Leo Torres

I was about to implement this an realize I don't know how the density is defined :sweat_smile: @maximelucas is the density at group size `d` equal to the number of...

> For the global density, I would've just summed over d. 2^n is not the total number of potential hyperedges, is it? Yes, it is the total number of possible...

> density = ncells/(nrows*ncols) This is the density of the incidence matrix, interesting. I guess this is more an indicator of how big the hyperedges are rather than an indicator...

https://www.wolframalpha.com/input?i=sum+for+k+%3D+0+to+n+of+%28n+choose+k%29

> So I'd say 2^n -1? That sounds like a good idea. > Also, when having multiple edges, it tells you how much their they intersect in terms of nodes,...

I guess in a way this density tells you how many stubs there are in the bipartite representation of the hypergraph divided by the total possible number of stubs *without...

So this is what we're looking at now: ```python3 xgi.density(H) # -> number of edges / (2^n - 1) xgi.density(H, max_order=k) # -> number of edges up to order k...

That's a good point. I guess I was thinking of a case where the two hypergraphs have the same number of nodes (in which case one would have isolated nodes)....

Are there any updates for supporting labels starting with anything other than `eq:`? For example, labels starting with `eqn:` are also very common. Otherwise, could you please kindly point to...

Thanks for your prompt response! For now my workaround is to call `G.remove_nodes_from(...)` before running the simulation by giving `rho` and not giving `initial_recovereds`. This might not work for everybody...