Tarik Sahin

@lululxvi For instance, `model` argument in `losses` function has already `net`. But I could not see that how can I calculate NN output from there.

For instance, for `torch` as backend, there is `forward` function in `fnn` module. What about `tensorflow.compat_v1`?

I guess you have to create your custom network, this is not supported.

@lululxvi Actually the parameter does not have to be in the PDE, I solved an example where the unknown parameter is in the set of a BC.

@lululxvi We generated a repository (namely [pinns_for_comp_mech](https://github.com/imcs-compsim/pinns_for_comp_mech)) based on DeepXDE with a focus on computational mechanics. Here's an example of the [Lame problem](https://github.com/imcs-compsim/pinns_for_comp_mech/blob/main/elasticity_2d/linear_elasticity/inverse_lame/Lame_quarter_inverse_mixed_variable.py) in elasticity. We define inlet pressure (`pressure_inlet_predicted`)...

Here is the outcome. Pressure (defined in BC), Youngs modulus (defined in PDE) and Poissons (defined in PDE) ratio are set as variables and calculated by solving the inverse problem....

Yes, I believe the Pytorch implementation has a bug that it does not improve loss accuracy using LBFGS so switch to tf. You will get NaN at the end as...

I had the same problem as well. Compiling the model twice is not allowed as I have seen.

you can use the output scale. ```python from deepxde import backend as bkd def output_transform(x,y): return bkd.sigmoid(y) # I assume that output is 1d net.apply_output_transform(output_transform) ```