MA1140 : Linear Algebra

Lecture 01

Dr. Mrinmoy Datta

Department of Mathematics
IIT Hyderabad
[email protected]/

January 2024

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Basic Information
Aim of the course
This core course aims at an introduction to the fundamental concepts of
linear algebra, including systems of linear equations, matrices, linear
transformations, vector spaces, eigenvalues, and eigenvectors.
Course contents and References
1 Gilbert Strang, Linear Algebra and its Applications, 4th Edition,
Cencage India, 2005.
2 S. Kumaresan, Linear Algebra - A Geometric Approach, Prentice
Hall of India.

Lectures and Office hours

Lectures: On E Slot.
Office Hours: Monday 12 noon - 2 pm, B 315. Appointments by email.
Final Exam
6th February.
Acknowledgement
I shall make use of the lecture notes prepared by Prof. B. V, Limaye and
Prof. Sudhir Ghorpade of a similar course in the past.
Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01
Basic Questions and Answers

Q. 1 What is Linear Algebra?

Ans. You will find out in this course!
Q. 2 Why should I study it?
Ans. Because it is beautiful and useful! In greater detail,
Linear Algebra is one of the basic areas in Mathematics, having at
least as great an impact as Calculus.
Provides a vital arena where the interaction of Mathematics and
machine computation is seen.
Many problems studied in Linear Algebra are amenable to systematic
and even algorithmic solutions, making them implementable on
computers.
Many geometric topics are studied using Linear Algebra.
Numerous Applications within and outside Mathematics. For
example, the Google page rank algorithm is based on notions and
results from Linear Algebra.

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Notation

N := {1, 2, 3, . . .}
Z := {. . . , −3, −2, −1, 0, 1, 2, 3, . . .}
R := the set of all real numbers
For n ∈ N, let us consider the Euclidean space

Rn := {(x1 , . . . , xn ) : xj ∈ R for j = 1, . . . , n}.

Elements of Rn will be called vectors. We let 0 := (0, . . . , 0). Also, for

x := (x1 , . . . , xn ) and y := (y1 , . . . , yn ) in Rn , and for α ∈ R, we define

(sum) x + y := (x1 + y1 , . . . , xn + yn ) ∈ Rn ,
(scalar multiple) α x := (α x1 , . . . , α xn ) ∈ Rn ,
(scalar product) x · y := x1 y1 + · · · + xn yn ∈ R.

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Matrices

Let m, n ∈ N. An m × n matrix A with real entries is a rectangular

array of real numbers arranged in m rows and n columns, written as
follows:

a11 ··· a1k ··· a1n

 
 .. .. .. 
 . . . 
··· ···
 
 aj1
A :=  ajk ajn  = [ajk ],
 . .. .. 
 .. . . 
am1 ··· amk ··· amn
where ajk ∈ R is called the (j, k)th entry of A for j = 1, . . . , m and
k = 1, . . . , n.
Let Rm×n denote the set of all m × n matrices with real entries. If
A := [ajk ] and B := [bjk ] are in Rm×n , then we say A = B ⇐⇒ ajk = bjk
for all j = 1, . . . , m and k = 1, . . . , n.

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Let 0 ≤ r < m, 0 ≤ s < n. By deleting r rows and s columns from A, we
obtain an (m − r )×(n − s) submatrix of A.

An n × n matrix, that is, an element of Rn×n , is called a square matrix

of size n.
A square matrix A = [ajk ] is called
symmetric if ajk = akj for all j, k.
skew-symmetric if ajk = −akj for all j, k.
a diagonal matrix if ajk = 0 for all j ̸= k.
A diagonal matrix A = [ajk ] is called a scalar matrix if all diagonal
entries of A are equal.
Two important scalar matrices are the identity matrix I in which all
diagonal elements are equal to 1, and the zero matrix O in which all
diagonal elements are equal to 0.

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

A square matrix A = [ajk ] is called upper triangular if ajk = 0 for all
j > k, and lower triangular if ajk = 0 for all j < k.
Note: A matrix A is upper triangular as well as lower triangular if and
only if A is a diagonal matrix.
Examples
 
1 2 3
The matrix 2 4 5 is symmetric, while the matrix
3 5 6
 
0 2 3
−2 0 5 is skew-symmetric.
−3 −5 0

Note: Every diagonal entry of a skew-symmetric matrix is 0 since

ajj = −ajj =⇒ ajj = 0 for j = 1, . . . , n.

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Examples
   
2 0 0 2 0 0
The matrix 0 3 0 is diagonal, while 0 2 0 is a scalar matrix.
0 0 4 0 0 2
 
2 1 −1
The matrix 0 3 1  is upper triangular,
0 0 4
 
2 0 0
while the matrix  1 3 0 is lower triangular.
−1 1 4

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

A row vector a of length n is a matrix with only one row consisting of n
real numbers; it is written as follows:
 
a = a1 · · · ak · · · an ,

where ak ∈ R for k = 1, . . . , n. Here a ∈ R1×n .

A column vector b of length n is a matrix with only one  column

b1
 .. 
.
 
bk  , where
consisting of n real numbers; it is written as follows: b =  
 .. 
.
bn
bk ∈ R for k = 1, . . . , n. Here b ∈ Rn×1 .
When n = 1, we may identify [α] ∈ R1×1 with α ∈ R.

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Operations on Matrices

Let m, n ∈ N, and let A := [ajk ] and B := [bjk ] be m × n matrices. Then

the m × n matrix
A + B := [ajk + bjk ]
is called the sum of A and B. Also, if α ∈ R, then the m × n matrix

αA := [αajk ]

is called the scalar multiple of A by α.

These operations follow the usual algebra rules:

A + B = B + A,
(A + B) + C = A + (B + C), which we write as A + B + C,
α(A + B) = αA + αB,
(α + β)A = αA + βA and
α(βA) = (αβ)A, which we write as αβA.
Also, we write (−1)A as −A, and define A − B := A + (−B).

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

The transpose of an m × n matrix A := [ajk ] is the n × m matrix

AT := [akj ]

(in which the rows and the columns of A are interchanged).

Clearly,
(AT )T = A,
(A + B)T = AT + BT and
(αA)T = αAT .
Note: A square matrix A is symmetric ⇐⇒ AT = A.
   
2 1 −1 1 0 2
Examples Let A := and B := .
0 3 1 −1 4 1
   
3 1 1 10 5 −5
Then A + B := and 5A = . Also
−1 7 2 0 15 5
 
2 0
AT =  1 3.
−1 1

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

In particular, the preceding operations can be performed on row vectors,
and also on column vectors since they are particular types of matrices.

Notice that the sum a1 + a2 of two row vectors a1 and a2 follows the
parallelogram law, and so does the sum b1 + b2 of two column vectors b1
and b2 . (Note: All vectors ‘originate’ from the zero vector.)

Also, note that the transpose of a row vector is a column vector, and vice
versa.  
b1
 .. 
.
  n×1
We shall often write a column vector b := bk  ∈ R , as

 .. 
.
bn
 T
b1 · · · bk · · · bn in order to save space.

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Let m, n ∈ N. Let α1 , . . . , αm ∈ R.
If a1 , . . . , am ∈ R1×n , then

α1 a1 + · · · + αm am ∈ R1×n

is called a (finite) linear combination of a1 , . . . , am .

Similarly, if b1 , . . . , bm ∈ Rn×1 , then

α1 b1 + · · · + αm bm ∈ Rn×1

is a (finite) linear combination of b1 , . . . , bm .

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

In particular, for k = 1, . . . , n, consider the column vector
 T
0 ··· 0 1 0 ··· 0
ek := |{z} ∈ Rn×1 ,
k-th entry

where the k-th entry is 1 and all other entries are 0.

 T
If b = b1 · · · bk · · · bn is any column vector of length n, then
it follows that b = b1 e1 + · · · + bk ek + · · · + bn en , which is a linear
combination of e1 , . . . , en . The vectors e1 , . . . , en are known as the basic
column vectors in Rn×1 .

Let A := [ajk ] ∈ Rm×n .

 
Then aj := aj1 · · · ajn ∈ R1×n is called the jth row vector of A for
 
a1
 .. 
j = 1, . . . , m, and we write A =  . .
am
 T
Also, ck := a1k · · · amk iscalled the kth  column vector of A for
k = 1, . . . , n, and we write A = c1 · · · cn .

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Matrix Multiplication

Before we discuss when and how a product AB of A by B may be

defined, we define the product of a row vector a by a column vector b of
the same size. Since every matrix can be written in terms of row vectors
and column vectors, we shall then consider the product AB.
 
b1
Let n ∈ N, a := a1 · · · an ∈ R1×n and b :=  ...  ∈ Rn×1 . Define
   

bn
the product of a row vector a with a column vector b as follows:
 
b1
  . 
ab = a1 · · · an  ..  := a1 b1 + · · · + an bn ∈ R.
bn

(This is like the scalar product of two vectors in Rn .)

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

  
a1
Let m ∈ N and A ∈ Rm×n . Then A =  ... , where a1 , . . . , am ∈ R1×n .
 

am
Recalling that b ∈ Rn×1 , we define
   
a1 a1 b
Ab =  ... b :=  ...  ∈ Rm×1 .
   

am am b

Finally, let p ∈ N and B ∈ Rn×p . Then B = [b1 · · · bp ], where

b1 , . . . , bp ∈ Rn×1 . Noting that Ab1 , . . . , Abp belong to Rm×1, we define

AB = A[b1 · · · bp ] := [Ab1 · · · Abp ] ∈ Rm×p .

Thus
   
a1 a1 b1 ··· a1 bp
AB =  ...  [b1 · · · bp ] =  ... ..  ∈ Rm×p .
  
. 
am am b1 ··· am bp

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

So if m, n, p ∈ N, A := [ajk ] ∈ Rm×n and B := [bjk ] ∈ Rn×p , then
AB ∈ Rm×p , and for j = 1, . . . , m; k = 1 . . . , p,
n
X
AB = [cjk ], where cjk := aj bk = ajℓ bℓk .
ℓ=1

Note that the (j, k)th entry of AB is a product of the jth row vector of
A with the kth column vector of B as shown below:
 
  b1k
 .. 



 . 

aj1 · · · ajℓ · · · ajn  
  bℓk 
 
  .. 
 . 
bnk
Clearly, the product AB is defined only when the number of columns of
A is equal to the number of rows of B.
Note: AI = A, IA = A, AO = O and OA = O whenever these products
are defined.

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Examples  
  1 6 0 2
2 1 −1
(i) Let A := and B := 2 −1 1 −2 .
0 3 1 2×3 2 0 −1 1 3×4
 
2 11 2 1
Then AB = .
8 −3 2 −5 2×4
(ii) Both products AB and BA are defined ⇐⇒ the number of columns
of A is equal to the number rows of B and the number columns of B is
equal to the number of rows of A, that is, when A ∈ Rm×n and
B ∈ Rn×m .    
1 0 0 1
In general, AB ̸= BA. For example, if A := and B := ,
0 0 0 0
   
0 1 0 0
then AB := , while BA := .
0 0 0 0
Note that BA = O, while AB = B ̸= O. Since A ̸= I,
we see that the so-called cancellation law does not hold.

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Important Remark
Let A := [ajk ] ∈ Rm×n , and let e1 , . . . , en be the basic column vectors in
Rn×1 . Then for k = 1, . . . , n,
 
a1k
 .. 
 . 
 
 ajk , which is the kth column of A.
A ek =  
 .. 
 . 
amk
This follows from our definition of matrix multiplication.
It follows that if B ∈ Rm×n , then
A = B if and only if Aek = Bek for each k = 1, . . . , n.
In particular, A = I ⇐⇒ Aek = ek for each k = 1, . . . , n.

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

  
c1
 .. 
 . 
  m×1
 cj  ∈ R
Let c :=  be a column vector of length m, and let

 .. 
 . 
cm
 
d := d1 · · · dk · · · dp ∈ R1×p be a row vector of length p. Then
 
c1 d1 ··· c1 dk ··· c1 dp
 .. .. .. .. .. 
 . . . . . 
m×p
 
 c j d1
cd =  ··· cj dk ··· c j dp  = [cj dk ] ∈ R .
 .. .. .. .. .. 
 . . . . . 
cm d1 ··· c m dk ··· cm dp

This is called the outer product of the column vector c and the row
vector d.

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Let A be an m × n matrix and B be an n × p matrix. We have defined
the matrix product AB in terms of the m row vectors of A and the p
column vectors of B. Now let us write A in terms of its n column vectors
 d1 , . . . , dn . Then
c1 , . . . , cn , and we write B in terms of its n rowvectors
a1ℓ
   .. 
d1  . 
A = c1 · · · cn and B =  ... , where cℓ = 
   
 ajℓ ,
  
 .. 
dn  . 
amℓ
 
dℓ = bℓ1 · · · bℓk · · · bℓp for ℓ = 1, . . . , n.
Hence
n n n
X X   hX i
cℓ dℓ = ajℓ bℓk = ajℓ bℓk = AB.
ℓ=1 ℓ=1 ℓ=1

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Thus
 
d1
 . 
cn  ..  = c1 d1 + · · · + cn dn ∈ Rm×p .

AB = c1 ···
dn
 
d1
 .. 
In particular, if p = 1, that is, if B :=  .  ∈ Rn×1 is a column vector,
dn
then
 
d1
 . 
cn  ..  = d1 c1 + · · · + dn cn ∈ Rm×1 .

AB = c1 ···
dn

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Properties of Matrix Multiplication

Consider matrices A, B, C and α ∈ R. Then

(A + B)C = AC + BC,
C(A + B) = CA + CB and
(αA)B = αAB = A(αB), if sums & products are well-defined.

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Matrix multiplication also satisfies the associative law:
Proposition
Let m, n, p, q ∈ N. If A ∈ Rm×n , B ∈ Rn×p and C ∈ Rp×q , then
A(BC) = (AB)C (which we shall write as ABC).

Proof. Let A := [ajk ], B := [bjk ] and C := [cjk ]. Also, let (AB)C := [αjk ]
and A(BC) := [βjk ]. Then
p X
X n  n
X p
X 
αjk = ajℓ bℓi cik = ajℓ bℓi cik = βjk
i=1 ℓ=1 ℓ=1 i=1

for j = 1, . . . , m and k = 1, . . . , q. Hence the result.

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Also, the transpose of a product is the product of the transposes in the
reverse order:
Proposition
Let m, n, p ∈ N. If A ∈ Rm×n and B ∈ Rn×p , then (AB)T = BT AT .

Proof. Let A := [ajk ], B := [bjk ] and AB := [cjk ].

′ ′
Also, let AT := [ajk ], BT := [bjk ] and (AB)T := [cjk′ ]. Then
n
X n
X
cjk = ajℓ bℓk and so cjk′ = ckj = akℓ bℓj
ℓ=1 ℓ=1

for j = 1, . . . , m; k = 1, . . . , p. Suppose BT AT := [djk ]. Then

n
X n
X
′ ′
djk = bjℓ aℓk = bℓj akℓ = cjk′
ℓ=1 ℓ=1

for j = 1, . . . , m; k = 1, . . . , p. Hence the result.

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Matrix Multiplication Revisited

Let m, n, p ∈ N, A ∈ Rm×n and B ∈ Rn×p .

Rows of AB
The j-th row of AB is a linear combination of the n row vectors of B
with coefficients aj1 , . . . , ajn provided by the jth row of A.

Can we come up with a similar characterization of columns of AB in

terms of the n column vectors of A?
Two things we should never forget:
 
a1 B
AB =  ...  = Ab1 · · · Abp
   

am B

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Example
As we have seen,

 
  1 6 0 2  
2 1 −1  2 11 2 1
2 −1 1 −2 =
 , where
0 3 1 8 −3 2 −5
2 0 −1 1

     
2 11 2 1 = 2 1 6 0 2 + 1 2 −1 1 −2
 
−1 2 0 −1 1 ,
     
8 −3 2 −5 = 0 1 6 0 2 + 3 2 −1 1 −2
 
+1 2 0 −1 1 ,
       
2 2 1 −1
= 1 +2 +2 ,
8 0 3 1
       
11 2 1 −1
= 6 −1 +0 , etc.
−3 0 3 1
.

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Linear System

Let m, n ∈ N. A linear system of m equations in the n unknowns

x1 , ..., xn is given by

a11 x1 + a12 x2 + · · · + a1n xn = b1 (1)

a21 x1 + a22 x2 + · · · + a2n xn = b2 (2)
.. .. .. .. ..
. . . . .
am1 x1 + am2 x2 + · · · + amn xn = bm , (m)

where ajk ∈ R for j = 1, . . . , m; k = 1, . . . , n and also bj ∈ R for

j = 1, . . . , m are given.
 T
Let A := [ajk ] ∈ Rm×n , x := x1 · · · xn ∈ Rn×1 and
 T
b := b1 · · · bm ∈ Rm×1 . Using matrix multiplication, we write the
linear system as
Ax = b.

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01

Food for Thought

We will continue the discussion of linear systems of equations in the next

lecture. Before that it may be a good idea for you to go over the slides of
this lecture so as to familiarize yourself with the notation and
terminology that we will be using in the course. Further, you may think
of the following
Question. Is it possible to write down linear systems that have: (i) no
solution,
(ii) exactly one solution,
(iii) exactly two solutions,
(iv) exactly 100 solutions,
(v) infinitely many solutions?
Note: You may think geometrically of solutions of linear systems of m
equations in n unknowns as intersections of m hyperplanes in the
n-dimensional euclidean space Rn .

Dr. Mrinmoy Datta , IIT Hyderabad Linear Algebra: Lecture 01