Professional Documents
Culture Documents
THE
OFFICIAL
JOURNAL
OF
VOLUME 5
NUMBER 1
CONTENTS
P A R T I ADVANCED
ON SUMMATION FORMULAS AND IDENTITIES
FOR FIBONACCI NUMBERS
. . . . . . .
David Zeitlin
D. A. Lind
44
45
59
69
PART II ELEMENTARY
RELATIONS INVOLVING LATTICE PATHS AND CERTAIN
SEQUENCES O F INTEGERS . . . . . . . .
Douglas R. Stocks, Jr.
81
87
. . . Dewey C. Duncan
. . .
FINITE D I F F E R E N C E
.
Brother Alfred Brousseau
FEBRUARY
j). A. Lind
89
91
. G. F. Feernan
99
. Edited by A. P. Hillman
107
1967
DEVOTED TO THE
WITH SPECIAL
PROPERTIES
EDITORIAL
B r o t h e r U. Alfred
H. L. Alder
M a r j o r i e Bicknell
John L. Brown, J r .
L. C a r l i t z
H. Wc Gould
Ae P . Hillman
ASSOC I A TION
BOARD
V. E Hoggatt, J r .
Donald E. Knuth
Douglas A, Lind
C. T. Long
Leo M o s e r
I.' D. Ruggles
M. N. S. Swamy
D. E. T h o r o
C h a r l e s H. King
L. H. Lange
Douglas Lind
J a m e s Maxwell
S i s t e r M. de Sales McNabb
C. D. Olds
D W. Robinson
A z r i e l Rosenfeld
MQ N Sc Swamy
John E. Vinson
Lloyd Walker
C h a r l e s Rc Wall
1.
n-i
L1
E ^ k + r =!^\.r+L^m)
< >
k=o
+ C
* '
i=o
with H0 = p and
<"> E ^ r
=H
k=o
s=o
* Hn+r+i 22
s=o
J^MJ^-MFji+iGin-s
+ Co
1=0
( r , m = 0, ! , ) ,
where
m
(1.3) C2 - - H r ^ ( - l ) i ( i I ) F 2 i G 1 m - H r + 1 5]M) 1 (i!)F 2 i + 1 G 1 m
1-0
1=0
( r , m = 0, !,)
and G
(see [2]) a r e Stirling numbers of the second kind with the properties
that Gj = 0 if i ^ 0, G\ = 1, i = 0 5 1 5 0 8 0 *
if i > s.
1
ON SUMMATION FORMULAS
[Feb*
S^k+r ^ ( " l ) W r t ^
k=o
) + C2 ,
i=o
We note that if Agfa.) = Ah(n)9 then g(n) = hfa) + C2o Thus 9 using the A operator on both sides of (1.4), we obtain
(1.5)
m
1
n m H n + r = ^ ^ ^ ^ i + n + r t A ^A,n +, I)
-xm
1=0
HI
^(-l) 1 H 2 i+n+r+i^ 1 (n in X
- /
1=0
n Hn+r
=^(-l)%2J+n+r+2Aj+1(iim)
3=0
'i=o
(a
) = 0, the right-hand
(1.7)
A^fa) = ( - D ^ t - l ^ f 1 Jg(D + k)
k=o
'
(i - 0,1, )
1967]
AV)
(1.8)
= { - l j ^ t - i ^ m & i + k) m
k=o
i
s=o
Z(T)s<-T(-k(Ot:
m
H1~S
s=o
s=o
#
k=o
m
HI
LmmJ
S=G
(1.9)
(-l)1^
=X)(-Dk(k)k:
(i =
0,l,-,m)
k=o
Buschman [3, p. 6, (12)] showed that
H2i+n+r+i
2i H n+r + F 2 i+iH n + r +i
(ia2)
n-i
k=o> % =
+
i=o
\ x{-i)iF*A%m)
H
n + r + iE ( - 1 ) i F 2i + i A i ( n m )
i=o
[Feb,
ON SUMMATION FORMULAS
n + r
0
m '
(k)H
1)1(Xl)F
a
Z
j(sHI i=s+i
k + r = \+rJjJj4
s=o i=o
w^|2>*)'-!E'j(iHs=o
1=0
n S + C3
j=s+i
( r , m = 0 , 1 , ),
where
m
(1.14)
C3 =
-HrZ<-1)i(i:)HZa3Gj!
1=0
3=1
r+iZ(-1)i(i!)F2i+1EaJGS
i=o
(j=i
(r,m = 0, ! , )
1967]
5
= 1, then
(1.13)
occurs when
m
P(k) = k ( m ) = k(k - 1)* (k - m + 1) = Y V
/
kj
where (see [2, p. 142]) S-1 are Stirling numbers of the first kind. Then, since
, (m)
, /k\
,0 m
k
= m! I I , we have
n-i-i
^k(m)Hk+r
n-l
= HI! ] T P\
k=o
Hk+r
1, m
2,;-.)
k=m
<n = in
we have
V
a . G Sdf sm^ 3 = \ / - /.)=\l
JL*J 3 3
L
m
if
m
!
=
l
f
if i f
mj
(1.15)
It should be noted that C3 in (1.14) was obtained from (1.13) for n = 0. For
P(k) = k
.16,
f V > H k + r - ><%
k=o
[Feb.
ON SUMMATION FORMULAS
We obtain f r o m (1.16)
(1.17)
vUl/
(-1)
2m+r+i ~
2m~nHn+r
we obtain
2m+i-nHn+r+i
m
(1.19)
(-l) (m! ) F 2 m n =
/ m
Z(-1)I(L')r E( J B) S iM
s=o i=o
(j=s+i
(n = 0, ! , , m - 1; m = 1 , 2 , - - ) ,
m
(1.20)
(-l) (m!)F2m+1n
= ^
s=o
i=o
(n = 0 , l ,
a , -
,m-l;
m = 1?29*).
m /
(1.21)
\H1/
(-lrCm!
)F2m+k_n = ^ T W t t )F2i+k ^ \ J ^
i=o
s=o lj=s+i
( J s ) S i G jl s
^
'
(k = 0 , 1 , ; n = 0 , 1 , v , m - 1; m = 1, 2, ) . '
Proof of T h e o r e m 20
proof of T h e o r e m 1, t h a t
Noting that A
P(n) = 0, w e find, i m i t a t i n g t h e
1967]
n-i
P(k)H.
k+r
jLmd
m
J ( - D ^ j + n + r + i ^P(n) + C3
k=o
1=0
- "
JLil
i=o
1=0
= H
Since
.Hi.
>(n) =
Hi
n5
X/j '
Aip(n) =
3=0
y^^
j=o
]TVD^aiAW) = ^ ( - D ^ i ^ a A1^)
i=o
i=o
3=0
-E^-EvEOK""
-E^>'-EE^(H.k
1=0
j=0
i~0
S=0
S=0
* }=S
I JLmsJ
s=o
i=o
3 \^/
3"
1 i=s+i
since
m
ni
y^y^f(s 9 j) = X ^ i C f ^ ] ) a n d
j=o s=o
j - s = i f .i - s < i .
s=o j=s
J)
Indeed*
k=o
ON SUMMATION FORMULAS
[Feb.
V*gCn) = A*g(n - i)
(i - 0 , 1 , ) ;
(2.3)
= A*(n - i ) m
(i = 0 , l , - - - , m + l ) .
(2.4)
(2.5)
] T k m F 2 k = ^ ( - D ^ + M V ^ n 1 1 1 ) + C5 ,
k=o
i^o
k=o
(2o6)
i=o
^k^zk-i =^W^^iVV
k=o
1 1 1
) +C
i=o
(2.7)
kmH
k=o
X i (-^Hn-^+iV^n111) + cf .
i=o
The following results contain (2.4), (2.5), (2.6) and (2.7) as special cases.
The notation is consistent with Theorems 1 and 2.
Theorem 3. For n = 0, ! , ; r = 0, 1, 2, , we have
10
ON SUMMATION FORMULAS
[Feb.
vV - (.
s=o
Since
m
j=o
w e have
ra
1=0
i=Q
j=0 S=0
i=o
s=o
j=s
s=o
\ j=s+i
1 3 :F
a+i
F . = (-1) TF .
and
= y^C-D^S^kJ
^(-k)(m)H2k+r
k=o
(~l)m(m!)
_^ / k + m - 1
H
2k+r
1967]
11
and
m
(2.10)
L-J
3 3
Z^ m j
[0
if
mj
we obtain
(using (1.10))
xHl
(2.11)
(2.12)
y^P(k)H 2 k+r
k=o
"m
E ( - i ) i ( i ! ) F i- 2
H<2n+r / ^
s=o
i=i
m
H
2n+r+l /
m
J
>=o 1
Eaj(Js)Gj-
J=S+1
( l)ife)r
Ei=o
( m
Hl a i(!) G i-
n S + C7
\ j=s+i
i=s+i
(m = 0,1, ) ,
where
m
(2.13)
C7 = H
a0
( 1)iw)Fi 2 a G
-X]
"
- E(j=iJ i
i=i
H +
- r iX!"
1=0
m
i)i(ii)Fi i a G 1
"E 3 3
J=i
12
ON SUMMATION FORMULAS
i=o
Ij=s+L
m
[Feb.
!
|
p-^wj .,('.)<
i=l
*j=s+i
. (s = 0 , ! , * , m - 1) ;
(2.15)
/ m
i=o
' j=s+i
'
(s = 0, !,-, m) .
For a. = S_^, j = 0,1, - , m, (2.14) and (2.15) with s = 09 yield (noting (2.10)),
respectively,
(2.16)
( - l ^ ' V l )Fm2 = ^ ( - D ^ !
i=o
(2.17)
Xm = 1, 2f ) ;
j=i
m
m
)F
m
i
(-l) em!)Fm_1 = ^ ( - l ) ( i ! ) F i _ 1 ^ ( - l ) j S ^ i G J
i=o
(m = 0 , 1 , )
j=i
m
i
(-l) (m! ) F m _ s = ^ ( - l ) ( i ! ) F i + 1 ^ ( - l ) J S ^ G J
i=o
Since L R = F n + 1 + F
j=i
(m = 1,2, ) .
1967]
(2.19)
13
HI
<-l) (m2)L m 2 = 2 ( - l ) a ! ) L i ^ ( - l ) j S ^ G J
i=o
(m = l f 2, ) .
j=i
m
i
=^(-D (il)Fi.1^H)iS^GJ
i=o
j=i
(nU)Fm_i[-l + (-l) ]
(m = 1,2,---) o
2n
i
(-l) (i!)Fi,1^(-l)jS^G| - 0
i=o-
(n = 1, 2, ) .
j=i
2n
~ - G 2n
-
(2.22)
2n
2n-2
= ^
(-1)^1 )Fi_ t ^
i=o
j=i
(-D^GJ
(n = 1, 2,
Suppose now
m
p(k> H k ( m ) =^]sikj
3=1
14
[Feb.
ON SUMMATION FORMULAS
in (2.12).
C7 = ( - l ) m + 1 ( m ! ) ( F m _ 2 H r + F m ^ H r + i ) = ( - l ) m + 1 ( m ! ) H r + m _ i
(2.23)
If we r e w r i t e (2.12) as
n
(2.24)
yV
m )
2 k
= L*(m,n)H2n+r + L * ( m , n ) H 2 n + r + 1 + C 7
k=o
we obtain f r o m (2.24)
( - l ) m ( m ! ) H r + m 1 = Lf(m,n)H 2 n +r
(2.25)
L*(m, n ) H 2 n + r + 1
(2.26)
r+m-i
m-2-2nH 2 n+r
(n=0,lv,m-1)
we obtain
m-i-2nH2n+r+i
an
m
(2.27)
(-l) (m!)Fm_2_2n =
s=o
i=i
I j=s+i
(n = 0 , 1 , , m - 1; m = l , 2 , ' - # )
111
(2.28)
ni
(-l) (ml)Fm_^
2n
s=o
i=o
111
''j=s+i
(n = 0 , 1 , , m - 1; m = 1, 2, ) .
Proof of T h e o r e m 4
It i s r e a d i l y verified that
1967]
(2,29)
15
^ P ( k ) H 2 k + r = y ^ ( - D 1 H m + r - i + i A*P(n) + C7
k=o
i=o
m
P(n) + ^ (-D 1 F i _ 2 ^Pfo)
H-2n+r
i=i
m
H g n + r + l ^ ( - 1 ) ^ 1 A*P{n) + C7
i=o
since H 2n + r _i+i = F ^ ^ n + r
Fi_iH2n+r+i5
The value
(2.30)
n
y ^ P ( k ) H*k+r
]
Since
= y Y l ) X ^ r + 2 4 j V W ) + C3
k=o
m
P(n) + ^ < - l ) i F i + 1 V i P ( i i )
- H,
i=i
m
+ H a + r + ^ ^ D V i ^ V ^ f c i ) + C3
i=o
we may now state
Theorem 5. For m = 0 5 l 9 e o o ; n = l 9 2 5 c
n-i
(2.31) X >
)H
k + r = n + r E C"1)
k=o
s=o
E,!'"|E<-uS(l)Qi1-
m
B-n+r+i 2 , ^~1^
s=o
+ C3
I j=s+i
"m
| m
I j = S+l
s v
16
[Feb.
ON SUMMATION FORMULAS
where
m
m
(2,32)
3 =
{ 1)Ja G
j J
- ^E
Ij=L
i!F
i + 2 Z(-1)JajGJ
1-0
\ j=i
(r, m = 0,1, )
Comments,, If we compare (1,13) with (231), we conclude that for arbit r a r y aJf
J = 0 , 1 , . . . ,m ,
(2.33)
' '
:
s+i
>" j=s+i
)
m
(s = 0,1, , m - 1) ;
m
(2.34)
XM^OF^J
i=o
a/nGH
I j=s+i
^ '
'
(2.35)
m
m
m ! F m + 1 = (-!)*(!! )F 2 i
i=l
(-D^G*
J=i
Cm = 1, 2, ) ;
1967]
(2.36)
17
m
1
(m = 0 , 1 , ) .
Suppose now
m
(m)
P(k) = (-k)
= S(-D 5 s4k j
in (2.31). Then
n-i
n-i
/ k + m - 1\
k=o
T-
k=i
k+r
'
C3 = - ( m ! ) ( F m + 1 H r + F m + 2 H r + 1 ) - - ( m I ) H m + r + 2
We note that (2.4) and (2.7). are special cases of (2.30)..
3. ADDITIONAL RESULTS
In terms of forward differences it is readily verified that
n-i
(3.1)
m
p k H
C8
m
3 n + r ^ (-l) i 2- i " 1 F 2 i _ 2 A^tn)
i=o
m
+ H3I1+r+i 2
1=0
(- 1 ) 1 2 " 1 ^ i - i ^ ( D ) + C8
18
ON SUMMATION FORMULAS
(3.2)
2 POOHsk+r
m
] T (-1) X 2
k=o
i=o
^n+r^iV'PW
+ C8
m
H3n+r^
(-DV^^^iV'pW
i=o
m
+
n-i
(3
3)
m
p(k)H
X)
k=o
3k+r
we have
|
( - D W - V , a. (J)G
sn+rS
i=o
s=o
'i=s+i
J=S+1
m
m
+ H 3 n + r + i 2^
s=o
'j=s+i
1=0
(m - 0 , 1 , )
where
(3.4)
C8 = -H r Y, C-Di(i02"i"1F2i2
a G
i S
J=i
1=0
"
Hr+1
i 1
J = l
F o r n = 0 3 1 , *; r = 0, 1 , 2, * , we have
1967]
(3.5)
m
lj=s+i
s=o
+ C8
I m
"HI
+ H.3n+r+i
19
Es=o <-'
^j=s+i
(m = 0, ! , - )
where
HI
m
(3.6)
I>>S G J
C8 = H r
3=1
i=o
/ HI
HI
^iZ^^VE'-^f!
J 3
1=0
3=i
(3.7)
a s + J ] (-l)i(iI)2-i-1F2i_2
i=o
a ^ G *
I j=s+i
i=s+i
m
^ T f % E (-S(lK
\ j=s+i
1=0
(s = 0 , 1 , , m ) ;
m
(3.8) j ; w w % j ; a lu;
1
1=0
HI
j=S+l
/HI
( _ D ^ ^ - - V i E <1=0
li=s+i
1)Ja
j(i) 3-s
(s
= o.i.-.>
20
[Feb,,
ON SUMMATION FORMULAS
F o r a. = S^ ,
j - 0 , 1 , - , m,
respectively
(3.9)
m! 2-m"1F2m_2 = J ] i! 2"i"1F1-i ]T t - D ^ G J
1=0
(m = 1, 2, ) ,
3=i
(3.10)
m!2
hf^-i
il 2
^ - i E
1=0
( 1)Js G
" m J
<m = 0,1, ) ,
13=i
I
(3.11)
>l
Fi_ 2 yield,
(ml)2~m V J
m
i
1=0
(3.12)
-(m!)2
(m = 1 2 o o o )
' '
3=i
^m^a
m
= ^(-l)i(i!)2-i-1F2i_ii;(-l)JsiGi
i=o
(m = 0 , 1 , )
3=i
L .
n
Suppose now
po
(m)
s k
= 2 s m kj
j=i
1967]
in (3o3)
H1 1
- (F2m_2Hr + F2m_iHr+1)
Efem+r-l ~
2m-2-3nH3n+r
If we s u b s t i t u t e for
H3n+r+i,
(3.13)
21
we obtain
2m-l-3n H 3n+r+i
H3n+r
and
(-l)m(ml )2-m"1F2m_23n
m
-z
s=o
i=o
'j=s+i
(n = 0 5 1 , , m - 1; m = 1, 2, ) ,
(3.14)
(-l)m(m!)2
"1F2m-l-3n
m
E
s=o
Ij=s+i
1=0
(n = 0, V n i . - l ; m = l , 2 , - - )
Suppose now
m
<-l)jS^
22
ON SUMMATION FORMULAS
in (3.5).
[Feb.
4.
GENERALIZATIONS
Un+2 -
a U n + 1 + bU n ,
(4.2)
U n + 2 = aUn+i + U n ,
(4.3)
ab = 1,
a f
-19
(n = 0 , 1 , - - )
a f
0,
(n = 0 , 1 , - - ) ,
b = 0,
(n = 0 , 1 , - )
, .
We have a l r e a d y
1967]
(4.4)
Up+ S - b ^ g U p - ! + * s + i U p
where $0
0> <i = 1,
and $ n + 2 = a$n+i
U - n .= (U 0 V n - U n ) / ( - b ) n
V 0 = 2,
Vi = a,
n
n-i
S
(n = l s 2 5 )
and V n + 2 = aV n +i + bV n ?
n = 05l,e" ,
Since
3k+r
= J^i-DV
+ b r ^ n + r - i - h s i ^ P ( n ) + C^
i=o
m
+ b ) " 1 " 1 ^ ^ A2P(n)
- bU3n+r ^ ( - D V
i=o
m
+U8n+r+i^(l)i(a2 + W
" ^ . ! A W ) + C8*
1=0
and
HI
<*)
as
We note
P{k)U
k=o
4 7
that
n = l,2,-- .
<
fofai* n = 0 9 1 9 . We note
r = 05 1 , 2 5 .
that $ _ n - - * n / { - b )
4 6)
F o r negative s u b s c r i p t s , we define
(4.5)
where
23
k=o
3k+r
= ^(-D^a
+ b)
h l s n + ^ - i ^ P C n ) + Cs
i=o
m
= bIJ
3 n + r S (""1)1(a2
i=o
m
1=0
W'^Wi^Pdi)
24
[Feb.
ON SUMMATION FORMULAS
y^P(k)U3k+r
k=o
m
bU3n+r V "
s=o
+
^(-lftilMa* + b ) " 1 - ^ J^
>
m "m
j (Js) GjL
'j=s+i
/ m
n
3n+r+i / j
s=o
C ft
lj=s+i
1=0
where
m
m
2
i 1
C* = - b U r ^ ^ D ^ i l X a + b)" " 2 i-2 E a j
(4.9)
1=0
/ m
i
and m = 0 , 1 , o o , we have
For n = 0 , 1 , ,
(4.10)
J=i
n
^P(k)U3k+I
k=o
"m
m
= bU
( m
1 1
^ 1)Ja j (3s) G j -
i=o
m
u
3n+r+i J> (- If
s=o
j=s+i
/ m
^ i i ( a 2 + w - ^ W i J ] ("1)Ja5 (i) G j^
1=0
i=s+i
+ C8 ,
1967]
25
where
(4.11)
Ca - U
( j=i
| m
- ur+iX/Ma2
+ b)
)_
\
1=0
'
3=i
(4.12)
n-i
^PU
2 k + r
m
= J ] ( - l ) i a " i ' 1 U 2 n + r . 1 + i A i P ( n ) + C?
k=o
1=0
U2n+r
P(n)
+ ^ ( . D V
"
^Pfo)
i=i
m
U g n + r + i ^ - l ) ^ " 1 " 1 ^ - ! A*P(n) + C?
1=0
and
(4.13)
n
^P(k)U2fcfr
k=o
m
^ ( - 1 ) l a " 1 " l u 2 n + r + i - i V 1 P ( n ) + C*
i=o
U^+r^VDV
*_ i V 1 P(n)
i=o
m
+ U a a + r + i ^ t - D V ^ U i - i ^ P f c ) + C7* ,
1=0
26
[Feb.
ON SUMMATION FORMULAS
Theorem 8. Let U
XX
(4.14)
P(k)U 2k+I
k=o
m "m
m
=
2n+r /
s=o i=i
m I m
u
2n+r+iy J
s=o
in
3=s+i
j m
n
+ C7
*j=s+i
1=0
where
m
(4.15)
C* = UT ~ u r a0
E^^^-E^i
3=i
i=i
HI
" Ur+iS^^tt^^Vi^ajGJ
1=0
]=i
(4.16)
y^P(k)U2k+r
k=o
m
(-1)
s=o
E'-^-'lEw'Sli)
i=0
+ c* ,
< s
| j= s +i
"m
( m
+ U2n+r+i/J(-DS
s=o
1=0
I j=s+i
ns
Mo-' ll
3W
3-s
1967]
27
where
(4.17)
C* - U7 - ur
1=0
I j=i
I in
Ur+i
1=0
For a = 1, (4.14) and (4.15) yield Theorem 4; and (4,16) and (4017) yield
Theorem 3.
(iii) Let U n satisfy (4e3)
(4018)
Since
n-1
m
^P(k)Uk+r = ^ ( - 1 ) ^
k=o
Hln+r+i^i ^Pfo) + C*
1=0
b U n + r ^ ( - l ) V 1 fy2i ^ P W
i=o
m
+ U n + r + 1 ^ ( - l ) V 1 " V 2 i + i ^ P ( n ) + C*
1=0
and
m
(4.19)
^ P ( k ) U k + r = ^(-Dhr1
k=o
V + r + a + i V ^ W + Cj
1=0
= bUn+r
b 1]P(n) + ( 1 ) J b i 1 i + l V i p & l )
m
Un+r+i^jM-l)^
1=0
V i + a ^ P W + C* ,
28
[Feb.
ON SUMMATION FORMULAS
we
have
n-i
(4o20)
y^P(k)Uk+r
k=o
- bUn+r
m Tm
.
m
=oli=o
s^o[i=o
m
m fTni
'j=s+i
( m
+ Un+r+1 ^ ^ ^
s^o I i=o
s , 0*
+ C3 ,
/ m
3=s+i
where
m
(4.21)
C* = -bUr^C-lJ^Ob-^WEa.Gjj
i=o
( j=i
"
Ur+1
/ m
( 1)i(il
]C "
^"^^i+i 5^ j J
3=i
For m = 0 , ! , ; n = 1,2, ,
n-i
fc=o
+r "
bu
n + r s=o
^ ( " 1 ) " i=i
u
m
n+r+l / ^(-D
s=o
+ c3
where
we have
]TW)U k
J
a G
i=o
(4,22)
(j=s+i
i~o
I j=s+i
\
'
1967]
(4.23)
j m
1
C3
-bUry^ilb" '"Vi+i
i=i
\
j
\^(-l) a.G!
\ j=i
J
m
, m
1=0
F o r b = 1,
29
]=i
(423)
yield T h e o r e m 5e
5.
(5.1)
oU n +p + b 1 u n + p - i + 8 8 + b p u n = 0
Let x b e a r e a l number*
(5.2)
3-i
2 Ev -E B
1=0
k=o
(b 0 b p 0)
Then
bfxi
and
bpn+i+k-j
k=o
i=o
p-i
k=0
j=o
n+i+k
30
ON SUMMATION FORMULAS
[Feb.
Tb.u, . :
(5.3)
> u,x
k
k=o
i=o
The series in (5.3) converges for | x | < I XI, where A is the root of bpxP +
+ bjx + b 0 = 0 with the smallest absolute value.
In [ 6 ] , (5.2) was used to obtain a closed form for
E*
k=o
Jji4 - >
i=o
then
^
k
ukx0
k=o
P(k) - y \ . k j ,
f 0 ,
j=o
and
, k = 0 , 1 , , where q = 1, 2, ; r = 0, 1,
1967]
(5.4)
31
Uk+2 - VqUk+1 + d f u k = 0
(x - afl)(x - j8Q) - x2 - Vqx + d,
(k = 0 , 1 , - . . ) ,
- cP + 0 n , n = 0 , 1 , %
where V
- 0 .
Since
2m+2
2
(x - Vqx + 4 )
m+1
b2m+2_sxS
- ^
s=o
we have that
m+1
(i - v qX + dfx^)
2m+2
y.3
j=0
(5.7)
b. = ( - ^ ^ ( ' ^ ^ ^ ( j - i ) ^
i=o
" ^ ^
CJ = o . V . 2
2
k=o
k
Pfl^w^^x
v
qk+r
32
ON SUMMATION FORMULAS
If w, = KL, then dt = d2 = - 1 , V q = L q 5
(5.8)
-(1 - Lqx +
[Feb.
(-1)^)m+^p(k)Hqk+rXl5
k=o
2m+i
^b.Pfo
+ 1 + k -
j)Hq(n+1+k_j)+r x
n+l+k
k=o
2m+i
k
^ b . P ( k - j)Hq(k-j)+r x
k=o
(n = 0 , 1 / " )
3=o
w h e r e (see (5.7))
m+i
(5.9)
2 m + 2).
If P(k) = k
= ml l ^ J
x,
2m+i
(5.10)
XX
mk
~J)
n+i+k
q(n+i+k-jHr
k=o
2m+i
H
k=o
j=o
q(k-j)+r
'
(n = 0 , 1 , , m - 1; m = 1, 2, )
OTYI-J-0
21X1+1
(5.11)
m u s t b e 0,
/
\
f 2m + 2 - j l H
= 0
3\
m
/ q(2m+2-j)+r
i. e.
(m = 1 , 2 , - . . )
1967]
33
a r b i t r a r y x and n = 0 5
2H1+1
(5.12)
V *
k=o
. i+k
|Hq(i+k-j)+r
3=0
2H1+1
x*
(m-= 1 , 2 , - - - )
j=0
k=o
In (5.12), t h e coefficient of x2m+2 m u s t be 05 i. e . ,
2K1+1
{5 13)
J2
i=o
,
b
j (
v
3 m
m 1 " j ) Hqfem-w-jHr
(m -
1,2,---)
with V n = 2 c o s n 0 ,
where
0 fi 09TT,
b.P(n
k=o
k-
)|
s i n
. n+i+k
rq(n + l
k - j ) + r ] 4
J^o
2m+i
k=o
Z^
]
3=0
P ( k
M s m [ q f e - j) + r ] 0 /
(n = 0 , 1 , )
w h e r e {see (5.7))
1X1+1
(5.15)
b. = <-!>J J ] ( m ^ / ( j - i ) (2 C S qe)2i_j
i=o
(j =
34
ON SUMMATION FORMULAS
[Feb.
The relative simplicity of our results, (5.14) and (5.15), may be compared with
the less general (as well as less elegant) results obtained by Schwatt [8,
pp0 217-219], who used the differential operator,
For choices of P{k) = k
or (~k)
(xd/dx)
(5 16)
Y^
LJ
(5ol6)
b ( 2m + 2 - j W cos [q(2m + 2 - j ) + r ] ^
j \
m
/isln[q(2m+2-j) + r]f
{Jl
2H1+1
(5.17)
y
JLJ
j=o
ni
y i sin [q(2m + 2 - j) + r ] 0 J
Identities (5016) and {5.17) may be transformed to hold for hyperbolic functions
by recalling that cosh (i#) = cos 0 and sinh (10) = i sin# e
As an application of (5.3), we have
(5.18)
2 ^ b , P ( k - j)w q (k-j)+r x
k=0
3=0
(5.19)
2m+2
(1 - Vqx + d f x 2 ) m + 1 - ^
b.x j
3=0
(m = 0 , 1 , - - - ) .
1987]
(5.20)
(j = 0 , 1 , . . . , m + 1) .
(5.21)
(d? - d2Vqx + d f ^ )
m+1
35
and so
2m+2
= J2 b j ^ ( m + l ) x j
5=o
Replacing x in (5.21) by x/dj?, we obtain (in reverse order)
2H1+2
m+1
(5.22)
(* - Vqx + d^)
2TCL+2
= J^ Y^ "^
b2m+
HxJ
and thus (5.20) is obtained by comparing the coefficients of x** in the sums in
(5.22).
Let t = 1 , 2 , . ' . ,
[gt-Hl( )]
= P(k)H q ^. +r
is given by
Since
t+
m+i
(t+i)(m+i)
[x wa/x)] = Y, v j
j=0
EP(k)Hqk+rxk
k=o
36
ON SUMMATION FORMULAS
[Feb.
and g 3 (x) - x 3 - 2x 2 - 2x + 1.
= 0,
Then x 3 g ( l / x ) = 1 - 2x - 2x 2 + x 3
and
3<m+i)
(1 - 2x - 2x 2 + i?)m+1
b x3
= ^
i=o
Using the binomial t h e o r e m and then applying (5. 7) (with t h e p r o p e r change of
notation for t h e coefficients) , w e obtain
m+i
(1 - 2x - 2x 2 + x 3 ) m + 1 = X K
+ 1
( W C l + x -
(K^/2)]1
i=o
m+i i
2i
3m+3
-Zm^Zv^-EV
i=o
'
k=o
j=o
where
i
E(s)( t - s 0 ( - 1/2)t
>
U I i i,
,k-s
U-V*J
(k = 0, ! , , 2i) ,
s=o
and
(5.23)
b. =
f 1 *) (-2)^.1
1=0
m+i
=
i=o
/ s=o
(j = 0 , 1 , - , 3m + 3)
we obtain (where b .
1967]
- ( 1 - 2x - 2x 2 +
(5,24)
x 3
>m+iy]p(k)H|
k+r
k=o
3m+2
k=o
37
i Y ^ b . P ( n + 1 + k - j)H*
n+i+k-j+r
J=o
n+i+k
3m+2
X>Jp*-J)HLjj+r
J=o
k=0
[x*
g t f j d / x ) ]im+1
b.x]
(m = 1,2,- )
F
Then
(5.26)
(t+i)(m+i)-l
y
/(t
l)(m +
l)-j)
i
q(tm+t+m+i-j)+r
= 0
j=o
(q,t,m = 1,2,---;
r = 0, 1 , 2, J
(t+i)(m+i)-i
(5.27)
J=o
-i(*
+ l)(m+l) - j
m
(q,t,m = 1,2,---;
1 + m\
r -
t
q(tm+t+m+l-j)+r
0, 1 , 2, ) .
3S
ON SUMMATION FORMULAS
[Feb.
j = 0 , ! , 5m - 1),
=1,
that
3
(6sl)
i*3 Z
kIF
k+iGf
= 0 1
' '")
k=o
2,j = Z k l F k + 2 G ^
0 = 0,1,-
k=o
(6 3)
3,3 =JjlF^
'
"j
t^0'1'-)
k=o
Thus, the assertion [9, (6e)] is valid only for i = 1,) (with j = 0,1, ) and
i = 3 (j = 1,2,-..). Since F k + . = F . ^ F ^ + F.__ 2 F k+l (see (1.10)), we obtain
from [9, (6b)], using (6.1) and (62) above, that
j
(6 4)
'
iJ
i-4
f c + 1)JF
M F
k+i? " E
k=o
i-3-k
(J = 1. 2, ) .
k=o
o, J
I>FkG?
k=o
= 0,1,
1967]
39
In t h e
S ( m s n - 1) = F P ( m , n ) + F
n
where (using
, P 0 ( m , n ) + C(m)
n~~ i
9, (2b), (3b) )
C(m) = ( - l ) m + 1 M
(6.6)
(m = 0 , l , - " ) .
2, m
(6 7)
39J
= ( i)j
- 2]("1)k(k!)F^Gf
(J = 0,1,
"") ?
(6.8)
1 J
<j = 0,1,---) .
k=o
Since M . = M . + M, . for Jj = 1, 2 , e " , w e obtain from (6.7) and (6.8) that
3SJ
2,3
1,3
(6.9)
Muj
1 3
Since F 2 k + i - 1 = F i 1 F 2 k 4 - 1
u s i n g (6.8) and (6.9), that
+ F
)F
i-2:F3k
2k G j ? :
(see
0 = i'2'"')
(la0))s
w e
(6.10)
k=o
obtain
f r o m
t9*
(6b)
'^i-S-k
k=o
(J = 1, 2, )
F r o m (6.4) and (6.10), we conclude that
j
j
j
k
= Ek'Fk+iG^
k=o
(j = 1, 2, ; i = 0,1, ).
40
ON SUMMATION FORMULAS
[ Feb-
It should be noted that [9, (7c)] was obtained from [9, (6a)], using [9,
(7a)]. Since 1.9, (7c)] is a linear difference equation of second order in i, its
solution is
i-3
(6.12) P.(m,n) = F . ^ ( m , n ) ,+ F . ^ P ^ m . n ) - ^ ( n - k ) m F . ^ 1 _ k
k=
ft
= 3.4..
P (m,n) + (-l) m + 1 M 0
S(m,n - h) = F P (m, n) + F
n
n+i
2,m
k=o
(6a). (7c)]) to
n
m
TYi
(6.14) ^ ( n - k + l ) mm
F u = M<
_ FF _ , A +
+ M
M n _ FF _ +
" n
i.m n+i
2,m n+2
k=i
(m = 1, 2, - . ) .
where Q(m,n)
Q(m.n) = ( - l ^ P ^ m , n )
J ] ( ^ )
^ f
we obtain the
1967 ]
(6.16)
= F ^ Q t m , ! ! + 1) + F.
41
Q(m,n)
1-3
J2{n+k)mFi-i-k
(i = 3 4 8
k=o
P.(m,n) = ^ T (-l) j f\ M . ^ ^
j=o
V /
(m = 0,1,<
k=o
k=o
^ '
(6.19)
M. jm - ] T V l ) 3 ( ^ ) P.O.n)!!111"5
(m = 0,1, ) .
(6.20)
Q(J5
MUm = Y^lf)
~n + * " 1)nm"j
5=0
From (1.19), we obtain for n = 0, recalling (6.9),
m
(6.21)
(-Dm^)F2m
(-D^i.j
j=l
(m = 1, 2, )
42
ON SUMMATION FORMULAS
[Feb,
r e c a l l i n g (6.8),
m
(622)
(^l) (ni!)F2m+1 = ] T ( - l ^ M ^
(m = 0 , 1 , ) .
3=0
F r o m (2.35), we obtain, r e c a l l i n g (6.9),
m
(6.23)
ml F
_,_, = 7 S j M, .
m+l
L^J
m 1,3
(m = 1, 2, ) .
j=i
F r o m (2.36), we obtain, r e c a l l i n g (6.8),
(6.24)
mSF
If we set
= V S j Mo .
m+2
L^j m 2,]
j=o
JO
(m =
b = 2 in (4.3), then U
so that a
0,1,-0.
= 1, a. - 0, j = 0, l s - , m - 1.
m
j
gives a closed f o r m for
In
Thus,
n-l
vk, m
^(~l)kkm
k=o
ACKNOWLEDGEMENT
T h i s p a p e r , consisting of the f i r s t five s e c t i o n s , w a s submitted to t h i s
Q u a r t e r l y on S e p t e m b e r 2 1 , 1964, after the author reviewed t h e p a p e r by Alfred
[l].
having r e a d the p a p e r by Ledin [ 9 ] . I wish to thank Dr. Hoggatt for the o p p o r tunity of r e a d i n g the p a p e r s [ l ] and [9 ] before t h e i r publication. As a r e s u l t ,
m y p r e s e n t p a p e r is m o r e complete,
1967 ]
43
REFERENCES
n
1. Brother U. Alfred, "Summation of T ^ F ,
k=i
k+r
3. R. Go Buschman, "Fibonacci Numbers, Chebyshev Polynomials, Generalizations, and Difference Equations," Fibonacci Quarterly, Vol. 1, (1963),
No. 4, pp. 1-7.
4. R. I. Reichman, "A Summation Formula Involving Fibonacci Numbers, n
Scripta Math. Vol. 20, March-June, 1954, pp. 111-112.
5 K. Subba Rao, "Some Summation Formulae Involving Fibonacci Numbers, "
Scripta Matho, Vol. 21, June-Sept., 1955, pp. 214-217.
n
kp
k=o
Soco , 15(1964), pp. 642-6470
and
n
,
(-1) kP, " Proc. Amer, Math.
k=o
7. D. Zeitlin, "On Summation Formulas for Fibonacci and Lucas Numbers, "
Fibonacci Quarterly, Vol. 2 (1964), pp. 105-107.
8. L J. Schwatt, An Introduction to the Operations with Series,
Chelsea,
T\
(ab)
TO
R =
["o i ] '
S=
["i ~]
But
RS = [I J] = Q .
the Q matrix. Now Basin and Hoggatt [l] have shown that
(RS)n
= Qn =
"n+i
^ I
"n-i
INTRODUCTION
The sum
n
S(m,n) = J ] k m F k
k=i
(where F, is the k
ues of me
The case
method
46
[ Feb.
(2b)
P 2 (m,n) = b m n m + b m 1 n m
+ . . . + b4n + b 0
Theorem 1.
C(m) = -b 0
Proof.
Take S(m, 0) = F ^ m , * ) ) + FoP^m.O) + C(m)
from (1))
0 = P 2 (m,0) + C(m) but P 2 (m,0) = b0 from (2b)) .
Inspection of the first few values of m (see Table I) leads us to the following determination of the polynomials (2a) and (2b).
m
P^m.n) = ( - l ) J ( ^ M l 7 J n H
(3a)
ILL
P2(m,n) ^ ( - ^ ( r ) ^ , ^
(3b)
n
P^m + l,n) = (m + 1) J P^m.njdx + a{
o
(4b)
P 2 (m + l,n) = (m + 1) J
n
P 2 (m,x)dx + bj
47
Table I
LIST O F FIBONACCI SUMS OF THE TYPE
S m
( > ) ^ Z k ^ ^ k
n+i
2(m'n)
F ^ O ^ n ) + C(m)
S(0,n) = F n + 1 ( l ) + F n ( l ) - 1
S ( l , n ) = F n + l ( n - 2) + F n ( n - 1) + 2
S(4 $ n) = F
+ 257) - 416
+ F (n5 - 5n4 + 50n3 - 310n2 + I285n - 2671) + 4322
6
8(6,11) = F
10
+ 8054177) - 13031936
S(99n) = F ,,(n9 - 18n8 + 288n7 - 4200n6 + 52416n5 - 544572n4 +
n+i
3
+ 4526592n
- 28220040n2 + 117287424n - 243733442) +
+ F (n9 - 9n8 + 180n? - 2604n6 + 32382n5 - 336546n4 +
n
+ 2797620n3 - 17440956n2 + 72487593n - 150635551)
+ 243733442
S(10,n) = F
(n10 - 20n9 + 360n8 - 6000n7 + 87360n6 - 1089144n5 +
n+i
+ 11316480n4-- 94066800n3 + 586487120n2 - 2437334420n +
+ 5064892768) + F (n10 - 10n9+ 225n8 - 3720n7 + 53970n6 - 673092n5 + 6994050n4 - 58136520n3 + 362437965n2 - 1506355510n + 3130287705) ~ 5064892768
48
LFeb.
NUMBERS
M i.5
M< 2 . 3
31
50
257
416
2671
4322
33305
53888
484471
783890
8.
8054177
13031936
150635551
243733442
10
3130287705
5064892768
(5a)
aj = 1 - (m + l)f
( P ^ m . x ) + P 2 (m,x))dx
(5b)
bj = 1 - (m + 1) J (Pi(m,x) + 2P 2 (m,x))dx
o
Proofo
P r o v e (4a) f i r s t .
xxx
(m + 1) { P i ( m , x)dx = (m +
iX
/
m
Cm + 1) X V D X j d) ^
H
QX
n
C
= E (-DV 3 ( m + 1) n+H
j=
1967]
49
(aj is determined for j = m + 1, a value which is missing from the summation sign.) A similar proof establishes (4b).
Now,
l
bj = P 2 (m + 1,0) = P 2 (m + 1,1) - (m + l ) / p 2 ( m , x ) d x
o
and since S(m + 1,1) = 1 = P 2 (m + 1,1) + Pj(m + 1,1) + C(m + 1)
(C(m + 1) =
1 = (m + 1) J Pi(m,x)dx + aj + (m + 1) J P 2 (m,x)dx
o
o
and the value of a* follows,, A similar manipulation yields the required value
of bj .
Corollary 1
dPjdn + l,n)
dn
dP 2 (m + l,n)
= (m + D P i1dVn , ! ! ) ;
Corollary 2
d r P!(m,n)
d r P 2 (m,n)
= m(m - 1)
dn
(m - r + l)P 2 (m - r, n)
Corollary 3
P 2 (m, 1) = a0
Example 1
Problem.
n
\ J kF,
k=i
50
[Feb.
Solution. We know
n
E F,k
= F ^ + F - 1
n+i
n
(m = 0) .
k=i
So the polynomials are Pt(0fn) = 1, P2(0sn) = 1. Now, applying Theorem 29
Pi(l,n) = J l d x + aj = n + aj
o
and
P 2 (l,n) = f
o
Idx + b^ = n + bj
aj = 1 - J* (1 + l)dx = 1 - 2 = 4
and
bj = 1 - / (1 + 2)dx = 1 - 3 = -2
(n - 2) + F (n - 1) + 2,
Example 2
Problem.
EkFk
k=i
= F
n+i{n"2)
+ F
n(n"
1) + 2
P 2 (l s n) = n - 2
Nowf applying
Theorem 2
1967 ]
51
(n2 - 4n + 8) + F (n2 - 2n + 5) - 8
Theorem 3.
If u, are the "generalized" Fibonacci numbers (i e, numbers obeying
the Fibonacci recurrence relation, but with different initial conditions) with
the properties u k + 2 = \
+ 1
u0 = q, uA = p, [ 7 ] , then
E kmu
k=i
where P 2 and Pi are polynomials defined as above (3a9 3b) and K(m) =
-(pb 0 + qa0 ).
In Theorem 3 we have stated a simple and useful result.
this theorem is trivial, since u, = pF^ + qF,
The proof of
M y + 1 > 2(j + l ) M y
(j > 1)
hold for all cases shown on Table II and can be proven by total induction using
the formulas developed for aj and bj
between these numbers; this relationship, and the fact that these numbers are
members of a whole class of numbers M y can be appreciated effectively in
Table IV Horizontal addition of two consecutive M y numbers is the basic
52
[Feb.
= L ^ P ^ n )
m = 0
T(0,n) = L n + 1 ( l ) + L n ( l ) - 3
m -
T ( l , n) = L n + l ( n - 2) + L n <n - 1) 4- 4
+ L^fr^n)
+ K(m)
m = 2
T ( 2 , n ) = L n + i ( n 2 - 4n + 8) + L n (n 2 - 2n + 5) - 18
m = 3
m = 4
m = 5
p r i n c i p l e in the c o n s t r u c t i o n of T a b l e IV; t h e r e s u l t s of s u c c e s s i v e h o r i z o n t a l
additions can be followed with t h e aid of t h e b r o k e n l i n e s . T h e following i l l u s t r a t i o n should clarify t h e p r o c e s s ;
1
\
^^
\
_
\
-31*"
^ \
M+5=6 -^ _ _
5+31=3
1
~ ~ ^ 6+36=42
1
2<L
2+6=8
_ _
\
^.8+42=50
50-
^8-
= M
+ M
(i = 3, 4, 5 , - - ;
- (i - 3) 3
j = 0, 1, 2,--- )
1967 ]
53
-31-
-257-
-2671
/
\
288-__
2928
- 42 - _ _ _ "^ - - 3 2 4 - _ _
-3216
/
\
^ - - 3 6 6 - _ _
-3540
7
X
/
\
^
^
3 9 0 6
/
x
6 ^ _
/ \
36 -
\ /
\
58-^_
68-_
/ \
^ - \
1 3 -/
-416-
-4322
466-__
- - 524- _ .
- - 592- _
\
/
io^.
/ \
-50-
\
\
\
/
- 8 1
4738
' - -5204
"^-5728
----6320
-6993
6 73*
1 6 ^ _
/ \
/
\
/
\
/
\
\ 9 4\/x - _ _ _
" - -110 - _ _
/ \
/
\
X
754-~^__
7666^
^ ~~ - 8 4 8 - _ _ __ - - 8 4 2 0
"--958--__' ^--9268
/
\
~"^-10226
'
\2 0
1/
130-
1088N
'
-11314
//
24^__
150^___
1218-__
12402^
\
^ - - 1 7 4 - ^ _ _ _ ^--1368-^ __^~^-13620
/
\
/
\
^--1542-_^"~~"^-14988
X
/
/
\
/
\
^-16530
\
\
/
\
/
\
/
/
\ /
\
/
\
/
\
29203-
1745^
/
[/
X /
34-^-___
232-___
1948-___
20020
/ \
^-266-__
^^-2180 _ _ _
"~~-21968
/
\
/
\
~^^--2446___ ^--24148
/
\
/
\ x
'
\
^--26594
/
\
/
/
\ //
X
\
/
402752-/
-306-
18275
\ 29346
N 3. 46 /- _ _ _
N 3M05 /8 X _ /
32098
^
3
9
2
_
_
_
^
3
4
0
4
^-35156
/
/ \
^
^--^3796- _
--38560
/
X
/
/
/
~^-42356x
x
/
\
/
\ N
/
\
/
\
| /
N
. 445
- X 4241^
46597
?7
5 3
4 6
/ ^
\
\
\
54
[ Feb.
i-4
(6b)
M. . = F . ]VL . + F . M, . - V * (k + 1)^ F . Q ,
i,j
l-i 2, j
i-2i,]
L^
i-3-k
k=o
w h e r e F . r e p r e s e n t s the i '
Fibonacci numbero
M2jj
is noted f r o m the f o r m u l a s
3
(60
MlfJ = E ( - 1 ) h ( h )
h=o
and
2,J-h
2,j
'
E
h=o
(h)
V
^"h
'
<6d)
M ^ . = (i - 1)1
(2j"h - 1) ( i ) M.jh
h=o
with j > 0, M. = 1,
J
*
i,o
M. = i > 1.
1,1
M. . = V h ! $ h F ^ .
(6e)
i.J
^
h=o
h+i
V\ and P 2 a r e , s i m i l a r l y ,
s p e c i a l c a s e s of a m o r e
g e n e r a l c a s e of polynomials.
1967 ]
m
P.(m f n) = Y (~1)3M. . H
(7a)
nm~j
isj I ] I
L^
55
X !
j=0
P. + h (m s n) = P . f r ^ n - h )
(7c)
P.(m3n) = P. .(m,n) + P.
(msn) - (n + 3 - i) m
(i = 3, 4, 5, )
These properties (7) enable us to obtain the following formula, thus generalize
(1):
(8)
+ nmF
n _ 1
and it seems quite natural* that we apply our results to the "convolution type1
sums of the form
n m F i + (n - l ) m F 2 + (n - 2 ) m F 3 + . . . + 2 m F n - 1 + F n
Theorem 40
(9)
(n
"
k + 1 ) m
R(m n)
'
k=i
s5mFn+i
+ M
2 3 m F n " P 3* (m > n)
56
where
and M
o, m
are particular
c a s e s of the
<i|,m
[ Feb.
M. . n u m b e r s
i,j
(see
T a b l e IV) and Pj^m, n) (the "conjugate" of the polynomial P 3 (m, n)) is defined
a s follows
in
(10)
P*(m,n) = ^ M 3 j j H
A l i s t of t h e s e "convolution-type" s u m s i s provided in T a b l e V.
Table V
n
2 >
k=i
m = 0
m = 1
m = 2
m = 3
m = 4
m = 5
V 1 ^ xF ^
" k + 1 ) ^k = xv
Rvu^
( m , niv) = iM
3jin n + 1
V i ? _ x F - P 3 ( m , n )
+ XM
2,m n
R(0,n) = F ^ + F - 1
n+l
n
R(l,n) = 3F _,_ + 2F - (n + 3)
n+l
n
R(2,n) = 13F ^ + 8F - (n2 + 6n + 13)
n+i
n
R(3,n) = 81F ^ + 50F - (n3 + 9n 2 + 39n + 81)
n+l
n
R(4,n) = 673F ^ + 416F - (n4 + 12n 3 + 78n 2 + 324n + 673)
n+l
n
R(5,n) = 6993F , A + 4322F - (n5 + 15n4 + 130n3 + 810n2 + 3365n + 6993)
n+l
n
Pt(m,n) = Q(m,n + i - l )
and
P.(m,n)
[ 8 j , then
= ( - l ) m Q ( m , - n + i - 1)
P.(m,-n).
(12)
S(m,n) = ( - l ) m J F n + 1 Q ( m 5 - n + 1) + F n Q ( m , - n ) - Q ( m , l ) }
these
1967 ]
57
(13)
G(r,s,n) =
hi p a r t i c u l a r ,
G ( l , 0 5 n) = G
h a s t h e following limiting b e h a v i o r :
G
-S-JL = l i m (G ^ - G ) = 1
G
^
n+i
n
n>oo
n
n> oo
lim
Some of t h e s e questions a r e :
(All we need
^2P.(m,n)
P.(-l,n)
i
2.
m?
lim
^
m*>o
dndni
m?
S(r,s,n) = krFjJ
k=i
58
F e b . 1967
ACKNOWLEDGEMENTS
The author w i s h e s to e x p r e s s h i s g r a t i t i d e to Dr. V e r n e r E. Hoggatt, J r .
and the r e f e r e e , who had v e r y useful and c o n s t r u c t i v e c o m m e n t s on s e v e r a l
a s p e c t s of t h i s p a p e r .
REFERENCES
1.
ff
4.
5.
6a
P r o b l e m H-57 of "Advanced P r o b l e m s and Solutions," The Fibonacci Q u a r t e r l y , Vol. 3 , No. 1, (1965) p . 45.
7.
P r o b l e m H-44 of "Advanced P r o b l e m s and Solutions," The Fibonacci Q u a r t e r l y , Vol. 2, No. 3, (1964) p . 205.
8.
10.
*****
In this paper we give a simple lattice point solution to a generalized permutation problem of Terquem and develop some elementary results for the
extended Fibonacci numbers associated with the permutation problems
The classical permutation problem of Terquem [12] has been stated by
Riordan [10, p 17, ex 15] in the following manner,.
Consider combinations
of n numbered things in natural (rising) order, with f (n,r) the number of r combinations with odd elements in odd position and even elements in even positions, or, what is equivalent, with f (n,r) the number of combinations with an
equal number of odd and even elements for r even and with the number of odd
elements one greater than the number of even for r odd.
It is easy to show that f(n,r) = f (n - 1, r - 1) + f(n - 2, r), with f(n, 0)
= 1, and explicitly
(1)
f(n,r) =
Moreover,
n
(2)
f (n) = ^T f (n, r) = f (n - 1) + f (n - 2)
r=o
60
stated a s follows.
[Feb.
Skolem T s g e n e r a l i z a t i o n m a y b e
de-
f(n,r;m) = N < a 1 a 2 o e a r :
a. = j (mod m)>
r + km,
with
k =
m
s i n c e r + k m < n i m p l i e s that t h e l a r g e s t i n t e g r a l v a l u e of k cannot exceed
(n - r ) / m 0
1, 2, , n which m a y a p p e a r in a combination,,
s i s t s of all t h o s e e l e m e n t s < n
m a y a p p e a r in the j
column con-
position.
(0,0)
1
#-$* X
1 +m l + 2m
i-
That i s , t h e j
1 + km
i.
2 + m -
-3 + m -
-2 + 2m
-3 + 2 m
-2 + k m
i.
3 + km
r +m
2m
-i.
-r + km #
B = (r,s)
Fig 0 1
1967]
from a = (0,0) to B = (r,s). MacMahon [7, VoL I, p 167] shows that this
number is precisely
(
hi our case s = [(n - r ) / m ] c
r +
f (n, r;m)
(4)
(m - l)r"
m\[-
HI
To illustrate, we
1r
2i
fiD
3 _ J ^--10*-
1 2
11
1 6
1
6
7
11
11
10
5
5
6
10
11
11
10
11
The clue
62
FH
1|
-5*-
11
*7
t
11
. Q , JL^^io^-
1 2
1
3
3
4
7
1 2 9
1 5
6
10
7
4
4
5 6
5 6
7
10
10
10
10
10
8 9
10
10
10
8 9
10
n
(5)
f(n) = f m (n) =
j*n + (m - l)r"
m
f m (n) - f m (n - 1) + f m (n - m)
t m - t111"1 - i - o
63
HI
(8)
(n) = \ ^ C
ZmJ
tn
rr
,
L
= a
n , ,n
+ b
for the F i b o n a c c i - L u c a s n u m b e r s . T h e c o n s t a n t s C
,
*
m a y b e d e t e r m i n e d from
t h e s y s t e m of m l i n e a r equations in C :
m
(9)
7 J C r t 3 r = j + 1,-
for
j = 0, 1, 2, , m - 1 .
r=i
F o r example 9 when m = 3, an a p p r o x i m a t e solution of the equation (7) i s given
by
w h e r e i2 = ' - 1 .
t j = 1.4655 ,
t2 -
-0.23275 + 0 o 792551 ,
t 3 = -0 o 23275 - 0 o 792551 ,
(11)
t2 = J - A + i
1
where
A + B
A^B
A - B ./-o
64
Let
n (t - t.) = t m - t m
(12)
- z
Then
m l
(13)
lL ]
j=i
\b>
" m) ^ '
n> 15
k=o
where
/ i.\
(a b)
k '
/ a + bk \
= rrbk (
xa = ^ T Ak(a,b)zk
(14)
with z =
,
k=o
expansion
x - 1
b
x
(15)
a s f i r s t noted by H. A. Rothe
we define t h e power s u m s of t h e
by
S(n) -
l)
65
+ z = t . , we find that
3
m
{t*" 1 + z t ^ 1 1 1 J =
S(n - 1) + zS(n - m) =
j=i
t ^ / t ^
j=i
+ z\
- J ]
t^t*1,
j=i
(17)
S(n)
n - (m - l ) k (
Z^
k=o
hi p a r t i c u l a r ,
|- (m
k
l)k
n > 1 ,
(18)
[r+ij
\-^
/ n - rk\
n- k(r+i)
satisfies
(19)
= ax
n - lJ
+ bx
n-r-i
bk
The classical
One
should also compare the Fibonacci-type relations here with the expansions
given in [5]
(n i k) r -2kzk
n+i
__ x
=
n+i
- y
x - y
k=o
and
V^ / n - k \
k n k
=r \ / -
k=o
n-2k-i k
n +, y n
x11
'
x + y
If the initial
values were the same then we would have found a formula for the permutation
problem not unlike (17) There are many papers (too numerous to mention) in
which complicated binomial sums are found by lattice point enumerations,, The
convolutions in 2, 3,4, 5J may mostly be found by such counting methods,, We
also note the recent papers of Greenwood |jf] and Stocks [ll] wherein the
Fibonacci numbers occur 0
The convolution addition theorem [2, 3,4, 5] of H Ae Rothe (1793)
(20)
^ A k ( a , b ) A n _ k ( c s b ) - A n (a + c,b) ,
k=o
valid for all real or complex a, b, c (being a polynomial identity in these), has
been derived several times by lattice point methods e We mention only a novel
67
derivation by Lyness [13j 0 Relation (20) has been rediscovered dozens of times
since 1793, and its application in probability., graph theory, analysis, and the
enumeration of flexagons, etc , shows that the theorem is very usefuL In fact,
it is a natural source of binomial identities,, We should like to raise the question here as to whether any analogous relation involving the generalized Terquem
coefficients (4) existSo It seems appropriate to study the generating function
defined by
T(x;a,b) = J^
(21)
fa + (b - l ) n 1 \
n=o
for as general a and b as possible., If b is a natural number and a is an
integer >0, the series terminates with that term where n = a, as is evident
from the fact that a + (b - l)n < bn for n > a and the fact that |
1 = 0 for
k < n when n > 0, provided k > 0o We also note that for arbitrary complex
a and | x | < 1
T(x;a,l) =
C n (a,c,b) -
which occurs in
^
k=o
/ f a + ( b - l ) k l \ / f c + (b - l')(n - k)] \
68
xnCn(asc?b)
Feb
n=o
for a r b i t r a r y b 6
REFERENCES
1.
Equation
2.
3e
4C
5.
6.
7.
8.
Convolution,"
Amer.
Fibonacci
Chelsea
11.
12o
13.
25(1941),
H-103
n
8
== FF
F 1
'3k+I
3k+I 3k+2
3k+2 66k+3
-'~
~3n+3
k=o
H-104
Proposed by VernerEo Hoggatt, J r s , San Jose State College, San Jose, Calif 0
Show
L
mX
m+
1 - 5Fm X ^ <-l) W
where L m andm F
H-105
are the m
k=o
5 (
^ mk
+ XL
^Dm)X2k '
(n + 1) P - n P = 6N + 1
where N is a positive integer.
Generalize.
69
70
Feb.
Show that
a)
k=o
k=0
Xx
'
k=o
k=0
H-107
'
'
T-^
/ \ 2
k=o
b)
'
k=o
Show that
F
p+2n
H-108
p+n
F ^0
q+2n
F ^
q+n
r+2n
for all integers p, q, r,
r+n
p
q
| - 0
and n.
Find the sides of a tetrahedron, the faces of which are all scalene t r i angles similar to each other, and having sides of integral lengths.
H-109 Proposed by George Ledin, Jr, San Francisco, California*
Solve
X2 + Y2 + l = 3XY
for all integral solutions and consequently derive the identity:
F2
+ F2
+ l = 3F
F
*6k+7
*6k+5
-1
^6k+7^6k+5
"
oo
l^ j-m
k=i |_ k J
1967 ]
71
w h e r e [ x ] i s the g r e a t e s t integer in x9
H-lll
[C
L n = "g" j l + 4 cos 2 ^ L z J :
m l
for n > 1 .
L*
- L 5 - L5
=
n+i
n
n-i
F^
- F5 - F5
= 5F
F F ,(2F 2 + ( - l ) n )
n+i
n
n-i
n+l n n - i
n
c)
L7
- L7 - L7
= 7L
T L
(2L2 - 5 ( - l ) n ) 2
n+l
n
n-i
n+l n n - i
n
F7
- F 7 - F 7 , = 7F
F F
(2F 2 + ( - l ) n ) 2
n+l
n
n-l
n+l n n - l
n
d)
5L
n+i n n - i
(2L2
- 5(-l)n)
SOLUTIONS
NO SOLUTIONS RECEIVED
H-59 Proposed by D. W . Robinson, Brigham Young University, Provo, Utah*
Show that, if m > 2,
1, 2, 3, - , F , * r e d u c e d modulo m is t w i c e t h e l e a s t p o s i t i v e i n t e g e r
mod 1 0 ,
then pt = 60,
, = F
i s the k
digit of the n
Q(n, k) i s p e r i o d i c ,
that i s q,
an explicit e x p r e s s i o n for q, .
H-62 Proposed by H. W Gould, W Virginia University, Morgantown, West
Virginia (corrected).
Find all polynomials f(x) and g(x),
of the f o r m
Find
72
an
[Feb.
integer
3=0
s
g(x) = Y b . x ,
b . an i n t e g e r
3=o
such that
2Jx 2 f 3 (x + 1) - (x + l) 2 g 3 (x)} + 3{x 2 f 2 (x + 1) - ( x + l) 2 g 2 (x)}
+ (2x + l){xf(x + 1) - (x + l)g(x)} = 0 .
LIMIT OF LIMITS
H-61 Proposed by P, F. Byrd, San Jose State College, San Jose, Calif.(corrected)
Let
f
=0
for
0 < n < k - 2,
f,
, , = 1
and
k
f
, = / f .,
n?k
-** n - j , k
3=i
for
n > k
Show that
n+i,k
Hence
f
lim
Hm
'
f
k( n-8^oo
_!.< 1
'
n+i,k
is
l
f (x) = x
Let r
1 ,
ljK.
r , , e , r, ,
2j]
k-i
-x
k-2
- x
.
- - x - 1
JijK.
such that
1967]
73
k
nsk
*->* j 9 k ],k
f(l) = 1 - k < 0,
f(2) = 1,
so that
and s i n c e f i s continuous,
Note
that
the r e m a i n i n g
root
(1 < j < k) ,
k,k
so t h a t
lim
^oo
, /f
'
k /
,J ,
n+1 k
'
I V b. , r11,
/
Y! b. , r. ,
A
= M/ 1/r,,
k,k "
\PL
^ k J' k // \ j-i J'k ^k k , I
lim
n^oo
Now
shows
f
Since
(m)<
i < j t < i
we have 2 > r, , > 2k/(k + 1),
JtC,.K
problem.
k.
74
[Feb.
is the n
Fibonacci n u m b e r . Show
n
2m-i
m-l
]T F(2m - l , n ) = 0 L 2 i ,
m > 1 .
n=o
i=o
Solution by Douglas Lind, University of Virginia, Charlottesville, Virginia.
n
Put
Sn = F<n>r)
r=o
and
(1)
= ^(-l)r(r+l)/2F(n,r)xn-r
r=o
f n (x)
where L
7i) = iL f
(-i)
2m+i
2m 2m-i
2m
*>
2r +
*)
m-i
m-l
E F < 2 m " 1. 2 r + 1) + i L 2 m J2 F ( 2 m - 1, 2r) ,
r=o
r=o
,
2m+i
= L
, which, with Si = 2 = L 0 ,
1
2 m 2H1-1
p r o v e s the proposition.
J-
where F
F
is the n
n+i
ini(l-2ico
JiL)
Fibonacci n u m b e r .
and d fi 0 be a r b i t r a r y r e a l
1967]
75
= W , n = 0 , 1 , , when W0 = 0 and Wj. = 1* In terms of Chebyshevpolynomials of the first kind, T (x), and of the second kind U (x), it is readily
verified that
(1)
Z + 1 , cn/2U (-$-)
n+1
\2V5 /
e 2cn/2T
(-*-)
\2VH /
Since
u n - 2* n (x - cs J t t , ) , T W - 2 - n (x - cos <5ti)n),
we obtain from (1)
<2>
V .
(3)
-"*&?-'-*)
V.
n11 -
"
If d = 1 and c = - 1 , then Z = F
n
n
"
"
<4)
(5)
n+1
,.a(1.Me0BfiLiJ)n)
11
3
Also solved by F, D. Parker, John L Brown, J r . , and the proposer.
the left of the remaining digits of M forming a new integer s Ns such that
N = 9Me Find the smallest M for which this is possible*
Solution by Robert H. Anglin, Danville, Va, and Murray Berg, Oakland, Calif*
n
M = 9 + X^x.10 1
i=i
76
[Feb.
n
N = x.10 1 + 9 - 1 0 n = 9M
1=1
= 9 ( l Q n + 1 - 1)
89
89999-991
89
10" V 2 k
m=l 10
w h e r e [ x ] i s the g r e a t e s t i n t e g e r function in x.
Also solved by Marjorie Bicknell, James Desmond, A . B. Western, J r . , C . B . A ,
Peck, and the proposer,
a,.
Lax3 = 0
j=o
be
rl9 r2, o o , r m
,n
* = y
. m.-i
ri. t
l
b..tj .
Vol,
1967]
77
Jy
<*)
b..ir
ij
and put
^
oo
YW
,n
r.t
= E w = E * " EYbf .
r
n=o n m
i=i
i=o
Now define n , v = n(n - ! ) (n - s + 1), n , ^ = 1,
"k
Y k (t) = e
E b,kj.t
so that
1J
j=o
oo
.(t) =
E E b,kj.
v!
v=o j=o
oo S k
= E E b ki
r>
Wv
'(j)
V=0 ]=0
F o r p = 0 , 1 , s s,
v+j
*
F^TJ!
T
+ JJ ) . .
put
oo
Ykk-npft) .= v=o
EVk(v
+ P)
f v+p
(P) ^rw
,v+p-n
,
n-p
n,,
-P\
t
_^
= b, r, ^ n , x = r, (b, n , x r, ^ ) .
^ ( p ) (v + p - n)! t=o
kp k
(p)
k v kp (p) k
T h u s applying the i n v e r s e t r a n s f o r m (393)9 we find
=
oo
y b
v(v
^
kp k x
yn = Y(n)(0) = g
s,
E YW (0) =
k=ip=o
^sP
g rn ^
i=i
j=0
-J
X
J {V
78
[Feb.
correct result. Also, since n^x = n 1 , his result will be correct if all roots of
multiplicity two are one, and there are no roots of greater multiplicity.
For
higher multiplicities his form almost never gives the correct result.
(i) We need only take a recurrence whose auxiliary equation does not satisfy
the conditions of (ii) to form a counterexample to (*).
Also solved by P. F. Byrd and D. Z e i t l i n .
Editorial Comment; The b.. in the first displayed equation above are arbitrary
constants. The b.. in the second displayed equation are also arbitrary constants. In this sense Jeske is correct. However, most readers would probably
incorrectly infer that after you have determined the specific constants for a
given problem one can then use these in the second displayed equation which, of
course, is not true in all cases.
V. E EL
AN INTERESTING ANGLE
H-67 Proposed by J . W . Gootherts, Sunnyvale, California.
S^JDL ,
|B|1V|*
respectively.
(2)
B. V = F ,
m+2n
|B| =
v^)
1967]
(3)
v
'
79
| V | = \/F
, F , ^ - F
F
v
I I
m+n m+n+i
m-i m
Thus
.
[B-V|2
|B|
(F
|8V |p2
2
m+2n)
2n
lI n) I v(Fm+n
- F m - i F~~)
A F m+n+i
M ^
m7
But
lim
n#-oo
)2
(F
m+n m+n+i
lim / 2 n \
n-^oo\n J
m-i
, ,
and hence
= GO
'
l i m
and
Y~
l i m c o s # = 0,
'
n-^-Qo
and
0 = ,r/2
n -#-oo
Also solved by the proposer.
MANY ROADS TO MORGANTOWN
H-68 Proposed by H. W . Gould, W , Virginia University, Morgonfown, W . Va
P r o v e that
n
n^
>
k=i k
with equality only for n = 1, 2
_ !
. n > 1
n+2
E ^ E B jJ = nEA.B. + | t t (\ - AJ )(B
- B.)
J
i=i
j=i
i=i
i=i j=i
i=i
j=i
(A. - A . ) 2
i
i=ij=i
3
i 3
1-1
j=i
80
F e b . 1967
with equality only when A. = A. for all 1 < i < n, 1 < j < n. T h e application
to the Fibonacci n u m b e r s F (with F ( J = F +F a and F i = 1, Fo = 1 )
n
n+i
n
n-l
i
* >
i s evident from the formula
F
i = Fn+2 "
1=1
so that we find
n
y l >
_B!
.^J F . " F
i=i
n+2
- 1
n2
fi B; * H
k=i
A
- H2 n - x
n+2
i - > =-5?
h L
(Zeitlin)
n n+i
n >1
(Hoggatt)
*****
Continued f r o m page 44.
REFERENCE
1.
*****
certain
F o r convenience of
P i s a l a t t i c e path,
by the step \XtY]
If
then
and Y.
A l a t t i c e path P in Euclidean 2 o r 3 - s p a c e i s said to be s y m m e t r i c with
r e s p e c t to the line k if and only if it i s t r u e that if X i s a point of s o m e s t e p
of P ,
then e i t h e r X i s a point of
k or t h e r e e x i s t s a point Y of s o m e step
(x 2 ,y 2 )] i s a step of
s o m e l a t t i c e path P in
[ X , Y ].
= 1 and x - d e c r e a s i n g
82
if x 2 - xt = - 1 . The t e r m s
defined,
y-increasing
and
[Feb.
y-decreasing
are similarly
A s t e p i s s a i d to be x y - i n e r e a s i n g if it i s both x - i n c r e a s i n g
y-increasing.
To say that
S is
x - i n c r e a s i n g only m e a n s that S i s
i n c r e a s i n g but n e i t h e r y - i n c r e a s i n g n o r y - d e c r e a s i n g
monotonically i n c r e a s i n g
P i s said to be
then 2 i s not x - d e c r e a s i n g .
2 i s a step of
A s t e p 2 i s said to be horizontal if it i s n e i t h e r
nor y-decreasing.
m e a n s that
xxP,
The t e r m y-monotonically i n c r e a s i n g i s s i m i -
l a r l y defined. A s t e p 2 i s said to be v e r t i c a l if it i s n e i t h e r x - i n c r e a s i n g
x-decreasing.
and
nor
y-increasing
i s both
x-monotonically i n c r e a s i n g
and
y-monotonically
increasing.
Part 2
In Euclidean 2 - s p a c e a path from
and
x + y = n
is
vertical.
A path having p r o p e r t y G will be called a
G-path.
T h e o r e m 1 (Greenwood)
Let g(0) = 1 and g(l) = 1.
The sequence
n > 4.
Each
G-path
or
from
(0,0) to (n - 1, n - 1) h a s a s i t s
n u m b e r of G-paths from
(0,0) to (n - 2, n - 2),
(1,0) to (n - l , n - 2) i s the
i. e. , g(n - 1).
[(0,0), (1,1) ],
then i t s
1967]
G-paths
from
(0,0)
to
83
(n - 3, n - 3), which is
g(n - 2)0
Thus
H-path.
from
(0,0)
to
the
following.
T h e o r e m 2
Let h(0) = 1 and,for each positive i n t e g e r n, let h(n) denote the n u m b e r of H - p a t h s from (0,0) to (n,n).
A method of e n u m e r a t i o n
dif-
i - 1
, where
z(2n,i) = z ( 2 n , i - 2) + z(2n - l , i - 2) ,
z(2n+l,i)
= z ( 2 n + l , i - 1) + z ( 2 n , i - 1)
84
[Feb.
let
i
f(i) = X/ ( k j i )
fc=i
In Euclidean 3-space, a
or
which have as their second term either (0,1,0) or (1,0,0) is the number of
F-paths from (0,0,0) to (n - 1, n - 1, n - 1). Hence,the number of F-paths
from (0,0,0) to (n,n,n) whose second term is either
(1,0,0) or (0,1,0)
is
2f(n - 1). Similarly, the number of F-paths from (0,0,0) to (n,n,n) whose
second term is (1,1,1) is f(n - 2). Hence, if n > 2, then f(n) = 2f(n - 1) +
f(n - 2).
1967]
85
f (x) = xf _ (x) + f
(x)
for
f0(x) = 0,
fj(x) = 1, and x = 2.
Using the methods of finite difference equations we may obtain an e x p r e s sion for calculating f(n) directly,,
C o n s i d e r again the r e c u r s i o n r e l a t i o n
f(n)
difference
equation
f(n + 2) - 2f(n + 1) - f(n) = 0 '.
The c o r r e s p o n d i n g c h a r a c t e r i s t i c equation
r 2 - 2r - 1 = 0
has roots
rt
= 1 + \f2
and
r 2 = 1 - <{2 .
Ct
and
C 2 a r e found to be
(A/2 + l)/2\l2
and
(N/2 -
l)/2\[2
f(n)
(1 + ^ ) n + 1 _ ; (1 - ^ )
2 \[2
n + 1
for
that a r e analogous to F p a t h s
86
Feb. 1967
REFERENCES
1. R. E. Greenwood, "Lattice Paths and Fibonacci Numbers, n The Fibonacci
Quarterly, Vol0 2, No. 1, pp. 13-14.
2. D. R. Stocks, J r . , "Concerning Lattice Paths and Fibonacci Numbers,"
The Fibonacci Quarterly, Vol. 3, No. 2, pp. 143-145.
3.
C. Jordan,
The references shown below are for "Iterated Fibonacci and Lucas Subscripts,"
which appears on page 89.
REFERENCES
1. EL H Ferns, Solution to Problem B-429 Fibonacci Quarterly, 2 (1964),
No, 4, p0 329.
20 I. D RugglesandV. E Hoggatt, Jr a , "A Primer on the Fibonacci Sequence, "
Fibonacci Quarterly, 1(1963), No 4, pp. 64-71
30 Raymond Whitney, Problem H-55, Fibonacci Quarterly,
Po 450
^
3(1965), No. 1,
brief
n+ii1
i = 0, 1, 2, , n , h, a r b i t r a r y positive i n t e g e r s . The t r i a n g l e
(F
n> W '
(F
n+2h s
nH-3h )s
(F
n+4h s
n+5h'
F
+ F F
- F F
- F F
2h sh
h 4h
sh 4h
*h2h
has the a r e a
'n+2h
" n+4h
n+5h
= F F
+ F
F
b a+i
b-i a
r e d u c e s to
AF*
+ BF
F + CF 2
n+i
n+l n
n
wherein
A = F ,F , + F,F. - F , F , - F,F.
,
2h 5h
h 4h
3h 4h
h 2h
B = F F
+ F F
+F
F
+ F ,
15
Jj
2h 5h-l
5h 2h-i
hr4h
*3h
F F
- F F
- F
- F F
sh 4h-i
r
4li 3h-i
h 2h-i
5h
C = F
F
+ F
F
+ F
u
2hi 5h-i
H-Ah-i
3h-i
F
F
- F
F
5h-i
*3h-r4h-i
h - i 2h-i
By u s e of the identity cited above 3 the fundamental r e l a t i o n s h i p
F
F2
n
F u r t h e r m o r e * since
1
+ F
n+i
88
2hF5h
hF4h -
hF2h -
3hF4h
F e b . 1967
E - D
COROLLARIES
lo
2.
= F
to wit, F 0 = 0, F
= 1, F
= -1, F
= 2 , and quite g e n e r a l l y , F
n+i
1
2
3
n
(-1)
F . Accordingly, the 2h chains of F i b o n a c c i - w i s e t r i a n g l e s extend
indefinitely in both d i r e c t i o n s .
3.
+ F
= F
*k+i
k+2
for
0 < k < 1,
and
F, = 1
k
for
1 < k < 2 .
~~
w h e r e i n individual
real
F i b o n a c c i n u m b e r s employed a s r e c t a n g u l a r c a r t e s i a n c o o r d i n a t e s of v e r t i c e s .
= Fg ,
a r e generalized
Fibonacci n u m b e r s introduced by H o r a d a m .
We will m a k e u s e of s e v e r a l i d e n t i t i e s .
2F ^ = F + L
n+i
n
n
(2)
-F
= 4(L- - F ) ,
l
n-i
n
n
(3)
L 2 - 5F 2 = 4 ( - l ) n
n
n
(4)
2L ^ = 5F + L
o
n+l
n
n
F r o m t h e s e H H F e r n s 11 h a s shown
(5)
(6)
Ln+1 = ^ 5 l
- 20(-l)n
Ln)
Equation (5) i m p l i e s
F Q _ I = {(sfo* + 4(-l)n - Fn)
(7)
We shall a l s o r e q u i r e
(8)
v
'
%
_,_ ^ = F F + F ^ . F , 4
m+n+l
m n
m+i n+l
m+n+l
m n
m+i n+i
it follows that
89
Since
s(n) =
90
(_l) n
F e b . 1967
(.D ^
F i r s t c o n s i d e r t h e sequence Y
= FTJ , w h e r e H obeys H , = H ,_
J
n
*%
n
n+2
n+l
+ H Then u s i n g (8), (7), and (5), we find
1
n+2
| F
H n -f 2
H n
( ^
Hn+i+Hn =
+ 4
Hn+rlFHn
Hn+1*Hn+i
( V ^ - F H n + i ) + i F H n + 1 (X/SF^ + 4(-l) H n
FHn)
n+l
5Y*
4 ( ~ l ) H n + 1 + Y ,_ V 5 Y 2 + 4 ( ~ l ) H n
n+l
n+l
n
H H = F 9 then Y = U and we h a v e
n
n
n
n
U N/SU2 +4(~l)s(n+i) + U ^>/5U2 + 4(-l)s(n)
n+i
n+l
n
Un+2
2 . n
(n > 0)
n+ 2 = f[ V n V , 5 V n + 1 + 4 (- 1 ) S ( n + 1 )
Now c o n s i d e r t h e s e q u e n c e Z
(9),
\+/5K
+ 4<-l)i
= LJJ , where H
Mxiy
is as before.
n+2
Hn+2
Hn+1+Hn = F H n + 1 - i L H n + F H n + 1 L H n + i
= TLHnLHn+1 -|LHnFHn+1 +
| L^H
+ ^V(L?_
T
H n ^LHn ^
+i^TVV^Hn+1
- 4 ( - l ) H ^ ) / 5 V5(LL - 4 ( - l ) ^ )
Z JL
+ V ( Z * , - 4 ( - l ) H n + 1 ) ( Z nu - 4 ( - l ) H n )
n+l n
n+i
Now if H
s(R+l)
)(X2 - 4 ( - l ) s ( n ) )
n+2 ~ h Xn+l Xn + \f(X\ n+i - 4 { ~ l )
n
W
W + V & 2 T ^ 4 ( - l ) D^51+55^172
^ ' r x 0 ( W 2 - At
4 ( - li\sfaK
)OUi/)
n+l n
n+i
n
See p a g e 88 for R e f e r e n c e s .
(n > 0)
Using
SUMMATION OF S j | = 1 k Tk+r
FINITE DIFFERENCE
APPROACH
! > %
k=i
or more generally
f n
[ - V)]
where f is a function of n and Fibonacci numbers involving n in their subscripts. We shall define
Af
[n'V|=
f (n+1) F
For example,
A(n 2 F n ) = (n + l ) 2 F n + 1 - n 2 F n
= n 2 F n _i + (2n + l ) F n + 1
91
SUMMATION OF 2 ^ k m F k +
92
[Feb.
n-i
(2n + D F n + 1 ] = n 2 F n + C
k FT ,
k+r
E
k=i
Then
n-l
k=i
Thus
^n'
X/^k+r
^"'(^Fn+r)
k=i
We need then simply to evaluate this inverse finite difference in order to obtain
an expression for the summation.
We develop certain relations for this purpose.
(1)
(2)
A ( n F n + r + 1 ) = (n + l ) F n + r + 2 - n F n + r + 1 = n F n + r + F n + r + 2
and in general
(3)
1967]
93
A ( n F n + r ) = n F n + r + 1 - A (F n + r + 2 ) = n F n + r + 1 - F n + r + 3 + C .
= J]' ( " 1 ) t A t ( n m ) F n + r + 2 t + 1 +
t=0
That this result is correct may be shown by calculating
AtA-Wn+r)]
from the summation in (6). The result is n m F n + r as can be readily seen from
the fact that apart from the first term in the expansion all succeeding terms
cancel in pairs.
The results for the first two terms will show the pattern,
A ( n m F n + r + 1 ) = n H i F n + r + A(nm)F n + r + 2
by (3)
AA(nm)F n + r + 3 ] = -A(n + l ) m F n + r + 4 + A ( n m ) F n + r + 3
= -A(nm)F n + r + 4 - A 2 ( n ^ ) F n + r + 4 + A(nm)F n + r + 3
= -A(nm)Fn+r+2 - A2(nm)Fn+r+4
Hence (6) provides the required formula apart from making explicit the coefficients in terms of n and calculating the undetermined constant. The former
are given subsequently in tables; the latter may be obtained as shown below for
the particular case in which m = 5.
We set n = 2 in (6) so that
94
kmFk+r
SUMMATION OF 2 ^
[Feb.
F r + 1 = 3 2 F r + 3 - 2 1 1 F r + 5 + 5 7 0 F r + 7 - 7 5 0 F r + 9 + 480 F r + 1 1 - 1 2 0 F r + 1 3 + C
Using the f o r m u l a s
Fn -
Fk+1Fn_k +
F ^ . ^
and
= (-D^FfeFn+fcH - Fk+1Fn+k)
Table 1
COEFFICIENTS O F A ( n m )
m
n2
n4
n3
n5
n6
n8
n7
10
10
15
20
15
21
35
35
21
28
56
70
56
28
36
84
126
126
84
36
10
10
45
120
210
252
210
120
n5
n*
9
45
10
Table 2
COEFFICIENTS OF A 2 (nm )
m
n2
n4
n3
n7
14
24
30
70
60
20
62
180
210
120
30
126
434
630
490
210
42
254
1008
1736
1680
980
336
56
510
2286
4536
5208
3780
1764
504
72
5100
;11430
15120
13020
7560
2940
720
10
1022
12
90
19671
FINITE D I F F E R E N C E APPROACH
95
Table 3
COEFFICIENTS O F
n4
n3
A3(nm)
n5
n*
36
24
150
180
60
540
900
540
120
7
8
1806
3780
3150
1260
210
5796
14448
15120
8400
2520
336
18150
52164
65016
45360
18900
4536
504
10
55980
181500
260820
216720
113400
37800
7560
720
Table 4
4
COEFFICIENTS O F A (nm)
n2
24
240
120
1560
1440
360
8400
10920
5040
n5
n4
n3
840
40824
67200
43680
13440
1680
186480
367416
302400
131040
30240
3024
10
818520
1864800
1837080
1008000
327600
60480
5040
Table 5
COEFFICIENTS OF A 5 ( n m )
m
5
n2
n3
n4
120
1800
720
16800
12600
2520
126000
134400
50400
6720
834120
1134000
604800
151200
10
5103000
8341200
5670000
2016000
15120
378000
30240
SUMMATION O F S ^ = 1 k m F k + r
96
Table 6
COEFFICIENTS O F
[Feb.
A 6 (n m )
n2
n3
720
15120
5040
191520
120960
20160
1905120
1723680
544320
60480
10
16435440
19051200
8618400
1814400
Table 7
COEFFICIENTS OF
m
151200
A 7 (n m )
n2
n3
5040
141120
40320
2328480
1270080
181440
10
29635200
23284800
6350400
604800
Table 8
COEFFICIENTS OF A 8 ( n m )
m
8
n'
40320
1451520
362880
10
30240000
14515200
1814400
Table 9
COEFFICIENTS O F A 9 (n m )
n
362880
10
16329600
3628800
Table 10
COEFFICIENTS O F A 1 0 ( n m )
m
10
3628800
1967]
97
Table 11
SUMMATION CONSTANTS
m
1
2
3
4
5
6
7
8
9
10
Summation Constants
F
r+3
-F
r+6
7F , + 5 F ,
r+5
r+4
-37 F , - 24 F ,
r+6
r+5
242 F _ + 147 F ^
r+7
r+6
-1861 F ^ o - 1139 F L^
r+8
r+7
16679 F ^ + 1 0 3 2 4 F i o
r+9
r+8
- 1 7 1 3 6 2 F , - 106089 F ,
r+10
r+9
1981723 F _ + 1224729 F ^ 4 A
r+ii
r+io
-25453505 F ^ - 15726832 F ,
r+i2
r+11
k7 FT ^ = n 7 F ^ _^ - v(7n6 + 21 n 5 + 35 n 4 + 35 n 3 + 21 n 2 + 7n + 1)F , ^
k+r
n+r+i
n+r+3
k=i
+ (v 4 2 n 5 + 2 1 0 n 4 + 4 9 0 n 3 + 630n 2 + 4 3 4 n + 126) F , ,
n+r+5
- ( 2 1 0 n 4 + 1260n 3 + 3150n 2 + 3 7 8 0 n + 1806) F , L
v
/ n+r+7
+ (840 n 3 + 5040 n2 + 10920 n + 8400)F ^ _,_
^
n+r+9
- (2520 n2 + 12600 n + 16800) F . , .. + (5040 n + 15120)F , ,. Q
n-i r+ 11
n+r+iu
- 5040 F ^ ^
n+r+15
CALCULATION BY FINITE D I F F E R E N C E S
Except for the s m a l l e r v a l u e s of m ,
k5 F
*k+7
98
SUMMATION OF 2 **
k m F,
[Feb.
50
312500000
51
345025251
52
380204032
53
418195493
54
459165024
55
503284375
Then
A [ k 5 ] k = 5 0 = 34502521 - 312500000 = 32525251
A2 k5
[ ]k=50
A3
380204032
A4 k
[ ]k=
459165024
"
A 5 [k 5 ]k=50
120
* * *
Recently the author has conducted in-service training sessions in mathematics for the elementary school teachers of the Williams town, Massachusetts
public schools,, During a session on the lowest common multiple and greatest
common divisor of two positive integers, two teachers observed that if the two
numbers are in the ratio 2:3, then the sum of the numbers is equal to the difference between their lowest common multiple and their greatest common divisor.
It is shown in [2] that this is the only ratio for which this relation holds.
Of course, one gets similar relations for other ratios.
For example, if
the two numbers are in the ratio 3:5, then twice their sum is equal to the sum
of their lowest common multiple and their greatest common divisor.
Again it
is shown in [2] that this is the only ratio for which this relation holds. This is
not always the case since, for example, both ratios 5:7 and 4:11 yield the
result that three times the sum of the numbers is equal to the sum of their lowest common multiple and their greatest common divisor.
If one specializes to the Fibonacci and the Lucas sequences, one gets
theorems of the type given below, in which families of such relations are exhibited and formulas for finding all ratios satisfying these relations a r e obtained.
Let {F n } be the sequence of Fibonacci numbers, where F t = 1, F 2 = 1
and F n + 2 = F n + F n + 1 for n > 1.'
Let {LJQ} be the sequence of Lucas numbers, where
and 1^+2 = L n + Ln+i for n ^ 10
The following known results are assumed.
L 1 = 1, L2 = 3
(See [1] or [ 3 ] , J
(i)
(ii)
F n + 1 = FnFn+2 + (-l)n .
(iii)
F 2n _i = F n F n + 1 - F n _ t F n _ 2
(iv)
(v)
F2n = FnLn
(vi)
Ln+1 = F n + Fn+2 .
99
100
ON RATIOS OF FIBONACCI
For the remainder of th^ article, let
[Feb.
for n >
2.
and
Then a = F n+1 k,
b = F n k, (a,b) = k
Then
F n + 1 (a + b) = F n ^(F n -f F n+1 )k = F n . 4 F n ^ 2 k = (F n + 1 - F n )F n + 2 k
= FttH(Fn+Fn+1)k - FnFn+2k
=
n F n+ik + (F2n+i - F n F n + 2 ) k
= [ a f b ] + (-l) n (a,b)
(2) If b = (c/d)a,
for n > 2 .
a-i(
+ b
>
Then
for n > 3
implies
F ^ f c + d)= cd + (-l) n
for which we wish to find all positive integral solutions. Solving for c, we get
I96t]
Fl_t
=
101
Fn_t
(-if
n-i
d -
Fn_{
s o that by (ii),
11-2-^11
c = Fn-i
We need T '
d -
Fn^
v-i
, then c < 0.
n~i5
The totsj : iu . i b r
.- ' . a e ^ ; -i roUifew^ a ^
divisors of
,,
v ir /
that if c - A,
rf
8^> f.i^f
P ^
\ f
so
b e r of distinct nc<)vil'r V
'- * r?-i j ?-d is one-half the n u m b e r of d i v i s o r s of
^n-2^n*
Fiu?v?y. iC rl ~ <lH t
ii-3-
<1
n-2-^n
=
x
and the r a t i o
Fn:FT1+1
E x a m p l e : If n = 8,
n-i
Fn_2
t h e n ' F n _ 2 = 8,
F n - 1 = 13,
F n = 21, and F n + 1 = 3 4 .
then
13(a + b) = [ a , b ] + (a 5 b)
(2) If b = (c/d)a,
then 13(a + b) = [ a , b ] + ( a , b )
implies that
168
c = 13 + d - 13
and
25:27,
102
ON RATIOS OF FIBONACCI
f Feb.
Lemma: F 2 n _ 1 = F j - ^ L ^ - L n L n + 1 for n ^ 2.
Proof: The proof is by induction.
2k-i
k L k+l
Then
F
2k+i
2k
kLk
2k-i
( F k+2 -
kLk
F
k+i L k+2 -
k) L k+2 -
= F k L k + Fk+2(Lk+3 "
kLkH
kLk+i
k+i) "
kLk+2 "
kLk+i
k L k + i - ( L k+2 -
k+i L k+2
k + i ( L k + i ~ F k+2 "
k+i)Lk+i
F
k)
But
L
k+i - Fk+2 ~ F k
by (vi), so that
F
2k+i
then
for n > 2 .
n+i( a
+ b
>
[a>b]
2n~i(a>b) for n ^ 3 .
1967]
103
For n 3,
(F^+j - F2n-.1)
Among
Then
a = L n+1 k,
b = L n k,
(a,b) = k
and
[a,b] = L n L n + 1 k
Then
F n + i (a,b) = F n + 1 (L n + L n + 1 )k = F n + 1 L n + 2 k
so that
as required.
(2) If
n-2
n-i
b =
then
a
n-ik >
n-2k >
(a>)
and
[a>b]
"
n-iFn-2k
104
ON RATIOS OF FIBONACCI
F n + 1 (a + b) = F ^ F ^
=
( F 2n-i
[Feb.
+ F n _ 2 )k F
Fn+1Fnk
n-iFn-2)k
= [ a , b ] +. F ^ a . b )
as required.
(3) If b = (c/d)a, where c and d are relatively prime, then, once
again
a = dk,
b = ck,
(a,b) = k
and
[ a , b ] = cdk
The relation
F n + 1 (a + b) = [ a , b ] + F 2n i(a 9 b)
implies
F n+1 (c + d) cd + F 2 I W
Solving this equation for c, we get
F n + 1 d - F 2 n -i
_ _ _ _ _
^
=
Fn+l
F n + 1 - F2n_1
^ - ^ r ^
d = L n is also a solution.
n+iLn+i ~ F 2n-i _
So let d = L n + 1 .
Ln+i ~ F n+i
Then
n L n+l
n+i ~ F n+i
"LnFn+i
By
n L n+i
n+i ~ F n+i
= Ln
1967 ]
105
The situation here differs from that in the second part of Theorem 1,
for not all solutions are obtained by considering the case d >F n + lo
example, let d = Fn_lffl
F
Q
n+iFn-i -
- F n+i F n-2
n-2
:F
n-i -
For
2 n-i
:
n*l
=T
, .
n+i
n-i
n-iFn-2
" ^
"
n-i
*s
n-i -
n+i
rn
solution,,
~2
This completes the proof of the
= 8, F n + 1
= 21 and F 2 n -i = 233.
(1) and (2): If a and b are in the ratio 29:47 or 5:8, then
21 (a + b) = [a s b] + 233(a,b)
(3): If b = (c/d)a,
then
21 +
M1^|33
d - 21
2 1 +
"
208
d - 21
The divisors of 208 are 1, 2, 4, 8, 13, 16, 26, 52, 104 and 208.
The solu-
tions are 22:229, 23:125, 25:73, 29:47, 34:37 and 5:8. Among these ratios
are 29:47 = L ^ L ^ i ' and 5:8 = F ^ i F ^ .
ACKNOWLEDGEMENTS
The final version of this article was written while the author held a National
Science Foundation Science Faculty Fellowship. The author wishes to thank the
reviewer and Professor V. E. Hoggatt, J r . , for their helpful suggestions and
comments.
106
Feb. 1967
Manuscripts intended
quests the Fibonacci Quarterly to be forwarded at first class rates to the new
address, he will not receive it. (This will usually cost about 30 cents for firstclass postage.) If possible, please notify us AT LEAST THREE WEEKS PRIOR
to publication dates: February 15, April 15, October 15, and December 15.
*
Uni-
Each p r o b l e m o r
Solutions should be
1 3
1 3
and G
b e r e s p e c t i v e l y the n
t e r m s of the s e q u e n c e s
Gil = 1,
Go
L = 2e
Mi = 3,
(of
and G n = G ^ +
P r o v e that
Mn = 1 + K
L V/ 5 G
m J1 ,
w h e r e m = 2 n - 1 and [ x ] i s t h e g r e a t e s t integer function,,
B-108 Proposed by V . E. Hoggatt, J r . , San Jose State College, San Jose, Calif.
Let Ui = p ,
+u .
y 2 = q,
and u n + 2 = u n + 1 + u n o Also l e t S n = Ui + u 2 +
B-109 Proposed by V . E. Hoggatt, J r . , San Jose State College, San Jose, Calif.
Let r and s b e t h e r o o t s of the q u a d r a t i c equation
(r f s ) .
Let U
= (r11 - s n ) / ( r - s) and V
= rn + sn.
V = U _, + qU ,
n
n+i
^ n-l
107
x 2 - px - q = 05
Show that
108
[Feb.
oo
oo
=V5V<^
X4 r 2 n + i
n=o
-*> L 2 n + i
n=o
J F 4 n + 2
oo
~* L 4n+2
SOLUTIONS
LUCAS NUMBERS MODULO 5
B-88 Proposed by John Wessner, Melbourne, Florida.
Show
that
L2k = 2 ( - l ) k
(mod 5)
A CLOSE APPROXIMATION
B-89 Proposed by Robert S. Seamons, Yakima Valley College, Yakima, Wash.
0,2m +
1967]
<
a2m
which i m p l i e s a2111 + / 3 2 m = [ l + # 2 m - / 3 2 m ] ,
^m
109
as desired.
ni
XL,
_*i
k=i
n
=
k=i
(b, + 2) = 5 n (bk + 2)
k
k=2
n-i
n-i
= 5 n ( b . + 1 + 2) = 5 n b 2
j=i
or
j=i
n-i
b
- 2 = 5 FI b 2
j=1 3
Thus,
c = b ,
and
n-i
d = II b . ,
n = 2,3,
110
[Feb.
CONVERGENCE OF SERIES
B-91 Proposed by Douglas Lind, University of Virginia, Charlottesville, V a ,
If F
is the n
I>/ F j>
i=i
converges while
Z(1/lnFj)
j=3
diverges.
Solution by Jeremy C* Pond, Sussex, England*
F
n+l
=
n
^ 1 +\/5
v
> 1 as
and so
Also,
(l/lnFn)/(l/n)-^l/ln(^-i^j
and so 1/ln F. and 1/n diverge together.
Also solved by C* B. A, Peck and the proposer*
> 0
n (
1967]
111
Show that
= F
m+n
F
+ F F
n-l m
n m+i
page 2 3 ,
, F ) = F,
* The d e s i r e d r e s u l t then follows I m m e d i m n
(m,n)x
ately from the easily e s t a b l i s h e d fact that (m,n) = (m,m + n) = (n,m + n).
MODULO n
n
112
Feb. 1967
(2j)=
(mod
> 0
l>2>'->
n / 2 ),
so that
\ 1 (S)/2""' = ^ '
>
= 1 (mod n) if
n is prime.
I have not been able to prove or disprove the converse of this statement.
A calculation by computer indicates that the converse is true for n < 700.
Also solved by the proposer who stated that the converse is false and gave 705,
2465, and 2737 as the first few composite values of n.
*****
NOTICE
George Ledin, J r . has been appointed by The Fibonacci Association to collect
and classify all existing Fibonacci Identities, Lucas Identities, and Hybrid
Identities. We request that readers send copies of their private lists (with
possible reference sources) to
George Ledin, J r .
445 Monticello
San Francisco, Calif. 94127
for inclusion in the planned booklet.
Verner E. Hoggatt, J r . ,
Director
*