Current Research
Gaussian Processes (GPs) are infinite-dimensional stochastic processes defined by a mean function and a kernel. Any measurements of a system are considered jointly Gaussian under this model admitting closed form posterior predictions. I'm exploring ways to optimise the injection efficiency of the Diamond Light Source synchrotron using GP models which act as simpler surrogates of unknown black-box functions.
In addition to my main focus on optimisation, I'm also fascinated by chaotic non-linear motion. GP latent variable models (GPLVMs) may offer a powerful way to predict chaotic exponents of particle beams, which would allow for efficient DA calculations etc.
Research interests
Gaussian Processes
Differential Geometry
Topology
Non-linear Beam Dynamics
Bayesian Optimisation