The sine-Gordon model carries an infinite tower of mutually commuting local conserved charges of odd Lorentz spin, and the corresponding generalized Gibbs ensemble (GGE) couples any finite subset of them to the density matrix through generalized inverse temperatures. Á. Hegedűs (Nucl. Phys. B 1025 (2026) 117385, arXiv:2510.25344) derived two nonlinear integral equation (NLIE) systems for this ensemble: NLIE II handles an arbitrary number of higher-spin charges but only in the repulsive regime p>1p>1, while NLIE I penetrates the attractive regime only in the window p>1−12N−1p>1-\frac{1}{2N-1}, where 2N−12N-1 is the largest spin included. That window is an artifact of the derivation: it demands that every included charge be individually bounded from below on one-particle breather states, a strictly stronger condition than boundedness of the full generalized Hamiltonian, which can hold on strictly larger coupling ranges. The open problem is to construct an analytically controlled and numerically practicable integral-equation (or equivalent compact) formulation of the GGE thermodynamics and its observables that is valid throughout the domain where the generalized Hamiltonian is bounded from below, together with the validity analysis (analytic continuation in the coupling, solvability of the equations) that establishes the extension.

Background

The sine-Gordon model is a paradigmatic integrable relativistic quantum field theory in 1+11+1 dimensions with Euclidean action S=∫dx dτ {12(∂μϕ)2−2κ2sin⁡(βSG2/8)cos⁡(βSGϕ)}S=\int dx\,d\tau\,\{\frac{1}{2}(\partial_\mu\phi)^2-\frac{2\kappa^2}{\sin(\beta_{SG}^2/8)}\cos(\beta_{SG}\phi)\} with 0<βSG2<8π0<\beta_{SG}^2<8\pi. The coupling is conventionally parameterized by p=βSG2/(8π−βSG2)p=\beta_{SG}^2/(8\pi-\beta_{SG}^2): for p>1p>1 (repulsive regime) the spectrum contains only solitons and antisolitons; for 0<p<10<p<1 (attractive regime) soliton–antisoliton bound states called breathers appear, with only the first breather B1B_1 present in the range 12<p<1\frac{1}{2}<p<1; p=1p=1 is the free-fermion point. Particle states are labeled by a rapidity θ\theta (with energy mcosh⁡θm\cosh\theta and momentum msinh⁡θm\sinh\theta).

Integrability supplies an infinite set of mutually commuting local conserved charges H^2j−1\hat{\mathcal{H}}_{2j-1}, j∈Zj\in\mathbb{Z}, of odd Lorentz spin 2j−12j-1. On a one-particle soliton or antisoliton state of rapidity θ\theta they act as H^2j−1∣θ,±⟩=e(2j−1)θ∣θ,±⟩\hat{\mathcal{H}}_{2j-1}|\theta,\pm\rangle=e^{(2j-1)\theta}|\theta,\pm\rangle, with the same eigenvalue on the soliton and the antisoliton; since jj ranges over all integers, the charge of spin −(2j−1)-(2j-1) has eigenvalue e−(2j−1)θe^{-(2j-1)\theta}. A generalized Gibbs ensemble (GGE) is the statistical ensemble with density matrix proportional to exp⁡(−∑jβ2j−1H^2j−1)\exp(-\sum_j\beta_{2j-1}\hat{\mathcal{H}}_{2j-1}), where the β2j−1\beta_{2j-1} are generalized inverse temperatures (also called potentials); it is the accepted description of the long-time steady state of generic non-equilibrium states in integrable systems and the baseline of generalized hydrodynamics (the hydrodynamic theory of integrable systems). Its partition function exists only if the generalized Hamiltonian H^gen(β‾)=1β∑jβ2j−1H^2j−1\hat{\mathcal{H}}_{gen}(\underline{\beta})=\frac{1}{\beta}\sum_j\beta_{2j-1}\hat{\mathcal{H}}_{2j-1} is bounded from below on the spectrum.

The thermodynamic Bethe ansatz (TBA) — the standard technique computing the free-energy density of integrable models in infinite volume through coupled integral equations for pseudoenergies — describes this GGE, but at generic coupling the equations are organized in strongly coupling-dependent, case-specific ways. At integer values p=2,3,…p=2,3,\dots the GGE TBA equations form a system for functions yj(x)y_j(x) structured by the Dp+1\mathcal{D}_{p+1} Dynkin diagram (Zamolodchikov's reflectionless ADE TBA), with source term SN(x∣β‾)=∑k=1N{β2k−1e(2k−1)x+β−(2k−1)e−(2k−1)x}\mathcal{S}_N(x|\underline{\beta})=\sum_{k=1}^{N}\{\beta_{2k-1}e^{(2k-1)x}+\beta_{-(2k-1)}e^{-(2k-1)x}\} built from the finitely many retained charges.

Nonlinear integral equation (NLIE) methods — originating with Klümper, Batchelor and Pearce and with Destri and de Vega — reorganize such TBA systems into a few coupled nonlinear integral equations whose unknowns are combinations of the TBA functions, with the coupling dependence absorbed into kernels, so that analytic continuation in the coupling becomes possible. The derivation passes through the YY-system and TT-QQ functional relations satisfied by the TBA functions and requires bookkeeping of analyticity strips: vertical strips in the complex rapidity variable inside which the unknown functions are analytic, nonzero, and free of singularities.

Á. Hegedűs (Nucl. Phys. B 1025 (2026) 117385, arXiv:2510.25344) derived two such NLIE systems for the sine-Gordon GGE. NLIE II, a hybrid NLIE that keeps the massive TBA node itself as one of the unknowns, is valid for 1<p<∞1<p<\infty with an arbitrary number of higher-spin charges. NLIE I, a Klümper–Batchelor–Pearce–Destri–de Vega type equation, is valid for 1−12N−1<p<∞1-\frac{1}{2N-1}<p<\infty where 2N−12N-1 is the largest spin included, and thus penetrates a coupling window of the attractive regime whose width shrinks as the maximal spin grows. Within their domains both formulations compute the GGE free-energy density and the free-energy flux (the generating current of generalized hydrodynamics), generate expectation values of conserved-charge densities and currents by differentiation, and give expectation values of vertex operators e±imβSGϕe^{\pm im\beta_{SG}\phi} through the fermionic-basis determinant formulas of Jimbo, Miwa and Smirnov (determinant representations arising from the hidden Grassmann structure of the XXZ/sine-Gordon family), with the β‾\underline{\beta}-dependent matrix elements expressed as NLIE integrals.

The restriction of NLIE I has a sharp physical characterization. In the attractive regime the first breather contributes to the partition function, and by the factorized-scattering bootstrap (constructing bound-state data from the soliton scattering amplitudes) its one-particle eigenvalues are HB1,±(2n−1)(θ)=(−1)n+1 2sin⁡[(2n−1)pπ2] e±(2n−1)θ\mathcal{H}_{B_1,\pm(2n-1)}(\theta)=(-1)^{n+1}\,2\sin[(2n-1)\frac{p\pi}{2}]\,e^{\pm(2n-1)\theta}, which are bounded from below in θ\theta in the window 1−12n−1<p1-\frac{1}{2n-1}<p below the free-fermion point (for a single high spin the oscillating sine factor is positive also in further windows, but for the full ladder of retained spins up to 2N−12N-1 simultaneous individual boundedness is equivalent to 1−12N−1<p1-\frac{1}{2N-1}<p, which is the range of validity of NLIE I). Consequently the collective generalized Hamiltonian can be bounded from below on coupling ranges where the individual higher-spin charges are not: for example H^gen=H^3+β9′H^9\hat{\mathcal{H}}_{gen}=\hat{\mathcal{H}}_{3}+\beta'_9\hat{\mathcal{H}}_{9} with β9′>0\beta'_9>0 is bounded from below for 49<p<23\frac{4}{9}<p<\frac{2}{3}, strictly outside the NLIE I window p>89p>\frac{8}{9} for spin-9 charges, because the spin-9 breather coefficient 2sin⁡(9pπ/2)2\sin(9p\pi/2) is positive throughout this range, so the faster-growing spin-9 term dominates the (individually unbounded) spin-3 one as θ→+∞\theta\to+\infty, while both contributions decay as θ→−∞\theta\to-\infty. The paper states that the straightforward NLIE extension is no longer valid in these regions, that a more sophisticated set of equations would be necessary, and explicitly does not exclude that a more refined approach than the TBA can be developed. Three further limitations are acknowledged: the analyticity strips of the NLIE I unknowns shrink as the maximal spin grows, so at large NN the integration contour must run close to lines where singularities condense, making the numerics stiff (while NLIE II loses efficiency as p↓1p\downarrow1 where its contour approaches the first pole of the kernel); the analytic continuation from the integer derivation points to continuous pp is asserted as straightforward rather than rigorously controlled, with solutions assumed to be smooth deformations of the constant β‾→∞\underline{\beta}\to\infty solution and no existence/uniqueness analysis; and the vertex-operator determinant formulas were not numerically evaluated beyond Gibbs-potential continuity checks at the free-fermion point.

Problem Statement

Provide a controlled description — capable of supporting quantitative computation — of the generalized Gibbs ensemble of the sine-Gordon model with higher-spin conserved charges throughout the parameter domain where the GGE exists, i.e. where the generalized Hamiltonian H^gen(β‾)\hat{\mathcal{H}}_{gen}(\underline{\beta}) is bounded from below on the particle spectrum. The description should take the form of a compact system of nonlinear integral equations for a few unknown functions, an equivalent analytically controlled reformulation, or a demonstrated reliable computational scheme. This domain strictly contains the NLIE I window 1−12N−1<p1-\frac{1}{2N-1}<p (with 2N−12N-1 the largest spin included). For example, for the ensemble H^3+β9′H^9\hat{\mathcal{H}}_3+\beta'_9\hat{\mathcal{H}}_9 with positive potentials the collective Hamiltonian is bounded from below on the soliton, antisoliton and first-breather spectrum for 49<p<23\frac{4}{9}<p<\frac{2}{3}, even though the individual spin-3 charge is unbounded from below there; in the sector 12<p<1\frac{1}{2}<p<1, where only the first breather contributes beyond the soliton and antisoliton, this is a coupling range in which the GGE exists but NLIE I does not apply, and at smaller pp, where further breathers enter the spectrum, the domain is delimited by their bootstrap eigenvalues wherever the GGE partition function exists.

The single common objective is to remove the individual-boundedness artifact of the existing NLIE derivations without losing their advantages. The source paper states that the straightforward extension of the NLIE method is no longer valid in the coupling regions where the generalized Hamiltonian remains bounded from below although the individual higher-spin charges are not, that a more sophisticated set of equations would be necessary to describe the thermodynamics of the model accurately there, and that it does not claim that no alternative to NLIE I exists in the regions where it fails.

References

  1. Á. Hegedűs, 'NLIE formulations for the generalized Gibbs ensemble in the sine-Gordon model', Nucl. Phys. B 1025 (2026) 117385, doi:10.1016/j.nuclphysb.2026.117385, arXiv:2510.25344, https://arxiv.org/abs/2510.25344
  2. Á. Hegedűs, 'Thermodynamics in the sine-Gordon model: the NLIE approach', Nucl. Phys. B 1020 (2025) 117155, doi:10.1016/j.nuclphysb.2025.117155, arXiv:2507.19200, https://arxiv.org/abs/2507.19200
  3. Á. Hegedűs and A. Roth, 'Leading UV Formula for Finite-Volume Vertex Operator Expectation Values in the Sine-Gordon Model from Kink NLIE', arXiv:2606.20018, https://arxiv.org/abs/2606.20018
  4. N. Brizio, T. Morone, N. Primi and R. Tateo, 'Graded S-matrices, generalised Gibbs ensembles and fractional-spin CDD deformations', JHEP 03 (2026) 137, doi:10.1007/JHEP03(2026)137, arXiv:2511.03791, https://arxiv.org/abs/2511.03791
  5. B. C. Nagy, M. Kormos and G. Takács, 'Full counting statistics in the sine-Gordon model', arXiv:2601.05079, https://arxiv.org/abs/2601.05079
  6. M. Tóth, J. H. Pixley, G. Takács and M. Kormos, 'Finite temperature correlation functions of the sine--Gordon model', arXiv:2604.12585, https://arxiv.org/abs/2604.12585
  7. B. C. Nagy, G. Takács and M. Kormos, 'Thermodynamic Bethe Ansatz and generalised hydrodynamics in the sine-Gordon model', SciPost Phys. 16 (2024) 145, doi:10.21468/SciPostPhys.16.6.145, arXiv:2312.03909, https://arxiv.org/abs/2312.03909
  8. M. Jimbo, T. Miwa and F. Smirnov, 'Hidden Grassmann structure in the XXZ model V: Sine-Gordon model', Lett. Math. Phys. 96 (2011) 325-365, doi:10.1007/s11005-010-0438-9, arXiv:1007.0556, https://arxiv.org/abs/1007.0556
  9. Á. Hegedűs, 'Finite volume expectation values in the sine-Gordon model', JHEP 01 (2020) 122, doi:10.1007/JHEP01(2020)122, arXiv:1909.08467, https://arxiv.org/abs/1909.08467
  10. C. Destri and H. J. de Vega, 'New thermodynamic Bethe ansatz equations without strings', Phys. Rev. Lett. 69 (1992) 2313-2317, doi:10.1103/PhysRevLett.69.2313, https://doi.org/10.1103/PhysRevLett.69.2313
  11. C. Destri and H. J. de Vega, 'Unified approach to thermodynamic Bethe Ansatz and finite size corrections for lattice models and field theories', Nucl. Phys. B 438 (1995) 413-454, doi:10.1016/0550-3213(94)00547-R, arXiv:hep-th/9407117, https://arxiv.org/abs/hep-th/9407117
  12. A. Klümper, M. T. Batchelor and P. A. Pearce, 'Central charges of the 6- and 19-vertex models with twisted boundary conditions', J. Phys. A 24 (1991) 3111-3133, doi:10.1088/0305-4470/24/13/025, https://doi.org/10.1088/0305-4470/24/13/025
  13. Al. B. Zamolodchikov, 'On the thermodynamic Bethe ansatz equations for reflectionless ADE scattering theories', Phys. Lett. B 253 (1991) 391-394, doi:10.1016/0370-2693(91)91737-G, https://doi.org/10.1016/0370-2693(91)91737-G