This thesis describes a framework for designing emergent behaviour. The framework is comprised of a simulation architecture, a rule migration process, and a rule combination language. Desired emergent behaviour is specified using a multi-level model which describes the behaviour at the appropriate level in the emergence hierarchy, using downward causation if appropriate. The rule migration process and rule combination language translate the multilevel specification to a design for a chosen target model. In the target model, all behaviour occurs as a result of upward causation, and therefore any emergent behaviour is true emergence. The multi-level model and target model are executed concurrently on the simulation architecture, and checked for equivalence.
The definition of the rule migration process is formalised by using it to design swarming behaviour in case studies for two different target models: a Multi-agent System, and a Cellular Automaton. These results are presented as a suitable framework for designing emergence.
Full thesis : PDF 10.7MB