R/likelihood2.R

In qpsy/clime2: Constrained L1-minimization for Inverse (covariance) Matrix Estimation

likelihood2 <- function(Sigma, Omega) {
  ot <- as.numeric(unlist(determinant(Omega)))
  if (ot[2]<=0) warning("Precision matrix estimate is not positive definite!")
  tmp <- (sum(diag(Sigma%*%Omega))  - ot[1] - dim(Omega)[1])
  if(is.finite(tmp)) {
    return(tmp)
  } else {
    return(Inf)
  }
}