The field of complex system modeling is moving towards a more integrated approach, combining machine learning techniques with physical principles to improve prediction accuracy and efficiency. This is particularly evident in areas such as energy management, power systems, and environmental modeling, where the incorporation of physical laws and constraints can enhance the performance of machine learning models. Notably, graph neural networks and attention mechanisms are being increasingly used to capture complex relationships and spatial heterogeneity in various domains. Some noteworthy papers in this regard include: The paper on Few-Shot Learning by Explicit Physics Integration, which introduces a Local-Global Convolutional Neural Network approach for modeling groundwater heat transport. The paper on Heterogeneous Graph Neural Networks for Short-term State Forecasting in Power Systems, which proposes the use of Heterogeneous Graph Attention Networks for modeling multi-domain, multi-rate power system state forecasting.