Sam studied Physics (MPhys) at Bath, graduating in 2023. His masters project looked at combining perovskite solar cell modelling with Bayesian Parameter Estimation to inversely derive physical device parameters from experimental cell measurements. Sam has also worked on projects modelling charge-transport in semi-conductor materials, incorporating a machine learning approach to quantify prediction uncertainty. Sam’s interests are applied modelling and its potential intersection with other areas of mathematics. Outside of maths Sam enjoys running, reading and Weetabix Crispy Minis.

Project title:
Numerical Analysis of Neural Differential Equations

Supervisor(s):
James Foster and Neill Campbell

Project description: 
Neural Differential Equations are a recently proposed model class in scientific modelling and machine learning. Neural Differential Equations are defined by setting the vector field terms of a differential equation to be neural networks. This construction presents many interesting benefits over discrete-time models.

There are many open topics in Neural Differential Equations that this project aims to investigate. These include computationally-efficient model training and alternative constructions of Neural PDEs and SDEs.