Babak Javani - Ph.D. student, School of Civil Engineering, University of Tehran, Tehran
Abbas Babazadeh - Assistant professor, School of Civil Engineering, University of Tehran

Recently, gradient projection (GP) has been extended as efficient algorithms for solving the traffic assignment problem (TAP). In this paper, two distinguished GP methods for TAP are investigated. Theyare an algorithm to apply Bertsekas`s GP method and an algorithm to apply Rosen's GP method. They confront the problem in the same way, but with various descent directions and step sizes. The efficiency of the algorithms is compared using large-scale test networks. Algorithms are developed ina same platform and compared based on the computational efforts for reaching given accuracies. The results show that the algorithm using Bertsekas`s GP converges more rapidly than another using Rosen's GP.

Traffic Assignment, Gradient Projection, Descent Direction Method, Algorithm