Thierry Deutsch is director of the MEM (Modelling and Exploration of Matter) laboratory at CEA-Grenoble, specialising in the simulation and modelling of materials at the atomic scale.
materials at the atomic scale.
He developed the BigDFT code, based on density functional theory, in collaboration with Luigi Genovese. This software uses wavelets to calculate the electronic structure of materials and molecules by based on density functional theory. He has taken part in more than seven European projects, mainly in the field of HPC and massively parallel and GPU computing.
He is involved in two major French national projects, the PEPR DIADEM project to accelerate the discovery of materials, and the PEPR NUMPEX project to build the software stack for exaflop-scale computers.
DIADEM (Material Acceleration Platforms) and NUMPEX (HPC) projects: Illustrating the application of BigDFT to proteins
In this talk, I will explain the objectives of the French national DIADEM project, which aims to accelerate the discovery of materials using artificial intelligence, with the implementation of a database and workflow infrastructure. I will then focus on the French national NUMPEX project, which aims to provide the software stack for the next French exascale computer.
Finally, I will illustrate the contribution of these projects on a concrete case of DFT calculation workflows for proteins containing several tens of thousands of atoms, with a reduction in complexity to provide quantities of interest to biologists.
References
[1] Saber, M., Kotan, H., Koch, C. C. & Scattergood, R. O. A predictive model for thermodynamic stability of grain size in nanocrystalline ternary alloys. Journal of Applied Physics 114, 103510 (2013).
