Michael Ketcha rotation work with Joshua Vogelstein
let's try communicating thru this, rather than email, and see how it goes?
The Dirilecht random variable should be constant among each of the instances generated from the model, correct? Its meant to represent internal deviations from the exact SBM, i,e, it shouldn't be treated as noise that will be random for each generated instance. In terms of implementation this will mean I use the same "P" matrix for each of the "M" adjacency matrices