arXiv:2610.02859
From Heat to Homology: Spectral Gap Transfer for Exact Quantum Gibbs Sampling at All Temperatures
Caesnan M. G. Leditto, Kuo-Chin Chen, Min-Hsiu Hsieh
Abstract
Preparing Gibbs states through dissipative dynamics requires controlling convergence for the chosen Hamiltonian $H$ and coupling operators. However, efficient implementation requires convergence guarantees from the spectral gap of the generator of the dynamics, which remains a crucial challenge. Recent results address the challenge by restricting the dynamics with structural assumptions about the Hamiltonian and its interactions. For the Chen-Kastoryano-Gilyen (CKG) construction and arbitrary finite-dimensional Hamiltonians, we prove an explicit comparison that converts a spectral gap estimate for the infinite-temperature generator into a lower bound on the spectral gap of the CKG quantum Gibbs sampler at every finite positive inverse temperature $β$. It allows a single spectral gap estimate at infinite temperature to certify convergence across Hamiltonians and temperatures. Such an estimate is often known from a classical spectral gap estimate, such as the classical random walk spectral gap. For Hodge Laplacians of simplicial complexes as Hamiltonians, we construct coupling operators whose generator at infinite temperature has the same spectral gap as a classical simplicial down-up walk. A positive spectral gap of the walk and bounded spectral width $W=λ_{\max}(H)-λ_{\min}(H)$ give a generator spectral gap independent of the number of simplices at every fixed temperature. We then establish conditions under which the sampler approximately prepares zero-energy (harmonic) states in polynomial evolution time. This provides a conditional convergence guarantee for the state preparation in the thermal approach to quantum topological data analysis.
AI Summary
The paper asks whether a spectral-gap estimate at infinite temperature can certify convergence of the exact Chen–Kastoryano–Gilyén (CKG) Gibbs sampler at finite temperature, without locality or Bohr-frequency assumptions. For couplings closed under adjoints up to phases and normalized by $\sum_a A_a^\dagger A_a=I$, let $\kappa_0>0$ bound the gap of $\mathcal L_{\mathcal A,0}(X)=\sum_a A_a^\dagger X A_a-X$ on the full operator space. For arbitrary finite-dimensional $H$, the Metropolis-weight CKG generator is primitive and satisfies $\operatorname{gap}(-\mathcal L_{\beta,\mathrm M})\ge\tfrac12e^{-\beta W}\kappa_0$, where $W=\lambda_{\max}(H)-\lambda_{\min}(H)$. The proof combines a stationary commutator identity, adjoint pairing, Plancherel’s identity, Gibbs-weight comparison, and KMS symmetry. A three-level barrier example establishes optimality of the exponential dependence. A separate perturbative bound also covers Gaussian transition weights. For simplicial Hodge Hamiltonians, measure-and-prepare couplings make the infinite-temperature quantum gap equal to the uniform down–up walk gap. Combining the comparison with Gibbs concentration gives polynomial evolution time to approximate the uniform harmonic state when the walk gap and harmonic-space fraction are inverse-polynomial and the smallest positive Hodge eigenvalue is a constant fraction of the spectral width; the required $\beta W$ is $O(\log n)$. The bound can be exponentially weak for extensive spectral width. A positive walk gap alone does not guarantee rapid low-temperature convergence; projected-ground-space obstructions and an explicit Hodge example demonstrate this. Efficient circuit implementation, topological-invariant estimation, and advantages over other preparation methods are not established.
OpenAI GPT-6 Astra was also used to generate the first manuscript draft, which the authors reviewed and polished into the final manuscript.
Error found: The abstract claims a size-independent generator gap from a positive classical walk gap and bounded spectral width, but the stated guarantee requires the classical walk gap to be bounded uniformly away from zero across the family.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.