Significant event simulation

Abstract

This paper compares a new method of simulation organization, called the significant event method, with an old one, called the clock pulse method, using as examples two automobile traffic models. The significant event method is found to be more efficient than the clock pulse method at low levels of system interaction and less efficient at high levels. A simple mathematical model for the trade-off in the relative running time of the two methods is developed. The model aids in choosing between the two simulation methods for a particular experiment. It is concluded that the significant event method can be of value in the simulation of some systems when computational efficiency is of sufficient importance.