FinNet: Solving Time-Independent Differential Equations with Finite Difference Neural Network
This work addresses a specific issue in computational physics for researchers dealing with PDEs, though it is incremental as it builds on existing deep learning approaches.
The paper tackled the problem of solving time-independent differential equations with few boundary constraints by introducing FinNet, a method that incorporates finite difference into deep learning, achieving low error rates and outperforming PINNs in some cases.
Deep learning approaches for partial differential equations (PDEs) have received much attention in recent years due to their mesh-freeness and computational efficiency. However, most of the works so far have concentrated on time-dependent nonlinear differential equations. In this work, we analyze potential issues with the well-known Physic Informed Neural Network for differential equations with little constraints on the boundary (i.e., the constraints are only on a few points). This analysis motivates us to introduce a novel technique called FinNet, for solving differential equations by incorporating finite difference into deep learning. Even though we use a mesh during training, the prediction phase is mesh-free. We illustrate the effectiveness of our method through experiments on solving various equations, which shows that FinNet can solve PDEs with low error rates and may work even when PINNs cannot.