- #1
Demon117
- 165
- 1
Homework Statement
For a symmetric matrix A, use the notation [tex]\lambda_{k}\left(A\right)[/tex] to denote the [tex]k^{th}[/tex] largest eigenvalue, thus
[tex]\lambda_{n}\left(A\righ)<=...<=\lambda_{2}\left(A\right)<=\lambda_{1}\left(A\right)[/tex]
Now suppose A and A+E are nxn symmetric matrices, prove the following results:
(a) [tex]\sum(\lambda_{i}\left(A+E\right)-\lambda_{i}\left(A\right))^{2}<=norm(E)_{F}^{2}[/tex] (Frobenius norm).
(b)[tex]|\lambda_{k}\left(A+E\right)-\lambda_{k}\left(A\right)|<=norm(E)_{2}[/tex] (2-norm) for k=1,...,n.
Homework Equations
Courant-Fischer Minimax theorem
[tex]\lambda_{k}\left(A\right)+\lambda_{k}\left(E\right)<=\lambda_{k}\left(A+E\right)<=\lambda_{k}\left(A\right)+\lambda_{1}\left(E\right)[/tex]
The Attempt at a Solution
I am not really sure how to use the Courant-Fischer Minimax theorem to establish that relationship in 2. If I had some pointers I am sure I could produce something.