In this paper we discuss two basic geometric techniques that can be used to speed up certain types of dynamic programs. We first present the algorithms in a general form, and then we show how these techniques can be applied to the economic lot-sizing problem and extensions. Furthermore, it is illustrated that the geometric techniques can be used to give elegant and insightful proofs of structural results, like Wagner and Whitin's planning horizon theorem. Finally, we present results of computational experiments in which new algorithms for the economic lot-sizing problem are compared with each other, as well as with other algorithms from the literature.
Keywords: Dynamic programming, computational analysis, lot-sizing, inventory