Discrete event simulation of large-scale spatial continuous systems

Abstract

Complex spatially-extended systems consist of numerous sub-systems leading to large simulation execution times. One approach to reducing these execution times is designing a simulation engine to allocate its attention to subsystems in proportion to their activity levels. In this paper, we consider a large scale simulation of a physics-based fire spread model. This model is discretized using a recently developed numerical method called quantization and implemented using discrete event simulation. In this paper, we provide comparisons between the quantization method and usual Euler discrete-time methods. The aim is to demonstrate the ability of quantization and discrete event simulation to focus on active sub-systems, thus significantly reducing execution time for large heterogeneous systems.