Assiut University

Faculty of Science

Department of Mathematics

Course Title: Mathematics (II)

Course Code: MATH121

Course hours per week:

Lecture Tutorial / Total
Practical
3 2 5

For
Faculty of Computers & Information
By
Professor
OSAMA RASHED SAYED

-0-
 Table of Contents
CHAPTER I: MATRICES
Matrices 6
Exercise Set (1.1) 19
Elementary Row Operations 21
Exercise Set (1.2) 26
Inverse matrix 27
Exercise Set (1.3) 37
Special matrices 36
Exercise Set (1.4) 46
CHAPTER II: SYSTEMS OF LINEAR EQUATIONS
Introduction of systems of linear equations 47
Exercise Set (2.1) 67
Homogeneous system of linear equations 70
Exercise Set (2.2) 76
CHAPTER III: DETERMINATES
Determinants 78
Exercise Set (3.1) 84
Cramer' Rule 85
Exercise Set (3.2) 87
CHAPTER VII: LINEAR SPACE
Euclidean n-space 89
Linear Space 91
Exercise Set (4.2) 103
Subspaces 106
Exercise Set (4.3) 113
Linear combination and spanning sets 115
Exercise Set (4.4) 125
Linear Independence 127
Exercise Set (4.5) 135
Basis and Dimension 136
Exercise Set (4.6) 150

-1-
CHAPTER 8: INNER PRODUCT SPACES
Inner product 154
Exercise Set (5.1) 162
Orthogonally 164
Exercise Set (5.2) 173
Orthonormal Basis 176
Exercise Set (5.3) 187
EigenValues and EigenVectors 188
Exercise Set (5.4) 200
CHAPTER VI: PARTIAL FRACTIONS & SEQUENCES
Partial Fractions Decomposition 202
Exercise Set (6.1) 210
Sequence 211
Exercise Set (6.2) 223
Monotone Sequences 227
Exercise Set (6.3) 236
CHAPTER IV: INFINITE SERIES
Infinite Series 239
Exercise Set (7.1) 250
Convergence; The Integral Test 253
Exercise Set (7.2) 262
Additional Convergence Tests 264
Exercise Set (7.3) 269
Applying the Comparison Test 272
Exercise Set (7.4) 281
Alternating Series; Conditional Convergence 284
Exercise Set (7.5) 297
Power Series 300
Exercise Set (7.6) 309
Taylor and Maclaurin Series 311
Exercise Set (7.7) 323

-2-
CHAPTER V: DIFFERENTIAL EQUATION
Introduction 326
Definitions and Terminology 326
Initial-Value and Boundary-Value Problems 339
Solutions of First Order Differential Equations 347
Exercise Set (8.1) 360
MISCELLANEOUS PROBLEMS 363
REFERENCES

-3-
-4-
 CHAPTER (I)
MATRICES
1. MATRICES

Idea of matrices was introduced by Arthur Caylet in 19th

century. Rectangular arrays of real numbers arise in many
contexts other than as augmented matrices for systems of
linear equations. In this section, we shall consider such
arrays as objects and develop some of their properties for
use in our later work.
Definition 1.
A matrix is a rectangular array of numbers or square grid
of numbers arranged into rows and columns. The numbers
in the arrays are called the entries or elements.
Example 1.
Some examples of matrices are.
1 2 −√2 𝜋 𝑒 1
[ 3 0] , [ 2 1 0 − 3] , [ 3 1⁄2 0] , [ 3] , [ 4 ]
−1 4 0 0 0

●If 𝑎𝑖𝑗 , where 1 ≤ 𝑖 ≤ 𝑚 and 1 ≤ 𝑗 ≤ 𝑛 is the elements

of the matrix 𝐴, then we write

-5-
●We say that 𝐴 is a matrix of size (order) 𝑚 × 𝑛 (or m by
n matrix) where 𝑚 is the number of the rows of 𝐴 and n is
the number of the columns of 𝐴. So, in Example 1 the
matrices are of size 3 × 2 , 1 × 4, 3 × 3, 2 × 1, and
1 × 1, respectively.
● We will use R to denote for row and C for column.
●We will use capital letters A, B, C, … to denote for
matrices.
●We write 𝐴 = [𝑎𝑖𝑗 ] or 𝐴 = [𝑎𝑖𝑗 ] if the size is known.
𝑚×𝑛

●We say that 𝐴 is a square matrix of size n, if 𝑛 = 𝑚.

●An 1 × 𝑛 matrix is called a row matrix (or, row vector)
and written as [𝑎1 𝑎2 … 𝑎𝑛 ] and 𝑛 × 1 matrix is called a

-6-
 𝑎1
𝑎2
column matrix (or, vertical vector) and written as [ ⋮ ]. In
𝑎𝑛
Example 1 the 2 × 1 matrix is a column matrix, the 1 × 4
matrix is a row matrix, and 1 × 1 matrix is both a row
matrix and a column matrix.
Definition 2.
Let 𝐴 = [𝑎𝑖𝑗 ] and 𝐵 = [𝑏𝑖𝑗 ] be two matrices of the same
size. Then 𝐴 = 𝐵 if and only if 𝑎𝑖𝑗 = 𝑏𝑖𝑗 for all 𝑖 and 𝑗.
Example 2.
𝑎 𝑏 1 1 3
The matrices 𝐴 = [ ] and 𝐵 = [ ] are not
𝑐 𝑑 −2 0 5
equal because they are not of the same size.
5 −1
But, 𝐶 = [ ] is equal to 𝐴 if and only if 𝑎 = 5,
7 0
𝑏 = −1, 𝑐 = 7 and 𝑑 = 0.◄
Definition 3.
Let 𝐴 = [𝑎𝑖𝑗 ] and 𝐵 = [𝑏𝑖𝑗 ] be two matrices of size
𝑚 × 𝑛. Then 𝐴 + 𝐵 is of size 𝑚 × 𝑛 and is defined as 𝐴 +
𝐵 = [𝑎𝑖𝑗 + 𝑏𝑖𝑗 ].

-7-
Example 3.
1 3 −3 −3 −2 0
[−1 4 ]+[ 0 4 ] = [−1 8].
0 −5 5 6 5 1
Example 4.
𝑎 3 4 8 1 11
Let [ ]+[ ]=[ ]. Find 𝑎, 𝑏 and 𝑐.
1 𝑏 𝑐 −1 5 −4
Solution:
𝑎 3 4 8 𝑎+4 11
Since [ ]+[ ]=[ ],
1 𝑏 𝑐 −1 1+𝑐 𝑏−1
𝑎 + 4 = 1, 1 + 𝑐 = 5, 𝑏 − 1 = −4. Therefore, we
obtain 𝑎 = 𝑏 = −3 and 𝑐 = 4.◄
Definition 4.
A matrix whose elements are zeros is called a zero-matrix
and is denoted by 𝝑 (or [0]).
Definition 5.
Let 𝐴 = [𝑎𝑖𝑗 ] be a matrix and 𝑘 ∈ ℝ. The matrix 𝑘𝐴 is
defined as 𝑘𝐴 = [𝑘𝑎𝑖𝑗 ].
Notes:
●If 𝑘 = −1 , we write – 𝐴 instead of (−1)𝐴.
●𝐴 − 𝐵 = 𝐴 + (−𝐵).

-8-
Theorem 1.
Let [𝑎𝑖𝑗 ] , 𝐵 = [𝑏𝑖𝑗 ] and 𝐶 = [𝑐𝑖𝑗 ] be three 𝑚 × 𝑛
matrices; and 𝑟, 𝑠 ∈ ℝ . Then
(1) 𝐴 + 𝐵 = 𝐵 + 𝐴;
(2) 𝐴 + (𝐵 + 𝐶 ) = (𝐴 + 𝐵) + 𝐶;
(3) 𝐴 + 𝝑 = 𝐴;
(4) 𝐴 + (−𝐴) = 𝝑;
(5) 𝑟(𝐴 + 𝐵) = 𝑟𝐴 + 𝑟𝐵;
(6) (𝑟 + 𝑠)𝐴 = 𝑟𝐴 + 𝑠𝐴;
(7) (𝑟𝑠)𝐴 = 𝑟(𝑠𝐴);
(8) 1𝐴 = 𝐴.

Proof.
(1) Note that:
𝐴 + 𝐵 = [𝑎𝑖𝑗 ] + [𝑏𝑖𝑗 ] = [𝑎𝑖𝑗 + 𝑏𝑖𝑗 ]
= [𝑏𝑖𝑗 + 𝑎𝑖𝑗 ] = 𝐵 + 𝐴.
Therefore 𝐴 + 𝐵 = 𝐵 + 𝐴.
(2) 𝐴 + (𝐵 + 𝐶 ) = [𝑎𝑖𝑗 ] + ([𝑏𝑖𝑗 ] + [𝑐𝑖𝑗 ])
= [𝑎𝑖𝑗 ] + [𝑏𝑖𝑗 + 𝑐𝑖𝑗 ]
= [𝑎𝑖𝑗 + (𝑏𝑖𝑗 + 𝑐𝑖𝑗 )]

-9-
 = [(𝑎𝑖𝑗 + 𝑏𝑖𝑗 ) + 𝑐𝑖𝑗 ]
= [𝑎𝑖𝑗 + 𝑏𝑖𝑗 ] + [𝑐𝑖𝑗 ]
= ([𝑎𝑖𝑗 ] + [𝑏𝑖𝑗 ]) + [𝑐𝑖𝑗 ]
= (𝐴 + 𝐵) + 𝐶.
Therefore 𝐴 + (𝐵 + 𝐶 ) = (𝐴 + 𝐵) + 𝐶.
(3) 𝐴 + 𝝑 = [𝑎𝑖𝑗 ] + [0] = [𝑎𝑖𝑗 + 0] = [𝑎𝑖𝑗 ] = 𝐴.
Therefore 𝐴 + 𝝑 = 𝐴.
(4) 𝐴 + (−𝐴) = [𝑎𝑖𝑗 ] + [−𝑎𝑖𝑗 ] = [𝑎𝑖𝑗 − 𝑎𝑖𝑗 ]
= [0] = 𝝑.
Therefore 𝐴 + (−𝐴) = 𝝑.
(5) 𝑟(𝐴 + 𝐵) = 𝑟([𝑎𝑖𝑗 ] + [𝑏𝑖𝑗 ]) = 𝑟[𝑎𝑖𝑗 + 𝑏𝑖𝑗 ]
= [𝑟(𝑎𝑖𝑗 + 𝑏𝑖𝑗 )] = [𝑟𝑎𝑖𝑗 + 𝑟𝑏𝑖𝑗 ]
= [𝑟𝑎𝑖𝑗 ] + [𝑟𝑏𝑖𝑗 ] = 𝑟[𝑎𝑖𝑗 ] + 𝑟[𝑏𝑖𝑗 ]
= 𝑟𝐴 + 𝑟𝐵.
Therefore 𝑟(𝐴 + 𝐵) = 𝑟𝐴 + 𝑟𝐵.
(6) (𝑟 + 𝑠)𝐴 = (𝑟 + 𝑠)[𝑎𝑖𝑗 ] = [(𝑟 + 𝑠)𝑎𝑖𝑗 ]
= [𝑟𝑎𝑖𝑗 + 𝑠𝑎𝑖𝑗 ] = [𝑟𝑎𝑖𝑗 ] + [𝑠𝑎𝑖𝑗 ]
= 𝑟[𝑎𝑖𝑗 ] + 𝑠[𝑎𝑖𝑗 ] = 𝑟𝐴 + 𝑠𝐴.
Therefore (𝑟 + 𝑠)𝐴 = 𝑟𝐴 + 𝑠𝐴.

- 10 -
(7) (𝑟𝑠)𝐴 = (𝑟𝑠)[𝑎𝑖𝑗 ] = [((𝑟𝑠)𝑎𝑖𝑗 )]

= [(𝑟(𝑠𝑎𝑖𝑗 ))] = 𝑟[𝑠𝑎𝑖𝑗 ] = 𝑟(𝑠[𝑎𝑖𝑗 ])

= 𝑟(𝑠𝐴).
Therefore (𝑟𝑠)𝐴 = 𝑟(𝑠𝐴).
(8) Obvious. ◄
Example 5.
0 −1 −1 0
Let 𝐴 = [2 5 ] and 𝐵 = [ 5 −1]. Find 3𝐴 − 𝐵.
1 3 3 2
Solution.
0 −1 −1 0 1 −3
3𝐴 − 𝐵 = 3 [2 5 ]−[ 5 −1] = [1 16 ].
1 3 3 2 0 7
Example 6.
Solve the following matrix equation.
0 −2 2 −1
𝐴 + 3[ ]=[ ].
1 −1 4 0
Solution.
Note that the size of the matrix 𝐴 should be 2 × 2. So, we
assume that.
𝑎 𝑏 2 −1 0 −6 2 5
𝐴=[ ]=[ ]−[ ]=[ ].◄
𝑐 𝑑 4 0 3 −3 1 3

- 11 -
Definition 6.
If 𝐴 = [𝑎𝑖𝑗 ] is any 𝑚 × 𝑛 matrix, then the transpose of
𝐴, denoted by 𝐴𝑇 , is the 𝑛 × 𝑚 matrix obtained from 𝐴 by
interchanging the rows and columns of 𝐴, 𝑖. 𝑒. 𝐴𝑇 = [𝑎𝑗𝑖 ].

Theorem 2.
Let 𝐴 = [𝑎𝑖𝑗 ] and 𝐵 = [𝑏𝑖𝑗 ] be an 𝑚 × 𝑛 matrices. The
following statements are true.
(1) (𝐴𝑇 )𝑇 = 𝐴.
(2) (𝑘𝐴)𝑇 = 𝑘𝐴𝑇 ,where𝑘 ∈ ℝ.
(3) (𝐴 + 𝐵)𝑇 = 𝐴𝑇 + 𝐵𝑇 .

𝑇
Proof. (𝐴𝑇 )𝑇 = [𝑎𝑗𝑖 ] = [𝑎𝑖𝑗 ] = 𝐴;
𝑇
(1) (𝑘𝐴)𝑇 = [𝑘𝑎𝑖𝑗 ] = [𝑘𝑎𝑗𝑖 ] = 𝑘[𝑎𝑗𝑖 ] = 𝑘𝐴𝑇 ;
𝑇
(2) (𝐴 + 𝐵)𝑇 = [𝑎𝑖𝑗 + 𝑏𝑖𝑗 ] = [𝑎𝑗𝑖 + 𝑏𝑗𝑖 ]
= [𝑎𝑗𝑖 ] + [𝑏𝑗𝑖 ] = 𝐴𝑇 + 𝐵𝑇 .
♣ We noted that sum of matrices is nearly like the addition
of real numbers. But the product of matrices is not like the

- 12 -
product of real numbers. Before we define the product of
matrices, we define the Euclidean product.
Definition 7.
𝑏1
𝑏
Let 𝐴 = [𝑎1 𝑎2 … 𝑎𝑛 ] be a row vector and 𝐵 = [ 2 ] be a
:
𝑏𝑛
column vector. We define the Euclidean product of 𝐴 and
B as follows:
𝐴𝐵 = 𝑎1 𝑏1 + 𝑎2 𝑏2 + ⋯ + 𝑎𝑛 𝑏𝑛 .
Definition 8.
Let 𝐴 = [𝑎𝑖𝑗 ] be an 𝑚 × 𝑛 matrix and 𝐵 = [𝑏𝑖𝑗 ] be an
𝑛 × 𝑝 matrix.
The product of 𝐴 and 𝐵, written as 𝐴𝐵, is the 𝑚 × 𝑝
matrix 𝐶 = [𝑐𝑖𝑗 ] whose 𝑖𝑗 - component is
𝑛

𝑐𝑖𝑗 = 𝑎𝑖1 𝑏1𝑗 + 𝑎𝑖2 𝑏2𝑗 + ⋯ + 𝑎𝑖𝑛 𝑏𝑛𝑗 = ∑ 𝑎𝑖𝑘 𝑏𝑘𝑗

𝑘=1

Note that 𝐴𝐵 is defined if and only if 𝐴 has the same

number of columns as 𝐵 has rows.
If 𝐴 is an 𝑛 × 𝑛 matrix, then 𝐴𝐴 is defined and denoted by
𝐴2 .

- 13 -
Example 6.
𝑎11 𝑎12
𝑏 𝑏12 𝑏13
𝐴 = [𝑎21 𝑎22 ], 𝐵 = ൤ 11 ൨.
𝑎31 𝑎32 𝑏21 𝑏22 𝑏23

𝐴𝐵
𝑎11 𝑏11 + 𝑎12 𝑏21 𝑎11 𝑏12 + 𝑎12 𝑏22 𝑎11 𝑏13 + 𝑎12 𝑏23
= ቎𝑎21 𝑏11 + 𝑎22 𝑏21 𝑎21 𝑏12 + 𝑎22 𝑏22 𝑎21 𝑏13 + 𝑎22 𝑏23 ቏
𝑎31 𝑏11 + 𝑎32 𝑏21 𝑎31 𝑏12 + 𝑎32 𝑏22 𝑎31 𝑏11 + 𝑎32 𝑏23
𝐵𝐴
𝑏11 𝑎11 + 𝑏12 𝑎21 + 𝑏13 𝑎31 𝑏11 𝑎12 + 𝑏12 𝑎22 + 𝑏13 𝑎32
=൤ ൨
𝑏21 𝑎11 + 𝑏22 𝑎21 + 𝑏23 𝑎31 𝑏21 𝑎12 + 𝑏22 𝑎22 + 𝑏23 𝑎32
Example 7.
1 2
0 −1 2
Let 𝐴 = [−1 0] and 𝐵 = [ ].
−4 1 0
0 3
Compute 𝐴𝐵 and 𝐵𝐴, if it is possible?
Solution.
Since 𝐴 is of size 3 × 2 and 𝐵 is of size 2 × 3, then 𝐴𝐵 is
of size 3 × 3 and 𝐵𝐴 is of size 2 × 2. Now
1 2 −8 1 2
0 −1 2
𝐴𝐵 = [−1 0] [ ]=[ 0 1 −2].
−4 1 0
0 3 −12 3 0

- 14 -
 1 2
0 −1 2 1 6
Also 𝐵𝐴 = [ ] [−1 0] = [ ].◄
−4 1 0 −5 −8
0 3
Notes.
(1) 𝐴𝐵 ≠ 𝐵𝐴, 𝑖. 𝑒., The product of matrices is not
commutative.
(2) It is possible that that 𝐴 ≠ 𝝑 and 𝐵 ≠ 𝝑 such that
𝐴𝐵 = 𝝑.
For example,
1 0 0 0 0 0
𝐴=[ ],𝐵 = [ ] and 𝐴𝐵 = [ ].
0 0 0 1 0 0

Theorem 3.
Let 𝐴 = [𝑎𝑖𝑗 ], 𝐵 = [𝑏 𝑖𝑗 ] and 𝐶 = [𝑐𝑖𝑗 ] be matrices with
suitable size such that the following operations are defined
and 𝑘 ∈ ℝ. Then:
(1) 𝐴(𝐵𝐶 ) = (𝐴𝐵 )𝐶
(2) 𝐴(𝐵 + 𝐶 ) = 𝐴𝐵 + 𝐴𝐶
(3) (𝐵 + 𝐶 )𝐴 = 𝐵𝐴 + 𝐶𝐴
(4) 𝑘 (𝐴𝐵) = (𝑘𝐴)𝐵 = 𝐴(𝑘𝐵 )
(5) (𝐴𝐵 )𝑇 = 𝐵 𝑇 𝐴𝑇

- 15 -
Proof.
(1) Let 𝐴, 𝐵, 𝐶 have the size 𝑚 × 𝑟, 𝑟 × 𝑠, 𝑠 × 𝑛,
respectively.

Therefore
𝑟 𝑠 𝑟

(𝐴𝐵)𝐶 = [∑ 𝑎𝑖𝑘 𝑏𝑘𝑗 ] 𝐶 = [∑ (∑ 𝑎𝑖𝑘 𝑏𝑘𝑡 ) 𝐶𝑡𝑗 ]

𝑘=1 𝑡=1 𝑘=1
𝑠 𝑟

= [∑ ∑ 𝑎𝑖𝑘 𝑏𝑘𝑡 𝑐𝑡𝑗 ]

𝑡=1 𝑘=1
𝑟 𝑠

= [∑ 𝑎𝑖𝑘 (∑ 𝑏𝑘𝑡 𝑐𝑡𝑗 )]

𝑘=1 𝑡=1
𝑠

= 𝐴 [∑ 𝑏𝑘𝑡 𝑐𝑡𝑗 ] = 𝐴(𝐵𝐶 ).

𝑡=1

(2) Let 𝐴, 𝐵, 𝐶 have the size 𝑚 × 𝑛, 𝑛 × 𝑝, 𝑛 × 𝑝,

respectively. Then

𝐴(𝐵 + 𝐶 ) = 𝐴[𝑏𝑖𝑗 + 𝑐𝑖𝑗 ]

𝑛

= [∑ 𝑎𝑖𝑘 (𝑏𝑘𝑗 + 𝑐𝑘𝑗 )]

𝑘=1

- 16 -
 𝑛 𝑛

= [∑ 𝑎𝑖𝑘 𝑏𝑘𝑗 + ∑ 𝑎𝑖𝑘 𝑐𝑘𝑗 ]

𝑘=1 𝑘=1
𝑛 𝑛

= [∑ 𝑎𝑖𝑘 𝑏𝑘𝑗 ] + [∑ 𝑎𝑖𝑘 𝑐𝑘𝑗 ]

𝑘=1 𝑘=1

= 𝐴𝐵 + 𝐴𝐶.
(3) Similar to (2)
(4) Let 𝐴 be of size 𝑚 × 𝑛 and 𝐵 of size 𝑛 × 𝑝. So
𝑛 𝑛

𝑘(𝐴𝐵 ) = 𝑘 [∑ 𝑎𝑖𝑟 𝑏𝑟𝑗 ] = [∑(𝑘𝑎𝑖𝑟 )𝑏𝑟𝑗 ] = (𝑘𝐴)𝐵

𝑟=1 𝑟=1

Similarly, 𝑘(𝐴𝐵 ) = 𝐴(𝑘𝐵).

(5) Let A be of size 𝑚 × 𝑛 and 𝐵 of size 𝑛 × 𝑝. Then
𝐴𝑇 = [𝑎𝑗𝑖 ] and 𝐵 𝑇 = [𝑏𝑗𝑖 ]. Therefore
𝐵𝑇 𝐴𝑇 = [∑𝑛𝑘=1 𝑏𝑘𝑗 𝑎𝑗𝑘 ] = [∑𝑛𝑘=1 𝑎𝑗𝑘 𝑏𝑘𝑗 ] = (𝐴𝐵)𝑇 .◄
♣An 𝑛 × 𝑛 matrix 𝐴 = [𝑎𝑖𝑗 ] is called diagonal if 𝑎𝑖𝑗 = 0
whenever 𝑖 ≠ 𝑗.
The diagonal matrix is called identity matrix if 𝑎𝑖𝑗 = 1
for 𝑖 = 1,2, … . , 𝑛. We denote 𝐼𝑛 for the identity matrix of
size 𝑛 × 𝑛 or 𝐼 if the size is known. For every 𝑛 × 𝑛
matrix, we have 𝐴𝐼𝑛 = 𝐼𝑛 𝐴 = 𝐴.

- 17 -
Definition 9.
Let 𝐴 = [𝑎𝑖𝑗 ] be an 𝑛 × 𝑛 matrix. The trace of 𝐴 is
denoted by 𝑡𝑟(𝐴) and is defined as follows:
𝑛

𝑡𝑟 (𝐴) = ∑ 𝑎𝑖𝑖
𝑖=1

Example 8.
1 3 −2
Let 𝐴 = [0 5 7 ].
1 2 −3
Then 𝑡𝑟(𝐴) = 1 + 5 + (−3) = 3.◄
Theorem 4.
Let 𝐴 and 𝐵 be an 𝑛 × 𝑛 matrices and 𝑘 ∈ ℝ. Then the
following statements are true.
(1) 𝑡𝑟(𝐴 + 𝐵) = 𝑡𝑟(𝐴) + 𝑡𝑟(𝐵);
(2) 𝑡𝑟(𝐴𝑇 ) = 𝑡𝑟(𝐴);
(3) 𝑡𝑟(𝐴𝐵) = 𝑡𝑟(𝐵𝐴);
(4) 𝑡𝑟(𝑘𝐴) = 𝑘 𝑡𝑟(𝐴);
2
(5) 𝑡𝑟(𝐴𝐴𝑇 ) = ∑𝑖≠𝑗 𝑎𝑖𝑗 .
Proof. Left as exercise.

- 18 -
 Exercise Set (1.1)
3 0
4 −1 1 4 2
Let 𝐴 = [−1 2] , 𝐵 = [ ],𝐶 =[ ]
0 2 3 1 5
1 1
1 5 2 6 1 3
𝐷 = [−1 0 1]and 𝐸 = [−1 1 2]
3 2 4 4 1 3
In Exercise (1) to (30) Find:
(1) 𝐷 + 𝐸 (2) 𝐷 − 𝐸
(3) 5𝐴 (4) −7𝐶
(5) 2𝐵 − 𝐶 (6) 4𝐸 − 2𝐷
(7) −3(𝐷 + 2𝐸) (8) 𝐶 − 𝐶 𝑇
(9) 𝑡𝑟(𝐷) (10) 𝑡𝑟(𝐷 − 3𝐸 )
(11) 4𝑡𝑟(7𝐵) (12) 𝑡𝑟(𝐴)
(13) 2𝐴𝑇 + 𝐶 (14) 𝐷 𝑇 − 𝐸 𝑇
(15) (𝐷 − 𝐸 )𝑇 (16) 𝐵𝑇 + 5𝐶 𝑇
1 1
(17) 𝐶𝑇 𝐴 (18) 𝐶 𝑇 − 𝐶 𝑇
4 2

(19) 3𝐸 𝑇 − 2𝐷 𝑇 (20) 𝐴𝐵
(21) 𝐵𝐴 (22) 𝐴(𝐵𝐶 )
(23) (𝐶 𝑇 𝐵)𝐴𝑇 (24) 𝑡𝑟(𝐶 𝑇 𝐴𝑇 + 2𝐸 𝑇 )
(25) 4𝐵𝐶 + 2𝐵 (26) 𝐵𝑇 (𝐶𝐶 𝑇 − 𝐴𝑇 𝐴)
(27) 𝐷𝑇 𝐸 𝑇 − (𝐸𝐷)𝑇 (28) 3𝐴 − 𝐵𝑇 𝐶 𝑇

- 19 -
(29) 𝑡𝑟(𝐷𝐸 + 𝐸𝐷) (30) 𝑡𝑟(𝐷𝐸 − 𝐸𝐷)
(31) Find 𝐴 of 𝐵 when 2𝐴 − 3𝐵 = 6(𝐴 + 3𝐵)
(32) Let 𝐵 be an 𝑚 × 𝑛 matrix such that 𝐵 + 𝐴 = 𝐴 for
every 𝑚 × 𝑛 matrix 𝐴. Prove that 𝐵 = 𝝑.
(33) Find all 2 × 2 matrices such that 𝐴2 = 𝐼.
(34) Prove that there is not an 2 × 2 matrices 𝐴 and 𝐵
such that 𝐴𝐵 − 𝐵𝐴 = 𝐼.

- 20 -
2-Elementary row operations
The following operations, called elementary row
operations, on the rows of a matrix:
1- Interchange two rows;
2- Multiply a row by a non-zero constant;
3- Replace one row by the sum of itself and a constant
multiple of another row.

Definition 1.
Let A and B be two matrices. We say that A and B are row
equivalent, written 𝐴~𝐵 if we can obtain one of them from
the another by a finite number of elementary row
operations.
Notes.
●It is obvious that if 𝐴~𝐵, then the size of A is same as B
●For ease of exposition we shall denote the elementary
row operations by a shorthand notation.
● 𝑅𝑖𝑗 will indicate that the 𝑖𝑡ℎ and 𝑗𝑡ℎ rows have been
interchanged.
● 𝑘𝑅𝑖 will indicate that the 𝑖𝑡ℎ row is multiplied by a
nonzero constant 𝑘.

- 21 -
●𝑘𝑅𝑖𝑗 will indicate that 𝑘 times the 𝑖𝑡ℎ row is added to the
𝑗𝑡ℎ row and the sum becomes the new 𝑗𝑡ℎ row.
●If we obtained B from A by elementary row operation,
then we can obtain B from A by the inverse of the
elementary row operation as follows:
1
(a') 𝑅𝑗𝑖 ; (b') 𝑅𝑖 ; (c') −𝑘𝑅𝑖𝑗 .
𝑘

Definition 2. We say that a matrix is in a row-echelon

form (REF, for short) if it has the following properties: -
(1) The first entry of the first column (leading entry) is
nonzero (1, if it is possible) and all other entries of this
column are zeros.
(2) The first column that has a nonzero entry that is not
in the first has a nonzero entry in the second row, and
every entry in this column below the second row equals
zero.
(3) For 𝑘 ≥ 3, continue in the manner described (2).
That is, the first column has a nonzero 𝑘 𝑡ℎ entry has
every entry below the 𝑘 𝑡ℎ entry to zero. Moreover, this
column occurs to the right of the first column that
contains a nonzero (𝑘 − 1)𝑡ℎ entry.

- 22 -
 13 5 7
1 45
01 1 0
In particular, [0 14] and [ ] are in row-
00 1 3
0 00
00 0 1
2 3 5
echelon form, while [4 0 1] is not.
0 0 0
♣ A matrix in row-echelon form that satisfy the following
two properties is said to be in reduced row-echelon form
(RREF, for short):
● If a row does not consist entirely of zeros then the
leading entry of that row, is 1.
● Any column containing the leading entry of some row
consists entirely of zeros except for that single leading
entry. For example,
1 0 2
1 0 0 1 4 7 0
0 1 0
[0 1 0], [ ] and [0 0 0 1]
0 0 0
0 0 1 0 0 0 0
0 0 0
are in reduced row-echelon form, while,
1 0 1
1 1 0 1 2 6 6
0 0 0
[0 1 3], [ ] and [0 0 2 1]are not.
0 1 0
0 0 1 0 0 0 0
0 0 0

- 23 -
♣ A systematic process that transforms any matrix to a
matrix in row-echelon form is as follows:
1. Interchanging rows (if necessary) to replace a nonzero
number in the first entry of the first row.
2. Use the second elementary row operation as many times
necessary to replace a zero as the first entry in every row
below the first.
3. Hold the first row fixed, and interchange other rows (if
necessary) to give the second row a nonzero entry as for as
left as possible.
4. Hold the first two rows fixed; use the second elementary
row operation to replace a zero in every row after the
second row directly beneath the first nonzero entry in the
second row.
5. Continue to hold the first two rows fixed and
interchange other rows (if necessary) to give the third row
a nonzero entry as for left as possible.
6. Hold the first three rows fixed; use the second
elementary row operation to replace a zero in every row
after the third row directly beneath the first nonzero entry
in the third row.
7. Continue in this fashion row-echelon forms obtained.

- 24 -
Example 1.
22 2 4
Transform the matrix 𝐴 = [7 3 4 4] to a matrix in row-
61 2 1
echelon form.
Solution.
1
𝑅 1112
2 1
𝐴→ [ 7 3 4 4]
61 2 1
−7𝑅12 ,−6𝑅13 1 1 1 2
→ [0 −4 −3 −10]
0 −5 −4 −11
1 1 2
− 𝑅2 1 1 3 5
4
→ [0 1 ]
0−5 4 2
−4 −11
5𝑅23 1 1 1 2
→ [0 1 3/4 5/2]
0 0 −1/4 3/2
−4𝑅3 1 1 1 2
→ [0 1 3/4 5/2].
0 0 1 −6
Therefore row-echelon form of the matrix is:
1 1 1 2
[0 1 3/4 5/2]. ◄
0 0 1 −6

- 25 -
Example 3.
Put the matrix
5 2 2
𝐴 = [7 3 4]
6 1 2
in reduced row echelon-form.
Solution.
1
𝑅 1 2/5 2/5
5 1
𝐴→ [7 3 4 ]
6 1 2
−7𝑅12, −6𝑅13 1
2/5 2/5
→ [0 1/5 6/5 ]
0 −7/5 −2/5

5𝑅21 2/5 2/5 7/5 𝑅23 1 2/5 2/5

→ [0 1 6 ]→ [0 1 6 ]
0 −7/5 −2/5 0 0 8
1/8 𝑅3 1 2/5 2/5
→ [0 1 6 ]
0 0 1
2
(− ) 𝑅21 1 0 −2
5
→ [0 1 6 ]
0 0 1
2𝑅31, −6𝑅32 1 0 0
→ [0 1 0]. ◄
0 0 1

- 26 -
 Exercise Set (1.2)
Transform the following matrices to the reduced row-
echelon form: -
1 2 1 1
1 −1 4 2
1. [ ];
2 1 3 1
−3 1 1 4
1 3 1 7
2. [3 −2 4 1];
4 1 9 6
0 2 4
3. [1 −3 5 ];
3 −1 −1
1 3 −2 0 2 0
2 6 −5 −2 4 −3
4. [ ];
0 0 5 10 0 15
2 6 0 8 4 18
1 3 −2 −1 0
2 6 3 1 1
5. [ ];
0 0 1 1 0
2 6 4 2 −2
1 2 1
1 3 5
6. [ ];
2 5 6
3 7 7
1 1
6 4 −5
7. [2 −1]; 8. [ ]
2 −3 2
1 2
- 27 -
3. Inverse of Matrix
Definition 1.
Let A be an 𝑛 × 𝑛 matrix. The matrix B is an inverse
matrix of A if B is an 𝑛 × 𝑛 matrix and 𝐴𝐵 = 𝐵𝐴 = 𝐼. In
this case, we say that A is invertible or nonsingular.
Example 1.
7 3 1 −3
Prove that 𝐵 = [ ] is an inverse of 𝐴 = [ ].
2 1 −2 7
Solution. Since
1 −3 7 3 1 0
𝐴𝐵 = [ ][ ]=[ ] = 𝐼 and
−2 7 2 1 0 1
7 3 1 −3 1 0
𝐵𝐴 = [ ][ ]=[ ] = 𝐼.
2 1 −2 7 0 1
Then B is the inverse of A.◄
Example 1.
0 1
Prove that the matrix 𝐴 = [ ] has no inverse.
0 1
𝑎 𝑏
Solution. Let 𝐵 = [ ] be the inverse of A. Then
𝑐 𝑑
0 1 𝑎 𝑏 1 0
𝐴𝐵 = [ ][ ]=[ ].
0 1 𝑐 𝑑 0 1
Therefore 𝑐 = 0 and 𝑐 = 1 which is impossible. Then A
has no inverse. ◄

- 28 -
Theorem 1.
The inverse of a square matrix A is unique.
Proof.
Let B and C are inverses of A, then 𝐴𝐶 = 𝐶𝐴 = 𝐼 and
𝐴𝐵 = 𝐵𝐴 = 𝐼. Therefore
𝐶 = 𝐶𝐼 = 𝐶 (𝐴𝐵 ) = (𝐶𝐴)𝐵 = 𝐼𝐵 = 𝐵.◄
Note.
We will write 𝐴−1 for the inverse of 𝐴
Definition 2.
Let A be an 𝑛 × 𝑛 matrix and k be positive integer. We
define 𝐴𝑘 inductively as: 𝐴𝑘+1 = 𝐴𝑘 𝐴 and 𝐴0 = 𝐼 ,
𝑘 ≥ 0. Also, 𝐴−𝑘 = (𝐴−1 )𝑘 .
Theorem 2.
(1) 𝐼−1 = 𝐼;
(2) (𝐴−1 )−1 = 𝐴
(3) If both A and B have an inverse, then AB has an
inverse and (𝐴𝐵 )−1 = 𝐵−1 𝐴−1 ;
(4) If 𝐴1 , 𝐴2 , … . , 𝐴𝑘 have an inverse, then 𝐴1 𝐴2 … 𝐴𝑘 has
−1 −1
an inverse (𝐴1 𝐴2 … 𝐴𝑘 )−1 = 𝐴−1
𝑛 … 𝐴2 𝐴1 .

- 29 -
(5) If A has an inverse, then 𝐴𝑘 has an inverse for all 𝑘 ≥
1 and (𝐴𝑘 )−1 = (𝐴−1 )𝑘
(6) If A has an inverse and 0 ≠ 𝑟 ∈ ℝ, then 𝑟𝐴 has an
1
inverse and (𝑟𝐴)−1 = 𝐴−1
𝑟

(7) If A has an inverse, then 𝐴𝑇 has an inverse and

(𝐴𝑇 )−1 = (𝐴−1 )𝑇

Proof.
(1) Since 𝐼𝐼 = 𝐼, then 𝐼 −1 = 𝐼.
(2) Since 𝐴−1 exists and 𝐴𝐴−1 = 𝐼 = 𝐴−1 𝐴, by
uniqueness of the inverse of 𝐴, 𝐴 is the inverse of 𝐴−1 .
Therefore (𝐴−1 )−1 = 𝐴.
(3) Since (𝐴𝐵)(𝐵−1 𝐴−1 ) = 𝐴(𝐵𝐵−1 )𝐴−1
= 𝐴𝐼𝐴−1
= 𝐴𝐴−1 = 𝐼
and
(𝐵−1 𝐴−1 )(𝐴𝐵) = 𝐵−1 (𝐴−1 𝐴)𝐵
= 𝐵−1 𝐼𝐵
= 𝐵𝐵−1 = 𝐼.
Then (𝐴𝐵 )−1 = 𝐵−1 𝐴−1 .

- 30 -
(4) We use induction.
The statement is true for 𝑘 = 1. Also, it is true for 𝑘 = 2.
Now, suppose that the statement is true at 𝑘 = 𝑚, 𝑚 > 2,
i.e.,
(𝐴1 𝐴2 … 𝐴𝑚 )−1 = 𝐴−1 −1 −1
𝑚 … 𝐴2 𝐴1 .

We prove the statement is true for 𝑘 = 𝑚 + 1.

(𝐴1 𝐴2 … 𝐴𝑚 𝐴𝑚+1 )−1 = 𝐴−1
𝑚+1 (𝐴1 𝐴2 … 𝐴𝑚 )
−1

= 𝐴−1 −1 −1 −1
𝑚+1 𝐴𝑚 … 𝐴2 𝐴1 .

So, the statement is true for all 𝑘.

(3) We put 𝐴1 = 𝐴2 = ⋯ = 𝐴𝑘 = 𝐴 in (4).
(4) Note that
1 1
(𝑟𝐴) ( 𝐴−1 ) = 𝑟 ( ) 𝐴𝐴−1 = 1𝐼 = 𝐼,
𝑟 𝑟
1 1
( 𝐴−1 ) (𝑟𝐴) = ( ) 𝑟𝐴−1 𝐴 = 1𝐼 = 𝐼.
𝑟 𝑟
1
Then (𝑟𝐴)−1 = 𝐴−1
𝑟

(5) 𝐴𝑇 (𝐴−1 )𝑇 = (𝐴−1 𝐴)𝑇 = 𝐼𝑇 = 𝐼,

(𝐴−1 )𝑇 𝐴𝑇 = (𝐴𝐴−1 )𝑇 = 𝐼𝑇 = 𝐼.
Therefore
(𝐴𝑇 )−1 = (𝐴−1 )𝑇 .◄

- 31 -
Definition 3.
Any matrix can be obtained from 𝐼𝑛 by using only one
elementary row operation is called an elementary matrix.
Example 3.
1 0
The matrix 𝐸 = [ ] is an elementary matrix, because
0 5
we obtained it from 𝐼2 by only one elementary row
operation (5𝑅2 ). Also,
1 0 5
𝐹 = [0 1 0]
0 0 1
is an elementary matrix, because we obtained it from 𝐼3 by
only one elementary row operation (5𝑅31 ).
1 0 2 3
Now let 𝐴 = [2 −1 3 6]. If 𝐵 is the matrix obtained
1 4 4 0
from 𝐴 by the elementary row operation (5𝑅13 ), 𝑖. 𝑒.,
1 0 2 3
𝐵 = [2 −1 3 6 ], then 𝐸𝐴 = 𝐵, where𝐸 is the
6 4 14 15
elementary matrix obtained from 𝐼3 by(5𝑅13 ). Thus
100 1 0 23 1 0 2 3
𝐸𝐴 = [0 1 0] [2 −1 3 6] = [2 −1 3 6 ] = 𝐵.◄
501 1 4 40 6 4 14 15

- 32 -
Theorem 3.
Let 𝐴 be an 𝑛 × 𝑛 matrix and 𝐸 be elementary matrix
obtained from 𝐼𝑚 . Then 𝐸𝐴 is a matrix obtained from 𝐴
by the same elementary row operation which done in 𝐼𝑚
to obtained 𝐸.
Proof. We prove the theorem for one elementary row
operation. For example, multiplying 𝑐𝑅𝑘 . That is,
𝑎11 𝑎12 … 𝑎1𝑛
𝑎21 𝑎22 … 𝑎2𝑛
suppose that 𝐴 = [ ⋮ ⋮ ⋱ ⋮ ]and 𝐸 is the
𝑎𝑚1 𝑎𝑚2 … 𝑎𝑚𝑛
elementary matrix obtained from 𝐼𝑚 . By 𝑐𝑅𝑘 , 𝑖. 𝑒.,
10 … 0 𝑎11 𝑎12 … 𝑎1𝑛
01 … 0 𝑎21 𝑎22 … 𝑎2𝑛
⋮⋮ : ⋮ ⋮ ⋮ ⋮ ⋮
𝐸= . Then 𝐸𝐴 =
00… 𝑐 …0 𝑐𝑎𝑘1 𝑐𝑎𝑘2 … 𝑐𝑎𝑘𝑛
⋮⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮
[ 00 … 1] [ 𝑎𝑚1 𝑎𝑚2 … 𝑎𝑚𝑛 ]
𝑎11 𝑎12 … 𝑎1𝑛
𝑎21 𝑎22 … 𝑎2𝑛
𝑐𝑅𝐾 ⋮ ⋮ ⋮ ⋮
and 𝐴 → 𝑐𝑎 𝑐𝑎 …𝑐𝑎 = 𝐸𝐴 .
𝑘1 𝑘2 𝑘𝑛
⋮ ⋮ ⋮ ⋮
[ 𝑎𝑚1 𝑎𝑚2 … 𝑎𝑚𝑛 ]
Similarly, we can prove the other two statements. ◄

- 33 -
Theorem 4.
Elementary matrices are nonsingular, and the inverse of
an elementary matrix is another elementary matrix.
Proof.
Let 𝐸 be the elementary matrix obtained from 𝐼𝑛 by
multiplying its 𝑖 𝑡ℎ row by a nonzero constant c.
Let 𝐹 be the elementary matrix obtained from 𝐼𝑛 by
1
multiplying its 𝑖 𝑡ℎ row by . Then 𝐸𝐹 is the matrix
𝑐
1
obtained from 𝐹 by multiplying its 𝑖𝑡ℎ row by .
𝑐

Therefore 𝐸𝐹 = 𝐼𝑛 . Then the matrix 𝐸 is nonsingular and

𝐹 = 𝐸 −1 .
The proof of the following theorem is left as an exercise.
Theorem 5.
Let 𝐴 be an 𝑛 × 𝑛 matrix the following are equivalent:
(1) 𝐴 has an inverse.
(2) The reduced row-echelon form of 𝐴 is 𝐼𝑛 .
(3) 𝐴 can be written as the product of finite number of
elementary matrices.

- 34 -
♣ There is a relatively easy way to determine whether an
𝑛 × 𝑛 matrix has an inverse and to compute 𝐴−1 whether
it exists. To begin we construct an 𝑛 × 𝑛, called a
partitioned matrix and denoted by [𝐴|𝐼𝑛 ], by 𝐴 and 𝐼𝑛
next to each other:
𝑎11 𝑎12 … 𝑎1𝑛 1 0 … 0
𝑎 𝑎22 … 𝑎2𝑛 0 1 … 0]
[𝐴|𝐼𝑛 ] = [ 21
: : … : |: : : :
𝑎𝑛1 𝑎𝑛2 … 𝑎𝑛𝑛 0 0 … 1
We now attempt to change the left side of this matrix into
𝐼𝑛 using elementary row operation.
If this can be done, the right side of resulting matrix is
𝐴−1 . If the left side of [𝐴|𝐼𝑛 ] cannot changed into 𝐼𝑛 ,
then 𝐴 has no inverse.
Example 4.
1 1 1
−1
Compute 𝐴 , where 𝐴 = [−1 3 2].
2 1 1
Solution.
𝟏 1 1 1 0 0
[𝐴|𝐼𝑛 ] = [−1 3 2 | 0 1 0]
2 1 1 0 0 1

- 35 -
 1𝑅12 , −2𝑅13 𝟏 1 1 1 0 0
→ [0 4 3 | 1 1 0]
0 −1 −1 −2 0 1

𝑅231 1 1 1 0 0
→ [0 −1 −1 |−2 0 1]
0 4 3 1 1 0
−1𝑅2 1 1 1 1 0 0
→ [0 𝟏 1 |2 0 −1]
0 4 3 1 1 0
−4𝑅23 1 1 1 1 0 0
→ [0 𝟏 1 | 2 0 −1]
0 0 −1 −7 1 4
−1𝑅3 1 1 1 1 0 0
→ [0 1 1 |2 0 −1]
0 0 𝟏 7 −1 −4
−1𝑅23 ,−1𝑅31 1 1 0 −6 1 4
→ [0 𝟏 0 |−5 1 3]
0 0 𝟏 7 −1 −4

−1𝑅21 1 0 0 −1 0 1
→ [0 1 0 |−5 1 3 ].
0 0 1 7 −1 −4
Therefore
−1 0 1
−1
𝐴 = [−5 1 3]
7 −1 −4

- 36 -
Example 5.
Find the inverse of the matrix (if it exists)
1 2 3
𝐴 = [4 5 6]
7 8 9
Solution.
1 2 3 1 0 0
[4 5 6 |0 1 0]
7 8 9 0 0 1
−4𝑅12 ,−7𝑅13 1 2 3 1 0 0
→ [0 −3 −6 |−4 1 0]
0 −6 −12 −7 0 1
1 1 0 0
− 𝑅23 1 2 3 4 1
3
→ [0 1 2 | − 0]
0 −6 −12 3 3
−7 −2 1

1 0 0
6𝑅23 1 2 3 4 1
→ [0 1 2| − 0]
0 0 0 3 3
1 −2 1
Since the last row in the left matrix is zero, the matrix 𝐴
has no inverse. ◄

- 37 -
 Exercise Set (1.3)
𝑎 𝑏
1. Let 𝐴 = [ ] and 𝑎𝑑 − 𝑏𝑐 ≠ 0. Prove that 𝐴 has an
𝑐 𝑑
1 𝑎 −𝑏
inverse and 𝐴−1 = [ ]
𝑎𝑑−𝑏𝑐 −𝑐 𝑎
2. Use Exercise 1 to find the inverse of the following
matrix (if it exists).
1 2
(a) [ ];
1 3
−1 3
(b) [ ];
2 5
1/2 0
(c) ൤ ൨;
0 1/3
3 6
(d) [ ].
2 4
3. Compute the inverse of the following matrices (if it
exists).
1 7 12
(a) [5 −1 2 ];
2 3 5
1 2 1
(b) [3 5 0];
1 7 1
1 1 2
(c) [1 1 1];
1 2 4

- 38 -
 1 4 2
(d) [2 3 3];
4 1 4

−1 4 5 2
0 0 0 −1
(e) [ ];
1 −2−2 0
0 −1−1 0

1 2 3 0 0
1 1 1 0 0
(f) 0 1 5 1 0.
0 0 1 2 2
[0 0 0 0 1]

4. Write the following matrices as a product of

elementary matrices (if it is possible).
2 1
(a) [ ];
3 1
5 0
(b) [ ];
3 2
1 0 2
(c) [0 1 1];
2 1 6

1 0 −3
(d) [ 0 1 4 ].
−2 2 15

- 39 -
5. Prove that 𝐸 𝑡 is an elementary matrix.

6. Let 𝐴 be an invertible matrix and 𝐴𝐵 = 𝐴𝐶. prove that

𝐵 = 𝐶.

7. Let 𝐴 be an invertible matrix and 𝐴2 − 3𝐴 + 𝐼 = 0.

Prove that 𝐴−1 = (3𝐼 − 𝐴)

8. Let 𝐴 and 𝐵 be matrices of size. Prove

(a) 𝐴(𝐼 + 𝐵𝐴) = (𝐼 + 𝐴𝐵)𝐴
(b) (𝐼 + 𝐵𝐴)𝐵 = 𝐵(𝐼 + 𝐴𝐵).

- 40 -
4.Special Matrices
Definition 1.
An 𝑛 × 𝑛 matrix 𝐴 = [𝑎𝑖𝑗 ] is called upper triangular if
𝑎𝑖𝑗 = 0 whenever 𝑖 > 𝑗 and lower triangular if 𝑎𝑖𝑗 = 0
whenever 𝑖 < 𝑗. A matrix is called triangular if it is
either upper or lower triangular.

Definition 2.
Let 𝐴 be a square matrix:
(1) 𝐴 is symmetric if 𝐴𝑇 = 𝐴.
(2) 𝐴 is anti(skew)-symmetric if 𝐴𝑇 = −𝐴.
Note.
We considered only the matrices of real numbers.
Sometimes we need to study the matrices of complex
number, i. e., its elements are complex numbers.

- 41 -
Example 1.
1 2 −1 𝑖 𝑖 1+𝑖
Both 𝐴 = [ 2 3 7 ] and 𝐵 = [ 𝑖 2𝑖 2 − 𝑖]
−1 7 5 1+𝑖 2−𝑖 1
1 −1
are symmetric matrices, but 𝐶 = [ ] is not.
3 2
Example 2.
Both the matrices
0 2 −1 0 𝑖 1+𝑖
𝐴 = [−2 0 7 ] and 𝐵 = [ −𝑖 0 2 − 𝑖]
1 −7 0 −1 − 𝑖 −2 + 𝑖 0
are anti-symmetric.
Theorem 1.
Let 𝐴 be a square matrix and 𝑘 ∈ ℝ. Then
(1) 𝐴𝐴𝑇 is symmetric;
(2) 𝐴 + 𝐴𝑇 is symmetric;
(3) 𝐴 − 𝐴𝑇 is anti-symmetric;
(4) If 𝐴 is symmetric, then 𝑘𝐴 is symmetric.
(5) If 𝐴 is anti-symmetric, then 𝑘𝐴 is anti-symmetric.
(6) There exists a unique matrix 𝐵 and 𝐶, where 𝐵 is
symmetric and 𝐶 is anti-symmetric such that
𝐴 = 𝐵 + 𝐶.

- 42 -
Proof.
(1) Since(𝐴𝐴𝑇 )𝑇 = (𝐴𝑇 )𝑇 𝐴𝑇 = 𝐴𝐴𝑇 , 𝐴𝐴𝑇 is symmetric.
(2) Since (𝐴+𝐴𝑇 )𝑇 = 𝐴𝑇 + (𝐴𝑇 )𝑇 = 𝐴𝑇 + 𝐴 = 𝐴 + 𝐴𝑇 ,
𝐴 + 𝐴𝑇 is symmetric.
(3) Since (𝐴−𝐴𝑇 )𝑇 = 𝐴𝑇 + (−𝐴𝑇 )𝑇 = 𝐴𝑇 − (𝐴𝑇 )𝑇 =
𝐴𝑇 − 𝐴 = −(𝐴−𝐴𝑇 ), 𝐴−𝐴𝑇 is anti-symmetric.

(4) Since A is symmetric, 𝐴𝑇 = 𝐴. Therefore (𝑘𝐴)𝑇 =

𝑘𝐴𝑇 = 𝑘𝐴. Hence 𝑘𝐴 is symmetric.

(5) Since 𝐴 is anti-symmetric, 𝐴𝑇 = −𝐴. Therefore

(𝑘𝐴)𝑇 = 𝑘𝐴𝑇 = 𝑘(−𝐴) = −𝑘𝐴. So, 𝑘𝐴 is anti-
symmetric.

1
(6) Note that 𝐵 = (𝐴 + 𝐴𝑇 ) is symmetric and 𝐶 =
2
1 1
(𝐴 − 𝐴𝑇 ) is anti-symmetric. Then 𝐵 + 𝐶 = (𝐴 +
2 2
1
𝐴𝑇 ) + (𝐴 − 𝐴𝑇 ) = 𝐴. To prove the uniqueness, suppose
2

𝐷 is symmetric and 𝐸 is anti-symmetric such that 𝐴 =

𝐷 + 𝐸. Now, 𝐴𝑇 = (𝐷 + 𝐸 )𝑇 = 𝐷 𝑇 + 𝐸 𝑇 = 𝐷 − 𝐸.
Furthermore 2𝐷 = 𝐴 + 𝐴𝑇 and 2𝐸 = 𝐴 − 𝐴𝑇 .

- 43 -
 1 1
Then 𝐷 = (𝐴 + 𝐴𝑇 ) = 𝐵 and 𝐸 = (𝐴 − 𝐴𝑇 ) = 𝐶.
2 2

Hence 𝐵 and 𝐶 are unique. ■

Example 3.
1 5 1
If 𝐴 = [ 3 3 −1], then the symmetric matrix 𝐵 and
−1 2 4
the anti-symmetric matrix C, where 𝐴 = 𝐵 + 𝐶 are.
1 4 0
1
𝐵 = (𝐴 + 𝐴𝑇 ) = [4 3
1
2] and
2 1
0 4
2
0 1 1
1
𝐶 = (𝐴 − 𝐴𝑇 ) = [−1 0 −3/2].◄
2
−1 3/2 0
Definition 3.

Let 𝐴 = [𝑎𝑖𝑗 ] be a complex matrix. The 𝐴 = [𝑎𝑖𝑗 ] is

called the conjugate matrix for 𝐴, where 𝑎𝑖𝑗 is conjugate
of the complex number 𝑎𝑖𝑗 .
There following theorem gives the properties of the
conjugate matrix. It easily be proved by using the
properties of the complex numbers. The proof is left to
the reader as exercise.

- 44 -
Theorem 2.
Let 𝐴 = [𝑎𝑖𝑗 ] and 𝐵 = [𝑏𝑖𝑗 ] be complex matrices.

(1) ̿𝐴 = 𝐴;
(2) (𝑘𝐴) = 𝑘 𝐴, 𝑘 ∈ ℂ;

(3) 𝐴 + 𝐵 = 𝐴 + 𝐵, where 𝐴, 𝐵 have the same size;

(4) 𝐴𝐵 = 𝐴̅ 𝐵̅ , where 𝐴 is of size 𝑚 × 𝑛 and 𝐵 is of size
𝑛 × 𝑝.
Definition 4.
Let 𝐴 be a complex matrix. The conjugate transpose of 𝐴
𝑇
is denoted by 𝐴∗ and is defined as: 𝐴∗ = (𝐴) = 𝐴𝑇 .
Example 4.
0 1+𝑖
𝑖 0 1−𝑖 −𝑖
If 𝐴 = [ ], then 𝐴 = [ ]
2 3 − 2𝑖
−𝑖 2 3 + 2𝑖 𝑖
𝑇
0 2
∗
And 𝐴 = (𝐴) = [1 − 𝑖 3 + 2𝑖 ].
−𝑖 𝑖
Theorem 3.
(1) (𝐴∗ )∗ = 𝐴;
(2) (𝐴 + 𝐵)∗ = 𝐴∗ +𝐵 ∗ ;
(3) (𝑘𝐴)∗ = 𝑘𝐴∗ , 𝑘 ∈ ℂ ;
(4) (𝐴𝐵)∗ = 𝐵 ∗ 𝐴∗ .
- 45 -
Proof.
𝑇
𝑇 ∗ 𝑇 𝑇 𝑇
(1) (𝐴∗ )∗
= [(𝐴) ] = ((𝐴) ) = [(𝐴̿) ] = 𝐴;
𝑇 𝑇
(2) (𝐴 + 𝐵)∗ = (𝐴 + 𝐵) = (𝐴 + 𝐵) .
𝑡 𝑡
= (𝐴) + (𝐵) = 𝐴∗ + 𝐵 ∗ ;
𝑇 𝑇
(3) (𝑘𝐴)∗ = (𝑘 𝐴) = 𝑘(𝐴) = 𝑘𝐴∗
𝑇 𝑇 𝑇
(4) (𝐴𝐵)∗ = (𝐴𝐵) = (𝐴̅𝐵̅)𝑇 = (𝐵) (𝐴) = 𝐵∗ 𝐴∗ .■

Definition 5.
Let 𝐴 be a nonsingular complex matrix. 𝐴 is said to be
unitary if 𝐴−1 = 𝐴∗ .
Example 5.
3 4
𝑖
5 5
Prove that the matrix 𝐴 = ቎−4 3 ቏ is unitary.
𝑖
5 5

Solution.
3/5 −4/5
𝐴∗ = 𝐴𝑇 = ൤ ൨.
−4/5𝑖 −3/5𝑖
3/5 4/5𝑖 3/5 −4/5 1 0
Also, 𝐴𝐴∗ = ൤ ൨൤ ൨=[ ].
−4/5 3/5𝑖 −4/5𝑖 −3/5𝑖 0 1

Hence 𝐴−1 = 𝐴∗ .◄

- 46 -
Remark:

If 𝐴 is real matrix, then 𝐴 = 𝐴. Therefore 𝐴∗ = 𝐴𝑇 .

Hence 𝐴 is unitary if and only if 𝐴−1 = 𝐴𝑇 . The real
unitary matrix is called orthogonal matrix.
Example 6.
3 2 6
7 7 7
6 3 2
The matrix 𝐴 = − 7 7 7
is orthogonal matrix
2 6 3
[ − ]
7 7 7

because 𝐴𝑇 𝐴 = 𝐼, i.e., 𝐴−1 = 𝐴𝑇 .◄

Definition 6.
A square complex matrix 𝐴 is said to be Hermitian
matrix if 𝐴∗ = 𝐴 and anti-Hermitian matrix if 𝐴∗ = −𝐴.
Example 7.
−2 1−𝑖
−1 + 𝑖
The matrix 𝐴 = [ 1 + 𝑖 0 3 ] is
−1 − 𝑖 3 5
𝑖 2 1 + 2𝑖
Hermitian matrix, while 𝐵 = [ −2 5𝑖 −2 + 𝑖]
−1 + 2𝑖 2 + 𝑖 0
is anti-Hermitian matrix.◄

- 47 -
 Exercise Set (1.4)
1. Let 𝐴 and 𝐵 be symmetric matrices of size 𝑛.
Prove that
(a) 𝐴𝐵 + 𝐵𝐴 is symmetric;
(b) 𝐴𝐵 − 𝐵𝐴 is anti-symmetric;
(c) 𝐴𝐵 is symmetric if and only if 𝐴𝐵 = 𝐵𝐴.

2. Let 𝐴 and 𝐵 be anti-symmetric matrices of size n.

Prove that 𝐴𝐵 is anti-symmetric if and only if
𝐴𝐵 = −𝐵𝐴.
3. Prove that if 𝐴 is symmetric, 2𝐴2 − 3𝐴 + 𝐼 is
symmetric.
4. We say that the matrix A is idempotent, if 𝐴2 = 𝐴.
Prove that if 𝐴𝑇 𝐴 = 𝐴, 𝐴 is symmetric and idempotent.
5. Let 𝐴 and 𝐵 be matrices of size n.
Prove that:
(a) 𝐴𝑇 𝐵 + 𝐵 𝑇 𝐴 is symmetric.
(b) 𝐴𝑇 𝐵 − 𝐵 𝑇 𝐴 is anti-symmetric.

6. Determine whether the following matrices is unitary

and find the inverse of unitary one.

- 48 -
 𝑖𝑒 −𝑖𝑒
൤ 𝑒 𝑖𝑒 𝑒
1
(a) ൨;
√ 𝑖𝑒
2 −𝑖𝑒 −𝑖𝑒
3 4𝑖
(b) [ ];
−4 3𝑖
1 1 1
(c) [ 𝑖 𝑖 ];
√2 −1 + 1+
√2 √2

3+𝑖 1 − 𝑖 √3
൤√
1
(d) ൨;
2√2 1 + 𝑖 √3 1 − √3

7. Prove that if 𝐴 is unitary, then 𝐴∗ is unitary.

8. The complex square matrix 𝐴 is normal if and only if
𝐴𝐴∗ = 𝐴∗ 𝐴.
Prove that:
1 1+𝑖
(a) [ ] is normal.
1−𝑖 −3
(b) If 𝐴 is Hermitian, then 𝐴 is normal.
(c) If 𝐴 is unitary, then 𝐴 is normal.

9. Find 𝑎, 𝑏 and 𝑐 such that the following matrix 𝐴 is

Hermitian:
−1 𝑎 −𝑖
𝐴 = [3 − 5𝑖 0 𝑐]
𝑏 2 + 4𝑖 2

- 49 -
10. Let 𝐴 be an invertible matrix. Prove that 𝐴∗ is
invertible and (𝐴∗ )−1 = (𝐴−1 )∗ .
11. Let 𝐴 be a square matrix. Prove that 𝐴 + 𝐴∗ is
Hermitian.
12. Which of the following matrices is Hermitian and
which anti- Hermitian:
0 𝑖
[ ]
𝑖 2

𝑖 𝑖
[ ]
−𝑖 𝑖

1 1+𝑖
[ ]
1−𝑖 −3

−2 1−𝑖 −1 + 𝑖
[ 1+𝑖 0 3 ]
−1 − 𝑖 3 5

- 50 -
- 51 -
 CHAPTER (II)
Systems of linear equations
1. Introduction of systems of linear equations
The study of systems of linear equations and their solutions
is one of the major topics in Mathematic (II). In this section
we shall introduce same basic terminology and discuss a
method for solving such systems. We suppose that the
student is familiar with a simple of linear system and its
solution. In general, an arbitrary system of 𝑚 linear
equations in 𝑛 unknowns can be written as.
𝑎11 𝑥1 + 𝑎12 𝑥2 + ⋯ + 𝑎1𝑛 𝑥𝑛 = 𝑏1
𝑎21 𝑥1 + 𝑎22 𝑥2 + ⋯ + 𝑎2𝑛 𝑥𝑛 = 𝑏2
⋮⋮⋮⋮
𝑎𝑚1 𝑥1 + 𝑎𝑚2 𝑥2 + ⋯ + 𝑎𝑚𝑛 𝑥𝑛 = 𝑏𝑚
Where 𝑥1 , 𝑥2 , … … , 𝑥𝑛 are the unknowns and the
subscripted a's and b's denote constants.
If we mentally keep track of the location of +′𝑠, the 𝑥′𝑠,
and the = ′𝑠, a system of 𝑚 linear equations in 𝑛
unknowns can be abbreviated by writing only the matrix
equation 𝑨𝒙 = 𝒃, where

- 52 -
 𝑎11 𝑎12 . . 𝑎1𝑛 𝑥1 𝑏1
𝑎21 𝑎22 . . 𝑎2𝑛 𝑥2 𝑏2
𝐴=[ : : .. : ] , 𝑥 = [ : ] 𝑏 = [ ]
:
𝑎𝑚1 𝑎𝑚2 . . 𝑎𝑚𝑛 𝑥𝑛 𝑏𝑚
𝐴 is the coefficients matrix, 𝑏 is the constants matrix 𝑥 is
the unknown's matrix, and [𝐴|𝑏] is the augmented
matrix.
Note that the number of the rows in 𝐴 is equal the
number of the equations in the system and the number of
the columns is equal the number of the unknowns.
For example, the augmented matrix for the system of
equations
𝑥1 + 𝑥2 + 2𝑥3 = 9
2𝑥1 + 4𝑥2 − 3𝑥3 = 1
3𝑥1 + 6𝑥2 − 5𝑥3 = 0
is
1 1 2 9
[2 4 −3| 1]
3 6 −5 0
Remark. When constructing an augmented matrix, the
unknown must be written in the same order in each
equation.

- 53 -
The basic method for solving a system of linear equations
is to replace the given system by a new system that has the
same solution set but is easier to solve. This new system is
generally obtained by the elementary row operations.
Now, we introduce two methods using the augmented
matrix to solve a system of linear equations.
(1) Gaussian elimination.
To solve the given system (1) by Gaussian elimination we
put the augmented matrix for the system (1) in row-
echelon form, then the solution set of the system will be
evident by inspection or after a few simple steps.
(2) Gauss-Jordan-elimination:

This method is like Gauss-elimination, but we put the

augmented matrix in a reduced row-echelon form.
Note that the reduced row echelon form of a matrix is
unique. No matter how the row operations are varied.
Before introducing examples on Gauss and Gauss-Jordan
we have that following definition.

- 54 -
Definition 1.
A system of linear equations is consistent if it has a
solution. Otherwise is inconsistent.
Example 1.
Solve the following system by Gauss and Gauss-Jordan
elimination.
2𝑥2 + 4𝑥3 = 3,
𝑥1 − 3𝑥2 + 5𝑥3 = 1,
3𝑥1 − 𝑥2 − 𝑥3 = 1
Solution.
(a) Gaussian-elimination
The augmented matrix for the given system is
0 2 4 3
[𝐴|𝑏] = [1 −3 5 | 1].
3 −1 −1 1
Now
𝑅12 1 −3 5 1
[𝐴|𝑏] → [0 2 4 | 3]
3 −1 −1 1
(−3)𝑅13 1 −3 5 1
→ [0 2 4 | 3]
0 8 −16 −2

- 55 -
 1
( )𝑅2 1 −3 5 1
2
→ [0 1 2 | 3/2]
0 8 −16 −2
(−8)𝑅23 1 −3 5 1
→ [0 1 2 | 3/2 ]
0 0 −32 −14
(−32)𝑅3 1 −3 5
1 1
→ [0 1 2| 3/2 ]
0 0 1 7/16
The last augmented matrix is on row-echelon form.
Therefore, the equivalent system to the given system is
𝑥1 − 3𝑥2 + 5𝑥3 = 1,
3
𝑥2 + 2𝑥3 = ,
2
7
𝑥3 = .
16
7
Substituting 𝑥3 = into the second equation yields
16
5
𝑥2 = .
8

Finally, substituting 𝑥2 and 𝑥3 into the first equation

yields
11
𝑥1 = .
16

Hence, the unique solution for the given system is

11 5 7
(16 , 8 , 16). ◄

- 56 -
(b) Gauss – Jordan elimination.

We complete where we end in (a) to put the augmented

matrix in reduced row-echelon form.
1 −3 5 1
[0 1 2| 3/2 ]
0 0 1 7/16

(−2)𝑅32 ,(−5)𝑅31 1 −3 0 −19/16

→ [0 1 0| 5/8 ]
0 0 1 7/16

3𝑅21 1 0 0 11/16
→ [0 1 0| 5/8 ]
0 0 1 7/16
The corresponding system of equation is
11 5 7
𝑥1 = , 𝑥2 = , 𝑥3 = .
16 8 16
Therefore, the unique solution of the system is
11 5 7
(16 , 8 , 16) which we obtained in (a). ◄

- 57 -
Example 2.
Use Gauss and Gauss-Jordan to solve the following: -
𝑥1 − 3𝑥2 + 2𝑥3 + 𝑥4 = 2
3𝑥1 − 9𝑥2 + 10𝑥3 + 2𝑥4 = 9
2𝑥1 − 6𝑥2 + 4𝑥3 + 2𝑥4 = 4
2𝑥1 − 6𝑥2 + 8𝑥3 + 𝑥4 = 7
Solution.
(a) Gauss-elimination
1 −3 2 1 2
[𝐴|𝑏] = ቎ 3 −9 10 2 |9቏
2 −6 4 2 4
2 −6 8 1 7
1 −3 2 1 2
(−3)𝑅12 , (−2)𝑅13
=቎0 0 4 −1 | 3቏
(−2)𝑅14 0 0 0 0 0
0 0 4 −1 3
1 1 −3 2 1 2
( )𝑅2
1 −1/4 3/4
[0 0
4
→ | ]
0 0 0 0 0
0 0 4 −1 3
1 −3 2 1 2
(−4)𝑅24 0 0 1 −1/4 3/4
→ [ | ]
0 0 0 0 0
0 0 0 0 0

- 58 -
The entire matrix is now in row-echelon form. The
corresponding system of equations is
𝑥1 − 3𝑥2 + 2𝑥3 + 𝑥4 = 2,
1 3
𝑥3 − 𝑥4 = .
4 4

Since 𝑥1 and 𝑥3 correspond to leading 1's in the

augmented matrix, we call them leading variables. The
non-leading variables (in this ease 𝑥2 and 𝑥4 ) are called
free variables.
Solving for the leading variables in terms of the free
variable gives
𝑥1 = 2 + 3𝑥2 − 2𝑥3 − 𝑥4 ,
3 1
𝑥3 = + 𝑥4 .
4 4
3 1
Substituting 𝑥3 = + 𝑥4 into the first equation yields
4 4
1 3
𝑥1 = + 3𝑥2 − 𝑥4 ,
2 2
3 1
𝑥3 = + 𝑥4 .
4 4

Assign arbitrary values to the free variables, we obtain

𝑥4 = 𝑡 and 𝑥2 = 𝑠.
So, the general solution is
1 3 3 1
𝑥1 = + 3𝑠 − 𝑡, 𝑥2 = 𝑠, 𝑥3 = + 𝑡, 𝑥4 = 𝑡.◄
2 2 4 4

- 59 -
 Remark.
The arbitrary values that are assigned to the free variables
are often called parameters. Although we shall generally
use the letters 𝑟, 𝑠, 𝑡, … for the parameters, any letters that
do not conflict with the variable names may be used.
(b) Gauss-Jordan elimination.
We start from the end of (a). We put the augmented matrix
in reduced row-echelon form.
1 −3 2 1 2
1 −1/4 3/4
[0 0 | ]
0 0 0 0 0
0 0 0 0 0
1 −3 0 3/2 1/2
−2𝑅21 0 0 1 −1/4 3/4
→ [ | ]
0 0 0 0 0
0 0 0 0 0
The last augmented matrix is in reduced row-echelon
form and the corresponding system is
3 1
𝑥1 − 3𝑥2 + 𝑥4 = ,
2 2
1 3
𝑥3 − 𝑥4 =
4 4
Putting 𝑥2 = 𝑠 and 𝑥4 = 𝑡 we obtain the set of solutions
3 1 3 1
𝑆 = {(3𝑠 − 𝑡 + , 𝑠, + 𝑡, 𝑡): 𝑡, 𝑠 ∈ ℝ}.◄
2 2 4 4

- 60 -
Example 3.
Solve
𝑥+𝑧=1
𝑦 − 𝑧 = −1
2𝑥 + 𝑦 + 𝑧 = 2
by Gauss and Gauss-Jordan elimination.
Solution.
(a) Gauss-Jordan elimination.
The corresponding augmented matrix is:
1 0 1 1 −2𝑅13 1 0 1 1
[𝐴|𝑏]=[ 0 1 −1| −1] → [ 0 1 −1| −1]
2 1 1 2 0 1 −1 0
−1𝑅23 1 0 1 1
→ [ 0 1 −1| −1]
0 0 0 1
The corresponding system is
𝑥 + 𝑧 = 1, 𝑦 − 𝑧 = −1, 0=1
Which is impossible. Therefore, the given system of
equations has no solution. Hence it is inconsistent.
(b) Gauss-Jordan elimination.
Note that the augmented matrix is in reduced row-echelon
form. Therefore, we obtain the same results as in (a). ◄

- 61 -
Remark.
Every system of linear equations has either no solutions,
exactly one solution, or infinitely many solutions.
Theorem 1.
If 𝐴𝒙 = 𝑏 is a consistent system of equations, then it has
exactly one solution, or infinitely many solutions.
Proof.
Let 𝑥1 and 𝑥2 be two solutions of 𝐴𝑥 = 𝑏 has infinitely
many solutions. We prove that 𝑥1 + 𝑘(𝑥1 − 𝑥2 ), 𝑘 ∈ ℝ is
also a solution of 𝐴𝑥 = 𝑏. Now,
𝐴(𝑥1 + 𝑘 (𝑥1 − 𝑥2 )) = 𝐴𝑥1 + 𝑘 (𝐴𝑥1 ) − 𝑘 (𝐴𝑥2 )
= 𝑏 + 𝑘𝑏 − 𝑘𝑏 = 𝑏.
So, 𝑥1 + 𝑘(𝑥1 − 𝑥2 ), 𝑘 ∈ ℝ is a solution of 𝐴𝑥 = 𝑏.■
The above three examples are for system of equations with
number equal the number of unknowns.
Now, we give examples for system of equations with
number does not equal the number of unknowns.

- 62 -
Example 3.
Use Gauss-Jordon to solve
3𝑥1 − 2𝑥2 = 4, 5𝑥1 + 𝑥2 = 1, 9𝑥1 + 7𝑥2 = −5
Solution.
3 −2 4 1/3𝑅1 1 −2/3 4/3
[𝐴|𝑏]=[5 1 | 1 ] → [5 1 | 1 ]
9 7 −5 9 7 −5
−5𝑅12 ,−9𝑅13 1 −2 3
⁄ 4/3
→ [0 13⁄3 | −17⁄13]
0 13 −17
3⁄13𝑅2 1 −2/3
4/3
→ [0 1 | −17⁄13]
0 13 −17
2 4
−13𝑅23 1 − 3 3
→ ቎0 1 | −17⁄13቏
0 0 0
2⁄3𝑅21 1 0
6/13
→ [0 1| −17⁄13]
0 0 0
The corresponding system of equations is 𝑥1 = 6⁄13 and
𝑥2 = −17⁄13.◄
Exercise.
Use Gauss elimination to solve
3𝑥1 + 2𝑥2 = 4, 9𝑥1 + 6𝑥2 = 12, −36𝑥1 − 24𝑥2 = −48

- 63 -
Example 3.
Use Gauss-Jordon to solve
𝑥1 − 2𝑥2 + 𝑥3 = 3, 𝑥1 − 2𝑥2 + 𝑥3 = 4
Solution.

[𝐴|𝑏]=[1 −2 1 3 −1𝑅12 1 −2 1 3
| ]→ [ | ].
1 −2 1 4 0 0 0 1
The corresponding system is
𝑥1 − 2𝑥2 + 𝑥3 = 3 and 0 = 1.
Therefore, the system has no solution. ◄
Remark.
If the number of unknowns is greater than the number of
equations, then the system either has infinitely many
solutions or it is inconsistent.
Example 4.
For which values of a will the following system have no
solutions? Exactly one solution? Infinitely many
solutions?
𝑥1 + 𝑥2 + 𝑥3 + 𝑥4 = 4
𝑥1 + 𝑎𝑥2 + 𝑥3 + 𝑥4 = 4
𝑥1 + 𝑥2 + 𝑎𝑥3 + (3 − 𝑎)𝑥4 = 6
2𝑥1 + 2𝑥2 + 2𝑥3 + 𝑎𝑥4 = 6.

- 64 -
Solution.
We use Gauss elimination to put the augmented matrix in
row-echelon form.
1 1 1 1 4
[𝐴|𝑏] = ቎1 𝑎 1 1 |4቏
1 1 𝑎 3−𝑎 6
2 2 2 𝑎 6
1 1 1 1 4
−1𝑅12 ,−1𝑅13 ,−2𝑅14 0 𝑎−1 0 0
→ ቎ |0቏
0 0 𝑎−1 2−𝑎 2
0 0 0 𝑎−2 −2
1 1 1 1 4
1𝑅43 0 𝑎−1 0 0
→ ቎ |0቏
0 0 𝑎−1 0 0
0 0 0 𝑎−2 −2
… … . . (∗)

If 𝑎 ≠ 1, 2 , then
(*)

1 1 1 1 1 1 1 4
𝑅 , 𝑅 , 𝑅 0
[0 1 0 0
𝑎−1 2 𝑎−1 3 𝑎−2 4
→ | 0 ]
0 0 1 0
0 0 0 1 −2/(𝑎 − 2)

- 65 -
 1 1 1 0 4 + 2/(𝑎 − 2)
−1𝑅41 0 1 0 0 0
→ [ | 0 ]
0 0 1 0
0 0 0 1 −2/(𝑎 − 2)
1 1 0 0 4 + 2/(𝑎 − 2)
−1𝑅31 0 1 0 0 0
→ [ | 0 ]
0 0 1 0
0 0 0 1 −2/(𝑎 − 2)
1 0 0 0 4 + 2/(𝑎 − 2)
−1𝑅21 0 1 0 0 0
→ [ | 0 ]
0 0 1 0
0 0 0 1 −2/(𝑎 − 2)
Therefore, the unique solution is
2 −2
(4 + , 0, 0, )
(𝑎 − 2 ) 𝑎−2
Now, if 𝑎 = 2 , then the matrix (*) becomes
1 1 1 1 4
0 1 0 0 0
[ | ]
0 0 1 0 2
0 0 0 0 −2
which implies 0 = −2, impossible. Then the system has
no solution. Finally, if 𝑎 = 1, then (*) becomes
1 1 1 1 4
0 0 0 0 0
[ | ]
0 0 0 0 0
0 0 0 −1 −2

- 66 -
which is equivalent to the following reduced row-echelon
form.
1 1 1 1 4
0 0 0 1 2
[ | ]
0 0 0 0 0
0 0 0 0 0
The corresponding system is
𝑥1 + 𝑥2 + 𝑥3 = 2, 𝑥4 = 2.
Let 𝑥3 = 𝑡 and 𝑥2 = 𝑠. Then the solution set is 𝑆 =
{(2 − 𝑠 − 𝑡, 𝑠, 𝑡, 2): 𝑠, 𝑡 ∈ ℝ} which is an infinitely many
solutions. ◄
Theorem 2.
If A is invertible 𝑛 × 𝑛 matrix, for each 𝑛 × 1 matrix 𝑏,
the system of equation 𝐴𝑥 = 𝑏 has exactly one solution,
namely 𝑥 = 𝐴−1 𝑏.
Proof.
Since 𝐴(𝐴−1 𝑏) = 𝑏, it follows that 𝑥 = 𝐴−1 𝑏 is a
solution of 𝐴𝑥 = 𝑏. To show that this is the only solution,
we will assume that 𝑥0 is an arbitrary solution and then
we show that 𝑥0 must be the solution 𝐴−1 𝑏.
If 𝑥0 is any solution, then 𝐴𝑥0 = 𝑏. Multiplying both
sides by 𝐴−1 , we obtain 𝑥0 = 𝐴−1 𝑏.■
- 67 -
Example 5.

Consider the system of linear equations

𝑥1 − 2𝑥2 + 2𝑥3 = 3,

2𝑥1 + 𝑥2 + 𝑥3 = 0,

𝑥1 + 𝑥3 = −2.

The matrix of this system can be written as 𝐴𝑥 = 𝑏:

1 −2 2 𝑥1 3
𝐴 = [2 1 1] , 𝑥 = [𝑥2 ] , 𝑏 = [ 0 ].
1 0 1 𝑥3 −2

By elementary row operation, we obtain that

1 2 −4
−1
𝐴 = [−1 −1 3 ].
−1 −2 5

Therefore, the unique solution is

𝑥1 1 2 −4 3 11
𝑥 = [𝑥2 ] = 𝐴 𝑏 = [−1
−1
−1 3 ] [ 0 ] = [ −9 ].◄
𝑥3 −1 −2 5 −2 −13

Corollary 1.
If is A is 𝑛 × 𝑛 matrix, then the following are equivalent.

- 68 -
(1) A is invertible.

(2) The reduced row-echelon form of A is In.

(3) A is expressible as a product of elementary matrices.

(4) 𝐴𝑥 = 𝑏 is consistent for every 𝑛 × 1 matrix b.

(5) 𝐴𝑥 = 𝑏 has exactly one solution for every 𝑛 × 1

matrix b.

(6) Det 𝐴 ≠ 0.

Example 6.
What conditions must 𝑏1 , 𝑏2 and 𝑏3 satisfy for the system
of equations
𝑥1 − 2𝑥2 + 5𝑥3 = 𝑏1
4𝑥1 − 5𝑥2 + 8𝑥3 = 𝑏2
−3𝑥1 + 3𝑥2 − 3𝑥3 = 𝑏3
to be consistent?
Solution.
The augmented matrix is:
1 −2 5 𝑏1
[𝐴|𝑏]=[ 4 −5 8 | 𝑏2 ]
−3 3 −3 𝑏3

- 69 -
which can be reduced to row-echelon form as follows.
𝑏1
−4𝑅12 1 −2 5
[𝐴|𝑏] [0 3 −12| 𝑏2 − 4𝑏1 ]
3𝑅13
0 −3 12 𝑏3 + 3𝑏1

1/3𝑅2 1 −2 5 𝑏1
→ [0 1 −4| (𝑏2 − 4𝑏1 )/3]
0 −3 12 𝑏3 + 3𝑏1

3𝑅23 1 −2 5 𝑏1
→ [0 1 −4| (𝑏2 − 4𝑏1 )/3].
0 0 0 𝑏3 + 𝑏2 −𝑏1
It is now evident from the third row in the matrix that
the system has a solution if and only if 𝑏1 , 𝑏2 , and 𝑏3
satisfy the condition 𝑏3 + 𝑏2 − 𝑏1 = 0 or 𝑏3 = 𝑏1 − 𝑏2 .
To express this condition another way, 𝐴𝑥 = 𝑏 is
consistent if and b is a matrix of the form
𝑏1
𝑏 = [ 𝑏2 ],
𝑏1 − 𝑏2
where 𝑏1 and 𝑏2 are arbitrary. ◄

- 70 -
 Exercise Set (2.1)
In exercise 1-15, find all solutions of the given systems
using Gauss and Gauss-Jordan elemination.
(1) 𝑥 + 𝑦 = 1; 𝑥 − 𝑦 = 0.
(2) 𝑥 + 2𝑦 = 2; 𝑥 − 4𝑦 = −1.
(3) 3𝑥 + 7𝑦 − 3𝑧 = 2;
2𝑥 + 5𝑦 + 𝑧 = −4;
2𝑥 + 6𝑦 + 10𝑧 = 3.
(4) 3𝑥 + 7𝑦 − 3𝑧 = 2;
2𝑥 + 5𝑦 + 𝑧 = −4;
2𝑥 + 6𝑦 + 10𝑧 = −20.
(5) 𝑥 + 𝑦 + 𝑧 = 4;
2𝑥 + 5𝑦 − 2𝑧 = 3;
𝑥 + 7𝑦 − 7𝑧 = 5.
(6) 𝑥 − 2𝑦 − 𝑧 = 1;
𝑥 + 𝑦 − 𝑧 = 2;
𝑥 + 2𝑦 − 2𝑧 = 2.
(7) 𝑥 + 𝑦 = 0; 2𝑥 − 𝑦 = 1; 𝑥 + 2𝑦 = −1.
(8) 2𝑥1 + 𝑥2 − 3𝑥3 = 1; 𝑥1 − 𝑥2 + 2𝑥3 = 2.
(9) 𝑥1 + 𝑥2 + 𝑥3 + 𝑥4 = 2; 2𝑥1 + 𝑥2 + 3𝑥3 − 𝑥4 = 1.

- 71 -
(10) 𝑥1 − 𝑥2 + 2𝑥3 − 𝑥4 = −1;
2𝑥1 + 𝑥2 − 2𝑥3 − 2𝑥4 = −2;
−𝑥1 − 2𝑥2 − 4𝑥3 + 𝑥4 = 1;
3𝑥1 − 3𝑥4 = −3.
(11) 𝑥1 − 2𝑥2 + 𝑥3 − 4𝑥4 = 1;
𝑥1 + 3𝑥2 + 7𝑥3 + 2𝑥4 = 2;
𝑥1 − 12𝑥2 − 11𝑥3 − 16𝑥4 = 5;
2𝑥1 + 3𝑥2 + 3𝑥3 + 5𝑥4 = 6.
(12) 2𝑥1 + 𝑥2 + 𝑥3 + 4𝑥4 = 2;
𝑥1 + 𝑥2 + 𝑥3 − 2𝑥4 = 0;
2𝑥2 + 𝑥3 − 4𝑥4 = 1.
(13) 𝑥2 + 𝑥3 = 1; 𝑥1 − 𝑥3 = 2; −2 𝑥1 − 𝑥2 = 3.
(14) 3𝑥 − 2𝑦 = 4; 5𝑥 + 𝑦 = 1; 9 𝑥 + 7𝑦 = −5.
(15) 𝑥1 + 3𝑥2 − 2𝑥3 + 2𝑥5 = 0;
2𝑥1 + 6𝑥2 − 5𝑥3 − 2𝑥4 + 4𝑥5 − 3𝑥6 = −1;
5𝑥3 + 10𝑥4 + 15𝑥6 = 5;
2𝑥1 + 6𝑥2 + 8𝑥4 + 4𝑥5 + 18𝑥6 = 6.
(16) Find conditions that b's must satisfy for the system to
be consistent:
(a) 2𝑥 + 3𝑦 + 5𝑧 = 𝑏1 ;
4𝑥 − 5𝑦 + 8𝑧 = 𝑏2 ;
3𝑥 + 3𝑦 + 2𝑧 = 𝑏3 .
- 72 -
(b) 𝑥1 + 𝑥2 + 2𝑥3 = 𝑏1 ;
𝑥1 + 𝑥3 = 𝑏2 ;
2𝑥1 + 𝑥2 + 3𝑥3 = 𝑏3 .
(c) 𝑥1 − 𝑥2 + 3𝑥3 + 2𝑥4 = 𝑏1 ;
−2𝑥1 + 𝑥2 + 5𝑥3 + 𝑥4 = 𝑏2 ;
−3𝑥1 + 2𝑥2 + 2𝑥4 − 𝑥4 = 𝑏3 ;
4𝑥1 − 3𝑥2 + 𝑥3 + 3𝑥4 = 𝑏4 .
(18) Find 𝑎, 𝑏, and 𝑐 such that the following system has the
unique solution (1, −1, 2).
𝑎𝑥 + 𝑏𝑦 − 3𝑧 = −3
−2𝑥 − 𝑏𝑦 + 𝑧 = −1
𝑎𝑥 + 3𝑦 − 𝑐𝑧 = −1

- 73 -
2. Homogeneous Systems of Linear Equations
A system of linear equation 𝐴𝑥 = 𝑏, where 𝐴 is a
matrix of size 𝑚 × 𝑛, 𝑥 is a matrix of size 𝑛 × 1, and b a
matrix of size 𝑚 × 1, is said to be homogeneous if the
matrix 𝑏 is a zero matrix, 𝑖. 𝑒. , 𝑏 = 𝝑. If 𝑏 ≠ 𝝑, then
𝐴𝑥 = 𝑏 is a non- homogeneous system.
Every homogeneous system of linear equations is
consistent, since (0, 0, … ,0) is a solution of this system.
This solution is called the trivial solution; if there are
other solutions, they are called nontrivial solutions,
nontrivial solutions.
Because a homogeneous linear system always has the
trivial solution, there are only two possibilities for its
solutions.
• The system has only the trivial solution.
• The system has infinitely many solutions in addition
to the trivial solution.

- 74 -
Theorem 1.
If 𝑥 and 𝑦 are solutions for the homogeneous system,
𝐴𝑥 = 𝝑, then both 𝑥 + 𝑦 and 𝑘𝑥 are solutions for the
system.
Proof.
Since 𝑥, 𝑦 are solutions for the system, then 𝐴𝑥 = 𝝑 and
𝐴𝑦 = 𝝑. Therefore 𝐴(𝑥 + 𝑦) = 𝐴𝑥 + 𝐴𝑦 = 𝝑 + 𝝑 = 𝝑
and 𝐴(𝑘𝑥 ) = 𝑘(𝐴𝑥 ) = 𝑘𝝑 = 𝝑. Hence both 𝑥 + 𝑦 and
𝑘𝑥 are solution of 𝐴𝑥 = 𝝑.■
Definition 2.
Let 𝑈1 , 𝑈2 , … … , 𝑈𝑘 be an 𝑛 × 1 (or 1 × 𝑛) matrices.
Then an 𝑛 × 1 (or 1 × 𝑛) matrix U is a linear
combination of 𝑈1 , 𝑈2 , … … , 𝑈𝑘 if there exist reals
𝛼1 , 𝛼2 , … … , 𝛼𝑘 such that
𝑈 = 𝛼1 𝑈1 + 𝛼2 𝑈2 + ⋯ + 𝛼𝑘 𝑈𝑘 .
Note that by the above theorem, any linear combination
of solutions of 𝐴𝑥 = 𝝑 is also a solution. We use Gauss
and Gauss-Jordan in solving homogeneous system. Also,
we write this solution as a linear combination of other
solutions.

- 75 -
Example 1.
Solve the following homogeneous system of linear
equations by Gauss-Jordan elimination:
𝑥1 − 2𝑥2 + 𝑥3 + 𝑥4 = 0
−𝑥1 + 2𝑥2 + 𝑥4 = 0
2𝑥1 − 4𝑥2 + 𝑥3 = 0
Write the solution as a linear combination of the two
solutions (1,0, −2, 1) and (2, 1, 0, 0).
Solution.
The augmented matrix for the system is
1 −2 1 1 0
[−1 2 0 1| 0].
2 −4 1 0 0
Reducing this matrix to reduced row-echelon form:
1 −2 0 1 0
[ 0 0 1 2 | 0]
0 0 0 0 0
The corresponding system of equations is
𝑥1 − 2𝑥2 − 𝑥4 = 0, 𝑥3 + 2𝑥4 = 0.
Solving for the leading variables yields
𝑥1 = 2𝑥2 + 𝑥4
𝑥3 = −2𝑥4

- 76 -
Thus, the general solution is
𝑥1 = 2𝑠 + 𝑡, 𝑥2 = 𝑠, 𝑥3 = −2𝑡, 𝑥4 = 𝑡.
Hence the set of solutions 𝑆 = {(2𝑠 + 𝑡, 𝑠, −2𝑡, 𝑡) ∶
𝑆, 𝑡 ∈ ℝ}. Therefore
(2𝑠 + 𝑡, 𝑠, −2𝑡, 𝑡) = (2𝑠, 𝑠, 0,0) + (𝑡, 0, −2𝑡, 𝑡)
= 𝑠(2,1,0,0) + 𝑡(0,0, −2,1).
Which is a linear combination of (2,1,0,0) and
(0,0, −2,1).◄
Theorem 2.
A homogeneous system of linear equations with more
unknowns than equations has infinitely many solutions.
Remark.
Note that the above theorem applies only to
homogeneous system. A non-homogeneous system with
more unknowns than equations need not be consistent;
however, if the system is consistent, it will have infinitely
many solutions.
Theorem 3.
If 𝑥∘ is a solution of 𝐴𝑥 = 𝑏, any solution of 𝐴𝑥 = 𝑏 is
in the form 𝑥∘ + 𝑥1 , where 𝑥1 is a solution of 𝐴𝑥 = 𝝑.

- 77 -
Proof. Since 𝐴(𝑥∘ + 𝑥1 ) = 𝐴𝑥∘ + 𝐴𝑥1 = 𝑏 + 𝝑 = 𝑏, then
any solution of 𝐴𝑥 = 𝑏 is in the form 𝑥∘ + 𝑥1 . Now
suppose that 𝑦 is a solution of 𝐴𝑥 = 𝑏. We prove that
𝑦 = 𝑥∘ + 𝑥1 . Put 𝑥1 = 𝑦 − 𝑥∘ . Then
𝐴𝑥1 = 𝐴(𝑦 − 𝑥∘ ) = 𝐴𝑦 − 𝐴𝑥∘ = 𝑏 − 𝑏 = 𝝑.■
Example 2.
If 𝑥∘ = (1, −1, 0, 0) is a solution of the system
𝑥1 + 2𝑥4 = 1, 𝑥2 − 𝑥4 = −1, 𝑥3 + 𝑥4 = 0, then find all
solutions of the system.
Solution.
By Gauss-Jordan, (−2𝑡, 𝑡, −𝑡, 𝑡) is a solution of the
homogenous system. Then the solutions of the given
system are (−2𝑡, 𝑡, −𝑡, 𝑡) + (1, −1, 0, 0), 𝑡 ∈ ℝ.◄
Corollary 1.
If A is a square matrix, then 𝐴𝑥 = 𝝑 has the trivial
solution if and only if A is invertible.
Example 3.
Prove that the following system has only the trivial
solution.
𝑥1 − 2𝑥2 + 2𝑥3 = 0, 2𝑥1 + 𝑥2 + 𝑥3 = 0, 𝑥1 + 𝑥3 = 0.

- 78 -
Solution.
1 −2 2
Since det 𝐴 = |2 1 1| = 1 ≠ 0, then 𝐴 is
1 0 1
invertible. Therefore, the given system has only the trivial
solution
We end this section by the following theorem. ◄

Theorem 4.

If A is an 𝑛 × 𝑛 matrix, then the following are equivalent.

(1) 𝐴 is in invertible.

(2) 𝐴𝑥 = 𝝑 has only the trivial solution.

(3) The reduced row-echelon from A is 𝐼𝑛 .

(4) 𝐴 is expressible as a product of elementary matrices.

(5) 𝐴𝑥 = 𝑏 is consistent for every 𝑛 × 1 matrix 𝑏.

(6) 𝐴𝑥 = 𝑏 has exactly one solution for every 𝑛 × 1

matrix 𝑏.

(7) Det (𝐴) ≠ 0.

- 79 -
 Exercise Set (2.2)
(1) Find a such that the following two systems has
infinitely many solutions?
(a) 𝑥 − 2𝑦 + 𝑧 = 0;
𝑥 + 𝑎𝑦 − 3𝑧 = 0;
−𝑥 + 𝑏𝑦 − 5𝑧 = 0.
(b) 𝑥 − 𝑦 − 𝑧 = 0;
𝑎𝑦 − 𝑧 = 0;
𝑥 + 𝑦 + 𝑎𝑧 = 0.
(3) Write the general solution of the following system
as a linear combination of particular solutions.
𝑥1 + 2𝑥2 + 𝑥3 − 𝑥4 + 3𝑥5 = 0
𝑥1 + 2𝑥2 + 2𝑥3 + 𝑥4 + 2𝑥5 = 0
2𝑥1 + 4𝑥2 + 2𝑥3 − 𝑥4 + 7𝑥5 = 0
(3) Prove that the following system has only the trivial
solution.
𝑥 + 𝑦 = 0;
𝑥 − 𝑦 = 0;
2𝑥 + 3𝑦 = 0.

- 80 -
- 81 -
 CHAPTER (III)
Determinates
1. Determinates
For any 𝑛 × 𝑛 matrix 𝐴 we can define the determinant of
𝐴 as function. Its domain is the set of 𝑛 × 𝑛 matrix and its
co-domain is the set of real numbers. The determinant
function is very important. It gives us some information
about the matrix, such as its inverse and how we compute
it. Also, we can solve a system of linear equation as we
will see. There are many methods to define the determinant
of the matrix. We choose the inductive definition, 𝑖. 𝑒., we
define the determinant for a matrix of degree 1 and 2, and
we define the determinant of an 𝑛 × 𝑛 matrix.
Let 𝐴 be an 𝑛 × 𝑛 matrix. The (𝑛 − 1) × (𝑛 − 1) sub-
matrix 𝑀𝑖𝑗 is obtained by deleting the 𝑖𝑡ℎ row and
𝑗𝑡ℎ column from 𝐴.
1 1 0
2 1
For example, if 𝐴 = [ 2 1 1], then 𝑀12 = [ ]
−1 3
−1 0 3
1 0
and 𝑀22 = [ ].
−1 3

- 82 -
Definition 1.
Let 𝐴 = [𝑎𝑖𝑗 ] be an 𝑛 × 𝑛 matrix the determinat of 𝐴 is
denoted by det 𝐴 or |𝐴| and is defined by inductive as:
(1) If 𝑛 = 1, then det 𝐴 = 𝑎11
(2) If 𝑛 = 2, then det 𝐴 = 𝑎11 𝑎22 − 𝑎12 𝑎21
(3) If 𝑛 > 2, then
det 𝐴 = 𝑎11 det 𝑀11 − 𝑎12 det 𝑀12 + ⋯ +
𝑛

(−1)𝑛+1 𝑎1𝑛 det 𝑀1𝑛 = ∑(−1)𝑗+1 𝑎𝑖𝑗 det 𝑀𝑖𝑗

𝑗=1

Example 1.
1 3 −2
If 𝐴 = [−1 2 4 ], then
3 1 0
2 4 −1 4 −1 2
det 𝐴 = 1 | | − 3| | + (−2) | |
1 0 3 0 3 1
= (0 − 4) − 3(0 − 12) − 2(−1 − 6)
= 46.◄
Example 2.
2 1 −3 1
−3−2 0 2
Let 𝐴 be the matrix [ ].
2 1 0 −1
1 0 1 2

- 83 -
Then
det 𝐴 = 2 det 𝑀11 − det 𝑀12 − det 𝑀13 − det 𝑀14 ,where
−2 0 2 −3 0 2
𝑀11 = [ 1 0 −1] , 𝑀12 = [ 2 0 −1],
1 0 2 1 1 2
−3 −2 −2 −3 −2 0
𝑀13 = [ 2 1 −1] , 𝑀14 = [ 2 1 0].
1 0 2 1 0 1
Therefore,
det 𝑀11 = 0, det 𝑀12 = 1, det 𝑀13 = 2, det 𝑀14 = 1.
Hence det 𝐴 = 2(0) − 1 − 3(2) − 1 = −8.◄
●Note that we defined the determinant by inductive using
the first row of the matrix. But we use any row or any
column.
Definition 2.
Let 𝐴 = [𝑎𝑖𝑗 ] be can 𝑛 × 𝑛 matrix. The minor of 𝑎𝑖𝑗 is
det 𝑀𝑖𝑗 and the cofactor 𝐶𝑖𝑗 of 𝑎𝑖𝑗 is (−1)𝑖+𝑗 det 𝑀𝑖𝑗 .
Example 3.
2 5 −2 1
1 −2 0 1
Let 𝐴 = [ ].Then
3 3 4 0
4 1 −1 6

- 84 -
 2 −2 1
4+2
𝐶42 = (−1) det 𝑀42 = [1 0 1] = −10.◄
3 4 0
● Note that det 𝐴 is the sum of the product of the

component of first row by their cofactors 𝑖. 𝑒.,

det 𝐴 = 𝑎11 𝑐11 + 𝑎12 𝑐12 + ⋯ + 𝑎1𝑛 𝑐1𝑛 .

In fact, we can find the determinant using any row or any

column in the matrix.

That is, for each 𝑖 and 𝑗, 1 ≤ 𝑖 ≤ 𝑛 and 1 ≤ 𝑗 ≤ 𝑛, we

have det 𝐴 = 𝑎𝑖1 𝑐𝑖1 + 𝑎𝑖2 𝑐𝑖2 + ⋯ + 𝑎𝑖𝑛 𝑐𝑖𝑛 ,

where the summation is called the cofactor expansion of

det 𝐴 along the 𝑖 𝑡ℎ row.

Similarly, for each 𝑖 and 𝑗, 1 ≤ 𝑖 ≤ 𝑛 and 1 ≤ 𝑗 ≤ 𝑛, we

have

det 𝐴 = 𝑎1𝑗 𝑐1𝑗 + 𝑎2𝑗 𝑐2𝑗 + ⋯ + 𝑎𝑛𝑗 𝑐𝑛𝑗 ,

where the summation is called the cofactor expansion of

det 𝐴 along the 𝑗𝑡ℎ column.

- 85 -
Example 4.
2 5 −2 1
1 0 0 1
If 𝐴 = [ ], then we choose the second
3 0 4 0
4 0 −1 2
column to compute det 𝐴 as it contains many zeros.
Therefore
det 𝐴 = 𝑎12 𝑐12 + 𝑎22 𝑐22 + 𝑎32 𝑐32 + 𝑎42 𝑐42
= 𝑎12 𝑐12 + 0 + 0 + 0
1 0 1
= −5 |3 4 0|
4 −1 2
= −5[1(8 + 0) − 0(6 − 0) + 1(−3 − 16)] = 55.◄

- 86 -
 Exercise set (3.1)
In exercise (1) – (8). Find the determinate of the given
matrix:
2 3 3𝑎 𝑎
(1) [ ] (2) [ ]
1 −2 3𝑏 𝑏
1 1 0 1 0 4
(3) [−1 2 3] (4) [0 −2 3]
3 −4 4 2 −2 2
00 1 0 1 −2 1 0
00 2 0 −1 0 −3 3
(5) [ ] (6) [ ]
0 3 −1 2 0 3 4 −1
5−2 3 −3 0 0 −1 2
𝑘−1 1
(7) Find 𝑘 such that | | = 0.
2 𝑘
1 −1 1
(8) Find 𝑘 such that |−1 𝑘 1| = 0.
−1 −1 1
𝑘−4 0 0
(9) Find 𝑘 such that | 0 𝑘 2 | = 0.
0 3 𝑘−1
1 0 −3
𝜆 −1
(10) Find 𝑘 such that | | = |2 𝜆 −6 |.
3 1−𝜆
1 3 𝜆−5
sin 𝜃 cos 𝜃 0
(11) Find | − cos 𝜃 sin 𝜃 0|.
sin 𝜃 − cos 𝜃 sin 𝜃 + cos 𝜃 1

- 87 -
3.2 Cramer' Rule
Theorem 1. (Cramer' Rule).
If 𝐴𝑥 = 𝑏 is a system of 𝑛 linear equations in 𝑛
unknowns such that det (𝐴) ≠ 0. Then the system has
a unique solution. This solution is
det (𝐴1 ) det (𝐴2 ) det (𝐴𝑛 )
𝑥1 = , 𝑥2 = , … , 𝑥𝑛 =
det (𝐴) det (𝐴) det (𝐴)
Where 𝐴𝑗 is the matrix obtained by replacing the
entries in the 𝑗𝑡ℎ column of 𝐴 by the entries in the
𝑏1
𝑏
matrix 𝑏 = [ 2 ]
⋮
𝑏𝑛
Proof.
Since det (𝐴) ≠ 0, then 𝐴 is invertible and𝑥 = 𝐴−1 𝑏
is the unique solution of 𝐴𝑥 = 𝑏.
Therefore
1
𝑥 = 𝐴−1 𝑏 = 𝑎𝑑j (𝐴). 𝑏
det(𝐴)
𝑐11 𝑐21 …𝑐𝑛1 𝑏1
1 𝑐12 𝑐22 …𝑐𝑛2 𝑏
= [ ⋮ ⋮ … ⋮ ] [ 2]
det (𝐴) ⋮
𝑐1𝑛 𝑐2𝑛 …𝑐𝑛𝑛 𝑏𝑛

- 88 -
 1
Therefore𝑥𝑖 = (𝑏1 𝑐1𝑖 + 𝑏2 𝑐2𝑖 + ⋯ + 𝑏𝑛 𝑐𝑛𝑖 ),
det(𝐴)

𝑖 = 1,2, … , 𝑛.
Also, computing det𝐴𝑖 using column 𝑖 we obtain
det 𝐴𝑖 = 𝑏1 𝑐1𝑖 + 𝑏2 𝑐2𝑖 + ⋯ + 𝑏𝑛 𝑐𝑛𝑖
det 𝐴𝑖
Therefore 𝑥𝑖 = , 𝑖 = 1,2, … , 𝑛.◄
det 𝐴

Example 1.
Use Cramer's Rule to solve
𝑥1 + 2𝑥3 = 6
−3𝑥1 + 4𝑥2 + 6𝑥3 = 30
−𝑥1 − 2𝑥2 + 3𝑥3 = 8
Solution.
1 0 2 6 0 2
𝐴 = [−3 4 6] , 𝐴1 = [30 4 6]
−1 −2 3 8 −2 3
1 6 2 1 0 6
𝐴2 = [−3 30 6] , 𝐴3 = [−3 4 30]
−1 8 3 −1 2 8
Therefore
det(𝐴1 ) −10 det (𝐴2 ) 72 18
𝑥1 = = , 𝑥2 = = =
det(𝐴) 11 det (𝐴) 44 11
𝑑𝑒𝑡 (𝐴3 ) 152 38
𝑥3 = − = .◄
det (𝐴) 44 11

- 89 -
 Exercise Set (3.2)
(1) Use Cramer's Rule to solve the following systems
(a) 2𝑥 − 3𝑦 = 7 (b) 8𝑥 − 6𝑦 = −4
8𝑥 + 𝑦 = −2 3𝑥 + 2𝑦 = 6
(c) 2𝑥 − 6𝑦 + 𝑍 = 2 (d) 𝑥 − 3𝑦 − 𝑍 = −7
𝑦+𝑍 =1 𝑥 − 𝑦 − 𝑍 = −2
𝑥−𝑦−𝑍 =0 𝑥 − 6𝑦 − 2𝑍 = −3
(e) 𝑥1 + 𝑥2 + 𝑥3 = 𝑎 (f) 6𝑥 − 𝑦 + 3𝑍 = −3
𝑥1 + (1 + 𝑎)𝑥2 + 𝑥3 = 2𝑎 9𝑥 + 5𝑦 + 2𝑍 = 7
𝑥1 + 𝑥2 + (1 + 𝑎)𝑥3 = 0 5𝑥 + 𝑦 − 8𝑍 = −2
(5) Is it possible to use Cramer's Ruler in solving the
following System?
2𝑥 + 3𝑦 + 𝑍 = 9
𝑥 + 2𝑦 + 3𝑍 = 6
𝑥 + 𝑦 + 2𝑍 = 3
Why? Is there a solution for the system?

- 90 -
- 91 -
1. Euclidean n-space
Definition 1.
Let 𝑛 ≥ 1 . The set of all ordered n- tuples of real
numbers is called Euclidean space of dimension n and
will be denoted by ℝ𝑛 . That is:
ℝ𝑛 = {(𝑎1 , 𝑎2 , … , 𝑎𝑛 ) ∶ 𝑎𝑖 ∈ ℝ, 1 ≤ 𝑖 ≤ 𝑛}
Note.
We will use the row or the column to express the
elements of ℝ𝑛 and will be called n- vector. The addition
and scalar multiplication on ℝ𝑛 are defined as follows.
Definition 2.
Let 𝒖 = (𝑢1 , 𝑢2 , … , 𝑢𝑛 ), 𝒗 = (𝑣1 , 𝑣2 , … , 𝑣𝑛 ) ∈ ℝ𝑛 and
𝛼 ∈ ℝ , then 𝒖 + 𝒗 = (𝑢1 + 𝑣1 , 𝑢2 + 𝑣2 , … , 𝑢𝑛 + 𝑣𝑛 ) and
𝛼𝒖 = (𝛼𝑢1, 𝛼𝑢2 , … , 𝛼𝑢𝑛 ). The zero element of ℝ𝑛 and
the inverse of 𝒖 (𝑖. 𝑒, −𝒖) are 𝝑 = (0,0, … ,0) and −𝒖 =
(−𝑢1 , −𝑢2 , … , −𝑢𝑛 ), respectively. Also,
𝒖 − 𝒗 = 𝒖 + (−𝒗) = (𝑢1 − 𝑣1 , 𝑢2 − 𝑣2 , … , 𝑢𝑛 − 𝑣𝑛 )

- 92 -
Example 1.
Let 𝒖 = (2, 6, −4) and 𝒗 = (−1, 3, 5) be elements of
ℝ3 . Find 𝒖 + 𝒗, 𝒖 − 𝒗, 3𝒖 and −2𝒗 .
Solution.
𝒖 + 𝒗 = (2, 6, −4) + (−1, 3, 5) = (1, 9, 1);
𝒖 − 𝒗 = 𝒖 + (−𝒗)
= (2, 6, −4) + (1, −3, −5)
= (3, 3, −9);
3𝒖 = 3(2, 6, −4) = (6, 18, −12);
−2𝒗 = −2(−1, 3, 5) = (2, −6, −10).◄
Exercise.
In ℝ5 let 𝒖 = (3, 7, −6, 2, −1), 𝒗 = (−1, 0, 2, 3, −6),
𝒘 = (1, 0, −2, −3, 6), and 𝒛 = (0, 0, 0, 0, 0). Calculate
(a) 𝒖 + 𝒗; (b) 𝒗 + 𝒖; (c) 𝒖 + 𝒛; (d) 3𝒖; (e) −7𝒗; (f) 2𝒛;
(g) 3𝒖 + 7𝒗; (h) 0𝒖; (i) 𝒗 + 𝒘.
Note.
For every positive integer n the set of ℝ𝑛 satisfies some
properties listed in the following theorem. Since these
properties are easy to establish, the proof of this theorem
is left as an exercise.

- 93 -
Theorem 1.
If 𝒖, 𝒗 and w are any vectors in ℝ𝑛 and if 𝛼, 𝛽 are any
real numbers, then
1. 𝒖 + 𝒗 is an n-vector;
2. 𝒖 + 𝒗 = 𝒗 + 𝒖 ;
3. (𝒖 + 𝒗) + 𝒘 = 𝒖 + (𝒗 + 𝒘);
4. There is a zero element 𝝑 in ℝ𝑛 such that
𝒖 + 𝝑 = 𝝑 + 𝒖 = 𝒖 for every u in ℝ𝑛 ;
5. There is an n-vector – 𝒖 such that 𝒖 + (−𝒖) = 𝝑 ;
6. 𝛼𝒖 is an n- vector;
7. (𝛼𝛽 )𝒖 = 𝛼 (𝛽𝒖);
8. (𝛼 + 𝛽 )𝒖 = 𝛼𝒖 + 𝛽𝒖 ;
9. 𝛼 (𝒖 + 𝒗) = 𝛼𝒖 + 𝛼𝒗 ;
10. 1. 𝒗 = 𝒗.

- 94 -
 2. Linear spaces
In Section 1 above, we found that ℝ𝑛 possesses several
arithmetic properties that do not depend on the positive
integer. The operation of addition is both commutative
and associative, while the operation of scalar
multiplication satisfies several distributive properties. In
addition, ℝ𝑛 contains a zero element and every vector in
ℝ𝑛 has an additive inverse. We now define an abstract
mathematical system that embodies the arithmetic
properties of ℝ𝑛 .
Definition 1.
A linear space (𝕍, ∗, ∘) over the set of real numbers is a
nonempty set 𝕍 with two laws of combination " ∗ " and
" ∘ ", called "addition" and "scalar multiplication",
respectively, satisfying the following conditions for all
𝒖, 𝒗 and w in 𝕍 and all real numbers 𝛼 and 𝛽.
1. To every pair 𝒖, 𝒗 in 𝕍 there is only one element in 𝕍,
called the sum of u and 𝒗, denoted 𝒖 ∗ 𝒗 (closure under
addition).

2. 𝒖 ∗ 𝒗 = 𝒗 ∗ 𝒖 (Commutative property of addition).

- 95 -
3. (𝒖 ∗ 𝒗) ∗ 𝒘 = 𝒖 ∗ (𝒗 ∗ 𝒘) (Associate property of
addition).
4. There is a zero element 𝝑 in 𝕍 such that
𝒖 ∗ 𝝑 = 𝝑 ∗ 𝒖 = 𝒖 for all u in 𝕍 (Existence of additive
identity).
5. To every element u in 𝕍 there corresponds an additive
inverse element – 𝒖 in 𝕍 such that 𝒖 ∗ (−𝒖) = 𝝑
(Existence of additive inverse).

6. To every real number 𝛼 and every element u in 𝕍 there

is associated a unique element in 𝕍, called the scalar
product of u and 𝛼, which denoted by 𝛼 ∘ 𝒖 (closure
under scalar multiplication).

7. (𝛼𝛽 ) ∘ 𝒖 = 𝛼 ∘ (𝛽 ∘ 𝒖) (Associate property of scalar

multiplication).

8. (𝛼 + 𝛽 ) ∘ 𝒖 = (𝛼 ∘ 𝒖) ∗ (𝛽 ∘ 𝒖) (Distributive property
of scalar multiplication).

9. 𝛼 ∘ (𝒖 ∗ 𝒗) = (𝛼 ∘ 𝒖) ∗ (𝛼 ∘ 𝒗) (Distributive property
of scalar multiplication).

10. 1 ∘ 𝒖 = 𝒖 (Scalar identity property).

- 96 -
By a scalar we mean a real number. If in the preceding
definition, we changed '' real numbers '' to '' complex
numbers "we would obtain a linear space over the
complex numbers. We shall restrict the discussion, in this
chapter, to linear spaces over the real numbers. A linear
space is also called a vector space.
Remark. It is important to realize that a vector space
consists of four entities:
1. A set 𝕍 of vectors.
2. A set of scalars. In this class, it will always be the set
of real numbers ℝ.
3. A vector addition denoted by " ∗ ".
4. A scalar multiplication denoted by " ∘ ".
If there is no confusion, we will use "+" and "⋅" instead
of " ∗ " and " ∘ ", respectively.
Example 1.
By Theorem 1 of Section 1 in this chapter, (ℝ𝑛 , +,⋅) of
all ordered n-tuples ℝ𝑛 of real numbers with the two
combinations "+" and "⋅" defined in Definition 2 is a
linear space over ℝ for 𝑛 ≥ 1.◄

- 97 -
Remark.
We express the elements of ℝ2 as vectors in the plane.
For example, 𝒖 = (a1 , a1 ) ∈ ℝ2 is the vector line with
initial point(0, 0) and end point (a1 , a1 ) as in the figure.

If 𝐚 = (a1 , a2 ) and 𝒃 = (𝑏1 , 𝑏2 ) ∈ ℝ2 , then

𝐚 + 𝒃 = (a1 + 𝑏1 , a2 + 𝑏2 ) is the vector with initial
(0, 0) and end (a1 + 𝑏1 , a2 + 𝑏2 ) as shown in the figure.

- 98 -
If 𝒗 = (𝑎, 𝑏) ∈ ℝ2 , then 𝛼𝒗 = α(𝑎, 𝑏) = (𝛼𝑎, 𝛼𝑏) is the
vector with initial (0, 0) and end (𝛼𝑎, 𝛼𝑏) .
If 𝛼 > 0, the direction of 𝛼𝒗 is the
same direction of 𝒗.
If 𝛼 < 0, the direction of 𝛼𝒗 is
opposite to the direction of 𝒗 as
shown in the following figure.
Example 2.
Is the set of ordered pairs 𝕍 = {(𝑥1 , 𝑥2 ): 𝑥1 ∈ ℝ, 𝑥2 ∈ ℝ}
a vector space?
Solution.
Until operations of vector addition and scalar
multiplication are specified, we cannot test for vector
space structure. Therefore, we do not have a vector space.
Example 3.
From Theorem 1 in Section 1 in Chapter (I) the set of all
𝑚 × 𝑛 matrices, with real entries, with addition and
scalar multiplication of matrices (𝕄𝑚×𝑛 , + , ∙) is a
linear space. ◄

- 99 -
Example 4.
Let 𝕍 be the set of all 2- vectors whose components total
zero. That is 𝕍 = {(𝑥, 𝑦): 𝑥 + 𝑦 = 0}. Determine
whether the set 𝕍 with the addition and scalar
multiplication defined in Definition 2 is a vector space.
Solution.
Since 𝝑 = (0, 0) is an element of 𝕍, then the set 𝕍 is
non-empty. The set 𝕍 also satisfied 2,3,7, 8,9 and 10 of
Definition 1 in this section, because all 2- vectors
satisfies them and 𝕍 is the set of 2-vectors.
We now verify that the remaining properties are satisfied.
Let 𝒖 = (𝑥1 , 𝑦1 ) and 𝒗 = (𝑥2 , 𝑦2 ) be any two elements
of 𝕍 so that 𝑥1 + 𝑦1 = 0 and 𝑥2 + 𝑦2 = 0. Then
𝒖 + 𝒗 = (𝑥1 + 𝑥2 , 𝑦1 + 𝑦2 ) and
(𝑥1 + 𝑥2 ) + (𝑦1 + 𝑦2 ) = (𝑥1 + 𝑦1 ) + (𝑥2 + 𝑦2 ) = 0.
Thus 𝒖 + 𝒗 is an element of 𝕍, and property 1 is
satisfied.
Since 𝝑 = (0, 0) is an element of 𝕍 and 𝒖 + 𝝑 = 𝒖 for
every vector u in 𝕍, property 4 is satisfied.
It is clear that – 𝒖 = (−𝑥, −𝑦) is an additive inverse for
𝒖 = (𝑥, 𝑦) since 𝒖 + (−𝒖) = (𝑥, 𝑦) + (−𝑥, −𝑦) =
- 100 -
(0, 0) = 𝝑 . Moreover, (−𝑥 ) + (−𝑦) = (−1)(𝑥 + 𝑦) =

(−1)(0) = 0 so that −𝒖 is an element of 𝕍. Therefore

property 5 is satisfied.
If 𝛼 is any number, then 𝛼𝒖 = 𝛼 (𝑥, 𝑦) = (𝛼𝑥, 𝛼𝑦) and
𝛼𝑥 + 𝛼𝑦 = 𝛼 (𝑥 + 𝑦) = 𝛼0 = 0. Thus 𝛼𝒖 is an element
of 𝕍, and property 6 is satisfied. We have shown that 𝕍
satisfied all the properties of Definition 1. Therefore, 𝕍 is
a linear space. ◄
Example 4.
Let 𝕍 be the set of all 2-vectors whose components total
one. That is 𝕍 = {(𝑥, 𝑦): 𝑥 + 𝑦 = 1} . Determine
whether the set 𝕍 with the addition and scalar
multiplication defined in Definition 2 is not a vector
space.
Solution.
Properties 1, 4, 5 and 6 of Definition 1 in this section are
not satisfied. In particular (0, 1) and (1, 0) are elements
of 𝕍, but (0, 1) + (1, 0) = (1, 1) is not. So, property 1 is
not satisfied. ◄

- 101 -
Note. The important example of linear spaces is the set
of all polynomials. So, we remember the basic properties
of polynomials functions.
We define the polynomial function of degree n in the
variable 𝑥 as follows:
𝑝(𝑥 ) = 𝑎0 + 𝑎1 𝑥 + 𝑎2 𝑥 2 + ⋯ + 𝑎𝑛 𝑥 𝑛 ,
where 𝑎0 , 𝑎1 , 𝑎2 , … , 𝑎𝑛 ∈ ℝ are the real coefficients and
𝑎𝑛 ≠ 0. The zero polynomial is a polynomial with all
coefficients are zeros and its degree is undefined.
Example 5.
Let ℙ be the set of all polynomials and
𝑝(𝑥 ) = 𝑎0 + 𝑎1 𝑥 + 𝑎2 𝑥 2 + ⋯ and
𝑞 (𝑥 ) = 𝑏0 + 𝑏1 𝑥 + 𝑏2 𝑥 2 + ⋯
be two elements of 𝕡 (may be with different degrees). We
define addition and scalar multiplication as follows.
𝑝(𝑥) + 𝑞(𝑥) = (𝑎0 + 𝑏0 ) + (𝑎1 + 𝑏1 )𝑥 + (𝑎2 + 𝑏2 )𝑥 2 + ⋯
𝛼𝑝(𝑥 ) = 𝛼𝑎0 + (𝛼𝑎1 )𝑥 + (𝛼𝑎2 )𝑥 2 + ⋯
It is obvious that if 𝑝(𝑥 ), 𝑞(𝑥) ∈ 𝕡 and 𝛼 ∈ ℝ , then
𝑝(𝑥 ) + 𝑞(𝑥) ∈ 𝕡 and 𝛼𝑝(𝑥) ∈ 𝕡. It is easy to verify that
𝕡 satisfies all properties of Definition 1 above. ◄

- 102 -
Example 6.
If 𝕍 denotes the set of all polynomial functions with real
coefficients of exactly degree 2. The addition and scalar
multiplication are defined as in Example 5. Then 𝕍 is not
a linear space, because the sum of two second-degree
polynomials need not be a second-degree polynomial. For
example, the sum of 𝑝(𝑥 ) = 𝑥 2 and 𝑞 (𝑥 ) = −𝑥 2 + 𝑥 is a
first-degree polynomial.
Hence property 1 of Definition 1 above is not satisfied.
The polynomial that is zero for all x is not a second-
degree polynomial, so property 4 is not satisfied.
Example 7.
There are important subsets of 𝕡 which is a linear space.
For example, if 𝕡𝑛 is the set of all polynomial functions
of degree less than or equal to n with the zero
polynomial. Then 𝕡𝑛 is a linear space for all 𝑛 ≥ 0 with
the addition and scalar multiplication which they are
defined in Example 5. ◄

- 103 -
Example 8.
Let 𝔽[𝑎, 𝑏] be the set of all real valued functions defined
on closed interval [𝑎, 𝑏], i.e., 𝔽[𝑎, 𝑏] = {𝑓: 𝑓: [𝑎, 𝑏] → ℝ}
It is easy to verify that 𝔽[𝑎, 𝑏] is a linear space, where
the addition and scalar multiplication is defined as
(𝑓 + 𝑔)(𝑥 ) = 𝑓 (𝑥 ) + 𝑔(𝑥 ) and (𝛼𝑓)(𝑥 ) = 𝛼 [𝑓(𝑥 )]for
all 𝑓, 𝑔 ∈ 𝔽[𝑎, 𝑏] , 𝛼 ∈ ℝ 𝑎𝑛𝑑 𝑥 ∈ [𝑎, 𝑏].◄
Example 9.
Let 𝕍 = {𝝑} be a set consists of only one element is 𝝑.
Then 𝕍 is a linear space which is called a zero space,
where the addition and scalar multiplication is defined as
𝝑 + 𝝑 = 𝝑, 𝛼𝝑 = 𝝑 for all 𝛼 ∈ ℝ.
All examples above on linear space are expected and
normal. Now, we introduce different examples.
Example 10.
Prove that the set 𝕍 = {𝑥 ∈ ℝ: 𝑥 > 0} with the addition
𝑥⨁𝑦 = 𝑥𝑦 and the scalar multiplication defined by
𝑟⨀𝑥 = 𝑥 𝑟 for all 𝑥, 𝑦 ∈ 𝕍 and 𝑟 ∈ ℝ is a linear space.

- 104 -
Solution.
1- Since the product of two positive real is positive real
then 𝑥⨁𝑦 ∈ 𝕍.
2- 𝑥⨁𝑦 = 𝑥𝑦 = 𝑦𝑥 = 𝑦⨁𝑥.
3- 𝑥⨁(𝑦⨁𝑧) = 𝑥⨁(𝑦𝑧) = 𝑥(𝑦𝑧)
= (𝑥𝑦)𝑧 = (𝑥⨁𝑦)𝑧 = (𝑥⨁𝑦)⨁𝑧.
4- Since 1⨁𝑥 = 1𝑥 = 𝑥, 1 ∈ 𝕍 is the additive identity.
1 1 1
5- Since 𝑥⨁ 𝑥 = 𝑥 𝑥 = 1 , 𝑥 ∈ 𝕍 is the additive inverse
of 𝑥.
6- Since for all 𝑥 > 0 and 𝑟 ∈ ℝ , we have 𝑥 𝑟 > 0 then
𝑟 ⊙ 𝑥 ∈ 𝕍.
7- 𝑟⨀(𝑥⨁𝑦) = 𝑟⨀(𝑥𝑦) = (𝑥𝑦)𝑟 = 𝑥 𝑟 𝑦 𝑟 = 𝑥 𝑟 ⨁𝑦 𝑟 =
(𝑟⨀𝑥 )⨁(𝑟⨀𝑦).
8- (𝑟 + 𝑠)⨀𝑥 = 𝑥 𝑟+𝑠 = 𝑥 𝑟 𝑥 𝑠 = 𝑥 𝑟 ⨁𝑥 𝑠 =
(𝑟⨀𝑥 )⨁(𝑠⨀𝑥 ).
9- (𝑟𝑠)⨀𝑥 = 𝑥 𝑟𝑠 = (𝑥 𝑠 )𝑟 = 𝑟⨀(𝑠⨀𝑥).
10- 1⨀𝑥 = 𝑥1 = 𝑥.◄
Therefore, all properties of Definition 1 above are
satisfied and 𝕍 is a linear space. ◄

- 105 -
Example 11.
Prove that the set 𝑉 = {𝑥 ∈ ℝ: 𝑥 > 0} with the addition
𝑥⨁𝑦 = 𝑥 + 𝑦 (Ordinary addition of real number) and the
scalar multiplication defined by 𝑟⨀𝑥 = 𝑟𝑥 (Ordinary
multiplication) for all 𝑥, 𝑦 ∈ 𝕍 and 𝑟 ∈ ℝ is not a linear
space. Verify?
Example 12.
Prove that 𝕍 = ℝ2 with addition defined as
(𝑎, 𝑏) + (𝑐, 𝑑) = (𝑎 + 𝑐, 𝑏 + 𝑑) and scalar
multiplication defined as 𝛼 (𝑎, 𝑏) = (𝛼𝑎, 𝑏) is not a linear
space.
Solution.
Property 8 (for example) of Definition 1 above is not
satisfied because
𝛼 (𝑎, 𝑏) + 𝛽 (𝑎, 𝑏) = (𝛼𝑎, 𝑏) + (𝛽𝑎, 𝑏)
= ((𝛼 + 𝛽)𝑎, 2𝑏).
While (𝛼 + 𝛽 )(𝑎, 𝑏) = ((𝛼 + 𝛽)𝑎, 𝑏).
Then 𝛼 (𝑎, 𝑏) + 𝛽 (𝑎, 𝑏) ≠ (𝛼 + 𝛽 )(𝑎, 𝑏).
Therefore, 𝕍 is not a linear space. ◄

- 106 -
We conclude this section with a theorem that lists several
useful properties of linear spaces.
From now on, we will use "𝒖 + 𝒗" for "𝒖 ∗ 𝒗" and "𝛼𝒗"
for "𝛼 ∘ 𝒗".
Theorem 1.
Let 𝑉 be a linear space, 𝒖, 𝒗, 𝒘 be elements of 𝕍 and 𝛼 be
a scalar. Then
1-There is precisely one zero element of 𝕍.
2- If 𝒖 + 𝒗 = 𝒖, then 𝒗 = 𝝑.
3- If 𝒖 + 𝒗 = 𝒖 + 𝒘, then 𝒗 = 𝒘.
4-There is exactly one element w of such that 𝒖 + 𝒘 = 𝝑.
5- 0𝒗 = 𝝑.
6- 𝛼𝝑 = 𝝑
7- If 𝛼𝒗 = 𝝑, then 𝛼 = 0 or 𝒗 = 𝝑.
8- (−𝛼 )𝒗 = −(𝛼𝒗) = 𝛼 (−𝒗)
9- (−1)𝒖 = −𝒖
Proof.
1-Assume that there are two elements 𝝑 and 𝝑′ such that
𝒖 + 𝝑 = 𝒖 and 𝒖 + 𝝑′ = 𝒖 for every u in 𝕍.

- 107 -
Setting 𝒖 = 𝝑′ in the first identity and 𝒖 = 𝝑 in the second
identity, we obtain 𝝑′ = 𝝑′ + 𝝑 = 𝝑 + 𝝑′ = 𝝑.
2- Suppose that 𝒖 + 𝒗 = 𝒖. By Property 5 of Definition 1
in this section, there is an inverse – 𝒖 for 𝒖.
Adding – 𝒖 to each side of 𝒖 + 𝒗 = 𝒖 we have
(−𝒖) + (𝒖 + 𝒗) = (−𝒖) + 𝒖,
or [(−𝒖) + 𝒖] + 𝒗 = (−𝒖) + 𝒖,
or 𝝑 + 𝒗 = 𝝑, or 𝒗 = 𝝑.
3- 𝒖 + 𝒗 = 𝒖 + 𝒘 ⇒
(−𝒖) + (𝒖 + 𝒗) = (−𝒖) + (𝒖 + 𝒘) ⇒
(−𝒖 + 𝒖) + 𝒗 = (−𝒖 + 𝒖) + 𝒘 ⇒
𝝑 + 𝒗 = 𝝑 + 𝒘 ⇒ 𝒗 = 𝒘.
4- Let 𝒘 and 𝒘′ be such that 𝒖 + 𝒘 = 𝝑 and 𝒖 + 𝒘′ = 𝝑
𝒘=𝒘+𝝑
= 𝒘 + (𝒖 + 𝒘′ )
= (𝒘 + 𝒖) + 𝒘′
= (𝒖 + 𝒘) + 𝒘′
= 𝝑 + 𝒘′ = 𝒘′.
5- Since 0𝒗 = (0 + 0)𝒗 = 0𝒗 + 0𝒗, then by adding
(−0𝒗) to both sides we have
0𝒗 + (−0𝒗) = (0𝒗 + 0𝒗) + (−0𝒗)
- 108 -
 or
0𝒗 + (−0𝒗) = 0𝒗 + (0𝒗 + (−0𝒗))
or
𝝑 = 0𝒗 + 𝝑 = 0𝒗
6- Since 𝛼𝝑 = 𝛼 (𝝑 + 𝝑) = 𝛼𝝑 + 𝛼𝝑, then by adding
(−𝛼𝝑) to both sides we have
𝛼𝝑 + (−𝛼𝝑) = (𝛼𝝑 + 𝛼𝝑) + (−𝛼𝝑), or
𝝑 = 𝛼𝝑 + (𝛼𝝑 + (−𝛼𝝑)) = 𝛼𝝑 + 𝝑.
Hence 𝛼𝝑 = 𝝑.
1 1
7- Let 𝛼𝒗 = 𝝑 and 𝛼 ≠ 0 . Then ( ) 𝛼𝒗 = ( ) 𝝑.
𝛼 𝛼

Therefore, 1. 𝒗 = 𝝑.
So, 𝒗 = 𝝑.
8- Note that (−𝛼 )𝒗 + (𝛼𝒗) = (−𝛼 + 𝛼 )𝒗 = 0𝒗 = 𝝑.
Hence (−𝛼 )𝒗 + (𝛼𝒗) = 𝝑.
By adding (−𝛼𝒗) to both sides we here
(−𝛼 )𝒗 = −(𝛼𝒗) .
Similarly, we can prove that 𝛼 (−𝒗) = −𝛼𝒗.
9- Let 𝛼 = 1 in (8). ◄

- 109 -
 Exercise Set (4.2)
In Exercise (1) – (20) determine whether the given set
with the two operations are linear space.
1. 𝕍 = ℝ2 ,
(𝑥1 , 𝑥2 ) + (𝑦1 , 𝑦2 ) = (𝑥2 + 𝑦2 , 𝑥1 + 𝑦1 ),
𝛼 (𝑥1 , 𝑥2 ) = (𝛼𝑥1 , 𝛼𝑥2 ).
2. 𝕍 = {(𝑥1 , 0, 𝑥3 ) ∶ 𝑥1 , 𝑥3 ∈ ℝ},
(𝑥1 , 0, 𝑥3 ) + (𝑦1 , 0, 𝑦3 ) = (𝑥1 + 𝑦1 , 0, 𝑥3 + 𝑦3 ),
𝛼 (𝑥1 , 0, 𝑥3 ) = (𝛼𝑥1 , 0, 𝑥3 ).
0 𝑏
3. 𝕍 = {[ ] ∶ 𝑎, 𝑏, 𝑐 ∈ ℝ} with summation and scalar
𝑎 𝑐
multiplication of matrices.
𝑎 𝑏
4. 𝕍 = {[ ] ∶ 𝑎 ≠ 0} with summation and scalar
𝑐 𝑑
multiplication of matrices.
5. 𝕍 = {𝐴: 𝐴 ∈ 𝑀2×3 , ∑ 𝑎𝑖𝑗 = 0} with summation and
scalar multiplication of matrices.
6. 𝕍 = {𝑃 ∈ 𝑃𝑛 ∶ 𝑃(1) + 𝑃(6) − 𝑃(2) = 0} with
summation andscalar multiplication of polynomials.
7. 𝕍 = {𝑃 ∈ 𝑃𝑛 ∶ 𝑃(1) + 𝑃(6) − 𝑃(2) = 1} with
summation and scalar multiplication of polynomials.

- 110 -
8. 𝕍 = ℝ𝟑 ,
(𝑥1 , 𝑦1 , 𝑧1 ) + (𝑥2 , 𝑦2 , 𝑧2 ) = (𝑥1 + 𝑥2 , 𝑦1 + 𝑦2 , 𝑧1 + 𝑧2 ),
𝛼 (𝑥, 𝑦, 𝑧) = (0,0,0).
9. 𝕍 = ℝ2 ,
(𝑥1 , 𝑦1 ) + (𝑥2 , 𝑦2 ) = (𝑥1 + 𝑥2 , 𝑦1 + 𝑦2 ),
𝛼 (𝑥, 𝑦) = (2𝑥, 2𝑦).
10. 𝕍 = {(𝑥, 0): 𝑥 ∈ ℝ} with the usual addition and scalar
multiplication on ℝ2 .
11. 𝕍 = {(𝑥, 0): 𝑥 ≥ 0} with the usual addition and scalar
multiplication on ℝ2 .
12. 𝕍 = ℝ2 ,
(𝑥, 𝑦) + (𝑥1 , 𝑦1 ) = (𝑥 + 𝑥1 + 1, 𝑦 + 𝑦1 + 1),
𝛼 (𝑥, 𝑦) = (𝑥, 𝑦).
𝑎 1
13. 𝕍 = {[ ] : 𝑎, 𝑏 ∈ ℝ} with summation and scalar
1 𝑏
multiplication of matrices.
14. 𝕍 = {𝑓 ∈ F[0,2]: 𝑓(1) = 0} with addition and scalar
multiplication in defined in Example 5 Section 2 above.
15. 𝕍 = {(𝑎, 𝑎, −𝑎): 𝑎 ∈ ℝ} with usual addition and
scalar multiplication on ℝ𝟑 .

- 111 -
16. 𝕍 = {(𝑎, 1, −𝑎): 𝑎 ∈ ℝ} with usual addition and
scalar multiplication on ℝ𝟑 .
17. 𝕍 = {𝑎0 + 𝑎1 (𝑥 + 1): 𝑎0 , 𝑎1 ∈ ℝ} with addition and
scalar multiplication of polynomials.
18. 𝕍 = {𝑎0 + 𝑥 + 1: 𝑎0 ∈ ℝ} with addition and scalar
multiplication of polynomials.
19. 𝕍 = {𝐴 ∈ 𝑀3×3 : 𝐴 = 𝐴𝑇 } with addition and scalar
multiplication of matrices.
20. 𝕍 = {(𝑎, 𝑏, 𝑐 ): ∈ ℝ3 } with addition and scalar
multiplication on ℝ𝟑 .
21. Let (𝕍,∗ ,∘) and (𝕎,⊕,⊙) be two vector spaces over
ℝ and 𝕍 × 𝕎 = {(𝑣, 𝑤): 𝑣 ∈ 𝕍, 𝑤 ∈ 𝕎}. Prove that
(𝕍 × 𝕎, ⨂, ⊚) is a vector space over ℝ, if for every
(𝑣1 , 𝑤1 ), (𝑣2 , 𝑤2 ) ∈ 𝕍 × 𝕎 and 𝛼 ∈ ℝ, we define
(𝑣1 , 𝑤1 )⨂(𝑣2 , 𝑤2 ) = (𝑣1 ∗ 𝑣2 , 𝑤1 ⊕ 𝑤2 ), and
𝛼 ⊚ (𝑣1 , 𝑤1 ) = (𝛼 ∘ 𝑣1 , 𝛼 ⊙ 𝑤1 ).
22. For any 𝛼 ∈ ℝ and 𝐱 = (𝑥1 , 𝑥2 )𝑇 , 𝐲 = (𝑦1 , 𝑦2 )𝑇 ∈
ℝ2 , define 𝐱 ⊕ 𝐲 = (𝑥1 + 𝑥1 + 1, 𝑥2 + 𝑥2 − 3)𝑇 and
α⨀𝐱 = (𝛼𝑥1 + 𝛼 − 1, 𝛼𝑥2 − 3𝛼 + 3)𝑇 . Then, ℝ2 is a
vector space with (−1, 3)𝑇 as the additive identity.

- 112 -
 3. Subspaces
It is possible for one linear space to be contained within a
larger space. For example, planes through the origin are
linear spaces that are contained within the larger linear
space ℝ𝟑 . In this section we shall study this important
concept in more detail.
Definition 1.
A subset 𝕎 of a linear space 𝕍 is called a subspace of 𝕍
if 𝕎 is itself a linear space under the addition and scalar
multiplication defined on 𝕍.
In general, one must verify the ten linear space axioms to
show that a set 𝕎 with addition and scalar multiplication
forms a linear space.
However, if 𝕎 is part of a larger set 𝕍 that is already
known to be a linear space, then certain axioms need not
be verified for 𝕎 because they are inherited from 𝕍.
For example, there is no need to check that 𝒖 + 𝒗 = 𝒗 +
𝒖 (Axiom 2) for 𝕎 because this holds for all vectors in 𝕍
and consequently for all vectors in 𝕎. other axioms
inherited by 𝕎 from 𝕍 are 3, 7, 8, 9 and 10. Thus, to

- 113 -
show that a set 𝕎 is a subspace of a linear space 𝕍, we
need only verify Axioms 1,4,5 and 6.
The following theorem shows that even Axioms 4 and 5
can be dispensed with.
Theorem 1.
If 𝕎 is a set of one or more vectors from a linear space
𝕍, then 𝕎 is a subspace of 𝕍 if and only if the following
conditions hold:
(a) If u and v are vectors in 𝕎, then 𝒖 + 𝒗 is in 𝕎.
(b) If k is any scalar and u is any vector in 𝕎, then 𝑘𝒖 is
in 𝕎.
Proof.
If 𝕎 is a subspace of 𝕍, then all the linear space axioms
are satisfied; in particular, Axioms 1 and 6 hold but these
are Precisely conditions (a) and (b).
Conversely, assume conditions (a) and (b) hold. Since
these conditions are linear space axioms 1 and 6, we need
only show that 𝕎 satisfies the remaining eight axioms.
Axioms 2,3,7,8,9, and 10 are automatically satisfied by
the vectors in 𝕎 since they are satisfied by all vectors in
𝕍. Therefore, to complete the proof, we need only verify
- 114 -
that Axioms 4 and 5 are satisfied by vectors in 𝕎. Let u
be any vector in 𝕎. By condition (b), 𝑘𝒖 is in 𝕎 for
every scalar 𝑘. Setting 𝑘 = 0, it follows from Theorem 1
in Section 2 of this chapter that 0𝒖 = 𝝑 is in 𝕎, and
setting 𝑘 = −1, it follows that (−1)𝒖 = −𝒖 is in 𝕎.◄
Remark.
A set 𝕎 of one or more vectors from a linear space 𝕍 is
said to be closed under a addition if condition (a) in
Theorem 1 above holds and closed under scalar
multiplication if condition (b) holds. Thus, Theorem 1
above states that a non-empty subset 𝕎 of 𝕍 is a
subspace of 𝕍 if and only if 𝕎 is closed under addition
and under scalar multiplication.

We can use one condition equivalent as both conditions

(a) and (b) in Theorem 1 above as follows:
A non-empty subset 𝕎 of 𝕍 is a subspace of 𝕍 if and
only if 𝛼𝒖 + 𝛽𝒗 ∈ 𝕎 for all 𝒖, 𝒗 ∈ 𝕎 and all 𝛼, 𝛽 ∈ ℝ
(verify?).

- 115 -
Example 1.
Every nonzero linear space 𝕍 has at least two subspaces,
𝕍 itself is a subspace, and the set {𝝑} consisting of just
the zero in 𝕍 is a subspace called the zero subspace. ◄
Example 2.
Let 𝕏 = {(𝑥, 0): 𝑥 ∈ ℝ} ⊆ ℝ2 . Prove that 𝕏 is a subspace
of ℝ2 .
Solution.
(1) 𝝑 = (0,0) ∈ 𝕏, i.e., 𝕏 ≠ 𝜙.
(2) Let (𝑥1 , 0), (𝑥2 , 0) ∈ 𝕏. Then
(𝑥1 , 0) + (𝑥2 , 0) = (𝑥1 + 𝑥2 , 0) ∈ 𝕏
(3) 𝛼 (𝑥, 0) = (𝛼𝑥, 0) ∈ 𝕏 for all (𝑥, 0) ∈ 𝕏 and all
α ∈ ℝ. From (1), (2) and (3), 𝑋 is a subspace of ℝ2 .
Similarly, 𝕐 = {(0, 𝑦): 𝑦 ∈ ℝ} is a subspace of ℝ2 .◄
Example 3.
Let 𝕎 = {(𝑥, 𝑦) ∈ ℝ2 : 𝑎𝑥 + 𝑏𝑦 = 0} ⊆ ℝ2 . Prove that
𝕎 is a subspace of ℝ2 .
Solution.
(1) Since 𝑎0 + 𝑏0 = 0, then (0,0) ∈ 𝕎 and 𝕎 ≠ 𝜙.

- 116 -
(2) Let (𝑥, 𝑦), (𝑥1 , 𝑦1 ) ∈ 𝕎. Then 𝑎𝑥 + 𝑏𝑦 = 0 and
𝑎𝑥1 + 𝑏𝑦1 = 0. Hence 𝑎(𝑥 + 𝑥1 ) + 𝑏(𝑦 + 𝑦1 ) =
(𝑎𝑥 + 𝑏𝑦) + (𝑎𝑥1 + 𝑏𝑦1 ) = 0 + 0 = 0.
Therefore (𝑥, 𝑦) + (𝑥1 , 𝑦1 ) = (𝑥 + 𝑥1 , 𝑦 + 𝑦1 ) ∈ 𝕎.
(3) Let (𝑥, 𝑦) ∈ 𝕎 and 𝛼 ∈ ℝ . Then 𝑎𝑥 + 𝑏𝑦 = 0.
Therefore 𝛼 (𝑎𝑥 + 𝑏𝑦) = 0. Thus 𝛼 (𝑥, 𝑦) ∈ 𝕎. From (1),
(2) and (3), we have 𝕎 is a subspace of ℝ2 . ◄
Note that the subspace in Example 3 above of ℝ2 is the
lines through the origin. So, the subspaces of ℝ2 are {𝝑},
lines through the origin and ℝ2 itself. Later, we will show
that these are the only subspaces of ℝ2 .◄
Example 4.
Let 𝕎 = {(𝑥, 𝑦, 𝑧) ∈ ℝ3 : 𝑎𝑥 + 𝑏𝑦 + 𝑐𝑧 = 0}. Prove that
𝕎 is a subspace of ℝ3 .
Solution.
(1) Since 𝑎0 + 𝑏0 + 𝑐0 = 0, (0,0,0) ∈ 𝕎 and 𝕎 ≠ 𝜙.
(2) Let (𝑥, 𝑦, 𝑧), (𝑥1 , 𝑦1 , 𝑧1 ) ∈ 𝕎. Then 𝑎𝑥 + 𝑏𝑦 + 𝑐𝑧 =
0 and 𝑎𝑥1 + 𝑏𝑦1 + 𝑐𝑧1 = 0. Therefore,
𝑎(𝑥 + 𝑥1 ) + 𝑏(𝑦 + 𝑦1 ) + 𝑐 (𝑧 + 𝑧1 ) = (𝑎𝑥 + 𝑏𝑦 + 𝑐𝑧) +
(𝑎𝑥1 + 𝑏𝑦1 + 𝑐𝑧1 ) = 0 + 0 = 0.

- 117 -
Thus, (𝑥, 𝑦, 𝑧) + (𝑥1 , 𝑦1 , 𝑧1 ) ∈ 𝕎.
(3) Let (𝑥, 𝑦, 𝑧) ∈ 𝕎 and α ∈ ℝ. Then 𝑎𝑥 + 𝑏𝑦 + 𝑐𝑧 =
0. Therefore 𝛼 (𝑎𝑥 + 𝑏𝑦 + 𝑐𝑧) = 0. Thus 𝛼 (𝑥, 𝑦, 𝑧) ∈ 𝕎.
Hence 𝕎 is a subspace of ℝ3 .
Note that the subspace 𝕎 is the planes through the origin.
Later, we will show that the only subspaces of ℝ3 are
{𝝑}, lines through the origin, planes through the origin,
and ℝ3 itself. ◄
Example 5.
𝑎 𝑏
Let 𝕄 = {[ ] ∶ 𝑎, 𝑏 ∈ ℝ} ⊆ 𝕄2×2 . Prove that 𝕄 is
0 −𝑎
a subspace of 𝕄2×2 .
Solution.
0 0
(1) 𝝑 = [ ] ∈ 𝕄 and 𝕄 ≠ 𝜙.
0 0
𝑎 𝑏 𝑐 𝑑
(2) Let [ ], [ ] ∈ 𝕄. Then
0 −𝑎 0 −𝑐
𝑎 𝑏 𝑐 𝑑 𝑎+𝑐 𝑏+𝑑
[ ]+[ ]=൤ ൨ ∈ 𝕄.
0 −𝑎 0 −𝑐 0 −(𝑎 + 𝑐)
𝑎 𝑏
(3) Let [ ] ∈ 𝕄 and α ∈ ℝ.
0 −𝑎
𝑎 𝑏 𝛼𝑎 𝛼𝑏
Then 𝛼 [ ]=[ ] ∈ 𝕄.
0 −𝑎 0 −𝛼𝑎

- 118 -
Therefore, 𝕄 is a subspace of 𝕄2×2 .◄
Example 6.
Let ℙ be the set of all polynomials and
ℙ = {𝑝(𝑥 ) ∈ ℙ: 𝑝(3) = 0} ⊆ ℙ.
Prove that 𝕎 is a subspace of 𝕡.
Solution.
(1) Since 𝝑(3) = 0, then 𝝑 ∈ ℙ.
(2) Let 𝑝(𝑥 ), 𝑞 (𝑥 ) ∈ ℙ. then 𝑝(3) = 𝑞 (3) = 0.
Therefore (𝑝 + 𝑞 )(3) = 𝑝(3) + 𝑞 (3) = 0 + 0 = 0. Thus
𝑝(𝑥 ) + 𝑞(𝑥) ∈ ℙ.
(3) Let 𝑝(𝑥 ) ∈ ℙ and α ∈ ℝ. Then 𝑝(3) = 0 and
α𝑝(3) = 0. Hence 𝛼𝑝(𝑥 ) ∈ ℙ.◄
Example 7.
The subset 𝕎 = {(𝑥, 𝑥 + 1): 𝑥 ∈ ℝ} is not a subspace of
ℝ2 since 𝝑 = (0,0) ∉ 𝕎. ◄
Example 8.
The subset 𝕎 = {𝑎0 + 𝑎1 𝑥 + 𝑎2 𝑥 2 : 𝑎0 ∈ ℤ} is not a
1
subspace of 𝕡2 , since, for example, 2 + 3𝑥 + 𝑥 2 ∈ 𝕎
3
1 1 1 2 1
and ∈ ℝ but (2 + 3𝑥 + 𝑥 2 ) = + 𝑥 + 𝑥 2 ∉ 𝕎.◄
3 3 3 3 9

- 119 -
Example 9.
The subset 𝕎 = {(𝑎, 𝑏) ∈ ℝ2 : 𝑎 ≠ 0, 𝑏 ≠ 0} is not a
subspace of ℝ2 , because, for example, (2,3), (−2,5) ∈ 𝕎
but (2,3) + (−2,5) = (0,8) ∉ 𝕎.
Definition 2.
If 𝐴𝒙 = 𝑏 is a system of linear equations, then each
vector 𝒙 that satisfies this equation is a solution vector of
the system. The following theorem shows that the
solution vectors of a homogeneous linear system form a
linear space, which we shall call the solution space of
the system.
Theorem 2.
If 𝐴𝒙 = 𝝑 is a homogeneous linear system of m
equations in n unknowns, then the set of solution vectors
is a subspace of ℝ𝑛 .
Proof.
Let 𝕊 be the solution vectors set. There is at least one
vector in 𝕊, namely 𝝑. To show that 𝕊 is closed under
addition and scalar multiplication, we must show that if
𝒙 and 𝒙′ are any solution vectors and k is any scalar, then

- 120 -
𝒙 + 𝒙′ and 𝑘𝒙 are also solution vectors. But if 𝒙 and 𝒙′
are solution vectors, then 𝐴𝒙 = 𝝑 and 𝐴𝒙′ = 𝝑. It
follows that 𝐴(𝒙 + 𝒙′) = 𝐴𝒙 + 𝐴𝒙′ = 𝝑 + 𝝑 = 𝝑 and
𝐴(𝑘𝒙) = 𝑘(𝐴𝒙) = 𝑘𝝑 = 𝝑. This proves that 𝒙 + 𝒙′ and
𝑘𝒙 are solution vectors. Therefore, 𝕊 is a subspace
ℝ𝑛 .◄

- 121 -
 Exercise set (4.3)
In Exercise (1)–(19) determine which of the following
subsets are subspaces of the given linear spaces.
1. 𝕎 = {𝑓: [a, b] ⟶ ℝ, 𝑓 is diffrentiable on [a, b]} ⊆
𝔽[𝑎, 𝑏]
2. 𝕎 = {𝑓: [𝑎, 𝑏] → ℝ, 𝑓 is diffrentiable twice on [a, b]}
⊆ 𝔽[𝑎, 𝑏] such that 𝑓 ′′ + 𝑓 = 0}
3. 𝕎 = {𝑎, 𝑏, 𝑎 − 2𝑏: 𝑎, 𝑏 ∈ ℝ} ⊆ ℝ3 .
𝑎 0
4. 𝕎 = {[ ] , 𝑎, 𝑏, 𝑐 ∈ ℝ} ⊆ 𝑀2×2 .
𝑏 𝑐
5. 𝕎 = {𝑎0 + 𝑎1 𝑥 + 𝑎2 𝑥 2 ∈ 𝕡2 : 𝑎0 + 𝑎1 + 𝑎2 = 0} ⊆
𝕡2 .
6. 𝕎 = {𝑓 ∈ 𝔽[0,2]: 𝑓 (1) = 0} ⊆ 𝔽[0,2].
7. 𝕎 = {(𝑥, 𝑦) ∈ ℝ2 : 2𝑥 + 3𝑦 = 0} ⊆ ℝ2 .
8. 𝕎 = {𝐴 ∈ 𝕄2×2 : det 𝐴 = 0} ⊆ 𝕄2×2 .
9. 𝕎 = {𝐴 ∈ 𝕄2×2 : 𝐴2 = 𝐴} ⊆ 𝕄2×2 .
10. 𝕎 = {(𝑥, 𝑦) ∈ ℝ2 : 𝑥 + 𝑦 = 1} ⊆ ℝ2 .
11. 𝕎 = {𝑎0 + 𝑎1 𝑥 + 𝑎2 𝑥 2 ∈ ℙ2 : 𝑎1 = 1 𝑎0 = 𝑎2 } ⊆
ℙ2 .
12. 𝕎 = {(𝑎, 𝑏, 𝑐) ∈ ℝ3 : 𝑎 + 𝑏 + 𝑐 = 0} ⊆ ℝ3 .
13. 𝕎 = {(𝑎, 𝑏, 𝑐) ∈ ℝ3 : 𝑎 − 𝑏 − 𝑐 = 0} ⊆ ℝ3 .

- 122 -
14. 𝕎 = {(𝑎, 𝑏, 𝑐) ∈ ℝ3 : 𝑎 + 𝑏 = 1, 𝑐 = 0} ⊆ ℝ3 .
15. 𝕎 = {(𝑎, 𝑏, 𝑐) ⊂ ℝ3 : 𝑎𝑏 = 0} ⊆ ℝ3 .
3 𝑎 3
16. 𝕎 = {[ ] : 𝑎, 𝑏 ∈ ℝ} ⊆ 𝕄2×3 .
𝑎 𝑏 𝑎
17. 𝕎 = {𝐴 ∈ 𝕄2×2 : 𝐴 = −𝐴𝑇 } ⊆ 𝕄2×2 .
1
18. 𝕎 = {𝑓 ∈ 𝔽[0,1] : ∫0 𝑓(x)dx = 0} ⊆ 𝔽[0,1].

- 123 -
4. Linear combination and spanning sets
Definition 1.
Let 𝕍 be a linear space and 𝒗1 , 𝒗2 , … , 𝒗𝑛 ∈ 𝕍. An
element w of 𝕍 is called a linear combination of
𝒗1 , 𝒗2 , … , 𝒗𝑛 if there are scalars 𝛼1 , 𝛼2 , … , 𝛼𝑛 ∈ ℝ such
that 𝒘 = 𝛼1 𝒗1 + 𝛼2 𝒗2 + ⋯ + 𝛼𝑛 𝒗𝑛 .
Example 1.
Prove that the vector 𝒗 = (7, −2, 2) ∈ ℝ3 is a linear
combination of the vectors 𝒗1 = (1, −1,0), 𝒗2 =
(0,1,1), 𝒗3 = (2,0,1).
Solution.
We want to find 𝛼1 , 𝛼2 , 𝛼3 ∈ ℝ such that
(7, −2, 2) = 𝛼1 (1, −1, 0) + 𝛼2 (0, 1, 1) + 𝛼3 (2, 0, 1).
Then
𝛼1 + 2𝛼3 = 7,−𝛼1 + 𝛼2 = −2, 𝛼2 + 𝛼3 = 2.
By solving this system, we obtain
𝛼1 = 1, 𝛼2 = −1, 𝛼3 = 3
Hence
(7, −2,2) = (1, −1,0) − (0,1,1) + 3(2,0,1).◄

- 124 -
Example 2.
Determine whether 𝒗 = (3, −1,4) ∈ ℝ3 is a linear
combination of 𝒗1 = (1, −1,0), 𝒗2 = (0,1,1), 𝒗3 =
(3, −5,2), or not?
Solution.
Let 𝛼1 , 𝛼2 , 𝛼3 ∈ ℝ such that 𝒗 = 𝛼1 𝒗1 + 𝛼2 𝒗2 + 𝛼3 𝒗3
Therefore,
(3, −1,4) = 𝛼1 (1, −1,0) + 𝛼2 (0,1,1) + 𝛼3 (3, −5, −2)
So, we have the following system of equation
𝛼1 + 3𝛼3 = 3;
−𝛼1 + 𝛼2 − 5𝛼3 = −1;
𝛼2 − 2𝛼3 = 4
By using Gauss method, we find the row–echelon form of
the augmented matrix is:
1 0 3 3
[0 1 −2| 2]
0 0 0 1
Hence the system is inconsistent. Therefore 𝒗 is not a
linear combination of 𝒗1 , 𝒗2 , 𝒗3 .◄

- 125 -
Example 3.
Consider 𝑓(𝑥 ) = 𝑥 2 + 𝑥, 𝑔(𝑥 ) = 𝑥 + 2,
ℎ(𝑥 ) = 3𝑥 2 + 2𝑥 − 2 and 𝑘 (𝑥 ) = 𝑥 3 in the linear space
of all functions defined on the interval (−∞, ∞), i.e.,
𝔽(−∞, ∞). The function ℎ is a linear combination of
𝑓 and 𝑔 since ℎ(𝑥 ) = 3𝑓(𝑥 ) − 𝑔(𝑥). But, 𝑘(𝑥 ) = 𝑥 3 is
not a linear combination of f and g.
This may be proved as follows.
Suppose that 𝑘 is a linear combination of 𝑓 and 𝑔. Then
there exists 𝛼1 , 𝛼2 ∈ ℝ with 𝑘 (𝑥 ) = 𝛼1 𝑓(𝑥 ) + 𝛼2 𝑔(𝑥).
Therefore 𝑥 3 = 𝛼1 (𝑥 2 + 𝑥 ) + 𝛼2 (𝑥 + 2). Comparing the
coefficients in both sides we have 0 = 1, which is
impossible hence 𝑘 (𝑥 ) is not a linear combination of 𝑓
and 𝑔.◄
Example 4.
Since cos(3 + x) = cos 3 cos x − sin 3 sin x, then
cos(3 + x) is a linear combination of sin x and cos x .
However, sin2 𝑥 is not a linear combination of sin x and
cos x . Indeed, let sin2 𝑥 be a linear combination of sin x
and cos x . Then sin2 𝑥 = 𝛼1 sinx + 𝛼2 cos x, where

- 126 -
𝛼1 , 𝛼2 ∈ ℝ for every x ∈ ℝ. In particular, sin2 0 =
𝛼1 sin 0 + 𝛼2 cos 0, 𝑖. 𝑒, 0 = 𝛼1 0 + 𝛼2 1, hence 𝛼2 = 0.
Therefore sin2 x = 𝛼1 sinx. If sin x ≠ 0, then sin x = α1
for every 𝑥 ∈ ℝ. This is impossible. So sin2 𝑥 is not a
linear combination of sinx and cos x. ◄
In Examples 1-3 above we need to solve system of
equations to show that a vector is a linear combination of
a given vectors. In special case that the linear space is a
subspace of ℝ𝑛 there is a mechanical procedure for
discovering whether an n-vector 𝑏 is a linear combination
of n- vectors 𝒗1 , 𝒗2 , . . 𝒗𝑛 . In fact, we will show that b is a
linear combination of 𝒗1 , 𝒗2 , . . 𝒗𝑛 if and only if the
equation 𝐴𝒙 = 𝑏 is consistent, where 𝐴 is the
matrix having 𝒗1 , 𝒗2 , . . 𝒗𝑛 as its columns. In order to
prove this, we first prove a theorem that relates the
concept of linear combinations to matrix multiplications.
Theorem 1.
Let 𝐴 = [𝑎𝑖𝑗 ] be an 𝑚 × 𝑛 matrix and 𝑥 =
[𝑥1 , 𝑥2 , … , 𝑥𝑛 ]𝑇 . If 𝐴1 , 𝐴2 , … , 𝐴𝑛 are the columns of 𝐴,
then 𝐴𝑥 = 𝑥1 𝐴1 + 𝑥2 𝐴2 + ⋯ + 𝑥𝑛 𝐴𝑛

- 127 -
Proof.
𝑎11 𝑎12 …𝑎1𝑛 𝑥1
𝑎21 𝑎22 …𝑎2𝑛 𝑥2
𝐴𝑥 = [ ⋮ ⋮ ⋮ ⋮ ]቎ ⋮ ቏
𝑎𝑚1 𝑎𝑚2 …𝑎𝑚𝑛 𝑥𝑛
𝑎11 𝑥1 + 𝑎12 𝑥2 + ⋯ + 𝑎1𝑛 𝑥𝑛
𝑎 𝑥 + 𝑎22 𝑥2 + ⋯ + 𝑎2𝑛 𝑥𝑛
= [ 21 1 ⋮ ]
𝑎𝑚1 𝑥1 + 𝑎𝑚2 𝑥2 + ⋯ + 𝑎𝑚𝑛 𝑥𝑛
𝑎11 𝑎12 𝑎1𝑛
𝑎21 𝑎22 𝑎2𝑛
= 𝑥1 [ ⋮ ] + 𝑥2 [ ⋮ ] + ⋯ + 𝑥𝑛 [ ⋮ ]
𝑎𝑚1 𝑎𝑚2 𝑎𝑚𝑛
= 𝑥1 𝐴1 + 𝑥2 𝐴2 + ⋯ + 𝑥𝑛 𝐴𝑛 . ■
Corollary1.
Let 𝐴 be an 𝑚 × 𝑛 matrix. The system 𝐴𝒙 = 𝑏 is
consistent if and only if 𝑏 is a linear combination of
columns of 𝐴.
Example 5.
Find all vectors 𝒃 = [𝑏1 𝑏2 𝑏3 𝑏4 ]𝑇 such that b is a linear
combination of 𝒗1 = [1 2 3 4]𝑇 and 𝒗2 = [4 3 2 1]T .

- 128 -
Solution.
Let 𝒃 = 𝛼1 𝒗1 + 𝛼2 𝒗2 , 𝛼1 , 𝛼2 ∈ ℝ .
1 4
Then 𝐴𝒙 = 𝒃, where 𝐴 = ቎2 3቏ , 𝑥 = [𝛼1 ].
3 2 𝛼2
4 1
From Corollary 1 above, it is enough to find 𝑏1 , 𝑏2 , 𝑏3 , 𝑏4
such that 𝐴𝒙 = 𝒃 is consistent. Therefore
1 4 𝑏1 1 4 𝑏1
[𝐴|𝑏] = ቎2 3቏ 𝑏2 ] ~ ቎0 −5቏ 𝑏2 − 2𝑏1
]
3 2 𝑏3 0 0 𝑏3 + 𝑏1 − 2𝑏2
4 1 𝑏4 0 0 𝑏4 + 2𝑏1 − 3𝑏2
This system is consistent if 𝑏3 + 𝑏1 − 2𝑏2 = 0 and
𝑏4 + 2𝑏1 − 3𝑏2 = 0 .
Hence
𝒃 = [𝑏1 𝑏2 2𝑏2 − 𝑏1 3𝑏2 − 2𝑏1 ]𝑇 . ◄
If 𝒗1 , 𝒗2 , … , 𝒗𝑟 are in vector space 𝕍, then generally
some vectors in 𝕍 may be a linear combinations of
𝒗1 , 𝒗2 , … , 𝒗𝑟 and others may not. The following theorem
shows that if we construct a set 𝕎 consisting of all those
vectors that are expressible as linear combinations of
𝒗1 , 𝒗2 , … , 𝒗𝑟 , then 𝕎 forms a subspace of 𝕍.

- 129 -
Theorem 2.
If 𝒗1 , 𝒗2 , … , 𝒗𝑟 are vectors in a vector space 𝕍, then
(a) The set 𝕎 of all linear combinations of 𝒗1 , 𝒗2 , … , 𝒗𝑟
is a subspace of 𝕍. i. e.,
(𝕎 = {𝒘: 𝒘 = 𝛼1 𝒗1 + ⋯ + 𝛼𝑟 𝒗𝑟 , 𝒗1 , … , 𝒗𝑟 ∈ 𝕍} is a
subspace of 𝕍).
(b) 𝕎 is the smallest subspace of 𝕍 that contains
𝒗1 , 𝒗2 , … , 𝒗𝑟 . In the since that every other subspace that
contains 𝒗1 , 𝒗2 , … , 𝒗𝑟 must contain 𝕎.
Proof.
(a) To show that 𝕎 is a subspace of 𝕍, we prove it is
closed under addition and scalar multiplication.
There is at least one vector in 𝕎, namely, 𝝑, since
𝝑 = 0𝒗1 + 0𝒗2 + ⋯ + 0𝒗𝑟 . If u and v are vectors in 𝕎,
then 𝒖 = 𝑐1 𝒗1 + 𝑐2 𝒗2 + ⋯ + 𝑐𝑟 𝒗𝑟 and 𝒗 = 𝑘1 𝒗1 +
𝑘2 𝒗2 + ⋯ + 𝑘𝑟 𝒗𝑟 , where 𝑐1 , 𝑐2 , … , 𝑐𝑟 , 𝑘1 , 𝑘2 , … , 𝑘𝑟 are
scalars. Therefore
𝒖 + 𝒗 = (𝑐1 + 𝑘1 )𝒗1 + (𝑐2 + 𝑘2 )𝒗2 + ⋯ + (𝑐𝑟 + 𝑘𝑟 )𝒗𝑟
and, for any scalar 𝑘, 𝑘𝒖 = (𝑘𝑐1 )𝒗1 + (𝑘𝑐2 )𝒗2 + ⋯ +
(𝑘𝑐𝑟 )𝒗𝑟 . Thus 𝒖 + 𝒗 and 𝑘 𝒖 are linear combinations of

- 130 -
𝒗1 , 𝒗2 , … , 𝒗𝑟 and consequently elements in 𝕎. Therefore,
𝕎 is closed under addition and scalar multiplication.
(b) Each vector 𝒗𝑖 is a linear combination of
𝒗1 , 𝒗2 , … , 𝒗𝑟 since we can write
𝒗𝑖 = 0𝒗1 + 0𝒗2 + ⋯ + 1𝒗𝑖 + ⋯ + 0𝒗𝑟
Therefore, the subspace 𝕎 contains each of the vectors
𝒗1 , 𝒗2 , … , 𝒗𝑟 . Let 𝕎′ be any other subspace that contains
𝒗1 , 𝒗2 , … , 𝒗𝑟 . Since 𝕎′ is closed under addition and
scalar multiplication, it must contain all linear
combinations of 𝒗1 , 𝒗2 , … , 𝒗𝑟 . Thus 𝕎′ contains each
vector of 𝕎. ■
We have the following definition.
Definition 2.
If 𝕊 = {𝒗1 , 𝒗2 , … , 𝒗𝑟 } is a set of vectors in a vector space
𝕍, then the subspace 𝕎 of 𝕍 consisting of all linear
combinations of vectors of 𝕊 is called the space spanned
by the vectors 𝒗1 , 𝒗2 , … , 𝒗𝑟 and we say that the vectors
𝒗1 , 𝒗2 , … , 𝒗𝑟 span 𝕎.
To indicate that 𝕎 is the space spanned by the vectors in
the set 𝕊 = {𝒗1 , 𝒗2 , … , 𝒗𝑟 }, we write 𝕎 = span (𝕊) or
𝕎 = span {𝒗1 , 𝒗2 , … , 𝒗𝑟 } .
- 131 -
Theorem 3.
Let 𝑆 = {𝒗1 , 𝒗2 , … , 𝒗𝑟 } ⊆ ℝ𝑛 and 𝐴 be an 𝑛 × 𝑘 matrix
whose columns are 𝒗1 , 𝒗2 , … , 𝒗𝑟 . Then 𝕊 spans ℝ𝑛 if
and only if the system 𝐴𝒙 = 𝑏 is consistent for every
𝑏 ∈ ℝ𝑛 .
Example 6.
Determine whether 𝒗1 = (1,1,2), 𝒗2 = (1,0,1), and
𝒗3 = (2,1,3) span the vector space ℝ3 , or not.?
Solution.
We must determine whether an arbitrary vector 𝒃 =
(𝑏1 , 𝑏2 , 𝑏3 ) in ℝ3 can be expressed as a linear
combination 𝒃 = 𝑘1 𝒗1 + 𝑘2 𝒗2 + 𝑘3 𝒗3 of the vectors
𝒗1 , 𝒗2 and 𝒗3 . Expressing this equation in terms of
components gives
(𝑏1 , 𝑏2 , 𝑏3 ) = 𝑘1 (1,1,2) + 𝑘2 (1,0,1) + 𝑘3 (2,1,3)
or
𝑘1 + 𝑘2 + 2𝑘3 = 𝑏1
𝑘1 + 𝑘3 = 𝑏2
2𝑘1 + 𝑘2 + 3𝑘3 = 𝑏3

- 132 -
This system is consistent for all 𝑏1 , 𝑏2 and 𝑏3 if and only
1 1 2
if the coefficient matrix 𝐴 = [1 0 1] is invertible.
2 1 3
But, Det(𝐴) = 0 (verify), so that 𝐴 is not invertible.
Consequently, 𝑣1 , 𝑣2 and 𝑣3 do not span ℝ3 . ◄
Theorem 4.
If 𝕊 = {𝒗1 , … , 𝒗𝑟 } and 𝕊′ = {𝒘1 , … , 𝒘𝑟 } are two sets of
vectors in a vector space 𝕍, then span{𝒗1 , … , 𝒗𝑟 } =
span{𝒘1 , … , 𝒘𝑟 } if and only if each vector in 𝕊 is a
linear combination of those in 𝕊′. And conversely each
vector in 𝕊′ is a linear combination of those in 𝕊.
Example 7.
Prove: 𝑝(𝑥 ) = 𝑥 2 + 1, 𝑞 (𝑥 ) = 𝑥 − 2 and r(𝑥 ) = 𝑥 + 3
span ℙ2
Solution.
Let 𝑚(𝑥 ) = 𝑎𝑥 2 + 𝑏𝑥 + 𝑐 ∈ ℙ2 such that 𝑚(𝑥 ) =
𝛼1 𝑝(𝑥 ) + 𝛼2 𝑞(𝑥 ) + 𝛼3 𝑟(𝑥) for 𝛼1 , 𝛼2 , 𝛼3 ∈ ℝ.
Therefore,
𝛼1 = 𝑎, 𝛼2 + 𝛼3 = 𝑏, 𝛼1 − 2𝛼2 + 3𝛼3 = 𝑐.
Hence the coefficient matrix is

- 133 -
 1 0 0
𝐴 = [0 1 1]
1 −2 3
is invertible (verify?).
Thus, the system above is consistent. So, 𝑚(𝑥) is a linear
combination of 𝑝, 𝑞, 𝑟 and then span 𝕡2 .◄
●The subspace 𝕎 in Theorem 2 above is called the
subspace spanned by 𝕊 and is denoted by ‫ۄ𝕊ۃ‬.
Remark:
(1) If 𝕊 spans 𝕍, then 𝕍 = ‫ۄ𝕊ۃ‬.
(2) If 𝕊 = 𝜙 or 𝕊 = {𝝑} , then ‫}𝝑{ = ۄ𝕊ۃ‬.
(3) If 𝕊 is a subspace of 𝕍, then ‫𝕊 = ۄ𝕊ۃ‬
Example 8.
Find the vectors that spans the subspace 𝕎 of ℝ3 , where
𝕎 = {(𝑥, 𝑦, 𝑧) ∈ ℝ3 : 𝑥 + 𝑦 + 𝑧 = 0}.
Solution.
Since 𝑥 + 𝑦 + 𝑧 = 0, then 𝑧 = −𝑥 − 𝑦. If (𝑥, 𝑦, 𝑧) ∈ 𝕎,
(𝑥, 𝑦, 𝑧) = (𝑥, 𝑦, −𝑥 − 𝑦)
= (𝑥, 0, −𝑥 ) + (0, 𝑦, −𝑦)
= 𝑥 (1,0, −1) + 𝑦(0,1, −1)
Hence 𝕊 = {(1, 0, −1), (0, 1, −1)} spans 𝕎. ◄

- 134 -
 Exercise Set (4.4)
In Exercise 1–8, express the first vector as a linear
combination of the other vectors.
1. (1, −7,5) ; (1, −1,1) , (1,0,1) , (1,1,0);
2. (−9, −7, −15) ; (2,1,4) , (1, −1,3) , (3,2,5);
3. (0,0,0) ; (2, 1, 4) , (1, −1, 3) , (3, 2, 5);
4. (2,3, −7,3) ; (2,1,0,3) , (3, −1,5,2) , (−1,0,2,1);
5. 𝑥 2 + 1 ; −𝑥 + 1 , 3𝑥 2 + 𝑥 + 2;
6. 2𝑥 2 − 3𝑥 + 1 ; 𝑥 + 1 , 𝑥 2 + 𝑥 , 𝑥 2 + 2;
7 19 2 4 3 1 0 10
7. [ ];[ ], [ ], [ ];
−3 8 0 2 3 2 −6 2
6 −8 4 2 1 −1 0 2
8. [ ];[ ],[ ], [ ];
−1 −8 −2 −2 2 3 1 4
9. Is 𝑤 = (−1, 4, 15) ∈ ‫ۄ𝕊ۃ‬, where
𝕊 = {𝒗1 = (1, 2, 8) , 𝒗2 = (3, 0, 1)} ⊆ ℝ3 .
10. Is 𝑝(𝑥 ) = 6𝑥 2 + 4𝑥 ∈ ‫ۄ𝕊ۃ‬, where
𝕊 = {𝑃1 (𝑥 ) = 𝑥 2 + 𝑥 + 1, 𝑃2 (𝑥 ) = 3𝑥 2 + 2𝑥 − 2,
𝑃3 (𝑥 ) = 6𝑥 3 + 5𝑥 2 + 5𝑥 + 1} ⊆ 𝕡2
05
11. Is 𝐴 = [ ] ∈ ‫ۄ𝕊ۃ‬, where
−4
−1
1 2 3 1
𝕊 = {𝐴1 = [ ] , 𝐴2 = [ ] ⊆ 𝕄2×2 }
−1 0 1 1

- 135 -
12. Which of the following subsets of ℝ3 spans ℝ3 .
(a) 𝕊1 = {(2,0,2), (3,1,1), (−3,5,5)}.
(b) 𝕊2 = {(1,5,1), (2,6,1), (3,3,0), (4,6,2)}.
13. Which of the following subsets of 𝕡2 spans 𝕡2
(a) 𝕊1 = {−𝑥 2 + 1, 𝑥 2 + 1, 6𝑥 2 + 5𝑥 + 2}.
(b) 𝕊2 = {3𝑥, 𝑥 + 1, 2𝑥 2 + 1}.
14. Which of the following subsets of 𝕄2×2 spans 𝕄2×2
1 0 1 0 0 1 1 1
(a) 𝕊1 = {[ ], [ ], [ ], [ ]}.
0 0 0 1 1 0 0 1
2 1 3 1 2 2 2 3
(b) 𝕊2 = {[ ], [ ], [ ], [ ]}.
−1 0 1 1 1 1 −4 −1

- 136 -
5. linear Independence
In the preceding section we learned that asset of vectors
𝕊 = {𝒗1 , 𝒗2 , … , 𝒗𝑟 } spans a linear space 𝕍 if every vector
in 𝕍 is expressible as a linear combination of the vectors
in 𝕊. In general, there may be more than one way to
express a vector in 𝕍 as a linear combination of vectors in
a spanning set. In this section we will study conditions
under which each vector in 𝕍 is expressible as a linear
combination of the spanning vectors in exactly one way.
Spanning sets with this property play a fundamental role
in the study of vector spaces.
Definition 1.
If 𝕊 = {𝒗1 , 𝒗2 , … , 𝒗𝑟 } is nonempty set of vectors, then the
vector equation
𝑘1 𝒗1 + 𝑘2 𝒗2 + ⋯ + 𝑘𝑟 𝒗𝑟 = 𝝑
has at least one solution, namely
𝑘1 = 0, 𝑘2 = 0, … , 𝑘𝑟 = 0
If this is the only solution, then 𝕊 is called a linearly
independent set. If there are other solutions, then 𝕊 is
called linearly dependent set.

- 137 -
Example 1.
If 𝒗1 = (2, −1, 0, 3), 𝒗2 = (1, 2, 5, −1) and 𝒗3 =
(7, −1, 5, 8), then the set vectors 𝕊 = {𝑣1 , 𝒗2 , 𝒗3 } is
linealy dependent, since 3𝒗1 + 𝒗2 − 𝒗3 = 𝝑 .◄
Example 2.
The polynomials 𝑃1 = 1 − 𝑥, 𝑃2 = 5 + 3𝑥 − 2𝑥 2 , and
𝑃3 = 1 + 3𝑥 − 𝑥 2 form a linearly dependent set in 𝕡2
since 3𝑃1 − 𝑃2 + 2𝑃3 = 𝝑 = 0.◄
Example 3.
Consider the vectors 𝒊 = (1,0,0), 𝒋 = (0,1,0), and 𝒌 =
(0,0,1) in ℝ3 . In terms of components the vector equation
𝑘1 𝒊 + 𝑘2 𝒋 + 𝑘3 𝒌 = 𝝑 becomes
𝑘1 (1,0,0) + 𝑘2 (0,1,0) + 𝑘3 (0,0,1) = (0,0,0)
Or equivalently, (𝑘1 , 𝑘2 , 𝑘3 ) = (0,0,0)
This implies that 𝑘1 = 0, 𝑘2 = 0 and 𝑘3 = 0. So, the set
𝕊 = {𝒊, 𝒋, 𝒌} is linearly independent.
●A similar argument can be used to show that the vectors
𝒆1 = (1,0,0, … ,0), 𝒆2 = (0,1,0, … ,0), …,
𝒆𝑛 = (0,0, … ,1) form a linearly independent set in ℝ𝑛 .◄

- 138 -
Example 4.
Determine whether the vectors
𝒗1 = (1, −2, 3), 𝒗2 = (5, 6, 1), 𝒗3 = (3, 2, 1)
from a linearly dependent set or linearly independent set.
Solution.
In terms of components the vector equation
𝑘1 𝒗1 + 𝑘2 𝒗2 + 𝑘3 𝒗3 = 𝝑
becomes
𝑘1 (1, −2,3) + 𝑘2 (5,6, −1) + 𝑘3 (3,2,1) = (0,0,0)
Equating corresponding components gives
𝑘1 + 5𝑘2 + 3𝑘3 = 0
−2𝑘1 + 6𝑘2 + 2𝑘3 = 0
3𝑘1 − 𝑘2 + 𝑘3 = 0
Thus, 𝒗1 , 𝒗2 and 𝒗3 form a linearly dependent set if this
system has a non-trivial solution or a linearly independent
set if it has only the trivial solution. Solving this system
1 1
gives 𝑘1 = − 𝑡, 𝑘2 = − 𝑡, 𝑘3 = 𝑡, 𝑡 ∈ ℝ
2 2

Thus, the system has non-trivial solutions and 𝒗1 , 𝒗2 and

𝒗3 form a linearly dependent set. ◄

- 139 -
Example 4.
Consider the vector space ℂ[−1, 1] of all continuous real-
valued functions defined on interval {𝑥 ∈ ℝ: −1 ≤ 𝑥 ≤
1} and the functions 𝑓, 𝑔, ℎ ∈ ℂ[−1, 1] defined as
0, −1 ≤ 𝑥 ≤ 0
𝑓(𝑥 ) = 1, −1 ≤ 𝑥 ≤ 1, 𝑔(𝑥 ) = {
𝑥, 0 ≤ 𝑥 ≤ 1
0, −1 ≤ 𝑥 ≤ 0
and ℎ(𝑥 ) = {
𝑥2, 0 ≤ 𝑥 ≤ 1
Prove that the functions 𝑓, 𝑔 and ℎ form a linearly
independent set.
Solution.
In terms of components the vector equation
𝑘1 𝑓(𝑥 ) + 𝑘2 𝑔(𝑥 ) + 𝑘3 ℎ(𝑥 ) = 𝝑(𝑥 ) = 0, −1 ≤ 𝑥 ≤ 1
At 𝑥 = 0, we have 𝑘1 = 0.
At 𝑥 = 1, we have 𝑘1 + 𝑘2 + 𝑘3 = 0.
1 1 1
At 𝑥 = , we have 𝑘1 + 𝑘2 + 𝑘3 = 0.
2 2 4

These equations have only the trivial solution 𝑘1 = 𝑘2 =

𝑘3 = 0. Therefore, the functions are linearly independent.
However, if we restrict these same functions to the
interval {𝑥 ∈ ℝ: −1 ≤ 𝑥 ≤ 0}, then they are dependent as
0 ∙ 𝑓(𝑥 ) + 1 ∙ 𝑔(𝑥 ) + 1 ∙ ℎ(𝑥 ) = 0. ◄

- 140 -
●Alternatively, we could show the existence of non-
trivial solutions without solving the system by showing
that the coefficient matrix has determinant Zero and
consequently is not invertible (verify).
Theorem 1.
A set 𝕊 with two or more vectors is:
(a) Linearly dependent if and only if at least one of the
vectors in 𝕊 is expressible as a linear combination of the
other vectors in 𝕊.
(b) Linearly independent if and only if no vector in 𝕊 is
expressible as a linear combination of the other vectors in
𝕊.
Proof:
(a) Let 𝕊 = {𝒗1 , 𝒗2 , … , 𝒗𝑟 } be a set with two or more
vectors. If we assume that 𝑆is linearly dependent, then
there are scalars 𝑘1 , 𝑘2 , … , 𝑘𝑟 , not all zero, such that
𝑘1 𝒗1 + 𝑘2 𝒗2 + ⋯ + 𝑘𝑟 𝒗𝑟 = 𝝑 (∗)
To be specific, suppose that 𝑘𝑗 ≠ 0, 1 ≤ 𝑗 ≤ 𝑟 then (*)
can be rewritten as
𝑘1 𝑘2 𝑘𝑗−1
𝒗𝑗 = (− ) 𝒗1 + (− ) 𝒗2 + ⋯ + (− ) 𝒗𝑗−1
𝑘𝑗 𝑘𝑗 𝑘𝑗
- 141 -
 𝑘𝑗+1 𝑘
+ (− ) 𝒗𝑗+1 + ⋯ + (− 𝑟 ) 𝒗𝑟 .
𝑘𝑗 𝑘𝑗

Which expresses 𝒗𝑗 as a linear combination of the other

vectors in 𝕊.
Conversely, assume that at least one of the vectors in 𝕊 is
expressible as a linear combination of the other vectors.
To be specific, suppose that
𝒗𝑗 = 𝑐1 𝒗1 + ⋯ + 𝑐𝑗−1 𝒗𝑗−1 + 𝑐𝑗+1 𝒗𝑗+1 + ⋯ + 𝑐𝑟 𝒗𝑟
So,
𝑐1 𝒗1 + ⋯ + (−1)𝒗𝑗 + ⋯ + 𝑐𝑟 𝒗𝑟 = 𝝑
It follows that 𝕊 is linearly dependent since the equation
𝑘1 𝒗1 + 𝑘2 𝒗2 + ⋯ + 𝑘𝑟 𝒗𝑟 = 𝝑
is satisfied by 𝑘1 = 𝑐1 , 𝑘2 = 𝑐2 , … , 𝑘𝑗 = −1, 𝑘𝑟 = 𝑐𝑟
which are all not zero.
(b) Similar to (a). ■

- 142 -
Example 5.
In Example 1 above we saw that the vectors 𝒗1 =
(2, −1, 0, 3), 𝒗2 = (1, 2, 5, −1), and 𝒗3 = (7, −1, 5, 8)
form a linearly dependent set. It follows from Theorem 1
above that at least one of these vectors is expressible as a
linear combination of the other two. In this example each
vector is expressible as a linear combination of the other
two since it follows from the equation 3𝒗1 + 𝒗2 − 𝒗3 =
1 1
𝝑 that 𝒗1 = − 𝒗2 + 𝒗3 , 𝒗2 = −3𝒗1 + 𝒗3 , and 𝒗3 =
3 3

3𝒗1 + 𝒗2 .◄
Example 6.
In Example 3 in this section we saw that the vectors 𝒊 =
(1,0,0), 𝒋 = (0,1,0), and 𝒌 = (0,0,1) form a linearly
independent set. Thus, it follows from Theorem 1 above
that none of these vectors is expressible as a linear
combination of the other two. To see this directly,
suppose that k is expressible as
𝒌 = 𝛼1 𝒊 + 𝛼2 𝒋.
Then, in terms of components,
(0,0,1) = 𝛼1 (1,0,0) + 𝛼2 (0,1,0).

- 143 -
Or
(0,0,1) = (𝛼1 , 𝛼2 , 0).
But this equation is not satisfied by any values of 𝛼1 and
𝛼2 . So, k cannot be expressed as a linear combination of i
and j. Similarly, i is not expressible as a linear
combination of j and k, and j is not expressible as a linear
combination of i and k.◄
Theorem 2.
(a) A finite set of vectors that contains the zero vector is
linearly dependent.
(b) A set with exactly two vectors is linearly independent
if and only if neither of the vectors is a scalar multiple of
the other.
Theorem 3.
Let 𝕊 = {𝒗1 , 𝒗2 , … , 𝒗𝑟 } be a set of vectors in ℝ𝑛 . If 𝑟 >
𝑛 , then 𝕊 is linearly dependent.
Remark.
The preceding theorem tells us that a set in ℝ2 with more
than two vectors is linearly dependent, and a set in ℝ3
with more than three vectors is linearly dependent.

- 144 -
Example 7.
Determine whether the following set of vectors in 𝕡2 is
linearly dependent
1 + 𝑥, 3𝑥 + 𝑥 2 , 2 + 𝑥 − 𝑥 2 .
Solution.
Let 𝛼1 (1 + 𝑥 ) + 𝛼2 (3𝑥 + 𝑥 2 ) + 𝛼3 (2 + 𝑥 − 𝑥 2 ) = 𝝑.
Then
𝛼1 + 2𝛼3 = 0
𝛼1 + 3𝛼2 + 𝛼3 = 0
𝛼2 − 𝛼3 = 0
The determinant of the coefficient's matrix A for this
system of equation is
1 0 2
det 𝐴 = |1 3 1 | = −2 ≠ 0.
0 1 −1
Therefore, A is invertible, and the only solution is the

trivial solution 𝛼1 = 𝛼2 = 𝛼3 = 0.Thus the vectors are

linearly independent. ◄

- 145 -
 Exercise Set (4.5)
In Exercise 1–5, determine whether the given set of
vectors is linearly dependent or independent in the given
vector's spaces:
1. 𝕊 = {(1, −1), (2, 0)}, 𝕍 = ℝ2 ;

2. 𝕊 = {(1, 1, 2), (1, 4, 5), (1, 2, 7), (−1, 8, 3)}, 𝕍 = ℝ4 ;

3. 𝕊 = {4𝑥 2 − 𝑥 + 2, 2𝑥 2 + 6𝑥 + 3, −4𝑥 2 + 10𝑥 + 2},

𝕍 = 𝕡2 ;

1 1 1 0 1 0
4. 𝕊 = {[ ], [ ], [ ]} , 𝕍 = 𝕄2×2;
0 1 1 1 0 1

5. Find x such that 𝕊 is linearly independent in ℝ3 , where

𝕊 = {(2, 𝑥, 1), (1,0,1), (0,1,3)}
6. Find 𝑎, 𝑏 such that the vectors (1, 𝑎, 2, 𝑏),
(1, −1,2,0), (1,2,1,3) are linearly dependent in ℝ4 .
7. Prove that any nonempty subset of linearly
independent set is linearly independent.
8. Let 𝕊1 , 𝕊2 be subsets of a vector space 𝕍 such that
𝕊1 ⊆ 𝕊2 and 𝕊1 is linearly dependent. Prove that 𝕊2 is
linearly dependent.

- 146 -
6. Basis and Dimension

Let 𝕊1 = {(2, 1), (1, −1), (3, −2)} and 𝕊2 =

{(1, −1), (1, 1)} be two subset of ℝ2 . It is easy to show
that 𝕊1 is linearly dependent and 𝕊2 is linearly
independent. Therefore, 𝕊2 is a basis for ℝ2 but 𝕊1 is not.
In this section, we discuss the notion of basis in vector
spaces. We will show that each basis for a vector space
contains the same number of elements. This number is
called the dimension of vector space. The number of
elements in a basis for a vector space 𝕍 depends only on
𝕍 and not on the way we chose the basis.
Definition 1.
If 𝕍 is any vector space and 𝕊 = {𝒗1 , 𝒗2 , … , 𝒗𝑛 } is a set
of vectors in 𝕍, then 𝕊 is called a basis for 𝕍 if the
following two conditions hold:
(a) 𝕊 is linearly independent;
(b) 𝕊 spans 𝕍.

- 147 -
Theorem 1.
If 𝕊 = {𝒗1 , 𝒗2 , … , 𝒗𝑛 } is a basis for a vector space 𝕍,
then every vector v in 𝕍 can be expressed in the form
𝒗 = 𝑐1 𝒗1 + 𝑐2 𝒗2 + ⋯ + 𝑐𝑛 𝒗𝑛 in exactly one way.
If 𝕊 = {𝒗1 , 𝒗2 , … , 𝒗𝑛 } is a basis for a vector space 𝕍, and
𝒗 = 𝑐1 𝒗1 + 𝑐2 𝒗2 + ⋯ + 𝑐𝑛 𝒗𝑛 is the expression for a
vector v in terms of the basis 𝕊, then the scalars
𝑐1 , 𝑐2 , … , 𝑐𝑛 are called the coordinates of v relative to the
basis 𝕊. The vector (𝑐1 , 𝑐2 , … , 𝑐𝑛 ) in ℝ𝑛 constructed from
these coordinates is called the coordinate vector of 𝕍
relative to 𝕊;
it is denoted by (𝒗)𝕊 = (𝑐1 , 𝑐2 , … , 𝑐𝑛 ).
Remark.
It should be noted that coordinate vectors depend not only
on the basis 𝕊 but also on the order in which the basis
vectors are written; a change in the order of the basis
vectors results in a corresponding change of order for the
entries in the coordinate vectors.

- 148 -
Example 1.
If 𝒊 = (1,0,0), 𝒋 = (0,1,0), and 𝒌 = (0,0,1) ,then it is
easy to show that 𝕊 = {𝒊, 𝒋, 𝒌} is a linearly independent
set in ℝ3 . This set also spans ℝ3 since any vector 𝒗 =
(𝑎, 𝑏, 𝑐) in ℝ3 can be written as
𝒗 = (𝑎, 𝑏, 𝑐)
= 𝑎(1,0,0) + 𝑏(0,1,0) + 𝑐(0,0,1)
= 𝑎𝒊 + 𝑏𝒋 + 𝑐𝒌 (1)
Thus, 𝕊 is a basis for ℝ3 ; it is called the standard basis
for ℝ3 .
Looking at the coefficients of i, j and k in (1), it follows
that the standard basis are 𝑎, 𝑏, and c.
So, (𝒗)𝑆 = (𝑎, 𝑏, 𝑐).
Comparing this result, we see that 𝒗 = (𝒗)𝕊 . ◄
●The result in the preceding example is a special case of
those in the next example.

- 149 -
Example 2.
As shown in the preceding section 𝕊 = {𝒆1 , 𝒆2 , … , 𝒆𝑛 } ,
where 𝒆1 = (1,0, … ,0), 𝒆2 = (0,1,0, … ,0) , … , 𝒆𝑛 =
(0,0, … ,1), is a linearly independent set in ℝ𝑛 . This set
also spans ℝ𝑛 . Since any vector 𝒗 = (𝑎1 , 𝑎2 , … , 𝑎𝑛 ) in
ℝ𝑛 can be written as
𝒗 = 𝑎1 𝒆1 + 𝑎2 𝒆2 + ⋯ + 𝑎𝑛 𝒆𝑛 (2)
Thus 𝕊 is a basis for ℝ𝑛 ; it is called the standard basis
for ℝ𝑛 . It follows from (2) that the coordinates of v
relative to the standard basis are
(𝒗)𝕊 = (𝑎1 , 𝑎2 , … , 𝑎𝑛 ) = 𝒗.◄
Remark.
We will see in a subsequent example that a vector and its
coordinate vector will not be the same; the equality that
we observed in the two preceding examples is a special
situation that occurs only with the standard basis for ℝ𝑛 .

- 150 -
Example 3.
Let 𝒗1 = (1, 2, 1), 𝒗2 = (2, 4, 0), and 𝒗3 = (3, 3, 4).
Show that the set 𝕊 = {𝒗1 , 𝒗2 , 𝒗3 } is a basis for ℝ3 .
Solution.
To show that the set 𝕊 spans ℝ3 , we must show that any
arbitrary vector 𝒃 = (𝑏1 , 𝑏2 , 𝑏3 ) can be expressed as a
linear combination 𝒃 = 𝑐1 𝒗1 + 𝑐2 𝒗2 + 𝑐3 𝒗3 of vectors
in 𝕊. Expressing this equation in terms of components
gives
(𝑏1 , 𝑏2 , 𝑏3 ) = 𝑐1 (1, 2, 1) + 𝑐2 (2, 4, 0) + 𝑐3 (3, 3, 4)
or on equating corresponding components
𝑐1 + 2𝑐2 + 3𝑐3 = 𝑏1
2𝑐1 + 9𝑐2 + 3𝑐3 = 𝑏2 (3)
𝑐1 + 4𝑐3 = 𝑏3
To show that 𝕊 spans ℝ3 , we must show the system (3)
has a solution for all choice of 𝒃 = (𝑏1 , 𝑏2 , 𝑏3 ).
To prove that 𝕊 is linearly independent, we must show
that the only solution of
𝑐1 𝒗1 + 𝑐2 𝒗2 + 𝑐3 𝒗3 = 𝝑 (4)
is 𝑐1 = 𝑐2 = 𝑐3 = 0.

- 151 -
As above, if (4) is expressed in terms of components, the
verification of independence reduces to showing that the
homogeneous system
𝑐1 + 2𝑐2 + 3𝑐3 = 0
2𝑐1 + 9𝑐2 + 3𝑐3 = 0 (5)
𝑐1 + 4𝑐3 = 0
has only the trivial solution. Observe that the systems (3)
and (5) have the same coefficient matrix. Thus, from
Theorem 4, in Section 2 of Chapter (III) we can
simultaneously prove that 𝕊 is linearly independent and
spans ℝ3 by demonstrating that in systems (3) and (5) the
matrix of coefficients
1 2 3
𝐴 = [2 9 3]
1 0 4
has a non-zero determinant. But Det (𝐴) = −1. So, 𝕊 is
a basis for ℝ3 . ◄

- 152 -
Example 4.
Let 𝕊 = {𝒗1 , 𝒗2 , 𝒗3 } be the basis for ℝ3 in the preceding
example.
(a) Find the coordinate vector of 𝒗 = (5, −1, 9) with
respect to 𝕊.
(b) Find the vector v in ℝ3 whose coordinate vector with
respect to the basis 𝕊 is (𝒗)𝕊 = (−1, 3, 2) .
Solution.
(a) We must find scalars 𝑐1 , 𝑐2 , 𝑐3 such that
𝒗 = 𝑐1 𝒗1 + 𝑐2 𝒗2 + 𝑐3 𝒗3
or, in terms of components,
(5, −1, 9) = 𝑐1 (1, 2, 1) + 𝑐2 (2, 9, 0) + 𝑐3 (3, 3, 4)
Equating corresponding components gives
𝑐1 + 2𝑐2 + 3𝑐3 = 5
2𝑐1 + 9𝑐2 + 3𝑐3 = −1
𝑐1 + 4𝑐3 = 9
Solving this system,
we obtain 𝑐1 = 1, 𝑐2 = −1, 𝑐3 = 2 (verify).
Therefore (𝒗)𝕊 = (1, −1, 2).

- 153 -
(b) Using the definition of the coordinate vector (𝒗)𝕊 ,
we obtain
𝒗 = (−1)𝒗1 + 3𝒗2 + 2𝒗3
= (−1)(1, 2, 1) + 3 (2, 9, 0) + 2(3, 3, 4)
= (11, 31, 7). ◄
Example 5.
(a) Show that 𝕊 = {1, 𝑥, 𝑥 2 , … , 𝑥 𝑛 } is a basis for the
vector space ℙ𝑛 of polynomials of the form 𝑎0 + 𝑎1 𝑥 +
⋯ + 𝑎𝑛 𝑥 𝑛 (Polynomials of degree n or less).
(b) Find the coordinate vector of the polynomial
𝒑 = 𝑎0 + 𝑎1 𝑥 + 𝑎2 𝑥 2 relative to basis 𝕊 = {1, 𝑥, 𝑥 2 } for
ℙ2 .
Solution.
(a) The student can easily show that 𝕊 spans ℙ𝑛 and 𝕊 is
linearly independent set. Thus, 𝕊 is a basis for ℙ𝑛 ; it is
called the standard basis for ℙ𝑛 .
(b) The coordinates of 𝒑 = 𝑎0 + 𝑎1 𝑥 + 𝑎2 𝑥 2 are the
scalar coefficients of the basis vectors 1, 𝑥, and 𝑥 2 so
(𝒑)𝑆 = (𝑎0 , 𝑎1 , 𝑎2 ).◄

- 154 -
Example 6.
1 0 0 1 0 0
Let 𝑴1 = [ ] , 𝑴2 = [ ] , 𝑴3 = [ ],
0 0 0 0 1 0
0 0
𝑴4 = [ ].
0 1
The set 𝕊 = {𝑴1 , 𝑴2 , 𝑴3 , 𝑴4 } is a basis for the vector
space 𝕄2×2 of all 2 × 2 Matrices.
Solution.
To see that 𝕊 spans 𝕄2×2 , note that an arbitrary vector
𝑎 𝑏
(matrix) [ ] can be written as
𝑐 𝑑
𝑎 𝑏 1 0 0 1 0 0 0 0
[ ] = 𝑎[ ]+𝑏[ ]+𝑐[ ]+𝑑[ ]
𝑐 𝑑 0 0 0 0 1 0 0 1
= 𝑎𝑴1 + 𝑏𝑴2 + 𝑐𝑴3 + 𝑑𝑴4
To see that 𝑆 is linearly independent, assume that
𝑎𝑴1 + 𝑏𝑴2 + 𝑐𝑴3 + 𝑑𝑴4 = 𝝑
That is,
1 0 0 1 0 0 0 0 0 0
𝑎[ ]+𝑏[ ]+𝑐[ ]+𝑑[ ]=[ ].
0 0 0 0 1 0 0 1 0 0
It follows that
𝑎 𝑏 0 0
[ ]=[ ]
𝑐 𝑑 0 0
Thus, 𝑎 = 𝑏 = 𝑐 = 𝑑 = 0. So, 𝕊 is linearly independent.
The basis 𝕊 in this example is called the standard basis
for 𝕄2×2 . ◄

- 155 -
●More generally, the standard basis for 𝕄𝑚×𝑛 consists
of the 𝑚 × 𝑛 different Matrices with a single l and zeros
for the remaining entries.
Definition 2.
A nonzero vector space 𝕍 is called finite-dimensional if
it contains a finite set of vectors {𝒗1 , 𝒗2 , … , 𝒗𝑛 } that
forms a basis. If no such set exists, 𝕍 is called infinite -
dimensional. In addition, we shall regard the zero-vector
space to be finite-dimensional.
Example 7.
The vector spaces ℝ𝑛 , ℙ𝑛 , and 𝕄𝑚×𝑛 are finite-
dimensional. The vector spaces 𝔽(−∞, ∞),
ℂ(−∞, ∞), ℂ𝑚 (−∞, ∞) and ℂ∞ (−∞, ∞) are infinite-
dimensional. ◄
Theorem 2.

If 𝕎 is a subspace of finite-dimensional vectors space 𝕍,

then dim(𝕎) ≤ dim (𝕍).

Moreover, if dim (𝕎) = dim (𝕍), then 𝕎 = 𝕍.

- 156 -
Theorem 3.
If 𝕍 is a finite-dimensional vector space and 𝕊 =
{𝒗1 , 𝒗2 , … , 𝒗𝑛 } is any basis. Then
(a) Every set with more than n vectors is linearly
dependent.
(b) No set with fewer than n vectors spans 𝕍.
Theorem 4.
All basis for a finite-dimensional vector space has the
same number of vectors.
Definition 3.
The dimension of a finite-dimensional vector space 𝕍,
denoted by dim (𝕍), is defined to be the number of
vectors in a basis for 𝕍. In addition, we define the zero-
vector space to have dimension zero.
Example 8.
dim (ℝ𝑛 ) = 𝑛 [The standard basis has n vectors].
dim (ℙ𝑛 ) = 𝑛 + 1[The standard basis has 𝑛 + 1 vectors].
dim (𝕄𝑚×𝑛 ) = 𝑚𝑛[The standard basis has 𝑚𝑛 vectors].
◄

- 157 -
Example 9.
Determine a basis for and the dimension of the solution
space of the homogenous system
2𝑥1 + 2𝑥2 − 𝑥3 + 𝑥5 = 0
−𝑥1 − 𝑥2 + 2𝑥3 − 3𝑥4 + 𝑥5 = 0
𝑥1 + 𝑥2 − 2𝑥3 − 𝑥5 = 0
𝑥3 + 𝑥4 + 𝑥5 = 0
Solution.
The general solution of the given system is
𝑥1 = −𝑠 − 𝑡 , 𝑥2 = 𝑠, 𝑥3 = −𝑡, 𝑥4 = 0, 𝑥5 = 𝑡 (Verify).
Therefore, the solution vectors can be written as
𝑥1 −𝑠 − 𝑡 −𝑠 −𝑡 −1 −1
𝑥2 𝑠 𝑠 0 1 0
𝑥3 = −𝑡 = 0 + −𝑡 = 𝑠 0 + 𝑡 −1
𝑥4 0 0 0 0 0
[𝑥5 ] [ 𝑡 ] [ 0 ] [ 𝑡 ] [0] [1]
Which shows that the vectors
𝒗1 = (−1, 1, 0, 0, 0) and 𝒗2 = (−1, 0, −1, 0, 1)
span the solution space. Since they are also linearly
independent (verify),{𝒗1 , 𝒗2 } is a basis and the solution
space is two-dimensional. ◄

- 158 -
 Exercise Set (4.6)
(1) Which of the following sets of vectors are basis for
ℝ2 ?
(a) {(1, 3) , (3, 5)};
(b) {(7, 1) , (10, 1)};
(c) {(7, 1) , (21, 3)}.
(2) Which of the following sets of vectors are basis for
ℝ3 ?
(a) {(0, 17, 8) , (1, 7, 5) , (2, −3, 2)};
(b) {(√2, 0, √2) , (√3, 0, √6) , (0, −√6, √3)};
(c) {(1, 2, 1) , (1, 3, 1) , (2, 3, 1)}.
(3) Which of the following sets of vectors are basis for
ℝ4 ?
(a) {(1,1,0,0) , (1,0,1,0) , (0,1,0,1) , (0,0,1,1)};
(b) {(3,0, −3,6) , (0,2,3,1) , (0, −2, −2,0) , (−2,1,2,1)}.
(4) Which of the following sets of vectors are basis for
ℙ2 ?
(a) {1, 𝑥 + 1, 𝑥 2 − 𝑥 };
(b) {𝑥 3 + 1 , 𝑥 3 − 1 , 𝑥 2 + 1 , 𝑥 2 − 1};
(c) {𝑥 2 + 3𝑥 + 5 , 5𝑥 2 + 3𝑥 + 1 , 𝑥 2 + 𝑥 + 1}.

- 159 -
(5) Which of the following sets of vectors are basis for
𝕄2×2 ?
1 0 0 −1 2 2 0 3
(a) 𝕊 = {[ ] ,[ ],[ ] ,[ ]};
0 1 −1 0 2 0 3 3
0 1 0 5 0 0 1 7
(b) 𝕊 = {[ ] ,[ ] ,[ ] ,[ ]};
4 4 5 0 2 −3 0 0
1 1 1 1 1 0 0 1
(c) 𝕊 = {[ ] ,[ ] ,[ ] ,[ ]}.
1 0 0 1 1 1 1 1
(6) Determine basis for and dimensions of the given
subspaces (≤ denotes subspace).
(a) 𝕎 = {(𝑎, 𝑏, 𝑐 ): 𝑎 + 𝑏 + 𝑐 = 0} ≤ ℝ3 ;
(b) 𝕎 = {(𝑎, 𝑏, 𝑐): 𝑎 + 𝑏 = 𝑏 + 𝑐 = 0} ≤ ℝ3 ;
(c) 𝕎 = {(𝑎 − 𝑏 , 𝑏 + 𝑐 , 𝑏, 𝑏 − 𝑐): 𝑎, 𝑏, 𝑐 ∈ ℝ} ≤ ℝ4 ;
(d) 𝕎 = {𝐴 ∈ 𝕄2×2 ∶ 𝐴𝑇 + 𝐴 = 0} ≤ 𝕄2×2 .
(e) 𝕎 = {𝐴 ∈ 𝕄2×2 : 2𝑎11 + 𝑎22 = 𝑎12 + 𝑎21 = 0} ≤
𝕄2×2
(f) 𝕎 = {𝑎(𝑥 + 1) + 𝑏(𝑥 2 + 𝑥 ): 𝑎, 𝑏 ∈ ℝ} ≤ ℙ2
(7) Determine the dimension of and a basis for the
solution space of the systems:
(a) 𝑥1 + 𝑥2 − 𝑥3 = 0; −2𝑥1 − 𝑥2 + 2𝑥3 = 0;
−𝑥1 + 𝑥3 = 0.
(b) 3𝑥1 + 𝑥2 + 𝑥3 + 𝑥4 = 0, 5𝑥1 − 𝑥2 + 𝑥3 − 𝑥4 = 0.

- 160 -
(8) Find the coordinate vector of w relative to the basis
𝕊 = {𝒖1 , 𝒖2 } for ℝ2 .
(a) 𝒖1 = (1, 0), 𝒖2 = (0, 1); 𝒘 = (3, −7);
(b) 𝒖1 = (1, 1), 𝒖2 = (0, 2) ; 𝒘 = (𝑎, 𝑏).
(10) Find the coordinate vector of p relative to the basis
𝕊 = {𝒑1 , 𝒑2 , 𝒑3 }, where
𝒑 = 4 − 3𝑥 + 𝑥 2 ; 𝒑1 = 1, 𝒑2 = 𝑥 , 𝒑3 = 𝑥 2
(11) Find the coordinate vector of A relative to the basis
𝕊 = {𝑨1 , 𝑨2 , 𝑨3 , 𝑨4 } , where
2 0 −1 1 1 1
𝑨=[ ] , 𝑨1 = [ ] , 𝑨2 = [ ],
−1 3 0 0 0 0
0 0 0 0
𝑨3 = [ ] , 𝑨4 = [ ].
1 0 0 1

- 161 -
- 162 -
1. Inner product
Definition 1.
An inner product on a linear space 𝕍 having the real
numbers as scalars is a function ‫ ∙ۃ‬, ∙‫ۄ‬: 𝕍 × 𝕍 → ℝ that
associates with each pair of elements u and v real number
‫𝒖ۃ‬, 𝒗‫ ۄ‬in such a way that the following axioms are
satisfied for all elements u , v and w of 𝕍 and 𝛼 ∈ ℝ:
(1) ‫𝒖ۃ‬, 𝒗‫𝒗ۃ = ۄ‬, 𝒖‫[ ;ۄ‬Symmetry axiom]
(2) ‫ 𝒖ۃ‬+ 𝒗, 𝒘‫𝒖ۃ = ۄ‬, 𝒘‫ ۄ‬+ ‫𝒗ۃ‬, 𝒘‫[ ;ۄ‬Additively axiom]
(3) ‫𝒖𝛼ۃ‬, 𝒗‫𝒖ۃ 𝛼 = ۄ‬, 𝒗‫[ ;ۄ‬Homogeneity axiom]
(4) ‫𝒗ۃ‬, 𝒗‫ ≥ ۄ‬0 [Positivity axiom]
(5) ‫𝒗ۃ‬, 𝒗‫ = ۄ‬0 if and only if 𝒗 = 𝝑.
A linear space 𝕍 with an inner product function ‫ ∙ۃ‬,∙‫ ۄ‬is
called an inner product space, denoted by (𝕍, ‫ ∙ۃ‬,∙‫)ۄ‬.
Example 1.
Let 𝕍 = ℝ𝑛 and 𝒖 = (𝑎1 , … , 𝑎𝑛 ) and 𝒗 = (𝑏1 , … , 𝑏𝑛 ) be
elements in 𝕍. Then the function defined by:
‫𝒖ۃ‬, 𝒗‫𝑎 = 𝒗 ∙ 𝒖 = ۄ‬1 𝑏1 + ⋯ + 𝑎𝑛 𝑏𝑛

- 163 -
is an inner product (called Euclidean Inner Product). In
this case, 𝕍 is called Euclidean Product Space.
Proof.
Let 𝒖 = (𝑎1 , … , 𝑎𝑛 ) , 𝒗 = (𝑏1 , … , 𝑏𝑛 ), and 𝒘 =
(𝑐1 , … , 𝑐𝑛 ) be elements in ℝ𝑛 and 𝛼 ∈ ℝ. Then
(1) ‫𝒖ۃ‬, 𝒗‫𝑎 = ۄ‬1 𝑏1 + ⋯ + 𝑎𝑛 𝑏𝑛
= 𝑏1 𝑎1 + ⋯ + 𝑏𝑛 𝑎𝑛
= ‫𝒗ۃ‬, 𝒖‫ۄ‬.
(2) ‫ 𝒖ۃ‬+ 𝒗, 𝒘‫𝑎( = ۄ‬1 + 𝑏1 )𝑐1 + ⋯ + (𝑎𝑛 + 𝑏𝑛 )𝑐𝑛

= 𝑎1 𝑐1 + 𝑏1 𝑐1 + ⋯ + 𝑎𝑛 𝑐𝑛 + 𝑏𝑛 𝑐𝑛

= (𝑎1 𝑐1 + ⋯ + 𝑎𝑛 𝑐𝑛 ) + (𝑏1 𝑐1 + ⋯ + 𝑏𝑛 𝑐𝑛 )

= ‫𝒖ۃ‬, 𝒘‫ ۄ‬+ ‫𝒗ۃ‬, 𝒘‫ۄ‬.

(3) ‫𝒖𝛼ۃ‬, 𝒗‫𝑎𝛼( = ۄ‬1 )𝑏1 + ⋯ + (𝛼𝑎𝑛 )𝑏𝑛

= 𝛼 (𝑎1 𝑏1 + ⋯ + 𝑎𝑛 𝑏𝑛 ) = 𝛼 ‫𝒖ۃ‬, 𝒗‫ۄ‬.

(4) ‫𝒖ۃ‬, 𝒖‫𝑎 = ۄ‬12 + ⋯ + 𝑎𝑛2 ≥ 0

(5) ‫𝒖ۃ‬, 𝒖‫ = ۄ‬0 ⟺ 𝑎12 + ⋯ 𝑎𝑛2 = 0 ⟺

𝑎1 = ⋯ = 𝑎𝑛 = 0 ⟺ 𝒖 = 𝝑.

Therefore, ℝ𝑛 is an inner product space. ■

- 164 -
Remark.
The Euclidian inner product is the most important inner
product on ℝ𝑛 . However, there are various applications
in which it is desirable to modify the Euclidean inner
product by weighting its terms differently. More
precisely, if 𝛼1 , … , 𝛼𝑛 are positive real numbers, which
we shall call weights, and if 𝒖 = (𝑢1 , … , 𝑢𝑛 ) and 𝒗 =
(𝑣1 , … , 𝑣𝑛 ) are vectors in ℝ𝑛 . Then it can be shown (left
as exercise) that the function

‫𝒖ۃ‬, 𝒗‫𝛼 = ۄ‬1 𝑢1 𝑣1 + ⋯ + 𝛼𝑛 𝑢𝑛 𝑣𝑛 (∗)

defines an inner product on ℝ𝑛 , it is called weighted

Euclidean inner product with weights 𝛼1 , … , 𝛼𝑛 .

It will always be assumed that ℝ𝑛 has the Euclidean inner

product unless some other inner product is specially
specified.

- 165 -
Example 2.

Let 𝒖 = (𝑎1 , 𝑎2 ) , 𝒗 = (𝑏1 , 𝑏2 ) ∈ ℝ2 . Verify the weighted

Euclidean inner product ‫𝒖ۃ‬, 𝒗‫ = ۄ‬7𝑎1 𝑏1 + 5𝑎2 𝑏2 satisfies
the five inner product axioms.

Solution.

Let 𝒖 = (𝑎1 , 𝑎2 ), 𝒗 = (𝑏1 , 𝑏2 ), 𝒘 = (𝑐1 , 𝑐2 ) ∈ ℝ2

and 𝛼 ∈ ℝ . Then

(1) ‫𝒖ۃ‬, 𝒗‫ = ۄ‬7𝑎1 𝑏1 + 5𝑎2 𝑏2 = 7𝑏1 𝑎1 + 5𝑏2 𝑎2 = ‫𝒗ۃ‬, 𝒖‫ۄ‬.

(2) ‫ 𝒖ۃ‬+ 𝒗, 𝒘‫ = ۄ‬7(𝑎1 + 𝑏1 )𝑐1 + 5(𝑎2 + 𝑏2 )𝑐2

= (7𝑎1 𝑐1 + 5𝑎2 𝑐2 ) + (7𝑏1 𝑐1 + 5𝑏2 𝑐2 )

= ‫𝒖ۃ‬, 𝒘‫ ۄ‬+ ‫𝒗ۃ‬, 𝒘‫ۄ‬.

(3) ‫𝒖𝛼ۃ‬, 𝒗‫ = ۄ‬7(𝛼𝑎1 )𝑏1 + 5(𝛼𝑎2 )𝑏2

= 𝛼 (7𝑎1 𝑏1 + 5𝑎2 𝑏2 )

= 𝛼 ‫𝒖ۃ‬, 𝒗‫ۄ‬.

(4) ‫𝒖ۃ‬, 𝒖‫ = ۄ‬7𝑎12 + 5𝑎22 ≥ 0.

(5) ‫𝒖ۃ‬, 𝒖‫ = ۄ‬0 ⇔ 7𝑎12 + 5𝑎22 = 0 ⟺ 𝑎1 = 𝑎2 = 0

⟺ 𝒖 = 𝝑.◄

- 166 -
Example 3.
Let 𝒑(𝑥 ) = 𝑎0 + 𝑎1 𝑥 + 𝑎2 𝑥 2 , 𝒒(𝑥 ) = 𝑏0 + 𝑏1 𝑥 +
𝑏2 𝑥 2 ∈ ℙ2 ,where ‫) 𝑥(𝒑ۃ‬, 𝒒(𝑥 )‫𝑎 = ۄ‬0 𝑏0 + 𝑎1 𝑏1 + 𝑎2 𝑏2 .
Prove that ℙ2 is an inner product space.
Solution.
Let 𝛼 ∈ ℝ and 𝒑(𝑥 ) = 𝑎0 + 𝑎1 𝑥 + 𝑎2 𝑥 2 , 𝒒(𝑥 ) = 𝑏0 +
𝑏1 𝑥 + 𝑏2 𝑥 2 and 𝒓(𝑥 ) = 𝑐0 + 𝑐1 𝑥 + 𝑐2 𝑥 2 ∈ ℙ2
(1) ‫𝒑ۃ‬, 𝒒‫𝑎 = ۄ‬0 𝑏0 + 𝑎1 𝑏1 + 𝑎2 𝑏2

= 𝑏0 𝑎0 + 𝑏1 𝑎1 + 𝑏2 𝑎2 = ‫𝒒ۃ‬, 𝒑‫ۄ‬.

(2) ‫ 𝒑ۃ‬+ 𝒒, 𝒓‫𝑎( = ۄ‬0 + 𝑏0 )𝑐0 + (𝑎1 + 𝑏1 )𝑐1 +

(𝑎2 + 𝑏2 )𝑐2

= (𝑎0 𝑐0 + 𝑎1 𝑐1 + 𝑎2 𝑐2 ) + (𝑏0 𝑐0 + 𝑏2 𝑐1 + 𝑏2 𝑐2 )

= ‫𝒑ۃ‬, 𝒓‫ ۄ‬+ ‫𝒒ۃ‬, 𝒓‫ ۄ‬.

(3) ‫𝒑𝛼ۃ‬, 𝒒‫𝑎𝛼( = ۄ‬0 )𝑏0 + (𝛼𝑎1 )𝑏1 + (𝛼𝑎2 )𝑏2

= 𝛼 (𝑎0 𝑏0 + 𝑎1 𝑏1 + 𝑎2 𝑏2 ) = 𝛼 ‫𝒑ۃ‬, 𝒒‫ ۄ‬.
(4) ‫𝒑ۃ‬, 𝒑‫𝑎 = ۄ‬02 + 𝑎12 + 𝑎22 ≥ 0
(5) ‫𝒑ۃ‬, 𝒑‫ = ۄ‬0 ⇔ 𝑎02 + 𝑎12 + 𝑎22 = 0 ⟺
𝑎0 = 𝑎1 = 𝑎2 = 0 ⟺ 𝒑 = 𝝑.
Therefore ℙ2 is an inner product space. ◄

- 167 -
Example 4.
Let 𝒇, 𝒈 ∈ ℂ[𝑎, 𝑏], where ℂ[𝑎, 𝑏] is the set consists of all
continuous function on the closed interval [𝑎, 𝑏].
𝑏
Prove that the function ‫𝒇ۃ‬, 𝒈‫ 𝑥𝑑) 𝑥(𝒈) 𝑥(𝒇 𝑎∫ = ۄ‬defines
an inner product space on ℂ[𝑎, 𝑏] .
Solution.
Let 𝒇, 𝒈, 𝒉 ∈ ℂ[𝑎, 𝑏] and 𝛼 ∈ ℝ. Then
𝑏 𝑏
(1) ‫𝒇ۃ‬, 𝒈‫𝒈ۃ = 𝑥𝑑) 𝑥(𝒇) 𝑥(𝒈 𝑎∫ = 𝑥𝑑) 𝑥(𝒈) 𝑥(𝒇 𝑎∫ = ۄ‬, 𝒇‫ۄ‬.

𝑏
(2) ‫ 𝒇ۃ‬+ 𝒈, 𝒉‫ ) 𝑥(𝒇[ 𝑎∫ = ۄ‬+ 𝒈(𝑥 )] 𝒉(𝑥 )𝑑𝑥
𝑏 𝑏
= ∫𝑎 𝒇(𝑥 )𝒉(𝑥 )𝑑𝑥 + ∫𝑎 𝒈(𝑥 )𝒉(𝑥 )𝑑𝑥
= ‫𝒇ۃ‬, 𝒉‫ ۄ‬+ ‫𝒈ۃ‬, 𝒉‫ۄ‬.
𝑏 𝑏
(3) ‫𝒇𝛼ۃ‬, 𝒈‫𝑥𝑑) 𝑥(𝒈) 𝑥(𝒇 𝑎∫ 𝛼 = 𝑥𝑑) 𝑥(𝒈)) 𝑥(𝒇𝛼( 𝑎∫ = ۄ‬
= 𝛼 ‫𝒇ۃ‬, 𝒈‫ۄ‬.

(4) Since f is continuous on [𝑎, 𝑏], then [𝒇(𝑥)]2 ≥ 0 for

all 𝑥 ∈ [𝑎, 𝑏] Therefore ‫𝒇ۃ‬, 𝒇‫ ≥ ۄ‬0.
𝑏
(5) ‫𝒇ۃ‬, 𝒇‫ = ۄ‬0 ⇔ ∫𝑎 [𝒇(𝑥)]2 𝑑𝑥 = 0 ⟺ 𝒇(𝑥 ) = 𝝑.
Thus ℂ[𝑎, 𝑏] is an inner product space. ◄

- 168 -
‫‪Theorem 4.‬‬
‫‪Let 𝕍 be an inner product space and 𝒖, 𝒗, 𝒘 ∈ 𝕍,‬‬
‫‪𝛼 ∈ ℝ. Then‬‬
‫;‪ = 0‬ۄ𝝑 ‪𝒗,‬ۃ = ۄ𝒗 ‪𝝑,‬ۃ )‪(1‬‬
‫;ۄ𝒘 ‪𝒖,‬ۃ ‪ +‬ۄ𝒗 ‪𝒖,‬ۃ = ۄ𝒘 ‪𝒖, 𝒗 +‬ۃ )‪(2‬‬
‫;ۄ𝒗 ‪𝒖,‬ۃ 𝛼 = ۄ𝒗𝛼 ‪𝒖,‬ۃ )‪(3‬‬
‫;ۄ𝒘 ‪𝒗,‬ۃ ‪ −‬ۄ𝒘 ‪𝒖,‬ۃ = ۄ𝒘 ‪𝒖 − 𝒗,‬ۃ )‪(4‬‬
‫‪.‬ۄ𝒘 ‪𝒖,‬ۃ ‪ −‬ۄ𝒗 ‪𝒖,‬ۃ = ۄ𝒘 ‪𝒖, 𝒗 −‬ۃ )‪(5‬‬
‫‪Proof.‬‬
‫‪. Then‬ۄ𝒗 ‪𝝑,‬ۃ ‪ +‬ۄ𝒗 ‪𝝑,‬ۃ = ۄ𝒗 ‪𝝑 + 𝝑,‬ۃ = ۄ𝒗 ‪𝝑,‬ۃ‪(1) Note that‬‬
‫‪ = 0.‬ۄ𝒗 ‪𝝑,‬ۃ ‪. Therefore‬ۄ𝒗 ‪𝝑,‬ۃ = )ۄ𝒗 ‪𝝑,‬ۃ‪ + (−‬ۄ𝒗 ‪𝝑,‬ۃ‬

‫ۄ𝒖 ‪𝒘,‬ۃ ‪ +‬ۄ𝒖 ‪𝒗,‬ۃ = ۄ𝒖 ‪𝒗 + 𝒘,‬ۃ = ۄ𝒘 ‪𝒖, 𝒗 +‬ۃ )‪(2‬‬

‫‪.‬ۄ𝒘 ‪𝒖,‬ۃ ‪ +‬ۄ𝒗 ‪𝒖,‬ۃ =‬

‫‪.‬ۄ𝒗 ‪𝒖,‬ۃ𝛼 = ۄ𝒖 ‪𝒗,‬ۃ 𝛼 = ۄ𝒖 ‪𝛼𝒗,‬ۃ = ۄ𝒗𝛼 ‪𝒖,‬ۃ )‪(3‬‬

‫ۄ𝒘 ‪(−𝒗),‬ۃ ‪ +‬ۄ𝒘 ‪𝒖,‬ۃ = ۄ𝒘 ‪𝒖 + (−𝒗),‬ۃ = ۄ𝒘 ‪𝒖 − 𝒗,‬ۃ )‪(4‬‬

‫‪.‬ۄ𝒘 ‪𝒗,‬ۃ ‪ −‬ۄ𝒘 ‪𝒖,‬ۃ =‬

‫ۄ𝒖 ‪𝒘,‬ۃ ‪ −‬ۄ𝒖 ‪𝒗,‬ۃ = ۄ𝒖 ‪𝒗 − 𝒘,‬ۃ = ۄ𝒘 ‪𝒖, 𝒗 −‬ۃ )‪(5‬‬

‫■‪ .‬ۄ𝒘 ‪𝒖,‬ۃ ‪ −‬ۄ𝒗 ‪𝒖,‬ۃ =‬

‫‪- 169 -‬‬

 Exercise Set (5.1)

In Exercise 1-8 determine which of the given functions

are inner products on ℝ3 , where 𝒖 = (𝑎1 , 𝑎2 , 𝑎3 ) and 𝒗 =
(𝑏1 , 𝑏2 , 𝑏3 ).
(1) ‫𝒖ۃ‬, 𝒗‫𝑎 = ۄ‬1 𝑏2 + 𝑎2 𝑏1 .
(2) ‫𝒖ۃ‬, 𝒗‫ = ۄ‬2𝑎1 𝑏1 + 5𝑎2 𝑏2 + 𝑎3 𝑏3 − 𝑎2 𝑏3 − 𝑎3 𝑏2 .
(3) ‫𝒖ۃ‬, 𝒗‫𝑎 = ۄ‬1 𝑏1 + 𝑎2 𝑏2 + 𝑎3 𝑏3 + 𝑎1 𝑎2 𝑏3 .
(4) ‫𝒖ۃ‬, 𝒗‫𝑎 = ۄ‬1 𝑏1 + 𝑎2 𝑏2 .
(5) ‫𝒖ۃ‬, 𝒗‫𝑎 = ۄ‬1 + 𝑏1 .
(6) ‫𝒖ۃ‬, 𝒗‫𝑎 = ۄ‬12 𝑏12 + 𝑎22 𝑏22 + 𝑎32 𝑏32 .
(7) ‫𝒖ۃ‬, 𝒗‫𝑎 = ۄ‬12 𝑏1 + 𝑎2 𝑏2 + 𝑎3 𝑏3 .
(8) ‫𝒖ۃ‬, 𝒗‫𝑎 = ۄ‬1 𝑏1 − 𝑎2 𝑏2 + 𝑎3 𝑏3 .
(9) Let 𝒑, 𝒒 ∈ ℙ2 . Prove that

‫𝒑ۃ‬, 𝒒‫(𝒑 = ۄ‬−1)𝒒(−1) + 𝒑(0)𝒒(0) + 𝒑(1)𝒒(1) is an

inner product function on ℙ2 .

(10) Let 𝒑(𝑥 ) = 𝑎0 + 𝑎1 𝑥 + 𝑎2 𝑥 2 , 𝒒(𝑥 ) = 𝑏0 + 𝑏1 𝑥 +

𝑏2 𝑥 2 ∈ ℙ2 . Determine if ‫𝒑ۃ‬, 𝒒‫𝑎 = ۄ‬0 𝑏0 − 𝑎1 𝑏1 + 𝑎2 𝑏2 is
an inner product function on ℙ2 , or not.?

(11) Let 𝕍 be an inner product space and 𝒖, 𝒗, 𝒘 ∈ 𝕍.

Simplify the following:

- 170 -
‫ 𝒖ۃ‬+ 𝒗, 𝒗 + 𝒘‫ۄ‬, ‫ۃ‬2𝒗 − 𝒘, 3𝒖 + 2𝒘‫ۄ‬, ‫ 𝒖ۃ‬− 𝒗 + 4𝒘, 3𝒖 − 𝒗‫ۄ‬.

(12) Let 𝒇, 𝒈 ∈ 𝔻[𝑎, 𝑏] (The linear space of all

differential function on [𝑎, 𝑏]). Is ‫𝒇ۃ‬, 𝒈‫𝒇 = ۄ‬′ (𝑥 ) + 𝒈′ (𝑥)
is an inner product function on 𝔻[𝑎, 𝑏].

(13) Let 𝒇, 𝒈 ∈ ℂ[0,1]. Is ‫𝒇ۃ‬, 𝒈‫ ۄ‬is an inner product on

ℂ[0,1].

(14) Let 𝑝, 𝑞 ∈ ℙ3 . Is ‫𝒑ۃ‬, 𝒒‫ ۄ‬is an inner product function

on 𝕡3 .

(15) Let 𝑨, 𝑩 ∈ 𝕄𝑛×𝑛 . Is ‫𝑨ۃ‬, 𝑩‫ )𝑩𝑨( 𝑟𝑡 = ۄ‬is an inner

product function on 𝕄𝑛×𝑛 .

(16) Let 𝑨, 𝑩 ∈ 𝕄𝑛×𝑛 . Is ‫𝑨ۃ‬, 𝑩‫)𝑩 𝑇𝑨( 𝑟𝑡 = ۄ‬is an inner

product function on 𝕄𝑛×𝑛 .

(17) Let A be an 𝑛 × 𝑛 non-singular matrix and 𝒖, 𝒗 ∈

ℝ𝑛 . Prove that ‫𝒖ۃ‬, 𝒗‫ )𝒗𝐴( 𝑇)𝒖𝐴( = ۄ‬is an inner product
function on ℝ𝑛 .

(18) Let 𝑨 = [𝑎𝑎1 𝑎𝑎2 ] , 𝑩 = [𝑏𝑏1 𝑏𝑏2 ] ∈ 𝕄2×2 and

3 4 3 4

‫𝑨ۃ‬, 𝑩‫𝑎 = ۄ‬1 𝑏1 + 𝑎2 𝑏2 + 𝑎3 𝑏3 + 𝑎4 𝑏4 .

Prove that 𝕄2×2 is an inner product space.

- 171 -
2.Orthogonally
Now, we define the notion of an angle between two
vectors in an inner product space and use this concept to
obtain basic relations between vectors in an inner product
and fundamental geometric concepts in ℝ2 .
Definition 1.
Let u be an element of an inner product space 𝕍. The
norm ‖𝒖‖ of 𝕍 is defined by ‖𝒖‖ = √‫𝒖ۃ‬, 𝒖‫ۄ‬.
If w is another element in 𝕍, then we define the distance
between u and w denoted by 𝑑 (𝒖, 𝒘) , as follows:

𝑑(𝒖, 𝒘) = ‖𝒖 − 𝒘‖ = √‫ 𝒖ۃ‬− 𝒘, 𝒖 − 𝒘‫ۄ‬

Example 1.
Let 𝒖 = (𝑎1 , … , 𝑎𝑛 ) , 𝒘 = (𝑏1 , … , 𝑏𝑛 ) ∈ ℝ𝑛 and the
inner product defined on ℝ𝑛 be the Euclidean inner
product. Then ‖𝒖‖ = √𝑎12 + ⋯ + 𝑎𝑛2 .

Also, 𝑑(𝒖, 𝒘) = √(𝑎1 − 𝑏1 )2 + ⋯ + (𝑎𝑛 − 𝑏𝑛 )2 .

If 𝑛 = 3, 𝒖 = (1, 2, 3) and 𝒘 = (−1, −2 ,4), then

‖𝒖‖ = √(1)2 + (2)2 + (3)2 = √14 and

𝑑 (𝒖, 𝒘) = √(1 + 1)2 + (2 + 2)2 + (3 − 4)2 = √21. ◄

- 172 -
Example 2.
Let 𝒖 = (𝑎1 , 𝑎2 ) , 𝒗 = (𝑏1 , 𝑏2 ) ∈ ℝ2 and the inner
product function defined on ℝ2 is
‫𝒖ۃ‬, 𝒗‫ = ۄ‬7𝑎1 𝑏1 + 5𝑎2 𝑏2 .
Then,
‖𝒖‖ = √7𝑎12 + 5𝑎22 .
Also,
𝑑 (𝒖, 𝒘) = √7(𝑎1 − 𝑏1 )2 + 5(𝑎2 − 𝑏2 )2 . ◄
Example 3.

Let 𝒑(𝑥 ) = 2 − 𝑥 + 𝑥 2 , 𝒒(𝑥 ) = 1 + 2𝑥 − 𝑥 2 ∈ 𝕡2 and

the inner product be as defined in Example 3 in Section 1
of this chapter. Then

‖𝒑(𝑥 )‖ = √‫)𝑥(𝒑ۃ‬, 𝒑(𝑥 )‫ۄ‬

= √(2)2 + (−1)2 + (1)2

= √6.

𝑑(𝒑(𝑥 ), 𝒒(𝑥 )) = √‫ )𝑥(𝒑ۃ‬− 𝒒(𝑥 ), 𝒑(𝑥 ) − 𝒒(𝑥)‫ۄ‬

= √(1)2 + (−3)2 + (2)2

= √14. ◄

- 173 -
Example 4.

Let 𝒇(𝑥 ) = cos 𝑥 , 𝒈(𝑥 ) = sin 𝑥 ∈ ℂ[0, 𝜋] and the inner

product on ℂ[0, 𝜋] be as defined in Example 4 of Section
1 in this chapter. Then
1 1
𝜋 2 𝜋 2
𝜋
‖𝒇(𝑥)‖ = (∫[𝒇(𝑥)]2 𝑑𝑥 ) = (∫ cos2 𝑥 𝑑𝑥 ) = √ .
2
0 0

1
𝜋 2
𝑑(𝒇(𝑥 ), 𝒈(𝑥 )) = (∫0 (cos 𝑥 − sin 𝑥) 𝑑𝑥 ) = √𝜋.◄
2

Theorem 1. (Cauchy – Schwartz Inequality).

If u and v are any vectors of an inner product space,

then|‫𝒖ۃ‬, 𝒗‫‖𝒗‖‖𝒖‖ ≤ |ۄ‬.

Proof.

If 𝒖 = 𝝑, then the result holds. Let 𝒖 ≠ 𝝑 and

‫𝒗ۃ‬, 𝒖‫ۄ‬
𝒘=𝒗− 𝒖.
‖𝒖‖2

Therefore,

‫𝒗ۃ‬, 𝒖‫ۄ‬ ‫𝒗ۃ‬, 𝒖‫ۄ‬

0 ≤ ‖𝒘‖2 = ‫𝒘ۃ‬, 𝒘‫ 𝒗ۃ = ۄ‬− 𝒖, 𝒗 − 𝒖‫ۄ‬
‖𝒖‖2 ‖𝒖‖2

- 174 -
 ‫𝒗ۃ‬, 𝒖‫ۄ‬ ‫𝒗ۃ‬, 𝒖‫ۄ‬ ‫𝒗ۃ‬, 𝒖‫ۄ‬2
= ‫𝒗ۃ‬, 𝒗‫ ۄ‬− ‫𝒗ۃ‬, 𝒖‫ ۄ‬− ‫ۃ‬ 𝒖, 𝒗‫ ۄ‬+ ‫𝒖ۃ‬, 𝒖‫ۄ‬
‖𝒖‖2 ‖𝒖‖2 ‖𝒖‖4

‫𝒗ۃ‬, 𝒖‫ۄ‬2 ‫𝒗ۃ‬, 𝒖‫ۄ‬2 ‫𝒗ۃ‬, 𝒖‫ۄ‬2

= ‖𝒗 ‖2 − − +
‖𝒖‖2 ‖𝒖‖2 ‖𝒖‖2
‫𝒗ۃ‬,𝒖‫ۄ‬2
= ‖𝒗‖2 − ‖𝒖‖2
.

Then ‖𝒖‖2 ‖𝒗‖2 − ‫𝒖ۃ‬, 𝒗‫ۄ‬2 ≥ 0 or ‫𝒖ۃ‬, 𝒗‫ۄ‬2 ≤ ‖𝒖‖2 ‖𝒗‖2 .

Thus |‫𝒖ۃ‬, 𝒗‫‖𝒗‖‖𝒖‖ ≤ |ۄ‬. ■

Remarks.
(1) Applying the Cauchy-Schwarz inequality to ℝ𝑛 we
obtain the inequality:
(𝑎1 𝑏1 + ⋯ + 𝑎𝑛 𝑏𝑛 )2 ≤ (𝑎12 + ⋯ + 𝑎𝑛2 )(𝑏12 + ⋯ + 𝑏𝑛2 )
(2) Applying the Cauchy-Schwarz inequality to ℂ[𝑎, 𝑏]
we obtain the inequality:
𝑏 2
(∫ 𝑓 (𝑥 )𝑔(𝑥 )𝑑𝑥 )
𝑎
𝑏 𝑏
2 2
≤ (∫ (𝑓(𝑥 )) 𝑑𝑥 ) (∫ (𝑔(𝑥 )) 𝑑𝑥 )
𝑎 𝑎

- 175 -
Lemma 1.
|‫𝒖ۃ‬, 𝒗‫ ‖𝒗‖‖𝒖‖ = |ۄ‬if and only if u and v be linearly
dependent.
Proof.
‫𝒗ۃ‬,𝒖‫ۄ‬
Let |‫𝒖ۃ‬, 𝒗‫ ‖𝒗‖‖𝒖‖ = |ۄ‬and 𝒘 = 𝒗 − ‖𝒖‖2 𝒖. Then
‫𝒗ۃ‬,𝒖‫ۄ‬
‫𝒘ۃ‬, 𝒘‫ = ۄ‬0. Therefore 𝒘 = 𝝑. Hence 𝒗 = 𝒖.
‖𝒖‖2

So, u and v are linearly dependent.

Conversely, assume that u and v are linearly dependent.
Hence 𝒗 = 𝛼𝒖 , where 𝛼 ∈ ℝ. Then
‫𝒖ۃ‬, 𝒗‫ۄ‬2 = ‫𝒖ۃ‬, 𝛼𝒖‫ۄ‬2 = ‫𝒖ۃ‬, 𝛼𝒖‫𝒖ۃۄ‬, 𝛼𝒖‫ 𝛼 = ۄ‬2 ‫𝒖ۃ‬, 𝒖‫𝒖ۃۄ‬, 𝒖‫ۄ‬
= ‫𝒖ۃ‬, 𝒖‫𝒖𝛼ۃۄ‬, 𝛼𝒖‫𝒖ۃ = ۄ‬, 𝒖‫𝒗ۃۄ‬, 𝒗‫‖𝒖‖ = ۄ‬2 ‖𝒗‖2 .
Therefore |‫𝒖ۃ‬, 𝒗‫‖𝒗‖‖𝒖‖ = |ۄ‬.■
Theorem 1.
If 𝕍 is an inner product space, then for every u and v in 𝕍
and every scalar 𝛼 ∈ ℝ.
(1) ‖𝒖‖ ≥ 0 ;
(2) ‖𝒖‖ = 0 if and only if 𝒖 = 𝝑;
(3) ‖𝛼𝒖‖ = |𝛼 |‖𝒖‖;
(4) ‖𝒖 + 𝒗‖ ≤ ‖𝒖‖ + ‖𝒗‖ (Triangular inequality).

- 176 -
Proof.

(1) Since ‫𝒖ۃ‬, 𝒖‫ ≥ ۄ‬0 , then ‖𝒖‖ = √‫𝒖ۃ‬, 𝒖‫ ≥ ۄ‬0.

(2) Since ‫𝒖ۃ‬, 𝒖‫ = ۄ‬0 if and only if 𝒖 = 𝝑. Then ‖𝒖‖ = 0
if and only if 𝒖 = 𝝑.
(3) ‖𝛼𝒖‖2 = ‫𝒖𝛼ۃ‬, 𝛼𝒖‫ 𝛼 = ۄ‬2 ‫𝒖ۃ‬, 𝒖‫ 𝛼 = ۄ‬2 ‖𝒖‖2 .
Therefore ‖𝛼𝒖‖ = |𝛼 |‖𝒖‖.
(4) ‖𝒖 + 𝒗‖2 = ‫ 𝒖ۃ‬+ 𝒗, 𝒖 + 𝒗‫ۄ‬
= ‫𝒖ۃ‬, 𝒖‫ ۄ‬+ 2‫𝒖ۃ‬, 𝒗‫ ۄ‬+ ‫𝒗ۃ‬, 𝒗‫ۄ‬
≤ ‫𝒖ۃ‬, 𝒖‫ ۄ‬+ 2|‫𝒖ۃ‬, 𝒗‫ |ۄ‬+ ‫𝒗ۃ‬, 𝒗‫ۄ‬
≤ ‖𝒖‖2 + 2‖𝒖‖‖𝒗‖ + ‖𝒗‖2
= (‖𝒖‖ + ‖𝒗‖)2 .
Therefore
‖𝒖 + 𝒗‖ ≤ ‖𝒖‖ + ‖𝒗‖. ■
Lemma 2.
Let 𝕍 be an inner product space, 𝒖, 𝒗, 𝒘 ∈ 𝕍 and
𝛼 ∈ ℝ. Then
(1) 𝑑(𝒖, 𝒗) ≥ 0 ;
(2) 𝑑(𝒖, 𝒗) = 0 if and only if 𝒖 = 𝒗 ;
(3) 𝑑(𝒖, 𝒗) = 𝑑(𝒗, 𝒖)
(4) 𝑑(𝒖, 𝒗) ≤ 𝑑(𝒖, 𝒘) + 𝑑(𝒘, 𝒗) (Triangular inequality).

- 177 -
Proof.
(1) 𝑑(𝒖, 𝒗) = ‖𝒖 − 𝒗‖ ≥ 0.
(2) 𝑑(𝒖, 𝒗) = 0 ⇔ ‖𝒖 − 𝒗‖ = 0 ⟺ 𝒖 − 𝒗 = 𝝑 ⟺
𝒖 = 𝒗.
(3) 𝑑(𝒖, 𝒗) = ‖𝒖 − 𝒗‖ = ‖𝒗 − 𝒖‖ = 𝑑(𝒗, 𝒖).
(4) 𝑑(𝒖, 𝒗) = ‖𝒖 − 𝒗‖ = ‖𝒖 − 𝒘 + 𝒘 − 𝒗‖
≤ ‖𝒖 − 𝒘‖ + ‖𝒘 − 𝒗‖
= 𝑑 (𝒖, 𝒘) + 𝑑(𝒘, 𝒗).■
♠Now, we show how to use Cauchy-Schwarz inequality
in defining the angle between two vectors in an inner
product space. Let 𝕍 be an inner product space and 𝒖, 𝒗 ∈
𝕍. From Cauchy Schwarz inequality we have
|‫𝒖ۃ‬, 𝒗‫|ۄ‬
|‫𝒖ۃ‬, 𝒗‫⟺ ‖𝒗‖‖𝒖‖ ≤ |ۄ‬ ≤1⟺
‖𝒖‖‖𝒗‖
‫𝒖ۃ‬, 𝒗‫ۄ‬
−1 ≤ ≤1
‖𝒖‖‖𝒗‖
Therefore, there exists unique angle 𝜃 such that
‫𝒖ۃ‬, 𝒗‫ۄ‬
cos 𝜃 =
‖𝒖‖‖𝒗‖
where 0 ≤ 𝜃 ≤ 𝜋.

- 178 -
Definition 2.
Let 𝕍 be an inner product space and 𝒖 and 𝒗 be a nonzero
vector in 𝕍. Then
‫𝒖ۃ‬,𝒗‫ۄ‬
(1) The cosine of 𝜃 between u and v is: cos 𝜃 = ‖𝒖‖‖𝒗‖.

(2) The vectors u and v are orthogonal if ‫𝒖ۃ‬, 𝒗‫ = ۄ‬0.

Example 5.
Let ℝ2 be the Euclidean inner product space. Then the
angle between 𝒖 = (1, 2) and 𝒗 = (−1, 3) satisfies the
(1)(−1)+(2)(3) 1
equation cos 𝜃 = = . The vectors 𝒖 =
√1+4√1+9 √2

(1, 2) and 𝒗 = (−2, 1) are orthogonal as ‫𝒖ۃ‬, 𝒗‫= ۄ‬

(−2)(1) + (1)(2) = 0. But 𝒖 = (1, 2) and 𝒗 = (−1, 1)
are not orthogonal if the inner product on ℝ2 is the inner
product defined in Example 2, in Section 2 of Chapter
(IV), where ‫𝒖ۃ‬, 𝒗‫ = ۄ‬7(1)(−2) + 5(2)(1) ≠ 0.◄
Example 6.

Let 𝑨 = [20 −4
0 0 1
] , 𝑩 = [−7 0
] ∈ 𝕄2×2 and the inner product
is as defined in Exercise (5.1)-(18). Then A and B are
orthogonal as
‫𝑨ۃ‬, 𝑩‫( = ۄ‬2)(0) + (0)(1) + 0(−7) + (−4)(0) = 0.◄

- 179 -
Example 7.
The vectors 𝒇(𝑥 ) = sin 𝑥 , 𝒈(𝑥 ) = cos 𝑥 ∈ ℂ[0, 𝜋] is
orthogonal, where the inner product on ℂ[0, 𝜋] is defined
in Example 4 of Section 1 in this Chapter as
𝜋 1 𝜋
‫𝒇ۃ‬, 𝒈‫= ۄ‬ 2
∫0 sin 𝑥 cos 𝑥 𝑑𝑥 = [ sin 𝑥] = 0.◄
2 0
Example 8.
Let 𝒑(𝑥 ) = 2 − 𝑥 + 4𝑥 2 , 𝒒(𝑥 ) = 3 + 5𝑥 − 𝑥 2 ∈ ℙ2 ,
where the inner product is defined in Example 3 of
Section 1 in this Chapter. The angle 𝜃 between 𝒑(𝑥) and
𝒒(𝑥) satisfies the following equation:
6−5−4 −3
cos 𝜃 = = .◄
√4+1+16√9+25+1 √735

Theorem 2.
Let u and v be any vectors of an inner product space 𝕍.
Then u and v are orthogonal if and only if
(‖𝒖 + 𝒗‖2 = ‖𝒖‖2 + ‖𝒗‖2 ) (*)
Proof.
‖𝒖 + 𝒗‖2 = ‫ 𝒖ۃ‬+ 𝒗, 𝒖 + 𝒗‫𝒖ۃ = ۄ‬, 𝒖‫ ۄ‬+ 2‫𝒖ۃ‬, 𝒗‫ ۄ‬+ ‫𝒗ۃ‬, 𝒗‫ۄ‬
= ‖𝒖‖2 + 2‫𝒖ۃ‬, 𝒗‫ ۄ‬+ ‖𝒗‖2
Thus, the identity (*) holds if and only if ‫𝒖ۃ‬, 𝒗‫ = ۄ‬0, i.e.,
if and only if u and v are orthogonal. ■

- 180 -
 Exercise Set (5.2)
In Exercise (1)-(5) determine the angle between the given
vectors and determine whose vectors orthogonal with
respect to the Euclidean inner produce on ℝ3 .
(1) 𝒖 = (−1, 3, 2), 𝒗 = (4, 2, −1);
(2) 𝒖 = (−2, −2, −2), 𝒗 = (−1, 1, −1);
(3) 𝒖 = (−1, 1, 0), 𝒗 = (4, 0, 9);
(4) 𝒖 = (−1, 5, 2), 𝒗 = (2, 4, −9);
(5) 𝒖 = (8, −4, −2), 𝒗 = (−4, 2, 1);
(6) Do the Exercise (1)-(5) if the inner product on ℝ3 is:
‫𝒖ۃ‬, 𝒗‫ = ۄ‬7𝑎1 𝑏1 + 3𝑎2 𝑏2 + 4𝑎3 𝑏3 .
In Exercise (7) – (11) determine the angle between the
given elements and determine whose are orthogonal with
respect to the Euclidean inner product ℝ4 .
(7) 𝒖 = (−4, 0, −10, 1), 𝒗 = (2, 1, −2, 9);
(8) 𝒖 = (−1, 1, 0, 2), 𝒗 = (1, −1, 3, 0);
(9) 𝒖 = (0, −2, 2, 1), 𝒗 = (−1, −1, 1, 1);
(10) 𝒖 = (1, 2, 0, 1), 𝒗 = (−2, 3, 1, 2);
(11) 𝒖 = (−2, 3, 1, 4), 𝒗 = (−3, 1, 4, 6).
(12) Do the Exercise (7)-(11) if the inner product on ℝ4
is defined as
- 181 -
‫𝒖ۃ‬, 𝒗‫𝑎 = ۄ‬1 𝑏1 + 𝑎2 𝑏2 + 2𝑎3 𝑏3 + 3𝑎4 𝑏4 .
In Exercise (13)-(17): Determine the angle 𝜃 between the
given vectors and determine whose are orthogonal where
the inner product on 𝕡2 is defined in Example 3 of
Section 1 in this Chapter
(13) 𝒑(𝑥 ) = 𝑥 2 + 2𝑥 + 3, 𝒒(𝑥 ) = 𝑥 2 + 1
(14) 𝒑(𝑥 ) = 2𝑥 2 + 𝑥 − 3, 𝒒(𝑥 ) = 4𝑥 2 − 𝑥 ;
(15) 𝒑(𝑥 ) = 2𝑥 2 + 5𝑥 − 1 , 𝒒(𝑥 ) = 9𝑥 2 + 2𝑥 + 1 ;
(16) 𝒑(𝑥 ) = 2𝑥 2 − 𝑥 + 1, 𝒒(𝑥 ) = 𝑥 2 + 2𝑥 ;
(17) 𝒑(𝑥 ) = 2𝑥 2 − 𝑥 + 1, 𝒒(𝑥 ) = 𝑥 2 + 2𝑥 ;
In Exercise (18)-(22): Find the angle 𝜃 between the given
vectors and determine whose are orthogonal, where the
inner product on 𝕄2×2 is defined in Exercise (5.1)-(18).
(18) 𝑨 = [21 −3
6
], 𝑩 = [01 23] ;

(19) 𝑨 = [−1 2
6 1
], 𝑩 = [30
31
];

(20) 𝑨 = [13 24], 𝑩 = [51 −3

2
];

(21) 𝑨 = [52 −4
6
], 𝑩 = [−2 −1
1 −5
];

(22) 𝑨 = [10 00], 𝑩 = [03 −2

5
].

- 182 -
(23) Prove that sin 3𝑥 , cos 4𝑥 ∈ ℂ[0, 𝜋] are orthogonal,
where the inner product defined in Example 4 of Section
1 in this Chapter.
(24) Let 𝕍 be an inner product space and , 𝒗, 𝒘 ∈ 𝕍 . If u
is orthogonal to v and w, prove that u and 𝛼𝒗 + 𝛽𝒘 are
orthogonal for every 𝛼, 𝛽 ∈ ℝ .
(25) Let 𝕍 be an inner product space and 𝒖, 𝒗 ∈ 𝕍. Prove
(1) ‖𝒖 + 𝒗‖2 − ‖𝒖 − 𝒗‖2 = 4‫𝒖ۃ‬, 𝒗‫;ۄ‬
(2) ‖𝒖 + 𝒗‖2 + ‖𝒖 − 𝒗‖2 = 2‖𝒖‖2 + 2‖𝒗‖2 ;
(3) ‖𝒖 + 𝒗‖2 = ‖𝒖‖2 + ‖𝒗‖2 + 2‖𝒖‖‖𝒗‖ cos 𝜃.
Where 𝜃 is the angle between u and v.
(4) 𝒖 − 𝒗 and 𝒖 + 𝒗 are orthogonal when ‖𝒖‖ = ‖𝒗‖.
(5) ‖𝒖‖ − ‖𝒗‖ ≤ ‖𝒖 − 𝒗‖.

- 183 -
3. Orthonormal Basis
In many problems involving vector spaces, the problem
solver is free to choose any basis for the vector space that
seems appropriate. In inner product spaces the solution of
a problem is often greatly simplified by choosing a basis
in which the vectors are orthogonal to one another. In this
section we shall show how such bases can be obtained.
Definition 1.
A set of vectors in an inner product space is called an
orthogonal set if all pairs of distinct vectors in the set are
orthogonal. An orthogonal set in which each vector has
norm 1 is called orthonormal.
Example 1.
1 1
Let 𝒖1 = (1, 1, 0, 1) , 𝒖2 = (1, −2, 0, 1) and
√3 √6
1
𝒖3 = (−1,0,2,1) be three vectors in the Euclidean
√6

inner product space ℝ4 . It follows that the set of vectors

𝑆 = {𝒖1 , 𝒖2 , 𝒖3 } is orthonormal since ‫𝒖ۃ‬1 , 𝒖2 ‫= ۄ‬
‫𝒖ۃ‬2 , 𝒖3 ‫𝒖ۃ = ۄ‬1 , 𝒖3 ‫ = ۄ‬0 and also,‖𝒖1 ‖ = ‖𝒖2 ‖ =
‖𝒖3 ‖ = 1.◄

- 184 -
Example 2.

1 3 3 2 1
Let 𝕊 = { , √ 𝑥, √ (𝑥 2 − )} ⊆ ℂ[−1,1], where the
2
√ 2 2 5 3

inner product in ℂ[−1,1] is defined in Example 4 of

Section 1 in this chapter. Then 𝕊 is orthonormal set since
1
1 3 1 √3 √3 2
‫ۃ‬ , √ 𝑥‫= ۄ‬ ∫−1 2 𝑥 𝑑𝑥 = [ 𝑥 ] = 0;
√2 2 4 −1

1 3 2 1 1 3 1
‫ۃ‬ , √ (𝑥 2 − )‫∫ = ۄ‬−1 (𝑥 2 − 3) 𝑑𝑥 = 0;
√2 2 5 3 2 √ 5

√3 3 2 1 1 3√6 1
‫ۃ‬ 𝑥, √ (𝑥 2 − )‫∫ = ۄ‬−1 (𝑥 3 − 3 𝑥) 𝑑𝑥 = 0;
2 2 5 3 4 √ 5

Also, we can show that

1 3 3 2 1
‖ ‖ = ‖√ 𝑥‖ = ‖ √ (𝑥 2 − )‖ = 1 .◄
2 √ 2 2 5 3

Remark.
If 𝕍 is a nonzero vector in an inner product space, then
1 1 1 1
‖𝒗‖
𝒗 has norm 1, as ‖‖𝒗‖ 𝒗‖ = |‖𝒗‖| ‖𝒗‖ = ‖𝒗‖ ‖𝒗‖ = 1.

The process of multiplying nonzero vector v by the

reciprocal of its length to obtain a vector of norm 1 is
called normalizing the vector v. An orthogonal set of

- 185 -
nonzero vectors can always be converted to an
orthonormal set by normalizing each of its vectors.
Example 3.
Let 𝒖1 = (0,1,0), 𝒖2 = (1,0,1), 𝒖3 = (1,0, −1) and
assume that ℝ3 has the Euclidean inner product. It
follows that the set of vectors 𝕊 = {𝒖1 , 𝒖2 , 𝒖3 } is
orthogonal since ‫𝒖ۃ‬1 , 𝒖2 ‫𝒖ۃ = ۄ‬1 , 𝒖3 ‫𝒖ۃ = ۄ‬2 , 𝒖3 ‫ = ۄ‬0.
Also, the Euclidean norms of the vectors are
‖𝒖1 ‖ = 1, ‖𝒖2 ‖ = √2, ‖𝒖3 ‖ = √2
Consequently, normalizing 𝒖1 , 𝒖2 , and 𝒖3 yields
𝒖1
𝒗1 = = (0,1,0),
‖𝒖1 ‖
𝒖 1 1
𝒗2 = ‖𝒖2 ‖ = ( , 0, ),
2 √2 √2
𝒖 1 −1
𝒗3 = ‖𝒖3 ‖ = ( , 0, ).
3 √2 √2

We leave it for you to verify that the set 𝕊′ = {𝒗1 , 𝒗2 , 𝒗3 }

is orthonormal by showing that:
‫𝒗ۃ‬1 , 𝒗2 ‫𝒗ۃ = ۄ‬1 , 𝒗3 ‫𝒗ۃ = ۄ‬2 , 𝒗3 ‫ = ۄ‬0
and
‖𝒗1 ‖ = ‖𝒗2 ‖ = ‖𝒗3 ‖ = 1.◄

- 186 -
Theorem 1.
If 𝕊 = {𝒗1 , … , 𝒗𝑛 } is an orthogonal set of nonzero
vectors in an inner product space, then 𝕊 is linearly
independent.
Proof.
Assume that k1 𝒗1 + ⋯ + k 𝑛 𝒗𝑛 = 𝝑. For each 𝒗𝑖 in 𝕊,
‫ۃ‬k1 𝒗1 + ⋯ + k 𝑛 𝒗𝑛 , 𝒗𝑖 ‫𝝑ۃ = ۄ‬, 𝒗𝑖 ‫ = ۄ‬0. Or equivalently,
k1 ‫𝒗ۃ‬1 , 𝒗𝑖 ‫ ۄ‬+ ⋯ + k 𝑖 ‫ 𝑖𝒗ۃ‬, 𝒗𝑖 ‫ ۄ‬+ ⋯ + k 𝑛 ‫ 𝑛𝒗ۃ‬, 𝒗𝑖 ‫ = ۄ‬0.
From the orthogonality of 𝕊 it follows that ‫ 𝑗𝒗ۃ‬, 𝒗𝑖 ‫ = ۄ‬0 .
when 𝑗 ≠ 𝑖, so that this equation reduces to
k 𝑖 ‫ 𝑖𝒗ۃ‬, 𝒗𝑖 ‫ = ۄ‬0
Since the vectors in 𝕊 are assumed to be nonzero,
‫ 𝑖𝒗ۃ‬, 𝒗𝑖 ‫ ≠ ۄ‬0 by the positivity axiom for inner product.
Therefore, k 𝑖 = 0. Since the subscript i is arbitrary, then
k1 = ⋯ = k 𝑛 = 0. Thus, 𝕊 is linearly independent. ■
● In an inner product space, a basis consisting of
orthonormal vectors is called an orthonormal basis, and
a basis consisting of orthogonal vectors is called
orthogonal basis.

- 187 -
Theorem 2.
If 𝕊 = {𝒗1 , … , 𝒗𝑛 } is an orthonormal basis for an inner
product space 𝕍, and u is any vector in 𝕍, then
𝒖 = ‫𝒖ۃ‬, 𝒗1 ‫𝒗ۄ‬1 + ‫𝒖ۃ‬, 𝒗2 ‫𝒗ۄ‬2 + ⋯ + ‫𝒖ۃ‬, 𝒗𝑛 ‫ 𝑛𝒗ۄ‬.
Proof.
Since 𝕊 = {𝒗1 , … , 𝒗𝑛 } is a basis, a vector u can be
expressed in the form 𝒖 = k1 𝒗1 + ⋯ + k 𝑛 𝒗𝑛 .
We complete the proof by showing that k 𝑖 = ‫𝒖ۃ‬, 𝒗𝑖 ‫ ۄ‬for
𝑖 = 1, … , 𝑛 . For each vector 𝒗𝑖 in 𝕊 we have
‫𝒖ۃ‬, 𝒗𝑖 ‫ۃ = ۄ‬k1 𝒗1 + ⋯ + k 𝑛 𝒗𝑛 , 𝒗𝑖 ‫ۄ‬
= k1 ‫𝒗ۃ‬1 , 𝒗𝑖 ‫ ۄ‬+ ⋯ + k 𝑖 ‫ 𝑖𝒗ۃ‬, 𝒗𝑖 ‫ ۄ‬+ ⋯ + k 𝑛 ‫ 𝑛𝒗ۃ‬, 𝒗𝑖 ‫ۄ‬.
Since 𝕊 = {𝒗1 , … , 𝒗𝑛 } is an orthonormal set, we have
‫ 𝑖𝒗ۃ‬, 𝒗𝑖 ‫‖ 𝑖𝒗‖ = ۄ‬2 = 1 and ‫ 𝑗𝒗ۃ‬, 𝒗𝑖 ‫ = ۄ‬0 if 𝑗 ≠ 𝑖.
Therefore, the above expression for ‫𝒖ۃ‬, 𝒗𝑖 ‫ ۄ‬simplifies to
‫𝒖ۃ‬, 𝒗𝑖 ‫ = ۄ‬k 𝑖
The scalars ‫𝒖ۃ‬, 𝒗1 ‫ۄ‬, ‫𝒖ۃ‬, 𝒗2 ‫ۄ‬, … , ‫𝒖ۃ‬, 𝒗𝑛 ‫ ۄ‬in the above
theorem are the coordinates of u relative to the
orthonormal basis 𝕊 = {𝒗1 , … , 𝒗𝑛 } and
(𝒖)𝕊 = (‫𝒖ۃ‬, 𝒗1 ‫ۄ‬, ‫𝒖ۃ‬, 𝒗2 ‫ۄ‬, … , ‫𝒖ۃ‬, 𝒗𝑛 ‫)ۄ‬
is the coordinate vector of u relative to this basis.■

- 188 -
Example 4.
4 3 3 4
Let 𝒗1 = (0,1,0), 𝒗2 = (− , 0, ) , 𝒗3 = ( , 0, ). It is
5 5 5 5

easy to check that 𝕊 = {𝒗1 , 𝒗2 , 𝒗3 } is an orthonormal

basis for ℝ3 with the Euclidean inner product. Express
𝒖 = (1, 1, 1) as a linear combination of the vectors in 𝕊
and find the coordinate (𝒖)𝕊 .
Solution.
1 7
‫𝒖ۃ‬, 𝒗1 ‫ = ۄ‬1, ‫𝒖ۃ‬, 𝒗2 ‫ = ۄ‬− and ‫𝒖ۃ‬, 𝒗3 ‫ = ۄ‬.
5 5

Therefore, by Theorem 2 in Section 3 of this chapter we

1 7
have 𝒖 = 𝒗1 − 𝒗2 + 𝒗3 . That is,
5 5
1 4 3 7 3 4
(1,1,1) = (0,1,0) − (− , 0, ) + ( , 0, ).
5 5 5 5 5 5

The coordinate vector of u relative to 𝕊 is

1 7
(𝒖)𝕊 = (‫𝒖ۃ‬, 𝒗1 ‫ۄ‬, ‫𝒖ۃ‬, 𝒗2 ‫ۄ‬, ‫𝒖ۃ‬, 𝒗3 ‫( = )ۄ‬1, − , ).◄
5 5

●We have seen that orthonormal bases enjoy variety

useful properties. Our next theorem shows that every
nonzero finite dimensional vector space has an
orthonormal basis. The proof of this result provides an
algorithm (called Gram-Schmidt Process) for converting
an arbitrary basis into orthonormal basis.

- 189 -
Theorem 3 (Gram-Schmidt algorithm).
Every nonzero finite-dimensional inner product space has
an orthonormal basis.
Proof.
Let 𝕍 be any nonzero finite-dimensional inner product
space and let {𝒖1 , … , 𝒖𝑛 } be any basis for 𝕍. It suffices to
show that 𝕍 has an orthogonal basis, since the vectors in
the orthogonal basis can by normalized to produce an
orthonormal basis for 𝕍. The following sequence of steps
will produce an orthogonal basis {𝒗1 , … , 𝒗𝑛 } for 𝕍. Let
Step 1. 𝒗1 = 𝒖1 .
‫𝒖ۃ‬2 ,𝒗1 ‫ۄ‬
Step 2. 𝒗2 = 𝑢2 − ‖𝒗1 ‖2
𝒗1 .
‫𝒖ۃ‬3 ,𝒗1 ‫ۄ‬ ‫𝒖ۃ‬3 ,𝒗2 ‫ۄ‬
Step 3. 𝒗3 = 𝒖3 − ‖𝒗1 ‖2
𝒗1 − ‖𝒗2 ‖2
𝒗2 .

⋮
‫ 𝑛𝒖ۃ‬, 𝒗𝑛−1 ‫ۄ‬ ‫ 𝑛𝒖ۃ‬, 𝒗1 ‫ۄ‬
𝒗 𝑛 = 𝒖𝑛 − 𝒗 − ⋯ − 𝒗
‖𝒗𝑛−1 ‖2 𝑛−1 ‖𝒗1 ‖2 1
Since {𝒖1 , … , 𝒖𝑛 } is linearly independence, then
{𝒗1 , … , 𝒗𝑛 } is nonzero.
Now, we prove by induction that {𝒗1 , … , 𝒗𝑛 } is
orthogonal.

- 190 -
Let 𝑛 = 2 , then
‫𝒖ۃ‬2 , 𝒗1 ‫ۄ‬
‫𝒗ۃ‬2 , 𝒗1 ‫𝒖ۃ = ۄ‬2 − 𝒗 ,𝒗 ‫ۄ‬
‖𝒗1 ‖2 1 1
‫𝒖ۃ‬2 ,𝒗1 ‫ۄ‬
= ‫𝒖ۃ‬2 , 𝒗1 ‫ ۄ‬− ‖𝒗1 ‖2
‫𝒗ۃ‬1 , 𝒗1 ‫ۄ‬

= ‫𝒖ۃ‬2 , 𝒗1 ‫ ۄ‬− ‫𝒖ۃ‬2 , 𝒗1 ‫ = ۄ‬0.

Therefore 𝒗1 , 𝒗2 are orthogonal.
Now, let {𝒗1 , 𝒗2 , … , 𝒗𝑛−1 } is orthogonal, where 𝑛 > 2 .
We prove that {𝒗1 , 𝒗2 , … , 𝒗𝑛 } is orthogonal.
For 𝑖 = 1,2, … , 𝑛 − 1, we have
‫ 𝑛𝒖ۃ‬, 𝒗𝑛−1 ‫ۄ‬
‫ 𝑛 𝒗 ۃ‬, 𝒗 𝑖 ‫ 𝑛𝒖 ۃ = ۄ‬, 𝒗 𝑖 ‫ ۄ‬− ‫𝑛𝒗ۃ‬−1 , 𝒗𝑖 ‫ ۄ‬−
‖𝒗𝑛−1 ‖2
‫ 𝑛𝒖ۃ‬, 𝒗1 ‫ۄ‬
…− ‫𝒗ۃ‬1 , 𝒗𝑖 ‫ۄ‬
‖𝒗1 ‖2
‫ 𝑛𝒖 ۃ‬, 𝒗 𝑖 ‫ۄ‬
= ‫ 𝑛 𝒖ۃ‬, 𝒗𝑖 ‫ ۄ‬− ‫ 𝑖𝒗 ۃ‬, 𝒗𝑖 ‫ = ۄ‬0
‖𝒗𝑖 ‖2
Therefore {𝒗1 , … , 𝒗𝑛 } is orthogonal. From Theorem 1 in
Section 3 of Chapter (IV) {𝒗1 , … , 𝒗𝑛 } is linearly
independent. Hence {𝒗1 , … , 𝒗𝑛 } is an orthogonal basis
𝒗
for the inner product space 𝕍. Finally, if 𝒘𝑖 = ‖𝒗𝑖 ‖ for
𝑖

𝑖 = 1, … , 𝑛, we obtain an orthonormal basis {𝒘1 , … , 𝒘𝑛 }

for the inner product space 𝕍.■
- 191 -
Example 5.
Consider the vector space ℝ3 with the Euclidean inner
product. Apply the Gram- Schmidt process to transform
the basis vectors 𝒖1 = (1,1,1) , 𝒖2 = (0,1,1) , 𝒖3 =
(0,0,1) into an orthogonal basis {𝒗1 , 𝒗2 , 𝒗3 } ; then
normalize the orthogonal basis vectors to obtain an
orthonormal basis {𝒘1 , 𝒘2 , 𝒘3 } .
Solution.
Step 1. 𝒗1 = 𝒖1 = (1,1,1).
‫𝒖ۃ‬2 ,𝒗1 ‫ۄ‬
Step 2. 𝒗2 = 𝒖2 − ‖𝒗1 ‖2
𝒗1 .
2 2 1 1
= (0,1,1) − (1,1,1) = (− , , ).
3 3 3 3
‫𝒖ۃ‬3 ,𝒗1 ‫ۄ‬ ‫𝒖ۃ‬3 ,𝒗2 ‫ۄ‬
Step 3. 𝒗3 = 𝒖3 − ‖𝑣1 ‖2
𝒗1 − ‖𝒗2 ‖2
𝒗2 .
1 1⁄3 2 1 1
= (0,0,1) − (1,1,1) − (− , , )
3 2⁄3 3 3 3
1 1
= (0, − , ).
2 2
2 1 1 1 1
Thus 𝒗1 = (1,1,1) , 𝒗2 = (− , , ) , 𝒗3 = (0, − , )
3 3 3 2 2

form an orthogonal basis for ℝ3 . The norms of these

√6 1
vectors are ‖𝒗1 ‖ = √3, ‖𝒗2 ‖ = , ‖𝒗3 ‖ = . So, an
3 √2

orthonormal basis for ℝ3 is

- 192 -
 𝒗 1 1 1 𝒗 2 1 1
𝒘1 = ‖𝒗1 ‖ = ( , , ) , 𝒘2 = ‖𝒗2 ‖ = (− , , )
1 √3 √3 √ 3 2 √6 √6 √6
𝒗 −1 1
𝒘3 = ‖𝒗3 ‖ = (0, , ).◄
3 √2 √2

Example 6.
Consider the subspace 𝕎 of ℝ5 spanned by {𝒗1 , 𝒗2 , 𝒗3 } ,
where 𝒗1 = (−1, −1,1,0,0), 𝒗2 = (0, −1,0,0,1) , 𝒗3 =
(1, −1,0,1,0). Find an orthonormal basis for 𝕎, where
the inner product is the Euclidean product.
Solution.
Step 1: 𝒖1 = 𝒗1 = (−1, −1,1,0,0).
‫𝒗ۃ‬2 ,𝒖1 ‫ۄ‬
Step 2: 𝒖2 = 𝒗2 − ‖𝒖1 ‖2
𝒖1
1
= (0, −1,0,0,1) − (−1, −1,1,0,0)
3
1 −2 1
=( , , − , 0,1).
3 3 3
‫𝒗ۃ‬3 ,𝒖1 ‫ۄ‬ ‫𝒗ۃ‬3 ,𝒖2 ‫ۄ‬ 4 −3 1 3
Step 3: 𝒖3 = 𝒗3 − ‖𝒖1 ‖2
𝒖1 − ‖𝒖2 ‖2
𝒖2 = ( , , , 1, )
5 5 5 5

Thus, {𝒖1 , 𝒖2 , 𝒖3 } is orthogonal basis. Therefore, the

orthonormal basis is
1 1 1 1 −2 1 √15
(− ,− , , 0,0) , ( , ,− , 0, ),
√3 √3 √3 √15 √15 √15 5

2 √15 √15 √15 √15

( ,− , , , − ). ◄
√15 10 30 6 10

- 193 -
Example 7.
Let 𝒑, 𝒒 ∈ ℙ4 and the inner product defined on ℙ4 is as
‫𝒑ۃ‬, 𝒒‫(𝒑 = ۄ‬−2)𝒒(−2) + 𝒑(−1)𝒒(−1) + 𝒑(0)𝒒(0) +
𝒑(1)𝒒(1) + 𝒑(2)𝒒(2) and 𝕎 be a subspace of ℙ4
spanned by {𝒑1 (𝑥 ) = 1, 𝒑2 (𝑥 ) = 𝑥, 𝒑3 (𝑥) = 𝑥 2 }.
Find an orthogonal basis for 𝕎.
Solution.
Step 1.
𝒒1 = 𝒑1 = 1.
Step 2.
‫𝒑ۃ‬2 , 𝒒1 ‫𝒑 = ۄ‬2 (−2)𝒒1 (−2) + 𝒑2 (−1)𝒒1 (−1) +
𝒑2 (0)𝒒1 (0) + 𝒑2 (1)𝒒1 (1) + 𝒑2 (2)𝒒1 (2) = 0.
‫𝒑ۃ‬2 ,𝒒1 ‫ۄ‬
Hence, 𝒒2 = 𝒑2 − ‖𝒒1 ‖2
𝒒1 = 𝑥 − 0 = 𝑥

Step 3.
‫𝒑ۃ‬3 ,𝒒2 ‫ۄ‬ ‫𝒑ۃ‬3 ,𝒒1 ‫ۄ‬
𝒒3 = 𝒑3 − ‖𝒒2 ‖2
𝒒2 − ‖𝒒1 ‖2
𝒒1 , where

‫𝒑ۃ‬3 , 𝒒1 ‫ = ۄ‬10, ‖𝒒1 ‖2 = ‫𝒒ۃ‬1 , 𝒒1 ‫ = ۄ‬5, and ‫𝒑ۃ‬3 , 𝒒2 ‫ = ۄ‬0.

10
Therefore, 𝒒3 = 𝑥 3 − 0 − (1) = 𝑥 2 − 2.
5

Hence, {1, 𝑥, 𝑥 2 − 2} is orthogonal basis for 𝕎.◄

- 194 -
 Exercise Set (5.3)
(1) Let the vector space 𝕡2 have the inner product
1
‫𝒑ۃ‬, 𝒒‫∫ = ۄ‬−1 𝒑(𝑥 )𝒒(𝑥 )𝑑𝑥
Apply the Gram-Schmidt process to transform the
standard basis 𝕊 = {1, 𝑥, 𝑥 2 } into orthonormal basis.
(2) Let ℝ3 be the inner product space, where the inner
function is ‫𝒖ۃ‬, 𝒗‫𝑎 = ۄ‬1 𝑏1 + 2𝑎2 𝑏2 + 3𝑎3 𝑏3 .
(3) Apply the Gram-Schmidt process to transform the
basis {𝒗1 , 𝒗𝟐 , 𝒗3 } to orthonormal basis.
(𝑎)𝒗1 = (1,1,1) , 𝒗2 = (−1,1,0) , 𝒗3 = (1,2,1);
(𝑏)𝒗1 = (1,1,1) , 𝒗2 = (1,1,0) , 𝒗3 = (1,0,0).
(4) Let A be an 𝑛 × 𝑚 matrix. The subspace of ℝ𝑛
spanned by the rows (columns) of A is called the row
(column) space. Apply the Gram-Schmidt process to find
an orthonormal basis for the column space of the matrix
A, where the inner product is the Euclidean
3 0 5 1 0 0
(𝑎) 𝐴 = [0 1 −2] , (𝑏) 𝐴 = [1 1 0]
4 1 1 1 1 1
(5) Find an orthonormal basis for the row space for the
matrix A in Exercise (4).

- 195 -
4. EigenValues and EigenVectors
Eigenvalues and eigenvectors arise, for example, when
studying vibrational problems, where the eigenvalues
represent fundamental frequencies of vibration and the
eigenvectors characterize the corresponding fundamental
modes of vibration.
They also occur in many other ways; in mechanics, for
example, the eigenvalues can represent the principal
stresses in a solid body, in which case the eigenvectors
then describe the corresponding principal axes of stress
caused by the body being subjected to external forces.
Also in mechanics, the moment of inertia of a solid body
about lines through its center of gravity can be
represented by an ellipsoid, with the length of a line
drawn from its center to the surface of the ellipsoid
proportional to the moment of inertia of the body
about an axis through the center of gravity of the body
drawn parallel to the line. In this case the eigenvalues
represent the principal moments of inertia of the body
about the principal axes of inertia, that are then
determined by the eigenvectors.

- 196 -
In matrix algebra you come across equations of the form:

The equation can be rearranged as follows:

Rearrange: 𝐴𝒙 − 𝜆𝒙 = 𝝑
Factorize: (𝐴 − 𝜆)𝒙 = 𝝑
Make 𝜆 into a matrix:(𝐴 − 𝜆𝐼)𝒙 = 𝝑
This form is useful for finding eigenvalues and
eigenvectors.
We see from the last version of the equation that 𝒙 = 𝝑 is
a solution (trivial).
To find the eigenvalues (𝜆), matrix theory tells us that we
must set the determinant of 𝐴 − 𝜆𝐼 equal to 0.
i.e., det (𝐴 − 𝜆𝐼) = 0
To find the eigenvectors, put the eigenvalues back into
the original equation and solve.
The characteristic polynomial for the matrix A is:
𝑝𝑛 (𝜆) = det (𝐴 − 𝜆𝐼) .
- 197 -
More precisely, if 𝐴 is an 𝑛 × 𝑛 matrix, the polynomial
𝑝𝑛 (𝜆) of degree 𝑛 in the scalar 𝜆 defined as
𝑝𝑛 (𝜆) = det (𝐴 − 𝜆𝐼)
is called the characteristic polynomial of 𝐴.
The roots of the equation
𝑝𝑛 (𝜆) = 0
are called the eigenvalues of matrix 𝐴, and the column
vectors 𝒙1 , 𝒙2 , . . . , 𝒙𝑛 satisfying the matrix equation
(𝐴 − 𝜆𝑖 𝐼)𝒙𝑖 = 𝝑 (∗)
are called the eigenvectors of matrix 𝐴.
i.e., the eigenvectors 𝒙𝑖 of the 𝑛 × 𝑛 matrix 𝐴,
corresponding to the eigenvalue 𝜆 = 𝜆𝑖 , is a solution of
the homogeneous equation (∗).
In general, a matrix with complex coefficients will have
complex eigenvalues, though even when the coefficients
of A are all real it is still possible for complex eigenvalues
to arise. This is because then the characteristic equation
will have real coefficients, so if complex roots occur, they
must do so in complex conjugate pairs.

- 198 -
If an eigenvalue 𝜆∗ is repeated r times, corresponding to
the presence of a factor (𝜆 − 𝜆∗ )𝒓 in the characteristic
polynomial 𝑝𝑛 (𝜆), the number r is called the algebraic
multiplicity of the eigenvalue 𝜆∗ . The set of all
eigenvalues 𝜆1 , 𝜆2 , . . . , 𝜆𝑛 of A is called the spectrum of
A, and the number 𝑅 = 𝐦𝐚𝐱 {|𝜆1 |, |𝜆2 |, . . . , |𝜆𝑛 |}, equal
to the largest of the moduli of the eigenvalues, is called
the spectral radius of A.
It can happen that an eigenvalue with algebraic
multiplicity 𝑟 > 1 only has s different eigenvectors
associated with it, where 𝑠 < 𝑟, and when this occurs
the number s is called the geometric multiplicity of the
eigenvalue. The set of all eigenvectors associated with an
eigenvalue with geometric multiplicity s together with the
null vector 𝝑 forms what is called the eigenspace
associated with the eigenvalue. When one or more
eigenvalues has a geometric multiplicity that is less than
its algebraic multiplicity, it follows directly that the
vector space associated with 𝐴 must have dimension less
than n.

- 199 -
Example 1.
Find the eigenvalues and eigenvectors for the matrix
6 3
𝐴= [ ].
−2 1
Solution.
Finding the characteristic polynomial
𝜆 0
We know that 𝜆𝐼 = [ ]. Thus
0 𝜆
6 3 𝜆 0 6−𝜆 3
𝐴 − 𝜆𝐼 = [ ]−[ ]=[ ]
−2 1 0 𝜆 −2 1−𝜆
Therefore
𝑝2 (𝜆) = |𝐴 − 𝜆𝐼| = (6 − 𝜆)(1 − 𝜆) − 3(−2)
= 6 − 7𝜆 + 𝜆2 + 6 = 𝜆2 − 7𝜆 + 12.

Finding the eigenvalues:

Put the characteristic polynomial equal to 0 and solve to
find that 𝜆 = 3 𝑜𝑟 𝜆 = 4.

Finding the eigenvectors:

Going back to the very 1st equation we know that:
𝐴𝒙 = 𝜆𝒙
6 3 𝑥 𝑥
This means that: [ ] [𝑦 ] = 𝜆 [𝑦 ]
−2 1
Multiply out each side:

- 200 -
 6𝑥 + 3𝑦 = 𝜆𝑥, −2𝑥 + 𝑦 = 𝜆𝑦
(The 2 eigenvalues we found were λ = 3 and λ = 4)
When 𝝀 = 𝟑: 6𝑥 + 3𝑦 = 3𝑥; −2𝑥 + 𝑦 = 3𝑦
Rearranging either of these gives −𝑥 = 𝑦.
So, setting 𝑥 = 𝑘, where k is an arbitrary real number (a
parameter) shows that the eigenvector 𝒙1 corresponding
𝑘 1
to the eigenvalue λ = 3 is given by [ ] = 𝑘 [ ]. As 𝑘
−𝑘 −1
is an arbitrary parameter, for convenience we set 𝑘 = 1
1
and as a result obtain [ ] is the eigenvector associated
−1
to the eigenvalue 𝜆 = 3.
When 𝝀 = 𝟒: 6𝑥 + 3𝑦 = 4𝑥; −2𝑥 + 𝑦 = 4𝑦
Rearranging either of these gives −2𝒙 = 𝟑𝒚.
So, setting 𝑥 = 𝑘, where k is an arbitrary real number (a
parameter) shows that the eigenvector 𝒙2 corresponding
𝑘 1
to the eigenvalue λ = 4 is given by [− 𝑘 ] = 𝑘 [− 2]. As
2
3 3

𝑘 is an arbitrary parameter, for convenience we set 𝑘 = 3

3
and as a result obtain [ ] is the eigenvector associated
−2
to the eigenvalue 𝜆 = 4.◄

- 201 -
Example 2.
Find the characteristic polynomial, eigenvalues and
1 2
eigenvectors for the matrix 𝐴 = [ ].
4 3
Solution.
𝜆 0 1 2 𝜆−1 −2
𝜆𝐼 − 𝐴 = [ ]−[ ]=[ ].
𝜆 0 4 3 −4 𝜆−3
The characteristic polynomial is
𝜆−1 −2
𝑝2 (𝜆) = | | = 𝜆2 − 4𝜆 − 5.
−4 𝜆−3
And the eigenvalues are 𝜆 = 5 and 𝜆 = −1.
1 2 𝑥 𝑥
The eigenvector for 𝜆 = 5 ⟹ [ ] [𝑦 ] = 5 [𝑦 ]
4 3
yields the equations 𝑥 + 2𝑦 = 5𝑥 and 4𝑥 + 3𝑦 = 𝑦.
Solving, we find that 𝑦 = 2𝑥; therefore, the eigenvector
1
is [ ]. Using the same method, we find the eigenvector
2
1
for 𝜆 = −1 is [ ].◄
−1

- 202 -
Example 3.
Find the characteristic polynomial, the eigenvalues, and
the eigenvectors of the matrix
2 1 −1
𝐴 = [3 2 −3].
3 1 −2
Solution.
The characteristic polynomial 𝑝3 (𝜆) is given by
2−𝜆 1 −1
𝑝3 (𝜆) = |𝐴 − 𝜆𝐼| = | 3 2−𝜆 −3 |
3 1 −2 − 𝜆
and after expanding the determinant we find that
𝑝3 (𝜆) = −𝜆3 + 2𝜆2 + 𝜆 − 2.
The characteristic equation 𝑝3 (𝜆) = 0 is
𝜆3 − 2𝜆2 − 𝜆 + 2 = 0
and inspection shows it has the roots 2, 1, and −1. So, the
eigenvalues of A are 𝜆1 = 2, 𝜆2 = 1, and 𝜆3 = −1.
To find the eigenvectors 𝒙𝑖 of A corresponding to the
eigenvalues 𝜆 = 𝜆𝑖 for 𝑖 = 1, 2, 3, it will be necessary to
solve the homogeneous system of algebraic equations
(𝐴 − 𝜆𝑖 𝐼)𝒙𝑖 = 𝝑 for 𝑖 = 1, 2, 3,where 𝒙𝑖 =
[𝑥1 , 𝑥2 , 𝑥3 ]𝑇 .

- 203 -
Case 𝝀𝟏 = 𝟐
The system of equations to be solved to find the
eigenvector 𝒙1 is
2−2 1 −1 𝑥1 0
[ 3 2−2 −3 ] [𝑥2 ] = [0]
3 1 −2 − 2 𝑥3 0
and this matrix equation is equivalent to the set of three
linear algebraic equations
𝑥2 − 𝑥3 = 0, 3𝑥1 − 3 𝑥3 = 0, 3𝑥1 + 𝑥2 − 4 𝑥3 = 0.
Solving this homogenous system of equations, we find
that 𝑥1 = 𝑥2 = 𝑥3 , so setting 𝑥3 = 𝑠, where s is an
arbitrary real number (a parameter) shows that the
eigenvector 𝒙1 corresponding to the eigenvalue 𝜆1 = 2
𝑠 1
is given by 𝒙1 = [𝑠] = 𝑠 [1].
𝑠 1
As s is an arbitrary parameter, for convenience we set
1
𝑠 = 1 and as a result obtain the eigenvector 𝒙1 = [1].
1
Case 𝝀𝟐 = 𝟏
The system of equations to be solved to find the
eigenvector 𝒙2 is

- 204 -
 2−1 1 −1 𝑥1 0
[ 3 2−1 −3 ] [𝑥2 ] = [0]
3 1 −2 − 1 𝑥3 0
and this matrix equation is equivalent to the set of three
linear algebraic equations
𝑥1 + 𝑥2 − 𝑥3 = 0,
3𝑥1 + 𝑥2 − 3 𝑥3 = 0,
3𝑥1 + 𝑥2 − 3 𝑥3 = 0.
Solving this homogenous system of equations, we find
that 𝑥2 = 0, 𝑥1 = 𝑥3 , so setting 𝑥3 = 𝑡, where 𝑡 is an
arbitrary real number (a parameter) shows that the
eigenvector 𝒙2 corresponding to the eigenvalue 𝜆2 = 1
𝑡 1
is given by 𝒙2 = [0] = 𝑡 [0].
𝑡 1
As 𝑡 is an arbitrary parameter, for convenience we set
1
𝑡 = 1 and as a result obtain the eigenvector 𝒙2 = [0].
1
Case 𝝀𝟑 = −𝟏
Setting, and proceeding as before, shows that the
elements of the eigenvector must satisfy the equations
3𝑥1 + 𝑥2 − 𝑥3 = 0,

- 205 -
 3𝑥1 + 3𝑥2 − 3 𝑥3 = 0,
3𝑥1 + 𝑥2 − 𝑥3 = 0.
Solving this homogenous system of equations, we find
that 𝑥1 = 0, 𝑥2 = 𝑥3 , so setting 𝑥3 = 𝑘, where 𝑘 is an
arbitrary real number (a parameter) shows that the
eigenvector 𝒙3 associated to the eigenvalue 𝜆3 = − 1 is
0 0
given by 𝒙3 = [𝑘] = 𝑘 [1].
𝑘 1
As 𝑘 is an arbitrary parameter, for convenience we set
0
𝑘 = 1 and as a result obtain the eigenvector 𝒙3 = [1].
1
These three eigenvectors form a basis for the three-
dimensional vector space associated with A.
The spectrum of A is the set of numbers −1, 1, 2, and the
spectral radius of A is seen to be 𝑅 = 2. ◄

- 206 -
Example 4.
Determine if the given vectors u and v are eigenvectors of
A? If yes, find the eigenvalue of A associated to the
eigenvector.
4 −1 6 −1 −3
𝐴 = [2 1 6] , 𝒖 = [ 2 ] , 𝒗 = [ 0 ]
2 −1 8 1 1
Solution.
4 −1 6 −3 −6 −3
𝐴𝒗 = [2 1 6] [ 0 ] = [ 0 ] = 2 [ 0 ] = 2𝒗.
2 −1 8 1 2 1
Hence, 𝐴𝒗 = 2𝒗 and thus v is an eigenvector of A with
corresponding eigenvalue 𝜆 = 2. On the other hand,
4 −1 6 −1 0
𝐴𝒖 = [2 1 6] [ 2 ] = [6].
2 −1 8 1 4
There is no scalar λ such that
0 −1
[6] = 𝜆 [ 2 ]
4 1
Therefore, u is not an eigenvector of A.◄

- 207 -
Theorem 1.
If 𝜆1 , … 𝜆𝑚 are mutually distinct eigenvalues of A with
eigenvectors 𝒙1 , … 𝒙𝑚 , then {𝒙1 , … 𝒙𝑚 } is a linearly
independent set.
Corollary 1.
If an 𝑛 × 𝑛 matrix A has n distinct eigen values, then A
has n linearly independent eigenvectors.
Theorem 2.
Let A be an 𝑛 × 𝑛 matrix. Then A is invertible if and
only if the number 0 is not an eigenvalue of A.
Example 6
6 8 4
𝐴 = [−2 −2 −2]
−1 −2 1
What are the eigenvalues and eigenvectors?
Solution. Outlines
1) Find the characteristic polynomial.
When it is zero the solution to it are the only possible
eigenvalues:
𝜆−6 −8 −4
|𝜆𝐼 − 𝐴| = | 2 𝜆−2 2 | = (𝜆 − 1)2 (𝜆 − 2)1
1 2 𝜆−1

- 208 -
has eigenvalues 𝜆 = 1, 1, 2.
Solve: 𝐴𝒙 = 1𝒙 and 𝐴𝒙 = 2𝒙.
𝐴𝒙 = 1𝒙 has eigenspace of dimension 1.
𝐴𝒙 = 2𝒙 has eigenspace of dimension 2. ◄
Example 6.
Find the characteristic polynomial, eigenvalues and
eigenvectors of the matrix A, where
5 4 2
𝐴 = [4 5 2].
2 2 2
Solution.
The characteristic polynomial is
𝜆−5 −4 −2
𝑝3 (𝜆) = 𝑑𝑒𝑡( 𝜆𝐼 − 𝐴) = | −4 𝜆−5 −2 |
−2 −2 𝜆−2
= (𝜆 − 1)2 (𝜆 − 10) = 0.⇒ 𝜆 = 1, 1, and 10.
As: 𝜆 = 1,
−4 −4 −2 𝑥1
(1 ⋅ 𝐼 − 𝐴)𝒙 = [−4 −4 −2] [𝑥2 ] = 0.
−2 −2 −1 𝑥3
⇔ 𝑥1 = −𝑠 − 𝑡, 𝑥2 = 𝑠, 𝑥3 = 2𝑡.
𝑥1 −𝑠 − 𝑡 −1 −1
𝑥
⇔ 𝒙 = [ 2 ] = [ 𝑠 ] = 𝑠 [ 1 ] + 𝑡 [ 0 ] , 𝑠, 𝑡 ∈ 𝑅.
𝑥3 2𝑡 0 2

- 209 -
Thus,
−1 −1
𝒙 = 𝑠 [ 1 ] + 𝑡 [ 0 ] , 𝑠, 𝑡 ∈ 𝑅, 𝑠 ≠ 0 or 𝑡 ≠ 0,
0 2
are the eigenvectors associated with eigenvalue 𝜆 = 1.
As: 𝜆 = 10,
5 −4 −2 𝑥1
(10 ⋅ 𝐼 − 𝐴)𝒙 = [−4 5 −2] [𝑥2 ] = 0.
−2 −2 8 𝑥3
⇔ 𝑥1 = 2𝑟, 𝑥2 = 2𝑟, 𝑥3 = 𝑟
𝑥1 2𝑟 2
⇔ 𝒙 = [𝑥2 ] = [2𝑟] = 𝑟 [2] , 𝑟 ∈ 𝑅.
𝑥3 𝑟 1
2
Thus 𝑟 [2] , 𝑟 ∈ 𝑅, 𝑟 ≠ 0, are the eigenvectors associated
1
with eigenvalue 𝜆 = 10.◄
Example 7.
Find the eigenvectors of the matrix A, where
0 1 2
𝐴 = [2 3 0].
0 4 5
Solution.
𝜆 −1 −2
𝑝3 (𝜆) = 𝑑𝑒𝑡( 𝜆𝐼 − 𝐴) = |−2 𝜆−3 0 |
0 −4 𝜆−5
- 210 -
 = (𝜆 − 1)2 (𝜆 − 6) = 0
⇒ 𝜆 = 1, 1, and 6.
−1 1 2 𝑥1
1. As 𝜆 = 1,(𝐴 − 1 ⋅ 𝐼)𝒙 = [ 2 2 0] [𝑥2 ] = 0.
0 4 4 𝑥3
𝑥1 1 1
𝒙 = [𝑥2 ] = 𝑡 [−1] , 𝑡 ∈ 𝑅. Thus 𝑡 [−1] , 𝑡 ∈ 𝑅, 𝑡 ≠ 0
𝑥3 1 1
are the eigenvectors associated with eigenvalue 𝜆 = 1.
−6 1 2 𝑥1
2. As 𝜆 = 6, (𝐴 − 6 ⋅ 𝐼)𝒙 = [ 2 −3 0 ] [𝑥2 ] = 0.
0 4 −1 𝑥3
𝑥1 3 3
𝒙 = [𝑥2 ] = 𝑟 [2] , 𝑟 ∈ 𝑅 or 𝑟 [2] , 𝑟 ∈ 𝑅, 𝑟 ≠ 0.
𝑥3 8 8
are the eigenvectors associated with eigenvalue 𝜆 = 6.◄

- 211 -
 Exercise Set (5.4)
Find the eigenvalues of the matrices in problems 1 - 9.
2 7
1. 𝐴 = ( )
7 2
3 −2
2. 𝐴 = ( )
1 −1
5 3
3. 𝐴 = ( )
−4 4
0 0 0
4. 𝐴 = (0 2 5 )
0 0 −1
4 0 0
5. 𝐴 = (0 0 0 )
1 0 −3
4 0 0
6. 𝐴 = (0 0 0 )
1 0 −3
4 −7 0 2
0 3 −4 6
7. 𝐴 = ( )
0 0 3 −8
0 0 0 1
5 0 0 0
8 −4 0 0
8. 𝐴 = ( )
0 7 1 0
1 −5 2 1
3 0 0 0 0
−5 1 0 0 0
9. 𝐴 = 3 8 0 0 0 .
0 −7 2 1 0
(−4 1 9 −2 3)

- 212 -
- 213 -
 Chapter VI

partial fractions & Sequences

1. Partial Fractions
Let 𝑁(𝑥) and 𝐷(𝑥) be two polynomials. Then a rational
function of x is any function of the form 𝑁(𝑥)/𝐷(𝑥). The
method of partial fractions involves the decomposition
of rational functions into an equivalent sum of simpler
terms of the type

𝑃1 𝑃2 𝑄1 𝑥+𝑅1 𝑄 𝑥+𝑅2
, (𝑎𝑥+𝑏)2
, … and , (𝐴𝑥 22 , …,
𝑎𝑥+𝑏 𝐴𝑥 2 +𝐵𝑥+𝐶 +𝐵𝑥+𝐶)2

where the coefficients are all real together with, possibly,

a polynomial in x. The steps in the reduction of a rational
function to its partial fraction representation are as
follows:
STEP 1 Factorize 𝐷(𝑥) into a product of linear factors
and quadratic factors with real coefficients with complex
roots, called irreducible factors. This is the hardest
step, and real quadratic factors will only arise when
𝐷(𝑥 ) = 0 has pairs of complex conjugate roots. Use the
result to express 𝐷(𝑥) in the form
- 214 -
𝐷(𝑥 ) = (𝑎1 𝑥 + 𝑏1 )𝑟1 … (𝑎𝑚 𝑥 + 𝑏𝑚 )𝑟𝑚 (𝐴1 𝑥 2 + 𝐵1 𝑥 +
𝐶1 )𝑠1 … (𝐴𝑘 𝑥 2 + 𝐵𝑘 𝑥 + 𝐶𝑘 )𝑠𝑘 ,
where 𝑟𝑖 is the number of times the linear factor
(𝑎𝑖 𝑥 + 𝑏𝑖 ) occurs in the factorization of 𝐷(𝑥), called its
multiplicity, and 𝑠𝑖 is the corresponding multiplicity of
the quadratic factor (𝐴𝑗 𝑥 2 + 𝐵𝑗 𝑥 + 𝐶𝑗 ).
STEP 2 Suppose first that the degree n of the numerator
is less than the degree d of the denominator. Then, to
every different linear factor (𝑎𝑥 + 𝑏) with multiplicity
r, include in the partial fraction expansion the terms
𝑃1 𝑃2 𝑃𝑟
+ + ⋯ + ,
(𝑎𝑥 + 𝑏) (𝑎𝑥 + 𝑏)2 (𝑎𝑥 + 𝑏)𝑟
where the constant coefficients 𝑃𝑖 are unknown at this
stage, and so are called undetermined coefficients.
STEP 3 To every quadratic factor (𝐴𝑥 2 + 𝐵𝑥 + 𝐶 )𝑠 with
multiplicity s include in the partial fraction expansion the
terms
𝑄1 𝑥+𝑅1 𝑄 𝑥+𝑅2 𝑄 𝑥+𝑅𝑠
+ (𝐴𝑥 22 + ⋯ + (𝐴𝑥 2𝑠 ,
𝐴𝑥 2 +𝐵𝑥+𝐶 +𝐵𝑥+𝐶)2 +𝐵𝑥+𝐶)𝑠

where the 𝑄𝑗 and 𝑅𝑗 for 𝑗 = 1, 2, . . . , 𝑠 are undetermined

coefficients.

- 215 -
STEP 4 Take as the partial fraction representation of
𝑁(𝑥)/𝐷(𝑥) the sum of all the terms in Steps 2 and 3.
STEP 5 Multiply the expression
𝑁(𝑥)/𝐷(𝑥) = Partial fraction representation in Step 4
by 𝐷(𝑥), and determine the unknown coefficients by
equating the coefficients of corresponding powers of x on
either side of this expression to make it an identity (that
is, true for all x).
STEP 6 Substitute the values of the coefficients
determined in Step 5 into the expression in Step 4 to
obtain the required partial fraction representation.
STEP 7 If 𝑛 ≥ 𝑑, use long division to divide the
denominator into the numerator to obtain the sum of a
polynomial of degree 𝑛 − 𝑑 of the form
𝑇0 + 𝑇1 𝑥 + 𝑇2 𝑥 + ⋯ + 𝑇𝑛−𝑑 𝑥 𝑛−𝑑 ,
together with a remainder term in the form of a rational
function 𝑅(𝑥) of the type just considered. Find the partial
fraction representation of the rational function 𝑅(𝑥) using
Steps 1 to 6. The required partial fraction representation
is then the sum of the polynomial found by long division
and the partial fraction representation of 𝑅(𝑥).

- 216 -
Case I. Distinct Roots (Factors):
Example 1.
𝑥+1
Find the partial fraction decomposition of .
𝑥(𝑥−1)(𝑥+2)

Solution.
All terms in the denominator are linear factors, so by Step
1 the appropriate form of partial fraction representation is
𝑥+1 𝐴 𝐵 𝐶
= + + .
𝑥(𝑥−1)(𝑥+2) 𝑥 𝑥−1 𝑥+2

Cross multiplying, we obtain

𝑥 + 1 = 𝐴(𝑥 − 1)(𝑥 + 2) + 𝐵𝑥 (𝑥 + 2) + 𝐶𝑥 (𝑥 − 1)
Setting 𝑥 = 0 makes the terms in B and C vanish and
gives 𝐴 = −1/2. Setting 𝑥 = 1 makes the terms in A
and C vanish and gives 𝐵 = 2/3, whereas setting
𝑥 = −2 makes the terms in A and B vanish and gives
𝐶 = − 1/6.
So, the partial fraction decomposition is
𝑥+1 −1/2 2/3 −1/6
= + +
𝑥 (𝑥 − 1)(𝑥 + 2) 𝑥 𝑥−1 𝑥+2
Note.
If you solve this problem using the book’s method, you
will need to solve a system of 3 equations.

- 217 -
Exercise 1.
𝑥 2 +2𝑥+5
Find the partial fraction decomposition of .
𝑥 3 +3𝑥 2 −𝑥−3

Case II Repeated Roots (Factors):

Example 2.
𝑥+1
Find the partial fraction decomposition of .
𝑥(𝑥−1)3

Solution.
𝑥+1 𝐴 𝐵 𝐶 𝐷
= + (𝑥−1)3 + (𝑥−1)2 + (𝑥−1)
𝑥(𝑥−1)3 𝑥

When we multiply by 𝑥 (𝑥 − 1)3 , this becomes

𝑥 + 1 = 𝐴(𝑥 − 1)3 + 𝐵𝑥 + 𝐶𝑥 (𝑥 − 1) + 𝐷𝑥 (𝑥 − 1)2
Setting 𝑥 = 0 makes the terms in B, C and D vanish and
gives 𝐴 = −1. Setting 𝑥 = 1 makes the terms in A, C
and D vanish and gives 𝐵 = 2,
To find C and D, we have to solve a system of equations.
Since we know A and B, we obtain
𝑥 + 1 = −(𝑥 − 1)3 + 2𝑥 + 𝐶𝑥(𝑥 − 1) + 𝐷𝑥(𝑥 − 1)2 (*)
If we substitute any number other than the ones already
used (say 𝑥 = 2) in (*), we obtain
𝐶+𝐷 = 0

- 218 -
If we substitute any other number (say 𝑥 = −1) in (*), we
obtain
𝐶 − 2𝐷 = −3
If we solve the system, we get 𝐷 = 1 and 𝐶 = −1
So, the partial fraction decomposition is
𝑥+1 −1 2 −1 1
= + + +
𝑥 (𝑥 − 1)3 𝑥 (𝑥 − 1)3 (𝑥 − 1)2 (𝑥 − 1)
Note:
If you solve this problem using the book’s method, you
will need to solve a system of 4 equations.

- 219 -
Case III Quadratic Roots:
Example 3.
𝑥+1
Find the partial fraction decomposition of .
𝑥(𝑥 2 −2𝑥−2)

Solution.
𝑥+1 𝐴 𝐵𝑥+𝐶
= +
𝑥(𝑥 2 −2𝑥−2) 𝑥 𝑥 2 −2𝑥−2

When we multiply by 𝑥 (𝑥 2 − 2𝑥 − 2), this becomes

𝑥 + 1 = 𝐴(𝑥 2 − 2𝑥 − 2) + 𝑥 (𝐵𝑥 + 𝐶 )
1
At 𝑥 = 0, we get 𝐴 = − .
2

Therefore,
1
𝑥 + 1 = − (𝑥 2 − 2𝑥 − 2) + 𝑥 (𝐵𝑥 + 𝐶 )
2
At this point, you can substitute any number for x.
If we substitute 𝑥 = 2, we obtain 2𝐵 + 𝐶 = 1.
1
If we substitute 𝑥 = −1, we obtain 𝐵 − 𝐶 = .
2
1
If we solve the system, we get 𝐵 = and 𝐶 = 0.
2

So, the partial fraction decomposition is

1 1
𝑥+1 −2 (2)𝑥
= + .◄
𝑥(𝑥 2 −2𝑥−2) 𝑥 𝑥 2 −2𝑥−2

- 220 -
 Exercise Set (6.1)
Express the rational functions in Exercises 1 through 7 in
terms of partial fractions.
2𝑥
(1) ;
4𝑥 2 +12𝑥+9
−10𝑥 2 +27𝑥−14
(2) (𝑥−1)3 (𝑥+2)
;
2𝑥 4 +4𝑥 3 −2𝑥 2 +𝑥+7
(3) ;
𝑥 3 +2𝑥 2 −𝑥−2
𝑥
(4) (𝑥 2 ;
−1)(𝑥+4)

𝑥 3 +1
(5) (𝑥−1)(𝑥−2);
1
(6) ;
2𝑥 3 +3𝑥 2 −2𝑥
1
(7) .
𝑥(𝑥 2 +1)

- 221 -
2. SEQUENCES
The "sequence" in mathematics is used to describe an
unending succession of numbers. Some possibilities are
1 ,2 ,3 ,4 ,…
2 ,4 ,6 ,8 ,…
1 1 1
1, , , , …
2 3 4

1 , −1, 1, −1, …
In each case, the dots are used to suggest that the
sequence continues indefinitely, following the obvious
pattern. The numbers in a sequence are called the terms
of the sequence. The terms may be described according to
the positions they occupy. Thus, a sequence has a first
term, a second term, a third term, and so forth. Because
a sequence continues indefinitely, there is no last term.
The most common way to specify a sequence is to give a
formula for the terms. To illustrate the idea, we have listed
in the table below the terms in the sequence 2, 4, 6, 8, …
together with their term numbers:
Term number: 1 2 3 4 …
Term 2 4 6 8 …

- 222 -
There is a clear relationship here: each term is twice its
term number. Thus, for each positive integer 𝑛 the nth term
in the sequence 2, 4, 6, 8, … is given by the formula 2𝑛.
This is denoted by writing
2 , 4 , 6 , 8 , … ,2𝑛 , …
or more compactly in bracket notation as
{2𝑛}+∞
𝑛=1

From the bracket notation, the terms in the sequence can

be generated by successively substituting the integer
values 𝑛 = 1, 2, 3, … into the formula 2𝑛.
Example 1.
List the first five terms of the sequence{2𝑛 }+∞
𝑛=1

Solution.
Substituting 𝑛 = 1, 2, 3, 4, 5 into the formula 2𝑛 yields
21 , 22 , 23 , 24 , 25 or, equivalently, 2, 4, 8, 16, 32.◄
Example 2.
Express the following sequences in bracket notation
1 2 3 4 1 1 1 1
(a) , , , ,… (b) , , , ,…
2 3 4 5 2 4 8 16
1 2 3 4
(c) 1, −1, 1, −1, … (d) ,− , ,− ,…
2 3 4 5

(e) 1, 3, 5, 7, …

- 223 -
Solution.
(a) Begin by comparing terms and term numbers:
Term number: 1 2 3 4 …
1 2 3 4
Term: …
2 3 4 5

In each term, the numerator is the same as the term

number, and the denominator is one greater than the term
𝑛
number. Thus, the nth term is and the sequence may
𝑛+1

be written
𝑛 +∞
{ }
𝑛 + 1 𝑛=1
(a) Observe that the sequence can be rewritten as
1 1 1 1
, , , ,…
2 22 23 24
and construct a table comparing terms and term numbers:
Term number: 1 2 3 4…
1 1 1 1
Term: …
2 22 23 24
1
From the table we see that the nth term is so the
2𝑛

sequence can be written as

1 +∞
{ 𝑛}
2 𝑛=1

- 224 -
(c) Observe first that (−1)𝑟 is either 1 or −1 according to
whether r is an even integer or an odd integer. In the
sequence 1, −1, 1, −1, … the odd-numbered terms are 1's
and the even-numbered terms are −1's. Thus a formula for
the nth term can be obtained by raising −1 to a power that
will be even when n is odd and odd when n is even. This
is accomplished by the formula (−1)𝑛+1 , so that the
sequence can be written as {(−1)𝑛+1 }+∞
𝑛=1

(d) Combining the results in parts (a) and (c), we may write
this sequence as
𝑛 +∞
{(−1 )𝑛+1 }
𝑛 + 1 𝑛=1
(e) Begin by comparing terms and term numbers:
Term number: 1 2 3 4 …
Term : 1 3 5 7…
From the table we see that each term is one less than twice
the term number. Thus, the nth term is 2𝑛 − 1 and the
sequence can be written as
{2 𝑛 − 1}+∞
𝑛=1

.◄

- 225 -
Frequently we shall want to write down a sequence without
specifying the numerical values of the terms. We do this
by writing
𝑎1 , 𝑎2 , … , 𝑎𝑛 , …
Or in bracket notation
{𝑎𝑛 }+∞
𝑛=1

Or sometimes simply {𝑎𝑛 }

The time has come to formally define the term "sequence".
When we write a sequence such as
2, 4, 6, 8, … , 2𝑛 , …
In bracket notation
{2𝑛}+∞
𝑛=1 (*)
We could rewrite (*) in functional notation as
𝑓(𝑛) = 2𝑛 , 𝑛 = 1, 2, 3, …
From this point of view, the notation
2, 4, 6, 8, … , 2𝑛, …
represents a listing of the function values
𝑓(1), 𝑓(2), 𝑓(3), … , 𝑓(𝑛), …
This suggests the following definition.

- 226 -
Definition 1.
A sequence (or infinite sequence) is a function whose
domain is the set of positive integers.
Because sequences are functions, we may inquire about
the graph of sequence. For example, the graph of the
𝑛+1 +∞
sequence { }
2𝑛2 𝑛=1

is the graph of the sequence

𝑛+1
𝑓(𝑛) = , 𝑛 = 1, 2, 3, …
2𝑛2

Because the right side of this equation is defined only for

positive integers values of n, the graph consists of a
succession of isolated points (Figure 1a). This is in
marked distinction to the graph of
𝑥+1
𝑓(𝑥) = ,𝑥 ≥ 1
2𝑥 2
which is a continuous curve (Figure 1 b).

- 227 -
In Figure 2 we have sketched the graphs of 3 sequences:
𝑛−1 +∞ 𝑛−1 +∞
(a) { } ; (b) {(−1)𝑛+1 ( )} ; (c) {3}+∞
𝑛=1 .
𝑛 𝑛=1 𝑛 𝑛=1

To say that a sequence{𝑎𝑛 }+∞

𝑛=1 approaches a limit L as n

gets large is intended to mean that eventually the terms in

the sequence become arbitrary close to the number L.
Thus, if we choose any positive number 𝜀, the terms in the
sequence will eventually be within 𝜀 units of L.
Geometrically, this means that if we sketch the lines 𝑦 =

- 228 -
𝐿 + 𝜀 and 𝑦 = 𝐿 − 𝜀 , the terms in the sequence will
eventually be trapped within the band between these lines,
and thus be within 𝜀 units of L (Figure 3).

The following definition expresses this idea precisely.

Definition 2.
A sequence {𝑎𝑛 }+∞
𝑛=1 is said to have the limit L if given any

𝜀 > 0 there is a positive integer N such that |𝑎𝑛 − 𝐿| < 𝜀

when 𝑛 ≥ 𝑁. If a sequence {𝑎𝑛 }+∞
𝑛=1 has a limit L, we say

that the sequence converges to L and write

lim 𝑎𝑛 = 𝐿
𝑛→+∞

A sequence that does not have a finite limit is said to

diverge.

- 229 -
Definition 3.
A sequence {𝑎𝑛 }+∞
𝑛=1 is said to be convergent or

divergent if lim 𝑎𝑛 is finite or not finite, respectively.

𝑛→+∞

If a sequence {𝑎𝑛 }+∞

𝑛=1 neither converges to a finite

number nor diverges to + ∞ or – ∞, it is called an

Oscillatory sequence.
Oscillatory sequences are of 2 types:
(i) A bounded sequence which does not converge, is said
to oscillate finitely.
(ii) An unbounded sequence which does not diverge, is
said to oscillate infinitely.
Example 3.
1 +∞
(1) Consider the sequence { 𝑛 } .
2 𝑛=1
1 1
Here 𝑎𝑛 = and lim 𝑎𝑛 = lim = 0 which is
2𝑛 𝑛→+∞ 𝑛→+∞ 2𝑛

1 +∞
finite. Therefore, the sequence { 𝑛 } is convergent.
2 𝑛=1

(2) Consider the sequences {𝑛2 }+∞ 𝑛 +∞

𝑛=1 and {−2 }𝑛=1 .

Here 𝑎𝑛 = 𝑛2 or 𝑎𝑛 = −2𝑛 .
lim 𝑎𝑛 = +∞ or lim 𝑎𝑛 = −∞.
𝑛→+∞ 𝑛→+∞

⇒ Both these sequences are divergent.

- 230 -
(3) Consider the sequence {(−1)𝑛 }+∞
𝑛=1 .

Here 𝑎𝑛 = (−1)𝑛 .
It is a bounded sequence because there exist two real
numbers k and K (𝑘 ≤ 𝐾) such that 𝑘 ≤ 𝑎𝑛 ≤ 𝐾 for
every 𝑛 ∈ ℕ.
{𝑎𝑛 } = {– 1, 1, – 1, 1, – 1, . . . . . . } ⇒ – 1 ≤ 𝑎𝑛 ≤ 1.
Now, lim 𝑎2𝑛 = lim (−1)2𝑛 = 1 and
𝑛→+∞ 𝑛→+∞

lim 𝑎2𝑛+1 = lim (−1)2𝑛+1 = −1.

𝑛→+∞ 𝑛→+∞

Thus, lim 𝑎𝑛 does not exist ⇒ The sequence does not

𝑛→+∞

converge. Hence this sequence oscillates finitely.

Note:
When we say lim 𝑎𝑛 = 𝑙 , it means lim 𝑎2𝑛 = 𝑙 and
𝑛→+∞ 𝑛→+∞

lim 𝑎2𝑛+1 = 𝑙.
𝑛→+∞

Example 3.
+∞
𝑛−1
Figure 2 suggests that the sequence {(−1)𝑛+1 { 𝑛 }}
𝑛=1
𝑛−1 +∞
oscillates finitely, while the sequences { } and
𝑛 𝑛=1

{3}+∞
𝑛=1 converge to 1 and 3, respectively; that is,
𝑛−1
lim = 1 and lim 3 = 3.◄
𝑛→+∞ 𝑛 𝑛→+∞

- 231 -
Theorem 3.
Suppose that the sequences {𝑎𝑛 } and {𝑏𝑛 } converge to the
limits 𝐿1 and 𝐿2 , respectively, and c is a constant. Then
(a) lim 𝑐 = 𝑐
𝑛→+∞
(b) lim 𝑐𝑎𝑛 = 𝑐 lim 𝑎𝑛 = 𝑐𝐿1
𝑛→+∞ 𝑛→+∞
(c) lim (𝑎𝑛 + 𝑏𝑛 ) = lim 𝑎𝑛 + lim 𝑏𝑛 = 𝐿1 + 𝐿2
𝑛→+∞ 𝑛→+∞ 𝑛→+∞
(d) lim (𝑎𝑛 − 𝑏𝑛 ) = lim 𝑎𝑛 − lim 𝑏𝑛 = 𝐿1 − 𝐿2
𝑛→+∞ 𝑛→+∞ 𝑛→+∞
(e) lim (𝑎𝑛 ∙ 𝑏𝑛 ) = lim 𝑎𝑛 ∙ lim 𝑏𝑛 = 𝐿1 𝐿2
𝑛→+∞ 𝑛→+∞ 𝑛→+∞
𝑎 lim 𝑎𝑛 𝐿1
(f) lim ( 𝑛 ) = 𝑛→+∞
= (if 𝐿2 ≠ 0).
𝑏
𝑛→+∞ 𝑛 lim 𝑏 𝐿2
𝑛→+∞ 𝑛
Example 4.
In each part, determine if the given sequence converges or
diverges. If it converges, find the limit.
𝑛 +∞
(a) { } ;
2𝑛+1 𝑛=1

𝑛 +∞
(b) {(−1)𝑛+1 } ;
2 𝑛+1 𝑛=1

1 +∞
(c) {(−1)𝑛+1 } ;
𝑛 𝑛=1

(d) {8 − 2 𝑛}+∞
𝑛=1 ;

𝑛 +∞
(e) { 𝑛} .
𝑒 𝑛=1

- 232 -
Solution.
(a) Dividing numerator and denominator by n yields
𝑛 1 1
lim = lim 1 = .
𝑛→+∞ 2𝑛+1 𝑛→+∞ (2+ ) 2
𝑛

𝑛 +∞ 1
Thus { } converges to .
2 𝑛+1 𝑛=1 2
𝑛 1
(b) From part (a), lim = .
𝑛→+∞ 2𝑛+1 2

Thus, since(−1)𝑛+1 oscillates between + 1 and −1, then

𝑛
the product (−1)𝑛+1 oscillates between positive and
2𝑛+1

negative values, with the odd-numbered terms

1
approaching and the even-numbered terms approaching
2

1 𝑛 +∞
− . Therefore, the sequence {(−1)𝑛+1 }
2 2 𝑛+1 𝑛=1

approaches no limit. It oscillates finitely.

1 1
(c) Since lim = 0, the product(−1)𝑛+1 ( )
𝑛→+∞ 𝑛 𝑛

oscillates between positive and negative values, with

the odd-numbered terms approaching 0 through
positive values and the even-numbered terms
approaching 0 through negative values.
1
Thus lim (−1)𝑛+1 = 0.
𝑛→+∞ 𝑛

- 233 -
So, the sequence converges to 0.
(d) We have lim (8 − 2𝑛) = −∞.
𝑛→+∞

So, {8 − 2𝑛}+∞
𝑛=1 diverges.
𝑛
(e) We want to find lim which is an indeterminate
𝑛→+∞ 𝑒 𝑛

from type ∞/∞ . Unfortunately, we cannot apply

L'Hopital's rule directly since 𝑒 𝑛 and n are not
differentiable (n assumes only integer values).
However, we can apply L'Hopital's rule to the related
𝑥 𝑥 1
problem lim 𝑥 to obtain lim = lim =0
𝑥→+∞ 𝑒 𝑥→+∞ 𝑒 𝑥 𝑥→+∞ 𝑒 𝑥
𝑛
We conclude from this that lim = 0 since the values
𝑛→+∞ 𝑒 𝑛
𝑛 𝑥
of and are the same when x is a positive integer.
𝑒𝑛 𝑒𝑥

𝑛 +∞
Therefore, { 𝑛 } converges to 0. ◄
𝑒 𝑛=1

Example 5.
𝑛
Show that lim √𝑛 = 1.
𝑛→+∞

Solution.
1
By L'Hopital's rule, lim ln 𝑛 = 0.
𝑛→+∞ 𝑛

So,

- 234 -
 1
𝑛
lim √𝑛 = lim 𝑛𝑛
𝑛→+∞ 𝑛→+∞
1
(𝑛) ln 𝑛
= lim 𝑒
𝑛→+∞

= 𝑒 0 = 1.◄

- 235 -
 Exercise Set (6.2)
Show the first five terms of the sequence, determine if the
sequence converges, and if so, find the limit. (When
writing out the terms of the sequence, you need not find
numerical values: leave the forms in the first form you
obtain).
+∞ +∞
𝑛 𝑛2
(1) { } ; (2) { } ; (3) {2}+∞
𝑛=1 ;
𝑛+2 𝑛=1 2 𝑛+1 𝑛=1

1 +∞ ln 𝑛 +∞ 𝜋 +∞
(4) {ln ( )} ; (5) { } ; (6) {𝑛 sin } ;
𝑛 𝑛=1 𝑛 𝑛=1 𝑛 𝑛=1
+∞
(−1)𝑛+1
(7) {1 + (−1)𝑛 }+∞
𝑛=1 ; (8) { } ;
𝑛2 𝑛=1
+∞
𝑛 2 𝑛3 𝑛 +∞
(9) {(−1) 3 } ; (10) { 𝑛 } ;
𝑛 +1 𝑛=1 2 𝑛=1
+∞
(𝑛+1)(𝑛+2) +∞ 𝜋𝑛 +∞ 1
(11) { } ; (12) { 𝑛 } ; (13) {𝑛 }𝑛 ;
2 𝑛2 𝑛=1 4 𝑛=1 𝑛=1

In Exercise (14)-(18), express the sequence in the

notation{𝑎𝑛 }+∞
𝑛=1 , determine if the sequence converges, and

if so, find its limit.

1 3 5 7 1 2 3
(14) , , , ,… (15) 0 , , , ,…
2 4 6 8 22 32 42
1 1 1 1
(16) , , , , … (17) −1 , 2 , −3 ,4 , −5 , …
3 9 27 81
1 1 1 1 1 1 1
(18) (1 − ) , ( − ) , ( − ) , ( − ),…
2 2 3 3 4 4 5

- 236 -
(19) (a) Let {𝑎𝑛 } be a sequence for which 𝑎1 = 3
and 𝑎𝑛 = 2 𝑎𝑛−1 when 𝑛 ≥ 2. Find the first eight terms.
(b) Let {𝑎𝑛 } be a sequence for which 𝑎1 = 1,𝑎2 = 1
and 𝑎𝑛 = 𝑎𝑛−1 + 𝑎𝑛−2 when 𝑛 ≥ 3. Find the first eight
terms.
(20) The nth term 𝑎𝑛 of the sequence 1, 2, 1, 4, 1, 6, … is
1 𝑛 𝑖𝑠 odd
best written in the form 𝑎𝑛 = { ,
𝑛 𝑛 𝑖𝑠 even
since it would be tedious to find one formula applicable to
all terms. By considering even and odd term separately,
find a formula for the nth term of the given sequence
1 1 1
1, ,3 , ,5, ,…
22 24 26

(21) Prove:
1 +∞
(a) The sequence { } converge to 0.
𝑛 𝑛=1

𝑛 +∞
(b) The sequence{ } converge to 1.
𝑛+1 𝑛=1

(22) Consider the sequence {𝑎𝑛 }+∞

𝑛=1 whose nth term is
𝑛−1
1 1
𝑎𝑛 = ∑ 𝑘 .
𝑛
𝑘=0 1 + 𝑛

Show that lim 𝑎𝑛 = ln 2

𝑛→+∞

- 237 -
3. MONOTONE SEQUENCES
Definition 1.
A sequence {𝑎𝑛 } is called
increasing if 𝑎1 < 𝑎2 < 𝑎3 < ⋯ < 𝑎𝑛 < ⋯
nondecreasing if 𝑎1 ≤ 𝑎2 ≤ 𝑎3 ≤ ⋯ ≤ 𝑎𝑛 ≤ ⋯
decreasing if 𝑎1 > 𝑎2 > 𝑎3 > ⋯ > 𝑎𝑛 > ⋯
nonincreasing if 𝑎1 ≥ 𝑎2 ≥ 𝑎3 ≥ ⋯ ≥ 𝑎𝑛 ≥ ⋯
A sequence that is either nondecreasing or nonincreasing
is called monotone, and a sequence that is increasing or
decreasing is called strictly monotone. Observe that a
strictly monotone sequence is monotone, but not
conversely.
Example 1.
1 2 3 𝑛
, , ,…, ,… is increasing
2 3 4 𝑛+1
1 1 1
1 , , ,…, ,… is decreasing
2 3 𝑛

1, 1, 2, 2, 3, 3, … is nondecreasing
1 1 1 1
1 ,1 , , , , , … is nonincreasing
2 2 3 3

All four of these sequences are monotone, but the sequence

1 1 1 1
1 , − , , − , … , (−1)𝑛+1 ,… is not. ◄
2 3 4 𝑛

- 238 -
The first and second sequences are strictly monotone.
Frequently, one can guess whether a sequence is
increasing, decreasing, nondecreasing, or nonincreasing
after writing out some of the initial terms. However, to be
certain that the guess is correct, the kind of analysis
illustrated in the next few examples is needed.
Example 2.
Show that
1 2 3 𝑛
, , ,…, ,…
2 3 4 𝑛+1
is an increasing sequence.
Solution.
Although it is intuitively clear that the sequence is
increasing, intuition cannot replace a definitive proof. To
𝑥
prove that the sequence is increasing, let 𝑓(𝑥 ) =
𝑥+1

So, the nth term in the given sequence is 𝑎𝑛 = 𝑓(𝑛). The

1
function f is increasing for 𝑥 ≥ 1 since 𝑓′(𝑥 ) = (𝑥+1)2. So

𝑓′(𝑥) > 0 for 𝑥 ≥ 1 . Thus 𝑎𝑛 = 𝑓(𝑛) < 𝑓(𝑛 + 1) =

𝑎𝑛+1 .
This proves that the given sequence is increasing.

- 239 -
Alternate Solution.
𝑛
Since 𝑎𝑛 = , we have by replacing n by 𝑛 + 1
𝑛+1
𝑛+1 𝑛+1
𝑎𝑛+1 = (𝑛+1)+1 = .
𝑛+2

Thus, for 𝑛 ≥ 1
𝑛 𝑛 + 1 𝑛2 + 2𝑛 − 𝑛2 − 2 𝑛
𝑎𝑛 − 𝑎𝑛+1 = − =
𝑛+1 𝑛+2 (𝑛 + 1)(𝑛 + 2)
1
= − (𝑛+1)(𝑛+2) < 0.

So that 𝑎𝑛 < 𝑎𝑛+1 for 𝑛 ≥ 1 . This proves that the

sequence is increasing.
●To prove that a sequence {𝑎𝑛 } is decreasing, one can
show either that 𝑓′(𝑥) < 0 for 𝑥 ≥ 1 [where 𝑓(𝑛) =
𝑎𝑛 ] or 𝑎𝑛 − 𝑎𝑛+1 > 0 for 𝑛 ≥ 1.
●If a sequence {𝑎𝑛 } has positive terms, then it is
increasing if
𝑎𝑛+1
>1 (1a)
𝑎𝑛

for 𝑛 ≥ 1, and it is decreasing if

𝑎𝑛+1
<1 (1b)
𝑎𝑛

for 𝑛 ≥ 1. ◄

- 240 -
Example 3.
Use (1a) to prove that the sequence in Example 2 above is
increasing.
Solution.
As shown in the alternate solution of Example 2,
𝑛
𝑎𝑛 =
𝑛+1
𝑛+1
⟹ 𝑎𝑛+1 =
𝑛+2
(𝑛+1)
𝑎𝑛+1 (𝑛+2)
⟹ = 𝑛
𝑎𝑛
(𝑛+1)

𝑛2 + 2 𝑛 + 1
=
𝑛2 + 2 𝑛
𝑎𝑛+1
That is > 1 for 𝑛 ≥ 1.
𝑎𝑛

This proves that the sequence is increasing. ◄

●In our subsequent work n! (n factorial) is the product of

the first n positive integers, that is 𝑛 ! = 1 . 2 . 3 … 𝑛 and

0 ! = 1.

- 241 -
Example 4.
Show that the sequence
𝑒 𝑒2 𝑒3 𝑒𝑛
, , ,…, ,…
2! 3! 4! (𝑛 + 1)!

is decreasing

Solution.

𝑒𝑛
We will use (1b). Since 𝑎𝑛 = (𝑛+1)!
, it follows on

replacing n by 𝑛 + 1 that

𝑒 𝑛+1 𝑒 𝑛+1
𝑎𝑛+1 = =
[(𝑛 + 1) + 1]! (𝑛 + 2)!

Thus

𝑒 𝑛+1
𝑎𝑛+1 (𝑛+2)! 𝑒 𝑛+1 (𝑛 + 1)! 𝑒
= 𝑒𝑛
= . =
𝑎𝑛 𝑎𝑛 (𝑛 + 2)! 𝑛 + 2
(𝑛+1)!

For 𝑛 ≥ 1, 𝑛 + 2 ≥ 3 > 𝑒(≈ 2 .718 … ).

𝑎𝑛+1 𝑒
So, = < 1 for 𝑛 ≥ 1.
𝑎𝑛 𝑛+2

This proves that the sequence is decreasing. ◄

- 242 -
Theorem 2.
For a nondecreasing sequence 𝑎1 ≤ 𝑎2 ≤ ⋯ ≤ 𝑎𝑛 ≤ ⋯
there are two possibilities:
(a) There is a constant M such that𝑎𝑛 ≤ 𝑀 for all n,
in which case the sequence converges to a limit L
satisfying 𝐿 ≤ 𝑀.
(b) No such constant exists, in which case
lim 𝑎𝑛 = +∞
𝑛→+∞

Theorem 3.
For a nonincreasing sequence 𝑎1 ≥ 𝑎2 ≥ ⋯ ≥ 𝑎𝑛 ≥ ⋯
there are two possibilities:
(a) There is a constant M such that 𝑎𝑛 ≥ 𝑀 for all n,
in which case the sequence converges to a limit L
satisfying 𝐿 ≤ 𝑀.
(b) No such constant exists, in which case
lim 𝑎𝑛 = −∞.
𝑛→+∞

Axiom 1.
If 𝑆 is a nonempty set of real numbers, and if there is some
real number that is greater than or equal to every number

- 243 -
in 𝑆, then there is a smallest real number that is greater than
or equal to every number in 𝑆.
For example, let 𝑆 be the set of numbers in the interval
open interval ]1, 3[. It is true that there exists a number 𝑢
greater than or equal to every number in 𝑆; some examples
are 𝑢 = 10, 𝑢 = 100 , and 𝑢 = 3.02. The smallest
number 𝑢 that is greater than or equal to every number in
𝑆 is 𝑢 = 3.
●Let us call a number 𝑢 an upper bound for a set 𝑆 if 𝑢 is
greater than or equal to every number in 𝑆.
Axioms 2.
If a nonempty set 𝑆 of real numbers has an upper bound,
then 𝑆 has a smallest upper bound.

The smallest upper bound of 𝑆 is commonly called the

least upper bound of S.

Example 5.
𝑛 +∞
As shown in Examples 2 and 3, the sequence { } is
𝑛+1 𝑛=1
𝑛
nondecreasing. Since 𝑎𝑛 = < 1 , 𝑛 = 1 , 2 , …, then
𝑛+1

the terms in the sequence have 𝑀 = 1 as an upper

- 244 -
bound. By Theorem 2 above the sequence must converge
to a limit 𝐿 ≤ 𝑀. This is indeed the case since
𝑛 1
lim = lim 1 = 1.◄
𝑛→+∞ 𝑛+1 𝑛→+∞ 1+𝑛

Because the limit of a sequence {𝑎𝑛 } describes the

behavior of the terms as n gets large, one can add or even
delete a finite number of terms in a sequence without
affecting either the convergence or the value of the limit.
That is, the original and the modified sequence will both
converge, or both diverge, and in the case of convergence
both will have the same limit. We omit the proof.
Example 6.
5𝑛 +∞
Show that the sequence { }
𝑛! 𝑛=1

converges.
Solution.
It is tedious to determine convergence directly from the
5𝑛
limit lim . Thus, we will proceed indirectly.
𝑛→+∞ 𝑛!
5𝑛 5𝑛+1
If we let 𝑎𝑛 = , then 𝑎𝑛+1 = (𝑛+1)!. So that
𝑛!

- 245 -
 5𝑛+1
𝑎𝑛+1 (𝑛+1)! 5𝑛+1 𝑛! 5
= 5𝑛
= 𝑛 . =
𝑎𝑛 5 (𝑛 + 1)! 𝑛 + 1
𝑛!
𝑎𝑛+1
For 𝑛 = 1, 2, and 3, the value of is greater than 1.
𝑎𝑛

So 𝑎𝑛+1 > 𝑎𝑛 . Thus 𝑎1 < 𝑎2 < 𝑎3 < 𝑎4 .

𝑎𝑛+1
For 𝑛 = 4 the value of is 1. So 𝑎4 = 𝑎5 .
𝑎𝑛
𝑎𝑛+1
For 𝑛 ≥ 5 the value of is less than 1, so
𝑎𝑛

𝑎5 > 𝑎6 > 𝑎7 > 𝑎8 > ⋯

Thus, if we discard the first four terms of the given
sequence (which won't affect convergence) the resulting
sequence will be decreasing. Moreover, each term in the
sequence is positive. So that by Theorem 3, the sequence
converges to some limit that is ≥ 0.

- 246 -
 SOLVED PROBLEMS
1. Give an example of a monotonic increasing sequence
which is (i) convergent (ii) divergent.
Solution.
1 2 3 𝑛
(i) The sequence , , ,…, , … is a monotonic
2 3 4 𝑛+1

increasing sequence which is convergent as

𝑛 1
lim 𝑎𝑛 = lim = lim 1 = 1, which is finite.
𝑛→+∞ 𝑛→+∞ 𝑛+1 𝑛→+∞ 1+𝑛

∴ The sequence converges to 1.

(ii) The sequence 1 , 2 , 3 , … , 𝑛 , … is a monotonic
increasing sequence which is divergent as
lim 𝑎𝑛 = lim 𝑛 = +∞, which is infinite.
𝑛→+∞ 𝑛→+∞

∴ The sequence diverges to +∞.◄

2. Give an example of a monotonic decreasing sequence
which is (i) convergent (ii) divergent.
Solution.
1 1 1
(i) The sequence 1 , , , … , , … is a monotonic
2 3 𝑛

decreasing sequence which is convergent as

1
lim 𝑎𝑛 = lim = 0, which is finite.
𝑛→+∞ 𝑛→+∞ 𝑛

∴ The sequence converges to 0.

- 247 -
(ii) The sequence −1 , −2 , −3 , … , −𝑛 , … is a monotonic
decreasing sequence which is divergent as
lim 𝑎𝑛 = lim −𝑛 = −∞, which is infinite.
𝑛→+∞ 𝑛→+∞

∴ The sequence diverges to −∞.◄

3. Discuss the convergence of the sequence {𝑎𝑛 }, where
𝑛 +∞
(i) { } ;
𝑛2 +1 𝑛=1

1 1 1 +∞
(ii) {1 + + +⋯+ } ;
3 32 3𝑛 𝑛=1

𝑛+1 +∞
(iii) { } .
𝑛 𝑛=1

Solution.
𝑛
(i) Here 𝑎𝑛 =
𝑛2 +1
𝑛+1 𝑛
𝑎𝑛+1 − 𝑎𝑛 = −
(𝑛 + 1)2 + 1 𝑛2 + 1
−𝑛2 −𝑛+1
= (𝑛2 < 0, ∀𝑛 ⟹ 𝑎𝑛+1 < 𝑎𝑛
+2𝑛+2)(𝑛2 +1)

⟹ {𝑎𝑛 } is a decreasing sequence.

𝑛
Also, 𝑎𝑛 = > 0 ∀𝑛
𝑛2 +1

⟹ {𝑎𝑛 } is bounded below by 0.

∴{𝑎𝑛 } is decreasing and bounded below, it is convergent.

- 248 -
 1
𝑛 𝑛
lim 𝑎𝑛 = lim = lim 1 = 0.
𝑛→+∞ 𝑛→+∞ 𝑛2 +1 𝑛→+∞ 1+ 2
𝑛

∴ The sequence {𝑎𝑛 } converges to zero.

1 1 1
(ii) Here 𝑎𝑛 = 1 + + +⋯+
3 32 3𝑛

= sum of (𝑛 + 1) terms of a G.P. whose first term is 1

and common ratio is 1.
1
1 (1 − ) 𝑎(1 − 𝑟 𝑛 )
3𝑛+1
= 1
|∵ 𝑆𝑛 =
(1 − 3) 1−𝑟

3 1
= (1 − 𝑛+1 ).
2 3
1 1 1 1
Now, 𝑎𝑛 = 1 + + + ⋯+ ++
3 32 3𝑛 3𝑛+1
1
∴ 𝑎𝑛+1 − 𝑎𝑛 = > 0∀𝑛 ⟹ 𝑎𝑛+1 > 𝑎𝑛 ∀𝑛
3𝑛+1
⟹ {𝑎𝑛 } is a increasing sequence.
3 1 3
Also, 𝑎𝑛 = (1 − 𝑛+1 ) < ∀𝑛
2 3 2
3
⟹ {𝑎𝑛 } is bounded above by .
2

∴{𝑎𝑛 } is increasing and bounded above, it is convergent.

3 1 3 3
lim 𝑎𝑛 = lim (1 − 𝑛+1 ) = (1 − 0) =
𝑛→+∞ 𝑛→+∞ 2 3 2 2
3
∴ The sequence {𝑎𝑛 } converges to .
2

- 249 -
 𝑛+1
(iii) Here 𝑎𝑛 =
𝑛
𝑛+2 𝑛+1
𝑎𝑛+1 − 𝑎𝑛 = −
𝑛+1 𝑛
−1
= < 0, ∀𝑛 ⟹ 𝑎𝑛+1 < 𝑎𝑛
𝑛(𝑛+1)

⟹ {𝑎𝑛 } is a decreasing sequence.

𝑛+1 1
Also, 𝑎𝑛 = = 1 + > 1 ∀𝑛
𝑛 𝑛

⟹ {𝑎𝑛 } is bounded below by 1.

∴{𝑎𝑛 } is decreasing and bounded below, it is convergent.
𝑛+1 1
lim 𝑎𝑛 = lim = lim (1 + ) = 1.
𝑛→+∞ 𝑛→+∞ 𝑛 𝑛→+∞𝑛

∴ The sequence {𝑎𝑛 } converges to 1.◄

4. Evaluate
𝑛! 𝑛𝑛
(i) lim ; (ii) lim ;
𝑛→+∞ 𝑛𝑛 𝑛→+∞ (2𝑛)!
2 2
2𝑛 2𝑛
(iii) lim ; (iv) lim .
𝑛→+∞ 𝑛! 𝑛→+∞ (2𝑛)!

Solution.
𝑛! 1∙2∙3∙∙∙𝑛 1∙𝑛∙𝑛∙∙∙𝑛 1
(i) Note that 0 ≤ = ≤ = .
𝑛𝑛 𝑛∙𝑛∙𝑛∙∙∙𝑛 𝑛∙𝑛∙𝑛∙∙∙𝑛 𝑛
1
Since lim = 0, then by squeeze theorem (Sandwich
𝑛→+∞ 𝑛
𝑛!
theorem) lim = 0.
𝑛→+∞ 𝑛𝑛

- 250 -
(ii) Note that
𝑛𝑛 𝑛𝑛 𝑛𝑛 𝑛𝑛 1
0≤ = ≤ = = .
(2𝑛)! (2𝑛)(2𝑛−1)∙∙∙(𝑛+1)𝑛! 𝑛∙𝑛∙𝑛∙∙∙𝑛! 𝑛𝑛 ∙𝑛! 𝑛!
1
Since lim = 0, then by squeeze theorem (Sandwich
𝑛→+∞ 𝑛!
𝑛𝑛
theorem) lim = 0.
𝑛→+∞ (2𝑛)!

(iii) Note that

2 2 2 2
2𝑛 2𝑛 2𝑛 2𝑛 2𝑛 𝑛
0≤ = ≥ = =( ) .
𝑛! 𝑛(𝑛−1)∙∙∙2∙1 𝑛∙𝑛∙∙∙𝑛 𝑛𝑛 𝑛

2𝑛 2𝑛 𝑛
Since > 2 for all 𝑛 ≥ 1, then ( ) ≥ 2𝑛 .
𝑛 𝑛
2
𝑛 2𝑛
Since lim 2 = +∞, then lim = +∞.
𝑛→+∞ 𝑛→+∞ 𝑛!

(iv) Note that

2 2 2 2
2𝑛 2𝑛 2𝑛 2𝑛
0≤ = ≥ = (2𝑛)2𝑛
=
(2𝑛)! 2𝑛(2𝑛−1)(2𝑛−2)∙∙∙2∙1 2𝑛∙2𝑛∙∙∙2𝑛

2𝑛 𝑛
(4𝑛2 ) .
2𝑛
Since lim = +∞ (by L'Hopital's rule), then
𝑛→+∞ 4𝑛2
𝑛 2
2𝑛 2𝑛
lim ( ) = +∞ . Hence lim = +∞.◄
𝑛→+∞ 4𝑛2 𝑛→+∞ (2𝑛)!

- 251 -
 Exercise Set (6.2)
In Exercise 1-6, use differentiation to show that the sequence is
strictly monotone and classify it as increasing or decreasing:
𝑛 +∞ 1 +∞ 1 +∞
(1) { } ; (2) {3 − } ; (3) { } ;
2𝑛+1 𝑛=1 𝑛 𝑛=1 𝑛+ln 𝑛 𝑛=1

ln(𝑛+2) +∞
(4) {𝑛𝑒 −2𝑛 }+∞
𝑛=1 ; (5) { } ; (6) {tan−1 𝑛}+∞
𝑛=1 .
𝑛+2 𝑛=1
In exercise (7) - (12), determine if the given sequence {𝑎𝑛 } is
𝑎
monotone by examining 𝑛+1. If so, classify it as increasing,
𝑎𝑛
decreasing, nonincreasing, or nondecreasing.
𝑛 +∞ 2𝑛 +∞ 𝑛 +∞
(7) { } ; (8) { } ; (9) { 𝑛 } ;
2𝑛+1 𝑛=1 𝑛! 𝑛=1 2 𝑛=1

2𝑛 +∞ 10 +∞ 𝑛𝑛 +∞
(10) { 𝑛} ; (11) { 𝑛2
} ; (12) { } .
1+2 𝑛=1 2 𝑛=1 𝑛! 𝑛=1

In Exercise 13 - 18, shows the sequence is monotone and apply

Theorem 2 or 3 above to determine if the sequence converges.
1 +∞ 2 +∞ 1 +∞
(13) {2 + } ; (14) {4 − } ; (15) {𝑛 − } ;
𝑛 𝑛=1 𝑛 𝑛=1 𝑛 𝑛=1

4 𝑛−1 +∞ 𝑛 +∞ 𝜋 +∞
(16) { } ; (17) { 𝑛 } ; (18) {cos } .
5 𝑛+2 𝑛=1 5 𝑛=1 2 𝑛 𝑛=1

(19) Find the limit (if it exists) of the sequence

1 1 1 1
(a) 1, −1 , 1 , , , , , …
2 3 4 5
1
(b) − , 0 ,0 ,0 , 1 , 2 , 3 , 4 , …
2

- 252 -
- 253 -
 Chapter (VII)
INFINITE SERIES

Definition 1.
An infinite series is an expression of the form
𝑢1 + 𝑢2 + 𝑢3 + ⋯ + 𝑢𝑘 + ⋯
or in sigma notation ∑∞
𝑘=1 𝑢𝑘

The numbers 𝑢1 , 𝑢2 , 𝑢3 , … called the terms of the series.

Informally speaking the expression ∑∞
𝑘=1 𝑢𝑘 directs us to

obtain the "sum" of the terms 𝑢1 , 𝑢2 , 𝑢3 , …. To carry out

this summation process we proceed as follows:
Let 𝑠𝑛 denote the sum of the first 𝑛 terms of the series.
𝑠1 = 𝑢1
𝑠2 = 𝑢1 + 𝑢2
𝑠3 = 𝑢1 + 𝑢2 + 𝑢3
⋮
𝑠𝑛 = 𝑢1 + 𝑢2 + 𝑢3 + ⋯ + 𝑢𝑛 = ∑𝑛𝑘=1 𝑢𝑘
The number 𝑠𝑛 is called the nth partial sum of the series
and the sequence {𝑠𝑛 }+∞
𝑛=1 is called the sequence of partial

sums.

- 254 -
Example 1.
For the infinite series
3 3 3 3
+ 2+ 3+ 4+⋯
10 10 10 10
the partial sums are
3
𝑠1 =
10
3 3 33
𝑠2 = + =
10 102 100
3 3 3 333
𝑠3 = + 2 + =
10 10 103 1000
3 3 3 3 3333
𝑠4 = + 2 + 3 + =
10 10 10 104 10000

⋮
.◄

- 255 -
Definition 2.
Let {𝑠𝑛 } be the sequence of partial sums of the series
∑∞
𝑘=1 𝑢𝑘 .

If the sequence {𝑠𝑛 } converges to a limit 𝑆, then the series

is said to converge, and 𝑆 is called the sum of the series,

i.e. lim 𝑠𝑛 = 𝑆.
𝑛→∞

We denote this by writing 𝑆 = ∑∞

𝑘=1 𝑢𝑘 .

If the sequence of partial sums of a series diverges, then

the series is said to diverge, i.e. lim 𝑠𝑛 = +∞ or

𝑛→∞

lim 𝑠𝑛 = −∞.. A divergent series has no sum.

𝑛→∞

If the sequence {𝑠𝑛 } is oscillatory sequence, then the

series ∑∞
𝑘=1 𝑢𝑘 is oscillatory series and:

(i) The series ∑∞

𝑘=1 𝑢𝑘 oscillate finitely when the

sequence {𝑠𝑛 } oscillate finitely.

(ii) The series ∑∞

𝑘=1 𝑢𝑘 oscillate infinitely when the

sequence {𝑠𝑛 } oscillate infinitely.

- 256 -
Example 2.
Determine if the series
3 3 3 3
+ 2 + 3 + ⋯+ +⋯
10 10 10 10𝑘

converges, oscillate finitely, oscillate infinitely or

diverges. If it converges find the sum.
Solution.
Let us verify that this is indeed the case. The nth partial
sum is
3 3 3
𝑠𝑛 = + +⋯+ (1)
10 102 10𝑛

The problem of calculating lim 𝑠𝑛 is complicated by the

𝑛→+∞

fact that the number of terms in (1) changes with n. For

purposes of calculation, it is desirable to rewrite (1) in
closed form as follows:
1
We multiply both sides of (1) by to obtain
10
1 3 3 3 3
𝑠𝑛 = 2 + 3 + ⋯+ 𝑛 + (2)
10 10 10 10 10𝑛+1

and then subtract (2) from (1) to obtain

1 3 3
𝑠𝑛 − 𝑠𝑛 = − .
10 10 10𝑛+1

Therefore
9 3 1
𝑠𝑛 = (1 − 10𝑛 ).
10 10
- 257 -
 1 1
𝑠𝑛 = (1 − 𝑛 ) (3)
3 10
1
Since → 0 𝑎𝑠 𝑛 → +∞, it follows from (3) that
10𝑛
1
𝑆 = lim 𝑠𝑛 = .
𝑛→+∞ 3
1
Thus, the given series converges, and its sum is ,
3

i.e.,
1 3 3 3 3
= + 2 + 3 + ⋯+ + ⋯◄
3 10 10 10 10𝑘

Example 3.
Determine the series
1 − 1 + 1 − 1 + …
converges, oscillate finitely, oscillate infinitely or
diverges. If it converges find the sum.
Solution.
The partial sums are
𝑠1 = 1 ,
𝑠2 = 1 − 1 = 0 ,
𝑠3 = 1 − 1 + 1 = 1,
𝑠4 = 1 − 1 + 1 − 1 = 0,
⋮
Thus, the sequence of partial sums is

- 258 -
 1, 0, 1, 0, 1, 0, …
Since this is an oscillatory sequence which oscillate
finitely, the given series oscillate finitely and consequently
has no sum. ◄
Example 4.
Determine if the series
∞
1 1 1 1
∑ = + + +⋯
𝑘(𝑘 + 1) 1.2 2.3 3.4
𝑘=1

converges, oscillate finitely, oscillate infinitely or

diverges. If it converges find the sum.
Solution.
The nth partial sum of the series is
𝑛
1 1 1 1 1
𝑠𝑛 = ∑ = + + + ⋯+
𝑘(𝑘 + 1) 1.2 2.3 3.4 𝑛(𝑛 + 1)
𝑘=1

To calculate lim 𝑠𝑛 we will rewrite 𝑠𝑛 in closed form.

𝑛→+∞

This may be accomplished by using the method of partial

fractions to obtain (verify):
1 1 1
= −
𝑘(𝑘+1) 𝑘 𝑘+1

from which it follows that

- 259 -
 1 1
𝑠𝑛 = ∑𝑛𝑘=1 ( − )
𝑘 𝑘+1
1 1 1 1 1 1 1
= (1 − ) + ( − ) + ( − ) + ⋯ + ( − )
2 2 3 3 4 𝑛 𝑛+1
1 1 1 1 1 1 1
= 1 + (− + ) + (− − ) + ⋯ + (− + ) − .
2 2 3 3 𝑛 𝑛 𝑛+1

The sum above is an example of a telescoping sum, which

means that each term cancels part of the next term, thereby
collapsing the sum (like a folding telescope) into only two
terms. After the cancellation,
1
𝑠𝑛 = 1 −
𝑛+1
So
1
lim 𝑠𝑛 = lim (1 − )=1
𝑛→+∞ 𝑛→+∞ 𝑛+1
Therefore
∞
1
1=∑ .
𝑘(𝑘 + 1)
𝑘=1

Example 5.
One of the most famous and important of all diverging
series is the harmonic series
∞
1 1 1 1
∑ =1+ + + +⋯
𝑘 2 3 4
𝑘=1

- 260 -
However, the divergence will become apparent when we
examine the partial sums in detail. Because the terms in the
series are all positive, the partial sums
𝑠1 = 1,
1
𝑠2 = 1 + ,
2
1 1
𝑠3 = 1 + + ,…
2 3

form an increasing sequence

𝑠1 < 𝑠2 < 𝑠3 < ⋯ < 𝑠𝑛 < ⋯
Thus, by Theorem 2 in Section 2 of Chapter VI we can
prove divergence by demonstrating that there is no
constant M that is greater than or equal to every partial
sum. To this end, we will consider some selected sums,
namely 𝑠2 , 𝑠4 , 𝑠8 , 𝑠16 , 𝑠32 , …. Note that the subscripts
are successive powers of 2, so that these are the partial
sums of the form 𝑠2𝑛 . These partial sums satisfy the
inequalities
1 1 1 2
𝑠2 = 1 + > + =
2 2 2 2
1 1 1 1 1 3
𝑠4 = 𝑠2 + + > 𝑠2 + ( + ) = 𝑠2 + >
3 4 4 4 2 2
1 1 1 1 1 1 1 1
𝑠8 = 𝑠4 + + + + > 𝑠4 + ( + + + )
5 6 7 8 8 8 8 8

- 261 -
 1 4
= 𝑠4 + >
2 2
1 1 1 1 1 1 1 1
𝑠16 = 𝑠8 + + + + + + + +
9 10 11 12 13 14 15 16
1 1 1 1 1 1 1 1
> 𝑠8 + ( + + + + + + + )
16 16 16 16 16 16 16 16

1 5
= 𝑠8 + >
2 2
⋮
𝑛+1
𝑠 2𝑛 > .
2

Now if M is any constant, we can certainly find a positive

(𝑛+1)
integer 𝑛 such that > 𝑀. But for this 𝑛
2
𝑛+1
𝑠2𝑛 > >𝑀
2
So, no constant M is greater than or equal to every partial
sum of the harmonic series. This proves divergence. ◄
Example 6.
Determine if the series
∞

∑ 𝑘 (−1)𝑘 = −1 + 2 − 3 + 4 − 5 + 6 …
𝑘=1

converges, oscillate finitely, oscillate infinitely or

diverges. If it converges find the sum.

- 262 -
Solution.
The partial sums are
𝑠1 = −1 ,
𝑠2 = −1 + 2 = 1 ,
𝑠3 = −1 + 2 − 3 = −2,
𝑠4 = −1 + 2 − 3 + 4 = 2,
𝑠5 = −1 + 2 − 3 + 4 − 5 = −3,
𝑠6 = −1 + 2 − 3 + 4 − 5 + 6 = 3,
⋮
Thus, the sequence of partial sums is
−1, 1, −2, 2, −3, 3, …
The nth partial sum of the series is
𝑠𝑛 = −1 + 2 − 3 + 4 − 5 + 6 + ⋯ + to 𝑛 terms
𝑛+1
−( ) , 𝑛 is odd
={ 2
𝑛
𝑛 is even.
2
The subsequence {𝑠2𝑛−1 } diverges to – ∞, while the
subsequence {𝑠2𝑛 }diverges to + ∞
∴ {𝑠𝑛 } oscillates infinitely.
⇒ ∑∞ 𝑘
𝑘=1 𝑘 (−1) oscillates infinitely.◄

- 263 -
●The series in Examples 2 and 3 are examples of

geometric series. A geometric series is one of the forms

𝑎 + 𝑎 𝑟 + 𝑎𝑟 2 + ⋯ + 𝑎𝑟 𝑘−1 + ⋯ (𝑎 ≠ 0)

where each term is obtained by multiplying the previous

one by a constant 𝑟. the multiplier 𝑟 is called the ratio for

the series.

Some examples of geometric series are:

1 + 2 + 4 + 8 + ⋯ + 2𝑘−1 + ⋯ (𝑎 = 1 , 𝑟 = 2);

3 3 3 1
3+ + +⋯+ + ⋯ (𝑎 = 3 , 𝑟 = );
10 102 10𝑘−1 10

1 1 1 1 1 1
− + − ⋯ + (−1)𝑘+1 + ⋯ (𝑎 = , 𝑟 = − );
2 4 8 2𝑘 2 2

1+ 1+⋯+1 +⋯ (𝑎 = 1 , 𝑟 = 1);

1 − 1 + 1 − ⋯ + (−1)𝑘+1 + ⋯ (𝑎 = 1 , 𝑟 = −1).

The following theorem is the fundamental result on

convergence of geometric series

- 264 -
Theorem 3.
A geometric series
𝑎 + 𝑎𝑟 + 𝑎𝑟 2 + ⋯ + 𝑎𝑟 𝑘−1 + ⋯ (𝑎 ≠ 0)
(i) converges if |𝑟| < 1 i.e., −1 < 𝑟 < 1;
(ii) diverges if 𝑟 ≥ 1;
(iii) oscillates finitely if 𝑟 = – 1;
(iv) oscillates infinitely if 𝑟 < – 1.
𝑎
When the series converges the sum is .
1−𝑟

Proof.
Let us treat the case|𝑟| = 1 first.
If 𝑟 = 1, then the series is
𝑎 + 𝑎 + ⋯ + 𝑎 + ⋯.
So that the nth partial sum is 𝑠𝑛 = 𝑛𝑎 and lim 𝑠𝑛 =
𝑛→+∞

lim 𝑛 𝑎 = ±∞ ( the sign depending on whether 𝑎 is

𝑛→+∞

positive or negative). This proves divergence.

If 𝑟 = −1, the series is
𝑎 − 𝑎 + 𝑎 − 𝑎 + ⋯.
So, the sequence of partial sums is
𝑎, 0, 𝑎, 0, 𝑎, 0, …
which oscillates finitely. So, the series oscillates finitely.
- 265 -
Now, let us consider the case where |𝑟| ≠ 1 . The nth
partial sum of the series is
𝑠𝑛 = 𝑎 + 𝑎 𝑟 + 𝑎𝑟 2 + ⋯ + 𝑎𝑟 𝑛−1 (*)
Multiplying both sides of (*) by 𝑟 yields
𝑟 𝑠𝑛 = 𝑎 𝑟 + 𝑎𝑟 2 + ⋯ + 𝑎𝑟 𝑛−1 + 𝑎𝑟 𝑛 (**)
and subtracting (**) form (*) gives
𝑠𝑛 − 𝑟𝑠𝑛 = 𝑎 − 𝑎𝑟 𝑛
or
(1 − 𝑟)𝑠𝑛 = 𝑎 − 𝑎𝑟 𝑛
Since 𝑟 ≠ 1 in the case we are considering, this can be
𝑎−𝑎𝑟 𝑛 𝑎 𝑎𝑟 𝑛
rewritten as 𝑠𝑛 = = − .
1−𝑟 1−𝑟 1−𝑟

If |𝑟| < 1, then lim 𝑟 𝑛 = 0. So, that {𝑠𝑛 } converges and

𝑛→+∞
𝑎
lim 𝑠𝑛 =
𝑛→+∞ 1−𝑟
𝑎
Therefore, the series converges, and its sum is .
1−𝑟

If |𝑟| > 1 then either 𝑟 > 1 or 𝑟 < −1.

In the case 𝑟 > 1, lim 𝑟 𝑛 = +∞ and the series
𝑛→+∞

diverges.

- 266 -
If 𝑟 < −1, 𝑟 𝑛 oscillates between positive and negative
values that grow in magnitude, so {𝑠𝑛 } oscillates
infinitely. Therefore, the series oscillates infinitely. ■
Example 1.
5 5 5
The series 5 + + + ⋯+ +⋯
4 42 4 𝑘−1
1
is a geometric series with 𝑎 = 5 and 𝑟 = .
4
1
Since |𝑟| = < 1, the series converges, and the sum is
4
𝑎 5 20
= 1 = .◄
1−𝑟 1− 3
4

Example 2.
Find the rational number represented by repeating decimal
0.784784784 …
Solution.
We can write
0.784784784 … = 0.784 + 0.000784 +
0.000000784 + …
so, the given decimal is the sum of a geometric series with
𝑎 = 0.784 and 𝑟 = 0.001. Thus
𝑎 0.784 0.784 784
0.784784784 … = = = = .◄
1−𝑟 1−0.001 0.999 999

- 267 -
 Exercise Set (7.1)
(1) In each part, find the first four partial sums, find a
closed form for the nth partial sum; determine if this
series converges, and if so, give the sum.
2 1 2𝑘−1
(a) ∑∞
𝑘=1 ; (b) ∑∞
𝑘=1 ; (c) ∑∞
𝑘=1 .
5𝑘−1 (𝑘+1)(𝑘+2) 4

In Exercise (2) - (16), determine if the series converges or

diverges. If it converges, find the sum.
1 3 𝑘−1 2 𝑘+2
(2) ∑∞ ∞
𝑘=1 5𝑘 ; (3) ∑𝑘=1 (− 4) ; (4) ∑∞
𝑘=1 (3) ;
7
(5) ∑∞
𝑘=1(−1)
𝑘−1
; (6) ∑∞
𝑘=1 4
𝑘−1
;
6𝑘−1

∞ 3 𝑘+1 1 1
(7) ∑𝑘=1 (− ) ; (8) ∑∞
𝑘=1 ( − );
2 𝑘+3 𝑘+4
1 1 1
(9) ∑∞
𝑘=1 ; (10) ∑∞
𝑘=1 ( 𝑘 − );
(𝑘+2)(𝑘+3) 2 2𝑘+1

1 1 4𝑘+2
(11) ∑∞
𝑘=1 ; (12) ∑∞
𝑘=1 ; (13) ∑∞
𝑘=1 ;
9𝑘 2 +3𝑘−2 𝑘 2 −1 7𝑘−1

𝑒 𝑘−1 1 𝑘 5
(14) ∑∞
𝑘=1 (𝜋) ; (15) ∑∞ ∞
𝑘=1 (− 2) ; (16) ∑𝑘=1 𝑘−2.

In Exercise (17) - (22), express the repeating decimal as a

fraction.
(17) 0.4444 … ; (18) 0.9999. ..;
(19) 5.373737 …; (20) 0.159159159 …;
(21) 0.782178217821 …; (22) 0.451141414 …
- 268 -
(23) Find a closed form for the nth partial sum of the series
1 2 3 𝑛
ln ( ) + ln ( ) + ln ( ) + ⋯ + ln ( )+⋯
2 3 4 𝑛+1
and determine if the series converges.
1
(25) Show ∑∞
𝑘=1 In (1 − ) = −In 2
𝑘2

√𝑘+1−√𝑘
(26) Show ∑∞
𝑘=1 =1
√𝑘 2 +𝑘

(27) Use geometric series to show:

1
(a) ∑∞ 𝑘 𝑘
𝑘=1(−1) 𝑥 = if − 1 < 𝑥 < 1
1+𝑥
1
(b) ∑∞ 𝑘
𝑘=1(𝑥 − 3) = if 2 < 𝑥 < 4
4−𝑥
1
(c) ∑∞ 𝑘 2𝑘
𝑘=1(−1) 𝑥 = if −1 < 𝑥 < 1.
1+𝑥 2

- 269 -
2. Convergence of an infinite series
Theorem 1.
(Necessary condition for convergence)
If an infinite series ∑ 𝑢𝑘 converges, then lim 𝑢𝑘 = 0.
𝑘→+∞

Proof.
The term 𝑢𝑘 can be written
𝑢𝑘 = 𝑠𝑘 − 𝑠𝑘−1 (#)
where 𝑠𝑘 is the sum of the first 𝑘 terms and 𝑠𝑘−1 is the sum
of the first 𝑘 − 1 terms.
If 𝑆 denotes the sum of the series, then lim 𝑠𝑘 = 𝑆.
𝑘→+∞

Since (𝑘 − 1) → +∞ as 𝑘 → +∞, we also have

lim 𝑠𝑘−1 = 𝑆. Thus from (#)
𝑘→+∞

lim 𝑢𝑘 = lim (𝑠𝑘 − 𝑠𝑘−1 ) = 𝑆 − 𝑆 = 0.■

𝑘→+∞ 𝑘→+∞

Theorem 2 (The Divergence test).

If lim 𝑢𝑘 does not exist or lim 𝑢𝑘 ≠ 0, then the series
𝑘→+∞ 𝑘→+∞

∑ 𝑢𝑘 diverges.
Proof.
It is just an alternate phrasing of the above theorem and
needs no additional proof. ■
- 270 -
Example 1.
The series
∞
𝑘 1 2 3 𝑘
∑ = + + +⋯+ +⋯
𝑘+1 2 3 4 𝑘+1
𝑘=1
𝑘 1
diverges since lim = lim 1 = 1 ≠ 0.◄
𝑘→+∞ 𝑘+1 𝑘→+∞ 1+𝑘

Example 2.
The series
∞
1
∑ cos ( )
𝑘
𝑘=1
1
diverges since lim cos ( ) = cos 0 = 1 ≠ 0.◄
𝑘→+∞ 𝑘

Example 3.
The series
∞
𝑘
∑√
𝑘+1
𝑘=1

𝑘 1
diverges since lim √ = lim √ 1 = 1 ≠ 0.◄
𝑘→+∞ 𝑘+1 𝑘→+∞ 1+ 𝑘

WARNING:
The converse of Theorem 1 is false. To prove that a series
converges it does not suffice to show that lim 𝑢𝑘 = 0.
𝑘→+∞

- 271 -
Since this property may hold for divergent as well as
convergent series. For example, the kth term of the
1 1
divergent harmonic series 1 + + … + + …tends to
2 𝑘
zero as 𝑘 → +∞ , and the kth term of the convergent
1 1 1
geometric series + 2 + ⋯ + 𝑘 + ⋯ tends to zero
2 2 2
as 𝑘 → +∞.
Theorem 3 (The proof left as exercise).
(a) If ∑ 𝑢𝑘 and ∑ 𝑣𝑘 are convergent series, ∑(𝑢𝑘 + 𝑣𝑘 )
are convergent series and the sums of these series are
related by ∑(𝑢𝑘 ± 𝑣𝑘 ) = ∑∞ ∞
𝑘=1 𝑢𝑘 ± ∑𝑘=1 𝑣𝑘 .

(b) If 𝑐 is a nonzero constant then the series ∑ 𝑢𝑘 and

∑ 𝑐 𝑢𝑘 both converge, or both diverge. In the case of
convergence, the sums are related by
∑∞ ∞
𝑘=1 𝑐 𝑢𝑘 = 𝑐 ∑𝑘=1 𝑢𝑘 .

(c) Convergence or divergence is unaffected by deleting a

finite number of terms from the beginning of a series;
that is, for any positive integer 𝐾, the series
∑∞
𝑘=1 𝑢𝑘 = 𝑢1 + 𝑢2 + 𝑢3 + ⋯

and
∑∞
𝑘=𝐾 𝑢𝑘 = 𝑢𝑘 + 𝑢𝑘+1 + 𝑢𝑘+2 + ⋯

both converge or both diverge.

- 272 -
Example 1.
3 2
Find the sum of the series ∑∞
𝑘=1 ( − ).
4𝑘 5𝑘−1

Solution.
3 3 3 3
The series ∑∞
k=1 (4 k ) = 4 + 4 2 + 4 3 + ⋯ is a convergent

3 1
geometric series (𝑎 = 4 , 𝑟 = 4) and the series
2 2 2 2
∑∞
𝑘=1 ( ) = 2 + 5 + 52 + 53 + ⋯ is also a convergent
5𝑘−1
1
geometric series (𝑎 = 2 , 𝑟 = ) . Thus, by part (a) of
5

Theorem 3 above, the given series converges and

∞ ∞ ∞
3 2 3 2
∑ ( 𝑘 − 𝑘−1 ) = ∑ 𝑘 − ∑ 𝑘−1
4 5 4 5
𝑘=1 𝑘=1 𝑘=1
3
2 3
= 4
1 − 1 = − .◄
1−4 1−5 2

Example 2.
5 5 5 5
The series ∑∞
𝑘=1 (𝑘 ) = 5 + 2 + 3 + ⋯ + 𝑘 + ⋯

diverges by part (b) of Theorem 3 above since

5 1
∑∞ ∞
𝑘=1 (𝑘 ) = ∑𝑘=1 5 (𝑘 ).

So, each term is a constant time the corresponding term of

the divergent harmonic series. ◄

- 273 -
Example 3.
The series
∞
1 1 1 1
∑ = + + +⋯
𝑘 10 11 12
𝑘=10

diverges by part (c) of Theorem 3 above, since this series

results by deleting the first nine terms from the divergent
harmonic series. ◄
Theorem 4.
If ∑ 𝑢𝑘 is a series with positive terms, and if there is a
constant 𝑀 such that
𝑠𝑛 = 𝑢1 + 𝑢2 + ⋯ + 𝑢𝑛 ≤ 𝑀
for every 𝑛 , then the series converges and the sum
𝑆 satisfies 𝑆 ≤ 𝑀 . If no such 𝑀 exists, then the series
diverges.
1
●If we have a series with positive terms, say ∑∞
𝑘=1 𝑘2
+∞ 1
and if we form the improper integral ∫1 𝑑𝑥 whose
𝑥2

integrand is obtained by replacing the summation index 𝑘

by 𝑥, then there is a relationship between convergence of
the series and convergence of the improper integral.

- 274 -
Theorem 5 (Cauchy’s Integral Test).
Let ∑ 𝑢𝑘 be a series with positive terms and let 𝑓(𝑥) be
the function that results when 𝑘 is replaced by 𝑥 in the
formula 𝑢𝑘 . If 𝑓 is decreasing and continuous for 𝑥 ≥ 1,
+∞
then ∑∞
𝑘=1 𝑢𝑘 and ∫1 𝑓(𝑥 )𝑑𝑥 both converge, or both
diverge.
Example 1.
1
Determine if ∑∞
𝑘=1 converges or diverges.
𝑘2

Solution.
1
If we replace 𝑘 by 𝑥 in the formula 𝑢𝑘 = we obtain the
𝑘2
1
function 𝑓(𝑥 ) = which is decreasing and continuous
𝑥2

for 𝑥 ≥ 1. Since
+∞ 1 ℓ 𝑑𝑥
∫1 𝑑𝑥 = lim ∫1
𝑥2 ℓ→+∞ 𝑥2

1 ℓ
= lim [− ]
ℓ→+∞ 𝑥 1
1
= lim [1 − ]
ℓ→+∞ ℓ

= 1, which is finite.
Then the integral converges and consequently the series
converges. ◄

- 275 -
Example 2.
The integral test provides another way to demonstrate
1
divergence of the harmonic series ∑∞
𝑘=1 . If we replace 𝑘
𝑘
1
by 𝑥 in the formula for 𝑎𝑘 = we obtain the function
𝑘
1
𝑓 (𝑥 ) = which satisfies the hypotheses of the integral
𝑥

test (Verify). Since

+∞ 1 ℓ1
∫1 𝑑𝑥 = lim ∫1 𝑑𝑥
𝑥 ℓ→+∞ 𝑥

= lim [ln ℓ − ln 1]
ℓ→+∞

= +∞, which is infinite.

Then the integral and the series diverge. ◄

- 276 -
Example 3.
Determine if the series
∞
1
∑
𝑘2 + 1
𝑘=1

converges or diverges.
Solution.
1
If we replace 𝑘 by 𝑥 in the formula for , we obtain the
𝑘 2 +1
1
function𝑓 (𝑥 ) =
𝑥 2 +1

For 𝑥 ≥ 1 , this function is non-negative, integrable, and

monotonically decreasing function. Thus, the hypotheses
of the integral test are met. But
+∞ ℓ
1 1
∫ 2 𝑑𝑥 = lim ∫ 2 𝑑𝑥
𝑥 +1 ℓ→+∞ 𝑥 +1
1 1

= lim [tan−1 𝑥]1ℓ

ℓ→+∞

= tan−1 ∞ − tan−1 1
𝜋 𝜋 𝜋
= − = which is finite.
2 4 4

Then the improper integral and the series converge. ◄

- 277 -
Example 4.
Determine if the series
∞
1
∑
𝑘 ln 𝑘
𝑘=2

converges or diverges.
Solution.
1
If we replace 𝑘 by 𝑥 in the formula for , we obtain the
𝑘 ln 𝑘
1
function𝑓 (𝑥 ) =
𝑥 ln 𝑥

For 𝑥 ≥ 2 , this function is non-negative, integrable and

monotonically decreasing function. Thus, the hypotheses
of the integral test are met. But
+∞ ℓ
1 1
∫ 𝑑𝑥 = lim ∫ 𝑑𝑥
𝑥 ln 𝑥 ℓ→+∞ 𝑥 ln 𝑥
2 2

= lim [ln ln 𝑥]ℓ2

ℓ→+∞

= ln ln ∞ − ln ln 2
= ∞ which is infinite.
Then the improper integral and the series diverges. ◄

- 278 -
Example 5.
Determine if the series
1 2 3 𝑘
+ + + ⋯ + 𝑘2 + ⋯
𝑒 𝑒4 𝑒9 𝑒
converges or diverges.
Solution.
𝑘
If we replace 𝑘 by 𝑥 in the formula for 2 , we obtain the
𝑒𝑘
𝑥 2
function 𝑓(𝑥 ) = 2 = 𝑥 𝑒 −𝑥 .
𝑒𝑥

For 𝑥 ≥ 1 , this function has positive values and is

continuous. Moreover, for 𝑥 ≥ 1 the derivative
2 2 2
𝑓 ′ (𝑥 ) = 𝑒 −𝑥 − 2𝑥 2 𝑒 −𝑥 = 𝑒 −𝑥 (1 − 2𝑥 2 )
is negative, so that f is decreasing for 𝑥 ≥ 1. Thus, the
hypotheses of the integral test are met. But
+∞ 2 ℓ 2 1 2 ℓ
∫1 𝑥 𝑒 −𝑥 𝑑𝑥 = lim ∫1 𝑥 𝑒 −𝑥 𝑑𝑥 = lim [− 𝑒 −𝑥 ]
ℓ→+∞ 2 ℓ→+∞ 1
1 2 1
= (− ) lim [𝑒 −ℓ − 𝑒 −1 ] =
2 ℓ→+∞ 2𝑒

The improper integral and the series converge. ◄

Remark.
If the summation index in a series ∑ 𝑢𝑘 does not begin with
𝑘 = 1, a variation of the integral test may still apply. It can

- 279 -
 +∞
be shown that lim 𝑢𝑘 and ∫𝑘 𝑓(𝑥 )𝑑 𝑥 both converge, or
𝑘=𝐾

both diverge provided the hypotheses of Theorem 5 hold

for 𝑥 ≥ 𝐾.
♣The harmonic series and the series in Example 1 are
special cases of a class of series called p-series or hyper
harmonic series. A p-series is an infinite series of the form
∞
1 1 1 1
∑ = 1 + + + ⋯ + +⋯
𝑘𝑝 2𝑝 3𝑝 𝑘𝑝
𝑘=1

where 𝑝 > 0.
Examples of p-series are
∞
1 1 1 1
∑ = 1 + + + ⋯+ + ⋯ (𝑝 = 1)
𝑘 2 3 𝑘
𝑘=1
∞
1 1 1 1
∑ = 1 + + + ⋯ + +⋯ (𝑝 = 2)
𝑘2 22 32 𝑘2
𝑘=1
∞
1 1 1 1 1
∑ =1+ + +⋯+ + ⋯ (𝑝 = )
√𝑘 √2 √3 √𝑘 2
𝑘=1

The following theorem tells when a p-series converges.

- 280 -
Theorem 6 (Convergence of p-series).
∞
1 1 1 1
∑ = 1 + + + ⋯ + +⋯
𝑘𝑝 2𝑝 3𝑝 𝑘𝑝
k=1

converge if 𝑝 > 1 and diverge if 0 < 𝑝 ≤ 1.

Proof.
To establish this result when 𝑝 ≠ 1 , we will use the
integral test.
+∞ 1 ℓ
∫1 𝑑𝑥 = lim ∫1 𝑥 −𝑝 𝑑𝑥
𝑥𝑝 ℓ→+∞
ℓ
𝑥 ℓ−𝑝
= lim [ ]
ℓ→+∞ 1−𝑝 1

ℓ1−𝑝 1
= lim [ − ]
ℓ→+∞ 1−𝑝 1−𝑝

For 𝑝 > 1, 1 − 𝑝 < 0 and ℓ1−𝑝 → 0 as ℓ → +∞. So,

the integral and the series converge.
For 0 < 𝑝 < 1, 1 − 𝑝 > 0 and ℓ1−𝑝 → +∞ as ℓ →
+∞, so the integral and the series diverge. The case 𝑝 =
1 is the harmonic series, which was previously shown to
diverge. ◄

- 281 -
Example 1.
1 1 1
1 + 3 + 3 + ⋯+ 3 + ⋯
√2 √3 √𝑘
1
diverge since it is a p-series with 𝑝 = < 1.◄
3

- 282 -
 Exercise Set (7.2)
In Exercise 1-4, find the sum of the series.
1 1 1 1
(1) ∑∞
𝑘=1 [ + ]; (2) ∑∞
𝑘=1 [ − ];
2𝑘 4𝑘 5𝑘 𝑘(𝑘+1)
1 7 7 6
(3) ∑∞
𝑘=2 [ − ]; (4) ∑∞
𝑘=1 [ + (𝑘+3)(𝑘+4)].
𝑘 2 −1 10𝑘−1 3𝑘

In Exercise (5) - (26), determine if the series converges or

diverges.
1 3
(5) ∑∞
𝑘=1 ; (6) ∑∞
𝑘=1 ;
𝑘+6 5𝑘
1 𝑘
(7) ∑∞
𝑘=1 ; (8) ∑∞
𝑘=1 ;
5 𝑘+2 1+𝑘 2
1 1
(9) ∑∞
𝑘=1 2; (10) ∑∞
𝑘=1 3 ;
1+9 𝑘
(4+2 𝑘)2
1 1
(11) ∑∞
𝑘=1 ; (12) ∑∞
𝑘=1 𝑘 ;
√𝑘+5 √𝑒
1 ln 𝑘
(13) ∑∞
𝑘=1 3 ; (14) ∑∞
𝑘=3 ;
√2 𝑘−1 𝑘
𝑘 2
(15) ∑∞
𝑘=1 ; (16) ∑∞
𝑘=1 𝑘 𝑒
−𝑘
;
ln (𝑘+1)

1 𝑘 2 +1
(17) ∑∞
𝑘=1 (𝑘+1)[ln 2 ; (18)
∑∞
𝑘=1 ;
(𝑘+1)] 𝑘 2 +3

1 𝑘 1
(19) ∑∞
𝑘=1 (1 + 𝑘 ) ; (20) ∑∞
𝑘=1 ;
√𝑘 2 +1
tan−1 𝑘
(21) ∑∞
𝑘=1 ; (22) ∑∞ 2
𝑘=1 sech 𝑘 ;
1+𝑘 2
1 3
(23) ∑∞ 2 2
𝑘=1 𝑘 sin (𝑘 ); (24) ∑∞ 2 −𝑘
𝑘=1 𝑘 𝑒 .

- 283 -
 1
(25) Prove ∑∞
𝑘=2 converges if 𝑝 > 1 and
𝑘 (ln 𝑘)𝑝

diverges if 𝑝 ≥ 1.
1
(26) Prove ∑∞
𝑘=3 converges if 𝑝 >
𝑘(ln 𝑘)[ln (ln 𝑘)]𝑝

1 and diverges if 𝑝 ≤ 1.
(27) Prove : If ∑ 𝑢𝑘 converges and ∑ 𝑣𝑘 diverges, then
∑(𝑢𝑘 + 𝑣𝑘 ) diverges and ∑(𝑢𝑘 − 𝑣𝑘 ) diverges.
(28) Find examples to show that ∑(𝑢𝑘 + 𝑣𝑘 ) and
∑(𝑢𝑘 − 𝑣𝑘 ) may converge or may diverge if ∑ 𝑢𝑘 and
∑ 𝑣𝑘 both diverge.
(29) With the help of Exercise (27) determine if the
given series converges or diverges.
1 2 𝑘−1
(a)∑∞
𝑘=1 ൤𝑘 + (3) ൨;

∞ 𝑘2 1
(b) ∑𝑘=1 [ 2 + ];
1+𝑘 𝑘(𝑘+1)

1 1
(c) ∑∞
𝑘=1 ൤ + 3 ൨;
3𝑘+2
𝑘2
1 1
(d) ∑∞
𝑘=2 [ − ].
𝑘(ln 𝑘)2 𝑘2

- 284 -
3. ADDITIONAL CONVERGENCE TESTS
In this section we will develop some additional
convergence tests for series with positive terms.
Theorem 1 (The Comparison Test).
Let ∑ 𝑢𝑘 and ∑ 𝑣𝑘 be series with positive terms and
suppose 𝑢1 ≤ 𝑢1 , 𝑢2 ≤ 𝑢2 , … , 𝑢𝑘 ≤ 𝑣𝑘 , …
(a) If the "bigger series" ∑ 𝑣𝑘 converges, then the
"smallest series" ∑ 𝑢𝑘 also converges.
(b) If the "smaller series” ∑ 𝑢𝑘 diverges, then the
"bigger series" ∑ 𝑣𝑘 also diverges.
Example 1.
Determine whether the series
1 1 1
(i) ∑∞ ∞ ∞
𝑘=1 (𝑘 𝑘 ); (ii) ∑𝑘=2 (ln 𝑘 ); (iii) ∑𝑘=1 (2𝑘 +𝑥) ∀𝑥 > 0.

converges or diverges.
Solution.
1 1
(i) For 𝑘 > 2, we have 𝑘 𝑘 > 2𝑘 . Hence < for 𝑘 > 2.
𝑘𝑘 2𝑘
1
Now, ∑∞
𝑘=1 is a geometric convergence series as its
2𝑘
1
common ratio = < 1 . Therefore, by comparison test,
2
1
∑∞
𝑘=1 (𝑘 𝑘 ) also converges.

- 285 -
 1 1
(ii) For 𝑘 ≥ 2, ln 𝑘 < 𝑘. Hence > for 𝑘 ≥ 2.
ln 𝑘 𝑘
1
Now, ∑∞
𝑘=1 is a harmonic divergence series. Therefore,
𝑘
1
by comparison test, ∑∞
𝑘=2 ( ) is also divergent.
ln 𝑘
1 1
(iii) For 𝑥 ≥ 0, 2𝑘 + 𝑥 > 2𝑘 . Hence < for 𝑥 > 0.
2𝑘 +𝑥 2𝑘
1
Now, ∑∞
𝑘=1 is a geometric convergence series as its
2𝑘
1
common ratio = < 1 . Therefore, by comparison test,
2
1
∑∞
𝑘=1 ( ) also converges. ◄
2𝑘 +𝑥

Example 2.
1 1
Test the convergence of the series ∑∞
𝑘=1 ( 2 + ).
𝑘 (𝑘+1)2

Solution.
1 1 1 1 1
(i) We have + < + < .
𝑘2 (𝑘+1)2 𝑘2 𝑘2 𝑘2
1
But, ∑∞
𝑘=1 is a convergent 2-series. Therefore, by
𝑘2
1 1
comparison test, we have ∑∞
𝑘=1 ( 2 + ) is a
𝑘 (𝑘+1)2

convergent series. ◄

- 286 -
Theorem 2 (D’ Alembert’s Ratio Test).
Let ∑ 𝑢𝑘 be a series with positive terms and suppose
𝑢𝑘+1
lim =𝜌
𝑘→+∞ 𝑢𝑘

(a) If 𝜌 < 1, the series converges.

(b) If 𝜌 > 1 or𝜌 = +∞ , the series diverges.
(c) If 𝜌 = 1, the series may converge or
diverge, so that another test must be tried.
Proof. The proof of (a) and (b) are omitted.
1 1
(c) The series ∑∞
𝑘=1 and ∑∞
𝑘=1 both have 𝜌= 1
𝑘 𝑘2

(verify). Since the first is the divergent harmonic series and

the second is a convergent p-series, the ratio test does not
distinguish between convergence and divergence when
𝜌= 1.■
Example 1.
1
The series ∑∞
𝑘=1 converges by the ration test since
𝑘!
1
𝑢𝑘+1 (𝑘+1)
𝜌 = lim = lim 1
𝑘→+∞ 𝑢𝑘 𝑘→+∞ 𝑘!
1
= lim =0
𝑘→+∞ 𝑘+1

so that 𝜌 < 1.◄

- 287 -
Example 2.
𝑘
The series ∑∞
𝑘=1 converges by the ration test since
2𝑘
𝑢𝑘+1
𝜌 = lim
𝑘→+∞ 𝑢𝑘

𝑘+1 2𝑘
= lim .
𝑘→+∞ 2𝑘+1 𝑘
1 𝑘+1
= lim
2 𝑘→+∞ 𝑘
1
= .
2

so that 𝜌 < 1.◄

Example 3.
𝑘𝑘
The series ∑∞
𝑘=1 𝑘! diverges by the ratio test since
𝑢𝑘+1
𝜌 = lim
𝑘→+∞ 𝑢𝑘

(𝑘+1)𝑘+1 𝑘!
= lim .
𝑘→+∞ (𝑘+1)! 𝑘𝑘

(𝑘+1)𝑘
= lim
𝑘→+∞ 𝑘𝑘

1 𝑘
= lim (1 + )
𝑘→+∞ 𝑘

= 𝑒.
Since 𝜌 = 𝑒 > 1, the series diverges. ◄

- 288 -
Example 4.
Determine whether the series
1 1 1 1
1+ + + +⋯+ +⋯
3 5 7 2𝑘−1
converges or diverges.
Solution.
The ratio test is of no help since
𝑢𝑘+1
𝜌 = lim
𝑘→+∞ 𝑢𝑘

1 2 𝑘−1
= lim .
𝑘→+∞ 2 (𝑘+1)−1 1
2 𝑘−1
= lim
𝑘→+∞ 2 𝑘+1

= 1.
However, the integral test proves that the series diverges
since
+∞ ℓ
𝑑𝑥 𝑑𝑥
∫ = lim ∫
2𝑥 − 1 ℓ→+∞ 2𝑥 − 1
1 1

1 ℓ
= lim ln (2𝑥 − 1)]
ℓ→+∞ 2 1

= +∞.◄

- 289 -
Example 5.
Determine whether the series
2! 4! 6! (2 𝑘)!
+ 2 + 3 + ⋯+ 𝑘 + ⋯
4 4 4 4
converges or diverges.
Solution.
(2 𝑘)! [2 (𝑘+1)]!
𝑢𝑘 = ⇒ 𝑢𝑘+1 = .
4𝑘 4 𝑘+1

Hence
𝑢𝑘+1
𝜌 = lim
𝑘→+∞ 𝑢𝑘
[2 (𝑘+1)]! 4𝑘
= lim . (2
𝑘→+∞ 4 𝑘+1 𝑘)!
(2 𝑘+2)! 1
= lim ( (2 𝑘)!
. )
𝑘→+∞ 4
1
= lim (2 𝑘 + 1)(2 𝑘 + 2)
4 𝑘→+∞

= +∞.
So, the ratio test proves that the series diverges. ◄

- 290 -
Example 6.
Determine whether the series
1 2 1∙2 2 1∙2∙3 2 1∙2∙3∙4 2
( ) +( ) +( ) +( ) +⋯
3 3∙5 3∙5∙7 3∙5∙7∙9
converges or diverges.
Solution.
2
1 ∙ 2 ∙ 3 ∙ 4 ∙∙∙ 𝑘
𝑢𝑘 = ( )
3 ∙ 5 ∙ 7 ∙ 9 ∙∙∙ (2𝑘 + 1)
2
1 ∙ 2 ∙ 3 ∙ 4 ∙∙∙ 𝑘 ∙ (𝑘 + 1)
⇒ 𝑢𝑘+1 =( )
3 ∙ 5 ∙ 7 ∙ 9 ∙∙∙ (2𝑘 + 1) ∙ (2𝑘 + 3)
𝑢𝑘+1
𝜌 = lim
𝑘→+∞ 𝑢𝑘

𝑘+1 2
= lim ( )
2𝑘+3
𝑘→+∞
1 2
1+
𝑘
= lim ( 3 )
𝑘→+∞ 2+𝑘

1 1
= = < 1.
22 4

Hence, by the Ratio test, the given series converges. ◄

- 291 -
Theorem 3 (Cauchy’s nth Root Test).
Let ∑ 𝑢𝑘 be a series with positive terms and suppose
𝜌 = lim 𝑘√𝑢𝑘 = lim (𝑢𝑘 )1/𝑘
𝑘→+∞ 𝑘→+∞

(a) If 𝜌 < 1, the series converges.

(b) If 𝜌 > 1 or 𝜌 = ∞, the series diverges.
(c) If 𝜌 = 1, the series may converge or diverge, so that
another test must be tried.
Example 1.
4𝑘−5 𝑘
The series ∑∞
𝑘=1 ( ) diverges by the root test since
2 𝑘+1
4 𝑘−5
𝜌 = lim 𝑢𝑘 1/𝑘 = lim = 2 > 1.◄
𝑘→+∞ 𝑘→+∞ 2 𝑘+1

Example 2.
1
The series ∑∞
𝑘=1 (ln converges by the root test, since
(𝑘+1)𝑘
1
lim (𝑢𝑘 )1/𝑘 = lim = 0 < 1.◄
𝑘→+∞ 𝑘→+∞ ln (𝑘+1)

Example 3.
1
The series ∑∞
𝑘=1 converges by the root test, since
𝑘𝑘
1
lim (𝑢𝑘 )1/𝑘 = lim = 0 < 1.◄
𝑘→+∞ 𝑘→+∞ k

- 292 -
Example 4.
2
∞ 𝑘 𝑘
The series ∑𝑘=1 ( ) converges by the root test, since
𝑘+1

𝑘 𝑘
lim (𝑢𝑘 )1/𝑘 = lim ( )
𝑘→+∞ 𝑘→+∞ 𝑘 + 1
𝑘
1 1
= lim ( 1 ) = 𝑒 < 1.◄
𝑘→+∞ 1+𝑘

Example 5.
𝑘
The series ∑∞
𝑘=1 5
−𝑘−(−1)
converges by the root test,
since
−𝑘−(−1)𝑘
lim (𝑢𝑘 )1/𝑘 = lim 5 𝑘
𝑘→+∞ 𝑘→+∞
(−1)𝑘
−(1+ 𝑘 ) 1
= lim 5 = 5−1 = < 1.◄
𝑘→+∞ 5

Remark.
Until now we have written most of our infinite series in the
form
∑∞
𝑘=1 𝑢𝑘 ($)
with the summation index beginning at 1. If summation
index begins at some other integer, it is always possible to
rewrite the series in form ($). Thus, for example, the series

- 293 -
 ∞
2𝑘 22 23
∑ =1+2+ + +⋯ ($$)
𝑘! 2! 3!
𝑘=0

can be written as
∞
2𝑘−1 22 23
∑ =1+2+ + +⋯ ($$$)
(𝑘 − 1)! 2! 3!
𝑘=1

However, for purpose of applying the convergence tests, it

is not necessary that the series have form ($). For example,
we can apply the ratio test to ($$) without converting to the
more complicated form ($$$). Doing so yields
𝑢𝑘+1
𝜌 = lim
𝑘→+∞ 𝑢𝑘

2𝑘+1 𝑘!
= lim .
𝑘→+∞ (𝑘+1)! 2𝑘
2
= lim =0
𝑘→+∞ 𝑘+1

which shows the series converges since 𝜌 < 1.◄

- 294 -
 Exercise Set (7.3)
In Exercise (1) - (6) apply the ratio test. According to the
test, does the series converge, does the series diverge, or
are the results inconclusive?
3𝑘
(1) ∑∞
𝑘=1 ; 𝑘!
4𝑘
(2) ∑∞
𝑘=1 ;
𝑘2
1
(3) ∑∞
𝑘=2 ;
5𝑘
𝑘!
(4) ∑∞
𝑘=1 ;
𝑘3
𝑘
(5) ∑∞
𝑘=1 .
𝑘 2 +1

In Exercise (6)-(9), apply the root test. According to the

test, does the series converge, does the series diverge, or
are the results inconclusive?
3 𝑘+2 𝑘
(6) ∑∞
𝑘=1 ( ) ;
2 𝑘−1

∞ 𝑘 𝑘
(7) ∑𝑘=1 ( ) ;
100
𝑘
(8) ∑∞
𝑘=1 𝑘 ; 5
1
(9) ∑∞
𝑘=1 .
𝑘2

- 295 -
In Exercise (10)-(31), use any appropriate test to
determine if the series converges, or not.
2𝑘 1
(10) ∑∞
𝑘=1 ; (11) ∑∞
𝑘=1 ;
𝑘3 𝑘2
7𝑘 1
(12) ∑∞ ∞
𝑘=0 ; (13) ∑𝑘=1 ;
𝑘! 2 𝑘+1
𝑘2 𝑘! 10𝑘
(14) ∑∞
𝑘=1 𝑘 ; (15)
∑∞
𝑘=1 ;
5 3𝑘
𝑘 2
(16) ∑∞ 50 −𝑘
𝑘=1 𝑘 𝑒 ; (17) ∑∞
𝑘=1 𝑘 3 +1;

2 𝑘
(18) ∑∞ ∞ 𝑘
𝑘=1 𝑘 (3) ; (19) ∑𝑘=1 𝑘 ;

1 2𝑘
(20) ∑∞
𝑘=2 ; (21) ∑∞
𝑘=1 ;
𝑘 ln 𝑘 𝑘 3 +1

4 𝑘 (𝑘 !)2 2𝑘
∞
∑𝑘=1 ( ∞
∑𝑘=1
(22)
7 𝑘−1
) ; (23) (2 𝑘+2)!
;
(𝑘 !)2 1
(24) ∑∞
𝑘=0 (2 ; (25) ∑∞
𝑘=1 ;
𝑘)! 𝑘 2 +25

1 𝑘 𝑘
(26) ∑∞
𝑘=1 1+√𝑘; (27) ∑∞
𝑘=1 𝑘 ! ;

ln 𝑘 𝑘!
(28) ∑∞
𝑘=1 ; (29) ∑∞
𝑘=1 2 ;
𝑒𝑘 𝑒𝑘
2
(𝑘+4)! 𝑘 𝑘
(30) ∑∞
𝑘=0 4 !𝑘 ! 4 𝑘 ; (31) ∞
∑𝑘=1 ( ) ;
𝑘+1

(32) Determine if the following series converges:

1 .2 1 .2 .3 1 .2 .3 .4
1+ + + +⋯
1 .3 1 .3 .5 1 .3 .5 .7

- 296 -
4. Applying the comparison test
Informal Principle 1.
Constants added or subtracted in the formula for 𝑢𝑘 may
usually be deleted without affecting the convergence or
divergence behavior of the series.
Example 1.
Guess whether the following series converge or diverge.
1 1 1
(a) ∑∞
𝑘=1 ; (b) ∑∞
𝑘=5 ; ∑∞
𝑘=1 1 3
.
2𝑘 +1 √𝑘−2 (𝑘+2)

Solution.
(a) Deleting the constant 1 suggest that
1 1
∑∞
𝑘=1 behaves like ∑∞
𝑘=1 𝑘 .
2𝑘 +1 2

The modified series is a convergent geometric series (ratio

1
< 1, so the given series is likely to converge.
2

(b) Deleting the −2 suggest that

1 1
∑∞
𝑘=5 behaves like ∑∞
𝑘=5 .
√𝑘−2 √𝑘
1
The modified series is a divergent p-series(𝑝 = ) , so the
2

given series is likely to diverge.

1
(c) Deleting the suggests that
2

- 297 -
 1 1
∑∞
𝑘=1 1 3
behaves like ∑∞
𝑘=1 .
(𝑘+2) 𝑘3

The modified series is a convergent p-series (𝑝 = 3), so

the given series is likely to converge. ◄
Informal Principle 2.
If a polynomial in 𝑘 appears as a factor in the numerator or
denominator of 𝑢𝑘 , all but the highest power of 𝑘 in the
polynomial may usually be deleted without affecting the
convergence or divergence behavior of the series.
Example 2.
Guess whether the following series converge or diverge.
1 6 𝑘 4 −2𝑘 3 +1
(a) ∑∞
𝑘=1 √𝑘 3 +2 𝑘; (b) ∞
∑𝑘=1 5 2
𝑘 +𝑘 −2 𝑘

Solution.
(a) Deleting the term 2𝑘 suggest that
1 1 1
∑∞
𝑘=1 behaves like ∑∞
𝑘=1 = ∑∞
𝑘=1
√𝑘 3 +2 𝑘 √𝑘 3 𝑘 3/2
2
Since the modified series is a convergent p-series (𝑝 = ),
3

the given series is likely to converge.

(b) Deleting all but the highest power of k in the
numerator and in the denominator suggests that

- 298 -
 6 𝑘 4 −2𝑘 3 +1 6 𝑘4 1
∑∞
𝑘=1 behaves like ∑∞
𝑘=1 = 6 ∑∞
𝑘=1 .
𝑘 5 +𝑘 2 −2 𝑘 𝑘5 𝑘

Since the modified series is a constant time the divergent

harmonic series, the given series is likely to diverge.
♠To prove ∑ 𝑢𝑘 converge by the comparison test, we must
find a convergent series ∑ 𝑣𝑘 such that
𝑢𝑘 ≤ 𝑣𝑘
for all k, frequently, 𝑣𝑘 is derived from the formula for 𝑢𝑘
by either increasing the numerator of 𝑢𝑘 , or decreasing the
denominator of 𝑢𝑘 , or both.
Example 3.
Use the comparison test to determine whether
∞
1
∑
2𝑘 2 + 𝑘
𝑘=1

converges or diverges.
Solution.
Using Principle 2, the series behaves like the series
1 1 1
∑∞
𝑘=1 = ∑∞
𝑘=1
2 𝑘2 2 𝑘2

which is a constant time a convergent p-series. Thus, the

given series is likely to converge. To prove the
convergence, observe that when we discard the k from the
- 299 -
 1
denominator of (2 𝑘 2 +𝑘)
, the denominator decreases and

the ratio increases, so that

1 1
<
2 𝑘 2 +𝑘 2 𝑘2

for 𝑘 = 1, 2, … since
1 1 1
∑∞
𝑘=1 = ∑∞
𝑘=1
2𝑘 2 2 𝑘2
1
converges, so does ∑∞
𝑘=1 by the comparison test.
2 𝑘 2 +𝑘

Example 4.
Use the comparison test to determine whether
1
∑∞
𝑘=1 2 𝑘 2 −𝑘

converges or diverges.
Solution.
Using Principle 2, the series behaves like the convergent
1 1 1
series ∑∞
𝑘=1 = ∑∞
𝑘=1 .
2 𝑘2 2 𝑘2

Thus, the given series is likely to converge. However, if

1
we discard the k from the denominator of (2 𝑘 2 −𝑘)
, the

denominator increases and the ratio decreases, so that

1 1
>
2 𝑘 2 −𝑘 2 𝑘2

- 300 -
Unfortunately, this inequality is in the wrong direction to
prove convergence of the given series. A different
approach is needed; we must do something to decrease the
denominator, not increase it. We accomplish this by
1 1 1
replacing k by 𝑘 2 to obtain ≤ = . Since
2𝑘 2 −𝑘 2 𝑘 2 −𝑘 2 𝑘2
1
∑∞
𝑘=1 is a convergent p-series, the given series
𝑘2

converges by the comparison test. ◄

♠To prove a series ∑ 𝑢𝑘 diverges by the comparison test,

we must produce a divergent series ∑ 𝑣𝑘 of positive terms

such that 𝑢𝑘 ≥ 𝑣𝑘 for all k.

Example 5.
Use the comparison test to determine whether
∞
1
∑ 1
𝑘=1 𝑘 − 4

converges or diverges.
Solution.
Using Principle 1, the series behaves like the divergent
1
harmonic series ∑∞
𝑘=1 . 𝑘

- 301 -
Thus, the given series is likely to diverge. Since
1 1
1 > for 𝑘 = 1 , 2 , …
𝑘−4 𝑘

1
and since ∑∞
𝑘=1 diverges, the given series diverges by
𝑘

the comparison test. ◄

Example 6.
1
Use the comparison test to determine whether ∑∞
𝑘=1 √𝑘+5

converges or diverges.
Solution.
Using Principle 1, the series behaves like the divergent p-
1
series ∑∞
𝑘=1 .
√𝑘

Thus, the given series is likely to diverge. For 𝑘 ≥ 25 we

1 1 1 1
have = ≥ and since ∑∞
𝑘=25 diverges
√𝑘+5 √25+√𝑘 2 √𝑘 2 √𝑘
1
(why?) the series ∑∞
𝑘=1 diverges by the comparison
√𝑘+5

test; consequently, the given series diverges by Theorem 3

(a) in Section 2.◄

- 302 -
Theorem 3 (Comparison Test (Limit Form Test)).
Let ∑ 𝑢𝑘 and ∑ 𝑣𝑘 be series with positive terms and
𝑢𝑘
suppose 𝜌 = lim . if 𝜌 is finite and 𝜌 > 0, then the
𝑘→+∞ 𝑣𝑘

series both converge, or both diverge.

Example 1.
Use the limit comparison test to determine if
3𝑘 3 −2𝑘 2 +4
∑∞
𝑘=1 converges or diverges.
𝑘 5 −𝑘 3 +2

Solution.
From Principle 2, the series behave like
3𝑘 3 3
∑∞
𝑘=1 𝑘 5 = ∑∞
𝑘=1 (*)
𝑘2

which converges since it is a constant time a convergent p-

series (𝑝 = 2). We apply the limit comparison test to series
(*) and the given series. We obtain
3 𝑘3 −2 𝑘2 +4
𝑘5 −𝑘3 +2 3 𝑘 5 −2𝑘 4 +4𝑘 2
𝜌 = lim 3 = lim = 1.
𝑘→+∞ 𝑘→+∞ 3 𝑘 5 −3𝑘 3 +6
𝑘2

Since 𝜌 > 0, the given series converges because series

(*) converges. ◄

- 303 -
 Exercise Set (7.4)
In Exercise (1)-(6) prove that the series converges by the
comparison test.
1 2 1
(1) ∑∞
𝑘=1 ; (2) ∑∞
𝑘=1 ; (3) ∑∞
𝑘=1 ;
3𝑘 +5 𝑘 4 +𝑘 5𝑘 2 −𝑘
𝑘 2𝑘 −1 5 sin2 𝑘
(4) ∑∞
𝑘=1 8 𝑘 2 +𝑘 2 −1; (5) ∞
∑𝑘=1 𝑘 ; (6) ∞
∑𝑘=1 .
3 +2 𝑘 𝑘!

In Exercises (7)-(12), prove that the series diverges by the

comparison test
3 1 9
(7) ∑∞
𝑘=1 1 ; (8) ∑∞
𝑘=1 ; (9) ∑∞
𝑘=1 ;
𝑘−4 √𝑘+8 √𝑘+1
𝑘+1 𝑘 4/3 𝑘 −1/2
(10) ∑∞
𝑘=1 𝑘 2 −𝑘; (11) ∑∞
𝑘=1 8 𝑘 2 +5 𝑘+1; (12) ∞
∑𝑘=1 .
2+sin2 𝑘
In exercises (13)-(18), use the limit comparison test to
determine if the series converges or diverges.
4𝑘 2 −2 𝑘+6 1
(13) ∑∞
𝑘=1 ; (14) ∑∞
𝑘=1 ;
8 𝑘 7 +𝑘−8 9 𝑘+6
5 𝑘 (𝑘+3)
(15) ∑∞
𝑘=1 ; (16) ∑∞
𝑘=1 (𝑘+1)(𝑘+2)(𝑘+5);
3𝑘 +1
1 1
(17) ∑∞
𝑘=1 3 ; (18) ∑∞
𝑘=1 (2 𝑘+3)17
.
√8 𝑘 2 −3𝑘

In Exercise (19)-(33), use any method to determine if the

series converges or diverges. In some cases, you may have
to use tests from earlier sections.
1 1
(19) ∑∞
𝑘=1 ; (20) ∑∞
𝑘=1 (3+𝑘)2/5 ;
𝑘 3 +2 𝑘+1
1 ln 𝑘
(21) ∑∞
𝑘=1 ; (22) ∑∞
𝑘=1 ;
9 𝑘−2 𝑘
- 304 -
 √𝑘 4
(23) ∑∞
𝑘=1 ; (24) ∑∞
𝑘=1 ;
𝑘 2 +1 2+𝑘3𝑘
1 2+(−1)𝑘
(25) ∑∞
𝑘=1 ; (26) ∑∞
𝑘=1 ;
√𝑘(𝑘+1) 5𝑘

2+√𝑘 4+|cos 𝑘|
(27) ∑∞ ∞
𝑘=1 (𝑘+1)3 ; (28) ∑𝑘=1 ;
−1 𝑘3

1 √𝑘ln 𝑘
(29) ∑∞
𝑘=1 ; (30) ∑∞
𝑘=1 ;
4+2𝑘 𝑘 3 +1

∞ tan−1 𝑘 ∞ 5𝑘 +𝑘
(31) ∑𝑘=1 2 ; (32) ∑𝑘=1 ;
𝑘 𝑘 !+3
ln 𝑘
(33) ∑∞
𝑘=1 ;
𝑘 √𝑘

(34) Use the limit comparison test to show that

𝜋
∑∞
k=1 sin (𝑘 ) diverges.

𝜋
[Hint: Compare with the series∑∞
𝑘=1 ]] 𝑘
ln 𝑘
(35) Use the comparison test to determine if ∑∞
𝑘=1 𝑘2

converges or diverges.
1
(36) Determine if ∑∞
𝑘=1 (In converges or diverges.
𝑘)2

(37) Let a, b and p be positive constants. For which values

1
of p does the series ∑∞
𝑘=1 (𝑎+𝑏 converge?
𝑘)𝑝

- 305 -
(38) Show that 𝑘 𝑘 ≥ 𝑘! and use this result to prove that the
series ∑∞
𝑘=1 𝑘
−𝑘
converges by the comparison test. Prove
convergence using the root test.
(39) Use the limit comparison test to investigation
(𝑘+1)2
convergence of ∑∞
𝑘=1 (𝑘+2)! .

(40) Use the limit comparison test to investigate

1 1 1
convergence of the series 1 + + + + ⋯
3 5 7
1
(41) Prove that ∑∞
𝑘=1 converges by comparison with a
𝑘!

suitable geometric series.

(42) Let ∑ 𝑢𝑘 and ∑ 𝑣𝑘 be series with positive terms
Prove:
𝑢
(a) If lim ( 𝑘 ) = 0 and ∑ 𝑣k converges, then ∑ 𝑢𝑘
k→+∞ 𝑣 𝑘

converges.
𝑢
(b) If lim ( 𝑘 ) = ∞ and ∑ 𝑣𝑘 diverges, then ∑ 𝑢𝑘
k→+∞ 𝑣 𝑘

diverges.

- 306 -
5. ALTERNATING SERIES; CONDITIONAL
CONVERGENCE
In this section we discuss series containing negative terms.
Of special importance are series whose terms are
alternately positive and negative. These are called
alternating series. Such series have one of two possible
forms:
𝑢1 − 𝑢2 + 𝑢3 − 𝑢4 + ⋯ + (−1)𝑘+1 𝑢𝑘 + ⋯
or
−𝑢1 + 𝑢2 − 𝑢3 + 𝑢4 − ⋯ + (−1)𝑘 𝑢𝑘 + ⋯
where the 𝑢𝑘 ′𝑠 are all positive.
The following theorem is the key result on convergence of
alternating series.
Theorem 1 (Alternating Series Test or Leibnitz’s Test).
An alternating series
𝑢1 − 𝑢2 + 𝑢3 − 𝑢4 + ⋯ + (−1)𝑘+1 𝑢𝑘 + ⋯
or
−𝑢1 + 𝑢2 − 𝑢3 + 𝑢4 − ⋯ + (−1)𝑘 𝑢𝑘 + ⋯
converges if the following two conditions are satisfied:
(a) 𝑢1 ≥ 𝑢2 ≥ 𝑢2 ≥ ⋯ ≥ 𝑢𝑘 ≥ ⋯ and

- 307 -
(b) lim 𝑢𝑘 = 0.
𝑘→+∞

Fact: If the even-numbered terms of a sequence tend

toward a limit L, and if the odd-numbered terms of the
sequence tend toward the same limit L, then the entire
sequence tend toward the limit L.
Example 1.
Test the convergence of the following series!
1 1 1 1
(i) 1 − + − + ⋯ + (−1)𝑘+1 + ⋯;
2 3 4 𝑘
1 1 1 1
(ii) 1 − + − + ⋯ + (−1)𝑘+1 + ⋯.
22 32 42 𝑘2

Solution.
1 1 1 1
(i) The series 1 − + − + ⋯ + (−1)𝑘+1 + ⋯ is
2 3 4 𝑘

called the alternating harmonic series.

Since
1 1
𝑢𝑘 = > = 𝑢𝑘+1
𝑘 𝑘+1
and
1
lim 𝑢𝑘 = lim =0
𝑘→+∞ 𝑘→+∞ 𝑘

this series converges by the alternating series test.

(i) Since

- 308 -
 1 1
𝑢𝑘 = > = 𝑢𝑘+1
𝑘 2 (𝑘 + 1)2
and
1
lim 𝑢𝑘 = lim =0
𝑘→+∞ 𝑘→+∞ 𝑘 2

Hence by Leibnitz’s test, the given series converges. ◄

Example 2.
Test the convergence of the following series.
∞
𝑘+3
∑ (−1)𝑘+1
𝑘(𝑘 + 1)
𝑘=1

Solution.
Requirement (b) of the alternating series test is satisfied
since
1 3
𝑘+3 + 2
𝑘 𝑘
lim 𝑢𝑘 = lim = lim =0
𝑘→+∞ 𝑘→+∞ 𝑘(𝑘 + 1) 𝑘→+∞ 1 + 1
𝑘

To see if requirement (a) is met, we must determine if the

sequence
𝑘 + 3 +∞
{𝑢𝑘 }+∞
𝑘=1 ={ }
𝑘 (𝑘 + 1) 𝑘=1
is nonincreasing sequence. Since

- 309 -
 𝑢𝑘+1 𝑘+4 𝑘(𝑘 + 1) 𝑘 2 + 4𝑘
= ∙ = 2
𝑢𝑘 (𝑘 + 1)(𝑘 + 2) 𝑘 + 3 𝑘 +5𝑘+6
𝑘 2 +4 𝑘
= (𝑘 2 < 1.
+4 𝑘)+(𝑘+6)

We have 𝑢𝑘 > 𝑢𝑘+1 the series converges by the

alternating series test. ◄
Definition 3 (Absolute Convergence).
A series ∑∞
𝑘=1 𝑢𝑘 = 𝑢1 + 𝑢2 + ⋯ + 𝑢𝑘 + ⋯ is said to

converge absolutely if the series of absolute values

∞

∑ |𝑢𝑘 | = |𝑢1 | + |𝑢2 | + ⋯ + |𝑢𝑘 | + ⋯

𝑘=1

converges.
Example 4.
The series
1 1 1 1 1 1
1− − 2+ 3+ 4− 5− 6+⋯
2 2 2 2 2 2
converges absolutely since the series of absolute values
1 1 1 1 1 1
1+ + 2+ 3+ 4+ 5+ 6+⋯
2 2 2 2 2 2
is a convergent series (Since it is a geometric series whose
1
common ratio is < 1).
2

- 310 -
On the other hand, the alternating harmonic series
1 1 1 1
1− + − + −⋯
2 3 4 5
does not converge absolutely since the series of absolute
values
1 1 1 1
1+ + + + +⋯
2 3 4 5
diverges. ◄
Theorem 4.
If the series
∞

∑ |𝑢𝑘 | = |𝑢1 | + |𝑢2 | + ⋯ + |𝑢𝑘 | + ⋯

𝑘=1

converges, then so does the series

∞

∑ 𝑢𝑘 = 𝑢1 + 𝑢2 + ⋯ + 𝑢𝑘 + ⋯
𝑘=1

In other words, if a series converges absolutely, then it

converges.
Example 5.
In Example 4, we showed that
1 1 1 1 1 1
1− − 2+ 3+ 4− 5− 6+⋯
2 2 2 2 2 2

- 311 -
converges absolutely. It follows from Theorem 4 above
that the series converges. ◄
Example 6.
cos 𝑘
Show that the series ∑∞
𝑘=1 converges.
𝑘2

Solution.
cos 𝑘 1 cos 𝑘
Since |cos 𝑘| ≤ 1 for all k, | |≤ . Thus ∑∞
𝑘=1 | |
𝑘2 𝑘2 𝑘2

converges by the comparison test, and consequently

cos 𝑘
∑∞
𝑘=1 converges absolutely and so it converges. ◄
𝑘2

♠If ∑|𝑢𝑘 | diverges, no conclusion can be drawn about the

convergence or divergence of ∑ 𝑢𝑘 . For example,
consider the two series
1 1 1 1
1− + − + ⋯ + (−1)𝑘+1 + ⋯ (∗)
2 3 4 𝑘
1 1 1 1
−1 − − − − ⋯ − − ⋯ (∗∗)
2 3 4 𝑘
Series (*), the alternating harmonic series, converges;
while series (**), being a constant time the harmonic
series, diverges. Yet in each case the series of absolute
1 1 1
values is 1 + + + ⋯ + + ⋯
2 3 𝑘

which diverges.

- 312 -
 The following version of the ratio test is useful for
investigating absolute convergence.
Theorem 5. (The Ratio Test for Absolute Convergence).
Let ∑ 𝑢𝑘 be a series with nonzero terms and suppose
|𝑢𝑘+1 |
lim =𝜌
𝑘→+∞ |𝑢𝑘 |

(a) If 𝜌 < 1, the series ∑ 𝑢𝑘 converges absolutely.

(b) If 𝜌 > 1 or if 𝜌 = +∞, then the series ∑ 𝑢𝑘 diverges.
(c) If 𝜌 = 1, no conclusion about converges can be drawn.
Example 7.
2𝑘
The series ∑∞
𝑘=1(−1)
𝑘
converges absolutely and so
𝑘!

converges since
|𝑢𝑘+1 | 2𝑘+1 𝑘!
𝜌 = lim = lim . 𝑘
𝑘→+∞ |𝑢𝑘 | 𝑘→+∞ (𝑘 + 1)! 2

2
= lim = 0 < 1.◄
𝑘→+∞ 𝑘+1

Example 8.
We proved earlier (Theorem 3 in Section 1 in this Chapter)
that a geometric series
𝑎 + 𝑎𝑟 + 𝑎𝑟 2 + ⋯ + 𝑎𝑟 𝑘−1 + ⋯

- 313 -
converges if |𝑟| < 1 and diverges if |𝑟| ≥ 1. However, a
stronger statement can be made - the series converges
absolutely if |𝑟| < 1. This follows from Theorem 5 in
Section 4 above since
|𝑢𝑘+1 | |𝑎𝑟 𝑘 |
𝜌 = lim = lim = lim |𝑟| = |𝑟|
𝑘→+∞ |𝑢𝑘 | 𝑘→+∞ |𝑎𝑟 𝑘−1 | 𝑘→+∞

so that 𝜌 < 1 if |𝑟| < 1.◄

Result: Every absolutely convergent series is convergent.
But the converse may not be true.
Definition 4 (Conditional Convergence)
A series which is convergent but not absolutely convergent
is called conditionally convergent series.
Example 9.
Test the convergence and absolute convergence of the
following series:
1 1 1 1
(i) 1 − + − + ⋯ + (−1)𝑘+1 + ⋯
2 3 4 𝑘
1 1 1 1
(ii) 1 − + − + ⋯ + (−1)𝑘+1 +⋯
22 32 42 𝑘2
1
(iii) ∑∞
𝑘=2(−1)
𝑘+1
.
log 𝑘

- 314 -
Solution.
1
(i) Since ∑∞
𝑘=1(−1)
𝑘+1
is convergent by Leibnitz’s test.
𝑘
1
Now, ∑∞ ∞
𝑘=1|𝑢𝑘 | = ∑𝑘=1 𝑘 is not convergent (As 𝑝 = 1)
Hence the given series is not absolutely convergent. This
is an example of conditionally convergent series.
1
(ii) The given series ∑∞
𝑘=1(−1)
𝑘+1
is convergent by
𝑘2

Leibnitz’s test.
1
Also, ∑∞ ∞
𝑘=1|𝑢𝑘 | = ∑𝑘=1 is convergent (𝑝 = 2 > 1).
𝑘2

Hence the given series is absolutely convergent.

(iii) The given series ∑∞
𝑘=2(−1)
𝑘+1
𝑢𝑘 .
1
Here 𝑢𝑘 = .
log 𝑘

Now, log 𝑥 is an increasing function ∀𝑥 > 0.

∴ log(𝑘 + 2) > log(𝑘 + 1)
1 1
∴ <
log(𝑘 + 2) log(𝑘 + 1)
∴ 𝑎𝑘+1 < 𝑎𝑘
1
Also, lim 𝑢𝑘 = lim = 0.
𝑘→+∞ 𝑘→+∞ log 𝑘

Hence by Leibnitz’s test, the given series is convergent.

Now, for absolute convergence, consider ∑∞
𝑘=2|𝑢𝑘 | =

- 315 -
 1
∑∞
𝑘=2 . It is a divergent series (by comparison test
log 𝑘
1
with the harmonic series ∑∞
𝑘=2 ). 𝑘

Hence the given series is not absolutely convergent. This

is an example of conditionally convergent series. ◄
Example 10.
Test the convergence of the series:
1 1
(i) ∑∞
𝑘=1(−1)
𝑘−1
[ 2 + (𝑘+1)2 ];
𝑘
1
(ii) ∑∞
𝑘=1(−1)
𝑘−1
.
𝑘2𝑘

Solution.
(i) The given series ∑∞
𝑘=1(−1)
𝑘−1
𝑢𝑘 .
1 1
Consider ∑∞ ∞
𝑘=1|𝑢𝑘 | = ∑𝑘=1 2 + (𝑘+1)2 .
𝑘
1 1 1 1 2
Now, + (𝑘+1)2 < + =
𝑘2 𝑘2 𝑘2 𝑘2
1
Since ∑∞
𝑘=1 is convergent p-series (As 𝑝 = 2 > 1).
𝑘2

Then by the comparison test ∑∞

𝑘=1|𝑢𝑘 | is also convergent.

Hence the given series is absolutely convergent and so

convergent.
(ii) The given series is ∑∞
𝑘=2(−1)
𝑘−1
𝑢𝑘 .
1
Consider ∑∞ ∞
𝑘=1|𝑢𝑘 | = ∑𝑘=1 .
𝑘2𝑘

- 316 -
 1
Here |𝑢𝑘 | = .
𝑘2𝑘
|𝑢𝑘+1 | 𝑘2𝑘 1
Now, lim = lim = < 1.
𝑘→+∞ |𝑢𝑘 | 𝑘→+∞ (𝑘+1)2𝑘+1 2

Hence, by Ratio test ∑∞

𝑘=1|𝑢𝑘 | is convergent or the given

series is absolutely convergent and hence convergent. ◄

Example 11.
Find the values of 𝑥 for which the series
𝑥3 𝑥5
𝑥− + −⋯
3 5
is absolutely convergent and conditionally convergent.
Solution.
The given series
𝑥 2𝑘−1
∑∞
𝑘=1(−1)
𝑘−1
= ∑∞
𝑘=1(−1)
𝑘−1
𝑎𝑘 .
2𝑘−1
𝑥 2𝑘−1
Then |𝑢𝑘 | = | |.
2𝑘−1
|𝑢𝑘+1 | 𝑥 2𝑘+1 2𝑘−1
Now, lim = lim | ∙ | = |𝑥 2 | = 𝑥 2 .
𝑘→+∞ |𝑢𝑘 | 𝑘→+∞ 2𝑘+1 𝑥 2𝑘−1

Thus , by Ratio test ∑∞ 2

𝑘=1|𝑢𝑘 | converges if 𝑥 < 1, i.e.,

|𝑥 | < 1, diverges if |𝑥 | > 1 and test fails if |𝑥 | = 1.

When |𝑥 | = 1, i.e., 𝑥 = 1 or 𝑥 = −1.
For 𝑥 = 1, the given series is

- 317 -
 1 1
1− + −⋯
3 5
which is convergent by Leibnitz’s test but not absolutely
convergent.
For 𝑥 = −1, the given series is
1 1
−1 + − +⋯
3 5
which is also convergent by Leibnitz’s test but not
absolutely convergent.
So, the given series is absolutely convergent for |𝑥 | < 1
and conditionally convergent for |𝑥 | = 1. ◄

- 318 -
 Review of Convergence Tests
NAME STATEMENT COMMENTS
Divergence test If lim 𝑢𝑘 ≠ 0 or lim 𝑢𝑘 does not If lim 𝑢𝑘 = 0, then
𝑘→+∞ 𝑘→+∞ 𝑘→+∞

exist, then ∑ 𝑢𝑘 diverges ∑ 𝑢𝑘 may or may not

converge.
Let ∑ 𝑢𝑘 be a series with positive Use this test when
terms and let 𝑓(𝑥) be the function 𝑓(𝑥) is easy to
that results when k is replaced by x integrate.
in the formula for 𝑢𝑘 . If f is
decreasing and continuous for 𝑥 ≥
+∞
1, then ∑∞
𝑘=1 𝑢𝑘 and ∫1 𝑓(𝑥)𝑑𝑥

both converges or both diverges.

Comparison test Let ∑ 𝑢𝑘 and ∑ 𝑣𝑘 be series with Use this test as a last
positive terms such that 𝑢𝑘 ≤ 𝑣𝑘 . reason easier to apply.
If ∑ 𝑣𝑘 converges, ∑ 𝑢𝑘 converges,
and if ∑ 𝑢𝑘 diverges, ∑ 𝑣𝑘 diverges.

- 319 -
 NAME STATEMENT COMMENTS
Root test Let be ∑ 𝑢𝑘 a series with positive Try this test when 𝑢𝑘

terms such that 𝜌 = lim 𝑘√𝑢𝑘 involves kth power.

𝑘→+∞

Series converges if 𝜌 < 1.

Series diverges if 𝜌 > 1 or 𝜌 =
+∞
No conclusion if 𝜌 = 1.
Limit Let ∑ 𝑢𝑘 and ∑ 𝑣𝑘 be series with This is easier to apply
comparison test positive terms such that than the comparison
𝑢𝑘 test, but still requires
𝜌 = lim
𝑘→+∞ 𝑣𝑘
some still in choosing
if 0 < 𝜌 < +∞, then both series
the series ∑ 𝑏𝑘 for
converges, or both diverges.
comparison.

Alternating The series 𝑢1 − 𝑢2 + 𝑢3 − 𝑢4 + ⋯ This test applies only

series test and −𝑢1 + 𝑢2 − 𝑢2 + 𝑢4 − ⋯ for alternating series.
converges if 𝑢1 ≥ 𝑢2 ≥ 𝑢3 ≥ ⋯
and lim 𝑢𝑘 = 0.
𝑘→+∞

Ratio test for Let ∑ 𝑢𝑘 be a series with nonzero The series need not
absolute |𝑢𝑘+1 | have positive terms
terms such that 𝜌 = lim
𝑘→+∞ |𝑢𝑘 |
convergence and need not be
(a) Series converges absolutely
alternating to use this
if 𝜌 < 1
test
(b) Series diverges if 𝜌 > 1 or
𝜌 = +∞
(c ) No conclusion if 𝜌 = 1.

- 320 -
 Exercise Set (7.5)
In Exercise (1)-(6), use the alternating series test to
determine if the series converges or diverges
(−1)𝑘+1
(1) ∑∞
𝑘=1 ;
2 𝑘+1
𝑘
(2) ∑∞
𝑘=1(−1)
𝑘+1
;
3𝑘
𝑘+1
(3) ∑∞
𝑘=1(−1)
𝑘+1
;
3 𝑘+1
𝑘+4
(4) ∑∞
𝑘=1(−1)
𝑘+1
;
𝑘 2 +𝑘

(5) ∑∞
𝑘=1(−1)
𝑘+1 −𝑘
𝑒 ;
ln 𝑘
(6) ∑∞
𝑘=3(−1)
𝑘
;
𝑘

In Exercise (7)-(12), use ratio test for absolute

convergence to determine if the series converges
absolutely or diverges.
3 𝑘
(7) ∑∞
𝑘=1 (− 5) ;

∞ 𝑘+1 2𝑘
(8) ∑𝑘=1(−1) ;
𝑘!

∞ 𝑘+1 3𝑘
(9) ∑𝑘=1(−1) ;
𝑘2
𝑘
(10) ∑∞ 𝑘
𝑘=1(−1) (5𝑘 );

𝑘3
(11) ∑∞ 𝑘
𝑘=1(−1) (𝑒 𝑘 ) ;

𝑘𝑘
(12) ∑∞
𝑘=1(−1)
𝑘+1
;
𝑘!

- 321 -
In Exercise (13)-(30), classify the series as: absolutely
convergent, conditionally convergent, or divergent.
(−1)𝑘+1
(13) ∑∞
𝑘=1 ;
3𝑘
(−1)𝑘+1
(14) ∑∞
𝑘=1 ;
𝑘 4/3
(−4)𝑘
(15) ∑∞
𝑘=1 ;
𝑘2
(−1)𝑘+1
(16) ∑∞
𝑘=1 ;
𝑘!
cos 𝑘𝜋
(17) ∑∞
𝑘=1 ;
𝑘
(−1)𝑘 ln 𝑘
(18) ∑∞
𝑘=3 ;
𝑘
𝑘+2
(19) ∑∞
𝑘=1(−1)
𝑘+1
( );
3 𝑘−1
(−1)𝑘+1
(20) ∑∞
𝑘=1 ;
𝑘 2 +1
𝑘+2
(21) ∑∞
𝑘=1(−1)
𝑘+1
;
𝑘 (𝑘+3)

(−1) 𝑘+1 2
𝑘
(22) ∑∞
𝑘=1 𝑘 3 +1 ;
𝑘𝜋
(23) ∑∞
𝑘=1 sin ;
𝑘3
sin 𝑘
(24) ∑∞
𝑘=1 ;
𝑘3
(−1)𝑘
(25) ∑∞
𝑘=2 ;
𝑘 ln 𝑘

- 322 -
 (−1)𝑘
(26) ∑∞
𝑘=1 ;
√𝑘 (𝑘+1)

1
(27) ∑∞
𝑘=2 (− );
ln 𝑘

∞ (−1)𝑘+1
(28) ∑𝑘=1 ;
√𝑘+1+√𝑘
(−1)𝑘 (𝑘 2 +1)
(29) ∑∞
𝑘=2 ;
𝑘 3 +2
𝑘 cos 𝑘𝜋
(30) ∑∞
𝑘=1 .
𝑘 2 +1

(31) Prove: If ∑ 𝑢𝑘 converges absolutely, then ∑ 𝑢𝑘2

converges.

- 323 -
6. POWER SERIES
If 𝑐0 , 𝑐1 , 𝑐2 , … are constants and x is a variable, then a
series of the form
∞

∑ 𝑐𝑘 𝑥 𝑘 = 𝑐0 + 𝑐1 𝑥 + 𝑐2 𝑥 2 + ⋯ + 𝑐𝑘 𝑥 𝑘 + ⋯
𝑘=0

is called a power series in 𝑥 (centered at 𝑥 = 0). Some

examples are:
∞

∑ 𝑥𝑘 = 1 + 𝑥 + 𝑥2 + 𝑥3 + ⋯
𝑘=0
∞
𝑥𝑘 𝑥2 𝑥3
∑ =1+𝑥+ + +⋯
𝑘! 2! 3!
𝑘=0
∞
𝑥 𝑘+1 𝑥2 𝑥3 𝑥4
∑ (−1 )𝑘 =𝑥− + −
𝑘+1 2 3 4
𝑘=0
∞
𝑥 2𝑘 𝑥2 𝑥4 𝑥6
∑ (−1 )𝑘 =1− + − +⋯
(2𝑘) ! 2! 4! 6!
𝑘=0
∞
𝑥 2𝑘+1 𝑥3 𝑥5 𝑥7
∑ (−1 )𝑘 =𝑥− + − +⋯
(2𝑘 + 1) ! 3! 5! 7!
𝑘=0

A fundamental problem. For what values of x does a given

power series, ∑ 𝑐𝑘 𝑥 𝑘 , converge?

- 324 -
Theorem 1.
For any power series in x, exactly of the following is true:
(a) The series converges only for 𝑥 = 0.
(b) The series converges absolutely for all x.
(c)The series converges absolutely for all x in some finite
open interval (−𝑅, 𝑅 ), and diverges if 𝑥 < −𝑅 or 𝑥 >
𝑅 . At 𝑥 = 𝑅 and 𝑥 = −𝑅 the series may converge
absolutely, converge conditionally, or diverge, depending
on the series.

In case (c), where the power series converges absolutely

for |𝑥 | < 𝑅 and diverges for |𝑥 | > 𝑅, we call 𝑅 the
radius of convergence. In case (a), where the series
converges only for 𝑥 = 0, we define the radius of
convergence to be 𝑅 = 0; and in case (b), where the
series converges absolutely for all 𝑥, we define the radius
of convergence to be 𝑅 = −∞.
The set of all values of 𝑥 for which a power series
converges is called the interval of convergence.

- 325 -
Example 1.
Find the interval of convergence and radius of
convergence of the power series
∞

∑ 𝑥𝑘 = 1 + 𝑥 + 𝑥2 + ⋯ + 𝑥𝑘 + ⋯
𝑘=0

Solution.
For every 𝑥, the given series is a geometric series with
ratio = 𝑥. By Example 8 of Section 5 in this chapter, the
series converges absolutely if |𝑥 | < 1, i.e., −1 < 𝑥 < 1
and diverges if |𝑥 | ≥ 1. So, the interval of convergence is
(−1, 1) and the radius of convergence is 𝑅 = 1.
Example 2.
Find the interval of convergence and radius of
𝑥𝑘
convergence of ∑∞
𝑘=0 .
𝑘!

Solution.
We will apply the ratio test for absolute convergence. for
every real number 𝑥,
𝑢𝑘+1
𝜌 = lim | |
𝑘→+∞ 𝑢𝑘

- 326 -
 𝑥 𝑘+1 𝑘!
= lim | ∙ 𝑘|
𝑘→+∞ (𝑘 + 1) ! 𝑥

𝑥
= lim | |=0
𝑘→+∞ 𝑘+1
Since 𝜌 < 1 for all 𝑥, the series converges absolutely for
all 𝑥.
Thus, the interval of convergence is (−∞ , +∞) and the
radius of convergence is 𝑅 = +∞ .◄
REMARK.
There is a useful byproduct of Example 2.
Since
∞
𝑥𝑘
∑
𝑘!
𝑘=0

converges for all 𝑥, then for all values of x

𝑥𝑘
lim
𝑘→+∞ 𝑘 !

We will need this result later.

- 327 -
Example 3.
Find the interval and radius of converges of
∞

∑ 𝑘 ! 𝑥𝑘
𝑘=0

Solution.
If 𝑥 = 0 , the series has only one nonzero term and
therefore converges. If 𝑥 ≠ 0, the ratio test yields
𝑢𝑘+1
𝜌 = lim | |
𝑘→+∞ 𝑢𝑘

(𝑘+1) ! 𝑥 𝑘+1
= lim | |
𝑘→+∞ 𝑘 ! 𝑥𝑘

= lim |(𝑘 + 1) 𝑥 | = +∞
𝑘→+∞

Therefore, the series converges if 𝑥 = 0, but diverges for

all other x.
Consequently, the interval of convergence is the single
point 𝑥 = 0 and the radius of convergence is 𝑅 = 0.◄
Example 4.
Find the interval of convergence and radius of
convergence of
∞
(−1)𝑘 𝑥 𝑘
∑ 𝑘
3 (𝑘 + 1)
𝑘=0

- 328 -
Solution.
Since |(−1)𝑘 | = |(−1)𝑘+1 | = 1, we obtain
𝑢𝑘+1
𝜌 = lim | |
𝑘→+∞ 𝑢𝑘
𝑥 +1 3𝑘 (𝑘+1)
= lim | ∙ |
𝑘→+∞ 3𝑘+1 (𝑘+2) 𝑥𝑘
|𝑥| 𝑘+1
= lim [ ∙ ( )]
𝑘→+∞ 3 𝑘+2
|𝑥| 1+1/𝑘
= lim ( )
3 𝑘→+∞ 1+2/𝑘
|𝑥|
= .
3

The ratio test for absolute convergence implies that the

series converges absolutely if |𝑥 | < 3 and diverges if
|𝑥 | > 3. The ratio test fails when |𝑥 | = 3.
So, the cases 𝑥 = −3 and 𝑥 = 3 need separate
analyses.
Substituting 𝑥 = −3 in the given series yields
∞ ∞ ∞
(−1)𝑘 (−3)𝑘 (−1)𝑘 (−1)𝑘 3𝑘 1
∑ 𝑘 =∑ = ∑
3 (𝑘 + 1) 3𝑘 (𝑘 + 1) 𝑘+1
𝑘=0 𝑘=0 𝑘=0

which is the divergent harmonic series

1 1 1
1 + + + + ⋯.
2 3 4

- 329 -
Substituting 𝑥 = 3 in the given series yields
∞ ∞
(−1)𝑘 3𝑘 (−1)𝑘
∑ 𝑘 =∑
3 (𝑘 + 1) 𝑘+1
𝑘=0 𝑘=0

which is the conditionally convergent alternating harmonic

series
1 1 1
1− + + −⋯
2 3 4
Thus, the interval of convergence for the given series is
(−3, 3] and the radius of convergence is 𝑅 = 3.◄

- 330 -
Example 5.
Find the interval and radius of convergence of the series
∞
𝑥 2𝑘
∑ (−1 )𝑘
(2 𝑘)!
𝑘=0

Solution.
Since|(−1)𝑘 | = |(−1)𝑘+1 | = 1, we have
𝑢𝑘+1 𝑥 2(𝑘+1) (2 𝑘) !
𝜌 = lim | | = lim | . 2𝑘 |
𝑘→+∞ 𝑢𝑘 𝑘→+∞ [2 (𝑘 + 1)] ! 𝑥
𝑥 2 𝑘+2 (2 𝑘) !
= lim | |
𝑘→+∞ (2 𝑘 + 2) ! 𝑥 2 𝑘

𝑥2
= lim | |
𝑘→+∞ (2 𝑘 + 2)(2 𝑘 + 1)

1
= 𝑥 2 lim
𝑘→+∞ (2 𝑘 + 2)(2 𝑘 + 1)

= 𝑥 2 .0 = 0.
Thus 𝜌 < 1 for all x, which means that the interval of
convergence is (−∞ , +∞) and the radius of convergence
is 𝑅 = +∞.◄

- 331 -
Theorem 2.
For a power series ∑ 𝑐𝑘 (𝑥 − 𝑎)𝑘 (centered at 𝑥 = 𝑎 ),
exactly one of the following is true:
(a) The series converges only for 𝑥 = 𝑎;
(b) The series converges absolutely for all 𝑥;
(c) The series converges absolutely for all 𝑥 in some finite
open interval (𝑎 − 𝑅, 𝑎 + 𝑅) and diverges if
𝑥 < 𝑎 − 𝑅 or 𝑥 > 𝑎 + 𝑅.
At the points 𝑥 = 𝑎 − 𝑅 and 𝑥 = 𝑎 + 𝑅 , the series
may converge absolutely, converge conditionally, or
diverge, depending on the series.
In case (a), (b) and (c) the series is said to have radius of
convergence, 0, +∞, and 𝑅 respectively.
The set of all values of 𝑥 for which the series converges is
called the interval of convergence.

- 332 -
Example 6.
Find the interval of convergence and radius of
convergence of the series
∞
(𝑥 − 5)𝑘
∑
𝑘2
𝑘=1

Solution.
We apply the ratio test for absolute convergence:
𝑢𝑘+1
𝜌 = lim | |
𝑘→+∞ 𝑢𝑘

(𝑥−5)𝑘+1 𝑘2
= lim | (𝑘+1)2
. (𝑥−5)𝑘 |
𝑘→+∞

𝑘 2
= lim ൤|𝑥 − 5| ( ) ൨
𝑘→+∞ 𝑘+1
2
1
= |𝑥 − 5| lim ( 1 )
𝑘→+∞ 1+
𝑘

= |𝑥 − 5|.
Thus, the series converges absolutely if |𝑥 − 5| < 1, or
−1 < 𝑥 − 5 < 1, or 4 < 𝑥 < 6.
The series diverges if 𝑥 < 4 or 𝑥 > 6.
To determine the convergence behavior at end points 𝑥 =
4 and 𝑥 = 6, we substitute these values in the given
series, if 𝑥 = 6 the series becomes
- 333 -
 ∞ ∞
1𝑘 1 1 1 1
∑ 2 = ∑ 2 =1+ 2+ 2+ 2+⋯
𝑘 𝑘 2 3 4
𝑘=1 𝑘=1

which is a convergent p-series (𝑝 = 2).

If 𝑥 = 4 the series becomes
∞
(−1)𝑘 1 1 1
∑ = −1 + − + −⋯
𝑘2 22 32 42
𝑘=1

which converges absolutely.

Therefore, the interval of converges for the given series is
[4, 6]. The radius of convergence is 𝑅 = 1.◄

- 334 -
 Exercise Set (7.6)
In Exercise (1)-(24), find the radius of convergence and the
interval of convergence.
∞ 𝑥𝑘
(1) ∑𝑘=0 ; (2) ∑∞ 𝑘 𝑘
𝑘=0 3 𝑥 ;
𝑘+1
(−1) 𝑥 𝑘 𝑘 𝑘!
(3) ∑∞
𝑘=0 ; (4) ∑∞
𝑘=0 𝑥𝑘;
𝑘! 2𝑘
5𝑘 𝑥𝑘
(5) ∑∞ 𝑘
𝑘=1 𝑘 2 𝑥 ; (6) ∞
∑𝑘=1 ;
ln 𝑘

∞ 𝑥𝑘 𝑘 𝑘+1
∞ (−2) 𝑥
(7) ∑𝑘=1 ; (8) ∑𝑘=0 ;
𝑘 (𝑘+1) 𝑘+1

𝑥𝑘 (−1)𝑘 𝑥 2𝑘
(9) ∑∞
𝑘=1(−1)
𝑘−1
; (10) ∑∞
𝑘=0 ;
√𝑘 (2𝑘) !

𝑥 2𝑘+1 𝑥 3𝑘
(11) ∑∞ 𝑘
𝑘=0(−1) (2𝑘+1) ; (12) ∑∞
𝑘=0(−1)
𝑘
;
! 𝑘 3/2

3𝑘 𝑥𝑘
(13) ∑∞
𝑘=0 𝑥𝑘; (14) ∑∞
𝑘=2(−1)
𝑘+1
;
𝑘! 𝑘 (𝑙𝑛 𝑘)2

𝑥𝑘 (𝑥−3)𝑘
(15) ∑∞
𝑘=0 2; (16) ∑∞
𝑘=0 ;
1+𝑘 2𝑘
(𝑥+1)𝑘 (𝑥−4)𝑘
(17) ∑∞
𝑘=1(−1)
𝑘+1
; (18)∑∞ 𝑘
𝑘=0(−1) (𝑘+1)2 ;
𝑘

∞ 3 𝑘 (2𝑘+1) !
(19) ∑𝑘=0 ( ) (𝑥 + 5)𝑘 ; (20) ∑∞
𝑘=1 (𝑥 − 2)𝑘 ;
4 𝑘3
(𝑥+1)2𝑘+1 (𝐼𝑛 𝑘)(𝑥−3)𝑘
(21) ∑∞
𝑘=1(−1)
𝑘
; (22) ∑∞
𝑘=1 ;
𝑘 2 +4 𝑘
𝜋𝑘 (𝑥−1)2 𝑘 (2 𝑥−3)𝑘
(23) ∑∞
𝑘=0 ; (24) ∑∞
𝑘=1 .
(2 𝑘+1) ! 4 2𝑘

- 335 -
(25) Use the root test to find the interval of
𝑥𝑘
convergence of ∑∞
𝑘=2 .
(ln 𝑘)𝑘

(25) Find the radius of convergence of

∞
1.2.3 … 𝑘
∑ (−1)𝑘 𝑥 2𝑘+1
1.3.5 … (2𝑘 − 1)
𝑘=1
(𝑥−𝑎) 𝑘
(26) Find the interval of convergence of ∑∞
𝑘=0 𝑏𝑘

where 𝑏 > 0.
(27) Find the radius of convergence of the power
(𝑝𝑘)!
series ∑∞
𝑘=0 (𝑘 𝑥 𝑘 , where p is a positive integer.
!)𝑝

(28) Find the radius of convergence of the power

(𝑘+𝑝)!
series ∑∞
𝑘=0 𝑥 𝑘 , where p and q are positive
𝑘!(𝑘+𝑞)!

integers.
(29) Prove: If lim |𝑐𝑘 |1/𝑘 = 𝐿 , where 𝐿 ≠ 0, then
𝑘→+∞

the power series ∑∞ 𝑘

𝑘=0 𝑐𝑘 𝑥 has radius of

convergence 1/𝐿.

- 336 -
7. TAYLOR AND MACLAURIN SERIES
In this section we will study the approximation of
function by polynomials and introduce an important class
of power series.
Suppose we are interested in approximating a function 𝑓
by a polynomial:
𝑝(𝑥 ) = 𝑐0 + 𝑐1 𝑥 + ⋯ + 𝑐𝑛 𝑥 𝑛 (1)
over an interval centered at 𝑥 = 0.
Because 𝑝(𝑥) has 𝑛 + 1 coefficients, it seems
reasonable that we will be able to impose 𝑛 + 1
conditions on this polynomial. We will assume that the
first 𝑛 derivatives of 𝑓 exist at 𝑥 = 0, and we will
choose these 𝑛 + 1 conditions to be:
𝑓(0) = 𝑝(0), 𝑓 ′ (0) = 𝑝′ (0), … , 𝑓 (𝑛) (0) = 𝑝(𝑛) (0) (2)
These conditions require that the value of 𝑝(𝑥) and its
first 𝑛 derivatives match the value of 𝑓(𝑥) and its first 𝑛
derivatives at 𝑥 = 0.
By forcing this high degree of "match" at 𝑥 = 0, it is
reasonable to hope that 𝑓(𝑥)and 𝑝(𝑥) will remain close
over some interval (possibly small) centered at 𝑥 = 0.
𝑝(𝑥 ) = 𝑐0 + 𝑐1 𝑥 + 𝑐2 𝑥 2 + 𝑐3 𝑥 3 + ⋯ + 𝑐𝑛 𝑥 𝑛
- 337 -
 𝑝′ (𝑥 ) = 𝑐1 + 2𝑐2 𝑥 + 3𝑐3 𝑥 2 + ⋯ + 𝑛𝑐𝑛 𝑥 𝑛−1
𝑝′′ (𝑥 ) = 2𝑐2 + 3 ∙ 2𝑐3 𝑥 + ⋯ + 𝑛(𝑛 − 1)𝑐𝑛 𝑥 𝑛−2
𝑝′′′ (𝑥 ) = 3 ∙ 2𝑐3 + ⋯ + 𝑛(𝑛 − 1)(𝑛 − 2)𝑐𝑛 𝑥 𝑛−3
⋮
𝑝(𝑛) (𝑥 ) = 𝑛(𝑛 − 1)(𝑛 − 2) … (1)𝑐𝑛
we obtain on substituting 𝑥 = 0
𝑝(0) = 𝑐0
𝑝′ (0) = 𝑐1
𝑝′′ (0) = 2𝑐2 = 2! 𝑐2
𝑝′′′ (0) = 3 ∙ 2𝑐2 = 3! 𝑐3
⋮
𝑝(𝑛) (0) = 𝑛(𝑛 − 1)(𝑛 − 2) … 𝑐𝑛 = 𝑛! 𝑐𝑛
Thus, from (2),
𝑓(0) = 𝑐0
𝑓 ′ (0) = 𝑐1
𝑓 ′′ (0) = 2! 𝑐2
𝑓 ′′′ (0) = 3! 𝑐3
⋮
𝑓 (𝑛) (0) = 𝑛 ! 𝑐𝑛
𝑓 ′′ (0) 𝑓 ′′′ (0)
So, 𝑐0 = 𝑓(0), 𝑐1 = 𝑓 ′ (0), 𝑐2 = , 𝑐3 = , …,
2! 3!

- 338 -
 𝑓 (𝑛) (0)
𝑐𝑛 =
𝑛!
Substituting these values in (1) yields a polynomial,
called the nth Maclaurin polynomial for 𝑓.
Definition 1.
If 𝑓 can be differentiated 𝑛 times at 0, then we define the
nth Maclaurin polynomial for 𝑓 to be
𝑓 ′ (0) 𝑓 ′′ (0) 2 𝑓 ′′′ (0)
𝑝𝑛 (𝑥 ) = 𝑓(0) + 𝑥+ 𝑥 + 𝑥3 + ⋯ +
1! 2! 3!
𝑓 (𝑛) (0)
𝑥 𝑛 (𝑛 = 0, 1, 2, … ) (3)
𝑛!

This polynomial has the property that its value and the
values of its first 𝑛 derivatives match the value of 𝑓(𝑥)
and its first 𝑛 derivatives when 𝑥 = 0.
Example 1.
Find the Maclaurin polynomials 𝑝0 , 𝑝1 , 𝑝2 , 𝑝3 and 𝑝𝑛 for
𝑒𝑥.
Solution.
Let 𝑓(𝑥) = 𝑒 𝑥 .
Thus 𝑓 ′ (𝑥 ) = 𝑓 ′′ (𝑥 ) = ⋯ = 𝑓 (𝑛) (𝑥 ) = 𝑒 𝑥 and 𝑓(0) =
𝑓 ′ (0) = 𝑓 ′′ (0) = ⋯ = 𝑓 (𝑛) (0) = 𝑒 0 = 1.
Therefore 𝑝0 (𝑥 ) = 𝑓 (0) = 1;
- 339 -
𝑝1 (𝑥 ) = 𝑓 (0) + 𝑓 ′ (0) = 1 + 𝑥;
𝑓 ′′ (0) 𝑥2
𝑝2 (𝑥 ) = 𝑓 (0) + 𝑓 ′ (0)𝑥 + 𝑥2 = 1 + 𝑥 + ;
2! 2!

′(
𝑓 ′′ (0) 2 𝑓 ′′′ (0) 3
𝑝3 (𝑥 ) = 𝑓 (0) + 𝑓 0) + 𝑥 + 𝑥
2! 3!
𝑥2 𝑥3
=1+𝑥+ + ;
2! 3!

′(
𝑓 ′′ (0) 2 𝑓 (𝑛) (0) 𝑛
𝑝𝑛 (𝑥 ) = 𝑓(0) + 𝑓 0) + 𝑥 +⋯+ 𝑥
2! 𝑛!
𝑥2 𝑥𝑛
=1+𝑥+ +⋯+ ;
2! 𝑛!

In the figure, we have sketched the graphs of 𝑒 𝑥 and its

Maclaurin polynomials of degree 1, 2, and 3.
Note that the graphs of 𝑒 𝑥 and 𝑝3 (𝑥) are virtually
indistinguishable over the interval from −0.5 𝑡𝑜 + 0.5.◄

- 340 -
Example 2.
Find the nth Maclaurin polynomial for ln (𝑥 + 1).
Solution.
Let 𝑓(𝑥) = ln (𝑥 + 1) and arrange the computations as
follows:
𝑓(𝑥 ) = ln (𝑥 + 1) ⟹ 𝑓(0) = ln 1 = 0;
1
𝑓 ′ (𝑥 ) = ⟹ 𝑓 ′ (0) = 1;
𝑥+1
1
𝑓 ′′ (𝑥 ) = − ⟹ 𝑓 ′′ (0) = −1;
(𝑥+1)2
2
𝑓 ′′′ (𝑥 ) = (𝑥+1)3 ⟹ 𝑓 ′′′ (0) = 2;
3 .2
𝑓 4 (𝑥) = (𝑥+1)4 ⟹ 𝑓 (4) (0) = −3 !;
4 .3 .2
𝑓 (5) (𝑥) = (𝑥+1)5 ⟹ 𝑓 (5) (0) = 4 !;

⋮ ⋮
(𝑛 − 1) !
𝑓 (𝑛) = (−1)𝑛+1 ⟹
(𝑥 + 1)𝑛
𝑓 (𝑛) (0) = (−1)𝑛+1 (𝑛 − 1) !.
Substituting these values in (3) yields
𝑥2 𝑥3 𝑥𝑛
𝑝𝑛 (𝑥 ) = 𝑥 − + − ⋯ + (−1)𝑛+1 .◄
2 3 𝑛

- 341 -
 Example 3.
In the Maclaurin polynomials for sin 𝑥 , only the odd
powers of 𝑥 appear explicitly.
To see this, let 𝑓 (𝑥 ) = sin 𝑥, thus
𝑓 (𝑥 ) = sin 𝑥 ⟹ 𝑓(0) = 0
𝑓′(𝑥) = cos 𝑥 ⟹ 𝑓′(0) = 1
𝑓"(𝑥) = − sin 𝑥 ⟹ 𝑓"(0) = 0
𝑓 ′′′ (𝑥 ) = − cos 𝑥 ⟹ 𝑓 ′′′ (0) = −1
Since 𝑓 (4) (𝑥) = sin 𝑥 = 𝑓(𝑥), the pattern 0, 1, 0, −1
will repeat over and over as we evaluate successive
derivatives at 0. Therefore, the successive Maclaurin
polynomials for sin x are
𝑝1 (𝑥) = 0 + 𝑥 = 𝑥 ;
𝑝2 (𝑥) = 0 + 𝑥 + 0 = 𝑥 ;
𝑥3 𝑥3
𝑝3 (𝑥 ) = 0 + 𝑥 + 0 − =𝑥− ;
3! 3!
𝑥3 𝑥3
𝑝4 (𝑥 ) = 0 + 𝑥 + 0 − +0=𝑥− ;
3! 3!
𝑥3 𝑥5 𝑥3 𝑥5
𝑝5 (𝑥 ) = 0 + 𝑥 + 0 − +0+ =𝑥− + ;
3! 5! 3! 5!
𝑥3 𝑥5 𝑥3 𝑥5
𝑝6 (𝑥 ) = 0 + 𝑥 + 0 − +0+ +0=𝑥− + ;
3! 5! 3! 5!

- 342 -
 𝑥3 𝑥5 𝑥7
𝑝7 (𝑥 ) = 0 + 𝑥 + 0 − + 0 + + 0 −
3! 5! 7!
𝑥3 𝑥5 𝑥7
=𝑥− + −
3! 5! 7!
In general, the Maclaurin polynomials for sin x are
𝑥3 𝑥5 𝑥7
𝑝2𝑛+1 (𝑥 ) = 𝑝2𝑛+2 (𝑥 ) = 𝑥 − + − + ⋯+
3! 5! 7!

𝑛 𝑥 2𝑛+1
(−1) (𝑛 = 0, 1, 2, … ).◄
(2𝑛+1)!

Example 4.
The successive Maclaurin polynomials for cos 𝑥 are
𝑝0 (𝑥 ) = 𝑝1 (𝑥 ) = 1;
𝑥2
𝑝2 (𝑥 ) = 𝑝3 (𝑥 ) = 1 − ;
2!
𝑥2 𝑥4
𝑝4 (𝑥 ) = 𝑝5 (𝑥 ) = 1 − =𝑥− ;
2! 4!
𝑥2 𝑥4 𝑥6
𝑝6 (𝑥 ) = 𝑝7 (𝑥 ) = 1 − + − ;
2! 4! 6!
𝑥2 𝑥4 𝑥6 𝑥8
𝑝8 (𝑥 ) = 𝑝9 (𝑥 ) = 1 − + − + .
2! 4! 6! 8!

In general, the Maclaurin polynomials for cos 𝑥 are

𝑥2 𝑥4 𝑥 2𝑛
𝑝2𝑛 (𝑥 ) = 𝑝2𝑛+1 (𝑥 ) = 1 − + − ⋯ + (−1)𝑛 (2𝑛)!,
2! 4!

(𝑛 = 0, 1, 2, … ).◄

- 343 -
 Definition 2.
If 𝑓 can be differentiated 𝑛 times at 𝑎, then we define the
nth Taylor polynomials for 𝑓 about 𝑥 = 𝑎 to be

′(
𝑓 ′′ (𝑎)
𝑝𝑛 (𝑥 ) = 𝑓(𝑎) + 𝑓 𝑎)(𝑥 − 𝑎) + (𝑥 − 𝑎)2
2!
𝑓 ′′′ (𝑎) 𝑓 𝑛 (𝑎)
+ (𝑥 − 𝑎)3 + ⋯ + (𝑥 − 𝑎)𝑛 (5)
3! 𝑛!

Observe that the Taylor polynomials include the

Maclaurin polynomials as a special case (𝑎 = 0).
Example 5.
Find the Taylor polynomials 𝑝1 (𝑥 ), 𝑝2 (𝑥 ), and 𝑝3 (𝑥) for
sin x about 𝑥 = 𝜋/3.
Solution.
Let 𝑓(𝑥 ) = sin 𝑥 . Thus
𝜋 √3
𝑓(𝑥 ) = sin 𝑥 ⇒ 𝑓(𝜋/3) = sin = ;
3 2
𝜋 1
𝑓 ′ (𝑥 ) = cos 𝑥 ⇒ 𝑓(𝜋/3) = cos = ;
3 2
𝜋 √3
𝑓 ′′ (𝑥 ) = − sin 𝑥 ⇒ 𝑓 ′′ (𝜋/3) = − sin = − ;
3 2
𝜋 1
𝑓 ′′′ (𝑥 ) = − cos 𝑥 ⇒ 𝑓 ′′′ (𝜋/3) = − cos = − .
3 2

Substituting in (5) with 𝑎 = 𝜋/3 yields

- 344 -
 𝜋 𝜋 𝜋 √3 1 𝜋
𝑝1 (𝑥 ) = 𝑓 ( ) + 𝑓 ′ ( ) (𝑥 − ) = + (𝑥 − ).
3 3 3 2 2 3
𝜋
𝜋 𝜋 𝜋 𝑓 ′′ ( 3 ) 𝜋 2
𝑝2 (𝑥 ) = 𝑓 ( ) + 𝑓′ ( ) (𝑥 − 3 ) + (𝑥 − 3 )
3 3 2!

√3 1 𝜋 √3 𝜋 2
= + (𝑥 − ) − (𝑥 − ) .
2 2 3 2.2 ! 3
𝜋
𝜋 𝜋 𝜋 𝑓 ′′ ( ) 𝜋 2
′ 3
𝑝3 (𝑥 ) = 𝑓 ( ) + 𝑓 ( ) (𝑥 − ) + (𝑥 − ) +
3 3 3 2! 3
𝑓 ′′′ (𝜋/3) 𝜋 3
3!
(𝑥 − 3 )

√3 1 𝜋 √3 𝜋 2 1 𝜋 3
= + (𝑥 − ) − (𝑥 − ) − (𝑥 − 3 ) .
2 2 3 2 .2 ! 3 2 .3 !

It is convenient to express the defining formula for the

Taylor polynomials in sigma notation. To do this, we use
the notation 𝑓 (𝑘) (𝑎) to denote the kth derivative of f at
𝑥 = 𝑎, and we make the added convention that 𝑓 (0) (𝑎)
denotes 𝑓(𝑎). This enables us to write
𝑛
𝑓 (𝑘) (𝑎)
∑ (𝑥 − 𝑎)𝑘 = 𝑓(𝑎) + 𝑓 ′ (𝑎)(𝑥 − 𝑎)
𝑘!
𝑥=0

𝑓 ′′ (𝑎) 2
𝑓 (𝑛) (𝑎)
+ (𝑥 − 𝑎) + ⋯ + (𝑥 − 𝑎)𝑛
2! 𝑛!

- 345 -
In particular, the nth Maclaurin polynomial for 𝑓(𝑥) is
𝑛
𝑓 (𝑘) (𝑎) 𝑘
∑ 𝑥
𝑘!
𝑘=0

′(
𝑓 ′′ (0) 2 𝑓 (𝑛) (0) 𝑛
= 𝑓 (0) + 𝑓 0) + 𝑥 +⋯ 𝑥
2! 𝑛!
Definition 3.
If f derivatives of all orders at a, then we define the Taylor
Series for f about 𝑥 = 𝑎 to be
∞
𝑓 (𝑘) (𝑎)
∑ (𝑥 − 𝑎)𝑘 = 𝑓(𝑎) + 𝑓 ′ (𝑎)(𝑥 − 𝑎)
𝑘!
𝑘=0
𝑓 ′′ (𝑎) 𝑓 (𝑘) (𝑎)
+ (𝑥 − 𝑎 )2 +⋯+ (𝑥 − 𝑎)𝑘 + ⋯ (6)
2! 𝑘!

Definition 4.
If f has derivatives of all orders at 0, then we define the
Maclaurin series for f to be
∞
𝑓 (𝑘) (0) 𝑘
∑ 𝑥
𝑘!
𝑘=0

′(
𝑓 ′′ (0) 2
= 𝑓(0) + 𝑓 0)𝑥 + 𝑥 +⋯
2!
𝑓 (𝑘) (0) 𝑘
+ 𝑥 + ⋯ (7)
𝑘!

- 346 -
Observe that the Maclaurin series for f is just the Taylor
series for f about 𝑎 = 0.
Example 6.
In Example 1, we found the nth Maclaurin polynomial
for 𝑒 𝑥 to be
∞
𝑥𝑘 𝑥2 𝑥𝑛
∑ =1+𝑥+ + ⋯+
𝑘! 2! 𝑛!
𝑘=0

Thus, the Maclaurin series for 𝑒 𝑥 is

∞
𝑥𝑘 𝑥2 𝑥3 𝑥𝑛
∑ =1+𝑥+ + +⋯+ +⋯
𝑘! 2! 3! 𝑛!
𝑘=0

From Example 3 the Maclaurin series for sin x is

∞
𝑥 2𝑘+1 𝑥3 𝑥5 𝑥7
∑ (−1 )𝑘 =𝑥− + − +⋯
(2𝑘 + 1) ! 3! 5! 7!
𝑘=0

and from Example 4 the Maclaurin series for cos x is

∞
𝑥 2𝑘 𝑥2 𝑥4 𝑥6
∑ (−1 )𝑘 =1− + − +⋯
(2𝑘) ! 2! 4! 6!
𝑘=0

- 347 -
Example 7.
1
Find the Taylor series about 𝑥 = 1 for .
𝑥

Solution.
1 1
Let 𝑓(𝑥 ) = so that 𝑓(𝑥 ) = ⇒ 𝑓(1) = 1;
𝑥 𝑥
1
𝑓 ′ (𝑥 ) = − ⇒ 𝑓 ′ (1) = −1;
𝑥2
2
𝑓 ′′ (𝑥 ) = ⇒ 𝑓 ′′ (1) = 2 !;
𝑥3
3 .2
𝑓 ′′′ (𝑥 ) = − ⇒ 𝑓 ′′′ (1) = −3 !;
𝑥4
4 .3 .2
𝑓 (4) (𝑥 ) = ⇒ 𝑓 (4) (1) = 4 !;
𝑥5

⋮ ⋮
𝑘!
𝑓 (𝑘) (𝑥 ) = (−1)𝑘 ⇒ 𝑓 (𝑘) (1) = (−1)𝑘 𝑘 !;
𝑥 𝑘+1

Thus, substituting in (6) with 𝑎 = 1 yields

∞ ∞
(−1)𝑘 𝑘 !
∑ (𝑥 − 1)𝑘 = ∑ (−1)𝑘 (𝑥 − 1)𝑘
𝑘!
𝑘=0 𝑘=0

= 1 − (𝑥 − 1) + (𝑥 − 1)2 − (𝑥 − 1)3 + ⋯◄

- 348 -
 Exercise Set (7.7)
In Exercise (1)-(12), find the fourth Maclaurin polynomial
(𝑛 = 4) for the given function.
1
(1) 𝑒 −2𝑥 ; (2) ; (3) sin 2𝑥; (4) 𝑒 𝑥 cos 𝑥; (5) tan 𝑥;
1+𝑥

(6) 𝑥 3 − 𝑥 2 + 2𝑥 + 1; (7) 𝑥 𝑒 𝑥 ; (8) tan−1 𝑥; (9) sec 𝑥;

(10) √1 + 𝑥; (11) ln (3 + 2 𝑥); (12) sinh 𝑥.
In Exercise (13)–(22), find the third Taylor polynomial
(𝑛 = 3) about 𝑥 = 𝑎 for the given function.
(13) 𝑒 𝑥 , 𝑎 = 1; (14) ln 𝑥 , 𝑎 = 1;
(15) √𝑥 , 𝑎 = 4; (16) 𝑥 4 + 𝑥 − 3 , 𝑎 = −2;
𝜋 𝜋
(17) cos 𝑥 , 𝑎 = ; (18) tan 𝑥 , 𝑎 = ;
4 3
1 𝜋
(19) sin 𝜋 𝑥 , 𝑎 = − ; (20) csc 𝑥 , 𝑎 = ;
3 2

(22) tan−1 𝑥 , 𝑎 = 1; (23) cosh 𝑥 , 𝑎 = ln 2;

In Exercise (23)–(31), find the Maclaurin series for the
given function. Express your answer in sigma notation.
1
(23) 𝑒 −𝑥 ; (24) 𝑒 𝑎 𝑥 ; (25) ; (26) 𝑥𝑒 𝑥 ;
1+𝑥
𝑥
(27) ln(1 + 𝑥);(28) sin 𝜋 𝑥; (29) cos ( );
2

(30) sinh 𝑥; (31) cosh 𝑥.

- 349 -
In Exercise (32)–(39), find the Taylor series about 𝑥 = 𝑎
for the given function. Express your answer in sigma
notation.
1 1
(32) , 𝑎 = 3; (33) , 𝑎 = −1;
𝑥 𝑥

(34) 𝑒 𝑘 , 𝑎 = 2; (35) ln 𝑥 , 𝑎 = 1;
𝜋 1
(36) cos 𝑥 , 𝑎 = ; (37) sin 𝜋 𝑥 , 𝑎 = ;
2 2
1
(38) , 𝑎 = 3; (39) sinh 𝑥 𝑥 , 𝑎 = ln 4.
𝑥+2

(40) Prove that the value of

𝑓 (𝑛) (𝑎)
𝑝𝑛 (𝑥 ) = ∑𝑛𝑘=0 (𝑥 − 𝑎)𝑘
𝑘!

and its first n derivatives match the value of 𝑓(𝑥) and its
first n derivatives at 𝑥 = 𝑎.

- 350 -
- 351 -
 CHAPTER (VIII)
Differential Equation
1. Introduction

The words differential and equations clearly indicate

solving equation involving derivatives. Differential

equations are interesting and important because they

express relationships involving rates of change. Such

relationships form the basis for developing ideas and

studying phenomena in sciences, economics, engineering,

finance, medicine and in short without any exaggeration

every aspect of human knowledge.

2. Definitions and Terminology

Definition 1.
An equation containing the derivatives of one or more
dependent variable, with respect to one or more
independent variables, is said to be a differential
equation (DE).

- 352 -
Definition 2.
A differential equation is said to be an ordinary
differential equation (ODE) if it contains only ordinary
derivatives of one or more dependent variables, with
respect to a single independent variable x.
Definition 3.
An equation involving the partial derivatives of one or
more dependent variables of two or more independent
variables is called a partial differential equation (PDE).
Example 1.
𝑑𝑦 𝑑2 𝑦
+ 10𝑦 = 𝑒 𝑥 , + 10𝑦 = sin 𝑥
𝑑𝑥 𝑑𝑥 2
𝑑2 𝑦 𝑑𝑦 𝑑𝑥 𝑑𝑦
− + 6𝑦 = 0, + = 2𝑥 + 𝑦
𝑑𝑥 2 𝑑𝑥 𝑑𝑡 𝑑𝑡
are examples of ordinary differential equation.
Example 2.
𝜕2 𝑢 𝜕2 𝑢 𝜕𝑦 𝜕2 𝑢 𝜕2 𝑢 𝜕2 𝑢
= ; = ; =−
𝜕𝑡 2 𝜕𝑥 2 𝜕𝑥 𝜕𝑡 2 𝜕𝑥 2 𝜕𝑦 2

𝜕2 𝑢 𝜕2 𝑢 𝜕𝑢 𝜕𝑢 𝜕𝑣
= −2 ; =−
𝜕𝑥 2 𝜕𝑡 2 𝜕𝑡 𝜕𝑦 𝜕𝑥

are examples of partial differential equation.

- 353 -
Definition 4.
The order of a differential equation (ODE) is the order of
the highest derivative in the equation.
Example 3.
𝑑𝑦
(i) 𝑥 − 𝑦 2 = 0 is an equation of the 1st order.
𝑑𝑥

𝑑2 𝑦 𝑑𝑦 4
(ii) + 5 ( ) − 4𝑦 = 𝑒 𝑥 is 2nd order.
𝑑𝑥 2 𝑑𝑥
𝑑3 𝑦 𝑑2 𝑦
(iii) + − 4𝑦 = 𝑒 −𝑥 is 3rd order.
𝑑𝑥 3 𝑑𝑥 2

Definition 5.
The degree of a differential equation (ODE) is the degree
of the highest order derivative in the equation.
Example 4.
𝑑𝑦 𝑑𝑦 2
(i) 𝑦 − √𝑥 ( ) − 5 = 0 is an ODE of 2nd degree.
𝑑𝑥 𝑑𝑥

𝑑2 𝑦 𝑑𝑦 2
(ii) − ( ) + 6𝑦 + 10 = 0 is an ODE of 1st degree.
𝑑𝑥 2 𝑑𝑥

Remarks 1.
(i) Very often notation 𝑦 ′ , 𝑦 ′′ , . . . , 𝑦 (𝑛) are respectively
𝑑𝑦 𝑑2 𝑦 𝑑3 𝑦 𝑑𝑛 𝑦
used for , , ,…,
𝑑𝑥 𝑑𝑥 2 𝑑𝑥 3 𝑑𝑥 𝑛

- 354 -
(ii) In symbols we can express an nth order ordinary
differential equation in one dependent variable y by the
general form
𝐹(𝑥, 𝑦, 𝑦 ′ , 𝑦 ′′ , . . . , 𝑦 (𝑛) ) = 0
where F is a real-valued function of 𝑛 + 2 variables
𝑥, 𝑦, 𝑦 ′ , 𝑦 ′′ , . . . , 𝑦 (𝑛) .
Definition 6.
An n th-order ordinary differential equation is said to be
linear in 𝑦 if it can be written in the form
𝑎𝑛 (𝑥)𝑦 (𝑛) + 𝑎𝑛−1 (𝑥)𝑦 (𝑛−1) + . . . + 𝑎1 (𝑥)𝑦’ + 𝑎0 (𝑥)𝑦
= 𝑓(𝑥)
Where 𝑎0 , 𝑎1 , 𝑎2 , … , 𝑎𝑛 and 𝑓 are functions of 𝑥 on some
interval, and 𝑎𝑛 (𝑥) ≠ 0 on that interval. The functions
𝑎𝑘 (𝑥), 𝑘 = 0, 1, 2, . . . . , 𝑛 are called the coefficient
functions.
A differential equation that is not linear is said to be non-
linear.

- 355 -
Example 5.
(i) 𝑦 ′′ = 4𝑦 ′ + 3𝑦 − 𝑥 4 and 𝑥𝑦′′ + 𝑦𝑒 𝑥 + 6 = 0
are linear differential equations.
(ii) (𝑦 − 𝑥)𝑑𝑥 + 4𝑥𝑑𝑦 = 0,
𝑑 𝑦 3 𝑑𝑦
𝑥3 3 − 4𝑥 + 12𝑦 = 𝑒 𝑥 , and
𝑑𝑥 𝑑𝑥

𝑦 ′′ − 4𝑦 ′ + 𝑦 = 0.
are linear differential equations.
𝑑2 𝑦
(iii) (1 + 𝑦)𝑦 ′ + 2𝑦 = 𝑒 𝑥 , + cos 𝑦 = 0, and
𝑑𝑥 2
𝑑3 𝑦
+ 𝑦 2 = 0 are non-linear.
𝑑𝑥 3

Remark 2.
An ordinary differential equation is linear if the following
conditions are satisfied.
(i) The unknown function y and its derivatives occur in
the first degree only.
(ii) There are no products involving either the unknown
function and its derivatives or two or more derivatives.
(iii) There are no transcendental functions involving the
unknown function or any of its derivatives.

- 356 -
Definition 7.
(i) A solution or a general solution of an nth-order
differential equation of the form
𝐹(𝑥, 𝑦, 𝑦 ′ , 𝑦 ′′ , . . . , 𝑦 (𝑛) ) = 0
on an interval 𝐼 = [𝑎, 𝑏] = {𝑥ℝ: 𝑎 ≤ 𝑥 ≤ 𝑏} is any
function possessing all the necessary derivatives, which
when substituted for 𝑥, 𝑦, 𝑦 ′ , 𝑦 ′′ , . . . , 𝑦 (𝑛) , reduces the
differential equation to an identity.
In other words, an unknown function is a solution of a
differential equation if it satisfies the equation.
(ii) A solution of a differential equation of order n will
have n independent arbitrary constants. Any solution
obtained by assigning particular numerical values to some
or all of the arbitrary constants is a particular solution.
(iii) A solution of a differential equation that is not
obtainable from a general solution by assigning particular
numerical values is called a singular solution.
(iv) A real function 𝑦 = (𝑥) is called an explicit
solution of the differential equation
𝐹(𝑥, 𝑦, 𝑦 ′ , 𝑦 ′′ , . . . , 𝑦 (𝑛) ) = 0 on [𝑎, 𝑏] if 𝐹(𝑥,  (𝑥),
 (𝑥)′ , . . . , (𝑛) (𝑥) ) = 0 on [𝑎, 𝑏].
- 357 -
(v) A relation 𝑔(𝑥, 𝑦) = 0 is called an implicit solution
of the differential equation 𝐹(𝑥, 𝑦, 𝑦 ′ , 𝑦 ′′ , . . . , 𝑦 (𝑛) ) = 0
on [𝑎, 𝑏] if 𝑔(𝑥, 𝑦) = 0 defines at least one real function
f on [𝑎, 𝑏] such that 𝑦 = 𝑓(𝑥) is an explicit solution on
this interval. We now illustrate these concepts through the
following examples:
Example 6.
(i) 𝑦 = 𝑐1 𝑒 𝑥+𝑐2 is a solution of the equation
𝑦’’ − 𝑦 = 0
This ODE is of order 2 and so its solution involves 2
arbitrary constants c1 and c2. It is clear that 𝑦 ′ = 𝑐1 𝑒 𝑥+𝑐2 ,
𝑦′′ = 𝑐1 𝑒 𝑥+𝑐2 and so 𝑐1 𝑒 𝑥+𝑐2 − 𝑐1 𝑒 𝑥+𝑐2 = 0.
Hence 𝑦 = 𝑐1 𝑒 𝑥+𝑐2 is a general solution or simply a
solution.
(ii) 𝑦 = 𝑐𝑒 2𝑥 is a solution of ODE 𝑦 ′ − 2𝑦 = 0,
because 𝑦′ = 2𝑐𝑒 2𝑥 and 𝑦 = 𝑐𝑒 2𝑥 satisfy the ODE. Since
given ODE is of order 1, solution contains only one
constant.
1
(iii) 𝑦 = 𝑐𝑥 + 𝑐 2 is a solution of the equation
2
1
(𝑦 ′ )2 + 𝑥𝑦 ′ – 𝑦 = 0.
2

- 358 -
To verify the validity, we note that 𝑦 ′ = 𝑐, and therefore
1 2 1
(𝑐) + 𝑐𝑥 − (𝑐𝑥 + 𝑐 2 ) = 0
2 2
(iv) 𝑦 = 𝑐1 𝑒 2𝑥 + 𝑐2 𝑒 −𝑥 is a general solution of the 2nd
order differential equation 𝑦 ′′ − 𝑦 ′ − 2𝑦 = 0.
To check the validity, we compute 𝑦 ′ and 𝑦 ′′ and put
values in this equation.
𝑦 ′ = 2𝑐1 𝑒 2𝑥 − 𝑐2 𝑒 −𝑥 , 𝑦′′ = 4𝑐1 𝑒 2𝑥 + 𝑐2 𝑒 −𝑥
L.H.S. of the given ODE
= (4𝑐1 𝑒 2𝑥 + 𝑐2 𝑒 −𝑥 ) − (2𝑐1 𝑒 2𝑥 − 𝑐2 𝑒 −𝑥 ) − 2(𝑐1 𝑒 2𝑥 +
𝑐2 𝑒 −𝑥 ) = 0 =
R.H.S. of the given ODE.
Example 7.
(i) Choosing 𝑐 = 1 we get a particular solution of
differential equation considered in Example 6 (iii).
(ii) For 𝑐1 = 1 we get a particular solution of differential
equation in Example 6 (i), that is,𝑦 = 𝑒 𝑥+𝑐2 is a
particular solution of 𝑦′′ − 𝑦 = 0.

- 359 -
Example 8.
1
(i) 𝑦 = − 𝑥 2 is a singular solution of differential equation
2

in Example 6 (iii).
(ii) 𝑦 = 0 is a singular solution of 𝑦 ′ = 𝑥𝑦1/2 .
Verification.
1
(i) 𝑦 = − 𝑥 2 is not obtainable from the general solution
2
1
𝑦 = 𝑐𝑥 + 𝑐 2 . However, it is a solution of the given
2

differential equation, can be checked as follows:

𝑦 ′ = −𝑥.
By putting values of y and 𝑦’ into the R.H.S of the equation
we get
1 1
(−𝑥 )2 + 𝑥 (−𝑥 ) − (− 𝑥 2 ) = 0.
2 2
1
(ii) The solution of this equation is 𝑦 = 𝑥 2 + 𝑐.
4

For 𝑐 = 0, we do not get the solution 𝑦 = 0.

Therefore, the solution 𝑦 = 0 of the equation is not
obtainable from the general solution. Hence 𝑦 = 0 is a
singular solution. ◄

- 360 -
Example 9.
(i) 𝑦 = sin 4𝑥 is an explicit solution of 𝑦 ′′ + 16𝑦 = 0
for all real x.
(ii) 𝑦 = 𝑐1 𝑒 𝑥 + 𝑐2 𝑒 −𝑥 is an explicit solution of the
equation 𝑦 ′′ − 𝑦 = 0.
Verification.
(i) 𝑦 ′ = −4 cos 4𝑥, 𝑦 ′′ = −16 sin 4𝑥.
Putting the value of y and 𝑦’’ in terms of x into the R.H.S
of the equation we get
– 16 sin 4𝑥 + 16 sin 4𝑥 = 0.
Hence the equation is satisfied for 𝑦 = sin 4𝑥.
Therefore 𝑦 = sin4𝑥 is an explicit solution of the given
equation.
(ii) 𝑦′ = 𝑐1 𝑒 𝑥 − 𝑐2 𝑒 −𝑥 , 𝑦′′ = 𝑐1 𝑒 𝑥 + 𝑐2 𝑒 −𝑥 .
Put values of y and 𝑦 ′′ in the R.H.S of the given equation
to get
(𝑐1 𝑒 𝑥 + 𝑐2 𝑒 −𝑥 ) − (𝑐1 𝑒 𝑥 + 𝑐2 𝑒 −𝑥 ) = 0.◄
Example 10.
(i) The relation 𝑥 2 + 𝑦 2 = 4 is an implicit solution of
𝑑𝑦 𝑥
the differential equation = − on the interval (−2,2).
𝑑𝑥 𝑦

- 361 -
(ii) The relation 𝑦 2 + 𝑥 − 4 = 0 is an implicit solution
of 2𝑦𝑦 ′ + 1 = 0 on the interval (−, 4).
Verification.
(i) By implicit differentiation of the relation
𝑑𝑦 𝑑𝑦 𝑥
𝑥 2 + 𝑦 2 = 4 we get 2𝑥 + 2𝑦 = 0 or =−
𝑑𝑥 𝑑𝑥 𝑦

Further,𝑦1 = √4 − 𝑥 2 and 𝑦2 = −√4 − 𝑥 2

2
satisfying the relation 𝑦 = 4 − 𝑥 2 and 𝑦 = ±√4 − 𝑥 2 .
𝑑𝑦 𝑥
are solutions of the differential equation =− .
𝑑𝑥 𝑦
1 1 𝑥
It is clear that 𝑦′1 = 2
(−2𝑥 ) = − and
2 √4−𝑥 𝑦1
1 1 𝑥
𝑦′2 = − 2
(−2𝑥 ) = − .
2 √4−𝑥 𝑦2

(ii) Differentiating 𝑦 2 + 𝑥 − 4 = 0 with respect to x, we

obtain 2𝑦𝑦 ′ + 1 = 0, which is the given differential
equation. Hence 𝑦 2 + 𝑥 − 4 = 0 is an implicit solution
if it defines a real function on (−, 4). Solving the
equation 𝑦 2 + 𝑥 − 4 = 0 for y, we get 𝑦 = ±√4 − 𝑥.
Since both 𝑦1 = √4 − 𝑥 and 𝑦2 = −√4 − 𝑥 and their
derivatives are functions defined for all x in the interval
(−, 4), we conclude that 𝑦 2 + 𝑥 − 4 = 0 is an
implicit solution on this interval. ◄
- 362 -
Remark 3.
It is very pertinent to note that a relation 𝑔(𝑥, 𝑦) = 0
can reduce a differential to an identity without
constituting an implicit solution of the differential
equation. For example, 𝑥 2 + 𝑦 2 + 1 = 0 satisfies 𝑦𝑦 ′ +
𝑥 = 0, but it is not an implicit solution as it does not
define a real-valued function. This is clear from the
solution of the equation 𝑥 2 + 𝑦 2 + 1 = 0 or 𝑦 =

±√−(1 + 𝑥 2 ), imaginary number.

The relation 𝑥 2 + 𝑦 2 + 1 = 0 is called a formal solution
of 𝑦𝑦 ′ + 𝑥 = 0. That is it appears to be a solution. Very
often we look for a formal solution rather than an implicit
solution. ◄

- 363 -
3. Formation of a differential Equation
Let us consider an equation containing 𝑛 arbitrary
constants. Then by differentiating it successively 𝑛 times
we get 𝑛 more equations containing 𝑛 arbitrary constants
and derivatives. Now, by eliminating 𝑛 arbitrary
constants from the above (𝑛 + 1) equations and obtaining
an equation which involves derivatives up to the nth
order, we get a differential equation of order 𝑛. The
concept of obtaining differential equations from a family
of curves is illustrated in following examples.
Example 1.
Find the differential equation of the family curves
𝑦 = 𝑐𝑒 2𝑥 .
Solution.
Differentiating we get 𝑦 = 𝑐𝑒 2𝑥 and 𝑦 ′ = 2𝑐𝑒 2𝑥 = 2𝑦.
Therefore, 𝑦 ′ − 2𝑦 = 0. Thus, arbitrary constant c is
eliminated and 𝑦 ′ − 2𝑦 = 0 is the required equation of
the family of curves g 𝑦 = 𝑐𝑒 2𝑥 . ◄

- 364 -
Example 2.
Find the differential equation of the family of curves
𝑦 = 𝑐1 cos 𝑥 + 𝑐2 sin 𝑥
Solution.
Differentiating (*) twice we get
𝑦 ′ = −𝑐1 sin 𝑥 + 𝑐2 cos 𝑥
𝑦 ′′ = − 𝑐1 cos 𝑥 − 𝑐2 sin 𝑥
c1 and c2 can be eliminated from the above to equations
and we obtain the different equation
𝑦 ′′ + 𝑦 = 0
Therefore 𝑦 ′′ + 𝑦 = 0 is the differential equation of the
family of curves given by 𝑦 = 𝑐1 cos 𝑥 + 𝑐2 sin 𝑥. ◄
Example 3.
Find the differential equation which has
𝑦 = 𝑐1 𝑒 𝑥 + 𝑐2 𝑒 −𝑥 + 3𝑥
as its general solution.
Solution.
Differentiate the given expression twice
𝑦 = 𝑐1 𝑒 𝑥 + 𝑐2 𝑒 −𝑥 + 3𝑥
𝑦 ′ = 𝑐1 𝑒 𝑥 − 𝑐2 𝑒 −𝑥 + 3

- 365 -
 𝑦 ′′ = 𝑐1 𝑒 𝑥 + 𝑐2 𝑒 −𝑥
Eliminate 𝑐1 and 𝑐2 we obtain 𝑦 − 𝑦 ′′ = 3𝑥 gives which
is the desired differential equation. Note that the desired
differential equation is free from the arbitrary
constants.◄

- 366 -
4. Initial-Value and Boundary-Value Problems
A general solution of an nth order ODE contains 𝑛
arbitrary constants. To obtain a particular solution, we are
required to specify 𝑛 conditions on solution function and
its derivatives and thereby expect to find values of 𝑛
arbitrary constants. There are two well-known methods
for specifying auxiliary conditions. One is called initial
conditions and other is said to be boundary conditions.
It may be observed that an ODE does not have solution or
unique solution. However, by imposing initial and
boundary conditions uniqueness can be ensured for
certain classes of differential equations.
Definition 1. (initial-Value Problem)
If the auxiliary conditions for a given differential
equation relate to a single 𝑥 value, the conditions are
called initial conditions. The differential equation with its
initial conditions is called an initial-value problem.

- 367 -
Definition 2. (boundary-value problem)
If the auxiliary conditions for a given differential
equation relate to two or more x values, the conditions are
called boundary conditions or boundary values. The
differential equation with its boundary conditions is
called boundary-value problem.

- 368 -
Example 1.
(i) 𝑦 ′ + 𝑦 = 3, 𝑦(0) = 2 is a first-order initial value
problem. Order of initial value problem is nothing but
order of the given equation. 𝑦(0) = 2 is an initial
condition.
(ii) 𝑦 ′′ + 2𝑦 = 0, 𝑦(1) = 2, 𝑦 ′ (1) = −3 is a second-
order initial value problem. Initial conditions are 𝑦(1) =
2 and 𝑦 ′ (1) = −3. Values of function 𝑦(𝑥) and its
derivative are specified for value 𝑥 = 1.
(iii) 𝑦 ′′ − 𝑦 ′ + 𝑦 = 𝑥 3 , 𝑦(0) = 4, 𝑦 ′ (1) = −2 is a
second-order boundary-value problem.
Boundary conditions are specified at two points namely
𝑥 = 0 and 𝑥 = 1.
One may specify boundary conditions for different values
of 𝑥 say 𝑥 = 2 and 𝑥 = 5. In this case the boundary-
value problem is
𝑦 ′′ − 𝑦 ′ + 𝑦 = 𝑥 3 , 𝑦(2) = 4, 𝑦 ′ (5) = −2.◄
● The following questions are quite pertinent as boundary
value and initial value problems represent important
phenomena in nature:

- 369 -
Problem 1.
When does a solution exist? That is, does an initial-value
problem or a boundary value problem necessarily have a
solution?
Problem 2.
Is a known solution unique? That is, is there only one
solution of an initial-value problem or a boundary-value
problem?
The following theorem states that under the specified
conditions, a first-order initial-value problem has a
unique solution.
Theorem 1.
𝜕𝑓
Let f and 𝑓𝑦 ( ) be continuous functions of 𝑥 and 𝑦 in
𝜕𝑦

some rectangle 𝑅 of the xy-plane, and let (𝑥∘ , 𝑦∘ ) be a

point in that rectangle. Then on some interval centred at
𝑥∘ there is a unique solution 𝑦 = (𝑥) of the initial value
problem:
𝑑𝑦
= 𝑓(𝑥, 𝑦), 𝑦(𝑥∘ ) = 𝑦∘
𝑑𝑥

- 370 -
Example 2.
(i) 𝑦 = 3 𝑒 𝑥 is a solution of the initial-value problem
𝑦’ = 𝑦, 𝑦(0) = 3. This means that the solution of the
differential equation 𝑦’ = 𝑦 passes through the point
(0, 3).
(ii) Find a solution of the initial-value problem 𝑦’ = 𝑦,
𝑦(1) = −2. That is, find a solution of differential
equation 𝑦’ = 𝑦 which passes through the point (1, −2).
Verification.
(i) Let 𝑦 = 𝑐 𝑒 𝑥 , where c is an arbitrary constant.
Then 𝑦′ = 𝑐 𝑒 𝑥 = 𝑦.Thus, 𝑦 = 𝑐 𝑒 𝑥 is a general
solution of the given equation 𝑦’ = 𝑦.
By applying initial condition, we get 3 = 𝑦(0) =
𝑐 𝑒 0 = 𝑐 or c = 3. Therefore 𝑦 = 3 𝑒 𝑥 is a solution of
the given initial value problem.
(ii) As seen in part (i) 𝑦 = 𝑐 𝑒 𝑥 is a solution of the
given equation. By imposing given initial condition we
get −2 = 𝑦(1) = 𝑐 𝑒 1 or 𝑐 = −2/𝑒.
2
Therefore 𝑦 = − 𝑒 𝑥 = −2𝑒 𝑥−1 is a solution of the
𝑒

initial-value problem. ◄

- 371 -
Example 3.
𝑑𝑦
= 𝑥𝑦1/2 , 𝑦(0) = 0
𝑑𝑥
has at least two solutions, namely 𝑦 = 0 and
𝑦 = 𝑥 4 /16.
Example 4.
(i) Does a solution of the boundary value problem 𝑦’’ +
𝑦 = 0, 𝑦(0) = 0, 𝑦() = 2 exist?
(ii) Show that the boundary value problem
𝑦’’ + 𝑦 = 0, 𝑦(0) = 0, 𝑦() = 0
has infinitely many solutions.
Solution.
(i) 𝑦 = 𝑐1 cos 𝑥 + 𝑐2 sin 𝑥 is a solution of the
differential equation 𝑦’’ + 𝑦 = 0. Using given boundary
conditions in 𝑦 = 𝑐1 cos 𝑥 + 𝑐2 sin 𝑥, we get
0 = 𝑐1 cos 0 + 𝑐2 sin 0
and
2 = 𝑐1 cos 𝜋 + 𝑐2 sin 𝜋
The first equation yields 𝑐1 = 0 and the second yields
𝑐1 = −2 which is absurd, hence no solution exists.

- 372 -
(ii) The boundary values yield
0 = 𝑐1 cos 0 + 𝑐2 sin 0
and
0 = 𝑐1 cos 𝜋 + 𝑐2 sin 𝜋
Both equations lead to the fact that 𝑐1 = 0. The constant
c2 is not assigned a value and therefore takes arbitrary
values. Thus, there are infinitely many solutions
represented by 𝑦 = 𝑐2 sin 𝑥.◄
Example 5.
Examine existence and uniqueness of a solution of the
following initial-value problems:
𝑦
(i) 𝑦 ′ = , 𝑦(2) = 1;
𝑥
𝑦
(ii) 𝑦 ′ = , 𝑦(0) = 3;
𝑥

(iii) 𝑦 ′ = −𝑥, 𝑦(0) = 2.

Solution.
(i) We examine whether conditions of Theorem 1 in this
section are satisfied. To check, we observe that
𝑦 𝜕𝑓 1
𝑓(𝑥, 𝑦) = and (𝑥, 𝑦) = .
𝑥 𝜕𝑦 𝑥

Both functions are continuous except at 𝑥 = 0.

- 373 -
 𝜕𝑓
Hence 𝑓 and satisfy the conditions of Theorem 1 in
𝜕𝑦

any rectangle 𝑅 that does not contain any part of the y -

axis(𝑥 = 0). Since the point (2, 1) is not on the y-axis,
1
there is a unique solution. One can check that 𝑦 = 𝑥 is
2

the only solution.

𝜕𝑓
(ii) In this problem neither 𝑓 nor is continuous at
𝜕𝑦

𝑥 = 0, which means that (0, 3) cannot be included in any

𝜕𝑓
rectangle 𝑅 where f and are continuous.
𝜕𝑦

Hence, we cannot conclude anything from Theorem 1.

However, it can be verified that 𝑦 = 𝑐𝑥 is a general
𝑦
solution of 𝑦’ = , but that a particular solution cannot be
𝑥

found whose graph passes through the point (0, 3).

(iii) Conditions of Theorem 1 are satisfied. Therefore,
the initial-value problem has a unique solution. ◄

- 374 -
5. Solutions of First Order Differential Equations
The systemic development of techniques for solving
differential equations logically begins with the equations
of the first order and first degree.
Equations of this type can in general, be written as
𝑑𝑦
= 𝑓(𝑥, 𝑦) (1)
𝑑𝑥

where𝑓(𝑥, 𝑦) is a given function. However, despite the

apparent simplicity of this equation, analytic solutions
are usually possible only when 𝑓 (𝑥, 𝑦) has simple forms.
Type (I) Exact (By direct integration)
If you are asked to solve the Differential Equation
𝑑𝑦
= cos 𝑥 + 3𝑥 2
𝑑𝑥
you can integrate immediately giving
𝑦 = sin 𝑥 + 𝑥 3 + 𝐴.
In the same way the following second order Differential
Equation can be solved directly by integration:
𝑑2 𝑦
2
= 3𝑒 3𝑥 + 4𝑥 − sin 𝑥
𝑑𝑥
𝑑𝑦
= 𝑒 3𝑥 + 2𝑥 2 + cos 𝑥 + 𝐴
𝑑𝑥

- 375 -
 𝑒 3𝑥 2𝑥 3
𝑦= + + sin 𝑥 + 𝐴𝑥 + 𝐵
3 3
Note that integrating twice has introduced the 2 arbitrary
constants A and B.◄
𝒅𝒏 𝒚
♣ If you have a DE of the form = 𝒇(𝒙), then you can,
𝒅𝒙𝒏

in theory, integrate 𝒏 times to find the solution. This

assumes that you can integrate at each step (remember
that you cannot integrate every function algebraically).
Your solution will contain 𝒏 arbitrary constants.
Example 1.
𝑑𝑦
Find the particular solution of the equation 𝑒 𝑥 = 4,
𝑑𝑥

given that 𝑦 = 3 when 𝑥 = 0.

Solution.
𝑑𝑦
First, we-rewrite the equation in the form = 4𝑒 −𝑥 .
𝑑𝑥

Then
𝑦 = ∫ 4𝑒 −𝑥 𝑑𝑥 = −4𝑒 −𝑥 + 𝐶.
Knowing that when 𝑥 = 0, 𝑦 = 3, we can evaluate 𝐶 in
this case, so that the required particular solution is
𝑦 = −4𝑒 −𝑥 + 7.◄

- 376 -
Type (II) Variables separable
If 𝑓(𝑥, 𝑦) = 𝑓(𝑥)𝑔(𝑦), 𝑓(𝑥) and 𝑔(𝑥) are respectively
functions of 𝑥 only and 𝑦 only, then, (1) becomes
𝑑𝑦
= 𝑓(𝑥)𝑔(𝑦)
𝑑𝑥
Since the variables x and y are now separable, we have
𝑑𝑦
∫ = ∫ 𝑓(𝑥)𝑑𝑥
𝑔(𝑦)
which expresses y implicitly in terms of x.
Example 1.
𝑑𝑦 𝑦+1
Solve the equation = .
𝑑𝑥 𝑥−1

Solution.
Rewriting the given equation, we get:
𝑑𝑦 𝑑𝑥
∫ =∫
𝑦+1 𝑥−1
or log 𝑒 ( 𝑦 + 1) = log 𝑒 ( 𝑥 − 1) + log 𝑒 𝐶
where 𝐶 is an arbitrary constant.
Hence
𝑦+1
=𝐶
𝑥−1
is the general solution. ◄

- 377 -
Example 2.
𝑑𝑦 2𝑥+1
Solve the equation = .
𝑑𝑥 𝑦

Solution.
Re-writing the given equation, we obtain
𝑦𝑑𝑦 = (2𝑥 + 1)𝑑𝑥

∫ 𝑦𝑑𝑦 = ∫(2𝑥 + 1)𝑑𝑥

Hence
1 2
𝑦 = 𝑥 2 + 𝑥 + 𝑐,
2
where c is an arbitrary constant.
Thus

𝑦 = √2𝑥 2 + 2𝑥 + 𝑐
is the general solution. ◄

- 378 -
Example 3.
Find the general solution of the equation
𝑑𝑦
( 𝑥 2 + 1) − 𝑥𝑦 = 0
𝑑𝑥
Solution.
This gives
𝑑𝑦 𝑥𝑦
= 2
𝑑𝑥 (𝑥 + 1)
Or
𝑑𝑦 𝑥𝑑𝑥
= 2
𝑦 ( 𝑥 + 1)

Integrating

𝑑𝑦 𝑥𝑑𝑥
∫ =∫ 2
𝑦 (𝑥 + 1)

1
ln|𝑦| = ln|𝑥 2 + 1| + 𝐴
2

And simplify, putting ln 𝐵 = 𝐴

ln|𝑦| = ln √𝑥 2 + 1 + ln 𝐵

or ln|𝑦| = ln(𝐵√𝑥 2 + 1).

Then 𝑦 = 𝐵√𝑥 2 + 1 is the general solution. ◄

- 379 -
Example 4.
Solve the equation
4𝑥 2 + 1 𝑑𝑦
= 𝑥𝑦
𝑦+1 𝑑𝑥
given 𝑦(1) = 0.
Solution.
This can be written as
4𝑥 2 + 1
𝑑𝑥 = 𝑦(𝑦 + 1)𝑑𝑦
𝑥
Simplify and integrate
1
∫ (4𝑥 + ) 𝑑𝑥 = ∫(𝑦 2 + 𝑦)𝑑𝑦
𝑥
Then the general solution is

2
𝑦3 𝑦2
2𝑥 + ln|𝑥 | = + + 𝐴.
3 2
When 𝑥 = 1, 𝑦 = 0, giving 𝐴 = 2. Therefore, the
particular solution is
𝑦3 𝑦2
2𝑥 2 + ln|𝑥 | = + + 2.◄
3 2

- 380 -
Type (III) Homogeneous differential equation
An expression of the nth degree in x and y is said to be
homogeneous of degree n, if when x and y are replaced
by 𝑡𝑥 and 𝑡𝑦, the result will be the original expression
multiplied by 𝑡 𝑛 ; symbolically 𝑓(𝑡𝑥, 𝑡𝑦) = 𝑡 𝑛 𝑓(𝑥, 𝑦).
Example 1.
Show that expression 𝑥 2 + 𝑥𝑦 − 𝑦 2 is homogeneous and
determine the degree.
Solution.
Replace x and y by 𝑡𝑥 and 𝑡𝑦 respectively, to obtain
(𝑡𝑥 )2 + (𝑡𝑥 )(𝑡𝑦) − (𝑡𝑦)2 = 𝑡 2 (𝑥 2 + 𝑥𝑦 − 𝑦 2 ).
Therefore, the given expression is homogeneous of
degree 2. ◄
♠ Consider the differential equation
𝑀(𝑥, 𝑦)𝑑𝑥 + 𝑁(𝑥, 𝑦)𝑑𝑦 = 0
The equation is said to be homogeneous in 𝑥 and 𝑦 if 𝑀
and 𝑁 are homogeneous functions of the same degree in
𝑥 and 𝑦. The technique for solving this equation is to
make the substitution 𝑦 = 𝑣𝑥 or 𝑥 = 𝑣𝑦 and is based
upon the following theorem.

- 381 -
Theorem 1.
Any homogeneous differential of the first order and first
degree can be reduced to the type of variables separable
by a substitution of 𝑦 = 𝑣𝑥 or 𝑥 = 𝑣𝑦.
Example 2.
Find the general solution of the differential equation
2𝑥𝑦𝑑𝑥 − (𝑥 2 − 𝑦 2 )𝑑𝑦 = 0.
Solution.
A simple check reveals that the equation is homogeneous.
Let 𝑦 = 𝑣𝑥 and 𝑑𝑦 = 𝑣𝑑𝑥 + 𝑥𝑑𝑣, and substitute into the
differential equation to obtain
2𝑥(𝑣𝑥)𝑑𝑥 − (𝑥 2 − 𝑣 2 𝑥 2 )(𝑣𝑑𝑥 + 𝑥𝑑𝑣) = 0.
𝑥 2 (𝑣 + 𝑣 3 )𝑑𝑥 + 𝑥 3 (𝑣 2 − 1)𝑑𝑣 = 0.
Dividing by 𝑥 3 (𝑣 + 𝑣 3 ) separates the variables:
𝑑𝑥 (𝑣 2 −1)𝑑𝑣
+ = 0.
𝑥 𝑣(1+𝑣 2 )

Integration yields
ln 𝑥 − ln 𝑣 + ln( 𝑣 2 + 1) = ln 𝑐
or 𝑥(𝑣 2 + 1) = 𝑐𝑣
𝑦
Re-writing the original variables by substituting 𝑣 = ,
𝑥

we obtain 𝑦 2 + 𝑥 2 = 𝑐𝑦 as the general solution. ◄

- 382 -
Example 3.
𝑑𝑦 𝑦(𝑥+𝑦)
Solve the equation = , given 𝑦(1) = 1.
𝑑𝑥 𝑥(𝑦−𝑥)

Solution.
Substitute 𝑦 = 𝑣𝑥 we obtain
𝑑𝑣 𝑣𝑥(𝑥+𝑣𝑥) 𝑣(1+𝑣)𝑥 2
𝑥 +𝑣 = = .
𝑑𝑥 𝑥(𝑣𝑥−𝑥) (𝑣−1)𝑥 2

Simplifying and integrating

1 𝑑𝑥
∫ (1 − ) 𝑑𝑣 = 2 ∫
𝑣 𝑥
𝑣 − ln|𝑣| = 2 ln|𝑥 | + 𝑐
𝑣 = ln|𝑣| + 2 ln|𝑥 | + ln 𝐴 = ln(𝐴𝑥 2 𝑣)
𝑦 𝑦
= ln (𝐴𝑥 2 ) = ln(𝐴𝑥𝑦)
𝑥 𝑥
𝑦 = 𝑥 ln(𝐴𝑥𝑦).
But 𝑦(1) = 1; substituting gives 1 = ln 𝐴 ⟹ 𝐴 = 𝑒.
The solution can be written as
𝑦 = 𝑥 ln(𝑒𝑥𝑦) or 𝑦 = 𝑥(1 + ln(𝑥𝑦)).◄
♠It is important to realize that there are often many
different ways of writing the final solution. It depends on
how it is to be used as to which is the most useful form.

- 383 -
Example 4.
Solve the equation
𝑑𝑦
𝑥 − 𝑦 = √𝑥 2 − 𝑦 2
𝑑𝑥
Solution.
This can be written as
𝑑𝑦 √𝑥 2 − 𝑦 2 + 𝑦
=
𝑑𝑥 𝑥
Substitute 𝑦 = 𝑣𝑥, we obtain

𝑑𝑣 √𝑥 2 − 𝑣 2 𝑥 2 + 𝑣𝑥 𝑥√(1 − 𝑣 2 ) + 𝑣𝑥
𝑥 +𝑣 = =
𝑑𝑥 𝑥 𝑥
Thus
𝑑𝑣
𝑥 + 𝑣 = √1 − 𝑣 2 + 𝑣
𝑑𝑥
Separate the variables
𝑑𝑣 𝑑𝑥
=
√1 − 𝑣 2 𝑥
Integrate sin−1 𝑣 = ln|𝑥 | + 𝑐.
𝑦
Put 𝑣 = and 𝑐 = ln 𝐴.
𝑥
𝑦
So sin−1 = ln|𝑥 | + 𝑐 = ln|𝐴𝑥 |.
𝑥

Therefore 𝑦 = 𝑥 sin(ln|𝐴𝑥 |) .◄

- 384 -
Example 5.
Solve the equation
𝑑𝑦
𝑥 =𝑥+𝑦
𝑑𝑥
Solution.
Re-arrange the given equation:
𝑑𝑦 𝑥 + 𝑦
=
𝑑𝑥 𝑥
Putting 𝑦 = 𝑣𝑥 gives:
𝑑𝑣 𝑥 + 𝑣𝑥
𝑥 +𝑣 = =1+𝑣
𝑑𝑥 𝑥
Simplify:
𝑑𝑣
𝑥 =1
𝑑𝑥
Separate the variables:
𝑑𝑥
𝑑𝑣 =
𝑥
Integrate:
1
𝑣 = ∫ 𝑑𝑥 = ln|𝑥 | + 𝐴
𝑥
𝑦 𝑦
Put 𝐴 = ln 𝐵 and 𝑣 = we obtain = ln|𝐵𝑥 |.
𝑥 𝑥

So, the general solution is 𝑦 = 𝑥 ln|𝐵𝑥 | .◄

- 385 -
Type (VI) Linear Equations
𝑑𝑦
A linear equation is one of the form + 𝑃(𝑥 )𝑦 = 𝑄 (𝑥 ).
𝑑𝑥

The method used to solve such equations depends, again,

on noticing a particular pattern.
If 𝑧 = 𝑦𝑒 𝑓(𝑥) then differentiating (using the product rule
and the chain rule
𝑑𝑧 𝑑𝑦 𝑓(𝑥) 𝑑𝑦
= 𝑒 + 𝑦𝑓′(𝑥 )𝑒 𝑓(𝑥) = 𝑒 𝑓(𝑥) ( + 𝑓′(𝑥 )𝑦)
𝑑𝑥 𝑑𝑥 𝑑𝑥
When compared to the linear equation above, this
suggests that if we can find an 𝑒 𝑓(𝑥) to multiply both
sides of the linear equation by then we’ll be able to
integrate the left-hand side immediately.
𝑑𝑦
Multiplying the linear equation + 𝑃(𝑥 )𝑦 = 𝑄 (𝑥 ) by
𝑑𝑥
𝑑𝑦
𝑒 𝑓(𝑥) gives 𝑒 𝑓(𝑥) ( + 𝑃(𝑥 )𝑦) = 𝑒 𝑓(𝑥) 𝑄(𝑥 )
𝑑𝑥
𝑑𝑦
Comparing left hand side with 𝑒 𝑓(𝑥) ( + 𝑓′(𝑥 )𝑦)
𝑑𝑥

(which can be integrated immediately) shows that we

want to find𝑓(𝑥), where𝑓 ′ (𝑥) = 𝑃(𝑥 ) ⇒ 𝑓(𝑥 ) =
∫ 𝑃(𝑥 )𝑑𝑥 i.e. multiply the equation by 𝑒 𝑓(𝑥) = 𝑒 ∫ 𝑃(𝑥)𝑑𝑥
which is called the Integrating Factor (I.F., for short).

- 386 -
Example 1.
Solve the equation
𝑑𝑦 2𝑥
− 3𝑦 = 𝑒 .
𝑑𝑥

Solution.
From above the Integrating Factor is
𝑒 ∫ −3𝑑𝑥 = 𝑒 −3𝑥
Multiplying through gives

−3𝑥
𝑑𝑦
𝑒 − 3𝑒 −3𝑥 𝑦 = 𝑒 −3𝑥 𝑒 2𝑥 = 𝑒 −𝑥
𝑑𝑥
The left-hand side is the differential of
𝑦𝑒 −3𝑥
Then
𝑑(𝑦𝑒 −3𝑥 )
= 𝑒 −𝑥
𝑑𝑥

𝑦𝑒 −3𝑥 = ∫ 𝑒 −𝑥 𝑑𝑥 = −𝑒 −𝑥 + 𝐴

Therefore, the general solution is

𝑦 = 𝑒 2𝑥 + 𝐴𝑒 3𝑥 .◄

- 387 -
Example 2.
Given
𝑑𝑦
2 cos 𝑥 + 4𝑦 sin 𝑥 = sin 2 𝑥.
𝑑𝑥
𝜋
Find 𝑦(𝑥) (y as a function of x), given that 𝑦 ( ) = 0.
3

Solution.
Divide through by 2 cos 𝑥 to write it in the form
𝑑𝑦
+ 𝑃(𝑥 )𝑦 = 𝑄 (𝑥 )
𝑑𝑥
𝑑𝑦 2 sin 𝑥 sin 2𝑥
+ 𝑦=
𝑑𝑥 cos 𝑥 2 cos 𝑥
𝑑𝑦
+ 2(tan 𝑥 )𝑦 = sin 𝑥
𝑑𝑥
The equation is linear, integrating factor
2𝑥
𝑒 ∫ 2 tan 𝑥 𝑑𝑥 = 𝑒 ln sec = sec 2 𝑥
Multiply by I. F. We obtain
𝑑𝑦
sec 2 𝑥 + (2 sec 2 𝑥 tan 𝑥 )𝑦 = sec 2 𝑥 sin 𝑥
𝑑𝑥
Then
𝑑 (𝑦 sec 2 𝑥 ) sin 𝑥
= sec 2 𝑥 sin 𝑥 =
𝑑𝑥 cos2 𝑥
sin 𝑥 1
Integrate 𝑦 sec 2 𝑥 = ∫ 𝑑𝑥 = + 𝐴.
cos2 𝑥 cos 𝑥

- 388 -
Hence
𝑦 = cos 𝑥 + 𝐴 cos 2 𝑥.
𝜋
But 𝑦 = 0 when 𝑥 = .
3

Hence
𝜋 𝜋
0 = cos ( ) + 𝐴 cos 2 ( )
3 3
1 1
0 = + 𝐴 ⇒ 𝐴 = −2
2 4
Particular solution
𝑦(𝑥) = cos 𝑥 − 2 cos 2 𝑥.◄
Example 3.
𝑑𝑦
Solve the equation −𝑦 =𝑥
𝑑𝑥

Solution.

The I. F. is 𝑒 ∫ −1𝑑𝑥 = 𝑒 −𝑥 giving

𝑑𝑦 𝑑(𝑦𝑒 −𝑥 )
𝑒 −𝑥 − 𝑦𝑒 −𝑥 = 𝑥𝑒 −𝑥 . Hence = 𝑥𝑒 −𝑥 .
𝑑𝑥 𝑑𝑥

So 𝑦𝑒 −𝑥 = ∫ 𝑥𝑒 −𝑥 𝑑𝑥.
Integrating the right side by parts we obtain
𝑦𝑒 −𝑥 = −𝑥𝑒 −𝑥 − 𝑒 −𝑥 + 𝐴.
Thus 𝑦 = 𝐴𝑒 𝑥 − 𝑥 − 1.◄

- 389 -
 Exercise Set (8.1)
1. Classify the given differential equation by order and
tell whether it is linear or non-linear.
(a) 𝑦’ + 2𝑥𝑦 = 𝑥 2 ;
(b) 𝑦’ (𝑦 + 𝑥) = 5;
(c) 𝑦 sin 𝑦 = 𝑦’’;
(d) 𝑦 cos 𝑦 = 𝑦’’’;
(e) cos 𝑦 𝑑𝑦 = sin 𝑥 𝑑𝑥;
(f) 𝑦’’ = 𝑒 𝑦 .
2. State whether the given differential equation is linear
or non-linear. Write the order of each equation.
(a) (1 − 𝑥 )𝑦’’ − 6𝑥𝑦’ + 9𝑦 = sin 𝑥;
𝑑3 𝑦 𝑑𝑦 2
(b) 𝑥 − 2 ( ) + 𝑦 = 0;
𝑑𝑥 3 𝑑𝑥

(c) 𝑦𝑦’ + 2𝑦 = 2 + 𝑥 2 ;
𝑑2 𝑦
(d) + 9𝑦 = sin 𝑦;
𝑑𝑥 2
1
2 2
𝑑𝑦 𝑑2 𝑦
(e) = (1 + ( ) );
𝑑𝑥 𝑑𝑥 2

𝑑2 𝑟 𝑘
(f) = − 2.
𝑑𝑡 2 𝑟

- 390 -
3. Verify that, the indicated function is a solution of the
given differential equation. In some cases, assume an
appropriate interval
(a) 2𝑦’ + 𝑦 = 0; 𝑦 = 𝑒 −𝑥/2 .
(b) 𝑦’ = 25 + 𝑦 2 ; 𝑦 = 5 tan 5𝑥.
(c) 𝑥 2 𝑑𝑦 + 2𝑥𝑦 𝑑𝑥 = 0; 𝑦 = −1/𝑥 2 .
(d) 𝑦’’’ − 3𝑦’’ + 3𝑦’ − 𝑦 = 0; 𝑦 = 𝑥 2 𝑒 𝑥 .
(e) 𝑦’ = 𝑦 + 1; 𝑦 = 𝑒 𝑥 − 1.
(f) 𝑦’’ + 9𝑦 = 8 sin 𝑥;
𝑦 = sin 𝑥 + 𝑎 cos 3𝑥 + 𝑏 sin 3𝑥.
4. Solve the following differential equations
𝑑𝑦
(i) 𝑥 = 𝑦 2 ;
𝑑𝑥
𝑑𝑦
(ii) 𝑥 (𝑥 2 + 1) = ;
𝑑𝑥
2𝑥𝑦 𝑑𝑦
(iii) = ;
𝑥 2 +1 𝑑𝑥
𝑑𝑦 𝑥 2 +𝑥
(iv) = ;
𝑑𝑥 𝑦 2 +𝑦
𝑑𝑦
(v) cos2 𝑥 = cos2 𝑦;
𝑑𝑥
𝑑𝑦
(vi) 𝑒 𝑥+𝑦 = 1;
𝑑𝑥
𝑑𝑦 𝑦−1
(vii) 𝑥 = − 𝑦;
𝑑𝑥 𝑦+1

- 391 -
 𝑑𝑦
(viii) (𝑥 + 1)𝑦 = 𝑥 , where 𝑦(1) = 1;
𝑑𝑥
𝑑𝑦 𝜋
(ix) tan 𝑥 = cot 𝑦 where 𝑦(0) = ;
𝑑𝑥 2

5. Find the equation of the curve which has gradient

𝑑𝑦 𝑦
function = and passes through the point (2, −1).
𝑑𝑥 𝑥 2 −1

6. Solve the following differential equations

𝑑𝑦 𝑥−2𝑦
(i) = ;
𝑑𝑥 𝑥
𝑑𝑦 𝑥 2 +𝑦2
(ii) = ;
𝑑𝑥 𝑥𝑦
𝑑𝑦
(iii) 2𝑥 2 = 𝑥 2 + 𝑦2;
𝑑𝑥
𝑑𝑦
(iv) 𝑥𝑦 2 = 𝑥 3 − 𝑦3 ;
𝑑𝑥
𝑑𝑦
(v) 𝑥 = 𝑦 + √𝑥 2 − 𝑦 2 ;
𝑑𝑥
𝑑𝑦
(vi) 𝑦 = −𝑥 + 2𝑦;
𝑑𝑥
𝑑𝑦
(vii) 𝑥 (𝑥 2 + 𝑦 2 ) = 𝑦(𝑦 2 − 𝑥 2 );
𝑑𝑥
𝑑𝑦
(viii) (𝑥 3 + 𝑥𝑦 2 ) = (𝑥 2 𝑦 − 𝑦 3 ).
𝑑𝑥

7. Solve the following differential equations

𝑑𝑦
(i) + 𝑦 = 𝑒 −𝑥 ;
𝑑𝑥
𝑑𝑦
(ii) − 𝑥𝑦 = 0;
𝑑𝑥

- 392 -
 𝑑𝑦 𝑦 1
(iii) + = ;
𝑑𝑥 𝑥 𝑥
𝑑𝑦
(iv) 𝑥 − 𝑦 = 𝑥 2;
𝑑𝑥
𝑑𝑦
(v) +𝑦 =𝑥;
𝑑𝑥
𝑑𝑦
(vi) − 2𝑦 = 𝑥;
𝑑𝑥
𝑑𝑦
(vii) + 𝑦 cot 𝑥 = cos 𝑥;
𝑑𝑥
𝑑𝑦
(viii) 𝑥 − 2𝑦 = 𝑥 3 ;
𝑑𝑥
𝑑𝑦
(ix) − 𝑦 tan 𝑥 = cos2 𝑥;
𝑑𝑥
𝑑𝑦
(x) (𝑥 2 − 1) + 2𝑥𝑦 = 𝑥.
𝑑𝑥

- 393 -
 References

[1] Axler, Sheldon (February 26, 2004), Linear Algebra Done

Right (2nd ed.), Springer, ISBN 978-0-387-98258-8

[2] Bretscher, Otto (June 28, 2004), Linear Algebra with

Applications (3rd ed.), Prentice Hall,

[3] Garner, Lynn E. (1991), Calculus and analytic geometry,

Dellen Publishing Company,.

[4] Hazewinkel, Michiel, ed. (2001), Differential equation,

ordinary, Encyclopedia of Mathematics, Springer, ISBN 978-1-

55608-010-4

[5] Kolman, Bernard; Hill, David R. (May 3, 2007), Elementary

Linear Algebra with Applications (9th ed.), Prentice Hall,

ISBN 978-0-13-229654-0

[6] Shores, Thomas S. (December 6, 2006), Applied Linear

Algebra and Matrix Analysis, Undergraduate Texts in

Mathematics, Springer, ISBN 978-0-387-33194-2

[7] Smith, Larry (May 28, 1998), Linear Algebra, Undergraduate

Texts in Mathematics, Springer, ISBN 978-0-387-98455-1

- 394 -