On the convergence of global rational approximants for stochastic discrete event systems (Q679030)

This paper investigates the convergence and convergence rates of the rational approximants of integer parameter functions, which are often encountered in the performance evaluation and analysis of stochastic discrete event systems (DES), such as computer systems, communication networks and general distributed and parallel processing systems. Two types of rational approximants, Type-1 and Type-2 [\textit{W. B. Gong} and \textit{H. Yang}, IEEE Trans. Comput. 44, No. 12, 1394-1404 (1995)], are introduced, which are used in global approximation for stochastic DES. It is shown that the convergence rates of Type-1 and [\(n/n\)]Type-2 approximants are given by orders \(O(1/\sqrt{n})\) and \(O(n^\alpha e^{-\beta \sqrt{n}})\), respectively. A numerical example of the global rational approximants approach is presented for the analysis of the Queue Inference Engine problem.