Kernel-Adaptive Discretization Improves Neural Network Solutions for Fredholm Integral Equations
A systematic benchmarking framework demonstrates that matching discretization strategies to kernel regularity dramatically improves physics-informed neural network accuracy and stability ...


