DPG Phi
Verhandlungen
Verhandlungen
DPG

Regensburg 2025 – scientific programme

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

MM: Fachverband Metall- und Materialphysik

MM 31: Data-driven Materials Science: Big Data and Worksflows

MM 31.10: Talk

Thursday, March 20, 2025, 17:30–17:45, H10

A systematic benchmark of G0W0 calculations — •Marc Thieme, Max Großmann, Malte Grunert, and Erich Runge — Technische Universität Ilmenau, Ilmenau, Germany

Accurate and efficient ab initio electronic structure calculations of semiconductors and insulators are a prerequisite for building large, high-quality databases for machine learning (ML). However, the "optimal" choice (speed vs. accuracy) of the approximations used, i.e. the exchange-correlation (XC) functional for density functional theory (DFT) calculations or a particular many-body perturbation theory, remains unclear. A systematic benchmark of band gaps of solids using several different DFT XC functionals by Borlido et al [1,2] showed that hybrid functionals perform exceptionally well and seem to be the functionals of choice. The present study addresses the question of whether G0W0 calculations provide a sufficient increase in accuracy to justify their increased computational cost compared to simpler DFT calculations with hybrid functionals. We calculate the band gaps for about 300 materials using the G0W0 method starting from LDA/PBE DFT calculations. The deviations between G0W0 and experimental band gaps are systematically compared with those of the best hybrid functionals

[1] Borlido et al., J. Chem. Theory Comput. 15, 9 (2019)

[2] Borlido et al., npj Comput. Mater. 6, 96 (2020)

Keywords: Benchmark; Band gap; DFT; GW; Workflow

100% | Mobile Layout | Deutsche Version | Contact/Imprint/Privacy
DPG-Physik > DPG-Verhandlungen > 2025 > Regensburg