Multiagent Models of Complex Socio-economic Systems

Agent-based models (ABMs) and computational simulations are considered important research areas in modern computer science. In relation to simulations, the ABMs have the potential to serve as computational laboratories permitting researchers to explore in a controllable (lab) environment how changes in structural conditions, institutional arrangements, and/or decision processes affect system outcomes over time. ABMs are also natural means for the exploration of complex adaptive systems (CAS), i.e., systems of large numbers of components that interact and adapt or learn. This branch of research provides researchers with vast opportunities to study various socio-economic problems, such as long-term sustainability, development of the society, decision-making on the level of the community or its parts, mutual interactions between communities, market interactions, and others. ABMs also represent an interesting alternative to more traditional equilibrium-based economic models, which is given by the fact that their agents interact, learn, and adapt over time, and the structure of the system is formed via the bottom-up approach. There is also a potential to study various emergent phenomena that occur when a large number of agents interact mutually with each other for various lengths of time. The ability of agents to learn and adapt should gradually lead to the rational development of such a community, which in this context means that the optimal utilization of available resources and maximization of its (future) utility value can be reached. The economic point of view helps to evaluate the performance through the competitiveness of the community expressed in the monetary value, but various other criteria can also be applied to measure the performance. The size of ABMs can be adjusted to the considered research questions/objectives and can vary from small models with several individual agents to large-scale communities consisting of thousands of agents. Our research is focused on the creation of large-scale, domain-independent, scalable agent-based computational laboratories for experimenting. The motivation behind this is to provide a systematic theoretical framework, which would work as a basis for further research in the areas where large numbers of individuals and groups interact, cooperate or compete in complex situations.