Multiplicity-generalized conformal-field-theory construction of trigonometric $R$-matrices and interaction-round-a-face weights (benchmark: the unrestricted $\mathfrak{su}(3)_k$ adjoint face model)

The 'conformal field theory (CFT) approach' constructs trigonometric solutions of the face-type Yang-Baxter equation - the Boltzmann weights of integrable interaction-round-a-face (IRF) lattice models - from rational conformal field theory data, by extracting ultraviolet-limit projectors and fixing eigenvalue-function ratios through conformal dimensions. In its present form it requires fusion rules without multiplicities. Belavin-Gepner-Ramos Cabezas-Runov (JHEP 2025) solved the unrestricted $\mathfrak{su}(3)_k$ face model with adjoint admissibility conditions by an independent quantum-group route and showed that the obstruction to extending the CFT approach is concrete: in the adjoint tensor product the submodule $(1,1)$ occurs twice, making the tensor-product decomposition non-unique so that the $R$-matrix mixes the two copies, while the multiplicity-one pair $(3,0)/(0,3)$ has coinciding ultraviolet eigenvalues; only the rationality in $e^u$ of the inter-copy mixing functions $M_{ij}^{(\lambda)}(u)$ is known. The open problem is to generalize the CFT-based construction to tensor-product (fusion) decompositions with multiplicities of both kinds - repeated submodules and coinciding eigenvalues - validated against the explicitly known $\mathfrak{su}(3)_k$ adjoint weights.

Background

An interaction-round-a-face (IRF, also solid-on-solid or face) model is a two-dimensional statistical lattice model whose fluctuating variables are 'heights' sitting on lattice sites; each elementary square ('face') with corner heights $a,b,c,d$ contributes a Boltzmann weight $w(a,b,c,d;u)$ depending on a spectral parameter $u$. In the models considered here the heights are integral weights of the affine Lie algebra $\mathfrak{su}(3)_k$ (the central extension of the loop algebra of $\mathfrak{su}(3)$, graded by an integer level $k$), and a face is admissible when its corners are related by the weight set of a chosen representation - here the adjoint representation $(1,1)$ of finite $\mathfrak{su}(3)$, the eight-dimensional representation carried by the Lie algebra's own generators. Concretely, a pair of neighboring heights $(a,b)$ is admissible when $b=a+\hat e_i$ for one of the eight affine-extended adjoint weights $\hat e_1,\dots,\hat e_8$, and a face is admissible when all four of its neighboring pairs are. The model is exactly solvable when the weights satisfy the face-type (dynamical) Yang-Baxter equation, a star-triangle relation that makes the row-to-row transfer matrices (the matrices of statistical weights of a complete row of faces, whose eigenvalues generate the thermodynamics) commute. The archetype is the eight-vertex SOS model of Andrews-Baxter-Forrester, whose critical phases are described by rational conformal field theories (RCFTs) - two-dimensional quantum field theories with finitely many primary fields (the elementary superselection sectors of the theory, whose operator products are governed by the fusion rules, i.e. the decomposition laws of tensor products of the corresponding representations) - and Jimbo-Miwa-Okado extended the construction to restricted face models of higher-rank affine algebras.

This lattice-model/RCFT correspondence motivated a 'CFT approach' to the weights: Gepner (1992) showed that for RCFTs whose fusion rules are multiplicity-free, the ratios of the eigenvalue functions of the $R$-matrix (the operator on the tensor product of two representation spaces that intertwines the two factors and satisfies the Yang-Baxter equation) are fixed by the conformal dimensions $\Delta$ of the RCFT primaries (the scaling exponents of the primary fields), $f_a(u)/f_b(u)=\prod_{r=b}^{a-1}\sinh(i\xi_r-u/2)/\sinh(i\xi_r+u/2)$ with $\xi_r=\pi(\Delta_{r+1}-\Delta_r)/2$, and the projectors onto the irreducible tensor-product submodules, being $u$-independent, can be extracted from the ultraviolet limit $u\to\pm\infty$ (the large-$|u|$ regime in which subleading eigenvalues are exponentially suppressed, so the eigenvalue decomposition degenerates). Belavin-Gepner-Ramos Cabezas-Runov applied this ansatz to the restricted face models of $\mathfrak{su}(2)_k$ and $\mathfrak{su}(3)_k$ at fixed levels (J. Phys. A 2024). In the 2025 sequel they solved, by a different route, the unrestricted $\mathfrak{su}(3)_k$ face model with adjoint admissibility (unrestricted: the heights run over all integral weights, not just the level-$k$ integrable weights of a restricted RSOS-type model): they constructed the quantum $R$-matrix of the quantum group $U_q(\mathfrak{sl}(3))$ (the $q$-deformed enveloping algebra, with $q=e^{\pi i/(k+3)}$) in the adjoint representation via Jimbo's scheme - constructing $R$ as an intertwiner of the quantum-group action by solving linear intertwining equations - and obtained the face weights through the Vertex-IRF correspondence, which expresses face weights as matrix elements of a vertex-model $R$-matrix (a vertex model being a lattice model whose degrees of freedom live on the edges rather than the sites).

For the adjoint representation the tensor product decomposes (in Dynkin labels for the highest weights of finite $\mathfrak{su}(3)$) as $(1,1)\otimes(1,1)=(2,2)\oplus(3,0)\oplus(0,3)\oplus(1,1)_1\oplus(1,1)_2\oplus(0,0)$: the adjoint appears twice, making the decomposition non-unique, and separately the multiplicity-one pair $(3,0)/(0,3)$ has coinciding eigenvalues at $u\to\pm\infty$. Thus two kinds of multiplicity - repeated irreducible submodules and coinciding ultraviolet eigenvalues - both occur, and the projector extraction at the heart of the CFT approach fails. Within the multiplicity sector the $R$-matrix is only constrained to the form $\hat R=\sum M_{ij}^{(\lambda)}(u)P_{ij}^{(\lambda)}$ with mixing functions between the copies that are known (so far) only to be rational in $e^u$. Comparing their explicit solution with Gepner's formula, the authors found agreement for the multiplicity-one subrepresentations despite the multiplicities, and concluded that it 'might be possible' to generalize the CFT construction to multiplicity cases. They also observed that face weights vanish for some admissible configurations involving the two null affine weights $\hat e_4=(0,0,0)_1$ and $\hat e_5=(0,0,0)_2$ (the two copies of the zero weight of the adjoint weight system, called null weights in the paper), which they interpret as an effect of multiplicity but do not derive. The importance of the CFT-side construction is that it is the only known route to face models based on extended current algebras (algebras obtained by extending the chiral current algebra - the symmetry algebra generated by the conserved currents of the CFT - by its higher-spin W-currents), for which no quantum-group $R$-matrix exists (Baver-Gepner).

Problem Statement

Generalize the CFT-based (RCFT ultraviolet-limit) construction of trigonometric $R$-matrices and IRF Boltzmann weights to tensor-product (fusion) decompositions that carry multiplicities of both kinds: repeated irreducible submodules and coinciding ultraviolet eigenvalues. Concretely, the required results are: (1) a principled procedure that determines, from RCFT data (ultraviolet-limit spectrum, i.e. conformal dimensions together with the fusion ring and its multiplicities), the full solution of the face-type Yang-Baxter equation - requiring both the spectral-parameter mixing functions $M_{ij}^{(\lambda)}(u)$ acting between repeated copies of a highest-weight submodule and an equivalent replacement for the projector extraction that fails under degeneracy, where a treatment of the coinciding-eigenvalue case alone does not satisfy this requirement; and (2) validation of the procedure on the unrestricted $\mathfrak{su}(3)_k$ face model with adjoint admissibility conditions, reproducing the Boltzmann weights already constructed there from the quantum-group $R$-matrix via the Vertex-IRF correspondence (arXiv:2409.05637, eq. (3.7) with the $R$-matrix elements (3.14)), up to the standard overall normalization. The source work asserts only that such a generalization 'might be possible'. The procedure should be formulated at the natural generality of multiplicity-carrying fusion rings - the prerequisite for the authors' stated aim of treating extended current algebras, for which no quantum-group $R$-matrix exists - but constructing weights for any particular extended algebra is not itself required.

References

1. V. Belavin, D. Gepner, J. Ramos Cabezas, B. Runov, "On different approaches to IRF lattice models. Part II" (arXiv preprint title: "On different approaches to integrable lattice models II"), JHEP 02 (2025) 200, arXiv:2409.05637, DOI 10.1007/JHEP02(2025)200, https://arxiv.org/abs/2409.05637
2. V. Belavin, D. Gepner, J. Ramos Cabezas, B. Runov, "On different approaches to integrable lattice models", J. Phys. A 57 (2024) 155201, arXiv:2308.10329, DOI 10.1088/1751-8121/ad33dc, https://arxiv.org/abs/2308.10329
3. D. Gepner, "Foundations of Rational Quantum Field Theory, I", preprint (1992), arXiv:hep-th/9211100, https://arxiv.org/abs/hep-th/9211100
4. E. Baver, D. Gepner, "Fusion rules for extended current algebras", Mod. Phys. Lett. A 11 (1996) 1929-1946, arXiv:hep-th/9601114, DOI 10.1142/S0217732396001934, https://arxiv.org/abs/hep-th/9601114
5. E. Baver, "New solvable lattice models from conformal field theory", Phys. Lett. B 387 (1996) 502-512, arXiv:hep-th/9607055, DOI 10.1016/0370-2693(96)01090-8, https://arxiv.org/abs/hep-th/9607055
6. G. E. Andrews, R. J. Baxter, P. J. Forrester, "Eight-vertex SOS model and generalized Rogers-Ramanujan-type identities", J. Stat. Phys. 35 (1984) 193-266, DOI 10.1007/BF01014383, https://doi.org/10.1007/BF01014383
7. M. Jimbo, T. Miwa, M. Okado, "Local state probabilities of solvable lattice models: An $A_{n-1}^{(1)}$ family", Nucl. Phys. B 300 (1988) 74-108, DOI 10.1016/0550-3213(88)90587-1, https://doi.org/10.1016/0550-3213(88)90587-1