Deal with having two copies of u

[lm2612]

Issue

We currently have two tensors: u for the main domain and uufor the domain plus the halo to include the boundary conditions. Ideally we would like to only have one copy of u.

Possible solutions

• Maybe with padding='same' in the nn.Conv?
• and also with padding_mode='replicate' ?
• https://pytorch.org/docs/stable/generated/torch.nn.Conv2d.html