Comments on Estimating Quantiles for Gaussian Functionals by Simulation.

Abstract

The construction of an optimum model-sampling simulation experimental design is often an art and clearly very much problem dependent. The accurate estimation of the distribution of any random variable by computer is almost always difficult, especially when reasonable accuracy is desired for its tail probabilities. If, in addition, each element of the computer generated pseudo-random sequence is itself the result of a stochastic limit or possibly a functional of a continuous-time process, it becomes most complicated to assess the final statistics. In this paper, we focus on the estimation of tail probabilities for the distribution function of the maximum on the unit interval of a continuous-time Wiener process approximated as a multivariate normal of increasing dimension. We critique recent approaches to sample-size determination for such distribution-sampling problems and build on insights from probability theory to find more reasonable run sizes. Keywords include: Simulation; Quantile estimation; Distribution function estimation; Gaussian functionals; Brownian motion.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Apr 01, 1985
Accession Number
ADA154302

Entities

People

  • C. M. Harris

Organizations

  • University of Virginia

Tags

DTIC Thesaurus Topics

  • Applied Mathematics
  • Brownian Motion
  • Business Administration
  • Computer Science
  • Computers
  • Data Science
  • Distribution Functions
  • Engineering
  • Information Science
  • Materials Science
  • Network Science
  • Probability
  • Probability Distributions
  • Random Variables
  • Statistical Algorithms
  • Statistics
  • Systems Engineering

Fields of Study

  • Mathematics

Readers

  • Computational Modeling and Simulation
  • Statistical inference.