Minimum variance criterion in stochastic dynamic programming. Abstract

0410870 - UTIA-B 20020084 GB eng A - Abstract
Sladký, Karel
Minimum variance criterion in stochastic dynamic programming. Abstract.
Edinburgh: UK Operational Research Society, 2002. International Federation of Operational Research Societies 2002. IFORS 2002. Abstracts. s. 28
[IFORS 2002. 08.07.2002-12.07.2002, Edinburgh]
R&D Projects: GA ČR GA402/02/1015; GA ČR GA402/01/0539
Institutional research plan: CEZ:AV0Z1075907
Keywords : stochastic dynamic programming * Markov decision chains * mean-variance
Subject RIV: BB - Applied Statistics, Operational Research

We investigate how the minimum variance criterion can work in discrete stochastic dynamic programming. We adapt notions and notation used in Markov decision chains and in contrast to the classical models we also consider variance of the obtained total reward. Alternative definitions of the reward variance along with their mutual connections are discussed. Attention is also focused on finding policies minimizing the average reward variance on condition that the average reward is not less than a given value.
Permanent Link: http://hdl.handle.net/11104/0130957