The question is: Prove/disprove $||A^2|| \leq ||A||^2$ where A is some nxn matrix.

I've played around with a while few matrices and I'm pretty sure that this is correct but I can't quite figure out how to prove it. My first guess is just define some arbitrary matrix A, and show that for any vector $x$ of size 1, $||A^2 x|| \leq ||Ax||^2$ however this method seems rather involved.

Any tips would be great! :)

Edit: $||A||= \max ||Ax||$, where $|x| = 1$

Edit: Wow this was rather fast. Thanks everyone.

  • $\begingroup$ Knowing about eigenvalues maybe provides a shortcut. Have you done eigenvalues yet? $\endgroup$ – mathreadler Jan 21 '16 at 1:25

Recall that by the definition of $\|A\|$ we have $$\|Ay\| \leq \|A\|\|y\|.$$

For any $x \in \mathbb{R}^n$, using this inequality twice (on $y = Ax$ and on $y = x$), we have $$\|A^2x\| = \|A(Ax)\| \leq \|A\| \|Ax\| \leq \|A\| \left( \|A\| \|x\| \right) = \|A\|^2 \|x\|.$$

Therefore $$\|A^2\| = \sup_{x \neq 0}\dfrac{\|A^2x\|}{\|x\|} \leq \|A\|^2.$$

Note that what the matrix norm is does not matter in the proof, as long as the norm is an induced norm.


It is true. In fact, for all $x \in \mathbb{R}^n$ with $\|x\| = 1$. If $Ax \neq 0$, then by definition, $$\|A^2x\| = \|A(Ax)\| \leq \|Ax\| \|A\|\frac{\|x\|}{\|Ax\|} = \|A\|\|x\| = \|A\|.$$ The above inequality also holds for $x$ such that $Ax = 0$. Thus $$\|A^2\| = \sup_{\|x\| = 1} \|A^2 x \| \leq \|A\|.$$


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.