Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
Asymptotic Linear Programming and Policy Improvement for Singularly Perturbed Markov
 

Summary: Asymptotic Linear Programming and Policy
Improvement for Singularly Perturbed Markov
Decision Processes
Eitan Altman \Lambda Konstantin E. Avrachenkov y Jerzy A. Filar z
Abstract
In this paper we consider a singularly perturbed Markov decision
process with finitely many states and actions and the limiting expected
average reward criterion. We make no assumptions about the under­
lying ergodic structure. We present algorithms for the computation
of a uniformly optimal deterministic control, that is, a control which
is optimal for all values of the perturbation parameter that are suf­
ficiently small. Our algorithms are based on Jeroslow's Asymptotic
Linear Programming.
Key Words: Markov Decision Processes, Singular Perturbations,
Asymptotic Linear Programming.
Running title: Asymptotic Linear Programming for Singularly
Perturbed MDP's.
1 Introduction
Singularly perturbed Markov decision processes (MDPs, for short), are dy­
namic stochastic systems controlled by a ''controller'' or a ''decision­maker''

  

Source: Avrachenkov, Konstantin - INRIA Sophia Antipolis

 

Collections: Computer Technologies and Information Sciences