시뮬레이션을 통한 시스템 최적화 과정에서 공통 난수 활용의 이점 분석

A Benefit Analysis of Using Common Random Numbers When Optimizing a System by Simulation Experiments

One of the primary goals of the simulation experiments is to understand the overall system behavior and to analyze the system, ultimately to optimize the system. Optimizing the system includes determining the optimum condition of the system parameters of interest. This paper is concerned with the simulation methodology for estimating the unknown objective function for the system of interest and optimizing the system with respect to the controllable factors. In the process of estimating the unknown objective function, which is assumed to be a second order spline function, we use common random numbers for different set of the controllable factors resulting in more accurate par neter estimation for the objective function. We will show some mathematical result for the benefit of using common random numbers.