Machine Learning to Search for Accreting Neutron Star Binary Candidates Using Chinese Space Station Telescope Photometric System

Abstract

Accreting neutron star binary (ANSB) systems can provide some important information about neutron stars (NSs), especially on the structure and the equation of state of NSs. However, only a few ANSBs are known so far. The upcoming Chinese Space Station Telescope (CSST) provides an opportunity to search for a large number of ANSB candidates. We aim to investigate whether or not a machine learning method may efficiently search for ANSBs based on CSST photometric system. In this paper, we generate some ANSBs and normal binaries under CSST photometric system by binary evolution and binary population synthesis method and use a machine learning method to train a classification model. We consider the classical multi-color disk and the irradiated accretion disk, then compare their effects on the classification results. We find that no matter whether the X-ray reprocessing effect is included or not, the machine learning classification accuracy is always very high, i.e., higher than 96%. If a significant magnitude difference exists between the accretion disk and the companion of an ANSB, machine learning may not distinguish it from some normal stars such as massive main sequence stars, white dwarf binaries, etc. False classifications of the ANSBs and the normal stars highly overlap in a color–color diagram. Our results indicate that machine learning would be a powerful way to search for potential ANSB candidates from the CSST survey.