OmniPolis: Traffic is central to modern city life.

In many large urban settings, regular daily, weekly and seasonal cyclic variations are occasionally upset by transient and unusually harsh traffic jams, which can involve a large multitude of vehicles. Understanding the mechanisms governing the behavior enables accurate modeling of both typical and extraordinary conditions, leading the way to the development of innovative services in numerous fields. Omnipolis provides a platform on which models can be run so as that they may be tailored to the analysis of specific aspects of the task at hand. Currently a rush hour model has been developed which is now being ported to a GP-GPU parallel processing environment with the aim of capturing the dynamics of 2 to 4 million vehicles. Sub-sampling in both time and space should provide accurate descriptions on the order of a minute and within a distance of a few car lengths.