5
$\begingroup$

I'm not sure if the set of symmetric positive-definite matrices is open in the set of symmetric matrices. I'm almost certain that it is not open in the larger set of n by n matrices. So far most of the proofs I've seen don't use symmetry, so I'm not sure if they are correct. Any hints? I've been trying to use that the set of positive-definite matrices is path connected, but I can't come up with a solution. Thanks in advance.

$\endgroup$
  • 4
    $\begingroup$ What does positive-definite mean if you're not a symmetric matrix? Anyway, yes - this is equivalent to saying that $x^T A x$ is positive for all nonzero $x$, which is equivalent to saying that $x_i^T Ax_i$ is positive for the basis vectors $x_i$. This last condition, then, is clearly an open condition. $\endgroup$ – user98602 Aug 1 '16 at 3:31
  • $\begingroup$ @Mike Miller, don't we need to show that any matrix within some distance of A is also positive-definite? $\endgroup$ – Saul Alinsky Aug 1 '16 at 3:38
  • 1
    $\begingroup$ Yes, Sylvester's criterion implies that any matrix $A'$ which is sufficiently close to a positive-definite matrix $A$ is positive-definite (of course, provided $A'$ is symmetric). $\endgroup$ – Alex Ravsky Aug 1 '16 at 3:44
  • 3
    $\begingroup$ Here's a more direct proof. Sylvester's criterion mentioned by @AlexRavsky shows that the set of positive-definite matrices is the intersection of the set of symmetric matrices and the set $\{A\mid\text{det}A_k>0,k=1,2,\ldots,n\}$, which is the intersection of inverse images of a open set under the continuous determinant map. $\endgroup$ – Cave Johnson Aug 1 '16 at 4:38
  • $\begingroup$ @MikeMiller why is it enough to check on a basis? $\endgroup$ – peter Dec 21 '18 at 21:43

Your Answer

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

Browse other questions tagged or ask your own question.