Abstract
Let\(l(v) = E_{v_1 } \{ L(Y,v_2 )\} \) be the expected performance measure of adiscrete event system (DES), whereL is the sample performance depending on the vector of parametersv 2 and driven by an input vectorY, which has a probability density function (pdf)f(y, v 1),v=(v 1,v 2) is a vector of parameters, and the subscriptv 1 in E\(E_{v_1 } L\) indicates that the expectation is taken with respect to the pdff(y, v 1). Suppose thatl(v) is not available analytically and we want to evaluate (estimate) it, as well as the associated sensitivities ∇k l(v),k=1, 2, ...simultaneously for different values ofv=(v 1,v 2) via simulation. In this paper, we show that in some cases interesting for applications, we can estimatel(v) and ∇k l(v),k=1,2, ... by using the so-called “push out” technique. More precisely, we show that it is possible to replace the original sample performance by an auxiliary one while “pushing out” the parameter vectorv 2 from the original sample performance functionL(Y,v 2) to a pdf\(\tilde f\)(x,v 1,v 2) associated with the original onef(y,v 1). We also show how both the auxiliary sample performance and the associated pdf can be obtained from their original counterparts and how to combine them together to perform sensitivity analysis for the original DES. Particular emphasis will be placed on the case where the sample performance functionL(y,v 2) isneither analytically available nor everywhere differentiable in v 2. We finally discuss the advantage of the proposed method and present numerical results supporting our theory.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
S. Asmussen,Applied Probability and Queues (Wiley, 1987).
S. Asmussen, Queueing simulation in heavy traffic, Math. Oper. Res. 17(1992)84–111.
S. Asmussen and R.Y. Rubinstein, The efficiency and heavy traffic properties of the score function method for sensitivity analysis of queueing models, Adv. Appl. Prob. 24(1992)172–201.
S. Asmussen and R.Y. Rubinstein, The performance of the likelihood ratio estimators via the score function method, Manuscript, Chalmers Institute of Technology, Sweden (1990).
S. Asmussen and R.Y. Rubinstein, Performance evaluation for the score function method in sensitivity analysis and stochastic optimization, Manuscript, AT&T Bell Laboratories, Holmdel, NJ 07733 (1991), to appear 1992.
B.L. Fox and P. Glasserman, Estimating derivatives via Poisson's equation, Manuscript, AT&T Bell Laboratories, Holmdel, NJ 07733 (1990).
P. Glasserman, Smoothing complements and randomized score function, Manuscript, AT&T Bell Laboratories, Holmdel, NJ 07733 (1990).
P. Glasserman, Gradient estimation via perturbation analysis, Manuscript, AT&T Bell Laboratories, Holmdel, NJ 07733 (1991).
P.W. Glynn, A GSMP formalism for discrete events systems, Proc. IEEE 77(1989)14–23.
P.W. Glynn, Likelihood ratio derivative estimators for stochastic systems, Commun. ACM (1990).
P.W. Glynn and D.I. Iglehart, Importance sampling in stochastic simulations, Manag. Sci. 35(1989)8.
P.W. Glynn and D.I. Iglehart, Simulation methods for queues. An overview, Queueing Systems 3(1988)221–255.
P.W. Glynn and W. Whitt, Indirect estimation viaL =λW, Oper. Res. 37(1989)82–103.
P. L'Ecuyer, A unified view of infintesimal perturbation analysis and likelihood ratios, Manag. Sci. 36(1991)1364–1384.
D.L. McLeish and S. Rollans, Conditioning for variance reduction in estimating the sensitivity in simulations, Ann. Oper. Res. (1992), this volume.
A. Perez-Luna, Sensitivity analysis and optimization of queueing networks by the score function method, Master Thesis, Technion, Haifa, Israel (1990).
C.Ch. Pflug, Sampling derivatives of probabilities, Computing 42(1989)315–328.
H. Rief, Monte Carlo uncertainty analysis,CRC Handbook on Uncertainty Analysis, ed. Y. Ronen (1988).
M.I. Reiman and A. Weiss, Sensitivity analysis for simulations via likelihood ratios, Oper. Res. 37(1990)830–844.
R.Y. Rubinstein, A Monte Carlo method for estimating the gradient in a stochastic network, Manuscript, Technion, Haifa, Israel (1976), unpublished.
R.Y. Rubinstein, Sensitivity analysis of computer simulation models via the efficient score, Oper. Res. 37(1989)72–81.
R.Y. Rubinstein, Monte Carlo methods for performance evaluation, sensitivity analysis and optimization of stochastic systems,Encyclopedia of Computer Science and Technology, Vol. 25, ed. A. Kent and J.G. Williams (Marcel Dekker, 1991), pp. 211–233.
R.Y. Rubinstein and A. Shapiro,Discrete Event Systems: Sensitivity Analysis and Stochastic Optimization via the Score Funtion Method (Wiley, 1992).
M. Shalmon and R.Y. Rubinstein, The variance of the regenerative estimators with special references to sensitivity analysis of queueing systems with Poisson arrivals,1st Int. Conf. on Operations Research in Telecommunications, Boca Raton, FL (1990).
R.A. Topia and J.R. Thompson,Nonparametric Density Estimation (Johns Hopkins University Press, Baltimore, 1978).
Author information
Authors and Affiliations
Additional information
Research supported by the L. Edelstein Research Fund of the Technion-Israel Institute of Technology and AT&T Bell Laboratories, Holmdel, NJ, USA.
Rights and permissions
About this article
Cite this article
Rubinstein, R.Y. Sensitivity analysis of discrete event systems by the “push out” method. Ann Oper Res 39, 229–250 (1992). https://doi.org/10.1007/BF02060943
Issue Date:
DOI: https://doi.org/10.1007/BF02060943