BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Hybrid Modelling of Interatomic Forces (Physics+Data+Mathematics) 
 - Christoph Ortner (University of Warwick)
DTSTART:20190502T134500Z
DTEND:20190502T143000Z
UID:TALK123904@talks.cam.ac.uk
CONTACT:Ferdia Sherry
DESCRIPTION:Accurate molecular simulation requires quantum chemistry model
 s that accurately capture the interaction between nuclei and electrons. Un
 fortunately\, their high computational cost makes these models unsuitable 
 for simulating complex material phenomena or large molecules. On the other
  hand\, interatomic potentials (IPs) are largely empirical models with poo
 r accuracy and narrow applicability. The past decade has seen a revival of
  IPs\, re-casting their construction as a “machine learning” instead o
 f a “modelling” problem.\n\nIn this talk I will backtrack from this tr
 end by returning to simple\nfunctional forms\, motivated by physics (and s
 omewhat justified by mathematics)\, but retain a key feature of the data-d
 riven/ML approach: models must be systematically improvable. I will demons
 trate how a combination of elementary ideas from analysis\,\napproximation
  theory\, and representation theory leads to a very competitive constructi
 on (in terms of fitting error/cost) that has a range of additional advanta
 ges\, for example that it can be effectively regularised. While my talk wi
 ll focus specifically on interatomic potentials\, the ideas should in prin
 ciple be applicable to a much wider range of particle models\, and at a mo
 re conceptual level also to continuum models.
LOCATION:Centre for Mathematical Sciences\, MR2
END:VEVENT
END:VCALENDAR