Maximizing distinguishability in Neutron Star Equation of State Model Selection Methods

We present a novel method, motivated by reparameterization of the sample space, well-suited for Bayesian inference of the equation of state (EOS) of neutron stars. We calculate the Bayesian evidence of a collection of EOS hypotheses using posterior distribution obtained from a single EOS agnostic parameter estimation run for each binary neutron star (BNS) event in a given population of observations. We perform parameter estimation runs using relative binning. We compute the evidence using published methods and compare their relative efficiency in selecting the correct EOS. Additionally, we assess their relative computational performance for both current and planned gravitational wave detectors. Our findings reveal that this new EOS model selection method, conducted within the subspace of the posterior probability distribution, yields an improved statistic for recognizing the correct EOS for a population of BNS events.