BEGIN:VCALENDAR
PRODID:-//eluceo/ical//2.0/EN
VERSION:2.0
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:9745be9a05dbdf97fa38d448f41fec91
DTSTAMP:20261022T190252Z
SUMMARY:Stochastic Analysis Seminar – Tobias Grafke (University of Warwic
 k)
DESCRIPTION:Tobias Grafke\nTitle: Sharp rare-event asymptotics and fluctuat
 ions around instantons\nAbstract\n\n\n\nRare and extreme events are notori
 ously hard to handle in complex stochastic systems: They are simultaneousl
 y too rare to be reliably observable in experiments or numerics\, but at t
 he same time often too impactful to be ignored. Large deviation theory pro
 vides a classical way of dealing with events of extremely small probabilit
 y\, but generally only yields the exponential tail scaling of rare event p
 robabilities. In this talk\, I will discuss theory\, and algorithms based 
 upon it\, that improve on this limitation\, yielding sharp quantitative es
 timates of rare event probabilities from a single computation and without 
 fitting parameters. Notably\, prefactor estimates\, which take into accoun
 t fluctuation corrections around the instanton\, require the computation o
 f determinants of differential operators. For multiplicative noise\, itera
 ted stochastic integrals introduce a second Wiener chaos contribution\, so
  that the resulting prefactor involves a Carleman-Fredholm determinant. Th
 roughout\, I will demonstrate the applicability of these methods to high-d
 imensional real-world systems\, for example coming from atmosphere and oce
 an dynamics.\n\n\n
URL:https://www.imperial.ac.uk/events/215529/stochastic-analysis-seminar-to
 bias-grafke/
DTSTART;TZID=Europe/London:20261027T150000
DTEND;TZID=Europe/London:20261027T160000
LOCATION:410\, Huxley Building\, South Kensington Campus\, Imperial College
  London\, London\, SW7 2AZ\, United Kingdom
END:VEVENT
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:STANDARD
DTSTART:20261027T150000
TZNAME:GMT
TZOFFSETTO:+0000
TZOFFSETFROM:+0000
END:STANDARD
END:VTIMEZONE
END:VCALENDAR