Classifying Unidentified X-ray Sources in the Chandra Source Catalog Using a Multi-wavelength Machine Learning Approach
Hui Yang1, Jeremy Hare2, Oleg Kargaltsev1, Steven Chen1, Igor Volkov1, Blagoy Rangelov3, Yichao Lin1,
1The George Washington University 2NASA GSFC 3Texas State University
CHECK our MUWCLASS paper
Related Papers: NGC 3532, 13 FGL-LAT source, Visualization tool, 4XMM-DR13 TD
Visualization Tools : CSCv2.0 Training Dataset, 4XMM-DR13 Training Dataset, 4FGL-DR4 catalog visualization, Classification of CSCv2.0 Sources within Unidentified 4FGL-DR4 Sources
contact huiyang@gwu.edu if you have any questions
This github repo provides the MUltiWavelength Machine Learning CLASSification Pipeline (MUWCLASS) and the classification results on the Chandra Source Catalog v2 (CSCv2).
The main components of this github repo are
demos/
- There are notebooks of demonstrations of classifying CSCv2 sources using MUWCLASS with CSCv2 and multiwavelength data
files/{CSC_TD_11042023_MW_allcolumns.csv, tbabs.data}
- Some other CSV files including the raw training dataset with more properties (CSC_TD_MW_remove.csv), the photoelectric absorption cross-section file (tbabs.data).
-
clone the MUWCLASS package to your local desktop
-
run the follow code to create a new conda environment muwclass; if you already have Python 3.9, you can use your own conda environment with additional Python packages installed from below
-
conda create -n muwclass python=3.9
-
run 'bash install-packages.sh' under muwclass environment to install all required packages
-
run 'pip install gdpyc' under muwclass environment
-
clone (NOT pip install) the NWAY package to your local desktop and change the nway_dir variable in nway_match.py line 25 in MUWCLASS repo to the directory where you clone the nway package