Skip to article frontmatterSkip to article content

Accelerating Plasma Physics Simulations with Pyccel: A PyGyro Case Study.

Authors
Affiliations
The UM6P Vanguard Center, Morocco
EPFL Switzerland
Max-Planck-IPP Germany

This demonstration presents Pyccel, a Python-to-C/Fortran translator designed to optimize performance-critical bottlenecks in scientific workflows. Pyccel transforms high-level Python code, including object-oriented structures, into efficient compiled routines, enabling significant performance improvements while preserving code readability and maintainability. PyGyro, a modular library for computational plasma physics, will be used as an example to illustrate Pyccel’s ability to accelerate numerical methods such as splines, advection operators, and solvers. We will demonstrate how to apply Pyccel to accelerate specific routines and seamlessly integrate the optimized code into existing workflows.

Repository

https://github.com/pyccel/pyccel

References
  1. Bourne, E., Güçlü, Y., Hadjout, S., & Ratnani, A. (2023). Pyccel: a Python-to-X transpiler for scientific high-performance computing. Journal of Open Source Software, 8(83), 4991. 10.21105/joss.04991
  2. Bourne, E., & Güçlü, Y. (2024). Highly parallel drift-kinetic semi-Lagrangian simulations in Python. ESAIM: Proceedings and Surveys, 77, 176–194. 10.1051/proc/202477176