Complex systems are collections of independent agents interacting to as to produce emergent, often unexpected, behaviour. Computer based simulation is one of the main ways of studying complex systems and a naive approach to such simulation is fraught with difficulty due to the scope for deadlock in various patterns of interaction between the agents which are of necessity sharing aspects of the computational platform.
Agent behaviour, though, can be entirely looked at from the point of view of the environments within which the agents interact. Structuring a simulation purely in this manner leads to a simulation that has essentially no tendency to deadlock and still behaves in the manner required.
A number of experiments are conducted to demonstrate the feasibility of this approach. These start with a simple flocking system and continue through an investigation of the ways in which multiple environments can best be combined. Finally, a larger scale experiment investigating the evolution of variety in a rich environment shows that interesting results can be obtained of a simulation constructed in this manner.
Full thesis : PDF 30.2MB