CTQW-GNN uses Continuous-Time Quantum Walks to solve over-smoothing

CTQW-GNN employs a unitary propagator e^(-iHt) to prevent the low-pass filtering bias common in standard GNNs. This method preserves feature norms and prevents Dirichlet energy decay, addressing both heterophily issues and over-smoothing in deep architectures.