With the start of the Rubin Observatory’s Legacy Survey of Space and
Time, the LSST AGN Science Collaboration is on the precipice of
discovering tens of millions of new quasars and Active Galactic
Nuclei. Just as machine learning (ML) applied to SDSS data increased
quasar discovery space by an order of magnitude, so too will it for
LSST. While lacking representative samples to train on, we are
encouraged by the latest advances in ML, which I will discuss.
Ultimately the goal is not to simply discover AGNs, but to use them to
investigate galaxy evolution and black hole physics. Lacking a
spectroscopic component, I will discuss how the LSST team hopes to
leverage time-domain information to be able to estimate mass-weighted
accretion rates from the photometry alone. How AGNs accrete and vary
is a function of their accretion disk physics and geometry. Thus, it
is of considerable interest that, while perhaps still a gold standard
for X-ray binaries, 2024 saw the demotion of the standard thin
accretion disk model as applied to AGNs – in favor of a much thicker
hyper-magnetized, flux-frozen model derived from following galaxy
formation processes to small scales. I’ll close with some commentary
on how this worked is shaped by AI and by considering how AGN
astronomers can think about winds and jets within the context of that
model.
