Title :
Successive approximation algorithms for stochastic games-Numerical comparisons
Author :
Dasgupta, P. ; Gruver, William A.
Author_Institution :
North Carolina State University, Raleigh, NC
Abstract :
In this paper we treat algorithmic methods for solution of stochastic games. A value iteration method incorporating bounds in a test for suboptimality is compared with policy iteration for three types of transition probability matrices. Numerical experiments demonstrate the superiority of the value iteration technique for problems with special structure.
Keywords :
Approximation algorithms; Stochastic processes; Testing;
Conference_Titel :
Decision and Control including the 17th Symposium on Adaptive Processes, 1978 IEEE Conference on
Conference_Location :
San Diego, CA, USA
DOI :
10.1109/CDC.1978.268108
