TOPICS OF ADVANCED MATHEMATICS FOR ECONOMICS – 2020/2021

Sheet 1. Diagonalization

1-1. Given the matrix  

2 4
A=
3 1
compute its eigenvalues, eigenvectors and diagonalize A.

Solution: The characteristic polynomial is

2−λ 4
|A − λI| = = λ2 − 3λ − 10
3 1−λ
The roots are √
3± 9 + 40
λ= = −2, 5
2
There are two different eigenvalues. The matrix is diagonalizable.
The eigenspace S(5) is the solution to the system of linear equations
  
−3 4 x
=0
3 −4 y
which is equivalent to the following one
3x − 4y = 0.
Therefore, S(5) < (4, 3) >.
The eigenspace S(−2) is the solution to the system of linear equations
  
4 4 x
=0
3 3 y
which is equivalent to the following one
x + y = 0.
Therefore, S(−2) =< (1, −1).
We conclude that A = P DP −1 with
   
5 0 4 1
D= P =
0 −2 3 −1

1-2. Given the following matrices

     
4 6 0 1 0 −2 4 5 −2
A =  −3 −5 0  B= 0 0 0  C =  −2 −2 1 
−3 −6 1 −2 0 4 −1 −1 1
(a) Compute its eigenvalues, eigenvectors and the eigenspaces.
(b) Diagonalize them, whenever possible.

Solution:
First, we find the eigenvalues of A. The characteristic polynomial is
4−λ 6 0
|A − λI| = −3 −5 − λ 0 = −(λ + 2)(λ − 1)2
−3 −6 1−λ
so the eigenvalues ar λ1 = −2 with multiplicity n1 = 1 and λ2 = 1 with multiplicity n2 = 2.
Now we compute the eigenspace S(1). We solve the following system of linear equations
 

3 6 0

(A − I) x y z =  −3 −6 0  x y z = 0
−3 −6 0
o sea,
3x + 6y = 0
−3x − 6y = 0
1
2

using y as the parameter, the solution is S(1) = {−2y, y, z) : y, z ∈ R} =< (−2, 1, 0), (0, 0, 1) > so dim S(1) =
2.
On the other hand, S(−2) is the set of solutions to the system of linear equations
 

6 6 0

(A + 2I) x y z =  −3 −3 0  x y z = 0
−3 −6 3
o sea,
6x + 6y =0
−3x − 3y =0
−3x − 6y + z = 0
so S(−2) = {−z, z, z) : z ∈ R} =< (−1, 1, 1) >.
The matrix A is diagonalizable and A = P DP −1 with
   
1 0 0 0 2 −1
D= 0 1 0  P = 0 −1 1 
0 0 −2 1 0 1
The characteristic polynomial of B is
1−λ 0 −2
|B − λI| = 0 −λ 0 = −λ2 (λ − 5)
−2 0 4−λ
so its eigenvalues are λ1 = 0 with multiplicity n1 = 2 and λ2 = 5 with multiplicity n2 = 1.
We compute S(0) by solving the linear system of equations
 

1 0 −2

(b − 0I) x y z =  0 0 0  x y z = 0
−2 0 4
that is,
x − 2z =0
−2x + 4z =0
We use y and z as parameters. The solution is x = 2z. Hence, S(0) = {2z, y, z) : y, z ∈ R} =<
(2, 0, 1), (0, 1, 0) >, so dim S(0) = 2.
Now, S(5) is the set of solutions to the system of linear equations
 

−4 0 −2

(B − 5I) x y z = 0 −5 0  x y z =0
−2 0 −1
that is,
−4x − 2z =0
−5y =0
−2x − z =0
Using x as the parameter, we find that y = 0, z = −2x. Hence, S(5) = {x, 0, −2x) : x ∈ R} =< (1, 0, −2) >.
The matrix B is diagonalizable and B = QDQ−1 with
   
0 0 0 0 2 1
D= 0 0 0  Q= 1 0 0 
0 0 5 0 1 −2
The characteristic polynomial of C. is
4−λ 5 −2
|C − λI| = −2 −2 − λ 1 = −(λ − 1)3
−1 −1 1−λ
so there is only one eigenvalue λ1 = 1 with multiplicity n1 = 3. The space S(1) is the set of solutions to the
system of linear equations
 

3 5 −2

(C − I) x y z =  −2 −3 1  x y z = 0
−1 −1 0
 3

that is,
3x + 5y − 2z = 0
−2x − 3y + z = 0
−x − y =0
the solution is y = −x, z = −x. Using z as the parameter S(1) = {−z, z, z) : z ∈ R} =< (−1, 1, 1) > so
dim S(1) = 1 < n1 = 3 and the matrix C is not diagonalizable.

1-3. What are the values of a for which the matrix

 
1 0 0
A= a 1 0 
1 1 2
is diagonalizable?

Solution: The characteristic polynomial is

1−λ 0 0
|A − λI| = a 1−λ 0 = (1 − λ)2 (2 − λ)
1 1 2−λ
There are two eigenvalues λ1 = 1, with multiplicity 2 and λ2 = 2 with multiplicity 1.
The matrix is diagonalizable if and only if dim S(1) = 2. The space S(1)is the set of solutions to the system
of linear equations     
x 0 0 0 x
(A − λI)  y  =  a 0 0   y  = 0
z 1 1 1 z
which is the same as
ax =0
x+y+z =0
If a 6= 0 the solutions are x = 0, y = −z. Hence, S(1) = {(0, −z, z) : z ∈ R} and we see that dim S(1) = 1.
Hence, if a 6= 0 A is not diagonalizable.
But, if a = 0, the system becomes
x+y+z =0
so S(1) = {(x, y, −x − y) : x, y ∈ R} and dim S(1) = 2. In this case, A is diagonalizable.

1-4. Show that

(a) If A is a diagonalizable matrix, so is An for each n ∈ N.
(b) A diagonalizable matrix A is regular if and only if none of its eigenvalues vanishes.
(c) If A has an inverse, then both A and A−1 have the same eigenvectors and the eigenvalues of A are the
reciprocal of the eigenvalues of A−1 .
(d) A and At have the same eigenvalues.
t
Solution: Since, |At − λI| = (A − λI) = |A − λI| the characteristic polynomials of A and At are the
same. Therefore, the eigenvalues are the same.
 
5 0 0
1-5. Study for which values of a and b the matrix A =  0 −1 a  is diagonalizable.
3 0 b
Solution: The characteristic polynomial is
5−λ 0 0
−1 − λ a
|A − λI| = 0 −1 − λ a | = (5 − λ) = (5 − λ)(1 + λ)(b − λ)
0 b−λ
3 0 b−λ
So, the eigenvalues are λ1 = 5, λ2 = −1 y λ3 = b. If b 6= 5 y b 6= −1 there are three different eigenvalues and
the matrix is diagonalizable.
If b = 5 then λ1 = 5 has multiplicity n1 = 2 and the other eigenvalue has multiplicity 1. The matrix is
diagonalizable or not depending on the dimension of S(5). This space is the set of solutions to the system of
linear equations     
x 0 0 0 x
(A − 5I)  y  =  0 −6 a   y  = 0
z 3 0 0 z
4

which is the same as

−6y + az =0
3x =0
Clearly, dim S(5) = 1 < n1 = 2, so A is not diagonalizable.
On the other hand, if b = −1 the eigenvalues are λ1 = 5, with multiplicity n1 = 1 and λ2 = −1 with
multiplicity n2 = 2. Now The matrix is diagonalizable or not depending on the dimension of S(−1). This
space is the set of solutions to the system of linear equations
    
x 6 0 0 x
(A + I)  y  =  0 0 a   y =0
z 3 0 0 z
that is,
6x =0
az =0
3x =0
and we see that 
1, si a 6= 0;
dim S(−1) =
2, si a = 0.
We could have done this in an easier way, by noting that dim S(−1) = 3 − rg(A + I) and
   
6 0 0 6 0 0 
2, if a 6= 0;
rg(A + I) = rg  0 0 a  = rg  0 0 a =
1, if a = 0.
3 0 0 0 0 0
Thus, if 
 b = 5, then A is not diagonalizable;
b = −1, then A is diagonalizable and only if a = 0;
b 6= 5 y b 6= −1, then A is diagonalizable.


1-6. Which of the following matrices are diagonalizable?

 
1 2 0    
−2 1 1 1
A =  −1 3 1  B= C
1 0 0 1
0 1 1
Solution: The characteristic polynomial of A is
1−λ 2 0
|A − λI| = −1 3−λ 1 = −(λ − 1)(λ − 2)2
0 1 1−λ
so the eigenvalues are λ1 = 1 with multiplicity n1 = 1 and λ2 = 2 with multiplicity n2 = 2.
The space S(2) is the set of solutions to the system of linear equations
    
x −1 2 0 x
(A − 2I)  y  =  −1 1 1   y  = 0
z 0 1 −1 z
that is,
−x + 2y =0
−x + y + z =0
y−z =0
The solution is y = z, x = 2y = 2z (z is the parameter). Hence, S(2) = {(2z, z, z) : z ∈ R} =< (2, 1, 1) > so
dim S(2) = 1 < n2 = 2 and A is not diagonalizable.
The characteristic polynomial of B is
−2 − λ 1
|B − λI| = = λ2 + 2λ − 1
1 −λ
so the eigenvalues are √ √
λ1 = −2+√ 4+4
= −1 + 2
2 √
λ2 = −2−2 4+4 = −1 − 2
all of the with multiplicity 1. Hence, B is diagonalizable.
 5

√
The space S(−1 + 2) is the set of solutions to the system of linear equations
√
 
√  x
    x
 −1 − 2 1√
A − (−1 + 2)I =  y 
y 1 1− 2
=0
that is, √
−(1 + 2)x√+y =0
x + (1 − 2)y = 0
√ √ √ √
the solution
√ is x = y/(1 + 2). √Hence, S(−1 + √2) = {(y/(1 + 2), y) : y ∈√R} =< (1/(1 + 2),
√1) >=<
(1, 1 + 2) >. Likewise, S(−1 + 2) = {(y/(1 − 2), y) : y ∈ R} =< (1/(1 − 2), 1) >=< (1, 1 − 2) >.
The diagonal form of B is B = P DP −1 with
   √ 
1√ 1√ −1 + 2 0√
P = D=
1+ 2 1− 2 0 −1 − 2
Finally, the characteristic polynomial of C is
 
1−λ 1
|C − λI| = = (λ − 1)2
0 1−λ
so there is a unique eigenvalue λ = 1 with multiplicity 2. The eigenspace S(1) is the set of solutions to the
system of linear equations   
0 1 x
=0
0 0 y
so y = 0. Hence, S(1) = {(x, 0) : x ∈ R} =< (1, 0) > and dim S(1) = 1. Therefore, the C is not diagonalizable.

 
1 0 0
1-7. The matrix  α + 1 2 0  is diagonalizable if and only α is...
0 α+1 1
Solution: The characteristic polynomial of
 
1 0 0
A= α+1 2 0 
0 α+1 1
is (λ − 1)2 (λ − 2). The eigenvalues are λ1 = 1 with multiplicity n1 = 2 and λ2 = 2 with multiplicity n2 = 1.
The matrix A is diagonalizable if and only if dim S(1) = 2. The subspace S(1) is the set of solutions to the
system of linear equations 
(α + 1)x + y = 0
(α + 1)y =0
If α 6= −1 the solution is x = y = 0. That is, S(1) = {(0, 0, z) : z ∈ R} and dim S(1) = 1. Therefore, if
α 6= −1 then A is not diagonalizable.
If α = −1 the linear system above reduces to y = [Link] this case, S(1) = {(x, 0, z) : x, z ∈ R} and
dim S(1) = 2. So, if α = −1 the matrix A is diagonalizable.

1-8. Consider the matrices

     
3 2 0 2 0 0 1 0 0
A =  −1 0 0  B= 1 1 −1  C= 0 1 1 
0 0 1 1 −1 1 0 0 2
Find whether they are diagonalizable and, whenever they are, compute their n-th power.

Solution: Let  
3 2 0
A =  −1 0 0 
0 0 1
The eigenvalues are λ1 = 1 with multiplicity n1 = 2 and λ2 = 2 with multiplicity n2 = 1. The eigenspaces
are S(1) =< (0, 0, 1), (−1, 1, 0) > y S(2) =< (−2, 1, 0) >. The matrix A is diagonalizable A = P DP −1 with
   
1 0 0 0 −1 −2
D= 0 1 0  P = 0 1 1 
0 0 2 1 0 0
6

So,
   −1
0 −1 −2 1 0 0 0 −1 −2
An =  0 1 1  0 1 0  0 1 1  =
n
1 0 0 0 0 2 1 0 0
−1 + 21+n −2 + 21+n
     
0 −1 −2 1 0 0 0 0 1 0
= 0 1 1   0 1 0   1 2 0  =  1 − 2n 2 − 2n 0 
n
1 0 0 0 0 2 −1 −1 0 0 0 1
Let  
2 0 0
B= 1 1 −1 
1 −1 1
The eigenvalues are λ1 = 0 with multiplicity n1 = 1 and λ2 = 2 with multiplicity n2 = 2. The eigenspaces
are S(0) =< (0, 1, 1) > and S(2) =< (1, 0, 1), (1, 1, 0) >. The matrix B is diagonalizable: B = P DP −1 with
   
0 0 0 0 1 1
D= 0 2 0  P = 1 0 1 
0 0 2 1 1 0
so,
   −1
0 1 1 0 0 0 0 1 1
B n =  1 0 1   0 2n 0   1 0 1 
1 1 0 0 0 2n 1 1 0
2n
     
0 1 1 0 0 0 −1 1 1 0 0
1
= 1 0 1   0 2n 0   1 −1 1  =  2n−1 2n−1 −2n−1 
2
1 1 0 0 0 2n 1 1 −1 2n−1 −2n−1 2n−1
The eigenvalues of  
1 0 0
C= 0 1 1 
0 0 2
are λ1 = 1 with multiplicity n1 = 2 and λ2 = 2 with multiplicity n2 = 1. The eigenspaces are S(1) =<
(1, 0, 0), (0, 1, 0) > and S(2) =< (0, 1, 1) >. The matrix C is diagonalizable: C = P DP −1 with
   
1 0 0 1 0 0
D= 0 1 0  P = 0 1 1 
0 0 2 0 0 1
so,
   −1
1 0 0 1 0 0 1 0 0
Cn =  0 1 1   0 1 0   0 1 1 
0 0 1 0 0 2n 0 0 1
     
1 0 0 1 0 0 1 0 0 1 0 0
=  0 1 1   0 1 0   0 1 −1  =  0 1 2n − 1 
0 0 1 0 0 2n 0 0 1 0 0 2n

1-9. The following are the characteristic polynomials of some square matrices. Determine which of them corre-
spond to diagonalizable matrices.
p(λ) = λ2 + 1 p(λ) = λ2 − 1
2
p(λ) = λ + α p(λ) = λ2 + 2αλ + 1
p(λ) = λ + 2λ + 1 p(λ) = (λ − 1)3
2

p(λ) = λ3 − 1

Solution:
1) p(λ) = λ2 + 1. The matrix is not diagonalizable because not all the roots are real numbers.
2) p(λ) = λ2 + α. If α > 0 the matrix is no diagonalizable because not all the roots are real numbers. If
α < 0 the characteristic polynomial has two different real roots, so the matrix is diagonalizable. If α = 0
there is a unique eigenvalue 0 with multiplicity 2. Hence, either all the entries in the matrix are 0, or else the
matrix is no diagonalizable.
3) p(λ) = λ2 + 2λ + 1 = (λ + 1)2 . We see that −1 is a double root. Therefore, either the matrix is −I , or
else the matrix is no diagonalizable.
4) p(λ) = λ3 − 1 = (λ − 1)(λ2 + λ + 1) has no real roots. The matrix is no diagonalizable.
 7

5) p(λ) = λ2 − 1 has two distinct real roots. The √

matrix is diagonalizable.
6) p(λ) = λ2 + 2αλ + 1. The roots are λ = −α ± α2 − 1. Thus,
• If |α| > 1, the matrix is diagonalizable.
• If |α| < 1, the matrix is not diagonalizable.
• If |α| = 1, we are in case 3).

1-10. Determine whether the following matrices are diagonalizable. Compute the n-th power whenever they are
diagonalizable.  
    1 2 3
α 0 α 1
A= B= C= 0 1 2 
1 α 1 α
0 0 1
Solution:
1) The matrix A is of order 2 and its unique eigenvalue is α of multiplicity 2. Therefore, A is not diagonalizable.
2) The characteristic polynomial of B is (λ − α)2 − 1. The roots are α ± 1 so B is diagonalizable. The
eigenvalues are
S(α − 1) =< (−1, 1) >, S(α + 1) =< (1, 1) >
−1
and B = P DP con    
−1 1 α−1 0
P = D=
1 1 0 α+1
Thus,
−1
(α − 1)n
  
−1 1 0 −1 1
Bn = =
1 1 0 (α + 1)n 1 1
(α − 1)n (α − 1)n + (α + 1)n −(α − 1)n + (α + 1)n
     
1 −1 1 0 −1 1 1
=
2 1 1 0 (α + 1)n 1 1 2 −(α − 1)n + (α + 1)n (α − 1)n + (α + 1)n
3) The eigenvalues of  
1 2 3
C= 0 1 2 
0 0 1
are λ1 = 1 with multiplicity n1 = 3. Since,
S(1) =< (1, 0, 0) >
the matrix is not diagonalizable.

1-11. Study for what values of the parameters the following matrices are diagonalizable. Find the eigenvalues and
eigenvectors.    
a b 0 1 −2 −2 − α
A =  0 −1 0  B= 0 1 α 
0 0 1 0 0 1
1-12. The matrix  
a 1 p
 b 2 q 
c −1 r
has (1, 1, 0), (−1, 0, 2) and (0, 1, −1) as eigenvectors. Compute its eigenvalues.

1-13. Determine whether the following matrices are diagonalizable. If possible, write their diagonal form.
     
5 4 3 −2 −1 −1 5 7 5
A =  −1 0 −3  B =  1 0 1  C =  −6 −5 −3 
1 −2 1 0 0 −1 4 1 0
     
3 2 0 −1 2 −2 5 −10 8
D =  −1 2 0  E =  0 2 0  F =  −10 2 2 
0 0 3 0 3 −2 8 2 11
     
1 −1 2 2 0 3 3 −1 0
G =  0 3 2  H =  0 1 0  I =  −1 3 0 
0 1 4 −1 0 −2 0 0 2
     
1 0 0 −1 2 −2 −9 1 1
J =  1 2 0  K =  0 2 0  L =  −18 0 3 
1 0 2 0 3 −2 −21 4 0
8

Solution:
1) The eigenvalues of  
5 4 3
A =  −1 0 −3 
1 −2 1
are −2, 4, 4. Also, S(−2) =< (−1, 1, 1) >, S(4) =< (1, −1, 1) >, so the matrix is diagonalizable.
2) The eigenvalues of  
−2 −1 −1
B= 1 0 1 
0 0 −1
are −1, −1, −1. Since, B is not already in diagonal form, it is not diagonalizable.
5) The eigenvalues of  
−1 2 −2
E= 0 2 0 
0 3 −2
are −2, −1, 2. Since they are all distinct then E is diagonalizable. Also, E = P DP −1 with
   
2 1 2 −2 0 0
P =  0 0 12  D =  0 −1 0 
1 0 9 0 0 2
12) The eigenvalues of  
−9 1 1
L =  −18 0 3 
−21 4 0
are −3, −3, −3. Since L is not already in diagonal form, it is not diagonalizable.