0
$\begingroup$

Consider a matrix $P\in \mathbb{R}^{n\times n}$ such that at most $m<n$ elements of each column are non zero and $P$ is symmetric. I would like to find the sufficient condition(s) such that $P^{-1}$ is positive definite. Given the positions of non zero elements in $P$ what are their values such that $P$ is positive definite.

$\endgroup$
  • 1
    $\begingroup$ do you need $P^{-1}$ to be sparse? Otherwise it is trivial... $\endgroup$ – Dima Pasechnik Mar 31 '16 at 20:06
  • $\begingroup$ I need specific elements of $P$ to be non zero, Jacobi Rotations on a positive definite diagonal matrix might work as user251257 said. $\endgroup$ – Cauchy Apr 1 '16 at 1:54
2
$\begingroup$

If $P$ is an invertible real symmetric matrix, $P^{-1}$ is positive definite iff $P$ is positive definite. There are many equivalent conditions to positive definiteness.

$\endgroup$
  • $\begingroup$ That is true but not sufficient for the sparsity . So How do I generate a positive definite sparse symmetric matrix? $\endgroup$ – Cauchy Mar 31 '16 at 6:43
  • $\begingroup$ @Cauchy: Google is your friend. Apply random Jacobi Rotations on a positive definite diagonal matrix... $\endgroup$ – user251257 Mar 31 '16 at 14:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.