Analytic foundations of the new Q-matrix for the eight-vertex model at roots of unity: regularity, functional relations, and the $t\neq 0$ similarity structure

For the eight-vertex model at roots of unity of the crossing parameter, $\eta=2mK/L$ with odd $L$ and an even number of sites, Baxter's original 1972 Q-matrix construction fails, and Fabricius (2007) constructed a new Q-matrix $\hat Q(v,t)$ , depending on a spectral parameter $v$ and a free parameter $t$ , whose regularity, functional relation, and commutation properties were verified only numerically. The open problem is to give analytic proofs of these properties: non-singularity of the right-building matrix $\hat Q_R(v,t)$ at a reference point, an unconditional proof of the conjectured Q-functional relation (Roan verified a closely related functional relation for his own Q-operator — under a different involution and at specific values of the free parameter — only under an unproven non-singularity hypothesis), an explicit construction of the similarity $A$ relating the left- and right-normalized Q-operators for $t\neq 0$ , a classification of the resulting non-Abelian symmetry action on degenerate transfer-matrix eigenspaces, and rigorous quasiperiodicity-based singularity criteria. Later literature through 2024 treats these questions as unresolved and builds conditional results on them.

Background

The eight-vertex model on a square lattice is solved by the transfer matrix $T(v)$ , a one-parameter commuting family of matrices built from local Boltzmann weights that are theta functions of the spectral parameter $v$ ; the anisotropy of the weights is encoded in the crossing parameter $\eta$ , and the theta functions live on a torus with real half-period $K$ and imaginary half-period $iK'$ (so $H(v)$ , $\Theta(v)$ are the Jacobi theta functions and $K$ , $K'$ the complete elliptic integrals). Via Baxter's mapping the model is equivalent to the one-dimensional anisotropic Heisenberg XYZ quantum spin chain with spin- $1/2$ sites, so statements about one transfer directly to the other. Baxter's solution introduces a commuting auxiliary matrix, the Q-matrix, satisfying the TQ equation $T(v)Q(v)=[\rho h(v-\eta)]^N Q(v+2\eta)+[\rho h(v+\eta)]^N Q(v-2\eta)$ with $h(v)=H(v)\Theta(v)$ and $N$ the number of sites; it yields the Bethe ansatz (the closed system of algebraic equations whose solutions determine the eigenvalues) and the full spectrum. Baxter gave two constructions: $Q_{72}$ , valid at the 'roots of unity' $2L\eta=2m_1K+im_2K'$ with integer $m_1,m_2,L$ , which for all $N$ exists only when $m_1$ and $m_2$ are not both even (Ann. Phys. 70 (1972) 193), and $Q_{73}$ , valid for generic $\eta$ but only for even $N$ (Ann. Phys. 76 (1973) 1). For generic $\eta$ the transfer-matrix spectrum is non-degenerate, but at rational $\eta/K$ the spectrum becomes highly degenerate and several inequivalent Q-matrices exist. Fabricius and McCoy conjectured (J. Stat. Phys. 111 (2003) 323) that at roots of unity $Q_{72}(v)$ satisfies an additional Q-functional relation,

e^{-N\pi i v/2K}\,Q(v-iK') \;=\; A\sum_{l=0}^{L-1} h^N\bigl(v-(2l+1)\eta\bigr)\,\frac{Q(v)}{Q(v-2l\eta)\,Q(v-2(l+1)\eta)},

with $A$ a $v$ -independent normalizing matrix commuting with $Q$ ; it implies that the dimensions of degenerate eigenspaces of $T$ are powers of two. A proof exists only for $L=2$ (Fabricius–McCoy, Publ. RIMS 40 (2004) 905). Crucially, $Q_{72}$ does not exist when $\eta=2mK/L$ with $L$ odd — the real-root-of-unity case $m_2=0$ , $m_1=2m$ of that condition, in which $m_1,m_2$ are both even — precisely a case needed to complete the functional-relation picture. Fabricius (J. Phys. A 40 (2007) 4075) closed this gap constructively: for even $N$ he built a right Q-matrix $\hat Q_R(v,t)$ from local $L$ -valued auxiliary intertwining matrices — matrices carrying an $L$ -dimensional auxiliary index that intertwine the transfer matrix with the auxiliary space — depending on $v$ and a free parameter $t$ , together with $\hat Q_L(v,t)$ and $\hat Q(v,t)=\hat Q_L^{-1}(v_0)\hat Q_L(v,t)$ , which satisfies the TQ equation and commutes with $T$ for $t=0$ . He also showed that a well-defined Q-operator requires $\hat Q_R$ to be invertible at some reference point $v_0$ ('regularity'), and that quasiperiodicity under $v\mapsto v+iK'$ forces singularity for certain parity classes of the integers $m_1,m_2,L,N$ — but only plausibility arguments were given. The discrete spin-reversal symmetries of the model, $S$ and $R$ , act on the degenerate $T$ -subspaces; $\hat Q(v,t\neq 0)$ fails to commute with $S$ , with $R$ , and with itself at different spectral parameters, so it acts as a non-Abelian symmetry on those subspaces. All load-bearing properties — regularity, the functional relation, the similarity between $\hat Q^{(L)}$ and $\hat Q^{(R)}$ , and the non-Abelian action — were established numerically only, for system sizes large enough to exhibit the transfer-matrix degeneracies.

Problem Statement

For the eight-vertex model with an even number of sites $N$ and crossing parameter $\eta=2mK/L$ with $L$ odd (the root-of-unity regime where Baxter's 1972 Q-matrix does not exist), establish analytically — replacing the numerical evidence and the plausibility arguments of Fabricius (2007) — the rigorous theory of the new Q-matrix $\hat Q(v,t)$ built from the right and left auxiliary matrices $\hat Q_R(v,t)$ , $\hat Q_L(v,t)$ . The common objective is that every property of $\hat Q$ on which the degeneracy structure of the transfer matrix rests be proven, not numerically supported. Specifically: (1) Prove that $\hat Q_R(v,t)$ is regular, i.e. non-singular at some reference point $v_0$ , for the parity cases where numerics indicate regularity (including $t=0$ and generic $t$ ) — Roan's formulation relies on a closely related but narrower hypothesis, the non-singularity of an auxiliary matrix $M(v_0)$ at specific values of the free parameter — so that $\hat Q(v,t)$ is well defined; and establish rigorously, with full classification of the parity cases of $m$ , $L$ , $N$ , the complementary criteria under which one-step $iK'$ quasiperiodicity forces $\hat Q_R$ to be singular. (2) Give an unconditional analytic proof of the Q-functional relation (the $N$ th QQ-relation / equation (3) conjectured by Fabricius–McCoy), currently verified numerically; Roan proved a related $N$ th QQ-relation for his own Q-operator — which coincides with $\hat Q$ only at specific values of the free parameter and whose involution differs from $v\mapsto v-iK'$ — conditional on the unproven non-singularity hypothesis for his $M(v_0)$ . (3) Construct or analytically characterize the similarity $A$ with $\hat Q^{(L)}(v,t)=A\,\hat Q^{(R)}(v,t)\,A^{-1}$ for $t\neq 0$ (this similarity is distinct from the $v$ -independent normalizing matrix $A$ of the Q-functional relation displayed in the Background) and explain its algebraic origin, replacing Baxter's commutation relation that fails for $t\neq 0$ . (4) Classify the non-Abelian symmetry algebra generated by the operators $\hat Q(v,t\neq 0)$ — which do not commute among themselves at different spectral parameters nor with the discrete symmetries $R$ and $S$ — acting on the degenerate eigenspaces of $T$ , and derive its consequences for the spectrum, including the power-of-two dimension of degenerate eigenspaces. Deliverables (1)–(4) are jointly linked: (1) is the load-bearing hypothesis (Roan's conditional result shows that functional relations of this kind can reduce to a non-singularity hypothesis, but the reduction of (2) for $\hat Q(v,t)$ at generic $t$ to (1) has not been established), and (3)–(4) describe the symmetry content of the same degenerate subspaces that the functional relation (2) characterizes, and the power-of-two dimension clause in (4) records the spectrum consequence that the Background already attaches to the Q-functional relation, to be re-derived for the new Q-matrix.

References

R. J. Baxter, 'Partition function of the Eight-Vertex lattice model', Annals of Physics 70 (1972) 193–228, DOI 10.1016/0003-4916(72)90335-1, https://doi.org/10.1016/0003-4916(72)90335-1
R. J. Baxter, 'Eight-vertex model in lattice statistics and one-dimensional anisotropic Heisenberg chain. I. Some fundamental eigenvectors', Annals of Physics 76 (1973) 1–24, DOI 10.1016/0003-4916(73)90439-9, https://doi.org/10.1016/0003-4916(73)90439-9
K. Fabricius and B. M. McCoy, 'New Developments in the Eight Vertex Model', Journal of Statistical Physics 111 (2003) 323–337, DOI 10.1023/A:1022213209641, https://doi.org/10.1023/A:1022213209641
K. Fabricius and B. M. McCoy, 'Functional Equations and Fusion Matrices for the Eight-Vertex Model', Publications of the Research Institute for Mathematical Sciences 40 (2004) 905–932, DOI 10.2977/prims/1145475496, https://doi.org/10.2977/prims/1145475496
K. Fabricius, 'A new Q-matrix in the eight-vertex model', Journal of Physics A: Mathematical and Theoretical 40 (2007) 4075–4086, DOI 10.1088/1751-8113-40-15-002, arXiv:cond-mat/0610481, https://arxiv.org/abs/cond-mat/0610481
S.-S. Roan, 'The Q-operator for root-of-unity symmetry in the six-vertex model', Journal of Physics A: Mathematical and General 39 (2006) 12303–12325, DOI 10.1088/0305-4470-39/40/002, arXiv:cond-mat/0602375, https://arxiv.org/abs/cond-mat/0602375
S.-S. Roan, 'The Q-operator and functional relations of the eight-vertex model at root-of-unity $\eta=2mK/N$ for odd $N$ ', Journal of Physics A: Mathematical and Theoretical 40 (2007) 11019–11044, DOI 10.1088/1751-8113-40-36-004, arXiv:cond-mat/0611316, https://arxiv.org/abs/cond-mat/0611316
K. Fabricius and B. M. McCoy, 'The TQ equation of the eight-vertex model for complex elliptic roots of unity', Journal of Physics A: Mathematical and Theoretical 40 (2007) 14893–14926, DOI 10.1088/1751-8113-40-50-001, https://doi.org/10.1088/1751-8113-40-50/001
K. Fabricius and B. M. McCoy, 'New Q Matrices and Their Functional Equations for the Eight Vertex Model at Elliptic Roots of Unity', Journal of Statistical Physics 134 (2009) 643–668, DOI 10.1007/s10955-009-9692-6, arXiv:0809.2802, https://arxiv.org/abs/0809.2802
K. Fabricius, 'Properties of the string operator in the eight-vertex model', Journal of Physics A: Mathematical and Theoretical 44 (2011) 135001, DOI 10.1088/1751-8113-44-13-135001, arXiv:1011.4455, https://arxiv.org/abs/1011.4455
T. Takebe, 'Q-Operators for Higher Spin Eight Vertex Models with an Even Number of Sites', Letters in Mathematical Physics 106 (2016) 319–340, DOI 10.1007/s11005-015-0813-7, arXiv:1509.08616, https://arxiv.org/abs/1509.08616
T. Takebe, 'Q-operators for higher spin eight vertex models with a rational anisotropy parameter', Letters in Mathematical Physics 109 (2019) 1867–1890, DOI 10.1007/s11005-019-01179-7, arXiv:1810.12969, https://arxiv.org/abs/1810.12969
C. Hagendorf and H. Rosengren, 'Nearest-Neighbour Correlation Functions for the Supersymmetric XYZ Spin Chain and Painlevé VI', Communications in Mathematical Physics 405 (2024) 97, DOI 10.1007/s00220-024-04977-w, arXiv:2208.13533, https://arxiv.org/abs/2208.13533
X. Zhang, A. Klümper, and V. Popkov, 'Pedestrian's way to Baxter's Bethe ansatz for the periodic XYZ chain', Physical Review B 109 (2024) 115411, DOI 10.1103/PhysRevB.109.115411, arXiv:2312.00161, https://arxiv.org/abs/2312.00161