Large-Scale Complex Adaptive Systems using Multi-Agent Modeling and Simulation

Modeling and analysis of large-scale complex adaptive systems (CAS) is critical to understanding their key properties such as self-organization, emergence, and adaptability. These properties are difficult to analyze in real-world scenarios due to performance constraints, metric design, and limitations in existing modeling tools. In our previous work, we proposed the Complex Adaptive Systems Language (CASL) and its associated framework. In this paper, we introduce CASL-SG, a Semantic Group extension for large-scale modeling using relational hierarchies. CASL-SG permits largescale simulations to be executed on modest hardware by enabling simulations to contain approximately twice as many agents. This achieves a 356% runtime improvement for a Game of Life model with 4 million cells. CASL-SG also enables designers to model behaviors of collectives that drive the system and entities towards self-organization and adaptability.