Dan Schult

Please look at [the previous question's answer](https://github.com/networkx/networkx/issues/7173#issuecomment-1862152218) as linked by @Schefflera-Arboricola . You **are** getting the classical algebraic adjacency matrix. But you aren't specifying the node order for the matrix....

So, if I understand right, we now have two related configuration settings: `backend_priority` and `backend_fallback`. The priority says what order to try backends when a NXGraph is the input. The...

Thanks for that description -- still digesting it. What is the difference between the backend in "fallback" and having that backend in the last spot of the "priority"?

> Thus, we have: > 1. If backend=backend_name is given then use that backend (and fail if can't) > 2. If you pass in a backend graph to an algorithm,...

Sorry for multiple posts instead of one... > nx.config.backend_priority = ["spam"] > nx.pagerank(G_spam) # Uses "spam" backend > nx.pagerank(G_eggs) # Tries to convert and use "spam" backend, then would try...

I haven't hardly even looked at this yet. But I wonder if that `or 1` near the end of `faux_inf` should be a 3. (the whole max expression gets multiplied...

Yes, a PR would help us understand and test the ideas discussed above, and especially your suggestion. Thanks!

For random graph generation, there is usually a distribution that is expected. For example, the degree distribution should follow a specified distribution. That is, for example, specific to the case...

I can make the recently suggested changes above if you don't have time atm. Didn't mean to dump a bunch of stuff on you. What do you think?

Thanks @tillahoffmann (and @rossbar reviews)! create_using in random_graphs.py and duplication.py is merged!