Generative adversarial network for stellar core-collapse gravitational waves

Abstract

We present a rapid stellar core-collapse waveform emulator built using a deep convolutional generative adversarial network (DCGAN). The DCGAN was trained on the Richers \textit{et al.~}\cite{richers:2017} waveform catalogue to learn the structure of rotating stellar core-collapse gravitational-wave signals and generate realistic waveforms. We show that the DCGAN learns the distribution of the training data reasonably well, and that the waveform emulator produces signals that appear to have the key features of core-collapse, bounce, early post-bounce, and ringdown oscillations of the early proto-neutron star. The pre-trained DCGAN can therefore be used as a phenomenological model for rotating stellar core-collapse gravitational-waves.