Classification of integrable nearest-neighbour spin chains with local site dimension $d>2$ and the graded/non-graded correspondence

de Leeuw, Pribytok and Ryan (2019) classified all fundamental integrable spin chains with two-dimensional local Hilbert space whose integrable structure is a regular $R$-matrix of difference form, finding 14 equivalence classes — 8 known six- or eight-vertex-type models and 6 apparently new ones, several non-diagonalisable or nilpotent — and, for even (XYZ-type) $R$-matrices, a bijection between graded and non-graded regular solutions. Their conclusions pose as the natural extension the same classification for local dimension $d>2$, where the generic $d=3$ Hamiltonian density already carries $9^2=81$ free parameters and the Cayley–Hamilton ansatz for the $R$-matrix requires all powers of the density up to $\mathcal{H}^8$, together with the question whether the graded/non-graded bijection generalises beyond the two-state case. In the intervening years the two-state programme has been completed — all regular $4\times4$ solutions including non-difference form (2024), and all $4\times4$ solutions including non-regular ones (2024) — but no corresponding exhaustive classification exists for $d\ge3$: only symmetry-restricted or vertex-ansatz classes are known. This problem asks for the complete classification of regular difference-form $R$-matrices on $\mathbb{C}^d$ for $d\ge3$ (with $d=3$ as the first mandatory case), with explicit Hamiltonians, $R$-matrices and a proof of exhaustiveness, together with the graded (supersymmetric) classification on $\mathbb{C}^{m|n}$ with $m+n\ge3$ and a proof or refutation of a graded/non-graded bijection in higher dimension. This core remains open: the classification programme's own 2024 completion paper still lists the higher-dimensional extension as an open avenue.

Background

A quantum spin chain is called integrable when it possesses an infinite tower of mutually commuting conserved operators $\mathbb{Q}_2,\mathbb{Q}_3,\mathbb{Q}_4,\ldots$, where $\mathbb{Q}_r$ acts on $r$ consecutive sites; the Hamiltonian is $\mathbb{Q}_2$, and when it has the form $H=\sum_n \mathcal{H}_{n,n+1}$ the interactions are called nearest-neighbour. The standard mechanism producing such towers is the quantum inverse scattering method: one starts from an $R$-matrix, an invertible operator $R_{ab}(u,v)\in\mathrm{End}(V_a\otimes V_b)$ obeying the Yang–Baxter equation (YBE) $R_{ab}(u,v)R_{ac}(u,w)R_{bc}(v,w)=R_{bc}(v,w)R_{ac}(u,w)R_{ab}(u,v)$, and builds commuting transfer matrices from it. For nearest-neighbour chains the $R$-matrix is regular, meaning $R_{ab}(u,u)=P_{ab}$ with $P_{ab}$ the permutation operator, so that the Hamiltonian density is the logarithmic derivative of the transfer matrix at the regular point; the $R$-matrix is of difference form when $R_{ab}(u,v)=R_{ab}(u-v)$. A fundamental chain has the same local space $V$ at every site, with $R$ acting on $V\otimes V$. A key structural fact is that for chains with a regular difference-form $R$-matrix the whole tower of charges is generated from the Hamiltonian alone by the boost operator $\mathcal{B}[\mathbb{Q}_2]$, the generator of lattice translations built from the charge density; this reduces the classification of such chains to a problem about the Hamiltonian.

de Leeuw, Pribytok and Ryan (2019) used this to classify all fundamental integrable spin chains with local space $V=\mathbb{C}^2$ and regular difference-form $R$-matrix. They made a general ansatz for the $4\times4$ Hamiltonian density, quotiented by integrability-preserving transformations (local basis transformations, the discrete maps $R\mapsto PRP$, $R\mapsto R^T$, $R\mapsto PR^TP$, affine reparametrisations of the spectral parameter, and the gauge $R\mapsto f(u)R$), imposed the necessary conditions that the boost-generated charges commute, $[\mathbb{Q}_r,\mathbb{Q}_s]=0$ for $r$ up to 6 (the lowest of these, $[\mathbb{Q}_2,\mathbb{Q}_3]=0$, is known in the literature as the Reshetikhin condition), and then constructed the $R$-matrix perturbatively from the YBE, using the Cayley–Hamilton theorem (any $4\times4$ matrix satisfies its own characteristic polynomial, so all higher powers reduce to $\mathbb{1},H,H^2,H^3$) to write the ansatz $R=P(f_0(u)\mathbb{1}+u f_1(u)H+u^2f_2(u)H^2+u^3f_3(u)H^3)$. They found 14 equivalence classes: 8 well-known models (six- and eight-vertex families including the XXX, XXZ and XYZ chains) and 6 apparently new ones, several with exotic algebraic properties — non-diagonalisable (having non-trivial Jordan block structure, i.e. no complete eigenbasis) or nilpotent. For even (XYZ-type) $R$-matrices they furthermore established a bijection between regular solutions of the ordinary YBE and of the graded Yang–Baxter equation on the supersymmetric local space $\mathbb{C}^{1|1}$, in which odd tensor factors pick up signs under permutation; this also classified the regular two-state supersymmetric chains.

The same paper's conclusions pose the natural extension: the classification of (supersymmetric) spin chains with local site dimension $d>2$. The authors warn this is substantially harder: the generic $d=3$ Hamiltonian density is a $9\times9$ matrix with up to $9^2=81$ free parameters before quotienting by integrability-preserving transformations, the Cayley–Hamilton ansatz would require all powers $\mathcal{H}^j$ for $j=0,\ldots,8$, and it is not guaranteed that a general solution admits such an ansatz at all. Subsequent work completed the two-state programme — all regular $4\times4$ solutions, including genuinely non-difference-form ones (Corcoran and de Leeuw, 2024), and then all $4\times4$ solutions including non-regular ones, which turn out to satisfy modified Yang–Baxter equations (de Leeuw and Posch, 2024) — but for $d\ge3$ only restricted classes are classified: symmetry-restricted families such as all $16\times16$ $\mathfrak{su}(2)\oplus\mathfrak{su}(2)$-symmetric solutions, vertex-type ansätze (fifteen-vertex and $(n+1)(2n+1)$-vertex models), SUSY-algebra Baxterizations of constant solutions, and constrained or anyonic Hilbert spaces. Whether an exhaustive list of regular difference-form solutions exists on $\mathbb{C}^d$ for $d\ge3$, and whether the graded/non-graded bijection survives beyond the two-state case, remains open.

Problem Statement

Classify all fundamental integrable spin chains with nearest-neighbour interactions and local Hilbert space $V=\mathbb{C}^d$ with $d\ge3$ whose integrable structure is a regular $R$-matrix of difference form: enumerate all invertible solutions $R_{ab}(u)$ of the Yang–Baxter equation on $\mathbb{C}^d\otimes\mathbb{C}^d$ of difference form ($R_{ab}(u,v)=R_{ab}(u-v)$), analytic in a neighbourhood of the regular point and satisfying $R_{ab}(0)=P_{ab}$ (regularity; $P_{ab}$ the permutation operator on $\mathbb{C}^d\otimes\mathbb{C}^d$), generalising the $d=2$ classification of de Leeuw, Pribytok and Ryan. A complete answer must deliver:

1. For $d=3$, the first open case, where the generic Hamiltonian density is a $9\times9$ matrix carrying up to $9^2=81$ free parameters before quotienting by integrability-preserving transformations: the complete list of equivalence classes of solutions — the equivalence relation being local basis transformations, the discrete maps $R\mapsto PRP$, $R\mapsto R^T$, $R\mapsto PR^TP$, affine reparametrisations of the spectral parameter, and the gauge $R\mapsto f(u)R$ — with an explicit representative Hamiltonian and $R$-matrix for each class and a proof of exhaustiveness (for example, the complete solution of the commuting-charge constraints $[\mathbb{Q}_r,\mathbb{Q}_s]=0$ for the boost-generated charges — the conserved charges $\mathbb{Q}_r$ obtained from the Hamiltonian by the boost operator, the generator of lattice translations — together with a constructive lift to genuine Yang–Baxter solutions).

2. For each class, a record of whether it is a known model in disguise (for example a higher-rank or symmetry-restricted version of a known vertex model) and of its algebraic properties — diagonalisability, nilpotency, Jordan structure — as the $d=2$ classification recorded for its classes.

3. For general $d>3$: either the complete classification, or a rigorous structural characterisation of the solution set — an explicit parametrisation of all families, or a proof that the equivalence classes form an infinite or wild set admitting no finite enumeration — either of which resolves this item.

4. The extension to graded (supersymmetric) local spaces $\mathbb{C}^{m|n}$ with $m+n\ge3$: the classification of regular difference-form graded solutions, and a proof or refutation of the statement that the bijection between graded and non-graded regular solutions found in the two-state XYZ-type case generalises to higher-dimensional graded spaces.

The difference-form setting is the source's own; additionally treating regular non-difference-form solutions at $d\ge3$, as the completed $d=2$ programme did, is natural but not required for completion.

References

1. Marius de Leeuw, Anton Pribytok, Paul Ryan (2019). Classifying integrable spin-1/2 chains with nearest neighbour interactions. J. Phys. A: Math. Theor. 52, 505201. DOI 10.1088/1751-8121/ab529f. https://arxiv.org/abs/1904.12005
2. R. S. Vieira (2019). Fifteen-vertex models with non-symmetric R matrices. arXiv:1908.06932. https://arxiv.org/abs/1908.06932
3. Marius de Leeuw, Chiara Paletta, Anton Pribytok, Ana L. Retore, Paul Ryan (2020). Classifying nearest-neighbour interactions and deformations of AdS. Phys. Rev. Lett. 125, 031604. DOI 10.1103/PhysRevLett.125.031604. https://arxiv.org/abs/2003.04332
4. R. S. Vieira, A. Lima-Santos (2021). Solutions of the Yang–Baxter equation for (n+1)(2n+1)-vertex models using a differential approach. J. Stat. Mech. (2021) 053103. DOI 10.1088/1742-5468/abf7be. https://arxiv.org/abs/2012.02543
5. Marius de Leeuw, Chiara Paletta, Anton Pribytok, Ana L. Retore, Paul Ryan (2021). Yang-Baxter and the Boost: splitting the difference. SciPost Phys. 11, 069. DOI 10.21468/SciPostPhys.11.3.069. https://arxiv.org/abs/2010.11231
6. Juan Miguel Nieto García, Leander Wyss (2022). Jordan blocks and the Bethe ansatz I: The eclectic spin chain as a limit. Nucl. Phys. B 981, 115860. DOI 10.1016/j.nuclphysb.2022.115860. https://arxiv.org/abs/2112.13883
7. Marius de Leeuw, Chiara Paletta (2022). The Bethe ansatz for a new integrable open quantum system. arXiv:2207.14193. https://arxiv.org/abs/2207.14193
8. Marius de Leeuw, Rafael I. Nepomechie, Ana L. Retore (2023). Flag Integrable Models and Generalized Graded Algebras. J. High Energy Phys. 2023, 113. DOI 10.1007/JHEP06(2023)113. https://arxiv.org/abs/2210.06495
9. Juan Miguel Nieto García (2024). Jordan blocks and the Bethe ansatz III: Class 5 model and its symmetries. Nucl. Phys. B 998, 116419. DOI 10.1016/j.nuclphysb.2023.116419. https://arxiv.org/abs/2309.10044
10. Chiara Paletta (2023). Yang-Baxter integrable open quantum systems. arXiv:2312.00064. https://arxiv.org/abs/2312.00064
11. Luke Corcoran, Marius de Leeuw (2024). All regular 4 x 4 solutions of the Yang-Baxter equation. SciPost Phys. Core 7, 045. DOI 10.21468/SciPostPhysCore.7.3.045. https://arxiv.org/abs/2306.10423
12. Marius de Leeuw, Vera Posch (2024). All 4 x 4 solutions of the quantum Yang-Baxter equation. arXiv:2411.18685. https://arxiv.org/abs/2411.18685
13. Luke Corcoran, Marius de Leeuw (2025). Integrable and critical Haagerup spin chains. Phys. Rev. B 111, L140408. DOI 10.1103/PhysRevB.111.L140408. https://arxiv.org/abs/2410.16356
14. Luke Corcoran, Marius de Leeuw, Balázs Pozsgay (2025). Integrable models on Rydberg atom chains. SciPost Phys. 18, 139. DOI 10.21468/SciPostPhys.18.4.139. https://arxiv.org/abs/2405.15848
15. Somnath Maity, Pramod Padmanabhan, Vladimir Korepin (2025). Non-hermitian integrable systems from constant non-invertible solutions of the Yang-Baxter equation. J. High Energy Phys. 2025, 206. DOI 10.1007/JHEP05(2025)206. https://arxiv.org/abs/2503.08109
16. Matthew Blakeney, Luke Corcoran, Marius de Leeuw (2026). Constrained integrability and anyonic chains. Int. J. Mod. Phys. A. DOI 10.1142/S0217751X26300140. https://arxiv.org/abs/2605.28946