A multiagent, dynamic rank-driven multi-deme architecture for real-valued multiobjective optimization
Introducing elitism into evolutionary multi-agent system for multi-objective optimization proofed to be smooth both conceptually and in realization. Simultaneously it allowed for obtaining results with comparable high quality to such referenced algorithms as Non-dominated Sorting Genetic Algorithm (NSGA-II) or Strength Pareto Evolutionary Algorithm (SPEA2). What is more, applying mentioned agent-based computational paradigm for solving multi-criteria optimization tasks in ldquonoisyrdquo environments mainly because of-characteristic for EMAS-based approach-a kind of soft selection allowed for obtaining better solutions than mentioned referenced algorithms. From the above observations the following conclusion can be drown: evolutionary multi-agent system (EMAS) (and being the subject of this paper elitist evolutionary multi-agent system (elEMAS) in particular) seems to be promising computational model in the context of multi-criteria optimization tasks. In previous works however the possibility of applying elEMAS for solving constrained multi-objective optimization task has not been investigated. It is obvious however that in almost all real-life problems constraints are a crucial part of multi-objective optimization problem (MOOP) definition and it is nothing strange that among (evolutionary) algorithms for multi-objective optimization a special attention is paid to techniques and algorithms for constrained multi-objective optimization and a variety-more or less effective-algorithms have been proposed. Thus, the question appears if effective constrained multi-objective optimization with the use of elitist evolutionary multi-agent system is possible. In the course of this paper preliminary answer for that question is given.