Thibault Cimic

Thibault Cimic

Research engineer

Sorbonne Université

INRIA

Biography

My name is Thibault Cimic, I’m a research engineer in applied mathematics specialized in linear algebra and PDE solvers. My job consists of turning mathematical ideas into efficient computational tools.

Current position

Since May 2025, I am on a 2-year contract at the CEA, jointly financed by two laboratories: L_Sim in Grenoble and the Maison de la Simulation in Saclay. This dual affiliation defines two complementary lines of work. On the L_Sim side, I am developing a new Kokkos backend for the psolver module of the BigDFT suite, a quantum chemistry code based on Density Functional Theory. On the Maison de la Simulation side, I contribute to the Kokkos ecosystem with new features that benefit both BigDFT and the broader community. This includes advancing the use of Kokkos within Fortran codebases and environments, and developing new features for KokkosFFT such as callbacks inside FFT computations to enable data manipulation mid-computation.

Previous positions

Before joining the CEA in Grenoble, I was a HPC research engineer at CEA/Saclay (February 2023 – August 2024) at Laboratoire Génie Logiciel, where I parallelized Ondo, a high-order spectral finite element solver for wave propagation modelling, distributing it across distributed-memory architectures.

Before that, I spent three years in the ALPINES team at INRIA Paris, a joint team with the Laboratoire Jacques Louis Lions (LJLL) at Sorbonne Université. There I developed 2-level preconditioning techniques for the map-making problem in cosmology (cosmic microwave background). More details on the dedicated page: B3DCMB.

Teaching

From 2018 to 2020, alongside my research at INRIA, I taught applied mathematics at Sorbonne Université: the finite element method in Python, ordinary differential equations in Python, and linear equation solvers in C++. Those courses are still available in the Teaching section of this site.

More recently I put together a Git tutorial aimed at researchers and students who want to get started with version control in a scientific workflow.

Useful resources

If you are interested in GPU programming with CUDA and Kokkos, have a look at Rémi Bourgeois’ blog.

InfoMath is a seminar series on digital tools for mathematicians (Git, formal mathematics with Lean, scientific computing, and more), hosted by Sorbonne Université and Paris-Dauphine. A great resource if you want to get up to speed on modern computational workflows in a mathematical context.

Download my resumé.

Interests
  • Linear algebra, linear solvers, eigen problem
  • PDE solvers
  • Preconditioning techniques
  • Navier-Stokes equations
  • Maxwell equations
Education
  • Research master in applied mathematics, 2016-2018

    Sorbonne Université Sciences - Campus Pierre et Marie Curie

  • License in applied mathematics, 2014-2016

    Sorbonne Université Sciences - Campus Pierre et Marie Curie

Skills

Cpp_Programming_Language
C++

70%

C_Programming_Language
C

70%

Open_MPI_logo
MPI

80%

Open_MP_logo
OpenMP

60%

python2
Python

80%

LaTeX_logo
LaTeX

80%

bash_logo
Bash

80%

Git_icon
Git

70%

LinaroDDT
DDT

70%

LinaroMAP
Map

70%

ScoreP
ScoreP

25%

Fortran

70%

Experience

 
 
 
 
 
CEA
Research Engineer
May 2025 – Present Grenoble
Porting psolver module of the BigDFT suite to the performance portability programming ecosytem Kokkos.
 
 
 
 
 
CEA
Research Engineer
Feb 2023 – Aug 2024 Saclay
Parallelization of Ondo a high order spectral finite element solver for NDT in the CIVA commercial software.
 
 
 
 
 
INRIA
Doctoral researcher
Jan 2018 – Sep 2021 Paris
  • Construction of a 2-lvl preconditionner for the map-making problem in cosmology.
  • Accelerate solvers of linear systems with ECG
 
 
 
 
 
Sorbonne Université
Professor of Applied mathematics
Sorbonne Université
Jan 2018 – Dec 2020 Paris
Teaching Finite Element Method in Python, ODE in Python, Linear Equation in C++

Contact