Deterministic Discrete Particle Swarm Optimizers with Collision and Insensitivity

This paper studies a novel particle swarm optimizer for multi-solution problems. The system is governed by a deterministic difference equation and has no stochastic parameter: it is suitable for analysis of dynamics. The particles move on discrete search space, can be insensitive and can collide with other particles. The collision and insensitivity can be effective to make a swarm diversity and to approach solutions. The algorithm efficiency is investigated in basic numerical experiments.