transGFM (version 1.0.2)

identify: Identify factor decomposition via SVD

Description

Identify factor decomposition via SVD

Usage

identify(M, r)

Value

List with F (row factors) and B (column factors)

Arguments

M

Matrix to decompose

r

Number of factors

Examples

# Generate Poisson data
set.seed(123)
n0 <- 50; p0 <- 50; r <- 2
F_true <- matrix(runif(n0 * r, min = -2, max = 2), n0, r)
B_true <- matrix(runif(p0 * r, min = -2, max = 2), p0, r)
F_true <- F_true / sqrt(r)
B_true <- B_true / sqrt(r)
M_true <- F_true %*% t(B_true)

# Decompose using identify
result <- identify(M_true, r = 2)
F_hat <- result$F
B_hat <- result$B

# Check reconstruction
M_reconstructed <- F_hat %*% t(B_hat)
print(max(abs(M_reconstructed - M_true)))  # Should be very small