A Lower Bound to the Complexity of Euclidean and Rectilinear Matching Algorithms

Abstract The worst-case time complexity of any exact algorithm for the Euclidean or rectilinear minimum-weight perfect matching problem, which takes as input the list of coordinates of n points in R k, is shown to be bounded below by the infimum of the worst-case time complexities of all algorithms which sort n real numbers. This result also applies to any heuristic algorithm for which the worst-case ratio of the weight of the approximate matching it produces to the weight of the optimal matching only depends upon n.