Abstract
This paper describes a process for determining the value of the gradient of the real outputs of a program with respect to its real parameters. CalledGradient Instrumentation, it is a mechanical process of insertion into the program's source code. The resulting program yields the gradient without the re-execution of the program. The sample path derivatives of many discrete event dynamical system simulations can be found using Gradient Instrumentation, by treating them as deterministic programs. The technique can also be applied to continuous simulations. The subject of a patent, Gradient Instrumentation yields derivatives of any order.
Similar content being viewed by others
References
Braude, E. J. 1991. An efficient method for optimization by simulation.Simulation Digest 21: 51–52.
Cao, X. R. 1987. A sample performance function of Jackson Queuing networks.Operations Research 9: 128–136.
Dijkstra, E. 1969.A Discipline of Programming. Englewood Cliffs, N.J.: Prentice Hall.
Glasserman, P. 1991.Gradient Estimation via Perturbation Analysis. Norwell, MA: Kluwer.
Ho, Y. C., and Cao, X. R. 1983. Optimization and perturbation analysis of queuing networks.Journal of Optimization Theory and Applications 40: 559–582.
Ho, Y. C., and Cao, X. R. 1991.Perturbation Analysis of Discrete Event Dynamic Systems. Norwell, MA: Kluwer.
Jeudes, D. 1991. A Taxonomy of Automatics Differentiation Tools,Automatic Differentiation of Algorithms: Theory, Implementation and Application, A. Griewank and C. F. Corliss, eds., SIAM, Philadelphia, PA.
L'Ecuyer, P. 1991. An overview of derivative estimation.Winter Simulation Conference, 207–217.
L'Ecuyer, P. 1990. A unified view of the IPA, SF, and LR gradient estimation techniques.Management Science 36: 1364–1383.
Rubenstein, R. Y. 1989. Sensitivity analysis and performance extrapolation for computer simulation models.Operations Research 37: 72–80.
Suri, R. 1987. Infinitesimal perturbation analysis for general discrete event systems.Journal of the Association for Computing Machinery 34: 686–717.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Braude, E.J. Obtaining sample path derivatives by source code instrumentation. Discrete Event Dyn Syst 6, 371–378 (1996). https://doi.org/10.1007/BF01797137
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/BF01797137