Hybrid Quantum Physics-informed Neural Network: Towards Efficient Learning of High-speed Flows

This study assesses the potential use of hybrid quantum physics-informed neural network (HQPINN) to model high-speed flows as an alternative to classical PINN and quantum neural network options. The model integrates parameterized quantum circuit (PQC) with classical neural network in parallel as input to a physics-based optimization. For problems with harmonic solutions, the HQPINN exhibits superior accuracy and trainability compared to both classical and quantum models at low parameter costs. For transonic flows, the hybrid approach yields modest results and additionally suffers from poor trainability if the quantum layer were under-parameterized. Our results highlight inherent limitations in deploying quantum neural networks in PINN applications, and potential use of hybrid architectures as a general tool in problems where the nature of the solution is not known a-priori.