BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Localized updates via machine learned flows to accelerate Markov C
hain Monte Carlo simulations
DTSTART:20240611T140000Z
DTEND:20240611T160000Z
DTSTAMP:20240723T154750Z
UID:indico-event-1398718@indico.cern.ch
DESCRIPTION:\n\nState-of-the-art simulations of discrete gauge theories ar
e based on Markov chains with local changes in the field space\, which how
ever at very fine lattice spacings are notoriously difficult due to separa
ted topological sectors of the gauge field. Hybrid Monte Carlo (HMC) algor
ithms\, which are very efficient at coarser lattice spacings\, suffer from
increasing autocorrelation times.\nAn approach\, which can overcome long
autocorrelation times\, is based on trivializing maps\, where a proposal o
f a new gauge configuration can be generated by mapping a configuration fr
om a trivial space to the target one\, distributed via the associated Bolt
zmann factor. \nI will discuss applications to the 2D Schwinger model and
strategies how to utilize the flow in large scale applications. One possi
ble way is to use the locality of the theory and only update local domains
. By defining local maps\, defects can be mapped to the target space\, whi
ch are able to unfreeze the topological charge in the simulation.\n\n\n\nh
ttps://indico.cern.ch/event/1398718/
LOCATION:CERN 4/2-037 - TH meeting room
URL:https://indico.cern.ch/event/1398718/
END:VEVENT
END:VCALENDAR