Exact solution of the Mangazeev–Bazhanov–Sergeev positive-weight 3D integrable lattice model: convergence of the partition function and rank-size duality of the Bethe Ansatz

The Mangazeev–Bazhanov–Sergeev model (J. Phys. A 46 (2013) 465206) is the first integrable lattice model on the three-dimensional cubic lattice with strictly non-negative Boltzmann weights. Its edge spins take infinitely many non-negative integer values, its q-hypergeometric vertex weights satisfy the tetrahedron equation, and its layer-to-layer transfer matrices form a two-parameter commuting family that decomposes into $U_q(\widehat{sl_N})$ sector transfer matrices related by a rank-size duality. The construction, however, left the model analytically unsolved: convergence of the full partition function $Z=\sum_{I\ge0}u^IZ_I$ was explicitly left open (the sum over the fixed-layer-spin index $I$ was said to require an additional study and possibly an $I$-dependent damping factor), and the implications of the rank-size duality for the Bethe Ansatz were left unexplored. This problem asks to complete the two analyses that the source itself left open: a rigorous convergence analysis of $Z$, and an explicit characterization of how the rank-size duality acts on the Bethe-Ansatz equations and eigenvalues of the sector transfer matrices.

Background

Exactly solvable two-dimensional (2D) lattice models, such as the six-vertex model, owe their solvability to the Yang–Baxter equation, an algebraic relation on vertex weights that guarantees that row-to-row transfer matrices (the matrices whose iterated matrix elements generate the partition function layer by layer) at different spectral parameters commute and can be diagonalized simultaneously, ultimately by the Bethe Ansatz, the standard technique that diagonalizes such transfer matrices by solving algebraic equations for auxiliary complex roots. The three-dimensional (3D) analogue is the tetrahedron equation, introduced by Zamolodchikov in 1981 and established as an integrability statement by Baxter in 1983 for the Zamolodchikov model: it encodes the commutativity of layer-to-layer transfer matrices of models defined on the cubic lattice. All known solutions, such as the finite-state model of Bazhanov and Baxter [1992] and the infinite-state $q$-oscillator model of Bazhanov and Sergeev [2006], involve non-positive (complex or sign-alternating) weights, so they cannot be interpreted as Boltzmann weights, the local statistical weights of a physical model.

Mangazeev, Bazhanov, and Sergeev [2013] constructed the first solution of the tetrahedron equation with strictly non-negative matrix elements. In their model the spin variables live on the edges of the cubic lattice and take infinitely many non-negative integer values. At each vertex the six surrounding edge spins obey two local conservation laws of ice type (constraints of the form $i+j=i'+j'$ and $j+k=j'+k'$ on the spins meeting at a vertex), and the vertex weight is an explicit $q$-hypergeometric expression (a basic hypergeometric series, a series built from $q$-shifted factorials) depending on a parameter $0<q<1$ and eight continuous parameters. Because the spins are unbounded, the partition function requires regularization: the paper introduces restricted partition functions $Z_I$ in which the total spin per horizontal layer is fixed to a non-negative integer $I$, weighted by a Gaussian-type damping factor $q^{\mu U}$, where $U$ is a positive quadratic form in the spin sums and $\mu$ a real parameter, and establishes convergence of the internal spin sums for $\mu>1$ (any $v$, $w$) or for $\mu=1$ with $v<1$, $w<1$. The full partition function is then defined as $Z=\sum_{I\ge0}u^IZ_I$ with $u<1$ (Eq. (58) of the paper), and the authors state that the convergence of this $I$-sum requires an additional study and could require an additional $I$-dependent damping factor.

The algebraic structure is rich. The composite vertex weight decomposes into an infinite direct sum of $R$-matrices (the building blocks of the Yang–Baxter equation) of the quantum affine algebra $U_q(\widehat{sl_N})$, the $q$-deformation of the affine extension of the Lie algebra $sl_N$, where $N$ is the lattice width, acting in symmetric-tensor representations (the representations obtained by symmetrizing tensor powers of the fundamental one, labeled by the symmetrization level). Consequently the layer-to-layer transfer matrix $\boldsymbol{T}(v,w)$ decomposes (Eq. (84)) as a direct sum over chains of such symmetric representations in the quantum (vertical) space, the direct sum running over row-sum sequences $I^{(N)}=\{\mathcal{I}^{(N)}_1,\dots,\mathcal{I}^{(N)}_M\}$ (Eqs. (46), (83)), summed over the auxiliary representation index $J$ with a horizontal fugacity $v^J$ (a weighting parameter for the auxiliary-sector index). Moreover $Z_I=\langle\Psi_I|\boldsymbol{T}(v,w)^L|\Psi_I\rangle$, a matrix element of the $L$-th power of the transfer matrix in the fixed-layer-spin sector, where $I=\sum_m\mathcal{I}^{(N)}_m$ is the layer total spin of Eq. (44) and $|\Psi_I\rangle$ is the flat sum over all layer configurations of total spin $I$ (Eq. (62)). By a symmetry of the weights, the same transfer matrix admits a second decomposition (Eq. (85)) in which the rank $N$ of the algebra is exchanged with the length $M$ of the chain, the spectral parameter with the fugacity, and the inhomogeneities (site-dependent parameter shifts) with the horizontal fields, up to transposition in the quantum space. The authors call this rank-size duality, note that it parallels one found earlier in the Bazhanov–Baxter [1992] and Bazhanov–Sergeev [2006] models, and state that it would be extremely interesting to understand its implications for the algebraic and analytic structure of the Bethe Ansatz. For each fixed sector the Bethe Ansatz equations are those of a $U_q(\widehat{sl_N})$ chain in symmetric representations, but no Bethe Ansatz is known for the combined object including the fugacity sum.

The authors promised a follow-up treating the $N=2$ case, which appeared as Mangazeev [2014a,b]: using two-layer projections of the 3D $R$-matrix, these works derived new expressions for higher-spin $U_q(\widehat{sl_2})$ $R$-matrices, for $Q$-operators (auxiliary transfer matrices whose eigenvalues encode the Bethe roots), and for their functional relations. Bošnjak and Mangazeev [2016] extended the construction to $R$-matrices for symmetric tensor representations of $U_q(\widehat{sl_n})$ for all $n$. These deliver the 2D sector building blocks but not the 3D model itself. Later work remains adjacent: Bazhanov and Sergeev [2023; 2024] constructed and solved new positive-weight 2D models; Wang and Hazzard [2023] exactly solved a positive-weight 3D Ising-type model whose interactions decouple into free-fermion-like (reducible to non-interacting fermions) 2D sectors, remarking that no physical 3D model with a genuinely 3D phase transition has been exactly solved; Iwao, Motegi, and Ohkawa [2025; 2026] computed partition functions of the older non-positive Bazhanov–Sergeev solution as Schur-type symmetric functions; and Bazhanov, Kashaev, Mangazeev, and Sergeev [2025] constructed new 3D models with quantum-dilogarithm weights (weights built from the quantum dilogarithm, a special function of $q$-analysis) and continuous spins and computed their partition function per site, but did not analyze the 2013 discrete model. No work has addressed the convergence of the 2013 model's $I$-sum or the action of its rank-size duality on the Bethe Ansatz.

Problem Statement

The overarching objective is to complete the two analyses that the source itself left open for the Mangazeev-Bazhanov-Sergeev positive-weight 3D lattice model of arXiv:1308.4773 (J. Phys. A 46 (2013) 465206): to turn its formally constructed partition function into a well-defined statistical-mechanical system, and to understand the structural consequences of its rank-size duality. Two linked questions, both posed explicitly in the source:

(1) Convergence of the full partition function. Fix $0<q<1$ and the model exactly as defined in the source (edge spins, ice-type constraints, weights, periodic boundary conditions, the $q^{\mu U}$ regularization of $Z_I$). Determine the domain of the parameters $(u,v,w,\mu)$ — with the lattice sizes $L,M,N$ fixed, and restricted to the region where each restricted partition function $Z_I$ itself converges (Eq. (57): $\mu>1$ at any $v$, $w$, or $\mu=1$ with $v<1$, $w<1$) — for which the full partition function $Z=\sum_{I\ge0}u^IZ_I$ (Eq. (58)) converges. In parameter regimes where the $I$-sum does not converge, decide whether an additional explicit $I$-dependent damping factor, the possibility the source names after Eq. (58), yields a convergent partition function; admissible factors are those of the source's Gaussian type — $q^{\lambda I(I+1)}$ with $\lambda>0$, the $q^{\mu U}$ regularization extended to the layer-spin index $I$ — and for these determine the range of $\lambda$ (as a function of $v,w,\mu$) for which the resulting partition function converges, and prove convergence in that range.

(2) Rank-size duality and the Bethe Ansatz. Give the explicit characterization of how the rank-size duality between Eqs. (84) and (85) acts on the algebraic and analytic structure of the Bethe Ansatz of the $U_q(\widehat{sl_N})$ symmetric-tensor sector transfer matrices $\mathbf{T}^{sl_N}_{I^{(N)},J}$ appearing in these decompositions: give the transformation of the Bethe-Ansatz equations, Bethe roots, and eigenvalues exchanging rank with chain length and spectral parameter with fugacity. Here the direct sum in Eq. (84) runs over row-sum sequences $I^{(N)}=\{\mathcal{I}^{(N)}_1,\dots,\mathcal{I}^{(N)}_M\}$ (Eqs. (46), (83)), while the fixed-layer-spin index $I$ of Eqs. (44), (55), (58) and (62) is the layer total spin $I=\sum_m\mathcal{I}^{(N)}_m$.

The two parts concern one and the same object: the partition function whose convergence is questioned in (1) is built from matrix elements $Z_I=\langle\Psi_I|\boldsymbol{T}(v,w)^L|\Psi_I\rangle$ of the transfer matrix whose sector decompositions (84)–(85) carry the rank-size duality studied in (2). Partial progress must not silently restrict the model, for example to a finite spin cutoff, without a controlled limit.

References

1. V. V. Mangazeev, V. V. Bazhanov, S. M. Sergeev, 'An integrable 3D lattice model with positive Boltzmann weights', J. Phys. A: Math. Theor. 46 (2013) 465206. DOI: 10.1088/1751-8113/46/46/465206, arXiv:1308.4773. URL: https://doi.org/10.1088/1751-8113/46/46/465206
2. A. B. Zamolodchikov, 'Tetrahedron equations and the relativistic S-matrix of straight strings in 2+1-dimensions', Commun. Math. Phys. 79 (1981) 489-505. DOI: 10.1007/BF01209309. URL: https://doi.org/10.1007/BF01209309
3. R. J. Baxter, 'On Zamolodchikov's solution of the tetrahedron equations', Commun. Math. Phys. 88 (1983) 185-205. DOI: 10.1007/BF01209476. URL: https://doi.org/10.1007/BF01209476
4. V. V. Bazhanov, R. J. Baxter, 'New solvable lattice models in three dimensions', J. Stat. Phys. 69 (1992) 453-485. DOI: 10.1007/BF01050423. URL: https://doi.org/10.1007/BF01050423
5. V. V. Bazhanov, S. M. Sergeev, 'Zamolodchikov's tetrahedron equation and hidden structure of quantum groups', J. Phys. A: Math. Gen. 39 (2006) 3295-3310. DOI: 10.1088/0305-4470/39/13/009, arXiv:hep-th/0509181. URL: https://doi.org/10.1088/0305-4470/39/13/009
6. V. V. Mangazeev, 'On the Yang-Baxter equation for the six-vertex model', Nucl. Phys. B 882 (2014) 70-96. DOI: 10.1016/j.nuclphysb.2014.02.019, arXiv:1401.6494. URL: https://doi.org/10.1016/j.nuclphysb.2014.02.019
7. V. V. Mangazeev, 'Q-operators in the six-vertex model', Nucl. Phys. B 886 (2014) 166-184. DOI: 10.1016/j.nuclphysb.2014.06.024, arXiv:1406.0662. URL: https://doi.org/10.1016/j.nuclphysb.2014.06.024
8. G. Bošnjak, V. V. Mangazeev, 'Construction of $R$-matrices for symmetric tensor representations related to $U_q(\widehat{sl_n})$', J. Phys. A: Math. Theor. 49 (2016) 495204. DOI: 10.1088/1751-8113-49-49-495204, arXiv:1607.07968. URL: https://doi.org/10.1088/1751-8113-49-49-495204
9. V. V. Bazhanov, S. M. Sergeev, 'An Ising-type formulation of the six-vertex model', Nucl. Phys. B 986 (2023) 116055. DOI: 10.1016/j.nuclphysb.2022.116055, arXiv:2205.10708. URL: https://doi.org/10.1016/j.nuclphysb.2022.116055
10. V. V. Bazhanov, S. M. Sergeev, 'A distant descendant of the six-vertex model', Nucl. Phys. B 1004 (2024) 116558. DOI: 10.1016/j.nuclphysb.2024.116558, arXiv:2310.08427. URL: https://doi.org/10.1016/j.nuclphysb.2024.116558
11. V. V. Bazhanov, R. M. Kashaev, V. V. Mangazeev, S. M. Sergeev, 'Quantum Dilogarithms and New Integrable Lattice Models in Three Dimensions', arXiv:2512.23338 (2025). URL: https://arxiv.org/abs/2512.23338
12. Z. Wang, K. R. A. Hazzard, 'Topological correlations in three-dimensional classical Ising models: An exact solution with a continuous phase transition', Phys. Rev. Research 5 (2023) 013086. DOI: 10.1103/PhysRevResearch.5.013086, arXiv:2202.11303. URL: https://doi.org/10.1103/PhysRevResearch.5.013086
13. S. Iwao, K. Motegi, R. Ohkawa, 'Tetrahedron equation and Schur functions', J. Phys. A: Math. Theor. 58 (2025) 015201. DOI: 10.1088/1751-8121/ad9675, arXiv:2405.10011. URL: https://doi.org/10.1088/1751-8121/ad9675
14. S. Iwao, K. Motegi, R. Ohkawa, 'Tetrahedral $L$-operators, tensor Schur polynomials and $q$-deformed loop elementary symmetric functions', arXiv:2604.22141 (2026). URL: https://arxiv.org/abs/2604.22141
15. A. Gorsky, M. Vasilyev, A. Zotov, 'Dualities in quantum integrable many-body systems and integrable probabilities. Part I', JHEP 04 (2022) 159. DOI: 10.1007/JHEP04(2022)159, arXiv:2109.05562. URL: https://doi.org/10.1007/JHEP04(2022)159
16. P. Gavrylenko, M. Semenyakin, Y. Zenkevich, 'Solution of tetrahedron equation and cluster algebras', JHEP 05 (2021) 103. DOI: 10.1007/JHEP05(2021)103, arXiv:2010.15871. URL: https://doi.org/10.1007/JHEP05(2021)103