In stochastic (dynamic) traffic assignment (DTA), route choice (RC) formulations with implicit route set require less memory than their explicit counterparts. To compute turn probabilities, however, a topological ordering of the nodes is needed, which negatively impacts the stability of convergence and consistency of equilibrium. This paper formulates a general framework for the RC module of the efficient stochastic DTA (ESDTA) without topological ordering; we demonstrate how its full fixed route yields smooth and consistent convergence. The theoretical framework requires a series of computational modifications to ensure that numerical errors remain limited and to make the formulation more efficient for use in practice. The resulting ESDTA's computation time to reach equilibrium scales linearly with respect to all relevant complexity dimensions (network size, demand zones, time resolution, and congestion level).

中文翻译：

---

具有稳定路由集的随机动态流量分配的一致迭代算法

在随机（动态）交通分配 (DTA) 中，具有隐式路由集的路由选择 (RC) 公式比其显式对应物需要更少的内存。然而，为了计算转弯概率，需要对节点进行拓扑排序，这会对收敛的稳定性和平衡的一致性产生负面影响。本文为没有拓扑排序的高效随机DTA（ESDTA）的RC模块制定了一个通用框架；我们展示了其完整的固定路线如何产生平滑且一致的收敛。理论框架需要一系列计算修改，以确保数值误差保持有限，并使公式在实践中更有效。由此产生的 ESDTA'