Technosignatures with TESS: Anomaly Detection and Data Exploration in Detail

The Transiting Exoplanet Survey Satellite (TESS) has revolutionized the study of stellar variability through its extensive collection of full frame images. With millions of stars observed at a high cadence, TESS has provided a rich dataset for investigating various celestial phenomena. While the primary focus has been on studying transiting exoplanets, spotted stars, and pulsators, the vastness of the dataset also presents an opportunity to uncover more anomalous variability, such as transiting megastructures.

Drawing inspiration from the discoveries made by astronomers and citizen scientists using TESS's predecessor, Kepler, we have embarked on a comprehensive search for anomalous variability within the first two years of TESS observations (sectors 1-26). To facilitate this search, we have employed statistical and machine learning methods to enable efficient detection and exploration of anomalies in the light curves. We present here the multi-step process employed to translate light curve data into anomaly scores.

For the top ~2% of potentially anomalous fading light curves returned by our procedures, we have also established a visual and statistical vetting process aimed at separating observing artifacts from truly odd variability. Our methods rely heavily on an interactive database based on SkyPortal architecture. What remains after this careful vetting are multiple useful data products. We present examples of problematic systematics by sector, typical types of known variables, and a handful of sources with as yet undetermined astrophysical processes which merit further investigation.