Minimax network location : theory and algorithms

Abstract

For a given network let P and N denote the set of all points and the set of all nodes respectively. Let G and T denote a cyclic network and a tree network respectively and let m denote the number of centers available. The categorization scheme P N/P N/m/G T, where the first and second cells refer to the possible locations of centers and demand generating points respectively, provides for compact identification of a variety of minimax network location problems. This dissertation presents algorithms which efficiently solve all problems in this class--for example, P/P/m/G-for virtually any size of network. Moreover, tree problems can usually be solved manually. Methodologically, the tree-based results are graph-theoretic while the general case, formulated in a mathematical programming framework, leads to a highly efficient strategy for a class of massive generalized set covering problems.