Tom Gustafsson

`Basis._detect_tensor_order`

Sorry for the delay in responding - I was busy with teaching tasks this autumn. I think it makes sense. Are you still willing to push this forward and proceed...

Here is what I've been testing: ```python from skfem import * from skfem.experimental.autodiff import * from skfem.experimental.autodiff.helpers import * import jax.numpy as np import numpy as onp import matplotlib.pyplot as...

Now here is something to consider: - How should `basis.N` be picked? (Mortaring vs. DG is different?) - Make sure mesh types can be different (for mortaring) - How about...

Fancy pics from the code above. One of the earlier iterations:  The final iteration: 

move to #1143

Here are the relevant bits: ```python np.save(dire + 'mt', m.t, allow_pickle=False) t = np.load(dire + 'mt.npy', allow_pickle=False) ```

I don't think this should work. Such matrix cannot be invertible. Do you think we should add explicit zeros to the diagonal of all inputs? What is the use case?