Long-Period High-Amplitude Red Variables in the KELT Survey

We present a sample of 4,132 Mira-like variables (red variables with long periods and high amplitudes) in the Kilodegree Extremely Little Telescope (KELT) survey. Of these, 814 are new detections. We used 2MASS colors to identify candidate asymptotic giant branch (AGB) stars. We tested for photometric variability among the sample and used Lomb-Scargle to determine the periodicity of the variable sample. We selected variables with high amplitudes and strong periodic behavior using a Random Forest classifier. Of the sample of 4,132 Mira-like variables, we estimate that 70% are Miras, and 30% are semi-regular (SR) variables. We also adopt the method of using (W_{RP} - W_{K_s}) vs. (J - K_s) colors (Lebzelter et al. 2018) in distinguishing between O-rich and C-rich Miras and find it to be an improvement over 2MASS colors.