Further Results on Fastest Paths in STDNs

Dijkstra's algorithm may be used to compute fastest paths in transportation networks represented as weighted graphs where edge weights denote travel times between locations. If travel times are both stochastic and time dependent, however, eecient dynamic programming methods cannot be used to compute fastest paths. Wellman et al. introduced a variant of priority-rst search that is capable of nding optimal paths in stochastic time-dependent networks (STDNs). This paper presents new computational results on the performance of Wellman's algorithm on a class of randomly generated STDNs. These results suggest that on a reasonable class of problems Wellman's algorithm eeciently computes optimal paths.