Does parameter-space modular invariance organize gapped integrable models? Virasoro characters, dualities, and the complex-argument extension of the XYZ chain

Ercolessi, Evangelisti, Franchini and Ravanini (Phys. Rev. B 88, 104418 (2013)) showed that Baxter's elliptic parametrization of the spin-1/2 XYZ chain can be analytically continued, via Landen transformations, from the principal regime to the whole twelve-region phase diagram, and that the resulting reorganizations of the couplings realize the spin-rotation symmetries of the model as a $\mathrm{PGL}(2,\mathbb{Z})$ action in parameter space, under which the partition function is invariant. The paper explicitly leaves open whether this parameter-space modular invariance is a general organizing principle of gapped integrable models: does it explain why partition functions and finite-size/finite-mass spectra of integrable models in their gapped phases organize into Virasoro characters with the mass setting the finite-size scale (a long-standing unexplained observation dating to Date-Jimbo-Miwa-Okado and Bauer-Saleur); does it encode the integrability-preserving perturbations; can it be exploited to classify dual-related operator contents and to compute correlation functions and Rényi/von Neumann entanglement entropies; and does the continuation of the elliptic argument to generic complex values — yielding non-Hermitian but integrability-preserving Hamiltonians — admit a physical interpretation?

Background

The spin-1/2 XYZ chain is the fully anisotropic Heisenberg chain, with Hamiltonian $H=\sum_n (J_x\sigma^x_n\sigma^x_{n+1}+J_y\sigma^y_n\sigma^y_{n+1}+J_z\sigma^z_n\sigma^z_{n+1})$, where $\sigma^{x,y,z}$ are Pauli matrices. It is integrable (it admits infinitely many conserved quantities and hence an exact solution) through its equivalence, via a Jordan-Wigner-type transformation, to the eight-vertex model of classical statistical mechanics solved by Baxter. Baxter's solution expresses the coupling ratios $J_y/J_x$ and $J_z/J_x$ through Jacobi elliptic functions — doubly periodic meromorphic functions generalizing sine and cosine, parametrized by an argument $u$ and a modulus $l$, or equivalently by theta functions (their quasi-periodic building blocks) with argument $z$ and modular parameter $\tau$ — but was originally obtained only in a 'principal regime' of the phase diagram. The phase diagram in the $(J_y/J_x,J_z/J_x)$ plane consists of twelve regions related by spin rotations exchanging the $x$, $y$, $z$ axes, and each region requires a different branch of the mapping to the eight-vertex parameters.

Ercolessi, Evangelisti, Franchini and Ravanini (2013) showed that the analytic continuation of Baxter's parametrization via Landen transformations (the classical modular transformations of the elliptic modulus) covers the whole phase diagram, and that the rotations connecting the twelve regions act on the parameter space of the elliptic functions as the group $\mathrm{PGL}(2,\mathbb{Z})$ — integer $2\times 2$ matrices with determinant $\pm 1$, taken modulo overall sign, acting by fractional-linear transformations on $\tau$, an index-two extension of the modular group $\mathrm{PSL}(2,\mathbb{Z})$ (the source paper calls it the 'double covering' of the modular group). The partition function of the XYZ chain is therefore invariant under this group acting in parameter space, in close analogy with the modular invariance of a conformal field theory (CFT) partition function on a torus, where the modular group acts on real space (and with Seiberg-Witten electromagnetic duality in supersymmetric gauge theory, where modular transformations also act on the space of couplings).

Two further strands of background are needed. First, in the corner transfer matrix (CTM) formulation of two-dimensional integrable lattice models — the corner piece of the full transfer matrix, whose diagonalization by Baxter yields exact one-point functions and, in modern use, exact bipartite entanglement entropies $S_n=\frac{1}{1-n}\log\mathrm{Tr}\rho_A^n$ (Rényi entropies) and their von Neumann limit $n\to 1$, where $\rho_A$ is the reduced density matrix of a subsystem — it has been known since Date-Jimbo-Miwa-Okado (1987) and Bauer-Saleur (1989) that the local (highest-weight) probabilities of gapped models organize into characters of Virasoro representations (graded traces over representations of the Virasoro algebra, the symmetry algebra of two-dimensional CFT), with the finite-size/temperature scale set by the mass rather than by the system size. The 2013 paper calls this a 'long-standing mystery' that 'has so far eluded fundamental explanation' and proposes that parameter-space modularity might provide the explanation, since real-space modular invariance is precisely what organizes the operator content of critical theories. Second, the same authors' exact entanglement calculations for the XYZ chain (2011, 2012) found an essential singularity at the Berezinskii-Kosterlitz-Thouless transition and 'unusual' corrections to scaling in the massive phase, with finite-size and finite-mass contributions entering with different exponents and violating naive scaling; the 2013 paper flags this as deserving further analytical and numerical study, and speculates that the modular picture may control these subleading terms. The closest known partial results lie outside the interacting XYZ setting: for quadratic (free-fermionic) chains the symmetries and dualities of the entanglement entropy have been shown to originate in Möbius transformations acting on the associated torus, and exact corner-transfer-matrix Rényi entropies have been computed for non-unitary Forrester-Baxter RSOS chains.

Finally, within the modular parametrization only purely real or purely imaginary arguments $u$ (modulo half-periods) guarantee real couplings and a Hermitian Hamiltonian; continuing $u$ to generic complex values defines non-Hermitian Hamiltonians that nevertheless preserve the structure responsible for integrability. The open question is whether the parameter-space modular structure is a genuine organizing principle of gapped integrable models — explanatory of the character organization, predictive of integrability-preserving perturbations, computationally useful for dual-related operator contents and entanglement — and whether this natural mathematical completion to complex couplings admits a physical interpretation.

Problem Statement

Determine to what extent the parameter-space modular invariance of the gapped XYZ chain is a general structural principle of gapped integrable models, and turn it into an explanatory and computational tool. The constituent questions share one objective — establishing the $\mathrm{PGL}(2,\mathbb{Z})$ modular structure of the elliptic parametrization as the organizing principle of the massive phases of integrable systems:

1. Derive (or refute) that the $\mathrm{PGL}(2,\mathbb{Z})$ invariance in parameter space constrains the organization of the gapped partition function and of the finite-size/finite-mass spectra into Virasoro(-like) characters — thereby explaining why integrable models in their gapped phases admit such a character organization with the finite-size/temperature scale set by the mass parameter, an observation left unexplained since Date-Jimbo-Miwa-Okado and Bauer-Saleur. In real-space modular invariance of critical theories, the modular group constrains the partition function and organizes the operator content; the question is whether parameter-space modularity does the analogous job for gapped integrable models.

2. Determine whether the modular properties of the elliptic parametrization encode the symmetries of the model and hence the class of integrability-preserving relevant perturbations that drive the system away from criticality (currently a conjecture), including how far this modular encoding extends beyond the XYZ chain.

3. Determine whether, and to what extent, the parameter-space dualities can be exploited computationally. Specifically, determine: (a) whether the fact that points with different couplings are related by modular transformations can be used to classify the operator contents related by these dualities; (b) whether it can be used to simplify the calculation of correlation functions; (c) whether it can be used to simplify the calculation of the Rényi and von Neumann entanglement entropies; and (d) whether, and if so how, the modular picture controls the subleading finite-size and finite-mass corrections that violate simple scaling. Each of (a)–(d) is resolved by a determinate positive or negative finding.

4. Explore the continuation of the elliptic argument $u$ to generic complex values — mathematically natural within the modular picture, yielding non-Hermitian Hamiltonians that preserve the integrable structure — and determine whether such non-unitary extensions have a physical interpretation and usefulness.

A complete answer establishes the general principle (for the XYZ chain and other elliptically parametrized integrable models, or by an explicit counterexample delimiting the range of parameters and models where it holds) together with its computational consequences; the source states the authors were 'currently exploring to which extent this speculation can hold true in general'. Items 1–3 are required constituents of a complete answer, whereas item 4 is an optional exploratory direction; for item 1, a single explicit counterexample delimiting where the character organization fails is by itself a complete resolution of that item; for item 3, all four parts (a)–(d) are required, and each is complete once it is settled by an explicit construction or a determinate negative finding.

References

1. E. Ercolessi, S. Evangelisti, F. Franchini, F. Ravanini, 'Modular invariance in the gapped XYZ spin-1/2 chain', Phys. Rev. B 88, 104418 (2013). DOI: 10.1103/PhysRevB.88.104418; arXiv:1301.6758; https://arxiv.org/abs/1301.6758
2. E. Ercolessi, S. Evangelisti, F. Franchini, F. Ravanini, 'Essential singularity in the Renyi entanglement entropy of the one-dimensional XYZ spin-1/2 chain', Phys. Rev. B 83, 012402 (2011). DOI: 10.1103/PhysRevB.83.012402; arXiv:1008.3892; https://arxiv.org/abs/1008.3892
3. E. Ercolessi, S. Evangelisti, F. Franchini, F. Ravanini, 'Correlation length and unusual corrections to entanglement entropy', Phys. Rev. B 85, 115428 (2012). DOI: 10.1103/PhysRevB.85.115428; arXiv:1201.6367; https://arxiv.org/abs/1201.6367
4. A. De Luca, F. Franchini, 'Approaching the restricted solid-on-solid critical points through entanglement: One model for many universalities', Phys. Rev. B 87, 045118 (2013). DOI: 10.1103/PhysRevB.87.045118; arXiv:1205.6426; https://arxiv.org/abs/1205.6426
5. E. Date, M. Jimbo, T. Miwa, M. Okado, 'Automorphic properties of local height probabilities for integrable solid-on-solid models', Phys. Rev. B 35, 2105 (1987). DOI: 10.1103/PhysRevB.35.2105; https://doi.org/10.1103/PhysRevB.35.2105
6. H. Saleur, M. Bauer, 'On some relations between local height probabilities and conformal invariance', Nucl. Phys. B 320, 591 (1989). DOI: 10.1016/0550-3213(89)90014-X; https://doi.org/10.1016/0550-3213(89)90014-X
7. R. J. Baxter, 'Exactly Solved Models in Statistical Mechanics', Academic Press, London (1982). https://inspirehep.net/literature/1120339 (INSPIRE record)
8. F. Ares, J. G. Esteve, F. Falceto, A. R. de Queiroz, 'On the Möbius transformation in the entanglement entropy of fermionic chains', J. Stat. Mech. (2016) 043106. DOI: 10.1088/1742-5468/2016/04/043106; arXiv:1511.02382; https://arxiv.org/abs/1511.02382
9. F. Ares, J. G. Esteve, F. Falceto, A. R. de Queiroz, 'Entanglement entropy and Möbius transformations for critical fermionic chains', J. Stat. Mech. (2017) 063104. DOI: 10.1088/1742-5468/aa71dc; arXiv:1612.07319; https://arxiv.org/abs/1612.07319
10. D. Bianchini, F. Ravanini, 'Entanglement entropy from corner transfer matrix in Forrester-Baxter non-unitary RSOS models', J. Phys. A 49, 154005 (2016). DOI: 10.1088/1751-8113/49/15/154005; arXiv:1509.04601; https://arxiv.org/abs/1509.04601
11. S. Murciano, G. Di Giulio, P. Calabrese, 'Symmetry resolved entanglement in gapped integrable systems: a corner transfer matrix approach', SciPost Phys. 8, 046 (2020). DOI: 10.21468/SciPostPhys.8.3.046; arXiv:1911.09588; https://arxiv.org/abs/1911.09588
12. V. Marić, S. M. Giampaolo, D. Kuić, F. Franchini, 'The frustration of being odd: how boundary conditions can destroy local order', New J. Phys. 22, 083024 (2020). DOI: 10.1088/1367-2630/aba064; arXiv:1908.10876; https://arxiv.org/abs/1908.10876
13. I. Gahramanov, 'Integrability from supersymmetric duality: a short review', Analysis and Mathematical Physics 16, 98 (2026). DOI: 10.1007/s13324-026-01224-4; arXiv:2201.00351; https://arxiv.org/abs/2201.00351
14. S. A. Jafari, 'Exact phase boundaries and topological phase transitions of the XYZ spin chain', Phys. Rev. E 96, 012159 (2017). DOI: 10.1103/PhysRevE.96.012159; arXiv:1703.01420; https://arxiv.org/abs/1703.01420
15. G. Pelegrí, J. Mompart, V. Ahufinger, A. J. Daley, 'Quantum magnetism with ultracold bosons carrying orbital angular momentum', Phys. Rev. A 100, 023615 (2019). DOI: 10.1103/PhysRevA.100.023615; arXiv:1907.11599; https://arxiv.org/abs/1907.11599
16. Ö. M. Aksoy, C. Mudry, A. Furusaki, A. Tiwari, 'Lieb-Schultz-Mattis anomalies and web of dualities induced by gauging in quantum spin chains', SciPost Phys. 16, 022 (2024). DOI: 10.21468/SciPostPhys.16.1.022; arXiv:2308.00743; https://arxiv.org/abs/2308.00743