Presentation #211.02 in the session “Computational Astronomy”.
With upcoming large all-sky surveys like LSST, accurate and fast classification will be necessary to work through the upcoming torrent of data, and spectroscopic surveys will be a valuable tool to aid in this classification of variable stars. The Time-Domain Spectroscopic Survey (TDSS) is a Sloan Digital Sky Survey (SDSS)-IV Extended Baryon Oscillation Spectroscopic Survey (eBOSS) subproject that has targeted a large sample of point sources selected based on variability in SDSS-I/II/III and PanSTARRS-1 photometry. The final TDSS sample includes optical spectra for over 50,000 stars selected as variable, spanning the entire variability tree across all spectra types, including both extrinsic and intrinsic variables. Of this sample, 23,595 stellar variables have photometric light curves in either the Catalina Sky Survey (CSS, Drake et al. 2014) or the Zwicky Transient Facility (ZTF, Bellm et al. 2019). We search all light curves for periodic signals and classify the objects found to be periodic variables, finding types such as EA, EB/EW, CVs, RRab, RRc, RRd, Delta Scuti. For non-periodic objects we provide a variety of variability statistics to search e.g., for flare and eruptive phenomena. For every star, we use the optical spectra to gain information about each source such as Hα emission, UVW space motions, effective temperatures, and absolute magnitudes. Our new PyHammer2.0 tool allows us to identify and classify many spectroscopic binaries in the sample. Using the spectral information together with the light curve statistics we are building a random forest classifier to detect and classify all TDSS variable stars.