On the approximability of some maximum spanning tree problems

Abstract

We study the approximability of some problems which aim at finding spanning trees in undirected graphs which maximize, instead of minimizing, a single objective function representing a form of benefit or usefulness of the tree. We prove that the problem of finding a spanning tree which maximizes the number of paths which connect pairs of vertices and pass through a common arc can be polynomially approximated within a factor of 1.5. It is known that this problem can be solved exactly in polynomial time if the graph is 2-connected [14]; we extend this result to graphs having at most two articulation points. We leave open whether in the general case the problem admits a polynomial time approximation scheme or is MAX-SNP hard and therefore not polynomially approximable whithin any constant 1+∈, ∈>0, unless P=NP. On the other hand we show that the problems of finding a spanning tree which has maximum diameter, or maximum height with respect to a specified root, or maximum sum of the distances between all pairs of vertices, or maximum sum of the distances from a specified root to all remaining vertices, are not polynomially approximable within any constant factor, unless P=NP. The same result holds for the problem of finding a lineal spanning tree with maximum height, and this solves a problem which was left open in [6].

Partially supported by the M.U.R.S.T. 40% Research Project “Algoritmi, Modelli di Calcolo e Strutture Informative”.