Open xubonan opened 1 year ago
If you are referring to equation (4.1)-(4.5) in the paper, then you may see that U,V are used in every hidden layer, as shown in equation (4.4)
I misunderstood that formula. This modified FNN have greatly improved the accuracy of PINN.
In the publication, only the last layer uses U and V, but in the code it seems that every layers used U and V