Particle-in-Fourier plasma simulation

This is a proof of concept for efficient Fourier-space particle simultion. For more information about the algorithm, see https://arxiv.org/abs/1808.03742. Note that, although this repository and several of the code samples are titled "FT-PIC", the name of the algorithm has been changed to "particle-in-Fourier" (PIF).

This code is compiled using the USFFT library, which is not open source. In order to build it, you'll need a copy of the compiled USFFT library, or you'll need to rewrite the code to use another library that provides support for fast nonuniform FFTs (which we might do in the future).

You'll also need the QDSP library to view phase plots, which can be obtained from https://github.com/matt2718/qdsp. A dummy QDSP header is provided in the include directory for building on headless systems.

Two source files are provided: ftpic.c contains our fast gridless algorithm, while oldpic.c contains a reference 1D PIC code with first-order deposition and interpolation for the purposes of comparison with PIF. They use the same options, which you can view with the -h option.

Analysis scripts

• eplot.py displays a plot of kinetic, field, and total energy over time based on the output file of a single simulation.

• modeplot.py displays a logarithmic plot of the growth or damping of one or more modes from one or more output files.

• ecomp.py plots the total energies over time for several simulations.

• npng.sh runs a scaling study varying the number of particles and the number of grid cells/modes.

• threadscale.sh runs a scaling study varying the number of threads.

• dtscale.sh and dtscale2d.sh run convergence studies by varying the timestep size, for 1D and 2D programs respectively. The same scripts can be used to analyze the output of convergence studies with the -a flag.

• getmax.py prints the maximum difference in total between two timesteps for some output file.

• convplot.py displays a convergence plot based on the output of dtscale.sh