Identifications of RR Lyrae Stars and Quasars from the Simulated Data of Mephisto-W Survey

Abstract We have investigated the feasibilities and accuracies of identifications RR Lyrae stars quasars from simulated data Multi-channel Photometric Survey Telescope (Mephisto) W Survey. Based on variable sources light curve libraries Sloan Digital Sky (SDSS) Stripe 82 observation history simulation Mephisto-W Scheduler, we uvgriz multi-band curves stars, other for first year Mephisto applied ensemble machine learning algorithm Random Forest Classifier (RFC) to identify quasars, respectively. build training test samples extract ∼150 features train two RFCs respectively star quasar classification. find that, our are able select with remarkably high precision completeness, purity = 95.4% completeness 96.9% RFC 91.4% 90.2% RFC. also derived relative importances extracted utilized classify quasars.