CEA IRESNE (CEA-Cadarache)
FrancePublication : 16 oct. 2026
LM2C (Laboratory for Fuel Behavior Modeling) - Atomic descriptors for AI-driven modeling of nuclear fuels
- Stage
- 4-6 mois
- Présentiel
- Rémunéré
- Échéance : 30 oct. 2026
- Machine Learning
- Artificial Intelligence/Data
- molecular dynamics
- Descriptors
- Simulation Python
- Materials science / Materials engineering
- generative models
Description du poste
Goal: Extend the PULSE generative AI method to describe atomic displacements in MOX and U‑Mo alloys using atomic descriptors, to better predict thermodynamic properties. Hosted at CEA Cadarache, IRESNE, LM2C.
Tasks:
- Generate atomic configurations via molecular dynamics at several temperatures; analyze displacements around equilibrium positions.
- Compare existing atomic descriptors (sensitivity to small displacements; reconstructability of positions).
- Propose and test ways to integrate descriptors into PULSE on model systems.
What you’ll learn: atomistic simulation and machine learning for materials; statistical and nuclear materials physics; scientific programming in Python; research methods.
Methods/tools: Molecular dynamics, variational autoencoders, atomic descriptors, PULSE, LAMMPS.
Keywords: artificial intelligence, nuclear fuels, computational materials, materials physics.
Work mode/location: Onsite at CEA Cadarache (IRESNE/LM2C). Thesis possible afterward (Yes).