a luminosity variable .
on the basis of 7.8 h of integrated light photometry on the mcdonald observatory 2.1 m otto struve telescope , these authors reported that sdss j1426 + 5752 has a light curve dominated by a single periodicity at 417.7 s with an amplitude of about 1.7% of the mean brightness of the star , accompanied by its first harmonic ( 208.9 s ) with a relatively large amplitude ( @xmath40.7% ) , and possibly also by its fourth harmonic as well ( @xmath40.3% ) .
quite interestingly , they also reported that no luminosity variations were detected in five other hot dq s that they surveyed .
using some theoretical arguments , montgomery et al .
( 2008 ) argued that the luminosity variations seen in sdss j1426 + 5752 and not in their other targets could be accounted for naturally in terms of pulsational instabilities .
if true , this means that sdss j1426 + 5752 is the prototype of a new class of pulsating white dwarfs after the gw vir , v777 her , and zz ceti types ( and see , e.g. , fontaine & brassard 2008 for a detailed review on these pulsators ) .
the hypothesis that the luminosity variations seen in sdss j1426 + 5752 are caused by pulsational instabilities associated with low - order and low - degree gravity - mode oscillations ( as in the known types of pulsating white dwarfs ) is backed by the exploratory nonadiabatic calculations carried out independently by fontaine , brassard , & dufour ( 2008 ) in parallel to the efforts of montgomery et al .
( 2008 ) . on the other hand , montgomery et al .
( 2008 ) also noted that the folded light curve of sdss j1426 + 5752 does not resemble those of pulsating white dwarfs showing nonlinearities in their light curves , but shows instead similarities with the folded pulse shape of am cvn , the prototype of the family of helium - transferring cataclysmic variables .
the am cvn stars are close interacting binaries consisting of ( probably ) two helium white dwarfs with orbital periods in the range 1000@xmath53000 s ( and see the reviews of warner 1995 or nelemans 2005 for a lot more details on these challenging objects ) . in these systems , the main photometric period , almost always accompanied by several harmonics , corresponds to the beat period between the orbital period and the precession period of the slightly elliptical accretion disk around the more massive white dwarf .
the dominant component of the light variability usually comes from the moving ( precessing ) optically thick accretion disk .
thus , on the basis of similarities in the folded light pulses between sdss j1426 + 5752 and am cvn , montgomery et al .
( 2008 ) proposed an alternative to pulsational instabilities for explaining its luminosity variations : the possibility that it is , in fact , a new type of close interacting binary , a carbon - transferring analog of am cvn . in this scenario ,
the observed spectrum of sdss j1426 + 5752 would originate from an optically thick carbon - oxygen accretion disk around the more massive white dwarf component in the system .
the pulse shape argument was again used recently by barlow et al .
( 2008 ) to favor the close interacting binary model after those other authors discovered two more luminosity variable hot dq s .
however , counterarguments , favoring this time the pulsation model , have been put forward by dufour et al .
( 2009 ) and fontaine et al .
( 2009 ) .
the third development concerning sdss j1426 + 5752 resulted from follow - up spectroscopic observations carried out by dufour et al .
( 2008c ) at the 6.5 m multiple mirror telescope ( mmt ) and at one of the 10 m keck telescopes .
this was motivated by the sole availability of a rather poor sdss spectrum , insufficient for quantitative analysis .
their objective was , firstly , to obtain a sufficiently good spectrum for detailed atmospheric modeling and , secondly , to search for the presence of helium as required to account for the observed pulsational instabilities according to the nonadiabatic calculations of fontaine et al .
the spectral analysis of the improved spectra readily revealed the presence of a substantial amount of helium in the atmosphere of sdss j1426 + 5752 , an abundance comparable to that of carbon .
this is in line with the expectations of nonadiabatic pulsation theory which requires an important `` helium pollution '' in the atmosphere / envelope of sdss j1426 + 5752 for it to pulsate at its current estimated effective temperature and surface gravity ( dufour et al . 2008b ) .
in addition to this finding , an unexpected surprise came out of the follow - up spectroscopic observations of dufour et al .
( 2008c ) .
indeed , it was found that the strong carbon lines seen in the optical spectrum of sdss j1426 + 5752 feature zeeman splitting , a structure that could not be seen in the original noisy sdss spectrum .
the observed splitting between the @xmath6 and @xmath7 components implies a large scale magnetic field of about 1.2 mg .
hence , if sdss j1426 + 5752 is really a pulsating star , it would be the first example of an isolated pulsating white dwarf with a large detectable magnetic field .
as such , it would be the white dwarf equivalent of a rapidly oscillating ap ( roap ) star .
the roap stars are main sequence ( or near main sequence ) magnetic a stars ( ap ) showing multiperiodic luminosity variations with periods in the range 5@xmath515 minutes caused by low - degree , high - order pressure - mode pulsational instabilities ( see , e.g. , kurtz 1990 ) . in view of the importance of sdss j1426 + 5752
, we carried out additional studies of that star .
in particular , we present and discuss in this paper the results of some 106.4 h of integrated light photometry gathered at the steward observatory 1.55 m kuiper telescope .
we also discuss the pros and the cons of the pulsational instabilities model versus those of the interacting binary model .
this includes a detailed search for possible radial velocity variations due to rapid orbital motion using mmt spectroscopy . at the end of the exercise ,
we unequivocally conclude in favor of the pulsation model .
our follow - up photometry was obtained with mont4k ( as in `` montral 4k@xmath84k camera '' ) , a new ccd camera designed and built at the steward observatory .
mont4k is a partnership between universit de montral and the university of arizona .
the instrument was designed primarily with differential time - series photometry in mind ( in the windowing mode ) , but the variety of filters available coupled with the excellent sensitivity of the chip and the large ( 9.7@[email protected]@xmath9 ) field of view make it ideal for many imaging projects
. it is used at the kuiper telescope on mount bigelow near tucson .
some details about the instrument can be found in randall et al .
( 2007 ) , but the interested reader will find more on the following web site : http://james.as.arizona.edu/ psmith/61inch / instrument s.html . at the outset , the relative faintness of sdss j1426 + 5752 posed a challenge for a small instrument such as the kuiper telescope , but we were pleasantly surprised at the quality of the light curves that we could actually obtain .
the time - series observations were taken on dark nights through a broadband schott 8612 filter , and an effective exposure time of 67 s ( on average ) was used as a compromise between the s / n and the need to sample adequately the luminosity variations as reported by montgomery et al .
altogether , we collected some 106.4 h of useful photometry over a 40 day span in the spring of 2008 .
this corresponds to a formal temporal resolution of 0.29 @xmath10hz and a modest duty cycle of 11.8% .
details of the observations are provided in table 1 .
cccc 2008 mar 31 & 6.6517743 & 399 & 7.88 + 2008 apr 01 & 7.6417324 & 468 & 8.73 + 2008 apr 03 & 9.6413232 & 466 & 8.69 + 2008 apr 04 & 10.6382536 & 449 & 8.52 + 2008 apr 05 & 11.6310329 & 223 & 4.15 + 2008 apr 06 & 12.6288716 & 487 & 9.05 + 2008 apr 09 & 15.6271095 & 364 & 6.55 + 2008 apr 10 & 16.6864942 & 108 & 1.99 + 2008 apr 11 & 17.7342303 & 328 & 6.11 + 2008 may 01 & 37.6670997 & 389 & 7.26 + 2008 may 02 & 38.6404921 & 424 & 7.88 + 2008 may
03 & 39.6412906 & 418 & 7.80 + 2008 may 04 & 40.6406385 & 275 & 6.29 + 2008 may 08 & 44.6459373 & 406 & 7.65 + 2008 may 10 & 46.6446166 & 420 & 7.82 + figure 1 shows the 15 nightly light curves that were obtained .
the original images were reduced using standard iraf reduction aperture photometry routines , except that we set the photometric aperture size individually for each frame to 2.0 times the fwhm in that image .
we computed differential light curves of sdss j1426 + 5752 on the basis of three suitable comparison stars well distributed around the target .
final detrending of the effects of differential extinction was made through a spline strategy .
a zoomed - in view of the light curve gathered on may 2 ( 2008 ) is provided in figure 2 .
again , in view of the magnitude of the target ( @xmath0 = 19.16 ) , the overall quality of the light curves is most gratifying .
we attribute this to the excellent sensitivity of the ccd and our optimized data pipeline .
the time - series photometry gathered for sdss j1426 + 5752 was analyzed in a standard way using a combination of fourier analysis , least - squares fits to the light curve , and prewhitening techniques ( see , e.g. , billres et al .
2000 for more details ) .
figure 3 shows the fourier amplitude spectrum of the full data set in the [email protected] mhz bandpass ( upper curve ) and the resulting transforms after prewhitening of two ( middle curve ) and three significant peaks ( lower curve ) .
these results first confirm the presence of a dominant oscillation with a period of 417.707 s and of its first harmonic at 208.853 s , in agreement with the report of montgomery et al .
note , in this context , that we did not detect the 2nd harmonic , again in agreement with their report , but also that we could not verify the presence of the 4th harmonic of the main periodicity as suggested in the data of montgomery et al .
( 2008 ) because our relatively large sampling time led to an effective nyquist frequency of 7.46 mhz , too small for picking up that high - frequency peak . on the other hand
, our results also reveal the presence of an additional significant oscillation with a period of 319.720 s as can be seen in the figure .
it is also quite likely that there are many other frequency components in the light curve of sdss j1426 + 5752 , independent oscillations or rotationally - split components for instance , but the sensitivity of our observations did not allow them to be detected .
the two upper curves in figure 4 provide a zoomed in view of the fourier amplitude spectrum in the vicinity of the 417.707 s peak before and after prewhitening .
likewise , the two lower curves give a similar view for the 319.720 s peak .
it is interesting to note that , within our measurement errors , the main peak at 417.707 s is a singlet . given that sdss j1426 + 5752 has a large scale magnetic field of some 1.2 mg
, one could have expected instead the presence of @xmath11 components due to magnetic splitting according to jones et al .
perhaps the multiplet components have much lower amplitudes than the main mode and are buried in the noise , or perhaps the field geometry is such that magnetic splitting can not be observed . either way , this remains an interesting curiosity .
cccc [email protected] & [email protected] & [email protected] & [email protected] + [email protected] & [email protected] & [email protected] & [email protected] + [email protected] & [email protected] & [email protected] & [email protected] + we thus were able to extract three distinct harmonic oscillations in the light curve of sdss j1426 + 5752 .
the basic characteristics of these oscillations are summarized in table 2 .
note that the phase is relative to an arbitrary point in time ; in our case , the beginning of the first run on ut 31 march 2008 .
the uncertainties on the period , frequency , amplitude , and phase of each oscillation as listed in the table were estimated with the method put forward by montgomery & odonoghue ( 1999 ) .
we point out , in this context , that the uncertainties on the amplitudes and phases obtained by our least - squares fits during the prewhitening stage were virtually the same as those derived with the montgomery & odonoghue ( 1999 ) method .
a basic quantity in that latter approach is the average noise level in the bandpass of interest , and we thus computed this mean value from the residual fourier transform ( the lower curve in fig .
3 ) spanning the [email protected] mhz interval .
the mean noise level in that range turned out to be 0.060% of the mean brightness of the star . in comparison ,
if we use the shuffling technique discussed by kepler ( 1993 ) for estimating the noise level , we find an almost identical value of 0.061% . in the context of being able to discriminate between the two possibilities put forward by montgomery et al .
( 2008 ) to explain the luminosity variations of sdss j1426 + 5752 pulsations versus interacting binary
our detection of a periodicity ( 319.720 s ) that is incommensurate with the periods of the main peak and its harmonics is potentially quite important ( see below ) .
hence , it is essential that its presence be well established .
formally , according to the results of table 2 , our detection of the 319.720 s periodicity is a 6.0 @xmath7 ( 0.288/0.048 ) result . in the more standard way ,
the derived amplitude is rather compared to the average noise level , in which case our detection would be seen as a 4.8 @xmath7 ( 0.288/0.060 ) result .
if we divide our runs into two `` seasons '' , i.e. , from march 31 through april 11 , and then from may 1 through may 10 , we find that the 319.720 s period is present in both sets of observations ( [email protected] s and [email protected] s ) , at the level of 4.2 @xmath7 according to the standard criterion .
we also verified explicitly that the 319.720 s periodicity is not present in the light curves of our comparison stars , thus ruling out a possible instrumental effect .
we note finally that montgomery et al .
( 2008 ) would not have been able to pick up the 319.720 s periodicity in their data assuming that it was present in their light curve with an amplitude comparable to our detection because their sensitivity was at least a factor of two lower than what we could achieve on a smaller telescope ( 1.55 m versus 2.1 m ) at a brighter site but at the price of much longer observations ( 106.4 h versus 7.8 h ) .
we investigated the stability of the amplitude and phase of each of the three frequencies we extracted from the light curve of sdss j1426 + 5752 by performing nightly measurements . a clue about possible amplitude variations on a daily timescale is first provided by figure 5 in which we show a montage of the nightly fourier amplitude spectra .
considering the main peak ( 2.3950 mhz , 417.707 s ) , the figure does suggest some possible amplitude variations .
however , things are a lot less suggestive for the two other peaks given the level of noise and their relatively small amplitudes .
a more quantitative and standard way of measuring the nightly amplitudes ( and phases ) is to fix the periods at their values given in table 2 and simultaneously perform least - squares sine fits with these periods for each nightly run .
the output are nightly amplitudes and phases with formal estimates of their uncertainties .
it is interesting to point out that the formal estimates of the uncertainties on the amplitudes and phases that came out of our least - squares exercise were , again , essentially the same as those obtained through the method of montgomery & odonoghue ( 1999 ) , which we explicitly used after the fact as a verification .
figure 6 summarizes our results in the case of the main periodicity found in the light curve of sdss j1426 + 5752 .
note that we explicitly excluded the values obtained from the short runs gathered on april 5 and april 10 because the associated uncertainties are significantly larger than those shown in the figure .
hence , the top panel in figure 6 displays the amplitudes of the 417.707 s peak along with their formal 1 @xmath7 uncertainties for 13 nightly runs .
the central dotted horizontal line represents the weighted average of the nightly amplitudes and the horizontal lines above and below the average value give the 1 @xmath7 uncertainty on that value . in the case of the phase ( lower panel of fig .
6 ) , a similar procedure was used , except that the average value was shifted to zero ( since the phase is arbitrary ) and ultimately expressed in units of cycle .
taken at face value , the results summarized in figure 6 do suggest some variations in amplitude and in phase over timescales of days .
for instance , on march 31 , we find a `` low '' amplitude of [email protected]% for the dominant periodicity at 417.707 s , while we find a `` high '' amplitude of [email protected]% on may 2 . in this context
, it may be appropriate to recall that amplitude modulation associated with the rotation of the star is typically seen in roap stars ( see , e.g. , kurtz 1990 ) .
this phenomenon ( and many others ) observed in roap stars has been explained within the framework of the very successful oblique pulsator model . in that model ,
the pulsations align themselves along the magnetic field axis which is itself inclined with respect to the rotation axis of the star .
the viewing aspect of the pulsations thus changes periodically with rotation , which produces amplitude modulation of a given mode . in the case of sdss j1426
+ 5752 , we do not yet know if its large scale magnetic field is aligned with the rotation axis .
although the material presented in this subsection is suggestive of possible amplitude and phase modulations with timescales of days for the dominant 417.707 s period , it is simply not possible at this stage to be certain about their reality . if , for example , we were to double the uncertainties on the derived nightly amplitudes and phases ( assuming that our least - squares approach or the method of montgomery & odonoghue 1999 underestimates the true errors by that factor ) , then we could only conclude that the nightly amplitudes and phases of the 417.707 s periodicity we extracted from the light curve of sdss j1426 + 5752 do not vary within our measurement errors .
things are even worse for the lower amplitude 319.720 s and 208.853 s oscillations in that the uncertainties on their nightly amplitudes and phases prevent us from concluding with any certainty about possible modulations although there are hints of variations . here then is a classic case of `` more observations are needed '' .
we have followed up on the remark made by montgomery et al .
( 2008 ) that the folded pulse shape of sdss j1426 + 5752 is different from that of pulsating white dwarfs and rather shows similarities with that of am cvn , the prototype of helium - transferring double degenerate binaries .
the top panel of figure 7 shows our 106.4 h long light curve of sdss j1426 + 5752 folded on the period of 417.707 s. to reach a decent s / n , we distributed the folded amplitudes in 10 different phase bins , each containing 572 points on average .
the error bars about each point in the folded light curve correspond to the errors of the mean in each bin .
given that the first harmonic of the 417.707 s periodicity in the light curve has a very high amplitude , about 30% of that of the main peak ( see table 2 ) , it is not surprising that the pulse shape illustrated in the top panel of figure 7 is highly nonlinear .
along with this , another striking characteristic with respect to known pulsating white dwarfs is the fact that it boosts a relatively flat maximum and a sharp minimum ( see montgomery et al .
2008 and the examples below ) . in the middle panel of figure 7
, we again display our folded light curve of sdss j1426 + 5752 , but only after having prewhitened the data of the first harmonic ( 208.853 s ) of the main periodicity .
if higher order harmonics have negligible amplitudes , and if other modes do not interfere in the folding process , the pulse shape in the middle panel should be that of a perfect sinusoid with an amplitude equal to that of the 417.707 s oscillation described in table 2 .
this is precisely what the dotted curve shows .
taking into account the uncertainties , the match between the folded pulse shape and the template is essentially perfect . in the lower panel
, we plotted the same template ( dashed curve ) corresponding to the 417.707 s sinusoid .
taking properly into account the phase difference , we also plotted a sinusoid ( dotted curve ) with the defining characteristics of the 208.853 s periodicity as given in table 2 .
the sum of these two sine waves gives the solid curve , the overall nonlinear pulse shape associated with the 417.707 s oscillation .
we reported this model pulse shape in the upper panel of figure 7 ( now as a dotted curve ) so that a direct comparison can be made with the observations .
again , within our measurement errors , the agreement is nearly perfect .
one can see that the relatively flat maximum and sharp minimum in the pulse shape is due to the fact that the first harmonic of the main oscillation falls nearly in phase at the minimum and in antiphase near the maximum of the main sinusoid .
this is particularly well illustrated in the lower panel of the figure . for comparison purposes
, we carried out similar folding exercises using representative light curves from the archives that one of us ( g.f . ) built up over the years using lapoune at the 3.6 m canada - france - hawaii telescope ( cfht ) .
lapoune is a portable three - channel photometer that uses photomultiplier tubes as detectors .
the archived light curves are integrated `` white light '' data and were generally obtained with a sampling time of 10 s. they include a short ( 2.04 h ) light curve of am cvn itself , originally taken as a test as part of a multisite campaign carried out on that star ( provencal et al .
1995 ; solheim et al .
1998 ) . in a format identical to that of figure 7 , figure 8 summarizes the results of our calculations after having folded that light curve of am cvn on the period of 521.105 s , which is the dominant photometric periodicity in that star ( see provencal et al .
although the model pulse shape ( dotted curve in upper panel and solid curve in lower panel ) based on the superposition of two sinusoids ( the 521.105 s oscillation and its first harmonic ) is far from perfect , it is nevertheless sufficient for illustrating the fact that am cvn does show a main pulse shape with a rounded top and a sharp bottom .
the lower panel of figure 8 explains why : the first harmonic ( dotted curve ) of the 521.105 s oscillation shows a relatively large amplitude compared to that of the main component , and it tends to be in phase and in antiphase at the minimum and maximum , respectively , of that component .
there is a significant offset however , between the extrema of the 521.105 s sinusoid and those of its first harmonic , and this largely explains the asymmetric shape of the pulse in this particular light curve .
in contrast to this behavior , known pulsating white dwarfs with nonlinear light curves ( the large amplitude ones ) quite generally display pulse shapes that have flatter minima and sharper maxima .
an example of this is provided by figure 9 which refers to the large amplitude zz ceti star pg 2303 + 242 , a pulsator with a light curve dominated by a periodicity at 712.250 s. in the present case , the model pulse shape based on only two components ( the dominant mode and its first harmonic ) does a very good job at explaining the folded light curve , as can be appreciated in the top panel of figure 9 .
a second example is provided by another large amplitude zz ceti star , gd 154 , as illustrated in figure 10 . in that case , we retained also the second harmonic along with the main oscillation ( 1186.085 s ) and its first harmonic in the construction of the model pulse shape .
although the top panel of the figure clearly demonstrates that the model could be improved ( gd 154 has a complicated multiperiodic light curve with higher - order harmonic terms ; see , e.g. , fig .
23 of fontaine & brassard 2008 ) , the results presented here are sufficient to make our point about the general shape of a light pulse in a large amplitude pulsating white dwarf .
another interesting example is shown in figure 11 .
it concerns balloon 090100001 , which is not a pulsating white dwarf at all , but instead belongs to the family of hot b subdwarf pulsators ( and see fontaine et al .
2006 for a brief review on these stars , if interested ) .
the light curve of that star , the largest amplitude variable of that type currently known , is dominated by a main pulsation ( the fundamental radial mode according to van grootel et al . 2008 ) with a period of 356.194 s. our model pulse shape , based on the superposition of the main periodicity and of its first two harmonics , gives a perfect fit to the folded light curve .
what is common in these examples , and what is generally true for large amplitude pulsating white warfs and hot subdwarfs , is that the harmonic components tend to be in phase at light maximum .
this produces relatively sharp maxima and flat minima .
in contrast , the first harmonic tends to be in antiphase with the main periodicity at light maximum in the light curve of sdss j1426 + 5752 and am cvn , and this now produces pulse shapes with rounded tops and sharp minima .
montgomery et al .
( 2008 ) interpreted this as evidence that the luminosity variations observed in sdss j1426 + 5752 could be caused , not by pulsational instabilities ( their preferred possibility ) , but by photometric activity in a carbon - transferring analog of am cvn .
we show below , however , that the pulsation hypothesis provides the better explanation . on the basis of our photometry , can we argue in favor of one or the other of the two possibilities put forward by montgomery et al .
( 2008 ) to account for the luminosity variations seen in sdss j1426 + 5752 ? to begin with , it is well known that flickering
incoherent light bursts arising on short timescales is a telltale sign of mass transfer in an interacting binary system ( warner 1995 ) . flickering is actually observed in the cfht / lapoune light curve of am cvn as can be readily seen in figure 1 of provencal et al .
( 1995 ) where the light curve has been displayed for the first time .
the sampling time was 10 s , sufficiently short to pick up some flickering .
however , when degraded to a sampling time of 60 s , flickering all but disappears in the light curve of am cvn .
we conclude from this that we could not have detected flickering in our photometric observations of sdss j1426 + 5752 if present , because of the large sampling time we used .
hence , we can not use this as a diagnostic . on the other hand ,
the light curves of am cvn systems are known to be unstable and irregular .
solheim et al .
( 1998 ) discuss this phenemenon in some detail in relation to their figure 2 , where the cfht / lapoune light curve ( binned in 40 s data points ) is again displayed .
the authors comment that , in spite of the fact that the cfht / lapoune light curve is essentially noise - free , it is irregular compared to that of isolated pulsating white dwarfs . for instance , the troughs in the light curve are irregularly spaced in time .
random flickering is blamed for this state of affairs .
in contrast to this , and not withstanding the noise , the light curves we have gathered ( see fig .
1 ) have kept the same appearance and regularity over a six week period . however , it is not clear if this is really significant since flickering is not coherent over long timebases .
the light curves of am cvn systems are generally dominated by a principal periodicity along with a suite of several harmonics of that dominant oscillation ( warner 1995 ) .
montgomery et al .
( 2008 ) reported the detection of the 417 s oscillation in sdss j1426 + 5752 along with its first harmonic and possibly also the 4th harmonic , but not the 2nd or 3rd one .
the detection of a period ideally several periods that would be incommensurate with such series of harmonically related oscillations would go against the interacting binary hypothesis and would favor the pulsations alternative .
this would be the best evidence for pulsations according to barlow et al .
our detection of the 319 s oscillation ( see subsection 2.2 ) therefore goes a long way in that direction . what about the pulse shape argument ?
we do not know at this stage why the folded light curve of sdss j1426 + 5752 features a maximum that is flatter than its minimum , contrary to what is observed in large amplitude pulsating white dwarfs and hot subdwarfs .
however , the fact that the pulse shape is unusual does not rule out by any means the possibility that pulsations are involved .
in fact , nature provides us with explicit examples of isolated pulsating stars with pulse shapes that qualitatively resemble that seen in sdss j1426 + 5752 in that they display rounded maxima and sharper minima .
these are some of the roap stars with the largest amplitudes that show harmonics of dominant modes in their light curves .
an excellent case is that of hr 3831 which exhibits a pulse shape with a rounded top and a sharp bottom as can be seen in figure 1 of kurtz , shibahashi , & goode ( 1990 ) .
the fast rise , rounded maximum , slower decline , and pointed minimum are very reminescent of what is observed in the light curve of the am cvn system cr boo in its high state as can be seen in figure 6 of warner ( 1995 ) for instance . yet , hr 3831 is a genuinely pulsating star .
another example is provided by the roap star hd 99563 studied by handler et al .
( 2006 ) . as can be seen in their figure 1 ( and see also the discussion in the text ) , the pulsations of that star show again flatter light maxima than minima .
the authors comment on the phasing of the harmonics with respect to the main mode in their paper .
sdss j1426 + 5752 has one thing in common with roap stars , and that is a large scale magnetic field sufficiently important in both cases to disrupt the atmospheric layers and influence the pulsations there .
it would therefore not surprise us if the magnetic field were responsible for the unusual pulse shape ( relative to nonmagnetic pulsating stars ) observed in the light curves of sdss j1426 + 5752 and large amplitude roap stars .
be that as it may , in view of the very existence of roap stars such as hr 3831 and hd 99563 , the argument that the pulse shape of sdss j1426 + 5752 is different from those seen in ( nonmagnetic ) pulsating white dwarfs can not be held against the pulsations interpretation .
dufour et al . ( 2008c ) presented follow - up spectroscopy of sdss j1426 + 5752 including , in particular , an optical spectrum obtained with the blue channel spectrograph at the mmt using a 500 line mm@xmath13 grating with a 1@xmath14 slit , resulting in a @xmath43.6 fwhm spectral resolution over a wavelength range of 3200@xmath56400 .
a total of 17 individual spectra were obtained over a timebase of 3.37 h , resulting in an overall exposure time of 180 minutes .
part of the final combined spectrum , taken from dufour et al .
( 2008c ) , is reproduced in figure 12 ( upper curve ) .
it shows a s / n@xmath475 per pixel at 4500 .
the most prominent features are cii lines which clearly show zeeman splitting in their cores . as a comparison , we also plotted the available sdss spectrum for the star ( lower curve ) .
if sdss j1426 + 5752 is truly a double degenerate carbon analog of am cvn systems , then one should be able to pick up radial velocity variations associated with orbital periods in the range 1000@xmath53000 s as found in most such systems .
this excludes , of course , the improbable configuration of having the orbital plane at nearly 90 degrees with respect to the line - of - sight .
the sampling and the baseline of 3.37 h used during the mmt observations of dufour et al .
( 2008c ) are well suited to search for such radial velocity variations .
in addition , the combination of the blue channel spectrograph and the mmt is well known to provide a very stable platform for radial velocity measurements , and one of us ( e.m.g . )
has developed over the years a large mmt program dedicated to radial velocity measurements in hot subdwarf stars .
we therefore went back to the mmt spectroscopic data and searched for possible radial velocity variations using the tools and expertise developed at steward observatory .
we provide some details on the procedure followed .
we first reinterpolated the 17 individual spectra of sdss j1426 + 5752 onto a logarithmic wavelength scale with identical starting and ending wavelengths ( 3215@xmath56357 ) .
after manually removing cosmic ray signatures from each spectrum , we fitted the continuum , divided by the fit , and subtracted 1.0 to get a flattened continuum with a mean level of zero .
we median - filtered the 17 resulting spectra into a single spectrum to create a higher s / n radial velocity template .
radial velocity cross - correlations of the individual continuum - removed spectra relative to the template spectrum were performed using the double precision version of iraf s fxcor task , fitting the cross - correlation peak with a gaussian .
a ramp filter was used prior to the cross - correlation to select the optimal range in fourier space , with adopted values of 125 , 275 , 1480 , and 1490 , respectively , for the cut - on , full - on , cut - off , and full - off points .
the high frequency noise cut - off corresponds to 2.75 pixels , which is slightly better than the instrumental resolution of 3.0 pixels , as determined by the fwhm s of the hearne comparison arc lines .
the low - frequency cut - on , corresponding to 33 pixels , is slightly wider than the observed fwhm ( 30 pixels ) of the strongest absorption lines in the spectrum .
the essential result that came out of this exercise is that the velocities are constant to within 7.1 km s@xmath13 ( the standard deviation ) , which is about 1/32 of the velocity resolution , and there is no sign of any velocity trend over the 3.37 h of the observations .
we point out that 1/32 of a resolution element represents quite a high level of accuracy for spectra at this s / n . in comparison ,
the typical radial velocity semi - amplitude expected in a am cvn system is @xmath15 sin @xmath16 km s@xmath13 .
this estimate uses the following representative values : @xmath17 for the mass of the helium degenerate primary , @xmath18 for the mass of the helium degenerate or semi - degenerate donor , and @xmath19 s for the orbital period , as can be inferred in the reviews of warner ( 1995 ) and nelemans ( 2005 ) and references therein .
actually , if we assume that the dominant photometric period of 417 s is approximately equal to the orbital period in a putative sdss j1426 + 5752 carbon system equivalent of am cvn ( this is the case for the known systems of the type , except for am cvn itself ) , then the estimate of the velocity semi - amplitude goes up to @xmath20 sin @xmath16 km s@xmath13 .
if , as in am cvn itself , the orbital period is rather approximately equal to twice the dominant photometric periodicity of 417 s ( 521 s in am cvn ) , the velocity semi - amplitude takes on the value @xmath21 sin @xmath16 km s@xmath13 . hence , unless the inclination of the orbital plane is quite low , this argues strongly against the interacting binary hypothesis as the explanation for the luminosity variations observed in sdss j1426 + 5752 .
along with the fact that we do not detect radial velocity variations , the available spectroscopic data on sdss j1426 + 5752 can further be used to build up the case against the interacting binary model . in
a am cvn - type system in its high state , most of the light does not come from the photosphere of the helium degenerate primary , but rather from an optically thick accretion disk orbiting around that star . according to warner ( 1995 ) , the hei lines seen in absorption have quite different profiles from those observed in isolated helium - atmosphere ( db ) white dwarfs .
their profiles are asymmetric , and they vary both in shape and depth , contrary to the symmetric and stable lines seen in db white dwarfs . in this context , we wish to point out that the available spectroscopy on sdss j1426 + 5752 shows that the broad ( carbon ) absorption lines are symmetric ( see , e.g. , fig .
furthermore , the spectrum appears quite stable , at least over a timescale of 20 h , which is the time lapse between the observations of sdss j1426 + 5752 carried out at the mmt and those gathered at the keck i telescope ( and see dufour et al .
the spectrum does not show any sign of accretion disk activity .
however , it does show clear zeeman splitting in the line cores , a feature that would presumably be washed away if sdss j1426 + 5752 were part of a interacting binary system because of the rapid orbital motion .
zeeman splitting is seen in both the mmt and keck spectra , and the spacings between the @xmath6 and @xmath7 components are the same within the measurement errors in both spectra . in contrast , zeeman splitting has never been reported for am cvn systems . finally , we beg to disagree with montgomery et al .
( 2008 ) who suggested that the broad absorption lines seen in a hypothetical carbon analog of am cvn could mimic those seen in the photosphere of an isolated white dwarf . and
indeed , it has already been established ( see , e.g. , odonoghue & kilkenny 1989 or warner 1995 ) , that the hei absorption lines seen in am cvn systems , although relatively broad by main sequence standards , have strengths that are more akin to those observed in log @xmath0 = 6 hot subdwarf stars than log @xmath0 = 8 db atmospheres .
this reflects the fact that they are formed in the inner region of the accretion disk orbiting the white dwarf primary where the physical conditions are similar to those encountered in log @xmath0 = 6 atmospheres .
in contrast , the spectral fits carried out by dufour et al .
( 2008b ) for hot dq stars totally rule out equivalent surface gravities of log @xmath0 = 6 , as the line strengths observed in these stars bear the clear signature of log @xmath0 = 8 environments , even log @xmath0 = 9 in the case of sdss j1426 + 5752 ( notwithstanding the presence of a magnetic field ) . in brief , the spectrum of sdss j1426 + 5752 can not be confused with that of the accretion disk in a hypothetical carbon analog of am cvn .
we have presented an analysis based on follow - up photometric and spectroscopic observations of the faint but highly interesting star sdss j1426 + 5752 . on the photometric front
, we carried out a campaign in integrated light over a baseline spanning some 40 days .
we used the kuiper / mont4k combination at the steward observatory mount bigelow station near tucson .
altogether , we acquired some 106.4 h of useful photometry during the campaign .
our analysis of this data set confirms that the light curve of sdss j1426 + 5752 is dominated by a periodicity at 417.707 s along with its first harmonic ( 208.853 s ) as found originally , but with less accuracy , by montgomery et al .
( 2008 ) in their discovery paper .
in addition , due to the higher sensitivity achieved in our campaign , we uncovered the presence of a new oscillation with a period of 319.720 s , a 4.8 @xmath7 result using the standard detection criterion ( a 6.0 @xmath7 detection at the formal level ) .
the characteristics of these three oscillations are summarized in table 2 .
we investigated the stability of the amplitude and phase of each of the three periodicities extracted from the light curve of sdss j1426 + 5752 .
our results suggest possible variations over timescales of days , and this is particularly true for the dominant 417.707 s periodicity .
however , this needs to be confirmed with higher s / n data since observing a @xmath0 = 19.16 star in integrated light photomery with a small telescope such as the kuiper remains challenging . on the basis of our current data , we can not be absolutely certain of the reality of the suggested amplitude and phase modulations in the light curve of sdss j1426 + 5752 .
we followed up on the suggestion made by montgomery et al .
( 2008 ) that the luminosity variations in sdss j1426 + 5752 may not be caused by pulsational instabilities , but rather be associted with photometric activity in a carbon - transferring analog of am cvn .
if true , sdss j1426 + 5752 would represent the prototype of a new class of cataclysmic variable . since their argument hinges on the shape of the light curve folded on the dominant periodicity of 417.707 s , we further exploited our 106.4 h data set in that direction . using this and archived light curves of known pulsating stars , we found , in agreement with montgomery et al .
( 2008 ) , that the folded pulse shape of sdss j1426 + 5752 is unusual compared to those of large amplitude pulsating white dwarfs and hot subdwarfs .
however , in view of the existence of isolated pulsators such as the roap stars hr 3831 and hd 99563 exhibiting pulse shapes with flatter light maxima than minima ( the opposite of what is seen in large amplitude pulsating white dwarfs ) , we emphasize that the pulse shape argument _ can not _ be used to discriminate against the pulsations interpretation .
since sdss j1426 + 5752 and these roap stars and others share the common property of having a magnetic field sufficiently strong to affect the pulsations in their atmospheric layers , we suggest instead that the magnetic field may be responsible for the different pulse shape as compared to those of ( nonmagnetic ) pulsating white dwarfs . on the other hand , arguments against the interacting binary hypothesis can be put forward .
for instance , the light curves we gathered have shown to be quite regular and stable over at least a six week period , a behavior that is not commonly observed in am cvn systems .
our discovery of a periodicity ( 319.720 s ) that is not harmonically related to the dominant oscillation of 417.707 s also goes against the interacting binary proposal .
likewise , on the spectroscopic front this time , our detailed radial velocity analysis of the available mmt spectrocopy has revealed that the velocities are constant to within 7.1 km s@xmath13 over a period of 3.37 h. this is to be compared with expected velocity semi - amplitudes in the range 80@xmath5100 sin @xmath16 km s@xmath13 .
furthermore , the spectrum of sdss j1426 + 5752 , unlike those of am cvn systems , exhibits well defined symmetric absorption lines and it has proven stable over at least a 20 h period . it shows sharp zeeman splitting in the line cores , indicative of the presence of a large scale magnetic field of 1.2 mg , a property never observed in a am cvn system . finally , the line strengths indicate a physical environment appropriate for log @xmath0 = 8 atmospheres , unlike that found in the inner region of an accretion disk around a white dwarf and more akin to what is found in a log @xmath0 = 6 stellar atmosphere . it should also be pointed out that the interacting binary hypothesis does not account well at all for the existence of the family of hot dq white dwarfs as a whole since , among other things , not all of them exhibit luminosity variations ( and see dufour et al .
in contrast , the single - star evolutionary scenario originally proposed by dufour et al .
( 2007 ) appears quite viable as demonstrated recently by althaus et al .
( 2009 ) . also , the nonadiabatic calculations of fontaine et al .
( 2008 ; also dufour et al .
2008b ) do predict a mixture of pulsators and nonpulsators in the hot dq population .
in addition , according to montgomery et al .
( 2008 ) themselves , citing the work of benz et al .
( 1990 ) , rasio & shapiro ( 1995 ) , and piersanti et al .
( 2003 ) , carbon - transferring am cvn - type systems are not expected to exist .
putting all these arguments together , we conclude with high confidence that , of the two possibilities put forward by montgomery et al .
( 2008 ) to account for the luminosity variations seen in sdss j1426 + 5752 , pulsational instabilities is the correct cause .
this is also the preferred solution of montgomery et al .
( 2008 ) .
randall , s.k . ,
green , e.m .
, van grootel , v. , fontaine , g. , charpinet , s. , lesser , m. , brassard , p. , sugimoto , t. , chayer , p. , fay , a. , wroblewski , p. , daniel , m. , story , s. , & fitzgerald , t. 2007 , a&a , 476 , 1317 | we present a follow - up analysis of the unique magnetic luminosity - variable carbon - atmosphere white dwarf sdss j142625.71 + 575218.3 .
this includes the results of some 106.4 h of integrated light photometry which have revealed , among other things , the presence of a new periodicity at 319.720 s which is not harmonically related to the dominant oscillation ( 417.707 s ) previously known in that star . using our photometry and available spectroscopy
, we consider the suggestion made by montgomery et al .
( 2008 ) that the luminosity variations in sdss j142625.71 + 575218.3 may not be caused by pulsational instabilities , but rather by photometric activity in a carbon - transferring analog of am cvn .
this includes a detailed search for possible radial velocity variations due to rapid orbital motion on the basis of mmt spectroscopy . at the end of the exercise ,
we unequivocally rule out the interacting binary hypothesis and conclude instead that , indeed , the luminosity variations are caused by @xmath0-mode pulsations as in other pulsating white dwarfs .
this is in line with the preferred possibility put forward by montgomery et al .
( 2008 ) . |
nucleons in light nuclei have typical momenta that are small compared to the characteristic qcd scale of 1 gev . at these low momenta , qcd
can conveniently be represented by a hadronic theory containing all possible interactions consistent with the qcd symmetries .
effective field theory ( eft ) provides a controlled framework for exploiting the separation of scales in nuclei .
it is crucial to formulate a power counting that justifies a systematic truncation of the lagrangian leading to observables with the desired accuracy .
nuclei offer a non - trivial challenge because one wants such a perturbative expansion in addition to the non - perturbative treatment of certain leading operators , which is required by the existence of shallow bound states . by now , mainly few - body systems have been studied within eft , and , while much remains to be understood , many successes have been achieved @xcite .
similar to other approaches , the extension of efts to larger nuclei faces computational challenges @xcite . as a first step in this extension
, we specialized to very low energies where clusters of nucleons behave coherently @xcite . even though many interesting issues of nuclear structure are by - passed , we can describe anomalously shallow ( `` halo '' or `` cluster '' ) nuclei and some reactions of astrophysical interest .
since they are strongly bound , alpha particles play a central role in this framework .
many nuclear states have energies close to thresholds for break - up into alpha particles and nucleons , the most famous being the excited ( `` hoyle '' ) state of @xmath2c near the triple - alpha ( @xmath3 ) threshold .
these states should be describable within the halo / cluster eft , which is formulated with contact interactions among nucleon ( @xmath4 ) and alpha - particle ( @xmath5 ) fields . together with the @xmath6 interaction ,
the @xmath0 interaction is an important input for such calculations .
while we have studied the @xmath6 interaction elsewhere through both neutron - alpha ( @xmath7 ) @xcite and proton - alpha ( @xmath8 ) @xcite scattering , we focus here on @xmath0 scattering .
consideration of this system is required before tackling other states with two or more alpha particles , such as @xmath9be and @xmath2c .
the internal alpha - particle dynamics is characterized by an intrinsic momentum scale @xmath10 associated with the binding mechanism .
a naive guess is that this scale is set by the pion mass @xmath11 mev .
the @xmath0 interaction consists of the long - range photon exchange and short - range strong interactions . at low energies ,
the latter can be represented by contact interactions .
the central issue is the relative importance of these contributions .
the coulomb interaction is non - perturbative for momenta smaller than around @xmath12 , where @xmath13 is the fine - structure constant , @xmath14 the reduced mass of the system , and @xmath15 , @xmath16 , the electromagnetic charge of the particles . here
@xmath17 and @xmath18 in terms of the mass and charge of the alpha particle , respectively , so @xmath19 mev . at momenta
much below 100 mev , the deviation from pure - coulomb @xmath0 scattering is dominated by the @xmath20 wave @xcite . the large near - threshold @xmath20-wave phase shift has been interpreted as resulting from a ( @xmath21 , @xmath22 ) = ( @xmath23 , 0 ) state @xcite at an energy @xmath24 mev above threshold in the center - of - mass frame , with a tiny width @xmath25 ev .
the momentum corresponding to this @xmath26be state sets a smaller scale @xmath27 mev , which must arise from the larger underlying scale @xmath10 by a fine - tuning of the parameters of the underlying theory . in the halo eft ,
our goal is not to explain the mechanism of this fine - tuning , but instead to exploit its existence in order to describe @xmath5-cluster systems at low energies .
we seek a description of these systems in an expansion in powers of the small ratio @xmath28 .
power counting is dependent on how the various parameters scale with @xmath29 and @xmath10 .
the physics of the low - energy @xmath20 state is conveniently discussed in the language of a dimeron field @xcite with the quantum numbers of the low - energy composite state .
this field is characterized in leading order by a fine - tuned mass @xmath30 and a non - derivative coupling @xmath31 to the @xmath0 state . in subleading orders more complicated couplings
appear .
it is not immediately obvious how the fine - tuned mass @xmath30 relates to @xmath29 .
the simplest assumption is @xmath32 @xcite . in the absence of coulomb interactions
, this reproduces the leading term in the effective - range expansion , and one has a shallow real or virtual bound state with a typical momentum @xmath33 .
strong interactions are non - perturbative for momenta of order @xmath34 and larger .
higher - order terms in the effective - range expansion appear as subleading corrections .
this scenario is appropriate for @xmath20-wave @xmath35 scattering at momenta below @xmath36 @xcite . for @xmath37 scattering , @xmath38 mev and @xmath39 mev .
the coulomb interaction can be included non - perturbatively in a straightforward way @xcite , providing calculable contributions plus a renormalization of @xmath40 .
the situation in @xmath0 scattering is somewhat different .
the extremely low energy of the @xmath20-wave resonance suggests that a smaller @xmath30 might be necessary .
an alternative fine - tuning assumes thus that @xmath41 .
such scaling has already appeared in @xmath42-wave @xmath6 scattering @xcite , and has striking consequences in @xmath20-wave @xmath0 scattering . in the absence of the coulomb interaction ,
the leading contribution for momenta @xmath43 comes entirely from the unitarity term @xmath44 in the inverse amplitude . to this order
, the @xmath26be system would be at the so - called unitary limit , exhibiting conformal invariance @xcite , and @xmath45c would acquire an exact efimov spectrum @xcite .
as we will discuss later , this exact pattern is modified by the coulomb interaction , which breaks scale invariance and thus moves the @xmath26be ground state away from threshold .
we will see that in leading order we can describe the ground state , and higher - energy , scattering data can be systematically accounted for in higher orders .
at this point , we simply fit the eft parameters to data . in the future , many - body methods @xcite should allow the calculation of these parameters from the underlying efts , which in turn will have their parameters obtained from lattice qcd @xcite . to the extent that the halo / cluster eft is built on fields for clusters , it is closely related to phenomenological few - body cluster models @xcite .
the latter have had many applications , ranging from the general analysis of halo structures @xcite to the successful confrontation with data for specific processes , such as decays of @xmath2c resonances @xcite .
the emphasis here is instead on a systematic expansion of the most general interactions allowed by the symmetries of qcd . in this paper
, we study the alternative expansion scenario and perform a detailed comparison to phase - shift data for the @xmath0 system .
as we will see , among the many findings our new method entails are a new expansion for the @xmath0 amplitude around the resonance , which differs from the effective - range expansion ; the existence of inconsistencies between recent resonance measurements and old scattering data ; a more precise determination of low - energy parameters ; and an extraordinary amount of fine - tuning .
all these findings could have implication in the study of other alpha - cluster systems .
the paper is organized as follows . in sec .
[ sec : npc ] , we briefly discuss the eft formulation of the two - body system with coulomb interactions , with some details relegated to app .
[ app : ceft ] .
the proposed power counting is discussed in sec .
[ sec : pc ] . in sec .
[ sec : fits ] , we present results of our fits to the @xmath23 resonance parameters and @xmath0 phase shifts , and discuss the required fine - tunings .
our conclusions are presented in sec .
[ sec : conc ] .
we start with a summary of some basic ideas in halo eft @xcite , extended to include the main equations needed to deal with coulomb interactions .
these equations are a straightforward generalization of kong and ravndal s formalism @xcite to include dimeron fields .
more details can be found in app .
[ app : ceft ] . in sect .
[ sec : pc ] we discuss the significant differences in scales between alpha - cluster systems and the @xmath37 system considered by kong and ravndal @xcite .
we consider here the scattering of two alpha particles of charge @xmath46 and reduced mass @xmath47 , at a center - of - mass ( cm ) energy @xmath48 .
we want to build an eft that provides a controlled expansion for observables at momenta around the @xmath23 resonance , @xmath49 mev @xmath50 .
the energy of the resonance , @xmath24 mev , is much smaller than either the alpha - particle excitation energy @xmath51 mev or any energy set by pion exchange , such as two - pion exchange between alpha particles or one - pion exchange among nucleons , @xmath52 mev .
thus , we expect an alpha particle to behave in a first approximation as a rigid entity ( `` core '' or `` cluster '' ) , which we represent by a scalar - isoscalar field @xmath53 .
the smaller effects from deformations and other core - structure properties are accounted for in a derivative expansion .
the breakdown momentum scale @xmath10 of this eft will be set by the lowest - energy degrees of freedom that are not incorporated explicitly in the lagrangian .
since these include the nucleons within the alpha particle which can be resolved with a momentum @xmath54 mev and the pions which can be resolved with momenta of the order of the pion mass @xmath11 mev a reasonable estimate is @xmath55 mev .
the eft provides an expansion of observables in powers of the ratio between the low - energy scales @xmath56 and @xmath29 and the high - energy scale @xmath10.be ground state .
this argument is based on an extension of the halo eft to include explicit pion fields .
in fact , it had been pointed out before @xcite that the long - range interaction of alpha particles is related to two - pion exchange , but in the context of nucleon scattering on an alpha particle built out of four nucleons .
since the pion mass is not much smaller than the typical nucleon momentum in the alpha particle , we see no rationale ( for realistic values of the quark masses ) for an expansion in powers of @xmath57 in a theory with `` elementary '' alpha particles . ] around the resonance @xmath58 , we expect an expansion parameter of the order of @xmath59 .
as the energy increases the expansion deteriorates and one should not expect the eft to be applicable to laboratory ( lab ) energies much above @xmath60 mev , a conservative estimate that corresponds to a cm momentum of about @xmath61 mev .
since at low energies only @xmath20 waves contribute significantly to @xmath0 scattering beyond pure coulomb scattering , we introduce an auxiliary scalar dimeron field @xmath62 with `` residual mass '' @xmath30 , which couples to two @xmath5 fields .
( observables are independent of the choice of fields . )
we thus start with the following strong - interaction effective lagrangian : @xmath63\phi + \sigma\,d^{\dagger}\bigg[i\partial_0+\frac{{{\mathbf \nabla}}^2}{4m_{\alpha } } -\delta\bigg]d+g\,\big[d^{\dagger}\phi\phi+(\phi\phi)^{\dagger}d\big ] + \ldots\ , , \label{eq : lolag}\end{aligned}\ ] ] where @xmath64 and @xmath31 is a coupling constant .
the `` @xmath65 '' represent interactions with higher derivatives .
the sign @xmath66 is used to match the sign of the effective range @xmath67 .
the existence of this sign merely reflects the non - physical , auxiliary character of the dimeron field @xcite . in momentum space ,
associated with the displayed bilinear terms are the @xmath5 propagator @xmath68 and the dimeron propagator @xmath69 other terms in the lagrangian ( [ eq : lolag ] ) can be treated as insertions of interaction vertices in feynman diagrams or corrections to the propagators ( [ eq : alprop ] ) and ( [ eq : dprop ] ) .
different assumptions about the dependence of @xmath30 on @xmath29 translate into different relative weights for the terms in the denominator of eq .
( [ eq : dprop ] ) , where @xmath70 . for example , in the present scenario all terms are of the same size , while in the @xmath35 case the energy and kinetic terms are of subleading order and the dimeron propagator reduces in leading order to @xmath71 .
( the latter form gives rise in the amplitude not to a resonance but to a bound - state pole @xcite at a momentum @xmath72 , where @xmath73 is given by eq .
( [ eq : ardef ] ) . ) in this section we illustrate the use of the lagrangian ( [ eq : lolag ] ) together with electromagnetic interactions to calculate @xmath0 scattering amplitudes .
we postpone a discussion of the relative importance of various terms until the next section .
electromagnetic interactions are introduced into the effective lagrangian ( [ eq : lolag ] ) in the standard way , that is , by both changing the derivatives into gauge - covariant ones , and introducing gauge - invariant interactions generated by the electromagnetic field strength . for practical calculations one usually works in a fixed gauge .
the coulomb gauge is very convenient , since it allows a clear separation between coulomb and transverse photons .
the former provides the leading electromagnetic interaction and is driven by the sommerfeld parameter @xmath74 with @xmath75 the fine - structure constant and @xmath76 the inverse of the `` bohr radius '' of the @xmath0 system .
this parameter is enhanced by the presence of @xmath14 in the numerator , which , as we are going to see , makes @xmath77 large around the resonance and requires a non - perturbative resummation of coulomb photons . on the other hand ,
transverse photons are suppressed @xcite by two powers of @xmath78 , and when in loops by extra powers of @xmath79 . since numerically @xmath80 , these interactions contribute to orders where they can be accounted for in perturbation theory beyond the precision we are working with .
they can become significant only at energies comparable to the excitation energy of the @xmath5 core .
the scattering amplitude for two particles interacting via coulomb plus a short - range interaction is given by @xcite ( for more details , see app .
[ app : ceft ] ) : @xmath81 where @xmath82 and @xmath83 are the pure - coulomb and the coulomb - distorted short - range amplitudes , respectively . considering only @xmath20-wave interactions ,
the coulomb - distorted short - range amplitude is parametrized in terms of the `` coulomb - corrected '' phase shift @xmath84 as @xmath85}\ , , \label{eq : tcstext}\ ] ] with the pure - coulomb phase shift @xmath86 given by eq .
( [ eq : sigma_l ] ) in the appendix and the sommerfeld factor @xmath87 the @xmath88 function is given by eq .
( [ eq : hdef ] ) . for real @xmath77
it can be expressed as @xmath89-\ln\eta+ \frac{i}{2\eta}c_{\eta}^2 \label{eq : h}\ ] ] in terms of the digamma function @xmath90 .
the other ( real ) term in the denominator of eq .
( [ eq : tcstext ] ) is the landau - smorodinsky @xmath91 function @xcite , @xmath92 at low energies it reduces to the effective - range expansion ( ere ) in the presence of coulomb interactions , @xmath93 \ , , \label{eq : kapprox}\ ] ] where @xmath73 , @xmath67 , @xmath94 , @xmath65 , are the scattering length , effective range , shape , @xmath65 , parameters . as shown in app .
[ app : ceft ] , the calculation using the eft lagrangian ( [ eq : lolag ] ) leads to @xmath95^{-1}+\dots\ , , \label{eq : tcsres}\end{aligned}\ ] ] where @xmath96 is the coulomb - renormalized mass parameter of the eft lagrangian , @xmath97 \right\}\ , , \label{eq : renorm}\ ] ] with @xmath98 the renormalization scale and @xmath99 the dimension of spacetime .
equation ( [ eq : tcsres ] ) is in the form of the coulomb - modified ere with scattering length and effective range given respectively by @xmath100 from eq .
( [ eq : tcs ] ) it is clear that the effect of a non - perturbative coulomb dressing of the strong - interaction amplitude apart from multiplying the amplitude by @xmath101 and from renormalizing the short - range parameters is to effectively replace the unitarity term @xmath44 by @xmath102 . in order to estimate the relative sizes of the various contributions to eq .
( [ eq : tcsres ] ) , we turn now to a discussion of power counting . as we are going to see , at each order
the eft in the two - body system is equivalent to a truncation of the coulomb - modified ere .
in this section we elaborate on the proposed power - counting scenario mentioned in the introduction .
we also give the @xmath0 amplitudes that are used in the next section for a comparison to phase - shift data . in the present strongly - tuned scenario
one considers @xmath103 and @xmath104 , with other couplings scaling with @xmath10 according to naive dimensional analysis . in this case
the contribution of the bare dimeron propagator to the scattering amplitude at momentum @xmath43 comes not only from the dimeron mass , but also from its kinetic term .
the simplest strong - interaction contribution is a bare dimeron propagator attached to initial and final external legs , which ( as seen by multiplying eq .
( [ eq : dprop ] ) by @xmath105 ) is of @xmath106 .
a non - coulomb bubble diagram times another dimeron propagator brings an extra factor of @xmath107 . as a consequence , the bubble - chain resummation is necessary .
the resulting denominator acquires the form @xmath108 .
interesting in this power counting is the fact that at momenta @xmath43 the first two terms are suppressed by @xmath28 compared to the last one .
therefore , all that is left at lo if the coulomb interaction is turned off is the unitarity term @xmath109 . in this limit , the @xmath26be system exhibits conformal invariance @xcite and the corresponding three - body system , @xmath45c , acquires an exact efimov spectrum @xcite .
this scenario is a possible realization of the unitary limit .
this picture is significantly modified when the coulomb interaction is turned on . in a non - relativistic system an @xmath110 potential breaks scale invariance and introduces the scale @xmath76 in the propagation of two charged particles . as we have seen in eq .
( [ eq : tcs ] ) , the unitarity term is modified .
the balance between strong - interaction terms and coulomb - modified propagation depends on both the intrinsic strong - interaction scale and @xmath76 .
we note that the scales are very different in @xmath0 than in @xmath37 scattering . while for the latter @xmath111 , for the @xmath0 system , @xmath112 mev and @xmath113 mev , so @xmath114 .
for momenta @xmath58 , we are therefore in the deep non - perturbative coulomb region .. ] this case corresponds to large values of @xmath77 , where the function @xmath115 is significantly different from the usual unitarity term @xmath116 . using stirling s series @xcite , @xmath117 in eq .
( [ eq : h ] ) gives @xmath118 the unitarity term is thus replaced by @xmath119 .
this term is now a factor @xmath120 smaller in magnitude than the unitarity term in the absence of coulomb , and comparable to the effective - range term coming from the dimeron kinetic term .
this can be captured automatically if we take @xmath121 , as it appears to be the case numerically .
if all higher - order parameters are natural , in the @xmath20 wave the shape - parameter term @xmath122 is down by @xmath123 compared to the effective - range term @xmath124 , and so should be each of the successive terms . on the other hand , the @xmath99-wave scattering-``length '' term @xmath125 is down by @xmath126 compared to the @xmath20-wave scattering length @xmath127 , and further energy dependence and higher angular momenta bring in further powers of @xmath123 . in this power counting , the coulomb - distorted short - range amplitude @xmath83 is given in lo by eq .
( [ eq : tcsres ] ) . including corrections in perturbation theory , we obtain up to nlo : @xmath128\ , , \label{eq : ampsc2}\end{aligned}\ ] ] where @xmath94 is given by a higher - derivative term in the lagrangian ( [ eq : lolag ] ) .
this corresponds to an expansion of the ere formula , @xmath129 for small @xmath94 . to this order ,
all waves higher than @xmath20 are purely coulombic .
higher orders can be calculated similarly .
equation ( [ eq : ampsc2 ] ) holds for generic momenta @xmath43 .
however , it fails in the immediate proximity of @xmath130 .
this situation is familiar from another application of the halo eft to a resonance @xcite .
the power counting works for @xmath43 except in the narrow region @xmath131 where the lo denominator nearly vanishes and a resummation of the nlo term , here associated with the shape parameter , is required .
as one gets closer to the resonance momentum @xmath130 , higher - order terms in the ere are kinematically fine - tuned as well .
this happens because the imaginary part of the denominator is exponentially suppressed by a factor @xmath132 and the real part gets arbitrarily small .
( for @xmath7 scattering , multiple kinematical fine - tunings are prevented in the @xmath133 wave by the presence of an imaginary term @xmath134 @xcite . )
this kinematical fine - tuning is not a conceptual problem . from the eft point of view , each new fine - tuning is equivalent to reshuffling the series and redefining the pole position . such a procedure works fine with a small number of kinematical fine - tunings , but is not practical in the present case .
a better alternative is to perform an expansion around the resonance pole position .
the situation here is analogous to the @xmath35 system , where we can choose to expand the amplitude around the bound - state pole @xcite rather than around zero energy . a great simplification results from the fact that the resonance lies in the deep coulomb regime , where eq .
( [ eq : happrox ] ) provides an accurate representation of @xmath88 up to the precision we are considering .
the real terms shown in eq .
( [ eq : happrox ] ) can be seen as an expansion in powers of @xmath135 .
of course in an asymptotic expansion at some point the remaining terms can no longer be expanded ; at that point the remainder should be treated exactly . in lowest orders ,
however , we can use the successive terms shown in eq .
( [ eq : happrox ] ) : numerically , the terms up to @xmath136 work to better than 3% for @xmath137 mev .
the expansion ( [ eq : happrox ] ) makes the physics around the resonance quite transparent . since the `` size '' of the resonance , @xmath138 , is much larger than the bohr radius @xmath139 ,
the coulomb interaction is effectively short ranged , and the real part of @xmath88 is ere - like .
this expansion matches well with the expansion for the landau - smorodinski function , which is in powers of @xmath140 . in @xmath83 , not only the @xmath141 terms , but also higher - order terms from strong and coulomb interactions have comparable sizes .
we can thus lump these terms together , defining @xmath142 and so on .
in this case we can rewrite up to nlo : @xmath143&= & -\frac{2\pi}{\mu}\,\frac{c_{\eta}^2\,e^{2i\sigma_0 } } { \tilde{r}_0(k^2-k_r^2)/2\!-\!\tilde{\cal p}_0(k^4\!-\!k_r^4)/4\ ! -\!ikc_{\eta}^2 } \nonumber\\[1mm]&= & -\frac{2\pi}{\mu}\,c_{\eta}^2\,e^{2i\sigma_0}\,\bigg [ \underbrace { \frac{1}{\tilde{r}_0(k^2-k_r^2)/2\!-\!ikc_{\eta}^2 } } _ { \rm lo\;\ ; term } + \underbrace { \frac{\tilde{\cal p}_0}{4}\ , \frac{(k^4\!-\!k_r^4 ) } { ( \tilde{r}_0(k^2-k_r^2)/2\!-\!ikc_{\eta}^2)^2 } } _ { \rm nlo\;\ ; correction}+\ldots\,\bigg ] , \label{eq : poletcs}\end{aligned}\ ] ] where @xmath144 from this expression one sees directly that , indeed , @xmath145 , with corrections of @xmath146 .
note that we keep the exact form of the imaginary term in eq .
( [ eq : poletcs ] ) : even though it is negligible at @xmath58 , it has an important exponential dependence on the energy responsible for keeping the phase shifts real in the elastic regime .
when @xmath147 and @xmath148 , and thus @xmath149 , we have @xmath150 and the two poles of eq . ( [ eq : poletcs ] ) are located in the lower half of the complex - momentum plane very near the real axis , as befits a very narrow resonance .
the amplitude @xmath83 can be written in terms of the resonance energy @xmath151 and the resonance width @xmath152 as @xmath153 one finds that @xmath154 \ , , \label{eq : widthstronge}\ ] ] where @xmath155 the width is very small because of the large value of @xmath156 in the exponential . in the form of eqs .
( [ eq : poletcswidth ] ) and ( [ eq : widthstronge ] ) we can keep @xmath157 and @xmath158 fixed at each order in the expansion .
note that these equations do not change to this order if one makes a different choice _e.g .
_ , @xmath159 instead of @xmath160 for the form of the @xmath161 term in eq .
( [ eq : poletcs ] ) . since at the resonance , @xmath162 and @xmath163_{e_r}=2/\gamma(e_r)$ ]
, the behavior of the phase shift around @xmath157 is fixed .
we turn now to a test of the expansion ( [ eq : poletcs ] ) mandated by our power counting .
in this section we briefly describe the experimental situation regarding @xmath0 scattering at low energies .
our predictions are then compared against the available data and discussed in detail , with a particular emphasis on the scalings of the eft parameters .
scattering data at low energies are not abundant . at energies below @xmath137
mev data were obtained , and a phase - shift analysis performed , in ref .
all later references that we were able to find @xcite start at higher energies .
for example , ref .
@xcite covers the region @xmath1376 mev , ref .
@xcite @xmath16412 mev , and ref .
@xcite @xmath1659 mev .
these data show that @xmath0 scattering at @xmath166 mev is dominated by the @xmath20 wave , thanks to the presence of a ( @xmath21 , @xmath22 ) = ( @xmath23 , 0 ) resonance immediately above threshold , identified as the @xmath26be ground state @xcite .
early determinations of the @xmath23 resonance energy were performed in reactions like @xmath167 @xcite .
later measurements and careful analysis of the scattering of @xmath168he atoms off @xmath168he@xmath169 ions @xcite improved the determination of the resonance energy and width to their currently accepted values , @xmath170 kev and @xmath171 ev .
the resonance cm momentum is thus @xmath172 mev . at @xmath173
mev the @xmath99-wave phase shift crosses @xmath174 , indicating the position of a broader resonance associated with the first excited , ( @xmath175 , 0 ) state @xcite .
the @xmath176 wave does not become comparable to the @xmath99 wave until @xmath177 mev , interpreted as the region of an even broader ( @xmath178 , 0 ) state @xcite .
the low - energy data can be studied with the ere . when used together with the @xmath23 resonance energy @xcite , the data from refs . @xcite
give for these parameters @xcite : @xmath179 fm , @xmath180 fm , and @xmath181 @xmath182 .
they provide a good description of the available phase shifts up to @xmath173 mev , which lends some credence to these numbers . however , it was pointed out @xcite that without input from the @xmath23 resonance width these parameters have large uncertainties .
inclusion of both resonance energy and width from ref .
@xcite reduces these uncertainties considerably @xcite : @xmath183 fm , @xmath184 fm , and @xmath185 @xmath182 . since the later , more precise data from ref .
@xcite are consistent with ref .
@xcite , the ere parameters from ref .
@xcite can be seen as a reasonable parametrization of the existing data .
here we use the phase shifts compiled in table ii of ref . @xcite .
since it is well - determined experimentally , and due to its relevance to the triple - alpha process , we use the @xmath23 state as an important constraint .
this is in line with the eft approach , where lower - energy observables have smaller theoretical errors .
it provides a relationship among our eft parameters and , consequently , reduces the number of variables to be adjusted at each order in the power counting .
below , we also use the ere from ref .
@xcite for orientation , and comment on the extremely large value of the scattering length @xmath73 , which suggests a large amount of fine - tuning in the parameters of the underlying theory away from the naturalness assumption .
in contrast , both @xmath186 mev and @xmath187 mev correspond to natural scales comparable to the pion mass . at this level
there is no evidence against our initial estimate of the expansion parameter @xmath188 .
note that here we do not include dimeron fields for resonances beyond the ground state , and therefore can not go beyond the energy region where the @xmath99-wave resonance is significant . in the power counting we are proposing for the @xmath0 system ,
the amplitude @xmath83 for generic momenta is given up to nlo by eq .
( [ eq : poletcs ] ) .
as previously discussed , this expression combines the deep - non - perturbative coulomb approximation ( [ eq : happrox ] ) for the function @xmath88 with the expansion around the resonance pole , which avoids the need for multiple kinematical fine - tunings . in lo , the two parameters @xmath73 and
@xmath189 can be obtained from a fit to the resonance position and width . at nlo
, scattering data are needed to determine @xmath161 .
figure [ fig : pshift ] shows the results of our fit to the available @xmath20-wave phase shifts below @xmath137 mev , including the resonance position and width .
the latter control the steep rise of the phase shift at very low energies . in the region above the resonance , where scattering data are shown , the lo curve is a prediction , which is consistent with the first few points but then moves away from the data .
the nlo curve has an extra parameter , which here was determined from a global @xmath190-fit to scattering data shown .
as expected from a convergent expansion , the description of the low - energy data improves with increasing order . at about 3 mev and above
, higher - order contributions are expected to be significant , as suggested by the discussion on the relevant scales and manifest in the growing difference between the nlo curve and both lo curve and data points . also shown
are results from a fit using the conventional ere formula , eq .
( [ eq : ere ] ) , in order to stress the differences between this and our eft approach .
the ere formula includes some of the contributions of higher order in the eft .
-wave phase shift @xmath84 as a function of the laboratory energy @xmath191 .
the eft results in lo and nlo are given by the ( blue ) dotted and ( red ) solid lines , respectively .
our ere fit using eq .
( [ eq : ere ] ) is given by the dash - dotted line and the empirical phase shifts @xcite as ( black ) solid circles with error bars .
right panel : analogous results for the landau - smorodinsky @xmath91-function .
, title="fig:",width=302,height=226 ] -wave phase shift @xmath84 as a function of the laboratory energy @xmath191 .
the eft results in lo and nlo are given by the ( blue ) dotted and ( red ) solid lines , respectively .
our ere fit using eq .
( [ eq : ere ] ) is given by the dash - dotted line and the empirical phase shifts @xcite as ( black ) solid circles with error bars .
right panel : analogous results for the landau - smorodinsky @xmath91-function .
, title="fig:",width=302,height=226 ] in table [ tab : erepar ] we give the values of ere parameters that we extract from the fits in fig .
[ fig : pshift ] , and compare them with the values @xcite obtained from effective range theory . at lo , @xmath73 and @xmath67 come out consistent with the values given in ref .
@xcite , @xmath67 in particular .
the changes due to nlo corrections , however , worsen this initial lo agreement .
the reason for this deviation is most likely due to the calculation of the width constraint in ref .
( 4 ) reads @xmath192 where @xmath193 $ ] . following this reference s procedure
, we were able to reproduce the quoted value of the width ( 6.4 ev ) only when @xmath194 was approximated by @xmath195 [ see eq .
( [ eq : happrox ] ) ] .
this amounts to ignoring the electromagnetic piece of @xmath161 in eq .
( [ eq : widthstrong ] ) , which is inconsistent since the strong piece contributes at the same order .
it explains why the value for @xmath67 in ref .
@xcite agrees quite well with ours at lo but disagrees at nlo . with this larger @xmath196
, one can also understand why ref .
@xcite obtains a smaller @xmath73 ( see the following discussion on the scaling of @xmath73 ) . when we repeated ref .
@xcite s procedure including the width constraint consistently , we obtained essentially the same values as in our eft fits .
this updated ere fit is also shown in table [ tab : erepar ] .
@xmath67 ( fm ) & @xmath94 ( @xmath182 ) + lo & @xmath197 & 1.083 & + nlo & @xmath198 & @xmath199 & @xmath200 + ere ( our fit ) & @xmath198 & @xmath201 & @xmath202 + ere @xcite & @xmath203 & @xmath204 & @xmath205 + in agreement with our expectations , the effective range shows a natural size , @xmath206 .
the shape parameter also has a natural size @xmath207 at nlo which is in agreement with our _ a priori _ estimate .
the relative errors in @xmath73 and @xmath196 at lo are estimated to be of the order of the eft parameter expansion , @xmath208% . at nlo , they are essentially given by the precision of the most recent measurement of the resonance width @xcite , which lies between 45% .
the uncertainty in @xmath161 , given by the @xmath190-fit , is of the same order .
note that the small relative error in @xmath67 compared to the one in @xmath196 is due to the former being an order of magnitude larger than the latter , as we discuss below .
the nlo errors found here are a factor of two smaller than the ones obtained by ref .
@xcite .
one should stress that the accurate value of the resonance width , @xmath209 ev , imposes tight constraints on our fits , through @xmath73 and @xmath67 .
there is a significant improvement in our nlo fit and overall agreement with data , but the theoretical curve is not able to cross the error bands of all scattering points below 3 mev , as reflected in a @xmath190/datum @xmath210 . in principle ,
a better agreement should be achieved by an n@xmath211lo calculation .
however , that introduces an extra parameter that is mostly determined by the scattering data , and this agreement could mask any possible inconsistencies that the phase shifts might have with the resonance parameters .
the high nlo @xmath190/datum suggests that the resonance width and the @xmath20-wave scattering data set are not compatible with each other or , at least , one of them has overestimated precision . as a test we performed fits to scattering data without the input from the resonance width , within our nlo eft and the conventional ere framework . in both cases , description of @xmath20-wave phase shifts is much better but the resonance width is underpredicted , @xmath212 ev with ere and @xmath213 ev with eft .
the ere result is still consistent with the measured @xmath214 thanks to its large error bar . in eft , where lower - energy data have higher priority , the discrepancy is amplified .
the problem is even more pronounced if the fit is performed using data up to 2.5 mev instead of 3 mev : the results @xmath215 ev with ere and @xmath216 ev with eft fall beyond the quoted experimental error bars .
oddly , this tendency continues as one lowers the upper limit in the fit .
reanalyses of the existing low - energy data or even new measurements seem necessary to resolve this discrepancy .
an astonishing feature of the @xmath0 system is the very large magnitude of @xmath73 , even if compared to the large scattering length observed in the two - nucleon system .
the latter is evidence of a fine - tuning in the qcd parameters , which gives rise to an anomalously low momentum scale .
( this fine - tuning can be seen as the proximity of the observed pion mass to the critical value @xmath217 mev where the @xmath35 bound states have zero energy @xcite and the triton displays the efimov effect @xcite . ) certainly this fine - tuning scale propagates to heavier systems . however , the enormous value of @xmath73 in the @xmath0 system is suggestive of a more dramatic fine - tuning , with electromagnetic interactions playing a crucial role .
the parameters of the @xmath23 resonance are indeed quite surprising . as we remarked , this resonance is associated with two poles of @xmath83 , the momenta of which are much smaller in magnitude than @xmath76 . as a consequence , there is an exponential suppression of the width , which is evident in eq .
( [ eq : widthstrong ] ) .
yet , the width is somewhat _
large _ given the position of the resonance .
it is remarkable that in order to fit both the resonance position and width one needs @xmath218 fm , nearly a factor 10 smaller in magnitude than the coulomb contribution to the energy - dependent term , @xmath219 fm .
that is , although both @xmath220 fm and @xmath221 fm as expected from dimensional analysis , they approximately cancel producing a balance that is about 10% of either .
this balance would itself have natural size if the width were a factor of 10 smaller than measured , or else if @xmath130 were 15% larger .
instead , the existing cancellation effectively leads to either @xmath222 or @xmath223 . since the position of the resonance determines to a first approximation @xmath224 , the seemingly accidental cancellation in @xmath189 translates into an exceptionally large @xmath73 , which is effectively @xmath225 or @xmath226 .
the coulomb - modified parameter @xmath96 is thus suppressed by another power of @xmath28 compared to the assumed scaling when coulomb interactions are turned off .
the values of ere parameters show that , as anticipated by the power counting we proposed , the strong - interaction effective - range term is comparable to both the scattering - length term and the electromagnetic @xmath88 .
a power counting based on the scalings appropriate for @xmath35 scattering would instead miss the resonance in lo unless one also counted @xmath227 as @xmath29 , which is a numerical stretch .
even so , the width would be far off , since @xmath228 , and convergence questionable . although the results presented in this section were based on eq .
( [ eq : poletcs ] ) , and thus assumed the validity of the expansion ( [ eq : happrox ] ) , we have checked that no qualitative changes arise if one bases the eft fits on eq .
( [ eq : ampsc2 ] ) instead . in the latter case , one is simply retaining at each order some terms that truly belong to higher orders , at the cost of much less transparent analytical expressions .
in this paper , we have studied @xmath0 interactions within the framework of halo / cluster eft . from this perspective , the @xmath0 system is most extraordinary .
eft is model independent to the extent that includes all interactions allowed by known symmetries .
it is equivalent in the @xmath0 system to a truncation of the ere .
the halo eft can be viewed as a formulation of the ere that can be used for other systems made of shallow - bound alpha particles and , together with results from refs .
@xcite , also nucleons .
we have designed a power - counting scheme that seems realistic for the @xmath0 system , but requires even more fine - tuning than the one in the @xmath35 system .
it assumes that the parameter @xmath30 in the eft lagrangian scales as @xmath229 in the absence of the coulomb interaction .
interesting in this scenario is the fact that at lo without coulomb the @xmath26be system would exhibit conformal invariance and @xmath45c would display an exact efimov spectrum .
these exact features are broken by the coulomb interaction but some remnants of this behavior are manifest in the experimental spectra , such as the shallowness of the @xmath23 @xmath26be resonance .
we have incorporated two new ideas in the halo / cluster eft .
the most effective way of implementing an eft with resonances seems to be an expansion around the resonance pole , since it avoids the need for multiple kinematical fine - tunings and improves convergence .
our study was significantly simplified by treating the coulomb function @xmath230 in a power series expansion in @xmath231 .
this simplification may be very relevant when treating interactions with several alpha - particle clusters .
we made use of the precise measurements of the @xmath23 resonance properties @xcite , together with the ( rather old ) existing @xmath20-wave scattering phase shifts @xcite in order to produce lo and nlo fits that fix our ere parameters .
the result is in slight disagreement with ref .
@xcite , probably due to an approximation made in its calculation of the width .
the uncertainties in our ere parameters are smaller by a factor of two in comparison with this reference .
a systematic improvement is seen in the theoretical phase shifts , and reasonably good overall description is obtained at nlo . yet
, the rather high @xmath190 when the resonance width is fitted and the deterioration of the calculated width as one restricts the fit to lower - energy scattering data suggest that either the resonance width or the @xmath20-wave phase shifts have overestimated precision . despite the phenomenological success of the proposed power counting
, the necessary cancellations between strong and coulomb interactions , which seem accidental , are puzzling . in the absence of coulomb interactions , the regulator - dependent part of the bare dimeron
mass @xmath30 is @xmath232 . as we can see from eq .
( [ eq : renorm ] ) , the electromagnetic @xmath98-dependent pieces do not change this expectation , since they are @xmath233 . therefore , barring cancellations between the finite pieces of @xmath30 and loops , we would expect the renormalized mass @xmath96 to have the same size , which would set a scale for @xmath234 at @xmath235 fm . yet
, the observed resonance energy ( together with a natural - sized effective range @xmath189 ) tells us that @xmath236 . a similar scaling for @xmath96 emerges in @xmath7 scattering @xcite , but here we need a further cancellation between the finite pieces in @xmath30 and coulomb loops .
this brings @xmath234 up by two orders of magnitude , or more than two inverse powers of the expansion parameter @xmath237 , to @xmath238 fm .
uncomfortable as this fine - tuning by a factor of nearly 100 in the energy - independent part of the amplitude @xmath83 might be , it is not the whole story . as we have shown , both the resonance width and the higher - energy phase shifts require a fine - tuning also in the energy dependence of the amplitude .
this @xmath239% additional cancellation ( together with the observed resonance energy ) further enhances @xmath234 by another factor of 10 , or about another inverse power of @xmath28 , leading effectively to @xmath240 fm , which is indeed the order of magnitude we obtain in our fit .
this fine - tuning of a factor of @xmath241 1000 in @xmath0 completely overshadows the fine - tuning of @xmath241 10 from @xmath242 ( or from the pion decay constant @xmath243 mev ) down to @xmath244 mev in the @xmath35 @xmath245 channel .
it has important consequences : for example , if the strong - interaction effective range @xmath67 were just 15 - 20% larger it would make the ground state of @xmath1be bound , presumably with far - reaching effects in nucleosynthesis .
this context frames a fascinating picture for the @xmath0 system : a nearly conformally invariant system that is plagued by cancellations between strong and coulomb interactions .
of course , fine - tuning has long been discussed in nuclear physics , but usually in connection to the position of the hoyle state of @xmath2c ( see , for example , ref .
an immediate question is to which extent the latter arises from the fine - tunings we discussed here .
together with previous and ongoing work on the @xmath7 @xcite and @xmath8 systems @xcite , our results provide a framework for the description of cluster states in nuclei using eft , which can be seen as a generalization of the ere to systems that involve more than two bodies .
an important extension of this work is to systems of more than two alpha particles .
the strong @xmath0 interaction obtained here , together with an exact treatment of the coulomb interaction , should provide the necessary ingredients .
since @xmath114 as discussed above , one might be able to simplify the calculation of coulomb effects by developing a perturbation scheme in powers of @xmath246 from the start .
moreover , we conjecture that the @xmath2c hoyle state is a remnant of an efimov state that appears in the unitary limit .
more complex @xmath5-cluster systems could also be studied with the eft presented here .
we would like to thank carlos bertulani for useful discussions and encouragement .
rh has also benefited from discussions with andreas nogga , and uvk with brett carlson , tobias frederico , boris gelman , and manuel malheiro .
this research was supported in part by the bundesministerium fr bildung und forschung under contract number 06bn411 ( rh , hwh ) , by the u.s .
department of energy ( uvk ) , by the nederlandse organisatie voor wetenschappelijk onderzoek ( uvk ) , and by brazil s fapesp under a visiting professor grant ( uvk ) .
uvk would like to thank the hospitality of the kernfysisch versneller instituut at rijksuniversiteit groningen , the instituto de fsica terica of the universidade estadual paulista , and the instituto de fsica of the universidade de so paulo , where part of this work was carried out .
the scattering amplitude for two particles in their center - of - mass ( cm ) system interacting via coulomb and a short - range interaction is given by @xcite @xmath247 where @xmath248 , @xmath249 , and @xmath250 represent free and ( incoming / outgoing ) states of momentum @xmath251 for pure - coulomb and coulomb - distorted short - range interactions , respectively , while @xmath252 ( @xmath253 ) is the coulomb ( short - range ) interaction operator . in coordinate space , the coulomb wave functions can be written as @xcite @xmath254 where @xmath255 is the kummer function . from @xmath256 @xcite ,
one obtains the important properties @xmath257 & & \chi^{(\mp)\,*}_{k'}(0)\,\chi^{(\pm)}_{k}(0)=e^{-\pi\eta}\gamma(1\pm i\eta)^2 = c_{\eta}^2\,e^{\pm 2i\sigma_0}\ , , \label{eq : psi0 - 2}\end{aligned}\ ] ] where @xmath86 is the coulomb phase shift for the partial wave @xmath258 . the general expression for the coulomb phase shift , @xmath259\ , , \label{eq : sigma_l}\ ] ]
is defined from the partial - wave expansion of the pure - coulomb amplitude @xmath260 the explicit solution @xmath261\ ] ] leads to the well - known mott scattering cross section , @xmath262 , which holds at very low energies .
the main ingredient in the calculation of the coulomb - distorted short - range amplitude is the coulomb green s function at energy @xmath263 ,
@xmath264 using the lippmann - schwinger equation , one is able to express @xmath265 in terms of multiple insertions of the operator @xmath266 acting on the coulomb states @xmath267 @xcite , allowing the amplitude to be written as the sum @xmath268 this amplitude can be decomposed into partial waves , @xmath269 in terms of the `` coulomb - corrected '' phase shifts @xmath270 .
if the deviation from the mott cross section is mostly @xmath20 wave , then the calculation of @xmath83 proceeds from the lagrangian ( [ eq : lolag ] ) similarly to ref .
the purely strong matrix element @xmath271 is defined as the amplitude @xmath272 for @xmath0 scattering via a bare dimeron two - point function , with the external legs amputated , evaluated in the cm : @xmath273 explicit evaluation of eq .
( [ eq : tcs1 ] ) using eqs .
( [ eq : psi0 - 1 ] ) , ( [ eq : psi0 - 2 ] ) , and ( [ eq : mtrx1 ] ) then gives @xmath274\,.\ ] ] this resummation of the coulomb exchanges is illustrated in fig .
[ fig : t0sum ] .
it leads to @xmath275^{-1}\,,\end{aligned}\ ] ] where @xmath276 is given by @xmath277\right\}\ , , \label{eq : kr - s-13a}\ ] ] with @xmath99 the dimensionality of spacetime , @xmath98 the renormalization scale , @xmath278 , @xmath279 the euler - mascheroni constant , and @xmath280 in terms of the digamma function @xmath281 , which obeys @xcite @xmath282={{{\rm re\ , } [ } } \psi(1+i\eta)]\ , , \qquad { { { \rm im\ , } [ } } \psi(i\eta)]=\frac{1}{2\eta}+\frac{\pi}{2}\,\coth\pi\eta\,.\ ] ] , represented by the ( red ) shaded rectangle .
the double line stands for the dimeron propagator .
the ( blue ) shaded ellipse represents the propagation in the presence of coulomb photons ( dashed line ) but absence of a dimeron .
, title="fig:",height=124 ] + , represented by the ( red ) shaded rectangle .
the double line stands for the dimeron propagator .
the ( blue ) shaded ellipse represents the propagation in the presence of coulomb photons ( dashed line ) but absence of a dimeron .
, title="fig:",height=45 ] p.f .
bedaque and u. van kolck , ann .
nucl . part .
sci . * 52 * , 339 ( 2002 ) ; e. epelbaum , prog .
nucl . part .
phys . * 57 * , 654 ( 2006 ) .
e. braaten and h .- w .
hammer , phys .
* 428 * , 259 ( 2006 ) .
i. stetcu , b.r .
barrett , and u. van kolck , phys . lett . *
b653 * , 358 ( 2007 ) ; i. stetcu , b.r .
barrett , u. van kolck , and j. vary , phys . rev .
a * 76 * , 063613 ( 2007 ) .
b. borasoy , e. epelbaum , h. krebs , d. lee , and u .- g .
meiner , eur .
j. a * 31 * , 105 ( 2007 ) .
bertulani , h .- w . hammer , and u. van kolck , nucl . phys . * a712 * , 37 ( 2002 ) .
u. van kolck , arxiv : hep - ph/9711222 , in _ proceedings of the workshop on chiral dynamics 1997 , theory and experiment _ , ed .
a. bernstein , d. drechsel , and t. walcher ( springer - verlag , berlin , 1998 ) ; nucl .
* a645 * , 273 ( 1999 ) ; d.b .
kaplan , m.j .
savage , and m.b .
wise , phys .
b424 * , 390 ( 1998 ) ; nucl
. phys . *
b534 * , 329 ( 1998 ) ; j. gegelia , phys . lett .
* b429 * , 227 ( 1998 ) .
x. kong and f. ravndal , phys
* b450 * , 320 ( 1999 ) ; nucl
* a665 * , 137 ( 2000 ) ; b.r .
holstein , phys .
d * 60 * , 114030 ( 1999 ) ; s. ando , j.w .
shin , c.h .
hyun , and s.w .
hong , phys .
c * 76 * , 064001 ( 2007 ) .
t. mehen , i.w .
stewart , and m.b .
wise , phys .
b474 * , 145 ( 2000 ) . s.r .
beane , p.f .
bedaque , m.j .
savage , and u. van kolck , nucl .
a700 * , 377 ( 2002 ) ; s.r .
beane and m.j .
savage , nucl .
a713 * , 148 ( 2003 ) ; nucl . phys .
* a717 * , 91 ( 2003 ) ; e. epelbaum , u .- g .
meiner , and w. glckle , nucl . phys . *
a714 * , 535 ( 2003 ) ; arxiv : nucl - th/0208040 .
e. braaten and h .- w .
hammer , phys .
lett . * 91 * , 102002 ( 2003 ) ; e. epelbaum , h .- w .
hammer , u .- g
. meiner , and a. nogga , eur .
j. c * 48 * , 169 ( 2006 ) ; h .- w .
hammer , d.r .
phillips , and l. platter , eur .
phys . j. a * 32 * , 335 ( 2007 ) . | we study the two - alpha - particle ( @xmath0 ) system in an effective field theory ( eft ) for halo - like systems . we propose a power counting that incorporates the subtle interplay of strong and electromagnetic forces leading to a narrow resonance at an energy of about 0.1 mev .
we investigate the eft expansion in detail , and compare its results with existing low - energy @xmath0 phase shifts and previously determined effective - range parameters .
good description of the data is obtained with a surprising amount of fine - tuning .
this scenario can be viewed as an expansion around the limit where , when electromagnetic interactions are turned off , the @xmath1be ground state is at threshold and exhibits conformal invariance .
we also discuss possible extensions to systems with more than two alpha particles . |
taking into account the recent technological advances , and needs of our modern society , the study of the magnetic properties of new materials is of fundamental importance to develop devices for different applications such as , for example , memory structures.@xcite in view of their characteristics , amorphous silicon ( _ a_-si ) thin films seems to be good candidates for such a purpose.@xcite an interesting way of studying the _ a_-si thin films magnetic response is to focus on the properties of the neutral dangling - bonds ( d@xmath3 ) present in these materials .
neutral dangling - bonds are paramagnetic centers that are excellent probes for the investigation of _ a_-si thin films .
moreover , silicon dangling - bonds are charge trapping centers@xcite that are more stable under the diamagnetic d@xmath5 form.@xcite in the present work we have studied the behaviour of the paramagnetic defects d@xmath3 in amorphous silicon nitride thin films doped with various rare - earth elements , _ a_-sin : re ( re = rare - earths : y , la , pr , nd , sm , gd , tb , dy , ho , er , yb , and lu ) . depending on the re dopant , these thin films present a relative strong and narrow light emission , even at room temperature.@xcite,@xcite as a consequence , re - doped _
a_-sin thin films
are expected to be ideal candidates to develop photonic devices . towards this end
, the study of their magnetic properties will certainly contribute to decide about the potential applications of these materials .
all films were prepared in a high vacuum chamber ( base pressure @xmath6 torr ) , by radio frequency ( @xmath7 mhz ) sputtering a si ( @xmath8 % pure ) target covered at random with small pieces of metallic re ( @xmath9 % pure ) elements .
polished crystalline ( _ c_-)si wafers and high - purity quartz plates were used as substrates in every deposition run . during deposition ,
the substrates were kept at @xmath10 @xmath11c under a constant total pressure of @xmath12 torr consisting of a mixture of high - purity ar + n@xmath13 gases .
the mean thickness of the films was @xmath14 nm . the atomic composition of the _ a_-sin : re films ( @xmath15 si , @xmath16 n , @xmath17 re ) were determined by _ rutherford _
backscattering spectrometry ( rbs ) in the case of si and re and by nuclear reaction analysis ( nra ) for n. a non - intentionally amount of hydrogen of @xmath18 h was detected by elastic recoil detection ( erd ) analysis in all _ a-_sin : re films .
the density of the films was estimated to be @xmath19 at .
the optical bandgap of these films were determined through optical transmission measurements in the visible - ultraviolet range using a commercial spectrophotometer and stays around @xmath21 ev.@xcite room-@xmath22 _ raman _ scattering measurements , using the @xmath23 nm line of an ar@xmath24 _ laser _ , were also performed and confirmed the amorphous structure of the films .
the electron spin resonance ( esr ) experiments were carried out in a _ bruker _ x - band ( @xmath25 ghz ) spectrometer using a room-@xmath26 te@xmath27 cavity .
all mesuarements have been taken at room temperature .
this work presents a new approach in the study of the d@xmath3 density of _ a_-sin thin films doped with res .
our main finding was the depletion of the density of d@xmath3 in the _
a_-sin matrix caused by the presence of magnetic re species .
we have observed that the insertion of magnetic re species dramatically suppresses the number of esr active d@xmath3 states and that such a decrease approximately scales with the spin component of the re magnetic moment .
table i displays the atomic concentrations [ re ] , [ si ] , [ n ] and [ h ] as determined from rbs , nra , and erd for all the films investigated in this work . from the d@xmath3
esr intensity measurements , and using as standard a kcl - pitch sample , we have estimated the [ d@xmath0 of each film . the d@xmath3 esr parameters and [ d@xmath0 are also given in table i. as can be seen from table i , the average density of d@xmath3 magnetic defects in these films was of , typically , @xmath28 @xmath29 .
figure @xmath30 shows the room-@xmath26 esr normalized spectra of d@xmath3 in _ a_-sin films doped with different re elements ( notice the different intensities ) . from a lorentzian lineshape fitting of the resonances for all the _ a_-sin : re thin films we have obtained approximately the same peak - to - peak linewidth , @xmath31 @xmath32 g , and field for resonance , @xmath33 g ( corresponding to @xmath34 ) .
an early esr study on undoped _
a-_si@xmath35n@xmath36 films pointed out a very weak d@xmath3 esr signal with @xmath37 and @xmath38 g.@xcite a comparison of these data with our larger linewidth and higher [ d@xmath1 suggests that the present _ a-_sin : re films are more disordered .
this is probably associated to the higher n / si ratio ( @xmath39 in our films.@xcite figure @xmath40 shows [ d@xmath0 for the various re elements investigated in this work .
it is noted that the magnetic res cause a dramatic depletion of [ d@xmath41 and the strongest suppressing effect is found for gd@xmath42 , at the middle of the re - series . within our experimental accuracy , the non - magnetic re elements do not cause a systematic change in [ d@xmath0 . the inset of fig .
@xmath40 presents the drop of [ d@xmath0 , or in another words , the number of inactive esr d@xmath3 , [ d@xmath3(re@xmath43)]@xmath44[d@xmath3(re@xmath45 ) ] , due to the presence of the magnetic re@xmath45 s relative to the average value for the non - magnetic re@xmath43 s .
notice that the minimum in fig . @xmath40
correlates quite well with the re s de gennes factor , @xmath46 , and/or the @xmath47 factor .
the striking result of figure @xmath40 suggests that the mechanism responsible for the depletion of [ d@xmath0 involves the spin part of the re magnetic moment and may be attributed to a strong exchange - like coupling , @xmath48 , between the re@xmath42 spin , @xmath49 , and the spin of the d@xmath3 , @xmath50 .
such a strong exchange coupling may probably shift and/or broaden the d@xmath3 resonance beyond the detection limit of our esr experimental facilities .
it is then possible that a coupling of this kind leads to a [ d@xmath0 decrease involving the de gennes factor , @xmath46 .
the existence of this factor has been largely confirmed in re - doped type ii superconductor compounds through the decrease of the superconducting temperature , @xmath4 , ( @xmath51 ) due to the cooper - pairs breaking property of the re ions .
@xcite @xcite @xcite at this point , we should also mention that the depletion of [ d@xmath0 could be approximately described by the spin part of the re - magnetic moment , @xmath47 , that also takes its highest value at the gd@xmath42 ion ( @xmath52 ) .
these two analyses are showed in the inset of fig . @xmath40 .
rare - earth doped amorphous silicon - nitrogen films were prepared by co - sputtering and investigated by means of different experimental techniques ( esr , @xmath53-susceptibility , ion - beam analyses , and _
raman _ scattering ) .
the main experimental results can be summarized as : @xmath54 a strong depletion in the density of d@xmath3 states induced by the presence of magnetic reions , and @xmath55 the correspondence between this depletion and the re s de gennes factor , @xmath46 , and/or the re s @xmath47 factor .
these results led us to propose a mechanism involving a strong exchange - like coupling between the re@xmath42 magnetic moment and the spin of the silicon dangling - bond d@xmath3 . this strong coupling may cause a large shift and/or broadening of the d@xmath3 resonance which are beyond the limit of detection of our esr spectrometer .
this work has been supported by fapesp , capes and cnpq .
table i - atomic composition of the a - sin : re films considered in this work , as determined from rbs ( _ rutherford _ backscattering spectrometry ) and nra ( nuclear reaction analysis ) .
@xmath56 , @xmath57 , @xmath58 , and [ d@xmath0 stand for the line - width , resonance field , @xmath58-value , and neutral si dangling - bond concentration , respectively .
_ _ @xmath59_n.a .
_ means not available and `` * '' means minimum detectable value .
@xmath60$]_@xmath61$]_@xmath62$]_@xmath63h_pp@xmath63h_r@xmath63g@xmath64 ^ 0@xmath65 \\ & ( at.\% ) & ( at.\% ) & ( at.\% ) & ( g ) & ( g ) & - & ( cm$]^-3@[email protected]@[email protected]@xmath6316(2)@xmath633378(2)@xmath632.004(1)@xmath633.210 ^ 20@[email protected]@[email protected]@xmath6320(2)@xmath633378(2)@xmath632.004(1)@xmath634.210 ^ 20@[email protected]@[email protected]@xmath6318(2)@xmath633378(2)@xmath632.005(1)@xmath632.710 ^ 20@[email protected]@[email protected]@xmath6316(2)@xmath633379(2)@xmath632.004(1)@xmath631.910 ^ 20@[email protected]@[email protected]@xmath6316(2)@xmath633378(2)@xmath632.005(1)@xmath639.610 ^ 19@[email protected]@[email protected]@xmath6315(2)@xmath633378(2)@xmath632.004(1)@xmath639.210 ^ 19@[email protected]@[email protected]@[email protected][email protected][email protected] ^ 17@[email protected]@[email protected]@xmath6319(2)@xmath633377(2)@xmath632.005(1)@xmath631.610 ^ 20@[email protected]@[email protected]@xmath6317(2)@xmath633377(2)@xmath632.005(1)@xmath631.710 ^ 20@[email protected]@[email protected]@xmath6316(2)@xmath633379(2)@xmath632.004(1)@xmath631.710 ^ 20@[email protected]@[email protected]@xmath6316(2)@xmath633379(2)@xmath632.004(1)@xmath631.710 ^ 20@[email protected]@[email protected]@xmath6321(2)@xmath633378(2)@xmath632.004(1)@xmath632.410 ^ 20@[email protected]@[email protected]@xmath6320(2)@xmath633377(2)@xmath632.005(1)@xmath634.710 ^ 20@xmath82 | amorphous silicon - nitrogen thin films doped with rare - earth elements ( _ a_-sin : re ; re = y , la , pr , nd , sm , gd , tb , dy , ho , er , yb , and lu ) have been prepared by co - sputtering and studied by means of electron spin resonance ( esr ) .
it was found that the neutral dangling - bond density [ d@xmath0 of _ a_-sin films decreases with the presence of magnetic res and the drop of [ d@xmath1 approximately scales with the spin and/or the de gennes factor of each rare - earth element .
these results suggest that a strong exchange - like interaction , @xmath2 , between the spin of the magnetic res and d@xmath3 may be responsible for this behaviour , similarly to the decrease of @xmath4 in re - doped superconductors . |
axially symmetric solutions of einstein s field equations corresponding to disklike configurations of matter are of great astrophysical interest , since they can be used as models of galaxies or accretion disks .
these solutions can be static or stationary and with or without radial pressure .
solutions for static disks without radial pressure were first studied by bonnor and sackfield @xcite , and morgan and morgan @xcite , and with radial pressure by morgan and morgan @xcite .
disks with radial tension have been considered in @xcite , and models of disks with electric fields @xcite , magnetic fields @xcite , and both magnetic and electric fields have been introduced recently @xcite .
solutions for self - similar static disks were analyzed by lynden - bell and pineault @xcite , and lemos @xcite .
the superposition of static disks with black holes were considered by lemos and letelier @xcite , and klein @xcite .
also bick , lynden - bell and katz @xcite studied static disks as sources of known vacuum spacetimes and bick , lynden - bell and pichon @xcite found an infinity number of new static solutions . for a recent survey on relativistic gravitating disks ,
see @xcite . the principal method to generate
the above mentioned solution is the `` displace , cut and reflect '' method .
one of the main problem of the solutions generated by using this simple method is that usually the matter content of the disk is anisotropic i.e. , the radial pressure is different from the azimuthal pressure . in most of the solutions
the radial pressure is null .
this made these solutions rather unphysical .
even though , one can argue that when no radial pressure is present stability can be achieved if we have two circular streams of particles moving in opposite directions ( counter rotating hypothesis , see for instance @xcite ) .
in this article we apply the `` displace , cut and reflect '' method on spherically symmetric solutions of einstein s field equations in isotropic coordinates to generate static disks made of a _ perfect fluid _ ,
i.e. , with radial pressure equal to tangential pressure and also disks of perfect fluid surrounded by an halo made of perfect fluid matter
. the article is divided as follows .
section ii gives an overview of the `` displace , cut and reflect '' method .
also we present the basic equations used to calculate the main physical variables of the disks . in section iii
we apply the formalism to obtain the simplest model of disk , that is based on schwarzschild s vacuum solution in isotropic coordinate .
the generated class of disks is made of a perfect fluid with well behaved density and pressure .
section iv presents some models of disks with halos obtained from different known exact solutions of einstein s field equations for static spheres of perfect fluid in isotropic coordinates . in section v we give some examples of disks with halo generated from spheres composed of fluid layers .
section vi is devoted to discussion of the results .
for a static spherically symmetric spacetime the general line element in isotropic spherical coordinates can be cast as , @xmath1 \mbox{.}\ ] ] in cylindrical coordinates @xmath2 the line element ( [ eq_line1 ] ) takes the form , @xmath3 the metric of the disk will be constructed using the well known `` displace , cut and reflect '' method that was used by kuzmin @xcite in newtonian gravity and later in general relativity by many authors @xcite-@xcite .
the material content of the disk will be described by functions that are distributions with support on the disk .
the method can be divided in the following steps that are illustrated in fig .
[ fig_schem1 ] : first , in a space wherein we have a compact source of gravitational field , we choose a surface ( in our case , the plane @xmath4 ) that divides the space in two pieces : one with no singularities or sources and the other with the sources .
then we disregard the part of the space with singularities and use the surface to make an inversion of the nonsingular part of the space .
this results in a space with a singularity that is a delta function with support on @xmath4 .
this procedure is mathematically equivalent to make the transformation @xmath5 , with @xmath0 constant . in the einstein tensor
we have first and second derivatives of @xmath6 . since @xmath7 and @xmath8 , where @xmath9 and @xmath10 are , respectively , the heaviside function and the dirac distribution .
therefore the einstein field equations will separate in two different pieces @xcite : one valid for @xmath11 ( the usual einstein s equations ) , and other involving distributions with an associated energy - momentum tensor , @xmath12 , with support on @xmath4 . for the metric ( [ eq_line2 ] ) , the non - zero components of @xmath13 are @xmath14 \mbox { , } \label{eq_qtt}\\ q^r_r & = q^{\varphi}_{\varphi } = \frac{1}{16 \pi } \left [ -b^{zz}+g^{zz}(b^t_t+b^r_r+b^z_z ) \right ] \mbox { , } \label{eq_qrr}\end{aligned}\ ] ] where @xmath15 denote the jump of the first derivatives of the metric tensor on the plane @xmath4 , @xmath16 and the other quantities are evaluated at @xmath17 .
the `` true '' surface energy - momentum tensor of the disk can be written as @xmath18 , thus the surface energy density @xmath19 and the radial and azimuthal pressures or tensions @xmath20 read : @xmath21 note that when the same procedure is applied to an axially symmetric spacetime in weyl coordinates we have @xmath22 , i.e. , we have no radial pressure or tension .
this procedure in principle can be applied to any spacetime solution of the einstein equations with or without source ( stress tensor ) .
the application to a static sphere of perfect fluid is schematized in fig .
[ fig_schem2 ] .
the sphere is displaced and cut by a distance @xmath0 less then its radius @xmath23 .
the part of the space that contains the center of the sphere is disregarded .
after the inversion of the remaining space , we end up with a disk surrounded by a cap of perfect fluid .
the properties of the inner part of the disk will depend on the internal fluid solution , but if the internal spherical fluid solution is joined to the standard external schwarzschild solution , the physical properties of the outer part of the disk will be those originated from schwarzschild s vacuum solution . in isotropic
coordinates the matching at the boundary of the fluid sphere leads to four continuity conditions metric functions @xmath24 and @xmath25 together with their first derivatives with respect to the radial coordinate should be continuous across the boundary .
in addition , to have a compact body the pressure at the surface of the material sphere has to drop to zero . also to have a meaningfully solution the velocity of sound , @xmath26
should be restricted to the interval @xmath27 .
the einstein equations for a static spherically symmetric space time in isotropic coordinates for a perfect fluid source give us that density @xmath28 and pressure @xmath29 are related to the metric functions by @xmath30 \mbox { , } \\ p & = \frac{e^{-\lambda}}{8 \pi}\left [ \frac{1}{4}(\lambda')^2+\frac{1}{2}\lambda'\nu'+\frac{1}{r}(\lambda'+\nu ' ) \right ] \mbox{,}\end{aligned}\ ] ] where primes indicate differentiation with respect to @xmath31 . also static spheres composed of various layers of fluid can be used to generate disks with halos of fluid layers ( see figure [ fig_schem3 ] ) .
the disk will then be composed of different axial symmetric `` pieces '' glued together .
the matching conditions at the boundary of adjacent spherical fluid layers in isotropic coordinates involves four continuity conditions : the two metric functions @xmath24 and @xmath25 , the first derivative of @xmath32 with respect to the radial coordinate , and the pressure should be continuous across the boundary . at the most external boundary ,
the metric functions @xmath24 and @xmath25 , and their first derivatives with respect to the radial coordinate should be continuous across the boundary ; also the pressure there should go to zero .
we first apply the `` displace , cut and reflect '' method to generate disks discussed in the previous section and depicted in fig.[fig_schem1 ] to the schwarzschild metric in isotropic coordinates @xmath33 , @xmath34 \mbox{. } \label{eq_metrica_sch}\ ] ] expressing solution ( [ eq_metrica_sch ] ) in cylindrical coordinates , and using eq .
( [ eq_qtt ] ) ( [ eq_disk_surf ] ) , we obtain a disk with surface energy density @xmath19 and radial and azimuthal pressures ( or tensions ) @xmath35 given by @xmath36 the total mass of the disk can be calculated with the help of eq.([eq_en_sch ] ) : @xmath37 eq .
( [ eq_en_sch ] ) shows that the disk s surface energy density is always positive ( weak energy condition ) .
positive values ( pressure ) for the stresses in azimuthal and radial directions are obtained if @xmath38 .
the strong energy condition , @xmath39 is then satisfied .
these properties characterize a fluid made of matter with the usual gravitational attractive property .
this is not a trivial property of these disks since it is known that the `` displace , cut and reflect '' method sometimes gives disks made of exotic matter like cosmic strings , see for instance @xcite .
another useful parameter is the velocity of sound propagation @xmath40 , defined as @xmath41 , which can be calculated using eq .
( [ eq_en_sch ] ) and eq .
( [ eq_p_sch ] ) : @xmath42 the condition @xmath43 ( no tachyonic matter ) imposes the inequalities @xmath44 or @xmath45 . if the pressure condition and the speed of sound less then the speed of light condition are to be simultaneously satisfied , then @xmath44
. this inequality will be valid in all the disk if @xmath46 . , ( b ) pressures @xmath35 , ( c ) sound velocity @xmath40 and ( d ) tangential velocity @xmath47 ( rotation curve or rotation profile ) with @xmath48 and @xmath49 , and @xmath50 as function of @xmath51 .
we use geometric units @xmath52.,title="fig : " ] , ( b ) pressures @xmath35 , ( c ) sound velocity @xmath40 and ( d ) tangential velocity @xmath47 ( rotation curve or rotation profile ) with @xmath48 and @xmath49 , and @xmath50 as function of @xmath51 .
we use geometric units @xmath52.,title="fig : " ] + , ( b ) pressures @xmath35 , ( c ) sound velocity @xmath40 and ( d ) tangential velocity @xmath47 ( rotation curve or rotation profile ) with @xmath48 and @xmath49 , and @xmath50 as function of @xmath51 .
we use geometric units @xmath52.,title="fig : " ] , ( b ) pressures @xmath35 , ( c ) sound velocity @xmath40 and ( d ) tangential velocity @xmath47 ( rotation curve or rotation profile ) with @xmath48 and @xmath49 , and @xmath50 as function of @xmath51 .
we use geometric units @xmath52.,title="fig : " ] , ( b ) pressures @xmath35 and ( c ) sound velocity @xmath40 with @xmath53 and @xmath54 , and @xmath55 as function of @xmath56.,title="fig : " ] , ( b ) pressures @xmath35 and ( c ) sound velocity @xmath40 with @xmath53 and @xmath54 , and @xmath55 as function of @xmath56.,title="fig : " ] + , ( b ) pressures @xmath35 and ( c ) sound velocity @xmath40 with @xmath53 and @xmath54 , and @xmath55 as function of @xmath56.,title="fig : " ] with the presence of radial pressure one does not need the assumption of streams of rotating and counter rotating matter usually used to explain the stability of static disk models
. however , a tangential velocity ( rotation profile ) can be calculated by assuming a test particle moves in a circular geodesic on the disk .
we tacitly assume that this particle only interact gravitationally with the fluid .
this assumption is valid for the case of a particle moving in a very diluted gas like the gas made of stars that models a galaxy disk . the geodesic equation for the @xmath57 coordinate obtained from metric ( [ eq_line2 ] ) is @xmath58 for circular motion on the @xmath4 plane , @xmath59 and @xmath60 , then eq . ( [ eq_geod ] ) reduces to @xmath61 the tangential velocity measured by an observer at infinity is then @xmath62 from the metric on the disk , @xmath63 we find that eq .
( [ eq_veloc_tang ] ) can be cast as , @xmath64 } \mbox{.}\ ] ] for @xmath65 , eq . ( [ eq_v_fi_sch ] ) goes as @xmath66 , the newtonian circular velocity . to determine the stability of circular orbits on the disk s plane
, we use an extension of rayleigh @xcite criteria of stability of a fluid at rest in a gravitational field @xmath67 where @xmath68 is the specific angular momentum of a particle on the disk s plane : @xmath69 using eq .
( [ eq_phi_t ] ) and the relation @xmath70 one obtains the following expression for @xmath68 : @xmath71 for the functions ( [ eq_sch_coef ] ) , eq . ( [ eq_h ] ) reads @xmath72 the stability criterion is always satisfied for @xmath73 . in figure [ fig_1 ] ( a)(d )
we show , respectively , the surface energy density , pressures , the sound velocity and curves of the tangential velocity ( rotation curves ) eq . ( [ eq_v_fi_sch ] ) with @xmath48 and @xmath49 , and @xmath50 as functions of @xmath51 .
figure [ fig_2 ] ( a)(c ) display , respectively , the surface energy density , pressures and sound velocity with parameters @xmath53 and @xmath74 as as functions of @xmath51 .
we see that the first three quantities decrease monotonically with the radius of the disk , as can be checked from eq .
( [ eq_en_sch ] ) , ( [ eq_p_sch ] ) and ( [ eq_sound_sch ] ) .
energy density decreases rapidly enough to , in principle , define a cut off radius and consider the disk as finite .
now we study some disks with halos constructed from several exact solutions of the einstein equations for static spheres of perfect fluid .
a survey of these class of solutions is presented in @xcite .
the first situation that we shall study is similar to the one depicted in fig .
[ fig_schem2 ] wherein we start with a sphere of perfect fluid
. this case will not be exactly the same as the one presented in the mentioned figure because the sphere has no boundary .
hence the generated disk will be completely immersed in the fluid .
an example of exact solution of the einstein equations that represent a fluid sphere with no boundary is the the buchdahl solution that may be regarded as a reasonably close analogue to the classical lane - emden index 5 polytrope @xcite .
the metric functions for this solution are : @xmath75 where @xmath76 and @xmath77 are constants .
far from the origin the solution goes over into the external schwarzschild metric , when @xmath78 .
the density , pressure and sound velocity are given by : @xmath79 the condition @xmath80 is satisfied for @xmath81 . using eq .
( [ eq_sol_buch ] ) and eqs .
( [ eq_qtt])([eq_disk_surf ] ) , we get the following expressions for the energy density , pressure and sound velocity of the disk : @xmath82 ^ 3 } \mbox { , } \label{eq_en_buch}\\ p & = \frac{aka^2}{2 \pi \left[-a+\sqrt{1+k(r^2+a^2 ) } \right ] \left[a+\sqrt{1+k(r^2+a^2 ) } \right]^3 } \mbox { , } \label{eq_p_buch}\\ v^2 & = \frac{a\left[-a+2\sqrt{1+k(r^2+a^2 ) } \right]}{3\left[a-\sqrt{1+k(r^2+a^2 ) } \right]^2 } \mbox{. } \label{eq_v_buch}\end{aligned}\ ] ] the conditions @xmath83 and @xmath84 are both satisfied if @xmath85 .
figure [ fig_3 ] ( a)(d ) shows , respectively , @xmath19 , @xmath35 , @xmath40 and rotation curves , eq .
( [ eq_v_fi_buch ] ) , as functions of @xmath51 for the disk calculated from buchdahl s solution . , eq .
( [ eq_en_buch ] ) , ( b ) the pressure @xmath35 eq .
( [ eq_p_buch ] ) , ( c ) the velocity of sound @xmath40 eq .
( [ eq_v_buch ] ) , ( d ) the tangential velocity @xmath47 eq .
( [ eq_v_fi_buch ] ) for the disk with @xmath86 as function of @xmath51.,title="fig : " ] , eq .
( [ eq_en_buch ] ) , ( b ) the pressure @xmath35 eq .
( [ eq_p_buch ] ) , ( c ) the velocity of sound @xmath40 eq .
( [ eq_v_buch ] ) , ( d ) the tangential velocity @xmath47 eq .
( [ eq_v_fi_buch ] ) for the disk with @xmath86 as function of @xmath51.,title="fig : " ] + , eq .
( [ eq_en_buch ] ) , ( b ) the pressure @xmath35 eq .
( [ eq_p_buch ] ) , ( c ) the velocity of sound @xmath40 eq .
( [ eq_v_buch ] ) , ( d ) the tangential velocity @xmath47 eq .
( [ eq_v_fi_buch ] ) for the disk with @xmath86 as function of @xmath51.,title="fig : " ] , eq .
( [ eq_en_buch ] ) , ( b ) the pressure @xmath35 eq .
( [ eq_p_buch ] ) , ( c ) the velocity of sound @xmath40 eq .
( [ eq_v_buch ] ) , ( d ) the tangential velocity @xmath47 eq .
( [ eq_v_fi_buch ] ) for the disk with @xmath86 as function of @xmath51.,title="fig : " ] in figure [ fig_4 ] ( a)(b ) we show , respectively , the density @xmath28 together with pressure @xmath29 , and sound velocity @xmath87 of the halo along the axis @xmath6 for @xmath88 . note that in this solution there is no boundary of the fluid sphere : the disk is completely immersed in the fluid . and pressure @xmath29 eq . ( [ eq_rhop_buch ] ) and ( b ) the velocity of sound v eq . ( [ eq_v_buch ] ) for the halo with @xmath88 along the axis @xmath6.,title="fig : " ] and pressure @xmath29 eq . ( [ eq_rhop_buch ] ) and ( b ) the velocity of sound v eq .
( [ eq_v_buch ] ) for the halo with @xmath88 along the axis @xmath6.,title="fig : " ] the tangential velocity @xmath47 calculated from metric coefficients ( [ eq_sol_buch ] ) is @xmath89^{3/2}+a[1+k(a^2-r^2)]\right\ } } \mbox{.}\ ] ] for @xmath65 , eq .
( [ eq_v_fi_buch ] ) goes as @xmath90 .
the specific angular momentum follows from eq .
( [ eq_h ] ) and eq .
( [ eq_sol_buch ] ) : @xmath91^{1/4 } } { \sqrt{[1+k(r^2+a^2)]^2 - 4akr^2\sqrt{1+k(r^2+a^2)}-a^2[1+k(a^2-r^2 ) ] } } \mbox{.}\ ] ] now we shall study the generation of a disk solution with an halo exactly as the one depicted in fig .
[ fig_schem2 ] .
we start with a solution of the einstein equations in isotropic coordinates that represents a sphere with radius @xmath23 of perfect fluid that on @xmath92 will be continuously matched to the vacuum schwarzschild solution .
narlikar , patwardhan and vaidya @xcite gave the following two exact solutions of the einstein equations for a static sphere of perfect fluid characterized by the metric functions @xmath93 and @xmath94 , @xmath95 ^ 2 & \text { for } k=-2+\sqrt{2 } \text { , } \label{eq_nar_sol1c}\end{aligned}\ ] ] where @xmath96 are constants , and @xmath97 .
we shall refer to these solutions as npv 1a and npv 1b , respectively .
the density , pressure and sound velocity for for the solutions @xmath98 and @xmath94 , will be denoted by @xmath99 and @xmath100 , respectively .
we find , @xmath101 \right .
\notag \\ &
\left.+b_{1a } \left[3k^2 + 8(1+n)+4k(n+3 ) \right ] r^{2n } \right\ } \mbox { , } \label{eq_p1_nar1}\\ p_{1b } & = \frac{a_{1b}+b_{1b } \ln ( r)+2\sqrt{2}b_{1b}}{16 \pi c [ a_{1b}+b_{1b } \ln ( r)]}\mbox { , } \label{eq_p2_nar1 } \\
\mathrm{v}_{1a}^2 & = \frac{1}{k(k+4)(a_{1a}+b_{1a}r^{2n})^2 } \left\ { a_{1a}^2 \left [ -3k^2 + 8(n-1)+4k(n-3 ) \right ] \right .
\notag \\ & \left .
-b_{1a}^2r^{4n } \left [ 3k^2 + 8(1+n)+4k(n+3 ) \right ] -2a_{1a}b_{1a}r^{2n}\left [ 3k(k+4)+8(1-n^2 ) \right ] \right\ } \mbox{,}\label{eq_v1_nar1}\\ \mathrm{v}_{1b}^2 & = \frac{2b_{1b}^2r^{\sqrt{2}}}{[a_{1b}+b_{1b } \ln ( r)]^2 } \mbox{. } \label{eq_v2_nar1}\end{aligned}\ ] ] the condition of continuity of the metric functions @xmath102 given by ( [ eq_nar_sol1a])([eq_nar_sol1c ] ) and the corresponding functions in eq .
( [ eq_metrica_sch ] ) at the boundary @xmath92 leads to following expressions : @xmath103 @xmath104 has its maximum at @xmath105 , and @xmath106 at @xmath92 .
condition @xmath107 is satisfied if @xmath108 . using eq .
( [ eq_nar_sol1a])([eq_nar_sol1c ] ) in eq .
( [ eq_qtt])([eq_disk_surf ] ) , we get the following expressions for the energy density , pressure and sound velocity of the disk : @xmath109}{\left [ a_{1a}+b_{1a}\mathcal{r}^{n}\right ] } \mbox { , } \label{eq_p1_nar1}\\ p_{1b } & = \frac{a}{4 \pi \sqrt{c}\mathcal{r}^{1/2+\sqrt{2}/4}[2a_{1b}+b_{1b } \ln ( \mathcal{r } ) ] } \left [ 2a_{1b}(\sqrt{2}-1)+2b_{1b } \right . \notag \\ & \left
. + b_{1b}(\sqrt{2}-1)\ln ( \mathcal{r})\right ] \mbox { , } \label{eq_p2_nar1}\\ v_{1a}^2 & = \frac{1}{k(k+4)\left [ a_{1a}\mathcal{r}^{-n/2}+b_{1a}\mathcal{r}^{n/2}\right]^2 } \left [ -b_{1a}^2\mathcal{r}^n(k^2 + 5k+nk+4n+4 ) \right . \notag \\ & \left .
+ a_{1a}^2\mathcal{r}^{-n } ( -k^2 - 5k+nk+4n-4 ) + 2a_{1a}b_{1a}(-k^2 - 5k+4n^2 - 4 ) \right ] \mbox{,}\label{eq_v1_nar1}\\ v_{1b}^2 & = \frac{1}{2(2a_{1b}+b_{1b } \ln ( \mathcal{r}))^2 } \left [ b_{1b}^2 \sqrt{2 } \ln^2 ( \mathcal{r } ) + 2b_{1b } ( 2b_{1b}+b_{1b}\sqrt{2}+2a_{1b}\sqrt{2})\right . \notag \\ & \left .
\ln ( \mathcal{r } ) + 4a_{1b}b_{1b}(2+\sqrt{2})+4\sqrt{2}a_{1b}^2 + 8b_{1b}^2 \right ] \mbox { , } \label{eq_v2_nar1}\end{aligned}\ ] ] where @xmath110 .
@xmath111 have their maximum values at @xmath112 . because the expressions are rather involved , the restrictions on the constants to ensure the velocities are positive and less then one is best done graphically .
the curves of @xmath19 , @xmath35 and @xmath40 as function of @xmath51 with parameters @xmath113 are displayed in figure [ fig_5 ] ( a)(c ) , respectively .
figure [ fig_6 ] ( a)(b ) show the density @xmath28 , pressure @xmath29 and velocity of sound v for the halo with parameters @xmath114 along the axis @xmath6 . the same physical quantities are shown in figures [ fig_7 ] ( a)(c ) and [ fig_8 ] ( a)(b ) with @xmath115 .
we note that @xmath19 and @xmath35 are continuous at the boundary between the internal and external parts of the disk , but the velocity of sound has a discontinuity .
( [ eq_en_nar1 ] ) , ( b ) the pressure @xmath35 eq .
( [ eq_p1_nar1 ] ) and ( c ) the velocity of sound @xmath40 eq .
( [ eq_v1_nar1 ] ) for the disk with @xmath116 as function of @xmath51.,title="fig : " ] eq .
( [ eq_en_nar1 ] ) , ( b ) the pressure @xmath35 eq .
( [ eq_p1_nar1 ] ) and ( c ) the velocity of sound @xmath40 eq .
( [ eq_v1_nar1 ] ) for the disk with @xmath116 as function of @xmath51.,title="fig : " ] + eq .
( [ eq_en_nar1 ] ) , ( b ) the pressure @xmath35 eq .
( [ eq_p1_nar1 ] ) and ( c ) the velocity of sound @xmath40 eq .
( [ eq_v1_nar1 ] ) for the disk with @xmath116 as function of @xmath51.,title="fig : " ] eq .
( [ eq_rho_nar1 ] ) and pressure @xmath29 eq .
( [ eq_p1_nar1 ] ) , ( b ) the velocity of sound v eq .
( [ eq_v1_nar1 ] ) for the halo with @xmath117 along the axis @xmath6.,title="fig : " ] eq .
( [ eq_rho_nar1 ] ) and pressure @xmath29 eq .
( [ eq_p1_nar1 ] ) , ( b ) the velocity of sound v eq .
( [ eq_v1_nar1 ] ) for the halo with @xmath117 along the axis @xmath6.,title="fig : " ] eq .
( [ eq_v_fi_nar1 ] ) and ( b ) the curves of @xmath118 eq . ( [ eq_h_nar1 ] ) with @xmath119 as function of @xmath51 .
a region of instability appears on the disk generated with parameter @xmath120.,title="fig : " ] eq .
( [ eq_v_fi_nar1 ] ) and ( b ) the curves of @xmath118 eq .
( [ eq_h_nar1 ] ) with @xmath119 as function of @xmath51 .
a region of instability appears on the disk generated with parameter @xmath120.,title="fig : " ] eq .
( [ eq_en_nar1 ] ) , ( b ) the pressure @xmath35 eq .
( [ eq_p2_nar1 ] ) , ( c ) the velocity of sound @xmath40 eq .
( [ eq_v2_nar1 ] ) for the disk with @xmath121 as function of @xmath51.,title="fig : " ] eq .
( [ eq_en_nar1 ] ) , ( b ) the pressure @xmath35 eq .
( [ eq_p2_nar1 ] ) , ( c ) the velocity of sound @xmath40 eq .
( [ eq_v2_nar1 ] ) for the disk with @xmath121 as function of @xmath51.,title="fig : " ] + eq .
( [ eq_en_nar1 ] ) , ( b ) the pressure @xmath35 eq .
( [ eq_p2_nar1 ] ) , ( c ) the velocity of sound @xmath40 eq .
( [ eq_v2_nar1 ] ) for the disk with @xmath121 as function of @xmath51.,title="fig : " ] eq .
( [ eq_rho_nar1 ] ) and pressure @xmath29 eq .
( [ eq_p2_nar1 ] ) , ( b ) the velocity of sound v eq .
( [ eq_v2_nar1 ] ) for the halo with @xmath122 along the axis @xmath6.,title="fig : " ] eq .
( [ eq_rho_nar1 ] ) and pressure @xmath29 eq . ( [ eq_p2_nar1 ] ) , ( b ) the velocity of sound v eq .
( [ eq_v2_nar1 ] ) for the halo with @xmath122 along the axis @xmath6.,title="fig : " ] eq .
( [ eq_v_fi_nar2 ] ) , ( b ) the curves of @xmath118 eq .
( [ eq_h_nar2 ] ) with @xmath123 as function of @xmath51 . as in the previous case , the same region of instability occurs.,title="fig : " ] eq .
( [ eq_v_fi_nar2 ] ) , ( b ) the curves of @xmath118 eq .
( [ eq_h_nar2 ] ) with @xmath123 as function of @xmath51 .
as in the previous case , the same region of instability occurs.,title="fig : " ] the tangential velocity @xmath47 is given by @xmath124 [ r^2(k+2)+2a^2 ] } \mbox { , } \label{eq_v_fi_nar1 } \\
\mathrm{v}_{c1b}^2 & = r^2\frac{2\sqrt{2}a_{1b}+b_{1b}[4+\sqrt{2}\ln ( r^2+a^2)]}{[2a_{1b}+b_{1b}\ln ( r^2+a^2 ) ] [ 2a^2+\sqrt{2}r^2 ] } \mbox { , } \label{eq_v_fi_nar2}\end{aligned}\ ] ] and the specific angular momentum @xmath68 , @xmath125 } { 2a^2[a_{1b}+b_{1b}\ln(\sqrt{r^2+a^2})]-2b_{1b}r^2}}\mbox{. } \label{eq_h_nar2 } \end{aligned}\ ] ] in figure [ fig_9 ] ( a)(b ) , the curves of tangential velocity eq . ( [ eq_v_fi_nar1 ] ) and @xmath118 eq .
( [ eq_h_nar1 ] ) , respectively , are displayed as functions of @xmath51 with @xmath126 .
the same quantities are shown in figure [ fig_10](a)(b ) with @xmath115 .
for @xmath120 the disks have a small region of unstable orbits immediately after the `` boundary radius '' . like in the previous subsections we study the generation of a disk solution with an halo exactly as the one depicted in fig .
[ fig_schem2 ] .
we also start with a solution of the einstein equations in isotropic coordinates that represents a sphere of radius @xmath23 of perfect fluid that on @xmath92 will the continuously matched to the vacuum schwarzschild solution .
we will use another two solutions found by narlikar , patwardhan and vaidya @xcite that we shall refer as npv 2a and npv 2b , respectivelly . that are characterized by the metric functions @xmath127 and @xmath128 , @xmath129 ^ 2}{\left ( a_1r^{1/\sqrt{2}}+a_2r^{-1/\sqrt{2}}\right)^2 } & \text { for } n=\sqrt{2}. \label{eq_nar_sol2c}\end{aligned}\ ] ] where the @xmath76 s and @xmath130 s are constants and @xmath131 .
the solution @xmath132 with @xmath133 corresponds to schwarzschild s internal solution in isotropic coordinates ( see for instance ref .
this solution has constant density and is conformally flat when @xmath134 .
the density , pressure and sound velocity for the solutions @xmath135 and @xmath128 , will be denoted by @xmath136 and @xmath137 , respectively .
we find , @xmath138 \mbox { , } \label{eq_rho_nar2}\\ p_{2a } & = \frac{1}{32 \pi } \left [ -12n^2a_1a_2+(3n^2 - 4)\left ( a_1r^{n/2}+a_2r^{-n/2}\right)^2\right . \notag \\ & \left .
+ \frac{2nx(b_{2a}-b_{1a}r^x)}{b_{2a}+b_{1a}r^x}\left ( a_1 ^ 2r^{n}-a_2 ^ 2r^{-n}\right ) \right ] \mbox { , } \label{eq_pa_nar2}\\ p_{2b } & = \frac{1}{16 \pi ( b_{1b}+b_{2b } \ln ( r ) ) } \left [ ( b_{1b}+b_{2b } \ln(r ) ) \left ( a_1 ^ 2r^{\sqrt{2}}+a_2 ^ 2r^{-\sqrt{2}}-10a_1a_2 \right ) \right . \notag \\ & \left .
+ 2\sqrt{2}b_{2b } \left ( a_2 ^ 2r^{-\sqrt{2}}-a_1 ^ 2r^{\sqrt{2 } } \right ) \right ] \mbox { , } \label{eq_pb_nar2}\\ \mathrm{v}_{2a}^2 & = \frac{1}{(4-n^2)\left ( a_1 ^ 2r^{n}-a_2 ^ 2r^{-n}\right)(b_{2a}+b_{1a}r^x)^2 } \left\ { b_{2a}^2 \left[a_1 ^ 2r^n(3n^2 + 2nx-4)\right . \right . \notag \\ & \left .
+ a_2 ^ 2r^{-n}(-3n^2 + 2nx+4 ) \right]+b_{1a}^2r^{2x } \left [ a_1 ^ 2r^n(3n^2 - 2nx-4 ) \right .
\right . \notag \\ & \left .
+ a_2 ^ 2r^{-n}(-3n^2 - 2nx+4 ) \right]+2b_{1a}b_{2a}r^x \left[a_1 ^ 2r^n(3n^2 - 2x^2 - 4)\right .
\right . \notag \\ & \left .
+ a_2 ^ 2r^{-n}(-3n^2 + 2x^2 + 4 ) \right]\right\ } \mbox { , } \label{eq_va_nar2}\\ \mathrm{v}_{2b}^2 & = \frac{1}{[b_{1b}+b_{2b } \ln ( r)]^2\left ( a_1 ^ 2r^{\sqrt{2}}-a_2 ^ 2r^{-\sqrt{2}}\right ) } \left\ { a_1 ^ 2r^{\sqrt{2 } } \left [ ( b_{1b}+b_{2b } \ln ( r ) ) \right . \right . \notag \\ & \left .
\times ( b_{1b}+b_{2b } \ln ( r)-2\sqrt{2}b_{2b})+2b_{2b}^2 \right]- a_2 ^ 2r^{-\sqrt{2 } } \right . \notag \\ & \left .
\times \left [ ( b_{1b}+b_{2b } \ln ( r))(b_{1b}+b_{2b } \ln ( r)+2\sqrt{2}b_{2b})+2b_{2b}^2 \right]\right\ } \mbox{. } \label{eq_vb_nar2}\end{aligned}\ ] ] the condition of continuity of the metric functions @xmath102 given by ( [ eq_nar_sol2a])([eq_nar_sol2c ] ) and the corresponding functions in eq .
( [ eq_metrica_sch ] ) at the boundary @xmath92 leads to following expressions : @xmath139 \text { , } \label{eq_cond_sol2a}\\ a_2 & = \frac{1}{r_b^{2-n/2}\left ( 1+\frac{m}{2r_b } \right)^3 } \left [ -\frac{m}{2n}+r_b \left ( \frac{1}{2}+\frac{1}{n}+\frac{m}{4r_b } \right ) \right ] \text { , } \label{eq_cond_sol2b}\\ b_{1a } & = \frac{-4r_b^2\left(1-\frac{2m}{r_b}\right)-m^2 + 2xr_b^2 \left(1-\frac{m^2}{4r_b^2}\right)}{4xr_b^{3+x/2}\left(1+\frac{m}{2r_b}\right)^4 } \mbox { , } \label{eq_cond_sol2c}\\ b_{2a } & = \frac{4r_b^2\left(1-\frac{2m}{r_b}\right)+m^2 + 2xr_b^2 \left(1-\frac{m^2}{4r_b^2}\right)}{4xr_b^{3-x/2}\left(1+\frac{m}{2r_b}\right)^4 } \mbox { , } \label{eq_cond_sol2d}\\ b_{1b } & = \frac{1}{4r_b^3\left ( 1+\frac{m}{2r_b } \right)^4 } \left[4r_b^2-m^2+(m^2 - 8mr_b+4r_b^2 ) \ln(r_b ) \right ] \mbox { , } \label{eq_cond_sol2e}\\ b_{2b } & = -\frac{m^2 - 8mr_b+4r_b^2}{4r_b^3\left ( 1+\frac{m}{2r_b } \right)^4 } \mbox{. } \label{eq_cond_sol2f}\end{aligned}\ ] ] @xmath140 has its maximum at @xmath92 , and @xmath141 at @xmath105 . using eq .
( [ eq_nar_sol2a])([eq_nar_sol2c ] ) in eq .
( [ eq_qtt])([eq_disk_surf ] ) , we get the expressions for the energy density , pressure and sound velocity of the disk : @xmath142 \text { , } \label{eq_en_nar2}\\ p_{2a } & = -\frac{a}{8 \pi \left ( b_{1a}\mathcal{r}^{1/2+x/4}+b_{2a}\mathcal{r}^{1/2-x/4 } \right ) } \left [ b_{1a}a_1(2 + 2n - x)\mathcal{r}^{(x+n)/4 } \right . \notag \\ & \left .
+ b_{1a}a_2(2 - 2n - x)\mathcal{r}^{(x - n)/4}+b_{2a}a_1(2 + 2n+x)\mathcal{r}^{(-x+n)/4 } \right .
\notag \\ & \left .
+ b_{2a}a_2(2 - 2n+x)\mathcal{r}^{-(x+n)/4 } \right ] \mbox { , } \label{eq_pa_nar2}\\ p_{2b } & = -\frac{a}{4 \pi [ 2b_{1b}+b_{2b}\ln ( \mathcal{r})]}\left\ { 2(1+\sqrt{2})b_{1b}a_1\mathcal{r}^{-1/2 + \sqrt{2}/4}\right . \notag \\ & \left .
+ 2(1-\sqrt{2})b_{1b}a_2\mathcal{r}^{-1/2-\sqrt{2}/4 } + \left [ ( 1+\sqrt{2 } ) \ln ( \mathcal{r})-2 \right ] b_{2b}a_1\mathcal{r}^{-1/2+\sqrt{2}/4}\right . \notag \\ & \left .
+ \left [ ( 1-\sqrt{2 } ) \ln ( \mathcal{r})-2 \right ] b_{2b}a_2\mathcal{r}^{-1/2-\sqrt{2}/4 } \right\ } \mbox { , } \label{eq_pb_nar2}\\ v_{2a}^2 & = \frac{1}{2(4-n^2)\left[a_2+a_1\mathcal{r}^{n/2}\right ] \left [ b_{2a}+b_{1a}\mathcal{r}^{x/2}\right]^2 } \left\ { a_1\mathcal{r}^{n/2 } \right . \notag \\ & \left .
\left [ b_{1a}^2(n-2)(2n - x+2)\mathcal{r}^{x}+b_{2a}^2(n-2)(2n+x+2 ) \right .
\right . \notag \\ & \left .
\left . -4b_{1a}b_{2a}(-n^2+n+x^2 + 2)\mathcal{r}^{x/2 } \right ] + a_2 \left [ b_{1a}^2(n+2)(2n+x-2)\mathcal{r}^{x } \right .
\right . \notag \\ & \left
+ b_{2a}^2(n+2)(2n - x-2)-4b_{1a}b_{2a}(-n^2-n+x^2 + 2)\mathcal{r}^{x/2 } \right ] \right\ } \mbox { , } \label{eq_va_nar2}\\ v_{2b}^2 & = - \frac{2b_{1b}+2b_{2b}+b_{2b}\ln ( \mathcal{r})}{2[2b_{1b}+b_{2b}\ln ( \mathcal{r})]^2[a_2+a_1\mathcal{r}^{1/\sqrt{2}}]}\left\ { \left [ \sqrt{2}\ln ( \mathcal{r})-4 \right]\right .
\notag \\ & \left .
\times b_{2b}a_1\mathcal{r}^ { 1/\sqrt{2}}-\left [ \sqrt{2}\ln ( \mathcal{r})+4 \right]b_{2b}a_2 + 2\sqrt{2}b_{1b}\left [ a_1\mathcal{r}^{1/\sqrt{2}}-a_2 \right ] \right\ } \mbox { , } \label{eq_vb_nar2}\end{aligned}\ ] ] where @xmath110 .
the curves of @xmath19 , @xmath35 and @xmath40 as function of @xmath51 with parameters @xmath143 are displayed in figure [ fig_11 ] ( a ) ( c ) , respectively .
figure [ fig_12 ] ( a ) ( b ) shows the density @xmath28 , pressure @xmath29 and velocity of sound v for the halo with parameters @xmath144 along the axis @xmath6 . the same physical quantities are shown in figures [ fig_13 ] and [ fig_14 ] with @xmath145 .
( [ eq_en_nar2 ] ) , ( b ) the pressure @xmath35 eq .
( [ eq_pa_nar2 ] ) , ( c ) the velocity of sound @xmath40 eq .
( [ eq_va_nar2 ] ) for the disk with @xmath146 as function of @xmath51.,title="fig : " ] eq .
( [ eq_en_nar2 ] ) , ( b ) the pressure @xmath35 eq .
( [ eq_pa_nar2 ] ) , ( c ) the velocity of sound @xmath40 eq .
( [ eq_va_nar2 ] ) for the disk with @xmath146 as function of @xmath51.,title="fig : " ] + eq .
( [ eq_en_nar2 ] ) , ( b ) the pressure @xmath35 eq .
( [ eq_pa_nar2 ] ) , ( c ) the velocity of sound @xmath40 eq .
( [ eq_va_nar2 ] ) for the disk with @xmath146 as function of @xmath51.,title="fig : " ] eq .
( [ eq_rho_nar2 ] ) and pressure @xmath29 eq . ( [ eq_pa_nar2 ] ) , ( b ) the velocity of sound v eq .
( [ eq_va_nar2 ] ) for the halo with @xmath147 along the axis @xmath6.,title="fig : " ] eq .
( [ eq_rho_nar2 ] ) and pressure @xmath29 eq .
( [ eq_pa_nar2 ] ) , ( b ) the velocity of sound v eq .
( [ eq_va_nar2 ] ) for the halo with @xmath147 along the axis @xmath6.,title="fig : " ] eq .
( [ eq_v_fi_nara ] ) , ( b ) the curves of @xmath118 eq .
( [ eq_h_nara ] ) with @xmath148 as function of @xmath51 .
the disks have no unstable orbits for these parameters.,title="fig : " ] eq .
( [ eq_v_fi_nara ] ) , ( b ) the curves of @xmath118 eq .
( [ eq_h_nara ] ) with @xmath148 as function of @xmath51 .
the disks have no unstable orbits for these parameters.,title="fig : " ] eq .
( [ eq_en_nar2 ] ) , ( b ) the pressure @xmath35 eq .
( [ eq_pb_nar2 ] ) , ( c ) the velocity of sound @xmath40 eq .
( [ eq_vb_nar2 ] ) for the disk with @xmath149 as function of @xmath51.,title="fig : " ] eq .
( [ eq_en_nar2 ] ) , ( b ) the pressure @xmath35 eq .
( [ eq_pb_nar2 ] ) , ( c ) the velocity of sound @xmath40 eq .
( [ eq_vb_nar2 ] ) for the disk with @xmath149 as function of @xmath51.,title="fig : " ] + eq .
( [ eq_en_nar2 ] ) , ( b ) the pressure @xmath35 eq .
( [ eq_pb_nar2 ] ) , ( c ) the velocity of sound @xmath40 eq .
( [ eq_vb_nar2 ] ) for the disk with @xmath149 as function of @xmath51.,title="fig : " ] eq .
( [ eq_rho_nar2 ] ) and pressure @xmath29 eq .
( [ eq_pb_nar2 ] ) , ( b ) the velocity of sound v eq .
( [ eq_vb_nar2 ] ) for the halo with @xmath150 along the axis @xmath6.,title="fig : " ] eq .
( [ eq_rho_nar2 ] ) and pressure @xmath29 eq .
( [ eq_pb_nar2 ] ) , ( b ) the velocity of sound v eq .
( [ eq_vb_nar2 ] ) for the halo with @xmath150 along the axis @xmath6.,title="fig : " ] eq .
( [ eq_v_fi_narb ] ) , ( b ) the curves of @xmath118 eq .
( [ eq_h_narb ] ) with @xmath151 as function of @xmath51 .
the disks have no unstable orbits for these parameters.,title="fig : " ] eq .
( [ eq_v_fi_narb ] ) , ( b ) the curves of @xmath118 eq .
( [ eq_h_narb ] ) with @xmath151 as function of @xmath51 .
the disks have no unstable orbits for these parameters.,title="fig : " ] the tangential velocity @xmath152 is given by @xmath153-a_1\mathcal{r}^{n/2}[b_{1a } ( n - x)\mathcal{r}^{x/2}+b_{2a}(n+x)]}{[b_{1a}\mathcal{r}^{x/2}+b_{2a}][a_1\mathcal{r}^{n/2}(2a^2-nr^2 ) + a_2(2a^2+nr^2 ) ] } \mbox { , } \label{eq_v_fi_nara } \\ \mathrm{v}_{c2b}^2 & = r^2\frac{a_2[2\sqrt{2}b_{1b}+b_{2b}(4+\sqrt{2}\ln ( \mathcal{r}))]-a_1\mathcal{r}^{\sqrt{2}/2 } [ 2\sqrt{2}b_{1b}+b_{2b}(-4+\sqrt{2}\ln ( \mathcal{r}))]}{[2b_{1b}+b_{2b}\ln ( \mathcal{r})][a_1\mathcal{r}^{\sqrt{2}/2 } ( 2a^2-\sqrt{2}r^2)+a_2(2a^2+\sqrt{2}r^2)]}\mbox { , } \label{eq_v_fi_narb}\end{aligned}\ ] ] and the specific angular momentum @xmath68 @xmath154 - a_1\mathcal{r}^{n/2}[b_{1a}(n - x)\mathcal{r}^{x/2}+b_{2a}(n+x ) ] \right\}^{1/2 } \mbox { , } \label{eq_h_nara}\\ h_{2b } & = \frac{r^2\mathcal{r}^{-1/2+\sqrt{2}/4}}{[a_1\mathcal{r}^{\sqrt{2}/2}+a_2]^{3/2 } \sqrt{4a^2b_{1b}+2b_{2b}[-2r^2+a^2\ln(\mathcal{r } ) ] } } \times \notag \\ & \left\ { a_2[2\sqrt{2}b_{1b}+b_{2b}(4+\sqrt{2 } \ln ( \mathcal{r } ) ) ] -a_1\mathcal{r}^{\sqrt{2}/2}[2\sqrt{2}b_{1b}+b_{2b}(-4+\sqrt{2}\ln ( \mathcal{r } ) ) ] \right\}^{1/2 } \mbox{. } \label{eq_h_narb}\end{aligned}\ ] ] in figure [ fig_15 ] ( a)(b ) , the curves of tangential velocities eq .
( [ eq_v_fi_nara ] ) and @xmath118 eq . ( [ eq_h_nara ] ) , respectively , are displayed as functions of @xmath51 with @xmath155 .
figure [ fig_16 ] shows the same quantities with @xmath145 . unlike solution 2 ,
no unstable circular orbits are present for the disks constructed with these parameters .
we study two examples of disks with halos constructed from spheres of fluids with two layers as the ones depicted in fig .
[ fig_schem3 ] .
let us consider a fluid sphere is formed by two layers : the internal layer , @xmath156 , will be taken as the internal schwarzschild solution ( solution 2a with n=2 ) , @xmath157 the external layer , @xmath158 is taken as the buchdahl solution , @xmath159 note that the external layer has no boundary , i.e. , this layer has infinite radius . according to the continuity conditions at @xmath160
, the constants are related through : @xmath161 } { \left [ ( 1+kr_1 ^ 2)^2 + 2c\sqrt{1+kr_1 ^ 2}+c^2(1-kr_1 ^ 2)\right ] } \mbox{,}\\ b_2 & = \frac{1-\frac{c}{\sqrt{1+kr_1 ^ 2}}}{\left ( 1+\frac{c}{\sqrt{1+kr_1 ^ 2}}\right)^3 } -b_1r_1 ^ 2 \mbox{. } \label{eq_cond_l2}\end{aligned}\ ] ] with these relations , one verifies that , using eq .
( [ eq_en_buch ] ) and eq .
( [ eq_en_nar2 ] ) , eq .
( [ eq_p_buch ] ) and eq .
( [ eq_pa_nar2 ] ) , both the energy density and the pressure are continuous at the radius @xmath162 of the disk .
figure [ fig_17 ] ( a)(c ) shows , respectively , @xmath19 , @xmath35 and @xmath40 for the disk obtained from fluid layers eq .
( [ eq_layer1])([eq_layer2 ] ) with parameters @xmath163 as function of @xmath51 .
the density @xmath28 , pressure @xmath29 and velocity of sound v for the halo along the @xmath6 axis with the same parameters for @xmath120 is shown in figure [ fig_18 ] . , ( b ) the pressure @xmath35 , ( c ) the velocity of sound @xmath40 for the disk generated from spherical fluid layers eq .
( [ eq_layer1])([eq_layer2 ] ) with @xmath164 as function of @xmath51.,title="fig : " ] , ( b ) the pressure @xmath35 , ( c ) the velocity of sound @xmath40 for the disk generated from spherical fluid layers eq .
( [ eq_layer1])([eq_layer2 ] ) with @xmath164 as function of @xmath51.,title="fig : " ] + , ( b ) the pressure @xmath35 , ( c ) the velocity of sound @xmath40 for the disk generated from spherical fluid layers eq .
( [ eq_layer1])([eq_layer2 ] ) with @xmath164 as function of @xmath51.,title="fig : " ] and pressure @xmath29 , ( b ) the velocity of sound v for the halo formed by fluid layers eq .
( [ eq_layer1])([eq_layer2 ] ) with @xmath165 along the axis @xmath6.,title="fig : " ] and pressure @xmath29 , ( b ) the velocity of sound v for the halo formed by fluid layers eq .
( [ eq_layer1])([eq_layer2 ] ) with @xmath165 along the axis @xmath6.,title="fig : " ] in figure [ fig_19 ] ( a)(b ) , the curves of tangential velocities and of @xmath118 , respectively , are displayed as functions of @xmath51 . ,
( b ) the curves of @xmath118 for the disk generated from fluid layers eq .
( [ eq_layer1])([eq_layer2 ] ) with @xmath164 as function of @xmath51 .
the disks obtained with these parameters have no unstable orbits.,title="fig : " ] , ( b ) the curves of @xmath118 for the disk generated from fluid layers eq .
( [ eq_layer1])([eq_layer2 ] ) with @xmath164 as function of @xmath51 .
the disks obtained with these parameters have no unstable orbits.,title="fig : " ] now we consider a sphere composed with two finite layers : the internal layer , @xmath166 , is taken as the npv solution 2b with @xmath145 , @xmath167 the external layer , @xmath168 , is taken as the npv solution 1b with @xmath115 , @xmath169 the spacetime outside the sphere , @xmath170 , will be taken as the schwarzschild s vacuum solution is isotropic coordinates .
@xmath171 in this case the pressure should be zero at @xmath172 .
the continuity conditions at @xmath160 and @xmath172 give the relations : @xmath173 \mbox{,}\\ b_1 & = \frac{r_1^{\sqrt{2}}}{2\sqrt{2c}}\left [ \left(a_3+b_3 \ln ( r_1)\right)\left ( 2\sqrt{2}r_1^{-\sqrt{2 } } -\ln ( r_1)+\ln ( r_1)r_1^{-\sqrt{2}}\right)-2\sqrt{2}b_3 \ln ( r_1 ) \right ] \mbox{. } \label{eq_cond_l4}\end{aligned}\ ] ] using eq .
( [ eq_en_nar1 ] ) and eq .
( [ eq_en_nar2 ] ) , the energy density of the disk at @xmath162 is continuous , but not the pressure .
the difference between eq .
( [ eq_pb_nar2 ] ) and eq .
( [ eq_p2_nar1 ] ) is @xmath174 \mbox{.}\ ] ] the pressure is continuous if @xmath175 .
figure [ fig_20 ] ( a)(c ) shows , respectively , @xmath19 , @xmath35 and @xmath40 for the disk obtained from fluid layers ( [ eq_layer3])([eq_layer4 ] ) with parameters @xmath176 as function of @xmath51 .
the density @xmath28 , pressure @xmath29 and velocity of sound v for the halo along the @xmath6 axis with the same parameters for @xmath177 is shown in figure [ fig_21 ] .
, ( b ) the pressure @xmath35 , ( c ) the velocity of sound @xmath40 for the disk generated from spherical fluid layers eq . ( [ eq_layer3])([eq_layer4 ] ) with @xmath178 as function of @xmath51.,title="fig : " ] , ( b ) the pressure @xmath35 , ( c ) the velocity of sound @xmath40 for the disk generated from spherical fluid layers eq .
( [ eq_layer3])([eq_layer4 ] ) with @xmath178 as function of @xmath51.,title="fig : " ] + , ( b ) the pressure @xmath35 , ( c ) the velocity of sound @xmath40 for the disk generated from spherical fluid layers eq .
( [ eq_layer3])([eq_layer4 ] ) with @xmath178 as function of @xmath51.,title="fig : " ] and pressure @xmath29 , ( b ) the velocity of sound v for the halo formed by fluid layers eq .
( [ eq_layer3])([eq_layer4 ] ) with @xmath179 along the axis @xmath6.,title="fig : " ] and pressure @xmath29 , ( b ) the velocity of sound v for the halo formed by fluid layers eq .
( [ eq_layer3])([eq_layer4 ] ) with @xmath179 along the axis @xmath6.,title="fig : " ] in figure [ fig_22 ] ( a)(b ) , the curves of tangential velocities and of @xmath118 , respectively , are displayed as functions of @xmath51 . in this case ,
regions of unstable orbits exist for parameters @xmath177 and @xmath180 . ,
( b ) the curves of @xmath118 for the disk generated from spherical fluid layers eq .
( [ eq_layer3])([eq_layer4 ] ) with @xmath181 as function of @xmath51 .
regions of unstable circular orbits appear for the disks obtained with parameters @xmath177 and @xmath182.,title="fig : " ] , ( b ) the curves of @xmath118 for the disk generated from spherical fluid layers eq .
( [ eq_layer3])([eq_layer4 ] ) with @xmath181 as function of @xmath51 .
regions of unstable circular orbits appear for the disks obtained with parameters @xmath177 and @xmath182.,title="fig : " ]
the `` displace , cut and reflect ''
method applied to solutions of einstein field equations in isotropic coordinates can generate disks with positive energy density and equal radial and azimuthal pressures ( perfect fluid ) . with solutions of static spheres of perfect fluid
it is possible to construct disks of perfect fluid surrounded also by perfect fluid matter . as far we know these are the first disk models of this kind in the literature .
all disks constructed as examples have some common features : surface energy density and pressures decrease monotonically and rapidly with radius . as the `` cut '' parameter @xmath0 decreases , the disks become more relativistic , with surface energy density and pressure more concentrated near the center .
also regions of unstable circular orbits are more likely to appear for high relativistic disks .
parameters can be chosen so that the sound velocity in the fluid and the tangential velocity of test particles in circular motion are less then the velocity of light .
this tangential velocity first increases with radius and reaches a maximum .
then , for large radii , it decreases as @xmath183 , in case of disks generated from schwarzschild and buchdahl s solutions .
the sound velocity is also a decreasing function of radius , except in solution npv 2a with @xmath184 , where it reaches its maximum value at the boundary . in principle
other solutions of static spheres of perfect fluid could be used to generate other disk + halo configurations , but it is not guaranteed that the disks will have the characteristics of normal fluid matter .
we believe that the presented disks can be used to describe more realistic model of galaxies than most of the already studied disks since the counter rotation hypothesis is not needed to have a stable configuration . in the table we list the seed metric coefficients , matching conditions at the boundaries and relevant physical quantities of all disks studied in this work .
the numbers refer to the equations presented along the paper and npv stands for narlikar , patwardhan and vaidya as before .
99 w. a. bonnor and a. sackfield , _ commun .
* 8 * , 338 ( 1968 ) .
t. morgan and l. morgan , _ phys .
rev . _ * 183 * , 1097 ( 1969 ) .
l. morgan and t. morgan , _ phys .
d _ * 2 * , 2756 ( 1970 ) .
g. gonzlez and p. s. letelier , _ class .
quantum grav . _ * 16 * , 479 ( 1999 ) .
t. ledvinka , m. zofka , and j. bik , in _ proceedings of the 8th marcel grossman meeting in general relativity _ , edited by t. piran ( world scientific , singapore , 1999 ) , pp .
339 - 341 .
p. s. letelier , _ phys .
* 60 * , 104042 ( 1999 ) .
j. katz , j. bik , and d. lynden - bell , _ class .
quantum grav . _
* 16 * , 4023 ( 1999 ) .
d. lynden - bell and s. pineault , mon . not .
r. astron .
185 , 679 ( 1978 ) j. p.
s. lemos , class .
grav . 6 , 1219 ( 1989 ) j. p.
s. lemos and p. s. letelier , class .
10 , l75 ( 1993 ) j. p. s. lemos and p. s. letelier , phys .
rev d49 , 5135 ( 1994 ) j. p. s. lemos and p. s. letelier , int .
d5 , 53 ( 1996 ) c. klein , class .
14 , 2267 ( 1997 ) j. bick , d. lynden - bell and j. katz , phys .
d47 , 4334 ( 1993 ) j. bick , d. lynden - bell and c. pichon , mon . not .
r. astron .
265 , 126 ( 1993 ) o. semerk , _ towards gravitating disks around stationary black holes _ , ( april , 2002 ) , available at http://xxx.lanl.gov/abs/gr-qc/0204025 . g. g. kuzmin , _ astron .
_ , * 33 * , 27 ( 1956 ) .
a. h. taub , j. math . phys .
21 , 1423 ( 1980 ) j.p.s .
lemos and p.s .
letelier , phys lett .
a 153 , 288 ( 1991 ) .
lord rayleigh , proc .
ser . a 93 , 148 ( 1916 ) .
l. d. landau , e. m. lifshitz , _ fluid mechanics _
( pergamon press , oxford , 1987 ) , 27
. b. kuchowicz , _ acta phys .
_ , * b * 3 , 209 ( 1972 ) .
h. a. buchdahl , _ astrophys .
j. _ , * 140 * , 1512 ( 1964 ) . | using the well - known `` displace , cut and reflect '' method used to generate disks from given solutions of einstein field equations , we construct static disks made of perfect fluid based on vacuum schwarzschild s solution in isotropic coordinates .
the same method is applied to different exact solutions to the einstein s equations that represent static spheres of perfect fluids .
we construct several models of disks with axially symmetric perfect fluid halos .
all disks have some common features : surface energy density and pressures decrease monotonically and rapidly with radius . as the `` cut '' parameter @xmath0 decreases ,
the disks become more relativistic , with surface energy density and pressure more concentrated near the center .
also regions of unstable circular orbits are more likely to appear for high relativistic disks .
parameters can be chosen so that the sound velocity in the fluid and the tangential velocity of test particles in circular motion are less then the velocity of light .
this tangential velocity first increases with radius and reaches a maximum . |
since the early fifties it is known from optical polarisation studies , that magnetic fields are constituents of the galactic interstellar medium .
the magnetic field strength is about a few @xmath2 g .
radio continuum observations clearly indicate synchrotron radiation originating high above the galactic plane .
thus , magnetic fields and cosmic rays are obviously constituents of the galactic halo .
but what is about the gas within the galactic halo ? already parker ( 1966 ) showed , that magnetic fields are always associated with the gaseous phase .
investigations in the uv - range show that highly ionised gas is common within the halo , but it is a long way from a pencil beam to the whole volume of the galactic halo .
recent investigations of the soft x - ray background data indicated the existence of a pervasive x - ray emitting plasma ( @xmath3k ) with a vertical scale height of about 4.4kpc ( pietz et al .
1998 ) within the halo .
moreover , a sensitive analysis of the leiden / dwingeloo survey gave evidence for a emission component with a high velocity dispersion of 60 @xmath4 ( kalberla et al .
1998 ) also detectable across the entire sky .
the discovery of both gas components within the galactic halo encouraged us to study the hydrostatic equilibrium model of the milky way once again . for this approach we studied recent all - sky surveys of gas , soft x - ray radiation , high energy @xmath0-ray emission , and radio - continuum emission .
to describe the large - scale properties of the milky way we used the approach of a hydrostatic halo model , as proposed by parker ( 1966 ) . to describe the gaseous disk - halo system , we identified 3 main constituents of the galactic interstellar medium , namely : the neutral interstellar gas with @xmath5 = 400 pc ( dickey & lockman , 1990 ) , the diffuse ionised gas ( dig ) with @xmath5 = 950 pc ( reynolds , 1997 ) , and halo gas with @xmath5 = 4.4 kpc ( kalberla et al .
1998 , and pietz et al . 1998 ) .
the major difference to the previous studies of the hydrostatic equilibrium of the milky way ( e.g. bloemen 1987 , boulares & cox 1990 ) is the detailed knowledge about the gas phase in the galactic halo .
in particular , the x - ray plasma in combination with the high - velocity dispersion component adds major physical parameters to our model .
1 displays the vertical density distributions of the gas phases ( diffuse neutral and ionised gas as well as the x - ray plasma ) in the solar vicinity .
2 gives an impression on the radial density distribution represented by the parameter @xmath6 according to taylor & cordes ( 1993 ) with @xmath7 kpc .
following parker s ( 1966 ) suggestion , we studied whether gas , magnetic fields and cosmic rays in the galactic halo may be in pressure equilibrium . indeed
, hydrostatic equilibrium models fit the all - sky - averaged observations best . in detail
we tested the hydrostatic equilibrium model by modelling the galactic synchrotron emission at 408 mhz as observed by haslam et al .
( 1982 ) , the @xmath0-ray emission as observed with _ egret _ at energies @xmath8 100 mev ( fichtel et al . 1994 ) as well as by modelling the galactic x - ray plasma distribution deduced from the all - sky survey data ( pietz et al .
a detailed discussion of the model calculations and a quantitative comparison with the observations are beyond the scope of this contribution ; for details we refer to kalberla & kerp ( 1998 ) . here
we summarise the main features of the model .
we found a pressure equilibrium between gas , magnetic fields and cosmic rays within the galactic halo .
the magnetic field of the galactic halo is globally regularly ordered and orientated parallel to the galactic plane .
in contrast to the halo the magnetic field within the disk is highly irregular and has only 1/3 of the gas pressure .
for a galaxy in hydrostatic equilibrium the 3-d distributions of gas pressure , density and gravitational potential are identical in size and shape .
accordingly , we can utilise our parameterisation of the milky way to deduce the gravitational potential _ and _ the dark matter content . in a simple view , the galaxy consists of 3 main parts : the galactic bulge , the stellar disk with a radial scale length of 4.5 kpc and the gaseous halo as described above . assuming that the gaseous halo traces the dark matter distribution we optimised the density of the gaseous halo component until the rotation velocity of the modelled distribution was in quantitative agreement with the observed rotation velocities ( i. e. fich el al . ,
1990 ) within galactocentric radii 3 @xmath9 25 kpc .
3 shows the corresponding rotation curve .
the total mass of the galaxy within @xmath10 = 50 kpc derived from our model is m=@xmath11 , consistent with m=@xmath12 ( little & tremaine , 1987 ) and also within the uncertainties with the results of kochanek ( 1996 ) of m=@xmath13 . in fig .
4 we show the gravitational acceleration @xmath14 in the solar neighbourhood as a function of @xmath15 deduced from our model in comparison to that of kuijken & gilmore ( 1989 ) and bienam et al .
( 1987 ) . within vertical distances of @xmath16 kpc our model ( solid line )
is in excellent agreement with @xmath14 derived by kuijken & gilmore ( 1989 ) ( dotted line ) and bienam et al .
( 1987 ) ( dashed line ) .
the differences at larger @xmath17 distance is because of different model assumptions on the dark matter distribution .
the turn - over of our model about 5 kpc above the disk is because of the radial dependence of @xmath14 , as shown in fig.5 ( the solar radius is marked by the dotted line ) .
the large scale properties of the galactic halo are very well modelled assuming that the main constituents of the interstellar matter , the gas , the magnetic fields , and the cosmic rays are in hydrostatic equilibrium .
we analysed recent all - sky surveys of gas , soft x - ray radiation , high energy @xmath0-rays and synchrotron radiation to test the model assumptions . in general
we find good quantitative agreement between model and data .
the assumption that the gaseous halo traces the dark matter in the galaxy leads to a total mass which is consistent with the observed rotation curve .
bienaym o. , robin a.c . , crz m. , 1987 , a&a 180 , 94 bloemen j.b.g.m . , 1987 , apj 322 , 694 boulares a. , cox d.p . , 1990 ,
apj 365 , 544 ( b&c ) buote d.a , canizares c.r . , 1996 ,
apj 457 , 565 dickey j.m . ,
lockman f.j . , 1990 , ara&a , 28 , 215 fich m. , blitz l. , stark a.a . , 1990 , apj 342 , 272 fichtel c.e . , bertsch d.l . ,
chiang , j. , et al . , 1994 ,
apjs 94 , 551 haslam c.g . , stoffel h. , salter c.j .
, wilson w.e . , 1982 ,
a&as 47 , 1 kalberla p.m.w . , kerp j. , 1998 , a&a 339 , 745 kalberla p.m.w . , westphalen g. , mebold u. , hartmann dap .
, burton w.b . , 1998 , a&a 332 , l61 little b. , tremaine s. , 1987 , apj 320 , 493 kochanek c.s .
, 1996 , apj 457 , 228 kuijken k. , gilmore g. , 1989 , mnras 239 , 605 little b. , tremaine s. , 1987 , apj 320 , 493 parker e.n .
, 1966 , apj 145 , 811 pietz j. , kerp j. , kalberla p.m.w . , et al . , 1998a , a&a 332 , 55 reynolds r.j .
1997 , in proc . of 156 .
we - heraeus - seminar on `` the physics of galactic halos '' , eds .
lesch et al . , akademie verlag , berlin , 57 taylor j.h .
, cordes j.m . ,
1993 , apj 411 , 674 leo blitz : i m concerned about the reality of the 60 km / s component .
first , i do nt see how it s possible to have a static ism with such highly supersonic velocities .
second , the component is not seen in the bell labs survey which has very good instrumental sidelobe response .
i can understand how baselines can introduce features in a spectrum , but i do nt see how a bad baseline can coincidentally remove a feature from all over the sky .
peter kalberla : the reality of the 60 km / s component was discussed in detail by kalberla et al .
( 1998 , a&a 332 , l61 ) . for a comparison between the bell labs and leiden / dwingeloo surveys concerning broad emission lines
see kalberla et al .
( 1997 , in proceedings of the iau colloquium no .
166 `` the local bubble and beyond '' , eds .
d. breitschwerdt , m.j .
freyberg , j. trmper , lecture notes in physics 506 , 475 ) .
the fact that the observed line width of 60 km / s exceeds the thermal hi line width significantly does not imply that the hi gas is in supersonic motion .
the hi gas in the halo has a volume filling factor of 0.12 only .
one has to consider the motion of individual hi eddies with respect to the surrounding plasma . since the sound velocity of this plasma is more than twice as high as the typical hi eddy velocity of 60 km / s , the motions of the hi clumps are clearly _ sub_sonic .
dissipative cloud - cloud collisions are of little relevance in a multiphase halo . | we investigated a hydrostatic equilibrium model of the milky way following parker ( 1966 ) , to constrain the large scale properties of the interstellar medium . in our approach
we found an excellent agreement between our simple hydrostatic equilibrium model of the milky way and the recent all - sky survey data rangeing from the @xmath0-ray to the radio regime . on large scales the galactic disk - halo system
is found to be stable against parker - instabilities .
pressure support from the galactic disk is essential to stabilise the halo .
in particular the diffuse ionised gas layer acts as a disk - halo interface . assuming that the distribution of the soft x - ray emitting plasma traces the gravitational potential , we derived the dark matter content of the milky way to be about @xmath1 .
our findings are consistent with the rotation curve of the galaxy . |
many of the x - ray sources in the rosat all - sky survey have been identified optically in the hamburg objective prism survey ( hagen et al . 1995 ) , among which are several cataclysmic variables ( cvs ) ( jiang et al .
the source rxj0944.5 + 0357 (= 1rxsj094432.1 + 035738 ; hereafter rxj0944 ) , in the constellation sextans , was observed spectroscopically by jiang et al . and found to have hi and hei emission lines typical of a cv .
further spectroscopic study by mennickent et al .
( 2002 ) showed the presence of absorption bands in the red , characteristic of a secondary with a spectral type near m2 .
observations by the vsnet group have identified two dwarf nova - like outbursts , in january and june 2001 , during which rxj0944 rose to v @xmath0 13 from its quiescent magnitude of v @xmath0 16.2 .
mennickent et al .
confirmed the spectroscopically determined orbital period ( @xmath1 ) of 0.1492 d ( 3.581 h ) reported to them by thorstensen & fenton .
mennickent et al .
also provided the first high speed photometry of rxj0944 in which large amplitude variations ( @xmath0 0.5 mag ) were found on time scales of 10 min to 2 h. they did not report any coherent signals in their photometry .
we have used the university of cape town ccd photometer ( odonoghue 1995 ) , attached to the 74-in and 40-in telescopes at the sutherland site of the south african astronomical observatory , to observe rxj0944 at time resolutions down to 6 s. table 1 gives the log of our photometric observations and figure [ fig1 ] shows the resulting light curves .
.observing log . [ cols="^,^,^,^,^,^,^ " , ] notes : ` : ' denotes an uncertain value , @xmath2 is the integration time .
[ tab1 ] a fourier transform ( ft ) of the entire data set shows no power at the spectroscopic period or its first harmonic , so we deduce that rxj0944 is of quite low inclination . from the radial velocity amplitude of 75 km
s@xmath3 mennickent et al .
reasoned that the inclination probably lies in the range @xmath4 ; our result indicates that it is probably at the lower end of this range .
a low inclination is also compatible with the weakness of the emission lines in the spectrum .
it was obvious early in our work that rxj0944 has a repetitive brightness modulation with a period @xmath0 2000 s. with further observations it could be seen that the feature is a double humped profile , with the two humps varying independently and rapidly in amplitude . in figure [ fig2 ]
we show the light curve of run s6324 on a larger scale , with the cyclic modulation marked , and its highly variable pair of peaks .
the ft for this run discloses a fundamental period at @xmath0 2220 s plus its first harmonic .
there are only six cycles of this modulation in the light curve , so the uncertainty of the period is large ( at least @xmath0 40 s ) .
the mean light curve , folded on the fundamental period of 2162 s as derived below , is given in figure [ fig3 ] and shows the double humped nature of the profile , and that the humps sit on plateaux with only short - lived dips between them .
( we removed the strong flare seen at hjd 2452356.418 in figure [ fig2 ] as being not representative ; it probably resulted from a sudden short - lived surge of mass transference . ) in the mean light curve , the two peaks occur at about phases 0.26 and 0.68 , respectively .
the peaks on the plateau appear as flares of variable width , so that adding more observations tends to even out their contributions , with the result that the mean light curve for the entire data set ( using the period of 2162 s ) , shown in figure [ fig4 ] , has largely lost the evidence for the doubling of the profile .
the ft for the full set of observations is given in figure [ fig5 ] , and shows clearly the humps of power near the @xmath0 2000 s fundamental and its first and second harmonics .
there is a great deal of complicated fine structure in the ft , beyond what is produced by the window pattern ; this is caused by the rapid amplitude modulation of the fundamental and its harmonics .
it is not possible to select unambiguous frequencies from the forest of aliases .
however , the highest peak in the neighbourhood of the fundamental modulation is at 2162 s and the highest peak at the first harmonic is 1079 s , which supports the choice of a fundamental period near 2160 s. there are other humps of power in the total ft , but by subdividing our data ( in particular , treating the march and april data sets separately ) we find that the ft is non - stationary only the 2160 s modulation and its harmonics are persistent features . given the high activity in the light curves ( figure [ fig1 ] ) it is not surprising that the ft is also very variable .
we find no evidence for rapid oscillations in brightness ( dwarf nova oscillations typically with periods in the range 550 s : see warner 1995 ) , but in run s6341 we find a quasi - periodic oscillation ( qpo ; see warner 1995 ) with a mean period of 351 s and amplitude 0.013 mag .
this is clearly seen in the light curve and maintains coherence for about 6 cycles between each major change of phase .
the presence of two distinct coherent periodicities in a cv is the recognised signature of an intermediate polar ( ip ) in which the non - orbital modulation is the spin period ( @xmath5 ) of the white dwarf primary , or its orbital side band ( see , e.g. , warner 1995 ) .
x - ray emission is another common feature of ips , resulting from accretion from the inner edge of the accretion disc onto the magnetic pole(s ) of the white dwarf . we therefore conclude that rxj0944 is most probably an ip with highly variable two - pole accretion . with @xmath1 = 3.581 h and @xmath5 = 36.0 min , rxj0944 is quantitatively similar to canonical ips such as fo aqr and tv col .
however , the double - humped light curve and other properties make it most similar to yy dra , as can be seen from the following brief review of the latter s properties .
yy dra is a dwarf nova at v @xmath0 16.0 quiescent magnitude with a mean outburst interval of 870 d and amplitude 5.5 mag , a @xmath1 of 3.96 h and a @xmath5 of 529.2 s. both the spin period and the orbital sideband ( at 550 s ) have been detected in the optical region ( patterson et al .
yy dra is the x - ray source 3a1148 + 719 and the spin modulation is seen in the x - ray emission ( patterson & szkody 1993 ) .
an m - type spectrum of the secondary is visible in the red , which is not normal for cvs with @xmath1 @xmath0 4h , suggesting a lower luminosity disc , probably the result if its central truncation by the magnetosphere of the primary .
hst observations of yy dra ( haswell et al . 1997 ) show that the uv emission line profiles are modulated at half the spin period , and that there is simultaneous presence of broad red and blue wings in civ emission , which is interpreted as evidence for two - pole accretion .
this is in accord with the double - humped pulse profiles in the optical and x - ray regions , and the occasional variations in height of the two peaks ( though yy dra is also notable for the near equality of its accretion pole luminosities most of the time ) . during an outburst of yy dra x - ray emission
greatly increased and the 529 s oscillation was usually visible , but near maximum disappeared , which is interpreted as possible due to the equality of accretion onto two extended poles ( szkody et al .
2002 ) .
there are other ips with evidence of two - pole accretion : v405 aur ( haberl et al .
1994 ) is similar in having its principal optical modulation at the first harmonic rather than the 545 s fundamental ( allan et al . 1996 ) ; the 1wga j1958.2 + 3232 ( @xmath5 = 1467 s ) has a double peaked profile which shows reversal of circular polarization between the peaks , confirming that it is a two - pole accretor ( norton et al .
2002 ) and still , duck & marsh ( 1998 ) have found spectroscopic evidence that rx j0558 + 5353 is a two - pole accretor .
the basic similarity of the optical photometric properties of rxj0944 and yy dra is evident , so we conclude that the same model , namely two - pole accretion , is the most probable description of rxj0944 , though with more variable and independent accretion rates onto the two poles .
the phases of the two peaks determined from the mean light curve ( figure [ fig3 ] ) are not half a cycle apart , indicating that the magnetic poles are not diametrically opposite on the surface of the primary . for an ip , however , the strengths of the heii and ciii / niii emission lines at 4686 and 4650 ( mennickent et al .
2002 ) are relatively weak but the fact that they are seen at all is uncharacteristic of an ordinary dwarf nova .
rxj0944 has the spectroscopic and photometric characteristics of an ip and would be worth studying in a pointed x - ray observation , in order to detect any modulation and thereby determine whether the 36.0 min periodicity in the optical is the white dwarf spin period or an orbital sideband .
a time - resolved spectroscopic study with the hst should also be undertaken .
rxj0944 also has interest as a dwarf nova
there are other intermediate polars that show full dwarf nova outbursts ( xy ari is an example ) and others that have abbreviated outbursts ( e.g. v1223 sgr ) .
high speed photometry during an outburst of rxj0944 could help to reveal the interaction between disc and magnetosphere as the rate of mass transfer increases and decreases .
we thank drs s. potter and p. rodriguez - gil for allowing us to use their light curve of rxj0944 .
paw is supported by funds from the university of cape town and the national research foundation ; bw is supported by funds from the university . | from optical photometry the cataclysmic variable rxj0944.5 + 0357 is shown to have a double - peaked pulse profile with a period @xmath0 2160 s. the two peaks vary rapidly in relative amplitude .
often most of the optical power is concentrated in the first harmonic of the 2160 s modulation ; rxj0944.5 + 0357 therefore probably belongs to the relatively rare class of two - pole accreting intermediate polars exemplified by yy dra and v405 aur . |
[ 1.1 ] a * quandle * @xmath0 is a magma satisfying the following properties : 1 . *
( idempotence ) * @xmath1 for all @xmath2 , 2 . *
( inverse ) * there exists @xmath3 such that @xmath4 for all @xmath5 , and 3 . *
( self - distributivity ) * @xmath6 for all @xmath7 .
the primordial example of a quandle arises when starting with a group @xmath8 and defining the quandle operation as @xmath9 for all @xmath10 .
the three properties satisfied by the elements of a quandle are the algebraic analogues of the conditions satisfied by quandle colorings under the three reidemeister moves as depicted in figures [ fig 1 ] and [ fig 2 ] . by relaxing some of the axioms of a quandle
, we obtain generalizations of quandles , including : 1 .
a * rack * is a magma which satisfies the first and second quandle axioms
. 2 . a *
spindle * is a magma which satisfies the first and the third axioms of a quandle .
a * shelf * is a magma which satisfies only the third quandle axiom .
4 . a * unital shelf * is a shelf @xmath0 equipped with an element @xmath11 , such that @xmath12 for all @xmath13 .
table [ assocspindle ] depicts an associative spindle that is not a quandle .
we continue with some basic observations , due to sam c. [ assocspindle ] @xmath14 [ 1.4 ] let @xmath0 be a unital shelf . then : 1 . for all @xmath15 2 . for all @xmath16 3 . for
all @xmath17 let @xmath5 , where @xmath18 is a unital shelf .
then : 1 . @xmath19 2 .
@xmath20 3 .
@xmath21 1 .
monoids satisfying condition ( 1 ) of proposition [ 1.4 ] are known as ` graphic monoids ' and have been studied in @xcite and @xcite .
2 . in proposition
[ 1.4 ] , the proofs of parts ( 0 ) and ( 1 ) require only the left - sided unity , whereas , for the proof of part ( 2 ) , both the left - sided and right - sided units are needed .
a shelf satisfying axioms ( 0 ) , ( 1 ) , and ( 2 ) of proposition [ 1.4 ] is called a * pre - unital shelf * , and a shelf satisfying properties ( 1 ) and ( 2 ) of proposition [ 1.4 ] is called a * proto - unital shelf*. proto - unital shelves are associative .
in particular , this implies unital shelves are associative .
let @xmath7 , where @xmath18 is a unital shelf .
then , @xmath22 we conclude by illustrating a relationship between pre - unital and unital shelves .
[ oneelt ] if @xmath23 is a pre - unital shelf , then @xmath24 is unital .
conversely , if @xmath0 is a unital shelf , then @xmath25 is pre - unital . extending a pre - unital shelf to a unital shelf requires verifying the self - distributivity property for all @xmath7 where at least one of these elements is the unit
this follows easily .
conversely , for two elements @xmath26 , neither of which are the unit , @xmath27 .
therefore , the result follows . in the sections that follow , we study various properties of associative shelves but not of associative racks and quandles , even though these two algebraic structures are more important from knot - theoretic point of view .
this is due to the fact that associate racks are trivial . if @xmath0 is an associative rack with @xmath5 , then @xmath28 . but
this implies , @xmath29 .
as @xmath0 is associative , we may ignore the parentheses .
hence , it follows that @xmath30 . [ 1.9 ] a * quasi - group * @xmath0 is a magma satisfying : 1 .
the equation @xmath31 has a unique solution in @xmath18 for all @xmath32 .
2 . the equation @xmath33 has a unique solution in @xmath18 for all @xmath32 .
finite quasi - groups are also known as ` latin squares ' and were introduced by euler in the late 1700 s .
the two axioms of a latin square @xmath18 of order @xmath34 essentially mean that the rows and columns of the multiplication table are permutations of the @xmath34 elements in @xmath18 .
table [ 2 ] illustrates a latin square of size 4 .
quandles satisfying the first axiom of definition [ 1.9 ] are called * quasi - group quandles*. @xmath35
let @xmath0 be an associative quasi - group .
if @xmath0 satisfies condition @xmath36 of proposition [ 1.4 ] ( in particular all elements of @xmath18 are idempotent ) , then @xmath18 has only one element .
let @xmath5 .
then we have @xmath37 . but
this implies @xmath38 .
hence , as @xmath39 was arbitrary , there is only one element in @xmath18 .
note that this result only requires the second property of quasi - groups .
using * microsoft visual c++ * , preliminary computations produce the data in table [ 3 ] regarding the number of associative shelves ( as ) , associative spindles ( asp ) , and unital shelves ( us ) of up to 4 elements .
however , much more has been computed for semi - groups satisfying different combinations of axioms ( @xcite ) .
in this section , we consider finitely generated objects and prove that they are finite .
moreover , for these objects , we find normal forms , count number of elements and provide generating functions .
we will denote our alphabet by @xmath41 , the monoid of words in alphabet @xmath41 with juxtaposition as the operation by @xmath42 , and the unit element of the monoid , namely the empty word , by @xmath43 .
if we delete the empty word we obtain the semi - group @xmath44 .
let @xmath45 .
then the * free associative shelf * on @xmath46 , denoted by @xmath47 is given by : @xmath48 where @xmath49 is an equivalence relation defined by : two words in @xmath50 are equivalent under @xmath49 if one can be transformed into the other by a finite number of right self - distributive operations with respect to juxtaposition .
note that it is enough to define @xmath49 on the generators of @xmath50 .
this can be proven by induction on the length of the words by considering three arbitrary words .
the structure of @xmath50 was studied by a czech group from prague .
the extensive survey of their work can be found in @xcite .
intuition suggests that to obtain a word in its reduced form , it is enough to consider reductive self - distributive operations .
however , this is not often the case .
we can reduce any word in @xmath51 uniquely to the normal form via the process described below .
the crucial observation is that doing so requires the relation @xmath52 : @xmath53 the above equations illustrate how associativity and right self - distributivity go hand in hand . for small values of @xmath34
, computations produce : @xmath54 @xmath55 table [ fas(2 ) ] displays the multiplication table of @xmath56 replacing the above words with integers from @xmath57 to @xmath58 in the order above .
similar calculations reveal the size of @xmath59 as 93 , which motivates the following proposition and theorem .
much more computational data appears in @xcite .
@xmath60 [ normal ] words in normal form in @xmath50 are characterized by the following family of words : 1 .
non - empty words in which every letter appears at most once , 2 .
words of the form @xmath61 , where @xmath62 is a word in which every letter appears at most once and the letter @xmath63 appears exactly once , and 3 .
words obtained from words in ( 2 ) by doubling the last letter , in particular the word @xmath64 are possible .
let @xmath65 be denoted by @xmath66 .
then : 1 .
@xmath66 is finite with cardinality @xmath67 2 .
@xmath66 satisfies the recursive relation @xmath68 3 .
@xmath66 satisfies the @xmath69-term recursive relation @xmath70 4 .
@xmath66 has exponential generating function @xmath71 1 .
this follows from proposition [ normal ] and @xcite .
this follows from the formula given in @xcite , appendix 2 .
this follows from ( 2 ) : @xmath72 4 .
this is straightforward from the theory of generating functions .
let @xmath45 .
then the * free proto - unital shelf * on @xmath46 , denoted by @xmath73 , is given by : @xmath74 where @xmath49 is an equivalence relation defined by : two words in @xmath73 are equivalent under @xmath49 if one can be transformed into the other using a finite number of the axioms of proto - unital shelves with respect to juxtaposition .
as noted before , any proto - unital shelf is associative but not vice versa .
the free proto - unital shelf @xmath75 has six elements : @xmath76 and is depicted in table [ fpus(2 ) ] .
@xmath77 let @xmath78 be the free proto - unital shelf over the alphabet @xmath41 .
then : 1 .
every word in @xmath78 can be uniquely reduced to a word of type @xmath63 , @xmath79 or @xmath62 , where @xmath62 has at least two letters and no letter is repeated in @xmath62 , 2 .
if @xmath80 is a finite alphabet of @xmath34 letters then , @xmath81 3 .
the exponential generating function of @xmath82 is @xmath83 1 .
this follows from lemma [ lemma 3.10 ] .
2 . this follows from counting elements in normal form .
this follows from a straightforward generating functions computation .
[ lemma 3.10 ] let @xmath41 be a set ( alphabet ) .
then the free proto - unital shelf @xmath78 is obtained from the semigroup of words over @xmath41 ( that is @xmath84 ) , by adding the following relations : 1 .
@xmath85 for any @xmath86 2 .
@xmath87 for any @xmath86 3 .
@xmath88 for any @xmath89 and @xmath90 in particular , every element of the free proto - unital shelf over the alphabet @xmath41 has a unique representative of type @xmath63 , @xmath79 or @xmath62 , where @xmath62 has at least two letters and no letter is repeated .
the main ingredient used here is the reduction of the word @xmath91 , because the number of letters actually increases before reducing : @xmath92 with the identity @xmath93 , the other reductions are immediate and a straightforward computation shows that words in normal ( reduced ) form with juxtaposition create a proto - unital shelf .
consider the subset @xmath94 composed of letters @xmath79 and @xmath62 without repetition .
the product here is defined by the composite @xmath95 where if @xmath96 for some @xmath89 , then @xmath97 and otherwise if @xmath95 uses more than one letter , we reduce it by deleting any letter which is not the last one of the type . for example , @xmath98 is reduced to @xmath99 . by construction , this subset is the largest possible so is thus @xmath78 .
let @xmath0 be a magma .
if there exists @xmath100 such that @xmath101 for all @xmath13 then @xmath102 is a * right @xmath103-fixed element*. if @xmath104 then @xmath105 is a * right zero * or *
right projector*. @xmath73 , has @xmath106 different right zeros ( right projectors ) , namely : 1 .
@xmath79 for @xmath107 , and 2 .
@xmath108 where @xmath109 is a permutation of alphabet letters from @xmath46 .
this follows from the structure of multiplication of words in normal form as described in the proof of lemma [ lemma 3.10 ] .
free pre - unital shelves and unital shelves are discussed together because , by the result of proposition [ oneelt ] , they differ only by one element ( namely the empty word in the presentation of the shelf by generators and relators ) .
let @xmath41 be an alphabet .
then the * free pre - unital shelf * on a , denoted by @xmath110 , is given by : @xmath111 where @xmath49 is an equivalence relation defined by : two words in @xmath110 are equivalent under @xmath49 if one can be transformed to the other using a finite number of the axioms of pre - unital shelves with respect to juxtaposition .
if @xmath80 is a finite set of @xmath34 elements , we write @xmath112 for @xmath110 . the free pre - unital shelf @xmath113 is illustrated in table [ fptildeus(2 ) ] .
@xmath114 for any alphabet , @xmath41 , we construct a pre - unital shelf @xmath110 and show that it is the free pre - unital shelf over alphabet @xmath41 , using the observation that pre - unital shelves are associative .
let @xmath41 be any set and @xmath110 is the set of of non - empty words in @xmath115 which have no repeating letters .
we define the * binary operation @xmath116 on @xmath110 * by : if @xmath117 then @xmath118 is obtained from @xmath95 ( juxtaposition ) by deleting from @xmath119 all letters which are also in @xmath120 .
[ 2.14 ] @xmath110 is a pre - unital shelf isomorphic to the free pre - unital shelf over alphabet @xmath41 .
first we show that @xmath110 is a pre - unital shelf .
self - distributivity and associativity are verified directly from definition .
namely , @xmath121 is obtained from @xmath122 by deleting from @xmath119 letters which are already in @xmath120 and deleting from @xmath118 letters which are already in @xmath123 .
when using the rules for @xmath116 operation we see that @xmath124 and @xmath125 give the same result .
furthermore , from the rule @xmath126 , idempotence follows immediately as does axiom ( 1 ) in proposition [ 1.4 ] : @xmath127 .
axiom ( 2 ) in proposition [ 1.4 ] follows from associativity and idempotence : @xmath128 . from proposition
[ 2.14 ] we can calculate the number of elements in @xmath112 and its exponential generating function .
let @xmath129 be denoted by @xmath130 .
then : 1 .
@xmath131 2 .
@xmath130 satisfies the recursion relation @xmath132 with @xmath133 3 .
@xmath130 has exponential generating function @xmath134 1 .
the irreducible ( normal form ) words in @xmath112 ( equivalently words in @xmath112 are all arrangements of @xmath135 different letters of the alphabet of @xmath34 letters ( empty word excluded ) .
we calculate the number as : @xmath136 , as desired .
this follows directly from the formula in ( 1 ) .
this follows from a straightforward generating functions computation . the free unital shelf @xmath137 is obtained from @xmath112 using proposition [ oneelt ] by adding empty word .
therefore , we have the following : 1 .
the number of elements in @xmath137 is one more than the number of elements in @xmath112 .
that is , @xmath138 2 .
the exponential generating function for @xmath139 is @xmath140 the material of this subsection was essentially proven by kimura in 1958 @xcite where he considered monoids satisfying the axiom @xmath141 ( where @xmath63 can be the empty word ) .
it was proven that free monoids with this axiom consist of words without repeating letters , hence in the case of @xmath34 generators , he obtained @xmath142 elements .
in this section , we first recall the definition of one - term distributive homology introduced in @xcite and two - term ( rack ) homology introduced in @xcite . following @xcite , we first define the chain modules ( denoted by @xmath143 ) and the face maps @xmath144 , satisfying @xmath145 for @xmath146 and @xmath147 , so that @xmath148 forms a pre - simplicial module . after defining the boundary map @xmath149 by @xmath150
, we obtain the chain complex @xmath151 .
we describe two pre - simplicial modules leading up to one - term and two - term ( rack ) homology , respectively .
let @xmath0 be a shelf and @xmath152 the free abelian group generated by the @xmath153-tuples @xmath154 of the elements in @xmath18 , that is @xmath155 .
we then have @xmath158 , and therefore @xmath159 is a pre - simplicial module . defining the boundary map as usual by @xmath160 , then @xmath161 is a chain complex whose @xmath162 homology group of @xmath18 is given by : @xmath163 to define two - term ( rack ) homology , we consider the trivial operation given by @xmath164 . notice that @xmath165 we define face maps @xmath166 by @xmath167 .
we then have @xmath168 , hence @xmath169 is a pre - simplicial module . defining the boundary map as usual by @xmath170 then @xmath171 is a chain complex whose @xmath162 two - term ( rack ) homology group of @xmath18 is given by : @xmath172 a similar , but stronger result also holds for two - term homology . before stating this result , namely theorem [ main theorem ] , we remark that in a shelf @xmath0 if @xmath102 is a right @xmath103-fixed element , then @xmath183 . therefore , @xmath103 is idempotent .
let @xmath0 be a shelf with a right @xmath103-fixed element @xmath102 .
let @xmath151 be the chain complex corresponding to the two - term ( rack ) homology .
then , there exists a sequence of maps @xmath185 which is a chain homotopy between the identity on @xmath143 and the constant map on @xmath186 , given by @xmath187 .
let @xmath0 be a shelf with an idempotent element @xmath103 and a right @xmath103-fixed element @xmath102 .
let @xmath188 be the chain complex corresponding to two - term ( rack ) homology of @xmath0 .
consider a sequence of maps ( chain homotopy ) @xmath185 , for all positive integers @xmath34 , given by : @xmath189 we will show that this sequence of maps is a chain homotopy between the identity chain map and the constant chain map @xmath190 .
let @xmath202 , be given by @xmath203 and let @xmath204 be an embedding given by @xmath205 .
then , @xmath206 and by lemma [ lemma1 ] , @xmath207 is homotopy equivalent to @xmath208 .
therefore , these chain complexes are chain homotopy equivalent .
let @xmath0 be a finite proto - unital shelf with at least two elements and no right zeros .
then there exist two distinct elements @xmath209 such that @xmath210 is not a right zero and @xmath211 .
otherwise , the only possibility is @xmath212 for all @xmath177 and @xmath213 for all @xmath214 .
but this is a contradiction since @xmath215 with @xmath216 implies @xmath217 .
therefore , @xmath218 , @xmath210 is not a right zero , @xmath219 and @xmath211 .
now , if @xmath220 is a right zero , we are done . therefore , suppose it is not .
thus , there exists @xmath221 such that @xmath222 . since @xmath18 is associative ,
parentheses can be ignored .
this implies @xmath223 since otherwise we have @xmath224 .
due to same argument as above , @xmath225 .
also , @xmath226 , since otherwise we have @xmath227 .
similarly , @xmath228 .
a similar inductive argument shows that @xmath229 can not be generated by @xmath230 . an analogous argument
is used for the element @xmath231 .
next , assume the element @xmath232 is not a right zero for some finite @xmath34 .
this implies there exists @xmath233 such that @xmath234 . as above , due to associativity of @xmath18 , parentheses can be ignored .
let @xmath235 , where @xmath236 so that we have @xmath237 if @xmath238 for some @xmath239 , then : @xmath240 the above equation follows from the second axiom of proto - unital shelves , @xmath241 for all @xmath5 . by the same argument
, we can reduce @xmath242 followed by @xmath243 until @xmath244 .
finally when @xmath245 , there are two possibilities . either @xmath246 , or they are equal . in the first case , the above argument can be used . for the second case ,
the reduction requires more work .
we have : this contradicts the fact that @xmath248 can be generated by the previous terms .
therefore , at every stage , it is possible to have a new element of @xmath18 which can not be generated by the previous elements .
but , this is a contradiction as we assumed @xmath0 was finite .
therefore , @xmath0 has at least one right zero element . 1
. @xmath137 has trivial one - term homology in all positive dimensions for all @xmath34 .
@xmath73 has two - term ( rack ) homology @xmath184 in all dimensions for any @xmath34 .
hence , in particular , the same holds for @xmath112 and @xmath137 as well .
choose a family of sets @xmath251 , not necessarily finite , and functions @xmath252 .
define the spindle product on @xmath253 for @xmath254 and @xmath255 by : @xmath256 subsets @xmath257 are called * blocks * of the spindle @xmath18 and the maps @xmath258 are called * block functions*. we write @xmath259 for the function induced by all the block functions .
one - term homology was initially thought to be torsion - free and hence less interesting than two - term ( rack ) homology . in @xcite
, there are several examples in which families of shelves are constructed . in @xcite , the one - term homology of one such infinite family was studied .
theorem 5.4 in @xcite states that any finite abelian group can be realized as the torsion subgroup of the second homology group of @xmath260-block spindles having a one element block .
but , every @xmath260-block spindle with a one element block has a right zero element .
therefore , there exist infinite families of spindles which can realize any finite commutative group as a torsion subgroup in one - term homology .
but , they have @xmath184 as their two - term homology group in all dimensions .
theorems [ 3.1 ] and [ main theorem ] can be further applied to determine the one - term and two - term ( rack ) homology groups of laver tables .
richard laver discovered these algebraic structures while studying set theory , and the combinatorial structure of laver tables has been extensively studied ( @xcite , @xcite and @xcite ) . in the following discussion ,
we use the convention of laver(@xcite , @xcite , @xcite , @xcite ) , that is the magmas are left self - distributive . in particular , to get right self - distributivity we transpose the laver tables .
it is not difficult to notice the pattern suggested from the data above , and thus the results for all other associative shelves of size four are not included .
this pattern holds for racks ( @xcite ) .
more computations on torsion in two - term homology for quandles were done in @xcite for example .
r. fenn , c. rourke , b. sanderson , an introduction to species and the rack space .
_ m.e.bozhuyuk ( ed ) , topics in knot theory ( proceedings of the topology conference , erzurum ) , nato adv .
inst . ser .
_ , 399 , kluver academic publishers , 33 - 35 , 1993 . j. jeek , t. kepka , selfdistributive groupoids part d1 : left distributive semigroups .
_ acta universitatis carolinae .
mathematica et physica _ , vol .
47 , 2006 , no . 1 , 15 - 56 .
e - print : https://eudml.org/doc/246241?lang=es&limit=15 v. lebed , cohomology of finite monogenic self - distributive structures . _
journal of pure and applied algebra , volume 220 , issue 2 , february 2016 , pages 711 - 734 .
e - print : http://arxiv.org/abs/1503.07030_ j. h. przytycki , s. y. yang , the torsion of a finite quasigroup quandle is annihilated by its order . _ journal of pure and applied algebra _ , 219 ( 10 ) , 2015 , 4782 - 4791 .
e - print : http://arxiv.org/abs/1411.1727 | homology theories for associative algebraic structures are well established and have been studied for a long time .
more recently , homology theories for self - distributive algebraic structures motivated by knot theory , such as quandles and their relatives , have been developed and investigated . in this paper , we study associative self - distributive algebraic structures and their one - term and two - term ( rack ) homology groups . .
] in may of 2015 , during a discussion of the origin of the word ` quandle ' on the _ n - category caf _ ( https://golem.ph.utexas.edu/category/ ) ,
sam c. remarked that unital shelves are associative and provided an elegant short proof .
the work herein was inspired by this post . in section
[ intro ] , we provide definitions and discuss some elementary propositions .
the first two of these propositions were proven by sam c. ( @xcite ) .
the latter propositions lay the pathway for the subsequent sections . following this
, we discuss a way to construct families of algebraic structures , in particular associative shelves and spindles .
section [ finitelygen ] addresses finitely generated associative shelves , their cardinalities and generating functions . following this , finitely generated proto - unital shelves , pre - unital shelves , and unital shelves are discussed . while some of the observations in this section are new , others are stated for completeness ( @xcite , @xcite and @xcite ) . in section [ homologyassoc ] , the two well - known homology theories for self - distributive algebraic structures , namely one - term distributive homology and two - term ( rack ) homology are defined and the main results concerning the homology groups of a special class of associative shelves are proven . in particular , we compute one - term and two - term ( rack ) homology of unital shelves .
the main theorem concerning two - term homology of shelves with right fixed elements extends to the two - term ( rack ) homology groups of laver tables ( @xcite , @xcite and @xcite ) and f - block spindles ( @xcite , @xcite ) . finally , section [ future ] and the appendix contain conjectures and computational data . |
in signal processing , a classical uncertainty principle limits the time - bandwidth product @xmath0 of a signal , where @xmath1 is the measure of the support of the signal @xmath2 , and @xmath3 is the measure of the support of its fourier transform @xmath4 ( cf .
, e.g. , @xcite ) .
a very elementary formulation of that principle is @xmath5 whenever @xmath6 and @xmath7 . in the inverse source problem ,
the far field radiated by a source @xmath8 is its _ restricted _ ( to the unit sphere ) _ fourier transform _ , and the operator that maps the restricted fourier transform of @xmath9 to the restricted fourier transform of its translate @xmath10 is called the _ far field translation operator_. we will prove an uncertainty principle analogous to , where the role of the fourier transform is replaced by the far field translation operator . combining this principle with a _
regularized picard criterion _ , which characterizes the non - evanescent ( i.e. , detectable ) far fields radiated by a ( limited power ) source supported in a ball provides simple proofs and extensions of several results about locating the support of a source and about splitting a far field radiated by well - separated sources into the far fields radiated by each source component .
we also combine the regularized picard criterion with a more conventional uncertainty principle for the map from a far field in @xmath11 to its fourier coefficients .
this leads to a data completion algorithm which tells us that we can deduce missing data ( i.e. on part of @xmath12 ) if we know _ a priori _ that the source has small support .
all of these results can be combined so that we can simultaneously complete the data and split the far fields into the components radiated by well - separated sources .
we discuss both @xmath13 ( least squares ) and @xmath14 ( basis pursuit ) algorithms to accomplish this .
perhaps the most significant point is that all of these algorithms come with bounds on their condition numbers ( both the splitting and data completion problems are linear ) which we show are sharp in their dependence on geometry and wavenumber .
these results highlight an important difference between the inverse source problem and the inverse scattering problem .
the conditioning of the linearized inverse scattering problem does not depend on wavenumber , which means that the conditioning does not deteriorate as we increase the wavenumber in order to increase resolution .
the conditioning for splitting and data completion for the inverse source problem does , however , deteriorate with increased wavenumber , which means the dynamic range of the sensors must increase with wavenumber to obtain higher resolution .
we note that applications of classical uncertainty principles for the one - dimensional fourier transform to data completion for band - limited signals have been developed in @xcite . in this classical
setting a problem that is somewhat similar to far field splitting is the representation of highly sparse signals in overcomplete dictionaries .
corresponding stability results for basis pursuit reconstruction algorithms have been established in @xcite . the numerical algorithms for far field splitting that we are going to discuss have been developed and analyzed in @xcite .
the novel mathematical contribution of the present work is the stability analysis for these algorithms based on new uncertainty principles , and their application to data completion . for alternate approaches to far field splitting that however , so far , lack a rigorous stability analysis we refer to @xcite
( see also @xcite for a method to separate time - dependent wave fields due to multiple sources ) .
this paper is organized as follows . in the next section we provide the theoretical background for the direct and inverse source problem for the two - dimensional helmholtz equation with compactly supported sources . in section [ sec : regpicard ]
we discuss the singular value decomposition of the restricted far field operator mapping sources supported in a ball to their radiated far fields , and we formulate the regularized picard criterion to characterize non - evanescent far fields . in section [ sec : uncertainty ] we discuss uncertainty principles for the far field translation operator and for the fourier expansion of far fields , and in section [ sec : l2corollaries ] we utilize those to analyze the stability of least squares algorithms for far field splitting and data completion .
section [ sec : l1corollaries ] focuses on corresponding results for @xmath15 algorithms .
consequences of these stability estimates related to conditioning and resolution of reconstruction algorithms for inverse source and inverse scattering problems are considered in section [ sec : conditioning ] , and in section [ sec : analyticexample][sec : numericalexamples ] we provide some analytic and numerical examples .
suppose that @xmath16 represents a compactly supported acoustic or electromagnetic source in the plane .
then the time - harmonic wave @xmath17 radiated by @xmath8 at _ wave number _ @xmath18 solves the _ source problem _ for the helmholtz equation @xmath19 and satisfies the _ sommerfeld radiation condition _
@xmath20 we include the extra factor of @xmath21 on the right hand side so that both @xmath22 and @xmath23 scale ( under dilations ) as functions ; i.e. , if @xmath24 and @xmath25 , then @xmath26 with this scaling , distances are measured in wavelengths wavelengths . ] , and this allows us to set @xmath27 in our calculations , and then easily restore the dependence on wavelength when we are done . the _ fundamental solution _ of the helmholtz equation ( with @xmath27 ) in two dimensions is @xmath28 so the solution to can be written as a volume potential @xmath29 the asymptotics of the hankel function tell us that @xmath30 where @xmath31 with @xmath32 , and @xmath33 the function @xmath34 is called the _ far field _ radiated by the source @xmath8 , and equation shows that the _ far field operator _ @xmath35 , which maps @xmath8 to @xmath34 is a _ restricted fourier transform _ , i.e. @xmath36 the goal of the inverse source problem is to deduce properties of an unknown source @xmath16 from observations of the far field .
clearly , any compactly supported source with fourier transform that vanishes on the unit circle is in the nullspace @xmath37 of the far field operator .
we call @xmath38 a _ non - radiating source _ because a corollary of rellich s lemma and unique continuation is that , if the far field vanishes , then the wave @xmath39 vanishes on the unbounded connected component of the complement of the support of @xmath8 .
the nullspace of @xmath35 is exactly @xmath40 neither the source @xmath8 nor its support is uniquely determined by the far field , and , as non - radiating sources can have arbitrarily large supports , no upper bound on the support is possible .
there are , however , well defined notions of lower bounds .
we say that a compact set @xmath41 _ carries _ @xmath34 , if every open neighborhood of @xmath42 supports a source @xmath43 that radiates @xmath34 .
convex scattering support _
@xmath44 of @xmath34 , as defined in @xcite ( see also @xcite ) , is the intersection of all compact convex sets that carry @xmath34 .
the set @xmath44 itself carries @xmath34 , so that @xmath44 is the smallest convex set which carries the far field @xmath34 , and the convex hull of the support of the `` true '' source @xmath8 must contain @xmath44 . because two disjoint compact sets with connected complements can not carry the same far field pattern ( cf .
* lemma 6 ) ) , it follows that @xmath44 intersects any connected component of @xmath45 , as long as the corresponding source component is not non - radiating . in @xcite , an analogous notion , the _ uwscs support _
, was defined , showing that any far field with a compactly supported source is carried by a smallest union of well - separated convex sets ( well - separated means that the distance between any two connected convex components is strictly greater than the diameter of any component ) .
a corollary is that it makes theoretical sense to look for the support of a source with components that are small compared to the distance between them . here , as in previous investigations @xcite
, we study the well - posedness issues surrounding numerical algorithms to compute that support .
if we consider the restriction of the source to far field map @xmath35 from to sources supported in the ball @xmath46 of radius @xmath47 centered at the origin , i.e. , @xmath48 we can write out a full singular value decomposition .
we decompose @xmath49 as @xmath50 where @xmath51 , @xmath52 , span the closed subspace of _ free sources _ , which satisfy @xmath53 and @xmath54 belongs to the orthogonal complement of that subspace ; i.e. , @xmath54 is a non - radiating source . with its continuation
to @xmath55 by zero whenever appropriate . ]
the restricted far field operator @xmath56 maps @xmath57 where @xmath58 denoting the fourier coefficients of a far field @xmath59 by @xmath60 so that @xmath61 and @xmath62 by parseval s identity , an immediate consequence of is that @xmath63 which has @xmath64-norm @xmath65 is the source with smallest 2-norm that is supported in @xmath66 and radiates the far field @xmath34 .
we refer to @xmath67 as the _ minimal power source _ because , in electromagnetic applications , @xmath67 is proportional to current density , so that , in a system with a constant internal resistance , @xmath68 is proportional to the input power required to radiate a far field .
similarly , @xmath69 measures the radiated power of the far field .
the squared singular values @xmath70 of the restricted fourier transform @xmath71 have a number of interesting properties with immediate consequences for the inverse source problem ; full proofs of the results discussed in the following can be found in appendix [ app : sn2 ] .
the squared singular values satisfy @xmath72 and @xmath73 decays rapidly as a function of @xmath74 as soon as @xmath75 , @xmath76 moreover , the odd and even squared singular values , @xmath73 , are decreasing ( increasing ) as functions of @xmath77 ( @xmath78 ) , and asymptotically @xmath79 where @xmath80 denotes the smallest integer that is greater than or equal to @xmath81 .
this can also be seen in figure [ fig : sn2 ] , where we include plots of @xmath73 ( solid line ) together with plots of the asymptote @xmath82 ( dashed line ) for @xmath83 ( left ) and @xmath84 ( right ) .
( solid line ) and asymptote @xmath82 ( dashed line ) for @xmath83 ( left ) and @xmath84 ( right).,title="fig:",height=207 ] ( solid line ) and asymptote @xmath82 ( dashed line ) for @xmath83 ( left ) and @xmath84 ( right).,title="fig:",height=207 ] the asymptotic regime in is already reached for moderate values of @xmath47 . the forgoing yields a very explicit understanding of the restricted fourier transform @xmath71 .
for @xmath85 the singular values @xmath86 are uniformly large , while for @xmath87 the @xmath86 are close to zero , and it is seen from as well as from figure [ fig : sn2 ] that as @xmath47 gets large the width of the @xmath74-interval in which @xmath86 falls from uniformly large to zero decreases .
similar properties are known for the singular values of more classical restricted fourier transforms ( see @xcite ) .
a physical source has _ limited power _ , which we denote by @xmath88 , and a receiver has a _ power threshold _ , which we denote by @xmath89 .
if the radiated far field has power less than @xmath90 , the receiver can not detect it .
because @xmath91 and the odd and even squared singular values , @xmath92 , are decreasing as functions of @xmath77 , we may define : @xmath93
so , if @xmath59 is a far field radiated by a limited power source supported in @xmath46 with @xmath94 , then , for @xmath95 @xmath96 accordingly , @xmath97 is below the power threshold .
so the subspace of detectable far fields , that can be radiated by a power limited source supported in @xmath66 is : @xmath98 we refer to @xmath99 as the subspace of _ non - evanescent far fields _ , and to the orthogonal projection of a far field onto this subspace as the _ non - evanescent _ part of the far field .
we use the term _ non - evanescent _ because it is the phenomenon of evanescence that explains why the the singular values @xmath92 decrease rapidly for @xmath87 , resulting in the fact that , for a wide range of @xmath90 and @xmath100 , @xmath101 , if @xmath47 is sufficiently large . as function of @xmath47 for different values of @xmath102 .
dotted lines correspond to @xmath103 and @xmath104.,title="fig:",height=207 ] as function of @xmath47 for different values of @xmath102 .
dotted lines correspond to @xmath103 and @xmath104.,title="fig:",height=207 ] this is also illustrated in figure [ fig : n - rpp ] , where we include plots of @xmath105 from for @xmath106 , @xmath107 , and @xmath108 and for varying @xmath47 .
the dotted lines in these plots correspond to @xmath103 and @xmath104 , respectively .
in the inverse source problem , we seek to recover information about the size and location of the support of a source from observations of its far field . because the far field is a restricted fourier transform , the formula for the fourier transform of the translation of a function : @xmath109 plays an important role .
we use @xmath110 to denote the map from @xmath111 to itself given by @xmath112 the mapping @xmath113 acts on the fourier coefficients @xmath114 of @xmath34 as a convolution operator , i.e. , the fourier coefficients @xmath115 of @xmath116 satisfy @xmath117 where @xmath118 and @xmath119 are the polar coordinates of @xmath120 . employing a slight abuse of notation , we also use @xmath113 to denote the corresponding operator from @xmath121 to itself that maps @xmath122 note that @xmath113 is a unitary operator on 2 , i.e. @xmath123 . the following theorem , which we call an _ uncertainty principle for the translation operator _
, will be the main ingredient in our analysis of far field splitting .
[ th : uncertainty1 ] let @xmath124 such that the corresponding fourier coefficients @xmath125 and @xmath126 satisfy @xmath127 and @xmath128 with @xmath129 , and let @xmath130 .
then , @xmath131 we will frequently be discussing properties of a far field @xmath34 and those of its fourier coefficients . the following notation will be a useful shorthand : @xmath132 the notation emphasizes that we treat the representation of the function @xmath34 by its values , or by the sequence of its fourier coefficients as simply a way of inducing different norms .
that is , both and describe different norms of the same function on @xmath12 .
note that , because of the plancherel equality , @xmath133 , so we may just write @xmath134 , and we write @xmath135 for the corresponding inner product . we will extend the notation a little more and refer to the support of @xmath34 in @xmath12 as its @xmath136-support and denote by @xmath137 the measure of @xmath138 .
we will call the indices of the nonzero fourier coefficients in its fourier series expansion the @xmath139-support of @xmath34 , and use @xmath140 to denote the number of non - zero coefficients . @xmath141 with this notation , theorem [ th : uncertainty1 ]
becomes [ th : uncertainty2 ] let @xmath142 , and let @xmath130 . then
, @xmath143 we refer to theorem [ th : uncertainty2 ] as an uncertainty principle , because , if we could take @xmath144 in , it would yield @xmath145 as stated , is is true but not useful , because @xmath146 and @xmath147 can not simultaneously be finite .
could have been radiated by a source supported in an arbitrarily small ball centered at the origin , or centered at @xmath120 , but rellich s lemma and unique continuation show that no nonzero far field can have two sources with disjoint supports .
] we present the corollary only to illustrate the close analogy to the theorem 1 in @xcite , which treats the discrete fourier transform ( dft ) on sequences of length @xmath148 : if @xmath149 represents the sequence @xmath150 for @xmath151 and @xmath152 its dft , then @xmath153 this is a lower bound on the _ time - bandwidth product_. in @xcite donoho and stark present two important corollaries of uncertainty principles for the fourier transform .
one is the uniqueness of sparse representations of a signal @xmath149 as a superposition of vectors taken from both the standard basis and the basis of fourier modes , and the second is the recovery of this representation by 1 minimization .
the main observation we make here is that , if we phrase our uncertainty principle as in theorem [ th : uncertainty2 ] , then the far field translation operator , as well as the map from @xmath34 to its fourier coefficients , satisfy an uncertainty principle . combining the uncertainty principle with the regularized picard criterion from section [ sec : regpicard ] yields analogs of both results in the context of the inverse source problem .
these include previous results about the splitting of far fields from @xcite and @xcite , which can be simplified and extended by viewing them as consequences of the uncertainty principle and the regularized picard criterion .
the proof of theorem [ th : uncertainty2 ] is a simple corollary of the lemma below : [ lmm : tcestimate ] let @xmath130 and let @xmath110 be the operator introduced in and .
then , the operator norm of @xmath154 , @xmath155 , satisfies @xmath156 whereas @xmath157 fulfills @xmath158 recalling , we see that @xmath113 is multiplication by a function of modulus one , so is immediate . on the other hand , combining with the last inequality from page 199 of @xcite ; more precisely , @xmath159 shows that @xmath160 using hlder s inequality and we obtain that @xmath161 we can improve the dependence on @xmath118 in under hypotheses on @xmath34 and @xmath162 that are more restrictive , but well suited to the inverse source problem .
[ th : uncertainty3 ] suppose that @xmath163 , @xmath164 with @xmath165 , and let @xmath130 such that @xmath166
. then @xmath167 because the 0-support of @xmath162 is contained in @xmath168 $ ] @xmath169 so @xmath170 and it follows from theorem 2 of @xcite , using the fact that @xmath171 , together with our hypothesis , which implies that @xmath172 , that @xmath173 ( see appendix [ app : proofkrasikov ] for details ) .
we now simply repeat the proof of theorem [ th : uncertainty2 ] , replacing the estimate for @xmath174 from with the estimate we have just established in , i.e. @xmath175,{\ensuremath{{l^{\infty}}}}[-m , m ] } \,\le\ , \frac{1}{|c|^{{\ensuremath{\frac{1}{2 } } } } } \,.\ ] ] we will also make use of another uncertainty principle .
a glance at reveals that the operator which maps @xmath34 to its fourier coefficients maps 2 to 2 with norm 1 , 1 to with norm @xmath176 , and its inverse maps 1 to , also with norm @xmath176 .
an immediate corollary of this observation is [ th : fourieruncertainty ] let @xmath177 and let @xmath130 .
then , @xmath178 combining hlder s inequality with and using the mapping properties of the operator which maps @xmath34 to its fourier coefficients we find that @xmath179
the regularized picard criterion tells us that , up to an 2-small error , a far field radiated by a limited power source in @xmath46 is 2-close to an @xmath34 that belongs to the subspace of non - evanescent far fields , the span of @xmath180 with @xmath181 , where @xmath95 is a little bigger than the radius @xmath47 .
this non - evanescent @xmath34 satisfies @xmath182 .
the uncertainty principle will show that the angle between translates of these subspaces is bounded below when the translation parameter is large enough , so that we can split the sum of the two non - evanescent far fields into the original two summands .
[ lm : l2split ] suppose that @xmath183 and @xmath184 with @xmath185 and that @xmath186 .
then , for @xmath187 @xmath188 we first note that and imply @xmath189 we now use , @xmath190 dropping the second term now gives for @xmath191 , and we may interchange the roles @xmath191 and @xmath192 in the proof to obtain the estimate for @xmath192 . the analogous consequence of theorem [ th : uncertainty3 ] is [ lm : l2splitws ] suppose that @xmath193 , @xmath194 for some @xmath195 , @xmath187 , and @xmath184 with @xmath196 and @xmath197 and that @xmath198
. then , for @xmath187 @xmath199 in our application to the inverse source problem , we will know that each far field is the translation of a far field @xmath200 , radiated by a limited power source supported in a ball centered at the origin , and therefore that all but a very small amount of the radiated power is contained in the non - evanescent part , the translation of the fourier modes @xmath201 for @xmath202 .
the estimate in the theorem below says that , if the distances between the balls is large enough , we may uniquely solve for the non - evanescent parts of the individual far fields , and that this split is stable with respect to perturbations in the data .
[ th : l2splitls ] suppose that @xmath203 , @xmath184 and @xmath204 such that @xmath196 and @xmath205 and let [ eq : split2ls ] @xmath206 then , for @xmath187 @xmath207 the notation in above means that the @xmath208 are the ( necessarily unique ) least squares solutions to the equations @xmath209 . recall that the far fields radiated by a limited power source from a ball have almost all , but not all , of their power ( 2-norm ) concentrated in the fourier modes with @xmath210 .
therefore the @xmath211 will typically not belong to the subspace that is the direct sum of @xmath212 , and therefore @xmath213 and @xmath214 will usually not solve equations exactly .
the estimate in is nevertheless always true , and guarantees that the pair @xmath215 is unique and that the absolute condition number of the splitting operator which maps @xmath216 to @xmath215 is no larger than @xmath217 .
each @xmath218 can be uniquely decomposed as @xmath219 where each @xmath220 belongs to the @xmath221-dimensional subspace @xmath222 and each @xmath223 is orthogonal to @xmath224 .
the definition of least squares solutions means that @xmath225 subtracting gives @xmath226 and applying the estimate yields @xmath227 finally , we note that @xmath228 which finishes the proof .
we also have corresponding corollaries of theorem [ th : fourieruncertainty ] , which tell us that , if a far field is radiated from a small ball , and measured on most of the circle , then it is possible to recover its non - evanescent part on the entire circle .
theorem [ th : l2completels ] below , describes the case where we can not measure the far field @xmath229 on a subset @xmath230 .
we measure @xmath231 , where @xmath232 .
the estimates imply that we can stably recover the non - evanescent part of the far field on @xmath42 .
before we state the theorem , we give the corresponding analogue of lemma [ lm : l2split ] and lemma [ lm : l2splitws ] .
[ lm : l2complete ] suppose that @xmath233 and @xmath130 with @xmath234 and that @xmath235 .
then [ eq : l2complete ] @xmath236 proceeding as in , but replacing by yields the result .
[ th : l2completels ] suppose that @xmath203 , @xmath130 , @xmath237 and @xmath238 such that @xmath239 , and let @xmath240 then [ eq : l2completels ] @xmath241 just as in , we decompose each @xmath218 @xmath242 where each @xmath220 belongs to the subspace @xmath243 and each @xmath223 is orthogonal to @xmath224 . proceeding as in , but using the estimates from , we find @xmath244 and then note that is true here as well to finish the proof .
a version of theorem [ th : l2splitls ] with multiple well - separated components is also true .
[ th : l2splitmultiplels ] suppose that @xmath203 , @xmath245 and @xmath195 , @xmath246 , such that @xmath247 for every @xmath248 and @xmath249 and let @xmath250 then , for @xmath246 @xmath251 as in , we decompose each @xmath218 @xmath242 where each @xmath220 belongs to the subspace @xmath252 and each @xmath223 is orthogonal to @xmath224 .
subtracting gives @xmath253 and applying shows that @xmath254 where @xmath255 , @xmath256 and @xmath257 .
since is true here as well this ends the proof .
we may include a missing data component as well .
[ th : l2completeandsplitmultiplels ] suppose that @xmath203 , @xmath245 , @xmath195 , @xmath246 , and @xmath258 such that @xmath247 for every @xmath248 and @xmath259 and let [ eq : l2completeandsplitmultiplels - lsp ] @xmath260 then @xmath261 as in , we decompose each @xmath218 @xmath242 where each @xmath220 belongs to the subspace @xmath262 and each @xmath223 is orthogonal to @xmath224 .
subtracting gives @xmath263 and thus @xmath264 proceeding as in , using and , and applying finishes the proof .
the results below are analogous to those in the previous section .
the main difference is that they do not require the _ a priori _ knowledge of the size of the non - evanescent subspaces ( the @xmath265 in theorems [ th : l2splitls ] through [ th : l2completeandsplitmultiplels ] ) . in theorem
[ th : l1split ] below , @xmath266 represents the ( measured ) approximate far field ; the @xmath267 are the non - evanescent parts of the true ( unknown ) far fields radiated by each of the two components , which we assume are well - separated . the constant @xmath268 in accounts for both the noise and the evanescent components of the true far fields .
condition requires that the optimization problem be formulated with a constraint that is weak enough so that the @xmath267 are feasible .
[ th : l1split ] suppose that @xmath269 and @xmath184 such that @xmath270 and @xmath271 if @xmath272 and @xmath273 with @xmath274 and @xmath275 then , for @xmath187 @xmath276 a consequence of is that the pair @xmath277 satisfies the constraint in , which implies that @xmath278 because @xmath279 is a minimizer .
additionally , with @xmath280 representing the @xmath139-support of @xmath281 and @xmath282 its complement , @xmath283 inserting into yields @xmath284 we now use together with , the constraint in and the fact that @xmath285 is an 2-isometry to obtain @xmath286 hlder s inequality , , and show @xmath287 using hlder s inequality once more yields @xmath288 which implies because @xmath289 . assuming that some a priori information on the size of the non - evanescent subspaces is available and that the distances between the source components is large relative to their dimensions , we can improve the dependence of the stability estimates on the distances .
[ th : l1splitws ] if we add to the hypothesis of theorem [ th : l1split ] : @xmath290 for some @xmath204 and replace with @xmath291 then , for @xmath187 @xmath292 replace by in .
the constraint in corollary [ th : l1splitws ] is much more restrictive than the corresponding assumption in theorem [ th : l2splitls ] , and also the estimate is weaker than .
however , if we add to the hypothesis of theorem [ th : l1split ] all _ a priori _ assumptions on the non - evanescent subspaces used in theorem [ th : l2splitls ] , the result improves as follows .
[ th : l1splitws - apriori ] if we add to the hypothesis of theorem [ th : l1split ] : @xmath293 for some @xmath204 and replace with @xmath294 the conclusion becomes @xmath295 the constraint in corollary [ th : l1splitws - apriori ] is now the same as the corresponding assumption in theorem [ th : l2splitls ] , but the estimate still differs from ( after taking the square root on both sides of these inequalities ) by a factor of two . however , the main advantage of the 1 approach is clearly that no a priori knowledge of the size of the non - evanescent subspaces is required .
if such a priori information is available , we recommend using least squares .
proceeding as in and applying , we find that @xmath296 where @xmath297 , @xmath298 and @xmath299 .
dropping the second term gives for @xmath191 , and we may interchange the roles of @xmath191 and @xmath192 in the proof to obtain the estimate for @xmath192 .
the analogue of theorem [ th : l2completels ] for data completion but without _ a priori _ knowledge on the size of the non - evanescent subspaces is [ th : l1complete ] suppose that @xmath300 , @xmath238 , @xmath301 and @xmath130 such that @xmath302 and @xmath303 if @xmath272 and @xmath273 with @xmath304 and @xmath305 then [ eq : l1complete ] @xmath306 proceeding as in we find that @xmath307 with @xmath224 representing the @xmath139-support of @xmath308 . applying similar arguments as in yields @xmath309 we now use hlder s inequality , , the mapping properties of the operator which maps @xmath34 to its fourier coefficients and to obtain @xmath310 dropping the second term gives for @xmath34 because @xmath311 , and we may interchange the roles of @xmath34 and @xmath162 when completing the square in the last line of to obtain the estimate for @xmath162 . if @xmath42 is unknown as well , then theorem [ th : l1complete ] can be adapted as follows .
[ th : l1completeno ] suppose that @xmath300 , @xmath238 , @xmath301 and @xmath130 such that @xmath312 and @xmath313 if @xmath272 and @xmath273 with @xmath304 and @xmath314 then @xmath315 proceeding as in we find that @xmath316 with @xmath224 representing the @xmath139-support of @xmath308 . applying similar arguments as in and together with yields @xmath317
this ends the proof because @xmath311 .
a possible application of corollary [ th : l1completeno ] is the problem of removing ( high - amplitude ) strongly localized noise from measured far field data .
next we consider sources supported on sets with multiple disjoint components .
[ th : l1splitmultiple ] suppose that @xmath318 and @xmath245 , @xmath246 such that @xmath319 and @xmath320 if @xmath272 and @xmath273 with @xmath304 and @xmath321 then , for @xmath246 @xmath322 proceeding as in we find that @xmath323 with @xmath324 representing the 0-support of @xmath325 .
applying similar arguments as in and using the inequality from appendix [ app : sumestimates ] and we obtain @xmath326 applying hlder s inequality and yields @xmath327 where @xmath328 .
as in corollary [ th : l1splitws ] we can improve these estimates , under the assumption that some a priori knowledge of the size of the non - evanescent subspaces is available and that the individual source components are sufficiently far apart from each other . [ th : l1splitmultiplews ] if we add to the hypothesis of theorem [ th : l1splitmultiple ] : @xmath329 for some @xmath330 , and replace with @xmath331 the conclusion becomes , for @xmath246 @xmath332 replace by in .
next we consider multiple source components together with a missing data component .
[ th : l1completeandsplitmultiple ] suppose that @xmath333 , @xmath245 , @xmath246 , @xmath238 and @xmath301 such that [ eq : l1completeandsplitpositivitymultiple ] @xmath334 and @xmath335 if @xmath272 and @xmath273 with @xmath304 and @xmath336 then [ eq : l1completeandsplitmultiple ] @xmath337 proceeding as in we find that @xmath338 with @xmath324 representing the 0-support of @xmath325 . applying similar arguments as in we obtain @xmath339 the third term on the right hand side of can be estimated as in , while for the last term we find using hlder s inequality , the mapping properties of the operator which maps @xmath34 to its fourier coefficients and that @xmath340 where @xmath328 . combining these estimates ends the proof .
again , including a priori information of the size of the non - evanescent subspaces and assuming that the individual source components are well separated , the result can be improved : [ th : l1completeandsplitmultiplews ] if we add to the hypothesis of theorem [ th : l1completeandsplitmultiple ] : @xmath341 for some @xmath330 , and replace with @xmath342 the conclusion becomes , for @xmath246 @xmath343 finally , we establish variants of theorem [ th : l1splitmultiple ] and theorem [ th : l1completeandsplitmultiple ] , where we replace the 1 minimization in and by a weighted 1 minimization in order to obtain better estimates for certain geometric configurations of the supports of the individual source components .
in contrast to theorem [ th : l1splitmultiple ] the constant in the stability estimate in theorem [ th : l1splitmultiplen ] below only depends on the distances of source components relative to the source component corresponding to the far field component appearing on the left hand side of the estimate . [
th : l1splitmultiplen ] suppose that @xmath318 and @xmath245 , and set @xmath344 @xmath246 .
assume that @xmath345 and @xmath346 if @xmath272 and @xmath273 with @xmath304 and @xmath347 then , for @xmath246 @xmath348 proceeding as in we find @xmath349 with @xmath324 representing the 0-support of @xmath325 .
abbreviating @xmath257 , the triangle inequality shows @xmath350 i.e. , @xmath351 thus our assumption implies @xmath352 and therefore , using @xmath353 now , applying similar arguments as in , using and , we obtain @xmath354 using hlder s inequality and we deduce @xmath355 where @xmath328 .
this ends the proof .
[ th : l1splitmultiplenws ] if we add to the hypothesis of theorem [ th : l1splitmultiplen ] : @xmath356 for some @xmath330 , and replace with @xmath357 the conclusion remains true .
[ th : l1completeandsplitmultiplen ] suppose that @xmath318 , @xmath245 , @xmath238 and @xmath301 , and set @xmath358 @xmath246 .
assume that @xmath359 and @xmath335 if @xmath272 and @xmath273 with @xmath304 and @xmath360 then @xmath361 proceeding as in we find that @xmath362 with @xmath324 representing the 0-support of @xmath325 .
applying similar arguments as in we obtain @xmath363 the third term on the right hand side of can be estimated as in , while for the last term we find using hlder s inequality , the mapping properties of the operator which maps @xmath34 to its fourier coefficients , and that @xmath364 applying hlder s inequality once more yields @xmath365 where @xmath328 . combining these estimates
ends the proof .
[ th : l1completeandsplitmultiplenws ] if we add to the hypothesis of theorem [ th : l1completeandsplitmultiplen ] : @xmath356 for some @xmath330 , and replace with @xmath357
the conclusion remains true .
so far , we have suppressed the dependence on the wavenumber @xmath366 .
we restore it here , and consider the consequences related to conditioning and resolution .
we confine our discussion to theorem [ th : l2splitls ] , assuming that the @xmath367 , @xmath368 , represent far fields that are radiated by superpositions of limited power sources supported in balls @xmath369 , @xmath187 , and that accordingly , for @xmath27 ( following our discussion at the end of section [ sec : regpicard ] ) , the numbers @xmath370 are just a little bigger than the radii of these balls .
this becomes @xmath371 when we return to conventional units , and the estimate then depends on the quantity @xmath372 writing @xmath373 and denoting by @xmath374 the orthogonal projection onto @xmath375 , @xmath187 , we have @xmath376 if @xmath377 , and the angle @xmath378 between these subspaces is given by @xmath379 a glance at the proof of lemma [ lm : l2split ] reveals that the square root of is just a lower bound for this cosine .
furthermore , the least squares solutions to can be constructed from simple formulas @xmath380 where @xmath381 and @xmath382 denote the projection onto @xmath383 along @xmath384 and vice versa .
these satisfy @xmath385 consequently @xmath386 is the absolute condition number for the splitting problem , and theorem [ th : l2splitls ] ( with our choice of @xmath387 and @xmath388 ) essentially says that @xmath389 we will include an example below to show that , at least for large distances , the dependence on @xmath366 in estimate in is sharp .
this means that , for a fixed geometry @xmath390 , the condition number increases with @xmath366 .
because resolution is proportional to wavelength , this means that we can not increase resolution by simply increasing the wavenumber without increasing the dynamic range of the sensors ( i.e. the number of significant figures in the measured data ) .
note that as @xmath366 increases , the dimensions of the subspaces @xmath391 increase .
the increase in the number of significant fourier coefficients ( non - evanescent fourier modes ) is the way we see higher resolution in this problem .
the situation changes considerably if we replace the limited power source radiated from @xmath392 by a point source with singularity in @xmath393
. then we can choose for @xmath383 a one - dimensional subspace of @xmath121 ( spanned by the zeroth order fourier mode translated by @xmath394 ) , and accordingly set @xmath395 .
consequently , the estimate reduces to @xmath396 since numerator and denominator have the same units , the conditioning of the splitting operator does not depend on @xmath366 in this case .
this has immediate consequences for the inverse scattering problem : qualitative reconstruction methods like the linear sampling method @xcite or the factorization method @xcite determine the support of an unknown scatterer by testing pointwise whether the far field of a point source belongs to the range of a certain restricted far field operator , mapping sources supported inside the scatterer to their radiated far field .
the inequality indeed shows that ( using these qualitative reconstruction algorithms for the inverse scattering problem ) one can increase resolution by simply increasing the wave number .
finally , if we replace both sources by point sources with singularities in @xmath393 and @xmath397 , respectively , then we can choose both subspaces @xmath383 and @xmath384 to be one - dimensional , and accordingly set @xmath398 .
the estimate reduces to @xmath399 i.e. , in this case the conditioning of the splitting operator improves with increasing wave number @xmath366 .
music - type reconstruction methods @xcite for inverse scattering problems with infinitesimally small scatterers recover the locations of a collection of unknown small scatterers by testing pointwise whether the far field of a point source belongs to the range of a certain restricted far field operator , mapping point sources with singularities at the positions of the small scatterers to their radiated far field . from we
conclude that ( using music - type reconstruction algorithms for the inverse scattering problem with infinitesimally small scatterers ) on can increase resolution by simply increasing the wave number and the reconstruction becomes more stable for higher frequencies .
the example below illustrates that the estimate of the cosine of the angle between two far fields radiated by two sources supported in balls @xmath392 and @xmath400 , respectively , can not be better than proportional to the quantity @xmath401 as pointed out in the previous section , we need only construct the example for @xmath27
. we will let @xmath8 be a single layer source supported on a horizontal line segment of width @xmath224 , and @xmath23 be the same source , translated vertically by a distance @xmath402 ( i.e. , @xmath403 and @xmath404 ) .
specifically , with @xmath405 denoting the heavyside or indicator function , and @xmath406 the dirac mass : @xmath407 the far fields radiated by @xmath8 and @xmath23 are : @xmath408 for @xmath409 . accordingly
@xmath410 and we can evaluate the first integral on the right hand side using the plancherel equality as @xmath411 is the fourier transform of the characteristic function of the interval @xmath412 $ ] , and estimate the second , yielding @xmath413 on the other hand , for @xmath414 , according to the principle of stationary phase ( there are stationary points at @xmath415 ) @xmath416 which shows that for @xmath417 @xmath418 which decays no faster than that predicted by theorem [ th : l2splitls ] .
next we consider the numerical implementation of the 2 approach from section [ sec : l2corollaries ] and the 1 approach from section [ sec : l1corollaries ] for far field splitting and data completion simultaneously ( cf .
theorem [ th : l2completeandsplitmultiplels ] and theorem [ th : l1completeandsplitmultiple ] ) .
since both schemes are extensions of corresponding algorithms for far field splitting as described in @xcite ( least squares ) and @xcite ( basis pursuit ) , we just briefly comment on modifications that have to be made to include data completion and refer to @xcite for further details .
given a far field @xmath419 that is a superposition of far fields @xmath420 radiated from balls @xmath369 , for some @xmath245 and @xmath421 , we assume in the following that we are unable to observe all of @xmath34 and that a subset @xmath238 is unobserved .
the aim is to recover @xmath422 from @xmath423 and a priori information on the location of the supports of the individual source components @xmath369 , @xmath246 .
we first consider the 2 approach from section [ sec : l2corollaries ] and write @xmath424 for the observed far field data and @xmath425 .
accordingly , @xmath426 i.e. , we are in the setting of theorem [ th : l2completeandsplitmultiplels ] . using the shorthand @xmath427 and @xmath428 , @xmath246 ,
the least squares problem is equivalent to seeking approximations @xmath429 and @xmath430 , @xmath246 , satisfying the galerkin condition @xmath431 the size of the individual subspaces depends on the a priori information on @xmath432 .
following our discussion at the end of section [ sec : regpicard ] we choose @xmath433 in our numerical example below . denoting by @xmath434 and @xmath435 the orthogonal projections onto @xmath436 and @xmath437 , respectively ,
is equivalent to the linear system @xmath438 p_ip_\omega{{\widetilde\beta}}+ p_ip_1t_{c_1}^*{{\widetilde\alpha}}_1 + \cdots + t_{c_i}^*{{\widetilde\alpha}}_i & \,=\ , p_i\gamma \ , .
\end{split}\ ] ] explicit matrix representations of the individual matrix blocks in follow directly from
( see ( * ? ? ?
* lemma 3.3 ) for details ) for @xmath435 and by applying a discrete fourier transform to the characteristic function on @xmath439 for @xmath434 .
accordingly , the block matrix corresponding to the entire linear system can be assembled , and the linear system can be solved directly .
the estimates from theorem [ th : l2completeandsplitmultiplels ] give bounds on the absolute condition number of the system matrix .
the main advantage of the 1 approach from section [ sec : l1corollaries ] is that no a priori information on the radii @xmath440 of the balls @xmath369 , @xmath246 , containing the individual source components is required .
however , we still assume that a priori knowledge of the centers @xmath441 of such balls is available . using the orthogonal projection @xmath434 onto @xmath442 , the basis pursuit formulation from theorem [ th : l1completeandsplitmultiple ]
can be rewritten as @xmath443 accordingly , @xmath444 is an approximation of the missing data segment .
it is well known that the minimization problem from is equivalent to minimizing the tikhonov functional @xmath445 \in \ell^2\times\cdots\times\ell^2 \,,\ ] ] for a suitably chosen regularization parameter @xmath446 ( see , e.g. , ( * ? ? ?
* proposition 2.2 ) ) .
the unique minimizer of this functional can be approximated using ( fast ) iterative soft thresholding ( cf .
@xcite ) . apart from the projection @xmath434 , which can be implemented straightforwardly , our numerical implementation analogously to the implementation for the splitting problem described in @xcite , and also the convergence analysis from @xcite carries over .
imply that the solution of and is very close to the true split . ]
we consider a scattering problem with three obstacles as shown in figure [ fig : numexa1 ] ( left ) , which are illuminated by a plane wave @xmath447 , @xmath448 , with incident direction @xmath449 and wave number @xmath27 ( i.e. , the wave length is @xmath450 ) .
.,title="fig:",height=188 ] .,title="fig:",height=188 ] assuming that the ellipse is sound soft whereas the kite and the nut are sound hard , the scattered field @xmath451 satisfies the homogeneous helmholtz equation outside the obstacles , the sommerfeld radiation condition at infinity , and dirichlet ( ellipse ) or neumann boundary conditions ( kite and nut ) on the boundaries of the obstacles .
we simulate the corresponding far field @xmath34 of @xmath451 on an equidistant grid with @xmath452 points on the unit sphere @xmath453 using a nystrm method ( cf .
@xcite ) . figure [ fig : numexa1 ] ( middle ) shows the real part ( solid line ) and the imaginary part ( dashed line ) of @xmath34 . since the far field @xmath34 can be written as a superposition of three far fields radiated by three individual smooth sources supported in arbitrarily small neighborhoods of the scattering obstacles ( cf . ,
e.g. , ( * ? ? ?
* lemma 3.6 ) ) , this example fits into the framework of the previous sections .
we assume that the far field can not be measured on the segment @xmath454 i.e. , @xmath455 .
we first apply the least squares procedure and use the dashed circles shown in figure [ fig : numexa1 ] ( left ) as a priori information on the approximate source locations @xmath369 , @xmath456 . more precisely , @xmath457 , @xmath458 , @xmath459 and @xmath460 , @xmath461 and @xmath462 .
accordingly we choose @xmath463 , @xmath464 and @xmath465 , and solve the linear system .
( left ) , reconstruction of the missing part @xmath422 ( middle ) , and difference between exact far field and reconstructed far field ( right).,title="fig:",height=188 ] ( left ) , reconstruction of the missing part @xmath422 ( middle ) , and difference between exact far field and reconstructed far field ( right).,title="fig:",height=188 ] ( left ) , reconstruction of the missing part @xmath422 ( middle ) , and difference between exact far field and reconstructed far field ( right).,title="fig:",height=188 ] figure [ fig : numexa2 ] shows a plot of the observed data @xmath216 ( left ) , of the reconstruction of the missing data segment obtained by the least squares algorithm and of the difference between the exact far field and the reconstructed far field . again
the solid line corresponds to the real part while the dashed line corresponds to the imaginary part .
the condition number of the matrix is @xmath466 .
we note that the missing data component in this example is actually too large for the assumptions of theorem [ th : l2completeandsplitmultiplels ] to be satisfied .
nevertheless the least squares approach still gives good results .
applying the ( fast ) iterative soft shrinkage algorithm to this example ( with regularization parameter @xmath467 in ) does not give a useful reconstruction .
as indicated by the estimates in theorem [ th : l1completeandsplitmultiple ] the 1 approach seems to be a bit less stable .
hence we halve the missing data segment , consider in the following @xmath468 i.e. , @xmath469 , and apply the 1 reconstruction scheme to this data .
( left ) , reconstruction of the missing part @xmath422 ( middle ) , and difference between exact far field and reconstructed far field ( right).,title="fig:",height=188 ] ( left ) , reconstruction of the missing part @xmath422 ( middle ) , and difference between exact far field and reconstructed far field ( right).,title="fig:",height=188 ] ( left ) , reconstruction of the missing part @xmath422 ( middle ) , and difference between exact far field and reconstructed far field ( right).,title="fig:",height=188 ] figure [ fig : numexa3 ] shows a plot of the observed data @xmath216 ( left ) , of the reconstruction of the missing data segment obtained by the fast iterative soft shrinkage algorithm ( with @xmath467 ) after @xmath470 iterations ( the initial guess is zero ) and of the difference between the exact far field and the reconstructed far field .
the behavior of both algorithms in the presence of noise in the data depends crucially on the geometrical setup of the problem ( i.e. on its conditioning ) .
the smaller the missing data segment is and the smaller the dimensions of the individual source components are relative to their distances , the more noise these algorithms can handle
we have considered the source problem for the two - dimensional helmholtz equation when the source is a superposition of finitely many well - separated compactly supported source components .
we have presented stability estimates for numerical algorithms to split the far field radiated by this source into the far fields corresponding to the individual source components and to restore missing data segments .
analytic and numerical examples confirm the sharpness of these estimates and illustrate the potential and limitations of the numerical schemes .
the most significant observations are : * the conditioning of far field splitting and data completion depends on the dimensions of the source components , their relative distances with respect to wavelength and the size of the missing data segment .
the results clearly suggest combining data completion with splitting whenever possible in order to improve the conditioning of the data completion problem . *
the conditioning of far field splitting and data completion depends on wave length and deteriorates with increasing wave number .
therefore , in order to increase resolution one not only has to increase the wave number but also the dynamic range of the sensors used to measure the far field data .
in the following we collect some interesting properties of the squared singular values @xmath70 , as introduced in , of the restricted fourier transform @xmath71 from .
we first note that @xcite implies that the squared singular values from satisfy @xmath471 and simple manipulations using recurrence relations for bessel functions [ eq : recurrencebessel ] @xmath472 ( cf . ,
e.g. , @xcite ) ) show that , for @xmath52 @xmath473 [ lmm : sumsn2 ] @xmath474 since @xmath475 ( cf .
@xcite ) , the definition yields @xmath476 the next lemma shows that odd and even squared singular values , @xmath73 , are monotonically decreasing for @xmath77 and monotonically increasing for @xmath78 .
[ lmm : monotonicitysn2 ] @xmath477 using the recurrence relations we find that @xmath478 thus , @xmath479 integrating sharp estimates for @xmath480 from @xcite , we obtain upper bounds for @xmath73 when @xmath481 . since @xmath482 , @xmath52 , it is sufficient to consider @xmath483 .
[ lmm : decaysn2 ] suppose that @xmath484 .
then @xmath485 from theorem 2 of @xcite we obtain for @xmath486 that @xmath487 substituting this into yields @xmath488 since @xmath489 is monotonically increasing for @xmath490 , we see @xmath491 on the other hand , the squared singular values @xmath92 are not small for @xmath492 .
[ th : besselasy ] suppose that @xmath493 , define @xmath494 by @xmath495 , and therefore @xmath496 , and assume @xmath497 .
then @xmath498 where the constant @xmath499 depends on the lower bound @xmath500 but is otherwise independent of @xmath74 and @xmath47 . by definition ,
@xmath501 with @xmath502 the phase function @xmath503 has stationary points at @xmath504 , and @xmath505 vanishes at @xmath506 and @xmath507 .
we will apply stationary phase in a neighborhood of each stationary point .
the neighborhood must be small enough to guarantee that @xmath508 is bounded from below there .
integration by parts will be used to estimate integral in regions where @xmath509 is bounded below .
the hypothesis that @xmath497 will guarantee that the union of these two regions covers the whole interval @xmath510 . to separate the two regions ,
let @xmath511 , @xmath512 , be a @xmath513 cutoff function satisfying @xmath514 with the @xmath515 independent of @xmath516 .
define @xmath517 , @xmath518 , then @xmath519 theorem 7.7.1 of @xcite tells us that for any integer @xmath520
@xmath521 with @xmath522 only depending on an upper bound for the higher order derivatives of @xmath503 . for the second inequality we have used and
the fact that all higher derivatives of @xmath503 are bounded by 1 .
we will estimate the remainder of the integral using theorem 7.7.5 of @xcite , which tells us that , if @xmath523 is the unique stationary point of @xmath503 in the support of a smooth function @xmath524 , and @xmath525 on the support of @xmath524 , then @xmath526 with @xmath527 depending only on the lower bound @xmath406 for @xmath508 and an upper bound for higher derivatives of @xmath503 on the support of @xmath524 .
we will set @xmath528 , which is supported in two intervals , one containing @xmath34 and the other containing @xmath529 , so , becomes @xmath530 as long as @xmath500 is chosen so that @xmath531 on the support of @xmath532 .
the following lemma suggests a proper choice of @xmath500 .
[ lmm : statphase ] let @xmath533 and @xmath34 belong to @xmath534 $ ] then @xmath535 implies that @xmath536 since @xmath537 we deduce @xmath538 consequently @xmath539 we choose @xmath540 and assume that @xmath541 , then lemma [ lmm : statphase ] gives @xmath542 .
we use this value of @xmath500 in , i.e. @xmath543 accordingly , @xmath544 now , adding and establishes . the calculation for is analogous with replaced by @xmath545 which has the same phase and hence the same stationary points .
the only difference is that the term @xmath546 in at @xmath547 will be @xmath548 rather than 1 .
we now combine ( [ eq : estbessel1 ] ) and ( [ eq : estbessel2 ] ) with the equality to obtain , for @xmath549 @xmath550 since equation is only a valid definition of the bessel function @xmath551 when @xmath74 is an integer is not an integer . ]
, we denote in the following by @xmath80 is smallest integer that is greater than or equal to @xmath81 , so that we can state a convergence result .
let @xmath52 and @xmath553 . theorem 2 of @xcite establishes that for @xmath554 , @xmath555 the following lemma shows that , under the assumptions of theorem [ th : uncertainty3 ] , the estimate implies .
let @xmath556 and @xmath557 , then @xmath558 since @xmath559 we may assume w.l.o.g .
that @xmath77 .
let @xmath560 .
then @xmath561 , i.e. @xmath562 and therefore our assumption @xmath557 implies for @xmath563 that @xmath564 accordingly , @xmath565 this shows that the assumptions of theorem 2 of @xcite are satisfied .
next we consider and further estimate its right hand side : @xmath566 since @xmath567 , applying and yields @xmath568 whence @xmath569
here we prove some elementary inequalities that we havent been able to find in the literature . | starting with far field data of time - harmonic acoustic or electromagnetic waves radiated by a collection of compactly supported sources in two - dimensional free space , we develop criteria and algorithms for the recovery of the far field components radiated by each of the individual sources , and the simultaneous restoration of missing data segments . although both parts of this inverse problem are severely ill - conditioned in general , we give precise conditions relating the wavelength , the diameters of the supports of the individual source components and the distances between them , and the size of the missing data segments , which guarantee that stable recovery in presence of noise is possible .
the only additional requirement is that a priori information on the approximate location of the individual sources is available .
we give analytic and numerical examples to confirm the sharpness of our results and to illustrate the performance of corresponding reconstruction algorithms , and we discuss consequences for stability and resolution in inverse source and inverse scattering problems . mathematics subject classifications ( msc2010 ) : 35r30 , ( 65n21 ) + keywords : inverse source problem , helmholtz equation , uncertainty principles , far field splitting , data completion , stable recovery + short title : uncertainty principles for inverse source problems |
Subsets and Splits