Neutron stars are important astrophysical objects for understanding and constraining the equation of state of nuclear matter. However, the challenge lies in mapping neutron star observations, along with their inherent uncertainties, into meaningful insights about their composition and equation of state. In this talk, a new approach based on Bayesian neural networks will be presented to address this challenge. Unlike traditional deep learning methods, these networks can not only map observations to the equation of state but also quantify the confidence in their predictions. This feature enables a more credible understanding of the results obtained and, consequently, the composition of neutron stars.
Organized by: Catarina Cosme