In two spatial dimensions, transfer matrices of exactly solvable lattice models are matrix product operators (MPOs) that can be diagonalized exactly through free-fermion methods, the Yang-Baxter equation, and the algebraic Bethe ansatz, yielding closed-form free energies and critical behavior (Onsager's solution of the two-dimensional Ising model being the prototype). The 2017 review by Haegeman and Verstraete formulated the two-dimensional generalization of this program as the field's 'real challenge': an algebraic and computational framework for projected entangled-pair operators (PEPOs), the two-dimensional transfer operators of three-dimensional models. As of 2026, no nontrivial family of PEPOs exhibiting a phase transition has been exactly diagonalized. Solutions of the Zamolodchikov tetrahedron equations have produced commuting three-dimensional transfer-matrix families and even exactly computed free energies (the Zamolodchikov model; the quantum-dilogarithm classes of 2025), but none has delivered an exactly characterized phase transition, and the simple-cubic Ising model remains without an exact solution. The problem is to construct - by tetrahedron-equation solutions or any other algebraic mechanism - a genuinely interacting three-dimensional lattice family whose two-dimensional transfer operators are diagonalized exactly in the thermodynamic limit, and to extract from that solution the exact location and leading critical singularities of its phase transition.

Background

A classical lattice model in dd spatial dimensions can be built by stacking (d−1)(d-1)-dimensional layers, each layer coupled to the next by a transfer operator; the free energy and correlation functions are governed by the eigenvalues and eigenvectors of this operator. For d=2d=2, the transfer matrix acting on a one-dimensional row is a matrix product operator (MPO) - an operator built as a chain of local tensors, each carrying one physical index and auxiliary ('bond') indices contracted with the neighboring tensors, with the auxiliary dimension called the bond dimension. The exact solution of the two-dimensional Ising model by Onsager (1944) is, in this language, an exact diagonalization of a family of one-dimensional transfer matrices: commuting transfer matrices, a closed-form free energy, and exact critical exponents at the order-disorder transition. The same structure underlies all exactly solvable two-dimensional models: local Boltzmann weights satisfying the Yang-Baxter equation (a consistency relation on three-spin configurations that implies commutativity of transfer matrices) make the algebraic Bethe ansatz available - an exact method that diagonalizes such commuting transfer matrices by constructing their eigenvectors with creation operators of an auxiliary algebra - and the resulting exact eigenstructure reveals phase transitions and their critical behavior. Matrix product states (MPS) - finite-bond-dimension variational wavefunctions built from the same chain-of-local-tensors structure - make this exact eigenstructure computationally usable.

For d=3d=3, the transfer operator acts on a whole two-dimensional layer and is naturally a projected entangled-pair operator (PEPO) - a two-dimensional network of local tensors with finite bond dimension, the natural generalization of an MPO. Haegeman and Verstraete's 2017 review of transfer-matrix and MPO diagonalization concluded that constructing an algebraic and computational framework for PEPOs is the outstanding open challenge of the field: 'As far as we are aware of, no exact diagonalizations have been reported for a family of PEPOs exhibiting phase transitions (such as the 3D Ising model).' The canonical target is the Ising model on the simple cubic lattice, whose critical point is known only numerically (kBTc/J≈4.5115k_B T_c/J \approx 4.5115, i.e. critical coupling Kc≈0.22165K_c \approx 0.22165, with exponents ν≈0.630\nu \approx 0.630 and η≈0.036\eta \approx 0.036 from high-precision Monte Carlo and conformal-bootstrap studies - the bootstrap being a method that fixes critical exponents from consistency conditions on the correlation functions of the critical theory -) - no closed-form counterpart of Onsager's solution exists.

The expected algebraic route was proposed by Zamolodchikov (1980-1981): the tetrahedron equations, a three-dimensional generalization of the Yang-Baxter equation expressing the consistency of the two ways of resolving the intersection of four plane worldsheets (each swept out by a straight string in three-dimensional spacetime), which imply commutativity of layer-to-layer transfer matrices. Baxter verified Zamolodchikov's solution and computed the partition function of the resulting three-dimensional Zamolodchikov model exactly (1984). However, that model lives on a special tetrahedral lattice, its exact solution stops at the free energy, and even its criticality was subsequently the subject of a separate numerical inquiry ('Is the Zamolodchikov model critical?', Baxter and Forrester, 1985) - no exactly characterized phase transition came out of it. A related 'first step' noted by the source review, the Walker-Wang exactly solvable three-dimensional quantum models, concern topological quantum order rather than thermal phase transitions of transfer operators.

Three classes of adjacent exact results must be distinguished from the requested nontrivial solution: (i) the three-dimensional spherical model is solved exactly and has a phase transition, but by a global constraint and Fourier methods, not by diagonalization of a commuting family of layer-to-layer transfer operators; (ii) models on hierarchical or fractal lattices admit exact recursions and phase transitions, but are not regular-lattice transfer-operator diagonalizations; (iii) layered or decoupled models that reduce to two-dimensional exact solutions are 'trivial' in this sense. The state of the art as of 2026 is that tetrahedron-equation solutions are abundant (quantum-group related constructions, quantum cluster algebras, Teichmüller topological quantum field theory, Clifford-algebra solvers), and the newest integrable three-dimensional classes (Bazhanov, Kashaev, Mangazeev and Sergeev, 2025) have exactly computed partition functions per site - but no exact critical point or critical exponent of any genuinely three-dimensional, non-mean-field universality class has ever been derived from a transfer-operator diagonalization.

Problem Statement

Construct a family of local Boltzmann weights for a three-dimensional lattice model on a regular lattice - equivalently, a family of two-dimensional transfer operators realized as projected entangled-pair operators (PEPOs) of finite bond dimension - satisfying all of the following: (i) the family is genuinely interacting, i.e., not reducible as a model to decoupled or layered two-dimensional exactly solved models, to models whose local Boltzmann weights are those of a free (Gaussian, free-fermion or free-boson) theory by construction, to globally constrained (spherical-model-type) constructions, or to hierarchical/fractal-lattice recursions - these exclusions concern degenerate model classes rather than solution techniques, so a fermionization-type exact diagonalization of a genuinely interacting three-dimensional family (including the simple-cubic Ising model, should it be solved) remains admissible; (ii) its layer-to-layer transfer operators admit exact diagonalization in the thermodynamic limit through a commuting family of transfer matrices - nontrivial solutions of the Zamolodchikov tetrahedron equations, or of an equivalent three-dimensional integrability structure, are the expected route, but any explicitly stated algebraic mechanism implying the required commutativity and diagonalizability is admissible; and (iii) at least one member (or limit point) of the family exhibits a genuine phase transition whose exact location and leading critical singularities - free energy and order-parameter critical behavior, in the sense delivered by the Onsager paradigm of the source review (closed-form free energy, exact critical point, exact order-parameter law) - are extracted exactly from that diagonalization. If the solved family contains the simple-cubic Ising model, the exact critical data must reproduce the numerically established benchmarks (Kc≈0.22165K_c \approx 0.22165, ν≈0.630\nu \approx 0.630, η≈0.036\eta \approx 0.036) within their published confidence intervals. The three clauses are jointly one objective - exact solvability of a genuinely three-dimensional phase transition via two-dimensional transfer-operator diagonalization, the three-dimensional analogue of Onsager's solution: clause (ii) supplies the mechanism, clause (iii) is the physical payoff that defines the transition as exactly characterized, and clause (i) excludes degenerate loopholes. Partial results do not complete the problem: an exact free energy without a phase transition (the Zamolodchikov-model precedent), tetrahedron-equation solutions without exact thermodynamics, commuting families without diagonalization, or improved numerical contraction schemes each leave the objective unmet.

References

  1. J. Haegeman and F. Verstraete, Diagonalizing Transfer Matrices and Matrix Product Operators: A Medley of Exact and Computational Methods, Annual Review of Condensed Matter Physics 8, 355-406 (2017), DOI 10.1146/annurev-conmatphys-031016-025507, arXiv:1611.08519, https://arxiv.org/abs/1611.08519
  2. L. Onsager, Crystal Statistics. I. A Two-Dimensional Model with an Order-Disorder Transition, Physical Review 65, 117-149 (1944), DOI 10.1103/PhysRev.65.117, https://doi.org/10.1103/PhysRev.65.117
  3. A. B. Zamolodchikov, Tetrahedra equations and integrable systems in three-dimensional space, Soviet Physics JETP 52, 325-336 (1980) [original Russian publication: Zh. Eksp. Teor. Fiz. 79, 641 (1980)]; pre-digital Soviet-journal article with no assigned DOI
  4. A. B. Zamolodchikov, Tetrahedron Equations and the Relativistic S Matrix of Straight Strings in (2+1)-dimensions, Communications in Mathematical Physics 79, 489-505 (1981), DOI 10.1007/BF01209309, https://doi.org/10.1007/BF01209309
  5. K. Walker and Z. Wang, (3+1)-TQFTs and topological insulators, Frontiers of Physics 7, 150-159 (2012), DOI 10.1007/s11467-011-0194-z, https://doi.org/10.1007/s11467-011-0194-z
  6. R. J. Baxter, On Zamolodchikov's solution of the tetrahedron equations, Communications in Mathematical Physics 88, 185-205 (1983), DOI 10.1007/BF01209476, https://doi.org/10.1007/BF01209476
  7. R. J. Baxter, Partition Function of the Three-Dimensional Zamolodchikov Model, Physical Review Letters 53, 1795-1798 (1984), DOI 10.1103/PhysRevLett.53.1795, https://doi.org/10.1103/PhysRevLett.53.1795
  8. R. J. Baxter and P. J. Forrester, Is the Zamolodchikov model critical?, Journal of Physics A: Mathematical and General 18, 1483-1497 (1985), DOI 10.1088/0305-4470/18/9/028, https://doi.org/10.1088/0305-4470/18/9/028
  9. V. V. Bazhanov and S. M. Sergeev, Zamolodchikov's tetrahedron equation and hidden structure of quantum groups, Journal of Physics A 39, 3295-3310 (2006), DOI 10.1088/0305-4470/39/13/009, arXiv:hep-th/0509181, https://arxiv.org/abs/hep-th/0509181
  10. V. V. Mangazeev, V. V. Bazhanov and S. M. Sergeev, An integrable 3D lattice model with positive Boltzmann weights, Journal of Physics A 46, 465206 (2013), DOI 10.1088/1751-8113/46/46/465206, arXiv:1308.4773, https://arxiv.org/abs/1308.4773
  11. V. V. Bazhanov, V. V. Mangazeev, Y. Okada and S. M. Sergeev, An Elliptic Parameterisation of the Zamolodchikov Model, Nuclear Physics B 871, 127-144 (2013), DOI 10.1016/j.nuclphysb.2013.02.011, arXiv:1212.6800, https://arxiv.org/abs/1212.6800
  12. V. V. Bazhanov, R. M. Kashaev, V. V. Mangazeev and S. M. Sergeev, Quantum Dilogarithms and New Integrable Lattice Models in Three Dimensions, arXiv:2512.23338 (2025), https://arxiv.org/abs/2512.23338
  13. R. Inoue, A. Kuniba and Y. Terashima, Tetrahedron equation and quantum cluster algebras, Journal of Physics A 57, 085202 (2024), DOI 10.1088/1751-8121/ad2224, arXiv:2310.14529, https://arxiv.org/abs/2310.14529
  14. P. Gavrylenko, M. Semenyakin and Y. Zenkevich, Solution of tetrahedron equation and cluster algebras, Journal of High Energy Physics 2021(05), 103 (2021), DOI 10.1007/JHEP05(2021)103, arXiv:2010.15871, https://arxiv.org/abs/2010.15871
  15. M. Shim, X. Sun, H. E. Wang and J. Yagi, Solving the tetrahedron equation by Teichmüller TQFT, JHEP 2026(06), 141 (2026), DOI 10.1007/JHEP06(2026)141, arXiv:2602.22060, https://arxiv.org/abs/2602.22060
  16. P. Padmanabhan, V. K. Singh and V. Korepin, Clifford Solver for the Tetrahedron Equation and Its Variants, Bulgarian Journal of Physics 52(s1), 126-131 (2025), arXiv:2510.23944, https://arxiv.org/abs/2510.23944
  17. A. Molnar, A. Ruiz de Alarcon, J. Garre-Rubio, N. Schuch, J. I. Cirac and D. Perez-Garcia, Matrix Product Operator Algebras I: Representations of Weak Hopf Algebras and Projected Entangled Pair States, Communications in Mathematical Physics 407, 203 (2026), DOI 10.1007/s00220-026-05670-w, arXiv:2204.05940, https://arxiv.org/abs/2204.05940
  18. W. Tang, F. Verstraete and J. Haegeman, Matrix product state fixed points of non-Hermitian transfer matrices, Physical Review B 111, 035107 (2025), DOI 10.1103/PhysRevB.111.035107, arXiv:2311.18733, https://arxiv.org/abs/2311.18733
  19. A. Ueda, L. Burgelman, L. Tagliacozzo and L. Vanderstraeten, Interface roughening in the 3-D Ising model with tensor networks, arXiv:2601.07829 (2026), https://arxiv.org/abs/2601.07829