Authors: Jean-Baptiste Soyez, Gildas Morvan, Rochdi Merzouki, Daniel Dupont
This article deals with the conception, modeling and simulation of complex systems, represented at different levels of analysis with respect to the agent-based modeling (ABM) paradigm and more precisely on the generic meta-model IRM4MLS. A methodology, using IRM4MLS, is proposed to save computational resources in multi-level agent-based simulations, representing only the relevant elements. It means that the structure of agents can be modified during simulation, by temporarily aggregating, removing or approximating their characteristics to maximize their life-cycles.