A CONVEX ANALYTIC FRAMEWORK FOR ERGODIC CONTROL OF SEMI-MARKOV PROCESSES

Authors
BHATNAGAR S BORKAR VS
Citation
S. Bhatnagar et Vs. Borkar, A CONVEX ANALYTIC FRAMEWORK FOR ERGODIC CONTROL OF SEMI-MARKOV PROCESSES, Mathematics of operations research, 20(4), 1995, pp. 923-936
Citations number
15
Categorie Soggetti
Operatione Research & Management Science",Mathematics,"Operatione Research & Management Science",Mathematics
Journal title
Mathematics of operations research → ACNP
ISSN journal
0364765X
Volume
20
Issue
4
Year of publication
1995
Pages
923 - 936
Database
ISI
SICI code
0364-765X(1995)20:4<923:ACAFFE>2.0.ZU;2-L
Abstract
The ergodic control problem for semi-Markov processes is reformulated as an optimization problem over the set of suitably defined 'ergodic o ccupation measures.' This set is shown to be closed and convex, with i ts extreme points corresponding to stationary strategies. This leads t o the existence of optimal stationary strategies under additional hypo theses. A pathwise analysis of the joint empirical occupation measures of the state and control processes shows that this optimality is in t he strong (i.e., almost sure) sense.