haku: @author Bean, J. C. / yhteensä: 7
viite: 3 / 7
| Tekijä: | Hopp, W. J. Bean, J. C. Smith, R. L. |
| Otsikko: | A new optimality criterion for nonhomogeneous Markov decision processes. |
| Lehti: | Operations Research
1987 : NOV-DEC, VOL. 35:6, p. 875-883 |
| Asiasana: | MARKOV CHAINS DECISION ANALYSIS |
| Kieli: | eng |
| Tiivistelmä: | A framework for solving nonhomogeneous Markov decision process is considered. To solve a problem of this type requires a new, more general definition of an optimal solution, called periodic forecast horizon, and a more robust optimal-solution-seeking technique. The concept of weak ergodicity is used to develop conditions that ensure that periodic forecast horizon optimal strategies exist and are average optimal. The results lead to the existence of forecast horizons in undiscounted as well as discounted problems. |
SCIMA