Different Weight Matrices — create_weight_matrix • CVN

This function generates different weight matrices for two external covariates, \(W\). There are several types:

full: All graphs are fully connected with weight 1
glasso: All graphs are disconnected with weight 0. This mimicks the GLASSO, where each graph is estimated independently
grid: A weight matrix for a \(k \times l\) grid
uniform-random: Fully-connected, but the entries are drawn from a uniform distribution

Usage

create_weight_matrix(
  type = c("full", "glasso", "grid", "uniform-random"),
  k,
  l,
  plot = FALSE
)

Arguments

type: The type of weight matrix
k: Number of categories in the first external covariate
l: Number of categories in the second external covariate
plot: If TRUE, the weight matrix is plotted (default: FALSE)

Value

Weight matrix

Examples

W <- create_weight_matrix(type="grid", k=3, l=2)