A method of clustering for discounted markovian decision problems
作者:
A. Hahnewald-busch,
V. Nollau,
期刊:
Mathematische Operationsforschung und Statistik. Series Optimization
(Taylor Available online 1981)
卷期:
Volume 12,
issue 1
页码: 137-147
ISSN:0323-3898
年代: 1981
DOI:10.1080/02331938108842713
出版商: Akademic-Verlag
数据来源: Taylor
摘要:
It was shown by Blackwell [2] that for discounted Markovian Decision Problems (MDP) with countable state space and finite action space there exists a minimal expected total loss function is hounded). In this note we apply an approximation procedure given in [b], [10] to such a discounted MDP, It is shown, that the sequence of solutions for the considered approximating problems converges to the solution of the given MDP. Simultaneously common bounds for the total loss of the approximative and the orignal Droblsms are obtained.
点击下载:
PDF (349KB)
返 回