1
$\begingroup$

Let $A \in \mathbb{R}^{m\times n}$ and $v, w \in \mathbb{R}^n$, $u\in \mathbb{R}^m$, where $u,v,w$ are all unit vectors. Prove of disprove that $\max |Av| = \max u\cdot (Aw)$. (Assuming $|\cdot|$ is vector 2-norm)

2 Answers 2

1

If $A \in \mathbb{R^{m \times n}}$ then we can make a singular value decomposition $A = U \Sigma V^T$. We can choose the row vector $v^T$ from $V^T$ that corresponds to the largest value of $\Sigma$ and the column vector $u$ from $U$ that corresponds to the same value.

now $\max\limits_{|r| = 1} |A r| = |A v| = u^TAv = \max\limits_{|q| = 1, |r|=1} qAr$

0

Let $A=\pmatrix{-1&-2\\-3&-4}$ then $$\max |A v|=\max\{2,4\}=4$$ But $$\max u\cdot(Aw)=\max_{i,j}A_{ij}=-1$$

  • 0
    BTW, I assumed that By $|\cdot|$ you meant **norm-1** and by _unit vactor_ you meant a vector that has only one nonzero element with value $1$2017-02-01
  • 0
    Actually I mean norm-2.2017-02-01
  • 0
    Either way the statement is false @Lewis2017-02-01