Proof of the Kotousov–Lukyanov conjectures for the scaling-limit norms of Bethe states in the critical XXZ Heisenberg chain

For the spin-1/2 XXZ Heisenberg chain in the critical regime (anisotropy $\Delta=\cos(\pi\beta^2)$, $0<\beta^2<1$), Kotousov and Lukyanov conjectured in 2019 the exact large-$L$ asymptotics of the norms $\mathfrak N$ of low-energy Bethe states: $\mathfrak N=\mathfrak N_\infty L^\eta \mathrm e^{A_2L^2+A_1L}(1+o(1))$, with explicit coefficients $A_2$, $A_1$, a state-dependent exponent $\eta$ expressed through the eigenvalue of the first local integral of motion of the continuum conformal field theory, and a fully factorized amplitude whose chiral factors involve an explicit special function $Z_+(p|\beta)$, eigenvalues of reflection operators, and two undetermined constants $C_0$, $C$ depending only on $\beta$. The conjectures rest on the ODE/IQFT correspondence — the identification of the spectral theory of certain Schrödinger-type differential equations with the integrable structure of the continuum quantum field theory — and on numerics at descendant levels $N\le 2$. The problem is to give a rigorous proof (or a controlled refutation) of these asymptotic formulas. The constants $C_0(\beta)$ and $C(\beta)$ are presently determined only numerically, exactly only at the free-fermion point $\beta^2=\tfrac12$. The 2019–2026 follow-up literature extends the conjectures to inhomogeneous and $\mathcal Z_r$-symmetric models and has rigorously classified the underlying monster-potential ODEs, but leaves the norm conjectures themselves unproven.

Background

The spin-$\frac12$ XXZ Heisenberg chain is a one-dimensional quantum lattice model of spins with nearest-neighbour interactions whose anisotropy $\Delta$ interpolates between the isotropic ferromagnet and antiferromagnet. Its low-energy eigenstates are constructed by the Bethe ansatz: each Bethe state is a superposition of plane waves whose momenta (encoded by rapidities $\zeta_j$) are fixed by the Bethe ansatz equations, and the state is labelled by the magnetization sector $S^z$ and a twist parameter $k$ describing quasi-periodic boundary conditions. The normalization sum $\mathfrak N=\sum|\Psi|^2$ over all spin configurations — the squared norm of the Bethe wave function — is the basic ingredient needed to normalize eigenstates and to compute matrix elements and correlation functions. Gaudin, McCoy and Wu expressed $\mathfrak N$ through an $M\times M$ determinant built from the two-body scattering phase (the Gaudin determinant), a formula later proved by Korepin; the result depends on the convention chosen for the two-body factors in the wave function.

In the critical regime $-1<\Delta<1$, written $\Delta=\cos(\pi\beta^2)$ with $0<\beta^2<1$, the chain is massless and its low-energy states flow, as the chain length $L\to\infty$, to states of a two-dimensional conformal field theory (CFT) — a relativistic quantum field theory invariant under local scale transformations, whose spectrum is organized by scaling dimensions. Here the limiting CFT is a free compactified boson: a free scalar field whose values are identified modulo a fixed period, so that field configurations may wind around a circle whose radius is fixed by $\beta$. Bazhanov, Lukyanov and Zamolodchikov (BLZ) uncovered the integrable structure of this CFT: an infinite family of mutually commuting local integrals of motion (IM) $I_{2n-1}$ — deformations of the Virasoro generators (the generators of infinitesimal conformal transformations), with $I_1$ essentially the conformal energy — supplemented by dual, non-local IMs and by $Q$-operators (generating operators whose eigenvalues encode the entire tower of IM eigenvalues) whose eigenvalues obey functional Bethe-ansatz-type equations. The scaling limit of the lattice can be organized into individual renormalization-group (RG) trajectories: each low-energy Bethe state (one whose Bethe quantum numbers differ from the vacuum values only at finitely many places near the edges of the root distribution) converges, after a suitable normalization, to a vector $|\bar{\boldsymbol w}\rangle\otimes|{\boldsymbol w}\rangle$ in a pair of free-boson Fock spaces (state spaces built by acting with creation operators on a vacuum) $\bar{\mathcal F}_{\bar p}\otimes{\mathcal F}_p$, one for each chiral (left-moving and right-moving) sector, where $\boldsymbol w=\{w_j\}_{j=1}^N$ is a finite set of complex parameters labelling the descendant level $N$.

The identification of the labels $\boldsymbol w$ is provided by the ODE/IQFT correspondence (ordinary-differential-equation/integrable-quantum-field-theory correspondence, also called ODE/IM) discovered by Dorey and Tateo and developed by BLZ: the eigenvalue of the CFT $Q$-operator on the state labelled by $\boldsymbol w$ coincides with the spectral determinant of a Schrödinger equation $[-\mathrm d^2/\mathrm dx^2+V(x)-E]\psi=0$ with an anharmonic potential. For descendant states one needs the monster potentials: potentials of the form $x^{2\alpha}$ (with $\alpha=\beta^{-2}-1$) plus an inverse-square term and $N$ additional apparent singularities at positions $\boldsymbol w=\{w_j\}_{j=1}^N$ — points where the potential has poles but all solutions are constrained to remain single-valued (monodromy-free) there — with the positions fixed by algebraic equations. The number of distinct monster potentials with $N$ singularities equals the number of integer partitions of $N$ — the dimension of the level-$N$ subspace of the CFT Fock space — a fact proved by Conti and Masoero.

Against this background, Kotousov and Lukyanov posed the question: what is the large-$L$ behaviour of the norms of the low-energy Bethe states? Their numerics (chains of $10^2$–$10^3$ sites) suggested $\mathfrak N \;=\; \mathfrak N_\infty\, L^{\eta}\, \mathrm e^{A_2 L^2 + A_1 L}\,\bigl(1+o(1)\bigr),$ with $A_2$ given by an explicit integral (eq. (3.6) of their paper) and $A_1=\log\!\bigl[2(1-\beta^2)/\sin(\pi\beta^2)\bigr]$ (eq. (3.7)). They then conjectured: (I) the exponent is $\eta=\frac16-\frac13\rho^2-2\rho\beta S^z-4I_1(\boldsymbol w)-4\bar I_1(\bar{\boldsymbol w})$, where $\rho=\beta^{-1}-\beta$ and $I_1(\boldsymbol w)$ is the eigenvalue of the first local IM on the continuum state, and the amplitude factorizes into chiral parts $\mathfrak N_\infty=\mathcal N_{\bar p}(\bar{\boldsymbol w})\mathcal N_p(\boldsymbol w)$; (II)–(III) the auxiliary amplitudes $K_p(\boldsymbol w)$ — of the auxiliary norm $\mathfrak K=\mathsf U^{-1}\mathfrak N$ (eq. (10.1) of their paper) and related to the chiral parts $\mathcal N_p(\boldsymbol w)$ of $\mathfrak N_\infty$ by eq. (10.6) there — have the form $K_p(\boldsymbol w)=C_0\,C^{h(p)+4N}\,V_p(\boldsymbol w)\,Z_+(p\,|\,\beta)$ with $h(p)=4p^2+\frac16(\rho^2-1)=4p^2+\frac16(\beta^2+\beta^{-2}-3)$, where $Z_+(p|\beta)$ is an explicit special function given by a convergent infinite product, $V_p(\boldsymbol w)$ is the diagonal of the Virasoro Hermitian form on the descendant states (an explicitly known Gram-type factor, eqs. (9.6), (9.1), (9.2) of their paper), $N$ is the descendant level, and $C_0$, $C$ are constants depending only on $\beta$; the eigenvalues $R_p(\boldsymbol w)$ of the reflection operators (the simplest dual non-local IMs) enter the fully factorized amplitude (eq. (11.11) of their paper) through $H_p=r_pV_p$ with $r_p=R_p(\boldsymbol w)/R_p^{(0)}$. These conjectures were verified numerically for levels $N\le 2$ and anisotropies roughly $0.06<\beta^2<0.8$, and are consistent with two special cases: at the free-fermion point $\beta^2=\frac12$ (where the chain is equivalent to non-interacting lattice fermions and the conjectures reproduce the exact values $\beta C=\pi$ and $C_0=(\pi\mathrm e)^{-1/12}A_G$, with $A_G$ the Glaisher–Kinkelin constant) and at $\Delta=-\frac12$, where the source paper derives a consistency relation with finite-$L$ combinatorial norm formulas (eqs. (11.12)–(11.13) there), which are themselves conjectural.

Nevertheless, all of these formulae remain conjectures. The constants $C_0(\beta)$ and $C(\beta)$ have no known analytic origin: they intertwine the lattice normalization conventions (the choice of two-body factors and twist) with the continuum Coulomb-gas normalization of the CFT (a normalization of CFT operators built from free-field integrals), and are determined only numerically (with conjectural limits $\beta C\to\mathrm e$ as $\beta\to0$ and $\beta C\to(\pi/2)^3$ as $\beta\to1$). In the regime $\tfrac23<\beta^2<1$, the dual non-local IMs govern the leading finite-$L$ energy corrections (eq. (7.3) of the source paper) used to identify individual Bethe states at finite $L$, and their excited-state eigenvalues are laborious to compute. The Hermitian structure that defines the norms — a star conjugation (an antilinear anti-involution fixing which lattice operators are mutually adjoint) on the finite chain that becomes the canonical Virasoro conjugation in the scaling limit, the lattice counterpart of the charge-at-infinity procedure (augmenting the Gaussian model by auxiliary degrees of freedom at infinity so that the inner product is well defined) — raises positivity and null-vector questions that matter for RSOS (restricted solid-on-solid; a lattice model whose heights are restricted to a finite set) restrictions of the six-vertex model and for alternating spin chains. A rigorous proof would constitute the first amplitude-level (as opposed to spectrum-level) validation of the ODE/IQFT correspondence.

Problem Statement

Prove or refute the Kotousov–Lukyanov conjectures on the large-$L$ asymptotics of the norms of the low-energy Bethe states of the spin-$\frac12$ XXZ Heisenberg chain in the critical regime $\Delta=\cos(\pi\beta^2)$, $0<\beta^2<1$, in a sector of fixed magnetization $S^z\ge 0$ and twist $k$ (with $-\tfrac12<k\le\tfrac12$). Specifically, for every low-energy Bethe state whose scaling limit is the Fock-space vector $|\bar{\boldsymbol w}\rangle\otimes|{\boldsymbol w}\rangle$ at descendant level $N$ (arbitrary $N\ge0$), establish rigorously — or disprove by a controlled counterexample — that:

1. the norm obeys $\mathfrak N=\mathfrak N_\infty L^{\eta}\mathrm e^{A_2L^2+A_1L}(1+o(1))$ as $L\to\infty$, with $A_2=\int_0^\infty\frac{\mathrm dt}{t}\,\frac{\sinh\!\big(\tfrac{\beta^2t}{1-\beta^2}\big)\,\sinh t}{2\sinh\!\big(\tfrac{t}{1-\beta^2}\big)\,\cosh^2 t}\,,\qquad A_1=\log\!\Big[\frac{2(1-\beta^2)}{\sin(\pi\beta^2)}\Big]$ (eqs. (3.6) and (3.7) of the source paper);

2. the exponent is $\eta=\frac16-\frac13\rho^2-2\rho\beta S^z-4I_1(\boldsymbol w)-4\bar I_1(\bar{\boldsymbol w})$ with $\rho=\beta^{-1}-\beta$ and $I_1$, $\bar I_1$ the eigenvalues of the first local integrals of motion of the continuum theory on the two chiral sectors, and the amplitude factorizes as $\mathfrak N_\infty=\mathcal N_{\bar p}(\bar{\boldsymbol w})\mathcal N_p(\boldsymbol w)$ (Conjecture I), the chiral sector labels being $2p=\beta S^z+\beta^{-1}(k+w)$ and $2\bar p=\beta S^z-\beta^{-1}(k+w)$ with $w$ the integer winding number of the sector;

3. the auxiliary amplitudes $K_p(\boldsymbol w)$ — of the auxiliary norm $\mathfrak K=\mathsf U^{-1}\mathfrak N$ (eq. (10.1) of the source paper) and related to the chiral factors $\mathcal N_p(\boldsymbol w)$ of $\mathfrak N_\infty$ by eq. (10.6) of the source paper — obey $K_p(\boldsymbol w)=C_0\,C^{h(p)+4N}V_p(\boldsymbol w)Z_+(p\,|\,\beta)$ with $h(p)=4p^2+\frac16(\rho^2-1)=4p^2+\frac16(\beta^2+\beta^{-2}-3)$, $V_p(\boldsymbol w)$ the diagonal of the Virasoro Hermitian form defined by eqs. (9.6), (9.1), (9.2) of the source paper, and $Z_+$ the convergent-product special function of the source paper (Conjectures II–III), so that $\mathfrak N_\infty$ takes the fully factorized form of eq. (11.11) there, in which the reflection-operator eigenvalues $R_p(\boldsymbol w)$ enter through $H_p=r_pV_p$ with $r_p=R_p(\boldsymbol w)/R_p^{(0)}$;

4. the constants $C_0(\beta)$ and $C(\beta)$ entering these amplitudes are independent of $p$, $\boldsymbol w$ and the descendant level $N$ and depend only on $\beta$, consistently with the exact free-fermion values $\beta C=\pi$ and $C_0=(\pi\mathrm e)^{-1/12}A_G$ ($A_G$ the Glaisher–Kinkelin constant) at $\beta^2=\tfrac12$ derived in the source paper.

A complete solution must cover arbitrary descendant level and every anisotropy with $0<\beta^2<1$. A refutation consists of a rigorously established (or numerically controlled with error bounds) violation of any of the conjectured formulas for some family of low-energy Bethe states.

References

1. G. A. Kotousov and S. L. Lukyanov, 'Bethe state norms for the Heisenberg spin chain in the scaling limit', Nucl. Phys. B 947 (2019) 114748, doi:10.1016/j.nuclphysb.2019.114748, arXiv:1906.07081, https://arxiv.org/abs/1906.07081
2. M. Gaudin, B. M. McCoy and T. T. Wu, 'Normalization sum for the Bethe's hypothesis wave functions of the Heisenberg-Ising chain', Phys. Rev. D 23 (1981) 417, doi:10.1103/PhysRevD.23.417, https://journals.aps.org/prd/abstract/10.1103/PhysRevD.23.417
3. V. E. Korepin, 'Calculation of norms of Bethe wave functions', Commun. Math. Phys. 86 (1982) 391, doi:10.1007/BF01212176, https://link.springer.com/article/10.1007/BF01212176
4. P. Dorey and R. Tateo, 'Anharmonic oscillators, the thermodynamic Bethe ansatz and nonlinear integral equations', J. Phys. A 32 (1999) L419, doi:10.1088/0305-4470/32/38/102, arXiv:hep-th/9812211, https://arxiv.org/abs/hep-th/9812211
5. V. V. Bazhanov, S. L. Lukyanov and A. B. Zamolodchikov, 'Spectral determinants for Schrödinger equation and Q-operators of conformal field theory', J. Stat. Phys. 102 (2001) 567, doi:10.1023/A:1004838616921, arXiv:hep-th/9812247, https://arxiv.org/abs/hep-th/9812247
6. V. V. Bazhanov, S. L. Lukyanov and A. B. Zamolodchikov, 'Higher-level eigenvalues of Q-operators and Schrödinger equation', Adv. Theor. Math. Phys. 7 (2003) 711-725, doi:10.4310/ATMP.2003.v7.n4.a4, arXiv:hep-th/0307108, https://arxiv.org/abs/hep-th/0307108
7. V. V. Bazhanov, G. A. Kotousov, S. M. Koval and S. L. Lukyanov, 'On the scaling behaviour of the alternating spin chain', JHEP 08 (2019) 087, doi:10.1007/JHEP08(2019)087, arXiv:1903.05033, https://arxiv.org/abs/1903.05033
8. G. A. Kotousov and S. L. Lukyanov, 'Spectrum of the reflection operators in different integrable structures', JHEP 02 (2020) 029, doi:10.1007/JHEP02(2020)029, arXiv:1910.05947, https://arxiv.org/abs/1910.05947
9. V. V. Bazhanov, G. A. Kotousov, S. M. Koval and S. L. Lukyanov, 'Scaling limit of the Z_2 invariant inhomogeneous six-vertex model', Nucl. Phys. B 965 (2021) 115337, doi:10.1016/j.nuclphysb.2021.115337, arXiv:2010.10613, https://arxiv.org/abs/2010.10613
10. G. A. Kotousov and S. L. Lukyanov, 'On the scaling behaviour of an integrable spin chain with Z_r symmetry', Nucl. Phys. B 993 (2023) 116269, doi:10.1016/j.nuclphysb.2023.116269, arXiv:2305.03620, https://arxiv.org/abs/2305.03620
11. S. Gehrmann, G. A. Kotousov and S. L. Lukyanov, 'Scaling limit of the ground state Bethe roots for the inhomogeneous XXZ spin-1/2 chain', Nucl. Phys. B 1006 (2024) 116624, doi:10.1016/j.nuclphysb.2024.116624, arXiv:2406.12102, https://arxiv.org/abs/2406.12102
12. G. A. Kotousov, S. L. Lukyanov and D. A. Shabetnik, 'Scaling limit of Q-functions for the Z_r invariant inhomogeneous XXZ spin-1/2 chain near free fermion point', Nucl. Phys. B 1024 (2026) 117315, doi:10.1016/j.nuclphysb.2026.117315, arXiv:2511.20530, https://arxiv.org/abs/2511.20530
13. R. Conti and D. Masoero, 'Counting monster potentials', JHEP 02 (2021) 059, doi:10.1007/JHEP02(2021)059, arXiv:2009.14638, https://arxiv.org/abs/2009.14638
14. R. Conti and D. Masoero, 'On solutions of the Bethe ansatz for the quantum KdV model', Commun. Math. Phys. 402 (2023) 335, doi:10.1007/s00220-023-04728-3, arXiv:2112.14625, https://arxiv.org/abs/2112.14625
15. D. Masoero and G. Ruzza, 'ODE/IM correspondence at the free-fermion point. Laguerre Wronskians, shifted symmetric functions, and quantum KdV', arXiv:2605.24563 (2026), https://arxiv.org/abs/2605.24563
16. H. Frahm, S. Gehrmann and G. A. Kotousov, 'Scaling limit of the staggered six-vertex model with U_q(sl(2)) invariant boundary conditions', SciPost Phys. 16 (2024) 149, doi:10.21468/SciPostPhys.16.6.149, arXiv:2312.11238, https://arxiv.org/abs/2312.11238