Further improvement of the Glowworm Swarm Optimization algorithm by adding a conservation of agent move direction

Glowworm Swarm Optimization is an algorithm, which can localize multiple optima of the multi-modal function during a single run. Unfortunately, original algorithm manifests several disadvantages: an agent can change its position only in the presence of other agents and exclusively in direction of randomly chosen neighboring agent. This paper shows how GSO algorithm can be significantly improved by simple modifications: agents receive an alternative method of changing the position by random jumps and they prefer directions consistent with the direction of the previous movement.