Optimal policies for constrained average-cost Markov decision processes

Juan González-Hernández ^[1] ; César E. Villarreal ^[2]
1. [1] Universidad Nacional Autónoma de México
  
  Universidad Nacional Autónoma de México
  
  México
2. [2] Universidad Autónoma de Nuevo León
  
  Universidad Autónoma de Nuevo León
  
  México
Localización: Top, ISSN-e 1863-8279, ISSN 1134-5764, Vol. 19, Nº. 1, 2011, págs. 107-120
Idioma: inglés
Enlaces
- Texto completo (pdf)
Resumen
- We give mild conditions for the existence of optimal solutions for a Markov decision problem with average cost, under m constraints of the same kind, in Borel actions and states spaces. Moreover, there is an optimal policy that is a convex combination of at most m+1 deterministic policies.