Dan Schult

I took the liberty of pushing up some tests. I think this is now ready. @rossbar can you take a look at the tests?

Thanks @mjschwenne!! Sorry to hear you will be out of action here for a long time. Best wishes on all your future projects! We'll be watching your career from afar....

This one occurs because there are no edges in the graph... It shouldn't matter how many nodes. So adding a check with `nx.is_empty(G)` and raising a [NetworkXPointlessConcept](https://networkx.org/documentation/stable/reference/exceptions.html) sounds like a...

Yes -- let's go with the "copy and morph" variant. But I would not do a complete copy... Just copy the adjacency (so data you don't need is left behind)....

For set subtraction, you would use `d = d.keys() - to_remove` which does create a new dict instead of acting in-place. The best way I know to get an in-place...

> d = d.keys() - to_remove which does create a new dict > > Wouldn't this return a set though, losing the property of being ordered for subsequent operations? Yes,...

Looks like `p2g.py` is not included in `__init__.py`. It also doesn't appear in the docs. I don't think this is a `lazy_import` problem. It is just an import problem. The...

Thanks for this comment @fernandodiazdiaz Do you have insight into when one might use the laplacian variant `A.sum() + A`? Or the normalized laplacians like: `I - (1 / np.sqrt(A.sum())...

The normalized cases are analogous to the unnormalized cases. But do we want this function to be able to provide the normalized cases too by choosing a variant? Or is...

> With the current implementation, we do not overwrite the input graph’s data in-place I see that those tests check are supposed to check for in-place changes. But this example...