Week 2: Exercises&Solutions

Date: September 17, 2024

Exercise 1. Consider the matrix  

2 −1
A= . (1)
−1 2
(a). Find the eigenvalues and eigenvectors of A.
(b). Compute A2 and its eigenvalues and eigenvectors.
(c). Compute A−1 and its eigenvalues and eigenvectors.
(d). Compute A + 4I and its eigenvalues and eigenvectors.
Solution 1. (a). Firstly, compute the eigenvalues of A. The eigenvalues of A comes from det(A−λI) = 0:
 
2 − λ −1
det(A − λI) =   = λ2 − 4λ + 3 = (λ − 1)(λ − 3) = 0 (2)
−1 2 − λ
The leads to the eigenvalues of A: λ1 = 1 and λ2 = 3.
Secondly, we compute the eigenvectors of A.
      
1 −1 x 0 1
λ1 = 1 : (A − I)x1 =     =   gives the eigenvector is x1 =   (3)
−1 1 y 0 1
      
−1 −1 x 0 1
λ2 = 3 : (A − 3I)x1 =     =   gives the eigenvector is x2 =   (4)
−1 −1 y 0 −1
(b). Suppose λ and x are eigenvalue and eigenvector of A: Ax = λx, then
A2 x = A(Ax) = Aλx = λAx = λ2 x (5)
 
1
Therefore, the eigenvalues are λ21 = 1 and λ22 = 9; the eigenvector keeps the same direction: x1 =  
1
 
1
and x2 =  .
−1
(c). Suppose λ and x are eigenvalue and eigenvector of A: Ax = λx , then
1
x = A−1 Ax = A−1 λx = λA−1 x =⇒ A−1 x = x (6)
λ
Therefore,  are 1/λ1 = 1 and 1/λ2 = 1/3; the eigenvectors keep the same direction:
  the eigenvalues
1 1
x1 =   and x2 =  .
1 −1

1
 (d). Suppose λ and x are eigenvalue and eigenvector of A: Ax = λx , then
(A + 4I)x = Ax + 4Ix = λx + 4x = (λ + 4)x (7)
Therefore,  are λ1 + 4 = 5 and λ2 + 4 = 7; the eigenvectors keep the same direction:
  the eigenvalues
1 1
x1 =   and x2 =  .
1 −1

Exercise 2. Consider the matrix

 
1 −1 0
 
A=
−1 .
−1
2 (8)
0 −1 1
(a). Find the eigenvalues and eigenvectors of A.
(b). Digonalize the matrix A.
(c). Compute A100 .

Solution 2. (a). First, compute the eigenvalues of A:

1−λ −1 0
det(A − λI) = −1 2−λ −1 = λ(λ − 1)(3 − λ) = 0 (9)
0 −1 1−λ
The eigenvalues are 0, 1 and 3.
For λ1 = 0, the corresponding eigenvector
      
1 −1 0 x 0 1
     1  
(A − 0I)x1 =  −1 2 −1 y  = 0
    gives the eigenvector is x1 = √  1 . (10)
3 
0 −1 1 z 0 1
For λ1 = 1, the corresponding eigenvector
      
0 −1 0 x 0 1
     1  
(A − I)x2 =  −1 1 −1 y  = 0
    gives the eigenvector is x2 = √  0 . (11)
2 
0 −1 0 z 0 −1
For λ1 = 3, the corresponding eigenvector
      
−2 −1 0 x 0 1
     1  
(A − I)x3 = −1 −1 −1 y  = 0
     gives the eigenvector is x3 = √ −2 . (12)
 6 
0 −1 −2 z 0 1
(b). Because A is symmetric, then we can concatenate all orthonormal vector to obtain eigenvector matrix.
 √ √ √ 
1/ 3 1/ 2 1 6
 √ √ 
Q= 1/ 3 0 −2/ 6. (13)
√ √ √
1/ 3 −1/ 2 1/ 6

2
 Then we can digonalize matrix A:
 
0 0 0
Λ = QT AQ = 
 
0 1 0
 (14)
0 0 3
(c).
A100 = (QΛQT )100 = QΛQT QΛQT · · · QΛQT = QΛ100 QT (15)
 √ √ √   √ √ √ 
1/ 3 1/ 2 1 6 0 0 0 1/ 3 1/ 3 1 3
 √ √   √ √ 
=1/ 3 0 −2/ 6  0 1
 0   1/ 2
 0 −1/ 2 (16)
√ √ √ √ √ √

1/ 3 −1/ 2 1/ 6 0 0 3 100 1/ 6 −2/ 6 1/ 6
 99 99

3 +1 3 −1
−399
 2 2

=  −3
 99 2×3 99 −399  (17)

399 −1 99 399 +1
2 −3 2

Exercise 3. Consider the matrix  

3 2 2
A= . (18)
2 3 −2
(a). Compute AT A.
(b). Find the eigenvalues and eigenvectors of AT A
(c). Find the eigenvalues and eigenvectors of AAT
(d). Find the Singular Value Decomposition (SVD) of A.

Solution 3. (a).
   
3 2   13 12 2
 3 2 2
AT A = 
  
 = 12 13 −2 (19)
2 3 

2 3 −2
 
2 −2 2 −2 8
(b). First, compute the eigenvalue of AT A:
 
13 − λ 12 2
 = −λ3 + 34λ2 − 225λ = −λ(λ − 25)(λ − 9) = 0
 
det(A − λI) =   12 13 − λ −2 
2 −2 8−λ
(20)
So the eigenvalues are 25, 9 and 0. For λ1 = 25,
 
−12 12 2
AT A − 25I = 
 
 12 −12 −2 
 (21)
2 −2 −17

3
    √ 
1 −1 0 1/ 2
   √ 
0 0 1. A unit-length eigenvector is v1 = 1/ 2. For λ = 9,
which row reduces to    

0 0 0 0
 
4 12 2
T
 
A A − 25I = 12 4 −2

, (22)
2 −2 −1
   √ 
1 0 −1/4 1/ 18
. A unit-length eigenvector is v2 = −1/√18 . For λ = 0,
   
which row-reduces to  0 1 1/4
√
   
0 0 0 4/ 18
 
13 12 2
T
 
A A − 0I =  12 13 −2 ,
 (23)
2 −2 8
   
1 0 2 2/3
   
0 1 −2 . A unit-eigevector is v3 = −2/3 .
which reduces to    

0 0 0 −1/3
(c).
 
  3 2  
3 2 2 17 8
AAT = 
 
 2 3  =  . (24)
2 3 −2 8 17
 
2 −2
The det(AAT − λI) = λ2 − 34λ + 225 = (λ − 25)(λ − 9) = 0, so the eigenvalues are λ1 = 25
√ √
and λ2 = 9. And the singular values are σ1 = 25 = 5 and σ2 = 9 = 3. We can compute ui by
σu = Avi :
 √ 
1/ 2
u1 = Av1 /σ1 =  √  (25)
1/ 2
 √ 
1/ 2
u2 = Av2 /σ2 =  √  (26)
−1/ 2
(d). The SVD of A is
 1/√2 √
 
 √ √  1/ 2 0
1/ 2 1/ 2 5 0 0  √ √ √ 
A = U ΣV T = √ √   1/ 18 −1/ 18 4/ 18 . (27)
1/ 2 −1/ 2 0 3 0
 
2/3 −2/3 −1/3