A Structural Theory of Quantum Metastability: Markov Properties and Area Laws

TL;DR

This work develops a universal structural theory for quantum metastability by modeling thermalization via a quasi-local, KMS-detailed-balanced Lindbladian. It proves that sufficiently metastable quantum states obey a Gibbs-like area law for mutual information and possess a local Markov property, extending equilibrium correlation structure to out-of-equilibrium metastable regimes. The authors introduce a cohesive framework linking approximate stationarity, local free-energy minima, non-commutative Fisher information, and approximate detailed balance, using the KMS-detailed-balanced Lindbladians to establish static characterizations and recoverability. These results have algorithmic implications for quantum thermal simulation, suggesting that metastable states can be efficiently probed and stabilized via time-averaging and local recovery maps, even in the presence of noise and measurements. Overall, the paper lays groundwork for a rigorous theory of thermal metastability with concrete targets for quantum simulation and a quantitative handle on the correlation structure in non-equilibrium quantum many-body systems.

Abstract

Statistical mechanics assumes that a quantum many-body system at low temperature can be effectively described by its Gibbs state. However, many complex quantum systems exist only as metastable states of dissipative open system dynamics, which appear stable and robust yet deviate substantially from true thermal equilibrium. In this work, we model metastable states as approximate stationary states of a quasi-local, (KMS)-detailed-balanced master equation representing Markovian system-bath interaction, and unveil a universal structural theory: all metastable states satisfy an area law of mutual information and a Markov property. The more metastable the states are, the larger the regions to which these structural results apply. Therefore, the hallmark correlation structure and noise resilience of Gibbs states are not exclusive to true equilibrium but emerge dynamically. Behind our structural results lies a systematic framework encompassing sharp equivalences between local minima of free energy, a non-commutative Fisher information, and approximate detailed balance conditions. Our results build towards a comprehensive theory of thermal metastability and, in turn, formulate a well-defined, feasible, and repeatable target for quantum thermal simulation.

Figures (4)

• Figure 1: (a) Given a metastable state $\bm{ \sigma}$ and a Hamiltonian $\bm{H}$, we prove that sufficiently small regions $\mathsf{A}$ admit an area law: the mutual information between $\mathsf{A}$ and its complement $\bar{\mathsf{A}}$ scales with the strength of the Hamiltonian on the boundary $\mathsf{I}(\mathsf{A}:\mathsf{\bar{A}})_{\bm{ \sigma}} \le 2\beta\cdot \Vert {\partial \bm{H}} \Vert$. (b) Underlying this behavior is an interpretation of metastable states as local minima of the free energy, in approximate thermal equilibrium within a "well" in the energy landscape. Local perturbations to the state take it out-of-equilibrium, triggering a local mixing process. So long as the lifetime of the metastable state (the time to escape the surrounding energy well) exceeds the local mixing timescale, the system re-equilibrates within the well.
• Figure 2: (a) Running Glauber dynamics on a restricted region $\mathsf{A}$, resamples the configuration on $\mathsf{A}$, from the Gibbs measure conditioned on the current boundary configuration. (b) In a quantum system, one can likewise define a resampling map for a region of qubits $\mathsf{A}$ by activating only a subset of the Lindbladian terms $\sum_{a\in \mathsf{A}} \mathcal{L}_a.$ However, the resulting map is only quasi-local, which may perturb far-away qubits.
• Figure 3: Copies of Ising "hard disks". Consider a $\mathsf{L}\times 2\mathsf{L}$ square lattice of bits, partitioned into $m\times m$ blocks, where ferromagnetic Ising interactions are placed only within each block. The left $\mathsf{L}\times \mathsf{L}$ square is initialized randomly to encode either 0 (up) or 1 (down) with suitable metastable states (Gibbs measure conditioned on majority 0 or 1). The right $\mathsf{L}\times \mathsf{L}$ square is simply a duplicate of the left, which maximally correlates the encoded $(\mathsf{L}/m)^2$ bits of information.
• Figure 4: The equivalent notions of metastability studied in this paper and their consequences. Our starting assumption captures stationarity in statistical distance, for arbitrary observables; the entropy production rate captures the stationarity of the free energy as a particular observable, for which we give an explicit Fisher information. This, in turn, reveals a more workable form of local static equilibrium, approximate detail balance.

Theorems & Definitions (100)

• Definition 2.1: Metastability as approximate stationarity
• Theorem 2.1: Metastability Implies an Area Law
• Example 2.1
• Theorem 2.2: Metastable States are Locally Markov
• Lemma 2.1: Time-averaging forces metastability
• Definition 2.2: Strongly Markov states
• Theorem 4.1: Local Recovery implies an Area Law
• Theorem 4.2: KMS Detailed Balance of $\mathcal{L}$ chen2023efficient
• Definition 4.1: The Entropy Production Rate
• Theorem 4.3: Entropy Production and the Fisher Information, \ref{['thm:integral_square_logs']} (Informal)