Location Games and Bounds for Median Problems

Abstract

We consider a two-person zero-sum game in which the maximizer selects a point in a given bounded planar region, the minimizer selects K points in that region,.and the payoff is the distance from the maximizer's location to the minimizer's location closest to it. In a variant of this game, the maximizer has the privilege of restricting the game to any subset of the given region. We evaluate/approximate the values (and the saddle point strategies) of these games for K = 1 as well as for K + , thus obtaining tight upper bounds (and worst possible demand distributions) for K-median problems.