UNIT 4

SYSTEMS OF LINEAR
EQUATIONS
Structure
Page Nos.
4.1 Introduction 5
Objective
4.2 Gaussian Elimination 6
4.3 Row Operations 8
4.4 Systems of Linear Equations 19
4.5 Summary 27
4.6 Solutions/Answers 27

4.1 INTRODUCTION
In this Block we continue our study of Linear Algebra with a discussion of
solution of simultaneous equations. The problem of solution of linear equations
efficiently is one of the fundamental problems of Mathematics. Efficient
methods of solution of systems of linear equations has many applications,
including in new areas like Data Science.

In Sec. 4.2, we discuss Gaussian elimination, a procedure for solution of

systems of linear equations invented by C.F. Gauss. In Sec. 4.3 we reformulate
the procedure in terms of operations on the rows of a matrix associated to the
system of equations. We will see that we can solve a system of linear
equations by reducing the matrix associated to the system to a special form by
row operations. This special form is called the Row Echelon Form. We further
refine the procedure and introduce you to Row Reduced Echelon form of a
matrix. in Sec. 4.4, we discuss the method for solution of systems of linear
equations

Objectives
After studying this unit, you should be able to:
• explain the Gaussian Elimination procedure;

• carry out basic row operations on a given matrix;

• identify whether a given matrix is in Reduced Echelon form or not; 5

Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
• identify whether a given matrix is in Row Reduced Echelon form or not;

• reduce a matrix to Row Echelon form or Row Reduced Echelon form using
row operations;

• find the solution set of a system of linear equations;

4.2 GAUSSIAN ELIMINATION

In your high school, you have solved simultaneous equations. In this section,
we will discuss a method for solving simultaneous equations called Gaussian
elimination procedure, which is named after Gauss. We start with an example.

Example 1: Solve the set of equations

x1 + x2 + x3 = 3 …(1)
x2 − x3 = 0 …(2)
x3 = 1 …(3)

Solution: This is an example of a triangular system of equations which is

easy to solve. From Eqn. (3), we get x3 = 1. Substituting the value of x3 in
Eqn. (2), we get x2 = 1. Substituting the value of x3 and x2 in Eqn. (1), we get
x1 = 1.
∗∗∗

We saw that the system of equations in Example 1 is a triangular system of

equations and we solved it using the process of back substitution. We call
the method back substitution because we substituted back the value of x3 we
got from Eqn. (3) into Eqn. (2) to get the value of x2 . We then substituted back
the value of x2 and x3 in Eqn. (1) to get the value of x3 . More generally, in a
triangular system, the equation with fewer variables follows an equation with
more variables.

As you may be already aware, not all the systems of simultaneous linear
equations are triangular. However, the good news is that we can reduce any
system of linear equations to a triangular system. This method of reducing a
system of simultaneous linear equations to a triangular system and solving it is
due to Gauss, hence the name Gaussian elimination. In the next example, we
will see how to reduce an arbitrary system of linear equations to a triangular
system and solve it.

Example 2: Solve the following system of equations:

x1 + x2 + x3 = 4 …(4)
2x1 + 2x2 − x3 = 2 …(5)
2x1 + 4x2 − 2x3 = 2 …(6)

Solution: To eliminate x1 from the second equation, we multiply Eqn. (4) by 2

6 and subtract it from Eqn. (5). The second equation becomes −3x3 = −6. We get
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Systems
. . . . . . . . . .of
. . linear
. . . . . . .equations
..........
the following system of equations which are equivalent to Eqn. (4), Eqn. (5)
and Eqn. (6).

x1 + x2 + x3 = 4 …(7)
−3x3 = −6 …(8)
2x1 + 4x2 − 2x3 = 2 …(9)

By equivalent equations, we mean that both the sets of equations have the
same set of solutions. We will see why these two systems of equations are
equivalent later. So, instead of Eqn. (4), Eqn. (5) and Eqn. (6), we can solve
the equations Eqn. (7), Eqn. (8) and Eqn. (9). Note that, the second set of
equations is simpler than the first set because the second equation in the
second set has two less variables; x1 and x2 have coefficients 0.

In a triangular system of equations, the equation with fewer variables follows

an equation with more variable. Here Eqn. (8) has only one variable, but
Eqn. (9) has three variables. So, we interchange Eqn. (8) and Eqn. (9) to get

x1 + x2 + x3 = 4 …(10)
2x1 + 4x2 − 2x3 = 2 …(11)
−3x3 = −6 …(12)

We eliminate the variable x1 from the second equation; we multiply Eqn. (10)
by 2 and subract it from the second equation to get

x1 + x2 + x3 = 4 …(13)
2x2 − 4x3 = −6 …(14)
−3x3 = −6 …(15)

1
Next, we multiply the second equation by to get
2

x1 + x2 + x3 = 2 …(16)
x2 − 2x3 = −3 …(17)
−3x3 = −6 …(18)

1
Next, we multiply the third equation by − to get
3

x1 + x2 + x3 = 4 …(19)
x2 − 2x3 = −3 …(20)
x3 = 2 …(21)

Back substituting, we get x3 = 2, x2 = 1 and x1 = 1

∗∗∗

Before we proceed further, solve the following exercise to see if you have
understood the Gaussian elimination procedure.

7
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
E1) Solve the following set of simultaneous equations using Gaussian
elimination:
x1 + 2x2 + x3 = 7
2x1 + x2 − x3 = 5
3x1 − x2 − x3 = 3

In the next section, we will look at the operations we performed in the Gaussian
elimination procedure as matrix operations by writing a system of equations in
matrix form.

4.3 ROW OPERATIONS

In this section, We will introduce you to row operations. We will see how to
use row operations to solve linear equations.

We go back to Example 2. We have reproduced the equations here for ready

reference.

x1 + x2 + x3 = 4
2x1 + 2x2 − x3 = 2
2x1 + 4x2 − 2x3 = 2

As we have seen in Unit 1, we can write this as the matrix equation

1 1 1 x1 4
[2 2 −1] [x2 ] = [2]
2 4 −2 x3 2

Another way of representing the system is using an augmented matrix.

Suppose matrix equation is Ax = b where A is an m × n matrix. The augmented
matrix is an m × (n + 1) matrix. The first n columns are the columns of A and b
forms the (n + 1)th column of the augmented matrix. For example, the
augmented matrix representation of the system of equations Example 2 is

1 1 1 4
[2 2 −1 2] …(22)
2 4 −2 2

The first operation we carried out was to multiply the first equation by 2 and
subtract it from the second equation. Equivalently, we replace the second row
in Eqn. (22) by

Second row − 2 × First row

We denote this operation by R2 → R2 − 2R1 . In general, we can replace the iith

row by

8 ith row − c × jth row

Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Systems
. . . . . . . . . .of
. . linear
. . . . . . .equations
..........
where c is a constant. We denote this operation by Ri → Ri − c × Rj . We call
these operations row operations. Carrying out R2 → R2 − 2R1 , we get

1 1 1 4
[0 0 −3 −6]
2 4 −2 2

Next operation we performed was to interchange the second and third rows.
We denote this operation by R2 ↔ R3 . We get

1 1 1 4
[2 4 −2 2 ]
0 0 −3 −6

In general, we denote the operation of interchanging the rows i and j by

Ri ↔ Rj .

Next, we multiply the first equation by 2 and subtract it from the second
equation. Again, we denote the operation by R2 → R2 − 2R1 . The matrix form
of the equations is

1 1 1 4
[0 2 −4 −6]
0 0 −3 −6
1
Next, we multiply the second row by to get
2

1 −1 1 4
[0 1 −2 −3]
0 0 −3 −6
1
Next, we multiply the last row by − . We get
3

1 1 1 4
[0 1 −2 −3] …(23)
0 0 1 2

Bearing in mind that the first three columns correspond to first three variables
and the last column corresponds to right hand side, we can reconstruct the
equations.
x1 + x2 + x3 = 4
x2 − 2x3 = −3
x3 = 2
As before we can use back substitution to solve these equations.

You can have one doubt. How can we be sure that the row operations don’t
alter the solutions of the system of equations? One way could be to check that
the solutions we get for the modified set of equations are also solutions for the
original equations. (Do this for the examples and exercises we have done so
far!) Don’t you think it will be better to prove that, in general, the row operations
do not change the solution set of a system of equation? We will show this in
the next proposition. 9
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
Before that, we will set up some definitions that will help us in better formulation
as well as the proof of this result.

Definition 1: Let
a11 x1 + a12 x2 + ⋯ + a1n xn = b1 ⎫
⎪
a21 x2 + a22 x2 + ⋯ + a2n xn = b2 ⎪
…(∗)
⎬ ⋮
⎪
⎪
am1 x1 + an2 x2 + ⋯ + amn xn = bm ⎭

be a system of simultaneous linear equations.

We say that (a1 , a2 , … , an ) ∈ ℝn is a solution to the (*) if

ai1 a1 + ai2 a2 + … + ain an = bi (1 ≤ i ≤ m)

The set

{ (a1 , a2 , … , an ) ∈ ℝn | (a1 , a2 , … , an ) is a solution of Eqn. (∗)}

is called the solution set of Eqn. (∗).

In the next proposition we prove that the row operations we carry out for
solving a system of linear equations do not change the solution set of the
system of linear equations.

Proposition 1: Suppose the set of simultaneous equations

a′11 x1 + ⋯ + a′1n xn = b′1

a′21 x1 + ⋯ + a′2n xn = b′2
} …(24)
⋮ ⋮ ⋮ ⋮ ⋮
a′m1 x1 + ⋯ + a′mn xn = b′m

is obtained from the set of equations

a11 x1 + ⋯ + a1n xn = b1
a21 x1 + ⋯ + a2n xn = b2
} …(25)
⋮ ⋮ ⋮ ⋮ ⋮
am1 x1 + ⋯ + amn xn = bm

by performing one of the following operations.

1. Multiplying the ith equation by a constant a and adding it to the jth equation.

2. Interchanging two equations.

3. Multiplying the ith equation by a nonzero constant.

The, Eqn. (25) and Eqn. (24) have the same set of solutions.
You can skip this proof in
first reading and come
Proof: 1. It is clear that interchanging equations doesn’t affect the solution
back later after going
through the entire course set. For example, if (a1 , a2 , … , an ) is a solution to the second equation, it
10
once. will still be a solution to the equation even if it become the third equation.
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Systems
. . . . . . . . . .of
. . linear
. . . . . . .equations
..........
2. Suppose we mulitply the jth equation in Eqn. (25) by 𝜆, 𝜆 ≠ 0. The new set
ot equations will be

a11 x1 + ⋯ + a1n xn = b1 ⎫
⎪
a21 x1 + ⋯ + a2n xn = b2 ⎪
⎪
⎪
⎪
⋮ ⋮ ⋮ ⋮ ⋮
…(26)
𝜆ai1 x1 + ⋯ + 𝜆ain xn = 𝜆bi ⎬
⎪
⎪
⋮ ⋮ ⋮ ⋮ ⋮ ⎪⎪
⎪
am1 x1 + ⋯ + amn xn = bm ⎭

Only the ith equations of Eqn. (25) and Eqn. (26) are different. To check
that Eqn. (25) and Eqn. (26) have the same solution, we need to show
that any solution to the ith equation of Eqn. (25) is a solution to Eqn. (26)
and vice versa.
Suppose (a1 , … , an ) is a solution to the ith equation

ai1 x1 + ai2 x2 + ⋯ + ain xn = bi

in Eqn. (25); that is

ai1 a1 + ai2 a2 + ⋯ + ain an = bi

Multiplying both sides of the above equation by𝜆, we get

𝜆ai1 a1 + 𝜆ai2 a2 + ⋯ + 𝜆ain an = 𝜆bi

In other words, (a1 , … , an ) is a solution to the equation

𝜆ai1 x1 + 𝜆ai2 x2 + ⋯ + 𝜆ain xn = 𝜆bi

which is the ith equation in Eqn. (26).

Conversely, suppose that (a′1 , a′2 , … , a′n ) is a solution to

𝜆ai1 x1 + 𝜆ai2 x2 + ⋯ + 𝜆ain xn = 𝜆bi

where 𝜆 ≠ 0. We have
′ ′ ′
𝜆ai1 a1 + 𝜆ai2 a2 + ⋯ + 𝜆ain an = 𝜆bi

or
′ ′ ′
𝜆 (ai1 a1 + ai2 a2 + ⋯ + ain an − bi ) = 0

Since 𝜆 ≠ 0, we have

ai1 a′1 + ai2 a′2 + ⋯ + ain a′n = bi

In other words (a′1 , … , a′n ) is a solution to

ai1 x1 + ai2 x2 + ⋯ + ain xn = bi

3. Suppose we multiply the jth equation by c and add it to the ith equation.
We can always assume that i > j since we saw that reordering the 11
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
equations does not alter the solution set. So, the new set of equations is

a11 x1 + ⋯ + a1n xn = b1 ⎫
⎪
a21 x1 + ⋯ + a2n xn = b2 ⎪
⎪
⎪
⎪
⋮ ⋮ ⋮ ⋮ ⋮
…(27)
(ai1 + caj1 ) x1 + ⋯ + (ain + cajn ) xn = bi + cbj ⎬
⎪ ⎪
⎪
⋮ ⋮ ⋮ ⋮ ⋮ ⎪
⎪
am1 x1 + ⋯ + amn xn = bm ⎭

We need to check the following:

(a) If (a1 , a2 , … , an ) is a solution to system in Eqn. (25), then it is a

solution to the system in Eqn. (27).
(b) If (a′1 , … , a′n ) is a solution to the system in Eqn. (27), it is a solution to
the system in Eqn. (25).

Let us prove part (a). Let (a1 , … , an ) be a solution to the system in

Eqn. (25). Note that, except for the ith equation, the rest of the equations
in the two systems Eqn. (25) and Eqn. (27) are the same. So, we just
need to check that (a′1 , … , a′n ) is a solution to the ith in Eqn. (27). We have

(ai1 + caj1 ) + ⋯ + (ain + cajn ) − (bi + cbj )

= (ai1 a1 + ai2 a2 + ⋯ + ain an − bi ) + c (aj1 a1 + aj2 a2 + ⋯ + ajn an − bj ) = 0

since (a1 , … , an ) satisfies the equations

ai1 x1 + ⋯ + ain = bi

and

aj1 x1 + ⋯ + ajn xn = bj

which are, respectively, the ith and jth equations of Eqn. (25).
Let us prove part (b). Suppose that (a′1 , … , a′n ) is a solution to the system
in Eqn. (27). In particular, (a′1 , … , a′n ) is a solution to the ith equation in the
system Eqn. (27). In other words,

(ai1 a′1 + ai2 a′2 + ⋯ + ain a′n − bi ) + c (aj1 a′1 + aj2 a′2 + ⋯ + ajn a′n − bj ) = 0 …(28)

Since (a′1 , … , a′n ) is a solution to the jth equation of the system in Eqn. (27),
that is

aj1 a′1 + aj2 a′2 + ⋯ + ajn a′n − bj = 0

from Eqn. (28) it follows that

ai1 a′1 + ai2 a′2 + ⋯ + ain a′n − bi = 0

From Proposition 1 it follows that the row operation we carry out to a matrix
12 representing a system of equations doesn’t alter the solution set of the system.
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Systems
. . . . . . . . . .of
. . linear
. . . . . . .equations
..........
Another question that you may have is the following: When we solve a system
of equations using Gaussian elimination, we know when we reach our goal of
reducing the set of equations to a triangular equation. When we write the
equations in the form of matrix equations, how can we find out that we have
reduced the equations to triangular form?

To understand this, let us take a close look at the matrix in Eqn. (23). See
Fig. 1.

1 1 1 2
⎡ ⎤
⎢
⎢ 0 1 −2 −3 ⎥
⎥
⎢ ⎥
⎢ ⎥
0 0 1 2
⎣ ⎦

Fig. 1: Example of Row Echelon Form

It has the following properties:

1. The first nonzero entry in each row, from the left, is one and all entries
below this nonzero entry in the column containing the entry is zero. We
have circled the first nonzero entry in the first row.

2. The first nonzero entry in each row, from the left, is to the right of the first
nonzero entry in the previous row.

So, we can stop our row operations and solve the system of equations using
back substitution once the matrix representing the system of equations is of the
above form. The matrices that have the above form are supposed in Row
Echelon Form. Echelon means ‘steps’. You can see in Fig. 1 that the first
nonzero elements in each row form a step like structure.

Let us now formally define the Row Echelon Form of a matrix.

Definition 2: We say that a matrix is in Row Echelon Form if the following

holds.

1. The first non zero entry in each row is to the right of the first non zero
entry in the previous row. We call the first non zero entry in each row the
leading entry of that row.

2. All entries below the leading entry in the column containing the leading entry are
zero. The position occupied by the leading entry is called the pivot
position. The value of the leading entry is called the pivot and the pivot
should be one. The column containing the leading entry is called the
pivot column.

3. A row with all the entries zero is called a zero row. All the zero rows occur
after all the nonzero rows. 13
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
For example, in Fig. 1, the first, second and third columns are pivot columns.
The pivots in first, second and third column are, respectively, 1, 2 and −3,
respectively.

Let us now look at an example to consolidate your understanding of REF.

Example 3: Which of the following are in REF? Justify your answer. For the
matrices which are in REF,identify the pivots.
1 2 1 0 0 1 −1 2 1 0
0 0 0 0 0 0 1 0 0 1
i) [ ] ii) [ ]
0 2 1 0 0 0 0 0 1 1
0 0 1 0 0 0 0 1 0 0
1 2 1 2 0 1 0 0 2
0 1 1 0 0 0 1 0 3
iii) [ ] iv) [ ]
0 −1 1 0 0 0 0 1 0
0 0 1 0 0 0 0 0 0

Solution:

i) The pivot position is in the first row, first column of the matrix. The pivot is
1. All the entries below the pivot position is zero.
The second rows is a zero row and there are nonzero rows after this row.
This is not in REF because the row of all zeros is appearing before
nonzero rows.

ii) First row, first column is the pivot position of the first row and the pivot is
1. All the entries below the pivot position are zero.
The pivot position in the second row should be in the second, third, fourth
or fifth column. The pivot position is in the second row, second column
and the pivot is 1. All the entries below the pivot position is zero.
The pivot position in the third row is in the fourth column. So, the pivot
position in the fourth row must be in the fifth column, but it is in the third
column.
This matrix is not in REF.

iii) The first nonzero entry in the first row is in the first column and all the
other entries below this entry in the first column are 0.
The pivot position in the second row is in second column and the pivot is
1. The entry below the pivot position is −1 and not zero. This is not in REF.

iv) The pivot position in the first row is in the second column and the pivot is
1. All the entries below the pivot position is zero.
The pivot position in the second row has to be in the third, fourth or fifth
column. The pivot position is in the third column and the pivot is 1. All the
entries below the pivot position are 0.
The pivot position in the third row, which should appear in the fourth or
fifth column is in the fourth column. The pivot is one. Further, all the
14 entries below the pivot position is zero.
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Systems
. . . . . . . . . .of
. . linear
. . . . . . .equations
..........
The zero row appears below all the nonzero rows. This is in REF.

0 1 0 0 2
0 0 1 0 3
[ ]
0 0 0 1 0
0 0 0 0 0

∗∗∗

Try the following exercise now to check your understanding of Example 3.

E2) Which of the following matrices are in REF? Justify your answer.
0 1 3 1 0 1 1 0 0 1
0 0 1 1 0 0 1 2 1 0
i) [ ] ii) [ ]
0 0 0 1 0 0 0 0 0 0
0 0 0 0 1 0 0 0 1 0
1 0 1 2 0 1 1 0 0 2
0 1 1 −2 1 0 0 1 0 0
iii) [ ] iv) [ ]
0 0 0 0 1 0 0 0 0 0
0 0 0 1 0 0 0 1 0 0

Let us now look at an example to understand the procedure for reducing a

matrix to Row Echelon Form.

Example 4: Reduce the following matrix to Row Echelon form:

1 2 1 8
[1 −1 1 2]
1 1 3 8

Solution: The left most nonzero column is the first column. The pivot position
in the first row is in the first column. We now carry out row operations
R2 → R2 − R1 , R3 → R3 − R1 , to get

1 2 1 8
[0 −3 0 −6]
0 −1 2 0

All the other entries below the pivot are zero.

We now look for the first nonzero entry in the second row. This is in the second
row, third column which is to the right of the previous pivot column, the first
column. (We have already found a pivot in the first row.) The second column
1
has the nonzero entry −3. Carrying out the row operation R2 → − R2 gives
3

1 2 1 8
[0 1 0 2]
0 −1 2 0 15
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
Carrying out the row operations R3 → R3 + R2 , gives

1 2 1 8
[0 1 0 2]
0 0 2 2

We now look for the first nonzero column which is to the right of the previous
pivot column, the second column, and has a nonzero entry in the third row. The
third column has the nonzero entry 2 in the third row. We divide the row by 2 to
get

1 2 1 8
[0 1 0 2]
0 0 1 1

There are no entries below the pivot entry. So, there is nothing more to do in
this row and there are no more rows left. The matrix is in Row Echelon Form.

∗∗∗

Let us now look at a bigger example.

Example 5: Reduce the following matrix to REF.

1 1 0 1 2
[2 2 −1 1 4]
1 1 0 1 1

Solution: The firt nonzero entry in the first row is in the first column. The
pivot is 1. The row operation R2 → R2 + (−2)R1 gives,

1 1 0 1 2
[0 0 −1 −1 0]
1 1 0 1 1

The row operation R3 → R3 + (−1)R1 gives

1 1 0 1 2
[0 0 −1 −1 0 ]
0 0 0 0 −1

Now, all the entries in the pivot in the first row are zero. We move on to the
second row. The first nonzero entry in the second row is in the third column
and it is −1. This is to the right of the first column which contains the pivot for
the first row. We carry out the row operation R2 → (−1)R2 . We get

1 1 0 1 2
[0 0 1 1 0]
0 0 0 0 −1

The entry below 1 is already zero. So, there is nothing to do.

1 1 0 1 2
[0 0 1 1 0]
16 0 0 0 0 −1
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Systems
. . . . . . . . . .of
. . linear
. . . . . . .equations
..........
We now look for nonzero column to the right of the third column which a
nonzero entry in the third row. The fourth column is nonzero, but the entry in
the third row fourth column is zero. However, the entry in the fifth column, third
row is −1 so we choose this column as the next pivot column. Carrying out row
operation R3 → (−1)R3 , we get

1 1 0 1 2
[0 0 1 1 0]
0 0 0 0 1

As there are no more rows left, this matrix is in the REF.

∗∗∗

Try the following exercise to check whether you have understood the method
for reducing a matrix to REF.

E3) Reduce the following matrix to REF.

1 1 1 3
[−1 1 3 4]
2 2 1 5

Let us go back to Eqn. (19), Eqn. (20) and Eqn. (21)

x1 − x2 + x3 = 4
x2 − 2x3 = −3
x3 = 2

Our plan now is to reduce these equations to three equations in which the first
equation involves only x1 , the second equation involves only x2 and the third
equation involves only x3 .

Multiplying third equation by 2 and adding to second equation, we get

x1 + x2 + x3 = 4 …(29)
x2 =1 …(30)
x3 = 2 …(31)

Subtracting Eqn. (31) from Eqn. (29), we get

x1 + x2 =2 …(32)
x2 =1 …(33)
x3 = 2 …(34)

Subtracting Eqn. (33) from Eqn. (32) we get

x1 =1
x2 =1

x3 = 2 17
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
Again the matrix form of these equations is

1 0 0 1
[0 1 0 1]
0 0 1 2

Again, the matrix is special form. Along with conditions for REF, the matrix
satisfies the following additional condition the in the pivot column all the entries
other than the pivot are zero.

This leads us to the following definitio,

Definition 3: An m × n matrix is in Row Reduced Echelon Form if is in row

echelon form and in the pivot column all the entries other than the pivot are
zero.

The next proposition says that the Row Reduced Echelon Form of matrix is
unique.

Proposition 2: The Row Reduced Echelon Form of an m × n matrix is unique.

Example 6: Which of the following matrices are in Row Reduced Echeton

Form. Justify your answer.
1 1 0 1 0
1 0 0 2 1
0 0 1 0 0
i) [0 1 2 2 0] ii) [ ]
0 0 0 1 0
0 0 −1 0 0
0 0 0 0 1
1 0 1 0 0
0 1 0 0 1
0 1 0 0 1
iii) [0 0 2 1 0] iv) [ ]
0 0 0 1 0
0 0 0 0 1
0 0 0 0 0

Solution:
i) The pivot in the first row is 1 and it is in the first column. The first nonzero
entry in the second row is 1. This should be in second, third, fourth or fifth
column and it is in second column. Also, all the entries in this column other
than this are not zero.
However, the pivot in the third row is −1 and not 1. So, this matrix is not in
RREF.
ii) The pivot in the first row is one and it is in the first column and all the
remaining entries in the column are zero.
The pivot in the second row should be in the second, third, fourth or fifth
column. It is one and is in the third column. All the other entries in the third
column, the pivot column, are zero.
The pivot in the third row must be in the fourth of fifth column, and it is in
the fourth column. However, the entry in the first row of the fourth column is
not zero. So this matrix is not in RREF.
iii) The pivot in the first row is 1 and it is in the second column and all the other
18 entries in that column are zero.
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Systems
. . . . . . . . . .of
. . linear
. . . . . . .equations
..........
The pivot in the second row should be in the third, fourth, or fifth column. It
1 and it is in the third column. However, the pivot is not 1. So the matrix is
not in RREF.
iv) The pivot in the first row is 1 and it is in the first column, and all the other
entries in that column are zero.
The pivot in the second row should be in second, third, fourth or fifth
column. It is in the second column and it is one. All the other entries in the
second column are zero.
The pivot in the third row should be one and it should be in the third, fourth
or fifth column. It is one and it is in the fourth column.
All the entries in the fourth row are zero and it appears after all nonzero
rows. Since all the conditions are satisfied the matrix is in RREF.
∗∗∗

Here is an exercise for you to check your understanding of the above example.

E4) Check whether the following matrix is in RREF:

1 1 0 0 0 2
0 0 1 0 0 1
[ ]
0 0 0 0 1 0
0 0 0 0 0 0

In all the systems of equations we considered, the number of variables were

equal to the number of solutions. In the next section we expand our study to
more general systems of equations. We will see how to solve them using row
reduction.

4.4 SYSTEMS OF LINEAR EQUATIONS

In this section, we take up the study of simultaneous linear equations. First,
we study systems of linear homogeneous equations. These are system of
linear equations in which the RHS is zero.

A homogeneous equation has always one solution, namely (0, 0, … , 0) ∈ ℝn .

(See Definition 1) What other solutions can it have? We will answer this
equation in this section. The main tool for this is RREF of the matrix
corresponding to the equations.

Let us now look at an example.

Example 7:

x1 + x2 + x3 + x4 = 0
x1 + 2x2 + x3 + x4 + x5 = 0
2x1 + x2 + 2x3 + 2x4 − x5 = 0
x1 + 6x2 + 2x3 + 2x4 + 4x5 = 0
19
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
Solution: The augmented matrix form is

1 1 1 1 0 0
1 2 1 1 1 0
[ ]
2 2 2 2 −1 0
2 6 2 2 4 0

The first nonzero entry is in the first column and choose this as the pivot entry.
R2 → R2 − R1 gives

1 1 1 1 0 0
0 1 0 0 1 0
[ ]
2 1 2 2 −1 0
2 6 2 2 4 0

R3 → R3 − 2R1 gives

1 1 1 1 0 0
0 1 0 0 1 0
[ ]
0 −1 0 0 −1 0
2 6 2 2 4 0

R4 → R4 − 2R1 gives

1 1 1 1 0 0
0 1 0 0 1 0
[ ]
0 −1 0 0 −1 0
0 4 0 0 4 0

We look for the pivot entry in the second column in the second, third and fourth
rows. The entry in the second row is nonzero and we choose this as the pivot.
R1 → R1 − R2 gives

1 0 1 1 −1 0
0 1 0 0 1 0
[ ]
0 −1 0 0 −1 0
0 4 0 0 4 0

R3 → R3 + R2 gives

1 0 1 1 −1 0
0 1 0 0 1 0
[ ]
0 0 0 0 0 0
0 4 0 0 4 0

R4 → R4 − R2 gives

1 0 1 1 −1 0
0 1 0 0 1 0
[ ]
0 0 0 0 0 0
0 0 0 0 0 0

This matrix is in Row Reduced Echelon Form. We see that the first and second
20 columns have pivots and there are no pivots in the other columns. We choose
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Systems
. . . . . . . . . .of
. . linear
. . . . . . .equations
..........
the variables corresponding to the columns without pivots to be free variables.
Writing as equations, we get

x1 = −x3 − x4 + x5
x2 = −x5

Writing x3 = 𝜆1 , x4 = 𝜆2 , x5 = 𝜆3 the solutions are

(x1 , x2 , x3 , x4 , x5 ) = (−𝜆1 − 𝜆2 + 𝜆3 , −𝜆3 , 𝜆1 , 𝜆2 , 𝜆3 )

So, the solution set is { (−𝜆1 − 𝜆2 + 𝜆3 , −𝜆3 , 𝜆2 , 𝜆3 )| 𝜆1 , 𝜆2 , 𝜆3 ∈ ℝ}

Writing the elements of the solution set as column vectors, we have

−𝜆1 − 𝜆2 + 𝜆3 −1 −1
1
⎡ −𝜆3 ⎤ ⎡ 0⎤ ⎡ 0⎤ ⎡−1⎤
⎢
⎢ ⎥
⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢ 𝜆1 ⎥ ⎢
= 𝜆1 ⎢ 1 ⎥ + 𝜆2 ⎢ 0 ⎥ + 𝜆3 ⎢ 0 ⎥
⎥ ⎢ ⎥ ⎢
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎥
𝜆2 0 1 0
⎣ 𝜆3 ⎦ ⎣0⎦ ⎣0⎦ ⎣1⎦

So, we see that the solution set is [S] where

−1 −1 1 ⎫
⎧
⎪ ⎪
⎪⎡ 0 ⎤ ⎡ 0 ⎤ ⎡−1⎤⎪
⎪ ⎪
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
S = ⎢⎢ 1 ⎥⎥ , ⎢⎢ 0 ⎥⎥ , ⎢⎢ 0 ⎥⎥
⎨ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎬
⎪
⎪ 0 1 0 ⎪ ⎪
⎪ ⎪
⎩⎣ 0 ⎦ ⎣ 0 ⎦ ⎣ 1 ⎦⎭

∗∗∗

We now explain the general method of solving a system of homogeneous

equations. We write the matrix corresponding to the equation and reduce to
Row Reduced Echelon Form.

We then designate as free variables the variables corresponding to these

columns, that do not contain a pivot. We then solve for the variables that
correspond to those columns in terms of the free variables.

Let us look at one more example:

Example 8: Solve the system of equations

x1 + x2 + x3 + 3x4 = 0
x1 + x2 + x3 + 5x4 = 0
x1 + x2 + x3 + 7x4 = 0

Solution: The agumented matrix form of the equations is

1 1 1 3 0
[1 1 2 5 0]
1 1 3 7 0 21
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
In the first column, the entry in the first row is 1 which is nonzero and choose
this as the pivot entry. R2 → R2 − R1 gives

1 1 1 3 0
[0 0 1 2 0]
1 1 3 7 0

R3 → R3 − R1 gives

1 1 1 3 0
[0 0 1 2 0]
0 0 2 4 0

We now look for a pivot in the second column. We cannot choose the entry in
the first row, second column as the pivot and there are no other nonzero entries
in the second column. So, we look for the pivot in the third column. The entry in
the second row, third column is nonzero and we choose this as the pivot entry.

R3 → R3 − 2ℝ2 gives

1 1 1 3 0
[0 0 1 2 0]
0 0 0 0 0

R1 → R1 − R2 gives

1 1 0 1
[0 0 1 2]
0 0 0 0

The matrix is in RREF. The second and fourth columns don’t have pivots, so
the free variables are x2 and x4 . We get the equations

x1 + x3 + x4 = 0
x3 + 2x4 = 0

or

x1 = −x2 − x4
x3 = −2x4

We set x2 = 𝜆1 and x4 = 𝜆2 . We have

(x1 , x2 , x3 , x4 ) = (−𝜆1 − 𝜆2 , 𝜆1 , −2𝜆2 , 𝜆2 )

So, the solution set is

{ (−𝜆1 − 𝜆2 , 𝜆1 , −2𝜆2 , 𝜆2 )| 𝜆1 , 𝜆2 ∈ ℝ}

So, writing in the form of a column vector, we can write every solution in the
form
−1 −1
1 0
𝜆1 [ ] + 𝜆2 [ ]
0 −2
22 0 1
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Systems
. . . . . . . . . .of
. . linear
. . . . . . .equations
..........
As before, the solution set is S where

−1 −1
1 0
S = {[ ] , [ ]}
0 −2
0 1

∗∗∗

Proposition 3: If the number equations is less than the number of variables, a

homogeneous system of equations will have infinitely many solutions.

Proof: The RREF of the matrix associated to the system can have only as
many columns with pivots as the number of rows in the matrix form of the
equations. This is because every row has at most one pivot position. So, the
number of pivot postions cannot be greater than the number of rows. But, the
number of rows in the matrix of the homogeneous system is the number of
equations in the system and the number of columns is the number of variables
in the system. If the number of equations is less than the number of variables,
the matrix associated with the system will have more columns than rows. So,
there will be at least one non-pivot column. This means there is at least one
free variable in the solution. Giving different values to the free variable, we get
infinitely many solutions to the system. ■

Try the following exercise to check your understanding of solutions of

homogeneous equations.

E5) Solve the following system of liner equations.

x1 + x2 + 2x4 = 0
2x1 + x2 + x4 = 0
x1 + x2 + x3 − x4 = 0

We will now discuss the solutions of nonhomogeneous equations, equations in

which the RHS is nonzero. Let us look an example.

Example 9: Solve the following system of equations using RREF:

x1 + 2x2 + x3 + 2x4 = 2
x1 + 3x2 + x3 + 3x4 = 3
−x1 + x2 − x3 + x4 = 2

Solution: The augmented matrix form is

1 2 1 2 2
[ 1 3 1 3 3]
−1 1 −1 1 2 23
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
The operations R2 → R2 − R1 , R3 → R3 + R1 , R1 → R1 − 2R2 and
R3 → R3 − 3R2 reduces the matrix to following RREF

1 0 1 0 0
[0 1 0 1 1]
0 0 0 0 0

(Check this!) There are two non-pivot columns (other than the column
corresponding to the RHS) corresponding to variables x3 and x4 . We choose
them as free variables. Reverting to equations, the system is

x1 + x3 =0
x2 + x4 = 1

Setting x3 = 𝜆1 and x4 = 𝜆2 , we get x1 = −𝜆1 and x2 = 1 − 𝜆2 . So, the solution is

(x1 , x2 , x3 , x4 ) = (−𝜆1 , 1 − 𝜆2 , 𝜆1 , 𝜆2 )

Writing as column vectors,

0 −1 0
0 −1 1
(−𝜆1 , 1 − 𝜆2 , 𝜆1 , 𝜆2 ) = 𝜆1 [ ] + 𝜆2 [ ] + [ ]
1 0 0
0 1 0

where 𝜆1 and 𝜆2 vary over ℝ. Let us write

−1 0
0 −1
u1 = [ ] , u2 = [ ] , S = {u1 , u2 } , W = [S]
1 0
0 1

Then, W is a subspace of ℝ4 and it is the solution set of the system

x1 + 2x2 + x3 + 2x4 = 0
x1 + 3x2 + x3 + 3x4 = 0
−x1 + x2 − x3 + x4 = 0

which is the corresponding homogeneous system of equations.

Further, the solution set our nonhomogeneous equation is the coset v + W

0
1
where v = [ ].
0
0
∗∗∗

We saw in the previous example that the solution set of a system of linear
equations is of the form w + v where w is a solution of the homogeneous
system of equation and v is a particular solution of the system. This true in
24 general. The solution of a linear system of equations (if it exists) is of the form
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Systems
. . . . . . . . . .of
. . linear
. . . . . . .equations
..........
w + v where w is a solution to the corresponding homogeneous system and v is
any solution to the given system. In particular if the subspace W of the system
of corresponding homogeneous system is not the zero space, if the system has
one solution, it has infinitely many solutions. If W = {0}, the system has a
unique solution. Try the following exercise now.

E6) Find the solution set of the following system of equations:

2x1 + 2x2 + 2x3 − 2x4 = 3

2x1 + 3x2 + 2x3 =3
x1 + 2x2 + x3 − x4 = 1

We will now look at example of an inconsistent system of equations, i.e. a

system of equations that has no solutions.

Example 10: Check whether the following system of equations is consistent:

x1 − x2 + x3 − x4 = −1
−x1 + 2x2 − x3 + 2x4 = 2
2x1 − x2 + 2x3 − x4 = 0

Solution: The augmented matrix form of the equation is

1 −1 1 −1 −1
[−1 2 −1 2 2 ]
2 −1 2 −1 0

The operations R2 → R2 + R1 , R3 → R3 − 2R1 , R1 → R1 + R2 , and

R3 → R3 − R2 reduce matrix to the following RREF:

1 0 1 0 0
[0 1 0 1 1]
0 0 0 0 1

Reverting to equations, we get

x1 + x3 =0 …(35)
x2 + x4 = 1 …(36)
0x1 + 0x2 + 0x3 + 0x4 = 1 …(37)

As you can see easily, this system has no solutions because the LHS of
Eqn. (37) will always be zero for any value of x1 , x2 , x3 and x4 while the RHS is
nonzero.

∗∗∗

What we have seen in Example 10 is true in general. In the RREF of a system

of equations, if there is a row in which all the entries are zero except the entry
in the last column and entry in the last column is nonzero, the system of
equations is inconsistent. 25
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
Proposition 4: A system of equations is consistent if and only if the RREF of
the associated augmented matrix has no row of the form (0 ⋯ 0 bk ) with bk ≠ 0.

Proof: If such a row is present, it represents the equation 0x1 + ⋯ + 0xn = bk

which has no solution.

Suppose no such row is present. By reordering the variable if necessary by

choosing the variables corresponding to the k pivot columns as the first k
variables, we can assume that the RREF of the augmented matrix of the
system has the form

1 r1,k+1 ⋯ r1,n b1
⎡ ⋱ ⋮ ⋮ ⋮ ⋮⎤
⎢
⎢ ⎥
⎢ 1 rk,k+1 ⋯ rk,n bk ⎥⎥
⎢ ⎥

⎣ ⎦

By giving any values to the free variables xk+1 , xk+2 , xk+1 , xk+2 , …, xn , if there is at
least one free variable, we can get infinitely many solutions to the system. If
there are no free variables, i.e. k = n, the system has an unique solution
x1 = b1 , x2 = b2 , …, xk = bk . In any case there is always a solution and the
system is consistent. ■

Quite often, Here is an exercise for you to try.

E7) Check whether the following system of equations is consistent:

x1 + x3 = −1
x1 + x2 + x3 + x4 = 2
2x1 + x2 + 2x3 + x4 = 0

Note that, while testing consistency of a system of equations, we need not

reduce a matrix completely to RREF. In the intermediate stage, if there is a row
in which all the entries in a row, except the entry in the last column, are zero
and the entry in the last column is nonzero we can stop the process and
declare the system as inconsistent. Here is an example that illustrates this.

Example 11: Check the following system of equations for consistency:

x1 − x2 + 2x3 = 1
2x1 + x2 + x3 = 5
x1 + x2 =4

Solution: The associated augmented matrix is

1 −1 2 1
[2 1 1 5]
26 1 1 0 1
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Systems
. . . . . . . . . .of
. . linear
. . . . . . .equations
..........
R2
Carrying out row operations R2 → R2 − 2R1 , R3 → R3 − R1 , R2 → ,
3
R1 → R1 + R2 and R3 → R3 − 3R2 , we get

1 0 1 2
[0 1 −1 1]
0 0 0 1

Note that the matrix is not in RREF because the entries above the pivot entry in
the fourth row are not zero. However, we can stop here because the last row
has zeros in the first three columns and one in the last column. If we translate
this to an equation, we get the equation

0x1 + 0x2 + 0x3 = 1

and this equation has no solutions. So, the system of equations is inconsistent.

∗∗∗

We saw that the solution set of the system in Example 7 is spanned by three
vectors. Can we span the solution set with just two vectors? How about just a
single vector? The answer is ‘no’. In fact, the spanning vectors of the solution
set is minimal in the sense that there can’t be a smaller spanning set of
vectors. In the next Unit, we will develop the concepts that will let us prove this
fact. vector space and the dimension of a vector space.

4.5 SUMMARY
In this unit, we

1. explained the Gaussian Elimination procedure;

2. saw how to carry out basic row operations on a given matrix;

3. saw how to identify whether a given matrix is in Reduced Echelon form or

not;

4. explained how to identify whether a given matrix is in Row Reduced

Echelon form or not;

5. explained how to reduce a matrix to Row Echelon form or Row Reduced

Echelon form using row operations;

6. how to find the solution set of a system of linear equations using row
reduction;

4.6 SOLUTIONS/ANSWERS
27
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
E1) We number the equations for convenience.

x1 + 2x2 + x3 = 7 …(38)
2x1 + x2 − x3 = 5 …(39)
3x1 − x2 − x3 = 3 …(40)

Multiplying Eqn. (38) by 2 and subtracting from Eqn. (39) we get the
following new set of equations:

x1 + 2x2 + x3 = 7 …(41)
− 3x2 − 3x3 = −9 …(42)
3x1 − x2 − x3 = 3 …(43)

Multiplying Eqn. (41) by three and subtracting from Eqn. (43), we get the
following new set of equations:

x1 + 2x2 + x3 = 7 …(44)
− 3x2 − 3x3 = −9 …(45)
− 7x2 − 4x3 = −18 …(46)

We divide Eqn. (45) by −3 to get the following set of equations:

x1 + 2x2 + x3 = 7 …(47)
x2 + x3 = 3 …(48)
− 7x2 − 4x3 = −18 …(49)

Mutltiplying Eqn. (48) by −7 and subtracting from Eqn. (48), we get

x1 + 2x2 + x3 = 7 …(50)
x2 + x3 = 3 …(51)
3x3 = 3 …(52)

1
Multiplying Eqn. (52) by , we get
3

x1 + 2x2 + x3 = 7 …(53)
x2 + x3 = 3 …(54)
x3 = 1 …(55)

Back substituting, we get x1 = 2, x2 = 2 and x3 = 1.

E2) i) The pivot in the first row appears in the second column and the pivot
is 1. All the entries in the column below the pivot position are zero.
The pivot in the second row, which should be in third, fourth or fifth
column, appears in the third column and the pivot is 1. All the other
entries in the third column below the pivot position are zero.
The pivot in the third row should appear in fourth or fifth column and it
appears in fourth column and the pivot is 1. All the other entries in the
28 fourth column below the pivot position are zero.
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Systems
. . . . . . . . . .of
. . linear
. . . . . . .equations
..........
The pivot in the fourth row should be in the fifth column and it appears
in the fifth column and the pivot is 1.
0 1 3 1 0
0 0 1 1 0
[ ]
0 0 0 1 0
0 0 0 0 1
This matrix is in REF.
ii) The pivot in the first row appears in the first column and all the other
entries in that column are zero.
The pivot in the second row must be in second, third, fourth or fifth
column and it appears in the second column. All the entries in the
second column below the pivot position are zero.
However, the third row is a zero row and it appears before the fourth
row which is not a zero row.
This is not in REF.
iii) The pivotis 1 and it is in the first row, first column and all the other
entries blow the below the pivot position in the first column are zero.
The pivot in the second row should be in the second, third, fourth
column or fifth column and it is the second column. The pivot is one
and all the entries below the pivot are zero.
The pivot in the third row should be in the third, fourth or the fifth
column, and it in the fifth column. The pivot is 1 and all the entries
below the pivot are zero.
The fourth row should be a row of zeros since the leading pivot, if it
exists, should be in the sixth column and the matrix has only five
columns. However there is a nonzero entry in the fourth column.
So, the matrix is not in REF.
iv) The leading entry in the first row is in first column, and all the entries
below the leading entry are zero.
In the second row the first nonzero entry is in the third column.
However, the entry in the fifth row, third column is not zero.
So, the matrix is not in REF.

E3) The first column is a nonzero column with a nonzero entry, 1, in the first
row. We choose the first column as the pivot column and the entry in the
first row as the pivot. Carrying out row operations R2 → R2 + R1 gives
1 1 1 3
[0 2 4 7]
2 2 1 5
Carrying out row operation R3 → R3 − 2R1 gives

E4) The pivot in the first row is 1 and it is in the first column, and all the other
entries in that column are zero.
The pivot in the second row should be in second, third, fourth or fifth
column. It is in the third column and it is one. All the other entries in the
second column are zero. 29
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
The pivot in the third row should be one and it should be in the third,
fourth or fifth column. It is one and it is in the fifth column.
All the entries in the fourth row are zero and it appears after all nonzero
rows. Since all the conditions are satisfied the matrix is in RREF.

E5) The matrix representation of the equations is

1 1 0 2
[2 1 0 1 ]
1 1 1 −1

ℝ2 → R2 R1 gives

1 1 0 2
[0 −1 0 −3]
1 1 1 −1

ℝ3 → R3 − R1 gives

1 1 0 2
[0 −1 0 −3]
0 0 1 −3

ℝ2 → (−1)R2 gives

1 1 0 2
[0 1 0 3 ]
0 0 1 −3

ℝ1 → R1 − R2 gives

0 0 0 −3
[0 1 0 3 ]
0 0 1 −3

The matrix is in RREF.

The corresponding equations are

x1 − 3x4 = 0
x2 + 3x4 = 0
x3 − 3x4 = 0

We set x4 = 𝜆.
Then we get

x1 = −3𝜆, x2 = −3𝜆, x3 = 3𝜆.

So, the solution set {( 3𝜆, −3𝜆, 3𝜆, 𝜆)| 𝜆 ∈ ℝ4 }.

E6) The matrix form is

2 2 2 −1 3
[2 1 2 0 3 ]
30 1 2 1 −1 1
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Systems
. . . . . . . . . .of
. . linear
. . . . . . .equations
..........
The operations R2 → R2 − 2R1 , R3 → R3 − R1 , R2 → −R2 , R1 → R1 − R2 ,
R
R3 → R3 − R2 , R3 → 2R3 , R1 → R1 − 3 , R2 → R2 + R3 reduces the matrix
2
to the following RREF

1 0 1 0 2
[0 1 0 0 −1]
0 0 0 1 −1

There is only one non-pivot column, the third column. Reverting to

equations, we get

x1 + x3 = 2
x2 = −1
x4 = −1

Setting x3 = 𝜆, the solution is

(x1 = 2 − 𝜆, x2 = −1, x3 = 𝜆, x4 = −1) = (2, −1, 0, −1) + 𝜆(−1, 0, 1, 0)

. Let

−1 2
0 −1
u = [ ],v = [ ],
1 0
0 −1

As before, the solution set is v + W where

W = {𝜆u ∣ 𝜆 ∈ ℝ}

E7) The augmented matrix form of the equation is

1 0 1 0 −1
[1 1 1 1 2 ]
2 1 2 1 0

Carrying out the operation R2 → R2 + R1 , we get

1 −1 1 −1 −1
[0 1 0 1 1]
2 −1 2 −1 0

Carrying out the operations R3 → R3 − 2R1 , R1 → R1 + R2 , R3 → R3 − R2 ,

we get

1 0 1 0 0
[ 0 1 0 1 1]
0 0 0 0 1

In the last row, all the entries except the right most entry is zero and the
right most entry is 1. So, the system doesn’t have a solution.

31
 UNIT 5

BASES AND DIMENSION

Structure
Page Nos.
5.1 Introduction 32
Objective
5.2 Linear Independence 33
Some Elementary results on Linear Independence
5.3 Basis and Dimension 39
Dimension
5.4 Dimensions of Some Subspaces. 48
5.5 Dimension of a Quotient Space. 54
5.6 Summary 56
5.7 Solutions/Answers 57

5.1 INTRODUCTION
In this Block we continue our study of Linear Algebra with a discussion of
solution of simultaneous equations. The problem of solution of linear equations
efficiently is one of the fundamental problems of Mathematics. Efficient
methods of solution of systems of linear equations has many applications,
including in new areas like Data Science.

In Sec. 5.2 we discuss the concept of linear independence of vectors and the
dimension of a vector space. In Sec. 5.3, we introduce you to two of the
important concepts in Linear Algebra, the concepts of basis and dimension of a
vector space. In Sec. 5.4, we will determine the dimensions of some
subspaces. In Sec. 5.5, we will determine the dimension of quotient space.
Objectives
After studying this unit, you should be able to:

• define a linearly independent set and give examples of linearly independent

and linearly dependent sets of vectors;

• determine whether a give set of vectors is linearly independent or not;

• define a basis for a vector space and give examples;

32 • determine whether a set forms a basis for a vector space or not;

Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Bases
. . . . . . . and
. . . . .Dimension
...........
• determine bases for well known vector spaces;

5.2 LINEAR INDEPENDENCE

Consider the two vectors (1, 0) and (0, 1). You can see that
𝛼 (1, 0) + 𝛽 (0, 1) = (0, 0) implies that 𝛼 = 0 and 𝛽 = 0 (where 𝛼, 𝛽 ∈ ℝ). We say
that {(1, 0) , (0, 1)} is a linearly independent subset of ℝ2 . Let us now define the
concept of a linearly independent set formally.

Definition 1: Let V be a vector space over a field F and let S = {v1 , v2 , … , vn } be

a subset of V. Then we say that S is a linearly independent set if,

𝛼1 v1 + 𝛼2 v2 + ⋯ + 𝛼n vn = 0

for 𝛼i ∈ F, 1 ≤ i ≤ n, then 𝛼i = 0.

If S is a subset of V, not necessarily finite, we say that S is a linearly

independent set if every finite, non-empty, subset of S is linearly independent.

A set is linearly dependent if it is not linearly indpendent.

Remark 1: Another way of defining a finite linearly independent set is as

follows: A finite set {v1 , v2 , … , vn } is linearly independent if and only if the only
way of writing the zero vector as a linear combination of {v1 , v2 , … , vn } is the
trivial way:

0v1 + 0v2 + ⋯ + 0vn = 0

Lecture in 3Blue1Brown
An alternative way of defining a linearly dependent set of vectors is as follows:
A finite set {v1 , v2 , … , vn } is linearly dependent if and only if we can of write the
zero vector as a linear combination of {v1 , v2 , … , vn } in a non-trivial way; in other
words, we can find 𝛼1 ,, 𝛼2 , …, 𝛼n , not all of them zero, such that

𝛼1 v1 + 𝛼2 v2 + ⋯ + 𝛼n vn = 0

Note that in two dimension, if two vectors v1 and v2 are linearly independent,
then there are 𝛼1 , 𝛼2 , not both zero, such that 𝛼1 v1 + 𝛼2 v2 = 0. Without loss of
𝛼
generality, we may assume that 𝛼1 ≠ 0. We then have v1 = − 2 v2 . So, v1 is a
𝛼1
scalar multiple of v2 and hence v1 and v2 are collinear. Thus, in ℝ2 , two vectors
linearly dependent iff they are collinear.

Similarly, if v1 , v2 and v3 are three vectors in ℝ3 which are linearly dependent,

we can find 𝛼1 , 𝛼2 and 𝛼3 , not all of them zero such that 𝛼1 v1 + 𝛼2 v2 + 𝛼3 v3 = 0.
𝛼 𝛼
Again, assuming that 𝛼1 ≠ 0, we get v1 = − 2 v2 − 3 v3 . Thus, in ℝ3 , three
𝛼1 𝛼1
vectors are linearly dependent iff they are coplanar.

A standard way of checking the linear independence of vectors {v1 , v2 , … , vn } is

to write down the equation 𝛼1 v1 + 𝛼2 v2 + ⋯ + 𝛼n vn and check whether the
equation has a non-trivial solution or not. It the equation has a non-trivial 33
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
solution, the set of vecgtors is linearly dependent; otherwise the set of vectors
is linearly independent. Let us now look at some examples to understand this
process.

Example 1: Check whether the following set of vectors are linearly

independent:
a) {(1, 0, 0), (0, 1, 0), (0, 0, 1)} b) {(1, 0, 0), (1, 1, 0), (1, 1, 1)}
c) {(1, 0, 2, 1), (−1, 1, 1, 2), (1, 2, 8, 7)}

Solution:

a) Writing as column vectors, suppose

1 0 0 0
𝛼1 [0] + 𝛼2 [1] + 𝛼3 [0] = [0]
0 0 1 0
or
𝛼1 0
[𝛼2 ] = [0]
𝛼3 0
Therefore, 𝛼1 = 0, 𝛼2 = 0 and 𝛼3 = 0. So, the set is linearly independent.

b) Writing as column vectors, suppose

1 1 1 0
𝛼1 [0] + 𝛼2 [1] + 𝛼3 [1] = [0]
0 0 1 0
or
𝛼1 + 𝛼2 + 𝛼3 0
[ 𝛼2 + 𝛼3 ] = [0]
𝛼3 0
So, 𝛼1 , 𝛼2 and 𝛼3 are solutions to the system of equations

𝛼1 + 𝛼2 + 𝛼3 = 0
𝛼2 + 𝛼3 = 0
𝛼3 = 0

This is a triangular system and we can easily solve this by back

substitution. We get 𝛼1 = 0, 𝛼2 = 0 and 𝛼3 = 0.

c) Writing as column vectors, suppose

1 −1 1 0
0 1 2 0
𝛼1 [ ] + 𝛼2 [ ] + 𝛼3 [ ] = [ ]
2 1 8 0
1 2 7 0
or
𝛼1 − 𝛼2 + 𝛼3 0
𝛼2 + 2𝛼3 0
[ ]=[ ] …(1)
2𝛼1 + 𝛼2 + 8𝛼3 0
𝛼1 + 2𝛼2 + 7𝛼3 0
34
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Bases
. . . . . . . and
. . . . .Dimension
...........
We can write the last vector equation as the following system of equations:

𝛼1 − 𝛼2 + 𝛼3 = 0
𝛼2 + 2𝛼3 = 0
2𝛼1 + 2𝛼2 + 8𝛼3 = 0
𝛼1 + 2𝛼2 + 7𝛼3 = 0

The matrix form is

1 −1 1 0
0 1 2 0
[ ]
2 1 8 0
1 2 7 0

Carrying out the the operations R3 → R2 − 2R1 , R4 → R4 − R1 ,

R3 → R3 − 3R2 , R4 → R4 − 3R2 , R1 → R1 + R2 , we get

1 0 3 0
0 1 2 0
[ ]
0 0 0 0
0 0 0 0

As before, we take 𝛼3 , the variable corresponding to the only non-pivot

column as the free variable and set x3 = 𝜆. The solution set is (−3𝜆, −2𝜆, 𝜆).
Taking 𝜆 = 1, we see that 𝛼1 = −3, 𝛼2 = −2, and 𝛼3 = 1 is a non-trivial
solution to Eqn. (1). So, the vectors are linearly dependent.

∗∗∗

Here are some exercises for you to try.

E1) Check whether the following vectors are linearly independent:

a) {(−1, 1, 1), (1, −1, 1), (1, 1, 0)} b) {(1, 3, 2), (2, 1, −1), (1, 2, 1)}.

We will now look at some more examples.

Example 2: In the real vector space of all functions from ℝ to ℝ determine

whether the set {sinx, ex } is linearly independent.

Solution: The zero element of this vector space is the zero function, i.e., it is
the function 0 such that 0 (x) = 0 ∀ x ∈ ℝ. So we have to determine a, b ∈ ℝ
such that, ∀ x ∈ ℝ, a sin x + bex = 0.

In particular, putting x = 0, we get a.0 + b.1 = 0, i.e., b = 0. So our equation

reduces to a sinx = 0. Then putting x = 𝜋/2, we have a = 0. Thus, a = 0, b = 0.
So, {sinx, ex } is linearly independent.

∗∗∗

5.2.1 Some Elementary Results on Linear

Independence 35
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
We shall study some simple consequences of the definition of linear
independence. An immediate consequence is the following theorem.

Theorem 1: If 0 ∈ {v1 , v2 , … , vn }, a subset of the vector space V, then the set

{v1 , v2 , … , vn } is linearly dependent.

Proof: 0 is one of the v′i s. We may assume that v1 = 0. Then

1.v1 + 0.v2 + 0.v3 + ⋯ + 0.vn = 0 + 0 + ⋯ + 0 = 0. That is, 0 is a linear combination of
v1 , v2 , … , vn in which all the scalars are not zero. Thus, the set is linearly
dependent. ■

Try to prove the following result yourself.

E2) Show that, if v is a non-zero element of a vector space V over a field F,

then {v} is linearly independent.

The next result is also very elementary.

Theorem 2: a) If S is a linearly dependent subset of a vector space V over F,

then any subset of V containing S is linearly dependent.

b) A subset of a linearly independent set is linearly independent.

Proof: a) Suppose S = {u1 , u2 , … , uk } and S ⊆ T ⊆ V. We want to show that

T is linearly dependent.
If S = T there is nothing to prove. Otherwise, let

T = S ∪ {v1 , … ., vm } = {u1 , u2 , … .., uk , v1 , … ., vm }

where m > 0.
Now S is linearly dependent. Therefore, for some scalars 𝛼1 , 𝛼2 , … , 𝛼k , not
all zero, we have
k
∑ 𝛼i ui = 0
i=1

But then, 𝛼1 u1 + 𝛼2 u2 + … . + 𝛼k uk + 0.v1 + 0.v2 + … .. + 0.vm = 0, with some

𝛼i ≠ 0. Thus, T is linearly dependent.

b) Suppose T ⊆ V is linearly independent, and S ⊆ T. If possible, suppose S

is not linearly independent. Then S is linearly dependent, but then by (a),
T is also linearly dependent, since S ⊆ T. This is a contradiction. Hence,
our supposition is wrong. That is S is linearly independent.

Now, what happens if one of the vectors in a set can be written as a linear
combination of the other vectors in the set? The next theorem states that such
36 set is linearly dependent.
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Bases
. . . . . . . and
. . . . .Dimension
...........
Theorem 3: Let S = {v1 , v2 , … , vn } be a subset of a vector space V over field F.
Then S is linearly dependent if and only if some vector of S is a linear
combination of the rest of the vectors of S.

Proof: We have to prove two statements here:

i) If some vi , say v1 , is linearly combination of v2 , … .., vn , then S is linearly

dependent.

ii) If S is linearly dependent, then some vi is linear combination of the other

v′i s.

Let us prove (i) now, suppose v1 is linear combination of v2 , … , vn , i.e.,

n
v1 = 𝛼2 v2 + ⋯ + 𝛼n vn = ∑ 𝛼i vi
i=2

where 𝛼i ∈ F ∀i . Then v1 − 𝛼2 v2 − 𝛼3 v3 − ⋯ − 𝛼n vn = 0, which shows that S is

linearly dependent.

We now prove (ii), which is the converse of (i). Since S is linearly dependent,
there exist 𝛼i ∈ F, not all zero, such that

𝛼1 v1 + 𝛼2 v2 + … + 𝛼n vn = 0.

Since some 𝛼i ≠ 0, suppose 𝛼k ≠ 0. Then we have

𝛼k vk = −𝛼1 v1 … ⋯ − 𝛼k−1 vk−1 − 𝛼k+1 vk+1 … . − 𝛼n vn

Since 𝛼k ≠ 0, we divide throughout by 𝛼k and get

n
𝛼1 −𝛼n 𝛼i
vk = (− ) v1 + ⋯ + ( ) vn = ∑ 𝛽i vi , 𝛽i = −
𝛼k 𝛼k i=1 𝛼k
i≠k

Thus, vk is a linear combination of v1 , v2 , … , vk−1 , vk+1 , … , vn . ■

We can also state Theorem 3 as : S is linearly dependent if and only if some

vector in S is in the linear span of the rest of the vectors of S.

Now, let us look at the situation in ℝ3 where we know the i, j are linearly
independent. Can you immediately prove whether the set {i, j, (3, 4, 5)} is
linearly independent or not? The following theorem will help you to do this.

Theorem 4: If S is linearly independent and v ∉ [S], then S ∪ {v} is linearly

independent.

Proof: Let S = {v1 , v2 , ⋯ , vn } and T = S ∪ {v}. If possible, suppose T is linearly

dependent, then there exist scalars 𝛼, 𝛼1 , 𝛼2 , ⋯ , 𝛼n , not all zero, such that

𝛼v + 𝛼1 v1 + ⋯ + 𝛼n vn = 0. 37
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
Now, if 𝛼 = 0, this implies that there exist scalars 𝛼1 , 𝛼2 , … , 𝛼n , not all zero, such
that

𝛼1 v1 + ⋯ + 𝛼n vn = 0.

But that is impossible as S is linearly independent. Hence 𝛼 ≠ 0. But then,

(−𝛼1 ) (−𝛼n )
v= v1 + ⋯ + vn ,
𝛼 𝛼
i.e., v is linear combination of v1 , v2 , … , vn , i.e., v ∈ [S], which contradicts our
assumption.

Therefore, T = S ∪ {v} must be linearly independent. ■

Using this theorem we can immediately see that the set {i, j, (3, 4, 5)} is linearly
independent, since (3, 4, 5) is not a linear combination of i and j.

Now try the following exercises.

E3) Ginen a linearly independent subset S of a vector space V, can we always

get a set T such that S ⊆ T and T is linearly independent? (Hint: Consider
the real space R2 and the set S = {(1, 0) , (0, 1)}.)

If you’ve done Exercise 3) you will have found that, by adding a vector to a
linearly independent set, it may not remain linearly independent. Theorem 4
tells us that if, to a linearly independent set, we add a vector which is not in
the linear span of the set, then the augmented set will remain linearly
independent. Thus, the way of generating larger and larger linearly
independent subsets of a non-zero vector space V is as follows:

1. Start with any linearly independent set S1 of V, for example, S1 = {v1 },

where 0 ≠ v1 ∈ V.

2. If S1 generates the whole vector space V , i.e., if [S1 ] = V, then every

v ∈ V is a linear combination of S1 .So S1 ∪ {v} is linearly dependent for
every v ∈ V. In this case S1 is a maximal linearly independent set, that is,
no larger set than S1 is linearly independent.

3. If [S1 ] ≠ V, then there must be a v2 ∈ V such that v2 ∉ S1 . Then,

S1 ∪ {v2 } = {v1 , v2 } = S2 (say) is linearly independent. In this case, we
have found a set larger than S1 which is linearly independent, namely, S2 .

4. If [S2 ] = V, the process ends. Otherwise, we can find a still larger set S3
which is linearly independent. It is clear that, in this way, we either reach a
set which generates V or we go on getting larger and larger linearly
independent subsets of V.

38 In the next example, we we will give an infinite set that is linearly independent.
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Bases
. . . . . . . and
. . . . .Dimension
...........
Example 3: Prove that the infinite subset S = {1, x, x2 , … … }, of the vector space
P of all real polynomials in x, is linearly independent.

Solution: Take any finite subset T of S. Then ∃ distinct, non-negative integers

a1 , a2 , … , ak , such that

T = {xa1 , xa2 , … , xak }

Now, suppose
k
ai
∑ 𝛼i x = 0, where 𝛼i ∈ ℝ ∀ i
i=1

In P, 0 is the zero polynomial, all of whose coefficients are zero. ∴𝛼i = 0 ∀ i.

Hence T is linearly independent. As every finite subset of S is linearly
independent, so is S.

∗∗∗

E4) Prove that {1, x + 1, x2 + 1, x3 + 1, …} is a linearly independent subset of the

vector space P.

We have seen in E6, Unit 2 of Block 1 that any vector in R2 is a linear

combination of the vectors e1 = (1, 0) and e2 = (0, 1). Further, in R3 , we can
write every vector as a linear combination of i = (1, 0, 0), j = (0, 1, 0) and
k = (0, 0, 1). So, a natural question is, in every vector space whether there is a
subset such that we can write any vector in the vector space as a linear
combination of vectors from this subset. The next question is can we find a
subset of the ”smallest size”. This leads us to the twin concepts of dimension
and basis. We will discuss them in the following sections.

5.3 BASIS AND DIEMENSION

We begin the subsection with the definition of a basis.

Definition 2: A subset B, of a vector space V, is called a basis of V, if

a) B is linearly independent, and

b) B generates V, i.e., [B] = V.

Note that b) implies that every vector in V is a linear combination of a finite

number of vectors from B.

Thus, B ⊆ V is a basis of V if B is linearly independent and every vector of V is Lecture by Prof. Strang
a linear combination of a finite number of vectors of B. on basis and dimension
39
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
You have already seen that we can write every element of R3 as a linear
combination of i = (1, 0, 0), j = (0, 1, 0) and k = (0, 0, 1). We proved in Example
1, that this set is linearly independent. So, {i, j, k} is a basis for ℝ3 . The
following example shows that ℝ2 has more than one basis.

Example 4: Prove that B = {v1 , v2 } is a basis of ℝ2 , where v1 = (1, 1),

v2 = (−1, 1).

Solution: Firstly, for 𝛼, 𝛽 ∈ ℝ, 𝛼v1 + 𝛽v2 = 0

⟹ (𝛼, 𝛼) + (−𝛽, 𝛽) = (0, 0) ⟹ 𝛼 − 𝛽 = 0, 𝛼 + 𝛽 = 0

⟹𝛼=𝛽=0

Hence, B is linearly independent.

Secondly, given (a, b) ∈ ℝ2 , we can write

b+a b−a
(a, b) = v1 + v2
2 2
Thus, every vector in ℝ2 is a linear combination of v1 and v2 . Hence, B is also a
basis of ℝ2 .

∗∗∗

Another important characteristic of a basis is that no proper subset of a basis

can generate the whole vector space. This is brought out in the following
example.

Example 5: Prove that {i = (1, 0)} is not a basis of ℝ2 .

Solution: By E5), since i ≠ 0, {i} is linearly independent.

Now, [{i}] = {𝛼i |𝛼 ∈ ℝ } = {(𝛼, 0) |𝛼 ∈ ℝ }. So, the second component of every

vector in [i] is zero. Therefore, (1, 1) ∉ [{i}]; so [{i}] ≠ ℝ2 .

Thus, {i} is not a basis of ℝ2 .

Note that {i} is a proper subset of the basis {i, j} of ℝ2 .

∗∗∗

E5) B = {u, v, w} is a basis of ℝ3 , where

u = (1, 2, 0) , v = (2, 1, 0) , w = (0, 0, 1).

E6) Prove that {1, x, x2 , x3 , … … .} is a basis of the vector space, P, of all

polynomials over a field F.

E7) Prove that {1, x + 1, x2 + 2x} is a basis of the vector space, P2 , of all
polynomials of degree less than or equal to 2.

40 E8) Prove that {1, x + 1, 3x − 1, x2 } is not a basis of the vector space P2 .

Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Bases
. . . . . . . and
. . . . .Dimension
...........

We have already mentioned that no proper subset of a basis can generate the
whole vector space. We will now prove another important characteristic of a
basis, namely, no linearly independent subset of a vector space can obtain
more vectors than a basis of the vector space. In other words, a basis
contains the maximum possible number of linearly independent vectors. In the
next section, we will discuss the dimensions of some subspaces.

Theorem 5: If B = {v1 , v2 , ⋯ , vn } is a basis of a vector space V over field F, and

S = {w1 , w2 , … .., wm } is a linearly independent subset of V, than m ≤ n.

Proof: Since B is a basis of V and w1 ∈ V, w1 is a linear combination of v1 , v2 ,

′
… , vn . Hence, by Theorem 3, S1 = {w1 , v1 , v2 , … , vn } is linearly dependent.
Since [B] = V and B ⊆ S′1 , we have [S′1 ] = V. Since S′1 is a linearly dependent
set we have
n
′
𝛽w1 + ∑ 𝛼i vi = 0
i=1

not all of 𝛽, 𝛼i , 1 ≤ i ≤ n are zero. If 𝛽 = 0, we have ∑ni=1 𝛼i′ vi = 0, not all of 𝛼i ,

1 ≤ i ≤ n are zero. This contradicts the assumption that B is a linearly
′
𝛼
independent set. So, writing 𝛼i = − i , we have
𝛽

n
w1 = ∑ 𝛼i vi , 𝛼i ∈ F ∀ i = 1, … , n.
i=1

Now, 𝛼i ≠ 0 for some i. (Because, otherwise w1 = 0. But , as w1 belongs to a

linearly independent set, w1 ≠ 0.)

Suppose 𝛼k ≠ 0. Then we can just recorder the elements of B, so that vk

becomes v1 . This does not change any characteristic of B. It only makes the
proof easier to deal with since we can now assume that 𝛼1 ≠ 0. Then
n
1 𝛼i
v1 = w1 − ∑ vi ,
𝛼1 i=2 𝛼1

that is, v1 is a linear combination of w1 , v2 , v3 , … , vn . So any linear combination

of v1 , v2 , … , vn can also be written as a linear combination of w1 , v2 , … , vn . Thus,
if S1 = {w1 , v2 , v3 , … , vn } , then [S1 ] = V.

Note that we have been able to replace v1 by w1 in B in such a way that the
new set still generates V. Next, let

S′2 = {w2 , w1 , v2 , v3 , … , vn } .

Then, as above, S′2 is linearly dependent and [S′2 ] = V.

As we argued in the case of w1 we have

w2 = 𝛽1 w1 + 𝛽2 v2 + ⋯ + 𝛽n vn , 𝛽i ∈ F ∀ i = 1 , … , n

Again, 𝛽i ≠ 0 for some i, since w2 ≠ 0. Also, it cannot happen that 𝛽1 ≠ 0 and

𝛽i = 0 ∀ i ≥ 2, since {w1 , w2 } is a linearly independent set (by Theorem 2(b)). So
𝛽i ≠ 0 for some i ≥ 2. 41
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
Again for convenience, we may assume that 𝛽2 ≠ 0. Then

1 𝛽1 𝛽3 𝛽n
v2 = w2 − w1 − v3 − ⋯ − vn ,
𝛽2 𝛽2 𝛽2 𝛽2

This shows that v2 is a linear combination of w1 , w2 , v3 , ⋯ , vn . Hence, if

S2 = {w2 , w1 , v3 , v4 , ⋯ , vn }, then [S2 ] = V.

So we have replaced v1 , v2 in B by w1 , w2 , and the new set generates V. It is

clear that we can continue in the same way, replacing v1 by w1 at this ith step.

Now, suppose n < m. Then, after n steps, we will have replaced all v′i s by
corresponding w′i s and we shall have a set Sn = {wn , wn−1 , … , w2 , w1 } with
[Sn ] = V. But then, this means that wn+1 ∈ V = [Sn ], i.e., wn+1 is a linear
combination of w1 , w2 , … , wn . This implies that the set {w1 , … , wn , wn+1 } is
linearly dependent. This contradicts the fact that {w1 , w2 , … , wm } is linearly
dependent. Hence, m ≤ n. ■

An immediate corollary of Theorem 5 gives us a very quick way of determining

whether a given set is basis of a given vector space or not.

Corollary 1: If B = {v1 , v2 , … , vn } is a basis of V, then any set of n linearly

independent vectors is a basis of V.

Proof: If S = {w1 , w2 , … ., wn } is linearly independent subset of V, then, as

shown in the proof of Theorem 5, [S] = V. As S is linearly independent and
[S] = V, S is a basis of V. ■

The following example shows how the corollary is useful.

Example 6: Show that (1,4) and (0,1) form a basis of ℝ2 over ℝ.

Solution: You know that (1,0) and (0,1) form a basis of ℝ2 over ℝ. Thus, to
show that the given set forms a basis, we only have to show that the 2 vectors
in it are linearly independent. For this, consider the equation
𝛼 (1, 4) + 𝛽 (0, 1) = 0, where 𝛼, 𝛽 ∈ ℝ. Then (𝛼, 4𝛼 + 𝛽) = (0, 0) ⟹ 𝛼 = 0, 𝛽 = 0.

Thus, the set is linearly independent. Hence, it forms a basis of ℝ2 .

∗∗∗

E9) Let V be a vector space over F, with {u, v, w, t } as a basis.

a) Is {u, v + w, w + t, t + u} a basis.
b) Is {u, t} a basis of V?

We now give two results that you must always keep in mind when dealing with
42 vector spaces. They depend on Theorem 5.
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Bases
. . . . . . . and
. . . . .Dimension
...........
Theorem 6: If one basis of a vector space contains n vectors, then all its bases
contain n vectors.

Proof: Suppose B1 = {v1 , v2 , … , vn } and B2 = {w1 , w2 , … , wm } are both bases of

V. As B1 is a basis and B2 is linearly independent, we have m ≤ n, by
Theorem 5. On the other hand, since B2 is a basis and B1 is linearly
independent, n ≤ m. Thus, m = n. ■
Theorem 7: If a basis of a vector space contains n vectors, then any set
containing more than n vectors is linearly dependent.

Proof: Let B1 = {v1 , … , vn } be a basis of V and B2 = {w1 , … , wn+1 } be a subset of

V. Suppose B2 is linearly independent. Then, by Corollary 1, wn+1 is a linear
combination of w1 , … , wn . This contradicts our assumption that B2 is linearly
independent. Thus, B2 must be linearly dependent. ■

Try the next exercises which illustrates the use of Theorem 7.

E10) Using Theorem 7, prove that the subset S = {1, x + 1, x2 , x3 + 1, x3 , x2 + 6} of

P3 , the vector space of all real polynomials of degree ≤ 3, is linearly
dependent.

So far we have been saying that “if a vector space has a basis, then …”. Now
we state the following theorem (without proof).

Theorem 8: Every non-zero vector space has a basis.

Note: The space {0} has no basis.

Let us look at the scalars in any linear combination of basis vectors.

Coordinates of a vector: You have seen that if B = {v1 , … , vn } is a basis of a

vector space V, then every vector of V is a linear combination of the elements
of B. We now show this linear combination is unique.

Theorem 9: If B = {v1 , v2 , … , vn } is a basis of the vector space V over a field F,

then every v ∈ V can be expressed uniquely as a linear combination of
v1 , v2 , … , vn .

Proof: Since [B] = V and v ∈ V, v is a linear combination of {v1 , v2 , … , vn }. To

prove uniqueness, suppose there exist scalars 𝛼1 , … , 𝛼n , 𝛽1 , … , 𝛽n such that

v = 𝛼1 v1 + ⋯ + 𝛼n vn = 𝛽1 v1 + ⋯ + 𝛽n vn .

Then (𝛼1 − 𝛽1 ) v1 + (𝛼2 − 𝛽2 ) v2 + ⋯ + (𝛼n − 𝛽n ) vn = 0.

But {v1 , v2 , … , vn } is linearly independent. Therefore, 𝛼i − 𝛽i = 0 ∀ i, i.e., 𝛼i = 𝛽i ∀ i.

This establishes the uniqueness of the linear combination. ■ 43

Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
This theorem implies that given a basis B of V, for every v ∈ V, there is one and
n
only one way of writing v = ∑ 𝛼i vi with 𝛼i ∈ F ∀ i.
i=1

Definition 3: Let B = {v1 , v2 , … , vn } be a basis of an n – dimensional vector

space V. Let v ∈ V. If the unique expression of v as a linear combination of
v1 , v2 , … , vn is v = 𝛼1 v1 + ⋯ + 𝛼n vn , then (𝛼1 , 𝛼2 , … , 𝛼n ) are called the coordinates
of v relative to the basis B, and 𝛼i is called the ith coordinate of v.

The coordinates of a vector will depend on the particular basis chosen, as can
be seen in the following example.

Example 7: For ℝ2 , consider the two bases B1 = {(1, 0) , (0, 1)},

B2 = {(1, 1) , (−1, 1)} (see Example 4). Find the coordinates of the following
vector in ℝ2 relative to both B1 and B2 .
a) (1,2) b) (0,0) c) (p,q)

Solution:

a) Now, (1, 2) = 1 (1, 0) + 2 (0, 1) . Therefore, the coordinates of (1,2) relative

to B1 are (1,2).
3 1
Also, (1, 2) = (1, 2) + (−1, 1). Therefore, the coordinates of (1, 2) are
2 2
3 1
( , ).
2 2

b) (0, 0) = 0 (1, 0) + 0(0, 1) and (0, 0) = 0 (1, 1) + 0 (−1, 1).

In this case, the coordinates of (0,0) relative to both B1 and B2 are (0, 0).

c) (p, q) = p (1, 0) + q(0, 1) and

q+p q−p
(p, q) = (1, 1) + (−1, 1)
2 2
Therefore, the coordinates of (p, q) relative to B1 are (p, q) and the
q+p q−p
coordinates of (p, q) relative to B2 are ( , ).
2 2

∗∗∗

Note: The basis B1 = {i, j} has the pleasing property that for all vectors (p, q)
and all the coordinates of (p, q) relative to B1 are (p, q). For this reason B1 is
called the standard basis of ℝ2 , and the coordinates of a vector relative to the
standard basis are called standard coordinates of the vector. In fact, this is
the basis we normally use for plotting points in 2-dimensional space.

In general, the basis

B = {(1, 0, … .., 0) , (0, 1, 0, … ., 0) , … … , (0, 0, … … , 0, 1)}

of ℝn over ℝ is called the standard basis of ℝn .

Example 8: Let V be the vector space of all real polynomials of degree at

most 1 in the variable x. Consider the basis B = {5, 3x} of V. Find the
coordinates relative to B of the following vectors:
a) 2x + 1 b) 3x − 5 c) 11 d) 7x.

44 Solution:
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Bases
. . . . . . . and
. . . . .Dimension
...........
a) Let 2x + 1 = 𝛼 (5) + 𝛽 (3x) = 3𝛽x + 5𝛼.
Then 3𝛽 = 2, 5𝛼 = 1. So, the coordinates of 2x + 1 relative to B are
(1/5, 2/3).

b) 3x − 5 = 𝛼 (5) + 𝛽 (3x) ⟹ 𝛼 = −1, 𝛽 = 1. Hence, the answer is (−1, 1).

c) 11 = 𝛼 (5) + 𝛽 (3x) ⟹ 𝛼 = 11/5, 𝛽 = 0. Thus, the answer is (11/5, 0).

d) 7x = 𝛼 (5) + 𝛽 (3x) ⟹ 𝛼 = 0, 𝛽 = 7/3. Thus, the answer is (0, 7/3).

∗∗∗

E11) Find a standard basis of ℝ3 and for the vector space P2 of all polynomials
of degree ≤ 2.

E12) For the basis B = {(1, 2, 0) , (2, 1, 0) , (0, 0, 1) of ℝ3 , find the coordinates of
(−3, 5, 2).

E13) Prove that, for any basis B = {v1 , v2 , … , vn } of a vector space V, the
coordinates of 0 are (0, 0, 0, … , ).

E14) For the basis B = {3, 2x + 1, x2 − 2} of the vector space P2 of all polynomial
of degree ≤ 2, find the coordinates of
a) 6x + 6 b) (x + 1)2 c) x2

E15) For the basis B = {u, v} of ℝ2 , the coordinates of (1, 0) are (1/2, 1/2) and
the coordinates of (2, 4) are (3, −1). Find u, v.

We now continue the study of vector space by looking into their ‘dimension’, a
concept directly related to the basis of a vector space.

5.3.1 Dimension
So far we have seen that, if a vector space has a basis of n vectors, then every
basis has n vectors in it. Thus, given a vector space, the number of elements in
its different bases remains constant.

Definition 4: If a vector space V over the field F has a basis containing n

vectors, we say that the dimension of V is n. We write dimF V = n or, if the
underlying field is understood, we write dim V = n.

If V = {0}, it has no basis. We define dim 0 = 0.

If a vector space does not have a finite basis, we say that it is

infinite-dimensional.

In Exercise 6), you have seen that P is infinite-dimensional. Also Exercise 7

says that dimℝ P2 = 3 . Earlier you have seen that dimℝ ℝ2 = 2 and dimℝ ℝ3 = 3.
In Theorem 8, you saw that every non-zero vector space has a basis. The next
theorem gives us a helpful criterion for obtaining a basis of a finite-dimensional
vector space. 45
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
Theorem 10: If there is a subset S = {V1 , … , vn } of a non-empty vector space V
such that [S] = V, then V is finite-dimensional and S contains a basis of V.

Proof: We may assume that 0 ∉ S because, if 0 ∈ S, then S ⧵ {0} will still

satisfy the conditions of the theorem. If S is linearly independent then, since
[S] = V, S itself is a basis of V. Therefore, V is finite-dimensional (dim V = n). If
S is linearly dependent, then some vector of S is a linear combination of the rest
(Theorem 3). We may assume that this vector is vn . Let S1 = {v1 , v2 , … ., vn−1 }.

Since [S] = V and vn is a linear combination of v1 , … , vn−1 , [S] = V.

If S1 is linearly dependent, we drop, from S1 , that vector which is a linear

combination of the rest, and proceed as before. Eventually, we get a linearly
independent subset Sr = {v1 , v2 , … , vn−r } of S, such that [Sr ] = V (This must
happen because {v1 } is certainly linearly independent.) So Sr ⊆ S is a basis of
V and dim V = n − r . ■

Example 9: Show that the dimension of ℝn is n.

Solution: The set of n vectors

{(1, 0, 0, … ., 0) , (0, 1, 0, … , 0) , … ., (0, 0, 0, … ., 0, 1)}

spans V and is obviously a basis of ℝn .

∗∗∗

E16) Let V be a vector space of dimension n. If S ⊂ V, [S] = V and |S| = n, then

S is a linearly independent set and forms a basis for V.

E17) Prove that the real vector space ℂ of all complex numbers has dimension
two.

E18) Prove that the vector space Pn , of all polynomials of degree at most n,
has dimension n + 1.

Int the next subsection, we will see how to complete a linearly independent set
in a vector space to a basis of vector space when the dimension of the vector
space is finite.

5.3.2 Completion of a Linearly Independent Set to

a Basis
We have seen that in an n-dimensional vector space, a linearly independent
subset cannot have more than n vectors Theorem 7. We now ask: Suppose
we have a linearly independent subset S of an n-dimensional vector space V.
Further, suppose S has m(< n) vectors. Can we add some vectors to S, so that
46 the enlarged set will be a basis of V? In other words, can we extend a linearly
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Bases
. . . . . . . and
. . . . .Dimension
...........
independent subset to get a basis? The answer is yes. But, how many vectors
would we have to add? Do you remember Corollary 1 of Theorem 5? That
gives the answer: n − m. Of course, any (n − m) vectors won’t do the job. The
vectors have to be carefully chosen. That is what the next theorem is about.

Theorem 11: Let W = {w1 , w2 , … , wm } be a linearly independent subset of an n

– dimensional vector space V. Suppose m < n. Then there exist vectors
v1 , … , v2 , … , vn−m ∈ V such that B = {w1 , w2 , … , wm , v1 , v2 , … , vn−m } is a basis of V.

Proof: Since m < n, W is not a basis of V (Theorem 6). Hence, [W] ≠ V. Thus,
we can find a vector v1 ∈ V such that v1 ∈ [W]. Therefore, by Theorem 4,
W1 = W ∪ {v1 } is a linearly independent set with n vectors in the n-dimensional
space V, so W1 contains m + 1 vectors. If m + 1 = n, W1 is linearly independent
set with n vectors in the n-dimensional space V, so W1 is a basis of V
(Theorem 5, Corollary 1). That is, {w1 , … , wm , v1 } is a basis of V. If m + 1 < n,
then [W1 ] ≠ V, so there is a v2 ∈ V such that v2 ∉ [W1 ]. Then W2 = W ∪ {v2 } is
linearly independent and contains m + 2 vectors. So, if m + 2 = n, then

W2 = W1 ∪ {v2 } = W ∪ {v1 , v2 } = {w1 , w2 , … , wm , v1 , v2 }

is a basis of V. If m + 2 < n, we continue in this fashion. Eventually, when we

have adjoined n − m vectors v1 , v2 , … , vn−m to W, we shall get a linearly
independent set B = {w1 , w2 , … , wm , v1 , v2 , … , vn−m } containing n vectors, and
hence B will be a basis of V. ■

Let us see how Theorem Theorem 12 actually works.

Example 10: Complete the linearly independent subset S = {(2, 3, 1)} of ℝ3 to

a basis of ℝ3 .

Solution: Since S = {(2, 3, 1)},

[S] = {𝛼(2, 3, 1) |𝛼 ∈ ℝ}

= {(2𝛼, 3𝛼, 𝛼) |𝛼 ∈ ℝ }

Now we have to find v1 ∈ ℝ3 such that v1 ∉ [S], i.e., such that v1 ≠ (2𝛼, 3𝛼, 𝛼)
for any 𝛼 ∈ ℝ. We can take v1 = (1, 1, 1). Then

S1 = S ∪ {(1, 1, 1)} = {(2, 3, 1) , (1, 1, 1)}

is a linearly independent subset of ℝ3 containing 2 vectors.

Now [S1 ] = {𝛼 (2, 3, 1) + 𝛽(1, 1, 1) |𝛼, 𝛽 ∈ ℝ}

= {(2𝛼 + 𝛽, 3𝛼 + 𝛽, 𝛼 + 𝛽) |𝛼, 𝛽 ∈ ℝ

Now select v2 ∈ ℝ3 such that v2 ∉ [S1 ]. We can take v2 = (3, 4, 0). How do we
‘hit upon’ this v2 ? There are many ways. What we have done here is to take
𝛼 = 1 = 𝛽, then 2𝛼 + 𝛽 = 3, 3𝛼 + 𝛽 = 4, 𝛼 + 𝛽 = 2. So (3, 4, 2) belongs to [S1 ] then,
by changing the third component from 2 to 0, we get (3, 4, 0), which is not in
[S1 ]. Since v2 ∉ [S1 ] , S1 ∪ {v2 } is linearly independent. That is,

S2 = {(2, 3, 1) , (1, 1, 1),(3, 4, 0)} 47

Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
is a linearly independent subset of ℝ3 . Since S2 contains 3 vectors and
dimℝ ℝ3 = 3 , S2 is a basis of ℝ3 .

∗∗∗

Note: Since we had a large number of choices for both v1 and v2 , it is obvious
that we could have extended S to get a basis of ℝ3 in many ways.

Example 11: For the vector space P2 of all polynomials of degree ≤ 2,

complete the linearly independent subset S = {x + 1, 3x + 2} to form a basis of P2 .

Solution: We note that P2 has dimension 3, a basis being {1, x, x2 } (see E19).
So we have to add only one polynomials to S to get a basis of P2 .

Now [S] = {a (x + 1) + b(3x + 2) |a, b ∈ ℝ}

= {(a + 3b) x + (a + 2b) |a, b ∈ ℝ}

This shows that [S] does not contain any polynomials of degree 2. So we can
choose x2 ∈ P2 because x2 ∉ [S]. So S can be extended to {x + 1, 3x + 2, x2 },
which is a basis of P2 . Have you wondered why there is no constant term in
this basis? A constant term is not necessary. Observe that 1 is linear
combination of x + 1 and 3x + 2, namely, 1 = 3 (x + 1) − 1(3x + 2). So, 1 ∈ [S] and
hence , ∀ 𝛼 ∈ ℝ, 𝛼.1 = 𝛼 ∈ [S].

∗∗∗

E19) Complete S = {(−3, 1/3)} to a basis of ℝ2 .

E20) Complete S = {(1, 0, 1) , (2, 3, −1) in two different ways to get two distinct
bases of ℝ3 .

E21) For the vector space P3 of all polynomials of degree = 3, complete

a) S = {2, x2 + x, 3x3 }
b) S = {x2 + 2, x2 − 3x}

to get a basis of P3 .

Let us now look at some properties of the dimensions of some subspaces.

5.4 DIMENSIONS OF SOME SUBSPACES

In Unit 3 you learnt what a subspace of a space is. Since it is a vector space
itself, it must have a dimension. We have the following theorem.

Theorem 12: Let V be a vector space over a field F such that dim V = n . Let W
48 be a subspace of V. Then dim W ≤ n .
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Bases
. . . . . . . and
. . . . .Dimension
...........
Proof: Since W is a vector space over F in its own right, it has a basis.
Suppose dim W = m . Then the number of elements in W’s basis is m. These
elements form a linearly independent subset of W, and hence, of V. Therefore,
by Theorem 7, m ≤ n. ■
Remark 2: If W is a subspace of V such that dim W = dim V = n then
W = V, since the basis of W is a set of linearly independent elements in V, and
we can appeal to Corollary 1.

Example 12: Let V be a subspace of ℝ2 . What are the possible subspaces of

V?

Solution: By Theorem 12, since dim ℝ2 = 2, the only possibilities for dim V
are 0,1 and 2.

If dim V = 2, then, by the remark above, V = ℝ2 .

If dim V = 1 , then {(𝛽1 , 𝛽2 )} is a basis of V, where (𝛽1 , 𝛽2 ) ∈ ℝ2 ⧵ 0. Then

V = {𝛼 (𝛽1 , 𝛽2 ) |𝛼 ∈ ℝ} .

The set of points on the line ax + by = 0, through the origin, is a subspace of ℝ2 .

In particular S = { (x, y) ∈ ℝ2 | 𝛽2 x − 𝛽1 y = 0} is a subspace of ℝ2 and V ⊆ S. So,
dim S ≥ dim V = 1.

To show that S = V, we need to show that S ⊆ V. Note that, since (𝛽1 , 𝛽2 ) ≠ 0,

at most one of 𝛽1 or 𝛽2 can be zero. We will assume that 𝛽1 ≠ 0, 𝛽2 ≠ 0. If
𝛽1 y y y
(x, y) ∈ S, we have x = y or x = 𝛽1 . Therefore, (x, y) = ( 𝛽1 , 𝛽2 ) = 𝛼(𝛽1 , 𝛽2 )
𝛽2 𝛽2 𝛽2 𝛽2
y
where 𝛼 = . The same argument works if 𝛽1 = 0, 𝛽2 ≠ 0. If 𝛽2 = 0, 𝛽1 ≠ 0, we
𝛽2
x
have y𝛽1 = 0 or y = 0. Then, (x, y) = (x, 0) = (𝛽1 , 𝛽2 ) = 𝛼 (𝛽1 , 𝛽2 ). Since we can
𝛽1
write the equation ax + by = 0 as ax − (−b)y = 0, it follows that we can
characterise the one dimensional subspaces of ℝ2 as lines through the origin.

If dim V = 0 , then V = {0}.

∗∗∗

Let us see what happens in three dimensions.

Example 13: Let V be a subspace of ℝ3 . What are the possibilities of its

structure?

Solution: dim V can be 0, 1, 2 or 3. dim V = 0 ⟹ V = {0}.

dim V = 1 ⟹ V = { 𝛼(𝛽1 , 𝛽2 , 𝛽3 )| 𝛼 ∈ ℝ}

for some (𝛽1 , 𝛽2 , 𝛽3 ) ∈ ℝ3 .

As we did in the previous example, let us first consider the case 𝛽1 ≠ 0, 𝛽2 ≠ 0

and 𝛽3 ≠ 0. Consider the set

x y z
S = {(x, y, z) ∈ ℝ3 | = = }
𝛽1 𝛽2 𝛽3 49
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
Let (x1 , y1 , z1 ), (x2 , y2 , z2 ) ∈ S. The set S is a line through the origin.

Further,
x1 y1 z1 x2 y2 z2
= = and = =
𝛽1 𝛽2 𝛽3 𝛽1 𝛽2 𝛽3

Check that
𝛼x1 + 𝛽x2 𝛼y1 + 𝛽y2 𝛼z1 + 𝛽z2
= =
𝛽1 𝛽2 𝛽2

This shows that S is a subspace of ℝ3 . Further V ⊆ S.

x y z
Also, if (x, y, z) ∈ S, writing = = = 𝛼, we have
𝛽1 𝛽2 𝛽3
(x, y, z) = (𝛼𝛽1 , 𝛼𝛽2 , 𝛼𝛽3 ) = 𝛼𝛼 (𝛽1 , 𝛽2 , 𝛽3 ). So, V = S. Suppose one of 𝛽1 , 𝛽2 or 𝛽3
is zero, say, 𝛽1 = 0. Then, we let

y z
S = {(x, y, z) ∈ ℝ3 |x = 0, = }
𝛽2 𝛽3

Then, S is a line through the origin that lies on the yz-plane. Again, we can
show that S is a subspace and

S = V = { 𝛼 (0, 𝛽2 , 𝛽3 )| 𝛼 ∈ ℝ}

Similarly, if 𝛽2 = 0 and 𝛽1 ≠ 0, 𝛽3 ≠ 0, we let

x z
S = {(x, y, z) ∈ ℝ3 |y = 0, = }
𝛽1 𝛽3

is a line in xz-plane. Again, we can show that S is a subspace and

S = V = { 𝛼 (𝛽1 , 0, 𝛽3 )| 𝛼 ∈ ℝ}

If 𝛽3 = 0 and 𝛽1 ≠ 0, 𝛽2 ≠ 0, we let

x y
S = {(x, y, z) ∈ ℝ3 |z = 0, = }
𝛽1 𝛽2

Again, we can show that S is a subspace and

S = V = { 𝛼 (𝛽1 , 𝛽3 , 0)| 𝛼 ∈ ℝ}

Similarly, you work out the cases where two of 𝛽1 , 𝛽2 or 𝛽3 is zero.

dim V = 2 ⟹ V is generated by two linearly independent space vectors.

Suppose u1 = (𝛼1 , 𝛼2 , 𝛼3 ), u2 = (𝛽1 , 𝛽2 , 𝛽3 ) are linearly independent vectors and

V = [u1 , u2 ]

i.e. {u1 , u2 } is a basis of V. Consider the equations

𝛼1 x + 𝛼2 y + 𝛼3 z = 0
𝛽1 x + 𝛽2 y + 𝛽3 z = 0

The system has two equations in three unkowns. So, by ??, this has a nonzero
solution (a, b, c). Let

50 S = { (x, y, z) ∈ ℝ3 | ax + by + cz = 0}
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Bases
. . . . . . . and
. . . . .Dimension
...........
Then, S is a subspace of ℝ3 and u1 , u2 ∈ S. Since S contains a basis of V, it
contains V.

Consider the equation ax + by + cz = 0, at least one of a, b or c is nonzero. Let

us suppose a ≠ 0. The same argument works when b or c is nonzero. Then,
b c
x = − y − z. So, if (x, y, z) ∈ S, we have
a a

b c
x − −
a a
[y] = y [ 1 ] + z [ 0 ]
z 0 1

b c
− −
a a
So, S is spanned by v1 = [ 1 ] and v2 = [ 0 ]. Check that {v1 , v2 } is linearly
0 1
independent. Therefore, dim S = 2. Since V ⊆ S and dim V = dim S, it follows
that S = V.

dim V = 3 ⟹ V = ℝ3 .

∗∗∗

Now let us go further and discuss the dimensions of the sum of subspace (See
Unit 3.). If U and W are subspaces of a vector space V, then so are U + W and
U ∩ W. Thus, all these subspaces have dimensions. We relate these
dimensions in the following theorem.

Theorem 13: If U and W are two subspaces of a finite-dimensions vector

space V over a field F, then

dim(U + W) = dim U + dim W − dim(U ∩ W)

Proof: We recall that U + W = {u + w |u ∈ U, w ∈ W }.

Let dim(U ∩ W) = r, dim U = m, dim W = n. We have to prove that

dim (U + W) = m + n − r.

Let {v1 , v2 , … , vr } be a basis of U ∩ W. Then {v1 , v2 , … , vr } is a linearly

independent subset of U and also of W. Hence, by, it can be extended to form a
basis

A = {v1 , v2 , … , vr , ur+1 , ur+2 , … , um }

of U and a basis

B = {v1 , v2 , … , vr , wr+1 , wr+2 , … , wn }

of W.

Now, note that none of the u′ s can be a w. For, if us = wt then us ∈ U, wt ∈ W,

so that us ∈ U ∩ W. But then us must be a linear combination of the basis
{v1 , … , vr } of U ∩ W. This contradicts the fact that A is linearly independent. 51
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
Thus,

A ∪ B = {v1 , v2 , … , vr , ur+1 , … , um , wr+1 , … , wn }

contains (m − r) + (n − r) vectors. We need to prove that A ∪ B is a basis of

U + W. For this, we need to prove that every vector of U + W is linear
combination of A ∪ B. So let
r m n
∑ 𝛼i vi + ∑ 𝛽j uj + ∑ 𝜏k wk = 0
i=1 j=r+1 k=r+1

where 𝛼i , 𝛽j , 𝜏k ∈ F ∀ i, j, k.

Then
r m n
∑ 𝛼i vi + ∑ 𝛽j uj = − ∑ 𝜏k wk …(2)
i=1 j=r+1 k=r+1

The vector on the left hand side of ?? is a linear combination of

{v1 , … .vr , ur+1 , … … , un }. So it is in U. The vector on the right hand side is in W.
Hence, the vectors on both side of the equation are in U ∩ W. But {v1 , … … , vr }
is a basis of U ∩ W. So the vectors on both sides of Eqn. (2) are a linear
combination of the basis {v1 , … .., vr } of U ∩ W.

That is,
r m r
∑ 𝛼i vi + ∑ 𝛽j uj = ∑ 𝛿i vi …(3)
i=1 j=r+1 i=1

and
n r
∑ 𝜏k wk = ∑ 𝛿i vi …(4)
k=r+1 i=1

where 𝛿i ∈ F ∀ i = 1, … , r

Eqn. (2) gives ∑ (𝛼i − 𝛿i ) vi + ∑ 𝛽i ui = 0.

But {v1 , … , vr , ur+1 , … , um } is linearly independent, so 𝛼i = 𝛿i and 𝛽j = 0 ∀ i, j.

Similarly, since by Eqn. (3)

∑ 𝛿i vi + ∑ 𝜏k wk = 0,

Since, we have already obtained 𝛼i = 𝛿i ∀ i, we get 𝛼i = 0 ∀ i.

Thus, ∑ 𝛼i vi + ∑ 𝛽j uj + ∑ 𝜏l wk = 0

⟹ 𝛼i = 0, 𝛽j = 0, 𝜏k = 0 ∀ i, j, k.

So A ∪ B is linearly independent.

Next, let u + w ∈ U + W. Then u = ∑ 𝛼i vi + ∑ 𝛽j uj and w = ∑ 𝛽i vi + ∑ 𝜏k wk , i.e., u + w

is a linear combination of A ∪ B.

∴, A ∪ B is a basis of U + W, and

dim (U + W) = m + n − r = dim U + dim W − dim(U ∩ W)

52 ■
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Bases
. . . . . . . and
. . . . .Dimension
...........
We give a corollary to Theorem 13 now.

Corollary 2: dim(U ⨁ W) = dim U + dim W.

Proof: The direct sum U ⨁ W indicates that U ∩ W = {0}. Therefore, dim

(U ∩ W) = 0. Hence, dim (U + W) = dim U + dim W . ■

Let us use Theorem 13 now.

Example 14: Suppose U and W are subspace of V,

dim U = 4 , dim W = 5, dim V = 7 . Find the possible values of dim(U ∩ W) .

Solution: Since W is a subspace of U + W, we must have

dim (U + W) ≥ dim W = 5 .i.e.,
dim U + dim W − dim(U ∩ W) ≥ 5 ⟹ 4 + 5 − dim(U ∩ W) ≥ 5 ⟹ dim(U ∩ W) ≤ 4 .
On the other hand, U + W is a subspace of V, so dim (U + W) ≤ 7.

⟹ 5 + 4 − dim(U ∩ W) ≤ 7

⟹ dim(U ∩ W) ≥ 2

Thus, dim (U ∩ W) = 2, 3 or 4.

∗∗∗

Example 15: Let V and W be the following subspace of ℝ4 :

V = {(a, b, c, d) |b − 2c + d = 0} , W = {(a, b, c, d) |a = d, b = 2c}

Find basses and the dimensions of V, W and V ∩ W. Hence prove that

4
ℝ = V + W.

Solution: We observe that

(a, b, c, d) ∈ V ⟺ b − 2c + d = 0.

⟺ (a, b, c, d) = (a, b, c, 2c − b)

= (a, 0, 0, 0) + (0, b, 0 − b) + (0, 0, c, 2c)

= a (1, 0, 0 , 0) + b (0, 1, 0, −1) + c(0, 0, 1, 2)

This shows that every vector in V is a linear combination of the three linearly
independent vectors (1, 0, 0, 0) , (0, 1, 0, −1) , (0, 0, 1, 2). Thus, a basis of V is

A = {(1, 0, 0, 00, (0, 1, 0, −1) , (0, 0, 1, 2)}

Hence, dim V = 3 .

Next, (a, b, c, d) ∈ W ⟺ a = d, b = 2c

⟺ (a, b, c, d) = (a, 2c, c, a) = (a, 0, 0, a) + (0, 2c, c, 0)

= a (1, 0, 0, 1) + c (0, 2, 1, 0) ,

This shows that W is generated by the linearly independent set

{(1, 0, 0, 1) , (0, 2, 1, 0)}. 53
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
∴, a basis for W is

B = {(1, 0, 0, 1) , (0, 2, 1, 0)},

and dim W = 2 .

Next, (a, b, c, d) ∈ V ∩ W ⟺ (a, b, c, d) ∈ V and (a, b, c, d) ∈ W

⟺ b − 2c + d = 0, a = d, b = 2c

⟺ (a, b, c, d) = (0, 2c, c, 0) = c(0, 2, 1, 0)

Hence, a basis of V ∩ W is {(0, 2, 1, 0)} and dim (V ∩ W) = 1 .

Finally, dim (V + W) = dim V + dim W − dim(V ∩ W)

= 3 + 2 − 1 = 4.

Since V + W is a subspace of ℝ4 and both have the same dimension,

4
ℝ = V + W.

∗∗∗

E22) If U and W are 2 – dimensional subspaces of ℝ3 , show that U ∩ W ≠ {0}.

E23) If U and W are distinct 4 –dimensional subspaces of a 6-dimensional

vector space V, find the possible dimensions of U ∩ W.

E24) Suppose V and W are subspaces of ℝ4 such that dim V = 3, dim W = 2.

prove that dim(V ∩ W) = 1 or 2.

E25) Let V and W be subspaces of ℝ4 defined as follows :

V = {(a, b, c) |b + 2c = 0}

W = {(a, b, c) |a + b + c = 0}

a) Find bases and dimensions of V, W, V ∩ W

b) Find dim (V + W).

Let us now look at the dimension of a quotient space. Before going further it
may help to revise Sec. 3.5.

5.5 DIMENSION OF A QUOTIENT SPACE

In unit 3 we defined the quotient space V/W for any vector space V and
subspace W. Recall that V/W = {v + W |v ∈ V }.

We also showed that it is a vector space. Hence, it must have a basis and a
54 dimension. The following theorem tells us what dim V/W should be.
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Bases
. . . . . . . and
. . . . .Dimension
...........
Theorem 14: If W is a subspace of a finite-dimensional space V, then
dim (V/W) = dim V − dim W.

Proof: Suppose dim V = n and dim W = m. Let {w1 , w2 , … , wm } be a basis of

W. Then there exist vectors v1 , v2 , … , vk such that {w1 , w2 , … , wm , v1 , v2 , … , vk } is
a basis of V, where m + k = n. (Theorem 11.)

We claim that B = {v1 + W, v2 + W, … , vk + W} is a basis of V/W. First, let us

show that B is linearly independent. For this, suppose ∑ki=1 𝛼i (vi + W) = W,
where 𝛼1 , … , 𝛼k are scalars

(note that the zero vector of V/W is W).

Then ∑ki=1 𝛼i vi + W = W
k
⟹ (∑ 𝛼i vi ) + W = W
i=1

k
⟹ ∑ 𝛼i vi ∈ W
i=1

But W = [{w1 , w2 , … ., wm }], so ∑ki=1 𝛼i vi = ∑m

j=1 𝛽j wj for some scalars 𝛽1 , … ., 𝛽m .

⟹ ∑ 𝛼i vi − ∑ 𝛽j wj = 0

But {w1 , … , wm , v1 , … .., vk } is a basis of V, so it is linearly independent. Hence

we must have

𝛽j = 0, 𝛼i = 0 ∀ j, i.

Thus,

∑ 𝛼i (vi + W) = W ⟹ 𝛼i = 0 ∀ i.

So B is linearly independent.

Next, to show that B generates V/W, let v + W ∈ V/W. Since v ∈ V and

m k
{w1 , … , wm , v1 , … , vk } is a basis of V, v = ∑1 𝛼i wi + ∑1 𝛽j vj , where the 𝛼i s and 𝛽j s
are scalars.

Therefore,

v + W = (∑ 𝛼i wi + ∑ 𝛽j vj , ) + W
i j

= {(∑ 𝛼i wi ) + W} + {(∑ 𝛽j vj ) + W}
i j
k
= W + ∑ 𝛽j (vj + W) since ∑ 𝛼i wi ∈ W
1
k
= ∑ 𝛽j (vj + W)
j=1

since W is the zero element of V/W. 55

Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
Thus, v + W is linear combination of {vj + W, j = 1, 2, , … … ., k}.

So, v + W ∈ [B].

Thus, B is a basis of V/W.

Hence, dim V/W = k = n − m = dim V − dim W . ■

Let us use this theorem to evaluate the dimensions of some familiar quotient
spaces.

Example 16: If Pn denotes the vector space of all polynomials of degree ≤ n,

exhibit a basis of P4 /P2 and verify that dim P4 /P2 = dim P4 − dim P2 .

Solution: Now P4 = {ax4 + bx3 + cx2 + dx + e |a, b, c, d, e ∈ ℝ} and

P2 = {ax2 + bx + c |a, b, c ∈ ℝ} .

Therefore, P4 /P2 = {(ax4 + bx3 ) + P2 |a, b ∈ ℝ },

Now,

(x4 + bx3 ) + P2 = (ax4 + P2 ) + (bx3 + P2 ) = a (x4 + P2 ) + b(x3 + P2 )

This shows that every element of P4 /P2 is a linear combination of the two
elements (x4 + P2 ) and (x3 + P2 ).

These two elements of P4 /P2 are also linearly independent because if

4 3 4 3
𝛼 (x + P2 ) + 𝛽 (x + P2 ) = P2 , then 𝛼x + 𝛽x ∈ P2 (𝛼, 𝛽 ∈ ℝ).

4 3 2
∴, 𝛼x + 𝛽x = ax + bx + c for some a, b, c ∈ ℝ

⟹ 𝛼 = 0, 𝛽 = 0, a = 0, b = 0, c = 0.

Hence a basis of P4 /P2 is {x4 + P2 , x3 + P2 }.

Thus, dim (P4 /P2 ) = 2 . Also dim (P4 ) = 5 , dim (P2 ) = 3, (see E19). Hence
dim (P4 /P2 ) = dim (P4 ) − dim (P2 ) is verified.

∗∗∗

Try the following exercise now.

E26) Let V be an n – dimensional real vector space. Find dim (V/V) and
dim V/{0}.

We end this unit by summarizing what we have covered in it.

5.6 SUMMARY
56 In this unit, we
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Bases
. . . . . . . and
. . . . .Dimension
...........
1. defined a linearly independent set and gave examples of linearly
independent and linearly dependent sets of vectors;

2. saw how to determine whether a give set of vectors is linearly

independent or not;

3. defined a basis for a vector space and give examples;

4. how to determine whether a set forms a basis for a vector space or not;

5. determined bases for well known vector spaces;

6. determined the dimension of the sum of two spacese, intersection of two

subspaces and quotient spaces.

5.7 SOLUTIONS/ANSWERS

E1) a) Writing as column vectors, suppose

−1 1 1 0
𝛼1 [ 1 ] + 𝛼2 [−1] + 𝛼3 [1] = [0]
1 1 0 0
or
−𝛼1 − 𝛼2 + 𝛼3 0
[ 𝛼1 − 𝛼2 + 𝛼3 ] = [0]
𝛼1 + 𝛼2 0

Writing in the form of equations

−𝛼1 + 𝛼2 + 𝛼3 = 0
𝛼1 − 𝛼2 + 𝛼3 = 0
𝛼1 + 𝛼2 =0

The matrix form is

−1 1 1 0
[1 −1 1 0]
1 1 0 0

Carrying out row operations R1 → −R1 , R2 → R2 − R1 , R3 → R3 − R1 ,

1 1 3
R2 ↔ R3 , R2 → R2 , R1 → R1 + R2 , R3 → R3 , R1 → R1 − R3 and
2 2 2
1
R2 → R2 − R3 gives the matrix (Check this!)
2

1 0 0 0
[0 1 0 0]
0 0 1 0

There are no non-pivot columns except the last column which doesn’t
correspond to a variable. The solution to the system is 𝛼1 = 0, 𝛼2 = 0
and 𝛼3 = 0. So, the vectors are linearly independent. 57
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
b) Writing as column vectors, suppose

1 2 1 0
𝛼1 [3] + 𝛼2 [ 1 ] + 𝛼3 [2] = [0]
2 −1 1 0
or

𝛼1 + 2𝛼2 + 𝛼3 0
[3𝛼1 + 𝛼2 + 2𝛼3 ] = [0] …(5)
2𝛼1 − 𝛼2 + 𝛼3 0

Writing as linear equations, we get

𝛼1 + 2𝛼2 + 𝛼3 = 0
3𝛼1 + 𝛼2 + 2𝛼3 = 0
2𝛼1 − 𝛼2 + 𝛼3 = 0

The matrix form is

1 2 1 0
[3 1 2 0]
2 −1 1 0

Carrying out the row operations R2 → R2 − 3R1 , R3 → R3 − 2R1 ,

R
R2 → − 2 , R1 → R1 − R2 and R3 → R3 + 5R2 , we get
5

3
1 0 0
5
1
[0 1 0]
5
0 0 0 0

The column corresponding te 𝛼3 is the non-pivot column. Setting

3 1
x3 = 𝜆, the solutions is 𝛼1 = − 𝜆, 𝛼2 = − 𝜆, 𝛼3 = 𝜆. In particular, setting
5 5
𝜆 = 5, we get a non-trivial solution 𝛼1 = −5, 𝛼2 = −3 and 𝛼3 = −1.

E2) Suppose 𝛼 ∈ F such that 𝛼v = 0. Then, from Unit 3 you know that 𝛼 = 0
or v = 0. But v ≠ 0…, 𝛼 = 0, and {v} is linearly independent.

E3) The set S = {(1, 0) , (0, 1)} is a linearly independent subset of ℝ2 . Now,
suppose ∃ T such that S ⊆ T ⊆ ℝ2 . Let (x, y) ∈ T such that (x, y) ∉ S. Then
we can always find a, b, c ∈ ℝ, not all zero, such that
a (1, 0) + b (0, 1) + c (x, y) = (0, 0). (Take a = −x, b = −y, c = 1, for example.)
∴S ∪ {(x, y)} is linearly independent. Since this is obtained in T, T is
linearly dependent. The answer to the question in this exercise is ‘No’.

E4) Let T be a finite subset of P. Suppose 1 ∉ T. Then, as in Example 3, ∃

non zero a1 , … … , ak such that

T = {xa1 + 1, … .., xak + 1}.

Suppose ∑ki=1 𝛼i (xa1 + 1) = 0, where 𝛼i ∈ ℝ ∀ i.

58 Then 𝛼1 xa1 + ⋯ + 𝛼k xak + (𝛼1 + 𝛼2 + ⋯ + 𝛼k ) = 0

Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Bases
. . . . . . . and
. . . . .Dimension
...........
⟹ 𝛼i = 0 = 𝛼2 = . ⋯ = 𝛼k , so that T is linearly independent.
If 1 ∈ T, then T = {1, xa1 + 1, … , xak + 1} for some non-zero a1 , … , ak .
k a
𝛽0 + ∑i=1 𝛽i (x i + 1) = 0, where 𝛽0 , 𝛽1 , … .., 𝛽k ∈ ℝ.
Then 𝛽0 + 𝛽1 + ⋯ + 𝛽k = 0, 𝛽1 xa1 + ⋯ + 𝛽k xak = 0.

⟹ 𝛽0 + 𝛽1 + ⋯ + 𝛽k = 0, 𝛽1 = 0 = 𝛽2 = ⋯ = 𝛽k

⟹ 𝛽0 = 0 = 𝛽1 = ⋯ = 𝛽k

⟹ T is linearly independent.
Thus, every finite subset of {1, x + 1, x2 + 1, …} is linearly independent.
Therefore, {1,x+1,….} is a linearly independent.

E5) B is linearly independent. For any (a, b, c) ∈ ℝ3 , we have

2b−a 2a+b
(a, b, c) = (1, 2, 0) + (2, 1, 0) + c(0, 0, 1). Thus, B also spans ℝ3 .
3 3

E6) Firstly, any element of P is of the form a0 + a1 x + a2 x2 + … . + an xn ,ai ∈ ℝ ∀ i.

This is a linear combination of {1, x, … ., xn }, a finite subset of the given set.
Therefore, the given set spans P. Secondly, Example 3 says that the
given set is linearly independent. Therefore, it is a basis of P.
E7) The set {1, x + 1, x2 + 2x} is linearly independent. It also spans P2 , since
any element a0 + a1 x + a2 x2 ∈ P2 can be written as
(a0 − a1 + 2a2 ) + (a1 − 2a2 ) (x + 1) + a2 (x2 + 2x). Thus, the set is a basis of
P2 .
E8) The set is linearly dependent, since 4 − 3 (x + 1) + (3x − 1) + 0.x2 = 0. .., it
can’t form a basis of P2 .

E9) a) We have to show that the given set is linearly independent.

Now au + b (v + w) + c (w + t) + d (t + u) = 0, for a, b, c, d ∈ F.

⟹ (a + d) u + bv + (b + c) w + (c + d) t = 0

⟹ a + d = 0, b = 0, b + c = 0 and c + d = 0. Solving for a, b, c and d,

we get a = 0, b = 0, c = 0 and d = 0. Since it has 4 vectors, it is a
basis of V.
b) No, since [{u, t}] ≠ V. For example, w ∉ [{u, t}] as {u, w, t} is a linearly
independent set by Theorem 2.

E10) You know that {1, x, x2 , x3 } is a basis of P3 , and contains 4 vectors. The
given set contains 6 vectors, and hence, by Theorem 7, it must be linearly
dependent.

E11) A standard basis of ℝ3 is {(1, 0, 0) , (0, 1, 0) , (0, 0, 1)} .{1, x, x2 } is a standard

basis for P2 , because the coordinates of any vector a0 + a1 x + a2 x2 , in P2 ,
is (a0 , a1 , a2 ).
13 11
E12) ( ,− , 2)
3 3
E13) Since 0 = 0.v1 + 0.v2 + … .. + 0.vn , the coordinates are (0, 0, … ., 0). 59
Block
. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Vector
. . . . . . . .Spaces-II
.........
E14) a) 6x + 6 = 1.3 + 3 (2x + 1) + 0.(x2 − 2). ∴, the coordinates are (1,3,0).
b) (2/3, 1, 1).
c) (2/3, 0, 1).
E15) Let u = (a, b) , v = (c, d). We know that
a+c b+d
(1, 0) = 1/2 (a, b) + 1/2 (c, d) = ( , ),
2 2
and (2, 4) = 3 (a, b) − (c, d) = (3a − c, 3b − d)
∴, a + c = 2, b + d = 0, 3a − c = 2, 3b − d = 4. Solving these equations gives
us a = 1, b = 1, c = 1, d = −1

∴, u = (1, 2) , v = (1, −2).

E16) By Theorem 10 it follows that there is a set S1 ⊆ S such that S1 is a basis

for V. If S1 ⊆ S, then |S1 | < |S| = n. But this contradicts Theorem 6. So,
S1 = S and S is a basis for V.
E17) ℂ = {x + iy |x, y ∈ ℝ }. Consider the set S = {1 + i0, 0 + i1}. This spans ℂ and
is linearly independent. Therefore, it is a basis of ℂ. ∴, dimℝ C = 2.
E18) Check that the set {1, x, … ..xn } is linearly independent and spans Pn .

E19) To obtain a basis we need to add one element. Now.

[S] = {𝛼91, 0, 1) + 𝛽(2, 3, −1) |𝛼, 𝛽 ∈ R }

= {(𝛼 + 2𝛽, 3𝛽, 𝛼 − 𝛽) |𝛼, 𝛽 ∈ R}

Then (1, 0, 0) ∉ [S] and (0, 1, 0) ∉ [S] . ∴{(1, 0, 1) , (2, 3, −1) , (1, 0, 0)} and
3
{(1, 0, 1) , (2, 3, −1) , (0, 1, 0)} are two distinct bases of R .

E20) a) Check that x ∉ [S].∴ S ∪ {x} is a basis.

b) 1 ∉ [S]. Let S1 = S ∪ {1}. Then x3 ∉ [S1 ]. Thus, a basis is
2 2 3
{1, x + 2, x − 3x, x }.

E21) dim U = 2 = dim W. Now U + W is a subspace of R3 .

∴dim(U + W) ≤ 3. i.e., dim U + dim W − dim(U ∩ W) ≤ 3 .
i.e., dim(U ∩ W) ≥ 1 ∴U ∩ W ≠ {0}.
E22) dim V = 6 , dim U = 4 = dim W and U ≠ W. Then dim(U + W) ≤ 6
⟹ 4 + 4 − dim(U ∩ W) ≤ 6 ⟹ dim(U ∩ W) ≥ 2 . Also,
dim(U + W) ≥ dim U ⟹ dim(U ∩ W) ≤ 4 . ∴, the possible dimensions of
U ∩ W are 2,3,4.
E23) Since V + W is a subspace of R4 , dim(V + W) ≤ 4 .
That is , dim V + dim W − dim(V ∩ W) ≤ 4 .

∴, dim(V ∩ W) ≥ 1.

Also V ∩ W is a subspace of W. ∴, dim(V ∩ W) ≤ dim W = 2 .

60 ∴, 1 ≤ dim(V ∩ W) ≤ 2 .
Unit
. . . . .4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Bases
. . . . . . . and
. . . . .Dimension
...........
E24) a) Any element of V is v = (a, b, c) with b + 2c = 0.

∴,v = (a, −2c, c) = a (1, 0, 0) + c(0, −2, 1).

∴, a basis of V is {(1, 0, 0) , (0, −2, 1)}.∴,dim V = 2 .

Any element of W is w = (a, b, c) with a + b + c = 0.

∴, w = (a, b, −a − b) = a (1, 0, −1) + b(0, 1, −1)

∴, a basis of W is {(1, 0, −1) , (0, 1, −1)}.

∴,dim W = 2 .

Any element of V ∩ W is x = (a, b, c) with b + 2c = 0 and a + b + c = 0.

∴,x = (a, −2c, c) with a − 2c + c = 0, that is, a = c.
∴, x = (c, −2c, c) = c(1, −2, 1). ∴, a basis of V ∩ W is {1, −2, 1}.

∴,dim (V ∩ W) = 1 .

b) dim (V + W) = dim V + dim W − dim (V ∩ W) = 3 . ∴,V + W = R3 .

E25) 0, n.

61
 UNIT 6
ELEMENTARY MATRICES AND ROW
REDUCTION

Structure Page No.

6.1 Introduction
Objectives
6.2 Elementary Matrices
6.3 Row Space, Column Space and Null Space of a Matrix
6.4 Summary
6.5 Solution/Answers

6.1 INTRODUCTION
In the previous Units, you have learned about several objects (mathematical!)
like vector space, subspaces, linear independence, bases and dimension. In
this Unit, the emphasis is on computing some of these objects. The main
technique that we are going to use is row reduction. In Sec.~6.2 we take a
closer look at row reduction. As we will see, row operations on a matrix are
equivalent multiplying the matrix by certain special matrices called elementary
matrices. We will study the properties of these matrices in Sec.~6.2. We will
also see how to find the inverse of a matrix through row operations. In
Sec.~6.3, we will see how to calculate bases and dimension of the subspaces
spanned by the columns or rows of a matrix. We introduce you to an
important concept, the rank of a matrix, in this section. Later in this section, we
will see how to find a bases for subspaces of n if we are given a spanning
set of the subspace of the subspace. We will also see how to find bases for
the sum and intersection of two subspaces of n .

Objectives
After studying this Unit, you will be able to:
• define elementary matrices and explain the row operations to them;
• prove that elementary matrices are invertible and give the inverse of an
elementary matrix;
• define the row rank and the column rank of a matrix;
• compute the rank of a matrix;
• state and apply the rank-nullity theorem;
• find the dimension and a basis for subspace of n given a spanning
set of the subspace;
• find bases for the sum and intersection of two subspaces of n .

6.2 ELEMENTARY MATRICES

Recall that, in Unit 4, we discussed three kinds of operations on a system of
equations.
62
1. Interchanging two equations.
2. Multiplying the ith equation by a nonzero constant.
3. Multiplying the ith equation by a constant a and adding it to the jth
equation.

All these operations correspond to row operations on the matrix associated

with the system. As we will see, the row operations on a matrix associated
with the system of equations is equivalent to multiplication of the matrix on the
left by certain matrices. We call these matrices elementary matrices.

Consider the operation of interchanging the ith row and jth row of an m  n
matrix. We can do this by multiplying the matrix on the left by the matrix
obtained by interchanging the ith and jth rows of an m  m identity matrix. For
example, to interchange the second and third rows of the matrix.
 1 1 1 4
 0 0 −3 −6  (1)
 2 4 −2 2
 
we multiply this matrix by the matrix
1 0 0
 0 0 1
0 1 0 
 
which is the matrix we get by interchanging the 2nd and 3rd rows of the 3  3
identity matrix. Check that
1 0 0 1 1 1 4  1 1 1 4 
 0 0 1 0 0 −3 −6  = 2 4 −2 2 
0 1 0  2 4 −2 2  0 0 −3 −6 
     
We can also look at this from the point of view of looking at Eqn. (1) as the
matrix equation

 1 1 1   x1   4 
 0 0 −3   x 2  =  − 6  (2)
 2 4 −2   x   2 
   3  
1 0 0
Multiplying both sides of Eqn. (2) by the matrix 0 0 1 , we have
0 1 0 
 

1 0 0  1 1 1   x1   1 0 0   x1   1 0 0   4 
 0 0 1 0 0 −3   x 2  = 0 0 1  x 2  = 0 0 1  −6 
0 1 0  2 4 −2   x  0 1 0   x  0 1 0   2 
    3   3   
or

 1 1 1   x1   4 
 2 4 −2   x 2  =  2 
0 0 −3   x   −6 
   3  
More generally, we have

63
 i j  a11 a12 ain b1 
1   a21 a22 a2n b2 
   
i 0 1  a ain bi 
   i1 ai2
   
j 1 0   a j1 a jn b j 
   
 1 a
 m1 am2 amn bm 

 a11 a12 ain b1 

 a21 a22 a2n b2 
 
a a j2 a jn b j 
=  j1 (3)
 
 ai1 ai2 ain bi 
 
a amn bm 
 m1 am2
where the first matrix on the LHS is the matrix we get by interchanging the ith
and jth rows of the n  n identity matrix. (All the entries along the diagonal,
except for the entries in the ith row, ith column and the jth row, jth column are 1.
All the other entries that are not filled are zero in the first matrix on the LHS.)

In terms of matrix equation, we have

i j
 a11 a12 ain   x1 
1   a21 a22 a2n   x2 
     
i 0 1  a ain  x 
   i1 ai2  i
     
j 1 0   a j1 a jn   xj 
     
 1 a
 m1 am2 amn  x 
 m
i j
 b1 
1   b2 
   
i 0 1  b 
=    i (4)
   
j 1 0   bj 
   
 1 b 
 m
So,
 a11 a12 ain   x1   b1 
 a21 a22 a2n   x 2   b2 
    
a a jn   xi   b j 
 j1 = (5)
    
 ai1 ai2 ain   x j   bi 
    
a amn   xm  bm 
 m1 am2

64
Consider the operation of multiplying the ith row of a matrix by a constant
c, Ri → cRi . For this we multiply the matrix corresponding to the set of
equations on the left by the matrix we get by replacing the 1 in the ith row ith
column by c. For example, to multiply the last row of the matrix

 1 −1 1 2
 0 2 −4 −6  (6)
 0 0 −3 −6 
 
1
by − we multiply this matrix on the left by the matrix
3
 
1 0 0 
0 1 0 
 1
0 0 − 
 3
Check that

 
 1 0 0   1 −1 1 2   1 −1 1 2 
0 1 0  0 2 −4 −6  = 0 2 −4 −6 
 1  0 0 −3 −6  0 0 1 2 
0 0 −  
 3
If we write Eqn. (6) in the form of a matrix equation, we have

 1 −1 1   x1   2 
0 2 −4   x 2  =  −6  (7)
0 0 −3   x   −6 
  3  
Multiplying both sides of Eqn. (7) by
 
1 0 0 
0 1 0 
 1
0 0 − 
 3
we get

 1 −1 1   x1   2 
0 2 −4   x 2  =  −6 
0 0 1   x   2 
  3  
More generally, we have
i
1 0 0 0   a11 a12 ain b1 
0 1 0 0   a21 a22 a2n b2 
  
i 0 0 c 0   ai1 ai2 ain bi 
 
  
0 0 1 am1 am2 amn bm 

65
  a11 a12 ain b1 
 a21 a22 a2n b2 
 
=
ac in cbi 
(8)
cai1 cai2
 
 am1 am2 amn bm 
In terms of matrix equation, we have
i
1 0 0 0   a11 a12 ain   x1 
0 1 0 0   a21 a22 a2n   x2 
    
i 0 0 c 0   ai1 ai2 ain  x 
 i
 
    
0 0 1 am1 am2 amn   xm 
i
1 0 0 0  b1 
0 1 0 0  b2 
    (9)
=  b 
i 00 c 0  i
 
   
0 0 1 bm 
So,

 a11 a12 ain   x1   b1 

 a21 a22 a2n   x 2   b2 
    
ca =

ac in   xi  cbi 
(10)
 i1 cai2
    
 am1 am2 amn   xm   bm 

Consider the operation multiplying the ith row by a constant c and adding it to
the jth row. For this we multiply by the matrix we get by replacing the zero
entry in the ith column of the jth row of the n  n identity matrix by c. For
example to multiply the first row of the matrix
 1 −1 1 2 
 2 4 −2 2  (11)
 0 0 −3 −6 
 
by −2 and add it to the second row, we multiply the matrix by
 1 0 0
 −2 1 0 
 0 0 1
 
Check that
 1 0 0  1 −1 1 2   1 −1 1 2 
 −2 1 0  2 4 −2 2  = 0 6 −4 −2 
 0 0 1 0 0 −3 −6  0 0 −3 −6 
     
Again, writing Eqn. (11) in the form of a matrix equation, we get

66
  1 −1 1   x1   2 
2 4 −2   x 2  =  2  (12)
0 0 −3   x   −6 
   3  
Multiplying both sides of Eqn. (12) by
 1 0 0
 −2 1 0 
 0 0 1
 
we get

 1 −1 1   x1   2 
0 6 −4   x 2  =  −2 
0 0 −3   x   −6 
  3  
More generally, we have
i j  a11 a12 a1n b1 
1  a a22 a2n b2 
   21 
i 1 
ain bi 
   ai1 aij
  
j c 1   a j1 a j2 a jn b j 
  
 1 
am1 am2 amn bn 

 a11 a12 ain b1

 a21 a22 a2n b2
 
 a ai2 ain bi 
= i1

 
a j1 + cai1 a j2 + cai2 a jn + cain b j + cbi 
 
 a bm 
 m1 am2 amn
In terms of matrix equation, we have
i j  a11 a12 a1n b1   x1 
1  a a22 a2n b2   x 2 
   21  
i 1 
ain bi   xi 
   ai1 aij
 
   
j c 1   a j1 a j2 a jn b j   x j 
   
 1 
am1 am2 amn bn   xm 
i j
 b1 
1   b2 
   
i 1  b 
=    i
   
j c 1   bj 
   
 1 b 
 m
That is

67
  a11 a12 a1n   x1   b1 
 a21 a22 a2n   x 2   b2 
     
 a ai2 ain  x   b 
 i1  i= i 
     
a j1 + cai1 a j2 + cai2 a jn + cain   x j  b j + cbi 
     
 a amn  x   b 
 m1 am2  m  m 

Let us now look at an example to better understand our discussion so far.

Example 1: The first column in the following table gives the order of the matrix
on which one of the row operations is to be performed, the second column
gives the row operation to be performed. Write down the elementary matrix
you will use to perform the row operation.

Table 1

No. Order of the matrix Row operation

1 43 R1  R3
2 45 R 2 → R 2 − 2R1
3 54 1
R4 → R4
3
4 44 1
R1 → R1 + R3
5

Solution:
1. The matrix is a 4  3 matrix. The elementary matrix should have four
columns so that we can multiply the matrix on the left by the elementary
matrix. Since the elementary matrices are square matrices, the required
elementary matrix will be a 4  4 matrix. Since we want to interchange
the first and third rows of the matrix, the elementary matrix will be the
matrix we get by interchanging the first and third rows of the 4  4
identity matrix. So, the matrix is
0 0 1 0 
0 1 0 0 
1 0 0 0
 0 0 0 1
 

2. Reasoning as before, the elementary matrix is a 4  4 square matrix.

Since we want to multiply the first row by −2 and add it to the second
row, we replace the zero in the second row, first column of the 4  4
identity matrix by −2. The required elementary matrix is
 1 0 0 0
 −2 1 0 0 
 0 0 1 0
 0 0 0 1
 
One easy way to remember this is to note that the second row is to be
changed, so we have to change the entry in the second row of the 4  4 .
The column will be one because we will be adding −2 times the first
row.

68
3. The elementary matrix will have order 5  5 . Since we are multiplying
1
the fourth row by , we replace the 1 in the fourth row, fourth column of
3
1
the 5  5 identity matrix by . So, the elementary matrix is
3
1 0 0 0 0
0 1 0 0 0
0 0 1 0 0
 1 
0 0 0 0
 3 
0 0 0 0 1

4. The elementary matrix will have order 4  4 . We are changing the first
1
row in the row operation and we want to add times the third row to the
5
first row. So, we change the element in the third column of the first row
1
of the 5  5 identity matrix from 0 to . So, the required matrix is
5
 1 
1 0 5
0
0 1 0 0
0 0 1 0 

0 0 0 1
***
Try the next exercise to check your understanding of Example 1.

E 1) The first column in the following table gives the order of the matrix on
which one of the row operations is to be performed, the second column gives
the row operation to be performed. Write down the elementary matrix you will
use to perform the row operation.

Table 2

No. Order of the matrix Row operation

i) 45 R1  R 4
ii) 53 R 2 → R 2 − 2R1
iii) 43 1
R4 → R4
3
iv) 44 1
R1 → R1 + R3
5

We can describe the elementary matrices in terms of matrix units E ij which

are n  n square matrices defined as follows:

E ij is an n  n matrix with 1 as the entry in the ith row, jth column and all the
0 0 0 
other entries are zero. For example, the 3  3 matrix unit E23 = 0 0 1 . We
0 0 0 
 
have the following rule for matrix units:
69
 Eij Ek =  jkEi (13)

where  jk is the Kronecker  function. It follows that

EijEij = 
0 if i  j
Eii if i = j
(14)

Note that {Eij | 1  i  n, 1  j  n} forms a basis for Mn ( ) as a vector space

n n
over . Indeed, we have (aij )nn =  aijEij .
i=1 j =1

We can write the elementary matrices in terms of matrix units as follows:

1) The n  n elementary matrix that multiplies the ith row by c and adds it
to the jth row is given by I + cE ji . For example consider the matrix
 1 0 0
 −2 1 0  which multiplies the first row by −2 and adds it to the second
 0 0 1
 
 1 0 0 0 0 0   1 0 0 
 
row. We have −2 1 0 = −2  1 0 0  + 0 1 0  = ( −2)E 21 + I.
 0 0 1  0 0 0  0 0 1 
     

2) The n  n elementary matrix that interchanges the ith and jth rows is
1 0 0
given by I − Eii − E jj + Eij + E ji . For example the matrix 0 0 1
1 0 0
 
interchanges the second and third row.
We have
 1 0 0   1 0 0   0 0 0   0 0 0  0 0 0  0 0 0 
0 0 1 = 0 1 0  − 0 1 0  − 0 0 0  + 0 0 1 + 0 0 0 
0 1 0  0 0 1 0 0 0  0 0 1 0 0 0  0 1 0 
           
= I − E22 − E33 + E23 + E32 .

3) The matrix that multiplies the ith row by c, c  0, is given by

 
1 0 0 
I + (c − 1)Eii . For example consider the matrix 0 1 0  that multiplies
 1
0 0 − 
 3
1
the third row by − . We have
3
   
 1 0 0   1 0 0  0 0 0  0 0 0 
 0 1 0  =  0 1 0  − 0 0 0  + 0 0 0 
 1  0 0 1 0 0 1  1
0 0 −   0 0 − 
 3  3
 1 0 0   4  0 0 0   1 
= 0 1 0  +  −  0 0 0  = I +  − − 1 E33 .
 0 0 1  3   0 0 1  3 
   

70
Recall that, an n  n matrix A is invertible if there is an n  n matrix B such
that AB = In , the n  n identity matrix. We denote the inverse matrix of A by
A −1 and write A −1 = B. Note that, if AB = I, we have BA = I.

We now prove a simple fact that will be useful later.

Lemma 1: If A and B are two m  m matrices, the AB is invertible if and

only if A and B are invertible. If AB is invertible, (AB)−1 = B−1A −1. If
A1, A 2 , , A n are m  m matrices, then the product A1A 2 A n is invertible
iff each A i , 1  i  n is invertible. If A1A 2 A n is invertible we have
(A1A 2 A n )−1 = A n−1 A1−1.

Proof: If A and B are invertible matrices, A −1 and B −1 exist and we have

ABB−1A −1 = A(BB−1 )A −1 = AA −1 = I. So, AB is invertible.

Suppose AB is invertible, that is there is an m  m matrix C such that

ABC = I. By associativity of matrix multiplication We have A(BC) = I.
Therefore A is invertible and BC is the inverse A. Also, since A(BC) = I, we
have (BC)A = I or B(CA) = I. So, B is also invertible and CA is the inverse of
B.

To prove the general result, we use induction on n. It is obviously true for

n = 1 and we have proved the result for n = 2. Let us now suppose that the
result is true for n  3. Suppose A1, A 2 , , A n+1 are invertible, m  m ,
matrices. By induction on n, A1 A n is invertible. Since the result is true for
n = 2, (A1A 2 A n )A n+1 is invertible.

Conversely, suppose that the product A1A 2 A n+1 is invertible. Writing

(A1A 2 A n )A n+1 = I, it follows that A1A 2 A n and A n+1 are invertible. By
induction, since the product A1A 2 A n is invertible, each of A1, A 2 , , An
are invertible.

The next proposition shows that all the elementary matrices are invertible and
their inverses are also elementary.

Proposition 1: All the elementary matrices are invertible. We have

a) (I + cEij )−1 = I − cEij when i  j.
b) (I − Eii − E jj + Eij + E ji )−1 = I − Eii − E jj + Eij + Eij , i  j.
1 
c) I + (c − 1)Eii = I +  − 1 Eij when c  0.
c 

Proof: a) (I + cEij )(I − cEij ) = I − cIEij + cEijI − c EijEij

2 2

= I − cEij + cEij + c 2EijEij.

Using Equation (14), we have EijEij 0. Therefore (I + cEij ) (I − cEij ) = I.

b) Let us write  = (Eii + E jj ),  = (Eij + E ji ). Then

71
  = EiiEij + EiiE ji + E jjEij + E jjE ji
= Eij + 0 + 0 + E ji = .

 = EijEii + EijE jj + E jiEii + E jiE jj

= 0 + Eij + E ji + 0 = 

2 = (Eii + E jj ) (Eii + E jj ) = EiiEii + EiiE jj + E jjEii + E jj + E jj

= Eii + 0 + 0 + E jj = 

 2 = (Eij + E ji ) (Eij + E ji ) = EijEij + EijE ji + E jiEij + E jiE ji

= 0 + Eii + E jj + 0 = .

(I − Eii − E jj + Eij + E ji ) (I − Eii − E jj + Eij + E ji ) = (I −  +  ) (I −  +  )

= I −  +  −  + 2 −  +  −  +  2
= I −  +  −  +  −  +  −  +  = I.

 1   1 
c) (I + (c − 1)Eii )  I +  − 1 Eii  = I +  − 1 Eii + (c − 1)Eii
 c   c 
1 
+(c − 1)  − 1 EiiEii
c 
1 1 
= I +  − 1 + c − 1 + 1 − c − + 1 Eii = I.
c c 

The next theorem gives a characterisation of inversible matrices.

Theorem 1: Let A be an m  m square matrix. Then the following are

equivalent.

a) A can be reduced to the identity matrix by a sequence of elementary row

operations.
b) A is a product of elementary matrices
c) A is invertible.
d) The system of homogenous equations Ax = 0 has only the trivial solution
Ax = 0.

Proof: a)  b). Since each row operation corresponds to multiplication on the

left by elementary matrices, they are elementary matrices E1,E2 ,...,Ek such
that Ek ,Ek −1,...,E2E1 A = I. Since elementary matrices are invertible,
multiplying both sides by E1−1 E2−1....Ek−1, we get A = E1−1E2−1....Ek−1. Since the
inverse of elementary matrices are also elementary matrices each of
E1−1,E2−1,...,Ek−1 are elementary matrices and we are done.

b)  c) Immediate from Lemma 1.

c)  d) Since A is invertible, multiplying both sides of the equation Ax = 0 by
A −1, we get x = 0.

72
d)  a). Let A  be the RREF of A. We claim that A  = 1. Since row operations
do not change the solution set of a system of homogenous equations, the only
solution of A x = 0 is x = 0. So, A  does not have any zero row; otherwise the
number of variables will be greater than the number of equations and the
system will have a solution different from the zero vector. This means that
each of the m rows of A  should have a pivot, so there are m pivots. So, each
column has exactly one pivot which is 1. Because the column containing the
pivot in a row is to the right of the column containing the pivot in the previous
row, the columns of A  form an m  m identity matrix. So, there are elementary
matrices E1,E2 ,...,Ek such that EkEk −1...E1A = A  = I or A = Ek−1Ek−−11...E1−1, a
product of elementary matrices.

Using theorem 1, we can find the inverse of a square matrix by row reduction
let us see how.

Suppose A is an m  m invertible, square, matrix. Consider the m  2m matrix

B =  A | I where I is the m  m identity matrix. We apply row operations to
B that reduces the submatrix A to RREF. B is transformed to B = [I | C] since
the RREF of A is the identity matrix. We claim that C = A −1. Suppose
A = Ek ...E1. Then A gets reduced to its RREF on multiplication by E1−1...Ek−1 on
the left, so we multiply B on the left by this product E1−1....Ek−1B = [I|E1−1....Ek−1 ]
so C = E1−1E2−1...Ek−1 = (EkEk −1...E1 )−1 = A −1 .

Let us now look at an example.

 1 −1 0 
Example 2: Find the inverse of the matrix  1 1 1 by row reduction.
 −1 0 1
 

Solution: We construct a 3  6 matrix by adding the columns of a 3  3 identiy

matrix. We get the matrix

 1 −1 0 1 0 0 
 1 1 1 0 1 0
 −1 0 1 0 0 1
 

Carrying out row operations R2 → R2 − R1,R3 → R1,R 2 → R 2/2 ,

2 R R
R1 → R1 + R2 ,R3 → R3 + R2 , R3 → R3 ,R1 → R1 − 2 , R2 → R2 − 3 we
3 2 R2
get

 1 0 0 1/ 3 1/ 3 −1/ 3 
0 1 0 −2 / 3 1/ 3 −1/ 3 
0 0 1 1/ 3 1/ 3 2 / 3 
 

−1
 1 −1 0   1/ 3 1/ 3 −1/ 3 
  1 1 1 =  −2 / 3 1/ 3 −1/ 3 
 −1 0 1  1/ 3 1/ 3 2 / 3 
   

Try the next exercise to check your understanding of Example 2.

73
  1 1 2
E1) Find the inverse of the matrix 0 −1 0  using row reduction.
 1 0 1
 

In the next section, we will consider the vector spaces spanned by the row
vectors of a matrix and the column vectors of a matrix. We will see how to find
their dimensions.

6.3 ROW SPACE, COLUMN SPACE AND NULL

SPACE OF A MATRIX
We begin this section by defining row space and column space of a matrix.

Definition 1: Let A = (aij )mn be an m  n matrix. Then, we can consider A as

a collection of row vectors u1 = (a11,a12 ,...,a1n ),u2 = (a21,a22 ,...,a2n ),...,
um = (am1,am2 ,...,amn ) where ui  n
,1  i  n. We define the row space of
A by

RS(A) =  {u1,u2 ,...,um } .

Alternately, we can consider A as a collection of column vectors

 a11   a21   an1 

 a21   a22   an2 
v1 =   ,v 2 =   ,...,v n =  
a  a  a 
 m2   m2   nm 

We define the column space of A by CS(A) = [{v1,v 2 ,...,v n }]

We call the dimension of RS(A) the row rank of A; we denote the row mark of
A by r (A) . We call the dimension CS(A) the column rank of A; we denote
the column rank of A by c (A). If A is an m  n , then r (A)  m and
c (A)  n.

You may have a question now. How do the row operations affect the row rank
and column rank of a matrix? The answer is that they are unaffected. First, we
investigate what is the effect of a row operation on the row rank of a matrix.

Proposition 2: Let A be an m  n matrix and B be an m  m matrix. Then

RS(BA)  RS(A). If B is inversible, RS(A) = RS(BA).

Proof: Let B = (bij )mm . Consider BA. Then, the rows of BA are
u1 = (b11u1 + b12u2 + ... + b1mum ),u2 = (b21u1 + b22u2 ,...,bm2um ),...
um = (bm1u1 + bm2u2 + ...bmmum ).

So, u1 ,u2 ,...,um   {u1,u2 ,...,um } . So,

RS(BA) = [{u1 ,u2 ,..,um }]  [{u1,u2 ,...,um }] = RS(A)
74
Suppose B is invertible. Let us write C = BA. Then RS(B−1C)  RS(C) or
RS(A)  RS(BA).
Since we have RS(BA)  RS(A) and RS(BA)  RS(A) we have
RS(BA) = RS(A) if B is invertible. In particular, r (A) = r (BA) since
RS(A) = R(BA) if B is invertible. ■

Corollary 1: Elementary row operations do not affect the row rank of a matrix.
In particular, if A is an m  n matrix and A  is its RREF, RS(A) = RS(A ) and
r (A) = r (A ).

Proof: We perform a row operation on A by multiplying it on the left by an

elementary matrix E. We have A  = Ek ....E1A and E1....Ek is invertible since
each Ei 1  i  k, is invertible. By Proposition 2, the result
follows. ■

The next proposition tells us how we can calculate the row rank of a matrix by
row reduction.

( )
Proposition 3: Let A be an m  n matrix and suppose A  = aij is the REF of
A. The non zero rows of A  form a basis for RS(A). Also, the row rank of A
is the number of non zero rows in the REF of A.

Proof: We know that RS(A) = RS(A ). RS(A ) is spanned by its non zero
rows since we can always omit zero rows from the spanning set. It remains to
show that the nonzero rows are linearly independent.

Each nonzero row contains a pivot, namely one. Suppose there are k nonzero
rows and the pivots are in columns j1  j2  ...  jk . Let u1,u2 ,...,uk be the
k
nonzero rows and suppose that u
i=1
i i = 0.

 +  2a2 j(1) + ... + k ak j(1) = 0.

Comparing the j(1) component, we get 1a1j(1)
Since A  is in REF and j(1) is a pivot column a1 j(1) = 1 . Further
a2 j(1),...,ak j(1) lie in the pivot column below the pivot, so, they are zero. It
follows that 1 = 0. Similarly, comparing j(2),..., j(k) components
successively, we can prove that  2 ,  3 ,..., k are all zero. Therefore {ui } is
also linearly independent and we are done. From the fact that the nonzero
rows form a basis for RS(A ) it follows that
dim(RS(A)) = dim(RS(A )) = Number of nonzero rows of A . But
r (A) = dim(RS(A)) and the result follows. ■

Let us now look at an example to understand how we can apply this

proposition, given a spanning set of a subspace in n .

Example 3: Find a basis for the vector space V which is spanned by the
vectors {( −1,1,0,1),(0,1,1,0),(2,3,1,2),(1, − 1,1,0)} .

Solution: We form a matrix with these vectors as rows:

75
  −1 1 0 1
 0 1 1 0
A=
−2 3 1 2
 1 −1 1 0 
 
The row operations R1 → −R1,R3 → R3 + 2R1,R1 → R 4 − R1,R1 → R1 + R 2 ,
R3 → R3 − R2 ,R3  R 4 ,R1 → R1 − R3 ,R2 → R2 − R3 , gives

1 0 0 −2 
0 1 0 −1
0 0 1 .
1
0 0 0 0 


The first three nonzero rows (1,0,0, −2),(0,1,0, −1),(0,0,1,1) forms a basis for
V. The dimension of V is three.

***

Try the next exercise to check your understanding of this example.

E2) Find a basis for the vector spanned by the vectors (1,1,0,1),(1,1,1,1),
(2,2, − 2,2),(0,0,01). What is the dimension of V ?

We now discuss the relationship between the column rank of a matrix and the
column rank of its row reduced echelon matrix. We convert a matrix into
RREF by multiplying it on the left by an invertible matrix. What is the effect on
the columns of an m  n matrix A when we multiply A on the left by an
m  m matrix B ?

Suppose v1...,v n  are the columns of A. Then, the columns of BA = A  are

{Bv1,Bv 2 ,...,Bv n } . We will use this fact to prove that the column vectors of
A and BA satisfy the same linear relations.

Lemma 2: Let A be an m  n matrix and B be an invertible m  m matrix.

n
Suppose {v1,v 2 ,...,v n } are the column vectors of A. Then  v
i=1
i i =w
n
  iBv i = Bw.
i=1

n n
In particular, ivi = 0  iBvi = 0.
i=1 i=1

 n
 n
Proof: We have B    v  =   Bv . (See Block 1, miscellaneous
i i i i
 i=1  i=1
k k
 k 
exercises, example 4) Therefore, if  v i i = 0,  iBv i = B   i v i  = Bw
i=1 i=1  i=1 

76
 k
 k 
If B is invertible and   iBv i = Bw. We have B   i v i  = Bw. Multiplying
i=1  i=1 
k
both sides of this equation by B −1, we get  v
i=1
i i = w.

Proposition 4: Let A be an m  n matrix and A  be the RREF. The columns of

A corresponding to pivot columns in A  form a basis for CS(A). Further
r (A) = c (A).

Proof: Suppose r (A) = k. Then A  is of the form


 a11 
a12  
a1n
a21 a22 a2n 
 
 a ak 2 akrn 
 k1
0 0 0 0 
 
 0 0 0 0 

That is, in each column vector the last k − r rows are zeros. Therefore, we can
write each column vector as a linear combination of e1,e2 ,...,ek , where e i is a
column vector of length m with a one in the ith row and 0 in the other rows:

 a1i 
 a2i   1 0 0
   0   1  0
   = a      
 
1i   + a 2i   + ... + aki  
 aki  0 0 1
0      
       
0 0 0 0
 

Therefore CS(A )  S[{e1,e2 ,...,ek }]

On the other hand, since there are k nonzero rows there are k pivot entries.
Since no two pivot entries are in the same row or same column, it follows there
are k distinct pivot columns. Each pivot column is e j for some j, 1  j  k,
since each pivot column has one in a particular row and zero in the other rows.
If v j1 ,v j2 ,...,v jk , j1  j2   jk , are the pivot columns in A , v j1 has one in the
first row and zeros in the other rows, v j2 has 1 in the second row and zeros in
the other rows etc. So, v ji = ei for 1  i  k. Therefore,
{vj1 ,vj2 ,...,vjk } = {e1,e2 ,...,ek }
Therefore {e1,e2 ,...,ek }  CS(A ). So, S[{e1,e2 ,...,er }]  CS(A ) and
therefore CS(A ) = S[{e1,e2 ,...,ek }] = S[{v j1 ,...,v jk }]. In particular c (A ) = k.

Since BA = A  , there are columns v j1 ,...,v jk in A such that vjp = Bv jp for

1  p  k.

77
   
We claim that v j1 ,...,v jk forms a basis for CS(A). Since vj ,...,v jk is a
1

linearly independent set, by lemma 2, {v j1 ,...,v jr } is also a linearly independent
set. We need to show that this set spans CS(A). Let w  CS(A). Then
k
Bw  CS(A ), therefore there are 1,  2 ,..., k such that   v
i=1
i ji = Bw or
k

  Bv = Bw. Multiplying both sides of the equation by B , we get

i=1
i ji
−1

  v = w. In particular, CS(A) has a basis consisting of r elements.

i ji

Therefore, c (A) = k = r (a). ■

We now define the rank of a matrix.
Definition We define the rank of a m  n matrix A, denoted by (A) , by
(A) = c (A) = r (A)
In other words, the rank of a matrix is the common value of row rank and
column rank.
Remark 1: While CS(A) and CS(A ) have the same dimension, they are not

. For example consider A = 

m 1 3
the same subsets . Since
2 6 
3  = 3  1 , we have CS(A) = S    1  . The RREF of A is A =  1 3  and
6  2   2  0 0 
  
3  = 3  1 . So, CS(A ) = S    1  .
0  0    0  
  

So, CS(A) and CS(A ) have the same dimension, but they are different
m
subspaces of .

In the next example we will see how to compute a basis for the column space
of a matrix.

Example 4: Find a basis for the column space of the matrix

 −1 1 0 −2
0 1 0 −1
A=
−2 3 1 1
 1 −1 0 0 


Solution: We saw in Example 4 that the RREF of this matrix is

1 0 0 −2 
0 1 0 −1
0 0 1 1
0 0 0 0 


The first three columns are the pivot columns of the RREF of A . So, by
Proposition 4, the first three columns of A, namely

 −1  1  0 
 0  1  1
 −2 ,  3  and  1
 1   −1  1
     
78
forms a basis for CS(A).
***

Try the next exercise to check your understanding of example?

1 1 0 1
1 1 1 1
E3) Find a basis for the column space of  .
22 −2 2
0 0 1
 0

According to Proposition 4, the column space of a matrix has a basis

consisting of some of the column vectors of the matrix. Using this, given a
spanning set of a vector space in m , we can find a basis of the vector space
which is a subset of the spanning set. We form a matrix whose column vectors
are the vectors in the spanning set. We find RREF of this matrix and find a
basis for the column space of the matrix using Proposition 4. This basis is the
basis of vector space we started with.

The important thing to note is that the basis we get is a subset of the spanning
set we were given. This is not true if we write vectors in the spanning set of
vectors of a subspace V as row vectors of a matrix, row reduce the matrix and
take the nonzero vectors of the REF as the basis for the subspace V. The next
example illustrates this procedure.

Example 5: Find a basis for the vector space spanned by the set
S = {(1, −2,1,0),(1, − 1,2,0),(0,1,1,0),(0,0, −2,1)} which is a subset of the
spanning set S.

Solution: Arranging the vectors as the columns of a matrix, we get

1 1 0 0
 −2 −1 1 0
A=
1 2 1 −2
0 1 
 0 0

Carrying out row operations R2 → R2 + 2R1,R3 → R3 − R1,R1 → R1 − R 2 ,

 1
R3 → R3 − R2 ,R3 → −   R3 ,R4 → R4 − R3 we get
2

1 0 −1 0 
0 1 1 0
0 0 .
0 1
0 0 0 0 


First, second and the fourth columns are pivot columns. So, writing the
corresponding columns in A as row vectors, we see that
{(1, − 2,1,0),(1, − 1,2,0),(0,0, − 2,1)} is a basis for V.
***

79
 E4) Find a basis for the vector space spanned by S = {(2, − 2, − 1, − 2),
(3, − 1, − 1, − 2),(1,1,0,0),( −2,1,1,2)} which is a subset of S.

We now turn our attention to the third space associated with a matrix; the Null
space of a matrix. We begin with the definition of Null space of a matrix. We
begin with the definition of Null space of a matrix.

Definition 2: Let A be an m  n matrix. Then, we define N(A), the null space

of A by N(A) = b   n

Ab = 0 . We define the nullity of A to be the
dimension of N(A). We denote the nullity of A by null (A).

Note that N(A) is the solution set of the system of homogeneous linear
equations Ax = 0. We have already seen that the solution set of a system of
homogeneous linear equations in n unknowns is a vector subspace of n .
From our definition of N(A) also, if Ab1 = 0 and Ab 2 = 0, we have
A ( b1 + b2 ) = Ab1 + Ab2 = 0. So, it follows that N(A) is a subspace of
n
.

In the next lemma we prove that the null space of a matrix and the null space
of the RREF of the matrix are the same.

Lemma 3: Let A be an m  n matrix and B be an m  m invertible matrix.

Then, N(A) = N(BA). In particular, N(A) = N(A ) where A  is the RREF of
A.

Proof: x  N(A), then Ax = 0. Multiplying both sides by B, we get BAx = 0

or x  N(BA). So, N(A)  N(BA). On the other hand if x  N(BA), we have
BAx = 0. Since B is invertible, multiplying both sides of the equation
BAx = 0 by B −1, we get Ax = 0, so, x  N(A). Therefore, N(BA)  N(A)
and N(A) = N(BA). In particular, there is an invertible matrix E such that
A  = EA. Therefore
N(A ) = N(EA) = N(A).

The next theorem gives the relationship between the rank and nullity of a
matrix.

Theorem 2 (Rank-Nullity): Let A be an m  n matrix. Then, we have

(A) + null(A) = Number of columns of A.

Proof: Let {v1, v 2 , , v n } be the columns of A. Then


col(A) = 1v1 +  2 v 2 + + n v n i  n

  1  
 
=  A   1,  2 , , n  
 n  

Let f1, , fr  be a basis of N(A). This is linearly independent set which can
be extended to a basis f1, f2 , , fn  of n
. Then,
80
 n
 n  n
v =  i fi  Av = A   i fi  =  i Afi
i=1  i=1  i=1
We have Afi = 0 for 1  i  r since f1, f2 , , fr  N(A). Therefore
n
Av =  Af . In other words, Af
i=r +1
i r +1, Afr + 2 , , Afn  spans col(A). We claim

that Afr +1, Afr + 2 , , Afn  is linearly independent set. Suppose there are
n
 n 
r +1, r + 2 , , n , such that  i Afi = 0. Then A   i fi  = 0. Therefore
i=r +1  i=r +1 
n

  f  N(A), and we can write this as a linear combination of f , f ,

i=r +1
i i 1 2 , fr .

 n
 r
Therefore, we have   i i  =  ifi for some 1, 2,
 f , r  .
 i=r +1  i=1
r n
Therefore  (− )f +   f = 0. Since f , f ,
i=1
i 1
i=r +1
i i 1 2 , fn  is a linearly

independent, it follows that i = 0 for 1  i  n. It follows that {Afr +1, , Afn } is

a basis for col(A). Therefore, rank (A) = dim col(A) = n − r where r is the
nullity of A. The theorem now follows.

We will now see a nice application of Theorem 2 in the next example.

Example 6: Let us now go back to Example 7 in Unit 4 where we discussed

the solution set for the system of equations
x1 + x 2 + x 3 + x 4 = 0

x1 + 2x 2 + x 3 + x 4 + x 5 = 0

2x1 + x 2 + 2x 3 + 2x 4 − x 5 = 0

x1 + 6x 2 + 2x 3 + 2x 4 + 4x 5 = 0.

We saw that the RREF of the augmented matrix was

1 0 1 1 −1 0 
0 1 0 0 1 0
0 0 0 0 .
0 0
0 0 0 0 0 0 

We can write the system of equations in the form Ax = 0, where

 x1 
1 1 1 1 0  x2 
1 2 1 1 1
A= , x =  x3 
2222 −1  
2 6 2 2 −4  x4 

 x 5 
Since the last column of the augmented matrix associated with the system of
equations in the zero column, it doesn’t play any role in the computation of
RREF. So, the RREF of A is

81
 1 0 1 1 −1
0 1 0 0 1
0 0 0 0 .
0
0 0 0 0 0 

The number of nonzero rows in this matrix is 2, so the rank of the matrix is
three. The number of columns is five. Therefore the nullity is 5 − 2 = 3.
The solution set of the system of equations is the null space of A. We have
already seen that the set

  −1  −1  1 
 0   0   −1 
  0  
S =  1  ,  0  ,
 0   1   0 
  0   0   
 1 
spans the solution set and therefore spans N(A). Since
| S | = dim(N(A)) = nullity of A = 3, it follows from E16), Unit 5, that S is
actually a basis for N(A).
***

We will now see how to compute a basis for the sum and intersection of two
subspaces of n , if we are given a spanning sets of the subspaces.

Let V1, V2 be subspaces of n

. Let U = {u1, u2 , , uk } and V = {v1, , vr }
be spanning sets of V1 and V2 .
 a1i   b1i 
a2i  b 
Let ui =   and v i =  2i  .
a  b 
 ni   ni 
We form the matrix A = [u1 u2 uk v1 v 2 v n ] and find the RREF of A  .
Then, the columns in A corresponding to the pivot columns in A form a basis
for V1 + V2 because the columns of A span V1 + V2 .
To find a spanning set for V1  V2 we find a basis for the null space N(A).
 c i1 
c i2 
 
 
Suppose the basis is {X1, X2 , , Xm }, where Xi = c ik . Then, we have
 
 di1 
 
 dir 
c i1u1 + c i2u2 + + c ikuk + di1v1 + di2 v 2 + dir v k = 0. Consider the set
k r k
{w1, w 2 , , w m } where w i =  c ijui =  dik v i  V1  V2 since c u  V
ij i 1
j=1 k =1 j=1
r
and d
k =1
ik v i  V2 .  [{w1, w 2 , , w m }]  V1  V2 .
k r
Suppose v  V1  V2 . Then, v =  piui =  qivi.
i=1 x=

82
  p1 
 
k r p 
  piui +  ( −qi )v i = 0. Therefore, x 0 =  k   N(A). So,
i=1 i=1  −q1 
 
 −qr 
 c j1 
c   p1 
 j2   
m m    pk 
x 0 =   y X j =   j c jk  =  .
− q
j=1 j =1  d j1   1 
   
   r  − q
 d jk 
m
If follows that pi =  C .
j=1
j ij

k  m 
k m
 k  m
 v =  p u =     jc j  u =   j   c iju  =   j w j .
=1 =1  j=1  j=1  =1  j=1
Therefore, V1  V2  [{w 1, w 2 , , w m }] . So, V1  V2 [{w 1, w 2 , , w m }].

Remark 2: If {u1, u2 , , uk } is a basis for V1 and {v1, v 2 , , v r } is a basis

for V2 , {w1, w 2 , , w m } is a basis for V1  V2 . Otherwise {w 1, , w m } is
only a spanning set for V1  V2 . If we need a basis for V1  V2 , we can use
the method in Example 5 find a basis for V1  V2 .

We now summarise the method for finding the sum and intersection of the
subspaces V1 and V2 , given spanning sets {u1, u2 , , v k } for V1 and
{v1, v 2 , , v r } of V2 .

1) We form a matrix A = [u1 u2 uk v1 v 2

v r ] and row reduce this to get
A . The columns in A corresponding to the pivot columns of A form a
basis for V1  V2 .
We then compute a basis for the null space of A. Since A has r + k
columns, N(A)  r +k . Suppose {X1, X2 , , Xm } is a basis for the null
space of A. We can write

 x i1 
 x i2 
 
Xi = xik 

 
 − y i1 
 
 − yir 

We then compute w i as
k r
w i =  xiju j or w i = − yij v j ,
j=1 j=1

whichever is convenient. Then {w1, w 2 , , w m } spans V1  V2 .

83
 Let us now look at an example to understand this method.

  0   1     1  0   
  0   0       −1  1 
Example 7: Let V1 =     ,    , V2 =     ,    . Find a basis for
1 0 −1 0
             
      
1 1    −1  1 
V1 + V2 . Also find a spanning set for V1  V2 .

Solution: We form the matrix A consisting of the spanning vectors of V1 and

V2 :

0 1 1 0 
0 0 −1 1
A= .
1 0 −1 0 
 1 1 −1 1
 
1 0 0 −1
0 1 0 1
The RREF of A is A  =  .
001 −1
0 0 0 0 

The first, second and the fourth columns of A are the pivot columns of A .
So, the first, second and fourth columns of A form a basis for A. Therefore,
 0   1  1  
 0  0   −1 
  1 , 0  ,  −1  forms a basis for V1 + V2 .
      
  1  1  −1 
The fourth column is the only non-pivot column and so we set x 4 = . So,
N(A) is given by x1 −  = 0, x 2 +  = 0, x 3 −  = 0, x 4 = . So,

 1 
  −1 
N(A) =      .
1
   
 1 
 1
0 
Therefore, u1 − u2 = −v1 − v 2 =   spans V1  V2 .
1
0 
 
***

Let us look at another example.

   1  1   1      1  1  2  
   0   1  0      1 0   0  
Example 8: Let V1 =     ,   ,    V2 =     ,   ,    . Find a
0 0 1 1 1 −1
                 
   1  1   1      1 0   0  
basis for V1 + V2 and spanning set for V1  V2 . Is the spanning set a basis for
V1  V2 ?

Solution: We form the matrix consisting of span vectors of V1 and V2 :

84
 1 1 1 1 1 2
0 1 0 1 0 0
A= .
0 0 111 −1
1 1 1 1 0 0 

1 0 0 −1 0 3
0 1 0 1 0 0
The RREF is A  =  .
001 1 0 −3 
0 0 0 2 
 0 1
The pivot columns are the first, second, third and the fifth columns. So, the
corresponding columns in A,

  1  1  1  1 
 0   1 0  0  
 0  , 0  ,  1 ,  1 
        
  1  1  1 0  
form a basis for V1 + V2 .
We compute a basis for N(A). The non-pivot columns are the fourth and sixth
columns. So, we set x 4 = , x 6 = . N(A) is given by
x1 −  + 3 = 0, x 2 +  = 0, x 3 +  − 3 = 0, x 4 = , x 5 + 2 = 0, x 6 = .
x1 =  − 3, x 2 − , x 3 = − + 3, x 4 = , x 5 = 2, x 6 = .

 x1    − 3  1  −3 
 x 2   −   −1 0
 x   − + 3   −1 3
 3 =   =   +  .
x4     1
 
0
 
 5 
x −2  0
  −
 2
 x 6      
0 1
1
 −1
 −1
Corresponding to   since it is convenient to compute in terms of v 1 here,
1
 
0
 
0
we have
 −1
 −1
w1 = ( −1)v1 + 0.v 2 + 0.v 3 = −v1 =   ( = u1 − u2 − u3 )
−1
 −1
 
 −3 
0
3
corresponding to   , we have
0
 
 −2 
1
w 2 = −3u1 + 3u3 = 0.v1 + 2v 2 + ( −1)v 3 = 2v 2 − v 3

 1  2   0 
0   0  0 
= 2   −   =  .
1 −1 3
0   0  0 
     
85
   −1 0  
  −1 0  
Therefore,    ,    span V1  V2 . Check that w1 and w 2 are linearly
−1 3
    
  −1 0  
independent and hence form a basis for V1  V2 .

***

Try the next exercise to check your understanding of examples 7 and 8.

  1  0   1    1  2  
  1  −1 0    0   0  
E5) Let V1 =    ,   ,    , V2 =    ,   .
0 −1 1 0 −1
           
      
1 1 0    0  
1
Find a basis for V1 + V2 and a generating set for V1  V2 . Is the
spanning set for V1  V2 a basis for V1  V2 ?

The method we discussed above is convenient for computing by hand. (with

some machine computing for RREF). We next discuss an algorithm due to
zassenhaus which involves row reduction of a matrix of larger size. But, there
are other advantages:
1) We need only Row Echelon form, not Row Reduced Echelon form.
2) The algorithm gives a basis for V1  V21 not just a generating set.

3) The bases can be read off from the row echelon form.

The algorithm is simple to describe. The proof is not difficult, but involves,
concepts from the material in the second volumne.

We now describe the algorithm. Although the algorithm works for general
vector spaces we restrict our attention to subspaces of n . Let V1 and V2 be
n
subspaces of and suppose the rows of
 a11 a12 a1n 
a a a2n 
M1 =  21 22 
a a arn 
 r1 r 2
span V1 and the rows of the matrix

b11 b21 b1n 

b b b2n 
M2 =  21 22 
b b brn 
 k1 k 2
span V2 . We form the matrix

86
  a11 212 a1n a11 a21 a1n 
a21 a22 a2n a21 a22 a2n 
 
 arn 
 M1 M1   ar1 ar 2 arn ar1 ar 2
=
M2 0  b11 = A
b21 b1n 0 0 0 
b21 b22 b2n 0 0 0 
 
bk1 bk 2 bkn 0 0 0 
We use row operations to get the row echelon form of the matrix:

Then, the vectors Ci = (c i1 c i2 c in ), 1  i  q form a basis for V1 + V2 . The

vectors Di = (di1 di2 din ), 1  i  form a basis V1  V2 .

Let us now look at an example to understand this algorithm.

   1  1  1   1   0   0  
   1 0  0    0   1   0  
Example 9: Let V1 =     ,   ,    , V2 =    ,   ,   .
0 1 0 0 −1 −1
               
        
0 0 1 −
   0   1  
1
Find a basis for V1 + V2 and V1  V2 using Zassenhaus’ algorithm.

1 1 0 0   1 0 0 −1
Solution: We form the matrices M1 = 1 0 1 0  and M2 = 0 1 −1 0  and
1 0 0 1 0 0 −1 1 
   
the matrix
1 1 0 0 1 1 0 0
1 0 1 0 1 0 1 0
 M1 M1   1 0 0 1 1 0 0 1
A= = .
 M2 0   1 0 0 −1 0 0 0 0 
0 1 −1 0 0 0 0 0 
0 0 −1 1 0 0 0 0 
1 1 0 0 1 1 0 0 
0 1 −1 0 0 1 −1 0 
0 0 1 −1 0 0 1 −1 
The REF is 
0 0 0 1 1/ 2 0 0 1/ 2 
.
 
0 0 0 0 0 −1 1 0 
0 0 0 0 0 0 1 −1 

The set of rows of the 3  4 matrix at the top left, {(1, 1, 0, 0), (0, 1, − 1, 0),

87
 (0, 0, − 1, − 1),(0,0,0,1)} form a basis for V1 + V2 . The set of rows of the 2  4
matrix at the bottom right, {(0, − 1, 1, 0), (0, 0, 1, − 1)} form a basis for V1  V2 .
***

Try the next exercise to check your understanding of Zassenhaus’ algorithm.

  1  0     1  1 
        
E6) Let V1 =  1 , −1  V2 =  0  ,  1  .
  1  0     1 0  
      
Find bases for V1 + V2 and V1  V2 using Zassenhaus’ algorithm.

We have reached the end of this Unit. In the next section, we will summarise
what we have learned in this unit

6.4 SUMMARY
In this Unit we
• defined elementary matrices and explained the row operations
corresponding to them;
• proved that elementary matrices are invertible and given the inverse of
an elementary matrix;
• defined the row rank and the column rank and of a matrix;
• saw how to compute the rank of a matrix using row reduction;
• stated the rank-nullity theorem and seen some of its applications;
• saw how to find dimension and a basis for subspace of n given a
spanning set of the subspace;
• saw how to find bases for the sum and intersection of two subspaces
of n .

6.5 SOLUTION/ANSWERS

E2) Carrying out row operations R3 → R3 − R1,R1 → R1 − R 2 ,

R3 → R3 + R2 ,R3 = −R3 ,R1 → R1 − 2R3 gives

 1 0 0 −1 −1 2 
0 1 0 0 −1 0 
0 0 1 1 1 −1
 

 −1 −1 2 
A −1 =  0 −1 0 
 1 1 −1
 

E3) We form a matrix with these vectors as rows:

1 1 0 1
1 1 1 1
2 2 −2 2
0 0 1
 0

88
 carrying out row operations
R2 → R2 − R1,R3 → R3 − 2R1,R3 → R3 + 2R2 ,
1 1 0 0
0 0 1 0 
R3  R 4 ,R1 → R1 − R3 gives  . The first three rows,
0 0 0 1
0 0 0 0 
 
(1,1,0,0),(0,0,1,1) and (0,0,0,1) forms for V. Since the basis has three
elements, dim V = 3.

1 1 0 0
0 0 1 0 
E4) As we saw in E? the RREF of this matrix is  .
0 0 0 1
0 0 0 0 
 
The pivot columns are the first, third and the fourth columns. So, the
 1  0   1
 1  1   1
corresponding columns in A,   ,   and 2 form basis for the
2 −2
0   0   1
     
column space of A.

E5) Arranging the vectors as columns of a matrix, we get the matrix, we get

 2 3 1 −2 
 −2 −1 1 1 
 − 1 −1 0 1  .
 0 −1 −1 3 
 

R2
Carrying out row operations R2 → ,R2 → R2 + 2R1,R3 → R3 + R1,
2
R 3R2 R
R2 = 2 ,R1 → R1 − ,R3 → R3 − 2 ,R 4 → R 4 + R2 ,R3 → 4R3 ,
2 2 2
R R 5R3
R1 → R1 + 3 ,R2 → R2 + 3 ,R4 → R4 − ,
4 2 2

1 0 −1 0 
0 1 1 0
We get  .
00 0 1
0 0 0 0 


The pivot columns are the first, second and fourth columns. Writing the
corresponding columns as row vectors, the set
{(2, − 2, − 1,0),(3, − 1, − 1, − 1), ( −2,1,1,3)} forms a basis for V.

E6) Arranging the spanning vectors in a matrix, we get

1 0 11 2
1 −1 0 0 0
A= .
0 −1 1 0 −1
1 0 
 1 01
 1 0 0 1/ 2 0 
0 1 0 1/ 2 0 
The RREF of A is A  =  .
0 0 1 1/ 2 0 
 0 0 0 0 1
 
89
 The first, second, third and the fifth columns are pivot columns in A . So,
  1  0   1  2  
  1  −1 0   0  
the corresponding columns,    ,   ,   ,    form a basis for
0 −1 1 −1
        

 1  1  0   0  
V1 + V2 .
Fourth column is the non-pivot column. So, we set x 4 = . The null
space is given by
1 1 1
x1 +  = 0, x 2 +  = 0, x 3 +  = 0, x 4 = , x 5 = 0. So,
2 2 2
  −1/ 2 
  −1/ 2 
   
N(A) =  −1/ 2   
  1  
  0  
 
 −1
 u1 + u2 + u3  0
w1 = −   = −v1 + 0v 2 =   spans V1  V2 . V1  V2  {0}
 2  0
 −1
 
  −1 
 0  
and is spanned by a single element. Hence     forms a basis for
0
  
−
  1 
V1  V2 .

E7) Writing the spanning vectors of V1 and V2 as rows of matrices, we get

M1 = 
1 1 1
, M2 = 
1 0 1
.
0 −11 1 1 0 
We form the matrix
1 1 11 1 1
 M M  0 −1 1 0 −1 1
A =  1 1 =  .
M2 0   1 0 10 0 0
1 1 00 0 0 
 1 1 1 1 1 1
0 1 0 0 1 0 
The row reduced form is A  =  .
0 0 1 1 1 1
0 0 0 −1 0 1
 
The set of rows of the 3  3 matrix at the top left,
{(1, 1, 1), (0, − 1, 1), (1, 0, 1)} form a basis for V1 + V2.
The set containing the row at the bottom right, ( −1,0,1) , forms a basis
for V1  V2 .

90
Block Miscellaneous
Examples Exercises and
MISCELLANEOUS EXAMPLES AND EXERCISES
The examples and exercises given below cover the concepts and processes
you have studied in this block. Doing them will give you a better understanding
of the concepts concerned, as well as practice in solving such problems.

As we saw in Unit 4, Gaussian elimination is equivalent to reducing the

associated augmented matrix to REF. We look at one more example of
Gaussian elimination.

Example 1: Solve the following system of equations by reducing the matrix

corresponding to the system of equations using Gaussian elimination.

x1 + 2x 2 + x 3 = 8
x1 − x 2 + x 3 = 2
x1 + x 2 + 3x 3 = 8

Solution: The corresponding matrix

1 2 1 8 
1 −1 1 2
1 1 3 8 
 

We saw in Unit 4, Example 4 that the REF of this matrix is

 1 2 1 8
 0 1 0 2
 0 0 1 1
 

Writing down the corresponding equations, we have

x1 + 2x 2 + x 3 = 8
x2 =2
x3 = 1

Back substitution gives x 3 = 1,x 2 = 2,x1 = 3.

***

E1) Solve the following system of equations by reducing the matrix

corresponding to the system of equations to REF.
x1 + x 2 + x 3 = 3
− x1 + x 2 + 3x 3 = 4
2x1 + 2x 2 + x 3 = 5

Example 2: In this example, we will see how to determine the current flow in
an electrical laws using Kirchoff’s laws. We begin with the simple circuit in
Fig. 1.

91
Block 2

Fig. 1
In the figure symbol denotes a voltage source, usually a battery. The
bigger symbol on the right is the positive terminal and the smaller symbol on
the left is the negative terminal. The current always flows from the positive
terminal of the battery to the negative terminal of the battery and this is taken

as the positive direction.The symbol denotes a resistor. There is a

voltage source that supplies 20 volts current.

There are two resistors of resistance 2 ohms and 3 ohms. When the current
passes through a resistor, there is a drop in voltage. The voltage drop is
governed by Ohm’s law V = IR where V is the voltage, R is the resistance
and I is the current. We have the Kirchoff’s voltage law which states the
following:

Kirchoff’s Voltage Law: In a closed path within an electrical circuit, the

sum of the voltage drop in any direction is equal to the sum of the
voltage sources in the same direction.

In the figure above, there are two voltage drops, one is of 2I and the other is of
4I Their sum equals the voltage supplied by the battery. The drop in voltage is
2I+4I=6I and the voltage supplied is 30 volts. So, 6I+30 and I, the current flow
30
is = 6 amperes.
5

Let us now look at a slightly more complicated situation. Consider the figure
below:

Fig. 2
In this circuit, there are junctions at A and B, indicated by dots. There
are three branches, each with its current flow. The current flow at the
junctions follow another law, called the Kirchoff’s Current Law which is
as follows:

Kirchoff’s Current Law: The current flow into any junction is equal
to the current flow out of the junction.

We consider the closed path ABCD. Total voltage supply in this path is
zero. I2 is in a direction opposite to I1. We have
92
Block Miscellaneous
Examples Exercises and
(2 + 2)I1 − I2 = 0 or 4I1 − 2I2 = 0. In the closed path AEFB, the voltage
supply is 42 volts. The voltage drop is 2I2 + I3 . Therefore, 2I2 + I3 = 42 .

Finally, at the junction B, the inflow (towards B) is I3 and the outflow

(away from B) is I1 + I2 . By Kirchoff’s Current Law I1 + I2 = I3 or
I1 + I2 − I3 = 0 . Therefore, we get the equations
4I1 − 2I2 = 0
2I2 + I3 = 42
I1 + I2 − I3 = 0
The corresponding augmented matrix is
 4 −2 0 0 
 0 2 1 42 
 1 1 −1 0 
 
The RREF of this matrix is
 1 0 0 6
0 1 0 12 
0 0 1 18 
 
Therefore I1 = 6, I2 = 12, I3 = 18.

E 2) Find the current flow in each branch of the following circuit:

Fig. 3

E 3) Find the current flow in each branch of the following circuit:

Fig. 4

Example 3: Solve the following system of homogeneous linear equations:

x1 − x 2 − x 3 + 2x 4 = 0
2x 2 + 4x 3 − 4x 4 = 0
x 2 + 2x 3 − 2x 4 = 0
2x1 + x 3 = 0
Solution: The corresponding matrix is 93
Block 2

1 −1 −1 2 0
0 2 4 −4 0
0 1 2 −2 0
2 0 
 0 1 0
R 4 → R 4 − 2R1 gives
1 −1 −1 2 0
0 2 4 −4 0
0 1 2 −2 0
0 2 3 −4 0 

1
R2 → R2 gives
2
1 −1 −1 2 0
0 1 2 −2 0
0 1 2 −2 0
0 2 3 −4 0 

R1 → R1 + R 2 gives

1 0 1 0 0
0 1 2 −2 0
0 1 2 −2 0
0 2 3 −4 0 

R3 → R3 − R 2 gives
1 0 1 0 0
0 1 2 −2 0
0 0 0 0 0
0 2 3 −4 0 

R 4 → R 4 − 2R2 gives

1 0 1 0 0
0 1 2 −2 0
0 0 0 0 0
0 2 3 −4 0 

The third row is an all zeros row and this is followed by a non-zero row. We
interchange the third and fourth rows to get
1 0 1 0 0
0 1 2 −2 0 
0 0 −1 0 0 
0 0 0 0 0 

R3 → ( −1)R3 gives

1 0 1 0 0
0 1 2 −2 0 
0 0 1 0 0
0 0 0 0 0 

R2 → R 2 − 2R3 gives

94
Block Miscellaneous
Examples Exercises and
1 0 1 0 0
0 1 0 −2 0 
0 0 1 0 0
0 0 0 0 0 

R1 → R1 − R3

1 0 0 0 0
0 1 0 0 0
0 0 1 0 0
0 0 0 0 0 

The rank is 3 and these are 4 variables. So, there are 4 − 3 = 1 free variable.
However, the column corresponding to the free variable is zero. So, we get a
unique solution x1 = 0,x 2 = 0,x 3 = 0,x 4 = 0 and x 5 = 0.

***
Try the next exercise to check your understanding of the above example.

E4) Solve the following system of equations.

x1 + x 2 + x 4 = 0

x1 + x 2 + 2x 3 + x 4 = 0

x1 + x 2 + x 3 + x 4 = 0

x1 + x 2 + 4x 3 + x 4 = 0

In the next example, we will look at one more application of homogeneous

system of linear equations, balancing chemical equations.

Example 4: You would have studied various chemical reactions in school. For
reation, there is a chemican equation. For example, when hydrogen and
oxygen react, we get water. The chemical formula for the hydrogen molecule
is H2 and the chemical formula for oxygen is O2. We represent the reaction by
the equation
2H2 + O2 → 2H2O
Note the factor 2 in H2 and H2O. Without these factors, the number of
hydrogen atoms and oxygen atoms in the reactants and the products will not
be equal. We have added these factors to balance the equation. More often
than not, we can balance a chemical equation by trial and error. However,
some of the reactions are so complicated that we need to use Linear Algebra
to balance the equations. Consider the reaction of sodium sulfite (Na2SO4)
with nitric acid (HNO3). The products are sodium nitrate (Na2NO3), sulphur
dioxide (SO2) and water (H2O). Let us write the reaction in the form
x1Na2SO3 + x 2HNO3 → x 3NaNO3 + x 4SO2 + x 5H2O
We now write down the number of atoms of each element in the reactants and
products.
Reactants Products
Sodium 2x1 x3
Hydrogen x2 2x5
Sulphur x1 x4
Oxygen 3x1+3x2 3x3+2x4+x5

95
Block 2

Since the number of atoms in products and reactants are the same, we get the
following system of homogeneous equations:

2x1 − x3 = 0
x2 − 2x 5 = 0
x1 − x4 = 0
3x1 + 3x 2 − 3x 3 − 2x 4 − x5 = 0

The associated matrix is

 2 0 −1 0 0 
 0 1 0 0 −2 
A=
1 0 0 −1 0 
 3 −3 −2 −1
 3
The RREF of the matrix is
 1 0 0 0 −1
 0 1 0 0 −2
A = 
0 0 1 0 −2
 1 −1
 0 0 0
The column corresponding to x5 is the non-pivot column. So, we take x5 as the
free variable and set x 5 = . The solution is x1 = , x 2 = 2, x 3 = 2, x 4 = 
and x 5 = . So, the solution set is

 (1,2,2,1,1)   
Since we want an integer solution, we set  = 1 to get the solution (1,2,2,1,1).
So, the balanced equation is
Na2SO3 + 2HNO3 → 2NaNO3 + SO2 + H2O
We can multiply both sides of the equation by any integer to get infinitely many
solutions, but they are essentially the same.
***
Here is an exercise for you to try.

E 5) Balance the following chemical equation:

C6H12O6 → CO2 + C2H5OH

We saw how to check the consistency of a system of equations in Unit 4. One

application of this is to check whether a vector b 
n
is in the linear span of a
set of v1,v 2 ,, v k  
n
. Let us write
 ai1   b1 
 ai2  b 
vi =   , b =  2 
a  b 
 1n   n
Then, if b is in the linear span of v1,v 2 ,,v k  if and only if there are
1,  2 ,, k  such that 1v1 +  2 v 2 +
n
+ k v k = b. This is equivalent to
saying that the matrix equation
 1a11 +  2a12 + + k a1k   b1 
1a21 +  2a22 + + k a2k  b2 
 = 
 a +  a + + k ank  bn 
 1 n1 n n2

This means that 1,  2 ,, k 

n
is a solution to the system of equations
96
Block Miscellaneous
Examples Exercises and
1a11 +  2a12 + + k a1k = b1
1a21 +  2a22 + + k a2k = b2

1an1 + nan2 + + k ank = bn

We can check whether the above system of equations is consistent using
RREF. If the system of equations is consistent, b is in the linear span of
v1,v 2 ,,v k  . If the system of equations is consistent, we can also solve for
1,  2 ,, k . Let us now look at some examples now.

Example 5: Check whether the vector b = (0,0,3) is in the linear span of

v1 = (1,1,1) , v 2 = (1,1, −1) and v 3 = (5,3,0).
Solution: The vector b will be in the linear span of v1,v 2 ,v 3  if there are
1,  2 , and 3 such that 1v1 +  2 v 2 + 3 v 3 = b. Writing the vectors as
column vectors, b  v1,v 2 , v 3  if and only if the system of equations
1 + 2 + 5 3 = 0
1 + 2 + 3 3 = 0
1 − 2 = 3
The associated matrix is
1 1 5 0
A = 1 1 3 0
1 −1 0 3 

The RREF of A is
 3
 1 0 0 2
 3
A  = 0 1 0 − 
 2
 0 0 1 0
 
There is no row in the RREF which has zeros in all but the last column and a
nonzero entry in the last column. So, the system is consistent. From the
3 3
RREF, we get the solution 1 = ,  2 = − and  3 = 0.
2 2
***
Here is another example.
Example 6: Check whether the vector b = (1,5,4) is in [S] where
S = (1,2,1),( −1,1,1),(2,1,0).
Solution: Since this is similar to the previous example, we sketch the solution.
The associated matrix is
 1 −1 2 1
A = 2 1 1 5 
 1 1 0 4
 
Carrying out row operations R 2 → R 2 − 2R1, R3 → R3 − R1,
In the last row, the entries in all but the entry in the last column are zero and
the entry in the last column is 1. So, the associated system of equations is not
consistent. Therefore b is not in [S].
***
Here is an exercise for you.

97
Block 2

E 6) In each of the following cases, check whether b is in [S]. If b is in

[S], write b as a linear combination of elements in [S].
a) b = (3,1,0) , S = (1,0,1),(0,1, −1),(0, −1,1).
b) b = (1,1,5), S = (1,1,2),( −1, −1,1),(2,2,1)
c) b = (0,3,1), S = (1,1,1),(2,1, −1),( −1,0,2)

Example 7: Show that, for any m  n matrix A, (A) =  A t . ( )

Solution: Note that RS(A) = CS(A ) . Therefore, we have
t

( ( )) =  ( A ) =  ( A )
( A ) = r (A) = dim (RS(A)) = dim CS A t c
t t

***

Next, we consider the problem of checking whether a given set of vectors are
n
linearly independent or not. If we are working in , we can use the concept
of rank of a matrix.
Example 8: Check whether the following subsets of 3 or 4 (as the case
may be) are linearly independent or not.

a) {u = (1,0,0),v = (0,0, − 5)}

b) {u = ( −1,6, − 12),v = (1/ 2, − 3,6)}
c) {u = (1,2,3,4),v = (4,3,2,1)}

Solution: a) Let au + bv = 0,a,b  .

Then, a(1,0,0) + b(0,0, −5) = (0,0,0)

i.e., (a,0,0) + (0,0, −5b) = (0,0,0)
i.e., (a,0, −5b) = (0,0,0)
i.e., a = 0, − 5b = 0,i.e.,a = 0,b = 0
,{u,v} is linearly independent.

b) Let au + bv = 0,a,b  .

b 
Then ( −a,6a, − 12a) +  , − 3b,6b) = (0,0,0) 
2 
b
i.e., −a + = 0,6a − 3b = 0, −12a + 6b = 0. Each of these equations is
2
equivalent to 2a − b = 0, which is satisfied by many non-zero values of a and
b (e.g., a = 1,b = 2).

Hence, {u,v} is linearly dependent.

c) Suppose au + bv = 0, a, b  . Then

(a + 4b,2a + 3b,3a + 2b,4a + b) = (0,0,0,0)

i.e.,a + 4b = 0 …(1)
2a + 3b = 0 …(2)
3a + 2b = 0 …(3)
4a + b = 0 …(4)
98
Block Miscellaneous
Examples Exercises and
Subtracting (2) from (3) we get a − b,i.e.,a = b. Putting this in (1), we have
5b = 0. ,b = 0, and so, a = b = 0. Hence, {u,v} is linearly independent.

***
You know that the set {1,x,x ,...,x }  P is linearly independent. For larger
2 n

and larger n, this set becomes a larger and larger linearly independent subset
P . This example shows that in the vector space P, we can have as large a
linearly independent set as we wish. In contrast to this situation look at the
following example, in which more than two vectors are not linearly
independent.

2
Example 9: Prove that in any three vectors form a linearly dependent set.

Solution: Let u = (a1,a2 ),v = (b1,b2 ),w = (c1,c 2 )  2

. If any of these is the
zero vector, say u = (0,0), then the linear combination 1.u + 0.v + 0.w, of
u,v, w, is the zero vector, showing that the set {u,v, w} is linearly dependent.
Therefore, we may suppose that u,v, w, are all non-zero.

We wish to prove that there are real numbers, , , , not all zero, such that
u + v + w = 0. That is, u + v = −w. This reduces to the pair of
equations.

a1 +  b1 = −c1
a2 + b2 = −c 2

We can solve this pair of equations to get values of ,  in terms of

a1,a2 ,b1,b2 ,c1,c 2 and  iff a1b2 − a2b1  0. So, if

a1b2 − a2b1  0, we get

(b1c 2 − b2c1 )
=
a1b2 − a2b1
(c1a2 − a2c1 )
=
a1b2 − a2b1

Then, we can give  a non-zero value and get the corresponding values of 
and . Thus, if a1b2 − a2b1  0 we see that {u,v, w} is a linearly dependent set.

Suppose, a1b2 − a2b1 = 0. Then one of a1 and a 2 is non-zero since u  0.

Similarly, one of b1 and b2  0. Let us suppose that a1  0,b1  0. Then,
observe that b1(a1,a2 ) − a1(b1,b2 )

= (b1a1,b1a2 ) − (a1b1,a1b2 )
= (0,0)
i.e., b1u − a1v + 0.w = 0 and a1  0,b1  0.
Hence, in this case, also {u,v, w} is a linearly dependent set.

Try the following exercises now.

99
Block 2

3
E7) Check whether each of the following subsets of is linearly
independent.

a) (1,2,3),(2,3,1),(3,1,2)
b) (1,2,3),(2,3,1),( −3, − 4,1)
c) ( −2,7,0),(4,17,2),(5, − 2,1)
d) ( −2,7,0),(4,17,2)
E8) Prove that in the vector space of all functions from to , the set
{sin x,cos x} is linearly independent, and the set
{sin x,cos x,sin(x +  / 6)} is linearly dependent.
E9) Determine whether each of the following subsets of P is linearly
independent or not.

a) {x 2 ,x 2 + 2}
b) {x 2 + 1,x 2 + 11,,2x 2 − 3}
c) {3,x + 1,x 2 ,x 2 + 2x + 5}
d) {1,x 2 ,x 3 + x 2 + 1}

SOLUTIONS/ANSWERS

E 1)The matrix corresponding to the system of equations is

 1 1 1 3
 −1 1 3 4 
 2 2 1 5
 
R2 → R 2 + R1 gives
1 1 1 3 
0 2 4 7 
0 0 −1 −1
 
1
The operations R2 → R2 and R3 → ( −1)R3 give
2
1 1 1 3 
 7
0 1 2 
0 0 1 2 1 

The corresponding system of equations are
x1 + x 2 + x 3 = 3
7
x 2 + 2x 3 =
2
x3 = 1
3 1
Back substitution gives. x 3 = 1 , x 2 = and x1 = .
2 2
E 2) Looking at the closed path ABCD, we get 5I1 − 5I2 = 0 since I1 and I2 are
in opposite directions. In AEFB, the supply is 30 volts, so we get 5I2 + 5I3 = 30.
100
Block Miscellaneous
Examples Exercises and
Looking at the junction B, we get I1 + I2 − I3 = 0 . To find I1, I2 and I3, we need to
solve the following system of equations:
5I1 − 5I2 = 0
2I2 + I3 = 30
I1 + I2 − I3 = 0
Solving, we get I1 = 2, I2 = 2 and I3 = 4 .
E 3) One new feature in this problem is there are voltage sources in both
the closed paths. Considering AEFB, we get −I2 + 2I3 = 24. Considering the
closed path ABCD, we get 2I1 + I2 = 48. Considering the junction A, we get
I1 − I2 − I3 = 0. So, the equations are
I1 − I2 − I3 = 0
− I2 + 2I3 = 24
2I1 + I2 = 48
Solving, we get I1 = 21, I2 = 6 and I3 = 15.
E 4) The matrix form of the system of equations.
1 1 0 1 0 
1 1 2 1 0 
1 1 1 1 0 
1 1 4 1 0 
 
R2 → R 2 − R1 gives
1 1 0 1 0
0 0 2 0 0 
1 1 1 1 0
1 1 4 1 0
 
R3 → R3 − R1 gives
1 1 0 1 0
0 0 2 0 0 
0 0 1 0 0 
0 0 4 0 0 
 
1
R2 → R2 gives
2
1 1 0 1 0
0 0 1 0 0 
0 0 1 0 0 
0 0 4 0 0 
 
R3 → R3 − R 2 gives
1 1 0 1 0
0 0 1 0 0 
0 0 0 0 0 
0 0 4 0 0 
 
1
R4 → R4 gives
4
1 1 0 1 0
0 0 1 0 0 
0 0 0 0 0 
0 0 1 0 0 
 
R 4 → R 4 − R1 gives
101
Block 2

1 1 0 1 0
0 0 1 0 0 
0 0 0 0 0 
0 0 0 0 0 
 
x 2 and x 4 are the columns without pivot elements. The solution is
x1 = − x 2 − x 4
x 3 = 0.
Writing 1 = x 2 ,  2 = x 4 . We have x1 = 1 +  2 ,x 3 = 0 . So, the solution set
is (1 +  2 ,0, 1,  2 ) | 1,  2  .
E 5) We have the equation
x1C6H12O6 → x 2CO2 + x 3C2H5OH
Comparing the number of carbon, hydrogen and oxygen atoms in
the reactants and the product, we get

Reactants Products
Carbon 6x1 x2+2x3
Hydrogen 12x1 6x3
Oxygen 6x1 2x2+x3
So, the system of equations is
6x1 − x2 − x3 = 0
12x1 − 5x 3 = 0
6x1 − 2x 2 − x3 = 0
The associated matrix is
 6 −1 −2
A = 12 0 −5 
 6 −2 −1
 
The RREF of A is
 1
 1 0 − 2
A  = 0 1 −1
0 0 0 
 
 
The non-pivot column corresponds to x3.Taking x 3 = , the

solution set is x1 = , x 2 = , and x 3 = , i.e.
2
 1  
  ,11 ,    .
 2  
Since we need a solution in natural numbers, we take  = 2 and
get the solution x1 = 1, x 2 = 2, and x 3 = 2. The balanced equation
is
C5H12O6 → 2CO2 + 2C2H5OH
E 6) a) The associated matrix is
 1 1 0 3
0 1 −1 1
 1 0 1 0
 
The row operations R3 → R3 − R1, R3 → R3 + R2 gives the
102 matrix
Block Miscellaneous
Examples Exercises and
 1 1 0 3
 0 1 −1 1
 0 0 0 −2 
 
In the last row, the entries in all the columns, except the
last column, are zero and the last column is −2.
Therefore, the system system of equations is inconsistent
and b is not in [S] .
b) The associated matrix is
 1 −1 2 1
 1 −1 2 1
2 1 1 5 
 
The RREF is
 1 0 1 2
 0 1 −1 1
 0 0 0 0
 
The equations are consistent. The third column is
the non-pivot column. So, we take the third
variable as the free variable. We get 1 = 2 − ,
 2 = 1 +  and  3 = . Taking  = 0, we get 1 = 1
 2 = 1 and  3 = 0. So,
2(1,1,2) + ( −1, −1,1) + 0(2,2,1) = (1,1,5)
c) The associated matrix is
 1 2 −1 0 
 1 1 0 3
 1 −1 2 1
 
The row operations R2 → R2 − R1, R3 → R3 − R1
gives the matrix
 1 2 −1 0 
 0 1 −1 −3 
 0 0 0 8
 
In the last row, entries in all the columns, except
the one in last column, is zero and the last column
is 8. So, the associated system of equations is
inconsistent. Therefore b is not in [S].

E 7) a) a(1,2,3) + b(2,3,1) + c(3,1,2) = (0,0,0)

 (a,2a,3a) + (2b,3b,b) + (3c,c,2c) = (0,0,0)
 (a + 2b + 3c,2a + 3b + c,3a + b + 2c) = (0,0,0)
 a + 2b + 3c = 0 ...........(1)
2a + 3b − 4c = 0 ...........(2)
3a + b + 2c = 0 ...........(3)

Then (1) + (2) − (3) gives 4b + 2c = 0, i.e., c = −2b. Putting this

value in (1) we get a + 2b − 6b = 0, i.e., a = 4b. Then (2) gives
8b + 3b − 2b = 0, i.e. b = 0. Therefore, a = b = c = 0 . Therfore, the
given set is linearly independent.

b) a(1,2,3) + b(2,3,1) + c( −3, −4,1) = (0,0,0)

 (a + 2b − 3c,2a + 3b − 4c,3a + b + c) = (0,0,0)
103
Block 2

 a + 2b − 3c = 0
2a + 3b − 4c = 0
3a + b + c = 0

On simultaneously solving these equations you will find that a,b,c

can have many non-zero values, one of them being
a = −1,b = 2,c = 1., the given set is linearly dependent.

c) Linearly dependent

e) Linearly independent.
E 8) To show that {sin x,cos x} is linearly independent, suppose a,b  such
that a sin x + b cos x = 0.

Putting x = 0 in this equation, we get b = 0. Now, take x =  / 2.

We get a = 0. Therefore, the set is linearly independent.
Now, consider the equation

a sin x + bcos x + c sin(x +  / 6) = 0.

Since sin(x +  / 6) = sin x cos  / 6 + cos x sin  / 6

= 3 / 2sin x + 1/ 2cos x, taking a = − 3 / 2,b = 1/ 2,c = 1, we get a
linear combination of the set {sin x,cos x,sin(x +  / 6)} which shows
that this set is linearly independent.
E 9) a) ax 2 + b(x 2 + 1) = 0  (a + b)x 2 + b = 0  a + b = 0,b = 0
 a = 0,b = 0.  , the given set is linearly independent.
b) Linearly dependent because, for example,
−5(x 2 + 1) + (x 2 + 11) + 2(2x 2 − 3) = 0 .
c) Linearly dependent.
d) Linearly dependent.

104