2018 PhD thesis in applied mathematics Docking Flexible Proteins using Polynomial Expansions, supervised by Valérie Perrier and Sergei Grudinin, Grenoble Alpes University, ljk, INRIA, Grenoble, France
2014 Master 2 Mathematics, Informatics and Applications Specialization: Modeling and scientific calculus, Joseph Fourier University, Grenoble, France
2025-now Research engineer Improving the performances of the SAMURAI software within the conctext of the NMPEX PEPR supervised by Loic Gouarin, CNRS/École Polytechnique, Paris Saclay, France
2022-2024 Post-doc position Studying the impact of numerical precision on Krylov methods supervised by Yves Durand, CEA-List, Grenoble, France, CEA, Grenoble, France
2020-2022 Post-doc position FWI uncertity quantification supervised by Ludovic Métivier and Romain Brossier, Grenoble Alpes University, ISTerre, UGA Grenoble, France
2018-2020 Post-doc position Developping new methods for the FWI problem supervised by Dimitri Komatitsch, Vadim Monteiller and Cédric Bellis, Aix Marseille Univesity, LMA, CNRS Marseille, France
I worked with Nano-D research team. Nano-D is a LJK and INRIA research team, which develops algorithms for Modelling and Simulation of Nanosystems. My doctoral thesis focused on algorithm designs for system modelling for Proteins and Macromolecular assemblies. My research consisted of two main parts.
FFT accelerated exhaustive search Our goal was to extend the FFT-based exhaustive search method, applied so far to 6D rigid bodies DOF, to the off-grid rigid and flexible DOF. We started by exhaustively sampling the rigid off-grid DOF and later implemented flexible global DOF, which were obtained using the NMA.
Large scale global molecular motions We proposed a method for nonlinear NMA called NOLB, which relies on the theoretical basis of the RTB method. Overall, our method produces better structures compared with the standard approach, especially at large deformation amplitudes. To the best of our knowledge, this was the first work on non-linear normal mode extrapolation in the Cartesian space. This work was applied to RMSD computation, flexible fitting and functional motions prediction.
Optmization techniques for FWI (2018-2020) I first worked on the optimization part of FWI within the LMA under the supervision of Dimitri Komatitsch, Vadim Monteiller and Cédric Bellis. We first proposed a robust FWI algorithm. The method produced excellent results on difficult synthetic problems with strong contrasts, even when low frequency data were not available. The standard method was unable to solve this problem du to cycle-skipping.
Uncertainty quantification for FWI (2020-2022) I then worked on the uncertainty quantification for the FWI within the ISTerre under the supervision of Ludovic Métivier and Romain Brossier. We worked on a time domain ensemble-FWI scheme that uses an ETKF to estimate the uncertainty of a FWI scheme. Our work was applied to a North Sea OBC data-set.
Extended precision for numerical methods (2022-now) I'm currently working on the impact of numerical precision on Krylov solvers CEA-List under the supervision of Yves Durand. We first showed how numerical precision can speedup the convergence of both BiCG and QMR solvers for the wave equation, especially when it is discretized with high-order SEM. We then showed, on a FWI inspired, how numerical precision could speed-up, and in some case enable the convergence of the block variant of the BiCG.
I taught the following courses at ENSIMAG in Grenoble:
I also taught the following courses at the bachelor of science and technology department (Département Licence Sciences et Technologies, DLST) of Grenoble Alpes University:
I taught Algorithmic and imperative programming (33h) during a Post-Doc at the bachelor of science and technology department (Département Licence Sciences et Technologies, DLST) of Grenoble Alpes University.