Machine Learning Acceleration of Neutron Star Pulse Profile Modeling

TL;DR

This paper introduces a neural network model that significantly speeds up the computation of neutron star pulse profiles, enabling faster data analysis for astrophysical research.

Contribution

A residual neural network was developed to predict neutron star flux profiles rapidly, overcoming computational bottlenecks of traditional ray tracing methods.

Findings

Achieves orders-of-magnitude speedup over traditional ray tracing.

Maintains high accuracy in flux predictions.

Handles complex emission geometries efficiently.

Abstract

Ray tracing algorithms that compute pulse profiles from rotating neutron stars are essential tools for constraining neutron-star properties with data from missions such as NICER. However, the high computational cost of these simulations presents a significant bottleneck for inference algorithms that require millions of evaluations, such as Markov Chain Monte Carlo methods. In this work, we develop a residual neural network model that accelerates this calculation by predicting the observed flux from the surface of a spinning neutron star as a function of its physical parameters and rotational phase. Leveraging GPU-parallelized evaluation, we demonstrate that our model achieves many orders-of-magnitude speedup compared to traditional ray tracing while maintaining high accuracy. We also show that the trained network can efficiently accommodate complex emission geometries, including non-circular and multiple hot spots, by integrating over localized flux predictions.