Evolutionary algorithms are very efficient tools to find a near-optimumsolution in many cases. Until now they have been mostly used to find results butin this article we argue that evolutionary algorithms can also be usedto simulate the evolution of complex systems. We modelcomplex systems as networks in which agents are connected by edges if theyinteract with each other. It is known that many networks of this kind exhibit stable properties despite the dynamic processes they are subject to.We show here how evolutionary processes on complex systems can be modeled with a new kind of evolutionary algorithm which we have presented in \cite{lk-eaftsoeon-05}. We will show that some evolutionary processes within this framework yieldnetworks with stable properties in reasonable time. An understanding ofwhat kind of evolutionary processes will produce what kind of network properties in what timeis vital to transfer evolutionary processes to technical ad-hocnetworks in order to improve their flexibility and stability in quickly changing environments. |
