Pade approximation for stochastic discrete-event systems

We show that Pade approximation can be effectively used for approximation of performance functions in discrete-event systems. The method is (1) obtaining the MacLaurin coefficients of the performance function and (2) finding a Pade approximant from the MacLaurin coefficients and use it to approximate the function. We use the method with the expected number of renewals in a random interval, GI/G/1 systems, and inventory systems. The results are very good.<
>