1
$\begingroup$

I would like to know what is the induced matrix norm of a matrix $A$, when the domain space is equipped with $l_2$ norm and the range space is equipped with $l_\infty$ norm, i.e., what is $\lvert| A \rvert|_{2 \infty}$? The expression for $\lvert| A \rvert|_{\infty 2}$ is given here.

$\endgroup$
1
$\begingroup$

In general, when $X$ is a normed space, the $X\to \ell_\infty$ norm of a matrix $A$ is computed as
$$ \sup_{\|x\|_X\le 1}\|Ax\|_\infty = \sup_{\|x\|_X\le 1} \max_i \left|\sum_{j} a_{ij}x_j\right| = \max_i \sup_{\|x\|_X\le 1} \left|\sum_{j} a_{ij}x_j\right| = \max_i \|A_{i*}\|_{X^*} $$ where $A_{i*}$ is the $i$th row of $A$, and $X^*$ is the dual of $X$. (Since $A_{i*}$ is a map from $X$ to scalars, its $X^*$ norm makes sense.) Simply put, when the target is $\ell_\infty$, the rows (components of the map) can be considered independently of one another.

In the special case $\ell_2\to\ell_\infty$ we end up with $$ \|A\|_{2\infty} = \max_i \sqrt{\sum_j |a_{ij}|^2} $$ which is the maximal Euclidean length of the rows of $A$.

$\endgroup$

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.