Dual signatures in binary light curves: eccentricity and flare activity in TESS data

Supervised machine learning methods require sufficiently representative samples of input data, particularly observational light curves or their characteristic parameters. To compile such samples, data from multiple astronomical surveys are typically employed. However, preparing a reliable training set is often a non-trivial task, as it demands considerable resources: time and the expertise of qualified astronomers to classify and determine key parameters necessary for building an effective machine learning model. In this work, we present a sample of objects with two distinctive features in their light curves, selected from the catalog of eclipsing binaries observed by the TESS space mis- sion. These features are (i) orbital eccentricity, which visually manifests as a deviation of the secondary eclipse from the midpoint between successive primary eclipses, and (ii) the simultaneous presence of flare activity. An initial set of 38 candidates was selected, of which 20 systems were ultimately included in the final sample. Based on calculated eclipse timings, we derived preliminary values of the secondary minimum phase (assuming the primary eclipse occurs at phase zero), as well as the times of peak flare events observed in the light curves of these binary systems.