eval_dcsbm_like: Log likelihood of a DCSBM (fast with poi = TRUE)
In aaamini/nett: Network Analysis and Community Detection

eval_dcsbm_like

R Documentation

Log likelihood of a DCSBM (fast with poi = TRUE)

Compute the log likelihood of a DCSBM, using estimated parameters B, theta based on the given label vector

eval_dcsbm_like(A, z, poi = TRUE, eps = 1e-06)

`A`	adjacency matrix
`z`	label vector
`poi`	whether to use Poisson version of likelihood
`eps`	truncation threshold for the Bernoulli likelihood, used when parameter phat is close to 1 or 0.

The log likelihood is calculated by

\ell(\hat B,\hat θ, \hat π, \hat z \mid A) = ∑_i \log \hat π_{z_i} + ∑_{i < j} φ(A_{ij};\hat θ_i \hat θ_j \hat B_{\hat{z}_i \hat{z}_j} )

where \hat B, \hat θ is calculated by estim_dcsbm, \hat{π}_k is the proportion of nodes in community k.

log likelihood of a DCSBM