BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Latent SDEs for Modelling Quasar Variability and Inferring Black H
 ole Properties
DTSTART;VALUE=DATE-TIME:20230630T140000Z
DTEND;VALUE=DATE-TIME:20230630T141500Z
DTSTAMP;VALUE=DATE-TIME:20260306T095813Z
UID:indico-contribution-841@cern.ch
DESCRIPTION:Speakers: Joshua Fagin (CUNY)\nActive galactic nuclei (AGN) ar
 e thought to be powered by the accretion of matter around supermassive bla
 ck holes at the centers of galaxies. The time-dependent variability of an 
 AGN's brightness can provide valuable insights into the physical character
 istics of its underlying black hole. The variability can be well modeled b
 y a damped random walk process described by a stochastic differential equa
 tion (SDE). Upcoming wide-field telescopes such as the Rubin Observatory L
 egacy Survey of Space and Time (LSST) are expected to observe 100 million 
 AGN in multiple bandpass filters\, so new methods need to be developed to 
 analyze the large volume of light curve data. Latent SDEs are variational 
 auto encoders (VAEs) with a neural SDE as the decoder. Latent SDEs are wel
 l suited for modeling the AGN time series\, as they explicitly model the u
 nderlying dynamics. We modify latent SDEs to jointly reconstruct the unobs
 erved portions of multivariate AGN light curves as well as infer their phy
 sical properties\, such as the black hole mass. We train our model on a re
 alistic physics-based simulation of ten-year LSST light curves and find ou
 r method outperforms a multi-output Gaussian process regression in light c
 urve reconstruction. Our method has the potential to provide a deeper unde
 rstanding of the physical properties of black holes and AGN variability an
 d may be applicable to a wide range of other astronomical times series.\n\
 nhttps://indico.unina.it/event/61/contributions/841/
LOCATION:Centro Congressi Federico II Aula Magna
URL:https://indico.unina.it/event/61/contributions/841/
END:VEVENT
END:VCALENDAR