For the free-fermion vertex models isolated in de Leeuw, Paletta, Pribytok, Retore and Torrielli (JHEP 02 (2021) 191), the transfer matrix can be rewritten in terms of canonical fermionic oscillators via a Bogoliubov transformation: this was established at arbitrary chain length N for the 6-vertex class-B sector, but only on two sites for the 8-vertex class-B model, where a specially constructed reference state (the pseudo-pseudo vacuum) is needed. The authors state explicitly that they do not know whether the required canonical map and pseudo-pseudo vacuum exist for N greater than 2. Separately, they conjecture that the free-fermion condition — the algebraic property making such a reformulation possible — holds in settings far beyond the regular nearest-neighbour 4×4 R-matrices, including non-nearest-neighbour interactions, deformed S-matrices, and higher-dimensional Korepanov-type constructions. It remains open whether the canonical map and pseudo-pseudo vacuum exist for N > 2, and whether the free-fermion condition holds beyond regular nearest-neighbour 4×4 R-matrices.

Background

An integrable two-dimensional vertex model is built from an R-matrix, a matrix R(θ₁, θ₂) acting on the tensor product of two two-dimensional spaces, which satisfies the Yang-Baxter equation (YBE) — the consistency condition ensuring that scattering of three excitations is independent of the order of pairwise collisions. From the R-matrix one builds a transfer matrix T_N(θ) = tr R_N(θ) · … · R_1(θ), a product of R-matrices along a chain of N sites traced over the auxiliary space; commuting transfer matrices at different spectral parameters θ generate an extensive family of conserved charges, including the spin-chain Hamiltonian. Diagonalising T_N is the central spectral problem of the theory, usually attacked with the algebraic Bethe ansatz (ABA), which requires a reference state (a pseudo-vacuum) annihilated by the off-diagonal entries of the monodromy matrix (the ordered product of R-matrices along the chain, of which the transfer matrix is the trace). The R-matrices relevant here are of non-difference form — R(θ₁, θ₂) depends on the two spectral parameters separately, not only on their difference — which is typical of the worldsheet S-matrices of string theory on anti-de Sitter backgrounds (AdS/CFT integrability), e.g. for strings on AdS₂ and AdS₃ supported by Ramond-Ramond flux (an antisymmetric background flux characteristic of AdS/CFT string theories); these theories describe massless and massive excitations and generally admit no standard ABA pseudo-vacuum. A model is of free-fermion type when its R-matrix coefficients satisfy the free-fermion condition, an algebraic constraint descending from the classical Fan-Wu condition for the eight-vertex model, under which the transfer matrix can be recast entirely in terms of canonical fermionic creation and annihilation operators — the mechanism behind the exact solvability of the two-dimensional Ising model. The starting point of the source paper is the classification of integrable nearest-neighbour spin-chain Hamiltonians by de Leeuw, Paletta, Pribytok, Retore and Ryan (Phys. Rev. Lett. 125 (2020) 031604), which organises regular 4×4 R-matrices — 'regular' meaning that the R-matrix reduces to the (graded) permutation at coincident spectral parameters, so that it defines a nearest-neighbour Hamiltonian — into classes A and B; the source proves that the two new class-B models of this classification, which contain all the AdS/CFT worldsheet S-matrices, satisfy the free-fermion condition, while the class-A models (essentially the usual XXZ and XYZ spin chains) satisfy instead a different Baxter condition, both conditions descending from a single universal functional relation. For the class-B six-vertex model (6vB) the source constructs the free-fermion form at arbitrary N by an inhomogeneous generalisation of the coordinate Bethe ansatz (the original form of the Bethe ansatz, which builds eigenstates as superpositions of wavefunctions labelled by particle positions) followed by a Bogoliubov transformation (a linear canonical redefinition mixing creation and annihilation operators, parametrised by Bogoliubov angles). For the class-B eight-vertex model (8vB), however, the construction is carried out only on two sites: the transfer matrix factorises there as T₂ = t₁ + t₂N₁ + t₃N₂ + t₄N₁N₂ in terms of number operators Nᵢ = ηᵢ†ηᵢ, with a factorisation condition t₂t₃ = t₁t₄ verified numerically, and the eigenstates are created by oscillators ηᵢ† acting on a reference state |vac⟩ = κ(|ψ⟩⊗|ψ⟩ + i coth(μ₂/2)|φ⟩⊗|φ⟩) that the authors call the pseudo-pseudo vacuum — it plays the role of the ABA pseudo-vacuum but is itself a nontrivial superposition built by the Bogoliubov map, since no ordinary pseudo-vacuum exists for these non-difference-form, non-triangular R-matrices. The source also verifies the free-fermion condition for the AdS₂, AdS₃ (massless, massive, mixed-flux) worldsheet S-matrices and studies 16×16 su(2)⊕su(2) models akin to AdS₅, and it notes that the massless AdS₂ R-matrix — which is not of regular nearest-neighbour type — still satisfies the condition, motivating the conjecture that the condition is far more general, possibly connected to Korepanov-type three-dimensional integrable structures (tetrahedral Zamolodchikov-algebra constructions that extend vertex models to three dimensions).

Problem Statement

The objective is to establish the true scope and completeness of free-fermion solvability for the integrable vertex models introduced and analysed in arXiv:2011.08217, through two open items recorded in the source, each an independent target in its own right: (i) completing their diagonalisation where it is currently only exemplified, and (ii) delimiting the class of integrable R-matrices for which free-fermion solvability holds at all. Concretely: (A) For the 8-vertex class-B (8vB) free-fermion R-matrix, for which the two-site transfer matrix is diagonalised by a Bogoliubov transformation onto canonical oscillators ηᵢ, ηᵢ† acting on the pseudo-pseudo vacuum |vac⟩ = κ(|ψ⟩⊗|ψ⟩ + i coth(μ₂/2)|φ⟩⊗|φ⟩), construct — or prove the impossibility of —, for each chain length N beyond the N=2 examples, a canonical (Bogoliubov) map that is global across the sites of the chain at that N and may itself depend on N (in the sense of the source's oscillators ηᵢ(N), ηᵢ†(N)), and that produces a pseudo-pseudo vacuum together with a complete set of normalised free-fermion eigenstates of the transfer matrix T_N. (B) Prove, refute, or systematically classify the source's conjecture that the free-fermion condition holds beyond regular nearest-neighbour 4×4 R-matrices — in particular for non-nearest-neighbour interactions (of which the massless AdS2 R-matrix is the observed example), deformed S-matrices, and higher-dimensional / Korepanov-type integrable constructions. The target of (B) is the general conjecture, and the settings just listed are illustrative instances of it rather than separate targets: settling one of them advances the item but does not by itself discharge it, and its classification branch must return an explicit output — a general criterion for the free-fermion condition on the stated class, or an explicit classified list of the free-fermion families together with the boundary of that class — rather than a collection of examples. Questions (A) and (B) are conceptually related — both bear on whether the free-fermion reformulation is a structural property of these models or an artefact of small examples — but they are distinct open items of the source, and a resolution of either constitutes substantive progress on its own. The source additionally records, without determinate completion criteria, the wish for compact closed-form coefficients in the massive AdS3 and 16×16 su(2)⊕su(2) sectors, for the explicit operator functions f, g, F, G relating on-shell ABA operators to the canonical oscillators at general N, and for a transparent crossing behaviour of the η variables; these are contextual desiderata that a complete answer would be expected to illuminate, but they do not by themselves define the completion boundary.

References

  1. M. de Leeuw, C. Paletta, A. Pribytok, A. L. Retore, A. Torrielli, 'Free Fermions, vertex Hamiltonians, and lower-dimensional AdS/CFT', JHEP 02 (2021) 191. DOI: 10.1007/JHEP02(2021)191, arXiv:2011.08217, https://arxiv.org/abs/2011.08217
  2. M. de Leeuw, C. Paletta, A. Pribytok, A. L. Retore, P. Ryan, 'Classifying Nearest-Neighbor Interactions and Deformations of AdS', Phys. Rev. Lett. 125 (2020) 031604. DOI: 10.1103/PhysRevLett.125.031604, https://doi.org/10.1103/PhysRevLett.125.031604
  3. M. de Leeuw, A. Pribytok, A. L. Retore, P. Ryan, 'Integrable deformations of AdS/CFT', JHEP 05 (2022) 012. DOI: 10.1007/JHEP05(2022)012, arXiv:2109.00017, https://arxiv.org/abs/2109.00017
  4. L. Corcoran, M. de Leeuw, 'All regular 4×4 solutions of the Yang-Baxter equation', SciPost Phys. Core 7, 045 (2024). DOI: 10.21468/SciPostPhysCore.7.3.045, arXiv:2306.10423, https://arxiv.org/abs/2306.10423
  5. B. Hoare, A. L. Retore, F. K. Seibold, 'Elliptic deformations of the AdS3×S3×T4 string', JHEP 04 (2024) 042. DOI: 10.1007/JHEP04(2024)042, arXiv:2312.14031, https://arxiv.org/abs/2312.14031
  6. A. Torrielli, 'On factorising twists in AdS3 and AdS2', J. Geom. Phys. 183 (2023) 104690. DOI: 10.1016/j.geomphys.2022.104690, arXiv:2203.15367, https://arxiv.org/abs/2203.15367
  7. J. M. Nieto García, R. Ruiz, A. Torrielli, 'Infinite-dimensional R-matrices for the relativistic scattering of massless modes on AdS2', JHEP 05 (2023) 229. DOI: 10.1007/JHEP05(2023)229, arXiv:2303.16108, https://arxiv.org/abs/2303.16108
  8. A. Torrielli, 'AdS3 integrability, the Sine-Gordon model, and fractional supersymmetry', Phys. Rev. D 113 (2026) 065013. DOI: 10.1103/zbtr-ztw6, arXiv:2508.12316, https://arxiv.org/abs/2508.12316
  9. B. Brubaker, D. Bump, H. P. A. Gustafsson, 'Quantum superalgebras and the free-fermionic Yang-Baxter equation', J. Algebra 696 (2026) 315-343. DOI: 10.1016/j.jalgebra.2026.01.046, arXiv:2503.24189, https://arxiv.org/abs/2503.24189
  10. A. Melikyan, 'On extension of the Yang-Baxter equation and the fermionic R-operator', Nucl. Phys. B 986 (2023) 116062. DOI: 10.1016/j.nuclphysb.2022.116062, arXiv:2212.09514, https://arxiv.org/abs/2212.09514