Artificial Intelligence for Gravitational Waveform Modeling

As catalogs of gravitational-wave observations grow and we prepare for next-generation gravitational-wave observatories, it is becoming increasingly important to have accurate and dense banks of template waveforms. Depending on the sources considered, such template banks can be produced using many different approaches including numerical relativity, perturbation theory, and analytic and semi-analytic models to name a few. For the purpose of improving speed and efficiency, many studies have recently used artificial intelligence in various ways to aid in the generation of these template banks. These methods include using neural networks to identify optimal numerical relativity simulation parameters, using machine learning to upscale simulation resolution, and using supervised learning to model neutron star remnants, among many more. In the coming years, such approaches will likely continue to gain speed, changing the way we approach waveform modeling.