Year: 2025
Type: article
Abstract: Overdamped Langevin dynamics are reversible stochastic differential equations which are commonly used to sample probability measures in high-dimensional spaces, such as the ones appearing in computational statistical physics and Bayesian inference. By varying the diffusion coefficient, there are in fact infinitely many overdamped Langevin dynamics which are reversible with respect to the target probability measure at hand. This suggests to optimize the diffusion coefficient in order to increase the convergence rate of the dynamics, as measured by the spectral gap of the generator associated with the stochastic differential equation. We analytically study this problem here, obtaining in particular necessary conditions on the optimal diffusion coefficient. We also derive an explicit expression of the optimal diffusion in some appropriate homogenized limit. Numerical results, both relying on discretizations of the spectral gap problem and Monte Carlo simulations of the stochastic dynamics, demonstrate the increased quality of the sampling arising from an appropriate choice of the diffusion coefficient.
Source: Mathematics of Computation
DOI: 10.1090/mcom/4098
Institutions: Institut national de recherche en sciences et technologies du numérique, CERMICS, Institut Polytechnique de Paris, MATHERIALS: MATHematics for MatERIALS, Imperial College London, Université Paris-Sud, Centre National de la Recherche Scientifique, Centre de Mathématiques Appliquées de l'École polytechnique, Laboratoire de Mathématiques et Modélisation d'Évry
Language: English
Subfield: Statistics and Probability
Field: Mathematics
Domain: Physical Sciences
Open Access status: green
Awards: G1111137553, G1215857969, G1620472780, G1814043814, G1886517568, G2176653565, G2361257913, G2664173715, G2671653789, G2798619229, G3010527862, G3173448015, G3783898869, G434337789, Simulation nonequilibrium stochastic dynamics, G4927318077, G6978771000, G7683707404, G7774426346, G8124668106, G8374943362, G8471469333, G8531439085, Quantitative Analysis of Metastable Processes, G896097225
OpenAlex ID: W4394973588