Reconstructing Core-Collapse Supernova Gravitational-Wave Signals with Transdimensional Bayesian Inference

Abstract

Core-collapse supernovae (CCSNe) are promising future sources of gravitational waves for current and next-generation observatories. Reconstructing CCSN gravitational-wave signals is challenging as they contain stochastic elements, have multiple complex features, and cover a wide frequency band. The stochasticity of the signal in particular motivates the need for morphology-independent reconstruction techniques which, once observed, will enable us to infer properties of the newly born proto-neutron star, the rotation, and the unknown CCSN explosion mechanism. In this work, we investigate the reconstruction of gravitational-wave signals from CCSNe using the transdimensional Bayesian inference framework tBilby. We demonstrate the method using simulated signals in synthetic Advanced LIGO detector noise at a range of signal-to-noise ratios. We reconstruct the signals using two types of wavelets: sine Gaussians and chirplets. We calculate overlaps between injected and reconstructed signals of up to 85%. We find that even when reconstruction overlap values are low, enough of the time-frequency structure of the dominant mode is captured to still make statements about the size of the evolving proto-neutron star. These capabilities establish tBilby as a powerful tool for gravitational-wave astronomy with burst sources.