The key idea of this project is that maybe what we call "magnetic fields" are just the 
result of speed-of-light delayed moving electric fields.  With modern GPUs I think we 
can simulate things well enough to test this idea.  We are not and will not use any direct
"magnetic" concepts in the simulation.  If we can get "electro magnetic waves" without
using magnetism, then "electon waves", the name of this project, seems like a good name.

I have collected some chatbot opinions of this idea in the file chatbot.opinions.  

Only when calculating forces between electrons do we have an N-squared problem.  This
is the most performance critical part.  This part is done in CUDA with the code in kernel.cu.   

Much of the rest of the computaion is done using CuPy vectors in Python in electronwaves.py.  
Some output/visualization code is done with Python/Numpy vectors so they can run 
on CPU cores in the background.

Several bash scripts to make video from output images etc.

README.vast explains how to use vast.ai to run this on rented GPUs.