Lecture 3 : Matrices

ENSIA, May 2022

 Contents
• Definitions
• Matrix operations
• Vector space of matrices
• Ring of square matrices
• Invertible matrices, Computing the inverse using RREF
• Matrix transpose
• Matrix trace
• Matrix of a linear map
• Matrix rank
• Change of basis
• Determinants
 Definition

Throughout this lecture, 𝐾 will denote an arbitrary field (usually ℝ or ℂ).

Definition 1
A matrix 𝐴 is an 𝒎 × 𝒏 array of scalars from a given field 𝐾 of the form
𝑎11 𝑎12 ⋯ 𝑎1𝑛
𝑎21 𝑎22 ⋯ 𝑎2𝑛
𝐴= ⋮ ⋮ ⋮ ⋮ .
𝑎𝑚1 𝑎𝑚2 ⋯ 𝑎𝑚𝑛
The individual values 𝑎𝑖𝑗 ∈ 𝐾 in the matrix are called entries.
The size of the array is 𝒎 × 𝒏, where 𝑚 is the number of rows and 𝑛 is the
number of columns.
 Row matrix and column matrix
Definition 2
A row matrix of size 1 × 𝑛 is a matrix of the form 𝑎11 ⋯ 𝑎1𝑛
𝑎11
A column matrix of size 𝑚 × 1 is a matrix of the form ⋮
𝑎𝑚1
Example 1
1 0 3 is a row matrix of size 1 × 3 ;
1
is a column matrix of size 2 × 1.
−1
 Notation
Notation
A matrix 𝐴 of size 𝑚 × 𝑛 is denoted by 𝐴 = 𝑎𝑖𝑗 1≤𝑖≤𝑚 or 𝑎𝑖𝑗 when we
1≤𝑗≤𝑛
know the number of rows and columns.
Example 2
𝐾=ℝ
1 2 0
𝐴=
3 −1 4
is a matrix of size 2 × 3.
1 −1
𝐵= .
0 2
is a matrix of size 2 × 2.
 Square matrices

If 𝑚 = 𝑛, the matrix is called square. In this case we have

a) A matrix 𝐴 is said to be diagonal if 𝑎𝑖𝑗 = 0 for 𝑖 ≠ 𝑗.
Example 3
1 0 0
0 0
0 −2 0 ; .
0 1
0 0 1
 Special matrices
b) A diagonal matrix 𝐴 may be denoted by 𝑑𝑖𝑎𝑔 𝑑1 , 𝑑2 , … , 𝑑𝑛 ,
where 𝑎𝑖𝑖 = 𝑑𝑖 and 𝑎𝑖𝑗 = 0 for 𝑖 ≠ 𝑗.
The diagonal matrix 𝑑𝑖𝑎𝑔(1, 1, … , 1) is called the identity matrix and
is denoted by 𝐼𝑛 . We have then
1 0 ⋯ 0
0 1 ⋯ 0
𝐼𝑛 = .
⋮ ⋮ ⋮ ⋮
0 0 ⋯ 1
0 = 𝑑𝑖𝑎𝑔(0, … , 0) is called the zero matrix.
 Triangular matrices
c) A square matrix 𝐴 is said to be lower triangular if 𝑎𝑖𝑗 = 0 for 𝑖 < 𝑗.
Example 4
0 0 0
2 1 0
3 −1 0
d) A square matrix A is said to be upper triangular if 𝑎𝑖𝑗 = 0 for 𝑖 > 𝑗.
Example 5
0 1 −1
0 1 3
0 0 4
 Symmetric matrices
e) A square matrix A is called symmetric if 𝑎𝑖𝑗 = 𝑎𝑗𝑖 .
Example 6
0 1 3
1 1 −1 .
3 −1 0
f) 𝐸𝑖𝑗 is the matrix with 1 in the (𝑖, 𝑗) position and zeros in all other
positions.
Example 7
0 0 0
0 0
𝐸23 = 0 0 1 ; 𝐸21 = .
1 0
0 0 0
 Notation

Notation
The set of all 𝑚 × 𝑛 matrices is denoted by ℳ𝑚,𝑛 𝐾 , where 𝐾 is the
underlying field. In the case where 𝑚 = 𝑛, we write ℳ𝑛 𝐾 to denote
the matrices of size 𝑛 × 𝑛.
 Matrix operations

Equality

Two matrices 𝐴 = 𝑎𝑖𝑗 1≤𝑖≤𝑚 , 𝐵 = 𝑏𝑖𝑗 1≤𝑖≤𝑚 are equal if, and only if,
1≤𝑗≤𝑛 1≤𝑗≤𝑛
𝑎𝑖𝑗 = 𝑏𝑖𝑗 for all 𝑖 ∈ 1,2, ⋯ , 𝑚 and 𝑗 ∈ 1,2, ⋯ , 𝑛 .
 Matrix operations
Sum of matrices
Let 𝐴 = 𝑎𝑖𝑗 1≤𝑖≤𝑚 , 𝐵 = 𝑏𝑖𝑗 1≤𝑖≤𝑚 ∈ ℳ𝑚,𝑛 𝐾 . Then
1≤𝑗≤𝑛 1≤𝑗≤𝑛

𝐴 + 𝐵 = 𝑎𝑖𝑗 + 𝑏𝑖𝑗
𝑖𝑗
Example 8
1 −1 0 1 0 3 2 −1 3
3 1 2 + −1 −2 5 = 2 −1 7
0 4 −2 3 0 −1 3 4 −3
 Matrix operations

Multiplication by a scalar
Let 𝐴 = (𝑎𝑖𝑗 )1≤𝑖≤𝑚 ∈ ℳ𝑚,𝑛 𝐾 and 𝛼 ∈ 𝐾. Then
1≤𝑗≤𝑛
𝛼𝐴 = 𝛼𝑎𝑖𝑗
𝑖𝑗
Example 9
1 −1 0 −2 2 0
−2 3 1 2 = −6 −2 −4
0 4 −2 0 −8 4
 Matrix operations
Product of two matrices
Let 𝐴 = 𝑎𝑖𝑗 1≤𝑖≤𝑚 ∈ ℳ𝑚,𝑛 𝐾 and 𝐵 = 𝑏𝑖𝑗 1≤𝑖≤𝑛 ∈ ℳ𝑛,𝑟 𝐾 .
1≤𝑗≤𝑛 1≤𝑗≤𝑟

The product 𝐴𝐵 is defined as follows :

𝐴𝐵 = 𝑐𝑖𝑗 1≤𝑖≤𝑚 ∈ ℳ𝑚,𝑟 𝐾 ,
1≤𝑗≤𝑟
where
𝑘=𝑛

𝑐𝑖𝑗 = 𝑎𝑖𝑘 𝑏𝑘𝑗 .

𝑘=1
 Product of matrices
Example 10
1 −1
2 3 1
Set 𝐴 = ∈ ℳ2,3 (ℝ) and 𝐵 = 2 1 ∈ ℳ3,2 (ℝ)
−3 0 −1
1 3

𝐴𝐵
2×1+3×2+1×1 2 × −1 + 3 × 1 + 1 × 3
=
−3 × 1 + 0 × 2 + (−1) × 1 −3 × −1 + 0 × 1 + (−1) × 3

9 4
𝐴𝐵 =
−4 0
 Product of matrices

Proposition 1
The product of matrices is associative.
 Product of matrices
The product of two matrices is not commutative.
Example 11
1 −1 −1 2
Set 𝐴 = ;𝐵 =
2 3 1 4
−2 −2
𝐴𝐵 =
1 16
and
3 7
𝐵𝐴 = .
9 11
We have 𝐴𝐵 ≠ 𝐵𝐴.
 Vector space of matrices

Theorem 1
ℳ𝑚,𝑛 𝐾 endowed with the addition and the multiplication by a scalar
is a 𝐾- vector space.
Theorem 2
The vector space ℳ𝑚,𝑛 𝐾 has dimension 𝑚𝑛.
The matrices 𝐸𝑖𝑗 , 1 ≤ 𝑖 ≤ 𝑚, 1 ≤ 𝑗 ≤ 𝑛, form a basis of ℳ𝑚,𝑛 𝐾 .
 Ring of square matrices

Theorem 3
ℳ𝑛 𝐾 , +,∙ is a non commutative ring.
Remark 1
ℳ𝑛 𝐾 , +,∙ has zero divisors.
Example 12
0 −1 2 −3
Set 𝐴 = and 𝐵 = .
0 5 0 0
0 0 0 0 0 0
We have 𝐴 × 𝐵 = with 𝐴 ≠ and 𝐵 ≠ .
0 0 0 0 0 0
 Invertible Matrices

Definition 4
Let 𝐴 ∈ ℳ𝑛 𝐾 . The square matrix 𝐴 is said to be invertible if there
exists a matrix 𝐵 ∈ ℳ𝑛 𝐾 such that 𝐴𝐵 = 𝐵𝐴 = 𝐼𝑛 .
In this case 𝐵 is called the inverse matrix of 𝐴.
If 𝐴 is invertible, 𝐵 is denoted by 𝐴−1 .
The set of invertibles matrices of size 𝑛 × 𝑛 is denoted by 𝐺𝐿(𝑛, 𝐾).
 Invertible Matrices

Example 13
1 −1
The inverse of 𝐴 = is
2 1
1 1
3 3 1 1 1
𝐵= −2 1 = .
3 −2 1
3 3
We have 𝐴𝐵 = 𝐵𝐴 = 𝐼2 .
 Invertible Matrices

Proposition 2
Let 𝐴, 𝐵 ∈ ℳ𝑛 𝐾 two invertible matrices. Then 𝐴𝐵 is invertible and
(𝐴𝐵)−1 = 𝐵−1 𝐴−1 .
Proposition 3
Let 𝐴, 𝐵 ∈ ℳ𝑛 𝐾 . If 𝐴𝐵 = 𝐼𝑛 , then 𝐵𝐴 = 𝐼𝑛 .
 Matrix power
Let 𝐴 be a matrix of size 𝑛 × 𝑛 and 𝑚 ∈ ℕ.
We define
𝐴𝑚 = 𝐴 ∙ 𝐴 ∙ ⋯ ∙ 𝐴 (𝑚 times),
and
𝐴0 = 𝐼𝑛 .
If 𝐴 is invertible, we define
𝐴−𝑚 = (𝐴−1 )𝑚 = 𝐴−1 ∙ 𝐴−1 ∙ ⋯ ∙ 𝐴−1 (𝑚 times).
 Matrix power

Theorem 4
Let 𝐴 be an invertible matrix. Then
1. 𝐴−1 is invertible and 𝐴−1 −1 = 𝐴
2. 𝐴𝑚 is invertible and 𝐴𝑚 −1 = 𝐴−𝑚 .
−1 1 −1
3. 𝑘𝐴 is invertible if 𝑘 ≠ 0 and (𝑘𝐴) = 𝐴 .
𝑘
 Row Reduced Echelon Form (RREF)
Definition 9
We say that a matrix in Row Echelon Form (REF) is in Row Reduced
Echelon Form (RREF) if it satisfies the following additional conditions :
1. All the pivots are 1.
2. Each pivot is the only nonzero entry in its column.
Theorem 5
The Row Reduced Echelon Form of a matrix is unique.
 Examples
Example 14
The matrix
1 0 1 2
0 1 3 4
0 0 0 0
is in RREF, while the matrix
1 0 0 0
0 1 2 0
0 0 1 0
Is not in RREF.
 Computing the inverse of a matrix using RREF

Gauss method for inverting a matrix

Let 𝐴 be an invertible square matrix of size 𝑛. To determine the inverse
of 𝐴, we use the Gauss method putting the matrix 𝐴 in the Row
Reduced Echelon Form, which is 𝐼𝑛 . We then do the same operations in
the same order on 𝐼𝑛 , and we end up with the inverse of A.
In practice, we do both operations at the same time, as we will see in
the next example.
 Computing the inverse of a matrix using RREF
Example 15

𝑅2 : = 𝑅2 − 4𝑅1 ; 𝑅3 : = 𝑅3 + 𝑅1

1 2 1 ⋮ 1 0 0 1 2 1 ⋮ 1 0 0
4 0 −1 ⋮ 0 1 0 ⟶ 0 −8 −5 ⋮ −4 1 0
−1 2 2 ⋮ 0 0 1 0 4 3 ⋮ 1 0 1
Computing the inverse of a matrix using RREF

−1
𝑅2 ≔ 𝑅2
8
1 2 1 ⋮ 1 0 0
1 2 1 ⋮ 1 0 0 5 1 −1
0 −8 −5 ⋮ −4 1 0 → 0 1 ⋮ 0
0 4 3 ⋮ 1 0 1 8 2 8
0 4 3 ⋮ 1 0 1
Computing the inverse of a matrix using RREF

𝑅3 : = 𝑅3 − 4𝑅2

1 2 1 ⋮ 1 0 0
1 2 1 ⋮ 1 0 0 5 1 −1
5 1 −1 0 1 ⋮ 0
0 1 ⋮ 0 ⟶ 8 2 8
8 2 8 1 1
0 4 3 ⋮ 1 0 1 0 0 ⋮ −1 1
2 2
Computing the inverse of a matrix using RREF

𝑅3 : = 2𝑅3

1 2 1 ⋮ 1 0 0
5 1 −1 1 2 1 ⋮ 1 0 0
0 1 ⋮ 0 5 1 −1
8 2 8 ⟶ 0 1 ⋮ 0
1 1 8 2 8
0 0 ⋮ −1 1 0 0 1 ⋮ −2 1 2
2 2
Computing the inverse of a matrix using RREF
5
𝑅2 : = 𝑅2 − 𝑅3 ; 𝑅1 ≔ 𝑅1 − 𝑅3
8
1 2 1 ⋮ 1 0 0 1 2 0 ⋮ 3 −1 −2
5 1 −1 7 −3 −5
0 1 ⋮ 0 ⟶ 0 1 0 ⋮
8 2 8 4 4 4
0 0 1 ⋮ −2 1 2 0 0 1 ⋮ −2 1 2
 Computing the inverse of a matrix using RREF
𝑅1 ≔ 𝑅1 − 2𝑅2

−1 1 1
1 2 0 ⋮ 3 −1 −2 1 0 0 ⋮
7 −3 −5 2 2 2
0 1 0 ⋮ ⟶ 0 1 0 ⋮
7 −3 −5 .
4 4 4
4 4 4
0 0 1 ⋮ −2 1 2 0 0 1 ⋮ −2 1 2
Then the inverse of 𝐴 is given by

1 −2 2 2
𝐴−1 = 7 −3 −5 .
4
−8 4 8
 Matrix transpose
Let 𝐴 = 𝑎𝑖𝑗 1≤𝑖≤𝑚 ∈ ℳ𝑚,𝑛 𝐾 .
1≤𝑗≤𝑛
Definition 5
The transpose of 𝐴, denoted by 𝐴𝑇 , is a matrix of size 𝑛 × 𝑚 defined by
𝐴𝑇 = 𝑎′𝑖𝑗 1≤𝑖≤𝑛 , 𝑎′𝑖𝑗 = 𝑎𝑗𝑖
1≤𝑗≤𝑚
Example 16
1 2
1 −1 0
𝐴 = −1 3 ; 𝐴𝑇 =
2 3 −2
0 −2
 Matrix transpose
Proposition 4
Let 𝐴, 𝐵 ∈ ℳ𝑚,𝑛 𝐾 .
1) (𝐴 + 𝐵)𝑇 = 𝐴𝑇 + 𝐵𝑇 .
2) (𝑘𝐴)𝑇 = 𝑘(𝐴)𝑇 .
3) (𝐴𝑇 )𝑇 = 𝐴.
4) If 𝐴 ∈ ℳ𝑚,𝑛 𝐾 and 𝐵 ∈ ℳ𝑛,𝑟 𝐾 , then (𝐴𝐵)𝑇 = (𝐵)𝑇 (𝐴)𝑇
5) If 𝑚 = 𝑛 and 𝐴 invertible, then (𝐴𝑇 )−1 = (𝐴−1 )𝑇 .
 Matrix Trace
Let
𝑎11 𝑎12 ⋯ 𝑎1𝑛
𝑎21 𝑎22 ⋯ 𝑎2𝑛
𝐴= ⋮ ⋮ ⋮ ⋮
𝑎𝑛1 𝑎𝑛1 ⋯ 𝑎𝑛𝑛
be a square matrix of size 𝑛 × 𝑛.
Definition 6
The Trace of 𝐴 denoted by 𝑇𝑟(𝐴) is defined as follows :
𝑛

𝑇𝑟 𝐴 = 𝑎𝑖𝑖
𝑖=1
 Matrix Trace
Example 17
Let
1 −1 0
𝐴= 3 1 2 ,
0 4 −2
then

𝑇𝑟 𝐴 = 1 + 1 + −2 = 0
 Matrix Trace Properties

Proposition 5
Let 𝐴, 𝐵 ∈ ℳ𝑛 𝐾 .
1) 𝑇𝑟 𝐴 + 𝐵 = 𝑇𝑟 𝐴 + 𝑇𝑟 𝐵 .
2) 𝑇𝑟 𝛼𝐴 = 𝛼𝑇𝑟 𝐴 , ∀𝛼 ∈ 𝐾.
3) 𝑇𝑟 𝐴𝑇 = 𝑇𝑟 𝐴 .
4) 𝑇𝑟 𝐴𝐵 = 𝑇𝑟 𝐵𝐴 .
 Matrix of a linear map
Let 𝑉 and 𝑊 be finite dimensional 𝐾- vector spaces and let 𝑓: 𝑉 ⟶ 𝑊
be a linear map.
Let 𝐵 = 𝑣1 , 𝑣2 , ⋯ , 𝑣𝑛 a basis of 𝑉 and 𝐵′ = 𝑤1 , 𝑤2 , ⋯ , 𝑤𝑚 a basis
of 𝑊.
Each vector 𝑓(𝑣𝑖 ), 𝑖 ∈ 1,2, ⋯ , 𝑛 , can be written with respect to the
basis 𝑤1 , 𝑤2 , ⋯ , 𝑤𝑚 of 𝑊 as follows :
𝑓(𝑣1 )=𝑎11 𝑤1 + 𝑎21 𝑤2 + ⋯ + 𝑎𝑚1 𝑤𝑚
𝑓(𝑣2 )=𝑎12 𝑤1 + 𝑎22 𝑤2 + ⋯ + 𝑎𝑚2 𝑤𝑚
⋮
𝑓 𝑣𝑛 = 𝑎1𝑛 𝑤1 + 𝑎2𝑛 𝑤2 + ⋯ + 𝑎𝑚𝑛 𝑤𝑚
 Matrix of a linear map

Definition 7
The matrix
𝑎11 𝑎12 ⋯ 𝑎1𝑛
′
𝑎21 𝑎22 ⋯ 𝑎2𝑛
ℳ(𝑓, 𝐵, 𝐵 ) = ⋮ ⋮ ⋮ ⋮ .
𝑎𝑚1 𝑎𝑚1 ⋯ 𝑎𝑚𝑛
is called the matrix representing the linear map 𝑓 with respect to the
bases 𝐵 and 𝐵’ of 𝑉 and 𝑊 respectively.
 Matrix of a linear map
Example 18
Let 𝑓: ℝ2 ⟶ ℝ3 be the linear map defined by
𝑓 𝑥, 𝑦 = 𝑥 − 𝑦, 2𝑥 + 𝑦, 𝑥 + 3𝑦 .
We have
𝑓 1,0 = 1,2,1 = 𝑒′1 + 2𝑒′2 + 𝑒′3 ;
𝑓 0,1 = −1,1,3 = −𝑒 ′1 + 𝑒 ′ 2 + 3𝑒′3 .
The matrix representing 𝑓 with respect to the standard bases 𝐵 = 𝑒1 , 𝑒2
and 𝐵′ = 𝑒′1 , 𝑒′2 ,𝑒′3 of ℝ2 and ℝ3 respectively is :

1 −1
2 1 = ℳ(𝑓, 𝐵, 𝐵′ )
1 3
 Matrix of the identity map

Remark 2
Let 𝑉 be a 𝐾-vector space of dimension 𝑛 and 𝐼𝑑𝑉 : 𝑉 ⟶ 𝑉 the identity
map. Then the matrix representing 𝐼𝑑𝑉 with respect to a basis 𝐵 of 𝑉 is
the identity matrix 𝐼𝑛 .
 Composed linear maps matrix
Proposition 6
Let 𝑈, 𝑉 and 𝑊 be finite-dimensional vector spaces over 𝐾 with bases
𝐵1 = 𝑢1 , ⋯ , 𝑢𝑛 , 𝐵2 = 𝑣1 , ⋯ , 𝑣𝑚 and 𝐵3 = 𝑤1 , ⋯ , 𝑤𝑟
respectively. Let 𝑓: 𝑈 → 𝑉 and 𝑔 ∶ 𝑉 → 𝑊 be two linear maps, 𝐴
the matrix representing 𝑓 with respect to 𝐵1 and 𝐵2 , and 𝐵 the matrix
representing 𝑔 with respect to 𝐵2 and 𝐵3 . Then 𝐵𝐴 is the matrix
representing the linear map 𝑔 ∘ 𝑓: 𝑈 → 𝑊 with respect to 𝐵1 and 𝐵3 .
 Composed linear maps matrix
Example 19
Let 𝑓: ℝ2 ⟶ ℝ3 , 𝑔: ℝ3 ⟶ ℝ2 be the linear maps defined by
𝑓 𝑥, 𝑦 = (𝑥 + 2𝑦, 𝑥 − 𝑦, 𝑥) and 𝑔 𝑥, 𝑦, 𝑧 = 𝑥 + 𝑦 + 𝑧, 𝑥 − 𝑦 − 𝑧 .
Let 𝐴, 𝐵, the matrices representing 𝑓 and 𝑔 with respect to the
standard bases of ℝ2 and ℝ3 . Then
1 2
1 1 1
𝐴 = 1 −1 and 𝐵 = .
1 −1 −1
1 0
2 2 3 1
The matrix representing 𝑔 ∘ 𝑓: ℝ ⟶ ℝ is 𝐵𝐴 =
−1 3
 The isomorphism between vector space of
matrices and vector space of linear maps
Theorem 6
Let 𝑉 and 𝑊 be finite-dimensional vector spaces over 𝐾 with (ordered) bases
𝐵 = 𝑣1 , 𝑣2 , ⋯ , 𝑣𝑛 and 𝐵′ = 𝑤1 , 𝑤2 , ⋯ , 𝑤𝑚 respectively. Then, there
exists an isomorphism 𝜑 from the 𝐾-vector space ℒ 𝑉, 𝑊 of linear maps to
ℳ𝑚,𝑛 𝐾 that sends a linear map 𝑓: 𝑉 ⟶ 𝑊 to the matrix representing 𝑓
with respect to 𝐵 and 𝐵′ .
 Example

Example 20
2 3 1
Let 𝐴 = the matrix representing the linear map
−3 0 −1
𝑓: ℝ3 ⟶ ℝ2 with respect to the standard bases 𝐵 = 𝑒1 , 𝑒2 ,𝑒3 and
𝐵′ = 𝑒′1 , 𝑒′2 of ℝ3 and ℝ2 respectively. Then

𝑥
𝑓 𝑥, 𝑦, 𝑧 = 𝐴 𝑦 = (2𝑥 + 3𝑦 + 𝑧, −3𝑥 − 𝑧)
𝑧
 Some properties

Theorem 7
A matrix 𝐴 ∈ ℳ𝑛 𝐾 is invertible if, and only if, the linear map
representing 𝐴 is bijective.
Proposition 7
Let 𝐴 ∈ ℳ𝑛 𝐾 . The matrix 𝐴 is invertible if, and only if, the column
vectors of 𝐴 are linearly independent.
 Matrix rank

Definition 8
The rank of a matrix 𝐴 ∈ ℳ𝑝,𝑞 𝐾 is the rank of the linear map
representing 𝐴. It is denoted by 𝑟 𝐴 .
Theorem 8
For any matrix 𝐴, we have 𝑟 𝐴 = 𝑟 𝐴𝑇 .
Proposition 8
Let 𝐴 ∈ ℳ𝑛 𝐾 . The matrix 𝐴 is invertible if, and only if, the rank of 𝐴
equals to 𝑛.
 Change of coordinates matrix

Let 𝑉 be a 𝐾-vector space of dimension 𝑛. Let 𝐵 = 𝑣1 , 𝑣2 , ⋯ , 𝑣𝑛 and

𝐵′ = 𝑣′1 , 𝑣′2 , ⋯ , 𝑣′𝑛 be two ordered bases of 𝑉 such that

𝑣′1 = 𝑎11 𝑣1 + 𝑎21 𝑣2 + ⋯ + 𝑎𝑛1 𝑣𝑛

𝑣′2 = 𝑎12 𝑣1 + 𝑎22 𝑣2 + ⋯ + 𝑎𝑛2 𝑣𝑛
⋮
𝑣′𝑛 = 𝑎1𝑛 𝑣1 + 𝑎2𝑛 𝑣2 + ⋯ + 𝑎𝑛𝑛 𝑣𝑛 .
 Change of coordinates matrix

Definition 10
The matrix
𝑎11 𝑎12 ⋯ 𝑎1𝑛
𝑎21 𝑎22 ⋯ 𝑎2𝑛
𝑃𝐵′ ⟶𝐵 = ⋮ ⋮ ⋮ ⋮
𝑎𝑛1 𝑎𝑛2 ⋯ 𝑎𝑛𝑛
changing 𝐵′ coordinates into 𝐵 coordinates is called the change of
coordinates matrix.
Proposition 9
The matrix 𝑃 is invertible.
 Change of basis

Theorem 9
Let 𝑉 and 𝑊 be 𝐾-vector spaces of dimension 𝑛. Let 𝑓: 𝑉 ⟶ 𝑊 be a linear
map. Let 𝐵 = 𝑣1 , 𝑣2 , ⋯ , 𝑣𝑛 , 𝐵′ = 𝑣′1 , 𝑣′2 , ⋯ , 𝑣′𝑛 be two ordered bases
of 𝑉 and 𝑅 = 𝑤1 , 𝑤2 , ⋯ , 𝑤𝑚 , 𝑅′ = 𝑤′1 , 𝑤′2 , ⋯ , 𝑤′𝑚 be two ordered
bases of 𝑊. Denote by 𝑃 the matrix changing 𝐵 coordinates into 𝐵′
coordinates and by 𝑄 the matrix changing 𝑅 coordinates into 𝑅′
coordinates. Then

ℳ 𝑓, 𝐵′ , 𝑅′ = 𝑄−1 ℳ 𝑓, 𝐵, 𝑅 𝑃.
 Determinant
Definition 11
Let 𝐴 = 𝑎𝑖𝑗 be a square matrix. Denote by 𝐴𝑖,𝑗 the submatrix
1≤𝑖,𝑗≤𝑛
of 𝐴 of size 𝑛 − 1 obtained by deleting row 𝑖 and column 𝑗. Then the
determinant of 𝐴, denoted by det(𝐴) or 𝐴 , is the scalar defined
recursively as :
• If 𝑛 = 1, det 𝐴 = 𝑎11 .
• For all 𝑛 ≥ 2, det 𝐴 = 𝑛𝑗=1(−1)1+𝑗 𝑎1𝑗 det 𝐴1,𝑗 .
Example 21
𝑎 𝑏
= 𝑎𝑑 − 𝑏𝑐.
𝑐 𝑑
 Minor and Cofactor
Definition 12
The scalar
det 𝐴𝑖,𝑗
is called a minor.
The scalar
(−1)𝑖+𝑗 det 𝐴𝑖,𝑗
is called a cofactor.
 Laplace expansion

Theorem 10
Expansion along row 𝑖 :
𝑛

det 𝐴 = (−1)𝑖+𝑗 𝑎𝑖𝑗 det 𝐴𝑖,𝑗 .

𝑗=1
Expansion along column 𝑗 :
𝑛

det 𝐴 = (−1)𝑖+𝑗 𝑎𝑖𝑗 det 𝐴𝑖,𝑗 .

𝑖=1
 Example
Let
1 −1 0
𝐴= 3 1 2 .
0 4 −2
Computing along the first row :
1 2 3 2 3 1
det 𝐴 = 1. − −1 + 0. = −10 − 6 = −16.
4 −2 0 −2 0 4
Computing along the second column :
3 2 1 0 1 0
det 𝐴 = −1 −1 + + −1 . 4
0 −2 0 −2 3 2
= −6 − 2 − 8 = −16.
 Numerical note
By today’s standards, a square matrix of size 25 is small. Yet it would be
impossible to calculate its determinant by cofactor expansion. In
general, a cofactor expansion requires more than 𝑛! multiplications,
and 25! is approximately 1,5 × 1025 .
If a computer performs one trillion multiplications per second, it would
have to run for more than 500 000 years to compute a determinant of
size 25 by this method. Fortunately, there are faster methods, as we’ll
soon discover.
 Properties of determinants

Theorem 11 (Row operations)

Let 𝐴 be a square matrix
a. If a multiple of one row of 𝐴 is added to another row to produce a matrix 𝐵,
then det 𝐵 = det(𝐴).
b. If two rows of 𝐴 are interchanged to produce 𝐵, then det 𝐵 = −det(𝐴).
c. If one row of 𝐴 is multiplied by 𝑘 to produce 𝐵, then det 𝐵 = 𝑘det(𝐴).
 Determinant of REF Matrix
Proposition 10
If 𝐴 is a triangular matrix, the det(𝐴) is the product of the entries of
the diagonal of 𝐴.
Corollary 1
Let 𝑈 be the REF of a square matrix 𝐴, and suppose that 𝑟 interchanges
have been performed. Then we have
(−1)𝑟 × product of pivots in 𝑈, when 𝐴 is invertible
det 𝐴 =
0 , when 𝐴 is not invertible.
 Numerical note

1. Most computer programs that compute det(𝐴) for a general matrix 𝐴 use
the method of corollary above.
2. It can be shown that evaluation of a determinant of size 𝑛 using row
operations requires about 2𝑛3 3 arithmetic operations. Any modern
microcomputer can calculate a determinant of size 25 in a fraction of a second,
since only about 10 000 operations are required.
 Example
Example 23
Let
1 −1 1
𝐴= 3 −3 2 .
2 4 −2
Then
1 −1 1 1 −1 1
det 𝐴 = 0 0 −1 = − 0 6 −4 = − 1 × 6 × −1 = 6.
0 6 −4 0 0 −1
 Determinant and Invertibility

Theorem 12
A square matrix 𝐴 is invertible if and only if det(𝐴) ≠ 0.
 Determinant and transpose

Theorem 13
If 𝐴 is a square matrix, then
det 𝐴𝑇 = det(𝐴).
 Determinants and matrix products

Theorem 14
If 𝐴 and 𝐵 are square matrices of size 𝑛, then
det 𝐴𝐵 = det 𝐴 det(𝐵).
Corollary 2
If 𝐴 is invertible, then
−1
1
det(𝐴 ) = .
det 𝐴
 An inverse formula

Definition 13
Let 𝐴 be an 𝑛 × 𝑛 matrix. The adjugate matrix of 𝐴, denoted by adj(𝐴), is the
transpose of the matrix of cofactors of 𝐴 ; that is,
𝑇
adj(𝐴)= (−1)𝑖+𝑗 det 𝐴𝑖,𝑗
1≤𝑖,𝑗≤𝑛
Theorem 15
Let 𝐴 be an invertible matrix. Then
−1
1
𝐴 = adj(A).
det 𝐴
 Example
Let
1 −1 0
𝐴= 3 1 2 .
0 4 −2
We have seen that det 𝐴 = −16. The cofactors
𝐶𝑖𝑗 = (−1)𝑖+𝑗 det 𝐴𝑖,𝑗
of 𝐴 are
1 2 3 2 3 1
𝐶11 = = −10, 𝐶12 = − = 6, 𝐶13 = = 12,
4 −2 0 −2 0 4
−1 0 1 0 1 −1
𝐶21 =− = −2, 𝐶22 = = −2, 𝐶23 = − = −4,
4 −2 0 −2 0 4
 Example

−1 0 1 0 1 −1
𝐶31 = = −2, 𝐶32 = − = −2, 𝐶33 = = 4.
1 2 3 2 3 1

Then the inverse of 𝐴 is

−1 −10 −2 −2
𝐴−1 = 6 −2 −2 .
16
12 −4 4