arXiv:2610.05848
Spectral Analysis Beyond the Spectral Gap: Rapid Mixing of Quantum Gibbs Samplers in One Dimension
Hongrui Chen, Zhiyan Ding
Abstract
We prove rapid mixing of a quantum Gibbs sampler for general one-dimensional non-commuting local spin systems at every fixed positive temperature. In particular, for a system of $n$ sites, we show that the Kubo--Martin--Schwinger (KMS) detailed-balanced Gibbs sampler of~\cite{CKG2023} converges from any initial state to trace-norm error $ε$ in time $O(\log^4(n)+\log(1/ε))$. This improves the previous $O(n+\log(1/ε))$ bound and establishes rapid mixing beyond high-temperature and weak-interaction regimes. Our proof explains why the slowest decay rate, captured by the spectral gap, can substantially underestimate the speed of equilibration. The key insight is that a state far from equilibrium must have a substantial component outside the span of low-weight Pauli operators, where stronger dissipation estimates apply. This yields two stages of relaxation: a double-exponential convergence from a bad initial state, followed by exponential convergence near equilibrium. We turn this mechanism into a framework for proving rapid mixing through spectral analysis beyond the gap. Our decay estimate is equivalent to exponential convergence in sandwiched Rényi-2 divergence. Together with the spectral gap, it also implies a modified logarithmic Sobolev inequality with rate $Ω(\log^{-4}(n))$.
AI Summary
The paper asks whether the CKG KMS-detailed-balanced Gibbs sampler mixes rapidly for general one-dimensional noncommuting spin systems beyond high-temperature and weak-interaction regimes. For chains with bounded finite-range interactions, fixed local dimension, and every fixed inverse temperature $\beta>0$, it proves worst-case trace-norm mixing time $O(\log^4 n+\log(1/\epsilon))$, improving the previous $O(n+\log(1/\epsilon))$ guarantee. The sampler uses single-site Pauli couplings. The proof works in KMS $L^2$ geometry rather than directly establishing entropy or Wasserstein contraction. It combines existing Dirichlet-form comparison and uniform spectral-gap estimates with a decomposition into low-Pauli-weight operators and their KMS-orthogonal complement. Buffered interval projections establish stronger dissipation for the complementary component, while positivity and state normalization bound the low-weight contribution. For $F=\chi^2(\eta\Vert\rho_\beta)$, this yields dissipation proportional to $(1+F)\log(1+F)/\log^3 n$. Consequently, $\log(1+F)$, equivalently sandwiched Rényi-2 divergence, contracts exponentially at rate $\Omega(\log^{-3}n)$, bringing exponentially large initial $F$ to order one in $O(\log^4 n)$ time. The constant spectral gap controls subsequent relaxation. An appendix derives an MLSI rate $\Omega(\log^{-4}n)$. Constants may depend on temperature and fixed interaction parameters; the theorem does not give uniform low-temperature scaling. Extensions to other generators require an analogous Dirichlet-form comparison, and extensions beyond one dimension remain prospective. The paper also notes subsequent work proving a stronger, size-independent complete MLSI for the same sampler, implying $O(\log n+\log(1/\epsilon))$ mixing by a different method.
the authors utilized GPT-6 to brainstorm the initial proof ideas and to assist with drafting preliminary versions of certain technical calculations appearing in this work.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 38 pages, 1 figure