0% found this document useful (0 votes)
52 views11 pages

A Generalization O.F The Connection Between The Fibonacci

Uploaded by

RaskoJovanovic
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
52 views11 pages

A Generalization O.F The Connection Between The Fibonacci

Uploaded by

RaskoJovanovic
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 11

A GENERALIZATION O.

F THE CONNECTION BETWEEN THE FIBONACCI


SEQUENCE AND PASCAL' S TRIANGLE
JOSEPH A. RAAB, WISCONSIN STATE COLLEGE

Before the main point of this paper can be developed, it is necessary to


review some elementary facts about the Fibonacci Sequence and Pascal's
triangle.
It is well-known that rectangles exist such that if a full-width square is
cut from one end, the remaining part has the same proportions as the original
rectangle.
K-

x -

>(

Assuming width to be unity and length x9 we have


1
x

x - 1

x - 1 = 0

(1)

The greatest root of (1) is the number <p, called the Golden Ratio, and the
rectangle defined is the Golden Rectangle of Greek geometry.

Each root of (1)

has the property that its reciprocal is itself diminished by 1, so that

cp

CD

21

A GENERALIZATION O F THE CONNECTION BETWEEN

22

Given any two. i n i t i a l

[Oct.

i n t e g r a l t e r m s Uj and u 2 not both z e r o f

Fibonacci Sequence i s defined r e c u r s i v e l y by


(2)

n-2

n-1

It i s a well-known p r o p e r t y of such s e q u e n c e s that


lim

If Uj = 0 and u2 = 1,

n+1

<P

we have the Fibonacci sequence.

if a r e c t a n g l e i s defined such that when an i n t e g r a l n u m b e r k of fullwidth s q u a r e s a r e cut from one end, the r e m a i n i n g p a r t has the s a m e p r o p o r tions as the original rectangle, then
1 - 0

ky

(3)

w h e r e 'the width i s unity and the length Is y.


\<r-~~~^

-9H

*|y_k

T
1

1V

\/

The r e c t a n g l e defined i s a golden-type r e c t a n g l e . The r o o t s of (3) behave much


like cps that i s s

l / y = y - k.

The g r e a t e s t r o o t In absolute value of (3) i s the


lim
n-^oo

where u

n+1

= ku ., + u n . In fact.9 It Is well-known that under c e r t a i n conn


n-1
n-2
ditions F i b o n a c c i - l i k e s e q u e n c e s defined by

1963 J

THE FIBONACCI SEQUENCE AND PASCAL'S TRIANGLE

(4)

23

u = au ., + bu 0
n
n-1
n-2

given initial t e r m s ut and u 2 not both z e r o , w h e r e a and b a r e real., have


the p r o p e r t y that
l i m n+1
= a
n-^oo u
n
w h e r e a i s the g r e a t e s t r o o t in absolute value of (See [3] )
x2 - ax - b = 0

(5)

The condition i s that a a n d b m u s t be such that the r o o t s of (5) a r e not both


distinct^ and equal in absolute value,
The above g e n e r a l r e s u l t can be e s t a b l i s h e d i n the following way: Con
th
s i d e r s e q u e n c e s such that the n t e r m u s a t i s f i e s
(6)

= cc/ 1 + d/5 n

By substitution In (4), a a n d (3 can b e d e t e r m i n e d so that s e q u e n c e s (8) will


satisfy (4) and b e F i b o n a c c i - l i k e s e q u e n c e s .
in 9

n?

We find that the coefficients of c

and d a r e a
(a - aa - b) and /3 (/3 - a/3~b) s r e s p e c t i v e l y . Sequences (6),
t h e r e f o r e , satisfy (4) if a and p a r e r o o t s of (5).
n-2 2
On the other hand f if a and /5 a r e r o o t s of (5), then ca
(a - %a - b)
n2
+ d/3
(J32 - a/3 - b) = 0 I s satisfied for any choice of c a n d d0 But then we
have ca + d/3 = a(ca " + d/5 ~ ) + b(ca
+ d/3
). Moreover^ if a ==| fi9
c and d can b e d e t e r m i n e d given initial t e r m s ut a n d u 2 . Hence a s e q u e n c e
satisfying (4) s a t i s f i e s (6) under the conditions stated.
u s e (6) to obtain the
lim

n+1
n

11m ca + d(6/a)
c + d(p/a)

If | a>\ > j p\9 we can

24

A GENERALIZATION OF THE CONNECTION BETWEEN

[Oct.

The above limit does not exist, of course, if a = -jS. If the roots of (5) are
equal 3 then we can set
n
, n
u = ca + nda
1 ;
n
.
and show by arguments similar to those above that (7) is a Fibonacci sequence
(7)

if and only if a is the root of (5) and so>. + 2/3 = 0. But the roots of (5) are
equal if and only if a = a/2 and b = - a 2 / 4 .

Therefore all requirements for

(7) being a Fibonacci sequence are met. It is now possible to solve for c and
d5 and to show that for sequences (7)9
lim

n+1

_ a
n
An interesting observation has been made about the array of numerals
n-*o

known as Pascal*s Triangle. If a particular set of parallel diagonals is designated as in Fig. 1, then the sequence resulting from the individual summations
of the terms of each diagonal is the Fibonacci sequence. [2 ]

Figure 1
Therefore, the limit of quotients of sums of terms on these parallel diagonals
of the triangle is a.

We shall now show that some generalizations of this

connection can be made.

1963]

THE FIBONACCI SEQUENCE AND PASCAL'S TRIANGLE

25

To begin s we note that the indicated diagonal sums in Fig. 2 are indeed
the first few terms (except the first) of (4) if nt = 0 and u2 = 1.

Other sets of parallel diagonals of Fig. 2 also have interesting properties.

It is possible to formalize the definition of the array given as Fig. 25 but

it will be more efficacious here to simply refer informally to the figure in the
arguments to follow.

We will assume only that a and b are real 9 and that

Fig. 2 is a Generalized Pascal's Triangle. The row index shall be j , and the
term index for each row ? 6 5 each ranging over the non - negative integers.
th
th
The j power of (a + b) is the sum of terms in the j row of Fig. 2.
Definition 1. A diagonal sum x. of the generalized Pascal's triangle
shall be given by

Lr+lJ
x.

r5

,1-60*1^

6-0

Counting from left to right in Fig. 2, the (<5 + l)th term of the diagonal sum is
the (6 + l)th term in the (j - r6 )th row of the triangle as 5 ranges over the
non-negative integers.

Hence x. ^ is a function of j and r

Note that the role of r is simply to determine which terms of the t r i angle are to be summed. This has the effect of defining a set of parallel diagonals for each r.

For example 3 if r = 1, the first term of x61 is the first

26

A GENERALIZATION O F THE CONNECTION BETWEEN

t e r m of the sixth row of Fig s 2e

[Oct.

The second t e r m of x 6 1 i s the second t e r m

of the fifth r o w of Fig 9 2 , and so on

If r = 3 9 the f i r s t t e r m of x 6 3 i s t h e

f i r s t t e r m of the sixth r o w of F i g . 2 S but the second t e r m of x 6 3 i s the second


t e r m of the t h i r d row f and so on. When r = 0, x . n i s the s u m of t e r m s on the
JO
.th
3""" row. A sequence {x-j r }. of diagonal s u m s i s uniquely d e t e r m i n e d by r .
J
' J th
Since for j = 0 the (j - r6)
row is defined for e v e r y r only when 6 = 0,
x = 1 for all r . F u r t h e r 9 x- = a if r > 0. If r = 2. the f i r s t f e w t e r m s
9
Or
Ir
>
of the r e s u l t i n g sequence a r e :
(1, a s a 2 ? a 3 + b 5 a 4 + 2ab, a 5 + 3a 2 b, * )
T h e o r e m 1. F o r s e q u e n c e s {x- } of s u m s of t e r m s on p a r a l l e l diagonjr j
als of the g e n e r a l i z e d P a s c a l ? s t r i a n g l e ,
(8)

j-r-1
r + 1 J/j~r(6 + 1 ) - 1

Proof:
.

ax,. n . + bx..
.
0l)r
( j - r -nl ) r

"jr

(j-r-l)r

(j~l)r

dmmmd

5=0

\
j-(5(r+l) b 6

Lr+lJ/ j -r6 - 1
6=1

6 - 1

Lr+1.

j - r6 - 1

j - r6 - 1

J-Hr+Dh6

!a J-S(r + D b 6

j-r6-l
, j-<5(r+l), 6
j
;
aJ v
b + aJ +

a j-<5(r + l) b a

6=1

Lr+1Ji

J+ 6=1
but

'j

- T 6 - 1 \

j-6(r+l)b6

1963]

THE FIBONACCI SEQUENCE AND PASCAL'S TRIANGLE


j - r<5 - 1 \

27

/ j - r<5 \
j -rfi

6-1

and

so

r + l j I / j - r<5
(j-r-l)r

0 - l ) r = a"

rf

(\

jr(5

a /

j , 5 ( r + 1)
j - r(5

a j-6(r+l) b <5

r(5

j^(r+l)b6

jr

5=1

In view of Theorem 1, any property of sequences defined recursively by


un = au n-11 + bu n - r --1

(9)

will be a property of sequences of sums of terms on diagonals of the generalized


Pascal's triangle.

Further s these diagonal sequences will all be of the special

: since r + 1 initial terms


case ut = 0, u2 = 1 3 u3 = a,. s5 u - = a
r+1
are required for (9) to generate a sequence. We note that diagonal sum x
(n-2)r
is u of (9) given the above initial terms.
As In the proof of

28

A GENERALIZATION OF THE CONNECTION BETWEEN


lim
n-*-oo

[Oct.

n+1 _
u
~ ^
n

given (2)3 we shall establish the existence of similar limits for the sequences
defined by (9). If we set
,., -.
(10)

n
n
n
u n = e 0 a 0 + exat + e2a2 + +

n
e ^

then substituting in (9) the coefficients of the e. are

ai

n - r - 1 , r+l
r
, w.
~ .,
,
(a.
- aa . - b ) (i = 0 , 1 , - , r)

and (9) is satisfied if the a. are the r + l

roots of

x r + 1 - ax r - b = 0

(11)

Conversely, given that the a, are the roots of (11), it follows that sequences
(9) can be written in the form of (10) if the e. can be determined. One can obtain from the given (r + 1) initial terms (r + 1) equations u^ = eQal+ ejcq
+ + e or (j = 1, 2, , r + l ) .

This set of equations has a non-trivial

solution for the e. , however, if and only if the a. are distinct. Whether or
not the terms of sequences defined by (9) can be written in the form of (10) depends, therefore, on whether or not we can determine conditions for the multiplicity of the roots of (11).
Suppose p is a root of (11) where a and b are both not zero.

Then

(11) may be written as (x - p) Q (x) = 0 where


^/ x
r
/
x r-1
,
r-2
.
,
, r-1
x
x ? r-3
Q(x) = x + (p - a)x
+ (p - a)px
+ (p - a)p^x
+ + (p - a)p
Clearly p is a multiple root of (11) if and only if it is a root of Q(x) = 0. But
then it is easily verified that

[1963]

THE FIBONACCI SEQUENCE AND PASCAL'S TRIANGLE

29

ar
r + 1
Now s i n c e p i s r e a l , at l e a s t all complex r o o t s of (11) a r e distinct.
DeGua's r u l e for finding i m a g i n a r y r o o t s s t a t e s that when 2m

succes-

sive t e r m s of an equation a r e absent, the equation h a s 2m i m a g i n a r y r o o t s ;


and when 2m - 1 s u c c e s s i v e t e r m s a r e absent, the equation has 2m - 2 o r
2m i m a g i n a r y r o o t s , a c c o r d i n g as the two t e r m s between which the deficiency
o c c u r s have like o r unlike s i g n s .

Accordingly, we s e e that (11) has at m o s t

t h r e e r e a l r o o t s , s i n c e t h e r e a r e r - 1 s u c c e s s i v e absent t e r m s and hence at


r+l
r
l e a s t r - 2 complex r o o t s .

F u r t h e r , if f(x) = x

cal n u m b e r s of f a r e z e r o and a r / ( r + l ) .
of f,

- ax

- b,

the two c r i t i -

Since f(ar/(r + l)) i s an e x t r e m u m

the g r e a t e s t multiplicity of any r e a l r o o t of (11) i s two. [ l ]


If b i s z e r o but a i s not 5 then the r o o t s of (11) a r e z e r o (of multiplicity

r),

and a.

Other c a s e s a r e t r i v i a l .

If the r e a l r o o t s of (11) a r e distinct and # 0 i s any r o o t such that


> jofj (i = 1,2, , r ) ,

lim V i
n-* o o u
n

then

n+1
n+1
+ e
l i m e o^Q
i"i
+ ' - + e
n-*co
n ,
n ,
,
e0a0 + eta t. + +
lim

'

n
e ^

e i i ( i / o ) n + '*'

e 0 + e1(alL/a0)

Therefore

n+1
*

+ +

e a a a
Y T( T/ of

er(a>T/a>0)

It i s c l e a r that a r / ( r + 1) i s a r o o t of (11) if and only if


r+l

b = - * r
(r+Dr+1

\a^

30

A GENERALIZATION O F THE CONNECTION BETWEEN

Suppose a 0 and a^ a r e this r o o t .


/-lov
(12)

Then we can s e t

n ,
n ,
n ,
u n = e 0 a 0 +.ne 1 of 0 + e2a2 + >- +

and u s e (9) to find the coefficients of the e..


nr-T

i =|= 1 i s a.
1

T*+1

(a.

[Oct.

n
e^a^

The coefficient of e.

where

T*

- a a . - b) and for e. we have

'

n-r-1 / r+1

, ^

r
^

^ b ( r + 1) A

It i s c l e a r that the r e q u i r e d condition i s that the a. be the r o o t s of (11) and


r
aq? 0 + b(r + 1) = o. But with a0 chosen a s above, this i s indeed the c a s e . A s
before,

(12) can be used to g e n e r a t e equations which enable us to find the e..

Finally
lim
n+1
n oo u
n
e x i s t s and i s the g r e a t e s t r o o t of (11) in absolute value.
Since (9) g e n e r a t e s a r e a l sequence given r e a l initial t e r m s 9 not only i s
lim
n-^oo

n+1
u
n

the g r e a t e s t r o o t of (11) in absolute v a l u e , but it m u s t a l s o be r e a l .

Hence the

g r e a t e s t r o o t in absolute value of (11) m u s t be r e a l .


If a, b , and r in (11) a r e such that two distinct r o o t s s h a r e the
absolute value of all r o o t s 5 then it i s e a s i l y shown that no l i m i t e x i s t s .
Employing s i m p l e unit t h e o r e m s , we can p r o v e that
lim
n-^oo

"n+s

s .
= a n it
u

u
n

We a r e now able to s t a t e that:

lim

n+1
= cxii

n-*oo

u
n

greatest

1963]

THE FIBONACCI SEQUENCE AND PASCAL'S TRIANGLE

31

Theorem 2. For all sequences formed by sums of terms on parallel diagonals of the generalized Pascal's triangle 9 and for all sequences defined by
(9) given r + 1 initial t e r m s ,
lim

n+s
n

exists and is the greatest root in absolute value of


r+1
x

r_
- ax - b = 0

provided this absolute value is not shared by two distinct roots.


REFERENCES
1. We Se Burnside and As Wa Panton f

Introduction to the Theory of Binary

Algebraic Forms 9 Dublin University P r e s s , 1918, p. 197,


2. L. E. Dickson, History of the Theory of Numbers, Washington, D. C. ,
Carnegie Institute, 1919 - 1923e
3. B. W. Jones, The Theory of Numbers, RInehart and Company, 1955, pp.
77-99.
REFERENCES
(Cont. from p. 14)

1. Ve E. Hoggatt and C, King, Prob. E1424, American Mathematical Monthly,


Vol. 66, 1959, pp. 129-130.
2. H. L. Alder, "The Number System In More General Scales, M Mathematical
Magazine, June 1962, pp. 147-148,
3. J. L. Brown, J r . , "Note on Complete Sequences of Integers," American
Mathematical Monthly, Vol. 68, 1961, pp. 557-560.

You might also like