Constructing quantum R-matrix blocks in tensor-product multiplicity sectors and rederiving them from rational CFT data

Exactly solvable interaction-round-a-face (IRF) lattice models can be built either from the fusion rules of a rational conformal field theory (RCFT) or from the quantum group R-matrix via the vertex-IRF correspondence, but both routes currently fail when a tensor-product decomposition contains an irreducible representation with multiplicity greater than one: the RCFT eigenvalue formulae only cover multiplicity-one sectors, and Jimbo's intertwining equations for the R-matrix become an underdetermined linear system for the matrix-valued coefficient functions that describe the spectral-parameter dependence in the multiplicity block. Belavin, Gepner, Ramos Cabezas and Runov (JHEP 02 (2025) 200) solved the su(3) adjoint case by brute force, extracted the explicit 2x2 multiplicity block, showed that no spectral-parameter-independent basis diagonalizes it, and posed the open problem: give a constructive, basis-independent procedure producing closed-form expressions for these blocks in general, and in particular rederive them from purely UV (RCFT) data without solving Jimbo's full linear system, as a step towards a general method for IRF models based on extended current algebras.

Background

A two-dimensional lattice model is exactly solvable when its local Boltzmann weights (BWs), the amplitudes assigned to each elementary face of the lattice, satisfy the Yang-Baxter equation (YBE), the algebraic consistency condition guaranteeing commuting transfer matrices and, at criticality, a direct connection to conformal field theory (CFT). In an interaction-round-a-face (IRF) model the fluctuating variables sit on lattice vertices and the BWs are supported on faces, subject to admissibility rules dictated by representation-theoretic data. The archetype is the Andrews-Baxter-Forrester eight-vertex solid-on-solid model, whose critical behavior realizes the unitary minimal CFT series; Pasquier identified the general mechanism by which affine Lie algebra (current algebra) representation theory, i.e. the spectrum of the associated Wess-Zumino-Witten-type rational CFT (RCFT), organizes such models, and Jimbo, Miwa and Okado gave a systematic construction of solvable face models from affine Lie algebras.

Two constructive approaches are available. In the CFT approach, initiated by Gepner, the BWs are fixed, up to normalization, by the fusion rules (the multiplication table of the representation ring, i.e. the rules determining which sectors combine into which in operator products) and conformal dimensions of an RCFT: the ratios of the scalar eigenvalue functions of the R-matrix obey a product formula over intermediate conformal dimensions. In the vertex-IRF correspondence approach, the BWs are expressed through the quantum group R-matrix $R(u)$ of $U_q(\mathfrak{sl}(n))$ , a trigonometric function of the spectral parameter $u$ satisfying the YBE; Jimbo showed that $\hat R(u)$ is characterized, up to normalization, by a finite linear system of intertwining relations with the quantum group generators. The R-matrix admits a spectral decomposition $\hat R(u)=\sum_j f_{\lambda_j}(u)\,P^{(\lambda_j)}$ over the irreducible submodules of the tensor product $V\otimes V=\bigoplus_j V_{\lambda_j}$ , where $P^{(\lambda_j)}$ are projectors and $f_{\lambda_j}(u)$ scalar eigenvalue functions.

Two kinds of multiplicity can occur in this decomposition, and both occur for $U_q(\mathfrak{sl}(3))$ at generic level $k$ . First, an irreducible representation can appear more than once in the tensor product (representation multiplicity). Second, the ultraviolet (UV, i.e. large- $u$ ) limits of distinct eigenvalues can coincide; the authors note this second kind is easier to handle, since the corresponding eigenspace projectors remain well defined. The first kind is the obstruction: when a representation $\lambda_m$ occurs with multiplicity $\#\lambda_m>1$ , the spectral decomposition must be generalized to $\hat R(u)=\sum_{m,i,j}M_{ij}^{(\lambda_m)}(u)\,P_{ij}^{(\lambda_m)}$ , where $P_{ii}^{(\lambda_m)}$ projects onto the $i$ -th copy, the off-diagonal $P_{ij}^{(\lambda_m)}$ map between the copies (they obey the matrix-unit algebra $P_{ij}^{(\lambda_m)}P_{kl}^{(\lambda_m)}=\delta_{jk}P_{il}^{(\lambda_m)}$ ), and the matrix-valued functions $M_{ij}^{(\lambda_m)}(u)$ carry the spectral-parameter dependence. Jimbo's relations then turn into an underdetermined linear system for these functions; only their rationality in $e^u$ is known in general.

The state of the art is the following. Part I of the series (Belavin, Gepner, Ramos Cabezas, Runov, 2024) solved the multiplicity-free cases, such as $\mathfrak{su}(3)_k$ at level $k=2$ with adjoint admissibility, by the CFT approach, and showed that for $k\ge 3$ multiplicities in the tensor products make that approach inapplicable, leaving these cases explicitly as open questions. Part II (2025) took the vertex-IRF route for the unrestricted $\mathfrak{su}(3)_k$ IRF model with adjoint admissibility: it constructed the full $64\times 64$ R-matrix of $U_q(\mathfrak{sl}(3))$ in the adjoint representation from Jimbo's system (a demanding computation in its own right), verified that the multiplicity-one eigenvalue functions agree with Gepner's RCFT product formula, and extracted the explicit multiplicity-two block: the tensor product decomposes as $(1,1)\otimes(1,1)=(2,2)\oplus(3,0)\oplus(0,3)\oplus(1,1)_1\oplus(1,1)_2\oplus(0,0)$ , and the block $M^{(1,1)}(u)$ is a $2\times 2$ matrix of explicit rational functions of $e^u$ and $q=e^{\pi i/(k+3)}$ . Two structural facts emerged: no basis independent of $u$ diagonalizes $M^{(1,1)}(u)$ , so scalar-function techniques cannot capture multiplicity sectors; and the agreement of the multiplicity-one eigenvalues with the RCFT formula suggests that the CFT construction might generalize to cases with multiplicities of both kinds, a speculation stated in the preprint version of the paper whose published conclusions pose the task directly: one needs to find general expressions, analogous to the eigenvalue-ratio product formula, for the matrices $M^{(\lambda_m)}$ , and the authors state they intend to address this problem in future works. The multiplicity-two block was obtained only by solving Jimbo's full system, not from CFT data.

Problem Statement

The problem, as posed by Belavin-Gepner-Ramos Cabezas-Runov (JHEP 02 (2025) 200) after their solution of the $\mathfrak{su}(3)_k$ adjoint IRF model, is to remove the representation-multiplicity obstruction in constructing R-matrices (and hence IRF Boltzmann weights) for exactly solvable lattice models, with one overarching objective: a constructive method that produces the R-matrix in sectors where an irreducible representation $\lambda_m$ occurs with multiplicity $\#\lambda_m>1$ in the tensor product $V\otimes V$ of two copies of the representation $V$ , ideally from rational CFT data alone. Concretely, for the generalized spectral decomposition

\hat R(u)=\sum_{m}\sum_{i,j=1}^{\#\lambda_m} M_{ij}^{(\lambda_m)}(u)\,P_{ij}^{(\lambda_m)},

where $P_{ii}^{(\lambda_m)}$ projects onto the $i$ -th copy of $V_{\lambda_m}$ and the off-diagonal $P_{ij}^{(\lambda_m)}$ map between copies (matrix units on the multiplicity space), the requested results are: (1) a constructive, basis-independent procedure to derive the off-diagonal intertwiners $P_{ij}^{(\lambda_m)}$ and closed-form general expressions for the matrix-valued coefficient functions $M_{ij}^{(\lambda_m)}(u)$ , generalizing the multiplicity-one eigenvalue product formula to multiplicity sectors; (2) specifically, a rederivation of these blocks from purely UV (RCFT) data, that is from the conformal dimensions, fusion rules and symmetry of the associated rational CFT, without solving Jimbo's full linear system, with the published $\mathfrak{su}(3)$ -adjoint multiplicity-two block $M^{(1,1)}(u)$ (equations (4.28)-(4.31) of the published JHEP version of the source; absent from arXiv v1) as the mandatory benchmark; and (3) as a further direction stated in the source, the extension of this construction towards a general method producing R-matrices and face weights in all cases with multiplicities, including models based on extended current algebras. The completion boundary is fixed as follows: (1) and (2) are the mandatory deliverables of the task, and the benchmark block of (2) is a required test case of the CFT/UV rederivation rather than a self-contained target, so reproducing the published $M^{(1,1)}(u)$ on its own does not resolve the task; (3) is recorded as a further direction of the source and is not a required deliverable. The required scope of (1) is fixed as follows: the multiplicity-block construction is required for tensor products $V\otimes V$ whose decomposition contains a representation with multiplicity greater than one, with the $\mathfrak{su}(3)_k$ adjoint case as the worked instance; only the extension named in (3) --- producing face weights and covering models based on extended current algebras --- lies outside the required scope.

References

V. Belavin, D. Gepner, J. Ramos Cabezas, B. Runov, 'On different approaches to IRF lattice models. Part II' (published title; the arXiv version is titled 'On different approaches to integrable lattice models II'), JHEP 02 (2025) 200, DOI 10.1007/JHEP02(2025)200, arXiv:2409.05637, https://arxiv.org/abs/2409.05637
V. Belavin, D. Gepner, J. Ramos Cabezas, B. Runov, 'On different approaches to integrable lattice models', J. Phys. A 57 (2024) 155201, DOI 10.1088/1751-8121/ad33dc, arXiv:2308.10329, https://arxiv.org/abs/2308.10329
D. Gepner, 'Foundations of Rational Quantum Field Theory, I', arXiv:hep-th/9211100, https://arxiv.org/abs/hep-th/9211100
M. Jimbo, 'Quantum R Matrix for the Generalized Toda System', Commun. Math. Phys. 102 (1986) 537-547, DOI 10.1007/BF01221646, https://doi.org/10.1007/BF01221646
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
V. Pasquier, 'Etiology of IRF Models', Commun. Math. Phys. 118 (1988) 355-364, DOI 10.1007/BF01466721, https://doi.org/10.1007/BF01466721
M. Jimbo, T. Miwa, M. Okado, 'An $A_{n-1}^{(1)}$ family of solvable lattice models', Mod. Phys. Lett. B 1 (1987) 73-79, DOI 10.1142/S0217984987000119, https://doi.org/10.1142/S0217984987000119
E. Baver, D. Gepner, 'Fusion rules for extended current algebras', Mod. Phys. Lett. A 11 (1996) 1929-1945, DOI 10.1142/S0217732396001934, arXiv:hep-th/9601114, https://arxiv.org/abs/hep-th/9601114