Detection and reification of emerging dynamical ecosystems from interaction networks
Résumé
In this paper, we present an hybrid ecosystem modeling based on emerging computation from interaction networks. Initially based on an individual-based modeling (IBM) simulation, we propose an automatic computation to detect predator-preys systems. After their detection, these systems are replaced by a differential system during the simulation. In this way, we can change the description level and improve both the computation time and the whole system analysis by detecting some emergent organizations. The description modification between IBM representation to differential one needs to identify the global coefficients of these differential equations. Due to the complexity of relations between these two kinds of representations, a genetic algorithm is proposed to solve this identification.