Closed d-schindler closed 1 year ago
I'm confused about the null model, in https://github.com/michaelschaub/generalizedLouvain/blob/1d3e73b30ec2ce8edac44a300f5a767050048230/MATLAB/quality_functions/louvain_signedLap.m#L49 it seems to depend on the scale.
We should also create an example notebook for signed graphs.
I implemented signed_combinatorial
now with 0 as the null model, as suggested by Michael.
I also added the signed network from the Harari data as an example and applied both signed_modularity
and signed_combinatorial
constructors. They seem to work and the clusters seem to correspond to the ones in the PRE paper according to the labels, but maybe @arnaudon can double-check?
I think this is ready to merge after your checks, @arnaudon
@arnaudon , is there anything else you want me to do here?
This PR is to implement a constructor based on the signed combinatorial Laplacian, see: https://github.com/michaelschaub/generalizedLouvain/blob/master/MATLAB/quality_functions/louvain_signedLap.m