Regensburg 2019 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
MM: Fachverband Metall- und Materialphysik
MM 33: Topical session (Symposium MM): Big Data Analytics in Materials Science
MM 33.2: Vortrag
Donnerstag, 4. April 2019, 10:45–11:00, H43
Reproducible massive calculations and data sharing with AiiDA and the Materials Cloud — •Giovanni Pizzi1, Leopold Talirz1, Snehal Kumbhar1, Aliaksandr Yakutovich1, Elsa Passaro1, Marco Borelli1, Sebastiaan P. Huber1, Martin Uhrin1, Spyros Zoupanos1, Fernando Gargiulo1, Ole Schuett2, Joost VandeVondele3, Thomas C. Schulthess3, Berend Smit1, and Nicola Marzari1 — 1NCCR MARVEL and EPFL, CH — 2Empa, Switzerland — 3CSCS and ETHZ, CH
We discuss the challenges and solutions to store data resulting from the modern, complex workflows of computational science, allowing for the search and dissemination of results according to the FAIR principles of sharing. We first show how a materials’ informatics framework like AiiDA [1] allows to automate all calculations and store their entire provenance. By uploading all data to the Materials Cloud (materialscloud.org), results can be disseminated seamlessly, DOIs are assigned to the datasets, and interactive (online or local) browsing of the provenance makes it possible to explore any element of the workflow guaranteeing its full reproducibility and enabling reuse of the results. Materials Cloud also provides intuitive web-based simulation services based on AiiDA, reducing the access barrier to HPC simulation tools.
[1] G. Pizzi et al., Comp. Mat. Sci. 111, 218 (2016), www.aiida.net