Efficient Agent Based Simulation Exploiting the Model Structure

Agent Based Models (ABM) simulation supports the study of systems whose overall evolution emerges from the behavior of several interacting agents. In [6] a compositional method based on the meta-formalism Agent Based Extended Stochastic Symmetric Nets (AB-ESSN) has been proposed to build ABMs that are then simulated in continuous time according to a next event time advance algorithm. In [22] a structure-based approach has been proposed to improve ABM simulation efficiency, leveraging the possibility of exploiting structural properties of Symmetric Nets (SN) to identify the interdependencies among the agents’ actions. The approach is now facilitated by the tool SNexpression and its recently added module [15, 16] supporting the computation of all dependency relations on a SN. In this paper, the possibility of automatizing the ABM simulation optimization process, thanks to the new features of the SNexpression tool, is discussed through an example.