Hi Tiago,
After minimising a LayeredBlockState with overlap = True, I extract the majority block assignment of each vertex in each layer from the edge blocks property map. I do this because I want partitions that potentially vary by layers, but I don’t want the layer-specific membership of each vertex to be mixed.
By creating a new edge block property map using the majority blocks, I can initiate a new LayeredBlockState with overlap = True for which I can compute the entropy. However, as I am not anymore allowing for within-layer mixed membership, I wonder if this practice renders the overlapping entropy function invalid? Can I still use this entropy function for model selection?
Thanks,
Arttu