A PROBLEM IN OPTIMAL SEARCH AND STOP
Abstract
We are told that an object is hidden in one of m(m < infinity) boxes and we are given prior probabilities p sub i superscript 0 that the object is in the i superscript th box. A search of box i costs c sub i and finds the object with probability alpha sub i if the object is in the box. Also, we suppose that a reward R sub i is earned if the object is found in the i superscript th box. A strategy is any rule for determining when to search and if so which box. The major result is that an optimal strategy either searches a box with maximal value of (alpha sub i)(p sub i)/(c sub i) or else it never searches those boxes. Also, if rewards are equal, then an optimal strategy either searches a box with maximal (alpha sub i)(p sub i)/(c sub i) or else it stops.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1968
- Accession Number
- AD0676132
Entities
People
- Sheldon M. Ross
Organizations
- University of California, Berkeley