[dmh2000]

"The trick was to use a Cholesky factor of the covariance matrix as the dependent variable in an equivalent Riccati equation. "

I'm going to use that as my next pickup line