Conditional variational autoencoder inference of neutron star equation of state from astrophysical observations

TL;DR

This paper introduces a conditional variational autoencoder framework that accurately reconstructs neutron star equations of state from astrophysical observations, enabling rapid inference of dense matter properties.

Contribution

The study develops a flexible, deep learning-based inference model that reconstructs neutron star equations of state from observational data, improving speed and robustness over traditional methods.

Findings

Robust reconstruction of neutron star equations of state from observations

Fast, instantaneous inference capability

Flexible model adaptable to various dense matter quantities

Abstract

We present a new inference framework for neutron star astrophysics based on conditional variational autoencoders. Once trained, the generator block of the model reconstructs the neutron star equation of state from a given set of mass-radius observations. While the pressure of dense matter is the focus of the present study, the proposed model is flexible enough to accommodate the reconstructing of any other quantity related to dense matter equation of state. Our results show robust reconstructing performance of the model, allowing to make instantaneous inference from any given observation set.