MA 110:

Lecture 06

Saurav Bhaumik
Department of Mathematics
IIT Bombay

Spring 2025

Saurav Bhaumik, IIT Bombay MA 110: Lecture 06

Let S be a set of vectors (in R1×n or in Rn×1 ). Recall:
S is linearly dependent if a nontrivial linear combination
of finitely many vectors in S is zero, i.e., there are
(distinct) vectors a1 , . . . , am in S and scalars α1 , . . . , αm ,
not all zero, such that α1 a1 + · · · + αm am = 0.
S is linearly independent if it is not linearly dependent.
Crucial Result: If S has s vectors, and each of them is a
linear combination of a (fixed) set of r vectors, with
s > r , then S is linearly dependent.
In particular, if S has more than n vectors, then S is
linearly dependent.
Let A ∈ Rm×n . The row rank of A is the maximum
number of linearly independent row vectors of A.
Elementary row operations do not alter the row rank.
row-rank(A) = number of nonzero rows in a REF of A
= the number of pivots in a REF of A.
Saurav Bhaumik, IIT Bombay MA 110: Lecture 06
Example
In a previous lecture, we had seen that the matrix
 
3 2 2 −5
A :=  0.6 1.5 1.5 −5.4 
1.2 −0.3 −0.3 2.4

can be transformed to a row echelon form

 
3 2 2 −5
′
A :=  0 1.1 1.1 −4.4 
0 0 0 0

by elementary row transformations of type I and type II. Since

the number of nonzero rows of A′ is 2, we see that the row
rank of A is 2. This shows that the 3 row vectors of A are
linearly dependent.
Saurav Bhaumik, IIT Bombay MA 110: Lecture 06
Column Rank of a Matrix
Definition
Let A ∈ Rm×n . The column rank of A is the maximum
number of linearly independent column vectors of A.
Clearly, column-rank(A) = row-rank(AT ).
Proposition
The column rank of a matrix is equal to its row rank.
Proof. Let A := [ajk ] ∈ Rm×n . Let r = row-rank(A) and
s :=column-rank(A). We will show that s ≤ r and r ≤ s.
Let a1 , . . . , am denote the row vectors of A, so that
 
aj := aj1 · · · ajk · · · ajn for j = 1, . . . , m.
Among these m row vectors of A, there are r row vectors
aj1 , . . . , ajr which are linearly independent, and each row vector
of A is a linear combination of these r row vectors.
Saurav Bhaumik, IIT Bombay MA 110: Lecture 06
For simplicity, let b1 := aj1 , . . . , br := ajr . Then for
j = 1, . . . , m and ℓ = 1, . . . , r , there is αjℓ ∈ R such that

aj = αj1 b1 + · · · + αjℓ bℓ + · · · + αjr br for j = 1, . . . , m.

 
If bℓ := bℓ1 · · · bℓk · · · bℓn for ℓ = 1, . . . , r , then the
above equations can be written componentwise as follows:

a1k = α11 b1k + ··· + α1ℓ bℓk + ··· + α1r brk

.. .. .. .. ..
. . . . .
ajk = αj1 b1k + ··· + αjℓ bℓk + ··· + αjr brk
.. .. .. .. ..
. . . . .
amk = αm1 b1k + · · · + αmℓ bℓk + · · · + αmr brk

for k = 1, . . . , n. These m equations show that the kth

column of A can be written as follows:

Saurav Bhaumik, IIT Bombay MA 110: Lecture 06

        
a1k α11 α12 α1r
 ..   ..   ..   .. 
 .   .   .   . 
 jk  = b1k  j1  + b2k  j2  + · · · + brk  αjr  .
a α α
       
 .   .   .   . 
 ..   ..   ..   .. 
amk αm1 αm2 αmr

for k = 1, . . . , n. Thus each of the n columns of A is a linear

 T
combination of the r column vectors α11 · · · αj1 · · · αm1 ,
 T  T
α12 · · · αj2 · · · αm2 , . . . , α1r · · · αjr · · · αmr .
By a crucial result we have proved earlier, no more than r
columns of A can be linearly independent. Hence s ≤ r , i.e.,
column-rank(A) ≤ row-rank(A).
Applying the above result to AT in place of A, we obtain
r ≤ s. Thus s = r , as desired.
Saurav Bhaumik, IIT Bombay MA 110: Lecture 06
In view of the above result, we define the rank of a matrix A
to be the common value of the row rank of A and the column
rank of A, and we denote it by rank A . Consequently,
rank AT = rank A .
Example  
1 2 3 4 5 6
0 0 0 7 8 9 
Let A′ :=   0 0 0 0 0 10 .


0 0 0 0 0 0
Since A′ is in REF, its row rank is equal to the nonzero rows,
namely 3. Also, we can easily check that the columns c1 , c4
and c6 are linearly independent, and so the column rank of A′
is at least 3. Further, each of the columns c2 = 2c1 , c3 = 3c1
and c5 = (3/7)c1 + (8/7)c4 is linear combination of the
columns c1 , c4 and c6 , and so by a crucial result, the column
rank of A′ is at most 3. Hence the column rank of A′ is also 3.
Saurav Bhaumik, IIT Bombay MA 110: Lecture 06
Let us restate some results proved earlier in terms of the newly
introduced notion of the rank of a matrix.
Let A ∈ Rm×n .
1. If A is transformed to A′ by EROs, then rank A′ = rank A.
2. If A is transformed to A′ by EROs and A′ is in REF, then

rank A = number of nonzero rows of A′

= number of pivotal columns of A′ .

(Note: Each nonzero row has exactly one pivot, and a zero
row has no pivot.)
3. Let m = n. Then

A is invertible ⇐⇒ rank A = n.

(Recall: A square matrix is invertible if and only if it can be

transformed to the identity matrix by EROs.)
Saurav Bhaumik, IIT Bombay MA 110: Lecture 06
Subspace, Basis, Dimension
From now on we shall consider only the column vectors .
Let n ∈ N. We have denoted the set of all column vectors
whose entries are real numbers and whose length is n by Rn×1 .
Definition
A nonempty subset V of Rn×1 is called a vector subspace, or
simply a subspace of Rn×1 if
(i) a, b ∈ V =⇒ a + b ∈ V , and
(ii) α ∈ R, a ∈ V =⇒ α a ∈ V .

Note: The zero column vector 0 is in every subspace V since

V ̸= ∅ and 0 = a + (−1)a for each a ∈ V .
Each linear combination of elements of a subspace V is in V .
{0} is the smallest and Rn×1 is the largest subspace of Rn×1 .
Saurav Bhaumik, IIT Bombay MA 110: Lecture 06
Two Important Examples of Subspaces Related to a Matrix
Let A ∈ Rm×n .

Definition
The null space of A is

N (A) := {x ∈ Rn×1 : Ax = 0}.

Definition
The column space of A is

C(A) := the set of all linear combinations of columns of A.

We shall later find an interesting relationship between the null

space N (A) and the column space C(A).

Saurav Bhaumik, IIT Bombay MA 110: Lecture 06

Basis and dimension of a subspace

Let V be a subspace of Rn×1 . Since every element of Rn×1 is

a linear combination of the vectors e1 , . . . , en , we see that no
linearly independent subset of V contains more than n vectors.
Definition
A subset S of V is called a basis of V if S is linearly
independent and S has maximum possible number of elements
among linearly independent subsets of V .

Clearly, a basis of V has at most n elements.

Definition
Let S ⊂ Rn×1 . The set of all linear combinations of elements
of S is denoted by span S and called the span of S.

Saurav Bhaumik, IIT Bombay MA 110: Lecture 06

Proposition
Let V be a subspace of Rn×1 , and let S ⊂ V . Then S is a
basis for V ⇐⇒ S is linearly independent and span S = V .

Proof. Suppose S is a basis for V . By the definition of a

basis, S is linearly independent. Let S have r elements. If
x ∈ V , but x is not a linear combination of elements of S,
then S∪{x} would be a linearly independent subset of V
containing r + 1 elements, which would be a contradiction.
Conversely, suppose S is linearly independent and span S = V .
Let S have r elements. Since every element of V is a linear
combination of elements of S, no more than r elements of V
can be linearly independent. Hence S is a basis for V .

Saurav Bhaumik, IIT Bombay MA 110: Lecture 06

Corollary
Let V be a subspace of Rn×1 and let S, S ′ be two bases of V .
Then S and S ′ have the same cardinality.

Proof. It is enough to prove that |S| ≤ |S ′ |, because we can

reverse the roles of S and S ′ and prove the other inequality.
By the last proposition, every element of S is a linear
combination of vectors from S ′ . If |S| > |S ′ |, then S would be
linearly dependent. As S is a basis, it is linearly independent,
which implies |S| ≤ |S ′ |.
Definition
The dimension of V is defined as the number of elements in a
basis of V . It is denoted by dim V .

Example: Clearly, dim Rn×1 = n and dim{0} = 0.

Saurav Bhaumik, IIT Bombay MA 110: Lecture 06

Corollary
Let V be a subspace of Rn×1 . Every linearly independent
subset of V can be enlarged to a basis for V .
Proof. Let S be a linearly independent subset of V . If
span S = V , then by the previous result, S is a basis for V .
Suppose span S ̸= V . Then there is x1 ∈ V such that x1 is not
a linear combination of elements of S. Now S1 := S ∪ {x1 } is
a linearly independent subset of V . If span S1 = V , then as
before, S1 is a basis for V . If span S1 ̸= V , then there is
x2 ∈ V such that x2 is not a linear combination of elements of
S1 . Now S2 := S1 ∪{x2 } is a linearly independent subset of V .
This process must end after a finite number of steps since the
number of elements of any linearly independent subset of V is
less than or equal to dim V , and dim V ≤ n.
Remark: All things we have defined above for column vectors
can also be defined for row vectors.
Saurav Bhaumik, IIT Bombay MA 110: Lecture 06
Another immediate consequence of the earlier result is the
unique representation of every element of a subspace in terms
of the elements belonging to a basis of that subspace.
Proposition
Let S := {c1 , . . . , cr } be a basis for a subspace V of Rn×1 ,
and let x ∈ V . Then there are unique α1 , . . . , αr ∈ R such
that x = α1 c1 + · · · + αr cr .
Proof. Since V is a basis for V , we obtain V = span S, and
so the vector x is a linear combination of vectors in S, that is,
there are scalars α1 , . . . , αr such that x = α1 c1 + · · · + αr cr .
Now suppose x = β1 c1 + · · · + βr cr for some β1 , . . . , βr ∈ R.
Then
(α1 − β1 )c1 + · · · + (αr − βr )cr = 0.
Since the set S is linearly independent,
α1 − β1 = · · · = αr − βr = 0, that is, β1 = α1 , . . . , βr = αr .
This proves the uniqueness.
Saurav Bhaumik, IIT Bombay MA 110: Lecture 06
Let us recall some definitions and results from the last lecture.
Let A ∈ Rm×n . The row rank of A is the maximum number of
linearly independent row vectors of A. The column rank of A
is the maximum number of linearly independent column vectors
of A. The two are equal, and we denote both by rank A .
Further, if A′ is any REF of A, then the number of nonzero
rows of A′ is equal to rank A′ and it is equal to rank A.
The column space of A is denoted by C(A) and the null space
of A by N (A).
Proposition
Let A ∈ Rm×n , and let rank A = r . Then dim C(A) = r and
dim N (A) = n − r .

Proof. Since the column rank of A is equal to r , there are r

linearly independent columns ck1 , . . . , ckr of A, and any other
column of A is a linear combination of these r columns.
Saurav Bhaumik, IIT Bombay MA 110: Lecture 06
Let x ∈ C(A). Then x is a linear combination of columns of
A, each of which in turn is a linear combination of ck1 , . . . , ckr .
Thus x is a linear combination of ck1 , . . . , ckr . This shows that
span{ck1 , . . . , ckr } = C(A). Hence {ck1 , . . . , ckr } is a basis for
C(A) and dim C(A) = r .
To find the dimension of N (A), let us transform A to a REF
A′ by EROs of type I and type II. Since the row rank of A is
equal to r , the matrix A′ has exactly r nonzero rows and
exactly r pivotal columns.
Let the n − r nonpivotal columns be denoted by cℓ1 , . . . , cℓn−r .
Then xℓ1 , . . . , xℓn−r are the free variables. For each
ℓ ∈ {ℓ1 , . . . , ℓn−r }, there is a basic solution sℓ of the
homogeneous equation Ax = 0, and every solution of this
homogeneous equation is a linear combination of these n − r
basic solutions. Let S denote the set of these n − r basic
solutions. Then span S = N (A).
Saurav Bhaumik, IIT Bombay MA 110: Lecture 06
We claim that the set S of the n − r basic solutions is linearly
independent. To see this, we note that each basic solution is
equal to 1 in one of the free variables and it is equal to 0 in
the other free variables. Let α1 , . . . , αn−r ∈ R be such that
x := α1 sℓ1 + · · · + αn−r sℓn−r = 0.
For j = 1, . . . , n, let xj denote the jth entry of x. Then for
each ℓ ∈ {ℓ1 , . . . , ℓn−r }, we see that αℓ · 1 = xℓ = 0. Hence S
is linearly independent. Thus S is a basis for N (A) and
dim N (A) = n − r , the number of elements in S.

The dimension of the null space N (A) of A is called the

nullity of A. Since r + (n − r ) = n, we have proved the
following Rank-Nullity Theorem.

Theorem
Let A ∈ Rm×n . Then rank A + nullity A = n.
Saurav Bhaumik, IIT Bombay MA 110: Lecture 06