查看文章

Traffic equilibrium problem, also known as the traffic assignment problem, is the core of many important transportation models. The traditional Frank-Wolfe (FW) algorithm has been criticized for its slow convergence speed to approach high precision level. Recently, various traffic equilibrium algorithms have been developed in literature. Basically, they can be classified into three main categories, ie link-based algorithms, path-based algorithms and origin-based algorithms, depending on the solution variables adopted in the problem. These new algorithms are expected to fulfill the requirement of the implementations of more advanced and comprehensive transportation models and analysis, which desire rapid convergence speed and highly precise equilibrium solutions.

Nowadays, distributed computing is getting popular in many fields, due to the increasing power of the computer hardware. Multiple machines, machines with multiple processors, or processor with multiple cores become more widely available and more affordable in recent years. Therefore, it is also desired to examine the possibility and efficiency of incorporating the power of distributed computing with the newly developed algorithms in order to further improve the performance of solving traffic equilibrium problems.