Swarm-NG package helps scientists and engineers harness the power of GPUs for solving computationally intensive problems in dynamics. In the early releases, Swarm-NG will focus on integration of ensembles of N-body systems evolving under Newtonian gravity.
Swarm-NG does not replicate existing libraries that calculate forces for large-N systems on GPUs, but rather focuses on highly optimized integrators for simultaneous integration of ensembles of many (10^3-10^5) systems of small N (2-10 bodies). This is of particular interest for astronomers who study the chaotic evolution of planetary systems, asteroids, producing orbit libraries, etc.
The preview release (v0.1) concentrates on showcasing the overall ensemble-integrator architecture and benchmarking GPU acceleration with commonly used integration algorithms (hermite, verlet, rk4, adaptive timestep hermite). In upcoming releases we plan to provide more complex integrator kernels (mixed-variable symplectic and Bulirsch-Stoer).
The code, extended description, and instructions on how to build it can be found here.
-- Aaron Boley, Eric Ford, Ameya Gancchha, Jianwei Gao, Mario Juric and Young In Yeo.