Curved schemes for stochastic differential equations on, or near, manifolds

John Armstrong, Tim King*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

3 Citations (Scopus)

Abstract

Given a stochastic differential equation (SDE) in Rn whose solution is constrained to lie in some manifold M Rn, we identify a class of numerical schemes for the SDE whose iterates remain close to M to high order. These schemes approximate a geometrically invariant scheme, which gives perfect solutions for any SDE that is diffeomorphic to n-dimensional Brownian motion. Unlike projection-based methods, they may be implemented without explicit knowledge of M. They can even be implemented if the solution merely remains close to M, without being exactly confined to it. Our approach does not require simulating any iterated Itô integrals beyond those needed to implement the Euler-Maryuama scheme. We prove that the schemes converge under a standard set of assumptions, and illustrate their geometric advantages in a variety of numerical contexts, including Monte Carlo simulation of the Riemannian Langevin equation.

Original languageEnglish
Article number20210785
JournalProceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences
Volume478
Issue number2262
DOIs
Publication statusPublished - 29 Jun 2022

Keywords

  • conserved quantities
  • flow-based numerical schemes
  • geometric invariance

Fingerprint

Dive into the research topics of 'Curved schemes for stochastic differential equations on, or near, manifolds'. Together they form a unique fingerprint.

Cite this