Gravity seminar : Stephen Green, University of Nottingham — Accelerating Gravitational Wave Science with Simulation-Based Inference

Observations of gravitational waves by LIGO-Virgo-KAGRA occur every few days, with detection rates set to rise sharply in the future. This surge in detections offers immense scientific potential but also creates computational challenges for analyzing each event using advanced theoretical models. In this talk, I will describe how deep learning techniques for amortized Bayesian inference address these challenges. The approach uses simulated data to train neural networks such as normalizing flows to represent the posterior. Trained models then enable extremely fast sampling, with analyses in just seconds. I will highlight recent advancements in population inference and binary neutron star analysis, demonstrating the potential of these methods for next-generation detectors.