VIETNAM NATIONAL UNIVERSITY-HCMC

International University

Chapter 1. Introduction to linear systems and matrices

Linear Algebra for FERM
Lecturer: Nguyen Minh Quan, PhD
[email protected]

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

1 / 102

Contents
1

Basic concepts

Linear systems of equations

Gaussian elimination and Gauss-Jordan elimination

Matrix Operations

Inverse matrices

The LU factorization

Applications: Leontief input-output Model

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

2 / 102

Introduction
Linear Algebra has become as basic and as applicable as calculus, and
fortunately it is easier. Gilbert Strang, MIT

The application of linear algebra to economics lies primarily in its use of

matrices. A matrix in economics is used to solve a large number of linear
equations at once, where the variables are economic indicators and factors.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

3 / 102

Introduction

Leontief, 1973 Nobel Prize in Economic Science, divided the U.S.

economy into 500 sectors, and analyzed a linear system of 500
equations in 500 unknowns.
Linear programming: Many important management decisions today
are made on the basis of linear programming models. The airline
industry, for instance, employs linear programs that schedule flight
crews, monitor the locations of aircraft,...

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

4 / 102

What is a Vector?
A vector, represented by an arrow, has both a direction and a
magnitude. Magnitude is shown as the length of a line segment.
Direction is shown by the orientation of the line segment, and by an
arrow at one end.

Equal vectors have the same length and direction but may have
different starting points.

Examples of vectors in nature are velocity, force, electromagnetic

fields,...
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

5 / 102

What is a Vector?
Consider the figure below:

Each of the directed line segments in the sketch represents the same
vector. In each case the vector starts at a specific point then moves 2
units to the left and 5 units up.
Notation: ~v = h2, 5i or ~v = (2, 5).
Be careful to distinguish vector notation ~v = (2, 5) from the notation we
use to represent coordinates of points, A(2, 5) .
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

6 / 102

Vectors
Given the two points A(a1 , a2 ) and B(b1 , b2 ), the vector with the
~ is AB
~ = (b1 a1 , b2 a2 ).
representation AB
The magnitude, or length, of the vector ~v = (a, b) is given by,
p
k~v k = a2 + b 2
Example, if ~v = (3, 5) then its magnitude

k~v k = 9 + 16 = 5
Any vector with magnitude of 1 is called a unit vector, e.g.,
~v1 = (0, 1), or ~v2 = (1, 0) (standard basis vectors).
Zero Vector, ~0 = (0, 0), is a vector that has no magnitude or
direction.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

7 / 102

Linear systems of equations

We begin with the central problem of linear algebra: solving linear
equations.
Any straight line in the xy-plane can be represented algebraically by
an equation of the form
a1 x + a2 y = b
where a1 , a2 and b are real constants, a1 , a2 not both zero.
In general, a linear equation in the variables x1 , x2 , , xn is one that can
be put in the form
a1 x1 + a2 x2 + ... + an xn = b
where a1 , a2 , ..., an and b are real constants. The variables in a linear
equation are sometimes called unknowns.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

8 / 102

System of linear equations

Example
The director of a trust fund has $100.000 to invest. The rules of the trust
state that both a certificate of deposit (CD) and a long-term bond must
be used. The directors goal is to have the trust yield $7800 on its investments for the year. The CD chosen returns 5% per annum, and the bond
9%. Find the amounts that the director invest in the CD and in the bond.
Let x and y be the amount to invest in the CD and in the bond,
respectively. We have

x + y = 100, 000
0.05x + 0.09y = 7800
To eliminate x, we add (0.05) times the first equation to the second,
obtaining 0.04y = 2800, this implies y = 70, 000. Thus, x = 30, 000.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

9 / 102

System of linear equations (n=2)

We consider another example
1x + 2y = 3

(1)

4x + 5y = 6

(2)

The two unknowns are x and y.

How to solve? Elimination
Eq. (2) 4 Eq.(1): 3y = 6 y = 2.
Back-substitution x = 1.
This method is called Gaussian Elimination. It is good to solve large
systems of equations. We will consider in detail this method in this section.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

10 / 102

The Geometry of Linear Equations

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

11 / 102

The Geometry of Linear Equations

Understand this subject is by example:
2x y = 1
x +y =5
First approach: Each equations stands for a straight line. In column form
(2nd approach: vectors):
 

  
2
1
1
x
+y
=
1
1
5

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

12 / 102

Examples of system of linear equations (n=3)

Consider a system of n = 3 equations:
x + 2y + 3z = 6
2x 3y + 2z = 14
3x + y z = 2
To eliminate x, we add (2) times the first equation to the second one
and (3) times the first equation to the third one, obtaining a system of
two equations in the unknowns y and z
7y 4z = 2
5y 10z = 20
We multiply the second equation by 1/5, yielding
7y 4z = 2
y + 2z = 4
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

13 / 102

Examples of system of linear equations (n=3)

Interchanging equations
y + 2z = 4
7y 4z = 2
We now eliminate y by adding 7 times the first equation to the second
one, to obtain 10z = 30. Thus, z = 3. Substitute z = 3 into y + 2z = 4
to get y = 2. Finally, we find x = 1.
We observe further that our elimination procedure has actually produced
the linear system which is equivalent to the original one.
x + 2y + 3z = 6
y + 2z = 4
z =3
The importance of this procedure is that the later system is easier to solve!
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

14 / 102

The Geometry of Linear Equations

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

15 / 102

Column Vectors and Linear Combinations

Consider a system of n = 3 equations (3 planes):

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

16 / 102

The Geometry of Linear Equations

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

17 / 102

The method of elimination

The method of elimination involves three manipulations that can be
performed on a linear system to convert it into an equivalent system:

We need the concept of matrix to investigate the method of Gaussian

Elimination.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

18 / 102

What is a Matrix?
A matrix is a set of elements, organized into rows and columns

The number aij , 1 i m, 1 j n, are called the entries (or elements)

of A.
Matrix is like a function that transforms the vectors on a plane:


   
ax + by = e
a b
x
e

=
cx + dy = f
c d
y
f
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

19 / 102

Example of matrix

The following matrix gives the airline distance between the indicated cities
(in statute miles):

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

20 / 102

Example of matrix

Suppose that a manufacturer has four plants, each of which makes three
products. If we let aij denote the number of units of product i made by
plan j in one week, then the 3 4 matrix

gives the manufacturers production for the week. For example, plant 2
makes 270 units of product 3 in one week.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

21 / 102

Matrices
A matrix with m rows and n columns is called an m n matrix (m by n
matrix) or a matrix of order m n.
A matrix consisting of a single column is called a column vector and a
matrix consisting of a single row is called a row vector.
If m = n the matrix is called a square matrix. For example,


1 2 3
is a 2 by 3 matrix
4 5 6

1 2 3
3 5 1 is a (square) 3 by 3 matrix
7
0 9
Equality: Two matrices are said to be equal if they have the same order
and all the corresponding entries are equal.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

22 / 102

Linear systems of equations

We often deal with several linear equations at the same time:
a11 x1 + a12 x2 + ... + a1n xn = b1
a21 x1 + a22 x2 + ... + a2n xn = b2
..
.
am1 x1 + am2 x2 + ... + amn xn = bm
A system with at least one solution is called consistent. Otherwise, it is
called inconsistent.
Theorem: Every system of linear equations has zero, one or infinitely many
solutions; there are no other possibilities.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

23 / 102

Linear systems of equations

Matrix of coefficients is a matrix of m rows and n

a11 a12 a1n

a21 a22 a2n

A= .
..
..
..
..
.
.
.

columns:

am1 am2 amn

Augmented matrix is

A=

a11
a21
..
.

a12
a22
..
.

..
.

a1n b1
a2n b2
..
.

am1 am2 amn bm

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

24 / 102

Example
Consider the linear system of equations
x1 2x2 + x3 = 0
2x2 8x3 = 8
4x1 + 5x2 + 9x3 = 9
Matrix of Coefficients is

1 2 1
0
2 8
4 5
9
Augmented matrix is


1 2 1  0
0
2 5  8
4 5
9  9

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

25 / 102

Linear systems of equations

Motivation: Original augmented matrix much simpler augmented

matrix of an equivalent system!
For any augmented matrix of a system of equations, the following
operations produce the augmented matrix of an equivalent system:
1. Interchanging any two rows, denoted by: Iik .
2. Multiply a row by a nonzero constant: Ri .
3. Add a multiple of one row to another: Ri + Rk .

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

26 / 102

Pivot-Matrix in Echelon Form

A pivot (leading entry) of a matrix is the first nonzero entry in a row. A
pivot column is a column of A that contains a pivot position.

Definition
A matrix is in (row) echelon form if
a. All rows that contain only zeros are grouped at the bottom of the
matrix
b. For each row that does not contain only zeros, the pivot appears
strictly to the right of the pivot of each row that appears above it
Intuitively, a matrix is in row echelon
staircase pattern like

0
0 0

0 0 0 0

0 0 0 0
0 0 0 0
Nguyen Minh Quan (HCMIU-VNU)

form if it has the appearance of a

0
0

Chap 1 Linear systems and Matrices

Spring 2014

27 / 102

Matrix in Row Echelon Form

Example
The following matrices are in row echelon form


0 0 2 1 4
a)
0 0 0 0 1

1 3 2 7
b) 0 3 1 7
0 0 2 9

4 2 1 7 9 1
0 0 5 9 1 2
c)
0 0
0 8 3 5
0 0
0 0 0 0

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

28 / 102

Matrix in Row Echelon Form

Example
The following matrices are NOT in row echelon form


0 0 0 0 0
a)
0 0 0 0 1

1 3 2 7
b) 0 0 1 7
0 2 2 9

4 2 1 7 9 1
0 0 5 9 1 2
c)
0 0
0 8 3 5
0 0
0 7 1 0

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

29 / 102

Reduced Echelon Form

The leading entry in each nonzero row is 1.
Each leading 1 is the only non-zero

1 0 0
0 1 0
0 0 1

1 0 0
0 1 0

0 0 1
0 0 0

entry in its column.

29
16
1

0
0

0
1

Remark:
(1) If the reduced augmented matrix is in echelon form, the method is
called Gauss Elimination.
(2) If the reduced augmented matrix is in reduced echelon form, the
method is called Gauss-Jordan Elimination.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

30 / 102

Matrix in Row Echelon Form

Example
Suppose that the augmented matrix for a linear system has been reduced
to the given matrix in row echelon form and the variables are also given.
Solve the system.

2 1 3
5
0 3 2
17
0 0
5 10
Variables: x, y , z
Solution The first step is to find the associated linear system
2x + y 3z = 5
3y + 2z = 17
5z = 10
The system is in triangular form and we solve it by backward substitution:
5z = -10 leads to z = -2. Thus, y=-7, x=3.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

31 / 102

Matrix in Row Echelon Form

Example
Suppose that the augmented matrix for a linear system has been reduced
to the given matrix in row echelon form and the variables are also given.
Solve the system.

2 3 1 5 2
0 3 2 1 2
0 0 2 8 4
Variables: x, y , z, w
Solution The associated linear system is
2x + 3y z + 5w = 2
3y + 2z w = 2
2z 8w = 4
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

32 / 102

Matrix in Row Echelon Form

Solution (Cont.)
The variables x, y, and z corresponding to the pivots of the augmented
matrix are called leading variables (or dependent variables). The remaining
variables are called free variables (or independent variables).
The second step is to move the free variables to the right-hand side of the
equations
2x + 3y z = 2 5w
3y + 2z = 2 + w
2z = 4 + 8w
We let w = t, the system becomes in triangular form
2x + 3y z = 2 5t
3y + 2z = 2 + t
2z = 4 + 8t
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

33 / 102

Matrix in Row Echelon Form

Solution (Cont.)
It can be solved by backward substitution.
2z = 4 + 8t, thus, z = 2 4t
3y + 2(2 4t) = 2 + t
3y = 6 + 9t, y = 2 + 3t
2x + 3(2 + 3t) (2 4t) = 2 5t x = 3 9t
The solution is
(x, y , z, w ) = (3 9t, 2 + 3t, 2 4t, t), where t is any real number

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

34 / 102

Matrix in Row Echelon Form

Exercises
Reduce the matrix A below to echelon form.

0 3 6 4
9
1 2 1 3
1
A=
2 3 0
3 1
1
4
5 9 7
Hint

1 4
5
0 2
4

0 5 10
0 3 6

1
0
Echelon Form
0
0
Nguyen Minh Quan (HCMIU-VNU)

9 7
6 6

15 15
4
9
4
2
0
0

5 9 7
4 6 6

0 5 0
0 0
0

Chap 1 Linear systems and Matrices

Spring 2014

35 / 102

Gaussian elimination
Example
Solve the following system
2x y z = 3
6x + 6y + 5z = 3
4x + 4y + 7z = 3.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

36 / 102

Gaussian elimination
Example
Solve the following system
2x y z = 3
6x + 6y + 5z = 3
4x + 4y + 7z = 3.
First, find the augmented matrix



2 1 1  3
2 1 1  3
3R1 +R2 R2
6 6
5  3
2  6
0 3
2R1 +R3 R3

4
4
7
3
0 6
9  3

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

36 / 102

Gaussian elimination

2 1 1  3
2 1 1  3
2R +R3 R3
0 3
2  6
2  6 2
0 3

3
0 0
5  15
0 6
9

The matrix is now in row echelon form, so we find the associated system
2x y z = 3
3y + 2z = 6
5z = 15.
Thus, z = 3, then y = 4 and x = 2.
The solution is (x, y , z) = (2, 4, 3).

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

37 / 102

Gaussian elimination
Example
Solve the following system with three equations and three unknowns:
2x + y + 5z = 1
x 3y + 6z = 2
3x + 5y + 4z = 0.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

38 / 102

Gaussian elimination
Example
Solve the following system with three equations and three unknowns:
2x + y + 5z = 1
x 3y + 6z = 2
3x + 5y + 4z = 0.
First, write down the augmented matrix


2 1 5  1
2
1
5
1/2R1 +R2 R2
1 3 6  2

0 7/2 7/2


3/2R1 +R3 R3
3 5 4  0
0 7/2 7/2


1

 3/2

 3/2

Nguyen Minh Quan (HCMIU-VNU)

Spring 2014

Chap 1 Linear systems and Matrices

38 / 102

Gaussian elimination

2
1
5  1
R +R3 R3
2
0 7/2 7/2  3/2
0
0
0  0

The last row gives us no information whatsoever. If we let z = t

arbitrarily, then y = t 3/7.
Substituting this into the first row, we get x = (1 y 5z)/2 = 5/7 3t.
Thus, this system has an infinite number of solutions parameterized by the
parameter t, i.e., for every value of t, there corresponds a solution to the
system.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

39 / 102

Gaussian elimination
Example
Solve the following system with three equations and four unknowns:
x 2y z w = 4
3x + y + z 2w = 11
x + 12y + 7z + w = 31.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

40 / 102

Gaussian elimination
Example
Solve the following system with three equations and four unknowns:
x 2y z w = 4
3x + y + z 2w = 11
x + 12y + 7z + w = 31.

1 2 1 1  4
1
3R +R2 R2
3 1
1 2  11 1
0
R1 +R3 R3
1 12 7
1  31
0

1 2 1 1
2R +R3 R3
4
1
2
0 7
0 0
0
0

2 1 1  4
7
4
1  23
14 8
2  35


 4

 23

 11

The last equation can not be solved. So the system has no solutions.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

40 / 102

The Cost of Elimination

Q: How many separate arithmetical operations does elimination require,
for n equations in n unknowns?
Since all the steps are known, we should be able to predict the number of
operations.

Theorem
If n is at all large, a good estimate for the number of operations is 1/3n3 .

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

41 / 102

The Cost of Elimination

Q: How many separate arithmetical operations does elimination require,
for n equations in n unknowns?
Since all the steps are known, we should be able to predict the number of
operations.

Theorem
If n is at all large, a good estimate for the number of operations is 1/3n3 .
Proof:
(1) Suppose we call each division, and each multiplication-subtraction, one
operation.
In column 1, takes n operations for every zero we achieve-one to find the
multiple, and the other to find the new entries along the row.
(2) There are n 1 rows underneath the first one, so the first stage of
elimination needs n(n 1) = n2 n operations.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

41 / 102

The Cost of Elimination

Proof (cont.):
(3) When the elimination is down to k equations, only k 2 k operations
are needed to clear out the column below the pivot.
(4) Altogether, the total number of operations is the sum of k 2 k over
all values of k from 1 to n.
n
X
k=1

n (n + 1) (2n + 1) n (n + 1)

k2 k =
6
2
=

n3 n
3

Therefore, a good estimate for the number of operations is 1/3n3 .

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

42 / 102

Exercises: Gaussian elimination

Exercises
Solve the following system
2x + 5y z = 15
x y + 3z = 4
3x + 3y 5z = 2.
Answer: (x, y , z) = (1, 3, 2).

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

43 / 102

Gauss-Jordan elimination
Gauss-Jordan method uses row operations to transform the augmented
matrix of a linear system of equations into one of the form from which the
solution can be immediately read.
For example, if we employ row operations of the augmented matrix A and
obtain the following


1 0 0  c1


Row operations
Row operations 0 1 0  c2
A ... . . .


.
.
.. .. . . ..  ..

0 0 1  cn
Then, the solution is
x1 = c1 , x2 = c2 , ..., xn = cn
Recall: In Gauss-Jordan elimination, the reduced augmented matrix is in
reduced echelon form.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

44 / 102

Gauss-Jordan elimination
Example
Use Gauss-Jordan method to solve the system
2x1 3x2 + x3 = 2
x1 2x2 + 3x3 = 1
3x1 2x2 + 4x3 = 2
Solution:
The augmented matrix is given by


2 3 1  2
A = 1 2 3  1
3 2 4  2

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

45 / 102

Gauss-Jordan elimination
Solution (Cont.)
Employ row operations, we obtain



1 2 3  1
2 3 1  2
I12
2 3 1  2
A = 1 2 3  1


2
3 2 4  2
3 2 4


1 2 3  1
2R +R2 R2
1 5  4
1
0
3 2 4  2


1 2 3  1
3R +R3 R3
1

0 1 5  4
0 8 13  5

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

46 / 102

Gauss-Jordan elimination
Solution (Cont.)


1 0 7  7
2R +R1 R1
2

0 1 5  4
8R2 +R3 R3
0 0 27  27


 7
1
0
7
1

R R3
27 3

0 1 5  4
0 0 1  1


1 0 0  0
5R +R2 R2
3

0 1 0  1
7R3 +R1 R1
0 0 1  1

Then, the solution is

x1 = 0, x2 = 1, x3 = 1

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

47 / 102

Gauss-Jordan elimination
Example
Use Gauss-Jordan method to solve the system
3x1 + 2x2 x3 = 3
x1 x2 + 2x3 = 4
2x1 + 3x2 x3 = 3
Hint:

3 2 1
1 1 2
2 3 1

 3
1 0 0

Row Operations
 4 0 1 0

 3
0 0 1

 1

 1

 2

The solution is
x1 = 1, x2 = 1, x3 = 2
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

48 / 102

Gaussian elimination and Gauss-Jordan method

Exercises
Solve the following system
x 2y + z = 0
2y 8z = 8
4x + 5y + 9z = 9
Answer: (x, y , z) = (29, 16, 3).

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

49 / 102

Gaussian elimination and Gauss-Jordan method

Exercises
Solve the following system
2x + 8y z + w = 0
4x + 16y 3z w = 10
2x + 4y z + 3w = 6
6x + 2y + 5z + w = 3
Answer: (x, y , z, w ) = (3, 1/2, 4, 2).

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

50 / 102

Gaussian elimination and Gauss-Jordan method

Exercises
1. Using Gaussian elimination method and Gauss-Jordan method to solve
the linear system
2x + 3y z = 7
x y +z =1
4x 5y + 2z = 3
2. A certain brand of razor blades comes in packages of 6, 12, and 24
blades, costing $2, $3, and $4 per package, respectively. A store sold 12
packages, containing a total of 162 razor blades and took in $35. How
many packages of each type were sold?
Answer: (x, y , z) = (5, 3, 4).

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

51 / 102

Exercises
It is known that three brands of fertilizer (Fertifun, Big Grow and Soakem)
are available that provide Nitrogen, Phosphoric Acid and Soluble Potash to
the soil. One bag of each brand provides the following units of each
nutrient:

Nitrogen
Phosphoric Acid
Potash

Fertifun BigGrow Soakem

1 2 3
3 1 2
2 0 1

The soil of Dong Nai farm needs 18 units of nitrogen, 23 units of

phosphoric acid and 13 units of potash per acre. The corresponding units
for An Giang are 31, 24, and 11. How many bags of each brand of
fertilizer should be used per acre for each farm?
Hint: a. Dong Nai: (x, y , z) = (5, 2, 3).
b. An Giang: (x, y , z) = (2, 4, 7).
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

52 / 102

Product of two vectors

Product of a row matrix
and a column matrix of the same dimension: Let
A = a1 a2 an and B = ( b1 b2 bn )T . The product
AB is the 1 1 matrix whose entry is
Xn
a1 b1 + a2 b2 + ... + an bn =
ai b i
1

i.e., it is the dot product of the row vector A and the column vector B.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

53 / 102

Product of two vectors

Product of a row matrix
and a column matrix of the same dimension: Let
A = a1 a2 an and B = ( b1 b2 bn )T . The product
AB is the 1 1 matrix whose entry is
Xn
a1 b1 + a2 b2 + ... + an bn =
ai b i
1

i.e., it is the dot product of the row vector A and the column vector B.

Example
If
u = (3, 6, 2) , v = (4, 2, 4) ,
then
u v = 3 4 + 6 2 + 2 4 = 32
u u = 3 3 + 6 6 + 2 2 = 49 = kuk2

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

53 / 102

Basic Matrix Operations

Addition, Subtraction, Multiplication: creating new matrices.
a. Addition: Just add elements

 
a b
e
+
c d
g

a+e
c +g

b+f
d +h

b. Subtraction: Just subtract elements


 
 
a b
e f
ae

=
c d
g h
c g

bf
d h

f
h


=

Generally, if both matrices A and B are of the

a11 b11 a12 b12

a21 b21 a22 b22

AB =
..
..

.
.

same size, then

..
.

a1n b1n
a2n b2n
..
.

am1 bm1 am2 bm2 amn bmn

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

54 / 102

Basic Matrix Operations

c. Scalar Multiplication

2 3
10 15
5 4 1 = 20 5
1/5 6
1
30

A = A =

a11
a21
..
.

a12
a22
..
.

..
.

a1n
a2n
..
.

am1 am2 amn

d. Multiplication: Multiply each row by each column


 

a b
e f
ae + bg af + bh
=
c d
g h
ce + dg cf + dh
Is AB = BA? Maybe, but maybe not! In general, multiplication is NOT
commutative!
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

55 / 102

Matrix Multiplication
The product AB of two matrices A and B is defined if and only if the
number of columns of A is equal to the number of rows of B.
(A)mn (B)np = (AB)mp
Furthermore, the entry in the i-th row and j-th column of AB is given by
the dot product of the i-th row of A and the j-th column of B. That is,
If A = (aij )mn , B = (bij )np , then AB = C = (cij )mp where
cij = ai1 b1j + ai2 b2j + + ain bnj , i.e.,
!
n
X
AB =
aik bkj
k=1

Nguyen Minh Quan (HCMIU-VNU)

mp

Chap 1 Linear systems and Matrices

Spring 2014

56 / 102

Scalar Multiplication

Example
Let

18.95
p = 14.75
8.60

be a 3-vector that represents the current prices of three items at a store.

Suppose that the store announces a sale so that the price of each item is
reduced by 20%.
(a) Determine a 3-vector that gives the price changes for the three items.
(b) Determine a 3-vector that gives the new prices of the items.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

57 / 102

Scalar Multiplication
Solution
(a) Since each item is reduced by 20%, the 3-vector

(0.20) (18.95)
3.79
(0.20) p = (0.20) (14.75) = 2.95
(0.20) (8.60)
1.72
gives the price changes for the three items.
(b) The new prices of the items are given by the expression

18.95
3.79
15.16
p 0.20p = 14.75 2.95 = 11.80
8.60
1.72
6.88

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

58 / 102

Examples: Matrix Multiplication

Example

5 2 3

2
1 = 10 2 12 = 4.
4

Example


2 3 1
1 2 4

Nguyen Minh Quan (HCMIU-VNU)



5 3
4 0 = 25 7
15 1
3 1

Chap 1 Linear systems and Matrices

Spring 2014

59 / 102

Examples: Matrix Multiplication

Example

If A =

3 4 1
2 0 1

then


AB =

1 1 2
3
, B = 1 2
2
3 1

1 8 17
0 5 5

Example
Let

A=

1 x 3
2 1 1

2
and B = 4
y


12
If AB =
, find x and y .
6
Answer: x = 2, y = 6.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

60 / 102

Examples: Matrix Multiplication

Last month, Nguyen bought 3 pencils, 3 notebooks, and 1 eraser. This

month, he bought 3 pencils, 2 notebooks, and 5 erasers. The prices at IU
are $2, $5 and $1 respectively. Hence
QUANTITY PRICE = COST

Nguyen Minh Quan (HCMIU-VNU)

3 3 1
3 2 5



2
5 = 22
21
1

Chap 1 Linear systems and Matrices

Spring 2014

61 / 102

Exercise

Multiply

2 1 3
4 7 8
4
6 ,B = 9 3
5
A= 0
6 10 5
1 1 2

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

62 / 102

Exercises
A store brand X and brand Y dishwashers. The following matrices give the
sales figures and costs of these items for three months. Use matrix
multiplication to determine the total dollar sales and total costs of these
items for the three months.

Brand X
Brand Y

Dec. Apr . Aug .



18 10 12
,
19 12 14

Nguyen Minh Quan (HCMIU-VNU)

Retail Price
Dealer Cost

Chap 1 Linear systems and Matrices

X Y

350 260
240 190

Spring 2014

63 / 102

Properties of Matrix Addition

Theorem
Let A, B, and C be m n matrices.
(a) A + B = B + A.
(b) A + (B + C ) = (A + B) + C .
(c) There is a unique m n matrix O such that A + O = A, for any m n
matrix A. The matrix O is called the m n zero matrix
(d) For each m n matrix A, there is a unique m n matrix D such that
A + D = O, where O is the m n zero matrix.
We will write D as A.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

64 / 102

Properties of Matrix Addition

Theorem
Let A, B, and C be m n matrices.
(a) A + B = B + A.
(b) A + (B + C ) = (A + B) + C .
(c) There is a unique m n matrix O such that A + O = A, for any m n
matrix A. The matrix O is called the m n zero matrix
(d) For each m n matrix A, there is a unique m n matrix D such that
A + D = O, where O is the m n zero matrix.
We will write D as A.
Proof:
(a) Let A = [aij ], B = [bij ], A + B = C = [cij ], and B + A = D = [dij ].
By the definition of matrix addition, we have cij = aij + bij and
dij = bij + aij . Since aij and bij are real numbers, we get
aij + bij = bij + aij , which implies cij = dij for all i, j.
Thus, C = D, or A + B = B + A.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

64 / 102

Properties of Matrix Multiplication

Theorem
If A, B, and C are matrices of the appropriate sizes then
(a) A(BC ) = (AB)C .
(b) (A + B)C = AC + BC .
(c) C (A + B) = CA + CB.
Proof:

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

65 / 102

Properties of Scalar Multiplication

Theorem
If r and s are real numbers and A, B, and C are matrices of the
appropriate sizes then
(a) r (sA) = (rs)A
(b) (r + s)A = rA + sA
(c) r (A + B) = rA + rB
(d) A(rB) = r (AB) = (rA)B.
Proof:

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

66 / 102

Example: (AB)T = B T AT

0 1
1 3 2
A=
,B = 2 2
2 1 3
3 1




12 5
12 7
T
AB =
, (AB) =
7 3
5 3


On the other hand,



1 2
0 2 3
T
T

A = 3 1 , B =
1 2 1
2 3
T

B A =

Nguyen Minh Quan (HCMIU-VNU)

12 7
5 3

= (AB)T

Chap 1 Linear systems and Matrices

Spring 2014

67 / 102

Remark

Remark: If a, b, and c are real numbers for which ab = ac and a 6= 0 it

follows that b = c. That is, we can cancel out the nonzero factor a.
However, the cancellation law does NOT hold for matrices, as the
following example shows.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

68 / 102

Remark

Remark: If a, b, and c are real numbers for which ab = ac and a 6= 0 it

follows that b = c. That is, we can cancel out the nonzero factor a.
However, the cancellation law does NOT hold for matrices, as the
following example shows.
Let


A=

1 2
2 4


,B =

2 1
3 2


, and C =


AB = AC =

8 5
16 10

2 7
5 1

but B 6= C .

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

68 / 102

Remark

Remark: We also note two other peculiarities of matrix multiplication. If a

and b are real numbers, then ab = 0 can hold only if a or b is zero.
However, this is not true for matrices.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

69 / 102

Remark

Remark: We also note two other peculiarities of matrix multiplication. If a

and b are real numbers, then ab = 0 can hold only if a or b is zero.
However, this is not true for matrices.
Let


A=

1 2
2 4


, and B =

but


AB =

Nguyen Minh Quan (HCMIU-VNU)

0 0
0 0

4 6
2 3

Chap 1 Linear systems and Matrices

Spring 2014

69 / 102

Special Types of Matrices

An n n matrix A = [aij ] is called a diagonal
i 6= j.

a11 0

.
0 a22 0 ..

..
..
.
.
0
0
0
0
ann

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

matrix if aij = 0 for

Spring 2014

70 / 102

Special Types of Matrices

An n n matrix A = [aij ] is called a diagonal
i 6= j.

a11 0

.
0 a22 0 ..

..
..
.
.
0
0
0
0
ann

matrix if aij = 0 for

An n n diagonal matrix whose entries on the diagonal are all 1 is

called an identity matrix (or a unit matrix). It is denoted by In or
simply I .

1 0 0
I3 = 0 1 0
0 0 1
It is easy to show that In A = AIn for any square n n matrix A.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

70 / 102

Special Types of Matrices

An n n matrix A = [aij ] is called a diagonal
i 6= j.

a11 0

.
0 a22 0 ..

..
..
.
.
0
0
0
0
ann

matrix if aij = 0 for

An n n diagonal matrix whose entries on the diagonal are all 1 is

called an identity matrix (or a unit matrix). It is denoted by In or
simply I .

1 0 0
I3 = 0 1 0
0 0 1
It is easy to show that In A = AIn for any square n n matrix A.
A scalar matrix is a diagonal matrix whose diagonal elements are
equal.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

70 / 102

Special Types of Matrices

Suppose that A is a square matrix. The powers of a matrix, Ap , for p
a positive integer, is defined by
Ap = A.A...A

(p factors)

We also define A0 by the n n identity matrix, In .

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

71 / 102

Special Types of Matrices

Suppose that A is a square matrix. The powers of a matrix, Ap , for p
a positive integer, is defined by
Ap = A.A...A

(p factors)

We also define A0 by the n n identity matrix, In .

An n n matrix A = [aij ] is called
Example

1
A= 0
0

Nguyen Minh Quan (HCMIU-VNU)

upper triangular if aij = 0 for i > j.

7 5
2 1
0 2

Chap 1 Linear systems and Matrices

Spring 2014

71 / 102

Special Types of Matrices

Suppose that A is a square matrix. The powers of a matrix, Ap , for p
a positive integer, is defined by
Ap = A.A...A

(p factors)

We also define A0 by the n n identity matrix, In .

An n n matrix A = [aij ] is called
Example

1
A= 0
0

upper triangular if aij = 0 for i > j.

7 5
2 1
0 2

An n n matrix A = [aij ] is called lower

Example:

1 0
B = 1 2
2 4
Nguyen Minh Quan (HCMIU-VNU)

triangular if aij = 0 for i < j.

0
0
2

Chap 1 Linear systems and Matrices

Spring 2014

71 / 102

Transpose, Symmetric
The transpose of an m n matrix A = [aij ], denoted by AT is
obtained by interchanging the rows and columns of A, i.e.
[aij ]T = [aji ]
Example: Let

A=

1 2 3
4 8 7

1 4
AT = 2 8
3 7

A matrix A is symmetric if AT = A.

1
A = 1
4
Nguyen Minh Quan (HCMIU-VNU)

Example:

1 4
0 2
2 6

Chap 1 Linear systems and Matrices

Spring 2014

72 / 102

Properties of Transpose
Theorem
If r and s are real numbers and A, B, and C are matrices of the
appropriate sizes then
(a) (AT )T = A
(b) (A + B)T = AT + B T
(c) (AB)T = B T AT
(d) (rA)T = r (AT ).
Proof:

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

73 / 102

Skew Symmetric
A matrix A with real entries is called
Example:

0
A = 2
3

skew symmetric if AT = A.

2 3
0 0
4 0

Remark:

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

74 / 102

Skew Symmetric
A matrix A with real entries is called
Example:

0
A = 2
3

skew symmetric if AT = A.

2 3
0 0
4 0

Remark:
If A is symmetric or skew symmetric, then A is a square matrix
(Why?).
If A is a symmetric matrix, then the entries of A are symmetric with
respect to the main diagonal of A.
A is symmetric if and only if aij = aji , and A is skew symmetric if and
only if aij = aji .
If A is an n n matrix, then we can show that A = S + K , where S is
symmetric and K is skew symmetric (Exercise).
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

74 / 102

Diagonal

Let A be a square, n n, matrix. Then its (principal) diagonal

consists of the entries a11 , a22 , ..., ann .

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

75 / 102

Remark on representing linear systems via matrices

Consider the linear system of m equations in n unknowns,
a11 x1 + a12 x2 + ... + a1n xn = b1
a21 x1 + a22 x2 + ... + a2n xn = b2
..
.
am1 x1 + am2 x2 + ... + amn xn = bm

Then the linear system can be re-written as AX = b, where

a11 a12 a1n

a21 a22 a2n

T
T
A= .
.. , X = (x1 , x2 , ..., xn ) and b = (b1 , b2 , ..., bn )
..
.
.
.
.
.
.
.
am1 am2 amn
Q: Can we find X as X = A1 b? And what is A1 ?
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

76 / 102

Inverse of a Matrix
Definition
Let A be an n n matrix. A matrix B is said to be the inverse of A if
AB = BA = In ,
where In is the n n identity matrix. In this case, we denote B by A1 .
And A is said to be invertible or nonsinglar.

Example


1 2
3 4

Thus,




1 2
3 4

2
1
3/2 1/2

1


=

Nguyen Minh Quan (HCMIU-VNU)


=

2
1
3/2 1/2

2
1
3/2 1/2



1 2
3 4


=

1 0
0 1

 
1 

2
1
1 2
,
=
3/2 1/2
3 4

Chap 1 Linear systems and Matrices

Spring 2014

77 / 102

Inverse of a Matrix
If AB = In , then BA = In (why?).
Thus, to verify that B is an inverse of A, we need verify only that
AB = In .

Theorem
The inverse of a matrix, if it exists, is unique.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

78 / 102

Inverse of a Matrix
If AB = In , then BA = In (why?).
Thus, to verify that B is an inverse of A, we need verify only that
AB = In .

Theorem
The inverse of a matrix, if it exists, is unique.
Proof: Let B and C be inverses of A. Then AB = BA = In and
AC = CA = In . We then have
B = BIn = B(AC ) = (BA)C = In C = C ,
which proves that the inverse of a matrix, if it exists, is unique.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

78 / 102

Inverse of a Matrix
Theorem
If A and B are both nonsingular n n matrices, then AB is nonsingular and
(AB)1 = B 1 A1
Proof:...
Hint: Show that (AB)(B 1 A1 ) = In and (B 1 A1 )(AB) = In .

Corollary
If A1 , A2 ,..., and Ak are nonsingular n n matrices, then A1 A2 ...Ak is
nonsingular and
1
1
(A1 A2 ...Ak )1 = A1
k Ak1 ...A1

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

79 / 102

Inverse of a Matrix
Theorem
If A is a nonsingular matrix, then A1 is nonsingular and (A1 )1 = A.
Hint: Show that (A1 )A = A(A1 ) = In !

Theorem
If A is a nonsingular matrix, then AT is nonsingular and
(A1 )T = (AT )1
Proof: Taking transposes of the equation AA1 = In both sides, we get
A1
Similarly, AT A1
inverse of AT

T

AT = InT = In

T

= (A1 A)T = In . These equation implies that the

T
is A1 .

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

80 / 102

Inverse of a Matrix
Note: If A is nonsingular. Then AB = AC implies that B = C (Exercise).

Theorem
Let


A=

a b
c d

If ad bc 6= 0 then A is nonsingular and



1
d b
1
A =
ad bc c a

Example


3 4
5 6

3
2
5/2 3/2

A=
1

A
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

81 / 102

An algorithm for finding A1

Theorem
An n n matrix A is invertible if and only if A is row equivalent to In , and
in this case, any sequence of elementary row operations that reduces A to
In also transforms In to A.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

82 / 102

An algorithm for finding A1

To obtain A1 for any n n invertible matrix A, follow these steps

Algorithm for finding A1

1. Form the matrix [A I ], where I is the identity matrix.
2. Perform row operations on [A I ] to get a matrix of the form [I B].
3. Matrix B is A1 .

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

83 / 102

Finding A1

Example
Find the inverse of the matrix

1 3 3
A= 1 4 3
2 6 7
Solution: Step 1: We form the matrix

1 3
[A|I ] = 1 4
2 6

Nguyen Minh Quan (HCMIU-VNU)

[A|I ]

3 
3 
7 

as

1 0 0
0 1 0
0 0 1

Chap 1 Linear systems and Matrices

Spring 2014

84 / 102

Solution (Cont.)
Step 2: Using Row Operations on [A|I ]


1 3 3
1 3 3  1 0 0
R2 R1 R2


0 1 0
1 4 3  0 1 0

[A|I ] =
R3 2R1 R3
0 0 1
2 6 7  0 0 1

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

 1 0 0

 1 1 0

 2 0 1

Spring 2014

85 / 102

Solution (Cont.)
Step 2: Using Row Operations on [A|I ]


1 3 3
1 3 3  1 0 0
R2 R1 R2


0 1 0
1 4 3  0 1 0

[A|I ] =
R3 2R1 R3
0 0 1
2 6 7  0 0 1

 1 0 0

 1 1 0

 2 0 1

1 0 3  4 3 0
R 3R2 R1
1

0 1 0  1 1 0
0 0 1  2 0 1

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

85 / 102

Solution (Cont.)
Step 2: Using Row Operations on [A|I ]


1 3 3
1 3 3  1 0 0
R2 R1 R2


0 1 0
1 4 3  0 1 0

[A|I ] =
R3 2R1 R3
0 0 1
2 6 7  0 0 1


1 0 3  4
R 3R2 R1
1

0 1 0  1
0 0 1  2


1 0 0  10 3
R 3R3 R1
1

0 1 0  1 1
0 0 1  2 0

Nguyen Minh Quan (HCMIU-VNU)

 1 0 0

 1 1 0

 2 0 1

3 0
1 0
0 1

3
0 = [I |A1 ]
1

Chap 1 Linear systems and Matrices

Spring 2014

85 / 102

Solution (Cont.)
Step 2: Using Row Operations on [A|I ]


1 3 3
1 3 3  1 0 0
R2 R1 R2


0 1 0
1 4 3  0 1 0

[A|I ] =
R3 2R1 R3
0 0 1
2 6 7  0 0 1


1 0 3  4
R 3R2 R1
1

0 1 0  1
0 0 1  2


1 0 0  10 3
R 3R3 R1
1

0 1 0  1 1
0 0 1  2 0

 1 0 0

 1 1 0

 2 0 1

3 0
1 0
0 1

3
0 = [I |A1 ]
1

Step 3: Conclusion

A1
Nguyen Minh Quan (HCMIU-VNU)

10 3 3
0
= 1 1
2 0
1

Chap 1 Linear systems and Matrices

Spring 2014

85 / 102

An algorithm for finding A1

Example
Find the inverse of the matrix

0 1 2
A = 1 0 3 ,
4 3 8
if it exists.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

86 / 102

An algorithm for finding A1

Example (Cont.)

Since A I , A is nonsingular and

9/2 7 3/2
4
1
A1 = 2
3/2 2 1/2
Check: AA1 = I3 (?).
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

87 / 102

An algorithm for finding A1

Example

1 2 1
6 ,
Find the inverse of the matrix A = 1 5
5 4 5

if it exists.

So [A I ] is row equivalent to a matrix of the form [B D], where B has a

row of zeros. Further row operations will not transform B into I . Thus, A
does NOT have the inverse.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

88 / 102

An algorithm for finding A1

Theorem
If A is an invertible (nonsingular) n n matrix , then for each b in Rn , the
equation Ax = b has the unique solution x = A1 b.

Example
Solve the system


3x1 + 4x2 = 3
5x1 + 6x2 = 7


This system is equivalent to Ax = b, where A =

1

x =A

Nguyen Minh Quan (HCMIU-VNU)


b=

3
2
5/2 3/2



3
7

Chap 1 Linear systems and Matrices


3 4
, so
5 6


=

5
3

Spring 2014

89 / 102

The LU factorization
The LU factorization described below is motivated by the fairly common
industrial and business problem of solving a sequence of equations, all with
the same coefficient m n matrix A:
Ax = b1 , Ax = b2 , ..., Ax = bk
When A is invertible, one could find A1 then compute A1 b1 ,
A1 b2 ,...,A1 bk .
However it is more efficient when apply LU factorization, by re-write A in
the form A = LU, where L is an m m lower triangular matrix with 1s on
the diagonal, and U is an m n echelon form of A.
The matrix L is invertible and is called a unit lower triangular matrix.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

90 / 102

The LU factorization
Example of an LU factorization

Why they are so useful? We have


Ax = b LUx = b

Ly = b
Ux = y

We thus can solve x by solving the pairs of equations


Ly = b
Ux = y
First solve Ly = b for y , and then solve Ux = y for x.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

91 / 102

The LU factorization

An LU Factorization Algorithm
1. Reduce A to an echelon form U by a sequence of row replacement
operations, if possible.
2. Place entries in L such that the same sequence of row operations
reduces L to I .

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

92 / 102

The LU factorization
Example
Find an LU factorization of

2
4 1 5 2
4 5 3 8 1

A=
2 5 4 1
8
6 0
7 3 1

Solution:
The first column of L is the divided by the top pivot entry:

1 0 0 0
2 1 0 0

L=
1
1 0
3
1
Compare the first columns of A and L. The row operations that create
zeros in the first column of A will also create zeros in the first column of L.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

93 / 102

The LU factorization
Solution (Cont.)
To make this same correspondence of row operations on A hold for the
rest of L, watch a row reduction of A to an echelon form U.

The highlighted entries above determine the row reduction of A to U. At

each pivot column, divide the highlighted entries by the pivot and place
the result into L.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

94 / 102

The LU factorization
Solution (Cont.)

An easy calculation verifies that this L and U satisfy LU = A.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

95 / 102

The LU factorization
Example
Use the LU factorization to solve Ax = b, where

3 7 2 2
3 5
1
0
A=
6 4 0 5
9 5 5 12
and

Nguyen Minh Quan (HCMIU-VNU)

9
5

b=
7
11

Chap 1 Linear systems and Matrices

Spring 2014

96 / 102

The LU factorization
Hint

1
0 0
1 1 0
L=
2 5 1
3 8 3

0
3 7 2 2

0
0 2 1 2
,U =
0 0 1 1
0
1
0 0
0 1

Solve Ly = b for y :

Solve Ux = y for x:

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

97 / 102

Leontief input-output Model

Wassily Leontief (Nobel Prize in Economics in 1973) conducted a

monumental input-output analysis of the United States economy in the
1930s.
Assumptions:
Suppose the nations economy is divided into many sectors, such as
service industries, manufacturing, communication, and entertainment.

Suppose that for each sector we know its total output for one year
and we know exactly how this output is divided or exchanged among
the sectors of the economy.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

98 / 102

Leontief input-output Model

E.g., the 2nd column says that the total output of Electric is divided as
follows: 40% to Coal, 50% to Steel, and the remaining 10% to Electric.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

99 / 102

Leontief input-output Model

How much should each industry sector produce (output) in order to meet
demand for their commodities (so called equilibrium prices)?
Solution Denote the prices (in US dollars) of the total annual output of
Coal, Electric, and Steel sectors by x, y, and z, respectively. We want to
find x, y, z such that
x = 0.4y + 0.6z
y = 0.6x + 0.1y + 0.2z
z = 0.4x + 0.5y + 0.2z
That is,

0 0.4 0.6
x
x
0.6 0.1 0.2 y = y
0.4 0.5 0.2
z
z

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

100 / 102

Leontief input-output Model

Solution (cont.)
x 0.4y + 0.6z = 0
0.6x 0.9y 0.2z = 0
0.4x 0.5y + 0.8z = 0

The general solution is x = 0.94z, y = 0.85z, and z is free. Thus, the

equilibrium prices vector has the form (x, y , z)T = z(0.94, 0.85, 1)T for
any nonnegative z.
Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

101 / 102

Exercises (B. Kolman-D. Hill)

Section 1.1: 1-4, 16, 17, 18, 30, 32
Section 1.2: 1, 2, 4, 6, 9, 18, 20.
Section 1.3: 2, 3, 12, 15, 30, 31, 48, 49, 50.
Section 1.4: 13, 14, 16, 18, 26, 27, 37.
Section 1.5: 8, 12, 22, 35, 38, 48, 50.
Section 2.1: 11, 12.
Section 2.2: 5-8, 14, 15.

Nguyen Minh Quan (HCMIU-VNU)

Chap 1 Linear systems and Matrices

Spring 2014

102 / 102