GPU-accelerated, multi-GPU traffic microsimulator for HPC and AI research at city scale: powers ICML 2026 deep-RL ride-hailing dispatch, MoE routing policies, and closed-loop LLM-driven fleet optimization over networks with hundreds of thousands of nodes and millions of agents.
simulator reinforcement-learning hpc cuda deep-reinforcement-learning high-performance-computing gpu-computing mobility multi-gpu icml traffic-simulation rl-environment ride-hailing large-scale-simulation ai-for-science
-
Updated
Jun 25, 2026 - Jupyter Notebook