BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Comparing gravitational-wave data and stellar-physics predictions 
 with deep learning - Prof. Davide Gerosa\, University of Milano-Bicocca
DTSTART:20230215T140000Z
DTEND:20230215T150000Z
UID:TALK197269@talks.cam.ac.uk
CONTACT:Isobel Romero-Shaw
DESCRIPTION:The catalog of gravitational-wave events is growing\, and so a
 re our hopes of constraining the underlying astrophysics of stellar-mass b
 lack-hole mergers by inferring the distributions of\, e.g.\, masses and sp
 ins. While conventional analyses parametrize this population with simple p
 henomenological models\, we propose an innovative physics-first approach t
 hat compares gravitational-wave data against astrophysical simulations. We
  combine state-of-the-art deep-learning techniques with hierarchical Bayes
 ian inference and exploit our approach to constrain the properties of repe
 ated black-hole mergers from the gravitational-wave events in the most rec
 ent LIGO/Virgo catalog. Deep neural networks allow us to (i) construct a f
 lexible population model that accurately emulates simulations of hierarchi
 cal mergers\, (ii) estimate selection effects\, and (iii) recover the bran
 ching ratios of repeated-merger generations. Among our results we find tha
 t: the distribution of host-environment escape speeds favors values <100 k
 m/s but is relatively flat\; first-generation black holes are born with a 
 maximum mass that is compatible with current estimates from pair-instabili
 ty supernovae\; there is multimodal substructure in both the mass and spin
  distributions due to repeated mergers\; and binaries with a higher-genera
 tion component make up at least 15% of the underlying population. The deep
 -learning pipeline we present is ready to be used in conjunction with real
 istic astrophysical population-synthesis predictions.
LOCATION:Both in-person (at MR2\, DAMTP) and online (details to be sent by
  email)
END:VEVENT
END:VCALENDAR