http://arxiv.org/abs/2101.06685
We construct five deep generative models of gravitational waveforms for the compact binary coalescence events. Our construction bases on the extensions of the conditional variational autoencoder (cVAE). By inputting just the masses of binary black holes, these trained generative models can produce the corresponding more than $95\%$ accurate inspiral-merger waveform in less than $10^{-3}$ seconds. Moreover, these models are also capable of extrapolation. That is, with mainly the low-mass-ratio training set, the resultant trained model is capable of generating large amount of accurate high-mass-ratio waveforms. Our result implies that the deep generative model is possible to speed up the generation of highly accurate gravitational waveforms of higher mass ratio by progressively self-training.
C. Liao and F. Lin
Tue, 19 Jan 21
88/92
Comments: 11 pages, 15 figures