Sensitivity Analysis of Discrete Event Systems by the 'Push Out' Method.

In this paper, we introduce a new technique called the "push out" technique and show how to combine it with the standard Score Function (SF) method for sensitivity analysis (evaluation of the gradient, Hesian, etc.) of a Discrete Event Systems (DES). The idea of our method is simple indeed. We merely replace (transform) the original sample performance by an auxiliary one while "pushing out" the vector of the parameters from the original sample performance to a pdf (probability density function), associated with the original one. We show how both the auxiliary sample performance and the associated pdf can be obtained from their original counterparts and how combining them together to perform sensitivity analysis for the underlying (original) DES. We also discuss the advantage of the proposed method and present numerical results supporting our theory.