Title: Constructing probabilistic templates for astronomical lightcurves
Authors: David Jones - Texas A&M University (United States) [presenting]
Sujit Ghosh - North Carolina State University (United States)
Ana-Maria Staicu - North Carolina State University (United States)
Ashish Mahabal - California Institute of Technology (United States)
Abstract: The lightcurve of an astronomical source (e.g. a star) is its light intensity as a function of time. New telescopes such as the Large Synoptic Survey Telescope (LSST) will observe billions of lightcurves at tens to hundreds of time points, and astronomers want to classify each and make decisions about follow-up observations. Many of the most interesting and informative astronomical sources have periodic lightcurves with shapes that are characteristic of the particular class of source. Given a sufficient number of pre-classified lightcurves we can infer the lightcurve shapes characteristic of different classes (or, more precisely, the distribution of characteristic lightcurve shapes). In this manner we can construct probabilistic templates for lightcurves from each class. These templates can then be input into methods for the key tasks of classifying new sources and optimally scheduling further observations (of sources with few observations). The success of the classification and scheduling methods depends highly on the development of accurate probabilistic templates and we discuss our data driven approach to construct bases that can efficiently capture important lightcurve features. We apply our method to data from the Catalina Real-time Transient Survey.