Are small swarms effective in PSO optimization tasks?

This manuscript studies the optimal size of a swarm in PSO-type algorithms. As the popularity of PSO algorithms has increased and they have been applied to an ever-increasing number of applications, some key aspects appear to be less than fully explored. One of the questions is what the optimal size of a swarm is. This is expressed alternatively as a question of how small a swarm can be before it fails to properly search for a solution and conversely whether a very large swarm can better search for a solution and still make efficient use of the computing resources required. In the present manuscript a pilot study is presented, that investigates a standard PSO algorithm and attempts to address the issue of optimizing the swarm size. The results obtained are analysed via statistical tests to determine their significance.