BEGIN:VCALENDAR
VERSION:2.0
CALSCALE:GREGORIAN
PRODID:UW-Madison-Physics-Events
BEGIN:VEVENT
SEQUENCE:0
UID:UW-Physics-Event-8530
DTSTART:20231219T210000Z
DURATION:PT1H0M0S
DTSTAMP:20260414T012211Z
LAST-MODIFIED:20231201T224445Z
LOCATION:5310 CH
SUMMARY:Cosmology at the Field Level with Probabilistic Machine Learni
 ng\, Graduate Program Event\, Adam Rouhiainen\, Physics PhD Graduate S
 tudent
DESCRIPTION:The large-scale structure is highly non-Gaussian at late t
 imes and small length scales\, making it difficult to describe analyti
 cally. Parameter inference\, data reconstruction\, and data generation
  are greatly aided by various machine learning models\, and this work 
 takes a field level approach to solving these problems. The probabilit
 y distribution of the large-scale structure is learned with normalizin
 g flows\, allowing Bayesian reconstruction of noisy fields with remove
 d foregrounds. The normalizing flow is trained to be conditional on co
 smological variables\, from which accurate parameter estimation can be
  done. Turning to highly expressive denoising diffusion models\, a sup
 er-resolution emulator is developed for large cosmological simulation 
 volumes\, allowing high-resolution simulation volumes to be conditiona
 lly generated from low-resolution volumes. The super-resolution emulat
 or is trained to perform outpainting\, and can thus upgrade very large
  cosmological volumes from low-resolution to high-resolution using an 
 iterative outpainting procedure.
URL:https://www.physics.wisc.edu/events/?id=8530
END:VEVENT
END:VCALENDAR