I am often looking for new PhD students, Postdocs, Masters and Bachelors students to join the team
BSc and MMath students normally get involved via final year projects.
Professor of Applied Mathematics
PhD Student, MIGSAA CDT programme, started 2017. Research interests- Developing new optimization schemes for the training of deep neural networks, numerical methods for stochastic differential equations, deep learning theory, molecular dynamics, sampling methods.
PhD Student, MAC-MIGS programme, started 2019. Studying distributed algorithms for deep learning and constraint-based neural network training. Co-supervised by Michele Ottobre (HWU) and Amos Storkey (Informatics).
PhD Student, MAC-MIGS programme, started 2020. Research topic - machine learning for molecular dynamics. Currently studying symmetry preservation in machine-learned force fields. Co-supervised by Ben Goddard (Mathematics), Antonia Mey (Chemistry), James McDonagh (IBM Research)
PhD Student, MAC-MIGS programme, started 2020. Research topic- scalable Bayesian inference methods for high dimensional stochastic models. Currently studying sampling methods on manifolds. Co-supervised by Daniel Paulin (Statistics).
PhD Student, MAC-MIGS programme, started 2020. Research topic- simplification of causal graphs and analysis of neural computational models. Co-supervised by Michal Branicki (Mathematics).
postdoc visiting the group from 2019-2021 studied constrained algorithms for neural network training
Worked on generalized Langevin equation, nonequilibrium models, and numerical methods. Moved on to postdocs first at SAMSI and Duke University, then to UBC where he is now working with Christoph Ortner.
numerical methods for dissipative particle dynamics, adaptive (noisy gradient) sampling algorithms, machine learning training algorithms
algorithms and software design for ensemble study of neural networks and complex models; the force behind TATi the Thermodynamic Analytics Toolkit.
studied algorithms based on diffusion maps for accelerating the sampling of molecular models; she also set up a collaboration with DNV GL to study classification methods for wind form SCADA data and worked with Anton Martinsson to create the acwind python package for this purpose; currently working in Paris as a machine learning developer.
high performance computing genius and chief architect of the MIST software suite for interfacing novel integration methods into existing molecular dynamics codes like LAMPPS and Gromacs. Studied a wide variety of algorithms using MIST. Defended his PhD in late 2020. Currently head of software development at a Perth startup.
collaborated on a variety of projects, include Langevin integrators and their application/analysis in molecular dynamics applications, including constrained MD methods. Moved to a postdoc at U. Chicago after his PhD and worked with J. Weare (further collaboration on an ensemble sampling method). Returned for projects in 2018-19 on wind data analysis, among other things. Now working in machine learning for an Edinburgh startup.