Professional Documents
Culture Documents
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
1342
1341
1261
1221
981
901
501
381
1742
1741
1661
1621
1381
1301
901
781
1942
1941
1861
1821
1581
1501
1101
981
sibility 1 , 5 , 1 as endings.
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
1.
INTRODUCTION
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.
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
[Oct.
(2)
= P
n
+ P
n-l
+ P
n-2
. n > 3.
n-3'
(4)
{Rj
where
R0 = P i - P 0 ,
(5)
Ri = P 2 - P i .
= P
n
, + P 0
n-l
n-2
we have
+ R
n-l
+ R
n-2
n-3
R4
R3 + R2 + Ri
R3
R2 + Rj +
RQ
(6)
= R
n
+ R
n-l
n-2
+ R 0 .
n-3
1967]
211
2 - p i
+ p
T n e
i will be
I that follows.
P j = P 2 = 1 and P 0 = 1,
~Pt = 0,
P2 = 1 r e -
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
} a
+ K
n-l
P -type
' , L ,
'.
Lj = K 2 - K^,
n-2
sequence and
JL
| an
-type
sequence,,
The sequence JK
Til.
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
evidently
such
212
[Oct.
among
since J P
(13)
Pn = B ^
Thus
+ B 2 x* + B 3 x^
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,
cumber-
("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
rp
n
(14)
"P2]
1 0 0
n-i
Pl
. oJ
0 1 0
f n
10
(15)
K , _, L . . K
n+1 n+l n
1
n-l
p 1 q.
{Pj.
"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,
+ K P
n n-2
214
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,
n+h+k =
+ K ^ P
p+r n - r - 2
h +k+l n - i
n+k n - l
"n+h+k
(22)
n+h
h+i
h+i
1 0
0 J L
Pi
i _
.po
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
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
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
ro i o*
i
-n
(29)
-n+i
L -n+2-
ii
0 0 1
I I
-1 -1
"Po"
Pi
.P2.
1- n f P o l
0 0
Pi
1 0
.PJ
(30)
P -n = P -n+3
. ^
J J 0 - P-n+2
J . - P J-n+1
n >0
are
216
R
[Oct.
= P
n
+ P
n-l
n-2
p , r , h,
k above.
LINEAR SUMS
| and j P
Since an exhaustive
No
(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
(330
E Lsi = K3n .
i=i
and
1967]
217
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
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
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),
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
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^
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
F i r s t , w e s t a t e the following l e m m a
of A multiplied by x X y; i e. ,
(Ax) X (Ay) -
(cofactor A) (x X y).
1967]
rp
1 0
-3J
P
n-2-
"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
PlP2
J
P 2 ( P ! - P3P1) .
E x a m p l e : Suppose we let P 0 = 0,
becomes
Po - 2Po + 2 .
Pl3
P2s
P
P2,
we see that t h e r e
P
P
+ 2P P
P
n+l n - i n-3
n n - i n-2
\,
220
[Oct.
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 -
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 ^UjLl ^ + 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
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'
nKn+k-i "
h+kKh-i "
hLh+k+i "
h+kLh+i *
'
so
1967]
221
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 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'
MISCELLANEOUS RESULTS
(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
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
D(P 0 , P l f P 2 , - , P n )
= 0
REFERENCE
1.
M. Agronomoff,
p. 126.
RESTRICTED COMPOSITIONS
S. G . M O H A N T Y
McMaster University, Hamilton, Ontario, Canada
(tl9t2, , t^.+1) of an
integer n (i e. ,
k+i
E t.
= n
and
t. > 1
1
i=i
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
RESTRICTED COMPOSITIONS
224
'
[Oct.
^'JL\-
i=i
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
Trivially,
N(n)
(1)
N(h;a,jL,_V^il) =
(2)
k- 1
/a + k -
l\
and
N(n;a 1 ,a 2 ,-"- ,&k) - 0
(3)
'
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
. . . , A y . Therefore,
1967]
(4)
RESTRICTED COMPOSITIONS
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
!()(-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.
RESTRICTED COMPOSITIONS
Theorem 1:
(6)
NCri^a^.-^ak)
/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
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,---,^)
1967
RESTRICTED COMPOSITIONS
-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
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
+ 1, a.+2, , a k ).
[Oct.
RESTRICTED COMPOSITIONS
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 ) .
A2
Ak
-i
i-
x1=o x 2 =: x 1
xk^ 0
x =0
xk-i"1
A2
2=xi
1 .
k=0
- 1 i s negative a r e z e r o .
(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 )
/
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]
RESTRICTED COMPOSITIONS
229
(:)
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
RESTRICTED COMPOSITIONS
[Oct.
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,-
(17)
/ a +
(
k
kb\
)
(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
(18,
('%*/,'.-, ""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]
RESTRICTED COMPOSITIONS
Proof: F o r c + q(d - 1) ^ qb,
(a - 1) + (i - l)(b - 1),
i = 1, 2, '
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
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,
(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;
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
RESTRICTED COMPOSITIONS
and aj =
be verified.
af3
[Oct
Then (21) immediately
a.t = a3 = = a, = y,
a\
i=2
(22)
< 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]
RESTRICTED COMPOSITIONS
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
(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
-,,
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)
234
RESTRICTED COMPOSITIONS
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
then [ r x l 5 rx 2 , , r x k J ^ L ^ A j , A 2 , , A^ ) .
REFERENCES
1.
F . Gobel,
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
4.
Orthogonal
T h e B r a c k e t . Function, and C o m -
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
Biometrische
T. V. N a r a y a n a and G. E. Fulton,
1.
INTRODUCTION
Let k be a p o s i t i v e integer*
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) >
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)
and
k,
(F 0 = 0,
F j = 1) .
put m - pk + r
pk+i
nlm
p,pk p,pk+i
n
w h e r e n1A is e i t h e r
0 or
1 and
235
2m
pm
p+i, m
(1 < r < k)
and
236
(1.9)
i,]+i
[Oct.
'
13 ~
1+1,3 -
x x x x x x
x x x
x x x x
X
X X ,
1 1 1
1 ,
1 1 0
0 ,
1 0 0
0 ,
0 0 0
0 ,
0 .
T(k,n) = F m (k)
(m,k = 1,2,3, ) .
(LID
ao = - | V j
j=o
Dk - I ] ( P 3
Dn ,
N n (k)x n = k ^ - 1
n^o
x - (1 - x)
(1.12)
1967]
237
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
i, m-i
matrix.
anc
* 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)
T(k,m) = m + 1
(1 < m < k) .
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.
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
lm(m+i)
2-f
n=o
qi(n;m)x
Then
(k) ,
l ) - k
tA
238
k > 1,
(2.4) r e d u c e s to (1.5).
However,
[Oct
Chaundy
q, (n;p)
i s not
known.
30
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,
n t l n l j k + 1 n ljP k-M n l m
n
j i """
n
n
jm
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
F o r example,
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)
(3.4)
(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]
239
we s e e that
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)
(3.6)
N k ( m - k,k) =
(-D
r=o
r=o
r /
/
'
Indeed, it follows
Note that (1.7) and
(3.5) imply
Ni(m,2) = 3 2 m " 2
(3.7)
(m > 2) .
we make u s e of s o m e r e -
Let p denote a p r i m i t i v 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 ,
k-i
N k (n,k) = Z A V
- Dn .
240
[Oct.
We show that
(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
(3.10)
Nk(n,k) = \
[ ( p " j + l ) k - l ] (p j - l ) n ,
so that
(3.1D
Nk(n)k)^E(^
s=o \
(j).
/ r=s(mod k) \
(3.12)
F , ( x ) = f; N(n,k)x n
J
JJ, = 0
(1 < j < k)
(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)
(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.
t h e conditions (1Q9).
function (3.15).
48
(4.1)
n. A n . + 1 + k
(J = 1, 2, 3, )
= 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)
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)
S2(n) = F n
(n = 1 , 2 , 3 ,
F
(k)
nk+j
(4.6)
1 =
r=o Wi*>
1 < j < k,
S, .(n*)
-k)
denote the n u m b e r of
nr ^ nr+l
n
let
(14 j
>
r+1
(r * j (mod k) ) ,
<r = 3 (mod k) ) .
Sk.(l) = j
( 1 4 j 4 k) ,
(4.9)
kjj+1<
) = 1
kjW
(1 4 j 4 k - 1) ,
1967]
2 43
n-i
(4.10)
S kl (n) = 1 + E Sj.fr) .
r=i
S kj fr + 1) = F r t f j - 1 ( k )
(1 < j < k) .
(n-0k(k)
+ F
nk-i
which implies
(4.D
S ki (n
1) = F n k (k)
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)
nr > nr+1
(r j (mod k) ) ,
n r > k + n r + 1 (r = j (mod k) )
244
[Oct
(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.
i s given by
(4.18)
R k j (n)
= N.(n,k)
(1 < j < k) .
, (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
(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)
N k _.(n,k)
]=o
V /
r=o
k-i
\ /
= (1 + ( - l ) k + 1 ) N k ( n - k,r)
which implies (5.1).
246
Oct. 1967
(5.4)
N (n,k) =
[V
r=Q
so that
k-l
(5.5)
N <n,k) =
J
s=o
(n;A(lAj-k;
V / r=s+j (mod k) \
+r(n
+ km - r,k) = ^ \V
s=o L
= V f"(p"s +
s=o L
=
2 1
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.
2.
3.
(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
<*>
kb
-Z2>' Z Z f(iJ)
j=oi=o
248
[Oct.
[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
a+b+k
( * ) (-l)J (J>) (1 + ^
k=o
a
j=o
b
k=o j=o
W W
m
Y\C\
n^o
v y v
a b
lr=0 i=n
k=o
j=o
n=Q
I \ i \ i
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]
249
,k
?> M#)()
and we shall show in a simple way that this generalizes to give the relation
say
^ = Z ai ^ 5
i=o
then
<
0, n < r ,
8)
E
^ ^ k ) ^ "J( (_n
-1)%!
k=0
Since I ^
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
250
Oct. 1967
ki
H:)t: )
3=0
so that
n
k= o7:
, i
.Y
WW\ n /
\k/
differences of
do
N^cn) = y y
(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.
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
252
[Oct
= x . f n (x) + fn_lW
fi(x) = 1,
If z
(i)
f2(x) = x .
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
?j Vk
Fmi+K = F
1=1
where F
i s the n
\ nTkj nk A
Fibonacci number,
"
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
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-
\
rI/ n~- :kk +Il n
; o< k<^j-i
a)
(2)
/n - k \
^^x r : ~ l ,
n - k 1 k J
Summed over k,
0<k<|
(1) and (2) give the Fibonacci and
k\
/ m - s + 1 \ / m - (k - s) + l \
k s
254
[Oct,
2
F2
m+2
=
~
R?]
V
yk / m - s + l \ / m - ( k - s ) + 1\
s
k
s=o\
A
~S
/
feo
(>)
for
b > a > 0
(4)
k\
s=<
)\
k s
fisr-.-r-t-n
F
m+2 m+3
m - s\
s
/m - (k - s)\
J m - (k - s) \
k-s
1967]
2
L
H k
s=o
('m s
ni
k
255
/ m - (k - s ) \
k
s
- * "^V
"
/ '
m
L
L
m
^4
m+l
fco
s^o
"
/ m ~ s\
s
m + l
m
- * -
s)
/ m - (k - s) + l \
+
For example,
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 ^
Bull. A m e r . Math
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.
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
[Oct.
n - 0
^ (x) = xB v(x 2 ) ,
2n+2
n
and
f
2n+l
<x) = b (x 2 )
n
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 /
Show
2n+i / _
, i \
=h )
for all i n t e g e r s k.
2k+2j+l
2n+2k+2
5W
^n
1967]
257
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
2n+i-2j
S(2"-/)
>. i
r i F2J+1
3=l
so that
Sfr )
F2j+i
"
5n
Similarly, we have
n / \
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
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
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
[Oct
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
6 0 0 0 0 0 1 1
8 16
32
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
(6)
q2 - q - 1 = 0
ri
= Lz^l
* -0.6180
and
r 2 = ^ y ^ - * 1.6180 ,
n " f n-i
(9)
then
< s e e E4]*
b
iri
2r2
1967]
261
FIBONACCI NUMBERS
jr^j < \ 9
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
(ID
and we w r i t e
0 > N)
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
H-(^)]
GENERALIZATION
(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 .
(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.
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.
x (1
^ ^k-x) >
Kkl-<
(18)
k,k *
(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
% Vii
1=1
(21)
k-i
(m =
,k =
? V?--
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
(23)
M (ri^ - W
Wi
(24)
uj;k* b k 4 k
(j > N) .
(25)
J->00
J,k
(26)
lim ! & k = r m
APPROXIMATIONS
We first note that as k>oo the sequence u. , , approaches the geoJ-K,K
264
(27)
lim r k k
k^oo
'
[Oct.
= 2 .
for k = 2 to 19,
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
u
(28)
lim
lim
k->oo
j-9>oo
3
U
,
*' K = 2 ,
j
j k
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
fe
j and k we m a y w r i t e
ujjk* rk
(30)
(31)
By (17)
w. , =
j,k
1 1 . . - ul ,
j,k
jsk|
1 L i 4 k - 1,
i=i
266
Oct.
k-i
(32)
l i m w. .
J->oo J
= 0 ,
will b e p r e p a r e d by c o m p u t e r s .
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.
2.
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.
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
r2,
r3j
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)
c2,
c3s
Fibonacci
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)
268
[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
71
8c
131
l c 3 ~ oZC-jC2Co + 2 4 c j C ^ + 1 2 c ^ c | -" 8c 2 c|
powers
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 "
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]
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
[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.
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
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
has a s i m i l a r computation.
(2,n)
x,
n n
riI>2
n n
riI>3
n n
r2r3
'
,
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,
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.
<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
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;
i = 6,
c2 = 1 1 , c3 = 6
yields
x 3 - 98x 2 + 1393x - 6 4 = 0
1967]
with r o o t s
l4,
24,
271
1, 2, and 3.
(BJ By i t e r a t i o n : To get c.
(l,0)'
(l,l)'
i =
(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 ,
0,
c(lj0)
3,
c(ljl)
1,
c(l^2)
First,
272
[oct.
(l,3)
1(3)
1(1)
'"'
1(3)
(l,n)
'
C
(1 4)
(l,n-1)
(l,n-2)
+ C
Then,
'
(l,n-3)-
[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
with r o o t s
+ x - l
= 0
[ 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'
'
is odd
Lucas
1967]
L 2 = 3,
Ln = ~Ln-i + L n _ 2
273
;ri
= a
and letting r 3 v a r y .
(1 + N / 5 ) / 2 ,
r2
(1 -
N/5)/2
/ofX
( 3!)
c ^ g 09
r,
n
+
r2
[n/3]
\-^
n
+
r3 =
(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
Oct. 1967
+ 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
rt
= a,
r2
p,
r3=
- I/N/5" ,
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.
P o w e r s of the Roots of a
pp. 170-178.
2.
* *
1, 2, 3S 5, 8.
the
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,
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.
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
general
276
(4)
^ n
[Oct.
*o
13
21
34
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
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]
where &
is d
*>J
277
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 )
2 + 7(1) +
W
1M
2
(1) +
W
IM(5)
3(2)
lUj
7(6)(5)(4)
4(3) (2)
(1
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 .
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 ,..,
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
[oct.
Since d 0 ,
series,
u ,
the z e r o - o r d e r of difference,
d 0 -;_! = U 4 1 ?
k+]
etc e
provided ref-
when j - k i s negative*
Thus,
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
Negative when j - k
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
7!_
7!U~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
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
a p p r o p r i a t e for u._,
or S u
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.
3.
G. C h r y s t a l ,
Textbook of A l g e b r a ,
Second Edition,
Chelsea Publishing
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.
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
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
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
v + 1 different primes, for any 1 < k < n. On the other hand, since a,
is
U ^
- U n + 1 - U n = 0; (n ^ 0) U0 = 0, U t = 1 ,
(2)
(3)
v
'
(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.
U. = b for all n ^ 1; a
i=n
281
= 1 .
282
[Oct.
V Y y ( l ) = be
] 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 L^U).
determined explicitly.
Since d
L^U) = b,
Then
n+p
(5)
H(V) = J2 d i-n V i
= C
'
i=n
and
n+p
(6)
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 !
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
satisfied
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
[Oct.
Theorem 1:
P
v == 1> a.n ; a
o
implies
P
t'-O^--'
J = '
3=o
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.
1967]
, 285
XN n i =
o
Thus
P
o
and by Theorem 1, the result follows0
Theorem 3: If L^U) is of order p, then a necessary and sufficient
condition that L^U) be self-conjugate is that
LI(U) =
( - D J ( P ) v n ^ = P!
^3 '
j=0
a 1 = a0 = - l ,
The
REFERENCES
1. James A. Jeske, "Linear Recurrence Relations P a r t I,"
Quarterly, Vol. 1, No. 2, April, 1963, pp. 69-74.
Fibonacci
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,
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.
in case a = c,
and
in case a ^ e,
>
>
>
x = 1, 2, 3, ,
is
N ^Id I, x 1 ^ 1, we have x2
>
x0 > N.
Since c ^ 1,
286
The
Each problem or
Solutions should be
Let Fj = 1 = F 2 and F n + 2 = F n + 1 + F
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 .
and
Mex.
and
288
[Oct.
2k0
+ 2 cos
+ + 2 cos
20
+ 1
x 2 + (x l ) 2
= 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
Let u , = u , + u - 1,
n+2
n+l
n
solution for u .
n
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.
= a11"1 - A b*-i + i =
V5
V5
2F
n A
+ i .
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
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
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 . ,
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 ) ?
= 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
^
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\
= l p k +1 -
2p p
k k + l - p kl =
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
F x11
n
n
v^
=
>
a
x
*-' n
n=i 1- - x n
n=l
1967]
291
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
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>
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.
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,
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
A PERIODIC SEQUENCE
B-105 Proposed by Phil Mana, University of New M e x . , Albuquerque, New Mex.
Let g
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.
M n
all n-sequences not included are not acceptable since they violate the given
restraints.
I
It follows that g^ = g ^ + g ^ .
Each
Now there are (s - l) 3 possible distinct triplets (a,b, c) modulo s such that
a, b, c $ 0 (mod s).
s > 1.
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
But
* *
Manuscripts intended