Berlin 2024 – scientific programme

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

MM: Fachverband Metall- und Materialphysik

MM 53: Data Driven Material Science: Big Data and Workflows VI

MM 53.2: Talk

Thursday, March 21, 2024, 10:30–10:45, C 243

SE(3)-Transformers for predicting the electronic structure of hydrogen molecules — •Bartosz Brzoza and Attila Cangi — CASUS/HZDR

In this work, we demonstrate the efficacy of a neural network model implemented as the Materials Learning Algorithms (MALA) package in predicting the electronic structure of a system of hydrogen molecules under various pressure and temperature conditions across the molecular liquid-solid phase boundary, demonstrating the potential of our methods for molecular systems. Additionally, we investigate the use of SE(3)-Transformer Graph Neural Networks to improve the generalizability and extrapolation capabilities of our models. Our results indicate that the MALA framework provides a powerful and efficient tool for accelerating Kohn-Sham density functional theory calculations in molecular systems. This work paves the way for future research in developing advanced machine-learning algorithms for accelerating electronic structure calculations both accurately and efficiently.

Keywords: MALA; Graph; Neural Networks; Hydrogen; Molecules