Combining physics-informed graph neural network and finite difference for solving forward and inverse spatiotemporal PDEs.
Hao ZhangLongxiang JiangXinkun ChuYong WenLuxiong LiYonghao XiaoLiyuan WangPublished in: CoRR (2024)
Keyphrases
- finite difference
- partial differential equations
- numerical solution
- neural network
- level set
- numerical scheme
- eikonal equation
- diffusion equation
- finite element
- image denoising
- hamilton jacobi
- image processing
- anisotropic diffusion
- numerical analysis
- differential equations
- exact solution
- image enhancement
- pattern recognition
- numerical methods
- diffusion process
- curve evolution
- multiscale
- boundary conditions
- energy functional
- level set method
- high order
- scale space
- image segmentation
- semi implicit