7.

Eigenvalues and
Eigenvectors
Learning Objectives

Given a square matrix A, students are expected to be able to

1. Determine eigenvalues and corresponding eigenvectors,
2. Diagonalize matrix A,
3. Identify matrix properties according to its eigenvalues,
4. Explain eigenspace as a nullspace of a particular matrix.

2
Scope

Eigenvalues and Eigenvectors

Eigenspace

Diagonalization

Orthogonal Diagonalization

3
Pre-test modul
Pre-test

• Answer the following question:

Given matrix A2x2 and vectors u, v, and w

2 0   5   1   0 
A=  u=  v=  w= 
4 1  4  4  4 

Find Au, Aw, Av.

Which of the following result is a vector that is parallel to the original

vector?

5
Answers to pre-test

2 0   5   1   0 
A=  u=  v=  w= 
4 1  4  4  4 

 2 0 5  10  
Au =  =
 4  24   ku untuk semua k  R u and Au are NOT parallel
 4 1    

 2 0 1  2 1  
Av =  =
    = 2 . 4  = 2v v and Av are parallel
4 1  4  8   

  2 0  0  0  
Aw =  =
    = 1 w w and Aw are parallel
4 1  4  4 

6
Eigenvalues and Eigenvectors

Eigenspaces

Diagonalization

Orthogonal Diagonalization

7.1 Eigenvalues and

Eigenvectors
Vector and Matrix Multiplication

2 0 5  10  5  2 0 1  1   2 0  0  0 
4 1  4  = 24   k 4  4 1  4  = 2 4  4 1  4  = 1  4 
                

Au ≠ ku Av = 2v Aw = 1w
10 
y Au =   y 2  y
4 Av =  
8 

1 
v= 
5  4  0 
u=  w = Aw =  
4  4 

x x x

8
Vector and Matrix Multiplication

A x = λx

Ax
x

Ax

x and Ax are parallel

9
Definition: eigenvalue and eigenvector

Definition 7.1: Eigenvalue and Eigenvector

Given Anxn, then a nonzero vector v in Rn is called an eigenvector of A if if
there is a scalar λ such that Av = λv,
λ is called an eigenvalue, x is called an eigenvector of A corresponding to
λ.
(v ≠ 0)

Av
v
v v
v Av
Av Av
(1) λ ≥ 1 (2) 0 ≤ λ ≤ 1 (3) -1 ≤ λ ≤ 0 (4) λ ≤ - 1

10
Exercise 1: zero vector and the zero
eigenvalue
a. Why is an eigenvector never a zero vector?
Answer:
Suppose vector v = 0, then every matrix Anxn, and scalar λ will satisfy
the equation Ax = λx, such that the equation would not give any
meaningful information.

b. Is it possible to have 0 as an eigenvalue?

Answer:
An eigenvalue can be zero. If a matrix have 0 as an eigenvalue, then
that matrix does not have an inverse.

11
Exercise 2: Homogeneous SLE Anxnx = 0

1. A is a square matrix and SLE Ax = 0 only has the trivial solution. What
is your conclusion about A?
Answer:
a. A has an inverse.
b. det(A) ≠ 0

2. A is a square matrix and SLE Ax = 0 has nontrivial solutions. What is

your conclusion about A and det(A)?
Answer:
a. A does not have an inverse.
b. det(A) = 0

12
The Eigenvector Problem

Given a square matrix A,

A x parallel to x

A x = λ x

Find al nonzero vectors x such that Ax is a scalar multiple of x

(or Ax is parallel to x).
or

Find all nonzero vectors x such that Ax = λx for a scalar λ

13
The Eigenvalue Problem

Given a square matrix A.

A x = λ x
x is a nonzero vector

Find all scalar λ such that Ax = λx for a nonzero vector x.

or
Find all scalar λ such that Ax = λx has nontrivial solutions.

14
Equivalent Statements

If A is an nxn square matrix, then the following statements are equivalent

1.  is an eigenvalue of A

2. There is a nonzero vector x such that Ax = x

3. SLE (A – I)x = 0 has nontrivial solutions

4.  is a solution to the equation det(A – I) = 0

Finding eigenvalues of A is equivalent to finding all solutions to det(I-A) = 0

15
Characteristic Equation
If it is expanded, det(A - λI) is an n degree polynomial in terms of λ,
p( ) = n + cn−1 n−1 + cn−2n−2 +  + c1  + c0 characteristic polynomial

The equation det( A − I ) = n + cn−1n−1 + cn−2n−2 +  + c1 + c0 = 0 is called a

characteristic equation

A - λI = A-λI

=  + cn−1 + cn−2 +  + c1  + c0 = 0
n n−1 n−2
det A-λI

A polynomial equation of degree n has at most n solutions, thus a nxn matrix

has at most n distinct eigenvalues.

16
Exercise 3: finding eigenvalues
2 0
Find all eigenvalues of A =  

 4 1 

Answer:
First, find the characteristic equation det(A- λ I) = 0
2 −  0 
det  = 0
 4 1 −  

Such that we have the following equation

(2 −  )(1 −  ) − 0.4 = 0
(2 −  )(1 −  ) = 0

The solution to that equation are eigenvalues of A i.e.:

1 = 2
2 = 1
17
Procedure to find eigenvalues

Given a square matrix A, then the eigenvalues of A can be obtained as

follows

1. Find the characteristic equation det(A - λI) = 0

p( ) = n + cn−1 n−1 + cn−2n−2 +  + c1 + c0 = 0

2. Modify the characteristic equation into an equivalent equation

where the polynomial is easier to solve.

3. Solve the equation to obtain eigenvalues of A.

18
Exercise 4: finding eigenvalues
 1 1 1
 
Find all eigenvalues of matrix A =  0 3 3
−2 1 1
 
Answer:
1. Find the characteristic equation det(A - λI) = 0
1 −  1 1 
 
det( A − I ) = det 0 3− 3  = (1 −  )2 (3 −  ) − 6 + 2(3 −  ) − 3(1 −  )
 −2 1 1 −  

2. Modify the characteristic equation into an equivalent equation that is easier to solve
(1 −  )2 (3 −  ) − 6 + 2(3 −  ) − 3(1 −  ) = 0
3. Solve the equation to obtain eigenvalues

(1 −  )2 (3 −  ) − (3 −  ) = 0 Eigenvalues of A:
1 = 0
 ( − 2)(3 −  ) = 0
2 = 2
3 = 3
19
Eigenvalues of diagonal matrices
2 0 0 0
 
0 5 0 0
Given a diagonal matrix A=
0 0 6 0
 
0 0 0 1 


2 −  0 0 0 
 
 0 5 −  0 0 
A − I = 
0 0 6− 0 
 
 0 0 0 1 −  
 
Characteristic equation: det( A − I ) = (2 −  )(5 −  )(6 −  )(1 −  ) = 0

The eigenvalues are 2, 6, 5, 1 (entries of main diagonal)

The eigenvalues of a diagonal matrix are its entries in the main diagonal.

20
Exercise 5 : finding eigenvalues
1. Given an upper triangular matrix below. Find its eigenvalues.
1 1 3 0 
 
0 5 2 0
A=
0 0 6 7
 
0 0 0 7
 
2. Find all eigenvalues of the following lower triangular matrix.
 a11 0  0 
 
 a21 a22  0 
A=
    
 
a an2  ann 
 n1

3. What is the eigenvalue of the identity matrix?

4. The characteristic polynomial of matrix B is (λ-1)3(λ-1) λ, what is the size of B?

21
Solution
1 1 3 0
 
1. Eigenvalues of matrix 0 5 2 0
are 1, 5, 6, 7
A=
0 0 6 7
 
0 0 0 7 


 a11 0  0 
 
 a21 a22  0 
2. Eigenvalues of matrix A =   are a11, a22,…ann
   
 
a an2  ann 
 n1

3. The eigenvalue of the identity matrix is 1

4. The characteristic polynomial of matrix B is (λ-1)3(λ-1) λ, then the size of B is 5x5

22
Is λ an eigenvalue of A?

2 2 0 Suppose the question is are 2, 4, 8 eigenvalues of matrix A,

 
A = 0 4 0 then it can be solved by substituting 2, 4, 8 into the characteristic
0 1 0 equation and check if it is a solution.
 

λ = 2, 4, 8. If det(A-λI) = 0, then λ is an eigenvalue, if ≠ 0, then λ is not an eigenvalue of A.

2 − 2 2 0 
 
( )
det A − 2I = det 0 4 −2 0 =0 2 is an eigenvalue of A
 0 1 0 − 2 


2 − 4 2 0 
 
det( A − 4I ) = det 0 4−4 0 =0 4 is an eigenvalue of A
 0 1 0 − 4 

2 − 8 2 0 
 
det( A − 0I ) = det 0 4−8 0  = −192 8 is not an eigenvalue of A
 0 1 0 − 8 

23
Exercise 6 : finding eigenvectors

• Given matrix A, nonzero vectors u, x, y and zero vector z. Determine if u, x, y, z

are eigenvectors of A.

 1 1 1  0  − 4  − 2  0  2
               
A =  0 3 3 , u =  − 2 , x =  − 6 , y =  1  z =  0 , w =  1 
−2 1 1  2   2   1  0 0
           
• Find and draw vectors u, x, y, z and Au, Ax, Ay, Az

• [Hint: x is an eigenvector corresponding to 2,

u is an eigenvector corresponding to 0,
y is not an eigenvector of A,
z is a zero vector, thus not an eigenvector.]

24
Exercise 7

Determine if the following statements are true or false.

1. A matrix may have zero as its eigenvalue, and a zero vector is not an
eigenvector for any matrix.
– Answer: TRUE
– The zero vector is not an eigenvector (by definition), eigenvalues can be
zero. Next, it will be proven that having zero as eigenvalue is an indicator
that it is a singular matrix (has no inverse)

2. Vectors x and y are eigenvectors of A, then x = y.

– Answer: FALSE

3. λ1 and λ2 are eigenvalues of A, then det(A - λ1I) = det(A - λ2I)

– Answer: TRUE

25
Eigenvalues of Power of a Matrix

Theorem 7.1:
If n is a positive integer, λ is an eigenvalue of A, then λn is an eigenvalue of An

• Given an arbitrary matrix A where λ is its eigenvalue. Then there is a

nonzero vector x such that
Ax = λx Multiply both sides with A
A.Ax = A λx
A2x = λ(Ax) substitute Ax with λx
A2x = λ2x Thus, λ2 is an eigenvalue A2

26
Eigenvalues of Power of a Matrix
 1 1 1
 
• Eigenvalues of A are 0, 2, and 3. A =  0 3 3
−2 1 1
 

 −1 5 5 
 
Find eigenvalues of A2 =  − 6 12 12 , A13 , A20
− 4 2 2 
 
Eigenvalues of A2 are: 0, 4, and 9
Eigenvalues of A13 are: 0, 213, and 313

Eigenvalues of A20 are 0, 220, and 320

• If eigenvalues of A are known, then eigenvalues of An can be determined

without having to find An first.

27
Eigenvalue of Singular Matrix
• Suppose  = 0 is an eigenvalue of A.
Then 0 is a solution to the characteristic equation:
By substituting  with 0, we have c0 = 0.
Because det(A- I) = 0, where = 0, then det(A) = c0 = 0.
And because det(A) = 0 then A does not have inverse.

• det(A) = det(A - I) where  = 0.

Thus, det(A) = c0.
If A does not have inverse, then det(A) = 0 = c0.
Such that  = 0 is a solution to the characteristic equation.
 = 0 is an eigenvalue of A.

Theorem 7.2:
0 is an eigenvalue of A if and only if A does not have an inverse.

28
Eigenvalues of Transpose of Matrix

det(B) = det(BT)
(A- I)T = (AT- I)

Suppose  is an eigenvalue of A, then det(A-  I)= 0.

Because the determinant of a matrix and its transpose are the same,
then det(A-  I)T= 0. Because (A-  I)T = (AT- I) , then det(AT-  I)= 0. Thus,  is also
an eigenvalue of AT. Conclusion: A and AT have the same eigenvalue.

Theorem 7.3:
A and AT have the same eigenvalue.

29
Exercise 8
Determine if the following statements are TRUE or FALSE!
1. If Ax – ax = 0 for a nonzero scalar a, then x is an eigenvector.
2. If a is not an eigenvalue of A, then SLE (aI –A)x = 0 has exactly one solution.
3. Every identity matrix has one eigenvalue, that is 1
4. A matrix that only has 1 as its eigenvalue is the identity matrix.
5. If the following is the characteristic polynomial of A
p( ) = ( − 1 )2 ( − 2)3 ( − 3), then A is a 5x5 matrix
6. If the following is the characteristic polynomial of matrix A
p( ) = ( − 1 )2 ( − 2)3 ( − 3) , then A2 does not have an inverse.

30
Solution to Exercise 8

Solution:
1. FALSE, if x is a zero vector, x is not an eigenvector
2. TRUE
3. TRUE
4. FALSE
5. FALSE
6. FALSE

31
Eigenvalues and Eigenvectors

Eigenspace

Diagonalization

Orthogonal Diagonalization

7.2 Eigenspace

32
Scalar Multiple of Eigenvectors
• Given A, suppose x is an eigenvector of A corresponding to 2. Find if 1/2x, 10x, 5x are
also eigenvectors of A
 1 1 1 − 4  − 2  − 40   − 20 
    1      
A =  0 3 3 x =  − 6 x =  − 3  10x =  − 60  5x =  − 30 
−2 1 1  2  2  1   20   10 
         

 1 1 1  − 4   − 8   1 1 1  − 40   − 80 
         
Ax =  0 3 3  − 6  =  − 12  = 2x A(10 x ) =  0 3 3  − 60  =  − 120  = 2(10 x )
 − 2 1 1  2   4   − 2 1 1  20   40 
         

Ax = 2 x A(10x) = 2 (10x)

A x = λ x

A (10) x = λ (10) x

33
Scalar Multiple of Eigenvectors (cont)

 1 1 1 − 4  − 2
    1  
A =  0 3 3 x =  − 6 x =  − 3
−2 1 1  2  2  1 
     

 1 1 1  − 4   − 8   1 1 1  − 2   − 4 
1      1
    
Ax =  0 3 3  − 6  =  − 12  = 2x A x  =  0 3 3  − 3  =  − 6  = 2 x 
2    1   2  2 
 − 2 1 1  2   4   − 2 1 1    
    
1 1
Ax = 2x A  x = 2  x
2 2

A x = λ x A (1/2) x = λ (1/2) x

Scalar multiple (nonzero) of an eigenvector is an eigenvector corresponding

to the same eigenvalue

34
Finding Eigenvectors of Eλ
• Given a matrix A and one of its eigenvalues, λ. Find all eigenvectors
corresponding to λ.

• Eigenvectors of A corresponding to λ = 3 can be obtained by solving SLE (A - λ I)x

= 0. These eigenvectors are elements of Null(A - λ I)

Null(A - λ I) The set of solutions to

SLE (A - λ I)x = 0

0 The set of eigenvectors

corresponding to λ

Null(A - λ I)-{0}

35
Eigenspace
• The eigenspace of A corresponding to λ consists of all eigenvectors
corresponding to λ and the zero vector

The space of solutions to (A - λ I)x = 0

Null(A - λ I)x

0
Eigenspace
Eλ

Null(A - λ I) = Eλ

• Finding Eλ is the same as finding the solutions to:

(A - λ I)x = 0

36
Procedure to find eigenspace

Given A and its eigenvalue λ.

E λ is the eigenspace of A corresponding to λ that is obtained by:

1. Making the homogeneous SLE (A - λI)x = 0

2. Solve the SLE, you will obtain the subspace: Null(A - λI)

3. Eigenspace Eλ is the set of solutions to SLE (A - λI)x = 0

E λ= Null(A – λI)

37
Example 1: Finding eigenspace Eλ
Given matrix A and one of its eigenvalue λ = 3. Find all eigenvectors corresponding
to λ = 3.
 1 1 1 1 − 3 1 1 
   
A =  0 3 3 A − I =  0 3−3 3 
−2 1 1  −2 1 1 − 3 
  

Given SLE ( A − 3I ) x = 0
1 − 3 1 1  x1   0  − 2 x1 + x2 + x 3 = 0
    
( A − 3I )x =  0 3 − 3 3  x2  =  0  3 x3 = 0
 −2 1 1 − 3  x3   0  − 2 x1 + x2 − 2 x 3 = 0

x1 = a  1  
 
General solution : x2 = 2a Solution set: a  2  , a  R 
x3 = 0  0 
   
 1  
   
The set of eigenvectors of A corresponding to  = 3: a  2  , a  0, a  R 
 0 
   

38
Exercise 9: Finding the eigenspace

1. What si the relation between eigenspace Eλ and Null(A - λ I)

2. Is every element in the eigenspace is an eigenvector?
3. Make a procedure to find the basis of the eigenspace of A
corresponding to eigenvalue λ.

Answer:
1. Eλ = Null(A - λ I)
2. NO, there is one element, i.e. 0 that is not an eigenvector.
The zero vector is not an eigenvector for any matrix.
3. [Hint: find basis of Null(A - λI)]

39
Eigenvalues and Eigenvectors

Eigenspace

Diagonalization

Orthogonal Diagonalization

7.3 Diagonalization
Diagonalization

Definition 7.2: diagonalization

A square matrix A is said to be diagonalizable if there exists an invertible
matrix P such that P-1AP = D is a diagonal matrix.

Example 2:
5 − 6 2 1  1 −1
A =   P =   −1
P =  
 3 − 4   1 1   − 1 2  Diagonal
matrix
 1 − 1  5 − 6  2 1   2 0 
    =  
 − 1 2  3 − 4  1 1   0 − 1 
P-1 A P D

A is diagonalizable

41
When is matrix A diagonalizable?
Theorem 7.3.
If A is a nxn matrix, then the following sentences are equivalent:
1. A is diagonalizable
2. A has n linearly independent eigenvectors

 a11 a12  a1 n 
Proof (1) → (2)  
 a21 a22  a2n 
A=
Given A    
 
a an2  ann 
 n1

 p11 p12  p1 n   p11 p12  p1 n 

   
Suppose A is diagonalizable, then  p21 p22  p2n   p21 p22  p2n 
P = AP = PD = 
there is a matrix P that has an        
   
p pn2  pnn  p pn2  pnn 
inverse  n1  n1

 1 0  0 
 
0 2  0 
Such that P-1 AP = D is a diagonal D= 
  
matrix 
0

 0   n 

42
When is matrix A diagonalizable? (cont)

P-1AP = D, multiply both sides with P,

AP = PD
 p11 p12  p1 n   1 0  0    1 p11  2 p12   n p1 n 
    
 p21 p22  p2n  0 2  0    1 p21  2 p22   n p2n 
PD =  =  = ( 1 p1  2p 2   np n )
         
    
p pn2  pnn  0 0   n    1 pn1  2 pn2   n pnn 
 n1

 a11 a12  a1 n  p11 p12  p1 n 

  
 a21 a22  a2n  p21 p22  p2n 
AP =  = (Ap1 Ap 2  Ap n )
       
  
a an2  ann  pn1 pn2  pnn 
 n1
AP = PD, thus Ap1 = 1p1 Ap2 = 2p2  Apn = npn
Because P is invertible, then its columns are not columns of zeros. According to the
definition of eigenvalues, then λ1, λ2, λ3, …,λn are eigenvalues of A, and columns of
P are eigenvectors of A that are linearly independent (because P has an inverse)

Proof for (2) → (1) may be done as exercise.

43
Procedure to Diagonalize a Matrix

Given matrix Anxn find matrix P such that P-1AP = D.

Procedure
1. Find n linearly independent eigenvectors of A, e.g. p1, p2, p3, …, pn

2. Define matrix P where its columns are p1, p2, p3, …, pn

3. Matrix D = P-1AP is a diagonal matrix where the entries of its main

diagonal are λ1, λ2, λ3, …,λn where λj is an eigenvalue corresponding to pj
for j = 1, 2, 3, …, n

44
Example 3: diagonalizing matrix
Given matrix Anxn find P such that P-1AP = D.
5 − 6
A =  
 3 − 4 
Procedure
1. Find 2 linearly independent eigenvectors A. First, find its eigenvalues, i.e. λ1= 2 and λ2= -1
(was previously obtained/calculated). Find eigenvectors corresponding to its eigenvalues,
by solving SLE (A - λ I)x =0 we have
2 1 
p 1 =   p 2 =  
1  1 
2. Define matrix P where its columns are the eigenvectors above.
2 1  1 −1
P =   P −1 =  
 1 1   − 1 2 
3. Matrix D = P-1A P is a diagonal matrix where the entries of its main diagonal are λ1, λ2
consecutively
 1 − 1  5 − 6  2 1   2 0 
D =     =  
 − 1 2  3 − 4  1 1   0 − 1 

45
The Diagonalization Problem and The
Eigenvector Problem
• The Eigenvector Problem
Given matrix Anxn, is there a basis of Rn that consists of eigenvectors of A?

• The Diagonalization Problem

Given matrix Anxn is there an invertible matrix P such that P-1AP is a diagonal
matrix?

Theorem 7.4:
Anxn is diagonalizable if and only if there are n linearly independent eigenvectors.

Moreover, every n linearly independent eigenvectors of Rn is a basis of Rn.

Conclusion: the eigenvector problem is equivalent to the diagonalization
problem

46
The Diagonalization Problem (cont)

Theorem 7.5:
If v1, v2, …,vk are eigenvectors of A corresponding to distinct eigenvalues λ1, λ2, …
λk, then { v1, v2, …,vk } is a linearly independent set.

The following theorem is the consequence of theorem 7.5.

Theorem 7.6.:
If matrix Anxn has n distinct eigenvalues, then A is diagonalizable.

47
Example : diagonalizable matrix
a. Matrix 0 1 0
 
A = 0 0 1
 4 − 17 8 
 
Has 3 distinct eigenvalues, i.e. λ = 4, λ = 2+3, λ = 2 - 3. Thus, A is
diagonalizable.

b. Matrix B has characteristic equation λ4 + 6λ3 + 11λ2 – 6λ = 0.

Thus B has eigenvalues: λ = 0, λ = 1, λ = 2, and λ = 3. then B
is diagonalizable.

c. Matrix C has characteristic equation (5-λ)4 (2-λ)3 = 0.

C has eigenvalues: λ = 5 and λ = 2. If dim(E5) = 4 AND dim (E2) = 3 then
C is diagonalizable.
Geometric and Algebraic Multiplicity

Definition 7.3.:
• If λ0 is an eigenvalue of a square matrix A, then dimension of the
eigenspace corresponding to λ0 is called the geometric multiplicity of λ0.
• The number of times λ – λ0 appears as a factor in the characteristic
polynomial of A is called the algebraic multiplicity of A.

Suppose A is a 6x6 matrix that has the characteristic equation λ2 (λ-1) (λ-2)3 = 0.
Algebraic multiplicity of each eigenvalues 0, 1, and 2 are 2, 1, dan 3.

If the dimension of eigenspace E0 is 2, then the geometric multiplicity of 0 is 2.

49
Geometric and Algebraic Multiplicity
Theorem 7.4:
If A is a square matrix, then:
1. For every eigenvalue of A, its geometric multiplicity is less than or equal to its
algebraic multiplicity.
2. A is diagonalizable if and only if the geometric multiplicity is equal to its
algebraic multiplicity for each of the eigenvalue of A.

50
Exercise 10: information from the characteristic
equation

Suppose A is a 6x6 matrix where its characteristic equation is

λ2 (λ-1) (λ-2)3 = 0. List the possible dimension of the eigenspaces of A?

Answer:

The eigenspace of λ = 0 may have the dimension of 1 or 2.

The eigenspace of λ = 1 have the dimension of 1.

The eigenspace of λ = 2 may have the dimension of 1, 2, or 3.

51
Geometric and Algebraic Multiplicity

The characteristic equation of A is (1-λ) (6-λ)2 = 0. Thus, A has eigenvalues:

λ = 1, λ = 6. Suppose the basis of the eigenspaces are:

1  2  1   2 1  

    3  4       
basis E1 : B1 = 1  basis E6 : B2 =  1 , 3 4   atau B*2 = 3, 3 
1   1   1    3  4  
            
Then A is diagonalizable because for each eigenvalue, geometric
multiplicity = algebraic multiplicity,

1 2 1 
 
P = 1 3 3
1 3 4 

52
Diagonalization
Diagonalization can be used to simplify the following expression:
1. If A is a square matrix and P is an invertible matrix, then:
(P-1AP)2 = P-1APP-1AP = P-1AIAP = P-1A2P

2. Generally, for a positive integer k:

(P-1AP)k = P-1AkP

3. If A is diagonalizable, and P-1AP = D is a diagonal matrix, then:

P-1AkP = (P-1AP)k = Dk

4. Such that Ak is diagonalizable into Dk, by matrix P:

Ak = PDkP-1

53
Exercise 12: diagonalization

Given matrix A
0 0 0
 
A = 0 0 0
3 0 1
 

• Determine if A is diagonalizable.

• If it is, find matrix P that diagonalizes A and find P-1AP.

54
Answer to Exercise 12

The characteristic equation is λ2 (λ - 1) = 0, such that its eigenvalues are λ

= 0 and λ = 1. The basis of the eigenspace corresponding to λ = 0
(Eλ = 0 ) is  1  0 
    
 0 , 1  
− 3 0 
    
0 
  
Basis of Eλ = 1 is .0 
1  
  

 1 0 0 0 0 0
Then    
P =  0 1 0 and P −1 AP = 0 0 0 = D
− 3 0 1  0 0 1 

55
Eigenvalues and Eigenvectors

Eigenspace

Diagonalization

Orthogonal Diagonalization

7.4 Orthogonal
Diagonalization
56
Orthogonal Diagonalization
Theorem 7.7:
If A is a square matrix, then the following sentences are equivalent:
a. A is orthogonally diagonalizable
b. A has an orthonormal set that consists of n eigenvectors
c. A is symmetric

Note: generally, symmetric matrix is diagonalizable.

57
Example 6: Orthogonal Diagonalization

Characteristic equation of A:
 − 4 − 2 −2 
 
det(I − A) = det  − 2  − 4 − 2  = ( − 2)2 ( − 8) = 0
 − 2 − 2  − 4

Such that, eigenvalues of A is λ1,2 = 2 and λ3 = 8. Find the eigenspaces of each

eigenvalue. After finding the eigenspaces, we have

− 1  − 1 
Vectors u1 =   and u2 =   form a basis of Eλ = 2.
1  0
 0   1 

Apply the Gram-Schmidt process on {u1,u2} such that

− 1  − 1 
 2  6
We obtain an orthonormal set: {v1 , v2 } where v1 =  1  dan v2 = − 1 
 2  6
 0   2 
 6 

58
Orthogonal Diagonalization (cont)
1  
  
The eigenspace of λ = 8 has basis B = { u3 }= 1  
1    1 
    3 
 1  
Apply normalization to {u3} such that it produces: B = {v3 }= 
 1 3  
 3  


Finally, use v1, v2, and v3 as columns, we obtain:

− 1 −1 1 
 2 6 3
P= 1 −1 1 
 2 6 3  that diagonalizes A orthogonally.
 0 2 1 
 6 3 

59
Orthogonal Diagonalization Procedure

The following steps is a procedure to diagonalize a symmetric matrix A

orthogonally:
1. For every eigenvalue of A, find the basis for its eigenspaces.
2. Apply the Gram-Schmidt process and normalize each of the basis to
form an orthonormal basis.
3. Form matrix P where the columns are vectors from basis in number 2.
then P can diagonalize A orthogonally

60
Reflection

Make a diagram that explains the relation between concepts you have
learned, i.e. Vectors in 2-space and 3-space, vector and matrix
multiplication, homogeneous SLE, determinant, polynomial equation,
eigenvalue, eigenvector, eigenspace, diagonalization.

61
Summary

Summarize the material you have just learned.

Check your summary, are the following key concepts mentioned in your
summary?
❑ eigenvalue
❑ eigenvector
❑ Characteristic equation
❑ eigenspace
❑ Geometric multiplicity
❑ Algebraic multiplicity
❑ diagonalization
❑ Orthogonal diagonalization

62
Post-test modul

Copyright © 2004-2005 NameOfTheOrganization. All rights reserved.

Post-test

1. How do you find eigenvalues of A?

2. Given matrix A and eigenvalue k, how do you find its eigenspace?

3. Does the eigenspace only consists of eigenvectors?

4. Two matrices have the same characteristic equation, what can you
conclude? Do they have the same eigenspaces?

5. There is a basis of Rn that consists of eigenvectors of A. What can you

conclude about A?

6. A is orthogonally diagonalizable, what can you conclude?

64
Congrats, you have completed module 7.
Prepare for the next module