The RATS Software Forum

Dear Tom,

I am trying to do a LDL' cholesky decomposition of a covariance matrix in RATS.

As you have mentioned in one of the previous posts that the L matrix can be computed by doing the Cholesky factor of the covariance matrix and dividing each column by its diagonal element.
But I am unable to get the diagonal D matrix for this. is this the same as diagonal of the covariance matrix?

Please help me on this.

Thanks

No. It's the square of the diagonal elements of the Cholesky factor. (It's only the diagonal element of the input matrix for the first row).

compute chol=%decomp(sigma)
compute d=%diag(%xdiag(chol).^2)
compute l=%ddivide(chol,%xdiag(chol))