Toward Robust Materials Simulations with Common Workflow Interfaces in AiiDA
DescriptionIn recent years, "materials by design" has become a very powerful approach, requiring to run large numbers of simulations on supercomputers to build databases of computed properties. I will describe the resulting challenges and how we address them in our Open Digital Infrastructure, based on AiiDA (https://aiida.net, to orchestrate simulation workflows and automatically track their full provenance) and Materials Cloud (https://materialscloud.org, to enable sharing of reproducible research data and run simulations on the cloud). I will show concrete examples of our materials databases and how they enable the discovery of novel materials and their properties. I will then discuss the development of common workflow interfaces providing identical input/output interface across different quantum simulation codes toward accessible, interoperable and robust simulations. These enabled our recent collaborative effort to test precision and transferability of DFT methods across chemistries (all elements from hydrogen to curium, in 10 prototypical crystal structures), resulting in a reference dataset of 960 equations of state cross-checked between two all-electron codes, then used to verify and improve multiple pseudopotential methods. I will conclude mentioning our plans to extend the common interfaces to more properties (bands, phonons).
 S. P. Huber et al., npj Comput. Mater. 7, 136 (2021).
TimeTuesday, June 2711:30 - 12:00 CEST