Professional Documents
Culture Documents
N. N.Vorob’ev
POPULAR LECTURES IN MATHEMATICS SERIES
Volume 2
FIBONACCI NUMBERS
TITLES IN THE POPULAR LECTURES IN MATHEMATICS SERIES
by
N. N. VOROB’EV
Translation Editor
IAN N. SNEDDON
Simson Professor of Mathematics
in the University of Glasgow
Copyright © 1961
Pergamon Press L td.
Printed in Great Britain by Pergamon Printing and Art Services Limited, London
CONTENTS
Foreword • • • • • • • a vii
Introduction . . . . . . . . I
v
Foreword
vi 1
Foreword
V ll]
INTRODUCTION
1
2 Fibonacci numbers
A p a ir
1
F irs t (Month)
2
Second
3
Third
5
Fourth
8
Fifth
13
Sixth
21
Seventh
34
Eighth
55
Ninth
89
Tenth
144
Eleventh
233
Twelfth
377
Fig. 1.
144 and 233) and obtain the total number of rabbits (i.e .
377); and i t is possible to do th is in th is order for an
in fin ite number of months”.
n* (1)
un un + un- 2* ( 2)
U1 “ u 2 + u 3 - u 4 +
U1 " u 2 + u 3 ~ u 4 + + ( - l ) n+\
( 8)
(- i ) n+1un_i + i
(9)
We note that
= u,u
12
Simplest properties 9
un+» u ,u
n-l n
+ u u ,.
n n+1 ( 10 )
n+1
un - l,u.1 + u u n_ 2= un - l, + u ,
T n’
u‘n + 2 = Un - l U2 + Un
Thus, le t
un+k. = u n - l,u,k + u u, ,
^ nfc+1
and
un+fc, + 1, = un - l,u,«+ ,1 + u u.
T n fc+2
lin +4 + 2 = Un - l U* + 2 + Un Uk +V
or
T un +,)
2n - un '(u n-1, + 1' . ( 11 )
Since
u n = u n +, 1 - u n-1,,
or
u 2, n = n+1n-1'
u 3n = u 3n + 1 + u n3 _ u n-1
3
U2 = U 1U3 - 1’
which is obvious.
2
u"+1 Un+lUn+2 = UnUn+2 + Ur,+lUn+2 + (-1)"
or
or
or
“.’ ♦J = * t- 1*"*1
c °o
c° c\
pO p i p2
z °2
3 ci c i <1
i. e. 1
1 1
1
w 1
/V s 4
1 5 IP 10 5 1
r 6 15 20 15 6 1
n 0 nl r2
Un-3’ n-4' n -6’
C° „ C1 .. C2
n-2 n-3 n-4
C°
n -2 = f°
Ln-1 = 1
and
t l _ k(k - 1) . . . (k - i + 1)
1.2........... i
k(k - 1) . . . (k - i + 1)(* - i)
+ ------------------------------------------
1.2........... i . ( i + 1 )
k(k - 1) . . . (k - i + 1) / * - i\
1.2................ i \ i + 1/
_ k( k - 1) . . . (fc - i + 1) i + 1+ - i
1.2........... i i + 1
(k + 1)*(* - 1) . . . (k - i + 1)
- r'i+l
1.2........... i . ( i + 1) - V i-
Simplest properties 15
i.e . the sum of the numbers lying on the nth diagonal of the
triangle.
Making UBe of formulae (4). (5), (6) and sim ilar ones, the
reader can easily obtain further id e n titie s connecting
Fibonacci numbers with binomial coefficients.
= Vn-2 + Vn- 1
cv n - CVn-2 +CVn - r
as was required.
v n' = V'n-1. T
+ V'n-2„
and
Vn = V l + Vn - 2-
For solutions V' and V" of (2) which are not propor
tio n al, le t
(16)
vl 2
V1 + v 2 _ v2
+ V
2
II
+ C V*
H-*
c i v i 2V1
(17)
+ = v 2
C 1V 2 C 2V 2
c = v l v 2 ~ V2V1 c = vl v2 ~ V2V1
1 v { v 2 ~ v "\v 2 ’ 2 u l v 2 ~ v l v2
1 + q - q2.
1 + VsT
The roots of th is quadratic equation, i.e . — ~ and
i -V s
— -----, w ill be the required common ra tio s of the pro
2
and
cl a + c20= u2’
i.e . from the simultaneous equations
C1 + c 2 “
20 Fibonacci numbers
whence
i + ^ /w ( i + ^ n-1 i - V 5 / 1 - v r \n-1
2 V5 V 2 / 2 V5 \ 2 J
i«e
1 -V 5 \n
un (19)
vT
We have
a3 - p 3
a 6 - p/>6 a3n - p/}3n
a
- /--- ^ -- + • • • + ---
30 w v r v ir
Simplest properties 21
_ J _ (a** - a 3 g * * 3 - P 3\
U3 +U0 + --- + “ 3n = V T a3- 1 j53 - 1
But
a 3 - 1 = cr + a 2 - 1 = a + a + 1 - 1 = 2a,
and sim ilarly j3 3 - 1 = j .
2 3 Therefore
_ 1 / q ^ 3 - a 3 f l3" * 3 - £ 3'
U3 +U6 + - " +U3 n = V i-! 2a 2/3
or a fte r cancellations
3n t 2 _ a 2 _^fj3n+2 + o 2 '
U3 + U6 + ‘ ‘ + U 3 n = v f
1 g 2 - j82,
2 \ vH T -v^T
1 “ lnj.9 - 1
=7 (U3"+2 " U2}
We note that
Therefore:
or, using formula (8 ) and the resu lts of the preceding section,
- 1
+ 3 [1 + ( - l ) r,+1un_1]j =
u 3n+2 + + 5
10
Theorem:
----- ■-------
The Fibonacci number un is the nearest whole
number to the nth term a o f the geometric progress ion
on n n
_p a a a" - r 1/31"
v ir vt VT V5
lim un an = 0.
n —» > o o I
376.9.
sTh
The nearest whole number to 376.9 is 377; th is is uJ4
25
26 Fibonacci numbers
a = bqQ + r y N
b = r l9l + r 2.
r l = r 2^2 + r 3-
r ( 20 )
rn-2 = rn -l9n-l + rn -
rn-l = rn % ‘
610 = 55 X 11 + 5,
55 = 5X1 1
a + c = (qQ + c i)6 + r r
u m .( « j + tl )x = u m + u mm j u m+1
,,.
C T3
th e ir difference un+1 - should be d iv isib le by d. Ai
since un - un = un then un-1 should be d iv isib le by
d. Similarly, we prove (induction!) th a t un_2, un_3,
e tc ., and fin a lly uj, w ill be d iv isib le by d. But uj = 1
therefore i t cannot be divided by d > 1. The incompati
b ility thus obtained proves the theorem,
r t-l = r f9f
(u
v «, un) = (unq^+r^, u n'),
(u , u ) = (u ,u +u u ,, u ),
»’ n ' ri ~ r i +* "
or by sections 7, 8
v( u*’, u n'
) = ( u' "9Q-1, u , un),"
or by sections 10, 5
(u
' n*, un') = '(u , un').
Similarly, we prove th a t
(u . u ) - (u , u ).
1 " r2 1
(u . u ) = (u u ),
r2 rl r3 r2
(u . u ) - (u u ).
f-1 t-2 t t- 1
Combining a ll these equations, we get
(u , un )/ = v(u r f’, ur t - l ),
' n’
u t+i ~ u t ~ u t - r
*0
9, +
9 o +-
(24)
93 +-
36
Continued fractions 37
a = b90 + r i-
b = r 1<7l + r 2.
rl = r 2<*2 + r 3’
r n - 2 = rn - l % - 1 + V
r n- 1 “ r n V
a _
b ^0 =%
r1
b
— = <7 + _r 2
L = q. + —
1
rl r l rl
so that
36 Fibonacci numbers
a
~b
r
q2 + - * = q2 + 1
r2 r2
and therefore
ji
6 9n +
9i +
q2 + 1
a_
b +
+
'+
%
2. Let
a) = +
(26)
91 +
q° ‘ q° + ' g°
0 <h_
So i
= 9n + ;
<?2 =*0
91 + —
?2
40 Fibonacci numbers
Pk P*+l
Note that the tra n sitio n from — to q---- is realized by
Qk v*+l
the replacement of the la s t of those p a rtia l denominators
which took part in the construction of th is convergent
1
fraction, i.e . q., by <j. + ----- .
**+l
P1 _ „ . 1 _ W l + 1
~n~ +
Qj <?! <7!
P2 _ 1 _ g0 ^ 1 g 2 + *> + g 2
(30)
Q2 9° + 9i + -±- gl g2 + 1
1 q2
Q2 = ql q2 + 1 = ^ 2 + <?()•
The equation
P2^1 ~ P1^2 “ 1
is easily verified.
P P
The tra n sitio n from _ *±I to <i'f2 according to the
<fc+1 4+2
42 Fibonacci numbers
pk « J k { i k + i + ^ 2 h pk .i
Qk+2 Qkfak+i + q ^ ) + ^ - i
P* + 2 _ Pk + l q k+2 + P k
@k +2 ^ k + l q k+2 + <?A (3 1 )
expression
P A+ 2 " Pk + l qk + 2 + Pk
and that
P k+2®k+l ~ Pk + l ^ k +2 ~
- P k + i ^ k ^ k + i + p kQk+i - p k + i ^ k ^ k + i - p k+iQk'
Corollary.
P. . P. (-1)*
------ h- = ------- . (33)
^4+1 Qfik+1
Since p a rtia l denominators of continued fractions are
positive whole numbers, i t follows from the above lemma
that:
a i • a ............ a i
are consecutive convergent fractions of a.
Let
■* %
As
a , = l = _L
and
i + J - _ JL
_ “n+l (35)
to - qo + 1 . .= .<7o. +
to 1
1
<7l + 9l + •
<72 + <72 +
Continued fractions 45
while
fo ^1_ ^2_
<?o Qi ' -
and the convergent fractions of ti/hy
p_o n p_2
<?;■
Prom the resu lts of the lemma of section 3 i t is easy to detect
th at in view of (36)
and
^ _ u t+1 . „ + u i Un-t+1
■ Mn-t+3 •,
. „ + u t-1
u tu n-t+3 . ,un - t•+ 1,
ui = 1 +
2+
n - 1 p a rtia l
A '+.—1
denominator 1
1 ___ 1 1
<• >=! +- = 1+ V-1 = 1 +
2 +. 2un + un- 1-
2+
« n -l
U u „ + un
— n+ 2
= 1+ n
Un+ 2 Un+ 2
or, putting uQ = 0
u2 un + 2 + ulun
to =
ulun+2 + Y n
Thus the basis of induction has been proved.
i p a rtial 1+
denominators 1+
+1+
2 +-
Continued fractions 47
_ Ui+lUn-i+3 + Ui Un - i+ l (37)
" Ui Un- i + 3 + Ui - l Un -i+l '
i + 1 partial . 1+
denominators 1 +,
'+ 1 +
2+
n-i-1
It can obviously be considered thus:
1 ±J_
i partial (38)
denominators •+ l + ■
2 +.
ln-i —
1
The continued fraction below the dotted line in (38) is, by
(37), equal to
u i+l
. ,un-1+2
■ r, + u.u
i n-1
i n-i+2 T i- l,un-i
u.u . „ + u.
,1 ' i 4-
_ (u. ,)un - . „ + (u.i- l, + u.)u
T u.i+l' 1 + 2 i n-i_
1+
i i.+ ,u .
l n-i+2
„ + u.u
i n-i
u. ,u . ,„
i + l n —1 + 2
+ u.u
i n-i
U.U . ,_ + u.
i n —i+2 i - l n-i
U. ,_U . n + u. „u
_ i + 2 n - i + 2 T i+l n_i
u i+l
. n - i. + „2 + u i un-i
Thus the inductive transition has been proved and so has the
whole lemma.
and s i m i l a r l y .
Q>
< Un+1
A substantial role is of course played here by the lemma of
section 3. on the basis of which we obtain only fractions in
th e ir lowest terms in the process of “contracting" a continued
fraction into a vulgar one. Therefore no diminution of
numerators and denominators of the fractions obtained due to
“cancelling" w ill take place.
+ 1
«1 (39)
9a +
% +
Let
0 (41)
and
Pi 2n + l
(42)
Qi ^2n +l
50 Fibonacci numbers
^2n+2^2n+l ^2n+1^2n °
This means th a t the sequence (41) is an increasing one.
In the same way. i t follows from
P2n+3 _ P2n+1 _ 1 1
^2n+3 *?2n+l ^+3^42 ^2n42^2n4l
th at the sequence (42) is a decreasing one.
*41
— > (43)
Ok Qk+1
and from the fact th at (41) increases and (42) decreases
i t follows that
(44)
and
*2*41 (45)
^2*41
Comparing (43), (44) and (45) we obtain
Continued fractions 51
P2n , P2m+1
''
Q2n ^ V2«+l
Let us prove now that any number can be thfe value of no more
than one continued fraction. / Let us take for th is purpose two
continued fractions to and to ( i t does not matter whether they
are fin ite or in fin ite ).
Let ijg, 9 ^, q^, . . . and q£, < 7 g2> ___ be th eir correspond
ing p a rtia l denominators. We shall show th at i t follows from
the equation co=o>that g^ = g^, g^ = g ', g2 = q'y ••• and so
1 1
90 and g ' t —r
to1
‘ VT * »-•
a ^ 1
+ 9 n+1
1im a = 1im U"+1 = lim V i"
n oo n n-*oo _n
V T ^ n
Continued fractions 53
a + 0 n+l nV 5
1ini a + @n+1 \
a n -» o ° a" /
= lim- 0nV5 1 *
n -» o o lim
1+ oo i +•
a"
But 0 n +1, Vjf is a bounded quantity ( its absolute value is
less them 2) and a continues to increase in d efin itely
as n tends to in fin ity (because a > l ) . This means
1im
n -fO o
lim ----------= 0,
rwoo
a
and we obtain
l i ma = a .
n-too n
The theorem that has been proved means th at the ra tio
of neighbouring Fibonacci numbers approaches a as th e ir
suffixes increase. This re su lt can be used for the
approximate calculation of the number a . (Compare the
calculation of un in I, section 12.) This calculation
produces a very small error, even when small Fibonacci
numbers are taken. For example (correct to the fifth
decimal place)
u 55
10 _
1.6176,
u9 34
Cg A Cf B
Fig. 2.
1 X (46)
X 1 — X
whence
x 2 = 1 - x. (47)
55
56 Fibonacci numbers
2R sin 360° .,
2 . 10
so that
sin 72° = 4 sin 18°X cos 18° ( 1 - 2 s in 218°). (*)
Since
sin 72° = cos 18° # 0,
then i t follows from (*) that
1 = 4 sin 18° ( 1 - 2 s in 2 18°),
and therefore sin 18° is one of the roots of the equation
1 = 4x(l - 2x2),
or
8x 3 - 4x + 1 = 0.
Factorizing the left-hand side of the l a tte r equation we
obtain
(2x - l ) ( 4 x 2 + 2x - 1) = 0,
whence
x = 1 r _ 1-1 +V5
■' ■ . - -1 -V"5
1 T
A,
3 " — :-----
Fig. 4.
AC _ A B _
AB BC~ a '
also holds.
Fig. 5.
A B
X
D F C
Fig. 6. Fig. 7.
Indeed,
also
AD = A E = EF,
since AEFD is a square.
Fig. 9.
62 Fibonacci numbers
U2n-2
U2n-1
Fig. 12.
Fig. 13.
— cm,
V I + 82
2 2
CONCLUSION
65
66 Fibonacci numbers