Infinite Horizon Optimization for Finite State Markov Chain (Q3773809)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 4033650
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | Infinite Horizon Optimization for Finite State Markov Chain |
scientific article; zbMATH DE number 4033650 |
Statements
Infinite Horizon Optimization for Finite State Markov Chain (English)
0 references
1987
0 references
infinite horizon control
0 references
finite state Markov chain
0 references
long run average
0 references
overtaking optimality criterion
0 references
stationary optimal strategy
0 references
0.8389574289321899
0 references
0.8300365805625916
0 references
This work is concerned with infinite horizon control of a finite state Markov chain from the point of view of i) the long run average and ii) the overtaking optimality criterion. In the framework of the author, the sets of strategies and admissible controls are larger than in previous work. Moreover, two hypothesis are introduced and summarized here: (A) For every initial state i at time \(k=0\) and a distribution \(p=(p_ 1,...,p_ m)\) with \(p_ i>0\), \(1\leq i\leq n\), there is a control u for which the distribution of the process at time \(k=1\) is p. (B) Any two states i, j communicate after a finite number of steps under the action of a stationary admissible strategy and the set of admissible control values is convex and compact. The main results of the paper are: i) the existence and computation of a stationary optimal strategy with a minimal long run average cost is proved under A or B. ii) Unter B there is a stationary strategy which is overtaking optimal in the class of all stationary strategies. iii) Under A every stationary strategy with a minimal long-run average cost is in fact overtaking optimal for every initial state.
0 references