Chapter.

1 Matrices and Systems of Equations

1.5 Elementary Matrices

1. Properties of Elementary Matrices

2. Theorem
3. Equivalent Systems
4. Triangular Factorization

1
 Elementary Matrices
In this section, we view the process of solving a linear system in terms of matrix
multiplications rather than row operations. Given a linear system 𝐴𝐱 = 𝒃, we can
multiply both sides by a sequence of special matrices to obtain an equivalent
system in row echelon form. The special matrices we will use are called
elementary matrices. We will use them to see how to compute the inverse of a
nonsingular matrix and also to obtain an important matrix factorization. We begin
by considering the effects of multiplying both sides of a linear system by a
nonsingular matrix.

2
 Elementary Matrices
Definition
An elementary matrix is a square matrix that differs from the identity matrix by one
single elementary row operation.
Or, If we start with the identity matrix I and then perform exactly one elementary row
operation, the resulting matrix is called an elementary matrix.
Note: Elementary matrix and their inverse are nonsingular.
Example: let us see whether matrix A is an elementary matrix or not.
1 0 1 0
A= ,I=
2 1 0 1
1 0
Solution:
0 1
1 0
~R 2 + 2R1 ∴ A is an elementary matrix
2 1 3
 Elementary Matrices

1 0 0 0 1 0
e.g. 0 1 0 ~𝑅1 ↔ 𝑅2 1 0 0 = 𝐸1 = Elementary matrix
0 0 1 0 0 1
4
See example.1-page.77 from book
 Elementary Matrices

1 0 0 1 0 0
e.g. 0 1 0 ~3𝑅3 0 1 0 = 𝐸2 = Elementary matrix
0 0 1 0 0 3
5
See example.2-page.78 from book
 Elementary Matrices

1 0 0 1 0 3
e.g. 0 1 0 ~R1 + 3R 3 0 1 0 = 𝐸3 = Elementary matrix
0 0 1 0 0 1
6
See example.3-page.78 from book
 1. Properties of Elementary Matrices
Property.1
If 𝐸1 is an elementary matrix as obtained earlier, and A is a 3 × 3 matrix, then
0 1 0 𝑎11 𝑎12 𝑎13 𝑎21 𝑎22 𝑎23
𝐸1 𝐴 = 1 0 0 𝑎21 𝑎22 𝑎23 = 𝑎11 𝑎12 𝑎13
0 0 1 𝑎31 𝑎32 𝑎33 𝑎31 𝑎32 𝑎33
𝑎11 𝑎12 𝑎13 0 1 0 𝑎12 𝑎11 𝑎13
𝐴𝐸1 = 𝑎21 𝑎22 𝑎23 1 0 0 = 𝑎22 𝑎21 𝑎23
𝑎31 𝑎32 𝑎33 0 0 1 𝑎32 𝑎31 𝑎33
Multiplying 𝐴 on the left by 𝐸1 interchanges the first and second rows of 𝐴.
Right multiplication of 𝐴 by 𝐸1 is equivalent to the elementary column operation of
interchanging the first and second columns.

7
 1. Properties of Elementary Matrices
Property.2
If 𝐸2 is an elementary matrix as obtained earlier, and A is a 3 × 3 matrix, then
1 0 0 𝑎11 𝑎12 𝑎13 𝑎11 𝑎12 𝑎13
𝐸2 𝐴 = 0 1 0 𝑎21 𝑎22 𝑎23 = 𝑎21 𝑎22 𝑎23
0 0 3 𝑎31 𝑎32 𝑎33 3𝑎31 3𝑎32 3𝑎33
𝑎11 𝑎12 𝑎13 1 0 0 𝑎11 𝑎12 3𝑎13
𝐴𝐸2 = 𝑎21 𝑎22 𝑎23 0 1 0 = 𝑎21 𝑎22 3𝑎23
𝑎31 𝑎32 𝑎33 0 0 3 𝑎31 𝑎32 3𝑎33

Multiplication on the left by 𝐸2 performs the elementary row operation of multiplying the third
row by 3, while multiplication on the right by 𝐸2 performs the elementary column operation of
multiplying the third column by 3.
8
 1. Properties of Elementary Matrices
Property.3
If 𝐸3 is an elementary matrix as obtained earlier, and A is a 3 × 3 matrix, then

1 0 3 𝑎11 𝑎12 𝑎13 𝑎11 + 3𝑎31 𝑎12 + 3𝑎32 𝑎13 + 3𝑎33

𝐸3 𝐴 = 0 1 0 𝑎21 𝑎22 𝑎23 = 𝑎21 𝑎22 𝑎23
0 0 1 𝑎31 𝑎32 𝑎33 𝑎31 𝑎32 𝑎33
𝑎11 𝑎12 𝑎13 1 0 3 𝑎11 𝑎12 3𝑎11 + 𝑎13
𝐴𝐸3 = 𝑎21 𝑎22 𝑎23 0 1 0 = 𝑎21 𝑎22 3𝑎21 + 𝑎23
𝑎31 𝑎32 𝑎33 0 0 1 𝑎31 𝑎32 3𝑎31 + 𝑎33

Multiplication on the left by 𝐸3 adds 3 times the third row to the first row. Multiplication on the
right adds 3 times the first column to the third column.

9
 1. Properties of Elementary Matrices
Theorem
If 𝐸 is an elementary matrix, then 𝐸 is nonsingular and 𝐸 −1 is an elementary matrix of the same
type.
Examples.
0 1 0 0 1 0 1 0 0
(i). 𝐸 ~𝑅1 ↔ 𝑅2 𝐸 ~𝑅1 ↔ 𝑅2 1 0 0 1 0 0= 0 1 0
0 0 1 0 0 1 0 0 1
1 0 0 1 0 0 1 0 0
1
(ii). 𝐸 ~3𝑅3 𝐸 ~ 𝑅3 0 1 0 0 1 0 = 0 1 0
3
0 0 3 0 0 1Τ3 0 0 1
1 0 3 1 0 −3 1 0 0
(iii). 𝐸(~R1 + 3R 3 )𝐸(~R1 − 3R 3 ) 0 1 0 0 1 0 = 0 1 0
0 0 1 0 0 1 0 0 1
10
 2. Row equivalent
Definition
A matrix 𝐵 is a row equivalent to a matrix 𝐴 if there exists a finite sequence 𝐸1 , 𝐸2 , … , 𝐸𝑘 of
elementary matrices such that
𝐵 = 𝐸𝑘 𝐸𝑘−1 ··· 𝐸1 𝐴.
Denoted by 𝐴 ≅ 𝐵
In other words, 𝐵 is row equivalent to 𝐴 if 𝐵 can be obtained from 𝐴 by a finite number of row
operations. In particular, if two augmented matrices (𝐴 | 𝒃) and (𝐵 | 𝒄) are row equivalent, then
𝐴𝐱 = 𝐛 and 𝐵𝐱 = 𝐜 are equivalent systems. The following properties of row equivalent matrices
are easily established.
i. If 𝐴 is row equivalent to 𝐵, then 𝐵 is row equivalent to 𝐴. i.e. 𝐴 ≅ 𝐵, ⇒ 𝐵 ≅ 𝐴
ii. If 𝐴 is row equivalent to 𝐵, and 𝐵 is row equivalent to 𝐶, then 𝐴 is row equivalent to 𝐶. i.e.
𝐴 ≅ 𝐵, 𝐵 ≅ 𝐶 ⇒ 𝐴 ≅ 𝐶
11
 2. Row equivalent
Theorem: Equivalent Conditions for Nonsingularity
Let 𝐴 be an 𝑛 × 𝑛 matrix. The following are equivalent:
(a). 𝐴 is nonsingular.
(b). 𝐴𝐱 = 𝟎 has only the trivial solution 0.
(c). 𝐴 is row equivalent to 𝐼
Proof:
(a)→ (b)
If 𝐴 is nonsingular (𝐴𝐵 = 𝐵𝐴 = 𝐼) and 𝐱ො is a solution of 𝐴ො𝐱 = 𝟎 then
𝐱ො = 𝑰ො𝐱 = 𝑨−𝟏 𝑨 𝐱ො = 𝑨−𝟏 𝑨ො𝐱 = 𝑨−𝟏 𝟎 = 𝟎
Thus, Ax = 0 has only the trivial solution.

12
 2. Row equivalent
Theorem: Equivalent Conditions for Nonsingularity
Proof: (continued)
(b)→ (c)
If we use elementary row operations, the system can be transformed into the form Ux = 0, where
U is in row echelon form. If one of the diagonal elements of U were 0, the last row of U would
consist entirely of 0’s. But then Ax = 0 would be equivalent to a system with more unknowns
than equations and hence, by Theorem 1.2.1, would have a nontrivial solution. Thus, U must be
a strictly triangular matrix with diagonal elements all equal to 1. It then follows that I is the
reduced row echelon form of A and hence A is row equivalent to I.
I
𝐴𝑛×𝑛 𝑟𝑟𝑒𝑓 𝑅𝑛×𝑛 ⇒ 𝑅𝑛×𝑛 → ቊ
There is at least one row with all zeros.
∵ 𝐴 can’t be an undetermined system,
∴𝐴→I
13
 2. Row equivalent
Theorem: Equivalent Conditions for Nonsingularity
(b)→ c (continued)
Example:
1 2 0 1 2 0 1 2 0 1 2 0
1 1
𝐴 = 2 1 0 ~𝑟2 − 2𝑟1 0 −3 0 = 𝑅1 , ~ −3 𝑟2 0 1 0 = 𝑅2 , ~ 3 𝑟3 0 1 0 = 𝑅3
0 0 3 0 0 3 0 0 3 0 0 1
𝑅1 = 𝐸1 𝐴 𝑅2 = 𝐸2 𝑅1 𝑅3 = 𝐸3 𝑅2
1 0 0
~𝑟1 − 2𝑟2 0 1 0 = 𝑅4 = 𝐸4 𝑅3 1 0 0
I= 0 1 0
0 0 1 0 0 1
1 0 0
1 0 0 1 2 0 1 2 0 𝐸1 = ~𝑟2 − 2𝑟1 −2 1 0
𝑅1 = 𝐸𝑟2 −2𝑟1 𝐴 = 𝐸1 𝐴 = −2 1 0 2 1 0 = 0 −3 0 0 0 1
0 0 1 0 0 3 0 0 3 14
 2. Row equivalent
Theorem: Equivalent Conditions for Nonsingularity
(b)→ c (continued)
Example:(continued)
1 0 0
1 1 2 0 1 2 0
𝑅2 = 𝐸 1 𝐴 = 𝐸2 𝑅1 = 0 0 0 −3 0 = 0 1 0
−3𝑟2 −3 0 0 3 0 0 3
0 0 1
rref(A)= 𝐸4 𝑅3 = 𝐸4 𝐸3 𝑅2 = 𝐸4 𝐸3 𝐸2 𝑅1 = 𝐸4 𝐸3 𝐸2 𝐸1 𝐴 = I

15
 2. Row equivalent
Theorem: Equivalent Conditions for Nonsingularity
Proof: (continued)
(c)→ (a)
𝐴 = 𝐸𝑘 𝐸𝑘−1 … 𝐸1 I = 𝐸𝑘 𝐸𝑘−1 … 𝐸1
But since 𝐸𝑖 is invertible, 𝑖 = 1, … , 𝑘, the product 𝐸𝑘 𝐸𝑘−1 … 𝐸1 is also invertible.
Hence, A is nonsingular and

𝐴−1 = (𝐸𝑘 𝐸𝑘−1 … 𝐸1 )−1 = 𝐸1−1 𝐸2−1 … 𝐸𝑘−1

Therefore A is nonsingular

16
 3. Equivalent Systems
Definition
Systems of equations that have the same solution are called equivalent systems.
Given a system of two equations, we can produce an equivalent system by
replacing one equation with the sum of the two equations, or by replacing an
equation with a multiple of itself.
Example:
x+y=1 … (a) 2x+2y=2 … (𝑐)
ቊ …(i) convert equation i to → ቊ ….(ii)
3x+6y=3 ...(b) x+2y=1 … (𝑑)
Sol (i): Putting (x,y)=(1,0) in system (ii)
(x,y)=(1,0) 2(1)+2(0)=2 => which satisfy equ(c)
(1)+2(0)=1 => which satisfy equ(d)
Therefore both systems of equations (i) and (ii) are equivalent.
17
 3. Equivalent Systems
Corollary
The system 𝐴𝐱 = 𝐛 of n linear equations in n unknowns has a unique solution if and only
if 𝐴 is nonsingular.
Method to find the inverse of 𝐴
If 𝐴 is nonsingular then 𝐴 is equivalent to I, and there exist elementary matrices
𝐸1 𝐸2 … 𝐸𝑘 such that
𝐸𝑘 𝐸𝑘−1 … 𝐸1 𝐴 = I
Multiply both sides by 𝐴−1
𝐸𝑘 𝐸𝑘−1 … 𝐸1 𝐴𝐴−1 = I𝐴−1 ⇒ 𝐸𝑘 𝐸𝑘−1 … 𝐸1 I = 𝐴−1
𝐸𝑘 𝐸𝑘−1 … 𝐸1 𝐴 ⋮ I = I ⋮ 𝐴−1
The reduce row echelon form of the augmented matrix 𝐴 ⋮ I will be I ⋮ 𝐴−1
18
 3. Equivalent Systems
Example

Thus,

19
 3. Equivalent Systems
Example

20
 4. Triangular Factorization
If an 𝑛 × 𝑛 matrix 𝐴 can be reduced to strict upper triangular form using only row
operation III, then it is possible to represent the reduction process in terms of a matrix
factorization.
The factorization of a matrix A into
a product of a unit lower triangular
matrix L times a Strictly upper
triangular matrix U is often
referred to as LU factorization.
0

21
 4. Triangular Factorization
Example
let us use only row operation III to carry out the reduction process.

1
~𝑟2 − 2 𝑟1 ~𝑟3 + 3𝑟2 = U = 𝐸3 𝐸2 𝐸1 𝐴 If you do the same operation in the
𝑟3 − 2𝑟1 identity matrix, You will get 𝐸1 , 𝐸2 , 𝐸3
1 0 0
I= 0 1 0
0 0 1
1 0 0
1 1
𝐸1 = ~𝑟2 − 𝑟1 − 1 0
2 2
0 0 1
22
 4. Triangular Factorization
Example (continued)
U = 𝐸3 𝐸2 𝐸1 𝐴

A = (𝐸3 𝐸2 𝐸1 )−1 U = LU

(𝐸3 𝐸2 𝐸1 )−1 = 𝐿, (𝐸3 𝐸2 𝐸1 )−1 = 𝐸1−1 𝐸2−1 𝐸3−1 , also see theorem, page.11

𝐸1−1 𝐸2−1 𝐸3−1 =

23
 4. Triangular Factorization
Example (continued)

A = LU =

This means we decompose/factorize matrix A into lower triangular matrix “L”

and upper triangular matrix “U”.

24