Yang-Baxter integrability of the odd eight-vertex model, singularity completeness, and finite-torus equivalence for sixteen-vertex weight sets on the square lattice

The sixteen-vertex model on the square lattice, with two arrow states per edge and sixteen unconstrained Boltzmann weights, splits by arrow parity into the classical even eight-vertex model (containing Baxter's solvable symmetric model) and the far less explored odd eight-vertex model (containing the close-packed dimer model). A 2017 survey by Assis recorded that no new integrable points had been found and that the full sixteen-vertex model had not been solved; it left open the existence of Brascamp-Kunz-type boundary conditions giving product-form finite-lattice partition functions for the free-fermion subcases and a Hurst-type theorem on the singularities of staggered free-fermion models, and recorded that algebraic-invariant equalities give only sufficient conditions for finite-torus equivalence. Later work resolved parts of this program: Li and Wang (2025) found the product-form boundary conditions for the even and odd free-fermion models, and Martins (2018) proved Yang-Baxter integrability of the symmetric odd eight-vertex model on Baxter's manifold, leaving the asymmetric case explicitly open, including whether the free-fermion odd model admits commuting transfer matrices at finite size. The surviving open problem is threefold: (A) determine whether the asymmetric odd eight-vertex model admits integrable manifolds with finite-size commuting transfer matrices beyond the symmetric solution on Baxter's manifold; (B) prove or disprove that the temperature-independent free-fermion conditions capture all thermodynamic-limit singularities on the principal sheet of staggered free-fermion models on the square lattice; (C) give necessary-and-sufficient finite-torus equivalence criteria for sixteen-vertex weight sets beyond the known sufficient algebraic conditions.

Background

Assign to each edge of the square lattice one of two states, drawn as an arrow pointing along the edge, and assign a Boltzmann weight to each configuration of the four edges meeting at a vertex. With no restriction beyond two states per edge there are $2^4=16$ vertex configurations and hence sixteen weights: this is the sixteen-vertex model. Configurations with an even number of arrows entering each vertex define the even eight-vertex model (weights $w_1,\ldots,w_8$ ); the complementary eight configurations, each with an odd number of entering arrows, define the odd eight-vertex model (weights $v_1,\ldots,v_8$ ); models with both even and odd weights nonzero are called mixed. 'Symmetric' weights are invariant under simultaneous reversal of all arrows (four independent weights $a,b,c,d$ ). The even model contains the six-vertex model and Baxter's symmetric eight-vertex model; the odd model contains the close-packed dimer model. The partition function is the sum, over all arrow configurations, of the product of vertex weights. On a torus it is built by the row-to-row transfer matrix; the strongest notion of exact solvability holds when transfer matrices at different values of a spectral parameter (an auxiliary complex argument labeling the weight family) commute, which follows if the $4\times 4$ weight matrix (the Lax operator $L$ ) admits an intertwiner $R$ satisfying the Yang-Baxter equation $R_{12}L_{13}L_{23}=L_{23}L_{13}R_{12}$ (in vertex-model language, the star-triangle relation), where subscripts label tensor factors. Baxter solved the symmetric even model on the elliptic-curve manifold $(ab-cd)/(ab+cd)=\Gamma$ , $(a^2+b^2-c^2-d^2)/(2(ab+cd))=\Delta$ ; the six-vertex and free-fermion cases are degenerations. For the asymmetric even eight-vertex model (no arrow-reversal symmetry), Galleas and Martins showed that the integrable manifolds occur precisely when the weights are symmetric or satisfy the free-fermion condition; the analogous question for the asymmetric odd model is open. A weaker, distinct solvability route is the free-fermion condition of Fan and Wu, $w_1w_2+w_3w_4=w_5w_6+w_7w_8$ (odd analogue $v_1v_2+v_3v_4=v_5v_6+v_7v_8$ ), under which the partition function is a Pfaffian (the square root of the determinant of an antisymmetric matrix) computable by dimer methods; this Pfaffian solvability is not equivalent to Yang-Baxter integrability, although Melotti constructed Yang-Baxter solutions for free-fermion eight-vertex models in a checkerboard setting where the two weight sets alternate between sublattices. The 'Krinsky-type extra constraints' are the three additional weight restrictions that, imposed on top of the free-fermion condition, give commuting transfer matrices in the even eight-vertex model, $w_6w_8/(w_5w_7)=\Delta_1$ , $(w_1w_4+w_2w_3)/(w_5w_7)=\Delta_2$ , $(w_1^2+w_4^2-w_2^2-w_3^2)/(w_5w_7)=\Delta_3$ , with free parameters $\Delta_1,\Delta_2,\Delta_3$ (Martins 2018, Eq. (41)); the odd-model analogue replaces $w_i$ by $v_i$ , and a manifold of a type other than this class is one defined by weight restrictions not of this form. Gaaff showed that the Ising model in an imaginary field $H=i\pi/(2\beta)$ maps to a solvable point of the sixteen-vertex model. Brascamp and Kunz exhibited fixed-arrow boundary conditions for the square-lattice Ising model under which the finite-lattice partition function collapses to a single double product. 'Staggered' models alternate two weight sets between rows (column staggering) or between the two sublattices (bi-partite staggering). Writing each weight as $w_i=e^{-\beta\epsilon_i}$ with $\beta$ the inverse temperature and $\epsilon_i$ the interaction energies, the temperature-independent free-fermion condition is the restriction of the free-fermion condition to $\epsilon_1+\epsilon_2+\epsilon_3+\epsilon_4=\epsilon_5+\epsilon_6+\epsilon_7+\epsilon_8$ , equivalently $w_1w_2w_3w_4=w_5w_6w_7w_8$ , a constraint on the energies alone that holds at every temperature; for a staggered model it is imposed on each weight set separately. Hurst proved that for the homogeneous even free-fermion model the only singularities of the free energy on the principal sheet (the sheet reached from the physical high-temperature region without crossing branch cuts in the complex temperature plane) are those obeying this temperature-independent form, so that no physical critical behavior is lost by imposing it; for staggered free-fermion models the corresponding statement has never been established. The weak-graph expansion and the polynomial invariants of the $SL(2)\times SL(2)$ transformation of the vertex weights, under which the toroidal partition function is invariant, relate models with equal finite-torus partition functions; these relations are sufficient, not necessary, for equivalence, and may map positive weights to negative or complex ones, so finite-torus equality need not preserve physical positivity.

Problem Statement

Determine the integrable structure and finite-size solvability of the odd eight-vertex model, the thermodynamic-limit singularity structure of staggered free-fermion models, and the finite-torus equivalence classes of sixteen-vertex weight sets on the square lattice. (A) Integrable manifolds: for the asymmetric odd eight-vertex model (weights $v_1,\ldots,v_8$ without arrow-reversal symmetry, in the weight-index convention of Figs. 1-2 of Assis and of Martins), determine all weight manifolds beyond Baxter's symmetric elliptic-curve manifold (which carries the symmetric odd model solved by Martins with explicit R-matrices) on which commuting transfer matrices exist at finite lattice size, or prove that no such manifold exists beyond the symmetric solution. Via the Wu-Kunz correspondence (Martins 2018, Eq. (4)), the odd-model partition function equals that of a bi-partite staggered even model with permuted weights, and the finite-size problem is formulated through the two explicit $4\times 4$ staggered Lax operators $L(v_1,\ldots,v_8)$ and $\mathbf{L}(v_1,\ldots,v_8)$ (Martins 2018, Eq. (44)) and the Yang-Baxter/intertwiner equation for the tensor product of these two types of Lax operators. In particular, settle the dichotomy stated by Martins for the odd free-fermion manifold ( $v_1v_2+v_3v_4=v_5v_6+v_7v_8$ ): either exhibit an integrable manifold of a type other than the Krinsky-type extra constraints excluded by his two-site commutator computation, or establish that the free-fermion odd eight-vertex model is an example in which the free energy is exactly computable in the continuum without the existence of commuting transfer matrices at finite lattice sizes. This component asks for the classification of all such manifolds; the free-fermion dichotomy is a mandatory special case of it, so deciding the free-fermion case in either direction is required but is not by itself sufficient. The component is complete only once the full set of manifolds beyond the symmetric solution is determined. A no-go theorem confined to a chosen ansatz class for the R-matrix does not resolve this component. (B) Singularity completeness: prove or disprove, in the spirit of Hurst's analysis of the homogeneous even free-fermion model, that no thermodynamic-limit singularities of the free energy beyond those implied by the temperature-independent free-fermion conditions occur on the principal sheet for the column- and bi-partite-staggered free-fermion eight-vertex models on the square lattice; that is, decide whether every principal-sheet singularity obeys the temperature-independent free-fermion conditions. This is a one-directional completeness statement, not a set equality between the singularity locus and the temperature-independent free-fermion locus, and - while Assis's corresponding open item also mentions free-fermion models on other lattices - the present component is explicitly restricted to the square lattice. (C) Finite-torus equivalence: give necessary-and-sufficient conditions for two sixteen-vertex weight sets (in Assis's index convention) to have identical finite-torus partition functions on all $M\times N$ toroidal lattices, going beyond the sufficient conditions supplied by weak-graph transformations and $SL(2)\times SL(2)$ -invariant equalities, which Assis recorded "do not fully specify a model's equivalency class", exhibiting mappings outside the invariant class. The three components correspond to separate open items recorded in the sources cited below and are logically independent.

References

M. Assis, 'The 16-vertex model and its even and odd 8-vertex subcases on the square lattice', Journal of Physics A: Mathematical and Theoretical 50, 395001 (2017), doi:10.1088/1751-8121/aa842c, https://arxiv.org/abs/1702.02110 (primary source of the open items underlying components (B) and (C))
M. J. Martins, 'Integrability of the odd eight-vertex model with symmetric weights', Journal of Physics A: Mathematical and Theoretical 51, 225002 (2018), doi:10.1088/1751-8121/aab9cc, https://arxiv.org/abs/1711.03131 (solves the Yang-Baxter equation for the symmetric odd model; states the asymmetric case as open)
De-Zhang Li and Xin Wang, 'Free-fermion approach to the partition function zeros: special boundary conditions and product form of solution', arXiv:2507.21943 (2025), https://arxiv.org/abs/2507.21943 (finds the Brascamp-Kunz-type boundary conditions resolving Assis's boundary-condition problem)
M. J. Martins, 'On the equivalence between n-state spin and vertex models on the square lattice', Nuclear Physics B 1005, 116610 (2024), doi:10.1016/j.nuclphysb.2024.116610, https://arxiv.org/abs/2405.07341 (mixed even/odd eight-vertex model equivalent to the Ising model in a field; non-difference R-matrix)
M. J. Martins, 'Embedding integrable spin models in solvable vertex models on the square lattice', Nuclear Physics B 1013, 116849 (2025), doi:10.1016/j.nuclphysb.2025.116849, https://arxiv.org/abs/2501.02377 (spin-to-vertex embedding machinery; background for weight-space mappings)
W. Galleas and M. J. Martins, 'Yang-Baxter equation for the asymmetric eight-vertex model', Physical Review E 66, 047103 (2002), doi:10.1103/PhysRevE.66.047103, https://arxiv.org/abs/nlin/0303036 (integrable manifolds of the asymmetric even eight-vertex model occur at symmetric or free-fermion weights; the even-model precedent for the open odd-model question)
C. Fan and F. Y. Wu, 'General lattice model of phase transitions', Physical Review B 2, 723-733 (1970), doi:10.1103/PhysRevB.2.723, https://doi.org/10.1103/PhysRevB.2.723 (origin of the free-fermion condition and the sixteen-vertex model)
R. J. Baxter, 'Eight-vertex model in lattice statistics', Physical Review Letters 26, 832-833 (1971), doi:10.1103/PhysRevLett.26.832, https://doi.org/10.1103/PhysRevLett.26.832 (solution of the symmetric even model via star-triangle relations)
F. Y. Wu and H. Kunz, 'The odd eight-vertex model', Journal of Statistical Physics 116, 67-78 (2004), doi:10.1023/B:JOSS.0000037206.47155.58, https://doi.org/10.1023/B:JOSS.0000037206.47155.58 (thermodynamic-limit solution of the odd free-fermion model via staggered even models)
C. A. Hurst, 'Solution of plane Ising lattices by the Pfaffian method', Journal of Chemical Physics 38, 2558-2571 (1963), doi:10.1063/1.1733541, https://doi.org/10.1063/1.1733541 (Pfaffian method; the model for the singularity-completeness proof requested by Assis)
H. J. Brascamp and H. Kunz, 'Zeroes of the partition function for the Ising model in the complex temperature plane', Journal of Mathematical Physics 15, 65-66 (1974), doi:10.1063/1.1666505, https://doi.org/10.1063/1.1666505 (boundary conditions giving double-product finite-lattice Ising partition functions)
A. Gaaff, 'The 2-d Ising model in a magnetic field i*pi/2 as a soluble case of the sixteen vertex model', Physics Letters A 49, 103-105 (1974), doi:10.1016/0375-9601(74)90690-2, https://doi.org/10.1016/0375-9601(74)90690-2 (imaginary-field Ising point of the sixteen-vertex model)
C. Ahn, M. Horibe and K. Shigemoto, 'Relation between Yang-Baxter and pair propagation equations in 16-vertex models', Modern Physics Letters B 8, 1729-1738 (1994), doi:10.1142/S0217984994001655, arXiv:hep-th/9312199, https://arxiv.org/abs/hep-th/9312199 (shows pair-propagation/free-fermion solvability is not equivalent to the Yang-Baxter equation)
P. Melotti, 'The free-fermion eight-vertex model: couplings, bipartite dimers and Z-invariance', Communications in Mathematical Physics 381, 33-82 (2020), doi:10.1007/s00220-020-03901-2, https://arxiv.org/abs/1811.02026 (Yang-Baxter solutions for checkerboard free-fermion eight-vertex models and exact free-fermion correlations; integrable structure within the free-fermion family in the staggered setting)
E. Pospíšilová, R. Krčmár, A. Gendiar and L. Šamaj, 'Full nonuniversality of the symmetric 16-vertex model on the square lattice', Physical Review E 102, 012125 (2020), doi:10.1103/PhysRevE.102.012125, https://arxiv.org/abs/2002.02757 (numerical transfer-matrix study confirming the symmetric sixteen-vertex model is not exactly solved)
De-Zhang Li, Xin Wang and Xiao-Bao Yang, 'Free-fermion models and two-dimensional Ising models under zero field and imaginary field i(pi/2)kBT', Entropy 27, 799 (2025), doi:10.3390/e27080799, https://arxiv.org/abs/2406.00232 (partition-function zeros of free-fermion models; companion work to Li and Wang 2025)