A comparison of perturbation analysis techniques

ABSTRACT

Perturbation analysis (PA) is a technique for estimating gradients of performance measures, particularly applicable to the simulation of discrete-event systems. Over the past two decades, various "versions" have been developed. In this paper, we compare and contrast some of these perturbation analysis techniques by applying them to a simple example. This example also serves to highlight the issue of process representation that can play a very crucial role in the application of perturbation analysis.

REFERENCES

• 1: P. Brémaud, Maximal coupling and rare perturbation sensitivity analysis, Queueing Systems: Theory and Applications, v.11 n.4, p.307-333, Dec. 1992 [doi>10.1007/BF01163859]
• 2: Pierre Brémaud , Felisa J. Vázquez-Abad, On the pathwise computation of derivatives with respect to the rate of a point process: the phantom RPA method, Queueing Systems: Theory and Applications, v.10 n.3, p.249-270, Jan. 1992 [doi>10.1007/BF01159209]
• 3: Dai, L.. and Y.C. Ho. 1995. Structural infinitesimal perturbation analysis for derivative estimation in discrete event dynamic systems, IEEE Transactions on Automatic Control 40:1154-1166.
• 4: Fu, M.C. 1994. Optimization via simulation: a review. Annals of Operations Research 53: 199-248.
• 5: Fu, M.C. and J. Q. Hu. 1992. Extensions and generalizations of smoothed perturbation analysis in a generalized semi-Markov process framework, IEEE Transactions on Automatic Control 37: 1483-1500.
• 6: Fu, M.C. and J. Q. Hu. 1996. Conditional Monte Carlo: Gradient Estimation and Optimization Applications, Kluwer Academic Publishers, forthcoming.
• 7: Gaivoronski, A., L.Y. Shi, and R.S. Sreenivas. 1992. Augmented infinitesimal perturbation analysis: an alternate explanation, Discrete Event Dynamic Systems: Theory and Applications, 2: 121-138.
• 8: Glasserman, P. 1991. Gradient Estimation Via Perturbation Analysis, Kluwer Academic.
• 9: Gong, W.B. and Y.C. Ho. 1987. Smoothed perturbation analysis of discrete-event dynamic systems, IEEE Transactions on Automatic Control 32: 858- 867.
• 10: Ho, Y.C. and X.R. Cao. 1991. Discrete Event Dynamic Systems and Perturbation Analysis, Kluwer Academic.
• 11: G. Ch. Pflug, Gradient estimates for the performance of Markov chains and discrete event processes, Annals of Operations Research, v.39 n.1-4, p.173-194, 1992 [doi>10.1007/BF02060941]
• 12: Shi, L. Y. 1996. Discontinuous perturbation analysis of discrete event dynamic systems, to appear in IEEE Transactions on Automatic Control.