Parallel Barnes-Hut Algorithm
Program for simulating N-body systems. The algorithm is a parallel implemenation of the Barnes-Hut algorithm inspired by Salmon, John K. Parallel hierarchical N-body methods. Diss. California Institute of Technology, 1991. The program can run on 2d process using MPI.
A simulation of two spinning disks and a visualization of the Barnes-Hut tree.
Install a MPI library such as OpenMPI or MPICH. On Mac OS X this can be done through Homebrew:
$ brew install mpich
Download the git repository and build the program in a some directory, e.g. build
$ git clone https://github.com/pbarkm/n-body.git
$ cd n-body
$ mkdir build
$ cd build
$ cmake ..
$ make
To check if the program runs (on 2 processes):
$ mpirun -np 2 ./n_body
The supplied python files for plotting require matplotlib and numpy.
To display the command line options run
$ mpirun -np 1 ./n_body -?
The following command runs a simulation of 1000 bodies for 10000 time steps. The sampled positions of the bodies are stored in positions.txt and the program is run with the verbose flag (-v).
$ mpirun -np 2 ./n_body -N 1000 -s 10000 -v -o positions.txt
A python script is supplied to animate the simulation.
$ python py/plot.py positions.txt