Regensburg 2025 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
CPP: Fachverband Chemische Physik und Polymerphysik
CPP 15: Poster Session I
CPP 15.64: Poster
Montag, 17. März 2025, 19:00–21:00, P4
Fourth-Generation High-Dimensional Neural Network Po- tentials for Molecular Chemistry in Solution — •Djamil Abdelkader Adel Maouene1,2, Moritz Richard Schäffer1,2, Moritz Gubler3, Stefan Goedecker3, and Jörg Behler1,2 — 1Theoretische Chemie II, Ruhr-Universität Bochum, Germany — 2Research Center Chemical Sciences and Sustainability, Research Alliance Ruhr, Germany — 3Department Physik, Universität Basel, Switzerland
Machine learning has become a powerful tool in chemistry and materials science, as high-dimensional neural network potentials (HDNNPs) provide accurate representations of multidimensional potential energy surfaces for atomistic simulations. In this study, we compare the performance of two types of HDNNPs; 2G-HDNNPs and 4G-HDNNPs, in modeling organic molecules in aqueous solution. While 2G-HDNNPs have proven effective in many systems in capturing local interactions based on atomic environments, they fail in scenarios where long-range charge transfer plays a critical role. These cases are better addressed by 4G-HDNNPs, which take into account atomic charge variations caused by structural or electronic changes even at distant regions in the system. Both methods are demonstrated using a model organic molecule.
Keywords: Atomistic Simulations; Machine Learning Potentials; High-Dimensional Neural Network Potential; 4G-HDNNPs; Molecular Dynamics