nx-cugraph: add CC for undirected graphs to fix k-truss #3965
Conversation
| @number_connected_components._can_run | ||
| def _(G): | ||
| # NetworkX <= 3.2.1 does not check directedness for us | ||
| try: | ||
| return not G.is_directed() | ||
| except Exception: | ||
| return False |
There was a problem hiding this comment.
I fixed this upstream here: networkx/networkx#7074
There was a problem hiding this comment.
changes LGTM. Should there be a test for k_truss that passes
G=nx.Graph([(0,1), (1,2), (2,0), (3,4), (4,5), (5,3)])
and asserts that NotImplementedError is raised, which we then xfail when k_truss is updated?
OTOH, maybe that's overkill for (hopefully) temporary code that I'm assuming was at least manually tested? I'm fine either way.
rlratzel
added
improvement
Thanks for providing a test case, but now we actually have the opposite problem: now we don't test the main branches of k_truss! All test graphs apparently have more than 1 connected components. |
Karate club has only 1 CC; maybe add a quick test for that too to make sure we at least exercise |
Excellent suggestion, although we'd need #3954 to test with karate club (b/c it has string node values). Test added, and code is exercised 👍 I'm looking forward to having more graph generators to enable more comparison testing of algorithms. |
| def _groupby( | ||
| groups: cp.ndarray, values: cp.ndarray, groups_are_canonical: bool = False | ||
| ) -> dict[int, cp.ndarray]: |
There was a problem hiding this comment.
I'm really happy with this groupby implementation btw. Using prepend with cp.diff is a little slow compared to cp.diff w/o it, but this isn't where the bulk of the time is spent.
|
/merge |
Fixes #3963 and add
connected_components,is_connected,node_connected_component, andnumber_connected_components.Also updated
_groupbyto handle groups that are not consecutive integers starting with 0.Also,
plc.weakly_connected_componentsdoes not handle isolated nodes well, and I needed to handle this at the Python layer as was done in #3897