Berlin 2024 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
MM: Fachverband Metall- und Materialphysik
MM 43: Data Driven Material Science: Big Data and Workflows V
MM 43.6: Vortrag
Mittwoch, 20. März 2024, 17:15–17:30, C 243
Accurate and Efficient Protocols for High-Throughput Computational Materials Science — •Gabriel M. Nascimento1, Flaviano José dos Santos2, Marnik Bercx2, Davide Grassano1, Giovanni Pizzi2, and Nicola Marzari1, 2 — 1École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland — 2Laboratory for Materials Simulations (LMS), Paul Scherrer Insititute, Villigen, Switzerland
Advances in simulation methods and workflow engines, and an ever-increasing computational power have enabled the paradigm change in computational materials science of high-throughput (HT) discovery. A significant challenge in these efforts is to automate the choice of parameters in simulation codes, where a consistent balance of accuracy and computational cost is required. We propose a rigorous methodology to assess the quality and sensitivity of self-consistent density-functional theory (DFT) calculations with respect to smearing and Brillouin zone sampling of 2D and 3D crystalline compounds. Our results enable the construction of unified protocols for the choice of parameters that guarantee k-point sampling convergence and control of entropic systematic errors for calculations within most classes of materials, while minimizing the required overall wall time. These protocols are integrated into automated workflows for DFT calculations using Quantum ESPRESSO, accessible through the open-source AiiDA framework. This work streamlines access to state-of-the-art computational frameworks, fostering HT research and supporting the development of computational materials databases.
Keywords: High-throughput workflows; Workflow engines; HT materials discovery