A* vs. Dijkstra

[geoHeil]

Why did you choose Dijkstra over A*?

[smattheis]

Sorry for the delay. For short routing distances as we have it in map matching even for 3 min sampling intervals, we encountered more overhead for the calculation of the heuristic distance function than a benefit from "guiding" the route search. Further, our Dijkstra implementation uses a single-source-multiple-target approach to save time/workload when we calculate some number of similar routes between matching candidates. This is not straight-forward with A* because for each target we must maintain a different value of the heuristic function per edge.