Wrong prior inclusion probabilities saved in the arguments list.

[sekulovskin]

When estimating a model with edge_prior = "Beta-Bernoulli" | "Stochastic-Block", the saved prior inclusion probabilities are incorrect. This issue does not happen when edge_prior = "Bernoulli".

Example code:

m_bern <- bgm(Wenchuan, edge_prior = "Bernoulli")
m_bern$arguments$inclusion_probability
m_sbm <- bgm(Wenchuan, edge_prior = "Stochastic-Block")
m_sbm$arguments$inclusion_probability

[MaartenMarsman]

I believe this is fixed now in the adaMala branch. The main issue was that we passed the matrix of inclusion probabilities by reference, and with the Beta-Bernoulli and the Stochastic-Block priors these values were updated. This matrix is not passed by reference any longer, so they should no longer be updated in the R environment.