Gravity seminar - Lucy Thomas Seminar

Abstract: In current gravitational wave data analysis, we heavily rely on waveform models to make possible the detection and parameter estimation of compact binary merger signals from LIGO data. These models must strike a delicate balance between accuracy and computational efficiency, as on average 10^7 model evaluations are needed for analysis of a single signal. In this talk, I will discuss recent developments in waveform modelling utilising machine learning to accelerate existing models, such as numerical relativity and effective-one-body models, by orders of magnitude without sacrificing accuracy. I will outline how they are constructed, their performance and limitations. In future when detections of mergers happen every few minutes from ground-based detectors, these techniques could form a crucial part of data analysis for these signals.