Perform an agglomerative merge step — collapse

Collapse n.merges blocks together via agglomerative merging. For each block, n.moves merges are proposed using the same move proposal algorithm used for the MCMC sweeps. The move proposals are ranked and n.merges moves are performed after accounting for duplicates. After the merge, n.sweeps MCMC sweeps are performed to stabilize the resulting partition.

Usage

collapse_step(
  graph,
  partition,
  degree_correction = FALSE,
  n.merges = 1,
  n.moves = 10,
  n.sweeps = 0,
  eps = 0.1,
  beta = 1,
  verbose = TRUE
)

Arguments

graph: An igraph graph.
partition: An vector of integers giving the block partition of nodes.
degree_correction: Whether to use degree correction.
n.merges: Number of block merges in collapse step.
n.moves: Number of merge trials per block.
n.sweeps: Number of sweeps after block merge.
eps: (optional) A number controlling the randomness of moves.
beta: (optional) A number controlling the greediness of moves.
verbose: Whether to print verbose output to console.