Computing Diameters in Slim Graphs

Abstract

With the use of large graphs with n vertices and m edges, the current approach for computing the diameter is not efficient. We have investigated a special graph class, namely slim graphs. Slim graphs are graphs whose diameter is at least some fixed fraction of the number of vertices. This constraint allows us to prove structural features in these special graphs. Using these features, we have developed three algorithms which are asymptotically superior to diameter computation in the general case. We present the following three algorithms, for a fixed 0 < k < 1/2: a (1 − k)- approximation algorithm of the diameter in O(n+m) time; a deterministic algorithm which computes the diameter in O(n2) time and a Monte Carlo algorithm which also computes the diameter in O(n2) time.