A practical PINN framework for multi-scale problems with multi-magnitude loss terms

Yong Wang, Yanzhong Yao, Jiawei Guo +1

For multi-scale problems, the conventional physics-informed neural networks (PINNs) face some challenges in obtaining available predictions. In this paper, based on PINNs, we propo…