Pygmo and Pykep: Open Source Tools for Massively Parallel Optimization in Astrodynamics (the Case of Interplanetary Trajectory Optimization)

At the intersection between computer science and astrodynamics lies a fertile ground for improving methodologies and performances of spacecraft trajectory computations. In this paper we present two open source projects (written in C++ and exposed to Python) that are focussed around computational efficiency and that allow to script massively parallel optimization of aerospace related problems. In particular, we will show their use for interplanetary trajectory optimization. After having described novel findings and technologies powering these two projects, we will show some use examples. We show results on asteroid selection for human mission to asteroids, on the Global Trajectory Optimization Competitions and, finally, on a novel idea to obtain on-line adaptive mesh during the direct optimization of interplanetary low-thrust problems.