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:Theory of Condensed Matter
SUMMARY:Four Generations of High-Dimensional Neural Networ
 k Potentials - Prof. Jörg Behler (Ruhr-Universität
  Bochum)
DTSTART;TZID=Europe/London:20250619T140000
DTEND;TZID=Europe/London:20250619T153000
UID:TALK227407AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/227407
DESCRIPTION:Machine learning potentials (MLPs) have become an 
 important tool for atomistic simulations in many f
 ields\, from chemistry to materials science. The r
 eason for the popularity of MLPs is their ability 
 to provide very accurate energies and forces\, whi
 ch are essentially indistinguishable from the unde
 rlying reference electronic structure calculations
 . Still\, the computational costs are much reduced
  enabling large-scale simulations of complex syste
 ms. Almost two decades ago\, in 2007\, the introdu
 ction of high-dimensional neural network potential
 s (HDNNP) by Behler and Parrinello paved the way f
 or the application of MLPs to condensed systems co
 ntaining a large number of atoms. Still\, the orig
 inal second-generation HDNNPs\, like most current 
 MLPs\, are based on a locality approximation of th
 e atomic interactions that are truncated at some f
 inite distance. Third-generation MLPs contain long
 -range electrostatic interactions up to infinite d
 istance and overcome this restriction to short-ran
 ge energies. Still\, there are surprisingly many s
 ystems in which long-range electrostatic interacti
 ons are insufficient for a physically correct desc
 ription\, since non-local phenomena like long-rang
 e charge transfer are essential. Such global effec
 ts can be considered in fourth-generation HDNNPs. 
 In this talk the evolution of HDNNPs will be discu
 ssed along with some key systems illustrating thei
 r applicability.
LOCATION:Seminar Room 3\, RDC
CONTACT:Bo Peng
END:VEVENT
END:VCALENDAR