Lecture 28 Monday March 30 Matrix Norms and Eigen values

# Lecture 28 Monday March 30 Matrix Norms and Eigen values -...

This preview shows pages 1–2. Sign up to view the full content.

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Lecture 28: Matrix Norms and Eigen values Matrix Norms Frobenius norm a i, j 2 ij 2 2-norm j u j max|v|1 | Av | 1 2 12 2 32 ... |A|2 = |UΣVT|2 = |Σ|2 = 1 |A|F = |UΣVT|F = |Σ|F = Theorem: (1)n ( n (a11 a22 ... ann ) n 1 ...) |A – Ak|F = (1) n (a11 a22 ... ann ) (1)n (11 22 ... nn ) min rank ( B )k | A B |F c(k ) 2 |A – A2|2 = min rank ( B ) k | A B |2 Let B be a square matrix. Find solutions of the form: B x = x Consider B = AAT = ( = i T jii j T j k 1 2 d u v )( u v i 1 T iii n n u v v u = i, j j 1 j j T j ) 2 T i ii uu Buj = i i 2 T i ii u u u j = 2u j j * ui uiT 1 when i j B is symmetric and positive semi definite xTBx 0 xTAATx = (ATx)2 0 What does the probability distribution of Eigen values of random matrix look like? (Wigner, 1957) The distribution of eigenvectors can tell you whether a matrix is random. Lemma: trace(A) = λ1 + λ2 + λ3 + … + λn trace(A2) = λ12 + λ22 + λ32 + … + λn2 Proof: There exists a nontrivial solution for x in (A – λI)x = 0 only if det(A - λI) = 0 det(A - λI) is polynomial in λ. (1) n ( ) (1) ( n i 1 i n n (1 2 ... n ) n1 ...) = (a11 – λ)det(n – 1) + … = (a11 – λ) (a22 – λ) (a33 – λ)… (ann – λ) + … = (1)n ( n (a11 a22 ... ann ) n1 ...) We want to equate λn-1 coefficients. The coefficient for λn-1 is (1)n (a11 a22 ... ann ) (1)n (a11 a22 ... ann ) (1)n (11 22 ... nn ) We want to show P(λ) is like 1 2 (semicircle) Normalize the Eigen values to range [-1, 1] 2 1 2 Find equal moments for both equations to show equality Kth moment: c( k ) 2 k 1 2 d ...
View Full Document

## This document was uploaded on 01/22/2010.

### Page1 / 2

Lecture 28 Monday March 30 Matrix Norms and Eigen values -...

This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online