Bethe Ansatz solution for a model of global-range interacting bosons on the square lattice

Jon Links · 2025

Read Bethe Ansatz solution for a model of global-range interacting bosons on the square lattice by Jon Links (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-01989.txt.

Introduction

Jon Links solves a bosonic lattice model with all-to-all interactions by the Bethe Ansatz, a method famed for one dimensional systems. Bosons are particles that can share states freely, unlike fermions. His model lives on a square lattice, breaking the one dimensional mold. Global range interaction, where every site talks to every other, opens the door to exact solutions in higher dimensions.

Exact solutions are given for open, cylindrical, and toroidal edge setups. A generalization to an integrable defect, a flaw that preserves solvability, also follows. That defect case resists a neat closed form, unlike the clean cases. The result widens where exact methods can reach in quantum many body theory. It may guide studies of long range interacting cold atom systems.

Catalog record

CREATORJon Links
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE29 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-01989

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Bethe Ansatz solution for a model of global-range interacting bosons on the square lattice Jon Links Abstract Quantum systems on a one-dimensional lattice are ubiquitous in the study of models exactly-solved by Bethe Ansatz techniques. Here it is shown that including global-range interaction opens scope for Bethe Ansatz solutions that are not constrained to one-dimensional quantum systems. A bosonic model on a square lattice is defined, and the exact Bethe Ansatz solution is provided for open, cylindrical, and toroidal boundary conditions. Generalising the result for an integrable defect leads to a Bethe Ansatz solution that is not exact. 1 Introduction For each of the fundamental particle properties known as spin, boson, and fermion, there exists a fundamental, one-dimensional model accommodating a Bethe Ansatz solution. These are, respectively, the Heisenberg spin-1/2 model [ 1 ] , the Bose gas [ 33 ] , and the Hubbard model [ 34 ] . While the Heisenberg and Hubbard models are defined on a lattice, the Bose gas Hamiltonian is formulated through continuous field operators. The natural discretised version of the Bose gas, the Bose-Hubbard model, is generally considered to not be integrable [ 6 , 16 , 28 , 29 , 30 , 39 , 40 ] . While there have been attempts to formulate alternative lattice discetisations of the Bose gas model, these have generally possessed complicating features, such as requiring intricate alternating-site expressions [ 5 , 26 ] , or being non-hermitian [ 15 , 31 , 48 ] . Moreover, while there are well-developed methods for generating and/or testing integrability in models for spin [ 13 , 18 , 22 , 43 , 47 ] or fermionic [ 11 , 37 ] degrees of freedom using boost (a.k.a. ladder) operator techniques and the Reshetikhin criterion, these do not readily extend to bosonic systems. Consequently, examples of exactly-solved bosonic lattice models are infrequently encountered. Here, a Hamiltonian is introduced describing bosons on a square lattice with nearest-neighbour hopping, and a global-range (a.k.a. infinite-range) collective interaction term. The form of this interaction was proposed in [ 32 ] to model the interaction mediated by an optical cavity in an experiment using 87 Rb cold atoms confined to a stack of square optical lattices. For a 2\times 2 four-site lattice, the same form of interaction appears in [ 19 ] for a system of 164 Dy atoms with long-ranged dipolar interactions, discussed in relation to NOON state formation; see also [ 3 , 20 ] . Other studies involving global-range interactions include [ 7 , 14 , 23 ] for bosonic models, and [ 24 , 41 , 49 ] for spin models. There has been a resurgence of interest in exactly-solvable models with long-range interactions in recent times; e.g. [ 2 , 12 , 17 , 27 ] . These works consider the systems to be one-dimensional spin chains. However, as will be discussed below, once long-range interactions are incorporated there is a decreased clarity in the notion of dimensionality. A case in point is provided by the Haldane-Shastry model [ 21 , 42 ] , the integrability of which does not stem from the typical transfer matrix approach associated with the Yang-Baxter equation [ 4 ] . The model is generally considered to be one-dimensional, but the geometry of the model can be viewed as being a ring with closed boundary conditions embedded in two-dimensional Euclidean space. This is due to the strength of spin interactions in the Hamiltonian being being dependent on the chord distance between sites on the ring [ 21 ] . In the next section the model with global-range interaction is formulated in terms of canonical boson operators associated with the sites of a square lattice. A key element in the analysis is the introduction of the adjacency matrix to describe the Hamiltonian. The Bethe Ansatz solution for the Hamiltonian of this study is obtained by undertaking a canonical transformation to a spectrally equivalent system with hopping on a collection of disjointed lattices. This simplifies the form of the Hamiltonian, and allows the solution to be read off from previously published results. The case of open, toroidal, and periodic boundary conditions are discussed. However it will be seen that this approach, developed for bosons on the square lattice, bears little resemblance to the typical transfer matrix method used for spin systems; e.g. [ 10 ] . Finally, the case of an integrable boundary defect. 2 Integrable model with global-range interaction The hopping interactions of the Hamiltonian will be formulated on an L\times L square lattice, where L is assumed to be even. The approach taken is aided by implementing the notion of an adjacency matrix , and by exploiting the bipartite nature of the lattice [ 25 ] . Let \{a_{j},\,a_{j}^{\dagger}:j=1,\dots,m\}\cup\{b_{j},\,b_{j}^{\dagger}:j=1,\dots,m\} , where m=L/2 , denote mutually commuting sets of canonical boson operators satisfying \displaystyle[a_{j},\,a_{k}^{\dagger}]=[b_{j},\,b_{k}^{\dagger}]=\delta_{jk}I, \displaystyle[a_{j},\,a_{k}]=[a_{j}^{\dagger},\,a_{k}^{\dagger}]=[b_{j},\,b_{k}]=[b^{\dagger}_{j},\,b_{k}^{\dagger}]=0. For adjacency matrix of the form \displaystyle A=\left(\begin{array}[]{ccc}0&|&B\\ -&&-\\ B&|&0\end{array}\right) where the submatrix B is a symmetric matrix, the Hamiltonian reads \displaystyle H \displaystyle=U(N_{a}-N_{b})^{2}+\sum_{j,k=1}^{m}B_{jk}(a_{j}^{\dagger}b_{k}+b_{j}^{\dagger}a_{k}) (4) where \displaystyle N_{a}=\sum_{j=1}^{m}a_{j}^{\dagger}a_{j},\,N_{b}=\sum_{j=1}^{m}b_{j}^{\dagger}b_{j} . The first term on the right-hand side of ( 4 ) is precisely the form of interaction used to model the effect of an optical cavity applied to bosonic cold atoms in an optical square lattice [ 7 , 14 , 23 , 32 ] . Note that N=N_{a}+N_{b} commutes with ( 4 ), indicating that the Hamiltonian conserves total particle number. In addition, there exists a set of conserved operators that follow from the Yang-Baxter integrability of ( 4 ), which have been derived in [ 25 ] . The conserved operators satisfying [C(y),\,C(z)]=0 are given by \displaystyle C(2p) \displaystyle=\sum_{j,k=1}^{m}{B}^{2p}_{jk}(a_{j}^{\dagger}a_{k}+b_{j}^{\dagger}b_{k}), (5) \displaystyle C(2p+1) \displaystyle=2U\sum_{i=0}^{2p}D(2p,i)+\sum_{j,k=1}^{m}{B}^{2p+1}_{jk}(a_{j}^{\dagger}b_{k}+b_{j}^{\dagger}a_{k}) (6) with \displaystyle D(2p,i)=\begin{cases}\displaystyle\sum_{j,k,r,q=1}^{m}{B}^{i}_{jk}{B}^{2p-i}_{rq}(a^{\dagger}_{j}a_{q}a^{\dagger}_{r}a_{k}+b^{\dagger}_{j}b_{q}b^{\dagger}_{r}b_{k}),\qquad i\,\,{\rm even},\\ \displaystyle\sum_{j,k,r,q=1}^{m}({B}^{i}_{jk}{B}^{2p-i}_{rq}+{B}^{2p-i}_{jk}{B}^{i}_{rq})a^{\dagger}_{j}a_{q}b^{\dagger}_{r}b_{k},\qquad i\,\,{\rm odd}.\end{cases} Above, B^{q}_{jk} denotes the (j,k) entry of the matrix power B^{q} . We have N=N_{a}+N_{b}=C(0) and H=C(1)-U(N^{2}+2(m-1)N) . 2.1 Bethe Ansatz solution The Bethe Ansatz solution is obtained by mapping the Hamiltonian ( 4 ) to an equivalent model via a unitary transformation. Let X denote a unitary matrix that diagonalises B , viz. \displaystyle\sum_{p=1}^{m}X^{\dagger}_{jp}X_{pk}=\delta_{jk},\qquad\sum_{p,q=1}^{m}X^{\dagger}_{jp}B_{pq}X_{qk}=\varepsilon_{j}\delta_{jk}, with \{\varepsilon_{j}:j=1,...,m\} denoting the spectrum of B . Introducing \displaystyle a_{k}=\sum_{j=1}^{m}X_{kj}c_{j},\quad b_{k}=\sum_{j=1}^{m}X_{kj}d_{j},\quad a^{\dagger}_{k}=\sum_{j=1}^{m}X^{\dagger}_{jk}c^{\dagger}_{j},\quad b^{\dagger}_{k}=\sum_{j=1}^{m}X^{\dagger}_{jk}d^{\dagger}_{j}, leads to \displaystyle N_{a}=\sum_{j=1}^{m}c_{j}^{\dagger}c_{j}=N_{c} , \displaystyle N_{b}=\sum_{j=1}^{m}d_{j}^{\dagger}d_{j}=N_{d} , and \displaystyle H \displaystyle=U(N_{c}-N_{d})^{2}+\sum_{j=1}^{m}\varepsilon_{j}(c_{j}^{\dagger}d_{j}+d_{j}^{\dagger}c_{j}). (7) Note that \displaystyle\hat{N}_{j}=c^{\dagger}_{j}c_{j}+d^{\dagger}_{j}d_{j} are conserved operators; let N_{j} denote their eigenvalues. Then \displaystyle\sum_{j=1}^{m}N_{j}=N gives the total number of particles. The effect of the transformation is that takes the square lattice model and maps it to the spectrally equivalent model ( 7 ). The transformed model is defined on a lattice that is the union of disconnected adjacency graphs, each with only two vertices and one edge. These adjacency graphs are weighted by \epsilon_{j} , the coupling for the tunneling strength between the two connected vertices. However observe that the form of the interaction term with coupling U is invariant under the transformation. A graphical representation the transformed Hamiltonian is provided in panel (d) of Figs. 1 , 2 , and 3 . Expressions for the energy eigenvalues of ( 7 ) are known from [ 35 ] , with the form \displaystyle E=UN^{2}+4U\sum_{j=1}^{m}\sum_{n=1}^{N}\frac{N_{j}\varepsilon_{j}^{2}}{v_{n}-\varepsilon_{j}^{2}} (8) subject to the parameters \{v_{n}:n=1,\dots,N\} satisfying the Bethe Ansatz equations \displaystyle\sum_{m\neq n}^{N}\frac{2v_{n}}{v_{n}-v_{m}}+\frac{\displaystyle\prod_{j=1}^{m}(v_{n}-\varepsilon_{j}^{2})^{N_{j}}}{\displaystyle 16U^{2}\prod_{m\neq n}^{N}(v_{n}-v_{m})} \displaystyle=N-1+\sum_{j=1}^{m}\frac{N_{j}\varepsilon_{j}^{2}}{v_{n}-\varepsilon^{2}_{j}} (9) for n=1,\dots,N . Note that both

Read the complete text