Speaker
Robin Chan
(UGent | Royal Observatory of Belgium)
Description
Gravitational wave parameter estimation demands millions of waveform evaluations per event, a bottleneck set to worsen as next-generation detectors like the Einstein Telescope and Cosmic Explorer increase detection rates and signal duration. We present a major expansion of ripple, a JAX-based waveform package offering GPU acceleration and exact autodifferentiation, which now covers higher-order-mode, precessing and tidal waveform families. We validate these implementations against LALSuite via mismatch analysis and against numerical relativity simulations from the SXS and CoRe catalogues, finding strong agreement overall. We also benchmark the speedup gained from GPU-based waveform generation with respect to CPU-based waveforms.
Author
Robin Chan
(UGent | Royal Observatory of Belgium)
Co-authors
Archisman Ghosh
(Ghent University)
Bruno Bertrand
(Université Catholique de Louvain (UCL) - CP3)
Chun-Fung Wong
(KU Leuven)
Harsh Narola
(Utrecht University)
Justin Janquart
(UCLouvain | Royal Observatory of Belgium)
Thibeau Wouters
(Utrecht University/Nikhef)
Thomas Ng
(Nikhef / Utrecht University)