Accelerating gravitational wave astrophysics with machine learning

Lead Research Organisation: University of Glasgow
Department Name: College of Science and Engineering


The recent discovery of gravitational waves from the coalescence of black holes and neutron stars has opened a new window on the universe, greatly advancing our understanding of compact binaries. As the LIGO and Virgo detectors continue to improve, they will see further into the universe, greatly increasing the rate of detections. As the noise level drops, the loudest signals will also stand out against the noisy data, reducing the statistical uncertainties in our analysis of their properties. At the same time, the new field of multi-messenger astronomy is demanding that we analyse the data quicker, to drive down the latency with which we pass on information for follow-up.

Meeting these challenges and delivering the best possible science from gravitational wave detections requires us to deliver results faster, from more complex models that reduce systematic uncertainties. An answer to these challenges may be found by making use of the recent major advances in machine learning through the use of neural networks. In particular, the aim of this project is to accelerate gravitational-wave inference using the methods of variational inference and neural-network enhances sampling algorithms. This in turn will also allow us to use more sophisticated models to avoid systematic errors, and ultimately deliver the most detailed and robust possible results from the ground-based network of gravitational wave observatories. This addresses several of STFC's strategic priorities, particularly in ensuring our continued leadership in gravitational wave astronomy, making use of the STFC-funded Advanced LIGO and A+ facilities. The methodology we will be using is also highly relevant for developing skills in artificial intelligence and data-intensive science that are aligned with STFC's goal of maintaining UK leadership in this area.


10 25 50

Studentship Projects

Project Reference Relationship Related To Start End Student Name
ST/T506102/1 01/10/2019 30/09/2023
2285031 Studentship ST/T506102/1 01/10/2019 31/03/2023 Michael Williams