Co-evolution of agents in self-organized criticality

Evolutionary computation has been applied successfully in evolutionary multi-agent system (MAS) design in electronic commerce (EC). Studies of evolutionary MAS are mainly fastened on "fastest find the fittest steady state" now. This is not adaptive or suitable when the problem goes much complicated or rapidly changing. This paper describes the analysis of studies in software evolution, especially of MAS, presents a united evolutionary model, proves reality system, information or software system and knowledge system are united in self-organized criticality state, and discusses the co-evolution rules of agents in self-organized criticality. This model covers the following features: the co-evolution agents for self-organizing and self-adaptive multi-agent systems, the united multi-agent system for reality system and knowledge management, and the dynamic information creation in self-organized criticality.