01 Linear Algebra 1 - ETL PDF

CHAPTER 7.
LINEAR ALGEBRA
2015.3
2012 .
Engineering Math, 7. Linear Algebram, Spring 2015
Seoul
National
Univ.
Contents
7.1 Matrices, Vectors : Addition and Scalar Multiplication
7.2 Matrix Multiplication
7.3 Linear Systems of Equations. Gauss Elimination
7.4 Linear Independence. Rank of Matrix. Vector Space
7.5 Solutions of Linear Systems : Existence, Uniqueness
7.6 Second- and Third-Order Determinants

7.7 Determinants. Cramers Rule
7.8 Inverse of a Matrix. Gauss-Jordan Elimination
Seoul
National
Univ.
7.1 MATRICES, VECTORS :

ADDITION AND SCALAR
MULTIPLICATION
Seoul
National
Univ.
Matrices
Matrix : a rectangular array of numbers (or functions)
Entries
enclosed in brackets.
(or sometimes the elements)
Examples)
1
5
0.3
0 0.2 16 ,
e
6x
e
rows
2x
4x
2
a11 a12
a
a
21
22
a31 a32
a13
a23 ,
a33
columns
square matrices
(no. of rows = no. of columns)

Seoul
National
Univ.
Matrices
We shall denote matrices by capital boldface letters A, B, C, , or by writing
the general entry in brackets ; thus A=[ajk], and so on. By an m x n matrix (read
m by n matrix) we mean a matrix with m rows and n columns rows come
always first!
m x n is called the size of the matrix. Thus an m x n matrix is of the form
a11
a
21
A [a jk ]

am1
a12
a22
am 2
a1n
a2 n
(2).

amn
Seoul
National
Univ.
a11
a
A [a jk ] 21

am1
Matrices
a12
a22
am 2
a1n
a2 n

amn
If m = n, we call A an n x n square matrix.
a11 a12
a
a
21
22
an1 an 2
a1n
a2 n

ann
main diagonal
If m n, we call A a rectangular matrix.

Seoul
National
Univ.
Vectors
Matrix : a rectangular array of numbers (or
functions) enclosed in brackets.
Vector : a matrix with only one row or column.
We shall denote vectors by lowercase boldface
letters a, b, or by its general component in
brackets, a=[ai] , and so on.
Examples)
column vector
row vector
a a1 a2
Components (entries)
a3 ,
b1
b b2
b3
Seoul
National
Univ.
Matrices
Ex 1) Linear Systems, a Major Application of Matrices
In a system of linear equations, briefly called a linear system, such as
4 x1 6 x2 9 x3 6
6 x1
2 x3 20
5 x1 8 x2 x3 10
the coefficients of the unknowns x1, x2, x3 are the entries of the coefficient
matrix, call it A,
9
4 6
A 6 0 2 .
5 8 1
The matrix
9 6
4 6
~
A 6 0 2 20
5 8 1 10
is called augmented matrix() of the system.
We shall discuss this in great detail, beginning in Sec. 7.3

Seoul
National
Univ.
Matrices
Ex 2) Sales Figures in Matrix Form
Sales Figures for Three products I, II, III in a store on Monday (M),
Tuesday (T), may for each week be arrange in a matrix.
Th
400 330 810 0 210 470 I
A 0 120 780 500 500 960 II .

100 0
0 270 430 780 III
If the company has ten stores, we can set up ten such matrices, on for
each store. Then by adding corresponding entries of these matrices we
can get a matrix showing the total sales of each product on each day.
Seoul
National
Univ.
Matrix Addition and Scalar Multiplication

Equality of Matrices
Two matrices A=[ajk] and B=[bjk] are equal, written A=B, if and only if they have
the same size and the corresponding entries are equal, that is, a11=b11, a12=b12,
and so on. Matrices that are not equal are called different. Thus, matrices of
different sizes are always different.
Ex3) Equality of Matrices
a11 a12
4 0
Let A
and B
a
a
3
21 22
Then A B if and only if
a11 4,
a12 0,
a21 3, a22 1.
Seoul
National
Univ.
10

Addition of Matrices
The sum of two matrices A=[ajk] and B=[bjk] of the same size is written A+B and
has the entries ajk + bjk obtained by adding the corresponding entries of A and B.
Matrices of different sizes cannot be added.
Ex4) Addition of Matrices and Vectors
If
4 6 3
5 1 0
1 5 3
A
and B
, then A B
.
0 1 2
3 1 0
3 2 2
If a 5 7 2 and b 6 2 0, then a b 1 9 2.
Seoul
National
Univ.
11

Scalar Multiplication (Multiplication by a Number)
The product of any m x n matrix A=[ajk] and any scalar c (number c) is written cA
and is the m x n matrix cA=[cajk] obtained by multiplying each entry of A by c.
Ex5) Scalar Multiplication
If
2.7 1.8
2.7 1.8
A 0
0.9 , then -A 0
0.9
zero matrix
9.0 4.5
9.0 4.5
3 2
0 0
10
A 0 1 , 0 A 0 0
9
10 5
0 0
Seoul
National
Univ.
12
0 : zero matrix (of size m x n matrix)
Rules for Matrix Addition and Scalar Multiplication
AB BA
(a)
(b) ( A B) C A (B C) written A B C
(3)
A0 A
(c)
(d )
A ( A) 0
(a) c( A B) cA cB
(b) (c k ) A cA kA
(4)
c(k A) (ck ) A written ckA
(c)
(d )
1A A
Seoul
National
Univ.
13
7.2 MATRIX MULTIPLICATION
Seoul
National
Univ.
14
Multiplication of a Matrix by a Matrix

n
c jk a jl blk a j1b1k a j 2b2 k a jnbnk (1)

l 1
p2
n3
m4
a11
a
21
a31
a41
a12
a22
a32
a42
a13
a23
a33
a43
b11
b
21
b31
A
B C
[m n] [n r ] [m r ]
j 1,, m
k 1,, p.
p2
c11 c12
b12
c
21
22
b22 (r 3)
m 4
c31 c32
b32
c
c
41 42
c21 a21b11 a22b21 a23b31
Seoul
National
Univ.
15
A
B C
[m n] [n r ] [m r ]
Ex 1) Matrix Multiplication
5 1 2 2 3 1 22 2 43
42
3
AB 4
0
2 5 0 7 8 26 16 14
6
6 3 2 9 4 1 1 9 4 37 28
c11
3 2 5 5 (1) 9 22
c23
4 3 0 7 2 1 14
The product BA is not defined.
Seoul
National
Univ.
16
[m n] [n r ] [m r ]
Ex 1) Matrix Multiplication
4 2 3 4 3 2 5 22
1 8 5 1 3 8 5 43

3
5

4 2
1 8
is not defined.
Ex3) Products of Row and Column Vectors
1
3 6 1 2 19
4
1
3 6 1
2 3 6 1 6 12 2.

12 24 4
4
Seoul
National
Univ.
17
A
B C
[m n] [n r ] [m r ]
Ex 4) CAUTION! Matrix Multiplication is Not Commutative (),

ABBA in General
1
1
1 1
A
,B
100
100
1
1 1 1 0 0
1
AB
100 100 1 1 0 0
1 1
BA
1 99
99
1
100 100 99 99
AB BA
It is interesting that this also shows that AB=0 does not necessarily imply BA=0 or A=0 or B=0.
Seoul
National
Univ.
18
Properties of Matrix Multiplication
(kA)B k ( AB ) A(kB) written kAB or AkB

(a)
(b)
written ABC
A(BC) ( AB )C
(2)
(c) ( A B)C AC BC
(d ) C( A B) CA CB
(2b) is called the associative law ()
(2c) and (2d) is called the distributive law ()
Seoul
National
Univ.
19
A
B C
[m n] [n r ] [m r ]
c jk a jl blk a j1b1k a j 2b2 k a jnbnk (1)

l 1
Since matrix multiplication is a multiplication of rows into columns, we can write the
defining formula (1) more compactly as
c jk a j b k (3)
j 1,, m; k 1,, p.
where aj : the jth row vector of A

bk : the kth column vector of B, so that in agreement with (1),
a j b k a j1
a j 2 a jn
b1k
a b a b a b
j1 1k
j 2 2k
jn nk

bnk
Seoul
National
Univ.
20

Ex 5) Product in Terms of Row and Column Vectors
If A=[ajk] is of size 3x3 and B=[bjk] is of size 3x4, then
a1b1 a1b 2
AB a 2b1 a 2b 2
a3b1 a3b 2
a1b3
a 2b 3
a3b3
a1b 4
a 2b 4
a3b 4
(4).
a1
A a 2 , B b1 b 2
a3
b3 b 4
Parallel processing of products on the computer is facilitated by a variant of (3)

for computing C=AB, which is used by standard algorithms.
In this method, A is used as given, B is taken in terms of its column vectors,
and the product is computed columnwise; thus,
AB A b1 b 2 b p Ab1
Ab 2 Ab p (5).
Seoul
National
Univ.
21
c jk a j bk

Ex 5) Product in Terms of Row and Column Vectors
a1b1 a1b 2
AB a 2b1 a 2b 2
a3b1 a3b 2
a1b3
a 2b 3
a3b3
a1b 4
a 2b 4
a3b 4
AB A b1 b 2 b p Ab1
(4).
Ab 2 Ab p (5).
Columns of B are then assigned to different processors (individually or

several to each processor), which simultaneously compute the
columns of the product matrix Ab1, Ab2, etc.
Seoul
National
Univ.
22

Ex 6) Computing Products Columnwise by (5)
AB A b1 b 2
b p Ab1 Ab 2
Ab p
(5).
4 1 3 0 7 11 4 34
AB
5
2
1
4
6
17
8
23
To obtain AB from (5), calculate the columns
4 1 3 11
5 2 1 17,
4 1 0 4
5 2 4 8,
4 1 7 34
5 2 6 23
of AB and then write them as a single matrix, as shown in the fist formula on
the right.
Seoul
National
Univ.
23
Applications of Matrix Multiplication

Ex 11) Computer Production. Matrix Times Matrix
Supercomp Ltd produces two computer models PC1086 and PC1186.
Matrix A : the cost per computer (in thousands of dollars)
Matrix B : the production figures for the year 2005 (in multiples of 1000 units)
Find a matrix C that shows the shareholders the cost per quarter (in millions of
dollars) for raw material, labor, and miscellaneous.
PC1086 PC1186
1.2 1.6 Raw Components
A 0.3 0.4 Labor

0.5 0.5 Miscellaneous
Quarter
1 2 3 4
3 8 6 9 PC1086
B
6
2
4
3
PC1186
Seoul
National
Univ.
24

Ex 11) Computer Production. Matrix Times Matrix
PC1086 PC1186
1.2 1.6 Raw Components

A 0.3 0.4 Labor
0.5 0.5 Miscellaneous
Quarter
1 2 3 4
3 8 6 9 PC1086
B
PC1186
6
2
4
3
Q?
Seoul
National
Univ.
25

Ex 12) Weight Watching. Matrix Times Vector
A weight-watching program
A person of 185lb burns 350 cal/hr in walking (3 mph)
500 cal/hr in bicycling (13 mph)
950 cal/hr in jogging (5.5 mph)
Bill, weighing 185 lb, plans to exercise according to the matrix shown.
Verify the calculations (W = Walking, B = Bicycling, J=Jogging).
W
MON 1.0
0.5
825
1.0 1.0 0.5 350 1325
WED
500
1000
FRI 1.5
0 0.5
950
2400
SAT 2.0 1.5 1.0
MON
WED
FRI
SAT
Seoul
National
Univ.
26

Ex 13) Markov Process. Powers of a Matrix. Stochastic Matrix
Suppose that the 2004 state of land use in a city of 60 mi2 of built-up area is
C : Commercially Used 25 % ,
I : Industrially Used 20 %
R : Residentially Used 55 %.
Transition probabilities for 5-year intervals : A and remain practically the same
over the time considered.
Find the states 2009, 2014, 2019
From C
To C
A:
To I
To R
0.7
0.2
0.1
From I
0.1
0.9
0
From R
0.2
0.8
Seoul
National
Univ.
27

A : stochastic matrix() : a square matrix
all entries nonnegative
all column sums equal to 1
Markov process : the probability of entering a certain state depends only on the
last state occupied (and the matrix A), not on any earlier state.
From C
To C
solution.
A:
To R
area
C
I
R
To I
0.7
0.2
0.1
From I
0.1
0.9
0
From R
0
0.2
0.8
0.7 25 0.1 20 0 55 0.7 0.1 0 25 19.5

0.2 25 0.9 20 0.2 55 0.2 0.9 0.2 20 34.0.
0.1 25 0 20 0.8 55 0.1 0 0.8 55 46.5
Seoul
National
Univ.
28
From C
To C
A:
To I
To R
0.7
0.2
0.1
From I
0.1
0.9
0
From R
0
0.2
0.8

area
C
I
R
0.7 25 0.1 20 0 55 0.7 0.1 0 25 19.5

0.2 25 0.9 20 0.2 55 0.2 0.9 0.2 20 34.0.

0.1 25 0 20 0.8 55 0.1 0 0.8 55 46.5
We see that the 2009 state vector (y) is the column vector
19.5
25
y 34.0 Ax A 20
46.5
55
The vector (x) is the given 2004 state

vector
The sum of the entries of y is 100%.

Similarly, you may verify that for 2014 and 2019 we get the state vectors.
Q?
Seoul
National
Univ.
29
a11
a
A 21

am1
Transposition
Transposition of Matrices and Vectors
a12
a22
am 2
a1n
a2 n

amn
The transposition of an m x n matrix A=[ajk] is the n x m matrix AT (read A

transpose)
: the first row of A as its first column
the second row of A as its second column, and so on.
Thus the transpose of A in (2) is AT=[akj] , written out
As a special case, transposition converts row vectors to column vectors and
conversely
a11
a
A T [akj ] 12

a1n
a21 am1
a22 am 2
(9)

a2 n amn
Seoul
National
Univ.
30
Transposition
Ex 7) Transposition of Matrices and Vectors
5 4
5 8 1
- 8 0
T
If A
,
then
A
4
0
0
1 0
Comparing this A little more compactly, we can write
5 4
5 8 1
- 8 0 ,
4 0 0
1 0
T
6
6 2 3T 2,
3
3 0
3 8
8 - 1 0 - 1,
6
2 6 2 3

3
Seoul
National
Univ.
31
Rules for transposition
( a )
( A T )T A
T
T
T
( A B) A B
(b)
T
T
(
c
)
(
c
A
)
c
A
(d ) ( AB)T BT A T .
Caution! Note that in (d) the transposed matrices are in reversed order.
Seoul
National
Univ.
32
Special Matrices
Symmetric matrix
AT A, (thus akj a jk )
Skew-Symmetric matrix
AT A, (thus akj a jk , hence a jj 0).
Ex 8) Symmetric and Skew-Symmetric Matrices

20 120 200
A 120 10 150
200 150 30
Symmetric matrix
00 1 3
B 1 0 2
3 2 0
Skew-Symmetric matrix
Seoul
National
Univ.
33
Triangular Matrices
Ex 9) Upper and Lower Triangular Matrices
1 3
0 2 ,
1 4 2
0 3 2 ,
0 0 6
Upper Triangular Matrices
0
2
8 1
7
6
0
0,
8
Lower Triangular Matrices
3 00
9 3
1 0
1 9
0 0
0 00
.
2 0
3 6
Upper triangular matrices

square matrices
nonzero entries only on and above the main diagonal
any entry below the diagonal must be zero.
Lower triangular matrices
nonzero entries only on and below the main diagonal
any entry on the main diagonal of a triangular matrix may be zero or not.
Seoul
National
Univ.
34
Diagonal Matrices
Ex 10) Diagonal Matrix D. Scalar Matrix S. Unit Matrix I.
2 0 0
D 0 3 0,
0 0 0
c 0 0
S 0 c 0,
0 0 c
1 0 0
I 0 1 0.
0 0 1
Diagonal Matrices
square matrices
nonzero entries only on the main diagonal
any entry above or below the main diagonal must be zero.
Seoul
National
Univ.
35
Diagonal Matrices
2 0 0
D 0 3 0,
0 0 0
c 0 0
S 0 c 0,
0 0 c
1 0 0
I 0 1 0.
0 0 1
S : scalar matrix,
AS SA cA(12).
I : unit matrix (or identity matrix)
a scalar matrix whose entries on the main diagonal are all 1
AI IA A(13).
Seoul
National
Univ.
36
REFERENCE COLUMN PICTURE AND LINEAR

EQUATIONS*
*Strang
G., Introduction
Engineering Math, 7.
Linear Algebram,
Spring 2015to
Linear Algebra, Third edition, Wellesley-Cambridge Press, 2003, Ch.2.1, p21
Seoul
National
Univ.
37
Vectors and Linear Equations

The central problem of linear algebra is to solve a system of equations.
x 2y 1
First example)
3x 2 y 11
3x 2 y 11
Slopes are important in calculus and
this is linear algebra.
1 x 2y 1
The point x = 3, y = 1 lies on both lines.
3 11
This is the solution to our system of linear

equation.
solution of first equation

solution of second equation
The row picture show two lines meeting

at a single point.
Seoul
National
Univ.
38

Linear system can be a vector equation. If you separate the original
system into its columns instead of its rows, you get
This has two column vectors on

the left side. The problem is to
find the combination of those
vectors that equals the vector on
the right.
1
2 1
x y b
3
2 11
1
11

3
9

We are multiplying the first column by x and the second column by y,

and adding. With the right choices x = 3, y = 1, this produces 3
(column 1) + 1 (column 2) = b.
The column picture combines the column vectors on the

left side to produce the vector b on the right side.
2
2

1
3

3x 2 y 11
x 2y 1
Column picture : A combination of columns

produces the right side (1, 11).
1 x 2y 1
3 11
3x 2 y 11
3
Seoul
National
Univ.
39
x 2y 1

1
2 1
x y b
3
2 11
3 x 2 y 11
x 3, y 1
The left side of the vector equation is a linear combination of the

columns.
The problem is to find the right coefficients x = 3 and y = 1.
We are combining scalar multiplication and vector addition into one
step. :
1 2 1
Linear combinatio n 3 1
3 2 11
Seoul
National
Univ.
40

- Three Equations in Three Unknowns
The three unknowns x, y, z. The linear equations Ax = b are
x 2 y 3z 6
2x 5 y 2z 4
6x 3y z 2
The row picture show three planes meeting at a single point.
L
6x 3 y z 2
x 2 y 3z 6
L
0
0

2
2x 5 y 2z 4
x
The usual result of two equations in
three unknowns is a intersect line L of
solutions.
The third equation gives a third plane. It cuts

the line L at a single point. That point lies on
all three planes and it solves all three
equations.
Seoul
National
Univ.
41

- Three Equations in Three Unknowns
x 2 y 3z 6
2x 5 y 2z 4
6x 3y z 2
The column picture starts with the vector form of the equations :
1
2
3 6
x 2 y 5 z 2 4
6
3
1 2
z
1
2 column 1

6
3
2

1
6
b 4 2 times
2
2
5 column 2

3
column 3
C The column picture combines three columns to

produce the vector (6,4,2)
Figure 2.4 Column picture : (x, y, z) = (0, 0, 2)
because 2(3, 2, 1) = (6, 4, 2) = b.
The coefficient we need are x = 0, y = 0 and z = 2.

This is also the intersection point of the three
planes in the row picture.
Seoul
National
Univ.
42
x 2 y 3z 6

- The Matrix Form of the Equations
Matrix equation :
2x 5 y 2z 4
6x 3y z 2
(3)
2 3 x 6
1
Ax 2
5 2 y 4 b
6 3 1 z 2
Coefficient matrix
unknown vector
We multiply the matrix A times the unknown vector x to get the right side
b.
Multiplication by rows : Ax
Multiplication by columns : Ax is
comes from dot products, each
a combination of column
row times the column x :
vectors :
(row 1) x
Ax (row 2) x.
(row 3) x
Ax x (column 1)
y (column 2) z (column 3)
Seoul
National
Univ.
43

- The Matrix Form of the Equations
2 3 x 6
1
Ax 2
5 2 y 4 b
6 3 1 z 2
Ax x (column 1) y (column 2) z (column 3)

When we substitute the solution x = (0, 0, 2), the multiplication Ax
produces b :
2 3 0
1
6
2
0 2 times column 3 4.
5
2

6 3 1 2
2
The first dot product in row multiplication is (1, 2, 3) (0, 0, 2) = 6. The
other dot products are 4 and 2.
Multiplication by columns is simply 2 times column 3.
Ax as a combination of the columns of A.
Seoul
National
Univ.
44
7.3 LINEAR SYSTEMS OF

EQUATIONS. GAUSS
ELIMINATION
Seoul
National
Univ.
45
Linear System, Coefficient Matrix, Augmented Matrix

A linear system of m equations in n unknowns x1, , xn is a set of equations of
the form
a11x1 a12 x2 a1n xn b1
given :
(1) a , a
11
1n
b ,b
am1 x1 am 2 x2 amn xn bm
m
1
a21x1 a22 x2 a2 n xn b2
coefficients
If b1==bm=0, homogeneous system,

else nonhomogeneous system
The system is called linear because each variable xj appears in the first power
only, just as in the equation of a straight line.
Seoul
National
Univ.
46

a21x1 a22 x2 a2 n xn b2
(1)
A solution of (1) is a set of numbers x1, , xn that satisfies all the m equations.
A solution vector of (1) : a vector x whose components form a solution of (1).
If the system (1) is homogeneous the trivial solution x1=0, , xn=0.
Matrix Form of the Linear system (1).
Ax b(2)
Seoul
National
Univ.
47
Ax b(2)

a21x1 a22 x2 a2 n xn b2
the coefficient matrix A=[ajk] : m x n matrix
a11
a
21
am1
m
a12
a22
am 2
x1
a1n
a2 n
, and x , and b ,

bm
amn
xn
A is not a zero matrix.

x has n components, whereas b has m components.
Seoul
National
Univ.
48

Augmented matrix()
a11
a
~ 21
A

am1
a21x1 a22 x2 a2 n xn b2
a12
a22
am 2
a1n b1
a2 n

amn bm
(1)
The dashed vertical line could be omitted.

The last column of does not belong to A.
The augmented matrix determines the system (1) completely because it
contains all the given numbers appearing in (1).
Seoul
National
Univ.
49

Ex 1) Geometric Interpretation. Existence and Uniqueness of Solutions
(a) precisely one solution if the lines intersect
If m = n =2, we have two equations in two

unknowns x1, x2
x2
x1 x2 1
a11x1 a12 x2 b1
a21x1 a22 x2 b2
If we interpret x1, x2 as coordinate in the x1 x2 plane,
each of the two equations represents a
straight line,
(x1, x2) is a solution if and only if the point P
with coordinates x1, x2 lies on both lines.
2 x1 x2 0
1
x1
1
3
(b) Infinitely many solutions if the lines coincide
x2
x1 x2 1
2 x1 2 x2 2
x1
(c) No solution if the lines are parallel
Hence there are three possible cases :
x2
x1 x2 1
x1 x2 0
If the system is homogenous case (c) cannot

happen
1
x1
Seoul
National
Univ.
50

(b) Infinitely many solutions if the lines
coincide
x2
x1 x2 1
2 x1 2 x2 2
1
x1 : independent variable
or design variable,
x2 : dependent variable.
x1
Changing the form of the equation as
x2 1 x1
if we choose a value of x1, then the

corresponding values of x2 is uniquely
determined.
There would be many solutions,

so we choose the best solution to obtain proper
result.
Seoul
National
Univ.
51

p1 : a11 x1 a12 x2 a13 x3 b1
Three equations in three unknowns interpreted as

planes in space
p2 : a21 x1 a22 x2 a23 x3 b2

p3 : a31 x1 a32 x2 a33 x3 b3
p3
p2
p3
p3
p2
p1
p2
p3
p2
p1
p1
Unique solution
p1
Infinitely many solutions
No solution
p2
p2
p2
p2
p3
p1
p3
p3
p1
p3
p1
p1
No solution
Seoul
National
Univ.
52
Gauss Elimination and Back Substitution

a standard elimination method for solving linear systems
If a system is in triangular form, we can solve it by back
substitution.
Triangular Matrices
1 4 2 2 0
3 2 , 8 1
0

0 0 6 7
6
0
8
Seoul
National
Univ.
53

x1 + 2x2 + x3 = 1
3x1 - x2 - x3= 2
row2 +
row1x(-3)
2x1+3x2 - x3= -3
row2 +
row1x(-3)
1 2 1 x1 1
3 1 1 x 2
2
2 3 1 x3 3
3x1 - x2 - x3= 2
-3x1-6x2-3x3=-3
-7x2 -4x3= -1
x1 + 2x2 + x3 = 1
-7x2 -4x3 = -1
2x1+3x2 - x3 = -3
1 2 1 x1 1
30 71 14 x 21
2
2 3 1 x3 3
Seoul
National
Univ.
54

1 x1 1
1 2
3 1 1 x 2
2
2 3 1 x3 3
x1 + 2x2 + x3 = 1
3x1 - x2 - x3 = 2
2x1 + 3x2 - x3 = -3
1 x1 1
1 2
0 7 4 x 1
2
2 3 1 x3 3
x1 + 2x2 + x3 = 1
-7x2 -4x3 = -1
2x1+3x2 - x3 = -3
row3 +
row1x(-2)
2x1+3x2 - x3 = -3
row3 +
row1x(-2)
-2x1 -4x2 -2x3 =-2
x1 + 2x2 + x3 = 1
- 7x2 - 4x3 = -1
- x2 - 3x3 = -5
-x2 -3x3= -5
1 x1 1
1 2
0 7 4 x 1
2
20 31 31 x3 53
Seoul
National
Univ.
55

1 x1 1
1 2
3 1 1 x 2
2
2 3 1 x3 3
x1 + 2x2 + x3 = 1
3x1 - x2 - x3 = 2
2x1 + 3x2 - x3 = -3
x1 + 2x2 + x3 = 1
-7x2 -4x3 = -1
-x2 -3x3 = -5
row 2 row 3
x1 + 2x2 + x3 = 1
1 x1 1
1 2
0 7 4 x 1
2
0 1 3 x3 5
1 x1 1
1 2
0 1 3 x 5
2
0 7 4 x3 1
-x2 -3x3 = -5
-7x2 -4x3 = -1
row 2x(-1)
x1 + 2x2 + x3 = 1
x2 +3x3 = 5
-7x2 -4x3 = -1
1 x1 1
1 2
0 1
x 5
3
2
0 7 4 x3 1
Seoul
National
Univ.
56

1 x1 1
1 2
3 1 1 x 2
2
2 3 1 x3 3
x1 + 2x2 + x3 = 1
3x1 - x2 - x3 = 2
2x1 + 3x2 - x3 = -3
1 x1 1
1 2
0 1
x 5
3
2
0 7 4 x3 1
x1 + 2x2 + x3 = 1
x2 +3x3 = 5
-7x2 -4x3 = -1
row 3+ row 2x7
x1 + 2x2 + x3 = 1
x2 +3x3 = 5
17x3 = 34
1 2 1 x1 1
0 1 3 x 5
2
0 0 17 x3 34
The last equations and matrix are equal to given equations.
Seoul
National
Univ.
57

x1 + 2x2 + x3 = 1
x1
x2 +3x3 = 5
x2
x3
17x3 = 34
34
x3
2
17
1 2 1 x1 1
0 1 3 x 5
2
0 0 17 x3 34
Back substitution
x2 3x3 x2 3 2 5
x2 1
x1 2 x2 x3 x1 2 (1) 2 1
x1 1
Seoul
National
Univ.
58

Since a linear system is completely determined by its augmented matrix,
Gauss elimination can be done by merely considering the matrices.
augmented matrix
1 2 1 x1 1
3 1 1 x 2
2
2 3 1 x3 3
1 2 1 x1 1
0 1 3 x 5
2
0 0 17 x3 34
1 2 1
3 1 1
2 3 1
1
2
3
1 2 1
0 1 3
0 0 17
1
5
34
Seoul
National
Univ.
59
Elementary Row Operations.

Row-Equivalent System
Elementary Row Operations for Matrices :

1) Interchange of two rows
2) Addition of a constant multiple of one row to another row
3) Multiplication of a row by a nonzero constant c
Elementary Operations for Equations :

1) Interchange of two equations
2) Addition of a constant multiple of one equation to another equation

3) Multiplication of an equation by a nonzero constant c
Seoul
National
Univ.
60
Elementary Row Operations.

Row-Equivalent () System
Solution set of S1
S1
Same
Row Operations
S2
a linear system S1
row-equivalent to a
linear system S2
Solution set of S2
: elimination : Solution of S2 found : Solution of S1 found

goal
The interchange, addition and multiplication of two equations does not alter
the solution set. Because we can undo it by a corresponding subtraction.
A linear system S1 row-equivalent to a linear system S2 if S1 can be obtained
from S2 by row operations.
Because of this theorem, systems having the same solution sets are often
called equivalent systems.
No column operations on the augmented matrix are permitted because they
would generally alter the solution set.
Seoul
National
Univ.
61
Gauss Elimination :
The Three Possible Cases of Systems
case 1 : Gauss Elimination if Infinitely Many Solutions Exist
three equations < four unknowns
2.0 5.0 8.0

3.0 2.0
0.6 1.5
1
.
5
5
.
4
2
.
7
1.2 0.3 0.3 2.4 2.1

3.0 2.0 2.0 5.0 8.0
0 1.1 1.1 4.4 1.1
0 1.1 1.1 4.4 1.1
3.0 x1 2.0 x2 2.0 x3 5.0 x4 8.0

0.6 x1 1.5 x2 1.5 x3 5.4 x4 2.7
1.2 x1 0.3x2 0.3x3 2.4 x4 2.1
Row2-0.2*Row1
Row3-0.4*Row1
3.0 x1 2.0 x2 2.0 x3 5.0 x4 8.0

1.1x2 1.1x3 4.4 x4 1.1
1.1x2 1.1x3 4.4 x4 1.1
Row3+Row2
3.0 2.0 2.0 5.0 8.0

0 1.1 1.1 4.4 1.1
0
0
0
0
0
3.0 x1 2.0 x2 2.0 x3 5.0 x4 8.0

1.1x2 1.1x3 4.4 x4 1.1
00
Seoul
National
Univ.
62
Gauss Elimination :
case 1 : Gauss Elimination if Infinitely Many Solutions Exist
3.0 2.0 2.0 5.0 8.0

0 1.1 1.1 4.4 1.1
0
0
0
0
0
3.0 x1 2.0 x2 2.0 x3 5.0 x4 8.0

1.1x2 1.1x3 4.4 x4 1.1
00
Back substitution.
From the second equation : x2 1 x3 4 x4
From the first equation : x1 1 x4

Since x3 and x4 remain arbitrary, we have infinitely
many solutions.
If we choose a value of x3 and a value of x4, then the corresponding values of x1
and x2 are uniquely determined.
Seoul
National
Univ.
63
Gauss Elimination :
case 2 : Gauss Elimination if no Solution Exists
3x1 2 x2 x3 3
2 x1 x2 x3 0
6x 2x 4x 6
2
3
1
3 2 1 3
2 1 1 0
6 2 4 6
2
1
3
3
0 1 / 3 1 / 3 2
0 2
2
0
2
1
3
3
0 1 / 3 1 / 3 2
0
0
0 12
Row2-2/3*Row1
Row3-2*Row1
Row3-6*Row3
x3 3
3x1 2 x2
1/ 3x2 1/ 3x3 2
2 x2 4 x3 0
x3 3
3x1 2 x2
1/ 3x2 1/ 3x3 2
0 12
The false statement 0=12 show that the system has no solution.
Seoul
National
Univ.
64
Row Echelon Form and Information From It

Row Echelon Form
At the end of the Gauss elimination (before the back substitution)
the row-echelon form( ) of the augmented matrix will be
a1n
a11 a12
c22
c2 n
Zero
k rr k rn
Zero
b1
~
b2

~
br (8)
~
br 1

~
bm
( r : no. of equations, n : no. of unknowns, m : no. of rows )
Here, r m and a11 0, c22 0, , krr 0, and all the entries in the blue triangle
as well as in the blue rectangle are zero.
Seoul
National
Univ.
65
Row Echelon Form and Information From It

Row Echelon Form
( r : no. of equations, n : no. of unknowns, m : no. of rows )
(a) Exactly one solution

if r = n and br 1 , bm , if present, are zero. no. of equations r = no. of unknowns n
Solve the nth equation corresponding to (8) (which is knnxn=bn) for xn, then the (n-1)st equation for xn-1,
and so on up the line.
(b) Infinitely many solutions
if r < n and br 1 , bm are zero. No. of equations < no. of unknowns
To obtain any of these solutions, choose values of xr-1, , xn arbitrary .
Then solve the rth equation for xr, then the (r-1)st equation for xr-1, and so on up the line.
(c) No solution
if r < m and one of the entries br 1 , bm
no. of equations < no. of unknowns
n3
r 3
n4
0
1 1 1
0 10 25 90
0 0 95 190
0
0
0 0
x1 x2 x3
0 x1 10 x2 25 x3
0 x1 0 x2 95 x3
0 x1 0 x2 0 x3
is not zero.
r2
m4
0
90
190
0
3.0 2.0 2.0 5.0 8.0

0 1.1 1.1 4.4 1.1
0
0
0
0
0
n3
r 2 3
m3
3.0 x1 2.0 x2 2.0 x3 5.0 x4 8.0
0 x1 1.1x2 1.1 x3 4.4 x4 1.1

0 x 0 x 0 x 0 x 0.0
1
2
3
4
3
0 1 / 3 1 / 3 2
0
0
0 12
2
m3
x3 3
3x1 2 x2
1/ 3x2 1/ 3x3 2
0 12
Seoul
National
Univ.
66
Example
Models of markets.
Determine the equilibrium solution (D1=S1, D2=S2) of the twocommodity market with linear model demand, supply, price; index
first commodity, index second commodity)
D1=40-2P1-P2
S1=4P1-P2+4
D2=5P1-2P2+16
S2=3P2 - 4
Seoul
National
Univ.
67
7.4 LINEAR INDEPENDENCE. RANK

MATRIX. VECTOR SPACE
OF
Seoul
National
Univ.
68
Linearly independent vectors
If the point is at the origin, the equation

becomes
j
i
1
0
0

i 0 , j 1 , k 0
0
0
1

We can express the location of the point

with i, j, k.
1 0 0
0
a 0 b 1 c 0 0 0.
0 0 1
0
The equation above is satisfied if and only if
a=b=c=0.
Then, i, j, k are linearly independent.
1 0 0

ai bj ck a 0 b1 c 0
0 0 1

Seoul
National
Univ.
69
Linear Independence and Dependence of Vectors

Given any set of m vectors a(1), , a(m) (with the same number of components),
a linear combination of these vectors is an expression of the form
c1a(1) c2a( 2) cma( m)
c1, c2, , cm are any scalars. Now consider the equation.

When
Vector
c1a(1) c2a( 2) cma( m) 0 (1)

c1 c2 cm 0
a(1) , a(2) , , a( m)
vectors linearly independent set or linearly independent.
Linear Dependence / Independence

Definition 3.1
A set of functions f1 ( x ), f 2 ( x ), , f n ( x )is said to be linearly
dependent on an interval I if there exist constant c1 , c2 ,...cn , not all zero
such that c1 f1 ( x) c2 f 2 ( x) cn f n ( x) 0
for every x in the interval.
If the set of functions is not linearly dependent on the interval, it is said to be
linearly independent
Function
In other words, a set of functions is linearly independent if the only constants for
are
c1 f1 ( x) c2 f 2 ( x) cn f n ( x) 0
c1 c2 cn 0
Seoul
National
Univ.
70
c1a(1) c2a( 2) cma( m) 0 (1)

If (1) also holds with scalars not all zero, we call these vectors linearly dependent, because
then we can express (at least) one of them as a linear combination of the others. For
instance, if (1) holds with, say, c1=0, we can solve (1) for a(1) :
a(1) k2a( 2) kma( m) ,
(where k j c j / c1 )
(Some kjs may be zero. Or even all of them, namely, if a(1)=0.)
Seoul
National
Univ.
71

Ex 1) Linear Independence and Dependence
Vector
Linear Systems
Matrix
0
2
2
3
6 42 24 54
21 21 0 15
0
2
3
6 42 24
21 21 0
0
2
2
3
0 42 28 58
21 21 0 15
0
2
3
0 42 28
21 21 0
2
2
3 0
0 42 28 58
0 21 14 29
2
3 0
0 42 28
0 21 14
3 0 2 2
0 42 28 58
0 0 0 0
3 0 2
0 42 28
0 0 0
2]
3x1 0 x2 2 x3 2
a(2) [6, 42, 24, 54]
6 x1 42 x2 24 x3 54
a(3) [21, 21,
21x1 21x2
a(1) [ 3,
0, 2,
0, 15]
6a(1) [18, 0, 12, 12]

1
a(2) [3, 21, 12, 27]
2
a(3) [21, 21, 0, 15]
1
6a(1) a(2) a(3) [0, 0, 0, 0]
2
0 x3 15
3x1 0 x2 2 x3 2
0 x1 42 x2 28 x3 58
21x 21x 0 x 15
2
3
1
3x1 0 x2 2 x3 2
0 x1 42 x2 28 x3 58
0 x 21x 14 x 29
1
2
3
3x1 0 x2 2 x3 2
0 x1 42 x2 28 x3 58
0 x 0 x 0 x 0
1
2
3
The three vectors are

linearly dependent
x2+
x(-7)+
x(0.5)+
The three equations are

linearly dependent
The three rows are

linearly dependent
Seoul
National
Univ.
72
Rank of a Matrix
Rank of a Matrix
The rank of a matrix A
: the maximum number of linearly independent row vectors of A. rank A.
Ex 2) Rank
The
matrix
0
2
2
3
A 6 42 24 54 (2)
21 21 0 15
has rank 2, because Example 1 shows that the first two row vectors are linearly
independent, whereas all three row vectors are linearly dependent.
Note further that rank A=0 if and only if A=0 (zero matrix).
Seoul
National
Univ.
73
Rank of a Matrix
Example 1
Rank of 3 x 4 Matrix
Consider the 3 x 4 matrix
1 1 1 3
A 2 2 6 8 .
3 5 7 8
With u1=(-1 1 -1 3), u2=(2 -2 6 8), and

u3=(3 5 -7 8) 4u1-1/2u2+u3=0.
the set u1, u2, u3 is linearly dependent.
u1 c u2 u1, u2 is linearly independent.
Hence by Definition, rank(A) = 2.
1 1 1 3
2
6
8
3 5 7 8
x1 x2 x3 3
2 x1 2 x2 6 x3 8
3x 5 x 7 x 8
2
3
1
1 1 1 3
4
8
2
3 5 7 8
x1 x2 x3 3
0 x1 4 x2 8 x3 2
3x 5 x 7 x 8
2
3
1
1 1 1 3
4
8
2
0 2 4 1
x1 x2 x3 3
0 x1 4 x2 8 x3 2
0 x 2 x 4 x 1
2
3
1
x(-2)+
x(-3)+
x(0.5)+
1 1 1 3
4
8
2
0 0 0 0
x1 x2 x3 3
0 x1 4 x2 8 x3 2
0 x 0 x 0 x 0
2
3
1
rank(A) = 2
Seoul
National
Univ.
74
Rank and Row-Equivalent Matrices

Theorem 1 : Row-Equivalent Matrices
Row-equivalent matrices have the same rank.
A1 row-equivalent to a matrix A2
rank is invariant under elementary row operations.
1 1 1
1 1 1
0 10 25
20 10 0
1 1 1
0 10 25
0 0 95
0
0
0
3.0 2.0 2.0 5.0

0.6 1.5 1.5 5.4
1.2 0.3 0.3 2.4
3 2 1
2 1 1
6 2 4
3.0 2.0 2.0 5.0

0 1.1 1.1 4.4
0
0
0
0
2
1
3
0 1/ 3 1/ 3
0
0
0
Seoul
National
Univ.
75
Rank and Row-Equivalent Matrices
Example 3
Linear Independence
/Dependence
Solution) Q ?
Determine whether the set of vectors

u1 2,1,1
u 2 0,3,0
u 3 3,1,2
in R3 in linearly dependent or linearly

independent.
Seoul
National
Univ.
76
Rank in Terms of Column Vectors

Theorem 3 : Rank in Terms of Column Vectors
The rank r of a matrix A equals the maximum number of linearly independent
column vectors of A. Hence A and its transpose AT have the same rank.
Let A be an m x n matrix of rank(A) = r
Proof)
Then by definition of rank, A has r linearly independent rows which we
denote by v(1), , v(r) and all the rows a(1), , a(m) of A are linear
combinations of those.
Seoul
National
Univ.
77
Rank in Terms of Column Vectors - 3 by 3 matrix

let rank A 3
a11 a12
A a21 a22
a31 a32
Note
a13 a1
a23 a2 ,
a33 a3
3(= rank A) linearly independent rows (basis) :
v1 v11 v12 v13

v 2 v21 v22 v23
v 3 v31 v32 v33
a1 a11
a a
2 21
a 3 a31
a12
a22
a32
a13 c11 v1 c12 v 2 c13 v 3

a23 c21 v1 c22 v 2 c23 v 3
a33 c31 v1 c32 v 2 c33 v 3
v1, v2, v3
A rank 3 linearly independent

.
A 3 Basis
. (ex : a , a , a )
Ex) 3 b
1
2
3
basis a1 , a 2 , a3 linear
combination
(b) A
. b la ma na
(l , m, n : const )
1
a1 , a2 , a3
basis 3 basis .
b la1 ma 2 na3
l (c11 v1 c12 v 2 c13 v3 ) m(c21 v1 c22 v 2 c23 v3 ) n(c31 v1 c32 v 2 c33 v3 )
(lc11 mc21 nc31 ) v1 (lc12 mc22 nc32 ) v 2 (lc13 mc23 nc33 ) v3
(l , m, n, c : const )
a1 , a2 , a3
2 basis
Ex) 3 b
basis v1 , v 2 , v3 linear
combination
b la1 ma 2 na3
l (c11 v1 c12 v 2 ) m(c21 v1 c22 v 2 ) n(c31 v1 c32 v 2 )
(lc11 mc21 nc31 ) v1 (lc12 mc22 nc32 ) v 2
(l , m, n, c : const )
Ex) v1 , v 2 3
b
Seoul
National
Univ.
78
v1 v11 v12 v13

v2 v21 v22 v23
v 3 v31 v32 v33

- 3 by 3 matrix
a1 a11
A a 2 a21
a3 a31
a12
a22
a32
a13 c11 v1 c12 v 2 c13 v3
a23 c21 v1 c22 v 2 c23 v3

a33 c31 v1 c32 v 2 c33 v3
a11 c11v11 c12v21 c13v31

a21 c21v11 c22v21 c23v31
a31 c31v11 c32v21 c33v31
a12 c11v12 c12v22 c13v32
a22 c21v12 c22v22 c23v32
a32 c31v12 c32v22 c33v32
a13 c11v13 c12v23 c13v33
a23 c21v13 c22v23 c23v33
a33 c31v13 c32v23 c33v33
a basis v
[a1 ] [a11 , a12 , a13 ] c11 v1 c12 v 2 c13 v3
c11[v11 , v12 , v13 ] c12 [v21 , v22 , v23 ] c13[v31 , v32 , v33 ]
[(c11v11 c12v21 c13v31 ), (c11v12 c12v22 c13v32 ), (c11v13 c12v22 c13v32 )]
Column
Vector of A
linearly independent basis
a1k
c11
c12
c13
a v c v c v c
2 k 1k 21 2 k 22 3k 23
a3k
c31
c32
c33
3 basis (c v ) . rank AT = rank A

Seoul
National
Univ.
79

- 3 by 3 matrix
a1 a11
A a 2 a21
a3 a31
a12
a22
a32
a13 c11 v1 c12 v 2 c13 v3
a23 c21 v1 c22 v 2 c23 v3

a33 c31 v1 c32 v 2 c33 v3
linearly dependent ?
a1k
a v
1k
2k
a3k
c11
c v
2k
21
c31
c12
c12
c v u c
3k
22
22
c32
c23
a1k
c11
c12
a v c (v v u ) c
2k
3k
2 k 1k 21
22
a3k
c31
c32
a1k
c11
c12
c13
a v c v c v c
2 k 1k 21 2 k 22 3k 23
a3k
c31
c32
c33
linearly independent basis
3 basis (c v
) .
rank AT = rank A
A basis v1 , v new
a1 a11 a12
A a 2 a21 a22
a3 a31 a32
a13 c11 v1 c12 v 2 uc12 v3 c11 v1 c12 (1 u )( v 2 v3 ) c11 v1 c12 (1 u ) v new

Seoul
National
Univ.
80
7.5 SOLUTIONS OF LINEAR SYSTEMS :

EXISTENCE, UNIQUENESS
Seoul
National
Univ.
81
Solutions of Linear Systems : Existence, Uniqueness

Theorem 1
(a) Existence : A linear systems of m equations in n unknowns x1, , xn

a21x1 a22 x2 a2 n xn b2
Is consistent, that is, has solutions, if and only if the coefficient matrix A
and the augmented matrix have same rank. Here
a11
a
A 21

am1
a12
a22
am 2
a1n
a2 n
,

amn
a11
~ a21
A

am1
a12
a1n
a22
a2 n
am 2
amn
b1

bm
Seoul
National
Univ.
82

(b) Uniqueness
The linear system has precisely one solution if and only if this common
rank r of A and equals n.
(c) Infinitely Many Solutions

If this common rank r is less than n, the system has infinitely many
solutions.
(d) Gauss Elimination

If solutions exist, they can all be obtained by the Gauss elimination;
Seoul
National
Univ.
83

Theorem
1 1
1 1
0 10
20 10
Rank of a Matrix by Row Reduction

If a matrix A is row equivalent to a row-echelon form B, then
i) the row space of A = the row space of B
ii) the nonzero rows of B from a basis for the row space of A, and
iii) rank(A) = the number of nonzero rows in B
0
0
90
80
1
1
25
0
0
1 1 1
0 10 25 90
0 0 95 190
rank : 3
x1 x2 x3
x1 11x2 24 x3
3x1 3x2 92 x3
2 x1 2 x2 2 x3
0
90
190
0
x1 x2 x3
0 x1 10 x2 25 x3
0 x1 0 x2 95 x3
0 x1 0 x2 0 x3
0
90
190
0
2.0 5.0 8.0

3.0 2.0
0.6 1.5
1.5 5.4 2.7
1.2 0.3 0.3 2.4 2.1
3.0 2.0 2.0 5.0 8.0
0 1.1 1.1 4.4 1.1

0
rank : 2
3 2 1 3
2 1 1 0
6 2 4 6
3
2
1
3
0 1 / 3 1 / 3 2
0
0 12
0
rank : 3
3.0 x1 2.0 x2 2.0 x3 5.0 x4 8.0
x1 0.3x2 0.3 x3 0.2 x4 2.3

1.5 x 1.0 x 1.0 x 2.5 x 4.0
1
2
3
4
3x1 2 x2 x3 3
9 x1 7 x2 2 x3 15
3x 2 x x 9
1
2
3
3.0 x1 2.0 x2 2.0 x3 5.0 x4 8.0
0 x1 1.1x2 1.1 x3 4.4 x4 1.1

0 x 0 x 0 x 0 x 0.0
1
2
3
4
x3 3
3x1 2 x2
1/ 3x2 1/ 3x3 2
0 12
Seoul
National
Univ.
84

x1 x2 1
1 1
1
4 1 x1 6
x
2 3 2 8
4 x1 x2 6
2 x1 3x2 8
1 1 1
A | B 4 1 6
2 3 8
1 0 1
0 1 2
0 0 16
row operation
1 1
A 4 1
2 3
row operation
Ax B
1 0 0
0 1 0
0 0 1
row operation
rank(A|B)=3
different
1 0
0 1
rank(A)=2
0 0
1 0 0
0 1 0
0 0 1
rank(A)rank(A|B)
x1 0 x2 1
No solution case
0 x1 x2 0
0 x 0 x 1
2
1
False statement
Seoul
National
Univ.
85

Theorem
Consistency of AX=B
A linear system of equations AX=B is consistent if and only if the rank of the
coefficient matrix A is the same as the rank of the augmented matrix of the
system
1 1
1 1
0 10
20 10
0
0
90
80
1
1
25
0
0
1 1 1
0 10 25 90
0 0 95 190
rank (A|B): 3
0 rank (A): 3
x1 x2 x3
x1 11x2 24 x3
3x1 3x2 92 x3
2 x1 2 x2 2 x3
2.0 5.0 8.0

3.0 2.0
0.6 1.5
1.5 5.4 2.7
1.2 0.3 0.3 2.4 2.1
90
190
x1 x2 x3
0 x1 10 x2 25 x3
0 x1 0 x2 95 x3
0 x1 0 x2 0 x3
rank (A|B) : 2
rank (A): 2
3
2
1
3
0 1 / 3 1 / 3 2
0
0 12
0
3.0 x1 2.0 x2 2.0 x3 5.0 x4 8.0
x1 0.3x2 0.3 x3 0.2 x4 2.3

1.5 x 1.0 x 1.0 x 2.5 x 4.0
1
2
3
4
3.0 2.0 2.0 5.0 8.0
0 1.1 1.1 4.4 1.1
3 2 1 3
2 1 1 0
6 2 4 6
Solution
- One solution
90
190
0
Solution
- Many solutions
3.0 x1 2.0 x2 2.0 x3 5.0 x4 8.0
0 x1 1.1x2 1.1 x3 4.4 x4 1.1

0 x 0 x 0 x 0 x 0.0
1
2
3
4
rank (A|B) : 3
rank (A): 2
3x1 2 x2 x3 3
9 x1 7 x2 2 x3 15
3x 2 x x 9
1
2
3
No Solution
x3 3
3x1 2 x2
1/ 3x2 1/ 3x3 2
0 12
Seoul
National
Univ.
86

Theorem 2
A homogeneous linear system
a11 x1 a12 x2 a1n xn 0

a21 x1 a22 x2 a2 n xn 0
am1 x1 am 2 x2 amn xn 0
A homogeneous linear system of m equations in n unknowns always
has the trivial solution.
Nontrivial solutions exist if and only if rank A =r <n.
If rank A =r <n, these solutions, together with x = 0, form a vector space

of dimension n r, called the solution space.
Linear combination of two solution vectors of the homogeneous linear

system, x = c1x(1) + c2x(2) with any scalars c1x(1) and c2, is a solution vector.
Seoul
National
Univ.
87

if x3 1, x4 0 x2 1, x1 0
Example
2
5 0
3 2
6 15 15 54 0
12 3 3 24 0
2
5 0
3 2
0 11 11 44 0
0 11 11 44 0
if x3 0, x4 1 x2 4, x1 1
[ x1 , x2 , x3 , x4 ] [1, 4, 0,1]
dimension of solution vectors (n - r) (4 - 2)
x (1) [ x1 , x2 , x3 , x4 ] [0, 1,1, 0]
x ( 2 ) [ x1 , x2 , x3 , x4 ] [1, 4, 0,1]
3 2 2 5 0
0 1 1 4 0
0 0 0 0 0
(a) rank (A|B) : 2
rank (A): 2
[ x1 , x2 , x3 , x4 ] [0, 1,1, 0]
(b) Linear combination of solution vectors
x c1x (1) c2 x ( 2 ) [c2 ,c1 4c2 , c1 , c2 ]

Nontrivial
solution exist!
(b) Solution space
3 x1 2 x2 2 x3 5 x4 0
x2 x3 4 x4 0
c2 0
2
5
3 2

6 15 15 54 c1 4c2 0
c
0
1
12 3 3 24

c2
0
A(c1x (1) c2 x ( 2 ) ) c1Ax (1) c2 Ax ( 2 ) 0

Seoul
National
Univ.
88

Theorem 3 : Homogeneous Linear System with Fewer Equations Than Unknowns
A homogeneous linear system with fewer equations than unknowns has

always nontrivial solutions.
Theorem 4 :Nonhomogeneous Linear System
If a nonhomogeneous linear system (1) is consistent, then all of its solutions are
obtained as
a21x1 a22 x2 a2 n xn b2
x = x0 + xh
(1)
a11x1 a12 x2 a1n xn 0

a21x1 a22 x2 a2 n xn 0
(2)
am1 x1 am 2 x2 amn xn 0
where x0 is any solution of the nonhomogeneous linear system (1) and xh runs
through all the solutions of the corresponding homogeneous system (2).
Proof ) xh = x - x0 Axh = A(x - x0)=b-b=0

Seoul
National
Univ.
89
Connection between the concept of

rank of a matrix and the solution of linear system
AX=0
Always consistent
Unique Solution :
X=0
rank(A) = n
Infinity of Solution :
Rank(A) < n
AX=B, B0
Consistent:
rank(A)=rank(A|B)
Unique Solution :
rank(A) = n
Inconsistent:
rank(A)<rank(A|B)
Infinity of Solution :
Rank(A) < n
Seoul
National
Univ.
90
7.6 SECOND- AND THIRD-ORDER

DETERMINANTS
Seoul
National
Univ.
91
Determinant of second- and third order

Determinant of second order
a11 a12 a11 a12

D det A det
a11a22 a12 a21
a21 a22 a21 a22
Determinant of third order
a11
a12
a13
D a21 a22
a23
a31 a32
a33
a11
a22 a23
a32
a33
a21
a12
a13
a32 a33
a31
a12
a13
a22 a23
Seoul
National
Univ.
92
7.7 DETERMINANTS. CRAMERS RULE
Seoul
National
Univ.
93
Determinant of Order n
Terms
In D we have n2 entries ajk, also n rows and n columns, and a main diagonal on
which a11, a12, , ann stand.
Mjk is called the minor of ajk in D, and Cjk the cofactor of ajk in D
For later use we note that D may also be written in terms of minors
D 1
j k
k 1
n
D 1
j 1
j k
a jk M jk
j 1, 2, , n
a jk M jk
k 1, 2, , n
C jk 1
j k
M jk
Seoul
National
Univ.
94
A determinant of order n is a scalar associated with an n x n matrix A=[ajk],
which is written
a11
a12 a1n
a21 a22 a2 n
D det A
an1
and is defined for n=1 by
an 2 ann
D a11
Seoul
National
Univ.
95
For n2 by
D a j1C j1 a j 2C j 2 a jnC jn
j 1, 2, , n
D a1k C1k a2k C2k ankCnk
k 1, 2, , n
or
Here,
C jk 1
j k
M jk
Mjk is a determinant of order n-1, namely, the determinant of the

submatrix of A obtained A by omitting the row and column of the entry
ajk, that is, the jth row and the kth column.
Seoul
National
Univ.
96
1) n=1
A a11
det A a11
2) n=2
a11 a12
A
a21 a22
det A a11
a11
a12
a21 a22
a12
a11
a12
a21 a22
a11a22 a12a21
Seoul
National
Univ.
97
3) n=3
a11 a12 a13
A a21 a22 a23

a31 a32 a33
a11
a12
a13
a11
a12
a13
a11
a12
a13
det A a11 a21 a22 a23 a12 a21 a22 a23 a13 a21 a22 a23
a31 a32 a33
a31 a32 a33
a31 a32 a33
a11a22a33 a23a32 a12 a21a33 a23a31

a13 a21a32 a22a31
Seoul
National
Univ.
98
C jk 1
Determinant :
j k
M jk
(Minors and Cofactors of a Third-Order Determinant)

Find minors and cofactors.
a11 a12 a13

A a21 a22 a23
a31 a32 a33
1) 1st row
a11
a13
M 12 a21 a22 a23

a31 a32
a33
C12 1 M12 M12

1 2
a11
a13
a23
M 13 a21 a22
a23
a33
a31 a32
a33
a13
M 11 a21 a22
a31 a32
C11 1 M11 M11

a12
a12
a12
11
a11
C13 1 M13 M13

13
Seoul
National
Univ.
99
C jk 1
Determinant :
a11 a12 a13

A a21 a22 a23
a31 a32 a33
2)
2nd
row
a11
a12
a13
M 21 a21 a22 a23

a31 a32
a33
C21 1 M 21 M 21
21
a11
a12
j k
M jk
a13
M 22 a21 a22 a23

a31 a32
C22 1
2 2
a11
a33
M 22 M 22
a12
a13
M 23 a21 a22 a23

a31 a32
C23 1
23
a33
M 23 M 23
Seoul
National
Univ.
100
C jk 1
Determinant :
a11 a12 a13

A a21 a22 a23
a31 a32 a33
a11
a12
j k
M jk
a13
M 32 a21 a22 a23

a31 a32
C32 1
3 2
a33
M 32 M 32
3) 3rd row
a11
a12
a13
M 31 a21 a22
a23
a31 a32 a33

31
C31 1 M 31 M 31
a11
a12
a13
M 13 a21 a22
a23
a31 a32
a33
C33 1 M 33 M 33
33
Seoul
National
Univ.
101
j k
Determinant :
D 1 a jk M jk j 1, 2, , n
k 1
(Minors and Cofactors of a Third-Order
Determinant)
n
(Expansions of a Third-Order Determinant)

Find determinant
3 0
det A 2
6 4
1 0 2
1) 1st rows
1
1 2
3 0
6 4 3 2
1 0 2
3 0
6 4 0 2
6 4
3 0
1 0 2
1 0 2
112 0 34 4 00 6 12
Seoul
National
Univ.
102
Determinant :
j 1, 2, , n
D 1 a M
n
k 1
j k
jk
jk

Find determinant.
3 0
det A 2
6 4
2)
2nd
1 0 2
rows
1
2 2
3 0
6 4 6 2
1 0 2
3 0
6 4 4 2
6 4
3 0
1 0 2
1 0 2
26 0 62 0 40 3 12
Seoul
National
Univ.
103
Determinant :
j 1, 2, , n
D 1 a M
n
k 1
j k
jk
jk

Find determinant.
3 0 0
det A 6
4 0
1 2 5
3 0 0
3 6
4 0 0 6
1 2 5
3
3 0 0
4 0
2 5
3 0 0
4 0 0 6
1 2 5
4 0
1 2 5
3 4 5 60
Seoul
National
Univ.
104
Behavior of an nth-Order Determinant under Elementary

Row Operations
Theorem 1. Behavior of an nth-Order Determinant under
Elementary Row Operations
(a) Interchange of two rows multiplies the value of the
determinant by -1.
(b) Addition of a multiple of a row to another row does not
alter the value of the determinant.
(c) Multiplication of a row by a nonzero constant c
multiplies the value of the determinant by c.
Seoul
National
Univ.
105

Row Operations
Proof. (a) Interchange of two rows multiplies the value of the
determinant by -1 by induction.
The statement holds for n=2 because
a b
c d
ad bc,
c d
a b
bc ad
(a) holds for determinants of order n-1(2) and show that it then holds
determinants of order n.
Let D be of order n. Let E be one of those interchanged. Expand D and E
by a row that is not one of those interchanged, call it the jth row.
n
D 1
k 1
j k
a jk M jk ,
E 1
k 1
j k
a jk N jk
Seoul
National
Univ.
106

Row Operations
n
D 1
j k
k 1
E 1
a jk M jk ,
j k
k 1
a jk N jk
Njk is obtained from the minor Mjk of ajk in D by interchange of those two rows
which have been interchanged in D.

Now these minors are of order n-1.
a
a
a
a
a
a
The induction hypothesis applies
11
12
Ex) D a21
a31
13
11
12
13
a22
a23 E a31
a32
a33
a32
a33
a22
a23
a21
M jk N jk
D 1
k 1
j k
a jk M jk
j k
1
a jk N jk E
k 1
Seoul
National
Univ.
107

Row Operations
Proof. (b) Addition of a multiple of a row to another row does
not alter the value of the determinant.
Add c times Row i to Row j.
Let ~ be the new determinant. Its entries in Row j are
D
ajk+caik.
a11
a12 a1n
ai1
ai 2
D
a j1
an1
ain
a j 2 a jn
an 2 ann
~
D
a11
a12
a1n
ai1
ai 2
ain
a j1 cai1
a j 2 cai 2 a jn cain
an1
an 2
ann
Seoul
National
Univ.
108

Row Operations
We can write
~
D
~
D
by the jth row.
a11
a12
a1n
ai1
ai 2
ain
a j1 cai1
an1
an 2
ann
j k
k 1
a jk M jk c 1
j k
k 1
D1
D2
k 1
j k
jk
caik M jk
aik M jk D1 cD2
Seoul
National
Univ.
109

Row Operations
a11
D1 1
j k
k 1
a12 a1n
ai1
ai 2
a jk M jk
a j1
ain
a j 2 a jn
an1 an 2 ann
Seoul
National
Univ.
110

Row Operations
It has aik in both Row i and Row j.
Interchanging these two rows gives D2 back,
a11
n
D2 1
k 1
j k
ai1
ai 2
aik M jk
a12 a1n
ain
ai1
ai 2
ain
Interchanging
,
Same
an1 an 2 ann
D2
a11
a12
a1n
ai1
ai 2
ain
ai1
ai 2
ain
an1 an 2
D2
ann
but on the other hand interchanging these two rows gives D2 by Theorem(a).
D2
Interchanging
,
D2
Together D2 = D2 = 0.
Seoul
National
Univ.
111
Behavior of an nth-Order Determinant under ElementaryD1 D

D2 0
Row Operations
~
D
a11
a12
a1n
a11
ai1
ai 2
ain
ai 2
ai1
a j1 cai1
an1
an 2
ann

a j1
a12 a1n
ain
a j 2 a jn
an1 an 2 ann
a11
a12
a1n
ai1
ai 2
ain
c
ai1
ai 2
ain
an1 an 2
ann
D1 cD2
D c0
D
Seoul
National
Univ.
112

Row Operations
Proof. (c) Multiplication of a row by a nonzero constant c multiplies the value of
the determinant by c
a11
D a j1
an1
a12 a1n
a12
a1n
an1
an 2
ann
~
D ca j1 ca j 2 ca jn
a j 2 a jn
a11
an 2 ann
Expand the determinant by the jth row.
n
~ n
j k
j k
D 1 ca jk M jk c 1 a jk M jk cD
k 1
k 1
Seoul
National
Univ.
113
Further Properties of nth-Order Determinants

Theorem 2. Further Properties of nth-Order Determinants
(d) Transposition leaves the value of a determinant
unaltered.
(e) A zero row or column renders the value of a determinant
zero.
(f) Proportional rows or columns render the value of a

determinant zero. In particular, a determinant with two
identical rows or columns has the value zero.
Seoul
National
Univ.
114

Proof.
(d) Transposition leaves the value of a determinant unaltered.
Transposition is defined as for matrices, that is, the jth row becomes the
jth column of the transpose.
Proof.
(e) A zero row or column renders the value of a determinant zero.
a11
a12 a1n
D 0
an1
j k
k 1
n
a jk M jk
1 0 M jk 0
an 2 ann
k 1
j k
Seoul
National
Univ.
115

Proof.
(f) Proportional rows or columns render the value of a determinant zero.
In particular, a determinant with two identical rows or columns has the
value zero.
a11
a12
a1n
a11
ai1
ai 2
ain
ai1
ai 2
cai1 cai 2 cain
a12 a1n
ai1
ai 2
an1
an 2
ann
an1
ain
an 2 ann
Theorem (1.c) Multiplication of a row by

a nonzero constant c multiplies the value
of the determinant by c
a12
a1n
ai1
ai 2
ain
ain
a11
x(-1)+
an1 an 2
c0 0
ann
Theorem. (1.b) Addition of a multiple of a

row to another row does not alter the
value of the determinant.
Seoul
National
Univ.
116
Determinant of a Triangular Matrix
a21 a22
a11
D det A
an1
M 11
It is also a determinant
of a triangular matrix.
an 2 ann
a11C11 a12C12 a1nC1n
a11C11 a12 a13 a1n 0

a11M11
a11 a22 ann
Seoul
National
Univ.
117
Rank in Terms of Determinants

Theorem 3. Rank in Terms of Determinants
Consider an m x n matrix A=[ajk]
1) A has rank r1 if and only if A has an r x r submatrix with nonzero
determinant.
2) The determinant of any square submatrix with more than r rows,
contained in A (if such a matrix exists!) has a value equal to zero.
3) In particular, if A is square, n x n, it has rank n if and only if det D 0
1 1 1
1 1 1
(1) Example
0 10 25
20 10 0
1 1 1
0 10 25
0 0 95
0
0
0
3 2 1
(3) Example 2 1 1
6 2 4
Rank=3
2
1
3
0 1/ 3 1/ 3 det D=0
Rank=2
0
0
0
Seoul
National
Univ.
118
Solving linear systems of two equations

Solve the linear systems of two equations
1. General Solution
a11x1 a12 x2 b1
a21x1 a22 x2 b2
a22 a12 :
a11a22 a12a21 x1
a21 a11 :
a11a22 a12a21 x1
b1a22 a12b2
a11b2 b1a21
b1a22 a12b2
x1
a11a22 a12a21
a11b2 b1a21
x2
a11a22 a12a21
a11a22 a12a21 0
a11a22 a12a21 0
Seoul
National
Univ.
119

a11x1 a12 x2 b1
a21x1 a22 x2 b2
x1
b1a22 a12b2
a11a22 a12a21
x2
a11b2 b1a21
a11a22 a12a21
D det A
2. Use Cramers rule
a11
a12
a21 a22
a11a22 a12a21
b1
a12
a22
x1
D
b1a22 a12b2
a11
b2
D 0
D 0
b1
a21 b2
x2
D
a11b2 b1a21
D 0
Seoul
National
Univ.
120
Solving linear systems of three equations
a11x1 a12 x2 a13 x3 b1

a21x1 a22 x2 a23 x3 b2
a31x1 a32 x2 a33 x3 b3
D1
x1
,
D
D2
x2
;
D
b1
a12
a13
a11
D1 b2
a22
a23 ,
D2 a21 b2
b3
a32
a33
b1
a31 b3
D3
x3
D
a13
a23 ,
a33
a11
a12
b1
D3 a21 a22 b2
a31 a32 b3
Seoul
National
Univ.
121

Note that D1, D2, D3 are obtained by replacing Columns 1, 2, 3.
a11x1 a12 x2 a13 x3 b1

a21x1 a22 x2 a23 x3 b2
a31x1 a32 x2 a33 x3 b3
a11 a12
a
a
21
22
a31 a32
a13 x1 b1
a23 x2 b2
a33 x3 b3
x
ab111 a12
A ab21
a22
2
ab313 a32
a13
b1
a23 , b b2
b3
a33
b1
a12
a13
D1 b2
a22
a23
b3
a32
a33
Seoul
National
Univ.
122

a11x1 a12 x2 a13 x3 b1

a21x1 a22 x2 a23 x3 b2
a31x1 a32 x2 a33 x3 b3
a11 a12
a
a
21
22
a31 a32
a13 x1 b1
a23 x2 b2
a33 x3 b3
x
a11 ab121
A a21 ab22
2
a31 ab32
3
a13
b1
a23 , b b2
b3
a33
a11
b1
a13
D2 a21 b2
a23
a31 b3
a33
Seoul
National
Univ.
123

a11x1 a12 x2 a13 x3 b1

a21x1 a22 x2 a23 x3 b2
a31x1 a32 x2 a33 x3 b3
a11 a12
a
a
21
22
a31 a32
A
a11 a12
A a21 a22
a31 a32
ab13
b1
1
ab23
2 , b b2
b3
ab33
3
a13 x1 b1
a23 x2 b2
a33 x3 b3
x
a11 a21 b1
D3 a21 a22 b2
a31 a23 b3
Seoul
National
Univ.
124
Cramers Rule
Cramers Theorem (Solution of Linear Systems by Determinants)
(a) If a linear system of n equations in the same number of unknowns x1, , xn

a21x1 a22 x2 a2 n xn b2
an1 x1 an 2 x2 ann xn bn
has a nonzero coefficient determinant D=det(A), the system has precisely one
solution. This solution is given by the formulas
Dn
D1
D2
x1
, x2
, , , xn
D
D
D
Where Dk is the determinant obtained from D by replacing in D the kth column
by the column with the entries b1, , bn.
Seoul
National
Univ.
125
Cramers Rule
Cramers Theorem (Solution of Linear Systems by Determinants)
a11 ab11k
a21 ab22k
Dk
a1n
b1
a2 n
b2
an1 abnk
ann
n
bn
replace
Dk b1C1k b2C2k bnCnk

(b) Hence if the system is homogeneous and D0, it has only the trivial solution
x1=0, , xn=0. If D=0, the homogeneous system also has nontrivial solutions.
Seoul
National
Univ.
126
Example
Q ? Solve by Cramers rule.
3y - 4z = 16
2x - 5y +7z = -27
-x - 9z
=9
Seoul
National
Univ.
127

a
a
D 0 D det A a11 a12 a11a22 a12a21 0
21
22
a11 x1
a12 x2 b1
a21 x1
a22 x2 b2
a22
a
x
a
x
b1
21 1 22 2
a
12
a x a x b
2
21 1 22 2
a22
a
x
a
x
b1
21
1
22
2
12
0 x 0 x a22 b b
1
2
1
2
a12
a11
a12 a21
a22
a12 a21
x1 a12 x2 b1
a22
Linearly independent equation : 1

Variables : 2
a21
a22
0
Homogeneous linear systems
a
A 11
a21
a12
a22
Trivial Solution x 0
a22
b1 b2 0 b
a12
a22
a
x
a
x
b1
21 1 22 2
a
12
0 x 0 x b
1
2
False statement
a21
a22
b1
a12
rank(A)=1=rank(A|B)
rank(A)=1<2 unknowns
Ax 0
a22
b1 b2 0
a12
a22
a
x
a
x
b1
21 1 22 2
a
12
0 x 0 x 0
1
2
when
a22
0
a22
b1
a12
rank(A)=12=rank(A|B)
a11 x1 a12 x2 0
a21 x1 a22 x2 0
det A
a11
a12
a21 a22
Ax 0
a
A 11
0
a12
0
Nontrivial many solutions

Seoul
National
Univ.
128
Cramers Rule - Proof

Proof
(a) The augmented matrix is of size nx(n+1). Hence its rank can be at
most n.
If D=det A 0, then rank A = n. Thus rank =rank A. The system has a
unique solution.
If kth column is replaced by lth
column
(b)
a11
a1k
D a j1
a jk
an1
ank
a1l a1n
a jl a jn
anl a jn
D a1k C1k a2k C2k ank Cnk

a1l
a11
a1l
D a j1
a jl
an1
anl
a1n
a jl a jn
anl a jn
D a1l C1k a2l C2k anl Cnk 0

Seoul
National
Univ.
129
Cramers Rule - Proof

We now multiply the first equation in the linear system by C1k on both sides, the
second by C2k ,the last by Cnk and add the resulting equations. This gives
C1k (a11 x1 a12 x2 a1n xn ) Cnk (an1 x1 an2 x2 ann xn )

b1C1k bnCnk
Collecting terms with the same xj, we can write the left side as
x1(a11C1k a21C2 k an1Cnk ) xn(a1nC1k a2nC2 k annCnk )

Only one term led by xj remains.
xk (a1k C1k a2k C2 k ank Cnk ) xk D

x1(a11C1k a21C2 k an1Cnk ) 0 since 1 k
xl(a1l C1k a2l C2 k anl Cnk ) 0 since l k
Therefore,
xk D b1C1k bnCnk Dk
a11 ab11k a1n

a21 ab22k a2 n
Dk
an1 abnk
ann
n
Seoul
National
Univ.
130
7.8 INVERSE
OF A
MATRIX. GAUSS-JORDAN
ELIMINATION
Seoul
National
Univ.
131
Notation of inverse matrix

In this inverse section, only square matrices are considered exclusively.
Notation of inverse of an nn matrix A=[ajk] : A-1
AA1 A1A I
, where I is the nn unit matrix.
Nonsingular matrix : A matrix that has an inverse.
(If a matrix has an inverse, the inverse is unique)

Singular matrix : A matrix that has no inverse.
Proof of uniqueness of inverse matrix

If B and C are inverses of A
We obtain
(AB I & CA I) ,
B IB (CA)B C(AB ) CI C
(the uniqueness of inverse)
Seoul
National
Univ.
132
Theorem1. Existence of the Inverse
Theorem 1. Existence of the Inverse

The inverse A-1 of an n x n matrix A exists if and only if
rank A = n, thus if and only if det A 0.
Hence A is nonsingular if rank A = n,
and is singular if rank A < n.
Seoul
National
Univ.
133
Inverse by the Gauss-Jordan Method

For Practical determination of the inverse A-1 of a nonsingular nn matrix A,
Gauss elimination can be used.
: This method is called Gauss-Jordan elimination
Step 1. Make augmented matrix.
~
A [ A I]
Step 2. Make Multiplication of AX=I by A-1

(by applying Gauss elimination to
~
A [ A I])
This gives a matrix of the form [U H]
U : upper triangular
Step 3. Reduce U by further elementary row operations to diagonal form.

(Eliminate the entries of U above the main diagonal and making the
diagonal entries all 1 by multiplication. See the example next page.)
Seoul
National
Univ.
134
Inverse of a Matrix. Gauss-Jordan elimination.

Determine the inverse A-1 of
1 1 2
A 3 1 1
1 3 4
Step 1. Make augmented matrix.
1 1 2 1 0 0
A I 3 1 1 0 1 0
1 3 4 0 0 1
Step 2. Make Multiplication of AX=I by A-1 by applying Gauss elimination
to
Seoul
National
Univ.
135
Inverse of a Matrix. Gauss-Jordan elimination

A1AX A1I
AX I
Gauss-Jordan
Elimination
[ A | I]
2
1 0 0
1 1
0 2
7
3
1
0
0 0 5 4 1 1
[I | A1 ]
1 1 0 0.6 0.4 0.4

0 2 0 2.6 0.4 1.4
0 0 5 4
1 1
1 1 2 1 0 0
A I 3 1 1 0 1 0
1 3 4 0 0 1
1 0 0
1 1 2
0 2 7
3
1
0
0 2 2 1 0 1
IX A1
Row2 + 3Row1
Row3 - Row1
1 0 0 0.7 0.2 0.3

0 2 0 2.6 0.4 1.4
0 0 5 4
1
1
Row1
+0.4Row3
Row2
+1.4Row3
Row1
-0.5Row2
diagonal matrix
Row3
Row2
1 0 0 0.7 0.2 0.3

0 1 0 1.3 0.2 0.7
0 0 1 0.8 0.2 0.2
-Row1
0.5Row2
-0.2Row3
Seoul
National
Univ.
136
Inverse of a Matrix. Gauss-Jordan elimination

0.2
0.3
1 0 0 0.7
0 1 0 1.3 0.2
0.7
0 0 1
0.8
0.2 0.2
A 1
Check the result.
Let
-1
AA = B
b11 b12 b13
b21 b22 b23
b31 b32 b33
b11 1 0.7 1 1.3 2 0.8 1
b12 1 0.2 1 0.2 2 0.2 0
b13 1 0.3 1 0.7 2 0.2 0
b21 3 0.7 1 1.3 1 0.8 0
b22 3 0.2 1 0.2 1 0.2 1
b23 3 0.3 1 0.7 1 0.2 0
b31 1 0.7 3 1.3 4 0.8 0

b32 1 0.2 3 0.2 4 0.2 0
b33 1 0.3 3 0.7 4 0.2 1
1
1
2
0
.
7
0
.
2
0
.
3
1 0 0
3 1 1 1.3 0.2
0.7
0 1 0
1 3 4 0.8
0.2 0.2
0 0 1
Seoul
National
Univ.
137
Useful formulas for Inverses

Theorem 2. Inverse of a Matrix
cofactor: C jk (1) j k M jk
M jk : a determinant of order n 1
Theorem 2. Inverse of a Matrix by determinant

The inverse of a nonsingular nn matrix A=[ajk] is given by
C11 C21 Cn1
1
1 C12 C22 Cn 2
1
T
(4)
A
[C jk ]
det A
det A .
.
.
.
C1n C2 n Cnn
Where Cjk is the cofactor of ajk in det A

CAUTION! Note well that in A-1, the cofactor Cjk occupies the same place
as akj (not ajk) does in A.)
Proof)
Let
C11
1 C12
B
det A .
C1n
C21 Cn1
C22 Cn 2
.
.
.
C2 n Cnn
and show that BA=I
Seoul
National
Univ.
138
Useful formulas for Inverses

Theorem 2. Inverse of a Matrix
Let
BA = G = [gkl](5)
Sec 7.7
(k l ), D a1k C1k a2 k C2 k ankCnk (9)
(k l ), a1l C1k a2l C2 k anlCnk 0(10)
Hence,
g kk
and then show that G = I.

By definition of matrix multiplication
and because of the form of B in (4)
n
Csk
asl
s 1 det A
1
a1l C1k anlCnk (6)
det A
g kl
In Sec 7.7 (9) and (10),
If l k , a1l C1k anlCnk det A
l k , a1l C1k anlCnk 0

1
det A 1
det A
glk 0
gkk are the entries of
main diagonal of matrix G.
and that means
only entries of main diagonal
is 1.
G = I
B = A-1
C11
1 C12
BA
det A .
C1n
C21 Cn1 a11

C22 Cn 2 a21
.
.
. .
C2 n Cnn an1
a12
a22
.
an 2
a1n
a2 n
.
.
ann
Seoul
National
Univ.
139
Inverse of 22 Matrix
Ex 2)
C jk 1
j k
M jk
C11 C21 Cn1

1 C12 C22 Cn 2
A1
.
.
.
det A .
C1n C2 n Cnn
3 1
A
2 4
det A 3 4 1 2 10
C11 4,
C12 2,
C21 1,
C22 3,
1 4 1
A
10 2 3
1
0.4 0.1
0
.
2
0
.
3
Seoul
National
Univ.
140
Inverse of 33 Matrix
Q?
1 1
A 3 1
1 3
2
1,
4
A 1 ?
C jk 1
j k
M jk
C11 C21 Cn1

1 C12 C22 Cn 2
A1
.
.
.
det A .
C1n C2 n Cnn
Seoul
National
Univ.
141
Inverse of Diagonal matrices

A=[ajk], ajk=0 when jk, have an inverse if and only if all ajj 0. Then A-1 is
diagonal, too, with entries 1/a11,, 1/ann.
Proof) For a diagonal matrix we have in (4)
a22 ann
C11
1
, etc.
D a11a22 ann a11
Ex 4) Inverse of Diagonal Matrix
0.5 0 0
A 0
4 0,
0
0 1
1
0.5
A 1 0
0
0
0
2
1
0 0 0.25 0
4
0
1
0
1
0
1
0
Seoul
National
Univ.
142
Inverse of Products
Products can be inverted by taking the inverse of each factor and multiplying
these inverses in reverse order,
Hence for more than two factors,
(AC) 1 C1A 1
(AC PQ) 1 Q1P 1 C1A 1
AC C1A1 ACC1A1 AIA 1

C1A1 AC C1A1AC C1IC
AA 1 I
C1C I
Seoul
National
Univ.
143
Cancellation Laws
Theorem 3. Cancellation Laws
Let A, B, C be n x n matrices. Then

(a) If rank A =n and AB = AC, then B = C
(b)If rank A =n, then AB =0 implies B =0. Hence if AB =0, but A0 as well as
B0, then rank A<n and rank B <n.
(c) If A is singular, so are BA and AB
Seoul
National
Univ.
144
Determinants of Matrix Products

Theorem 4. Determinant of a Product of Matrices
For any n x n matrices A and B,
det(AB ) det(BA) det A det B

If A or B is singular, so are AB and BA
Now A and B be nonsingular. Then we can reduce A to a diagonal matrix =
[ajk] by Gauss-Jordan steps.
a11 a12
a
a22
21
an1 an 2
a1n
a11 0 0
0 a
a2 n
0
0
22

.
, A

0
0 0
ann
0
0
a
nn
A and are row-equivalent

matrices.
Seoul
National
Univ.
145
A and are row-equivalent matrices.
a11 a12
a
a22
21
an1 an 2
a11
a
Ax 21

an1
a11
0
x
A
0
a1n
a11 0 0
0 a
a2 n
0
0
22

.
, A

0
0 0
ann
0 ann
0
a12 a1n x1 b1
a22 a2 n x2 b2

an 2 ann xn bn
0 0 x1 b1

a 22 0 0 x2 b2
0 0

0 a nn xn bn
1b
x A 1b A
Seoul
National
Univ.
146

Addition of a multiple of a row to another row does not alter the value of the
determinant.
B)
det(AB ) det(A
a11 0
0 a
22
B
A
0
0
0
0
a11b11
a 22b21
det(AB)
a nnbn1
0 b11
0 0 b21
0
a nn bn1
a11b12
a 22b22
b12 b1n a11b11

b22 b2 n a 22b21

bn 2 bnn a nnbn1
a11b1n
b11
a 22b2 n
b21
a11a 22 a nn
a nnbn 2 a nnbnn
a11b12 a11b1n
a 22b22 a 22b2 n
a nnbn 2 a nnbnn
b12 b1n
b22 b2 n
bn1 bn 2 bnn
Seoul
National
Univ.
147
D 1
j k
k 1
ca jk M jk cD
B)
det(AB ) det(A
B)
det(A
a11b11 a11b12 a11b1n

a 22b21 a 22b22 a 22b2 n
a nnbn1
b11
a11a 22 a nn
a nnbn 2 a nnbnn
b12 b1n
b21 b22 b2 n
bn1 bn 2 bnn
a11a22 ann det(B)

) det(B) det( A) det(B)
det( A
det(AB ) det(A) det(B)
Seoul
National
Univ.
148

01 Linear Algebra 1 - ETL PDF

Transféré par

Informations du document

Description originale:

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

01 Linear Algebra 1 - ETL PDF

Transféré par

Droits d'auteur :

Formats disponibles

CHAPTER 7.

7.6 Second- and Third-Order Determinants

Engineering Math, 7. Linear Algebram, Spring 2015

7.1 MATRICES, VECTORS :

Engineering Math, 7. Linear Algebram, Spring 2015

(no. of rows = no. of columns)

Engineering Math, 7. Linear Algebram, Spring 2015

If m = n, we call A an n x n square matrix.

If m n, we call A a rectangular matrix.

Engineering Math, 7. Linear Algebram, Spring 2015

is called augmented matrix() of the system.

We shall discuss this in great detail, beginning in Sec. 7.3

400 330 810 0 210 470 I

A 0 120 780 500 500 960 II .

Engineering Math, 7. Linear Algebram, Spring 2015

Matrix Addition and Scalar Multiplication

Then A B if and only if

Engineering Math, 7. Linear Algebram, Spring 2015

Matrix Addition and Scalar Multiplication

Engineering Math, 7. Linear Algebram, Spring 2015

Matrix Addition and Scalar Multiplication

Ex5) Scalar Multiplication

Engineering Math, 7. Linear Algebram, Spring 2015

0 : zero matrix (of size m x n matrix)

Rules for Matrix Addition and Scalar Multiplication

7.2 MATRIX MULTIPLICATION

Engineering Math, 7. Linear Algebram, Spring 2015

Multiplication of a Matrix by a Matrix

c jk a jl blk a j1b1k a j 2b2 k a jnbnk (1)

c21 a21b11 a22b21 a23b31

Engineering Math, 7. Linear Algebram, Spring 2015

Multiplication of a Matrix by a Matrix

The product BA is not defined.

Engineering Math, 7. Linear Algebram, Spring 2015

Multiplication of a Matrix by a Matrix

Ex3) Products of Row and Column Vectors

Multiplication of a Matrix by a Matrix

Ex 4) CAUTION! Matrix Multiplication is Not Commutative (),

Engineering Math, 7. Linear Algebram, Spring 2015

Properties of Matrix Multiplication

(kA)B k ( AB ) A(kB) written kAB or AkB

Engineering Math, 7. Linear Algebram, Spring 2015

Properties of Matrix Multiplication

c jk a jl blk a j1b1k a j 2b2 k a jnbnk (1)

where aj : the jth row vector of A

Engineering Math, 7. Linear Algebram, Spring 2015

Properties of Matrix Multiplication

If A=[ajk] is of size 3x3 and B=[bjk] is of size 3x4, then

Parallel processing of products on the computer is facilitated by a variant of (3)

Properties of Matrix Multiplication

Columns of B are then assigned to different processors (individually or

Engineering Math, 7. Linear Algebram, Spring 2015

Properties of Matrix Multiplication

To obtain AB from (5), calculate the columns

Engineering Math, 7. Linear Algebram, Spring 2015

Applications of Matrix Multiplication

1.2 1.6 Raw Components

A 0.3 0.4 Labor

Applications of Matrix Multiplication

1.2 1.6 Raw Components

Engineering Math, 7. Linear Algebram, Spring 2015

Applications of Matrix Multiplication