3
$\begingroup$

Assume matrices $A$, $B$, and $C$ are of same dimensions, does $\|A\|_2\leq \|B\|_2$ imply $\|CA\|_2\leq \|CB\|_2$ or $\|AC\|_2\leq \|BC\|_2$?

$\|A\|_2$ denotes by $\lambda_{max}\sqrt{A^TA}$, and $\lambda_{max}$ is the largest eigenvalue.

$\endgroup$
4
$\begingroup$

No. Consider $A = \begin{bmatrix}1&0\\0&0\end{bmatrix}$, $B = \begin{bmatrix}0&0\\0&2\end{bmatrix}$, $C = \begin{bmatrix}1&0\\0&0\end{bmatrix}$. Clearly, $\|A\|_2 = 1 \le 2 - \|B\|_2$.

However, since $AC = CA = \begin{bmatrix}1&0\\0&0\end{bmatrix}$ and $BC = CB = \begin{bmatrix}0&0\\0&0\end{bmatrix}$, we have $\|AC\|_2 = 1 > 0 = \|BC\|_2$ and $\|CA\|_2 = 1 > 0 = \|CB\|_2$.

Therefore, $\|A\|_2 \le \|B\|_2$ does not imply either $\|CA\|_2 \le \|CB\|_2$ or $\|AC\|_2 \le \|BC\|_2$.

$\endgroup$
4
  • $\begingroup$ thank you so much for your answer, does it imply any other matrix norm inequality related to them? $\endgroup$
    – wayne
    Dec 24 '15 at 9:29
  • $\begingroup$ and under what conditions, the above conclusions may hold $\endgroup$
    – wayne
    Dec 24 '15 at 9:32
  • $\begingroup$ I'm not sure if there are any reasonable set of constraints on $A,B,C$ which gives us $\|A\|_2 \le \|B\|_2 \implies \|CA\|_2 \le \|CB\|_2$. $\endgroup$
    – JimmyK4542
    Dec 24 '15 at 9:39
  • $\begingroup$ okay, thank you so much. $\endgroup$
    – wayne
    Dec 24 '15 at 9:40
4
$\begingroup$

Very general counterexample: let be $P$, $Q$ the projections on two orthogonal (nontrivial) subspaces. Wlog we can suppose $0<\|P\|\le\|Q\|$. Now, $$\|PP\|=\|P\|\not\le 0 =\|PQ\|,\qquad \|PP\|=\|P\|\not\le 0 =\|QP\|.$$

$\endgroup$

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.