Solving 1D Diffusion Equation with PINN: A Practice of Physics-Informed Neural Networks from Forward to Inverse Problems

Introduction / Main Post: Solving 1D Diffusion Equation with PINN: A Practice of Physics-Informed Neural Networks from Forward to Inverse Problems

A project using Physics-Informed Neural Networks (PINN) to solve both forward and inverse problems of the 1D diffusion equation, achieving an estimation error of 0.17% for the diffusion coefficient under sparse observation data with only 1% noise.