Gradient-based adaptative sampling in NeuralPDE

Hey everyone,

I am testing NeuralPDE to simulate hydrodynamic shocks and I would like to try the residual/gradient adaptative sampling of data, as described in the article in the following link : Physics-informed neural networks with residual/gradient-based adaptive sampling methods for solving partial differential equations with sharp solutions | Applied Mathematics and Mechanics.

Is it possible to do this with the current version of NeuralPDE, or to adapt it to make it work?

Thank you for your answers!

That algorithm isn’t currently in there but it could be added. Open an issue.

1 Like