Our latest article has been published in Computer Physics Communications:

"Learnable activation functions in physics-informed neural networks for solving partial differential equations"

The paper investigates the use of learnable activation functions to address key challenges in Physics-Informed Neural Networks (PINNs), such as spectral bias and convergence instability. Various network architectures and PDE types are studied using Neural Tangent Kernel and Hessian analyses.

🔗 Article link: https://doi.org/10.1016/j.cpc.2025.109139

Related Github repository can be found at  https://github.com/afrah/pinn_learnable_activation