Deep Learning with Quantized Neural Networks for Gravitational-wave Forecasting of Eccentric Compact Binary Coalescence

We present the first application of deep learning forecasting for eccentric compact binary coalescence. We consider binary neutron stars, neutron star-black hole systems, and binary black hole mergers that span an eccentricity range e<=0.9. We train neural networks that describe these astrophysical populations, and then quantify how many seconds in advance they can predict the merger of eccentric binaries whose gravitational waves are injected in advanced LIGO noise available at the \texttt{Gravitational Wave Open Science Center}. Our findings indicate that the rich structure and complex morphology of eccentric signals enables deep learning to predict the coalescence of these sources from several seconds up to two minutes prior to merger. A quantized version of our neural networks achieved 4x reduction in model size, and up to 2.5x inference speed up. These novel algorithms enable gravitational wave forecasting of compact binaries that may exist in dense stellar environments, and whose early detection is relevant for time-sensitive Multi-Messenger Astrophysics campaigns.