Skip to contents

Nearest positive definite stabilisation

Source: R/covariance_stabilisation.R
stabilize_nearest_pd.Rd

Stabilises a covariance matrix by projecting it to the nearest positive definite matrix using Matrix::nearPD().

Usage

stabilize_nearest_pd(Sigma, keep_diag = TRUE)

Arguments

Sigma

A square symmetric covariance matrix.

keep_diag

Logical. If TRUE, keeps original diagonal where possible.

Value

A symmetric positive definite covariance matrix.

Examples

Sigma <- matrix(c(1, 1, 1, 1), nrow = 2)
eigen(stabilize_nearest_pd(Sigma))$values
#> [1] 2e+00 2e-08