The aim of this project is to obtain new results which will help reduce the time and computational effort required when using the computer simulation models for optimization found in many discrete-event systems. The focus is on the enhancement of the Score Function and Likelihood Ratio methods to provide accurate estimates for derivative and perturbation, respectively. The goals are improvements of the existing variance reduction techniques, theoretical unification, some extensions, and advancements in the practical implementation of these `single run` methods. The project will develop procedures and algorithms, investigate their accuracy versus execution time behavior, and present numerical results for a wide variety of queuing, inventory, and reliability models.
