Variance Reduction in the Simulation of Stochastic Activity Networks.
Abstract
This paper describes a Monte Carlo method based on the theory of quasirandom points for estimating the distribution functions and means of network completion time and shortest path time in a stochastic activity network. In particular, the method leads to estimators whose variances converge faster than 1/K, where K denotes the number of replications collected in the experiment. The paper demonstrates how accuracy diminishes for a given K with increasing dimensionality of the network and shows how a procedure that uses a cutset of the network together with convolution can reduce dimensionality and increase accuracy. Two examples illustrate the benefits of using quasirandom points together with a cutset and then convolution. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1983
- Accession Number
- ADA124251
Entities
People
- George S. Fishman
Organizations
- University of North Carolina at Chapel Hill