SpinFlow: A Physics-Informed Spin Field Framework for Traffic Phase Inference and Transition Detection

ArXi:2605.23306v1 Announce Type: cross Active traffic management (ATM) is frequently hindered by traditional macroscopic models and rigid empirical thresholds that fail to capture metastable phase precursors, resulting in delayed, reactive interventions. To address this, we propose SpinFlow, a physics-informed spin-field framework unifying Kerner's three-phase theory with statistical physics for continuous macroscopic traffic phase inference.