Maxime Lucas

I think it would! As I mentioned in the related issue, just wondering about the call syntax with many dots.

I've never looked for a definition but that's what I was thinking to do for size `d`, yes. For the global density, I would've just summed over `d`. `2^n` is...

Ok yea my bad. Although `2^n` counts the empty set which I assume we don't want. So I'd say `2^n -1`? (It also counts all singletons edges which I always...

I did it wrong 👍🏻

That makes sense! I would add also an option to not count singleton edges, like `count_singletons=False`, in which case, I would divide by `2^n-1 - n`. Otherwise the density of...

> I don't think this is true. If you have exactly two edges of size `a` and `b`, on `n` nodes, then this density is always equal to `(a+b)/2n`, regardless...

Great summary! I'm just unsure between having `incidence` as an argument, and having a separate function `density_incidence()`. In the first case it might get complicated to explain all the cases...

Partially solved by #204. `density_incidence` not implemented yet.

Thanks Tim! Yea my first thought is exactly what you wrote: if you need to reuse the positions, it's best to pre-compute them outside of `draw` and then pass them...

Good question. In my mind it could be Explanation (bottom left). For example, I see this [page](https://pandas.pydata.org/docs/user_guide/reshaping.html#reshaping) as something that explains some functionalities of the Reference. The Reference tells you...