scientific article; zbMATH DE number 3849001
[For the entire collection see Zbl 0521.00029.] The paper is concerned with the usual model of stochastic dynamic programming. The state space is finite. The criterion is essentially the expected total reward for the infinite horizon. However, it is not assumed that the expected total n-stage reward converges as n tends to infinity. This total reward model is then treated by methods known for the average reward case. The main results concern the existence of optimal stationary policies and the characterization of optimal policies by two optimality equations. The results seem to be new even for the classical positive case and are applied to optimal stopping in a forthcoming paper.
- Bounded-parameter Markov decision processes
- Markov-achievable payoffs for finite-horizon decision models.
- scientific article; zbMATH DE number 5946235 (Why is no real title available?)
- scientific article; zbMATH DE number 3847226 (Why is no real title available?)
- scientific article; zbMATH DE number 3898636 (Why is no real title available?)
- Finite state dynamic programming with the total reward criterion
- scientific article; zbMATH DE number 4020871 (Why is no real title available?)
- scientific article; zbMATH DE number 52986 (Why is no real title available?)
- Finite-stage stochastic decision processes with recursive reward structure I: optimality equations and deterministic strategies
- scientific article; zbMATH DE number 1786118 (Why is no real title available?)
- Finite Horizon Markov Decision Processes with Uncertain Terminal Payoffs
- An analysis of transient Markov decision processes
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3318569)