# Cholesky decomposition and permutation matrix

Dear all, Let's assume that I have a symmetric matrix $\Sigma$ and a permutation matrix $A$. Is there a relationship between the Cholesky decompositions of $\Sigma$ and of $A^T \Sigma A$ ? Many thanks, Samuel

-

The Cholesky decomposition allows you to write (for real symmetric positive definite matrix $\Sigma$) $$\Sigma = U^T U$$ with $U$ an upper triangular matrix. Therefore, $$A^T \Sigma A = A^T U^T U A = B^T B$$ with $B= UA$. Note however that $B$ is (in general) not upper triangular anymore such that $B^T B$ is not the standard Cholesky decomposition (but of course very similar to a Cholesky decomposition).
@J. M.: of course you are right. However depending on what you want to do afterwards, the decomposition $B^T B$ might be enough (and you don't need the additional property that B is upper triagonal). –  Fabian May 5 '11 at 9:41
You guys are right. Let's say $\Sigma=U^T U$ and $A^T \Sigma A=V^T V$ where $U$ and $V$ are upper triangular. I wanted to know if there was any realtionship between $U$ and $V$. –  S4M May 5 '11 at 10:13