A Multiwavelength Machine-learning Approach to Classifying X-ray Sources in the Fields of Unidentified 4FGL-DR4 sources

Q: What problem does the study address?

A large fraction of 4FGL-DR4 gamma-ray sources remains unidentified. The study targets improving identification through X-ray counterparts using machine learning.

Q: How is the machine-learning pipeline improved and used?

The pipeline is updated with astrometric corrections, probabilistic crossmatching to lower-frequency counterparts, and a more realistic oversampling method. It classifies 1206 X-ray sources within 73 unidentified 4FGL regions.

Q: What types of counterparts does the study identify for specific gamma-ray sources?

It reports plausible X-ray counterparts for many GeV targets, including isolated neutron star candidates, active galactic nucleus candidates, sources linked to star-forming regions, and ambiguous cases. For the remaining unidentified 4FGL sources, no plausible X-ray counterpart is found or they are too close to the Galactic Center.

Aug 5, 2026

Upload by Noah

A large fraction of Fermi-LAT sources in the fourth Fermi-LAT 14yr catalog (4FGL-DR4) remain unidentified. This work improves a multiwavelength machine-learning pipeline and applies it to classify 1206 Chandra X-ray sources (S/N > 3) within 73 unidentified 4FGL regions using Chandra Source Catalog 2.0. Pipeline gains include astrometric corrections, probabilistic crossmatching to lower-frequency counterparts, and a more realistic oversampling method. X-ray sources are categorized into eight astrophysical classes, and additional spectral/variability checks refine results, yielding likely counterparts for gamma-ray emitters.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages34

Reading Time1 h 26 min

AccessFree to download

Abstract

Introduction

Fermi-LAT catalogs and identification challenges

Machine learning for gamma-ray source classification

Role of Chandra and the Chandra Source Catalog

arXiv :2403 .05068v2 [ astro-ph .HE] 17 Aug 2024 Draft version August 20, 2024 Typeset using LATEX twocolumn style in AASTeX631 A Multiwavelength Machine-learning Approach to Classifying X-ray Sources in the Fields of Unidentified 4FGL-DR4 sources Hui Yang,1 Jeremy Hare,2, 3, 4 and Oleg Kargaltsev1 1 Department of Physics, The George Washington University, 725 21st St, NW, Washington, DC 20052, USA 2 Astrophysics Science Division, NASA Goddard Space Flight Center, 8800 Greenbelt Rd, Greenbelt, MD 20771, USA 3 Center for Research and Exploration in Space Science and Technology, NASA/GSFC, Greenbelt, MD 20771, USA 4 The Catholic University of America, 620 Michigan Ave. , N. E. Washington, DC 20064, USA ABSTRACT A large fraction of Fermi-Large Area Telescope (LAT) sources in the fourth Fermi-LAT 14yr catalog (4FGL) still remain unidentified (unIDed) . We continued to improve our machine-learning pipeline and used it to classify 1206 X-ray sources with signal-to-noise ratios > 3 located within the extent of 73 unIDed 4FGL sources with Chandra X-ray Observatory observations included in the Chandra Source Catalog 2.0 . Recent improvements to our pipeline include astrometric corrections, probabilistic crossmatching to lower-frequency counterparts, and a more realistic oversampling method. X-ray sources are classified into eight broad predetermined astrophysical classes defined in the updated training data set, which we also release. We present details of the machine-learning classification, describe the pipeline improvements, and perform an additional spectral and variability analysis for brighter sources. The classifications give 103 plausible X-ray counterparts to 42 GeV sources. We identify 2 GeV sources as isolated neutron star candidates, 16 as active galactic nucleus candidates, seven as sources associated with star-forming regions, and eight as ambiguous cases. For the remaining 40 unIDed 4FGL sources, we could not identify any plausible counterpart in X-rays, or they are too close to the Galactic Center. Finally, we outline the observational strategies and further improvements in the pipeline that can lead to more accurate classifications. Keywords: X-ray sources (1822), Classification (1907), Active galactic nuclei (16), Compact objects (288), Catalogs (205), Neutron stars (1108), Astronomical object identification (87), Astrostatistics tools (1887), X-ray surveys (1824), Gamma-ray sources (633), X-ray binary stars (1811), Random Forests (1935) 1. INTRODUCTION Over the last 15yr, the Large Area Telescope (LAT) on board the Fermi γ-ray Space Telescope has revealed a large number of γ-ray sources (Atwood et al. 2009; Abdollahi et al. 2022) . However, only a small fraction, comprising 14% for Galactic sources with |b| < 5◦ or 6% for all sources, are confidently identified in the fourth FermiLAT 14yr point-source catalog (4FGL-DR4, hereafter 4FGL; Ballet et al. 2023), among a total of 7,195 sources. The 4FGL catalog also includes the associated source category (≈60% of sources) where the γ-ray sources have probable counterparts at other wavelengths based [huiyang@gwmail.gwu.edu](huiyang@gwmail.gwu.edu) purely on positional coincidence. The matched counterpart may or may not belong to a known source class. Among the identified Galactic sources, pulsars standout as one of the most prevalent types. The number of γ-ray pulsars has surged from six during the EGRET era to 297 in the third Fermi-LAT pulsar catalog (Smith et al. 2023) . While many pulsars are identified through their γ-ray and radio pulsations, imaging X-ray observations can also unveil young pulsar (PSR) candidates. This identification can be achieved either through their distinct spectral properties and the absence of multiwavelength (MW) counterparts or by resolving extended emission from pulsar wind nebulae (PWNe, see, e.g. , Kargaltsev et al. 2012; Hare et al. 2021a) . Recent research has leveraged machine learning (ML) for cla

FAQs

What problem does the study address?

A large fraction of 4FGL-DR4 gamma-ray sources remains unidentified. The study targets improving identification through X-ray counterparts using machine learning.

How is the machine-learning pipeline improved and used?

The pipeline is updated with astrometric corrections, probabilistic crossmatching to lower-frequency counterparts, and a more realistic oversampling method. It classifies 1206 X-ray sources within 73 unidentified 4FGL regions.

What types of counterparts does the study identify for specific gamma-ray sources?

It reports plausible X-ray counterparts for many GeV targets, including isolated neutron star candidates, active galactic nucleus candidates, sources linked to star-forming regions, and ambiguous cases. For the remaining unidentified 4FGL sources, no plausible X-ray counterpart is found or they are too close to the Galactic Center.

Aug 5, 2026

Upload by Noah