Line 6 in the pseudocode indicates that the algorithm involves n 2 1 iterations. This is based on the fact that a shortest path in an n-node graph can theoretically have n 2 1 hops. However, in actual practice, the maximum number of hops in any shortest path in a graph is far less than n. This observation has been confirmed through extensive software simulations [14]. Therefore the iterations can be terminated as soon the algorithm converges resulting in a significant degree of speed-up. For graphs with no negative-weight links, the convergence of the algorithm can be confirmed if the status of none of the nodes is modified during an iteration. Other single-source shortest-path algorithms such as Dijkstra’s algorithm do not permit this property of graphs to be exploited to improve performance.