Drawing Graphs to Speed Up Shortest-Path Computations

We consider the problem of (repeatedly) computing singlesource single-target shortest paths in large, sparse graphs. Previous investigations have shown the practical usefulness of geometric speed-up techniques that guarantee the correctness of the result for shortest-path computations. However, such speed-up techniques utilize a layout of the graph which typically comes from geographic information. This paper examines the question how geometric speed-up techniques can be used in case there is no layout given. We present an extensive computational study analyzing the usefulness of methods from graph drawing as foundation for such techniques. It turns out that using appropriate layout algorithms, a significant speed-up can be achieved.