The open spin-1/2 XXZ chain with diagonal boundary fields admits an exact quantum-transfer-matrix representation of its surface (boundary) free energy whose boundary-field-independent O(1) terms - Fredholm-determinant-type contributions arising from the rapidity density of states and the quadratic fluctuations around the saddle point - are missed by the standard thermodynamic Bethe Ansatz (TBA) treatments of the open chain. Whether the TBA, properly extended, can reproduce the full boundary free energy (the lattice analogue of the exact g-function of boundary quantum field theory) has been open since Kozlowski and Pozsgay posed the question in 2012. Subsequent exact results obtained by the overlap/quantum-transfer-matrix route at zero bulk magnetic field (Pozsgay and Rákos, 2018) confirm the presence and structure of the two O(1) Fredholm determinants but leave the TBA-side derivation, the boundary-level equivalence between the quantum-transfer-matrix and TBA frameworks, and the entire nonzero-bulk-field regime - where string excitations enter and the overlap method breaks down - unresolved.

Background

The spin-12\frac{1}{2} XXZ chain is the paradigmatic exactly solvable one-dimensional quantum lattice model, with nearest-neighbour interactions whose anisotropy is parametrized by η\eta (massive/easy-axis regime for real η\eta, massless/easy-plane regime for purely imaginary η\eta). On a chain of MM sites with open ends one may add diagonal boundary magnetic fields, parametrized by angles ξ±\xi^{\pm}, without destroying integrability. The surface or boundary free energy is the volume-independent part of the free energy, fsurf(T)=lim⁡M→∞(Fopen(M,T)−M fbulk(T))f_{\rm surf}(T)=\lim_{M\to\infty}\big(F_{\rm open}(M,T)-M\,f_{\rm bulk}(T)\big); it is the lattice counterpart of the boundary free energy (or logarithm of the g-function) of an integrable quantum field theory with boundaries, which measures the boundary degrees of freedom.

Two finite-temperature frameworks exist for Bethe-ansatz-solvable models. The thermodynamic Bethe Ansatz (TBA) rewrites the finite-volume Bethe equations in the thermodynamic limit as integral equations for a dressed energy (the excitation energy including the backflow effect of the surrounding quasiparticle sea) or counting function, with source terms for the boundaries, and evaluates the partition function at its saddle point. For the open XXZ chain this requires the string hypothesis, i.e. the organization of the Bethe roots (the rapidities parametrizing eigenstates) into complexes with a common real part and regularly spaced imaginary parts. The alternative quantum transfer matrix (QTM) method represents the partition function as the dominant eigenvalue of a transfer matrix in the imaginary-time direction, obtained through a Suzuki-Trotter decomposition (a slicing of the imaginary-time evolution into small discrete steps); sending the Trotter number to infinity leads to non-linear integral equations (NLIE) for a single auxiliary function. For the bulk free energy the equivalence of the two frameworks was established by Takahashi, Shiroishi and Klümper (2001); no analogous statement is known at the level of the boundary free energy.

Early TBA computations of the surface free energy of the open chain (de Sa and Tsvelik 1995; Frahm and Zvyagin 1997; Zvyagin and Makarova 2004) captured the boundary-field-dependent part but disagreed with field-theory predictions in the low-temperature limit. The origin of the discrepancy was identified by Woynarovich (2004, arXiv:cond-mat/0402129; 2011, arXiv:1007.1148) and Pozsgay (2010, arXiv:1003.5542): a careful evaluation of the Bethe-ansatz partition function - keeping both the density of states in rapidity space and the Gaussian (quadratic) fluctuations around the TBA saddle point - produces additional volume-independent O(1) contributions, expressible through Fredholm determinants, i.e. determinants det⁡(1+K^)\det(1+\hat K) of integral operators acting in rapidity space. In continuum (relativistic) Bethe-ansatz systems with diagonal scattering these O(1) terms coincide with the exact g-function of Dorey, Fioravanti, Rim and Tateo (2004, arXiv:hep-th/0404014), and Pozsgay (2010) proved that the TBA does generate them in that setting. None of these methods had been applied to the lattice XXZ chain.

Göhmann, Bortz and Frahm (2005) initiated an exact route: a finite-Trotter-number representation of the surface free energy in terms of expectation values of a finite-temperature boundary operator in the dominant QTM eigenvector. Kozlowski and Pozsgay (2012) recast this representation through Tsuchiya's determinant formula for the six-vertex model with reflecting ends (the six-vertex model is the two-dimensional classical statistical-mechanics counterpart of the XXZ chain, and reflecting ends encode the open boundary) and took the infinite-Trotter-number limit, obtaining an exact representation of fsurff_{\rm surf} in terms of the solution a(ω)a(\omega) of the NLIE governing the periodic chain. Besides the boundary-field-dependent pieces B(ξ±)B(\xi^{\pm}), this representation contains boundary-field-independent O(1) terms: a Fredholm-type determinant Tln⁡det⁡C[I+K]T\ln\det_{\mathcal C}[I+K], logarithmic terms such as T2ln⁡[1−areg2(0)]\tfrac{T}{2}\ln[1-a_{\rm reg}^{2}(0)], and an infinite series F\mathcal F of nested multiple integrals - exactly the type of contributions absent from the earlier TBA results. The authors concluded that it was not clear whether the TBA in its present setting is capable of providing the full boundary free energy, and left the confrontation of the two frameworks for future work.

The exact answer itself subsequently advanced from outside the TBA. Pozsgay and Rákos (2018) derived, at zero bulk magnetic field but for arbitrary boundary fields, a compact exact formula for the boundary free energy - single integrals plus two simple one-dimensional Fredholm determinants, i.e. the O(1)-complete answer - using the QTM together with an overlap formula between QTM eigenstates and boundary states (the states obtained by joining the two open ends of the chain, whose overlaps with finite-volume eigenstates encode the boundary partition function; conjectural for generic boundary K-matrices, the operators encoding the boundary fields, and proven in the diagonal case). They stressed that a TBA derivation would be considerably more complicated because the fluctuation operators act also on string indices, and that the TBA is the natural route at nonzero bulk field, where the pair structure of Bethe roots on which the overlap method relies is lost. In the adjacent continuum setting, He and Jiang (2024) computed the exact sine-Gordon g-function - a scalar part plus a universal ratio of Fredholm determinants - from an integrable lattice regularization and an NLIE for the counting function, bypassing the TBA after Vu, Kostov and Serban (2019, arXiv:1906.01909) found that a naive boundary TBA diverges for non-diagonal scattering theories (perturbed Wess-Zumino-Novikov-Witten (WZNW) models, where only ultraviolet-infrared differences are finite); Kostov, Serban and Vu (2019, arXiv:1809.05705) reproduced the diagonal-scattering g-function from a matrix-tree expansion. A proper modification of the TBA for boundary free energies is explicitly listed as an open direction.

Problem Statement

Can the thermodynamic Bethe Ansatz, extended to keep track of the sub-leading volume-independent O(1) contributions - the density-of-states and quadratic-fluctuation terms, including their action on string degrees of freedom - reproduce the full surface free energy of the open spin-1/2 XXZ chain with diagonal boundary fields, and can the equivalence between the quantum-transfer-matrix and TBA frameworks be established at the level of the boundary free energy? The common objective is a TBA-side representation of the complete boundary free energy, with the following jointly supporting deliverables:

(i) At zero bulk magnetic field, derive within the TBA the boundary-field-independent O(1) terms - the Fredholm-determinant contributions identified exactly by the quantum-transfer-matrix/overlap method - thereby proving (or disproving) that the TBA is a complete framework for boundary free energies of the lattice chain, in both the massive and massless regimes.

(ii) At nonzero bulk magnetic field - where string excitations participate and the overlap/quantum-transfer-matrix simplifications relied upon at zero field break down - produce the TBA prediction for the full boundary free energy, boundary-field-dependent and O(1) parts alike, and confront it with the exact representation of Kozlowski-Pozsgay (2012) (analytic identity, arbitrary-precision numerical comparison, or controlled finite-size benchmarks from exact diagonalization or finite-Trotter transfer-matrix data).

Components (i) and (ii) jointly constitute a positive resolution of this question, and a derivation in one regime alone is a partial result. The boundary magnetization within the TBA, which Kozlowski and Pozsgay suggested as a less ambitious intermediate target, is an admissible milestone but does not by itself constitute a solution.

References

  1. K. K. Kozlowski and B. Pozsgay, 'Surface free energy of the open XXZ spin-1/2 chain', J. Stat. Mech. (2012) P05021, DOI: 10.1088/1742-5468/2012/05/P05021, arXiv:1201.5884, https://arxiv.org/abs/1201.5884
  2. F. Göhmann, M. Bortz and H. Frahm, 'Surface free energy for systems with integrable boundary conditions', J. Phys. A: Math. Gen. 38 (2005) 10879-10892, DOI: 10.1088/0305-4470/38/50/001, arXiv:cond-mat/0508377, https://arxiv.org/abs/cond-mat/0508377
  3. P. de Sa and A. M. Tsvelik, 'Anisotropic spin-1/2 Heisenberg chain with open boundary conditions', Phys. Rev. B 52 (1995) 3067-3070, DOI: 10.1103/PhysRevB.52.3067, https://doi.org/10.1103/PhysRevB.52.3067
  4. H. Frahm and A. A. Zvyagin, 'The open spin chain with impurity: an exact solution', J. Phys.: Condens. Matter 9 (1997) 9939, DOI: 10.1088/0953-8984/9/45/021, https://doi.org/10.1088/0953-8984/9/45/021
  5. A. A. Zvyagin and A. V. Makarova, 'Bethe-ansatz study of the low-temperature thermodynamics of an open Heisenberg chain', Phys. Rev. B 69 (2004) 214430, DOI: 10.1103/PhysRevB.69.214430, https://doi.org/10.1103/PhysRevB.69.214430
  6. F. Woynarovich, 'O(1) contribution of saddle point fluctuations to the free energy of Bethe Ansatz systems', Nucl. Phys. B 700 (2004) 331, DOI: 10.1016/j.nuclphysb.2004.08.043, arXiv:cond-mat/0402129, https://arxiv.org/abs/cond-mat/0402129
  7. F. Woynarovich, 'On the normalization of the partition function of Bethe Ansatz systems', Nucl. Phys. B 852 (2011) 269-286, DOI: 10.1016/j.nuclphysb.2011.06.015, arXiv:1007.1148, https://arxiv.org/abs/1007.1148
  8. B. Pozsgay, 'On O(1) contributions to the free energy in Bethe Ansatz systems: the exact g-function', JHEP 08 (2010) 090, DOI: 10.1007/JHEP08(2010)090, arXiv:1003.5542, https://arxiv.org/abs/1003.5542
  9. P. Dorey, D. Fioravanti, C. Rim and R. Tateo, 'Integrable quantum field theory with boundaries: the exact g-function', Nucl. Phys. B 696 (2004) 445-467, DOI: 10.1016/j.nuclphysb.2004.06.045, arXiv:hep-th/0404014, https://arxiv.org/abs/hep-th/0404014
  10. B. Pozsgay and O. Rákos, 'Exact boundary free energy of the open XXZ chain with arbitrary boundary conditions', J. Stat. Mech. (2018) 113102, DOI: 10.1088/1742-5468/aae5a5, arXiv:1804.09992, https://arxiv.org/abs/1804.09992
  11. Y.-J. He and Y. Jiang, 'Exact g-function without strings', arXiv:2412.12869 (2024), https://arxiv.org/abs/2412.12869
  12. D.-L. Vu, I. Kostov and D. Serban, 'Boundary entropy of integrable perturbed SU(2)_k WZNW', JHEP 08 (2019) 154, DOI: 10.1007/JHEP08(2019)154, arXiv:1906.01909, https://arxiv.org/abs/1906.01909
  13. I. Kostov, D. Serban and D.-L. Vu, 'Boundary TBA, trees and loops', Nucl. Phys. B 949 (2019) 114817, DOI: 10.1016/j.nuclphysb.2019.114817, arXiv:1809.05705, https://arxiv.org/abs/1809.05705
  14. K. K. Kozlowski and V. Terras, 'Multi-point Correlation Functions in the Boundary XXZ Chain at Finite Temperature', Ann. Henri Poincare 25 (2024), DOI: 10.1007/s00023-023-01310-4, arXiv:2212.08824, https://arxiv.org/abs/2212.08824
  15. M. Takahashi, M. Shiroishi and A. Klümper, 'Equivalence of TBA and QTM', J. Phys. A: Math. Gen. 34 (2001) L187-L194, DOI: 10.1088/0305-4470/34/13/105, arXiv:cond-mat/0102027, https://arxiv.org/abs/cond-mat/0102027
  16. Y. Qiao, J. Cao, W.-L. Yang, K. Shi and Y. Wang, 'Exact surface energy and helical spinons in the XXZ spin chain with arbitrary nondiagonal boundary fields', Phys. Rev. B 103 (2021) 220401(R), DOI: 10.1103/PhysRevB.103.L220401, arXiv:2102.02643, https://arxiv.org/abs/2102.02643
  17. P. Lu, J. Cao, W.-L. Yang, I. Marquette and Y.-Z. Zhang, 'Exact surface energies and boundary excitations of the Izergin-Korepin model with generic boundary fields', JHEP 02 (2025) 087, DOI: 10.1007/JHEP02(2025)087, arXiv:2407.21258, https://arxiv.org/abs/2407.21258