hegendroffer

Hi, I am struggling with the following proof. I think I know how to prove the right hand side inequality |lambda| <= ||A||, but I still don't know how to prove the left hand side inequality.

Could you please give me some hint ?

Attached Images

Proof.jpg (13.7 KB, 23 views)

chiro

Hey hegendroffer.

Hint - Use the sub-multiplicative norm with respect to the eigen-decomposition of the matrix and consider the result of the inverse of the matrix (in terms of how the eigen-values become reciprocated).

The eigen-decomposition should have three matrices of PDP_inverse and the sub-multiplicative law of matrix norms in combination with how eigen-values are determined for the original matrix and the inverse relate to that result.

zylo

Let ui and $\displaystyle \lambda_{i}$ be eigenvectors and eigenvalues of A
x=x1u1+..xnun, ||x||=1
||A||=max||Ax||
||Ax||=||$\displaystyle \lambda_{1}$x1u1+...$\displaystyle \lambda_{n}$xnun|| $\displaystyle \geq$ |$\displaystyle \lambda_{1}$|,..,|$\displaystyle \lambda_{n}$|
||A|| $\displaystyle \geq |\lambda|$

Au=$\displaystyle \lambda$ u
A$\displaystyle ^{-1}$u=1/$\displaystyle \lambda$ u
||A$\displaystyle ^{-1}$||$\displaystyle \geq$ 1/|$\displaystyle \lambda$|