Algebra Through Problem Solving

by
Abraham P. Hillman Gerald L. Alexanderson
University of New Mexico Santa Clara University

This PDF file designed by Mervin E. Newton of Thiel College while on sabbatical at Los Alamos
National Laboratory. An interactive version of this text is available on the World Wide Web at
http://WWW.Education.lanl.gov/resources/ATPS/. Revised 5/24/00

i
 Copyright ©

Algebra Through Problem Solving

by
Abraham P. Hillman and Gerald L. Alexanderson

Copyright © 1966 by Allyn and Bacon, Inc.

150 Tremont Street, Boston

Copyright assigned to Abraham P. Hillman and Gerald L. Alexanderson on January 25, 1971;
assignment recorded in Library of Congress Vol. 1404, Pages 109-111.

Produced in the United States of America.

ii
 To

Josephine Hillman

iii
 PREFACE

This book is an outgrowth of the authors' work in conducting problem solving seminars for
undergraduates and high school teachers, in directing mathematics contests for undergraduates and high
school students, and in the supervision of an undergraduate research participation program. Their
experience has shown that interest in and knowledge of mathematics can be greatly strengthened by an
opportunity to acquire some basic problem solving techniques and to apply these techniques to
challenging problems for which the prerequisite knowledge is available.

Many students who have not had this opportunity lose confidence in themselves when they try
unsuccessfully to solve non-routine problems such as those in the Mathematics Magazine or in the
Putnam Intercollegiate Mathematics Competitions, conducted by the Mathematical Association of
America. Those who gain self-confidence by work on challenging material at the proper level also
generally have increased motivation for mastering significant mathematical concepts and for making
original contributions to mathematical knowledge.

The topics chosen for this book are particularly appropriate since they are at a fairly elementary
level and exhibit the interdependence of mathematical concepts. Many generalizations are suggested in
the problems; the perceptive reader will be able to discover more.

The authors express their debt to all who have influenced this effort. We are especially grateful
to Leonard Klosinski, Roseanna Torretto, and Josephine Hillman for their invaluable assistance.

Albuquerque, New Mexico Abraham P. Hillman

Santa Clara, California Gerald L. Alexanderson

iv
 TABLE OF CONTENTS

COVER PAGE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i

COPYRIGHT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii

DEDICATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

PREFACE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv

INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii

1. THE PASCAL TRIANGLE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Problems for Chapter1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2. THE FIBONACCI AND LUCAS NUMBERS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

Problems for Chapter 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3. FACTORIALS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Problems for Chapter 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

4. ARITHMETIC AND GEOMETRIC PROGRESSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

Problems for Chapter 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

5. MATHEMATICAL INDUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Problems for Chapter 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

6. THE BINOMIAL THEOREM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

Problems for Chapter 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

7. COMBINATIONS AND PERMUTATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

Problems for Chapter 7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

8. POLYNOMIAL EQUATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
8.1 The Factor and Remainder Theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
8.2 Integral Roots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
8.3 Rational Roots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Problems for Sections 8.1, 8.2, and 8.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
8.4 Symmetric Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Problems for Section 8.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

v
9. DETERMINANTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
9.1 Determinants of Order 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
9.2 Determinants of Order 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Problems for Sections 9.1 and 9.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
9.3 Determinants of Order n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Problems for Section 9.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
9.4 Vandermonde and Related Determinants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Problems for Section 9.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

10. INEQUALITIES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
10.1 Elementary Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Problems for Section 10.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
10.2 Further Inequalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
Problems for Section 10.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
10.3 Inequalities and Means . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
10.4 The Cauchy-Schwarz Inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
Problems for Sections 10.3 and 10.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

ANSWERS TO THE ODD-NUMBERED PROBLEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

INDEX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

vi
 INTRODUCTION

Knowledge of mathematics together with the ability to apply this knowledge to non-routine
problems will be very valuable in the more and more automated world we face. Routine problems will
tend to be solved mechanically, while new and challenging problems arise at a rapid rate for human
minds to solve.

We hope that this book will be helpful in the process of mastering some aspects of mathematics
and becoming proficient in using this knowledge. The topics selected contain important ideas that are
often lost in the regular curriculum. An effort has been made to develop most of the theoretical material
through sequences of related and progressively more sophisticated problems that follow the necessary
text material and illustrative examples. In later chapters some proofs of a considerably more involved
nature are omitted or dealt with only in special cases. The sets of problems provide opportunities for
recognizing mathematical patterns and for conjecturing generalizations of specific results.

And now some words of advice to prospective problem solvers: some of the problems of this
book are easy, some may take longer to solve than any previously encountered, and some may prove
to be too difficult. If a problem is difficult, it may be helpful to look at the surrounding problems. One
does not have to do all of the problems in a chapter before going on. In fact, some of the hard
problems will appear to be easier if one returns to them after progressing through later chapters. The
statements in problems preceded by the symbol "R" are required for subsequent work and should be
specially noted.

Answers or hints are provided for the odd-numbered problems except starred problems and
those that contain the answer in the problem. However, the greatest benefit comes from trying the
problems; one should postpone looking at the answers as long as possible.

vii
This page left blank on purpose.

viii
 Chapter 1

THE PASCAL TRIANGLE

In modern mathematics, more and more stress is placed on the context in which
statements are true. In elementary mathematics this generally means an emphasis on a clear
understanding of which number systems possess certain properties. We begin, then, by describing
briefly the number systems with which we will be concerned. These number systems have
developed through successive enlargements of previous systems.
At one time a "number" meant one of the natural numbers: 1,2,3,4,5,6,... . The next
numbers to be introduced were the fractions:1/2, 1/3, 2/3, 1/4, 3/4, 1/5,..., and later the set of
numbers was expanded to include zero and the negative integers and fractions. The number
system consisting of zero and the positive and negative integers and fractions is called the system
of rational numbers, the word "rational" being used to indicate that the numbers are ratios of
integers. The integers themselves can be thought of as ratios of integers since 1 = 1/1, -1 = -1/1,
2 = 2/1, -2 = -2/1, 3 = 3/1, etc.
The need to enlarge the rational number system became evident when mathematicians
proved that certain constructible lengths, such as the length 2 of a diagonal of a unit square,
cannot be expressed as rational numbers. The system of real numbers then came into use. The
real numbers include all the natural numbers; all the fractions; numbers such as 2, 3, 5,

3
and (2% 6)/3 which represent constructible lengths; numbers such as 2 and B which do
not represent lengths constructible from a given unit with compass and straightedge; and the
negatives of all these numbers. Modern technology and science make great use of a still larger
number system, called the complex numbers, consisting of the numbers of the form a + bi with
a and b real numbers and i2 = -1.
Our first topic is the Pascal Triangle, an infinite array of natural numbers. We begin by
considering expansions of the powers (a + b)n of a sum of two terms. Clearly, (a + b)2 =
(a + b)(a + b) = a2 + ab + ba + b2 = a2 + 2ab + b2. Then (a + b)3 = (a + b)2(a + b) =
(a2 + 2ab + b2)(a + b). We expand this last expression as the sum of all products of a term of
a2 + 2ab + b2 by a term of a + b in the following manner:

a2 + 2ab + b2
a +b
a3 + 2a2b + ab2
a2b + 2ab2 + b3
a3 + 3a2b + 3ab2 + b3

1
Hence (a + b)3 = a3 + 3a2b + 3ab2 + b3. If a … 0 and b … 0 (a is not equal to zero and b is not
equal to zero), this may be written

(a + b)3= a3b0 + 3a2b1 + 3a1b2 + a0b3

The terms of the expanded form are such that the exponent for a starts as 3 and decreases by one
each time, while the exponent of b starts as 0 and increases by one each time. Thus the sum of
the exponents is 3 in each term.
One might guess that by analogy the expansion of (a + b)4 involves a4, a3b, a2b2, ab3, and
b4. This is verified by expanding (a + b)4 = (a + b)3(a + b)= (a3 + 3a2b + 3ab2 + b3)(a + b) as
follows:

(1) a3 + 3a2b + 3ab2 + b3

(2) a +b
(3) a4 + 3a3b + 3a2b2 + ab3
(4) a3b + 3a2b2 + 3ab3 + b4
(5) a4 + 4a3b + 6a2b2 + 4ab3 + b4

Thus we see that a4, a3b, a2b2, ab3, and b4 are multiplied by 1, 4, 6, 4, 1 to form the terms
of the expansion. The numbers 1, 4, 6, 4, 1 are the coefficients of the expansion. Examination of
expressions (1) to (5), above, shows that these coefficients are obtainable from the coefficients 1,
3, 3, 1 of (a + b)3 by means of the following condensed versions of (3), (4), and (5):

(3*) 1 3 3 1
(4*) 1 3 3 1
(5*) 1 4 6 4 1

We now tabulate the coefficients of (a + b)n for n = 0, 1, 2, 3, 4, ... in a triangular array:

n Coefficients of (a + b)n
0 1
1 1 1
2 1 2 1
3 1 3 3 1
4 1 4 6 4 1
... . . . . . .

2
One may observe that the array is bordered with 1's and that each number inside the border is the
sum of the two closest numbers on the previous line. This observaion simplifies the generation of
additional lines of the array. For example, the coefficients for n = 5 are 1, 1 + 4 = 5, 4 + 6 = 10,
6 + 4 = 10, 4 + 1 = 5, and 1.
The above triangular array is called the Pascal Triangle in honor of the mathematician
Blaise Pascal (1623-1662). A notation for the coefficients of (a + b)n is

n n n
(6) , , ... , .
0 1 n

For example, one writes (a + b)4 in this notation as

4 4 4 3 4 2 2 4 4 4
a % a b % a b % ab 3 % b
0 1 2 3 4

4 4 4 4 4
where ' 1 ' , ' 4 ' , and ' 6.
0 4 1 3 2

A two-term expression is called a binomial, and an expansion for an expression such as

(a + b)n is called a binomial expansion. The coefficients listed in (6) above are called binomial
coefficients.
n
Note that the symbol denotes the coefficient of an-kbk, or of akbn-k, in the expansion
k

3
of (a + b)n. Thus is the coefficient 3 of a2b or of ab2 in the expansion of (a + b)3,and
1

 4 n
  is the coefficient 6 of x2y2 in (x + y)4. One reads as "binomial coefficient n choose k"
 2 k
or simply as "n choose k." The reason for this terminology is given in Chapter 7.
n
In Figure 1, (on page 4) we see how n and k give us the location of in the Pascal
k

n
Triangle. The number n in is the row number and k is the diagonal number if one adopts
k
the convention of labeling the rows or diagonals as 0, 1, 2, ... .

3
 FIGURE 1

n n
The formulas ' 1 ' recall the fact that the Pascal Triangle is bordered with
0 n
1's. The rule that each number inside the border of 1's in the Pascal Triangle is the sum of the two
closest entries on the previous line may be written as

n%1 n n
' % .
k k&1 k
Example 1. Expand (2x + 3y2)3.

Solution: The expansion (a + b)3= a3 + 3a2b + 3ab2 + b3 is an identity which remains true
when one substitutes a = 2x and b = 3y2 and thus obtains

(2x + 3y2)3 = (2x)3 + 3(2x)2(3y2) + 3(2x)(3y2)2 + (3y2)3

= 8x3 + 3(4x2)(3y2) + 3(2x)(9y4) + 27 y6
= 8x3 + 36x2y2 + 54xy4 + 27y6.

4 5 6 7 8 9
Example 2. Show that % % % % ' .
0 1 2 3 4 4

4 5
Solution: Using the fact that ' 1 ' and the formula
0 0

n n n%1
% ' we see that
k k&1 k

4
 8 7 6 5 4 8 7 6 5 5
% % % % ' % % % %
4 3 2 1 0 4 3 2 1 0

8 7 6 6
' % % %
4 3 2 1

8 7 7
' % %
4 3 2

8 8
' %
4 3

9
' .
4

Problems for Chapter 1

5
1. Give the value of , that is, of the coefficient of a3b2 in (a + b)5.
2

5
2. Give the value of .
4

5 5
3. Find s if ' and s is not 4.
4 s

5 5
4. Find t if ' and t is not 0.
t 0

5. Obtain the binomial coefficients for (a + b)3 from those for (a + b)2 in the style of lines (3*),
(4*), (5*) on page 2.

6. Obtain the binomial coefficients for (a + b)6 from those for (a + b)5 in the style of lines (3*),
(4*), (5*) on page 2.

7. Generate the lines of the Pascal Triangle for n = 6 and n = 7, using the technique described at
the top of page 3.

5
 8 8 8 8 8
8. Find , , , , and .
0 1 2 3 4

9 9 9 9
9. Use ' 9 and ' 36 to find and .
1 2 7 8

9 9 10 10
10. Use ' 9 and ' 36 to find and .
1 2 2 8

11. Expand (5x + 2y)3.

12. Expand (x2 - 4y2)3 by letting a = x2 and b = -4y2 in the expansion of (a + b)3.

13. Show that:

(a) (x - y)4 = x4 - 4x3y + 6x2y2 - 4xy3 + y4.

5 5 5 4 5 3 2 5 2 3 5 5 5
(b) (x - y)5 = x & x y % x y & x y % xy 4 & y .
0 1 2 3 4 5

14. Show that:

6 6 6 4 6 2 6
(a) (x + 1)6 + (x - 1)6 = 2 x % x % x % .
0 2 4 6

6 5 6 3 3 6
(b) (x + y)6 - (x - y)6 = 2 x y % x y % xy 5 .
1 3 5

15. Show that (x + h)3 - x3 = h(3x2 + 3xh + h2).

100 99 100 98 100 97 2 100 99

16. Show that (x + h)100 - x100 = h x % x h% x h %ÿ% h .
1 2 3 100

17. Find numerical values of c and m such that cx3ym is a term of the expansion of (x + y)8.

6
18. Find d and n such that dx5y4 is a term of (x + y)n.

19. Find each of the following:

2 2 2
(a) % % .
0 1 2

3 3 3 3
(b) % % % .
0 1 2 3

4 4 4
(c) 2 % 2 % .
0 1 2

5 5 5
(d) 2 % % .
0 1 2

6 6 6 6
(e) 2 % % % .
0 1 2 3

7 7 7 7
(f) 2 % % % .
7 6 5 4

20. Find the sum of the 101 binomial coefficients for n = 100 by assigning specific values to a and
b in the identity
100 100 100 99 100 98 2
(a % b)100 ' a % a b% a b %ÿ
0 1 2

100 2 98 100 100 100

% a b % ab 99 % b .
98 99 100

21. Find each of the following:

2 2 2
(a) & % .
0 1 2

4 4 4 4 4
(b) & % & % .
0 1 2 3 4

7
22. Find each of the following:

100 100 100 100 100 100

(a) & % & % ... & % .
0 1 2 3 99 100

101 101 101 101 101 101

(b) & % & %... % & .
0 1 2 3 100 101

23. Find each of the following:

4 4 4
(a) % % .
0 2 4

5 5 5
(b) % % .
0 2 4

6 6 6 6
(c) % % % .
0 2 4 6

7 7 7 7
(d) % % % .
1 3 5 7

24. Find each of the following:

1000 1000 1000 1000

(a) % % % ... % .
0 2 4 1000

1000 1000 1000 1000

(b) % % % ... % .
1 3 5 999

25. Find r, s, t, and u, given the following:

2 3 4 5 r
(a) % % % ' .
2 2 2 2 3

2 3 4 5 6 s
(b) % % % % ' .
0 1 2 3 4 4

8
 1 2 3 4 5 t
(c) % % % % ' .
1 1 1 1 1 2

3 4 5 6 u
(d) % % % ' .
0 1 2 3 3

4 5 6 7 100
26. Express % % % %þ% as a binomial coefficient.
4 4 4 4 4

5 6 7 995
27. Express % % %þ% as a binomial coefficient.
0 1 2 990

28. Show that

0 1 2 3 n&1 n
(a) n ' % % % %þ% ' .
0 0 0 0 0 1

1 2 3 n n%1
(b) % % %þ% ' .
1 1 1 1 2

n n n n %2
(c) %2 % ' for 0 # k # n & 2.
k k% 1 k% 2 k% 2

9 5 9 6 9 7
29. Show that ' ' .
4 3 3 4 2 3

10 9 7 10 6 3 10 8 4
30. Show that ' ' .
1 2 3 4 3 2 2 4 1

31. Expand (x + y + z)4 by expanding (w + z)4, then replacing w

by x + y, and expanding further.

32. Expand (x + y - z)4.

9
 2 2 2 2
3 3 3 3 2m
33. The sum of squares % % % is expressible in the form .
0 1 2 3 m
Find m.

2m
34. Express each of the following in the form :
m

4 4 4 4 4 4 4 4 4 4
(a) % % % % .
0 4 1 3 2 2 3 1 4 0

2 2 2
5 5 5
(b) 2 % % .
0 1 2

2 2 2 2
6 6 6 6
(c) % 2 % % .
3 2 1 0

35. Use (x3 + 3x2 +3x + 1)2 = [(x + 1)3]2 = (x + 1)6 to show the following:

2
3 6
(a) ' .
0 0

3 3 3 3 6
(b) % ' .
0 1 1 0 1

3 3 3 3 3 3 6
(c) % % ' .
0 2 1 1 2 0 2

3 3 3 3 3 3 3 3 6
(d) % % % ' .
0 3 1 2 2 1 3 0 3

2 2 2 2
100 100 100 100 2m
36. Express % % % þ % in the form .
0 1 2 100 m

n
37. How many of the 3 binomial coefficients with n = 0 or 1 are odd?
k

10
 n
38. How many of the 10 binomial coefficients with n = 0, 1, 2, or 3 are odd?
k

n
39. How many binomial coefficients are there with n = 0, 1, 2, 3, 4, 5, 6, or 7, and how
k
many of these are odd?

7 7 7 7 7
40. Show that all eight of the coefficients , , , ,ÿ, are odd.
0 1 2 3 7

n n n n
41. For what values of n among 0, 1, 2,..., 20 are all the coefficients , , ,ÿ,
0 1 2 n
on row n of the Pascal Triangle odd?

42. If n is an answer to the previous problem, how many of the binomial coefficients on row
n + 1 of the Pascal Triangle are odd?

* n
43. How many binomial coefficients are there with n = 0, 1, 2,..., 1022, or 1023, and
k
how many of these are even?

11
 Chapter 2

THE FIBONACCI AND LUCAS NUMBERS

The great Italian mathematician, Leonardo of Pisa (c. 1170-1250), who is known today as
Fibonacci (an abbreviation of filius Bonacci), expanded on the Arabic algebra of North Africa and
introduced algebra into Europe. The solution of a problem in his book Liber Abacci uses the sequence

(F) 0, 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, ÿ

One of the many applications of this Fibonacci sequence is a theorem about the number of steps
in an algorithm for finding the greatest common divisor of a pair of large integers. We study the sequence
here because it provides a wonderful opportunity for discovering mathematical patterns.

The numbers shown in (F) are just the beginning of the unending Fibonacci sequence. The rule
for obtaining more terms is as follows:

RECURSIVE PROPERTY. The sum of two consecutive terms in (F) is the term immediately
after them.

For example, the term after 55 in (F) is 34 + 55 = 89 and the term after that is 55 + 89 = 144.

To aid in stating properties of the Fibonacci sequence, we use the customary notation F0 , F1 , F2,
F3 , ÿ for the integers of the Fibonacci sequence. That is, F0 = 0, F1 = 1, F2 = F0 + F1 = 1, F3 = F1 + F2 =
2, F4 = F2 + F3 = 3, F5 = F3 + F4 = 5, F6 = F4 + F5 = 8, and so on. When Fn stands for some term of the
sequence, the term just after Fn is represented by Fn+1 , the term after Fn+1 is Fn+2 , and so on. Also, the
term just before Fn is Fn-1, the term just before Fn-1 is Fn-2 and so on.

We now can define the Fibonacci numbers formally as the sequence F0 , F1 , ÿ having the two
following properties.

INITIAL CONDITIONS F0 = 0 and F1 = 1.

RECURSION RULE Fn + Fn+1 = Fn+2 for n = 0, 1, 2, ÿ.

Next let S n stand for the sum of the Fibonacci numbers from F0 through Fn . That is,

S 0 = F0 = 0
S 1 = F0 + F1 =0+1 =1
S 2 = F0 + F1 + F2 =0+1+1=2
S 3 = F0 + F1 + F2 + F3 = S 2 + F3 = 2 + 2 = 4

and in general, S n = F0 + F1 + F2 + ÿ + Fn = S n-1 + Fn .

We tabulate some of the values and look for a pattern.

12
 n 0 1 2 3 4 5 6 7 ÿ
Fn 0 1 1 2 3 5 8 13 ÿ
Sn 0 1 2 4 7 12 20 33 ÿ

Is there a relationship between the numbers on the third line of this table and the Fibonacci
numbers? One pattern is that each of the terms of the sequence S 0 , S 1 , S 2 , ÿ is 1 less than a Fibonacci
number. Specifically, we have

S 0 = F2 - 1 = 0
S 1 = F3 - 1 = 1
S 2 = F4 - 1 = 2
S 3 = F5 - 1 = 4
S 4 = F6 - 1 = 7
S 5 = F7 - 1 = 12

and might conjecture that S n = Fn+2 - 1 for n = 0, 1, 2, ÿ. Does this formula hold for n = 6? Yes, since

S6 ' F0 % F1 % F2 % F3 % F4 % F5 % F6
' S 5 % F6
' (F 7 & 1) % F6
' (F 6 % F7) & 1
' F8 & 1.

The first steps in proving our conjecture correct for all the terms in the unending sequence S 0 , S 1 ,
S 2 , ÿ are rewriting the recursion formula Fn + Fn+1 = Fn+2 as Fn = Fn+2 - Fn+1 and then using this to replace
each Fibonacci number in the sum S n by a difference as follows:

S n ' F0 % F1 % F2 % ÿ % Fn&1 % Fn
S n ' (F2 & F1) % (F3 & F2) % (F4 & F3) % ÿ % (Fn%1 & Fn) % (Fn%2 & Fn %1).

Next we rearrange the terms and get

S n ' &F1 % (F2 & F2) % (F3 & F3) % ÿ % (Fn%1 & Fn%1) % Fn%2
S n ' &F1 % 0 % 0 % ÿ % 0 % Fn%2
S n ' Fn%2 & F1 ' Fn%2 & 1.

Thus we have made our conjecture (that is, educated guess) into a theorem.

The fundamental relation Fn = Fn+2 - Fn+1 can also be used to define Fn when n is a negative
integer. Letting n = -1 in this formula gives us F-1 = F1 - F0 = 1 - 0 = 1. Similarly, one finds that F-2 = F0 -
F-1 = 0 - 1 = -1 and F-3 = F-1 - F-2 = 1 - (-1) = 2. In this way one can obtain Fn for any negative integer n.

13
Some of the values of Fn for negative integers n are shown in the following table:

n ... -6 -5 -4 -3 -2 -1
Fn ... -8 5 -3 2 -1 1

Perhaps the greatest investigator of properties of the Fibonacci and related sequences was
François Edouard Anatole Lucas (1842-1891). A sequence related to the Fn bears his name. The Lucas
sequence, 2, 1, 3, 4, 7, 11, 18, 29, 47, ..., is defined by

L0 = 2, L1 = 1, L2 = L1 + L0 , L3 = L2 + L1 , ..., Ln+2 = Ln+1 + Ln ,... .

Some of the many relations involving the Fn and the Ln are suggested in the problems below.
These are only a very small fraction of the large number of known properties of the Fibonacci and Lucas
numbers. In fact, there is a mathematical journal, The Fibonacci Quarterly, devoted to them and to
related material.

Problems for Chapter 2

1. For the Fibonacci numbers Fn show that:

(a) F3 = 2F1 + F0 . (b) F4 = 2F2 + F1 . (c) F5 = 2F3 + F2 .

2. The relation Fn+2 = Fn+1 + Fn holds for all integers n and hence so does Fn+3 = Fn+2 + Fn+1 .
Combine these two formulas to find an expression for Fn+3 in terms of Fn+1 and Fn .

3. Find r, given that Fr = 2F101 + F100 .

4. Express F157 + 2F158 in the form Fs .

5. Show the following:

(a) F4 = 3F1 + 2F0 . (b) F5 = 3F2 + 2F1 .

6. Add corresponding sides of the formulas of the previous problem and use this to show that
F6 = 3F3 + 2F2 .

7. Express Fn+4 in terms of Fn+1 and Fn .

8. Find s, given that Fs = 3F200 + 2F199 .

9. Find t, given that Ft = 5F317 + 3F316 .

10. Find numbers a and b such that Fn+6 = aFn+1 + bFn for all integers n.

14
11. Show the following:

(a) F0 + F2 + F4 + F6 = F7 - 1. (b) F0 + F2 + F4 + F6 + F8 = F9 - 1.
(c) F1 + F3 + F5 + F7 = F8 . (d) F1 + F3 + F5 + F7 + F9 = F10 .

12. The relation Fn+2 = Fn+1 + Fn can be rewritten as Fn+1 = Fn+2 - Fn . Use this form to find a
compact expression for Fa + Fa+2 + Fa+4 + Fa+6 + ... + Fa+2m .

13. Find p, given that Fp = F1 + F3 + F5 + F7 + ... + F701 .

14. Find u and v, given that Fu - Fv = F200 + F202 + F204 + ... + F800 .

15. Show the following:

(a) F4 = 3F2 - F0 . (b) F5 = 3F3 - F1 . (c) F6 = 3F4 - F2 .

16. Use the formulas Fn+4 = 3Fn+1 + 2Fn and Fn+1 = Fn+2 - Fn to express Fn+4 in terms of Fn+2 and Fn .

17. Show the following:

(a) 2(F0 + F3 + F6 + F9 + F12 ) = F14 - 1. (b) 2(F0 + F3 + F6 + F9 + F12 + F15 ) = F17 - 1.

18. Show the following:

(a) 2(F1 + F4 + F7 + F10 + F13 ) = F15 . (b) 2(F1 + F4 + F7 + F10 + F13 + F16 ) = F18 .

19. By addition of corresponding sides of formulas of the two previous problems, find expressions for:

(a) 2(F2 + F5 + F8 + F11 + F14 ). (b) 2(F2 + F5 + F8 + F11 + F14 + F17 ).

20. (i) Prove that Fn+3 - 4Fn - Fn-3 = 0 for n $ 3.

(ii) Prove that Fn+4 - 7Fn + Fn-4 = 0 for n $ 4.
*(iii) For m $ 4, find a compact expression for Fa + Fa+4 + Fa+8 + ... + Fa+4m .

21. Evaluate each of the following sums:

2 1 3 2
(a) % . (b) % .
0 1 0 1

4 3 2 5 4 3
(c) % % . (d) % % .
0 1 2 0 1 2

6 5 4 3 7 6 5 4
(e) % % % . (f) % % % .
0 1 2 3 0 1 2 3

15
 9 8 7 6 5
22. Find m, given that % % % % ' F m.
0 1 2 3 4

23. Find r, s, and t given that:

2 2 2 2 2 2
(a) F % F % F ' F r. (b) F % F % F ' F s.
0 0 1 1 2 2 0 1 1 2 2 3

2 2 2
(c) F % F % F ' F t.
0 7 1 8 2 9

24. Find u, v, and w, given that:

3 3 3 3
(a) F % F % F % F ' F u.
0 0 1 1 2 2 3 3

3 3 3 3
(b) F % F % F % F ' F v.
0 1 1 2 2 3 3 4

3 3 3 3
(c) F % F % F % F ' F w.
0 7 1 8 2 9 3 10

25. Find r,s, and t, given that:

(a) (F7)2 % (F8)2 ' F r. (b) (F8)2 % (F9)2 ' F s. (c) (F9)2 % (F10)2 ' F t.

26. Find u, v, and w, given that:

(a) (F3)2 & (F2)2 ' F u Fu%3. (b) (F4)2 & (F3)2 ' F v Fv%3. (c) (F9)2 & (F8)2 ' F w Fw%3.

27. Let L0, L1, L2 ,... be the Lucas sequence. Prove that

L0 % L1 % L2 % L3 % ... % Ln ' Ln%2 & 1.

28. Find r, given that Lr ' 2L100 % L99.

29. Find s, given that Ls ' 3L201 % 2L200.

30. Find t, given that Lt ' 8L999 % 5L998.

31. Show that L0 % L2 % L4 % L6 % L8 % L10 ' L11 % 1.

32. Find m, given that L0 % L2 % L4 % L6 % ... % L400 ' Lm % 1.

16
33. Derive a formula for L1 % L3 % L5 % L7 % ... % L2m%1.

34. Conjecture, and test in several cases, formulas for:

(a) L0 % L3 % L6 % L9 % ... % L3m. (b) L1 % L4 % L7 % L10 % ... % L3m%1 .

n n n n
(c) L2 % L5 % L8 % L11 % ... % L3m%2. (d) L % L % L % ... % L .
0 k 1 k%1 2 k%2 n k%n

35. Find r, s, and t, given that F2L2 ' Fr, F3L3 ' F s, and F4L4 ' Ft.

36. Find u, v, and w, given that F10/F5 ' Lu, F12/F6 ' Lv, and F14/F7 ' Lw .

37. Evaluate the following:

(a) (F1)2 & F0F2. (b) (F2)2 & F1F3. (c) (F3)2 & F2F4. (d) (F4)2 & F3F5.

38. Evaluate the expressions of the previous problem with each Fibonacci number replaced by
the corresponding Lucas number.

39. Which of the Fibonacci numbers F800, F801, F802, F803 , F804, and F805 are even?

40. Which of the Fibonacci numbers of the previous problem are exactly divisible by 3?

17
 Chapter 3

FACTORIALS

A sequence which occurs frequently in mathematics is

1, 1, 1 @ 2, 1 @ 2 @3, 1@ 2@3 @ 4, ÿ .

We tabulate this in the form

n 0 1 2 3 4 5 6 ...
n! 1 1 2 6 24 120 720 ...

where the notation n! (read as n factorial) is used for the number on the second line that is thus
associated with n. Clearly 0! = 1, 1! = 1, 2! = 2, 3! = 6, 4! = 24, etc. The definition of n! can be
given as follows:

0! = 1, 1! = 1(0!), 2! = 2(1!),
3! = 3(2!), ... , (n + 1)! = (n + 1)(n!), ... .

The expression n! is not defined for negative integers n. One reason is that the relation (n + 1)! =
(n + 1)(n!) becomes 1 ' 0@(&1)! when n = -1, and hence there is no way to define (-1)! so that
this relation is preserved.

Problems for Chapter 3

1. Find the following:

(a) 7!.
(b) (3!)2.
(c) (32)!.
(d) (3!)!.

2. Find the following:

(a) 8!.
(b) (2!)(3!).
(c) (2 @ 3)!.

18
 5 7
3. Show that (2!)(3!) ' 5! and (3!)(4!) ' 7!.
2 3

6 8
4. Find c and d, given that (2!)(4!) ' c! and (3!)(5!) ' d!.
2 3

5. Write as a single factorial:

(a) 3!@ 4 @ 5.
(b) 4!@ 210.
(c) n!(n + 1).

6. Express a!(a2 + 3a + 2) as a single factorial.

7. Find a and b such that 11@ 12@ 13@ 14 ' a!/b!.

8. Find e, given that (n % e)!/n! ' n 3 % 6n 2 % 11n % 6.

9. Express (n + 4)!/n! as a polynomial in n.

10. Find numbers a, b, c, d, and e such that (n + 5)!/n! = n5 + an4 + bn3 + cn2 + dn + e.

11. Calculate the following sums:

(a) 1!@ 1 % 2!@ 2 % 3!@ 3.

(b) 1!@ 1 % 2!@ 2 % 3!@ 3 % 4!@ 4.
(c) 1!@ 1 % 2!@ 2 % 3!@ 3 % 4!@ 4 % 5!@ 5.

12. Conjecture a compact expression for the sum 1!@ 1 % 2!@ 2 % 3!@ 3 % þ % n!@ n and test it for
several values of n.

13. Show that (n % 1)! & n! ' n!@ n.

14. Show that (n + 2)! - n! = n!(n2 + 3n + 1).

15. Find numbers a, b, and c such that (n % 3)! & n! ' n!(n 3 % an 2 % bn % c) holds for
n = 0, 1, 2, ... .

16. Use the formula in Problem 13 to derive a compact expression for the sum in Problem 12.

19
17. Use the formula in Problem 14 to derive a compact expression for

0! + 11(2!) + 29(4!) + ... + (4m2 + 6m + 1)[(2m)!].

18. Derive a compact expression for

5(1!) + 19(3!) + 41(5!) + ... + (4m2 + 2m - 1)[(2m - 1)!].

19. Derive compact expressions for:

(a) 0! % 5(1!) % 11(2!) % ... % (n 2 %3n % 1)(n!).

*
(b) 0! % 2(1!) % 5(2!) % ... % (n 2 % 1)(n!).

1 2 3 n
20. Derive a compact expression for % % % ... %
2! 3! 4! (n % 1)!

21. Show that:

(a) 6! ' 3!@ 23 @ 3@ 5.

(b) 8! ' 4!@ 24 @ 3@ 5@ 7.

(c) 10! ' 5!@ 25 @ 3@ 5@ 7@ 9.

22. Express r, s, and t in terms of m so that

1@ 3@ 5@ 7@ þ@ (2m & 1) ' r!/(s!@ 2t).

20
 Chapter 4

ARITHMETIC AND GEOMETRIC PROGRESSIONS

A finite sequence such as

2, 5, 8, 11, 14, ... , 101

in which each succeeding term is obtained by adding a fixed number to the preceding term is
called an arithmetic progression. The general form of an arithmetic progression with n terms is
therefore

a, a % d, a % 2d, a % 3d, ... , a % (n & 1)d

where a is the first term and d is the fixed difference between successive terms.
In the arithmetic progression above, the first term is 2 and the common difference is 3.
The second term is 2 % 3@1, the third term is 2 % 3@2, the fourth is 2 % 3@3, and the nth is 2 +
3(n - 1). Since 101 & 2 ' 99 ' 3@33 or 101 = 2 + 3(34 - 1), one has to add 3 thirty-three times
to obtain the nth term. This shows that there are thirty-four terms here. The sum S of these
thirty-four terms may be found by the following technique. We write the sum with the terms in
the above order and also in reverse order, and add:

S ' 2 % 5 % 8 %þ % 98 % 101
S ' 101 % 98 % 95 %þ % 5 % 2
2S ' (2 % 101) % (5 % 98) % (8 % 95) %þ % (98 % 5) % (101 % 2)
' 103 % 103 % 103 %þ % 103 % 103
2S ' 34@103 ' 3502.

Hence S = 3502/2 = 1751.

Using this method, one can show that the sum

Tn = 1 + 2 + 3 + 4 + ... + n

of the first n positive integers is n(n + 1)/2. Some values of Tn are given in the table which
follows.

n 1 2 3 4 5 6 ...
Tn 1 3 6 10 15 21 ...

The sequence Tn may be defined for all positive integers n by

21
 T1 = 1, T2 = T1 + 2, T3 = T2 + 3,

T4 = T3 + 4, ..., Tn+1 = Tn + (n + 1), ... .

The values 1, 3, 6, 10, 15, ... of Tn are called triangular numbers because they give the number
of objects in triangular arrays of the type shown in Figure 2.

Figure 2

An arithmetic progression may have a negative common difference d. One with a = 7/3,
d = -5/3, and n = 8 is:

7/3, 2/3, -1, -8/3, -13/3, -6, -23/3, -28/3.

The average (or arithmetic mean) of n numbers is their sum divided by n. For example,
the average of 1, 3, and 7 is 11/3. If each of the terms of a sum is replaced by the average of the
terms, the sum is not altered. We note that the average of all the terms of an arithmetic
progression is the average of the first and last terms, and that the average is the middle term when
the number of terms is odd, that is, whenever there is a middle term.
If a is the average of r and s, then it can easily be seen that r, a, s are consecutive terms of
an arithmetic progression. (The proof is left to the reader.) This is why the average is also called
the arithmetic mean.
A finite sequence such as

3, 6, 12, 24, 48, 96, 192, 384

in which each term after the first is obtained by multiplying the preceding term by a fixed number,
is called a geometric progression. The general form of a geometric progression with n terms is
therefore

a, ar, ar2, ar3, ..., arn-1.

Here a is the first term and r is the fixed multiplier. The number r is called the ratio of the
progression, since it is the ratio (i.e., quotient) of a term to the preceding term.
We now illustrate a useful technique for summing the terms of a geometric progression.

22
Example. Sum 5 % 5@22 % 5@24 % 5@26 % 5@28 % ... % 5@2100.
Solution: Here the ratio r is 22 = 4. We let S designate the desired sum and write S and
rS as follows:

S ' 5 % 5@22 % 5@24 % 5@26 % ... % 5@2100

4S ' 5@22 % 5@24 % 5@26 % ... % 5@2100 % 5@2102.
Subtracting, we note that all but two terms on the right cancel out and we obtain

3S ' 5@2102 & 5

or
3S = 5(2102 - 1).

Hence we have the compact expression for the sum:

5(2102 & 1)
S= .
3

If the ratio r is negative, the terms of the geometric progression alternate in signs. Such a
progression with a = 125, r = -1/5, and n = 8 is

125, -25, 5, -1, 1/5, -1/25, 1/125, -1/625.

The geometric mean of two positive real numbers a and b is ab , the positive square
3
root of their product; the geometric mean of three positive numbers a, b, and c is abc. In
general, the geometric mean of n positive numbers is the nth root of their product. For example,
3 3 3
the geometric mean of 2, 3, and 4 is 2@3@4 ' 8@3 ' 2 3.

Problems for Chapter 4

1. (a) Find the second, third, and fourth terms of the arithmetic progression with the first term
-11 and difference 7.
(b) Find the next three terms of the arithmetic progression -3, -7, -11, -15, ... .

2. (a) Find the second, third, and fourth terms of the arithmetic progression with the first term 8
and difference -3.
(b) Find the next three terms of the arithmetic progression 7/4, 1, 1/4, -1/2, -5/4, -2, ... .

23
3. Find the 90th term of each of the following arithmetic progressions:

(a) 11, 22, 33, 44, ... .

(b) 14, 25, 36, 47, ... .
(c) 9, 20, 31, 42, ... .

4. For each of the following geometric progressions, find e so that 3e is the 80th term.

(a) 3, 9, 27, ... .

(b) 1, 3, 9, ... .
(c) 81, 243, 729, ... .

5. Find x, given that 15, x, 18 are consecutive terms of an arithmetic progression.

6. Find x and y so that 14, x, y, 9 are consecutive terms of an arithmetic progression.

7. Sum the following:

(a) 7/3 + 2/3 + (-1) + (-8/3) + (-13/3) + ... + (-1003/3).

(b) (-6) + (-2) + 2 + 6 + 10 + ... + 2002.
(c) The first ninety terms of 7/4, 1, 1/4, -1/2, -5/4, -2, ... .
(d) The first n odd positive integers, that is, 1 + 3 + 5 + ... + (2n - 1).

8. Sum the following:

(a) 12 + 5 + (-2) + (-9) + ... + (-1073).

(b) (-9/5) + (-1) + (-1/5) + 3/5 + 7/5 + 11/5 + ... + 2407.
(c) The first eighty terms of -3, -7, -11, -15, ... .
(d) The first n terms of the arithmetic progression a, a + d, a + 2d, ... .

9. Find the fourth, seventh, and ninth terms of the geometric progression with first term 2 and
ratio 3.

10. Find the fourth and sixth terms of the geometric progression with first term 2 and ratio -3.

11. Find the next three terms of the geometric progression 2, 14, 98, ... .

12. Find the next three terms of the geometric progression 6, -2, 2/3, -2/9, ... .

13. Find both possible values of x if 7, x, 252 are three consecutive terms of a geometric
progression.

14. Find all possible values of y if 400, y, 16 are three consecutive terms of a geometric
progression.

24
15. Find a compact expression for each of the following:

(a) 1 + 7 + 72 + 73 + ... + 7999.

(b) 1 - 7 + 72 - 73 + ... - 7999.
(c) 1 + 7 + 72 + 73 + ... + 7n-1.

16. Find a compact expression for each of the following:

1 1 1 1
(a) 1 % % % % ... % .
2 4 6
3 3 3 388

8 8 8 8
(b) 8 % % % % ... % .
2 4 6
3 3 3 3188

(c) 8 % 8 @3&2 % 8 @3&4 % 8 @3&6 % ... % 8 @3&2m.

n
17. Find n, given that 1 + 2 + 3 + 4 + 5 + 6 + 7 + 8 = .
2

* m
18. Find m, given that 1 + 2 + 3 + 4 + ... + 1000 = .
m&2
19. Given that a is the average of the numbers r and s, show that r, a, s are three consecutive
terms of an arithmetic progression and that their sum is 3a.

20. Show that r3, r2s, rs2, s3 are four consecutive terms of a geometric progression and that their
sum is (r4 - s4)/(r - s).

21. Find the geometric mean of each of the following sets of positive numbers:

(a) 6, 18.
(b) 2, 6, 18, 54.
(c) 2, 4, 8.
(d) 1, 2, 4, 8, 16.

22. Find the geometric mean of each of the following sets of numbers:

(a) 3, 4, 5.
(b) 3, 4, 5, 6.
(c) 1, 7, 72, 73.
(d) a, ar, ar2, ar3, ar4.

23. Find the geometric mean of 8, 27, and 125.

25
24. Find the geometric mean of a4, b4, c4, and d4.

25. Let b be the middle term of a geometric progression with 2m + 1 positive terms and let r be
the common ratio. Show that:

(a) The terms are br-m, br-m+1, ... , br-1, b, br, ... , brm.
(b) The geometric mean of the 2m + 1 numbers is the middle term.

26. Show that the geometric mean of the terms in a geometric progression of positive numbers
is equal to the geometric mean of any two terms equally spaced from the two ends of the
progression.

27. Find a compact expression for the sum xn + xn-1y + xn-2y2 + ... + xyn-1 + yn.

28. Find a compact expression for the arithmetic mean of xn, xn-1y, xn-2y2, ..., xyn-1, yn.

29. A 60-mile trip was made at 30 miles per hour and the return at 20 miles per hour.

(a) How many hours did it take to travel the 120 mile round trip?
(b) What was the average speed for the round trip?

30. Find x, given that 1/30, 1/x, and 1/20 are in arithmetic progression. What is the relation
between x and the answer to Part (b) of problem 29?

31. Verify the factorization 1 - x7 = (1 - x)(1 + x + x2 + x3 + x4 + x5 + x6) and use it with x = 1/2 to
find a compact expression for

2 3 4 5 6
1 1 1 1 1 1
1% % % % % % .
2 2 2 2 2 2

32. Use the factorization 1 + x99 = (1 + x)(1 - x + x2 - x3 + x4 - ... + x98) to find compact
expressions for the following sums:

(a) 1 - 5-1 + 5-2 - 5-3 + ... - 5-97 + 5-98.

(b) a - ar + ar2 - ar3 + ... - ar97 + ar98.

33. Let a1, a2, a3, ..., a3m be an arithmetic progression, and for n = 1, 2, ..., 3m let An be the
arithmetic mean of its first n terms. Show that A2m is the arithmetic mean of the two
numbers Am and A3m.

34. Let g1, g2, ..., g3m be a geometric progression of positive terms. Let A, B, and C be the
geometric means of the first m terms, the first 2m terms, and all 3m terms, respectively.
Show that B2 = AC.

26
35. Let S be the set consisting of those of the integers 0, 1, 2, ..., 30 which are divisible exactly
by 3 or 5 (or both), and let T consist of those divisible by neither 3 nor 5.

(a) Write out the sequence of numbers in S in their natural order.

(b) In the sequence of Part (a), what is the arithmetic mean of terms equally spaced from
the two ends of the sequence?
(c) What is the arithmetic mean of all the numbers in T?
(d) Find the sum of the numbers in T.

36. Find the sum 4 + 5 + 6 + 8 + 10 + 12 + 15 + ... + 60,000 of all the positive integers not
exceeding 60,000 which are integral multiples of at least one of 4, 5, and 6.

37. Let u1, u2, ..., ut satisfy un+2 = 2un+1 - un for n =1, 2, ..., t - 2. Show that the t terms are in
arithmetic progression.

38. Find a compact expression for the sum v1 + v2 + ... + vt in terms of v1 and v2, given that
vn+2 = (vn+1)2/vn for n = 1, 2, ... ,t - 2.

39. Let an = 2n be the nth term of the geometric progression 2, 22, 23, ... , 2t. Show that
an+2 - 5an+1 + 6an = 0 for n = 1, 2, ... , t - 2.

40. For what values of r does the sequence bn = rn satisfy bn+2 - 5bn+1 + 6bn = 0 for all n?

41. Let a be one of the roots of x2 - x - 1 = 0. Let the sequence c0, c1, c2, ... be the geometric
progression 1, a, a2, ... . Show that:

(a) cn+2 = cn+1 + cn.

(b) c2 = a2 = a + 1.
(c) c3 = a3 = 2a + 1.
(d) c4 = 3a + 2.
(e) c5 = 5a + 3.
(f) c6 = 8a + 5.

42. For the sequence c0, c1, ... of the previous problem, express c12 in the form aFu + Fv, where
Fu and Fv are Fibonacci numbers, and conjecture a similar expression for cm.

27
*
43. In the sequence 1/5, 3/5, 4/5, 9/10, 19/20, 39/40, ... each succeeding term is the average of
the previous term and 1. Thus:

3 1 1 4 1 3 9 1 4
' % 1 , ' % 1 , ' % 1 , ÿ.
5 2 5 5 2 5 10 2 5

1
(a) Show that the twenty-first term is 1 & .
5@218
(b) Express the nth term similarly.

(c) Sum the first five hundred terms.

*
44. In the sequence 1, 2, 3, 6, 7, 14, 15, 30, 31, ... a term in an even numbered position is
double the previous term, and a term in an odd numbered position (after the first term) is
one more than the previous term.

(a) What is the millionth term of this sequence?

(b) Express the sum of the first million terms compactly.

28
 Chapter 5

MATHEMATICAL INDUCTION

In mathematics, as in science, there are two general methods by which we can arrive at
new results. One, deduction, involves the assumption of a set of axioms from which we deduce
other statements, called theorems, according to prescribed rules of logic. This method is
essentially that used in standard courses in Euclidean geometry.
The second method, induction, involves the guessing or discovery of general patterns from
observed data. While in most branches of science and mathematics the guesses based on
induction may remain merely conjectures, with varying degrees of probability of correctness,
certain conjectures in mathematics which involve the integers frequently can be proved by a
technique of Pascal called mathematical induction. Actually, this technique in not induction, but is
rather an aid in proving conjectures arrived at by induction.

THE PRINCIPLE OF MATHEMATICAL INDUCTION: A statement concerning positive

integers is true for all positive integers if (a) it is true for 1, and (b) its being true for any integer
k implies that it is true for the next integer k + 1.

If one replaces (a) by (a'),"it is true for some integer s," then (a') and (b) prove the
statement true for all integers greater than or equal to s. Part (a) gives only a starting point; this
starting point may be any integer - positive, negative, or zero.
Let us see if mathematical induction is a reasonable method of proof of a statement
involving integers n. Part (a) tells us that the statement is true for n = 1. Using (b) and the fact
that the statement is true for 1, we obtain the fact that it is true for the next integer 2. Then (b)
implies that it is true for 2 + 1 = 3. Continuing in this way, we would ultimately reach any fixed
positive integer.
Let us use this approach on the problem of determining a formula which will give us the
number of diagonals of a convex polygon in terms of the number of sides. The three-sided
polygon, the triangle, has no diagonals; the four-sided polygon has two. An examination of other
cases yields the data included in the following table:

n = number of sides 3 4 5 6 7 8 9 ... n ...

Dn = number of diagonals 0 2 5 9 14 20 27 ... Dn ...

The task of guessing the formula, if a formula exists, is not necessarily an easy one, and
there is no sure approach to this part of the over-all problem. However, if one is perspicacious,
one observes the following pattern:

2D3 ' 0 ' 3@0

2D4 ' 4 ' 4@1
2D5 ' 10 ' 5@2
2D6 ' 18 ' 6@3
2D7 ' 28 ' 7@4.

29
This leads us to conjecture that

2Dn = n(n - 3)

or

n(n&3)
Dn =
2
Now we shall use mathematical induction to prove this formula. We shall use as a starting
point n = 3, since for n less than 3 no polygon exists. It is clear from the data that the formula
holds for the case n = 3. Now we assume that a k-sided polygon has k(k - 3)/2 diagonals. If we
can conclude from this that a (k + 1)-sided polygon has (k + 1)[(k + 1) - 3]/2 = (k + 1)(k - 2)/2
diagonals, we will have proved that the formula holds for all positive integers greater than or
equal to 3.
Consider a k-sided polygon. By assumption it has k(k - 3)/2 diagonals. If we place a
triangle on a side AB of the polygon, we make it into a (k + 1)-sided polygon. It has all the
diagonals of the k-sided polygon plus the diagonals drawn from the new vertex N to all the

Figure 3

vertices of the previous k-sided polygon except 2, namely A and B. In addition, the former side
AB has become a diagonal of the new (k + 1)-sided polygon. Thus a (k + 1)-sided polygon has a
k(k&3)
total of % (k & 2) % 1 diagonals. But:
2
k(k&3)
% (k & 2) % 1
2
k 2 &3k % 2k & 2
'
2

30
 k2 & k & 2
'
2
(k % 1)(k & 2)
'
2
(k % 1)[(k % 1) &3]
'
2
This is the desired formula for n = k + 1.
So, by assuming that the formula Dn = n(n - 3)/2 is true for n = k, we have been able to
show it true for n = k + 1. This, in addition to the fact that it is true for n = 3, proves that it is
true for all integers greater than or equal to 3. (The reader may have discovered a more direct
method of obtaining the above formula.)
The method of mathematical induction is based on something that may be considered one
of the axioms for the positive integers: If a set S contains 1, and if, whenever S contains an
integer k, S contains the next integer k + 1, then S contains all the positive integers. It can be
shown that this is equivalent to the principle that in every non-empty set of positive integers there
is a least positive integer.

Example 1. Find and prove by mathematical induction a formula for the sum of the first n cubes,
that is, 13 + 23 + 33 + ... + n3.
Solution: We consider the first few cases:

13 ' 1
1 % 23
3
' 9
13 % 23 % 33 ' 36
13 % 23 % 33 % 43 ' 100.

We observe that 1 = 12, 9 = 32, 36 = 62, and 100 = 102. Thus it appears that the sums are the
squares of triangular numbers 1, 3, 6, 10, ... . In Chapter 4 we saw that the triangular numbers
are of the form n(n + 1)/2. This suggests that

n(n % 1) 2
13 % 23 % 33 % ... % n 3 ' .
2

It is clearly true for n = 1. Now we assume that it is true for n = k:

k(k % 1) 2
13 % 23 % 33 % þ % k 3 ' .
2

Can we conclude from this that

31
 (k % 1)[(k % 1) % 1] 2
13 % 23 % 33 % þ % k 3 % (k % 1)3 ' ?
2

We can add (k + 1)3 to both sides of the known expression, obtaining:

2
k(k % 1)
13 % 23 % 33 % þ % k 3 % (k % 1)3 ' % (k % 1)3
2
k2
' (k % 1)2 % (k % 1)3
4
(k % 1)2 (k 2 % 4k % 4)
'
4
(k % 1) (k % 2)2
2
'
4
(k % 1)(k % 2) 2
'
2

(k % 1)[(k % 1) % 1] 2
' .
2

Hence the sum when n = k + 1 is [n(n + 1)/2]2, with n replaced by k + 1, and the formula is
proved for all positive integers n.
Our guessed expression for the sum was a fortunate one!

Example2. Prove that a - b is a factor of an - bn for all positive integers n.

Proof: Clearly, a - b is a factor of a1 - b1; hence the first part of the induction is verified,
that is, the statement is true for n = 1. Now we assume that ak - bk has a - b as a factor:

a k & b k ' (a & b)M.

Next we must show that a - b is a factor of ak+1 - bk+1. But

a k%1 & b k%1 ' a@a k & b@b k

' a@a k & b@a k % b@a k & b@b k
' (a & b)a k % b(a k & b k).

Now, using the assumption that ak - bk = (a - b)M and substituting, we obtain:

32
 a k%1 & b k%1 ' (a & b)a k % b(a & b)M
' (a & b) [a k % bM].

We see from this that a - b is a factor of ak+1 - bk+1 and hence a - b is a factor of an - bn for n equal
to any positive integer. It is easily seen that the explicit factorization is
a n & b n ' (a & b)(a n&1 % a n&2b % a n&3b 2 % ... % ab n&2 % b n&1).
Example 3. Prove that n(n2 + 5) is an integral multiple of 6 for all integers n, that is, there is an
integer u such that n(n2 + 5) = 6u.
Proof: We begin by proving the desired result for all the integers greater than or equal to
0 by mathematical induction.
When n = 0, n(n2 + 5) is 0. Since 0 ' 6@0 is a multiple of 6, the result holds for n = 0.
We now assume it true for n = k, and seek to derive from this its truth for n = k + 1.
Hence we assume that

(1) k(k 2 % 5) ' 6r

with r an integer. We then wish to show that

(2) (k % 1)[(k % 1)2 % 5] ' 6s

with s an integer. Simplifying the difference between the left-hand sides of (2) and (1), we obtain

(3) (k % 1)[(k % 1)2 % 5] & k(k 2 % 5) ' 3k(k % 1) % 6.

Since k and k + 1 are consecutive integers, one of them is even. Then their product k(k + 1) is
even, and may be written as 2t, with t an integer. Now (3) becomes

(4) (k % 1)[(k % 1)2 % 5] & k(k 2 % 5) ' 6t % 6 ' 6(t % 1).

Transposing, we have

(k % 1)[(k % 1)2 % 5] ' k(k 2 % 5) % 6(t % 1).

Using (1), we can substitute 6r for k(k2 + 5). Hence

(k % 1)[(k % 1)2 % 5] ' 6r % 6(t % 1) ' 6(r % t % 1).

Letting s be the integer r + t + l, we establish (2), which is the desired result when n = k + 1. This
completes the induction and proves the statement for n $ 0.

33
 Now let n be a negative integer, that is, let n = -m, with m a positive integer. The
previous part of the proof shows that m(m2 + 5) is of the form 6q, with q an integer. Then

n(n2 + 5) = (-m)[(-m)2 + 5] = -m(m2 + 5) = -6q = 6(-q),

a multiple of 6. The proof is now complete.

We have seen that binomial coefficients, Fibonacci and Lucas numbers, and factorials may
be defined inductively, that is, by giving their initial values and describing how to get new values
from previous values. Similarly, one may define an arithmetic progression a1, a2, ... , at as one for
which there is a fixed number d such that an+1 = an + d for n = 1, 2, ..., t - 1. Then the values of a1
and d would determine the values of all the terms. A geometric progression b1, ... , bt is one for
which there is a fixed number r such that bn+1 = bnr for n = 1, 2, ... , t - 1; its terms are determined
by b1 and r.
It is not surprising that mathematical induction is very useful in proving results concerning
quantities that are defined inductively, however, it is sometimes necessary or convenient to use an
alternate principle, called strong mathematical induction.

STRONG MATHEMATICAL INDUCTION: A statement concerning positive integers is true

for all the positive integers if there is an integer q such that (a) the statement is true for 1, 2, ... ,
q, and (b) when k $ q, the statement being true for 1, 2, ... , k implies that it is true for k + 1.

As in the case of the previous principle, this can be modified to apply to statements in
which the starting value is an integer different from 1.
We illustrate strong induction in the following:

Example 4. Let a, b, c, r, s, and t be fixed integers. Let L0, L1,... be the Lucas sequence. Prove
that

(A) rLn%a ' sLn%b % tLn%c

is true for n = 0, 1, 2, ... if it is true for n = 0 and n = 1.

Proof: We use strong induction. It is given that (A) is true for n = 0 and n = 1. Hence, it
remains to assume that k $ 1 and that (A) is true for n = 0, 1, 2, ..., k, and to use these
assumptions to prove that (A) holds for n = k + 1.
We therefore assume that

rL a ' sLb % tLc

rL1%a ' sL1%b % tL1%c
rL2%a ' sL2%b % tL2%c
þ
rLk&1%a ' sLk&1%b % tLk&1%c
rLk%a ' sLk%b % Lk%c
and that there are at least two equations in this list. Adding corresponding sides of the last two of

34
these equations and combining like terms, we obtain

r(Lk%a % Lk&1%a ) ' s(Lk%b % Lk&1%b ) % t(Lk%c % Lk&1%c ).

Using the relation Ln+1 + Ln = Ln+2 for the Lucas numbers, this becomes

rLk%1%a ' sLk%1%b % tLk%1%c

which is (A) when n = k + 1. This completes the proof.

Problems for Chapter 5

In Problems 1 to 10 below, use mathematical induction to prove each statement true for
all positive integers n.

1. The sum of the interior angles of a convex (n + 2)-sided polygon is 180n degrees.

2. 13 % 33 % 53 % ... % (2n & 1)3 ' n 2(2n 2 & 1).

3. (a) 12 % 32 % 52 % ... % (2n & 1)2 ' n(4n 2 & 1)/3.

(b) 1@3 % 3@5 % 5@7 % ... % (2n & 1) (2n % 1) ' n(4n 2 % 6n & 1)/3.
1 1 1 1 n(3n % 5)
(c) % % % ... % ' .
1@3 2@4 3@5 n(n % 2) 4(n % 1)(n % 2)

(d) 1 % 2a % 3a 2 % ... % na n&1 ' [1 & (n % 1)a n % na n%1]/(1 & a)2.

4. (a) 12 % 22 % 32 % ... % n 2 ' n(n % 1)(2n % 1)/6.

(b) 1@3 % 2@4 % 3@5 % ... % n(n % 2) ' n(n % 1)(2n % 7)/6.

5 1 7 1 9 1 2n % 3 1 1
(c) @ % @ % @ % ... % @ ' 1 & .
1@2 3 2@3 32 3@4 33 n(n % 1) 3n 3n(n % 1)

5. (13 % 23 % 33 % ... % n 3) % 3(15 % 25 % 35 % ... % n 5) ' 4(1 % 2 % 3 % ... % n)3.

6. (15 % 25 % 35 % ... % n 5) % (17 % 27 % 37 % ... % n 7) ' 2(1 % 2 % 3 % ... % n)4

35
*
7. 3n % 7n & 2 is an integral multiple of 8.

*
8. 2@7n % 3@5n & 5 is an integral multiple of 24.

9. x 2n & y 2n has x + y as a factor.

10. x 2n%1 % y 2n%1 has x % y as a factor.

11. For all integers n, prove the following:

(a) 2n3 + 3n2 + n is an integral multiple of 6.

(b) n5 - 5n3 + 4n is an integral multiple of 120.

12. Prove that n(n2 - 1)(3n + 2) is an integral multiple of 24 for all integers n.

13. Guess a formula for each of the following and prove it by mathematical induction:
1 1 1 1
(a) % % % ... % .
1@2 2@3 3@4 n(n % 1)
n n
(b) (x % y)(x 2 % y 2)(x 4 % y 4)(x 8 % y 8)...(x 2 % y 2 ).

14. Guess a formula for each of the following and prove it by mathematical induction:
(a) 1@2 % 2@3 % 3@4 % ... % n(n % 1).

1 1 1 1
(b) % % % ... % .
1@3 3@5 5@7 (2n & 1)(2n % 1)

15. Guess a simple expression for the following and prove it by mathematical induction:

1 1 1 1
1 & 1 & 1 & þ 1 & .
22 32 42 n2

16. Find a simple expression for the product in Problem 15, using the factorization

x 2 & y 2 ' (x & y)(x % y).

36
17. Prove the following properties of the Fibonacci numbers Fn for all integers n greater than or
equal to 0:
(a) 2(Fs % Fs%3 % Fs%6 % ... % Fs%3n) ' Fs%3n%2 & Fs&1.

(b) F&n ' (&1)n%1Fn.

n n n n
(c) F % F % F % ... % F ' Fs%2n.
0 s 1 s%1 2 s%2 n s%n

18. Discover and prove formulas similar to those of Problem 17 for the Lucas numbers Ln.

19. Use Example 4, in the text above, to prove the following properties of the Lucas numbers
for n = 0, 1, 2, ... , and then prove them for all negative integers n.
(a) Ln%4 ' 3Ln%2 & Ln.

(b) Ln%6 ' 4Ln%3 % Ln.

(c) Ln%8 ' 7Ln%4 & Ln.

(d) Ln%10 ' 11Ln%5 % Ln.

20. State an analogue of Example 4 for the Fibonacci numbers instead of the Lucas numbers and
use it to prove analogues of the formulas of Problem 19.

21. In each of the following parts, evaluate the expression for some small values of n, use this
data to make a conjecture, and then prove the conjecture true for all integers n.
2
(a) Fn%1 & Fn Fn%2.

2 2
Fn%2 & Fn%1
(b) .
Fn

(c) Fn-1 + Fn+1.

22. Discover and prove formulas similar to the first two parts of the previous problem for the
Lucas numbers.

23. Prove the following for all integers m and n:

(a) Lm%n%1 ' Fm%1Ln%1 % FmLn.

(b) Fm%n%1 ' Fm%1Fn%1 & FmFn.

37
24. Prove that (Fn+1)2 + (Fn)2 = F2n+1 for all integers n.

25. Let a and b be the roots of the quadratic equation x2 - x - 1 = 0. Prove that:
an & bn
(a) Fn ' .
a &b

(b) Ln ' a n % b n.

(c) FnLn ' F2n.

(d) a n ' aFn % Fn&1 and b n ' bF n % Fn&1.

26. The sequence 0, 1, ½, 3/4, 5/8, 11/16, ... is defined by

u1 % u0 un%1 % un
u0 ' 0, u1 ' 1, u2 ' , ÿ, un%2 ' , ÿ.
2 2
Discover and prove a compact formula for un as a function of n.

27. The Pell sequence 0, 1, 2, 5, 12, 29, ... is defined by

P0 = 0, P1 = 1, P2 = 2P1 + P0, ..., Pn+2 = 2Pn+1 + Pn , ... .

2 2 2 2
Let xn ' Pn%1 & Pn , yn ' 2Pn%1Pn, and z n ' Pn%1 % Pn . Prove that for every positive
integer n the numbers xn, yn, and zn are the lengths of the sides of a right triangle and that xn
and yn are consecutive integers.

28. Discover and prove properties of the Pell sequence that are analogous to those of the
Fibonacci sequence.

29. Let the sequence 1, 5, 85, 21845, ... be defined by

c1 = 1, c2 = c1(3c1 + 2), ..., cn+1 = cn(3cn + 2), ... .

n&1
42 & 1
Prove that c n ' for all positive integers n.
3

30. Let a sequence be defined by d1 = 4, d2 = (d1)2, ..., dn+1 = (dn)2, ... .

Show that dn = 3cn + 1, where cn is as defined in the previous problem.

n(n % 1)(n % 2)(n % 3)
31. Prove that 1@2@3 % 2@3@4 % ... % n(n % 1)(n % 2) ' .
4

38
*
32. Certain of the above formulas suggest the following:

n(n % 1)...(n % m)
1@2þm % 2@3þ(m % 1) % þ % n(n % 1)þ(n % m & 1) ' .
m % 1
Prove it for general m.
*
33. Prove that n5 - n is an integral multiple of 30 for all integers n.
*
34. Prove that n7 - n is an integral multiple of 42 for all integers n.
*
35. Show that every integer from 1 to 2n+1 - 1 is expressible uniquely as a sum of distinct
powers of 2 chosen from 1, 2, 22, ... , 2n.

3n%1 & 1 3n%1 & 1

*
36. Show that every integer s from & to has a unique expression of the
2 2
form
s ' c0 % 3c1 % 32c2 % ... % 3ncn

where each of c0, c1, ..., cn is 0, 1, or -1.

39
 Chapter 6

THE BINOMIAL THEOREM

n
In Chapter 1 we defined as the coefficient of an-rbr in the expansion of (a + b)n, and
r
tabulated these coefficients in the arrangement of the Pascal Triangle:

n Coefficients of (a + b)n
0 1
1 1 1
2 1 2 1
3 1 3 3 1
4 1 4 6 4 1
5 1 5 10 10 5 1
6 1 6 15 20 15 6 1
... . . . . . . . .

n n
We then observed that this array is bordered with 1's; that is, ' 1 and ' 1 for n =
0 n
0, 1, 2, ... . We also noted that each number inside the border of 1's is the sum of the two closest
numbers on the previous line. This property may be expressed in the form

n n n%1
(1) % ' .
r&1 r r

This formula provides an efficient method of generating successive lines of the Pascal Triangle,
but the method is not the best one if we want only the value of a single binomial coefficient for a
100
large n, such as . We therefore seek a more direct approach.
3
It is clear that the binomial coefficients in a diagonal adjacent to a diagonal of 1's are the

40
 n
numbers 1, 2, 3, ... ; that is, ' n. Now let us consider the ratios of binomial coefficients
1
to the previous ones on the same row. For n = 4, these ratios are:

(2) 4/1, 6/4 = 3/2, 4/6 = 2/3, 1/4.

For n = 5, they are

(3) 5/1, 10/5 = 2, 10/10 = 1, 5/10 = 1/2, 1/5.

The ratios in (3) have the same pattern as those in (2) if they are rewritten as

5/1, 4/2, 3/3, 2/4, 1/5.

It is easily seen that this pattern also holds on the line for n = 8, and that the coefficients on that
line are therefore:

8 8 7 8 7 6 8 7 6 5
(4) 1, , @ , @ @ , @ @ @ , ... .
1 1 2 1 2 3 1 2 3 4

8 8@7@6
The binomial coefficient ' can be rewritten as
3 1@2@3

8@7@6@5@4@3@2@1 8!
' .
1@2@3@5@4@3@2@1 3!5!

8 8! 8 8!
Similarly, ' and ' .
2 2!6! 4 4!4!

n n!
This leads us to conjecture that ' holds in all cases. We prove this by
r r!(n & r)!
mathematical induction in the following theorem.

THEOREM: If n and r are integers with 0 # r # n, then

n n!
' .
r r!(n & r)!

0
Proof: If n = 0, the only allowable value of r is 0 and = 1. Since
0

41
 n! 0!
' ' 1
r!(n & r)! 0!0!

the formula holds for n = 0.

Now let us assume that it holds for n = k. Then

k k! k k!
' , ' .
r&1 (r&1)!(k & r % 1)! r r!(k & r)!

Using (1), above, we now have

k%1 k k k! k!
' % ' %
r r&1 r (r & 1)!(k & r % 1)! r!(k & r)!
k!r k!(k & r % 1)
' %
(r & 1)!r(k & r % 1)! r!(k & r)!(k & r % 1)
k!(r % k & r % 1)
'
r!(k & r % 1)!

k!(k % 1)
'
r!(k & r % 1)!

(k % 1)!
' .
r!(k & r % 1)!

Since the formula

k%1 (k % 1)!
'
r r!(k & r % 1)!

is the theorem for n = k + 1, the formula is proved for all integers n $ 0, with the exception that
our proof tacitly assumes that r is neither 0 nor k + 1; that is, it deals only with the coefficients
inside the border of 1's. But the formula

k%1 (k % 1)!
'
r r!(k & r % 1)!

42
 k%1 k%1 (k % 1)!
shows that each of and is ' 1.
0 k%1 0!(k % 1)!
Hence the theorem holds in all cases.

The above theorem tells us that the coefficient of x ry s in (x % y)n is

n!
.
r!s!

Since this expression has the same value when r and s are interchanged, we again see that the
binomial coefficients have the symmetry relation

n n
' .
r n&r

By writing out the factorials more explicitly, we see that

n n!
'
r r!(n & r) !
n(n & 1)(n & 2)...(n & r % 1)(n & r)(n & r & 1)...2@1
' .
1@[email protected](n & r)(n & r & 1)...2@1

Cancelling common factors, we now have

n n(n & 1)(n & 2) ... (n & r % 1)

' .
r 1@2@3 ... r

This is the alternate form of the theorem illustrated for n = 8 in (4), above.

We can now rewrite the expansion of (a + b)n in the form

n(n & 1) n&2 2

(a % b)n ' a n % na n&1b % a b % ...
1@2
n(n & 1)...(n & r % 1) n&r r
% a b % þ % b n.
1@2 ... r

This last formula it generally called the Binomial Theorem.

43
 The formulas

n n n(n & 1)...(n & r % 1)

' 1, ' for r > 0
0 r 1@2 ... r

n
enable us to extend the definition of , previously defined only for integers n and r with
r

2 &2
0 # r # n, to allow n to be any integer. We then have, for example, ' 0, ' &8,
5 7
&3
and ' 45.
8

It can easily be shown that the formula

n n n%1
% '
r&1 r r

holds with the extended definition as it did with the original definition.

Now the identity

m m m(m & 1)
2 % ' 2 % m ' m2 & m % m ' m2
2 1 2

holds for all integers m, and we can use the formulas

1 2 3 n n%1
% % % ... % '
1 1 1 1 2

1 2 3 n n%1
% % % ... % '
2 2 2 2 3

to show that

44
 12 % 22 % ... % n 2

1 1 2 2 n n
' 2 % % 2 % % ... % 2 %
2 1 2 1 2 1

1 2 n 1 2 n
' 2 % % ... % % % % ... %
2 2 2 1 1 1

n%1 n%1
' 2 %
3 2
(n % 1)n(n & 1) (n % 1)n
' 2 %
6 2

2n 3 & 2n 3n 2 % 3n
' %
6 6

2n 3 % 3n 2 % n
'
6

n(n % 1)(2n % 1)
' .
6

Frequently in mathematical literature a short notation for sums is used which involves the Greek
capital letter sigma, written '. In this notation,

a1 % a2 % ... % an

is written as

j ai
n

i ' 1

and the auxiliary variable i is called the index of summation. Thus, for example,

j i ' 1 % 2 % 3 % 4 % 5 ' 15
5

i ' 1

j 1 ' 1 % 1 % 1 % 1 % 1 % 1 ' 6
6

i ' 1

j j ' 1 % 2 % 3 % ... % (n & 1) .

n & 1
2 2 2 2 2
j ' 1

Under the capital sigma, one indicates the symbol that is used as the index of summation and the

45
initial value of this index. Above the sigma, one indicates the final value. The general
polynomial a0x n % a1x n&1% ... % an&1x % an of degree n can be written as

j ak x .
n
n&k
k ' 0

One easily sees that

j ai % j bi ' j (ai % b i)
n n n

i ' 1 i ' 1 i ' 1

since

j ai % j bi ' (a1 % a2 % ... % an) % (b1 % b2 % ... % bn)

n n

i ' 1 i ' 1
' (a1 % b1) % (a2 % b2) % ... % (an % bn)

' j (ai % bi).

n

i ' 1

Also, j (ca i) ' c j a i, the proof of which is left to the reader. However,
n n

i ' 1 i ' 1

j ai j bi … j (aib i)
n n n

i ' 1 i ' 1 i ' 1

as can easily be shown by counterexample. (See Problem 19 of this chapter.)

A corresponding notation for products uses the Greek letter pi:

k a i ' a1a2 ... an.

n

i ' 1

In this notation, n! for n $ 1 can be expressed as k k.

n

k ' 1

In solving problems stated in terms of the sigma or pi notation, it is sometimes helpful to

rewrite the expression in the original notation.

46
Problems for Chapter 6

1. Find each of the following:

(a) The coefficient of x4y16 in (x + y)20.

(b) The coefficient of x5 in (1 + x)15.
(c) The coefficient of x3y11 in (2x - y)14.

2. Find each of the following:

(a) The coefficient of a13b4 in (a + b)17.

(b) The coefficient of a11 in (a - 1)16.
(c) The coefficient of a6b6 in (a - 3b)12.

n
3. Find integers a, b, and c such that 6 ' n 3 % an 2 % bn % c for all integers n.
3

n
4. Find integers p, q, r, and s such that 4! ' n 4 % pn 3 % qn 2 % rn % s for all
4
integers n.

n
5. Prove that ' 0 for n ' 0, 1, 2.
3

n
6. Given that k is a positive integer, prove that ' 0 for n ' 0, 1, ... , k & 1.
k

&1
7. Find for r ' 0, 1, 2, 3, 4, and 5. .
r

&2
8. Find for r ' 0, 1, 2, 3, 4, and 5. .
r

&3 r%2
9. Prove that ' (&1)r for r ' 0, 1, 2, ... .
r 2

47
 &4 r%3
10. Prove that ' (&1)r for r ' 0, 1, 2, ... .
r 3

&m
11. Let m be a positive integer and r a non-negative integer. Express in terms of a
r

n
binomial coefficient with 0 # k # n.
k

n
12. In the original definition of as a binomial coefficient, it was clear that it was always an
r
integer. Explain why this is still true in the extended definition.

n n&a n!
13. Show that ' for integers a, b, and n, with
a b a!b!(n & a & b)!

a $ 0, b $ 0, and n $ a % b.

14. Given that n = a + b + c + d and that a, b, c, and d are non-negative integers, show that

n n&a n&a&b n&a&b&c n!

' .
a b c d a!b!c!d!

15. Express j [a % (k & 1)d] as a polynomial in n.

n

k'1

n
16. Express A (2k) compactly without using the A notation.
k'1

n n&1
17. Show that A a k ' A aj%1.
k'1 j'0

18. Show that j bk ' j bi&2.

n&2 n

k'1 i'3

48
19. Evaluate j ai j bi and j (aib i ) and show that they are not always equal.
2 2 2

i'1 i'1 i'1

n n n
20. Show that A ai A bi ' A (aibi ).
i'1 i'1 i'1

21. Prove by mathematical induction that j

n
s%i s%1%n
' .
i'0 s s%1

22. Prove that j

n
s%j s%1%n
' .
j'0 j n

23. Express j
n&2
k(k % 1)
as a polynomial in n.
k'1 2

24. Express j
n&2
k%1
as a polynomial in n.
k'1 k&1

n n n n
25. Write 6 % % as a polynomial in n, and then use the fact that is
3 2 1 r
always an integer to give a new proof that n(n2 + 5) is an integral multiple of 6 for all
integers n.

n n n n
26. (a) Write 4! % % % as a polynomial in n.
4 3 2 1

(b) Show that n4 - 2n3 + 11n2 + 14n is an integral multiple of 24 for all integers n.

27. Find numbers s and t such that n3 = n(n - 1)(n - 2) + sn(n - 1) + tn holds for n = 1 and n = 2.

n n n
28. Find numbers a and b such that n 3 ' 6 %a %b for all integers n.
3 2 1

49
29. Find numbers r, s, and t such that n4 = n(n - 1)(n - 2)(n - 3) + rn(n - 1)(n - 2) + sn(n - 1) + tn
for n = 1, 2, and 3. Using these values of r, s, and t, show that

n n n n
n 4 ' 24 % 6r % 2s %t
4 3 2 1
for all integers n.

30. Find numbers a, b, c, and d such that

n n n n n
n 5 ' 5! %a %b %c %d .
5 4 3 2 1

31. Express j k 4 as a polynomial in n.

n

k'1

32. Express j k 5 as a polynomial in n.

n

k'1

33. We define a sequence S0, S1, S2, ... as follows: When n is an even integer 2t, let

Sn ' S2t ' j . When n is an odd integer 2t + 1, let Sn ' S2t%1 ' j
t t
t%j t%1%j
.
j' 0 t&j j' 0 t&j

Prove that S2t + S2t+1 = S2t+2 and S2t+1 + S2t+2 = S2t+3 for t = 0, 1, 2, ... .

34. For the sequence defined in Problem 33, prove that Sn is the Fibonacci number Fn+1.
*
35. Prove the following property of the Fibonacci numbers:

j
n
n
(&1)j Fs%2n&2j ' Fs%n.
j'0 j

*
36 Prove an analogue of the formula of Problem 35 for the Lucas numbers.

37 Find a compact expression, without using the sigma notation, for

1@n % 2(n & 1) % 3(n & 2) % þ % (n & 1)@2 % n@1,

that is, for j (k % 1)(n & k).

n&1

k'0

50
 Chapter 7

COMBINATIONS AND PERMUTATIONS

We have seen in the previous chapter that (a + b)n can be written as

n n n n&1 n n&k k n n
a % a b%þ% a b %þ% b
0 1 k n

where we have the specific formula for the binomial coefficients:

n n! n(n & 1)(n & 2) ÿ(n & k % 1)

' ' .
k k!(n & k)! 1@2@3 ÿk

n
We now look at a different interpretation of these numbers and will see why is
k
called "n choose k." Let

M = (1 + x1)(1 + x2)(1 + x3).

Expanding this product, we get

M = 1 + x1 + x2 + x3 + x1x2 + x1x3 + x2x3 + x1x2x3.

The terms of this expansion correspond to the subsets of S = {x1, x2, x3}. That is, we can
associate the term 1 with the empty subset of S; the terms x1, x2, and x3 with the singleton subsets
of S; the terms x1x2, x1x3, and x2x3 with the doubleton subsets; and x1x2x3 with S itself. (S is the
only subset with 3 elements.)

Next we replace each of x1, x2, and x3 by x in our two expressions for M. This results in

(1 + x)3 = 1 + 3x + 3x2 + x3.

3
Thus we see that for k = 0, 1, 2, 3 is the number of ways of choosing a subset of k
k
elements from a set S of 3 elements. Similarly, one can see that the number of ways of choosing k
n
elements from a set of n elements is .
k

51
 5
For example, the set {1, 2, 3, 4, 5} with 5 elements has subsets having 3 elements.
3
Since
5 5@4@3
' ' 10,
3 1@2@3

it is not too difficult to write out all ten of these subsets as

{1, 2, 3}, {1, 2, 4}, {1, 2, 5}, {1, 3, 4}, {1, 3, 5},
{1, 4, 5}, {2, 3, 4}, {2, 3, 5}, {2, 4, 5}, {3, 4, 5}.

If we drop the braces enclosing the elements of each subset, the resulting sequence is said to be a
combination of 3 things chosen from the set {1, 2, 3, 4, 5}. Thus the ten combinations of 3
things chosen from this set of 5 objects are

1, 2, 3; 1, 2, 4; 1, 2, 5; 1, 3, 4; 1, 3, 5;
1, 4, 5; 2, 3, 4; 2, 3, 5; 2, 4, 5; 3, 4, 5.

Note that changing the order in which the objects of a combination are written does not change
the combination. For example, 1, 2, 4 is the same combination as 1, 4, 2.

n n
The formula ' tells us that the entries on row n of the Pascal Triangle
k n& k
read the same left to right as they do right to left. The combinatorial significance of this formula
is that the number of ways of choosing k elements from a set of n elements is equal to the number
of ways of omitting n - k of the elements.

A problem analogous to that of finding the coefficients of a binomial expansion is that of

finding the coefficents in

(x + y + z)n.

These coefficients, called the trinomial coefficients, are naturally more complicated but,
fortunately, can be expressed in terms of the binomial coefficients in the following way. Let us
look for the coefficient of x6y3z1 in (x + y + z)10. From this product of ten factors we must choose
10
six x’s, three y’s, and one z. We can choose six x’s from a set of ten in ways and three
6

52
 4
y’s from the remaining four factors in ways, and the z from the remaining factor in
3

1
way. Therefore the trinomial coefficient of x6y3z1 in (x + y + z)10 can be written as
1

10 4 1 1 10 4
, or, since ' 1, as . We can obtain an alternate
6 3 1 1 6 3

10
representation of this number, however, by choosing the z first in ways, then the six x’s
1
from the remaining nine, and finally the three y’s from the remaining three. Thus the coefficient
10 9 3 10 9 10 4 10 9
would appear as or . Hence ' . By
1 6 3 1 6 6 3 1 6
choosing the y’s first, one can see that this coefficient could also be expressed as
10 7 10 7
or . The reader may find other forms of the coefficient.
3 6 3 1
This problem can be generalized similarly to find the coefficients of (x % y % z % þ % w)n;
they are called multinomial coefficients. It can readily be shown that the coefficient of
x ay bz cÿw d in the expansion of (x % y % z % þ % w)n is

n!
a!b!c!ÿd!

where, of course, the sum a % b % c % þ % d of the exponents must be n.

Another interesting problem, and one with frequent applications, is that of finding the
number of ways in which one can arrange a set of objects in a row, that is, the number of
permutations of the set. Let us consider the set of four objects a, b, c, d. They can be arranged
in the following ways:

abcd bacd cabd dabc

abdc badc cadb dacb
acbd bcad cbad dbac
acdb bcda cbda dbca
adbc bdac cdab dcab
adcb bdca cdba dcba

53
 Rather than write them all out, if we are only interested in the number of arrangements, we
may think of the problem thus: We have four spaces to fill. If we put, for example, the b in the
first, we have only the a, c, and d to choose from in filling the remaining three. And if we put the
d in the second, we have only a and c for the remaining; and so forth. So we have four choices
for the first space, three for the second, two for the third, and one for the fourth. This gives us
4@3@2@1, or 4! arrangements of four objects. This argument can be used to show that there are n!
arrangements of n objects.

We may also consider the possibility of arranging, in a row, r objects chosen from a set of
n. We have n choices for the first space, n - 1 for the second, n - 2 for the third, and so on.
Finally we have n - r + 1 choices for the rth space, giving a total of n(n - 1)(n - 2)ÿ(n - r + 1)
possible arrangements (or permutations). This can be written in terms of factorials as follows:

n(n & 1)(n & 2)ÿ(n & r % 1)

n(n & 1)(n & 2)ÿ(n & r % 1)(n & r)(n & r & 1)ÿ3@2@1
'
(n & r)(n & r & 1)ÿ3@2@1

n!
' .
(n & r)!

It should be noted that this is not the number of combinations of r objects taken from a set of n,
since in permutations order is important; in combinations it is not. For example, if we consider
the three objects a, b, and c, the number of permutations of two objects chosen from them is 3@2,
3!
the arrangements ab, ba, bc, cb, ca, ac. However, the number of combinations is ' 3,
2!1!
and the combinations are a and b, b and c, and a and c.

Next we define even and odd permutations of 1, 2, ..., n; this topic is used in Chapter 9
and in higher algebra.

We begin with the case n = 3, that is the numbers 1, 2, 3. With each permutation

i, j, k

of these three numbers, we associate the product of differences

p = (j - i)(k - i)(k - j).

If p is positive, the associated permutation is called even; if p is negative, the associated

permutation is odd. Three of the 3! permutations of 1, 2, 3 are even and three are odd. The even
ones are listed in the first column, and the odd ones in the second column:

54
 1, 2, 3 1, 3, 2
2, 3, 1 2, 1, 3
3, 1, 2 3, 2, 1

For general n, a permutation i, j, h, k, ... , r, s of 1, 2, 3, ... , n is associated with the product

p = [(j - i)][(h - i)(h - j)][(k - i)(k - j)(k - h)] ...

[(s - i)(s - j)(s - h)(s - k)...(s - r)]

of all the differences of two of i, j, h, k, ... , r, s in which the number that appears first is
subtracted from the other. If the permutation i, j, h, k, ... , r, s is written in the notation a1, a2, a3,
a4, ... , an - 1, an, then the product p takes the form

p ' [(a2 & a1)][(a3 & a1)(a3 & a2)][(a4 & a1)(a4 & a2)(a4 & a3)]ÿ
[(a n & a1)(an & a2)(an & a3)(an & a4)ÿ(a n & an & 1)].

If the product p is positive, the permutation is even; if p is negative, the permutation is odd.

Problems for Chapter 7

1. Write out all the combinations of two letters chosen from a, b, c, d, and e.

2. Write out all the combinations of three letters chosen from a, b, c, d, and e.

3. Write out all the permutations of two letters chosen from a, b, c, d, and e.

4. Write out all the permutations of three letters chosen from a, b, c, d, and e.

5. Find the positive integer that is the coefficient of x3y7z2 in (x + y + z)12.

6. Express the trinomial coefficient of the previous problem in six ways as a product of two
binomial coefficients.

7. How many combinations are there of 1, 2, 3, or 4 elements from a set of 5 elements?

8. How many non-empty proper subsets are there of a set of n elements? That is, how many
combinations are there of 1, 2, ... , or n - 1 elements?

9. Express the coefficient of x3y7w2 in (x + y + z + w)12 in six different ways as a product of two
binomial coefficients.

55
 10. Express the coefficient of x2y3z4w2 in (x + y + z + w)11 in six different ways as a product of
three binomial coefficients.

11. Find the coefficient of x2y9z3w in (2x + y - z + w)15.

12. Find the coefficient of xryzw in (x + y + z + w)r+3.

13. Show that x5y2z9 has the same coefficient in (x + y + z + w)16 as in (x + y + z)16.

14. What is the relation between the coefficient of xy7z2 in (x + y - z)10 and its coefficient in
(x - y + z + w)10?

15. Let a, b, and n be positive integers, with n > a + b. Show that

n n& a n n&b n n& a n %1 n& a% 1

% % ' .
a b b a&1 a b& 1 a b

16. Express the coefficient of x2y4z6 in (x + y + z)12 as the sum of three of the trinomial
coefficients in the expansion of (x + y + z)11.

17. What is the sum of all the trinomial coefficients in (x + y + z)100?

18. What is the sum of the coefficients in each of the following:

(a) (x + y - z)100?

(b) (x - y + z - w)100?

19. List the even permutations of 1, 2, 3, 4.

20. List the odd permutations of 1, 2, 3, 4.

R 21. Let P be a permutation i, j, h, ... k of 1, 2, 3, ... , n.

(a) Show that if i and j are interchanged, P changes from odd to even or from even to odd.

(b) Show that if any two adjacent terms in P are interchanged, P changes from odd to even
or from even to odd.

(c) Show that the interchange of any two terms in P can be considered to be the result of
an odd number of interchanges of adjacent terms.

(d) Show that if any two terms in the permutation P are interchanged, P changes from odd
to even or from even to odd.

56
 (e) Given that n $ 2, show that half of the permutations of 1, 2, ... , n are even and half

n!
are odd, that is, that there are even permutations and the same number of odd
2
ones.

R 22. (a) Let P be a permutation a, b, c, d of the numbers 1, 2, 3, 4. Let d = 4 and let Q be the
associated permutation a, b, c of 1, 2, 3. Show that P and Q are either both even or
both odd.

(b) Let R be a permutation i, j, ..., h, n of the numbers 1, 2, ..., n-1, n in which the last
term of R is n. Let S be the associated permutation i, j, ..., h of 1, 2, ..., n-1 obtained
by dropping the last term of R. Show that R and S are either both even or both odd.

23. How many triples of positive integers r, s, and t are there with r < s < t and:

(a) r + s + t = 52?

(b) r + s + t = 352?

1 2 4 7
24. The arrangement has the property that the numbers increase as one goes down
3 5 6 8
or to the right.

(a) How many other arrangements are there of the numbers 1, 2, ..., 8 in 2 rows and 4
columns with this property?

(b) How many arrangements are there of the numbers 1, 2, ..., 14 in 2 rows and 7 columns
with this property?

(c) How many arrangements are there of the numbers 1, 2, ..., 12 in 3 rows and 4 colunms
with this property?

57
 Chapter 8

POLYNOMIAL EQUATIONS

An nth degree polynomial with complex coefficients is of the form

a0xn + a1xn -1 + ... + an -1x + an

where the ai are complex numbers and a0 … 0. (Of course, the coefficients a0, ..., an may be

real numbers, since a real number is a special case of a complex number.)

Thus a first degree polynomial is of the form ax + b and a second degree polynomial of the
form ax2 + bx + c, with a … 0 in both cases. A non-zero constant a is a polynomial of degree
zero; the constant zero is also a polynomial, but it is not assigned a degree.
The polynomial equation y = x2 - 6x + 1 defines y to be a function of x on the domain of
all complex numbers; that is, it provides a rule for assigning a unique complex number y to each
complex number x. The table

x 4 3 2 1 0 i 2i
y = x2 - 6x + 1 -7 -8 -7 -4 1 -6i -3 - 12i

shows that this functional rule assigns -4 to 1, 1 to 0 , -6i to i, etc. A rule that makes y a function
of x assigns precisely one value y to a fixed x; however, the same number y may be assigned to
more than one x, as is seen here with -7 assigned to 2 and to 4.
It is sometimes convenient to represent the rule that defines y to be a function of x by the
symbol f(x). This notation enables one to express in a simple way the number assigned to a given
x by the function. For example, f(1), f(2), and f(3) stand for the numbers assigned to 1, 2, and 3,
respectively. If f(x) = x2 - 6x + 1, then f(1) = -4, f(2) = -7, f(3) = -8,

2
f 2 ' 2 & 6 2 % 1,

f(a + b) = (a + b)2 - 6(a + b) + 1,

and
f(x + 1) = (x + 1)2 - 6(x + 1) + 1.

Notice that f(a + b) is not necessarily the same as f(a) + f(b), since f(a + b) is the result of
replacing x in x2 - 6x + 1 by a + b and is not f times a + b.
If several functions are involved in a given discussion, one may use g(x), F(x), p(x), q(x),
and so on, as alternates for f(x).

58
8.1 THE FACTOR AND REMAINDER THEOREMS

If an nth degree polynomial p(x) = a0xn + a1xn -1 + ... + an can be factored in the form

p(x) = a0(x - r1)(x - r2)...(x - rn), a0 … 0,

then the roots of the polynomial equation p(x) = 0 are found by setting each of the factors equal
to zero, since a product of complex numbers is zero if and only if at least one of the factors is
zero. Therefore, the roots are r1, ..., rn. We wish to establish a form of converse to this result: we
wish to show that if r is a root of a polynomial equation p(x) = 0 then it follows that x - r is a
factor of p(x); that is p(x) can be expressed in the form

p(x) = (x - r)q(x)

where q(x) is a polynomial in x.

THE FACTOR THEOREM: Let

p(x) = a0xn + a1xn -1 + ... + an

be a polynomial in x. If r is a root of p(x), that is, if p(r) = 0, then x - r is a factor of p(x).

Proof: Using the hypothesis that p(r) = 0, we have

p(x) ' p(x) & 0

p(x) ' p(x) & p(r)
p(x) ' (a0x n % a1x n&1 % þ % an) & (a0r n % a1r n&1 % þ % a n)
(1) p(x) ' a0(x n & r n) % a1(x n&1 & r n&1) % þ % (an & a n).

Since x - r is a factor of xn - rn, xn -1 - rn -1, and so on (see Example 2, Chapter 5), it follows that
x - r is a factor of the entire right side of equation (1), and so is a factor of p(x).

We next use this theorem to obtain information concerning the case in which r is not a
root of p(x).

THE REMAINDER THEOREM: Let p(x) be a polynomial. Then for every complex number r
there is a polynomial q(x) such that

(2) p(x) = (x - r)q(x) + p(r)

Proof: Let us define a new polynomial f(x) by

59
 f(x) = p(x) - p(r)

Then f(r) = p(r) - p(r) = 0. Hence r is a root of f(x) and, by the Factor Theorem, above, x - r is a
factor of f(x), and so there is a polynomial q(x) such that

f(x) = (x - r)q(x)

Now p(x) - p(r) = (x - r)q(x), since both sides are equal to f(x); equation (2) is then obtained by
transposing p(r).
The polynomial p(r) is the remainder in the division of p(x) by x - r. In specific cases, the
quotient polynomial q(x) of (2), above, may be found by long division or by a more compact
form of division called synthetic division. We first illustrate these techniques on the example in
which

p(x) = x3 - 7x2 + 4x + 9, r =2.

Dividing p(x) by x - 2, we have

x2 - 5x -6
x-2 x3 - 7x2 + 4x + 9
x3 - 2x2
- 5x2 + 4x
- 5x2 + 10x
- 6x + 9
- 6x + 12
- 3
This shows that

x3 - 7x2 + 4x + 9 = (x - 2)(x2 - 5x - 6) - 3.

That is, p(x) = (x - 2)q(x) + p(2), with q(x) = x2 - 5x - 6 and p(2) = -3.
The synthetic form of the division is as follows:

2 1 -7 4 9
2 -10 -12
1 -5 -6 -3

60
The steps in this synthetic form of the division are explained in the treatment of the general case
which follows.

The synthetic division of

p(x) = a0xn + a1xn -1 + ... + an -1x + an

by x - h is in the form

h a0 a1 a2 ... an -1 an
b1 b2 ... bn -1 bn
c0 c1 c2 ... cn -1 cn

where c0 = a0, b1 = hc0, c1 = a1 + b1, b2 = hc1, c2 = a2 + b2, ..., bn = hcn -1, cn = an + bn. In general,
each b is h times the previous c, c0 = a0, and each succeeding c is the sum of the a and b above it.
The last c, cn , is the value of p(h), and the other c’s are the coefficients of q(x) in the formula
p(x) = (x - h)q(x) + p(h); they give us the expression

p(x) = (x - h)(c0xn -1 + c1xn -2 + ... + cn -2x + cn -1) + cn .

Example 1. Express p(x) = x5 + 25x2 + 7 in the form (x + 3)q(x) + p(-3).

Solution: We note that h = -3 and that a0 = 1, a1 = 0, a2 = 0, a3 = 25, a4 = 0, and a5 = 7 in

this problem. The synthetic division is therefore written

-3 1 0 0 25 0 7
-3 9 -27 6 -18
1 -3 9 -2 6 -11

Hence:

x5 + 25x2 + 7 = (x + 3)(x4 - 3x3 + 9x2 - 2x + 6) - 11.

Example 2. Use synthetic division to show that 5 is a root of p(x) = 2x3 - 40x - 50 = 0, and use
this fact to solve the equation.

Solution: We divide p(x) by x - 5 with the object of showing that the remainder p(5) is
zero. Thus:

61
 5 2 0 -40 -50
10 50 50
2 10 10 0

This shows us that p(x) = (x - 5)(2x2 + 10x + 10). The roots of p(x) = 0 are therefore obtained
from

x - 5 = 0, 2(x2 + 5x + 5) = 0

as 5 and (&5 ± 25 & 20)/2; we have, then

5, (&5 % 5)/2, and (&5 & 5)/2.

Example 3. Let f(x) = 9x3 + x2 - 7x + 4. Find numbers a, b, c, and d such that

(3) f(x) ' a % b(x % 1) % c(x % 1)2 % d(x % 1)3.

We give two solutions.

First solution: Letting x = -1 in (3), we see that a = f(-1). We therefore use synthetic
division to express f(x) in the form (x + 1)g(x) + f(-1) and find that g(x) = 9x2 - 8x + 1 and
a = f(-1) = 3. Now (3) becomes

(x + 1)(9x2 - 8x + 1) + 3 = 3 + b(x + 1) + c(x + 1)2 + d(x + 1)3.

On each side we subtract 3 and then divide by x + 1, thus obtaining

(4) g(x) ' 9x 2 & 8x % 1 ' b % c(x % 1) % d(x % 1)2.

Letting x = -1, we see that b = g(-1). We therefore treat g(x) as f(x) was treated above, and find
that g(x) = (x + 1)(9x - 17) + 18. Hence b = 18. Then (4) becomes

(x + 1)(9x - 17) + 18 = b + c(x + 1) + d(x + 1)2.

This leads to

9x - 17 = c + d(x + 1)

or

62
 9(x + 1) - 26 = c + d(x + 1).

Hence c = -26 and then d = 9.

Alternate solution: Let x + 1 = y. Then x = y - 1 and

f(x) = f(y - 1) = 9(y - 1)3 + (y - 1)2 - 7(y - 1) + 4.

Expanding and collecting like terms, we obtain

f(x) ' 3 % 18y & 26y 2 % 9y 3

' 3 % 18(x % 1) & 26(x % 1)2 % 9(x % 1)3.

8.2 INTEGRAL ROOTS

Let the coefficients ai of the polynomial equation

a0xn + a1xn -1 + ... + an -1x + an = 0

be integers. Then it can be shown that the only possibilities for integral roots are the integral
divisors of the last coefficient an. For example, an integer that is a root of

x4 + x3 + x2 + 3x - 6 = 0

would have to be one of the eight integral divisors ±1, ±2, ±3, ±6 of -6. Trial of each of these
eight integers, as in Example 2 in Section 8.1, would show that 1 and -2 are the only integral
roots. The work can be reduced, when one root is found, by substituting the quotient polynomial
for the original polynomial in further work. Thus

1 1 1 1 3 -6
1 2 3 6
1 2 3 6 0

shows that x4 + x3 + x2 + 3x - 6 = (x - 1)(x3 + 2x2 + 3x + 6). Hence, 1 is a root and the other roots
are the roots of the equation x3 + 2x2 + 3x + 6 = 0.

8.3 RATIONAL ROOTS

We now consider a polynomial equation

63
 a0xn + a1xn -1 + ... + an -1x + an = 0, a0 … 0

of degree n with integer coefficients ai. It can be shown that if there is a rational root p/q, with p
and q integers having no common integral divisor greater that 1, then p must be an integral divisor
of an and q must be an integral divisor of a0. For example, if the rational number p/q in lowest
terms is a root of

6x4 - x3 - 6x2 - x - 12 = 0

then p must be one of the twelve integral divisors ±1, ±2, ±3, ±4, ±6, ±12 of -12 and q one
of the integral divisors of 6. Without losing any of the possibilities, we may restrict q to be
positive, that is, to be one of the integers 1, 2, 3, 6. The possible rational roots, therefore, are

±1, ±2, ±3, ±4, ±6, ±12, ±1/2, ±3/2, ±1/3, ±2/3, ±4/3, ±1/6.

Trials would show that 3/2 and -4/3 are the only rational roots.

Example. Prove that 2 % 3 is not a rational number.

Solution: Let a ' 2 % 3. Then

a & 2 ' 3
2
a & 2 2a % 2 ' 3
a2 & 1 ' 2 2a
a 4 & 2a 2 % 1 ' 8a 2
a 4 & 10a 2 % 1 ' 0.

Hence a is a root of x4 - 10x2 + 1 = 0. This fourth degree polynomial equation has integer
coefficients. The rule on rational roots tells us that the only possible rational roots are 1 and -1.
Substituting, we see that neither 1 nor -1 is a root. Hence there are no rational roots. Since a is a
root, it follows that a is not rational.

Problems for Sections 8.1, 8.2, and 8.3

1. Express p(x) = x4 + 5x3 - 10x - 12 in the form (x + 2)q(x) + p(-2).

2. Express f(x) = 5x5 - x4 - x3 - x2 - x - 2 in the form (x - 1)g(x) + f(1).

3. Show that -1 is a root of x3 + 3x2 - 2 = 0, and find the other roots.

64
4. Show that 2 is a root of x3 - 6x + 4 = 0, and find the other roots.

5. Find a, given that -4 is a root of 5x6 - 7x5 + 11x + a = 0.

6. Find b, given that 3 is a root of x7 - 10x5 + 8x3 + 4x2 - 3x + b = 0.

7. Find all the integral roots of x4 - 2x3 - x2 - 4x - 6 = 0, and then find the other roots.

8. Find all the integral roots of x5 - 8x4 + 15x3 + 8x2 - 64x + 120 = 0, and then find the other
roots.

9. Let f(x) = (x - a)3 - x3 + a3. Find f(0) and f(a), and use this information to find two factors of
f(x).

10. Let g(x) = (x - a)5 - x5 + a5. Show that f(x) is divisible by x and by x - a, and find the other
factors.

11. Find all the integral roots of 3x4 + 20x3 + 36x2 + 16x = 0, and then find the other roots.

12. Let f(x) = a0xn + a1xn -1 + ... + an -1x; that is, let an = 0. Also, let the ai be integers. Show that
any non-zero integral root of f(x) = 0 is an integral divisor of an -1.

13. Find a rational root of 3x3 + 4x2 - 21x + 10 = 0, and then find the other roots.

14. Find all the roots of 6x4 + 31x3 + 25x2 - 33x + 7 = 0.

15. Find all the roots of 81x5 - 54x4 + 3x2 - 2x = 0.

16. Let f(x) = a0xn + a1xn -1 + ... + an -1x with the ai integers. State a necessary condition for a
non-zero rational number to be a root of f(x) = 0.

17. Given that a and b are integers, what are possibilities for rational roots of
x3 + ax2 + bx + 30 = 0?

18. Let f(x) = xn + a1xn -1 + ... + an -1x + an with the ai integers. Note that a0 = 1. Show that any
rational root of f(x) = 0 must be an integer.

19. Let f(x) be a polynomial. Let r and s be roots of f(x) = 0 and let r … s. Show that there
exist polynomials g(x) and h(x) such that all of the following are true:

(a) f(x) = (x - r)g(x).

(b) g(s) = 0.
(c) g(x) = (x - s)h(x).
(d) f(x) = (x - r)(x - s)h(x).

65
20. Let f(x) = 0 be a polynomial equation with distinct roots r, s, and t. Show that
f(x) = (x - r)(x - s)(x - t)p(x), with p(x) a polynomial.

21. Prove that if r1, r2..., rn are distinct roots of a polynomial equation f(x) = 0, then f(x) is a
multiple of (x - r1)(x - r2)...(x - rn).
22. Prove that 3 & 2, 2 & 3, and & 2 & 3 are all irrational.

23. Prove that 5 % 3, 5 & 3, & 5 % 3, and & 5 & 3 are all irrational.

3
24. Prove that 14 is irrational.
25. Find an eighth-degree polynomial equation with integer coefficients that has
2 % 3 % 7 as a root.

26. If f(x) is a function of x, the notation )f(x) represents f(x + 1) - f(x). Show that

)x 2 ' 2x % 1 and )x 3 ' 3x 2 % 3x % 1.

27. Let )f(x) ' f(x % 1) & f(x). Find )f(x) for each of the following:

(a) f(x) = a + bx.

(b) f(x) = a + bx + cx2.
(c) f(x) = a + bx + cx2 + dx3.
(d) f(x) = xn, with n a positive integer.

28. Find f(x + 2) - 2f(x + 1) + f(x) for:

(a) f(x) = a + bx.

(b) f(x) = a + bx + cx2.

29. Find f(x + 3) - 3f(x + 2) + 3f(x + 1) - f(x) for:

(a) f(x) = a + bx.

(b) f(x) = a + bx + cx2.
(c) f(x) = a + bx + cx2 + dx3.
30. Let )nf(x), with n a positive integer, be defined inductively by

66
 )1f(x) ' )f(x) ' f(x % 1) & f(x),
)2f(x) ' )[)f(x)] ' )[f(x % 1) & f(x)]
' [f(x % 2) & f(x % 1)] & [f(x % 1) & f(x)],
) f(x) ' )[) f(x)],
3 2

...,
)m% 1f(x) ' )[)mf(x)],
... .

[The function )nf(x) is called the nth difference of f(x).] Show that

)nf(x) ' ' (&1)k

n
n
f(x % n & k).
k'0 k

31. Let )nf(x) be defined as in Problem 30 above and show:

(a) )nf(x) ' 0, if f(x) is a polynomial of degree less than n.

(b) )nf(x) ' n!a0, if f(x) = a0xn + a1xn -1 + ... + an.

32. Let f(x) = a + bx + cx2. Let r = f(0), s = f(1) - f(0), and t = f(2) - 2f(1) + f(0).

(a) Show that f(x) = r + sx + tx(x - 1)/2.

(b) Generalize this problem.

33. Let f(x) = 5x4 - 6x3 - 3x2 + 8x + 2. Use repeated synthetic division to find numbers a, b, c, d,
and e such that

f(x) = a + b(x - 2) + c(x - 2)2 + d(x - 2)3 + e(x - 2)4.

34. Use the method of the alternate solution for Example 3 in Section 8.1 to do Problem 33.

35. Let f(x) = x3 + ax2 + bx + c, and let a, b, c, and r be complex numbers. Show that

f(x) = f(r) + (3r2 + 2ar + b)(x - r) + s(x - r)2 + (x - r)3,

and express s in terms of a and r.

36. Let f(x) = x3 + ax2 + bx + c, and let r be a root of f(x) = 0. Show that f(x) is divisible by
(x - r)2 if and only if 3r2 + 2ar + b = 0.

67
37. Let f(x) = x3 + ax2 + bx + c, g(x) = 3x2 + 2ax + b, and h(x) = 6x + 2a. Show that
f(x) = (x - r)3 if and only if f(r) = g(r) = h(r) = 0.

38. Let f(x) = x4 + ax3 + bx2 + cx + d. Find s, t, and u in terms of a, b, c, and r such that

f(x) = f(r) + s(x -r) + t(x - r)2 + u(x - r)3 + (x - r)4.

39. Do the methods of this chapter enable you to solve x3 - 3x + 1 = 0?

8.4 SYMMETRIC FUNCTIONS

If we multiply out (x - a)(x - b)(x - c)(x - d), we obtain an expression of the form
x4 - s1x3 + s2x2 - s3x + s4, where

s1 = a + b + c + d,
s2 = ab + ac + ad + bc + bd + cd,
s3 = abc + abd + acd + bcd,
s4 = abcd.

We note that sk is the sum of all products of a, b, c, and d taken k at a time. It is also clear that
the sk are symmetric functions of a, b, c, and d; that is, they do not change value when any two
of a, b, c, and d are interchanged.
The Fundamental Theorem of Algebra, the proof of which is too advanced for this
book, states that the general nth degree polynomial with complex coefficients

a0xn + a1xn-1 + ... + an-1x + an, a0 … 0

has a factorization into linear factors

a0(x - r1)(x - r2) ... (x - rn)

where the ri are complex numbers. One can then see that (-1)kak/a0 is the sum of all the products
of k factors chosen from r1, r2, ..., rn.
The absolute value of a real number x is written as |x| and is defined as follows: If
x $ 0, then |x| = x; if x < 0, then |x| = -x.

Problems for Section 8.4

1. Let 3(x - r)(x - s) = 3x2 - 12x + 8. Find the following:

(a) r + s. (d) r2 + s2.

(b) rs. (e) r2 - 2rs + s2.
(c) (r + s)2. (f) |r - s|.

68
2. Find the sum, product, and absolute value of the difference of the roots of 5x2 + 7x - 4 = 0.

3. Let (x - r)(x - s) = x2 + x + 1. Show the following:

(a) r = -(s + 1), s = -(r + 1).

(b) r3 + r2 + r = 0 = s3 + s2 + s.
(c) r = s2, s = r2.
(d) r-1 + s2 = -1, s-1 + r2 = -1.
(e) r4 + r-1s-1 + s4 = 0.
(f) r9 - r6 + r3 - 1 = 0 = s9 - s6 + s3 - 1.
(g) r10 + s7 + r4 + s = -2 = s10 + r7 + s4 + r.
(h) (r2 - r + 1)(s2 - s + 1)(r4 - r2 + 1)(s4 - s2 + 1) = 16.

4. Let r be a root of x2 + x + 1 = 0. Show the following:

(a) x3 - a3 = (x - a)(x - ar)(x - ar2).

(b) x3 + y3 + z3 - 3xyz = (x + y + z)(x + ry + r2z)(x + r2y + rz).

5. Let a, b, and c be the roots of x3 + 3x + 3 = 0. Find (a + 1)(b + 1)(c + 1).

6. Given that (x - a)(x - b) = x2 - px + q, express each of the following in terms of p and q:

(a) a + b.
(b) ab.
(c) a2 + 2ab + b2.
(d) a2 + ab + b2.
(e) ab(a2 + ab + b2).
(f) a3b3.
(g) The coefficients of the expansion of (x - a2)(x - ab)(x - b2).

7. Let (x - r)(x - s) = x2 - px + q, and let (x - r3)(x - r2s)(x - rs2)(x - s3) = x4 - ax3 + bx2 - cx + d.
Express a, b, c, and d in terms of p and q.

8. Let (x - a)(x - b) = x2 - ex + f, (x - c)(x - d) = x2 - gx + h, and (x - ac)(x - ad)(x - bc)(x - bd) =

x4 - px3 + qx2 - rx + s. Find p, q, r, and s in terms of e, f, g, and h.

9. Let (x - a)(x - b)(x - c) = x3 - 3x + 1. Find each of the following:

(a) 2(a + b + c).

(b) (a + b)(a + c) + (a + b)(b + c) + (a + c)(b + c).
(c) (a + b)(a + c)(b + c).
(d) the equation y3 - py2 + qy - r = 0 whose roots are a + b, a + c, and b + c.

10. Do Problem 9 with (x - a)(x - b)(x - c) = x3 + 3x - 1.

69
11. Let s1 = a + b + c, and s2 = ab + ac + bc, and s3 = abc. Find numbers x, y, z, t, u, v, and w
such that for all a, b, and c:

(a) a3 + b3 + c3 = xs3 + ys1s2 + zs13.

(b) a4 + b4 + c4 = ts1s3 + us22 + vs12s2 + ws14.

12. Let (x - r)(x - s)(x - t) = x3 - ax2 + bx - c. Express r5 + s5 + t5 in terms of a, b, and c.

13. Let (x -1)(x -2)(x - 3)...(x - n) = xn - s1xn-1 + s2xn-2 - ... + (-1)nsn. Show the following:
n%1
(a) s1 ' .
2
(b) sn = n!.

(c) 2s2 = (13 + 23 + ... + n3) - (12 + 22 + ... + n2).

70
 Chapter 9

DETERMINANTS

Let us consider a pair of simultaneous equations:

(A) 2x - 3y = 9
(B) 8x + 5y = 7

We seek numbers x and y which satisfy (A) and (B) simultaneously, that is, a pair x,y for which
both (A) and (B) become true statements. Such a pair also satisfies the equations

(A') 10x - 15y = 45

(B') 24x + 15y = 21

obtained by respectively multiplying both sides of (A) by 5 and both sides of (B) by 3. A pair x, y
satisfying both (A') and (B') has to satisfy

(C) 34x = 66

which is obtained by adding corresponding sides of (A') and (B'). The only root of (C) is x =
66/34 = 33/17.
If one replaces x by 33/17 in (A), one finds that

2(33/17) & 3y ' 9

66/17 & 3y ' 9
(66/17) & 9 ' 3y
&87/17 ' 3y
&29/17 ' y

Hence the only pair of numbers x,y which might satisfy (A) and (B) simultaneously is x = 33/17,
y = -29/17. Our method of obtaining these numbers shows that they do satisfy (A). We check
that they also satisfy (B) by substituting in the left side as follows:
33 &29 264 &145 119
8@ % 5@ ' % ' ' 7
17 17 17 17 17
This shows that x = 33/17, y = -29/17 is the unique pair that satisfies (A) and (B)
simultaneously.
An equation of the form ax + by = c in which a, b and c are known numbers is called a
first-degree equation in x and y, or a linear equation in x and y. Thus (A) and (B) are an
example of a pair of simultaneous linear equations. The method illustrated above for solving
such a pair is called elimination. More specifically, we eliminated y to obtain equation (C).

71
 Eliminating y (or x) from the simultaneous equations

(D) 6x - 15y = -10

(E) 4x - 10y = -7

leads, upon multiplying both sides of (D) by 2 and both sides of (E) by -3, to

(D') 12x - 30y = -20

(E') -12x + 30y = 21

and, upon adding corresponding sides of (D') and (E'), to

(F) 0=1

Assuming that there is a pair x, y satisfying (D) and (E) simultaneously has led us to the false
conclusion that 0 = 1. Hence there is no pair x, y which satisfies (D) and (E) simultaneously.

9.1 DETERMINANTS OF ORDER 2

Let us now examine the general pair of simultaneous first-degree equations:

a1x % b1y ' c1

(G)
a2x % b2y ' c2

We are going to apply the elimination technique discussed above to (G) and then introduce the
related concept of a determinant, which has important applications in the theory of systems of
equations and in other fields.
If we multiply both sides of the first equation in (G) by b2 and both sides of the second
equation by b1, we obtain
a1b2x % b1b2y ' c1b2
a2b1x % b1b2y ' c2b1

Whence, by subtraction, we obtain:

(1) (a1b2 - a2b1)x = c1b2 - c2b1.

If we use the equations in (G) to eliminate x rather than y, the result is

(2) (a1b2 - a2b1)y = a1c2 - a2c1.

If a1b2 - a2b1 is not zero, we see that the solution of the system (G) is found from (1) and
(2) in the form

72
 c1b2 & c2b1 a1c2 & a2c1
(3) x ' , y '
a1b2 & a2b1 a1b2 & a2b1

We note that the denominators are the same, that the numerator in the expression for x is obtained
from the denominator by replacing the coefficients a1 and a2 of x in (G) with c1 and c2,
respectively, and that the numerator in the expression for y is obtained from the denominator by
replacing the coefficients b1 and b2 of y with c1 and c2.
This motivates us to introduce the notation

/0 / ' ad & bc
00 c d 000
a b
(4)

The square array bordered by vertical lines in (4) is called a two-by-two (2x2) determinant.
With this notation, the equations in (3) can be rewritten as

/0 /0 /0 /0
c1 b1 a1 c1
00c b2000 00a c2000
x ' 0 2 , y ' 0 2
(5)
/0 /0 /0 /0
a1 b1 a1 b1
00a b2000 00a b2000
0 2 0 2

provided that the common denominator is not zero.

Thus we see that the solution of a system of simultaneous first-degree equations can be
written as ratios of determinants in which the denominator is the determinant made up of
coefficients of x and y in the order in which they appear in the equations, while the numerator for
x is the same determinant with the coefficients of x replaced by the constants, and the numerator
for y is the determinant of the denominator with the coefficients of y replaced by the constants.
This technique is called Cramer's Rule. The common denominator is called the determinant of
the system.

Example 1. Solve by determinants:

3x + 2y = 5
x - 7y = 2

Solution: We first evaluate the determinant in the denominator (the determinant of the
system), as follows:

73
 /0 / ' 3(&7) & 1@2 ' &21 & 2 ' &23
001 &7000
3 2

Since this determinant is not zero, the system has the unique solution:

/0 /
002 &7000
5 2
5(&7) & 2@2 &35 & 4 &39 39
x ' ' ' ' '
&23 &23 &23 &23 23

/0 /
001 2000
3 5
3@2 & 1@5 6&5 1 1
y ' ' ' ' ' & .
&23 &23 &23 &23 23
Example 2. Use determinants to investigate the simultaneous equations:

10x & 14y ' 5

(H) 15x & 21y ' 8

Solution: The determinant of the system is

D ' /0 /0 ' 10(&21) & 15(&14) ' &210 % 210 ' 0.

10 &14
0015 &2100

Since D = 0, we cannot solve for x and y in the form of (3) above. However, we can consider
forms (1) and (2), since they do not involve division by zero. Evaluating the other determinants,
we have

/0 / ' 5(&21) & 8(&14) ' &105 % 112 ' 7

008 &21000
5 &14

/0 / ' 10@8 & 15@5 ' 80 & 75 ' 5.

8000
10 5
0015

Thus (1) and (2) lead to

0 ' 0@x ' 7

(I) 0 ' 0@y ' 5.

Because the contradictory equations (I) are implied by the system (H), this latter system has no

74
simultaneous solution.
It can be shown that if the determinant of system (G) is zero, then (G) has no solution or
an infinite number of solutions.

9.2 DETERMINANTS OF ORDER 3

When the elimination technique is applied to the general system

a1x % b1y % c1z ' d1
(1) a2x % b2y % c2z ' d2
a3x % b3y % c3z ' d3

of first-degree equations in the three unknowns x, y, and z, one obtains equations of the form

(2) Dx = E, Dy = F, Dz = G

where D is the three-by-three determinant (or determinant of order 3) defined by

/0 1 1 1/0
a b c
0 0
D ' 000a2 b2 c2000 ' a1b2c3 & a1b3c2 % a2b3c1 & a2b1c3 % a3b1c2 & a3b2c1.
00 0
00a b c 000
0 3 3 30

In the equations above, E, F, and G are obtained by substituting the column of d's for the column
of a's, b's, or c's, respectively, in D. For example,

/0 1 1 1/0
a d c
0 0
F ' 000a2 d2 c2000 ' a1d2c3 & a1d3c2 % a2d3c1 & a2d1c3 % a3d1c2 & a3d2c1.
00 0
00a b c 000
0 3 3 30

If D … 0, it follows from the equations (2) above that the system of simultaneous equations (1)
has the unique solution
E F G
(3) x ' , y ' , z ' .
D D D

As in the case of 2 by 2 determinants, it can be shown that if D = 0, then the simultaneous

equations (1) either have no common solution or an infinite number of common solutions. The
determinant D is called the determinant of the system. The technique of expressing the solution
of a system of simultaneous linear equations in terms of ratios of determinants when the

75
determinant of the system is not zero, as in (3) above, is called Cramer's Rule.
We note that, according to the definition

/0 1 1 1/0
a b c
0 0
D ' 00a2 b2 c200 ' a1b2c3 & a1b3c2 % a2b3c1 & a2b1c3 % a3b1c2 & a3b2c1
00 00
00 0
00a3 b3 c3000
given above, the three-by-three determinant D consists of a sum of products of the form
± aibjc k where i, j, k is a permutation of 1, 2, 3, and the plus sign is chosen when the
permutation is even and the minus sign when it is odd. (For a definition of even and odd
permutations, see Chapter 7.) Since there is a term corresponding to each permutation, the
number of terms is 3! = 6, half preceded by a plus sign and half by a minus sign. (See Problem 21,
Chapter 7.) Is should be noted that these observations also apply to two-by-two determinants

/0 / ' a1b2 & a2b1

a1 b1
00a b 000
0 2 20
in that here the permutation 1, 2 is even and the permutation 2, 1 is odd, so that the 2! = 2 terms
are preceded by the appropriate signs.

Example. Evaluate the determinant of the following system and thus show that the system has a
unique solution:
2x & 3z ' 10
5x % 4y ' 11
y & 6z ' &3

Solution:

/0 /0
2 0 &3
D ' 0005 4 0 000
00 0
000 1 &6000

' 2@4@(&6) & 2@1@0 % 5@1@(&3) & 5@0@(&6) % 0@0@0 & 0@4@(&3)
' &48 & 15 ' &63

Since D … 0, there is a unique solution.

We now introduce a double subscript notation for the entries in a determinant, with the
first subscript giving the row and the second giving the column. In this notation, a determinant of
order 3 can be written as

76
 /0 11 12 13/0
a a a
0 0
D ' 00a21 a22 a2300
00 00
00 0
(4)

00a31 a32 a33000

and the expanded value is
(5) D ' a11a22a33 % a12a23a31 % a13a21a32 & a11a23a32 & a12a21a33 & a13a22a31.

Each term in (5) is of the form ±a1ia2ja3k with the plus sign used if i, j, k is an even permutation

of 1, 2, 3 and the minus sign if i, j, k is an odd permutation. The terms in (5) can be grouped in
several ways, one of which is
(6) D ' a11(a22a33 & a23a32) & a12(a21a33 & a23a31) % a13(a21a32 & a22a31).

The coefficient a21a32 - a22a31 of a13 in (6) is the value of the 2 by 2 determinant that results when
the entire first row and third column of D are removed; the coefficients of a11 and -a12 in (6) can
be characterized similarly.

This motivates the following definitions: Let aij be a given entry in the determinant D of (4) and
let Mij be the 2 by 2 determinant obtained by deleting the ith row and the jth column of D. This
determinant Mij is called the minor of the entry aij. The cofactor Cij of the entry aij is defined by
the formula

Cij = (-1)i+jMij.

For example, the minor of a23 is

M23 ' /0 / ' a11a32 & a12a31

a11 a12
00a a 000
0 31 32 0
and the cofactor of a23 is

C23 = (-1)2+3M23 = -(a11a32 - a12a31).

It now can easily be seen that equation (6) may be rewritten as

D = a11M11 - a12M12 + a13M13

or as

D = a11C11 + a12C12 + a13C13.

77
Problems for Sections 9.1 and 9.2

1. Solve the following system by determinants, that is, by using Cramer’s rule.

9x - 7y = 11
5x + 2y = -4

2. Show by determinants that the following system has no solution.

15x + 20y = -13

21x + 28y = 5

3. Solve the system given in the example of Section 9.2.

4. Use Cramer’s rule to solve the following system:

7x - 3y - z = 0
2x - 2y + z = 5
-x + y + 2z = 25

5. Use Cramer’s rule to solve the following system:

x - y + 3z = 15
2x - 4y + z = 6
3x + 3y - 6z = -3

6. Solve the following system for x, y, and z:

2x + y = 0
y + 2z = 0
x+ z=0

7. Solve the following system for x, y, and z in terms of a, b, and c:

x+y=c
y+z=a
x+z=b

8. Show the following for the determinant

/0 1 1 1/0
a b c
0 0
D ' 000a2 b2 c2000.
00 0
00a b c 000
0 3 3 30

78
 (a) D = a1M1 - a2M2 + a3M3, where M1, M2, and M3 are the minors of a1, a2, and a3
respectively.
(b) D = a2A2 + b2B2 + c2C2, where A2, B2, and C2 are the cofactors of a2, b2, and c2,
respectively.

9. Show the following for the determinant

/0 11 12 13/0
a a a
0 0
D ' 00a21 a22 a2300.
00 00
00 00
00 31 32 3300
a a a

(a) a31M21 - a32M22 + a33M23 = 0.

(b) a11C13 + a21C23 + a31C33 = 0.

10. Show the following:

/ / / /.
a1 b1 a1 a2
(a) 0
00a b 000 000b b 000
'
0 2 20 0 1 20

/0 1 1 1/0 /0 1 2 3/0
a b c a a a
0 0 0 0
(b) 000a2 b2 c200 ' 000b1 b2 b300.
00 00 0 00
00a b c 000 000c c c 000
0 3 3 30 0 1 2 3 0

/0 1 /0 /0 1 1 1/0
a b1 c1 a b c
0 0 0 0
11. (a) Show that 000ka2 kb2 kc200 ' k00a2 b2 c200.
00 0
00 0
00 00
00 a b c 00 00a b c 000
0 3 3 30 0 3 3 30

(b) Show that if each element of a fixed row of a three by three determinant D is multiplied
by a factor k, the new determinant equals kD.

/0 /0
0 0 0
12. (a) Evaluate 000a2 b2 c2000 .
00 0
00a b c 000
0 3 3 30

79
 (b) Show that if a three by three determinant D has a row of zeros, then D = 0.

(c) Show that if a three by three determinant D has a column of zeros, then D = 0.

13. Show the following:

(a) /0 /0 ' &/0 /.

a1 b1 b1 a1
00a b 00 00b a 000
0 2 20 0 2 20

/ / / /.
a1 b1 a2 b2
(b) 0
00a b 000
' &0
00a b 000
0 2 20 0 1 10

/0 1 1 1/0 /0 1 1 1/0
a b c a b c
0 0 0 0
(c) 00a2 b2 c200 ' & 00a3 b3 c300.
00 00 00 00
00 00 00 0
00 3 3 300
a b c 00 2 2 2000
a b c

/0 1 1 1/0 /0 1 1 1/0
a b c c b a
0 0 0 0
(d) 000a2 b2 c200 ' & 000c2 b2 a200.
00 0
0 00 00
00a b c 000 00c b a 000
0 3 3 30 0 3 3 30

/0 1 1 1/0 /0 1 2 3/0
a b c b b b
0 0 0 0
(e) 00a2 b2 c200 ' & 00a1 a2 a300.
00 00 00 00
00 00 00 0
00a3 b3 c300 00c1 c2 c3 000

/0 /0
a b c
14. Evaluate 000a b c000.
00 0
00d e f 000

15. Show that in a three by three determinant if the elements of one row are a constant k times
the elements of another row, then the determinant equals zero.

16. (a) Use the definition of a determinant to show that

80
 /0 /0 /0 / / /
00a % a ) b % b ) c % c )00 ' 00a b c 000 % 000a ) b ) c )000.
a1 b1 c1 a1 b1 c1 a1 b1 c1

00 2 2 2
00
20
00 000 2 2 2000 000 2 2 2 000
00 a 00 00a b c 00 00 0
00 00 3 3 300 00a3 b3 c3 000
2 2

0 3 b 3 c3

(b) Show that if the elements of a given row of a three by three determinant D are f1 + g1,
f2 + g2, f3 + g3, then D = D1 + D2 where D1 results from D by replacing the given row by
f1, f2, f3 and D2 by replacing the given row by g1, g2, g3.

17. Show that if two rows of a three by three determinant D are u1, u2, u3 and v1, v2, v3,
respectively, and if D* is the same as the determinant D except that the row u1, u2, u3 is
replaced by u1 + kv1, u2 + kv2, u3 + kv3 where k is a constant, then D* = D.

9.3 DETERMINANTS OF ORDER n

We have defined the 2 by 2 determinant

D ' /0 /0
a b
00c d 00
(1)

to be the number ad - bc obtained from the square array

a b
(2)
c d
of 4 numbers in two rows and two columns. Thus, bordering the array (2) with vertical lines
converts the array into a symbol for the number D. Similarly, a 3 by 3 determinant is a number
obtained in a certain manner from a square array of 9 numbers.
Our next objective is to define an n by n determinant. More precisely, we seek an
unambiguous rule for obtaining a number from an n by n square array of numbers and want this
rule to agree with the previous definitions when n = 2 or 3. Let

a11 a12 a13 þ a1n

a21 a22 a23 þ a2n
a31 a32 a33 þ a3n
(3) þ
þ
þ
an1 an2 an3 þ ann

81
be an array of n2 numbers aij, where the first subscript designates the row and the second
designates the column.
There are n! products
a1ia2ja3hþank

with exactly one factor from each row and exactly one factor from each column. The determinant
D associated with the array (3) is the sum of the n! terms
±a1ia2ja3hþank

where the plus sign is used when the permutation

i, j, h, ... , k

is even, and the minus sign is used when the permutation is odd. As before, the array (3) is
bordered with vertical lines in writing the symbol

/0 /0
a11 a12 a13 þ a1n
00 a2n 000
00 þ
00
00
a21 a22 a23
00 a3n 000
00 00
a31 a32 a33 þ
00 00
00 00
00 00
(4) þ
00 00
00 00
þ
00 00
00 00
00
þ
0
00 ann 00
0
an1 an2 an3 þ

for a determinant D of order n.

The minor Mij of the entry aij in D is the determinant of order n - 1 obtained by deleting
the ith row and jth column of D. The cofactor of aij is defined by

Cij = (-1)i+jMij .

Associated with the determinant D of (4) is the determinant

/0 /0
a11 a21 a31 þ an1
00 an2000
00 00
00
a12 a22 a32 þ
0
D ) ' 000 an3000
00 00
a13 a23 a33 þ
00 00
00 00
00
þ
0
00 a1n a2n a3n þ ann000

82
in which the entries in the ith column are the corresponding entries in the ith row of D. The
determinant D' is called the transpose of D; it is easily seen that D is then the transpose of D'.
For example

/0 /0 /0 /
00d e f 00 and 00b e h 000
a b c a d g

00 0 00 0
00g h i 000 00c f i 000
0 0 0 0
are transposes of each other.
The following theorem enables one to prove results involving columns of a determinant
from the corresponding results on the rows and vice versa.

THE TRANSPOSE THEOREM: Let D' be the transpose of a determinant D. Then D' = D.

Proof: Let ars and brs be the entries of D and D', respectively, in the rth row and sth
column. Since D' is the transpose of D, brs = asr. By definition, D is the sum of n! terms
(5) ± a1ia2jþanh

where the plus sign is used if the permutation

i, j, ..., h

is even and the minus sign if it is odd. Also, D' is the sum of n! terms
(6) ± b1xb2yþbnz

where the sign is plus if

(7) x, y, ÿ, z

is an even permutation, and minus otherwise. Since brs = asr, each term (6) can be rewritten as
(8) ± ax1ay2þazn

The terms (8) are all the products, with signs attached, in which there is exactly one factor from
each row and from each column of D; hence the terms (8) are the terms of the expansion of D,
except that the signs may not agree. We will therefore prove that D = D' by showing that (8) has
the same sign as a term of the expansion of D that it has in the expansion of D'. We do this by
describing a method of determining the sign whether or not the row numbers of the entries are in
the order 1, 2, ..., n. Let
(9) ± apuaqvþarw

be a term of the expansion of D with its factors in any order. Then the row subscript numbers and
the column subscript numbers give us the two permutations:

83
 p, q, ÿ, r
(10) u, v, ÿ, w
We wish to prove that the plus sign should be used in (9) if both of the permutations of (10) are
even or both are odd, and that the minus sign should be used when one is even and the other odd.
When the entries in (9) have their row numbers in the normal order, the permutations (10) are of
the form
1, 2, ..., n
i, j, ..., h

with the top one even, and hence the new 2-permutation rule indicates that the sign should be plus
when the bottom permutation is even and minus otherwise. This agrees with the definition of a
determinant and shows that the new rule is correct in this case.
We can go from the order a1ia2j...anh of the factors to any order apuaqv...arw by means of a
number of interchanges of adjacent factors. Whenever one such interchange is made, the row
subscript permutation and column subscript permutation will each change from even to odd or
from odd to even. This means that the new rule will continue to indicate the correct sign as these
interchanges are made. We are especially interested in the case in which the column numbers are
in order, that is, the case in which (9) is (8)
± ax1ay2þazn .
The permutations (10) then become
x, y, ÿ, z
(11)
1, 2, ÿ, n
with the bottom one even. Hence the 2-permutation rule indicates that the plus sign is used if and
only if the permutation x, y, ..., z of (11) is even. This is exactly the rule for determining the sign
of (8)
± ax1ay2þazn
as a term of D'. Hence the sign for (8) is the same either as a term of D' or of D, since the sign in
both cases agrees with the 2-permutation rule. This shows that D' = D and completes the proof.
Let

a11x1 % a12x2 % a13x3 % þ % a1nx n ' b1

a21x1 % a22x2 % a23x3 % þ % a2nx n ' b2
(12) a31x1 % a32x2 % a33x3 % þ % a3nx n ' b3
þ
an1x1 % an2x2 % an3x3 % þ % annxn ' bn

be a system of n simultaneous first-degree equations in n unknowns, x1, x2, ..., xn. The
determinant D of (4) is the determinant of the system for (12).
Let D1, D2, ..., Dn be the determinants that result when the column of b’s in (12) is
substituted for the first, second, ..., nth column, respectively, of the D of (4). The general
Cramer’s Rule states that if D … 0, then the system (12) has the unique solution

84
 D1 D2 Dn
(13) x1 ' , x2 ' , ÿ, x n '
D D D

and that if D = 0, then the system either has no solution or an infinite number of solutions. We do
not give the proof of this rule for general n.
When the number of equations in (12) is large, it becomes very difficult to evaluate the n + 1
determinants in (13) by the methods discussed in this book. More advanced texts describe
variations of the elimination techniques that are practical for the numerical approximation of
determinants of large order or in the solution of systems (12) with n large. (For example, see the
description of Crout’s Method in the appendix of F. B. Hildebrand’s Methods of Applied
Mathematics, Prentice-Hall, 1952.)

Problems for Section 9.3

In Problems 1-7, below, D represents a determinant of order n. Prove each statement

either from the definition of an n by n determinant, by using the Transpose Theorem, or by using
previous results.

R 1. If all the entries on a given row (or column) of D are multiplied by a fixed number k, the
value of D is multiplied by k.

R 2. If each entry in a given row (or column) of D is zero, then D = 0.

R 3. (a) If any two columns of a determinant D are interchanged, the resulting determinant D1
equals -D. (See Problem 21, Chapter 7.)
(b) If any two rows of a determinant D are interchanged, the resulting determinant D2
equals -D.

R 4. If the entries of a row (or column) of D are a constant k times the corresponding entries of
another row (or column), then D = 0.

R 5. If the entries of a given row (or column) of D are f1 + g1, f2 + g2, ..., fn + gn, then D = D1 + D2,
where D1 results from D by replacing the given row (or column) by f1, f2, ..., fn and D2 by
replacing the given row (or column) by g1, g2, ..., gn.

R 6. Let u1, u2, ..., un and v1, v2, ..., vn be the entries of two rows (or columns) of D, and let D*
result from replacing v1, v2, ..., vn in D by v1 + ku1, v2 + ku2, ..., vn + kun, respectively. Then
D* = D.

R 7. Let aij be the entry in the ith row and jth column of D.

(a) If S is the sum of all the terms of the expansion of D that involve ann, then S = annMnn =
annCnn, where Mnn and Cnn are the minor and cofactor of ann. (See Problem 22, Chapter 7.)

85
 (b) Let T be the sum of all the terms of the expansion of D that involve a fixed entry ahk.
Then T = (-1)h+kahkMhk = ahkChk. (Use Problem 3, above, and Part (a) of this problem.)

(c) If h is one of the numbers 1, 2, ...., n, then each term of the expansion of D has one and
only one of the entries ah1, ah2, ..., ahn as a factor.

(d) D ' (&1)h%1ah1Mh1 % (&1)h%2ah2Mh2 % þ % (&1)h%nahnMhn.

(e) D ' ah1Ch1 % ah2Ch2 % þ % ahnChn.

(f) If k is any one of the numbers 1, 2, ..., n then D ' a1kC1k % a2kC2k % þ % ankCnk .

8. (a) Find specific numbers a, b, c, and d such that the polynomial f(x) = ax3 + bx2 + cx + d has
the values listed in the table below. Check, using f(5) = 165.

x 1 2 3 4
f(x) 1 10 35 84

(b) Follow the instructions of the previous part for the table below. Check, using f(4) = 30.

x 0 1 2 3
f(x) 0 1 5 14

9. (a) Show that

/0 / ' a11a22
a11 a12
00 0 a 000
0 220

and that

/0 11 12 13/0
a a a
00 0 a a 00
00 2300 ' a11a22a33.
00 0
00 0 0 a 000
22

0 330

(b) Let D be an n by n determinant, with the element aij in the ith row and jth column 0 if
i > j. Show that D = a11a22a33...ann.

86
 / /
1 a a2
10. (a) Evaluate /0 /0 and 000 a 1 a 000.
1 a
00a 100 00 0
00a 2 a 1 000
0 0

/0 /0 /0 /
000 1 & a 2 a & a 3 a 2 & a 4000
1 a a2 a3 1 a a2 a3
00 a a a 2000
(b) Show that 000 00 ' 000 00.
0
1
00a 2 1 a 000 000 3 00
00 00 00 00
00 3 00 0
2
a 0 1 & a a & a
00a a2 a 1 00 0 000 0 0 1 & a 000
2

/0 /0
1 a a2 a3 a4
00 a a a 2 a 3000
00 00
0
1
(c) Evaluate 000a 2 a 1 a a 2000.
00 00
00a 3 a a 1 a 000
00 00
00 4
2

00a a a a 1 000
3 2

(d) In the determinants of Parts (a), (b), and (c) of this problem, the element cij in the ith
row and jth column is a|i-j|. Establish a compact formula for the value of the n by n
determinant with cij = a|i-j|. (See Section 8.4 for a definition of |x|.)

/0 /0
r&s s&t t&r
11. Show that 000 s & t t & r r & s000 ' 0.
00 0
00 t & r r & s s & t 000

/0 2 3 1 3 1 2/0 /0 1 2 3/0
x %x x %x x %x x x x
0 0 0 0
12. Show that 000y2 % y3 y1 % y3 y1 % y2000 ' 2 000y1 y2 y3000.
00 0 00 0
00z % z z % z z % z 000 00z z z 000
0 2 3 1 3 1 20 0 1 2 30

/0 /0
a b c
13. Show that 000c a b000 ' a 3 % b 3 % c 3 & 3abc.
00 0
00b c a000

/0 /
a b c d
00d a b c 000
00 00 ' (a % b % c % d)(a & b % c & d)[(a & c)2 % (b & d)2].
00c d a b000
00
14. Show that

00b 0
0 c d a000

87
 /0 /0
x%y z z
15. Evaluate 000 x y % z x 000.
00 0
00 y y z % x000

16. Evaluate

/0 /
1 1 1 1
01 0 0 0000
(a) 000 00.
001 0 1 0000
00 0
001 0 0 1000
0

/0 /
1%a 1%a 1%a 1%a
001 % a a a 000
(b) 00 0.
001 % a a 1 % a a 000
a

00 00
001 % a a 00
0 a 1 % a 0

17. Let F0, F1, F2, ... be the Fibonacci sequence.

(a) Show that /0 /0 ' & /0 /.

Fn % 2 Fn % 1 Fn % 1 Fn
00F 00 00 F F 000
0 n%1 F n 0 0 n n & 10
(b) Find numbers x, y, and z such that F2n = xFn2 + yFnFn+1 + zFn+12.
(c) Find x, y, z, and w such that F3n = xFn3 + yFn2Fn+1 + zFnFn+12 + wFn+13.
(d) Find analogues of the formulas above for the Lucas numbers.

18. Evaluate:

/0 /
1 x y z%w
01 y z w % x000
(a) 000 00.
001 z w x % y 000
00 0
0 w x y % z 000
001

/0 /
x y z w
0y x z w000
(b) 000 00.
00y x w z 000
00 0
00x y w z 000
0

88
19. Let D be an n by n determinant with cij the entry in the ith row and jth column. Show that
D = 0 if n is odd and cij + cji = 0 for all i and j.

20. Evaluate the n by n determinant with the entry cij in the ith row and jth column satisfying
each of the following conditions: (It may be helpful to begin with small values of n and to
try to find a pattern which suggests a proof.)

i%j&2
(a) cij ' .
j&1
(b) cij = c1j if i > j.
(c) cij = a + |i -j|d. (See Section 8.4 for a definition of |x|.)
(d) cij = 1 if j - i is -1, 0, or a positive even integer, and cij = 0 for other values of j - i.
(e) cij = a + x if j > i, cij = b + x if j < i, and cii = ri + x.
1
(f) cij ' for i < j % 2 and cij ' 0 for i $ j % 2.
(j % 2 & i)!

9.4 VANDERMONDE AND RELATED DETERMINANTS

Determinants in which the elements of each column (or row) are the terms 1, r, r2, ..., rn-1
of a geometric progression are called Vandermonde determinants, named for Alexandre-
Théophile Vandermonde (1735-1796), who was the first to give a systematic treatment of the
theory of determinants.
Let us evaluate the 4 by 4 Vandermonde determinant

/0 /
c 000
1 1 1 1
00 x
D ' 000 2 00.
a b
00x 200
c 0
00 00
a2 b2
00x 3 c 3000
0 a3 b3

The expansion by minors of first column entries as outlined in Problem 7, Section 9.3, yields the
following:

/0 2 2 2/0 /0 2 2 2/0 / / / /
0 0 0
D ' 00a b c 00 & x 00a b c 00 % x 2 00 a b c 00 & x 3 00 a b c 000.
a b c 1 1 1 1 1 1 1 1 1

00 0 00 0 00 0 00 0
00a 3 b 3 c 3000 00a 3 b 3 c 3000 00a 3 b 3 c 3000 00a 2 b 2 c 2000
0 0 0 0 0 0 0 0

Letting r, s, t, and u stand for the 3 by 3 determinants in this expression, we may write

D = f(x) = r - sx + tx2 - ux3.

If we let x = a in D, two columns become identical and, by Problem 4, Section 9.3, D becomes
zero. This means that f(a) = 0, and it follows from the Factor Theorem that x - a is a factor of
f(x). Similarly, x - b and x - c are factors of f(x).

89
 If two of the numbers a, b, and c are equal, D has identical columns and thus is zero. We
therefore assume that a, b, and c are distinct numbers. Then f(x) is a multiple of the product of
x - a, x - b, and x - c. Since f(x) is a polynomial of degree 3 or less and has -u as the coefficient of
x3, this means that f(x) must be -u(x - a)(x - b)(x - c). This can be written as

u(a - x)(b - x)(c - x).

We leave it as an exercise for the reader (in Problem 1 below) to show that the 3 by 3 determinant
u is expressible as (b - a)(c - a)(c - b). Substituting this for u in the above gives the result:

D = (a - x)(b - x)(c - x)(b - a)(c - a)(c - b).

Problems for Section 9.4

/0 /
R 1. Show that 00 a b c 000 ' (b & a)(c & a)(c & b).
1 1 1

00 0
00a 2 b 2 c 2000
0 0

/0 /0
1 a a3
2. Show that 0001 b b 3000 ' (b & a)(c & a)(c & b)(a % b % c).
00 0
001 c c 3 000
0 0

/0 /0 / /
1 a2 a3
3. Show that 0001 b 2 b 3000 ' 000a 2 b 2 c 2 000 ' (b & a)(c & a)(c & b)(bc % ca % ab).
a b c

00 0 0 00
001 c 2 c 3 000 000 bc ca ab00
0 0 0 0

/0 2 2 2/0
x y z
4. Express 00x y z 00 as a product of 6 first-degree factors.
00 0
00x 3 y 3 z 3000
0 0

/0 /0
1 &x x 2 &x 3
001 &y y 2 &y 3 000
5. Express 000 00 as a product of 6 first-degree factors.
001 3 00
00 &z z &z 0
00
00
2

001 300
&w w &w 0
2

90
 /
00 /0
1 a % x b % cx % x 2
6. Evaluate /0 / and 001 a % y b % cy % y 000 in factored form.
001 a % y000
1 a%x
00 0
001 a % z b % cz % z 2 000
2

0 0

/0 /0
1 2 & x 3 % 4x % x 2 5 & x 3
001 2 & y 3 % 4y % y 2 5 & y 3 000
7. Evaluate 000 00 in factored form.
001 2 & z 3 % 4z % z 2 5 & z 3 000
00 00
00 300
001 2 & w 3 % 4w % w 5 & w 0
2

8. Show the following:

/0 /
1 x x2
000 1 2x000 ' (y & x)2.
00 0
001 y y 2000
(a)

0 0

/0 /
y 000
1 0 0 1
00 x 1 0
00 00 ' 2(y & x)3.
00x 2 2x 2 200
00 y 0
(b)
00 3 00
00x 3x 2 6x y 3000

9. Find integers r, s, and t such that

/0 /
z 000
1 0 1 1
00 x 1 y
00 00 ' (y & x)r(z & x)s(z & y)t.
00x 2 2x y 2 z 2000
00 00
00 3 300
00x 3x 2 y 3 z 0

/0 /0
1 x x2 x3 x4
000 1 2x 3x 2 4x 3 000
00 00
0
10. Show that 0000 0 2 6x 12x 00 20 ' 2(y & x)6.
00 00
001 y y 2 y 3 y 4 000
00 00
00
000 1 2y 3y 2 4y 3 000

91
 /0 /
00 x 1 y 1 z 000
1 0 1 0 1

00 00
11. Find integers r, s, and t such that 0 0
0 20
00 ' (y & x)r(z & x)s(z & y)t.
00 0
00x 3 3x 2 y 3 3y 2 z 3000
2 2
x 2x y 2y z

00 00
00 4 0
00x 4x y 4y z 000
3 4 3 4

12. Solve the following system of equations if a < b < c:

x% y% z ' 3
ax % by % cz ' a % b % c
a 2x % b 2y % c 2z ' a 2 % b 2 % c 2.

13. (a) Show that the system

x% y% z ' 3
a x % b 2y % c 2z ' a 2 % b 2 % c 2
2

a 3x % b 3y % c 3z ' a 3 % b 3 % c 3

has a solution if the a, b, c are distinct and ab % bc % ca … 0.

(b) Calculate the solution.

14. Evaluate:

/0 /0
1 a a 2 % 2bc
(a) 0001 b b 2 % 2ca000.
00 0
001 c c 2 % 2ab000
0 0

/0 /0
1 a a 2 a 3 % bcd
001 b b 2 b 3 % cda000
(b) 000 00.
001 c c c % dab000
00 00
00
2 3

001 d d 2 d 3 % abc000

92
 /0 /0
1 a a 2 a 3 % 2bcd
001 b b 2 b 3 % 2cda000
(c) 000 00.
001 c c c % 2dab000
00 00
00
2 3

001 d d d % 2abc000
2 3

15. Let D be the general n by n Vandermonde determinant with the entry cij in the ith row and
jth column given by cij = aji-1. Show that D is the product

(a2 - a1)(a3 - a1)(a3 - a2)...(an - an-1)

of all the different ar - as with 1 # s < r # n.

93
 Chapter 10

INEQUALITIES

We deal only with real numbers in this chapter.

10.1 ELEMENTARY PROPERTIES

If a and b are real numbers, a < b (read "a is less that b") is defined to mean that b - a is
positive. This definition and the following three properties can be used in proving elementary
properties of inequalities:

(1) Closure of the positive numbers. If a and b are positive numbers, then a + b and ab
are positive numbers.

(2) Trichotomy. One and only one of the following is true for a given real number a:
(a) a is zero; (b) a is positive; (c) -a is positive.

(3) Roots. If p is a positive number and n is a positive integer, then there is exactly one
positive number r such that rn = p. (This number r is called the positive nth root of p or the
principal nth root of p.)

We can write the statement a < b in the form b > a (read "b is greater than a"). The
notation a # b (read "a is less than or equal to b") means that either a < b or a = b,
and b $ a is defined analogously. The notation x < y < z or z > y > x means that x < y and y < z
are true simultaneously.
As mentioned is Section 8.4, the absolute value of a real number x is written as |x| and is
defined as follows: If x $ 0, then |x| = x; if x < 0, then |x| = -x.

Example 1. Show that if x < y and y < z, that is x < y < z, then x < z.

Solution: If x < y then y - x = p, a positive number, and similarly y < z implies that
z - y = q where q is positive. Hence (z - y) + (y - x) = q + p, or

z - x = q + p.

Since z - x is the sum of the positive numbers q and p, it is positive by the closure property and
thus x < z by the definition.

Example 2. Show that if x < y and p > 0, then px < py.

Solution: If x < y, then y - x is positive and, by closure, the product p(y - x) of positive
numbers is positive; that is, py - px is positive. Now we have px < py by definition.

94
 Example 3. Show that if m and n are integers and m < n, then n & m $ 1.
Solution: Since m < n, it follows that n - m is positive. Since the difference of integers is
an integer and the least positive integer is 1, n - m is at least 1; that is n & m $ 1.

Problems for Section 10.1

R 1. (a) Show that if x < y, then x + z < y + z.

(b) Show that if x < y, then x - w < y - w.

R 2. Show that if x < y and q < 0, then qx > qy.

R 3. (a) Show that if x > 0 or x < 0, then x2 > 0.

(b) Show that for all real x, x 2 $ 0.
(c) Show that 1 > 0.

1 1
R 4. (a) Show that if x > 0, then > 0, and if x < 0, then < 0.
x x
1 1
(b) Show that if 0 < x < y or x < y < 0, then > .
x y

R 5. Show the following:

(a) If 0 < x < y and n is a positive integer, then x2n-1 < y2n-1.
(b) If x < 0 < y and n is a positive integer, then x2n-1 < y2n-1.
(c) If x < y < 0 and n is a positive integer, then x2n-1 < y2n-1.

R 6. (a) Show that if 0 < x < y and n is a positive integer, then x2n < y2n.
(b) Show that if y < x < 0 and n is a positive integer, then x2n < y2n.

R 7. Show that if n is a positive integer and x2n-1 < y2n-1, then x < y. (See Problem 5.)

R 8. (a) Show that if x2n < y2n and y > 0, then &y < ±x < y. (See Problem 6.)
(b) Show that if x2n < y2n and y < 0, then y < ±x < &y.
(c) Use Parts (a) and (b) to show that if x2n < y2n, then &|y| < ±x < |y|.

R 9. Prove the following by mathematical induction:

(a) If a1, a2, ..., an are positive, so is a1 + a2 + ... + an.

(b) If a1, a2, ..., an are positive, so is a1a2...an.
(c) If a1 < b1, a2 < b2, ..., an < bn, then a1 + a2 + ... + an < b1 + b2 + ... + bn.
(d) If 0 < a1 < b1, 0 < a2 < b2, ..., 0 < an < bn, then a1a2...an < b1b2...bn.

95
R 10. Show that:

(a) x 2 & 2xy % y 2 $ 0.

(b) x 2 % y 2 $ 2xy.

(See Problem 3.)

11. Given that x … y, show that:

(a) x & y … 0.
(b) x 2 & 2xy % y 2 > 0.
(c) x 2 % y 2 > 2xy.

12. Find all the integers n such that 2n2 - 3 < 8n.

13. (a) Given that 0 < a < b, show that a2 < ab < b2.
(b) Given that 0 < a < b, show that 3a2 < a2 + ab + b2< 3b2.

a%b
14. (a) Given that a < b, show that a < < b.
2
2a % b a % 2b
(b) Given that a < b, show that a < < < b.
3 3

15. Given that 0 < x < y, show the following:

x&1 y&1
(a) < .
x y
x y
(b) < .
x%1 y%1

x y z
16. Given that 0 < x # y # z and that x + y > z, show that % > .
x%1 y%1 z%1
a%b
17. (a) Given that 0 < a < b # c, show that c > .
2
a%b
(b) Given that 0 < a # b < c, show that c > .
2
a%b
(c) Given that 0 < a # b # c and a < c, show that c > .
2

R 18. Given that 0 < a1 # a2 # a3 # ÿ #an & 1 # an and a1 < an, show that
an > (a1 + a2 + ... + an-1)/(n - 1).

96
19. Find integers a, b, and c such that 0 < a < b < c, a + b > c, and c is as small as possible.

20. Let m and n be positive integers, and let 1, m, and n be the lengths of the sides of a
triangle. Show that m = n.

21. Given that x > 0 and y > 0, show that (x + y)n > xn-1(x + ny) for all integers n $ 2.

22. Given that 1 % x $ 0, prove by mathematical induction that (1 % x)n $ 1 % nx for all
positive integers n.

1
23. Prove that 2 x < for all positive real numbers x.
x%1& x

2
24. Prove that n n < 1 % 2 % 3 % þ % n for all positive integers n.
3
(4n % 3) n
25. Prove that 1 % 2 % þ % n < for all positive integers n.
6
26. Use the fact that 1<b and x<y imply bx<by to prove the inequalities 2 # an < an%1 < 2
for the sequence a1, a2, ... defined by

a1 an
a1 ' 2, a2 ' 2 , ÿ, an%1 ' 2 , ÿ.

27. Use the fact that 0 < b < 1 and x < y imply bx > by to prove the inequalities

u1 < u2, u2 > u3, ÿ, u2k&1 < u2k, u2k > u2k%1,ÿ

for the sequence u1, u2, ... defined by

u1 un
1 1 1
u1 ' , u2 ' , ÿ, un%1 ' , ÿ.
2 2 2

10.2 FURTHER INEQUALITIES

In this section we develop a technique for investigating the range of values assumed by a
quadratic function. In subsequent work we shall assume as known the results of the examples in
Section 10.1 and of Problems 1 to 8 in Section 10.1.

Example: Let f(x) = ax2 + bx + c, where a, b, and c are real numbers and a … 0. Let D be
2
the discriminant b - 4ac. Show that if D > 0, then f(x) takes on both positive and negative

97
values.

Solution: Completing squares, we obtain

4a 2x 2 % 4abx % 4ac
f(x) ' ax 2 % bx % c '
4a
2 2 2 2
4a x % 4abx % b & (b & 4ac)
'
4a
2
(2ax % b) & D
' .
4a

If x = -b/2a, 2ax + b = 0, and so f(-b/2a) = -D/4a. We first consider the case in which a > 0. This
and D > 0 imply that f(x) = -D/4a < 0. We wish to show that f(x) also takes on positive values.
We consider values of x greater that (&b% D)/2a. Then

&b % D
x >
2a
2ax > &b % D
2ax % b > D > 0
(2ax % b)2 > D
(2ax % b)2 & D > 0
(2ax % b)2 & D
f(x) ' > 0.
4a

Thus we have proved the desired result for the case a > 0. If a < 0, let g(x) = -ax2 - bx - ac.
Since the coefficient of x2 in g(x) is positive, g(x) takes on both positive and negative values by
the previous case. Then so does f(x) = -g(x).

Problems for Section 10.2

1. Let a and b be real numbers. Prove that a 2 % b 2 $ 0 and that a 2 % b 2 ' 0 if and only if
a = b = 0.

2. Let c1, c2, ..., cn be real numbers. Prove that c12 % c22 % þ % c n2 $ 0 and c12 % c22 % þ % c n2 ' 0 if
and only if each ci = 0.

3. Let f(x) = ax2 + bx + c, where a, b, and c are real numbers and a > 0. Let D be the
discriminant b2 - 4ac. Show the following:

98
 (a) If D < 0, then f(x) > 0 for all x, and f(x) = 0 has no real roots.
(b) If D = 0, then f(x) $ 0 for all x, and f(x) = 0 has one real root.
&b & D &b % D &b & D
(c) If D > 0, then f(x) < 0 for < x < , f(x) > 0 for x < or
2a 2a 2a
&b % D
x > , and f(x) = 0 has two real roots.
2a

4. Let f(x) = ax2 + bx + c, where a, b, and c are real numbers and a < 0. Let D = b2 - 4ac. Show
the following:

(a) If D < 0, then f(x) < 0 for all x, and f(x) = 0 has no real roots.

(b) If D = 0, then f(x) # 0 for all x, and f(x) = 0 has one real root.

&b & D &b % D &b & D

(c) If D > 0, then f(x) > 0 for > x > , f(x) < 0 for x > or
2a 2a 2a
&b % D
x < , and f(x) = 0 has two real roots.
2a
*
5. Let F1, F2, F3, ... be the sequence of Fibonacci numbers 1, 1, 2, 3, 5, ... and let Rn = Fn+1/Fn for
n = 1, 2 , 3, ... . Do the following:

1
(a) Show that Rn%1 ' 1 % .
Rn
(b) Prove that R2n-1 < R2n+1 < R2n and R2n+1 < R2n+2 < R2n for all positive integers n, that is, that
R1 < R3 < R5 < R7 < ... < R8 < R6 < R4 < R2.

10.3 INEQUALITIES AND MEANS

We recall that the arithmetic mean of a1, a2, ..., an is

a1 % a2 % þ % a n
n

and the geometric mean is

n
a1a2þan .

We restrict a1, ..., an to be positive in discussing geometric means, since otherwise the definition
might express the mean as an even root of a negative number.
We shall use An for the arithmetic mean of a1, a2, ..., an and Gn for the geometric mean.

99
THEOREM: Let a1, a2, ..., an be positive real numbers. Then

A n $ Gn

that is,

a1 % a2 % þ % an n
$ a1a2þa n .
n

Proof: We proceed by mathematical induction. When n = 1, it is clear that A1 = a1 and G1 = a1.

Hence A1 = G1, and the theorem holds for n = 1.
We next prove it for n = 2. Since a1 and a2 are positive, a1 and a2 exist in the real
number system by the roots property of Section 10.1, and so

2
a1 & a2 $ 0

by Problem 3(b) of section 10.1. It follows that

a1 & 2 a1 a2 % a2 $ 0
a1 % a2 $ 2 a1 a2
a1 % a2
$ a1a2 .
2

This is precisely the statement A2 $ G2 of the theorem for n = 2.

We now assume the inequality true for n = k, that is, we assume that Ak $ Gk , and with
this as a basis shall prove that Ak%1 $ Gk%1.
If a1 = a2 = ... = ak+1 then Ak+1 = a1,Gk+1 = a1, and so Ak%1 $ Gk%1. It remains to
investigate the case in which the ai are not all equal. Without loss of generality, we may assume
that the ai are numbered so that
0 < a1 # a2 # a3 # þ #ak%1.
The fact that the a’s are not all equal implies that a1 < ak+1. It now follows from Problem 18 of
Section 10.1 that ak+1 > Ak. Since
a % a2 % þ % a k
Ak ' 1
k
we have kAk = a1 + a2 + ... + ak, and hence

100
 a1 % a2 % þ % ak % ak%1
Ak%1 '
k%1
kAk % ak%1
'
k%1
(k % 1)Ak % (ak%1 & A k)
'
k%1
ak%1 & A k
'A k % .
k%1

Let (ak+1 - Ak)/(k + 1) = p. We have seen above that ak+1 > Ak; this implies that p > 0. Now Ak+1 =
Ak + p. We raise both sides of this equality to the (k + 1)th power, obtaining

(Ak%1)k%1 ' (Ak % p)k%1

k%1
' (Ak)k%1 % (k % 1)(Ak)kp % (Ak)k&1p 2 % þ % p k%1.
2
Since p > 0 and Ak > 0, all the terms in the binomial expansion on the right side are positive.
There are k + 2 terms in this expansion; hence there are at least 4 terms. Now (Ak+1)k+1 is greater
than the sum of the first two terms:

(Ak+1)k+1 > (Ak)k+1 + (k + 1)p(Ak)k.

Since (k + 1)p = ak+1 - Ak, this becomes

(Ak%1)k%1 > (Ak)k%1 % ak%1(Ak)k & (Ak)k%1

(Ak%1)k%1 > ak%1(Ak)k.

Having assumed above that Ak $ Gk, we now have

(Ak%1)k%1>ak%1(Ak)k$ak%1(G k)k'ak%1(a1a2þak)
(Ak%1)k%1>(Gk%1)k%1.

From Problems 7 and 8, Section 10.1, it follows that Ak+1 > Gk+1, and the theorem is proved.
We have actually proved more than is stated in the theorem; we have shown that An > Gn
unless all the ai are equal.
There is a third type of mean that is used quite often: the harmonic mean. The harmonic
mean of numbers a1, a2, ..., an, none of which is zero, is given by

n
Hn ' .
1 1 1
% %þ%
a1 a2 an

101
In a harmonic progression, defined as a progression of the form

1 1 1 1
, , , ÿ,
a a % d a % 2d a % (n & 1)d

each term, except the end ones, is the harmonic mean of its two adjacent terms; that is, if a1, a2,
..., an are in harmonic progression, then

2
' a k.
1 1
%
ak&1 ak%1

The proof is left to the reader.

10.4 THE CAUCHY-SCHWARZ INEQUALITY

Another famous inequality is given various names in different texts, although in the United
States it is usually referred to as the Cauchy - Schwarz Inequality (named for Augustin Cauchy,
1789-1857; and Hermann Amandus Schwarz, 1843-1921). Some call it the Schwarz Inequality,
while others, including the Russians, call it the Cauchy-Buniakowski Inequality.

THEOREM: Let a1, a2, ..., an and b1, b2, ..., bn be any real numbers. Then

2 2 2 2 2 2
(a1 % a2 % þ % an )(b1 % b2 % þ % bn ) $ (a1b1 % a2b2 % þ % anbn)2,

that is,

' ' ' aibi .

2
n n n
2 2
ai bi $
i'1 i'1 i'1

Proof: We define a polynomial function f(x) by

f(x) = (a1x + b1)2 + (a2x + b2)2 + ... + (anx + bn)2.

Clearly f(x) is positive or zero for all real numbers x, since it is a sum of squares. Now

2 2 2 2 2 2
f(x) ' (a1 x 2 % 2a1b1x % b1 ) % (a2 x 2 % 2a2b2x % b2 ) % þ % (a n x 2 % 2a nbnx % bn )
2 2 2 2 2 2
' (a1 % a2 % þ % a n )x 2 % 2(a1b1 % a2b2 % þ % a nbn)x % (b1 % b2 % þ % bn ).

2 2 2 2 2 2
Let a1 % a2 % þ % an ' A, a1b1 % a2b2 % þ % anbn ' B, and b1 % b2 % þ % b n ' C so that

102
f(x) = Ax2 +2Bx + C. Since f(x) $ 0 for all x, the discriminant D # 0, since D > 0 implies that
f(x) is sometimes positive and sometimes negative. (See Section 10.2.) Now

D ' (2B)2 & 4AC ' 4B 2 & 4AC # 0.

Hence B 2 & AC # 0, and so AC $ B 2. Translating this back into our original notation, we
have the Cauchy-Schwarz Inequality:

2 2 2 2 2 2
(a1 % a2 % þ % an )(b1 % b2 % þ % bn ) $ (a1b1 % a2b2 % þ % anbn)2.

Examination of the above proof shows that

j j j aibi
n n n
2 2
ai bi '
i'1 i'1 i'1

if and only if there is a fixed number x such that aix + bi = 0 for all i, that is, the ai and bi are
proportional.
The hypothesis for the inequality on the arithmetic and geometric means is that the
numbers are all positive. The numbers in the Chauchy-Schwarz Inequality need not be positive.
In fact,

j ai j bi
n n
2 2

i'1 i'1

is unaltered by changes in the signs of the ai and bi, while

j aibi
n

i'1

is largest when all the signs are positive.

Problems for Sections 10.3 and 10.4

1. Given that a, b, c, d, x, y, z, and w are positive real numbers, prove the following
[from (a) to (z)]:

(a) If x + y = 2, then xy # 1.
(b) If xyz = 1, then x % y % z $ 3.
1 1 1
(c) If xyz = 1, then % % $ 3.
x y z

103
 1 1 1
(d) If x + y + z = 1, then % % $ 9.
x y z
a%b%c 3
(e) $ abc.
3
(f) (a % b % c % d)4 $ 256abcd.
(g) (x % y)(x & y)2 $ 0.
(h) x 3 % y 3 $ x 2y % xy 2.
(i) x 4 % y 4 $ x 3y % xy 3 $ 2x 2y 2.
(j) x 5 % y 5 $ x 4y % xy 4 $ x 3y 2 % x 2y 3.
1
(k) x % $ 2.
x
x y z
(l) % % $ 3.
y z x
x x y y z z
(m) % % % % % $ 6.
y z z x x y
(n) xy(x % y) % yz(y % z) % zx(z % x) $ 6xyz.
x y z w
(o) % % % $ 4.
y z w x
(p) a % b % c $ bc % ca % ab.
(q) 3(a % b % c % d) $ 2 ab % ac % ad % bc % bd % cd .
(r) (x % y)(y % z)(z % x) $ 8xyz.
(s) [(x % y)(x % z)(x % w)(y % z)(y % w)(z % w)]2 $ 4096(xyzw)3.
(t) If x + y + z = 1, then (1 & z)(1 & x)(1 & y) $ 8xyz.
1 1 1
(u) If x + y + z = 1, then &1 &1 & 1 $ 8.
x y z
(v) (y % z % w)(x % z % w)(x % y % w)(x % y % z) $ 81xyzw.
(w) If x + y + z + w = 1, then (1 & x)(1 & y)(1 & z)(1 & w) $ 81xyzw.
(x) (ab % xy)(ax % by) $ 4abxy.
(y) [(ab % cd)(ac % bd)(ad % bc)]2 $64(abcd)3.
1
(z) If x + y = 1, then x 2 % y 2 $ .
2
2. Given that a, b, and c are positive real numbers, show that

(a 2b % b 2c % c 2a)(a 2c % b 2a % c 2b) $ 9a 2b 2c 2.

Is this true for all real numbers a, b, and c?

3. Show that if a1, a2, ..., an are positive real numbers, then

104
 a1 a2 a3 an&1 an
% % %þ% % $ n.
a2 a3 a4 an a1

4. Let a1, a2, ..., an-1, an be positive, and let ai, aj, ..., ah, ak be a permutation of these n numbers.
a a a a
Show that 1 % 2 % þ % n&1 % n $ n.
ai aj ah ak

5. Let a, b, x and y be real numbers, with a2 + b2 = 1 and x2 + y2 = 1. Show that:

(a) (ax % by)2 # 1.

(b) (ax & by)2 # 1.
(c) &1 # ax % by #1.
(d) &1 # ax & by #1.

6. Let a, b, c, x, y, and z be real numbers with a2 + b2 + c2 = 1 = x2 + y2 + z2. Show that:

(a) (ax % by % cz)2 # 1.

(b) &1 # ax % by % cz # 1.

7. Let a, b, c, d, and e be real numbers. Show the following:

(a) a 2 % b 2 $ 2ab.
(b) a 2 % b 2 % c 2 $ bc % ac % ab.
(c) 3(a 2 % b 2 % c 2 % d 2) $ 2(ab % ac % ad % bc % bd % cd).
(d) 2(a 2 % b 2 % c 2 % d 2 % e 2) $ a(b % c % d % e) % b(c % d %e) % c(d % e) % de.

8. Show that a 2 % b 2 % 1 $ b % a % ab for all real numbers a and b.

9. Show that if x and y are positive real numbers with x + y = 1, then

2 2
1 1 25
x% % y% $ .
x y 2

10. Show that if x, y and z are positive real numbers with x + y + z = 1, then

2 2 2
1 1 1 100
x% % y% % z% $ .
x y z 3

11. Let a, b, and c be positive real numbers. Show that 3(bc % ca % ab) # a % b % c.

105
12. Let a, b, c, and d be positive. Show that

2 6(ab % ac % ad % bc % bd % cd) # 3(a % b % c % d).

13. Let An, Gn, and Hn be the arithmetic, geometric, and harmonic mean, respectively, of
positive numbers a1, a2, ..., an. Assuming An $ Gn, show that An $ Gn $ H n.

14. Show that n n $ 1@3@5þ(2n & 1).

15. Show that (1k % 2k % þ % n k)n $ n n(n!)k for all positive integers n and k.

16. Show the following:

(a) (n % 1)n $ 2@4@6þ(2n).

2n
n%1
(b) n n $ (n!)3 for all positive integers n.
2

17. Show the following:

n%2
(a) n@1 % (n & 1)@2 % (n & 2)@3 % þ % 2(n & 1) % 1@n ' .
3
n/2
(n % 1)(n % 2)
(b) $ n!.
6

18. Show the following:

(a) 1 + 2 + 22 + ... + 2n-1 = 2n - 1.

n&1
(b) 2n $1 % n 2 for all positive integers n.

19. Do the following:

n
1 1 1 nn
(a) Show that 1 % % %þ% $ .
2 3 n n!

1
(b) Show that n % 1 & n > for all positive integers n.
2 n%1
1 1 1
(c) Show by mathematical induction that 2 n > 1 % % %þ% .
2 3 n
106
 n
n
(d) Show that n! > .
4

n&1
n
n 2n & 2
20 Show that A # for n $ 2.
k'0 k n&1

21. Do the following:

(a) Find the arithmetic mean of a1, a2, ..., a100, given that a1 = 1 and
a2 = a3 = ... = a100 = 100/99.

99/100
100 101
(b) Prove that < .
99 100

99 100
1 1
(c) Prove that 1% < 1% .
99 100

n n%1
1 1
(d) Prove that 1% < 1% for all positive integers n.
n n%1

22. Do the following:

(a) Find the arithmetic mean of a0, a1, a2, ..., a100, given that a0 = 1 and
a1 = a2 = ... = a100 = 99/100.

(b) Prove that 100201 > 99100@101101.

n%1 n%2
1 1
(c) Prove that 1% > 1% for all positive integers n.
n n%1

23. Do the following:

(a) Find the arithmetic and geometric means of the roots of x4 - 8x3 + 18x2 - 11x + 2 = 0,
given that all the roots are positive.

(b) Given that all the roots of x6 - 6x5 + ax4 + bx3 + cx2 + dx + 1 = 0 are positive, find a, b, c,
and d.

(c) Find all the roots of x11 - 11x10 + ... - 1 = 0, given that each root is positive.

107
24. Given tht a, b, and c are the lengths of the sides of a triangle, show that

3(bc % ac % ab) # (a % b % c)2 < 4(bc % ac % ab).

25. For all real numbers a, b, c, x, y, and z show that

a2 % b2 % c2 % x2 % y2 % z2 $ (a % x)2 % (b % y)2 % (c % z)2.

26. For all real numbers a1, a2, ..., an, b1, b2, ..., bn, show that

2 2 2 2 2 2
a1 % a2 % þ % an % b1 % b2 % þ % bn $ (a1 % b1)2 % (a2 % b2)2 % þ % (a n % bn)2.

27. Show that if a and b are positive real numbers and m and n are positive integers, then

m mn n a mb n
$ .
(m % n)m%n (a % b)m%n

28. Let Fn and Ln be the nth Fibonacci and nth Lucas number, respectively. Prove that

n
F4n
> L2L6L10ÿL4n&2
n

for all integers n $ 2.

108
 ANSWERS TO THE
ODD-NUMBERED PROBLEMS

Chapter1, page 5

1. 10.

3. 1.

5. 1 2 1
1 2 1
1 3 3 1.

7. 1, 1 + 5 = 6, 5 + 10 = 15, 10 + 10 = 20, 10 + 5 = 15, 5 + 1 = 6, and 1;

1, 1 + 6 = 7, 6 + 15 = 21, 15 + 20 = 35, 20 + 15 = 35, 15 + 6 = 21, 6 + 1 = 7, and 1.
9 9 9 9
9. ' ' 36, ' ' 9.
7 2 8 1
11. 125x 3 + 150x 2y + 60xy2 + 8y3.

17. c = 56, m = 5.

19. (a) 4. (b) 8. (c) 16. (d) 32. (e) 64. (f) 128.

21. (a) 0. (b) 0.

23. (a) 8. (b) 16. (c) 32. (d) 64.

25. r = 6, s = 7, t = 6, u = 7.
996
27. .
990
31. x 4 % 4x 3 y % 6x 2 y 2 % 4xy 3 % y 4 % 4x 3 z % 12x 2 yz % 12xy 2 z %
4y 3 z % 6x 2 z 2 % 12xyz 2 % 6y 2 z 2 % 4xz 3 % 4yz 3 % z 4 .

33. 3.

37. 3.

39. 36, of which 27 are odd.

109
41. 0, 1, 3, 7, and 15.

Chapter 2, page 14

3. 103.

7. Fn+4 = 3Fn+1 + 2Fn.

9. 321.

13. 702.

19. (a) F16 - 1. (b) F19 - 1.

21. (a) 2. (b) 3. (c) 5. (d) 8. (e) 13. (f) 21.

23. r = 4, s = 5, and t = 11.

25. r = 15, s = 17, and t = 19.

29. 204

33. L2m+2 - 2.

35. r = 4, s = 6, and t = 8.

37. (a) 1. (b) -1. (c) 1. (d) -1.

Chapter 3, page 18

1. (a) 5040. (b) 36. (c) 362,880. (d) 720.

5. (a) 5!. (b) 7!. (c) (n + 1)!.

7. a = 14 and b = 10 or a = 24024 and b = 24023.

9. n4 + 10n3 + 35n2 + 50n + 24.

11. (a) 23. (b) 119. (c) 719.

15. a = 6, b = 11, and c = 5.

110
17. (2m + 2)! - 1.

19. (a) (n + 2)! + (n + 1)! - 2.

Chapter 4, page 23

1. (a) -4, 3, 10. (b) -19, -23, -27.

3. (a) 990. (b) 993. (c) 988.

5. 33/2.

7. (a) -33,698. (b) 501,994. (c) -11,385/4. (d) n2.

9. 54, 1458, 13122.

11. 686, 4802, 33614.

13. ± 42.

15. (a) (71000 -1)/6. (b) (1 - 71000)/8. (c) (7n - 1)/6.

17. 9.

21. (a) 6 3. (b) 6 3. (c) 4. (d) 4.

23. 30.

27. (x n+1 - yn+1)/(x - y).

29. (a) 5. (b) 24 miles per hour.

31. 2[1 - (1/2)7].

35. (a) 0, 3, 5, 6, 9, 10, 12, 15, 18, 20, 21, 24, 25, 27, 30.
(b) 15. (c) 15. (d) 16 • 15 = 240.

Chapter 5, page 35

1 2 n%1 2 n%1
13. (a) 1 & . (b) x & y .
n% 1 x&y

111
15. (n + 1)/2n for n > 1.

21. Using mathematical induction, one can show that:

2
(a) Fn%1 & F n Fn%2 ' (&1)n .

2 2
Fn%2 & Fn%1
(b) ' Fn% 3 .
Fn

(c) Fn-1 + Fn+1 = Ln.

Chapter 6, page 47

1. (a) 4,845. (b) 3,003. (c) -2,912.

3. a = -3, b = 2, c = 0.

7. 1, -1, 1, -1, 1, -1.

&m r % m& 1
11. ' (&1)r .
r m &1

15. (d/2)n2 + [a - (d/2)]n.

j ai j b i ' a1 b1 % a1 b2 % a2 b1 % a2 b2 and j (a ibi) ' a1 b1 % a2 b2 . These are not

2 2 2
19.
i'1 i'1 i'1
always equal, since, for example, they are unequal for a1 = a2 = b1 = b2 = 1.

23. (1/6)n3 - (1/2)n2 + (1/3)n.

25. n3 + 5n.

27. s = 3, t = 1.

29. r = 6, s = 7, t = 1.

31. (1/5)n5 + (1/2)n4 + (1/3)n3 - (1/30)n.

112
37. (n3 + 3n2 + 2n)/6.

Chapter 7, page 55

1. a, b; a, c; a, d; a, e; b, c; b, d; b, e; c, d; c, e; d, e.

3. a, b; b, a; a, c; c, a; a, d; d, a; a, e; e, a; b, c; c, b; b, d; d, b; b, e; e, b; c, d; d, c; c, e; e, c;
d, e; e, d.

5. 7,920.

7. 30.

12 9 12 9 12 5 12 5 12 10 12 10
9. , , , , , .
3 7 3 2 7 3 7 2 2 7 2 3

11. -1,201,200.

17. 3100.

19. 1,2,3,4; 2,1,4,3; 3,1,2,4; 4,1,3,2;

1,3,4,2; 2,4,3,1; 3,2,4,1; 4,3,2,1;
1,4,2,3; 2,3,1,4; 3,4,1,2; 4,2,1,3.

23. (a) 200. (b) 10,150.

Chapter 8

Sections 8.1, 8.2 and 8.3, page 64

1. (x + 2)(x 3 + 3x 2 - 6x + 2) - 16.

3. &1 ± 3.

5. a = -27,604.

7. &1, 3, 2 i, & 2 i.

113
 9. f(0) = 0 = f(a). Two factors are x and x - a.

11. 0, -2, -4, -2/3.

13. 5 , &3 % 17 , &3 & 17 .

3 2 2

1 2 1 % 3 i 1 & 3i
15. 0, & , , , .
3 3 6 6

17. ±1, ±2, ±3, ±5, ±6, ±10, ±15, ±30.

25. x 8 - 48x 6 + 536x 4 - 1728x 2 + 400 = 0.

27. (a) b.
(b) c(2x + 1) + b.
(c) d(3x 2 + 3x + 1) + c(2x + 1) + b.
n n&1 n n& 2 n n&3 n n
(d) x % x % x %þ% x% .
1 2 3 n& 1 n

29. (a) 0. (b) 0. (c) 6d.

33. a = 38, b = 84, c = 81, d = 34, e = 5.

39. No.

Section 8.4, page 68

4 3
1. (a) 4. (b) 8/3. (c) 16. (d) 32/3. (e) 16/3. (f) .
3

5. 1.

7. a = p3 - 2pq, b = p4q - 3p2q2 + 2q3, c = p3q3 - 2pq4, d = q6.

9. (a) 0. (b) -3. (c) 1. (d) y3 -3y - 1 = 0.

11. (a) x = 3, y = -3, z = 1.

(b) t = 4, u = 2, v = -4, w = 1.

114
Chapter 9

Sections 9.1 and 9.2, page 78

1. x = -6/53, y = -91/53.

3. x = 103/21, y = -71/21, z = -4/63.

5. x = 5, y = 2, z = 4.

7. x = (b + c - a)/2, y = (c + a - b)/2, z = (a + b - c)/2.

Sections 9.3, page 85

15. 4xyz.

17. (b) x = -1, y = 2, z = 0.

(c) x = 2, y = -3, z = 3, w = 0.

/0 n%2 n%1 /0 ' &/0 n%1 /0 ,

(d) L L L Ln
00 L 0 0
00 Ln Ln&1 000
00 n%1 L n 000 0 0
2 2
3Ln & 2L n Ln%1 % 2Ln%1
L2n ' ,
5

2Ln3 & 3L n2 Ln%1 % 3Ln L n%1

2

L3n ' .
5

Section 9.4, page 90

5. (w - x)(w - y)(w - z)(z - x)(z - y)(y - x).

7. (w - x)(w - y)(w - z)(z - x)(z - y)(y - x).

9. r = 2, s = 2, t = 1.

11. r = 4, s = 2, t = 2.

13. (b) x =1, y = 1, z = 1.

115
Chapter 10

Section 10.1, page 95

19. a = 2, b = 3, c = 4.

Sections 10.3 and 10.4, page 103

21. (a) 101/100.

4
23. (a) The arithmetic mean is 2; the geometric mean is 2.
(b) a = 15, b = -20, c = 15, d = -6.
(c) The roots are equal, and so all equal to 1, since their arithmetic mean is the same as
their geometric mean.

116
 INDEX
A Combinations, 51-57
Complex coefficients, 58, 68
Absolute value of a real number, 68, 89, 94 Complex numbers, 1, 58, 59, 68
Advice to the problem solver, vii Consecutive integers, 33, 38
Analogue of a formula, 16-17, 37, 38, 50 Consecutive terms of a progression, 22-25
Angles of a polygon, 35 Constructible lengths, 1
Arithmetic mean, 22, 25-28, 99-101, 103, Convex polygons, 29, 35
106, 107 Cramer's Rule, 73, 76, 78, 84-85
Arithmetic progression, 21-26, 34 Crout's Method, 85
Arrangements, 53-55, 57 Cube roots, 1, 23
Average (see Arithmetic mean)
Axiomatic approach, 29 D

Deduction, 29
B Degree of a polynomial, 58, 59, 64, 67
Determinant, 71-93
Binomial, 3 Determinant of the system, 73, 75, 84
Binomial coefficients, 2-11, 15-17, 25, 34, Determinants of large order, 81-93
37, 40-45, 47-53, 56, 106 Diagonal:
Binomial expansion (see Binomial of the Pascal Triangle, 3-4, 40
coefficients) of a polygon, 29-31
Binomial Theorem, 41-43 of a square, 1
Difference:
C of an arithmetic progression, 21-23
of a function, 66-67
Cauchy, Augustin (1789-1857), 102 Discriminant of a quadratic, 97-99
Cauchy-Buniakowski Inequality, 102-108 Division, synthetic, 60-63, 67
Cauchy-Schwarz Inequality, 102-108 Divisor, integral, 63-65
Closure, 94 Double subscript notation, 76, 81
Coefficients:
binomial, 2-11, 15-17, 25, 34, 37,
40-45, 47-53, 56, 106 E
complex, 58, 68
integral, 63-66 Elimination technique, 71-73, 75, 85
of linear equations, 71-73 Equations:
multinomial, 53-56 linear, 71-92
of a polynomial, 58, 61, 63-66, 68-70 polynomial, 38, 58-70, 98-99, 107
trinomial, 53-56 Even permutation, 54-57, 76, 77, 82, 83, 84
Cofactor, 77, 79, 82, 85-86

117
 F Induction, 29-39, 41-43, 49, 66, 95, 97, 100,
106
Factor Theorem, 59, 89 Inequalities, 94-108
Factorials, 18-20, 34, 41-43, 46, 48, 51-54, Initial value in a summation, 46
82 Integers, 1, 12, 13, 21, 27, 29-39, 41-49, 56, 57,
Factorizations, 26, 32-33, 36, 59, 65-66, 68, 63-66, 95-97, 106, 107
89-93 Integral coefficients, 63-66
Fibonacci (c. 1170-1250), 12 Integral divisor, 63-65
Fibonacci numbers, 12-17, 27, 34, 37-38, 50, Integral multiple, 27, 33, 36, 39, 49
88, 99, 108 Integral roots, 63-65
Fibonacci Quarterly, The, 14 Irrational roots, 64, 66
Fibonacci sequence (see Fibonacci numbers)
Final value in a summation, 46
Finite sequence, 21, 22 L
First-degree equation, 71
Fixed multiplier, 22 Leonardo of Pisa (c. 1170-1250), 12
Fractions, 1 Liber Abacci, 12
Function, 58, 66-70, 97-99, 102 Linear equations, 71-92
Fundamental Theorem of Algebra, 68 Lucas, Francois Edouard Anatole (1842-1891),
14
Lucas numbers, 14-17, 34-35, 37, 38, 50, 88,
G 108
Lucas sequence(see Lucas numbers)
General polynomial, 46, 58
Geometric mean, 23, 25, 26, 99-101, 103,
106, 107 M
Geometric progression, 22-27, 34
Mathematical induction, (See Induction)
Mean:
H arithmetic, 22, 25-28, 99-101, 103, 106,
107
Harmonic mean, 101, 106 geometric, 23, 25, 26, 99-101, 103, 106,
Harmonic progression, 102 107
harmonic, 101, 106
Minor of a determinant, 77, 79, 82, 85-86, 89
I Multinomial coefficients, 53-56
Multiple, integral, 27, 33, 36, 39, 49
Identity, 44 Multiple roots, 67-68
Index of a summation, 45 Multiplier, fixed, 22

118
 N Progression:
arithmetic, 21-26, 34
Natural numbers, 1 geometric, 22-27, 34
nth roots, 1, 23, 94, 100 harmonic, 102
Number systems, 1 Proper subset, 55
Numbers:
complex, 1, 58, 59, 68
Fibonacci, (See Fibonacci numbers) Q
Lucas, (See Lucas numbers)
natural, 1 Quadratic equation, 38, 98-99
negative real, 1, 97-98, 99, 103 Quadratic function, 97-99
rational, 1, 64-65 Quotient, 22, 60, 63
real, 1, 23, 58, 94-108
triangular, 22, 31
Numerical approximation of a determinant, R
85
R (before problem numbers), vii
O Ratio of a geometric progression, 22-26
Rational numbers, 1, 64-65
Odd permutation, 54-57, 76, 77, 82, 83 Rational roots, 63-65
Order of a determinant, 72, 75, 81, 85 Real numbers, 1, 23, 58, 94-108
Remainder in a division, 60-63
Remainder Theorem, 59
P Roots:
cube, 23
Pascal, Blaise (1623-1662), 3, 29 integral, 63-65
Pascal Triangle (see Binomial coefficients) irrational, 64, 66
Pell sequence, 38 multiple, 67-68
Permutations, 53-57, 76, 77, 82-84, 105 nth, 1, 23, 94, 100
Pi notation, 46, 48, 49, 107 of a polynomial equation, 38, 59-69,
Polygons, 29-31, 35 98-99
Polynomial equations, 38, 58-70, 98-99, 107 rational, 63-65
Polynomials, 19, 46, 48, 49, 58-70, 86, 90, square, 23, 97, 98, 99, 106, 108
97-99, 102 Row number of a binomial coefficient, 3-4
Product notation, 46, 48, 49, 107

119
 S Symmetric function, 68-70
Symmetry of the Pascal Triangle, 43, 52
Schwarz, Hermann Amandus (1843-1921), Synthetic division, 60-63, 67
102 System of linear equations (see simultaneous
Schwarz Inequality, 102-108 linear equations)
Sequence: Systems, number, 1
arithmetic (progression), 21-26, 34
factorial, 18
Fibonacci, (See Fibonacci numbers) T
geometric (progression), 22-27, 34
Lucas, (See Lucas numbers) Theorems, deduction of, 29
miscellaneous, 28, 38, 50 Transpose of a determinant, 83, 85
Pell, 38 Transpose Theorem, 83, 85
Set, 27, 31, 55 Triangle, sides of a, 97, 108
Sigma notation, 45-50, 67, 102-103 Triangular numbers, 22, 31
Simultaneous linear equations, 71-92 Trichotomy, 94
Square roots, 23, 97, 98, 99, 106, 108 Trinomial coefficients, 53-56
Strong mathematical induction, 34-35
Subsets, 51, 55
Sum: V
of an arithmetic progression, 21, 24
of a geometric progression, 22-26 Vandermonde, Alexandre-Theophile
of squares of binomial coefficients, 10 (1735-1796), 89
of squares of real numbers, 102-103 Vandermonde determinants, 89-93
Summation notation, 45-50, 67, 102-103

120