Max T.

Any leads on how to address this? I seem to be suffering from the same or a similar issue. My error ``` Expression: select(dbConn, relatedEntries, sqlCondition, oneEntry.id) [1] dbConn: DbConn...

Hm, someone more familiar with the code base needs to help me with the `numpy` import :/ Edit: Contrary to my expectations, there was no measurable performance deficit with simply...

> I'm -1 on replacing the existing implementation. I can definitely see why. Is there precedence in networkx for implementing the same functionality but faster with worse-quality results? If so,...

> There is precedent for a "method" kwarg that allows users to choose between implementations I went ahead and implemented this for `approximation.maximum_independent_set` and `approximation.max_clique` as I really like that...

> OTOH, if this algorithm really only makes sense for graphs that are so large that the added memory footprint of a copy is prohibitive for standard applications then that...

Alright, thank you for the advice! The last open question for me is non-determinism. Testing multiple times on the "genes" graph, I get different results (between size 71 and 75)...

> using a dict instead of a set Thanks! This does work, but it definitely hurts readability (see #3912) and complicates some parts that would otherwise be simple set subtractions....

I wasn't aware of `.keys()` but as far as I can tell from searching the internet, there's no simple way of actually removing the keys from the dict. Including handling...

> `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? > The best...

The benefit of having the sets be ordered (until the end of the algorithm) is that the algorithm is deterministic which wouldn't be the case for unordered sets. It's not...