Welcome to the MixNet software website. MixNet stands for Erdös-Renyi Mixture for Networks, a probabilistic model for random graphs. This model is a clustering model based on the hypothesis that nodes of real networks are spread into hidden classes (or colors) which show specific connectivity patterns. We propose various algorithms to estimate the model's parameters as well as probabilities of class membership for each node. We also propose a statistical criterion to select the number of classes.

A complete description at http://stat.genopole.cnrs.fr/logiciels/mixnet

Two software package are available:

  • MixNet, the Unix/Mac/Linux software package, with a command line interface and multiple options
  • MixeR, the R package, with less options but additionnal methods and graphical tools