Sensitivity Analysis for Simulations via Likelihood Ratios

We present a simple method of estimating the sensitivity of quantities obtained from simulation with respect to a class of parameters. Here sensitivity means the derivative of an expectation with respect to a parameter. The class of parameters includes, for example, Poisson rates, discrete probabilities, and the mean and variance of a Normal distribution. The method is extremely well suited to regenerative simulation, and can be implemented on extant simulations with little effort, increase in running time, or memory requirements. It is based on some change-of-measure formulas derived from likelihood ratios. In addition to the theorems that underly the technique, we present some numerical examples.

[1]  R. Tolman,et al.  The Principles of Statistical Mechanics. By R. C. Tolman. Pp. xix, 661. 40s. 1938. International series of monographs on physics. (Oxford) , 1939, The Mathematical Gazette.

[2]  H. Chernoff.  A Measure of Asymptotic Efficiency for Tests of a Hypothesis Based on the sum of Observations , 1952 .

[3]  Patrick Billingsley,et al.  Statistical inference for Markov processes , 1961 .

[4]  J. Hammersley,et al.  Monte Carlo Methods , 1966 .

[5]  D. Siegmund.  Importance Sampling in the Monte Carlo Study of Sequential Tests , 1976 .

[6]  P. Bickel,et al.  Mathematical Statistics: Basic Ideas and Selected Topics , 1977 .

[7]  Donald L. Iglehart,et al.  Regenerative Simulation of Response Times in Networks of Queues , 1980, J. ACM.

[8]  P. Brémaud.  Point Processes and Queues , 1981 .

[9]  K. G. Ramakrishnan,et al.  A class of closed Markovian queuing networks: integral representations, asymptotic expansions, and generalizations , 1981 .

[10]  Christos G. Cassandras,et al.  A new approach to the analysis of discrete event dynamic systems , 1983, Autom..

[11]  Marie Cottrell,et al.  Large deviations and rare events in the study of stochastic algorithms , 1983 .

[12]  P. J. Fleming,et al.  An approximate analysis of sojourn times in the M/G/1 queue with round-robin service discipline , 1984, AT&T Bell Laboratories Technical Journal.

[13]  Peter W. Glynn,et al.  Stochastic approximation for Monte Carlo optimization , 1986, WSC '86.

[14]  Donald R. Smith,et al.  An Asymptotic Analysis of a Queueing System with Markov-Modulated Arrivals , 1986, Oper. Res..

[15]  Y. Ho,et al.  Smoothed (conditional) perturbation analysis of discrete event dynamical systems , 1987 .

[16]  Xi-Ren Cao.  Sensitivity estimates based on one realization of a stochastic system , 1987 .

[17]  Martin I. Reiman,et al.  Simterpolations: estimating an entire queueing function from a single sample path , 1987, WSC '87.

[18]  Xi-Ren Cao,et al.  Convergence properties of infinitesimal perturbation analysis , 1988 .

[19]  Martin I. Reiman,et al.  An Interpolation Approximation for Queueing Systems with Poisson Input , 1988, Oper. Res..

[20]  R. Suri,et al.  Perturbation analysis gives strongly consistent sensitivity estimates for the M/G/ 1 queue , 1988 .

[21]  Alan Weiss,et al.  Light traffic derivatives via likelihood ratios , 1989, IEEE Trans. Inf. Theory.

[22]  Martin I. Reiman,et al.  Open Queueing Systems in Light Traffic , 1989, Math. Oper. Res..

[23]  Reuven Y. Rubinstein,et al.  Sensitivity Analysis and Performance Extrapolation for Computer Simulation Models , 1989, Oper. Res..