Algebra

(Selected Topics)

BY

Dr. Mohamed Ismail Mohamed Hessein

Professor Of Mathematics
Faculty Of Engineering ( Shoubra )
Benha University
Algebra Preface

PREFACE
Algebra, which has a glorious history of more than
three thousand years, might be very well called a
universal language of civilization. It provides a
foundation upon which higher mathematics is built and it
is the language of modern Science and Technology. Many
problems, which would provide difficult to one armed
only with arithmetic, become far easier when translated
into algebra.
Algebra of a sort was however known to the
ancient Egyptians, and Ahmes, a scribe who lived more
than 1000 years B.C., used a method similar to that of the
solution of simple equations. For the unknown number in
his equations he used a symbol that for the word “heap”.
This method of symbolical representation was not
followed up by the classical Greeks but Diaphanous of
Alexandria (about 350 A.D.) invented a kind of algebra
and used symbols freely for arithmetical numbers and also
to express certain other ideas or operations such as
equality, addition, subtraction.
Western algebra arose with the introduction of the
Science from India to the Arabs by Mohamed Ben Musa
Alkorismi about 830 A.D. It was fostered by Arabs and

3
Algebra Preface

brought to Europe by Leonardo of Pisa (born about 1175)

in an Italian work based upon the treatise of Alkorismi.
The Arabs had real ability in devising proper problems
and methods for instruction. For further centuries
numerous mathematical texts continued to present the
Arabic material frequently quite unaltered.
In the twentieth century algebraic studies expanded
considerably and algebra as we already know occupies a
very special place of honour in mathematics. New
divisions of algebra have sprung up.
This book presents brief introductions to some
selected topics in algebra. The emphasis is on the
computational aspects of the subject keeping the
abstractions down to a minimum. Thus we sometimes
omit proofs of difficult or less rewarding theorems while
amply illustrating them with examples.
The contents are divided into seven chapters.
Chapter 1 deals with Mathematical Induction. Chapter 2
devoted to Partial Fractions. Chapter 3, deals with Finite
Series. Chapter 4, devoted to the Binomial Theorems.
Chapter 5 deals with Matrices and their applications,
Determinants in a brief manner eigenvalues and
eigenvectors settles the digitalisation problem for

4
Algebra Preface

symmetric matrices. Chapter 6, deals with theory of

equations specially Horner’s Method, Bi-quadratic
equations, Cubic Equations (Cardan's method), Quartic
Equations Ferrari's Solution and Reciprocal Equations.
Chapter 7, devoted to find the roots of Algebraic and
transcendental equations by different methods.
Even after the most careful checking it is quite
possible that some errors have been overlooked and I shall
be greatly indebted to anyone submitting corrections. It
has been gratifying to receive letters from students who
have used this book giving us their reactions to it pointing
out errors and misprints in it and offering suggestions for
its improvement. Should you be inclined to do so we
would be glad to hear from you also? Now good luck and
every success.
Dr. Mohamed Ismail Mohamed Hessein
Professor of Mathematics
Faculty of Engineering (Shoubra)
Benha University

5
Algebra Contents

CONTENTS

Chapter Page
PREFACE……………………………………………….….. 3
CONTENTS……………………………………………….... 6
1 MATHEMATICAL INDUCTION……………….. 9
EXERCISES 1………………………………………………. 15
2 Partial Fractions……………………………. 18
§ 2.1 Polynomials…………………………………………………... 18
Resolution of rational functions into partial 25
§ 2.2
fractions……………………………………………….………
§ 2.3 Methods for determining coefficients……………... 26
§ 2.4 EXERCISES 2………………………………………………. 38
3 FINITE SERIES…………………………………………. 40
§ 3.1 Summation of series………………………………………. 41
§ 3.2 The difference method…………………………………… 45
§ 3.3 EXERCISES 3………………………………………………. 52
4 BINOMIAL EXPANSION………………………….. 56
Binomial expansion for negative and fractional 56
§ 4.1
indices……………………………………………….…………..
§ 4.2 Determination of a binomial function from its
expansion……………………………………………….……... 62
EXERCISES 4………………………………………………. 69

6
Algebra Contents

CONTENTS
5 MATRICES………………………………………………… 76
§ 5.1 Matrix Algebra……………………………………………… 77
§ 5.2 Products by partitioning…………………………………. 82
§ 5.3 Determinants…………………………………………………. 84
§ 5.4 Inverse of a matrix………………………………………… 86
§ 5.5 Cramer's Rule………………………………………………... 87
§ 5.6 Linear dependence of vectors…………………………. 89
§ 5.7 The rank of a matrix ……………………………………... 92
§ 5.8 System of linear equations……………………………... 100
§ 5.9 Eigenvalues and eigenvectors………………………… 107
Reduction of a matrix to Diagonal canonical
§ 5.10
form………………………………………………...……………. 110
§ 5.11 Quadratic form………………………………………………. 117
EXERCISES 5………………………………………………. 121
6 Theory of Equations……………………….. 132
§ 6.1 Some Elementary Properties…………………………... 132
Equations with given Roots……………………………. 132
Symmetric Relations between Roots and
§ 6.2
Coefficients……………………………………………….…... 134
§ 6.3 Evaluation of a Polynomial……………………………. 140
Repeated Roots……………………………………………… 143
§ 6.4 Transformation of Equations………………………….. 144

7
Algebra Contents

CONTENTS
Horner’s Method of Reducing an Algebraic
Equation………………………………………………...……… 149
Equations with Integer Coefficients………………... 153
To Find the Rational Roots of a Rational
Equation………………………………………………...……… 154
Bi-quadratic equations…………………………………… 157
Solution of the Bi-quadratic Equation……………... 158
Equation reduced to quadratic equation…………... 159
Binomial Equations……………………………………….. 160
De Moiver’s Theorem……………………………………. 162
Cubic Equations (Cardan's method) ……………….. 166
Quartic Equations Ferrari's Solution……………….. 173
Reciprocal Equations……………………………………... 178
EXERCISES 6………………………………………………. 183
7 Calculations methods…………………… 190
§ 7.1 Descartes' Rule of Signs………………………………… 190
Detection of Imaginary Roots from Descartes’
Rule……………………………………………………………… 191
§ 7.2 Approximate Methods…………………………………… 194
§ 7.2.1- Methods to Algebraic equations
only………………………………………………...……………. 195

8
Algebra Contents

CONTENTS

Chapter Page
§ 7.2.2-Simplifications in the case of Horner’s
Method………………………………………………...……….. 199
§ 7.2.3-Roots nearly Equal……………………………... 202
Methods to Algebraic and transcendental
§ 7.3
equations……………………………………………….……… 204
§ 7.3.1- graphical method………………………………. 204
§ 7.3.2- Method of halving the interval…………… 208
§ 7.3.3- Iteration method
(successive approximations) ………………………….. 213
§ 7.3.4- Newton- Raphson’s Method………………. 216
§ 7.3.5- The false position method (Regula falsi
method)………………………………………………...……… 221
§ 7.4 Special cases……………………………………………….… 223
§ 7.4.1- Fixed point method…………………………… 223
§ 7.4.2- The secant (chord) method………………… 224
EXERCISES 7………………………………………………. 229
REFERENCES……………………………………………… 233

9
Algebra (Chapter 1) Mathematical Induction

CHAPTER I
MATHEMATICAL INDUCTION
As early as 1889 Peano showed that all the properties of
the natural numbers N = {1, 2, 3, ...} can be derived from five
axioms which now bear his name Peano’s axiom system:
1) 1∈ N.
+
2) Every number n ∈N has precisely one successor n .
+
i.e. ∀ n ∈ N ⇒ n = n + 1∈ N.
3) 1 is not the successor of any natural number.
+
i.e. if n = n + 1 = 1 ⇒ n ∉N.
4) Distinct numbers have distinct successors.
+ +
If m, n ∈ N, m ≠ n then m ≠ n .
5) If A be a subset of natural numbers A ⊂ N, contains the
number 1 and contains, together with any number, also its
successor then it is the set of all natural numbers i.e. A = N.
According to these axioms every natural number other 1
can be described as a successor or as a successor of a successor of
+ + +
... of a successor of 1. But instead of the notation 1, 1 , (1 ) , …
one uses simply the decade positional system and denotes the
numbers by 1, 2, 3, ... . In particular Axiom 5 gives the
justification for the method of mathematical induction.
§1.1 Mathematical Induction:
This method of reasoning is used to prove that a

10
Algebra (Chapter 1) Mathematical Induction

proposition (or theorem or formula) P(n) concerning natural

numbers n is valid or not for all natural numbers. In using
mathematical induction as a method of proof one considers the set
M ⊂ N of all natural numbers for which P(n) is true. If one can
show:
1) That 1∈M, in words that the proposition is valid for n =1, (basis
of induction) and
2) That from m ∈M it follows that m+1∈M, in words that the
validity of the preposition for any natural number m (inductive
hypothesis) implies its validity for the successor m (inductive
step) i.e. the truth of P(m) implies the truth of P(m+1) then by
Peano’s fifth axiom M must be the set of all natural numbers.
i.e. M = N. i.e. all the propositions P(1), P(2), P(3), ..., P(m),... are
true. i.e. P(n) is true for every positive integer n.
To apply this principle we need first to show that P(1) is true and
second to show that P(m)⇒P(m+1) (the symbol ⇒ is read
“implies”).
Example 1.1:
Using mathematical induction to show that for all positive integral
values of n , 1 + 2 + 3+ ... + n = 1 n(n+1). (1)
2

Solution:
In the present example let P(n) be the proposition that
formula (1) is correct for the integer n.

11
Algebra (Chapter 1) Mathematical Induction

Step 1: The formula (1) is true for n = 1, since ( 1 = 1 (1)(2) )

2
i.e. P(1) is correct.
Step 2: Assume that the formula is true for n = m.
i.e. we assume that m is an integer for which P(m) is true:
1 + 2 + 3+ … + m = 1 m (m+1). (2)
2
If we add (m+1) to both sides of this equality we get:
1 + 2 + 3+ … + m + (m+1) = 1 m (m+1)+(m+1)= 1 (m+1)(m+2),
2 2
which is the value of 1 n(n+1) when (m+1) is substituted for n.
2
Hence if the formula is true for n = m we have proved it to be true
for n = m+1. But the formula holds for n = 1; hence it holds for
n = 2. Then since it holds for n = 2 it holds for n = 2 + 1 = 3, and
so on. Thus the formula is true for all positive integral values.
Example 1.2:
Prove by mathematical induction that for all positive integral
2
values of n , 1 + 3 + 5 + … + (2n–1) = n .
Solution:
Step 1: The formula is true for n = 1 since 1 = 1.
Step 2: Assume that formula is true for n = m,
2
then 1 + 3 + 5 + … + (2m–1) = m . (1)
th
Add (m+1) term, which is (2m+1) to both sides of the above
equation. Then:
2 2
1 + 3 + 5 + … + (2m–1) + (2m+1) = m + (2m+1)= (m+1) .

12
Algebra (Chapter 1) Mathematical Induction
2
The right hand side of this equation (m+1) , which is the value of
2
n when n is replaced by (m+1). Hence the truth of P(m+1)
follows from P(m) provided it is true. But the formula holds for
n = 1; hence it holds for the next integer n = 2. Then since it holds
for n = 2 it holds for the next integer n = 2+1= 3 and so on, by the
principle of M.I. then the formula is true for all positive integral
values of n.
Example 1.3:
n
Prove by M.I. that a – 1, is divisible by a – 1, a ≠ 1, when n is any
(2n+2)
positive integer then deduce that 3 – 8n–9 is divisible by 64.
Solution:
Step 1: The theorem is true for n = 1, since a – 1 is divisible by
itself .
Step 2: Assume that the theorem is true for n = m.
m m+1
Then a – 1 is divisible by a – 1. We must show that a – 1 is
divisible by a – 1.
From the identity:
m+1 m m+1
a – a + a – 1 = a(a – 1) + (a – 1), it follows that a – 1 is
m
divisible by a – 1 if a – 1 is.
Hence if the theorem is true for n = m, we have proved it
to be true for n = m + 1. But the theorem holds for n = 1 hence it
holds for the next integer n = 2. Then, since it holds for n = 2 it
holds for the next integer n = 3, and so on. Thus the theorem is

13
Algebra (Chapter 1) Mathematical Induction
2(n+1)
true for all n. To prove the second part: let f(n) = 3 – 8n – 9.
4
Step 1: f(1) is divisible by 64 since f(1) = 3 – 8 – 9= 81 –17= 64.
Step 2: Assume that f(m) is divisible by 64. Then
2(m+2) 2(m+1)
f(m+1) – f(m) = [3 – 8(m+1) – 9]– [3 – 8m – 9]=
2(m+1) (m+1)
= [3 (9–1) – 8] = 8[9 – 1].
(m+1)
From the first part (take a = 9); 9 is divisible by a – 1=8.
Thus f(m+1) – f(m) is divisible by 8 × 8 = 64.
Hence f(m+1) is divisible by 64 since f(m) is. By the principle of
M.I. f(n) is divisible by 64.
Example 1.4:
Discuss the validity of the formula
2 2 2
1 + 2 +…+ n = 1 (n+1)(7n – 4) (1)
6

Solution:
The formula (1) is true for n = 1, since 1 = 1 (2)(3)= 1.
6
The formula (1) is also true for n = 2 since 5= 1 (3)(10) = 5, we
6

have verified the truth of P(1) and P(2). Such verification is

necessary only for P(1) but the additional checking was easy and
not harmful.
Now we see if we can show that P(m) ⇒ P(m+1) for any positive
integer m. To test whether this is so, we assume that m is an
integer for which P(m) is true:
2 2 2
1 + 2 +…+ m = 1 (m+1)(7m – 4) (2)
6
2
If we add (m +1) to both sides of this equation we get

14
Algebra (Chapter 1) Mathematical Induction
2 2 2 2 2
1 + 2 +…+ m + (m+1) = 1 (m+1)(7m – 4) + (m+1) =
6
= 1 (m+1)[(7m – 4) + 6(m+1)] = 1 (m+1)(13m+2) (3)
6 6

Now P(m+1) is the proposition that the sum of the squares on the
left above should equal 1 (n+1)(7n–4) with n = m+1 ; in other
6
words it should equal 1 (m+2)(7m+3).
6

Hence the truth of P(m+1) follows from P(m) provided it is true

that 1 (m+2)(7m+3)= 1 (m+1)(13m+2).
6 6
2 2
i.e. 7m + 17m + 6 = 13m + 15m + 2 ,
2 2
or 6m – 2m – 4 = 0 or 3m – m – 2 = 0
or (3m+2) (m–1)= 0
i.e. m = 1, and we exclude m = – 2 . Hence the truth of P(m +1)
3

follows from P(m) provided that m = 1 only. Therefore the

formula is true when n ∈ {1 , 2}.

15
Algebra (Chapter 1) Mathematical Induction

EXERCISES 1
By the method of mathematical induction prove the validity of the
following statements for all positive integers n.
3 3 3 2
1) 1 + 2 +…+ n = [ 1 n(n+1)] .
2
2) 1.2 + 3.4 + 5.6 +... +(2n–1)(2n) = 1 n(n+1)(4n–1).
3
n 2 2 n-1 n
3) i) ∑ (2k − 1) = n , ii) 1 + 3 + 3 +… + 3 = 1 (3 – 1).
2
k =1
n 1 n
4) ∑ =
k =1 ( 2k − 1)(2k + 1) 2n + 1
n 1 3 2n + 3
5) i) ∑ = – ,
k =1 k ( k + 2) 4 2(n + 1)(n + 2)
1 1 1 1 n
ii) + + + ... + = .
2.5 5.8 8.11 (3n − 1)(3n + 2) 6n + 4
n
6) i) ∑ k 2 = 1 n(n+1)(2n+1).
6
k =1
1 1 1 n (n + 3)
ii) + + ... + = .
1.2.3 2.3.4 n (n + 1)(n + 2) 4(n + 1)(n + 2)
2 2 2 2 n–1 2 n–1
7) 1 – 2 + 3 – 4 +…+ (–1) n = 1 (–1) n(n+1).
2
n
8) ∑ r(r + 2) = 16 n(n+1)(2n+7).
r =1
2 2 (2n − 1) 3n +1 + 3
n
9) 1.3 + 2.3 + 3.3 +... + n. 3 = .
4
n
2 1 n(4n2–1).
10) ∑ (2k − 1) = 3
k =1
2 2 2 2
11) 1 + 4 + 7 + … to n terms = 1 n(6n – 3 n – 1).
2

16
Algebra (Chapter 1) Mathematical Induction
n
12) ∑ (n − r)(r − 1) = 16 n(n – 1)(n – 2).
r =1
3
13) Prove that n – n is divisible by 3 for every positive
integer n.
3 3 2
[Hint: Use (m + 1) – (m + 1) = (m – m) + 3(m + m)]
[Note: This is a special case of a theorem of Fermat that states that
p
n – n is divisible by p, if p is a prime.]
14) Using mathematical induction to prove the validity of the
following inequalities:
n
i) (1+ h) > 1+ n h, if h > –1, n ≥ 2 Bernoulli’s inequality.
n 2
ii) (1+ h) ≥ 1 + n h + 1 n(n–1) h , if h ≥ 0.
2
n 2
iii) (1– h) ≤ 1 – n h + 1 n(n–1) h , if 0 ≤ h ≤ 1.
2

15) Given that ur+1 = 2ur + 1, where r is a positive integer

n–1
prove that un = 2 (u1 + 1 ) – 1, and if u1 = 1 then prove that
n n+1
∑ ur = 2 – (n+2).
r =1
3 3 3 3
16) Given that 4= = u1 + = u2 + = u3 + =… prove
u1 u2 u3 u4
3n +1 − 3
that un = .
3n +1 − 1
17) Given that un+1 = 3un – 2un–1 and u0 = 2, u1 = 3, prove that
n
un = 2 + 1.
n
18) Let un+1= 3un – 2un–1 and u0 = 0, u1 = 1, prove that un = 2 +1.
19) Prove by mathematical induction that when n is any positive

17
Algebra (Chapter 1) Mathematical Induction

integer:
n n
i) a – b , is divisible by a – b,
2n 2n
ii) a –b , is divisible by a + b,
2n+1 2n+1
iii) a +b , is divisible by a + b.
3 2
iv) Prove that for all integers n ≥ –2, 2n – 3n + n + 31 ≥ 0.
20) Prove the binomial theorem for positive integral values of n.
21) Discuss the validity of the following statements:
n 2
a) ∑ r 2 = 1 n (n+1) ,
2
r =1
2 n-1 n
b) 1 + 2 + 2 +… + 2 = (2 – 1).
1 1 1 2 3
c) + + ... + =n [ − 1].
1.2 2.3 n (n + 1) n +1
22) Prove that:
4 3 2
i) n + 2n + n , is divisible by 4,
n
ii) 4 – 1, is divisible by 3,
3
iii) n + 2n, is divisible by 3 .
23) Prove that:
2n
i) f(n) = 3 + 7, is divisible by 8,
2
ii) g(n) = (2n +1) – 1, is divisible by 8,
3n
iii) h(n) = 2 – 7n – 1, is divisible by 49.
24) Prove that:
3n+3
i) f(n) = 2 – 7n – 8, is divisible by 49,
2n 2
ii) g(n) = 6 – 35n – 1, is divisible by (35) .

18
Algebra (Chapter 2) Partial Fractions

CHAPTER 2
Partial Fractions
In the preparatory schools, we learned how to combine
fractions over a common denominator. Now, it is desirable to
reverse the process and split a function into a sum of fractions
with simpler denominators. The technique for doing so is called
the method of partial fractions. We shall look at some examples,
present the method in general terms, and then describe a shortcut
called the Heaviside “Cover-up” method (invented by Oliver
Heaviside [1850 – 1925], a pioneer in electrical engineering and
vector analysis).
Example 2.1:
1 4 ( x − 2) + 4( x + 1) 5x + 2
+ = = .
x +1 x − 2 ( x + 1)( x − 2) 2
x −x−2
Adding the fraction on the left produces the fraction on the right.
The reverse process consists in finding constants A and B such
5x + 2 A B
that: = + .
x2 − x − 2 x +1 x − 2
(Pretend, for a moment, we don't know that A = 1, B = 4 will
work.) We call A and B undetermined coefficients.
§ 2.1 Polynomials:
An expression of the form
n n–1
a0 x + a1 x +… + an–1 x + an , where n is a natural number, the
coefficients ar are arbitrary real numbers, and a0 ≠ 0, is called a

19
Algebra (Chapter 2) Partial Fractions

polynomial of degree n. ℜ is always regarded as the domain of

definition.
Theorem 2.1:
A polynomial can only equal zero for all values of x, if the
coefficient of each power of x is equal to zero, and if the sum of
the terms not involving x is equal to zero i.e. ar = 0, r = 0,1, 2 ...n.
Theorem 2.2:
If two polynomials of x are equal for all values of x, then:
i) The highest power of x in each is the same and,
ii) The coefficients of the corresponding powers of x are equal.
Theorem 2.3:
If a polynomial of x can be divided by x – a, the remainder will be
the same function of a as the dividend is of x.
Corollary 1: If a polynomial of x becomes zero when a, is

substituted for x then, (x – a) is a factor of the polynomial.

Corollary 2: If a polynomial of x becomes zero when a, b, c,

etc., are in turn substituted for x, it is divisible by x – a, x– b, x – c

etc., and consequently by the continued product of these factors.
The values of x for which P(x) vanishes are called the
zeros of P(x); they are the roots of the equation P(x) = 0. ,
Identities & Equations: We must distinguish carefully

between an identity and an equation. When a polynomial or other

20
Algebra (Chapter 2) Partial Fractions

function P(x) is thrown into another form Q(x) the relation P(x) =
Q(x) is called an identity, as for example:
2
2x – 5x + 4 = (2x–1) (x–2),
2
It is clear that, (x+1) (x–1) – x + 1 = 0, is a quadratic identity.
2
On the other hand, (x+1) (x–1) – x + 2x = 0, is not a quadratic
but a linear equation.
Divisors:
In general, a natural number a, is said to be divisible by another b
if there exists a natural number n such that a = n . b; b as well as n
are then called divisors of a. On the other hand, a is called a
multiple of b and of n. Every number a ≠ 0 is divisible by 1 and by
itself; these divisors are called improper.
Uniqueness of polynomial representation:
The assumption that two different polynomials can represent the
same rational function leads to a contradiction. Because the two
polynomials:
n n–1
P(x) = a0 x + a1 x +… + an–1 x + an ,
n n–1
Q(x) = b0 x + b1 x +… + bn–1 x + bn ,
are assumed to be different.
Factorisation of polynomials:
A polynomial P(x) of degree n ≥ 1 is called reducible if it can be
expressed as a product of polynomials of lower degree. If such a
representation is not possible, the polynomial is called irreducible.

21
Algebra (Chapter 2) Partial Fractions

Polynomials of degree zero are constants; they are excluded from

this classification, since they are neither reducible nor irreducible.
Polynomials of the first degree a x + b, are then always
irreducible.
If a polynomial of degree n is reducible, that is, if it can be
split up as a product P(x) = Q(x) R(x), then the polynomials Q(x)
and R(x) must be of degree at least equal to 1 and necessarily
smaller than n. If Q(x) or R(x) is reducible, the process can be
repeated; after at most n steps the polynomial is factorised into a
product. P(x) = P1 (x) P2 (x) ... Pm (x), m ≤ n.
The factorisation of a reducible polynomial is unique apart from
the order and up to constant factors.
Example 2.2:
2 2
x – 1 is reducible and x + 1 is irreducible over the domain of
2
real numbers. (Whereas it factorise into x + 1 = (x+i) (x–i) over
the domain of complex numbers).
If arbitrary complex numbers are allowed for the coefficients of
the polynomial then the fundamental theorem of algebra shows
that each polynomial function of degree n can be factorised into n
linear factors ( x – αk), k = 1, 2, ... ,n. The numbers αk are the
zeros of the function. In the case of polynomials with real
coefficients if one of these numbers α = a + ib is complex then
the conjugate complex number α = a – ib also occurs as a zero.

22
Algebra (Chapter 2) Partial Fractions

The product of the corresponding linear factors is then

2 2 2
(x – α )(x – α ) = (x – a – ib)(x – a + ib) = x – 2ax + (a + b ),
that is, a quadratic polynomial with real coefficients. If all pairs of
conjugate complex linear factors are collected together in this
way, then over the field of real numbers the irreducible factors are
either real linear factors or quadratic polynomials with real
coefficients. When the coefficients are restricted to the field of
real numbers it follows that all irreducible polynomials are of
degree at most 2.
Theorem 2.4: Factors of polynomials:
Every polynomial whose coefficients are real can be resolved into
real factors of the first and second degree.
If a+ib is a root of the equation the conjugate number a–ib
is also a root. The corresponding factors are (x–a–ib)(x–a+ib) =
2 2
= (x–a) + b . Thus the two roots give rise to a real quadratic
factor. If the factor (x–a–ib) is repeated k times then it follows that
factor (x–a–ib) must also be repeated k times. These factors give
2 2 k
rise to the real factor [(x–a) + b ] . Thus the factors of Q must be
represented in one or other of the following forms, where a, b, c,
are real.
i) A non-repeated simple factor of the form x–a;
k
ii) A repeated simple factor of the form (x–a) ;
2
iii) A non-repeated quadratic factor of the form ax + bx + c;

23
Algebra (Chapter 2) Partial Fractions
2 k
iv) A repeated quadratic factor of the form [ax + bx + c] .
Example 2.3:
3 2
The polynomial P(x) = x – 5x + 7x – 3 has the zero x = 3.
2
Division by (x – 3) gives the quotient x – 2x + 1, so that the
polynomial can be expressed in the form:
2 2
P(x) = (x – 3) (x – 2x + 1) = (x – 3) (x – 1) .
Theorem 2.5:
A polynomial P(x) of degree n has at most n distinct zeros.
Multiplicity of a zero: It can happen that a polynomial
2
with a zero a is divisible not only by (x – a) but also by (x–a) ,
3
(x–a) , or a still higher power of (x–a).
k k+1
If P(x) is divisible by (x–a) , but not by (x–a) , then a is called
a k-fold zero or a zero of multiplicity k (k ≥ 1, an integer) of P(x).
If k =1, a simple zero, if k = 2, a double zero.
Example 2.4:
4 3 2
The polynomial P(x) = x – 9x + 27x – 31x +12, has a double
2 3
zero for x = 1, that is, it is divisible by (x–1) , but not by (x–1) .
2 2 2
∴ P(x) = (x–1) (x – 7x + 12) = (x–1) (x–3)(x–4).
Prime polynomials:
Two polynomials are said to be prime to each other or relatively
prime if they have no common factor apart from numerical
2
factors. Thus e.g. 3(x+1), 6(x – x + 1) are prime to each other

24
Algebra (Chapter 2) Partial Fractions

(co-prime).
Rational functions:
A function F(x) = P(x)/Q(x), where P(x) and Q(x) are
polynomials is called a rational fraction.
If the degree of P(x) is less than the degree of Q(x); F(x) is
called proper; otherwise F(x) is called improper. Division in the
usual way may express an improper fraction as the sum of a
polynomial and a proper fraction. If the degree of P and Q are
equal this polynomial will be a constant. Thus e.g.
x3 x x2 + x + 1 x+2
=x+ and =1+ .
x2 −1 x2 −1 x2 −1 x2 −1

Theorem 2.6: Equality of two rational functions:

Let P, P′; Q , Q′; R, R′ be polynomials such that
P P' P P'
+R= + R′, where and are proper fractions.
Q Q' Q Q'
P P'
Then R = R′ , ≡ .
Q Q'
Theorem 2.7: Resolution into partial fractions:
R
If P , Q are co-prime the rational fraction can be expressed
P.Q
X Y
uniquely in the form T + + , where T is a polynomial and
P Q
X Y
, are proper and irreducible.
P Q

25
Algebra (Chapter 2) Partial Fractions

Repeated applications of it show that:

If P1, P2, ..., Pn are polynomials every two of which are co-prime,
R
the rational fraction can be expressed uniquely in the
P1.P2 ...Pn
X X X
form T + 1 + 2 +…+ n , where T is a polynomial and
P1 P2 Pn
X1 X 2 X
, ,…, n are proper and irreducible.
P1 P2 Pn
§ 2.2 Resolution of rational functions
Into partial fractions:
If the given rational function is improper it should first be
expressed as the sum of a polynomial and a proper fraction. We
now have the following theorem:
Theorem 2.8:
i) To the non-repeated factor (x – a) of Q there corresponds a
A
fraction of the form .
x−a
k
ii) To the repeated factor (x–a) , corresponds a group of terms:
A1 A2 A3 Ak
+ + +…+ .
(x − a ) (x − a ) 2 (x − a )3 (x − a ) k
2
iii) To the non-repeated factor ax + bx + c, corresponds a fraction
Bx + C
of the form: .
2
ax + bx + c
2 k
iv) To the repeated factor (ax + bx + c) , corresponds a group of
B1x + C1 B2 x + C 2 Bk x + C k
terms + +…+ .
(ax 2 + bx + c) (ax 2 + bx + c) 2 (ax 2 + bx + c) k

26
Algebra (Chapter 2) Partial Fractions

The A’s, B’s and C’s are constants i.e. are independent of x.
§ 2.3 Methods for determining coefficients:
There are various methods for determining equations for
the purpose of calculating the coefficients in the identity, which
expresses the rational function as the sum of partial fractions.
In any particular case it may be convenient to use more than one
method.
i) The substitution of particular real values of variable. In many
cases the choice as determined by the denominators of the partial
fractions can provide equations, which contain only one
coefficient.
ii) Equating of coefficients of corresponding powers of the
variable.
F(a ) 1
iii) The use of the formula , for the coefficient , where
Q' (a ) x−a
F(a )
is a proper fraction and (x – a) is a non-repeated factor of
Q' (a )
Q(x).
To prove (iii) we may proceed as follows:
Let us consider Q(x) = (x – a) R(x), where R(a) ≠ 0.
Then Q′(x) = (x – a) R′(x) + R(x), giving Q′(a) = R(a).
F( x ) F( x ) A g(x )
Now = = +
Q( x ) ( x − a ) R ( x ) x−a R (x)
i.e. F(x) = A R(x) + (x – a)g(x), where A denotes a constant which
is different from zero. Then F(a) = A R(a) = A Q′(a) ≠ 0.

27
Algebra (Chapter 2) Partial Fractions
F(a ) F(a )
Hence = .
Q' (a ) R (a )
This result justifies the following rule for finding the numerator A
A
of a partial fraction , when (x–a) is not a repeated factor of
x−a
the denominator. Remove (x–a) from the denominator and
F(a )
substitute a, for x in the remaining fraction. i.e. A = .
R (a )
iv) The substitution of a complex number to provide an equation
in which only two coefficients remain. These being real can then
be found by equating real and imaginary parts.
P( x )
The determination of the partial fractions of is
Q( x )
particularly simple when there are no repeated factors in Q(x). It is
easier to deal with linear than with quadratic factors the use of
complex algebra, quadratic factors can be avoided but the
corresponding partial fractions have to be combined in pairs and
further reduction may be necessary.
Case 1 : Non-repeated linear factors:
Example 2.5:
x2 + 1
Express in partial fractions.
( x − 1)( x − 2)( x − 3)
Solution:
2
The degree of P(x) = x +1 is less than the degree of Q(x), where
Q(x) = (x–1)(x–2)(x–3). According to case (i) of theorem 2.8, we

28
Algebra (Chapter 2) Partial Fractions

have three non-repeated linear factors then we may write:

x2 + 1 A B C
= + + . (1)
( x − 1)( x − 2)( x − 3) x − 1 x − 2 x − 3
Clearing (1) of fractions and equating the numerator in both sides
2
gives x + 1 ≡ A(x–2)(x–3) + B(x–1)(x–3) + C (x–1)(x–2). (2)
To determine the coefficients in the identity (2), there are various
methods.
Method 1: (This method is valid for all cases)
Since (2) is an identity in x it is necessary and sufficient that the
coefficient of each power of x, be the same on the left side of the
equation as it is on the right side.
Equating these coefficients leads to the following equations:
2 2
x + 1 ≡ (A+ B + C) x –(5A+4B+3C)x + 6A + 3B + 2C
2
Coefficient of x : 1 = A + B + C,
Coefficient of x : 0 = –5A – 4B – 3C,
0
Coefficient of x : 1 = 6A + 3B + 2C.
Solving we get A = 1, B = –5 C = 5.
Method 2: (This method valid for this case only)
If we multiply both sides of identity (1) by (x–1) to get
x2 + 1 B( x − 1) C( x − 1)
=A+ + .
( x − 2)( x − 3) x−2 x −3
The goal is to make convenient substitutions whenever possible.
Now set x = 1, the resulting equation gives the value of A: A = 1.
Thus the value of A is the number we would obtained if we had

29
Algebra (Chapter 2) Partial Fractions

covered the factor (x – 1) in the denominator of the original

fraction and evaluated the rest at x = 1:
2
(1) + 1 2
A= = = 1.
(x–1) (1–2) (1–3) (–1)(–2)
Similarly we can find the value of B by covering the factor
(x–2) in the original fraction and evaluating the rest at x = 2:
2
(2) + 1 5
B= = = –5.
(2–1) (x–2) (2–3) (1)(–1)
Finally C can be formed by covering the factor (x–3) and
evaluating the rest at x = 3:
2
(3) + 1 10
C= = = 5.
(3–1) (3–2) (x–3) (2)(1)
P( x )
In general to expand a quotient of polynomial by partial
Q( x )
fractions when the degree of P(x) is less than the degree of Q(x),
and Q(x) is the product of n linear factors each to the first power
one may proceed in the following way. First write the quotient
with Q(x) factored:
P( x ) P( x )
= . (2.1)
Q( x ) ( x − a1 )( x − a 2 )...( x − a n )
Then cover the factors (x–ai) of Q(x) in (2.1) one at a time each
time replacing all the uncovered x’s by the number ai. This gives a
number Ai for each root ai:
The partial fractions expression of P(x)/Q(x) is then:
P( x ) A1 A2 An
= + +…+ . (2.2)
Q( x ) ( x − a1 ) ( x − a 2 ) (x − a n )

30
Algebra (Chapter 2) Partial Fractions

There are some problems in the other cases in which this method
is not fully succeeded.
Method 3: (Valid for all cases)
In many problems it is easiest to assign small values to x
like x = 0, ±1, ± 2, … to get equations in A, B, C, D and so on.
In identity (2),
2
x + 1 ≡ A(x–2)(x–3) + B(x–1)(x–3) + C (x–1)(x–2). (2)
Let x = 1, 2, 3 successively to find A, B and C:
2
x = 1 : 1 + 1 = A(1–2)(1– 3) + 0 + 0 i.e. A = 1,
2
x = 2 : 2 + 1 = 0 + B(2–1)(2– 1) + 0 i.e. B = –5,
2
x = 3 : 3 + 1 = 0 + 0 + C(3–1)(3– 2) i.e. C = 5,
Conclusions:
x2 + 1 1 5 5
= – + .
( x − 1)( x − 2)( x − 3) x − 1 x − 2 x − 3
Example 2.6:
5x 2 − 10x + 2
Express in partial fractions
3 2
x − 3x + 2 x
Solution:
2
The degree of P(x) = 5x – 10x + 2, is less than the degree of
3 2
Q(x) = x – 3x + 2x = x(x–1)(x–2). The roots of Q(x) are a1 = 0,
a2 = 1 , a3 = 2 we find

0+0+2 2
A1 = = = 1,
x (0–1) (0–2) (–1)(–2)

31
Algebra (Chapter 2) Partial Fractions

5 – 10 + 2 –3
A2 = = = 3,
1 (x–1) (1–2) (1)(–1)

5(4) – 10(2) + 2 2
A3 = = = 1.
2 (2–1) (x–2) (2)( 1)
5x 2 − 10x + 2 1 3 1
Therefore = + + .
x 3 − 3x 2 + 2 x x x −1 x − 2

Example 2.7:
2 x 4 − 5x 3 + 6 x 2 − 8x + 2
Express , in partial fractions.
x ( x − 1)( x − 2)
Solution:
The degree of the numerator is greater than the degree of the
denominator. Hence we divide first, obtaining:
2 x 4 − 5x 3 + 6 x 2 − 8x + 2 5x 2 − 10 x + 2
= 2x + 1 +
x ( x − 1)( x − 2) x ( x − 1)( x − 2)
By the result of example 2.6 we have the remainder term
2 x 4 − 5x 3 + 6 x 2 − 8x + 2 1 3 1
therefore: = 2x + 1 + + + .
x ( x − 1)( x − 2) x x −1 x − 2
Case 2: A Repeated linear factor in the denominator:
Example 2.8:
x+3
Express in partial fractions.
2
x ( x − 1)
Solution:
2
Since the denominator has a repeated linear factor (x–1) , we must

32
Algebra (Chapter 2) Partial Fractions

express the fraction in the form

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

B C
You need both terms and , the method will not work
x −1 ( x − 1) 2
otherwise. Clearing (1) of fractions gives:
2
x + 3 ≡ A (x–1) + B x (x–1) + C x. (2)
Method 1:
2
x + 3 ≡ (A + B) x – (2A + B – C) x + A.
Equating the coefficient of each power of x we get:
A = 3, 2A + B – C = – 1, A + B = 0.
Solving we get A = 3, B = –3, C = 4.
Method 2:
The value of A in (1) is the number we would have obtained if we
had covered the factor x in the denominator of the original
fraction and evaluated the rest at x = 0: A = 3.
2
If we multiply both sides of equation (1) by (x–1) , to get
x+3 A 2
= (x–1) + B(x–1)+ C, and set x = 1, the resulting equation
x x
gives the value of C: C = 4. i.e. the value of C in (1) is the number
2
we would have obtained if we had covered the factor (x–2) in the
denominator of the original fraction and evaluated the rest at x =1:
C = 4.
To determine the value of B, set x = any value other than 0, 1 in
equation (2).
33
Algebra (Chapter 2) Partial Fractions

Let x = 2 we get: 5 = A + 2B + 2C, i.e. 5 = 3 + 2B + 8, ∴ B = –3.

Also if we differentiate both sides of (2) with respect to x to get
1 = 2 A (x–1) + B x + B (x–1) + C. (3)
Substituting x = 1, shows that 1 = B + C. Hence B = –3.
Method 3:
We may substitute any three values for x.
Example 2.9
x +1
Express in partial fractions.
3
( x − 1)
Solution:
x +1 A B C
Let = + + .
( x − 1) 3 x − 1 ( x − 1) 2 ( x − 1)3

We first clear the fractions to get

2
x + 1 ≡ A(x–1) + B(x– 1) + C.
Substituting x = 1, shows that C = 2. We then differentiate both
sides with respect to x to get: 1 = 2A(x– 1) + B.
Substituting x = 1, shows B = 1. We differentiate again to get
x +1 1 2
2A = 0 which shows A = 0. Hence: = + .
( x − 1) 3 ( x − 1) 2 ( x − 1) 3
Case 3: A Quadratic factor in the denominator:
Example 2.10:
2 x 2 − 3x − 4
Express , in partial fractions
2 2
( x − 1)( x + 2 x + 2)

34
Algebra (Chapter 2) Partial Fractions

Solution:
2 x 2 − 3x − 4 A B Cx + D
Let + = + .
( x 2 − 1)( x 2 + 2 x + 2) x − 1 x + 1 x 2 + 2 x + 2
We first clear the fraction to get:
2 2 2 2
2x –3x–4 ≡A(x+1)(x +2x+2) + B(x–1) (x +2x+2)+(Cx+D)(x –1).
Our goal is to make convenient substitutions whenever possible.
Set x = 1, we get A = – 1 . Set x = –1, we get B = – 1 .
2 2
Set x = 0, we get –4 = 2A – 2B – D i.e. D = 4.
Set x = 2, we get 8 – 6 – 4 = –2 = 30A + 10B + 6C ,
which shows C = 1.
2 x 2 − 3x − 4 −1
2
−1
2 x+4
i.e. = + + .
( x 2 − 1)( x 2 + 2 x + 2) x − 1 x + 1 x 2 + 2 x + 2
Case 4: Repeated quadratic factors:
Example 2.11:
1
Express in partial fractions.
2 2
x ( x + 1)
Solution:
1 A Bx + C Dx + E
Let = + + .
x ( x 2 + 1) 2 x x 2 + 1 ( x 2 + 1) 2
Bx + C Dx + E
Here, we need both terms and .
2 2 2
x +1 ( x + 1)
By clearing fractions we get:
2 2 2
1 ≡ A(x + 1) + x (x + 1)(B x + C) + x (D x + E)

35
Algebra (Chapter 2) Partial Fractions
2 2 2 2 2 2
≡ A(x + 1) +B x (x + 1) + C x (x + 1) + D x + Ex. (1)
We need five values of x because there are five unknowns:
x = 0 : 1 = A,
x = 1 : 1 = 4A + 2B + 2C + D + E,
x = –1 : 1 = 4A + 2B – 2C + D – E,
x = 2 : 1 = 25A + 20B + 10C + 4D + 2E,
x = –2 : 1 = 25A + 20B – 10C + 4D – 2E.
This is a system of five linear algebraic equations with five
unknowns. Solving such system, the solution is A = 1, B = D = –1,
and C = E = 0.
1 1 x x
= – – .
x ( x 2 + 1) 2 x x 2 + 1 ( x 2 + 1) 2
Method 1:
By expanding equation (1) and collecting like terms we have
4 3 2
1 ≡ (A + B) x + C x + (2A + B + D) x + (C + E) x + A.
Now we can equate the coefficients of like terms on opposite sides
of the equation. Thus we have:
A = 1, C + E = 0 , 2A + B + D = 0 , C = 0 , A + B = 0 .
The solution is: A = 1, B = D = –1, and C = E = 0.

Example 2.12:
x 2 − 8x + 9
Express , in partial fractions
( x + 1)( x − 2) 3

36
Algebra (Chapter 2) Partial Fractions

Solution:
First method:
x 2 − 8x + 9 A B C D
Assume = + + + ,
( x + 1)( x − 2) 3 ( x + 1) ( x − 2)3 ( x − 2) 2 ( x − 2)
2 3 3
then x – 8x + 9 ≡ A (x–2) + B(x+1)+ C(x+1)(x–2)+D(x+1)(x–2) .
Set x = –1, then A = –2/3 . Set x = 2 then B = –1.
3
Equating coefficients of x , to zero, A + D = 0, ∴D = 2/3.
Set x = 0, 0 = –8A + B – 2C + 4D, ∴ C = –1.
Therefore:
x 2 − 8x + 9 −2 1 1 2
= – – + . (1)
( x + 1)( x − 2) 3 3( x + 1) ( x − 2)3 ( x − 2) 2 3( x − 2)
Second method:
If, as here, the denominator contains a factor repeated three or
more times it may be shorter to proceed as follows:
Put x – 2 = y, then the original fraction becomes:
x 2 − 8x + 9 ( y + 2) 2 − 8( y + 2) + 9 y 2 − 4 y − 3
= = .
( x + 1)( x − 2) 3 y 3 ( y + 3) y 3 ( y + 3)
2 2 2
But y – 4 y – 3 = –3 – 4y + y = –(3+y) –3y + y =
2 2 3
= –(3+y) – y(3+y) + 2y = –(3+y) – y(3+y) + 2 y (3+y)– 2 y .
3 3

y 2 − 4y − 3 1 1 2 2
Therefore: =– – + + ,
y 3 ( y + 3) y3 y2 3y 3( y + 3)

which gives the same result as before in (1).

Note: The process of expressing functions in partial fractions

37
Algebra (Chapter 2) Partial Fractions

although elementary is often laborious. In such cases the

alternative methods given in example 2.12, are preferable.
Third method:
After finding the value of A and B:
x 2 − 8x + 9 −2 1 C D
Put = – – + .
( x + 1)( x − 2) 3 3( x + 1) ( x − 2)3 ( x − 2) 2 ( x − 2)

Since
2 3 2 3
3(x – 8x + 9) + 2(x–2) + 3(x+1) = 3(x – 7x + 10) + 2(x–2) =
3 2
= 3(x–2) (x–5)+ 2(x–2) = (x–2)[3(x–5) + 2(x–2) ] =
2
= (x – 2)[2x – 5x – 7] = (x – 2)(x + 1)(2x – 7).
Therefore:
( x − 2)( x + 1)(2 x − 7) C D
= +
3( x + 1)( x − 2) 3 ( x − 2) 2 ( x − 2)
2x − 7 C D
or = +
3( x − 2) 2 ( x − 2) 2 ( x − 2)
2( x − 2) − 3 2 1 C D
or = – = +
3( x − 2) 2 3( x − 2) ( x − 2) 2 ( x − 2) 2 ( x − 2)

i.e. C = – 1, D = 2/3.
x 2 − 8x + 9 −2 1 1 2
∴ = – – + .
( x + 1)( x − 2) 3 3( x + 1) ( x − 2)3 ( x − 2) 2 3( x − 2)

38
Algebra (Chapter 2) Partial Fractions

EXERCISES 2
x2 2 2
1) In the fraction , we would first divide x by x –1 to get
x2 −1
x2 1 2
= 1+ , suppose that we ignore the fact that x and
x2 −1 x2 −1
2 x2 A B
x –1, have the same degree and try to write =
+ .
x2 −1 x −1 x + 1
What goes wrong? Find out by trying to solve for A and B.
Split into partial fractions:
2x + 5 x+5 1
2) , 3) , 4) ,
( x + 2)( x + 3) 2
x + 7 x + 12 x ( x + 1)(2 x + 1)
3 2
2x +5x −3x −1 4(x −1) 2x − 5
5) , 6) , 7) ,
2
2x +5x + 2 (x + 2)( x − 2) 2
( x + 3)( x + 1) 2
x −1 4 1
8) , 9) , 10) ,
( x + 1)( x 2 + 1) 1 − x4 2
(1+x )(1−x+x ) 2

2 x3 + x + 1 6 x 3 + 5x 2 − 7
11) , 12) , 13) ,
x (2 + x 2 ) 2 (1 − x ) 4 3x 2 − 2 x − 1
x 1 16) 5x2 + 6x − 23 .
14) , 15) , 3 2
2
( x − 1) 2 2
( x − 1) 2 (x −1) (x +1) (x − 2)

2x + 5 1 1
[Answers: 2) = + ,
( x + 2)( x + 3) x + 2 x + 3
x+5 2 1
3) = – ,
x 2 + 7 x + 12 x + 3 x + 4
1 1 1 4
4) = + – ,
x ( x + 1)(2 x + 1) x x + 1 2 x + 1
2x3 +5x2 −3x −1 1 3
5) =x+ – ,
2
2x +5x + 2 2 x + 1 x + 2
39
Algebra (Chapter 2) Partial Fractions
4(x −1) 1 3 3
6) = + – ,
(x + 2)(x − 2)2 ( x − 2) 2 4( x − 2) 4( x + 2)
2x − 5 7 11 11
7) =– + – ,
( x + 3)( x + 1) 2 2( x + 1) 2 4( x + 1) 4( x + 3)
x −1 x 1
8) = – ,
( x + 1)( x 2 + 1) x 2 + 1 x + 1
4 1 1 2
9) = – + ,
1− x 4 x + 1 x − 1 2
x +1
1 x x −1
10) = – ,
(1+ x 2 )(1− x + x 2 ) x 2 + 1 x 2 − x + 1
2 1 x x
11) = – – ,
2 2 2 2 2
x (2 + x ) 2 x ( x + 2) 2( x + 2)
x3 + x + 1 3 4 3 1
12) = + + + ,
(1 − x ) 4 ( x − 1) 4 ( x − 1) 3 ( x − 1) 2 x −1
6 x 3 + 5x 2 − 7 5 1
13) = 2x + 3 + + ,
3x 2 − 2 x − 1 3x + 1 x − 1
x 1 1
14) = – ,
( x 2 − 1) 2 4( x − 1) 2 4( x + 1) 2
1 1 1 1 1
15) = + + – ,
( x 2 − 1) 2 4( x + 1) 2 4( x + 1) 4( x − 1) 2 4( x − 1)
5x 2 + 6x − 23 3 4 1 1
16) = + + – –
(x − 1)3 (x + 1)2 (x − 2) ( x − 1)3 ( x − 1) 2 x − 1 ( x + 1) 2
2 1
– + .]
x +1 x − 2

40
Algebra (Chapter 3) Finite Series

CHAPTER 3
FINITE SERIES
A set of numbers arranged according to a definite law of
formation is called a sequence.
A general representation of a sequence can be made as u1,
u2...un... Where the suffixes follow a definite law. The sequence
may be finite or infinite. If the definite law of formation of
sequence is known, we can find the general term of the set, with
the help of which any terms can be written.
When the terms of a sequence are connected together with
plus or minus signs, we get a series. A series of terms is
th
completely defined if the r term is given as a function of r, for all
positive integral values of r. This term may be represented by ur or
vr.
Also if Sn = u1 + u2 + ...+ un, Sn is called the sum to n terms of the
n n
series, and we write Sn = ∑ u r , or Sn = ∑ u r .
r =1 1
The Greek letter Σ (sigma) is shorthand method of representing
the sum of a number of terms of the same type; the value of r at
the foot indicates the first term and the number at the top indicates
the last term.
Properties of Σ notation:
n n n
i) ∑ (u r + v r ) = ∑ u r + ∑ v r ,
r =1 r =1 r =1

41
Algebra (Chapter 3) Finite Series
n n n
ii) ∑ a u r = a ∑ u r , iii) ∑ 1 = n .
r =1 r =1 r =1
Thus u1 + u2 + ...+ un , is a series which may be expressed in the
n ∞
form ∑ u r . When we write ∑ u r , it means n →∞.
r =1 r =1
§ 3.1 Summation of series:
There is no general method for summing series, nor is it
even always possible to express the sum to n terms as an
elementary function of n. No such expression, for example, exists
for the sum of the harmonic progression, 1 + 1 + 13 +…+ n1 .
2

The methods for the summation of a series vary according to the

series and are classified as follows:
A. Inspection Method (order of testing)
1) Test for an arithmetic progression (a.p.):
n
ur = a + (r–1)d , ∑ u r = n [ a+ 1 d(n–1)]. (3.1)
2
r =1
2) Test for a geometrical progression (g.p.):
(r–1) n a (1 − h n )
ur = a h , ∑ur = . (3.2)
r =1 1− h
∞ a
If |h| < 1 , then ∑ u r = . (3.3)
r =1 1− h
3) Test for an arithmetic-geometrical progression:
The general arithmetic-geometrical progression of the form
2 3 n
a + (a+d)h + (a+2d)h + (a+3d)h +…+ [a + (n –1)d ] h ,
where the coefficients of the powers of h are in a.p., and the h

42
Algebra (Chapter 3) Finite Series

portions of the terms are in g.p. The method of summation of this

series is the same as for g.p. namely:
n
Let S = ∑ [a + (r − 1)d ]h r −1 . (3.4)
r =1
n n
Then S h = h ∑ [a + (r − 1)d ]h r −1 = ∑ [a + (r − 1)d ]h r . (3.5)
r =1 r =1
Subtracting (3.5) from (3.4), we get
2 n–1
S = a + (a+d)h + (a+ 2d)h + ... +[a+ (n–1)d]h ,
2 3 n
S h = a h + (a+d)h + (a+ 2d)h + ... +[a+ (n–1)d]h .
2 n–1
S(1– h) =a + (a+d)h + (a+ 2d)h + ... +[a+ (n–1)d]h –
2 3 n–1 n
–{ah + (a+d)h + (a+ 2d)h +... +[a+ (n–2)d]h +[a+ (n–1)d]h } =
2 n–1 n
= a + dh[1 + h + h + …+ h ] –[a+ (n–1)d]h =
1 − h n −1 n
= dh + a –[a+ (n–1)d]h
1− h
1 − h n −1 a − [a + (n − 1)d]h n
Therefore S = dh + (3.6)
(1 − h ) 2 1− h

4) Test for a binomial expansion:

n n n n–1 n n–2 2 n n–r r n
(a + x) = a + C1 a x + C2 a x + …+ Cr a x +…+ x =
n 0 n n–1 n n–2 2 n n–r r 0 n
= a x + C1 a x+ C2 a x +…+ Cr a x +…+ a x =
n
n n −r r
= ∑ Cr a x . (3.7)
r =0
n (n − 1)(n − 2)...(n − r + 1) n!
Where n C r = = . (3.8)
r! r! (n − r )!
n n n–1 n (n − 1) n–2 2 n
(a + x) =a + n a x+ a x +... + x
2!

43
Algebra (Chapter 3) Finite Series
n n (n − 1) 2 n (n − 1)(n − 2) 3
(1 ± x) = 1 ± n x + x ± x +… . (3.9)
2! 3!
5) Test for an exponential series (infinite series):
x2 x3 x4 xk x
1+x+ + + +…+ +…=e , (3.10)
2! 3! 4! k!
x2 x3 x4 k x
k
–x
1–x+ – + +…+(–1) +…=e , (3.11)
2! 3! 4! k!
x3 x5 x7 x 2k −1
x+ + + +…+ +…= cosh x, (3.12)
3! 5! 7! (2k − 1)!
x2 x4 x6 x 2k −2
1+ + + +…+ +…= sinh x, (3.13)
2! 4! 6! (2k − 2)!
2k −2
x2 x4 x6 k +1 x
1– + – +…+ (−1) +…= cos x, (3.14)
2! 4! 6! (2k − 2)!
2k −1
x3 x5 x7 k +1 x
x– + – +…+ (−1) +…= sin x, (3.15)
3! 5! 7! (2k − 1)!
There are valid for all values of x.
6) Test for a logarithmic series (infinite series):
If –1< x ≤ 1, then
x2 x3 x4 k +1 x
k
x– + – +…+ (−1) + … = An (l + x), (3.16)
2 3 4 k
If –1 ≤ x < 1, then
x2 x3 x4 xk
–x– – – –…– – … = An (l – x). (3.17)
2 3 4 k
B. Summation of types not obvious by inspection:
(Summation by means of undetermined coefficients)

44
Algebra (Chapter 3) Finite Series

Example 3.1:
Find the sum of the squares of the first n natural numbers.
Solution:
n 2 2 2 2
Let Sn = ∑ r 2 = 1 + 2 +3 ... + n
r =1
2 3
≡ A0+A1n +A2n +A3n +… (3.18)
Replacing n by n+1,
2 2 2 2 2
Sn+1 = 1 + 2 +3 ... + n + (n+1) ≡
2 3
≡ A0+A1(n+1) +A2(n+1) +A3(n+1) +… (3.19)
Subtracting (3.19) from (3.18) we get
2 2
(n+1) ≡ A1 +A2(2n+1) +A3(3n + 3n+ 1)+… (3.20)
Since (3.20) is an identity, and the highest power of n on the
L.H.S. is the square, it follows that A4 , A5 , etc., must be zero.
Equating coefficients on the two sides of the identity (3.20),
2
n : 1 = 3 A3 . ∴ A3 = 13 ,
1
n : 2 = 2A2 + 3A3 = 2A2 + 1 , ∴ A2= 2
,
0 1 1 5
Unity u : 1 = A1 +A2 +A3 = A1 + 2
+ 3
= A1+ 6
,
∴ A1= 1 .
6

Therefore
2 2 2 2 1 2 3
1 + 2 +3 ... + n = A0 + 1 n + 2
n + 13 n .
6
1
Using n =1 in this identity, 1 = A0 + 1 + 2
+ 13 , ∴ A0 = 0.
6

Therefore
2 2 2 2 1 2 3
1 + 2 +3 ... + n = 1 n + 2
n + 13 n = 1 n(n+1)(2n+1).
6 6

45
Algebra (Chapter 3) Finite Series

Exercise:
By the same method prove that
n
1
i) ∑ r = 2
n(n+1) ,
r =1
n 2
ii) ∑ r 3 = [ 12 n(n+1)] ,
r =1
n
iii) ∑ r (r + 1) = 13 n(n+1)(n+2).
r =1
§ 3.2 The difference method:
This method is based on the following simple idea. If we able to
find a function f(r) such that
ur = f(r+1) – f(r) (r = 1,2,3,...,n) (3.21)
Then, on giving r the values 1,2...n in succession,
u1 = f(2) – f(1),
u2 = f(3) – f(2),
u3 = f(4) – f(3),
………………
un = f(n+1) – f(n).
When we adding we get
n
∑ u r = u1 + u2 + u3 +…+ un = f(n+1) – f(1), (3.22)
r =1
If it is possible to discover a function f(r) satisfies (3.21), it is easy
to sum the series to n terms and the relation gives this sum (3.22).
Also, if we able to find a function g(r) such that,
ur = g(r+1) – g(r–1) (3.23)

46
Algebra (Chapter 3) Finite Series

Hence
u1 = g(2) – g(0),
u2 = g(3) – g(1),
u3 = g(4) – g(2),
u4 = g(5) – g(3),
…………………
un–1 = g(n) – g(n–2).
un = g(n+1) – g(n–1).
When adding we get
n
∑ u r = u1 + u2 + u3 +…+ un = g(n) + g(n+1) – g(0) – g(1). (3.24)
r =1

Example 3.2:
n
To find the sum of the first n natural numbers Sn = ∑ r .
r =1

Solution:
2 2 2
i) Let ur = r , it is clear that , (r+1) – r = 2r + 1 , then let f(r) = r .
n n n
Therefore ∑ (2r + 1) = 2 ∑ r + ∑ 1= 2 Sn + n, (3.25)
r =1 r =1 r =1
n n
Also ∑ (2r + 1) = ∑ [f(r + 1) − f(r)] = f(n+1) – f(1) =
r =1 r =1
2 2 2
= (n+1) – 1 = n + 2 n = n(n+2), (3.26)
From (3.25) and (3.26), we get
2Sn + n = n(n+2) , or 2Sn = n(n+1),
Therefore Sn = 12 n (n+1).

47
Algebra (Chapter 3) Finite Series
2 2
ii) Using the well known fact that, (r+1) – (r–1) = 4r,
2 2
from which it follows that r = 1 (r+1) – 1 (r–1) .
4 4
Let ur = r, g(r) = 1 r2 .
4
n n
Therefore Sn = ∑ r = ∑ [ g(r + 1) − g(r − 1)]
r =1 r =1
= g(n) + g(n+1) – g(0) – g(1).
n 2 2 1
i.e. Sn = ∑ r = 1 n + 1 (n+1) – 1 = 2
n(n+1).
4 4 4
r =1
2 2
iii) It is easy to see also that, (2r+1) – (2r–1) ≡ 8 r.
Similarly to (i) and (ii) , you can complete the solution.
Example 3.3:
Find the sum to n terms of the series 1.2 + 2.3 + 3.4 +…+ n(n+1).
Solution:
th
The r term is ur = r(r+1).
But r(r+1)(r+2)–(r–1) r(r+1) = r(r+1) [r + 2 – r + 1] = 3r(r+1)=3 ur,
Then if we put f(r) = 13 (r–1) r(r+1), we get u = f(r+1) – f(r).
n
Therefore ∑ u r = f(n+1) – f(1) = 13 n(n+1) (n+2).
r =1

Example 3.4:
n
Find the sum Sn = ∑ r 2 .
r =1

Solution:
i) The series can be summed by means of the identity

48
Algebra (Chapter 3) Finite Series
3 3 2
(r + 12 ) – (r– 12 ) ≡3 r + 1.
4

But to avoid fractions we write

3 3 2
(2r+1) – (2r–1) ≡ 24r + 2.
Therefore ur = f(r+1) – f(r) – 1 ,
12
1 3
Where f(r) = (2r – 1) .
24

Thus as before putting r=1,2...n and adding we have

n n n
1
∑ u r = ∑ [f(r + 1) − f(r) ] – ∑ 12
= f(n+1) – f(1) – n =
12
r =1 r =1 r =1
3
=1 (2n + 1) – 1 – n = 1 n(n+1)(2n+1). (3.27)
24 24 12 6

ii) Another proof, using a similar adaptation of the difference

method.
2 3 3 2
Let ur = r . Now (r+1) – r = 3r + 3r + 1.
3
Put f(r) = r .
n n
Hence ∑ [3r 2 + 3r + 1 ] = ∑ [f(r + 1) − f(r) ] =
r =1 r =1
3
= f(n+1) – f(1) = (n+1) – 1. (3.28)
n n
But L.H.S.= ∑ [3r 2 + 3r + 1 ] = 3Sn + 3 ∑ r + n =
r =1 r =1
3
= 3Sn + 2
n(n+1) + n. (3.29)

From (3.28) and (3.29) we get

3
3Sn + 32 n(n+1) + n = (n+1) – 1.
3
Therefore: Sn = 13 (n+1) – 13 – 12 n(n+1) – 13 n = 1 n(n+1)(2n+1).
6

iii) The results obtained above in Example 3.3 may be used for

49
Algebra (Chapter 3) Finite Series
n
summing the series ∑ r 2 .
r =1
2
Since r = r (r+1) – r
Therefore
n n n
2 1
∑ r = ∑ r (r + 1) – ∑ r = 13 n(n+1)(n+2) – 2
n(n+1)=
r =1 r =1 r =1
= 1 n(n+1)(2n+1).
6

Example 3. 5:
1 1 1 1
Find the sum to n terms of the series + + +…+ .
1.2 2.3 3.4 n (n + 1)
Solution:
th 1 A B 1 1
The r term ur = = + = – .
r (r + 1) r r + 1 r r + 1
1
Therefore ur = f(r) – f(r+1), where f(r) = .
r
n 1 n
Hence ∑ u r = f(1) – f(n+1) = 1 – = . (3.30)
r =1 n + 1 n + 1
Examples 3.3 and 3.5, illustrate two general types of series to
which many others can be reduced. In Example 3.3, all terms
contain the same number of factors, which are successive terms of
an a.p., and the initial factors are successive terms of the same a.p.
In Example 3.5, the terms are the reciprocals of terms of the type
illustrated in Example 3.3. The methods given above for
discovering the required difference can be applied similarly to all
series of these types. The general terms of the series summed in

50
Algebra (Chapter 3) Finite Series

Examples 3.3 and 3.5, can each be reduced to the one of the
following forms :
i) ur = [a + (r–1)d] [a + rd] ... [a + (r + k – 2)d] ,
1 1
ii) vr = = .
[a + (r − 1)d] [a + rd] ... [a + (r + k − 2)d] u r
Where each term contains k factors, which are successive terms of
an a.p. and the initial factor are successive terms of the same a.p.
for type (i) see Example 3.3, the required difference is obtained by
writing down ur and inserting the next factor at the end and put
1
f(r)= [a+(r–1)d] [a + rd]...[a+(r + k–2)d] [a + (r + k – 1)d],
(k + 1)d
It is easy to see that ur = f(r) – f(r–1).
n
Therefore ∑ u r = f(n) – f(0). (3.31)
r =1
For type (ii) see Example 3.5, the required function g(r) is
obtained by writing down vr and removing the first factor and put
1 1
g(r)= , k > 1.
(k + 1)d [a + rd][a + (r + 1)d]...[a + (r + k − 2)d]
It is easy to see that vr = g(r–1) – g(r)
n
Therefore ∑ v r = g(0) – g(n). (3.32)
r =1

Example 3.6:
Find the sum of the series 2.5 + 5.8 + 8.11+…+(3n –1)(3n + 2).
Solution:
th
The r term ur = (3r–1)(3r + 2).

51
Algebra (Chapter 3) Finite Series
Put f(r) = (3)(1 3) (3r–1) (3r+2) (3r+5), ∴ f(r) – f(r–1) = ur .
n
1 10
Therefore ∑ u r = f(n) – f(0) = 9
(3n–1)(3n+2)(3n+5) + 9
.
r =1

Example 3.7:
Find the sum to n terms of the series
1 1 1 1
+ + +…+ .
2.5 5.8 8.11 (3n − 1)(3n + 2)
Solution:
th 1
The r term vr = .
(3r − 1)(3r + 2)
1 1
Put g(r) = 3
, therefore g(r–1) – g(r) = vr ,
(3r + 2)
n 1
and ∑ v r = g(0) – g(n) = 1 – .
r =1
6 3(3n + 2 )

Example 3.8:
n 1
Find ∑ .
r =1 (r + 1)(r + 3)

Solution:
1 1 1
ur = = –
(r + 1)(r + 3) 2(r + 1) 2(r + 3)
1
Put f(r+1) = , then ur = f(r–1) – f(r+1).
2(r + 3)
n
It is clear that ∑ u r = f(0) + f(1)– f(n) – f(n+1).
r =1
1 1 1 5 (2n + 5)
= +1 – – = – .
4 6 2(n + 2) 2(n + 3) 12 2(n + 2)(n + 3)

52
Algebra (Chapter 3) Finite Series

This result can be obtained also as follows:

n 1 n 1 1 n 1 1 n +1 1 1 n + 3 1
∑ur = ∑ – ∑ = ∑ – ∑ =
r =1 2 r =1 r + 1 2 r =1 r + 3 2 r = 2 r 2 r = 4 r
1 1 1 1 1 (2n + 5)
= [ + – – ]= 5 – .
2 2 3 n+2 n+3 12 2(n + 2)(n + 3)

53
Algebra (Chapter 3) Finite Series

EXERCISES 3
The student should take a paper and a pencil and fill in the details
for himself.
2 2
1) Verify that (2r+1) – (2r–1) = 8r, and deduce the sum of the
n
1
series ∑ r . [Answer: 2
n (n+1)]
r =1
2 2
2) Verify that (r+1)r – r(r–1) = r(3r–1),
2 2
(r+1) r – r (r–1) = r(3r+1), and deduce that
n 2 n 2
∑ r (3r − 1) = n (n+1), ∑ r (3r + 1) = n (n+1) .
r =1 r =1
3) Verify that
(2r–1)(2r+1)(2r+3) – (2r–3)(2r–1)(2r+1) = 6(2r–1)(2r+1),
and deduce that
1.3 + 3.5 + ... + (2n–1)(2n+1) = 1 [(2n-1)(2n+1)(2n+3) + 3].
6

4) Prove by working similar to that of problem 3, that

n
∑ (3r − 1)(3r + 2) = 1
9
[(3n–1)(3n+2)(3n+5) + 10] .
r =1
1 1 1 1
5) Verify that = [ – ],
(2r − 1)(2r + 1) 2 2r − 1 2r + 1
1 1 1 1 n
and deduce that + + +…+ = .
1.3 3.5 5.7 (2n − 1)(2n + 1) (2n + 1)
n 1
Hence or otherwise find the sum of the series ∑ .
2 2
r =1 (2r) − 1
1 1 1 n
6) Prove that + +…+ = .
1.4 4.7 (3n − 2)(3n + 1) (3n + 1)
Hence find the sum to infinity

54
Algebra (Chapter 3) Finite Series

7) Sum the following series to n terms

3 3 3 2
a) 1 + 2 + ... + n . [Answer: { 12 n (n+1)} ]
3 3 3 3 n–1 3
b) 1 – 2 + 3 – 4 + ... + (–1) n .
n–1 3 3 2
[Answer: (–1) ( 12 n + 4
n – 18 )– 18 ]
2 2 2 2 n–1 2 1 n–1
c) 1 – 2 + 3 – 4 + ... + (–1) n . [Answer: 2
(–1) n (n+1) ]

8) Sum the following series to n terms:

a) 1.2 + 3.4 + 5.6 + ... [Answer: 13 n(n+1)(4n–1)]
1
b) 1.2.3 + 2.3.4 +... + n(n+1)(n+2). [Answer: 4
n(n+1)(n+2)(n+3)]
c) 1.4.7 + 2.5.8 + ... + n(n+3)(n+6). [Answer: 14 n(n+1)(n+6)(n+7)]
2
d) 1.2.5 + 3.4.7 + 5.6.9 + ... [Answer: 13 n(n+1)(6n + 14n–5)]

9) Sum the following series to n terms then find the sum up to

infinity:
1 1 1
a) + + +…,
1.4.7 4.7.10 7.10.13
1 1 1
[Answer: 1 ( – ), 24 ]
6 1.4 (3n + 1)(3n + 4)

1 1 1 1
b) + + +…, [Answer: 12 1
– 1
), 12 ]
1.3.5 3.5.7 5.7.9 4(2n + 1)(2n + 3)
4 5 6 2n + 5
c) + + +…, [Answer: 54 – ), 54 ]
1.2.3 2.3.5 3.4.5 2(n + 1)(n + 2)
n 4 2
10) Prove that ∑ =1– .
r =1 r(r + 1)(r + 2) (n + 1)(n + 2)
Deduce the sum to infinity of the series
1 1 1
+ + +…, [Answer: 1
4
]
1.2.3 2.3.4 3.4.5
11) Sum the following series to n terms:

55
Algebra (Chapter 3) Finite Series
n 2
a) ∑ r(r + 1)(2r + 1) , [Answer: 1
2
n(n+1) (n+2)]
r =1
n 2r + 1 1
b) ∑ 2 2
, [Answer: 1 – ]
r =1 r (r + 1) (n + 1) 2
n 1
c) ∑
r =1 (2r − 1)(2r + 1)(2r + 3)(2r + 5)
1 1
[Answer: 1 ( – )]
6 1.3.5 ( 2n + 1)(2n + 3)(2n + 5)

Sum the following series to n terms :

1 1 1
12) + +…+ .
(1 + x)(1 + 2x) (1 + 2x)(1 + 3x) [1 + nx][1 + (n + 1)x]
n
[Answer: ]
[1 + x][1 + (n + 1)x]
1 1 1
13) + +…+ .
(1 + x)(1 + ax) (1 + ax)(1 + a 2 x) (1 + a n −1x)(1 + a n x)
1 1 an
[Answer: { – }]
(1 − a) 1 + x (1 + a n x)
1 x x2
14) + + +…+
(1 + x)(1 + x 2 ) (1 + x 2 )(1 + x 3 ) (1 + x 3 )(1 + x 4 )
x n −1 1 1 1
. [Answer: { – }]
(1 + x n )(1 + x n +1 ) x(1 − x) 1 + x n +1 1 + x
[Hint: If we introduce the factor x(1–x) in the general term
x r −1 x(1 − x)
. , in both numerator and denominator
(1 + x r )(1 + x r + 1 ) x(1 − x)
then adding one and subtracting one in the numerator then resolve
the given fraction into partial fraction.
i.e. Resolve x r − 1 (1 − x) + 1 − 1 ].
x(1 + x r )(1 + x r + 1 )(1 − x)

56
Algebra (Chapter 4) Binomial Expansion

CHAPTER 4
BINOMIAL EXPANSION
The student is already familiar with the binomial
expansion for n a positive number. The following expansion:
n n–1 n (n − 1) n–2 2 n (n − 1)(n − 2) n–3 3
a + n a x+ a x + a x +... , (4.1)
2! 3!
terminates automatically after (n+1) terms if n is a positive integer
since each later coefficient contains the factor (n–n) = 0; and the
n
sum of this finite series is (a + x) . The general term is taken to be
th
the (r+1) term namely ur+1,
()
n n–r r
ur+1= a x =
r
n (n − 1)(n − 2)...(n − r + 1) n–r r
r!
a x . (4.2)

The factors
n
()
are called the binomial coefficients read n over r. It
r

is to be noted that the degree in a, and x combined in n and there

are r factors in both numerators and denominator of the coefficient
()
n
r
. The binomial coefficient can only vanish if n – r + 1 = 0,

i.e. if r = n + 1. But r is always a positive integer so that this is

only possible when n is a positive integer. Hence there will be an
infinite number of terms if n is fractional or negative.
§ 4.1 Binomial expansion for negative
And fractional indices:
The truth of the binomial expansion for negative and fractional
indices will be assumed i.e. for all rational values of n it will be
taken that:

57
Algebra (Chapter 4) Binomial Expansion
n n (n − 1) 2 n (n − 1)(n − 2) 3
(1 + x) = 1 + n x + x + x +… +
2! 3!
n (n − 1)(n − 2)...(n − r + 1) r
+ x +…. (4.3)
r!
The expansion (4.3) can be called an infinite series. In
process of finding the sum of an infinite series practical
difficulties may be faced, as the addition becomes an endless task.
But we may use the word “divergent” to denote a series whose
sum is infinitely great.
But when the sum tends to a finite limit as the number of terms
increases greatly we say the series is “convergent”. However an
example will make the meaning clear.
Let us discuss the expansion (4.3):
When n = –1 , x = –a :
–1 1 2 3
i.e. (1 – a) = = 1 + a + a + a +… (4.4)
1− a
If a = 2 the R.H.S. takes the form:
2 3
S = 1 + 2 + 2 + 2 +…, and it is easy to see that
S > 1 + 2 + 2 + ... → ∞. i.e. divergent series.
–1
But the L.H.S. = ( –1 ) = –1. i.e. L.H.S. ≠ R.H.S. when a = 2.
i.e. the expansion (4.3) is not valid for all values of x. The series
on the R.H.S. of (4.4) is a g.p., whose common ratio a, if |a|<1,
1
then its sum is finite and equals to .
1− a
Expansion (4.3) is called the binomial theorem (expansion) and

58
Algebra (Chapter 4) Binomial Expansion

was established by Sir Isaac Newton in 1676 for the cases in

which n is a positive integer and extended by Euler to negative
and fractional indices about 100 years later.
If n is a positive integer, the binomial coefficients ( ) can
n
r
n n! n
be written in the form Cr or ( Cr gives the number of
r! (n − r )!
combinations of n things taken r at a time). If n is not a positive
n
integer there is no meaning to the symbol n! and Cr and we
can not use one of them and we use only.
n
According to the usual convention the function (1+x) ,
where (1+x) > 0, has two values if n = p/q, where p and q are
positive or negative integers, and q is even. Thus
− 12 1
9 –½ ⎛ 25 ⎞ ⎛ 16 ⎞ 2 4
(1 + ) =⎜ ⎟ =⎜ ⎟ =± .
16 ⎝ 16 ⎠ ⎝ 25 ⎠ 5
But the sum to infinity of the binomial series for any given value
of x for which it is convergent has only one value. We shall use
n n
the symbol (1 + x) to denote the positive value of (1 + x) .
Theorem 4.1:
If n is a positive fraction or any negative number, and if |x| <1, the
series (4.3) is convergent and equals to the positive value of
n
(1 + x) . Moreover the series (4.3) is also convergent for x = 1, if
n > –1, and for x = – 1 if n > 0.
n
If we wish to expand a binomial of the form (a + b) ,

59
Algebra (Chapter 4) Binomial Expansion

where n is not a positive integer we must first write it in the form

n n
(1 + x) , where |x| < 1. Thus if |b| < |a| and if a exists then:
r
n n b n n ∞ ⎛ n ⎞⎡ b ⎤ b
(a + b) = a (1+ ) = a ∑ ⎜⎜ ⎟⎟ ⎢ ⎥ , provided < 1.
⎢⎣ a ⎥⎦
a r=0 ⎝ r ⎠ a

Thus we expand it in ascending powers of b.

n
Again if |a| < |b| and if b exists, then
r
n a n
n n ∞ ⎛ n ⎞⎡ a ⎤ a
(a + b) = b (1+ ) = b ∑ ⎜⎜ ⎟⎟ ⎢ ⎥ , provided < 1.
⎢⎣ b ⎥⎦
b r=0 ⎝ r ⎠ b

Thus we expand it in ascending powers of a.

Theorem 4.2:
–n
If n , x are positive, then all terms in the expansion of (1 – x)
are positive.
proof:
–n
ur+1 in the expansion of (1 – x) is
(−n)(−n − 1)(−n − 2)...(− n − r + 1)
ur+1 = (− x ) r =
r!
2r (n)(n + 1)(n + 2)...(n + r − 1)
= (–1) xr =
r!
n(n + 1)(n + 2)...(n + r − 1) r
= x =
r!
= Positive quantity for all values of r.
–n
Taking the expansion of (1 – x) , we have:
–n n (n + 1) 2 n (n + 1)(n + 2) 3
(1 –x) = 1 + n x + x + x +… +
2! 3!

60
Algebra (Chapter 4) Binomial Expansion
n(n + 1)(n + 2)...(n + r − 1) r
+ x +…. (4.5)
r!
Here: 1) The indices of the powers of x increases regularly by
unity the index of x in any term being the number of factors in the
numerator or denominator of its coefficient;
2) The numerator of the coefficient is the product of an
arithmetical progression of which the first is n, and the common
difference is 1,and
3) The denominator of the coefficient is also the product if an
arithmetical progression of which the first term is 1, and the
common difference is 1.
n
In the expansion of (1+x) , we have:
n n (n − 1) 2 n (n − 1)(n − 2) 3
(1+x) = 1 + n x + x + x +…+
2! 3!
n (n − 1)(n − 2)...(n − r + 1) r
+ x +…. (4.6)
r!
–n
Here: 1) The same remark (1) in the expansion of (1– x) ,
2) The numerator of the coefficient of ur+1, is the product of an
arithmetical progression of which the first is n and the common
difference is 1, and
3) The denominator of the coefficient is also the product of an
arithmetical progression of which the first term is r, and the
common difference is –1.
Example 4.1:
–3
Expand (1 – x) given |x| < 1. Find also which is the greatest term
61
Algebra (Chapter 4) Binomial Expansion
3 4
in the expansion, if (i) x = , (ii) x = 7
.
4

Solution:
–3 (−3)(−4)
(1 – x) = 1 + (– 3)(– x) + (− x ) 2 +
1.2
(−3)(−4)(−5)
+ (− x ) 3 +…
1.2.3
2 3
= 1 + 3x + 6 x + 10 x +…,
u n − r +1 − 3 − r +1 r+2
and r +1 = (− x ) = (–x)= x
ur r r r
3(r + 2)
i) If x = 3 , ur+1 > ur , so long as > 1, or so long as
4 4r
3r + 6 > 4r, that is, if r < 6 .
If r = 6, ur+1 = ur , and if r > 6 , ur+1 < ur .
Therefore u6 and u7 are equal to each other and are greater than all
other terms i.e. u1 < u2 < u3 < u4 < u5 < u6 = u7 ,
u6 = u7 > u8 > u9 > u10 > …
4
ii) If x = 7
, ur+1 > ur so long as 4(r+2) > 7r , that is if 3r < 8
8 2
i.e. r < = 2 <3
3 3
If r < 2, ur+1 > ur , but if r ≥ 3 , ur+1 < ur.
Therefore u3 is greater than all the other terms.
u 1 < u2 < u3 ; u 3 > u4 > u5 > …
Example 4.2:
1
Expand to four terms the binomial expression (2 + x ) 2 ,
2
i) in ascending powers of x, ii) in descending powers of x.
62
Algebra (Chapter 4) Binomial Expansion

Solution:
1 1 1
i) (2 + x ) 2 = (2) 2 (1 + x ) 2 =
2 4
1 (− 1 ) 1 ( − 1 )(− 3 )
1
= 2[ 1 + ( )+ x 2 2 ( )2 + 2 2
x 2 ( x ) 3 +…]=
2 4 2! 4 3! 4
2 3
= 2 [ 1 + x – x + x –…] , |x| < 4.
8 128 1024
1 1 1
ii) (2 + x)2 = x
( )2 (1 + 4)2 =
2 2 x
1 (− 1 ) 1 ( − 1 )(− 3 )
= x [ 1 + 1 ( 4 ) + 2 2 ( 4 )2 + 2 2 2 ( 4 ) 3 +…]=
2 2 x 2! x 3! x

= x [ 1 + 2 – 2 + 4 –…] , |x| > 4.

2 x x2 x3

Example 4.3:
Find to five decimal places the value of 3.96 .
Solution:
–½
3.96 = 4 × 0.99 = 2(1– 0.01) =
1 (− 1 ) 1 ( − 1 )( − 3 )
=2[1+ (−0.01) +
1 2 2
(−0.01) + 2 2 2 (−0.01) 3 +…]=
2
2 2! 3!
= 2[1– 0.005– 0.0000125– 0.0000001+…]
= 2[0.9949874] = 1.98997 to 5 decimal places
N.B. With the factor two outside the bracket the quantity inside
the bracket must be taken to seven decimal places accuracy.
§ 4.2 Determination of a binomial function
from its expansion:
So far we have been concerned mainly with the problem of

63
Algebra (Chapter 4) Binomial Expansion

determining the expansion from the given function. We now

consider the converse problem i.e. given a binomial series to
determine the function from which it is derived.
If we are given a power series in x we can determine by a
study of the coefficients whether or not the series is a binomial
expansion. It must be remembered that the complete expansion
may not be given and further that we may not be dealing with a
n
direct binomial expansion of the form (1+x) . The student must be
guided by his previous experience gained in dealing with
particular binomial expansions.
If we are given a numerical series the problem is in general
less simple than the previous case. For now another unknown
factor has been introduced, viz. the particular numerical value of
x, which has been substituted in the power series to given the
particular numerical series.
Example 4.4:
1 1.3 1.3.5
Sum the following series: S = – x+ x 2 –…
2 4 (4!) 25 (5!) 2 6 (6!)
Solution:
The general form of the coefficients suggests a binomial
expansion. First we observe that the factors of the numerators of
successive coefficients increase by 2. This suggests that the
denominator of the index of the power of the binomial expression

64
Algebra (Chapter 4) Binomial Expansion

is 2. This is confirmed by the presence of the powers of 2 in the

denominator of the coefficients.
Next we observe that the terms are alternately positive and
negative and that the powers of x increase by unity. This indicates
n
that the expansion will be in the form (1 + x) as distinct from
n
(1 – x) . The fact that the terms are alternately positive and
negative from the beginning does not indicate precisely whether n
is positive or negative. For some of the terms of the complete
binomial expansion may be missing from the beginning of the
given expansion.
The next point to be observed is that the “factorial” terms
in the denominators of the coefficients begin with 4! This shows
1
clearly that four terms of the series are missing and that is
2 4 (4!)
th
the coefficient of the 5 term of the complete binomial expansion.
4
Thus we must multiply each term of the given expansion by x ,
4
and divide the resulting sum by x .
1 th
Again if is the coefficient of the 5 term there
2 4 (4!)
must be four factors corresponding to n(n–1)(n–2)(n–3). Three
additional factors must then be inserted in each numerator to
secure that the number of factors in the numerator is the same as
that in the denominator. In order that the factors of the numerator
may remain in arithmetical progression the additional factors

65
Algebra (Chapter 4) Binomial Expansion

(which should be the same in each term) must be 5, 3, 1. Now, we

can deduce that the factors, which are missing, must be 5/2, 3/3,
and ½. Hence we must multiply each term of the expansion by
5.3.1, and dividing the resulting sum by 15. The given expansion
5/2
must form part of the binomial expansion of (1+x) . The precise
setting out of the working is as follows:
1 5.3.1.1 4 5.3.1.1.3 5 5.3.1.1.3.5 6
S = [ x – x + x –…]=
15x 4 2 4 (4!) 25 (5!) 2 6 (6!)
5 . 3 . 1 .(− 1 ) 5 . 3 . 1 .(− 1 )(− 3 )
−1 2 2 2 2
= [ x + 2 2 2 2 2 x5 +
4
15x 4 (4!) (5!)
5 . 3 . 1 .(− 1 )(− 3 )(− 5 )
+ 2 2 2 2 2 2 x 6 +…]=
(6!)
−1 5/2
= [(1+x) – 1– 5 x – 15 x 2 – 5 x 3 ] =
15x 4 2 8 16
1 5/2
= [1+ 5 x + 15 x 2 + 5 x 3 – (1+x) ].
15x 4 2 8 16

Example 4.5:
Sum to infinity S = 1 – 1.4 + 1.4.7 –…
12 12.18 12.18.24

Solution:
The factors in the numerators form an arithmetical progression
with common difference 3; we therefore divide each of these by 3.
The factors in the denominators form an arithmetical progression
with common difference 6; we therefore divide each of these by 6.
Hence we have:

66
Algebra (Chapter 4) Binomial Expansion
(1) ( 1 )( 4 ) ( 1 )( 4 )( 7 )
S= 3
( 3) – 3 3
( 3)2 + 3 3 3
( 3 )3 –
2 6 2.3 6 2.3.4 6
The factors of the denominators are in arithmetical progression,
but they do not begin with 1. Hence one additional factor namely
unity, has to be introduced into the denominator of each
coefficient; and as the number of factors in the numerator is the
same as that of the factors in the denominator, we have to
introduce an additional factor in the numerator also, which factor
is clearly – 2 . Then:
3
( − 2 )( 1 ) ( − 2 )( 1 )( 4 ) ( − 2 )( 1 )( 4 )( 7 )
– 2S= 3 3
(1) – 3 3 3
( 1 )2 + 3 3 3 3
( 1 ) 3 –…
3 2! 2 3! 2 4! 2
Again since the index of x [=(½ )] in every term must be the same
as the number of factors in the numerator or denominator of the
coefficient we have:
– 2 1 S = –1S =
3 2 3
( 2 )( − 1 ) ( 2 )( − 1 )( − 4 ) ( 2 )( − 1 )( − 4 )( − 7 )
= 3 3 1 2
( ) – 3 3
( ) + 3 3 3 3 ( 1 ) 3 –…
3 1 3
2! 2 3! 2 4! 2
Here the first two terms are evidently wanting. When we supply
these we find that:

(2 )
2
2/3
– 1 S = (1 + ½ ) – [1 + ( 2 )( 1 ) ] = 3 3 – 4 .
3 3 2 3

(2 )
2
Therefore S = 4 – 3 3 3 = 4 – 3 3 9 = 4 – 3 3 18 =4 – 3 3 18 .
4 8 2

Example 4.6:
n
If (2 + 3 ) = N + F, where N and n are positive integers and F is

67
Algebra (Chapter 4) Binomial Expansion

a positive fraction less than unity. Show that: (1– F )( N + F ) = 1,

and N is an odd integer.
Solution:
n
Let the integral part of (2 + 3 ) be denoted by N and its rational
part F. Then:
N+F=
n n n–1 n n–2 n n–3 n
=2 + C1 2 3 + C2 2 3 + C3 2 3 3 +…+( 3 ) . (4.7)
Now 2 – 3 is obviously a positive number less than 1, and
n
therefore (2 – 3 ) is also a positive fraction less than 1. Let this
be denoted by T. Therefore
n n n–1 n n–2 n n–3
T = 2 – C1 2 3 + C2 2 3 – C3 2 3 3 +
n n
+…+(–1) ( 3 ) (4.8)
Adding (4.7) and (4.8) makes the irrational term on the right side
disappear. Hence:
n n n–2 n n–4 2
N + F + T = 2 [2 + C2 2 3 – C4 2 3 +… .]
= an even integer. (4.9)
But F and T each is less than one and N being an integer we have
F + T = 1. (4.10)
From (4.9) and (4.10) we have N + 1 = an even integer. Therefore
n n
N is an odd integer. Again N + F = (2 + 3 ) and T = (2 – 3 ) .
n n n
Therefore (N + F) T = (2 + 3 ) (2 – 3 ) = (4 – 3) =1. (4.11)
But by (4.10) we have T = 1 – F, so that by (4.11) we have
(N + F)(1 – F) = 1.

68
Algebra (Chapter 4) Binomial Expansion

Example 4.7:
3n
Show that (2 – 7n–1) is divisible by 49.
Solution:
n n (n − 1) 2 n (n − 1)(n − 2) 3 n
Since (1+x) = 1 + n x+ x + x +…+ x .
2! 3!
n 2 n ( n − 1) n (n − 1)(n − 2) n-2
Then (1+x) – n x – 1 = x [ + x +…+ x ]
2! 3!
n 2
Hence [(1 + x) – n x–1] is divisible by x
n
If we put x = 7 we get 8 – 7n – 1 is divisible by 49
3 3 n
But 8 = 2 , so that: (2 ) –7n – 1, is divisible by 49
3n
i.e. 2 –7n – 1, is divisible by 49.
Similarly any number of problems can be built up by substituting
different values of x.

69
Algebra (Chapter 4) Binomial Expansion

EXERCISES 4
1) Write down the first five terms and the coefficient of x in the
expansions in ascending powers of x of the following functions.
In each case state when the expansion is valid.
–2 –½
i) (1 – x) , ii) (1–3x) , iii) 4 + x2
iv) 3 (8 − x 3 ) 2 , v) x 2 .
1+ x
2 3 4
[Answers: i) 1+ 2x + 3x + 4x + 5x +… ; r+1 ; |x| < 1,
2 3 4 ( 2 r )! 3 r
ii) 1 + 3 x + 27 x + 135 x + 2835 x +…; 2
( ) ; |x| < 1 ,x=– 1 ,
2 8 16 128 ( r!) 4 3 3
( 2p − 2)! 1 2p −1
iii) 2+ 1 x – 1 x + 1 x – 57 x +… ; (− 1)p −1
2 4 6 8
( ) ,
4 64 512 4 ( p −1) p! 4

if r = 2p; |x| < 2 and x = ±2.

3 6 9 7 x12+…];
iv) 2[2– 1 x – 1 x – 1 x–
6 288 (64)(81) 2.(12) 5
1.4.7...(3p − 5)
(−8) if r = 3p; |x| < 2, and x=±2.
( 24) p ( p!)
3 5 7 9 (r–1)/2
v) x –x + x – x + x +… ; r odd ; (–1) ; |x| < 1]
2) Find the named terms in the expansions of the following:
th –3 9
i) 10 term in (1– 4x) , if |x| < ¼. [Answer: u10 = 55(4x) ]
th 5/2
ii) 4 term in (1+3x) , if |x| < 1/3. [Answer: u4 = 135 x 3 ]
16
3) What is the first negative coefficient in the expansion of
[Answer: u5 = − 5 x 4 ]
5/2
(1+ x) , where |x| < 1.
128
21
4) How many terms are positive in the expansion of (1 − x 2 ) 4 ,
where |x| < 1? [Answer: two]
7 1
5) How many terms are negative in the expansion of (1 − x 2 ) 2 ,

70
Algebra (Chapter 4) Binomial Expansion

where |x| < 1? [Answer: Four]

3
6) Expand in ascending powers of x, as far as x ,
–1 –½ –1/3 3
(1 – x) – 2(1 – 2x) + (1–3x) . [Answer: 2 x .]
3
7) Which is the numerically greatest term (or terms) in the
expansions of the functions:
21/2
i) (1 + x) , when x = 3 , [Answer: u5]
4
–10 3
ii) (1 – x) , when x = , [Answer: u27 and u28]
4
–½
iii) (1 + x) , when x = ½ , [Answer: u1]
−3 1
iv) (1 + x ) 4, when x = 0.8 . [Answer: u9 and u10]
½
8) By expanding (100 – 2) , obtain the value of 2 to 5 places of
decimals. [Answer: 1.41421]
2n
9) If n is a positive integer, prove that the middle term of (1+x) ,
th –½
is the same as (n+1) term of (1 – 4x) .
10) Prove that:
1 1
(1 + x) 2 − (1 + x 2 ) 2
i) lim 1 1
= 1,
x →0
(1 − x 2 ) 2 − (1 − x) 2
1 1
(2 + x) 2 − (4 + 3x) 2 2
ii) lim 1 2
=– .
x →0 16
(1 + x) 3 − (1 + 2x) 3
Find the sum to infinity the following series
11) 1 + 10 + 10.16 + 10.16.22 +…
9 9.18 9.18.27
[Answer: x = – 2 , n = – 5 , S = 3 3 9 ]
3 3
12) 1 – 1.3 + 1.3.5 –…
24 24.32 24.32.40

71
Algebra (Chapter 4) Binomial Expansion

[Answer: x = 1 , n = 3 , S = 1 (179–80 5 )]
4 2 3
5 5.7 5.7.9
13) 2+ + + +…
2!.3 3!.32 4!.33
[Answer: x = 2 , n = 3 ; S= 3 3 ]
3 2
14) 1 + 6 ( 4 ) + 6.7 ( 4 ) 2 + 6.7.8 ( 4 ) 3 +…
5 1.2 5 1.2.3 5
6
[Answer: x = − 4 , n = – 6; S = 5 ]
5
2
15) 1 + 2( ) + 3( ) + … [Answer: x = – 1 , n = –2 ; S = 4]
1 1
2 2 2
16) 1 + 1 ( 3 ) + 1.3 ( 3 ) 2 + 1.3.5 ( 3 ) 3 +…
2 4 2.4 4 2.4.6 4
[Answer: x = – 3 , n = – 1 ; S=2]
4 2
m ( m +1 ) m ( m +1)( m + 2 )
17) 1 + m 2+ 22 + 23 +…
3 3.6 3.6.9
m
[Answer: x = – 2 , n = –m; S = 3 ]
3
18) 1 –1+ 1.4 – 1 .4 .7 +…
5 5.10 5.10.15
[Answer: x = – 3 , n = – 1 ; S = 1 3 5 ]
5 3 2
19) 1 . 1 – 1 . 1 + 1.3 . 1 –…
2 2 4 2.4 28 2.4.6 212
[Answer: x = 14 , n = 1 ; S = 1 17 –1]
2 2 4
20) 1 + 1.3 + 1.3.5 +…[Answer: x = – 2 , n = – 1 ; S = 3 – 1]
3 3.6 3.6.9 3 2
21) 3 ( 1 ) + 3.5 ( 1 ) 2 + 3.5.7 ( 1 ) 3 +…
2 3 2.3 3 2.3.4 3
[Answer: x = – 2 , n = – 1 ; S= 3 3 – 4]
3 2
22) 3 + 3.4 + 3.4.5 + …
2.4 2.4.6 2.4.6.8
[Answer: x = – 1 , n = –2 ; S = 1]
2
23) 1 + 1.3 x 2 + 1.3.5.7 x 4 +… ; |x| < 1
2.4 2.4.6.8

72
Algebra (Chapter 4) Binomial Expansion
–½ –½
[Answer: S = 1 {(1+x) + (1–x) }]
2
n
24) The expansion of (1 + x) is valid
if x = 1, when n + 1 > 0 ; if x = –1, when n > 0.
Prove that the sum to infinity of
1 + 3 + 3.5 + 3.5.7 + … [Answer: x = 1 , n = 3 ; S=2]
8 8.10 8.10.12 2
1
is double that of + 1.3 + 1.3.5 1
+… [Answer: x = 1, n = ; S = 1]
4 4.6 4.6.8 2
25) 1 + 1 .4 + 1.4.7 + …[Answer: x = – , n=– ; S = 3 2 –1]
1 1
6 6.12 6.12.18 2 3
n
26) If (3 + 7 ) = N + F, where N and n are positive integers and
F is a positive fraction less than one, show that N is an odd integer
n
and that (1 – F)(N + F) = 2 .
n
27) If (5 + 2 6 ) = N + F, where N and n are positive integers
and F is a positive fraction less than one, show that N is an odd
integer and that (1 – F)(N + F) = 1.
n
28) If (7 + 4 3 ) = N + F, where N and n are positive integers
and F is a positive fraction less than one, show that N is an odd
integer and that (1 – F)(N + F) = 1.
n
29) If (8 + 3 7 ) = N + F, where N and n are positive integers
and F is a positive fraction less than one, show that N is an odd
integer and that (1 – F)(N + F) = 1.
30) Show that
4n
i) 2 – 15n – 1, is divisible by 225,
2n
ii) 2 – 3n – 1, is divisible by 9.
3n+3
iii) 2 – 7n – 1, is divisible by 49.

73
Algebra (Chapter 4) Binomial Expansion
2n 2
iv) 6 – 35n – 1, is divisible by (35) .
31) Find constants a, b, c such that
x2 − 5x + 1 a b c
= + + .
(x + 1)(x+ 2)(x+ 3) (x +1) (x +1)(x+ 2) (x + 1)(x+ 2)(x+ 3)
[Answer: a = 1, b = –10, c = 25]
32) Prove the identity: 16(x+2)(x–4)(x–6) =
−15 15 7 9
=(x–1)(x–3)(x–5)(x–7)*[ + + + ]
x −1 x − 3 x − 5 x − 7
8x + 25
33) Resolve the expression into partial fractions.
(1+ 2x)2(3− x)
Find the first three terms in the expansion of the expression in
descending powers of x.
2 6 1 2
[Answer: + + ; 8 1 – 27 8 x + 80 1 x ]
1+ 2x (1 + 2x) 2 1− x 3 9 27

Prove the following results:

34) 1 + 1 + 1.3 + 1.3.5 + … = 2.
4 4.8
4.8.12
35) 3 3 = 1 + 1 + 1.4 1 + 1.4.7 1 + …
2 32 1.2 3 4 1.2.3 36
36) 2 = 1 + 1 + 1.3 2 + 1.3.5 3 + …
2.2 2.3.3 2.3.4.3
37) 8 3 = 1 + 3 + 3 . 5 + 3 . 5 . 7 + …
9 8 8 16 8 16 24
2
38) If x be so small that x and higher powers of x may be
½ –2½
neglected, show that (1 + 2x) (1 – 4x) = 1 + 11 x.
r 2 –1
39) What is the coefficient of x in the expansion of (1 + x + x ) .
[Note that 1 + x + x = 1− x ]
2 3
1− x
[Answer: 1 , – 1 , 0 ; if r = 3p, 3p+1, 3p+2]
74
Algebra (Chapter 4) Binomial Expansion

40) What is the coefficient of x in the expansion of 1+ x in

1− x

ascending powers of x?
( 2 n )!
[Answer: , where n = 1 r or 1 (r–1) as r is even or odd]
4 n ( n!) 2 2 2
1+ 2 x
41) Find the approximation for 3 1+ 3x
, if x is so small that terms
3
involving x may be neglected. [Answer: 1 + 1 x]
2
42) If x is large find an approximation for x2 + 1– x2 −1.
[Answer: x 2 + 1 – x 2 − 1 ≅ 1 , neglecting 15 ]
x x

75
Algebra (Chapter 5) Matrices

CHAPTER 5
MATRICES
In this chapter we discuss the properties of matrices.
Unlike a determinant, we cannot attach a value to a matrix.
Matrices are not functions, but we shall find that, with certain
important exceptions, they obey the ordinary laws of algebra, if
the operations of addition, multiplication, etc., of matrices are
suitable defined.
The student is already familiar with some operations on
matrices. A number of proofs will be omitted, and for further
details the interested student may consult the references given at
the end of this book. ,
A matrix of order m × n, or m by n matrix is a rectangular
array of numbers having m rows and n columns. It can be written
in the form:
⎡ a11 a12 ... a1n ⎤ column j
⎢a a 22 ... a 2n ⎥
A = ⎢ 21 ⎥ (5.1) row i aij
⎢ ... ... ... ... ⎥
⎢ ⎥
⎣a m1 a m 2 ... a mn ⎦

Each number a in this matrix is called an element. The subscripts i

and j indicate respectively the row and column of the matrix in
which the element appears. We shall often denote a matrix by a
letter, such as A in (5.1), or by the symbol [aij], which shows a
representative element. A matrix having only one row is called a

76
Algebra (Chapter 5) Matrices

row matrix (or row vector), while a matrix having only one
column is called column matrix (or column vector). If the numbers
of rows m and columns n are equal, the matrix is called a square
matrix of order n × n or briefly n.
§ 5.1 Matrix Algebra:
1. Equality of matrices: Two matrices A = [aij] and B = [bij] of
the same order are equal if and only if aij = bij for all i , j.
2. Addition: If A = [aij] and B = [bij] have the same order, we
define the sum of A and B as A + B = C where, cij = aij + bij.
Note that A + B has the same order as A and B. The sum of two
matrices with different orders is not defined. Note that matrices
satisfy the commutative and associative laws for addition.
i.e. for any three matrices A, B, C of the same order:
A + B = B + A , A + (B + C) = (A + B) + C .
3. Multiplication by a number: If A = [aij] and k is any number,
we define the product of A by k as kA = Ak = [k aij ]. Note that A
and kA have the same order. We also introduce the following
notation –A = (–1) A.
4. Subtraction: A – B is defined as A + (–B). It is called the
difference of the two matrices.
5. Null matrix: A matrix having every element zero is called a
null matrix or zero matrix, and is written 0. For any matrix A
having the same order as the matrix 0, we have A + 0 = 0 + A = A.
6. Multiplication of matrices: If A = [aij] is an m × n matrix

77
Algebra (Chapter 5) Matrices

while B = [bij] is an n × p matrix, then we define the product A . B

or AB of A and B as the matrix C = [cij] of order m × p, where
k =n
cij = ∑ a ik b kj . (5.2)
k =1
Note that matrix multiplication is defined if and only if the
number of columns of A is the same as the number of rows of B.
Such matrices are sometimes called conformable. Note also that in
general AB ≠ BA. i.e. the commutative law for multiplication of
matrices is not satisfied in general. However:
A . (B .C) = (A . B) . C , (5.3)
A . (B + C) = A . B + A . C , (5.4)
(B + C) . A = B . A + C . A . (5.5)
Assuming that A, B, C are conformable for the indicated sums and
products.
A matrix can be multiplied by itself if and only if it is a
2
square matrix. The product A . A can in such case be written A .
3 2
Similarly we define powers of a square matrix, i.e. A = A . A ,
4 3
A = A . A , etc. The points of difference between ordinary a, b
and matrices A, B are i) whereas a b = b a, A B is not usually the
same as B A; ii) Whereas a b = 0 implies that either a or b (or
both) is zero, the equation A B = 0 does not necessary imply that
either A or B is zero. ,
⎡a b ⎤ ⎡ b 0⎤ ⎡0 0 ⎤
For example: if A = ⎢ ⎥ , B=⎢ ⎥ , then AB= ⎢ ⎥ = 0,
⎣ 0 0 ⎦ ⎣ − a 0 ⎦ ⎣ 0 0 ⎦

78
Algebra (Chapter 5) Matrices
⎡ ab b2 ⎤
BA = ⎢ ⎥.
⎣− a − ab⎦
2

i.e. the product AB is the zero matrix, although neither A nor B is

the zero matrix.
iii) Whereas if a ≠ 0, a b = a c imply b = c, the equation A B = A C
does not necessary imply B = C.
For example,
⎡a b ⎤ ⎡ b 0⎤ ⎡0 − b ⎤
if A = ⎢ ⎥ , B=⎢ ⎥ , C=⎢ ⎥ , then A B = A C = 0
⎣ 0 0 ⎦ ⎣ − a 0 ⎦ ⎣ 0 a ⎦
and B ≠ C. ,
7. Transpose of a matrix: If we interchange rows and columns of
a matrix A, the resulting matrix is called the transpose of A and is
t t
denoted by A or A′. i.e. if A = [aij ], then A = A′ = [aji ]. It is easy
t t t t t t t t
to prove that: (A + B) = A + B , (A B) = B A , (A ) = A.
⎡− 3 2⎤
⎡− 3 0 2⎤
The transpose of A = ⎢ 0 1 ⎥ is A = ⎢
t
⎢ ⎥ ⎥ .
⎢⎣ 2 ⎣ 2 1 4 ⎦
4 ⎥⎦

8. Symmetric and skew-symmetric matrix:

t
A square matrix A is called symmetric if A =A and skew-
t
symmetric if A = – A.
⎡ 1 − 3⎤ ⎡ 0 1⎤
If A = ⎢ ⎥ ,B ⎢
0⎥⎦
then A is symmetric and B is
⎣− 3 2 ⎦ ⎣−1
skew-symmetric.
If A is any real square matrix, then,

79
Algebra (Chapter 5) Matrices
t t
A = S + K , where S = 1 (A + A ) , K= 1 (A – A ) .
2 2
t t t t t
But since S = 1 (A + A ) = 1 (A + A) = 1 (A + A ) = S,
2 2 2
1 t
it follows that S = = (A + A ) is symmetric. Also since
2
t 1 t t t t
K = (A – A ) = (A – A) = – 1 (A – A ) = –K,
1
2 2 2
t
it follows that K= 1 (A – A ) is a skew-symmetric, i.e. any real
2
square matrix can always be expressed as the sum of a real
symmetric matrix and a real skew-symmetric matrix.
9. Principal diagonal and trace of matrix:
If A = [aij] is a square matrix, then the diagonal which contains all
elements a for which i = j is called the principal or main diagonal
and the sum of all such elements is called the trace of A.
The bold letters indicates the principal diagonal of the matrix,
⎡1 4 5 ⎤
⎢7 2 6⎥ and the trace of the matrix is 1 + 2 + 3 = 6. ,
⎢ ⎥
⎢⎣8 9 3⎥⎦
A matrix for which aij = 0 when i ≠ j is called a diagonal
⎡d11 0 ... 0 ⎤
⎢ 0 d 22 ... 0 ⎥
matrix. The matrix D = ⎢ ⎥ is called a diagonal
⎢ ... ... ... ... ⎥
⎢ ⎥
⎣ 0 0 0 d nn ⎦
matrix. It will frequently be written as D = diag. (d11, d22 ... dnn).
If in the diagonal matrix, d11= d22 =...= dnn = k, D is called a
scalar matrix; if in addition, k = 1, the matrix is called the identity
matrix and is denoted by In.

80
Algebra (Chapter 5) Matrices

10. Unit matrix:

A square matrix n × n in which all elements of the
principal diagonal are all 1, while all other elements are zero is
called the unit matrix and is denoted by In. When the order is
evident or immaterial, “I” will denote an identity matrix. An
important property of “I” is that:
m
A I = I A = A, I = I, m = 1, 2, 3, ... .
The unit matrix plays a role in matrix algebra similar to that
played by the number one in ordinary algebra.
11. Complex conjugate of a matrix:
If all elements aij of a matrix A are replaced by their complex
conjugates a ij , the matrix is called the complex conjugate of A
and is denoted by A i.e. [a ij ] = [ a ij ].
⎡1 + 2i 4 + 3i ⎤ ⎡1 − 2i 4 − 3i ⎤
Let A = ⎢ ⎥ , we have A = ⎢2 − 5i 3 + 4i ⎥ .
⎣2 + 5i 3 − 4i ⎦ ⎣ ⎦
It is easy to see that (A) = A.
12. Hermitian and skew-hermitian matrices:
A square matrix A which is the same as the complex conjugate of
t′
its transpose, i.e. A = A = A , is called hermitian.
t ′
If A = – A = – A ,then A is called skew-hermitian.
If A is real, these reduce to symmetric and skew-symmetric
matrices respectively:
⎡ 0 i⎤ t ⎡0 − i ⎤
Let A = ⎢ ⎥ , we have A = A′ = ⎢ i 0 ⎥ and,
⎣ − i 0 ⎦ ⎣ ⎦

81
Algebra (Chapter 5) Matrices
t ′ ⎡0 i⎤
A =A = ⎢ ⎥ =A. Thus A is hermitian.
⎣− i 0⎦
Exercise:
1) If A is a square matrix, then A + A t is hermitian and A – A t is
skew-hermitian.
2) Every square matrix A with complex elements can be written as
t
the sum of a hermitian matrix B = 1 (A + A ) and a skew-
2
t
hermitian matrix C = = 1 (A – A ).
2
3) (A) = A, 4) (kA ) = k A , 5) A + B = A + B
t
6) A.B = A . B 7) A = A t .
§5.2 Products by partitioning:
Let A = [aij] be of order k × m and B = [bij] be of order m × n. In
forming the product AB, the matrix A is in effect partitioned into
k matrices of order 1 × m and B into n matrices of order m × 1.
Other partitions may be used. For example, let A and B be
partitioned into matrices of indicated orders by drawing a dotted
lines as:
⎡ (k1 × m1 ) (k1 × m 2 ) (k1 × m 3 ) ⎤ ⎡ (m 1 × n1 ) (m 1 × n 2 ) ⎤
⎢
A = ⎢ − − − − − − − − − − − − − − − −⎥ , B =
⎢
⎢ − − − − − − − − − − ⎥
⎥ ⎥
⎢(m 2 × n1 ) (m 2 × n 2 )⎥
⎢⎣(k 2 × m1 ) (k 2 × m 2 ) (k 2 × m 3 ) ⎥⎦ ⎢ ⎥
⎢ − − − − − − − − − − ⎥
⎢(m × n ) ( m 3 × n 2 ) ⎥⎦
⎣ 3 1

⎡ A11 A12 A13 ⎤ ⎡ B 11 B 12 ⎤

or A = ⎢− − − − − − − − − − − −⎥ , B = ⎢
⎢− − − − − − − −⎥ , then
⎢ ⎥ ⎥
⎢ B 22 ⎥
⎢⎣ A 21 A 22 A 23 ⎥⎦ ⎢
B 21
⎥
⎢− − − − − − − −⎥
⎢ B 31 B 32 ⎥⎦
⎣

82
Algebra (Chapter 5) Matrices
⎡A B + A B + A B A11 B12 + A12 B22 + A13 B32 ⎤
AB = ⎢ 11 11 12 21 13 31 ⎥
⎣A21 B11 + A22 B21 + A23 B31 A21 B12 + A22 B22 + A23 B32⎦
⎡C C12 ⎤
= ⎢ 11 ⎥ = C.
⎣C 21 C 22 ⎦
In any such partitioning, it is necessary that the columns of A and
the rows of B be partitioned in exactly the same way; however, k1,
k2 ; n1 , n2 may be any nonnegative (including 0) integers such
that k1 + k2 = k and n1 + n2 = n.
Example 5.1:
Compute AB given
⎡1 2 2⎤
⎢1 1 ⎡2 3 1⎤
3⎥ ⎢1 2 0⎥ . Partitioning so that
A= ⎢ ⎥ , and B =
⎢1 1 1⎥ ⎢ ⎥
⎢ ⎥ ⎢⎣0 0 1⎥⎦
⎣0 0 2⎦

We have

⎡ ⎡1 2 ⎤ ⎡ 2 3⎤ ⎡ 2⎤ ⎡1 2⎤ ⎡1⎤ ⎡2⎤ ⎤
⎢⎢ + [0 0] ⎢1 1 ⎥ ⎢0⎥ + ⎢3⎥[1]⎥
⎢ ⎣1 1 ⎥⎦ ⎢⎣1 2⎥⎦ ⎢⎣3⎥⎦ ⎣ ⎦⎣ ⎦ ⎣ ⎦ ⎥
AB = ⎢ ⎥=
⎢ ⎡1 1 ⎤ ⎡ 2 3 ⎤ ⎡1 ⎤ ⎡1 1⎤ ⎡1⎤ ⎡1 ⎤ ⎥
⎢⎢ + [0 0] ⎢0 0⎥ ⎢0⎥ + ⎢2⎥[1]⎥
⎣ ⎣0 0⎥⎦ ⎢⎣1 2⎥⎦ ⎢⎣2⎥⎦ ⎣ ⎦⎣ ⎦ ⎣ ⎦ ⎦

83
Algebra (Chapter 5) Matrices
⎡ ⎡4 7 ⎤ ⎡0 0 ⎤ ⎡1⎤ ⎡2⎤ ⎤
⎢ ⎢ + ⎢1⎥ + ⎢3⎥ ⎥ ⎡4 7 3⎤
⎢ ⎣3 5⎥⎦ ⎢⎣0 0⎥⎦ ⎣ ⎦ ⎣ ⎦⎥ ⎢3 5 4⎥
=⎢ ⎥= ⎢ ⎥
⎢ ⎡3 ⎢3 5 2⎥
5 ⎤ ⎡0 0 ⎤ ⎡1⎤ ⎡1 ⎤ ⎥ ⎢ ⎥
⎢⎢ + + ⎢0 ⎥ + ⎢ 2 ⎥ ⎥
⎣ ⎣0 0⎥⎦ ⎢⎣0 0⎥⎦ ⎣ ⎦ ⎣ ⎦⎦ ⎣0 0 2⎦

Example 5.2:

Using the indicated partitioning, we have,

⎡ ⎡3⎤ ⎡1 1 ⎤ ⎤
⎢[1 2]⎢ ⎥ + [− 1][2] [1 2]⎢ ⎥ + [− 1][0 1] ⎥
⎢ ⎣1 ⎦ ⎣3 2 ⎦ ⎥
AB = ⎢ ⎥=
⎢ ⎡3⎤ ⎡1 1 ⎤ ⎥
⎢ [3 0]⎢ ⎥ + [2][2] [3 0]⎢ ⎥ + [− 2 ][0 1 ]⎥
⎣ ⎣1⎦ ⎣3 2⎦ ⎦
⎡5 − 2 [7 5] + [0 − 1]⎤ ⎡ 3 7 4⎤
=⎢ ⎥ = ⎢13 3 5⎥
⎣9 − 4 [3 3] + [0 2 ] ⎦ ⎣ ⎦

§ 5.3: Determinants:
The student is already familiar with the determinants. If the matrix
A is square, then we associate with A, a number denoted by |A|,
a11 a12 ... a1n
a 21 a 22 ... a 2n
|A| = det. A =
... ... ... ...
a n1 a n 2 ... a nn

Called the determinant of A of order n.

84
Algebra (Chapter 5) Matrices

1. Minor:
Given any element aij, of A we associate a new determinant of
th
order (n–1) obtained by removing all elements of the i row and
th
j column called the minor of aij.
2. Co factor:
i+j
If we multiply the minor of aij by (–1) , the result is called the co
factor of aij and is denoted by Aij.
The value of the determinant is then defined as the sum of
the products of the elements in any row (or column) by their
corresponding co factors and is called the Laplace expansion. In
symbols:
n
|A| = det. A = a1j A1j +a2j A2j +...+anj Anj = ∑ a ijA ij =
i =1
n
= aj1 Aj1 +aj2 Aj2 +...+ajn Ajn = ∑ a ji A ji .
i =1

Theorem 5.1:
The sum of the products of the elements of any raw (or column)
by the co factors of another row (or column) is zero.
In symbols,
n n
∑ a ijA kj = 0 , or ∑ a ji A jk = 0, if i ≠ k.
j=1 j=1

If i = k, the sum is det. A.

n n
i.e. ∑ a ijA kj = ∑ a ji A jk = |A| δik ,
j=1 j=1

85
Algebra (Chapter 5) Matrices
⎧1 if i = k,
where δij = ⎨
⎩0 if i ≠ k.
§ 5.4 Inverse of a matrix:
If for a given square matrix A there exists a matrix B such that
–1
A B = I, then B is called an inverse of A and is denoted by A .
The following theorem is fundamental.
Theorem 5.2:
If A is a non-singular matrix of order m (i.e. |A| ≠ 0), then there
–1 –1 –1
exists a unique inverse A , such that AA =A A = I, and we
–1
can express A in the following form:
⎡ A11 A 21 ... A n1 ⎤
–1
A =
[ ] [ ]
A adj A ji
=
t
=
⎢
1 ⎢A12 A 22 ... A n 2 ⎥
⎥.
A A A ⎢ ... ... ... ... ⎥
⎢ ⎥
⎣A1n A 2n ... A nn ⎦
adj t
Where [A ] = [Aij ] is the matrix of co factors Aij and [Aji] =

=[Aij] is its transpose. It is easy to see that

[ ]
A adj
A = I.
A
If A, B are two non-singular matrices then,
–1 –1 –1 –1 –1
(AB) = B A , (A ) = A.

Orthogonal matrices:
–1 t
A non-singular matrix A is called orthogonal if A =A.
Of course, we can say that a non-singular matrix is orthogonal if
t
A A = I.

86
Algebra (Chapter 5) Matrices
⎡1 2 −2 ⎤ ⎡1 −2 2⎤
⎢3 3 3 ⎥ ⎢3 3 3⎥
If A= ⎢ − 2 2 1 ⎥ , then A–1= ⎢ 2 2 1 ⎥ = At.
⎢ 32 3
1
3 ⎥
2
⎢ −32 3
1
3⎥
2
⎢⎣ 3 3 3 ⎥⎦ ⎢⎣ 3 3 3 ⎥⎦

It is easy to check it.

Corollaries:
1. If A is an orthogonal matrix, then it is easy to show that:
|A| = ±1.
2. The product of two or more orthogonal matrices is also
orthogonal matrix.
3. The inverse and the transpose of an orthogonal matrix are
orthogonal.
4. If A is orthogonal matrix, then its column vectors (row
vectors) are orthonormal vectors, and conversely.
§ 5.5 Cramer's Rule:
A set of equation having the form
a11 x1 + a12 x2 + … + a1n xn = b1 , ⎫
⎪
a21 x1 + a22 x2 + … + a2n xn = b2 , ⎪
⎬ (5.6)
… … … … … … … … ⎪
⎪⎭
am1 x1 + am2 x2 + … + amn xn = bm .
is called a system of m linear equations in the n unknowns x1,
x2...xn. If b1, b2 ...bn are all zero the system is called
homogeneous. If they are not all zero it is called non-
homogeneous. Any set of numbers x1, x2...xn which satisfies (5.6)

87
Algebra (Chapter 5) Matrices

is called a solution of the system.

In matrix form (5.6) can be written:
⎡ a11 a12 ... a1n ⎤ ⎡ x1 ⎤ ⎡ b1 ⎤
⎢a a 22 ... a 2n ⎥ ⎢ x 2 ⎥ ⎢ b 2 ⎥
⎢ 21 ⎥ ⎢ ⎥= ⎢ ⎥, (5.7)
⎢ ... ... ... ... ⎥ ⎢ ... ⎥ ⎢ ... ⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎣a m1 a m 2 ... a mn ⎦ ⎣ x n ⎦ ⎣b m ⎦
Or more briefly A X = B (5.8)
–1
If m = n, and if A is a non-singular matrix so that A exists, we
–1
can solve (5.7) or (5.8) by writing: X = A B, (5.9)
and the system has a unique solution.
Alternatively we can express the unknown x1, x2...xn as
x1 = |A|1/|A| , x2 = |A|2/|A| , and xn = |A|n/|A|, where |A| is the
determinant of the system, and |A|k , k = 1, 2, 3…n is the
determinant obtained from |A| by removing the k column and
replacing it by the column vector B. The rule expressed in (5.10)
is called Cramer's Rule.
The following four cases can arise:
Case 1: |A| ≠ 0 , B ≠ 0. In this case there will be a unique
solution where not all xk will be zero.
Case 2: |A| ≠ 0, B = 0. In this case the only solution will be x1 =
= x2 = ... = xn = 0. i.e. X = 0. This is often called the trivial
solution.
Case 3: |A| = 0, B = 0. In this case there will be infinitely many
solutions other than the trivial solution. This means that at least

88
Algebra (Chapter 5) Matrices

one of the equations can be obtained from the others, i.e. the
equations are linearly dependent.
Case 4: |A| = 0, B ≠ 0. In this case infinitely many solutions will
exist if and only if all the determinants |A|k are zero. Otherwise,
there will be no solution.
We have a specific method for the calculation of the
inverse matrix and to solve system of linear equations (Cramer's
Rule). In both cases the evaluation of a number of determinants is
required. Unfortunately, unless n is quite small, the evaluation of
determinants is a very tedious and inefficient procedure, and so
the methods just mentioned for solution of equations and inversion
of matrices are very undesirable. In the following sections we
proceed to describe alternative methods which are to be preferred
for such calculations.
§ 5.6: Linear dependence of vectors:
By an n-dimensional vector or n-vector X is meant an ordered set
of elements xi as, X = [x1, x2, ... , xn ]. The elements x1 , x2, ... ,xn
are called respectively the first, second,..., n components of X. It is
more convenient to write the components of a vector in a column
⎡ x1 ⎤
t ⎢x ⎥
as X = [x1 ,x2 ...xn ] = ⎢ 2 ⎥ .
t
⎢ .... ⎥
⎢ ⎥
⎣x n ⎦
t
We shall speak of X as a row vector and X as a column vector.

89
Algebra (Chapter 5) Matrices

We may, then, consider the m × n matrix A as defining m row

vectors or as defining n column vectors.
The m vectors X1, X2...Xm are said to be linearly
dependent provided there exist m constants k1 ,k2...km, not all
zero, such that:
k1 X1 + k2 X2 + ... + km Xm = 0. (5.10)
Otherwise, the m vector are said to be linearly independent.
Example 5.3:
Consider the following vectors
X = [1, –4, 3], Y = [2, –3, 2], Z = [–4, 1, 0], T = [–4, 6, –4].
It is easy to see that
i) 2X – 5Y =2[1, –4, 3] – 5[2, –3, 2] =
=[2, –8, 6] – [10, 15, 10] = [–8, 7, –4],
ii) 2Y + T = 2[2, –3, 2] + [–4, 6, –4] =
= [4, –6, 4] + [–4, 6, –4] = [0, 0, 0] = 0,
iii) 2X – 3Y – Z = 0,
iv) 2X – Y – Z + T = 0.
From (ii) the vectors Y and T are linearly dependent. So also are
X, Y, Z by (iii) and the entire said by (iv). The vectors X and Y,
however, are linearly independent:
Let a X + b Y = 0 , then
a [1, –4, 3] + b [2, –3, 2] = [a + 2b, 4a – 3b, 3a + 2b] = [0 , 0 , 0].
i.e. a + 2b = 0, (1)

90
Algebra (Chapter 5) Matrices

4a + 3b = 0, (2)
3a + 2b = 0 . (3)
Equation (3) – Equation (1) gives a = 0 and then b = 0
i.e. X and Y are linearly independent.
Any n-vector X and n-zero vector 0 are linearly dependent.
A vector Y is said to expressible as a linear combination of
the vectors X1 , X2 … Xm if there exist k1 , k2 … km such that
Y = k1 X1 + k2X2 +… +kmXm (5.12)
Basic Theorems:
1. If m vectors are linearly dependent, some one of them may
always be expressed as a linear combination of the others.
2. If m vectors X1 , X2 … Xm are linearly independent while the
set obtained by adding another vector Y is linearly dependent,
then Y can be expressed as a linear combination of X1 , X2 … Xm.
3. If among the m vectors X1 , X2 … Xm there is a subset of r < m
vectors which are linearly dependent, the vectors of the entire set
are linearly dependent.
Example 5.4:
From example 5.3, the vectors X and Y are linearly independent
while X, Y and Z are linearly dependent, satisfying the relation
2X – 3Y – Z = 0. Clearly Z = 2X – 3Y.
The set of vectors X and Y are linearly independent; the four
vectors X, Y, Z and T linearly dependent vectors.

91
Algebra (Chapter 5) Matrices

§ 5.7: The rank of a matrix:

A nonzero matrix A is said to have rank r if at least one of
its r-square minors is different from zero, while every (r+1)-
square minor, if any is zero. A zero matrix is said to have rank 0.
Example 5.4:
⎡1 2 3⎤
The rank of A = ⎢2 3 4⎥ is r = 2, since
1 2
= –1 ≠ 0,
⎢ ⎥ 2 3
⎢⎣2 4 6⎥⎦

while |A| = 0.
An n-square matrix A is called non-singular if its rank r = n, that
is, if |A| ≠ 0. Otherwise, A is called singular.
From |A B| = |A| |B| follows: The product of two or more non-
singular n-square matrices is non-singular; the product of two or
more n-square matrices is singular if at least one of the matrices is
singular.

Elementary operations:
The following operations on a matrix called elementary
operations:
th th
1) The interchange of the i and j rows, denoted by Rij . The
th th
interchange of the i and j columns, denoted by Cij.
th
2) The multiplication of every element of the i row by a nonzero
scalar k, denoted by Ri (k). The multiplication of every element of
th
the i column by a nonzero scalar k, denoted by Ci(k).
92
Algebra (Chapter 5) Matrices
th
3) The addition to the elements of the i row of k, a scalar, times
th
the corresponding elements of the j row denoted by Rij (k). The
th
addition to the elements of the i column of k, a scalar, times the
th
corresponding elements of the j column, denoted by Cij(k).
The operations R are called elementary row operations; the
operations C are called elementary column operations. It is easy to
see that an elementary operation can’t alter the order of a matrix.
Theorem 5.3:
The rank of a matrix is not altered by any sequence of elementary
operations.
It is interesting and important to note that an elementary
operation involving the rows of a matrix A can be accomplished
by pre-multiplying A by a unit matrix on whose rows the same
elementary operation has been performed and any elementary
operation involving the columns of A can be accomplished by
post multiplying A by a unit matrix on whose columns the same
elementary operation has been performed:

Example 5.5:
⎡1 2 0⎤ ⎡1 2 0⎤
When A = ⎢ 2 3 − 1⎥ , R23 A = ⎢− 1 − 1 2 ⎥ ,
⎢ ⎥ ⎢ ⎥
⎢⎣− 1 − 1 2 ⎥⎦ ⎢⎣ 2 3 − 1⎥⎦
Interchanges the second and third rows

93
Algebra (Chapter 5) Matrices
⎡1 0 0⎤ ⎡1 0 0⎤ ⎡1 0 0⎤
Let B = R23 I = ⎢0 0 1⎥ , D = ⎢0 3 0 ⎥ , E = ⎢3 1 0⎥ ,
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣0 1 0⎥⎦ ⎢⎣0 0 1⎥⎦ ⎢⎣0 0 1⎥⎦
⎡1 0 0⎤ ⎡ 1 2 0⎤ ⎡1 2 0⎤
and BA = ⎢0 0 1⎥ ⎢ 2 3 − 1⎥ = ⎢− 1 − 1 2 ⎥ = R23 A.
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢⎣0 1 0⎥⎦ ⎢⎣− 1 − 1 2 ⎥⎦ ⎢⎣ 2 3 − 1⎥⎦
Multiplying the second row by 3, then
⎡1 0 0⎤ ⎡ 1 2 0⎤ ⎡1 2 0⎤
DA = ⎢0 3 0⎥ ⎢ 2 3 − 1⎥ = ⎢ 6 9 − 3⎥ =R2(3) A.
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢⎣0 0 1⎥⎦ ⎢⎣− 1 − 1 2 ⎥⎦ ⎢⎣− 1 − 1 2 ⎥⎦

Adds to the second row of A three times the first row; then
⎡1 0 0⎤ ⎡ 1 2 0⎤ ⎡1 2 0⎤
⎢
EA = 3 1 0 2 ⎥ ⎢ ⎥
3 −1 = 5 ⎢ 9 − 1⎥ = R12(3) A.
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢⎣0 0 1⎥⎦ ⎢⎣− 1 − 1 2 ⎥⎦ ⎢⎣− 1 − 1 2 ⎥⎦

Adds to the second column of A, three times the first column; then
⎡1 2 0 ⎤ ⎡1 3 0⎤ ⎡ 1 5 0⎤
AF = ⎢ 2 3 − 1⎥ ⎢0 1 0⎥ = ⎢ 2 9 − 1⎥ = AC12(3).
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢⎣− 1 − 1 2 ⎥⎦ ⎢⎣0 0 1⎥⎦ ⎢⎣− 1 − 4 2 ⎥⎦

Elementary matrices:
The matrix, which results when an elementary row (column)
operation is applied to the identity matrix I is called an elementary
row (column) matrix. Here, an elementary matrix will be denoted
by the symbol introduced to denote the elementary operation,
which produces the matrix.

94
Algebra (Chapter 5) Matrices

In example 5.5; B = R23 = C23 , D = R2 (3) = C2 (3),

E = R12 (3) = C21 (3), F = R21 (3) = C12 (3). ,
In general; Rij = Cij , Ri (k)=Ci (k), Rij (k)=Cji (k).
It is easy to see that every elementary matrix is non-singular. In
fact, the inverse of each elementary matrix is again an elementary
matrix.
R ij−1 = R ij , Cij−1 = Cij ; [Ri(k)] = Ri( 1 ) , [Ci(k)] = Ci( 1 );
–1 –1
k k
–1 –1
[Rij(k)] = Rij(–k) , [Cij(k)] = Cij(–k).
The effect of applying an elementary operation to an m × n matrix
A can be produced by multiplying A by an elementary matrix. To
effect a given elementary row operation on A of order m × n,
apply the operation Im to form the corresponding elementary
matrix R* and multiply A on the left by R* i.e. R* A. To effect a
given elementary column operation on A of order m × n, apply the
operation to In to form the corresponding elementary matrix C*
and multiply A on the right by C* i.e. A C*.
Equivalent matrices:
Two matrices A and B are called equivalent, A∼B, if one can be
obtained from the other by a sequence of elementary operations.
Equivalent matrices have the same order and the same rank.
Example 5.6:
Applying in turn the elementary operations R12(–3); R13 (1);
R21(1) ; R23 (1); R2 (–½) on the matrix A, we get

95
Algebra (Chapter 5) Matrices
⎡1 2 −1 3 ⎤ ⎡1 2 −1 3 ⎤
A= ⎢ 3 4 0 − 1⎥ ~ ⎢0
− 2 3 − 10⎥ ~
⎢ ⎥ ⎢ ⎥
⎢⎣− 1 0 −2 7 ⎥⎦ ⎢⎣02 − 3 10 ⎥⎦
⎡1 0 2 − 7 ⎤ ⎡1 0 2 − 7⎤
~ ⎢0 − 2 3 − 10⎥ ~ ⎢0
1 − 3 5 ⎥ = B.
⎢ ⎥ ⎢ 2 ⎥
⎢⎣0 0 0 0 ⎥⎦ ⎢⎣00 0 0 ⎥⎦
1 0
Since all 3-square minors of B are zero, while ≠ 0, the rank
0 1
of B is 2; hence the rank of A is 2. This procedure of obtaining
from A an equivalent matrix B from which the rank is evident by
inspections to be compared with that of computing the various
minors of A. The equivalent matrix of B obtained is not unique. In
particular, since only row operations were used, the student may
obtain others by using only column operations. When the elements
are rational numbers, there generally is no gain in mixing row and
column operations.
The normal form of a matrix:
By means of elementary operations, any matrix A of rank r > 0
can be reduced to one of the forms

called its normal canonical form. A zero matrix is its own normal
form.
Basic results:
1) Let the rank of the matrix A , is r,

96
Algebra (Chapter 5) Matrices
⎡ a11 a12 ... a1n ⎤
⎢a a 22 ... a 2n ⎥
A = ⎢ 21 ⎥ , m ≤ n.
⎢ ... ... ... ... ⎥
⎢ ⎥
⎣a m1 a m 2 ... a mn ⎦
Then associated with the m row vectors X1, X2...Xm , m > r, there
are exactly r vectors of the set which are linearly independent
while each of the remaining m–r vectors can be expressed as a
linear combination of these r vectors.
2) A necessary and sufficient condition that the vectors X1,
X2...Xm by linearly dependent is that the matrix A of the vectors
be of rank r < m. If the rank is m, the vectors are linearly
independent.
3) The set of vectors X1, X2...Xm is necessary linear dependent if
m > n.
4) If the set of vectors X1, X2...Xm is linearly independent so also
is every subset of them.
Example 5.7:
Examine the following set of vectors for linear dependence or
independence. Determine a maximum subset of linearly
independent vectors and express the others as linear combinations
of these: X = [1, 2, 4, 3], Y = [3, –1, 1, 2] and Z = [1, –5, –7, 8].
Solution:
Compute the rank of the matrix A by using only row elementary

97
Algebra (Chapter 5) Matrices

operations
⎡1 2 4 − 3⎤ ⎡1 2 4 − 3⎤ ⎡1 2 4 − 3⎤
1 2
A= ⎢3 − 1 1 2 ⎥ ~ ⎢0 − 7 − 11 11 ⎥ ~ ⎢0 − 7 −11 11⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣1 − 5 − 7 8 ⎥⎦ ⎢⎣0 − 7 − 11 11 ⎥⎦ ⎢⎣0 0 0 0 ⎥⎦
In step 1, we make operations R12 (–3), R13 (–1). In step 2, we
apply the operation R23 (–1).
Now, make the same elementary operations on the matrix.
⎡X ⎤ 1 ⎡ X ⎤ 2 ⎡ X ⎤
A = ⎢Y ⎥ ~ ⎢Y − 3X ⎥ ~ ⎢ Y − 3X ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣ Z ⎥⎦ ⎢⎣ Z − X ⎥⎦ ⎢⎣ Z − X − Y + 3X = 0⎥⎦

i.e. Z – Y + 2X = 0.
Here rank A = 2, there are two linearly independent vectors, say
X, Y. Then Z = Y – 2X.
Theorem 5.4:
Any non-singular matrix of order n can be obtained by performing
elementary row operations, or equally well, elementary column
operations, on In.
i.e. there are elementary matrices E1 ,E2 …Ek such that:
Ek Ek–1 …E2 E1 A = In .
It is easy to see that, the matrix A will be equal
–1 –1 –1 –1 –1
to A = (Ek Ek–1 …E2 E1 ) = E1 E2 … Ek–1 Ek ,
which is a product of elementary matrices.
–1
Therefore A = Ek Ek–1 …E2 E1.
i.e. If A is reduced to I by a sequence of row operations alone,

98
Algebra (Chapter 5) Matrices
–1
then A is equal to the product in reverse order of the
corresponding elementary matrices.
A systematic way to both keep track of and multiply the
elementary matrices is the following: Begin with the partitioned
matrix [A | In ] and perform elementary row operations on it until
the sub matrix on the left “A” is the identity matrix. The sub
–1
matrix on the right “ I ” will be the inverse sought A .
Example 5.8:
⎡1 0 0⎤
Find A if A = ⎢2 2 − 1⎥
–1
⎢ ⎥
⎢⎣1 − 1 1 ⎥⎦

Solution:
⎡1 0 0 1 0 0⎤ ⎡1 0 0 1 0 0⎤
⎢ ⎥ 1 ⎢0 2 − 1 − 2 1 0 ⎥ 2
⎢2 2 − 1 0 1 0⎥ → ⎢ ⎥→
⎢⎣1 − 1 1 0 0 1⎥⎦ ⎢⎣0 − 1 1 − 1 0 1⎥⎦
⎡1 0 0 1 0 0⎤ ⎡1 0 0 1 0 0⎤
⎢ ⎥ 4
2 ⎢0 − 1 −1 1 ⎥ 3 −1
⎥ → ⎢0 1 2 − 1 0⎥
1 0 1
→⎢ 2 2
⎢ 2
⎥→
⎢⎣ 0 − 1 1 − 1 0 1 ⎦⎥ 1 − 2 1
⎢⎣ 0 0 2 2
1⎥
⎦
⎡1 0 0 1 0 0⎤ ⎡1 0 0 1 0 0⎤
4 ⎢0 ⎢ –1
1 −1 − 1 1 0 ⎥ 5 ⎢ 0 1 0 − 3 1 1 ⎥ = [In | A ].
→⎢ 2 2 ⎥→ ⎥
⎢⎣ 0 0 1 − 4 1 2 ⎥⎦ ⎢⎣ 0 0 1 − 4 1 2 ⎥⎦

⎡ 1 0 0⎤
i.e. A = ⎢ − 3 1 1 ⎥ .
–1
⎢ ⎥
⎢⎣− 4 1 2⎥⎦

99
Algebra (Chapter 5) Matrices

§ 5.8: System of linear equations:

Let us consider the system (5.6) of m linear equations in the n
unknowns x1, x2…xn. By a solution of the system is meant any set
of values x′1, x′2,…, x′n which satisfy simultaneously the given m
equations. When the system has a solution, it is said to be
consistent ;otherwise, the system is said to be inconsistent.
A consistent system has either just one solution or infinitely many
solutions.
Two systems of linear equations in the same number of
unknown are called equivalent if every solution of either system is
a solution of the other. A system of equations equivalent to (5.6)
may be obtained from it by applying one or more of the following
operations:
i) Interchanging any two of the equations,
ii) Multiplying any equation by any nonzero constant,
iii) Adding to any equation a constant multiply of another
equation.
The system (5.6) may be written as (5.7) or (5.8),
A X = B, (5.8)
where A = [aij] is the coefficient matrix
t t
X = [x1, x2 …xn] , and B = [b1, b2 …bm ] .
The following matrix [A | B] is called the augmented matrix. To
solve the system (5.6), we proceed by elementary row operations
to replace A by row equivalent canonical matrix.
100
Algebra (Chapter 5) Matrices

Example 5.9:
Solve the system
x + 2y + z = 2 ,
3x + y – 2z = 1,
4x – 3y – z = 3 ,
2x + 4y + 2z = 4 .
Solution:
The augmented matrix is [A | B] =
⎡1 2 1 2⎤ ⎡1 2 1 2⎤
⎢ ⎢
⎢ 3 1 −2 1 ⎥ 1 ⎢0 − 5 −5 − 5⎥ 2
= ⎥ ~ ⎥~
⎢4 − 3 − 1 3⎥ ⎢0 − 11 −5 − 5⎥
⎢ ⎥ ⎢ ⎥
⎣2 4 2 4 ⎦ ⎣0 0 0 0⎦
⎡1 2 1 2 ⎤ ⎡1 0 − 1 0 ⎤
⎢ ⎢
2 0 1 1
⎢ 1 ⎥ 3 ⎢0 1 1 1 ⎥4
~ ⎥~ ⎥~
⎢0 11 5 5 ⎥ ⎢0 0 − 6 − 6⎥
⎢ ⎥ ⎢ ⎥
⎣0 0 0 0 ⎦ ⎣0 0 0 0 ⎦
⎡1 0 − 1 0⎤ ⎡1 0 0 1⎤
4 ⎢0 1 1 1 ⎥ 5 ⎢0 1 0 0⎥
~⎢ ⎥~⎢ ⎥.
⎢0 0 1 1 ⎥ ⎢0 0 1 1⎥
⎢ ⎥ ⎢ ⎥
⎣ 0 0 0 0 ⎦ ⎣0 0 0 0⎦
In step 1, we apply the operation R12(–3), R13 (–4), R14 (–2).
In step 2, we apply R2 (–1/5), R3(–1).
In step 3, applying R23 (–11), R21 (–2).
In step 4, R3 (–1/6) and in the last step 5, we apply the operation

101
Algebra (Chapter 5) Matrices

R31(1), R32 (–1). From the last matrix the equivalent system is
x = 1, y = 0 and z = 1.
Example 5.10:
Solve the system x + y = 2, 2x + 2y = 5.
Solution:
The augmented matrix is
⎡1 1 2⎤ 1 ⎡1 1 2⎤
[A | B] = ⎢ ⎥ ~ ⎢0 0 1 ⎥⎦
.
⎣ 2 2 5 ⎦ ⎣
This is equivalent to the two equations x + y = 2, 0x + 0y = 1
In the first equation: There is many values such as x = a, y = 2 –
a which satisfied for all values of a.
In the second equation: There is no x and y to satisfy the second
equation. The given system is inconsistent. It is easy to see that
the rank of the coefficient matrix = 1, while the rank of the
augmented matrix = 2. It is a very important remark, that in the
case of inconsistent system the rank of the coefficient matrix ≠ the
rank of the augmented matrix.
i) Solution of a set of homogeneous linear equation:
If A is an m×n matrix and X a column matrix, the matrix equation
A X = 0, is equivalent to a set of m homogeneous linear equations
in the n unknowns x1, x2…xn. X = 0, i.e. x1=x2=…=xn =0, clearly
satisfies the equations, whatever the matrix A. This solution is
called trivial. We seek nontrivial (i.e. nonzero) solutions these

102
Algebra (Chapter 5) Matrices

equations when they exist. The solution will not be unique. For,
suppose that x1, x2…xn satisfy the equations. Then, since the
equations are homogeneous, kx1, kx2…kxn also satisfy them, for
all values of k. i.e. If X is a solution, so is k X for all values of the
scalar k.
Example 5.11:
Solve the system:
x + y + 2z + 2t = 0 ,
x + 3y + 4z + 8t = 0 ,
2x + 2z – 2t = 0 .
Solution:

⎡1 1 2 2 ⎤ ⎡1
1
1 22⎤
2
⎡1 1 2 2⎤ 3
⎢1 3 4 8 ⎥ ~ ⎢0 2 2 6 ⎥~ ⎢0 2 2 6 ⎥ ~
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣2 0 2 − 2⎥⎦ ⎢⎣0 − 2 − 2 − 6⎥⎦ ⎢⎣0 0 0 0⎥⎦
⎡1 1 2 2⎤ ⎡1 0 1 − 1⎤
3 4
~ ⎢0 2 2 6 ⎥ ~ ⎢0 1 1 3⎥
⎢ ⎥ ⎢ ⎥
⎢⎣0 0 0 0⎥⎦ ⎢⎣0 0 0 0 ⎥⎦
In step 1, we apply the operations R12 (–1), R13 (–2);
and in step 2, R23 (1); in step 3; R2 (½) and in step 4, R21(–1).
Thus the given system is equivalent to the following system of
equations:
x + z – t = 0, ⎫
⎬ (1)
y + z + 3t = 0. ⎭

103
Algebra (Chapter 5) Matrices

Setting z = λ, t = µ , we get
x = –λ + µ , y = –λ – 3 µ , z=λ, t=µ.
In this case, all solutions of the given system are given by
assigning arbitrary values to λ, µ i.e. z , t. Once the values of z, t
have been assigned, the equations (1) determine x, y uniquely.
Fundamental sets of solutions: Take λ = 1, µ = 0,
t
we get x = –1, y = –1, z = 1, t = 0. i.e. X1 = [–1, –1, 1, 0] .
Take λ = 0, µ = 1, we get x = 1, y = –3, z = 0, t = 1
t
i.e. X2 = [1, –3, 0, 1] .
It is easy to see that X1, X2 are linearly independent and any
solution X can be written in the form X = λ X1 + µ X2.
That is to say, (a) every solution of (1) is a sum of multiples of the
particular solutions, X1, X2.
Moreover (b) no one X1 or X2 is a sum of multiples of the other. A
set of solutions that has the properties (a) and (b) is called a
fundamental set of solutions.
For the system of homogeneous equations A X = 0, X=0
that is x1 = x2 = ... = xn = 0 is always a solution; it is called the
trivial solution. If the rank of A is n, then the system has only the
trivial solution. If the rank of A is r < n, then the system has
exactly (n–r) linearly independent solutions such that every
solution is a linear combination of these (n–r) and every such
linear combination is a solution.

104
Algebra (Chapter 5) Matrices

(ii) Solution of non-homogeneous equations:

A system of m simultaneous algebraic equations in n unknowns,
A X = B, has a solution if and only if the coefficient matrix A and
the augmented matrix [A | B] have the same rank r.
When solutions exist, the number of arbitrary constants in
any complete solution is (n–r).
Let X1 , X2 be two distinct solutions of A X = B.
Then A X1 = B, A X2 = B and A(X1 – X2 ) = 0. Thus Y = X1 –X2
is nontrivial solution of A X = 0.
Conversely, if Z is any nontrivial solution of A X = 0 and
if Xp is any particular solution of A X = B, then X = Xp + Z is
also a solution of A X = B.
Theorem 5.5:
If the system of non-homogeneous equations A X = B is
consistent, a complete solution of the system is given by the
complete solution of AX=0 plus any particular solution of AX= B.
Example 5.2:
In the system:
x – 2y + 3z = 4, ⎫
⎬ (1)
x + y + 2z = –5. ⎭
Set z = 0; then x = –2, y = –3.
t
A particular solution Xp = [–2, –3, 0] .
The complete solution of the homogeneous system

105
Algebra (Chapter 5) Matrices
t
x – 2y + 3z = 0, x + y + 2z = 0 is [–7a, a, 3a] , where a is
arbitrary.
Then the complete solution of the given system (1) is:
t t t
X = [–7a, a, 3a] + [2, 3, 0] = [–7a + 2 , a + 3 , 3a] .
The above procedure may be rearranged in:
⎡1 − 2 3 4 ⎤ 1 ⎡1 − 2 3 4⎤2
[A | B] = ⎢ ⎥ ⎢0 3 − 1 − 9 ⎥ ~
~
⎣1 1 2 − 5⎦ ⎣ ⎦
2 ⎡1 − 2 3 4 ⎤ ⎡1 0 7 / 3
3 − 2⎤
~⎢ ⎢
− 3⎥⎦ ⎣0 1 −1 / 3 − 3⎥⎦
~
⎣0 1 −1 / 3
It is easy to see that x + 7 z = –2, y – (– 1 )z = –3.
3 3
Set z = 3a, we get x = –2 – 7a, y = –3 + a , z = 3a.
Example 5.3:
Solve:
x – 2y + z + 3t = 3,
x – 2y – 2z – 6t = –3,
2x – 4y – 3z – 9t = –4.
Solution:
The augmented matrix is
⎡1 − 2 1 3 3 ⎤ ⎡1 − 2 1 3 3 ⎤
1 2
⎢ ⎥ ⎢
[A | B] = ⎢1 − 2 − 2 − 6 − 3 ~ ⎢0 0 − 3 − 9 −6⎥ ~
⎥ ⎥
⎢⎣2 − 4 − 3 − 9 − 4⎥⎦ ⎢⎣0 0 − 5 −15 −10⎥⎦
⎡1 − 2 1 3 3⎤ ⎡1 − 2 0 0 1⎤
2 3
⎢ ⎢
~ ⎢ 0 0 1 3 2 ⎥ ~ ⎢0 0 1 3 2⎥ .
⎥ ⎥
⎢⎣0 0 1 3 2⎥⎦ ⎢⎣0 0 0 0 0⎥⎦

106
Algebra (Chapter 5) Matrices

Then x – 2y = 1 and z + 3t = 2.
Take y = a and t = b, where a and b are arbitrary; the complete
solution may be given as:
x = 1 + 2a, y = a, z = 2 – 3b, t = b.
§ 5.9: Eigenvalues and eigenvectors:
Let A = [aij ] be an n × n matrix and X a column vector. The
equation A X = λ X, where λ is a number can be written as:
⎡ a11 a12 ... a1n ⎤ ⎡ x1 ⎤ ⎡ x1 ⎤
⎢a ⎥ ⎢
a 22 ... a 2n x 2 ⎥ ⎢x ⎥
⎢ 21 ⎥ ⎢ ⎥= λ⎢ 2⎥ (5.14)
⎢ ... ... ... ... ⎥ ⎢ ... ⎥ ⎢ ... ⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎣a n1 a n 2 ... a nn ⎦ ⎣ x n ⎦ ⎣x n ⎦
or (A – λ I) X = 0. (5.15)
(A – λ I) is an n × n matrix. Equation (5.14) has a nontrivial
solution, if and only if, the rank of A – λ I is less than n.
i.e. |A – λ I| = 0 (5.16)
Thus a solution X ≠ 0 exists if, and only if, λ satisfies the
equation:
a11 − λ a12 ... a1n
a 21 a 22 − λ ... a 2n
=0 (5.17)
... ... ... ...
a n1 a n2 ... a nn − λ

This is an equation of degree n in λ, called the characteristic

equation of A. Its roots are called eigenvalues (latent roots) of A.
Corresponding to each eigenvalue λ there is a nontrivial solution
X of the equation A X = λ X, known as eigenvector.
107
Algebra (Chapter 5) Matrices

If X satisfies the equation, any scalar multiple kX of X also

satisfies it.
t
Now, if | A – λI| = 0, then | (A – λI) | = 0, also.
t t t
Since (A – λI) = A –λI , so |A – λI| = 0.
t
Hence A, and A have the same eigenvalues.

Theorem 5.6: (Caley-Hamilton)

A matrix satisfies its own characteristic equation.
Example 5.4:
⎡− 4 − 5⎤
Given A= ⎢
3 ⎥⎦
, use the fact that A satisfies its own
⎣ 2
2 3
characteristic equation to compute A and A ; also, since A is
–1 –2
non-singular, to compute A and A . Find the eigenvalues and
eigenvectors of the matrix A.
Solution:
−4−λ −5
The characteristic equation of A is =0
2 3−λ
2
i.e. (4+ λ )(3 – λ ) – 10 = 0 or λ + λ – 2 = 0, ∴ (λ–1)(λ+2)=0.
The roots are clearly 1 and –2. Thus the eigenvalues of A are 1
and –2. Since A satisfies its own characteristic equation, then:
2
A + A – 2I = 0. (1)
2 ⎡2 0⎤ ⎡− 4 − 5⎤ ⎡ 6 5⎤
Then: A = 2I – A = ⎢ ⎥– ⎢ 2 ⎥= ⎢− 2 − 1⎥ (2)
⎣ 0 2 ⎦ ⎣ 3 ⎦ ⎣ ⎦
3 2
On pre-multiplying (1) by A, we get: A + A – 2A = 0.

108
Algebra (Chapter 5) Matrices
3 2 ⎡− 8 − 10⎤ ⎡ 6 5⎤ ⎡− 14 − 15⎤
Then: A = 2A – A = ⎢
6 ⎥⎦ ⎢⎣− 2 − 1⎥⎦ ⎢ 6 7 ⎥⎦
– = (3)
⎣4 ⎣
–1
On pre-multiplying (1) by A , we get:
–1 –1
A + I – 2A = 0. Then A = 1 [A + I] , (4)
2

–1 ⎡− 4 − 5⎤ 1 ⎡1 0⎤ 1 ⎡− 3 − 5⎤ ⎡− 3 − 5⎤
= 1⎢ ⎥ =⎢ 2 2 ⎥ (5)
3 ⎥⎦ 2 ⎢0 1 ⎥ = 2 ⎢ 2
and A +
2 2
⎣ ⎣ ⎦ ⎣ 4 ⎦ ⎢⎣ 1 2 ⎥⎦
–1
On pre-multiplying (4) by A , we get: ,
− 5 ⎤ ⎡⎢− 4 − 5⎤
1
–2 –1 ⎡1 0⎤ 1 ⎡− 3 4⎥ .
= 1 ( I+ A ) = 1 ⎢ ⎢ 2 2⎥ =
1⎥⎦ 2
A +
2 2 0
⎣ ⎣⎢ 1 2 ⎦⎥ ⎢ 1 3 ⎥
⎣ 2 2 ⎦
To find the eigenvector corresponding to eigenvalue 1, we have to
find a non-zero solution of the equation A X = X.
t
Let X = [ x , y ] , i.e. – 4x – 5y = x, 2x + 3y = y
i.e. –5x – 5y = 0, 2x + 2y = 0.
t
Both of these equations yields x = –y. Hence [1, –1] is an
eigenvector of A corresponding to the eigenvalue of 1. In the same
t
way we may find that [5, –2] is an eigenvector corresponding to
the eigenvalue of –2.
Let B = ⎢
⎡1 5 ⎤ –1 1 ⎡ − 2 − 5⎤ .
⎥ , then B =
3⎢ 1 1 ⎥⎦
⎣ − 1 − 2⎦ ⎣
⎡ − 4 − 5⎤ ⎡ 1 5 ⎤ ⎡ 1 − 10⎤
Therefore A B = ⎢ ⎥ ⎢ ⎥ = ⎢− 1 4 ⎥ ,
⎣ 2 3 ⎦⎣ − 1 − 2 ⎦ ⎣ ⎦
–1 ⎡− 2 − 5⎤ ⎡ 1 − 10⎤ ⎡1 0 ⎤
and B A B = 1 ⎢ ⎥ ⎢− 1 4 ⎥ = ⎢0 − 2⎥ = diag (1,–2).
3 1
⎣ 1 ⎦⎣ ⎦ ⎣ ⎦

109
Algebra (Chapter 5) Matrices

Example 5.15:
⎡ 0 1⎤ 2
Let A = ⎢ ⎥ . The characteristic polynomial of A is λ +1 = 0,
⎣− 1 0⎦
which has no real roots, thus A has no real eigenvalues.
§ 5.10: Reduction of a matrix to
Diagonal canonical form:
Let A be a non-singular n × n matrix.
Theorem 5.7:
If λ1, λ2…λk are distinct eigenvalues of a matrix A and if X1,X2
,…Xn are non-zero eigenvectors associated respectively with these
roots, the X’s are linearly independent.
Let B be the matrix whose columns X1,X2 ,…Xn are the n
linearly independent eigenvectors of A. Then B is non-singular so
–1
that B exists. Now the equations:
A Xi = λ i Xi , i = 1,2…n may be written
A B = B D, (5.18)
where D is the n × n diagonal matrix diag (λ1 , λ2 … λ n ).
–1
Multiplying equation (5.18) on the left by B , we have:
–1 –1
B AB=B BD=D (5.19)
Similarly, if C is the matrix with rows are eigenvectors of A,
–1
C A = D C and C A C = D. (5.20)
Thus, by means of transformations,
–1 –1
B AB=D=CAC , (5.21)

110
Algebra (Chapter 5) Matrices

the matrix A is reduced to diagonal form D.

This proof depends upon the fact that the eigenvectors are
linearly independent. Now this may be the case when the
characteristic equation has no multiple roots. If λ is an eigenvalue
of multiplicity k, it can be shown that there are not more than k
linearly independent eigenvectors corresponding to λ. Thus there
may be n linearly independent vectors, but this is not necessarily
true. If there are, then the matrix B whose columns are these
–1
vectors in non-singular, and as before B A B = D.
Example 5.16:
Determine the eigenvalue and associated eigenvectors, given: ,
⎡ 5 7 − 5⎤
A = ⎢0 4 − 1⎥ and transform it into diagonal form.
⎢ ⎥
⎢⎣2 8 − 3⎥⎦

Solution:
The characteristic equation: | A – λ I | = 0
5−λ 7 −5
i.e. 0 4−λ −1 =0,
2 8 −3−λ
3 2
λ – 6λ + 11λ – 6 = 0 or (λ – 1)(λ – 2)(λ – 3) = 0
Then the eigenvalues are λ = 1 , 2, 3.
t
a) When λ = 1, let X1 = [x1, x2, x3] , then equation (5.13)
becomes:
4x1 + 7x2 – 5x3 = 0,
111
Algebra (Chapter 5) Matrices

3x2 – x3 = 0,
2x1 + 8x2 – 4x3 = 0.
⎡ −2 ⎤
⎡4 7 − 5⎤ 1 ⎡1 4 − 2⎤ 2 ⎡1 4 − 2⎤ 3 ⎢1 0 3 ⎥
⎢ 0 3 − 1 ⎥ ~ ⎢ 0 3 − 1 ⎥ ~ ⎢ 0 3 − 1 ⎥ ~ ⎢0 1 −1 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ 3 ⎥
⎢⎣2 8 − 4⎥⎦ ⎢⎣4 7 − 5⎥⎦ ⎢⎣0 − 9 − 3⎥⎦ ⎢0 0 0⎥
⎣ ⎦
We find x1 = 2 x3 , x2 = 1 x3 , let x3 = 3t
3 3
then x1 = 2t , x2 = t , x3 = 3t. ,
t
or simply [2, 1, 3] since any eigenvector is scalar multiple of this
vector. Similarly, corresponding to λ = 2, we have the eigenvector
t
[1 , 1 , 2] and corresponding to λ = 3, we obtain the eigenvector
t
[–1, 1, 1] Let B be the matrix whose columns the eigenvectors,
⎡2 1 − 1⎤ ⎡− 1 − 3 2 ⎤
i.e. B = ⎢1 1 1 ⎥ , |B| = 1 , B = ⎢ 2 5 − 3⎥
–1
⎢ ⎥ ⎢ ⎥
⎢⎣3 2 1 ⎥⎦ ⎢⎣− 1 − 1 1 ⎥⎦
–1
and B A B = diag (1, 2, 3).
Example 5.17:
Determine the eigenvalues and associated eigenvectors, given: ,
⎡2 2 1⎤
A = ⎢1 3 1 ⎥ and transform it into diagonal form.
⎢ ⎥
⎢⎣1 2 2⎥⎦

Solution:
2−λ 2 1
The characteristic equation: 1 3−λ 1 = 0.
1 2 2−λ

112
Algebra (Chapter 5) Matrices
3 2
i.e. λ – 7λ + 11 λ – 5 = 0 , and the eigenvalues are
λ1 = 5, λ2 = 1 , λ3 = 1.
When λ = λ1 = 5, equation (5.3) becomes:
⎡− 3 2 1 ⎤ ⎡ x1 ⎤ ⎡ 0 ⎤
⎢ 1 − 2 1 ⎥ ⎢ x ⎥ = ⎢0⎥ =0
⎢ ⎥ ⎢ 2⎥ ⎢ ⎥
⎢⎣ 1 2 − 3⎥⎦ ⎢⎣ x 3 ⎥⎦ ⎢⎣0⎥⎦
⎡ 1 −2 1 ⎤ 1 ⎡1 − 2 1 ⎤ 2 ⎡1 − 2 1 ⎤ 3 ⎡1 0 − 1⎤
⎢− 3 2 1 ⎥~ ⎢0 − 4 4 ⎥ ~ ⎢0 1 − 1⎥ ~ ⎢0 1 − 1⎥ .
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣ 1 2 − 3⎥⎦ ⎢⎣0 4 − 4⎥⎦ ⎢⎣0 0 0 ⎥⎦ ⎢⎣0 0 0 ⎥⎦
Therefore x1 – x3 = 0, x2 – x3 = 0, i.e. x1 = x2 = x3.
t
Then the eigenvector is X1 = [1, 1, 1] .
When λ = λ2 = 1 , equation (5.13) becomes:
⎡1 2 1⎤ ⎡ y1 ⎤ ⎡0⎤ ⎡1 2 1⎤ 1 ⎡1 2 1⎤
⎢1 2 1⎥ ⎢ y ⎥ = ⎢0⎥ = 0 ∴ ⎢1 2 1⎥ ~ ⎢0 0 0⎥
⎢ ⎥ ⎢ 2⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣1 2 1⎥⎦ ⎢⎣ y 3 ⎥⎦ ⎢⎣0⎥⎦ ⎢⎣1 2 1⎥⎦ ⎢⎣0 0 0⎥⎦
Therefore y1 + 2y2 + y3 = 0 .
To choose two linearly independent solutions:
t
i) Set y2 = 1, y3 = 0 ; then y1 = –2 i.e. X2 =[–2, 1, 0] .
t
ii) Set y2 = 0, y3 = 1 ; then y1 = –1 i.e. X3 =[–1, 0, 1] .
t
It is easy to see that the three vectors, X1 = [1, 1, 1] ,
t t
X2 =[–2, 1, 0] , X3 =[–1, 0, 1] are linearly independent. Take:
⎡1 − 2 − 1⎤ ⎡1 2 1⎤
⎢
B= 1 1 0 , then B = 1 − 1 2 − 1⎥
⎥ –1 ⎢
⎢ ⎥ 4 ⎢ ⎥
⎢⎣1 0 1 ⎥⎦ ⎢⎣− 1 − 1 3 ⎥⎦

113
Algebra (Chapter 5) Matrices
⎡5 0 0 ⎤
and B A B = ⎢0 1 0⎥ .
–1
⎢ ⎥
⎢⎣0 0 1⎥⎦

Example 5.18:
⎡1 1⎤
Let A = ⎢ ⎥
⎣0 1⎦
The eigenvalues here are λl = 1 and λ2 = 1 .
t
Eigenvectors associated with λ1 and λ2 are vectors [x, y] such
⎡x ⎤ ⎡0 1 ⎤ ⎡ x ⎤
that: (A – I) ⎢ ⎥ = 0 i.e. ⎢0 0⎥ ⎢ y ⎥ = 0.
⎣ y⎦ ⎣ ⎦⎣ ⎦
t
i.e. y = 0. Therefore the eigenvectors are of the form [k, 0] , where
k is any real number. Since A does not have two linearly
independent eigenvectors, we conclude that A is not
diagonalizable. The following is a useful theorem because it
identifies a large class of matrices that can be diagonalised and we
omit the proof of it.
Theorem 5.8:
If A is a symmetric n × n matrix, then there exists an orthogonal
–1
matrix B, such that B A B = D, a diagonal matrix. The
eigenvalue of A lie on the main diagonal of D.
Example 5.19:
⎡0 1 1 ⎤
Let A = ⎢1 0 1⎥
⎢ ⎥
⎢⎣1 1 0⎥⎦

114
Algebra (Chapter 5) Matrices

The characteristic polynomial of A is

−λ 1 1
1 −λ 1 =0
1 1 −λ
3 2
i.e. λ – 3λ – 2 = 0 or ( λ + 1) (λ – 2) = 0.
So the eigenvalues are λl = –1, λ2 = –1 and λ3 =2.
That is, –1 is an eigenvalue whose multiplicity is 2. To find the
eigenvectors associated with λl , λ2, we solve the homogeneous
linear system (A + I) X = 0:
⎡1 1 1⎤ ⎡a ⎤ ⎡0⎤
⎢1 1 1⎥ ⎢b⎥ = ⎢0⎥ =0. i.e. a + b + c = 0.
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢⎣1 1 1⎥⎦ ⎢⎣ c ⎥⎦ ⎢⎣0⎥⎦
t
i) Set b = 1, c = 0, we get a = –1 then X1 = [–1, 1, 0] .
t
ii) To obtain X2 = [a, b, c] which satisfies also a + b + c =0 and
choose X2 such that X1 . X2 = 0. i.e. X1 ⊥ X2.
i.e. –a + b = 0 or a = b therefore c = –2a.
t
Set c = –2, a = b = 1, we get X2 = [1, 1, –2] .
To find the eigenvector associated with λ3 = 2, we solve
( A – 2I) X = 0.
⎡− 2 1 1 ⎤ ⎡α ⎤ ⎡0⎤
⎢ 1 −2 1 ⎥ ⎢β ⎥ = ⎢0⎥ =0
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣ 1 1 − 2⎥⎦ ⎢⎣ γ ⎥⎦ ⎢⎣0⎥⎦

We find that α = β = γ = k any constant (k = 1 say).

t
Therefore X3 = [1, 1, 1] . Take

115
Algebra (Chapter 5) Matrices
⎡− 1 1 1⎤ ⎡− 3 3 0 ⎤
B= ⎢1 1 1⎥ , then B = 1 ⎢ 1 1 − 2⎥ ,
–1
⎢ ⎥ 6 ⎢ ⎥
⎢⎣ 0 − 2 1⎥⎦ ⎢⎣ 2 2 2 ⎥⎦
⎡− 1 0 0 ⎤
and B A B = ⎢ 0 − 1 0 ⎥ .
–1
⎢ ⎥
⎢⎣ 0 0 − 2⎥⎦
The matrix B is not orthogonal matrix. Can you make it
orthogonal? It is very easy try it!
Remarks:
Some remarks about non-symmetric matrices are in order at this
point. The matrix A is diagonalizable if all the roots of its
characteristic polynomial are real and distinct. We also studied
examples of non-symmetric matrices with repeated eigenvalues
that were diagonalizable (see Example 5.17) and others that were
not diagonalizable (see example 5.18). There are some striking
differences between the symmetric and non-symmetric cases,
which we now summarize. If A is non-symmetric, then the roots
of its characteristic polynomial need not all be real numbers; if an
eigenvalue λ has multiplicity k, then the maximum number of
linearly independent eigenvectors may be less than k; if the roots
of the characteristic polynomial of A are all real, it is still possible
for A not to have n linearly independent eigenvectors (which
means that A can not be diagonalised); eigenvectors associated
with distinct eigenvalues need not be orthogonal. Thus in Example

116
Algebra (Chapter 5) Matrices

5.17, the eigenvectors X1 and X2 associated with the eigenvalues

λl = 5 and λ2 = 1 are not orthogonal.
We now turn to the general situation for a symmetric
matrix; even if A has eigenvalues whose multiplicities are greater
than one, it turns out that we can still diagonalised A. See
Theorem 5.8 and Example 5.19.

§ 5.11: Quadratic form:

A homogeneous algebraic expression of the second degree in three
real variables x, y, z, given by:
2 2 2
ϕ (x, y, z)=a x + b y + c z + 2 f y z + 2 g z x + 2 h x y (5.22)
is called a quadratic form in three variables.
t
i.e. ϕ (x, y, z) = X A X , where
⎡a h g⎤ ⎡x ⎤
A = ⎢h b f ⎥ , X= ⎢ y ⎥ , X = [ x, y, z].
t
⎢ ⎥ ⎢ ⎥
⎢⎣g f c ⎥⎦ ⎢⎣ z ⎥⎦

Since A is symmetric, we can find an orthogonal matrix B such

–1
that D = B AB is diagonal. Now let us apply the orthogonal
⎡ x *⎤
transformation X = B X , where X = ⎢ y *⎥ .
* *
⎢ ⎥
⎢⎣ z *⎥⎦
* –1 –1 t
Then X =B X and since B is orthogonal matrix, then B = B.
It is easy to see that:
t * t * *t t *
ϕ (x, y, z) = X A X = (B X ) A (B X ) = X B A B X =

117
Algebra (Chapter 5) Matrices
*t –1 * *t *
=X B ABX =X DX =
*2 *2 *2
= d1 x + d2 y + d3 z .
where D = diag (d1 , d2 , d3 ) .
The equation of a central conic, referred to its centre as
origin, is of the form
2 2
a x + 2h x y + b y = 1. (5.23)
This may be written in the form:
⎡a h ⎤ ⎡x ⎤
[x y] ⎢ ⎥ ⎢ ⎥=1
⎣h b ⎦ ⎣ y ⎦
⎡a h ⎤
The left-hand side is a quadratic form with matrix A = ⎢ ⎥,
⎣ h b ⎦
t
and the equation (5.23) may be written X A X = 1, where
t
X = [x , y] .
One problem of analytic geometry is that of reducing, by
translation and rotation of axes, the equation of a quadratic curve
(or surface) to simplest form. The main tasks are to locate the
centre and to determine the principle directions, i.e. the direction
of the axes after rotation.
Without attempting to justify the steps, we show here the role of
two matrices in such a reduction of the equation of a central conic
section (quadratic equation).

Example 5.20:
2 2
Trace the curve S(x, y) = x – 3x y + y + 10x – 10y + 21 = 0.

118
Algebra (Chapter 5) Matrices

Solution:
See Example 7.3, page 241; Analytic Geometry (plane and solid)
by the author.
2
a = b = 1, h = – 3 , h – a b = 9 – 1 = 5 > 0.
2 4 4
It is hyperbola. To find its centre:
Solving Sx = Sy = 0, we get(–2 , 2).
Set x = –2 + X, y = 2 +Y .
2 2
The equation of the hyperbola is X – 3XY + Y + 1 =0 (1)
⎡ 1 − 3 ⎤ ⎡X ⎤
i.e. [X , Y] ⎢ 3 2⎥
⎢ ⎥ = –1.
⎢− 1 ⎥ ⎣Y ⎦
⎣ 2 ⎦
The characteristic equation of A is | A – λ I| = 0.
⎡1 − λ − 3 ⎤
i.e. ⎢ − 3 2 ⎥ = 0 or ( 1 – λ )2 – 9 = 0 .
⎢ 1 − λ⎥ 4
⎣ 2 ⎦
2
i.e. 4 λ – 8 λ – 5 = 0 or (2λ – 5)(2 λ + 1) = 0 .
i.e. λ1 = 5 , λ2 = – 1 .
2 2
i) When λ1 = 5 :
2
⎡1 − 5 − 3 ⎤ ⎡a ⎤ ⎡0 ⎤
⎢ 2 2 ⎥
⎢b⎥ = ⎢0⎥ = 0 i.e. a + b = 0 .
⎢ − 3 1 − 5⎥
⎣ 2 2⎦ ⎣ ⎦ ⎣ ⎦
t
Therefore the eigenvector is [1, –1] , and also the unit vector
t
[ 1 ,– 1 ].
2 2
ii) When λ2 = – 1 :
2

119
Algebra (Chapter 5) Matrices
⎡1 + 1 −3 ⎤
⎢ 2 2 ⎥ ⎡c ⎤ = ⎡0⎤
⎢ ⎥ ⎢0⎥ = 0. i.e. c – d = 0.
⎢ − 3 1 + 1 ⎥ ⎣d ⎦ ⎣ ⎦
⎣ 2 2⎦
t
Therefore the eigenvector is [1 , 1] , and also the unit vector
t
[ 1 , 1 ] .
2 2
⎡ 1 1 ⎤t
Now form the orthogonal matrix B = ⎢ 21 2⎥
⎢− 1 ⎥
⎣ 2 2⎦

whose columns are the two vectors after normalization.

⎡X ⎤ ⎡X'⎤ ⎡X'⎤ t ⎡X ⎤
The transformation ⎢ ⎥ = B ⎢Y'⎥ and ⎢Y'⎥ =B ⎢Y ⎥ ,
⎣Y ⎦ ⎣ ⎦ ⎣ ⎦ ⎣ ⎦
X = 1 X′ + 1 Y′ , Y=– 1 X′ + 1 Y′
2 2 2 2
and X′ = 1 X – 1 Y , Y′ = 1 X + 1 Y.
2 2 2 2
⎡X'⎤ t
This transformation reduces (1) to : [X′ , Y′] B A B ⎢ ⎥ = –1.
⎣Y'⎦
2 2 Y'2 X'2
i.e. X′ – Y′ = –1,
5 1 – = 1.
2 2 2 2
5
The conic is a hyperbola, whose semi-axes are 2 and 2.
5

The equations of its axes are X′ = 0 and Y′ = 0.

i.e. X′ = 1 X – 1 Y = 0, or X – Y = 0, or x – y + 4 = 0,
2 2
Y′ = 1 X + 1 Y = 0, or X + Y = 0, or x + y = 0.
2 2

120
Algebra (Chapter 5) Matrices

EXERCISES 5
1) Express the following matrix as a sum of symmetric and skew-
⎡− 1 7 1⎤
symmetric matrix: ⎢ 5 0 5⎥ .
⎢ ⎥
⎢⎣ 2 3 4⎥⎦
⎡ 1 −1 1 ⎤ ⎡1 2 3⎤
2) Given A = ⎢ − 3 2 − 1⎥ and B = ⎢ 2 4 6⎥ ,
⎢ ⎥ ⎢ ⎥
⎢⎣− 2 1 0 ⎥⎦ ⎢⎣1 2 3⎥⎦

Compute AB and BA, hence deduce that AB ≠ BA generally.

3) Given
⎡1 − 3 2 ⎤ ⎡1 4 1 0 ⎤ ⎡ 2 1 − 1 − 2⎤
A= ⎢2 1 − 3⎥ , B = ⎢2 1 1 1 ⎥ ,C = ⎢3 − 2 − 1 − 1⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣4 − 3 − 1⎥⎦ ⎢⎣1 − 2 1 2⎥⎦ ⎢⎣2 − 5 − 1 0 ⎥⎦

Show that AB = AC. Thus AB = AC does not necessarily imply

B = C.
4) Explain why in general:
2 2 2 2 2
i) (A ± B) = A ± 2AB + B , ii) i) A – B ≠ (A–B)(A+B).
5) Show by means of an example that in matrices AB = 0 does not
necessarily mean that either A = 0 or B = 0 where 0 stands for null
matrix.
6) Given
⎡ 2 − 3 − 5⎤ ⎡− 1 3 5⎤ ⎡ 2 − 2 − 4⎤
A = ⎢− 1 4 5 ⎥,B= ⎢ 1 − 3 − 5⎥ , C= ⎢− 1 3 4⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣ 1 − 3 − 4⎥⎦ ⎢⎣− 1 3 5 ⎥⎦ ⎢⎣ 1 − 2 − 3⎥⎦

a) Show that AB = BA = 0 , AC = A , CA = C.

121
Algebra (Chapter 5) Matrices

b) Use the result of (a) to show that A C B = C B A.

2 2 2 2 2
A – B = (A – B)(A + B), (A ± B) = A + B .
7) Compute A B given: (Using indicated partitioning)

⎡0 0 4 1⎤
⎡ 2 1 0⎤ ⎢0 0
⎡ − 2 6⎤ 2 0⎥
[Answer: a) ⎢1 2 0⎥ , b) ⎢
⎢− 2 5⎥ , c) ⎢0 1
⎥]
⎢ ⎥ ⎣ ⎦ 0 0⎥
⎢⎣1 1 0⎥⎦ ⎢ ⎥
⎣2 2 0 0⎦
⎡ 2 − 2 − 4⎤ ⎡1 1 3⎤
8) Given A = ⎢− 1 3 4 ⎥,B= ⎢ 5 2 6 ⎥.
⎢ ⎥ ⎢ ⎥
⎢⎣ 1 − 2 − 3⎥⎦ ⎢⎣− 2 − 1 − 3⎥⎦
2
Prove that: i) A = A, A is called idempotent.
2
ii) B = 0, B is called nilpotent of order (index) 3.
k+1
9) A matrix A for which A = A, where k is a positive integer is
k+1
called periodic. If k is the least positive integer for which A =
= A, then A is said to be of period k. Show that:

122
Algebra (Chapter 5) Matrices
⎡ 1 − 2 − 6⎤
i) A = ⎢− 3 2 9 ⎥ is periodic of period 2.
⎢ ⎥
⎢⎣ 2 0 − 3⎥⎦
2
Is that means A = I? Why?
⎡3 − 3 4⎤
ii) A = ⎢2 − 3 4⎥ is periodic of period 4.
⎢ ⎥
⎢⎣0 − 1 1 ⎥⎦
4
Is that means A = I? Why?
⎡1 2 ⎤ ⎡2 2 ⎤ ⎡1 2⎤
10) Let A = ⎢ ⎥, B= ⎢3 − 1⎥ , C= ⎢ ⎥.
⎣ 4 − 3⎦ ⎣ ⎦ ⎣ 0 1 ⎦
2
i) Compute A + 2A – 11 I,
2
ii) Compute B – B – 8 I,
n
iii) Compute C .
11) Without expanding show that:
1 a a2
1 b b 2 = (a – b)(b – c)(c – a).
1 c c2
a b 0 0
c d 0 0 a b e f
12) Prove that: = .
0 0 e f c d g h
0 0 g h

13) Without evaluating, show that:

1 a a 2 a3 bcd 1 a a 2
1 b b2 b3 acd 1 b b 2
= =
1 c c2 c3 abd 1 c c2
1 d d2 d3 abc 1 d d 2

123
Algebra (Chapter 5) Matrices

= (a – b)(a – c)(a – d)(b – c)(b – d)(c – d).

14) Show that the adjoint of
⎡ − 1 − 2 − 2⎤
A= ⎢2 1 − 2⎥ , is 3A and the adjoint of B where :
t
⎢ ⎥
⎢⎣ 2 − 2 1 ⎥⎦
⎡− 4 − 3 − 3⎤
B= ⎢ 1 0 1 ⎥ , is B itself.
⎢ ⎥
⎢⎣ 4 1 3 ⎥⎦
15) Find the inverse of each of the following matrices:
⎡1 3 3⎤ ⎡ 7 − 3 − 3⎤ ⎡2 3 1 ⎤
A = ⎢1 4 3⎥ , B = ⎢− 1 1 0 ⎥ , C = ⎢1 2 3 ⎥ ,
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣1 3 4⎥⎦ ⎢⎣− 1 0 1 ⎥⎦ ⎢⎣3 1 2⎥⎦
⎡2 4 3 2⎤
⎡ 1 − 5 − 7⎤ ⎢3 6 5 2⎥
D= 1 ⎢ 7 1 − 5⎥ , E = ⎢ ⎥.
18 ⎢ ⎥ ⎢2 5 2 − 3⎥
⎢⎣− 5 7 1 ⎥⎦ ⎢ ⎥
⎣4 5 14 14 ⎦
16) Solve the system:
⎡1 1 1 ⎤ ⎡ x u ⎤ ⎡ 9 2 ⎤
⎢2 5 7 ⎥ ⎢ y v ⎥ = ⎢52 15 ⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢⎣2 1 − 1⎥⎦ ⎢⎣ z w ⎥⎦ ⎢⎣ 0 − 1⎥⎦

[Answers: x = 1, y = 3 z = 5; u = –1, v = 2, w = 1].

17) If A is skew-symmetric matrix and I + A is non-singular then
–1
B = (I – A)(I + A) is orthogonal matrix.

18) Use problem (17) to obtain the orthogonal matrix B given

124
Algebra (Chapter 5) Matrices
⎡ 0 1 2⎤
⎡ 0 5⎤
a) A = ⎢ ⎥, b) A = ⎢ − 1 0 3⎥
⎣ − 5 0⎦ ⎢ ⎥
⎢⎣− 2 − 3 0⎥⎦
⎡ 5 − 14 2 ⎤
1 ⎡− 12 − 5 ⎤ 1⎢
[Answers: a) ⎢ ⎥ , b) − 10 − 5 − 10⎥ .]
13 ⎣ 5 − 12⎦ 15 ⎢ ⎥
⎢⎣ 10 2 − 11⎥⎦
19) If A is a square matrix prove that:
–1 –1
i) A(I + A) = (I + A)A, A (I + A) = (I + A)A ,
2
ii) (I + A)(I – A) = (I – A)(I + A) = I – A ,
–1 –1
iii) A(I + A) = (I + A) A, where (I + A) is non singular,
t –1 –1 t
iv) A (I + A) = (I +A) A , where (I +A) is non-singular
and A is orthogonal matrix.
20) Prove: If A is orthogonal and I + A is non-singular then
–1
B =(I – A)(I + A) , is skew-symmetric.
21) Show that each triple of vectors:
a) X1 = [1, 2, 3, 4], X2 = [3, –1, 2, 1 ], X3 = [1, –5, 8, –7] ,
b) X1 = [2, 3, 1, –1], X2 = [2, 3, 1, –2] , X3 = [4, 6, 2, –3] .
is linearly dependent. In each determine a maximum subset of
linearly independent vectors and express the others as linear
combinations of these.
[Answers: a) X3 = –2X1+X2 ; X3 = X1 + X2. ]
22) Examine each of the following sets of vectors for linear
dependence or independence. In each dependent set select a
maximum linearly independent subset and express each of the

125
Algebra (Chapter 5) Matrices

remaining vectors as a linear combination of these.

a) X1 = [2, –1, 3, 2] X2 = [1, 3, 4, 2], X3 = [3, –5, 2, 2].
b) X1 = [1, 2, 1], X2 = [2, 1, 4] , X3=[4, 5, 6], X4 = [1, 8, 3].
c) X1= [2, 1, 3, 2, –1], X2 = [4, 2, 1, –2, 3],
X3 = [0, 0, 5, 6, –5], X4 = [6, 3, –1, –6, 7].
[Answers: a) X3 =2X1 – X2 , b) X3 = 2X1 +X2 , X4 = 5X1–2X2 ;
c) X3 =2X1 –X2, X4 = 2X2 – X1]
23) If the polynomials of either system are linearly dependent find
a linear combination, which is identically zero.
3 2 2 3 2
a) P1 = x – 3x + 4x–2, P2 = 2x – 6x +4 , P3 = x –2x + x.
4 3 3 2
b) Q1= 2x + 3x – 4x + 5x + 3, Q2 = x + 2x – 3x + 1,
4 3 2
Q3 = x + 2x – x + x + 2 .
[Answers: a) 2P1 + P2 – 2P3 = 0, b) Q1+ Q2 – 2Q3 = 0]
24) Show that:
⎡1 2 3⎤ ⎡ 2 1 3⎤ ⎡0 3 3 ⎤
A = ⎢3 2 4⎥ , B = ⎢3 4 2⎥ , and C = ⎢3 0 6⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣1 3 2⎥⎦ ⎢⎣2 2 1 ⎥⎦ ⎢⎣0 4 3⎥⎦

are linearly dependent.

25) Show that any 2×2 matrix can be written as a linear
combination of the matrices:

⎡1 0⎤ ⎡0 1 ⎤ ⎡0 0 ⎤ ⎡0 0 ⎤
A= ⎢ ⎥ , B = ⎢0 0 ⎥ , C = ⎢1 0⎥ , D = ⎢0 1 ⎥ .
⎣0 0 ⎦ ⎣ ⎦ ⎣ ⎦ ⎣ ⎦
26) Evaluate the rank of the following matrices:

126
Algebra (Chapter 5) Matrices
⎡1 3 1⎤ ⎡1 4 2 4 ⎤ ⎡1 1 3 3⎤
⎢2 4 0⎥ ⎢1 3 1 2 ⎥ ⎢0 2 2 4⎥
a) ⎢ ⎥ , b) ⎢ ⎥ , c) ⎢ ⎥
⎢2 4 0⎥ ⎢1 2 0 0 ⎥ ⎢1 0 2 1⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎣1 3 1⎦ ⎣0 − 1 − 1 − 2 ⎦ ⎣1 1 3 3⎦
[Answers: a) 2 , b) 2 , c) 2 ]
27) Find the inverse of each of the following matrices:
⎡1 0 − 1⎤ ⎡ 2 1 0⎤
⎡ 2 − 1⎤
A= ⎢ ⎥ , B = ⎢0 1 1 ⎥ , C = ⎢1 1 0 ⎥ .
⎣− 1 1 ⎦ ⎢ ⎥ ⎢ ⎥
⎢⎣1 − 1 1 ⎥⎦ ⎢⎣0 0 1⎥⎦
⎡1 1 2⎤ ⎡2 1 − 1⎤
⎡ ⎤
D = ⎢1 2 2⎥ , E = ⎢ , F = ⎢1 1 1 ⎥.
3 2
⎢ ⎥ ⎥ ⎢ ⎥
⎢⎣2 2 3⎥⎦ ⎣ − 2 4⎦ ⎢⎣3 2 1 ⎥⎦
–1 ⎡1 1 ⎤ –1 1 ⎡⎢ 2 1 1 ⎤
[Answers: A = ⎢ ⎥ , B =
3⎢ 1 2 − 1⎥ ,
⎣1 2⎦ ⎥
⎣⎢ − 1 1 1 ⎦⎥
⎡ 1 − 1 0⎤ ⎡− 2 −1 2⎤
–1⎢ ⎥
C = −1 2 0 , D = −1
–1 ⎢
1 0 ⎥,
⎢ ⎥ ⎢ ⎥
⎢⎣ 0 0 1⎥⎦ ⎢⎣ 2 0 − 1⎥⎦
⎡− 1 −3 2⎤
⎡ − ⎤
,F = ⎢2 − 3⎥ .
–1 4 2 –1
E = 1 ⎢ ⎥ 5
16 2
⎣ 3 ⎦ ⎢ ⎥
⎢⎣− 1 −1 1 ⎥⎦
28) Solve:
a) x+2y+z =2, 3x+ y–2z = 1, 4x – 3y – z = 3, 2x + 4y + 2 z = 4.
b) 2x + y + 5z + t = 5, c) 2x + 3y + z = 9,
x + y – 3z – 4t = –1, x + 2y + 3z = 6,
3x + 6y – 2z + t = 8, 3x + y + 2z = 8.
2x + 2y + 2 z – 3t = 2.

127
Algebra (Chapter 5) Matrices

d) x + 2y – 3z – 4t = 6, e) x + y + z + t = 0,
x + 3y + z – 2t = 4, x + 3y + 2z + 4t = 0,
2x + 5y – 2z – 5t = 10. 2x + z – t = 0.
[Answers: a)x = 1, y = 0, z = 1; b) x=2, y = 1 , z = 0, t = 4 ;
5 5
35 29 5
c) x = , y = , z = ; d) x = 10 + 11k , y = –2– 4k, z = k, t = 0;
18 18 18
e) x =– a + b , y = –a –3b, z = 2a , t = 2b.
29) Find all solutions of:
a) x – 2 y + z – 4 t = 1. b) x + y + z = 4,
2x + 5y – 2z = 3.
c) x + y + z = 4, d) x + y + z + t = 0,
2x + 5y – 2z = 3, x + y + z – t = 4,
x + 7y – 7z = 5. x + y – z + t = –4,
x – y + z + t = 2.
[Answers: a) x = 1+2a – b + 3c, y = a , z = b , t = c ;
b) x = –7a/3 + 17/3 , y = 4a/3 – 5/3 , z = a ; c) φ ;
d) x = – y = 1, z = –t = 2]
30) Find all nontrivial solutions of:
a) x – 2 y + 3z = 0, b) 2x – y + 3z = 0,
2x + 5 y + 6z = 0. 3x + 2y + z = 0,
x – 4y + 5z = 0.
c) x + 2 y + 3 z = 0, d) 4x – y + 2z + t = 0,
2x + y + 3 z = 0, 2x + 3 y – z – 2 t = 0,
3x + 2y + z = 0. 7 y – 4z – 5t = 0,
2x – 11 y + 7z + 8t = 0.

128
Algebra (Chapter 5) Matrices

[Answers: a) x = –3a, y = 0 , z = a , b) x = –y = –z = a,
c) only trivial solution, d) x =–5a+3b, y = 8a, z =14a–10b , t = 8b.]
31) For the following matrices determine the eigenvalues and the
corresponding eigenvectors:
⎡1 0 − 1⎤ ⎡1 2 2⎤ ⎡1 1 − 2⎤ ⎡0 1 0 ⎤
a) ⎢1 2 1 ⎥ , b) ⎢ 0 2 1 ⎥ ,c) ⎢− 1 2 1 ⎥ ,d)
⎢0 0 1 ⎥ ,
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣2 2 3 ⎥⎦ ⎢⎣− 1 2 2⎥⎦ ⎢⎣ 0 1 − 1 ⎥⎦ ⎢⎣1 − 3 3⎥⎦
⎡3 2 2 − 4⎤ ⎡1 − 4 − 1 − 4⎤
⎡2 1 1⎤ ⎢2 − 1⎥ ⎢2 5 − 4⎥
e) ⎢1 2 1⎥ , f)
3 2 0
⎢ ⎥ , g) ⎢ ⎥.
⎢ ⎥ ⎢1 1 2 − 1⎥ ⎢− 1 1 −2 3 ⎥
⎢⎣0 0 1⎥⎦ ⎢ ⎥ ⎢ ⎥
⎣2 2 2 − 1⎦ ⎣− 1 4 −1 6 ⎦
t t t
[Answers: a) 1, [1,–1, 0] ; 2, [2, –1, –2] ; 3 , [1, –1, –2] .
t t
b) 1, [1 , 1, –1] ; 2 , [2, 0, 1] .
t t t t
c) –1, [1, 0, 1] ; 2, [1, 3, 1] ; 1 , [3, 2, 1] . d) 1 , [1, 1, 1] .
2 t t t
e) 1 , [1,0 , –1] , [0, 1, –1] ; 3 , [1, 1, 0] .
2 t t t
f) 1 , [1, 0 , –1, 0] , [1, –1, 0, 0] ; 2 , [–2, 4, 1, 2] , 3 , [0, 3, 1, 2].
3 t t
g) 1 , [3, 6 , –4, –5] ; 2 , [2, 3, –2, –3]
32) Find the characteristic equation of the matrix:
⎡ 2 −1 1 ⎤
A = ⎢− 1 2 − 1⎥ and hence obtain the inverse of it A.
⎢ ⎥
⎢⎣ 1 − 1 2 ⎥⎦
3 2 –1 2
[Answers: A – 6A + 9A – 4I = 0; 4A = A – 6A + 9I,
⎡3 1 − 1⎤
–1 1 ⎢
A = 1 3 1 ⎥ .]
4 ⎢ ⎥
⎢⎣− 1 1 3 ⎥⎦

129
Algebra (Chapter 5) Matrices
–1
33) Find an orthogonal matrix B such that B A B is diagonal and
has as diagonal elements the characteristic roots of A given:
⎡ 7 −2 1 ⎤ ⎡ 2 0 − 1⎤
a) A= ⎢− 2 10 − 2⎥ , b) A = ⎢ 0 2 0 ⎥
⎢ ⎥ ⎢ ⎥
⎢⎣ 1 − 2 7 ⎥⎦ ⎢⎣− 1 0 2 ⎥⎦
⎡ 2 −4 2 ⎤ ⎡ 3 2 2⎤
c) A = ⎢− 4 2 − 2⎥ , d) A = ⎢2 2 0⎥ .
⎢ ⎥ ⎢ ⎥
⎢⎣ 2 − 2 − 1⎥⎦ ⎢⎣2 0 4⎥⎦
⎡ 1 1 1 ⎤
⎢ 2 3 6⎥
[Answers: a) B = 0 ⎢ 1 − 2 ⎥ , B–1AB= diag(6, 6, 12).
⎢ 3 6⎥
⎢ −1 1 1 ⎥
⎢⎣ 2 3 6 ⎥⎦
⎡ 1 0 1 ⎤
⎢ 2 2⎥ –1
b) B = ⎢ 0 1 0 ⎥ , B AB= diag(1, 2, 3).
⎢ 1 −1 ⎥
⎢⎣ 2 0 2 ⎥⎦
⎡2 1 2⎤
⎢ 3 3 3 ⎥ –1
c) B = ⎢ − 2 2 1 ⎥ , B AB= diag(7, –2, –2).
⎢ 13 32 −32 ⎥
⎢⎣ 3 3 3 ⎥⎦
⎡2 1 2⎤
⎢3 3 3⎥
d) B = ⎢ − 2 2 1 –1
⎥ , B AB= diag(0, 3, 6).
⎢ −31 −32 23 ⎥
⎢⎣ 3 3 3 ⎥⎦

34) Write the quadratic form

5x 12 – 2x 22 – 3x 32 + 12x1x2 – 8 x1x3 + 20 x2x3

in the matrix form. Find the eigenvalues and eigenvectors of A.

130
Algebra (Chapter 5) Matrices

Find the matrix B and the transformation equations X = BU when

t
U=[u1, u2, u3] so that the given quadratic form is reduced to
canonical form. Write the new quadratic form.
t t t
[Answers: 6, 9, –15; [–2, 1, 2] , [2, 2, 1] , [1, –2, 2] ,
x1 = 1 (–2u1+ 2u2+ u3), x2 = 1 (u1+ 2u2–2 u3),
3 3
x3 = 1 (2u1+ u2+ 2u3), 6u 12 + 9u 22 – 15u 32 .]
3
35) Given the quadratic form
2x 12 + 4x 22 + 5x 32 – 4x1x2

Answer all questions in problem (34) with respect to that

quadratic form.
t t t
[Answers: 1, 4, 6; [2, 0, 1] , [0, 1, 0] , [1, 0, –2] ,
x1 = 1 ( 2u1+ u2), x2 = u2 , x3 = 1 ( u1 – 2 u3), u 12 + 4u 22 + 6u 32 .]
5 5

36) a) Verify that the eigenvalues of:

⎡ 2 1 − 1⎤
A = ⎢− 1 0 1 ⎥ , are 1, 1, 2.
⎢ ⎥
⎢⎣− 1 1 2 ⎥⎦
t
b) Show that eigenvector corresponding to λ = 2 is [1, –1, 1] and
that there are two linearly independent6leigenvectors namely
t t
[1, 0, 1] , [0, 1, 1] , corresponding to the single eigenvalue λ = 1.
c) Can Theorem 5.7 and formula 5.19 be used to transform the
matrix A to diagonal form? Justify your answer.

131
Algebra (Chapter 6) Theory of equations

CHAPTER 6
Theory Of Equations
§ 6.1 Some Elementary Properties:
n n–1 n–2
Let f(x) ≡ a0x + a1x + a2x + … + an–1x + an denote a
th
polynomial of the n degree in x in which the coefficients a0, a1,
a2,…, an–1, an are all real, a0 > 0; an ≠ 0. Suppose further that
f(αi) = 0, i = 1, 2,…,n. Then α1, α2,…, αn are the roots of f(x) = 0.
The quantities αi may not be real and need not all be
r
distinct. Thus if f(x) = (x–α) ϕ(x), where ϕ(α) ≠ 0 then x = α is
an r-multiple root of f(x) = 0. In particular when r = 2, f (x) = 0 is
said to have a double root at x = α.
Equations with given Roots
There are two points of view. The first is “if given the values of
the roots, then we can construct the equation, in which each root
satisfy it.” The second is “given the equation and it is require
solving it. i.e. finding its roots.”
It is easy to construct an equation, which has roots of given
values; the degree of the equation equals the number of roots,
repeated roots being so counted.
Example 6.1:
Construct the equation whose roots are
i) –1, 5/2 , 3;
(ii) – 1, 5/2, and 2 roots equal to 3.

132
Algebra (Chapter 6) Theory of equations

(i) The equation may be written

(x + 1)(2x –5)(x – 3) = 0.
When multiplied out, this becomes an equation of degree 3 in x.
(ii) The equation may be written
2
(x + 1)(2x –5)(x – 3) = 0.
When multiplied out, this becomes an equation of degree 4 in x.
Example 6.2:
Construct the equation whose roots are
(i) α , β ; the equation may be written, (x – α)(x – β) = 0.
2
When multiplied out, this becomes x – (α + β) x + αβ = 0.
ii) Similarly, if α , β, γ are the roots, then the equation may be
written (x – α)(x – β) (x – γ) = 0.
When multiplied out, this becomes
3 2
x – (α + β + γ) x + ( αβ + αγ + βγ)x – αβγ = 0,
and this is expressed more conveniently in the form,
3 2
x – x ∑ α + x ∑ αβ – α β γ = 0
iii) Similarly, if α , β, γ , δ are the roots, then the equation may be
written (x – α)(x – β) (x – γ) ( x – δ) = 0. When multiplied out,
this becomes
4 3 2
x – (α + β + γ + δ) x + ( αβ + αγ + αδ + βγ + βδ + γδ)x –
– (αβγ + αβδ + αγδ + βγδ) x + αβγδ = 0,
and this is expressed more conveniently in the form,
4 3 2
x – x ∑ α + x ∑ αβ – x ∑ αβγ + αβγδ= 0

133
Algebra (Chapter 6) Theory of equations

This example shows the relations that connect the roots of

quadratic, cubic, and quartic equations with the coefficients, and
suggests the extension to equations of any degree.
§ 6.2 Symmetric Relations between Roots
And Coefficients:
If x = α is a root of the equation
n n–1 n–2
f(x) = a0x + a1x + a2x + … + an–1x + an = 0
then f(α) = 0, and therefore, by the remainder theorem, x – α is a
factor of f(x).
Hence, if α1, α2,…, αn are n unequal roots of this equation, it
follows that x – α1, x – α2,…, x – αn are each factors of f(x).
∴ f(x) = A (x – α1)(x – α2)… (x – αn), where A is independent of
n
x since f(x) is of degree n in x; by equating coefficients of x , it
follows that A = a0.
∴ a0 (x – α1)(x – α2)… (x – αn) =
n n–1 n–2
= a0x + a1x + a2x + … + an–1x + an
But the expansion of the left side of this identity is
n n–1 n–2 n
a0[ x – x ∑ (α1 ) + x ∑ (α1α 2 ) + … + (–1) α1α2α3…αn]
Equating coefficients, we have
∑ (α1 ) = – a1/a0 ,
∑ (α1α 2 ) = a2/a0 ,
∑ (α1α 2 α 3 ) = – a3/a0 ,
…………
n
α1 α2 α3 … αn = (–1) an/a0.

134
Algebra (Chapter 6) Theory of equations

These relations express the sums of the products of the roots,

taken 1, 2, 3,…, n at a time, in terms of the coefficients. It is easy
to see that they are symmetric relations in the roots α1, α2,…, αn.
The general relations, established above, remain true if any of the
roots α1, α2,…, αn are equal to one another.
Example 6.3:
If α , β are the roots of the quadratic equation,
2
a x + b x + c = 0,
2 2
x – (α + β) x + αβ = 0,
We have α + β = –b/a, αβ = c/a.
The symmetrical relations between the roots of an equation and
the coefficients do not by themselves provide a means of solving
the equation; e.g. the substitution by the value of β from the first
2
relation in the second gives α(–b/a – α) = c/a, i.e. aα + b α+ c = 0,
2
merely leads back to the original equation, aα + b α+ c = 0, with
α instead of x.
Now, it is easy to see that
2 2 2 2 2
(α – β) = α + β – 2αβ = α + β + 2αβ – 4αβ =
2 2 b 2 − 4ac
= ( α + β) – 4 αβ = (–b/a) – 4 c/a= .
a2
b 2 − 4ac b 2 − 4ac
i.e. |α – β| = . Let α ≥ β , then α – β = .
a a
− b + b 2 − 4ac − b − b 2 − 4ac
Thus we have α = ,β= .
2a 2a

135
Algebra (Chapter 6) Theory of equations
th
In fact the roots of the general equation of the n degree
cannot be expressed by elementary functions of the coefficients if
n > 4. Of course, there is a great difference between “the roots
cannot be expressed by elementary functions of the coefficients
if n > 4”, and the existence of the roots. e.g. the roots of the
following equation
5 4 3 2
x + 3 x – 11x – 15 x + 46 x – 24 = 0 are – 4 , – 3 , 1 , 1 , 2 .
But if some additional unsymmetrical relation is known to exist
between two or more of the roots, it may be possible to solve the
equation or to find some of its roots. By means of these relations,
any symmetric function of the roots of an equation can be
expressed in terms of the coefficients.
Example 6.4:
3
If α, β, γ are the roots of the cubic equation, a x + c x + d = 0.
2 2 2 3 3 3
Find the values of (i) α + β + γ ; (ii) α + β + γ ;
4 4 4 5 5 5 1 1 1 1 1 1
iii) α + β + γ ; iv) α + β + γ ; v) + + ; vi) + + .
α β γ α 2 β2 γ 2
Solution:
We have α + β + γ = 0 , αβ + αγ + βγ = c/a , α β γ = – d/a.
2 2 2 2
i) ∴ α + β + γ = (α + β + γ) – 2 (αβ + αγ + βγ) =
= –2c/a .
3 3
ii) Since a α + c α + d = 0, then a α = – c α – d.
3 3
Similarly, a β =–cβ –d,aγ = – c γ – d.

136
Algebra (Chapter 6) Theory of equations

Adding, we get
3 3 3
a(α + β + γ ) = – c(α + β + γ) – 3 d = – 3d.
3 3 3
∴ α + β + γ = – 3d/a .
3 3
iii) Since a α + c α + d = 0, then a α = – c α – d.
4 2
Multiplying by α, we get a α = – c α – d α.
4 2 4 2
Similarly, a β = – cβ – d β, a γ = – cγ – d γ.
4 4 4 2 2 2
Adding, we get a(α + β + γ ) = –c(α + β + γ ) – d (α + β + γ) .
4 4 4 2 2 2 2 2
∴ α + β + γ = – c(α + β + γ )/a = 2 c /a .
3 2
iv) Since a α = – c α – d, then multiplying by α , we get
5 3 2
aα =–cα –dα .
5 3 2 5 3 2
Similarly, a β = – c β – d β , a γ = – c γ – d γ .
Adding, we get
5 5 5 3 3 3 2 2 2
a(α + β + γ ) = – c(α + β + γ ) – d (α + β + γ ).
5 5 5 2
∴ (α + β + γ ) = 5 c d/a .
1 1 1 βγ + αγ + αβ ca c
v) + + = =– =–
α β γ αβγ ad d
1 1 1 (βγ ) 2 + (αγ ) 2 + (αβ) 2
vi) + + =
α2 β2 γ2 (α β γ ) 2
2 2 2 2
Since (αβ) + (αγ) + (βγ) =(αβ + αγ + βγ) – 2 α β γ(α + β + γ)=
2 2
= c /a , then
1 1 1 (βγ ) 2 + (αγ ) 2 + (αβ) 2 c 2 a 2 c 2
+ + = = = .
α 2 β2 γ 2 (α β γ ) 2 a 2 d2 d2

Example 6.5:
3 2
Solve x + 3 x – 4 = 0, given that it has two equal roots.

137
Algebra (Chapter 6) Theory of equations

Solution:
Suppose the roots are α , α , β.
2 2
Then 2α + β = –3, α + 2αβ = 0, α β = 4.
2 2
∴ β = –3 – 2α, α + 2αβ = α + 2α( –3 – 2 α) = 0 ;
2
and this reduces to α + 2 α = 0 or α (α + 2) = 0
∴ α = 0 or α = – 2 . Hence α = 0 , β = –3 or α = –2 , β = 1·
2
Also α β = 4 is satisfied by the second pair of values of α and β,
but not by the first. Therefore the roots are –2 , –2 , 1.
Example 6.6:
3 2
Solve x – 3x – 6x + 8 = 0, given that the roots are in arithmetical
progression.
Solution:
Suppose the roots are α – β , α , α + β.
Then (α – β) + α + (α + β) = 3, ∴ 3 α = 3, i.e. α = 1.
(α – β) α + (α + β)(α – β) + α (α + β) = – 6
2 2 2 2
∴ 3 α – β = – 6. Hence β = 3α + 6 = 9, i.e. β = ± 3.
Therefore the roots are – 2 , 1 , 4.
Example 6.7:
3 2
Solve 2x – 21 x + 63 x – 54 = 0, given that the roots are in
geometrical progression
Solution:
Suppose the roots are α/β , α , α β .

138
Algebra (Chapter 6) Theory of equations
2
Then α/β + α + α β = 21/2, α ( β + 1/β + 1) = 63/2 ,
3
α = 54/2 = 27 ∴ α = 3.
α/β + α + α β = 21/2 ∴ 1/β + 1 + β = 7/2 i.e. 1/β + β = 5/2
2
Thus 2 β – 5 β + 2 = 0 ∴(2β – 1)(β – 2) = 0 or β = 2 or β = ½ .
Therefore the roots are 3/2, 3, 6.
Example 6.8:
4 3 2
Solve 6x – 3x + 8 x – x + 2 = 0, having given that it has a pair
of roots whose sum is zero.
Solution:
Let the roots are α, β , γ , δ , then
α + β + γ + δ = ½,
αβ + αγ + αδ + βγ + βδ + γδ = 8/6 = 4/3,
αβγ + αβδ + αγδ + βγδ = 1/6,
αβγδ = 2/6= 1/3.
We may write α + β = 0 i.e. β = – α , substituting we get
γ+δ=½, (1)
2
– α + αγ + αδ – αγ – αδ + γδ = 4/3,
2
i.e. – α + γδ = 4/3, (2)
2 2
–α γ – α δ + αγδ – αγδ = 1/6,
2
i.e. – α ( γ + δ) = 1/6, (3)
2
–α γ δ = 1/3. (4)
2
From (1) and (3) α = –1/3, α = ± i .
3

From (4) γ δ = 1. (5)

139
Algebra (Chapter 6) Theory of equations
2
From (1) and (5) γ + 1/γ = ½ ∴ 2 γ – γ + 2 = 0 (6)
1 ± 1 − 16 1
∴ γ= = ( 1 ± i 15 )
4 4
Thus the four roots are ± i , 1 ( 1 ± i 15 ).
3 4
§ 6.3 Evaluation of a Polynomial.
It is very important to make a facility in calculating the
value of f(x) for special values of x. How, for example, can we
3 2
best evaluate f(4) when f(x) = 2x – 3x + 4x – 5? Instead of direct
substitution, which is often troublesome and liable to errors, the
following process, introduced in 1819 by Horner, is to be
recommended, particularly as it furnishes the technique required
also for the actual numerical solution of the equation f(x) = 0. The
work is arranged as follows:
3 2
f(x) = 2x – 3x + 4x – 5.
2 – 3 4 – 5 (5
10 35 195
7 39 190 f(5) = 190
The coefficients 2, –3, 4, –5 with their proper signs are
placed in a row according to descending powers of x. Missing
powers must be indicated by zero coefficients. Thus we have four
columns (n + l in general). The argument 5 of the desired f(5) is
then entered, as if it were the quotient of a long division sum,
beyond the final column. Beginning at the left the leading entry 2
is now multiplied by 5, the product, 10, being written below the
next entry on the right and added. The result 7 is again multiplied

140
Algebra (Chapter 6) Theory of equations

by 5 and the product, 35, is added to the coefficient in the third

column, giving 39. This again is multiplied by 5 and the product,
195, is added to the coefficient in the fourth, the last column. The
process is then complete and the result is f(5) = 190.
3 2
f(x) = 2x – 3x + 4x – 5.
2 –3 4 –5 (5
10 35 195
7 39 190 f(5) = 190
10 85
17 124 f ′(5)= 124
The explanation is simple: and it is only necessary to carry out the
same procedure with letters to convince ourselves of its
correctness:
a b C d (x
2 3 2
ax ax + bx ax + bx + cx
2 3 2
ax+b ax + bx + c ax +bx + cx + d = f(x)
2
ax 2ax +bx
2
2ax+b 3ax +2bx+c=f ′(x)
According to the rule the last entry in the final column, f(x) =
3 2
a x + b x + c x + d, should be the value of the polynomial whose
coefficients in order are a, b, c, d and whose argument is x, and
this is indeed the case. The method depends on the simple identity
f(x)= [(a x + b) x + c] x + d,
where brackets are introduced so as to resolve all powers of x into
single factors. This can obviously be done for the general

141
Algebra (Chapter 6) Theory of equations

polynomial of degree n.
3 2
The student will find it instructive to divide 2 x – 3 x + + 4x – 5
2
by x – 4, using long division. The quotient is 2x + 5 x + 24 and
the remainder 91. Horner’s method (sometimes called the method
of synthetic division) provides all the materials for the quotient
and remainder more compactly than the ordinary method: and it
has the psychological advantage of employing addition rather than
subtraction as the staple operation.
Horner's method yields the quotient and remainder at once
whenever a polynomial f(x) is divided by a linear divisor x – α.
For the more general divisor α x + β, divide f(x) first by α and
then by x + β/α. This yields the correct quotient, but the remainder
R is relative to x + β/α. It must be re-multiplied by α to give the
correct remainder.
Examples 6.9:
5 4
If f(x) = x – 2x + 3, find f(–3) , f ′(–3).
Solution:
1 –2 0 0 0 3 (–3
–3 15 –45 135 –405
–5 15 –45 135 –402 = f(–3)
–3 24 –117 486
–8 39 –162 621 = f′(–3)
Find also f(5), f(4), f(0.1), f(2), f(3) and try other values of x.

142
Algebra (Chapter 6) Theory of equations

Repeated Roots:
k
Let x = α be a repeated root of f(x) = 0. If f(x) = (x – α) ϕ(x),
where ϕ ( α) ≠ 0 , we say that f(x) = 0 has k- multiple roots α.
k k–1
f′ (x) = (x – α) ϕ′(x) + k(x – α) ϕ(x) =
k–1 k–1
=(x – α) [(x – α) ϕ′(x) + k ϕ(x)] = (x – α) g(x).
The second factor g(x) is different from zero when x = α. It
follows that a k- multiple root of f(x) = 0 is a (k–1)-multiple root
of f′(x) = 0.
Example 6.10:
3 2
Solve 12x – 28 x + 3x + 18 = 0, given that it has two equal roots.
Solution:
3 2
If f(x) = 12x – 28 x + 3x + 18,
2
f′(x) = 36x – 56 x + 3 = (2x – 3)(18x – 1).
Since f(x) = 0 has a repeated root, either 3/2 or 1/18 is the repeated
one. By trial, using synthetic division, we get:
12 –28 3 18 (3/2
18 –15 –18
– 10 –12 0 (3/2
18 12
8 0
2
It is easy to see that, f(x) = (x – 3/2) (12x + 8).
i.e. The roots are 3/2, 3/2, –2/3.
Example 6.11:
4 3 2
Solve 4x – 20 x + 13 x – 1 = 0, given that there are two equal

143
Algebra (Chapter 6) Theory of equations

roots.
Solution:
4 3 2
If f(x) = 4x – 20 x + 13 x – 1,
3 2 2
f′(x) = 16 x – 60 x + 26 x = 2x(8 x – 30 x + 13)=
= 2x(2x – 1)(4x – 13).
Since f(x) = 0 has a repeated root, either 0 or 1/2 or 13/4 is the
repeated one. The first x = 0 is excluded. By trial, using synthetic
division, we get:
4 –20 13 0 –1 (1/2
2 –9 2 1
–18 4 2 0 (1/2
2 –8 –2
–16 –4 0
2 2 2 2
Then f(x) = (x – 1/2) (4x – 16 x – 4) = (2x – 1) (x – 4 x – 1).
i.e. The roots are 1/2, 1/2, 2 ± 5 .
§ 6.4 Transformation of Equations:
Properties of the roots of an equation are often best investigated
by constructing another equation whose roots are related to those
of the given equation by some law.
n n–1 n–2
Let f (x) ≡ a0x + a1x + a2x + … + an–1x + an = 0 (6.1)
Theorem 6.1:
n
f(x) ≡ a0 ∏ ( x − α i ) . (6.2)
i =1

144
Algebra (Chapter 6) Theory of equations

Theorem 6.2:
The algebraic equation of degree n has n and only n roots.
Theorem 6.3: complex roots
The imaginary (or complex) roots of f(x) = 0 occur in pairs.
Theorem 6.4: irrational roots
If the coefficients a0, a1, a2, …, an are rational, the irrational roots
of f(x) = 0 occur in pairs.
Theorem 6.5: Reverse the sign
If α1, α2, α3,…, αn are the roots of f(x) then the roots of f(–x) = 0
are – α1 , – α2 , – α3 ,…, – αn .
To apply this theorem on the cubic equation:
3 2
Let α , β , γ are the roots of a x + b x + c x + d = 0.
To form the equation whose roots are the inverse additives of the
roots?
3 2
Put y = –x, then –α,–β,–γ are the roots of – ay + by – c y + d =0.
Thus, the equation whose roots are the inverse additives of the
3 2 3 2
roots of x – 3x – 6x + 8 = 0 is – x – 3x + 6x + 8 = 0.
3 2
i.e. x + 3x – 6x – 8 = 0.
The roots of the first are –2, 1, 4 ; but of the second are –4, –1, 2.
Increasing or decreasing all the Roots by a Constant Multiple.
Theorem 6.6: Multiplying the roots by k
If α1, α2,…, αn are the roots of f(x) then the roots of f(x/k) = 0

145
Algebra (Chapter 6) Theory of equations

are k α1, k α2, k α3,…, k αn .

3 2
Let α , β , γ are the roots of a x + b x + c x + d = 0.
To form the equation whose roots are the k multiples of the roots?
Put y = k x, then k α , k β , k γ are the roots of
3 2
a ⎛⎜ ⎞⎟ + b ⎛⎜ ⎞⎟ – c ⎛⎜ y ⎞⎟ + d = 0, that is
y y
⎝k⎠ ⎝k⎠ ⎝k⎠
3 2 2 3
a y + b k y + c k y + d k = 0.
This process is called “multiplying the roots of an equation by k.”
This is of special numerical use when k is a power of 10.
Thus, the equation whose roots are 10 times the roots of
3 2 3 2
x – 3x – 6x + 8 = 0 is y – 30 y – 600 y + 8000 = 0.
The roots of the first are –2,1, 4; but of the second are –20, 10, 40.
Theorem 6.7: reciprocals of the roots
If α1, α2,…, αn are the roots of f(x) then the roots of f 1 = 0 are (x )
1/α1, 1/α2, 1/α3,…, 1/αn .
3 2
Let α , β , γ are the roots of a x + b x + c x + d = 0.
To form the equation whose roots are the reciprocals of the roots?

x
()
Put y = 1 , then x = ⎛⎜ 1 ⎞⎟ ,
⎝ y⎠
3 2
a ⎛⎜ 1 ⎞⎟ + b ⎛⎜ 1 ⎞⎟ – c ⎛⎜ 1 ⎞⎟ + d = 0, that is
⎝ y⎠ ⎝y⎠ ⎝y⎠
2 3
a + b k + c y + d y = 0;
3 2
i.e. 1 , 1 , 1 are the roots of d y + c y + b y + a = 0.
α β γ

Thus, the equation whose roots are the reciprocals of the roots of

146
Algebra (Chapter 6) Theory of equations
3 2 3 2
x – 3x – 6x + 8 = 0 is 8 y – 6 y – 3y + 1 = 0.
The roots of the first are –2,1, 4; but of the second are –½ , 1 , ¼ .
Theorem 6.8: squaring of the roots
If α1, α2,…, αn are the roots of f(x) then the roots of f( x ) = 0,
2 2 2 2
when expressed in rational form are α1 , α2 , α3 ,…, αn .
3 2
Let α , β , γ are the roots of a x + b x + c x + d = 0.
To form the equation whose roots are the squares of the given
roots?
2 3/2
Put x = y , then x = y; a y +by+c y +d=0
3/2
ay +c y = –(b y + d) , that is
2 2
y [a y + c] = –(b y + d), i.e. y [a y + c] = (b y + d) .
2 3 2 2 2 2 2
a y + c y + 2 a c y = b y + d + 2bdy.
2 3 2 2 2 2
a y – (b – 2 a c) y + (c – 2bd) y – d = 0.
Thus, the equation whose roots are the squares of the roots of
3 2 3 2 2
x – 3x – 6x + 8 = 0 is y + [2(–6)– 9]y +[36 –2(–3)(8)]y – 8 = 0
3 2
i.e. y – 21y + 84 y – 64 = 0.
The roots of the first are –2,1, 4; but of the second are 4, 1, 16.
4 3 2
Let α , β , γ , δ are the roots of a x + b x + c x + d x + e = 0.
To form the equation whose roots are the squares of the given
roots?
2 2 3/2
Put x = y , then x = y; a y + b y +cy+d y + e = 0.
2 3/2
[a y + cy + e ] = –[ b y +d y ] = – y [b y + d],
2 4 3 2 2 2 2 3 2 2
a y + 2 a c y + (2ae + c ) y + 2cey + e = b y + 2b d y + d y

147
Algebra (Chapter 6) Theory of equations
2 4 2 3 2 2 2 2
a y – ( b – 2 a c) y + (c + 2ae – 2b d )y – (d – 2ce) y + e = 0.
Thus, the equation whose roots are the squares of the roots of
4 3 2
x – 6x + 3 x + 26 x – 24 = 0 is
4 3 2
y – 30 y + 273 y – 820 y + 576 = 0.
The roots of the first are –2,1, 3, 4; but of the second are 4,1,9, 16.
Theorem 6.9: diminished (or increased) the roots
If α1, α2,…, αn are the roots of f(x) then the roots of f (x + k) = 0
are α1– k , α2 – k , α3 – k ,…, αn – k .
If k > 0 the last transformation is called reducing (diminishing) the
roots of an equation by k. If k < 0 it is called increasing by k.
n n–1
Let f(x) = a0x + a1x +…+ an–1x + an= 0. (1)
Be the given equation whose roots are α1, α2,…, αn, and it is
require the equation whose roots are α1– k, α2 – k, …, αn – k.
Suppose that g(x – k) ≡ f(x) has been expressed in the form
n n–1
g(x– k) = b0(x – k) +b1(x – k) +…+ bn–1(x – k) + bn= 0. (2)
Changing the notation by writing y instead of x – k, we have
n n–1
g(y) = b0y + b1y +…+ bn–1y + bn = 0. (3)
Now the roots of g(y) = 0 are those of f(x) diminished by k. Hence
the roots of
n n–1
g(y) = b0y + b1y +…+ bn–1y + bn = 0 (3)
are α1– k , α2 – k , …, αn – k.
f(x) and g(y) are to be identically equal when y = x – k. The
coefficients ai are supposed known; so too is k. We require finding

148
Algebra (Chapter 6) Theory of equations

the coefficients bi. The method of calculating the values of b0,

b1,…, bn–1, bn has been known by Horner’s method.
Horner’s Method of Reducing an Algebraic Equation.
For the purpose of solving a numerical equation by
successive approximations - a method which is exemplified in
ordinary arithmetic when we calculate 2 as 1.4142…, digit by
digit - we need a practical way of expressing f(x) as a polynomial
g(y) where y = x + k.
Suppose, for example, that we wish to express
4 3 2
f(x) = x – 6x + 3 x + 26 x – 24 = 0
4 3 2
in the form b0(x–2) + b1(x–2) + b2(x–2) + b1(x–2) + b0.
We apply Horner's method, dividing f(x) by x – 2. From the
Horner scheme
1 –6 3 26 –24 (2
2 –8 –10 32
–4 –5 16 8
3 2
We infer that the quotient is x –4x –5 x + 16 and that the
remainder is 8. In fact
4 3 2 3 2
x – 6x + 3 x + 26 x – 24 = (x – 4x –5 x + 16)(x – 2) + 8.
The student will find it instructive to reverse the process
and start by multiplying out this identity. The above scheme will
then be seen from another point of view.
3 2
Now let x – 4x –5 x + 16 be treated similarly and divided
by x – 2. From the scheme
149
Algebra (Chapter 6) Theory of equations

1 –4 –5 16 (2
2 –4 –18
–2 –9 –2
3 2 2
We infer that x – 4x –5 x + 16 = (x – 2 x – 9)(x – 2) – 2.
Again, from
1 –2 –9 (2
2 0
0 –9
2
we infer that x – 2 x – 9 = x (x – 2) – 9 ,
and finally from
1 0 (2
2
2
that x = (x – 2) + 2.
We can telescope these results and write
4 3 2
2x – 3x + 4x – 5x + 6 = {[(y + 2) y – 9] y – 2}y + 8
4 3 2
= y + 2 y – 9 y – 2 y + 8,
where y = x – 2, and this is the required form. In fact
b0 = 1, b1 = 2, b2 = – 9 , b3 = – 2 , b4 = 8.
We notice that the successive remainders have furnished
the coefficients bi from right to left, except for b0, which is the
same as a0. Clearly too the process is true in general: the bi (are
the remainders on dividing by y (= x – k) first f(x) and then its
successive quotients.
Furthermore–and this is the chief advantage of the

150
Algebra (Chapter 6) Theory of equations

method–we can tabulate the whole process in one scheme:

1 –6 3 26 –24 (2
2 –8 –10 32
–4 –5 16 8
2 –4 –18
–2 –9 –2
2 0
0 –9
2
2
4 3 2 4 3 2
f(x) = x – 6x + 3 x + 26 x – 24 = y + 2 y – 9 y – 2 y + 8,
where y = x – 2.
It will be seen that this scheme incorporates all that has been said,
but without needless repetition. The method is easy to memorize
directly from such a scheme. Build the scheme from left to right
by rows. At each step multiply the balance in a column by k (k = 2
above) and add it to the balance in the next column to the right.
Perform the step once in the last column, twice in the last but one,
and so on. Rule off with a vertical line at the completion of each
stage. Thus Horner's Method is simply the systematic use of a
very familiar process in elementary arithmetic.
Example 6.12:
Find the equation whose roots are those of
4 3 2
x + 3 x – 2 x – 12 x – 8 = 0,

151
Algebra (Chapter 6) Theory of equations

increased by 2, and hence solve the equation.

Solution:
In this case k = – 2
1 3 –2 –12 –8 (–2
–2 –2 8 8
1 –4 –4 0
–2 2 4
–1 –2 0
–2 6
–3 4
–2
–5
4 3 2
The required equation is y – 5 y + 4 y = 0
2 2 2
i.e. y [y – 5 y + 4] = 0 . Hence y (y – 1)(y – 4) = 0
The roots are 0 , 0 , 1 , 4.
It follows that the roots of the original equation are –2, –2, –1, 2.
Example 6.13:
4 3 2
Given that x + 3 x – 2 x – 12 x – 8 = 0, has the product of two
of its roots equal to 2, find them.
Solution:
Since the product of two roots is 2, then the product of the other
two roots is – 4. Then the given equation can be factorised to the
2 2
two factors (x + m x + 2) and (x + n x – 4)
4 3 2 2 2
Then x + 3 x – 2 x – 12 x – 8 ≡ (x + m x + 2) (x + n x – 4) ≡
4 3 2
≡ x + (m + n) x + ( m n – 2) x + 2(n – 2m) x – 8 = 0.

152
Algebra (Chapter 6) Theory of equations

Equating the corresponding coefficients, we get

m + n = 3 , m n – 2 = – 2 , 2( n – 2m) = – 12
i.e. m + n = 3 , m n = 0 , n – 2m = – 6
Solving these equations we have m = 3 , n = 0.
2 2
Thus we have (x + 3 x + 2) (x – 4) = 0,
i.e. (x + 2)(x + 1)(x – 2)(x + 2) = 0.
Hence the roots are –2 , – 2 , – 1 , 2.
This example shows that when a relation between two of the roots
is known, two dimensions may reduce the degree of the equation.

Equations with Integer Coefficients.

We may always reduce a given equation to the form
n n–1
f(x) = x + a1x +…+ an–1x + an = 0. (1)
A case of special interest arises when the coefficients a1, a2,…,an
are all integers, for the roots of the equation are then integers. If
all the coefficients a1, a2,…,an–1, an of an algebraic equation are
integers the reduction to the still more special form (1), in which
a0 = 1, can always be carried out; for we have only to take z = a0 x
as a new variable to obtain an equation in z, namely,
n n–1 2 n–2 n
z + a0 a1z + a0 a2z +…+ a0 an = 0.
which is evidently of the form (1).
2
For example, if 7x – 15 x + 2 = 0, and z = 7x, we have
2 2
49 x – 105 x + 14 = 0, that is, z – 15 z + 35 = 0.

153
Algebra (Chapter 6) Theory of equations

Theorem 6.10:
Each rational root (if it is exists) of an equation with integer
coefficients in the form (1) is necessarily an integer.
Proof:
Let p/q, rational number and in its lowest terms, be a root of
equation (1). Then
n n–1
(p/q) + a1(p/q) +…+ an–1(p/q) + an = 0. (2)
n–1
Multiply through by q . Then every term is an integer except the
n
first, which p /q, a fraction. This is impossible: hence p/q must be
an integer.
To Find the Rational Roots of a Rational
Equation.
If a0 ≠ 0 we first bring the equation to the standard form (1). By
n
the factorised form of a polynomial f(x) ≡ ∏ ( x − α i ) ,
i =1
is now the product of the roots of the equation, and these if
rational are integers, as we have just seen. Thus the only possible
rational roots are to be found among the factors of an, with
positive or negative sign affixed, and these must be examined
one by one.

Example 6 .14:
4 3 2
Solve 4 x – 12 x + x + 6 x – 2 = 0.

154
Algebra (Chapter 6) Theory of equations

Solution:
3
Instead of multiplying by 4 it is enough here to multiply
throughout by 4 and to put 2 x = z. This gives
4 3 2
16 x – 48 x + 4 x + 24 x – 8 = 0.
4 3 2
f(z) = z – 6 z + z + 12 z – 8 = 0.
The factors of – 8 are ±1, ±2, ±4, ±8.
Calculate f(±1) , f(±2) , f(±4) , f(±8) .
1 –6 1 12 –8 (–1
–1 7 –8 –4
–7 8 4 –12
∴ f(–1) = – 12
1 –6 1 12 –8 (1
1 –5 –4 8
–5 –4 8 0
∴ f(1) = 0.
3 2
Therefore f(z) = (z – 1)( z – 5 z – 4 z + 8) = (z – 1)g(z),
3 2
where g(z) = z – 5 z – 4 z + 8
Once more the factors of 8 are also ±1, ±2, ±4, ±8.
Calculate g(±1) , g(±2) , g(±4) , g(±8) .
1 –5 –4 8 (1
1 –4 –8
–4 –8 0
2 2
Then f(z) = (z – 1) ( z – 4 z – 8 )
2
Actual roots are 1, 1, so that (z – 1) is a factor of f(z). The other
2
factor is z – 4 z – 8, which leads to irrational roots. Hence the
roots of f(z) are 1 , 1 , 2( 1 + 3 ) , 2( 1 + 3 ).

155
Algebra (Chapter 6) Theory of equations

Since z = 2 x , the roots of the original equation are

½,½,1+ 3, 1 – 3.
Remark:
Theorem 6.10 stated that: “Each rational root of an equation with
integer coefficients in the form (1) is necessarily an integer.
This means that if the rational root exists it must be integer. The
following equation has integer coefficients and in the form (1):
2
x + 4x – 1 = 0 , it is easy to see that its roots are x = –2 ± 5.
See also that Example 6.21.
Example 6.15:
4 2
Solve 4 x – 5 x – 10 x + 6 = 0.
Solution:
3
Instead of multiplying by 4 it is enough here to multiply
throughout by 4 and to put 2 x = z. This gives
4 2
16 x – 20 x – 40 x + 24 = 0.
4 2
f(z) = z – 5 z – 20 z + 24 = 0.
The factors of 24 are ±1, ±2, ±3, ±4, ±6, ±8, etc.
Calculate f(±1) , f(±2) , f(±3) , f(±4) etc.
1 0 –5 –20 24 (1
1 1 – 4 –24
1 –4 –24 0
∴ f(1) = 0
3 2
Therefore f(z) = (z – 1)( z + z – 4 z – 24) = (z – 1)g(z),

156
Algebra (Chapter 6) Theory of equations
3 2
where g(z) = z + z – 4 z – 24
Once more the factors of 24 are ±1, ±2, ±3, ±4, ±6, ±8, etc.
Calculate g(±1) , g(±2) , g(±3) , g(±4), etc.
It is easy to see that g(±1) ≠ 0 , g(±2) ≠ 0 and f(3) = 0
1 1 –4 –24 (3
3 12 24
4 8 0
Actual roots are 1, 3, so that (z – 1)(z – 3) is a factor of f(z). The
2
other factor is z + 4 z + 8, which leads to complex roots.
2
Then f(z) = (z – 1) (z – 3) ( z + 4 z + 8 )
Thus the roots of f(z) are 1 , 3 , – 2 + 2i , – 2 – 2i. Since z = 2 x ,
the roots of the original equation are ½ , 3/2 , – 1 + i , – 1 – i.
Bi-quadratic equations:
There are certain types of equation, which are reducible to
equations of lower degree and root extractions. It is easy, for
example, to reduce equations such as
2n n 3n 2n n
ax + b x + c = 0 , or ax +bx + c x + d = 0,
n
by the substitution y = x , which reduces to
2 3 2
ay + b y + c = 0 , or a y + b y + c y + d = 0.
4 2
Bi-quadratic equations is in the form Ax + B x + C = 0, which
2 2
reduced to the quadratic A y + B y + C = 0 , y = x .
− B ± B 2 − 4AC 2
Its solution is given by y1,2 = =x .
2A
− B + B 2 − 4AC − B − B 2 − 4AC
Thus x1,2 = ± , x3,4 = ± .
2A 2A

157
Algebra (Chapter 6) Theory of equations

Example 6 .16:
4 2
Solve x – 13 x + 36 = 0.
Solution:
2 2
Put y = x , we get y – 13 y + 36 = 0. ∴ (y – 4)(y – 9) = 0.
Therefore the roots of the original equation are –3, –2, 2 , 3.
Solution of the Bi-quadratic Equation:
The bi-quadratic expression may be represented as the product of
two quadratic expressions. By equating to zero each of the
quadratic factors the four roots of the bi-quadratic equation are
obtained. In practice it may not be necessary to proceed with the
general method in order to determine the quadratic factors as they
may sometimes be obtained by inspection. Further, if any relation
between the roots is known, the equation can always be reduced to
one of lower degree.
Example 6.17:
4 3 2
Solve the bi-quadratic equation x – x + 2x + x + 3 = 0.
Solution:
4
Since the coefficient of x is unity and the free term is 3 = ±1× ± 3
the factors may be of the form
2 2
(x + a x ± 1) (x + b x ± 3),
like signs corresponding.
Taking the – signs the product is

158
Algebra (Chapter 6) Theory of equations
4 3 2
x + (a + b)x + (a b – 4)x + (3a + b) x + 3 . If this expression
equated to zero is identical with the given equation
a + b = –1, a b – 4 = 2, 3a + b = – 1.
The first and last equations give a = 0 and this obviously cannot
satisfy the second. Thus the three equations are inconsistent.
Taking the + signs we obtain
4 3 2
x + (a + b)x + (a b + 4)x + (3a + b) x + 3 = 0.
If this is identical with the given equation, a + b = –1, 4 + a b = 2,
3a + b = 1. The first and third equations give a = 1, b = –2. These
values satisfy the second equation. Hence the given equation is
equivalent to
2 2
(x + x + 1) (x – 2 x + 3) = 0.
2
The roots being – 1 [1 ± 3 ] , 1 ± i 2 i.e. ω, ω , 1 ± i 2 .
2
Equation reduced to quadratic equation:
Equation in the form: (x – a) (x – b) (x – c) (x – d) = k.
If a + b = c + d = α , then
2 2
(x – a) (x – b) = x –(a + b) x + a b = x – α x + a b,
2 2
(x – c) (x – d) = x –(c + d) x + c d = x – α x + c d,
2 2
Thus (x– a)(x– b)(x– c)(x– d)= (x –α x + a b)( x – α x + c d) = k.
This equation can be reduced to quadratic equation in the form
2
( z + a b)( z + c d) = k, where z = x – α x.
Example 6 .18:
Solve (x – 2)(x – 3)(x + 4)(x + 5) = 144.

159
Algebra (Chapter 6) Theory of equations

Solution:
It is clear that 2 – 4 = – 2 = 3 – 5, then
2 2
(x – 2)(x + 4) = x + 2 x – 8, (x – 3)(x + 5) = x + 2 x – 15.
2
Put z = x + 2 x, then (z – 8)(z – 15) = 144,
2
z – 23 z – 24 = 0 ∴ (z – 24)(z + 1) = 0.
i.e. z = 24 or z = –1
2 2
If z = 24 , then x + 2 x = 24 i.e. x + 2 x – 24 = 0
∴(x – 4)(x + 6) = 0, i.e. x = –6 or 4.
2 2 2
If z = – 1 , then x + 2 x = – 1 i.e. x + 2 x + 1 = 0 ∴ (x + 1) = 0
∴ x = –1 , –1.
Then the four roots are –6 , –1 , –1 , 4.
Binomial Equations:
If each coefficient of
n n–1 n–2
f (x) ≡ a0x + a1x + a2x + … + an–1x + an = 0, (1)
vanishes except two, the equation is binomial in form, and reduce,
m
apart from zero roots, to x – a = 0, which is called the binomial
equation. By means of De Moiver’s theorem this can always be
solved. The roots of such equations have been considered in
n th
complex variables. The roots of x – 1= 0, are the n roots of
unity.
2
Consider the case n = 2. Then since x – 1 =(x – 1) (x + 1),
the square roots of unity are –1, 1.
2 2 2
Since x + 1 ≡ x – i = (x – i) (x + i),

160
Algebra (Chapter 6) Theory of equations

the square roots of –1 are –i, i.

Consider the case n = 3. Then since
3 2
x – 1 =(x – 1) (x + x + 1),
2
the cube roots of unity are 1, 1 (–1 ± i 3 ). i.e. 1, ω , ω .
2
Let ω denote one of the imaginary roots, i.e. one of the roots of
2 2 3
x + x + 1 = 0. Then ω + ω + 1 = 0 , ω = 1.
2
Also since the product of the roots of x + x + 1 = 0 is unity it
2
follows that if one root is ω the other is ω . Thus any product
involving real quantities and powers of ω can be represented in
2 2
one of the forms A + Bω , A + Bω , Aω + Bω
where A and B are real quantities. In particular
2 2 3 3 3
(a + b + c) (a + ωb + ω c) (a + ω b + ωc) = a + b + c – 3abc.
2 2
For (a + ωb + ω c) (a + ω b + ωc) =
2 3 2 3 2 2 4 2 2
=a + ω b +ω c + a b(ω + ω ) + b c (ω + ω ) + c a(ω + ω) =
2 2 2
≡ a + b + c – a b – b c – c a,
3 4 3 2
since ω = 1, ω = ω .ω = ω , ω + ω = – 1.
Multiplying both sides of the identity by (a + b + c) we obtain the
required form.
Consider the case n = 4. Then since
4 2 2
x – 1 =(x – 1) (x + 1) ≡(x + 1) (x – 1) (x + i) (x – i)
the fourth roots of unity are ± 1, ± i.
If in the above discussion we replace unity by any quantity k we
obtain the equations

161
Algebra (Chapter 6) Theory of equations
n
x – k = 0.
th
The n roots of these equations are found in a similar way.
1/3
In particular if k is real and k ) is the ordinary arithmetical cube
3 1/3 1/3 2 1/3
root of k then the three roots of x – k = 0 are k , ωk , ω k .
Note that the three roots may be obtained from any one of them by
2
multiplying it by 1, ω , ω .
3 2
Since x + 1 =(x + 1) (x – x + 1),
2 1± − 3 1± i 3
The roots of x – x + 1 = 0, are x1,2 = =
2 2
2 3 2
i.e. –ω,–ω . Therefore the roots of x + 1 = 0 are –1, – ω and –ω .
De Moiver’s Theorem:
For all real values of n, cos nθ + i sin nθ is the value, or
n
one of the values of (cos θ + i sin θ) .
Let a be written in the complex form a = r (cos θ + i sin θ),
n th
Where r > 0; and let b = r, where b = n r , the arithmetical n root
n
of the modulus r of a. If we now put x = by the equation x = a
n
reduces to y = (cos θ + i sin θ), which is solved by taking
y = cos θ + 2kπ + i sin θ + 2kπ , k = 0, 1, 2, …, n –1.
n n
For the n given values of k, y takes exactly n different values,
which are represented by points on a circle of unit radius whose
amplitudes differ successively by 2π/n; and such points lie at the
vertices of a regular polygon inscribed in the circle. Hence each
th
possible root of this equation of the n degree has been identified,
for there cannot be more than n roots. Furthermore, if k takes

162
Algebra (Chapter 6) Theory of equations

higher values n, n+1,…, or else negative values, the same vertices

are repeated.
n
General solutions of x = a.
By combing the above results we obtain the n values
x = |a| (cos θ + 2kπ + i sin θ + 2kπ ), k = 0, 1, 2, …, n –1,
1/n
n n
where |a| denotes the modulus of a, and θ is its amplitude.
Examples:
3
1) x – 1 = 0
Here a = 1 = cos 0 + i sin 0 and so θ = 0,
x = cos 0 + i sin 0 = 1, cos 2π + i sin 2π , cos 4π + i sin 4π .
3 3 3 3
2
i.e. x = 1 , (–1 + i 3 ), – (1 + i 3 ) i.e. 1, ω , ω .
1 1
2 2
The regular polygon in this case is an equilateral triangle with one
vertex at the point (1 , 0).
3
2) x + 1 = 0
Here a = –1 = cos π + i sin π and so θ = π,
x = cos π + i sin π , cos π + i sin π, cos 5π + i sin 5π .
3 3 3 3
2
= 1 (1 + i 3 ) , – 1, 1 (1 – i 3 ) i.e. –ω, –1, –ω .
2 2
The regular polygon in this case is an equilateral triangle with one
vertex at the point (–1 , 0).
3
3) x + 8 = 0.
Here a = –8 = 8(cos π + i sin π) and so
x = 2(cos π + i sin π ) , 2(cos π + i sin π), 2(cos 5π + i sin 5π )
3 3 3 3

163
Algebra (Chapter 6) Theory of equations
2
= 1 + i 3 , – 2, 1 – i 3 = –2ω , –2, –2ω .
The regular polygon in this case is an equilateral triangle with one
vertex at the point (–2 , 0).

4
4) x + 1= 0
Here a = –1 = (cos π + i sin π) and so θ = π,
( 2k +1) π ( 2k +1) π
x = cos + i sin , k = 0, 1, 2 , 3.
4 4
x = cos π + i sin π , cos 3π + i sin 3π , cos 5π + i sin 5π ,
4 4 4 4 4 4
cos 7 π + i sin .7 π
4 4
i.e. x = 1 + i 1 , – 1 +i 1 , – 1 –i 1 , 1 –i 1 .
2 2 2 2 2 2 2 2
i.e. the roots are ± 1 [1 + i] , ± 1 [1 – i].
2 2
7
5) x = 1.
Here a = 1 , θ = 0 , x = cos 2kπ +i sin 2kπ , k = 0,1,…,6.
7 7
i.e. x = 1, cos + i sin , cos + i sin 4π , cos 6π + i sin 6π ,
2 π 2 π 4 π
7 7 7 7 7 7
cos 8 π 8 π
+ i sin , cos 10 π + i sin 10 π , cos 12 π + i sin 12 π .
7 7 7 7 7 7
5
6) x = i.
( 4k +1) π ( 4k +1) π
Here a = i , θ = π , x = cos + i sin , k = 0 , 1 ,…, 4.
2 10 10
i.e. x = cos + i sin , cos + i sin ≡ i , cos 9π + i sin 9π ,
π π 5 π 5 π
10 10 10 10 10 10
cos 13π + i sin 13π , cos 17 π + i sin 17 π .
10 10 10 10
n
7) The real roots of x –1 = 0 are ±1, if n is even, and +1
n
only, if n is odd. On the other hand, x + 1 = 0 has a single real
root –1 if n is odd, and no real root if n is even.
164
Algebra (Chapter 6) Theory of equations

Special properties:
n
(a) If Ω is an imaginary root of x – 1 = 0, and m is any positive
m n
integer, then Ω is also a root of x – 1 = 0·
m n mn n m m m
Now (Ω ) = Ω = (Ω ) =(1) = 1. Thus Ω is also a root of
the equation.
n
(b) If n is a prime number, Ω any imaginary root of x – 1 = 0,
then the n roots may be expressed in the form
2 n–1
1,Ω,Ω ,…,Ω .
(c) If n and m denote positive integers which have no common
n m
factor except unity then the equations x – 1 = 0; x – 1 = 0, have
no common root except unity.
(d) If p is the highest common factor of the positive integers n and
n m
m, then the roots common to the equations x – 1 = 0, x – 1 = 0
p
are the roots of x – 1 = 0.
It will be observed that if in this result we write p = 1, we obtain
(c).
th
A special n root of unity is defined to be a root of
n m
x – 1 = 0, and which satisfies no other equation x – 1 = 0, where
n
m < n. It will then follow that all the roots of x – 1 = 0 will be
2 n–1 th
given by the sequence 1 , α , α , … , α where α is a special n
root of unity.
Example 6.19:
6
Find the special roots of x – 1 = 0.

165
Algebra (Chapter 6) Theory of equations

Now 6 may be written as the product of two positive integers (> 1)

6
in only one way 2 × 3. Hence the equation x – 1 = 0, has roots in
2 3
common with x – 1 = 0, x – 1 = 0.
2 3 6
Now (x – 1), (x – 1) are factors of x – 1 = 0. Thus in
6
determining the special roots of x – 1 = 0, it is sufficient to
6 2
determine the roots of x – 1 = 0 which are not roots of x – 1 = 0,
3
x – 1 = 0.
6 3 3
Now x – 1 = (x – 1)(x + 1). Thus the required roots will be
3 2
those of x + 1 = 0 which are not roots of x – 1 = o.
2 3
It is easy to see that –1, –ω , –ω are the roots of x + 1 = 0. Then
2 2 2 3
the special roots are –ω ,–ω . Take –ω ,(–ω) = ω , (–ω) = –1 ,
4 5 2 6
(–ω) = ω, (–ω) = – ω , ( –ω) = 1.
6
Thus there are two special roots of x – 1 = 0, and these are the
2 3 3 2
roots of x – x + 1 = 0, since x + 1 = (x + 1)( x – x + 1).
2
i.e. –ω , –ω .
Cubic Equations:
The classical method of solving a cubic equation was
invented by Ferro and by Tartaglia, but was first published in
1548 by Cardan who had obtained it from Tartaglia after giving a
solemn promise not to reveal it. In spite of these facts, it is
generally called Cardan's method.
The solution of the general cubic
3 2
a x + b x + c x + d = 0,

166
Algebra (Chapter 6) Theory of equations

may be made to depend upon that of a simpler cubic by the

substitution y = x + h, i.e. x = y – h . This gives
3 2
a (y – h) + b (y – h) + c (y – h) + d = 0
3 2 2 3 2 2
a[y – 3 y h + 3 y h – h ] + b[y – 2 y h + h ]+ c y – c h + d = 0
3 2 2 3 2
a y + y [b – 3a h] + y[ 3a h – 2bh + c] –a h + b h – c h + d = 0
Now choose b – 3a h = 0, i.e. h = b/3a, then we get
3 2 3 2
a y + y[ 3a h – 2bh + c] – ah + b h – c h + d = 0
This equation can be written in the form
3
z + p z + q = 0, (1)
To solve this equation, put z = u + v. Then
3 3 2 2 3 3 3
z = u + 3 u v + 3 u v + v = u + v + 3 u v (u + v)=
3 3
= u + v + 3 u v z. (2)
Equation (2) may be regarded as a cubic equation in z and is in
2
reduced form, since there is no term in z . Equation (2) can be
written in the form
3 3 3
z – 3 u v z – (u + v ) = 0. (3)
Equations (1) and (3) are the same provided that
3 3
u + v = – q , 3 u v = – p. (4)
To solve this pair of simultaneous equations for u and v, write
3 3 3 3 3
u v = – p /27, so that u and v are the roots of the quadratic
equation, where the sum of the roots is –q, and the product of the
3 3
roots, is – p /27 = –(p/3) .
2 3
i.e. t + q t – (p/3) = 0 , (5)

167
Algebra (Chapter 6) Theory of equations
3 3
Hence t = u or v ; that is
q q 2 + (4p 3 / 27) q ⎛ q ⎞ 2 ⎛ p ⎞3 3 3
t=– ± = – ± ⎜ ⎟ + ⎜ ⎟ = u or v . (6)
2 2 2 ⎝2⎠ ⎝3⎠
Since u and v have entered the formulas symmetrically, it does not
matter which is which: and so we let
3 q ⎛ q ⎞ 2 ⎛ p ⎞3 3 q ⎛ q ⎞ 2 ⎛ p ⎞3
u =– + ⎜ ⎟ +⎜ ⎟ ,v =– – ⎜ ⎟ +⎜ ⎟
2 ⎝2⎠ ⎝3⎠ 2 ⎝2⎠ ⎝3⎠
Finally we take the cube roots and add, so that
1 1
⎡ 2 3 ⎤3 ⎡ 2 3 ⎤3
− q ⎛ q ⎞ ⎛ p ⎞ − q ⎛ q ⎞ ⎛ p ⎞
z= ⎢ + ⎜ ⎟ + ⎜ ⎟ ⎥ +⎢ − ⎜ ⎟ +⎜ ⎟ ⎥ . (7)
⎢ 2 ⎝2⎠ ⎝3⎠ ⎥ ⎢ 2 ⎝2⎠ ⎝3⎠ ⎥
⎣ ⎦ ⎣ ⎦
3 3
Since every t (= u or v ) has three distinct cube roots, we have
evidently obtained several values of z by Cardan’s formula. It
appears that there are nine possibilities in the formula. We choose
the values which satisfies the relation u v = – p/3. Only three ways
will satisfy u v = –p/3. The three roots for z are given by the
2 2
formulae z = u + v , ω u + ω v , ω u + ω v , where u , v are the
2
cube root of t and ω , ω = 1 (–1 ± i 3 ),
2
2 πi 4 πi
2
ω = cis 2π/3 = e 3 , ω = cis 4π/3 = e 3 .
If ∆= ( ) ( ) > 0, then formula (7) gives real values of
q 2 p 3
2
+
3

both u and v, so that z = u + v is also real , and the other two

2 2
values of z are ω u + ω v and ω u + ω v; i.e. we have two
complex roots and only one real root.

168
Algebra (Chapter 6) Theory of equations

If ∆= ( ) ( ) 0, then we have three real roots.

q 2
2
+
p 3
3
<

If ∆= ( ) ( ) = 0, then we have two equal roots.

q 2
+
p 3
2 3

In the latter two cases the three roots are actually real, so that the
method is not convenient for a cubic equation in which all the
roots are real.
Example 6 .20:
3 2
Solve x + 6 x – 12 x + 32 = 0.
Solution:
2
To remove the term in x , put h = b / 3a = 6/3 = 2 , x = y – 2.
Then reduce the roots of the given equation by –2
1 6 –12 32 (–2
–2 –8 40
4 –20 72
–2 –4
2 –24
–2
0
3
The transformed equation is y – 24 y + 72 = 0.
() ()
q 2 p 3 ⎛ 72⎞ ⎛ − 24 ⎞
2 3
2 3
[It is easy to see that ∆= + =⎜ ⎟ +⎜ ⎟ =(36) –(8) > 0. i.e.
2 3 ⎝2⎠ ⎝ 3 ⎠

we have two complex roots and only one real root.]

Now suppose that y = u + v, where
3 3 3 3
u + v = –q = – 72 , u v = – p/ 3 = 24/3 = 8 ; u v = 512.
3 3
Thus u , v are the roots of the quadratic equation

169
Algebra (Chapter 6) Theory of equations
2
t + 72 t + 512 = 0, i.e. (t + 8)(t + 64) = 0.
The roots are t = – 8 , t = – 64.
3 3
Taking u = –8 , u = –2 ; v = – 64 , v = – 4.
Thus the roots of the transformed equation, are u + v = –6 ,
2 2 2 2
ωu + ω v = –2( ω +2ω ), ω u + ω v = –2(ω +2ω).
3 2
Hence the roots of the given equation x + 6 x – 12 x + 32 = 0,
2 2 2
are – 8 , –2(1+ω +2ω ) , –2(1+ ω +2ω), i.e. –8 , –2ω , –2ω.
We shall use in the following examples the well-known results in
complex numbers:
eiθ = (cos θ + i sin θ) = cis θ, e − iθ = (cos θ – i sin θ) = cis (– θ),
iθ − iθ
cos θ = e + e = 1 [cis θ + cis (– θ)],
2 2
iθ
− iθ
sin θ = e − e = 1 [cis θ + cis (– θ)] .
2i 2i

Example 6 .21:
3
Solve x – 6 x – 4 = 0.
Solution:
[It is easy to see that
() ()
q 2 p 3 ⎛−4⎞ ⎛−6⎞
2 3
2 3
∆= + = ⎜ ⎟ + ⎜ ⎟ =(2) –(2) = – 4 < 0.
2 3 ⎝ 2 ⎠ ⎝ 3 ⎠
i.e. we have three real roots.]
3
Let x + p x + q = 0, p = – 6 , q = – 4.
Now suppose that x = u + v, where
3 3 3 3
u + v = – q = 4 , u v = – p/ 3 = 6/3 = 2 ; u v = 8.

170
Algebra (Chapter 6) Theory of equations
3 3
Thus u , v are the roots of the quadratic equation
2
t – 4 t + 8 = 0, i.e. t = 2 ± 2i.
3 3
Taking u = 2 + 2i , v = 2 – 2i, where u v = 2.
1 1
) = 2 2 (cos π + i sin π ) =
3
But u = 2 + 2i = 2 2 ( +i
2 2 4 4

= 2 2 (cis π ).
4
u v = 2/u
2 (cos π + i sin π ) ≡ 2 cis( π ), 2 (cos π – i sin π ),
12 12 12 12 12
π π
2 (cos +i sin )≡ 2 cis ,
3 3 3π 2 (cos – i sin 3π ),
3 π
4 4 4 4 4
2 (cos 17 π +i sin 17 π )≡ 2 cis 17 π , 2 (cos 17 π – i sin 17 π ).
12 12 12 12 12

Choose u, v such that u v = 2, then the roots are

u1 + v1 = 2 2 cos π = 2.73205081 = 1 + 3 ,
12
u2 + v2 = 2 2 cos 3π = –2,
4
u3+v3 = 2 2 cos 17 π =– 0.73205081 = 1 – 3.
12

3
Algebraic method: x – 6 x – 4 = 0
Since the free term – 4 has factors ± 1 , ± 2 , ± 4.]
It is easy to see that x = – 2 is a root, and the equation can be
2
written in the form (x + 2)(x – 2 x – 2) = 0, and the roots are
x = –2, 1 ± 3 .i.e. –2 , –0.73205081, 2.73205081.

Example 6 .22:
3 2
Solve x – 3x – 24 x + 80 = 0.

171
Algebra (Chapter 6) Theory of equations

Solution:
2
To remove the term in x , 1 –3 –24 80 (1
put h = b / 3a = –3/3 = –1 , 1 –2 –26
x = y + 1. –2 – 26 54
Then diminishing the roots of 1 – 1
the given equation by 1. The
–1 –27
transformed equation is
3 1
y –27y+54 = 0, p =– 27, q = 54.
0

() ()
q 2 p 3 ⎛ 54⎞ ⎛ −27⎞
2 3
2 3
[It is easy to see that ∆= + =⎜ ⎟ +⎜ ⎟ = (27) –(9) = 0.
2 3 ⎝2⎠ ⎝ 3 ⎠

i.e. we have two equal roots and of course three real roots.]
Now suppose that y = u + v, where
3 3 3 3
u + v = – q = –54 , u v = – p/ 3 = 27/3 = 9 ; u v = 729.
3 3
Thus u , v are the roots of the quadratic equation
2 2
t + 54 t + 729 = 0, i.e. (t + 27) = 0.
The roots are t = – 27 , t = – 27.
3 3
Taking u = –27 , u = –3 ; v = – 27 , v = – 3.
3
Hence the roots of the given equation y – 27y + 54 = 0, are
2 2
u + v = – 6 , ωu + ω v = –3(ω + ω ) = 3,
2 2
ω u + ω v = –3( ω + ω) = 3. i.e. – 6, 3, 3.
Since x = y + 1, then the roots of the given equation
3 2
x – 3x – 24 x + 80 = 0, are – 5 , 4, 4.

Example 6 .23:
3 2
Solve x – 3 x – x + 3 = 0.

172
Algebra (Chapter 6) Theory of equations

Solution:
2 1 –3 –1 3 (1
To remove the term in x , put
1 –2 –3
h = b / 3a = –1 , x = y + 1. Then –2 –3 0
reduce the roots of the given 1 –1
equation by 1. The transformed –1 –4
3 1
equation is y – 4 y = 0.
0
2
It is clear that y(y – 4) = 0 , i.e. y(y – 2)( y + 2) = 0
i.e. y = 0 or y = –2 or y = 2. Hence x = 1 or x = –1 or x = 3.
Try to solve it by Cardan’s method to see why it is not convenient
in that case.
Quartic Equations: Ferrari's Solution:
This method deals with a systematic way to factorise the given
quartic equation into two quadratic equations. In this method we
2
add a perfect square like (α x + β) , to both sides of the given
equation. Then trying to make the new left hand side, a perfect
2 2
square like (x + A x + B) , by a suitable choice of α and β. This
gives us two quadratic equations. The four roots of them are also,
the roots of the original equation.
The general quartic equation is in the following form
4 3 2
a0 x + a1 x + a2 x + a3 x + a4 = 0. (1)
Dividing by a0, it reduced to
4 3 2
x +2ax + bx +cx+d=0. (2)
2
Adding to both sides the (perfect square) quantity (α x + β) .

173
Algebra (Chapter 6) Theory of equations

Now choosing α, β such that the left hand side is perfect square.
4 3 2 2 2
x + 2 a x + b x + c x + d + ( α x + β) = ( α x + β) ,
4 3 2 2 2 2
or x + 2 a x + (b + α )x + (c + 2α β) x + d + β = (α x + β) ,(3)
The left hand side may be written in the form
2 2 4 3 2 2 2
(x + a x + λ) = x + 2a x + (2λ + a )x + 2a λ x + λ . (4)
Equating the corresponding coefficients, in (3) and (4), we get
2 2 2 2
b + α = 2λ + a , c + 2α β = 2 a λ and d + β = λ .
2 2 2 2
i.e. α = 2λ + a – b , β = λ – d , 2α β = 2a λ – c
Eliminating α , β in the above three equations, we get
2 2 2 2 2
4 α β = (2 a λ – c) = 4(2λ + a – b) (λ – d).
This is a cubic equation in λ, and it has at least one real root.
2 2 2
Equations (4) and (5) gives (x + a x + λ) = (α x + β)
2
i.e. x + a x + λ = ± (α x + β).
Solving the last two quadratic equations, we get the four roots of
the original equation.
Example 6 .24:
4 3 2
Solve x – 2x – 12 x + 10 x + 3 = 0.
Solution:
2
Adding (α x + β) to both sides, we get
4 3 2 2 2
x – 2x – 12 x + 10 x + 3 + (α x + β) = (α x + β)
4 3 2 2 2 2
x – 2x + (α – 12) x + (2α β + 10 ) x + (3 + β ) = (α x + β)
2 2 4 3 2 2
(x – x + λ) = x – 2 x + (2λ + 1)x – 2 λ x + λ . (4)
2 2
2λ + 1 = α – 12 ⇒ α = 2λ + 13,

174
Algebra (Chapter 6) Theory of equations

–2λ = 2α β + 10 ⇒ 2α β = –2λ – 10, i.e. α β = –λ – 5

2 2 2 2
λ =3+β ⇒ β =λ –3
2 2
(λ – 3)( 2λ + 13)= (λ + 5) .
3 2 2
Or 2λ + 13λ – 6 λ – 39 = λ + 10 λ + 25
3 2
i.e. λ + 6 λ – 8 λ – 32 = 0.
It is easy to see that λ = – 2 is the real root of that equation.
2
α = 2λ + 13 = 13 – 4 = 9 ⇒ α = ± 3,
2 2
β = λ – 3 = 4 – 3 = 1 ⇒ β = ± 1,
α β = –λ – 5 = 2 – 5 = –3
∴ α = 3 , β = –1 or α = – 3 , β = 1
2
∴ (x – x + λ) = ± (α x + β)
The two quadratic equations are
2 2
x – x – 2 = –3 x + 1 ⇒ x + 2 x – 3 = 0 ⇒ (x + 3)(x – 1) = 0.
2 2
x – x – 2 = 3 x –1 ⇒ x – 4x – 1 = 0, i.e. x = 2 ± 5 .
Now, it is easy to see that
4 3 2 2 2
x – 2x – 12 x + 10 x + 3 = (x + 2 x – 3)( x – 4x – 1).
The four roots are 1, –3 , 2 ± 5.
Another solution:
This equation can be solved by trial. The free term is 3 .The
factors of 3 are ±1, ±3. Calculate f(±1) , f(±3), where
4 3 2
f(x) = x – 2x – 12 x + 10 x + 3 .
To calculate f(1): 1 –2 –12 10 3 (1
1 –1 –13 –3
–1 –13 –3 0
Hence f(1) = 0.

175
Algebra (Chapter 6) Theory of equations
4 3 2 3 2
∴ x – 2x – 12 x + 10 x + 3 = (x – 1)( x – x – 13 x – 3)
To calculate f(–1): 1 –1 –13 – 3 (–1
–1 2 11
–2 –11 8
Hence f(–1) ≠ 0.
To calculate f(–3): 1 –1 –13 –3 (–3
–3 12 3
–4 –1 0
Hence f(–3) = 0.
4 3 2 2
∴ x – 2x – 12 x + 10 x + 3 = (x – 1)(x + 3)( x –4x – 1),
2 4 ± 16 + 4
To solve x – 4x – 1 = 0 ∴ x = = 2 ± 5.
2
Therefore the four roots are – 3, 1, 2 ± 5.

Example 6 .25:
4 3 2
Solve x + x – 7 x – x + 6 = 0.
Solution:
This equation can be solved by trial. The free term is 6 .The
factors of 6 are ±1, ±2, ±3, ± 6.
Calculate f(±1) , f(±2) , f(±3) , f(±6).
Using synthetic division to calculate f(1);we get
1 1 –7 –1 6 (1
1 2 –5 –6
2 –5 –6 0
Hence f(1) = 0.
4 3 2 3 2
∴ x + x – 7 x – x + 6 = (x – 1)( x + 2x – 5 x – 6)

176
Algebra (Chapter 6) Theory of equations
To calculate f(–1): 1 2 –5 –6 (–1
–1 –1 6
1 –6 0
Hence f(–1) = 0 and
4 3 2 2
x + x – 7 x – x + 6 = (x – 1) (x + 1) ( x + x – 6) = ⇒
= (x – 1) (x + 1) (x – 2) (x + 3).
Therefore the four roots are – 3, – 1, 1, 2. Try to solve it by
Ferrari’s Method.
No general method exists for the solution of equations of higher
degree than 4 in terms of the literal coefficients.
Example 6 .26:
4 3 2
Solve x + 2x – 5 x + 6 x – 3 = 0.
Solution:
2
Adding (α x + β) to both sides, we get
4 3 2 2 2
x + 2x – 5 x + 6 x – 3 + (α x + β) = (α x + β)
4 3 2 2 2 2
x + 2x + (α – 5) x + (2α β + 6 ) x + ( β – 3) = (α x + β)
2 2 4 3 2 2
(x + x + λ) = x + 2 x + (2λ + 1)x + 2 λ x + λ . (4)
2 2
2λ + 1 = α – 5 ⇒ α = 2λ + 6,
2λ = 2α β + 6 ⇒ 2α β = 2λ – 6, i.e. α β = λ – 3,
2 2 2 2
λ = β –3 ⇒ β = λ + 3,
2 2
(λ + 3)( 2λ + 6) = (λ – 3) .
3 2 2
Or 2λ + 6 λ + 6 λ + 18 = λ – 6 λ + 9
3 2
i.e. 2 λ + 5 λ + 12 λ + 9 = 0.
It is easy to see that λ = – 1 is the real root of that equation.

177
Algebra (Chapter 6) Theory of equations
2 2 2
α = 2λ + 6 = 4 ⇒ α = ± 2, β = λ + 3 = 4 ⇒ β = ± 2,
αβ= λ–3=–4
∴ α = 2 , β = –2 or α = – 2 , β = 2.
2
∴ (x + x + λ) = ± (α x + β)
The two quadratic equations are
2 2 1± 1− 4 1± i 3
x +x–1=2x–2⇒x – x+1=0⇒x= = ,
2 2
2 2 −3± 9+12 − 3 ± 21
x + x – 1 = –2 x + 2 ⇒ x + 3x – 3 = 0⇒ x = = .
2 2
Now, it is easy to see that
4 3 2 2 2
x + 2x – 5 x + 6 x – 3 = (x – x + 1)( x + 3x – 3).
1 ± i 3 − 3 ± 21
The four roots are , .
2 2
Reciprocal Equations:
An equation, which is unaltered when x is replaced by l/x, is
called a reciprocal equation.
n n–1 n–2
If f (x) ≡ a0x + a1x + a2x + … + an–1x + an = 0,
is reciprocal, it is equivalent to
n n–1 n–2
an x + an–1x + a2n–2 x + … + a1x + a0 = 0
a0 ar an
∴ = =
a n a n −r a0
From the equality of the first and last ratio it follows that:
either (i) a r = a n–r (r = 0 , 1 , 2 ,…, n),
or (ii) a r = – a n–r (r = 0 , 1 , 2 ,…, n).
(i) ar = an–r : i.e. the coefficients of the corresponding terms

178
Algebra (Chapter 6) Theory of equations

taken from the beginning and end are equal.

a) If the equation is of odd degree, n = 2m+1; it is satisfied by
x = –1; and when the factor (x + l) is removed, there remains a
reciprocal equation of even degree 2m of the form
2m 2m–1 m
a0 (x + 1) + a1(x + x ) + … + am x = 0.
m
Divided by x , this may be written in the form:
m m–1
a0 (x + 1m ) + a1(x + m1−1 ) + … +am (x+ 1 )+ am = 0,
x x x
and the substitution of y for x + 1 reduces it to an equation of
x
degree m in y.
b) If the equation is of even degree, n=2m; the same substitution
is used without the removal of a factor. i.e. y = x + 1 to reduces it
x
to an equation of degree m in y.
2 2
It is easy to see that, if x + 1 = y , x + 12 = y – 2 ,
x x
3 1 3 1 4 4 2
x + 3 = y – 3 y , x + 4 = y – 4 y + 2, etc.
x x
(ii) a r = – a n–r: i.e. the coefficients of the corresponding terms

taken from the beginning and end are equal in magnitude but
opposite in sign.
a) If the equation is of odd degree, n = 2m+1; it is satisfied by
x = 1; and when the factor (x – l) is removed, there remains a
reciprocal equation of even degree 2m of the form (type (i) a).
b) If the equation is of even degree, n = 2m; it is satisfied by x =
2
1 and x = –1; and when the factor x – 1 is removed there remains

179
Algebra (Chapter 6) Theory of equations

a reciprocal equation of even degree of type (i).

It is clear that in a reciprocal equation the roots may be grouped in
pair; for if α is a root so is 1/α.
Every reciprocal equation can be reduced to a reciprocal equation
of even degree in which the coefficients of the first and the last
terms are equal (of type i).
Example 6 .27:
5 4 3 2
Solve 12 x – 8 x – 45 x + 45 x + 8 x – 12 = 0.
Solution:
This is reciprocal equation of odd degree (type (ii) a). Hence x – 1
is a factor of the left hand side. Dividing by ( x – 1), we get
12 –8 –45 45 8 –12 (1
12 4 –41 4 12
4 –41 4 12 0
4 3 2
i.e. 12 x + 4 x – 41 x + 4 x + 12 = 0
2
Dividing by x , we get
2
12(x + 12 ) + 4 (x + 1 ) – 41 = 0
x x
2 2
2 2
Put y = x + 1 , hence y = x + 12 + 2. i.e. x + 12 = y – 2.
x x x
2
Therefore we have 12(y – 2) + 4 y – 41 = 0.
2
i.e. 12 y + 4 y – 65 = 0 ⇒ (2 y + 5)(6y – 13) = 0
Hence y =13/6 or – 5/2.
2
If y = 13 : x + 1 = 13 , ⇒ 6 x – 13x + 6 = 0 ⇒ (2x – 3)(3x – 2)=0.
6 x 6
2
If y =– : x + = – 5 , ⇒ 2x + 5x + 2 = 0 ⇒ (2x + 1)(x + 2)=0.
5 1
2 x 2

180
Algebra (Chapter 6) Theory of equations

The roots of the original equation are – 2 , – 1 , 2 , 1 , 3 .

2 3 2

Example 6 .28:
6 5 4 2
Solve 6 x – 25 x + 31 x – 31 x + 25 x – 6 = 0.
Solution:
This is reciprocal equation of even degree (type (ii) b). Hence
2
x – 1 is a factor of the left hand side. Dividing by ( x – 1), we get
6 –25 31 0 –31 25 –6 (1
6 –19 12 12 –19 6
– 19 12 12 –19 6 0 (–1
– 6 25 –37 25 –6
6 – 25 37 –25 6 0
2 4 3 2
i.e. (x – 1)(6 x – 25 x + 37 x – 25 x + 6) = 0
4 3 2
∴ 6 x – 25 x + 37 x – 25 x + 6 = 0
2
Dividing by x , we get
2
6(x + 12 ) – 25 (x + 1 ) + 37 = 0
x x
2
2 22
Put y = x + 1 , hence y = x + 12 + 2. i.e. x + 12 = y – 2.
x x x
2
Therefore we have 6(y – 2) – 25 y + 37 = 0.
2
i.e. 6 y – 25 y + 25 = 0 ⇒ (3 y – 5)(2y – 5) = 0
Hence y = 5 or 5 .
3 2
2 5 ± i 11
If y = 5 , x + 1 = 5 , ⇒ 3 x – 5x + 3 = 0 i.e. x = .
3 x 3 6
2
If y = 5 : x + 1 = 5 , ⇒ 2x – 5x + 2 = 0 ⇒ (2x – 1)(x – 2)=0.
2 x 2
5 ± i 11
The roots of the original equation are 1, – 1 , , 2, – 1 .
6 2

181
Algebra (Chapter 6) Theory of equations

Example 6 .29:
Find the equation whose roots are those of
4 3 2
12 x – 44 x + 19 x + 50 x – 25 = 0.
diminish by 1, and hence solve the equation.
Solution:
12 –44 19 50 –25 (1
12 –32 –13 37
–32 –13 37 12
12 –20 –33
–20 –33 4
12 –8
– 8 –41
12
4
4 3 2
The transformed equation is 12y + 4 y – 41 y + 4 y + 12 = 0.
This is reciprocal equation, writing z = y + 1/y, this equation
2
becomes 12(z – 2) + 4 z – 41 = 0.
2
i.e. 12 z + 4 z – 65 = 0 ⇒ (2 z + 5)(6z – 13) = 0
Hence z =13/6 or – 5/2.
2
If z = 13 : y + 1 = 13 , ⇒ 6 y – 13y + 6 = 0 ⇒ (2y – 3)(3y – 2)=0.
6 y 6
2
If y =– 5 : y + 1 = – 5 , ⇒ 2y + 5y + 2 = 0 ⇒ (2y + 1)(y + 2)=0.
2 y 2
4 3 2
The roots of the equation 12y + 4 y – 41 y + 4 y + 12 = 0
are – 2 , – 1 , 2 , 3 .
2 3 2
Therefore the roots of the original equation
4 3 2
12 x – 44 x + 19 x + 50 x – 25 = 0, are –1 , 1 , 5 , 5 .
2 3 2

182
Algebra (Chapter 6) Theory of equations

EXERCISES 6
1) Construct the equations whose roots are
i) 1, 2, 3; ii) 0, 1, 2, 3 ; iii) ±1, ±2 , 3 ;
iv) –1/2, –1/4, ¾ ; v) 3 + 2 I; vi) 2 + 3 + i.
3 2
[Answers: i) x – 6 x + 11x – 6 = 0,
4 3 2
ii) x – 6 x + 11x – 6 x = 0,
5 4 3 2
iii) x – 3 x – 5 x + 15 x + 4x – 12 = 0,
3
iv) 32 x – 14 x – 3 = 0,
4 2
v) x – 2 x + 25 = 0,
8 6 4 2
vi) x – 16 x + 88 x + 192 x + 144 = 0 .]
2) Construct the equation, which has 4 roots equal to 1 and 3 roots
4 3
equal –1. [Answer: (x – 1) (x + 1) = 0.]
3 2
3) If α, β, γ are the roots of the equation x – 4x + 2 x – 1 = 0,
2 2 2 3 3 3
find the values of (i) α + β + γ ; (ii) α + β + γ ;
1 1 1 1 1 1
iii) + + ,iv) + + .[Answers: i)12, ii) 43, iii) 2, iv)–4]
α β γ α 2 β2 γ 2
3 2
4) Solve 12 x – 28 x + 3 x + 18 = 0, given that it has two equal
roots. [Answers: 3/2 , 3/2 , –2/3]
5 4 3 2
5) Solve 3x + 2 x + x – 6 x – 5 x – 4 = 0, given that ω is a
2 2
repeated root. [Answers: ω, ω , ω, ω , 4/3]
3 2
6) Solve the equation 3 x – 13x – x + 6 = 0, has the sum of two
of its roots equal to five. [Answers:–2/3 , 1 (5 ± 13 )]
2
4 3 2
7) Solve the equation x – 3 x – 5x + 17x – 6 = 0, has the sum of
two of its roots equal to five. [Answers: 2 , 3 ,–1 ± 2 ]
4 3 2
8)Solve x – 6x + 11x – 2x – 10 = 0, if (1– 3 ) is a given root to
it. [Answers: 1 ± 3 , 2 ± i]

183
Algebra (Chapter 6) Theory of equations
3 2
9) If α, β, γ are the roots of the equation x – 4x – x + 3 = 0, find
2 2 2 3 3 3
the values of (i) α + β + γ ; (ii) α + β + γ ;
1 1 1 1 1 1
iii) + + , iv) + + .
α β γ α 2 β2 γ 2
[Answers: i) 18, ii) 67, iii) 1 , iv) 25 ]
3 9
3 2
10) If f(x) = x –7x + 17x – 15, find f(x + 3) then solve f(x) = 0.
3 2
[Answers: f(x + 3)= x + 2x + 2x ; 3 , 2 ± i]
5 4 2
11) Solve x – x + 8 x – 9 x – 15 = 0, given that 3 , 1–2i are
two roots of it. [Answers: –1, ± 3 , 1 ± 2i]
3 2
12) Solve x –11x + 37x – 35 = 0, given that one root is 3+ 2 .
[Answers: 3± 2 , 5]
4 3 2
13) Solve x – 10x + 26x – 10x + 1 = 0, given that one root is
2 – 3. [Answers: 2± 3 , 3± 2 2 ]
3 2
14) Solve 4x – 24 x + 23 x + 18 = 0, given that the roots are in
arithmetical progression. [Answers: –1/2, 2 , 9/2]
3 2
15) Solve x – 6 x + 3 x + 10 = 0, given the roots are in
arithmetical progression. [Answers: –1, 2, 5]
3 2
16) Solve 2 x + 6 x + 5 x + 1 = 0, given the roots are in
arithmetical progression. [Answers:–1± 1/ 2 , –1]
3 2
17) Solve 2x – 21 x + 42 x – 16 = 0, given the roots are in
geometrical progression. [Answers: 2, ½, 8]
3 2
18) Find m in the equation 3 x – 26 x + m x – 24 = 0, if the roots
are in geometrical progression and solve it.
[Answers: m = 52 , 2/3 , 2 , 6]
4 2
19) Find c in the equation x – 2 x + c = 0, if it has one repeated
root and solve it. [Answers: c = 0; 0, 0, ± 2 & c = 1; 1, 1,–1,–1]
3 2
20) Find the values of c such that the equation x + 3x + c = 0,

184
Algebra (Chapter 6) Theory of equations

has one repeated root, then solve it.

[Answers: c = 0 ; 0, 0, –3 & c = – 4 ; –2 , –2 , 1]
3 2
21) Solve the equation x – 21x + 126 x – 216 = 0, whose roots
are in geometrical progression. [Answers: 3, 6, 12]
3 2
22) Solve x – 7 x + 36 = 0, given that one root is double another.
[Answers: –2, 3, 6]
3 2
23) Solve x – 5 x – 2 x + 24 = 0, given that the product of two of
the roots is 12. [Answers: –2 , 3 , 4]
4 3 2
24) Solve x + x – 16x – 4 x + 48 = 0, given that the product of
two of the roots is 6. [Answers: ±2 , 3 ,–4]
4 3 2
25) Solve 8x + 4 x – 18 x + 11x – 2 = 0, if it has three equal
roots [Answers: ½ , ½ , ½ , –2]
3 2
26) Solve 4 x + 4x – 7 x + 2 = 0, if it has repeated root.
[Answers: ½,½,–2]
3 2
27) Solve 3x – 7 x – 36 x – 20 = 0, given that the sum of two of
the roots is 3. [Answers: –2/3, –2, 5]
4 3 2
28) Solve x + 5x – 7 x – 29 x + 30 = 0, given that the product of
two of the roots is 15. [Answers: –5, –3, 1 , 2]
3 2
29) Solve x – 4x – 3 x + 18 = 0, given that two of the roots are
equal. [Answers: –2, 3 , 3]
4 3 2
30) Increase the roots of x + 4x – 7x – 22 x + 24 = 0, by 1;
4 2
hence solve this equation. [Answers: x –13x +36 = 0; –4,–3, 1 , 2]
3 2
31) Solve x – x – 14 x + 24 = 0, given that the roots are rational.
[Answers: – 4 , 2 , 3]
3 2
32) Given that the roots of 6x + 19 x + x – 6 = 0 are rational,
multiply the roots by some constant k so as to obtain an equation
whose roots are integral. Hence solve the equation.

185
Algebra (Chapter 6) Theory of equations
3 2
[Answers: k = 6 ; z + 19 z + 6 z – 216 = 0 ; ½ , –2/3 ,– 3]
2
33) Use the transformation x + 3 x = y, to solve
2 2 2
i) (x + 3 x + 3) = 2x + 6 x + 5,
2 2 2
ii) (x + 3 x – 1)(x + 3 x – 3) + 2x + 6 x – 11= 0.
[Answers: i) –1, –1, –2, –2 , ii) – 4, –2, –1, 1]
2 2
34) Solve (x + 5 x + 3)(x + 5 x – 5) = 9. [Answers: – 6, –4, –1, 1]
35) Solve the following equations:
i) x(2x+1)(x–2)(2x+5)+21=0,[Answers: 1,–3/2, 1 (–1± 57 )]
4
ii) x(2x+1)(x–2)(2x–3)=63, [Answers: 3,–3/2 , 1 (3± 47 i)]
4
iii) x(2x+3)(2x–1)(x+2)=63, [Answers:–3, 3/2 , (–3± 47 i)]
1
4
iv) (x–7)(x–3)(x+5)(x + 1) = 105, [Answers: 0, 2 , (1± 39 )]
v) (x–7)(x+6)(x–5)(x+ 4)= 720, [Answers: –1,2, 1 (1± 241 )]
2
vi) x(3x–1)(3x+2)(x+1) = 30,
[Answers: 1 (1± 19 ), 1 (–1± 14 i)]
3 3
vii) (x–2)(x–3)(x+ 4)(x +5)= 2640. [Answers: 7, –9,–1±i 39 )]
3 2
36) One root of x + x – 24 x + 16 = 0 is an integer. Find the
other two roots correct to 2 places of decimals.
[Answers: 4 , 0.70, –5.70]
4 3 2
37) Solve x – 2x – 6x – 2x + 1 = 0 [Answers: –1,–1, 2± 3 ]
4 3 2
38) Solve x –5x –12x – 5x + 1 = 0 [Answers: –1,–1, 1 (7±3 5 )]
2
4 3 2
39) Solve x – 10x + 26x – 10x+1= 0. [Answers: 2± 3 , 3± 2 2 ]
3 2
40) Solve 3 x + 4x + 4 x + 3 = 0. [Answers: –1, 1 (–1 ± 35 )]
2
5 4 3 2
41) Solve 6 x – 31 x + 62x – 62x + 31x – 6 = 0.
[Answers: 1, 2, ½, 1 (5± 11 i)]
6
6 5 4 2
42) Solve 6 x – 25 x + 31 x – 31x + 25 x – 6 = 0.
186
Algebra (Chapter 6) Theory of equations

[Answers: 1,–1, 2, ½, 1 (5± 11 i)]

6
4 3 2
43) Solve x – 2 x – 6 x – 2x + 1 = 0. [Answers: –1,–1, 2 ± 3 ]
4 3 2
44) Solve x – 2 x – 33 x – 2x + 1 = 0.
[Answers: 1 (7± 45 ), 1 (–5± 21 )]
2 2
4 3 2
45) Solve 6x – 5 x – 38 x – 5x + 6 = 0. [Answers: – 1 ,–2, 3 , 1 ]
2 3
5 4 3 2
46) Solve 6x – x – 43 x + 43 x + x – 6 = 0.
[Answers: 1 , 2 , –3 , – 1 , 1]
2 3
5 4 3 2
47) Solve 6x + 29 x + 27 x – 27 x – 29x – 6 = 0.
[Answers:– 1 ,–2,–3,– 1 , 1]
2 3
5 4 3 2
48) Solve 2x + 3 x – x + x – 3x – 2 = 0. [Answers:– 1 ,–2,i,–i , 1]
2
6 5 4 2
49) Solve 4x – 4x – 11 x + 11 x + 4 x – 4 = 0.
[Answers: 1, –1, 2 , 1 , 1 (–3± 7 i)]
2 4
6 5 4 2
50) Solve x + x – 5 x + 5 x – x – 1 = 0.
[Answers: 1, –1, 1, 1, 1 (–3± 5 )]
2
3 2
51) Solve 5x – 4 x + 4x – 5 = 0. [Answers: 1, (–1± 26 )]
1
10
3 2
52) Solve 7 x + 6 x + 6x + 7 = 0. [Answers: –1, (1± 195 i)]
1
14
5 4 3 2
53) Solve 6x –31 x + 62 x – 62 x + 31x – 6 = 0.
[Answers: 1, 2, 1 , 1 (5 ± i 11 )]
2 6
6 5 4 2
54) Solve 6x – 25x + 31 x – 31 x + 25 x – 6 = 0.
[Answers: 1, –1, 2, 1 , 1 (5 ± i 11 )]
2 6
3 2
55) Diminish the roots of 2x – 15x + 31x – 12 = 0, by 3. Hence
3 2
solve the equation. [Answers: 2x + 3x – 5x = 0; 4 , 3 , ½ ]
3 2
56) Diminish the roots of 2x – 15x + 31x – 12 = 0, by 4. Hence

187
Algebra (Chapter 6) Theory of equations
3 2
solve the equation. [Answers: 2z + 9z + 7z = 0, 4 , 3 , ½ ]
57) Solve by Cardan’s method the following equations:
3 2
i) x – 3x – 3x + 9 = 0 , [Answers: 3 , ± 3 ]
3
ii) x – 9x + 28 = 0 , [Answers: –4 , – 3– 4ω , 1 – 2ω]
3 2
iii) x – 15x – 33x + 847 = 0, [Answers: –7 , 11 , 11]
3 2 2 2
iv) x + x – 9x + 12 = 0, [Answers: –4 , –ω – 2ω , –ω –2ω]
3 2 2
v) x – 18x + 35 = 0, [Answers:–5 , –2ω – 3ω , –2ω –3ω]
3 2 2
vi) x – 6x – 9 = 0, [Answers: 3 , ω + 2ω , ω + 2ω]
3 2
vii) x + 3x – 24 x + 28 = 0, [Answers: –7, 2 , 2]
3 2
viii) x – 3x – 9 x + 27 = 0. [Answers: –3, 3 , 3]
58) Solve by Ferrari’s method the following equations:
4 3 2
i) x + 4 x + 8 x + 7 x + 4 = 0.
3 3 7
[Answers: λ=5/2, α = ±1, β=± 3/2; roots are – 1 ± i, – ± ]
2 2 2 2
4 3 2
ii) x – 2 x – 7 x + 8 x + 12 = 0.
[Answers: λ = 4, α = ±4, β =∓2; roots are 2 , 3 ,–2 , –1]
4 3 2
59) Solve by Ferrari’s method x + 2x – 12 x – 10 x + 3 = 0.
[Answers: λ = –2, α = ±3, β =±1; roots are – 1, 3, –2 ± 5 ]
6 4 2 2
60) Prove that f(x) = x – 7x + 15 x – 9 is divisible by x –1, and
that the quotient is a perfect square. Hence solve the equation
f(x) = 0, and write down the equation whose roots are the
reciprocals of the roots of this equation.
2 4 6
[Answers: roots are ±1 , ± 3 , ± 3 ; 1 – 7y + 15 y – 9 y = 0]
4 3 2
61) Increase the roots of x + 4x – 19 x – 106x – 120 = 0 by 3.
Hence solve the equation.
4 3 2
[Answers: y – 8y – y + 8 y = 0; –4, –3, –2, 5]
3 2
62) Solve by Cardan’s method x + 3x – 3x – 9 = 0 .

188
Algebra (Chapter 6) Theory of equations

[Answers: –3 , ± 3 ]
4 3
63) Solve by Ferrari’s method x + 4x – 11 x – 4 = 0.
[Answers:λ=–3/2,α=±1,β=±5/2;roots are 1 (–3± 5 ), 1 (–1 ± 17 )]
2 2
4 3 2
64) Solve by Ferrari’s method x + 4x – 6x + 20x + 8 = 0.
[Answers: λ = 3, α= ±4, β = ∓1 ; roots are –3 ± 7 , 1 ± i 3]
4
65) Solve by Ferrari’s method x – 12 x – 5 = 0.
[Answers: λ = 3, α= ± 4, β = ± 1 ; roots are 1 ± 2 , –1 ± 2i]
4 2
66) Solve by Ferrari’s method x – 18 x + 16 x – 3 = 0 .
[Answers: λ = –1, α= ± 4, β = ∓ 1 ; roots are –2 ± 7,2± 3]
67) Solve by Ferrari’s method
4 3 2
i) x – 2 x – 9 x + 6x + 9 = 0.
[Answers: λ=–3, α=± 2, β= 0 ; roots are 1 (3± 21 ), 1 (–1± 13 )]
2 2
4 3 2
ii) x + 5x – 7 x – 29 x + 30 = 0.
[Answers: λ = 8.5; α=± 5.5, β= ± 6.5 ; roots are –5, –3, 1 , 2]
68) Solve:
6 5 4 2
i) 3x + x – 27 x + 27 x – x – 3 = 0.
[Answers: ±1, –3, –1/3 , 1 (3 ± 5 )]
2
5 4 3 2
ii) 2x – 15 x + 37x – 37x + 15x – 2 = 0.
[Answers: 1, 2, 1 , 2 ± 3 ]
2
69) Form the equation whose roots exceed by unity those of
5 4 3 2
x – x – 7 x + 2x + 10 x + 4 = 0. Hence solve it.
5 4 3 2
[Answers: y –6y +7y +7y – 6 y + 1 = 0; –2, 1 ± 3 , 1 (1 ± 5 )]
2
2 3 2 2
70) Solve 4(x – x + 1) – 27 x (x – 1) = 0.
[Answers: 2, 2, 1 , 1 , –1 , –1 ]
2 2

189
Algebra (Chapter 7) Calculation Methods

Chapter 7
Calculation Methods
In this chapter we discuss theorems, which will determine
the nature of the roots, and their position when real. The nature of
the roots is determined when we know how many are imaginary
and how many are real.
The position of a real root is determined by its position in
the scale of real numbers. Thus, e.g. as a first approximation we
might determine two consecutive integers between which a root
lies. It will be seen, however, that the theorems considered in this
chapter do provide us in many cases with a good deal of
information.
Descartes' Rule of Signs:
Consider a polynomial whose terms are arranged in the
n n–1 n–2
form f(x) ≡ a0x + a1x + a2x + … + an–1x + an.
i.e. in descending powers of x. Then a change of sign is said to
occur when any particular term has the opposite sign to the term,
which immediately precedes it. Thus, e.g. in the polynomial
4 3 2
5x + 7 x – 3 x + 9x – 2,
we can represent the sequence of signs as
+ + – + –
2
Here there are three changes in sign, for – 3 x is of opposite sign
3 2
to 7 x , 9x is of opposite sign from – 3 x , and –2 is of opposite

190
Algebra (Chapter 7) Calculation Methods

sign to 9x. The three changes are + to – ; – to + ; and + to –.

Now let α be any positive real number, f (x) any polynomial. Then
the product (x – α) f (x) when expanded has at least one more
change of sign than f (x). We may now deduce the rule of signs.
Theorem 7.1: Descartes’ Rule of signs:
The equation f(x) = 0 cannot have more positive roots than there
are changes of sign in f(x) or more negative roots than there are
changes of sign in f(–x).
i.e. (i) The number of positive roots cannot exceed the number of
change of signs from positive to negative and from negative to
positive in f(x).
(ii) The number of negative roots cannot exceed the number of
variations in f(–x).
Detection of Imaginary Roots from Descartes’ Rule:
Let f (x) be a polynomial of degree n and suppose that f (x)
has r changes and f (–x) has s changes in sign. Then if r + s < n,
we conclude that f(x) = 0 has at least n – r – s imaginary roots.
It will be seen in what follows that the method is of service
only if some of the terms are missing from f (x), i.e. some of the
coefficients are zero.
Example 7.1:
7 4 3
Prove that the equation x – 3x + 5x – 2 = 0 has at least four
imaginary roots.

191
Algebra (Chapter 7) Calculation Methods

Solution:
7 4 3
Write f(x) ≡ x – 3x + 5x – 2. The number of changes in sign in
f(x) are 3. Hence there are at most three positive roots.
7 4 3 7 4 3
Again f(–x) =(–x) –3(–x) + 5(–x) –2 =–x – 3x – 5x – 2.
Since there are no changes in sign in f(–x) it follows that
there are no negative roots.
Thus there are at most three real roots, i.e. at least four imaginary
roots.
Let p denote the number of changes in sign in f (x), q the
number of changes in sign of f(–x), h the number of positive roots,
k the number of negative roots of f(x)= 0.
Then if none of the coefficients are zero it is clear that the
number of changes in sign in f(x) plus the number of changes in
sign in f(–x) is exactly n. In this case p + q = n.
Again, if some of the coefficients are zero it is clear that
the number of changes in sign cannot exceed n. Hence in general
p + q ≤ n.
Also from Descartes’ rule, h ≤ p, k ≤ q. So that
h+k≤ p+q≤n (i)
If all the roots of the equation f(x) = 0 are real then h = p, k = q.
In this case h + k = n and hence from (i): h + k = p + q = n.
Suppose that h < p so that k > q. This implies that the
number of negative roots is greater than the number of changes in

192
Algebra (Chapter 7) Calculation Methods

sign of f(–x), a result which contradicts Descartes’ Rule.

Since h ≤ p it follows that h = p. Similarly k = q.
Some General Theorems:
Let f(x) denotes the polynomial
n n–1 n–2
f(x) ≡ a0x + a1x + a2x + … + an–1x + an , a0 > 0 , (7.1)
while the letters a , b will denote real numbers.
Theorem 7.2:
If f(a) and f(b) have opposite signs, then there exists at least one
real root of f (x) = 0 between a and b.
3 2
For example let f(x) = (x – 1)(x – 3)(x – 5) = x – 9 x + 23x – 15.
It is easy to see that
x 0 1 2 3 4 5 6
f(x) –15 0 +3 0 –3 0 +15
Since f(x) is a polynomial it is continuous for all values of x.
Hence f(x) can only change sign by passing through the value
zero, i.e. by passing through a real root of f (x) = 0.
Theorem 7.3:
If n is odd the equation f(x) = 0 has at least one real root whose
sign is opposite to that of an, the last term.
3 2
For example let f(x) = x – 1 = (x – 1)(x + x + 1).
It has real root x = 1 , a3 = –1.
3 2
If g(x) = x + 1 = (x + 1)(x – x + 1).
It has real root x = –1 , a3 = 1.

193
Algebra (Chapter 7) Calculation Methods

Theorem 7.4:
If n is even and an < 0 then f(x) = 0 has at least two real roots, one
positive and one negative.
4 2
For example let f(x) = x – 1 = (x – 1)(x + 1)(x + 1).
It has real roots x = ± 1 , a4 = –1 < 0.
4
Let g(x) = x + 1, g(x) has no real root.
Theorem 7.5:
If f(a) and f(b) have opposite signs then there are an odd number
of real roots between a and b: If f(a) and f (b) have the same sign
then either there is an even number of real roots between a and b
or there are no real roots at all between a and b .
Theorem 7.6:
The largest root of the equation (7.1) may be obtained
approximately by the root of the equation a0x + a1=0, or by the
root of the quadratic expression such that it is larger in absolute
value.
3 2
For example let f(x) = (x – 1)(x – 3)(x – 5) = x – 9 x + 23x – 15.
The largest root is 5. Take the equation x – 9 = 0 ⇒ x = 9, which
can be taken as approximated value to the root x = 5.
§7.2: Approximate Methods:
There are several ways of obtaining approximate values of the
irrational roots of equations having numerical coefficients; we
shall illustrate some of these methods by examples.
194
Algebra (Chapter 7) Calculation Methods

§7.2.1: Methods to Algebraic equations only:

Horner’s Method of Approximation:
The method applies whether the roots can be represented by a
terminating decimal or not and is concerned in the first instance
with positive roots. The procedure is to determine the root, figure
by figure. Thus the integral part is first determined; then the first
decimal place is determined, then the second, and so on, until the
root terminates or the root has been obtained to the required
degree of approximation. The methods involved have already been
considered. The two transformations required are
(a) Diminution of the roots of an equation, and
(b) Multiplication of the roots of an equation by 10.
The first step is to determine between what two integers
the interested, lies. Having determined the integral part of the root
we now transform the given equation by reducing the roots by this
integral value. In the new equation the root must lie between 0 and
1. In order to avoid decimals in the working, the transformed
equation now has its roots multiplied by 10 so that the figure
required lies between 0 and 10. This integral value is determined,
and thus the first decimal figure of the required root is determined.
Proceeding in this way we may obtain any number of decimal
places or else the root terminates.
The method and presentation will be clear from the examples

195
Algebra (Chapter 7) Calculation Methods

given below.
Example 7.2:
3 2
Find an approximate solution of x + 4 x – 5 x – 20 = 0.
Solution:
3 2
Let ϕ(x) = x + 4 x – 5 x – 20. It is easy to see that ϕ(2) = – 6< 0,
ϕ(3)= 28 > 0. Hence there is a root between 2 and 3. i.e. the root is
2+ a . Diminish the roots of the equation by 2; we obtain a new
10
equation with a root between 0 and 1.( i.e. 0 < a < 1).
10
3 2
The working may be set out as the following: x + 4x –5 x–20 = 0
1 4 –5 – 20 (2
2 12 14
6 7 – 6
2 16
8 23
2
10
3 2
Therefore the transformed equation is y + 10 y + 23y – 6 = 0. It
has a root between 0 and 1, i.e. (i.e. 0 < a < 1).
10
Multiply the roots of this equation by 10; it follows that the
3 2
transformed equation f(y) = y + 100 y + 2300 y – 6000 = 0.
It has a root between 0 and 10 (i.e. 0 < a < 10). By trial, we find
that: f(2) = –992 < 0, f(3) = 1827 > 0.
i.e. the root lies between 2 and 3, i.e. 2 < a < 3, i.e. a = 2+ b .
10
[[If we stopped here, it is evident that the first approximation is

196
Algebra (Chapter 7) Calculation Methods

given by 2300y–6000 = 0, i.e. a = y = 6000/2300 = 60/23 ≅ 2.608

the root is x ≅ 2 + a ≅ 2.2608.]]
10
The process can be continued as before. Since f(2) < 0, f(3) > 0 ,
then diminish the roots of the equation by 2.
1 100 2300 –6000 (2
2 204 5008
102 2504 – 992
2 208
104 2712
2
106
3 2
∴ The transformed equation y + 106 y + 2712 y – 992 = 0, has a
root between 0 and 1, i.e. 0 < b < 1 or 0 < b < 10.
10
Multiplying the roots by 10, it follows that
3 2
g(y) = y + 1060 y + 271200 y – 992000 = 0.
It has a root between 0 and 10. By trial, we find that
g(3) = –168833 < 0 , g(4) = 109824 > 0,
i.e. it has a root lies between 3 and 4, i.e. 3 < b < 4 ,or b = 3 + c .
10
Diminish the roots of the equation by 3.
1 1060 271200 –992000 (3
3 3189 823167
1063 274389 –168833
3 3198
1066 277587
3
1069
3 2
i.e. y + 1069 y + 277587 y – 168833 = 0.

197
Algebra (Chapter 7) Calculation Methods

It has a root between 0 and 10 i.e. 0 < c < 10. Multiplying the
roots by 10, it follows that
3 2
h(y) = y + 10690 y + 27758700 y – 168833000 = 0.
It has a root between 0 and 10. By trial, we find that
h(6) = –189574 < 0 , h(7) = 26002053 > 0,
i.e. it has a root lies between 6 and 7, i.e. 6 < c < 7 ,or c = 6 + d .
10
Diminish the roots of the equation by 6.
1 10690 27758700 –168833000 (6
6 64176 166937256
10696 27822876 – 1895744
6 64212
10702 27887088
6
10708
3 2
y + 10708 y + 27887088 y – 1895744 = 0,
It is easy to see that the root lies between 0 and 1. The process can
then be continued as before, but it is evident that a good
approximation is given by 27887088 y – 1895744 = 0
d ≅ y = 1895744/27887088 = 0.068
Since the root is 2+ a , a = 2+ b , b = 3 + c , c = 6 + d ,
10 10 10 10
d = 0.068
∴ a root of the original equation is x ≅ 2+ a ≅ 2.236068
10
It is easy to verify that the roots of the original equation are ± 5 ,
– 4. But we are seeking the root 5 ≅ 2.236067.
If Horner’s method is used to find the approximate value

198
Algebra (Chapter 7) Calculation Methods

of a negative root, the equation should first be transformed by

putting x = –y; the method is then applied to find the
corresponding positive root of the transformed equation.
§7.2.2: Simplifications in the case of
Horner’s Method:
Example 7.2 shows that if it is necessary in each case to determine
the required figure by actual substitution, the numerical work
involved may be considerable. This is, however, unnecessary and
one of the practical advantages of Horner’s method is that after the
second or third figure of the root has been determined the
transformed equation itself suggests by inspection the next figure
for the root. This is called the principle of the trial-divisor. The
second last coefficient of each transformed equation (after the
roots have been multiplied by 10) is called the trial divisor. This
divided into the constant term give the suggested value.
(b) It is clear that beyond a certain stage in the working the second
last coefficient and the constant term must have opposite signs; for
as the process of evolution of the root proceeds the influence of
the other terms becomes smaller and smaller. After the first
transformation, i.e. after the equation has had its roots reduced by
the integral part of the root under consideration, the absolute term
must not change sign. The sign may change at the first
transformation. Thus, e.g. the equation

199
Algebra (Chapter 7) Calculation Methods
2 3 2
(x – 3)(2x – 7) = 2x – 7x – 6 x + 21 = 0
has a root between 3 and 4. Reducing the roots by 3 we obtain
3 2
2x + 11x + 6 x – 6 = 0.
The reason for the change in sign is that we have passed
over another root of the equation, viz. that between 1 and 2. But
assuming that the original equation has only one root between 3
3 2
and 4, the transformed equation 2x + 110 x + 600 x – 6000 = 0,
has one and only one root between 0 and 10. Thus further
transformations of the type considered will not affect the sign of
the constant term. We have thus the following rule. The figure to
be taken is the highest number, which in the process of
transformation will leave, unchanged the sign of the constant term.
If the number taken were too small the next figure suggested
would be greater than 9, as in ordinary division.
Example 7.3:
3
Find an approximate solution of x + 3x – 7 = 0.
Solution:
1 0 3 –7 (1
As in Example 7.2, we see that
1 1 4
there is a root between 1 and 2. 1 4 –3
i.e. the root is 1+ a . Diminish 1 2
10 2 6
the roots of the equation by 1; 1
3
we obtain a new equation with a root between 0 and 1.

200
Algebra (Chapter 7) Calculation Methods

i.e. 0< a < 1.

10
Multiply the roots of this equation by 10; it follows that the
3 2
transformed equation f (y) = y + 30 y + 600 y – 3000 = 0,
( i.e. 0 < a < 10). It has a root between 0 and 10. By trial, we find
that: f(4) = –56 < 0, f(5) = 875 > 0
i.e. the root lies between 4 and 5, i.e. 4 < a < 5, i.e. a = 4+ b .
10
Diminish the roots of the equation by 4.
1 30 600 –3000 (4
4 136 2944
34 736 – 56
4 152
38 888
4
42
3 2
∴ The transformed equation y + 42 y + 888 y – 56 = 0,
has a root between 0 and 1, i.e. 0 < b < 1 or 0 < b < 10 .
10
Multiplying the roots by 10, it follows that
3 2
g(y) = y + 420 y + 88800 y – 56000 = 0,
has a root between 0 and 10. By trial, we find that
g(0) = –56000 < 0 , g(1) = 1 + 420 + 88800 – 56000 = 33221 > 0,
i.e. it has a root lies between 0 and 1, i.e. 0 < b < 1,or b = 0+ c .
10
Multiplying the roots by 10, it follows that
3 2
y + 4200 y + 8880000 y – 56000000 = 0,
has a root between 0 and 10 i.e. 0 < c < 10. By trial, we find that
this root lies between 0 and 1. The process can then be continued

201
Algebra (Chapter 7) Calculation Methods

as before, but it is evident that a good approximation is given by

8880000 y – 56000000 = 0, i.e. 888 y – 5600 = 0
that is, y ≅ 0.63.
∴ a root of the original equation is a = 4 + b = 4 + 0 + c ,
10 10 100 10 100 1000
c= 6.3 ∴ x ≅ 1+ a ≅ 1+ 0.4b ≅ 1.40c ≅ 1.4063.
10
§7.2.3: Roots nearly Equal:
The method of procedure in such cases is to separate the roots and
then proceed as before.
Example 7.4:
3 2
Equation ϕ(x) = x – 3x – 4x + 13 = 0, has two roots between 2
and 3. Find each of them correct to two decimal places.
Solution:
Diminishing the roots of the equation by 2 and multiplying
the roots of the new equation by 10 we obtain
1 –3 –4 13 (2
2 –2 –12
–1 –6 1
2 2
1 –4
2
3
3 2
Equation ϕ(x) = x + 3 x – 4 x + 1= 0, has two roots between 0
3 2
and 1. Therefore equation f(x) = x + 30 x – 400 x + 1000 = 0,
must have two roots between 0 and 10.
Now f(3) =97 >0, f(4)= – 56< 0; f (6)= – 104 < 0, f (7)= 13 > 0 .

202
Algebra (Chapter 7) Calculation Methods

It follows that the transformed equation has one root between 3

and 4 and another between 6 and 7. Thus the two roots are 2.3…,
2.6… .
Beginning by the first root, diminishing the roots of the equation
f(x) = 0, by 3 and multiplying the roots of the new equation by 10
we obtain
1 30 –400 1000 (3
3 99 –903
33 –301 97
3 108
36 –193
3
39
3 2
g (x) = x + 39 x – 193 x + 97 = 0,
g(0) = 97 > 0 , g(1) = –56 < 0 ; g(3) = –104 < 0, g(4) = 13 >0 .
3 2
Let h(x) = x + 390 x – 19300 x + 97000 = 0,
h(6) = 10375 , h(7) = – 4544.
Thus the root correct to two decimal places is 2.36
Now since f(6) < 0 , f(7) > 0, then the equation f(x) = 0, has
another root between 6 and 7. i.e. the root is 2.6…. Diminishing
3 2
the roots of the equation f(x) = x + 30 x – 400 x + 1000 = 0, by 6
and multiplying the roots of the new equation by 10 we obtain
1 30 –400 1000 (6
6 216 –1104
36 –184 – 104
6 252
42 68
6
48

203
Algebra (Chapter 7) Calculation Methods
3 2
G (x) = x + 48 x + 68 x – 104 = 0,
G(0) = –104 < 0 , G(1) = 13 > 0
3 2
H(x) = x + 480 x + 6800 x – 104000 = 0,
H(9) = –3191 , h(10) = 13000.
Thus the second root correct to two decimal places is 2.69.
§7.3: Methods to Algebraic and
transcendental equations:
Now we shall discuss few methods, which are applicable
to both algebraic and transcendental equations. Sometimes a rough
approximation of a root can be found by graph (1- graphical
method) and more accurate results by the following methods:
2- Method of halving the interval,
3- Iteration method (successive approximations),
4- Newton- Raphson’s Method (The tangent method),
5- False position method (Regula falsi method),
Special cases:
a) Fixed point method:
b) The secant (chord) method:
§7.3.1: Graphical Method:
To solve the equation f(x) = 0 graphically, find a small interval
[a , b] between which the root of the equation lies. i.e. f(a).f(b)< 0.
Prepare a table for the different values of x between a and b, and
y = f(x). We draw the graph of the function y = f(x) with respect

204
Algebra (Chapter 7) Calculation Methods

to x and y-axes and then obtain x-co-ordinates of those points for

which y-co-ordinates are zero. These x-co-ordinates will then
determine the roots of the equation f(x) = 0. In other words the
roots may be interpreted as the x-co-ordinates of the points of
intersection of the curve y = f(x) with x-axis. In case f(x) involves
difference of two functions, then we usually write f(x) = 0 as
g(x) = h(x) where g(x) and h(x) are both functions of x. To solve
the above equation, we draw the graphs of the two functions
y = g(x) and y = h(x) and the x-co-ordinates of the points of
intersection of the two curves will then determine the roots of the
given equation f(x) = 0.
Example 7.4:
Solve graphically the equation f(x) = sin x – x + 1= 0.
Solution:
Let us consider the two functions g(x) = sin x , h(x) = x – 1
To find a small interval between which the root of the equation
lies. It is easy to see that f(0) = 1 > 0 , f(π) = 0 – π + 1 < 0.
Consider the interval [0 , π].
x 0 0.3 0.6 0.9 1.2 1.5 1.8 1.85
g(x) 0 0.3 0.6 0.8 0.9 1 0.97 0.96
h(x) –1 –0.7 –0.4 –0.1 0.2 0.5 0.8 0.85

x 1.9 1.95 2 2.1 2.4 2.7 3 π

g(x) 0.94 0.92 0.91 0.86 0.68 0.43 0.14 0
h(x) 0.9 0.95 1 1.1 1.4 1.7 2 2.14

205
Algebra (Chapter 7) Calculation Methods

Draw the graphs of the two

functions y = x – 1 and y = sin x
on the same axes as shown in the
figure, and then find the abscissa
of the point of intersection of the
two curves. This gives the approximate value of the root as
x = 1.95.
Example 7.5:
Find the approximate value of the smallest real root of
x
f(x) = e sin x – 1 = 0.
Solution:
–x
Let us consider the two functions g(x) = sin x , h(x) = e .
To find a small interval between which the root of the equation
lies. It is easy to see that f(0) = –1 < 0, f(1) = 1.29 > 0.
Consider the interval [0 , 1]
x 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
g(x) 0 0.1 0.2 0.3 0.39 0.48 0.56 0.64 0.72 0.78 0.84
h(x) 1 0.9 0.82 0.74 0.67 0.61 0.55 0.5 0.45 0.41 0.37

We can rewrite the given

x
equation e sin x= 1 in the form
–x
sin x = e . Draw the graphs of
the two functions y = sin x and
–x
y=e on the same axes as

206
Algebra (Chapter 7) Calculation Methods

shown in the figure, and then find the abscissa of the point of
intersection of the two curves. This gives the approximate value of
the root as x = 0.58.
In the following we study numerical methods to find the
roots of an equation. Much of algebra is devoted to the “solution
of equations.” In simple situations, this consists of a
rearrangement to exhibit the value of the unknown variable as a
simple arithmetic combination of the constants of the equation.
For second-degree polynomials, this can be expressed by the
familiar quadratic formula. For third- and fourth-degree
polynomials, formulas exist but are so complex as to be rarely
used; for higher-degree equations it has been proved that finding
the solution through a formula is impossible. Most transcendental
equations (involving trigonometric or exponential functions) are
likewise intractable.
Even though it is difficult if not impossible to exhibit the
solution of such equations in explicit form, numerical analysis
provides a means where a solution may be found, or at least
approximated as closely as desired. Many of these numerical
procedures follow a scheme that may be thought of as providing a
series of successive approximations, each more precise than the
previous one, so that enough repetitions of the procedure
eventually give an approximation which differs from the true

207
Algebra (Chapter 7) Calculation Methods

value by less than some arbitrary error tolerance. Numerical

procedures are thus seen to resemble the limit concept of
mathematical analysis.
The first numerical procedure that we shall study is that of halving
the interval. The method, also known as the Bolzano-Weierstrass
method, is of ancient origin. Some authors call it the bisection
method.
§7.3.2: Method Of Halving The Interval:
To solve the equation f(x) = 0, find a small interval [a , b] between
which the root of the equation lies. i.e. f(a).f(b)< 0. Suppose that
the root α is isolated on this interval [a , b]. We shall assume that
f(x) is continuous on the interval [a, b]. For definiteness we shall
assume that f (a) < 0, f (b) > 0. If we divide the interval [a , b] in
half by the point c = ½(a + b), two cases may occur: (i) f(c) = 0 the
value of the function at the middle point of the interval [a, b] is
equal to zero (in that case the required root is obtained i.e. α = c),
(ii) f(c) ≠ 0 i.e. f(c) > or < 0. In this case one half of the closed
intervals [a , c] , [c , b] possessing the property f(a) f(c) < 0 or
f(c) f(b) < 0. If f(a) f(c) < 0, we take the interval [a , c], but if
f(c) f(b) < 0 we take the other interval [c , b]. We shall designate
this half as [a1 , b1]. It is evident that f (a1) < 0, f(b1) > 0. We shall
deal with the interval [a2, b2] in the same way as with the interval
[a , b], i.e. divide it in half. Proceeding with analogous reasoning,

208
Algebra (Chapter 7) Calculation Methods

we shall have two possibilities: (i) either the process described

above will terminate because in the middle of some interval the
value of the function will turn out to be zero (in that case the
required root x = α is obtained), (ii) or the process described can
be continued indefinitely and we shall receive a nested system of
closed intervals [a , b] ⊃ [a1,b1] ⊃ [a2,b2] ⊃ …, with f(an) < 0,
n
f(bn) > 0 for any number n. Since bn– an= (½) (b – a), the length
of the interval [an , bn] tends to zero as n→∞, then there exists a
point x0 contained in all the interval [an , bn], such that lim an =
lim bn = α. i.e. this nested system of closed intervals has one
common point α to which each of the sequences {an} and {bn}
converges.
Example 7.6:
Find the approximate value of the positive real root of the cubic
2 3 2
equation f(x) = (x – 3)(x + 2) = x + 2 x – 3x – 6 = 0.
Solution:
It is easy to see that f(1) = –6 , f(2) = 4. Then it is obvious that the
change in sign of the function between x = 1 and x = 2 guarantees
at least one root on the interval [1 , 2].
Suppose we now evaluate the function at the mid-point of
this interval x = 1.5 and compare the result to f(1) and f(2). Since
f(1.5) = –21/8 = –2.625 the function changes sign between x = 1.5
and x = 2, a root lies between these values. i.e. in the interval

209
Algebra (Chapter 7) Calculation Methods

[1.5, 2]. We can obviously continue halving the interval to

determine a smaller and smaller interval within which a root must
lie. For this example, continuing the process leads eventually to an
approximation to the root at x = 3 = 1.732050807568877….
3 2
Applying the method to f(x) = x + 2 x – 3x – 6 = 0, we get the
results of the following Table.
3 2
Table of halving the interval for f(x) = x + 2x – 3x – 6 = 0.
Maximum
n an cn bn f(an) f(cn) f(bn) error in cn
1 1 (1.5 2) -6 -2.625 4 0.5
2 (1.5 1.75) 2 -2.625 0.234375 4 0.25
3 1.5 (1.625 1.75) -2.625 -1.30273 0.234375 0.125
4 1.625 (1.6875 1.75) -1.30273 -0.56177 0.234375 0.0625
5 1.6875 (1.71875 1.75) -0.56177 -0.17068 0.234375 0.03125*
6 (1.71875 1.734375) 1.75 -0.17068 0.030087 0.234375 0.015625
7 1.71875 (1.72656 1.73437) -0.17068 -0.07077 0.030022 0.00781

*Actual error in α, after 5 iterations is 0.0133.

From the table the root = 1.72656 with error 0.0078.
The repetition of our algorithm is called iteration and the
successive approximations are termed “ the iterates”. The entries
in the Table indicate the necessity of representing values of the
argument, a, b, as well as of the function, f(x), only approximately
when we carry a limited number of decimal figures. The
distinction between numerical methods and numerical analysis is
that the latter term implies the consideration of errors in the
procedure used. Certainly the blind use of any calculation method
without concern for its accuracy is foolish.
210
Algebra (Chapter 7) Calculation Methods

In addition to the limitation on accuracy because we retain

only a limited number of figures in our work, there is an obvious
limitation if we terminate the procedure itself too soon. One
important advantage of the halving interval method, beyond its
simplicity, is our knowledge of the accuracy of the current
approximation to the root. Since a root must lie between the x-
values where the function changes sign, the error in the last
approximation can be no more than one-half the last interval of
which it is the midpoint, and this interval is known exactly since
the original difference, |bn – an|, is halved at each iteration. The
accuracy of a computed value is usually expressed either as the
absolute error (true value – approximate value) or as the relative
error (absolute error divided by true value). The relative error is
often the better measure of accuracy for very large or very small
values. Sometimes the accuracy is expressed as the number of
digits that are correct; in other cases, the number of correct digits
after the decimal point is used. When the true value is unknown, it
is impossible to express the accuracy with exactness, and
approximate accuracy must be specified. Frequently we will put
bounds on the size of the error.
Example 7.6:
Find the approximate value of the positive real root of the
x
equation f(x) = e – 2x – 1= 0.

211
Algebra (Chapter 7) Calculation Methods

Solution:
It is easy to see that f(1) < 0 , f(2) > 0. Then it is obvious that the
change in sign of the function between x = 1 and x = 2 guarantees
at least one root on the interval [1 , 2].
Suppose we now evaluate the function at the mid-point of
this interval x = 1.5 and compare the result to f(1) and f(2). Since
f(1.5) = – 0.01831the function changes sign between x = 1.5 and x
= 2, a root lies between these values. i.e. in the interval [1.5, 2].
We can obviously continue halving the interval to determine a
smaller and smaller interval within which a root must lie. For this
example, continuing the process leads eventually to an
approximation to the root at x = 1.256837 with error 0.0009765.
x
Table of halving the interval for f(x) = e – 2x – 1= 0.

Maximum
n an cn bn f(an) f(cn) f(bn)
error in cn

1 (1 1.5) 2 -0.28172 0.481689 2.389056 0.5

2 1 (1.25 1.5) -0.28172 -0.00966 0.481689 0.25
3 (1.25 1.375) 1.5 -0.00966 0.205077 0.481689 0.125
4 (1.25 1.3125) 1.375 -0.00966 0.090451 0.205077 0.0625
5 (1.25 1.28125) 1.3125 -0.00966 0.038638 0.090451 0.03125
6 (1.25 1.265625) 1.28125 -0.00966 0.014058 0.038638 0.015625
7 (1.25 1.257813) 1.265625 -0.00966 0.002093 0.014058 0.0078125
8 1.25 (1.253907 1.257813) -0.00966 -0.00381 0.002094 0.0039065
9 1.253907 (1.25586 1.257813) -0.00381 -0.00086 0.002094 0.001953
10 (1.25586 1.256837) 1.257813 -0.00086 0.000613 0.002094 0.0009765

212
Algebra (Chapter 7) Calculation Methods

§7.3.3: Iteration Method

(successive approximations):
This method is used when we can find the value of x in terms of a
function of x i.e. x = F(x) , from f(x) = 0.
In this method, we first obtain the approximate value of the root,
say x0, of the given equation and then substitute it in F(x) giving
the second approximation as x1= F(x0). Again putting x = x1, in
F(x), we obtain the third approximation as x2 = F(x1).Continuing
in this manner, we can get xn = F(xn–1).
In this way, we obtain a sequence of successive approximation x1,
x2,… which may converge to the desired root.
In order to obtain the root of f(x) = 0, i.e. x = F(x), find the
abscissa of the point of intersection of the line y = x with the carve
y = F(x) which may or may not intersect. In case the two curves
do not intersect then no real root exists of the equation f(x) = 0.
We shall say that the sequence x0, x1, x2, …, xn,… is
iterative if for any n ≥ 1 the term xn can be expressed through the
term xn–1 by means of the recurrence formula xn = F(xn–1), and
any number from the domain of
definition of the function F(x) is
taken as x0.
Remark 1. Suppose that the
function F(x) is continuous on the
closed interval [a, b] and assume
that all the terms of the iterative Figure 4

213
Algebra (Chapter 7) Calculation Methods

sequence x0, x1, x2, …, xn,… belong to that interval. Then, if that
sequence converges to a certain number α, the indicated number α
is a root of equation f(x) = 0.
Remark 2. If | F′(x)| ≤ k < 1, on some interval [a , b], then xn→α .
One of the points a and b can be taken, as x0. More specifically,
that of the points a and b should be taken as x0, for which the
approximation xl = F(x0) is confined to the closed interval [a, b].
Example 7.7:
Use the method of iteration to find the approximate value of the
root of the equation 2 – log x – x = 0 with an accuracy to within
0.00001.
Solution:
We find the isolation interval of the real
root of the equation, represent the given
equation in the form log x = –x + 2 and
construct the graphs of the functions
y = log x and y = –x + 2. The abscissa
of the point M of intersection of the
graphs lies in the interval [1, 2], therefore we can take x0 = 1 as
the initial value of x. Let us write the original equation as
x = 2 – log x. Here F(x) = 2 – log x,
F′ (x) = – (log e)/x, i.e. |F′(x)| < 1 in the interval [1, 2] and,
therefore, the iteration method is applicable here. We find the first

214
Algebra (Chapter 7) Calculation Methods

approximate value:
x1 = 2 – log x0 = 2 – log 1 = 2.
Then we find the second and the following approximations:
x2 = 2 – log xl = 2 – log 2 = 2 – 0.30103= 1.69897;
x3 = 2 – log 1.69897 = 2 – 0.23019 = 1.76981;
x4 = 2 – log 1.76981 = 2 – 0.24793 = 1.75207;
x5 = 2 – log 1.75207 = 2 – 0.24355 = 1.75645;
x6 = 2 – log 1.75645 = 2 – 0.24464 = 1.75536;
x7 = 2 – log 1.75536 = 2 – 0.24437 = 1.75563.
x8 = 2 – log 1.75563 = 2 – 0.24443 = 1.75557.
x9 = 2 – log 1.75557 = 2 – 0.24442 = 1.75558.
x10 = 2 – log 1.75558 = 2 – 0.24442 = 1.75558.
We can put the above results in the following table.
n xn F(xn) n xn F(xn)
1 1.00000 2.00000 6 1.75645 1.75536
2 2.00000 1.69897 7 1.75536 1.75563
3 1.69897 1.76981 8 1.75563 1.75557
4 1.76981 1.75207 9 1.75557 1.75558
5 1.75207 1.75645 10 1.75558 1.75558
Thus, the sought-for root is x ≅ 1.75558.
Example 7.8:
Find the approximate value of the positive real root of the cubic
2 3 2
equation f(x) = (x – 3)(x + 2) = x + 2 x – 3x – 6 = 0.
Solution:
It is easy to see that f(1) < 0 , f(2) > 0. Then it is obvious that the
change in sign of the function between x = 1 and x = 2 guarantees
at least one root on the interval [1 , 2].

215
Algebra (Chapter 7) Calculation Methods

To put f(x) = 0 in the form, x = F(x) , there are more than one way
in which this can be done. Some possibilities are
3 2
x = 1 [x + 2 x – 6] , x = 1 6 + 3x − x 3 , ( or – 1 6 + 3x − x 3 )
3 2 2
2 3
6 + 3x − 2 x 6 + 3x − x 3
x = ,x= , x = 6 + 3x − 2 x 2 ,…
x2 2x
Any one of the above assumptions can be converge in the
3
iterations? Tray to take x = 6 + 3x − 2 x 2 = F(x)
3 − 4x
∴ F′(x) = 1 < 0.66 < 1.
3 2 23
(6 + 3x − 2 x )
In the interval [1 , 2], | F′(x)| < 0.66 < 1.
We can put the results in the following table.
n xn F(xn) n xn F(xn)
1 1 1.912931 6 1.739219 1.728905
2 1.912931 1.641144 7 1.728905 1.733421
3 1.641144 1.769094 8 1.733421 1.731452
4 1.769094 1.715418 9 1.731452 1.732312
5 1.715418 1.739219 10 1.732312 1.731937
Thus, the sought-for root is x ≅ 1.731937.
The exact root at x = 3 = 1.732051….
§7.3.4: Newton- Raphson’s Method
(The tangent method):
This is one of the most efficient methods of approximate
calculation of the roots of the equation f(x) = 0. Newton’s Method
for approximating the zeros of a function is based on two facts.
(i) If f is differentiable at x, then the tangent line at (x, f(x)) can be

216
Algebra (Chapter 7) Calculation Methods

used to approximate the values of f near x.

(ii) The point where a line crosses the
x-axis is easy to determine.
Let x0, be the first approximation. An
equation of the tangent line at
(x0, f(x0)) is y – f(x0) = f′(x0)(x – x0)
Let x1, be the point where this line crosses the x-axis; that
is, when x = x1, then y = 0. Thus, 0 – f(x0) = f′(x0)(x1 – x0), and if
f (x 0 )
f′(x0)≠0, then x1 = x0 –
f '(x 0 )
Now let x2, be the second approximation, and repeat the
f ( x1 )
process to obtain x2 = x1 – provided f′(x2) ≠ 0.
f ' ( x1 )
Continue with x3, as the third approximation, to obtain
f (x 2 )
x3 = x2 – provided f′(x2) ≠ 0 and so on. At each step, if xn,
f ' (x 2 )
th
is the n approximation, then
f (x n )
xn+1 = xn – provided f′(xn) ≠ 0. (7.2)
f '(x n )

Continuing this process indefinitely, we shall construct a sequence

x0, x1, …, xn,… of the approximate values of the desired root.
Formula (7.2) determines the algorithm of the tangent method.
Thus, the tangent method is the method of successive
approximations (iteration method) constructed by means of the
recurrence formula (7.2).

217
Algebra (Chapter 7) Calculation Methods

We must note that the equation x = F(x), where

f (x)
F (x) = x – , (7.3)
f '(x)
possesses on the interval [a, b] only one root α coinciding with the
root of the equation f(x) = 0. Therefore, we shall solve equation
(7.3) instead of the equation f(x) = 0. For that purpose, having
taken some x0, we shall construct an iterative sequence
f (x n )
xn+1 = F (xn) = xn – (7.4)
f '(x n )

Note that the recurrence formula (7.4) exactly coincides with the
recurrence formula (7.2).
To prove the convergence of the iterative sequence {xn} to the
required root α, it is sufficient to prove that in some interval [a , b]
containing the root α the derivative F′(x) satisfies the condition
| F′(x)| ≤ k < 1 where x0 ∈[a , b].
[f ' ( x )]2 − f ( x )f " ( x ) f ( x )f " ( x )
Since we have F′(x) = 1 – = ,
[f ' ( x )]2 [f ' ( x )]2
then F′(α) = 0, and
[f ' ( x )]2 [f ( x ).f " ' ( x ) + f ' ( x )f " ( x )] − 2f ( x )f ' ( x )[f " ( x )]2
F″(x) = .
4
[f ' ( x )]
f " (α )
Since f(α) = 0 , then F″(α) = ≠ 0 in general. Then from
f ' (α )
Taylor’s expansion, the Newton’s iteration is of the second order.
Example 7.9:
–x
Find the point of intersection of the curves y = sin x and y = e ,
218
Algebra (Chapter 7) Calculation Methods

0 ≤ x ≤ π/2 .
Solution:
–x
The curves will intersect when sin x = e ; that is, when
–x
sin x – e = 0. Therefore, the x-coordinate of the point of
–x
intersection is a zero of the function f(x) = sin x – e that lies
–x
between 0 and π/2. Since f′(x) = cos x + e , it follows from (7.2)
sin x n − e − x n
that xn+1 = xn – .
cos x n + e − x n
Now sketch the curves to make a
reasonable first approximation. From
the Figure, it appears that x is about
one-third of the way to π/2 ≈ 1.5, so let
x1 = 0.5 be the first approximation.
The entries below were computed with an 8-digit calculator using
equation (7.4).
– xn –xn
xn sin xn – e cos xn + e xn+1
0.5 –0.12710512 1.48411322 0.58564382
0.58564382 –0.00401128 1.3901037 0.58852941
0.58852941 –0.00000462 1.38690103 0.58853274
0.58853274 0 1.38689733 0.58853274
Observe that x4 = x5. When xn+1 = x n, it follows from (7.2) that
f(xn) = 0 (at least to 8 digits). Therefore, 0.58853274 is the x-
coordinate of the intersection point. The y-coordinate is
–0.58853274
sin 0.58853274 (or e ). Thus the intersection of the two

219
Algebra (Chapter 7) Calculation Methods

curves is (0.58853274, 0.55514122).

Example 7.10:
Find the approximate value of the positive real root of the cubic
2 3 2
equation f(x) = (x – 3)(x + 2) = x + 2 x – 3x – 6 = 0.
Solution:
It is easy to see that f(1) < 0 , f(2) > 0. Then it is obvious that the
change in sign of the function between x = 1 and x = 2 guarantees
at least one root on the interval [1 , 2].
3 2 2
Since f(x) = x + 2 x – 3x – 6 , then f′(x) = 3x + 4 x – 3
and formula (7.1) takes the form
x 3n + 2 x 2n − 3x n − 6
xn+1 = xn –
3x 2n + 4 x n − 3
n xn f(xn) f'(xn) f(xn)/f'(xn) xn+1
1 1 –6 4 –1.5 2.5
2 2.5 14.625 25.75 0.567961 1.932039
3 1.932039 2.881296 15.92648 0.180912 1.751127
4 1.751127 0.24924 13.20384 0.018876 1.73225
5 1.73225 0.002578 12.93107 0.000199 1.732051
6 1.732051 0 12.9282 0 1.732051
Thus, the sought-for root is x ≅ 1.732051.
The exact root at x = 3 = 1.732051….
This iteration formula (xn+1 is defined in terms of xn) is
particularly easy to use on a programmable calculator; store xn and
then recall it to make the calculation above that produces xn+1;
then the process repeats to produce xn+2 and so on.

220
Algebra (Chapter 7) Calculation Methods

You can stop the process when two successive

approximations are equal because it follows from (7.2) that when
xn+1 = xn, then f(xn)= 0. That is, xn is a zero of f (at least to within
the accuracy limits of your calculator).
Example 7.11:
3
Find an approximate solution of x + 3x – 7 = 0.
Solution:
3
If f(x) = x + 3x – 7 = 0 , f(l) = – 3 and f(2) = 7 ;
∴ f(x) = 0 has one root between 1 and 2.
3 2
Since f(x) = x + 3x – 7 , then f′(x) = 3x + 3
and formula (7.1) takes the form
x 3n + 3x n − 7 2x 3n + 7
xn+1 = xn – =
3x 2n + 3 3x 2n + 3
n xn f(xn) f'(xn) f(xn)/f'(xn) xn+1
1 1 –3 6 – 0.5 1.5
2 1.5 0.875 9.75 0.089744 1.410256
3 1.410256 0.03552 8.966469 0.003961 1.406295
4 1.406295 0.000066 8.932997 0.000007 1.406288
5 1.406288 0 8.932934 0 1.406288
Therefore x =1.406288.
§7.3.5: The false position method:
The false position method is one of the most, favourable methods
of approximation used for solving the equation f(x)= 0. Now we
shall describe the method. (It is known as “Regula falsi method”):

221
Algebra (Chapter 7) Calculation Methods

Let y = f(x) be represented by

the curve AB. The curve AB
cuts the x-axis at the point
whose abscissa α. The real root
of f(x) = 0 is α. The false
position of the curve AB is
taken as the chord AB.
The chord AB cuts the x-axis at the point whose abscissa x1. The
approximate root of f(x) = 0 is x1. To find x1:
Let A(a, f(a)), B(b, f(b)) be the extremities of the chord AB, where
f(a). f(b) < 0.The equation of the chord AB is
f ( b ) − f (a )
y – f(a) = (x – a) (Two points form)
b−a
f ( b ) − f (a )
To find x1, put y = 0 ∴ – f(a) = (x1 – a)
b−a
af (b) − bf (a )
∴ x1 = the approximate value of the root.
f ( b ) − f (a )
It is easy to see that [a , b] = [a , x1] ∪ [x1 , b] , and the root α lies
on only one of the two subintervals [a , x1] or [x1 , b].
If f(a).f(x1) < 0, then α∈[a , x1] but if If f(x1).f(b) < 0, then
α∈[x1, b]. Repeat the above rule on the subinterval over which the
function f(x) has opposite signs on its end points.
As in the figure, since f(x1).f(b) < 0, thus draw a chord through the
points A1, with the abscissa x1, and B of the graph of the function.
The abscissa x2, of the point of intersection of the chord A1B with

222
Algebra (Chapter 7) Calculation Methods

the x-axis will be taken as a second approximation.

x f (b) − bf ( x1 )
x2 = 1
f ( b ) − f ( x1 )
Continuing this process indefinitely, we shall construct a sequence
x0, x1, x2, … , xn,… of the approximate values of the desired root.
§7.4 : Special cases:
§7.4 .1: Fixed point method:
If we fix the point B in all iteration the method is called fixed-
point method. In practical applications it is more convenient to
obtain a recurrence formula expressing xn+l in terms of xn. For that
y − f (x n ) x − xn
purpose we shall take the equation = of the
f ( b) − f ( x n ) b − x n
chord passing through points An(xn , f(xn)) and B(b , f(b)), and
calculate the abscissa xn+1 of the point of intersection of that chord
with the x-axis. We shall obtain
(b − x n )f ( x n ) x n f (b) − bf ( x n )
xn+1 = xn – = (7.5)
f ( b) − f ( x n ) f ( b) − f ( x n )

Formula (7.5) determines the algorithm of the fixed-point method.

Thus, the fixed-point method is the method of iterations, which
can be constructed by means of the recurrence formula (7.5).
If α is an exact root of the equation f(x) = 0, isolated on
the interval [a, b], and ξ is an approximate value of the root found
by the chord method, then the estimation of the approximation
error is

223
Algebra (Chapter 7) Calculation Methods

f (a ).f (b) f "(x )

| α – ξ| < – . max .
2 [ f ' ( x )]3
[a , b ]

§7.4.2 : The secant (chord) method:

The advantage of the Newton-Raphson process is its rapidity of
convergence. However it requires the calculation of the derivative
of the function. The secant method (chord) dispenses with the
calculation of the derivative and, although not quite so rapidly
convergent as Newton-Raphson, is better in this respect than the
method false position method. This method replace f′(xn) in the
Newton-Raphson method by the slope of the secant (chord) passes
through the two points (xn–1, f(xn–1)) , (xn, f(xn)). i.e. replace f′(xn)
f ( x n ) − f ( x n −1 )
by . Then the formula for xn+1 become
x n − x n −1
f ( x n )[ x n − x n −1 ] x n −1f ( x n ) − x n f (x n −1 )
xn+1 = xn – = . (7.6)
f ( x n ) − f ( x n −1 ) f ( x n ) − f ( x n −1 )
Unlike the Newton-Raphson method, it contains two functional
values. Thus, the calculation of xn+1 requires a knowledge of f(xn)
and f(xn–1). At the beginning, assuming n = 1,
f ( x1 )[ x1 − x 0 ] x 0 f ( x1 ) − x1f (x 0 )
x2 = x1 – = ,
f ( x1 ) − f ( x 0 ) f ( x1 ) − f ( x 0 )
x0, x1 are taken as two values of x for which the function is
numerically small. If we start with the two points (a , f(a)) and
(b , f(b)) , the formula becomes
f (b)[b − a ] af (b) − bf (a )
x2 = b – = .
f ( b ) − f (a ) f ( b ) − f (a )

224
Algebra (Chapter 7) Calculation Methods

Example 7.12:
–x
Find the root of the equation f(x) = sin x – e = 0 , 0 ≤ x ≤ π/2 .
Solution:
–x
The root of f(x) = sin x – e , lies between 0 and π/2 ≈ 1.570796.
Applying false position method we get:
af (b) − bf (a )
x1 = =1.570796/1.792120 = 0.876502 .
f ( b ) − f (a )
Since f(0.876502) = 0.352269 > 0; f(0) < 0, we again apply the
false position method to the interval [0 , 0.876502].
x f (b) − bf ( x1 )
x2 = 1 = 0.876502/1.352269 = 0.648171.
f ( b ) − f ( x1 )
Since f(0.648171) = 0.080728 > 0; f(0) < 0, we again apply the
false position method to the interval [0 , 0.648171] and etc.
We can put the results in the following table.
n an bn af(b)–bf(a) f(b) – f(a) xn+1 f(xn+1)
0 0 1.570796 1.570796 1.792120 0.876502 0.352269
1 0 0.876502 0.876502 1.352269 0.648171 0.080728
2 0 0.648171 0.648171 1.080728 0.599754 0.015493
3 0 0.599754 0.599754 1.015493 0.590604 0.002870
4 0 0.590604 0.590604 1.002870 0.588914 0.000528
5 0 0.588914 0.588914 1.000528 0.588603 0.000097
6 0 0.588603 0.588603 1.000097 0.588546 0.000018
7 0 0.588546 0.588546 1.000018 0.588535 0.000003
8 0 0.588535 0.588535 1.000003 0.588533 0.000001
9 0 0.588533 0.588533 1.000001 0.588533 0
Observe that x9 = x10. When xn+1 = x n, it follows that f(xn) = 0 (at
least to 6 digits). Therefore, 0.588533 is the root of the equation

225
Algebra (Chapter 7) Calculation Methods
–x
f(x) = sin x – e .
Note that in the above example 7.11; an = 0 for all values of n, the
above results are the same as of the fixed-point method.
Now applying the secant method:
f ( x n )[ x n − x n −1 ] x n −1f ( x n ) − x n f (x n −1 )
xn+1 = xn – = . (7.6)
f ( x n ) − f ( x n −1 ) f ( x n ) − f ( x n −1 )

Example 7.13:
Using the secant (chord) method, find the root of the equation
–x
f(x) = sin x – e = 0, 0 ≤ x ≤ π/2 ,with an accuracy to within 0.01.
Solution:
The positive root belongs to the interval (0 , π/2) since f(0)= –1< 0
and f(π/2) > 0. The first approximate value of the root is
af (b) − bf (a )
x1 = =1.570796/1.792120 = 0.876502 .
f ( b ) − f (a )
Since f(0.876502) = 0.352269 > 0; f(0) < 0, we again apply the
secant method to the interval [0 , 0.876502].
af ( x1 ) − x1f (a ) x1
x2 = = = 0.876502/1.352269=0.648171;
f ( x1 ) − f (a ) f ( x1 ) + 1
f(0.648171) = 0.080728 > 0. We find the third approximate value:
af ( x 2 ) − x 2 f (a ) x2
x3 = = =0.648171/1.080728=0.599754;
f ( x 2 ) − f (a ) f (x 2 ) + 1
f(0.599754) = 0.015493 > 0. The fourth approximate value:
af ( x 3 ) − x 3f (a ) x3
x4 = = = 0.599754/1.015493=0.590604;
f ( x 3 ) − f (a ) f (x 3 ) + 1
f(0.590604) = 0.00287 > 0. Consequently, with accuracy to within
0.01, the desired root is equal to 0.590604.

226
Algebra (Chapter 7) Calculation Methods
Let us denote to λn = xn–1f(xn) – xn f(xn–1) ; µn = f(xn) – f(xn–1),
∴ xn+1 = λn / µn
n a b λn µn xn+1 f(xn+1)

0 0 1.570796 1.570796 1.79212 0.876502 0.352269

1 0 0.876502 0.876502 1.352269 0.648171 0.080728
2 0 0.648171 0.648171 1.080728 0.599754 0.015493
3 0 0.599754 0.599754 1.015493 0.590604 0.00287
4 0 0.590604 0.590604 1.00287 0.588914 0.000528
5 0 0.588914 0.588914 1.000528 0.588603 0.000097
6 0 0.588603 0.588603 1.000097 0.588546 0.000018
7 0 0.588546 0.588546 1.000018 0.588535 0.000003
8 0 0.588535 0.588535 1.000003 0.588533 0
9 0 0.588533 0.588533 1.000001 0.588533 0

Example 7.14:
Using the secant (chord) method, find the positive real root of the
2 3 2
equation f(x) = (x – 3)(x + 2) = x + 2 x – 3x – 6 = 0.
–x
f(x) = sin x – e = 0, 0 ≤ x ≤ π/2 ,with an accuracy to within 0.01.
Solution:
It is easy to see that f(1) < 0 , f(2) > 0. Then it is obvious that the
change in sign of the function between x = 1 and x = 2 guarantees
at least one root on the interval [1 , 2].
3 2
f(x) = x + 2 x – 3x – 6 , and formula (7.5) takes the form
227
Algebra (Chapter 7) Calculation Methods
x n −1f ( x n ) − x n f (x n −1 )
xn+1 = . (7.6)
f ( x n ) − f ( x n −1 )
n an bn λn µn λ/µ= xn+1 f(xn+1)
0 1 2 16 10 1.6 –1.584
1 1.6 2 9.568 5.584 1.713467 –0.23778
2 1.713467 2 7.32942 4.237776 1.729544 –0.03236
3 1.729544 2 6.982902 4.032363 1.731715 –0.00435
4 1.731715 2 6.935549 4.004345 1.732006 –0.00058
5 1.732006 2 6.929188 4.000582 1.732045 –0.00007
6 1.732045 2 6.928335 4.000078 1.73205 –0.00001
7 1.73205 2 6.928221 4.00001 1.732051 0
8 1.732051 2 6.928206 4.000001 1.732051 0
Thus, the sought-for root is x ≅ 1.732051.
The exact root at x = 3 = 1.732051….

228
Algebra (Chapter 7) Calculation Methods

EXERCISES 7
Using Horner’s method find approximate solutions of the
following equations:
3
1) x – 2 = 0. [Answer: 1.259921]
3
2) x – 2x – 5 = 0. [Answer: 2.09455]
3
3) x – 5 x + 3 = 0, (3 roots).
[Answers: 0.65662, 1.83424, –2.49086]
3
4) x – 7x – 1 = 0, (3 roots).
[Answers: 2.71448, –0.14328, –2.5712]
3
5) 2 x – 3x – 6 = 0, [Answers: 1.78377]
3
6) x – 9x + 14 = 0, [Answers – 3.59141]
3
7) x – 8x – 1 = 0, (3 roots).
[Answers: 2.88897, –0.12524, –2.76372]
3
8) x + x – 1000 = 0, [Answer: 9·96667]
4
9) x – 12x + 7 =0, [Answers: 0.59369 , 2.04728]
3 2
10) x – 3 x + 5x – 43 = 0, [Answer: 4.22524]
3 2
11) x – 8 x – 6x + 9 = 0, (3 roots),
[Answers: 0.77563, 8.57719, –1.35283]
5
12) x – 7 = 0, [Answer: 1.47577]
3 2
13) x + 2 x – 5x – 7 =0, positive root only,
[Answer: 2.064435]
3
14) x – 7x + 7 = 0, two roots between 1 and 2,
[Answers: 1.356894, 1.69202, –3.04892]
5
15) x – 4x – 2000 = 0. [Answer: 4.58140]
16) Determine the isolation intervals of the real roots of the

229
Algebra (Chapter 7) Calculation Methods
3 2
equation x – 9x + l8 x – 1 = 0. [Answers: (0, 1),(2, 3) ,(6, 7)]
17) Determine the isolation intervals of the real roots of the
3
equation x – 12x + 1 = 0. [Answers: (–4, –3), (0, 1), (3, 4)]
18) Find graphically the real roots of the equations:
i) 4 x – 7 sin x = 0,
[Answers: 0, 1.73]
ii) x ℓn x – 14 = 0,
[Answer: 7.13]
iii) 4 x – cos x = 0,
[Answer: 0.24]
iv) x ℓog x = 1,
[Answer: 2.506]
x
v) 2 = 4 x.
[Answers: 0.31 , 4]
19) Using method of halving the interval find approximate
solutions of the following equations correct to three decimal
places:
3
i) x – x + 1 = 0, [Answers: –1.325 ]
3
ii) x – 4x – 1 = 0, [Answers: –1.860, –0.255, 2.114]
3
iii) x – 2x – 5 = 0, [Answers: 2.094]
3
iv) x – 5x + 0.1 = 0, [–2.246, 0.020 and 2.226]
3
v) x – x – 2 = 0, [Answers: 1.521]
3
vi) x – 3x + 1 = 0, [Answers: –1.88, 0.347, 1.531]
4 2
vii) x + x – 2x – 2 = 0, [Answers: –0.673, 1.306]
3 2
viii) x – 2x + 3x – 5 = 0, [Answers: 1.843]
20) Use the method of iteration to find the approximate value of
230
Algebra (Chapter 7) Calculation Methods

the root of the equation 3x + ℓog x = 2, with an accuracy to within

0.0001. [Answer: 1.752]
3
21) Use the iteration method to solve the equation x + 2x – 7= 0
with an accuracy to within 0.00001. [Answer: 1.5689]
22) Find the approximate value of the root of the equation
3 2
x + x – 11 = 0, isolated on the interval (1 , 2), with an accuracy
to within 0.001. [Answer: 1.936]
By the iteration method solve the following equations with
accuracy to within 0.0001:
3
23) x – 12x – 5 = 0. [Answer: 3.6562]
3 2
24) x –2x – 4x – 7 = 0. [Answer: 3.6317]
Solve the following equations by the method of halving the
interval to within 0.001:
x
25) x + e = 0. [Answer: –0.567]
5
26) x – x – 2 = 0. [Answer: 1.267]
27) Applying the iteration method five times, find the approximate
root of the equation 2x – cos x = 0 isolated on the interval [0 , 0.5]
accurate to the third significant digit. [Answer: 0.450]
28) Using Newton- Raphson’s Method, find a positive root of the
4
equation x – 2x – 4 = 0 with an accuracy to within 0.01.
[Answer: 1.64]
29) Find the least positive root of the equation tan x – x = 0, with
accuracy up to 0.0001 applying Newton-Raphson’s method.
[Answer: 4.4934]
30) Find the real root of the equation 3x + ℓog x = 2, by applying
Newton-Raphson’s method. [Answer: 0.7151924]
231
Algebra (Chapter 7) Calculation Methods

31) Find the real root of the equation 2x – 4 – ℓn x = 0, by

applying Newton-Raphson’s method. [Answer: 2.447]
32) Applying three times Newton Raphson method, find the
4
approximate value of the real root of the equation x – 8x + 1 = 0,
isolated on the interval [1.6 , 2]. [Answers: x3=1.956465]
33) Applying Newton Raphson method to approximate a solution
4
of x = 3 – x. [Answer: 1.1640351]
34) Find the approximate value of the root of the equation
3x – 1 + sin x = 0. [Answer: 0.391847]
3 2
35) Find the real root of the equation x – 2x + 3x – 5 = 0 , with
–4
an accuracy up to 10 : by two different methods.[Answer:1.8437]
36) Applying the iteration method find the real roots of the
equation x – sin x = 0.25 accurate up to three decimal places.
[Answer: 1.170]
37) Applying the iteration method find the real roots of the
2
equation x = sin x, accurate up to three decimal places.
[Answers: 0 , 0.877]
38) Applying twice the method of false position, find the
3
approximate value of the real root of the equation x – 10 x + 5 = 0
isolated on the interval [0 , 0.6]. The approximations x1 and x2
must be calculated to the second decimal place. Estimate the error
of the approximation x2. [Answer: x1 = 0.51867, x2 = 0.51382]
By the methods of false position solve the following equations
with accuracy to within 0.01:
4
39) x + 3x – 20 = 0. [Answer: 1.94]
3
40) x – 2x – 5 = 0. [Answer: 2.09]
232
Algebra (Chapter 7) Calculation Methods
4
41) x – 3x + 1 = 0. [Answers: 0.33, 1.30]
3
42) x + 3x + 5 = 0. [Answer: –1.15]
4
43) x + 5x – 7 = 0. [Answer: 1.103]

233
Algebra References

REFERENCES
SUGGESTIONS FOR FURTHER READING
1) The Tutorial Algebra, volumes I and II,
by W. Briggs, G.H. Bryan and G. Walker.
2) Advanced Algebra volumes, I, II and III.
By C.V. Durell and A. Robson.
3) College Algebra, by N.H. Phadke and P.K. RAO.
4) College Algebra, (Theory and Problems)
Schaum's outline series by Murray R. Spiegel.
5) Finite Mathematics, (Theory and Problems)
Schaum's outline series by Seymour Lipschutz.
6) Matrices (Theory and Problems)
Schaum's outline series by Frank Ayres.
7) Advanced Mathematics for Engineers and Scientists
(Theory and problems) Schaum’s outline series
by Murray R. Spiegel.
8) Numerical Analysis, (Theory and Problems)
Schaum's outline series by Francis Scheid.
9) Operations Research (Theory and Problems)
Schaum's outline series by Richard Branson.
10) Introductory Linear Algebra with Applications
by Bernard Kolman.

233
‫ﺭﻗﻢ ﺍﻹﻳﺪﺍﻉ ﺑﺪﺍﺭ ﺍﻟﻜﺘﺐ ﻭﺍﻟﻮﺛﺎﺋﻖ ﺍﻟﻘﻮﻣﻴﺔ‬

Algebra
(Selected Topics)
By Dr. Mohamed Ismail
Mohamed Hessein
18012/2001
Algebra (Chapter 8) Linear programming

CHAPTER 8
Linear programming
§ 8.1 Optimisation problems:
In optimisation problems one seeks to minimize or maximize a
specific quantity called objective, which depends on a finite
number of variables. These variables may be independent of one
another or they may be related through one or more constraints.
A mathematical program is an optimisation problem in
which the objectives and constraints are given as mathematical
functions and functional relationships. Mathematical program has
the form:
Optimise (Maximize or Minimize) Z
Z = f (x1, x2 ... xn) (8.1)
Subject to:
g1 (x1, x2 ... xn ) ~ b1, ⎫
⎪
… … … … … … ⎬ (8.2)
⎪
gm (x1, x2 ... xn ) ~ bm ⎭

Where ~ stands for one of the relations ≤ , ≥ , = (not necessarily

the same, one for each i). The constant bi may always be assumed
non-negative. In any mathematical program we seek a solution. If
a number of equally optimal solutions exist then any one will do.
There is no preference between equally optimal solutions if there
is no preference stipulated in the constraints. For general

234
Algebra (Chapter 8) Linear programming

mathematical optimisation there are no restrictions on the given

functions f and gi ; for quadratic optimisation f(xj) is quadratic in
the xj and g (xj) is linear; for linear optimisation f(xj) and gi(xj) are
linear functions; and for convex optimisation f(xj) is a convex
function and g(xj) defined convex region.
Convex region:
A region (set) S is called convex if the line segment joining any
two points in S also lies entirely in S.
The convex region S may be finite or infinite. A cube,
parallelepiped, tetrahedron, sphere and ellipsoid are examples of
convex regions.

Convex sets in (i), (ii) and (iii) (Figures 1)

Non convex sets in (iv), (v) and (vi) (Figures 2)

Convex function:
A function f(x) defined on a convex region S is said to be convex
for any two vectors x′ and x″ in S and for all 0 ≤ k ≤ 1,

235
Algebra (Chapter 8) Linear programming

f (kx′ + (1– k)x″) ≤ k f(x′) + (1– k)f(x″). (8.3)

If (8.3) holds with the inequality reversed, then f(x) is
concave. Thus, the negative of a convex function is concave and
conversely.
The curve of a convex function lies on or above of its
tangents.
Theorem 8.1:
If f(x) is twice-differentiable on the interval I, then f(x) is convex
on I if and only if f″(x) ≥ 0 for all x in I.
The simplest example of such a function is a linear
function; in this case the equality sign is attained in (8.3).
The sum of a finite number of convex downward functions
is again a convex downward function.
§ 8.2 Linear programming linear inequalities:
Linear programming is a new area of mathematics developed in
1941 to solve a number of numerous problems in many areas. In
many problems in business and industry we are interested in
making decisions that will maximize or minimize some quantity.
For example, a plant manager may want to determine the most
economical way of shipping his goods from his factory to his
markets a hospital may want to design a diet satisfying certain
nutritional requirements at a minimum cost, a manufacturer may
wish to blend ingredients subject to given specifications to

236
Algebra (Chapter 8) Linear programming

maximize his profit.

It was in 1947 that George B. Dantzig and his associates
found out techniques for solving military problems while they
were working on a project for U.S. Air Force. This technique
consisted of representing the various activities of an organization
as a linear programming (L.P.) model and arriving at the optimal
program by minimizing a linear objective function. Afterwards
Dantzig suggested this approach for solving business and
industrial problems. He also developed a powerful mathematical
tool known as simplex method to solve linear programming
problems.
Linear inequities:
The linear equation a x + b y = c,
partitions the plane R2 into three
subsets:
I) Those points which satisfy the
given equation a x + b y = c ;
II) Those points which satisfy
(Figure 3)
the inequality a x + b y < c.
III) Those points which satisfy the inequality a x + b y > c.
The first set, determine a straight line; the other two sets
are called open half planes.

237
Algebra (Chapter 8) Linear programming

Example 8.1:
To draw the graph of the
inequality x + 2y ≤ 6, first plot the
line x + 2y = 6. Then choose an
arbitrary point that doesn't belong
to the line say (0 , 0) and test it to
see if it belongs to the solution set.
Since (0 , 0) satisfy the given (Figure 4)
inequality the point on that side of the line belongs to the solution
set.
Example 8.2:
To determine the region defined by the inequities:
x + 2y ≤ 6 , x – y ≤ 6 , x ≥ 1 , y ≥ –3.

(Figures 5)
The set of points satisfying the four inequities is the intersection of
the regions in figure 5 (a, b, c, d). This set of points is shown in
figure 6. The intersection of two boundary lines is called a corner
(an extreme) point. There are four corner points belonging to the

238
Algebra (Chapter 8) Linear programming

solution set. A(1 , –3) the

solution of x = 1, y = –3;
B (3 , –3) the solution of
x – y = 6, y = –3 ;
C (6 , 0) the solution of
x – y = 6, x + 2y = 6 ;
D (1 , 5 ), the solution of
2
x + 2 y = 6, x = 1.
There are other two corner
points not belonging to the (Figure 6)
solution set. E (1 , –6) the solution of x – y = 6 , x = 1; doesn’t
satisfy y ≥ –3.
F(12,–3) the solution of x+2y= 6, y = –3; doesn’t satisfy x –y = 6.
Feasible solution:
A point satisfying the given inequities is called a feasible solution;
the set of all such points is the feasible region.
It should be noted that not every
given inequities has a feasible region
such as y ≥ 2 and y ≤ 1. Here the solution
is empty set and we say that the system
is inconsistent.
(Figure 7)

239
Algebra (Chapter 8) Linear programming

§ 8.3 Two-dimensional linear programming:

We seek the maximum or minimum value of a linear function:
f = d1x + d2 y,
where the unknowns are subject to the following linear constraints
a1x + b1y ≤ c1,
a2x + b2y ≤ c2,
… … … …
amx + bmy ≤ cm.
Theorem 8.2:
Let S be the feasible region of a linear programming problem.
a) If S is a convex polygon then the objective function
f = f(x , y) = d1x + d2 y. assumes both a maximum and minimum
value of S; these values occur at a corner points of S.
b) If S not a convex polygon (S is unbounded) then there may be
or may be not a maximum or minimum value of S. If a maximum
or minimum value does exist on S it must occur at a corner point.
Thus when the feasible region S of linear programming
problem (L.P.P.) is a convex polygon a method for solving the
problem consists in finding the corner (extreme) point of S and
evaluating the value of the objective function f(x , y) at each
corner (extreme) point. An optimal solution is a corner point at
which the value of f(x , y) is a maximum (or minimum).

240
Algebra (Chapter 8) Linear programming

Example 8.3:
Find the maximum and minimum values of the function
f = 2x + 3y over the convex polygon S given by:
x + 2y ≤ 6 , x – y ≤ 6 , x ≥ 1 , y ≥ 3.
The extreme (corner)
points of S are as follows
(see Example 8.2):
A(1,–3), B(3, –3), C(6, 0),
D (1 , 5 ). Evaluate f at
2
each of these points:
f(A) = f(1,– 3) = –7 ,
f(B) = f(3 , –3) = –3 , ( Figure 8)
f(C) = f(6 , 0) = 12 , f(D) = f(1 , 5 ) = 19 .
2 2
Thus the maximum value of f over S is 12 and occurs at C(6 , 0)
and the minimum value of f is –7 and occurs at A(1 , –3).
We show how this result follows geometrically. We can
consider the equation f = 2x + 3y, as a one parameter family of
lines in the plane. We seek the maximum (or minimum) of f with
the condition that the line 2x + 3y = f, intersects the given convex
polygon S. As indicated by the diagram (Figure 8) starting with
lines of the family with large values of f the family will first
intersect S at the point C. Similarly starting with lines having
small values of f the family will first intersect S at point A. In

241
Algebra (Chapter 8) Linear programming

other words the linear function f = 2x + 3y, will take on its

maximum and minimum values over S at the points C and A,
respectively.
Example 8.4:
Find the maximum and minimum values of the functions:
g = –x + 2y , h = 2x – y, z = x – y, over the convex polygon S of
Example 8.3 .
Solution:
The corner (extreme) points of S are as follows:
A(1 , – 3), B(3 , –3), C(6 , 0) and D (1, 5 ).
2
Evaluate g at each of these points:
g(A) = 7 , g(B) = –9 , g(C) = – 6 , g(D) = 4.
Thus the maximum of g over S is 4 and occurs at D and the
minimum value is 9 and occurs at B.
Also h(A) =5, h(B) = 9 , h(C) = 12 , h(D) = – 1 .
2
The maximum of h over S is 12 and occurs at C and the minimum
value is – 1 and occurs at D.
2
Similarly z(A) = 4 , z(B) = 6 , z(C) = 6 , z(D) = – 3 .
2
The minimum value of z over S is – 3 and occurs at D and the
2
maximum value of z over S is 6 and occurs at both B and C.
i.e. occur at all points on the segment BC.

242
Algebra (Chapter 8) Linear programming

Example 8.5:
Find the extreme points of the polygonal convex set S determined
by the system: 3x + 2y ≤ 15, g = 2x – 3y, h = 3x + y,
Z = 3x + 7y , T = –3x – 7y, and U = 3x + 2y.
Solution:
Find the intersection of each
pair of lines
L1 = 3x + 2y –15 = 0 ,
L2 = y – x = 0, L3 = x – 7 = 0,
and test the point in the other
inequality. (see Figure 9)
i) Since L1∩L2={A} , A(3 , 3)
(Figure 9)
and it does satisfy the
inequality x ≤ 7 hence A(3 , 3) is an extreme point of S.
ii) Since L1∩L3 = {B}, B(7,–3)
and it does satisfy the
inequality y – x ≤ 0 , hence
B( –7, 3) is an extreme point of
S. (see Figure 9)
iii) Since L2∩L3= {C} ,C(7, 7)
but C doesn’t satisfy the
inequality 3x + 2y ≤ 15, hence
C isn’t an extreme point of S. (Figure 10)

243
Algebra (Chapter 8) Linear programming

The set S and the family of

parallel lines determined by
the function f = 3x – 2y, are
plotted in Figure 10. As
indicated by the diagram, the
lines f = 3x – 2y will intersect
S for every value of f.
Thus f takes on no maximum
(Figure 11)
value and no minimum value over S.
The set S and the family of parallel lines determined the
function g = 2x – 3y are plotted in Figure 11. As indicated by the
diagram, the lines g = 2x – 3y will intersect S for positive values
of g and so g will not take a maximum value over S. But g will
take on a minimum value over S at the point A(3 , 3).
Thus g(A) = g(3 , 3) = 3 is the minimum value of g over S.

(Figure 12) (Figure 13)

The set S and the family of lines determined by the

244
Algebra (Chapter 8) Linear programming

function h = 3x + y are plotted in figure 12. As indicated by the

diagram the lines h = 3x + y
will intersect S for negative
values of h and so h will not
take on a minimum value of S.
But h will take on a maximum
value of S over the point
B(7 , –3). Thus
h(B) = h(7,–3) = 18 is the (Figure 14)
maximum value of h over S.
In figure 13, the lines Z = 3x + 7y will take on a maximum
value over S at the point A(3 , 3).
Thus Z(3 , 3) = 30 is the maximum value of Z over S.
Similarly T = –3x – 7y = –Z will take on a minimum value over S
at the point A(3 , 3) and T(3 , 3) = –30 is the minimum value of T
over S.
In Figure 14, the lines U = 3x + 2y are parallel to the
bound AB i.e. parallel to L1 = 3x + 2y – 15 = 0. It is easy to see
that u will take on a maximum value over S at the two points
A(3 , 3), B(7, –3) and U(3 , 3) = U(7 , –3) = 15.
Hence the maximum value will occur at every point on the line
segment joining A and B.

245
Algebra (Chapter 8) Linear programming

Example 8.6: (Maximum problem)

2 2
A tailor has 27 m of cotton material 4 m of silk material and
2 2 2
24 m of wool material. A jacket requires 3 m of cotton, 1 m of
2 2 2
silk and 1 m of wool and a dress requires 1 m of cotton, 2 m of
2
silk and 4 m of wool. If a jacket sells for L.E. 50 and dress sells
for L.E. 250. How many of each garment should the tailor make to
maximize his income?
Solution:
The following table summarize the given facts:

Kind
Materials Available
Jacket Dress
Cotton 3 1 27
Silk 1 2 14
Wool 1 4 24
Price in L.E. 150 250

Let x be the number of jackets and y the number of dresses the

tailors makes. We seek to maximize f:
f = 50x + 250 y, (1)
subject to
3x + y ≤ 27, (2)
x + 2y ≤ 14, (3)
x + 4y ≤ 24, (4)

246
Algebra (Chapter 8) Linear programming

x ≥ 0, (5)
y ≥ 0. (6)
Observe that the inequality 3x + y ≤ 27, corresponds to the fact
2 2
that 3 m of cotton is required for each jacket and 1 m for each
2
dress, but only 27 m is available. The other inequities
x + 2y ≤ 14, and x + 4y ≤ 24 correspond to the analogous facts
about the silk and the wool.
The graph of the convex polygon
S is given in figure 15. The corner
points of S are O(0, 0), A(9, 0),
B(8, 3), C(4, 5) and D(0, 6). The
following aren’t corner (extreme)
points of S: (Figure 15)
(0 , 7) doesn’t satisfy x + 4y ≤ 4, (see Figure 16); (0 , 27) doesn’t
satisfy x + 2y ≤ 14, (14 , 0) doesn’t satisfy 3x + y ≤ 24, (24 , 0)
doesn’t satisfy 3x + y ≤ 24, ( 84 , 45 ) doesn't satisfy x + 2y ≤ 14.
11 11
The values of the objective
function f = 50x + 250y, at the
corner points are f(O) = 0,
f(A) = 350, f(B) = 950 , f(C) = 50,
f(D) = 1500. Thus the maximum
of f over S is 1950, and occurs at
the point B(8 , 3).
(Figure 16)
247
Algebra (Chapter 8) Linear programming

In other words the maximum amount of money the tailor can get
is L.E. 1950 and this will happen if he makes 8 jackets and 3
dresses.
Example 8.7: (minimum problem)
Suppose that 8, 2 , 9 units of proteins carbohydrates and fats
respectively are the minimum weekly requirements for a person.
Food A contains 2, 6 and 1 units of proteins, carbohydrates and
fats respectively per kg; and food B contains 1 , 1 and 3 units of P,
C and F respectively per kg. i) If A costs L.E. 1.1 per kg and B
costs L.E. 0.5 per kg, how many kg of each should he buy per
week to minimize his cost and still meet his minimum
requirements?
ii) If the price of A drops to L.E. 0.9 per kg while B still sells at
L.E. 0.5 per kg, how many kg of each should be then buy?
Solution:
The following table summarize the given facts:
Substances Kind Requirements
Food A Food B
Proteins 2 1 8
Carbohydrates 6 1 12
Fats 1 3 9
Cost (i) in
1.1 0.5
L.E.
Cost (ii) in
0.9 0.5
L.E.

248
Algebra (Chapter 8) Linear programming

Let x denote the number of kg bought from food A and y the

number of kg bought from food B. We seek to minimize the
functions:
i) g = 1.1 x + 0.5y,
ii) h = 0.9x + 0.5y,
subject to the conditions :
2x + y ≥ 8, (1)
6x + y ≥ 12, (2)
x + 3y ≥ 9, (3)
and the fact that x and y are non
( Figure 17)
negative: x ≥ 0 , y ≥ 0.
The graph of the polygonal convex set S (of this minimum linear
programming) is given in Figure 17.
The corner points of S are A(9 , 0) B(3 , 2), C(1 , 6) and D(0 , 12).
But the following points (0, 8) , (0 , 3) , (2 , 0), (4 , 0) ( 27 , 42 )
17 17
doesn’t satisfy the given inequities hence they don't belong to the
convex polygon. Although S isn't bounded the objective functions
g and h is never negative over S and so takes on a minimum value
at the corner point of S. Since g(A) = 9.9, g(B) = 4.3, g(C) = 4.1,
g(D) = 6; h(A) = 8.1, h(B) = 3.7, h(C) = 3.9 and h(D) = 6.
Then minimum value of g over S is L.E. 4.1 and occurs at
C(1 , 6). In other words he should buy 1 kg from food A and 6 kg
from food B. The minimum value of h over S is L.E. 3.7 and

249
Algebra (Chapter 8) Linear programming

occurs at B(3 , 2). In other words he should buy 3 kg from food A

and 2 kg from food B.
§ 8.4 The simplex method:
Problems discussed in § 8.3 involved two variables only.
The problems were two-dimensional and were simple to be
represented (by two axes lying in a plane) and solved graphically.
Now, as the number of variables increases to 3, 4, …, we come
across 3-dimensional, 4-dimensional, … problems which become
quite laborious to be represented and solved by graphical methods.
In such cases simplex technique helps us in:
i) Starting with a feasible solution and
ii) Searching for an optimal solution in a systematic way.
We shall refer to the following L.P.P. as a standard linear
programming problem. Find the values of x1, x2, …, xn that will
maximize f,
f = c1 x1+ c2 x2+ …+ cnxn , (8.4)
subject to the constraints :
a11 x1 + a12 x2 + a13 x3 +… a1n xn ≤ b1,
⎫
a21 x1 + a22 x2 + a23 x3 +… a2n xn ≤ b2, ⎪
⎪
⎬ (8.5)
… … … … … ⎪
am1 x1 + am2 x2 + am3 x3 +… amn xn ≤ bm. ⎪⎭

xi ≥ 0 for i = 1, 2, …,n.
Every L.P.P. can be transformed into a standard L.P.P.

250
Algebra (Chapter 8) Linear programming

Every minimization problem can be viewed as a

maximization problem and conversely. This follows from the
observation that:
Minimum of c1 x1+ c2 x2+ …+ cnxn =
= Maximum of [–(c1 x1+ c2 x2+ …+ cnxn )].
Slack variables and surplus variables:
n
A linear constraint of the form ∑ a ij x j ≤ b i can be
i =1
converted into equality by adding a new non-negative variable ui
to the left-hand side of the inequality. Such a variable is
numerically equal to the difference between the right-hand, left-
hand-sides of the inequality and is known as a slack variable (it
represents the waste involved in that phase of the system modelled
by constraint.)
n
∑ a ij x j + u i = b i
i =1
ui is called a slack variable since it takes up the slack between the
two sides of the inequality. i.e. the slack variable ui is defined by:
n
u i = b i − ∑ a ij x j ≥ 0 .
i =1
n
A linear constraint of the form ∑ a ij x j ≥ b i can be
i =1
converted into equality by subtracting a new non-negative variable
from the left-hand side of the inequality. Such a variable is
numerically equal to the difference between the left and right-hand
251
Algebra (Chapter 8) Linear programming

sides of the inequality and is known as a surplus variable. It

represents excess input into that phase of the system modelled by
the constraint.
After all linear constraints (with non negative right-hand
sides) has been transformed into equalities by introducing slack
and surplus variables necessary, add a new variable called an
artificial variable, to the left-hand side of each equation that
doesn’t contain a slack variable. Each constraint equation will then
contain either one slack variable or one artificial slack variable
and each artificial variable equal to the right-hand side of the
equation in which it appears and setting all other variables,
including the surplus variables, equal to zero.
Example 8.8:
The set of constraints:
x + 3y ≤ 2, 5x + 7y ≥ 3, and 2x + 3y = 5 (1)
is transformed into a system of equations by adding a slack
variable, u, to the left-hand side of the first constraint and
subtracting a surplus variable, v, from the left-hand side of the
second constraint. The new system is:
x + 3y + u = 2, ⎫
⎪
5x + 7y – v = 3, ⎬ (2)
⎪
2x + 3y = 5. ⎭

If now artificial variables w and z are respectively added to the

252
Algebra (Chapter 8) Linear programming

left-hand sides of the last two constraints in system (2), the

constraints without a slack variable, the result is
x + 3y + u = 2, ⎫
⎪
5x + 7y –v+w = 3, ⎬ (3)
⎪
2x + 3y + z = 5. ⎭

A non-negative solution to this system (3) is u = 2, w = 3, z = 5,

and x = y = v = 0. (Notice, however that x = 0, y = 0 is not a
solution to the original set of constraints.).
General idea of the simplex method:
In the graphical solution of § 8.3, we observed that the
optimum solution is always associated with a corner (or extreme)
point of the solution space. The simplex method is based
fundamentally on this idea. It employs an iterative process that
starts at a feasible corner point, normally the origin, and
systematically moves from one feasible extreme point to another
until the optimum point is eventually reached.
The general idea of the method will be illustrated in terms
of the maximum problem (Example 8.6). The simplex algorithm
starts at the origin, which is usually referred to as the starting
solution. (see Figure 15). It then moves to an adjacent corner
point, which could be either A(9, 0) or D(0, 6). The specific
choice of either point depends on the coefficients of the objective
function. Since y has a larger coefficient than x and since we are

253
Algebra (Chapter 8) Linear programming

maximizing, the solution will move in the direction in which y

increases until extreme point D(0, 6) is reached. At D, the process
is repeated to see if there is another extreme point that can
improve the value of the function. Again by using the information
in the objective function we can decide whether or not this point
exists. Eventually, the solution will stop at B(8 , 3), the optimum
extreme point.
There are two rules that govern the selection of the next
extreme (corner) point.
i) The next extreme point must be adjacent to the current one. For
example, in Figure 15, the algorithm cannot move from O to C
directly. Instead, it must follow the edges of the solution space:
O→D and then D→C.
ii) The algorithm can never go back to a previously considered
extreme point e.g. the algorithm cannot “regress” from D→O.
To summarize the ideas of the simplex method, we notice
that the method always starts at a feasible corner point, and always
passes through an adjacent feasible corner point, checking each
point for optimality before moving to a new one.
In Figure 15 of example 8.6, where the method starts at O, passes
through D, and locates the optimum at B, we say that it took four
iterations (O, D, C, and B) to reach the optimum. We shall
illustrate this method using Example 8.6.

254
Algebra (Chapter 8) Linear programming

Example 8.9:
Find the values of x and y that maximize
f = 150x + 250 y, (1)
subject to
3x + y ≤ 27, (2)
x + 2y ≤ 14, (3)
x + 4y ≤ 24, (4)
x ≥ 0, (5)
y ≥ 0. (6)
We now change each of the constraints (2),(3) and (4) into an
equation by introducing a non-negative slack variables u, v, and
w. Thus our new problem can be stated as follows:
Find the values of x y u v w that maximize:
f = 150x + 250 y + 0u + 0v + 0w, (7)
subject to
3x + y + u = 27, (8)
x + 2y +v = 14, (9)
x + 4y + w = 24, (10)
2
A vector X in R satisfying (2) - (6) is called a feasible solution to
the given problem, and a feasible solution maximizing the
objective function (1) is called an optimal solution.
n
The vector X in R is called a basic solution of the new
problem if it is obtained by setting two of the variables equal to

255
Algebra (Chapter 8) Linear programming

zero and solving for the remaining three variables. The three
variables that we solve for are called basic variables, and the two
variables set equal to zero are called non-basic variables.
To solve a L.P.P. we need only to search for basic feasible
solutions, we can select two of the five variables x, y, u, v, w as
non basic variables by setting them equal to zero and solve for the
remaining three variables; that is we solve for the basic variables.
Thus if x = y = 0, then u = 27, v = 4, and w = 24.
t
The vector [0, 0, 27, 14, 24] is a basic feasible solution, which
t
given rise to the feasible solution [0, 0] to the original problem.
The variables x and y are non-basic and variables u, v, and w are
t
basic. The vector [0 , 0] corresponds to the corner point O(0,0).
The value of the objective function for this initial basic feasible
solution is f = 0 (see equation (7)). It is clear that this solution is
not optimal.
The value of f in (7), can be increased by increasing either
x or y since both of these variables appears in (7) with positive
coefficients. The largest increases in f per unit increase in a
variable occur for the variable with the larger coefficient. The
variable to be increased is called the entering variable, because in
the next iteration it will become a basic variable, there by entering
the set of basic variables. If there are several candidates for
entering variables choose one. An increasing in y must be

256
Algebra (Chapter 8) Linear programming

accompanied by a decrease in some of the other variables. This

can be seen if we solve equation (8), (9) and (10) for u, v, w:
u = 27 – 3x – y, (11)
v = 14 – x – 2y, (12)
w = 24 – x – 4y. (13)
Since we only increase y, we keep x = 0, and obtain:
u = 27 – y , v = 14 – 2y, and w = 24 – 4y, (14)
so that as y increases : all u, v, and w decrease.
Equations (14) also show by how much we can increase y. That is,
since u, v, and w must be non-negative, we must have:
y ≤ 27 = 27 , y ≤ 14 = 7 and y ≤ 24 = 6.
1 2 2
We now see that the allowable increase in y can be no longer than
the smaller of the ratios 27 , 14 , and 24 . Taking y as 6 we obtain
1 2 2
the new basic feasible solution:
x = 0 , y = 6 , u = 21 , v = 21 and w = 0 .
The basic variables are y, u and v; the variables x and w are non-
basic. The objective function for this solution now has the value:
f = 150(0) + 250(6) + 0(21) + 0(2) + 0(0) = 1500,
which is much better than the earlier value of 0. This solution
yields the corner point D(0, 6) in Figure 15, which is adjacent to
O. Solving equation (13) for y, (since we choose the value of y = 6
from that equation) we obtain:
y = 6 – 1 x – 1 w, (15)
4 4

257
Algebra (Chapter 8) Linear programming

and substituting this expression for y in equation (11), (12) and

(7), we have:
u + 11 x– 1 w – 21 = 0, (16)
4 4
v + 1 x – 1 w – 2 = 0, (17)
2 2
f– 175 x + 125 w – 1500 = 0. (18)
2 2
From equation (18):
f = 1500 + 175 x – 125 w .
2 2
The value of, f can be increased by increasing only x.
Solving equation (15), (16) and (17) for y, u, and v we obtain:
y=6– 1x– 1w, (15)
4 4
u = 21 – 11 x + 1 w, (19)
4 4
1
v=2– x+ w. 1 (20)
2 2
Since we only increase x, we put w = 0, and obtain
y = 6 – 1 x , u = 21 – 11 x ,v = 2 – 1 x ,
4 4 2
so that as x increases all y, u, and v decrease. Since y ≥ 0 , u ≥ 0,
v ≥ 0 then we must have:
x ≤ 6/( 1 ) = 24, x ≤ 21 /( 11 ) = 84 , x ≤ 2/( 1 ) = 4.
4 4 11 2
The allowable increase in x can be no longer than the smaller of
the ratios 6/( 1 ), 21 /( 11 ) and 2/( 1 ). i.e. x = 4.
4 4 2
Therefore x = 4, y = 5 , u = 10, v = 0 and w = 0 .
The basic variables are x, y, and u. The non-basic variables are v
and w. The objective function for this solution now has the value:

258
Algebra (Chapter 8) Linear programming

f = 1500 + 175 x – 125 w . (18′)

2 2
i.e. f = 1500 + 175 (4) = 1850,
2
which is much better than the earlier value of D(0, 6). This
solution yields the corner point C(4 , 5) in figure 15, which is
adjacent to D(0 , 6). Solving equation (20) for x, we obtain:
x = 4 – 2v + w, (21)
and substituting this expression for x in equations (18), (15), and
(19), we have:
f = 1850 + 25w – 175v, (22)
y = 5 + 1 v – 1 w, (23)
2 2
u = 10 + v – 5 w,
11 (24)
2 2
The value of f can be, increased by increasing w.
Put v = 0, we obtain: y = 5 – 1 w, u = 10 – 5 w, x = 4 + w.
2 2
Since y ≥ 0, u ≥ 0, x ≥ 0, then we must have:
w ≤ 5( 1 ) = 10 w ≥ – 4, w ≤ 10/( 5 ) = 4.
2 2
Since w must be non-negative, we have w ≥ – 4, which put no
limitation at all on how far w can be increased. Thus, in forming
the ratios we only use the positive entries. The allowable increase
in w can be no longer than the smaller of the ratios 5/( 1 ) and
2
5
10/( ) i.e. w = 4. Therefore x = 8, y = 3, u = 0 , v = 0 and w = 4.
2
The basic variables are x, y, and w. The non-basic variables are u
and v. The objective function for this solution has the value

259
Algebra (Chapter 8) Linear programming

f = 1850 + 25w – 175 v, (22)

i.e. f = 1850 + (25)(4) = 1950,
which is much better than the earlier value of C(4 , 5). This
solution yields the corner point B(8 , 3) in figure 15, which is
adjacent to C(4 , 5). Solving equation (24) for w, we obtain:
w = 4 + 11 v – 2 u, (25)
5 5
and substituting this expression for w in equations (21), (22) and
(23) we have :
f = 1950 – 10u – 120v, (26)
y = 3 + 1 u – 3 v, (27)
5 5
2 1
x = 8 – u + v. (28)
5 5
Since f has no positive entries with variables u and v, we conclude
that we are finished and that the indicated solution is optimal.
Thus the optimal solution is:
x = 8 , y = 3, u = 0, v = 0 and f = 1950,
which corresponds to the corner B(8 , 3). Thus the simplex
method started from the corner point O(0 , 0) moved to the
adjacent corner point D(0 , 6) and then to the corner point C(4 , 5)
which is adjacent to D, and finally to the adjacent corner point
B(8 , 3), which is adjacent to C. The value of the objective
function increased from 0 to 1500 to 1850 and to 1950
respectively.
It is convenient to develop a tabular method for displaying

260
Algebra (Chapter 8) Linear programming

the given problem and the initial basic solution. Let us consider
the original L.P.P., (8.1) and (8.2). The initial tableau of the
maximum L.P.P. is as follows:
All
Basic x1 x2 … xn u1 u2 … um *
u1 a11 a12 … a1n 1 0 … 0 b1
u2 a21 a22 … a2n 0 1 … 0 b2
… … … … … … … … … …
um am1 am2 … amn 0 0 … 1 bm
f –cn –cn … –cn 0 0 … 0 0

indicators
The entries in the last row excluding the last entry are called
indicators and are the negatives of the coefficients of the objective
function f in Equation (8.4). The entry in the lower right-hand
corner is zero it corresponds to the value of the objective functions
f at the origin.
Pivot entry of a simplex tableau:
A pivot entry or simply a 3pivot of a simplex tableau is obtained
as follow:
i) Select a column that contains the most negative indicator this
column is called pivotal column, let it be j column.
ii) Form the ratios of the entries bi above the objective row in the
right most column of the tableau by the corresponding entries in
the pivotal column a (only positive numbers) i.e. bi /aij.

261
Algebra (Chapter 8) Linear programming

iii) the entry aij which yields the smallest ratio in step (ii) is the
pivot, its column is called the pivotal column and its row the
pivotal row.
If the tableau has no negative indicators then it is a
terminal tableau and has no pivot. Properly interpreted the
terminal tableau yields a solution to our problem.
iv) Locate and circle the pivot. If the pivot is k, multiply the
pivotal row by a/k, making the entry that was the pivot a “1”.
v) Add appropriate multiplies of the pivotal row to all other rows
(including the objective row) so that all elements in the pivotal
column except for the “1” where the pivot was located become
zero.
iv) In the new tableau replace the label on the pivotal row by the
variable corresponding to the pivotal column.
This process is repeated successively until there is no more
negative indicators. i.e. we have a terminal tableau. The resulting
solution is termed as the basic feasible solution or simple optimal
solution. Suppose after performing the algorithm the simplex
method on the L.P.P., we obtain the following terminal tableau:
xi
Some labeled *… *… *
* *…*
with xi xi 1 pi
f *… *… * q1q2…qm v

262
Algebra (Chapter 8) Linear programming

Then i) v, the entry in the right-hand corner of the tableau is the

maximum value of f.
ii) The point P(P1, P2 , …, Pn ), where
⎧the last entry of a row labeled x i
pi ⎨
⎩ 0 if no row is labeled x i
is the optimum solution of the maximum problem.
iii) There is a minimum problem corresponds to our maximum
problem known by dual problem and so by the duality theorem in
§ 8.5, the minimum value of the objective function of the dual
problem g is v. i.e. min g = v.
iv) The point Q(q1, q2,…,qm) is the optimum solution of the
minimum problem; its complements are the last entries in the
column corresponding to the slack variables.
Example 8. 10:
Find the optimum solution Q of the minimum problem, the
optimum solution P of the maximum problem and v = max f =
= min g, according to each of the following terminal tables:
i) ii)

i) Q(1, 6, 4) , P(3, 0, 5), v = max f = min g = 23.

263
Algebra (Chapter 8) Linear programming

Note that the second component of P is 0 since no row is labelled

x2.
ii) Q(7, 9), P(3, 0, 0) , v = max f = min g = 17.
Note that second and third components of P are each 0 since no
row is labelled x2 or x3.
Example 8.11:
Applying the simplex method to Example 8.6, we obtain the
following sequence of tables:
Equations (7) – (10) of Example 8.9 give us the initial table:

(Table 1)
The variables x, y, u, v, w, and f are written in the top row as
labels on the corresponding columns. Constraints (8), (9) and (10)
are entered in the top three rows followed by equation (7) in the
bottom row (which is called objective row). Along the left-hand
side of the table we indicate which variable is a basic variable in
the corresponding equation. In the table, the value of the basic
variable is explicitly given in the column № 7, (which is before
the right most column). A basic variable can also be described as a
variable other than f, which is present in exactly one equation and
264
Algebra (Chapter 8) Linear programming

there it appears with a coefficient of 1. The initial table shows the

values of the basic variable u, v, and w, and therefore the non-
basic variables have values x = 0 and y = 0. The value of the
objective function for this initial basic feasible solution is:
f – (150)(0) – (25)(0) + u(0) + v(0) + w(0) = 0. i.e. f = 0.
Which is the entry in the objective row and column № 7. It is clear
that this solution is not optimal, for using the bottom row of the
initial table. The entries in the last row, excluding the last entry,
are called indicators and are the negatives of the coefficients of the
objective function f in equation (7). The entry in the lower right-
hand corner is zero; it corresponds to the value of the objective
function f at the origin.
We remark that the marked 4 is a pivot entry, since it
appears in a column with the most negative indicator –250 ≤ –150,
and 24/4 is less than both 14/2 and 27/2. We now calculate the
new table step by step:
i) Divide each of the pivotal row (row of w) by the pivot 4 to
obtain the new row.
ii) Perform elimination of the other entries in the pivotal column
the new table is:

265
Algebra (Chapter 8) Linear programming

(Table 2)
In table 2, the marked 1/2 is a pivot entry, divide each entry of
the pivotal row by the pivot 1/2 to obtain the row 1, 0, 0, 2, –1, 4.
The new table is:

(Table 3)
In table 3, it is clear that the marked 5/2 is a pivot entry, divide
each entry of the pivotal row by the pivot 5/2 to obtain the row 0,
0, 2/5, –11/2 , 1, 4.
The new table is:

(Table 4)
266
Algebra (Chapter 8) Linear programming

Note that the new table (4) is a terminal table, i.e. it has no
negative indicators. Thus by the terminal table 4, the optimum
solution is x = 8 , y = 3 and the maximum value of f is 1950. This
agrees with the previous solution of the problem. Observe that the
numbers 0 , 1500 , 1850 and 1950 which appear respectively in
the lower right-hand corners of the tables 4 correspond precisely
to the values of the objective function at the points O(0, 0), D(4, 5)
and B(8 , 3) in Figure 15.
The simplex method begins with the initial table and then
calculates new tables one after the other until a terminal table is
obtained. We remark that in each step at most one row, the pivotal
row, changes its label (basic variable); the columns keep the same
label throughout.
We remark that, in table 1, if we choose the column of x as
the pivotal column and so obtain a different pivot, then we would
calculate a different new table. This corresponds to the fact that in
moving from corner to corner there may be more than one edge on
which the objective function increases in value; hence there is
more than one way to finally reach the optimum solution. This
will happen if we don’t restrict ourselves to choose the most
negative indicator, unfortunately, this new routine has mainly a
theoretical value and of no further interest. There isn’t a practical
gain. Thus we proceed in normal simplex style.

267
Algebra (Chapter 8) Linear programming

§ 8.5 Duality method:

Duality theory is one of the most important and interesting
concepts in mathematics in general and in linear programming as a
special case. The basic idea behind the duality theory is that every
L.P.P. has associated with it another (related) problem involving
the same data that describes the original problem. The given
original problem is called the primal (P). This problem can be
solved, by transposing the rows and columns of its algebraic
statements. Inverting the problem in this way results in the dual
problem (D). It may be noted that, in general, either the problem
may be considered the primal, with the other as the dual. The two
problems have very closely related properties so that optimal
solution to the dual gives complete information about the optimal
solution to the primal and vice versa.
Let us write the original L.P.P. in the following way:
Maximize f = c1 x1+ c2 x2+ …+ cnxn , (8.4)
subject to the constraints :
a11 x1 + a12 x2 + a13 x3 +… a1n xn ≤ b1,
⎫
a21 x1 + a22 x2 + a23 x3 +… a2n xn ≤ b2, ⎪
⎪
⎬ (8.5)
… … … … … ⎪
am1 x1 + am2 x2 + am3 x3 +… amn xn ≤ bm. ⎪⎭

xi ≥ 0 for i = 1, 2, …,n.
If the above problem is referred to as the primal then its associated
dual will be:
268
Algebra (Chapter 8) Linear programming

Minimize g = b1 y1+ b2 y2+ …+ bmym, (8.6)

subject to:
a11 y1 + a21 y2 + a31 y3 +… am1 ym ≥ c1,
⎫
a12 y1 + a22 y2 + a32 y3 +… am2 ym ≥ c2, ⎪
⎪
⎬ (8.7)
… … … … … ⎪
a1n y1 + a2n y2 + a3n y3 +… anm ym ≥ cn. ⎪⎭

yj ≥ 0 for j = 1, 2, …,m.
Because all of x’s are greater than or equal to zero we can
multiply the successive inequality of (8.7) by x1, x2, …, xn and
add we obtain after rearrangement of terms:
x1 (a11 y1 + a21 y2 + a31 y3 +… am1 ym ) +
+ x2 (a12 y1 + a22 y2 + a32 y3 +… am2 ym ) + …
+ xn (a1n y1 + a2n y2 + a3n y3 +… anm ym ) ≥
≥ c1x1+ c2x2+ …+ + cnxn = f
i.e.
y1 (a11 x1 + a12 x2 + a13 x3 +… a1n xn) +
+ y2 (a21 x1 + a22 x2 + a23 x3 +… a2n xn) + …
+ yn (am1 x1 + am2 x2 + am3 x3 +… amn xn ) ≥
≥ c1x1+ c2x2+ …+ + cnxn = f . (8.8)
Also from (8.5) each of the quantities in brackets is less than or
equal to the corresponding b1, b2,…, bm.
It follows that:
g = b1 y1+ b2 y2+ …+ bmym ≥

269
Algebra (Chapter 8) Linear programming

≥ y1 (a11 x1 + a12 x2 + a13 x3 +… a1n xn) +

+ y2 (a21 x1 + a22 x2 + a23 x3 +… a2n xn) + …
+ yn (am1 x1 + am2 x2 + am3 x3 +… amn xn ) ≥
≥ c1x1+ c2x2+ …+ + cnxn = f , (8.9)
that is g ≥ f for all (x1, x2,…, xn ) and (y1, y2,…, ym) satisfying the
conditions (8.5) and (8.7).
It follows that if we find a set of x’s and a set of y’s for
which g = f , we have the maximum f and the minimum g. If g
weren’t a minimum the minimum g would be less than some f
which is impossible and conversely.
The constraints relationships of primal and its dual can be
represented in a single table shown below:

(8.10)

The above table (8.10) represents the primal constraints if

read horizontally and the dual ones if read vertically. The
variables x1, x2,…, xn are written above the corresponding
columns, while the variables y1, y2,…, ym are placed to the left of
the corresponding rows. Constraint coefficients associated with
each row are the primal constraint coefficients, while those

270
Algebra (Chapter 8) Linear programming

associated with each column are the dual constraint coefficients.

Example 8.12:
Construct the dual of the following problem:
Minimize g = 3y1 – 2y2 + 4y3.
Subject to
3y1 + 5y2 + 4y3 ≥ 7, (1)
6y1 + y2 + 3y3 ≥ 4, (2)
7y1 – 2y2 – y3 ≤ 10, (3)
4y1 + 7y2 – 2y3 ≥ 2, (4)
y1 , y2 , y3 ≥ 0 . (5)
Solution:
As the given problem is of minimization all constraints
should be of ≥ type. Multiplying the third constraint (3)
throughout by –1,
–7y1 + 2y2 + y3 ≥ –10, (3′)
The dual of the given problem will be:
Maximize f = 7x1 + 4x2 – 10 x3 + 2x4.
Subject to
3x1 + 6x2 – 7x3 + 4x4 ≤ 3,
5x1 + x2 + 2x3 + 7x4 ≤ –2,
4x1 + 3x2 + x3 – 2x4 ≤ 4,
where x1, x2, x3 and x4 are the dual variables associated with the
first, second third and fourth constraint respectively of the primal.

271
Algebra (Chapter 8) Linear programming

Example 8.13:
We solve the problem in example 8.7, by the simplex method :
i) Minimize g = 1.1 x + 0.5 y , subject to
2x + y ≥ 8,
6x + y ≥ 12,
x + 3y ≥ 9,
x ≥ 0 , y ≥ 0.
ii) Minimize h = 0.9x + 0.5y,
subject to the same constraints of (i).
Solution:
Minimize g = 1.1 x + 0.5 y ,
subject to
2x + y ≥ 8, Primal
6x + y ≥ 12,
x + 3y ≥ 9,
x ≥ 0 , y ≥ 0.
Maximize G = 8X + 12Y + 9Z, Dual
subject to 2X + 6Y + Z ≤ 1.1,
X + Y + 3Z ≤ 0.5,
X, Y, Z ≥ 0 .
We now solve the dual problem independently by the
simplex method. A comparison of the optimal primal and dual
tables would show how the optimum solution of one problem

272
Algebra (Chapter 8) Linear programming

could be obtained directly from the other. The following table

presents the simplex iterations leading to the optimal primal
solution:
X Y Z U V G
2 6 1 1 0 1.1
1 1 3 0 1 0.5
–8 –12 –9 0 0 0
1 1 1 11
Y 3 1 6 6 0 60
2 17 –1 19
3 0 6 6 1 60

–4 0 –7 2 0 2.2
5 3 –1 14
Y 17 1 0 17 17 85
4 –1 6 19
Z 17 0 1 17 17 170
– 40 0 0
27 42 707
17 17 17 170

Y 0 1 –5 1 –1 1
4 4 2 40
17 –1 3 19
X 1 0 4 4 2 40
0 0 10 1 6 4.1
The optimal dual solution obtained directly by the simplex method
is given in the terminal table as
X = 19 , Y = 1 , Z = 0, G = 4.1,
40 40
and the optimal primal solution is given as :
x = 1 , y = 6 , g = 4.1 ,
which are the same values as those obtained from the graphical

273
Algebra (Chapter 8) Linear programming

method in example 8.7.

ii) The following table presents the simplex iterations leading to
the optimal primal solution of the second case:
ii) X Y Z U V H
2 6 1 1 0 0.9
1 1 3 0 1 0.5
–8 –12 –9 0 0 0
1 1 1 3
Y 3 1 6 6 0 20
2 17 –1 7
3 0 6 6 1 20

–4 0 –7 2 0 1.8
5 3 –1 22
Y 17 1 0 17 17 170
4 –1 6 21
Z 17 0 1 17 17 170
– 40 0 0
27 42 453
17 17 17 170

Y 0
17
0
3 –1 22
5 5 2 50
–4 –1 31 1
X 1 5 1 5 85 50
0 8 0 3 2 3.7
From the terminal table of the case (ii):
X = 22 , Y = 0 , Z = 1 , H = 3.7
50 50
and the optimal primal solution is given as:
x = 3 , y = 2 , h = 3.7,
which are the same values as those obtained from the graphical
method in example 8.7.

274
Algebra (Chapter 8) Linear programming

EXERCISES 6
1) Sketch the set of points satisfying the given system of
inequalities:
a) x ≤ 6, y ≥ 2 , y ≤ 4, y ≥ 1 & x + y ≤ 6.
b) 2x – y ≤ 6 , 2x + y ≤ 10, x ≥ 0 , y ≥ 0.
c) x + y ≤ 3, 5x + 4y ≥ 20 , x ≥ 0 , y ≥ 0.
d) x + y ≥ 4, x + 4y ≥ 8 , x ≥ 0, y ≥ 0.
2) Solve the following problem graphically:
Maximize f = 3x + 2y ,
subject to 2x – 3y ≤ 6, x + y ≤ 4, x ≥ 0 , y ≥ 0.
[Answer: x = 18/5, y = 2/5, o.v.f. = 58 /5]
3) Solve the following problem graphically:
Minimize g = 3x – y,
subject to –3x + 2y ≤ 6, 5x + 4y ≥ 20, 8x +3y ≤ 24, x ≥ 0, y ≥ 0.
[Answer: x = 8/11, y = 45/11, o.v.g. = –21 /11]
4) Solve the following problem graphically:
a) Maximize f = 5x + 2y ,
subject to x + y ≤ 10, x = 5, x ≥ 0 , y ≥ 0.
[Answer: x = y = 5, o.v.f. = 35 ]
b) If the constraint x = 5 is changed to x ≤ 5, determine all the
feasible extreme points and find the optimum by evaluating the
objective function numerically at each point. Repeat the procedure
with x = 5 replaced by x ≥ 5.

275
Algebra (Chapter 8) Linear programming

[Answer: x = 5 : {(5, 0) (5, 5)}; optimum at (5 , 5) with f = 35.

x ≤ 5:{(0,0), (5, 0), (5, 5), (0 , 10)}; optimum at (5, 5) with f = 35.
x ≥ 5: {(5,0), (5, 5), (10 , 0)}; optimum at (10 , 0) with f = 50.]
5) Consider the following problem:
Maximize f = 6x – 2y, subject to x – y ≤ 1, 3x – y ≤ 6, x ≥ 0 , y≥ 0.
Show graphically that at the optimal solution the variables x and y
can be increased infinitely while the value of the objective
function remains constant.
6) Consider the following linear programming problem:
Maximize f = 4x + 4y,
subject to 2x + 7y ≤ 21, 7x + 2y ≤ 49, x, y ≥ 0.
Find the optimal solution (x*, y*) graphically. What are the ranges
of variation of the coefficients of the objective function f = ax +by
that will keep (x*, y*) optimal? [Answer: 2/7 ≤ a/b ≤ 7/2]
7) Solve the following problem graphically:
Maximize f = 5x + 6y ,
subject to x – 2y ≥ 2, –2x + 3y ≥ 2 , x, y unrestricted in sign.
[Answer: x = –10, y = –6, f = –86]
8) Sketch the set of points satisfying the given system of
inequities:
x + y ≤ 5 , y – x ≤ 1, x – 2y ≤ 2 , y ≥ 0 , y ≤ 3 , x ≥ 0 , x ≥ 1, and
identify all redundant constraints. Determine the optimum solution
assuming that the objective function is as given below:

276
Algebra (Chapter 8) Linear programming

a) Minimize g1 = 2x + 6y,
b) Maximize f1 = –3x + 4y,
c) Minimize g2 = 3x + 4y,
d) Minimize g3 = x – 2y,
e) Minimize g4 = x,
f) Maximize f2 = x..
[Answers: a) at (1, 0), g1 = 2; b) at (2, 3), f = 6; c) at (1, 0), g2 = 3;
d) at (2, 3); g = – 4, e) at (1, 0) or (1, 2); g4 =1; f) at (4,1 ), f = 4]
9) Solve the following problem graphically:
Maximize f = 2x + 3y,
subject to 3x + 5y ≤ 6, 2x + 3y ≤ 7, x ≥ 0 , y ≥ 0.
[Answer: x = 2 , y = 0 , f = 4 ]
10) Solve the following problem graphically:
Maximize f = 2x + 5y ,
subject to 2x – 3y ≤ 4 , x – 2y ≤ 6 , x ≥ 0 , y ≥ 0.
[Answer: No finite optimal solution.]
Solve the following linear programming problems by the simplex
method or dual simplex method:
11) Maximize f = 2x – 4y + 5z,
subject to 3x + 2y + z ≤ 6, 3x – 6y + 7z ≤ 9, x ≥ 0, y ≥ 0, z ≥ 0.
[Answer: x = 0, y = 33/20, z = 27/ 0 ,o.v.f. = 69/10.]
12) Maximize f = x + 2y – z + 5t,
subject to 2x + 3y + z – t ≤ 8 , 3x + y – 4z + 5t ≤ 9 ,

277
Algebra (Chapter 8) Linear programming

x ≥ 0, y ≥ 0, z ≥ 0 , t ≥ 0.
[Answer: x = y = 0, z = 49 , t = 41, o.v.f. = 156.]
13) Maximize f = 8x + 5y ,
subject to 2x + y ≤ 7, 4x – y ≤ 4 , 4x + 3y ≤ 5, x ≥ 0, y ≥ 0.
[Answer: x = 17/16, y = ¼, o.v.f = 39/4. ]
14) Minimize g = 7x + 4y + 5z ,
subject to 2x + 4y + 4z ≥8 , x – y + 3z ≥ 5, x ≥ 0, y ≥ 0, z ≥ 0.
[Answer: x = 0, y = ¼, z = 7/4, o.v.f = 39/4]
15) Maximize f = 4x – 2y – z,
subject to x + y + z ≤ 3, 2x + 2y + z ≤ 4, x – y ≤ 0,
x ≥ 0, y ≥ 0, z ≥0. [Answer: x =1, y =1, z = 0 o.v.f = 2]
16) Maximize f = 2x –3y + z,
subject to 3x + 6y + z ≤ 6, 4x + 2y + z ≤ 4, x – y + z ≤ 3,
x ≥ 0, y ≥ 0 , z ≥ 0. [Answer: x = 1/3 , y = 0, z = 8/3, o.v.f =10/3.]
17) Maximize f = 2x – y + 5z,
subject to 2x + 4y + z ≤ 6, 2x – 2y +3z ≤ 5, x +3y + 2z ≤ 2,
x ≥ 0, y ≥ 0, z ≥ 0.
[Answer: x = 0, y = 0, z =1, max f = 5.] .
18) Minimize g = 6x + 5y + 2z,
subject to x + 3y + 2z ≥ 5, 2x +2y + z ≥ 2, 4x –2y +3z ≥–1,
x ≥ 0, y ≥ 0, z ≥ 0. [Answer: x = 0, y = 0, z = 5/2, min g = 5]
19) Maximize f = 9x + 2y + 5z,
subject to 2x + 3y – 5z ≤ 12, 2x – y + 3z ≤ 3, 3x + y – 2z ≤ 2,

278
Algebra (Chapter 8) Linear programming

x ≥ 0, y ≥ 0, z ≥ 0. [Answer: x = 0, y = 12, z = 5, max f = 49.]

20) Minimize g = 4x + 5y + z,
subject to x + y + z ≥ 3, 2x + 2y – z ≥ 2, x + 2y ≥ 4,
x ≥ 0, y ≥ 0, z ≥ 0. [Answer: x = 0, y = 2, z = 1, min. f = 11.]
Solve the following L.P.P. by the graphical method:
2 2
21) A tailor has 80 m of cotton material and 120 m of wool
2 2
material. A suit requires 1 m of cotton and 3 m of wool, and a
2
dress requires 2 m of each. How many of each garment should the
tailor make to maximize his income if:
i) A suit and dress each sells for L.E. 150,
ii) A suit sells for L.E. 200 and a dress for L.E. 100,
iii) A suit sells for L.E. 50 and a dress for L.E. 100.
[Answers: i) 20 suits, 30 dresses, income L.E. 7500,
ii) 40 suits, and no dresses, income L.E. 8000,
iii) Either 20s, 30d or 40s and no dresses, income L.E. 6000]
22) A truck rental company has two types of trucks; type A has 2
3 3
m of refrigerated space and 4 m of non-refrigerated space, and
3
type B has 3 m each of refrigerated and non-refrigerates space. A
3 3
food plant must ship 90 m of refrigerated produce and 120 m of
non-refrigerated produce. How many trucks of each type should
the plant rent so as to minimize cost if:
i) Truck A rents for L.E. 0.3 per km and B for L.E. 0.4 per km,
ii) Truck A rents for L.E. 0.3 per km and B for L.E. 0.5 per km,

279
Algebra (Chapter 8) Linear programming

iii) Truck A rents for L.E. 0.2 per km and B for L.E. 0.3 per km?
[Answers: i) Type A = 5, type B = 20, total cost of L.E. 12.5 per
km, ii) Type A = 45 and no one of type B, total cost of L.E. 13.5
per km. iii) Either type A = 5, type B = 20, or type A = 45,
type B = 0, total cost of L.E. 9 per km.]
23) A company owns two mines; mine A produces 1 ton of high-
grade ore, 3 tons of medium grade ore and 5 tons of low grade ore
each day; and mine B produces 2 tons of each of the three grades
of are each day. The company needs 80 tons of high-grade ore,
160 tons of medium grade ore, and 200 tons of low-grade ore.
How many days should each mine be operated if it costs L.E. 200
per day to work each mine?
[Answer: 40 days mine A, 20 days mine B, for total cost of L.E.
12000]
24) Suppose in the preceding problem, 23, the production cost at
mine A is L.E. 250 per day and at mine B is L.E. 150 per day.
How should the production schedule be changed?
[Answer: 20 days mine A, 50 days mine B, for total cost of L.E.
2500]
25) A baker has 50, 90, and 50 units of ingredients A, B, and C
respectively. A loaf of bread requires 1, 1, and 2 units of A, B, and
C respectively; and a cake requires 5, 2, and 1 units of A, B, and C
respectively.

280
Algebra (Chapter 8) Linear programming

i) If a loaf of bread sells for L.E. 0.35 and a cake for L.E. 0.8, how
many of each should he bake and what would his gross income
be?
ii) If the price of bread is increased to L.E. 0.45 per loaf and a
cake still sells for L.E. o.8, how many of each should he then bake
and what would be the gross income?
[Answers: i) 50 leaves of bread and 20 cakes for L.E. 33.5,
ii) 70 leaves of bread and 0 cakes for L.E. 39.5]
3
26) An oil company requires 9000, 2000, and 26000 m of high,
medium and low-grade oil, respectively. It owns two oil refineries,
say A and B. A produces 100, 300, and 400 m of high, medium
and low-grade oil, respectively per day; and B produces 200, 100,
3
and 300 m of high, medium and low grade oil respectively per
day.
i) If each refinery costs L.E. 200 per day to operate, how many
days should each be run to minimize the cost and meet the
requirements?
ii) If the cost at A rises to L.E. 300 per day and B remains at L.E.
200 per day, how many days should each be operated?
[Answers: i) A run 50 days and B run 20 days at cost = L.E.
14000, ii) A run 20 days and B run 60 days at cost = L.E. 18000]
27) A manufacturer has 240, 370 and 180 kg of wood (w), plastic
(P) and steel (S), respectively. Product A requires 1, 3 and 2 kg of

281
Algebra (Chapter 8) Linear programming

W, P, and S respectively; and product B requires 3, 4, and 1 kg of

W, P, and S respectively.
i) If A sells for L.E. 4 and B for L.E. 6, how many of each should
be made to obtain the maximum gross income? ,
ii) If the price of B drops to L.E. 5 and A remains at L.E. 4, how
many of each should be then be made to obtain the maximum
gross income?
[Answers: i) 30 of A and 70 of B for L.E. 540, ii) 70 of A and 40
of B for L.E. 480]
28) Suppose that 8, 2, and 9 units of proteins (P), carbohydrates
(C) and fats (F) respectively are the minimum weekly
requirements for a person. Food A contains 2, 6, and 1 units of P,
C, and F respectively per kg; and food B contains 1, 1, and 3 units
respectively per kg.
i) If A costs L.E. 0.85 per kg and B costs L.E. 0.4 per kg, how
many kg of each should he buy to minimize his cost and still meet
his minimum requirements?
ii) If the price of A drops to L.E. 0.75 per kg while B still sells at
L.E. 0.4 per kg, how many kg of each should he then buy?
[Answers: i) 1 of A and 6 of B for L.E. 3.25, ii) 3 of A and 2 of B
for L.E. 3.05]
29) A carpenter has 90, 80, and 50 linear meters of plywood (W),
pine (P) and birch (B), respectively. Product A requires 2, 1, and 1

282
Algebra (Chapter 8) Linear programming

linear meters of W, P, and B respectively; and product B requires

1, 2, and 1 linear meters of W, P, and B respectively.
i) If A sells for L.E. 12 and B sells for L.E. 10, how many of each
should he make to obtain the maximum gross income?
ii) If B sells for L.E. 12 and A sells for L.E. 10, how many of each
should he make?
[Answers: i) 40 of A and 10 of B for L.E. 580, ii) 20 of A and 30
of B for L.E. 560]
30) Old hens can be bought for L.E. 2 each but young ones cost
L.E. 5 each. Old hens lay 3 eggs per week and young ones lay 5
eggs per week, each being worth L.E. 0.3. A hen costs L.E. 1 per
week to be fed. If I have only L.E. 80 to spend on the hens, how
many of each kind should I buy to give a profit of more than L.E.
6 per week, assuming that I can’t house more than 20 hens?
[Answer: x = 0, y = 16, max. f = L.E. 8]

283
Algebra (Chapter 8) Linear programming

SUGGESTIONS FOR FURTHER READING

REFERENCES
1) The Tutorial Algebra, volumes I and II,
by W. Briggs, G.H. Bryan and G. Walker.
2) Advanced Algebra volumes, I, II and III.
By C.V. Durell and A. Robson.
3) College Algebra, by N.H. Phadke and P.K. RAO.
4) College Algebra, (Theory and Problems) Schaum's outline
series by Murray R. Spiegel.
5) Finite Mathematics, (Theory and Problems) Schaum's outline
series by Seymour Lipschutz.
6) Matrices (Theory and Problems) Schaum's outline series by
Frank Ayres.
7) Advanced Mathematics for Engineers and Scientists (Theory
and problems) Schaum’s outline series by Murray R. Spiegel.
8) Numerical Analysis, (Theory and Problems) Schaum's outline
series by Francis Scheid.
9) Operations Research (Theory and Problems) Schaum's outline
series by Richard Branson.
10) Introductory Linear Algebra with Applications by Bernard
Kolman.

284
Algebra References

REFERENCES
SUGGESTIONS FOR FURTHER READING
1) The Tutorial Algebra, volumes I and II,
by W. Briggs, G.H. Bryan and G. Walker.
2) Advanced Algebra volumes, I, II and III.
By C.V. Durell and A. Robson.
3) College Algebra, by N.H. Phadke and P.K. RAO.
4) College Algebra, (Theory and Problems)
Schaum's outline series by Murray R. Spiegel.
5) Finite Mathematics, (Theory and Problems)
Schaum's outline series by Seymour Lipschutz.
6) Matrices (Theory and Problems)
Schaum's outline series by Frank Ayres.
7) Advanced Mathematics for Engineers and Scientists
(Theory and problems) Schaum’s outline series
by Murray R. Spiegel.
8) Numerical Analysis, (Theory and Problems)
Schaum's outline series by Francis Scheid.
9) Operations Research (Theory and Problems)
Schaum's outline series by Richard Branson.
10) Introductory Linear Algebra with Applications
by Bernard Kolman.

233
‫ﺭﻗﻢ ﺍﻹﻳﺪﺍﻉ ﺑﺪﺍﺭ ﺍﻟﻜﺘﺐ ﻭﺍﻟﻮﺛﺎﺋﻖ ﺍﻟﻘﻮﻣﻴﺔ‬

Algebra
(Selected Topics)
By Dr. Mohamed Ismail
Mohamed Hessein
18012/2001