search

thunil / Deep-Flow-Prediction

A framework for fluid flow (Reynolds-averaged Navier Stokes) predictions with deep learning
Apache License 2.0
286 stars 117 forks source link

Computation of v_norm wrong? #13

Closed SamiBidier closed 2 years ago

SamiBidier commented 2 years ago

Hi, pls correct me if i'm wrong, but I think there is a small error in the data normalisation, where the norm of v is computed with the pressure values instead of v1.

Would be glad if you could verify. Thanks

https://github.com/thunil/Deep-Flow-Prediction/blob/c3123d1bac6e79fca0d3bba5bcc32cd08bd16464/train/dataset.py#L117

thunil commented 2 years ago

Good question, but the code is correct. The inputs are [in-x, in-y, mask], and hence indices 0 and 1 are used for computing v_norm.

SamiBidier commented 2 years ago

thanks! understood, my mistake

thunil commented 2 years ago

No worries, thanks for checking! It's easy to get wrong, it's actually explained incorrectly here at the moment: http://physicsbaseddeeplearning.org/supervised-airfoils.html (will also be fixed soon).