arXiv:2609.36016
Is rapid mixing stable?
Jordi A. Montañà-López, Barbara Roos, Sebastian Stengele, Ángela Capel, Rahul Trivedi
Abstract
Several physically interesting Lindbladians, such as those describing generic open quantum system models as well as engineered state preparation processes, are often rapidly mixing, that is, they relax to steady states in time scaling polylogarithmically with the system size. Owing to the short mixing times, it is physically expected that local perturbations do not substantially increase the mixing time. In this work we question this expectation and ask whether rapid mixing survives. We provide counterexamples showing that rapid mixing is not stable in general, even when the perturbation does not change the steady state. We also show that arbitrarily weak local Hamiltonian perturbations can destroy rapid mixing. Complementing these results, we develop an array of sufficient conditions which imply stable rapid mixing, including several classes of quasifree fermionic systems, commuting Lindbladians, and local spin systems. We also identify conditions under which modified logarithmic Sobolev inequalities (MLSIs), a stronger form of convergence, lead to stable rapid mixing. As a further implication of our results, we show that rapid mixing does not imply a positive MLSI constant.
AI Summary
The paper asks when local perturbations preserve rapid mixing of Lindbladians, defined as worst-case trace-norm mixing in $\operatorname{polylog}N$ time, distinguishing this from exponential convergence with a size-independent decay rate. Counterexamples show that two rapidly mixing generators can have a slowly mixing sum even with the same unique maximally mixed stationary state. A translation-invariant, finite-range fermionic construction has uniform constant-rate rapid mixing for both summands and their restrictions, but $\tau_{\mathrm{mix}}=\Omega(N^2)$ for their sum. Purely dissipative examples give $\Omega(N)$ mixing after adding another dissipator or an on-site Hamiltonian; these perturbed examples have degenerate stationary spaces. Another construction uses arbitrarily weak potentials to localize modes away from boundary dissipation, changing polynomial to exponential mixing in physical chain length $m$. Its advertised polylogarithmic-to-polynomial slowdown uses an auxiliary parameter $N$ with $m=\Theta(\log N)$, so it does not establish instability of rapid mixing measured against physical size. Positive results use generalized oscillator norms and “bulk-dissipative covers”: sufficiently fast local relaxation dominating noncommuting interactions yields size-independent perturbation thresholds. Applications include high-temperature Gibbs samplers and fixed-point injective MPDO dynamics. Quasifree results bound full-state convergence through the single-particle propagator, including non-Gaussian inputs. MLSI is preserved under stationary-kernel inclusion or compatible stationary projections, while constant-rate rapid mixing alone need not imply any positive MLSI constant. Instability under arbitrarily weak perturbations remains unresolved for uniformly rapidly mixing families and for constant-decay-rate rapid mixing. Quasifree stability guarantees restrict perturbations to quasifree generators; extensions beyond high temperature and MPDO renormalization fixed points remain open.
Central claim may not hold: The arbitrarily weak Hamiltonian construction has m = Θ(log N) physical sites and an unperturbed mixing time at least Ω(m), so its reference family is not rapidly mixing under the paper's physical-system-size definition. Consequently, the presented results do not establish the abstract's claim that arbitrarily weak local Hamiltonian perturbations can destroy rapid mixing, although the order-one counterexamples still establish general instability.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.