Edge-region 2-string deviations, special-object numerics, and counting integers in the excited-state nonlinear integral equations of the spin-1 XXZ chain

The integrable spin-1 XXZ (Fateev-Zamolodchikov / Izergin-Korepin 19-vertex) chain admits a nonlinear-integral-equation (NLIE) description of its excited-state finite-size spectrum and Bethe roots. In this framework the imaginary deviations of the quasi two-strings from the string-hypothesis positions are given analytically by $\delta_j = (1/N)\,\mathrm{Re}\,K(x_j)/Z_0'(x_j)$, but only where the 2-string density $Z_0'$ is not too small — the bulk $|x| \lesssim \ln N$ of the root distribution. Three linked gaps are left open by the founding papers of the method (J. Suzuki 2004; Á. Hegedűs 2007) and remain open in the literature: (i) no analytic formulas or even qualitative structure for the 2-string deviations and root configurations in the sparse edge region of the distribution, where the density vanishes and ordinary special objects occur; (ii) no stable numerical scheme for solving the NLIE in the presence of virtual or ordinary special objects (the standard large-$N$ iteration fails to converge there); and (iii) no systematic rule fixing the integers $K_W^{\pm}$, $N_{\tilde a\pm}$ and $N_{II}^{\pm}$ that enter the kink and counting equations for arbitrary excited-state patterns. Closing these gaps is the prerequisite for extracting Bethe-root data for arbitrary excited states at large $N$, the input required by the determinant form-factor series for correlation functions.

Background

The integrable spin-1 XXZ chain is the lattice model obtained from the Izergin-Korepin 19-vertex solution of the Yang-Baxter equation (also called the Fateev-Zamolodchikov chain); its Hamiltonian is the logarithmic derivative of the fused (spin-2) transfer matrix of the 19-vertex model, and it appears both as the one-loop anomalous-dimension operator of self-dual large-$N_c$ QCD operators and as a lattice regularization of the $\mathcal{N}=1$ supersymmetric sine-Gordon model. It is diagonalized by the Bethe ansatz: eigenstates are parametrized by $M$ complex roots $\theta_j$ solving algebraic Bethe equations, with $N$ the chain length ($S^z = N - M$ is the conserved third component of the total spin). In the antiferromagnetic thermodynamic limit the roots organize into 'strings' — regular complex patterns. The string hypothesis of Takhtajan, Babujian and Kirillov-Reshetikhin predicts that in the ground state the roots form exact 2-strings, i.e. pairs $\theta = x \pm i\pi/2$ with $x$ real, whose deviation from the exact string position is exponentially small in $N$. Numerical and analytic work in the late 1980s (Avdeev-Dörfel; de Vega-Woynarovich; Klümper-Batchelor-Pearce) showed this is false: because the number of 2-strings is $N/2$ minus $O(1)$, the accumulated effect of their mutual interaction makes the deviations only of order $1/N$, e.g. $\delta_j = (\ln 2/2N)\cosh x_j$ for the ground state. These $1/N$ deviations therefore survive the large-$N$ limit and must be controlled exactly in any finite-size computation.

The tool of choice for finite-size effects is the Destri-de Vega type nonlinear integral equation (NLIE), a single integral equation for an auxiliary counting function whose analytic structure encodes the whole spectrum of scaling dimensions. J. Suzuki (J. Phys. A 37 (2004) 11957) derived excited-state NLIE for this chain in the repulsive regime, and Á. Hegedűs (J. Phys. A 40 (2007) 12007) extended them to the attractive regime, where the anisotropy $\gamma = \pi/(p+2)$ obeys $\pi/3 < \gamma < \pi/2$ ($0 < p < 1$). In these equations an arbitrary excited state is encoded by finitely many 'source objects': holes (missing roots), close and wide complex source pairs, self-conjugated roots, and two types of 'special objects' — virtual special objects (contour artifacts whose number depends only on the chosen integration contour) and ordinary special objects (solutions of an auxiliary counting equation that typically live at the rarely populated edges of the root distribution, where the driving term becomes small). The framework yields the counting function $Z_0(\theta)$ of the 2-string centers, and linearizing the Bethe quantization condition around the string position gives the closed formula $\delta_j = (1/N)\,\mathrm{Re}\,K(x_j)/Z_0'(x_j)$, where $K$ is a function determined by the source-object pattern. Since $Z_0'$ is the density of 2-string centers, the formula is valid precisely where that density is not small. The same analysis proves, for all two-hole states in the repulsive regime, that the deviations change sign at the positions of the holes — a statement verified numerically long ago by Avdeev and Dörfel and expected to persist for multi-hole states. But at the outer edges of the root distribution, $|x| \gtrsim \ln N$, the density $Z_0'$ vanishes (or becomes negative where ordinary special objects sit), the linearization fails, the $O(\delta^2)$ terms are no longer controlled, and the source paper states explicitly that 'nothing definite can be said' there. The same edges and special objects are responsible for a second defect: whenever any special object appears in the NLIE, the standard large-$N$ iterative solution fails to converge, so the equations can then only be manipulated analytically. Finally, the kink equations that govern the conformal limit and the general counting formulae contain integer constants ($K_W^{\pm}$, $N_{\tilde a\pm}$, $N_{II}^{\pm}$) whose values are fixed only by the relative positions of the wide or complex effective roots; the papers determine them case by case, with no general rule.

The motivation for completing this description is quantitative: the form factors of local spin operators in higher-spin XXZ chains are determinants whose entries are known analytic functions of the Bethe roots of the eigenstates involved, so correlation functions can be assembled as form-factor series — but only if the roots of many excited states can be produced reliably at large $N$, which requires exactly the three missing ingredients above.

Problem Statement

For the integrable spin-1 XXZ chain (Fateev–Zamolodchikov / Izergin–Korepin 19-vertex model) of even length $N$ with anisotropy $\gamma = \pi/(p+2)$, $0 < p$, complete the analytic and numerical description of arbitrary excited states by the excited-state nonlinear integral equations (Suzuki 2004 for the repulsive regime; Hegedűs 2007 for the attractive regime), resolving the three linked gaps that these papers explicitly leave open:

(1) Edge region of the root distribution. The linearized deviation formula $\delta_j = (1/N)\,\mathrm{Re}\,K(x_j)/Z_0'(x_j)$ for quasi 2-strings centered at real $x_j$ holds only where the 2-string density $Z_0'$ is not small, i.e. for $|x_j| \lesssim \ln N$. Determine the analytic behaviour of the 2-string deviations and of the full root configuration in the sparse outer region $|x_j| \gtrsim \ln N$ where the density $Z_0'$ vanishes, becomes negative, or ordinary special objects occur: derive the leading finite-$N$ formulas (or prove their form), and decide precisely when and how the sign of the deviations at the positions of holes is or is not modified by ordinary special objects, extending the bulk statement that deviations change sign at hole positions.

(2) Stable numerics with special objects. Construct a numerical scheme that solves the NLIE in formulations containing virtual or ordinary special objects, for which the standard large-$N$ iteration provably fails to converge, and by which the individual Bethe roots (in particular the 2-string deviations, including those near the edges) of large families of excited states can be extracted from the NLIE solution.

(3) Systematic integers. Replace the case-by-case determination of the integers $K_W^{\pm}$, $N_{\tilde a\pm}$ and $N_{II}^{\pm}$ that enter the kink equations and the counting formulae (and hence the conformal weights and the classification of source-object patterns) by a systematic rule valid for arbitrary excited-state root configurations.

The single objective of these three components is a convergent, uniformly valid NLIE (or provably equivalent) description of every excited state of the spin-1 XXZ chain, delivering its Bethe-root data at large $N$ — the prerequisite for the determinant form-factor series of correlation functions. An answer must state its results for the homogeneous chain in both the repulsive ($1 < p$) and attractive ($0 < p < 1$) regimes, and must verify them against direct numerical solution of the Bethe ansatz equations for finite chains (as in Avdeev–Dörfel and Vlijm–Caux type computations) and against the known bulk limits and the ground-state formula $\delta_j = (\ln 2/2N)\cosh x_j$. The extension to higher-spin chains is an acknowledged outlook of the source, not a required deliverable.

References

1. Á. Hegedűs, Finite size effects and 2-string deviations in the spin-1 XXZ chains, J. Phys. A 40 (2007) 12007–12032. DOI: 10.1088/1751-8113/40/40/001, https://doi.org/10.1088/1751-8113/40/40/001 (preprint arXiv:0706.1411, https://arxiv.org/abs/0706.1411)
2. J. Suzuki, Excited state nonlinear integral equations for an integrable anisotropic spin-1 chain, J. Phys. A 37 (2004) 11957–11969. DOI: 10.1088/0305-4470/37/50/002, https://doi.org/10.1088/0305-4470/37/50/002 (preprint hep-th/0410243, https://arxiv.org/abs/hep-th/0410243)
3. H. J. de Vega and F. Woynarovich, Solution of the Bethe ansatz equations with complex roots for finite size: the spin S ≥ 1 isotropic and anisotropic chains, J. Phys. A 23 (1990) 1613–1626. DOI: 10.1088/0305-4470/23/9/022, https://doi.org/10.1088/0305-4470/23/9/022
4. 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
5. L. A. Takhtajan, The picture of low-lying excitations in the isotropic Heisenberg chain of arbitrary spins, Phys. Lett. A 87 (1982) 479–482. DOI: 10.1016/0375-9601(82)90764-2, https://doi.org/10.1016/0375-9601(82)90764-2
6. H. M. Babujian, Exact solution of the isotropic Heisenberg chain with arbitrary spins: thermodynamics of the model, Nucl. Phys. B 215 (1983) 317–336. DOI: 10.1016/0550-3213(83)90668-5, https://doi.org/10.1016/0550-3213(83)90668-5
7. A. N. Kirillov and N. Yu. Reshetikhin, Exact solution of the integrable XXZ Heisenberg model with arbitrary spin. I. The ground state and the excitation spectrum, J. Phys. A 20 (1987) 1565–1585. DOI: 10.1088/0305-4470/20/6/038, https://doi.org/10.1088/0305-4470/20/6/038
8. L. V. Avdeev and B.-D. Dörfel, The Bethe ansatz equations for the isotropic Heisenberg antiferromagnet of arbitrary spin, Nucl. Phys. B 257 (1985) 253–270. DOI: 10.1016/0550-3213(85)90345-1, https://doi.org/10.1016/0550-3213(85)90345-1; and L. V. Avdeev and B.-D. Dörfel, Solutions of the Bethe ansatz equations for XXX antiferromagnet of arbitrary spin in the case of a finite number of sites, Theor. Math. Phys. 71 (1987) 528–542. DOI: 10.1007/BF01028653, https://doi.org/10.1007/BF01028653
9. F. C. Alcaraz and M. J. Martins, Conformal invariance and the operator content of the XXZ model with arbitrary spin, J. Phys. A 22 (1989) 1829–1858. DOI: 10.1088/0305-4470/22/11/023, https://doi.org/10.1088/0305-4470/22/11/023
10. Á. Hegedűs, F. Ravanini and J. Suzuki, Exact finite size spectrum in super sine-Gordon model, Nucl. Phys. B 763 (2007) 330–353. DOI: 10.1016/j.nuclphysb.2006.11.006, https://doi.org/10.1016/j.nuclphysb.2006.11.006 (preprint hep-th/0610012, https://arxiv.org/abs/hep-th/0610012)
11. R. Murgan, On the NLIE of (inhomogeneous) open spin-1 XXZ chain with general integrable boundary terms, JHEP 02 (2011) 016. DOI: 10.1007/JHEP02(2011)016, https://doi.org/10.1007/JHEP02(2011)016 (preprint arXiv:1012.1055, https://arxiv.org/abs/1012.1055)
12. C. Matsui, Spinon excitations in the spin-1 XXZ chain and hidden supersymmetry, Nucl. Phys. B 913 (2016) 15–33. DOI: 10.1016/j.nuclphysb.2016.09.002, https://doi.org/10.1016/j.nuclphysb.2016.09.002 (preprint arXiv:1607.04317, https://arxiv.org/abs/1607.04317)
13. R. Vlijm and J.-S. Caux, Computation of dynamical correlation functions of the spin-1 Babujan–Takhtajan chain, J. Stat. Mech. (2014) P05009. DOI: 10.1088/1742-5468/2014/05/P05009, https://doi.org/10.1088/1742-5468/2014/05/P05009 (preprint arXiv:1401.4450, https://arxiv.org/abs/1401.4450)
14. T. Deguchi and P. Ranjan Giri, Exact quantum numbers of collapsed and non-collapsed two-string solutions in the spin-1/2 Heisenberg spin chain, J. Phys. A 49 (2016) 174001. DOI: 10.1088/1751-8113/49/17/174001, https://doi.org/10.1088/1751-8113/49/17/174001 (preprint arXiv:1509.00108, https://arxiv.org/abs/1509.00108)
15. T. Imoto, J. Sato and T. Deguchi, Exact regimes of collapsed and extra two-string solutions in the two down-spin sector of the spin-1/2 massive XXZ spin chain, J. Phys. A 52 (2019) 035203. DOI: 10.1088/1751-8121/aaf29f, https://doi.org/10.1088/1751-8121/aaf29f (preprint arXiv:1807.08885, https://arxiv.org/abs/1807.08885)