Nick Papior

Agreed, so it should return `jo + no * index_of_supercell`, no?

> I don't know what you agreed to, but to me it doesn't look like we agree 😅 `geom.close(ia)` by default returns `ja` which is a supercell index. That was...

But I also think it should be here as well, why not? From the supercell coupling it is very easy to convert to orbital couplings, and even getting the isc...

well, the entirety of `sisl` is based on the `nsc` concept, and it may be used to gauge the range of couplings. So I don't necessarily agree, while I can...

I can partially see where you are coming from, but then you suggest an underlying change of how sisl shows interactions. I am not against this, but I think it...

But isn't your argument that `close` returns also in supercell indices, which is a bit counter-intuitive? I think it would be nice if we could streamline this, no?

You are probably right. Then I think we should split the return values? It is more clearer that `isc[:, 0]` refers to offsets along the first lattice vector.

Custom class is fine for me ;)

> Cool, I think it's the best approach seeing that there might be different users interested in different things. And also it makes sure that the user knows what they...

Indeed, however, I am not sure whether it should be a default of both merged? Consider a case where a user has done `pbc = (T, T, F)` but then...