Chapter 5

Eigenvalues, Eigenvectors, and

Diagonozation

(除了標註※之簡報外，其餘採用李宏毅教授之投影片教材)
Eigenvalues and
eigenvectors
(Chapter 5.1)
Eigenvalues and Eigenvectors
• If ( is a vector, is a scalar)
• is an eigenvector of A excluding zero vector
• is an eigenvalue of A that corresponds to

A must be square Eigen value

Eigen vector
Eigenvalues and Eigenvectors
• If ( is a vector, is a scalar)
• is an eigenvector of A excluding zero vector
• is an eigenvalue of A that corresponds to
• T is a linear operator. If ( is a vector,
is a scalar)
• is an eigenvector of T excluding zero vector
• is an eigenvalue of T that corresponds to
Eigenvalues and Eigenvectors
• Example: Shear Transform

(x,y) (x’,y’)

This is an eigenvector.
Its eigenvalue is 1.
Eigenvalues and Eigenvectors
• Example: Reflection
reflection operator T about the line y = (1/2)x

y = (1/2)x b1 is an eigenvector of T

= Its eigenvalue is 1.

b2 is an eigenvector of T

=− Its eigenvalue is -1.

 Eigenvalues and Eigenvectors
• Example:
2 0 Eigenvalue is 2
Expansion and 0 2
Compression

All vectors are

eigenvectors.

0.5 0
0 0.5 Eigenvalue is 0.5
Eigenvalues and Eigenvectors
Source of image:
https://twitter.com/circleponiponi
• Example: Rotation /status/1056026158083403776

Do any n x n matrix or linear operator have

eigenvalues?
How to find eigenvectors
(given eigenvalues)
(Chapter 5.1)
Eigenvalues and Eigenvectors
• An eigenvector of A corresponds to a unique
eigenvalue.
• An eigenvalue of A has infinitely many eigenvectors.
Example: 1 0
= 0 = 1
0 −1

−1 0 0 1 −1 −1 0 0 0 0
0 1 2 0 = 0 0 1 2 1 = −1
0 2 1 0 0 0 2 1 −1 1
Eigenvalue= -1 Eigenvalue= -1
Do the eigenvectors correspond to the same
eigenvalue form a subspace?

0
Eigenspace
• Assume we know is the eigenvalue of matrix A
• Eigenvectors corresponding to
Eigenvectors corresponding to
Av = v are nonzero solution of
(A  In)v = 0
Av  v = 0
Eigenvectors corresponding to
Av  Inv = 0 = (A  In) 
eigenspace
(A  In)v = 0
Eigenspace of :
matrix Eigenvectors corresponding to +
Check whether a scalar is
an eigenvalue
(Chapter 5.1)
 (A  In):
Check Eigenvalues eigenspace of 

• How to know whether a scalar  is the eigenvalue

of A?

Check the dimension of eigenspace of 

If the dimension is 0
Eigenspace only contains {0}

No eigenvector

is not eigenvalue
 (A  In):
Check Eigenvalues eigenspace of 

• Example: to check 3 and 2 are eigenvalues of the

linear operator T

(A  3In)=? (A + 2In)=?
−6 −3 −2 2 −2
−3 −2 1 −3 3
9 1
−2
3
−2
−2
 (A  In):
Check Eigenvalues eigenspace of 

• Example: check that 3 is an eigenvalue of B and find

a basis for the corresponding eigenspace

find the solution set of (B  3I3)x = 0

find the RREF of

B  3I3

=
Looking for eigenvalues
(Chapter 5.2)
Looking for Eigenvalues
A scalar is an eigenvalue of A
Existing such that
Existing such that
Existing such that
has multiple solution
The columns of are independent
Dependent
is not invertible
Looking for Eigenvalues
• Example 1: Find the eigenvalues of

A scalar is an eigenvalue of A

=0

t = -3 or 5

The eigenvalues of A are -3 or 5.

Looking for Eigenvalues
• Example 1: Find the eigenvalues of

The eigenvalues of A are -3 or 5.

Eigenspace of -3

find the solution

Eigenspace of 5

find the solution

Looking for Eigenvalues
• Example 2: find the eigenvalues of linear operator

standard
matrix

A scalar is an eigenvalue of A
−1 − 0 0
− = 2 −1 − −1
0 0 −1 −

− = −1 −
Looking for Eigenvalues
• Example 3: linear operator on R2 that rotates a
vector by 90◦

A scalar is an eigenvalue of A

standard matrix of the 90◦-rotation:

No eigenvalues, no eigenvectors
Characteristic Polynomial
A scalar is an eigenvalue of A
A is the standard matrix of linear operator T
Characteristic polynomial of A
linear operator T
Characteristic equation of A
linear operator T

Eigenvalues are the roots of characteristic

polynomial or solutions of characteristic equation.
Characteristic Polynomial
• In general, a matrix A and RREF of A have different
characteristic polynomials. Different Eigenvalues
• Similar matrices have the same characteristic
polynomials The same Eigenvalues
− = − =

= −
= −
1
= − = −
Characteristic Polynomial
• Question: What is the order of the characteristic polynomial
of an nn matrix A?
• The characteristic polynomial of an nn matrix is indeed
a polynomial with degree n
• Consider det(A  tIn) ?− ⋯ ?
⋮ ⋱ ⋮
? ⋯ ?−

• Question: What is the number of eigenvalues of an nn

matrix A?
• Fact: An n x n matrix A have less than or equal to n
eigenvalues
• Consider complex roots and multiple roots
Characteristic Polynomial
vs. Eigenspace
• Characteristic polynomial of A is
Factorization
multiplicity

Eigenvalue:

Eigenspace:
(dimension) ≤ ≤ ≤
Characteristic Polynomial
• The eigenvalues of an upper triangular matrix are
its diagonal entries.
Characteristic Polynomial:

The determinant of an upper triangular matrix

is the product of its diagonal entries.
Diagonalization
(Chapter 5.3)
Review
• If ( is a vector, is a scalar)
• is an eigenvector of A excluding zero vector
• is an eigenvalue of A that corresponds to
• Eigenvectors corresponding to are nonzero
solution of (A  In)v = 0
Eigenvectors Eigenspace of :
corresponding to
Eigenvectors
= (A  In)  corresponding to +
eigenspace
• A scalar is an eigenvalue of A
Review
• Characteristic polynomial of A is
Algebraic
Factorization
multiplicity

Eigenvalue:

Eigenspace:
(dimension) ≤ ≤ ≤
Outline
• An nxn matrix A is called diagonalizable if

• D: nxn diagonal matrix

• P: nxn invertible matrix
• Is a matrix A diagonalizable?
• If yes, find D and P
Diagonalizable
A2 = 0 (?)
• Not all matrices are diagonalizable

If A = PDP1 for some invertible P and diagonal D

A2 = PD2P1 = 0 D2 = 0 D=0
D is diagonal
A= 0

=0 =0
⋯ 0 ⋯ 0
= ⋮ ⋱ ⋮ = ⋮ ⋱ ⋮
0 ⋯ 0 ⋯
=0 =0
Diagonalizable
• If A is diagonalizable

is an eigenvector of A corresponding to eigenvalue

Diagonalizable
• If A is diagonalizable
is an eigenvector of A
corresponding to eigenvalue
=

There are n eigenvectors that form an invertible matrix

=

There are n independent eigenvectors

=

The eigenvectors of A can form a basis for Rn.

Diagonalizable
• If A is diagonalizable
is an eigenvector of A
corresponding to eigenvalue

How to diagonalize a matrix A?

Find n independent eigenvectors corresponding if
Step 1:
possible, and form an invertible P

Step 2: The eigenvalues corresponding to the

eigenvectors in P form the diagonal matrix D.
 Diagonalizable
A set of eigenvectors that correspond to distinct
eigenvalues is linearly independent.

Factorization

Eigenvalue: ……
Eigenspace: ≤ ≤ …… ≤
(dimension)

Independent
 Diagonalizable
A set of eigenvectors that correspond to distinct
eigenvalues is linearly independent.
Eigenvalue: …… Assume dependent
Eigenvector: …… a contradiction
vk = c1v1+c2v2+  +ck1vk1
Avk = c1Av1+c2Av2+  +ck1Avk1
kvk = c11v1+c22v2+  +ck1k1vk1 (k)
- kvk = c1kv1+c2kv2+  +ck1kvk1
0 = c1(1k) v1+c2(2k)v2+  +ck1(k1k)vk1
Not c1 = c2 =  = ck1 = 0 Same eigenvalue a contradiction
 Diagonalizable
• If A is diagonalizable
is an eigenvector of A
corresponding to eigenvalue

Eigenvalue: ……
Eigenspace: ……

Basis for Basis for Basis for

Independent Eigenvectors
You can’t find more!
 Diagonalizable - Example
 1 0 0 
• Diagonalize a given matrix A   0 1 2 
 0 2 1 

characteristic polynomial is (t + 1)2(t  3) eigenvalues: 3, 1

eigenvalue 3  0  A = PDP1, 0 1 0 
  
P  1 0 1 
where
B1 =  1  
 1   1 0 1
  
eigenvalue 1
 1   0   3 0 0 
     D  0 1 0 
B2 =  0  ,  1  
 0   1  0 0 1
    
Test for a Diagonalizable Matrix
• An n x n matrix A is diagonalizable if and only if
both the following conditions are met.
The characteristic polynomial of A factors into
a product of linear factors.

Factorization

For each eigenvalue  of A, the multiplicity of 

algebraic multiplicity) equals the dimension of the
corresponding eigenspace (geometric multiplicity).
Independent Eigenvectors
An n x n matrix A is diagonalizable

=
The eigenvectors of A can form a basis for Rn.

Eigenvalue: = ……
Eigenspace: = = …… =
(dimension)
 Geometric Meaning of
Diagonalization

Rotation Scaling Rotation

Red and blue axes correspond to the directions of two eigenvectors

※
 How to Cope with Non-
diagonalizable Matrices
i.e.,
, can we write , where is
“near diagonal”?

Schur Decomposition: Any square matrix A can be

written as , where U is an “orthonormal
matrix” and T is “upper triangular” with eigenvalues
on the diagonal.

T=

※
 Jordan Normal Form

where J is in
Jordan Normal Form.
Jordan Normal Form

2 0
Eigenspace of 2
Jordan Block has dim = 1 1 0 0
 1 0  0 0 2 0 0 0
0 
 1 0
 J  0 0 3 1 0
J        

0   1
 Eigenspace of 3
 0 0 0 3 0
0 has dim = 2
0 0 0   0 0 0 0 3
kxk
※
Application of Diagonalization
• If A is diagonalizable,
= =
• Example: .15
.85 Study IG .97
.03

.85 .85
…… Study Study Study

.727 .15 .85 .15

.03
…… IG IG
.273 .97
.15
 From
Study IG .15
To Study .85 Study IG .97
IG .03

.85 .85
…… Study Study Study

.727 .15 .85 .15

.03
…… IG IG
.273 .97
.15
Diagonalizable
• Diagonalize a given matrix

 1
 p1   
RREF  1

 1 (invertible)
 p2   
RREF 5
When ,
The beginning
condition does
not influence.
 Diagonalization
of linear operators*
(Chapter 5.4)
Diagonalization of Linear Operator
8 +9
• Example 1: = −6 − 7
3 +3 −

det 8 -t 9 0
The standard matrix is = −6 −7 -t 0
3 3 −1 -t
 the characteristic polynomial is (t + 1)2(t  2)
 eigenvalues: 1, 2

Eigenvalue -1: Eigenvalue 2:

−1 0 3  B1  B2 is a basis of R 3
B1= 1 , 0 B2= −2  T is diagonalizable.
0 1 1
Diagonalization of Linear Operator
− + +2
• Example 2: = −
0
det −1 -t 1 2
The standard matrix is = 1 −1 -t0
0 0 0 -t
 the characteristic polynomial is t2(t + 2)
 eigenvalues: 0, 2 1 −1 0
 the reduced row echelon form of A  0I3 = A is 0 0 1
1 0 0 0
− =0
= 1
=0
0
 the eigenspaces corresponding to the eigenvalue 0 has the
dimension 1 < 2 = algebraic multiplicity of the eigenvalue 0
 T is not diagonalizable.
Review
Simple Function
Another output’
Input’
coordinate
system

Cartesian
coordinate Input Output
system Complex Function
Flowchart
B
B B
B coordinate
system

Cartesian
coordinate
system

B B
similar similar
• Example: reflection operator T about the line y = (1/2)x

0.4 0.2 2 −1
= =
−1 −0.2 0.4 1 2
=
2 y = (1/2)x
B
B B
2
=
1

0 y = (1/2)x
= =?
1

1
=
0
B
Diagonalization of Linear Operator
• Reference: Chapter 5.4
B
B B
simple

Properly Properly
selected selected
Diagonalization of Linear Operator
• If a linear operator T is diagonalizable

B
B B
simple
Eigenvectors form
the good system
Properly Properly
selected selected
Diagonalization of Linear Operator
-1: 2:
8 +9 −1 0 3
= −6 − 7 B1= 1 , 0 B2= −2
3 +3 − 0 1 1

B
B B
−1 0 0
0 −1 0
−1 0 3 0 0 2 −1 0 3
1 0 −2 1 0 −2
0 1 1 8 9 0 0 1 1
−6 −7 0
3 3 −1