Ultra-fast Traffic Nowcasting and Control via Differentiable Agent-based Simulation

ArXi:2603.25068v1 Announce Type: cross Traffic digital twins, which inform policymakers of effective interventions based on large-scale, high-fidelity computational models calibrated to real-world traffic, hold promise for addressing societal challenges in our rapidly urbanizing world. However, conventional fine-grained traffic simulations are non-differentiable and typically rely on inefficient gradient-free optimization, making calibration for real-world applications computationally infeasible.