Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Mean-field approximation, Gibbs relaxation, and cross estimates

Armand Bernou, Mitia Duerinckx

TL;DR

The paper addresses the coupled questions of propagation of chaos and Gibbs relaxation for $N$ Brownian particles with weak mean-field interactions under Langevin dynamics. It develops a detailed BBGKY-hierarchy analysis together with hypocoercivity to prove uniform-in-$N$ exponential Gibbs relaxation and uniform-in-time chaos propagation, and introduces a cross estimate that clamps the joint deviation by a product of the two effects, yielding an $O(N^{-1} e^{-ct})$ cross term. In translation-invariant settings on the torus, this leads to time-accelerated propagation of chaos with the one-particle error decaying as $O(N^{-1} e^{-ct})$, while the framework extends to overdamped dynamics and, in certain regimes, to non-perturbative interaction strengths with vanishing cross-error tails. The results provide new quantitative insights into Gibbs relaxation beyond Lipschitz forces and furnish a unified mechanism connecting short-time mean-field behavior with long-time equilibration, with potential applications to kinetic models, opinion dynamics, and interacting particle systems. Overall, the work advances the understanding of how mean-field approximations persist and relax in large interacting systems, offering rigorous cross-compatibility bounds between microscopic chaos and macroscopic Gibbs equilibration.

Abstract

We study the propagation of chaos and relaxation to Gibbs equilibrium for a system of $N$ classical Brownian particles with weak mean-field interactions. It is well known that propagation of chaos holds uniformly in time with rate $O(N^{-1})$ and that Gibbs relaxation holds uniformly in $N$ with exponential rate $O(e^{-ct})$. We go one step further by establishing a cross estimate that simultaneously captures both effects: the joint deviation between chaos propagation and Gibbs relaxation is of order $O(N^{-1}e^{-ct})$. In particular, for translation-invariant systems, this yields an accelerated propagation of chaos, with the mean-field approximation error at the level of the one-particle density improving from $O(N^{-1})$ to $O(N^{-1}e^{-ct})$. Our approach relies on a detailed analysis of the BBGKY hierarchy for correlation functions, and applies to both underdamped and overdamped Langevin dynamics with merely bounded interaction forces. In addition, we obtain new quantitative results on Gibbs relaxation and provide partial extensions beyond the weak interaction regime.

Mean-field approximation, Gibbs relaxation, and cross estimates

TL;DR

The paper addresses the coupled questions of propagation of chaos and Gibbs relaxation for Brownian particles with weak mean-field interactions under Langevin dynamics. It develops a detailed BBGKY-hierarchy analysis together with hypocoercivity to prove uniform-in- exponential Gibbs relaxation and uniform-in-time chaos propagation, and introduces a cross estimate that clamps the joint deviation by a product of the two effects, yielding an cross term. In translation-invariant settings on the torus, this leads to time-accelerated propagation of chaos with the one-particle error decaying as , while the framework extends to overdamped dynamics and, in certain regimes, to non-perturbative interaction strengths with vanishing cross-error tails. The results provide new quantitative insights into Gibbs relaxation beyond Lipschitz forces and furnish a unified mechanism connecting short-time mean-field behavior with long-time equilibration, with potential applications to kinetic models, opinion dynamics, and interacting particle systems. Overall, the work advances the understanding of how mean-field approximations persist and relax in large interacting systems, offering rigorous cross-compatibility bounds between microscopic chaos and macroscopic Gibbs equilibration.

Abstract

We study the propagation of chaos and relaxation to Gibbs equilibrium for a system of classical Brownian particles with weak mean-field interactions. It is well known that propagation of chaos holds uniformly in time with rate and that Gibbs relaxation holds uniformly in with exponential rate . We go one step further by establishing a cross estimate that simultaneously captures both effects: the joint deviation between chaos propagation and Gibbs relaxation is of order . In particular, for translation-invariant systems, this yields an accelerated propagation of chaos, with the mean-field approximation error at the level of the one-particle density improving from to . Our approach relies on a detailed analysis of the BBGKY hierarchy for correlation functions, and applies to both underdamped and overdamped Langevin dynamics with merely bounded interaction forces. In addition, we obtain new quantitative results on Gibbs relaxation and provide partial extensions beyond the weak interaction regime.

Paper Structure

This paper contains 33 sections, 17 theorems, 140 equations, 1 figure.

Table of Contents

  1. Introduction
  2. General overview
  3. Propagation of chaos
  4. Relaxation to equilibrium
  5. Cross error estimate
  6. Overview of main results
  7. Main results
  8. Extension 1: overdamped Langevin dynamics
  9. Extension 2: beyond the weak interaction regime
  10. Examples
  11. Preliminary
  12. Mean-field relaxation
  13. BBGKY hierarchy for marginals
  14. Gibbs relaxation
  15. Correlation functions
  16. ...and 18 more sections

Key Result

Proposition 1.1

Let $K \in L^{\infty}(\mathbb R^{2d})^d$, let $A$ satisfy eq:confinement-A, and let $\mu_\circ \in \mathcal{P}(\mathbb X) \cap L^2(\omega)$. Consider the unique global weak solution $F^N\in C(\mathbb R_+; \mathcal{P}(\mathbb X^N) \cap L^2(\omega^{\otimes N}))$ of the Liouville equation eq:Liouville_

Figures (1)

  • Figure 1: This diagram displays the estimates for mean-field and relaxation errors in $L^2(\omega^{\otimes m})$ norms. Our main result in this work is that the estimates for mean-field and relaxation errors further hold relatively to one another in form of a cross error estimate, cf. \ref{['eq:cross-err']}.

Theorems & Definitions (29)

  • Proposition 1.1
  • Theorem 1.2: Cross mean-field/relaxation error
  • Corollary 1.3: Accelerated propagation of chaos for translation-invariant systems
  • Theorem 1.4: Cross mean-field/relaxation error
  • Corollary 1.5: Accelerated propagation of chaos for translation-invariant systems
  • Theorem 1.6
  • Remark 1.7
  • Lemma 2.1
  • proof
  • Remark 2.2
  • ...and 19 more