Convergence of adaptive biasing potential methods for diffusions

DOI10.1016/J.CRMA.2016.05.011MaRDI QIDQ2630183zbMATH OpenOpenAlexFDO

Authors Michel Benaïm, Charles-Edouard Bréhier

Publication date 25 July 2016

Published in Comptes Rendus. Mathématique. Académie des Sciences, Paris (Search for Journal in Brave)

Copyright license Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International

Full work available at URL https://arxiv.org/abs/1603.08088

zbMATH Keywords

adaptive importance samplingalmost sure convergencediffusionsstochastic approximation

Mathematics Subject Classification ID

Strong limit theorems (60F15) Functional limit theorems; invariance principles (60F17) Stochastic ordinary differential equations (aspects of stochastic analysis) (60H10) Diffusion processes (60J60) Interacting random processes; statistical mechanics type models; percolation theory (60K35)

Abstract: We prove the consistency of an adaptive importance sampling strategy based on biasing the potential energy function $V$ of a diffusion process $d{X}_t^0=-ablaV(X_t^0)dt+d{W}_t$; for the sake of simplicity, periodic boundary conditions are assumed, so that $X_t^0$ lives on the flat $d$-dimensional torus. The goal is to sample its invariant distribution $$. The bias $V_t-V$, where $V_t$ is the new (random and time-dependent) potential function, acts only on some coordinates of the system, and is designed to flatten the corresponding empirical occupation measure of the diffusion $X$ in the large time regime. The diffusion process writes $d{X}_t=-abla{V}_t(X_t)dt+d{W}_t$, where the bias $V_t-V$ is function of the key quantity $overline{mu}_t$: a probability occupation measure which depends on the past of the process, {it i.e.} on $(X_s{)}_{sin[0,t]}$. We are thus dealing with a self-interacting diffusion. In this note, we prove that when $t$ goes to infinity, $overline{mu}_t$ almost surely converges to $mu$. Moreover, the approach is justified by the convergence of the bias to a limit which has an intepretation in terms of a free energy. The main argument is a change of variables, which formally validates the consistency of the approach. The convergence is then rigorously proven adapting the ODE method from stochastic approximation.

Recommendations

• Convergence analysis of adaptive biasing potential methods for diffusion processes
• Long-time convergence of an adaptive biasing force method
• Long-time convergence of an adaptive biasing force method: variance reduction by Helmholtz projection
• The adaptive biasing force algorithm with non-conservative forces and related topics
• Adaptive force biasing algorithms: new convergence results and tensor approximations of the bias

Cites work

• Free energy computations. A mathematical perspective
• Molecular dynamics. With deterministic and stochastic numerical methods
• Self-interacting diffusions.

Cited in

(7)

• Convergence analysis of adaptive biasing potential methods for diffusion processes
• Adaptive force biasing algorithms: new convergence results and tensor approximations of the bias
• Analysis of an adaptive biasing force method based on self-interacting dynamics
• Convergence of metadynamics: discussion of the adiabatic hypothesis
• Existence, uniqueness and convergence of a particle approximation for the adaptive biasing force process
• Birth–death dynamics for sampling: global convergence, approximations and their asymptotics
• Convergence rates for an adaptive biasing potential scheme from a Wasserstein optimization perspective

This page was built for publication: Convergence of adaptive biasing potential methods for diffusions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2630183)