Adaptive Quantum PINNs Enhance Fluid Dynamics Solutions

A new hybrid quantum-classical framework has been introduced to significantly improve the accuracy and efficiency of Quantum Physics-Informed Neural Networks (QPINNs) for solving nonlinear partial differential equations (PDEs). Traditional PINNs often struggle with high-dimensional or multiscale systems, and QPINNs face limitations not just from quantum circuit expressivity but also from optimization bottlenecks. The proposed method enhances QPINNs by dynamically prioritizing collocation points in regions with high PDE residuals or steep solution gradients, thereby mitigating the inherent spectral bias of conventional PINNs. Additionally, a trainable loss-weighting scheme is incorporated to balance the contributions from physics residuals, boundary conditions, and data fidelity during the training process. By integrating these strategies with quantum computing techniques, including variational quantum circuits and quantum gradient estimation, the framework achieved at least a 60% improvement in solution accuracy for benchmark fluid flows and reaction-diffusion systems under specific conditions. The research emphasizes that merely increasing model expressivity is insufficient for complex PDEs; structural optimization limitations of classical PINNs also need to be addressed in QPINNs.