Orateur
Description
Blazars are among the most violent non-thermal sources in the Universe, exhibiting dramatic variability across the electromagnetic spectrum. Yet, a key question remains: how tightly coupled are their optical and $\gamma$-ray emissions, and what does this reveal about the physical processes powering relativistic jets?
In this work, we present a multi-year study of optical–$\gamma$-ray correlations in blazars using data from the Fermi Large Area Telescope (LAT) and the Zwicky Transient Facility (ZTF). We develop a similarity metric to quantify cross-band correlations together with a robust flare-detection algorithm. Applying these methods to a large sample of blazars, we find that $\sim23\%$ show $>3 \sigma$ correlation hints between optical and $\gamma$-ray variability, often with short or absent time lags, suggesting possible co-spatial emission regions. We also investigate whether optical–$\gamma$-ray correlation signatures can help distinguish between different emission processes.
We further show that optical variability can be used to trigger high-energy observations. Our real-time flare-detection algorithm reaches a purity of 69% in identifying gamma-ray flares and 99% in detecting optical low states. These results demonstrate that large scale optical surveys can provide reliable triggers for high-energy monitoring. We implemented this methodology within the Fink broker to enable real-time identification of blazar high and low states from alert streams, providing automated triggers that can complement monitoring by Fermi-LAT and enable rapid follow-up with facilities such as the imaging atmospheric Cherenkov telescopes.
Furthermore, we are developing a more general machine-learning pipeline aimed at improving the identification of flaring states in real-world, multiwavelength light curves. This framework relies on unsupervised anomaly-detection and clustering methods to distinguish intrinsic variability from potential flares without requiring labeled training data. Designed to handle heterogeneous cadences and observational gaps across bands, it combines optical observations from ZTF with $\gamma$-ray data from Fermi-LAT and will ultimately be integrated into Fink to identify optical patterns and predict high-energy activity from changes detected in LSST.