Edge fluctuations of the melting domain wall in the XXZ spin-1/2 chain at nonzero anisotropy: the Saenz–Tracy–Widom conjecture

Saenz, Tracy, and Widom (2022) derived exact finite-$N$ Bethe-ansatz formulas for the distribution of the left-most up-spin in the Heisenberg–Ising (XXZ) spin-1/2 chain with domain-wall initial conditions. At zero anisotropy the melting-front edge provably converges, under $t^{1/3}$ rescaling, to the GUE Tracy–Widom distribution $F_2$; at nonzero anisotropy they reduced the problem to a conjectural Airy-kernel series (their Conjecture 8.1) whose limit they could not control — the multiplicative finite-$N$ factor $F(\sigma,S)$ is uncontrolled and the saddle-point error terms are only assumed negligible — while the physics literature disagreed on whether the front fluctuations scale as $t^{1/2}$ or $t^{1/3}$. The open problem is to convert this conjecture into a theorem, or refute it: control $F(\sigma,S)$ and the error terms in the joint limit $t\ll N\to\infty$ and identify the limiting law of the rescaled edge $(X_1(t)+2t)/t^{1/3}$ at $\Delta\ne0$. Fujimoto and Sasamoto (Phys. Rev. Lett. 136, 087102 (2026)) proved the analogous Tracy–Widom law for the folded XXZ model with an alternating domain wall and their numerics sharpen rather than settle the question for the XXZ chain itself; the conjecture remains open.

Background

The Heisenberg–Ising (XXZ) spin-1/2 chain is the one-dimensional quantum spin Hamiltonian $H_\Delta=\sum_x 2(X_{x+1}X_x+Y_{x+1}Y_x+\Delta Z_{x+1}Z_x)$ on the lattice $\mathbb{Z}$, where $X_x,Y_x,Z_x$ are spin-1/2 operators at site $x$ and the real parameter $\Delta$ is the anisotropy. At $\Delta=0$ the chain is the free-fermion (XX) model: the Jordan–Wigner transformation, a standard change of variables mapping spins to fermions, makes it noninteracting. A domain-wall (step) initial state prepares all spins up on one side of the origin and all spins down on the other; under unitary time evolution the wall melts. Viewing each up-spin as a particle and each down-spin as a hole, the extreme edge of the melting region is tracked by the position $X_1(t)$ of the left-most up-spin, whose distribution is the gap probability $F_N(x,t)=P(X_1(t)\ge x)$ (also called the one-point function).

At $\Delta=0$ the problem is exactly solvable: the $N\to\infty$ limit of $F_N$ is governed by the discrete Bessel kernel (the kernel that controls the length of the longest increasing subsequence of a random permutation), and the edge rescaling $(X_1(t)+2t)/t^{1/3}$ converges to the Tracy–Widom distribution $F_2$ of the Gaussian unitary ensemble (GUE) — the universal law for the largest eigenvalue of large GUE random matrices, expressible as a Fredholm determinant $\det(I-K)$ (the determinant of the identity operator plus a trace-class integral operator) of the Airy kernel $K_{\mathrm{Ai}}$ (Tracy and Widom, 1994). The same edge law was found to govern the full counting statistics of the propagating free-fermion front (Eisler and Rácz, 2013).

Saenz, Tracy, and Widom (STW) attacked the interacting case $\Delta\ne0$. Using the coordinate Bethe ansatz (an exact scattering representation of many-body wavefunctions as sums over permutations), an antisymmetrization identity of Cantini, Colomo, and Pronko for the six-vertex model, and the Izergin–Korepin determinant (a closed determinant formula for the six-vertex model with domain-wall boundary conditions), they reduced $F_N(x,t)$ to a finite-$N$ double series of contour integrals (their Theorem 4), and then, by residue extraction and deformation of contours onto steepest-descent paths, to a conjectural Airy-kernel series (their Conjecture 8.1): in the joint limit $t\ll N\to\infty$ (time and particle number both diverging with $t/N\to0$, so the left edge does not yet feel the right end of the initial block of $N$ up-spins), with $x=-2t-s\,t^{1/3}$ and initial positions $y_j+1=v_j t^{1/3}$,

$F_N(x,t)\;\to\; \lim \sum_{\sigma\in S_N}(-1)^{\sigma}\sum_{S\subset[N]}(-1)^{|S|}t^{-|S|/3}\,F(\sigma,S)\prod_{k\in S}K_{\mathrm{Ai}}\bigl(s+v_{\sigma(k)},\,s+v_k\bigr),$

where the multiplicative factor $F(\sigma,S)$ is an explicit $N$-fold contour integral (their Eq. (92)) and $S_N$ is the symmetric group. If $F(\sigma,S)$ factorized as $\prod_{j\in S^c}Q(\sigma(j),j)$ for some kernel $Q$, the series would collapse to a Fredholm determinant; at $\Delta=0$ this factorization holds with $F=\mathbf{1}(\sigma|_{S^c}=\mathrm{Id}_{S^c})$ and the limit provably reduces to $F_2$, but at $\Delta\ne0$ nothing is known. STW state verbatim that they are not able to control the limit, that the main obstacle is the term $F(\sigma,S)$, that the contour-deformation and saddle-point error terms (including remainders of order $e^{-Ct^{1-3\alpha}}$ with $1/4<\alpha<1/3$) are only assumed negligible, and — citing numerical, generalized-hydrodynamic (an exact framework for ballistic transport in integrable systems), and semiclassical analyses that disagree — that even the fluctuation scaling, $t^{1/2}$ versus $t^{1/3}$, is undecided, with the limiting distribution still unclear.

Problem Statement

Resolve Conjecture 8.1 of Saenz–Tracy–Widom for the Heisenberg–Ising (XXZ) spin-1/2 chain at nonzero anisotropy. Precisely: for the XXZ chain with $N$ up-spins initially at sites $y_1<\dots<y_N$ in a sea of down-spins, let $X_1(t)$ be the left-most up-spin and $F_N(x,t)=P(X_1(t)\ge x)$. In the joint limit $t\ll N\to\infty$ with $x=-2t-s\,t^{1/3}$ and $y_j+1=v_j\,t^{1/3}$, determine whether $F_N(x,t)$ converges, and identify the limit. This requires:

(i) rigorously justifying that the contour-deformation and saddle-point error terms (including the $O(e^{-Ct^{1-3\alpha}})$ remainders with $1/4<\alpha<1/3$) do not contribute uniformly in the joint limit, so that the exact finite-$N$ series is faithfully replaced by the Airy-kernel series of the conjecture;

(ii) controlling the multiplicative finite-$N$ factor $F(\sigma,S)$ — either via the factorization $F(\sigma,S)=\prod_{j\in S^c}Q(\sigma(j),j)$ (or an asymptotic form of it), which would make the series a Fredholm determinant, or by another method establishing the convergence of the double series over permutations $\sigma\in S_N$ and subsets $S\subset[N]$ with this factor present;

(iii) identifying the resulting limiting distribution of the rescaled edge $(X_1(t)+2t)/t^{1/3}$ for $\Delta\ne0$, and thereby settling whether the melting-front fluctuations are $t^{1/3}$ Airy-type as conjectured (as rigorously established at $\Delta=0$, where the limit is $F_2$) or belong to a different class, for example $t^{1/2}$ Gaussian scaling as suggested by some hydrodynamic analyses.

The common scientific objective is a complete, rigorous derivation of the long-time edge statistics of the melting domain wall at nonzero anisotropy; items (i)–(iii) are one linked asymptotic program, not independent deliverables — (i) and (ii) are the technical gateway without which no limiting law can be asserted, and (iii) is the payoff that settles the physical controversy. Refutation is an admissible outcome on equal footing: a rigorous demonstration that the conjectured limit fails, with the correct scaling and limit law identified where determinable, resolves the problem. Results restricted to the folded XXZ model or to alternating (diluted) domain-wall initial conditions do not by themselves answer the question for the dense domain wall of the XXZ chain, though they may serve as components of a solution.

References

1. Axel Saenz, Craig A. Tracy, Harold Widom, 'Domain Walls in the Heisenberg-Ising Spin-1/2 Chain', arXiv:2202.07695 (2022); published in 'Toeplitz Operators and Random Matrices: In Memory of Harold Widom', Operator Theory: Advances and Applications, Springer/Birkhäuser (2022), pp. 9-47, DOI: 10.1007/978-3-031-13851-5_2. https://arxiv.org/abs/2202.07695
2. Kazuya Fujimoto, Tomohiro Sasamoto, 'Quantum Transport in Interacting Spin Chains: Exact Derivation of the Tracy-Widom Distribution', Physical Review Letters 136, 087102 (2026), DOI: 10.1103/lbl9-n9ws; arXiv:2412.20147 (arXiv v1 preprint title: 'Quantum Transport in Interacting Spin Chains: Exact Derivation of the GUE Tracy-Widom Distribution'). https://arxiv.org/abs/2412.20147
3. Eric I. Corwin, Nikolaus Elsaesser, Axel Saenz, 'Coordinate-energy transformation and the one-point function for the Heisenberg-Ising XXZ spin-1/2 chain on the ring', arXiv:2506.14171 (2025). https://arxiv.org/abs/2506.14171
4. Vir B. Bulchandani, Christoph Karrasch, 'Subdiffusive front scaling in interacting integrable models', Physical Review B 99, 121410 (2019), DOI: 10.1103/physrevb.99.121410; arXiv:1810.08227. https://arxiv.org/abs/1810.08227
5. Mario Collura, Andrea De Luca, Jacopo Viti, 'Analytic solution of the domain-wall nonequilibrium stationary state', Physical Review B 97, 081111 (2018), DOI: 10.1103/physrevb.97.081111; arXiv:1707.06218 (arXiv v1 preprint title: 'Analytic solution of the Domain Wall non-equilibrium stationary state'). https://arxiv.org/abs/1707.06218
6. Jean-Marie Stéphan, 'Return probability after a quench from a domain wall initial state in the spin-1/2 XXZ chain', Journal of Statistical Mechanics: Theory and Experiment (2017) 103108, DOI: 10.1088/1742-5468/aa8c19; arXiv:1707.06625. https://arxiv.org/abs/1707.06625
7. Jean-Marie Stéphan, 'Free fermions at the edge of interacting systems', SciPost Physics 6, 057 (2019), DOI: 10.21468/scipostphys.6.5.057; arXiv:1901.02770. https://arxiv.org/abs/1901.02770
8. Viktor Eisler, Zoltán Rácz, 'Full Counting Statistics in a Propagating Quantum Front and Random Matrix Spectra', Physical Review Letters 110, 060602 (2013), DOI: 10.1103/physrevlett.110.060602; arXiv:1211.2321. https://arxiv.org/abs/1211.2321
9. Adam J. McRoberts, Thomas Bilitewski, Masudul Haque, Roderich Moessner, 'Domain Wall Dynamics in Classical Spin Chains: Free Propagation, Subdiffusive Spreading, and Soliton Emission', Physical Review Letters 132, 057202 (2024), DOI: 10.1103/physrevlett.132.057202; arXiv:2306.15351. https://arxiv.org/abs/2306.15351
10. Grégoire Misguich, Nicolas Pavloff, Vincent Pasquier, 'Domain wall problem in the quantum XXZ chain and semiclassical behavior close to the isotropic point', SciPost Physics 7, 025 (2019), DOI: 10.21468/scipostphys.7.2.025; arXiv:1905.08756. https://arxiv.org/abs/1905.08756
11. Luigi Cantini, Filippo Colomo, Andrei G. Pronko, 'Integral Formulas and Antisymmetrization Relations for the Six-Vertex Model', Annales Henri Poincaré 21, 865 (2020), DOI: 10.1007/s00023-019-00856-6; arXiv:1906.07636. https://arxiv.org/abs/1906.07636
12. Filippo Colomo, Andrei G. Pronko, 'Evaluation of integrals for the emptiness formation probability in the square-ice model', arXiv:2405.04358 (2024). https://arxiv.org/abs/2405.04358
13. Jean-Marie Stéphan, 'Exact time evolution formulae in the XXZ spin chain with domain wall initial state', arXiv:2112.12092 (2021). https://arxiv.org/abs/2112.12092
14. Craig A. Tracy, Harold Widom, 'Level-spacing distributions and the Airy kernel', Communications in Mathematical Physics 159, 151 (1994), DOI: 10.1007/bf02100489; arXiv:hep-th/9211141. https://arxiv.org/abs/hep-th/9211141