Closed fcdimitr closed 2 years ago
Sorry for the long delay in answering!
Unfortunately, within the framework of this package that particular check cannot be performed. In particular, that can only be easily done for CPM and Modularity, but not for some of the other quality functions. This package is aimed at being able to optimise arbitrary quality functions (i.e. it only requires a diff_move
function essentially). This limitation is also explained in the documentation, please see https://leidenalg.readthedocs.io/en/stable/reference.html#optimiser for more details.
If the particular guarantee of subpartition γ-density is important to you, you can either use the Java implementation available from https://github.com/CWTSLeiden/networkanalysis, or you can use the implementation in igraph
itself, see https://igraph.org/python/api/latest/igraph.Graph.html#community_leiden. These implementations follow the original paper exactly, but are less flexible than this package.
I'll close the issue now, but if you have any follow-up questions, let me know and feel free to re-open the issue.
Could anyone point us to the location in the source code for checking "well-connected" communities? The well-connected property is supposedly a signature feature of the Leiden algorithm.
The checks are described at lines 34 and 37 of algorithm A.2 in the supplementary material of the paper "From Louvain to Leiden" by van Traag, Waltman, and van Eck, 2019.
Many thanks in advance.