Linear algebra

Youming Chen([email protected])
Chongqing University of Technology

October 9, 2020

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 1 / 64
 Chapter 1 Matrices and Systems of Equations(›!Üê
ß|)

1.1 Systems of Linear Equations(Ç5êß|)

Consider the linear system of m equations in n variables(C˛)

8
>
> a11 x1 + a12 x2 + · · · + a1n xn = b1 ,
>
>
>
>
<a x + a x + · · · + a x = b ,
21 1 22 2 2n n 2
>
> ············
>
>
>
>
:am1 x1 + am2 x2 + · · · + amn xn = bm ,

We call it the m ⇥ n linear system. The linear system is called inconsistent

if it has no solution, otherwise it is called consistent.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 2 / 64
 2 ⇥ 2 linear system

First we consider a much more simple case

8
<a x + a x = b ,
11 1 12 2 1
:a x + a x = b .
21 1 22 2 2

When a11 a22 a12 a21 6= 0, Using ”Elimination Method(û!{)” we could

obtain the unique solution
b1 a22 b2 a12 a11 b2 a21 b1
x1 = , x2 = .
a11 a22 a21 a12 a11 a22 a21 a12

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 3 / 64
 We call a11 a22 a21 a12 the second order determinant(#"1!™) and
denote it by
a11 a12
a11 a22 a21 a12 = .
a21 a22
aij is called the entry(!É) of the determinant, the index i is called the
row index(1çI), and the index j is called the column index(!çI).

Therefore if we denote

a11 a12 b1 a12 a11 b1

D= , D1 = , D2 = ,
a21 a22 b2 a22 a21 b2

then when D 6= 0, the 2 ⇥ 2 linear system has the unique solution

D1 D2
x1 = , x2 = .
D D

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 4 / 64
 Example
Solve the 2 ⇥ 2 linear system
8
<3x 2x2 = 12,
1
:2x + x = 1.
1 2

Solution

3 2 12 2 3 12
D= = 7 6= 0, D1 = = 14, D2 = = 21.
2 1 1 1 2 1

Then the 2 ⇥ 2 linear system has the only solution

D1 D2
x1 = = 2, x2 = = 3.
D D

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 5 / 64
 3 ⇥ 3 linear system

We call

a11 a12 a13

a21 a22 a23 = a11 a22 a33 + a21 a32 a13 + a31 a12 a23
a31 a32 a33
a11 a32 a23 a21 a12 a33 a31 a22 a13

the third order determinant which could be obtained by the diagonal

rule(È#Ç{K)

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 6 / 64
 Example
1 2 1
Calculate 2 1 3 .
1 1 1

Example
1 1 1
Solve the equation 2 3 x =0
4 9 x2

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 7 / 64
 Consider the 3 ⇥ 3 linear system
8
>
> a x + a12 x2 + a13 x3 = b1 ,
>
< 11 1
a21 x1 + a22 x2 + a23 x3 = b2 ,
>
>
>
:a31 x1 + a32 x2 + a33 x3 = b3 .

a11 a12 a13

Then when D = a21 a22 a23 6= 0, the 3 ⇥ 3 linear system has the
a31 a32 a33
only solution
D1 D2 D3
x1 = , x2 = , x3 = ,
D D D
where

b1 a12 a13 a11 b1 a13 a11 a12 b1

D1 = b2 a22 a23 , D2 = a21 b2 a23 , D3 = a21 a22 b2 .
b3 a32 a33 a31 b3 a33 a31 a32 b3
Youming Chen([email protected]) Chongqing University
Linearof algebra
Technology () October 9, 2020 8 / 64
 Ideas of Linear Algebra

Philosophically, we may solve all linear systems by the Elimination Method.

We will explain this procedure by a new theory-”Matrix Theory”.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 9 / 64
 Definition
The Elementary Row Operations(–"1CÜ) of linear systems are the
following three operations:
(1) Interchange two rows;
(2) Multiply a row by a nonzero real number;
(3) Replace a row by its sum with a multiple of another row.

Definition
Two systems of equations are said to be equivalent("d) if one could
obtained by applying finite elementary row operations on another.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 10 / 64
 Matrix(›!)

Definition
A m ⇥ n array of numbers with m rows and n columns
0 1
a11 a12 · · · a1n
B C
B a21 a22 · · · a2n C
B
A=B . .. C
. .. C = (aij )
@ . . . A
am1 am2 · · · amn

is called a m ⇥ n matrix(›!). aij is called the entry(!É) of A, the

index i is called the row index(1çI), and the index j is called the
column index(!çI). The n ⇥ n matrix is also called square(ê!)of
order n. A 1 ⇥ n matrix is called row vector(1ï˛). A m ⇥ 1 matrix is
called column vector(!ï˛).

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 11 / 64
 Example
(1) Omn = (aij ), aij = 0;
(2) Diagonal matrix(È#›!)
0 1
1 0 ··· 0
B C
B 0 2 ··· 0 C
diag( 1 , · · · , n ) = B
B .. .. .. C;
C
@ . . . A
0 0 ··· n

(3) Identity matrix(¸†›!) E = I = diag(1, · · · , 1).

0 1
0 0 ··· n
B . .. .. C
B .. . . C
(4) Antidiagonal Matrix(gÈ#›!) B B
C,
C
@ 0 2 ··· 0 A
1 0 ··· 0

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 12 / 64
 0 1
a11 a12 · · · a1n
B C
B 0 a22 · · · a2n C
(5) Upper triangular matrix(˛n#›!) B
B .. .. .. C,
C
@ . . . A
0 0 ··· ann
0 1
a11 0 · · · 0
B C
B a21 a22 · · · 0 C
(6) Lower triangular matrix(en#›!) B
B .. .. .. C.
C
@ . . . A
an1 an2 · · · ann

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 13 / 64
 For a m ⇥ n linear system, Let
0 1 0 1 0 1
x1 b1 a11 a12 · · · a1n b1
B C B C B C
B x2 C B b2 C B a21 a22 · · · a2n b2 C
A = (aij ), X = B . C , b = B
B C
B .. C,B = B
C B .. .. .. .. C
C,
@ .. A @ . A @ . . . . A
xn bn an1 an2 · · · ann bn

then A is called the coefficient matrix(XÍ›!), X is called the Unknown

matrix(ô#Í›!), b is called the constant matrix(~Íë›!), B is
called augmented matrix£O2›!§.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 14 / 64
 Definition
The Elementary Row Operations(–"1CÜ) of matrices are the
following three operations:
(1) Interchange two rows, for example ith row and jth row which is
denoted by ri $ rj ;
(2) Multiply a row by a nonzero real number, for example, multiply ith
row by ri which is denoted by ri ⇥ k;
(3) Replace a row by its sum with a multiple of another row, for example,
add k times of jth row to ith row which is denoted by ri + krj .

Obviously all elementary row operations are invertible(å_#).

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 15 / 64
 Definition
We say two matrixes A and B are equivalent if one could obtained by
applying finite elementary operations on another. Then we will denote this
relation by A ⇠ B.

Property
The equivalent relation of matrices has the following properties:
(1) Reflexive property(á#5) A ⇠ A;
(2) Symmetric property(È°5) If A ⇠ B, then B ⇠ A;
(3) Transitive property(D45) If A ⇠ B, B ⇠ C , then A ⇠ C .

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 16 / 64
 1.2 Row Echelon Form(1"F/™)

Example (Underdetermined Systems)

Solve the system of equations
8
>
> 2x1 x2 x3 + x4 = 2,
>
>
>
>
<x + x 2x3 + x4 = 4,
1 2
(1)
>
> 4x1 6x2 + 2x3 2x4 = 4,
>
>
>
>
:3x1 + 6x2 9x3 + 7x4 = 9.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 17 / 64
 Solution
8
>
> x1 + x2 2x3 + x4 = 4,
>
>
>
>
< 2x2 2x3 + 2x4 = 0,
(1) !
>
> 5x2 + 5x3 3x4 = 6,
>
>
>
>
: 3x2 3x3 + 4x4 = 3,
8
>
> x1 + x2 2x3 + x4 = 4,
>
>
>
>
< x2 x3 + x4 = 0,
!
>
> 2x4 = 6,
>
>
>
>
: x4 = 3,

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 18 / 64
 8
>
> x1 + x2 2x3 + x4 = 4,
>
>
>
>
< x2 x3 + x4 = 0,
!
>
> x4 = 3,
>
>
>
>
: 0 = 0.
8
>
> x1 + 0 x3 + 0 = 4,
>
>
>
>
< x2 x3 + 0 = 3,
!
>
> x4 = 3,
>
>
>
>
: 0 = 0.
0 1 01 0 1 0 1
x1 c +4 1 4
B C B C B C B C
B x2 C B c +3 C B 1 C B 3 C
Therefore x = B
B
C=B
C B c C
C=cB
B
C+B
C B
C , c 2 R.
C
@ x3 A @ A @ 1 A @ 0 A
x4 3 0 3
Youming Chen([email protected]) Chongqing University
Linearof algebra
Technology () October 9, 2020 19 / 64
 Equivalently,
0 1 0 1
2 1 1 1 2 1 1 2 1 4
B C r1 $ r2 , B C
B 1 1 2 1 4 C r3 ⇥ 12 B 2 1 1 1 2 C
B = B C ⇠ B C
B 4 6 2 2 4 C B 2 3 1 1 2 C
@ A @ A
3 6 9 7 9 3 6 9 7 9
0 1 0 1
r2 r3 , 1 1 2 1 4 r2 ⇥ 12 , 1 1 2 1 4
r3 2r1 , B C r3 + 5r2 , B C
r4 3r1 B 0 2 2 2 0 C r4 3r2 B 0 1 1 1 0 C
⇠ B C ⇠ B C
B 0 5 5 3 6 C B 0 0 0 2 6 C
@ A @ A
0 3 3 4 3 0 0 0 1 3
r3 $ r4 , 0 1
r4 2r3 , 1 0 1 0 4
r1 r2 , B C
r2 r3 B 0 1 1 0 3 C
⇠ B C
B 0 0 0 1 3 C
@ A
0 0 0 0 0

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 20 / 64
 Definition (Row Echelon Form)
A matrix is said to be in row echelon form if
(1) The first nonzero entry in each nonzero row is 1;
(2) If the k-th row does not consist entirely of zeros, the number of
leading zero entries in the k + 1-th row is bigger than the number of
leading zero entries in the k-th row;
(3) If there are rows whose entries are all zero, they are below the rows
having nonzero entries.

Definition (Reduced Row Echelon Form)

A matrix is said to be in reduced row echelon form if
(1) The matrix is in row echelon form;
(2) The first nonzero entry in each row is the only nonzero entry in its
column.
Youming Chen([email protected]) Chongqing University
Linearof algebra
Technology () October 9, 2020 21 / 64
 Definition (Gaussian elimination)
The process of applying the elementary operations to transform a linear
system into another whose augmented matrix is in row echelon form is
called Gaussian elimination.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 22 / 64
 A system of linear equations is said to be homogeneous(‡g#) if the
constants on the righthand side are all zero.

Theorem
An m ⇥ n homogeneous system of linear equations has a nontrivial
solution if n > m.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 23 / 64
 Homework

Page 10-11, 6(a), 6(d), 6(h), 9.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 24 / 64
 1.3 Matrix Arithmetic(›!#$é) and 1.4 Matrix
Algebra(›!ìÍ)

Fix a natural number n. Denote Rn by the set of all n column vector, i.e.
0 1
x1
B . C
R := {x = B
n
@ .
. C |x1 , · · · , xn 2 R}.
A
xn

An m ⇥ n matrix A = (aij )mn has m row vectors ↵i = (ai1 , ai2 , · · · , ain ),

and A can be represented by
0 1
↵1
B . C
A=B . C
@ . A.
↵m
Youming Chen([email protected]) Chongqing University
Linearof algebra
Technology () October 9, 2020 25 / 64
 0 1
a1j
B C
B a2j C
Analogously A = (aij )mn has n column vectors j =B
B .. C, and A can
C
@ . A
amj
be represented by
A=( 1, · · · , n ).

Definition
Two m ⇥ n matrix A = (aij )mn and B = (bij ) are called to be equal if
aij = bij for each i and j.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 26 / 64
 Definition (Scalar Multiplication)
We call matrix
0 1
a11 a12 ··· a1n
B C
B a21 a22 ··· a2n C
C = (Cij )mn = B
B .. .. .. C
C
@ . . . A
am1 am2 · · · amn

the scalar multiplication(Ͳ¶») of matrix A = (aij )mn and scalar and

we denote it by C = A.

Property
(1) Distribution Law(©""): ( 1 + 2 )A = 1A + 2 A;

(2) Distribution Law(©""): (A + B) = A + B;

(3) Associative Law ((‹"): 1 ( 2 A) =( 1 2 )A;

(4) Identity(¸†!): 1A = A.
Youming Chen([email protected]) Chongqing University
Linearof algebra
Technology () October 9, 2020 27 / 64
 Definition (Matrix Addition)
Suppose
0 1 0 1
a11 a12 ··· a1n b11 b12 ··· b1n
B C B C
B a21 a22 ··· a2n C B b21 b22 ··· b2n C
A=B
B .. .. .. C,B = B
C B .. .. .. C
C
@ . . . A @ . . . A
am1 am2 · · · amn bm1 bm2 · · · bmn

are two m ⇥ n matrices. Then the matrix

0 1
a11 + b11 a12 + b12 · · · a1n + b1n
B C
B a21 + b21 a22 + b22 · · · a2n + b2n C
C = (cij )mn = B
B .. .. .. C
C
@ . . . A
am1 + bm1 am2 + bm2 · · · amn + bmn

is called the sum of A and B, and is denoted by C = A + B.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 28 / 64
 The matrix A = ( aij ) is called the additive inverse(K›!) of A. The
subtraction of the matrices A and B is defined by

A B = A + ( B).

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 29 / 64
 The matrix A = ( aij ) is called the additive inverse(K›!) of A. The
subtraction of the matrices A and B is defined by

A B = A + ( B).

Property
(1) Associative Law ((‹"): A + (B + C ) = (A + B) + C ;
(2) Commutative law(%Ü"): A + B = B + A;
(3) Identity(¸†!): A + O = A;
(4) A A = O.

Both scalar multiplication and addition of matrices are called Linear

operation.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 29 / 64
 Consider the linear systems
8
>
> a11 x1 + a12 x2 + · · · + a1n xn = b1 ,
>
>
>
>
<a x + a x + · · · + a x = b ,
21 1 22 2 2n n 2
>
> ············
>
>
>
>
:am1 x1 + am2 x2 + · · · + amn xn = bm ,

whose constant matrix is b, coefficient matrix is A = (aij )mn with n

column vector j . Then the linear systems could be rewritten as

x1 1 + · · · + xn n = b.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 30 / 64
 Some Motivation of Matrix Multiplication

Suppose we want to solve a system of linear equations about variables

x1 , · · · , xn
8
>
> b11 x1 + b12 x2 + · · · + b1n xn = y1 ,
>
>
>
>
<b x + b x + · · · + b x = y ,
21 1 22 2 2n n 2
>
> ············
>
>
>
>
: bm1 x1 + bm2 x2 + · · · + bmn xn = ym .

We denote its coefficient matrix by B = (bij ). However y1 , · · · , ym are

determined by variables z1 , · · · , zl via linear systems

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 31 / 64
 8
>
>
>a11 y1 + a12 y2 + · · · + a1m ym = d1 ,
>
>
>
<a y + a y + · · · + a y = d ,
21 1 22 2 2m m 2
>
> ············
>
>
>
>
: a y 1 + a y2 + · · · + a ym = d ,
l1 l2 lm l

whose coefficient matrix is denoted by A = (aij ). Then we may directly

solve the linear systems
8
>
> c11 x1 + c12 x2 + · · · + c1n xn = d1 ,
>
>
>
>
<c x + a x + · · · + c x = d ,
21 1 22 2 2n n 2
>
> ············
>
>
>
>
: c x1 + c x 2 + · · · + c xn = d ,
l1 l2 ln l

Pm Pm
where cij = ai1 b1j + · · · + aim bmj = k=1 k=1 aik bkj .

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 32 / 64
 Definition
The product of matrices A = (aik )lm and B = (bkj )mn is the matrix
C = (cij )ln = AB, where
m
X
cij = ai1 b1j + ai2 b2j + · · · + aim bmj = aik bkj .
k=1

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 33 / 64
 Example
0 1
0 1 1 2 0 1
4 1 2 1 B C 9 9
B CB 0 1 C B
C= @ 2 9 C
@ 1 1 0 3 AB
B C A.
@ 3 0 A
0 3 1 4 1 11
1 2

Example
! ! !
1 1 1 1 0 0
(1) = ;
1 1 1 1 0 0
! ! !
1 1 1 1 2 2
(2) = .
1 1 1 1 2 2

Multiplication of matrices is not commutative. The product of the

matrices may be zero matrix, hence the Elimination law(û#") is NOT
always satisfied.
Youming Chen([email protected]) Chongqing University
Linearof algebra
Technology () October 9, 2020 34 / 64
 Definition
The powers(ò) of square matrix A are defined as

A1 = A, · · · , Ak+1 = Ak A.

Obviously, Ak Al = Ak+l , (Ak )l = Akl . However, generally (AB)k 6= Ak B k .

Property
(1) (AB)C = A(BC );
(2) (AB) = ( A)B = A( B);
(3) A(B + C ) = AB + AC , (B + C )A = BA + CA;
(4) OA = O, AO = O;
(5) EA = A, AE = A;

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 35 / 64
 Example
Consider the linear systems
8
>
> a11 x1 + a12 x2 + · · · + a1n xn = b1 ,
>
>
>
>
<a x + a x + · · · + a x = b ,
21 1 22 2 2n n 2
>
> ············
>
>
>
>
:am1 x1 + am2 x2 + · · · + amn xn = bm ,

whose coefficient matrix is A = (aij )mn , Unknown matrix is X , constant

matrix is b. Then the linear systems could be rewritten as

AX = b.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 36 / 64
 Definition (Matrix Inversion)
An n ⇥ n matrix A is said to be nonsingular(ö¤…#) or invertible(å_
#) if there exists another matrix B such that

AB = BA = E .

The matrix B is called the multiplicative inverse of A and is denoted by

A 1 . Otherwise n ⇥ n is called singular(¤…#).

If A is invertible, then its inverse is unique since if both B, C are the

inverses of A, then B = EB = CAB = C .

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 37 / 64
 Property
If A, B are invertible matrices, 6= 0, then
(1) (A 1) 1 = A;
(3) A is invertible and A = 1 A 1;

(4) AT is invertible and (AT ) 1 = (A 1 )T ;

(5) AB is invertible and (AB) 1 =B 1A 1.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 38 / 64
 Definition
The transpose(=ò) of a matrix A = (aij ) is the matrix
0 1
a11 a21 · · · am1
B C
B a12 a22 · · · am2 C
A =B
T
B .. .. .. C
C
@ . . . A
a1n a2n · · · amn

Property
(1) (AT )T = A;
(2) (A + B)T = (AT + B T );
(3) (AB)T = B T AT ;
(4) (cA)T = cAT .

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 39 / 64
 Example
0 1
! 1 7 1
2 0 1 B C
Suppose A = ,B = @ 4 2 3 A , then calculate
1 3 2
2 0 1
(AB)T .

Solution
10 0 1 0 1
1 4 2 2 1 0 17
B CB C B C
(AB)T = B T AT = @ 7 2 0 A @ 0 3 A = @ 14 13 A .
1 3 1 1 2 3 10

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 40 / 64
 Definition
A square matrix A is called symmetric(È°#) if AT = A, i.e. aij = aji . A
square matrix A is called skew-symmetric(áÈ°#) if AT = A, i.e.
aij = aji .

Example
Suppose X = (x1 , · · · , xn )T satisfies X T X = 1 and let H = E 2XX T .
Then prove H is symmetric and HH T = E .

Proof.
H T = (E 2XX T )T = E T 2(XX T )T = E 2XX T = H, hence H is
symmetric. Moreover,

HH T = H 2 = (E 2XX T )2
= E 4XX T + 4XX T XX T
= E.
Youming Chen([email protected]) Chongqing University
Linearof algebra
Technology () October 9, 2020 41 / 64
 Example
Prove that any square matrix A could be represented by the sum of some
symmetric matrix and skew-symmetric matrix.

Proof.
Given any square matrix A, let

A + AT A AT
B= ,C = ,
2 2
A+AT A AT
then A = 2 + 2 = B + C,

(A + AT )T (A AT ) T AT A
BT = = B, C T = = = C
2 2 2
which means that B is symmetric and C is skew-symmetric.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 42 / 64
 Homework

Page 45-46, 2, 13.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 43 / 64
 1.5 Elementary Matrices(–"›!)

Three elementary matrices correspondent to three elementary operations:

(1) Interchanging ith and jth rows of identity matrix
0 1
1
B .. C
B . C
B C
B C
B 1 C
B C
B 0 ··· 1 C
B C
B C
B 1 C
B C
B .. .. .. C
E (i, j) = B . . . C,
B C
B 1 C
B C
B C
B 1 ··· 0 C
B C
B 1 C
B C
B . C
B .. C
@ A
Youming Chen([email protected]) Chongqing University
Linearof algebra
Technology () 1 9, 2020 44 / 64
October
 0 1
a11 a12 ··· a1n
B . .. .. C
B .. . . C
B C
B C
B aj1 aj2 ··· ajn C
B C
B . .. .. C
E (i, j)A = B .. . . C ,
B C
B a ··· C
B i1 ai2 ain C
B . .. .. C
B .. . . C
@ A
am1 am2 ··· amn
is equivalent to interchanging the ith and jth rows of A;

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 45 / 64
 (2) Multiplying ith row of E by a nonzero constant :
0 1
1
B .. C
B . C
B C
B C
B 1 C
B C
E (i( )) = B
B
C,
C
B C
B 1 C
B C
B .. C
@ . A
1

then multiplication on the left by E (i( )) is equivalent to multiplying to

the ith row of A;

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 46 / 64
 (3) Adding times of jth row to ith row of E :
0 1
1
B .. C
B . C
B C
B C
B 1 ··· C
B C
B .. . C
E (i, j( )) = B . .. C,
B C
B 1 C
B C
B .. C
B . C
@ A
1

then multiplication on the left by E (i, j( )) to A is equivalent to adding

times of jth row to ith row of A.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 47 / 64
 Property
Suppose A is a m ⇥ n matrix. Then applying a elementary row operation
on A is equivalently to multiplying the corresponding elementary matrix on
the left to A.

Obviously all elementary matrices are invertible and

E (i, j) 1 = E (i, j), E (i( )) 1 = E (i( 1 )), E (i, j( )) 1 = E (i, j( )).

Corollary
The square matrix A is invertible if and only if there exist finite elementary
matrices P1 , · · · , Pr such that A = P1 · · · Pr .

Corollary
The square matrix A is invertible if and only if A ⇠ E .

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 48 / 64
 Theorem
Let A be a square matrix. Then following conditions are equivalent:
(1) A is invertible;
(2) AX = 0 has only the trivial solution;
(3) A ⇠ E .

Corollary
Let A be a square matrix. Then AX = b has a unique solution if and only
if A is invertible.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 49 / 64
 A method to calculate the inverse

If A ⇠ B, then there exists an invertible matrix P such that PA = B.

Therefore
P(A, E ) = (B, P) , (A, E ) ⇠ (B, P).

To get P, it is suffice to apply elementary operations on (A, E ) since when

A become B, E become P. Especially when B = E , P = A 1 .

Example
0 1
0 2 1
B C 1.
Suppose A = @ 3 0 2 A . Then compute A
2 3 0

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 50 / 64
 Solution

(A, E )
0 1 r1 $ r2 0 1
0 2 1 1 0 0 3 0
r3 ⇥ 3 2 0 1 0
B C B r3 + 2r1 C
=@ 3 0 2 0 1 0 A ⇠ @ 0 2 1 1 0 0 A
2 3 0 0 0 1 0 9 4 0 2 3
0 1 0 1
r3 ⇥ 2 3 0 2 0 1 0 r1 + 2r3 3 0 0 18 9 12
r3 + 9r2 B C r2 r3 B C
⇠ @ 0 2 1 1 0 0 A ⇠ @ 0 2 0 8 4 6 A
0 0 1 9 4 6 0 0 1 9 4 6
0 1
r1 ⇥ 13 1 0 0 6 3 4
r2 ⇥ 12 B C
⇠ @ 0 1 0 4 2 3 A
0 0 1 9 4 6

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 51 / 64
 0 1
6 3 4
1 B C
Hence A = @ 4 2 3 A.
9 4 6

Example
8
>
> x x2 x3 = 2,
>
< 1
Solve linear system 2x1 x2 3x3 = 1,
>
>
>
:3x1 + 2x2 5x3 = 0.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 52 / 64
 Solution
0 1 0 1
1 1 1 2 r2 2r1 1 1 1 2
B C r3 3r1 B C
(A, b) = @ 2 1 3 1 A ⇠ @ 0 1 1 3 A
3 2 5 0 0 5 2 6
r1 + r2 0 1 0 1
r3 5r2 1 0 2 1 r1 + 2r3 , 1 0 0 5
r3 ⇥ 13 B C r2 + r3 B C
⇠ @ 0 1 1 3 A ⇠ @ 0 1 0 0 A
0 0 1 3 0 0 1 3
1 0
5
B C
Hence the linear system has the only solution x = A 1 b = @ 0 A .
3

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 53 / 64
 Example
0 1 0 1
2 1 3 1 1
B C B C
Suppose A = @ 1 2 2 A,B = @ 2 0 A. Solve equation
1 3 2 2 5
AX = B.

Solution

(A, B)
0 1 r1 $ r2 0 1
2 1 3 1 1 r2 2r1 1 2 2 2 0
B C r3 + r1 B C
= @ 1 2 2 2 0 A ⇠ @ 0 3 1 3 1 A
1 3 2 2 5 0 5 0 0 5
r3 ⇥ 15 0 1 0 1
r2 $ r3 1 2 2 2 0 1 0 0 4 2
r3 + 3r2 B C r1 2r2 + 2r3 B C
⇠ @ 0 1 0 0 1 A ⇠ @ 0 1 0 0 1 A
0 0 1 3 2 0 0 1 3 2
Youming Chen([email protected]) Chongqing University
Linearof algebra
Technology () October 9, 2020 54 / 64
 1 0
4 2
B C
Therefore X = A 1 B = @ 0 1 A .
3 2

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 55 / 64
 Example
!
a b
Discuss when A = is invertible?
c d

Solution
Without loss out generality we assume that a 6= 0.

(A, E )
! r1 ⇥ 1a !
b 1
a b 1 0 r2 c · r1 1 a a 0
= ⇠ ad bc c
c d 0 1 0 a a 1

Hence only when ad bc 6= 0, A is invertible. In this case,

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 56 / 64
 r2 ⇥ ad a bc !
d b
r1 b ·r
a 2 1 0 ad bc ad bc
(A, E ) ⇠ c a
.
0 1 ad bc ad bc

Hence when and only when ad 6 0, A is invertible, and its inverse

bc =
! !
d b
1 ad bc ad bc 1 d b
A = c a
= · .
ad bc ad bc
ad bc c a

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 57 / 64
 Homework

Page 68-69, 3(a), 7(a), 10(f).

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 58 / 64
 1.6 Partitioned Matrices(©¨›!)

(1) If both A and B are m ⇥ n matrices with the same Partition

0 1 0 1
A11 · · · A1t B11 · · · B1t
B . .. C B .. C
A=B .. C , B = B ... . C
@ . A @ A,
As1 · · · Ast Bs1 · · · Bst

where Aij and Bij are mi ⇥ nj matrices, then

0 1
A11 + B11 · · · A1t + B1t
B .. .. C
A+B =B @ . . C;
A
As1 + Bs1 · · · Ast + Bst

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 59 / 64
 0 1
A11 · · · A1t
B . .. C
B
(2) If A = @ .. . C
A, is a constant, then
As1 · · · Ast
0 1
A11 · · · A1t
B .. .. C
A=B
@ . . CA;
As1 · · · Ast

(3) If A is a l ⇥ m matrix, B is a m ⇥ n matrix, A and B are partitioned by

0 1 0 1
A11 · · · A1t B11 · · · B1r
B . .. C B .. C
A=B .. C , B = B ... . C
@ . A @ A,
As1 · · · Ast Bt1 · · · Btr

where the column number of Aik equates the row number of Bkj , then
Youming Chen([email protected]) Chongqing University
Linearof algebra
Technology () October 9, 2020 60 / 64
 0 1
C11 · · · C1t
B . .. C
AB = B
@ .
. . C
A,
Cs1 · · · Cst
Pt
where Cij = Aik Bkj ;
k=1
0 1 0 1
A11 · · · A1t AT
11 · · · ATs1
B . .. C B . .. C
(4) If A = B
@ .
. . C , then AT = B ..
A @ . CA;
As1 · · · Ast AT
1t · · ·
T
Ast

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 61 / 64
 (5) If A is a square matrix and it could be partitioned as
0 1
A1 O
B C
B A2 C
A=B
B ..
C,
C
@ . A
O As

where all Ai are square matrix, then A is invertible if and only if all Ai
are square matrices, and
0 1
A1 1 O
B C
B A2 1 C
A =B
1
B ..
C.
C
@ . A
O As 1

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 62 / 64
 Example
1 0
5 0 0
B C
Compute the inverse of A = @ 0 3 1 A.
0 2 1

Solution
!
A1 O
Partition A = , then
O A2
! !
3 1 1 1
A1 = (5), A1 1 = ( 15 ), A2 = , A2 1 = . Hence
2 1 2 3
0 1
1
5 0 0
1 B C
A =@ 0 1 1 A.
0 2 3

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 63 / 64
 Homework

Page 79-80, 13, 15, 16.

Youming Chen([email protected]) Chongqing University

Linearof algebra
Technology () October 9, 2020 64 / 64