Dynamic programming and the maximum principle for control of piecewise deterministic Markov processes
Date
2010-04-07T21:23:14Z
Authors
Ye, J. J.
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
This paper studies optimal control of piecewise deterministic Markov processes. A stochastic maximum principle expressed in terms of adjoints given by deterministic differential inclusions is formulated. The relationship between dynamic programming optimality conditions and the stochastic maximum principle is given. Conditions are given under which the maximum principle can be stated in the normal form, the value function appeared in the maximum principle is eliminated, and some asymptotic transversality conditions hold in the case where the boundary hitting time is infinite. Implications of the maximum principle to computations are also discussed.
Description
Keywords
optimal control, piecewise deterministic Markov process, dynamic programming, maximum principle, value function, asymptotic transversality conditions