A Novel Hierarchy of Quantum Kernel Networks on Smoothed Particle Hydrodynamics
This study proposed the hierarchy of quantum kernel networks by combing multi quantum networks with smoothed particle hydrodynamics (SPH). The Lagrangian quantum network model was further developed based on an improved quantum multilayer perceptron (QMLP). A sequential hybrid quantum-classical framework was constructed to ensure robust particle gradient-based optimization and mitigate barren plateaus for computational particle dynamics. This approach combines smoothing kernels with quantum learning, establishing a novel quantum intelligent particle paradigm. The framework was validated through some benchmarks on multifarious quantum neural networks, static multi-level vortex reconstructions and transient scalar advective transports. Numerical results show that while elementary quantum circuits struggle with generalization in unstructured domains, the hybrid crossed-QMLP matches the fitting accuracy of classical SPH in quantum optimized space. Despite current limitations in computational efficiency and hardware implementation, this work paves the way for a new investigation on quantum-particle approach by mapping unstructured Lagrangian particle topologies into integrated quantum networks.