Asynchronous and Synchronous–Asynchronous Particle Swarms

Particle swarm optimization is a valuable concept for the approximate solution of global optimization problems. It belongs to the class of metaheuristics and has its origin in the simulation of bird flocks. Particle swarm optimization is applicable in many ways and has an impact on a variety of research areas, including multiobjective optimization and machine learning. In this chapter, we introduce an asynchronous and synchronous–asynchronous framework for particle swarms. We compare those concepts to judge performance and adaptability, e.g., given different kinds of position and velocity limitations, adjustable update parameters, and swarm properties, as well as several possible end conditions. Our benchmark results show good agreement compared to existing implementations and demonstrate that the introduced framework can lead to significant performance improvements with respect to execution time and the number of iterations.