DeepAI AI Chat
Log In Sign Up

Random walk approximation for irreversible drift-diffusion process on manifold: ergodicity, unconditional stability and convergence

06/02/2021
by   Yuan Gao, et al.
0

Irreversible drift-diffusion processes are very common in biochemical reactions. They have a non-equilibrium stationary state (invariant measure) which does not satisfy detailed balance. For the corresponding Fokker-Planck equation on a closed manifold, via Voronoi tessellation, we propose two upwind finite volume schemes with or without the information of the invariant measure. Both two schemes enjoy stochastic Q-matrix structures and can be decomposed as a gradient flow part and a Hamiltonian flow part, which enable us to prove unconditional stability, ergodicity and error estimates. Based on two upwind schemes, several numerical examples - including sampling accelerated by a mixture flow, image transformations and simulations for stochastic model of chaotic system - are conducted. These two structure-preserving schemes also give a natural random walk approximation for a generic irreversible drift-diffusion process on a manifold. Thus they can be adapted to manifold-related computations induced from high dimensional molecular dynamics.

READ FULL TEXT

page 18

page 19

page 20

page 21

page 23

page 24

05/18/2020

Inbetweening auto-animation via Fokker-Planck dynamics and thresholding

We propose an equilibrium-driven deformation algorithm (EDDA) to simulat...
03/10/2023

Lower error bounds and optimality of approximation for jump-diffusion SDEs with discontinuous drift

In this note we prove sharp lower error bounds for numerical methods for...
03/31/2023

Microcanonical Langevin Monte Carlo

We propose a method for sampling from an arbitrary distribution exp[-S()...
10/06/2019

Convergence of the likelihood ratio method for linear response of non-equilibrium stationary states

We consider numerical schemes for computing the linear response of stead...
02/06/2023

Stochastic Gradient Descent-induced drift of representation in a two-layer neural network

Representational drift refers to over-time changes in neural activation ...