Uploaded June 2011 | Updated November 2026, 3 hours ago
1M bodies under gravitational self-attraction in two dimensions (with 3-dimensional kernel---sneaky!) 8000 time steps ran in 34 hours on 1 core of my 3.6 GHz Phenom II. The method uses linked lists for both particles and boxes (totally inefficient by today's standards). Euler time-stepping, Barnes-Hut-style quadtree-code, bucket size was 20, Krasny's delta-smoothing parameter was 0.01, and opening criterion was theta=2.0. The code is entirely mine, was written in 2000-2006, and is available here: markjstock.org/part-nd
1M bodies under gravitational self-attraction in two dimensions (with 3-dimensional kernel---sneaky!) 8000 time steps ran in 34 hours on 1 core of my 3.6 GHz Phenom II. The method uses linked lists for both particles and boxes (totally inefficient by today's standards). Euler time-stepping, Barnes-Hut-style quadtree-code, bucket size was 20, Krasny's delta-smoothing parameter was 0.01, and opening criterion was theta=2.0. The code is entirely mine, was written in 2000-2006, and is available here: markjstock.org/part-nd










