Inference of point sources from NuSTAR X-ray observations using probabilistic cataloging

Abstract

An important task in astronomy is in locating astronomical objects, known as sources, in images and describing the sources' properties such as size, spatial information, intensity, and color. This task is known as cataloging, and suffers from several shortcomings such as an inability to capture sources fainter than the background, or correctly modeling sources that overlap in what's known as a crowded field. Probabilistic cataloging is a technique based on Bayesian inference that allows sampling from the posterior distribution of catalogs, producing a set of valid hypotheses rather than one, allowing for robust calculations of uncertainty. In this work, we propose methods to produce a probabilistic catalog of the X-ray sky observed by the NuSTAR X-ray telescope.