Dresden 2020 – wissenschaftliches Programm
Die DPG-Frühjahrstagung in Dresden musste abgesagt werden! Lesen Sie mehr ...
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
MM: Fachverband Metall- und Materialphysik
MM 59: Computational Materials Modelling - Potentials
MM 59.3: Vortrag
Donnerstag, 19. März 2020, 16:15–16:30, IFW A
Machine Learning Augmented Density Functional Tight Binding Theory — •Adam McSloy1, Benjamin Hourahine2, David Yaron3, and Reinhard J. Maurer1 — 1Department of Chemistry, University of Warwick — 2Department of Physics, University of Strathclyde — 3Department of Chemistry, Carnegie Mellon University
Density functional tight binding theory (DFTB) is a cost-effective electronic structure method which excels at reaching the size and time scales normally off limits to Density Functional Theory and other electronic structure methods. However, the derivation of parameter sets for systems beyond simple semiconductors and organic molecules is a non-trivial task. Here, we present a deep machine learning framework to construct DFTB parameter sets and apply it to construct a new parameter set for hybrid organic-metallic materials. Furthermore, we show that machine learning offers a way to extend the DFTB method beyond the two-centre approximation, which enables a more accurate treatment of reactive chemistry and non-equilibrium dynamics.