Journal of Physics A: Mathematical and Theoretical Related content PAPER Time evolution of the atomic inversion for the generalized Tavis–Cummings model—QIM approach - The su(1,1) Tavis-Cummings model Andrei Rybin, Georg Kastelewicz, Jussi Timonen et al. To cite this article: N M Bogoliubov et al 2017 J. Phys. A: Math. Theor. 50 464003 - The Quench Action Jean-Sébastien Caux - Exact solution of generalized Tavis Cummings models in quantum optics N M Bogoliubov, R K Bullough and J Timonen View the article online for updates and enhancements. This content was downloaded from IP address 129.8.242.67 on 28/10/2017 at 00:16 Journal of Physics A: Mathematical and Theoretical J. Phys. A: Math. Theor. 50 (2017) 464003 (20pp) https://doi.org/10.1088/1751-8121/aa8c6a Time evolution of the atomic inversion for the generalized Tavis–Cummings model—QIM approach N M Bogoliubov1,2 , I Ermakov2,3 and A Rybin2 1 St.-Petersburg Department of V. A. Steklov Mathematical Institute RAS Fontanka 27, St.-Petersburg, 191023, Russia 2 ITMO University, Kronverkskiy 49, 197101, St.Petersburg, Russia 3 Saint Petersburg State University, University Embankment, 7-9, St Petersburg, Russia Received 1 March 2017, revised 9 September 2017 Accepted for publication 13 September 2017 Published 23 October 2017 Abstract A model describing the interaction between two-level atoms and a single mode field in an optical cavity enclosed by a medium with Kerr nonlinearity is considered in our paper. We study the model within the framework of the analytical solution obtained by application of the quantum inverse method (QIM). Time evolution of the atomic inversion is calculated and transition elements of photons are represented in the determinant form. The obtained answers depend on the solutions of the Bethe equations. We provide the numerical solutions of these equations for the different parameters of the model and study the behaviour of certain dynamical correlation functions. Keywords: Tavis–Cummings model, Kerr-nonlinearity, quantum optics, quantum inverse method, exactly solvable models, Bethe equations, dynamical correlation functions (Some figures may appear in colour only in the online journal) 1. Introduction The exactly solvable models of quantum nonlinear optics [1–6] allow one to study the behaviour of strongly correlated systems in ways that otherwise would be impossible. One of the most fundamental models in cavity quantum electrodynamics—exactly solvable for its eigenstates and eigenvalues—is the Jaynes–Cummings (JC) model [7]. It describes the interaction of the single mode of a cavity field with the two-level atom. In spite of its simplicity the model exhibits many nonclassical features caused by its intrinsic nonlinearity, such as, for example, collapses and revivals of the atomic inversion [8–11]. The JC model has been 1751-8121/17/464003+20$33.00 © 2017 IOP Publishing Ltd Printed in the UK 1 N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 thoroughly studied both theoretically [12] and experimentally [13]. The fiftieth anniversary of this model was recently noted [14]. Due to the progress of circuit quantum electrodynamics [15] the JC model has again attracted the attention of physicists [16]. Many interesting applications of the JC model concerned with its generalizations go far beyond its original concept. Natural generalization of the JC model is to consider an ensemble of N noninteracting twolevel atoms coupled to a single mode of a cavity field. The model describing such an ensemble is known as the Tavis–Cummings (TC) model. It was solved by Tavis and Cummings [17] at exact resonance, and by Hepp and Lieb for finite detuning [18]. The TC model may also be expanded in many other ways. For example, one can consider an array of coupled cavities, the so-called Jaynes–Cummings–Hubbard model [19], or abandon the rotating wave approx imation [20]. Recently, models that include either a nonideal cavity or a dynamical Stark shift have again attracted attention [21–24]. Nonlinear effects in cavities can provide new tools in quantum state engineering[25–28]. If the cavity is not ideal an effective Hamiltonian can be derived [29–31], which adds a fourth-order term in the boson operators to the simple TC Hamiltonian. We consider the integrable case of such a model—the integrable generalized Tavis–Cummings (IGTC) model. The presence of a Kerr-like medium changes the behaviour of the system significantly. For example, it crucially changes the Rabi oscillations in the JC model [30, 31]. Kerr-type systems can also be applied to the description of the nonlinear oscillators [32, 33] and Bose–Hubbard dimers [34–36]. The TC model belongs to a class of integrable models known as Gaudin–Richardson systems [37, 38]. The IGTC model belongs to a different set of integrable models connected with the so called XXX rational R-matrix. It was solved exactly in [39] by the Quantum Inverse Method (QIM) [40–42]. The QIM allows us to obtain the exact expressions for the energy spectrum of the model and its dynamical correlation functions. In this work we show that by using the so-called determinant representation [43] it becomes possible to obtain some manageable expressions for correlation functions. In the QIM approach it is considered that the model is solved if the analytical expression for the eigenenergies, eigenvectors and correlation functions are expressed through the solutions of Bethe equations. The Bethe equations being a set of coupled nonlinear algebraic equations dependent on the model under consideration. Because of their nonlinearity, Bethe equations are not easy to solve analytically or even numerically. Several approaches to this problem have been developed, for example [44], which have been successfully applied to the investigation of Bethe equations of the Gaudin and Richardson models [45]. For the XXX Heisenberg chain with spin-12 another technique was developed—the theory of deformed strings (for its extensive review see [46]). In this paper we present some exact numerical solutions of Bethe equations for the IGTC model in the TC limit. The paper is organized in the following way. In the second section we provide a description of the IGTC model as a generalization of the TC model. In the third section we discuss in detail the application of the QIM to the model. We present the Bethe equations, the eigenvectors and find the spectrum of the model. In the fourth section the analytical expression for the dynamical correlation functions are obtained and presented in determinant form. In the fifth, a numerical analysis of the Bethe equations and correlation functions is presented in detail. In the appendix we provide a number of explicit examples of the solutions of Bethe equations and present the numerical values of all the quantities we need in the calculation. 2 N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 2. Integrable generalized Tavis–Cummings model As the starting point we consider an ensemble of N two-level non interacting atoms of one sort in the cavity. We assume that only one mode of the cavity field has to be taken into account, whereas the other ones are suppressed. Each atom has ground |ψi− and exited |ψi+ states. The TC model model is defined by the Hamiltonian (in the units = c = 1): HTC = ωa† a + ω0 Sz + g(a† S− + aS+ ), (1) where ω and ω0 are frequencies of the cavity mode and the atomic system, g is the cavity-atom coupling constant. a and a† are the usual annihilation and creation operators for the cavity field, which satisfy usual commutation relations [a, a† ] = 1. We also introduced the collective N-atom Dicke operators S± , Sz (spin operators for which total spin S N/2 ) as: N N N N 1 i Sz ≡ Siz = S± ≡ Si± = σi± , σz , (2) 2 i=1 where Si+ = |ψi+ ψi− |, i=1 i=1 Si− = |ψi− ψi+ |, Siz = i=1 1 − (|ψ ψi− | − |ψi+ ψi+ |), 2 i (3) where σi± = 12 (σ x ± iσ y ), and σix , σiy , σiz are the Pauli matrices acting each on ith site. Operators Sz and S± satisfy the commutation relations: + − [S , S ] = 2Sz , [Sz , S± ] = ±S± , (4) of the su(2) algebra. In the Hamiltonian (1) the first term describes single quantized mode of the cavity field, the second one describes the atomic inversion of the whole system, and the first term of the interaction part describes an atomic transition from the excited state to the ground state accompanied by the emission of a photon, whereas the second one describes the reverse process. The number of excitations M and the Casimir operator S2: (5) M ≡ Sz + a† a, 2 S ≡ S+ S− + Sz (Sz − 1) (6) are two nontrivial constants of motion [HTC , M] = [HTC , S2 ] = 0. In this paper we consider the exactly solvable extension of the TC model that describes the system in a complex environment and takes into account the spin–spin interaction. The Hamiltonian for the IGTC model can be written in the form [39]: HK = ωa† a + ω0 Sz + g(a† S− + aS+ ) + γ(a† a† aa + (Sz )2 ). (7) Here, the fourth-order term in the boson operators describes a Kerr-like medium, whereas (Sz )2 describes the spin–spin coupling. When γ = 0, HK reduces to the TC model (1) and this becomes the JC model [7] for N = 1 (S = 12 ). When S = 12 , (Sz )2 = 1 and HK with γ = 0 is the JC model with Kerr nonlinearity [30]. It is easy to verify that operator M commutes with the Hamiltonian HK : [HK , M] = 0 , so we can consider another Hamiltonian H = g−1 (HK + (γ − ω)M − γM2 ) and [HK , H] = 0. We can then write 3 N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 H = ∆Sz + (a† S− + aS+ ) + ca† aSz , (8) where c = −2g−1 γ is a new effective coupling constant and ∆ = g−1 (γ + ω0 − ω) is the frequency shifted detuning. One should note that the last term on the right-hand side of (8) causes photon number dependent changes in the atomic transitions and therefore describes a Stark shift. Henceforth, we shall consider H, but the same results can immediately be extended to the model with Kerr nonlinearity, Hamiltonian (7), through the mapping given above. 3. The determinant representation of correlation functions To apply QIM to the solution of the model we consider the two 2 × 2 matrix operators LB (λ), LS (λ) for bosons and spins respectively [39]: λ − ∆ − c−1 − ca† a a† LB (λ) = , (9) a −c−1 λ − cSz cS+ L (λ) = , (10) S cS− λ + cSz in which λ is a complex number. Note that the elements of LB (λ) and LS (λ) mutually commute. For the monodromy matrix of the QIM T(λ) we can now set A(λ) B(λ) T(λ) = LB (λ)LS (λ) = , (11) C(λ) D(λ) so that B(λ) = λX − Y; X = a† + cS+ , Y = (1 + c∆)S+ − ca† Sz + c2 a† aS+ ; (12) [X, Y] = 0, and We also have C(λ) = λa − (S− + caSz ) ≡ λa − Z. (13) A(λ) = (λ − ∆ − c−1 − ca† a)(λ − cSz ) + ca† S− , (14) D(λ) = caS+ − c−1 (λ + cSz ). Multiplying L-operators in an order different than in (11) we obtain the following expression for the monodromy matrix A(λ) B(λ) T(λ) = LS (λ)LB (λ) = . (15) C(λ) D(λ) The entries of this matrix satisfy the involution relations [36] B(λ) = C+ (λ∗ ), C(λ) = B+ (λ∗ ). (16) The introduced monodromy matrices satisfy the intertwining relation R(λ, µ)T(λ)T(µ) = T(µ)T(λ)R(λ, µ) (17) R(λ, µ)T̃(λ)T̃(µ) = T̃(µ)T̃(λ)R(λ, µ) 4 N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 with the rational R-matrix f (µ, λ) 0 0 0 0 g(µ, λ) 1 0 , R(λ, µ) = (18) 0 1 g(µ, λ) 0 0 0 0 f (µ, λ) with the enries f (µ, λ) = 1 − c , µ−λ g(µ, λ) = − c . µ−λ The traces of the monodromy matrices (11)and (15) coincide: TrT(λ) = TrT(λ) = τ (λ), and commute for arbitary complex numbers λ, µ: [τ (λ), τ (µ)] = 0. (19) In the explicit form: τ (λ) = λ2 − λ(cM+∆ + 2c−1 ) + cH, (20) where M is the number operator (5) and H is the Hamiltonian (8). It can be checked that H = c−1 τ (0), (21) ∂τ (λ) |λ=0 −c−1 ∆ − 2c−2 . M = −c−1 ∂λ It can be shown that MB(λ) = B(λ)(M + 1), and likewise S2 B(λ) = B(λ)S2 . So B(λ) acts as a creation operator for the quasi-particles, while C(λ) is an annihilation operator. The M-particle state vectors are constructed in the usual fashion for the QIM method M M |(22) ΨS,M ({λ}) = B(λj ) | ΩS = (λj X − Y) | ΩS , j=1 j=1 where the vacuum state |ΩS = |0|S, −S (a|0 = 0; S− |S, −S = 0, with 2 z S |S, −S = S(S + 1)|S, −S, and S |S, −S = −S|S, −S) is annihilated by operator C C(λ) | ΩS = 0, and is the eigenstate of the operators A, D : where A(λ) | ΩS = a(λ) | ΩS , D(λ) | ΩS = d(λ) | ΩS , (23) a(λ) = (λ − ∆ − c−1 )(λ + cS), d(λ) = −c−1 (λ − cS). (24) In formula (22) a short-hand notation {x} ≡ {λ1 , λ2 , . . . , λM } is used. In the explicit form the state vector (22) has the form M |(25) ΨS,M ({λ}) = (−1)M−m em Xm YM−m | ΩS , m=0 5 N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 where em = λi1 λi2 . . . λim i1 <i2 <...<im is the elementary symmetric function [47]. The conjugated M-particle state vectors are equal to ΨS,M ({λ}) | = ΩS | M j=1 C(λj ) = ΩS | M j=1 [λj a − (S− + caSz )] (26) M = ΩS | (−1)M−m em am ZM−m m=0 and ΩS | B(λ) = 0. By the construction the state vectors (22) and (26) are symmetric functions of their arguments {λ} ≡ {λ1 , λ2 , ..., λn }. The state vectors (22), (26) are eigenvectors of the transfer matrix, and thus of the Hamiltonian if {λ} are the roots of the Bethe equations, which here take the form, for n = 1, 2, . . . , M , M λn − λj + c λn + cS = (1 + ∆c − cλ ) , n (27) λn − cS λn − λj − c j=1,j=n where 0 S N2 . Evidently, there are K = min(2S, M) + 1 (modulo the permutation group) sets of solutions of these M Bethe equations. The complex valued roots are pairwise conjugated [48]. The eigenvalues of the transfer matrix (20): M M c c θ(28) (1 − ) − d(µ) (1 + ). S,M (µ) = a(µ) µ − λj µ − λj j=1 j=1 From the equation (21) we see that the M-particle eigenenergies of the Hamiltonian (8) are equal to: H | ΨS,M ({λ}) = ES,M | ΨS,M ({λ}), M M S S c c (29) . ES,M = − S∆ + 1− 1+ c λj c λj j=1 j=1 The ground state of the Hamiltonian (8) corresponds to the minimal value of eigenenergy. The set of solutions of Bethe equations that defines this state will be labeled σg : {λσg }. The form of the Bethe equations drastically depends on the parameter c. In the case of vanishing nonlinearity c → 0 the Bethe equations (27) transform to another set of equations (30), which refers to the well-studied TC-model [37, 38]. M 2S 2 − λ̃n + ∆ = . (30) λ̃n λ̃ − λ̃j j=1,j=n n 6 N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 The energy spectrum of the TC-model can be expressed in the following form: M 1 σ . ES,M = −S ∆ + 2 (31) σ λ̃ j=1 j Consider the state vectors constructed by operators (16) M M j ) | ΩS , Ψ j ). S,M ({λ}) = S,M ({λ}) |= ΩS | |(32) Ψ B(λ C(λ j=1 j=1 It was proved in [41] that on the solutions of Bethe equations S,M ({λ}) = νM | ΨS,M ({λ}), Ψ S,M ({λ}) |= ν −1 ΨS,M ({λ}) |, |(33) Ψ M where for the model under consideration M M λj − cS −1 = (1 + ∆c − cλ ) = ν . j (34) M λj + cS j=1 j=1 Note, that from Bethe equations (27) it follows that M λn + cS = 1. (1 + ∆c − cλn ) (35) λn − cS n=1 In [39] it was proved that Bethe state vectors form a complete orthogonal set ΨS,M ({λσ1 }) | ΨS,M ({λσ2 }) ∼ δσ1 ,σ2 | ΨS,M ({λσ })ΨS,M ({λσ }) | (36) = I. Nσ2 σ Index σ denotes the independent sets of solutions of Bethe equations (27), and the summation is over all K sets of solutions. The scalar product of the Bethe state vectors for M = 1, 2 can be calculated explicitly: and S1 (µ, λ) ≡ ΨS,1 (µ)|ΨS,1 (λ) = −c2 s2 + s(2 + c(2∆ − λ − µ)) + λµ, (37) S2 (µ1 , µ2 , λ1 , λ2 ) ≡ ΨS,2 (µ1 , µ2 )|ΨS,2 (λ1 , λ2 ) = 2c4 s4 − 4c4 s3 + 2c4 s2 − (λ1 + λ2 ) −2c3 s3 + 2c3 s2 + c 4s2 − 2s (c∆ + 1) − (µ1 + µ2 ) −2c3 s3 + 2c3 s2 + 4cs2 (c∆ + 1) − 2cs(c∆ + 1) + 4c2 s2 (c∆ + 1) + 2c2 λ1 λ2 s2 + 2c2 µ1 µ2 s2 + (c∆ + 1) −8c2 s3 + 4c2 s2 − 2c2 s + 8s2 − 4s (c∆ + 1) + 2 (λ1 + λ2 ) (µ1 + µ2 ) s(c∆ + 1) − 2c (λ1 + λ2 ) µ1 µ2 s (38) The further straightforward calculation of scalar products for M > 2 is rather difficult because the number of commutation relations (one needs to evaluate in order to get the answer) grows exponentially. To obtain the analytical expression for the correlators for the arbitrary M we shall use Slavnov’s formula for the scalar products [49–51]. This formula, adopted for the model under consideration, has the form: − 2cλ1 λ2 (µ1 + µ2 ) s + 2λ1 λ2 µ1 µ2 . 7 N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 SM ({µ}, {λ}) = ΨS,M ({µ}) | ΨS,M ({λ}) M cM j=1 d(λj ) (39) det T({µ}, {λ}), = j>k (µk − µj ) α<β (λβ − λα ) where the entries of the M × M matrix T({µ}, {λ}) are ∂ Tab ≡ T(µa , λb ) = −c−1 θ(µa , {λ}) ∂λb M M c c 1 a(µa ) 1− − d(µa ) 1+ . = (λb − µa )2 µ a − λj µ a − λj j=1,j=b (40) j=1,j=b It is supposed here and below that {λ} are the solutions of Bethe equations, while {µ} is the set of arbitrary parameters. Functions a(µ), d(µ) are defined in (23). The square of the norm of the eigenvectors is calculated by the Gaudin formula [37, 43] which is (39) in the limit {µ} → {λ}: M λα − λβ + c N = SM ({λ}, {λ}) = c d2 (λj ) det Φ({λ}), (41) λ α − λβ 2 M j=1 α=β where the entries of the M × M matrix Φ are M ∂ a(λa ) λa − λk − c . Φ ln (42) ab = − ∂λb d(λa ) λk − λa − c k=1,k=a The determinant representation (39) may be used in calculation of the transition element of the photon annihilation operator n Ψ (43) S,M−n ({λ }) | a | ΨS,M ({λ}), where {λ} and {λ } are the solutions of Bethe equations (27) for M and M − n particles respectively. We note that from the definition (13) it follows that limλ→∞ λ−1 C(λ) = a and hence SM ({µ}, {λ}) n Ψ lim . (44) S,M−n ({µ}) | a | ΨS,M ({λ}) = µ1 ,µ2 ,...,µn →∞ µ1 µ2 . . . µn Replacing the arbitrary parameters {µ} by the solutions of Bethe equations {λ } we obtain the answer for (43). Denoting µ−M+a−1 ∂ a−1 a T(µa , λb ) µa →∞ (a − 1)! ∂µa−1 a Vab = lim we obtain the following answer AM,n ({µ}, {λ}) ≡ ΨS,M−n ({µ}) | an | ΨS,M ({λ}) M−n M M (45) j=1 α=1 (µj − λα ) M det T(n) ({µ}, {λ}). =c d(λj ) j>k>n (µk − µj ) α<β (λβ − λα ) j=1 8 N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 Here set {µ} ≡ {µ1 , µ2 , ..., µM−n } has the length M − n, whereas the set {λ} ≡ {λ1 , λ2 , ..., λM } has the length M. The entries of the M × M matrix T(n) are equal to Vab for a n and are Tab (40) for a > n, 1 b M . To obtain the answer for the transition element (43) we have to change parameters {µ} on the solutions of Bethe equations for M − n particles. For instance, for n = 1 the answer for the transition element is ΨS,M−1 ({λ }) | a | ΨS,M ({λ}) M−1 M M j=1 α=1 (λj − λα ) M det T(1) ({λ }, {λ}), d(λj ) =c − λ ) (λ − λ ) (λ β α j α<β j>k>1 k (46) j=1 where the entries of M × M matrix −1 T 21 T(1) ({λ }, {λ}) = .. . TM1 −1 T22 .. . TM2 −1 T23 .. . TM3 ... ... .. . ... −1 T2M .. . . TMM To find the transition element of the creation operator one can take the complex conjugation of the transition element of the annihilation operator and obtain: ΨS,M−n ({λ }) | an | ΨS,M ({λ})∗ = ΩS | = ΩS | = ΩS | M B+ (λj )(a† )n j=1 M j=1 M−n j=1 j )(a† )n C(λ M−n j=1 M−n j=1 C(λj )an M j=1 B(λj ) | ΩS ∗ C+ (λj ) | ΩS ) | ΩS B(λ j ν (47) = M−n ΨS,M ({λ}) | (a† )n | ΨS,M−n ({λ }), νM where the definition (32) and the properties (16), (33) have been used, and λ are the solutions of the Bethe equations (27) with M − n particles. The obtained formula allows us to express the transition element of the creation operator in the determinantal form: A†M,n ({λ}, {λ }) ≡ΨS,M ({λ}) | (a† )n | ΨS,M−n ({λ }) νM (48) = A∗M,n ({µ}, {λ}), νM−n where A∗M,n ({µ}, {λ}) is the complex conjugated coefficient (45) on the solutions of the Bethe equations. A similar determinant representation for the transition elements (45) and (48) for the bimodal Bose–Hubbard model has been studied in [36]. The obtained representations for the transition elements (45) and (48) allow us to calculate different n-photon time-dependent correlation functions. We define the n-photon Green function an (t)(a† )n M as the average taken over the M-particle ground state of the model | ΨS,M ({λσg }): 9 N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 an (t)(a† )n M = σ = g σ −ES,M eit(ES,M ) σ 1 ΨS,M ({λσg }) | e−iHt an eiHt (a† )n | ΨS,M ({λσg }) Nσ2g ΨS,M ({λσg }) | an | ΨS,M+n ({λσ }) 2 Nσ2g Nσ(49) × ΨS,M+n ({λσ }) | (a† )n | ΨS,M ({λσg }) σg σ 2 ν σ −ES,M eit(ES,M ) M+n σg σ n = ({λ }) | a | Ψ ({λ }) Ψ σg , S,M S,M+n 2 N2 N νM σ σ g σ where the sum is taken over K = min(2S, M + n) + 1 sets of solutions of Bethe equations (27) for M + n particles λσ labeled by σ , and the complete orthogonal set (36) of eigenstates of the Hamiltonian (8) was used. Substituting formulas (29), (34), (41) and (45) into (49) we obtain the final answer. The technique developed allows us to calculate different dynamical correlation functions of the considered model and, respectively, of the TC model in the c → 0 limit. 4. Time evolution of the atomic inversion Knowing the projection of the initial state on the state vectors of the model one can obtain the answers for the correspondent dynamical correlation functions. Let us consider the time evolution of the coherent states. Let us assume that in the initial state | Φ0 the spin system is in the ground state | S, −S (S− | S, −S = 0 ) and the field is in the coherent state | α α ∈ C (a | α = α | α) |α|2 † |(50) Φ0 =| α | S, −S = e− 2 eαa | 0 | S, −S. The time evolution of the initial state can be obtained by application of the evolution operator U(t) = exp(iHt), so we have | Φ(t) = U(t) | Φ0 . We may use representations (26) and (25) to find the projections of the initial states on the Bethe state vectors: ΨS,M ({λ}) | Φ0 = (−1)M+1 (cαS)M e− |α|2 2 M m=0 em (cS)−m ≡ fM ({λ}), (51) M |α|2 M − 2 Φ0 | ΨS,M ({λ}) = (cᾱS) e (−1)M−m em (cS)−m ≡ gM ({λ}). m=0 The evolution of field intensity can be constructed through the complete set of eigenfunctions | ΨS,M ({λσ }): σ σ eit(ES,M1 −ES,M3 ) Φ(t) | (a ) a | Φ(t) = Nσ21 Nσ22 Nσ23 (52) M σ ,σ ,σ † n n 1 σ1 × fM ({λ })gM ({λ σ3 2 3 })A†M,n ({λσ3 }, {λσ2 })AM,n ({λσ2 }, {λσ1 }). The summation here is over all sets of solutions of correspondent Bethe equations. Knowing the answer for the field intensity we can find the evolution of atomic inversion applying equality Φ(t) | Sz | Φ(t) = Φ(t) | M | Φ(t) − Φ(t) | a† a | Φ(t) = n − S − Φ(t) | a† a | Φ(t), 10 (53) N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 Figure 1. Temporal evolution of the atomic inversion Sz . The atoms are initially in the ground state with the total spin S = 32 , and the field is in the coherent state with n = 20. The parameters are equal to c = 0.01/0.5/2.0 and ∆ = 0.0/1.0/2.0/3.0 . where M is the number operator (5), and n = |α|2 is the average number of photons in a cavity. The examples of the numerical results for Sz (t) for different model parameters are given in the figure 1. For the sake of simplicity we consider the case when the cavity contains only 3 atoms, so the total spin S changes in the range [− 32 , 32 ]. In the initial state | Φ0 =| α | S, −S all the atoms are in ground state. It is evident that the state | Φ0 is not an eigenstate, so it rapidly decays in time. We see that the collapses and revivals of Rabi oscillations for small coupling constants (c = 0.01) are similar to those of the Jaynes-Cumming model [9] and the time of the first revival is proportional to the average number of photons in a cavity t ∼ n. As the effective coupling constant c increases the revivals become more localized in time, their amplitudes and the time of the first revival decrease because of the dominating role of a Kerr medium. The presence of detuning affects the form of the revivals for small values of the effective coupling c. Examples of neat evaluation of the Sz (t) are discussed in the next section. 11 N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 Figure 2. The roots distribution λσ i of the Bethe equations (27) with the parameters c = 0.5, M = 35, ∆ = 1.0 and S = 32 . Note that not all the roots are presented in this figure. Figure 3. The roots distribution λσ i of the Bethe equations (27) with the parameters c = 0.0, M = 35, ∆ = 1.0 and S = 32 . 12 N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 1 Figure 4. Behaviour of the single solution {λ } of the Bethe equations (27) for different c. The represented solutions refer to the ground state of the system with ∆ = 1.0, S = 32 , M = 35. 5. Numerical analysis of the Bethe equations and evaluation of the correlation functions The solutions of the Bethe equations (27) give us complete information about the system under consideration. The roots distribution of the Bethe equations (27) and the spectrum for IGTC for the fixed values of parameters c, ∆, S, M are presented in figure 2, and in tables A5 and A6 of the appendix. The roots distribution of the Bethe equations (30) and the spectrum for the TC model for the fixed values of parameters ∆, S, M are presented in figure 3. Solutions of the equations (27) become close to the solutions of the equations (30), when c → 0. The example of such transformation is presented in figure 4, for the solution {λ1 } which refers to the ground state of the system. We use an iterative approach to the solution of the Bethe equations (27). Within this framework it is crucial to choose the correct initial assumption for the roots distribution in the first iteration. The algorithm for finding the correct initial assumption is the following: (I) Solve the equations (30) for ∆ = 0.0, taking the initial assumption in a string form with all roots having the same real part {a ± ib · 1, a ± ib · 2, a ± ib · 3, ...}. By varying the parameters a and b we find the correct initial assumption for equations (30) which do not lead to nonphysical solutions containing coinciding roots. (II) Use the solution of the equations (30) as an initial assumption for the equations (27) with small c and Δ (the optimal step can be found experimentally). (III) Vary c and Δ slowly until they take the desired values. In table A1 we present the sets of solutions of the Bethe equations (27) for this system, σ eigenenergies E3/2,M , and norms of the Bethe wavefunctions Nσ2. Using the data from table A1 we evaluate the numerical values of the scalar products f15 ({λσ }), g15 ({λσ }), presented in 13 N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 table A2, and we also evaluate the numerical values for the photonic transition elements: A15,1 ({λσ }, {λσ }) and A†15,1 ({λσ }, {λσ }), presented in tables A3 and A4. 6. Conclusions In this paper the QIM approach to the IGTC model was considered. We have obtained the determinant representation for the norm of the Bethe wavefunctions Nσ2 and for the transition elements of the photons AM,n ({λσ }, {λσ }), A†M,n ({λσ }, {λσ }). The knowledge of these elements allows us to investigate any dynamical correlation function of the model provided the roots of the Bethe equations are known. The solutions of the Bethe equations for different values of parameters of the model were calculated numerically. The obtained results were applied for the evaluation of the atomic inversion Φ(t)|Sz |Φ(t). Acknowledgments This work was supported by the Russian Science Foundation (grant № 16-11-10218). Appendix Here we provide the solutions of the Bethe equations and the numerical values of the photonic transition elements and norms of wavefunctions. Table A1. Solutions of the Bethe equations for c = 0.5, ∆ = 1.0, S = 32 , M = 14, 15. M 14 15 λ11 λ12 λ13 λ14 λ15 λ16 λ17 λ18 λ19 −0.239 824 −0.750 193 −0.241 867 0.176 368 1 E3/2,M N12 λ21 λ22 λ23 λ24 λ25 λ26 λ27 λ28 λ29 −0.750 262 0.167 747 1.773 790 3.798 26 ± 4.182 16 I 2.422 63 ± 2.016 87 I 1.977 72 ± 1.038 23 I 3.022 76 ± 3.038 48 I 4.856 00 ± 5.581 94 I 2.812 93 ± 2.333 51 I 3.444 29 ± 3.334 88 I 4.247 73 ± 4.467 06 I 2.325 25 ± 1.401 43 I 5.333 30 ± 5.858 46 I 2.014 00 ± 0.477 29 I 14.3938 15.2107 22 1.328 07 × 10 −0.738 31 −0.365 95 3.128 92 + 1.922 48 I 2.736 71 + 1.071 69 I 2.512 42 + 0.333 28 I 3.678 73 + 2.886 82 I 4.411 73 + 4.000 60 I 5.430 90 + 5.382 55 I 1.111 14 × 1025 −0.740 738 −0.353 435 2.598 74 2.727 31 + 0.628 46 I 3.487 14 + 2.220 83 I 3.037 37 + 1.372 10 I 4.848 37 + 4.287 62 I 5.900 50 + 5.663 55 I 4.079 92 + 3.179 73 I (Continued ) 14 N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 Table A1. (Continued ) M 14 15 4.805 44 2 E3/2,M N22 5.0842 25 1.053 91 × 10 9.918 35 × 1027 −0.830 452 5.016 24 + 3.900 16 I 3.332 98 + 1.1325 I 6.012 43 + 5.252 18 I 3.021 86 + 0.475 188 I 4.298 96 + 2.825 74 I 3.751 56 + 1.914 65 I 2.988 33 + 0.158 96 I 4.102 89 + 2.188 17 I 3.639 48 + 1.393 31 I 6.474 53 + 5.530 95 I 4.690 08 + 3.104 76 I 3.277 31 + 0.710 51 I 5.443 43 + 4.1804 I −5.387 34 −5.659 03 3.872 31 + 1.188 78 I 5.601 29 + 3.852 93 I 4.3222 + 1.937 11 I 3.506 08 + 0.553 672 I 4.883 24 + 2.812 31 I 6.588 51 + 5.173 88 I 4.676 24 + 2.191 07 I 3.447 65 + 0.216 031 I 4.190 98 + 1.426 75 I 3.789 25 + 0.767 612 I 5.271 76 + 3.076 48 I 6.023 49 + 4.122 97 I 7.045 06 + 5.446 49 I −16.3119 −17.1358 −0.823 073 2.849 73 λ31 λ32 λ33 λ34 λ35 λ36 λ37 λ38 λ39 3 E3/2,M N32 4.496 29 × 1027 3.014 53 3.250 13 λ41 λ42 λ43 λ44 λ45 λ46 λ47 λ48 λ49 4 E3/2,M N42 2.0875 × 10 4.9616 × 1030 2.996 98 30 2.718 93 × 1033 Table A2. Numerical values for the functions c = 0.5, ∆ = 1.0, S = 3 2, M = 15. f15 ({λσ }) σ g15 ({λσ }) −3.542 25 × 109 1.880 59 × 1012 −4.846 51 × 1014 8.586 41 × 1016 1 2 3 4 f15 ({λσ }) and g15 ({λσ }) for −5.716 61 × 1011 1.314 35 × 1013 −1.474 09 × 1014 1.094 85 × 1015 Table A3. Numerical values for the transition element A15,1 ({λσ }, {λσ }) for c = 0.5, ∆ = 1.0, S = 32 , M = 15. σ, σ 1 2 3 4 1 2 3 4 −1.459 83 × 1024 1.657 28 × 1024 −2.810 49 × 1023 3.759 25 × 1022 3.739 37 × 1023 −1.194 78 × 1027 1.282 74 × 1027 −1.716 35 × 1026 −6.080 01 × 1022 2.306 75 × 1026 −5.405 08 × 1029 5.624 94 × 1029 7.897 56 × 1021 −2.996 71 × 1025 8.056 98 × 1028 −2.6862 × 1032 15 N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 † σ σ Table A4. Numerical values for the transition element A15,1 ({λ }, {λ }) for c = 0.5, ∆ = 1.0, S = 32 , M = 15. σ, σ 1 1 2 3 4 −1.260 67 × 1024 6.8671 × 1022 −5.578 47 × 1020 3.413 26 × 1018 2 3 4 7.456 56 × 1024 −1.143 15 × 1027 5.879 13 × 1025 −3.598 44 × 1023 −2.785 92 × 1025 5.071 59 × 1027 −5.692 46 × 1029 2.709 89 × 1028 8.631 94 × 1025 −1.571 59 × 1028 2.024 05 × 1030 −3.0869 × 1032 Table A5. Example of the numerical solutions of the Bethe equations for c = 0.5, ∆ = 0.1, M = 4 , and different values of S. Note: this example is easy to reproduce. S 1 2 3 2 1 λ11 1.048 08 ± 0.508 37 i 1.349 52 ± 0.426 617 i λ12 1.291 19 ± 1.890 55 i 1.532 19 ± 1.617 69 i 2 1.577 16 ± 0.365 265 i 1.758 06 ± 0.310 326 i 1.758 01 i ± 1.408 80 i 1.954 84 ± 1.240 21 i −6.420 34 −8.225 79 λ13 1 −2.328 55 ES,4 λ21 λ22 −0.299 68 ± 0.401 56 i 0.692 35 ± 1.446 50 i 0.435 412 ± 2.039 59 i 0.878 76 λ23 2 ES,4 λ31 λ32 λ33 3 ES,4 −4.463 41 2.078 55 −0.733 00 −0.230 48 1.167 94 ± 1.042 22 i 1.534 92 ± 0.872 592 i 1.244 67 1.528 05 −1.114 21 −1.550 59 −2.216 33 −0.248 57 ± 1.948 60 i −0.449 67 ± 1.030 01 i −3.847 29 1.312 63 ± 0.402 60 i −0.948 36 ± 0.529 91 i −1.074 33 −1.345 80 ± 0.584 813 i 3.693 89 −0.733 658 0.987 374 1.236 30 −0.863 032 ± 1.974 70 i −1.244 56 ± 1.277 96 i λ41 −1.462 16 ± 0.578 535 i 1.340 77 λ42 λ43 4.900 38 4 ES,4 −1.667 71 2.016 06 −1.917 37 ± 0.611 298 i λ51 −1.377 97 ± 2.0319 i λ52 λ53 5.790 68 5 ES,4 16 N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 Table A6. Example of the numerical solutions of the Bethe equations for c = 0.5, ∆ = 0.1, M = 15, and different values of S. S 1 2 1 3 2 2 λ11 λ12 λ13 λ14 λ15 λ16 λ17 λ18 λ19 2.327 41 ± 0.353 73 i 2.651 60 ± 0.953 04 i 3.058 14 ± 1.657 19 i 3.562 51 ± 2.463 49 i 4.187 08 ± 3.385 09 i 4.974 93 ± 4.461 72 i 6.039 78 ± 5.810 56 i 2.061 13 2.475 83 ± 0.259 97 i 2.800 88 ± 0.826 68 i 3.196 18 ± 1.503 89 i 3.682 14 ± 2.288 67 i 4.284 84 ± 3.194 15 i 5.049 01 ± 4.258 94 i 6.087 96 ± 5.599 00 i 2.092 62 2.602 27 ± 0.168 87 i 2.942 25 i ± 0.709 56 i 3.330 06 ± 1.362 10 i 3.801 12 ± 2.125 04 i 4.384 31 ± 3.012 80 i 5.126 06 ± 4.063 80 i 6.139 58 ± 5.393 12 i 2.098 94 3.076 94 ± 0.599 44 i 3.459 53 ± 1.229 65 i 3.918 47 ± 1.971 24 i 4.484 44 ± 2.840 44 i 5.205 25 ± 3.876 15 i 6.194 09 ± 5.193 00 i 2.654 82 2.099 86 2.747 37 1 −5.464 07 ES,15 −10.846 27 −16.150 21 −21.379 55 λ21 λ22 λ23 λ24 λ25 λ26 λ27 λ28 λ19 1.536 27 ± 0.810 20 i 1.915 51 ± 1.628 51 i 2.426 62 ± 2.502 33 i 3.068 13 ± 3.467 91 i 3.876 83 ± 4.575 37 i 4.965 49 ± 5.947 66 i −0.498 71 2.121 58 ± 0.128 31 i 2.410 32 ± 0.672 32 i 2.758 30 ± 1.351 72 i 3.210 46 ± 2.149 73 i 3.790 58 ± 3.071 52 i 4.539 77 ± 4.152 31 i 5.568 37 ± 5.507 53 i −0.810 31 2.577 21 ± 0.550 29 i 2.917 33 ± 1.194 37 i 3.348 63 ± 1.963 57 i 3.902 05 ± 2.864 94 i 4.622 48 ± 3.932 12 i 5.620 50 ± 5.278 39 i 2.342 32 2.113 76 −5.388 47 −10.545 03 1.773 69 ± 0.297 77 i 2.029 09 ± 0.967 17 i 2.426 13 ± 1.727 41 i 2.940 90 ± 2.573 21 i 3.583 30 ± 3.522 57 i 4.392 18 ± 4.619 27 i 5.480 81 ± 5.983 46 i −0.266 25 2 ES,15 5.214 07 λ31 λ32 λ33 λ34 λ35 λ36 λ37 λ38 λ19 3 ES,15 4 λ1 λ42 λ43 λ44 λ45 λ46 λ47 λ48 λ19 −0.032 60 1.370 64 −0.136 68 1.973 76 ± 0.219 89 i 2.228 42 ± 0.807 99 i 2.593 30 ± 1.527 68 i 3.073 33 ± 2.352 55 i 3.683 57 ± 3.290 84 i 4.462 59 ± 4.381 58 i 5.521 61 ± 5.742 75 i −0.536 45 9.982 96 1.817 76 ± 0.620 933 i 2.119 19 ± 1.376 46 i 2.572 49 ± 2.234 50 i 3.171 01 ± 3.196 82 i 3.944 28 ± 4.305 03 i 5.000 11 ± 5.679 89 i −0.327 67 1.702 80 −0.744 50 4.719 66 0.929 48 ± 0.622 77 i 1.368 96 ± 1.540 94 i 1.897 87 ± 2.446 14 i 2.546 98 ± 3.426 90 i 3.359 65 ± 4.543 50 i 4.450 75 ± 5.922 23 i −0.750 16 0.165 66 −0.241 92 14.319 02 4 ES,15 17 −1.087 52 2.052 22 ± 0.487 06 i 2.326 58 ± 1.169 72 i 2.732 67 ± 1.992 29 i 3.287 59 ± 2.940 09 i 4.022 90 ± 4.042 95 i 5.043 27 ± 5.416 79 i 1.916 06 −0.984 28 −0.637 40 −0.424 88 1.440 14 ± 0.334 64 i 1.617 67 ± 1.185 14 i 2.049 60 ± 2.109 40 i 2.645 23 ± 3.106 05 i 3.418 65 ± 4.234 86 i 4.475 19 ± 5.624 12 i −0.123 58 −1.000 49 −0.480 820 9.111 88 (Continued ) N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 Table A6. (Continued ) S 1 2 3 2 1 2 λ51 λ52 λ53 λ54 λ55 λ56 λ57 λ58 λ19 0.207 63 ± 0.682 92 i 0.784 26 ± 1.509 81 i 1.357 24 ± 2.418 91 i 2.022 65 ± 3.405 99 i 2.843 01 ± 4.527 17 i 3.938 44 ± 5.909 29 i −0.309 78 5 ES,15 18.237 59 −1.000 65 −0.433 68 ORCID iDs N M Bogoliubov https://orcid.org/0000-0002-5834-5066 I Ermakov http://orcid.org/0000-0002-2535-5540 References [1] Braak D 2011 Integrability of the Rabi model Phys. Rev. Lett. 107 100401 [2] Batchelor M T and Zhou H-Q 2015 Integrability versus exact solvability in the quantum Rabi and Dicke models Phys. Rev. A 91 053808 [3] Babelon O and Talalaev D 2007 On the bethe ansatz for the Jaynes–Cummings–Gaudin model J. Stat. Mech. P06013 [4] Rybin A, Kastelewicz G, Timonen J and Bogoliubov N 1998 The su (1, 1) Tavis–Cummings model J. Phys. A: Math. Gen. 31 4705 [5] Wolf F A, Vallone F, Romero G, Kollar M, Solano E and Braak D 2013 Dynamical correlation functions and the quantum Rabi model Phys. Rev. A 87 023835 [6] Chaichian M, Ellinas D and Kulish P 1990 Quantum algebra as the dynamical symmetry of the deformed Jaynes–Cummings model Phys. Rev. Lett. 65 980 [7] Jaynes E T and Cummings F W 1963 Comparison of quantum and semiclassical radiation theories with application to the beam maser Proc. IEEE 51 89–109 [8] Cummings F 1965 Stimulated emission of radiation in a single mode Phys. Rev. 140 A1051 [9] Eberly J H, Narozhny N and Sanchez-Mondragon J 1980 Periodic spontaneous collapse and revival in a simple quantum model Phys. Rev. Lett. 44 1323 [10] von Foerster T 1975 A comparison of quantum and semi-classical theories of the interaction between a two-level atom and the radiation field J. Phys. A: Math. Gen. 8 95 [11] Narozhny N, Sanchez-Mondragon J and Eberly J 1981 Coherence versus incoherence: Collapse and revival in a simple quantum model Phys. Rev. A 23 236 [12] Scully M O and Zubairy M S 1999 Quantum Optics (Berlin: Springer) [13] Haroche S and Kleppner D 1989 Cavity quantum electrodynamics Phys. Today 42 24–30 [14] Greentree A D, Koch J and Larson J 2013 Fifty years of Jaynes–Cummings physics J. Phys. B: At. Mol. Opt. Phys. 46 220201 [15] Girvin S, Devoret M and Schoelkopf R 2009 Circuit qed and engineering charge-based superconducting qubits Phys. Scr. T 137 014012 [16] Pimenta H and James D F 2016 Characteristic-function approach to the Jaynes–Cummings-model revivals Phys. Rev. A 94 053803 [17] Tavis M and Cummings F W 1968 Exact solution for an n-molecule-radiation-field Hamiltonian Phys. Rev. 170 379 18 N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 [18] Hepp K and Lieb E H 1973 On the superradiant phase transition for molecules in a quantized radiation field: the Dicke maser model Ann. Phys. 76 360–404 [19] Angelakis D G, Santos M F and Bose S 2007 Photon-blockade-induced mott transitions and x y spin models in coupled cavity arrays Phys. Rev. A 76 031805 [20] Agarwal S, Rafsanjani S H and Eberly J 2012 Tavis-Cummings model beyond the rotating wave approximation: quasidegenerate qubits Phys. Rev. A 85 043815 [21] Hovsepyan G, Shahinyan A and Kryuchkyan G Y 2014 Multiphoton blockades in pulsed regimes beyond stationary limits Phys. Rev. A 90 013839 [22] Baghshahi H, Tavassoly M and Behjat A 2014 Entropy squeezing and atomic inversion in the k-photon Jaynes–Cummings model in the presence of the stark shift and a Kerr medium: a full nonlinear approach Chin. Phys. B 23 074203 [23] Rohith M and Sudheesh C 2014 Fractional revivals of superposed coherent states J. Phys. B: At. Mol. Opt. Phys. 47 045504 [24] Rodríguez-Lara B and Moya-Cessa H 2013 The exact solution of generalized dicke models via Susskind–Glogower operators J. Phys. A: Math. Theor. 46 095301 [25] Leoński W and Tanaś R 1994 Possibility of producing the one-photon state in a kicked cavity with a nonlinear Kerr medium Phys. Rev. A 49 R20 [26] Khalaidovski A, Thüring A, Rehbein H, Lastzka N, Willke B, Danzmann K and Schnabel R 2009 Strong reduction of laser power noise by means of a Kerr nonlinear cavity Phys. Rev. A 80 053801 [27] Imamoglu A, Schmidt H, Woods G and Deutsch M 1997 Strongly interacting photons in a nonlinear cavity Phys. Rev. Lett. 79 1467 [28] Wang H, Goorskey D and Xiao M 2002 Controlling the cavity field with enhanced Kerr nonlinearity in three-level atoms Phys. Rev. A 65 051802 [29] Werner M and Risken H 1991 Quasiprobability distributions for the cavity-damped Jaynes– Cummings model with an additional Kerr medium Phys. Rev. A 44 4623 [30] Gora P and Jedrzejek C 1992 Nonlinear Jaynes–Cummings model Phys. Rev. A 45 6816 [31] Joshi A and Puri R 1992 Dynamical evolution of the two-photon Jaynes–Cummings model in a Kerr-like medium Phys. Rev. A 45 5056 [32] Gevorgyan T, Shahinyan A and Kryuchkyan G Y 2009 Quantum interference and sub-Poissonian statistics for time-modulated driven dissipative nonlinear oscillators Phys. Rev. A 79 053828 [33] Hovsepyan G, Shahinyan A, Chew L Y and Kryuchkyan G Y 2016 Phase locking and quantum statistics in a parametrically driven nonlinear resonator Phys. Rev. A 93 043856 [34] Milburn G, Corney J, Wright E M and Walls D 1997 Quantum dynamics of an atomic BoseEinstein condensate in a double-well potential Phys. Rev. A 55 4318 [35] Olsen M, Chianca C and Dechoum K 2016 Quantum correlations in pumped and damped Bose– Hubbard dimers Phys. Rev. A 94 043604 [36] Bogoliubov N 2016 Time-dependent correlation functions for the bimodal Bose–Hubbard model J. Math. Sci. 213 [37] Gaudin M 1983 La fonction d’onde de Bethe (New York: Masson) [38] Bogoliubov N and Kulish P 2013 Exactly solvable models of quantum nonlinear optics J. Math. Sci. 192 14–30 [39] Bogoliubov N, Bullough R and Timonen J 1996 Exact solution of generalized Tavis-Cummings models in quantum optics J. Phys. A: Math. Gen. 29 6305 [40] Faddeev L 1995 Quantum completely integrable models in field theory 40 Years in Mathematical Physics (Singapore: World Scientific) pp 187–235 [41] Korepin V E, Bogoliubov N M and Izergin A G 1997 Quantum Inverse Scattering Method and Correlation Functions vol 3 (Cambridge: Cambridge university press) [42] Kulish P P and Sklyanin E K 1982 Quantum spectral transform method recent developments Integrable Quantum Field Theories (Berlin: Springer) pp 61–119 [43] Korepin V E 1982 Calculation of norms of Bethe wave functions Commun. Math. Phys. 86 391–418 [44] Dorey P, Dunning C and Tateo R 2007 The ode/im correspondence J. Phys. A: Math. Theor. 40 R205 [45] Faribault A, El Araby O, Sträter C and Gritsev V 2011 Gaudin models solver based on the correspondence between Bethe ansatz and ordinary differential equations Phys. Rev. B 83 235124 [46] Hagemans R and Caux J-S 2007 Deformed strings in the Heisenberg model J. Phys. A: Math. Theor. 40 14605 19 N M Bogoliubov et al J. Phys. A: Math. Theor. 50 (2017) 464003 [47] Macdonald I G 1998 Symmetric Functions and Hall Polynomials (Oxford: Oxford University Press) [48] Vladimirov A A 1986 Proof of the invariance of the Bethe-ansatz solutions under complex conjugation Theor. Math. Phys. 66 102–5 [49] Slavnov N 1989 Calculation of scalar products of wave functions and form factors in the framework of the alcebraic Bethe Ansatz Theor. Math. Phys. 79 502–8 [50] Kitanine N, Maillet J and Terras V 1999 Form factors of the xxz Heisenberg spin-12 finite chain Nucl. Phys. B 554 647–78 [51] Kitanine N, Kozlowski K, Maillet J M, Slavnov N and Terras V 2007 On correlation functions of integrable models associated with the six-vertex r-matrix J. Stat. Mech. P01022 20

1/--страниц