2
$\begingroup$

Suppose that matrix $A$ is strictly diagonally dominant, show that $$\|A^{-1}\|_{\infty}\leq\left[\min_i\left(|a_{ii}|-\left|\sum_{\substack{j\neq i}} a_{ij}\right|\right)\right]^{-1}.$$

$\endgroup$
1
$\begingroup$

The bound is not true. Consider $$ A=\left[\begin{array}{rrr}4&1&-2\\-1&5&3\\-1&0&3\end{array}\right]. $$ Then $$ \|A^{-1}\|_\infty=\frac{31}{50}=0.64, \quad \min_i\left(|a_{ii}|-\left|\sum_{j\neq i}a_{ij}\right|\right)=2, \quad \text{but $0.64\not\leq 1/2$.} $$

Something similar is true, however. We show that with $$ \gamma:=\min_i\left(|a_{ii}|-\sum_{j\neq i}|a_{ij}|\right) \quad \text{($\gamma>0$ since $A$ is diagonally dominant)}, $$ we have $\gamma\|x\|_\infty\leq\|Ax\|_\infty$ for all $x$ and the inverse of this $\gamma$ is what you should have on the right-hand side of your statement. Indeed, let $i$ be such that $\|x\|_\infty=|x_i|$. Then $$ \begin{split} \|Ax\|_\infty&\geq|(Ax)_i|=\left|a_{ii}x_i+\sum_{j\neq i}a_{ij}x_j\right| \geq |a_{ii}||x_i|-\sum_{j\neq i}|a_{ij}||x_j|\geq|a_{ii}||x_i|-\sum_{j\neq i}|a_{ij}||x_i|\\ &=\left(|a_{ii}|-\sum_{j\neq i}|a_{ij}|\right)\|x\|_\infty\geq\gamma\|x\|_\infty. \end{split} $$ Consequently, $$ \|A^{-1}\|_\infty=\max_{x\neq 0}\frac{\|A^{-1}x\|_\infty}{\|x\|_\infty}=\max_{x\neq 0}\frac{\|x\|_\infty}{\|Ax\|_\infty}\leq\frac{1}{\gamma}. $$

$\endgroup$

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.