5 3

1967 AS THE SUM OF SQUARES
BROTHER ALFRED BROUSSEAU, St. Mary's College, California
With t h e coming of a new y e a r , t h e r e i s always a tendency to find out

whether the n u m b e r that indicates it h a s s o m e s p e c i a l m a t h e m a t i c a l p r o p e r t i e s .
1967 i s , of c o u r s e , no square 0
But t h e r e is a t h e o r e m that s t a t e s that any
i n t e g e r can b e r e p r e s e n t e d a s t h e s u m of at m o s t four s q u a r e s . Let u s i n v e s t i gate t h e m i n i m u m n u m b e r of s q u a r e s that will add up to 1967.

F i r s t , we note that t h e s q u a r e of e v e r y even n u m b e r is divisible by 4 and
t h e s q u a r e of e v e r y odd n u m b e r on being divided by 4 gives a r e m a i n d e r of one.
If 1967 i s to b e t h e s u m of the s q u a r e s of two n u m b e r s , one m u s t b e odd and
one even, o t h e r w i s e t h e r e could not b e an odd sum.
But t h e s u m of two such
s q u a r e s on being divided by 4 would give a r e m a i n d e r of one, while 1967 on

being divided by 4 gives a r e m a i n d e r of 3. T h u s 196 7 cannot b e t h e s u m of two
squares.
F o r t h r e e s q u a r e s , 1967 would have to be the s u m of the s q u a r e s of t h r e e
odd n u m b e r s s i n c e the r e m a i n d e r on division by 4 i s 3. Now the s q u a r e s of the
odd n u m b e r s ending in 1, 3, 5, 7, 9 end r e s p e c t i v e l y i n l , 9, 5, 9, and 1. T a k ing t h e s e endings t h r e e at a t i m e , it can be easily shown that only the c o m b i n a tions 1 , 5 , 1 and 9, 9, 9 give a l a s t digit of 7.
So one s y s t e m a t i c way to p r o -
c e e d i s to c o n s i d e r the v a r i o u s c a s e s c o r r e s p o n d i n g to n u m b e r s ending f i r s t of
all in 5, n a m e l y : 5, 15, 25, 35. Subtracting out 2 5 , 2 2 5 , 625, and 1225
r e m a i n d e r s of 1942, 1742, 1342, 742.
1
9
11
19
21
29
31
742
741
661
621
381
301
gives
We can p r o c e e d by table as follows:
1342
1341
1261
1221
981
901
501
381
1742
1741
1661
1621
1381
1301
901
781
1942
1941
1861
1821
1581
1501
1101
981
We can stop at 31 s i n c e this b r i n g s u s to the halfway point with the l a r g e s t

n u m b e r 1942.
Since no s q u a r e s a p p e a r in the t a b l e , this d i s p o s e s of the p o s -
sibility 1 , 5 , 1 as endings.
Similar c o n s i d e r a t i o n s apply for 9, 9, 9. T h u s , the
only possibility i s four s q u a r e s .
One such r e p r e s e n t a t i o n i s :
1967 = 6 2 + 9 2 + 25 2 + 35 2
* *** *
208
ANOTHER GENERALIZED FIBONACCI SEQUENCE

MARCELLUS E. WADDILL A N D LOUIS SACKS
Wake Forest College, Winston Salem, N . C . , and University of Pittsburgh, Pittsburgh, Pa.
1.
INTRODUCTION
Recent issues of numerous periodicals have given indication of a renewed

interest in the well-known Fibonacci sequence, namely
(1)
1, 1, 2, 3, 5, 8, - , C n ,
where
Cn = Cn-i' + Cn-2 . n > 3, with C*1 = Cl2 = 1 .
Some recent generalizations have produced a variety of new and extended results.
A search of the literature seems to reveal that efforts to generalize the
Fibonacci sequence have consisted of either (a) changing the recurrence relation while preserving the initial terms, or (b) altering the initial terms but
maintaining the recurrence relation.
A combination of these two techniques
will be employed here.

Heretofore, all generalizations of the Fibonacci sequence appear to have
restricted any given term to being a function (usually sum) of the two preceding
terms.
In this paper we shall extend this by considering sequences in which
any given term is the sum of the three preceding it.

Since the set of all algebraic integers, i.e., all y such that y satisfies
some monic polynomial equation,
/ \
p(x) = x
+ a
+*<>+
c\
aiX
aQ
= o,
with integral coefficients and of degree greater than zero, is an integral domain
under the operations of addition and multiplication, it was considered worthwhile to examine sequences in which the initial terms (hence all succeeding
terms) are algebraic integers. It will be shown that certain special cases of
such sequences a r e especially useful in the examination of the more general
case.
209
210

2.
THE GENERALIZED SEQUENCE { P
[Oct.
Specifically we c o n s i d e r the sequence
(2)
| P n | = P0, Pi, P2, P n ,
w h e r e P 0 , P i , P 2 a r e given, a r b i t r a r y a l g e b r a i c i n t e g e r s , not all z e r o , and

(3)
= P
n
+ P
n-l
+ P
n-2
. n > 3.
n-3'
It will a l s o be convenient t o c o n s i d e r a companion sequence, so to speak,
(4)
{Rj
= R 0 , Ri, R 2 . . Rn> "
where
R0 = P i - P 0 ,
(5)
Ri = P 2 - P i .
= P
n
and for n > 2,
, + P 0
n-l
n-2
F r o m (5) and (3), when n > 5,
we have
R n = P n - l + P n-2 = (P n-2 + P n-3J + ( P n-4 + P n-30 ) + (P n-4 + P n-5J

= R
+ R
n-l
+ R
n-2
n-3
Using (5) and (3) further, we have
R4
R3 + R2 + Ri
R3
R2 + Rj +
RQ
Hence for n > 3,
(6)
= R
n
+ R
n-l
n-2
+ R 0 .
n-3
1967]
211
Thus | R | is actually t h e s p e c i a l c a s e of (2) in which R 0 = P i - P 0 ,

-Rl
2 - p i
+ p
T n e
evident in the development of j P
usefulness of the sequence j R
i will be
I that follows.
Two other s p e c i a l c a s e s of (2) should be mentioned at t h i s t i m e ; n a m e l y

the c a s e s in which P 0 = 0,
P j = P 2 = 1 and P 0 = 1,
~Pt = 0,
P2 = 1 r e -
spectively, to give the s e q u e n c e s

(7)
0, 1, 1, 2, 4, 7, 13, 24, 44, , K n ,
and
(8)
1, 0, 1, 2, 3, 6, 1 1 , 20, 37,
We s e e i m m e d i a t e l y t h a t L 0 = Ki - K 0 ,
(9)
= K
n
Hence we might call J K
} a
+ K
n-l
P -type
' , L ,
'.
Lj = K 2 - K^,
and for n > 2,
n-2
sequence and
JL
| an
-type
sequence,,
The sequence JK
Til.
| w a s defined and d i s c u s s e d briefly b y M. Agronomoff
T h e following t h r e e r e l a t i o n s involving v a r i o u s t e r m s of t h i s s e q u e n c e
w e r e d i s c o v e r e d and proved by h i m :
(10)
x
'
x
(11)
'
(12)
x
'
= K
n+p
K + (K
+ K )K
+ K K 0 ,
p+l n
p-l
p n-i
p n-2
2
K 2ri
<* n = K n - i, + Knx (Kn + i + Kn - 1 + Kn _ 2 ' ) ,
, = K2 + K 2
+ 2K
K
2n-l
n
n-l
n - i n-2
T h e r e i s only one b a s i c identity h e r e b e c a u s e the l a t t e r two a r e
evidently
s p e c i a l c a s e s of the f i r s t one upon setting p = n and p = n - 1 r e s p e c t i v e l y .

F u r t h e r , it w a s conjectured in [ l ] that even though t h e sequence (7) w a s
a F i b o n a c c i - t y p e sequence, it quite p o s s i b l y would p o s s e s s few of t h e i n t e r e s t ing p r o p e r t i e s which t h e Fibonacci sequence h a s , and even if it should,
such
p r o p e r t i e s would b e much m o r e difficult to find due to the m o r e complex n a t u r e

of the r e c u r r e n c e r e l a t i o n d e t e r m i n i n g t h e sequence.
212
[Oct.
We t u r n now to an investigation of the sequence (2) and c o n s i d e r ,
among
o t h e r f a c t s , how (10), (11), and (12) o c c u r a s s p e c i a l c a s e s of m o r e g e n e r a l

relations,,
P a r a l l e l i n g t h e u s u a l t r e a t m e n t of the Fibonacci sequence, w e obtain a
closed e x p r e s s i o n for P
since J P
(13)
Pn = B ^
I satisfies a difference equation.
Thus
+ B 2 x* + B 3 x^
w h e r e xj,, x 2 , x 3 a r e the t h r e e distinct r o o t s of the equation

x 3 - x 2 - x - 1 = 0,
and B j , B 2 , B 3 a r e constants depending on t h e s e r o o t s a s well a s P 0 ,
P2,
Pi,
and a r e d e t e r m i n e d by the s y s t e m
T h e values of x*, x 2 ,
Bt + B 2 + B 3 = P 0
BtX! + B 2 x 2 + B 3 x 3 = Pi
Bpdj + B 2 x| + B3x32 = P 2
x3,
B j , B 2 , B 3 a r e such a s to m a k e (12) too
cumber-
s o m e to be of any f u r t h e r p r a c t i c a l u s e in the succeeding development and hence

will not be w r i t t e n h e r e .
A much m o r e useful way of r e p r e s e n t i n g the r e c u r r e n c e r e l a t i o n f o r j P
may be found a s follows: In the notation of v e c t o r s and m a t r i c e s , we have by

(3),
" 1 1 1"]
("Ps"
p2
pr
p3
LP2.
10
P21
Pi
LPi.
.0 1 oj LPOJ
" l i i i
ps"
10
p2
_ 0 1 Oj L P I _
" i l l "
=
10
_ 0 1 0_
"p 2 ]
Pi
.Poj
1967]
213
and by finite induction

" 1 1 1" n-2
rp
n
(14)
"P2]
1 0 0
n-i
Pl
. oJ
0 1 0
F u r t h e r , a s i m p l e induction proof gives

"1 1
f n
10
(15)
K , _, L . . K
n+1 n+l n
1
n-l
- K n-i L n-i K n-J
p 1 q.
so it might b e said t h a t j K I and i L | a r i s e " n a t u r a l l y " in t h e i n v e s t i g a tion of
{Pj.
Using (14) and (15), we find for n,
"Kp+l
^
" n+p
(16)
"n+p-i
L n+p-2.
p positive i n t e g e r s that
^
p+i
P
p
.,,x
P - 1 II n -Hi
K
L
K
p-i p - i p - 2 J U n-2
from which we i m m e d i a t e l y s e e t h a t
P ^ = K ^JP + L , P
+ K P
n+p
p+l n
p+l n - l
p n-2
(17)
(18)
2n
(19)
2n-i
Now setting P 0 = 0,
= K
P + L
P
n+i n
n+l n - l
= K
P + (K + K
)P
+ K P o
n+l n
n
n-l n-l
n n-2
= K P + (K
+ K JP
+ K .P 0
n n
n-l
n-2' n-l
n - i n-2
P j = P 2 = 1,
of (17), (18), and (19), r e s p e c t i v e l y ,

Since
+ K P
n n-2
we have (10), (11), (12) as s p e c i a l c a s e s
214

K . .. L . ,J K .
p+r+1 p+r+1 p+r
n+p
P
K ^
p+r
n+p-i
L ^
p+r
p+r--i
L n+p-2 J
n-r
K _,_
p+r-i
p+r-i
"n-r-i
p+r-2J
n-r-2
w e a l s o have
(20)
P u .
n+p
for n,
p,
K ^ _ , P
+ L ^
P
p+r+i n - r
p+r+1 n - r - i
r positive integers,
S i m i l a r l y for n,
(21)
h,
r < n - 2,
k p o s i t i v e i n t e g e r s , we can show that

*h+k+i P n
n+h+k =
+ K ^ P
p+r n - r - 2
h +k+l n - i
n+k n - l
Using (20) and (21), we have the following useful e x p r e s s i o n :

-,n-2
xx u " p
2
h+k+i hi+k+i^+k 1 ri I r
"n+h+k
(22)
n+h
h+i
h+i
1 0
0 J L
Pi
i _
.po
It can be shown quite e a s i l y that t h e s e q u e n c e

(23)
P l f R2, P 2 , R3, P 3 , ' , P n , R n + 1 ,
i s g e n e r a t e d by t h e m a t r i x
ri
i
[i
0 1
0 OJ
that is,
rp
n
R
n.
(24)
L
" I 1 <r n-2 " P 2 l
ri-i-J
10
R2
10
. Pi J
[Oct.
1967]
215
It is an interesting and useful fact that this matrix is the transpose of the
generating matrix for JP }.
Using (24) in away analogous to that in which we established (21), we
prove that
(25)
P n+h+k
a.^, = K.h+k+l
M _^P + K, _,. R
n
h+k n
(26)
n+h+k
h+k+i P n
+ L
h+k R n
+ li+k-i
IC , Pn-lA
+
^ k ^ n - i
two relations which are not only interesting in themselves but which also give
n+h+k
(27)
Rn+h
h+k+i K h + k
h+k-i]
h+i
h-i
- n
hi order to define P
ri i o"
10
1 1
<U
"p2
R2
0 J [ i o o,
.Pi
for negative n, we use (14) for n > 0 written
in the form
rp
k+i
(28)
"0
0 0 1
Pi
_1 1 1_
J>i\
P
L n+2 J
1 0 " n "Pol
Replacing n by -n in (28), we have for n > 0,

rp
ro i o*
i
-n
(29)
-n+i
L -n+2-
ii
0 0 1
I I
which together with (14) determines
-1 -1
"Po"
Pi
.P2.
1- n f P o l
0 0
Pi
1 0
.PJ
for all n since P 0? Pj, P2
given. The same result is obtained upon replacing n by -n in (3) to get
(30)
P -n = P -n+3
. ^
J J 0 - P-n+2
J . - P J-n+1
n >0
are
216
R
[Oct.
i s a l s o defined for negative n by (29) and (30) s i n c e

R
= P
n
+ P
n-l
n-2
T h i s allows u s to r e m o v e the r e s t r i c t i o n placed on n,

3
p , r , h,
k above.
LINEAR SUMS
A l a r g e n u m b e r of what we shall call l i n e a r s u m r e l a t i o n s on t e r m s of

the sequences j R
| and j P
f w e r e found and proved.
Since an exhaustive
l i s t i s not our aim, only a few of the m o r e i n t e r e s t i n g ones a r e listed.
No
proofs will be given h e r e s i n c e the proofs m a y all b e m a d e r a t h e r e a s i l y by

finite induction.
(31)
n
1
T P . = -i (P
+ P + P0 - P2)
u
f-'
l
2
n+2
n
*
1=0
(32)
3n-i
E P3i = E ?i
i=i
<33>
(34)
Po ,
i=o
n
E ^3i = Psn " Po
i=i
n
E R 3i+l = Psn+l " P i
i=i
T h e s e r e l a t i o n s obviously have s p e c i a l c a s e s for the s e q u e n c e s

F o r example (33) b e c o m e s
n
(330
E Lsi = K3n .
i=i
and
1967]

48
217
QUADRATIC AND CUBIC RELATIONS
An attempt to p a r a l l e l the q u a d r a t i c r e l a t i o n s of the Fibonacci sequence

failed.
A different approach w a s n e c e s s a r y and this w a s found in t h e u s e of
t h e v e c t o r - m a t r i x r e p r e s e n t a t i o n of P .
We have t h e following i n t e r e s t i n g
quadratic form:
(35)
P^ + P ^
+ 2 P n - i P n _ 2 = P2P2n_2 + R 2 P 2 n - 3 + Pi P2n_4
T h e proof of (35) follows by c o n s i d e r i n g the left side of the r e l a t i o n a s

t h e s c a l a r product of the v e c t o r s f p , R , P ,1 and f p J, P
, P
1
^
L n
n
n-ij
L n
n - i ' n-2 J
(recall R = P
+P
), and then u s i n g (14) and (24), we have
n
n-i
n-2
p2 + p2
+
n
n-i
P
= fp , R , P
J
n - i n-2
I n
n
n-lj 'n-i
n-2- 1
ri i r 2n-4 "P2"j
TP2, R2, P J 1 0
L i .
Pi
rp2n-" 2 1
[p 2 , R2J Pt]
p 2n - -3
Lp2n--*J
PoJ
P2?2n-2 + Rj^Zn-S + PlP 2 n-4
For
P 0 = 0,
PA = P 2 = 1,
(35) b e c o m e s
K2 + K2
+ 2K
K
= K H
n
n-l
n - l n-2
2n-l
(35*
which is (12),
It w a s shown that (12) i s a l s o a s p e c i a l c a s e of (19), but (35) is
not obtainable from (19) n o r v i c e v e r s a .

One of t h e m o s t i n t e r e s t i n g r e l a t i o n s involving t e r m s of the Fibonacci
sequence i s t h e one
C
c
- C2 =
n - i n+i
n
(-l)n
218
T h e r e is a r e l a t i o n of t h i s n a t u r e for the s e q u e n c e j P
[Oct.
1; however a s m a y
h a v e been s u s p e c t e d , it h a s a cubic r a t h e r than a q u a d r a t i c form,, T h e d e s i r e d

r e l a t i o n is
(36)
P2 P
+ P3
P2 P ^ P
P
P
-2PP
P
P30 + 2Pf
n+i n - lH n-3
n n - l n-2
n-2 n+i
n n-3
n-l
+ P 3 + 2PPi + 2P 0 P 2 i + P 2 P 2 - 2 P ! P i - 2 P 0 P i P 2 - P 0 P^
Before proving (36), we note that for P 0 = 0, V1 = P 2 = 1, (36) b e c o m e s

(37)
K2K
+ K3
+ K2 K
- K
K
K
- 2 K K K
= 1
n n-3
n-i
n-2 n+i
n+l n - l n-3
n n - l n-2
T h e proof of (37) follows from (9) and (15) by t h e u s e of d e t e r m i n a n t s
since
K2K
+ K3
+ K2 K _, - K ^ K
K
- 2K K
K
n n-3
n-l
n-2 n+l
n+i n - l n-3
n n - l n-2
K
n+l
K
K
n-l
n-2
n-l
n-3
n
n-l
n-2
^ L a., K
n+l n+l n
K
L
K H =
n
n-l
K
L
K
n - l n - l n-2
1 1 1
10
10
Proof of (36): Even though (36) m a y b e verified in v e r y m u c h t h e s a m e

m a n n e r a s (37), we adopt a different method of proof s i n c e this is m o r e e a s i l y
u s e d in a g e n e r a l i z e d v e r s i o n of (36).
F i r s t , w e s t a t e the following l e m m a
whose proof t h e r e a d e r can r e a d i l y supply,

L e m m a : Let A be any 3 x 3 m a t r i x and l e t x and y b e t h r e e - d i m e n s i o n a l
v e c t o r s ; then the c r o s s product
(Ax) X (Ay) i s equal to the cofactor m a t r i x
of A multiplied by x X y; i e. ,
(Ax) X (Ay) -
(cofactor A) (x X y).
Now the left s i d e of (36) can b e c o n s i d e r e d as the t r i p l e s c a l a r p r o d u c t of the

three vectors | P , . P , P
1, | P
. P
, P
1, and f p n , Pn 1 , Pn 2 1.
L n+l
n
n - i j [_ n - l
n-2
n-3j
|_
"
" J
By (14) and the l e m m a ,
1967]
rp
1 0
-3J
P
n-2-
" i l l " n-3
"P2"
. 0 1 0 .
10
Pi
n-i
n-2
L n
*1 1
" P
n
n-i
f n-3
.0
-Po-
219
"P31
P2
-PJ
10.
0 0 1
v\
1 0 - 1
P3P0 - P1P 2
Pi
1-1
- P2P0
- P3P1-
Therefore
p2p
n n-3
"n-i
+ P:
n-2 n+l
rp
n+i
P , P
P
- 2P P
P
n+i n - l n-30
n n - i n-2n
-1
rp
"Pn-i"
X
P
n
L P n-i-
n
P
n-i
n-2
. P n-J
P
L n-3 J
ri 1 o" n - 3 . 0 0
= [ p 4 , ps> p 2 ]1
1"
n-31-
P? - P2P0
0 1
0 0 -1
p s3-^0
p,
L1 .
0 1 -!_
p i - P3P1
= P4(P2i - P 2 P 0 ) + P3(P 3 Po - P2P1)
PlP2
J
P 2 ( P ! - P3P1) .
which r e d u c e s to the r i g h t side of (36)

I>1 = 1;
E x a m p l e : Suppose we let P 0 = 0,
then t h e r i g h t side of (36)
becomes
Po - 2Po + 2 .
Setting t h i s e x p r e s s i o n equal to z e r o and solving for

exist a l g e b r a i c i n t e g e r s , say P 0 3
p2p
+ p3
+ p2 p
n n-3
n-l
n-2 n+l
Pl3
P2s
P
P2,
we see that t h e r e
such that for t h e s e q u e n c e J P
P
P
+ 2P P
P
n+l n - i n-3
n n - i n-2
\,
220
[Oct.
The l e m m a and (22) m a y b e u s e d a s in the p r e v i o u s method of proof to

show that for h,
k,
n,
m,
integers
(38) P
P
P
+ P P
P
+ P P
P
v ;
n+h n+m n+h+k+t
n n+h+t n+h+k+m
n+t n+h+k n+h+m
-PP
P
-P
P
P
- P P P
n n+h+m n+h+k+t
n+m n+h+k n+h+t
n+h n+t n+h+k+m
= (K h L
h+k+i
h+kLh+i)[Pt+2^PlPm '
opm+i)
Pt+i(Po p m+2
+
t(
2pm+i -
T h e r e a r e many i n t e r e s t i n g s p e c i a l c a s e s of this relation,,

a few.
(39)
v
'
If P 0 = 0,
Pt = P 2 = 1,
2Pm)
p p
i m+2)]
We mention
(38) b e c o m e s
K , , K . K . , , , ,, + K K ^ J K ÛjLl ^ + K ^ K ^ ^ K ^, ^
n+h n+m n+h+k+t
n n+h+t n+h+k+m
n+t n+h+k n+h+m
n n+h+m n+h+k+t
n+m n+h+k n+h+t
n+h n+t n+h+k+m

v
If k = h = t,
(40)
v
m = 1,
m-r
(39) b e c o m e s
K , K _,_, K . , + K K __ K _ , + K ^ . K . ^ K _ - K K ^ . K . .
n+i n+h n+3h
n n+2h n+2h+i
n+h n+h+i n+2h
n n+h+i n+3h
- K
K2
_ K 2 , K _,_,_, = K. K , K ,
- K2 K , ;
n+i n+2h
n+h n+2h+i
n - i h 2h-l
h - i 2h
and if t = h,
(41)
v
'
h+k h - l / v t - i m
h h+k~i
k = m - h,
(39) r e d u c e s to
K K _,_ , K ^
+ 2K ^ , K ^ K ^, ^ - K K2 ^ - K2
K _
n n+2h n+2m
n+h n+m n+h+m
n n+h+m
n+m n+2h
- K2 K ^
n+h n+2m
-(K.K
-K K
J2
\ h m-i
m n-i'
In o r d e r that the above r e s u l t s b e valid, we m u s t choose h and k

that
nKn+k-i "
h+kKh-i "
hLh+k+i "
h+kLh+i *
'
so
1967]
221
for in the proof of (38), we a s s u m e that the m a t r i x
h+k+i Lh+k+i K h+k
h+i
h+i
h
0
i s n o n - s i n g u l a r
Using (27) we can find r e l a t i o n s h i p s involving t e r m s of both the s e q u e n c e s
| R J and | P j which r e d u c e to an e x p r e s s i o n independent of n0 F o r e x a m p l e , i t may be p r o v e d that
n+h+k+t^ n-tfi n+m
n n+h+nr
n+h+t^ n n+h+k+m
n+m n+h+k'
]
4- p
CP
R
- R
P
~^h+k-i^'VkV
n+t v n+h+k n+h+m
n+h n+h+k+m
+
+
[ p ^ P ^ m - P0Pm+i) P t + 1 ( P 0 P m ^ - P 2 P m )
Pt(P2Pm+i " pipm+2)]
i'
It should be noted that no t e r m s of the sequence j R I a p p e a r on the r i g h t side

of (42) and a l s o that the second factor on the r i g h t side of the equality sign i n
(42) i s the s a m e a s the second factor on the r i g h t side of (38).
5.
MISCELLANEOUS RESULTS
We conclude with some m i s c e l l a n e o u s r e s u l t s .
The following limiting
r e l a t i o n s may be e s t a b l i s h e d u s i n g (13) and the fact that rl9 r 2 , the two c o m p l e x r o o t s of

x3 - x2 - x - 1 = 0 ,
a r e such that
r,
(43)
(44)
'4-
= ro < 1
V3V33
- 19 + V 19 + 3V33
limit - =
i> oo
n
limit
n >oo
"n+h
1 - V 3 yg3" - 19 + V 19 + 3 Vg3
3
)
222
Oct. 1967
By induction the following t h e o r e m m a y be e s t a b l i s h e d :

T h e o r e m : F o r e v e r y positive n,
K 4 n = KAn_t
K4n-2 =
= 0 (mod 2)
4n-3 = 1 ( m o d
If we l e t D(P 0 , P j , P 2 , , P n )
Po P i
P2
PiI P^ 2.
P.,
n+i
(mod 4)
K4n = 0
be the d e t e r m i n a n t
n+i
n+2
i t can be shown that for n > 3,
D(P 0 , P l f P 2 , - , P n )
= 0
T h i s m a t e r i a l i s taken from Some G e n e r a l i z a t i o n s and E x t e n s i o n s of the

Fibonacci Sequence, a t h e s i s submitted to the U n i v e r s i t y of P i t t s b u r g h by the
f i r s t author in p a r t i a l fulfillment of r e q u i r e m e n t s for the Ph D. d e g r e e .
REFERENCE
1.
M. Agronomoff,
p. 126.
Une s e r i e r e c u r r e n t e , " M a t h e s i s , s e r . 4, vol. 4 (1914),

RESTRICTED COMPOSITIONS
S. G . M O H A N T Y
McMaster University, Hamilton, Ontario, Canada
As a continuation of [6] and [ 7 ] , this paper deals with a restricted set

of compositions of an integer (to be defined below) and presents extensions of
some results of Gould [ 2 ] , [ 3 ] , [ 4 ] , by interpreting the compositions through
the corresponding lattice paths.
By the definition in [ ? ] , a (k + 1)-composition
(tl9t2, , t^.+1) of an
integer n (i e. ,
k+i
E t.
= n
and
t. > 1
1
i=i
for every i) dominates another (k + 1)-composition (tj,tj, ,t^ + 1 ) of n if

and only if
3
S
i=i
\ - Z) *!
for
j =
1.2,---,k + l .
i=i
Using the 1:1 correspondence in [ 6 ] , we associate with each (k + ^ - c o m p o s i tion of n a minimal lattice path (onward and upward path through lattice points)
from (0,0) to (n - k - l,k) such that the directed distance measured along the
positive direction of x-axis, of the point (n - k - l , k - j), j = 1,2, , k
from the path is
3
i=i
Without any ambiguity, denote this path by

223
224
'
[Oct.
^'JL\-
i=i
Thus, it is evident that to the set C(n;sil9 a2, , a^) of (k + 1)-compositions

of n, dominated by the (k + 1-composition
(al3 a2, , aj^+j) of n c o r r e s -
ponds the set L(A1? A 2 J ' , A^) of lattice paths which do not cross to the left
or above the path
I Al9 A 2 ,*' , A^ I
aj - 1, at + a2 - 2, , ^2
a.
i=i
Let the number in the set C (equivalently in L) be represented by N(n;al9 a2,

,ajj) for k > 1, and by N(n) for k = 0.
Trivially,
N(n)
(1)
N(h;a,jL,_Vîl) =
(2)
k- 1
/a + k -
l\
and
N(n;a 1 ,a 2 ,-"- ,&k) - 0
(3)
if any a. is either zero or negative.

Now consider the path
[AJ,AJ ' " "

2
'
such that A\ < A. for all i0 Every path in L passes through one of the points
(n - k - A! + - 2, k - i),
(n - k - A!
i = 0 , 1 , 2, ,k,
(A'
= AM before moving
to
- 1, k - i) and then reaches (n - k - l,k) not crossing [AJ, A2 ,
. . . , A y . Therefore,
1967]
(4)
2 25
N ( n ; a 1 , a 2 , - - - , a k ) = N(n;aj - aj 5 a 2s s a k )N(n)
+ N(n;ai + a2 - &[ - aj, a3, , a k )N(n;aJ)
+ N(n;ai + a2 + a3 - a J - aj - a^, a4, , ak)N(n;ai, aj)
/ k
' + N b ; E
\ iFi
k \
i " E allNCnsai, aj, , a^
i=i
+ N(n)N(n;aJ,a5,...,aj 5 .)
We note that whenever A! = A.,
I
r
N(n;ai + . . . + aj - a\ - - aj, a| +1 , , a k ) = 0
It may be pointed out that relation (4) in some sense is a generalization of

VandermondeT s c onvolution
!()(-H x ; y )'
a further discussion of which is given later.
By setting ^ = A! + 1 and a2 = a 3 = = a k = 1 in (4) and using
(2) , we get the recursive formula
(5)
N(n;ai,aJ,-- ,a^) =
/AJ + k \
y \
k - i / A J - A! + k - i \
k
X
k
i +1
J- E (
JN(Q;ai,aJ,...,al. 1 )
which is the same as (9) in [ l ] and (2) in [ 8 ] , The solution of (5) is stated in
the following theorem.
226
[Oct.
Theorem 1:
(6)
NCriâ^.-âk)
/Ak-A1+k-l^
A,
+k - 2
k-2
k - 2
/Ak_2-A1+k-3\
>
A k
/Ak_3-A1+k-4X
AT + k - 3
k-3k - 3
)
( A, 7)
(A
, -
A l )
1
0
0
Another way of expressing the number in L leads to
Ai
(7)
N(n;a1,a2,-..,ak) =
A2
Ak
E
E ...
'
1
Xi=o x2=Xi
xk=xk-i
Ao
A,
a-i
=E E -
Xi=o x 2 = Xl
Ai A2
xk=xk_A
Ak
+ E
'00
Xj=a X 2 = X l
i
E
=x
1,0 < a < Aj + 1 .

k-l
Substituting x. - a = x! for i = 1, 2, " ,k, the second term on the right hand
side becomes
Aj-a A 2 -a
(8)
E
E
x}=0 xj==xj[
A, -a
x
2
E
=X
k k-i
N(n;at-a^g,---,^)
On the other hand, the first term can be written as
1967
-l
A2
Ak
(9) E E ...
A2
A3
Ak
i = E E E E 1
E
X
Xi=0 X 2 = X l
a-1
227
k=Xk-l
Xl=
2=
a-l
3=Xl
Xj-1
k=Xk-l
A3
Ak
- E E E
Xf=l X2=0 X 3 = X l
E 1.
X ^ k . j
whereas the last term in (9) can again be expressed as

-l
Xj-i
A3
A4
E ...
- E E E
x
l~1
2=0
3=0
Ajj
4=x3
0-1
Xj-i x 2 - i
A4
E '1+ E E E
x
k=xk-l
X =2
2 = 1 xl=0
A^
E
x
4=x3
E i.
x
k=xk-l
When we proceed in the above manner, the final expression for (9) is
E ("Di+1 h)
(10)
N(n;A.+i + 1, a.+2, , ^ )
by noting that
x p i - i x2=i-2
E i = E
Xj=o
E E i = m
x-pO x2=o
xi=o
\ /
and
i+i
i+2
E
E
E
1 = N(n;A.
X
=X
V f i+2 i-M
V*k-i
Thus it follows from (7), (8) and (10) that
+ 1, a.+2, , a k ).
[Oct.
228
k
. / v
g (-1) 1 / " J N ( n ; A i + i + 1, a. + 2 , ' " , \ )
(ID
= N ( n ; a i - a, a 2 , - - . , a k ) .
An a l t e r n a t i v e way of simplifying the f i r s t t e r m on the r i g h t of (9) i s

a-i
A2
Ak
-i
i-
x1=o x 2 =: x 1
xk^ 0
x =0
xk-i"1
A2
2=xi
w h e r e the s u m s in the l a s t t e r m for which x,
1 .
k=0
- 1 i s negative a r e z e r o .
Repetition of this p r o c e s s yields
(12)
for
k
/
(-D1+1 (Ak+i-i
i=i
\
i
c = a1#
1>
J N ( n ; a 1 , a 2 , - . - , a k _ i ) = N(n;a 1? a 2 , , a k )
/
Relation (12) h a s been obtained e a r l i e r in [ 7 ] , which i s e q u i v a -
lent to (3) in [ l ] .
When c = SL1$ the solution of e i t h e r (11) o r (12) i s s t a t e d a s T h e o r e m 2,
for which a d i r e c t e l e m e n t a r y proof i s p r o v i d e d below.
T h e o r e m 2:
(13)
N(n;a1,a2,---,ak)
-r) (v )
* (v)
Proofs
Obviously
( \
+ l
fr.v)
1
ft:. )
1967]
229
(:)
Using this in (7), we see that

Ai
A2
xf=o x2=xt
Ai
A2
Ak-i
=x
Xk-l k-2
V i
+*
+1
(\ )
Ak-i
E Z
Xj=0 X2=X!
(VHY'XH-C)
Or) > (:;)
ft-)-0
k-2
(kX-12)|
(?)
230
[Oct.
T h e proof i s complete when the s u m m a t i o n i s continued to the end,

T h e o r e m s 1 and 2 give r i s e to an i n t e r e s t i n g combinational identity on
d e t e r m i n a n t s , the d i r e c t proof of which i s not obvious.
We check e i t h e r f r o m the t h e o r e m s o r d i r e c t l y that
(14)
N ( n ; a 1 , a 2 , * - , a k ) + N ( n ; a ! + a 2 , a 3 , - - , a k ) = N(n;ai + l , a 2 , , ak)
(15)
N(n;l,a2,--",ak)
= N(n;a 2 , a 3 , , a k )
and
(16)
N ( ] a. + j ; a ^ a ^ - ' ^ k
= NI
aj + 1; a l 9 a 2 , - - , ak J
j = 1,2,-
A few i m p o r t a n t special c a s e s a r e c o n s i d e r e d below.

C o r o l l a r y 1.
-u
x.v
a
N(n;a,b^b) = ^
k- 1
KT,
(17)
/ a +
(
k
kb\
)
This i s d i r e c t l y verifiable f r o m e i t h e r one of the t h e o r e m s 0
(Also s e e T h e o -
r e m 1 in [6])
In the next, we evaluate
N
(a,b;c,d)
= N(n;a,b,- % b , c , d ^
p - 1
\*^d)
q - 1
which h a s been obtained by a different method a s T h e o r e m 3 in [ 6 ] ,

C o r o l l a r y 2.
(18,
Np>q<a,b;c,d, - < V ^^^^^

x
('%*/,'.-, ""J
(q - i + l ) b - c - (q - i)d / ( q - i + l ) b - c - q d + i \
(q - i + l ) b - c - qd + i \
i
/
1967]
Proof: F o r c + q(d - 1) ^ qb,
(a - 1) + (i - l)(b - 1),
i = 1, 2, '
When c + q(d - 1) < qb,

c + s(d - 1) ^ sbo
the r e s u l t i s i m m e d i a t e , by taking A! =
,p + q
in
(a,b;c,d)
and
N(n;a,b, ,b),
^Tn^T
p
help of (4), w h e r e
(a - 1) + (i - l)(b - 1)
i
(4) and applying C o r o l l a r y 1.
let s(p i. s 4= p + q) b e t h e l a r g e s t i n t e g e r s o that
PJQ
231
i =
e x p r e s s e d with t h e
l,2,',s,
) (a - 1) + (p - l)(b - 1) + (c - 1) + (i - p - l)(d - 1) i = s + 1,
s + 2 , 0 8 * , p + q,
lead to (18), after s o m e simplification.

F o r c o m p l e t e n e s s , we p r e s e n t two m o r e special c a s e s which a r e known
and can e a s i l y b e derived.
C o r o l l a r y 3;
(19)
N
(a,b;c,l)=(a
P>q
\
2 + (
y^
1 ) ( b
P ;
"" i= 4+i
1 ) +
<i
a + (p + q
""
i)b
P ^ )
/
/a+(p+q-i)bVc +q - ( q - i + l ) b - l
V P + c *-i A
C o r o l l a r y 4;
In his p a p e r [ 2 ] , Gould h a s defined
A <B y) -=p
\v,y)
and h a s shown t h a t A, 0,y)
(21)
p
+ yk
1 +
yI * k
yk
j)
s a t i s f i e s the r e l a t i o n
k
A.^,r)Ak_.(8,y)
i=o
= Ak(/3 + 8 , 7 )
232
Suppose that /3, y and 8 are non-negative integers 0

follows from (4) and (17) by putting aA = /3 + 8,
= 0,
and aj =
be verified.
af3
[Oct
Then (21) immediately
a.t = a3 = = a, = y,
a\
= = a* = y in (4). Relation (11) in [2] can similarly
Also, the convolution (5.5) in [3] for t = 0 can be compared with
(11) and their equivalence is easily established.

In what follows, the results on restricted compositions are analogous to
those on unrestricted compositions in Gould's paêr [4j (Theorems 1 and 5 or
equivalentry Theorem 6) Fix a2, a3, , a^ and let
i=2
From (14), (15) and (16) we infer that
(22)
N(m + a4 + 1; a1? a2, , a^) =
< m + i; a2 - 1 + i, a3, , a k )
i=i
j3 L m + i J
N {m + i; a 2 - l + i, a 3 ,- , akf
where fz] is the greatest integer less than or equal to z and NJm + i;a2 - 1
+ i, a3, , a^| is the number of compositions in the set S(m + i;a2 - 1 + i, a3,
, a k ) which is defined as follows: For i negative or equal to -zero,
S(m + i; a2 - 1 + i, a3, , a k ) is empty;
S(m + 1; a 2 j a 3 > . ,ak) = C(m + 1; a2, a3, , a k ) ;
For i > 2, S(m + i;a2 - 1 + i, a3, , a k ) is the subset of C(m + i;a2 - 1
+ i, a3? , ak) with the property that if (x1? x 2 ) " , , x ] { ) E S(m + u;a 2 - 1 + u, a3,
, ak), u = 1, 2, , i - 1, then for r a positive integer (rx1? rx 2 , , rx,)
( S(m + i;a2 - 1 + i, a3, , ak). Expression (22) corresponds to Theorem 1
in [4].
1967]
00
00
j=1
j=i
[mr4]
^m
233
i; a2 - 1 + i> as, , a k }
oo
i|iN{m
I;a1-l
J
i,^...,ak}|[S+J.] x m+i
EN{m
i;a2-l
1=1
by (3) in [ 4 j .
(23)
m+i
i,a3,-.-,ak[
X m +
(l-x)(l-X
Therefore,
m+i
E N { m + i; a 2 - 1 + i, a 3? , a k }
m
x + i
i=i
(1 - x
)
= X) N(m + i + l;i, a 2 /-- J ak)x m + 1 (l-x)
1=1
00
= X)-N(m + i ; a 2 r - l + i, a3, , a k ) x m
i=l
by (14), (15) and (16). But (23) can be written as

00
(24)
i
N { i ; a 2 - m - l + i,a5,---,ak}-xr
i=m+l
1- x
oo
H N d j a a - m - l + ijagj-'-jakJx1
i=m+i
In order to extend the summation to i = 1, 2, , m in (24), define

M-*/.
-,,
x _ / 0 for
i = 1,2, , m
N (i;a2-m-l + i,a3,---,ak) ~ } N ( i ; a 2 _ m _
Thus, following the procedure in [ 4 ] ,
(25)
1 + if
g^...
>ak)
for i = m+l, m+2-
N|n;n + a 2 - m - l , a 3 , 8 9 % a k } - ^ N^(n;n + a 2 - m - l , a 3 , ' - % a k ) ^ | ~ I ,

iln
\ /
234
Oct. 1967
which i s s i m i l a r to that of T h e o r e m 5 in [ 4 ] .
We finally r e m a r k that such r e s u l t s can a l s o b e obtained for t h e n u m b e r
of l a t t i c e p a t h s in the s e t L.(A l 5 A 2 , ,A, ) defined a s follows:
L0(A1,A2,-",Ak) = L(Ai,A2,'-',Ak);
L.(A 1? A 2 , , A k ) i s t h e subset
of
L(AA + i, A 2 + i, , A, + i) such that if [ x l 9 x 2 , , x ^ l E L (A l5 A 2 , , A k ) ,

u = 0 , 1 , . . , i - 1,
then [ r x l 5 rx 2 , , r x k J ^ L ^ A j , A 2 , , A^ ) .
REFERENCES
1.
F . Gobel,
S o m e R e m a r k s on Ballot P r o b l e m s , " R e p o r t S321a,
Math
C e n t r u m , A m s t e r d a m , 1964.
20
H. W. Gould,
of Vandermonde T s Convolutions,"
"Some G e n e r a l i z a t i o n s
A m e r . Math. Monthly, Vol. 63, 1956, pp. 8 4 - 9 1 .

3.
H. W. Gould, "A New Convolution F o r m u l a and Some N e w

Relations for I n v e r s i o n of S e r i e s ,
4.
Orthogonal
Duke Math. J o u r n a l , Vol, 29, 1962,
H. W. Gould, "Binomial Coefficients,
T h e B r a c k e t . Function, and C o m -
positions with Relative P r i m e Summands, " Fibonacci Q u a r t e r l y , Vol. 2,

1964, pp. 241-260.
5.
S. G. Mohanty, "Some P r o p e r t i e s of Compositions and T h e i r Application

to the Ballot P r o b l e m , " Can. Math. Bull. Vol. 8, 1965, pp. 359-372.
6.
S. G. Mohanty and T. V. N a r a y a n a ,
a n d T h e i r Application t o
"Some P r o p e r t i e s of Compositions
P r o b a b i l i t y and Statistics I , "
Biometrische
Zeitschrift, Vol. 3, 1961, pp. 252-258.

7.
T. V. N a r a y a n a and G. E. Fulton,
"A Note on the Compositions of an
Integer, " Can. Math. Bull. , Vol. 1, 1958, pp. 169-173.

8.
P . Switzer, "Significance P r o b a b i l i t y Bounds for Rank O r d e r i n g s , " Ann.

Math. Statist. , Vol. 35, 1964, pp. 891-894.

ENUMERATION OF CERTAIN TRIANGULAR ARRAYS

D, P. ROSELLE
University of Maryland, College Park, Maryland
1.
INTRODUCTION
Let k be a p o s i t i v e integer*
We define t h e n u m b e r s F (k) and N (k)
by m e a n s of the r e c u r r e n c e s
F (k)
n
(1.1)
- F
(k) + F . (k)
n-l
n-k
(n > k) ,
'
k
(1.2)
. (k) =
n-k
^ ( ^ N n i<W
J
i=o
(n
V /
^k) >
with the initial conditions

(1.3)
F (k) = n + 1
(1.4)
N n (k) = (
Note that
(0 < n < k) ,
(1.5)
Fn(l)
(1.6)
( 0 < n < k ) ,
'
= Nn(l) = 2 n ,
F n (2) = F n + 2
N n (2) = 3 . 2 n _ 1 ,
(1.7)
w h e r e F . denotes the u s u a l Fibonacci n u m b e r

Given p o s i t i v e i n t e g e r s
and
k,
(F 0 = 0,
F j = 1) .
put m - pk + r
let T(k, m) denote the n u m b e r of a r r a y s

(1. 8 )
n u n!kn l 9 k+i 9 e nu p\^nu

n
pk+i
nlm
2, k+i ' ' n2> pk n 2, pk+i " '
p,pk p,pk+i
n
w h e r e n1A is e i t h e r
0 or
p+l ? pk+l " '
1 and
235
2m
pm
p+i, m
(1 < r < k)
and
236
(1.9)
n.. > n. .,, > 0;

i]
i,]+i
[Oct.
n.. > n._,_, . > 0 ,
'
13 ~
1+1,3 -
For example, T(2,5) and T(2,6) are the number of arrays

x x x x x
x x x x x x
x x x
x x x x
X
X X ,
where each x is either 0 or 1 subject only to the conditions (1.9). As a

further example, we have T{2,3) = 5, the arrays being
1 1 1
1 1 1
1 ,
1 1 0
0 ,
1 0 0
0 ,
0 0 0
0 ,
0 .
Indeed, we find that

(1.10)
T(k,n) = F m (k)
(m,k = 1,2,3, ) .
The numbers N (k) also occur in connection with triangular arrays of

zeros and ones. We prove that
(LID
ao = - | V j
j=o
Dk - I ] ( P 3
Dn ,
N n (k)x n = k ^ - 1
nô
x - (1 - x)
(1.12)
where p denotes a primitive k

root of unity.
Finally, we have included some one-line arrays which can be enumerated
in terms of the numbers F (k) and N (k).
n
n
The author wishes to thank Professor Carlitz for his aid in the preparation of this paper.
1967]

2.
237
THE NUMBERS F (k)

n
F o r given p o s i t i v e i n t e g e r s m and k,
let T(k,m)
denote
the n u m b e r
of a r r a y s (1.8) subject to t h e conditions (1.9),

To evaluate T(k,m),
'''
i, m-i
matrix.
anc
we f i r s t note that if n m i = 1 in (1,8), then n ^
* t h e r e a r e T(k,m - k) a r r a n g e m e n t s
On t h e o t h e r hand, if n l m = 0,
t h e r e a r e T(k, m - 1) a r r a y s p o s s i b l e .
(2.1)
of the
resulting
then n 2 m = = n p + 1 J i n = 0
and
T h i s evidently yields
T ( k , m ) = T ( k , m - 1) + T ( k , m - k)
(m > k)
In the next p l a c e , it follows at once from (1.8) and (1.9) that

(2.2)
T(k,m) = m + 1
(1 < m < k) .
T h i s evidently c o m p l e t e s the proof of

T h e o r e m 1.
The n u m b e r of a r r a y s (1.8) subject to the conditions (1.9) is
given by
(2.3)
T(k,m) = F m ( k )
(m,k = 1 , 2 , 3 , - - - )
As an i m m e d i a t e c o r o l l a r y of (2.3) we have
T h e o r e m 2.
Let q, (n;p) denote the n u m b e r of p a r t i t i o n s of n into at
m o s t p p a r t s , s u c c e s s i v e p a r t s differing by at l e a s t k.
M
q k (n;p + 1) = F
n=o
(2.4)
w h e r e m = kp + r
(1 < r < k) and M = m(p +
Indeed, using the g e n e r a t i n g function f"2l
lm(m+i)
2-f
n=o
qi(n;m)x
Then
(k) ,
l ) - k
tA
238
w e e a s i l y verify that, for k = 1,

l j h a s noted that, for
k > 1,
(2.4) r e d u c e s to (1.5).
However,
the g e n e r a t i n g function for
[Oct
Chaundy
q, (n;p)
i s not
known.
30
THE NUMBERS N (k)

n
Given positive i n t e g e r s
for
1 4 j 4 k,
(3.1)
and
k,
put
m = pk + r ( l 4 r 4 k) and,
let N.(m,k) denote the n u m b e r of a r r a y s
n t l n l j k + 1 n ljP k-M n l m
n
j i """
n
n
j,k+i n j,pk+i " *'

j+k,k+i ' ' "
the conditions (1.9).
jm
j+k,pk+i ' '
w h e r e n ^ is e i t h e r 0 o r
j+k,m
j+pk,pk+l n j+pk,m
1 and the n y a r e non-negative i n t e g e r s subject to
F o r example,
Ni(5,2) and N 2 (5,2) a r e t h e n u m b e r of
arrays
X X
X X X
X X X X X
X X X
X X X X X
X X X
X X X
X
X
X X X
,
X ,
respectively,,
It follows from (3.1) and (1.9) that
(
. 1|
("V
)
J
(3.2)
N.(m,k) =
(1 4 m 4 k) ,
(3.3)
N.(m,k) = N.(m - l,k) + N.__ (m,k)
(3.4)
Ni(m,k) = Ni(m - l,k) + N R (m - k,k)
(2 4 j 4 k; m > k) ,
(m > k) .
T h e proof of (3.2) is not difficult; (3.3) and (3.4) a r e proved in exactly the s a m e
way a s (2.1).
1967]

Using (3.3) with j = k,
239
we s e e that
N k _ 1 (m,k) = N k (m,k) - N k ( m - l,k)
and, in g e n e r a l ,
(3.5)
N k _.(m,k) =
( - l ) r 0)
N k ( m - r,k)
(1 < j < k - 1)
C o m p a r i n g (3.4) and (3.5), we obtain the r e c u r r e n c e
(3.6)
N k ( m - k,k) =
(-D
r=o
r=o
r /
/
k(m " r'k)
'
which should b e c o m p a r e d with (1.2).

F o r k = 1,2
the r e c u r r e n c e (3.6) is easily handled.
from (3.2) that (3.6) is in a g r e e m e n t with (1.5) and (1.7).
Indeed, it follows
Note that (1.7) and
(3.5) imply
Ni(m,2) = 3 2 m " 2
(3.7)
(m > 2) .
T o solve the r e c u r r e n c e (3.6) for g e n e r a l k,
we make u s e of s o m e r e -
s u l t s from the calculus of finite differences [3J.
Let p denote a p r i m i t i v e
root of unity and note that t h e c h a r a c t e r i s t i c polynomial of the r e c u r r e n c e
is
(x - l ) k - 1 ,
whose r o o t s a r e p
- 1 (j = 0 , 1 , 2 , , k - 1).
T h u s t h e r e a r e constants A 0 ,
Aj, , A k - 1 such that

(3.8)
k-i
N k (n,k) = Z A V
- Dn .
240
[Oct.
We show that
*j = E [(p"5 + 1)k " x ] (0- j - k " 1] '
(3.9)
first noting that we may extend the recurrence (3.6) and define N, (0,k) = 1.
To prove (3.9), we have, for 0 4 r 4 k - 1,
E r<P-i
j=o
D k - i l (pj + D r = E ( J ) E ( t r ) E p j(t " s)
s=o V / t=o \ z I j=o
-'(00M k+ * r )which, using (3.2), implies (3.9).

It follows from (3.8) and (3.9) that
(3.10)
Nk(n,k) = \
[ ( p " j + l ) k - l ] (p j - l ) n ,
so that
(3.1D
Nk(n)k)Ê(^
s=o \
(j).
/ r=s(mod k) \
If we define generating functions
(3.12)
F , ( x ) = f; N(n,k)x n
J
JJ, = 0
(1 < j < k)
then it is clear from (3.2) and (3.3) that

(3.13)
(1 - x J ^ F y f e ) = F kl (x)
0 = 2f3,---,k)
1967]
'
241
M o r e o v e r , u s i n g (3.4), we have
= x - k ( l - x)Fki(x) - * * * _ - /
Fkk(x)
C o m p a r i s o n with (3.13) then yields
(3.14)
F. (x) -
(3.15)
F. .(x) =
(x
(X
kJ
- 1)(1 - x)
x - (1 - x)
" , 1 ) ( 1 " X) ,
x - (1 - x ) k
(1 L j 4 k) .
We s u m m a r i z e t h e r e s u l t s of t h i s section by stating
T h e o r e m 3.
Let N.(n,k) denote the n u m b e r of a r r a y s (3.1) subject to
t h e conditions (1Q9).
Then N.(n,k) s a t i s f i e s (3.6), (3.10), and h a s g e n e r a t i n g
function (3.15).
48
SOME ONE-LINE ARRAYS
Let S, (nj) denote the n u m b e r of o n e - l i n e a r r a y s

n 1 n 2 n 3 n 4 " 0
(4.1)
w h e r e the n. a r e non-negative i n t e g e r s , subject to the conditions

(4.2)
n. A n . + 1 + k
(J = 1, 2, 3, )
It is c l e a r from (4.1) and (4.2) that

S k (n)
= 1
n-k
r=o
(n < k) ,
242
[Oct.
which i m p l i e s
S k (n) = S k (n - 1) + S k (n - k)
(n > k)
Thus an easy induction e s t a b l i s h e s

T h e o r e m 4
T h e n u m b e r of a r r a y s (4.1) subject to the conditions (4.2)
is given by
(4.3)
S k (n) = 1
(1 4 n 4 k) ,
(4.4)
S k (n) = F n _ k ( k )
(n > k)
In p a r t i c u l a r note that (4.3) and (4.4) yield

(4.5)
S2(n) = F n
(n = 1 , 2 , 3 ,
Returning to the n u m b e r s F (k), we s e e from (1.1) and (1.3) that
F
(k)
nk+j
(4.6)
1 =
In the next p l a c e , for

a r r a y s (4.1), w h e r e the n
(4.7)
r=o Wi*>
1 < j < k,
S, .(n*)
-k)
denote the n u m b e r of
a r e non-negative i n t e g e r s subject to the conditions
nr ^ nr+l
n
let
(14 j
>
r+1
(r * j (mod k) ) ,
<r = 3 (mod k) ) .
It is i m m e d i a t e from (3.7) that

(4.8)
Sk.(l) = j
( 1 4 j 4 k) ,
(4.9)
kjj+1<
) = 1
kjW
(1 4 j 4 k - 1) ,
1967]
ENUMERATION OF CERTAIN TRMNGULAR ARRAYS
2 43
n-i
(4.10)
S kl (n) = 1 + E Sj.fr) .
r=i
We shall show that

<4-n)
S kj fr + 1) = F r t f j - 1 ( k )
(1 < j < k) .
The proof of (4.11) is by induction, the case r = 0 being in agreement with

(4.8).
Assuming (4.11) for r < n - 1, we see from (4.10) that
(n-0k(k)
+ F
nk-i
which implies
(4.D
S ki (n
1) = F n k (k)
Using (4.6), (4.9), and (4.12), we obtain successively
k ) j + 1 < n + 1) = 1 + E
^ ^ ( k )
F n k + j (k)
which proves
Theorem 5. The number of arrays (4.1) subject to the conditions (4.7)
is given by (4.11).
Finally, we can use the numbers N.(n,k) to enumerate certain one-line
arrays.
(4,13)
where
(4-14)
For 1 < j < k, let R, .(n) denote the number of arrays

n ni n2 n3
nr > nr+1
(r j (mod k) ) ,
n r > k + n r + 1 (r = j (mod k) )
244
[Oct

It follows that
(4.15)
R. .(]
kj
-(v)
(4.16)
(4.17)
k j
(0 4 n L k)
% Rk|j.i<s)
s=o
n-k
(n) = ^ R v(s)
kk
s=o
(2 ^ ] 4 k) ,
(n > k)
and we deduce
T h e o r e m 6.
The n u m b e r of a r r a y s (4.13) subject to the conditions (4.14)
i s given by
(4.18)
R k j (n)
= N.(n,k)
(1 < j < k) .
F o r convenience of r e f e r e n c e , we give t h e following t a b l e s of F
, (k)
n+k v
and N.(n,k).
16
32
64
128
256'
13
21
34
55
89
13
19
28
41
60
10
14
19
26
36
50
11
15
20
26
34
45
12
16
21
27
34
43
10
13
17
22
28
^NJ
Wk):
35
43
1967]
K1
j
x
N.(n,k):
i 12
245
16
32
64
128
256
192
11
12
24
48
96
H3
12
24
48
96
192
384
18
38
76
150
10
18
36
74
150
300
10
20
38
74
148
298
598
1'
4 1
10
25
60
130
10
15
25
50
110
240
10
20
35
60
110
220
460
15
35
70
130
240
460
920
5.
ADDITIONAL P R O P E R T I E S
T h e above t a b l e of v a l u e s for N.(n,k) s u g g e s t s the f o r m u l a s
(5.1)
(5.2)
ui
i <- (?)
N
n+r
0
Nk_
(2
2N k (n - k, k)
( n + km - r,k) = N n _ r ( n + km + r, k)
N (km + r , k) -
(5.3)
k,
n > k) ,
(2 K k,
n > k) ,
(n,k)
(n A r)
2N _ ^ k m + r , k)
To p r o v e (5.1), we have, using (3.5) and (3.6),
N k _.(n,k)
]=o
V /
r=o
< - r ( r ) N k < t t - r ' k ) k j=0

|"1<-1>r,"J(kIr)
^k+i
k-i
= (-D^ 1 E ("I)"fc)Nk(n r,k)

r=o
\ /
= (1 + ( - l ) k + 1 ) N k ( n - k,r)
which implies (5.1).
246
Oct. 1967
In the next place, it follows from (35) and (3o10) that

k-l
(5.4)
N (n,k) =
[V
r=Q
D k " l l (PT + l) n + j " k P" j r (1 4 j 4 k; n A k )
so that
k-l
(5.5)
N <n,k) =
J
s=o
(n;A(lAj-k;
V / r=s+j (mod k) \
It is clear from (5.4) that
+r(n
+ km - r,k) = ^ \V
s=o L
= V f"(p"s +
s=o L
=
2 1
^ - (P* + l ) 2 n + k m - r ] p - s < n + r >
D211"1"1^111
_ / p - s + 1 )2n+km-rl p s(n+r)
E1 r ^ s + D 2 n + k m - ( p s + D 2 n + k m - r i p- s fo- r )
which completes the proof of (5.2). We remark that (53) is an immediate corollary of (5.2).
Note that (5.2) requires only that n ^ r. This follows because (5.4) is
valid for all non-negative j
REFERENCES
1.
T. W. Chaundy, "Partition-Generating Functions, n Quarterly Journal of

Mathematics (Oxford), Vol. 2 (1931), pp. 234-40.
2.
G. H. Hardy and E. M. Wright, An Introduction to the Theory of Numbers,

Oxford, 1954.
3.
Charles Jordan, Calculus of Finite Differences, Chelsea, New York, 1947e

*****
NOTE ON A COMBINATORIAL IDENTITY IN THE THEORY OF BI-COLORED GRAPHS

H. W. GOULD*
West Virginia University, Morgantown, W . V a .
In connection with an enumeration problem arising in the theory of

labelled bi-colored graphs, Ce Y. Lee [2] has obtained the following identities. Defining N(a,b;n) by the expansion
(1)
ab
a
v
2 > a , b ; n ) x n = < - l ) a 4 l H k f J j { l - (1
n=o
k=o
^ '
x)k}b
and noting the lemma

b kj
<*>
kb
-Z2>' Z Z f(iJ)
j=oi=o
where <x>.= the smallest integer >x,
i=o j = < ^ >

Lee was able to show that
from which as a special case he deduced the apparently novel formula
It may be of interest therefore, to point out that the formulas may be

written in much simpler form inasmuch as the introduction of <x> leads to
* Supported by National Science Foundation Research Grants G-1409 5 and GP-482.
247
248
NOTE ON A COMBINATORIAL IDENTITY
[Oct.
unnecessarily complicated relations. Indeed it will be shown that relation (4)

is essentially trivial and may be generalized by methods of finite differences.
The simple nature of relation (4) was also missed in reviews of the paper [4l,
[5].
In order to determine N from expansion (1) it is not necessary to invoke
(2) and instead we shall merely make use of the fact that
( ; ) -
for
m < p
when m and p are integers with p > 0, m > 0.

From (1) we have in fact
ab
|>a,b ; n)x = (-l)

n=o
a+b+k
( * ) (-l)J (J>) (1 + ^
k=o
a
j=o
b
k=o j=o
W W
m
Y\C\
nô
v y v
a b
lr=0 i=n
k=o
j=o
n=Q
I \ i \ i
provided only that m ^ ab.,

Consequently we have
without any essential need for the restriction on range of summation in (3).
Of course, some terms are zero, but it is convenient to allow these to stand
in the indicated formula.
Then when we choose a = b = n in order to obtain the identity (4) found
by Lee, we see that this would appear more elegantly in the form
1967]
IN THE THEORY OF BI-COLORED GRAPHS
249
,k
?> M#)()
and we shall show in a simple way that this generalizes to give the relation
for any real value of ce

As for the proof, this comes from the familiar fact that when f(x) is a
polynomial of degree < n i n x ,
say
^ = Z ai ^ 5
i=o
then
<
0, n < r ,
8)
E
^ ^ k ) ^ "J( (_n
-1)%!
k=0
Since I ^
n d x I is a polynomial of degree n in x we have
z<-^( -*)
k=0
'
0, n < r ,
, (-d) . r = n ,
this being true for all real values of c and d. The identity is not new, and
appears for example in Schwatt [3, 104] and has been used by the writer [ l ]
in another connection0
Thus
NOTE ON A COMBINATORIAL IDENTITY

IN THE THEORY OF BI-COLORED GRAPHS
250
Oct. 1967
ki
H:)t: )
3=0
so that
n
k= o7:
, i
.Y
WW\ n /
\k/
but this is clearly n! by the familiar Euler theorem about n

n
differences of
powers of the natural numbers, or we may again apply (8).

If we define
do
a+b+k ' aa\ / b b\ / c +c k+ jk\r

^/ V V
(-i)
k=o j=o
vkAj
A n /
N^cn) = y y
with N(a,b, 0,n) = N(a,b;n), then we have the extension of (1) as
(11)
] T N(a,b,c,n)x n = (1
n=o
x)C^
( - l ) a W " Vl
k=o
Vk/
- (1
x) k f
and it would be of interest to know whether this yields any interesting result
about labelled bi-colored graphs.
REFERENCES
1. H. W. Gould, "Some Generalizations of Vandermonde T s Convolution,"
Amer. Math. Monthly, 63(1956), pp. 84-91.
2. C. Y. Lee, "An Enumeration Problem Related to the Number of Labelled
Bi-Coloured Graphs, " Canadian J. Math. , 13(1961), pp. 217-220.
3. I. J. Schwatt, An Introduction to the Operations with Series, Univ. pf
Pennsylvania P r e s s , 1924; Chelsea Reprint, 1962.
4. H. Kunneth, Review in Zentralblatt fur Mathematik, 97(1962), p. 391.
5. G. R. Livesay, Review in Math. Reviews, 25(1963), No. 1112.
ftDVANCED PROBLEMS AND SOLUTIONS

Edited By
RAYMOND E. WHITNEY
Lock Haven State College, Lock Haven, Pennsylvania
Send all communications concerning Advanced Problems and Solutions to

Raymond Whitney, Mathematics Department, Lock Haven State College, Lock
Haven, Pennsylvania, 17745* This department especially welcomes problems
believed to be new or extending old results.
Proposers should submit solu-
tions or other information that will assist the editor,
To facilitate their con-
sideration, solutions should be submitted on separate signed sheets within three

months after publication of the problems,,
H-119
Proposed by L* C a r i s f z , Duke U n i v e r s i t y , Durham, N o * C a r o l i n a .
Put
H<.n.^|||;-^*(-i)()-)(-"-')(--_-i)
/
/n - j +p - k \ / p - k + i \
\
p - k
A
*
;
Show that H(m,n,p) = 0 unless m, n,p are all even and that
xT/o
O O \
H(2m,2n,2p) =
min(m,n*p)
x^
t i\T
(-1)
v.
(m + n + p - r )J
r , r , ( m _ r ) , g _ r ) , fe _
t
r),=
(The formula
/m + n\
H(2m, 2n)
{ m )
where
-.-> - s- i ^( i r ) )(""j 1 + J )C-=r , X m "i-"" J )

251
252
ADVANCED PROBLEMS AND SOLUTIONS
[Oct
is p r o v e d in the Fibonacci Q u a r t e r l y , Vol. 4 ( 1 9 6 6 ) , pp. 323-325.)

H-120 Proposed by M . N . S . Swamy, Nova Scotia Technical College, Halifax,,
Canada.
T h e Fibonacci polynomials a r e defined by
f n + 1 (x)
= x . f n (x) + fn_lW
fi(x) = 1,
If z
(i)
f2(x) = x .
= f (x) . f (y) then show that

z
s a t i s f i e s the r e c u r r e n c e r e l a t i o n ,
2
2
z ,. - xv
+ 2)z , - xJ y z , + z
J z lo - x(x + Jy
n+4
n+3
n+2
n+l
n
(ii)
x
H-121
n
(x + Jy) 2 ]T z
V
r
=0.
= (z , - z
) - v(xy
- l)(z
_,_, - z ).
J
x
n+2
n-l
n+l
n
Proposed by H, H. Ferns, University of Victoria,, Victoria, B . C . , Canada.

Prove the following identity.
?j Vk
Fmi+K = F
1=1
where F
i s the n
\ nTkj nk A
Fibonacci number,
"
< m * k> >
m, A a r e any i n t e g e r s o r z e r o and
k is an even i n t e g e r o r z e r o .
W r i t e t h e f o r m t h e identity t a k e s if k i s an odd i n t e g e r .
Find an analogous identity involving L u c a s n u m b e r s .
H-122
Proposed by R. E. Whitney, Lock Haven State College, Lock Haven, Pa

Let F
denote t h e n
Fibonacci n u m b e r e x p r e s s e d in b a s e 2. Conn
th
sider the ordered a r r a y F - ^ F s .
Let g denote t h e n
digit of t h i s
a r r a y . Find a formula for g . If p o s s i b l e , g e n e r a l i z e for any b a s e .
1967]
H-70
2 53
Proposed by C . A e C h u r c h , J r . , W e V i r g i n i a U n i v 9 / M o r g a n t o w n , W . V i r g i n i a .
For n = 2m5 show that the total number of k-combinations of the first
n natural numbers such that no two elements i and i + 2 appear together in
the same selection is F 2 , and if n = 2m + 1, the total is F
F
,
m+2
m+2 m+35
Solution and comments by the proposer-
For his quick solution of the "probleme desmenages" Kaplanskyf2l gives

two results for combinations with restricted positions. We state them in the
following form:
The number of k-combinations of the first n natural numbers, on a
line, with no two consecutive is
\
rI/ n~- :kk +Il n
; o< k<^j-i
a)
if arranged on a circle, so that 1 and n are consecutive, the number is

n
(2)
/n - k \
^^x r : ~ l ,
n - k 1 k J
See also [4, p0 198Je
Summed over k,
0<k<|
(1) and (2) give the Fibonacci and
Lucas numbers, respectively,,

For the problem as stated we use (1).
The restriction that i and i + 2 cannot appear in any selection can be
stated as (a) no two consecutive even integers appear and (b) no two consecutive odd integers appear s
If n = 2m, a k-combination with the stated restrictions will be made
up of s integers from among the m even, no two consecutive, and k - s
from among the m odd, no two consecutive
k
(3)
k\
Thus there are
/ m - s + 1 \ / m - (k - s) + l \
k s
k-combinations of the first 2m natural numbers such that i and i + 2 do not

appear.
254
[Oct,
Summing (3) over k we get the total number
2
F2
m+2
=
~
R?]
V
yk / m - s + l \ / m - ( k - s ) + 1\
s
k
s=o\
A
~S
/
feo
with the usual condition that
(>)
for
b > a > 0
For n = 2m + 1 we choose s from among the m even integers, no

two consecutive, and k - s from among the m + 1 odd integers, no two consecutive, to get that there are
k / m - s + l \ / m - (k - s) + 2 \
yx
(4)
k\
s=<
)\
k s
k-combinations of the first 2m + 1 natural numbers such that i and i + 2 do

not appear.
Summed over k,
(4) gives the total number

m+l
fisr-.-r-t-n
F
m+2 m+3
It is also of interest to consider the circular analog of this problem by

way of (2)0
For n = 2m, 2 and 2m are taken to be consecutive as are 1 and 2m.
- 1. By the same argument as before we find that there are
k
v-^
^ m m
- s
s=o
m - s\
s
/m - (k - s)\
J m - (k - s) \
k-s
circular k-combinations such that i and i + 2 do not appear and a total of
1967]
2
L
H k
s=o
('m s
ni
k
255
/ m - (k - s ) \
k
s
- * "^V
"
/ '
F o r n = 2m + 1, 2 and 2m a r e consecutive as a r e 1 and 2m + 1 and

we have the total
m
L
L
m
^4
m+l
fco
sô
"
/ m ~ s\
s
m + l
m
- * -
s)
/ m - (k - s) + l \
+
Mixed r e s u l t s can also be obtained using both (1) and (2).
For example,
one can take l i n e a r combinations on the evens and c i r c u l a r combinations on

the odds.
Remarks.
T h e p r o b l e m posed in H-70 f i r s t appeared in t h e l i t e r a t u r e in
a p a p e r by N. S. Mendelsohn ["3J; an explicit formula was not obtained.

f i r s t explicit formula w a s given by M. A b r a m s o n [l,
solution for t h e n u m b e r of
k-combinations
l e m m a 3j
such that
and
The
AbramsonTs
i + 2 do not
a p p e a r t o g e t h e r is
/ n - 2k + s + 2 \
k s
k\
/ k -
s\
M )
REFERENCES
1.
M. A b r a m s o n ,
"Explicit E x p r e s s i o n s for a C l a s s of P e r m u t a t i o n P r o b -
lems, " C a n a d ^ J M a ^
7(1964), pp. 345-350.
2. I. Kaplansky, Solution of the " P r o b l e m e d e s Menages,
Bull. A m e r . Math
Soco , 49(1943), 784-785.

3.
N. S. Mendelsohn, "The Asymptotic S e r i e s for a C e r t a i n C l a s s of P e r m u tation P r o b l e m s , Canad. J. of Math. , 8(1956), pp. 234-244.
4.
J . Riordan, An Introduction to Combinatorial A n a l y s i s , New York,
1958.
H-73 Proposed by V . E . Hoggatt,Jr 8 / San Jose State ColIege / San Jose, Calif.
Let
f0(x) = 0,
f t (x)
= 1
256
and
f n+2 (x) = xf n + i (x) + f n (x) ,
[Oct.
n - 0
and let b (x) and B (x) be the polynomials in H-69; show

f
^ (x) = xB v(x 2 ) ,
2n+2
n
and
f
2n+l
<x) = b (x 2 )
n
T h u s t h e r e is an intimate r e l a t i o n s h i p between the Fibonacci polynomials, f (x)

and the Morgan-Voyce polynomials b (x) and B (x)
Solution by Douglas L i n d , U n i v . o f V i r g i n i a , C h a r l o t t e s v i l l e , V i r g i n i a .
U s i n g t h e explicit r e p r e s e n t a t i o n s of B (x) and b (x) given in H-69, and

of f (x) given in B - 7 4 , we find
2n-2r
2n-2r+i
c
t .
= f 2n +2( x ) >
,v
T h e s e r e l a t i o n s have been given by Rc Ae Hayes \_ "Fibonacci and Lucas P o l y n o m i a l s , " ( M a s t e r ' s T h e s i s ) ; equations (3.4-1) and ( 3 . 4 - 2 ) ] .
H-77
Proposed by V . E . H o g g a t t , J r 0 /
San Jose State C o l l e g e , San Jose, C a l i f .
Show
2n+i / _
, i \
=h )
for all i n t e g e r s k.
2k+2j+l
2n+2k+2
Set k = -(n + 1) and d e r i v e
5W
^n
1967]
257
a r e s u l t of S. G. Guba P r o b l e m No* 174, I s s u e No, 4, J u l y - A u g u s t 1965, p .

' 73 of Matematika V. SkSle.
Solution by L. Carlifz, Duke University, Durham, North Carolina.
Since
Ln = a
+ ^B s,
Fn =
n
0n
a - p '
]8 = I
(1 - V 5 )
where
a = |
(1 + V 5 ) 5
1 + Q'2 = o^V5 ,
1 + p2 = -pV5
it follows that
/ \
k+2
/ \
a k ( l + a2)
- k ( T + j3 2 ) n
a - j8
=
( a k + n - (-l)n/3k+n) (V5)n
V5
( 5(n-l)/2L,A
_ J
k+n
( 5
Fk+n
thus g e n e r a l i z i n g the s t a t e d result 0
(")*Note that
(n odd)
(n even) ,
In p a r t i c u l a r , for k = - n ,
2.5M/2
j
J
(nodd)
.
(n even) .
we get
258
2
f72n+l\F
_f/2n+l\F
[Oct.
^/2n+l\
-2n-i+2j
-5("; l )w^(T 1 )',
2n+i-2j
S(2"-/)
>. i
r i F2J+1
3=l
so that
Sfr )
F2j+i
"
5n
Similarly, we have
n / \
';) L k+2 . = k a + C 2) n + /s k a + /s2)r
SO
fck+n
i
In particular, since
+ (-l)n^k+n)(V5)n
5
L, ,
(n even)
k+n
V
k+n
(n odd) .
(n+1)/
gfflw-W $(?)--*-?(T)
-2n+2j
so that
LATE ACKNOWLEDGEMENT: Problems H-64, H-71, H-72, H-73, and H-77

were also solved by M. N. S. Swamy.
****-
DIRECT CALCULATION OF k-GENERALIZED FIBONACCI NUiBERS

IVAN FLORES
Brooklyn, New York
SUMMARY
A formula i s developed for d i r e c t calculation of any k - g e n e r a K z e d F i b onacci n u m b e r u. ,
without i t e r a t i o n .
3sK
DEFINITIONS
The o r d i n a r y Fibonacci n u m b e r u.
i s defined by
3*2
(1)
V ;
u.
= u.
+ u.
3*2
]-l,2
J-252
U(i
> 2)
~ 7
with the additional conditions u s u a l l y i m p o s e d

(2)
u 0 . 2 = 0;
uU2
= 1.
T h e k - g e n e r a l i z e d Fibonacci n u m b e r u. ,
k
i s defined a s the s u m of i t s
predecessors
'
j,k
j~i s k
j-2 # k
j-ksk
i=j-k
t o g e t h e r with the initial conditions
(4)
A table of
ujjk = 0
u. ,K
3
(0 L j L k - 2);
V l j k
= 1
from k = 1 to 7 and j = 0 to 15 i s found in T a b l e 1.

259
[Oct
DIRECT CALCULATION OF k-GENERALIZED
260
Table 1
Fibonacci Numbers u.] k, for Various Values of i and k
5
iP
1 0 1 2 3 4 5 6
13 14 15
233
377 610
89 144
12
11
9 10
16
2 0 1 1 2 3 5 8 13 21 34 55
3 0 0 1 1 2 4 7 13 24 44 81 149 274 504 927 1705
4 0 0 0 1 1 2 4 8 15 29 56 108 208 401 773 1490
5 0 0 0 0 1 1 2
8 16 31
61 120 236 464 912
6 0 0 0 0 0 1 1
8 16
32
63 125 248 492
7 0 0 0 0 0 0 1
16
32
64 127 253
TERM RATIO
The key to direct calculation is the existence of a fixed ratio r, between
successive u. , ! s so that in the limit we have
lim
(5)
]+i ; k
n>oo u. ,
If such a ratio can be founds
an approximate calculation is simple.
Vorob'ev [ s ] has shown that for k = 2, this requires the solution of

q^ = q^"1 + q^~2
(6)
which for q f 0 reduces to

(7)
q2 - q - 1 = 0
for which the roots are

.(8)
ri
= Lz^l
* -0.6180
and
r 2 = ^ y ^ - * 1.6180 ,
where s& means "approximately equal to* "

If f
f
n " f n-i
(9)
is any Fibonacci sequence obeying

obe
the difference equation f
3
then
has the form
< s e e E4]*
b
iri
2r2
1967]
261
FIBONACCI NUMBERS
Since ( r ^ < 0.7,
jr^j < \ 9
such that for all n > N s
so that | r p | < l / 2
un
u
(10)
Hence t h e r e e x i s t s an N
is the g r e a t e s t i n t e g e r to b 2 rf,
js2
* V2
To evaluate the constants ht

b
(ID
and we w r i t e
0 > N)
and b 2 , we u s e the initial conditions
l + b2 = u0j2 = 0
h1T1 + b 2 r 2 = u l s 2 = 1 ,
which yield
(12)
An exact e x p r e s s i o n for u.
(13)
hl
V5
bi = V5: ,
b 2 = V
5
i s hence the f a m i l i a r Binet f o r m
H-(^)]
GENERALIZATION
To find an e x p r e s s i o n for the k - g e n e r a l i z e d Fibonacci n u m b e r s , let u s

f i r s t s e e k solutions to (3) which f o r m a g e o m e t r i c p r o g r e s s i o n say aq . Then
(3) l e a d s to a g e n e r a l f o r m of (6),
(14)
J- 1
aq
= aq
j~2
+ aq
,
j-k
+ + aq J
Thus
(15)
aq
j-k, k
k-i
(q - q
q - 1) = 0 .
Since we a r e looking for solutions which a r e not identically zero,,

a s s u m e a ^ 0 and q f 0o
(16)
T h e r e f o r e we s e e
k
k-i
q - q
- q - 1 = 0
we can
262
,th
This k
r, , .
[Oct.
degree equation has k complex roots, say r t ^, r 2 ^5 ,
Now Miles [5j has shown that these roots are distinct, that all but one
of them lie within the unit circle in the complex plane, and that the remaining
root is real and lies between 1 and 2.
Hence with a suitable choice of sub-
scripts we may write

(17)
x (1
^ ^k-x) >
Kkl-<
(18)
k,k *
Since the roots are distinct, the Vandermond determinant

1
(19)
i,k
k-i
l,k
k-i
r
2,k
i,k
r
2,k
r.k,k.
2,k
rfk,k.
f 0 ,
k-i
k,k
and Jeske f4j has shown that the general solution can be written
(20)
b r
E
*-* -I -I,
i=i
u.j , k
To evaluate the constants b., we use the initial conditions

i
% Vii
1=1
(21)
k-i
(m =
,k =
? V?--
' 1 ' , , ' k - 2) >
i=l
This system has a unique solution by (19) which can be found using Cramer's
rule. This yields
1967]
FIBONACCI NUMBERS
k
< 22)
i -
<ri,k ' V k r l
QF=1
so that (20) becomes
(23)
M (ri^ - W
Wi
Recalling (17) and (18), we remark that as j becomes large ri , becomes

the dominant term in (23), so that as before there exists an N such that for
all j > N, u. . is the nearest integer to b, rj" , . We may therefore write
(24)
uj;k* b k 4 k
(j > N) .
It follows from (24) that

u.
lim - C M = r .
u. ,
k,k
(25)
J->00
J,k
and more generally
(26)
lim ! & k = r m
APPROXIMATIONS
We first note that as k>oo the sequence u. , , approaches the geoJ-K,K
metric progression of powers of two,

1, 2, 4, 8, 16, 32, 64,
as can be seen from Table 1. It follows that
264
(27)
lim r k k
kôo
'
[Oct.
= 2 .
See T a b l e 2 for calculated v a l u e s of the p r i n c i p a l root r. ,
for k = 2 to 19,
which gives s t r i k i n g verification of (27).

Table 2
Fibconacci Roots
1.6180340
3
4
1.8392868
1.9275621
1.9659483
6
7
1.9835829
1.9960312
1.9980295
10
1.9990187
11
1.9995105
12
1.9997556
13
1.9998779
14
1.9999390
15
1.9999695
16
1.9999845
17
1.9999925
18
1.9999962
19
1.9999981
1.9919642
F r o m (25) we get then
u
(28)
lim
lim
k->oo
j-9>oo
3
U
,
*' K = 2 ,
j
j k
which w a s stated in an equivalent form by P . F . B y r d [ l l . We shall now show

-k
that b, i s a p p r o x i m a t e l y r, , in the s e n s e that
1967
FIBONACCI NUMBERS
(29)
lim
^
265
b. / r " . = 1 .
k / k,k
To p r o v e (29), f i r s t r e c a l l that
(r
(r
k,k~ ^ k ^ "
k,k-
k-i,k)
Since
x
- x
k-i
x - 1 = (x - r l j k ) (x - r
)
k,k' '
and
f(x) = ( x - l ) ( x k - x k " 1
x-1)
=x
k + 1
-2x
= ( x - l ) ( x - r i k ) - . - ( x - r. . ) ,
k,k'
we find
.
, 1
f ' ( r k ) k ) = (k + l ) r ^ k - 2 k r k = ( r ^ - iHr^
- r1>k)-. ( r ^ - rk_1>k)
Hence
k,k
k
"
(k+i)rk)k-2krk;k
f r o m which (29) follows, s i n c e r , + k - 2 r k
fe
= - 1 . Then for sufficiently l a r g e
j and k we m a y w r i t e
ujjk* rk
(30)
Call the approximation for u. . in (24) u! , . Then using (20), the e r r o r

c o m m i t t e d by this approximation i s
k-i
(31)
By (17)
w. , =
j,k
Ir. , | < 1 for
1 1 . . - ul ,
j,k
jsk|
1 L i 4 k - 1,
i=i
so the t r i a n g l e inequality shows
266

FIBONACCI NUMBERS
k-i
Oct.
k-i
w.j 5 ,k ~^ Ej IMi | | lr-i,k|i I < .Et j I|b-|

i|
(32)
Note that the f i r s t inequality in (32) shows that

(33)
l i m w. .
J->oo J
= 0 ,
so that for fixed k the e r r o r tends to z e r o a s j b e c o m e s l a r g e , giving f o r m a l justification to (24).

EXTENSION
In the n e a r future t a b l e s of b,
will b e p r e p a r e d by c o m p u t e r s .
t o g e t h e r with r, , will p r o v i d e an excellent approximation for u. ,
These,
using an
analytic p r o c e d u r e .
T h e author e x p r e s s e s h i s a p p r e c i a t i o n to D a A. Lind and V Eo Hoggatt,
J r . , for helping in p r e p a r i n g t h i s p a p e r .
REFERENCES
1.
P . F . Byrd, P r o b l e m H - 6 1 , Fibonacci Q u a r t e r l y , 3(1965), 201.

5(1967).
2.
M a r k F e i n b e r g , "New Slants, " Fibonacci Q u a r t e r l y , 2(1964), 223-227.
3.
Mark Feinberg,
No. 3, 71-74,
4.
J a m e s A. J e s k e , " L i n e a r R e c u r r e n c e R e l a t i o n s ,
Q u a r t e r l y , 1(1963), No. 2, 69-74.
5.
E. P . M i l e s , J r . , " G e n e r a l i z e d F i b o n a c c i N u m b e r s and A s s o c i a t e d M a t r i c e s , "

A m e r i c a n Math. Monthly, 67(1960), 745-57.
6.
N. N. V o r o b ' e v ,
"Fibonacci-Tribonacci,
Fibonacci N u m b e r s ,
* * *
Fibonacci Q u a r t e r l y ,
Part I,"
Solution,
1(1963),
Fibonacci
B l a i s d e l l , New York, 1961.
EQUATIONS WHOSE ROOTS ARE T I E nth POWERS

OF T I E ROOTS OF A GIVEN CUBIC EQUATION
N . A , DRAiMand MARJORIE BICKNELL
Ventura, C a l i f , and Wilcox High School, Santa Clara, Calif.
Given the cubic equation

x3 - c ^ 2 + c2x - c 3 = 0
with r o o t s r l 5
._
(1)
r2,
r3j
the p r o b l e m of this p a p e r i s to w r i t e the equation
o
, n
n
IK o
, n n
nn
n IK
x6 - (ri + r 2 + r 3 )x* + ( r ^ + r ^ g + r 2 r 3 )x =
whose r o o t s a r e r ^ ,
of the coefficients
r2,
cl5
x3
C(l?n)x2
n n n
r ^ ^
C
(2sn)X~C(35n)
r 3 , and whose coefficients a r e e x p r e s s e d in t e r m s
c2,
c3s
of the given equation 0
T h i s p a p e r extends to the cubic equation a study initiated by the solution

of a s i m i l a r p r o b l e m for the q u a d r a t i c by the s a m e a u t h o r s [ 1 1 . J u s t a s a
s p e c i a l q u a d r a t i c equation l e a d s to a r e l a t i o n s h i p between the n
Fibonacci
n u m b e r and a s u m of binomial coefficients, so does a s p e c i a l cubic equation

th
r e l a t e the n
m e m b e r of a Tribonacci sequence to a s u m of p r o d u c t s of b i nomial coefficients. Some L u c a s identities also follow.
The s u m m a t i o n for initial values of p o w e r s of r o o t s by e l e m e n t a r y t h e o r y
y i e l d s the f i r s t five e n t r i e s in the table below.
Examination of this sequence
r e v e a l s an i t e r a t i v e p a t t e r n ; namely, that if
c
(ljn)
i + r2 + r 3 ?
n > 0
then
(l,n)
C C
i (l9n-l) "
2C(l3n-2)
3C(l3n-3)
which i s easily p r o v e d , s i n c e each r o o t , and hence s u m s of the r o o t s , s a t i s f i e s

the original equation,,
267
268
EQUATIONS WHOSE ROOTS ARE THE nth POWERS

Sums through eighth p o w e r s appear in the table below.
[Oct.
The r i g h t - h a n d
column gives the s u m of the absolute v a l u e s of the coefficients for each value
of n.
a>
n
n)
n
i
+ r2
Coefficient
Sums for n
C? -
2c 2
cl
30^2
*\
4c
5
6
cf c? -
c7i
\ -
i c 2 + 2c2
50^2 + 5lc2
6cfc2 +
+ 3c 3
+ 40^3
11
+ 5 0 ^ 3 - 5C2C3
21
9c 2 lC 2 - 2 c | + 6c^c 3 -
+ 3c 3 2
120^203
39
- 7c5c 2 + Ucfcjj - 7 0 ^ 1 + 7cjc 3 - 21c5c 2 c 3 + 7c \cz + lci cr 23
71
8cfc2 + 20c^c| - 16cfc| + 2c

+
8c
131
l c 3 ~ oZC-jC2Co + 2 4 c j C ^ + 1 2 c ^ c | -" 8c 2 c|
It i s p o s s i b l e to p e r c e i v e the g e n e r a l i z e d number p a t t e r n for s u m s of n
powers
of the r o o t s by extending the table above and b r e a k i n g down the s u m in t e r m s of

coefficients of p o w e r s of c 3 .
If ip
i s the coefficient of c 3 / n !
n
l
n
2 +
in the s u m
n
3
then
0 = c^ - nc^" 2 c 2 + n(n - 3 ) c f " 4 c | / 2 !
ip1 = n c j "
- n(n - 4)(n - 5 ) c ^ " 6 c | / 3 ! + . . .
- n(n - 4 ) ^ " c 2 + n(n - 5)(n - 6)c1~
cf/21
- n(n - 6)(n - 7)(n - 8 ) c f " 9 c f / 3 ! + - ip2 = n(n - 5)cf* 6 - n(n - 6)(n - 7 ) c ? ' 8 c 2 " + n(n - 7)(n - 8)(n - 9)c?~ i 0 cjj/2!
- n(n - 8)(n - 9)(n - 10)(n - l l ) c f ~ 1 2 c ^ / 3 ! + - . .
1967]
O F THE ROOTS O F A GIVEN CUBIC EQUATION
leading to the t h r e e equivalent e x p r e s s i o n s below 0
269
F o r CjCgCg ^ 0,
[n/3j
(2)
r ? + r f + r3
c3%k/k!
k=o
^k
[(n-3k)/2]
\~^
, ^ m n-2m-3k m .
_.
.,.. ..
_ rt. .,
=
2^
t" 1 ) c i
c 2 n(n - m - 2k - l ) ! / ( n - 2m - 3k)!m; ,
m=o
[n/3J [ (n-3k)/2
(3)
r? + rf +
r3n
(4)
T1
= J]
k=o
m=o
(-1) n(n - m - 2k - 1)1 n-2m-3k m k

(n - 2m - 3k)Jm!k! C l
2 3
[n/3][(n-3k)/2]
+ r 2 + r 33 =
V*
2Lt
2^
k=o
(-1) n
/ n-m-2k-l\/n-m-3k\
n - m - 3k \
m=o
) \
Ky
X
where
[n]
i s the g r e a t e s t i n t e g e r ^ n
and
n-2m-3k m k
i
c2 c3 ,
J i s a binomial coefficient.
Notice that the coefficients of c$ a r e the s a m e as the coefficients which a r o s e

in studying the r o o t s of the q u a d r a t i c in [ 1 J .
the t e r m s
cn
The r e i t e r a t i v e r e l a t i o n s h i p of
s u g g e s t s a proof by m a t h e m a t i c a l induction for the t h r e e
f o r m u l a e l i s t e d , and such a proof has been w r i t t e n by the a u t h o r s .

s a k e of b r e v i t y , the proof i s omitted.
F o r the
A d e r i v a t i o n of the above f o r m u l a s
could a l s o be done using W a r i n g ' s formula and Newton's identities (see [ 2 ] ) .

Thus f a r , we have found a way to e x p r e s s the coefficient for x 2 in our
general problem.
The coefficient for
has a s i m i l a r computation.
(2,n)
x,
n n
riI>2
n n
riI>3
n n
r2r3
'
In the auxiliary cubic equation
,
n nw
n nw
n n.
_
(x - rir2 )(x - rtT3 )(x - r 2 r 3 ) = 0
notice that
c.
. is the coefficient of
(z, n;
b e c o m e s , upon multiplication,
x2.
When n = 1,
the above cubic
EQUATIONS WHOSE ROOTS ARE THE n t h
270
POWERS
[Oct.
x 3 - ( r ^ + r 1 r 3 + r 2 r 3 )x 2 + (rl|r 2 r 3 + r 1 r ^ r 3 + r 1 r 2 r | ) x - r ^ r ^ r |
=
X3 -
C 2 X 2 + CjCgX -
0.
C o m p a r i n g this equation with the equations of our original p r o b l e m , we can

apply
the t h r e e formulae a l r e a d yJ derived for cM
to find c / 0 s if we r e ^ J
( l , n )v
(2,n)
p l a c e c 1 by c 2 ? c 2 by c^Cg, and c 3 by c | . F o r e x a m p l e , from Equation (4),
if c ^ C g fi 0,
<K
(5)
n n
vv
rtr2
our f o r m u l a for c , 2
n n
+ vr + r r
+ r i r 3 + r2r3
. becomes
[n/3] [(n-3k)/2]
V
V
(-1) n(n - m - 2k - 1)!
^
Z,
^ n ~ ^ n ~ ^ 3 k ) ! m! k!
k=o
m=o
m n-2m-3k 2k+m
In p r a c t i c e , when r a i s i n g the r o o t s of a given equation, It is s i m p l e r to utilize

the method of i t e r a t i n g functions than to substitute into the f o r m u l a e , especially
as n becomes larger.
An example worked by each method follows.
Given the equation

x 3 - 6x2 + l l x - 6
0 ,
w r i t e the cubic whose r o o t s a r e the fourth p o w e r s of the r o o t s of the given e q u a tion, without solving for the r o o t s .
(A) By substitution;
F r o m the table given e a r l i e r or from Equations (3)
and (5), the d e s i r e d cubic Is

x 3 - (cf - 4CjC2 + 2c 2 + 4c 1 c 3 )x 2 + (e2 - 4 0 ^ 0 ^ + 2G\C\ + 4c 2 c 3 )x - c 3 = 0.
Substituting
c
i = 6,
c2 = 1 1 , c3 = 6
yields
x 3 - 98x 2 + 1393x - 6 4 = 0
1967]
OF THE ROOTS O F A GIVEN CUBIC EQUATION
with r o o t s
l4,
24,
271
and 3 4 . A s a check, the r o o t s of the given equation a r e
1, 2, and 3.
(BJ By i t e r a t i o n : To get c.
(l,0)'
(l,l)'
i =
. , we wish to w r i t e the sequence
(l,3)'
( l , 2 ) = 21 "
2C
(l,2)'
(l,4)'
Now
(l,0)
'
(l,l)
>
2 = 36 - 22 = 14
By the i t e r a t i o n r e l a t i o n s h i p ,
(1,3)
C C
i (l52)-C2C(l,l)
(l 4)=
( l 5
6(6)
0)
~ n(14)
(36)
Similarly,
c / o m = 3,
'(2,0) " "'
c / 0 1N = 1 1 , c / 0 Q^ = cr22 - 2 0 ^ 3 = 49.
"(2,1) " " '
"(2,2)
Since
(n,2)
C2C
( 2 , n - l ) ~ C * C 3 c (2 5 n-2)
3)
= 251 and c
3c(2,n-3)
substitution y i e l d s
c
= 1393 ,
yielding the s a m e cubic as in (A).

Next let us t u r n our attention to s e v e r a l special cubic equations.
for the cubic
x3 - x2 - x - 1 -
0,
c(lj0)
3,
c(ljl)
1,
c(l^2)
First,
EQUATIONS WHOSE ROOTS ARE THE n t h POWERS.
272
[oct.
and oux r e p e a t i n g m u l t i p l i e r s in the i t e r a t i v e r e l a t i o n s h i p a r e 1, 1, 1.
(l,3)
1(3)
1(1)
'"'
1(3)
(l,n)
'
C
(1 4)
(l,n-1)
(l,n-2)
+ C
Then,
'
(l,n-3)-
F o r this p a r t i c u l a r equation we have a s p e c i e s of T r i b o n a c c i n u m b e r s , any t e r m

after the t h i r d being the s u m of the t h r e e p r e c e d i n g t e r m s , with the entry t e r m s
3,1,3.
By Equation (4), the n
[n/3] [(n-3k)/2]
n
iJ
JL-J
k=o
m=o
term
(-l)mn
n - m
m -3k
in this T r i b o n a c c i sequence i s
/ n - m - 2 k - l \ / n - m - 3 k \
k
/\
m
/
Notice that the s u m s of the coefficients in the table given for. c,., . a r e t h e s e
(l,n)
s a m e n u m b e r s . It is i n t e r e s t i n g to r e c a l l that the special equation
x2 - x - 1 = 0
led to a f o r m u l a r e l a t i n g the n
m e m b e r of the Fibonacci sequence to a s u m
of binomial coefficients in the e a r l i e r study of the q u a d r a t i c equation [ 1 ] ,

Considering the s p e c i a l equation
x3-x
with r o o t s
+ x - l
= 0
1, V^T, we can w r i t e the following from Equation (4):
[ n / 3 ] [(n-3k)/2]
Zu
k=o
Z.
m=o
n -
m - 3k ^
H
'
I l if
.f
I
'
L , defined by
J
n'
'
Of m o r e i n t e r e s t , however, a r e the following identities for the n

number
is odd
Lucas
1967]
O F THE ROOTS O F A GIVEN CUBIC EQUATION

Li = 1,
L 2 = 3,
Ln = ~Ln-i + L n _ 2
273
We substitute in Equation (3), using
;ri
= a
and letting r 3 v a r y .
(1 + N / 5 ) / 2 ,
r2
(1 -
N/5)/2
If r 3 = 1, Equation (3) cannot be used directly b e c a u s e

c 2 = a/3 + (3 + a = 0,
and 0 i s not defined,

s e e n that, if c 2 = 0,
/ofX
( 3!)
c ^ g 09
r,
But, by following the derivation for Equation (3), it i s
n
+
r2
[n/3]
\-^
n
+
r3 =
n(n - 2k - 1)1 n-3k k
(n - 3k)!k!
k=o
Since
c 1 = a + fi + 1 = 2,
c 3 = a/3 = - 1 ,
and
= a
JI
+ Bp
substitution gives
K3]
L
LJ
k n-3k
(-1) 2
n(n - 2k - 1)]
(n - 3k)I k!
k=o
In g e n e r a l , if
r 3 = P,
P ^ 1,
P -1,
P ^ 0 ,
274
EQUATIONS WHOSE ROOTS ARE THE n

POWERS
OF THE ROOTS O F A GIVEN CUBIC EQUATION
Oct. 1967
Equation (3) gives
+ nn -
[n/3]
V
2-J
[(n-3k)/ 2 ]
V
m + k
(-1)
2-J
k=o
. ^ m - s k ,
^mk
n(n - m - 2k - 1)1 (p + 1)
(p-1) p
(n - 2m - 3k>! m* k!
m=o
S i m i l a r l y , Equation (31) gives the following two identities using
rt
= a,
r2
p,
r3=
- I/N/5" ,
and the known r e l a t i o n s h i p for Fibonacci n u m b e r s ,

F n = {an-/3n)/M5
Below,
n i s taken to be 2s + 1 and 2s
F 92S+1 ~
1/58+1
=
~
s+i _
X/
T
+ i / ,
L 2 g + 1/5
_
-
respectively.
f ( 2 S + l ) / 3 J ,n
,w
^k,2S-3k + i
v-*
(2s + 1) (2s - 2k) ! (-1) 4
/_j
/rt
[2s/3J
V
>
^
s-k+i
(2s - 3k + l ) ! k ! 5 ^
k
2s(2s
- 2k - 1) I (-if
:
(2s - 3k)J k! 5 S K
? s-3k
42S 3 k
REFERENCES
1.
N. A. D r a i m and M, Bicknell, "Sums of the n

Given Q u a d r a t i c Equation,
P o w e r s of the Roots of a
The Fibonacci Q u a r t e r l y , 4:2, A p r i l , 1966y
pp. 170-178.
2.
W. S. B u r n s i d e and A. W. Panton, An Introduction to the T h e o r y of Binary

A l g e b r a i c F o r m s , Dover, New York, 1960, Chapter 15.
* *
_ J t i A T E ! J I INTERESTS IN THE FIBONACCI SERIES II

CALCULATION OF FIBONACCI NU1BEIS AND SUMSFROM THE BINOIIAL
JOSEPH MANDELSON
U.S. Army Edgewood Arsenal, Maryland
A s mentioned in an e a r l i e r p a p e r 5 my i n t e r e s t in the Fibonacci s e r i e s

s t e m m e d f r o m the o b s e r v a t i o n (in 1959) that the p r e f e r r e d r a t i o s developed in
the r e s e a r c h of my colleague., EL E l l n e r s and l a t e r included in D e p a r t m e n t of
Defense Handbook H109 [ l j s w e r e
1, 2, 3S 5, 8.
When the supposition was
tested^ that all p r e f e r r e d r a t i o s would c o m e from the Fibonacci s e r i e s ^

next r a t i o was c a l c u l a t e d and was found to b e 13.
the
Then it was noted that the
s a m p l e sizes., Acceptable Quality L e v e l s (AQL f s) s and lot s i z e r a n g e s of all

s a m p l i n g s t a n d a r d s since the o r i g i n a l work in this field by Dodge and Romig
2] w e r e s e r i e s a p p r o x i m a t e l y of the t y p e :
(1)
u ^ = u ,. + u
n+2
n+i
n
v ;
It s e e m e d self-evident that, in s o m e way,, the Fibonacci s e r i e s m u s t be i n t i m a t e l y connected with s o m e p r o b a b i l i t y distribution such a s the Binomial e x pansion,
T h r o u g h a little a l g e b r a i c juggling such a connection was quickly
e s t a b l i s h e d a s follows :
The method of finite differences^ d e s c r i b e d in C h r y s t a l [ 3 ] yields i n t e r esting results.
If s u c c e s s i v e differences be taken between adjacent Fibonacci
numbers:
(2)
v ;
= u ,, - u
n+i
n
i,n
a s e r i e s of f i r s t - o r d e r differences
dt
i s generated*
In the s a m e way 5 a
s e r i e s of s e c o n d - o r d e r differences may be g e n e r a t e d :
(3)
v
'
= d
2,n
_,, - d,
i, n+i
i,n
H i g h e r - o r d e r differences m a y b e g e n e r a t e d in a c c o r d a n c e with the

relationship:
275
general
276
CALCULATION OF FIBONACCI NUMBERS
(4)
^ n
[Oct.
(k~i)*(n+l) " d (k-l),n
Taking the Fibonacci series itself, u , to constitute the zero order of

differences d0 n
and if j is some given value of n, and k is the order of
differences, we get the following table;

1
*o
13
21
34
55 89 144 233 377
dt
13
21
34 55
d2
13
21 34
55
89 144
^3
-1
13 21
34
55
89
<*4
-1
8 13
21
34
55
d5
-3
-1
13
21
34
<*6
-3
-1
13
21
d,
-8
-3
-1
13
d8
13
-8
-3
-1
j=
d9
d
10
10 11
-21
13
-8
-3
-1
34
-21
13
-8
-3
-1
12
13
14... et<
u
n
89 144 233
It may be seen that as k, the order of differences, and j are increased

without limit, the table of d,
and j forms, both horizontally and vertically,
four Fibonacci series centering on each zero, such that the two series, one
above and one to the right of each zero are positive in all their terms while the
series to the left and below each zero have alternate negative terms.
In e s -
sence, the latter series constitute the negative branch of the Fibonacci series,
u
-n
(5)
.
We can calculate u n , n = k +j, from the differences d . as follows:
Ts+j
= d.
0,]
kd. + J % ^ d .
1,]
2!
2,3
* - y
3!
- 2> d .
3,3
+... +
dt.
k!
k,j
1967]
AND SUMS FROM THE BINOMIAL
where &
is d
*>J
277
of t h e o r d e r k a s shown in t h e table and u. = d .. T h e
coefficients of t h e d,
(a + b ) K .
t e r m s r e p r e s e n t t h o s e of t h e Binomial
0,1
Expansion,
E x a m p l e 1,
Calculate u , + . when k = 7, and j = 3 (u, . = u 1 0 )
U 7+3 = d 0 j 3 + 7d 1 ? , + ^ p d2}3 + ^ | d, 53 + . . + | [ d 7j3

=
2 + 7(1) +
W
1M
2
(1) +
W
IM(5)
3(2)
7(6) (5) (4) (3)

5(4) (3) (2)
lUj
7(6)(5)(4)
4(3) (2)
(1
7(6) (5) (4) (3) (2)

(6) (5) (4) (3) (2) l
l x;
'
7[
7! [ ^
_
42
210
840
2520
,5040 , 9 . . - , Q,
- 2 + 7 + (1) + (0) + - ^ (1) +
(-1) + ~ ^ (2) + l ( - 3 )
= 2 + 7 + 2 1 + 0 + 3 5 - 2 1
+ 1 4 - 3 = 55 = u 1 0 .
The s u m of consecutive t e r m s of the Fibonacci s e r i e s i s given by?

n=j+k-i
E x a m p l e 2a
Calculate t h e s u m of k = 9 consecutive Fibonacci n u m b e r s s t a r t i n g with
uj U = 4 ) .
n=j-Hk-i
u
n=
n=j
n=9+4-l=l2
9 x 8 .. . 9 x 8 x 7 ... , 9 x 8 x 7 x 6 ,..,
u n =9(3) + - T -(2) + - T l r i -(l) +
4 x 3 x 2
9 x 8 x 7 x 6 x 5 . ,
W
5x4x3x2
9x8x7x6x5x4
6x5x4x3x2
9 x 8 x 7 x 6 x 5 x 4 x 3 x 2
8 x 7 x 6 x 5 x 4 x 3 x 2
l )
{)
9x8x7x6x5x4x3
7x6x5x4x3x2
9[ , ,
9! ( }
= 27 + 72 + 84 + 126 + 0 + 84 - 36 + 18 - 3 = 372
n=i2
J^ u
n=4
(1)
n=4
= 3 + 5 + 8 + 13 + 2 1 + 34 + 55 + 89 + 144 = 372
278
CALCULATION O F FIBONACCI NUMBERS
[oct.
It i s noted that d. . = d , . such that when Ji - k = p

dn i s the s a m e n u m F 33
k,3
o,k-j
e r i c a l l y and i s positive when p i s p o s i t i v e o r when p i s negative and o d d

However,
d 0 i s negative when p i s negative and even.
Since d 0 ,
series,
u ,
the z e r o - o r d e r of difference,
equations 5 and 6 may be w r i t t e n in t e r m s of u
e r e n c e i s made to the p r o p e r sign of u. ,

examining the table of d.
=
d 0 -;_! = U 4 1 ?
k+]
i s the s a m e a s the Fibonacci
etc e
provided ref-
when j - k i s negative*
Thus,
and j f o r m s , it may b e s e e n that d 0 ; = u-j, d 1 *
Hence, equations 5 and 6 may be r e c a s t a s follows:
3-1
2!
3-2
3!
3-3
k!
j-k
and
n=j+(k-l)
(8r
T
LJ
n=j
v ;
u = k u .
n
3
^ | r ^ u .
+^
2!
]-i
^ P r o v i d e d the sign of u. ,
Also,
3!
. . .
j =2
k!
^ u .
,.
k! j - ( k - i )
is:
P o s i t i v e when j - k
i s positive
P o s i t i v e when j - k
i s negative and odd
Negative when j - k
i s negative and even.
u 0 = 0o
E x a m p l e 3,
Let j = 3 and k = 7.
,7x6
,7x6x5
Calculate u , + . = u 10
,7x6x5x4
7x6 x 5 x 4 x 3
u 10 - u 3 + 7u2 + - F -u 1 + - T F r u 0 + - r x 3 x 2 *-i + -TTTZTirr*

7 x 6,x 5 x 4 x 3 x 2
6 x 5 x 4 x 3 x 2 U~3
7!_
7!U~4
= 2 + 7(1) + 21(1) + 35(0) + 35(1) + 21(-1) + 7(2) + l(-3)

= 2 + 7 + 21 + 0 + 3 5 - 2 1
E x a m p l e 4.
+ 14-3
Let j = 3 and k = 7,
n=j+(k-i)
T.
n=3
= 55 = u 10
Calculate
n=3+7-i=9
n=3
Oct. 1967
CALCULATION O F FIBONACCI NUMBERS

AND SUMS FROM THE BINOMIAL
7 x 6
V * U,1 - 77nu 4 4. 7 x 6 x 5 , 7 x 6 x 5 x 4
E
n3+-2-^2+-3^y-u1+
4x3x2
7x6x5x4x3x2
6x5x4x3x2
U 2
7x6x5x4x3
5x4x3x2
0+
7[
7!
279
~3
= 7(2) + 21(1) + 35(1) + 35(0) + 21(1) + 7(-l) + 1(2)

'= 14 + 21 + 35 + 0 + 21 - 7 + 2 = 86
= u3 + u4 + u 5 + u6 + u7 + u 8 + u 9
= 2 + 3 + 5 + 8 + 13 + 21 + 34 = 86
NB.
The n u m b e r s in p a r e n t h e s e s in E x a m p l e s 3 and 4 a r e the n u m e r i c a l v a l u e s
a p p r o p r i a t e for u._,
with signs a s provided above,
It i s a g r e e d that the above equations do not p r o v i d e the l e a s t l a b o r i o u s

way of calculating u
or S u
but they do show that t h e r e i s a r e l a t i o n between
the Fibonacci s e r i e s and the Binomial.
REFERENCES
1.
D e p a r t m e n t of Defense Handbook H1Q9, Statistical P r o c e d u r e s for D e t e r mining "Validity of Suppliers 1 A t t r i b u t e s Inspection, 6 May I960,
2.
Dodge and Romig, Sampling Inspection T a b l e s , Second Edition, 1959, John

Wiley and Sons, Inc 0
3.
G. C h r y s t a l ,
Textbook of A l g e b r a ,
Second Edition,
Chelsea Publishing
House, Vol. II, p. 398 e

The Fibonacci A s s o c i a t i o n invites Educational Institutions to apply for a c a d e m i c

M e m b e r s h i p in the A s s o c i a t i o n .
The m i n i m u m s u b s c r i p t i o n fee i s $25 annually.
(Academic M e m b e r s will r e c e i v e two copies of each i s s u e and will have t h e i r

n a m e s l i s t e d in the J o u r n a l . )
* *
EXISTENCE OF ARBITRARILY LONG SEQUENCES OF CONSECUTIVE MEMBERS

IN ARITHMETIC PROGRESSIONS DIVISIBLE BY ARBITRARILY MANY DIFFERENT PRIMES
DOV JARDEN
Hebrew University, Jerusalem, Israel
It is well known that there exist arbitrarily long sequences of consecutive

positive integers that are all composite, e. g. , (n + 1)! + 2, (n + 1)! + 3, ,
(n + 1)! + (n + 1). This statement can also be formulated thus: for any given
positive integer n there exist n consecutive composite positive integers
each of which has at least one prime divisor.
The following is a twofold gen-
eralization of the last statement.

Theorem.
(1)
In any infinite arithmetic progression
ax + b,
a, b integers,
a ^ 0,
x = 1,2,3,-
and for any two positive integers, n, v, there exist n consecutive members
each of which is divisible by at least v different primes.
Proof.
(By induction on v ). Since a ^ 0, we have a < 1 or a 1.
We may suppose, without loss of generality,
a ^ 1, since if a < 1 we can
consider the progression -ax - b, the members of which have the same absolute values as the corresponding members of (1).
Thus for
x > (1 - b)/a,
(1) is an increasing sequence of positive integers >1. Since any integer > l i s divisibleby at least one prime, our statement is valid for v = i . From the validity of the statement for v we shall prove its validity for
fact, let 2 < aA < a2 < < a
v+ 1. As a matter of
be n consecutive members of (1) each of which
is divisible by at least v different primes. Consider the sequence of n consecutive positive integers (a )l 2 a+a l9 (a )\ 2 a+a 2 ,*",(a )\ 2a+a . For 2 L &i L a, 4 a
we have
2
(a n )! a + a k = a k
r ,_
^2,
(a n )! 2a
a
|"(2 . 3 4 - a k _i . a k . a k + 1 . . . a n )(a n )! a
W
J
= a k [(2- 2. 4 . . . a k - i - a ^ - . . a n )(2- 3- 4- - a k - . - a n )a-
The sum in brackets is composed of two terms, one divisible by a k , the other
being 1. Thus, this sum is coprime with a, , and since it is greater than 1,
Hence (a n )! 2 a + a k is divisible by
it is divisible by a prime not dividing a,.
v + 1 different primes, for any 1 < k < n. On the other hand, since a,
is
a member of (1), thus of the form ax + b, we have (a n )! a + a k = b (mod a),

thus (a n )! 2 a + a k is a member of (1), which completes the proof of the theorem.

280
A PROPERTY OF LINEAR RECURSION RELATIONS

RAYMOND E. WHITNEY
Lock Haven State College, Lock Haven, Pennsylvania
If one selects the basic Fibonacci recursion relation,

(1)
U ^
- U n + 1 - U n = 0; (n ^ 0) U0 = 0, U t = 1 ,
and applies the well-known series transformation f l ] ,
y(t) =XI u k t k / k ! '
(2)
one obtains a linear differential equation with a characteristic equation,

V = n2 - n - 1 = 0 .
n
(3)
v
'
It is easily verified that the recursion relation satisfied by |V I is
(4)
v
'
^ - 2 V
n+2
n+l
+ V = 2 .
n
It seemed reasonable to consider the relationship between (1) and (4). When
the relationship was investigated, a rather unusual result was obtained.
characterization of recursion relations of polynomials was also obtained from

the results. To carry out the investigation, it was expedient to introduce some
terminology.
Let a linear recursion relation of order p with constant coefficients be
denoted by
n+p
Li(U) = \ ^ a.
U. = b for all n ^ 1; a
i=n
281
= 1 .
282
[Oct.
When we apply the series transform (2) to the above, we obtain
V Y y ( l ) = be
The characteristic equation of this differential equation is,
] C ainl
0 .
Let
P
Vn = ^ a . n 1 ;
o
Define
(n > 1) V0 = a0 .
n+p
**<*> = E
i-n V i = C; d;p = 1 ,
i=n
as the conjugate recursion relation of LÛ).
determined explicitly.
Since d
The dTs and C will shortly be
= 1, it may be shown, as follows, that for
fixed order p, L2(V) is unique.

Clearly for any particular recurrence relation,
unique. Suppose L2(V) were not unique.
LÛ) = b,
Then
n+p
(5)
H(V) = J2 d i-n V i
= C
'
i=n
and
n+p
(6)
VUV) = Y ^J dlf.f n V. = C,f; d! = dn = 1

/
~ i
P
P
IV 1 is
1967]
283
are two distinct normalized recurrence relations for j V L For any fixed p 5
the series transformed differential equations for (5) and (6)s effected by (2),
th
would be distinct p order linear differential equations with identical boundary
conditions and the same solution^
y(t) = J] v i t i / i !
Note that the series for y(t) converges for all t,
since
is dominated
by
x^\
a.i /
Max
i=o? , p
Vn = 0(n P ) e
By linearity properties of the solutions of linear differential equations
and thus
with constant coefficients^ this is impossible.
Hence (5) and (6) are Identical,
Thus L2(V) is the normalized recursion relation of order p,
satisfied
by JV 1. We shall say that LÛ) i s self-conjugate If and only if

Li(U) - L2(U) .
Before we state and prove the central theorem^ we shall need a lemma and two
preliminary theorems,
Lemma:
X>j(n+p-j)p-k|
3=0
0 if k = 1,2,.-. , p
p! if k = 0 .
The proof of the above is an elementary albeit tedious exercise in induction and
Is given as elementary problem^ E12535 in f2jB
284
A PROPERTY OF LINEAR RECURSION RELAHONS
[Oct.
Theorem 1:
P
v == 1> a.n ; a
o
implies
P
t'-O^--'
J = '
3=o
Proof. If we use the polynomial expression for V , in J, the coefficient of a , in J is

p-k
j=
By the lemma,
J = p!a
= p!
QED.
Theorem 2:
Li(U) = b; a
= 1
implies
P
P-'O
L 2 (V) = ; ( ( - D J ( J v n + p . . = P :
j=0
Proof.
The characteristic equation of the series transform of Lj(U) is
1967]
PROPERTY OF LINEAR RECURSION RELATIONS
, 285
XN n i =
o
Thus
P
o
and by Theorem 1, the result follows0
Theorem 3: If LÛ) is of order p, then a necessary and sufficient
condition that LÛ) be self-conjugate is that
LI(U) =
( - D J ( P ) v n ^ = P!
^3 '
j=0
The proof follows from Theorem 2 and the uniqueness of L2(V).

In the light of the above we have
Corollary; Every polynomial of degree p has the same recursion relation and a recursion relation of the type given in Theorem 3 yields a polynomial
expression in closed form,,
If we choose p = 2,
a 1 = a0 = - l ,
b = 0 we obtain (1) above.
The
conjugate of (1) is (4)a Thus the Fibonacci relation is not self-conjugate,,

It would be interesting to see if there are other classes of functions which
yield a fixed recursion relation for all members of some subclasses of the
class of functions,,
Since the types of solutions of linear recursion formulae
with constant coefficients are quite restricted,
one would have to consider
more general relations to obtain results of much consequence.
REFERENCES
1. James A. Jeske, "Linear Recurrence Relations P a r t I,"
Quarterly, Vol. 1, No. 2, April, 1963, pp. 69-74.
Fibonacci
2. American Math. Monthly, Vol. 64, No. 8, October, 1957, p. 594.

*
SIMULTANEOUS PRIME AND COMPOSITE MEMBERS

IN TWO ARITHMETIC PROGRESSIONS
D O V a n d MOSHE JARDEN
Hebrew University, Jerusalem, Israel
Theorem, Any one of two non-identical infinite reduced arithmetic

progressions has an infinitude of prime members the corresponding members
of which in the other arithmetic progression are composite.
Proof.
Be
(1)
ax + b,
(a,b) = 1,
x = 1,2,3,...
. j
/ JX
-i
i o o
a ^ c or b ^ d
(2)
ex + d,
(c,d) = 1,
x = 1,2,3,-**
two non-identical infinite arithmetic progressions. We may suppose, without
/ON
loss of generality,
a ~ 1, c 1, since if, say,
a ^ 1,
we can consider
the progression -ax - b, the members of which have the same absolute values as the corresponding members of (1). Suppose, contrary to the assertion
of the theorem, that one of the progressions, say (1), has only a finite number
of prime members the corresponding members of which in the other progression are composite.
Thus, there is a positive integer N such that N >) d J
in case a = c,
and
in case a ^ e,
and such that, for any positive integer x ^ N, ex + d is a
prime if ax + b is a prime. ByDirichlet, (1) has an infinitude of prime members.
Hence, there is a positive integer x0
>
N such that ax0 + b is a prime,
whence, by the assumption, also cx0 + d is a prime. If a = c, then b d,

and ax + b ^ ex + d for any x. If a ^ c, then ax + b - ex + d only for
x = (d - b)/a - c). Since x0
>
N ^~ (d - b)/(a - c) we have ax0 + b ^ cx0 + d.
Thus the arithmetic progression a(cx0 + d)x + (ax0 + b),

reduced.
+ d)xj + (ax0 + b) is a prime.

x0
>
x = 1, 2, 3, ,
is
Hence, by Dirichlet, there is a positive integer xt such that a(cx0

Now put x2 = (cx0 + d)x1 + x0.
N Îd I, x 1 ^ 1, we have x2
>
x0 > N.
Since c ^ 1,
Thus ax2 + b = a(cx0 + d)xt
(ax0 + b) is a prime with x2 ^ N, while cx2 + d = c(cx0 + d)x1 + (cx0 + d) =

(cx0 + d)(cx1 + 1) is evidently composite with x2 > N,
being a prime, and cxt + 1,
since both cx0 + d,
with c ^ 1, xt ^ 1, are integers > 1 .
contradiction to the assumption thus obtained proves the theorem.

286
The
ELEMENTARY PROBLEMS AND SOLUTIONS

Edited By
A . P. HILLMAN
University of New Mexico, Albuquerque, New Mexico
Send all communications regarding Elementary Problems and Solutions

to Professor A. P. Hillman, Department of Mathematics and Statistics, University of New Mexico, Albuquerque, New Mexico, 87106o
Each problem or
solution should be submitted in legible form, preferably typed in double spacing,

on a separate sheet or sheets in the format used below.
Solutions should be
received within three months of the publication date,,

B-118
Proposed by J . L. Brown, J r . , Pennsylvania State U n i v e r s i t y ,

State C o l l e g e , Pa.
Let Fj = 1 = F 2 and F n + 2 = F n + 1 + F
for n > 1. Show for all n >
1 that
E (FKk /2 k ) < 2 .
k=i
B-119
Proposed by J i m W o o l u m , C l a y t o n V a l l e y High S c h o o l , C o n c o r d , C a l i f .
What is the area of an equilateral trapezoid whose bases are F
and
F , and whose lateral side is F ?

n+i
n
B-120
Proposed by Phil M a n a , University o f N e w M e x i c o , A l b u q u e r q u e , N .
Mex.
Find a simple function g such that g(n) is an integer when n is an

integer and g(m + n) - g(m) - g(n) = mn,
B-121
Proposed by Phil M a n a , University o f N e w M e x i c o , A l b u q u e r q u e , N . M e x .
Let n, q, d, and r be integers with n > 0, d > 0, n = qd + r,

0 < r < d. Prove that
F n = ( F d + 1 ) q F r (mod F d ) .
287
and
288
[Oct.
B-122 Proposed by A . J . Mont leaf, Univ. of New M e x . , Albuquerque, N . M e x ,

Show that
sin |(2k + 1)0 / s i n 6 = 2 cos
2k0
+ 2 cos
2(k - 1 ) 0 + 2 cos |2(k - 2)0 |
+ + 2 cos
20
+ 1
and obtain the analogous

f o r m u l a for F / r t l , x / F
in t e r m s of Lucas n u m b e r s ,
&
(2k+i)m/ m
B-123
(From B - 1 0 2 , Proposed by G . L. Alexanders on, U n i v . of Santa C l a r a ,

Santa C l a r a , C a l i f o r n i a .
x 2 + (x l ) 2
Show that all the p o s i t i v e i n t e g r a l solutions of
= z2 a r e
given by
= (p
n+l
where P
)2 _ (p )2 .
n '
= (p
n+i
)2 + (p )2
n '
is the P e l l n u m b e r defined by P i = 1,
J
= !
2,
'
P? = 2,
L
and P , = 2 P (J
n+2
n+i
+ P .
n
SOLUTIONS
A NON-HOMOGENEOUS D I F F E R E N C E EQUATION
B-100
Proposed by J . A . H . Hunter, Toronto, C a n a d a .
Let u , = u , + u - 1,
n+2
n+l
n
solution for u .
n
with ui = 1 and u?L = 3.

v
Solution by F. D . Parker, S t . Lawrence U n i v e r s i t y , C a n t o n ,
Find the &g e n e r a l
N.Y.
n
n
T h e g e n e r a l solution of t h e difference equation i s u = c i a + c 2 b + 1,
w h e r e cj and c 2 a r e a r b i t r a r y c o n s t a n t s , a = (1 + V ) / 2 and b = (1 - V 5 ) /
2.
Since uj = 1 and u 2 = 3, we find t h e p a r t i c u l a r solution to b e
= a11"1 - A b*-i + i =
V5
V5
2F
n A
+ i .
Also solved by L. C a r l i t z i ; Herta T . F r e i t a g ; W i l l i a m T . Jackson; Douglas L i n d ;

W i l l i a m C . Lombard; C . B . A . Peck; L t . A . G . Shannon, R . A . N ; David Z e i t l i n ;
and the proposer.
1967]
289
A SEQUENCE OF SEQUENCES
B-101 Proposed by Thomas P. Dence, Bowling Green State Univ., Bowling Green,
Ohio.
Let x.
b e defined by
x,
= 1, x
= n,
J
i,n
i,n
2,n
E x p r e s s x.
as a function of F
and n0
^
i,n
n
and x. = x.
+ x. .
i+2, n
i+l,n
i,n
Solution by Douglas Lind, University of Virginia, Charlottesville, Virginia

We c l a i m x.
ljn
all n.
= F . + (n - 1)F.
li
. This i s c l e a r l y t r u e for
i = 1, 2 and
Since both e x p r e s s i o n s obey the s a m e s e c o n d - o r d e r r e c u r r e n c e r e l a -
tion in i and a g r e e in the f i r s t two v a l u e s , they m u s t coincide for all i and

n.
Also solved by Gerald Edgar, Herta T. Freitag, William C. Lombard, John W.
Milson, F. D. Parker, David Z e i t l i n , and the proposer.
NOTE:
The p r o b l e m editor m i s s t a t e d t h e p r o b l e m a s
E x p r e s s x.
in
and n
i n s t e a d of
E x p r e s s x.
in t e r m s of n and F . . M
n
^
i,n
I
T h e p r o p o s e r intended t h a t F .
in the solution p r i n t e d above be e x p r e s s e d
t e r m s of
in t e r m s of F . ,
as one might do, for e x a m p l e , u s i n g t h e r e s u l t of B - 4 2 .

PELL-PYTHAGOREAN TRIPLES
B-102 Proposed by Gerald L. Alexanderson, Univ. of Santa Clara, Santa Clara,

Calif.
T h e P e l l sequence 1, 2, 5 , 1 2 , 29, is defined by P i = 1,
P 2 = 2 and
P , = 2 P , + P . Let (P ^ + i P ) 2 = x + iy
, with x and Jy r e a l and
J
n+2
n+i
n
n+i
n
n
n
n
n
l e t z = x + iy
1. P r o v e that t h e n u m b e r s x ., Jy , and z a r e t h e lengths
J
to
n
I n
n|
n n
n
of the s i d e s of a right
t
r
i
a
n
g
l
e
and
that
x
and
a
r
e
consecutive
i
n
t
e
Jy
&g e r s
&
&
n
n
for every positive i n t e g e r n e
of x 2 + (x I) 2 - z 2 than
A r e t h e r e any o t h e r p o s i t i v e i n t e g r a l solutions
(x, z) = (x , z ) ?
Solution by Herta T. Freitag, Hollins, Virginia*

(A)
\
= I x + iy
= Vxn 2 +J y
; h e n c e x , J y , and z
J
n
n
I n
n|
'
n n
n
t e d a s lengths of the s i d e s of a right t r i a n g l e
(B)
N
To show that Jy - x = 1:
n
n
Since x - P 2 , - P 2 and y
n
J
n
n+l
n
may
b e interpre
J
^
= 2 P . P , we need to show that

n+l n
290
[Oct.
J2P V - p2
+ p2 I = 1 .
I n+l n
n+l
n|
Proof by mathematical induction:
(1) 2P 2 Pi - v\ + v\
= 1, hence the statement is correct for n = 1.
(2) Assume the formula correct for n = k, i. e. , assume that:

,,P fk - P*k+i + P,k2 II = 1
I 2P,k+i
Then,
| 2 P k + 2 P k + i " P k + 2 + P k + 1 | = | 2 <2P k + 1 + P k ) P k + 1 - <2P k+1 + P k ) 2 + P k + 1 |
= l p k +1 -
2p p
k k + l - p kl =
This, however, means that correctness of the statement for n = k causes

its correctness for n = k + 1, and the query is settled.
(C)
No, there are no other positive integral solutions of x2 + (x l) 2 = z2

than (x, z) = (x , z ). This, however, is only a hunch; I was unable to
establish the unicifTT
Also solved by the proposer.
NOTE: See proposed problem B-123.

AN INCREASING SEQUENCE
B-103
Proposed by Douglas f u n d , U n i v . o f V i r g i n i a , C h a r l o t t e s v i l l e , V a .
Let
a = ] C F, (n ^ 1),
n
~
din
where the sum is over all divisors d of n. Prove that j a I is a strictly

increasing sequence. Also show that
00
F x11
n
n
v^
=
>
a
x
*-' n
n=i 1- - x n
n=l
1967]
291
Solution by Gerald Edgar, Boulder, Colorado.

F o r n ^ 1,
we have
Afr^SJlX
n+1
d ~
n+l
n+l
d (n+l)
Observe that
ai = 1 = F 2 ,
a2 = 2 = F 3 ,
a3 and t h a t for n > 3,
3 = F4 ,
s i n c e (n - 1) | n and (n - 2) | n,
n-3
= Z
F , 4 F +. Y
F. = F + F
- 1 < F ^
n
Ti
d
n
f-j
l
n
n-l
n+l
din
i=l
so that in all c a s e s for n y 1,

Therefore,
we have a
< F tj .
n
n+l
a ^ a , , so that \ a \ is a s t r i c t l y in(
J
n
n+l
n>
for all n > 1,
c r e a s i n g sequence,,
Also, we have
2 anxn = / E FA xn
n=i
n=i
\d|n
00
/ 00
zL
d=i
I ]C
Idln
00
/i
;(s-")*
d=:
KIT?)'oo
d
F ,x
d=i 1 - x a
(rearranging terms)
292
[Oct.
Also solved by the proposer.

TELESCOPING SERIES
B-104
Proposed by H . H . Ferns, V i c t o r i a , British C o l u m b i a .
Show that
oo
F
v^
2n+l
^-< L L
L ^
n=i
n n + i n+2
where F
and L a r e the n
n
n
respectively.
1
3
Fibonacci and n
Lucas numbers,
Solution by L. Carlitz, Duke University, Durham, N . C .

It is easily verified that
F
L
- F
L = F
n+i n+2
n+2 n
2n+l
Thus
N
n^l
F^,,
2n+i
L
nLn+lLn+2
Nf
A
/
f
F t J
n+l
L
nLn+l
F._,
n+2
L
n+l L n+2 I
F 22
F,
N+2
L L
i 2
N+i L N+2
and t h e r e f o r e
oo
2n+i
L L
L ^
n=i
n n+l n+2
1.
3
Also solved by Douglas L i n d , F. D . Parker, L t . A . G . Shannon, David Z e i t l i n ,

and the proposer.
A PERIODIC SEQUENCE
B-105 Proposed by Phil Mana, University of New M e x . , Albuquerque, New Mex.
Let g
be the n u m b e r of finite s e q u e n c e s c1} c 2 , o , o , c n ,

c
each c. in | o , l } , ( ^
i+1)
n e v e r (0,0),
and (c., c. + 1 ? c . + 2 )
with ci = 1,
never
(0,1,0).
1967]
Prove that for every integer s > 1 there is an integer t with t < s 3 - 3 and
g an integral multiple of s.
Solution by Douglas L i n d , University o f V i r g i n i a , C h a r l o t t e s v i l l e /
Va.
Acceptable sequences of length n can be produced by appending a

to all sequences of length n - 1, and a
M n
110" to those of length n - 3 Then
all n-sequences not included are not acceptable since they violate the given
restraints.
I
It follows that g^ = g ^ + g ^ .
determines the entire sequence
Put I k = fek> gk+1 g k + 2 )-
Each
by using the above recurrence relation,.
Thus modulo s > 1, if I. = I , then j g
is periodic with period S j - k .
Now there are (s - l) 3 possible distinct triplets (a,b, c) modulo s such that
a, b, c $ 0 (mod s).
Also (s - I) 3 < s 3 - 5 for
s > 1.
Thus either one of
li, I 2 , ' " , I s _ 5 contains a 0, in which case there is a t < s 3 - 3 such that g
= 0 (mod s), or I. = I
(mod s) for some j , k < s3 - 3 with j 4= kB
But
then | g 1 has period t = j - k > 0, so g, = g0 = 0 (mod s), where here

t < s 3 - 3,
Also solved by Robert L. M e r c e r and the proposer
* *
All subscription correspondence should be addressed to Brother U0 Alfred,

St. Mary f s College, Calif. All checks ($4o00 per year) should be made out to
the Fibonacci Association or the Fibonacci Quarterly.
Manuscripts intended
for publication in the Quarterly should be sent to Verner E9 Hoggatt9 Jr. ,

Mathematics Department, San Jose State College, San Jose, Calif. All manuscripts should be typed, double-spaced. Drawings shouldbe made the same size
as they will appear in the Quarterly, and should be done in India ink on either
vellum or bond paper. Authors should keep a copy of the manuscripts sent to
the editors.
*

5 3

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

5 3

Uploaded by

Copyright:

Available Formats

1967 AS THE SUM OF SQUARES

BROTHER ALFRED BROUSSEAU, St. Mary's College, California

With t h e coming of a new y e a r , t h e r e i s always a tendency to find out

But t h e r e is a t h e o r e m that s t a t e s that any

i n t e g e r can b e r e p r e s e n t e d a s t h e s u m of at m o s t four s q u a r e s . Let u s i n v e s t i gate t h e m i n i m u m n u m b e r of s q u a r e s that will add up to 1967.

But t h e s u m of two such

s q u a r e s on being divided by 4 would give a r e m a i n d e r of one, while 1967 on

We can p r o c e e d by table as follows:

We can stop at 31 s i n c e this b r i n g s u s to the halfway point with the l a r g e s t

Since no s q u a r e s a p p e a r in the t a b l e , this d i s p o s e s of the p o s -

Similar c o n s i d e r a t i o n s apply for 9, 9, 9. T h u s , the

only possibility i s four s q u a r e s .

ANOTHER GENERALIZED FIBONACCI SEQUENCE

Recent issues of numerous periodicals have given indication of a renewed

A combination of these two techniques

will be employed here.

In this paper we shall extend this by considering sequences in which

any given term is the sum of the three preceding it.

ANOTHER GENERALIZED FIBONACCI SEQUENCE

THE GENERALIZED SEQUENCE { P

Specifically we c o n s i d e r the sequence

| P n | = P0, Pi, P2, P n ,

w h e r e P 0 , P i , P 2 a r e given, a r b i t r a r y a l g e b r a i c i n t e g e r s , not all z e r o , and

It will a l s o be convenient t o c o n s i d e r a companion sequence, so to speak,

= R 0 , Ri, R 2 . . Rn> "

and for n > 2,

F r o m (5) and (3), when n > 5,

R n = P n - l + P n-2 = (P n-2 + P n-3J + ( P n-4 + P n-30 ) + (P n-4 + P n-5J

Using (5) and (3) further, we have

Hence for n > 3,

ANOTHER GENERALIZED FIBONACCI SEQUENCE

Thus | R | is actually t h e s p e c i a l c a s e of (2) in which R 0 = P i - P 0 ,

evident in the development of j P

usefulness of the sequence j R

Two other s p e c i a l c a s e s of (2) should be mentioned at t h i s t i m e ; n a m e l y

spectively, to give the s e q u e n c e s

0, 1, 1, 2, 4, 7, 13, 24, 44, , K n ,

Hence we might call J K

and for n > 2,

| w a s defined and d i s c u s s e d briefly b y M. Agronomoff

T h e r e i s only one b a s i c identity h e r e b e c a u s e the l a t t e r two a r e

s p e c i a l c a s e s of the f i r s t one upon setting p = n and p = n - 1 r e s p e c t i v e l y .

p r o p e r t i e s would b e much m o r e difficult to find due to the m o r e complex n a t u r e

ANOTHER GENERALIZED FIBONACCI SEQUENCE

We t u r n now to an investigation of the sequence (2) and c o n s i d e r ,

o t h e r f a c t s , how (10), (11), and (12) o c c u r a s s p e c i a l c a s e s of m o r e g e n e r a l

I satisfies a difference equation.

w h e r e xj,, x 2 , x 3 a r e the t h r e e distinct r o o t s of the equation

B j , B 2 , B 3 a r e such a s to m a k e (12) too

s o m e to be of any f u r t h e r p r a c t i c a l u s e in the succeeding development and hence

may be found a s follows: In the notation of v e c t o r s and m a t r i c e s , we have by

ANOTHER GENERALIZED FIBONACCI SEQUENCE

and by finite induction

F u r t h e r , a s i m p l e induction proof gives

- K n-i L n-i K n-J

so it might b e said t h a t j K I and i L | a r i s e " n a t u r a l l y " in t h e i n v e s t i g a tion of

Using (14) and (15), we find for n,

of (17), (18), and (19), r e s p e c t i v e l y ,

we have (10), (11), (12) as s p e c i a l c a s e s

ANOTHER GENERALIZED FIBONACCI SEQUENCE

k p o s i t i v e i n t e g e r s , we can show that

Using (20) and (21), we have the following useful e x p r e s s i o n :

It can be shown quite e a s i l y that t h e s e q u e n c e