A Hybrid Neural Network and Finite Difference Method for Solving Stokes Equations with Singular Forces on an Embedded Interface
The proposed hybrid method combines the expressive power of neural networks and the convergence of finite difference schemes to efficiently solve Stokes equations with singular forces on an embedded interface in regular domains.