Simpler Solutions by Means of Larger Formulations in Eynamic

Programming and Supply Chain Management

ABSTRACT

In dynamic programming, one can always use a formulation that employs a state space which is much larger than necessary. We show that such non-parsimonious formulations are sometimes powerful. We provide an illustration in the context of a battery management problem for a communication satellite. We then examine the inventory control problem for multiechelon uncapacitated supply chains. We show that this technique results in a very simple derivation of classical results (optimality of basestock policies), efficient computational algorithms, and several new results.