SYSTEMS OF LINEAR EQUATIONS

1. Main definitions
Definition 1.1. Let m ≥ 1 and n ≥ 1 be two integers. A system of m linear
equations with n variables x1 , x2 , ..., xn is a set of m equations of the following
form:


 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 the coefficients ai,j for 1 ≤ i ≤ m and 1 ≤ j ≤ n and the coefficients bi for
1 ≤ i ≤ m are fixed real numbers. To solve the system is to find the set of all
vectors  
x1
 x2 
−
→  
x = 
 ... 
 ... 
xn
whose coordinates x1 , x2 , ..., xn satisfy simultaneously all the m equations.

The set of all the vectors −

→
x which satisfy simultaneously the m equations is
called the set of solutions or the space of solutions of the system.
By multiplying the matrix A and the column vector − →
x , we see that the above
system of linear equations is equivalent to only one matrix equation:
−
→
A−→
x = b
where A is the coefficient matrix of the system:
 
a1,1 a1,2 ... ... a1,n
 a2,1 a2,2 ... ... a2,n 
A=  ...


... ... ...
am,1 am,2 ... ... am,n
−
→
and b is the column vector:  
b1
 b2 
→ 
− 
b =
 ... .

 ... 
bm
Remark 1.2. The number of equations m and the number of variables n don’t
have to be the same. So the m × n matrix A is not necessarily a square matrix.
1
 We will also use the following matrix B, obtained by putting the matrix A and
−
→
the vector b side by side. B has m rows and n + 1 columns, and is called the
augmented matrix of the system:
 
a1,1 a1,2 ... ... a1,n b1
 a2,1 a2,2 ... ... a2,n b2 
B=
 ...
.

... ... ... ...
am,1 am,2 ... ... am,n bm

Example 1.3. The following is a system of 2 linear equations with 3 variables:

{
x1 + x2 − x3 = 1
3x2 − 2x3 = 0
and we have:
( ) ( ) ( )
1 1 −1 −
→ 1 1 1 −1 1
A= , b = , B= .
0 3 −2 0 0 3 −2 0
In the following sections, we will describe several methods for solving systems of
linear equations. The augmented matrix B will play an important role.

2. The substitution method

In this section, we will describe the most straightforward method for solving
systems of linear equations. It is called the substitution method. The idea is
to compute one of the variables in one of the equations as a function of the other
variables, and substitute it in the other equations. After iterating this process at
most n − 1 times, we will be able to compute the set of solutions of the system.

Let us begin with the case m = n = 2. These are systems of two equations in
two variables, of the type:
{
a1,1 x1 + a1,2 x2 = b1
a2,1 x1 + a2,2 x2 = b2
and everybody has already studied them at least once.

The easiest way to solve this system is by substitution. First, we compute x1 in

the first equation as a function of x2 and we replace x1 in the second equation.

Three cases can occur. If x2 can be computed in the second equation, then
we compute x2 and replace it back in the first equation, and compute x1 : the
system has a unique solution. If x2 disappears from the second equation, then
the second equation becomes something of the form a = b where a and b are real
numbers. If a is different from b, we get a contradiction: this equation can never be
satisfied. This means that the system has no solution. If a is equal to b, we get
an equation which is always satisfied, so we can delete it without changing the set
of solutions. The system is equivalent to only one linear equation: it has infinitely
many solutions. We give three examples below, of the three different cases.

2
Example 2.1.

Example 1:
{
x1 + x2 =3
3x1 − x2 =5
{ { { {
x1 = 3 − x2 x1 = 3 − x2 x1 = 3 − x2 x1 = 2
⇐⇒ ⇐⇒ ⇐⇒
3(3 − x2 ) − x2 = 5 9 − 4x2 = 5 x2 = 1 x2 = 1
The system has a unique solution:
( ) ( )
x1 2
= .
x2 1
Example 2:
{
x1 + 2x2 =7
−x1 − 2x2 =2
{ {
x1 = 7 − 2x2 x1 = 7 − 2x2
⇐⇒
−(7 − 2x2 ) − 2x2 = 2 −7 = 2
The second equation yields a contradiction. Therefore, the system has no solution.
We say that the set of solutions is empty or that the system is inconsistent.

Example 3:
{
x1 + 2x2 =7
−x1 − 2x2 = −7
{ {
x1 = 7 − 2x2 x1 = 7 − 2x2
⇐⇒
−(7 − 2x2 ) − 2x2 = −7 −7 = −7
Here, the second equation is tautological: it is always satisfied. Therefore, this
equation can be deleted without changing the space of solutions of the system. The
system is equivalent to only one equation:
x1 = 7 − 2x2 .
This means that x2 can take any real value and x1 is equal to 7 − 2x2 . We can
express this clearly by setting x2 = t to be any real number. The set of solutions
is the set of all ( ) ( )
x1 7 − 2t
= , t ∈ R.
x2 t
It is an infinite set: the system has infinitely many solutions.
The substitution method can be used to solve any system of linear equations.
We give two more examples, of larger systems.
Example 2.2. 
 x1 + x2 + x3 =0
2x1 − x2 − x3 =6

x1 + 2x2 − x3 =7
3
First, we compute x1 in the first equation, as a function of x2 and x3 , and we
substitute it in the second and third equations.
 
 x1 = −x2 − x3  x1 = −x2 − x3
2(−x2 − x3 ) − x2 − x3 = 6 ⇐⇒ −3x2 − 3x3 = 6
 
(−x2 − x3 ) + 2x2 − x3 = 7 x2 − 2x3 = 7
Now, we don’t touch at the first equation. The second and third equations form
a system of two equations with two variables. We solve it by computing x2 in the
second equation as a function of x3 and substituting it in the third equation.
  
 x1 = −x2 − x3  x1 = −x2 − x3  x1 = 2
x2 = −2 − x3 ⇐⇒ x2 = −2 − x3 ⇐⇒ x2 = 1
  
(−2 − x3 ) − 2x3 = 7 x3 = −3 x3 = −3
Once we have computed x3 in the third equation, we substitute it back into the
second equation, to compute x2 , and then we substitute x2 and x3 in the first
equation to compute x1 . The system has a unique solution:
   
x1 2
 x2  =  1  .
x3 −3
Example 2.3. 

 x1 + x2 =3

3x1 − x2 −4x4 =5

 −x1 + 2x2 − x3 =0

3x1 + 2x2 − x3 −4x4
=8
 

 x1 = 3 − x2 
 x1 = 3 − x2
 
3(3 − x2 ) − x2 − 4x4 = 5 9 − 4x2 − 4x4 = 5
⇐⇒

 −(3 − x 2 ) + 2x 2 − x 3 = 0 
 −3 + 3x2 − x3 = 0
 
3(3 − x2 ) + 2x2 − x3 − 4x4 = 8 9 − x2 − x3 − 4x4 = 8
 
 x1 = 3 − x2





x1 = 3 − x2
x2 + x4 = 1 x2 = 1 − x4
⇐⇒ ⇐⇒

 3x 2 − x 3 = 3 
 3(1 − x4 ) − x3 = 3
 
−x2 − x3 − 4x4 = −1 −(1 − x4 ) − x3 − 4x4 = −1
 

 x1 = 3 − x2 
 x1 = 3 − x2
 
x2 = 1 − x4 x2 = 1 − x4
⇐⇒ ⇐⇒
 −x3 − 3x4 = 0
  x3 = −3x4

 
−x3 − 3x4 = 0 0=0
The last equation can be deleted and x4 can take any real value. We substitute
back to get the result: 
 x1 = 2 + x4
x2 = 1 − x4

x3 = −3x4
So the system has infinitely many solutions, and the set of solutions is:
   
x1 2+t
 x2   1 − t 
   
 x3  =  −3t  , t ∈ R.
x4 t
4
 3. The row reduction method
Here, we introduce a different, more algorithmic method to solve systems of
linear equations, by performing ”row operations” on the system.

Proposition 3.1. The following elementary row operations on a system of

linear equations do not change the space of solutions of the system.
• Operation 1: Exchanging two equations.
• Operation 2: Multiplying one equation by a nonzero constant.
• Operation 3: Adding to one equation the product of another equation by
a constant.
Proof. Operation 1 clearly doesn’t change the set of solutions, since the order of
the equations that have to be satisfied simultaneously is irrelevant. Operation 2
consists of replacing an equation by an equivalent equation, which clearly doesn’t
change the space of solutions. Our only problem is Operation 3. Given two linear
equations, we need to prove that the systems:
{
a1 x1 + ... + an xn = b
a′1 x1 + ... + a′n xn = b′
and {
a1 x1 + ... + an xn = b
a′1 x1 + ... + a′n xn + λa1 x1 + ... + λan xn = b′ + λb
are equivalent for any scalar λ ∈ R. However, this is obvious, since any solution
of the first system is clearly a solution of the second one, and any solution of the
second one is a solution of the first one. □

We will show that any system of linear equations can be solved by using only
these three elementary row operations.

The method is as follows: first, by using the operations, we make the coefficient
of x1 in the first row equal to 1. Then, by using operation 3, we ”remove” x1 from
equations 2 to m (this amounts to substituting x1 in equations 2 to m). Then, we
don’t touch at line 1 anymore, and we repeat the process with x2 in line 2. After
a finite number of steps, either we encounter a contradiction, in which case the
system has no solution, or we can compute one variable in the last equation and
”substitute” it back (using operation 3) to find the set of solutions.

We give the same examples as in the previous section, using this new method.
The reader is encouraged to compare the two methods.
Example 3.2. (Example 2.2 of section 2)

 x1 + x2 + x3 = 0
2x1 − x2 − x3 = 6

x1 + 2x2 − x3 = 7
  
 x1 + x2 + x3 = 0  x1 + x2 + x3 = 0  x1 + x2 + x3 = 0
−3x2 − 3x3 = 6 ⇐⇒ x2 + x3 = −2 ⇐⇒ x2 + x3 = −2
  
x2 − 2x3 = 7 x2 − 2x3 = 7 −3x3 = 9
5
   
 x1 + x2 + x3 = 0  x1 + x2 = 3  x1 = 2
x2 + x3 = −2 ⇐⇒ x2 = 1 ⇐⇒ x2 = 1
  
x3 = −3 x3 = −3 x3 = −3
We can see already on this example that, when performed correctly, this method
is quicker and more elegant than the substitution method used in Example 2.2.
Example 3.3. (Example 2.3 of section 2)


 x1 + x2 =3

3x1 − x2 −4x4 =5

 −x1 + 2x2 − x3 =0

3x1 + 2x2 − x3 −4x4 =8
Here, we start by adding the first row to the third one, and subtracting the
second row from the fourth one to ”remove ” the variable x1 .


 x1 + x2 = 3

3x1 − x2 − 4x4 = 5

 3x2 − x3 = 3

3x2 − x3 = 3
Then we add to the second row the first row multiplied by −3, to remove x1 from
the second equation, and we subtract the third row from the fourth row.


 x1 + x2 = 3

−4x2 − 4x4 = −4

 3x 2 − x3 = 3

0=0
Now we divide the second equation by −4 to make the coefficient of x2 equal to 1
in the second equation, and we delete the fourth equation, which is 0 = 0.

 x1 + x2 = 3
x2 + x4 = 1

3x2 − x3 = 3
Finally, we add to the third row the second row multiplied by (−3) to remove x2
from the second equation. 
 x1 + x2 = 3
x2 + x4 = 1

−x3 − 3x4 = 0
Finally, we multiply the third equation by (−1) to make the coefficient of x3 equal
to 1, and we subtract the second row from the first one to remove x2 from the first
equation: 
 x1 − x4 = 2
x2 + x4 = 1

x3 + 3x4 = 0
From here, x4 = t can take any real value, and the set of solutions is:


 x1 = 2 + t

x2 = 1 − t
t ∈ R.

 x 3 = −3t

x4 = t
6
 4. Elementary row operations on matrices
Recall that to each system of m linear equations with n variables, we can asso-
ciate a matrix B with m rows and n + 1 columns, which is called the augmented
matrix of the system (see Section 1). Conversely, to each matrix B, we can asso-
ciate the system of linear equations whose augmented matrix is B.

In the following sections, we will show how to solve a system of linear equations
by performing the elementary row operations on the augmented matrix B instead
of performing them on the system itself.

This has the advantage of being quicker, since there is no need to write all over
again the variables x1 , x2 , x3 .... We deal only with the coefficients of the variables
and perform all the operations on them.

Definition 4.1. Let B be a matrix. The following three operations are called
elementary row operations on B:
• Operation 1: Exchanging two rows.
• Operation 2: Multiplying one row by a nonzero constant.
• Operation 3: Adding to one row the product of another row by a constant.
Definition 4.2. Two matrices B and B ′ are said to be row equivalent if B ′ is
obtained from B by performing a finite number of elementary row operations. We
write
B ∼ B′
when B and B ′ are row equivalent.
Proposition 4.3. Let B and B ′ be row equivalent matrices with m rows and n + 1
columns. The systems of linear equations associated with B and with B ′ have the
same set of solutions.
Proof. The proof is a direct consequence of the proof of Proposition 3.1. □
Example 4.4. (Examples 2.2 and 3.2)

 x1 + x2 + x3 =0
2x1 − x2 − x3 =6

x1 + 2x2 − x3 =7
The augmented matrix of this system is:
 
1 1 1 0
B =  2 −1 −1 6 
1 2 −1 7
and we perform the elementary row operations on B as follows. These are exactly
the operations that we performed on the system of example 3.2.
       
1 1 1 0 1 1 1 0 1 1 1 0 1 1 1 0
 2 −1 −1 6  ∼  0 −3 −3 6  ∼  0 1 1 −2  ∼  0 1 1 −2 
1 2 −1 7 0 1 −2 7 0 1 −2 7 0 0 −3 9
     
1 1 1 0 1 1 0 3 1 0 0 2
∼  0 1 1 −2  ∼  0 1 0 1  ∼  0 1 0 1 
0 0 1 −3 0 0 1 −3 0 0 1 −3
7
Finally, we obtain the matrix
 
1 0 0 2
B′ =  0 1 0 1 
0 0 1 −3
which is row-equivalent to B. The system of linear equations associated with B’ is:

 x1 = 2
x2 = 1

x3 = −3
and this gives us the solution.

5. Row-echelon forms
In this section, we explain how to transform a matrix B, by elementary row
operations, into a ”more convenient” matrix B ′ . This process is called Gaussian
elimination.

Let B be a matrix with m rows and n + 1 columns. For each i (1 ≤ i ≤ m), let
Bi be the i-th row of B:
Bi = (bi,1 , bi,2 , ..., bi,n+1 ).
Bi is a row vector with n + 1 coordinates.

If all the coordinates bi,1 , bi,2 , ..., bi,n+1 are equal to zero, we say that Bi is a
zero row. Otherwise, the first nonzero coordinate of Bi to the left is called the
leading entry or the pivot of the row Bi .
Example 5.1. Here are a few possibilities for matrices with 5 columns:
• Bi = (2, 0, −1, 0, 7) : the leading entry is bi,1 = 2;
• Bi = (0, 0, 4, −5, 0) : the leading entry is bi,3 = 4;
• Bi = (0, 0, 0, 0, −1) : the leading entry is bi,5 = −1;
• Bi = (0, 0, 0, 0, 0) is the zero row: there is no leading entry.
Definition 5.2. We say that a matrix B is in reduced row-echelon form if the
folloowing four conditions are satisfied:
(1) All the zero rows are at the bottom of the matrix.
(2) For any nonzero row, the leading entry of that row is to the left of all the
leading entries in the rows below it.
(3) All leading entries are equal to 1.
(4) In each column containing a leading entry, all the other entries are zero.
If conditions (1) and (2) are satisfied but conditions (3) and (4) are not, we say
that B is in row-echelon form (but not in reduced row-echelon form).
Remark 5.3. For a matrix in row-echelon form, the leading entries form the shape
of a ”staircase” with zeros below it.
Example 5.4. Here, the matrices B, B ′ and B ′′ are row equivalent. B ′ is in
row-echelon form and B ′′ is in reduced row-echelon form:
     
2 2 4 6 2 2 4 6 1 0 2 1
B =  0 3 0 6  , B ′ =  0 3 0 6  , B ′′ =  0 1 0 2  .
2 2 4 6 0 0 0 0 0 0 0 0
8
 We will now prove that any matrix B can be transformed, by elementary row
operations, into a matrix B ′ which is in reduced row-echelon form. This is one of
the main results in our course.
Theorem 5.5. For any matrix B there exists a matrix B ′ which is in reduced
row-echelon form and which is row equivalent to B. B ′ is called the reduced
row-echelon form of B.
Proof. We will prove it by induction on the number of rows m of the matrix B.
• If m = 1, B = B1 is a row vector:
B = (b1,1 , ..., b1,n+1 ).
If B is the zero row, we are done. Otherwise, let b1,i be the leading entry of
the row. By dividing B by b1,i we obtain a new row vector whose leading
entry is 1, and we are done.
• Now suppose the theorem is true for all matrices with m rows and n + 1
columns, for a given m ≥ 1 , and for all n ≥ 0, and let B be a matrix with
m + 1 rows and n + 1 columns.
– If the first column of B is zero, then by applying the induction hy-
pothesis to the matrix
 
b2,2 ... ... b2,n+1
 ... ... .. ... 
b2,m+1 ... ... bm+1,n+1
which has m rows, we are done.
– If the first column of B is nonzero, then choose a number i such that
bi,1 ̸= 0 and exchange row i and row 1. Then, divide the first row by
bi,1 to make the leading entry of the first row equal to 1. Then, using
operation 3, for each row, from row 2 to row m + 1, add to that row
the first row multiplied by an appropriate constant, in order to make
all the entries in the first column equal to zero, except the top one
which remains equal to 1. We obtain a new matrix B ′ which is row
equivalent to B. The first column of B ′ is:
 
1
 0 
 
 0 
 
 ... 
0
and we can apply the induction hypothesis to the matrix:
 
b′2,2 ... ... b′2,n+1
 ... ... .. ... 
b′2,m+1 ... ... b′m+1,n+1
which has m rows, so we are done.
□

9
Remark 5.6. We will prove in later chapters that all the row-echelon forms of
a matrix B have the same number of zero rows. The rank of a matrix B is
defined as the number of nonzero rows of the row-echelon forms of B and it is
denoted by rank(B). In Example 5.4, you can check that rank(B) = 2.

6. Describing the set of solutions of a system of linear equations

We will now describe the systematic method for solving a system of linear equa-
tions by using only elementary row operations performed on the aug-
mented matrix B of the system.

We proceed in the following four steps:

(1) Write down the augmented matrix B of the system.
(2) Transform the matrix B by Gaussian elimination into its reduced row-
echelon form B ′ (this is the main step).
(3) Write down the system of linear equations associated with B ′ and interpret
the result.
(4) Write down the set of solutions.
Steps (1) and (2) have been described in the previous section. Let us now explain
steps (3) and (4).

After writing down the system of linear equations associated with B ′ , we first
look at the last equation of the system. If the last nonzero row of B ′ is equal to
(0, 0, ..., 0, 1)
then the last equation of the system becomes:
0=1
which is impossible. The system has no solution.

Actually, during the process of Gaussian elimination, if one of the rows of the
matrix becomes of the type
(0, 0, ..., 0, a)
with a being a nonzero real number, then the corrsponding equation will be:
0=a
which is impossible. The system has no solution. The process can be stopped at
this stage, without going all the way to computing the reduced row-echelon form
of B.
Definition 6.1. A system of linear equations is called consistent if it has at least
one solution. Otherwise, if the system has no solution, it is called inconsistent.
When the system is consistent, each nonzero row of the matrix B ′ gives an
equation of the type:
xk = αk+1 xk+1 + ... + αn xn + βk
where the coefficients αi and βk are real numbers (eventually zero). The variable
xk on the left side of the above equation, which corresponds to the leading entry
1 in the row, is called a leading variable. The variables which are not leading
10
variables are called free variables.

Since every column of B ′ with a leading entry has zeros everywhere else, the free
variables and the leading variables are well defined.

The free variables can take any real value, and the leading variables are computed
with respect to the free variables to give the set of solutions of the system.
Theorem 6.2. A system of linear equations has either no solution or a unique
solution or infinitely many solutions.
Proof. If the last nonzero row of B ′ is
(0, 0, ..., 0, 1)
then the system has no solution. Otherwise, the system associated with B ′ has k
free variables and n − k leading variables, where k is an integer between 0 and n.
If k = 0, then there are no free variables, and the system has a unique solution.
If k ≥ 1, the k free variables can take all real values, so the system has infinitely
many solutions. □

7. Examples
Example 7.1. (Examples 2.3 and 3.3)


 x1 + x2 =3

3x1 − x2 −4x4 =5

 −x1 + 2x2 − x3 =0

3x1 + 2x2 − x3 −4x4 =8
The augmented matrix of the system is:
 
1 1 0 0 3
 3 −1 0 −4 5 
B=  −1

2 −1 0 0 
3 2 −1 −4 8
We perform Gaussian elimination on B.
     
1 1 0 0 3 1 1 0 0 3 1 1 0 0 3
 3 −1 0 −4 5   0 −4 0 −4 −4   0 1 0 1 1 
   ∼ 
 −1 2 −1 0 0  ∼  0 3 −1 0 3   0 3 −1 0 3 
3 2 −1 −4 8 0 −1 −1 −4 −1 0 −1 −1 −4 −1
     
1 1 0 0 3 1 1 0 0 3 1 0 0 −1 2
 0 1 0 1 1   0 1 0 1 1   0 1 0 1 1 
∼  
 0 0 −1 −3 0  ∼  0 0
∼
  0 0

1 3 0 1 3 0 
0 0 −1 −3 0 0 0 0 0 0 0 0 0 0 0
The reduced row-echelon form of B is
 
1 0 0 −1 2
 0 1 0 1 1 
B′ = 
 0

0 1 3 0 
0 0 0 0 0
11
so rank(B) = 3. The system of linear equations corresponding to B ′ is:

 x1 − x4 = 2
x2 + x4 = 1

x3 + 3x4 = 0

There is one free variable x4 and three leading variables x1 , x2 , x3 . The system has
infinitely many solutions and the set of solutions is:
   
x1 2+t
 x2   1 − t 
   
 x3  =  −3t  , t ∈ R.
x4 t

Example 7.2.


 2x1 − x2 − x3 + 3x4 =5

−x1 − x2 + 2x3 + x4 =1

 3x1 − 3x3 + 2x4 =4

x1 − 5x2 + 4x3 + 9x4 =8
The augmented matrix of the system is:
 
2 −1 −1 3 5
 −1 −1 2 1 1 
B=  
3 0 −3 2 4 
1 −5 4 9 8

By Gaussian elimination:
   
2 −1 −1 3 5 1 1 −2 −1 −1
 −1 −1 2 1 1   2 −1 −1 3 5 
 ∼ 
 3 0 −3 2 4   3 0 −3 2 4 
1 −5 4 9 8 1 −5 4 9 8

   
1 1 −2 −1 −1 1 1 −2 −1 −1
 0 −3 3 5 7   0 3 −3 −5 −7 
∼
 0
∼ 
−3 3 5 7   0 0 0 0 0 
0 −6 6 10 9 0 0 0 0 −5
At this stage, we can see that our system of four equations has become equivalent
to the following system:


 x1 + x2 − 2x3 − x4 = −1

3x2 − 3x3 − 5x4 = −7

 0=0

0 = −5

The last equation is impossible, therefore the system is inconsistent: it has no

solution. In this example, we don’t need to compute the reduced row-echelon form
of B.

12
Example 7.3.


 2x1 − 3x2 + x3 =8

x1 − x2 + 2x4 + x5 =3

 −3x 1 + x2 + x5 = −7

x1 − 4x2 + x3 + 4x4 + 3x5 =7
The augmented matrix is:
 
2 −3 1 0 0 8
 1 −1 0 2 1 3 
B=
 −3

1 0 0 1 −7 
1 −4 1 4 3 7
and we obtain:
   
2 −3 1 0 0 8 1 −1 0 2 1 3
 1 −1 0 2 1 3   2 −3 1 0 0 8 
   
 −3 1 0 0 1 −7  ∼  −3 1 0 0 1 −7 
1 −4 1 4 3 7 1 −4 1 4 3 7
   
1 −1 0 2 1 3 1 −1 0 2 1 3
 0 −1 1 −4 −2 2   0 1 −1 4 2 −2 
∼ 0 −2 0 6
∼ 
4 2   0 −2 0 6 4 2 
0 −3 1 2 2 4 0 −3 1 2 2 4
   
1 −1 0 2 1 3 1 −1 0 2 1 3
 0 1 −1 4 2 −2   0 1 −1 4 2 −2 
∼ 0 0 −2 14 8 −2  ∼  0 0
  
1 −7 −4 1 
0 0 −2 14 8 −2 0 0 0 0 0 0
   
1 −1 0 2 1 3 1 0 0 −1 −1 2
 0 1 0 −3 −2 −1   0 1 0 −3 −2 −1 
∼  
 0 0 1 −7 −4 1  ∼  0 0 1 −7 −4 1 


0 0 0 0 0 0 0 0 0 0 0 0
The reduced row-echelon form of B is
 
1 0 0 −1 −1 2
 0 1 0 −3 −2 −1 
B′ = 
 0 0 1 −7 −4 1 


0 0 0 0 0 0
and rank(B) = 3. The corresponding system of linear equations is:

 x1 − x4 − x5 = 2
x2 − 3x4 − 2x5 = −1

x3 − 7x4 − 4x5 = 1
It has two free variables, x4 and x5 , and three leading variables, x1 , x2 , x3 . The
free variables x4 = s and x5 = t can independently take all real values. The system
has infinitely many solutions and the set of solutions is:
   
x1 2+s+t
 x2   −1 + 3s + 2t 
   
 x3  =  1 + 7s + 4t  , t ∈ R, s ∈ R.
   
 x4   s 
x5 t
13
Example 7.4.


 x1 + x2 + x3 − x4 − x5 =0

3x1 − x2 − x3 + x4 + x5 =8

 x1 + x2 + x3 − x4 +2 x5 =9

−x1 + x2 + 2x3 + x4 − x5 = −8
The augmented matrix of the system is:
 
1 1 1 −1 −1 0
 3 −1 −1 1 1 8 
B=  1

1 1 −1 2 9 
−1 1 2 1 −1 8
and by Gaussian elimination:
     
1 1 1 −1 −1 0 1 1 1 −1 −1 0 1 1 1 −1 −1 0
 3 −1 −1 1 1 8   0 −4 −4 4 4 8   0 1 1 −1 −1 −2 
 ∼ ∼ 
 1 1 1 −1 2 9   0 0 0 0 3 9   0 2 3 0 −2 −8 
−1 1 2 1 −1 8 0 2 3 0 −2 −8 0 0 0 0 1 3
   
1 1 1 −1 −1 0 1 0 0 0 0 2
 0 1 1 −1 −1 −2   0 1 1 −1 −1 −2 
∼  0 0 1 2
∼ 
0 −4   0 0 1 2 0 −4 
0 0 0 0 1 3 0 0 0 0 1 3
   
1 0 0 0 0 2 1 0 0 0 0 2
 0 1 1 −1 0 1   0 1 0 −3 0 5 
∼ 
 0 0 1 2 0 −4  ∼∼  0 0 1 2
 
0 −4 
0 0 0 0 1 3 0 0 0 0 1 3
The reduced row-echelon form of B is
 
1 0 0 0 0 2
 0 1 0 −3 0 5 
B′ =  0 0 1 2 0 −4 


0 0 0 0 1 3
and rank(B) = 4. The system associated with B ′ is:


 x1 = 2

x2 − 3x4 = 5

 x3 + 2x4 = −4

x5 = 3
There is one free variable x4 and there are four leading variables x1 , x2 , x3 and
x5 . The variable x4 = t can take any real value. The system has infinitely many
solutions and the set of solutions is:
   
x1 2
 x2   5 + 3t 
   
 x3  =  −4 − 2t  , t ∈ R.
   
 x4   t 
x5 3

14