Fédération de recherche PLAS@PAR

Workshop on Computational Plasma Physics

The workshop is organized by PLAS@PAR and will take place on December 12 at Sorbonne Université, in the Charpak Amphitheater on the Pierre and Marie Curie Campus, (Entrance at the base of Tower 22 - Level SB) 4 Place Jussieu, 75005, Paris. Metro Jussieu.

This one-day workshop will bring together researchers to exchange ideas and share recent advances in computational approaches to plasma physics. 

Publié le 6/11/2025 - Mis à jour le 19/11/2025

Rationale. Numerical modeling is central to advancing our understanding of plasmas across laboratory, space, and astrophysical environments. The growing diversity of algorithms and computing architectures calls for a new generation of hybrid approaches, both in term of models and hardware. Examples are coupling kinetic and fluid descriptions, integrating quantum and classical computation, and leveraging data-driven methods such as machine learning. This workshop, organized by Research Federation Plasmas à Paris - PLAS@PAR, brings experts focusing on computational physics, artificial intelligence, and high-performance computing to discuss novel strategies for multi-scale and multi-physics modeling of plasmas in the hybrid computing era.

There are no fees for participating, but registration using this link is mandatory.

Time Title Abstract Speaker
09:00 welcome coffee    
09:20 Overview of HPC ⓘ A broad tour of the major evolutions in high-performance computing: heterogeneous architectures at exascale, performance portability, and the convergence/divergence of HPC, AI, and quantum. The talk highlights challenges and opportunities for scientific applications and large-scale codes. Pascal Tremblin, CEA, Director Maison de la Simulation
10:00 Plasma Physics with Smilei: from physical models to high performance computing ⓘ Smilei is an open-source Particle-in-Cell (PIC) simulation code, designed with versatility, high performance and user-friendliness in mind, combining advanced physical modeling and efficiency for many applications in plasma physics, including high intensity laser-matter interaction, astrophysics, fusion and plasma acceleration. This talk will give an overview of its main physical modules, as well as the code’s parallelization strategy on CPU and recent developments in its porting for heterogeneous computing architectures, aiming to ensure efficient performance on modern high-performance computing machines. Francesco Massimo, LPGP
10:40 PHARE: Simulating Plasmas with Adaptive Mesh and Model Refinement ⓘ PHARE is a new open-source simulation framework designed to model multi-scale magnetized plasma environments and processes. This presentation introduces the code’s capabilities, including its ability to solve Hybrid PIC, hyper-resistive Hall-MHD, and resistive MHD equations on a hierarchy of nested Cartesian grids that dynamically adapt to evolving regions of interest. An overview of the project’s roadmap will also be discussed. Nicolas Aunai, LPP
11:20 Entity: An open-source, coordinate-agnostic PIC code for high-energy astrophysics ⓘ Particle-in-cell (PIC) simulations have long been key to modeling plasma processes in high-energy astrophysics. I will present Entity, a recent open-source, coordinate-agnostic particle-in-cell (PIC) code specifically targeted at studying plasma physics in relativistic astrophysical systems, allowing for arbitrary grid geometries. The code is highly modular and written in an architecture-agnostic manner, utilizing the Kokkos performance portability library to leverage device parallelization on various CPU and GPU architectures efficiently. The code incorporates QED processes, turbulence schemes, and generalized higher-order schemes to improve accuracy and numerical stability in extreme regimes. (https://entity-toolkit.github.io/wiki/) Arno Vanthieghem, LUX
12:00 Lunch    
14:00 An informal introduction to future quantum computing for plasma physics.   Fabrice Debbasch, LUX
14:40 Quantum Molecular Dynamics ⓘ The warm dense matter (WDM) regime spans a broad range of temperature–density conditions where thermal, collective, and quantum electronic effects all play pivotal roles. Such conditions arise in diverse environments—including planetary and stellar interiors as well as inertial confinement fusion (ICF) experiments—making a thorough understanding of WDM essential yet demanding both theoretically and experimentally. Quantum molecular dynamics (QMD) simulations grounded in density functional theory (DFT) now furnish accurate descriptions of these complex states, delivering a rich set of observables such as equations of state, phase diagrams, transport coefficients, and optical properties. Recent methodological advances have pushed the frontier further, enabling investigations of ever more extreme regimes: higher temperatures, lower densities, and larger simulation cells. These gains stem from novel DFT frameworks and the deployment of hybrid CPU–GPU codes that exploit modern high performance computing architectures. In this talk we will review the fundamentals of QMD and illustrate how these latest developments open pathways to explore previously inaccessible physics within the warm dense matter domain. Francois Soubiran, CEA DAM-DIF
15:20 High-order moment models and asymptotic regimes of the kinetic equation of low-temperature plasmas of noble gases ⓘ Low-temperature plasmas often present non-equilibrium distribution functions due to the collisions with the background gas and the presence of strong electric fields. This non-equilibrium is beyond classical fluid models, often requiring computationally-intensive kinetic simulations. In our work, we study high-order moment models in order to capture the non-equilibrium state with a macroscopic set of equations, which is more computationally efficient than kinetic simulations. Depending on the pressure regime and the electric field, the kinetic equation of the charged species present different asymptotic regimes that can help to simplify and develop high-order moment hierarchies. In this work, we compare numerical simulations of different moment closures: Grad's closure, the hyperbolic quadrature method of moments, the extended quadrature method of moments, and a method based on entropy maximization. We assess the different closures for plasma applications and propose efficient numerical discretizations. The numerical solution of the high-order moment models is compared to kinetic simulations of an argon plasma at different pressure regimes, from nearly collisionless to collisionally-dominated. In general, all the high-order moment closures capture the transport with high fidelity as compared to the kinetic simulations, providing an improvement as compared to classical fluid models. Classical fluid closures such as the Fourier law for the heat flux is shown to be not suitable to capture the sheath or the low pressure regime. Alejandro Alvarez, LPP
16:00 coffee    
16:30 Interpretable Machine Learning for Physics-Guided Discovery and Modeling of Flow Dynamics ⓘ Interpretable machine learning offers new avenues for uncovering physical mechanisms and building models directly from data. In this talk, I will present recent advances in unsupervised learning for discovering fluid flow patterns and in symbolic regression for deriving interpretable closure equations. These approaches open new paths toward physics-guided, data-driven modeling of complex flow dynamics. Paola Cinnella, IJRA
17:10 Multi-phase plasma model   Fabien Tholin, ONERA
17:50 closing