Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I've been trying to understand matrix norms (full disclosure: school assignment, not looking for answers, just clarity!), and how they follow from vector norms - been awhile since I did much linear algebra, so i'm struggling a bit with the notation, in particular I'm solving in the general case that for matrix A (and any nonzero vector x)

$$\frac{||Ax||_1}{||x||_1} \le C$$

For C = maximum column sum of A

The part I don't think I understand is what ${||Ax||_1}$ actually means, relative to matrix A. Could someone help me to understand a bit better the notation, and to apply the matrix norm?

share|improve this question
Note that $Ax$ is a vector, to be evaluated wrt I norm (sup?). If the question is about matrix norms though, you are likely asked to solve the inequality for the maximum non-zero vector $x$. –  gnometorule Jan 30 '13 at 2:57
So $Ax$ would be, in a general case that max non-zero vector? or one of the vectors in A? I just don't really understand how $Ax$ and $x$ are related, I guess. –  Stephen Young Jan 30 '13 at 3:07
What I meant is that the usual definition of a matrix norm is $\sup_{x \neq 0} \frac{||Ax||}{||x||} = \sup_{x: ||x|| = 1} ||Ax||$. It's the same $x$ in numerator and denominator, as you take the $\sup$ of the fraction. So write it down, forgetting about the $\sup$, for some such $x$ (just calculate the vector $Ax$); do some thinking for this fixed $x$; then take the supremum over all such $x$. It's probably easier to use the 2nd, equivalent definition ($||x|| = 1$). –  gnometorule Jan 30 '13 at 3:20
add comment

1 Answer

up vote 2 down vote accepted

Since it looks like you are interested in the matrix norm which is induced by a particular vector norm (take a look here) , there is a nice geometric interpretation for $\|A\|$:

$$\|A\|:=\max_{\|x\|=1}\|Ax\|,$$ that is, if you start with any unit vector, compute its image under the transformation $A$, and compute the norm of that image (which is a vector, so its computed via a vector norm), then $\|A\|$ is the largest such resulting value. In other words, $\|A\|$ is the maximum "stretch" that $A$ "does" to a unit vector.

As an example, suppose $A=\begin{bmatrix}1 & 2\\0 & 3\end{bmatrix}$, so $A:\mathbb{R^2}\to\mathbb{R}^2$, and we will consider $\mathbb{R}^2$ with the 2-norm. Then the matrix norm induced by the (vector) 2-norm described above is summarized graphically with this figure:

Mathematica graphics

Note the unit vectors on the left and then some representative images under $A$. The length of the longest such image is $\|A\|$ (induced by this vector norm).

share|improve this answer
awesome - this is a much better description than from my class notes, thanks! –  Stephen Young Jan 30 '13 at 3:32
add comment

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.