To come.
Group of work in the continuity of the corresponding internship.
Reconstruction of sparse data with reduced basis methods using an ML autoencoder to reduce the online optimization procedure down to a few POD coefficients instead of all of them.
Implementation in MATLAB of an a posteriori error estimator for heterogeneous diffusion on TrioCFD. Enables local mesh refinement recovering optimal convergence for low-regularity solutions.
Proposed to use a DG variational formulation of Finite Volumes hoping to enable a posteriori error estimators for Reduced Basis generation defined in the FEM framework.
Modelling and statistical analysis of critical branching processes — Lévy flights, Brownian motion — and their clustering statistics in neutronic and avalanche cases.
Research project and lecture-format presentation with Rachelle RUELLE on approximation spaces for deep neural networks.
Research project and lecture-format presentation with Maxence MICHOT on general solutions of the Schrödinger equation.
Literature review of Sobolev spaces and representation theorems with Dimitry BERGEAULT. Written report and oral presentation.
With Junior ROGNON. Partial proof of GAN correctness via Lebesgue Measure Theory, applied to breast cancer data.
Numerical optimisation of rectangle packing in a constrained space, testing multiple methods for reliable results within short computation times.
Email: raphael.lecoq [at] ens-rennes.fr
Updated: April 2026
GitHub
LinkedIn