Research Interests


I am interested in stochastic numerics, differential equations and their applications to machine learning.


Since my time as a graduate student, I have particularly enjoyed the numerical analysis of Brownian motion and Stochastic Differential Equations (SDEs). This research has focused on developing numerical methods and applying them to prominent SDEs in data science, such as Langevin dynamics and Neural SDEs. You can find some of my numerical methods in the Diffrax library (full credit goes to Andraž Jelinčič and Patrick Kidger for making this possible).


Alongside my interest in SDEs, I have worked on machine learning projects in collaboration with members of the DataSig team, where we introduced new differential equation models and algorithms, inspired by rough path theory, for multivariate time series problems.

Publications and preprints


Efficient, Accurate and Stable Gradients for Neural ODEs

Sam McCallum, and James Foster
Preprint [Slides]

Approximating the signature of Brownian motion for high order SDE simulation

James Foster
Preprint

On the convergence of adaptive approximations for stochastic differential equations

James Foster and Andraž Jelinčič
Preprint [Slides]

Single-seed generation of Brownian paths and integrals for adaptive and high order SDE solvers

Andraž Jelinčič, James Foster and Patrick Kidger
Preprint

Generative Modelling of Lévy Area for High Order SDE Simulation

Andraž Jelinčič, Jiajie Tao, William F. Turner, Thomas Cass, James Foster and Hao Ni
Preprint [Slides]

High order splitting methods for SDEs satisfying a commutativity condition

James Foster, Gonçalo dos Reis and Calum Strange
SIAM Journal on Numerical Analysis (2024)  [Slides]

Brownian bridge expansions for Lévy area approximations and particular values of the Riemann zeta function

James Foster and Karen Habermann
Combinatorics, Probability and Computing (2023)

An asymptotic radius of convergence for the Loewner equation and simulation of SLE traces via splitting

James Foster, Terry Lyons and Vlad Margarint
Journal of Statistical Physics (2022)

The shifted ODE method for underdamped Langevin MCMC

James Foster, Terry Lyons and Harald Oberhauser
Preprint [Slides]

Efficient and Accurate Gradients for Neural SDEs

Patrick Kidger, James Foster, Xuechen Li and Terry Lyons
Neural Information Processing Systems (2021)

The Signature Kernel is the solution of a Goursat PDE

Cristopher Salvi, Thomas Cass, James Foster, Terry Lyons and Weixin Yang
SIAM Journal on Mathematics of Data Science (2021)

Neural SDEs as Infinite-Dimensional GANs

Patrick Kidger, James Foster, Xuechen Li, Harald Oberhauser and Terry Lyons
International Conference on Machine Learning (2021)  [Slides]

Neural Rough Differential Equations for Long Time Series

James Morrill, Cristopher Salvi, Patrick Kidger, James Foster and Terry Lyons
International Conference on Machine Learning (2021)  [Slides]

Neural Controlled Differential Equations for Irregular Time Series

Patrick Kidger, James Morrill, James Foster and Terry Lyons
Neural Information Processing Systems, Spotlight (2020)  [Slides]

An Optimal Polynomial Approximation of Brownian Motion

James Foster, Terry Lyons and Harald Oberhauser
SIAM Journal on Numerical Analysis (2020)
[Slides] [Poster]

Teaching


Contact


Email: jmf68@bath.ac.uk

Dr James Foster
Department of Mathematical Sciences
University of Bath
Bath
BA2 7AY
United Kingdom

Office: 4 West 3.37