The nearest-neighbour Ising model on the cubic lattice has a vertex-model representation whose weight matrix satisfies a twisted tetrahedron equation — a variant of the Zamolodchikov tetrahedron equation in which one side is conjugated by a basis-change operator — constructed by Talalaev in 2018. Unlike the original tetrahedron equation, this twisted relation is not known to imply any integrability structure: no commuting family of transfer matrices, no conserved quantities, and no exact partition function follow from it, and the twist breaks the standard commutativity proof. The problem is to determine whether the twisted tetrahedron equation can nevertheless be used to derive a complete set of conserved charges for the 3D Ising model — concretely, a non-degenerate commuting family of layer-to-layer transfer matrices built from the weight matrix — sufficient to compute the model's thermodynamics exactly in the thermodynamic limit, or to establish that it does not determine such a structure. To date, no commuting family has been derived from the twisted equation and no obstruction is known.

Background

Statistical mechanics distinguishes a handful of exactly solved lattice models whose partition functions are known in closed form. The Ising model — spins σi=±1\sigma_i=\pm1 on lattice sites with Hamiltonian H(σ)=∑σiσjH(\sigma)=\sum\sigma_i\sigma_j summed over nearest-neighbour pairs, and partition function Z(t)=∑σexp⁡(tH(σ))Z(t)=\sum_\sigma\exp(tH(\sigma)) at coupling tt — is the archetype. In two dimensions it was solved by Onsager (1944), and the modern understanding is that such solvability rests on integrability: the 2D partition function is the trace of a transfer matrix (the operator propagating one row of the lattice to the next), and solvability follows when the transfer matrices belong to a commuting family — an infinite family of mutually commuting operators that can be diagonalized together and act as conserved charges. For 2D vertex models — lattice models whose degrees of freedom live on the edges, with local Boltzmann weights attached to each vertex as a function of the surrounding edge states — such a family is guaranteed when the local Boltzmann weights satisfy the Yang–Baxter equation, a cubic identity for a matrix RR acting on the tensor square of an auxiliary vector space. No comparable structure has ever been found for the nearest-neighbour Ising model on the cubic lattice: the 3D model is the canonical unsolved problem of the field, and its critical data are known only numerically (series expansions, Monte Carlo, and the conformal bootstrap — a numerical programme determining critical exponents from consistency conditions on correlation functions, as in El-Showk et al. 2012).

The natural three-dimensional analogue of the Yang–Baxter equation is the tetrahedron equation, introduced by Zamolodchikov (1980) in the theory of factorized scattering in three dimensions. It is an identity for a matrix acting on the tensor cube of an auxiliary space; when the Boltzmann weights of a 3D vertex model satisfy it, the layer-to-layer transfer matrices — the 3D analogues, propagating one two-dimensional layer of the lattice to the next — commute and the machinery of exactly solved models carries over (Maillet and Nijhoff 1989; Talalaev 2017). Known solutions define special exactly solvable 3D models — Zamolodchikov's original model and its descendants, including the quantum-dilogarithm models of Bazhanov–Kashaev–Mangazeev–Sergeev (2025) and the quantum-cluster-algebra solutions of Inoue–Kuniba–Sun–Terashima–Yagi (2024) — but none of them is the 3D Ising model.

Talalaev (2018, published in Journal of Geometry and Physics 2020) approached this gap from the Ising side. Introducing spin variables on the edges of the cubic lattice Λ\Lambda and passing to the dual lattice Λ∗\Lambda^* (whose vertices are the 3-cubes of Λ\Lambda), he attaches to each 2-face the 16-dimensional space Vf=(C2)⊗4V_f=(\mathbb{C}^2)^{\otimes4} and defines a weight matrix WW — a linear operator on Vf⊗3V_f^{\otimes3} whose nonzero entries are Ising Boltzmann weights exp⁡(t(σ1+σ2+σ3))\exp(t(\sigma_1+\sigma_2+\sigma_3)) on the admissible edge-spin configurations of a 3-cube. He proves that the 3D Ising partition function equals the product of these weight matrices over the dual lattice, so that the 3D Ising model is a vertex model in this formulation, and that WW satisfies a twisted tetrahedron equation: a product of four WW's conjugated by a fixed basis-change operator AA (acting in each tensor component and exchanging incoming and outgoing edge labels within each 2-face) equals a product of four unconjugated WW's, the two sides being related by the simultaneous index transposition 1↔41\leftrightarrow4, 2↔32\leftrightarrow3. The construction arises from a recursion on the spaces of solutions of the n-simplex equations (higher analogues of the Yang–Baxter equation living on n-simplices) and from the combinatorics of the 4-dimensional hypercube; the admissible configurations form subgraphs of the 4-cube related to Gray codes and (n,k)(n,k)-chain codes, including a (4,2)(4,2)-code — a side remark of the source with potential independent interest in coding theory.

The twist is precisely what breaks the standard integrability argument: proofs that tetrahedron-equation solutions yield commuting transfer matrices require all factors to act in one common basis, whereas here the two sides of the equation act in different bases, so no commuting family is known to follow. Two further facts frame the question. Istrail (2000) proved that computing Ising partition functions on non-planar lattices is NP-complete in general, so any exact solution must exploit the specific geometry of the cubic lattice rather than a universal algorithm. And the general status of 3D Ising integrability is open: the author's own 2021 review (Russian Mathematical Surveys 76:4) describes it as a long-standing question whose status remains uncertain, with the twisted tetrahedron equation the only relation of integrability type exhibited so far for the 3D Ising model's own weights.

Problem Statement

The problem posed by this line of work (Talalaev, J. Geom. Phys. 148 (2020) 103545) is whether the twisted tetrahedron equation satisfied by the 3D Ising weight matrix can be promoted to a full integrability structure for the model. Concretely: the vertex-model weight matrix WW of the isotropic nearest-neighbour Ising model on the cubic lattice, acting on Vf⊗3V_f^{\otimes3} with Vf=(C2)⊗4V_f=(\mathbb{C}^2)^{\otimes4}, satisfies W653A(1,2,3) W642A(1,2,4) W541A(1,3,4) W321A(2,3,4)=W356(2,3,4) W246(1,3,4) W145(1,2,4) W123(1,2,3),W^{A}_{653}(1,2,3)\,W^{A}_{642}(1,2,4)\,W^{A}_{541}(1,3,4)\,W^{A}_{321}(2,3,4)=W_{356}(2,3,4)\,W_{246}(1,3,4)\,W_{145}(1,2,4)\,W_{123}(1,2,3), where the superscript AA denotes conjugation by a fixed basis-change operator acting in each tensor component, and the two sides are related by the index transposition 1↔41\leftrightarrow4, 2↔32\leftrightarrow3 (equivalently, after relabelling, by conjugation with the transpositions P16P25P_{16}P_{25} of the six tensor factors). Determine whether from this equation one can construct a complete set of conserved quantities for the 3D Ising model — specifically, a non-degenerate commuting family of layer-to-layer transfer matrices (or an equivalent commuting algebra of charges) built from WW, containing the transfer matrix whose trace yields the partition function Z(t)=∏(α,β,γ)∈Λ∗WαβγZ(t)=\prod_{(\alpha,\beta,\gamma)\in\Lambda^{*}}W_{\alpha\beta\gamma} but not generated by it alone (the family is non-degenerate: its members are not all functions of a single transfer matrix) — and use it to derive an exact solution of the cubic nearest-neighbour Ising model, in the standard sense of closed-form thermodynamics (at minimum the free energy per site as a function of tt and the critical point) in the thermodynamic limit. Alternatively, establish the negative: that the twisted tetrahedron equation does not determine such an integrability structure, i.e. that no commuting family of the stated kind can be derived from it.

References

  1. D. V. Talalaev, 'Towards integrable structure in 3d Ising model', Journal of Geometry and Physics 148 (2020) 103545. DOI: 10.1016/j.geomphys.2019.103545; arXiv:1805.04138. https://arxiv.org/abs/1805.04138
  2. D. V. Talalaev, 'Tetrahedron equation: algebra, topology, and integrability', Russian Mathematical Surveys 76:4 (2021) 685-721. DOI: 10.1070/RM10009. https://www.mathnet.ru/eng/rm10009
  3. D. V. Talalaev, 'Asymmetric Hopfield neural network and twisted tetrahedron equation', arXiv:1806.06680 (2018). DOI: 10.48550/arXiv.1806.06680. https://arxiv.org/abs/1806.06680
  4. D. V. Talalaev, 'Zamolodchikov Tetrahedral Equation and Higher Hamiltonians of 2d Quantum Integrable Systems', SIGMA 13 (2017) 031. DOI: 10.3842/SIGMA.2017.031. https://doi.org/10.3842/SIGMA.2017.031
  5. A. B. Zamolodchikov, 'Tetrahedra equations and integrable systems in three-dimensional space', Soviet Physics JETP 52 (1980) 325-336. ADS bibcode 1980JETP...52..325Z. https://ui.adsabs.harvard.edu/abs/1980JETP...52..325Z/abstract
  6. J.-M. Maillet and F. Nijhoff, 'Integrability for multidimensional lattice models', Physics Letters B 224 (1989) 389-396. DOI: 10.1016/0370-2693(89)91466-4. https://doi.org/10.1016/0370-2693(89)91466-4
  7. L. Onsager, 'Crystal Statistics. I. A Two-Dimensional Model with an Order-Disorder Transition', Physical Review 65 (1944) 117-149. DOI: 10.1103/PhysRev.65.117. https://doi.org/10.1103/PhysRev.65.117
  8. S. Istrail, 'Statistical mechanics, three-dimensionality and NP-completeness', Proceedings of the 32nd ACM Symposium on Theory of Computing (STOC '00), pp. 87-96. DOI: 10.1145/335305.335316. https://doi.org/10.1145/335305.335316
  9. S. El-Showk, M. F. Paulos, D. Poland, S. Rychkov, D. Simmons-Duffin, A. Vichi, 'Solving the 3D Ising model with the conformal bootstrap', Physical Review D 86 (2012) 025022. DOI: 10.1103/PhysRevD.86.025022. https://doi.org/10.1103/PhysRevD.86.025022
  10. P. Padmanabhan and V. Korepin, 'Majorana fermions solve the tetrahedron equations as well as higher simplex equations', Nuclear Physics B 1014 (2025) 116865. DOI: 10.1016/j.nuclphysb.2025.116865; arXiv:2410.20328. https://arxiv.org/abs/2410.20328
  11. V. V. Bazhanov, R. Kashaev, V. V. Mangazeev, S. M. Sergeev, 'Quantum Dilogarithms and New Integrable Lattice Models in Three Dimensions', arXiv:2512.23338 (2025). DOI: 10.48550/arXiv.2512.23338. https://arxiv.org/abs/2512.23338
  12. B. Bychkov, A. Kazakov, D. Talalaev, 'Functional Relations on Anisotropic Potts Models: from Biggs Formula to the Tetrahedron Equation', SIGMA 17 (2021) 035. DOI: 10.3842/SIGMA.2021.035; arXiv:2005.10288. https://arxiv.org/abs/2005.10288
  13. D. Zhang, 'Exact Solution for Three-Dimensional Ising Model', Symmetry 13 (2021) 1837. DOI: 10.3390/sym13101837. https://doi.org/10.3390/sym13101837 — a disputed claim, unrelated to tetrahedron-equation structures.
  14. R. Inoue, A. Kuniba, X. Sun, Y. Terashima, J. Yagi, 'Solutions of Tetrahedron Equation from Quantum Cluster Algebra Associated with Symmetric Butterfly Quiver', arXiv:2403.08814 (2024). DOI: 10.48550/arXiv.2403.08814. https://arxiv.org/abs/2403.08814