Approximate Solutions for Certain Optimal Stopping Problems.
Abstract
This paper presents simple approximate methods which lead to arbitrarily accurate numerical approximations to the optimal continuation regions of optimal stopping problems involving a zero drift standard Wiener process. The methods involve approximating the Wiener process by a simple random walk, solving the analogous problem for the random walk and subsequently applying a correction for continuity due to Chernoff and Petkau to the solution of the discrete problem. The methods are illustrated in a problem due to Van Moerbeke for which the exact solution is known and in the one-armed bandit problem which has arisen in several different contexts in the statistical literature. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 21, 1977
- Accession Number
- ADA049556
Entities
People
- A. John Petkau
Organizations
- Massachusetts Institute of Technology