报告人：Fabrice Deluzet高级研究工程师

报告题目：Particle-in-Cell Methos with Embedded Sparse Grid Reconstructions

报告摘要：

Particle-in-Cell (PIC) methods are among the most widely used discretization schemes for plasma simulation. They combine a Lagrangian discretization of phase space—using numerical (macro)particles—with an Eulerian discretization of the field equations.

These methods have gained widespread popularity due to their ease of implementation, scalability with respect to problem dimensionality, and their ability to provide physical insight—thanks to the close correspondence between numerical and physical particles.

However, PIC methods suffer from a major limitation: they are prone to statistical noise, which arises from sampling the distribution function with a finite number of numerical particles. Moreover, this noise decays slowly, scaling as the inverse square root of the average number of particles per mesh cell. This makes three-dimensional simulations and mesh refinement computationally expensive.

In this lecture, I will introduce Sparse-PIC methods, in which the standard Cartesian grid used in conventional PIC methods is replaced by a sparse grid hierarchy composed of anisotropic subgrids. The goal is to reduce the inherent statistical noise in PIC methods while preserving accuracy relative to mesh resolution. I will also highlight the computational efficiency of Sparse-PIC methods across different computing architectures.

报告时间：2025.10.29上午10:00-12:00

报告地点：理学楼401室

报告人简介：Fabrice Deluzet，法国国家科学研究中心的高级研究工程师，从事等离子体物理的数值模拟研究二十余年，在相关领域发表学术论文50余篇，包括Mathematical Models & Methods in Applied Sciences、SIAM Journal on Scientific Computing、SIAM Multiscale Modeling and Simulation、Journal of Computational Physics等顶级期刊。