General method to compute log-likelihood for ghype models.

logl(
  object,
  xi = NULL,
  omega = NULL,
  directed = NULL,
  selfloops = NULL,
  adj = NULL,
  multinomial = NULL,
  ...
)

# S3 method for ghype
logl(
  object,
  xi = NULL,
  omega = NULL,
  directed = NULL,
  selfloops = NULL,
  adj = NULL,
  multinomial = NULL,
  ...
)

# S3 method for matrix
logl(
  object,
  xi = NULL,
  omega = NULL,
  directed = NULL,
  selfloops = NULL,
  adj = NULL,
  multinomial = NULL,
  ...
)

Arguments

object

either an adjacency matrix or ghype model If a ghype model is passed, then `xi`, `omega`, `directed`, `selfloops` are ignored If an adjacency matrix is passed, then `adj` is ignored

xi

matrix, combinatorial matrix to build ghype model, considered only if object is an adjacency matrix

omega

matrix, propensity matrix to build ghype model, considered only if object is an adjacency matrix

directed

boolean, is ghype model directed? considered only if object is an adjacency matrix

selfloops

boolean, has ghype model selfloops? considered only if object is an adjacency matrix

adj

optional matrix, adjacency matrix of which to compute log-likelihood, considered only if object is ghype model If adj is not passed, and object is a ghype model, the log-likelihood is computed for the original adjacency matrix stored in object.

multinomial

optional boolean. Force multinomial approximation? If not chosen, multinomial chosen for large graphs.

...

additional parameters passed to and from internal methods

Value

loglikelihood value

Methods (by class)

  • logl(ghype): Computes log-likelihood for ghype models from model object

  • logl(matrix): Computes log-likelihood for ghype models from adjacency.

Examples

data('adj_karate')
model <- scm(adj_karate, FALSE, FALSE)
logl(object = model)
#> [1] -434.5449
new_adj <- adj_karate
new_adj[3,4] <- 10
logl(object=model, adj=new_adj)
#> [1] -445.8136