Missions

This project aims to gain a better understanding of atomic descriptors and cost functions, and their role in the fitting of machine-learned interatomic potentials (MLIPs). Focusing on linear models, we will explore various sets of descriptors from the literature and evaluate their performance in materials and molecular systems. Key evaluation criteria will include numerical cost, stability during molecular dynamics and robustness. Different cost functions will also be tested. A central aspect of the project will be a mathematical analysis aimed at identifying the relationship between descriptors and the fitting of complex quantities calculated from MLIPs, possibly directly from electronic structure models. By improving the explicability of atomic descriptors, we aim to improve the accuracy and efficiency of linear MLIPs. We also seek to theoretically understand the role of the cost function in the fitting procedure.

Activités

- DFT-based database construction
- MLIP construction
- MLIP validation
- Mathematical analysis of results

Compétences

The ideal candidate will hold a PhD in mathematics, physics, chemistry or a related field. We are particularly interested in candidates with experience in one or both of the following areas:
- Numerical analysis / mathematical analysis / modeling for materials science,
- Machine learning force fields / data-driven techniques for materials science.

Contexte de travail

The position is located in an area covered by the protection of scientific and technical potential (PPST) and therefore requires, in accordance with regulations, that your arrival be authorized by the competent MESR authority.”

Le poste se situe dans un secteur relevant de la protection du potentiel scientifique et technique (PPST), et nécessite donc, conformément à la réglementation, que votre arrivée soit autorisée par l'autorité compétente du MESR.

Contraintes et risques

Nothing to declaire