Dynamic Model Updating in Simulation with Multimodels: A Taxonomy and a Generic Agent-Based Architecture

Abstract

Dynamic model update and replacement are critical capabilities for modeling problems, the underlying processes of which are uncertain. The trajectory of most complex social phenomena, including conflicts, for instance, is never fixed and predictable. Symbiotic simulation is yet another emergent challenge that requires model adaptivity in response to emerging conditions within the actual system. In such experiments simulation parameters, as well as scenarios may shift, invalidating a priori assumptions or analysis. Multimodel formalism and runtime dynamic model updating are suggested as potential approaches to deal with uncertainty. A taxonomy of multimodels, a generic agent-based architecture for static structure, and single aspect multimodels with exploratory (goal-directed) behavior are presented to explore issues in multimodel simulation.