Minimize Function

[kotserge]

I do not quite understand your implementation/wrap-around of graph-tools minimize_blockmodel_dl in wsbm.minimize function.

Graph tool does not expect the parameters you are providing, like B_min=1, B_max=30. These should be provided in the dictionaries state_args and multilevel_mcmc_args. It also does not expect the parameter overlap at all.

Also you provide an argument pure_w, which does nothing. Why?

In its current state this implementation breaks!

A possible fix to the current implementation would be:

def minimize(graph: gt.Graph, distribution: str = "discrete-binomial", deg_corr: bool = False):
    return gt.minimize_blockmodel_dl(
        graph,
        state_args=dict(deg_corr=deg_corr, recs=[s_gt.ep.orig_weight], rec_types=[distribution],  overlap=overlap),
        multilevel_mcmc_args=dict(B_min=1, B_max=30, niter=100, entropy_args=dict(adjacency=False, degree_dl=False))
    )

Could you maybe elaborate on why you choose to do it this way and if it was intentional, why you provide these arguments, as this would be very helpful.

[kicasta]

I'll check it and come back to you as soon as I have time

[kicasta]

At the moment of this implementation graph-tool expected those parameters. Thanks for pointing out, that the function signature changed. Code will be updated accordingly.