Auxiliary-weight elimination and the general Yang–Baxter equation in the spin-to-vertex embedding of integrable lattice models

A 2025 paper by M. J. Martins (Nucl. Phys. B 1013, 116849) showed that an integrable $n$-state edge (spin) model on the square lattice whose Boltzmann weights satisfy the star-triangle and inversion relations can be embedded in an equivalent solvable vertex model: the Lax operator and R-matrix are built explicitly from the spin edge weights, the Yang-Baxter algebra follows from the star-triangle relations, and unitarity follows from the vertical inversion relation. Two gaps in this construction define the present problem. First, the extension to arbitrary pairs of spin-weight sets — the route to R-matrices without the rapidity-difference property — passes through doubled star-triangle relations, a system of $2n^3$ polynomial relations from which auxiliary weights and a scalar must be eliminated; the existence and explicit parametrization of such eliminations is unknown. Second, the full Yang-Baxter equation for the R-matrix built from rapidity-difference weights has been verified only by direct substitution for four reflection-symmetric models, and a general proof from the star-triangle and inversion relations alone, without symmetry assumptions, is missing. A 2026 preprint by Z. Zhang constructs a three-spectral-parameter R-operator for the chiral Potts model, treating the most prominent special case by an adapted route, but the general elimination problem and the general Yang-Baxter theorem remain open, as does the spectral theory of the resulting non-difference-form R-matrices.

Background

Two-dimensional lattice models of statistical mechanics are conventionally divided by the location of their local degrees of freedom. In edge (spin) models, spins taking values in $\{1,\dots,n\}$ sit on the vertices of the square lattice and each bond carries a Boltzmann weight — a horizontal weight $W_h(i,j|x)$ or a vertical weight $W_v(i,j|x)$ for an edge joining neighboring spins $i,j$, depending on a spectral (rapidity) parameter $x$. In vertex models the state variables sit on the lattice edges and each vertex carries a weight depending on the four edge states meeting there. Both classes are solved through commuting families of transfer matrices, but the solvability mechanism differs. For edge models the key is the star-triangle relation (STR), a local identity on a triangular cell of three weights, $\sum_{d=1}^n W_v(d,c|y)\,W_v(b,d|x-y)\,W_h(a,d|x)=\mathcal{R}(x,y)\,W_h(a,b|y)\,W_h(a,c|x-y)\,W_v(b,c|x),$ together with a companion relation with permuted spin indices, where $\mathcal{R}(x,y)$ is independent of the spins $a,b,c$; it guarantees that the diagonal-to-diagonal transfer matrices of the model commute. For vertex models the key is an invertible R-matrix satisfying the Yang-Baxter algebra (YBA) $R_{12}(x,y)\,\mathbb{L}_{13}(x)\,\mathbb{L}_{23}(y)=\mathbb{L}_{23}(y)\,\mathbb{L}_{13}(x)\,R_{12}(x,y),$ where $\mathbb{L}$ is the Lax operator encoding the vertex weights and the subscripts indicate on which factors of a triple tensor product the operators act. This is the setting of the quantum inverse scattering method (QISM), the operator framework underlying the algebraic Bethe ansatz. The associativity (consistency) of the YBA is encoded in the Yang-Baxter equation (YBE) $R_{12}(x,y)R_{13}(x,z)R_{23}(y,z)=R_{23}(y,z)R_{13}(x,z)R_{12}(x,y)$.

Martins bridged the two frameworks. Building on the earlier proof that an $n$-state spin model and a specific $n$-state vertex model have identical partition functions on finite tori, he expressed the Lax operator of the equivalent vertex model directly in terms of the spin weights, $\mathbb{L}_{12}(x)=\sum_{i,j,k=1}^{n}W_h(j,i|x)\,W_v(j,k|x)\,e_{ik}\otimes e_{ji}$, where $e_{pq}$ is the $n\times n$ matrix unit with a single entry 1 in row $p$, column $q$, and proved that, when the weights depend on the difference of spectral parameters (the rapidity-difference property), the R-matrix $\mathrm{R}_{12}(x,y)=\sum_{i,j,k=1}^{n}\frac{W_h(j,i|x)\,W_v(j,k|x-y)}{W_h(k,i|y)}\,e_{ik}\otimes e_{ji}$ satisfies the YBA as a consequence of the STRs. If the vertical weights also obey the inversion relation $\sum_{k=1}^n W_v(i,k|x)\,W_v(k,j|-x)=\rho_2(x)\,\delta_{i,j}$, the R-matrix is unitary, $\mathrm{R}_{12}(x,y)\mathrm{R}_{21}(y,x)=\rho_2(x-y)\,I_n\otimes I_n$, hence invertible away from the zeros of the scalar factor $\rho_2$. The exact solution of a star-triangle-solvable spin model thereby becomes accessible to the QISM. The construction was carried out for the scalar $n$-state Potts model, where the R-matrix rewrites in terms of the generators of the Temperley-Lieb algebra (the algebra generated by operators $e_i$ with $e_i^2=\delta\,e_i$ and $e_ie_{i\pm1}e_i=e_i$), and for the integrable Ashkin-Teller model (two coupled Ising models), giving an R-matrix with sixteen distinct theta-function weights. Although the weights have the rapidity-difference property, the R-matrix depends on $x$ and $y$ separately rather than only on $x-y$, so it is not itself of difference form; Martins used this to define generalized vertex transfer matrices with an extra coupling and integrable quantum spin chains with additional interactions.

Two gaps were left explicit. First, the YBE for the R-matrix above was checked only by substituting the explicit weights of four models with rapidity-difference weights — the scalar Potts, the self-dual Ashkin-Teller, the $\mathbb{Z}_n$-symmetric Fateev-Zamolodchikov model with trigonometric weights, and its elliptic-weight generalization, the Kashiwara-Miwa model — all of which have the reflection symmetry $W_{h,v}(i,j|x)=W_{h,v}(j,i|x)$; whether the YBE follows from the STR and inversion relations alone was stated as plausible but not proven. Second, to go beyond rapidity-difference weights, Martins formulated an abstract scheme for two edge-weight sets $\mathbf{w}$ and $\mathbf{w}'$ (assumed in the source to be solvable in such a way that both weight sets lie on the same algebraic variety): find auxiliary weights $W''_{h,v}$ and a scalar $\mathcal{R}$ satisfying doubled star-triangle relations, and eliminate them to obtain explicit R-matrices. The solvability of this elimination problem was left open, with the chiral Potts model — an integrable $\mathbb{Z}_N$-symmetric edge model whose Boltzmann weights depend on two points of an algebraic curve of genus greater than one and therefore admit no global single rapidity parameter — named as the most prominent target. The subsequent literature sharpened the landscape: Zhang constructed a three-spectral-parameter R-operator for the chiral Potts model directly from its STR and connected it to the Bazhanov-Stroganov four-parameter R-matrix (an intertwiner of cyclic representations of $U_q(\mathfrak{sl}_2)$ at a root of unity), while de Leeuw and Posch completed the classification of $4\times4$ YBE solutions and showed that an R-matrix obtained from fundamental commutation relations can fail the YBE in non-regular cases — so the general implication at issue is not automatic.

Problem Statement

The problem is to complete the general theory of the edge-to-vertex embedding: to make it effective (explicit R-matrices obtained by solving the doubled star-triangle elimination problem) and rigorous (a general Yang-Baxter theorem), going beyond the four rapidity-difference, reflection-symmetric models for which the construction has been verified case by case. The two parts below serve this single objective: Part A produces the explicit generalized R-matrices, and Part B establishes the general Yang-Baxter theorem for the R-matrix built from rapidity-difference weights; the analogous theorem for the R-matrices of Part A arises once those are constructed.

Part A (the elimination problem). For two $n$-state edge-weight sets $\mathbf{w}=\{W_{h,v}(i,j)\}$ and $\mathbf{w}'=\{W'_{h,v}(i,j)\}$ — which may depend on spectral parameters or on points of an algebraic curve, the relations below being required to hold identically in those parameters — consider the doubled star-triangle relations of the source paper, $\sum_{d=1}^n W'_v(d,c)\,W''_v(b,d)\,W_h(a,d)=\mathcal{R}\,W'_h(a,b)\,W''_h(a,c)\,W_v(b,c),$ $\sum_{d=1}^n W'_v(c,d)\,W''_v(d,b)\,W_h(d,a)=\mathcal{R}\,W'_h(b,a)\,W''_h(c,a)\,W_v(c,b),$ for all $a,b,c\in\{1,\dots,n\}$ — a system of $2n^3$ polynomial relations in the auxiliary weights $W''_{h,v}(i,j)$ and the spin-independent scalar $\mathcal{R}$ (which may depend on the spectral or curve parameters but not on the spins). The degenerate branch $W''_{h,v}\equiv 0$, $\mathcal{R}=0$ is excluded: admissible solutions have $\mathcal{R}\neq 0$ and yield, via (c), a nonzero R-matrix which, following the source's integrability requirement, is invertible for most values of the spin edge weights. The source formulated the scheme under the assumption that $\mathbf{w}$ and $\mathbf{w}'$ lie on the same algebraic variety; whether that restriction is necessary is itself part of question (a). Determine: (a) for which pairs $(\mathbf{w},\mathbf{w}')$ nondegenerate solutions exist; (b) an explicit parametrization of the solutions in the cases where they exist; (c) the resulting explicit R-matrices $\mathrm{R}_{12}(\mathbf{w},\mathbf{w}')=\sum_{i,j,k=1}^n\frac{W_h(j,i)\,W''_v(j,k|\mathbf{w},\mathbf{w}')}{W'_h(k,i)}\,e_{ik}\otimes e_{ji},$ with the denominators $W'_h(k,i)$ nonvanishing wherever the ratio is used; and (d) for which solutions the auxiliary-weight unitarity constraint $\sum_{k=1}^n W''_v(i,k|\mathbf{w},\mathbf{w}')\,W''_v(k,j|\mathbf{w}',\mathbf{w})=\rho_2(\mathbf{w},\mathbf{w}')\,\delta_{i,j}$ holds, which guarantees invertibility of the R-matrix away from the zeros of $\rho_2$. Features (a)--(d) are not logically independent: an explicit parametrization of all nondegenerate solutions (b) already determines for which pairs such solutions exist (a) and, through the formula above, the resulting R-matrices (c), so it suffices to provide the parametrization (b) together with the unitarity classification (d); the existence criterion (a) and the matrices (c) then follow. The central cases are weights without the rapidity-difference property, in particular models whose rapidities live on higher-genus algebraic curves such as the chiral Potts model, for which an adapted three-spectral-parameter construction is known but the auxiliary-weight elimination itself is not solved.

Part B (the general Yang-Baxter theorem). Prove, or disprove by an explicit model, that the R-matrix of the source paper built from rapidity-difference weights, $\mathrm{R}_{12}(x,y)=\sum_{i,j,k=1}^n\frac{W_h(j,i|x)\,W_v(j,k|x-y)}{W_h(k,i|y)}\,e_{ik}\otimes e_{ji},$ satisfies the Yang-Baxter equation $\mathrm{R}_{12}(x,y)\,\mathrm{R}_{13}(x,z)\,\mathrm{R}_{23}(y,z)=\mathrm{R}_{23}(y,z)\,\mathrm{R}_{13}(x,z)\,\mathrm{R}_{12}(x,y)$ whenever the weights satisfy the star-triangle relations and the inversion relations, without assuming the reflection symmetry $W_{h,v}(i,j|x)=W_{h,v}(j,i|x)$ used in all existing verifications.

References

1. M. J. Martins, "Embedding integrable spin models in solvable vertex models on the square lattice", Nuclear Physics B 1013 (2025) 116849, DOI 10.1016/j.nuclphysb.2025.116849, arXiv:2501.02377, https://arxiv.org/abs/2501.02377
2. Zhao Zhang, "The Yang-Baxter Equation for the Chiral Potts Model and Integrable Parafermions", arXiv:2606.11146 (2026), https://arxiv.org/abs/2606.11146 (preprint submitted to SciPost Physics)
3. Marius de Leeuw and Vera Posch, "All 4×4 solutions of the quantum Yang-Baxter equation", arXiv:2411.18685 (2024), https://arxiv.org/abs/2411.18685
4. M. J. Martins, "Bethe equations for the critical three-state Potts spin chain with toroidal boundary conditions", Nuclear Physics B 1024 (2026) 117323, DOI 10.1016/j.nuclphysb.2026.117323, arXiv:2512.07763, https://arxiv.org/abs/2512.07763
5. Akash Sinha, Tinu Justin, Pramod Padmanabhan, and Vladimir Korepin, "The Yang-Baxter integrability of the critical Ising chain", Journal of Statistical Mechanics: Theory and Experiment 2025 (2025) 103102, DOI 10.1088/1742-5468/ae0690, arXiv:2506.03668, https://arxiv.org/abs/2506.03668
6. M. J. Martins, "On the equivalence between n-state spin and vertex models on the square lattice", Nuclear Physics B 1005 (2024) 116610, DOI 10.1016/j.nuclphysb.2024.116610, arXiv:2405.07341, https://arxiv.org/abs/2405.07341
7. Yuan Miao and Eric Vernier, "Integrable Quantum Circuits from the Star-Triangle Relation", Quantum 7 (2023) 1160, DOI 10.22331/q-2023-11-03-1160, arXiv:2302.12675, https://arxiv.org/abs/2302.12675
8. V. V. Bazhanov and Yu. G. Stroganov, "Chiral Potts model as a descendant of the six-vertex model", Journal of Statistical Physics 59 (1990) 799-817, DOI 10.1007/BF01025851, https://doi.org/10.1007/BF01025851
9. H. Au-Yang, B. M. McCoy, J. H. H. Perk, S. Tang, and M.-L. Yan, "Commuting transfer matrices in the chiral Potts models: Solutions of star-triangle equations with genus > 1", Physics Letters A 123 (1987) 219-223, DOI 10.1016/0375-9601(87)90065-X, https://doi.org/10.1016/0375-9601(87)90065-X
10. R. J. Baxter, J. H. H. Perk, and H. Au-Yang, "New solutions of the star-triangle relations for the chiral Potts model", Physics Letters A 128 (1988) 138-142, DOI 10.1016/0375-9601(88)90896-1, https://doi.org/10.1016/0375-9601(88)90896-1