Decoding Long-duration Gravitational Waves from Binary Neutron Stars with Machine Learning: Parameter Estimation and Equations of State

Gravitational waves (GWs) from binary neutron stars (BNSs) offer valuable understanding of the nature of compact objects and hadronic matter.

However, their analysis requires substantial computational resources due to the challenges in Bayesian stochastic sampling.

The third-generation (3G) GW detectors are expected to detect BNS signals with significantly increased signal duration, detection rates, and signal strength, leading to a major computational burden in the 3G era.

We demonstrate a machine learning-based workflow capable of producing source parameter estimation and constraints on equations of state (EOSs) for hours-long BNS signals in seconds with minimal hardware costs.

We employ efficient compressions on the GW data and EOS using neural networks, based on which we build normalizing flows for inferences.

Given that full Bayesian analysis is prohibitively time-intensive, we validate our model against (semi-)analytical predictions.

Additionally, we estimate the computational demands of BNS signal analysis in the 3G era, showing that the machine learning methods will be crucial for future catalog-level analysis.

;Comment: 8 pages, 6 figures