Scribd
Upload
16 views17 pages

Lecture 4

Uploaded by

drabiak2001
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
16 views17 pages

Lecture 4

Uploaded by

drabiak2001
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 17

Spectral Theorem

DIAGONALIZATION OF
SYMMETRIC MATRICES
SYMMETRIC MATRIX

▪ A symmetric matrix is a matrix A such that AT


= A.

▪ Such a matrix is necessarily square.

▪ Its main diagonal entries are arbitrary, but its other


entries occur in pairs—on opposite sides of the main
diagonal.
SYMMETRIC MATRIX
▪ Theorem 1: If A is symmetric, then any two
eigenvectors from different eigenspaces are
orthogonal.
▪ Proof: Let v1 and v2 be eigenvectors that correspond
to distinct eigenvalues, say, λ1 and λ2.
▪ To show that v1 v 2 = 0, compute
λ1v1 v 2 = (λ1v1 )T v 2 = ( Av1 )T v 2 Since v1 is an eigenvector

= (v1T AT )v 2 = v1T ( Av 2 ) Since A = A


T

= v (λ 2 v 2 )
T
1
Since v2 is an eigenvector

= λ 2 v v 2 = 2 v1 v 2
T
1
SYMMETRIC MATRIX
▪ Hence (λ1 − λ 2 )v1 v 2 = 0.
▪ But λ1 − λ 2  0, so v1 v2 = 0.
▪ An n  n matrix A is said to be orthogonally
diagonalizable if there are an orthogonal matrix P
(with P = P ) and a diagonal matrix D such that
−1 T

A = PDPT = PDP −1 ----(1)


▪ Such a diagonalization requires n linearly
independent and orthonormal eigenvectors.
▪ When is this possible?
▪ If A is orthogonally diagonalizable as in (1), then
A = ( PDP ) = P D P = PDP = A
T T T TT T T T
SYMMETRIC MATRIX

▪ Thus A is symmetric.
▪ Theorem 2: An n  n matrix A is orthogonally
diagonalizable if and only if A is symmetric matrix.

▪ Example 1: Orthogonally diagonalize the matrix


 3 −2 4 
A =  −2 6 2  , whose characteristic equation is
 
 4 2 3
0 = −λ3 + 12λ 2 − 21λ − 98 = −(λ − 7) 2 (λ + 2)
SYMMETRIC MATRIX

▪ Solution: The usual calculations produce bases for


the eigenspaces:
 1  −1/ 2   −1 
    
λ = 7 : v1 = 0 , v 2 = 1 ;λ = −2 : v 3 = −1/ 2 
     
 1  0   1 
▪ Although v1 and v2 are linearly independent, they are
not orthogonal.
v 2 v1
▪ The projection of v2 onto v1 is v1 .
v1 v1
SYMMETRIC MATRIX
▪ The component of v2 orthogonal to v1 is
 −1/ 2  1   −1/ 4 
v 2 v1   −1/ 2    
z2 = v2 − v1 = 1 − 0 = 1
v1 v1   2    
 0  1   1/ 4 

▪ Then {v1, z2} is an orthogonal set in the eigenspace


for λ = 7.

▪ (Note that z2 is linear combination of the eigenvectors


v1 and v2, so z2 is in the eigenspace).
SYMMETRIC MATRIX

▪ Since the eigenspace is two-dimensional (with basis


v1, v2), the orthogonal set {v1, z2} is an orthogonal
basis for the eigenspace, by the Basis Theorem.

▪ Normalize v1 and z2 to obtain the following


orthonormal basis for the eigenspace for λ = 7:

1/ 2   −1/ 18 
   
u1 =  0  , u 2 =  4 / 18 
1/ 2   
   1/ 18 
SYMMETRIC MATRIX
▪ An orthonormal basis for the eigenspace for λ = −2 is
 −2   −2 / 3
1 1   
u3 = 2v3 = −1 = −1/ 3
2v3 3   
 2   2 / 3

▪ By Theorem 1, u3 is orthogonal to the other


eigenvectors u1 and u2.

▪ Hence {u1, u2, u3} is an orthonormal set.


SYMMETRIC MATRIX

▪ Let
1/ 2 −1/ 18 −2 / 3 7 0 0 
 
P =  u1 u2 u3  =  0 4 / 18 −1/ 3 , D =  0 7 0
 
   0 0 −2 
1/ 2 1/ 18 2 / 3

▪ Then P orthogonally diagonalizes A, and A = PDP .


−1
THE SPECTRAL THEOREM
▪ The set of eigenvalues of a matrix A is sometimes
called the spectrum of A, and the following
description of the eigenvalues is called a spectral
theorem.

▪ Theorem 3: The Spectral Theorem for Symmetric


Matrices
▪ An n  n symmetric matrix A has the following
properties:
a. A has n real eigenvalues, counting
multiplicities.
THE SPECTRAL THEOREM

b. The dimension of the eigenspace for each


eigenvalue λ equals the multiplicity of λ as a
root of the characteristic equation.

c. The eigenspaces are mutually orthogonal, in


the sense that eigenvectors corresponding to
different eigenvalues are orthogonal.

d. A is orthogonally diagonalizable.
SPECTRAL DECOMPOSITION
▪ Suppose A = PDP −1
, where the columns of P are
orthonormal eigenvectors u1,…,un of A and the
corresponding eigenvalues λ1,…,λn are in the
diagonal matrix D.
▪ Then, since P = P ,
−1 T

 λ1 0   u1T 
   
A = PDP =  u1
T
un 
  
 0 λ n   u n 
 T

 u1T 
 
=  λ1 u 1 λnu n   
 u Tn 
SPECTRAL DECOMPOSITION
▪ Using the column-row expansion of a product, we can
write
A = λ1u1u1 + λ 2 u 2 u 2 + + λ n u n u n ----(2)
T T T

▪ This representation of A is called a spectral


decomposition of A because it breaks up A into pieces
determined by the spectrum (eigenvalues) of A.
▪ Each term in (2) is an n  n matrix of rank 1.
▪ For example, every column of λ1u1u1 is a multiple of u1.
T

T
▪ Each matrix u j u j is a projection matrix in the sense that
n T
for each x in , the vector (u j u j )x is the orthogonal
projection of x onto the subspace spanned by uj.
SPECTRAL DECOMPOSITION

▪ Example 2: Construct a spectral decomposition of the


matrix A that has the orthogonal diagonalization

7 2   2 / 5 −1/ 5   8 0   2 / 5 1/ 5 
A=  =    
 2 4   1/ 5 2 / 5  0 3  −1/ 5 2 / 5 

▪ Solution: Denote the columns of P by u1 and u2.

▪ Then A = 8u1u1T + 3u 2 u T2
SPECTRAL DECOMPOSITION
▪ To verify the decomposition of A, compute
 2 / 5   4 / 5 2 / 5
u 1u 1 =    2 / 5 1/ 5  = 
T

1/ 5   2 / 5 1/ 5 
 −1/ 5   1/ 5 −2 / 5
u 2u =    −1/ 5 2 / 5  = 
T

 −2 / 5 4 / 5
2
 2 / 5 

and
32 / 5 16 / 5  3 / 5 −6 / 5 7 2 
8u u + 3u 2 u = 
T T
 +  =  =A
16 / 5 8 / 5  −6 / 5 12 / 5  2 4 
1 1 2

You might also like