0% found this document useful (0 votes)
36 views24 pages

Chapter One and Two

TL ça veut dire Théorie de language

Uploaded by

marmiyasmine143
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
36 views24 pages

Chapter One and Two

TL ça veut dire Théorie de language

Uploaded by

marmiyasmine143
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 24

Determinants

1
1.1 Definition and properties
Determinant is a scalar value that can be calculated from a given matrix.
The matrix has to be square (same number of rows and columns).
Definition 1.1.1. Let A ∈ Mn (K), that is, the matrix A be square. Then
we define the matrix Aij as the matrix extracted from the matrix A, which
we obtain by removing the row i and the column j of the matrix A.
Example 1.1.1. Let
1 2 7
 

A= 0 4 −5 
.

−2 −4 2
Extract the following matrices A11 , A23 , A31

4 −5 1 2 2 7
! ! !
A11 = , A23 = , A31 = .
−4 2 −2 −4 4 −5

Remark 1.1.1. It is clear that if the matrix A is of degree n, then the matrix
Aij is of degree n − 1, then

A ∈ Mn (k) ⇐⇒ Aij ∈ Mn−1 (k).

1.1.1 Definition of Determinant


The definition of determinant is the focus of this subsection. Determinant’s
definition is extremely abstract, and calculating determinants will take a
while.

1
1.1. Definition and properties 1

Definition 1.1.2. (determinant) The determinant can be defined as a linear


application noted by f defined from the vector space Mn (K) into K, satisfying:

1. If the matrix is single-element, that is A = (a), so we have f (A) = a.


!
a11 a12
2. If the matrix is second degree, i.e, A = , then f (A) =
a21 a22
a11 f (A11 ) − a21 f (A21 ).

3. In the general case, that is, if the matrix is of degree n, then we have
n
f (A) = (−1)i+j aij f (Aij ).
X

i,j=1

Remark 1.1.2. We denote the determinant of the matrix A by det(A) or


|A|.

Example 1.1.2. Let the square matrix A defined by

1 −1
!
A= ,
2 5

by definition, we have

det(A) =| A |= 1 × 5 − 2 × (−1) = 7.

Example 1.1.3. Let the square matrix A defined by

1 2 1
 

A =  3 0 6 ,
 

−4 7 −2

we fix one of the lines or one of the columns (i.e., the choice is qualitative
and the result is the same), let it be the first line:
3
det(A) = (−1)1+j a1j det(A1j ) = a11 det(A11 ) − a12 det(A12 ) + a13 det(A13 ).
X

j=1

Note that det(A11 ) is the determinant of the matrix A11 , which we obtained
by limiting the first line (the anchored line) and the first column. det(A12 ) is
the determinant of the matrix A12 , which we obtained by limiting the first line
(the anchored line) and the second column, and det(A13 ) is the determinant

2
1.1. Definition and properties 1

of the matrix A13 , which we obtained by limiting the first line (the fixed line)
and the third column. From there, we can easily find

0 6 3 6 3 0
det(A) = 1 −2 +
7 −2 −4 −2 −4 7
= (0 × (−2) − 7 × 6) − 2(3 × (−2) − (−4) × 6) + (3 × 7 − (−4) × 0)
= −42 − 36 + 21 = −57.

Remark 1.1.3. Choosing a line or column is qualitative, and for this reason
we must choose the line or column that makes calculations easier for us, such
as choosing the line or column that contains the largest number of zeros.

Remark 1.1.4. There is a short way to calculate a determinant of the third


degree using the Sarrus rule. First, we add the first and second columns, then
we multiply the elements that are in the same direction as the arrow, then we
attach the red ones with a sign +, but for those that have a blue arrow, we
attach a sign to them, and then we add them all and find the value. Specified
 
a11 a12 a13
A =  21 a22 a23  ,
a
 

a31 a32 a33

so we find
   
a11 a12 a13 a11 a12 a11 a12 a13 a11 a12
 a21 a22 a23 a21 a22   21 a22 a23 a21 a22  ,
a
   

a31 a32 a33 a31 a32 a31 a32 a33 a31 a32

then
det(A) = a11 × a22 × a33 + a12 × a23 × a31 + a13 × a21 × a32
− a31 × a22 × a13 − a32 × a23 × a11 − a33 × a21 × a12 .

2 1 0
 

 1 −1 3 .
Example 1.1.4. Let’s calculate the determinant of the matrix A = 

3 2 1
By the Sarrus rule:

det(A) = 2 × (−1) × 1 + 1 × 3 × 3 + 0 × 1 × 2
− 3 × (−1) × 0 − 2 × 3 × 2 − 1 × 1 × 1 = −6.

Attention! This method does not apply to matrices larger


than 3.

3
1.1. Definition and properties 1

1.1.2 Geometric Interpretation of Determinant


We will see that in dimension 2, determinants correspond to areas,
! and in
a
dimension 3, to volumes. Let’s give us two vectors: v1 = and v2 =
c
!
b
of the plan R2 . These two vectors v1 , v2 determine a parallelogram.
d

Proposition 1.1.1. The area of the parallelogram is given by the absolute


value of the determinant:
!
a b
A = |det (v1 , v2 )| = det .
c d
Similarly, three vectors of the space R3 :
     
a11 a12 a13
v1 =  a21  v2 =  a22  v3 =  a23 
     

a31 a32 a33


define a parallelepiped.

From these three vectors, we define, by joining the columns, a matrix and a
determinant:  
a11 a12 a13
det (v1 , v2 , v3 ) = det 
 a21 a22 a23  .

a31 a32 a33

4
1.1. Definition and properties 1

Proposition 1.1.2. The volume of the parallelepiped is given by the absolute


value of the determinant:

V = |det (v1 , v2 , v3 )| .

1.1.3 Properties of Determinants


In this section, we study some properties of determinants of square matrices.

• The determinant of the null matrix 0n is zero.

• The determinant of the identity matrix In is one.

We will see the behavior of the determinant against the elementary operations
on the columns.

Theorem 1.1.1. We multiply the determinant of the matrix by −1 whenever


we switch between two columns or lines.

Example 1.1.5. Let

2 6 1 6 2 1 2 6 1
     

A =  1 1 5  , B =  1 1 5  , C =  0 0 −1 
    
,
0 0 −1 0 0 −1 1 1 5

so we have
det(A) = 4.
We note that the matrix B is a matrix resulting from the matrix A by replac-
ing the first column with the second column, and therefore

det(B) = det(A) × −1 = −4.

As for the matrix C, it is the result of the matrix A by replacing the second
line with the third line, and from it,

det(C) = det(A) × −1 = −4.

Attention! Switching two columns changes the determi-


nant sign.

Theorem 1.1.2. The value of the determinant does not change if a linear
blend is added to any column (line) for the rest of the columns (lines).

5
1.1. Definition and properties 1

Example 1.1.6. Let the matrix A define by

2 5 1
 

A =  −1 2 5 ,

1 −1 −1

calculate the determinant of the matrix A and then derive the value of deter-
minant of the following matrix

8 5 1
 

B= 6 2 5 ,

−1 −1 −1

we have det(A) = 25. As for the matrix B, it is clear that it is the result
of the matrix A by adding the second and third columns to the first column.
Therefore, according to the previous property, det(B) = 25.
Proposition 1.1.3. Let A ∈ Mn (K) be a matrix with the columns C1 , C2 , . . . , Cn .
We note A′ is the matrix obtained by one of the elementary operations on the
columns, which are:
1. Ci ← λCi avec λ ̸= 0 : A′ is obtained by multiplying a column of A by
a non-zero scalar. Then det A′ = λ det A.

2. Ci ← Ci + λCj avec λ ∈ K (et j ̸= i ) : A′ is obtained by adding to a


column of A a multiple of another column of A. Then det A′ = det A.

3. Ci ↔ Cj : A′ is obtained by exchanging two separate columns of A.


Then det A′ = − det A.
Theorem 1.1.3. If the columns or lines are linearly related, the determinant
of the matrix is null.
Example 1.1.7. Let the matrix A

1 2 5
 

A =  3 6 15 
 

−1 3 5

We note
(1, 3, −1) = −2(2, 6, 3) + (5, 15, 5),
that the columns are linearly related, then det(A) = 0.
Theorem 1.1.4. If all elements of one row or (one column) of square matrix
A are zero, then det(A) = 0.

6
1.1. Definition and properties 1

Example 1.1.8. Let


−6 8 3
 

A =  2 1 −9  ,
 

0 0 0
we note all elements of row number 3 of matrix A equal zero, then we
have by theorem (1.1.4) det(A) = 0.

Theorem 1.1.5. If there exist two rows or two columns of a square matrix
A are equal, then det(A) = 0.

Example 1.1.9. Let


−6 8 3
 

A =  2 1 −9 

,
2 1 −9
by theorem (1.1.5), we get det(A) = 0, because the elements of row number
3 equal the elements of row number 2.

Theorem 1.1.6. If one row or one column in matrix A is a multiple of


another row or column, then det(A) = 0.

Example 1.1.10. Let


4 6 8
 

A =  2 1 −9  ,
 

2 3 4
by theorem (1.1.6), we get det(A) = 0, because row one is a double of row
three.

Theorem 1.1.7. If A and B are two square matrices, then det(A.B) =


det(A). det(B).
In general,

det(A1 .A2 .....An ) = det(A1 ). det(A2 )....... det(An ), n ∈ N.

Example 1.1.11. Let

1 2 2 3 10 5
! ! !
A= , B= , A.B = ,
2 3 4 1 16 9

det(A) = −1, det(B) = −10, by theorem (1.1.7), we have

det(A.B) = det(A). det(B) = 10.

7
1.2. Finding the inverse matrix 1

Attention! If A, B, and C are three positive semidefinite


matrices of equal size, then the following equation holds
along with the corollary

det(A + B) ≥ det(A) + det(B)

,
A, B, C ≥ 0, det(A + B + C) + det C ≥ det(A + B) +
det(B + C).
Theorem 1.1.8. If A is a square matrix and a triangular matrix, then the
determinant of the matrix A is equal to the product of the elements of the
main diagonal.
Example 1.1.12. Let
1 2 3 5 0 0
   

A =  0 3 2 ,b =  1 2 0 ,
   

0 0 4 1 2 4
by theorem (1.1.8), we get det(A) = 1 × 3 × 4 = 12, det(B) = 5 × 2 × 4 =
40, because the matrices A and B are triangular matrices.
Theorem 1.1.9. if A is a square matrix, then det(A) = det(At ).
Example 1.1.13. Let
1 4 1 2
! !
A= , A =t
2 3 4 3
by theorem (1.1.9), we get, det(A) = −5 = det(At ).
Theorem 1.1.10. If A is n × n invertible matrix, then det(A−1 ) = 1
det(A)
.

1.2 Finding the inverse matrix


Let the matrix A be a square matrix of degree n, that is, A ∈ Mn (K) such
that its determinant is non-zero i.e (det(A) ̸= 0), then we know the inverse of
a matrix, which we denote by the symbol A−1 , by the following relationship
1
A−1 = (B)t ,
det(A)
Where the matrix B = (bij ) is of the same order as the matrix A and the
elements of this matrix are defined by the following:
bij = (−1)i+j det(Aij ).

8
1.2. Finding the inverse matrix 1

Example 1.2.1. Let it be a quadratic matrix of degree two


1 2
!
A= ,
−1 1
Show that the matrix A is invertible and find its inverse?
Solution 1.2.1. In order for the matrix A to be invertible, it is sufficient
for its determinant to be non-zero i.e
det(A) ̸= 0
so, we calculate the determinant of the matrix A
1 2
det(A) = = 1 × 1 − (−1) × 2 = 3 ̸= 0.
−1 1
Therefore, the matrix A is invertible and its inverse is given by the following
relationship
1
A−1 = (B)t ,
det(A)
whereas
bij = (−1)i+j det(Aij )
That is, it is enough to calculate the values bij :
b11 = (−1)1+1 × 1 = 1,
b12 = (−1)1+2 × (−1) = 1,
b21 = (−1)2+1 × 2 = −2,
b22 = (−1)2+2 × 1 = 1,
we obtain
1 1 1 −2
! !
B= =⇒ (B) =t
,
−2 1 1 1
so, we have
1 1 −2
!
A−1
= .
3 1 1
Verify that the result is correct
1 2 1 1 −2 1 0
! ! !
AA −1
= = = I2 .
−1 1 3 1 1 0 1
Example 1.2.2. Same question for the following matrix
1 −1 0
 

A= 1 2 1 .

2 −2 −1

9
Solving Linear Systems
2
2.1 Introduction
Linear algebra is an essential tool for all branches of applied mathematics, in
particular when it comes to modeling and then numerically solving problems
from various fields: physical or mechanical sciences, life sciences, chemistry ,
economics, engineering sciences,...
Linear systems are used in many application contexts because they form
the computational basis of linear algebra. They also make it possible to treat
a good part of the theory of linear algebra in finite dimensions. This is why
this course begins with a study of linear equations and their solution. This
chapter has an essentially practical goal: to solve linear systems.

2.2 Definitions
Linear equations:

Each equation of the form:

a1 x1 + ...an xn = b,

is called a linear equation, where xi for i = 1..., n, are unknowns and ai


for i = 1..., n, are transaction,
as for the value b, it is the second side.

Example 2.2.1. Let the following equation be defined as follows

2x − y + 4z − 3t = −3,

10
2.2. Definitions 2

It is a linear equation with four unknowns. We also note that the vector
(1, 0, 2, 1) from the vector space R4 is one of the solutions to the previous
equation.

Example 2.2.2. Let the following equation define as

2x2 + 2xy = 2,

It is not a linear equation.

Linear System with Two Variables

Real-world applications are often modeled using more than one variable
and more than one equation. A system of equations consists of a set of
two or more equations with the same variables. In this part, we will study
linear systems, which contain two linear equations each with two variables.
For example,
2x − 3y = 0
(

−4x − 2y = 8
A solution to a linear system, or simultaneous solution, is an ordered
pair (x, y) that solves both of the equations. In this case, (3, 2) is the only
solution. To check that an ordered pair is a solution, substitute the corre-
sponding x and y values into each equation and then simplify to see if you
obtain a true statement for both equations.

System of linear equations

A linear equations m in which n is unknown is called a linear system and


is written in the following form:

a11 x1 + a12 x2 + ... + a1n xn = b1




a21 x1 + a22 x2 + ... + a2n xn = b2



,

 ................................
am1 x1 + am2 x2 + ... + amn xn = bm

Example 2.2.3. Let a linear system

2x + y − z = 1



x − 3y + t = 0
y + z − 7t = 0

It is a linear system with three equations and four unknowns.

11
2.2. Definitions 2

2.2.1 A system of homogeneous linear equations:


If the second side is zero, then the system is called a system of homogeneous
linear equations, meaning that the system is of the following form
a11 x1 + a12 x2 + ... + a1n xn = 0


a21 x1 + a22 x2 + ... + a2n xn = 0



,

 ................................
am1 x1 + am2 x2 + ... + amn xn = 0

Example 2.2.4. Let a linear system


2x − 3y = 0



−5x + y = 0
x−y =0

This is a linear system with three equations and two unknowns, and it is a
homogeneous system.
Remark 2.2.1. All homogeneous linear system have at least one solution,
which is the zero solution

2.2.2 Solving linear equations


To find out if there are one or more solutions to a linear system.

Resolution by substitution
a first method is substitution. For example for the system:
3x + 2y = 1
(

2x − 7y = −2

We rewrite the first line 3x + 2y = 1 Under the form y = 21 − 32 x. And


we replace (we substitute) the y of the second equation, with the expression
y = 12 − 23 x. We obtain an equivalent system:

y = 21 − 32 x
(

2x − 7( 12 − 23 x) = −2
The second equation is now an expression that contains only x, and we
can solve it
y = 21 − 32 x y = 25
( (
8

2x − 7( 12 − 23 x) = −2 x = 25
3

System therefore admits a unique solution ( 25


3 8
, 25 ).

12
2.2. Definitions 2

Graphing method for solving linear systems


To summarize, linear systems described in this section consist of two linear
equations each with two variables. A solution is an ordered pair that corre-
sponds to a point where the two lines intersect in the rectangular coordinate
plane. Therefore, one way to solve linear systems is by graphing both lines
on the same set of axes and determining the point where they cross. This
describes the graphing method for solving linear systems.
When graphing the lines, take care to choose a good scale and use a
straightedge to draw the line through the points; accuracy is very important
here.
Example 2.2.5. Solve by graphing:

x − y = −4
(
,
2x + y = 1

Write the equivalent system and graph the lines on the same set of axes form

x − y = −4 y =x+4
( (
⇐⇒
2x + y = 1 y = 1 − 2x

Use the graph to estimate the point where the lines intersect and check to
see if it solves the original system. In the above graph, the point of intersec-
tion appears to be (−1, 3).

13
2.2. Definitions 2

2.2.3 Matrix writing for linear system


Every linear system can be written in matrix form i.e
a11 x1 + a12 x2 + ... + a1n xn = b1


a21 x1 + a22 x2 + ... + a2n xn = b2



⇐⇒ AX = B

 ................................
am1 x1 + am2 x2 + ... + amn xn = bm

Where
   
x1 b1
x2 b2
   
   
A = (aij )1≤i≤n,1≤j≤m , X = =
   

 . ,B


 . 


 . 


 . 

xn bm
We note that the number of lines of the matrix A is equal to the number
of equations, while the number of columns of the matrix A is equal to the
number of unknowns in the linear system.
Example 2.2.6. Let be the following linear system
2x + y − z = 1



x − 3y + t = 0
y + z − 7t = 0

This linear system can be written in the following matrix form, i.e
x
 
2 1 −1 0 1
   
 y 
 1 −3 0
AX + B ⇔  1  = 0 
  
 z
  
0 1 1 −7 0

t
Remark 2.2.2. We note that the first column is the coefficients of the first
unknown, while the coefficients of the second column are the coefficients of
the second unknown, the third column is the coefficients of the third unknown,
and the last column is the coefficients of the last unknown.

2.2.4 Applying Cramer’s method to solve a linear sys-


tem
The following theorem, called Cramer’s rule, gives an explicit formula for
the solution of certain systems of linear equations with as many equations as
unknown. We say that a system is a Cramer system if satisfies this following
:

14
2.2. Definitions 2

• The number of equations should be equal to the number of unknowns,


meaning that the matrix accompanying the system should be square,
meaning the number of columns should be equal to the number of lines.

• The determinant of the matrix accompanying the linear system must


be non-zero.

We consider the system of linear equations with n equations and n un-


known as follows:
+ a12 x2 + · · · + a1n xn = b1

 a11 x1


+ a22 x2 + · · · + a2n xn = b2

 a21 x1



 ···

an1 x1 + an2 x2 + · · · + ann xn = bn

This system can also be written in matrix form AX = B where

a11 a12 · · ·
     
a1n x1 b1

a21 a22 · · · a2n  
x2  
b2 
A= .. .. .. ∈ Mn (K), X= .. et B= .. .
     
    
. . . . .
     
     
an1 an2 · · · ann xn bn

We define the matrix Aj ∈ Mn (K) by


 
a11 . . . a1,j−1 b1 a1,j+1 . . . a1n

a21 . . . a2,j−1 b2 a2,j+1 . . . a2n 
Aj = .. .. .. .. .. 
 

. . . . . 
 

an1 . . . an,j−1 bn an,j+1 . . . ann

In other words, Aj is the matrix obtained by replacing the j column of A


with the second member B. The Cramer rule will allow us to calculate the
solution of the system in the case that det A ̸= 0 based on the determinants
of the matrices A and Aj .

Theorem 2.2.1 (Cramer rule). Let

AX = B

a system of n unknown equations. Suppose det A ̸= 0. So the only solution


(x1 , x2 , . . . , xn ) from the system is given by:

det A1 det A2 det An


x1 = x2 = ... xn = .
det A det A det A

15
2.2. Definitions 2

Example 2.2.7. We will solve the following system :

x1 + 2x3 = 6



−3x1 + 4x2 + 6x3 = 30
−x1 − 2x2 + 3x3 = 8

We have
1 0 2 6
   

A =  −3

4 6 
 B =  30 
 

−1 −2 3 8
6 0 2 1 6 2 1 0 6
     

A1 =  30 4 6  , A2 =  −3 30 6  , A3 =  −3 4 30 
     

8 −2 3 −1 8 3 −1 −2 8
and
det A = 44, det A1 = −40, det A2 = 72, det A3 = 152.
Then, the solution is
det A1 40 10 det A2 72 18 det A3 152 38
x1 = =− =− x2 = = = x3 = = = .
det A 44 11 det A 44 11 det A 44 11
Remark 2.2.3. The Cramer method is not the most effective method for
solving a system, but is useful if the system contains parameters.

2.2.5 Gauss Elimination with Pivoting for Linear Sys-


tems
First case: Squared matrices
Let AX = b such that A is a matrix (n × n).

The principle of the method


We will give the algorithm for the construction of an upper triangular matrix
M A without having to know the matrix M at any moment, i.e., we will
transform the matrix A into an upper triangular matrix.
We note : A′ = M A.

• We assume that the matrix A is invertible, i.e., det(A) ̸= 0.

For this, we construct the following transformation:

[A | b]−→
transformation [A | b ] ,
′ ′

16
2.2. Definitions 2

equivalently,

a11 a12 · · · a1n | b1 a′11 a′12 · · · a′1n | b′1


   

a21 a22 · · · a2n | b2  
0 a′22 · · · a′2n | b′2 
.. .. .. ..  .. .. . . .. .. 
   

..  → 

 . . . . . 

 . . . . . 

an1 an2 · · · ann | bn 0 0 · · · a′nn | b′n

We pose A = A(1) and b = b(1) .


Step 1:
(1)
(1) (2) (1) (2) (1) ai1 (1)
∗ If a11 ̸= 0, L1 = L1 , and for i = 2, n, Li = Li − (1) · L1 .
a11
Therefore, we get:
(1) (1) (1) (1) (2) (2) (2) (2)
   
a11 a12 · · · a1n | b1 a11 a12 · · · a1n | b1
(1) (1) (1) (1) (2) (2) (2)
0
   
 a21 a22 · · · a2n | b2   a22 · · · a2n | b2 
.. .. .. .. .. → .. .. .. .. ..
   
. . . . . . . . . .
   
   
   
(1) (1) (2)
an1 an2 · · · ann | b(1)
(1)
n 0 an2 · · · ann | b(2)
(2)
n
that is,
h i h i
A(1) | b(1) → A(2) | b(2)

(2) (1)

 a1j = a1j , j = 1, n.
where (2) (1) ai1
(1)
(1)

 aij = aij − (1) · a1j , i = 2, n and j = 1, n.
a11

(2) (1)
 b1
 = b1 ,
and (2) (1)
(1)
ai1 (1)
 bi
 = bi − (1) b1 , i = 2, n.
a11

(1)
∗ If a11 = 0, we are looking for a row L(1)
p with 2 ≤ p ≤ n such that
(1) (1)
ap1 ̸= 0, then, we permute the rows L1 and L(1) p , and we apply
(1)
the transformations similar to the case a11 ̸= 0, which is studied
above. In this step, all coefficients under the diagonal of the first
column are equal 0.

Step k:
(k)
∗ If akk ̸= 0, we make the following changes:
(k+1) (k)
L(1) = L1

17
2.2. Definitions 2

(k+1) (k)
L(2) = L2
..
.
(k+1) (k)
Lk = Lk
(k)
(k+1) (k) aik (k)
Li = Li − (k) · Lk i = k + 1, n
akk

(k+1) (k)
aij
 = aij i = 1, k, j = 1, n
with  (k+1) (k)
(k)
aik (k)
aij = aij − (k) akj i = k + 1, n, j = 1, n
akk

(k+1) (k)
 bi
 = bi , i = 1, k
and  (k+1) (k)
(k)
aik (k)
 bi = bi − (k) · bk , i = k + 1, n
akk

(k) (k)
∗ If akk = 0, we permute the rows Lk and Lp(k) where L(k) p is a
(k)
line with index p with k + 1 ≤ p ≤ n, such that apk ̸= 0, and
we apply transformations analogous to those corresponding to the
(k)
case akk ̸= 0 which is studied above.

We calculate the determinant as follows:


n
det{A} = (−1)p det{A′ } = (−1)p a′kk ,
Y

k=1

where p is the number of permutations.


We were assumed that det{A} ̸= 0, hence the system A′ X = b′ has a
unique solution X.

Solving the system A′ X = b′

We solve the system A′ X = b′ ( whose solution is exactly the solution


of the system AX = b ).

18
2.2. Definitions 2

(n) (n) (n) (n)


 
a11 a12 · · · a1n | b1
(n) (n) (n)
0
 
h i  a22 · · · a2n | b2 
We pose [A′ | b′ ] = A(n) | b(n) = .. .. .. .. .. =
 
.
 

 . . . . 

0 0 · · · a(n)
nn | b (n)
n

a′11 a′12 · · · a′1n | b′1


 

0 a′22 · · · a′2n | b′2 
.. .. . . .. .. 
 


 . . . . . 

0 0 · · · a′nn | b′n
Then, AX = b ⇔ A(1) X = b(1) ⇔ A(2) X = b(2) ⇔ . . . ⇔ A(n) X = b(n) ⇔
A ′ X = b′

x1 = [b′1 − a′12 x2 − . . . − a′1n xn ]



1
a′11


..


h .



hence, i




x n−1 = ′
1
an−1n−1
b ′
n−1 − a ′
n−1n x n
 xn = ′1 · b′


a nn
n

We determine xn then xn−1 , etc., until obtaining x1 ( we call this process


resolution by backtracking).
(k)
Remark 2.2.4. The coefficients akk are called pivots.
Example 2.2.8. Solve the following system by the Gauss method
 x1 + x2 + 3x4 = 4 1 1 0 3 4

x1
    

 2x + x − x + x = 1  2 1 −1 1  1
  x2 
    
=
1 2 3 4

  
3x − x − x + 2x = −3  3 −1 −1 2   x3  −3
  

 1 2 3 4  
−x1 + 2x2 + 3x3 − x4 = 4 −1 2 3 −1 4

x4

 →L
1 1 0 3 4 1

 2 1 −1 1 1  → L2 

 3 −1 −1 2 −3  → L3
 

−1 2 3 −1 4 → L4
We have a11 = 1 ̸= 0, then,
−a21 −2
ℓ21 = = = −2
a11 1
−a31 −3
ℓ31 = = = −3
a11 1
−a41 1
ℓ41 = = =1
a11 1

19
2.2. Definitions 2

   
L′2 = ℓ21 · L1 + L2 ⇔ L′2 = −2 1 1 0 3 4 + 2 1 −1 1 | 1
 
= 0 −1 −1 −5 | −7
   
L′3 = ℓ31 ·L1 +L3 ⇔ L′3 = −3 1 1 0 3 4 + 3 −1 −1 2 −3
 
= 0 −4 −1 −7 | −15
   
L′4 = ℓ41 · L1 + L4 ⇔ L′4 = 1 1 1 0 3 4 + −1 2 3 −1 | 4
 
= 0 3 3 2 |8


1 1 0 3 4 → L′1 = L1

 0 −1 −1 −5 −7  → L′
  2
0 −4 −1 7 −15  → L′
 

3
0 3 3 2 8 → L′4
We have a22 = −1 ̸= 0, then,
(2)
−a32 4
ℓ32 = (2)
= = −4
a22 −1
(2)
−a42 −3
ℓ42 = (2)
= =3
a22 −1

   
L′′3 = ℓ32 ·L′2 +L′3 ⇔ L′′3 = −4 0 −1 −1 −5 −7 + 0 −4 −1 −7 | −15
 
= 0 0 3 13 | 13

   
L′′4 = ℓ42 ·L′2 +L′4 ⇔ L′′4 = 3 0 −1 −1 −7 −15 + 0 3 3 2 8
 
= 0 0 0 −13 − 13

20
2.2. Definitions 2

1 1 0 3 4 → L′1 = L1
 

 0 −1 −1 −5 −7 
 → L′′2 = L′2
0 0 3 13 13 → L′′3
 
 
0 0 0 −13 | −13 → L′′4
1 1 0 3 4 x1 + x2 + 3x4 = 4

x1
    

0 −1 −1 −5 −x2 − x3 − 5x4 = −7

x2 −7
     

= ⇐⇒
    
0 0 3 13 x3 13 3x3 + 13x4 = 13
    
     

0 0 0 −13 −13x4 = −13

x4 −13

(4) =⇒ x4 = −13
−13
=1
(3) =⇒ 3x3 = 13 − 13x4 = 0 =⇒ x3 = 0
(2) =⇒ x2 = −x3 − 5x4 + 7 =⇒ x2 = 2
(1) =⇒ x1 = 4 − x2 − 3x4 = −1
The solution is :
−1
 
 2 
X=
 
k0

 
1
4
det{A} = (−1)p det{A′ } = (−1)P a′kk = (−1)0 (1)(−1)(3)(−13) = 39
Y

k=1

• Now, we assume that A is not invertible, i.e., det{A} = 0.


(k)
In this case, we find a coefficient akk equal 0, so we can not solve the
system A′ X = b′ , we have two cases: the system has no solution or has
an infinity of solutions, but the Gauss elimination is still valid.

Second case: Matrices n × m, (n ̸= m)


The Gauss elimination method is a useful way to solve any system without
any condition whether the matrix is square or non square (i.e The equations
are equal to the unknowns or no). Also, it is a useful way to calculate the rank
of a matrix. In this case, the matrix is not square, so we can not calculate
the determinant and we need to define the rank of a matrix, in order to know
whether the system has a solution or not

Definition 2.2.1. Let A ∈ Mn×m (R), we have the rank defined as follows:

rank(A) ≤ min(n, m)

Remark 2.2.5. The previous definition is valid if the matrix is square or


not.

21
2.2. Definitions 2

There are several ways to know the rank of a matrix, the most important
of which are:

• It is the largest non-zero determinant in the matrix.

• It is the largest number of rays arranged linearly (we take columns or


lines as rays).

• Perform initial transformations on the matrix with the intention of


showing up zeros so that some columns or rows are zero, and in the
end, we conclude that the rank of a matrix is the number of rows or
columns that are non-zero.

We are interessting in the third way because we can apply the Gauss elimi-
nation method.
Now, we have the following system in the matrix form:

AX = B

We have the following cases according to the rank:

1. If rank(A) ̸= rank(A|B), we have in this case that the system has no


solution.

2. If rank(A) = rank(A|B), in this case the system has a unique solution


or an infinity of solutions, and we have the following cases:

(a) If rank(A) = n, n is the number of the unknowns, in this case the


system has a unique solution.

(b) If rank(A) ≤ n, in this case the system has an infinity of solutions.

x + 2y − z = 1



Example 2.2.9. We have the following system:  2x − 3z = 3 (S)

−x − 2y + z = 1
We write the system in the matrix form:

1 2 −1 1
    
x
 2
(S) ⇔ AX = B ⇔  0  y  =  3 
3     

−1 −2 1 z 1

22
2.2. Definitions 2

Next,
1 2 −1 1
 

(A | B) =  2 0 3 3 


−1 −2 1 1
We do some changes, we get:
1 2 −1 1 1 2 −1 1
   

(A | B) =  2 0 3 3  ∼  0 −2 5 1 
  

−1 −2 1 1 0 0 0 2
We remark that rank(A) ̸= rank(A | B).
1 2 −1 1  x + 2y − z = 1
  

(A | B) ∼  0 −2 5 1  ⇔ (S ′ ) −2y + 5z = 1
 

0 0 0 2 0=2

Example 2.2.10. We have the following linear system:


 x + 2y − z = 1


2x + 3z = 3


(S)

 −x − 2y + z = −1
3x + 2y + 4z = 4

Then, we write it in the matrix system.


1 2 −1   1
   
x
 2 0 3    3
  
(S) ⇔ AX = B ⇔   y  = 
 
−1 −2 1   −1

 
z
3 2 4 4
Now, we write (A | B)
1 2 −1 1
 
 2 0 3 3 
(A | B) = 
 
−1 −2 1 −1

 
3 2 4 4
1 2 −1 1 1 2 −1 1
   
 2 0 3 3   0 −4 5 1 
(A | B) =  ∼
   
−1 −2 1 −1 0 0 0 0 
 
  
3 2 4 4 0 −4 7 1
We do some changes, we get:
1 2 −1 1 1 2 −1 1
   
 0 −4 5 1   0 −4 5 1 
(A | B) ∼  ∼
  
0 0 0 0   0 0 0 0 
 

0 −4 7 1 0 0 2 0

23
2.2. Definitions 2

x + 2y − z = 1  x + 2y − z = 1  x= 4 = 2
6 3
  

  
Finally, we have: (S)  −4y + 5z = 1 ⇔  −4y + 5z = 1 ⇔  y = − 14

z=n
3 o

2z = 0 
z=0
Therefore, the set of solutions is: 3
2
, − 14 , 0

Example 2.2.11. We have the following system:

x + 3y − z + t = 1



2x − 4y + z − t = 0
x − y + 4z + t = 1

• Solve this system if it is possible.

24

You might also like