pygmo2

Why we included this project

pygmo gives you one consistent Python API for a large collection of optimization solvers and problem definitions, so you do not have to glue together separate algorithm libraries by hand. It ships with dozens of ready-to-use methods, including differential evolution, particle swarm, CMA-ES, simulated annealing, NSGA-II, MOEA/D, and local solvers wrapped from NLopt, SciPy, and Ipopt, plus standard test problems such as Rosenbrock, Rastrigin, and the ZDT/DTLZ suites. The generalized island model is the part that stands out: you can run many optimization instances in parallel across cores or cluster nodes and let them exchange information as they go, which helps on expensive or stubborn problems. Metaheuristics researchers, people benchmarking new algorithms against established ones, and engineers with real optimization tasks will appreciate not having to build the parallel plumbing themselves.