Simulate data from known precision matrices for multiple groups

Generates synthetic data from K groups with shared and differential graph structure. Useful for evaluating the performance of ssjgl.

Usage

simulate_ssjgl_data(
  K = 2,
  p = 20,
  n = 100,
  graph_type = c("band", "random", "hub", "scale-free"),
  edge_prob = 0.1,
  perturb_prob = 0.05,
  signal = 0.3,
  seed = NULL
)

Arguments

K

Integer number of groups. Default 2.

p

Integer number of variables. Default 20.

n

Integer number of observations per group (scalar applied to all groups, or vector of length K). Default 100.

graph_type

Character string specifying the base graph type. One of "band" (AR-2 banded), "random" (Erdos-Renyi), "hub" (star topology), or "scale-free" (Barabasi-Albert). Default "band".

edge_prob

Numeric edge probability for "random" graph type. Default 0.1.

perturb_prob

Numeric probability of perturbing (flipping) each edge in groups 2, ..., K relative to group 1. Default 0.05.

signal

Numeric signal strength for off-diagonal entries of the precision matrix. Default 0.3.

seed

Integer random seed for reproducibility. Default NULL.

Value

A list with elements:

data_list

List of K data matrices, each n_k x p.

Omega_list

List of K true precision matrices (p x p).

adj_list

List of K binary adjacency matrices (p x p, 0 diagonal).

Sigma_list

List of K true covariance matrices (p x p).

K

Number of groups.

p

Number of variables.

n

Vector of sample sizes.

graph_type

The graph type used.

Examples

sim <- simulate_ssjgl_data(K = 2, p = 10, n = 50, graph_type = "band", seed = 42)
str(sim, max.level = 1)
#> List of 8
#>  $ data_list :List of 2
#>  $ Omega_list:List of 2
#>  $ adj_list  :List of 2
#>  $ Sigma_list:List of 2
#>  $ K         : num 2
#>  $ p         : num 10
#>  $ n         : num [1:2] 50 50
#>  $ graph_type: chr "band"