I'm trying to solve this question here. Thank you in advance for your help.

Prove that $k(A) \geq \rho(A)/\min |\lambda|$ and that $k(A) \geq \rho(A) \rho(A^{-1}) $

We assume the matrices $A$ and $A^{-1}$ exist. Any size $A$ matrix.

I know $k(A) = ||A||\cdot ||A^{-1}||$ and $\rho(A) = \max_{\lambda\in\sigma(A)} |\lambda|=\lim_{n\to\infty}\|A^n\|^{1/n}$ with $\rho$ being the spectral radius and $k(A)$ being the condition number of the matrix.

I'm not sure what exactly I need to set it up in order to get to both needed results.

  • $\begingroup$ Your hint doesn't look right. $\endgroup$ – user10354138 Sep 29 '18 at 6:43
  • $\begingroup$ @Brahadeesh It's any size matrix $A$. Just a general matrix. The same with $\lambda$ and $\rho$. It's using the definitions of them to obtain the inequalities. $\endgroup$ – MelMarieHA Sep 29 '18 at 7:19
  • $\begingroup$ @user10354138 $\rho$ is the spectral radius, and I double checked my hint/note and that is correct. $\endgroup$ – MelMarieHA Sep 29 '18 at 7:20
  • $\begingroup$ your hint is $k(A)=\rho(A)\rho(A^{-1})$ is definitely not correct, otherwise you wouldn't be asked to prove $k(A)\geq\rho(A)\rho(A^{-1})$. $\endgroup$ – user10354138 Sep 29 '18 at 7:37
  • $\begingroup$ @user10354138 Very true! I'll remove it then! Does it make more sense? $\endgroup$ – MelMarieHA Sep 29 '18 at 7:53

We will prove $k(A) \geq \rho(A)\rho(A^{-1}) $

We know, by definition, let $ \lambda $ be the eigenvalue for which $ \rho(A) = \max_{\lambda\in\sigma(A)} |\lambda| $. We let $x$ be an eigenvector, $ ||x||_{v} = 1 $.

$\Rightarrow \rho(A) = \max_{\lambda\in\sigma(A)} |\lambda| = ||\lambda x||_{v}$

$\Rightarrow ||\lambda x||_{v} \leq ||Ax||_{v} \leq ||A||\cdot||x||_{v} = ||A|| $

$\Rightarrow \rho(A) \leq ||A||$

We apply the same with $A^{-1}$:

$\Rightarrow \rho(A^{-1}) = \max_{\lambda\in\sigma(A^{-1})} |\lambda| = ||\lambda x||_{v}$

$\Rightarrow ||\lambda x||_{v} \leq ||A^{-1}x||_{v} \leq ||A^{-1}||\cdot||x||_{v} = ||A^{-1}|| $

$\Rightarrow \rho(A^{-1}) \leq ||A^{-1}||$

We combine both inequalities and get $k(A) \geq \rho(A)\rho(A^{-1})$.

To get $ k(A) \geq \rho(A)/\min_{\lambda\in\sigma(A)} |\lambda| $, we note $ k(A) \geq \rho(A)\rho(A^{-1})$. We know $\rho(A) = \max_{\lambda\in\sigma(A)} |\lambda|$ and the same with $\rho(A^{-1})$. We note that the eigenvalues of $A^{-1}$ are the reciprocals of those of $A$.

$\Rightarrow k(A)\geq \rho(A)\rho(A^{-1})$

$\Rightarrow k(A) \geq \rho(A)/\min_{\lambda\in\sigma(A)}|\lambda|$.


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.