Complex systems are often difficult to understand because of various issues that influence the behavior and interaction of entities in the system. In many situations researchers, policy makers and governing bodies would like to understand complex systems and predict the impact of regulations, interventions or other measures. E.g. one would like to investigate whether informing car drivers about traffic violations and specific sanctions increases the road safety and throughput, whether creating a vegetarian option in all restaurants or having a vegetarian menu once in a while would increase the number of vegetarians, and whether specific public transportation will stimulate the spread of diseases. Agent based simulation is a computational approach for modelling complex systems, where individuals (e.g., human, animals, autonomous systems or human collectives) are modelled as software agents.
In our research group we investigate how large scale complex agent based simulations can be developed. In the social simulations that we investigate the individual software agents represent humans or groups of humans (families, companies, etc.). However in general the agents can represent any complex reasoning entity whose behavior is determined by various issues such as information, preferences, options, norms, regulations and emotions. We look at the interaction of new regulations with existing norms, practices and other structures and emerging patterns of behavior. This is being used in the policies around the smoking prohibition in public places, in fishery management policies, traffic management, transportation, epidemic, and energy. We investigate context aware agents for these simulations that generate realistic emerging behavior of the population giving insight in the factors involved. Specifically we investigate complex practices in terms of routines in the decision making process of the agents and the influence of information, preferences, values, norms, emotions, and social identity in the agent decision making process.