Yang-Baxter algebra of the n-state mixed vertex model with the line-conservation constraint on the square lattice

M. J. Martins (2024) established two correspondences between arbitrary n-state spin models and n-state vertex models on the toroidal square lattice under which the partition functions coincide exactly at finite lattice size; the equivalent vertex model has only $n^3$ non-null Boltzmann weights, subject to the line-conservation constraint $w(i_1,i_2|i_3,i_4)=0$ for $i_1\neq i_4$ . For $n=2$ this recovers the Ising model in a magnetic field as a 'mixed' eight-vertex model whose weights combine even and odd arrow configurations. The Yang-Baxter algebra of this mixed model class - the object that would embed it in the quantum inverse scattering framework and produce commuting transfer matrices - has been analyzed only in two special cases: the arrow-reversal-symmetric $n=2$ submanifold and the 27-vertex model equivalent to the three-state Fateev-Zamolodchikov spin chain. Subsequent literature gives a direct derivation of the arbitrary- $(n,L)$ transfer-matrix identity underlying the first correspondence (Martins 2025) and extends the embedding to the chiral Potts model (Zhang 2026), but the Yang-Baxter algebra analysis of the general n-state mixed vertex model - in particular the second correspondence's weight map and the non-symmetric $n=2$ case - remains untouched. The problem posed is to determine the invertible R-matrices solving the Yang-Baxter algebra for this Lax-operator class for general $n$ and unrestricted weights, to identify the integrable weight manifolds, and to delimit which parts of the class admit no nontrivial solution.

Background

Two dual families of lattice models are studied on the square lattice. In a spin model, a variable $\sigma_{i,j}$ taking $n$ values sits on each site and Boltzmann weights $W_h(\sigma,\sigma')$ and $W_v(\sigma,\sigma')$ are attached to horizontal and vertical nearest-neighbor edges; the Ising model ( $n=2$ ) is the prototype. In a vertex model, the variables sit on the edges (links) of the lattice and a weight $w(i_1,i_2|i_3,i_4)$ is assigned to each configuration of the four links meeting at a vertex, where $i_1,i_2$ are the horizontal (left, right) and $i_3,i_4$ the vertical (top, bottom) edge states. With two-state edges there are sixteen possible vertex configurations. The even eight-vertex model, in which only configurations with an even number of arrows pointing into the vertex are retained, is the celebrated exactly solvable case solved by Baxter through the Yang-Baxter equation; Fan and Wu, and independently Kadanoff and Wegner, mapped the zero-field Ising model to it (a two-to-one correspondence), while Suzuki and Fisher showed that the Ising model in a magnetic field can be reformulated as a sixteen-vertex model. That reformulation has not made the field-driven Ising model solvable: its free energy remains one of the famous unknowns of two-dimensional statistical mechanics.

The partition function on an $L\times L$ torus is computed with transfer matrices. The row-to-row transfer matrix adds one full row of the lattice at a time, while the diagonal-to-diagonal transfer matrix, natural for spin models built from integrable edge weights, advances along lattice diagonals. Exact solvability of a vertex model means a family of commuting transfer matrices, which follows from the Yang-Baxter algebra

$R_{12}\,\mathbb{L}_{13}\,\mathbb{L}_{23}=\mathbb{L}_{23}\,\mathbb{L}_{13}\,R_{12},$

where the Lax operator $\mathbb{L}$ is the $n^2\times n^2$ matrix encoding the vertex weights as an operator on the tensor product of an auxiliary (horizontal) and a quantum (vertical) $n$ -dimensional space (the subscripts indicate on which of the three tensor factors the operators act), and $R$ is an intertwiner that must be invertible - in practice through the unitarity relation $R_{12}R_{21}=\rho\,\mathrm{I}$ with $\rho$ a non-vanishing scalar, where $R_{21}$ denotes the R-matrix with its two tensor factors and its two spectral arguments interchanged. When such an $R$ exists, the transfer matrices at different spectral parameters (also called rapidities) commute, which is the gateway to exact solutions via the quantum inverse scattering method - the algebraic framework that extracts exact spectra and conserved quantities of a one-dimensional quantum spin chain from the Yang-Baxter algebra. For spin (edge) models the analogous role is played by local star-triangle relations among the edge weights - functional equations on the weights around an elementary triangle that are sufficient for transfer matrices at different spectral parameters to commute; solvable edge models include the scalar n-state Potts (the Ising model with each spin taking $n$ instead of two values), the Ashkin-Teller model (a two-copy Ising generalization), the Fateev-Zamolodchikov (FZ) model (the $Z_n$ -symmetric chain whose trigonometric weights depend on the difference of two spectral parameters), and the Kashiwara-Miwa models. The chiral Potts model - the $Z_N$ -symmetric clock generalization of the Ising model - is the notable case whose weights live on an algebraic curve of genus larger than one for $N>2$ (the genus counting the handles of the curve), so that no single rapidity-difference parametrization exists.

Martins (2024) established, at the level of extensive finite-size verification together with a general Hamiltonian-limit argument (the limit in which the logarithmic derivative of the transfer matrix at a special spectral point yields the associated quantum spin chain), that any n-state spin model on the toroidal square lattice has an equivalent n-state vertex model with exactly $n^3$ non-null weights obeying the line-conservation constraint $w(i_1,i_2|i_3,i_4)=0$ for $i_1\neq i_4$ (each horizontal line label passes through every vertex unchanged). Two correspondences were given, differing in how the spin edge weights combine: the first, $w=W_h(i_3,i_1)W_v(i_3,i_2)\delta_{i_1,i_4}$ , under which the vertex row-to-row transfer matrix equals the spin diagonal-to-diagonal one up to a ninety-degree lattice rotation; and the second, $w=W_v(i_3,i_1)W_h(i_1,i_2)\delta_{i_1,i_4}$ , under which the vertex transfer matrix equals the spin row-to-row one. In both cases the partition functions coincide exactly on the finite torus, improving on earlier Ising-vertex mappings that were two-to-one or used twice as many lattice points. For $n=2$ the equivalent model is a mixed eight-vertex model whose eight weights comprise four even and four odd arrow configurations; the anisotropic Ising model in a magnetic field is a particular point of its weight space. Within this class the Yang-Baxter algebra has been solved only in two cases: the arrow-reversal-symmetric $n=2$ submanifold (vertex weights invariant under simultaneously reversing all arrows around a vertex), where the Lax operator lies on the same elliptic curve as Baxter's even eight-vertex model but the R-matrix cannot be written in terms of the difference of two rapidities and lies on a quartic surface argued to be of K3 type (a distinguished class of algebraic surfaces); and, via the first correspondence, the 27-vertex model equivalent to the integrable three-state FZ spin chain. Martins (2025) then derived the general- $(n,L)$ transfer-matrix identity of the first correspondence directly, and constructed the R-matrix of the equivalent vertex model for every spin model whose edge weights satisfy star-triangle relations with the rapidity-difference property, and Zhang (2026) extended the embedding to the chiral Potts model. What no work has analyzed is the Yang-Baxter algebra of the general n-state mixed vertex model itself - beyond the symmetric $n=2$ submanifold, beyond the FZ point, and in particular for the second correspondence's weight map and for non-symmetric weights at $n=2$ .

Problem Statement

Determine the solutions of the Yang-Baxter algebra for the Lax operators of the general n-state mixed vertex model on the square lattice - the vertex model with the line-conservation constraint $w(i_1,i_2|i_3,i_4)=0$ for $i_1\neq i_4$ ( $n^3$ non-null weights) - for arbitrary $n\ge 2$ and unrestricted, not necessarily arrow-reversal-symmetric, weights. In the clean conventions of the 2025 follow-up paper the two Lax operators induced by the two spin-vertex correspondences are

$\mathbb{L}^{(1)}_{12}=\sum_{i,j,k=1}^{n}W_h(j,i)\,W_v(j,k)\,e_{ik}\otimes e_{ji}$

for the first correspondence and

$\mathbb{L}^{(2)}_{12}=\sum_{i,j,k=1}^{n}W_v(j,i)\,W_h(i,k)\,e_{ik}\otimes e_{ji}$

for the second one, where $e_{ik}$ denotes the elementary $n\times n$ matrix with a single non-vanishing entry, equal to one, in row $i$ and column $k$ ; in both formulas the first tensor factor acts on the horizontal (auxiliary) pair of edge labels $(i_1,i_2)=(i,k)$ and the second on the vertical (quantum) pair $(i_3,i_4)=(j,i)$ , so that the line-conservation constraint is already built in through the repeated index $i$ . For $\mathbb{L}=\mathbb{L}^{(2)}$ (the case on which no analysis exists), and equally for $\mathbb{L}^{(1)}$ away from the already-solved family of star-triangle-satisfying, rapidity-difference edge weights, find all invertible $n^2\times n^2$ matrices $R_{12}$ satisfying

$R_{12}\,\mathbb{L}_{13}\,\mathbb{L}_{23}=\mathbb{L}_{23}\,\mathbb{L}_{13}\,R_{12},$

where the two Lax operators $\mathbb{L}_{13}$ and $\mathbb{L}_{23}$ are evaluated at two independent copies $\mathbf{w},\mathbf{w}'$ of the vertex weights (two spectral parameter values), the R-matrix $R_{12}=R_{12}(\mathbf{w},\mathbf{w}')$ may depend on both copies, $R_{21}=R_{21}(\mathbf{w}',\mathbf{w})$ is the R-matrix with the two tensor factors and the two weight arguments interchanged; the unitarity relation $R_{12}R_{21}=\rho\,\mathrm{I}_{n^2}$ , with $\rho$ a non-vanishing scalar, is one standard way to enforce the required invertibility; identify the algebraic manifolds of vertex weights on which such solutions exist, and delimit the parts of the weight space that admit no nontrivial solution (a nontrivial solution being one whose R-matrix is not proportional to the identity or the permutation operator). A complete answer gives the explicit R-matrices as functions of the weights; identifying the weight manifolds on which solutions exist and the region admitting none, without the explicit R-matrices everywhere, is an independently acceptable sub-result toward that complete answer. The common objective uniting these deliverables is the Yang-Baxter algebra analysis of the mixed vertex model class, exactly as proposed in the outlook of the source paper. A physically important sub-question, not a restriction of scope, is whether the closure of any integrable manifold of the $n=2$ mixed model that admits a nontrivial solution contains the point equivalent to the Ising model in a magnetic field.

References

M. J. Martins, 'On the equivalence between n-state spin and vertex models on the square lattice', Nuclear Physics B 1005 (2024) 116610; arXiv:2405.07341; doi:10.1016/j.nuclphysb.2024.116610; https://arxiv.org/abs/2405.07341
M. J. Martins, 'Embedding integrable spin models in solvable vertex models on the square lattice', Nuclear Physics B 1013 (2025) 116849; arXiv:2501.02377; doi:10.1016/j.nuclphysb.2025.116849; https://arxiv.org/abs/2501.02377
Zhao Zhang, 'The Yang-Baxter Equation for the Chiral Potts Model and Integrable Parafermions', arXiv:2606.11146 (2026); https://arxiv.org/abs/2606.11146
A. Sinha, T. Justin, P. Padmanabhan, V. Korepin, 'The Yang-Baxter integrability of the critical Ising chain', Journal of Statistical Mechanics: Theory and Experiment 2025 (2025) 103102; arXiv:2506.03668; doi:10.1088/1742-5468/ae0690; https://arxiv.org/abs/2506.03668
M. J. Martins, 'Bethe equations for the critical three-state Potts spin chain with toroidal boundary conditions', Nuclear Physics B 1024 (2026) 117323; arXiv:2512.07763; doi:10.1016/j.nuclphysb.2026.117323; https://arxiv.org/abs/2512.07763
C. Fan and F. Y. Wu, 'General Lattice Model of Phase Transitions', Physical Review B 2 (1970) 723-733; doi:10.1103/PhysRevB.2.723; https://doi.org/10.1103/PhysRevB.2.723
L. P. Kadanoff and F. J. Wegner, 'Some Critical Properties of the Eight-Vertex Model', Physical Review B 4 (1971) 3989-3993; doi:10.1103/PhysRevB.4.3989; https://doi.org/10.1103/PhysRevB.4.3989
R. J. Baxter, 'Partition function of the Eight-Vertex lattice model', Annals of Physics 70 (1972) 193-228; doi:10.1016/0003-4916(72)90335-1; https://doi.org/10.1016/0003-4916(72)90335-1
M. Suzuki and M. E. Fisher, 'Zeros of the Partition Function for the Heisenberg, Ferroelectric, and General Ising Models', Journal of Mathematical Physics 12 (1971) 235-246; doi:10.1063/1.1665583; https://doi.org/10.1063/1.1665583
F. Y. Wu and H. Kunz, 'The Odd Eight-Vertex Model', Journal of Statistical Physics 116 (2004) 67-78; doi:10.1023/B:JOSS.0000037206.47155.58; https://doi.org/10.1023/B:JOSS.0000037206.47155.58