Regensburg 2016 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

O: Fachverband Oberflächenphysik

O 3: Focus Session: Morphology Prediction at Interfaces

O 3.3: Topical Talk

Montag, 7. März 2016, 11:30–12:00, S051

Simulations of Surfaces and Interfaces Using High-Dimensional Neural Network Potentials — •Jörg Behler — Lehrstuhl für Theoretische Chemie, Ruhr-Universität Bochum, 44780 Bochum, Germany

In recent years a lot of progress has been made in the development of interatomic potentials based on machine learning techniques. These potentials are very efficient and thus allow to extend the length and time scales of molecular dynamics simulations significantly. Further, they are close to the accuracy of the underlying reference electronic structure methods and equally applicable to all types of bonding making them particularly useful for complex systems like interfaces. After a short overview about the available approaches, specifically the class of high-dimensional neural network potentials will be discussed. Several applications including water and solid-liquid interfaces will be presented to illustrate the advantages and current limitations of this method.