Manifestations d'intention > Projet ENIGME-équipement

Objective: Development and powering of a high-performance computing (HPC) cluster for ab initio calculations.

Context: The main objective of the ENIGME project is to get insights, thanks to a strong synergy between ab initio calculations and various experimental techniques, about the mechanisms (today still unknown) of the first colour photochromatic images of Edmond Becquerel. This fundamental understanding is crucial for the conservation itself of these clichés and represents one of the main goals of the Centre de la Recherche sur la Conservation (CRC) of the Muséum National d’Histoire
Naturelle, which is a partner of the ENIGME project.

ENIGME intervenes in two thematic axes of the DIM-MAP: i) Matériaux d’artistes et matérialités des œuvres; ii) Signal, calcul et statistiques.

Preliminary theoretical and numerical results, compared with the experimental results conducted by the CRC, have permitted to validate the level of theory and approximations required to tackle the systems under study, notably AgCl and related systems (crystal deformation of the AgCl, nanocluster of Silver, etc.). Not too astonishingly, the level of theory required is cutting edge: state-of-the-art theoretical approaches, like many-body theory in the domain of Green’s functions, are indeed required to describe the optical properties, absorption spectra and excitonic characteristic that are inherently related to the
definition of colors in materials.

This imply: i) further theoretical+numerical developments in our approaches and codes, for faster and more effective application of the theory; ii) cumbersome calculations on rather big systems (the description of crystal deformations requires the use of supercell approaches, the nanocluster contains hundreds of atoms, etc.).

Plan: for all these reasons we have decided to invest in developing a HPC cluster dedicated to our ENIGME project. We plan to ask help to acquire 4 nodes to dedicate to code developments and calculations involved in the ENIGME project. It is important to underline that this is completely independent and parallel to the annual resources that our group obtains in national and European computing centres. The importance of computational resources devoted to ENIGME is due instead to needs that would be in total contrast to typical rules in computing centres like GENCI or PRACE:

• possibility to send multiple jobs with testing and developing purposes (sometimes not producing results)
• possibility of long run (at GENCI the maximum allowed time is one day)
• specially dedicated high-memory nodes (we are aiming at 1-Tb ram nodes, while the maximum present at GENCI is 128-Gb ram nodes) for excitonic calculations.

Partners: we are open to partners who would like to extend their computing capabilities, especially for ab initio calculations, implying memory consuming tasks.