DCM_Exp: loss of nodes about nemtropy HOT 2 OPEN

Hi, thanks for the comment.

Sampling from the probability distribution of the maximum entropy models means that the number of edges and the degrees of the nodes are not preserved exactly, but only on average: a single sampled network will have a number of edges very different from the original.

For what regards the nodes, the sampled edgelists do not contain edges relative to nodes with degree 0, which is 100% the case if your nodes had degree 0 in the original network and can (and typically will) happen for nodes with small degree as well. In the sampled network they are isolated nodes, so since you pass only the edgelist to networkx they don't appear in the networkx graph, resulting in an apparent loss of nodes. The "lost" nodes are just isolated in a particular sampling.

I might suggest initializing the nodes of the networkx graph with the complete nodelist of the original graph before giving in input the edgelist.
Hoping this is clear now, I'll close the issue in a few days.

from nemtropy.

Thank you for the clarification!

from nemtropy.