Dissertation
Data-driven exploration in x-ray astronomy: investigating intermediate polar candidates found among unusual objects selected by an unsupervised outlier identification algorithm
University of Iowa
Doctor of Philosophy (PhD), University of Iowa
Summer 2023
DOI: 10.25820/etd.006877
Abstract
Astronomers are increasingly faced with a deluge of information, and finding worthwhile targets of study in the sea of data can be difficult. Outlier identification studies are a method that can be used to focus investigations by presenting a smaller set of sources that could prove interesting because they do not follow the trends of the underlying population. We apply a principal component analysis (PCA) and an unsu- pervised random forest algorithm (uRF) to sources from the Chandra Source Catalog v.2 (CSC2). We present 119 high-significance sources that appear in all repeated ap- plications of our outlier identification algorithm (OIA). We analyze the characteristics of our outlier sources and crossmatch them with the SIMBAD database. Among the 119 outlier sources are 34 unclassified X-ray point sources located in the Galactic cen- ter (GC). Previous population studies of thousands of GC X-ray point sources suggest that the majority of the observed X-ray flux originates from class of magnetic cata- clysmic variable (CV) known as an intermediate polar (IP). We conduct a spectral analysis of five of the 34 OIA-flagged IP candidates (IPCs) by fitting their 0.5 − 8.0 keV spectra with one- and two-temperature plasma models as well as a continuum plus Gaussians model for the Fe line emission observed between 6.0 − 7.0 keV. We use the best-fitting plasma temperatures and the Fe line intensity ratios to estimate white dwarf mass (MWD) for each IPC. We find MWD for three IPCs consistent with estimates for GC sources and nearby IPs, while two IPCs can be interpreted as IPs with low MW D or as a class of non-magnetic CV known as a dwarf nova.
Details
- Title: Subtitle
- Data-driven exploration in x-ray astronomy: investigating intermediate polar candidates found among unusual objects selected by an unsupervised outlier identification algorithm
- Creators
- Dustin K Swarm
- Contributors
- Casey T DeRoo (Advisor)Shea Brown (Committee Member)Kenneth G. Gayley (Committee Member)Keri Hoadley (Committee Member)Cornelia C. Lang (Committee Member)
- Resource Type
- Dissertation
- Degree Awarded
- Doctor of Philosophy (PhD), University of Iowa
- Degree in
- Physics
- Date degree season
- Summer 2023
- Publisher
- University of Iowa
- DOI
- 10.25820/etd.006877
- Number of pages
- xiii, 122 pages
- Copyright
- Copyright 2023 Dustin K Swarm
- Language
- English
- Date submitted
- 07/19/2023
- Description illustrations
- color illustrations
- Description bibliographic
- Includes bibliographical references (pages 116-122).
- Public Abstract (ETD)
Applying “big-data” methods enables discovery amid the exponential growth in data happening in modern astronomy. My research combines machine learning techniques with spectral analysis to perform data-driven exploration in X-ray astronomy. I apply an algorithm to identify strange “outlier” sources among a large data set from the Chandra X-ray Observatory. I then investigate a subset of the outliers located in the galactic center to learn about how stars end their lives in this astronomical environment.
- Academic Unit
- Physics and Astronomy
- Record Identifier
- 9984454435202771
Metrics
7 File views/ downloads
21 Record Views