Study of filtering processes of light curves obtained with the CoRoT and Kepler satellites

Abstract

Light curves are astronomical data used for the study of the brightness variability of stars, which can potentially result in the discovery of exoplanets, due to the eclipses (detected in the light curves) caused by the passage of the exoplanets between the observed star and the observer. Light curves show high-frequency noise that can make it difficult to detect eclipses and to determine parameters (such as the radii of exoplanets and their distances relative to the stars they orbit) from the modeling of these astronomical data. In this work, we propose to evaluate, using available data from the CoRoT and Kepler space telescopes, which are the best filtering techniques capable of removing high-frequency noise from light curves, without compromising the analyses to be applied, involving exoplanets. In order to check the efficacy of the filtering methods, we will determine the parameters of exoplanets, from the modeling of light curves, before and after the filtering techniques are applied, with the purpose of evaluating the effect of the filtering in the values (and uncertainties) of such parameters. Using these same data, we also intend to check the efficacy of different machine learning techniques in the analysis of light curves, with the purpose of differentiating the light curves of stars with eclipsing binaries, with exoplanets or without eclipses. (AU)