sfparticles 0.5.0

Simulation toolkit for particle tracking in strong fields.

Introduction

sfparticles is a simulation toolkit for particle tracking. Unklike particle-in-cell simulations, sfparticles does not solve field equations but instead tracks individual particles in pre-defined fields, without considering the inter-particle interactions. This allows for fast 3D simulations of large numbers of particles when plamsa denisty is low.

Features

• Numba Acceleration: sfparticles leverages Numba to accelerate simulations by compiling Python code to machine code.
• Cascade Simulation: Optimized for exponentially growing number of particles.
• Flexible Field Configuration: sfparticles allows you to define complex electromagnetic fields using python functions.
• Particle Tracking: Track the trajectories of particles. see example.

Installation

from github:

git clone https://github.com/xsgeng/sfparticles.git
cd sfparticles
pip install --user -e .

from pypi:

pip install sfparticles

Usage

Basic Simulation

To run a basic simulation, you can refer to the example scripts provided in the examples directory. For instance, to simulate a particle cascade, you can use the cascade.py script:

python examples/cascade.py
# set number of threads
NUMBA_NUM_THREADS=64 python examples/cascade.py

For trajectory simulations, you can use the trajectory.py script:

python examples/trajectory.py

Advanced Settings

You can set environment variables to configure the simulation. For example, to use an optical depth radiation model, set the SFPARTICLES_OPTICAL_DEPTH environment variable:

SFPARTICLES_OPTICAL_DEPTH=1 python examples/cascade.py

GPU Acceleration

For even faster simulations, sfparticles supports GPU acceleration. To enable this feature, you need to install CUDA and set SFPARTICLES_USE_GPU=1 environment variable.

Note: GPU is slow for small number of particles.

Installing CUDA

conda install -c conda-forge cupy cudatoolkit=11.2

Enabling GPU Acceleration

SFPARTICLES_USE_GPU=1 python examples/cascade.py

For Slurm job schedulers, you can use the following command:

SFPARTICLES_USE_GPU=1 srun -p gpu -G 1 -u python examples/cascade.py