Refactor Leiden #3327
Refactor Leiden #3327
Conversation
…iden, need to add more to make it complete
…ce_e_by_dst_key to support reduce op on tuple data types
…er during development
…e and construct decision graph
…to leiden-on-22-12
… leiden-aggregated graph
…s, fixes mis implementation
|
|
||
| # FIXME: the networkx.Graph type used in the type annotation for | ||
| # induced_subgraph() is specified using a string literal to avoid depending on |
There was a problem hiding this comment.
This comment needs to be updated since it still mentions induce_subgraph(). Maybe just rewrite it so it doesn't mention an API at all if it's going to be copy-pasted elsewhere.
| # Temporarily suppress warnings till networkX fixes deprecation warnings | ||
| # (Using or importing the ABCs from 'collections' instead of from | ||
| # 'collections.abc' is deprecated, and in 3.8 it will stop working) for |
There was a problem hiding this comment.
I think we should start removing this Nx warning supression since we should know about deprecations until Nx is completely removed from our tests.
| # Test data | ||
| # ============================================================================= | ||
|
|
||
| _test_data = { |
There was a problem hiding this comment.
I'm assuming black did this, but can the test data be reformatted to have >1 number per line? This is taking up hundreds of lines with no improvement to readability. You should be able to do this:
Black reformats entire files in place. It doesn’t reformat lines that end with # fmt: skip or blocks that start with # fmt: off and end with # fmt: on.
|
/merge |
A CAPI implementation of Leiden is available and this PR:
Implement PLC Leiden by leveraging the CAPI.
Refactor the python SG implementation of Leiden by leveraging the PLC implementation.
Add a python MG implementation of Leiden with tests.
closes #2487
closes #2489
closes #2490
closes #2491