Improved event-scanning mechanisms for discrete event simulation

Abstract

Simulation models of large, complex “real-world” applications have occasionally earned the reputation of eating up hours of computer time. This problem may be attributed in part to difficulties such as slow stochastic convergence. However, an additional problem lies in the fact that a significant amount of bookkeeping time is required to keep future events in their proper sequence. This paper presents a method for significantly reducing the time spent scanning future events lists in discrete event simulations. Three models are presented, all of which improve in effectiveness as the events-list scan problem becomes more burdensome.