Variance reduction of quantile estimates via nonlinear control

ABSTRACT

Linear controls are a well known technique for achieving variance reduction in computer simulation. Unfortunately the effectiveness of a linear control depends upon the correlation between the statistic of interest and the control which is often low. Since statistics are often nonlinear functions of the control this implies that nonlinear controls offer a means for improvement over linear controls. Nonlinear controls have had success in increasing the variance reduction over a linear control. This current work focuses on the use of nonlinear controls for reducing the variance of quantile estimates. The paper begins with a short discussion of linear controls. It describes nonlinear controls and the possibility for improved performance. The final sections discuss quantiles as controls and the potential of nonlinear controls for variance reduction in quantile estimation.

REFERENCES

• 1: Beale, E.M.L. (1985). Regression: A bridge between analysis and simulation. The Statistician 34, 141--154.
• 2: Breiman, Leo and Friedman, Jerome H. (1985) Estimating optimal transformations for multiple regression and correlation. Journal of the American Statistical Association 80, 391, 580--619.
• 3: Efron, Bradley and Gong, Gail (1983). A leisurely look at the Bootstrap, the Jackknife and Cross-Validation. The American Statistician 37, 1, 36 - 48.
• 4: Gallant, A. R. (1975). Nonlinear Regression. The American Statistician 29, 2, 73--81.
• 5: Lancaster, H. O. (1966). Kolmogorov's remark on the Hotelling canonical correlations. Biometrika 53, 585--588.
• 6: Lavenberg, Stephen, S. and Welch, Peter D. (1981). A perspective on the use of control variables to increase the efficiency of monte carlo simulations. Management Science 27, 3, 322--335.
• 7: Lewis, P. A. W. and Orav, E. J. (1989). Simulation Methodology for Statisticians, Operations Analysts and Engineers. Wadsworth & Brooks/Cole, Pacific Grove, California.
• 8: Lewis, P.A.W., Ressler, Richard L. and Wood, R. Kevin (1989). Variance reduction using nonlinear controls and transformations. Communications in Statistics 18B, 2.
• 9: Nelson, Barry (1988). Control variate remedies. Working Paper Series No. 1988--004, Department of Industrial and Systems Engineering, The Ohio State University, Columbus, Ohio.
• 10: Weiss, Lionel (1963). On the asymptotic joint normality of quantiles from a multivariate distribution. Journal of Research of the National Bureau of Standards 68B, 2, 65--66.

INDEX TERMS

Primary Classification:
  I. Computing Methodologies
  I.6 SIMULATION AND MODELING

Additional Classification:
  G. Mathematics of Computing
  G.1 NUMERICAL ANALYSIS
      G.1.6 Optimization
          Subjects: Linear programming

General Terms:
Algorithms, Design, Performance