Calibrating global behaviour of equation of state by combining nuclear and astrophysics inputs in a machine learning approach

TL;DR

This paper introduces a machine learning approach using symbolic regression to efficiently infer neutron star equation of state parameters from observational and experimental data, significantly speeding up Bayesian inference processes.

Contribution

The study develops symbolic regression models that accurately approximate TOV solutions, enabling rapid Bayesian inference of EoS parameters from diverse datasets.

Findings

SRMs closely match TOV-based Bayesian posteriors

Speed-up of approximately 100 times in inference process

Effective integration of nuclear and astrophysical data

Abstract

We implemented symbolic regression techniques to identify suitable analytical functions that map various properties of neutron stars (NSs), obtained by solving the Tolman-Oppenheimer-Volkoff (TOV) equations, to a few key parameters of the equation of state (EoS). These symbolic regression models (SRMs) are then employed to perform Bayesian inference with a comprehensive dataset from nuclear physics experiments and astrophysical observations. The posterior distributions of EoS parameters obtained from Bayesian inference using SRMs closely match those obtained directly from the solutions of TOV equations. Our SRM-based approach is approximately 100 times faster, enabling efficient Bayesian analyses across different combinations of data to explore their sensitivity to various EoS parameters within a reasonably short time.