Generating the index files is a bit slow. Here I am implementing a parallel version for updating the particles (finding new / removed particles is still serial).