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

O: Fachverband Oberflächenphysik

O 35: Poster Session II (Polymeric biomolecular films; Nanostructures; Electronic structure; Spin-orbit interaction; Phase transitions; Surface chemical reactions; Heterogeneous catalysis; Particles and clusters; Surface magnetism; Electron and spin dynamics; Surface dynamics; Methods; Electronic structure theory; Functional molecules)

O 35.132: Poster

Dienstag, 27. März 2012, 18:15–21:45, Poster B

Construction of a Neural Network Potential-Energy Surface for R,R-Tartaric Acid — •Sinja Klees, Tobias Morawietz, and Jörg Behler — Lehrstuhl für Theoretische Chemie, Ruhr-Universität Bochum, D-44780 Bochum, Germany

The availability of accurate interatomic potentials is a necessary condition for studying complex systems for example by molecular dynamics simulations. Typically, a number of physical approximations are introduced to obtain potential-energy surfaces (PESs), which are sufficiently efficient to study large systems. Artificial Neural Networks (NN) offer a very general way to construct accurate PESs without the introduction of any constraint on the functional form. They are extremely flexible and allow to interpolate reference energies and forces obtained from first-principles calculations. To date, most NN potentials for molecules have been reported only for very small systems containing up to six atoms. Using R,R-tartaric acid as a benchmark system we show that also high-dimensional NN PESs for larger molecules can be constructed in a systematic way.