A two-phase shortest path algorithm for networks with node coordinates

This paper presents a new algorithm for finding the shortest path from a source to a single sink in a network, in which the location in the plane of each node is known. The algorithm consists of two phases. In the first phase a heuristic solution to the shortest path problem is found. In the second phase the upper bound provided by the heuristic solution is utilized in a modification of a standard shortest path algorithm. Estimates based on computational tests show that on average the computation time of the presented algorithm is on the order of 40–60% of the computation time required if the information on node locations is not utilized.