0
$\begingroup$

I am stuck on this problem. I have no idea how to proceed, so any help would be welcome.

Let $A$ be symmetric positive definite matrix $n\times n$ prove that : $$\operatorname{trace}(A) * \operatorname{trace}(A^{-1}) \ge n^2$$

$\endgroup$
  • $\begingroup$ Hint: What is the relationship between trace and eigenvalues? What can you say about eigenvalues of a positive definite matrix? $\endgroup$ – Christian Sykes Sep 23 '18 at 18:45
  • $\begingroup$ They are positive ? $\endgroup$ – AdnanM91 Sep 23 '18 at 18:49
  • $\begingroup$ That answers the second question. What about the first? $\endgroup$ – Christian Sykes Sep 23 '18 at 18:50
  • $\begingroup$ Sum of eigenvalues is same as trace $\endgroup$ – AdnanM91 Sep 23 '18 at 18:51
  • $\begingroup$ Exactly. Finally, what are the eigenvalues of the inverse? You can put those three facts together to get what you need. $\endgroup$ – Christian Sykes Sep 23 '18 at 18:53
1
$\begingroup$

Hint: Let $D$ be a $n\times n$ diagonal matrix with eigenvalues $0<\lambda_1 \leq \lambda_2 \leq \cdots \leq \lambda_n$. Then we see that \begin{align} \operatorname{tr}(D)\operatorname{tr}(D^{-1}) =&\ \left(\lambda_1+\cdots +\lambda_n \right)\left(\frac{1}{\lambda_1}+\ldots+\frac{1}{\lambda_n} \right)\geq\ n^2 \end{align} where the last inequality follows from the AM-HM inequality.

$\endgroup$

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.