RESOLVE: Rare Event Surrogate Likelihood for Gravitational Wave Paleontology Parameter Estimation
read the original abstract
The first detection of gravitational waves, recognized by the 2017 Nobel Prize in Physics, has opened up a new research field: gravitational-wave paleontology. When massive stars evolve into black holes and collide, they create gravitational waves that propagate through space and time. These gravitational-waves, now detectable on Earth, act as fossils tracing the histories of the massive stars that created them. Estimating physics parameters of these massive stars from detected gravitational-waves is a parameter estimation task, with the primary difficulty being the extreme rarity of collisions in simulated binary black holes. This rarity forces researchers to choose between prohibitively expensive simulations or accepting substantial statistical variance. In this work, we present RESOLVE, a rare event surrogate model that leverages polynomial chaos expansion (PCE) and Bayesian MCMC to emulate this rare formation efficiency. Our experimental results demonstrate that RESOLVE is the only surrogate model that achieves proper statistical coverage, while effectively learning the underlying distribution of each physics parameter. We construct a likelihood function incorporating both the emulated formation efficiency and LIGO's gravitational wave observations, which we then minimize to produce community-standard credible intervals for each physics parameter. These results enable astronomers to gain deeper insights into how the universe transformed from simple gases into the complex chemical environment that eventually made life possible.
This paper has not been read by Pith yet.
Forward citations
Cited by 1 Pith paper
-
Probabilistic Data-Driven Modelling of Astrophysical Transients: The Neural Process Family for Ultrafast and Class-Agnostic Light Curve Reconstruction with NightLANP
Attentive Neural Processes outperform Gaussian Processes and neural networks on light curve interpolation quality, feature recovery, calibration, and speed for 15 transient classes under realistic Rubin cadences.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.