Discrete time adaptive impulsive control theory

This paper considers both impulsive and adaptive control of discrete time Markov processes. The first part deals with impulsive control with a long run average cost criterion. Under suitable assumptions the optimal impulsive strategies are characterized. In the second part, based on the impulsive control results obtained earlier, the optimal adaptive controller is constructed. With the use of so-called modified maximum likelihood estimation, value consistency is obtained.Feller Markov process * impulsive control * maximum likelihood estimation * martingale stability theorem