BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Nonstationary Gaussian process emulators with covariance mixtures 
 - Daniel Williamson (University of Exeter)
DTSTART:20180209T100000Z
DTEND:20180209T110000Z
UID:TALK100384@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:Routine diagnostic checking of stationary Gaussian processes f
 itted to the output of complex computer codes often reveals nonstationary 
 behaviour. There have been a number of approaches\, both traditional and m
 ore recent\, to modelling or accounting for this nonstationarity via the f
 itted process. These have included the fitting of complex mean functions t
 o attempt to leave a stationary residual process (an idea that is often ve
 ry difficult to get right in practice)\, using regression trees or other t
 echniques to partition the input space into regions where different statio
 nary processes are fitted (leading to arbitrary discontinuities in the mod
 elling of the overall process)\, and other approaches which can be conside
 red to live in one of these camps. In this work we allow the fitted proces
 s to be continuous by modelling the covariance kernel as a finite mixture 
 of stationary covariance kernels and allowing the mixture weights to vary 
 appropriately across parameter space. We introduce our method and compare 
 its performance with the leading approaches in the literature for a variet
 y of standard test functions and the cloud parameterisation of the French 
 climate model. This is work led by my final-year PhD student\, Victoria Vo
 lodina.
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR