Completing the rigorous low-temperature spectral theory of the quantum transfer matrix of the massless XXZ spin-1/2 chain

The spin-1/2 XXZ chain with anisotropy $-1<\Delta<1$ (the massless regime) is the archetype of a Luttinger liquid. In the quantum transfer matrix (QTM) approach its thermodynamics and correlation lengths are governed by the spectrum of a finite-dimensional auxiliary transfer matrix whose Bethe-Ansatz analysis Faulmann, Göhmann and Kozlowski (Probability and Mathematical Physics 6 (2025) 1507-1658) recently made rigorous at low temperature. Their solvability and classification theory is, however, restricted to $0\le\Delta<1$, obtained under working hypotheses on the Bethe roots (repulsion, simplicity, exclusion of neighbourhoods of the singularities at $\pm\mathrm{i}\zeta/2$), valid only for $1/(NT^4)$ small, and conditional on two conjectures — non-degeneracy and dominance of the largest QTM eigenvalue, and commutativity of the thermodynamic and Trotter limits — previously proven only at high temperature. The open problem is to complete this programme: classify all Bethe-root solutions of the QTM equations without the working hypotheses (including roots collapsing to the singularities and coincident roots); extend the classification to the remaining massless range $-1<\Delta<0$, in particular proving the conjectured thermal $r$-string classification of the source (its Conjecture 6.6); prove the dominance and limit-commutativity conjectures at low temperature; and establish that the constructed eigenvalue class exhausts the low-temperature spectrum.

Background

The spin-1/2 XXZ chain is the archetype of one-dimensional quantum integrable models. Its Hamiltonian on $L$ sites with periodic boundary conditions is $H \;=\; J\sum_{a=1}^{L}\Big\{\sigma_a^x\sigma_{a+1}^x+\sigma_a^y\sigma_{a+1}^y+\Delta\big(\sigma_a^z\sigma_{a+1}^z+1\big)\Big\}\;-\;\frac{h}{2}\sum_{a=1}^{L}\sigma_a^z\;,$ with exchange coupling $J>0$, anisotropy $\Delta=\cos\zeta$ parametrised by $\zeta\in(0,\pi)$, Pauli matrices $\sigma^\alpha$, and magnetic field $h$. For $-1<\Delta<1$ and $0<h<4J(1+\Delta)$ the zero-temperature spectrum is massless (gapless): the low-energy behaviour is that of a Luttinger liquid described by the $c=1$ free-boson conformal field theory, and the Fermi zone of the underlying Bethe quasiparticles ends at points $\pm q$ determined by the field.

Thermodynamics at temperature $T$ can be computed by the quantum transfer matrix (QTM) method of Suzuki. Expressing the partition function through a Trotter decomposition with $2N$ factors ($N$ is the Trotter number), the trace over the chain is rewritten as the trace of the $L$-th power of a single $2^{2N}$-dimensional auxiliary operator, the QTM $t_q(0)$, acting on a tensor product of $2N$ two-dimensional spaces. In the limits of infinite chain length followed by infinite Trotter number, the per-site free energy is determined by the dominant (largest-modulus) eigenvalue of $t_q(0)$, while the ratios of sub-dominant to dominant eigenvalues yield the correlation lengths of the chain. Since the QTM is the transfer matrix of an inhomogeneous six-vertex-type model, its spectrum is accessible by the Bethe Ansatz — the algebraic method that produces eigenvalues parametrised by solutions of a finite system of algebraic Bethe equations. Making this physics folklore rigorous is difficult. The only earlier fully rigorous result is the high-temperature analysis of Göhmann, Goomanee, Kozlowski and Suzuki (2020), who proved for $T$ large enough that the QTM has a non-degenerate dominant eigenvalue given by a specific Bethe solution and that the thermodynamic and Trotter limits commute; in the antiferromagnetic massive regime $\Delta>1$, the low-temperature QTM spectrum was classified earlier only at the level of physics derivations by Dugave, Göhmann, Kozlowski and Suzuki (2015).

Faulmann, Göhmann and Kozlowski (2025) initiated the rigorous low-temperature theory for the massless regime. They reformulate the Bethe equations of the QTM as a non-linear integral equation (NLIE) — a Destri–de Vega-type equation whose unknown is a function on a contour in the complex spectral plane — supplemented by quantisation conditions, and they rigorously prove existence and uniqueness of solutions, classify the structure of the associated particle and hole roots, construct a large class of sub-dominant QTM eigenvalues with sharp low-temperature, large-$N$ asymptotics, prove that the corresponding Bethe vectors (the eigenvectors produced by the algebraic Bethe Ansatz) are non-zero, and show that the correlation lengths that diverge as $T\to 0^+$ are, to leading order in $T$, in one-to-one correspondence with the spectrum of the $c=1$ free-boson conformal field theory. The classification involves thermal $r$-strings: complexes of $r$ particle roots whose consecutive members are spaced by $\mathrm{i}\zeta$ up to exponentially small corrections, characterised through the dressed energy $\varepsilon_c$, an auxiliary function solving a linear integral equation on a curved contour in the complex plane.

The restrictions under which these results hold define the problem. First, the analysis assumes Hypotheses 2.1 and 2.2 of the source on the Bethe roots: the roots stay a fixed distance from the Fermi points $\pm q$ and a distance of order $T$ from a reference contour, avoid neighbourhoods (of radius of order $T$ in one hypothesis and of small fixed radius in the other) of the singularities $\pm\mathrm{i}\zeta/2$ and $-3\mathrm{i}\zeta/2$ of the counting function (the logarithmic phase function of the Bethe equations), are simple, and repel each other at scale $T$. The authors state that dealing with coincident or weakly separated roots would make the analysis bulkier but remains within reach, while other relaxations would require genuinely new techniques. Second, the whole analysis is carried out for $0\le\Delta<1$ ($0<\zeta\le\pi/2$); the extension to $-1<\Delta<0$ is blocked, in the authors' words, by the lack of sufficient control of the auxiliary function $\varepsilon_c$ in that regime, and the expected structure there — for which only numerics and a stated conjecture (Conjecture 6.6 of the source) exist — involves thermal $r$-strings of lengths $r>1$, which cannot occur for $\zeta\le\pi/2$. Third, the identification of the constructed largest Bethe-Ansatz eigenvalue with the true dominant QTM eigenvalue, and the resulting low-temperature free-energy formula, are proven only conditionally on two conjectures (a non-degenerate dominant eigenvalue; commutativity of the $L\to\infty$ and $N\to\infty$ limits), established to date only at high temperature. Fourth, the authors construct an explicit class $\mathcal{B}_{\mathrm{NLIE}}(B,\widetilde{B})$ of Bethe solutions which they believe to be physically complete — every QTM eigenvalue outside the class should have modulus below any preassigned $\varepsilon>0$ — but this is stated as a belief, not proven. One subtlety is recorded by the source itself: completeness of the Bethe Ansatz at finite Trotter number, which is open for the discretisation used, does follow from the general results of Tarasov and Varchenko for a generic inhomogeneous twisted six-vertex transfer matrix obtained from a slightly different Trotter discretisation — but such completeness information gives no access to the large-$N$ behaviour of the solutions, which is the substantive content. The rigorous summation of the thermal form-factor series (eigenvalue-plus-matrix-element expansions of correlation functions) that this spectral theory is designed to enable is explicitly deferred by the source to subsequent work.

Problem Statement

For the antiferromagnetic massless XXZ spin-1/2 chain ($-1<\Delta<1$, i.e. $0<\zeta<\pi$, with $0<h<4J(1+\Delta)$ and, as assumed by the source unless stated otherwise, $\zeta$ generic in the sense $\zeta\notin\pi\mathbb{Q}$), complete the rigorous low-temperature spectral theory of the quantum transfer matrix $t_q(0)$ at Trotter number $N$ initiated by Faulmann, Göhmann and Kozlowski (2025), which at present is established only for $0\le\Delta<1$, for $T$ and $1/(NT^4)$ small enough, and under the working Hypotheses 2.1-2.2 on the Bethe roots. The problem has four jointly required components which together constitute one objective — the unrestricted classification and dominance theory of the QTM spectrum:

(i) Classification without the working hypotheses. Prove existence, uniqueness and structure of the solutions to the quantisation conditions for the QTM Bethe equations with Hypotheses 2.1-2.2 removed — in particular for solutions with coincident or weakly separated (non-repulsive) roots, and for solutions whose roots collapse toward the logarithmic singularities at $\pm\mathrm{i}\zeta/2$ of the finite-$N$ counting function, including the determination of whether such solutions exist and what eigenvalues they produce in the $N\to\infty$ limit.

(ii) Full anisotropy range. Extend the classification to the remaining massless range $-1<\Delta<0$ ($\pi/2<\zeta<\pi$), where the required control of the dressed energy $\varepsilon_c$ is presently missing; in particular prove the conjectural classification of the allowed thermal $r$-strings for general $\zeta$ (Conjecture 6.6 of the source: particle roots group into $r$-strings spaced by $\mathrm{i}\zeta$ whose tops lie on curves defined through the functions $\varepsilon_{c;k}^{(-)}$, with the allowed lengths $r$ depending on $\zeta$), or establish the correct classification with proof.

(iii) Dominance and limit commutativity at low temperature. Prove that, for $T$ small enough (not only for $T$ large, where this is known from 2020 work), the QTM admits a non-degenerate dominant eigenvalue which coincides with the Bethe-Ansatz-constructed largest eigenvalue, and that the thermodynamic ($L\to\infty$) and Trotter ($N\to\infty$) limits of the free energy commute (Conjectures 1.2 and 1.3 of the source), so that the conditional low-temperature free-energy formula and the identification with the $c=1$ conformal predictions become unconditional theorems.

(iv) Spectral exhaustion. Prove that the constructed class of Bethe solutions is physically complete in the sense stated by the source: for any $\varepsilon>0$ there exist fixed $B,\widetilde{B}>0$ such that every QTM eigenvalue not described by the constructed class has modulus below $\varepsilon$, so that the constructed family exhausts the low-temperature spectrum relevant for correlation functions.

A resolution must address all four components for the standard QTM of the chain (a proof for a surrogate, e.g. inhomogeneous-twisted, discretisation is admissible only if its equivalence for the spectral conclusions is itself proven). The rigorous summation of the thermal form-factor series for correlation functions, which this spectral theory is designed to enable, is the subsequent objective of the programme and not part of the completion boundary of this problem.

References

1. S. Faulmann, F. Göhmann, K. K. Kozlowski, "Low-temperature spectrum of the quantum transfer matrix of the XXZ chain in the massless regime", Probability and Mathematical Physics 6 (2025) 1507-1658. DOI: 10.2140/pmp.2025.6.1507, arXiv:2305.06679, https://arxiv.org/abs/2305.06679
2. F. Göhmann, S. Goomanee, K. K. Kozlowski, J. Suzuki, "Thermodynamics of the spin-1/2 Heisenberg-Ising chain at high temperatures: a rigorous approach", Communications in Mathematical Physics 377 (2020) 623-673. DOI: 10.1007/s00220-020-03749-6, arXiv:1811.12020, https://arxiv.org/abs/1811.12020
3. M. Dugave, F. Göhmann, K. K. Kozlowski, J. Suzuki, "Low-temperature spectrum of correlation lengths of the XXZ chain in the antiferromagnetic massive regime", Journal of Physics A: Mathematical and Theoretical 48 (2015) 334001. DOI: 10.1088/1751-8113-48-33-334001, arXiv:1504.07923, https://arxiv.org/abs/1504.07923
4. S. Faulmann, F. Göhmann, K. K. Kozlowski, "Dressed energy of the XXZ chain in the complex plane", Letters in Mathematical Physics 111 (2021) 132. DOI: 10.1007/s11005-021-01473-3, arXiv:2106.06732, https://arxiv.org/abs/2106.06732
5. F. Göhmann, K. K. Kozlowski, "Space-like asymptotics of the thermal two-point functions of the XXZ spin-1/2 chain", Journal of Statistical Mechanics 2024 (2024) 033101. DOI: 10.1088/1742-5468/ad267c, arXiv:2311.17196, https://arxiv.org/abs/2311.17196
6. F. Göhmann, K. K. Kozlowski, M. D. Minin, "Thermal form-factor expansion of the dynamical two-point functions of local operators in integrable quantum chains", Journal of Physics A: Mathematical and Theoretical 56 (2023) 475003. DOI: 10.1088/1751-8121/ad058a, arXiv:2307.13789, https://arxiv.org/abs/2307.13789
7. K. K. Kozlowski, V. Terras, "Multi-point correlation functions in the boundary XXZ chain at finite temperature", Annales Henri Poincaré 25 (2024) 1007-1046. DOI: 10.1007/s00023-023-01310-4, arXiv:2212.08824, https://arxiv.org/abs/2212.08824
8. T. S. Tavares, G. A. P. Ribeiro, "Finite temperature properties of an integrable zigzag ladder chain", Nuclear Physics B 995 (2023) 116333. DOI: 10.1016/j.nuclphysb.2023.116333, arXiv:2307.11279, https://arxiv.org/abs/2307.11279
9. V. Tarasov, A. Varchenko, "Completeness of Bethe vectors and difference equations with regular singular points", International Mathematics Research Notices 1995 (1995) 637-669. DOI: 10.1155/S1073792895000377, https://doi.org/10.1155/S1073792895000377 (preprint arXiv:q-alg/9504011, titled "Bases of Bethe Vectors and Difference Equations with Regular Singular Points", https://arxiv.org/abs/q-alg/9504011)
10. F. Göhmann, "Bethe Ansatz" (survey, 2023). arXiv:2309.02008, https://arxiv.org/abs/2309.02008
11. P. N. Bibikov, "Thermodynamics of the XX spin chain in the QTM approach" (2025). arXiv:2503.21831, https://arxiv.org/abs/2503.21831