Chapter 4 - MATRIX ALGEBRA • A matrix is upper triangular if all its elements under the

diagonal are zero

4.1. Matrix Operations
• A matrix is lower triangular if all its elements over the
 
a11 a12 ... a1j ... a1n diagonal are zero
 a 
 21 a22 ... a2j ... a2n 
 .. .. ..  • The set of all possible matrices of dimension (m × n)
 
A=  whose entries are real numbers is refered to as Rm×n
 ai1 ai2 ... aij . . . ain 
 . .. .. 
 . 
• The set of all possible matrices of dimension (m × n)
am1 am2 . . . amj . . . amn whose entries are complex numbers is refered to as Cm×n
 
• The entry in the ith row and the jth column of a matrix   2 0 0 0  
A is refered to as (A)ij . 0 1 0   2 2
   0 0 0 0    3×2
 0 1 −1     7 1 ∈K
 0 0 −1 0 
EXAMPLE: In the matrix above, 0 0 −1 3 −3
0 0 0 4

Algebra 2021/2022 4-1 Algebra 2021/2022 4-3

• A zero matrix is a matrix, written 0, whose entries are all

zero. • OPERATIONS:
• A square matrix has the same number of rows than Only for matrices with the same dimensions:
columns. ◦ Equality. Two matrices are equal if and only if their
corresponding entries are equal.
• In general (m 6= n), matrices are rectangular.
     
3 −1
• The (main) diagonal of a matrix, or its diagonal entries, 6= 6=
1 0
are the entries
• A diagonal matrix has all its nondiagonal entries equal ◦ Addition. A matrix whose entries are the sum of the
to zero. corresponding entries of the matrices.
       
    1 0 0 0
0 1 0 0 0   0 −1 1 −1
     0 −1 0 0       
 1 0 −1   0 0     1 0  +  −1 0  =  
 0 0 0 0  2 0 −1 2
−1 0 0 0 0
0 0 0 1

Algebra 2021/2022 4-2 Algebra 2021/2022 4-4

  
x1
 
◦ Scalar Multiplication. A matrix whose entries are the If we write B = [ b1 b2 . . . bp ] and x =  .. , then:
corresponding entries of the matrix multiplied by the scalar. xp

   
0 −1 Bx = x1 b1 + x2 b2 + · · · + xp bp
   
2 1 0  =  
2 0 A(Bx) = A(x1 b1 + x2 b2 + · · · + xp bp)

• PROPERTIES: = A(x1 b1) + A(x2 b2) + · · · + A(xp bp)

Let A, B and C be matrices of Km×n and λ, µ ∈ K:
= x1 Ab1 + x2 Ab2 + · · · + xp Abp
◦ A+B =B+A ◦ λ (A+B) = λ A+λ B
= =
◦ A+(B +C) = (A+B)+C ◦ (λ+µ) A = λ A+µ A
◦ A+0=A ◦ λ (µ A) = (λ µ) A

Algebra 2021/2022 4-5 Algebra 2021/2022 4-7

Matrix Multiplication • Let A be an (m × n) matrix and let B be an (n × p)

matrix with columns b1, b2, . . . , bp. The matrix product
Kp Kn Km of A by B is the (m × p) matrix AB whose columns are
Ab1, Ab2, . . . , Abp.
That is,

AB = A [ b1 b2 . . . bp ] = [ Ab1 Ab2 . . . Abp ]

The product of two matrices is defined so it corresponds

to the

One wonders: Warning: The dimensions of the matrices involved in a product must verify
p
Does C exist | C x = ABx ∀x∈ K ?
A B = C
PROBLEM: What dimensions would C have? (m × n) (n × p) ( )
Algebra 2021/2022 4-6 Algebra 2021/2022 4-8
EXAMPLE: EXAMPLE:
    " #" #
2 3 4 3 6  
= 2 3 4 3 6 ? ?
1 −5 1 −2 3 =
1 −5 1 −2 3 ? ? ?
(2 × 2) (2 × 3) ⇒ ( )
1st row 3rd column → ( 1 , 3 ) entry
          
2 3 4 2 3 3 2 3 6
= = " #" #  
1 −5 1 1 −5 −2 1 −5 3 2 3 4 3 6 ? ? ?
=
          1 −5 1 −2 3 ? ?
= =
2nd row 1st column → ( 2 , 1 ) entry

Algebra 2021/2022 4-9 Algebra 2021/2022 4-11

Row-Column Rule for computing AB

PROBLEM: Find the 2nd row of AB.
Consider A ∈ Km×n, and B = [ b1 . . . bp ] ∈ Kn×p such
that (A)ik = aik , and (B)kj = bkj .  
2 −5 0  
  4 −6
AB = [ Ab1 · · · Abj · · · Abp ]  −1 3 −4   
AB =    7 1 
z }| {  6 −8 −7 
    3 2
a11 a12 . . . a1n   ?1 −3 0 9
 .. ..  b1j  .. 
  b2j   
 ai1 ai2 . . . ain   ..  =  ?i 
 . ..    −→ (AB)ij
 .  .. 
bnj PROBLEM: Compute
am1 am2 . . . amn ?m
That is,  
    1 1  
1 −1 2   1 −1 0 0
b1j  2 −1 
  3 0 1 0 0 1 1
  b2j  X 1 0
(AB)ij = ai1 ai2 . . . ain  . = aik bkj
 . 
k
bnj
Algebra 2021/2022 4-10 Algebra 2021/2022 4-12
 WARNING: In general, AB = AC ⇒
/ B=C
• PROPERTIES:
EXAMPLE:
    
Let A be an (m × n) matrix, and B and C matrices of 1 0 0 −1
=
0 0 1 0
appropriate dimensions:     
1 0 0 −1
=
0 0 −1 0
◦ A(BC) = (AB)C
◦ A(B + C) = AB + AC WARNING: In general, AB = 0 ⇒
/ A = 0 or B=0

◦ (B + C)A = BA + CA EXAMPLE:     
1 −2 4 −2
◦ µ (AB) = (µ A) B = A (µ B) ∀µ∈K =
−1 2 2 −1

◦ Im A = A = A In where Ik is the (k × k) identity matrix

  WARNING: In general, AA = 0 ⇒
/ A=0

      
EXAMPLE: 1 1 1 1
Ik = 


 =
→ 4.3
−1 −1 −1 −1

Algebra 2021/2022 4-13 Algebra 2021/2022 4-15

• If AB = BA, we say that A and B commute with each other

• The kth power of a matrix is defined:
WARNING: In general, AB 6= BA
Ak = A A A · · · A
EXPANSION AXIS X ROTATION 30◦ 1st EXPANSION + 2nd ROTATION | {z }
  √   
B=
2 0
A = 12
−1
3√
AB =
k times
0 1 1 3

This only makes sense if A is a matrix and k is a

nonnegative integer.
• For convenience, we define A0 = I.

ROTATION 30o EXPANSION AXIS X 1st ROTATION + 2nd EXPANSION PROBLEM: Compute
√     
−1
3√ 2 0
A = 21 B=
0 1
BA =  
1 3 1 −1
0 1
 
cos α − sin α
sin α cos α
→ 4.4
Algebra 2021/2022 4-14 Algebra 2021/2022 4-16
Transpose of a Matrix • A symmetric matrix verifies AT = A.
• The transpose of an (m × n) matrix A is the (n × m) EXAMPLE:
matrix AT whose columns are the rows of A.  
1 −1 3
That is,  
A =  −1 0 i 
(AT )ij = (A)ji
3 i 2
EXAMPLE:  

−5 2 ∀ C ∈ K(m×n)
 
B =  1 −3  ⇒ BT =
0 4 • An antisymmetric matrix verifies AT = −A.
EXAMPLE:
EXAMPLE:  
A = [ a1 a2 · · · an ] ⇒ AT = 0 −1 1
 
A =  1 0 −i 
−1 i 0


EXAMPLE: Row A = ( Row AT = ) ∀ C ∈ K(n×n)
→ 4.7
Algebra 2021/2022 4-17 Algebra 2021/2022 4-19

• PROPERTIES:
Conjugate Transpose of a Matrix
Let A and B be matrices of appropriate dimensions and µ ∈ K:
◦ (AT )T = A ◦ (A + B)T = AT + B T • The conjugate transpose of an (m × n) matrix A is the
(n × m) matrix A∗, or AH , whose elements verify:
◦ (µ A)T = µ (AT ) ◦ (AB)T = B T AT
(A∗)ij = (A)ji.
Proof: Let be A ∈ Km×n and B ∈ Kn×q

n
X Notice: If A is a real matrix, then A∗ = AT
T
(AB) ij
= (AB)ji = (A)jk (B)ki
k=1
n n
EXAMPLE:  
X X
= T T
(A )kj (B )ik = T T
(B )ik (A )kj = −5 2−i
k=1 k=1
B= i 3  ⇒ B∗ =
0 4

PROBLEM: Prove that (ABC)T = C T B T AT .

(ABC)T = A = [ a 1 a2 · · · a n ] ⇒ A∗ =

Algebra 2021/2022 4-18 Algebra 2021/2022 4-20

  
• PROPERTIES: cos φ − sin φ
EXAMPLE: This matrix is invertible: A =
Let A and B be matrices of appropriate dimensions and µ ∈ K: sin φ cos φ
◦ (A∗)∗ = A ◦ (µ A)∗ = µ̄ (A∗)
Rotation Rotation
◦ (A + B)∗ = A∗ + B ∗ ◦ (AB)∗ = B ∗ A∗ A= ⇒ A−1 =
φ in R2 −φ in R2

• A Hermitian matrix verifies A∗ = A    

−1 cos(−φ) − sin(−φ)
EXAMPLE:   Thus, A = =
sin(−φ) cos(−φ)
1 −1 3−2i
A =  −1 0 i 
3+2i −i −2 EXAMPLE: The matrix B has no inverse and is, therefore, a
∗
• An antihermitian matrix verifies A = −A singular matrix:
 
EXAMPLE:   Projection
i 1 3+2i B= =
in x (R2)
A =  −1 0 i 
−3+2i i −2i → 4.9
→ 4.8
Algebra 2021/2022 4-21 Algebra 2021/2022 4-23

4.2. Inverse of a Matrix Theorem 4.1. If A is an invertible (n × n) matrix, then the

equation Ax = b has the unique solution x = A−1 b, ∀ b ∈ Kn.
• A square (n × n) matrix A is invertible, or nonsingular,
if there exists a matrix B such that
Proof:
◦ That x = A−1 b is a solution ∀ b can be checked by a mere
AB = In
substitution: Ax = .
Theo. 3.10
This matrix is called inverse of A and is denoted by A−1.
◦ As it has a solution ∀ b ⇒ A must have a pivot
• A noninvertible or singular matrix has no inverse. A square No free variables
⇒ A has a pivot ⇒
 
2 5
EXAMPLE: This matrix is invertible: A =
−3 −7
Warning:
      
−7 −5 2 5 −7 −5 Finding A−1 in order to compute x = A−1 b is not an
Because C = verifies AC = =
3 2 −3 −7 3 2 advisable method to solve the linear system Ax = b.
Algebra 2021/2022 4-22 Algebra 2021/2022 4-24
 Theorem 4.2. If A and B are square matrices of Kn×n, Theorem 4.5. If A is invertible, AT is also invertible and
(AT )−1 = (A−1)T .
AB = I ⇔ BA = I

Proof: ( AB = I ⇒ BA = I )
Theorem 4.6. If A is invertible, A∗ is also invertible and
(A∗)−1 = (A−1)∗.
◦ Suppose that BA = X⇒ BA = X
⇒ A = AX ⇒ A − AX = 0 ⇒ Proof: A∗ (A−1)∗ =

◦ Let’s define M = I − X = [ m1 m2 · · · mn ]. EXAMPLE:

As AM = 0 ⇒ [ Am1 Am2 · · · Amn ] = 0.     
1+i 1+2i −1−i −1−2i 1 0
That is, Ami = 0 ∀ i. =
−1 −1−i 1 1+i 0 1
Teo. 4.1
◦ But now, Ami = 0 ⇒ mi = ⇒ M =  −1  
−1 + i 1
then, =
Leading to X = I ⇒ BA = I. −1 + 2i 1 − i
Algebra 2021/2022 4-25 Algebra 2021/2022 4-27

Theorem 4.7. If A and B are invertible (n × n) matrices,

Theorem 4.3. If A is an invertible matrix, then A−1 is
then AB is invertible and (AB)−1 = B −1 A−1.
invertible and (A−1)−1 = A.
Proof: AB(AB)−1 = AB
Proof: It follows from the above theorem.
EXAMPLE: If A = ROTATE , B = EXPAND , then,
AB = ROTATE EXPAND =
Theorem 4.4. If exists, the inverse of a matrix is unique.
(in this order!) and the inverse is
Proof: Let A be an invertible matrix, and B a matrix such that (AB)−1 = =
−1 −1

AB = I (that is, B = A−1). Suppose there exists C such

that AC = I (in other words, suppose that A has another
PROBLEM: If A, B and C are nonsingular matrices of equal
inverse).This inverse will verify . And then,
size, show that (ABC)−1 = C −1 B −1 A−1.
C=
ATTENTION: In general, (A + B)−1 6=
→ 4.11
Algebra 2021/2022 4-26 Algebra 2021/2022 4-28
 • An elementary matrix is one that is obtained by ◦ I ∼ E2 ( r2 ↔ r3)
    
performing one elementary row operation on an identity matrix.
1 0 0 a d
    
E2 A =  0 0 1  b e  =  
EXAMPLE:
0 1 0 c f

      ◦ I ∼ E3 ( r2 → r2 − 4r1)
1 0 0 1 0 0 1 0 0     
      1 0 0 a d
E1 =  0 1 0  E2 =  0 0 1  E3 = −4 1 0      
0 0 5 0 1 0 0 0 1 E3 A = −4 1 0  b e  =  
0 0 1 c f

Notice: These matrices have a clear geometrical interpretation. They

correspond to an expansion along an axis, a reflection (interchange of axes)
and a shear transformation, respectively.
A ∼ A ∼ A ∼
r3 → 5 r3 r2 ↔ r3 r2 → r2 −4 r1

Algebra 2021/2022 4-29 Algebra 2021/2022 4-31

Theorem 4.9. Every elementary matrix E is invertible and

Theorem 4.8. If an elementary row operation if performed its inverse E −1 is the elementary matrix corresponding to the
on an (m × n) matrix A, the resulting matrix can be written row operation that transforms E back into I.
as EA, where E is the (m × m) elementary matrix created by
EXAMPLE: The matrix E1 multiplies the 3rd row by five:
performing the same operation on Im.  
1 0 0
  E1 =  0 1 0 
a d 0 0 5
 
EXAMPLE: Consider the (3 × 2) matrix A=b e
c f Its inverse E1−1 is the matrix that divides the 3rd row by five:
 
◦ I ∼ E1 ( r3 → 5 r3) E1−1 =  
    
1 0 0 a d
     Check: E1 E1−1 = · · · = I
E1 A =  0 1 0  b e  =  
0 0 5 c f
→ 4.12 PROBLEM: Find the matrices E2−1 and E3−1 .
Algebra 2021/2022 4-30 Algebra 2021/2022 4-32
    
1 0 3 0 1 0 1 0 3 0 1 0
Theorem 4.10. An (n × n) matrix A is invertible if and only    
 0 1 2 1 0 0  ∼  0 1 2 1 0 0  ∼
if A is row equivalent to In. In this case, any sequence 0 −3 −4 0 −4 1 0 0 2 3 −4 1
of elementary row operations that transforms A into In also | {z } | {z } | {z } | {z }
transforms In in A−1.
I
Proof:    
Theo. 4.1 A square
−1 1 0 3 0 1 0 1 0 3 0 1 0
◦ ∃A ⇒ Ax = b has a unique solution ∀ b ⇔    
 0 1 0 −2 4 −1  ∼  0 1 0 −2 4 −1  ∼
◦ A ∼ In ⇒ A ∼ E1 A ∼ · · · ∼ Ep Ep−1 · · · E2 E1 A = In 0 0 2 3 −4 1 0 0 1 32 −2 21
Then, Ep Ep−1 . . . E2 E1 = and, therefore, ∃ A−1
   
1 0 0 − 92 7 − 32
◦ On our way, we have found an expression for A−1:    
 0 1 0 −2 4 −1  ⇒ A−1 =  
A−1 = Ep Ep−1 . . . E2 E1 I 0 0 1 32 −2 12
→ 4.14 | {z } | {z }
Algebra 2021/2022 4-33 Algebra 2021/2022 4-35

An Algorithm for finding A−1 PROBLEM: If exists, find the inverse of the matrix
 
1 0 −2
◦ Construct the matrix [ A I ]  
C =  3 1 −2 
−5 −1 9
◦ Find its reduced echelon form.  

◦ If this matrix has the form [ I B ] , then A−1 = B .  

[ C I ] = 
Otherwise, A does not have an inverse.

EXAMPLE:
   
0 1 2 1 0 0 1 0 3 0 1 0
   
 1 0 3 0 1 0  ∼  0 1 2 1 0 0  ∼
4 −3 8 0 0 1 4 −3 8 0 0 1
| {z } | {z } | {z } | {z } Check: C C −1 =
A I → 4.16

Algebra 2021/2022 4-34 Algebra 2021/2022 4-36

 Theorem 4.11. (The Square Matrix Theorem) 4.3. Partitioned (or Block) Matrices
If A ∈ Kn×n
, the following statements are equivalent:
1. A is an invertible matrix.
2. There exists C ∈ Kn×n such that AC = In . EXAMPLE:  
3. There exists D ∈ Kn×n such that DA = In. 3 0 −1 5 9 −2
4. A is row equivalent to In.  
5. A has n pivots. A =  −5 2 4 0 −3 1 
6. The equation Ax = 0 has only the trivial solution. −8 −6 3 1 7 −4
7. The columns/rows of A are linearly independent.
8. The equation Ax = b has a (unique) solution ∀ b ∈ Kn.  
9. The columns/rows of A span Kn.  
10. The columns/rows of A form a basis of Kn  A11 A12 A13 
A=  =
11. AT is invertible.
12. A∗ is invertible. A21 A22 A23
13. The linear transformation x → Ax is bijective.
14. Col A = Row A = Kn where      
15. dim Col A = dim Row A = n 3 0 −1 5 9 −2
A11 = , A12 = , A13 =
16. rank A = n −5 2 4 0 −3 1
17. Nul A = {0}      
18. dim Nul A = 0 A21 = −8 −6 3 , A22 = 1 7 , A23 = −4

Algebra 2021/2022 4-39

EXAMPLE: Social web of 6 persons in 3 groups

• A transformation T : Kn −→ Kn is called
invertible if there exists a transformation S : Kn −→ Kn
such that

S(T (x)) = x Adjacency Matrix
∀ x ∈ Kn.
T (S(x)) = x

The transformation S is called the inverse of T .

 
Theorem 4.12. Let T : Kn −→ Kn be a linear 0 1 1 0 0 0
 1 0 1 0 0 0   
transformation and A its canonical matrix. T is invertible   M11 M12 M13
 1 1 0 1 0 1 
if and only if A is nonsingular. In this case, S(x) = A−1x. M =

 =  M21 M22 M23 

 0 0 1 0 0 0 
  M31 M32 M33
0 0 0 0 0 1
→ 4.17
0 0 1 0 1 0

Algebra 2021/2022 4-38 Algebra 2021/2022 4-40

EXAMPLE: Jefferson High School
• PROPERTIES:
◦ Addition: Matrices of equal size and identical partition can
be summed block by block:
   
A11 A12 A13 B11 B12 B13
A+B = +
A21 A22 A23 B21 B22 B23
 
A11 +B11 A12 +B12 A13 +B13
=
A21 +B21 A22 +B22 A23 +B23

◦ Scalar Multiplication:
   
A11 A12 A13 λA11 λA12 λA13
λA = λ =
A21 A22 A23 λA21 λA22 λA23

Algebra 2021/2022 4-41 Algebra 2021/2022 4-43

EXAMPLE: Trade share matrix between countries ◦ Transpose of a matrix:

   
  AT11 AT21
A11 A12 A13  T   
A= ⇒ AT =

A12 AT22  6=
 


A21 A22 A23
AT13 AT23

◦ Conjugate transpose of a matrix:

 
  A∗11 A∗21
A11 A12 A13  ∗ 
A= ⇒ A∗= A12 A∗22 
A21 A22 A23  
A∗13 A∗23
EXAMPLE:
   
2 0 8 2 1 0
   
A =  1 −5 3  ⇒ AT =  0 −5 −2 
0 −2 7 8 3 7

Algebra 2021/2022 4-42 Algebra 2021/2022 4-44

◦ Multiplication of partitioned matrices: Two matrices A EXAMPLE: Let A be a block upper triangular matrix:
and B of respective dimensions (m × n) and (n × p) are  
conformable for block multiplication when the number of A11 A12
A= .
columns of each partition of A is equal to the number of 0 A22
rows of the corresponding partition of B.
Assuming that A is invertible, A11 is (p × p) and A22 is (q × q),
 
  6 4 find a formula for A−1.
2 −3 1 0 −4   −2 1 

   Call B = A−1. Partition B in such a way that we can write:
AB =  1 5 −2 3 −1  −3 7 
      
0 −4 −2 7 −1  −1 3  A11 A12 B11 B12 I 0
5 2 AB = = .
0 A22 B21 B22 0 I
    
A11 A12 B11 The dimensions of the matrices involved are:
= =      
A21 A22 B21 A11 A12 B11 B12 Ix 0
=
0 A22 B21 B22 0 Iy
(Attention: The order of the products is important)
Algebra 2021/2022 4-45 Algebra 2021/2022 4-47

The equation can be written:

Concentrate on the dimensions of the blocks:
   
I 0
   = .
   0 I
(2 × 3) (2 × 2) (3 × 2)
(3 × 5) (5 × 2) = =
(1 × 3) (1 × 2) (2 × 2)
Equating components, we obtain:
 
(2 × 3)(3 × 2) + (2 × 2)(2 × 2) (a) = I
= =
(1 × 3)(3 × 2) + (1 × 2)(2 × 2) (b) = 0
(c) = 0
    (d) = I
(2 × 2) + (2 × 2) (2 × 2)  
= = = (3 × 2)
(1 × 2) + (1 × 2) (1 × 2)
We have to solve 4 matrix equations, which represent a
linear system of (p + q)2 equations with (p + q)2 unknowns.
Algebra 2021/2022 4-46 Algebra 2021/2022 4-48
 A22 square
◦ (d) A22B22=I ⇒ Theorem 4.14. A diagonal matrix is invertible if and only if
none of its diagonal elements is zero.
∃ A−1
22
◦ (c) A22B21=0 ⇒  −1  
a11 0 . . . 0 1/a11 0 . . . 0
   
B21 = 0  0 a22 0   0 1/a22 0 
 .  = . 
 .   . 

··

··
◦ (a) A11B11+A12B21=I ⇒

·
0 0 ann 0 0 1/ann
If so,
∃ A−1
11
◦ (b) A11B12+A12B22=0 ⇒ PROBLEM: Determine under what conditions the following
matrix is invertible and, in that case, find its inverse:
 
Obtaining, " # Im 0
.
A−1 = . A In
→ 4.19

Algebra 2021/2022 4-49 Algebra 2021/2022 4-51

4.4. Determinants
Theorem 4.13. A block diagonal matrix is invertible if and
only if each of the diagonal blocks is invertible. • Given an (m × n) matrix A, we define the minor Aij as
the ((m−1) × (n−1)) matrix obtained by removing the ith row
Proof: The case of two blocks follows from the above result and the jth column of the matrix A.
when A12 = 0. The general case of n blocks can be reduced EXAMPLE:  
1 5 0
recursively to that case. A= 2 4 −1  ⇒ A 1 2 = A23 =
0 −2 0
 −1  
 C11 0 ... 0   0 ... 0 







 • Let A be an (n × n) matrix whose entry (A)ij = aij .
   
    We define the determinant of A as
 0 C22 0   0 0 
   
   
    n n

 ..


= 



X X
... j+1
 .    det A = |A| = (−1) a1j det A1j = a1j C1j ,
··

··

   
·

   
    j=1 j=1
   
 0 0 Cnn   0 0 
where Cij = (−1)i+j det Aij is refered to as the ij cofactor of A.
Algebra 2021/2022 4-50 Algebra 2021/2022 4-52
That is, the determinant of A is defined as the cofactor
expansion along the first row of A. Theorem 4.16. If A is an (n × n) triangular matrix,
its determinant is the product of its diagonal entries.

Theorem 4.15. The determinant of an (n × n) matrix A can

 
be expressed as the cofactor expansion along any row of the a11 0 0 0 0  
matrix  ? a 0 0 0  a22 0 0 0
   22   
Xn Xn    ? a33 0 0 
det A = (−1) k+j
akj det Akj = akj Ckj
along the det  ? ? a33 0 0  = a11 det   =
kth row    ? ? a44 0 
j=1 j=1  ? ? ? a44 0 
? ? ? a55
? ? ? ? a55

WARNING:  
◦ Determinants are only defined for square matrices. a33 0 0
 
◦ The factor (−1)k+j only depends on the position of akj but is = a11 a22 det  ? a44 0  = · · · =
of the actual sign of akj . ? ? a55

Algebra 2021/2022 4-53 Algebra 2021/2022 4-55

 
EXAMPLE: 1 5 0 Theorem 4.17. Let A be an (n×n) matrix.
 
det  2 4 −1  If we obtain a matrix B,
0 −2 0 ◦ By adding to a row of A the multiple of another row,
1st row: det B = det A.
= ◦ By multiplying one row of A by λ,
det B = λ det A.
◦ By interchanging two rows of A,
det B = − det A.
3rd row:
EXAMPLE:
= −3 3 −3
−2 2 −1 = = = =
1 0 1

Algebra 2021/2022 4-54 Algebra 2021/2022 4-56

 Theorem 4.19. If A and B are square matrices,
Theorem 4.18. Let A be a square matrix and U an echelon
det(AB) = det A det B.
matrix obtained from A by adding multiples of rows and r row
interchanges (but without multiplying any row by a scalar!).
Then, Theorem 4.20. If A is a square matrix,

 0 if A is not invertible |AT | = |A| and |A∗| = |A|
det A =  
 (−1)r · product of if A is invertible
the pivots Proof:
◦ For elementary matrices, it’s easy to see that |E| = |E T |.
◦ If we obtain an echelon form of a matrix A:
Proof: A ∼ · · · ∼ U = EpEp−1 . . . E2E1A
→ 4.20
Leading to UT =
Note: This would add a new statement to theorem 4.11: ◦ Now, as U is a triangular matrix, |U T | = |U | and, consequently

19. The determinant of A is nonzero.

→ 4.24

Algebra 2021/2022 4-57 Algebra 2021/2022 4-59

Determinants as Area and Volume

WARNING: In general,
A ∼ B ⇒
/ det A = det B. Theorem 4.21. If A = [ a1 a2 ] is a (2 × 2) real matrix, the
Check theorem 4.17! area of the parallelogram formed by a1 and a2 is |det A|

Proof:
WARNING: In general,
det(A + B) 6= det A + det B.

EXAMPLE: If it was true, all determinants would be zero:

           0 0
a b a 0 0 b 0 0 0 0 0 0 a b
det = det + + + A = |a d | = det = | det[ Rφa1 Rφa2 ]| =
c d 0 0 0 0 c 0 0 d 0 d0
       
a 0 0 b 0 0 0 0
6= det + det + det + det =
0 0 0 0 c 0 0 d

Algebra 2021/2022 4-58 Algebra 2021/2022 4-60

EXAMPLE: Compute the area of the following triangle:
Theorem 4.24. Let A be the canonical matrix of a linear transformation
T : R3 → R3 . If S is a parallepided in R3, then
V T (S) = | det A| × VS

Theorem 4.25. Let A be the canonical matrix of

Finding a linear transformation T : Rn → Rn .
If S is a bounded reggion of Rn, then
A= V T (S) = | det A| × VS

Theorem 4.22. If A is a (3 × 3) real matrix, the

volume of the parallelepiped determined by the
columns of A is |det A|

Algebra 2021/2022 4-61 Algebra 2021/2022 4-63

x2 y 2 z 2
Theorem 4.23. Let A be the canonical matrix of EXAMPLE: Determine the volume of the ellipsoid + +
a2 a2 c2
≤ 1

a linear transformation T : R2 → R2 . If S is a
parallelogram in R2, then
A T (S) = | det A| × AS
e = Ax:
The linear transformation x
    
e
x 0 0 x
 ye  =  0 0  y 
ze 0 0 z

Proof: 4π
e 2 + ye 2 + ze 2 ≤ 1 whose volume is
transforms the ellipsiod into the sphere x 3
S = {x = } ⇒ AS = | det[ a1 a2]|
Therefore,
T (S) = {x = ∀ α, β ∈ [ 0, 1]} VSPHERE = | det A| × VELLIPSOID
⇒ AT (S) = | det[ T (a1) T (a2)]| = | det[ Aa1 Aa2]| leading to

VELLIPSOID =
= | det A[ a1 a2] | = | det A| × | det[ a1 a2]| → 4.25

Algebra 2021/2022 4-62 Algebra 2021/2022 4-64