Bayesian model selection with gravitational waves from supernovae

This thesis concerns inferring core collapse supernova physics using gravitational waves. The mechanism through which the supernova is re-energised is not well understood and there are many theories of the physical processes behind the so called supernova mechanism. Gravitational waves provide an opportunity to see through to the core of a collapsing star. This thesis provides an algorithm that will analyse a detected gravitational waveform from a core collapse supernova and identify the supernova mechanism. This is achieved through the use of Bayesian model selection and a nested sampling algorithm. This Bayesian data analysis algorithm is called the Supernova Model Evidence Extractor (SMEE). SMEE is designed to classify detected gravitational waveforms from core-collapse supernovae and 3 different versions which employ different types of data have been developed. These 3 versions utilise time domain (which has been fast Fourier transformed into the frequency domain), power spectrum domain and spectrogram data and the success of each version is investigated. Firstly, results for a simplified idealised version of SMEE are discussed. In this scenario only a single gravitational wave detector is considered and the effect of the sky position of the source are ignored. Next, techniques which can be employed to improve SMEE are investigated. Finally, SMEE is tested using 3 gravitational wave detectors and the full effect of the time delay between detectors and the antenna response on each detector is included. As well as this, recoloured detector noise from the Science runs from both LIGO and Virgo are utilised here. This thesis demonstrates that each version of SMEE is successful and are able to infer the supernova mechanism for a galactic supernova. The spectrogram version of SMEE is deemed the most accurate and it is recommended that this technique should be further explored in the future