An empirical work is described which compares the optimization levels produced by a group of economic agents versus those of a similar group of economic agents which additionally employ a genetic algorithm (GA) to attain a higher level of optimization. The problem domain is multimodal. It incorporates multiple hard and soft constraints and dynamical behaviors. It also has areas of infeasibility and non-linear behaviors. The simulated model environment provides several types of sensors, actuators and opportunities for inter-agent resource mediation. Evidence is offered to support the theory that multiple weak methods operating in concert, on a shared problem, can produce better results than the individual weak methods acting alone. The problem area is resistant to the use of strong methods.
展开▼