Building a prior distribution for the high density neutron star equation of state

The equation of state of dense matter is not well known at densities relevant to the cores of neutron stars. Bayesian inference can constrain the equation of state using multiple experimental and astrophysical observations, but requires assuming some prior distribution of equations of state. In this work, we use the non-parametric approach of Gaussian Processes to build a model-agnostic set of zero-temperature and beta-equilibrated equations of state.

We use training data from a few dozen nuclear equations of state available in the literature to condition our Gaussian Processes. We then modify the initial distribution using several Gaussian Process hyperparameters to explore the pressure and energy density plane.

To understand the impact of the initial training data and the hyperparameter choices, we explore how they change the resulting distribution in pressure and energy density. We also apply the hydrostatics equations of general relativity to determine how the training data and hyperparameters affect the prior distribution of mass-radius relationships. Finally, we discuss the impact of these prior choices on Bayesian inference of the equation of state using radio, X-ray and gravitational wave measurements.