N-body simulations in C++.

• NBodyRK4: Integrates a system of N gravitationally interacting bodies using the 4th-order Runge-Kutta method, with adaptive step sizes. This is a direct, particle-particle integration with O(N^2) complexity -- prohibitively slow for large N. (Stay tuned for faster integrators.)

• Input parameter file: params
• Output data file: results

$ ./src/integrate -p params -w results

(Note: standard output is switched off if writing to file)

• First row: Number of bodies, number of dimensions, integration duration, initial time step, desired accuracy
• Each remaining row: Mass of body, initial position in each dimension, initial speed in each dimension

An example Python script (requires matplotlib) plot_realtime_N3.py is included for viewing the orbits in real time. One could pipe the integration data directly from the simulator like so:

$ ./src/integrate -p params | ./plot_realtime.py [frameskip]

or read it in from a results file:

$ ./plot_realtime.py < results

N.B. This script includes a plethora of hard-coded parameters (for N=3), and may have to be modified to suit your particular system.

• Faster methods (tree code, multipole, mean field, etc.)
• Generalized interaction function