Integrating evolutionary models in an event-scheduling simulation engine

Resumen

The research developed introduces a new set of methods and tools to construct evolutionary systems, First, we present characteristics and definitions of evolutionary systems, which will be useful for the generalized modeling process. This knowledge will be used to compare different models and thus define the correct formalism in each case. Because systems of this type are constantly changing, the model may evolve differently than the system if the simulation expands in terms of time. We have therefore defined a metric for determining when an evolutionary simulation model differs too much from the evolution of the system. Next, we introduce a method to find equivalences between DEVS and SDL. DEVS is perhaps one of the most powerful formalisms for describing simulation models; in fact, its purpose is primarily to represent dynamic systems. The graphical representation of SDL allows an easier model understanding. The proposed method resulting of our research permits to switch systematically from SDL to DEVS and vice versa. We show that this switch can always be performed. We present different solutions that can represent the major features of evolutionary systems, such as environment and behavior. For the environment, a generalization of the common cellular automaton is presented allowing simple GIS data use. Our cellular automata can also be implemented over continuous space in addition of the discrete space represented by a matrix. Next, a wildfire simulation model is presented using the SDL formalism. In this modeling problem, which deals with complex behavior, an intelligent agent is specified and a new event is defined - the generic event - that allows complex behaviors to be modified without modifying the simulator kernel. Finally we show the implementation of the different elements, highlighting the different practical solutions to implement the methodological solutions shown in the previous sections. These implementations, based on the LeanSim system, describe