Hybrid systems modeling

Abstract

The 'Advances in the State of the Art of Modeling and Simulation' special issue is in two parts. The first part comprises five papers on hybrid systems modeling, 1-3 which is the combined application of modeling and simulation with methods and techniques from disciplines such as computer science/applied computing, business analytics, data science, operations research, systems engineering, economics, humanities, and psychology. Examples include the combined application of qualitative system dynamics, problem structuring methods, forecasting, classical optimization techniques, process mining, data-mining, machine learning, game-theoretic modeling, etc. with computer simulation. The mixed method approach could be applied to different stages of a modeling and simulation study (problem formulation, model implementation, verification and validation, experimentation, etc.).The second part of the special issue comprises two papers on discrete event system specification. 4,5 These two papers will be published in the next issue of the journal. The five hybrid systems modeling papers from this special issue are now introduced, together with an outline of specific methods and techniques that have been combined with modeling and simulation approaches.Collins and Frydenlund 6 present a hybrid systems modeling approach that combines game theory, a branch of economics that is devoted to the study of conflict and cooperation between rational decision-makers, with modeling and simulation. The authors explore strategic group formation by drawing on cooperative game theory in agent-based modeling for group decision-making. They consider a game setting in which agents compete for resources with their neighbors. The authors state that by placing strategic group selection behavior in an agent environment, they benefit from empirical results that overcome some of the limitations of standard cooperative game theory (e.g., computational complexity). This is indeed the objective of hybrid systems modeling -to combine modeling and simulation approaches with methods and techniques from other disciplines with the view of developing the best possible representation of the underlying system under study.The paper by Bell and Mgbemena 7 is again an example of a hybrid systems modeling approach, in which the authors combine data-mining techniques with computer simulation. More specifically, the authors have used agent-based modeling with classification and regression trees (CART), a type of decision tree, to understand customer behavior specific to retention in mobile telecommunication marketspace. The paper presents the customer-agent decision tree (CADET) method, which is a data-driven approach to agent-based modeling and simulation utilizing large and rapidly changing datasets.The third paper in the special issue, by Kazi and Wainer, 8 not only has resonance with topics such as sustainability and circular economy, but the authors go further in contextualizing their work to the spatial analysis of ecosystem services. The authors have used a comb...