Maxime Lucas

I think the way to do this is rather to have generators of lattice models that also output the positions of the nodes. Those positions can then be passed to...

Exactly. For example networkx has a `with_positions` argument: `nx.triangular_lattice_graph(m, n, with_positions=True)` https://networkx.org/documentation/stable/reference/generated/networkx.generators.lattice.triangular_lattice_graph.html

Networkx does not have one, as far as I know. So you need to manually collect all the positions, and pass them to the `draw` function with the `pos` argument....

Thanks for unearthing this. After PR #247, `cleanup()` does not raise an error anymore, but I feel its effect is not fully correct (or should be discussed). It can remove...

I have a vague memory that it was supposed to be similar to the node speed-up, and that it was just that nobody implemented it. @nwlandry @iaciac ?

If I'm not wrong, we have a function that computes faces. What do you mean by "method for links"? I don't think we have cofaces.

If I remember correctly, we decided to simply add an explanation in the docstring for this, right?

Great, thanks Thomas! I think the `all_stats` could also be useful.

Is this a bug though? `H.edges` contains edges IDs, so your example is not so surprising. Although a user might write ```python In [14]: H1.edges == H2.edges ``` and expect...

What do we want if the two have the same hyperedges (based on members) but different IDs?