BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Multilevel Emulation and History Matching of EAGLE: an expensive h
 ydrodynamical Galaxy formation simulation. - Ian Vernon (University of Dur
 ham)
DTSTART:20180412T133000Z
DTEND:20180412T140000Z
UID:TALK103732@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:We discuss strategies for performing Bayesian uncertainty anal
 yses for extremely expensive simulators. The EAGLE model is one of the mos
 t (arguably the most) complex hydrodynamical Galaxy formation simulations 
 yet performed. It is however extremely expensive\, currently taking approx
 imately 5 million hours of CPU time\, with order of magnitude increases in
  runtime planned. This makes a full uncertainty analysis involving the exp
 loration of multiple input parameters along with several additional uncert
 ainty assessments\, seemingly impossible. We present a strategy for the re
 solution of this problem\, which incorporates four versions of the EAGLE m
 odel\, of varying speed and accuracy\, within a specific multilevel emulat
 ion framework that facilitates the incorporation of detailed judgements re
 garding the uncertain links between the physically different model version
 s. We show how this approach naturally fits within the iterative history m
 atching process\, whereby regions of input parameter space are identified 
 that may lead to acceptable matches between model output and the real univ
 erse\, given all major sources of uncertainty. We will briefly discuss the
  detailed assessment of such uncertainties as observation error and struct
 ural model discrepancies and their various components\, and emphasise that
  without such assessments any such analysis rapidly loses meaning. 
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR