BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Machine learning potentials and excited-state molecular dynamics
DTSTART:20170210T100000
DTEND:20170210T110000
DTSTAMP:20260407T095821Z
UID:67ea1b06e5c9f98d2f406349faee7615929a987ec0433af4e5014ad2
CATEGORIES:Conferences - Seminars
DESCRIPTION:Professor Philipp Marquetand\, University of Vienna\, Institut
 e of Theoretical Chemistry\nArtificial neural networks can learn the relat
 ionship between the nuclear geometry of a molecule\nand the corresponding 
 potential energy and\, in this way\, serve as a highly accurate and extrem
 ely\nfast tool for predicting potential energy surfaces [1]. As an applica
 tion\, the simulation of an organic\nreaction with such a machine learning
  algorithm is presented [2]. Furthermore\, with the right neural\nnetwork 
 setup\, the chemical locality is inherently exploited\, where the local re
 gion of a large\nmolecule is only weakly influenced by the atoms that are 
 far from the region of interest [3]. As a\nresult\, a vast amount of compu
 tational time can be saved when generating the so-called reference\ndata f
 rom which the neural network learns the shape of the potential energy surf
 aces. Such\nimprovements will enable us to perform on-the-fly dynamics cal
 culations of large systems and long\ntime scales in the future.\nCurrently
 \, we still use highly accurate but time-consuming ab initio calculations 
 in semiclassical\nmolecular dynamics. Here\, we study molecules in their e
 xcited electronic states by the surface\nhopping method\, where the nuclei
  move classically on potentials computed by quantum mechanics.\nWe have mo
 dified the surface hopping approach to incorporate not only kinetic dynami
 cal\ncouplings but also any other arbitrary coupling in our so-called SHAR
 C (surface hopping including\narbitrary couplings) method [4\, 5]. The cor
 responding code is also publicly available [6]. Especially\nspin-orbit cou
 plings can now be treated in on-the-fly simulations. The method is applied
  to a variety\nof systems\, showing e.g. that intersystem crossing can tak
 e place on a femtosecond timescale even\nin organic molecules with only re
 latively small spin-orbit couplings (see [5] and references therein).\nLef
 t: Artificial neural networks are algorithms inspired by the network of ne
 urons in the human brain. Right: Logo of\nthe SHARC (Surface Hopping inclu
 ding ARbitrary Couplings) software.\nReferences\n[1] J. Behler\, J. Phys.:
  Condens. Matter\, 26 (2014) 183001.\n[2] M. Gastegger\, P. Marquetand J. 
 Chem. Theory Comput.\, 11 (2015) 2187-2198.\n[3] M. Gastegger\, C. Kauffma
 nn\, J. Behler\, P. Marquetand\, J. Chem. Phys.\, 144 (2016) 194110.\n[4] 
 M. Richter\, P. Marquetand\, J. González-Vázquez\, I. Sola\, L. Gonzále
 z\, J. Chem. Theory\nComput.\, 7 (2011) 1253-1258.\n[5] S. Mai\, P. Marque
 tand\, L. González\, Int. J. Quantum Chem.\, 115 (2015) 1215-1231.\n[6] S
 . Mai\, M. Richter\, M. Ruckenbauer\, M. Oppel\, P. Marquetand\, L. Gonzá
 lez\, SHARC\nProgram Package for Non-Adiabatic Dynamics\, sharc-md.org (20
 14).
LOCATION:BCH 3118 https://plan.epfl.ch/?room==BCH%203118
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR