BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20210916T132447Z
LOCATION:Michel Mayor
DTSTART;TZID=Europe/Stockholm:20210705T153000
DTEND;TZID=Europe/Stockholm:20210705T160000
UID:submissions.pasc-conference.org_PASC21_sess119_msa246@linklings.com
SUMMARY:Four Generations of Neural Network Potentials
DESCRIPTION:Minisymposium\n\nFour Generations of Neural Network Potentials
 \n\nBehler\n\nA lot of progress has been made in recent years in the devel
 opment of atomistic potentials employing machine learning (ML). Neural net
 work potentials (NNPs), which have first been proposed about two decades a
 go, are an important class of ML potentials. While first generation NNPs h
 ave been restricted to small molecules with only a few degrees of freedom,
  second generation NNPs relying on local atomic environments are applicabl
 e to high-dimensional systems containing thousands of atoms. Long-range el
 ectrostatic interactions beyond the local environments can be included bas
 ed on environment-dependent charges in third generation NNPs. Recently, se
 veral limitations of these local approaches have been identified resulting
  in the development of the fourth generation of NNPs, which is now able to
  describe long-range charge transfer and is applicable to systems in multi
 ple charge states. In this talk the evolution of NNPs is discussed with a 
 focus on constructing NNPs for high-dimensional condensed systems includin
 g long-range interactions.\n\nDomain: Chemistry and Materials
END:VEVENT
END:VCALENDAR