DPG Phi
Verhandlungen
Verhandlungen
DPG

Berlin 2024 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

MM: Fachverband Metall- und Materialphysik

MM 17: Poster Ib

MM 17.3: Poster

Monday, March 18, 2024, 18:30–20:30, Poster F

From ab-inito to experiments: A Python workflow for constructing neuroevolution potentials — •Eric Lindgren1, Adam Jackson2, Zheyong Fan3, Christian Müller4, Jan Swenson1, Thomas Holm-Rod5, and Paul Erhart11Department of Physics, Chalmers University of Technology, Gothenburg, Sweden — 2Centre for Sustainable Chemical Technologies and Department of Chemistry, University of Bath, United Kingdom — 3College of Physical Science and Technology, Bohai University, Jinzhou, People's Republic of China — 4Department of Chemistry and Chemical Engineering, Chalmers University of Technology, Gothenburg, Sweden — 5ESS Data Management and Software Center, Copenhagen, Denmark

Neuroevolution potentials, NEPs, as implemented in the GPUMD package are a highly accurate and computationally efficient type of machine-learned interaction potentials, enabling large scale MD simulations with system sizes up to millions of atoms with ab-initio level accuracy. Here, we present a Python workflow for constructing and sampling NEPs using the `calorine` package, and how the resulting trajectories can be analysed with the `dynasor` package to predict observables from scattering experiments. We focus on our recent work on crystalline benzene as an example system, but the approach is readily extendable to other systems.

Keywords: machine learning; force fields; potentials; molecular dynamics; neural network

100% | Mobile Layout | Deutsche Version | Contact/Imprint/Privacy
DPG-Physik > DPG-Verhandlungen > 2024 > Berlin