LinearAlgebra

andApplications

M244

MongiBlel
2025
 King Saud University
College of Sciences
Department of Mathematics

Linear Algebra and Applications

M-244

CLASS NOTES
2025

1
Mongi BLEL

Professor

1 E-mail : mblel@[Link], [Link]

 CONTENTS

I Matrices and Matrix Operations iii

1 Matrices and Matrix Operations . . . . . . . . . . . . . . . . . iii
1.1 Matrix Operations . . . . . . . . . . . . . . . . . . . . . iv
2 Transpose of Matrices . . . . . . . . . . . . . . . . . . . . . . . vi
2.1 The Elementary Row Operations . . . . . . . . . . . . . vii
2.2 Algorithm for Reducing a Matrix to Row-Echelon Form x
3 The Inverse Matrix . . . . . . . . . . . . . . . . . . . . . . . . . xi
3.1 The Inverse Matrix . . . . . . . . . . . . . . . . . . . . . xi
3.2 Matrix Inversion Algorithm . . . . . . . . . . . . . . . . xiv

II The Determinants xvii

1 Definition of the Determinant . . . . . . . . . . . . . . . . . . . xvii
2 Properties of the Determinants . . . . . . . . . . . . . . . . . . xx
3 The Adjoint Matrix . . . . . . . . . . . . . . . . . . . . . . . . xxiii

III System of Linear Equations xxvii

1 Introduction to System of Linear Equations . . . . . . . . . . . xxvii
1.1 Operations on Linear Systems . . . . . . . . . . . . . . . xxix
1.2 Gauss And Gauss Jordan Methods . . . . . . . . . . . . xxx
2 Homogeneous Linear Systems . . . . . . . . . . . . . . . . . . . xxxii
3 Crammer Method . . . . . . . . . . . . . . . . . . . . . . . . . . xxxiii
4 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxxv
5 Solutions of the Exercises . . . . . . . . . . . . . . . . . . . . . xxxvi

IV The Vector Spaces xli

1 Introduction to Vector Spaces . . . . . . . . . . . . . . . . . . . xli
1.1 Some Basic Identities in a Vector Space . . . . . . . . . xlii

i
ii

2 Vector Sub-Spaces . . . . . . . . . . . . . . . . . . . . . . . . . xlii

3 Linear Combination and Generating Sets . . . . . . . . . . . . xliii
4 Linear Dependence and Independence . . . . . . . . . . . . . . xlvii
5 Bases and Dimension . . . . . . . . . . . . . . . . . . . . . . . . xlix
6 Coordinate System and Change of Bases . . . . . . . . . . . . . liii
7 The Null Space, the Row Space and the Column Space of Matrix liv

V Inner Product Spaces and Orthogonality lix

1 Inner Product . . . . . . . . . . . . . . . . . . . . . . . . . . . . lix
2 The Orthogonality . . . . . . . . . . . . . . . . . . . . . . . . . lxi
3 The Orthonormal Bases . . . . . . . . . . . . . . . . . . . . . . lxiii

VI Linear Transformations lxvii

1 Definition of Linear Transformations . . . . . . . . . . . . . . . lxvii
1.1 Basic Properties of Linear Transformations . . . . . . . lxvii
2 Kernel and Image of Linear Transformations . . . . . . . . . . lxx
3 Matrix of Linear Transformation and Change of Basis . . . . . lxxii

VII Diagonalization of Matrix lxxix

1 Eigenvalues and Eigenvectors . . . . . . . . . . . . . . . . . . . lxxix
2 Diagonalization . . . . . . . . . . . . . . . . . . . . . . . . . . . lxxxii
CHAPTER I

MATRICES AND MATRIX OPERATIONS

1 Matrices and Matrix Operations

Definition 1.1

A real matrix is a rectangular array whose entries are real numbers.

These numbers are organized on rows and columns. An m × n matrix
will refer to one which has m rows and n columns, and the collection
of all m × n matrices of real numbers will be denoted by Mm,n (R). We
adopt the notation, in which the (j, k)th entry of the matrix A (that in
row j and column k) is denoted by aj,k and the matrix A = (aj,k ).
A matrix in Mm,n (R) is called a matrix of dimension (or of type) (m, n).

Definition 1.2

• Two matrices A = (aj,k ) and B = (bj,k ) in Mm,n (R) are called equal
if aj,k = bj,k for all j, k
• A matrix in M1,n (R) is called a row matrix or a row n−vector.
• A matrix in Mm,1 (R) is called a column matrix or a column vector.
• If the entries of a matrix are zero, we denote this matrix (0) or 0
• A matrix in Mn,n (R) is called a square matrix of type n and Mn,n (R)
will be denoted by Mn (R)
• A square matrix A = (aj,k ) ∈ Mn (R) is called diagonal if aj,k = 0 if

iii
iv

 
1 0 0
̸ k, example A = 0 0
j= 0.
0 0 3

Definition 1.3

• A square matrix A = (aj,k ) ∈ Mn,n (R) is called upper triangular if

aj,k = 0 if j > k
• A square matrix A = (aj,k ) ∈ Mn,n (R) is called lower triangular if
aj,k = 0 if j < k
• A diagonal matrix A = (aj,k ) in Mn (R), where aj,j = 1 is called the
identity matrix and denoted by In .

1.1 Matrix Operations

Matrix algebra uses three different types of operations.

1. Matrix Addition: If A = (aj,k ) and B = (bj,k ) have the same dimension

(or the same type), then the sum A + B is given by A + B = (aj,k + bj,k ).

Example 1 :
   
2 3 −1 4 −2 3
If A = and B = , then
−1
 1 3 2 −1 −2
6 1 2
A+B = .
1 0 1

2. Scalar Multiplication: If A = (aj,k ) is a matrix and α a scalar (real

number), the scalar product of α with A is given by αA = (αaj,k ).

Example 2 :
   
2 3 −1 4 6 −2
If A = , then 2A = .
0 1 5 0 2 10
3. Matrix Multiplication:
(a) If A ∈ M1,n (R) is a row
matrix,
 A = (a1 , . . . , an ) and B ∈ Mn,1 (R)
b1
a column matrix, B =  ... , we define the product A.B by: AB =
 

bn
a1 b1 + · · · + an bn .
 v

This matrix is of type (1, 1) (one column and one row) and called
the inner product of A and B.
(b) If A = (aj,k ) ∈ Mm,n (R) and B = (bj,k ) ∈ Mn,p (R), then the
product AB is defined as AB = (cj,k ) ∈ Mm,p (R), where cj,k is the
inner product of the j th row of A and the k th column of B
n
X
cj,k = aj,ℓ bℓk .
ℓ=1

Example 3 :  
  0 1 −1  
2 3 −1 7 3 7
If A = and B =  2 1 3 , then AB = .
0 1 5 −3 11 3
−1 2 0
Remarks 1 :

 i.e. AB ̸= BA
1. The multiplication operation ofmatrixis not commutative

0 1 0 0
in general. For example A = and B = . Then AB =
0 0 1 0
   
1 0 0 0
and BA = .
0 0 0 1
 
0 1
2. If A = , then A2 = 0.
0 0
The matrix operations satisfy the following properties:

Theorem 1.4

Let A, B, C denote matrices in Mm,n (R), and a, b ∈ R.

1. A + B = B + A,
2. A + (B + C) = (A + B) + C,
3. a(A + B) = aA + aB,
4. (a + b)A = aA + bA,
5. (ab)A = a(bA),
6. A + 0 = A, where 0 is the null matrix in Mm,n (R).
7. Im A = A and AIn = A,
8. If D ∈ Mn,p (R), E ∈ Mp,q (R) and F ∈ Mr,m (R), then
A(DE) = (AD)E,
vi

9. (A + B)D = AD + BD,
10. F (A + B) = F A + F B,

Remark 2 :
Let A ∈ Mm,n (R) and B ∈ Mn,p (R). If B1 , . . . , Bp are the columns of B, then
the columns of AB are A.B1 , . . . , [Link] .
B1 B2 . . . B p AB1 AB2 . . . ABp
b b1,2 . . . b1,p  c c1,2 . . . c1,p 
1,1 1,1
If B = ... ..
.
..
.
.. , then AB = ..
. .
..
.
..
.
.. , where
.
bn,1 bn,2 . . . bn,p cm,1 cm,2 . . . cm,p

n
X
cj,k = aj,i bi,k .
i=1

2 Transpose of Matrices
Definition 2.1

If A = (aj,k ) is a matrix in Mm,n (R), the transpose of A is the matrix

AT defined by AT = (bj,k ), where bj,k = ak,j . The matrix AT is in
Mn,m (R).

Theorem 2.2

Let A, B ∈ Mm,n (R) and C ∈ Mn,p (R), then

1. (A + B)T = AT + B T ,
2. (AC)T = C T AT ,

T
3. AT = A.
 vii

Definition 2.3

A square matrix A is called symmetric if AT = A and A is called

antisymmetric if AT = −A.

2.1 The Elementary Row Operations

Definition 2.4

There are three elementary matrix row operations:

1. (Interchange) Interchange two rows,
2. (Scaling) Multiply a row by a non-zero constant,

3. (Replacement) Replace a row by the sum of the same row and a

multiple of different row.

Definition 2.5

We denote the elementary row operations as follows:

1. The switches of the j th and the k th rows is indicated by: Rj,k

2. The multiplication of the j th row by r ̸= 0 is indicated by: r · Rj .

3. The addition of r times the j th row to the k th row is indicated by:

rRj,k .

Definition 2.6

A matrix A is called row-equivalent to matrix B of same type if B can

be obtained from A by a finite sequence of elementary row operations.
If A is row-equivalent to a matrix B, then B is row-equivalent to A.
viii

Definition 2.7

A matrix in Mm,n (R) is called in row echelon form if it has the following
properties:

1. The first non-zero element of a nonzero row must be 1 and is called

the leading entry.
2. All non-zero rows are above any rows of all zeros.
3. Each leading entry of a row is in a column to the right of the
leading entry of the row above it.

Definition 2.8: [Reduced Echelon Form]

A matrix in Mm,n (R) is called in reduced row echelon form if it has the
following properties:

1. The matrix is in row echelon form.

2. Each leading number is the only non-zero entry in its column.

Example 4:
 
1 −1 2
1. 0 1 3 is in row echelon form but is not reduced:
0 0 0
 
1 0 2
2. 0 1 5 is in reduced row echelon form:
0 0 0
 
1 −1 2
3. 0 1 5 is not in row echelon form.
3 0 0

Example 5 :    
2 3 −1 2 3 −1
(−1)R1,2
3 1 2  −→ 1 −2 3 
4 1 0 4 1 0
   
1 −2 3 1 −2 3
R1,2 (−2)R1,2 ,(−4)R1,3
−→ 2 3 −1 −→ 0 7 −7 
4 1 0 0 9 −12
 ix

   
1 1 −2 3 1 −2 3
7 R2 (−9)R2,3
−→ 0 1 −1  −→ 0 1 −1
0 9 −12 0 0 −3

     
1
− 31 R3
−2 3 1 −2 0 1 0 0
(−3)R3,1 ,1.R3,2 2R2,1
−→ 0 1 −1 −→ 0 1 0 −→ 0 1 0
0 0 1 0 0 1 0 0 1

Example6 :   
2 −3 4 −2 0 2 −3 4 −2 0
 3 −1 2 −3 2 1 2 −2 −1 2
  (−1)R
−→
1,2
 
−2 −3 4 3 2 −2 −3 4 3 2
−3 1 0 3 1 −3 1 0 3 1
   
1 2 −2 −1 2 1 2 −2 −1 2
R1,2  2 −3 4 −2 0 (−2)R 1,2
 0 −7 8 0 −4
−→  −2 −3 −→  
4 3 2   −2 −3 4 3 2
−3 1 0 3 1 −3 1 0 3 1

   
1 2 −2 −1 2 1 2 −2 −1 2
2R1,3 ,3R1,4 0 −7 8 0 −4 0 1 0 1 6
−→   R2,3−→
,1R2,4
 
0 1 0 1 6 0 −7 8 0 −4
0 7 −6 0 7 0 0 2 0 3
   
1 2 −2 −1 2 1 2 −2 −7 2
7R2,3 0 1 0 1 6  R3,4 0 1 0 1 6
−→  0 0 8
 −→  
7 38 0 0 2 0 3
0 0 2 0 3 0 0 8 7 38

   
1 2 −2 −1 2 1 2 −2 −1 2
1 1 0
(−4)R3,4 0 1 0 1 6 2 R3 , 7 R4 1 0 1 6
−→    −→  
3 
0 0 2 0 3 0 0 1 0 2
26
0 0 0 7 26 0 0 0 1 7
   
1 0 −2 −3 −10 1 0 0 −3 −7
(−2)R2,1 0 1 0 1 6  0 1 0 1 6
−→   2R3,1
3  −→
 
3 
0 0 1 0 0 0 1 0
2 2
26 26
0 0 0 1 7 0 0 0 1 7
29 29
   
1 0 0 0 7 1 0 0 0 7
3R4,1 0 16 
1 0 1 6 (−1)R4,2 0 1 0 0
−→  0
 −→ 
3 
7 
3 
0 1 0 2
0 0 1 0 2
26 26
0 0 0 1 7 0 0 0 1 7
x

Fractions can be avoided as follows:

   
1 2 −2 −1 2 7 14 −14 −7 14
(−4)R3,4 0 1 0 1 6 0 7 0 7 42
−→   7R−→
1 ,7R2
 
0 0 2 0 3 0 0 2 0 3
0 0 0 7 26 0 0 0 7 26

   
7 14 −14 0 40 7 0 0 0 29
1.R4,1 ,−1R4,2 0 7 0 0 16 0 7 0 0 16
−→   7R3,1 ,(−2)R
−→
2,1
 
0 0 2 0 3 0 0 2 0 3
0 0 0 7 26 0 0 0 7 26

1 0 0 0 29
 
7
1 1 1 1
7 R1 , 7 R2 , 2 R3 , 7 R4 0
 1 0 0 16 
−→ 0 0 1 0 3 
7 
2
0 0 0 1 26 7

Theorem 2.9

Each matrix is row equivalent to one and only one reduced echelon
matrix.

2.2 Algorithm for Reducing a Matrix to Row-Echelon

Form
Consider a non zero (m, n) matrix, A.
1. Determine the leftmost nonzero column (this is called a pivot column,
and the topmost position in this column is called the leading entry).
2. Use elementary row operations to put 1 in the leading entry.
3. Use elementary row operations to put 0 below the leading entry.
4. If there are no more nonzero rows below the leading entry, the matrix is in
row-echelon form. Otherwise apply the previous steps to the sub-matrix
consisting of the rows that lie below the leading entry.

Example7 : 
3 −2 7 0 7
1 −1 3 −1 4
Let A = 
1
 . We look for the reduced echelon form of A.
0 1 1 0
2 −2 6 0 7
 xi

   
3 −2 7 0 7 1 0 1 1
0
1 −1 3 −1 4 R3,1 ,(−1)R1,2 0 −1 2 −2
4
  −→  
1 0 1 1 0 (−3)R1,3 ,(−2)R1,4 0 −2 4 −3
7
2 −2 6 0 7 0 −2 4 −2
7
 
1 0 1 1 0
(−2)R2,3 0 −1 2 −2 4
−→ 
0

(−1)R3,4 0 0 1 1
0 0 0 1 0
 
1 0 1 1 0
−R2 ,(−1)R3,4 0 1 −2 2 −4
−→  
0 0 0 1 −1
0 0 0 0 1
 
1 0 1 0 0
(1)R4,3 ,(4)R4,2 0 1 −2 0 0
−→ 
0

(−2)R3,2 ,(−1)R3,1 0 0 1 0
0 0 0 0 1

3 The Inverse Matrix

3.1 The Inverse Matrix
Definition 3.1

A square matrix A of order n is called invertible if there is a square

matrix B of order n such that AB = BA = In .
We denote A−1 the inverse matrix of A if it exists.

Theorem 3.2

1. The inverse matrix if it exists is unique.

2. The inverse matrix of In is In .
3. (A−1 )−1 = A.
4. If A and B are invertible in Mn (R), then (AB)−1 = B −1 A−1 .
5. If A1 , . . . , Ak are invertible in Mn (R), then
(A1 . . . . .Ak )−1 = A−1 −1
k . . . . .A1 .
xii

6. If A is invertible, then (rA)−1 = 1r A−1 , for all r ∈ R∗ .

7. If A is invertible, then AT is invertible and (AT )−1 = (A−1 )T .

Example 8 :
Let A be a square matrix such that (A − I)2 = 0. Then A is invertible. Indeed:
(A − I)2 = A2 − 2A + I = 0 ⇐⇒ A(2I − A) = I. Then A is invertible and
A−1 = 2I − A.
Example 9 :   
−4 7 2 3
Let B = and C = . We look for a matrix A of order 2 such
1 −2 1 1

−1
that B −1 A = C.
−1

−1 −1 −1
We have:
 B A = C ⇐⇒ A B= C ⇐⇒ A = BC .
−1 3 11 −26
C −1 = and A = BC −1 = .
1 −2 −3 7

Definition 3.3

A matrix E of order n is called an elementary matrix if it is obtained

from In by an elementary row operation.

Remarks 3 :
   
1 −1 0 1 0 0
1. Let A = 2 1 3 and the elementary matrix E = 0 0 1
1 −2 1 0 1 0
which is the result of switching
 the second
 and the third rows of I3 .
1 −1 0
We have R2,3 A = EA = 1 −2 1.
2 1 3
   
1 0 2 3 1 0 0
2. Let A = 2 −1 3 6 and the elementary matrix E = 0 1 0 =
1 4 4 0 5 0 1
5R1,3 I3 .  
1 0 2 3
We have 5R1,3 A = EA = 2 −1 3 6 .
6 4 14 15
In general we have
 xiii

Theorem 3.4

Let A ∈ Mm,n (R) and E an elementary matrix obtained by performing

an elementary row operation on the matrix Im . If the sam operation is
performed on the matrix A, the resulting matrix is EA.

Theorem 3.5

For any elementary row operation R there exists an elementary row

operation R′ of the same type as R, such that R′ (R(A)) = R(R′ (A)) =
A for any A, i.e. the inverse operation of an elementary row operation
exists and is an elementary row operation of the same type.

Proof .
1. Suppose R = rRj is the operation which multiplies the j th row of a
matrix by r ̸= 0. Let R′ = 1r Rj .
2. Suppose R = rRj,k is the operation which replaces row k by row k plus
r times row j, j ̸= k, then R′ = (−r)Rj,k
3. If R = Rj,k interchanges rows j and k, let R′ = R.

Corollary 3.6

If E is an elementary matrix, then E has an inverse and its inverse is

an elementary matrix.
The matrix E −1 is the matrix obtained from Im by the inverse row
operation that produced E from Im .

Theorem 3.7

If A is a square matrix of order n. The following conditions are equiva-

lent:

1. The matrix A has an inverse.

2. The reduced row echelon form of the matrix A is In .

xiv

3. There is a finite number of elementary matrices E1 , . . . , Em in

Mn (R) such that A = E1 . . . Em .

3.2 Matrix Inversion Algorithm

Let A ∈ Mn (R) be an invertible matrix, there exists a sequence of elemen-
tary row operations that carry A to In . This same series of elementary row
operations carries In to A−1 . The algorithm car be summarized as follows:
[A|In ] −→ [In |A−1 ].

1. Let [B|C] be the reduced row echelon form of the matrix [A|I] ∈ Mn,2n (R).
2. If B = In , then C = A−1 .
3. If B ̸= In , the matrix A is not invertible.

Example 10 :    
2 1 −1 1
If A satisfies (AT − 2I)−1 = , then AT − 2I = , AT =
3 1 3 −2
   
1 1 1 3
and A = .
3 0 1 0
Example11 :
0 21 − 12


Let A = 1 0 1 
1
2 2 1

0 21 − 12 1 0 0
   
1 0 1 0 1 0
R1,2 ,R2,3 1
1 0 1 0 1 0 −→ 2
2 1 0 0 1
2 12 1 0 0 1 0 1
2 − 12 1 0 0
   
1 0 1 0 1 0 1 0 1 0 1 0
(−2)R1,2 (−1)R2,3
−→ 0 21 −1 0 −2 1 −→ 0 1
2 −1 0 −2 1
0 12 − 12 1 0 0 0 0 1
2 1 2 −1
   
1 0 1 0 1 0 1 0 0 −2 −3 2
2R2 ,2R3 (−1)R3,1 ,2R3,2
−→ 0 1 −2 0 −4 2  −→ 0 1 0 4 4 −2
0 0 1 2 4 −2 0 0 1 2 4 −2
 
−2 −3 2
Then the inverse matrix of A is the matrix  4 4 −2.
2 4 −2
0 21 − 12
    
−2 −3 2 1 0 0
1 0 1  4 4 −2 = 0 1 0
2 12 1 2 4 −2 0 0 1
 xv

Example12 : 
1 3 2 1
2 3 3 1
Let A = 
3
.
3 4 2
1 1 1 1

1 3 2 1 1 0 0 0 1 3 2 1 1 0 0 0
   
2 3 3 1 0 1 0 0 0 −3 −1 −1 −2 1 0 0
−→ −→
3 3 4 2 0 0 1 0 0 −6 −2 −1 −3 0 1 0
1 1 1 1 0 0 0 1 0 −2 −1 0 −1 0 0 1
1 3 2 12 1 0 0 0 1 3 2 1 1 0 0 0
   
0 3 1 1 2 −1 0 0  0 1 0 1 1 −1 0 1 
−→ −→
0 6 2 1 3 0 −1 0  0 0 0 −1 −1 2 −1 0 
0 2 1 0 1 0 0 −1 0 2 1 0 1 0 0 −1
1 3 2 1 1 0 0 0 1 3 2 1 1 0 0 0
   
0 1 0 1 1 −1 0 1  0 1 0 0 0 1 −1 1 
−→ −→
0 0 0 −1 −1 2 −1 0  0 0 0 1 1 −2 1 0 
0 0 1 −2 −1 2 0 −3 0 0 1 0 1 −2 2 −3
1 3 2 1 1 0 0 0 1 0 0 0 −2 3 −2 3
   
0 1 0 0 0 1 −1 1  0 1 0 0 0 1 −1 1 
−→ .
0 0 1 0 1 −2 2 −3 0 0 1 0 1 −2 2 −3
0 0 0 1 1 −2 1 0 0 0 0 1 1 −2 1 0
 
−2 3 −2 3
0 1 −1 1
The inverse matrix of the matrix A is  
1 −2 2 −3
  1 −2 1 0
1 1 1 1
0 1 1 1
Let A = 
1
.
0 1 1
1 1 0 1

1 1 1 1 1 0 0 0 1 1 1 1 1 0 0 0
   
0 1 1 1 0 1 0 0 0 1 1 1 0 1 0 0
−→
1 0 1 1 0 0 1 0 0 −1 0 0 −1 0 1 0
1 1 0 1 0 0 0 1 0 0 −1 0 −1 0 0 1
1 0 0 0 1 −1 0 0 1 0 0 0 1 −1 0 0
   
0 1 1 1 0 1 0 0 0 0 0 1 −2 1 1 1
−→  −→
0 −1 0 0 −1 0 1 0 0 −1 0 0 −1 0 1 0
0 0 −1 0 −1 0 0 1 0 0 −1 0 −1 0 0 1
1 0 0 0 1 −1 0 0
 
0 1 0 0 1 0 −1 0 
−→ 
0 0 1 0 1 0 0 −1
0 0 0 1 −2 1 1 1
 
1 −1 0 0
1 0 −1 0
Then A−1 = .
1 0 0 −1
−2 1 1 1
CHAPTER II

THE DETERMINANTS

We associate to any square matrix a number which is zero if and only if the
matrix is not invertible.

1 Definition of the Determinant

Definition 1.1

If A = (aj,k ) is a square matrix of type n. Denote Aj,k the square

matrix of type n − 1 obtained by deleting the j th −row and the k th −
column of A.

Example 13 : 
1 2 3  
1 2
If A = −1 2 −4 , then A2,3 =
 .
2 −3
2 −3 4

Definition 1.2

 
a b
1. If A = , the determinant of A is defined by: |A| =
c d
 
a b
det(A) = = ad − bc.
c d

xvii
xviii

 
a b c
2. If A = d e f , the determinant of A is defined by:
g h i

a b c
e f d f d e
|A| = det(A) = d e f =a −b +c .
h i g i g h
g h i

Definition 1.3

3. If  
a1,1 a1,2 ... a1,n
 a2,1 a2,2 ... a2,n 
A= ,
 
..
 . 
an,1 am,2 ... an,n
the determinant of A is defined by:

|A| = det(A) = a1,1 detA1,1 + . . . + (−1)n+1 a1,n detA1,n

Xn
= (−1)j+1 a1,j detA1,j .
j=1

Example 14 :
 
4 5
1. The determinant of the matrix A = is
2 3

4 5
|A| = det(A) = = 4.3 − 5.2 = 2.
2 3
 
1 3 5
2. The determinant of the matrix A = 2 4 6 is
0 2 4

1 3 5
4 6 2 6 2 4
|A| = det(A) = 2 4 6 =1 −3 +5
2 4 0 4 0 2
0 2 4
 xix

 
1 3 5
3. The determinant of the matrix A = 2 4 6 , is
0 2 4

1 3 5
4 6 2 6 2 4
|A| = det(A) = 2 4 6 =1 −3 +5 = 0.
2 4 0 4 0 2
0 2 4

Definition 1.4

If A is a square matrix of order n, the determinant detAj,k is called

the minor of the entry aj,k or the (j, k)th minor of A and the number
Cj,k = (−1)j+k detAj,k is called the cofactor of the entry aj,k or the
(j, k)th cofactor of the matrix A.

Remark 4 :
1. If A is a square matrix of order n, the determinant of the matrix A is
equal to

n
X
detA = a1,j C1,j .
j=1

2. By rearranging the boundaries we conclude to

n
X n
X
detA = a1,j C1,j = ak,j Ck,j
j=1 j=1
Xn
= ak,j Ck,j .
k=1

Theorem 1.5: [The Sarrus’s Theorem]

 
a1,1 a1,2 a1,3
If n = 3 and the matrix A = a2,1 a2,2 a2,3 , then
a3,1 a3,2 a3,3
xx

a1,1 a1,2 a1,3 a1,1 a1,1

a2,1 a2,2 a2,3 a2,1 a2,1

a3,1 a3,2 a3,3 a3,1 a3,1

detA = a1,1 (a2,2 · a3,3 − a2,3 · a3,2 )

−a1,2 (a2,1 · a3,3 − a2,3 · a3,1 )
+a1,3 (a2,1 · a3,2 − a2,2 · a3,1

Example 15 : 
3 −4 0
If A = 0 7 6,
2 −6 1

3 −4 0 3 −4
detA = 0 7 6 0 7
2 −6 1 2 −6
= 3 · 7 · 1 + (−4) · 6 · 2 − (−6) · 6 · 3 = 81.

2 Properties of the Determinants

Theorem 2.1

1. If A is a square matrix, detAT = detA.

2. If a square matrix A contains a zero row or column, then its de-
terminant is 0.
3. If the matrix A = (aj,k )1≤j,k≤n is upper (lower) triangular, then
its determinant is equal to:

a1,1 . . . an,n .

4. If a square matrix A contains a row which is a multiple of a dif-

ferent row, then its determinant is 0.
 xxi

Theorem 2.2

5. If a matrix B is obtained by multiplying a row of a matrix A by

a number c, then |B| = c|A| (i.e. |[Link] A| = c|A|).
6. If a matrix B is obtained by interchanging two rows of a matrix
A, then detB = −detA (i.e. |Rj,k A| = −|A|).
7. If a matrix B is obtained by adding a multiple of a row to another
row of a matrix A, then detB = detA. (i.e. |cRj,k A| = |A|).

Example 16 :

1 3 2 2 1 3 2 1
2 3 3 1 (−2)R1,2 ,(−3)R1,3 0 −3 −1 −1
=
3 3 4 2 (−1)R1,4 0 −6 −2 −1
1 1 1 1 0 −2 −1 0
3 1 1
= − 6 2 1
2 1 0
3 1 1
(−1)R1,2 3 1
= − 3 1 0 =− = −1.
2 1
2 1 0

Example 17 :
xxii

1 −2 5 −2 −1 1 −2 5 −2 −1
−2 3 −1 1 0 0 −1 9 −3 −2
2R1,2 ,−3R1,3
3 −3 2 0 −1 = 0 3 −13 6 2
−1R1,4 ,−1R1,5
1 −1 2 1 −4 0 1 −3 3 −3
1 −2 4 −3 1 0 0 −1 −1 2
−1 9 −3 −2
3 −13 6 2
=
1 −3 3 −3
0 −1 −1 2
−1 9 −3 −2
3R1,2 ,1R1,3 0 14 −3 −4
=
0 6 0 −5
0 −1 −1 2
14 −3 −4 14 6 1
= 6 0 −5 = −3 0 1
1 1 −2 −4 −5 −2
14 6 1
(−1)R1,2 ,2R1,3 −17 −6
= −17 −6 0 = |
24 7
24 7 0
1R1,2 −17 −6
= = 25.
7 1

Example 18 :

1 1 1 1 1 1 1 1
1 −1 2 1 0 −2 1 0
=
1 0 1 1 0 −1 0 0
1 2 1 −2 0 1 0 −3
1 1 1
1 1
= 0 1 0 = = −3.
0 −3
0 0 −3

Example 19 :
1 a a2 1 a a2
1 b b2 = 0 b−a (b − a)(b + a)
1 c c2 0 c−a (c − a)(c + a)
1 b+a
= (b − a)(c − a)
1 c+a
= (b − a)(c − a)(c − b).
 xxiii

Theorem 2.3

If A and B are in Mn (R), then

det(AB) = detA detB.

Theorem 2.4

A square matrix A is invertible if and only if detA ̸= 0.

Remarks 5 :

1. If A is a square matrix of order n, then |cA| = cn |A|.

2. Let A be a square matrix and B a row echelon form of A. Then there is a
finite elementary matrices E1 , . . . , Em such that E1 . . . Em A = B. Then

det(E1 ) . . . det(Em )det(A) = det(B).

3 The Adjoint Matrix

Definition 3.1

Let A be a square matrix. The adjoint matrix associated to the matrix

A is adj(A) = (Cj,k )T , where (Cj,k ) is the cofactor matrix of A.

Theorem 3.2

Let A be a square matrix of order n, then

(adj(A))A = A(adj(A)) = (detA)In .

Theorem 3.3

If A is an invertible matrix, then

xxiv

1
A−1 = adj(A).
detA
1
detA−1 = .
detA

Example 20 :
     
2 −1 1 1 1 1 1
1. A = , detA = 5, adj(A) = and A−1 = .
3 1 −3 2 5 −3 2
 
2 1 0
2. A =  1 −2 3, detA = −13
−1 0 2
 T  
−4 −5 −2 −4 −2 3
adj(A) = −2 4 −1 = −5 4 −6
3 −6 −5 −2 −1 −5
 
4 2 −3
1 
and A−1 = 5 −4 6 .
13
2 1 5
 
3 1 1 1
1 3 1 1
1 1 3 1, detA = −24
3. A =  

1 1 1 3 
−20 −4 −4 −4
 −4 −20 −4 −4 
adj(A) = 
 −4 −4 −20 −4 ,


−4 −4 −4 −20
 
5 1 1 1
1 1 5 1 1
and A−1 =  
6 1 1 5 1
1 1 1 5

Example 21 :
Let A, B be matrices of size (3, 3) such that A is not invertible and |B| = 2.
Find Aadj(A) + 2B −1 .
8
Aadj(A) = 0 and Aadj(A) + 2B −1 = = 4.
|B|
Example 22 :
 xxv

 
3 1 2
If adj(A) =  0 3 2, find the matrix A.
−1 2 1
We have Aadj(A) = I3 , then A = |A|(adj(A))−1 and
|A| |adj(A)|
p = |A|3 , then |adj(A)| = |A|2 . Therefore
A = |adj(A)|(adj(A))−1 .  
−1 3 −4
|adj(A)| = 1, the |A| = 1 and A = −2 5 −6
3 −7 9

Theorem 3.4

Let A ∈ Mn (R) be an invertible matrix, with n ≥ 2. Then

|adj(A)| = |A|n−1

1
(adj(A))−1 = A,
|A|
and
adj(adj(A)) = (detA)n−2 A.

Proof .
From the relation Aadj(A) = |A|In , we have |adj(A)| = |A|n−1 and (adj(A))−1 =
1
|A| A.
Let B = adj(A). We have: Badj(B) = |B|In = |A|n−1 In ,
adj(B) = |A|n−1 B −1 = |A|n−2 A and adj(adj(A) = (detA)n−2 A.

Remark 6 :
Let A ∈ Mn (R). The matrix A is invertible if and only if adj(A) is invertible.
1
If the matrix A is invertible, then A−1 = adj(A). Hence adj(A) is invertible.
|A|
Also, if the matrix adj(A) is invertible and the matrix A is not invertible, then
Aadj(A) = 0 and Aadj(A)(adj(A))−1 = 0.
Then A = 0 this is absurd, because if A = 0 then adj(A) = 0.
CHAPTER III

SYSTEM OF LINEAR EQUATIONS

1 Introduction to System of Linear Equations

Definition 1.1

1. A linear equation in the variables x1 , . . . , xn is an equation that

can be written in the form

a1 x1 + . . . + an xn = b,

where b and a1 , . . . , an are real numbers that are known in ad-

vance.
2. A system of linear equations (or a linear system) is a collection
of one or more linear equations involving the same variables say,
x1 , . . . , xn .


 a1,1 x1 + a1,2 x2 + · · · + a1,n xn = b1
 a2,1 x1 + a2,2 x2 + · · · + a2,n xn = b2

.. .. .. . ..


 . . . + .. + .
am,1 x1 + am,2 x2 + · · · + am,n xn = bm .


where b1 , . . . , bm , (aj,k ) are known real numbers, (1 ≤ j ≤ m, 1 ≤

k ≤ n) (called also the data of the system).

This linear system can be written more concisely as the vector equation:
AX = B, where

xxvii
xxviii

     
a1,1 a1,2 ··· a1,n b1 x1
 a2,1 a2,2 ··· a2,n   b2   x2 
A= . ..  , B =  . , X =  . .
     
.. ..
 .. . . .   ..   .. 
am,1 am,2 ··· am,n bm xn

Definition 1.2

A solution to a linear system is a choice of scalar values for the un-

knowns that satisfies every equation. The collection of all solutions of
a particular system is its solution set.

We will see that any system of linear equations has

1. no solution, or
2. exactly one solution, or
3. infinitely many solutions.

Definition 1.3

A system of linear equations is said to be consistent if it has either one

solution or infinitely many solutions.
A system of linear equation is said to be inconsistent if it has no solution.

Example 23 :
The following linear system with two variables

4x − y = 5
−7x + 2y = 3
can be interpreted as the intersection in the plane of the straight lines of equa-
tions respectively 4x − y = 5 and−7x + 2y= 3.   
4 −1 x1 5
This system can also be written =
−7 2 x2 3
Example 24 :
The following linear system with three variables

x + y −3 z = 1
2x + y − z = 3
can be interpreted as the intersection in the space of the planes of equations
x + y − 3z = 1 and 2x + y − z = 3.
 xxix

The solution of this system is {(2, 1, 0) + z(−2, −5, 1); z ∈ R}. This is the
equation of the line passing through the point A of coordinates (2, 1, 0) and
parallel of the vector v of coordinates (−2, −5, 1).
Example 25 :    
1 −1 0 1 2
Consider the matrices A = and C = and we look for a
1 −2 1 2 3
matrix 
of order (2,
 3) such that AB = C.
x y z
If B = we find the following linear system:
t u v


 x−t = 0
x − 2t = 1




y−u = 1


 y − 2u = 2
z−v = 2




z − 2v = 3


The solution of this system is (−1, 0, 1, −1, −1, −1)

x = t = −1, y = 0, u = −1, z = 1, v = −1.

1.1 Operations on Linear Systems

Definition 1.4

Two linear systems are called equivalent if they have the same set of
solutions.

The basic strategy for solving a linear system is to replace one system with
an equivalent system (i.e., one with the same solution set) that is easier to
solve.
The following operations on a system of linear equations do not change the
solution set are:

1. Exchange of two equations.

2. Multiply an equation by a nonzero real number.

3. Add an equation to another.

When we write a linear system in matrix form, AX = B, these actions corre-

spond to operations on the rows of the matrix.
xxx

Definition 1.5

The augmented matrix associated to the linear system AX = B is the

matrix obtained by adding the column vector B as a new last column
of A.

Example 26 :
The augmented matrix associated to the linear system

x + 2y − z = 1
2x + y + 3z = 3
 
1 2 −1 1
is .
2 1 3 3
The elementary row operations on the augmented matrix of a system pro-
duce the augmented matrix of an equivalent system.

1.2 Gauss And Gauss Jordan Methods

The method of solving a linear system by reducing the augmented matrix to
a row echelon form is called Gauss method and the method of solving a linear
system by reducing the augmented matrix to a reduced row echelon form is
called Gauss Jordan method.
The Gauss (respectively Gauss-Jordan) elimination method to solve a system
of linear equations is described in the following steps:

1. Write the augmented matrix of the system.

2. Use elementary row operations to transform the augmented matrix in a
row echelon form (respectively reduced echelon form).
3. Solve the obtained triangular system.

Example 27 :
Consider
 the following linear system
x + 2y − z = −4
−x + y = −2
y−z = −4

The augmented matrix of the system is
 
1 2 −1 −4
−1 1 0 −2
0 1 −1 −4
 xxxi

and the matrix  

1 2 −1 −4
0 1 −1 −4
0 0 1 3
is a row echelon form of this matrix.
Using Gauss method, the system has a unique solution which is x = 1, y =
−1, z = 3.
The reduced row echelon form of this matrix is
 
1 0 0 1
0 1 0 −1 .
0 0 1 3
Using Gauss Jordan method, the system has a unique solution which is x =
1, y = −1, z = 3.
Example 28 :
Consider
 the following linear system
 x + 2y − z + t = 1
3x − y + 5z − t = 2 ; m ∈ R.
5x + 3y + 3z + t = m

 
1 2 −1 1 1
The matrix 0 1 − 78 47 1
7
 is a row echelon form of the matrix of the
0 0 0 0 m−4
system.
If m ̸= 4, the system, is inconsistent.
If m = 4, the system has infinite solutions :
5 9 1 1 8 4
{( − z + t, + z − t) : z, t ∈ R2 }.
7 7 7 7 7 7


 −2y + 3z = 0
2x − 4y + 2z = 1

Consider the following linear system ,

 −x − 2y + 5z = 0
x − 2y = 1

The
 augmented matrix of the system is:
0 −2 3 0
 2 −4 2 1
 
−1 −2 5 0
1 −2 0 1  
1 −2 0 1
0 −2 3 1
A row echelon form of the augmented matrix is  0 0

1 1
0 0 0 3
The system is inconsistent.
Example 29 :
xxxii

Give the relations between the numbers a, b and c so that the following linear
system
 is consistent.
 x + y + 2z = a
x+z = b
2x + y + 3z = c

 
1 1 2 a
The augmented matrix of the system is: 1 0 1 b .
2 1 3 c
 
1 1 2 a
A row echelon form of the augmented matrix is 0 1 1 a − b .
0 0 0 c−a−b
The system is consistent if and only if c − a − b = 0.
Example 30 :
Consider the following linear system

 x + my + (m − 1)z = m+1
3x + 2y + mz = 3
(m − 1)x + my + (m + 1)z = m − 1

The determinant of the system is m2 (m − 4).
If m = 0, the augmented
  matrix of
 the system is:

1 0 −1 1 1 0 −1 1
 3 2 0 3 . The matrix 0 1 3 0 is a row echelon form of the
2
−1 0 1 −1 0 0 0 0
augmented matrix of the system.
The system has an infinity of solutions {(1 + z, − 32 z, z); z ∈ R}.
 
1 4 3 5
1
If m = 4, a row echelon form of the augmented matrix is 0 1 2 0 .

0 0 0 12
The system is inconsistent.

2 Homogeneous Linear Systems

Definition 2.1

A linear system AX = B is called homogeneous if B = 0.

Remarks 7 :
1. Any homogeneous linear system is consistent. 0 is a solution of the sys-
tem.
2. If X1 and X2 are solutions of the homogeneous system AX = 0, then
X1 + λX2 is also a solution of the linear system for all λ ∈ R.
 xxxiii

3. If the homogeneous linear system AX = 0 has a non zero solution, it has

an infinite number of solutions.

Theorem 2.2

If X0 is a solution of the linear system AX = B, then for any solution X

of the system AX = B, there exists a solution X1 of the homogeneous
system. AX = 0 such that X = X0 + X1 .

Theorem 2.3

Any homogeneous system of linear equations where the variables are

more than the equations, this system has an infinitely many solutions.

Theorem 2.4

Any consistent system of linear equations has a unique solution or an

infinite number of solutions.

3 Crammer Method
Theorem 3.1

If A is an invertible 
matrix,
 then the linear system AX = B has a
x1
unique solution X =  ... , with
 

xn
detA1 detAn
x1 = , . . . , xn = .
detA detA
where Aj the matrix obtained by replacing the j th column of A by the
column matrix B.

Example 31 :
xxxiv


 3x − 2z = 2
Use Crammer method to solve the following system: −2x + 3y − 2z = 3
−5x + 4y − z = 1

3 0 −2 2 0 −2
|A| = −2 3 −2 = 1, |A1 | = 3 3 −2 = −8 = x,
−5 4 −1 1 4 −1
3 2 −2 3 0 2
|A2 | = −2 3 −2 = −13 = y, |A3 | = −2 3 3 = −13 = z.
−5 1 −1 −5 4 1
Example 32 : 
x + 2y − z = −4
Consider the system −x + y = −2
y−z = −4


1 2 −1 −4
The augmented matrix of the system is −1 1 0 −2 and the matrix
  0 1 −1 −4
1 0 0 1
0 1 0 −1 is the reduced echelon form of this matrix. Then the unique
0 0 1 3  
1
solution of this system is −1.
3
Example 33 :
Consider
 the system of linear equations:
 x + my + (m − 1)z = m+1
3x + 2y + mz = 3
(m − 1)x + my + (m + 1)z = m − 1

The augmented matrix of the
  system is:
1 m m−1 m+1
 3 2 m 3 . This matrix is row equivalent to the matrix
 m − 1 m m + 1 m −1 
1 m m−1 m+1
0 2 − 3m 3 − 2m −3m .
0 m(2 − m) m(3 − m) m(1 − m)
If m = 0, the set of solutions is {(1 + z, − 23 z, z) : z ∈ R}.
If m ̸= 0, the augmented
  matrix is row equivalent to the matrix
1 m m−1 m+1
0 2 1+2m 
1 m .
0 0 m − 4 m+2 m
If m = 4, no solutions of the system and if m ̸= 4 and m ̸= 0, the system has
a unique solution.
Example 34 :
 xxxv

Consider
 the system of linear equations:
 ax + by − 3z = −3
−2x − by + z = −1 . If (1, −1, 2) is a solution of the linear system, then
 ax + 3y − cz = −1

 a − b − 6 = −3
−2 + b + 2 = −1 ⇐⇒ b = −1, a = 2, c = 0. With these values of a, b, c,
a − 3 − 2c = −1

(1, −1, 2) is the unique solution of the linear system because the determinant
2 −1 −3
of the system −2 1 1 = 16 ̸= 0,
2 3 0

4 Exercises
Solve the following linear systems

 x + 2y − z + t = 1
1. 3x − y + 5z − t = 2 .
5x + 3y + 3z + t = m



 −2y + 3z = 0
2x − 4y + 2z = 1

2. ,

 −x − 2y + 5z = 0
x − 2y = 1


 x + 2y + z = 1
3. 2x + (a + 3)y + 3z = 3
x + (−a + 3)y + (a − 2)z = 0


Use Gauss Jordan method to solve the following linear systems


 x + y + 2z = 8
1. −x − 2y + z = 1
3x − 7y + 4z = 10


 x − y + 2z − w
 = −1
2x + y − 2z − 2w = −2

2.

 −x + 2y − 4z + w = 1
3x − 3w = −3


 2x + 2y + 2z = 0
3. −2x + 5y + 2z = 0
−7x + 7y + z = 0

xxxvi


2x − 3y = −2
4. 2x + y = 1
3x + 2y = 1



 3x + 2y − z = −15
5x + 3y + 2z = 0

5.

 3x + y + 3z = 11
11x + 7y = −30


 4x − 8y = 12
6. 3x − 6y = 9
−2x + 4y = −6



 x + y + 2z + t = 11
x + 2y + z + t = 9

7.

 x + y + z + 2t = 6
2x + y + z + t = 14


1. Give conditions on the numbers a, b and c such that the following linear
system is consistent.

 x + y + 2z = a
x+z = b
2x + y + 3z = c


2.  that if ad −
Prove   bc ̸= 0, the reduced row echelon form of the matrix
a b 1 0
is .
c d 0 1

Find a, b, c such that (1, −1, 2) is a solution of the following linear system:
1. 
 ax + by − 3z = −3
−2x − by + z = −1 .
ax + 3y − cz = −1


2. Prove that (1, −1, 2) is the unique solution of the system.

   
1 2 3 a
Let the matrices A = 2 3 4 and B =  b .
3 4 5 c
Prove that the linear system AX = B is consistent if and only if b − a = c − b.

5 Solutions of the Exercises

1.  
1 2 −1 1 1
0 1 − 87 4
7
1

7
0 0 0 0 m−4
 xxxvii

is a row echelon form of the augmented matrix of the system.

If m ̸= 4, the system is inconsistent. If m = 4 the system has an infinite
many solutions.
5 9 1 1 8 4
{( − z + t, + z − t, z, t), z, t ∈ R2 }.
7 7 7 7 7 7
 
0 −2 3 0
 2 −4 2 1
2. The augmented matrix of the system is:  −1 −2 5 0. This ma-


1 −2 0 1
 
1 −2 0 1
0 2 −3 0 
trix is row equivalent to the matrix  . The system is
0 0 1 −1
0 0 0 1
inconsistent.
 
1 1 2 8
3. The augmented matrix of the system is: −1 −2 1 1 
3 −7 4 10
A row echelon form:
 
1 1 2 8
1R1,2 , −3R1,3 0 −1 3 9 
0 −10 −2 −14
 
1 1 2 8
−R2 , 10R2,3 : 0 1 −3 −9 
0 0 −32 −104
 
1 1 2 8
1
− 32 r3 : 0 1 −3 −9
0 0 1 134
 
1 0 5 17
(−1)R2,1 : 0 1 −3 −9
0 0 1 13 4 
1 0 0 34


3R3,2 , (−5)R3,1 : 0 1 0 34 .
0 0 1 134
3 3 13
The solution is x = , y = , z = .
4 4 4
 
1 −1 2 −1 −1
2 1 −2 −2 −2
1. The augmented matrix of the system is: 
−1
.
2 −4 1 1
3 0 0 −3 3
xxxviii

A row echelon form :  

1 −1 2 −1 −1
0 3 −6 0 0
(−2)R1,2 , 1R1,3 , (−3)R1,4 :  
0 1 −2 0 0
0 3 −6 0 0
 
1 −1 2 −1 −1
0 1 −2 0 0
R2,3 , (−3)R2,3 , (−3)R2,4 : 
0 0

0 0 0
 0 0
 0 0 0
1 0 0 −1 −1
0 1 −2 0 0 
1R2,1 : 
0 0 0

0 0
0 0 0 0 0
The solution is x = u − 1, y = 2s, z = s, t = u.

1. The solution is x = −3t, y = −4t, z = 7t.

2. The system is inconsistent.

3. x = −4, y = 2, z = 7.
4. x = 3 + 2t, y = t.
 
1 1 2 a
5. The augmented matrix of the system is: 1 0 1 b .
2 1 3 c
 
1 1 2 a
A row echelon form : (−1)R1,2 , (−2)R1,3 : 0 −1 −1 b − a .
0 −1 −1 c − 2a
 
1 1 2 a
(−1)R2 , 1R2,3 : 0 1 1 a − b .
0 0 0 c−a−b
The system is consistent if and only if −a − b + c = 0.

1. (1, −1, 2) is a solution for the linear system if and only if a = 2, b = −1

and c = 0.
2. As the matrix of the linear system

 2x − y − 3z = −3
−2x + y + z = −1
2x + 3y = −1


has an inverse, the linear system has a unique solution.

(The determinant of the matrix is equal to 16)
 xxxix

 
4 6 −2
• B = adj(A) = −1 0 1  and |A| = 2.
−2 −2 2

• A−1 = 12 B.

 
−1 1 2
A − B = −2 2 2
4 −2 −4
A(A − B) = 2I3 .
A−1 = 12 (A − B).
The augmented matrix of the system is

     
1 1 2 1 11 1 1 2 1 11 1 1 2 0 6
1
 2 1 1 9 0
 −→  1 −1 0 −2 0
 −→  1 0 0 1 
1 1 1 2 6 0 0 −1 1 −5 0 0 1 0 3
2 1 1 1 14 0 −1 −3 −1 −8 0 0 0 1 −2

Then x = 6, y = 1, z = 3, t = −2.
CHAPTER IV

THE VECTOR SPACES

1 Introduction to Vector Spaces

Definition 1.1

A real vector space V is a non empty set whose elements are called
vectors and an operation (+) on the vectors called addition and scalar
multiplication (.) satisfying:

1. For any u and v in V , u + v is also in V ;

2. For any u, v, w in V , u + (v + w) = (u + v) + w.
3. There is an element in V called the zero or null vector, which we
denote by 0, such that for all u in V we have 0 + u = u.
4. For every u in V , there is a vector called the negative of u and
denoted −u, such that −u + u = 0.
5. For any u and v in V , u + v = v + u.

1. If k is any scalar in R and u is any vector in V , then ku is a vector in V .

2. For any scalar k in R and any vectors u and v in V , k(u + v) = ku + kv.

3. For any scalars k and ℓ in R and any vector u in V , (k + ℓ)u = ku + ℓu.
4. For any scalars k and ℓ in R and any vector u in V , k(ℓu) = (kℓ)u.

xli
xlii

5. For any vector u in V , 1u = u.

Example 35 :

1. Rn is a vector space.

2. The set {(x, y, 2x + 3y); x, y ∈ R} is a vector space.

3. The set of polynomials P = R[X] is a vector space.
Also the set of polynomials of degree less then n, Pn = Rn [X] is a vector
space.

1.1 Some Basic Identities in a Vector Space

Theorem 1.2

Let V be a vector space. The following statements are always true.

1. 0u = 0;
2. k0 = 0;
3. (−1)u = −u;
4. If ku = 0, then k = 0 or u = 0.

2 Vector Sub-Spaces
Definition 2.1

Let (V, +, .) be a vector space and F a nonempty subset of V . F is called

a sub-space of V if F is itself a vector space with the same operations
of the vector space V .

Theorem 2.2

Let (V, +, .) be a vector space and F be a nonempty subset of V . F is

a sub-space of V if and only if
1. For any u and v in F , u + v is also in F ;
 xliii

2. For any k in R and any vector u in F , the vector ku is in F .

Example 36 :
 
a b
1. The set F = { ; a, b ∈ R} is a sub-space of V = M2 (R).
0 2a − b
2. Let A ∈ Mm,n (R) be a matrix and F = {X ∈ Rn ; AX = 0}.
F is sub-space of V = Rn . F is the set of solutions of the homoge-
neous system AX = 0. F is also called the null space of A and denoted
nullspace(A).
3. The set F = {(x, x + 1); x ∈ R} is not a sub-space of R2 since (0, 0) ̸∈ F .
4. The set W = {A ∈ Mn : A = −AT } is a sub-space of Mn (R). (W is the
subspace of anti-symmetric matrices.)
Indeed: if A, B ∈ W and λ ∈ R, (A + B)T = AT + B T = −A − B and
(λA)T = λAT = −λA. Then W is a sub-space of Mn .
5. The set F = {A ∈ Mn : A = AT } is a sub-space of Mn (R). (F is the
subspace of symmetric matrices.)
6. The set E = {(x, y) ∈ R2 ; xy = 0} is not a sub-space since (1, 0) ∈ E
and (0, 1) ∈ E but (1, 0) + (0, 1) = (1, 1) ̸∈ E.

3 Linear Combination and Generating Sets

In this section, we give another way to construct subspaces.

Definition 3.1

Let v1 , . . . , vn be a set of vectors in a vector space V . A linear combi-

nation of the vectors v1 , . . . , vn is an expression of the form

x1 v1 + . . . + xn vn ,

where x1 , . . . , xn ∈ R.
A vector space v ∈ V is a linear combination of the vectors v1 , . . . , vn if
there is x1 , . . . , xn ∈ R such that

v = x1 v1 + . . . + xn vn .

Example 37 :
xliv

1. The vector (4, 1, 1) is a linear combination of the vectors (1, 0, 2),(2, −1, 3),
(0, −1, 1) because (4, 1, 1) = −2(1, 0, 2) + 3(2, −1, 3) − 4(0, −1, 1).
2. The vector (1, 1, 2) is not a linear combination of the vectors (1, 0, 2),
(0, −1, 1) because the linear system (1, 1, 2) = x(1, 0, 2)+y(0, −1, 1) don’t
have a solution.

Theorem 3.2

Let S = {v1 , . . . , vn } be a set of vectors in a vector space V , then

1. the set W of linear combinations of the vectors of S is a linear

sub-space in V .

2. W is the smallest sub-space of V which contains S.

This sub-space is called the sub-space generated (or spanned) by
the set S and denoted by ⟨S⟩ or Vect(S).

Example 38 :
 
a b
1. Let F = { ; a, b ∈ R}.
 0 2a −b
   
a b 1 0 0 1
= a +b . Then F is the sub-space of
0 2a − b 0 2 0 −1 
1 0 0 1
V = M2 (R) spanned by , .
0 2 0 −1

2. In R4 , consider the vectors v1 = (1, 2, 3, 4) and v2 = (1, −2, 3, −4). The

vector (a, 1, b, 1) is not a linear combination of the vectors v1 and v2 .
Indeed, the vector (a, 1, b, 1) ∈ Vect{v
 1 , v2 } if and onlyif the
 linear system
1 1 a
2 −2 1
AX = B is consistent with A =  3 3  and B =  b .
  

4 −4 1
The system is not consistent because the second and the forth equations
can not be true in the same time. ((2a − 2b = 1, 4a − 4b = 1)).
The vector (a, 1, 1, b) ∈ Vect{v 1 , v2 } ifand only if the
 linear
 system AX =
1 1 a
2 −2
 and B = 1. The system is
 
B is consistent with A =  3 3  1
4 −4 b
1
consistent and has a unique solution only if a = and b = 2.
3
 xlv

3. Let E = Vect{v1 , v2 }, where v1 = (2, 3, −1) and v2 = (1, −1, −2) in R3

and F = Vect{u1 , u2 }, where u1 = (3, 7, 0) and u2 = (5, 0, −7). The
sub-spaces E and F are equal. It suffices to prove that v1 , v2 ∈ F and
u1 , u2 ∈ E.  
2 1 3 5
For this consider the augmented matrix  3 −1 7 0 . This ma-
 −1 −2 0 −7
1 0 2 1
trix is row equivalent to the matrix 0 1 −1 3. We have a unique
0 0 0 0
solution. Then E = F .
4. In what conditions on a, b ∈ R such that the vector v = (−2, a, b, 5) is
in the sub-space of R4 generated by the vectors u = (1, −1, 1, 2) and
v = (−1, 2, 3, 1).
The vector v = (−2, a, b, 5) ∈ Vect{u, v} if the following linear system is
consistent AX = B, where
   
1 −1 −2
−1 2  a
A= 1
 and B =  . This system is consistent if and only
3  b 
2 1 5
if 3 = a − 2 = b+24 .
Then a = 5 and b = 10.

Theorem 3.3
 
x1
Let A be the matrix of type (m, n) and let X =  ...  be the matrix
 

xn
of type (n, 1). If C1 , . . . , Cn are the columns of the matrix A, then

AX = x1 C1 + . . . + xn Cn .

Corollary 3.4

Let A be a matrix of type (m, n). The linear system AX = B is consis-

tent if and only if the matrix B is a linear combination of the columns
of the matrix A.
xlvi

Definition 3.5

Let S = {v1 , . . . , vn } be a set of vectors in a vector space V .

We say that the vector space V is generated (or spanned) by the set S
if any vector in V is a linear combination of the vectors v1 , . . . , vn . (We
say also that S is a spanning set of V ).

Theorem 3.6

Let S = {v1 , . . . , vn } ⊂ Rm and A the matrix of type (m, n) with

columns v1 , . . . , vn . (We mean the coordinates of the vectors in the
standard basis of R4 )
The set S spans the vector space Rm if and only if the system AX = B
is consistent for all B ∈ Rm .

Example 39 :
1. Determine whether the vectors v1 = (1, −1, 4), v2 = (−2, 1, 3), and v3 =
(4, −3, 5) span R3 .
We solve
 the following  linearsystem
 AX = B, where
1 −2 4 a
A = −1 1 −3, B =  b  for arbitrary a, b, c ∈ R.
4 3 5 c
A
 reduced of the augmented
 matrix is given by:
1 0 2 −a − 2b
0 1 −1 −a − b .
0 0 0 7a + 11b + c
This system has a solution only when 7a + 11b + c = 0. Thus, the vectors
do not span R3 .
   
1 1 2 1
2. Determine whether the vectors v1 = and v2 = , span the
  0 1 0 3
a b
vector space F = { ; a, b ∈ R}.
0 2a − b

  x + 2y = a
a b
= xv1 + yv2 ⇐⇒ x+y = b .
0 2a − b
x + 3y = 2a − b

This system has the unique solution x = 2b − a and y = a − b.
 xlvii

4 Linear Dependence and Independence

Definition 4.1

A set of vectors v1 , . . . , vn in a vector space V are called linearly inde-

pendent if the equation

x1 v1 + . . . , +xn vn = 0

has 0 as unique solution.

Example 40 :
1. The vectors u = (1, 1, −2), v = (1, −1, 2) and w = (3, 0, 2) are linearly
independent in R3 .

 x + y + 3z = 0
xu + yv + zw = (0, 0, 0) ⇐⇒ x−y = 0.
−2x + 2y + 2z = 0

This system has 0 as unique solution.
 
1 1 3
The matrix of this system is  1 −1 0 and its determinant is −4.
−2 2 2

2. The set of vectors {P1 = 1 + x + x2 , P2 = 2 − x + 3x2 , P3 = x − x2 } is

linearly independent in P2 .
2
aP1 + bP 2 + cP3 = 0 ⇐⇒ (a + 2b) + (a − b + c)x + (a + 3b − c)x =
 a + 2b = 0
0 ⇐⇒ a − b + c = 0.
a + 3b − c = 0


Definition 4.2

A system of vectors {v1 , . . . , vn } in a vector space V is called linearly

dependent if they are not linearly independent.

Example 41 :
The vectors u = (0, 1, −2, 1), v = (1, 0, 2, −1) and w = (3, 2, 2, −1) are linearly
dependent in R4 .
xlviii



 y + 3z = 0
x + 2z = 0

xu + yv + zw = (0, 0, 0, 0) ⇐⇒ . This system has

 −2x + 2y + 2z = 0
x−y−z = 0

infinite many solutions.  
0 1 3 0
1 0 2 0
The extended matrix of this system is 
  and the reduced row
−2 2 2 0
  1 −1 −1 0
1 0 2 0
0 1 3 0
0 0 0 0.
form of this matrix is :  

0 0 0 0

Theorem 4.3

Let S = {v1 , . . . , vn } be a set of vectors in a vector space V , with n ≥ 2.

The set S is linearly independent if and only if no vector in the set is a
linear combination of the other vectors in the set.
The set S is linearly dependent if and only if one of the vectors in the
set is a linear combination of the other vectors in the set.

Theorem 4.4

Let S = {v1 , . . . , vn } ⊂ Rm and A the matrix of type (m, n) with

columns the vectors v1 , . . . , vn .
The set S is linearly independent if and only if the homogeneous system
AX = 0 has 0 as unique solution.

Example 42 :
1. If A is a matrix of type (m, n) with m < n. Then the homogeneous
system AX = 0 has an infinite many solutions.

2. If S = {v1 , . . . , vn } ⊂ Rm with m < n, then the set S is linearly depen-

dent.
 xlix

5 Bases and Dimension

Definition 5.1

Let S = {v1 , . . . , vn } be a set of vectors in a vector space V .

The set S is called a basis of the vector space V if :

1. The set S generates the vector space V

2. The set S is linearly independent.

Theorem 5.2

If S = {v1 , . . . , vn } is a basis of the vector space V , then any vector

v ∈ V can be written uniquely as a linear combination of vectors in the
basis S.

Proof .
If v = x1 v1 +. . .+xn vn = y1 v1 +. . .+yn vn , then (x1 −y1 )v1 +. . .+(xn −yn )vn =
0. Since the system S is linearly independent, xj = yj for all j = 1, . . . , n.
Remark 8 :
Let S = {e1 , . . . , en } in Rn , where e1 = (1, 0, . . . , 0), . . . , en = (0, . . . , 0, 1). The
set S is a basis of Rn and called the standard basis of Rn .
Example 43 :
1. The system S = {1, X, . . . , X n } is a basis of the vector space Pn .

2. Consider the system S = {v1 , v2 , v3 } in R3 , where v1 = (λ, 1, 1), v2 =

(1, λ, 1) and v3 = (1, 1, λ).
The system S is a basis of the vector space R3 if and only if λ ∈
̸ {−2, 1}.

Theorem 5.3

Let S = {v1 , . . . , vn } be a basis of the vector space V and let T =

{u1 , . . . , um } be a set of vectors.
If m > n, then T is linearly dependent.

Proof .
l

n
X
Consider uj = aj,k vk , for all j = 1, . . . , m.
k=1
If
m m n
!
X X X
0= xj uj = xj aj,k vk
j=1 j=1 k=1
 
n
X m
X
=  xj aj,k  vk .
k=1 j=1

m
X
Then xj aj,k = 0 for all k = 1, . . . , n. This system is an homogeneous
j=1
system of n equations and m variables and m > n. This system has an infinite
many solutions. Then T is linearly dependent.

Corollary 5.4

All bases of a finite dimensional vector space have the same number of
elements.

Definition 5.5

If S = {v1 , . . . , vn } is a basis of the vector space V then the number of

vectors n of S is called the dimension of the vector space V and denoted
by: dimV .

Theorem 5.6

Let V be a vector space of dimension n and S = {v1 , . . . , vn } a system

of vectors in V . The following conditions are equivalent:

1. S is a basis of V ;
2. S is linearly independent;
3. S generates the vector space V .
 li

Theorem 5.7

Any generating system S = {v1 , . . . , vn } of vectors contains a basis of

the vector space.

Consider the vector subspace F of Rn generated by a system of vectors

S = {v1 , . . . , vm }. We give two algorithms to construct a basis of F .
First Algorithm:

1. Construct the matrix A with rows the coordinates of the vectors of S.

2. The non zeros rows of any row echelon form of the matrix A form a basis
of the vector space F .
Second Algorithm:

1. Construct the matrix A with columns the coordinates of the vectors of

S.
2. Take any row echelon form C of A.
3. Let Ck1 , . . . Ckp be the columns which contain a leading number of C and
k1 < . . . < kp . Then the system of vectors {vk1 , . . . , vkp } form a basis of
the vector space F .

Theorem 5.8

1. If S = {v1 , . . . , vn } is a generating system of the vector space V ,

then S contains a basis of V .
2. If S = {v1 , . . . , vn } is a linearly independent system of vectors in
the vector space V , then there is a basis T of V which contain the
set S.

Proof .
In the second part of the theorem, we add to the system S any basis {e1 , . . . , em }
of V . We use the first algorithm to extract a basis from the generating system
{v1 , . . . , vn , e1 , . . . , em }. This basis contains S.
Example 44 :
1. Let W be the sub-space of R5 generated by the set of vectors: S = {v1 =
(1, 0, 2, −1, 2), v2 = (2, 0, 4, −2, 4), v3 = (1, 2, −1, 2, 0), v4 = (1, 4, −4, 5, −2).}
lii

(a) Find a basis of the sub-space W in {v1 , v2 , v3 , v4 }.

(b) Find a basis of R5 and contains {v1 , v3 }.
 
1 2 1 1
0 0 2 4
 
(a) Consider the matrix A =  2  4 −1 −4  with columns the
−1 −2 2 5
2 4 0 −2
coordinates of the vectors v1 , v2 , v3 , v4 in the standard
 basis of R5 .
1 2 0 −1
0 0 1 2 
 
The reduced row echelon form of the matrix A is  0 0 0 0 .

0 0 0 0 
0 0 0 0
Then {v1 , v3 } is a basis of the sub-space W .
(b) Consider the standard basis of R5 : e1 = (1, 0, 0, 0, 0), e2 = (0, 1, 0, 0, 0),
e3 = (0, 0, 1, 0, 0), e4 = (0, 0, 0, 1, 0), e5 = (0, 0, 0, 0, 1). We prove
that {v1 , v3 , e1 , e2 , e3 } is a basis of R5 and contain {v1 , v3 }.

2. Consider the vector sub-space of R4 ; W = {(x, y, z, t) ∈ R4 ; 2x + y + z =

0, x − y + z = 0}. We look for a basis of W .  
2 1 1 0
u = (x, y, z, t) ∈ W ⇐⇒ AX = 0, where A = and
1 −1 1 0
 
x
y 
X=  z .


t
 
2x + y + z = 0 x = −2y
AX = 0 ⇐⇒ ⇐⇒
x−y+z =0 z = 3y
   
−2 0
1 0
⇐⇒ X = y   3  + t 0.
  

0 1
Then {(−2, 1, 3, 0), (0, 0, 0, 1)} is a basis of the vector sub-space W .
 liii

6 Coordinate System and Change of Bases

Definition 6.1

If S = {v1 , . . . , vn } is a basis of the vector space V and v ∈ V such that

v = x1 v1 + . . . xn vn . Then (x1 , . . . xn ) are called the
 coordinate
 system
x1
of the vector v in the basis S. We denote [v]S =  ... 
 

xn

Theorem 6.2

If B = {v1 , . . . , vn } and C = {u1 , . . . , un } are two bases of the vector

space V . The matrix C PB ∈ Mn (R) with columns [v1 ]C , . . . , [vn ]C is
called the transition matrix from the basis B to the basis C. This matrix
−1
C PB is invertible, C PB = B PC and

[v]C = C PB [v]B , for all v ∈ V.

(B PC is the transition matrix from the basis C to the basis B.)

Example 45 :

1. Let B = {v1 = (0, 1, 1), v2 = (1, 0, −2), v3 = (1, 1, 0)} be a basis of the
vector space R3 and let C = {u1 = (1, 0, 0), u2 = (0, 1, 0), u3 = (0, 0, 1)}
3
be the standard
 basis
 of R .In this case
   
0 1 1 −2 2 −1 1
C PB =
1 0 1 and B PC = −1 1 −1. If [v]C =  0 ,
1 −2 0 2 −1 1 −1
 
−1
then [v]B = B PC [v]C =  0 .
1
2. Consider two bases
 B and Cof vector space V such that the transition
1 −1 −2
matrix C PB = 0 −1 1 . Let u be a vector in V such that [u]C =
  1 2 −2  
x 6y + 3z
y , then [u]B = B PC [u]C = C P −1 [u]C = 1  −x + z .
B 3
z −x + 3y + z
liv

7 The Null Space, the Row Space and the Col-

umn Space of Matrix
Definition 7.1

Let A be a matrix of type (m, n).

The vector sub-space of Rn spanned by the rows of the matrix A is
called the row vector space of the matrix A and denoted by: row(A).
The vector sub-space of Rm spanned by the columns of the matrix A is
called the column vector space of the matrix A and denoted by: col(A).

Theorem 7.2

Let A be a matrix and B a matrix which is a result of some row opera-

tions on the matrix A, then row(A) = row(B).

Theorem 7.3

Let A be a matrix and B any row echelon form of the matrix A. Then
the set of non zero rows of the matrix B is a basis of the row vector
space of A.

Definition 7.4

The dimension of the row vector space of a matrix A is called the rank
of A.

rank(A) = dim(row(A)).

Remark 9 :
The rank of a matrix A is the numbers of leading numbers in any row echelon
form of the matrix A.
 lv

Theorem 7.5

For any matrix A:

rank(A) = dim(row(A)) = dim(col(A)).

Corollary 7.6

For any matrix A: rank(A) = rank(AT ).

Corollary 7.7

If A is a matrix of type (m, n) and P is any invertible matrix of type m

and Q an invertible matrix of type n, then

rank(A) = rank(P AQ).

Proof .
There exist some elementary matrices E1 , . . . , Ep of order m such that P =
E1 . . . Ep .
We know that if E is a elementary matrix which corresponds to an elementary
row operation R, then EA is the result of this elementary row operation R on
the matrix A. Hence
rank(A) = rank(P A).
Also rank(P AQ) = rank(P AQ)T = rank(QT AT P T ) = rank(AT P T ) = rank(P A) =
rank(A).

Theorem 7.8

If A is a matrix of type (m, n). We have the equivalence of the following

statements:

1. The homogeneous system AX = 0 has 0 as unique solution.

2. The columns of the matrix A are linearly independent.
3. rank(A) = n.
4. The matrix AT A is invertible.
lvi

Theorem 7.9

Let A be a matrix of type (m, n). We have the equivalence of the

following statements

1. The system AX = B is consistent for all B ∈ Rm .

2. The columns of the matrix A generates the vector space Rm .

3. rank(A) = m.
4. The matrix AAT is invertible.

Definition 7.10

Let A be a matrix of type (m, n). The vector sub-space {X ∈ Rn ; AX =

0} is called the nullspace of the matrix A and denoted by: N (A). The
dimension of N (A) is denoted by nullity(A).
Also the vector sub-space {AX; X ∈ Rn } is called the image (or the
range) of the matrix A and denoted by: Im(A).

Theorem 7.11

For any matrix A, Im(A) = col(A).

Theorem 7.12: [Rank-Nullity Theorem]

For any matrix A ∈ Mm,n (R), nullity(A) + rank(A) = n.

How to find bases for the null space, row space and column space of a matrix
Given a matrix A of type (m, n)
1. Reduce the matrix A to a row echelon form R.
2. Solve the system RX = 0. Find a basis for the solutions space. The same
basis for the solution space of RX = 0 is a basis for the null space of A.

3. Consider the non-zero rows of R. They form a basis for the row space of
R. The same basis for the row space of R is a basis for the row space of
 lvii

A.

4. Take the columns of R with leading 1s . They form a basis for the column
space of R. The corresponding column vectors in A form a basis for the
column space of A.

Example 46 :
 
1 2 −1 −1
0 −1 2 3
1. Consider the matrix A = 
2

3 0 1
1 1 1 2

(a) Find a basis of the vector space N (A).

(b) Find a basis of the vector space Col(A).
(c) Find the rank of the matrix A.
 
1 0 3 5
0 1 −2 −3
The reduced row echelon form of the matrix A is 
0
.
0 0 0
0 0 0 0

(a) (−3, 2, 1, 0), (−5, 3, 0, 1) is basis of the vector space N (A).

(b) (0, 1, 2, 1), (−1, 2, 3, 1) is a basis of the vector space Col(A).
(c) The rank of the matrix A is 2.

2. Consider in R3 the vectors, u1 = (1, 2, 1), u2 = (1, 3, 2), u3 = (1, 1, 0) and

u4 = (3, 8, 5).
Let F = Vect(u1 , u2 ) and G = Vect(u3 , u4 ). We claim that F = G.
As the vectors u1 , u2 are linearly independent and also the vectors u3 , u4 ,
then dimE = dimF = 2.  
1 2 1
1 3 2
F = G if and only if the rank of the following matrix is 2, A = 
1 1 0.


 3 8 5

1 0 −1
0 1 1 
The reduced row form of this matrix is   . Then F = G.
0 0 0
0 0 0
CHAPTER V

INNER PRODUCT SPACES AND ORTHOGONALITY

1 Inner Product
Definition 1.1

An inner product on a vector space V is a function that assigns a real

number ⟨x, y⟩ to every pair of vectors x, y in V in a such away that
satisfies the following properties:
1. For all u, v ∈ V : ⟨u, v⟩ = ⟨v, u⟩;

2. For all u, v, w ∈ V : ⟨u + v, w⟩ = ⟨u, w⟩ + ⟨v, w⟩;

3. For all u, v ∈ V and r ∈ R: ⟨ru, v⟩ = r⟨u, v⟩;
4. ⟨u, u⟩ > 0 for all u ̸= 0.

Example 47 :

1. The standard inner product on Rn defined by:

n
X
⟨u, v⟩ = xj yj = x1 y1 + . . . + xn yn ,
j=1

where u = (x1 , . . . , xn ) and v = (y1 , . . . , yn ).

2. If E = C([0, 1]) the vector space of continuous functions on [0, 1]. For all
f, g ∈ E, we define the inner product of f and g by:

lix
lx

Z 1
⟨f, g⟩ = f (t)g(t)dt.
0

3. On R2 , define for u = (x, y) and v = (x′ , y ′ ), the function:

⟨u, v⟩ = 2xx′ + yy ′ − xy ′ − yx′ .

⟨ , ⟩ is an inner product on R2 . We only prove that ⟨u, u⟩ ≥ 0 and

⟨u, u⟩ = 0 ⇐⇒ u = 0. The other properties are left to the reader.
⟨u, u⟩ = 2x2 + y 2 − 2xy = (x − y)2 + x2 ≥ 0 and ⟨u, u⟩ = 0 ⇐⇒ u = 0.
4. On R3 , define for u = (x, y, z) and v = (x′ , y ′ , z ′ ) the function:

⟨u, v⟩ = 2xx′ + yy ′ + 3zz ′ − xy ′ − yx′ + yz ′ + zy ′ .

⟨ , ⟩ is an inner product on R3 . We only prove that ⟨u, u⟩ ≥ 0 and

⟨u, u⟩ = 0 ⇐⇒ u = 0. The other properties are left to the reader.

⟨u, u⟩ = (y + z − x)2 − (z − x)2 + 2x2 + 3z 2

= (y + z − x)2 + (x + z)2 + z 2 ≥ 0

⟨u, u⟩ = 0 ⇐⇒ z = x = y = 0 ⇐⇒ u = 0.

5. On R3 , define for u = (x, y, z) and v = (x′ , y ′ , z ′ ) the function:

⟨u, v⟩ = 2xx′ + yy ′ + zz ′ − xy ′ − yx′ + yz ′ + zz ′ .

⟨ , ⟩ is not an inner product on R3 .

⟨u, u⟩ = (y + z − x)2 − (z − x)2 + 2x2 + z 2

= (y + z − x)2 + x2 + 2xz
= (y + z − x)2 + (x + z)2 − z 2 .

If A = (aj,k ) ∈ Mn (R), we define the trace of the matrix A by: tr(A) =

6. P
n
j=1 aj,j .
The function (A, B) 7−→ ⟨A, B⟩ = tr(AB T ) is an inner product on the
vector space Mn (R).
7. ⟨u, v⟩ = xx′ + yy ′ + 2zz ′ + yx′ + 2xy ′ + yz ′ + y ′ z is not an inner product
on R3 because ⟨u, v⟩ − ⟨v, u⟩ = xy ′ − yx′ .
8. ⟨u, v⟩ = xy ′ + yx′ + yz ′ + zy ′ + 3xz ′ + 3zx′ is not an inner product
on R3 because ⟨u, u⟩ = 2xy + 2yz + 6xz = 2(x + z)(y + 3z) − 6z 2 =
(x + y + 4z)2 − (x − y − 2z)2 − 6z 2 .
 lxi

9. ⟨u, v⟩ = xx′ + yy ′ + zz ′ + yx′ + xy ′ + yz ′ + y ′ z + zx′ + xz ′ is not an inner

product on R3 because

⟨u, u⟩ = x2 + y 2 + z 2 + 2xy + 2yz + 2xz = (x + y + z)2 .

10. ⟨u, v⟩ = xx′ + yy ′ + zz ′ − yz ′ − zy ′ + xz ′ + x′ z is not an inner product on

R3 because

⟨u, u⟩ = x2 + y 2 + z 2 − 2yz + 2xz

= (x + z)2 + y 2 − 2yz
= (x + z)2 + (y − z)2 − z 2

11. Find the values of a, b such that

⟨(x, y), (x′ , y ′ )⟩ = xx′ + yy ′ + axy ′ + byx′

is an inner product on R2 .
⟨(x, y), (x′ , y ′ )⟩ = ⟨(x′ , y ′ ), (x, y)⟩ if a = b.

⟨(x, y), (x, y)⟩ = x2 + y 2 + 2axy

= (x + ay)2 + y 2 (1 − a2 ).

Then ⟨ , ⟩ is an inner product on R2 if and only if |a| < 1.

Remark 10 :
If (E, ⟨ , ⟩) is an inner product space and u, v, w, x ∈ E, a, b, c, d ∈ R, we have:

⟨u + v, w + x⟩ = ⟨u, w⟩ + ⟨u, x⟩ + ⟨v, w⟩ + ⟨v, x⟩.

⟨au + bv, cw + dx⟩ = ac⟨u, w⟩ + ad⟨u, x⟩ + bc⟨v, w⟩ + bd⟨v, x⟩.

2 The Orthogonality
Definition 2.1

Let (E, ⟨ , ⟩) be an inner product space and u, v ∈ E. We define

p
1. the norm of u by: ∥u∥ = ⟨u, u⟩;

2. the distance between u and v by: d(u, v) = ∥u − v∥;

3. the angle 0 ≤ θ ≤ π between the vectors u, v ∈ E by:
lxii

⟨u, v⟩
cos θ =
∥u∥.∥v∥

Example 48 :
Consider the inner product space M2 (R), ⟨ , ⟩) defined by:
⟨A, B⟩ = tr(AB T ).
     
1 −1 2 1 T 1 0
Let A = and B = . We have AB = , ∥A∥2 = 15,
2 3 1 1 7 5
√
2 3
∥B∥2 = 7. Then if θ is the angle between the matrices A and B, cos θ = √ .
35

Theorem 2.2: [Cauchy-Schwarz Inequality]

If (E, ⟨ , ⟩) is an inner product space and u, v ∈ E,

then
|⟨u, v⟩| ≤ ∥u∥∥v∥. (2.1)
We have the equality in (2.1) if and only if the vectors u, v are linearly
dependent.

Proof .
Let Q(t) be the polynomial
Q(t) = ∥u + tv∥2 = ∥u∥2 + 2t⟨u, v⟩ + t2 ∥v∥2 .
Since Q(t) ≥ 0 for all t ∈ R, then the discriminant of Q(t) is non positive.
Then ⟨u, v⟩2 ≤ ∥u∥2 ∥v∥2 . If |⟨u, v⟩| = ∥u∥∥v∥, the discriminant of Q is zero
and there exists t ∈ R such that Q(t) = 0. Then u + tv = 0.

Theorem 2.3

If (E, ⟨ , ⟩) is an inner product space and u, v ∈ E, then

∥u + v∥ ≤ ∥u∥ + ∥v∥.

Proof .
∥u + v∥2 = ∥u∥2 + ∥v∥2 + 2⟨u, v⟩
≤ ∥u∥2 + ∥v∥2 + 2∥u∥ ∥v∥ = (∥u∥ + ∥v∥)2 .
 lxiii

Definition 2.4

Let (E, ⟨ , ⟩) be an inner product space and u, v ∈ E. The vectors u, v

are called orthogonal and we denote u ⊥ v, if ⟨u, v⟩ = 0.

Theorem 2.5: [Pythagor’s Theorem]

u ⊥ v if and only if ∥u + v∥2 = ∥u∥2 + ∥v∥2 .

Proof .
∥u + v∥2 = ∥u∥2 + ∥v∥2 + 2⟨u, v⟩ = ∥u∥2 + ∥v∥2 .

3 The Orthonormal Bases

Definition 3.1

Let (E, ⟨ , ⟩) be an inner product space. A set of non zeros vectors

S = {e1 , . . . , en } is called orthogonal if

⟨ej , ek ⟩ = 0, ∀1 ≤ j ̸= k ≤ n.

and S called normal if ∥ej ∥ = 1, ∀1 ≤ j ≤ n.

The system S is called orthonormal if

⟨ej , ek ⟩ = δj,k , ∀1 ≤ j, k ≤ n.

(δj,k = 0 for all j ̸= k and δj,j = 1.)

Theorem 3.2

Any set of non zero orthogonal vectors is linearly independent.

lxiv

Theorem 3.3:

Let (E, ⟨ , ⟩) be an inner product space and S = {e1 , . . . , en } an or-

thonormal basis of E, then for all u ∈ E

u = ⟨u, e1 ⟩e1 + . . . + ⟨u, en ⟩en .

Proof .
n
X Pn
If u = aj ej , then ⟨u, ek ⟩ = j=1 aj ⟨ej , ek ⟩ = ak .
j=1

Theorem 3.4: [Gramm-Schmidt Algorithm]

Let (E, ⟨ , ⟩) be an inner product space and (v1 , . . . , vn ) a linearly

independent set of vectors in E. Then there exists a unique orthonormal
set (e1 , . . . , en ) such that

1. for all k ∈ {1, . . . , n},

Vect{e1 , . . . , ek } = Vect{v1 , . . . , vk };

2. for all k ∈ {1, . . . , n}, ⟨ek , vk ⟩ > 0.

Proof .
Consider the orthogonal set of vectors {u1 , . . . , un } defined by iteration : u1 =
k−1
X ⟨ui , vk ⟩
v1 and for all 1 ≤ k ≤ n, uk = vk − ui . The set {e1 , . . . , en } defined
i=1
∥ui ∥2
by: ek = ∥uukk ∥ , k ∈ {1, . . . , n} is orthonormal.
Example 49 :
1. Let F be the vector sub-space of R4 spanned by the vectors S = {u =
(1, 1, 0, 0), v = (1, 0, −1, 0), w = (0, 0, 1, 1)}. S is a basis of F . In use of
Gramm-Schmidt algorithm, we construct an orthonormal basis of F with
respect to the Euclidean inner product .
u1 = √12 (1, 1, 0, 0), u2 = √16 (1, −1, −2, 0), u3 = √112 (1, −1, 1, 3).
B = {u1 , u2 , u3 } is an orthonormal basis of the sub-space F .
Consider the vector v = (2, 3, −4, 1).
v = ⟨v, u1 ⟩u1 + ⟨v, u2 ⟩u2 + ⟨v, u3 ⟩u3 = √52 u1 + √76 u2 − √13 u3 .
 lxv

2. Consider the base {u1 = (1, −1), u2 = (1, 2)} of R2 and the following
inner product

⟨(x, y), (x′ , y ′ )⟩ = xx′ + xy ′ + yx′ + 2yy ′ .

The vector u1 is unitary with respect to the inner product and using the
Gramm-Schmidt algorithm, the vector v2 = (1, 0) is orthogonal to u1 and
{v1 = (1, −1), v2 = (1, 0)} is an orthonormal basis of R2 .
3. Consider the basis
 S = {u1 ,  } of the
u2 , u3 , u4  space 
M2 (R)defined by:
1 −1 1 0 1 0 0 1
u1 = , u2 = , u3 = , u4 = . In use of
0 1 1 1 0 2 1 1
Gramm-Schmidt algorithm, we  construct  an orthonormal basis from the
1 1 −1
basis S and , we find: v1 = √ .
3 0 1   
2 1 1 2 1 1 2
⟨u2 , v1 ⟩ = √ , u2 − ⟨u2 , v1 ⟩v1 = . Then v2 = √ .
3 3 3 1 15 3 1
√ 3
⟨u3 , v1 ⟩ = 3, ⟨u3 , v2 ⟩ = √
15    
1 −1 3 1 −1 3
u3 − ⟨u3 , v1 ⟩v1 − ⟨u3 , v2 ⟩v2 = . Then v3 = √ .
5 −3 4 35 −3 4
6 4
⟨u4 , v1 ⟩ = 0, ⟨u4 , v2 ⟩ = √ , ⟨u4 , v3 ⟩ = √
15 35  
1 −10 −39
u4 − ⟨u4 , v1 ⟩v1 − ⟨u4 , v2 ⟩v2 − ⟨u4 , v3 ⟩v3 = . Then v4 =
  35 −29 −29
1 2 1
√ .
7 −1 −1

4. Let F be the vector sub-space of the Euclidean space R4 spanned by the

following vectors u1 = (1, 2, 0, 2), u2 = (−1, 1, 1, 1).
Using Gramm-Schmidt algorithm we construct an orthonormal basis of
the vector sub-space F .
1
v1 = u1 , ⟨u2 , v1 ⟩ = 1,
3
u2 − ⟨u2 , v1 ⟩v1 = (0, 3, 1, −1) − 13 (−1, 1, 1, 1) = 13 (−4, 1, 3, 1). Then v2 =
1
√
3 3
(−4, 1, 3, 1).
(v1 , v2 ) is an orthonormal basis of the vector sub-space F .
The set F ⊥ = {u ∈ R4 : ⟨u, v⟩ = 0, ∀v ∈ F } is vector sub-space of
R4 . Indeed: If v1 , v2 ∈ F ⊥ , α, β ∈ R and u ∈ F , then ⟨αv1 + βv2 , u⟩ =
α⟨v1 , u⟩ + β⟨v2 , u⟩ = 0. Now we construct an orthonormal basis of F ⊥ .
Let u = (x, y, z, t) ∈ R4 .
 
⟨u, u1 ⟩ = 0 x + 2y + 2t = 0
u ∈ F ⊥ ⇐⇒ ⇐⇒
⟨u, u2 ⟩ = 0 −x + y + z + t = 0
lxvi

x = 23 z
 
x + 2y + 2t = 0
⇐⇒
−x + y + z + t = 0 y = − z3 − t

Then u ∈ F ⊥ ⇐⇒ u = − z3 (−2, 1, −3, 0) + t(0, −1, 0, 1).

The vectors e1 = (−2, 1, −3, 0), e2 = (0, −1, 0, 1) is an orthogonal basis of
the vector sub-space F ⊥ .
w1 = √114 e1 , ⟨w1 , e2 ⟩ = − √114 ,
1
e2 − ⟨e2 , w1 ⟩w1 = 14 (2, 13, 3, 14).
Then ( √14 (−2, 1, −3, 0), 3√142 (2, 13, 3, 14)) is an orthonormal basis of the
1

vector sub-space F ⊥ .
CHAPTER VI

LINEAR TRANSFORMATIONS

1 Definition of Linear Transformations

Definition 1.1

Let V and W be two vector spaces. A function T : V −→ W is called a

linear transformation if for all u, v ∈ V and k ∈ R

1. T (u + v) = T (u) + T (v), (additivity).

2. T (ku) = kT (u), (homogeneity).

If V = W , the linear transformation from V to V is called a linear

operator.

1.1 Basic Properties of Linear Transformations

If T : V −→ W is a linear transformation, then

1. T (0) = 0. 3. T (u − v) = T (u) − T (v).

2. T (−u) = −T (u).

Example 50 :
1. T : R2 −→ R3 defined by: T (x, y) = (x + 3y, 2x − y, x + 5y);
2. T : Mm,n (R) −→ Mn,m (R) defined by: T (A) = AT ;

lxvii
lxviii

3. T : P2 (R) −→ P3 (R) defined by:

T (a + bX + cX 2 ) = (a + b − 2c) + cX + (a + c)X 2 + (a + b)X 3 ;
4. T : P2 (R) −→ R3 defined by:
T (a + bX + cX 2 ) = (a + 3b − c, b − c, 2a − b + 3c);

Example 51 :
The following functions are not linear transformations

1. T : R3 → R2 , defined by: T (x, y, z) = (xy, z) because

T (2, 2, 0) = (4, 0) ̸= 2T (1, 1, 0) = (2, 0).
2. T : R3 → R2 , defined by: T (x, y, z) = (x + y − 3z, z + y − 1) because
T (0) ̸= 0;

3. T : R3 → R3 , defined by: T (x, y, z) = (x+y, z+y, x2 ) because T (2, 0, 0) =

(2, 0, 4) ̸= 2T (1, 0, 0) = (2, 0, 2).
4. T : Mn (R) −→ R defined by: T (A) = detA because det(A + B) ̸= detA +
detB in general.

Theorem 1.2

If T : V −→ W is a mapping, then T is a linear transformation if and

only if

T (αu + βv) = αT (u) + βT (v) ∀u, v ∈ V, α, β ∈ R.

Theorem 1.3

If T : V −→ W is a linear transformation, then

T (α1 u1 + . . . + αn un ) = α1 T (u1 ) + . . . + αn T (un ),

for all α1 , . . . , αn ∈ R and u1 , . . . , un ∈ V .

Remarks 11 :
1. If S = {u1 , . . . un } is a basis of the vector space V . A linear transforma-
tion T : V −→ W is well defined if T (u1 ), . . . , T (un ) are defined.
2. The unique linear transformations T : R −→ R are T (x) = ax, a ∈ R.
 lxix

3. The unique linear transformations T : R2 −→ R are T (x, y) = ax + by,

a, b ∈ R.

Example 52 :
Consider the basis {v1 = (1, 1, 0), v2 = (1, 1, 1), v3 = (1, 0, 1)} of R3 and con-
sider the linear transformation T : R3 −→ R2 such that T (v1 ) = (2, 1), T (v2 ) =
(−1, 3) and T (v3 ) = (1, 2). If v = (x, y, z) ∈ R3 , then v = (x − z)v1 + (z − x +
y)v2 +(x−y)v3 and T (x, y, z) = (x−z)(2, 1)+(z −x+y)(−1, 3)+(x−y)(1, 2) =
(4x − 2y − 3z, y + 2z).

Theorem 1.4

If A ∈ Mm,n (R), then the mapping TA : Rn −→ Rm defined by:

TA (X) = AX for all X ∈ Rn is a linear transformation. TA is called
the linear transformation associated to the matrix A.

Theorem 1.5

Let T : Rn −→ Rm be a linear transformation and let B = (e1 , . . . , en )

be a basis of the vector space Rn and C = (u1 , . . . , um ) a basis of the
vector space Rm . Then T = TA , where A =∈ Mm,n (R) with columns
[T (e1 )]C , . . . , [T (en )]C .
The matrix A is called the matrix of the linear transformation T with
respect to the basis B and C.

Theorem 1.6

Let V, W be two vector spaces and S = {v1 , . . . , vn } a basis of the vector

space V and {w1 , . . . , wn } a set of vectors in the vector space W .
There is a unique linear transformation T : V −→ W such that T (vj ) =
wj for all 1 ≤ j ≤ n.
lxx

2 Kernel and Image of Linear Transformations

Definition 2.1

Let T : V −→ W be a linear transformation. The set {v ∈ V ; T (v) =

0} is called the kernel of the linear transformation T and denoted by:
ker(T).
The set {T (v); v ∈ V } is called the range or the image of the linear
transformation T denoted by: Im(T).

Theorem 2.2

If T : V −→ W is a linear transformation, then ker(T) is a vector sub-

space of V and Im(T) is a vector sub-space of W .

Definition 2.3

If T : V −→ W is a linear transformation then dimension the vector

space ker(T) is called the nullity of the linear transformation T and
denoted by: (nullity(T )).
The dimension of the vector space Im(T) is called the rank of the linear
transformation T and denoted by: (rank(T )).

Example 53 :
If A ∈ Mm,n (R) and TA : Rn −→ Rm the linear transformation defined by:
TA (X) = AX, then rank(T ) = rankA, and Im(T ) = colA.
Example 54 :
Let T : R3 −→ R2 be the linear transformation defined by T (x, y, z) = (2x −
y + 3z, x − 2y + z). 
2x − y + 3z = 0
(x, y, z) ∈ ker(T ) ⇐⇒
x − 2y + z = 0
 
2 −1 3 0
The extended matrix of this linear system is: .
1 −2 1 0
Then (x, y, z) ∈ ker(T ) ⇐⇒ x = 5y, z = −3y. Hence
ker(T ) = Vect{(5, 1, −3)}.
 lxxi

As T (x, y, z) = x(2, 1) + y(−1, −2) + z(3, 1), then

Im(T ) = Vect{(2, 1), (−1, −2), (3, 1)} = Vect{(2, 1), (−1, −2)}.

Theorem 2.4

If T : V −→ W is a linear transformation and {v1 , . . . vn } is a basis of

the vector space V , then the set {T (v1 ), . . . T (vn )} generates the vector
space Im(T ).

Theorem 2.5: [The Dimension Theorem of the Linear Trans-

formations]

If T : V −→ W is a linear transformation and if dimV = n, then

nullity(T ) + (rank(T ) = n.
i.e.
dimker(T ) + dimIm(T ) = n.

Definition 2.6

If T : V −→ W is a linear transformation,

1. T is called injective if T (u) = T (v) ⇒ u = v, for all u, v ∈ V .

2. T is called surjective if Im(T ) = W .

Theorem 2.7

If T : V −→ W is a linear transformation. The linear transformation T

is injective if and only if ker(T ) = {0}.
lxxii

Corollary 2.8

If T : V −→ W is a linear transformation and dim V = dim W = n.

Then the linear transformation T is injective if and only if T is surjective.

Example 55 :
Consider the linear transformation T : R4 −→ R4 defined by:

T (x, y, z, t) = (x − y, 2z + 3t, y + 4z + 3t, x + 6z + 6t).

(x, y, z, t) ∈ Ker(T ) ⇐⇒ x = y = 3t = −2z. Then (6, 6, −3, 2) is a basis the

kernel of T .
The range of T is the column space of the matrix
 
1 −1 0 0
0 0 2 3
0 1 4 3 .
 

1 0 6 6

The row reduced form of this matrix is

 
1 −1 0 0
0 1 4 3 
0 0 1 3  .
 
2
0 0 0 0

Then {(1, 0, 0, 1), (−1, 0, 1, 0), (0, 2, 4, 6)} is a basis of the range of T .
Example 56 :
Let V, W be two vector spaces and T : V −→ W a linear transformation. If T
is injective and S = {u1 , . . . , un } is a set of linearly independent, then the set
{T (u1 ), . . . , T (un )} is linearly independent.

x1 T (u1 ) + . . . + xn T (un ) = 0 ⇐⇒ T (x1 u1 + . . . + xn un ) = 0

⇐⇒ x1 u1 + . . . + xn un = 0

since T is injective and the set S is linearly independent, then x1 = . . . = xn =

0.

3 Matrix of Linear Transformation and Change

of Basis
 lxxiii

Definition 3.1

Let T : V −→ W be a linear transformation, B = (u1 , . . . , un ) be a basis

C
of V and C = (v1 , . . . , vm ) a basis of W . The matrix [T ]B with columns
[T (u1 )]C , . . . , [T (un )]C is called the matrix of the linear transformation
T with respect to the basis B and the basis C. This matrix satisfies
C
[T (v)]C = [T ]B [v]B ; ∀v ∈ V.
C
If V = W and B = C we write the matrix [T ]C instead of [T ]B .

Example 57 :
Let T : R3 −→ R2 be the linear transformation defined by the following:
T (x, y, z) = (2x − y + 3z,
 x − 2y +z). The matrix of T with respect to the
2 −1 3
standard basis of R3 is:
1 −2 1
Example 58 :
Consider the linear transformation of R3 defined by: T1 ((1, 0, 0)) = (1, 1, 1),
T1 ((0, 1, 0)) = (1, 2, 2), T1 ((0,
0, 1)) = (1,
 2, 3). The matrix of T with respect to
1 1 1
the standard basis of R3 is 1 2 2 and T1 (x, y, z) = (x + y + z, x + 2y +
1 2 3
2z, x + 2y + 3z).

Theorem 3.2

If T : V −→ V is a linear transformation and B and C are basis of the

vector space V , then

[T ]B = B PC [T ]C C PB .

Example 59 :
Let T : R3 −→ R3 be the linear transformation such that its matrix with respect
to the standard basis C of the vector space R3 is
 
−3 2 2
[T ]C = −5 4 2 .
1 −1 1

Consider the basis B = {u = (1, 1, 1), v = (1, 1, 0), w = (0, 1, −1)} of R3 .

lxxiv

 
1 1 0
C PB =
1 1 1 . Then the matrix of T with respect to the basis S
1 0 −1
and the basis B is
 
−1 1 1
−1  2 −1 −1
B PC = S PB =
−1 1 0
and  
1 0 0
[T ]B = B PC [T ]C C PB = 0 −1 0 .
0 0 2

Example 60 :
Consider the linear transformation T : R3 −→ R3 defined by:
T (x, y, z) = (3x + 2y, 3y + 2z, 9x − 4z).  
3 2 0
The matrix of the linear transformation T is A = 0 3 2 .
9 0 −4
 
3 2 0 0
The extended matrix of the linear system AX = 0 is: 0 3 2 0. This
9 0 −4 0
 
3 2 0 0
matrix is equivalent to the matrix 0 1 0 0.
0 0 1 0
Then ker(T ) = {0} and the range of T is R3 .
Consider the basis S = {(0, 0, 1), (0, 1, 1), (1, 1, 1)} of rb3 .
The
 matrix the linear transformation
 T with respect to the basis S is P −1 AP =

−6 −9 0 0 0 1
2 3 0, with P = 0 1 1 and
0  2 5  1 1 1
0 −1 1
P −1 = −1 1 0.
1 0 0
Example 61 :
Consider u1 = 13 (1, 2, 2), u2 = 13 (2, 1, −2), u3 = 13 (2, −2, 1).
1 2 2
As the determinant 2 1 −2 = −27, then the set{u1 , u2 , u3 } is a basis of
2 −2 1
R3 . Also, we have ∥u1 ∥ = ∥u2 ∥ = ∥u3 ∥ = 1 and ⟨u1 , u2 ⟩ = ⟨u1 , u3 ⟩ = ⟨u2 , u3 ⟩ =
0, then the set {u1 , u2 , u3 } is an orthonormal basis of R3 .
Define the linear transformation T : R3 7→ R3 by: T (e1 ) = u1 , T (e2 ) = u2
and T (e3 ) = u3 , where {e1 , e2 , e3 } the standard basis of R3 . The matrix of the
 lxxv

linear transformation T with respect to the basis {e1 , e2 , e3 } is

 
1 2 2
1
P = 2 1 −2
3
2 −2 1

and
1
T (x, y, z) = (x + 2y + 2z, 2x + y − 2z, 2x − 2y + z).
3
Define the linear transformation S : R3 7→ R3 by:
S(x, y, z) = (−x + 2z, y  + 2z, 2x + 2y).
 The matrix of S with respect to the
−1 0 2
basis {e1 , e2 , e3 } is A =  0 1 2 . The matrix of S with respect to basis
2 2 0
{u1 , u2 , u3 } is B = P −1 AP . As P −1 = P T = P , hence
   
3 0 0 1 0 0
B = 0 −3 0 , B n = 3n 0 (−1)n 0 and An = P B n P.
0 0 0 0 0 0

If u = xu1 + yu2 + zu3 , then S(u) = 3xu1 − 3u2 .

Example 62 :  
2 −2 3
Let the matrix A = −2 2 3 . We define the linear transformation
3 3 −3
T : R3 7→ R3 defined by the matrix A with respect to the standard basis
(e1 , e2 , e3 ) of the vector space R3 .

1. Find T (x, y, z).

2. Find an orthogonal basis (u1 , u2 , u3 ) of the vector space R3 such that
T (u1 ) = 3u1 and T (u2 ) = 4u2 .
3. Find the matrix of the linear transformation T with respect to the basis
(u1 , u2 , u3 ).
4. We define the linear transformation S : R3 −→ R3 by the following:
S(e1 ) = u1 , S(e2 ) = u2 and S(e3 ) = u3 .
Find the matrix P of the linear transformation S with respect to standard
basis.

1. Prove that the matrix P has an inverse and find P −1 .

2. Let the linear transformation U defined by the matrix P −1 with respect
to the standard basis.
Find U (uk ) for all k = 1, 2, 3.
lxxvi

3. Let F = U ◦ T ◦ S.
Find F (e1 ), F (e2 ), F (e3 ).
Find the matrix of the linear transformation F and conclude the value
An for all n ∈ N.
Solution
1.
T (x, y, z) = (2x − 2y + 3z, −2x + 2y + 3z, 3x + 3y − 3z).

2. Let u = (x, y, z).


−x − 2y + 3z = 0
T (u) = 3u ⇐⇒ −2x − y + 3z ⇐⇒ x = y = z.
3x + 3y − 6z = 0


We take u1 = (1, 1, 1).


−2x − 2y + 3z = 0 
x = −y
T (u) = 4u ⇐⇒ −2x − 2y + 3z ⇐⇒ .
z=0
3x + 3y − 7z = 0


We take u2 = (1, −1, 0) and we can choose u3 = (1, 1, −2).

3. The matrix of the linear transformation T with respect to basis (u1 , u2 , u3 )
is  
3 0 0
[T ] = 0 4 0
0 0 −6
4.  
1 1 1
P = 1 −1 1
1 0 −2

1. the matrix P has an inverse,

 then (u1 , u2 , u3 ) is a basis.
2 2 2
P −1 = 61 3 −3 0 .
1 1 −2
2. U (u1 ) = (1, 0, 0), U (u2 ) = (0, 1, 0), U (u3 ) = (0, 0, 1).
3. F = U ◦ T ◦ S.
F (e1 ) = U ◦ T (u1 ) = 3U (u1 ) = 3(1, 0, 0), F (e2 ) = U ◦ T (u2 ) = 4U (u2 ) =
4(0, 1, 0), F (e3 ) = U ◦ T (u3 ) = −6U (u3 ) = −6(0, 0, 1).
The matrix of the linear transformation F is
 
3 0 0
D = 0 4 0  .
0 0 −6
 lxxvii

An = P Dn P −1 .
CHAPTER VII

DIAGONALIZATION OF MATRIX

1 Eigenvalues and Eigenvectors

Definition 1.1

If A ∈ Mn (R) and λ ∈ R.
λ is called an eigenvalue of the matrix A if there is X ∈ Rn \ {0} such
that
AX = λX.
The corresponding nonzero X are called eigenvectors of the matrix A.

Example 63 :    
1 1 1
If A is the matrix A = , then the vector X = is an eigenvector
−3 5 1
for A because AX = 2X. The corresponding eigenvalue is λ = 2.
Remark 12 :
Note that if AX = λ and c is any real number, then A(cX) = cAX = c(λX) =
λ(cX). Then, if X is an eigenvector of A, then so is cX for any nonzero number
c.
The eigenvalue equation can be rewritten as (A−λI)X = 0. The eigenvalues
of A are the values of λ for which the above equation has nontrivial solutions.
There are nontrivial solutions if and only if det(A − λI) = 0.

lxxix
lxxx

Theorem 1.2

If A ∈ Mn (R) and λ ∈ R. λ is an eigenvalue the matrix A if and only if

det(A − λI) = 0.

Definition 1.3

If A ∈ Mn (R), the polynomial

qA (λ) = |A − λI|

is called the characteristic polynomial of the matrix A and the equation

qA (λ) = 0 is called the characteristic equation of A. The eigenvalues of
A are the roots of its characteristic polynomial.

Example 64 :  
1 3
Find all of the eigenvalues and eigenvectors of A = .
2 2
1−λ 3
Compute the characteristic polynomial qA (λ) = = (λ+1)(λ−4).
2 2−λ
 
3
The roots of qA (λ) are −1 and 4. X1 = is an eigenvector for A with
  −2
1
respect to the eigenvalue −1 and X2 = is an eigenvector for A with respect
1
to the eigenvalue 4.
Example 65 :
Find theeigenvalues of 
the following
 matrix
  
1 −1 0 5 4 2 −1 4 −2
A = −1 2 −1, A = 4 5 2, A = −3 4 0 .
0 −1 1 2 2 2 −3 1 3

Definition 1.4

If A is a matrix with characteristic polynomial qA (λ), the multiplicity

of a root λ of qA is called the algebraic multiplicity of the eigenvalue λ.

Example 66 :
 lxxxi

 
1 0 0
Let A = −1 1 −1. The characteristic function of the matrix A is
1 0 2
λ−1 0 0
qA (λ) = 1 λ−1 1 = (λ − 1)2 (2 − λ).
−1 0 λ−2
The eigenvalue λ = 1 has algebraic multiplicity 2, while λ = 2 has algebraic
multiplicity 1.

Definition 1.5

Let A ∈ Mn (R) and λ an eigenvalue of the matrix A. The set

Eλ = {X ∈ Rn ; AX = λX}
is called the eigenspace of A associated to the eigenvalue λ.

Remark 13 :
If λ is an eigenvalue of the matrix A ∈ Mn (R), then Eλ = {X ∈ Rn ; AX = λX}
is vector sub-space of Rn . Its dimension is called the the geometric multiplicity
of λ.
The geometric multiplicity of λ is the number of linearly independent eigen-
vectors corresponding to λ.

Theorem 1.6

The geometric multiplicity of an eigenvalue is less than or equal to its

algebraic multiplicity.

Definition 1.7

A matrix that has an eigenvalue whose geometric multiplicity is less

than its algebraic multiplicity is called defective.
lxxxii

Theorem 1.8

If A ∈ Mn (R) and X1 , . . . , Xm are eigenvectors for different eigenvalues

λ1 , . . . , λm , then X1 , . . . , Xm are linearly independent.

Proof .
The proof is by induction. The result is true for m = 1.
Assume the result true for m and let X1 , . . . , Xm+1 be eigenvectors for different
eigenvalues λ1 , . . . , λm+1 .
If a1 X1 + . . . am Xm + am+1 Xm+1 = 0, then
a1 λ1 X1 + . . . am λm Xm + am+1 λm+1 Xm+1 = 0. Also we have
a1 λm+1 X1 + . . . am λm+1 Xm + am+1 λm+1 Xm+1 = 0. Then
a1 (λ1 − λm+1 )X1 + . . . + am (λm − λm+1 )Xm = 0. Since
(λj − λm+1 ) ̸= 0 for all j = 1, . . . m, then a1 = . . . = am = 0 and so am+1 = 0.

2 Diagonalization
Definition 2.1

A matrix A ∈ Mn (R) is called diagonalizable if there exists an invertible

matrix P ∈ Mn (R) such that the matrix P −1 AP is diagonal.

Remark 14 :
If X1 , . . . , Xn are the columns of the matrix P , then the columns of the matrix
AP are: AX1 , . . . , AXn .
Moreover if  
λ1 0 . . . . . . 0
. 
 0 λ2 0 . . . .. 

 
D =  ... .. .. .. 

 0 . . . 

. . .
 .. .. . . . . . 0 


0 . . . . . . 0 λn
then the columns of the matrix P D are: λ1 X1 , . . . , λn Xn .
Then P −1 AP = D ⇐⇒ P D = AP and the columns of the matrix P form a
basis of Rn of eigenvectors of the matrix A.
 lxxxiii

Theorem 2.2

A matrix A ∈ Mn (R) is diagonalizable if and only if it has n eigenvectors

linearly independent. These vectors form a basis of the vector space Rn .

Example 67 :
Prove that the following matrices are diagonalizable and find an invertible
−1 15
matrix P∈ Mn (R) suchthat the
matrix P AP is  diagonal and
 find A .
1 −1 0 5 4 2 −1 4 −2
A = −1 2 −1, A = 4 5 2, A = −3 4 0 .
0 −1 1 2 2 2 −3 1 3

Theorem 2.3

If A ∈ Mn (R) and the characteristic function

qA (λ) = C(λ − λ1 )m1 . . . (λ − λp )mp

then A is diagonalizable if and only if the algebraic and geometric mul-

tiplicities are equal.

Remark 15 :
For example, if a matrix A ∈ Mn (R) and has n different eigenvalues, then A is
diagonalizable.
Example 68 :  
5 4
Consider the matrix A = . The characteristic polynomial of the
−4 −3
matrix A is
5−λ 4
qA (λ) = = (1 − λ)2 .
−4 −3 − λ
Then the matrix is not diagonalizable.
Example 69 :  
−10 −6
Consider the matrix A = . The characteristic polynomial of the
18 11
matrix A is

−10 − λ −6
qA (λ) = = (λ − 2)(1 + λ).
18 11 − λ
Then the matrix is diagonalizable.
E−1 = ⟨(−2, 3)⟩ and E2 = ⟨(1, −2)⟩.
lxxxiv

 
−1 0
The diagonal matrix is D =
 0 2
−2 1
and the matrix P is P = .
3 −2
Example 70 :  
5 0 4
Consider the matrix A =  2 1 5 . The characteristic polynomial of
−4 0 −3
the matrix A is
5−λ 0 4
qA (λ) = 2 1−λ 5 = (1 − λ)3 .
−4 0 −3 − λ

Then the matrix is not diagonalizable.

Example 71 :  
1 0 0
Consider the matrix A = −1 1 −1. The characteristic polynomial of
1 0 2
the matrix A is
1−λ 0 0
qA (λ) = −1 1−λ −1 = (1 − λ)2 (2 − λ).
1 0 2−λ

E1 = ⟨(0, 1, 0), (1, 0, −1)⟩ and E2 = ⟨(0, 1, −1)⟩.

Then the matrix is diagonalizable.
 
1 0 0
the diagonal matrix is D = 0 1 0
  0 0 2
0 1 0
and P = 1 0 1 .
0 −1 −1
Example 72 :  
5 −3 0 9
0 3 1 −2
Consider the matrix A = 0
.
0 2 0
0 0 0 2
The characteristic polynomial of the matrix A is

5−λ −3 0 9
0 3−λ 1 −2
qA (λ) = = (5 − λ)(3 − λ)(2 − λ)2 .
0 0 2−λ 0
0 0 0 2−λ
 lxxxv

The matrix is diagonalizable if and only if the dimension of the vector space
E2 is 2.
E2 = ⟨(1, 1, −1, 0), (−1, 2, 0, 1)⟩.
Then the matrix A is diagonalizable.
E5 = ⟨(1, 0, 0, 0)⟩ and E3 = ⟨(3,
 2, 0, 0)⟩. 
5 0 0 0
0 3 0 0
The diagonal matrix is D =  0 0 2 0


0 0 0 2
 
1 3 1 −1
0 2 1 2
and P = 0 0 −1 0 .


0 0 0 1
Example 73 :  
2 2 −1
Consider the matrix A =  1 3 −1.
−1 −2 2
The characteristic polynomial of the matrix A is

2−λ 2 −1
qA (λ) = 1 3−λ −1 = −(λ − 1)2 (λ − 5).
−1 −2 2−λ
E1 = ⟨(1, 0, 1), (−2, 1, 0)⟩, E5 = ⟨(1, 1, −1)⟩.
Then the matrix A is diagonalizable.
   
1 0 0 1 −2 1
The diagonal matrix is D = 0 1 0 and P = 0 1 1
0 0 5 1 0 −1
Example 74 :  
7 4 16
Consider the matrix A =  2 5 8 .
−2 −2 −5
The characteristic polynomial of the matrix A is

7−λ 4 16
qA (λ) = 2 5−λ 8 = −(λ − 3)2 (λ − 1).
−2 −2 −5 − λ
E3 = ⟨(1, −1, 0), (4, 0, −1)⟩, E1 = ⟨(2, 1, −1)⟩.
Then the matrix A is diagonalizable.
   
1 0 0 2 1 4
The diagonal matrix is D = 0 3 0 and the matrix P is P =  1 −1 0
0 0 3 −1 0 −1

Example 75 :
lxxxvi

−1 0 21
 
2
0 1 0 12 
Consider the matrix A = 
−1
. The characteristic polynomial
1 1 −1
1 −1 1 3
of the matrix A is
1
2−λ −1 0 2
1
0 1−λ 0
qA (λ) = 2 = (1 − λ)(2 − λ)3 .
−1 1 1−λ −1
1 −1 1 3−λ

The matrix is diagonalizable if and only if the dimension the vector space E2
is 3.
E2 = ⟨(−1, 1, 0, 2), (−1, 0, 1, 0)⟩. Then the matrix is not diagonalizable.