DISCRETIZATION PROCEDURES FOR CONTINUOUS TIME DECISION PROCESSES.

Research output: Chapter in Book / Report / Conference proceedingConference contributionAcademicpeer-review

Abstract

This paper considers decision processes with countable state space and continuous time parameter. The permitted policies are not restricted to the class of memoryless policies. In order to construct the probability law of a decision process ruled by a fixed non-memoryless policy, a sequence of discrete-time stochastic processes is defined by a discretization procedure, and the weak convergence of this sequence of processes is proved. Using this weak convergence property it can be proved that to any policy depending on the past states and with weak regularity conditions and any initial state there exists a memoryless (randomized) policy such that the one-dimensional marginal distributions at time t of the both induced processes are the same for all t. Refs.
Original languageEnglish
Title of host publicationThin Solid Films
PublisherD. Reidel Publ Co
Pages143-154
Publication statusPublished - 1978
Externally publishedYes
EventTrans of the Prague Conf on Inf Theory, Stat Decis Funct, Random Processes, 8th -
Duration: 28 Aug 19781 Sept 1978

Publication series

NameThin Solid Films
ISSN (Print)0040-6090

Conference

ConferenceTrans of the Prague Conf on Inf Theory, Stat Decis Funct, Random Processes, 8th
Period28/08/781/09/78

Fingerprint

Dive into the research topics of 'DISCRETIZATION PROCEDURES FOR CONTINUOUS TIME DECISION PROCESSES.'. Together they form a unique fingerprint.

Cite this