bipartite_matching_pseudolikelihood: Bipartite to unipartite matching via penalized...

In jesusdaniel/rBipartiteUnipartiteMatch: Graph Matching between Bipartite and Unipartite Networks

Description

This function performs graph matching between a bipartite and a unipartite graphs that are assumed to share a common set of vertices. The method performs an alignment of the non-zero parameters of a graphical model for the bipartite graph and the nonzero entries of the unipartite adjacency matrix via pseudolikelihood maximization.

Usage

bipartite_matching_pseudolikelihood(
  A,
  B,
  Q_true = NULL,
  family = "binomial",
  lambdas = 10^(-2:0),
  MAX_ITER = 20,
  verbose = FALSE,
  seeds = NULL,
  gamma = 0
)

Arguments

A	Adjacency matrix of the unipartite graph.
B	Bipartite graph incidence matrix, where rows correspond to the vertices in common with the unipartite graph. Entries of B need to be binary in 0, 1.
Q_true	A permutation matrix with the true solution, if known. When the true solution is unknown, the default is NULL.
lambdas	Penalty values for the optimization algorithm.
MAX_ITER	Maximum number of iterations.
verbose	If TRUE, the method prints an output after each iteration.
seeds	If some vertices have known correspondence, a vector containing the indexes of these vertices can be passed through this parameter, and the corresponding rows of A and B are assumed to be aligned. The algorithm will then match the remaining vertices. The default is NULL if no seeds are available.
gamma	Additional penalty for all variables. A positive small constant is needed when the sample size

jesusdaniel/rBipartiteUnipartiteMatch documentation built on Dec. 20, 2021, 11:06 p.m.