THE AUGMENTED LAGRANGIAN METHOD FOR SOLVING DYNAMIC NETWORK TRAFFIC ASSIGNMENT MODELS IN DISCRETE-TIME

We develop and test an augmented Lagrangian method for solving dynamic traffic assignment models formulated as optimal control problems. Our presentation is in terms of the discrete time, system optimal traffic assignment problem. However, the basic ideas presented here are readi ly applied to continuous time models and to other behavioral assumptio ns regarding traffic assignment which may be expressed as optimal cont rol problems. The proposed algorithm obviates the need for path enumer ation and exploits the natural decomposition of the traffic assignment problem by time period which is possible when an optimal control form ulation is employed.