Solution Assignment 1-Linear Algebra

AIMS Senegal-2024-2025
December 2, 2024

Solution-Exercise 1.

1. Two examples of subsets of R2 which are not subspaces:

• Let
C = {(x, y) ∈ R2 | x2 + y 2 = 1}.
We have (1, 0) ∈ C and (0, 1) ∈ C, since:

12 + 02 = 1 and 02 + 12 = 1.

However, their sum is:

(1, 0) + (0, 1) = (1, 1),

and:
12 + 12 = 2 ̸= 1.
Therefore, (1, 1) is not on the unit circle, so C is not closed under addition.
We could also remark that the unit circle does not contain the zero vector
(0, 0). Thus, C is not a subspace of R2 .
• Let
S = {(x, y) ∈ R2 | x, y ∈ Z}.
We have (1, 0) ∈ S. Multiply it by the scalar 21 gives:
 
1 1
· (1, 0) = ,0 ∈/ S.
2 2

Therefore, S is not closed under scalar multiplication. Thus, S is not a

subspace of R2 .

2. An example of subspaces X and Y of R2 such that X ∪ Y is not a subspace:

Let X = {(x, 0) : x ∈ R} and Y = {(0, y) : y ∈ R}. X and Y of R2 are subspaces
of R2 . However, X ∪ Y is not a subspace of R2 . indeed, (1, 0) ∈ X ⊂ X ∪ Y and
(0, 1) ∈ Y ⊂ X ∪ Y . But (1, 0) + (0, 1) = (1, 1) ∈
/ X ∪Y.

1
 3. Let S = {v1 , v2 , v3 , . . . , vn , . . . } be an infinite set of vectors. Assume S is linearly
independent. Let F = {v1 , v2 , . . . , vn } be a finite subset of S. Let λ1 , λ2 , . . . , λn ∈
X n
K such that λi vi = 0. Since F is a finite subset of the infinite independent set
i=1
S, then, we have λ1 = λ2 = · · · = λn = 0. Hence, F is linearly independent.
Now, assume that every finite subset of S is linearly independent.
Suppose by contradiction that S is not linearly independent. This means that
there exists a finite subset F = {v1 , v2 , . . . , vn } of S and scalars λ1 , λ2 , . . . , λn such
that not all of the λi are 0 and

λ1 v1 + λ2 v2 + · · · + λn vn = 0.

The set {v1 , v2 , . . . , vn } is a finite subset of S, and by assumption, every finite

subset of S is linearly independent. Therefore, the only solution to the equation

λ1 v1 + λ2 v2 + · · · + λn vn = 0

must be the trivial solution λ1 = λ2 = · · · = λn = 0, which contradicts our

assumption that at least one of the λi ’s is nonzero.
Thus, our assumption that S is not linearly independent leads to a contradiction,
so S must be linearly independent.

4. Let’s show that if V = Y1 + · · · + Ym , then the sum is a direct sum if and only if
the subspaces Y1 , . . . , Ym are linearly independent.

• Assume that the V = Y1 · · · Ym . Let yi ∈ Yi , i = 1, . . . , m such that

L L
Xm
yi = 0. Since V is a direct sum, then, Y1 ∩ · · · ∩ Yn = {0}. This means
i=1
that y1 = · · · = yn = 0.
• Assume that Y1 , . . . , Ym are linearly independent. That is if yi ∈ Yi , i =
Xm
1, . . . , m such that yi = 0, then y1 = · · · = yn = 0. This implies that
i=1 L L
Y1 ∩ · · · ∩ Yn = {0}. Therefore, V = Y1 · · · Ym .

Solution-Exercise 2.

Let E and F be two R-vectors spaces and φ a linear map defined from E to F . Let
A := {x1 , x2 , . . . , xn } ⊂ E.

1. Assume that A is linearly dependent. Then, there are scalars λ1 , λ2 , . . . , λn ∈ R

n
X n
X X n
such that not all of the λi are 0, and λi xi = 0. Thus, we have λi φ( λ i xi ) =
i=0 i=0 i=0
n
X
0. By the linearity of φ, we obtain λi φ(xi ) = 0. Since at least one of the λi ̸= 0,
i=0
we conclude that φ(A) = {φ(x1 ), φ(x2 ), . . . , φ(xn )} is linearly dependent.

2
 2. Assume that φ(A) is linearly independent. Let λ1 , λ2 , . . . , λn ∈ R such that
Xn n
X
λi xi = 0. By the linearity of φ, we have λi φ(xi ) = 0. Since φ(A) is
i=0 i=0
linearly independent, we conclude that λ1 = λ2 = · · · = λn = 0. Therefore, A is
linearly independent.

3. Assume that A is linearly independent and φ is injective. Let λ1 , λ2 , . . . , λn ∈ R

n
X Xn
such that λi φ(xi ) = 0. By the linearity of φ, we obtain φ( λi xi ) = 0 = φ(0).
i=0 i=0
n
X
Since φ is injective, this implies that λi xi = 0. The fact that A is linearly
i=0
independent implies that λ1 = λ2 = · · · = λn = 0. Therefore, φ(A) is linearly
independent.

Solution-Exercise 3.

E = R3 with canonical basis β = (e1 , e2 , e3 ). u is an endomorphism of R3 define by

the images of the vectors of β:

u(e1 ) = −2e1 + 2e3 , u(e2 ) = 3e2 , u(e3 ) = −4e1 + 4e3 .

1. The matrix of u relative to the canonical basis of R3 : we know that

u(e1 ) = (−2, 0, 2), u(e2 ) = (0, 3, 0), u(e3 ) = (−4, 0, 4).

Then,  
−2 0 −4
M (u) = (u(e1 ), u(e2 ), u(e3 )) =  0 3 0  .
2 0 4

2. We have
ker(u) = {X ∈ R3 : AX = 0}.
 
x
Let X = y . Then, AX = 0 implies that
z

 −2x − 4z = 0
3y = 0
2x + 4z = 0.


 
−2z
Then, x = −2z and y = 0. Thus X =  0 . Therefore,
z
 
 −2 
ker(u) = span  0  .
1
 

3
  
−2
Then, a basis of ker(u) is  0 .
1
 
0
Since, ker(u) ̸= 0, then, u is not injective.
0
 
 −2x − 4z = 1 
The equation 3y = 0 does not have a solution.
2x + 4z = 0
 
 
1
−1  
That is u 0 = {}. Therefore, u is not surjective.
0

3. A basis of Im(u) : consider

 
−2 0 −4 L1
0 3 0 L2
2 0 4 L3 .

We have
L′1 = L1
 
−2 0 −4
 0 3 0  L′2 = L2
0 0 0 L′3 = L3 + L1 .
This means that the first colon and the third one 
arelinearly
 dependent.
 We
 −1 0 
conclude that the rank of u is 2 and we can choose  0  , 3 as a basis
1 0
 
of Im(u).

4. Let A be the matrix given by the elements of the basis of ker(u) and Im(u). Then,
−2 −1 0
|A| = 0 0 3 = 3 ̸= 0. Then, the 3 vectors form a linearly independent
1 1 0
system. Since dim(E) = dim(R3 ) = 3, therefore, E = ker(u) Im(u).
L

Solution-Exercise 4.

1. A spanning set (or a generator) of the following subset of R3 :

• F = {(x, y, z) ∈ R3 : x+2y −z = 0}. Let (x, y, z) ∈ F . Then, x+2y −z = 0.

That is z = x + 2y. This implies that

(x, y, z) = (x, y, x + 2y) = x(1, 0, 1) + y(0, 1, 2).

Therefore,     
 1 0 
F = span 0 , 1 .
1 2
 

4
 • G = {(x, y, z) ∈ R3 : x − y + z = 0 and 2x − y − z = 0}. Let (x, y, z) ∈ G.
Then, 
x − y + z = 0 L1
2x − y − z = 0 L2
2L1 − L2 gives y = 3z. L2 − L1 gives x = 2z. Thus (x, y, z) = (2z, 3z, z) =
z(2, 3, 1). Therefore,  
 2 
F = span 3 .
1
 

2. The rank of the matrix

 
−2 3 0 0
−3 4 0 0
A=  ∈ M4 (R).
1 1 1 0
9 −5 −1 3
We have
−2 3 0 0
−3 4 0 0
det(A) = = 3 ̸= 0.
1 1 1 0
9 −5 −1 3
Then, the rank of A is 4.
(a) Invertibility of A. Since the rank of A is maximal, therefore, A is invertible.
The inverse of A is given by
1
A−1 = Cof(A)t .
det(A)
We have
 
4 0 0 −3 0 0 −3 4 0 −3 4 0
 1
 1 0 − 1 1 0 1 1 0 − 1 1 1 

 −5 −1 3 9 −1 3 9 −5 3 9 −5 −1 
 

 3 0 0 −2 0 0 −2 3 0 −2 3 0 

− 1 1 0 1 1 0 − 1 1 0 1 1 1 
 
 −5 −1 3 9 −1 3 9 −5 3 9 −5 −1 
Cof(A) =  .
 3
 0 0 −2 0 0 −2 3 0 −2 3 0 

 4
 0 0 − −3 0 0 −3 4 0 − −3 4 0 

 −5 −1 3 9 −1 3 9 −5 3 9 −5 −1 
 

 3 0 0 −2 0 0 −2 3 0 −2 3 0 

 − 4 0 0 −3 0 0 − −3 4 0 −3 4 0 
1 1 0 1 1 0 1 1 0 9 1 1
By simple computations, we get
 
12 9 −21 −28
−9 −6 15 22 
Cof(A) = 
0
.
0 3 1 
0 0 0 1

5
 Therefore,  
12 −9 0 0
1 9 −6 0 0
A−1 =  .
3  −21 15 3 0
−28 22 1 1
(b) The eigenvalues and the eigenvectors of A. The characteristic polynomial of
A is:
−2 − x 3 0 0
−3 4−x 0 0
PA (x) = det(A − xI4 ) =
1 1 1−x 0
9 −5 −1 3 − x
4−x 0 0 −3 0 0
= (−2 − x) 1 1−x 0 −3 1 1−x 0
−5 −1 3 − x 9 −1 3 − x
1−x 0 1−x 0
= (−2 − x)(4 − x) − 3(−3)
−1 3 − x −1 3 − x
= (x − 3)(x − 1)3 .

PA (x) = 0 implies that the eigenvalus of A are 1 and 3.

The eigenvectors of A :
For the
  eigenvalue 1, we need to slve (A − I)X = 0 where X ∈ R4 . Let
x
y 
X= z . Then, we solve


t
    
−3 3 0 0 x 0
−3 3 0 0 y  0
   =  .
1 1 0 0 z  0
9 −5 −1 2 t 0

That is 
 −3x + 3y = 0
x+y = 0
9x − 5y − z + 2t = 0

 
0
0
This implies that x = y = 0 and z = 2t. Then, .X =  2t for all t ∈ R.


t
 
0
0
Therefore, the eigenvector associated to the eigenvalue 1 is 
2 .


6
For the
 eigenvalue
 3, we need to slve (A − 3I)X = 0 where X ∈ R4 . Let
x
y 
X= z . Then, we solve


t
    
−5 3 0 0 x 0
−3 1 0 0  y  0
   =  .
1 1 −2 0 z  0
9 −5 −1 0 t 0

That is 

 −5x + 3y = 0
−3x + y = 0


 x + z − 2z = 0
9x − 5y − z = 0

 
0
0
This implies that x = y = z = 0. Then, .X =  0 for all t ∈ R. Therefore,


t
 
0
0
the eigenvector associated to the eigenvalue 3 is 0 .
