Professional Documents
Culture Documents
7 1
7 1
"'SUMMER I
CONTENTS
PART I - ADVANCED
Products of Fibonacci and Lucas Numbers .
. . . . . .
. . . . . .
KH. Ferns
A . F. Horadam
14
H. W. Gould
23
Joseph Arkin
41
PART H - ELEMENTARY
Identities Involving Generalized
Fibonacci Numbers
. . . . .
Grolden Triangles, Rectangles,
and Cuboids
.
Muthulakshmi R. Iyer
. . . . . . .
73
Muthulakshmi R. Iyer
92
Book Reviews
66
FEBRUARY
J- W. Gootherts 104
Y A- F - Hillman 107
1969
H. L. Alder
Marjorie Bicknell
John L. Brown 9 J r .
Brother A. Brousseau
L. Carlitz
H. W. Eves
H. W. Gould
A. P . Hillman
V. E. Hoggatt* J r .
Donald E Knuth
George Ledin, J r .
D. A. Lind
C. T. Long
Leo Moser
I. D. Ruggles
M. N. S. Swamy
D. E. Thoro
Let Ux
u u
xt
x2
u
xn
L
Xi
= F
X2
F
Xi
X2
X2
+ (-1)X2 F
Xi+X 2
= F ^
L
Xi
Xi~X 2
- (-l)* 2 F
Xi+X 2
= L
,
Xi~X 2
+(1)X2L
Xi+X 2
Xi-X2
Fx F
- 1 ti A - (-D* 2 L
i x2
5 L xi+x2
xi-x 2
The
[Feb.
or F
ator applied to these numbers it reduces their subscripts by 2x.. With this in
mind, we may write
L
Xj
X2
Xl
F
xt
__,
X2
Xl
Xt4x2
= F
Xi+X2
= (1 + X2 )L
X2
= F _,_ + (~1)X2F
X!+X2
= ( 1 - X 2 ) F
= (1 + X 2)F ^
= (1 - cX2)L ^
x2
- (~1)X2F
Xi+X2
= L
Xi+X2
Xi-X 2
Xj-X2
+(-l)X2L
Xi4x2
Xi~X2
= i|L ^ - (-1)X2L
xj+x2
5[ Xi+x2
1
xrx2J
L
xi
X3
x2
.
X3
= F
+ (-1)X2 L
F
X3
Xi-X2
- (-if 3 F _,
Xi+X 2 +X 3
- (-1) X3 F
Xi+X 2 +X 3
X F
Using
__, - (-if 3 L
|_ Xi-X 2 +X 3
+ (~1)X2
Xi+X 2 -X 3
X TF
= F M
X3
Xi-X 2 -X 3 J
+ (-if 2
Xi+X 2 -X 3
- (-1) X2+X3 L
xi-x 2 +x 3
xrx2-x3
1969]
= (1 + X 2 )(1 - X 3 )F
Xi X2 X3
,V
;Jr
Xi+X2+X 3
(1 + X 2 )F
- (-1)X3F
X!+X2-+X3
Xi+X 2 -X 3
- (-1)X3F
+ (-1)X2F
"xi-^2^3
'
Xi+X 2 -X 3
K }
- (~1)X2+X3 X
*XrX2+X3
1}
X F
xrx2~x3
Since
(1 + X 2 ) ( 1 - X 3 )
= 1
X
+ 2
_ X 3 _ e X 2 + X 3
>
L L F
= ( l + X 2 - e X 3 - e X 2 + X 3 ) F
xi x2 x3
x1+x2-bc3
X F
Xi-x 2 -x 3
L L
= ( 1 + X 2 ) ( 1 + X 3 )F . _,
xj x 2 x 3
xi+x2+x3
F
Xi
L
Xi
= ( 1 - X 2 ) ( 1 + X 3 )F
L
X2
L
X2
X3
= (1 + X 2 ) ( 1 X 3 ) F
F
X3
Xi+X 2 +X 3
_, .
Xi+X 2 +X 3
x t x 2 x3
[Feb.
= i ( l - X 2 ) ( l - X 3 ) F ^
5
Xi+x2+x3
L F F
= i ( l - X 2 ) ( l - X 3 ) L _, ^
xi x 2 x 3
5
Xi+x2+x3
L F
= 4 (v l + X 2n) ( l _ x 3 ) L
xi x 2 x3
5
' X!+x2+x3
F L
= i ( l - X 2 ) ( l + X3)L ^ M
xi x 2 x3
5
Xi+x2+x3
L L L
- (1+X 2 )(1+X 3 )L ./
xi x 2 x 3
Xi+x2+x3
The preceding resuJ.ts are the bases for the following conjecture.
Let U x . represent a Fibonacci or a Lucas number. Let p be the number of Fibonacci numbers in a product of both Fibonacci and Lucas numbers*
Let
Xi+x2+- +x n
(1 - ^ )
or
(1 + 1)
1969]
= i ( l - i 2 ) ( F 4 5 - F 2 9 + F 2 5 F 9 )
= i (F 45 - F 2 9 + F 2 5 - F 9 - F 2 i + F 5 - F j + F_ 15 )
= | (F45 - F 2 9 + F 2 5 - F 2 i + F 1 5 - F 9 + F 5 - F i )
The above rule also applies if the product consists entirely of Fibonacci
or of Lucas numbers each with the same subscript.
L
x=
For example,
^ ^ S x
= ( l + 4 e x + 6 2 x + 4 3 x + * 4 x ) L 5 x
5x
+ 4( 1)XL
3x
+ 6( 1)2XL
+ 4
<AX
+ (
"1)teL-3x
= L 5 x + [4(-l) X + (-1)X] L 3 x + [ 6 ( - l ) 2 x + 4 ( - l ) 2 x ] L x
= L5x
5(-l) X L 3 x
10 L x
[Feb.
(1 + e^r L
nx
= L +
nx
2 X
(n-4)x
-2)x
Since
L_ k = (-1)-L k ,
we get
LA = L +
x
nx
'(v)-(::ij
(VM^j
(Q-2)X
* *[(QUH)]fc\
e2xL
( n \ + ( n
)
I ml
In - mI
= (n
\
+
m
)
/
Ln = L
x
nx
_, / n \ xT
_,_ / n \
{l) L (n-2)x + {2}
2xT
^
, f n - l ) A 2 J1
L
+
+
(n-4)x "" \ /
n~l
2
n = 1, 3, 5,
i=0
'
(n-4)x
1969]
i - E K 1 (") v * J -<-|x f I )
-
i = 0
(n, even)
n-1
2
<x+1,i
- i 2i E
<-
()
i=0
(n9 odd)
.1
F 11 =
htl,i
4.2 E
<- (i)v 2i)x
i=0
| (x+D / n
+ 2(-l)
fi')
(ns even)
The proof of the rule which has been used to express products of Fibonacci and Lucas numbers as linear functions of those numbers is a proof by
induction.
We have seen that it is true for n = 2 and n = 3. Assume it is true for
all integral values of n up to and including k. Then, if p is even
k
TT u x.
(l)
"fin
(1
* * 2) '"
(1 k)
xj-bt2+ +xfc
i=l
TT".
i-i '
- we get
bi
(1 X2 ) (1 k )
S]
y
(T,
-J- (-1)
^ Xi+x 2 + ' + x k+i
L
x 4 +X2+--Hx k -x k + 1
A
K<ti
f*2
(1 +
2) (1 K ) ( l + fc+li
)L.
Xi+x2+- - - + x k + 1
U
x
F
i
= -*
k+l
x
2)
(i+
i 2
+8
e+x
xk+1
( H e* 2 ) (1 e X k ) X
F
x +x
5L
- we get
X
k)L
... ( i
[Feb.
.
X
- (-1)
Xl+x2+-
k+1
+x k + 1
F
Xi+x2+- .+x k -x k + 1
1
( 1 X 2 ) ^ - ( l X k ) ( l - k + 1 ) F _, ^ ^
[E]
Xi+x 2 +--+x k+1
Since both of these results agree with that given by the general rule for
n = k + 1 the induction is complete for the case in which
U
= L
XffX2+* +X n
Xi+X2+ -+X n
= F
Xi+X2+ +X n
Xi+X2+ +Xn
, _L
and
i 2 * * a+xn
x +x +
, ,
XJ-HX2+" *+xn
1969 ]
factors which are Fibonacci numbers and n - i which are Lucas numbers*
The number of such factors is 9 of course
( " )
For example s
Sf = F
*
F
Xi
+F
X2
L L + F
X2
X3
X4
X3
X|
L L + F
Xi
X3
X2 X4
+F
X4
X2
X4
Xi
F
Xj
L
X4
+F
X3
L
X2
F
X3
+
X3
L
X4
Xi
L
X2
SnL
1
n+l
n
1
1 F
1
"
= Sn+1
l
V>-1
(H-K; )
For the case n = 2 we readily prove (using Binet f s formulas) that
x1+x2
1
2
(L
x1Fx8
x1Lxa)
1 2
= 2; s |
^t-^ - i ^ ^ + 5 ^ V
\
(S20 + 5S22)
10
F
[ Feb.
= F
Xi+X 2 +X 3
(X1+X2)+X3
ifL
Xi+X 2
+F ^
X3
Xj+X 2
X3
L L F +5F F F + L F L +F
L L
[ Xi x 2 x3
xt x2 x 3
xj x 2 x3
Xi x 2 x3
2 L
= fe + 5 S|]
^X^+Xs
(Xi+X 2 )+X 3
L.
Xi+X 2
+5F
X3
Xi+X 2
F
X3
i(L
2 xt L x2 + 5 F XiF x)2 L x3 +|(L
2 xi F x2 +F xi Lx2) Fx3
L
2 2 L[ x i
22
L L + 5 F F L + 5 F L F +5L F F
x2 x 3
Xi x2 x3
Xi x2 x 3
xi x2 x3J
So + 5 S2
22 L
" Xi+X 2 +X 3 +X 4
23
Si + 5 S3
F Xi+X2+X3+X4+X5 = \sl
4
+ 5 S | + 5 2 S|
2 L
Xi+X 2 +X 3 +X 4
2L
+ 5 S^ + 52 sj
1969]
11
SA + 5SSJ. + 5 2 S |
x 1 +x 2 +x 3 +x 4 +x 5
(2)
F
xi+x 2 +- +x n
(3)
L
Xi+x2+ 9 +x n
n-2
2
n
^n , _ ^n t _o 0 n ,
, J5
S - (n, even)
n-1
n - 1 St + 5 S 3 + 5* S 5 + + /v
n-1
5 2
Sn
(n, odd)
n
9n-l
!S5? + 5 ^ + 5 2 S? + -
2gn
n
+ 1
(n, even)
n-1
5
(n, odd) .
S*
n-1
"nx
n-l
M F L*"
\1)
s2
+ 5 (*W L " + . .
y3y
d,. 2
.2 /n\
nx
/ MB*"1!
In - I I
n , r- / n \ ^2 _ n - 2
1
j + 51 0 F L
,n-l x
\2j
x x
n-2
Fn
"I
(n 9 odd)
(n, even)
\n-l
W L
/
(n, odd)
2x =
xLx
and
2x = i H
+5F
(n 3 even)
12
[ Febo
We have now to prove the identities (1) and (2). The proof is by induction
on n. Both identities are true for n - 2, We assume they are valid for all
integral values of n up to and including n = k.
Then
k-2
(4)
F.
k-1
< i
k-l_
(k, even)
k-1
2
Gk
Sk _
(k, odd)
k
(5)
xi+x2+"+xk
lr
52SJ*
IT
k - 1 SQ + 5 S f + 5< S f +
k-1
5
Sk
Now
(6)
<k. even)
(k, odd)
k-lJ
= L,
x
x1+x2+- +x k +x k + 1 = (xi+x2+ -+x k )+x k + 1
T
X +X
2 + ' 8+Xk
+ 5F
F
Xi+X2+- +Xk X k + 1 <
k+1
(7)
__, L
L ^
xi+x 2 + +x k x k + 1
Jc-1
S?L
+5S^L
k4-l
+
k+l
52S^L
K
+ {
(k, even)
k+l
k-1
s{^ - L
k - 1 x,
fc+1
(k, odd)
1969]
(8)
F
Xi+x 2 +-..+xk
xk+1
13
1
x
2k-l
k+l
k+l
k-2
2
|5
Sk , F
k X
+ I
"
(k, even)
*k+l
k-1
2
SkF
(k, odd)
k x,
*k+l J
Substituting in (6) from (7) and (8) and r e g r o u p i n g we get the following:
= ST1 + 5(SJL
+S^F
\
k+1
Xi+X 2 +- '+Xk+I
)
k+l/
+ 52(s?L
+ S^F
\ +
x
x
\
k+l
k+l/
52 S k L
k-1
2
+ Sk
k
k+l
-1
(k, even)
k+i/
S, F. M
k k+1
(k, odd)
Hence
Xi+x2 '-
+x
52Sk+1
(k + 1,
even)
S k *J (k + 1, odd)
ERRATA FOR
PSEUDO-FIBONACCI NUMBERS
EL H. F e r n s
V i c t o r i a , BC , Canada
P l e a s e m a k e the following changes in the above-entitled a r t i c l e a p p e a r i n g
in Vol. 6, No. 6:
p . 305: in Eq. (3), 0 . + 1 should r e a d : C^
bottom: B ~ k + 1 should reads B k + 1 .
r e a d : 2X0
; in Eq, (13),
p . 312: ( X - 2 ) 0 2
On p. 306, the 6
should r e a d : 3 0 2
should r e a d :
40? should r e a d : 4 0 ? .
1
30
_r
+1
should
. Equation (17),
atonr/u
An
^n+^d
should r e a d X( X - 2 ) 0 9
* *
on
^n
and
1.
INTRODUCTION
[n/2]
(1.1)
u n (p ; q) =
^(-l)k(nk=0
)pn-
2 k
and
(1.2)
[n/2]
vn(p,q) = ^(-1)
,
- ^
k=0
v
k
jp
with reciprocals [ 3 ]
[n/2]
*-gK0-(-2
and
14
V2k<P'*q
FPH 1Q6Q
'
1K
15
[n/2]
(la4)
pB =
= 1}
k=0
respectively.
(1.5)
2^ \ k J
k=0
ZLr n - k, \ k J
k=0
as
in
-i ~>
, _
(1.7)
V \
i i k / n - k\
wn(a,b,p,q) = a ^ f l )
( k )p
+
n-2k k
q
^ \T*
^"
pa)
zLr
k=0
2. TSCHEBYSCHEFF FUNCTIONS
Write
(2.1)
x = cos 0
(2.2)
p = 2x , q = 1
so that
(2.3)
d = 2i sin 0
(i = V 1 ! ) .
, . v k / n - l - k X n-l-2k fc
*(
k )p
1969]
[n/2]
(2.11)
v n (2x,l) = 2 c o s n =
17
(-l) k ^
/ n " k j (2x)n" 2k
k=0
with reciprocals
&/2] r
(2.12)
(2x) 1
V 2 k (2x,l)
k=0
and
[n/2]
(2.13)
<2x)
=
k=0
That
is,
(2.14)
w n ( l , 2 x ; 2 x , l ) = u n (2x,l) = Un(x) = ^ x U ^ x ) - U ^ x )
and
(2.15)
(2.16)
w n ( l , l ; l , l ) = u n ( l , l ) = U n (i) = U H ( i ) - U n - 2 ( i )
and
(2.17)
18
[Feb.
Generally,
(2.18)
wn(a,b;2x,l) =
bu
nl
(x)
"~ a U n-2 ( x )
'
By means of the w -notation, relationships among Tschebyscheff polynomials may be conveniently expressed. Recalling the known result [ 8 ] , for
instance, that
(2.19)
w n = w n (2,2x; 2x, 1) ,
- 2 X < V l - o>n_2 .
-1.1
>
>
>
w
2
n-r 2
u
- w = eq
T_i
2
2
a | cos (n + r)0 cos (n - r)o - cos n$ \
(2.22)
2
\
sin^rfl
2
2
+ (2 - a) cot e jsin(n+r)0 sin(n~r)0 - s i n n6 J = e ~ 2
sin 6
(2.23)
2
2
sin (n + r + 1)6 sin (n - r + 1)6 - sin (n+ 1)0 = -sin r0
1969]
19
and
(2.24)
2
cos (n+ r)0 cos (n - r)0 - cos n0
2
=
-sin
r0
in which e is given by (2.8) and (2.9), respectively. Both results (2,23) and
2
2
(2.24) are easy to verify. The particular result w + eu
= aw
((4.6) [5])
derived by setting r = m implies the identity
2
. 2
cos 2n0 - cos n0 = -sin n0
in (2.24).
Other trigonometrical identities are not hard to detect, but it is interesting to discover just how they are disguised. As further examples, we note that
pw n + 2 - (p 2 - q)w n + 1 + q 2 w n - 1 = 0
2
2 cos0 sin (n + 3)0 - (4 cos 0 - l)sin(n+2)0 + s i n n 0 = 0
(2.25)
1
(2.26)
and
(2.27)
v
' ;
J Bln(n+r+l)fl+Bln(n-r+l)g_
I
sin (n+1)0
(2.28)
I
l
coB(n + T ) g + c o s ( n - r ) g .
cos n0
cog
2 cos
r$
where, in each pair of identities, the first refers to ju | and the second to
jv j . A formula also worth investigation is
20
((4.1) [5]),
sions for
[Feb.
cos k
k=0
and
n-1
V ^ sin (k + 1)0
k=0
Similar remarks apply to the formulae for sums of squares and cubes.
Instead of (2.1)-(2.3), we may put
(2.29)
(2.30)
p = 2y, q = 1
so that
(2.31)
3. COMBINATORIAL FUNCTIONS
L (x) used in
1969]
21
[n/2]
(3.1)
u n (l > -x)=
kk
xk=L
n-l(x)'
k=0
(3.2)
k=0
2 V
>
2n"1g2
w n ( l , l ; 1,-x) = u n ( l , - x )
L n _ l ( x) = L n _ 2 (x) + xL n _ 3 (x)
[L Q = 1,1^ = 1 + x]
while
w
(3.5)
2n
( 2 j l ; lj
"X)
= V
2n(1'~x)
(3.7)
L n _ l ( l) = fn
(n > 1) .
JJ
(3.8)
_,.
fi
M n (l) = l 2 n .
4. OTHER FUNCTIONS
Besides these combinatorial f u n c t i o n s
(themselves involving trigonometrical and hyperbolic functions), other functions are related to the Fibonacci-type recurrences. In this respect, a recent
article by Byrd \_2j is worth emphasizing, particularly a s , it seems, his work
offers possibilities for generalization
REFERENCES
1. R. Buschman,
ff
1, No. 1, 1963, p .
16.
3. L. Carlitz, "Generating Functions for Powers of Certain Sequences of
Numbers," Duke Math. J . , Vol. 29, No. 4, 1962, pp. 521-538.
4. H. Gould, "ANew Series Transform with Applications toBessel, Legendre
and Tschebyscheff Polynomials," Duke Math. J..,
pp. 325-334.
5. A. Horadam, "Basic Properties of a Certain Generalized Sequence of
Numbers," The Fibonacci Quarterly, VoL 3, No. 3 9 p. 161.
6. A. Horadam, "Special Properties of the Sequence jw (a, b; p,q)|* M The
Fibonacci Quarterly, Vol. 5, No. 5, pp. 424-434.
7. E. Lucas, Theorie des Nombres, P a r i s , 1961, Chapter 18.
8. J . Riordan, An Introduction to Combinatorial Analysis, New York, 1958.
* *
INTRODUCTION
In 1915 Georges Fontene (1848-1928) published a one-page note [ 4 ] suggesting a generalization of binomial coefficients, replacing the natural numbers
by an arbitrary sequence A
damental recurrence relation for these generalized coefficients and noted that
for A = n we recover the ordinary binomial coefficients, while for A - q11
- 1 we obtain the q-binomial coefficients studied by Gauss (as well as Euler,
Cauchy, F . H, Jackson, and many others later).
These generalized coefficients of Fontene were later rediscovered by the
late Morgan Ward (1901-1963) in a short but remarkable paper [16] in 1936
which developed a symbolic calculus of sequences. He does not mention
Fontene", Failing to find other pioneers we shall call the generalized coefficients Fontene-Ward generalized binomial coefficients.
is the Fibon-
= F +F
n
n~l
with
F0
= 0,
Fx = 1 .
This idea seems to have originated with Dov Jarden [ll] in 1949. He actually
states the more general definition but only considers the Fibonomial case.
Fibonomial coefficients have been quite a popular subject in this Quarterly
since 1964 as references [ l ] , [ 9 ] , [lO] , [13], and [15] will tell.
23
See also
^4
Ward coefficients cited above, and because so many properties may be obtained in the most general case, we shall develop below a number of very
striking general theorems which include a host of special cases among the
references at the end of this paper. Despite an intensive study of all available
books and journals for twenty years, it is possible that some of our results
have been anticipated or extended. Indeed certain notions below are familiar
in variant form and we claim only a novel presentation of what seems obvious.
However a large body of the results below extend apparently new results of the
author [7], [8] and we obtain the following elegant general results: Representation of Fontene-Ward coefficients as a linear combination of greatest integer
(bracket function) terms; Representation of the bracket function as a linear
combination of Fontene-Ward coefficients; A Lambert series expansion of a
new number-theoretic function; A powerful inversion theorem for series of
Fontene'-Ward coefficients; and some miscellaneous identities including a brief
way to study Fontene-Ward multinomial coefficients by avoiding a tedius argument of Kohlbecker [ l 3 ] .
The present paper originated out of discussions with my colleagues,
Professors R. P . Agarwal and A. M. Chak, about the feasibility of extending
Ward ! s ideas to broader areas of analysis and number theory. Chak [3] has
developed and applied Ward ! s symbolic calculus of sequences to discuss numerous generalized special functions.
Every result below can be immediately applied to the Fibonacci triangle,
or new variants thereof, and the inversion theorem given below is expected to
be especially useful to Fibonacci enthusiasts. Such inversion theorems are
valuable tools in analysis and have not been previously introduced or applied
for Fibonomial coefficients.
to be the
1969]
25
{"} - o
The sequence A
whenever
k < 0 or k > n
A i- 0 for n > 1 8 Ward [16] took A = 1 9 and there is no loss of generality in doing that. However we cannot in general simplify very much and we
shall retain A1 as arbitrary,, One has only to multiply Ward's sequence by
A- to obtain our results. When no confusion can occur as to our choice of the
basic sequence A we shall omit the subscript A in our notation (1). We use
braces to set our coefficients apart from ordinary and q-binomial coefficients.
With this definition we can now exhibit the Fontene'-Ward Triangle:
1
1
A
2
.-*
A
l
tl
A1
1A2
l
A
V l
1A2
V i
AXA2
4 3
A
A ^
t
l
A A
Ag
A
1 2
A
5A4
1 2
A
6 5 4
6 5
tl
A A
A A
A
1 2 3
1 2
l
A A A
A A A
A A
7 6 5
7 6 5
7 6
AjAjjAg
A^Ag
A ^
A1
26
[Feb<
fc}- |kjl
[nw- kjl
(4)
where
M '
Vn-i
" Vi
wlth
C]'- = i
(5)
ft} {)) - {l ft :
(6)
Ak
-{!:-M^
In general A - A, J A
when we set A. = (qJ - 1) /(q - 1). Fontene" is correct that we get q-binomial coefficients with A. = qJi - 1, but it is better to
include the factor q - 1 in the denominator so that we can also assert that
lim. A.
q ->i J
= j
1969]
(8)
f(n,k)
27
- F.
K
n-k
It is easily verified that f satisfies the recurrence
(9)
f(n+ r, k+ r) = F ^ f ^ k ) + Frf(n-l,k-l) .
(11)
F
1
f(n5k) = F - d l _ +
n-k
which may also be derived directly from (8) and the relation
<12>
n=
k F n-fcf 1 + F k - l F n - k '
There are then an abundance of ways to modify f(n9k) using known Fibonacci
relations, and the particular way we might interpret f(n,k)
determines the
nature of the Fibonomial relations which will followfrom our general theorems.
An important observation is this: f(n,k) is independent of n in the case
of ordinary and q-binomial coefficients, but not in the Fibonomial case. This
makes the possibility of having certain expansions generalize depend on the way
in which we can modify the recurrence.
We return to relation (6) and sum both sides with respect to the upper
index. Clearly we obtain the relation
28
,
[Ken.
3=k
i4)
<
Lk*...,k>
l'"2'
*'V
k-
"A.
1=1 l
k
r.2
'
TTA^TTA^-TTA.
1=1
L = l
1 = 1
<15>
{a"b} - {1}
in terms of our original definition (1). Moreover, trinomial and higher order
coefficients are products of ordinary Fontene-Ward generalized binomial
coefficients:
w>
<"
(18)
and the general result follows at once by induction. This is a well-known device for ordinary multinomial coefficients and the application here is that once
one proves that the Fontene-Ward binomial coefficient is an integer for some
sequence A , then the Fontene-Ward multinomial coefficients, by the above
relations, are integers, being just products of integers. This circumvents
1969]
29
the tedius argument of Kohlbecker [13] for multinomial Fibonomial coefficients, for example.
Making use of the ideas developed so far and paralleling the steps in a
previous paper [8], we are now in a position to state and prove our first major
result. We have
Theorem 1. The Fontene-Ward generalized binomial coefficient may be
expressed as a linear combination of bracket functions by the formula
ft)-[5]tEB]V'''-E[!]v.i.
j=k+i
j=k
m<x
and apply this to formula (13) precisely as was done in [8]. The result follows
at once. It is easily seen that R,(k,A) = 1. There will be no confusion of
R (j,A) with R k (j,q) in the former paper if we merely make a convention that
whenever we have a sequence we denote it by a capital letter and then (20) is
meant. Thus R k (j,F) would mean the Fibonomial case. Thus our first theorem expands the Fibonomial coefficient as a linear combination of bracket
functions.
The expansion inverse to this requires a little more c a r e . It was found
in [8] by means of a certain inversion theorem for q-binomial coefficients.
We must pause and establish the corresponding inversion principle for the
30
[Feb.
(21)
k=0
Then we find: g(0) = f(0).
f(1) = -g(0) + g(l) ,
whence
g(l) = f ( l ) + f ( 0 )
Then
A
f(2) = g(0) - /
g(l) + g ( 2 ) ,
- l ) f(0)
g(3) = f(3)
%L f(2) + ^ ( ^
- l)fd) + ( l + ^ ( ^
" 2 ) j f(0)
and it appears that the B, are independent of n and any number may be found
in succession. This is quite correct, for we may readily solve the system of
equations necessary to determine such B,
1969]
31
4 = "
1+
;(-fc(S-i)-S@
4
A
Put
(22)
Wlth B
k=0
0= B l=
lj
B 2 = A 2 /A
- 1, etc.
(23)
g(n+i) =
n+1
B f(n+i k =
~ ) E{nr}Bn+i-jf)
ETk1 k
k=0
]=0
whence
k
i
-.^^5:^{':
}Eft>v*-kpO
j=0
i=o
k=j
-<* E2>"*{,,i1}fi}BM
32
B. . At the same
time we apply the identity (5) and we have the result that
<24>
(25)
k=0
From this it is easily seen that we can summarize our recurrence for
B
(26)
e i )
LJ
k=0
l x + y } n = S{k} x ky n -k
k=0
1969]
(28)
Br
(-1) n
1-1
- O
{1} -{
1-1
' \ n - 2(
n
i) , -H)
{?}. <-i) $.
f(n) = 2 ( - 1 ) n ' k { S } g W
(29)
k=0
if and only if
g(n) =
(30)
We have
k=0
k=0
j=0
k=0
S4
- ,
L*
k=0
The reader should have no difficulty in showing that (29) implies (30),
relation (31) again being what is needed to cancel out unwanted t e r m s .
These relations are nothing more than extensions of the familiar inversions given in [ 6 ] , [ ? ] , [ 8 ] .
The application and use of Theorem 2 for Fibonomial expansions needs
little elaboration. It allows often to solve for something given implicitly under the summation sign.
As was done in [6] and [8] we need some small variations of Theorem 2.
It is easy to see that the theorem can be stated in the equivalent form
f(n) = ^ K ) k { k } g ( k )
(32)
k=0
if and only if
g(n) =
(33)
K) k {1} Bnkf(k)
k=0
(34)
f(n) = Y^{k}
k=0
if and only if
g(k)
1969]
35
*<-> - E ** {1} B n . t f .
(35)
k=0
j=fcM
w h e r e the coefficients
(37)
]=k
Q, (j,A) a r e given by
A =
Then by the i n v e r s i o n
and writing
j for
in (34)-(35), the r e s u l t is i m m e d i a t e ,
Hence a s a Fibonacci i t e m , this t h e o r e m allows one to e x p r e s s
the
orthogonal in g e n e r a l .
L a m b e r t s e r i e s for
and
are
R and an i n v e r s e s e r i e s for
Qe
The binomial t h e o r e m
R,
36
oo
38
<>
ZE
f(d n) =
'
n=l din
oo
LFeb'
oo
ZZf(d,md)
d=l m=l
This is merely the limiting case of relation (20) in [7] for example.
Theorem 4. The Lambert series expansion for
x1
00
R, (j,A)
is given by
A.. - A .
& r5-?^(-^
(39)
~ , *,
din
kW = (:!) =ZX(d>
din
found in [ 7] .
Multiply (40) through by x11 and sum both sides on n, We find
1969]
VA
37
-A
A)
K
n=l
n=l
go
din
WJ
]C S
xindR
d=l m=l
_.
dA
= v' >E
d=l
xmd =
m=l
(d A
XX > > T 2 ^
d=l
L fc J
(l - x) (1 - x*)
(42)
n=k
j=k
j-k
n=j
The last inner sum is not conveniently put into closed form by a binomial theorem, but we can transform it as follows:
38
n=i
n=i+l
[Feb.
co
The r e s u l t is the f o r m u l a
x A
GO
- A
'-IEUK-S'I;:!}- ^1
,43,
n=j
n=j
00
(1 - x )
(1 - x) 3
- , ,
n=j
u s e d in [ 7 , p p . 2 4 1 , 2 5 2 ] ,
The c o r r e s p o n d i n g
(1 x)
- E[j]xn
= x3
q-analog
in [ 8 , p . 4 0 7 ]
was
fl" (1 - xqi)
n=j
i=l
The coefficients
j=k
Q,(j,A)
n=j
satisfy the g e n e r a t i n g
expansion
1969]
XX G ' A ) ~~J
(45)
j=k
= f(X k)
'
39
'
and
00
(46)
Jj
^Q^^Jf^xJ)
= ^
n=j
n=j
and we may now see easily that substitution of (45) into (46), and conversely,
yields our desired orthogonality of R and Q9 Thus we evidently have
Theorem 6. The functions R and Q as defined by (20) and (37) satisfy
the orthogonality relations
n
XX(j'A)Qj(n'A)
(48)
= 8
k=^Qk'A)Rj(nsA) '
j=k
j=k
Consequently, we also have proved the very general inversion theorem for two
sequences that held for the previous cases [ ? ] , [ 8 ] . That is we have
Theorem 7. For two sequences f(x s k,A), g(x,k,A), then
(49)
f(x,k,A) =
^
k<j<x
if and only if
40
(50)
g(x,k,A) =
[Feb.
2 ^ f(x,j,A)Q k (j,A) .
k<j<x
CONCLUSION
In the present paper we have given a sequence of seven main theorems,
generalizing all of the corresponding results previously found for ordinary and
q-binomial coefficients to the most general situation for Fontene-Ward generalized binomial coefficients. As a single byproduct we have results universally
valid for the popular Fibonomial triangle. The inversion theorems given here
are expected to suggest other inversion theorems in the most general setting,
which can then be applied to any special case that is covered by the FonteneWard Triangle.
REFERENCES
1. Terrance A. Brennan, "Fibonacci Powers and Pascal's Triangle in a
Matric," Fibonacci Quarterly, 2 (1964), pp. 93-104, 177-184,
2. Raoul Bricard, "Necrology of Georges Fontene," Nouv. Ann. Math. 9 (5)
1 (1923), pp. 361-363.
3. A. M. Chak, "An Extension of a Class of Polynomials," to appear.
4. G. Fontene, "Generalisation d'une formule connue," Nouv. Ann. Math.,
(4) 15 (1915), pp. 112.
5. H. W. Gould and L. Carlitz, "Bracket Function Congruences for Binomial
Coefficients," Math. Mag., 37 (1964), pp. 91-93.
6. H. W. Gould, "The Operator
1.
INTRODUCTION
w1
oo
\ ^ (k) w
Ev* -E
1
w=0
w=0
instead of c
w
.)
w
We define
Evw =
F(x)
w=0
for a finite f,
t
W=0
W=l
2.
THEOREM 1
If
F(x)/Q(x) = 2 ] u w x w
w=0
41
42
[Feb.
then
(2.1)
lim iu / u .1
n*>oo j n / n-j j
I.
lim u / u ,
n-^>oo n / n-i
converges to some root (r) in Q(x). (We must then prove that this root (r)
in Q(x) is
(rjj.)
Let
m
(2.3)
MW = J J (l-r w x)
w=i
where the p
di +
Pi
= d2 + p 2 = . . .
- p w + dw = k
(k = 1, 29 3, )
for a finite w = 1, 2, 3, , m.
Then*
m
M(x)Q(x) = " P J (1 - r w x ) k = 0 k (x)
w=i
so that
(2.4)
F(x)M(x)/Q(x)M(x) = F(x)M(x)/<^(x)
- Ev* - !>*'
w=o
w=o
1969]
43
where it is evident
u n = c(k,n)
Now let
c
4(x) = \
(where v is finite) ,
w=0
(2.5)
FWMW/^W
oo.
- ^
c(k .- 1, w)xW
w=0
Vw=0
/ \w=0
cfc-l.n) =^cfcn-w)c w =
W=0
c.
w w
W=0
k = 2,3,4,- .
k+1
(where
k = 1,2,3, )
where
(2.5.2)
c(k+l,n) = gk+ic(k+l,n-l) .
44
[Feb.
c ( k , n ) / c ( k , n - 1) = g k
v
/ v
= y c(k+l,n-w)c
iy
c(k + l , n - w - l ) c
w=0
/ w=0
=
V i
>
so that
(2-5.4)
gk+1
= gk = . . .
gI
(^(x))"
= T T (1 - i ^ x ) "
oo
1
W=l
= Y ^ e(m,w)x w
(for a finite m)
W=0
j are
I W j
(for a finite j = 0 , 1 , 2, ) ,
in place of
Inn
19^9]
For m = 1, we have
00
(i nx)"1 = ^
(2.7)
e(1
w=0
where
e(l, n) = r n ,
so that
e(l,n)/e(l,n-j)
rj
For m = 2, we have
00
[(1 - r l X ) (1 - rax)]"" 1 = ^
e(2,w)
w=0
(2.8)
w
X
where
e(2 9 n) = (rn'
- rn
)/(vt - r 2 )
so that
e(2,n)/e(2,n - j) = rj
It now remains to consider for finite m = 3, 4, 5,
/ ^
s=0
V1
t
s=l
46
[Feb.
(U
Un = X X ~ s V s
s=l
= 1
>
and
Ui = U 0 a t r a l , U2 = Uia t r a l + U
t-1
at^%Ut=^ U ^
s=0
t-1
=i
TTVv> -E a s xt " s
s=l
s=0
we may write
t
(2.1D
TfVv =
xt
s=l
t-1
-Zvs
s=0
- ^
E ( v E V H r s=2 V
r=l
Uixt = Ui
t-1
s=0
) x t-s
r
1969]
47
in (2.11) l e a d s to
x t+i
(2.13)
U 2 x t
"
1 +
E (
r=o
t-r-2
3 - E U r ^ s V s ^ j
s=o
'
+ Utao .
Now, multiplying (2.13) by x and combining the r e s u l t with
sx
^ s=o
(2.14)
t+2
1 +
(Urhl-ZUrt-sVs-i)
. tr-Y-2
r=o
+ a0U2
We continue in the exact way we found (2.13) and (2.14) for n - 1 s t e p s to get
(2.15)
t-3
J**- - U n x + 5 ] u.n+r+1
r=o
+U
n-i
JLJ
s=o
n+r-sat-s-i I
'
xt-r-2
r +n_i =
'
V*"1
+ R(ri) +
(2.16)
r t+n-i
t-l
n t
R(
V l a '
48
[Feb.
These
equations
UD2
= D l ( n) ,
- t+n--1
(2.18)
t-2
Dt(n) =
r
*r -
t+n--1
t
t~i
(2.19)
D,
rj
t-2
ri
r*
'
ri
rj
ll
=
t-2
t
"
IT
D
n - i L2
Dj1 (n - 1) ,
Since the r
/U
n-i
Dl(n) /Dl(n
"
1]
n/Un-i
t+n-2^
t+n-2 ^
(Di(n)/rp^)/(D1(n-l)/rr11^)
t+n7
Now in (2.22) we l e t r i
' (in the n u m e r a t o r ) divide e v e r y t e r m of the f i r s t
t+n2
column in (2.18) and rA
(in the denominator) divide e v e r y t e r m in the f i r s t
column of (2.18) (with n r e p l a c e d by n - 1).
dent that
Then if we l e t n ->oo i t i s e v i -
1969]
(2.23)
n lim
->oo|tU n /U n-i ! = I| r 1J| .
.
n lim
-> oo IUn-j. /U n-j-i
rJ
(j = 0 , 1 , 2, , t - 1) ,
so that
(2.24)
n lim
^ool Un /U n-tlJ
uu
S=0
in (2.9), we write
C xS ,
s
(2.25)
w=o
/ \ s=o
s=o
(2.26)
Z-^
s=o
n-i
s
b
s
C = ri Y ^ U
b = ri C
1
n
* / J n-s-i s
n-i *
s=o
49
50
[Feb.
where combining this with the r ^ 0 and are distinct (so that we may add that
th r, have distinct moduli), leads to the completion of the proof for Theorem
the
1.
From (2.7), (2.8), and (2.17), the following corollary is immediate:
Corollary. If
t
s=i
where the r
s=o
(2.27)
(n = 0 , 1 , 2,* )
as a function of the r .
s
SECTION 3
Let
-k
w=i
(CQ
r
oo
w=l
w=o
(k)
w
c
w x
s(x)
w=i
r=w
= Z E a^r^n+w-r"
-c ~xW_1 =
and
b=
Z a wa x;"7
W-2
W=2
19691
51
c 0 = 1, ci = aic 0 , c 2 = atcj + a 2 c 0 ,
t-i
3
t ~JL
w+i
t-w-i
w=o
and
p.. = a4(k + n - j )
m+i
(j = 1, 2, 3, , n) ,
b n
* " m^2k + n ~m ~ ^
(m = l , 2 , 3 , - - - , n - l )
then
nc(k) / c ( k ) = E /G
n ' n-i
n' n
(3.1)
(k, n = 1, 2, 3, ) ,
9
(3.1.1)
Pi
02
-1
P2
qa
-1
P3
q4
-i
P4
q5
-i
-1
n-l
*It should be noted that since the a ! s are constant for a fixed t, that the root
xi will be determined as a variable* since it is a function of the c and will,
n
of course, change values for different n-
[Feb
52
(3.1.2)
Proof.
(3.2)
G =
n
|P2
qs
0 |
1-1
P3
Q4
-1
p4
q5
-1
p5
q6
-1
V i
-1
Let
IvjfE
1 = 1
W=l
where the a
\\
and the c
Xw
(for a finite n) ,
' \ W=0
power, we write
vk
(3.3)
1" =
a xw
w
vw=o
w=i
(k=2,3,--),
(3.3.1)
(k)
1 = (1 - aix - bx'
\w=o
XW+J(K)
k(a lX + 2bx2) I ^
<w=o
nc k) x n + W(x)
1969]
53
(3.4)
ai(k
+ n - l ) c ^ + b(2k + n - 2)cf^
for
n = 2, 3, - , e 0 (k) = 1 and c | k ) = ajk .
k = 2, 3 , . . . ,
(k)
, we get
nc(k)
n
-70c
n-i
/i .
i\ _. M2k + n - 2)(n - 1)
= a!(k+n-l)+
Tgr
(n - l ) c
n-i
, .
v
(n,k = 02, 03, ) ,
~
n-2
= 1 and ci
= aik,
nc
(3 5)
(k)
q2
+, qs ,+ + , q n:- i .+ %. = K(n)
Jr =^ ^ ?r - ^- n -:i ^
'n
= En /Gn ;
In
\-k
TT-v-k-h-Zvw
1
w=i
i t i s always p o s s i b l e to solve for
w=l
=1+
00
- , ^
= 1+ >
w=i
(k) w
e x
w
54
(3.6)
when t = 2, 3,4,
nk)/cSi
K(n)
n/Gn
(k
and
[Feb.
n = 2s 3 f
' ' *)
or 5, if the r
( 1 - r )"k = f V
X
k )
(d<k) = l,
r^O)
w=o
Now, differentiating, we have
xkrf >
= f^wd^k)xw
dr'VM
w=0
'
w=i
d(k+1)
rk
n-i
so that
n
Jfwdfw=i
>
= r
n-i
jj(
w=o
k +
^k+n-w)
n-w-l)d<
w
and
and G
1969]
55
REFERENCES
1. L. M. Milne-Thomson, The Calculus of Finite Differences, Macmillan and
Co., Ltd., London, 1960, p. 526.
2. G. Chrystal, Textbook of Algebra, Vol. II, Dover Publications, Inc., New
York, 1961, p. 502.
The author wishes to thank L. Carlitz and V. E. Hoggatt, J r .
for their encouragement.
*.
Lock Haven, Pennsylvania 17745. This department especially welcomes p r o b lems believed to be new or extending old results. Proposers should submit
solutions or other information that will assist the editor. To facilitate their
consideration, solutions should be submitted on separate signed sheets within
two months after publication of the problems.
such that
m times
For s = o- + it
let
P(s) =
P" S >
F e b . 1969
57
oo
y~/(n)n~S = [H-P(s)]- 1 ,
n=l
00
\^b(n)n" s = [l-P(s)J" 1 .
n=l
D e t e r m i n e the coefficients
Show that
n-1
25
Z) 2 u^-
where F
is the n
p=l
1 = F4n+(n/3)(5n
*~ '
q=i I ^ T
Fibonacci number.
Put
WW
2
(1-ax
Show that
,
2 '
-bxy-cy )
\^A
B^
Put
\ ^
=
^
m 9 n=0
,nx
00
m n
y
_1
n
nx
l4)
= k - 2bx+ (b - 4 a c ) x > *9
58
[Feb.
00-
(1 - ax - bxy - cy)~
= }
Bm
nx
yn
m,n=0
Show that
X X , ^
n=0
{<l-bx)2-4aox}-*
H-152 Proposed by Verner E. Hoggatt, Jr., San Jose State College, San Jose, Calif.
the n
Fibonacci number.
1 k ^
- '
,m ^ m
n
n'
TV1
m ~
n >n=1
2 ~
copies
n = 7 afeck ,
a. = 0 or 1 if k = 1, 2, , p - 1
a
= 1
P
C.f. V. E . Hoggatt, J r . , and C. King, Problem E1424, American Mathematical Monthly, Vol. 67 (1960) r p . 593 and J . L. Brown, J r . , "Note on Complete
Sequences of Integers s n American Mathematical Monthly, Vol. 67 (1960), pp.
557-560.
1969]
59
SOLUTIONS
POWER PLAY
+ F2
+ 1 = 3F
(x-3y/2) 2 - 5(y/2)2 = -1 .
The general solution of (1) is therefore given by
(2)
for p =
60
[Feb.
TRIG OR TREAT
H-lll
Proposed by John L. Brown, Jr., Pennsylvania State University, State College, Pa.
Show that
[n/2 ] ,
L. = | |
h=1
k=l
|l +4cos2^l(rf)>for
n>l
K n IB odd, then
2n+l
L
2n+l
j=l '
2n+1
>
k=n+2
.h-tM.S*$j
2n+l
i=l
<
'
k=n+2
19691
61
"2n+l
5=1
(1)
-TT{ 1 + 4
cos
2 2j 1
7T )
2n + T " - 2 >
3=1
SimilarlyS
(2)
2n
1=1
Hence f r o m (1) and (2) we have the r e q u i r e d r e s u l t .
Also solved by Charles Wall, Douglas bind, and David Zeitlin.
VIVA LA D I F F E R E N C E
H-112 Proposed by L. Carlitz, Duke University, Durham, N. Carolina.
Show t h a t , for n > 1,
.5
, .n
n+1
.2
T5
L
- L , = 5L ^ L L ,(2LT
n
n-1
n+1 n n - l x Jn
.2
x- .- F - F - = 5 F , . F F - ( 2 F * +
n+1
n
n-1
n+1 n n-V
n
L7
- L 7 - L 7 n = 7L X 1 L L , v(2L 2 n+1
n
n-1
n+1 n n - l
n
a)
b)
c)
5(-l)n)
^ ' '
(-l)n)
' '
5 x( - l ) n ) 2
' '
v
F7
- F 7 - F 7 - = 7F . , - F F , x( 2 F 2 + v( - l ) n ) 2
n+1
n
n-1
n+1 n n - l
n
' '
d)
c) and d),
-x
-y
take x = L , y = L
2
7xy (x + y) (x
in the identity
2 *
+ xy + y )
Since
L 2 + L L - + L 2 n = 2 L 2 - 5 (x- l ) n ,
n
n n-1
n-1
n
' '
we get
2
L7
- L7 - L7
= 7L , n L L . (2L
- 5 v( - l ) n ) 2
n+1
n
n-1
n+1 n n - 1 1 n
' '
Similarly, since
62
[Feb.
F 2 + F F - + F 2 - = 2 F 2 + v(-l) n ,
n
n n-1
n-1
n
' *
we get
F
n+1
F7
= 7F . - F F - ( 2 F 2 + v( - l ) n ) 2 .
n-1
n+1 n n-V
n
' ' *
MINOR EXPANSION
H-117 Proposed by George Ledin, Jr., San Francisco, Calif.
Prove
F
F
2n+6
2n
(a - b - c + d) (a ~ b + c - d) (a + b - c - d)(a+b+c + d)
VVl+Vl>WFn+4+Fn+2>
-;
1969]
63
2n
= F L
n n
GOOD COMBINATION
H-119 Proposed by L. Carlitz, Duke University, Durham, N. Carolina
Put
,+i+k
! +k k
Yn - j + p - k \ / p - k + i \
64
[Feb.
(The formula
\2
H(2m f 2n)
'
( - ;
)2
where
m
l*"',l6
l
u
L
u
3 l 4 l 5 l6
3 U 4 U5 U 6
[l - U]_ - u 2 - u 3 - u 4 - u 5 - u 6 + U l u 4 + u ;1 u 5 + u 2 u 4 + u 2 u 5 + u2u(
+ u3u5 + u3u6 + u4u6 -
Ulu3u5
- u 2 u 4 u 6 J 2 - 4Ulu2u3u4u5u6 J"5
~Ur
5= "V
6 = ~U3
1969]
2
2
{(1 - u 2 - v 2 - w 2 ) 2 + 4 u2
v2
w''
H(m,n,p) u m v n vP =
65
2 P
m,n,p=o
2r 2r 2r
" w
00
00
E (-1)
.
nvr/2r\
2
u 2 r v 2 rw 2 r V ^ / 2 r + n V :U
n=0
i-O
2 ^ ( n )< +v
2r 2r 2r w
v w
X
r=0
\ ^
(2r + i + j + k)T. 2i 2j 2k
Z ^
(2r)U T jlkl
U,k=0
oo
V^
Z.J
2m 2n 2p
v w
2-f
m 9 njp=0
x
min(m,n 9 p)
\T*
,
ix
( }
r=0
(m + n + p - r ) l
rfe r i (m - r)l (n - r)l (p - r)i
C o m p a r i n g coefficients we get
min(m 9 n,p)
H(2m,2n.2p) =
r=0
(-l)r
,rL
(lf-
r H ( 7 - " r ) l ^ - r)l
2 2 n
+w )
L INTRODUCTION
Ke Subba Rao [4], and more recently V, C. Harris [l] have obtained
some identities involving Fibonacci Numbers F
Ft = 1,
F 2 = 1,
defined by
F n = F n _i + F n 2
n * 3 .
Our object in this paper is to obtain similar results for the generalized Fibonacci Numbers
as defined by A. F Horadam [ 2 ] ,
H4 = p, H2 = p + q
and
H = H
n
1
n~l
+ H
0
n2
n - ^ 3.
= - i 2 *JE
[lan-mbn]
[3]
where
1 = 2(p - qb), m = 2(p - qa)
and a and b are the roots of the quadratic equation x2 - x ~ 1 = 0 ,
1+
N/5
so that
66
1-
N/5
We call
ab = - 1 ,
a - b =
*JI>
ii
= pF
^ qF
- where' F
n-i
is the n
th
Fibonacci
number given by
an
-b
N/5
SECTION 2
In this section we obtain certain identities for the generalized Fibonacci
numbers. From result (9) of [2] we have the identity
H
r-1
+ H
~ *> H 2 r - 1 "
r = ^
ss
e F
2r-1 "
r=l
r = F n[<P +
+ qF
2q)H
2r-2
n
LJ
r=l
n *
eF
n-l 1 * M [ ( -
S0 tiiat
n
xarX
l f
" ^
'
68
(2)
E H2r-1 =
r=l
(3)
r=l
2n " H 2
2r = H2n
2n+l
+1 "
+ H
XX
E H 3r-2
(4)
2L H 3n" H 2 + H lJ
r=l
XX
(5)
r=l
xx
(6)
H3r
r=l
2 LH3n+2"H2j
n
(7)
(8)
ZH4r-3 =
r=l
2n-lH2n " H2
E H4r-2 =
r=l
2n H 2n
n
JLJ
r=l
4r-l
2n 2n+l
+ H
[Feb.
H4r =
r=l
2n+lH2n+l "
69
n
(11
>
X
r=l
2 r - 1 = I t H 2n
(H
2n-l
(H
2n + *W
+ 2ne + q(q
*W
" 2 *> )]
<
12)
2r = I tH2n+l
r=l
Let us now consider product terms as follows:
n
<
13)
2r-2H2r-l
r=l
n
<
14)
r=l
n
<15>
E
r=l
2r-lH2r+l
7 [H2n+l<H2n
+ H
2n+2 > +
3ne
~^
+ 2q)
n
<16)
Z H2rH2r+2 = \ [H2n+2<H2n+l
r=l
+ H
2n+3> "
3ne
" <P + ^
<^
r " W r
= <"1)r_kFk
70
(17)
(18)
E H2r~2H2iM-2
r=l
I [ H 2n + l ( H 2n + H 2 n + 2 } " 7 " ^
(19)
E H 2r-l H 2r + 3
I E H 2n + 2 (i Wl
+ 1 0
* 7ne " ^ H 3 p + q ) ]
^ ^
r=l
Evaluating the quantity
i r H k + i H , + 2 we get
H k H k + l H k + 2 = Hk+1 + ( - 1 ) k _ 1
(20)
eH
k+l
Therefore
H
2r-lH2rH2r+l
+ eH
2r
+e
Hence
n
n
H
E
2:r-1 2 r 2r+1 = E
L
r=l
r=l
E H2r '
r=l
< 21)
(22)
E
r=l
+e(H
2n+l "
i>]
E^iWr'X^^-^A-iH
r=l
1969]
Now
3
Ho2r = (pF
F 0 + qF0
2r ^ 2 r - l- ) .
On expanding the right side* taking the sum from r = 1 to n and simplifying
we get the relation
n
23
< >
E
r=l
n
<24>
E
r=l
2rH2r~l
I [ ( H 2 n H L l " *\>
+e
2nl "
n
<25>
E
r=l
2rH2r^l
J ^ a n - l ^ L l " Hlq2}
+e(H
2 n " H 2 >1
n
<26>
E H L-l=lf ( H 2n-^ )
r=l
+ 3 e < H
2n-^
n
^
E
r=0
rH
= nH
(28)
E <-1>rrHr = t("1):
r=0
(n + 1)H
n-1
H n - 2 + (3q - 2p)]
n
(29)
( - r H a r = | [ (-D n + 1 (H 2 n
r=0
+H
2 n + 2 > " ( P + *1
Feb. 1969
(30)
r=0
(31)
rH
r=0
(32)
rH
r=0
(33)
<-l)rrH2r = ^(-^((n + D H ^ + n H ^ ) - ^ - ^ ) ]
r=0
n
(34)
<-r'H2r+1 = i[(-l)n((n+l)H2n+1+nH2ri+3)-Hl]
r=0
It is easy to see that the list of identities given by K. Subba Rao can be extended
to Fibonacci Quaternions defined by
Q = F + iF , . + JiF ^ + kF ^ 0 .
^n
n
n+1
n+2
n+3
The author is very grateful to Dr. J. Sethuraman for valuable suggestions.
REFERENCES
1. V. C. Harris, "On Identities Involving Fibonacci Numbers," The Fibonacci
Quarterly, Vol. 3, No. 3, 1965, pp. 214-218.
2.
3.
4.
1.
INTRODUCTION
One of the most famous of all geometric figures is the Golden Rectangle,
which has the ratio of length to width equal to the Golden Section,
<t>= (l + V 5 ) / 2 .
The proportions of the Golden Rectangle appear consistently throughout c l a s s i cal Greek art and architecture.
fully known by either artists or psychologists, the Golden Rectangle holds great
aesthetic appeal. Surprisingly enough, the best integral lengths to use for
sides of an approximation to the Golden Rectangle are adjacent members of
the Fibonacci series: 1, 1, 2, 3, 5, 8, 13, , and we find 3 x 5 and
5x8
< = a + V5)/2 ,
then we will call the triangle a Golden Triangle.
(See [2], [ 3 ] . )
Now, consider the isosceles triangle with a vertex angle of 36. On bisecting the base angle of 72, two isosceles triangles are formed, and ABDC
is similar to A ABC as indicated in the figure:
73
74
[Feb.
Since AABC-ABDC,
AB
BD
BC
DC '
or,
x
I =
x
y - x
so that
j r - yx - xd
= 0
l! _ I > i
Dividing through by x 2 / 0,
f = (1 + V5)/2 = 0
as the positive root, so that A ABC is a Golden Triangle. Notice also, that,
using the common altitude from B, the ratio of the area of AABC to AADB
i s <f>*
1969]
75
shows
4> = (1 + \/5)/2 .
Also, in a regular pentagon, the angle at a vertex between two adjacent diagonals is 36. By reference to the figure above,, the ratio of a diagonal to a side
of a regular pentagon is also <j>.
2. A TRIGONOMETRIC PROPERTY OF THE ISOSCELES GOLDEN TRIANGLE
The Golden Triangle with vertex angle 36 can be used for a surprising
trigonometric application. Few of the trigonometric functions of an acute angle
have values which can be expressed exactly,, Usually, a method of approximation is used; mostvalues in trigonometric tables cannot be expressed exactly
as terminating decimals, repeating decimals, or even square roots, since they
are approximations to transcendental numbers, which are numbers so i r r a tional that they are not the root of any polynomial over the integers.
The smallest integral number of degrees for which the trigonometric
functions of the angle can be expressed exactly is three degrees. . Then, all
multiples of 3 can also be expressed exactly by repeatedly using formulas
such as sin(A + B). Strangely enough, the Golden Triangle can be used to derive the value of sin 3,
In our Golden Triangle, the ratio of the side to the base was
y/x = (1 + V5)/2 .
Suppose we let AB = y = 1. Then
1/x = (l + V 5 ) / 2 ' ,
or,
x = (\/5 - l)/2 .
Redrawing the figure and bisecting the 36 angle,
76
[Feb.
= I =
AC
Then,
V5 - 1
J_
2*
o
cos 18 =
y/lO + 2V5
\A/5<ft
= V l . Yl _ I . YI
2
V6 - V2
cos 15
o _ \/6 + \/2
4
s i n(18-i5)
= ( ^ i ^ ^ p l ) .
1969]
77
= (1 + y/E)/2
c
w
/-w
If rectangles ABCD and BPRA have the same ratio of length to width, then
k =
= 1
/- w
w "
which has
(l+V5)/2
as its positive root. If
78
[Feb.
1 = (1 + V 5 ) / 2 = <t>,
W
then
w
i. i
w
0-1 = 0
from rectangle
figure:
II
/-*
JC
Then
area ABCD
area PCDR
But,
fw
wi [ _ y
ABCD
as in the
1969]
'w
=1 -
wTT^x)
79
if and only if
w
/-
or w = / - x or PCDR is a square. Thus, our second theorem is a consequence of the first theorem,
Analogous theorems hold for Golden Triangles.
Theorem.
a/b = k > 1. A triangle with side equal to b can be removed to leave a t r i angle similar to the first if and only if k = (1 + \ / 5 ) / 2 .
Proof. Remove AABD from A ABC.
DC
AC
or
b
a
a - b
1969]
80
a/b = (1 + VB)/2 ,
then
DC/AC = (a - b)/b = a/b - 1 = (\/5 - l ) / 2
and
AC/BC = b / a =* 2/(1 + \/5) = (VJ5 - l ) / 2 = DC/AC .
Since
triangle similar to the first can be removed to leave a triangle such that the
ratio of the areas of the original triangle and the triangle remaining is k, if
and only if k = (1 + y/Z)/2a
Proof.
BC/AC = AC/DC = k.
If the ratio of areas of the original triangle and the one remaining is k,
since there is a common altitude from A,
1969]
ir = area ABAC
areaABDA
(BC)(h/2)
(BC - DC)(h/2)
81
BC/AC
_
k
BC/AC - DC/AC " k - 1/k
'
BC/(BC - DC)
(1 + \ / 5 ) / 2 ,
becomes
(1 + \/5)/2
when divided
through by AC and then simply substituting the values of BC/AC and DC/AC.
If
k = (1 + y/E)/2
= BC/AC ,
BC/AC
BC/AC - DC/AC
k
k-x
which leads to
x = k - 1
or
so that
AC/DC = (1 + V ) / 2
and ABAC is similar to A ADC.
4. THE GENERAL GOLDEN TRIANGLE
Unlike the Golden Rectangle 9 the Golden Triangle does not have a unique
shape. Consider a line segment CD of length
82
[Feb.
= (1 + \ / 5 ) / 2 .
Then9 ED = 0 - 1 and
CE/ED = 1 ( 0 - 1) = 0 ,
CF/DF = (20+ 1)(0 + 1)
= 0 3 /02 = 0 ,
Then CH/HD = 0 ,
= h02/2
and when AHDG is removed* the area of the remaining ACHD is h.0%
that the areas have ratio 0 .
so
are an infinite number of Golden Triangles because H can take an infinite number of positions on circle G.
1969]
83
If we choose H so that CH = <f> + 1, then we have the isosceles 36-7272 Golden Triangle of decagon fame. If we erect a perpendicular at D and
let H be the intersection with the circle of ApoUonius* then we have a right
golden triangle by applying the Pythagorean theorem and its converse. In our
right golden triangle ACHG, CH = $ W s
vW/29
(V?/2)$, (V$/2)tf* .
Before going on9 notice that the right golden triangle ACHG provides an
unusual and surprising configuration.
pairs of angles of ACHG and ACDH are congruent, yet ACHG is not congruent to ACDH! Similarly for ACDH and AHDG.
(e-
4 + i
- _ - _ _ ^
<fr:
l:<f>
84
[Feb.
and four of the six faces of the cuboid are Golden Rectangles.
If two cuboids of dimension
0 " 1 X 1 X 0""1
are removed from the Golden Cuboid, the remaining cuboid is similar to the
original and is also a golden cuboid,
and
then
b : c : d = <f>
so that
c = d<f), b = d$9
a = d<}? .
(4> -< )d
=
3
<fr3
3
< -l
2 + V 5 =_ 3 + V 5
1 + V5
,2
1969]
85
y - x = y - 1/y = k,
or s
y2~ky-l
= 0 .
-H
Let us consider only Lucas golden-type rectangles. That is, let k = L
2m+l'
st
where L rt
, is the (2m+l)
Lucas number defined by
2m+l
'
L
= 2, L
= 15 L
= L
+ L 0
n-1
n-2
A known identity is
- x- 1 = 0
86
[Feb.
In our problem, if
2m+1 '
then
y 2 - ky - 1 = 0
becomes
y2"-L2m+1y-l
= 0
so that
2m+l
y
= a
or
a
2m+l
y =P
but
y
2m+l
= a
or ky - y
=1
or y - ky + 1 = 0.
1969]
k-y
87
->j
2m
Now, let k = L
; then y =a
2ni
o r y = fS
2m
. H e r e , of c o u r s e , y =/3
zm
so that
k ~ y = L0
2m
n2m
- j3
=
c = dk,
2
b = dk ,
3
a = dk
Then
the volume r e m o v e d is
bed = k 3 d 3
88
[Feb.
k3-i
(k -kV
Now, let this ratio equal
k2/LQ = k2/2
which leads to
0 = k 3 - 2k - 1 = (k + l)(k 2 - k - 1)
with roots
k = -1,
(1 Vij/2
and having
k = (1 + VS)/2
as its only positive root.
Now consider ahypercuboidin ahyperspace of 6 dimensions, with dimensions a : b : c : d : e : f = k. Remove a hypercuboid of dimensions
b :c :d :e :f :g
abcdef
_
abcdef - bcdefg ~
2 1
g k
_
k
6 21 . 15 " . 6 .,
g /T(k - k )
k - 1
since
2
3
4
5
6
f = kg, e = k g, d = k g, e = k g, b = k g, a = k g
1969]
k6-l
which leads to
_ 4k3 - 1 = 0
with roots
2
2
where a> and a> are cube roots of unity. Then
k = a = (1 + VE)/2
is the only positive real root.
Suppose we have a cuboid in a hyperspace of 4m + 2 dimensions. Let
this have edges
=
1
: a
2
: a
3
4m+2
a : a : a : : a
:a
2
3
4
4m+2
4m+3
. 4m+3-n
= k
a. ^ 0
n
4m+3
90
2
a
4m
[Feb.
k4m+2
k4m+2 - 1
4m+3
,
'
+1
k 4 m + 2 - L9 + 1 k 2 m + 1 - 1 = 0
2m+l
whose only positive root is
a = (l + \ / 5 )/2
.
2
The proof is very neat* Since a(3 = -1 for a and /3 the roots of x - x - 1
= 0 and since L = a +Bni we can write
n
^ '
,
= a2m+l
2 m + l + f l 2 m + l _ a 4m+2
p
'
= a2m+lL
_ a 4m+2
2m+l
fe4m+2
2 m + 1 - L2m+1)
- 2m+l
2m+lwl2m+l 02m+lx
= (k
-a
)(k
-0
) = 0 .
Thus 9 k-t(js).,
(2m+l)
st
k = a = (l + \ / 5 ) / 2
is the only positive real root.
1969]
91
l "
la29"a4m+2 ~
- 4m+2 1 2 3
^4m+l
V 0 - a 0 a 0 aA , n a , , 0 = a
a a a * a
2
2 3
4m+2 4m+3
making the volume of the cuboid remaining
V
where T
XT
l "
is the n
1V
"" X
T
L
^,2m+l
2m+Ia
volume
4m+2 1 2 3
4m+l ,,
a a a a
( L 2m+1
2m+l,
* '
Thus we have generalized the Golden Cuboid of Huntley [5] and also the goldentype rectangle of Raab [ 6 ] .
REFERENCES
1. William R. Ransom, Trigonometric Novelties, J. Weston Walch, Portland,
Maine, 1959, pp. 22-23,
2. Verner E. Hoggatt, J r e , Fibonacci and Lucas Numbers
Company, Boston, 1969.
[Continued on p. 98. ]
Houghton Mifflin
1.
INTRODUCTION
(n 2)
AA
"
" *& .
-R
a - p '
aQf
"
* " a - p
u n (p,q)
w n (2,p; p,q)
v n (p,q)
w n ( r , r + s; 1,-1) = h n (r, s)
w (1,1; 1,-1) = f '= u (1, -1) = h (1, 0)
n '
'
n
n.
n
w (2,1; 1,-1) = 1 = v (1,-1) = h2 (2, -1)
L
n
n
n
'
wherein F n and L n are the famous Fibonacci and Lucas sequences respectively.
92
Feb. 1969
93
SECTION 2
In this paper our object is to derive some relations connecting the sums
of the above sequences up to n terms.
We shall derive aformulafor the sum of the most general sequence {w }
and thereby obtain the sums of the other sequences,
Theorem:
A
Lwr =
a +
r=o
bTQ - aqT
1 -p + q
where
Tn = 1 - An
and
A = u - Hqu
n
n
n-i
Consider
r=o
r=o
b - a/3 an+1 - 1
a - p
a - 1
aa - b pR+i - 1
a - p
p - 1
(a + ft) = p,
ap = q,
a p = d
[(a + b - ap) + aq(u n _ i - q u ^ ) - b(u n - q u ^ J j / d - p + q)
Set.
un - qu
^ n-i = A n
94
[Feb.
let now
1
" \
"
(1 )
bT - aqT 4
^ T w = a + ^n r - r r rn-i
^
r
1 - p + q
r=o
2 ur(Ps q)
r=o
is obtained by letting a = 1, b = p in (1)
n
LVP'^
pT
1+
- qT
l - p+q
r=o
(2)
J]u r (P.q) =
r=o
n+i/(1
_P + q)
'"
1969]
95
E V n (p' <*>
r=o
can be obtained by putting a = 2, b = p, p, q in. (1)
r=o
(3)
r=o
In particular,
Sw r (l, 1; 1,-1)
= 2 F r = S u r ( l f - 1 ) = Sh r (l f 0)
and
S w r ( 2 , 1 ; 1,-1) = XL r = s v r ( l f - l ) = S h r ( 2 , - 1 ) .
y^ urd, -i)
a)
r ==o
(1 - u
Therefore
) + (1 - u )
2ut(li.l)-l+Hf?T^^
r=o
i-
[d-u+1)
n+i
a-u
)]
96
2>*a.-i> = w 1 - * * * - 1
[Feb.
[ ]
* "'V
r=o
This can be verified for any n.
(ii) To get 2 v (1 -1) let a = 2, b = p = 1, q = - 1 in (1). Here also
A. = u ... So
n
n+i
*
(l-u
I/ r <l,-l) = * +
)+ 2(l-u)
f-1-1
r=o
= 2 - L[ 3 - 2un - . un+i
x i J]
= u + u _^ - 1
n
n+2
= v . - 1
n+2
(LJ
n
]Cv p , q )
r=o
set
a = p, b = p + q, p = 1, q = -1
K
Then
Vi
Vl
1969]
97
J j i r ( p , q ) = p - [ ( p + q ) ( l - F n + 1 ) + p(l - F n )J
r=o
= (P + ^ F n + l
PFn~
= (pFn+2+qFn+1)-(p
(p + q)
+
q)
] T h r ( p , q ) = h n+2 - (p + q) by [2]
. . . <Lm)
r=o
l s 1.5 (1..), (1...) can be proved for all (+ve) integers n by induction. We
l
ii
ill
k
--
bT, - aqT,
r
1 - p + q
r=o
Next let us add w,
k+1
2 ^ ^w r
bT - aqT
=
P
r-i +
- a- + ~
i _ p +r -"q
w,
"k+i
r=0
= a +
' b ( 1 " \ + ^ k - i ) ~ a q ( 1 ~ Vi
1-
- p + q
^Uk-2 )
+ Aa k + 1 + B|S k+1
*,
TT
(4)
2_, r
=a+
b(l-u
+qU
)-aq(l-u
+quk_2)
rrrrr
r=o
+ bu k - a q u ^
C
"a
^!1
Z^
r=o
1 - -P + q C b(1 " V l
b T . ^ - aqT.
w =a+
r
kL_U&
1 - p + q
+ qU
k) "
aq(1
- \
Vi}3
98
Feb. 1969
Equation (4) is of the same form as (l) f with k replaced by k + 1. Hence, etc*
Similarly other results can be proved for all positive integral values of
n.
REFERENCES
1* A. F, Horadam, "Basic Properties of a Certain Generalized Sequence of
Numbers, " The Fibonacci Quarterly, Vol. 3, 1965, pp. 161-176.
2. A. F. Horadam, "Complex Fibonacci Numbers and Fibonacci Quaternions,"
Amer. Math. Monthly, Vol. 70, 1963, pp. 289-291.
3. K. Subba Rao, "Some Properties of Fibonacci Numbers," Amer. M.ath.
Monthly, Vol. 60 (1953), pp. 680-684.
* * *
M A R RECURSION RELATIONS
LESSON THREE-M BINE? FORMULAS
BROTHER ALFRED BROUSSEAU
St. Mary's College, California
T ,.,
n+1
T + T , %,
n
n-1
(2)
or
x2 - x - 1 = 0
(3)
(4)
r = ^
and
s =
r + s = 1 and
rs = -1 .
It follows from what has been developed in the previous lesson that the terms
of any Fibonacci sequence can be written in the form:
(5)
T n - a,xn + b s n ,
99
100
[Feb.
-UK
5 = a r + bs
These give solutions:
aa
15^5.
.b
aana
nd
1Q
1 5 - V5
1Q
so that
10
10
with solutions
1
a
v^
and
b = --i
\/5
so that
1969]
(6)
101
n
n
- Z-L
1 and L = 3 ,
1 = a r + bs
3 = a r + bs
(7)
w
= r +s
*s
1 :1- x ,
i = o .
r =
i + Vs"
zr
102
[Feb.
Vs"-1
is also considered the Golden Section Ratio by some authors. This is a matter
of point of view: whether one is taking the ratio of the larger segment to the
smaller segment or vice-versa.)
1
1
2
3
5
8
13
21
34
55
1
3
4
7
11
18
29
47
76
123
4 L 4 = 3 X7 = 21 = F
F C L , = 5 X 1 1 = 55 = F.
10
5 5
and in general it appears that:
1969]
Why is this s o ?
FQ
.
2n
T.
=
2n
103
. n
n.
(r - s )
V5 = J \/5 -J, /=
. /r
F?
2n
- s
, n . nv
^ T
(r + s ) = F L
v
;
n n
Example 2
kn
F
kn
kn
, k , n ., k / 1
ZA =
(LA^ZJ^L.
V5
V5
has a factor
k
"
k
-
Vs"
which p r o v e s that if k is a d i v i s o r of the s u b s c r i p t of a F i b o n a c c i n u m b e r
F
then F, divides F .
k
m
m9
Example 3
By taking s u c c e s s i v e
values of
k9
formula:
F +
n+k
n-k
2 =
n
_D^+1F2
'
k
(
x
n+k n - k
^2
n
P + k
-s
y/r~
n + k
'
rn-k-sn^
Fe
This gives:
(r Q - s n ) 2
5
- k s n + k - r2n + 2 r V - s2n
the
104
Feb. 1969
PROBLEMS
1,
Prove that
2.
LQ = L 2 + 2 ( - l ) n + 1 .
2n
n
* '
Using the Binet formulas, find the value of:
L F - - F L - .
n n-1
n n-1
3.
4.
Fr
5.
6.
7.
3n = F V(
).
= F (
ir
nv
).
).
5n
3n
).
in terms of
2
2
determine an equivalent single Fibonacci number for F + F ,- using the
Binet formula.
[Continued on p. 106. ]
EiRiTA FOR
I L B i AUEHA CONSTRUCTED FROM FIBONACCI SEQUENCES
J. W. GOOTHERTS
Lockheed Missiles & Space Company, Sunnyvale, Calif.
Please make the following changes in the above-entitled article, appearing in Vol. 6, Noa 5, November 1968:
On page 36., change the eighth line from the end to read:
Definition 1.5. For U , V G ? , UV = (u0v0 + u ^ u ^ + UJVQ + UJVJ).
Equation (3) on p. 38 should read:
au + bu
=0
n
m
(38)
au . - + bu , - = 0 .
n+1
m+1
On p. 42, 11 lines from the end, change the " F " to a script *3\
On p. 49, in the equation preceding Eq. (10), change a. to ft).
BOOK REVIEWS
BROTHER ALFRED BROUSSEAU
St. Mary's College, California
The booklet takes up aspects of number theory that are within the range
106
Feb. 1969
BOOK REVIEWS
point: figurate numbers, the Euclidean algorithm for finding the greatest common divisor, perfect numbers, amicable numbers, the Pythagorean problem,
ancient systems of numeration, and Mersenne numbers.
On the other hand, up-to-date developments are not neglected. There is
an interesting discussion of the largest primes discovered by the factorization
of Mersenne numbers. In connection with number bases, computers and their
mode of arithmetic are introduced.
Finally, the author has introduced interest features throughout the book:
magic squares, games with digits, days of the week as related to congruences,
tournament schedules.
The book contains problems to be solved and has a section entitled "Solutions to Selected Problems. t f
The list price is $1.95.
2n+^)
5.
6.
- 10 + V 5
lL
= 2
n + 10_-_V5 s n
5
-n+l
n
10.
2n+l
14
19
13
22
21
1
20
17
15
24
25
18
23
10
16
12
4
= ( 1) P
"
aPb
= 1, and P
be the n
Fibonacci number, G = F , - 2n9 and H be the
n
* n
4n
' n
remainder when G is divided by
J 10,
n
107
108
[Feb,
ing block.
(b) The l a s t two digits of G g and G - 4 give Fibonacci n u m b e r s 34 and 89
respectively.
A r e t h e r e any o t h e r c a s e s ?
be the n
Fibonacci n u m b e r and j g
} any s e q u e n c e .
Show
that
n
/ ,tek*2
k=l
+ g
k t l ~gk)Fk
n+2
n+lFn+l
"
= [ ( n 2 + n + 2 ) F 2 + 2 - (n 2 + 3n + 2 ) F 2 + r (n 2 + 3n + 4 ) F 2 ] / 2 .
k=l
be the n
E u l e r totient.
Showthat
be the n
= 2 P ' - + P .- S h o w t h a t
n+1
n
P e l l n u m b e r defined by
P
J
= 1,
'
P 0 = 2, and P ^
2
'
n+2
1969]
n+1
+ p
p
2n+l
109
ANOTHER P E L L IDENTITY
B-137 Proposed by Phil Mana, University of New Mexico, Albuquerque, N. Mex.
Let P
- <-Dn.
and s = 1 - V2
form
9
e1(r")
with constant c ! s .
+ c 2 ( r s ) n + c3(s )
n+3
Vn+2
lVl
+ k
0yn
'
and 3 .
Also solved by Clyde A. Bridger, Herta T. Freitag, J. E. Homer, John Ivie, D. V. Jaiswal
(India), Bruce W. King, C. B. A, Peck, A. G. Shannon (Australia), M. N. S. Swamy (Canada),
Gregory Wulczyn, Michael Yoder, and the proposer.
there
110
[Feb.
1.
) is 1.
T h e r e e x i s t s an
n-by-n
F,+
k+n^
k+l
k+2
0
F
*k+2
F
k+3
0
k+3
0
k+4
0
"k+n-2
0
k+n-1
0
k+n
\+n-l
k+n-2
...
k+3
"k+3
k+2
k+2
k+1
+-
1 1 + b- + + b
for n = 1,2, .
1969]
a n + 1 < 1 + S n - a.
for
1< i < 6
and
n > i
and
(B)
a n + 1 > 1 + S n - a.
for
i > 6
and some
n>i,
V l
= *il >
1 + F
< 1 - 2 2 = - 3 , and
^-l- F k~l = 1 + S 2 k - 1 =
1+ S
2fc- a 2k=
1+S
n"ai '
(A) is
m>
a ^ = F 2 , . < 1 + S - a = S - 3 - 2F F , - - ' 3
n+1
m+1
n
6
n
m m+1
or if
F
or if
F
_ (F - F -)'
> 3,
m+l v m
m-17 - 9
112
Feb. 1969
F +F
F
= F
m n
m-1 n-1
m+n-1
>F
ab
a+b-1
= F F + F
F
>F F
a^b
a - l V r
a^b'
is an even
n
perfect number.
Solution by the proposer.
Recall that an even perfect number greater than 6 must leave a remainder
of 1 upon division by 9 and must be a multiple of 4. An even perfect number
greater than 28 must be a multiple of 16.
If F ' s 1 (mod 9), then n = 1, 2, 10, 18, or 23 (mod 24); if 16JFn
then n = 0 (mod 12). These two sets have no common elements.
If L n = 1 (mod 9), then n = 1 or
n == 3
SUSTAINING MEMBERS
*H. L . A l d e r
V. V. A l d e r m a n
G L e A l e x a n d e r s o n
R, H, Anglin
^Joseph Arkin
Col. R . S 8 B e a r d
*Marjorie Bicknell
J . H. Biggs
F r a n k Boehm
J . L0 B o h n e r t
M. B . B o i s e n , J r .
C s Ae B r i d g e r
Leonard Bristow
*Bro e A 9 B r o u s s e a u
* J . L0 B r o w n , J r .
N . S. C a m e r o n
L. Carlitz
P , V. Char land
P . J e Cocuzza
J . L. Crenshaw
D. E . Day kin
J . W. DeCelis
F . DeKoven
J . E . Desmond
A e We Dickinson
N. A. Draim
D. C. Duncan
M. H. E a s t m a n
C9 F e E l l i s
Merritt Elmore
H. W. Eves
R 8 A. F a i r b a i r n
A 8 J 8 Faulconbridge
*H8 H e F e r n s
D. C. F i e l d e r
E. T. Frankel
H 8 M8 Gehman
G9 R 8 Glabe
E . L . Godfrey
*H9 W8 Gould
Nicholas Grant
G, B . G r e e n e
*J. H9 Halton
V. C, H a r r i s
Cletus H e m s t e g e r
*A. P 9 Hillman
B r u c e H 9 Hoelter
*V. E 9 Hoggatt, J r e
*A. F . H o r a d a m
D 8 F e Howe lis
*Dov J a r d e n
*S. K s J e r b i c
Kenneth Kloss
Sidney Kravitz
George Ledin, J r .
Hal Leonard
*C. T . Long
A, F . Lopez
F 9 W. Ludecke
J 9 S8 Madachy
*Sister M . D e S a l e s McNabb
John Mellish, J r .
M r s . Lucille Morton
Stephen Nytch
F8 J8 Ossiander
Miss Ann Pape
M 8 M 6 Risque no
*D9 W8 Robinson
*Azriel Rosenfeld
T8 J . Ross
F , G . Roth we 11
H. J . Schafer
J . A , Schumaker
B . B . Sharpe
L , R . Shenton
G8 Singh
A , N . Spitz
H 9 L 8 Umansky
*C 8 R. Wall
*L8 A . Walker
V. T 8 White
P . A 8 Willis
C h a r l e s Ziegenfus
^Charter Members
THE CALIFORNIA
MATHEMATICS COUNCIL
VALLEJO UNIFIED
SCHOOL DISTRICT