BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Isaac Newton Institute Seminar Series
SUMMARY:MSG Design of Experiments Seminar Series: Simulati
on-based Bayesian experimental design for computat
ionally intensive models - Xun Huan (Sandia Nation
al Laboratories\; University of Michigan)
DTSTART;TZID=Europe/London:20180620T145500
DTEND;TZID=Europe/London:20180620T154500
UID:TALK107146AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/107146
DESCRIPTION:Selecting and performing experiments that produce
the most useful data is extremely valuable in eng
ineering and science applications where experimen
ts are costly and resources are limited. Simulatio
n-based experimental design thus provides a rigor
ous mathematical framework to systematically quan
tify and maximize the value of experiments while
leveraging the existing knowledge and predictive c
apability of an available model. \;We are pa
rticularly interested in design settings that acc
ommodate nonlinear and computationally intensive m
odels\, such as those governed by ordinary and pa
rtial differential equations. Employing principle
s from Bayesian statistics to characterize and qua
ntify uncertainty\, we seek experiments that maxi
mize the expected information gain. Computing the
se optimal designs using conventional approaches\,
however\, is generally intractable. Major challe
nges include high dimensional parameter spaces\,
expensive model simulations\, and numerical appro
ximation and optimization of the expected informat
ion gain. We thus describe practical numerical me
thods to help overcome these obstacles\, includin
g global sensitivity analysis\, surrogate modeling
via polynomial chaos\, and stochastic optimizati
on. \;The overall methodology is demonstrate
d through the design of combustion experiments for
optimal learning of chemical rate parameters\, a
nd of configurations for a supersonic jet engine
to obtain measurements most informative on turbul
ent flow parameters.

LOCATION:Seminar Room 1\, Newton Institute
CONTACT:INI IT
END:VEVENT
END:VCALENDAR