Abstract
A class of controlled semi-Markov jump processes is defined in this paper. Conditions are found which guarantee that satisfaction of the dynamic programming equations for stochastic control is necessary and sufficient for the minimization of the expected discounted cost of a controlled semi-Markov process over a random time. Terminal costs are included, and the controls are allowed to depend on both the present state of the process and the length of time the process has been in that state.

This publication has 13 references indexed in Scilit: