MaX offers a materials informatics ecosystem for automated high-throughput calculations and the automatic storage of data in graph databases.
The combination of AiiDA and Materials Cloud enables to:
- implement, run and share workflows and turnkey solutions, as well as the resulting raw and curated data;
- guarantees complete automatic provenance tracking, storage and preservation;
- ensures reproducibility of computational research, its reuse, as well as data analytics.
These are joint efforts performed in close collaboration with MARVEL, a long-term Swiss effort coordinated by EPFL, and supported by a number of other H2020 projects: NFFA, the EMMC, and the MarketPlace.
The high-throughput environment: AiiDA
Thanks to advances in simulation codes and the exponential increase of computational power (Moore’s law), researchers can now run thousands of simulations simultaneously to understand, predict or design materials properties.
However, managing these high-throughput simulation projects and the resulting data is not trivial. An ideal tool would ease data management, storage and retrieval, help ensure that simulations are reproducible and encourage the sharing and cross-validation of results by the larger research community.
All these ideas have been integrated into AiiDA, an automated interactive infrastructure and database built on the four “ADES” pillars of automation, data, environment and sharing (Pizzi et al, 2016).
AiiDA can manage and automate the simulations both on local computer resources and on big supercomputers. At the same time, it stores automatically input and output data and, most importantly, the whole provenance (i.e., the connection between data and calculations, with a full description of how data was generated) ensuring full reproducibility.
Moreover, it provides a powerful workflow engine that makes it possible not only to reproduce calculations, but also to run the same type of simulations on new materials or with different parameters. This is an enabler technology for the implementation of turn-key solution workflows, where materials properties can be computed with minimal inputs also by people that are not expert of the simulation codes. These workflows encode the knowledge and expertise of computational scientists and can perform automatically the choice of parameters to obtain a converged result.
Finally, AiiDA simplifies the task of sharing the results between researchers, as well as of disseminating research results thanks to its coupling with the Materials Cloud.
Education, dissemination and tutorials
Large efforts are invested in education and teaching of the tools and platforms, with 3-4 events per year, focusing on different user categories: from new users to experienced ones, up to focused coding weeks for developers.
The full list of past and current events can be found here, and many of the tutorials were co-sponsored by MaX. You can check for instance the reports from the June 2016 tutorial and the May 2017 tutorial, or the video report from the second AiiDA coding week in October 2017. As shown also in the reports, the feedback of these events from participants was always very positive, indicating that they were effective and useful, and suggesting to continue organizing similar tutorials in the future.
Info or subscribe to our mailing list