Partenaires

CNRS
Logo tutelle
Logo tutelle Logo tutelle


Rechercher

Sur ce site

Sur le Web du CNRS


Accueil du site > Jobs - PostDoc - These > PhD position at Université de Lorraine on machine learning for materials science

PhD position at Université de Lorraine on machine learning for materials science

par Nord & Ile de France - 18 juin 2020

Applications are invited for a PhD position starting in October 2020 at the LPCT laboratory of the Université de Lorraine. The proposed PhD topic is summarized below. Candidates should have a strong background in theoretical physics or quantum chemistry and possibly experience with numerical methods and programming. Application materials should include a cover letter and a cv with names of at least two references. For questions or to apply contact Dario Rocca at dario.rocca@univ-lorraine.fr.


Chemically accurate simulations by machine learning correlated approximations. Supervisor : D. Rocca / LPCT In ab initio molecular dynamics (MD) calculations, a finite-temperature simulated experiment is performed by computing quantum-mechanical forces from density functional theory (DFT) approximations. However, DFT does not systematically reach chemical accuracy. More sophisticated correlated methods overcome some of the limitations of DFT but are too expensive to be directly applied in MD. Within this PhD project the following methodology will be developed : Starting from the configurations generated by a numerically cheap MD, a machine learning model will be trained on a small number of expensive correlated calculations ; the predictions of this model will then be coupled with thermodynamic perturbation theory to compute highly accurate ensemble averages. Practical applications of this method will include calculations of enthalpies of adsorption of molecules in zeolites.