The square-lattice eight-vertex model at crossing parameter η=π/3\eta=\pi/3 — the supersymmetric or 'combinatorial' point, where the vertex weights satisfy (a2+ab)(b2+ab)=(c2+ab)(d2+ab)(a^2+ab)(b^2+ab)=(c^2+ab)(d^2+ab) — has a transfer matrix with a remarkably simple doubly degenerate eigenvalue Θn=(a+b)2n+1\Theta_n=(a+b)^{2n+1} for an odd number L=2n+1L=2n+1 of vertical lines with periodic boundary conditions; this 2001 conjecture of Stroganov was proven by Hagendorf and Liénardy in 2018 using the lattice supersymmetry of a related XYZ spin chain. Razumov and Stroganov (2010) conjectured the inhomogeneous generalisation: with arbitrary column inhomogeneities u1,…,u2n+1u_1,\dots,u_{2n+1}, the transfer matrix has the doubly degenerate eigenvalue Θn(u∣u1,…,u2n+1)=∏ir(ui−u)\Theta_n(u|u_1,\dots,u_{2n+1})=\prod_i r(u_i-u) with r(u)=a(u)+b(u)r(u)=a(u)+b(u), reducing to (a+b)2n+1(a+b)^{2n+1} in the homogeneous limit. Zinn-Justin (2013) strengthened this into a full characterisation of the eigenspace (eigenvector entries that are theta functions of degree 2n2n with prescribed quasi-periodicity). This conjecture is the load-bearing assumption of the sum-rule program for the supersymmetric eight-vertex model, yet it remains unproven: the homogeneous proof relies on a supercharge argument with no known inhomogeneous extension, and the standard QQ-operator functional relations are singular at η=π/3\eta=\pi/3. The problem asks for a proof (or an explicit refutation) of the inhomogeneous eigenvalue conjecture, comprising both the eigenvalue assertion and the eigenspace characterisation.

Background

The eight-vertex model of planar statistical mechanics places arrows on the edges of a square lattice and assigns a Boltzmann weight to each vertex; the allowed configurations (an even number of arrows pointing into each vertex) are encoded in four vertex weights a,b,c,da,b,c,d arranged in the integrable RR-matrix. In the standard theta-function parametrisation, the weights depend on a spectral parameter uu, a crossing parameter η\eta and an elliptic nome p=eiπτp=e^{\mathrm{i}\pi\tau} with Im⁡τ>0\operatorname{Im}\tau>0 (Jacobi theta functions are the elliptic special functions entering this parametrisation). The RR-matrix satisfies the Yang–Baxter equation — the algebraic compatibility condition behind integrability — which implies that transfer matrices with different spectral parameters commute. On a lattice with LL vertical lines and periodic boundary conditions in the horizontal direction the transfer matrix is T(u∣u1,…,uL)=tr⁡0R0,L(uL−u)⋯R0,1(u1−u)T(u|u_1,\dots,u_L)=\operatorname{tr}_0 R_{0,L}(u_L-u)\cdots R_{0,1}(u_1-u); the parameters uiu_i are inhomogeneities, letting each column carry a different spectral parameter, and u1=⋯=uL=0u_1=\cdots=u_L=0 gives the homogeneous model.

At η=π/3\eta=\pi/3 the weights obey, identically in uu, the relation (a(u)2+a(u)b(u))(b(u)2+a(u)b(u))=(c(u)2+a(u)b(u))(d(u)2+a(u)b(u))(a(u)^2+a(u)b(u))(b(u)^2+a(u)b(u))=(c(u)^2+a(u)b(u))(d(u)^2+a(u)b(u)); this is called the supersymmetric eight-vertex model because, via the transfer matrix, it corresponds to an odd-length XYZ spin-1/2 chain (the anisotropic Heisenberg chain with elliptic couplings) whose Hamiltonian admits lattice supersymmetry: a nilpotent 'supercharge' operator whose anticommutator reproduces the Hamiltonian. This is also called the combinatorial line, because in the homogeneous limit the model's ground-state data are conjecturally tied to the enumeration of alternating sign matrices — the square tables of 00 and ±1\pm1 with alternating partial sums that play the same role here that they play in the six-vertex model at Δ=−1/2\Delta=-1/2.

For the homogeneous model, Stroganov conjectured in 2001 that for odd L=2n+1L=2n+1 the transfer matrix possesses the doubly degenerate eigenvalue Θn=(a+b)2n+1\Theta_n=(a+b)^{2n+1} — for positive weights, its largest eigenvalue. Hagendorf and Liénardy proved this in 2018: the eigenspace of Θn\Theta_n is spanned by the supersymmetry singlets (ground states) of the related XYZ Hamiltonian. Their conclusion formulates the inhomogeneous generalisation that remained open and is the subject of this problem.

For the inhomogeneous model, Razumov and Stroganov (2010) conjectured, from the inversion relations of the transfer matrix and exact diagonalisation at small sizes, that for L=2n+1L=2n+1 and periodic boundary conditions the transfer matrix has the doubly degenerate eigenvalue Θn(u∣u1,…,u2n+1)=∏i=12n+1(a(ui−u)+b(ui−u))\Theta_n(u|u_1,\dots,u_{2n+1})=\prod_{i=1}^{2n+1}\bigl(a(u_i-u)+b(u_i-u)\bigr), a product of local factors r(ui−u)=a(ui−u)+b(ui−u)r(u_i-u)=a(u_i-u)+b(u_i-u) that reduces to (a+b)2n+1(a+b)^{2n+1} in the homogeneous limit. Zinn-Justin (2013) reformulated the conjecture as an eigenvalue problem with a spin-reversal constraint and strengthened it to a precise characterisation of the eigenspace (part (2) of the statement below). A 'sum rule' here means a closed-form identity for the sum of the components, or the squared norm, of the ground-state eigenvector; Zinn-Justin used the conjecture — as an unproven input — to derive such a sum rule for the XYZ ground state, and the entire scalar-product and sum-rule program of Brasseur and Hagendorf (2021) rests on it. In the trigonometric limit p→0p\to0 the corresponding six-vertex statement is a proven theorem, but in the elliptic regime no proof is known: the 2018 homogeneous proof does not extend because its supercharge and cohomology argument is tied to translation invariance, and the standard Baxter TQTQ-relation route is obstructed at η=π/3\eta=\pi/3 (no QQ-operator analytic in a neighbourhood of this point is known).

Problem Statement

Prove the Razumov–Stroganov–Zinn-Justin conjecture for the inhomogeneous supersymmetric eight-vertex model with periodic boundary conditions, in the following form (or refute it by an explicit counterexample).

Let p=eiπτp=e^{\mathrm{i}\pi\tau} with Im⁡τ>0\operatorname{Im}\tau>0, let η=π/3\eta=\pi/3, and let R(u)∈End⁡(V⊗V)R(u)\in\operatorname{End}(V\otimes V), V=C2V=\mathbb{C}^2, be the eight-vertex RR-matrix, which in the basis ∣↑↑⟩,∣↑↓⟩,∣↓↑⟩,∣↓↓⟩|\uparrow\uparrow\rangle,|\uparrow\downarrow\rangle,|\downarrow\uparrow\rangle,|\downarrow\downarrow\rangle takes the form R(u)=(a(u)00d(u)0b(u)c(u)00c(u)b(u)0d(u)00a(u)),R(u)=\begin{pmatrix}a(u)&0&0&d(u)\\0&b(u)&c(u)&0\\0&c(u)&b(u)&0\\d(u)&0&0&a(u)\end{pmatrix}, with weights given, up to an overall normalisation (under which every statement below is invariant), in terms of the Jacobi theta functions ϑi(u,q)\vartheta_i(u,q) in the Whittaker–Watson notation by a(u)=ϑ4(2η,p2)ϑ1(u+2η,p2)ϑ4(u,p2),b(u)=ϑ4(2η,p2)ϑ4(u+2η,p2)ϑ1(u,p2),a(u)=\vartheta_4(2\eta,p^2)\vartheta_1(u+2\eta,p^2)\vartheta_4(u,p^2),\qquad b(u)=\vartheta_4(2\eta,p^2)\vartheta_4(u+2\eta,p^2)\vartheta_1(u,p^2), c(u)=ϑ1(2η,p2)ϑ4(u+2η,p2)ϑ4(u,p2),d(u)=ϑ1(2η,p2)ϑ1(u+2η,p2)ϑ1(u,p2),c(u)=\vartheta_1(2\eta,p^2)\vartheta_4(u+2\eta,p^2)\vartheta_4(u,p^2),\qquad d(u)=\vartheta_1(2\eta,p^2)\vartheta_1(u+2\eta,p^2)\vartheta_1(u,p^2), so that at η=π/3\eta=\pi/3 the supersymmetry relation (a(u)2+a(u)b(u))(b(u)2+a(u)b(u))=(c(u)2+a(u)b(u))(d(u)2+a(u)b(u))(a(u)^2+a(u)b(u))(b(u)^2+a(u)b(u))=(c(u)^2+a(u)b(u))(d(u)^2+a(u)b(u)) holds identically in uu. For L=2n+1L=2n+1 vertical lines, n⩾0n\geqslant0 (the case n=0n=0 is trivial), and arbitrary inhomogeneity parameters u1,…,u2n+1u_1,\dots,u_{2n+1}, let T(u∣u1,…,u2n+1)=tr⁡0 R0,2n+1(u2n+1−u)⋯R0,1(u1−u)T(u|u_1,\dots,u_{2n+1})=\operatorname{tr}_0\, R_{0,2n+1}(u_{2n+1}-u)\cdots R_{0,1}(u_1-u) act on V⊗(2n+1)V^{\otimes(2n+1)}, where R0,j(u)R_{0,j}(u) denotes R(u)R(u) acting on an auxiliary space V0≅C2V_0\cong\mathbb{C}^2 and on the jj-th tensor factor, and tr⁡0\operatorname{tr}_0 is the trace over V0V_0; and let F=∏j=12n+1σjxF=\prod_{j=1}^{2n+1}\sigma_j^x denote spin reversal (σjx\sigma_j^x the Pauli matrix on site jj). Prove:

(1) Eigenvalue (Razumov–Stroganov 2010): for every spectral parameter uu, T(u∣u1,…,u2n+1)T(u|u_1,\dots,u_{2n+1}) has the doubly degenerate eigenvalue Θn(u∣u1,…,u2n+1)=∏i=12n+1r(ui−u),r(u)=a(u)+b(u)=ϑ4(0,p2)ϑ1(u+η,p2)ϑ4(u+η,p2),\Theta_n(u|u_1,\dots,u_{2n+1})=\prod_{i=1}^{2n+1} r(u_i-u),\qquad r(u)=a(u)+b(u)=\vartheta_4(0,p^2)\vartheta_1(u+\eta,p^2)\vartheta_4(u+\eta,p^2), which reduces to (a+b)2n+1(a+b)^{2n+1} in the homogeneous limit u1=⋯=u2n+1=0u_1=\cdots=u_{2n+1}=0.

(2) Eigenspace characterisation (Zinn-Justin 2013): the space of solutions of T(u∣u1,…,u2n+1)∣Ψ⟩=Θn(u∣u1,…,u2n+1)∣Ψ⟩,F∣Ψ⟩=(−1)n∣Ψ⟩,T(u|u_1,\dots,u_{2n+1})|\Psi\rangle=\Theta_n(u|u_1,\dots,u_{2n+1})|\Psi\rangle,\qquad F|\Psi\rangle=(-1)^n|\Psi\rangle, understood as an identity in uu with a uu-independent vector ∣Ψ⟩|\Psi\rangle, is one-dimensional, spanned by a vector ∣Ψn⟩|\Psi_n\rangle whose components in the spin-zz basis are entire functions of each uiu_i, generically non-vanishing and without a common factor depending on the uiu_i — equivalently, theta functions of degree 2n2n and nome p2p^2 in each variable — satisfying the quasi-periodicity relations ∣Ψn(…,ui+2πτ,… )⟩=p−4nzi2n∏j≠izj−1∣Ψn(…,ui,… )⟩,∣Ψn(…,ui+π,… )⟩=∏j≠iσjz ∣Ψn(…,ui,… )⟩,|\Psi_n(\dots,u_i+2\pi\tau,\dots)\rangle=p^{-4n}z_i^{2n}\prod_{j\neq i}z_j^{-1}|\Psi_n(\dots,u_i,\dots)\rangle,\qquad |\Psi_n(\dots,u_i+\pi,\dots)\rangle=\prod_{j\neq i}\sigma_j^z\,|\Psi_n(\dots,u_i,\dots)\rangle, with zi=e−2iuiz_i=e^{-2\mathrm{i}u_i}.

Part (2) implies part (1): the transfer matrix commutes with both FF and the spin parity P=∏j=12n+1σjzP=\prod_{j=1}^{2n+1}\sigma_j^z, while FP=−PFFP=-PF for odd L=2n+1L=2n+1, so PP maps the (−1)n(-1)^n spin-reversal sector isomorphically onto the opposite sector inside the Θn\Theta_n-eigenspace; the one-dimensionality of part (2) therefore yields exactly the two-fold degeneracy of part (1), the second eigenvector being, up to normalisation, P∣Ψn⟩P|\Psi_n\rangle. The two parts form one scientific objective — settling the inhomogeneous eigenvalue problem for Θn\Theta_n on which the conditional sum-rule results are built; a proof of part (1) alone resolves the conjecture in the form posed by Razumov–Stroganov and reiterated as open by Hagendorf–Liénardy (2018), while the full conjecture additionally fixes the analytic structure of the eigenspace.

References

  1. C. Hagendorf and J. Liénardy, On the transfer matrix of the supersymmetric eight-vertex model. I. Periodic boundary conditions, J. Stat. Mech. (2018) 033106. DOI: 10.1088/1742-5468/aab01d; arXiv:1711.04397. https://arxiv.org/abs/1711.04397
  2. A. V. Razumov and Yu. G. Stroganov, A possible combinatorial point for the XYZ spin chain, Theoretical and Mathematical Physics 164 (2010) 977–991. DOI: 10.1007/s11232-010-0078-3; arXiv:0911.5030. https://arxiv.org/abs/0911.5030
  3. P. Zinn-Justin, Sum rule for the eight-vertex model on its combinatorial line, in K. Iohara, S. Morier-Genoud and B. Rémy (eds.), Symmetries, Integrable Systems and Representations, Springer Proceedings in Mathematics & Statistics, Springer (2013), pp. 599–637. DOI: 10.1007/978-1-4471-4863-0_26; arXiv:1202.4420. https://arxiv.org/abs/1202.4420
  4. S. Brasseur and C. Hagendorf, Sum rules for the supersymmetric eight-vertex model, J. Stat. Mech. (2021) 023102. DOI: 10.1088/1742-5468/abda28; arXiv:2009.14077. https://arxiv.org/abs/2009.14077
  5. C. Hagendorf and J. Liénardy, On the transfer matrix of the supersymmetric eight-vertex model. II. Open boundary conditions, J. Stat. Mech. (2020) 033104. DOI: 10.1088/1742-5468/ab7748; arXiv:1911.03348. https://arxiv.org/abs/1911.03348
  6. 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
  7. Yu. Stroganov, The importance of being odd, J. Phys. A: Math. Gen. 34 (2001) L179–L185. DOI: 10.1088/0305-4470/34/13/104. https://doi.org/10.1088/0305-4470/34/13/104