Regensburg 2025 – scientific programme

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

HL: Fachverband Halbleiterphysik

HL 3: Focus Session: Machine Learning of semiconductor properties and spectra

HL 3.8: Talk

Monday, March 17, 2025, 12:45–13:00, H17

Learning an effective Hamiltonian for large-scale electronic-structure calculations — •Martin Schwade and David A. Egger — TUM School of Natural Sciences, Technical University of Munich, 85748 Garching, Germany

Exploring the optoelectronic properties of large-scale systems across various temperatures using conventional density functional theory (DFT) often encounters significant computational challenges. Recent advancements in machine learning force fields (ML-FFs) have made it easier to generate atomic trajectories at different temperatures. However, determining the electronic structure with temperature dependence remains a difficult task. Building on our earlier work involving a temperature-transferable tight-binding (TB) model [1] to learn an effective Hamiltonian, we introduce an extension of this method that leverages machine learning techniques to increase the accuracy and transferability of this approach. By integrating ML with TB models, this strategy offers a promising pathway to evaluate temperature-dependent material properties with reduced computational demands. [1] M. Schwade, M.J. Schilcher C. Reverón Baecker, M. Grumet, D. A. Egger, J. Chem. Phys. 160, 134102 (2024)

Keywords: Hamiltonian-learning; Machine learning; tight binding; electronic structure; machine learning force fields