Chapter 2

Matrices
Chapter 2 Matrices

Section 2.1
Introduction to Matrices
Matrices (Definition 2.1.1)

A matrix (plural matrices) is a rectangular array of

numbers.
The numbers in the array are called entries in the matrix.
The size of a matrix is given by m × n where m is the
number of rows and n is the number of columns.
The (i, j )-entry of a matrix is the number which is in the i th
row and j th column of a matrix.
Examples (Example 2.1.2)

(1, 2)-entry
1 2 4 is a 1 × 1 matrix.
3 4 is a 3 × 2 matrix.
A 1 × 1 matrix is usually
0 −1 treated as a number in
(3, 1)-entry computation.

2 1 0 is a 1 × 3 matrix. 1
1 is a 3 × 1 matrix.
2 3.1 −2 2
1
3 2
0 is a 3 × 3 matrix.
0 π 0
 (Definition 2.1.3
Column and row matrices & Example 2.1.4)

A column matrix (or a column vector) is a matrix with only

one column.
A row matrix (or a row vector) is a matrix with only one
row.

1
1 is a column matrix.
2

2 1 0 is a row matrix.

4 is both a column and row matrix.

Notation of matrices (Notation 2.1.5)

In general, an m × n matrix can be written as

a11 a12 ··· a1n
a21 a22 ··· a2n
⁞ ⁞ ⁞
am1 am2 ··· amn

or simply A = (ai j )m × n where ai j is the (i, j )-entry of A.

If the size of the matrix is already known, we may just
write A = (ai j ).
Examples (Example 2.1.6)

Let A = (ai j )2 × 3 where ai j = i + j.

2 3 4
Then A = .
3 4 5
1 if i + j is even
Let B = (bi j )3 × 2 where bi j =
−1 if i + j is odd.
1 −1
Then B = −1 1 .
1 −1
Square matrices (Definition 2.1.7.1 & Example 2.1.8.1)

A matrix is called a square matrix if it has the same

number of rows and columns.
In particular, an n × n matrix is called a square matrix of
order n.
The following are some examples of square matrices:
2 1 6 2
1 2 3
0 4 0 3 9 −1
4 −1 3 2
1 2 1 0 0 0
0 0 2
−2 3 0 1
Diagonal entries (Definition 2.1.7.2)

Given a square matrix A = (ai j ) of order n, the diagonal

of A is the sequence of entries a11, a22, …, ann.
Each entry ai i is called a diagonal entry
while ai j , with i ≠ j, is called a non-diagonal entry.

diagonal a11 a12 ··· a1n

entries
a21 a22 ··· a2n
⁞ ⁞ ⁞
an1 an2 ··· ann

non-diagonal
entries
Diagonal matrices (Definition 2.1.7.2 & Example 2.1.8.2)

A square matrix is called a diagonal matrix if all its non-

diagonal entries are zero,
i.e. A = (ai j )n × n is a diagonal matrix
⇔ ai j = 0 whenever i ≠ j.

The following are some examples of diagonal matrices:

2 0 0 0
1 0 0
0 0 0 3 0 0
4 0 3 0
0 2 0 0 0 0
0 0 2
0 0 0 1
Scalar matrices (Definition 2.1.7.3 & Example 2.1.8.3)

A diagonal matrix is called a scalar matrix if all its diagonal

entries are the same,
i.e. A = (ai j )n × n is a scalar matrix
0 if i ≠ j
⇔ ai j = for a constant c.
c if i = j

The following are some examples of scalar matrices:

3 0 0 0
1 0 0
2 0 0 3 0 0
4 0 1 0
0 2 0 0 3 0
0 0 1
0 0 0 3
Identity matrices (Definition 2.1.7.4 & Example 2.1.8.4)

A diagonal matrix is called an identity matrix if all its

diagonal entries are 1.
We use In to denote the identity matrix of order n.
Sometimes we write I instead of In when there is no
danger of confusion.
The following are some examples of identity matrices:
1 0 0 0
1 0 0
1 0 0 1 0 0
I1 = 1 I2 = I3 = 0 1 0 I4 =
0 1 0 0 1 0
0 0 1
0 0 0 1
Zero matrices (Definition 2.1.7.5 & Example 2.1.8.5)

A matrix with all entries equal to zero is called a zero

matrix.
We use 0m × n to denote the zero matrix of size m × n.
Sometimes we write 0 instead of 0m× n when there is no
danger of confusion.
The following are some examples of zero matrices:
0 0 0
0 0 0 0 0 0 0
01 × 1 = 0 02 × 4 = 04 × 3 =
0 0 0 0 0 0 0
0 0 0
Symmetric matrices (Definition 2.1.7.6 & Example 2.1.7.6)

A square matrix (ai j ) is called symmetric if

ai j = aj i for all i, j.

The following are some examples of symmetric matrices:

2 1 6 −2
1 −1 0
0 4 1 3 0 −1
4 −1 3 2
4 2 6 0 0 0
0 2 2
−2 −1 0 1
Triangular matrices (Definition 2.1.7.7 & Example 2.1.7.7)

A square matrix (ai j ) is called upper triangular if

ai j = 0 whenever i > j.
A square matrix (ai j ) is called lower triangular if
ai j = 0 whenever i < j.
Both upper and lower triangular matrices are called
triangular matrices.
The following are some examples of triangular matrices:
lower triangular
2 0 0 0
0 4 1 −1 0 1 3 0 0
4 0 3 2
0 2 6 0 0 0
0 0 2 −2 −1 0 1
upper triangular
Chapter 2 Matrices

Section 2.2
Matrix Operations
Equal (Definition 2.2.1 & Example 2.2.2)

Two matrices are said to be equal if they have the same

size and their corresponding entries are equal.
Given A = (ai j )m × n and B = (bi j )p × q,
A is equal to B if m = p, n = q and ai j = bi j for all i, j.

1 x 1 −1 1 −1 0
Let A = , B= and C = .
2 4 2 4 2 4 0
Then A = B if and only if x = −1;
B≠C
and A ≠ C for any values of x.
Matrix addition (Definition 2.2.3.1)

Given A = (ai j )m × n and B = (bi j )m × n,

A + B = (ai j + bi j )m × n ,

a11 a12 ··· a1n b11 b12 ··· b1n

a21 a22 ··· a2n b21 b22 ··· b2n
i.e. A + B = +
⁞ ⁞ ⁞ ⁞ ⁞ ⁞
am1 am2 ··· amn bm1 bm2 ··· bmn

a11 + b11 a12 + b12 ··· a1n + b1n

a21 + b21 a22 + b22 ··· a2n + b2n
= .
⁞ ⁞ ⁞
am1 + bm1 am2 + bm2 ··· amn + bmn
An example (Example 2.2.4.1)

2 3 4 1 2 3
Let A = , and B = .
4 5 6 −1 −1 −1
2+1 3+2 4+3
Then A + B =
4 + (−1) 5 + (−1) 6 + (−1)
3 5 7
= .
3 4 5
Matrix subtraction (Definition 2.2.3.2)

Given A = (ai j )m × n and B = (bi j )m × n,

A − B = (ai j − bi j )m × n.

a11 a12 ··· a1n b11 b12 ··· b1n

a21 a22 ··· a2n b21 b22 ··· b2n
i.e. A − B = −
⁞ ⁞ ⁞ ⁞ ⁞ ⁞
am1 am2 ··· amn bm1 bm2 ··· bmn

a11 − b11 a12 − b12 ··· a1n − b1n

a21 − b21 a22 − b22 ··· a2n − b2n
= .
⁞ ⁞ ⁞
am1 − bm1 am2 − bm2 ··· amn − bmn
An example (Example 2.2.4.2)

2 3 4 1 2 3
Let A = , and B = .
4 5 6 −1 −1 −1
2−1 3−2 4−3
Then A − B =
4 − (−1) 5 − (−1) 6 − (−1)
1 1 1
= .
5 6 7
Scalar multiplication (Definition 2.2.3.3)

Given A = (ai j )m × n and a real constant c,

cA = (cai j )m × n
where the constant c is usually called a scalar.

a11 a12 ··· a1n ca11 ca12 ··· ca1n

a21 a22 ··· a2n ca21 ca22 ··· ca2n
i.e. cA = c = .
⁞ ⁞ ⁞ ⁞ ⁞ ⁞
am1 am2 ··· amn cam1 cam2 ··· camn
An example (Example 2.2.4.3)

2 3 4
Let A = .
4 5 6
4·2 4·3 4·4
Then 4A =
4·4 4·5 4·6
8 12 16
= .
16 20 24
Some remarks (Remark 2.2.5)

1. Given a matrix A, we normally use −A to denote the

matrix (−1)A.
2. The matrix subtraction can be defined using the matrix
addition:
Given two matrices A and B of the same size,
A − B is defined to be the matrix A + (−B).
Some basic properties (Theorem 2.2.6)

Let A, B, C be matrices of the same size and c, d are

scalars.
1. Commutative law for matrix addition: A + B = B + A.
2. Associative law for matrix addition:
A + (B + C) = (A + B) + C.
3. c(A + B) = cA + cB.
4. (c + d )A = cA + dA.
5. (cd )A = c(dA) = d(cA).
6. A + 0 = 0 + A = A.
0 is the zero matrix of
7. A − A = 0.
the same size as A.
8. 0A = 0.
0 is the number zero.
Proof of A + (B + C) = (A + B) + C (Theorem 2.2.6.2)

To prove A + (B + C) = (A + B) + C:
Recall that two matrices are equal if
(i) they have the same size and
(ii) their corresponding entries are equal.

(i) Since A, B, C are matrices of the same size, by

the definition of the matrix addition,
A + (B + C) and (A + B) + C
have the same size.
Proof of A + (B + C) = (A + B) + C (Theorem 2.2.6.2)

(ii) Let A = (ai j )m × n, B = (bi j )m × n and C = (ci j )m × n.

For any i, j,
the (i, j )-entry of A + (B + C)
= ai j + [ the (i, j )-entry of B + C ]
= ai j + [ bi j + ci j ] by the associative law for
= [ ai j + bi j ] + ci j real number addition
= [ the (i, j )-entry of A + B ] + ci j
= the (i, j )-entry of (A + B) + C.
By (i) and (ii), A + (B + C) = (A + B) + C.
The associative law (Remark 2.2.7)

Let A1, A2, …, Ak be matrices of the same size.

By the Associative Law for Matrix Addition, we can write
A1 + A2 + ··· + Ak
to represent the sum of the matrices without using any
parentheses to indicate the order of the matrix addition.
Matrix multiplication (Definition 2.2.8)

Given A = (ai j )m × p and B = (bi j )p × n,

The product AB is defined to be an m × n matrix whose
(i, j )-entry is
p
ai 1b1 j + ai 2b2 j + ··· + ai pbp j = Σ
k=1
ai kbkj .

for i = 1, 2, ..., m and j = 1, 2, ..., n.

(Remark 2.2.10.1)
We can only compute the product AB when the
number of columns of A is equal to the number of rows
of B.
Examples (Example 2.2.9.1)

1 1
1 2 3
2 3
4 5 6
−1 −2

1 · 1 + 2 · 2 + 3 · (−1) 1 · 1 + 2 · 3 + 3 · (−2)
=
4 · 1 + 5 · 2 + 6 · (−1) 4 · 1 + 5 · 3 + 6 · (−2)

2 1
= .
8 7
Examples (Example 2.2.9.2)

1 1
1 2 3
2 3
4 5 6
−1 −2

1·1 + 1·4 1·2 + 1·5 1·3 + 1·6

= 2·1 + 3·4 2·2 + 3·5 2·3 + 3·6
(−1) · 1 + (−2) · 4 (−1) · 2 + (−2) · 5 (−1) · 3 + (−2) · 6

5 7 9
= 14 19 24
−9 −12 −15
Multiplication isn’t commutative (Remark 2.2.10.2)

The matrix multiplication is not commutative,

i.e. in general, AB and BA are two different matrices
even the products exists.

−1 0 1 2
For example, let A = and B = .
2 3 3 0

−1 −2 3 6
Then AB = and BA = .
11 4 −3 0

Hence AB ≠ BA.
Multiplication isn’t commutative (Remark 2.2.10.3)

It would be ambiguous to say “the multiplication of a

matrix A to another matrix B” since it could mean AB
or BA.
To distinguish the two, we refer to
AB as the pre-multiplication of A to B
and BA as the post-multiplication of A to B.
When the product is a zero matrix (Remark 2.2.10.4)

When AB = 0, it is not necessary that A = 0 or B = 0.

0 1 1 1
For example, let A = and B = .
0 1 0 0
We have A ≠ 0, B ≠ 0 and
0 1 1 1 0 0
AB = = = 0.
0 1 0 0 0 0
Some basic properties (Theorem 2.2.11)

1. Associative law for matrix multiplication:

If A, B and C are m × p, p × q and q × n
matrices respectively, then
So we can write the product as
A(BC) = (AB)C. ABC without using parentheses.

2. Distribution laws for matrix addition and multiplication:

If A, B1 and B2 are m × p, p × n and p × n
matrices respectively, then
A(B1 + B2) = AB1 + AB2.
If A, C1 and C2 are p × n, m × p and m × p
matrices respectively, then
(C1 + C2)A = C1A + C2A.
Some basic properties (Theorem 2.2.11)

3. If A, B are m × p, p × n matrices, respectively, and

c is a scalar, then
c(AB) = (cA)B = A(cB).
4. If A is an m × n matrix, then
A0n × q = 0m × q,
0p × m A = 0p × n
and A In = Im A = A.
Proof of A(B1 + B2) = AB1 + AB2 (Theorem 2.2.11.2)

To prove A(B1 + B2) = AB1 + AB2:

Recall that two matrices are equal if
(i) they have the same size and
(ii) their corresponding entries are equal.

(i) Since the size of A is m × p and the size of

B1 + B2 is p × n, the size of A(B1 + B2) is m × n.
On the other hand, the sizes of both AB1 and AB2
are m × n and hence the size of AB1 + AB2 is
m × n.
Thus A(B1 + B2) and AB1 + AB2 have the same
size.
 Proof of A(B1 + B2) = AB1 + AB2 (Theorem 2.2.11.2)

(ii) Let A = (ai j )m × p, B1 = (bi j )p × n and B2 = (bi j' )p × n.

For any i, j,
the (i, j )-entry of A(B1 + B2)
= ai 1 [ the (1, j )-entry of B1 + B2 ]
+ ai 2 [ the (2, j )-entry of B1 + B2 ]
+ ···
+ ai p [ the (p, j )-entry of B1 + B2 ]
= ai 1 [ b1j + b1j' ] + ai 2 [ b2 j + b2 j' ] + ··· + ai p [ bp j + bp j' ]
= ai 1b1j + ai 1b1j' + ai 2b2 j + ai 2b2 j' + ··· + ai pbp j + ai pbp j'.

by the distributive law for real numbers

Proof of A(B1 + B2) = AB1 + AB2 (Theorem 2.2.11.2)

On the other hand,

the (i, j )-entry of AB1 + AB2
= [ the (i, j )-entry of AB1 ] + [ the (i, j )-entry of AB2 ]
= [ ai 1b1j + ai 2b2 j + ··· + ai pbp j ]
+ [ ai 1b1j' + ai 2b2 j' + ··· + ai pbp j' ]
= ai 1b1j + ai 1b1j' + ai 2b2 j + ai 2b2 j' + ··· + ai pbp j + ai pbp j'
= the (i, j )-entry of A(B1 + B2). from the
previous slide
By (i) and (ii), A(B1 + B2) = AB1 + AB2.
 (Definition 2.2.12
Powers of square matrices & Example 2.2.13)

Let A be a square matrix and n a nonnegative integer.

We define An as follows: the identity matrix
I if n = 0
An =
AA ··· A if n ≥ 1.
n times

1 2
For example, let A = .
1 3

1 2 1 2 1 2 11 30
Then A3 = = .
1 3 1 3 1 3 15 41
Powers of square matrices (Remark 2.2.14)

1. Let A be a square matrix and n, m nonnegative

integers. Then AmAn = Am + n.
2. Since matrix multiplication is not commutative, in
general, for two square matrix A and B of the same
size, (AB)2 and A2B2 may be different.
1 0 0 1
For example, let A = and B = .
0 2 1 0
2 0
Then (AB)2 = (AB)(AB) = ABAB =
0 2
1 0
and A2B2 = (AA)(BB) = AABB = .
0 4
Some useful notation (Notation 2.2.15)

Given A = (ai j )m × p, we can write

a1
a2
A= where ai = ai 1 ai 2 ··· ai p
⁞
am is the i th row of A.

Given B = (bi j )p × n, we can write B = b1 b2 ··· bn

b1 j
b2 j
where bj = is the j th column of B.
⁞
bp j
Some useful notation (Notation 2.2.15)

Then a1b1 a1b2 ··· a1bn

a2b1 a2b2 ··· a2bn
AB =
⁞ ⁞ ⁞
amb1 amb2 ··· ambn

b1 j
b2 j The i th row of A
where ai bj = ai 1 ai 2 ··· ai p pre-multiply to
⁞ the j th column of B.
bp j

= ai 1b1 j + ai 2b2 j + ··· + ai pbp j

which is the (i, j )-entry of AB.
Some useful notation (Notation 2.2.15)

Also we can write

AB = A b1 b2 ··· bn = Ab1 Ab2 ··· Abn ,

where Abj is the j th column of AB;

or
a1 a1B
a2 a2B
AB = B=
⁞ ⁞
am amB

where ai B is the i th row of AB.

 The j th column of AB is
a11b1j + a12b2j + ··· + a1pbpj a11 a12 ··· a1p b1j
a21b1j + a22b2j + ··· + a2pbpj a21 a22 ··· a2p b2j
= = Abj .
⁞ ⁞ ⁞ ⁞ ⁞
am1b1j + am2b2j + ··· + ampbpj am1 am2 ··· amp bpj

The i th row of AB is
ai1b11 + ai2b21 + ··· + aipbp1 ai1b12 + ai2b22 + ··· + aipbp2 ··· ai1b1n + ai2b2n + ··· + aipbpn

b11 b12 ··· b1n

b21 b22 ··· b2n
= ai 1 ai 2 ··· ai p = ai B.
⁞ ⁞ ⁞
bp1 bp2 ··· bpn
An example (Example 2.2.16)

1 2 3
Let A = with a1 = 1 2 3 and a2 = 4 5 6 ;
4 5 6

1 1 1 1
B= 2 3 with b1 = 2 and b2 = 3 .
−1 −2 −1 −2

a1
Then A = , B= b1 b2 and
a2
a1b1 a1b2 a1B 2 1
AB = = Ab1 Ab2 = = .
a2b1 a2b2 a2B 8 7
 Linear systems (Remark 2.2.17)

Consider the system of linear equations:

a11x1 + a12x2 + ··· + a1nxn = b1
a21x1 + a22x2 + ··· + a2nxn = b2
⁞ ⁞
am1x1 + am2x2 + ··· + amnxn = bm .

Using matrix multiplication, the system can be rewritten

as:
a11 a12 ··· a1n x1 a11x1 + a12x2 + ··· + a1nxn b1
a21 a22 ··· a2n x2 a21x1 + a22x2 + ··· + a2nxn b2
= = .
⁞ ⁞ ⁞ ⁞ ⁞ ⁞
am1 am2 ··· amn xn am1x1 + am2x2 + ··· + amnxn bm
Linear systems (Remark 2.2.17)

a11 a12 ··· a1n x1 b1

a21 a22 ··· a2n x2 b2
= .
⁞ ⁞ ⁞ ⁞ ⁞
am1 am2 ··· amn xn bm

x1 b1
x2 b2
Let A = (ai j )m × n, x = and b = .
⁞ ⁞ b is called the
xn bm constant matrix.

We represent the system by the matrix equation Ax = b.

A is called the
x is called the
coefficient matrix.
variable matrix.
Linear systems (Remark 2.2.17)

Write A = c1 c2 ··· cn where cj is the j th column of

A.
The linear system can also be represented by
a11 a12 a1n b1
a21 a22 a2n b2
x1 + x2 + ··· + xn = ,
⁞ ⁞ ⁞ ⁞
am1 am2 amn bm
n
i.e. x1c1 + x2c2 + ··· + xncn = b or Σ xj cj = b.
j=1
An example (Example 2.2.18)

The system of linear equations

4x + 5y + 6z = 5
x− y =2
y− z=3

4 5 6 x 5
can be written as 1 −1 0 y = 2
0 1 −1 z 3

4 5 6 5
or x 1 + y −1 +z 0 = 2 .
0 1 −1 3
Solutions to linear systems

We follow the notation of Remark 2.2.16, an n × 1 matrix

u is said to be a solution to the linear system Ax = b if
the equation is satisfied when we substitute x = u into
the equation, i.e. Au = b.
For example, in Example 2.2.17,
3 4 5 6 x 5
u= 1 is a solution to 1 −1 0 y = 2
−2 0 1 −1 z 3

4 5 6 3 5
because 1 −1 0 1 = 2 .
0 1 −1 −2 3
 (Definition 2.2.19 & Example 2.2.20
Transposes & Remark 2.2.21.1)

Let A = (ai j ) be an m × n matrix.

The transpose of A, denoted by AT (or At ), is an
n × m matrix whose (i, j )-entry is aj i.

1 5
1 2 3 4 2 6
Let A = . Then AT = .
5 6 7 8 3 7
4 8

Note that the rows of A are the columns of AT and vice

versa.
Symmetric matrices (Remark 2.2.21.2)

Recall that a square matrix (ai j ) is called symmetric if

ai j = aj i for all i, j.

Thus a square matrix A is symmetric if and only if

A = AT.
Some basic properties (Theorem 2.2.22)

Let A be an m × n matrix.
1. (AT)T = A.
2. If B be an m × n matrix, then (A + B)T = AT + BT.
3. If c is a scalar, then (cA)T = cAT.
4. If B be an n × p matrix, then (AB)T = BTAT.
Proof of (AB)T = BTAT (Theorem 2.2.22.4)

To prove (AB)T = BTAT:

Recall that two matrices are equal if
(i) they have the same size and
(ii) their corresponding entries are equal.

(i) Since the size of AB is m × p, the size of (AB)T is

p × m.
On the other hand, the sizes of BT and AT are
p × n and n × m, respectively, and hence the size of
BTAT is p × m.
Thus (AB)T and BTAT have the same size.
Proof of (AB)T = BTAT (Theorem 2.2.11.2)

(ii) Let A = (ai j )m × n and B = (bi j )n × p .

For any i, j,
the (i, j )-entry of AB = ai 1b1j + ai 2b2 j + ··· + ai n bn j.
Thus
the (i, j )-entry of (AB)T
= the ( j, i )-entry of AB
= aj 1b1i + aj 2b2 i + ··· + aj n bn i.
Proof of (AB)T = BTAT (Theorem 2.2.11.2)

On the other hand, write AT = (ai j' )n × m and

BT = (bi j' )p × n where ai j' = aj i and bi j' = bj i.
For any i, j,
the (i, j )-entry of BTAT
= bi 1' a1j' + bi 2' a2 j' + ··· + bi n' an j'
= b1i aj 1 + b2 i aj 2 + ··· + bn i aj n
= aj 1b1i + aj 2b2 i + ··· + aj n bn i
= the (i, j )-entry of (AB)T. from the
previous slide
By (i) and (ii), (AB)T = BTAT.
Chapter 2 Matrices

Section 2.3
Inverses of Square Matrices
Inverses (Discussion 2.3.1)

Let a and b be two real number such that a ≠ 0. Then

the solution of the equation
ax = b
b
is x = a = a−1b.
Let A and B be two matrices. It is much harder to solve
the matrix equation
AX = B
because we do not have “division” for matrices.
However, for some square matrices, we can find their
“inverses” which have the similar property as a−1 in the
computation of the solution of ax = b above.
Inverses of square matrices (Definition 2.3.2)

Let A be a square matrix of order n.

Then A is said to be invertible if there exists a square
matrix B of order n such that
AB = I and BA = I.
The matrix B here is called an inverse of A.
A square matrix is called singular if it has no inverse.
Examples (Example 2.3.3.1)

2 −5 3 5
Let A = and B = .
−1 3 1 2
Then
2 −5 3 5 1 0
AB = = = I,
−1 3 1 2 0 1

3 5 2 −5 1 0
BA = = = I.
1 2 −1 3 0 1

So A is invertible and B is an inverse of A.

Examples (Example 2.3.3.2)

2 −5 4
X=
−1 3 0

3 5 2 −5 3 5 4
⇒ X=
1 2 −1 3 1 2 0

12
⇒ IX =
4

12
⇒ X=
4
By Theorem 2.2.11.3, I X = X.
Examples (Example 2.3.3.3) In the next section, we
shall learn a
systematic method to
1 0 check whether a
Show that A = is singular. square matrix is
1 0
invertible.
(Proof by contradiction)
a b
Assume the contrary, i.e. A has an inverse B = ,
c d
1 0
i.e. BA = I = .
0 1
a b 1 0 a+b 0
On the other hand, BA = = .
c d 1 0 c+d 0
It is impossible.
Hence our assumption is wrong, i.e. A is singular.
Cancellation laws (Remark 2.3.4)

1. Cancellation laws for matrix multiplication:

Let A be an invertible m × m matrix.
(a) If B1 and B2 are m × n matrices such that
AB1 = AB2, then B1 = B2.
(b) If C1 and C2 are n × m matrices such that
C1A = C2A, then C1 = C2.
2. If A is not invertible, the cancellation laws may not
hold.
1 0 2 0
For example, let A = , B1 = and
1 0 0 1
2 0
B2 = . Then AB1 = AB2 but B1 ≠ B2.
1 3
Uniqueness of inverses (Theorem 2.3.5)

If B and C are inverses of a square matrix A, then

B = C.

Proof: By the definition of inverses (Definition 2.3.2),

AB = I, BA = I and AC = I, CA = I.

So AB = I ⇒ CAB = CI ⇒ I B = C ⇒ B = C.
(Notation 2.3.6)

Given an invertible matrix A, since there is only one

inverse of A, we use the symbol A−1 to denote this
unique inverse of A.
 2 x 2 matrices (Example 2.3.8)

a b
Let A = . If ad − bc ≠ 0, then A is invertible and
c d
d −b
1 d −b ad − bc ad − bc
A = ad − bc
−1 = −c a
.
−c a
ad − bc ad − bc
d −b
ad − bc ad − bc
Proof: Let B = −c a
.
ad − bc ad − bc (Remark 2.3.7)
To show that A is invertible and B is the inverse of A,
by Definition 2.3.2, we need to check AB = I and BA = I.
(By Theorem 2.4.12 in the next section, we shall see that we only need
to check one of the two conditions: AB = I or BA = I.)
2 x 2 matrices (Example 2.3.8)

d −b ad − bc −ab + ba
a b ad − bc ad − bc 1ad 0− bc ad − bc
AB = −c a
= cd − dc
= I,
−cb + da
c d 0 1
ad − bc ad − bc ad − bc ad − bc

d −b da − bc db − bd
ad − bc ad − bc a b 1ad0− bc ad − bc
BA = −c
= = I.
a c d −ca
0 1 + ac −cb + da
ad − bc ad − bc ad − bc ad − bc

Hence A is invertible and A−1 = B.

Some basic properties (Theorem 2.3.9)

Let A, B be two invertible matrices and c a nonzero

scalar.
1 −1
1. cA is invertible and (cA)−1 = cA .
2. AT is invertible and (AT)−1 = (A−1)T.
3. A−1 is invertible and (A−1)−1 = A.
4. AB is invertible and (AB)−1 = B−1A−1.
(Remark 2.3.10)

By Part 4, if A1, A2, …, Ak are invertible matrices,

then A1A2 ··· Ak is invertible and
(A1A2 ··· Ak)−1 = Ak−1 ··· A2−1A1−1.
 AT is invertible (Theorem 2.3.9.2)

To prove that AT is invertible and (AT)−1 = (A−1)T:

By Remark 2.3.7, we only need to show that
AT(A−1)T = I and (A−1)TAT = I.
(By Theorem 2.4.12 in the next section, we shall see
that we only need to check one of the two conditions:
by Theorem 2.2.22.4 AT(A−1)T = I or (A−1)TAT = I.)

AT(A−1)T = (A−1A)T = I T = I
and similarly (A−1)TAT = (AA−1)T = I T = I.
So AT is invertible and (AT)−1 = (A−1)T.
 (Definition 2.3.11
Powers of square matrices & Example 2.3.12)

Let A be an invertible matrix and n a positive integer.

We define A−n as follows:
A−n = (A−1)n = A−1A−1 ··· A−1.
n times

1 2 3 −2
For example, let A = . Then A−1 = .
1 3 −1 1

3 −2 3 −2 3 −2
So A−3 = (A−1)3 =
−1 1 −1 1 −1 1
41 −30
= .
−15 11
Some basic properties (Remark 2.3.13)

Let A be an invertible matrix.

1. ArAs = Ar + s for any integers r and s.
2. An is invertible and (An)−1 = (A−1)n.
Chapter 2 Matrices

Section 2.4
Elementary Matrices
Some useful terms from Chapter 1 (Definition 2.4.1)

In Chapter 1, the following concepts are defined for

augmented matrices:
elementary row operations,
row equivalent matrices,
row-echelon forms,
reduced row-echelon forms,
Gaussian Elimination,
Gauss-Jordan Elimination.
From now on, these terms will also be used for matrices.
Elementary row operations (Discussion 2.4.2.1)

Multiply a row by a constant:

1 0 2 3 1 0 2 3
2R2
A = 2 −1 3 6 B= 4 −2 6 12
1 4 4 0 1 4 4 0

1 0 0
Let E1 = 0 2 0 .
0 0 1
1 0 0 1 0 2 3 1 0 2 3
Then E1A = 0 2 0 2 −1 3 6 = 4 −2 6 12 = B.
0 0 1 1 4 4 0 1 4 4 0
Elementary row operations (Discussion 2.4.2.1)

In general, let A be an m × n matrix and E an m × m

matrix such that
i column
th

1
0
1
E= k . i th row
1
0
1

kRi
Then A EA.
Elementary row operations (Discussion 2.4.2.1)

If k ≠ 0, then E is invertible
and
i column
th

1 0
1
E −1 = k
. i th row
1
0
1
1
k Ri
Note that A E −1A.
Elementary row operations (Discussion 2.4.2.2)

Interchange two rows:

1 0 2 3 1 0 2 3
R2 ↔ R3
A = 2 −1 3 6 B= 1 4 4 0
1 4 4 0 2 −1 3 6

1 0 0
Let E2 = 0 0 1 .
0 1 0
1 0 0 1 0 2 3 1 0 2 3
Then E2A = 0 0 1 2 −1 3 6 = 1 4 4 0 = B.
0 1 0 1 4 4 0 2 −1 3 6
 Elementary row operations (Discussion 2.4.2.2)

In general, let A be an m × n matrix and E an m × m

matrix such that
i th column j th column

1
i th row 0
1
0 1
1
Ri ↔ Rj
E= . Then A EA.
1
1 0 E is invertible and
j th row
1 E−1 = E.
0
1
Elementary row operations (Discussion 2.4.2.3)

Add a multiple of a row to another row:

1 0 2 3 1 0 2 3
R3 + 2R1
A= 2 −1 3 6 B= 2 −1 3 6
1 4 4 0 3 4 8 6

1 0 0
Let E3 = 0 1 0 .
2 0 1
1 0 0 1 0 2 3 1 0 2 3
Then E3A = 0 1 0 2 −1 3 6 = 2 −1 3 6 = B.
2 0 1 1 4 4 0 3 4 8 6
Elementary row operations (Discussion 2.4.2.3)

In general, let A be an m × n matrix and E an m × m

matrix such that if i < j,
i th column j th column

1
0
1
1
E= ;
k 1 j th row
1
0
1
Elementary row operations (Discussion 2.4.2.3)

and if i > j, j th column i th column

1
0
1
1 k j th row
E= .
1
1
0
1

Rj + kRi
Then A EA.
Elementary row operations (Discussion 2.4.2.3)

E is invertible
and if i < j, i th column j th column

1
0
1
1
E−1 = ;
−k 1 j th row
1
0
1
Elementary row operations (Discussion 2.4.2.3)

j th column i th column
and if i > j,
1
0
1
1 −k j th row
E−1 = .
1
1
0
1

Rj − kRi
Note that A E−1A.
Elementary matrices (Definition 2.4.3 & Remark 2.4.4)

A square matrix is called an elementary matrix if it can be

obtained from an identity matrix by performing a single
elementary row operation.

The three types of matrices E described in the previous

slides (Discussion 2.4.2) are elementary matrices
and every elementary matrix belongs to one of these
three types.

All elementary matrices are invertible and their inverses

are also elementary matrices.
 An example (Example 2.4.5)

0 4 2 1 0 2
R1 ↔ R3
A = −2 1 −3 −2 1 −3
E1
1 0 2 0 4 2

0 0 1 0 4 2 1 0 2
E1A = 0 1 0 −2 1 −3 = −2 1 −3
1 0 0 1 0 2 0 4 2

0 0 1
E1 = 0 1 0
1 0 0
 An example (Example 2.4.5)

1 0 2 1 0 2
R1 ↔ R3 R2 + 2R1
A −2 1 −3 0 1 1
E1 E2
0 4 2 0 4 2

1 0 0 1 0 2 1 0 2
E2E1A = 2 1 0 −2 1 −3 = 0 1 1
0 0 1 0 4 2 0 4 2

0 0 1 1 0 0
E1 = 0 1 0 E2 = 2 1 0
1 0 0 0 0 1
 An example (Example 2.4.5)

1 0 2 1 0 2
R1 ↔ R3 R2 + 2R1 R3 − 4R2
A 0 1 1 0 1 1
E1 E2 E3
0 4 2 0 0 −2

1 0 0 1 0 2 1 0 2
E3E2E1A = 0 1 0 0 1 1 = 0 1 1
0 −4 1 0 4 2 0 0 −2

0 0 1 1 0 0 1 0 0
E1 = 0 1 0 E2 = 2 1 0 E3 = 0 1 0
1 0 0 0 0 1 0 −4 1
 An example (Example 2.4.5)

1 0 2 1 1 0 2
R1 ↔ R3 R2 + 2R1 R3 − 4R2 − 2 R3
A 0 1 1 0 1 1 =B
E1 E2 E3 E4
0 0 −2 0 0 1

1 0 0 1 0 2 1 0 2
E4E3E2E1A = 0 1 0 0 1 1 = 0 1 1
1
0 0 −2 0 0 −2 0 0 1

0 0 1 1 0 0 1 0 0 1 0 0
E1 = 0 1 0 E2 = 2 1 0 E3 = 0 1 0 E4 = 0 1 0
1
1 0 0 0 0 1 0 −4 1 0 0 −2
 An example (Example 2.4.5)

0 4 2 1 0 2 0 0 1 1 0 0
A = −2 1 −3 B= 0 1 1 E1 = 0 1 0 E2 = 2 1 0
1 0 2 0 0 1 1 0 0 0 0 1
We have seen that 1 0 0 1 0 0
E4E3E2E1A = B. E3 = 0 1 0 E4 = 0 1 0
1
0 −4 1 0 0 −2
0 0 1 1 0 0
E1−1 = 0 1 0 E2−1 = −2 1 0 Then
1 0 0 0 0 1 A = (E4E3E2E1)−1B
1 0 0 1 0 0 = E1−1E2−1E3−1E4−1B.
E3−1 = 0 1 0 E4−1 = 0 1 0
0 4 1 0 0 −2
System of linear equations (Remark 2.4.6)

Recall Theorem 1.2.7: If augmented matrices of two

systems of linear equations are row equivalent, then the
two systems have the same set of solutions.

linear system same linear system

Ax = b solutions Cx = d

⇑
augmented matrix row augmented matrix
(A|b) equivalent (C|d)
System of linear equations (Remark 2.4.6)

Proof: Since ( A | b ) and ( C | d ) are row equivalent,

one can be obtained by the other by a series of
elementary row operations.
Thus (by Discussion 2.4.2) there exists elementary matrices
E1, E2, …, Ek such that
Ek ··· E2E1( A | b ) = ( C | d )
⇒ ( Ek ··· E2E1A | Ek ··· E2E1b ) = ( C | d )
⇒ Ek ··· E2E1A = C and Ek ··· E2E1b = d.
Note that A = (Ek ··· E2E1)−1C = E1−1E2−1 ··· Ek−1C
and b = (Ek ··· E2E1)−1d = E1−1E2−1 ··· Ek−1d.
System of linear equations (Remark 2.4.6)

If x = u is a solution to Ax = b, then
Au = b ⇒ Ek ··· E2E1Au = Ek ··· E2E1b
⇒ Cu = d
and hence x = u is a solution to Cx = d.
If x = v is a solution to Cx = d, then
Cv = d ⇒ E1−1E2−1 ··· Ek−1Cv = E1−1E2−1 ··· Ek−1d
⇒ Av = b
and hence x = v is a solution to Ax = b.
So we have shown that Ax = b and Cx = d has the
same set of solutions.
Invertible matrices (Theorem 2.4.7)

Let A be a square matrix.

The following statements are equivalent:
1. A is invertible.
2. The linear system Ax = 0 has only the trivial solution.
3. The reduced row-echelon form of A is an identity
matrix.
4. A can be expressed as a product of elementary
matrices.
Invertible matrices (Theorem 2.4.7)

1 ⇒ 2: If A is invertible, then
Ax = 0 ⇒ A−1Ax = A−10
⇒ Ix = 0
⇒ x = 0.
and hence the system Ax = 0 has only the trivial
solution.
2 ⇒ 3: Suppose the system Ax = 0 has only the trivial
solution.
The augmented matrix of the system is ( A | 0 ).
 Invertible matrices (Theorem 2.4.7)

Since Ax = 0 has only the trivial solution, every column of its row echelon
form is a pivot column (see Remark 1.4.8.2).
the number of nonzero rows
⊗ * ··· * 0 = the number of leading entries
Gaussian 0 ⊗ ··· * 0 = the number of pivot columns
(A|0)
Elimination ⁞ ⁞ =n
0 ··· 0 ⊗ 0 Hence there is no zero row.
Suppose the size
of A is n × n. 1 0 ··· 0 0
Gauss-Jordan 0 1 ··· 0 0
=(I|0)
Elimination ⁞ ⁞
0 ··· 0 1 0
Thus the reduced row-echelon form of A is an
identity matrix.
Invertible matrices (Theorem 2.4.7)

3 ⇒ 4: Since the reduced row-echelon form of A is an

identity matrix I, there exists elementary matrices
E1, E2, …, Ek such that
Ek ··· E2E1A = I.
Then A = (Ek ··· E2E1)−1I
= (Ek ··· E2E1)−1 = E1−1E2−1 ··· Ek−1
where E1−1, E2−1, …, Ek−1 are also elementary
matrices.
4 ⇒ 1: Suppose A is a product of elementary matrices.
Since elementary matrices are invertible, A is
invertible (see Remark 2.3.10).
Finding inverses (Discussion 2.4.8)

Let A be an invertible matrix of order n.

There exists elementary matrices E1, E2, …, Ek such
that
Ek ··· E2E1A = I ⇒ Ek ··· E2E1AA−1 = I A−1
⇒ Ek ··· E2E1I = A−1
⇒ Ek ··· E2E1 = A−1.
Construct an n × 2n matrix ( A | I ).
Then Ek ··· E2E1( A | I ) = ( Ek ··· E2E1A | Ek ··· E2E1I )
= ( I | A−1 ).
Thus Gauss-Jordan
(A|I) ( I | A−1 ).
Elimination
An example (Example 2.4.9)

1 2 3
Let A = 2 5 3 .
1 0 8

1 2 3 1 0 0 1 0 0 −40 16 9
Gauss-Jordan
2 5 3 0 1 0 0 1 0 13 −5 −3
Elimination
1 0 8 0 0 1 0 0 1 5 −2 −1

−40 16 9
So A−1 = 13 −5 −3 .
5 −2 −1
To check whether invertible (Remark 2.4.10)

Let A be a square matrix.

Recall that if A is invertible, then
⊗ * ··· * 0
Gaussian 0 ⊗ * * 0
(A|0) .
Elimination ⁞ ⁞
0 ··· 0 ⊗ 0 no zero row

Hence if a row-echelon form of A has at least one zero

row, A is singular (i.e. not invertible).
Examples (Example 2.4.11)

1. 1 2 0 1 2 0
Gaussian
A= 2 4 4 0 0 4
Elimination
3 6 3 0 0 0
So A is singular.
a b
2. Let A = .
c d
We already know that if ad − bc ≠ 0, then A is
invertible (see Example 2.3.8).
Actually, A is invertible if and only if ad − bc ≠ 0.
(By using Remark 2.4.10, we can show that if A is invertible,
then ad − bc ≠ 0. Read our textbook for the detailed proof.)
Invertible matrices (Theorem 2.4.12)

Let A and B be a square matrix of the same size.

If AB = I, then
(i) A is invertible,
(ii) B is invertible,
(iii) A−1 = B,
(iv) B−1 = A,
(v) BA = I.
Invertible matrices (Theorem 2.4.12)

Proof: Consider the linear system Bx = 0:

Bx = 0 ⇒ ABx = A0 ⇒ I x = 0 ⇒ x = 0.
As the linear system Bx = 0 has only the trivial solution,
(by Theorem 2.4.7) B is invertible.

Since B is invertible,
AB = I ⇒ ABB−1 = I B−1 ⇒ AI = B−1 ⇒ A = B−1.
Then (by Theorem 2.3.9.3) A is invertible
and A−1 = (B−1)−1 = B.
Finally, BA = BB−1 = I.
An example (Example 2.4.13)

Suppose A is a square matrix such that

A2 − 3A − 6I = 0.
Then A2 − 3A = 6I ⇒ AA − A(3I ) = 6I
⇒ A(A − 3I ) = 6I
1
⇒ 6 A(A − 3I ) = I
1
⇒ A 6 (A − 3I ) = I.
1
So A is invertible and A−1 = 6 (A − 3I ).
Singular matrices (Theorem 2.4.14)

Let A and B be square matrices of the same size.

If A is singular, then both AB and BA are singular.
Elementary column operations (Discussion 2.4.15)

Let A be an m × n matrix and E an n × n elementary

matrix.
an elementary
Then A EA.
row operation

How is A related to AE ?
Answer: AE can be obtained from A by doing an
“elementary column operation”.
(Read our textbook for more details.)
Chapter 2 Matrices

Section 2.5
Determinants
Invertible matrices (Discussion 2.5.1)

We know that a 2 × 2 matrix

a b
c d
is invertible if and only if ad − bc ≠ 0 (see Example 2.4.11.2).
We have a similar formula to determine whether a square
matrix of higher order is invertible.
The formula involves a quantity called “determinant”.
Determinants (Definition 2.5.2 & Notation 2.5.3)

Let A = (ai j ) be an n × n matrix.

Let Mi j be an (n − 1) × (n − 1) matrix obtained from A
by deleting the i th row and the j th column.
Then the determinant of A is defined to be
a11 if n = 1
det(A) =
a11A11 + a12A12 + ··· + a1n A1n if n > 1
where Ai j = (−1)i + j det(Mi j ) which is called the
(i, j )-cofactor of A.
a11 a12 ··· a1n
a21 a22 ··· a2n
Sometimes, we also write det(A) as .
⁞ ⁞ ⁞
an1 an2 ··· ann
2 × 2 matrices (Example 2.5.4.1)

a b
Let A = .
c d
− +
Then M11 = d , M12 = c ,
A11 = (−1)1 + 1 det(M11) = d
and A12 = (−1)1 + 2 det(M12) = −c.

a b
So det(A) = = a11A11 + a12A12 = ad − bc.
c d
Examples (Example 2.5.4.2)

−3 −2 4
Let B = 4 3 1 .
0 2 4

3 1 4 1 4 3
Then det(B) = (−3) − (−2) +4
2 4 0 4 0 2
= −3(3 · 4 − 1 · 2) + 2(4 · 4 − 1 · 0) + 4(4 · 2 − 3 · 0)
= 34.
Examples (Example 2.5.4.3)

0 −1 2
0
2 −3 3
−2
Let C = .
0 2 4
0
0 0 2
−1
−3 3 −2 2 3 −2
Then det(C) = 0 2 4 0 − (−1) 0 4 0
0 2 −1 0 2 −1
2 −3 −2 2 −3 3
+2 0 2 0 −0 0 2 4
0 0 −1 0 0 2
Examples (Example 2.5.4.3)

2 3 −2 2 −3 −2
det(C) = 0 4 0 + 2 0 2 0
0 2 −1 0 0 −1

4 0 0 0 0 4
= 2 −3 + (−2)
2 −1 0 −1 0 2

2 0 0 0 0 2
+2 2 − (−3) + (−2)
0 −1 0 −1 0 0

= [ 2 · (−4) − 3 · 0 − 2 · 0 ] + 2 [ 2 · (−2) + 3 · 0 − 2 · 0 ]
= −16.
3 × 3 matrices (Remark 2.5.5)

a b c
Let A = d e f .
g h i
Then det(A) = aei + bfg + cdh − ceg − afh − bdi.
The formula in can be easily a b c a b
remembered by using diagram on d e f d e
the right: g h i g h
− − − + + +
Warning: The method shown here cannot be
generalized to higher order.
Geometrical interpretation

(c, d ) (a + c, b + d )

(0, 0) (a, b)

a b
The area of the parallelogram is ad − bc = .
c d
Geometrical interpretation

(g, h, i ) (d + g, e + h, f + i )

(a + d + g, b + e + h, c + f + i )
(a + g, b + h, c + i )

(0, 0, 0)
(d, e, f )

(a, b, c)
(a + d, b + e, c + f )

The volume of the parallelepiped is a b c

aei + bfg + cdh − ceg − afh − bdi = d e f .
g h i
Cofactor expansions (Theorem 2.5.6)

Let A = (ai j ) be an n × n matrix.

Recall that Mi j be an (n − 1) × (n − 1) matrix obtained
from A by deleting the i th row and the j th column
and Aij = (−1)i + j det(Mi j ).
Then for any i = 1, 2, ..., n and j = 1,2, ..., n,
det(A) = ai1Ai1 + ai 2Ai 2 + ··· + ai n Ai n cofactor expansion
= a1j A1j + a2j A2j + ··· + an j An j . along the i th row

cofactor expansion
(Since the proof involves some deeper along the j th column
knowledge of determinants, we use this
result without proving it.)
An example (Example 2.5.7)

−3 −2 4
Let B = 4 3 1 .
0 2 4

−2 4 −3 4 −3 −2
Then det(B) = − 4 +3 −1 = 34
2 4 0 4 0 2
4 3 −3 −2 −3 −2
=4 −1 +4 = 34.
0 2 0 2 4 3
Triangular matrices (Theorem 2.5.8)

If A is an n × n triangular matrix, i.e.

a11 a12 ··· a1n a11 0 ··· 0

0 a22 ··· a2n a21 a22 ⁞
A= or ,
⁞ ⁞ ⁞ ⁞ 0
0 ··· 0 ann an1 an2 ··· ann

then det(A) = a11a22 ··· ann.

Triangular matrices (Theorem 2.5.8)

Proof: A proper proof using

a11 0 ··· 0 mathematical induction
a21 a22 ⁞ can be found in our
Let A = . textbook.
⁞ ⁞ 0
an1 an2 ··· ann

Then det(A) = a11A11 + 0A12 + ··· + 0A1n

a22 0 ··· 0
a22 a33 ⁞
= a11 .
⁞ ⁞ 0
an2 an2 ··· ann
The result follows by repeating this process.
Examples (Example 2.5.9)

1 0 ··· 0
0 1 ⁞
det(I ) = = 1 · 1 ··· 1 = 1.
⁞ 0
0 ··· 0 1

−2 3.5 2
0 5 −26 = (−2) · 5 · 2 = -20.
0 0 2

−2 0 0
99 0 0 = (−2) · 0 · 2 = 0.
10 −4.5 2
Transposes (Theorem 2.5.10 & Example 2.5.11)

If A is a square matrix, then det(AT ) = det(A).

0 −1 2 0 0 2 0 0
2 −3 3 −2 −1 −3 2 0
Let C = . Then C =
T .
0 2 4 0 2 3 4 2
0 0 2 −1 0 −2 0 −1
−1 2 0
Then det(CT ) = −2 2 4 2
0 0 −1
= −2 [ (−1) · 4 · (−1) + 2 · 2 · 0 + 0 · 2 · 0
− 0 · 4 · 0 − (−1) · 2 · 0 − 2 · 2 · (−1) ]
= −16 = det(C) (see Example 2.5.4.3).
 (Theorem 2.5.12
Identical rows or columns & Example 2.5.13)

1. The determinant of a square matrix with two identical

rows is zero.
2. The determinant of a square matrix with two identical
columns is zero.

The following matrices have zero determinant:

1 0 0 1
1 2 4
4 −2 −1 −3 −3 9
−1 10 4
4 −2 2 4 4 0
1 2 4
0 −2 −2 −1
Elementary row operations (Discussion 2.5.14.1)

a b c a b c
kR2
A= d e f B= kd ke kf
g h i g h i

det(B) = a(ke)i + b(kf )g + c(kd )h − c(ke)g − a(kf )h − b(kd )i

= k(aei + bfg + cdh − ceg − afh − bdi )
= k det(A).
Elementary row operations (Discussion 2.5.14.1)

1 0 0
Let E = 0 k 0 .
0 0 1

Then B = EA.
Since det(E ) = 1 · k · 1 = k,
det(E ) det(A) = k det(A) = det(B) = det(EA).
Elementary row operations (Discussion 2.5.14.2)

a b c a b c
R2 ↔ R3
A= d e f B= g h i
g h i d e f

det(B) = ahf + bid + cge − chd − aie − bgf

= −(aei + bfg + cdh − ceg − afh − bdi )
= −det(A).
Elementary row operations (Discussion 2.5.14.2)

1 0 0
Let E = 0 0 1 .
0 1 0

Then B = EA.
0 1
Since det(E ) = 1 = 0 · 0 − 1 · 1 = −1,
1 0
det(E ) det(A) = −det(A) = det(B) = det(EA).
Elementary row operations (Discussion 2.5.14.3)

a b c a b c
R3 + kR1
A= d e f B= d e f
g h i g + ka h + kb i + kc

det(B) = ae(i + kc) + bf(g + ka ) + cd(h + kb)

− ce(g + ka ) − af(h + kb) − bd(i + kc)
= aei + bfg + cdh − ceg − afh − bdi
+ k(aec + bfa + cdb − cea − afb − bdc )
= aei + bfg + cdh − ceg − afh − bdi
= det(A).
Elementary row operations (Discussion 2.5.14.3)

1 0 0
Let E = 0 1 0 .
k 0 1

Then B = EA.
Since det(E ) = 1 · 1 · 1 = 1,
det(E ) det(A) = det(A) = det(B) = det(EA).
Elementary row operations (Theorem 2.5.15)

Let A = (ai j ) be an n × n matrix.

B1 det(B1) = k det(A)
kRi
A B2 det(B2) = −det(A)
Ri ↔ Rj
Rj + kRi B3 det(B3) = det(A)

Furthermore, if E is an elementary matrix of the same

size as A, then det(EA) = det(E ) det(A).
Proof of det(B3) = det(A) (Theorem 2.5.15)

To prove det(B3) = det(A):

Since B3 is obtained from A by adding k times of the
i th row of A to j th row,
a11 a12 ··· a1n
⁞ ⁞ ⁞
aj −1, 1 aj −1, 2 ··· aj −1, n
B3 = aj1 + kai1 aj 2 + kai 2 ··· aj n + kai n .
aj +1, 1 aj +1, 2 ··· aj +1, n
⁞ ⁞ ⁞
am1 am2 ··· amn
 a11 a12 ··· a1n a11 a12 ··· a1n
⁞ ⁞ ⁞ ⁞ ⁞ ⁞
aj −1,1 aj −1,2 ··· aj −1,n aj −1, 1 aj −1, 2 ··· aj −1, n
A= aj1 aj 2 ··· aj n B3 = aj1 + kai1 aj 2 + kai 2 ··· aj n + kai n
aj +1,1 aj +1,2 ··· aj +1,n aj +1, 1 aj +1, 2 ··· aj +1, n
⁞ ⁞ ⁞ ⁞ ⁞ ⁞
am1 am2 ··· amn am1 am2 ··· amn

After deleting the j th row, the matrices on both sides look exactly the same.

a11 ··· a1t −1 a1t +1 ··· a1n

⁞ ⁞ ⁞ ⁞
aj −1,1 ··· aj −1,t −1 aj −1,t +1 ··· aj −1,n
The (j, t)-cofactor of B3 = (−1)j +t aj +1,1 ··· aj +1,t −1 aj +1,t +1 ··· aj +1,n
⁞ ⁞ ⁞ ⁞
am1 ··· am t −1 am t +1 ··· amn

= the (j, t)-cofactor of A = Ajt.

 Proof of det(B3) = det(A) (Theorem 2.5.15)

a11 a12 ··· a1n

⁞ ⁞ ⁞
aj −1, 1 aj −1, 2 ··· aj −1, n
det(B3) = aj1 + kai1 aj 2 + kai 2 ··· aj n + kai n
aj +1, 1 aj +1, 2 ··· aj +1, n
⁞ ⁞ ⁞ cofactor expansion
am1 am2 ··· amn along the j th row

= (aj1 + kai1)Aj1 + (aj 2 + kai 2)Aj 2 + ··· + (aj n + kai n)Aj n

= aj1Aj1 + aj 2Aj 2 + ··· + aj n Aj n + k(ai1Aj1 + ai 2Aj 2 + ··· + ai n Aj n)
= det(A) + k(ai1Aj1 + ai 2Aj 2 + ··· + ai n Aj n).
 cofactor expansion
ai1Aj1 + ai 2Aj 2 + ··· + ai n Aj n = ? along the j th row
Compare with det(A) = aj1Aj1 + aj 2Aj 2 + ··· + aj n Aj n.
a11 a12 ··· a1n a11 a12 ··· a1n
⁞ ⁞ ⁞ ⁞ ⁞ ⁞
aj −1,1 aj −1,2 ··· aj −1,n aj −1,1 aj −1,2 ··· aj −1,n
A= aj1 aj 2 ··· aj n C= ai1 ai 2 ··· ai n
aj +1,1 aj +1,2 ··· aj +1,n aj +1,1 aj +1,2 ··· aj +1,n
⁞ ⁞ ⁞ ⁞ ⁞ ⁞
am1 am2 ··· amn am1 am2 ··· amn

a11 ··· a1t −1 a1t +1 ··· a1n

⁞ ⁞ ⁞ ⁞
aj −1,1 ··· aj −1,t −1 aj −1,t +1 ··· aj −1,n
The (j, t)-cofactor of C = (−1)j +t aj +1,1 ··· aj +1,t −1 aj +1,t +1 ··· aj +1,n
⁞ ⁞ ⁞ ⁞
am1 ··· am t −1 am t +1 ··· amn
= the (j, t)-cofactor of A = Ajt.
 Proof of det(B3) = det(A) (Theorem 2.5.15)

Now, we compute det(C) in two different ways:

There exists a11 a12 ··· a1n
two identical ⁞ ⁞ ⁞
rows. ai1 ai 2 ··· ai n
⁞ ⁞ ⁞
0 = aj −1,1 aj −1,2 ··· aj −1,n = ai1Aj1 + ai 2Aj 2 + ··· + ai n Aj n.
ai1 ai 2 ··· ai n
aj +1,1 aj +1,2 ··· aj +1,n cofactor expansion
⁞ ⁞ ⁞ along the j th row
am1 am2 ··· amn

So det(B3) = det(A) + k(ai1Aj1 + ai 2Aj 2 + ··· + ai n Aj n)

= det(A).
Examples (Remark 2.5.16 & Example 2.5.17.1)

Given a square matrix A, we can use elementary row

operations to transform a square matrix to a triangular
matrix (e.g. using the Gaussian Elimination) and then
compute the determinant accordingly.
3 −1 1 1 3 −1 1 1 3 −1 1 1
R2 − R 1 R2 ↔ R3
3 −1 2 1 0 0 1 0 0 2 4 0
= = −
0 2 4 0 0 2 4 0 0 0 1 0
0 0 2 −1 0 0 2 −1 0 0 2 −1
3 −1 1 1
R4 − 2R3
0 2 4 0
= − = −3 · 2 · 1 · (−1) = 6.
0 0 1 0
0 0 0 −1
Examples (Remark 2.5.17.2)

5 0 8 −1
2
R3 + 9 R1 R2 ↔ R3 4R2 0 −2 1 0
A B=
0 0 1 9
1
0 0 0 3
Find det(A).
1
5 · (−2) · 1 · 3 = det(B) = 1 · (−1) · 4 det(A)

10 5
Thus − 3 = −4 det(A) and hence det(A) = 6 .
Elementary column operations (Remark 2.5.18)

Instead of performing elementary operations on rows of a

matrix, we can also perform these operations on columns.
Similar to the previous theorem (Theorem 2.5.15), one can
use elementary column operations to compute the
determinants of square matrices.
(Please read our textbook for the details.)
 Invertible matrices (Theorem 2.5.19)

A square matrix A is invertible if and only if det(A) ≠ 0.

Proof: Let B be the reduced row-echelon form of A,

i.e. Gauss-Jordan
A B = Ek ··· E2E1A.
Elimination
for some elementary matrices E1, E2, …, Ek.
Then det(B) = det(Ek ··· E2E1A)
= det(Ek) ··· det(E2) det(E1) det(A).
by Theorem 2.5.15.4
If A is invertible, then (by Theorem 2.4.7) B = I
and 1 = det(I ) = det(B) = det(Ek) ··· det(E2) det(E1) det(A)
and hence det(A) ≠ 0.
Invertible matrices (Theorem 2.5.19)

Suppose A is singular.
Then (by Remark 2.4.10) B has at least one zero row,
i.e.
* * ··· *
0 * ··· *
B= .
⁞ ⁞
0 ··· 0 0 zero row

So 0 = det(B) = det(Ek) ··· det(E2) det(E1) det(A).

Since det(Ei) ≠ 0 for all i, det(A) = 0.
Scalar multiplication (Theorem 2.5.22.1)

If A is a square matrix of order n and c a scalar, then

det(cA) = cn det(A).

Proof:
cR1 cR2 cRn
A ··· cA.

Thus det(cA) = c · c ··· c det(A) = cn det(A).

n times
Matrix multiplication (Theorem 2.5.22.2)

If A and B are square matrices of the same size, then

det(AB) = det(A) det(B).

Proof: We have to consider the cases that A is singular

and A is invertible separately.
Case 1: A is singular.
We learnt that (by Theorem 2.4.14) AB is also singular.
Then
det(A) = 0
det(B) = 0 · det(B) = 0 = det(AB).
Matrix multiplication (Theorem 2.5.22.2)

Case 2: A is invertible.
Then (by Theorem 2.4.7) A = E1E2 ··· Ek for some
elementary matrices E1, E2, …, Ek.
Thus
det(AB) = det(E1E2 ··· EkB)
= det(E1) det(E2) ··· det(Ek) det(B)
= det(E1E2 ··· Ek) det(B)
= det(A) det(B).
Invertible matrices (Theorem 2.5.22.3)

If A is an invertible matrix, then

1
det(A−1) = det(A)
.

Proof: Since AA−1 = I,

det(A) det(A−1) = det(AA−1) = det(I ) = 1.
1
So det(A−1) = det(A) .
Examples (Example 2.5.23)

−3 −2 4
Let A = 4 3 1 . Note that det(A) = 34.
0 2 4

1. det(2A) = 23 det(A) = 23 · 34 = 272.

1 0 1
2. Let B = 0 −1 0 . Note that det(B) = −1.
1 0 2
Then det(AB) = det(A) det(B) = 34 · (−1) = −34.
1 1
3. det(A−1) = det(A)
= 34
.
Adjoints (Definition 2.5.24)

Let A be a square matrix of order n.

The (classical) adjoint of A is the n × n matrix
T
A11 A12 ··· A1n A11 A21 ··· An1
A21 A22 ··· A2n A12 A22 ··· An2
adj(A) = =
⁞ ⁞ ⁞ ⁞ ⁞ ⁞
An1 An2 ··· Ann A1n A2n ··· Ann

where Ai j which is the (i, j )-cofactor of A.

Adjoints (Theorem 2.5.25)

1
If A is an invertible matrix, then A−1 = det(A) adj(A).

1
Proof: It suffices to show A det(A) adj(A) = I.
1
Then (by Theorem 2.4.12) A−1 = det(A) adj(A).
Let A = (ai j )n × n and A [ adj(A) ] = (bi j )n × n .
As a11 a12 ··· a1n A11 A21 ··· An1
a21 a22 ··· a2n A12 A22 ··· An2
A [ adj(A) ] = ,
⁞ ⁞ ⁞ ⁞ ⁞ ⁞
an1 an2 ··· ann A1n A2n ··· Ann
bi j = ai1Aj1 + ai 2Aj 2 + ··· + ai n Aj n.
 Adjoints (Theorem 2.5.25)

A [ adj(A) ] = (bi j )n × n
where bi j = ai1Aj1 + ai 2Aj 2 + ··· + ai n Aj n.

Note that bi i = ai1Ai1 + ai 2Ai 2 + ··· + ai n Ai n = det(A).

For i ≠ j, a11 a12 ··· a1n cofactor expansion

⁞ ⁞ ⁞ along the i th row of A
ai1 ai 2 ··· ai n
⁞ ⁞ ⁞
0 = aj −1,1 aj −1,2 ··· aj −1,n = ai1Aj1 + ai 2Aj 2 + ··· + ai n Aj n.
ai1 ai 2 ··· ai n = bi j
There exists aj +1,1 aj +1,2 ··· aj +1,n
two identical ⁞ ⁞ ⁞ cofactor expansion
rows. am1 am2 ··· amn along the j th row
Adjoints (Theorem 2.5.25)

A [ adj(A) ] = (bi j )n × n
where bi j = ai1Aj1 + ai 2Aj 2 + ··· + ai n Aj n.

Thus A [ adj(A) ] = (bi j )n × n

det(A) 0 ··· 0
0 det(A) ⁞
=
⁞ 0
0 ··· 0 det(A)

= det(A) I.
1 1
So det(A) A [ adj(A) ] = I ⇒ A det(A) adj(A) = I.
Examples (Example 2.5.26.1)

a b
Let A = with ad − bc ≠ 0.
c d

Then M11 = d , M12 = c , M21 = b , M22 = a ,

A11 = (−1)1 + 1 det(M11) = d, A12 = (−1)1 + 2 det(M12) = −c,

A21 = (−1)2 + 1 det(M21) = −b, A22 = (−1)2 + 2 det(M22) = a,
T T
A11 A12 d −c d −b
So adj(A) = = =
A21 A22 −b a −c a

1 1 d −b
and A−1 = adj(A) = ad − bc
.
det(A) −c a
Examples (Example 2.5.26.2)

1 −1 1 −3 3 1
Let B = 0 −1 0 . B−1 = 1 0 2 0 .
det(B)
−2
1 0 3 1 −1 −1
T
−1 0 0 0 0 −1
−
0 3 1 3 1 0 T
−3 03 11
−1 1 1 1 1 −1
adj(B) = − − = 3
0 22 −10 .
0 3 1 3 1 0
1 0−1−1
−1
−1 1 1 1 1 −1
−
−1 0 0 0 0 −1
Cramer’s Rule (Theorem 2.5.27)

Suppose Ax = b is a linear system

x1 b1
x2 b2
where A = (ai j )n × n, x = and b = .
⁞ ⁞
xn bn
Let Ai be the n × n matrix obtained from A by
replacing the i th column of A by b,
a11 ··· a1, i −1 b1 a1, i +1 ··· a1n
a21 ··· a2, i −1 b2 a2, i +1 ··· a2n
i.e. Ai = .
⁞ ⁞ ⁞ ⁞
an1 ··· an, i −1 bn an, i +1 ··· ann
 Cramer’s Rule (Theorem 2.5.27)

If A is invertible, then the system has only one solution

det(A1)
1 det(A2)
x = det(A) .
⁞
det(An)
1
Proof: Ax = b ⇔ x = A−1b = det(A) adj(A) b.

A11 A21 ··· An1 b1 b1 A11 + b2A21 + ··· + bn An1

A12 A22 ··· An2 b2 b1 A12 + b2A22 + ··· + bn An2
adj(A) b = =
⁞ ⁞ ⁞ ⁞ ⁞
A1n A2n ··· Ann bn b1 A1n + b2A2n + ··· + bn Ann
Cramer’s Rule (Theorem 2.5.27)

So the solution to the system is

b1 A11 + b2A21 + ··· + bn An1
1 b1 A12 + b2A22 + ··· + bn An2
x=
det(A) ⁞
b1 A1n + b2A2n + ··· + bn Ann
where for i = 1, 2, ..., n,
a11 ··· a1, i −1 b1 a1, i +1 ··· a1n
a21 ··· a2, i −1 b2 a2, i +1 ··· a2n
det(Ai) =
⁞ ⁞ ⁞ ⁞
an1 ··· an, i −1 bn an, i +1 ··· ann
cofactor expansion
along the i th column = b1 A1i + b2A2i + ··· + bn An i .
An example (Example 2.5.28)

Consider the linear system

x + y + 3z = 0
2x − 2y + 2z = 4
3x + 9y = 3.

First, we rewrite the system as

1 1 3 x 0
2 −2 2 y = 4 .
3 9 0 z 3
An example (Example 2.5.28)

Then by Cramer’s rule,

0 1 3
4 −2 2
3 9 0
x= = 2.2. 1 0 3
1 1 3 2 4 2
2 −2 2 3 3 0
3 9 0 y= = −0.4. 1 1 0
1 1 3 2 −2 4
2 −2 2 3 9 3
3 9 0 z= = −0.6.
1 1 3
2 −2 2
3 9 0