Dissertation: Multiwavelength, Machine Learning, and Parallax Studies of X-ray Binaries in Three Local Group Galaxies

(Cassiopeia – Spring / printemps 2019)

by Dr. Robin Arnason
Thesis defended on December 7, 2018
Department of Physics and Astronomy, Western University
Thesis advisor: Prof. Pauline Barmby

Abstract

X-ray binary stars are rare systems consisting of a black hole or neutron star and a main-sequence companion star. They are useful probes of galaxy properties and interesting laboratories for extreme physical conditions. In this thesis, I investigated the X-ray binary population of three galaxies in the Local Group.

The Sculptor Dwarf Spheroidal Galaxy offers the chance to study a primordial low-mass X-ray binary (LMXB) population in an isolated, low-metallicity environment. Combining X-ray, optical, and infrared observations, I have studied nine previously-identified and discovered four additional LMXB candidates in this galaxy. Of these candidates, all but one are either background galaxies or foreground stars, meaning that Sculptor is presently effectively devoid of bright LMXBs. If Sculptor is able to retain primordial LMXBs at a similar rate to globular clusters, it is likely that bright XRBs in globular clusters observed in the present day were dynamically formed.

The Andromeda Galaxy has the largest catalogue of Chandra-studied X-ray sources of any nearby galaxy. I have used this population to test a proof-of-concept method for identifying X-ray binary candidates using machine learning algorithms trained on known sources. After testing a variety of commonly used algorithms, I find that the best-performing random forest algorithm can identify X-ray binary candidates with 85% accuracy. I have identified 16 new strong X-ray binary candidates and find that 4 sources classified as X-ray binaries by this method coincide with star clusters identified by the Panchromatic Hubble Andromeda Treasury project.

The Milky Way’s X-ray binary population is the easiest to study but the most challenging for which to accurately measure distance. I have crossmatched Galactic X-ray binary catalogs to the second data release of the Gaia mission, finding candidate counterparts for 86 Galactic X-ray binaries. Distances to Gaia candidate counterparts are systematically smaller than those measured using Type I X-ray bursts, suggesting that these bursts do not consistently reach the Eddington limit. High-mass X-ray binaries are correlated with the Galaxy’s spiral arms and low-mass X-ray binaries are anti-correlated with the Galaxy’s spiral arms at a low level of significance.

Bookmark the permalink.

Comments are closed.