Discovery of emergent natural laws by hierarchical multi-agent systems

摘要

This paper defines an approach to simulation of natural systems, inspired by complex systems theory. A complex natural system is modeled as a multi-agent simulation system, agents representing living organisms, physical entities or environmental processes. Agents and their interactions can be aggregated to higher-level group agents. The properties and behavior of these group agents are determined by, or emerge from, the properties and behavior of the individual agents composing the group. Group agents discover macro-level natural laws implied by the properties and behavior of individual agents modeling micro-level natural entities. Such a system can be implemented in a distributed programming environment, exploiting emergence, hierarchy, and concurrency to perform large-scale simulations.