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:20190719T085744Z
LOCATION:HG EO Nord
DTSTART;TZID=Europe/Stockholm:20190613T195000
DTEND;TZID=Europe/Stockholm:20190613T215000
UID:submissions.pasc-conference.org_PASC19_sess179_post116@linklings.com
SUMMARY:CHM03 - Computing <em>Ab Initio</em> Molecular-Dynamics Trajectori
 es Using Artificial Neural Networks with Different Architectures and Featu
 re Spaces
DESCRIPTION:Poster\n\n\nCHM03 - Computing <em>Ab Initio</em> Molecular-Dyn
 amics Trajectories Using Artificial Neural Networks with Different Archite
 ctures and Feature Spaces\n\nBöselt, Riniker\n\nAccurate calculation of th
 e potential of mean force (PMF) of chemical reactions requires long <em>ab
  initio</em> molecular-dynamics (AIMD) trajectories using a computationall
 y expensive description of the physical interactions like density-function
 al theory (DFT) or wave-function (WF) methods. Artificial neural networks 
 (ANNs) belong to the supervised machine-learning approaches, which can be 
 used to accelerate AIMD simulations and thus, can facilitate the computati
 on of PMFs. However, different ANN architectures and feature spaces can in
 fluence the training, the convergence, and the final performance of the AN
 N significantly. In this work, we compare, how the use of ANNs with differ
 ent architectures and feature spaces reduces the complexity of AIMD simula
 tions. The performance in terms of accuracy is assessed by calculating PMF
 s of simple chemical reactions.
END:VEVENT
END:VCALENDAR