Numerical Methods for Bayes Sequential Decision Problems
Abstract
The general approach to sequential decision-theoretic problems where sums of successive observations are approximated by a continuous time Wiener process has a number of fundamental advantages. Simple numerical techniques which can be employed to obtain explicit descriptions of the solutions of the resulting continuous time optimal stopping problems are described. The techniques are not well adapted for very precise results, but are surprisingly effective for reasonably accurate approximations. Special features of particular problems can be exploited to reduce the necessary computational effort. The techniques are illustrated in a number of problems thereby clearly indicating their properties.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1984
- Accession Number
- ADA140418
Entities
People
- A. J. Petkau
- H. Chernoff
Organizations
- Massachusetts Institute of Technology