HdR Maria Pruzhinskaya

jeudi 10 sept. 2026, 14:00 → 17:00 Europe/Paris

Amphi de recherche (Eupi)

Supernovae, anomaly detection, and time-domain astronomy

This Habilitation thesis summarises my research trajectory from supernova cosmology to the search for rare and anomalous objects in modern time-domain surveys. Starting from Type Ia supernovae as cosmological distance indicators, my work has addressed several factors that affect their standardisation, including host-galaxy environment, galactocentric distance, peculiar velocities, and the physical interpretation of light curves through their modelling.

This interest in supernovae naturally led to a broader question: how can rare, peculiar, or previously unnoticed astrophysical phenomena be found in survey data that are too large for visual inspection? Within the SNAD project and the Fink broker, my research has focused on answering this question by combining machine-learning methods, expert feedback, follow-up observations, and astrophysical analysis. Applied first to ZTF data releases and then to real-time alert streams, this approach has led to the discovery and study of unusual supernovae, cataclysmic variables, M-dwarf flares, microlensing events, rare variable stars, and other anomalous sources. Overall, the thesis shows that anomaly detection becomes scientifically productive when it is embedded in a complete discovery cycle: machine-learning algorithms select and rank unusual candidates, experts interpret their astrophysical significance, and follow-up observations provide the additional information needed to understand their physical nature.