Generating Gravitational Waveforms with Deep Neural Networks

Modeling gravitational waveforms given the parameters of the source is a difficult task since numerical relativity simulations are computationally expensive. We are investigating methods using generative deep neural network to interpolate accurately in order to densely populate the parameter space of signals, given a sparse distribution of waveforms from numerical relativity simulations as training data.

Project Members: Daniel George, Hongyu Shen, Eliu Huerta