Avoid negative solutions

[MariaPaulaDelPopolo]

Hi! Thanks for the nice example! I am new to PINNs and was checking the results and saw that the minimum value is a negative value. I wanted to do basically the same example but to model the diffusion of a certain substance, so I cannot have negative concentrations. I thought about using an activation function after the final layer (a sigmoid or a softplus), but in any case the output is really inaccurate after this implementation. I also tried some other activation functions for the hidden layers, but it was not helpful. I was wondering if you had any idea on how to solve this issue. Thanks in advance!

[NicholasDewberryOfficial]

I'm not sure what you meant by this, could you give some more detail? From what I can tell, it accurately maps the output from 6^{-3x} into a workable 3d graph using a neural network.

There shouldn't be any negative numbers as the equation doesn't cross the X axis (check second image)