Abstract for: Adaptive sampling to explore the behavioral spectrum of system dynamics models: The case of mineral/metal scarcity

This paper describes a method for searching for undiscovered behaviors dynamically complex models could generate. Random sampling methods are sometimes used for determining the breadth of behaviors a system dynamics model could generate. However, as illustrated in this paper, an adaptive sampling strategy allows for the enumeration of a greater number of behaviors compared to preexisting methods. The method is used to analyze a previously studied scarcity model and to identify new behaviors not exhibited by random sampling. Furthermore, the application of the adaptive sampling method allows for a set of new visualization techniques, which allow to capture the large-scale behavioral patterns of the model.