Machine Learning and Gravitational Wave Astronomy

Project description

I am looking to supervise PhD students interested in either gravitational physics / general relativity, or machine learning applied to scientific problems. Projects will be in the area of gravitational-wave (GW) astronomy, with focus on modeling and/or data analysis for gravitational systems such as black hole binaries. We aim to develop fast and accurate models and inference techniques, which we will then use to analyze real data from the LIGO-Virgo GW observatories and to make predictions for future detectors (LISA, ET). We can thereby precisely characterize sources, derive constraints on alternative gravity theories, and map out the GW universe.

To this end, my main research has been in developing probabilistic deep-learning techniques for Bayesian inference with GW data. This trains neural networks to analyze data in a fraction of the time of classical methods, with comparable levels of accuracy. Example projects could be either on the machine-learning side (developing faster and more accurate techniques, which could in principle be applied more broadly in science) or on applications (to future detectors, to new sources, and to populations of sources). On the theory side, I am interested primarily in modeling nonlinear effects using black hole perturbation theory, to study the behavior of black holes as they equilibrate following a merger, but I am happy to supervise interesting projects more broadly in general relativity.