Stage Data / Mathématiques Appliquées Hauts-de-Seine entre mars et juin 2022 5 mois
IFP Energies nouvelles (IFPEN) est un acteur majeur de la recherche et de la formation dans les domaines de l’énergie, du transport et de l’environnement. De la recherche à l’industrie, l’innovation technologique est au cœur de son action, articulée autour de quatre priorités stratégiques : Mobilité Durable, Energies Nouvelles, Climat / Environnement / Economie circulaire et Hydrocarbures Responsables.
Dans le cadre de la mission d’intérêt général confiée par les pouvoirs publics, IFPEN concentre ses efforts sur :
Partie intégrante d’IFPEN, l’école d’ingénieurs IFP School prépare les générations futures à relever ces défis.
Development of efficient catalysts to enhance chemical reactions occurring in catalysis, water treatment and battery processes is a major research area at IFP Energies Nouvelles. To this end, numerical simulations of chemical processes at the atomic level are essential for a thorough understanding of the physical phenomena involved.
Since chemical reactions involve the formation and breaking of electronic bonds, electronic structures must be precisely modeled using the time independent Schrödinger equation. Approximation methods such as density functional theory (DFT) have been developed in order to solve this equation and compute the energy of a given molecular structure and the associated forces acting on each atom. Unfortunately, DFT based methods are computationally demanding and limit the atomic configurations under study to a small number of atoms for very short time scales below those required to describe a chemical reaction.
In order to alleviate these concerns, the use of Deep Learning has been extensively studied in the recent literature. The idea is to train models once based on accurate electronic structure calculations such as DFT, and then use these models in large simulations to directly predict the energy and forces for a given atomic configuration at a fraction of the DFT cost. This approach uses domain specific representations of atomic configurations in combination with state-of-the-art deep learning advances such as convolutional graph neural networks to achieve impressive results for a large range of applications.
The objective of this internship is to leverage the available packages allowing deep learning for electronic structure calculations  in order to study hydratation of γ-alumina surfaces, γ-alumina being one of the most used catalytic support. To do so, we will use previously existing DFT data obtained at IFPEN on γ-alumina surface models by using the periodic DFT code, VASP [3, 4]. The intern will be supervised by researchers specialized in ab-initio calculations as well as in deep learning.
The applicant should be pursuing a Master of Science or equivalent in data science, computer science, engineering or a related field.
IFP Energies nouvelles - Sciences et Technologies du Numérique