Speaker
Description
Gravitational wave astronomy in the era of third-generation (3G) detectors will pose significant computational challenges. While standard parameter estimation methods may remain technically feasible, the demand for more efficient inference algorithms is on the rise. We present a sequential neural simulation-based inference algorithm that merges neural ratio estimation (NRE) with nested sampling (NS) for advanced analysis in the field of gravitational waves. Building upon the principles of PolySwyft, which introduced the NSNRE algorithm, this framework leverages the power of JAX and seamlessly integrates the NRE algorithm with a state-of-the-art blackjax implementation of nested sampling. This integrated approach enables efficient and accurate parameter estimation, and will enable the continuation of breakthrough science in the 3G era and beyond.
| AI keywords | simulation-based inference |
|---|
