In an optimization book I am following, the induced norm is defined as the maximum of the norms of the vectors $Ax$ where the vector $x$ runs over the set of all vectors with unit norm.

Now, it says that because of the continuity of a vector norm for each matrix $A$ the maximum $\max_{\left|\left|x\right|\right|=1}\left|\left|Ax\right|\right|$ is attainable, that is, a vector $x_0$ exists such that $\left|\left|x_0\right|\right|=1$ and $\left|\left|Ax_0\right|\right|=\left|\left|A\right|\right|$. The book also says that this follows from the theorem of Weierstrass. I have not understood this part.

Also, it is not clear to me why for any $m \times n$ matrix $A$ and any $x \in \Bbb R^n$ the following inequality gets satisfied :

$$\left|\left|Ax\right|\right|_{(m)} \leq \left|\left|A\right|\right|\,\left|\left|x\right|\right|_{(n)}$$ assuming a above definition of norm.


The maximum is attained by compactness of the set ${x\in V| \,||x||=1}$ and continuity of the norm and linear maps in finite dimension. (See the extreme value theorem on wikipedia) This is why you can replace $\sup$ (found in most textbooks) by $\max$.


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.