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:DAMTP Statistical Physics and Soft Matter Seminar
SUMMARY:Multilevel Monte Carlo simulation for scale-separa
ted systems - Paul B. Rohrbach\, DAMTP
DTSTART;TZID=Europe/London:20200526T130000
DTEND;TZID=Europe/London:20200526T140000
UID:TALK142054AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/142054
DESCRIPTION:Designing efficient Monte Carlo simulations for mu
ltiscale physical systems is often a difficult pro
blem. The underlying Markov chain can typically on
ly modify the system locally on a small-scale leve
l. This means that the long length-scale degrees o
f freedom of the system tend to decorrelate very s
lowly. We propose a structured approach to circumv
ent slow simulations in this situation. Assume tha
t the degrees of freedom corresponding to the long
length scales give rise to a coarse grained syste
m which can be simulated efficiently. Instead of s
imulation the full system\, we first compute the c
oarse-grained approximation and subsequently corre
ct the coarse-graining error by a simulation of th
e remaining fine degrees of freedom. This is inspi
red by the multilevel Monte Carlo approach that ha
s been developed in recent mathematical studies fo
r applications in Bayesian inference and mathemati
cal finance. We develop a convergence theory of ou
r method and apply it to the Asakura-Oosawa (AO) m
odel and size-asymmetrical binary hard-sphere syst
ems. This allows us to investigate equilibrium pro
perties of systems with a large size-ratio where a
naive simulation would have been computationally
infeasible.
LOCATION:Zoom: https://zoom.us/j/5916271322 Meeting ID: 591
-627-1322
CONTACT:Etienne Fodor
END:VEVENT
END:VCALENDAR