Using Machine-Learning to obtain Be star candidates

Abstract

Classical B-type emission line (Be) stars are nonsupergiant stars whose spectra exhibit or have ever exhibited Hα or more Balmer lines in emission (Jaschek et al. 1981). The properties observed in the Hα line profile of Be stars are consistent with the presence of an optically thin gaseous circumstellar disk (flattened in the plane perpendicular to the axis of rotation) in which the gas is ionized by ultraviolet radiation from the central object. Although the definition of a star is based on its spectroscopic properties, it is possible to detect them photometrically as long as the emission in the H line is very intense. One technique commonly used to identify α classical Be stars candidates is the use of photometric 2-color diagrams (2CDs) that utilize a narrowband filter centered on H and an associated filter that samples the nearby continuum region (e.g., Raddi et al. 2015), for example, (r – H) versus (r – i).