Optimal control of Markovian jump processes with different information structures
FacultiesFakultät für Mathematik und Wirtschaftswissenschaften
LicenseStandard (Fassung vom 03.05.2003)
We consider a stochastic control problem where only groups of states of a Markovian jump process and not the complete state process is observable. In particular the Hidden-Markov-model and the Bayes-model are included. This model under incomplete information is transformed into an equivalent one with complete information with the help of the filter technique. An explicite construction of the processes and the filter is given. Then we propose two solution techniques. First, a generalized verification technique with the Hamilton-Jacobi-Bellman-equation. Second, a formulation of an equivalent time-discrete Markovian decision process. All results are illustrated for a parallel queueing model.
Subject HeadingsMarkov-Sprungprozess [GND]
Hidden Markov models [LCSH]
Jump processes [LCSH]
Stochastic control theory [LCSH]