Fast Simulation of Multicomponent Dynamic Systems

Abstract

A computer simulation has to be fast to be helpful, if it is employed to study the behavior of a multicomponent dynamic system. This paper discusses modeling concepts and algorithmic techniques useful for creating such fast simulations. Concrete examples of simulations that range from econometric modeling to communications to material science are used to illustrate these techniques and concepts. The algorithmic and modeling methods discussed include event-driven processing, ``anticipating'' data structures, and ``lazy'' evaluation, Poisson dispenser, parallel processing by cautious advancements and by synchronous relaxations. The paper gives examples of how these techniques and models are employed in assessing efficiency of capacity management methods in wireless and wired networks, in studies of magnetization, crystalline structure, and sediment formation in material science, in studies of competition in economics.