This article is cited in 22 papers

Short Communications

On a class of policies in general Markov decision models

A. A. Yushkevich


Abstract: The paper studies stationary policies which, under some final reward, become optimal on each time interval $[0, n]$ and provide a total gain linearly dependent on $n$. Necessary and sufficient conditions for the existence of such policies are given in the form of equations (4), (5). These equations appeared previously in various cases as sufficient optimality conditions for the average-per-unit-time criterion.

Received: 08.01.1973

 English version: Theory of Probability and its Applications, 1974, 18:4, 777–779

Bibliographic databases:

• 
•