Classifying X-ray Sources with Machine Learning and Searching for Compact Objects
Open Access DepositedHigh-energy astrophysics explores the most energetic astrophysical phenomena under extreme physical conditions such as neutron stars and black holes, often referred as compact objects (COs). Extensive efforts have been dedicated to explore and understand the diverse population of COs within our Galaxy, shedding light on their evolution and the fundamental physical processes associated with these extreme objects. X-ray and γ-ray observations play a pivotal role in identifying these objects, emitting radiation predominantly at high energies. However, the vast volume of serendipitous data, comprising over a million X-ray sources and thousands of γ-ray sources, presents a challenge in identifying COs using conventional methods. In this thesis, I present the development of an automated and efficient multiwavelength machine learning pipeline, MUWCLASS, designed to classify X-ray sources across various environments, with a primary goal of identifying COs. Alongside, I also look for particle accelerators in γ-ray sources and classify many X-ray sources for other types. Significant efforts have been dedicated to constructing training datasets comprising reliably classified sources of diverse types from multiple major X-ray catalogs, along with developments aimed at improving classification accuracy. The MUWCLASS pipeline has been successfully applied on the Chandra Source Catalog version 2.0, classifying over 66,000 sources, representing ∼ 21% of the entire catalog. To address the enigmatic nature of a large fraction of unidentified Galactic γ-ray sources, the MUWCLASS pipeline has been used to classify X-ray sources associated with 86 unidentified Fermi-LAT sources. I will also discuss a possibility of utilizing larger scales of high-energy and multiwavelength catalogs, holding promising prospects for identifying more high-energy sources, thereby offering valuable insights into their populations and underlying physics.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.