A Link-Based Algorithm for Stochastic User Equilibrium with Bounded Choice and Local Detour Threshold
This study develops a novel solution method for stochastic user equilibrium (SUE) under the bounded choice model with local detour threshold (BCM-LDT SUE; Rasmussen et al., 2024), which restricts the choice set in line with observed route choice behavior. The model assigns zero probability to paths whose total cost exceeds a bound or whose local detour exceeds a threshold; thus, the path choice set varies with link costs. The proposed algorithm updates link flows while implicitly enumerating the path choice set via a depth-first search (DFS) pruned with the BCM-LDT bounds. This distinguishes it from the original algorithm, which updates path flows, and enables quasi-Newton-type acceleration. Since paths are never stored, the method can handle choice sets totaling on the order of $10^7$ paths. On a personal computer, it reached equilibrium in a few to about fifteen minutes, not only for the Anaheim network under the original settings but also for larger networks such as Winnipeg and Barcelona.