Method cooperation in optimization process

This paper describes some ideas for method cooperation in optimization. Many methods for optimization have problems in some cases. Usually those problems don´t persist if we will change method. We wanted to show how methods can be forced to cooperate in solving optimization problem. As one method we will use Particle Swarm optimization (PSO) method. This method can get stuck in local minimum. There are improvements of this method like Repulsive Particle Swarm optimization method (RPSO). Basic PSO is working with “best known solution”. This solution is usually found by particles from swarm but it can be injected by external cognition. We will present how the system for optimization can be designed as an open system. This design can accept any method that will be coded in future. The open system will be able to accept different methods and use them as solvers. Designed open system has subsystem which is mastering set of optimization methods (mentioned PSO, random shooter, etc.). Those methods are working together like multiagent system. The cooperation of agents-methods can combine advantages and eliminate disadvantages of individual methods.