PySwarms: a research toolkit for Particle Swarm Optimization in Python

Abstract

Particle swarm optimization (PSO) is a heuristic search technique that iteratively im-proves a set of candidate solutions given an objective measure of fitness (Kennedy and Eberhart 1995b). Although vanilla implementations of PSO can be found in some Python evolutionary algorithm toolboxes (Fortin et al. 2012; Biscani, Izzo, and Märtens 2017), a PSO-specific library that focuses on the said technique is still an open challenge. PySwarms is a research toolkit for Particle Swarm Optimization (PSO) that provides a set of class primitives useful for solving continuous and combinatorial optimization problems. It follows a black-box approach, solving optimization tasks with few lines of code, yet allows a white-box framework with a consistent API for rapid prototyping of non-standard swarm models. In addition, benchmark objective functions and parameter-search tools are included to evaluate and improve swarm performance. It is intended for swarm intelligence researchers, practitioners, and students who would like a high-level declarative interface for implementing PSO in their problems.