For finding a shortest path in a network the bidirectional~A* algorithm is a widely known algorithm. An A* instance requires a heuristic estimate, a real-valued function on the set of nodes. %This algorithm distinguishes between the main phase and the postprocessing phase. %As long as the search processes of the two sides do not meet, we are in the main phase. %As soon as a meeting point is obtained, the post-phase is in progress. \\\\ The version of bidirectional~A* that is considered the most appropriate in literature hitherto, uses so-called balanced heuristic estimates. This means that the two estimates of the two directions are in balance, i.e., their sum is a constant value. In this paper, we do not restrict ourselves any longer to balanced heuristics. A generalized version of bidirectional A* is proposed, where the heuristic estimate does not need to be balanced. This new version turns out to be faster than the one with the balanced heuristic.

Additional Metadata
Keywords bidirectional search, road network search, shortest path
Publisher Erasmus School of Economics (ESE)
Persistent URL hdl.handle.net/1765/16100
Citation
Pijls, W.H.L.M., & Post, H.. (2009). Yet another bidirectional algorithm for shortest paths (No. EI 2009-10). Report / Econometric Institute, Erasmus University Rotterdam (pp. 1–9). Erasmus School of Economics (ESE). Retrieved from http://hdl.handle.net/1765/16100