Random Numbers and their Effect on Particle Swarm Optimization

Particle Swarm Optimization is a relatively new evolutionary computation technique. It is based on the social behaviour of birds flocking or fish schooling. It was designed to find optimal regions in a search space, with the biological idea of swarms in mind. In the Particle Swarm Optimization algorithm, particles which ‘fly’ around the search space have velocities associated with them. Updating these particles velocities can be described in one succinct equation which essentially updates the velocities of the particles in the system. A large part of this formula is affected by random numbers, as is it is clear from the formula that depending on the size, the random number generated can have a big say in the change of the velocity. The aim of this paper is to explore the effects of using different random number generation techniques, and specifically if it has an impact on the performance of Particle Swarm Optimization on three common optimization problems.