Dear Tiago, Thanks for your reply very much. Two follow-up questions would be: if I have some partition prior, say, for example, bipartiteness, at this step state = state.copy(bs=bs, sampling=True) do I need change this step to: state_copy = state.copy(bs=bs, sampling=True, state_args = state.state_args) to prevent that nodes will not be moved to the group that is not allowed to be grouped before performing equilibration step?
From the documentation, my answer is no . But if I do not pass this parameter, from my experiments, after the equilibration the results do not make sense. Since at one side of the bipartite network it has much larger groups even at the second highest possible level.
Additionally, can I call minimize_nested_blockmodel_dl() again and pass bs = state_copy.get_bs() in the function to get the final state instead of writing a callback as I mentioned in the previous email? Best, Terry -- Sent from: http://main-discussion-list-for-the-graph-tool-project.982480.n3.nabble.com/