digital text front

45
FIBONACCI AND LUCAS NUMBERS 2015 ALFIYA A

Upload: alfiya1710

Post on 14-Apr-2017

286 views

Category:

Education


1 download

TRANSCRIPT

[Type the company name]

10/25/2015

FIBONACCI AND LUCAS NUMBERS

2015

ALFIYA A

FIBONACCI AND LUCAS NUMBERS

By

ALFIYA A

(MSc Mathematics)

Preface

The Fibonacci sequence and the Lucas sequence are the two shining

stars in the vast array of integer sequences. This book is intended for

college undergraduate and graduate students often opt to study

Fibonacci and Lucas numbers because they find them challenging and

exciting. The book contains details about Fibonacci and Lucas

identities, its divisibility properties and also its generalization.

Suggestion for the further improvement of this book will

be highly appreciated.

ACKNOWLEDGEMENT

I would like to express my gratitude to many people

who gave me their comments and suggestions which influenced

the preparation of the text. I want to thank teachers and friends

for their effective cooperation and great care for preparing this

book.

CONTENTS

CHAPTER TITLE PAGE NO.

CHAPTER โ€“ 1 FIBONACCI AND LUCAS NUMBERS 1

CHAPTER โ€“ 2 FIBONACCI AND LUCAS IDENTITIES 7

CHAPTER โ€“ 3 DIVISIBILITY PROPERTIES 25

CHAPTER โ€“ 4 GENERALIZATIONS 37

REFERENCES 44

CHAPTER โ€“ 1

FIBONACCI AND LUCAS NUMBERS

In the realm of mathematics, many concepts have applications

in multiple mathematical fields. Without these important concepts, every field of

mathematics would be seemingly disjointed and unrelated to topics from other

mathematical fields. One of these concepts was discovered by a man named

Leonardo Pisano during the early 13th century. This particular concept is known

today as the Fibonacci sequence. Since its official introduction to the world by

Leonardo Pisano, the Fibonacci sequence has become one of the most fascinating

concepts in the entire realm of mathematics through its remarkable characteristics;

its useful applications to various mathematical fields such as number theory,

discrete mathematics, and geometry; and its clear demonstration of the aesthetic

nature of God.

The Fibonacci series was derived from the solution to a problem

about rabbits. The problem is:

Suppose there are two new born rabbits, one male and the other female. Find

the number of rabbits produced in a year if

โ€ข Each pair takes one month to become mature:

โ€ข Each pair produces a mixed pair every month, from the second month:

โ€ข All rabbits are immortal

Suppose, that the original pair of rabbits was born on January 1. They take a

month to become mature, so there is still only one pair on February 1. On March 1,

they are two months old and produce a new mixed pair, so total is two pair. So

continuing like this, there will be 3 pairs in April, 5 pairs in May and so on.

It is then seen that at the end of the ๐‘›th month, all of the pairs during the

๐‘› โˆ’ 2 th month has given birth to a new pair of rabbits. So then, the total number

of pairs at the end of the ๐‘›th month is the number from the ๐‘› โˆ’ 1 th month plus a

new pair for each pair of rabbits at the end of the ๐‘› โˆ’ 2 th month. In some

simpler terms, the number of pairs of rabbits at the end of ๐‘› months is the sum of

the numbers at the end of the two previous months.

Without using rabbits, it is said that in the Fibonacci sequence of numbers,

each number is the sum of the previous two numbers, where the first two numbers

of the sequence are 0 and 1. The numbers generated by the Fibonacci sequence are

one of the most famous examples of a recurrence relation. A recurrence relation is

a relation which uses previous values in the relation and is usually defined by some

fixed numbers in the relation, known as the initial conditions. Writing the

Fibonacci sequence as a function, it is shown that ๐น0 = 0, ๐น1=1, ๐น๐‘›+2 = ๐น๐‘›+1 +

๐น๐‘› . The recurrence is ๐น๐‘›+2 = ๐น๐‘›+1 + ๐น๐‘› and the initial conditions are ๐น0 = 0,

๐น1=1.

Then the Fibonacci numbers are 0,1, 1, 2, 3, 5, 8,13,...

This yield the following recursive definition of the ๐‘›th Fibonacci number ๐น๐‘› .

๐น1 = 1

๐น2 = 1

โ‹ฎ

๐น๐‘› = ๐น๐‘›โˆ’1 + ๐น๐‘›โˆ’2, ๐‘› โ‰ฅ 3

Using the formula given above where ๐น๐‘›โˆ’1 and ๐น๐‘›โˆ’2 must be calculated

before adding them together to find ๐น๐‘› is the most obvious method for calculating

a Fibonacci number. This method requires calculating all of the Fibonacci numbers

with indexes less than ๐‘› before the value of ๐น๐‘› can be calculated. There is a fairly

simple formula named after the French mathematician Jacques Phillipe Marie

Binet to find the value of a given Fibonacci number without performing all of the

tedious calculations. If ๐น๐‘› is the ๐‘›th Fibonacci number , then

๐น๐‘› =1

5

1 + 5

2

๐‘›

โˆ’ 1 โˆ’ 5

2

๐‘›

. The โ€œspanโ€ or โ€œdegreeโ€ of the recursion is the difference between the

highest and lowest subscripts in the recursion ( ๐‘› + 1 โˆ’ ๐‘› โˆ’ 1 = 2 in the

Fibonacci recursion). Lucas discovered that if the degree is d, there are always d

and at most d โ€œindependentโ€ sequences satisfying the same recursion (with

different starting values).

In the case where the degree is 2, like the Fibonacci numbers, the recursion can

be written as:

๐‘ˆ(๐‘› + 1) = ๐ด โˆ— ๐‘ˆ(๐‘›) + ๐ต โˆ— ๐‘ˆ(๐‘› โˆ’ 1)

There are two independent sequences (that is, one is not a multiple of the other)

satisfying this. We can always pick one such that U(1) = 1 and U(2) = 1, and also a

second one such that U(1) = 1 and U(2) = 3 . The second sequence which is

independent of the Fibonacci sequence and starts 1, 3,... is now called the Lucas

sequence after Edouard Lucas. Why choose 2, 1, ... for the start? It is probably

because of the Binet formulas for the Fibonacci and Lucas numbers:

๐น๐‘› = ๐›ผ๐‘› โˆ’ ๐›ฝ๐‘› ๐›ผ โˆ’ ๐›ฝ

๐ฟ๐‘› = ๐›ผ๐‘› + ๐›ฝ๐‘›

where ๐‘› โ‰ฅ 0, ๐›ผ = (1 + 5)/2 , and ๐›ฝ = (1 โˆ’ 5)/2. Then ๐›ผ and ๐›ฝ are the two

solutions to ๐‘ฅ2 โˆ’ ๐‘ฅ โˆ’ 1 = 0. The fact that the coefficients of ๐›ผ๐‘› and ๐›ฝ๐‘› are equal

in magnitude and opposite in sign in the first formula, and equal in magnitude and

sign in the second, is probably why this is the simplest and easiest second,

independent sequence to take. The first few terms of the Lucas sequence look like

this: 2, 1, 3, 4, 7, 11, 18, 29, 47, 76, 123, ...

Lucas numbers ๐ฟ๐‘› are defined recursively as follows

๐ฟ1 = 1

๐ฟ2 = 3

โ‹ฎ

๐ฟ๐‘› = ๐ฟ๐‘›โˆ’1 + ๐ฟ๐‘›โˆ’2, ๐‘› โ‰ฅ 3

CHAPTER - 2

FIBONACCI AND LUCAS IDENTITIES

Both Fibonacci and Lucas numbers satisfy numerous identities that have been

discovered over the centuries. In this chapter we explore several of these

fundamental identities.

In doing so, we notice the following interesting pattern:

๐น1 = 1 = 2 โˆ’ 1 = ๐น3 โˆ’ 1

๐น1 + ๐น2 = 2 = 3 โˆ’ 1 = ๐น4 โˆ’ 1

๐น1 + ๐น2 + ๐น3 = 4 = 5 โˆ’ 1 = ๐น5 โˆ’ 1

๐น1 + ๐น2 + ๐น3 + ๐น4 = 7 = 8 โˆ’ 1 = ๐น6 โˆ’ 1

๐น1 + ๐น2 + ๐น3 + ๐น4 + ๐น5 = 12 = 13 โˆ’ 1 = ๐น7 โˆ’ 1

Following this pattern, we conjecture that

๐น๐‘–

๐‘›

1

= ๐น๐‘›+2 โˆ’ 1

We shall establish the validity of this formula in two ways, but we first state it as a

theorem.

Theorem 2.1 (Lucas, 1876)

๐น๐‘–

๐‘›

1

= ๐น๐‘›+2 โˆ’ 1 (2.1)

Proof:

Using the Fibonacci recurrence relation, we have:

๐น1 = ๐น3 โˆ’ ๐น2

๐น2 = ๐น4 โˆ’ ๐น3

๐น3 = ๐น5 โˆ’ ๐น4

โ‹ฎ

๐น๐‘›โˆ’1 = ๐น๐‘›+1 โˆ’ ๐น๐‘›

๐น๐‘› = ๐น๐‘›+2 โˆ’ ๐น๐‘›+1

Adding these equations, we get:

๐น๐‘– = ๐น๐‘›+2 โˆ’

๐‘›

1

๐น2 = ๐น๐‘›+2 โˆ’ 1

AN ALTERNATE METHOD

An alternate method of proving Identity (2.1) is to apply the Principle of

Mathematical Induction. Since ๐น1 = ๐น3 โˆ’1, the formula works for ๐‘› = 1.

Now assume it is true for an arbitrary positive integer ๐‘˜ โ‰ฅ 1 :

๐น๐‘– = ๐น๐‘˜+2 โˆ’ 1

๐‘˜

1

Then

๐น๐‘–

๐‘˜+1

1

= ๐น๐‘– +

๐‘˜

1

๐น๐‘˜+1

= (๐น๐‘˜+2 โˆ’ 1) + ๐น๐‘˜+1,by the inductive hypothesis

= (๐น๐‘˜+1 + ๐น๐‘˜+2) โˆ’ 1

= ๐น๐‘˜+3 โˆ’ 1

Thus, by Principle of Mathematical Induction, the formula is true for every

positive integer ๐‘›.

For example,

๐น๐‘– =

10

1

๐น12 โˆ’ 1 = 144 โˆ’ 1 = 143

Theorem 2.2 (Lucas, 1876)

๐น2๐‘–โˆ’1

๐‘›

1

= ๐น2๐‘› (2.2)

Proof:

Using the Fibonacci recurrence relation, we have

๐น1 = ๐น2 โˆ’ ๐น0

๐น3 = ๐น4 โˆ’ ๐น2

๐น5 = ๐น6 โˆ’ ๐น4

โซถ

๐น2๐‘›โˆ’3 = ๐น2๐‘›โˆ’2 โˆ’ ๐น2๐‘›โˆ’4

๐น2๐‘›โˆ’1 = ๐น2๐‘› โˆ’ ๐น2๐‘›โˆ’2

Adding these equations, we get

๐น2๐‘–โˆ’1 = ๐น2๐‘› โˆ’ ๐น0 =

๐‘›

1

๐น2๐‘› , since ๐น0 = 0

For example,

๐น2๐‘–โˆ’1 = ๐น16 =

8

1

987

Corollary 2.1 (Lucas, 1876)

๐น2๐‘– =

๐‘›

1

๐น2๐‘›+1 โˆ’ 1 (2.3)

Proof:

๐น2๐‘– =

๐‘›

1

๐น๐‘– โˆ’

2๐‘›

1

๐น2๐‘–โˆ’1

๐‘›

1

= (๐น2๐‘›+2 โˆ’1) โˆ’ ๐น2๐‘› , by theorems 2.1 and 2.2

= (๐น2๐‘›+2 โˆ’ ๐น2๐‘›) โˆ’ 1

= ๐น2๐‘›+1 โˆ’ 1 , by the Fibonacci recurrence relation

Theorem 2.3 (Cassiniโ€™s Formula)

๐น๐‘›โˆ’1๐น๐‘›+1 โˆ’ ๐น๐‘›2 = (โˆ’1)๐‘› , where ๐‘› โ‰ฅ 1 (2.4)

Proof:

Since ๐น0๐น2 โˆ’ ๐น12 = 0 ยท 1 โˆ’ 1 = โˆ’1 = (โˆ’1)1, the given statement is clearly

true when ๐‘› = 1.

Now we assume it is true for an arbitrary positive integer k: :

๐น๐‘˜โˆ’1๐น๐‘˜+1 โˆ’ ๐น๐‘˜2 = (โˆ’1)๐‘˜ . Then,

๐น๐‘˜๐น๐‘˜+2 โˆ’ ๐น๐‘˜+12 = (๐น๐‘˜+1 โˆ’ ๐น๐‘˜โˆ’1)(๐น๐‘˜ + ๐น๐‘˜+1) โˆ’ ๐น๐‘˜+1

2

= ๐น๐‘˜๐น๐‘˜+1 + ๐น๐‘˜+12 โˆ’ ๐น๐‘˜๐น๐‘˜โˆ’1 โˆ’ ๐น๐‘˜โˆ’1๐น๐‘˜+1 โˆ’ ๐น๐‘˜+1

2

= ๐น๐‘˜๐น๐‘˜+1 โˆ’ ๐น๐‘˜๐น๐‘˜โˆ’1 โˆ’๐น๐‘˜2 โˆ’ (โˆ’1)๐‘˜ by the inductive hypothesis

= ๐น๐‘˜๐น๐‘˜+1 โˆ’ ๐น๐‘˜ ๐น๐‘˜โˆ’1 + ๐น๐‘˜ + (โˆ’1)๐‘˜+1

= ๐น๐‘˜๐น๐‘˜+1 โˆ’ ๐น๐‘˜๐น๐‘˜+1 + (โˆ’1)๐‘˜+1

= (โˆ’1)๐‘˜+1

Thus the formula works for ๐‘› = ๐‘˜ + 1. So, by the Principle of Mathematical

Induction, the statement is true for every integer ๐‘› โ‰ฅ 1.

Corollary 2.2

Any two consecutive Fibonacci numbers are relatively prime; that is,

(๐น๐‘›+1,๐น๐‘› ) = 1 for every ๐‘›.

Proof:

Let p be a prime factor of both ๐น๐‘› and ๐น๐‘›+1. Then, by Cassini's formula,

p|ยฑ1 , which is a contradiction.

Thus (๐น๐‘›+1,๐น๐‘› ) = 1.

Theorem 2.4 (Lucas, 1876)

๐น๐‘–2 = ๐น๐‘›๐น๐‘›+1

๐‘›

1

Proof:

When ๐‘› = 1,

L. H. S = ๐น๐‘–2

1

1

= ๐น12 = 1 = 1 ยท 1 = ๐น1 ยท ๐น2 = R. H. S

So the result is true when ๐‘› = 1.

Assume it is true for an arbitrary positive integer ๐‘˜ :

๐น๐‘–2

๐‘˜

1

= ๐น๐‘˜ ๐น๐‘˜+1

Then,

๐น๐‘–2 =

๐‘˜+1

1

๐น๐‘–2 +

๐‘˜

1

๐น๐‘˜+12

= ๐น๐‘˜๐น๐‘˜+1 + ๐น๐‘˜+12 , by the inductive hypothesis

= ๐น๐‘˜+1 (๐น๐‘˜ + ๐น๐‘˜+1)

= ๐น๐‘˜+1๐น๐‘˜+2 , by the Fibonacci recurrence relation.

So the statement is true when ๐‘› = ๐‘˜ + 1.Thus it is true for every positive integer ๐‘›.

For example,

๐น๐‘–2 = ๐น15

15

1

๐น16 = 610 ยท 987 = 602070

Interestingly enough, identities 2.1 through 2.5 have analogous results for

Lucas number also.

Result 2.1

๐ฟ๐‘– =

๐‘›

1

๐ฟ๐‘›+2 โˆ’ 3 (2.6)

Proof:

Using the Lucas recurrence relation, we have,

๐ฟ1 = ๐ฟ3 โˆ’ ๐ฟ2

๐ฟ2 = ๐ฟ4 โˆ’ ๐ฟ3

๐ฟ3 = ๐ฟ5 โˆ’ ๐ฟ4

โซถ

๐ฟ๐‘›โˆ’1 = ๐ฟ๐‘›+1 โˆ’ ๐ฟ๐‘›

๐ฟ๐‘› = ๐ฟ๐‘›+2 โˆ’ ๐ฟ๐‘›+1

Adding these equations, we get ,

๐ฟ๐‘– =

๐‘›

1

๐ฟ๐‘›+2 โˆ’ ๐ฟ2

= ๐ฟ๐‘›+2 โˆ’ 3

Result 2.2

๐ฟ2๐‘–โˆ’1 =

๐‘›

1

๐ฟ2๐‘› โˆ’ 2 (2.7)

Proof:

Using the Lucas recurrence relations, we have,

๐ฟ1 = ๐ฟ2 โˆ’ ๐ฟ0

๐ฟ3 = ๐ฟ4 โˆ’ ๐ฟ2

๐ฟ5 = ๐ฟ6 โˆ’ ๐ฟ4

โซถ

๐ฟ2๐‘›โˆ’3 = ๐ฟ2๐‘›โˆ’2 โˆ’ ๐ฟ2๐‘›โˆ’4

๐ฟ2๐‘›โˆ’1 = ๐ฟ2๐‘› โˆ’ ๐ฟ2๐‘›โˆ’2

Adding these equations, we get,

๐ฟ2๐‘–โˆ’1 =

๐‘›

1

๐ฟ2๐‘› โˆ’ ๐ฟ0

= ๐ฟ2๐‘› โˆ’ 2

For example,

๐ฟ2๐‘–โˆ’1 = 15125 = 15127 โˆ’ 2 = ๐ฟ20 โˆ’ 2

10

1

Result 2.3

๐ฟ2๐‘– =

๐‘›

1

๐ฟ2๐‘›+1 โˆ’ 1 (2.8)

Proof:

๐ฟ2๐‘– =

๐‘›

1

๐ฟ๐‘–

2๐‘›

1

โˆ’ ๐ฟ2๐‘–โˆ’1

๐‘›

1

= (๐ฟ2๐‘›+2 โˆ’ 3) โˆ’ (๐ฟ2๐‘› โˆ’ 2)

= ๐ฟ2๐‘›+2 โˆ’ ๐ฟ2๐‘› โˆ’ 1

= ๐ฟ2๐‘›+1 โˆ’ 1 , by the Lucas recurrence relation.

Result 2.4

๐ฟ๐‘›โˆ’1๐ฟ๐‘›+1 โˆ’ ๐ฟ๐‘›2 = 5(โˆ’1)๐‘›โˆ’1 (2.9)

Proof:

Since ๐ฟ๐‘œ๐ฟ2 โˆ’ ๐ฟ12 = 2 ห‘ 3 โˆ’ 12

= 6 โ€“ 1

= 5

= 5(โˆ’1)1โˆ’1

The given statement is clearly true when ๐‘› = 1. Now we assume it is true for an

arbitrary positive integer ๐‘˜, ๐ฟ๐‘˜โˆ’1๐ฟ๐‘˜+1 โˆ’ ๐ฟ๐‘˜2 = 5(โˆ’1)๐‘˜โˆ’1

Then,

, ๐ฟ๐‘˜๐ฟ๐‘˜+2 โˆ’ ๐ฟ๐‘˜+12 = ๐ฟ๐‘˜+1 โˆ’ ๐ฟ๐‘˜โˆ’1 ๐ฟ๐‘˜ + ๐ฟ๐‘˜+1 โˆ’ ๐ฟ๐‘˜+1

2

= ๐ฟ๐‘˜๐ฟ๐‘˜+1 + ๐ฟ๐‘˜+12 โˆ’ ๐ฟ๐‘˜๐ฟ๐‘˜โˆ’1 โˆ’ ๐ฟ๐‘˜โˆ’1๐ฟ๐‘˜+1 โˆ’ ๐ฟ๐‘˜+1

2

= ๐ฟ๐‘˜๐ฟ๐‘˜+1 โˆ’ ๐ฟ๐‘˜๐ฟ๐‘˜โˆ’1 โˆ’ ๐ฟ๐‘˜2 โˆ’ 5 โˆ’1 ๐‘˜โˆ’1

= ๐ฟ๐‘˜๐ฟ๐‘˜+1 โˆ’ ๐ฟ๐‘˜ ๐ฟ๐‘˜โˆ’1 + ๐ฟ๐‘˜ + 5(โˆ’1)๐‘˜

= ๐ฟ๐‘˜๐ฟ๐‘˜+1 โˆ’ ๐ฟ๐‘˜๐ฟ๐‘˜+1 + 5(โˆ’1)๐‘˜

= 5(โˆ’1)๐‘˜

Thus the formula works for ๐‘› = ๐‘˜ + 1.

So, by the Principle of Mathematical Induction, the statement is true for every

integer ๐‘› โ‰ฅ 1.

Result 2.5

๐ฟ๐‘–2 =

๐‘›

1

๐ฟ๐‘›๐ฟ๐‘›+1 โˆ’ 2 (2.10)

Proof:

When ๐‘› = 1,

L. H. S = ๐ฟ๐‘–2

1

1

= ๐ฟ12

= 1 = 1 ยท 3 โˆ’ 2

= ๐น1๐น2 โˆ’ 2

= R. H. S

So the result is true when ๐‘› = 1.

Assume it is true for an arbitrary positive integer ๐‘˜, ๐ฟ๐‘–2 =

๐‘˜

1

๐ฟ๐‘˜๐ฟ๐‘˜+1 . Then,

๐ฟ๐‘–2 =

๐‘˜+1

1

๐ฟ๐‘–2 +

๐‘˜

1

๐ฟ๐‘˜+12

= ๐ฟ๐‘˜๐ฟ๐‘˜+1 โˆ’ 2 + ๐ฟ๐‘˜+12 , by the inductive hypothesis

= ๐ฟ๐‘˜+1 ๐ฟ๐‘˜ + ๐ฟ๐‘˜+1 โˆ’ 2

= ๐ฟ๐‘˜+1๐ฟ๐‘˜+2 โˆ’ 2 , by the Lucas recurrence relation

So the statement is true when ๐‘› = ๐‘˜ + 1.

Thus it is true for every positive integer ๐‘›.

For example,

๐ฟ๐‘–2 =

๐‘›

1

24475 = 24477 โˆ’ 2 = 123 ยท 199 โˆ’ 2 = ๐ฟ10๐ฟ11 โˆ’ 2

Theorem 2.5

๐น๐‘›+๐‘š = ๐น๐‘›โˆ’1๐น๐‘š + ๐น๐‘›๐น๐‘š+1

Proof:

We shall prove the theorem by induction on ๐‘š.

For ๐‘š = 1, we get ๐น๐‘›+1 = ๐น๐‘›โˆ’1๐น1 + ๐น๐‘›๐น1+1 = ๐น๐‘›โˆ’1 + ๐น๐‘› which is true.

Suppose that it is true for ๐‘š = ๐‘˜ and ๐‘š = ๐‘˜ + 1,we shall prove it is also true for

๐‘š = ๐‘˜ + 2.

Let ๐น๐‘›+๐‘˜ = ๐น๐‘›โˆ’1๐น๐‘˜ + ๐น๐‘›๐น๐‘˜+1 and ๐น๐‘›+(๐‘˜+1) = ๐น๐‘›โˆ’1๐น๐‘˜+1 + ๐น๐‘›๐น๐‘˜+2.

Adding these two equations, we get

๐น๐‘›+(๐‘˜+2) = ๐น๐‘›โˆ’1๐น๐‘˜+2 + ๐น๐‘›๐น๐‘˜+3

Hence, ๐น๐‘›+๐‘š = ๐น๐‘›โˆ’1๐น๐‘š + ๐น๐‘›๐น๐‘š+1.

Note:

To derive new identities, we now present an explicit formula for ๐น๐‘› .

Let ๐›ผ and ๐›ฝ be the roots of the quadratic equation ๐‘ฅ2 โˆ’ ๐‘ฅ โˆ’ 1 = 0,

so, ๐›ผ = (1 + 5)/2 and ๐›ฝ = (1 โˆ’ 5)/2

Then ๐›ผ + ๐›ฝ = 1 and ๐›ผ๐›ฝ = โˆ’1

Besides, ๐›ผ2 = ๐›ผ (1 โˆ’ ๐›ฝ) = ๐›ผ โˆ’ ๐›ผ๐›ฝ = ๐›ผ + 1

๐›ผ3 = ๐›ผ(๐›ผ + 1) = ๐›ผ2 + ๐›ผ = 2๐›ผ + 1

and ๐›ผ4 = ๐›ผ 2๐›ผ + 1 = 2๐›ผ2 + ๐›ผ = 2(๐›ผ + 1) + ๐›ผ = 3๐›ผ + 2

Thus we have;

๐›ผ = 1๐›ผ + 0

๐›ผ2 = 1๐›ผ + 1

๐›ผ3 = 2๐›ผ + 1

๐›ผ4 = 3๐›ผ + 2

Notice an interesting pattern emerging. The constant term and the coefficient of

๐›ผ on the RHS appear to be adjacent Fibonacci numbers.

Accordingly, we have the following result.

Lemma 2.1

๐›ผ๐‘› = ๐›ผ๐น๐‘› + ๐น๐‘›โˆ’1 , where ๐‘› โ‰ฅ 0.

Corollary 2.3

๐›ฝ๐‘› = ๐›ฝ๐น๐‘› + ๐น๐‘›โˆ’1 , where ๐‘› โ‰ฅ 0.

Note:

Let ๐‘ข๐‘› = ๐›ผ๐‘› โˆ’ ๐›ฝ๐‘› / 5 , where ๐‘› โ‰ฅ 1. Then,

๐‘ข1 = ๐›ผโˆ’๐›ฝ

5 =

5

5 = 1 and

๐‘ข2 = ๐›ผ2โˆ’๐›ฝ2

5 =

๐›ผ+๐›ฝ (๐›ผโˆ’๐›ฝ)

5 = 1

Suppose ๐‘› โ‰ฅ 3. Then,

๐‘ข๐‘›โˆ’1 + ๐‘ข๐‘›โˆ’2 = ๐›ผ๐‘›โˆ’1โˆ’ ๐›ฝ๐‘›โˆ’1

5 +

๐›ผ๐‘›โˆ’2โˆ’ ๐›ฝ๐‘›โˆ’2

5

= ๐›ผ๐‘›โˆ’2 ๐›ผ+1 โˆ’๐›ฝ๐‘›โˆ’2 ๐›ฝ+1

5

= ๐›ผ๐‘›โˆ’2๐›ผ2โˆ’ ๐›ฝ๐‘›โˆ’2๐›ฝ2

5

= ๐›ผ๐‘›โˆ’ ๐›ฝ๐‘›

5

= ๐‘ข๐‘›

Thus ๐‘ข๐‘› satisfies the Fibonacci recurrence relation and the two initial conditions.

This gives us an explicit formula for ๐น๐‘› : ๐น๐‘› = ๐‘ข๐‘›

Theorem 2.6

Let ๐›ผ be the positive root of the quadratic equation ๐‘ฅ2 โˆ’ ๐‘ฅ โˆ’ 1 = 0 and ๐›ฝ its

negative root. Then

๐น๐‘› = ๐›ผ๐‘›โˆ’๐›ฝ๐‘›

๐›ผโˆ’๐›ฝ , where ๐‘› โ‰ฅ 1

This explicit formula for ๐น๐‘› is called Binetโ€™s formula, after the French

mathematician Jacques-Phillipe-Marie Binet, who discovered it in 1843.

Corollary 2.4 (Lucas, 1876)

๐น๐‘›+12 + ๐น๐‘›

2 = ๐น2๐‘›+1 (2.11)

๐น๐‘›+12 โˆ’ ๐น๐‘›โˆ’1

2 = ๐น2๐‘› (2.12)

For example,

๐น92 + ๐น8

2 = 342 + 212 = 1156 + 441 = 1597 = ๐น17

๐น112 โˆ’ ๐น9

2 = 892 โˆ’ 342 = 7921โˆ’1156 = 6765 = ๐น20

Corresponding to Binetโ€™s formula for ๐น๐‘› , there is one for ๐ฟ๐‘› also.

Theorem 2.7

Let ๐‘› โ‰ฅ 1. Then ๐ฟ๐‘› = ๐›ผ๐‘› + ๐›ฝ๐‘›

Corollary 2.5

๐น2๐‘› = ๐น๐‘›๐ฟ๐‘› (2.13)

๐น๐‘›โˆ’1 + ๐น๐‘›+1 = ๐ฟ๐‘› (2.14)

๐น๐‘›+2 โˆ’ ๐น๐‘›โˆ’2 = ๐ฟ๐‘› (2.15)

๐ฟ๐‘›โˆ’1 + ๐ฟ๐‘›+1 = 5๐น๐‘› (2.16)

For example,

๐น22 = 17711 = 89 ยท 199 = ๐น11๐ฟ11

๐น15 + ๐น17 = 610 + 1597 = 2207 = ๐ฟ16

๐น15 โˆ’ ๐น11 = 610 โˆ’ 89 = 521 = ๐ฟ13

๐ฟ12 + ๐ฟ14 = 322 + 843 = 1165 = 5 ยท 233 = 5 ยท ๐น13

Identity (2.13) implies that when ๐‘› โ‰ฅ 3, every Fibonacci number

๐น2๐‘› with an even subscript has nontrivial factors. According to identity (2.14), the

sum of any two Fibonacci numbers that are two units away is a Lucas number.

Likewise, by identity (2.15), the difference of any two Fibonacci numbers that lie

four units away is also a Lucas number.

Identity (2.13) has an interesting by-product.

Let 2๐‘› = 2๐‘š , where ๐‘š โ‰ฅ 1. Then,

๐น2๐‘š = ๐ฟ2๐‘šโˆ’1 ๐น2๐‘šโˆ’1

= ๐ฟ2๐‘šโˆ’1 (๐ฟ2๐‘šโˆ’2 ๐น2๐‘šโˆ’2 )

= ๐ฟ2๐‘šโˆ’1 ๐ฟ2๐‘šโˆ’2 ๐น2๐‘šโˆ’2

= ๐ฟ2๐‘šโˆ’1 ๐ฟ2๐‘šโˆ’2 (๐ฟ2๐‘šโˆ’3 ๐น2๐‘šโˆ’3 )

= ๐ฟ2๐‘šโˆ’1 ๐ฟ2๐‘šโˆ’2 ๐ฟ2๐‘šโˆ’3 ๐น2๐‘šโˆ’3

Continuing like this we get,

๐น2๐‘š = ๐ฟ2๐‘šโˆ’1 ๐ฟ2๐‘šโˆ’2 โ€ฆโ€ฆโ€ฆ๐ฟ8 ๐ฟ4 ๐ฟ2 ๐ฟ1

For example,

๐น16 = ๐ฟ8๐ฟ4๐ฟ2๐ฟ1 = 47 ยท 7 ยท 3 ยท 1 = 987

Theorem 2.8

A positive integer ๐‘› is a Fibonacci number if and only if 5๐‘›2 ยฑ 4 is a perfect

square.

Proof:

We have,

(โˆ’1)๐‘Ÿ + ๐น๐‘Ÿ2 = ๐น๐‘Ÿ+1๐น๐‘Ÿโˆ’1 [Cassiniโ€™s Formula]

and ๐ฟ๐‘Ÿ = ๐น๐‘Ÿ+1 + ๐น๐‘Ÿโˆ’1

โˆด ๐ฟ๐‘Ÿ2 โˆ’ 4 โˆ’1 ๐‘Ÿ + ๐น๐‘Ÿ

2 = ๐น๐‘Ÿ+1 + ๐น๐‘Ÿโˆ’1 2 โˆ’ 4๐น๐‘Ÿ+1๐น๐‘Ÿโˆ’1

= (๐น๐‘Ÿ+1 โˆ’ ๐น๐‘Ÿโˆ’1)2

= ๐น๐‘Ÿ2

๐ฟ๐‘Ÿ2 = 5๐น๐‘Ÿ

2 +4(โˆ’1)๐‘Ÿ

Thus if ๐‘› is a Fibonacci number, then 5๐‘›2 ยฑ 4 is a perfect square.

Conversely, let 5๐‘›2 ยฑ 4 be a perfect square ๐‘š2 . Then,

๐‘š2 โˆ’ 5๐‘›2 = ยฑ4

๐‘š+๐‘› 5

2.๐‘šโˆ’๐‘› 5

2 = ยฑ1

Since ๐‘š and ๐‘› have the same parity (both odd or both even), both

(๐‘š + ๐‘› 5)/2 and (๐‘šโˆ’ ๐‘› 5)/2 are integers in the extension field ๐‘„( 5) =

{ ๐‘ฅ + ๐‘ฆ 5|๐‘ฅ,๐‘ฆ โˆˆ ๐‘„}, where ๐‘„ denotes the set of rational numbers. Since their

product is ยฑ1, they must be units in the field. But the only integral units in ๐‘„( 5)

are of the form ยฑ๐›ผยฑ๐‘– . Then,

๐‘š+๐‘› 5

2 = ๐›ผ๐‘– =

1

2 [(๐›ผ๐‘– + ๐›ฝ๐‘– ) + (๐›ผ๐‘– โˆ’ ๐›ฝ๐‘–) ]

= ๐ฟ๐‘–+๐น๐‘– 5

2

Thus ๐‘› = ๐น๐‘– , a Fibonacci number.

NUMBER OF DIGITS IN ๐‘ญ๐’ AND ๐‘ณ๐’

Binetโ€™s formula can be successfully employed to predetermine the number of

digits in ๐น๐‘› and ๐ฟ๐‘› . We can show this by writing ๐น๐‘› as ๐น๐‘› = ๐›ผ๐‘›

5 [1 โˆ’

๐›ฝ

๐›ผ ๐‘›

]

Since |๐›ฝ| < |๐›ผ| , ๐›ฝ ๐›ผ ๐‘› โ†’ 0 as ๐‘› โ†’ โˆž. Therefore, when ๐‘› is sufficiently large,

๐น๐‘› โ‰ˆ ๐›ผ๐‘›

5

log ๐น๐‘› โ‰ˆ ๐‘› log ๐›ผ โˆ’ (log 5)/2

Number of digits in ๐น๐‘› = 1 + characteristic of log ๐น๐‘› = log๐น๐‘›

= ๐‘› log๐›ผ โˆ’ (log 5 )/2

= ๐‘› log( 1 + 5 ) โˆ’ log 2 โˆ’ (log 5 )/2

For example,

The number of digits in ๐น25 is given by,

25 log 1 + 5 โˆ’ log 2 โˆ’ log 5 /2 = 4.875206004 = 5

Note that F25 = 75025 , i.e, it exactly have 5 digits which establishes the

result.

Since ๐ฟ๐‘› = ๐›ผ๐‘› + ๐›ฝ๐‘› , it follows that when ๐‘› is sufficiently large, ๐ฟ๐‘› โ‰ˆ ฮฑn, so that

log ๐ฟ๐‘› โ‰ˆ ๐‘› log ๐›ผ . Thus the number of digits in ๐ฟ๐‘› is given by

log๐ฟ๐‘› = ๐‘› log๐›ผ = ๐‘› [log 1 + 5 โˆ’ log 2]

For example,

The number of digits in ๐ฟ28 is given by,

28 log 1 + 5 โˆ’ log 2 = 5.851653927 = 6

Note that ๐ฟ28 = 710647, i.e, it exactly have 6 digits which establishes the result.

Theorem 2.9

A positive integer ๐‘› is a Lucas number if and only if 5๐‘›2 ยฑ 20 is a

perfect square.

Proof:

Let ๐‘› = ๐ฟ2๐‘š+1. Then

5๐‘›2 + 20 = 5(๐›ผ2๐‘š+1 + ๐›ฝ2๐‘š+1)2 + 20

= 5[๐›ผ4๐‘š+2 + ๐›ฝ4๐‘š+2 + 2 ๐›ผ๐›ฝ)2๐‘š+1 + 20

= 5[๐›ผ4๐‘š+2 + ๐›ฝ4๐‘š+2 โˆ’ 2(๐›ผ๐›ฝ)2๐‘š+1]

= 5(๐›ผ2๐‘š+1 โˆ’ ๐›ฝ2๐‘š+1)2

= 5( 5๐น2๐‘š+1)2 = 25๐น2๐‘š+1 2

On the otherhand, let ๐‘› = ๐ฟ2๐‘š . Then

5๐‘›2 โˆ’ 20 = 5(๐›ผ2๐‘š + ๐›ฝ2๐‘š)2 โˆ’ 20 = 5[๐›ผ4๐‘š + ๐›ฝ4๐‘š + 2 ๐›ผ๐›ฝ)2๐‘š โˆ’ 20

= 5[๐›ผ4๐‘š + ๐›ฝ4๐‘š โˆ’ 2 ๐›ผ๐›ฝ)2๐‘š = 25๐น2๐‘š 2

Thus, if ๐‘› is a Lucas number, then 5๐‘›2 ยฑ 20 is a perfect square.

Because the proof of the converse is a bit complicated, we omit it.

For example,

Let ๐‘› = 1364 = ๐ฟ15 . Then 5๐‘›2 + 20 = 5 ยท 13642 + 20 = 9302500 = 30502, a

perfect square.

Let ๐‘› = 322 = ๐ฟ12 . Then 5๐‘›2 โˆ’ 20 = 5 ยท 3222 โˆ’ 20 = 518400 = 7202, a

perfect square.

With Binet's formulas at hand, we can extend the definitions of ๐น๐‘› and ๐ฟ๐‘› to

negative subscripts also. If we apply the Fibonacci recurrence relation to the

negative side, we get

. . . ๐นโˆ’4 ๐นโˆ’3 ๐นโˆ’2 ๐นโˆ’1 ๐น0 ๐น1 ๐น2 ๐น3 ๐น4 . . . . . . โˆ’3 2 โˆ’ 1 1 0 1 1 2 3 . . .

So, it appears that ๐นโˆ’๐‘› = (โˆ’1)๐‘›+1๐น๐‘› ,๐‘› โ‰ฅ 1. To prove this, assume Binet's formula holds for negative exponents

๐นโˆ’๐‘› = ๐›ผโˆ’๐‘›โˆ’ ๐›ฝโˆ’๐‘›

5 =

(โˆ’๐›ฝ)๐‘›โˆ’ (โˆ’๐›ผ)๐‘›

5 since ๐›ผ๐›ฝ = โˆ’1

= (โˆ’1)๐‘› (๐›ฝ๐‘›โˆ’ ๐›ผ๐‘› )

5 =

(โˆ’1)๐‘›+1 (๐›ผ๐‘›โˆ’ ๐›ฝ๐‘› )

5

= (โˆ’1)๐‘›+1๐น๐‘› (2.18)

Likewise,

๐ฟโˆ’๐‘› = (โˆ’1)๐‘› ๐ฟ๐‘› (2.19)

Thus, ๐นโˆ’๐‘› = ๐น๐‘› if and only if ๐‘› is odd, and ๐ฟโˆ’๐‘› = ๐ฟ๐‘› if and only if ๐‘› is even.

A formula for ๐›ผโˆ’๐‘› can now be derived easily.

Since ๐›ผ๐‘› = ๐›ผ ๐น๐‘› + ๐น๐‘›โˆ’1

(Lemma 2.1), it follows that

๐›ผโˆ’๐‘› = ๐›ผ ๐นโˆ’๐‘› + ๐นโˆ’๐‘›โˆ’1

= ๐›ผ (โˆ’1)๐‘›+1 ๐น๐‘› + (โˆ’1)๐‘›+2 ๐น๐‘›+1

= (โˆ’1)๐‘›+1 (๐›ผ ๐น๐‘› โˆ’ ๐น๐‘›+1)

= ๐›ผ๐น๐‘› โˆ’ ๐น๐‘›+1 if ๐‘› is odd ๐น๐‘›+1 โˆ’ ๐›ผ ๐น๐‘› otherwise

(2.20)

For example,

ฮฑโˆ’12 = F11 โˆ’ ฮฑF12 = 89 โˆ’ 144ฮฑ

Formula 2.20 can also be established by Principle of Mathematical Induction or by

showing that ๐›ผ๐‘› ๐›ผ๐น๐‘› โˆ’ ๐น๐‘›+1 = โˆ’1 ๐‘›+1, using Binet's formula.

Likewise,

๐›ฝโˆ’๐‘› = ๐›ฝ๐น๐‘› โˆ’ ๐น๐‘›+1 if ๐‘› is odd๐น๐‘›+1 โˆ’ ๐›ฝ๐น๐‘› otherwise

(2.21)

Notice two intriguing patterns that emerge from ๐›ผโˆ’๐‘› :

ฮฑโˆ’1 = 1 ยท ฮฑ โˆ’ 1

ฮฑโˆ’2 = 2 โˆ’ 1 ยท ๐›ผ

ฮฑโˆ’3 = 2 ยท ฮฑ โˆ’ 3

ฮฑโˆ’4 = 5 โˆ’ 3 ยท ฮฑ

ฮฑโˆ’5 = 5 ยท ฮฑ โˆ’ 8

โ‹ฎ

They are indicated by the two crisscrossing arrows: The absolute values of the

coefficients of ๐›ผ are consecutive Fibonacci numbers, and so are the absolute values

of the various constants.

The summation Formulas (2.1) through (2.3) are a special case of the generalized

summation formula, given in the next theorem. In addition, the theorem yields an

array of fascinating formulas as by-products. Its proof is a consequence of Binet's

formulas and the geometric summation formula

๐‘Ÿ๐‘–๐‘›โˆ’1

๐‘–=0

=๐‘Ÿ๐‘› โˆ’ 1

๐‘Ÿ โˆ’ 1 where ๐‘Ÿ โ‰  1

Theorem 2.10 (Koshy, 1998)

Let ๐‘˜ โ‰ฅ 1 and ๐‘— any integer. Then

๐น๐‘˜๐‘–+๐‘—

๐‘›

๐‘–=0

=

๐น๐‘›๐‘˜+๐‘˜+๐‘— โˆ’ โˆ’1 ๐‘˜ ๐น๐‘›๐‘˜+๐‘— โˆ’ ๐น๐‘— โˆ’ โˆ’1 ๐‘— ๐น๐‘˜โˆ’๐‘—

๐ฟ๐‘˜ โˆ’ โˆ’1 ๐‘˜ โˆ’ 1 if ๐‘— < ๐‘˜

๐น๐‘›๐‘˜+๐‘˜+๐‘— โˆ’ โˆ’1 ๐‘˜ ๐น๐‘›๐‘˜+๐‘— โˆ’ ๐น๐‘— + โˆ’1 ๐‘˜ ๐น๐‘—โˆ’๐‘˜

๐ฟ๐‘˜ โˆ’ โˆ’1 ๐‘˜ โˆ’ 1 otherwise

(2.22)

Proof:

๐น๐‘˜๐‘–+๐‘— = ๐›ผ๐‘˜๐‘–+๐‘— โˆ’ ๐›ฝ๐‘˜๐‘–+๐‘—

5

๐‘›

๐‘–=0

๐‘›

๐‘–=0

= 1

5 ๐›ผ๐‘— ๐›ผ๐‘˜๐‘– โˆ’ ๐›ฝ๐‘— ๐›ฝ๐‘˜๐‘–

=1

5 ๐›ผ๐‘— .

๐›ผ๐‘›๐‘˜+๐‘˜ โˆ’ 1

๐›ผ๐‘˜ โˆ’ 1โˆ’ ๐›ฝ๐‘— .

๐›ฝ๐‘›๐‘˜+๐‘˜ โˆ’ 1

๐›ฝ๐‘˜ โˆ’ 1

= ๐›ผ๐‘›๐‘˜ +๐‘˜+๐‘— โˆ’ ๐›ผ ๐‘— ๐›ฝ๐‘˜โˆ’ 1 โˆ’ ๐›ฝ๐‘›๐‘˜ +๐‘˜+๐‘— โˆ’ ๐›ฝ ๐‘— ๐›ผ๐‘˜ โˆ’ 1

5 ๐›ผ๐›ฝ ๐‘˜ โˆ’ ๐›ผ๐‘˜+๐›ฝ๐‘˜ + 1

= โˆ’๐น๐‘›๐‘˜ +๐‘˜+๐‘— + (โˆ’1)๐‘˜ ๐น๐‘›๐‘˜ +๐‘— + ๐น๐‘—+ (๐›ผ๐‘˜๐›ฝ ๐‘— โˆ’ ๐›ผ ๐‘—๐›ฝ๐‘˜)/ 5

(โˆ’1)๐‘˜ โˆ’ ๐ฟ๐‘˜+ 1

But ๐›ผ๐‘˜๐›ฝ๐‘— โˆ’ ๐›ผ๐‘—๐›ฝ๐‘˜ = ๐›ผ๐›ฝ ๐‘— ๐›ผ๐‘˜โˆ’๐‘— โˆ’ ๐›ฝ๐‘˜โˆ’๐‘— if ๐‘— < ๐‘˜

๐›ผ๐›ฝ ๐‘˜ ๐›ฝ๐‘—โˆ’๐‘˜ โˆ’ ๐›ผ๐‘—โˆ’๐‘˜ otherwise

= โˆ’1 ๐‘— 5๐น๐‘˜โˆ’๐‘— if ๐‘— < ๐‘˜

โˆ’1 ๐‘˜+1 5๐น๐‘—โˆ’๐‘˜ otherwise

โˆด ๐น๐‘˜๐‘–+๐‘— =

๐น๐‘›๐‘˜+๐‘˜+๐‘— โˆ’ โˆ’1 ๐‘˜๐น๐‘›๐‘˜+๐‘— โˆ’ ๐น๐‘— โˆ’ โˆ’1 ๐‘—๐น๐‘˜โˆ’๐‘—

๐ฟ๐‘˜ โˆ’ โˆ’1 ๐‘˜ โˆ’ 1 if ๐‘— < ๐‘˜

๐น๐‘›๐‘˜+๐‘˜+๐‘— โˆ’ โˆ’1 ๐‘˜๐น๐‘›๐‘˜+๐‘— โˆ’ ๐น๐‘— โˆ’ โˆ’1 ๐‘—๐น๐‘—โˆ’๐‘˜

๐ฟ๐‘˜ โˆ’ โˆ’1 ๐‘˜ โˆ’ 1 otherwise

๐‘›

๐‘–=0

Letting ๐‘— = 0 in this formula yields the following result.

Corollary 2.6 (Koshy, 1998)

๐น๐‘˜๐‘– = ๐น๐‘›๐‘˜+๐‘˜ โˆ’ (โˆ’1)๐‘˜ ๐น๐‘›๐‘˜ โˆ’ ๐น๐‘˜

๐ฟ๐‘˜ โˆ’ (โˆ’1)๐‘˜ โˆ’ 1

๐‘›

๐‘–=1

(2.23)

CHAPTER - 3

DIVISIBILITY PROPERTIES

Theorem 3.1

๐น๐‘š |๐น๐‘š๐‘›

Proof:

We can prove this theorem by applying the Principle of Mathematical

Induction.

The given statement is clearly true when ๐‘› = 1.

Now assume it is true for all integers 1 through ๐‘˜, where ๐‘˜ โ‰ฅ 1.

i.e. ๐น๐‘š |๐น๐‘š๐‘– for every ๐‘–, where 1 โ‰ค ๐‘– โ‰ค ๐‘˜.

To show that ๐น๐‘š |๐น๐‘š ๐‘˜+1 , we use the identity: ๐น๐‘Ÿ+๐‘  = ๐น๐‘Ÿโˆ’1๐น๐‘  + ๐น๐‘Ÿ๐น๐‘ +1

โˆด ๐น๐‘š (๐‘˜+1) = ๐น๐‘š๐‘˜ +๐‘š = ๐น๐‘š๐‘˜โˆ’1๐น๐‘š + ๐น๐‘š๐‘˜ ๐น๐‘š+1

Since ๐น๐‘š |๐น๐‘š๐‘˜ , by the induction hypothesis, it follows that ๐น๐‘š |๐น๐‘š ๐‘˜+1 .

Thus by the strong version of the Principle of Mathematical Induction,

the result is true for all integers ๐‘› โ‰ฅ 1.

For example,

๐น7 = 13, and ๐น28 = 317811. Since 7|28, it follows by the theorem that 13|317811.

Particularly,

๐น3 ,๐น6 ,๐น9 โ€ฆโ€ฆโ€ฆ are all divisible by ๐น3

๐น4 ,๐น8 ,๐น12 โ€ฆโ€ฆโ€ฆ are all divisible by ๐น4

๐น5 ,๐น10 ,๐น15 โ€ฆโ€ฆโ€ฆ are all divisible by ๐น5

Theorem 3.2

If ๐น๐‘š |๐น๐‘› , then ๐‘š|๐‘›.

Proof:

By the division algorithm, ๐‘› = ๐‘ž๐‘š + ๐‘Ÿ, where 0 โ‰ค ๐‘Ÿ < ๐‘š. Suppose ๐น๐‘š |๐น๐‘› .

Then by theorem 1.2 and by the identity ๐น๐‘› = ๐น๐‘›โˆ’๐‘š+1๐น๐‘š + ๐น๐‘›โˆ’๐‘š๐น๐‘šโˆ’1, we get

๐น๐‘š |๐น๐‘›โˆ’๐‘š๐น๐‘šโˆ’1. But ๐น๐‘š ,๐น๐‘šโˆ’1 = 1, so ๐น๐‘š |๐น๐‘›โˆ’๐‘š .

Similarly, ๐น๐‘š |๐น๐‘›โˆ’2๐‘š . Continuing like this, ๐น๐‘š |๐น๐‘›โˆ’๐‘ž๐‘š , i.e, ๐น๐‘š |๐น๐‘Ÿ .

This is possible unless ๐‘Ÿ = 0.โˆด ๐‘› = ๐‘ž๐‘š.

Thus ๐น๐‘š |๐น๐‘› implies ๐‘š|๐‘›.

Corollary 3.1

๐น๐‘š |๐น๐‘› if and only if ๐‘š|๐‘›.

Corollary 3.2

If ๐‘š, ๐‘› = 1, then ๐น๐‘š๐น๐‘› |๐น๐‘š๐‘› .

Proof:

By theorem 3.1, ๐น๐‘š |๐น๐‘š๐‘› and ๐น๐‘› |๐น๐‘š๐‘› . Therefore, ๐น๐‘š ,๐น๐‘› |๐น๐‘š๐‘› .

But ๐น๐‘š ,๐น๐‘› = ๐น ๐‘š ,๐‘› = ๐น1 = 1, so ๐น๐‘š ,๐น๐‘› = ๐น๐‘š๐น๐‘› . Thus ๐น๐‘š๐น๐‘› |๐น๐‘š๐‘› .

For example,

5,6 = 1,๐น5 = 5,๐น6 = 8, and ๐น30 = 832040. we can verify that 5 ยท 8|832040;

i.e, ๐น5๐น6|๐น30 .

Lemma 3.1

(๐น๐‘ž๐‘›โˆ’1 ,๐น๐‘›) = 1.

Proof:

Let ๐‘‘ = (๐น๐‘ž๐‘›โˆ’1,๐น๐‘›). Then ๐‘‘|๐น๐‘ž๐‘›โˆ’1 and ๐‘‘|๐น๐‘› . Since ๐น๐‘› |๐น๐‘ž๐‘› by

Theorem 3.1, ๐‘‘|๐น๐‘ž๐‘› . Thus ๐‘‘|๐น๐‘ž๐‘›โˆ’1 and ๐‘‘|๐น๐‘ž๐‘› . But (๐น๐‘ž๐‘›โˆ’1,๐น๐‘ž๐‘› ) = 1,by Corollary

2.2. Therefore ๐‘‘|1, so ๐‘‘ = 1 . Thus (๐น๐‘ž๐‘›โˆ’1,๐น๐‘›) = 1.

Lemma 3.2

Let ๐‘š = ๐‘ž๐‘› + ๐‘Ÿ. Then (๐น๐‘š ,๐น๐‘›) = (๐น๐‘› ,๐น๐‘Ÿ ).

Proof:

๐น๐‘š ,๐น๐‘› = ๐น๐‘ž๐‘›+๐‘Ÿ , ๐น๐‘›

= ๐น๐‘ž๐‘›โˆ’1๐น๐‘Ÿ + ๐น๐‘ž๐‘›๐น๐‘Ÿ+1๐น๐‘› , ๐น๐‘› by theorem (2.5)

= ๐น๐‘ž๐‘›โˆ’1๐น๐‘Ÿ , ๐น๐‘›

= ๐น๐‘Ÿ ,๐น๐‘› by Lemma 3.1

= ๐น๐‘› , ๐น๐‘Ÿ

Theorem 3.3

๐น๐‘š ,๐น๐‘› = ๐น ๐‘š ,๐‘›

Proof:

Suppose ๐‘š โ‰ฅ ๐‘›. Applying the Euclidean algorithm with ๐‘š as the dividend

and ๐‘› as the divisor, we get the following sequence of equations:

๐‘š = ๐‘ž0๐‘› + ๐‘Ÿ1 0 โ‰ค ๐‘Ÿ1 < ๐‘›

๐‘› = ๐‘ž1๐‘Ÿ1 + ๐‘Ÿ2 0 โ‰ค ๐‘Ÿ2 < ๐‘Ÿ1

๐‘Ÿ1 = ๐‘ž2๐‘Ÿ2 + ๐‘Ÿ3 0 โ‰ค ๐‘Ÿ3 < ๐‘Ÿ2

โซถ

๐‘Ÿ๐‘›โˆ’2 = ๐‘ž๐‘›โˆ’1๐‘Ÿ๐‘›โˆ’1 + ๐‘Ÿ๐‘› 0 โ‰ค ๐‘Ÿ๐‘› < ๐‘Ÿ๐‘›โˆ’1

๐‘Ÿ๐‘›โˆ’1 = ๐‘ž๐‘›๐‘Ÿ๐‘› + 0

By Lemma 3.2, ๐น๐‘š ,๐น๐‘› = ๐น๐‘› ,๐น๐‘Ÿ1 = ๐น๐‘Ÿ1

,๐น๐‘Ÿ2 = โ‹ฏ ยทยทยท = ๐น๐‘Ÿ๐‘›โˆ’1

,๐น๐‘Ÿ๐‘› .

But ๐‘Ÿ๐‘› |๐‘Ÿ๐‘›โˆ’1, so ๐น๐‘Ÿ๐‘› |๐น๐‘Ÿ๐‘›โˆ’1, by Theorem 3.1. Therefore, (๐น๐‘Ÿ๐‘›โˆ’1

,๐น๐‘Ÿ๐‘› ) = ๐น๐‘Ÿ๐‘› . Thus

(๐น๐‘š ,๐น๐‘›) = ๐น๐‘Ÿ๐‘› .

But, by Euclidean algorithm, ๐‘Ÿ๐‘› = (๐‘š,๐‘›) ; therefore, (๐น๐‘š ,๐น๐‘›) = ๐น(๐‘š ,๐‘›).

For example,

F16 , F24 = ๐น 16,24 = ๐น8 = 21. That is, 987, 46368 = 21

Corollary 3.3

If ๐‘š and ๐‘› are relatively prime, then so are ๐น๐‘š and ๐น๐‘› .

For example,

19, 27 = 1, so F19, F27 = 4181, 196418 = 1

Corollary 3.4

If ๐น๐‘š |๐น๐‘› , then ๐‘š|๐‘›.

Proof:

Suppose ๐น๐‘š |๐น๐‘› . Then ๐น๐‘š ,๐น๐‘› = ๐น๐‘š = ๐น(๐‘š ,๐‘›), by Theorem 3.3;

โˆด ๐‘š = ๐‘š,๐‘› . Thus ๐‘šว€๐‘›.

Corollary 3.5

There are infinitely many primes.

Proof:

Suppose there is only a finite number of primes, ๐‘1 ,๐‘2 ,โ€ฆ,and ๐‘๐‘˜ .

Then consider the Fibonacci numbers ๐น๐‘1,๐น๐‘2

,โ€ฆ,and ๐น๐‘๐‘˜ . Clearly, they are pair

wise relatively prime.

Since there are only ๐‘˜ primes, each of these Fibonacci numbers has exactly

one prime factor, that is, each is a prime.

This is a contradiction, since ๐น19 = 4181 = 37 ยท 113.

Thus our assumption that there are only finitely many primes is false.

In other words, thus there are infinitely many primes.

A quick look at Lucas numbers shows that every third Lucas number is even,

that is, 2|๐ฟ3๐‘› . This is, in fact, always true.

Theorem 3.4

๐ฟ๐‘š |๐น๐‘› if and only if 2๐‘š|๐‘›, where ๐‘š โ‰ฅ 2.

For example,

12|24, so ๐ฟ6|๐น24 ; 18|46368

Theorem 3.5

๐ฟ๐‘š |๐ฟ๐‘› if and only if ๐‘› = 2๐‘˜ โˆ’ 1 ๐‘š, where ๐‘š โ‰ฅ 2 and ๐‘˜ โ‰ฅ 1.

For example,

Let ๐‘š = 3 , and ๐‘› = 5 ยท 3 = 15. We have ๐ฟ3 = 4 and ๐ฟ15 = 1364. Clearly,

๐ฟ3|๐ฟ15 .

Theorem 3.6 (Freeman, 1967)

1. Let ๐‘Ž: ๐‘ = ๐น๐‘› :๐น๐‘›+1. Then ๐‘Ž + ๐‘ ๐น๐‘›โˆ’1 = ๐‘Ž, ๐‘ + โˆ’1 ๐‘› ๐‘Ž, ๐‘ ,where

๐‘› โ‰ฅ 2.

2. Let (๐‘,๐‘‘) = 1 such that ๐‘Ž: ๐‘ = ๐‘:๐‘‘. Let (๐‘Ž + ๐‘)๐น๐‘›โˆ’1 = [๐‘Ž, ๐‘] +

โˆ’1 ๐‘› ๐‘Ž, ๐‘ , where ๐‘› โ‰ฅ 3. Then the number of solutions of the ratio ๐‘:๐‘‘

equals one-half the number of positive factors of ๐น๐‘›๐น๐‘›โˆ’2, one of them being

๐น๐‘›๐น๐‘›+1.

Proof:

1. Let ๐‘Ž: ๐‘ = ๐น๐‘› :๐น๐‘›+1.

Then, since (๐น๐‘› ,๐น๐‘›+1) = 1, ๐‘Ž = ๐น๐‘›๐‘˜ , ๐‘ = ๐น๐‘›+1๐‘˜, ๐‘Ž, ๐‘ = ๐‘˜, ๐‘Ž, ๐‘ =

๐น๐‘›๐น๐‘›+1๐‘˜ for some positive integer ๐‘˜.

โˆด ๐‘Ž + ๐‘ ๐น๐‘›โˆ’1 = ๐น๐‘›โˆ’1 ๐น๐‘› + ๐น๐‘›+1 ๐‘˜ = ๐น๐‘›โˆ’1๐น๐‘›+2๐‘˜

= ๐น๐‘›+1 โˆ’ ๐น๐‘› ๐น๐‘›+2๐‘˜ = ๐น๐‘›+1 ๐น๐‘› + ๐น๐‘›+1 ๐‘˜ โˆ’ ๐น๐‘›๐น๐‘›+2๐‘˜

= ๐น๐‘›๐น๐‘›+1๐‘˜ + ๐น๐‘›+12 โˆ’ ๐น๐‘›๐น๐‘›+2 ๐‘˜

= ๐‘Ž, ๐‘ + โˆ’1 ๐‘› ๐‘Ž, ๐‘ by Cassiniโ€™s

rule

2. Let ๐‘Ž: ๐‘ = ๐‘:๐‘‘, where (๐‘ ,๐‘‘) = 1 .

Then ๐‘Ž = ๐‘๐‘˜, ๐‘ = ๐‘‘๐‘˜, (๐‘Ž, ๐‘) = ๐‘˜ and [๐‘Ž, ๐‘] = ๐‘๐‘‘๐‘˜ for some positive integer ๐‘˜.

Since (๐‘Ž + ๐‘) ๐น๐‘›โˆ’1 = [๐‘Ž, ๐‘] + โˆ’1 ๐‘› (๐‘Ž, ๐‘), we have

๐‘ + ๐‘‘ ๐น๐‘›โˆ’1 = ๐‘๐‘‘ + โˆ’1 ๐‘›

This yields

๐‘ =๐‘‘๐น๐‘›โˆ’1 โˆ’ โˆ’1 ๐‘›

๐‘‘ โˆ’ ๐น๐‘›โˆ’1

= ๐น๐‘›โˆ’1 +๐น๐‘›โˆ’1

2 โˆ’ โˆ’1 ๐‘›

๐‘‘ โˆ’ ๐น๐‘›โˆ’1

= ๐น๐‘›โˆ’1 +๐น๐‘›๐น๐‘›โˆ’2

๐‘‘ โˆ’ ๐น๐‘›โˆ’1 (3.1)

If 0 < ๐‘‘ < ๐น๐‘›โˆ’1, then ๐‘ < 0; so ๐‘‘ > ๐น๐‘›โˆ’1.

Since ๐‘ is an integer, ๐‘‘ โˆ’ ๐น๐‘›โˆ’1|๐น๐‘›๐น๐‘›โˆ’2.

Thus Eq. (3.1) yields a value of ๐‘ for every positive factor of ๐น๐‘›๐น๐‘›โˆ’2.

But, if ๐‘ = ๐ด, ๐‘‘ = ๐ต is a solution of the ratio ๐‘:๐‘‘, then so is ๐‘ = ๐ต,๐‘‘ = ๐ด.

Thus the number of distinct values of the ratio ๐‘:๐‘‘ equals the number of

p

positive factors of ๐น๐‘›๐น๐‘›โˆ’2.

In particular, let ๐‘‘ = ๐น๐‘›+1. Then

๐‘ = ๐น๐‘›โˆ’1 +๐น๐‘›๐น๐‘›โˆ’2

๐น๐‘›+1 โˆ’ ๐น๐‘›โˆ’1

= ๐น๐‘›โˆ’1 +๐น๐‘›๐น๐‘›โˆ’2

๐น๐‘›= ๐น๐‘›

Thus ๐‘:๐‘‘ = ๐น๐‘› :๐น๐‘›+1 is also a value of the ratio.

Example 3.2

1. Let ๐‘Ž: ๐‘ = ๐น11 :๐น12 = 89: 144 , so ๐‘› = 11. Let ๐‘Ž = 445,๐‘ = 720, so

๐‘Ž: ๐‘ = 89: 144

๐‘Ž, ๐‘ + โˆ’1 ๐‘› ๐‘Ž, ๐‘ = 64080 โˆ’ 5 = 64075 = 445 + 720 ยท 55 = ๐‘Ž + ๐‘ ๐น10

2. Since (๐‘Ž + ๐‘)๐น10 = [๐‘Ž, ๐‘] + โˆ’1 11 (๐‘Ž, ๐‘), it follows that

๐‘ = ๐น10 + ๐น11๐น9

๐‘‘โˆ’๐น10

= 55 +89 ยท 34

๐‘‘ โˆ’ 55= 55 +

3026

๐‘‘ โˆ’ 55

Since 3026 = 2 ยท 17 ยท 89, 3026 has eight positive factors: 1, 2, 17, 34,89, 178, 1513 and 3026. So ๐‘‘ has eight possible values: 56, 57, 72,89, 144, 233, 1568 and 3081. Consequently, the various values of ๐‘:๐‘‘ are

3081: 56, 1568: 57, 233: 72, 144: 89, 89: 144, 72: 233, 57: 1568 and 56: 3081. Since one-half of them are duplicates, the four distinct values of

๐‘:๐‘‘ are 89: 144, 72: 233, 57: 1568 and 56: 3081, keeping the numerator to

be smaller. Notice that one of the ratios is 89: 144 = ๐น11 :๐น12, as expected.

Lemma 3.3

๐น2๐‘›โˆ’1 = ๐น๐‘›+1๐ฟ๐‘›+2 โˆ’ ๐ฟ๐‘›๐ฟ๐‘›+1 , ๐‘› โ‰ฅ 2.

Theorem 3.7

1. Let ๐‘Ž: ๐‘ = ๐ฟ๐‘› : ๐ฟ๐‘›+1. Then ๐‘Ž + ๐‘ ๐น๐‘›+1 = ๐‘Ž, ๐‘ + ๐‘Ž, ๐‘ ๐น2๐‘›โˆ’1, ๐‘› โ‰ฅ 2.

2. Let ๐‘Ž: ๐‘ = ๐น๐‘›โˆ’2:๐น๐‘›โˆ’1. Then ๐‘Ž + ๐‘ ๐น๐‘›+1 = ๐‘Ž, ๐‘ + ๐‘Ž, ๐‘ ๐น2๐‘›โˆ’1, ๐‘› โ‰ฅ 3.

3. Let (๐‘,๐‘‘) = 1 such that ๐‘Ž: ๐‘ = ๐‘:๐‘‘. If ๐‘Ž + ๐‘ ๐น๐‘›+1 = ๐‘Ž, ๐‘ + ๐‘Ž, ๐‘ ๐น2๐‘›โˆ’1

where ๐‘› โ‰ฅ 2, then the ratios ๐‘:๐‘‘ are determined by the positive factors of

๐น๐‘›+12 โˆ’ ๐น2๐‘›โˆ’1, one of them being ๐ฟ๐‘› : ๐ฟ๐‘›+1. For ๐‘› โ‰ฅ 3, ๐น๐‘›โˆ’2:๐น๐‘›โˆ’1 is also a

solution.

Proof:

1. Let ๐‘Ž: ๐‘ = ๐ฟ๐‘› : ๐ฟ๐‘›+1. Since ๐ฟ๐‘› , ๐ฟ๐‘›+1 = 1, ๐‘Ž = ๐‘˜๐ฟ๐‘› , ๐‘ = ๐‘˜๐ฟ๐‘›+1, ๐‘Ž, ๐‘ =

๐‘˜, and ๐‘Ž, ๐‘ = ๐ฟ๐‘›๐ฟ๐‘›+1๐‘˜ for some positive integer ๐‘˜. Then

(๐‘Ž + ๐‘)๐น๐‘›+1 = (๐ฟ๐‘› + ๐ฟ๐‘›+1)๐‘˜๐น๐‘›+1 = ๐น๐‘›+1๐ฟ๐‘›+2๐‘˜

= (๐น2๐‘›โˆ’1 + ๐ฟ๐‘›๐ฟ๐‘›+1)๐‘˜

= [๐‘Ž, ๐‘] + (๐‘Ž, ๐‘)๐น2๐‘›โˆ’1

as desired.

2. Suppose ๐‘Ž โˆถ ๐‘ = ๐น๐‘›โˆ’2:๐น๐‘›โˆ’1. Then = ๐‘˜๐น๐‘›โˆ’2, ๐‘ = ๐‘˜๐น๐‘›โˆ’1, (๐‘Ž , ๐‘) = ๐‘˜ , and

[๐‘Ž, ๐‘] = ๐น๐‘›โˆ’1๐น๐‘›โˆ’2๐‘˜ for some positive integer ๐‘˜. Then

(๐‘Ž, ๐‘)๐น๐‘›+1 = (๐น๐‘›โˆ’2 + ๐น๐‘›โˆ’1)๐‘˜๐น๐‘›+1 = ๐น๐‘›๐น๐‘›+1๐‘˜

= ๐น2๐‘›โˆ’1 + ๐น๐‘›โˆ’1๐น๐‘›โˆ’2 ๐‘˜

Since ๐น2๐‘›โˆ’1 = ๐น๐‘›๐น๐‘›+1 โˆ’ ๐น๐‘›โˆ’2๐น๐‘›โˆ’1

= [ ๐‘Ž, ๐‘ ] + ( ๐‘Ž , ๐‘ ) ๐น2๐‘›โˆ’1

again as desired.

3. Let ๐‘Ž โˆถ ๐‘ = ๐‘ โˆถ ๐‘‘, where ๐‘ ,๐‘‘ = 1. As before, ๐‘Ž = ๐‘๐‘˜ , ๐‘ = ๐‘‘๐‘˜, ๐‘Ž , ๐‘ = ๐‘˜, and [ ๐‘Ž , ๐‘ ] = ๐‘๐‘‘๐‘˜ for some positive integer ๐‘˜.

Since ๐‘Ž + ๐‘ ๐น๐‘›+1 = ๐‘Ž, ๐‘ + ๐‘Ž , ๐‘ ๐น2๐‘›โˆ’1

๐‘ + ๐‘‘ ๐น๐‘›+1 = ๐‘๐‘‘ + ๐น2๐‘›โˆ’1

๐‘ =๐‘‘๐น๐‘›+1 โˆ’ ๐น2๐‘›โˆ’1

๐‘‘ โˆ’ ๐น๐‘›+1 (3.2)

= ๐น๐‘›+1 + ๐น๐‘›+1

2 โˆ’๐น2๐‘›โˆ’1

๐‘‘โˆ’๐น๐‘›โˆ’1

Since ๐‘ and ๐‘‘ are positive integers, it follows that the ratio ๐‘:๐‘‘ is

determined by the positive factors of ๐น๐‘›+12 โˆ’ ๐น2๐‘›โˆ’1. d

In particular, let ๐‘‘ = ๐น๐‘›+1. Then by Lemma 3.3,

๐‘ = ๐น๐‘›+1 + ๐น๐‘›+1

2 โˆ’๐น๐‘›+1๐ฟ๐‘›+2+๐ฟ๐‘›๐ฟ๐‘›+1

๐ฟ๐‘›+1โˆ’๐น๐‘›+1

= ๐น๐‘›+1 ๐ฟ๐‘›+1โˆ’๐ฟ๐‘›+2 +๐ฟ๐‘›๐ฟ๐‘›+1

๐ฟ๐‘›+1โˆ’ ๐น๐‘›+1

= ๐ฟ๐‘›๐ฟ๐‘›+1โˆ’๐ฟ๐‘›๐น๐‘›+1

๐ฟ๐‘›+1โˆ’๐น๐‘›+1 = ๐ฟ๐‘›

Thus ๐ฟ๐‘› : ๐ฟ๐‘›+1 is a solution of the ratio ๐‘:๐‘‘. ( By symmetry, ๐ฟ๐‘›+1: ๐ฟ๐‘› is also

a solution).

Unlike Theorem 3.7, not all solutions are obtained by considering the case

๐‘‘ > ๐น๐‘›+1.. For instance, let ๐‘‘ = ๐น๐‘›โˆ’1.Then by Eq.(3.2).

๐‘ =๐น๐‘›โˆ’1๐น๐‘›+1 โˆ’ ๐น2๐‘›โˆ’1

๐น๐‘›โˆ’1 โˆ’ ๐น๐‘›+1

= ๐น๐‘›โˆ’1๐น๐‘›+1โˆ’ ๐น๐‘›๐น๐‘›+1โˆ’ ๐น๐‘›โˆ’2๐น๐‘›โˆ’1

โˆ’๐น๐‘›

=โˆ’๐น๐‘›+1 ๐น๐‘›โˆ’1โˆ’ ๐น๐‘› + ๐น๐‘›โˆ’2๐น๐‘›โˆ’1

๐น๐‘›

=โˆ’๐น๐‘›โˆ’2๐น๐‘›+1+ ๐น๐‘›โˆ’2๐น๐‘›โˆ’1

๐น๐‘›

= = ๐น๐‘›โˆ’2(๐น๐‘›โˆ’1โˆ’๐น๐‘›+1)

๐น๐‘› = ๐น๐‘›โˆ’2

Thus ๐น๐‘›โˆ’2:๐น๐‘›โˆ’1 is also a solution of the ratio.

Example 3.3.

Let ๐‘› = 11. We have ๐น๐‘›+1 = ๐น12 = 144 and ๐น2๐‘›โˆ’1 = ๐น21 = 10946.

1. Let ๐‘Ž: ๐‘ = ๐ฟ๐‘› : ๐ฟ๐‘›+1 = ๐ฟ11 : ๐ฟ12 = 199: 322. Let ๐‘Ž = 597 and ๐‘ = 966.

Then

๐‘Ž, ๐‘ + ๐‘Ž, ๐‘ ๐น2๐‘›โˆ’1 = 597, 966 + 597, 966 ยท 10946

= 192234 + 3 ยท 10946 = 225072

= 597 + 966 ยท 144

= (๐‘Ž + ๐‘)๐น๐‘›+1

2. Let ๐‘Ž: ๐‘ = ๐น๐‘›โˆ’2:๐น๐‘›โˆ’1 = ๐น9:๐น10 = 34: 55. Let ๐‘Ž = 272 and ๐‘ = 440.

Then ๐‘Ž, ๐‘ + ๐‘Ž, ๐‘ ๐น2๐‘›โˆ’1 = 272 ,440 + 8 ยท 10946

= 102528 = 272 + 440 ยท 144

= ๐‘Ž + ๐‘ ๐น๐‘›+1

3. Let ๐‘Ž: ๐‘ = 208: 240 = 13: 15, where ๐‘:๐‘‘ = 13: 15 and 15, 17 = 1.

Then ๐‘ = ๐น๐‘›+1 + ๐น๐‘›+1

2 โˆ’๐น2๐‘›โˆ’1

๐‘‘โˆ’๐น๐‘›+1

= 144 + 1442โˆ’10,946

๐‘‘โˆ’144

= 144 + 9790

๐‘‘โˆ’144

Since 9790 = 2 ยท 5 ยท 11 ยท 89, 9790 has 16 positive factors: 1, 2, 5, 10, 11,

22, 55, 89, 110, 178, 445, 890, 979, 1958, 4895 and 9790.The

corresponding ratios are 145: 9934, 146: 5039, 149: 2102, 154: 1123,

155: 1034, 166: 589, 199: 322, 233: 254, 254: 233, 322: 199, 589: 166,

1034: 155, 1123: 154, 2102: 149, 5039: 146 and 9934: 145. These yields 8

distinct ratios ๐‘:๐‘‘ with (๐‘, ๐‘‘) = 1,namely, 145:9934, 146:5039, 149:2102,

154:1123, 155:1034, 166:589, 199:322 and 233:254. Notice that 34: 55 is

also a solution. Among these ratios we find ๐ฟ11 : ๐ฟ12 = 199: 322 and

๐น9:๐น10 = 34: 55 as expected.

AN ALTERNATE FIBONACCI SEQUENCE

In 1971, Underwood Dudley and Bessie Tucker of DePauw University in Indiana

investigated a slightly altered Fibonacci sequence, defined by ๐บ๐‘› = ๐น๐‘› + โˆ’1 ๐‘› ,

where ๐‘› โ‰ฅ 1. They made an interesting observation ,as table 2.1 shows : The

1st,3

rd,5

th,โ€ฆentries (see the circled numbers) in the (๐บ๐‘› ,๐บ๐‘›+1)-row are the 2

nd , 4

th,

6th

,โ€ฆ. Fibonacci numbers; and the 2nd

, 4th

, 6th

,โ€ฆ.. entries are the 3rd

, 5th , 7

th,โ€ฆ..

Lucas numbers.

TABLE 3.1.

To establish these two results, we need the following theorem.

Theorem 3.8 (Dudley and Tucker, 1971)

(1) ๐น4๐‘› + 1 = ๐น2๐‘›โˆ’1๐ฟ2๐‘›+1 (2) ๐น4๐‘› โˆ’ 1 = ๐น2๐‘›+1๐ฟ2๐‘›โˆ’1

(3) ๐น4๐‘›+1 + 1 = ๐น2๐‘›+1๐ฟ2๐‘› (4) ๐น4๐‘›+1 โˆ’ 1 = ๐น2๐‘›๐ฟ2๐‘›+1

(5) ๐น4๐‘›+2 + 1 = ๐น2๐‘›+2๐ฟ2๐‘› (6) ๐น4๐‘›+2 โˆ’ 1 = ๐น2๐‘›๐ฟ2๐‘›+2

(7) ๐น4๐‘›+3 + 1 = ๐น2๐‘›+1๐ฟ2๐‘›+2 (8) ๐น4๐‘›+3 โˆ’ 1 = ๐น2๐‘›+2๐ฟ2๐‘›+1

Proof:

The proof requires the following identities:

๐น๐‘š+๐‘› + ๐น๐‘šโˆ’๐‘› = ๐น๐‘›๐ฟ๐‘š if ๐‘› is odd๐น๐‘š๐ฟ๐‘› otherwise

๐น๐‘š+๐‘› โˆ’ ๐น๐‘šโˆ’๐‘› = ๐น๐‘š๐ฟ๐‘› if ๐‘› is odd ๐น๐‘›๐ฟ๐‘š otherwise

Then

๐น4๐‘› + 1 = ๐น4๐‘› + ๐น2 = ๐น 2๐‘›+1 + 2๐‘›โˆ’1 + ๐น 2๐‘›+1 โˆ’ 2๐‘›โˆ’1

= ๐น2๐‘›โˆ’1๐ฟ2๐‘›+1

and

๐น4๐‘›+1 + 1 = ๐น4๐‘›+1 + ๐น1 = ๐น 2๐‘›+1 +2๐‘› + ๐น 2๐‘›+1 โˆ’2๐‘›

= ๐น2๐‘›+1๐ฟ2๐‘›

The other formulas can be established similarly.

The following corollary, observed in 1971 by Hoggatt, follows easily from this

theorem.

Corollary 3.6

(1) ๐น4๐‘›+1 + 1, ๐น4๐‘›+2 + 1 = ๐ฟ2๐‘› (2) ๐น4๐‘›+1 + 1,๐น4๐‘›+3 + 1 = ๐น2๐‘›+1

(3) ๐น4๐‘›+1 โˆ’ 1, ๐น4๐‘›+2 โˆ’ 1 = ๐น2๐‘› (4) ๐น4๐‘›+1 โˆ’ 1,๐น4๐‘›+3 โˆ’ 1 = ๐ฟ2๐‘›+1

(5) ๐น4๐‘›โˆ’1 โˆ’ 1, ๐น4๐‘›+1 + 1 = ๐น2๐‘› (6) ๐น4๐‘›โˆ’1 + 1,๐น4๐‘›+1 + 1 = ๐ฟ2๐‘›

(7) ๐น4๐‘›+3 + 1, ๐น4๐‘› โˆ’ 1 = ๐น2๐‘›+1 (8) ๐น4๐‘›+3 + 1,๐น4๐‘›+2 โˆ’ 1 = ๐น2๐‘›

(9) ๐น4๐‘›+4 โˆ’ 1,๐น4๐‘›+3 โˆ’ 1 = ๐ฟ2๐‘›+1

Although it is not yet known whether or not the Fibonacci sequence contains

infinitely many primes, this theorem establishes their finiteness in the sequences

๐น๐‘› + 1 and ๐น๐‘› โˆ’ 1 , as the next corollary shows.

Corollary 3.7

๐น๐‘› + 1 is composite if ๐‘› โ‰ฅ 4, and ๐น๐‘› โˆ’ 1 is composite if ๐‘› โ‰ฅ 7.

Proof:

When ๐‘› = 1, ๐น4๐‘›+ 1 = 4 is composite. When ๐‘› โ‰ฅ 2, it follows from Theorem 2.9

that ๐น4๐‘›+1 + 1, ๐น4๐‘›+2 + 1, and ๐น4๐‘›+3 + 1 have nontrivial factors. Thus ๐น๐‘› + 1 is

composite if ๐‘› โ‰ฅ 4. Likewise, ๐น๐‘› โˆ’ 1 is composite if ๐‘› โ‰ฅ 7.

Notice that ๐น๐‘› + 1 is a prime if ๐‘› < 4 and ๐น๐‘› โˆ’ 1 is a prime if ๐‘› < 7.

Corollary 3.8

๐บ4๐‘› ,๐บ4๐‘›+1 = ๐ฟ2๐‘›+1 , ๐บ4๐‘›+1,๐บ4๐‘›+3 = ๐ฟ2๐‘›+1, and ๐บ4๐‘›+2,๐บ4๐‘›+3 = ๐น2๐‘›+2

where ๐‘› โ‰ฅ 1

Proof:

By theorem 3.9,

๐บ4๐‘› ,๐บ4๐‘›+1 = ๐น4๐‘›+1 + 1,๐น4๐‘›+1 โˆ’ 1

= ๐น2๐‘›โˆ’1๐ฟ2๐‘›+1 ,๐น2๐‘›๐ฟ2๐‘›+1

= ๐ฟ2๐‘›+1 ๐น2๐‘›โˆ’1,๐น2๐‘›

= ๐ฟ2๐‘›+1

Similarly, we can prove the other two parts.

Corollary 3.9

Let ๐ป๐‘› = ๐น๐‘› โˆ’ โˆ’1 ๐‘› . Then ๐ป4๐‘› ,๐ป4๐‘›+1 = ๐น2๐‘›+1 , ๐ป4๐‘›+1,๐ป4๐‘›+3 = ๐น2๐‘›+1 and

๐ป4๐‘›+2,๐ป4๐‘›+3 = ๐ฟ2๐‘›+2, where ๐‘› โ‰ฅ 1.

CHAPTER โ€“ 4

GENERALIZATION

We can study properties common to Fibonacci and Lucas numbers by

investigating a number sequence that satisfies the Fibonacci recurrence relation,

but with arbitrary initial conditions.

GENERALIZED FIBONACCI NUMBERS

Consider the sequence ๐บ๐‘› , ๐บ1 = a, ๐บ๐‘› = b, and ๐บ๐‘› = ๐บ๐‘›โˆ’1 + ๐บ๐‘›โˆ’2, ๐‘› โ‰ฅ 3. The

ensuing sequence

๐‘Ž, ๐‘, ๐‘Ž + ๐‘,๐‘Ž + 2๐‘, 2๐‘Ž + 3๐‘, 3๐‘Ž + 5๐‘, . ..

is called the generalized Fibonacci sequence (GFS).

Take a close look at the coefficients of ๐‘Ž and ๐‘ in the various terms of this

sequence. They follow an interesting pattern: The coefficients of ๐‘Ž and ๐‘ are

Fibonacci numbers. In fact, we can pinpoint these two Fibonacci coefficients, as

the following theorem shows.

Theorem 4.1

Let ๐บ๐‘› denote the nth term of the GFS. Then ๐บ๐‘› = ๐‘Ž๐น๐‘›โˆ’2 + ๐‘๐น๐‘›โˆ’1, ๐‘› โ‰ฅ 3.

Proof:

Since ๐บ3 = ๐‘Ž + ๐‘ = ๐‘Ž๐น1 + ๐‘๐น2 , the statement is true when ๐‘› = 3.

Let k be an arbitrary integer โ‰ฅ 3. Assume the given statement is true for all

integers ๐‘–, where 3 โ‰ค ๐‘– โ‰ค ๐‘˜: ๐บ๐‘– = ๐‘Ž๐น๐‘–โˆ’2 + ๐‘๐น๐‘–โˆ’1. Then:

๐บ๐‘˜+1 = ๐บ๐‘˜ + ๐บ๐‘˜โˆ’1

= ๐‘Ž๐น๐‘˜โˆ’2 + ๐‘๐น๐‘˜โˆ’1 + ๐‘Ž๐น๐‘˜โˆ’3 + ๐‘๐น๐‘˜โˆ’2

= ๐‘Ž ๐น๐‘˜โˆ’2 + ๐น๐‘˜โˆ’3 + ๐‘ ๐น๐‘˜โˆ’1 + ๐‘๐น๐‘˜โˆ’2

= ๐‘Ž๐น๐‘˜โˆ’1 + ๐‘๐น๐‘˜

Thus, by the principle of mathematical induction the formula holds for every

integer ๐‘› โ‰ฅ 3.

Notice that this theorem is in fact true for all ๐‘› โ‰ฅ 1.

Theorem 4.2

๐บ๐‘˜+๐‘–

๐‘›

๐‘–=1

= ๐บ๐‘›+๐‘˜+2 โˆ’ ๐บ๐‘˜+2

Proof:

By Theorem 4.1,

๐บ๐‘˜+๐‘–

๐‘›

๐‘–=1

= ๐‘Ž ๐น๐‘˜+๐‘–โˆ’2

๐‘›

๐‘–=1

+ ๐‘ ๐น๐‘˜+๐‘–โˆ’1

๐‘›

๐‘–=1

= ๐‘Ž ๐น๐‘›+๐‘˜ โˆ’ ๐น๐‘˜ + ๐‘ ๐น๐‘›+๐‘˜+1 โˆ’ ๐น๐‘˜+1

= ๐‘Ž๐น๐‘›+๐‘˜ + ๐‘๐น๐‘›+๐‘˜+1 โˆ’ ๐‘Ž๐น๐‘˜ + ๐‘๐น๐‘˜+1

= ๐บ๐‘›+๐‘˜+2 โˆ’ ๐บ๐‘˜+2

Theorem 4.3 (Koshy, 1998)

๐บ๐‘–๐บ๐‘–+1

๐‘›

๐‘–=1

= ๐‘Ž2 ๐น๐‘›โˆ’22 โˆ’ ๐‘ฃ + ๐‘2 ๐น๐‘›โˆ’1

2 โˆ’ ๐‘ฃ + 1

+ ๐‘Ž๐‘ ๐ฟ2๐‘›โˆ’1 + 5๐น๐‘›โˆ’1๐น๐‘› + ๐‘ฃ + 1 /5

where ๐‘ฃ = 1 if ๐‘› is odd 0 otherwise

Theorem 4.4 (Binet's formula).

Let ๐‘ = ๐‘Ž + ๐‘Ž โˆ’ ๐‘ ๐›ฝ and ๐‘‘ = ๐‘Ž + (๐‘Ž โˆ’ ๐‘)๐›ผ. Then

๐บ๐‘› =๐‘๐›ผ๐‘› โˆ’ ๐‘‘๐›ฝ๐‘›

๐›ผ โˆ’ ๐›ฝ

Proof:

By Theorem 4.1,

๐บ๐‘› = ๐‘Ž๐น๐‘›โˆ’2 + ๐‘๐น๐‘›โˆ’1

5๐บ๐‘› = ๐‘Ž ๐›ผ๐‘›โˆ’2 โˆ’ ๐›ฝ๐‘›โˆ’2 + ๐‘ ๐›ผ๐‘›โˆ’1 โˆ’ ๐›ฝ๐‘›โˆ’1

= ๐›ผ๐‘› ๐‘Ž

๐›ผ2+

๐‘

๐›ผ โˆ’ ๐›ฝ๐‘›

๐‘Ž

๐›ฝ2+

๐‘

๐›ฝ

= ๐›ผ๐‘› ๐‘Ž๐›ฝ2 โˆ’ ๐‘๐›ฝ โˆ’ ๐›ฝ๐‘› ๐‘Ž๐›ผ2 โˆ’ ๐‘๐›ผ

= ๐›ผ๐‘› ๐‘Ž + ๐‘Ž โˆ’ ๐‘ ๐›ฝ โˆ’ ๐›ฝ๐‘› ๐‘Ž + ๐‘Ž โˆ’ ๐‘ ๐›ผ

โˆด ๐บ๐‘› =๐‘๐›ผ๐‘› โˆ’ ๐‘‘๐›ฝ๐‘›

๐›ผ โˆ’ ๐›ฝ

as desired.

Notice that

๐‘๐‘‘ = [๐‘Ž + (๐‘Ž โˆ’ ๐‘)๐›ฝ][๐‘Ž + (๐‘Ž โˆ’ ๐‘)๐›ผ]

= ๐‘Ž2 + (๐‘Ž โˆ’ ๐‘)2๐›ผ๐›ฝ + ๐‘Ž(๐‘Ž โˆ’ ๐‘)(๐›ผ + ๐›ฝ)

= ๐‘Ž2 โˆ’ (๐‘Ž โˆ’ ๐‘)2 + ๐‘Ž(๐‘Ž โˆ’ ๐‘)

= ๐‘Ž2 + ๐‘Ž๐‘ โˆ’ ๐‘2

This constant occurs in many of the formulas for generalized Fibonacci numbers. It

is called the characteristic of the GFS. We denote it by the Greek letter ฮผ (mu):

๐œ‡ = ๐‘Ž2 + ๐‘Ž๐‘ โˆ’ ๐‘2

The characteristic of the Fibonacci sequence is 1, and that of the Lucas sequence is

โˆ’5.

Theorem 4.5

๐บ๐‘›+1๐บ๐‘›โˆ’1 โˆ’ ๐บ๐‘›2 = ๐œ‡ โˆ’1 ๐‘›

Proof:

5 ๐บ๐‘›+1๐บ๐‘›โˆ’1 โˆ’ ๐บ๐‘›2 = ๐‘๐›ผ๐‘›+1 โˆ’ ๐‘‘๐›ฝ๐‘›+1 ๐‘๐›ผ๐‘›โˆ’1 โˆ’ ๐‘‘๐›ฝ๐‘›โˆ’1 โˆ’ ๐‘๐›ผ๐‘› โˆ’ ๐‘‘๐›ฝ๐‘› 2

= โˆ’๐‘๐‘‘ ๐›ผ๐‘›+1๐›ฝ๐‘›โˆ’1 + ๐›ผ๐‘›โˆ’1๐›ฝ๐‘›+1 + 2๐‘๐‘‘ ๐›ผ๐›ฝ ๐‘›

= โˆ’๐œ‡ ๐›ผ๐›ฝ ๐‘›โˆ’1 ๐›ผ2 + ๐›ฝ2 + 2๐œ‡ ๐›ผ๐›ฝ ๐‘›

= 5๐œ‡ โˆ’1 ๐‘›

Therefore, ๐บ๐‘›+1๐บ๐‘›โˆ’1 โˆ’ ๐บ๐‘›2 = ๐œ‡ โˆ’1 ๐‘› .

In particular, ๐ฟ๐‘›+1๐ฟ๐‘›โˆ’1 โˆ’ ๐ฟ๐‘›2 = 5 โˆ’1 ๐‘›โˆ’1.

Theorem 4.6

Let ๐ด๐ต๐ถ be a triangle with AC = ๐บ๐‘˜๐บ๐‘˜+3, BC=2๐บ๐‘˜+1๐บ๐‘˜+2 and AB = ๐บ2๐‘˜+3. Then

โˆ†ABC is a right triangle with hypotenuse AB.

In Chapter 2, we found that the sum of any 10 consecutive Fibonacci

numbers is 11 times the seventh number in the sequence. Also, notice that the first

10 terms of the generalized Fibonacci sequence are , ๐‘, ๐‘Ž + ๐‘,๐‘Ž + 2๐‘, 2๐‘Ž +

3๐‘, 3๐‘Ž + 5๐‘, 5๐‘Ž + 8๐‘, 8๐‘Ž + 13๐‘, 13๐‘Ž + 21๐‘, and 21๐‘Ž + 34๐‘. Their sum is

55๐‘Ž + 88๐‘, which is clearly 11 times the seventh term 5๐‘Ž + 8๐‘. Interestingly

enough, 11 = ๐ฟ5. Thus

Gi

10

1

= ๐ฟ5 . G7

where ๐ฟ5 = (55,89 โ€“ 1) = (๐น10 ,๐น11 โˆ’ 1)

When ๐‘› = 10, this sum is divisible by ๐ฟ5, as we just observed. Consequently,

let us look for a way to factor this sum. Since ๐‘Ž and ๐‘ are arbitrary, we look for the

common factors of ๐น๐‘› and ๐น๐‘›+1 โˆ’ 1 . [Although (๐น๐‘› ,๐น๐‘›+1) = 1,๐น๐‘› and ๐น๐‘›+1 โˆ’ 1

need not be relatively prime.]

Table 4.1 shows a few specific values of ๐น๐‘› ,๐น๐‘›+1 โˆ’ 1 , and their factorizations; we

have omitted the cases where (๐น๐‘› ,๐น๐‘›+1 โˆ’ 1) = 1.

TABLE 4.1

It is apparent from the table that when n is of the form 4k + 2, (๐น๐‘› ,๐น๐‘›+1 โˆ’ 1) is

a Lucas number and the various quotients are consecutive Fibonacci numbers; and

when n is of the form 4k, (๐น๐‘› ,๐น๐‘›+1 โˆ’ 1) is a Fibonacci number and the various

quotients are consecutive Lucas numbers.

Next we proceed to confirm these two observations, for which we need the

following facts from Theorem 2.9:

๐น4๐‘›+1 โˆ’ 1 = ๐ฟ2๐‘›+1๐น2๐‘› and ๐น4๐‘›+2 โˆ’ 1 = ๐ฟ2๐‘›๐น2๐‘›+2

Case 1. Let n be of the form 4k + 2.Then

๐บ๐‘–

4๐‘˜+2

1

= ๐‘Ž๐น4๐‘˜+2 + ๐‘(๐น4๐‘˜+3 โˆ’ 1)

= ๐‘Ž๐ฟ2๐‘˜+1๐น2๐‘˜+1 + ๐‘๐ฟ2๐‘˜+1๐น2๐‘˜+2

= ๐ฟ2๐‘˜+1(๐‘Ž๐น2๐‘˜+2 + ๐‘๐น2๐‘˜+2)

= ๐ฟ2๐‘˜+1๐บ2๐‘˜+3

Thus ๐บ๐‘–

4๐‘˜+2

1

can be obtained by multiplying ๐บ2๐‘˜+3 with ๐ฟ2๐‘˜+1

In particular, ๐บ๐‘–

10

1

= ๐ฟ5 .๐บ7 = 11 . ๐บ7

as observed earlier. This is an interesting case, since multiplication by 11 is

remarkably easy. Likewise,

we can compute ๐บ๐‘–

30

1

by multiplying ๐บ17 with ๐ฟ15 = 1364

Case 2. Let n be of the form 4๐‘˜. Then

๐บ๐‘–

4๐‘˜

1

= ๐‘Ž๐น4๐‘˜ + ๐‘ ๐น4๐‘˜+1 โˆ’ 1

= ๐‘Ž๐ฟ2๐‘˜๐น2๐‘˜ + ๐‘๐ฟ2๐‘˜+1๐น2๐‘˜

= ๐น2๐‘˜ ๐‘Ž๐ฟ2๐‘˜ + ๐‘๐ฟ2๐‘˜+1

= ๐น2๐‘˜ ๐‘Ž ๐น2๐‘˜โˆ’1 + ๐น2๐‘˜+1 + ๐‘ ๐น2๐‘˜ + ๐น2๐‘˜+2

= ๐น2๐‘˜ ๐‘Ž๐น2๐‘˜โˆ’1 + ๐‘๐น2๐‘˜ + ๐‘Ž๐น2๐‘˜+1 + ๐‘๐น2๐‘˜+2

= ๐น2๐‘˜ ๐บ2๐‘˜+1 + ๐บ2๐‘˜+3

Thus we can realize ๐บ๐‘–

4๐‘˜

1

by multiplying the sum ๐บ2๐‘˜+1 + ๐บ2๐‘˜+3 with ๐น2๐‘˜ .

For instance, we can obtain ๐บ๐‘–

4๐‘˜

1

by multiplying the sum ๐บ11 + ๐บ13 with 55.

REFERENCES

1. Thomas Koshy , โ€œFibonacci and Lucas Numbers with Applicationsโ€, A

Wiley-Interscience Publication

2. Apostol T.M., โ€œIntroduction to Analytic Number Theoryโ€ , Springer

International Student Edition, Narosa Publishing House (1989).

3. Burton D.M. โ€œElementary Number Theoryโ€ , Tata McGraw-Hill

Edition, Sixth Edition (2006).