Paper published in Phys. Rev. E
The latest contribution of the Networks section is available in the June issue of Physical Review E.
In the paper, entitled “Finite size analysis of the detectability limit of the stochastic block model”, we propose a comprehensive information theoretical framework to study, well, the detectability limit of the stochastic block model, in finite networks. Considering the networks as finite allow us to visualize the parameter space of the model more completely, and this naturally leads to notion of symmetries (what transformations of the parameters maintain the difficulty of the detectability problem?), and conservation (what are the surface of constant detectability in the parameter space?).
The paper contains a thorough case study in a simple case (planted partition model), but the tools apply to the general SBM.
Links to the the full text, software, and preprints are also available on the Publications page.