Mobility estimation for langevin dynamics using control variates
File(s)PSV2023.pdf (1.49 MB)
Published version
Author(s)
Pavliotis, Grigorios A
Stoltz, Gabriel
Vaes, Urbain
Type
Journal Article
Abstract
The scaling of the mobility of two-dimensional Langevin dynamics in a periodicpotential as the friction vanishes is not well understood for nonseparable potentials. Theoreticalresults are lacking, and numerical calculation of the mobility in the underdamped regime is chal-lenging because the computational cost of standard Monte Carlo methods is inversely proportionalto the friction coefficient, while deterministic methods are ill-conditioned. In this work, we proposea new variance-reduction method based on control variates for efficiently estimating the mobility ofLangevin-type dynamics. We provide bounds on the bias and variance of the proposed estimatorand illustrate its efficacy through numerical experiments, first in simple one-dimensional settingsand then for two-dimensional Langevin dynamics. Our results corroborate prior numerical evidencethat the mobility scales as\gamma - \sigma , with 0<\sigma \leqslant 1, in the low friction regime for a simple nonseparablepotential.
Date Issued
2023-06-30
Date Acceptance
2023-01-20
Citation
SIAM: Multiscale Modeling and Simulation, 2023, 21 (2), pp.680-715
ISSN
1540-3459
Publisher
Society for Industrial and Applied Mathematics
Start Page
680
End Page
715
Journal / Book Title
SIAM: Multiscale Modeling and Simulation
Volume
21
Issue
2
Copyright Statement
© 2023 Society for Industrial and Applied Mathematics.
Identifier
http://dx.doi.org/10.1137/22m1504378
Publication Status
Published
Date Publish Online
2023-06-01