Application of GPU-accelerated FDTD method to electromagnetic wave propagation in plasma using MATLAB Parallel Processing Toolbox

TL;DR

This paper enhances MATLAB-based GPU-accelerated FDTD simulations for electromagnetic wave propagation in plasma, achieving 20x speedup by optimizing parallel computing techniques, thus enabling more efficient plasma modeling.

Contribution

It introduces specific techniques to improve MATLAB GPU parallel computing performance for FDTD, significantly accelerating electromagnetic simulations in plasma.

Findings

GPU-accelerated FDTD runs 20 times faster than serial code

Optimized MATLAB parallel techniques improve simulation speed

Effective modeling of plasma wave propagation using coupled FDTD and kinetic theory

Abstract

Since numerical computing with MATLAB offers a wide variety of advantages, such as easier developing and debugging of computational codes rather than lower-level languages, the popularity of this tool is significantly increased in the past decade. However, MATLAB is slower than other languages. Moreover, utilizing MATLAB parallel computing toolbox on the Graphics Processing Unit (GPU) face some limitations. The lack of attention to these limitations reduces the program execution speed. Even sometimes, parallel GPU codes are slower than serial. In this paper, some techniques in using MATLAB parallel computing toolbox are studied to improve the performance of solving complex electromagnetic problems by the Finite Difference Time Domain (FDTD) method. Implementing these techniques allows the GPU-Accelerated Parallel FDTD code to execute 20x faster than (basic) serial FDTD code. Eventually, GPU-Accelerated Parallel FDTD code is utilized to optimize the computational modeling of electromagnetic waves propagating in plasma. In this simulation, kinetic theory equations for plasma are used (excluding inelastic collisions), and temporal evolution is studied by the FDTD method (coupled FDTD with kinetic theory).